PhD Thesis by Stuart P Edwardson

Department of Engineering

PhD Thesis

A Study into the 2D and 3D Laser Forming of Metallic Components

Thesis submitted in accordance with the requirements of The University of Liverpool for the degree of

Doctor in Philosophy

By

Stuart Paul Edwardson

March 2004 Stuart P. Edwardson Laser Group Department of Engineering The University of Liverpool Liverpool, UK L69 3GH Email: [email protected] Web: www.lasers.org.uk


Stuart P. Edwardson PhD Thesis i

Declaration

I hereby declare that all of the work contained within this dissertation has not been

submitted for any other qualification.

Signed:

Date:

Word Count = 91360


Stuart P. Edwardson PhD Thesis ii

Abstract

The work presented in this thesis is primarily concerned with the process of laser

forming or laser bending of metal sheet material with a high power infra-red

defocused laser beam.

Presented in this thesis are results of investigations into the 2D and 3D laser

forming of metallic components. 2D laser forming encompasses laser forming

operations that utilise two dimensional out-of-plane bends to produce three

dimensional results e.g. a fold. 3D laser forming encompasses laser forming

operations that can utilise combinations of multi-axis two dimensional out-of plane

bends and in-plane localised shortening to produce three dimensional spatially

formed parts e.g. a dome.

There has been a considerable amount of work completed on 2D laser

forming to date, however due to the many variables in the process and numbers of

materials and material types that can be laser formed a full understanding of the

process is some way off. The work on 2D laser forming presented in this thesis aims

to increase the knowledge and understanding of the process, in particular the

transient thermo-mechanical and asymmetrical effects plus aspects for closed loop

controlled LF. Materials investigated include mild steel, aluminium AA1050,

aluminium AA6061, Ti6Al4V and newly developed Metal Laminate Composite

Materials sometimes referred to as Fibre Metal Laminates.

In order to advance the laser forming process still further for realistic forming

applications and for straightening and aligning operations in a manufacturing

environment it is necessary to consider 3D laser forming. Less work has been

completed in this field compared to 2D laser forming, however the process has been

shown to have a great deal of potential. In order to compete directly with

conventional forming techniques though, such as die forming the process must be

proven to be reliable, repeatable, cost effective and flexible. The work presented in

this thesis on 3D laser forming aims to prove the viability of this technique as a

direct manufacturing tool and as a means of post-conventional forming (or

processing e.g. chemical etching) distortion removal. To this aim progress towards

repeatable closed loop controlled 3D LF is presented. The materials investigated

were mild steel and Ti6Al4V.


Stuart P. Edwardson PhD Thesis iii

Acknowledgements

The author gratefully acknowledges all of the contributions and help given in order

complete this work. In particular acknowledgements are given to the following

people and organisations:

I would like to thank all of the members of the Laser group past and present

that have contributed to this work. In particular my supervisors/advisors Professor

Ken Watkins, Dr Geoff Dearden and my original supervisor Dr Jonathan Magee for

help and support throughout my PhD. Thanks are also given to the lab manger Andy

Snaylam and the mechanical and electronics technicians John McCulloch and Dave

Blanchard without whom the research wouldn’t exist.

In addition I would like to thank Professor Wesley Cantwell of the impact

research centre for contributions and collaborations within this research in more

recent times.

I would like to thank everybody at the Lairdside Laser Engineering Centre

(LLEC) for the use of their facilities and expertise; these include Dr Martin Sharp,

Dr Paul French (also of the laser group) and Anthony Walker.

An acknowledgement and thanks for the contribution to the development of a

3D laser forming model and closed loop system are given to Dr Andrew Moore of

Heriot Watt University, Edinburgh.

I would like to thank all of the Students, both undergraduate and post-

graduate, that I have worked with during the course of my PhD for the contributions

made to it, these include; Heather Tjia, Jonathan Howard, Chiung-Hao Chen (Jason),

Ian McArthy, Paul Simpson, Konstantinos Baltas, Marcus Rashford, Gabriel Cooke

and Tejas Voralia.

I would like to thank the EPSRC, BAE SYSTEMS and Rolls-Royce for the

funding of the project and providing materials. In particular at BAE SYSTEMS I

would like to thank Professor Len Cooke, Dr Jagjit Sidhu and Dr Neil Calder and Dr

Jeff Allen at Roll-Royce for their help and assistance throughout.

Finally I would like to thank all my family and friends for supporting me

over the years, special thanks go to Eleanor without whom I couldn’t have done it.


Stuart P. Edwardson PhD Thesis iv

Table of Contents

Declaration i

Abstract ii

Acknowledgements iii

Table of Contents iv

List of Figures viii

List of Tables xxiv

List of Symbols xxvi

1.0 Introduction 1 2.0 Literature Review 4 2.1 Introduction 4 2.2 Process Origins 4 2.3 Laser Forming Mechanisms 5 2.3.1 The Temperature Gradient Mechanism (TGM) 7 2.3.2 The Buckling Mechanism (BM) 10 2.3.3 The Shortening or Upsetting Mechanism (UM) 13 2.4 Analytical Models 15 2.4.1 Two Layer Models for the TGM 15 2.4.2 The Residual Stress Model for the TGM 20 2.4.3 The Buckling Mechanism 25 2.4.4 The Shortening Mechanism 27 2.5 Numerical Models 28 2.5.1 Temperature Gradient Mechanism 29 2.5.2 The Buckling Mechanism 30 2.5.3 The Shortening Mechanism 32 2.5.4 Further Numerical Modelling 33 2.6 Previous Experimental Work 35 2.6.1 Fundamental Investigations 35 2.6.2 Magee ’98 41 2.6.3 Recent research in macro-scale 2D LF 47 2.6.4 Recent advances in 2D LF for micro-scale applications 49 2.6.5 Developments towards 3D LF capability 52 2.6.6 Material and Metallurgical Studies 54 2.7 Potential Applications & Competing Processes 57

2.7.1 Projections for Potential Applications of Laser Forming in Shipbuilding 58

2.7.2 Potential Applications in the Aerospace Sector 60 2.8 State of the Art 62 2.9 Synopsis for Present Research 63 3.0 Experimental Procedure 64

3.1 General Set-up 64


Stuart P. Edwardson PhD Thesis v

3.1.1 Hardware 64 3.1.2 Software 73 3.1.3 Absorptive Coatings 78 3.2 2D Laser Forming 83

3.2.1 Empirical Study - Characterisation of the Laser Forming Process 83

3.2.1.1 Mild Steel CR4 84 3.2.1.2 Ti6Al4V 86 3.2.1.3 AA 1050 88 3.2.1.4 AA 6061 89 3.2.2 Thermal Analysis 92 3.2.2.1 Thermocouple Study 92 3.2.2.2 Thermal (IR) Imaging Study 94 3.2.2.3 Forced Cooling Study 97 3.2.3 Displacement / Time Analysis 98

3.2.4 Strain Gauge Analysis 99 3.2.4.1 Transverse Strain 100 3.2.4.2 Longitudinal Strain 102

3.2.5 Finite Element Analysis (FEA) 102 3.2.6 Metallurgical Study 105 3.2.7 Closed Loop Control 108 3.2.8 Thick Section and Large Area 2D Forming for Ship Building 109 3.2.9 Laser Forming of Novel Materials –

Metal Laminate Composite (MLC) Materials 112 3.2.9.1 Materials 112 3.2.9.2 Experimental 114

3.2.10 Application Example – Aero Engine Strut 116 3.3 3D Laser Forming 119 3.3.1 Empirical Study 121 3.3.2 Development of a Geometry based Model for

3D Laser Forming using Matlab 124 3.3.3 3D Laser Forming Demonstrator System 126

4.0 2D Laser Forming – Results and Discussion 128 4.1 Empirical Study - Characterisation of the Laser Forming Process 128 4.1.1 1.5mm Mild Steel CR4 129 4.1.2 0.9-3.2mm Ti6Al4V 137 4.1.3 0.9mm AA1050 156 4.1.4 1.6mm AA 6061 O/T4/T6 160 4.2 Thermal Analysis 171 4.2.1 Thermocouple Study 171 4.2.2 Thermal (IR) Imaging Study 177 4.2.3 Forced Cooling Study 185 4.3 Displacement / Time Analysis 189

4.4 Strain Gauge Analysis 195 4.4.1Transverse Strain 195 4.4.2 Longitudinal Strain 204 4.5 Finite Element Analysis (FEA) 211 4.5.1 Development of a Graded Mesh Model 211 4.5.2 Thermal Analysis 213


Stuart P. Edwardson PhD Thesis vi

4.5.3 Displacement 222 4.5.4 Transverse Strain E11 226 4.5.5 Longitudinal Strain E22 230 4.5.6 Transverse Stress S11 233 4.5.7 Longitudinal Stress S22 236 4.6 Metallurgical Study 239 4.6.1 1.5mm Mild Steel CR4 239 4.6.2 1.6mm Al6061 O/T4/T6 252 4.7 Closed Loop Control 259 4.8 Thick Section and Large Area 2D Laser Forming for Ship Building 265 4.9 Laser Forming of Novel Materials –

Metal Laminate Composite (MLC) Materials 274 4.9.1 Feasibility Study 274 4.9.2 Laser Forming Characteristics of MLC Materials 276 4.9.3 Implications of Laser Forming on Material Integrity 276 4.9.4 Laser Forming of More Complex MLC Components 281 4.9.5 Laser Forming of Thermosetting MLC Materials 283

4.10 Application Example – Aero Engine Strut 287 5.0 3D Laser Forming – Results and Discussion 293 5.1 Empirical Study 293 5.1.1 The Saddle Shape 293 5.1.2 The Pillow Shape 303 5.1.3 The Twisted Shape 305 5.1.4 Thick Section 3D Laser Forming for Ship Building 310

5.2 Development of a Geometry based Model for 3D Laser Forming using Matlab 314

5.2.1 Initial Predictions and Results of Scan Paths for the Pillow Shape 314

5.2.2 Application of the Model to the Saddle Shape 323 5.2.3 Developable and Non-Developable Surfaces –

Bending Strain and In-Plane Strain Requirements for 3D Laser Forming 326

5.3 3D Laser Forming Demonstrator System 330 6.0 Conclusions and Future Work 343 6.1 Conclusions 343

6.1.1 2D Laser Forming Empirical Study 343 6.1.2 Thermal Analysis 345 6.1.3 Displacement / Time Analysis 347 6.1.4 Strain Gauge Analysis 348 6.1.5 Finite Element Analysis 349 6.1.6 Metallurgical Study 350 6.1.7 2D Closed Loop Control 351 6.1.8 Thick Section and Large Area 2D Forming for Ship Building 352 6.1.9 Laser Forming of Metal Laminate Composite Materials 353 6.1.10 Application Example – Aero Engine Strut 354 6.1.11 3D Laser Forming Empirical Study 354 6.1.12 Development of a Geometry based Model for

3D Laser Forming using Matlab 355


Stuart P. Edwardson PhD Thesis vii

6.1.13 3D Laser Forming Demonstrator System 357 6.2 Future Work 359

References 361 List of Publications to Date by the Author 373 Appendix 375 A1. Matlab Code I A2. Abaqus Input Files IV A3. Beam Diameter Prediction X A4. Safety Interlocks & System Layout for the Electrox Workstation No. 2 XIII A5. MEL M5 & M1 Laser Range Finder Specifications XV A6. Example Galil CNC code XVIII

A Study into the 2D and 3D Laser Forming of Metallic Components List of Figures

Stuart P. Edwardson PhD Thesis viii

List of Figures

Figure 1.1: Examples of 2D forming to produce a 3D part, and 3D forming to produce a spatially formed part. 2

Figure 1.2: Laser formed examples of 2D forming to produce a 3D part, and 3D forming to produce a spatially formed part, both in Aluminium. 3

Figure 2.3.1: Laser Forming Set-up & Process Variables 5

Figure 2.3.2: The Laser Forming Mechanisms 6

Figure 2.3.3: Energy conditions required for the TGM 7

Figure 2.3.4: Principle of the Temperature Gradient Mechanism (TGM) 9

Figure 2.3.5: Stages in the Buckling Mechanism (BM) 11

Figure 2.3.6: The Upsetting (Shortening) Mechanism (UM) 13

Figure 2.4.1: Forces and moments acting in the two layer model 15

Figure 2.4.2: Comparison of solutions for the two layer models 20

Figure 2.4.3: Layout for the residual stress model 20

Figure 2.4.4: Critical operating region for the TGM 25

Figure 2.4.5: Model Geometry for the Buckling Mechanism 26

Figure 2.5.1: Development of the bending angle during Buckling Mechanism 31

Figure 2.5.2: Distribution of the upper and lower surface temperatures and displacements during the Buckling Mechanism 31

Figure 2.5.3: Distribution of the upper and lower surface strains during the Buckling Mechanism 31

Figure 2.5.4: Plastic restraining in extrusion bending 32

Figure 2.6.1: Time run of the strain development 38 Figure 2.6.2: Time run of the bend angle 38 Figure 2.6.3: Decreasing bend rate with increasing scans over an identical track 39 Figure 2.6.4: Bend angle with increasing traverse velocity for Ti6Al4V using a

large beam diameter 42 Figure 2.6.5: Bend angle with increasing traverse velocity for AA 2024 T3 42 Figure 2.6.6: Bend angle with increasing number of scans over the same track 43 Figure 2.6.7: Ideal bend angle and exaggerated view of edge effects 43 Figure 2.6.8: Demonstrator Part 45 Figure 2.6.9: Circle line system with square root radius increase

(inside to out), and resulting contour plot of sample 46 Figure 2.6.10: Actuator for CD lens adjustment by micro LF 50

Figure 2.6.11: LF of 50µm thick beams in wet-etched silicon micro-scale structures 51

Figure 2.6.12: 3-D Laser Forming: routes to practical realisation and key elements required 53

Figure 2.6.13: SEM micrographs of Ti6Al4V formed in (a) air and (b) argon. (Forming parameters: 760W / 30mms-1). 55

Figure 2.6.14: Hardness variation with depth through the sheet thickness for Ti6Al4V (760W 30mm/s 6mm beam). 55

Figure 2.7.1: Some Current Forming Techniques in Shipbuilding 58 Figure 2.7.2: Bulbous Bow from the QM2 59 Figure 2.7.3: Hot creep formed ‘A’ frame strut, possible to

manufacture using LF 62

Page No.


Stuart P. Edwardson PhD Thesis ix

Figure 3.1.1: Electrox 1.5kW CO2 Laser (exterior enclosure) 67 Figure 3.1.2: Laser Cavity, Heat Exchanger and Cavity Discharge 67 Figure 3.1.3: Electrox 1.5kW laser beam energy profile, PyroCam III image 68 Figure 3.1.4: Workstation 2, 3 Axis beam manipulation 68 Figure 3.1.5: Workstation 2, CAD Drawing of layout 69 Figure 3.1.6: MEL M5 Laser Range Finder 70 Figure 3.1.7: Centre Clamp 70 Figure 3.1.8: Edge Clamp 70 Figure 3.1.9: Corner Clamp 71 Figure 3.1.10: Un-clamped with guides 71 Figure 3.1.11: Burn prints in wood at 5mm Z steps, 127mmFL lens

130mm – 220mm stand-off 71 Figure 3.1.12: Power offset calibration graph 72 Figure 3.1.13: Agilent 34970A Data Acquisition unit 73 Figure 3.1.14: Control Application User Interface 74 Figure 3.1.15: Basic bend angle measurement using two height readings 75 Figure 3.1.16: Improved bend angle measurement accounting for any initial angle 75 Figure 3.1.17: User Interface for the automated 2D laser forming of 80x80mm

coupons 76 Figure 3.1.18: Co-ordinate Measuring Machine (CMM) User Interface 77 Figure 3.1.19: Example CMM output 77 Figure 3.1.20: Reflectivity of various metals as a function of wavelength 78 Figure 3.1.21: Absorptivity of various metals as a function of wavelength at room

temperature 79

Figure 3.1.22: Absorption of CO2 laser light on steel at room temperature dependent on surface condition 80

Figure 3.1.23: Dependence of coupling rate of coated surfaces on interaction time and incident intensity 81

Figure 3.1.24: Graphite Coated Sample 82 Figure 3.1.25: Example of coating degradation, Ti6Al4V, 20 Passes,

740W, 5.5mm∅, 45mm/s 82 Figure 3.2.1: Experimental set-up for 2D laser forming characterisation 83 Figure 3.2.2: Thermocouple locations used on the 80x200mm sample 92 Figure 3.2.3: Thermocouple attachment using Thermo-pads. 93 Figure 3.2.4: Thermocouple study experimental set-up. 93 Figure 3.2.5: The Thermovision® 880 Infrared Detector Set-up 94 Figure 3.2.6: Optical & IR images of a graphite coated sample 95 Figure 3.2.7: Optical & IR images of a masked graphite coated sample

showing differences in emissivity causing false readings 96 Figure 3.2.8: Forced cooling experimental set-up 97 Figure 3.2.9: Displacement / Time analysis experimental set-up 98 Figure 3.2.10: Quarter Bridge Circuit 99 Figure 3.2.11: Experimental Set-Up for Strain Gauge Analysis 100 Figure 3.2.12: Schematic Showing Strain Gauge Locations for the

Transverse Strain Study 101 Figure 3.2.13: Strain Gauges Attached to a Coupon 101 Figure 3.2.14: Schematic Showing Strain Gauge Locations for the

Longitudinal Strain Study 102 Figure 3.2.15: Locations for Hardness tests in the AA6061 study 107 Figure 3.2.16: Locations for Hardness tests in the mild steel CR4 study 107


Stuart P. Edwardson PhD Thesis x

Figure 3.2.17: Software User Interface for closed loop 2D laser forming 108 Figure 3.2.18: Initial study Set-up 110 Figure 3.2.19: Ferranti 8kW CO2 laser, 0.9x1.5m table,

800x400mm sample 110 Figure 3.2.20: 5 Axis beam delivery system 110 Figure 3.2.21: Schematic of the MLC lay-ups used in the investigation 113 Figure 3.2.22: 4/3 Polyamide based FML as Received Section 114 Figure 3.2.23: MLC Experimental Set-up 116 Figure 3.2.24: CAD drawing of an RR Trent 700 Aero Engine ‘A’ frame strut

component. 117

Figure 3.2.25: CAD drawing of an RR Trent 700 Aero Engine ‘A’ frame strut component (magnified). 117

Figure 3.2.26: Set-up for the laser forming of the strut section from 200x100mm 1.6mm Ti64 sheet. 118

Figure 3.2.27: Set-up for the full scale laser forming of the strut section from 3.2mm Mild Steel sheet. 118

Figure 3.3.1: 3D Primitive, ‘The Saddle Shape’ 119 Figure 3.3.2: 3D Primitive, ‘The Pillow Shape’ 119 Figure 3.3.3: 3D Primitive, ‘The Twisted Shape’ 120 Figure 3.3.4: Experimental Set-up for the 3D Laser Forming empirical study 121 Figure 3.3.5: Set-up for thick section 3D LF 122 Figure 3.3.6: 3D Laser Forming using a Laser Dyne 890 5 Axis Gantry 123 Figure 3.3.7: Improved 3D Laser Forming Set-up 125 Figure 4.1.1: 2D LF process map for 1.5mm mild steel CR4, 3mm Beam Dia 129 Figure 4.1.2: 2D LF process map for 1.5mm mild steel CR4, 5.5 mm Beam Dia. 129 Figure 4.1.3: 2D LF process map for 1.5mm mild steel CR4, 8mm Beam Dia. 129 Figure 4.1.4: 1.5mm mild steel CR4, 3mm Beam Dia., 760W, 55mm/s,

30 pass 132 Figure 4.1.5: 1.5mm mild steel CR4, 5.5mm Beam Dia., 760W, 30mm/s,

30 pass 132 Figure 4.1.6: 1.5mm mild steel CR4, 8mm Beam Dia., 760W, 20mm/s,

30 pass 132 Figure 4.1.7: Laser forming of 1.5mm mild steel CR4, 3mm Beam Dia.,

at various inter-pass time delays 134 Figure 4.1.8: Effect of inter-pass time delay on the laser forming of

1.5mm mild steel CR4, 3mm Beam Dia. 135 Figure 4.1.9: Laser forming of 1.5mm mild steel CR4, 5.5mm Beam Dia.,


1.5mm mild steel CR4, 5.5mm Beam Dia. 135 Figure 4.1.11: Laser forming of 1.5mm mild steel CR4, 8mm Beam Dia.,


1.5mm mild steel CR4, 8mm Beam Dia. 136 Figure 4.1.13: 2D LF process map for 0.9mm Ti64, 3mm Beam Dia. 137 Figure 4.1.14: 2D LF process map for 0.9mm Ti64, 5.5mm Beam Dia. 137 Figure 4.1.15: 2D LF process map for 0.9mm Ti64, 8mm Beam Dia. 138 Figure 4.1.16: 0.9mm Ti64, 3mm Beam Dia., 500W, 40mm/s, 30 pass 139 Figure 4.1.17: 0.9mm Ti64, 5.5mm Beam Dia., 500W, 30mm/s, 20 pass 140 Figure 4.1.18: 0.9mm Ti64, 8mm Beam Dia., 900W, 40mm/s, 20 pass 140


Stuart P. Edwardson PhD Thesis xi

Figure 4.1.19: Graphite coating condition after 20 passes, 0.9mm Ti64, 5.5mm Beam Dia., 500W, 30mm/s 141

Figure 4.1.20: 0.9mm Ti64, 5.5mm Beam Dia., 500W, 30mm/s, 30 pass, Coating re-spray at 20 passes 141

Figure 4.1.21: 0.9mm Ti64, 8mm Beam Dia., 900W, 40mm/s, 40 pass, Coating re-spray at 20 passes 142

Figure 4.1.22: 2D LF process map for 1.4mm Ti64, 3mm Beam Dia 142 Figure 4.1.23: 2D LF process map for 1.4mm Ti64, 5.5mm Beam Dia. 143 Figure 4.1.24: 2D LF process map for 1.4mm Ti64, 8mm Beam Dia 143 Figure 4.1.25: 1.4mm Ti64, 3mm Beam Dia., 900W, 50mm/s, 20 pass 144 Figure 4.1.26: 1.4mm Ti64, 5.5mm Beam Dia., 900W, 45mm/s, 20 pass 144 Figure 4.1.27: 1.4mm Ti64, 8mm Beam Dia., 900W, 30mm/s, 20 pass 144 Figure 4.1.28: 1.4mm Ti64, 5.5mm Beam Dia., 900W, 45mm/s,

30 pass, Coating re-spray at 20 passes 145 Figure 4.1.29: 1.4mm Ti64, 8mm Beam Dia., 900W, 30mm/s,

30 pass, Coating re-spray at 20 passes 145 Figure 4.1.30: 2D LF process map for 1.6mm Ti64, 3mm Beam Dia. 146 Figure 4.1.31: 2D LF process map for 1.6mm Ti64, 5.5mm Beam Dia. 146 Figure 4.1.32: 2D LF process map for 1.6mm Ti64, 8mm Beam Dia. 146 Figure 4.1.33: 1.6mm Ti64, 3mm Beam Dia., 740W, 40mm/s, 20 pass 147 Figure 4.1.34: 1.6mm Ti64, 5.5mm Beam Dia., 740W, 30mm/s, 20 pass 147 Figure 4.1.35: 1.6mm Ti64, 8mm Beam Dia., 740W, 20mm/s, 20 pass 147 Figure 4.1.36: 1.6mm Ti64, 5.5mm Beam Dia., 740W, 30mm/s,

30 pass, Coating re-spray at 20 passes 148 Figure 4.1.37: 1.6mm Ti64, 8mm Beam Dia., 740W, 20mm/s, 30 pass,

Coating re-spray at 20 passes 149 Figure 4.1.38: 2D LF process map for 2mm Ti64, 5.5mm Beam Dia. 149 Figure 4.1.39: 2D LF process map for 2mm Ti64, 8mm Beam Dia. 150 Figure 4.1.40: 2mm Ti64, 5.5mm Beam Dia., 1200W, 25mm/s, 20 pass 150 Figure 4.1.41: 2mm Ti64, 5.5mm Beam Dia., 1200W, 25mm/s,

Surface condition after 20 passes 151 Figure 4.1.42: 2mm Ti64, 5.5mm Beam Dia., 900W, 30mm/s, 15 passes 151 Figure 4.1.43: 2mm Ti64, 8mm Beam Dia., 1200W, 25mm/s, 15 passes 151 Figure 4.1.44: 2mm Ti64, 5.5mm Beam Dia., 900W, 30mm/s,

Re-spray at pass 15 152 Figure 4.1.45: 2mm Ti64, 5.5mm Beam Dia., 900W, 30mm/s,

Re-spray every 5 passes 152 Figure 4.1.46: 2mm Ti64, 8mm Beam Dia., 1200W, 25mm/s,

Re-spray at pass 15 152 Figure 4.1.47: 2mm Ti64, 8mm Beam Dia., 1200W, 25mm/s,

Re-spray every 5 passes 153 Figure 4.1.48: Single & Double Pass Comparison, 3.2mm Ti64 Sheet 154 Figure 4.1.49: Thermocouple Analysis Single Pass, 3.2mm Ti64 Sheet 155 Figure 4.1.50: Thermocouple Analysis Double Pass, 3.2mm Ti64 Sheet 155 Figure 4.1.51: 2D LF process map for 0.9mm AA1050, 3mm Beam Dia. 156 Figure 4.1.52: 0.9mm AA1050, 3mm Beam Dia., 300W, 35mm/s, 30 passes 158 Figure 4.1.53: 0.9mm AA1050, 3mm Beam Dia., 800W, 85mm/s, 30 passes 158 Figure 4.1.54: 0.9mm AA1050, 3mm Beam Dia., 800W, 85mm/s, 30 passes,

Repeatability test 158 Figure 4.1.55: 2D LF process map for 1.6mm AA6061 O, 3mm Beam Dia. 160


Stuart P. Edwardson PhD Thesis xii

Figure 4.1.56: 2D LF process map for 1.6mm AA6061 T4, 3mm Beam Dia. 160 Figure 4.1.57: 2D LF process map for 1.6mm AA6061 T6, 3mm Beam Dia. 161 Figure 4.1.58: Effect of heat treatment condition, AA6061,

3mm Beam Dia. 55mm/s, 500W, 30s interval, 30 pass 162 Figure 4.1.59: Effect of incident laser power, AA6061 O,

3mm Beam Dia. 55mm/s, 30s interval, 30 pass 163 Figure 4.1.60: Effect of incident laser power, AA6061 T4,

3mm Beam Dia. 55mm/s, 30s interval, 30 pass 164 Figure 4.1.61: Effect of incident laser power, AA6061 T6,

3mm Beam Dia. 55mm/s, 30s interval, 30 pass 164 Figure 4.1.62: Effect of processing speed, AA6061 O,

3mm Beam Dia. 500W, 30s interval, 30 pass 166 Figure 4.1.63: Effect of processing speed, AA6061 T4,

3mm Beam Dia. 500W, 30s interval, 30 pass 166 Figure 4.1.64: Effect of processing speed, AA6061 T6,

3mm Beam Dia. 500W, 30s interval, 30 pass 166 Figure 4.1.65: Effect of inter-pass time delay, AA6061 O,

3mm Beam Dia. 500W, 55mm/s, 30 pass 167 Figure 4.1.66: Effect of inter-pass time delay, AA6061 T4,

3mm Beam Dia. 500W, 55mm/s, 30 pass 167 Figure 4.1.67: Effect of inter-pass time delay, AA6061 T6,

3mm Beam Dia. 500W, 55mm/s, 30 pass 167 Figure 4.1.68: Effect of coating re-spray interval, AA6061 O,

3mm Beam Dia. 500W, 55mm/s, 30 pass, 30s interval 169 Figure 4.1.69: Effect of coating re-spray interval, AA6061 T4,

3mm Beam Dia. 500W, 55mm/s, 30 pass, 30s interval 169 Figure 4.1.70: Effect of coating re-spray interval, AA6061 T4,

3mm Beam Dia. 500W, 55mm/s, 30 pass, 30s interval 170 Figure 4.2.1: Thermocouple Output at various locations, 3mm Beam

Dia. 55mm/s, 760W, 1 pass 171 Figure 4.2.2: Thermocouple Output at various locations, 5.5mm Beam

Dia. 30mm/s, 760W, 1 pass 172 Figure 4.2.3: Thermocouple Output at various locations, 8mm Beam

Dia. 20mm/s, 760W, 1 pass 172 Figure 4.2.4: Thermocouple Output at various locations, 3mm Beam

Dia. 55mm/s, 760W, 6 pass, 60 second intervals 173 Figure 4.2.5: Thermocouple Output at various locations, 3mm Beam

Dia. 55mm/s, 760W, 6 pass, 24 second intervals 173 Figure 4.2.6: Thermocouple Output at various locations, 5.5mm Beam



Dia. 20mm/s, 760W, 3 pass, 24 second intervals 174 Figure 4.2.9: Thermocouple Output, 8mm Beam Dia. 20mm/s, 760W,

10 pass, 40 second intervals 176 Figure 4.2.10: 2D Thermal Images Obtained for the 3mm Beam Diameter

with Laser Power 760W and Scan Velocity 55mm/s 178 Figure 4.2.11: 2D Thermal Images Obtained for the 5.5mm Beam Diameter

with Laser Power 760W and Scan Velocity 30mm/s 179


Stuart P. Edwardson PhD Thesis xiii

Figure 4.2.12: 2D Thermal Images Obtained for the 8mm Beam Diameter with Laser Power 760W and Scan Velocity 20mm/s 180

Figure 4.2.13: Comparison of the Temperature Distributions for the 3mm, 5.5mm and 8mm Diameter Laser Beams 181

Figure 4.2.14: Temperature Distribution for the 3mm Incident Beam 181 Figure 4.2.15: Temperature Distribution of the 5.5mm Incident Beam 182 Figure 4.2.16: Temperature Distribution of the 8mm Incident Beam 182 Figure 4.2.17: Thermocouple Output, 3mm Beam Dia. 55mm/s,

760W, 4 pass, 40 second intervals, no cooling 185 Figure 4.2.18: Thermocouple Output, 3mm Beam Dia. 55mm/s,

760W, 4 pass, 40 second intervals, With cooling 185 Figure 4.2.19: Thermocouple Output, 5.5mm Beam Dia. 30mm/s,

760W, 4 pass, 40 second intervals, no cooling 186 Figure 4.2.20: Thermocouple Output, 5.5mm Beam Dia. 30mm/s,

760W, 4 pass, 40 second intervals, With cooling 186 Figure 4.2.21: Thermocouple Output, 8mm Beam Dia. 20mm/s,

760W, 4 pass, 40 second intervals, no cooling 186 Figure 4.2.22: Thermocouple Output, 8mm Beam Dia. 20mm/s,

760W, 4 pass, 40 second intervals, With cooling 186 Figure 4.2.23: 3mm Beam Dia. 55mm/s, 760W, 30 pass, 40 second

intervals, with and without cooling 187 Figure 4.2.24: 5.5mm Beam Dia. 30mm/s, 760W, 30 pass, 40 second

intervals, with and without cooling 187 Figure 4.2.25: 8mm Beam Dia. 20mm/s, 760W, 30 pass, 40 second

intervals, with and without cooling 188 Figure 4.3.1: Displacement/Time, 3mm Beam Dia. 760W, 55mm/s,

All 6 passes, 60s int. 189 Figure 4.3.2: Displacement/Time, 3mm Beam Dia. 760W, 55mm/s,

pass 1, 60s int. 189 Figure 4.3.3: Displacement/Time, 3mm Beam Dia. 760W, 55mm/s,

pass 6, 60s int. 189 Figure 4.3.4: Displacement/Time, 5.5mm Beam Dia. 760W, 30mm/s,

All 6 passes, 60s int. 190 Figure 4.3.5: Displacement/Time, 5.5mm Beam Dia. 760W, 30mm/s,

pass 1, 60s int. 190 Figure 4.3.6: Displacement/Time, 5.5mm Beam Dia. 760W, 30mm/s,


All 6 passes, 60s int. 190 Figure 4.3.8: Displacement/Time, 8mm Beam Dia. 760W, 20mm/s,


pass 6, 60s int. 191 Figure 4.3.10: Schematic of possible reasons for ‘S’ curve bend angle

development 194 Figure 4.4.1: Strain Gauge Output at 46mm Top Surface, 10mm from 1st edge 195 Figure 4.4.2: Strain Gauge Output at 46mm Top Surface, 30mm from 1st edge 196 Figure 4.4.3: Strain Gauge Output at 46mm Top Surface, 50mm from 1st edge 196 Figure 4.4.4: Strain Gauge Output at 46mm Top Surface, 70mm from 1st edge 196


Stuart P. Edwardson PhD Thesis xiv

Figure 4.4.5: Strain Gauge Output at 46mm Bottom Surface, 10mm from 1st edge 196




Figure 4.4.9: Strain Gauge Output at 10mm Top Surface, 10mm from 1st edge 199 Figure 4.4.10: Strain Gauge Output at 10mm Top Surface, 40mm from 1st edge

(Centreline) 199 Figure 4.4.11: Strain Gauge Output at 10mm Top Surface, 70mm from 1st edge 199 Figure 4.4.12: Strain Gauge Output at 10mm Bottom Surface,

10mm from 1st edge 200 Figure 4.4.13: Strain Gauge Output at 10mm Bottom Surface,

40mm from 1st edge (Centreline) 201 Figure 4.4.14: Strain Gauge Output at 10mm Bottom Surface,

70mm from 1st edge 201 Figure 4.4.15: Visualisation of the strain output close to the scan line

at the start of a pass 202 Figure 4.4.16: Visualisation of the strain output close to the scan line

at the end of a pass 203 Figure 4.4.17: Output from gauges on the top surface at 46mm from the

scan line, longitudinal strain 204 Figure 4.4.18: Output from gauges on the Bottom surface at 46mm

from the scan line, longitudinal strain 204 Figure 4.4.19: Exaggerated view of edge effects 205 Figure 4.4.20: Output from gauge at 10mm from 1st edge on the top

surface at 10mm from the scan line, longitudinal strain 206 Figure 4.4.21: Output from gauge on the centreline on the top

surface at 10mm from the scan line, longitudinal strain 206 Figure 4.4.22: Output from gauge at 70mm from 1st edge on the top

surface at 10mm from the scan line, longitudinal strain 207 Figure 4.4.23: Output from gauge at 10mm from 1st edge on the lower

surface at 10mm from the scan line, longitudinal strain 207 Figure 4.4.24: Output from gauge on the centreline on the lower surface

at 10mm from the scan line, longitudinal strain 207 Figure 4.4.25: Output from gauge at 70mm from 1st edge on the lower

surface at 10mm from the scan line, longitudinal strain 207 Figure 4.4.26: Visualisation of the longitudinal strain output close to the

scan line at the start of a pass 209 Figure 4.4.27: Visualisation of the longitudinal strain output close to the

scan line at the end of a pass 210 Figure 4.5.1: Initial 1200 element FEA model developed 211 Figure 4.5.2: 580 element graded mesh model 212 Figure 4.5.3: Variation in peak upper surface temperature with absorption

coefficient (model output). 214 Figure 4.5.4: Temperature output from the FEA model at 10 and 22mm

from the scan line for a) Upper surface b) Lower surface 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 214


Stuart P. Edwardson PhD Thesis xv

Figure 4.5.5: Thermocouple measurements at 10 and 22mm from the scan line for a) Upper surface b) Lower surface 5.5mm beam dia. 760W, 30mm/s, single pass 215

Figure 4.5.6: Model Output, 3D contour plot of temperature at; a) Mid-pass b) End of pass c) t=4.2s d) t=46.2s 3mm beam dia. 760W, 55mm/s, single pass, A=0.85 216

Figure 4.5.7: Temperature output at various distances from the scan line along the centreline of the plate, Upper Surface 3mm beam dia. 760W, 55mm/s, single pass, A=0.85 216

Figure 4.5.8: Temperature output at various distances from the scan line along the centreline of the plate, Lower Surface

3mm beam dia. 760W, 55mm/s, single pass, A=0.85 216 Figure 4.5.9: Model Output, 3D contour plot of temperature at;

a) Mid-pass b) End of pass c) t=4.5s d) t=28.6s 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 217

Figure 4.5.10: Temperature output at various distances from the scan line along the centreline of the plate, Upper Surface 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 217

Figure 4.5.11: Temperature output at various distances from the scan line along the centreline of the plate, Lower Surface 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 217

Figure 4.5.12: Model Output, 3D contour plot of temperature at; a) Mid-pass b) End of pass c) t=5.4s d) t=34.4s 8mm beam dia. 760W, 20mm/s, single pass, A=0.85 218

Figure 4.5.13: Temperature output at various distances from the scan line along the centreline of the plate, Upper Surface

8mm beam dia. 760W, 20mm/s, single pass, A=0.85 218 Figure 4.5.14: Temperature output at various distances from the

scan line along the centreline of the plate, Lower Surface 8mm beam dia. 760W, 20mm/s, single pass, A=0.85 218

Figure 4.5.15: Temperature output at various distances from the scan line along at Edge 1, Upper Surface 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 220

Figure 4.5.16: Temperature output at various distances from the scan line along at Edge 2, Upper Surface 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 221

Figure 4.5.17: Widening of the HAZ near the edge in mild steel 5.5mm beam dia. 760W, 30mm/s 221

Figure 4.5.18: Final displacement output, magnification factor =30 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 222

Figure 4.5.19: Model Output, 3D contour plot of temperature and displacement at; a)Start of pass b) Mid-pass c) End of pass d) t=49s 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 223

Figure 4.5.20: Displacement/time output, free end of the plate on the centreline 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 224

Figure 4.5.21: Height contour plots of the formed plate, magnification factor =30 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 224

Figure 4.5.22: Displacement/time output, free end of the plate 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 225


Stuart P. Edwardson PhD Thesis xvi

Figure 4.5.23: 3D contour plots of E11 at; a) Start b) Near Start c) Near End of pass d) t=29.5s 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 226

Figure 4.5.24: Transverse strain E11 at ~10mm from scan line near edge 1 Upper surface: a) Model output b) Strain gauge output 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 227

Figure 4.5.25: Transverse strain E11 at ~10mm from scan line at the centre Upper Surface: a) Model output b) Strain gauge output 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 228

Figure 4.5.26: Transverse strain E11 at ~10mm from scan line near edge 2 Upper Surface: a) Model output b) Strain gauge output 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 228

Figure 4.5.27: 3D contour plots of E22 at; a) Start b) Mid Pass c) End of pass d) t=19.5s 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 230

Figure 4.5.28: Longitudinal strain E22, centre of the scan line near edge 1 Upper Surface

5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 231 Figure 4.5.29: Longitudinal strain E22, centre of the scan line, plate centre

Upper Surface 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 231


5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 232 Figure 4.5.31: 3D contour plots of S11 at;

a) Start b) Mid Pass c) End of pass d) t=49.5s 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 233

Figure 4.5.32: Schematic of the stress distribution around the laser beam during laser forming 234

Figure 4.5.33: Transverse Stress S11, centre of the scan line near edge 1 Upper Surface

5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 234 Figure 4.5.34: Transverse Stress S11, centre of the scan line, plate centre



5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 235 Figure 4.5.36: 3D contour plots of S22 at;

a) Start b) Mid Pass c) End of pass d) t=29.5s 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 236


5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 237 Figure 4.5.38: Transverse Stress S22, centre of the scan line, plate centre



5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85 238


Stuart P. Edwardson PhD Thesis xvii

Figure 4.6.1: Iron-Carbon Equilibrium Phase Diagram with some typical microstructures 239

Figure 4.6.2: Microstructure of the ‘as-received’ coupon (x500 magnifications) 241

Figure 4.6.3: 1.5mm Mild Steel, 3mm beam diameter 760W, 55mm/s, 1 pass, Top Middle and Bottom of the HAZ section (x500 magnifications) 242

Figure 4.6.4: 1.5mm Mild Steel, 3mm beam diameter 760W, 55mm/s, 10 passes, Top Middle and Bottom of the HAZ section (x500 magnifications) 242


Figure 4.6.6: 1.5mm Mild Steel, 5.5mm beam diameter 760W, 30mm/s, 1 pass, Top Middle and Bottom of the HAZ section (x500 magnifications) 243

Figure 4.6.7: 1.5mm Mild Steel, 5.5mm beam diameter 760W, 30mm/s, 10 passes, Top Middle and Bottom of the HAZ section (x500 magnifications) 244

Figure 4.6.8: 1.5mm Mild Steel, 5.5mm beam diameter 760W, 30mm/s, 30 passes, Top Middle and Bottom of the HAZ section (x500 magnifications) 244

Figure 4.6.9: 1.5mm Mild Steel, 8mm beam diameter 760W, 20mm/s, 1 pass, Top Middle and Bottom of the HAZ section (x500 magnifications) 245



Figure 4.6.12: Typical microstructure of AA 6061 (x250 optical) 253 Figure 4.6.13: AA 6061 O

‘As Received’ 253

Figure 4.6.14: AA 6061 O After 5 passes 253

Figure 4.6.15: AA 6061 O After 30 passes 254

Figure 4.6.16: AA 6061 T4 ‘As Received’ 254

Figure 4.6.17: AA 6061 T4 After 5 passes 254


Figure 4.6.19: AA 6061 T6 ‘As Received’ 254



Figure 4.6.22: AA 6061 O - a) 0 pass, b) 5 pass c) 30 pass 257


Stuart P. Edwardson PhD Thesis xviii

Figure 4.6.23: AA 6061 T4 - a) 0 pass, b) 5 pass c) 30 pass 257 Figure 4.6.24: AA 6061 T6 - a) 0 pass, b) 5 pass c) 30 pass 257 Figure 4.7.1: Laser forming of 1.5mm mild steel CR4, 3mm beam dia.

760W 10 – 70mm/s process map 260 Figure 4.7.2: Closed loop laser forming of 1.5mm mild steel CR4,

3mm beam dia. 760W, 20° target, attempt 1 261 Figure 4.7.3: Closed loop laser forming of 1.5mm mild steel CR4,

3mm beam dia. 760W, 20° target, attempt 2 261 Figure 4.7.4: Laser forming of 0.9mm AA1050, 3mm beam dia.

300W 10 – 90mm/s process map 262 Figure 4.7.5: Closed loop laser forming of 0.9mm AA1050,

3mm beam dia. 300W, 20° target, attempt 1 263 Figure 4.7.6: Closed loop laser forming of 0.9mm AA1050,

3mm beam dia. 300W, 20° target, attempt 2 264 Figure 4.7.7: Closed loop laser forming of 0.9mm AA1050,

3mm beam dia. 300W, 30° target 264 Figure 4.8.1: Part-cylinder formed from 390x180x5mm mild steel plate 266 Figure 4.8.2: CMM 3D contour plot of part-cylinder geometry formed from

390x180x5mm mild steel plate 266 Figure 4.8.3: Schematic of the LF strategy used to form a part-cylinder along

the Y axis in an 800x400mm sheet. 267 Figure 4.8.4: 800x400x5mm mild steel sheet formed into a

part-cylinder. 267 Figure 4.8.5: Height measurements along the two longer X axis edges of an

800x400x5mm mild steel sheet formed into a part-cylinder. 268 Figure 4.8.6: Height measurements along the two shorter Y axis edges of an

800x400x5mm mild steel sheet formed into a part-cylinder. 268 Figure 4.8.7: Schematic of the LF strategy used to form a part-cylinder

along the longitudinal X axis 800x400x5mm mild steel sheet. 269 Figure 4.8.8: Laser forming a part-cylinder along the longitudinal X axis from

800x400x5mm mild steel sheet. 269 Figure 4.8.9: Height measurements along the two longer X axis edges of an

800x400x5mm mild steel sheet formed into a part-cylinder along the X axis. 270

Figure 4.8.10: Height measurements along the two shorter Y axis edges of an 800x400x5mm mild steel sheet formed into a part-cylinder along the X axis. 270

Figure 4.8.11: Thermocouple measurement locations 271 Figure 4.8.12: Thermocouple output, 1 double pass, 800x400x5mm

mild steel sheet 272 Figure 4.8.13: Thermocouple output, 1 double pass, 800x400x5mm

mild steel sheet 272 Figure 4.9.1: Treating the section as a metallic solid results in a buckling

of the Upper Laminate due to non-TGM parameters and excessive heating 275

Figure 4.9.2: Laser forming the upper laminate alone results in a positive bend, no melting and no obvious damage 275

Figure 4.9.3: Laser Forming of 1.38mm 2/1 glass reinforced polyamide based MLC 275

Figure 4.9.4: Repeatability Test, 1.38mm 2/1 Polyamide based MLC 276


Stuart P. Edwardson PhD Thesis xix

Figure 4.9.5: The Effect of Increasing No. of Layers on the Laser Forming of Polyamide based MLC 276

Figure 4.9.6: The Effect of Increasing No. of Layers on the Laser Forming of Self-Reinforced Polypropylene MLC 277

Figure 4.9.7: The Effect of Increasing No. of Layers on the Laser Forming of Glass-Reinforced Polypropylene MLC 277

Figure 4.9.8: The Effect of Fibre Orientation on the Laser Forming of Glass-Reinforced Polypropylene based MLC 278

Figure 4.9.9: Thermocouple Output for a 0.3mm Al 2024 Foil, Centreline Bottom Surface 279

Figure 4.9.10: 2/1 Polyamide based MLC after 5 passes, 200W, 90mm/s, 2.5mmØ 280

Figure 4.9.11: Upper layer cracked due to non-optimum excessive heating. 281 Figure 4.9.12: De-lamination due to failure in bonding layer. 281 Figure 4.9.13: 200x100mm Part-Cylinder formed from 2/1 polyamide

based MLC 282 Figure 4.9.14: 240x80mm Part-Cylinder formed from 2/1 polypropylene

based MLC 282 Figure 4.9.15: Laser forming 2/1 GLARE type materials, initial feasibility test 284 Figure 4.9.16: Laser forming 2/1 GLARE type materials at various

processing speeds 284 Figure 4.9.17: Laser forming a multiple scan line large radii bend,

2/1 GLARE type material, 240x80mm 285 Figure 4.9.18: Laser forming a multiple scan line large radii bend,

2/1 GLARE type material, 240x80mm (reverse angle) 286 Figure 4.10.1: Initial method to produce the ‘A’ frame strut from

400x200mm mild steel sheet. 287 Figure 4.10.2: Result of initial attempt to produce the ‘A’ frame strut from

400x200mm mild steel sheet. 288 Figure 4.10.3: Method used to produce the ‘A’ frame strut section from

200x100x1.6mm Ti64 sheet. 288 Figure 4.10.4: ‘A’ frame strut section production from 200x100x1.6mm

Ti64 sheet. Forming the sharp bends at the edges first 289 Figure 4.10.5: ‘A’ frame strut section production from 200x100x1.6mm

Ti64 sheet. Forming the gradual large radii bend at the centre to complete the geometry. 290

Figure 4.10.6: Method used to produce the full sized ‘A’ frame strut from 574x175x3.2mm mild steel sheet. 291

Figure 4.10.7: U channel formed first in 574x175x3.2mm mild steel sheet 291 Figure 4.10.8: Large radii bend at the centre added to complete the geometry 291 Figure 5.1.1: Scan Strategy 1, Speed 15mm/s 294 Figure 5.1.2: 3D Contour Plot Strategy 1 294 Figure 5.1.3: Contour Plot Strategy 1 294 Figure 5.1.4: Contour Plot Strategy 1 (end view) 294 Figure 5.1.5: Strategy 2, Speed 20mm/s 295 Figure 5.1.6: 3D Contour plot Strategy 2 295 Figure 5.1.7: Contour Plot Strategy 2 295 Figure 5.1.8: Contour Plot Strategy 2 (end view) 295 Figure 5.1.9: Strategy 3, Speed 30 mm/s 296


Stuart P. Edwardson PhD Thesis xx

Figure 5.1.10: 3D Contour Plot Strategy 3 296 Figure 5.1.11: Contour Plot Strategy 3 296 Figure 5.1.12: Contour Plot Strategy 3 (end view) 296 Figure 5.1.13: Strategy 4, Speed 30mm/s 297 Figure 5.1.14: 3D Contour Plot Strategy 4 297 Figure 5.1.15: Contour Plot Strategy 4 297 Figure 5.1.16: Contour Plot Strategy 4 (end view) 297 Figure 5.1.17: Strategy 5, Speed 20mm/s 298 Figure 5.1.18: 3D Contour Plot Strategy 5 298 Figure 5.1.19: Contour Plot Strategy 5 299 Figure 5.1.20: Contour Plot Strategy 5 (end view) 299 Figure 5.1.21: Strategy 5: square plate 20mm/s 299 Figure 5.1.22: 3D Contour Plot Strategy 5 (square plate) 299 Figure 5.1.23: 3D Contour Plot (side) Strategy 5 (square) 300 Figure 5.1.24: Contour Plot Strategy 5 (square) 300 Figure 5.1.25: 1.6mm Ti64. Strategy 5 300 Figure 5.1.26: 1.6mm Ti64. Strategy 5 300 Figure 5.1.27: 1.6mm Ti64. Strategy 5 contour plot 300 Figure 5.1.28: Strategy 6: 5.5mm beam dia. 40mm/s

400x200x1.5mm Mild Steel 301 Figure 5.1.29: 3D Contour Plot Strategy 6 (pass1) 301 Figure 5.1.30: 3D Contour Plot Strategy 6 (pass1) 301 Figure 5.1.31: Contour Plot Strategy 6 (pass1) 301 Figure 5.1.32: 3D Contour Plot Strategy 6 (pass2) 302 Figure 5.1.33: 3D Contour Plot Strategy 6 (pass2) 302 Figure 5.1.34: Contour Plot Strategy 6 (pass2) 302 Figure 5.1.35: 3D Contour Plot Strategy 6 (pass3) 302 Figure 5.1.36: 3D Contour Plot Strategy 6 (pass3) 302 Figure 5.1.37: Contour Plot Strategy 6 (pass3) 302 Figure 5.1.38: Pillow Shape Strategy: 5.5mm beam dia. 40mm/s 304 Figure 5.1.39: 3D Contour Plot Pillow Shape Strategy 304 Figure 5.1.40: 3D Contour Plot Pillow Shape Strategy 304 Figure 5.1.41: Contour Plot Pillow Shape Strategy 304 Figure 5.1.42: Distorted pillow shape due to over forming. 305 Figure 5.1.43: Twisted Shape Strategy 1 306 Figure 5.1.44: 3D Contour Plot Twisted Shape Strategy 1 306 Figure 5.1.45: 3D Contour Plot Twisted Shape Strategy 1 306 Figure 5.1.46: Contour Plot Twisted Shape Strategy 1 306 Figure 5.1.47: Twisted Shape Strategy 2, 760W, 50mm/s, 5.5mm beam dia. 306 Figure 5.1.48: 3D Contour Plot Twisted Shape Strategy 2

(upper surface, pass1) 307 Figure 5.1.49: 3D Contour Plot (side) Twisted Shape Strategy 2

(upper surface, pass1) 307 Figure 5.1.50: Contour Plot Twisted Shape Strategy 2

(upper surface, pass1) 307 Figure 5.1.51: 3D Contour Plot Twisted Shape Strategy 2

(lower surface, pass1) 308 Figure 5.1.52: Contour Plot Twisted Shape Strategy 2

(lower surface, pass1) 308


Stuart P. Edwardson PhD Thesis xxi

Figure 5.1.53: 3D Contour Plot Twisted Shape Strategy 2 (lower surface, pass2) 308

Figure 5.1.54: Contour Plot Twisted Shape Strategy 2 (lower surface, pass2) 308

Figure 5.1.55: 3D Contour Plot Twisted Shape Strategy 2 (lower surface, pass3) 309

Figure 5.1.56: 3D Contour Plot (side) Twisted Shape Strategy 2 (lower surface, pass3) 309

Figure 5.1.57: Contour Plot Twisted Shape Strategy 2 (lower surface, pass3) 309

Figure 5.1.58: 3D Contour Plot Saddle Shape, 5mm Mild Steel 310 Figure 5.1.59: 3D Contour Plot Saddle Shape, 5mm Mild Steel 310 Figure 5.1.60: Contour Plot Saddle Shape, 5mm Mild Steel 310 Figure 5.1.61: Saddle Shape, 5mm Mild Steel, image of longer edge 310 Figure 5.1.62: Scaled ‘race track’ strategy for the 800x400x5mm

mild steel plates 311 Figure 5.1.63: 800x400x5mm mild steel, height measurements

along shorter edges 312 Figure 5.1.64: 800x400x5mm mild steel, height measurements

along longer edges 312 Figure 5.1.65: 800x400x5mm mild steel plate after processing

with ‘race track’ 312 Figure 5.2.1: The Bezier surface patch 314

Figure 5.2.2: Matlab output showing a Bezier surface patch for a pillow shape 315

Figure 5.2.3: Contour plots of constant gradient values in X and Y for the Bezier interpolated surface of the pillow shape 316

Figure 5.2.4: Matlab data point output of the (overlaid) gradient based scan strategy for the pillow shape and forming results. dz/dy then dz/dx, alternating directions, 5.5mm beam diameter, 760W and 50mm/s 317

Figure 5.2.5: Quiver plot and contour plot of the resultant gradient vector and magnitude in X and Y for the pillow shape 318

Figure 5.2.6: Constant gradient vector direction based scan strategy for the pillow shape 318

Figure 5.2.7: Constant gradient vector direction based scan strategy forming result 318

Figure 5.2.8: Illustration of required forming direction for a given gradient vector 319

Figure 5.2.9: Quiver plot of resultant gradient vector rotated by 90° for pillow shape 319

Figure 5.2.10: Matlab data point output of contour lines of constant height for the pillow shape and forming results. 9 contours, 5.5mm beam diameter, 760W and 50mm/s 320

Figure 5.2.11: Schematic of possible reason why lines of constant height give a usable scan pattern for a surface 321

Figure 5.2.12: Height contour plot of pillow surface with an indication of the required gradient vector magnitude at points along the contour lines 322

Figure 5.2.13: Matlab output showing a Bezier surface patch for a saddle shape, based on rotated and interpolated control point data 323


Stuart P. Edwardson PhD Thesis xxii

Figure 5.2.14: Contour plots of constant gradient values in X and Y for the Bezier interpolated surface of the saddle shape 323

Figure 5.2.15: Quiver plot of resultant gradient vector rotated by 90° for saddle shape 324

Figure 5.2.16: Height contour plot of saddle surface with an indication of the required gradient vector magnitude at points along the contour lines. Blue indicates positive deflection, Red indicates negative deflection 325

Figure 5.2.17: Developable and non-developable surfaces, analogous to the 3D laser forming of continuous surfaces 326

Figure 5.3.1: CNC File Generation by Matlab 331 Figure 5.3.2: Elliptic paraboloid or pillow shape 332

Figure 5.3.3: Hyperbolic paraboloid or saddle shape 332

Figure 5.3.4: Matlab output using an elliptic paraboloid definition for the pillow shape 333

Figure 5.3.5: Matlab output using a hyperbolic paraboloid definition for the saddle shape 333

Figure 5.3.6: Predicted scan strategy and speed distribution for the pillow shape 333

Figure 5.3.7: Laser formed elliptic paraboloid based pillow shape, 5.5mm beam diameter, 760W, 45-55mm/s 334

Figure 5.3.8: Repeatability tests 2 and 3 334

Figure 5.3.9: Standard deviation between each of the repeatability tests 334

Figure 5.3.10: Desired 20mm max deflection pillow shape and error plot between it and the flat unformed sheet 336

Figure 5.3.11: Predicted scan strategy and speed distribution for pass 1 336

Figure 5.3.12: Speed selection based on 2D LF data for a 5.5mm beam diameter and 760W. 50mm/s selected as a minimum speed. All other speeds distributed in the range 50 to 85mm/s 337

Figure 5.3.13: Pass 1 forming result, 5.5mm beam diameter and 760W. Maximum forming ~8mm 337

Figure 5.3.14: Comparison between formed surface after pass 1 and desired shape, ~12mm difference. Error plot gives a prediction for the next pass 337

Figure 5.3.15: Scan strategy prediction for pass 2. Calibration with pass 1 data gives a strain calibration scaling factor for the speed based on the current plate’s forming characteristics 338

Figure 5.3.16: Speed distribution used for pass 2. As there is less required forming the minimum process speed has automatically increased to 67mm/s so as to avoid overshoot. The predicted induced strain has also been adjusted according to the pass 1 data 338

Figure 5.3.17: Pass 2 results, 5.5mm beam diameter and 760W. ~17mm maximum deflection 338

Figure 5.3.18: Comparison between formed surface after pass 2 and desired shape, ~4.5mm difference. Error plot now gives a prediction for the next pass. More forming along the longer edges now is required. 339

Figure 5.3.19: Scan strategy prediction for pass 3. No further calibration is performed after the pass 1 data. The Galil controller can easily reproduce smooth motion based on the complex scan prediction 339


Stuart P. Edwardson PhD Thesis xxiii

Figure 5.3.20: Speed distribution used for pass 3. As only fine adjustments are required a speed range of 73.13 to 85.4mm/s is predicted. 339

Figure 5.3.21: Pass 3 results, 5.5mm beam diameter and 760W. ~21mm maximum deflection (slight overshoot) 340

Figure 5.3.22: Comparison between formed surface after pass 3 and desired shape, +/- ~2.5mm error. Small overshoot has occurred 340

Figure 5.3.23: Predicted scan strategy for pass 4 suggests forming on the reverse side of the plate (red dots) to account for the overshoot. Forming had to be ended here as this capability was not yet included in the system 340

Figure 5.3.24: Image of a laser formed 400x200x1.5mm mild steel plate showing the complex scan patterns realised over the surface. 341

A Study into the 2D and 3D Laser Forming of Metallic Components List of Tables

Stuart P. Edwardson PhD Thesis xxiv

List of Tables

Table 2.3.1: Outline of the 3 main LF mechanisms 14 Table 2.7.1: Degree of application potential for LF in various stages

of a general product life-cycle (not specific to component scale, material or geometry) 57

Table 3.1.1: Typical values of reflectivity of various surfaces

to 10.6µm radiation at normal angles of incidence 80 Table 3.2.1: Material designation according to different

international standards. (Mild Steel CR4) 84 Table 3.2.2: Material composition by weight percentage of Mild Steel CR4 84 Table 3.2.3: Mechanical properties of Mild Steel CR4. 84 Table 3.2.4: Thermal Properties of Mild Steel CR4. 85 Table 3.2.5: Material designation according to different

international standards. (Ti-6Al-4V) 87 Table 3.2.6: Material composition by weight percentage of Ti-6Al-4V. 87 Table 3.2.7: Mechanical properties of Ti-6Al-4V. 87 Table 3.2.8: Thermal Properties of Ti-6Al-4V. 87 Table 3.2.9: Material designation according to

different international standards. (1050-H14) 88 Table 3.2.10: Material composition by weight percentage 88

of Aluminium 1050-H14. Table 3.2.11: Mechanical properties of Aluminium 1050-H14 89 Table 3.2.12: Thermal Properties of Aluminium 1050-H14. 89 Table 3.2.13: Material designation according to

different international standards (AA6061) 90 Table 3.2.14: Material composition by weight percentage of AA6061 90 Table 3.2.15: Mechanical properties of AA6061 in three different tempers 91 Table 3.2.16: Thermal Properties of AA 6061 in three different tempers 91

Page No.

A Study into the 2D and 3D Laser Forming of Metallic Components List of Tables

Stuart P. Edwardson PhD Thesis xxv

Table 3.2.17: Technical Data for the Thermovision® 880 Infrared Detector 95 Table 3.2.18: Lens Specifications for the Infrared Detector 95 Table 3.2.19: AA6061 Samples considered in study 105 Table 4.6.1: 1.5mm Mild Steel, ‘As received’ Vickers hardness 248

Table 4.6.2: 1.5mm Mild Steel,3mm beam diameter 760W,

55mm/s, 1 pass, Vickers hardness 248

Table 4.6.3: 1.5mm Mild Steel,3mm beam diameter 760W,

55mm/s, 10 passes, Vickers hardness 248

Table 4.6.4: 1.5mm Mild Steel, 3mm beam diameter 760W,

55mm/s, 30 passes, Vickers hardness 248 Table 4.6.5: 1.5mm Mild Steel, 5.5mm beam diameter 760W,


Table 4.6.6: 1.5mm Mild Steel, 5.5mm beam diameter 760W,


Table 4.6.7: 1.5mm Mild Steel, 5.5mm beam diameter 760W,








Table 4.6.11: Hardness results for AA 6061 O/T4/T6 256 Table 4.6.12: Irradiated section thickness measurements

for AA 6061 O/T4/T6 258

A Study into the 2D and 3D Laser Forming of Metallic Components List of Symbols

Stuart P. Edwardson PhD Thesis xxvi

List of Symbols S.I. Units

A - Absorption (constant)

b - Breadth of plastic zone

Cp - Specific heat capacity

d1 - Laser beam diameter

D - Ratio of depth of plastic zone to sheet thickness

E - Elastic Modulus

Enn – Strain in the n direction, n=1, 2, 3 (Abaqus Notation)

F - Force

Fn - Fourier number

f - Lens focal length

I - Moment of Inertia

I0 - Intensity at centre of laser beam

kf - Temperature dependent Yield stress

k , λ - Thermal conductivity

1 - Length

lh - Length of heated zone

11 - Length of plastically strained zone

12 - Length of elastically strained zone

M - Bending moment

M2 - Beam Quality Factor

m - Mass

N - In-plane force

P, p1 - Laser power

Q - Dimensionless power

Q1 - Average energy input

R - Radius of curvature

r1 - Laser beam radius

S - Dimensionless velocity

Snn – Stress in the n direction, n=1, 2, 3 (Abaqus Notation)

s0 - Sheet thickness

s1 - Depth of plastic zone

A Study into the 2D and 3D Laser Forming of Metallic Components List of Symbols

Stuart P. Edwardson PhD Thesis xxvii

T - Temperature

Tc - Critical temperature for plastic flow

t - time

u - displacement

v1 - Velocity

W0 - Minimum Beam Waist

W(z) - Beam Waist at distance z

w - Displacement of a plate

x, y, z - Cartesian co-ordinates

Y - Yield Strength

α - Thermal diffusivity

α b - Bend angle

α th - Coefficient of thermal expansion

γxy - Shear Strain in the xy plane

∆T - Time of heating

∆T - Average temperature of heated zone

∆T’ - Temperature increase

ε - Strain

εn - Strain in n direction (n = x, y, z etc.)

ε in - Inherent strain (maximum plastic strain less elastic strain during heating)

ε pm - Maximum plastic strain

κ - Thermal diffusivity

λ - Wavelength

ρ - Mass density

σ - Stress

υ - Poisson’s ratio

Chapter 1 Introduction

Stuart P. Edwardson PhD Thesis - 1 -

Chapter 1

Introduction

The work presented in this thesis is primarily concerned with the process of laser

forming or laser bending of metal sheet material with a high power infra-red

defocused laser beam.

The laser forming process (LF) has become viable for the shaping of metallic

components, as a means of rapid prototyping and of adjusting and aligning. Laser

forming is of significant value to industries that previously relied on expensive

stamping dies and presses for prototype evaluations. Relevant industry sectors

include aerospace, automotive, shipbuilding and microelectronics. In contrast with

conventional forming techniques, this method requires no mechanical contact and

thus promotes the idea of ‘Virtual Tooling’. It also offers many of the advantages of

process flexibility and automation associated with other laser manufacturing

techniques, such as laser cutting and marking 1, 2, 3.

Laser forming can produce metallic, predetermined shapes with minimal

distortion. Investigations are also ongoing into the removal of unwanted distortion

from other manufacturing processes. The process has its origins in flame bending for

ship construction, with the earliest work on LF beginning in the mid-1980s 4, 5. The

process has similarities to the well-established torch flame bending used on large

sheet material in the shipbuilding industry 6, 7, 8, 9, but a great deal more control of the

final product can be achieved. The process employs a defocused laser beam to

induce thermal stresses without melting in the surface of a workpiece in order to

produce controlled distortion. These internal stresses induce plastic strains, bending

or shortening the material, or result in a local elastic plastic buckling of the work

piece depending on the mechanism active 10, 11. The exact mechanisms of the process

are outlined in the next chapter.

It can be argued that the use of a defocused laser to form could be replicated

by cheaper more cost-effective means, e.g. a plasma torch 12. It could also be argued



that laser forming would be a secondary process when considering the cost-

effectiveness of a laser system, in that a system would be purchased for primarily a

cutting or welding operation, proven to be cost effective and competitive, and used

for laser forming as a bonus additional process. However, there are circumstances

where the unique capabilities of laser forming alone can achieve the desired result

such as micro-forming 13.

The range of metals and other materials that can be laser formed is

considerable. As there is only localised heating involved, below the melting

temperature, the bulk properties are not altered and good metallurgical properties are

retained in the irradiated area 14, 15. Materials of particular interest are specialist high

strength alloys 16. These include titanium and aluminium alloys. These materials are

widely used in the aerospace industry where the implementation of laser bending as

a replacement of existing manufacturing processes is under investigation 17, 18, 19 as

well as other industry areas 20.

Presented in this thesis are results of investigations into the 2D and 3D laser

forming of metallic components. 2D laser forming encompasses laser forming

operations that utilise two dimensional out-of-plane bends to produce three

dimensional results e.g. a fold. 3D laser forming encompasses laser forming

operations that can utilise combinations of multi-axis two dimensional out-of plane

bends and in-plane localised shortening to produce three dimensional spatially

formed parts e.g. a dome. Examples of these two types of forming are given in

figures 1.1 and 1.2.

There has been a considerable amount of work completed on 2D laser

forming to date (outlined in the next chapter). However, due to the many variables in

the process and numbers of materials and material types that can be laser formed, a

Forming or bending Lines

2D Forming 3D Forming

Figure 1.1: Examples of 2D forming to produce a 3D part, and 3D forming to produce a spatially formed part.



full understanding of the process is some way off. The work on 2D laser forming

presented in this thesis aims to increase the knowledge and understanding of the

process, in particular the transient thermo-mechanical and asymmetrical effects plus

aspects for closed loop controlled LF. Materials investigated include mild steel,

aluminium AA1050, aluminium AA6061, Ti6Al4V and newly developed Metal

Laminate Composite Materials.



environment, it is then necessary to consider 3D laser forming. Less work has been

completed in this field compared to 2D LF, however the process has been shown to

have a great deal of potential (discussed next chapter). In order to compete directly

with conventional forming techniques though, such as die forming, the process must

be proven to be reliable, repeatable, cost effective and flexible. It is the potential

flexibility of 3D laser forming that offers the greatest benefits, in that a change to a

required part geometry could be implemented easily through the CAD driven process,

this can be compared to the expensive and in-flexible hard tooling requirements of

the die forming process. The work presented in this thesis on 3D laser forming aims

to prove the viability of this technique as a direct manufacturing tool and as a means

of correcting unwanted distortion (perhaps from processes such as chemical etching).

To this aim progress towards repeatable closed loop controlled 3D LF is presented.

The materials investigated were mild steel and Ti6Al4V.

The work presented in this thesis contributed to a larger EPSRC funded

research programme entitled ‘Laser Forming of Aerospace Alloys – A Direct

Fabrication Technique’. The research programme involved a consortium of 3

universities; The University of Liverpool, Heriot Watt University and Cambridge

University; and 2 industrial partners; BAE SYSTEMS and Rolls-Royce plc.

Figure 1.2: Laser formed examples of 2D forming to produce a 3D part, and 3D forming to produce a spatially formed part, both in aluminium.

Chapter 2 Literature Review


Chapter 2

Literature Review 2.1 Introduction This chapter presents some background to laser forming. It reviews the mechanisms

and models for laser forming currently available in the literature, previous

experimental work of note and the potential and current applications of the process.

A synopsis for the current research is also given.

2.2 Process Origins Laser forming originates from the similar process of flame bending or “line heating”

which uses an oxy-acetylene torch as the heat source 8, 21. Flame bending has been

used extensively for profiling and straightening heavy engineering components such

as beams and girders for construction purposes and decking and hull plates for the

shipbuilding industry 6, 7, 9. The diffuse nature of the flame used in line heating

makes the process rely heavily on operator skill. A flame heat source produces a

constant temperature at the surface of the workpiece and it is difficult to establish a

steep thermal gradient (which is often necessary for the process) in thin sections and

materials with a high thermal conductivity. Consequently the operator must spend

much time learning about the heating conditions which will produce the desired

result by trial and error. The heating rates associated with laser beams impinging on

metallic objects are high and steep thermal gradients are easily achieved. In addition

the laser beam can be applied to a very localised region as opposed to the flame.

These advantages along with the potential for automation have led to research into

laser forming.



2.3 Laser Forming Mechanisms The laser forming process is realised by introducing thermal stresses into the

surface of a work piece by heating the surface with a laser beam. These internal

stresses induce plastic strains that result in local elastic-plastic buckling of the

workpiece. The practical application and processing variables of laser forming are

shown in figure 2.3.1

As discussed already, a conventional method - flame bending - has been

known for some years. This technique was traditionally practiced in shipbuilding

where thermal stresses were introduced (often into large sheet panels) by heating the

workpiece using a torch. There are some important differences to laser forming.

Firstly, the laser induces a constant heat flux through the surface, resulting in very

high temperatures at the surface which makes high thermal gradients possible even

in very thin sheets of materials with high thermal conductivity like copper. In

contrast, a torch gives a constant temperature at the surface of the workpiece. The

heat flux depends on the sheet temperature itself and the sheet surface temperature

cannot increase above the flame temperature. This makes high temperature gradients

in materials with a high thermal conductivity impossible. The second important

difference between the flame and laser techniques is the controllability. The spot

diameter and the total energy flux of a laser beam can be controlled from some tenth

of a mm to some cm and from some milliwatts up to some kilowatts, respectively.

Figure 2.3.1: Laser Forming Set-up & Process Variables11



The control of a flame is much more problematic. The energy flux or flame

temperature depends on the oxygen content of the gas mixture which is difficult to

control. In addition, the flame diameter is much larger than that of a laser beam and

also very hard to control11.

Due to the very good control offered by the laser beam, different types of

temperature fields can be generated, yielding different forming mechanisms and

results. These mechanisms are described below.

There are three main mechanisms for laser forming of sheet, tubes and extrusions

(figure 2.3.2), the Temperature Gradient (TGM), Buckling (BM) and Shortening or

Upsetting (UM) mechanisms

As its name suggests, the temperature gradient mechanism depends on

maintaining a high temperature gradient across the sheet thickness. A fourth

mechanism, the point source mechanism, is essentially similar to the temperature

gradient mechanism except that heating takes place at a point rather than over a line

and will not be further considered here.

The buckling mechanism is active if the temperature gradient across the sheet

thickness is small and the diameter of the heated area is large. This mechanism can

result in a bending towards or away from the laser beam. It is essential to control the

direction of bending.

The upsetting (or shortening) mechanism is based on nearly homogeneous

heating of the material. If buckling is avoided due to geometrical reasons or restraint,

a simple shortening (combined with an increase in thickness) of the material results.

This shortening is used in two different ways for forming. Either plane sheets are

Figure 2.3.2: The Laser Forming Mechanisms11



treated with this mechanism resulting in spatially formed parts or extrusions are

treated, giving specially bent extrusions11.

2.3.1 The Temperature Gradient Mechanism (TGM)

The temperature gradient mechanism proceeds in the following steps:

• Heating of the surface and thermal expansion against the cold bulk material

• Development of counter bending

• Further heating and plastic compression of the surface

• Cooling of the surface and thermal contraction

• Development of the bending angle

The conditions for the temperature gradient mechanism are energy

parameters that lead to a steep temperature gradient across the sheet thickness

direction (Figure 2.3.3). The beam is typically of the same order as the sheet

thickness, or slightly less. The path feed rate has to be chosen to be large enough that

a steep temperature gradient can be maintained. It has to be increased if materials are

used which have a high thermal conductivity. The laser path on the sheet surface is

typically a straight line across the whole sheet. This straight line is incident with the

bending edge.

The first step of the temperature gradient mechanism is a heating of the

surface which leads to purely elastic strains. If the heating is stopped in this range,

the elastic strain would recover and the process would be fully reversible; no plastic

strains would remain in the workpiece. Due to the thermal expansion of the surface

layer there is a counter bending of the part, resulting in a bending away from the

laser beam. The amount of the counter bending is very small as only the heated area,

which is approximately the size of the laser spot on the surface, has to generate

Figure 2.3.3: Energy conditions required for the TGM11



forces which produce the counter bending of the whole sheet. The counter bending

effect is detrimental for the development of a plastic bending angle towards the laser

beam. This is so because the counter bending is identical to a relaxation of the

surface stresses at the heated surface. So the thermal expansion leads to lower

surface stresses and therefore the fraction of the thermal strain which is converted

into plastic strain is less than without counter bending.

Further heating leads to a decrease of the flow stress in the heated area and a

further increase of the thermal expansion of the surface layer. At a certain

temperature which depends on the material and the geometry and the amount of

counter bending, the thermal strains reach the elastic strain which can be carried by

the material at the given temperature. A further increase of the temperature results in

a conversion of the thermal expansion into plastic compressive strains. These plastic

compressive strains are accumulated until the heating stops or surface melting

occurs. The heating of a certain point of the surface stops after the laser beam has

passed this point. Then cooling sets in.

In contrast to the heating part of the cycle, where the heat flow is through the

surface due to the coupling of the laser energy, cooling proceeds by heat conduction

in the part. Energy losses by radiation and heat conduction into the environment are

of less importance and can be neglected. Cooling is mainly due to self-quenching

which is also observed in laser surface treatments. The heat flows into the

surrounding sheet metal and gives cooling rates which lead to a cooling of the heated

area within some seconds, typically 10-20 s, which has to be compared with heating

times of about 0.5s. During cooling a shrinkage of the heated material sets in. Due to

the fact that the surface was plastically compressed during heating it is shorter after

cooling to room temperature compared to the non-heated layers of the sheet. Due to

the different length of the surface layer and the lower layer of the sheets bending

angle towards the laser beam develops. The magnitude of the bending angle depends

on the coupled energy, the geometry of the part and the thermal and mechanical

properties of the material. It lies typically between 0.1 degrees and 3 degrees after

one laser pass.

The asymmetry of the process is the reason why the thermal expansion leads

to plastic compression of the sheet and the thermal contraction does not lead to a

plastic tension of the material, this would cancel the plastic compression and

therefore hinder a development of a bending angle (Figure 2.3.4).



During heating there is heat flow through the surface which increases the

temperature in the region below the surface. Thermal expansion leads to plastic

compression of the material. The counter bending is hindered by the cold material

which has a large elastic modulus. During cooling the heat flow is now into the

surrounding material so that the thermal contraction of the material and the thermal

stresses are relieved by the thermal expansion of the surrounding material which is

heated by the heat flow. This material also tends to expand due to heating but it is

also hindered by the surrounding cold material. So compressive stresses are

produced which are superimposed onto the tensile stresses which develop in the

cooling region. In addition the section modulus which determines the amount of the

bending angle towards the laser beam is now lower than during heating as a large

amount of the cross section is heated and hence the elastic modulus is low in this

region.

So far this is an explanation for a two-dimensional representation of the

sample cross section. In addition an explanation for the asymmetry and the effects

due to that can be given from a three-dimensional observation of the sample. During

heating only a small area with a cross-section equal to the square of the beam

diameter is heated and exposed to thermal compressive stresses. These compressive

stresses have to counterbalance the section modulus of the whole sheet. Therefore

Figure 2.3.4: Principle of the Temperature Gradient Mechanism (TGM)11



the counter-bending is very small. In contrast, during cooling the whole edge of the

sheet is cooling simultaneously. This is so because the heating is very fast but

cooling is much slower. Therefore, the moments produced by the cooling of a small

section have to counter-balance the local stiffness of the part only, which is much

smaller than the global stiffness during heating. So there is nearly no cancellation of

the plastic compression and large bending angles can develop. This asymmetry of

the heating and cooling phase is essential for the development of a bending angle for

this mechanism11.

2.3.2 The Buckling Mechanism (BM)

The buckling mechanism operates if the laser beam diameter is large

compared to the sheet thickness and the processing speed is low resulting in a small

temperature gradient across the sheet thickness. These conditions may be realised by

different parameter combinations. One possibility is to irradiate a high alloyed steel

foil (e.g. 100µm in thickness), using a low power laser and a low processing speed.

Another possibility is to use material with a high thermal conductivity like copper.

The buckling mechanism proceeds by the following steps:

• Heating of a large area of the sheet metal and development of compressive

stresses

• Onset of buckling

• Growth of the buckle

• Shifting the buckle through the whole sheet

• Relaxation of the elastic stresses.

The principle is shown in figure 2.3.5. At the beginning the sheet metal is

heated by a laser beam with a beam diameter which is large compared to the sheet

thickness, also the processing speed is low, this results in only a small temperature

gradient across the sheet thickness and hence the thermal expansion of the material

results in compressive stresses in the heated area.

If the heated area is large enough and if there is a small natural deviation

from perfect flatness (which normally exists in real metal sheets) an instability

develops. This instability is similar to the buckle of a sheet metal in the flange during

deep drawing if no blank holder or too low blank holding forces are used. The

direction of the buckling is determined by different factors, these are for example the

pre-bending of the sheet and the relaxation of residual stresses.



In the centre of the buckle the temperature is very high so that the flow stress

is low in this region. Therefore the bending in this region is nearly totally plastic. In

contrast the root of the buckle which is far away from the centre of the beam is

heated to a much lesser extent. So the temperature is low and the flow stress in this

region is high. Therefore the bending of the sheet in this region is fully elastic.

Due to further heating the thermal expansion of the material increases the

height of the buckle. As the laser beam is guided across the surface with the

processing speed the buckle is also shifted along the bending edge. Now the existing

buckle predetermines the direction of the buckling and the remaining part of the

sheet buckles in the same direction as the sheet has done at the beginning. While the

beam is guided across the surface the stiffness of the part is changed. At the

beginning of the buckling process the bending arms were held in the original plane

due to the stiff surrounding material. As an increasing amount of the sheet is formed

by the buckle the forces that hold the bending arms straight, decrease. Therefore the

elastic part of the buckle relaxes and only the plastic part remains in the sheet. This

leads to the development of the bending angle which can be seen after irradiating the

whole bending edge. After finishing the irradiation the elastic strains are fully

relaxed so that an angular section remains. The buckling mechanism results typically

in bending angles between 1 and 15 degrees. This is significantly larger than

observed for the temperature gradient mechanism. This is not a result of a higher

degree of performance but a result of the fact that using the buckling mechanism

Figure 2.3.5: Stages in the Buckling Mechanism (BM)11



more energy can be coupled into the workpiece in one step. Trying the same for the

temperature gradient mechanism either surface melting or buckling would occur,

depending on the spot size of the laser beam. Therefore the energy which can be

coupled into the workpiece is restricted for the temperature gradient mechanism.

Another typical feature of the buckling mechanism is that the bending angle

can be positive (concave bending towards the laser beam), or negative (convex

bending away from the laser beam). Of course it is essential to control the direction

of bending for a reliable process. Otherwise this bending mechanism cannot be used

in manufacturing. For a reliable control of the bending direction the controlling

parameters on the bending direction must be understood. In practice there are two

main important factors which determine the direction of bending. These are the

plastic pre-bending of the sheet and the relaxation of the residual stresses. A plastic

pre-bending often occurs as a result of storing the sheet in a coil or a handling

operation like cutting. These operations give a pre-bending which is first well

defined but the direction of pre-bending may be lost during arbitrary handling of the

parts. Therefore plastic pre-bending will lead with a probability of 50 % each to a

positive and negative bending angle. But it is possible of course to produce a plastic

pre-bending in a well-controlled manner. This may be done by pre-bending the sheet

using the temperature gradient mechanism which always gives positive bending

angles. So a pre-bending may be defined giving also 100 % positive bending angles

after the irradiation with parameters according to the buckling mechanism. This is

the usual way to get well defined bending using the buckling mechanism.

The second important parameter that determines the direction of bending is

the asymmetric relaxation of residual stresses. After rolling there are normally

compressive stresses in the surface and tensile stresses in the core of the sheet.

When working with a relatively high processing speed we get an asymmetric

relaxation of these residual stresses. Due to the temperature gradient produced by

the radiation, the stresses in the heated surface are relaxed first. In this moment the

compressive stresses are removed from the surface resulting in a positive curvature

of the sheet due to the remaining residual stresses. Even if the relaxation of the

residual stresses proceeds and a symmetric state is reached after a certain time, the

asymmetric relaxation at the beginning gives an instability such that the buckling is

always away from the laser beam yielding a bending angle towards the laser beam11.



2.3.3 The Shortening or Upsetting Mechanism (UM)

If the laser beam diameter is of the same order or greater than the sheet

thickness, the path feed rate is low, the thermal conductivity of the material is

relatively high and in addition the geometry of the part does not allow buckling of

the material the Upsetting (Shortening) Mechanism may operate. This is true for

thick sheets and for extrusions and other stiff structures. If these conditions are

fulfilled, the upsetting mechanism proceeds by the following steps:

• Heating of the cross section and thermal expansion.

• Further thermal expansion that exceeds the elastic strain, resulting in a plastic

compression of the cross section.

• Cooling of the material without or with small tensile straining.

These steps are shown in figure 2.3.6. Using a low processing speed the sheet

is heated nearly homogeneously across the thickness direction. Due to the

temperature increase the flow stress decreases in the heated area and the thermal

strains approach the elastic strains at the yield stress. Further heating leads to plastic

compression of the heated materials it is hindered in free expansion by the

surrounding material. Therefore a large amount of the thermal expansion is

converted into a plastic compression. Due to the low temperature gradient there is

also a very small gradient in the plastic strain across the thickness direction. During

cooling the material contracts.

Figure 2.3.6: The Upsetting (Shortening) Mechanism (UM)11

Thermal and Plastic Strain Profiles



The plastic compressive strain remains in the sheet for the same reason which

hinders plastic straining of the compressed material during cooling in the case of the

temperature gradient mechanism: The material was heated along a line. During

heating the expansion is only local and is hindered strongly by the surrounding

material. So the thermal expansion is converted into a plastic compression. During

cooling the cooling is active along the whole line which was heated. Therefore, the

contraction is hindered less than the thermal expansion. Therefore nearly no plastic

straining of the material occurs. The compressive strain remains in the sheet. Of

course, due to the constancy of the volume there is an increase of the sheet thickness

in the compressed area.

This mechanism maybe used in different ways for a wide range of forming

results. Plane sheets may be treated according the upsetting mechanism along radial

paths so that this results in a spatially formed part. The mechanism can also be used

for shortening of small frames. This is useful for aligning operations in micro parts

production. The third application of this mechanism is the forming of extrusions and

pipes, in that the sections can be made to bend out of plane by the careful selection

of the sequence of irradiations11.

An outline of all of the mechanisms discussed is given in table 2.3.1

Mechanism Procedure Forming

efficiency Results

Temperature

Gradient

(TGM)

Spot diameter ~ thickness

Higher traverse speeds

Applicable to thin sections

~1-2° bending

per pass

High control

Low efficiency

Buckling

(BM)

Spot diameter > thickness

Lower traverse speeds

Applicable to thin sections

~15° bending

per pass

High efficiency

Reduced control

Shortening

(Upsetting)

Spot diameter ~ thickness

Applicable to stiff

geometries (can’t buckle)

Microns of

shrinkage per

pass

Shortening

Thickening

Table 2.3.1: Outline of the 3 main LF mechanisms22



2.4 Analytical Models

There are a number of analytical models available in the literature that have been

developed to describe and expand on the three main LF mechanisms which were

outlined in the previous section. The key results and concepts are detailed in the

following sections.

2.4.1 Two Layer models for the TGM

A number of models have been proposed for the TGM 10, 23-28. In particular

Vollerstsen’s two layer model 24 has been widely quoted and a number of

comparative studies performed 24, 29.

A simple beam model was proposed and an energy approach to the

temperature field was assumed. The nomenclature used and principle of the model is

given in figure 2.4.1. The bend angle can be defined by the geometry and the

difference in the strains between the upper (ε1) and lower (ε2) layers.

( )22 0

21

slb εεα −

= (2.4.1)

Figure 2.4.1: Forces and moments acting in the two layer model 24



Using beam theory, the strain within each layer may be calculated. Beam theory

states:

zRE

IM σ

== (2.4.2)

Where:

( )∫= 2

0

2h

dzzzBI (2.4.3)

R is the radius of curvature, M is the bending moment, I is the moment of inertia, E

is the Elastic Modulus and σ is the stress in the beam at z.

Considering the geometry illustrated in figure 2.4.1 the strain in the upper layer

(assuming it is compressed) is given by:

( ) TzEIM

AEF

th∆+−= αε 11

1

111 (2.4.4)

An important assumption made here is that all the thermal expansion of the upper

layer given by the is converted into plastic compression. In reality this is not the case

as some energy is used to elastically strain the material up to its temperature

dependent yield point. However an appreciable amount of elastic straining does not

occur because the free thermal expansion is greatly hindered by the cold and rigid

surrounding material. In addition, the yield stress of the heated zone is reduced to

almost zero during heating since it is temperature dependent. It may be acceptable

under these conditions to omit these parameters and assume that all the thermal

expansion is converted into plastic compression. During cooling as the heat flows

into the surrounding regions there may be a tensile plastic restraining of the

previously compressed zone.

After the plastic compression development the strain of the upper layer is given by:

( )'

11

1

111 Tz

EIM

AEF

th∆−−= αε (2.4.5)

∆T’ is the maximum temperature difference between the upper and lower layers of

the sheet. The strain of the lower layer is given by:

( ) 22

2

221 z

EIM

AEF

−=ε (2.4.6)



Here I2 is the second moment of area. From the previously described beam theory

the fraction of the moment M and the product El yield the local curvature. For large

curvatures it is assumed that:

( ) ( ) ( ) lEIM

EIM

EIM b

22

2

1

1 α=== (2.4.7)

The bending angle αb can then be found by combining the equations for the upper

and the lower layers (eq 2.4.5 and 2.4.6) along with equation 2.4.1:

∆−+−−−= '1

11

2

220 224 T

lz

AEF

lz

AEF

sl

thbb

b ααα

α (2.4.8)

Where the force F is given by:

lsEI

sMF b

002α

== (2.4.9)

And substituting z1-z2 = s0/2 into 2.4.8 and 2.5.9 yields:

( )( ) 0

'

22110

2211 44s

TlAEAEs

AEAEEI thb

bb

∆++

+=

αα

αα (2.4.10)

The cross sections of the beams are described by:

11 bsA = (2.4.11)

And

( )102 ssbA −= (2.4.12)

The section moment is given by:

12

30bsI = (2.4.13)

Assuming E1 = E2 the equation for the bend angle is then given by:

( )30

101'12

ssslsTth

b−∆= αα (2.4.14)

To calculate the bend angle with this formula requires knowledge of the length of the

heated zone l, the depth of heating s1, and the temperature rise of the upper layer ∆T’.

This requires the co-ordinates to be found as a function of the temperature. i.e. in the

form of:

l=f(T) (2.4.15)



For laser processing this is not possible analytically, since the expressions for

temperature contain transcendental functions or Bessel functions which cannot be

suitably inverted 30. Vollertsen adopted an energy approach to the solution of the

temperature field instead. In his energy approach all three factors l, s1 and ∆T’ are

calculated simultaneously 24 assuming the beam diameter is the same as the sheet

thickness. This factor is represented by:

( )101' sslsT −∆=ξ (2.4.16)

This can be done because the parameters l and s1 determine the extent of the heated

zone which is governed by the thermal conductivity. The temperature increase ∆T’ is

controlled by the heated area, the heat capacity and the energy input from the laser

beam. This approach was adopted because, as the thermal conductivity increases, the

extent of the heated area increases also but the average temperature increase is

lowered. From this it was assumed that the thermal expansion remains constant.

Using these assumptions the energy input Q1 is given by the time of heating, t, the

laser power, p1 and the absorption, A:

AtpQ 11 5.0 ∆= (2.4.17)

0.5 is used as only one half of the heated area is used for the calculations. The

heating time ∆t is given by the fraction of the laser spot size and the processing

velocity:

1

1

vdt =∆ (2.4.18)

The average temperature increase of the upper layer is given by the fraction of the

energy input, Q, and the heat capacity, Cp. The mass of the heated area is determined

from the volume of the heated zone and the density, ρ:

ρ11sldm = (2.4.19)

Combining equations 2.4.16 – 2.4.19 results in:

ρξ

1

101

2)(

vCssAp

p

−= (2.4.20)

Introducing this equation into equation (2.4.14) gives an expression for the bend

angle in terms of known parameters only:



201

1 13sv

ApC p

thb ρ

αα = (2.4.21)

In this work by Vollertsen 24 experimental data from other authors was presented and

compared with the analytical results, for the results presented there was a reasonable

agreement. Although substantial improvement in the agreement between this model

and experimental work was achieved (compared to previous analytical models for

the flame forming process) some of the basic concepts were still omitted. The model

assumed that all of the energy was used for plastic deformation and this ignored the

energy used for the elastic straining.

In Yau’s model 28 the two layer model approach was extended to include the

counter-bending effect in order to account for some of the purely elastic straining.

This modification resulted in two equations, one for the counter-bending angle and

one for the bend angle at the end of the cooling cycle. The final equation for the

bending angle (positive bend angle less counter-bend angle) including the

temperature field equation in Yau’s model is:

EY

sl

svAp

C p

thb

0201

1 362713 −

=

ρα

α (2.4.22)

Comparing equation 2.4.21 with equation 2.4.22, Yau’s solution (equation 2.4.22)

includes some material and geometrical parameters which reduce the calculated bend

angle compared to Vollertsen’s solution. Y is the Yield Strength and E the Young’s

Modulus of the material to be formed. Both solutions were implemented and they

differ only slightly for a single pass29. This is because under the conditions of the

temperature gradient mechanism the counter-bending angle is very small and

combined with the simplifying assumptions of the model the difference in the

models is less than expected originally. A comparison between the two models and

verification with experimental data (presented in a later chapter) can be seen in

figure 2.4.2. It can be seen in this figure that over an increasing number of passes

both models over predict the bend angle. There a number of possible reasons for this,

firstly no account of the effect of the thermal conductivity or beam diameter was

included directly in either of these models and this implies that simplifications about

the temperature field were made, additionally no account is taken of the temperature

dependent properties such as heat capacity and the coefficient of thermal expansion,



assumed constant. Another factor in these models is that they predict a constant bend

angle increase with increasing numbers of passes, it has been seen in a number of

published studies that the bend angle rate falls off with increasing numbers of passes 29, to be discussed later, these equations do not take into account factors such as

coating degradation (absorption dependent) and section thickening29.

2.4.2 The Residual Stress Model for the TGM

Vollertsen extended the work on the temperature gradient mechanism in another

model to include the effects of a realistic temperature field analytically, 25 and a

more realistic strain distribution was included. This model used the residual stress

approach often used in welding analysis. The strains in the y & z directions were the

only strains considered in the analysis, i.e. an infinitesimal strip in the direction in

which the laser beam is scanning (x) is considered (figure 2.2.3).

Figure 2.4.2: Comparison of solutions for the two layer models

Figure 2.4.3: Layout for the residual stress model



Initially it is assumed that that there is a boundary temperature or isotherm TB above

which the thermal expansion that is hindered by the surrounding material leads to

plastic compression.

An elliptical strain distribution was assumed and was given by:

221

1

)( zss

z in −=ε

ε (2.4.23)

εin is the inherent strain, this is the maximum plastic strain due to thermal expansion

less the purely elastic strain during heating, such as:

)()(

TETk

T fthin −∆=αε (2.4.24)

Plastic strain occurs if the strain due to the thermal expansion exceeds the purely

elastic strain. The elastic strain is governed by temperature dependent properties, in

that the flow or yield stress and Young’s modulus fall as the temperature increases,

thus making it easier to produce a plastic compression and hence bend a material.

s1 represents the depth of the plastic zone. If the plastic zone is less than the sheet

thickness, s0, then integration of the local strains results in the local bending moment,

the plate is said to bend about this depth s1. The bending moment can be determined

from beam theory, equation 2.4.2, substituting for I:

∫= Bdzzz

M B2σ

(2.4.25)

Substituting for σ and B yields:

dzzszdxEMs

B ∫

−= 1

00

2)(ε (2.4.26)

Substituting eq. 2.4.23 in 2.4.26 and integrating yields:

−=

38

21

10sssdxEM inB

πε (2.4.27)

From geometrical conditions it is known that the bending angle is given by the

fraction of twice the length of the bent zone and the curvature. From the previously

described beam theory (eq. 2.4.2) the fraction of the transverse bending moment

and the Elastic modulus times the moment of inertia yields the inverse of the

curvature, such that:



Rl

b2

=α (2.4.28)

and

EIM

RB=

1 (2.4.29)

where:

12

30dxsI = (2.4.30)

Substituting these equations into eq. 2.4.27 and rearranging for αb yields:

( )1030

1 83 ssslsin

b −= πεα (2.4.31)

To calculate the bend angles with these formulas requires knowledge of the depth of

the plastic zone, s0 and the length of the plastic zone, l. These may be obtained from

the proposed temperature field calculation. The solution to the temperature field

however, was an approximation of the Fourier equation of three dimensional heat

conduction for a finite area source. As the length and depth of the plastic zone were

required for the solution, the co-ordinates were required as a function of the

temperature. An approximate solution in the range relevant to laser bending was

used as it was not possible to invert this form of the heat equation in a suitable

fashion analytically to give the depth co-ordinate. Details of this approximation can

be found in the reference25.

The depth of the plastic zone was given by:

22

34ln 32

1

atPP

NTs f

ff

c

−= − (2.4.32)

The length of the heated zone was given by:

−= − 23

34ln

214 f

f

c

ff P

NT

PatPl (2.4.33)



This model showed the importance of the thermal conductivity on the process.

A slight change in the thermal conductivity changes the thermal expansion and the

position of the elastic-plastic interface, as the average temperature in the irradiated

zone is sensitive to slight changes in the thermal conductivity. Consequently it is

possible that the bend angle itself is sensitive to small changes in the thermal

conductivity. This is contrary to what was reported in the two 1ayer model. In

addition with this model the contribution of the thermal strain to the plastic bending

was found by subtracting the fraction of the yield stress and the elastic modulus from

the thermal expansion (equation 2.4.24). As mentioned both the yield stress and the

elastic modulus are temperature dependent, this required the function which related

those parameters to temperature to be known in order to calculate this contribution

accurately.

Provided with accurate information about the temperature dependent

mechanical properties the model can predict the bend angles with reasonable

accuracy for an analytical route, comparison with experimental data showed this 25.

Mucha et al 26 also modelled the TGM and has provided bend angle

equations for rectangular, triangular, elliptical and circular shaped plastic zones.

These shapes depend on the materials thermal properties and the laser processing

parameters used. Again the y & z directions (assuming the same co-ordinate system

as Vollertsen, figure 2.4.3) are the only directions considered relevant for the

analysis. The formulas for the bend angles for the different shaped plastic zones were

found to be:

Rectangular:

( ) 30

10116s

ssTlsthb −∆= αα (2.4.34)

Triangular:

( ) 30

101123s

ssTlsthb −∆=αα (2.4.35)

Elliptical:

( ) 30

101 2183s

ssTlsthb −∆= παα (2.4.36)



Circular:

( ) 30

102

1183s

ssTsthb −∆= παα (2.4.37)

In this work 26 the temperature was then calculated from the solution for a moving

point source, taken from Duley 31.

Then the bending angle was found for the case of semi - circular isotherms by

introducing dimensionless variables for laser power and traverse velocity into the

temperature field equation.

Dimensionless power:

( )ThkApQ∆

=π2

1 (2.4.38)

Dimensionless velocity:

κ21hvSvl = (2.4.39)

Where κ is the thermal diffusivity, k is the thermal conductivity, Ap1 is the absorbed

laser power and h = s0.

Also using: D = s1/s0 and Rdim = r/s0

These enabled the calculation of the maximum depth of an isotherm, and hence the

final bend angle, which was given by:

( )DTDthb 832 −∆= παα (2.4.40)

The usefulness of this model is found in the trend it presents between the

dimensionless laser power and traverse velocity and the resulting bend angle (figure

2.4.4). This assists with determining the critical conditions which give rise to the

temperature gradient mechanism (TGM). However this analytical route in common

with the previously described models calculates the bend angle at the end of the

process and does not describe the transient stages. Knowledge of the transient stages

is useful for successful process control 29.



In analytical work by Magee 29 it was agued that the above models for the

TGM, although they have advanced the understanding of the process on a

rudimentary level, are incomplete in terms of practical laser forming due to gross

simplifications. The mechanics of the process are defined in terms of the engineering

theory of bending and specifically beam theory. The transverse bending moment has

been used to find the curvature and the temperature field was approximated using an

energy approach, or by an approximation of the solution for the temperature field

from a static laser beam impinging on a thin sheet. Future forming operations will be

concerned with forming an initially flat sheet into a final geometry which is three

dimensional, therefore a model which analyses transverse bending moments only,

and predicts only and angular deformation is of limited use for practical forming. It

was shown in this work that there are other forces and moments acting in laser

forming apart from the transverse bending moments. This indicates that there should

in theory be two bend angles in laser forming under certain conditions, the angle

transverse to the direction of scanning and the angle parallel to the direction of

scanning. 2.4.3 The Buckling Mechanism As described previously the requirements for the initiation of the BM on sheet metal

are that the laser beam diameter on the surface of the sheet is approximately an order

of magnitude greater than the sheet thickness and that the material has a suitable

thermal conductivity so that the laser processing parameters employed do not result

Figure 2.4.4: Critical operating region for the TGM26



in a temperature gradient in the depth direction of the sample. Using the large beam

diameter results in a large amount of thermo-elastic strain which initiates the growth

of an elastic-plastic buckle. The strain near the centre of the laser beam is plastic and

the strain away from the centre of the beam is considered elastic in Vollertsen’s

model 32. The elastic strain is released when the laser beam traverses the exiting edge

of the sample and the plastic strain results in a curvature and a bend angle. The

model is derived from the geometrical conditions (figure 2.4.5).

The bend angle is given by:

2

2

1

1

2 rl

rlb ==

α (2.4.41)

The radius r2 in region 2 (elastic) is given by elastic bending theory (eq. 2.4.2).

2

30

12rEbsM el = (2.4.42)

For the plastic region 1, the moment is given by:

201)(

41 bsTkM fpl = (2.4.43)

At the elastic plastic interface Mel = Mpl, such that r2 can be now given by:

( )1

02 3 Tk

Esrf

= (2.4.44)

l2 can be calculated from the new geometry (figure 2.4.5):

Lll ∆+= 2.02 (2.4.45)

Figure 2.4.5: Model Geometry for the Buckling Mechanism32



where:

=

2sin22.0

brl α (2.4.46)

and

hth lTl ∆=∆ α (2.4.47)

Where ∆T is the average temperature of the heated zone of dimensions lh, ∆x and s0. ∆T is calculated from the absorbed laser power Ap1 the processing velocity v1 and the heat capacity and density, ρcp of the material.

10

1

2 vscApTpρ

=∆ (2.4.48)

l2 may be calculated from:

10

122 22sin

vscApfrl

p

thb

ραα ′

+

= (2.4.49)

ƒ’ is the fraction of the thermal expansion that leads to an expansion of region 2. Using a value of 0.5 results in:

120

11

4)(3

2sin

2 vcEsTkAp

p

fthbb

ρααα

+

= (2.4.50)

Using the sine series expansion, the last expression for the bend angle was simplified. The final equation for the bending angle was:

31

201

11 1)(36

=

svAp

EcTk

p

fthb ρ

αα (2.4.51)

Of note here is the much lower dependency of the bend angle on the temperature

gradient which is consistent with the buckling mechanism theory described earlier.

2.4.4 The Shortening Mechanism

Kraus 33 has modelled box section laser bending. Using the upsetting mechanism box

sections or extrusions can be made to bend out of plane by careful selection of the

sequence of irradiations. A similar approach has been used to Vollertsen’s models

where a geometry / strain relationship is drawn between the processing parameters



and the bending angle. The final equation describing the bend angle in this case was

found to be:

( )( )( )

−

−=

1

1120011

1

221

TEdTk

ssdcvbAp

bf

p

thb ρ

αα (2.4.52)

The model assumed that three of the four sides of the box section were heated

simultaneously to initiate the bending. In reality the sides are usually irradiated

sequentially. However for the purposes of an analytical model this effect would be

very difficult to include. Numerical studies into the sequence of irradiations in

extrusion bending have also been carried out by Kraus using finite element methods.

2.5 Numerical Models

Given the complexity of analytically modelling forming processes such as laser

bending where the workpiece temperatures, dimensions and properties are changing

both in time and space and which depend on many variables (figure 2.3.1), the

numerical approach is often more beneficial for modelling these situations29. The

improvement in computational efficiency in recent years has made such large scale

numerical studies more viable. Numerical models have been available for a number

of years for the flame forming process 8, 34, 35, 36, 37. However it is only in recent times

that emphasis has been placed on laser bending. With the release of more user

friendly numerical modelling software packages such as ALGOR and ABAQUS

coupled with faster computers, the use of numerical models as a research tool for

both academic and industrial sectors is becoming more prevalent, indeed one such

model is included in this thesis. The ability to investigate a complex process in a

non-destructive manor in any situation is extremely useful, however as with any

model assumptions are made and the quality of the output data is only as good as the

quality of the input data.

A number of the numerical models for each mechanism available in the literature are

outlined in the next sections.



2.5.1 Temperature Gradient Mechanism

Vollertsen developed a finite difference model 38 for a two dimensional (2D) analysis

of the process. The temperature dependent material parameters were included in the

model by taking values at particular temperatures of interest and linearly

interpolating between them, and then those functions were used to relate the

temperature to the material properties. A rectangular shaped source was used to

represent the laser beam and the resultant two dimensional (2D) temperature field

was used to calculate the thermal expansion, strains and stresses in the elements.

Then, accounting for the stiffness of the whole sheet, a calculation was made to

assess in which elements the stress exceeded the temperature dependent yield stress.

Then the elastic strain in those elements was reduced by the amount that exceeded

the yield stress.

The amount of strain that exceeded the maximum elastic strain at the yield stress was

converted into a plastic strain. A loop was initiated which continued with this

calculation until there was equilibrium of forces and moments. After the thermal

field had finished being computed, the bending angle was calculated from the length

of the upper and lower layers of an element in conjunction with the sheet thickness.

This model provided a very fast means of calculating the effects of various process

parameters, but the simple boundary conditions that limited this approach led to the

modelling with the finite element method (FEM) 39.

In Alberti et al model 40 of the TGM the finite element method was used first

to evaluate the temperature field and then the results of this analysis were input into

a mechanical analysis. Illustrations for the temperature field and the deformation

were provided at various stages of the process. Emphasis was placed on the

importance of the temperature dependence of the yield stress of the material. A

constant decay law was assumed for the relationship between increasing temperature

and decreasing yield stress. Steel plates were considered in the analysis. Another

numerical simulation by the same author looked at the combined process of thermal

and mechanical bending 41. As this is a process and not solely laser bending no

further details are given here.

Hsiao et al have used the commercial package ABAQUS to model the

process 42. They used the modelling in their work to emphasise the importance of the



specimen size. Their results state that the angular distortion obtained on a short

specimen is much smaller than for a long specimen. This agrees with experimental

results by Vollertsen.23 They also studied the effect of the fraction of the laser power

and the square root of the velocity times the plate thickness as a parameter:

10

1

vsp

(2.5.1)

In conjunction with this study Firth et al have used the code TOPAZ3D/NIKE3D for

analysis. Results from this study were compared with experimental results and it was

reported that the model predicted the trends correctly, but the absolute angles

predicted were considerably smaller (about a factor of 3) in some cases.

2.5.2 The Buckling Mechanism

In work by Holzer et al 43 the buckling mechanism (BM) was modelled using the

commercial finite element package ABAQUS. It was assumed that the sheet was flat

and free of residual stresses. The elements used in ABAQUS for the heat analysis

were DC3D8 (8 -node 3D cubic heat diffusion elements) and the stress analysis used

C3D8 elements. A user defined FORTRAN function was used to model the heat

input from a non-uniform heat flux. The intensity at the centre of the TEM00

gaussian beam was given by:

21

10

2rApIπ

= (2.5.2)

Where I0 is the laser beam intensity at the centre of a TEM beam, Ap1 is the

absorbed power and r1 is the laser beam radius.

The intensity at a distance r from the centre of the laser beam was given by:

1

2

0)( rr

eIrI−

= (2.5.3)

Figure 2.5.1 illustrates the development of the bending angle from this analysis.43 It

is shown at times 0.88s, 126s, and at the end of the process. In figures 2.5.2 and 2.5.3 43 the distributions of the upper and lower surface temperatures, and the elastic and

plastic strains are shown. As can be seen in the case where the sheet bends in a



convex direction away from the laser beam, the plastic strain at the non irradiated

side of the sheet is greater.

Figure 2.5.1: Development of the bending angle during Buckling Mechanism43

Figure 2.5.2: Distribution of the upper and lower surface temperatures and displacements during the Buckling Mechanism43

Figure 2.5.3: Distribution of the upper and lower surface strains during the Buckling Mechanism43



2.5.3 The Shortening Mechanism

Kraus has carried out a finite element analysis into extrusion forming.33 Important

information about the temporal development of the process resulted from this work

which could not be determined experimentally. For example during the cooling

phase a contraction in the irradiated zone takes place, and tensile stresses build up if

the thermal contraction is hindered by the surrounding material and the workpiece

stiffness. These stresses can reach the yield stress depending on the process

parameters employed and a plastic restraining may occur (see figure 2.5.4).

This effect is particularly noticeable in extrusion bending where the moment

of inertia of the workpiece is high. From this analysis Kraus found that there is an

upper limit to the plastic strain which should be induced in order to minimise plastic

restraining. The sequence of irradiations was also optimised using the finite element

method (FEM).

Figure 2.5.4: Plastic restraining in extrusion bending33



2.5.4 Further Numerical Modelling

A number of other numerical studies have been conducted of note that don’t

necessarily fit into the above categories 44 - 55. With improvements to computational

time and reliability researchers are increasingly regarding numerical models as

essential (if not the only) research tools. This is reflected in the increase in

publications of numerical based research in recent years. A summary of the results

and conclusions of a number of papers of note is given here.

In 1999 Yu et al 44 at MIT published numerical based research of laser

forming. Presented was an Abaqus based finite element model for thermo-

mechanical analysis of the LF process. A rezoning or re-meshing technique

(redrawing the fine mesh around the beam as it moves) was employed to greatly

reduce the simulation time yet preserve the required accuracy. A comparison of the

numerical results and experimental data on 2.53cm thick mild steel using 2.6 kW of

CO2 laser power, obtained from Penn State University, showed the effectiveness of

the model. However the observed errors between the model and experimental data

were attributed to the inaccurate estimation of the heat absorption rate (coupling rate)

and the heat convection and radiation coefficients. It was concluded that a more

accurate estimation of these parameters is essential for FEA modelling.

Li & Yao 45 at Columbia University in 1999 presented numerical based work

on the effects of strain rate in laser forming. An FEA model was created in Abaqus

of an 80 x 40 x 0.89mm mild steel coupon laser formed using a CO2 laser source

(Gaussian distribution), only half the coupon was modelled (symmetry assumed). It

was found in this study that with strain rate increase, the thermal-induced distortion

decreases and the bend angle reduces. The bend angle decreases by about 30% for

nearly doubled strain rate under the conditions used. Residual stress in the Y

direction (transverse) increases moderately with strain rate, with a doubled strain rate

residual stress increases by about 15%. From coupled experimental work it was

found that with a strain rate increase the hardness of the formed sample decreases

due to the reduced work hardening. This numerical/experimental approach was

continued at Columbia with work on laser forming with constant line energy 46 and

analysis and prediction of edge effects in laser bending 47.

In 2000 Li & Yao 50 presented numerical (FEA) work on the use of laser

forming to bend tubes, a development on extrusion bending 33. The mechanism was



found to be a combination of thickening (shortening mechanism) of the laser scanned

region due to thermally induced axial compressive stress, and a slightly outward

displacement of the region caused by a component of the thermally induced

circumferential stress. As a result bending is primarily achieved through the

thickening of the scanned region instead of the thinning of the un-scanned region.

The absence, when compared to conventional tube bending, of appreciable wall

thinning is one of the major advantages of laser bent tubes. It was concluded from

this study that the bending efficiency increases with the maximum scanning angle

(distance scanned around the tube) up to a critical point. In addition the asymmetry

of the LF process can be reduced by varying the scanning speed or employing a two-

segment scanning scheme.

In 2001 Cosenza et al 52 presented an explicit fully coupled thermo-

mechanical FE analysis of the LF process, again using Abaqus. This study proposed

a new FE modelling approach utilizing a dynamic explicit algorithm as opposed to

the traditional implicit models. This permits the reduction of CPU times because of

the linearity and the independence of the final set of equations. The FEA model of

the LF of 140 x 20 x 3mm Fe360 sheet using a 6kW CO2 laser source showed

reasonable agreement with experimentally found data.

In 2002 Lee et al 55 in Taiwan published a study into the pulsed LF of thin

sheets (20x10x1mm 304 Stainless Steel) using a single (or multiple) CO2 laser pulse

of an elliptical beam the width of the sample. This has applications in micro laser

forming (discussed later). An Abaqus FE model was developed to simulate this

unique set-up, and a good agreement was found with experimental data. The

conclusions drawn from this study were that the bend angle increases with laser

power and the laser radiation time. The bending angle decreases with the thickness

of the specimen, provided the peak temperature of the specimen is below the melting

point. It was found that if there is a high temperature gradient between the upper and

lower surfaces a positive bend angle is produced, for a low thermal gradient a

negative bend angle is produced. Finally it was concluded that the mechanisms of

pulsed laser forming are dependent upon a number of operation parameters, the main

influences are the laser power, the heating time (pulse length), the clamping

arrangement, the thermal properties and the residual stress state of the specimen.



2.6 Previous Experimental Work The following sections summarise important experimental work that has been carried

out to date in the field of laser forming. Included is early fundamental research and

more recent developments in 2D and 3D, macro and micro laser forming of

numerous materials.

2.6.1 Fundamental Investigations In 1985 Namba 56 published one of the first experimental papers on laser forming.

The materials investigated in this work included Ti, Al, AISI 304 stainless steel and

carbon steel. The materials were irradiated with a 1.5kW CO laser using a defocused

beam with traverse speeds in the range of 5 - 15m/min. Namba claimed the

deformation is caused by the steep thermal gradient which results in thermal

expansion, thermal stress and plastic deformation. The following parameters were

described as affecting the bend angle 56

1) Incident laser beam power

2) Laser beam diameter

3) Power density distribution of the laser beam

4) Absorptivity of laser beam on a material surface

5) Scanning speed of laser beam

6) Number of repetitions of laser beam scans

7) Density of the material, specific heat capacity of the material

8) Thermal expansion coefficient

9) Yield strength

10) Young’s modulus

11) Poisson’s ratio

12) Strain hardening coefficient

13) The dimensions of the workpiece

14) The melting temperature of the material and the fracture strength of the material.

In 1987 Scully determined that the positive bend angle (hence no buckling) is

equal to the fraction of the power and the square root of the traverse speed times the

plate thickness.4 That relationship was taken from earlier work by Masubuchi et al 57

on flame forming and can be seen in equation 2.6.1.



1

1

vsp

oB =α (2.6.1)

Later work in 1994 by Vollertsen has shown that there is a strong linear

dependence of the bend angle on the laser power. The dependence of the bend angle

on the processing speed has been reported in further work by Vollertsen as well 23.

Vollertsen is credited with producing a considerable amount of the early

fundamental research on laser forming, including naming the individual mechanisms

(although the names are not in use by every research group around the world), a

summary of the key results and conclusions drawn from this research 23 is given here.

In the experiments carried out relating to processing speed dependence, a

power law was assumed between the bend angle and the processing speed. A linear

dependence was obtained for 3.5mm sheet with scanning speeds in the range 7 -70

mm/s, the gradient was found to be -0.63. A negative slope was to be expected as

this is derived from the notion that the increase in processing speed decreases the

coupled energy. As the bend angle is proportional to the coupled energy it is

expected that the bend angle should decrease linearly with increasing speed.

However at lower speeds it was found that this is not the case. It was found that the

bend angle continues to increase up to a point with increasing traverse speed. This

behaviour may be attributed to the fact that the temperature gradient is increased

with increasing velocity and the time for heat conduction in the depth direction of the

sheet is reduced. Ultimately this results in the difference of the plastic strains

between the upper and the lower layer of the sheet being more pronounced and a

greater bend angle per unit time may be achieved. Of course if the velocity is

increased to a very high value then the temperature increase will be small and only

an elastically reversible bending may occur. Also of note from this work is the

concept of a threshold energy for the process. It was shown that no plastic

deformation occurs below a given energy input. Therefore the boundary energy

which will produce the onset of bending can be related to the temperature the

material must reach at the limit of the thermal strain at the yield point stress.

The thermal conductivity of the workpiece material is of vital importance in

laser bending. The thermal conductivity determines the temperature field and hence

the development of the thermal strain. If a materia1 is a good conductor it is unlikely

that a thermal gradient of sufficient magnitude can be created to initiate the bend



with the temperature gradient mechanism unless the heat can be put into the sheet

sufficiently fast. In general it is better that the material should be a relatively poor

conductor in order for the temperature gradient mechanism to occur.

The role of thermal conductivity on the buckling and shortening mechanisms

is slightly different. If the workpiece material has a high thermal conductivity then

the size of the irradiated area will increase rapidly thus decreasing the average

temperature of the material, the plastic straining and hence the bend angle. The

distinction is drawn however as forming will still occur for the latter mechanisms

(not thermal gradient dependent) but no bending will occur in the former case (TGM)

if the temperature gradient is diminished to a large extent by a high thermal

conductivity. The effect has been physically modelled by Vollertsen. It is considered

of great importance not only due to the previously described factors but also due to

the fact that the strength of the material changes with differing thermal conductivities.

The conductivity can be related to the age hardening state of the material. Time age

hardenable state then has a role in influencing the thermal conductivity and the

elastic reversible bending of the material, and ultimately the bend angle. Clearly

there are complex dependencies for the thermal conductivity.

Further work showed a linear influence of the fraction of the coefficient of

thermal expansion and the specific heat times the mass density when plotted against

the bend angle. This shows the influence of the material parameters on the bend

angle. This is useful as the two layer model 24 assumes the same relationship

between these parameters. The amount of forming depends critically on the thermal

expansion. The thermal expansion is determined from the temperature increase and

the coefficient of thermal expansion, the temperature increase of a volume is

indirectly proportional to the volumetric heat capacity 23.

In other research an early program in laser forming was that of the Laser line

Heating (LLH) which formed part of the Navy Manufacturing Technology program

(ManTech) in the U.S.A. The material investigated in this program was heavy duty

6.25 - 25 mm mild steel plate4. Important results from this work show how the

temperature changes as a function of time for the given material and how the micro

strain changes as a function of time also. The micro strain was obtained by means of

strain gauges mounted on the bottom of the plate surface. Scully noted the change in

the strain between the heating and cooling cycles as shown in figure 2.6.1.



The primary process parameters required for accurate control were identified

as the laser power P, the traverse velocity V, and the plate thickness t. The

Temperature Gradient Mechanism, although not specifically named was described as

the driving force behind the bending in this instance. Vollertsen has also studied the

temporal development of the bend angle experimentally23. Figure 2.6.2 illustrates

this. It is in agreement qualitatively with the results by Scully described above.

In work by both Scully 4 and Masubuchi 57 a linear relationship between the

number of passes over an identical track and the resultant bend angle was reported.

In later work the linear dependency of the bend angle on the number of passes has

not been found for a range of materials. Sprenger 58 showed that there is a decreasing

bend angle rate with increasing scans due to the strain hardening of the material

(figure 2.6.3). As the sheet deforms the outside of the bend cold works and the

Time [sec]

Microstrain

Figure 2.6.1: Time run of the strain development 4

Figure 2.6.2: Time run of the bend angle 23



orientation of the dislocations in the material are changed. This results in strain

hardening and each successive pass of the laser will increment the bend angle by a

smaller amount than the previous scan. In addition the work also showed that the

change in volume along the bend edge of the workpiece decreased the bend angle

rate for subsequent scans. A third reason proposed was a decrease in the coefficient

of absorption decreasing the bend angle achievable with further scans.

A summary of the influences on the bend angle with increasing scans is given:

(1) Effect of the change in thickness along the bend edge

As described by Namba 56 the upper layer thickens as the material plastically

compresses. The thermal expansion which is converted into plastic deformation is

not cancelled during cooling and a bend angle results. How ever as the material has

thickened due to the plastic compression the modulus of the section is augmented

and for the same laser parameters for subsequent scans the angle achievable will

diminish each time.

(2) Effect of strain hardening of the material

In materials with a large strain hardening coefficient and which are relatively thick it

has been shown that the cold working of the underside of the sheet which causes

strain hardening contributes significantly to the decreasing bending rate. Cold work

occurs when the temperature gradient mechanism plastically compresses the upper

layers of the sheet by thermal strain and cold works the outside of the bending edge.

Cold working increases the strength of the material. In Sprenger’s work 58 it was

shown that for AA2014 and for Ti6A14V in the mill annealed condition, the lower

layers of the sheet exceed the elastic limit after the fifth irradiation and the bend

Figure 2.6.3: Decreasing bend rate with increasing scans over an identical track 57



angle showed a greater linear dependence on the number of scans prior to this. With

subsequent passes the material will strain harden from the outside layer, layer by

layer to the neutral layer of the material. The layer between the neutral layer and the

heated layer is plastically compressed as well and consequently strain hardened.

Figure 2.6.3 shows the decreasing bend rate.

A number of other geometrical influences have been investigated. It has been

shown that the thickness of the sheet is one of the major variables in the

development of the bend angle. The bend angle is related linearly to the inverse of

the square of the sheet thickness for the temperature gradient mechanism23. The

volume of material to be heated increases with increasing thickness of the sheet.

Even with one nominal thickness the thickness of the sheet increases with each pass.

The increasing thickness is due to the plastic compression of the uppermost layer of

the sheet.

The length of the bending edge is also of significance for the development of

the bend angle. If the length of the bending edge is increased from 5 to 13 mm then

the bend angle is increased by a factor of 3 23. This is due to the changing section

modulus with changing length and the difference in the temperature field due to the

change of length in the lateral direction.

The length of the bending leg also affects the bend angle achieved per unit

time. If the bending leg is short then the cooling of the workpiece is restricted to one

side 23 and the temperature gradient decreases and hence the bending decreases. If

the bending leg is long then the gravitational forces acting on the length will affect

the bend angle. The weight of the leg results in tensile stresses in the surface of the

sheet thus reducing the compressive stresses from heating and diminishing the bead

angle.

Secondary geometrical effects were reported by Scully et al 4. Less distortion

occurs near the edges of plates according to this work. This is because the heat flow

pattern is altered in comparison to the innermost part of the plate where the heat flow

is to surrounding material. This results in less distortion near the edge of the

workpiece 4. This was also attributed to the rigidity of the plate becoming non

symmetric near the plate edge. These effects have not been investigated in depth.



2.6.2 Magee ‘98

The experimental work summarised in this section consisted of empirical work

carried out on AA 2024 T3 aluminium alloy and Ti6Al4V titanium alloy. Parametric

investigations were carried out by J. Magee29, a PhD researcher in the University of

Liverpool, into the single and multi-pass, large and small beam diameter 2D laser

forming of these materials, which led to the development of a 2D laser forming

demonstrator system for a part cylinder shape. Development of scan strategies for

the 3D laser forming of dish shapes was also carried out. This work was part of a

joint research programme between the University of Liverpool and BAE Systems.

This work formed a precursor for the present research within this thesis and as such a

more detailed account is given of the work.

PARAMETRIC STUDY – This work investigated the factors influencing the

angular dimensions of laser formed 80x80mm 0.8-1mm gauge plates of Ti6Al4V

and AA 2024 T3, commonly used aerospace alloys. The plates were clamped at one

end, graphite coated and irradiated with a PRC 10.6µm CO2 laser. Altering the

power density and the interaction time of the laser beam incident on the samples

varied the energy input to the plate surface.

The experimental results revealed that the bend angle development is

critically dependent on the energy supplied to the plate surface. Two distinct studies

were carried out (using a large and small beam diameter) on two materials (a

titanium and aluminium alloy), with different thermal and mechanical properties. In

the case of the titanium alloy it was found that the temperature gradient mechanism

was active for both studies, both for the large and small laser beam diameter to sheet

thickness ratios. This was attributed to the low thermal conductivity of the titanium

alloy. An optimum traverse velocity in terms of maximising the bend angle achieved

per scan was identified for this material when the beam diameter was in the order of

12 times the sheet thickness (figure 2.6.4).

Below the optimum velocity the bend angle dropped due to the loss of a high

temperature gradient and hence a smaller amount of differential straining through the

thickness direction. These results support the idea that the temperature gradient, and

the efficiency of the process, increase as the processing speed increases. This



efficiency increase is offset by a reduction in the bend angle after the optimum point.

This is because the increasing velocity results in less coupled energy, less thermal

expansion, and a smaller reduction of the flow stress in the heated zone. Since all of

these factors contribute to overcoming the elastic share of the bending, the bend

angle begins to drop off again.

For the aluminium alloy it was found that for one laser scan using a large

beam diameter the bend angle is decreasing with increasing traverse velocity (figure

2.6.5), this is in contrast with the titanium alloy where a peak occurs. Since the

thermal conductivity of the aluminium alloy is high, the temperature gradient in the

depth direction of samples was small for the lower traverse speeds. Under these

conditions the buckling mechanism (BM) was thought to be active. In the higher

velocity range, for the small beam diameter to sheet thickness ratio the TGM was

active, the bend angle continued to drop sharply. This is attributed as before to the

reduction in coupled energy and the elastic effects becoming more pronounced.

Figure 2.6.4: Bend angle with increasing traverse velocity for Ti6Al4V using a large beam diameter 29, 59

Figure 2.6.5: Bend angle with increasing traverse velocity for AA 2024 T3 29, 59



The experimental results were compared with calculations from the existing

two-layer model for the TGM 24. There were considerable differences in the results

obtained compared with the empirical data, these differences were attributed to no

account being taken of the strength of the material, the elastic counter bending and

the determination of the temperature field.

The decrease in bend angle with number of scans was also investigated

(figure 2.6.6). The cause of this reduction has been reported as being due to the strain

hardening of the material 58 and a change in absorption as the number of scans

increase. This study concluded that for the materials studied the effect of sheet

thickness increase in the irradiated area per scan, is of greater significance than strain

hardening.

This study also looked at edge effects or the changing bend angle along the

length of the bending edge in laser forming. The laser forming process is asymmetric

about the laser beam, as a result the bend angle cannot be constant along the entire

bending edge until the laser beam has completely scanned the sample. Ideally, after

the process, the bend angle would be constant along the bending edge, however

normally the bend angle varies with plate location (figure 2.6.7).

Figure 2.6.6: Bend angle with increasing number of scans over the same track29

Figure 2.6.7: Ideal bend angle and exaggerated view of edge effects 29, 60



This is attributed to the changing mechanical restraint, which is available to

hinder the free thermal expansion with distance from the edge of the sample and the

temperature dependent material properties. The effect is also attributed to the

contraction of the material in the direction the laser beam scans. This behaviour is

termed an edge effect. It was found that these effects could be minimised by varying

the energy supplied to the plate surface, with in plate location, by varying the speed.

The speed was varied in order to balance the thermal strain required to cause the

same amount of yielding, as the mechanical restraint hindering the thermal

expansion changed with distance from the edges of the plate and the temperature

dependent material properties.

A metallurgical study into the implications of the laser forming process using

the titanium and aluminium alloys was also carried out. It was concluded that in

order to apply laser forming to aerospace components it is necessary to restrict the

process parameter envelope to a range which does not adversely affect the

metallurgical or mechanical properties of the alloys. For the titanium alloy it was

found that oxygen uptake during processing in air contributes to the formation of an

alpha case and an increase in micro-hardness on the upper surface. To avoid this it

was concluded that processing should be carried out in an inert atmosphere such as

argon. In the case of the aluminium alloy the as received microstructure could be

maintained when an Average Energy Density (AED) of less than 25 J/mm2 was used

for forming. At higher AED re-crystallisation occurred and at extremes (greater than

133 J/mm2) a cast dendritic structure resulted from melting underneath the pure

aluminium clad layer on the surface of this alloy. A fluctuation in the micro-hardness

level about the as received value was found in samples processed at AED less than

25 J/mm2. This oscillatory nature can be explained by the re-crystallisation and

precipitation theory for this alloy.29, 61

2D LASER FORMING DEMONSTRATOR SYSTEM – A laser forming

demonstrator system was developed to demonstrate the process on a large primitive

2D shape. Data from the parametric and metallurgical study on the smaller tokens

discussed earlier was used to develop the processing parameters for the system. The

demonstrator part after some initial trials with larger parts was chosen as a flat

rectangular AA2024 T3 sheet of dimensions 450x225 x0.8mm that was to be formed

into a part-cylinder of radius 900mm (figure 2.6.8). The part was large in terms of



laser forming operations to date, and the shallow radius of curvature is almost at the

spring-back limit of conventional forming operations. The system was set up using a

CO2 CW laser, CNC tables, a pneumatic clamping system and a 3-D CAM laser

stripe measurement system.

The primary objectives of the demonstration were to obtain:

1. Geometrical accuracy, surface smoothness, and reproducibility.

2. Metallurgical integrity.

3. Specifications of the processing information required to automate the part.

With this system the scan conditions were set and then the program instructions were

executed. The surface was then profiled using the 3D CAM laser stripe and this

information was used to give the heights at various points over the sheet surface.

This data was used to give a measure of:

1. The radius of curvature

2. Any deviations in the radius of curvature along the length of the bending

edge, i.e. any longitudinal distortion or curvatures in the wrong direction.

The demonstrator system then relied on user intervention in order to determine what

the next processing steps were. These steps included:

• Next scan pattern

• Next Starting point and direction for scan pattern

• Next clamping location

• Next energy input

This adaptive approach was taken because the part produced by a constant

scan pattern, direction and clamping arrangement was twisted and distorted. The part

produced by altering these parameters and the energy input at different stages of the

process had increased accuracy, surface smoothness and reproducibility.17, 29, 62

Figure 2.6.8: Demonstrator Part 17, 29



3D LASER FORMING OF DISH SHAPES – A case study was also made by J.

Magee et al 29, 63 into the 3D laser forming of a dish shape from flat circular 2mm

gauge mild steel CR4 sheet. The objective of the investigation was to establish rules

about the positioning and sequencing of the laser irradiation lines for the

symmetrical laser forming of such a dish shape. The scan patterns investigated

employed radial or circular scan lines, or a combination of both to form the part. The

samples were verified using a co-ordinate measuring machine.

It was found that in order to achieve a smooth and symmetrical dish shape:

1. Geometrical symmetry should be reached as soon as possible after the initial

irradiations.

2. A symmetrical temperature distribution over the plate surface should be realised.

3. Any pre-orientation bend should be avoided

4. The laser beam parameters, particularly the irradiation angle of incidence and the

irradiation spot diameter, should be held constant.

To these ends the circle line system with square root radius increase, irradiating

from inside to out, was found to be one of the best strategies (figure 2.6.9). This

strategy employed the upsetting mechanism along the concentric circular scan lines

to achieve the forming result. This work was also continued successfully in larger

samples. 64

Figure 2.6.9: Circle line system with square root radius increase (inside to out), and resulting contour plot of sample.29, 63



2.6.3 Recent research in macro-scale 2D LF 65

There have been a number of other 2D laser forming experimental investigations

recently published that are of note 66-81. The experiments and conclusions of a

number of them are outlined here.

Chan et al 66 in Hong Kong published work in 2000 on the laser forming by a

low power Nd:YAG laser (90W max) of thin stainless steel (0.25-1mm thick).

Although some of the results had been observed before in other studies there were a

few interesting findings: A threshold heat input was observed in the process, below

which no bending occurs. In addition when the energy input was above critical value

the bend angle stops increasing with any further increase in heat input, possibly due

to the loss of thermal gradient through the thickness.

In 2001 Mucha et al 67 from Poland presented a paper on a comparative study

of the laser forming of plates using circular and rectangular beam cross-sections. It

was shown that the shape of the incident laser beam has a large effect on the LF

process, for the same energy parameters the bend angle generated by the rectangular

beam is 1.3 to 2.5 times greater than the one generated by the circular beam. In

addition it was argued that a greater control of the process could be achieved using a

rectangular beam. Also in the work was an extensive analytical model, it was argued

that the dimensionless form of the derived dependencies from this model would be a

useful method of selecting appropriate processing parameters for any material.

Liqun et al 69,70 from the Harbin Instiute of Technology, China, have

published a number of interesting studies on LF. There has been an increase in recent

years of the number of papers to come out of China in the laser materials processing

field in general. Research on using different cooling methods in LF and an update on

the factors involved in multi-pass LF have been presented. It was found that the use

of an effective cooling method can significantly increase the process efficiency of LF.

A number of cooling methods were employed including water jet and high pressure

CO2 gas cooling from the underside of the plate. It was found that cooling can

increase the temperature gradient through the thickness, however the peak

temperature in the sample is decreased (hence the bend angle is also) and so the

benefits of having cooling during processing are lessened. In addition the high

temperature difference between the top and bottom of the plate can cause unwanted



metallurgical effects such as an increase in hardness. Effective cooling after each

pass would be the answer to improve process efficiency. In another study work was

presented on multi-pass laser forming and the factors influencing the bend angle fall

off at higher numbers of passes, although many of the results had been observed in

previous work a number of key results were of interest. The work was conducted on

2mm thick aluminium using a 2kW Nd:YAG laser and a 10mm beam diameter, thus

no coating was required. It was observed that there was a significant thickening of

the sample in the irradiated area and that the thickening became more pronounced

after 20 irradiations (0.5mm thickness increase). It was argued that this thickening

effect was a large factor in the bend angle rate fall off after 20 or more irradiations, a

thicker sheet is harder to bend as it were. It was also found that the tensile strength of

the sample decreased with increasing irradiations, thus ruling out this factor as a

possible reason for bend angle fall off.

The industrial viability of LF depends on the process efficiency and speed

compared to other competitive techniques, in combination with repeatable accuracy.

Mechanical forming of thin sheet material is relatively fast and can produce large

deformations in a single process step, but it is inaccurate (due to variable ‘spring-

back’ and tool wear). In contrast, LF is comparatively slow, but offers a high degree

of control and remote application. To exploit the combined attributes of LF and

mechanical forming, a hybrid forming process has been proposed and demonstrated

by Magee and De Vin, in which LF is applied as a secondary process to adjust

mechanically formed parts 73, 74.

A high degree of control is required for LF to be an industrially viable

process, especially for 3-D LF, which is in its infancy. To address this for 2-D LF,

Thompson and Pridham at Dundee investigated a closed-loop control system, for the

case of laser bending to a pre-defined bend angle with some success. 75, 76, 77 Other

approaches to this problem have been taken, Cheng and Lin in Taiwan have

published work on using a neural network to predict and hence control the bend

angle during the laser forming of 304 stainless steel.78 This intelligent approach

could be a useful method of accounting for the many variables and unknowns in the

process. Peck and Jones79 are developing this approach into a commercial system for

the manufacture of single unit windshield wiper blades for Trico Products Corp. The

system employs a high power diode laser to laser form and heat treat a continuous

metal strip fed underneath it, this metal strip when cut to length forms the curved



backbone of the windshield wiper. A trained neural network provided by NA

Technologies monitors the process and can make adjustments on the fly to the power

output of the diode laser and hence the bend angle, this type of laser is well suited to

this dynamic power requirement.

The use of pulsed laser energy for LF has also been under investigation by

some researchers80, 81, and this has lead to possibilities for new LF mechanisms.

Laser Peen Forming (an extension of Laser Peening 80) or Laser Shock Forming81, in

which the application of a negative residual stress to one side of a component results

in a bending effect, is a key example that is now being realised through the

emergence of high pulse energy Nd:Glass lasers.

2.6.4 Recent advances in 2D LF for micro-scale applications 65

Pioneering studies on micro-scale LF by Hoving and co-workers at Philips 13 helped

establish some important results for precise adjustment of components by the

shortening or Upsetting Mechanism. Following on from this, the Philips research

team have developed a number of micro-adjustment applications for LF 13, 82, with

the first concept being the laser adjustment of digital audio head mounting frames. A

second application under development is the laser adjustment of reed switches. Here,

out-of-plane LF (laser bending) is used for adjustment of the 10-50 micron gap

between the two nickel-iron reed elements, which then determines the value of the

magnetic field above which the reed switch closes in operation. The process is

carried out using a 30W cw Argon-Ion laser beam which is able to pass through the

sealed glass tube, so that the operation can be carried out after manufacture of the

sealed envelope switch (both on new switches and for the adjustment of previously

rejected units). The precision of adjustment results in a reduced spread of ampere

winding values, with a corresponding reduction in customer sort / reject rates.

Thirdly, actuator frames have been designed and investigated by the Philips team for

micron level adjustments of a lens by LF during the final assembly of a CD player,

one of a number of applications for this product’s optical train. Through a study of

several actuator designs, one resembling a ‘Basket Ball Basket’ (figure 2.6.10) was

found to give the greatest flexibility of movement & reduced number of process

steps to achieve the desired lens alignment. Freedom of movement in a number of



axes is achieved by applying laser energy in small strokes over several areas of the

actuator frame, thereby inducing microns of either in-plane or out-of-plane

deformations, as required by the alignment task. Other groups are also developing

fixturing for micro adjustments using LF 83, 84, 85.

The technique heralds a fresh approach to product design, allowing flexible

and efficient adjustment of key components in the final stages of product assembly.

An additional concept under investigation is the development of novel on-board

actuation techniques, using miniature low power laser devices such as diode lasers,

together with control feedback, to allow the product to carry out self-adjustment of

its key components by LF.

In 1997, Tam and co-workers at IBM Almaden developed and implemented

in manufacture a Laser Curvature Adjust Technique (LCAT) system for adjusting

the curvature of magnetic head sliders in disk drives using a novel laser micro-

bending technique 20. This development addressed a need for precise and highly

controlled adjustment of the positive camber curvature of a slider, to improve its

tribological properties and allow reduced flying heights (below 25nm) above the disk

surface to provide increased disk storage density. Conventional lapping techniques

for slider fabrication had become inadequate and could result in an unpredictable or

undesired form of curvature change. The LCAT process involves scribing

microscopic patterns on the reverse side of the slider, which induce surface stress

changes in the alumina-based ceramic material to produce a corresponding curvature

change at the front side air bearing surface (ABS) pads. The scribing is performed

with a compact, pulsed, diode-pumped Nd:Vanadate laser and is integrated with an

Figure 2.6.10: Actuator for CD lens adjustment by micro LF82



optical monitoring technique for closed-loop control of slider shape (curvature) to

accuracies of within a few nanometres.

In joint work at Chemnitz and Mittweida Universities in Germany, by

Gaertner and co-workers, laser bending has been used for plastic reshaping of wet-

etched silicon micro-scale structures 86, 87. Figure 2.6.10 shows the result of LF 50

micron thick, 960 micron wide silicon beams using a cw Nd:YAG laser beam

focused to a 2mm spot. Here, LF offers a non-contact process for localised reshaping

of the structures, without the need to heat the whole structure to >700ºC in a furnace,

or the need for associated special forming tools. Out-of-plane bending by TGM is

established by scanning the laser spot across the width of the ‘beam’ structure.

Results show that, during the plastic deformation stage, dislocations are generated in

the near-dislocation-free mono-crystalline silicon that then affect the mechanical and

physical properties of the material and thus allow bending or reshaping to take place.

The dislocations only occur on reaching the yield point of the material, unlike in the

regions of elastic deformation. Applications of this laser micro-bending process that

are under investigation include micro-mirrors for optical circuits and micro-scale

grippers or ‘staples’ for semiconductor chips.

The laser forming of plastic using an Nd:YAG laser has been experimentally

investigated by Uno and co-workers at the West Japan Railway Co. in Osaka 88, with

the aim of changing the design shape of plastic components produced by injection

and compression moulding. The bending direction is controlled by selective painting

of the plastic surface with a black resin, on the side whose surface is to undergo

shrinkage in the LF process. A cw Nd:YAG laser of a few watts average power is

applied from one side only, since the plastic is transparent to the laser wavelength,

Figure 2.6.11: LF of 50µm thick beams in wet-etched silicon micro-scale

structures 86, 87 (Photos courtesy of E. Gaertner, Technical

University of Chemnitz, Germany).



with the resin then transferring its absorbed heat to the material from the surface to

which it is adhered.

A number of other studies fall into the micro scale LF area 89-92, these involve

the laser forming of thin metallic foils using low power lasers. Complex structures

can be formed in these thin materials with very little power, thus applications can be

found in the electronics industry for alignment (mentioned earlier) and for the

manufacture of mounts and housings for components. In work by Yoshioka et al 90,

91 92 at the Chiba institute in Japan a method of sample holding was developed to

reduce unwanted distortion when forming thin foils. A sample was held in place

using a glass plate over it, the sample was then irradiated through the glass and not

allowed to deform. Once the plate was lifted after processing, the part sprung into the

desired shape, thus eliminating any asymmetric or temporal effects. For more

complex shapes a mask was used to hold the sample down during processing.

2.6.5 Developments towards 3D LF capability 65

A considerable amount of research recently in LF has been aimed at prototyping of

3-D components and structures for applications in aerospace, automotive and artistic

design 18, 63, 64, 93-107

It has been shown that laser forming shows great potential for the

manufacturing of metallic components, using a 2D straight line or 3D spatial

forming approach. However in order to advance the process further for realistic

forming applications and for straightening and aligning operations in a

manufacturing industry it is necessary to develop systems for accurate and repeatable

part production. Figure 2.6.12 outlines the possible routes and key elements required

to the practical realisation of 3D laser forming 22. A predictive or an adaptive

approach can be taken.

Intelligent predictive systems, perhaps based on Knowledge-Based Systems

(KBS), neural networks or thermo-mechanical models can achieve predictability

through a knowledge of the material (including its stress history) combined with a

developed, highly tuned process model / control algorithm. Systems of this type have

been reported by a number of groups 98, 99, 100.



In an adaptive system the use of sensors to provided accurate controlled

feedback coupled with the development of intelligent control software e.g. neural

network, provides an incremental or even real time closed loop method of accurate

3D laser forming, based on the current part characteristics independent of material

variability e.g. residual stress. Systems of this type have been reported by a number

of groups. 102 - 107

It is likely that future 3D laser forming systems would include elements of

both these approaches, in that an initial prediction for a scan strategy would be made

based on a knowledge base of known data, the part’s geometry would be monitored,

and the scan strategy would be adapted either in process or for subsequent passes so

as to achieve the desired result. Work is ongoing in this area at a number of research

groups including Reutzel el al at ARL in Penn State University on ship hull

components with some successful results.106 Research in this area is also presented in

this thesis.

Predictive system Adaptive system

IntelligenceSensors

KBS

Parameter look-up

Neural Network

Repeatableaccuracy

Accurate model

Control feedback

Response tuning

Multi-axis LF process

Rapid data processing

2-D LF Basics

Real-time /incremental

Process mechanisms Material properties

Residual stresses (history)

Figure 2.6.12: 3-D Laser Forming: routes to practical realisation and key elements required 22



2.6.6 Material and Metallurgical Studies

As has been mentioned LF has great potential as a tool for prototyping, aligning and

removal of distortion or as a direct manufacturing tool. Key to the future success of

LF is what effects if any the process has on a material’s integrity and properties. In

order to prove the process a number of material and metallurgical studies have been

conducted on a variety of materials. 4, 14, 15, 61, 62, 108-118 In particular the effect of the

rapid and repeated heating and cooling cycles below melting points associated with

LF. Initial work in the field was mainly carried out on steel 4, 108, but has since been

extended to other materials including titanium and its alloys 14, 15, 109, 110, aluminium

and its alloys 61, 62, 111, 112, aluminium-matrix composites 113 and chromium114. Other

fundamental investigations have also been reported, including investigations into

material anisotropy 115-118. A summary of the more relevant publications to this study

are presented here.

In work by Thompson and Pridham 108 at the University of Dundee on laser

formed mild steel, it was shown from mechanical and metallurgical tests that LF

parts (in mild steel) are likely to perform at least as well as conventionally formed

equivalents. It was reported that in general laser forming increases the yield strength

of the material locally to the irradiated area. This increase in strength may not be

utilised fully, since the bulk of the material will not have been altered by the process,

but most significantly LF does not weaken structures. The slight loss of ductility

reported would mean that a laser formed part may not be suitable for large amounts

of subsequent manual forming. This was not felt to be a problem, since LF is likely

to be used as the sole forming operation or a fine adjustment after conventional

bending.

In work by Shackle et al 15 from UMIST on 2mm gauge Ti-6Al-4V (Ti64)

sheet (a study that the author of this thesis had some input), an investigation on the

metallurgical implications of LF on this aerospace alloy was reported. The effect of a

post-forming heat treatment was also investigated. The LF samples were processed

in air and in argon. O2 readily diffuses into the surface of Ti64 at temperatures

exceeding 550°C and produces a brittle α-case (figure2.6.13a), this can weaken the

material as crack propagation points can form in the surface region. Due to this



factor LF on this material has to be carried out in an inert atmosphere, the α-case was

not present in the samples processed in argon (figure 2.6.13b).

Ti64 is a dual phase, α+β alloy, where the β-transus temperature is at 982°C.

It was found by optical microscopy, FEGSEM and TEM methods that after LF a

HAZ is produced that consists of a fully martensitic region, where the temperature

has exceeded the β-transus, surrounded by a partially transformed zone where the

temperature has risen above the Ms (martensitic) temperature (~800°C) into the α+β

phase field. No microstructural changes where found at lower temperatures. Within

the HAZ a complex, refined, martensitic structure was produced due to the very high

heating and cooling rates into and from the β and α+β phase fields (reported as a

maximum of 9100Ks-1 as derived from a finite volume model). However, because of

the very rapid nature of the thermal cycle the original solute distribution in the parent

material was little altered and could still be seen within the fully transformed region.

An increase in hardness was reported in the irradiated region reducing into the

thickness (figure 2.6.14), however the application of a post-forming heat treatment

(PFHT 700°C for 4 hours in Ar) resulted in an overall reduction in hardness of the

HAZ due to a re-precipitation of Vanadium rich β at the martensitic plate boundaries

and the disappearance of the majority of the martensitic plates in most regions within

the HAZ.

Figure 2.6.14: Hardness variation with depth through the sheet thickness for Ti6Al4V (760W 30mm/s 6mm beam).15

Figure 2.6.13: SEM micrographs of Ti6Al4V formed in (a) air and (b) argon. (Forming parameters: 760W / 30mms-1).15



A tensile test on the LF samples was also reported, it was found that during

tensile deformation the higher hardness of the HAZ acted as a local constraint on

plastic deformation and failure always occurred in the parent material away from the

irradiated area. The bulk tensile properties of the Ti64 sheet consequently remained

relatively unaffected by the LF process. Studies in other alloys of Titanium14, 109, 110

also confirmed the viability of LF with these materials.

Work on aluminium and aluminium alloys by Merklien et al 111, 112 at the

University of Erlangen revealed the microstuctural development and mechanical

properties in laser formed Al1050 and an Al6082 in two heat treatment conditions

T41 and T61. The work was conducted on 80mm wide 1mm thick samples using a

1kW cw Nd:YAG laser, a graphite coating was still used however. SEM and TEM

methods were used for analysis. Changes in the mechanical behaviour as well as in

the microstructure were observed. The soft and annealed AA-1050 showed

hardening due to the LF process. This was proved by hardness tests and by

SEM/TEM images showing the dislocation motion and changes in microstructure.

For the two heat treatments of the Al6082 there were little or no differences in

forming characteristics over a number of irradiations found between them. For the

artificially aged T61 alloy, the hardness produced by ageing is lost in the HAZ and

immediate area after LF and is comparable to the naturally aged T41 values. For the

T41 alloy only a slight decrease in hardness is observed in the HAZ.

In 2003 Yao et al 118 from Columbia University presented research on the

effect of material anisotropy on the laser forming process, both numerical and

experimental results. Cold rolled sheet metal exhibit anisotropic properties which are

mostly caused by preferred orientations of grains developed during rolling reductions.

The anisotropic index or R value of a material in a particular orientation was

determined using an ASTM standard tensile test. The grain textures in the formed

samples were determined using an electron back-scatter diffraction method (EBSD).

It was found that there was a significant difference in the laser forming

characteristics of the cold rolled AISI 1010 mild steel depending on the orientation

of the scan line to the rolling direction in the sample. It was found that the

anisotropic effects increased with increased rolling reductions i.e. thinner materials.

It was also concluded that the higher the temperature achieved in the sample the less

the materials’ anisotropy has an effect.



2.7 Potential Applications & Competing Processes Laser forming has potential for prototyping, aligning and removal of distortion or as

a direct manufacturing tool in the industry sectors aerospace, automotive,

shipbuilding and microelectronics. A number of possible applications in these

sectors have already been discussed in the previous sections.

The full potential of LF will only be realised through improved process

knowledge and associated system developments. With many of the currently

identified limitations already being addressed in ongoing research, the process has

significant potential for use in a broad range of industrial applications and sectors,

including shipbuilding. Table 1 is a summary of the short-term degree of application

potential in various stages of a (component non-specific) product life cycle.

Compared to other forming processes, LF has the advantage of process flexibility, in

that it could be carried out alongside other laser processes (cutting, welding & others)

by multi-purpose laser systems. For large scale LF of metals e.g. in shipbuilding, the

high equipment costs and safety requirements are currently key concerns, but these

should be alleviated by the continuing development of cheaper, more compact and

more efficient sources (diode and fibre lasers) and automated LF systems.

Stage of industrial application (Product Life Cycle) Degree of application potential

Design & Development

Manufacture (processing)

Product assembly

In-service operation

Repair & Maintenance

High Rapid Prototyping

Forming (Hybrid LF)

Precision alignment & adjustment

Medium Distortion & shape

correction

On-board automatic correction

Damage & distortion correction

Low Forming (LF)

Table 2.7.1: Degree of application potential for LF in various stages of a general

product life-cycle (not specific to component scale, material or geometry) 22

Laser forming at a macro level is developing from a knowledge base of basic

2D mechanisms to a practical realisation of 3D laser forming of complex structures

routes to practical 3D laser forming may encompass elements of both predictive and



adaptive systems. Promising applications for LF are in rapid prototyping, net shape

production and distortion correction. It is likely that a hybrid system would be the

most practical manufacturing solution, e.g. using a laser to selectively heat a

component prior to a mechanical forming operation so as to reduce the yield stress

and hence the required force to deform it.

2.7.1 Projections for Potential Applications of Laser Forming in

Shipbuilding22

The current picture of shipbuilding technology and projections for the near future

suggest a continuing demand for metal forming processes. Currently, various types

of sheet metal forming processes are employed in shipyards, these mainly being

mechanical (such as roll bending). Figure 2.7.1 shows examples of some steel plate

forming technology currently in operation in UK shipyards (Courtesy of BAE

SYSTEMS, Glasgow).

BAE SYSTEMS on the Clyde have recently provided the authors with some

figures on the approximate levels of cost and time of forming in naval shipbuilding

today, with some projections for BAE’s demand for forming over the coming 10

years. Taking the general figures first, it is possible to estimate the percentage cost of

steel plate and section forming. Considering that (i) steelwork fabrication is around

15% of the total labour spend on a vessel; (ii) steel preparation is approximately 15%

of the steelwork fabrication labour spend; and (iii) steel plate and section forming is

approximately 8% of the total steel preparation time, then (multiplying these factors

together) it can be seen that forming constitutes approximately 0.2% of the total

Figure 2.7.1: Some Current Forming Techniques in Shipbuilding



Figure 2.7.2: Bulbous Bow from the QM222

labour spend on a vessel. While this may appear to be a small fraction, an example of

its true value is the projection from BAE Systems that on the Clyde the cost of steel

forming over the next 10 years is estimated to be in the order of £600k for labour

alone – not accounting for any re-work which has been reported to be considerable.

Some projections for potential applications of laser forming in shipbuilding can be

made by considering it either as a substitution process for existing forming methods

(for reasons of enhanced flexibility, increased control etc), but also for wholly novel

techniques that perhaps could not have been considered with other forming processes

and which may even provide unique advantages. With this in mind, the potential

applications under consideration for future work at Liverpool are in the following

areas:

Hull section fabrication (and alignment in future assembly operations)

Correction of distortion (due to welding and other processes)

Shaft / propeller alignment

In the fabrication of hull sections, the main area of interest is to use 2-D and

3-D laser forming to replace mechanical methods of steel plate bending, for a

material thickness range of up to 20-25mm (1inch). The most straightforward case is

the 2-D laser forming of part-

cylinder shapes for hull skin panels

to be subsequently welded together.

However, as the capabilities of 3D

laser forming begin to evolve, it

will be possible to consider using

the process to produce primitive 3D

shapes involving various double-

curvatures (‘saddles’ and ‘pillows’),

which would then be patched together as elements of a larger, more complex

structure. A key example of this concept is the ‘bulbous bow’, which has been

quoted by a number of yards as being one of the most difficult and time-consuming

parts of a ship to construct (Figure 2.7.2).

The correction of distortion (chiefly that due to welding operations) in

shipbuilding remains a significant issue, even though the last few years have seen the

introduction of a number of advanced ‘reduced-distortion’ welding techniques



(including laser-based or laser hybrids 119). For many larger projects, the manual re-

working of weld distortion can effectively use up 30% of the total ship production

labour costs and the new techniques referred to above can often only be applied to

ship deck flat panel construction, where component geometry allows. Therefore,

there must be significant potential for using an advanced 3-D laser forming system to

address at least some of this need. As an example, BAE Systems report that they

perceive potential issues around plate distortion in projects such as the Type 45

Destroyer programme, which requires a light hull structure in order to deliver its

specified sea speed. Since laser cutting is already being used and laser welding is

currently under investigation, there is now growing interest in laser forming for thin

plate if it can deliver productivity, throughput or cost savings.

Shaft alignment –Investigations have be reported using laser forming for the

straightening of rod and cylindrical tubes at smaller scales, where a glancing

incidence of the laser beam allows an almost self-straightening effect to be

established as the component is rotated and the laser beam moved along the

component length. It would be of interest to see if this could be scaled up to marine

shaft parts.

Ships propellers are large components cast from special alloys and their

performance characteristics depend strongly on the curvature variations along their

surfaces. Once cast, if modifications are required either during production or at a

later maintenance / repair phase, this requires large-scale and expensive machining

capability. This is a further area for investigation of laser forming

2.7.2 Potential Applications in the Aerospace Sector As mentioned LF can be useful for prototyping, aligning and removal of distortion or

as a direct manufacturing tool. Current forming practices in the Aerospace sector

include traditional die forming, super-plastic forming (of Ti) and hot creep forming.

Lf can certainly compete with die forming in terms of process flexibility and cost

(apart from initial laser investment), in particular for low volume parts and

prototypes, where the cost of the die or hard tooling can be in excess of £5K. The

structures that can be formed using the vacuum or blown forming technique in super-

plastic forming of titanium alloys would be difficult to achieve using LF, however

the components can have considerable distortions post-forming and a large amount



of man hours are used in the manual removal of distortion, usually by a skilled

hammer. LF (3D LF) could be used here for the removal of distortion; however there

may be a limit on the complexity of distortion that could be rectified further research

would reveal this. The distortion removal aspect of LF also could be used in the

chemical etching process, where the removal of material from a component, usually

to reduce weight, can produce unwanted distortion. Again these parts are invariably

manually straightened and so LF could be used to automate the process, particularly

where the distortions are uniform and reasonably repeatable.

An area where LF has the most potential to replace an existing process is hot

creep forming. This process involves the forming of material, usually titanium alloy,

using a heated die and press tool. A plate (sprayed with Boron Nitride to aid thermal

diffusivity) is placed in the press tool and die, the plate is then heated to 800°C and

formed over the die using much less force and inducing much less stress in the

component when compared to cold forming. If Ti64 is used, the plate is then sand

and vapour blasted prior to chemical etching to remove the α-case (mentioned earlier)

due to the process taking place in air. The disadvantages of this process specific to

LF are:

Long tool change-over times (1/2 day)

Long warm up/cool down times of the die (16 Hours)

Removal of α-case

Cost of effluent disposal

Cost of tooling

Storage of tooling

Inflexibility

Cooling of dies between components

High energy consumption (20Kw/H)

LF has a number of advantages over hot creep forming in these areas:

Increased flexibility

Reduced tooling costs

Possible Single piece forming of components

Reduced need for etching (Argon atmosphere or local shrouding)

Faster changeover times (batches of 1 possible)

Simple product changes



Figure 2.7.3: Hot creep formed ‘A’ frame strut, possible to manufacture using LF

There are disadvantages of the LF such as a longer cycle time and initial cost of

equipment, however, it is felt that the potential benefits of LF would outweigh these

negatives.

A possible hot creep formed component which could be manufactured by LF

is an ‘A’ frame strut section from a Roll-Royce Trent 700 Aero engine. The

completed strut can be seen in figure 2.7.3 (picture courtesy of Rolls-Royce). The

straight sections of the strut are hollow U channels made up of long 2D bends

(~600mm long) and so it should be possible to laser form them. The channels are

then electron beam welded together and to the end sections, this is a further possible

use for a laser, in that the part could be formed a welded in the same workstation. A

study into the manufacture of this component by LF is given in this thesis.

2.8 State of the Art

The ‘state of the art’ in LF is in the research that has taken the process out of the lab

and is on the verge of a breakthrough into the manufacturing environment. There are

two areas where this is happening, the use of LF for micro adjustments in the

electronics industry particularly in hard drive manufacture 20, and in the shipbuilding

industry for the manufacture of hull components 103-106, thus automating a manual

‘black art’ operation. This has been brought about by an improved understanding of

the LF process and the integration of sensors and control systems to improve the

repeatability of the process.



2.9 Synopsis for Present Research

There has been a considerable amount of work completed on 2D laser forming to

date, however due to the many variables in the process and numbers of materials and

material types that can be laser formed a full understanding of the process is some

way off. The work on 2D laser forming presented in this thesis aims to increase the

knowledge and understanding of the process, in particular the transient thermo-

mechanical and asymmetrical effects plus aspects for closed loop controlled LF.

Materials investigated include mild steel, aluminium 1050, aluminium 6061 (in three

heat treatments O, T4 & T6), Ti6AL4V and newly developed Metal Laminate

Composite Materials (or Fibre Metal Laminates).



environment it is necessary to consider 3D laser forming. Less work has been

completed in this field compared to 2D LF, however the process has been shown to

have a great deal of potential. In order to compete directly with conventional forming

techniques though, such as die forming the process must be proven to be reliable,

repeatable, cost effective and flexible. It is the potential flexibility of 3D laser

forming that offers the greatest benefits, in that a change to a required part geometry

could be implemented easily through the CAD driven process, this can be compared

to the expensive and in-flexible hard tooling requirements of the die forming process.

The work presented in this thesis on 3D laser forming aims to prove the viability of

this technique as a direct manufacturing tool and as a means of post-conventional

forming (or processing e.g. chemical etching) distortion removal. To this aim

progress towards repeatable closed loop controlled 3D LF is presented. The materials

investigated were mild steel and Ti6Al4V.

Chapter 3 Experimental Procedure


Chapter 3

Experimental Procedure This chapter reviews the equipment, set-up and procedures for each of the studies

undertaken for this thesis. For clarity of presentation, the studies are divided into

sections entitled ‘2D laser forming’ and ‘3D laser forming’. As mentioned earlier,

2D laser forming encompasses laser forming operations that utilise two dimensional

out-of-plane bends to produce three dimensional results e.g. a fold. 3D laser forming

encompasses laser forming operations that can utilise combinations of multi-axis two

dimensional out-of plane bends and in-plane localised shortening to produce three

dimensional spatially formed parts such as a dome.

3.1 General Set-up This section details the general experimental set-up used, not specific to any of the

individual investigations, such as the laser, beam manipulation and software tools.

The hardware and software developed for this work are detailed in the following

sub-sections.

3.1.1 Hardware Except where stated, all of the experimental laser forming studies reported in this

thesis were performed on a specially designed and constructed CO2 laser system that

was developed for laser forming operations as part of this research.

As the process involves the use of a high power laser, some background is

given here on their construction and type:



A laser beam is a high intensity monochromatic (single wavelength) coherent

beam of light, which is generated by a laser cavity. A laser cavity is an optical

oscillator made up of two mirrors placed parallel to each other and containing an

active medium (can be a gas, solid or liquid). If energy is supplied to the medium

(normally in the form of a DC or RF electrical power supply or in the form of

focussed pulses of light, or in the form of a chemical reaction, depending on laser

type) it must be capable of amplifying the light passing between the two mirrors by

the mechanism of stimulated emission, hence the name LASER is an acronym for

the process. Light Amplification by the Stimulated Emission of Radiation

One of the mirrors that make up the cavity is only partially reflective while

the other mirror is totally reflective. This means that a fraction of the light oscillating

between the two mirrors can be allowed to escape along the optical axis as the

working beam (usually a shutter prevents the beam propagating when not required).

For the CO2 laser used in this study the active medium in this laser type is

carbon dioxide gas. The carbon dioxide molecule is made up of a carbon atom

covalently bonded to two oxygen atoms. While constrained by the atomic bonds

between them, these atoms naturally oscillate about each other as a result of thermal

energy. The molecule can exist in a number of discrete energy states which depend

on the orientation of the oxygen atoms with respect to the carbon atom in the

molecule. The principle of the process is that a photon (or energy) of a particular

wavelength (10.6µm in the case of this laser type) is produced as the CO2 molecule

transforms from an upper energy state to a lower (conservation of energy principle).

The requirement to produce a laser cavity in which there is amplification of the light

energy produced (lasing) is that the lifetime of the upper laser level is higher than

that of the lower laser level, which is the case for the carbon dioxide molecule. Then

population inversion can be achieved by the selective excitation of the upper laser

level.

The production of carbon dioxide molecules in the various energy states is

achieved in the cavity by subjecting the gas to a high voltage electric discharge. It is

in this way that energy is pumped into the system to eventually obtain the laser beam

with which useful work can be carried out. In order to maintain a high population of

upper laser level molecules a process of selective excitation to the upper laser level is

achieved by addition of nitrogen to the carbon dioxide. Nitrogen has the property

that it only has one excited state and the energy level of this excited state is very



close to that of the upper energy state of carbon dioxide. Hence, if excited nitrogen

molecules collide with carbon dioxide in its natural (unexcited) state, a transition

directly to the upper energy state of carbon dioxide will occur. This leads to high

efficiency by aiding the ease of formation of the population inversion that is required

for stimulated emission. However, the carbon dioxide must be kept cool in order for

this transformation to take place. As a result, the design of carbon dioxide lasers

must include features that ensure that the gas is kept cool (heat exchangers) and that

the exited molecules are brought back down to the non-exited state, from an

intermediate state, such that a bottleneck in the cycle is avoided. This is overcome by

adding helium to the cavity gas mixture. The action of He is to absorb energy by

collision with the carbon dioxide molecules in the bottleneck intermediate state and

transfer this energy as heat to the walls of the laser cavity where it is removed. This

is aided by the high heat conductivity of He. The optimum gas composition for a

carbon dioxide laser is then:

He - 77% - Cooling, N2 - 13% - Excitation, CO2- 10%- Lasing Medium

There are three main types of CO2 laser, depending on how they are cooled and how

the gases are circulated:

• Slow flow (SF) – cooling by conduction through the cavity wall.

• Fast axial flow (FAF)

• Transverse flow (TF) –cooling by convection

The 10.6µm wavelength radiation of the CO2 laser is reflected by most metals

(~90% reflection) such that the energy coupled into a workpiece is only a fraction of

the total energy incident on it. Painting or roughening or changing the angle of

incidence of the surface can improve the absorption but the small fraction absorbed

is generally sufficient for materials processing.

The CO2 Laser is by far the most commonly used laser for materials

processing, it has been estimated that more than ten thousand with a beam power

above 1kW are employed around the world. This is due to its excellent beam quality

and unlimited beam power (available as 45kW continuous wave versions) and also

the fact that it has become rugged and reliable in a workshop environment 120.

The laser used for the majority of the studies reported in this thesis was an

Electrox 1.5kW CO2 fast axial flow continuous wave (CW) laser with a Class 4

safety designation. The laser can be seen in figures 3.1.1 and 3.1.2. As with most



high power industrial gas lasers the cavity is of a folded construction (using mirrors)

so as to conserve space (figure 3.1.2), the longer the cavity the higher the laser power

output. The cavity, vacuum pump, turbine, heat exchanger and power supply

(transformer) are self contained within the cabinet (figure 3.1.1).

Fed into the laser enclosure are the three constituent gases, via large gas

bottles (figure 3.1.1), a high voltage (HV) power supply and cooled water from a

chiller unit (Coolmation) for the heat exchanger and optics cooling. Although the

laser beam itself is invisible to the human eye (infra-red), the HV power supply

across cavity causes a discharge or ionisation of the gases, for the combination of

gases used in a CO2 laser the discharge colour is pink (figure 3.1.2). Changing the

combination of gases the discharge colour would also change.

Figure 3.1.1: Electrox 1.5kW CO2 Laser (exterior enclosure)

Figure 3.1.2: Laser Cavity, Heat Exchanger and Cavity Discharge



This laser has an M2 of approximately 2.5 (M2 is the beam quality factor, a

measure of divergence, 1 is a theoretical Gaussian beam). The laser beam has a ‘top

hat’ energy distribution, an image of the beam can be seen in figure 3.1.3. This

image was taken with a Spiricon PyroCam III laser beam analyser (LBA).

Although not a perfect Gaussian beam the ‘top hat’ energy distribution lends

itself well to the laser forming process where an even energy distribution across the

heated area so as to avoid excessive heating and melting is desirable.

The laser is shared between two workstations via a pneumatically operated

turning mirror (water cooled gold coated copper mirror). One of these workstations

(Workstation 2) has been developed as part of this work for purely laser forming

operations (figure 3.1.4). The laser beam was fed to this workstation via turning

mirrors and enclosed flight tubes (~3.5m run) to a processing head containing a

water cooled focusing optic and co-axial nozzle arrangement, for 10.6µm radiation

zinc selenide (ZnSe) is used as the transmissive optic.

Figure 3.1.3: Electrox 1.5kW laser beam energy profile, PyroCam III image

Figure 3.1.4: Workstation 2, 3 Axis beam manipulation



The workstation consists of a 3 axis (x, y, z) CNC table for work piece

movement (figure 3.1.4), for this setup the beam remains stationary and a part is

moved around it. The processing head containing the focusing optics is attached to

the vertically mounted Z axis with 300mm travel; this provides focus control and

beam size selection (de-focused beam), this axis has an in-built brake to prevent any

unwanted movement. The X and Y axes each have 435mm travel and are driven by

DC servo motors. A CAD drawing of the workstation layout is given in figure 3.1.5.

The tables and control system were purchased from Naples Coombe Ltd and

were based around their Servostep 1700 system. This employs a Galil DMC1730 PC

based controller (ISA card) interfaced with the tables via servo amplifiers housed in

a Servostep chassis. As the controller is pc based this allows custom software to be

written to automate a number of processes, this will be discussed in the next section.

The control card has full digital I/O plus Analog input capabilities. Control of the

shutter (normally manually push button operated) can be given to the controller via

the I/O line, a control relay and a key operated switch, more detail of the system is

given in Appendix 4. Integrated with the system is an MEL M5 laser range finder

with a range of 100mm, a resolution of 30µm (on a white surface) and produces a +/-

10Volt analog signal corresponding to +/- 50mm from a reference point 220mm

Figure 3.1.5: Workstation 2, CAD Drawing of layout



from the sensor. This analog signal is fed to the analog I/O on the controller, which

has a 12-bit analog to digital converter. The laser range finder is mounted behind the

lens holder (figures 3.1.5 and 3.1.6), the sensor operates on the principle of

triangulation between a red laser diode spot and a photo-sensitive diode. More

technical information on the range finder is given in Appendix 5.

The laser range finder was integrated into the system in order to give an

online single point non-contact method of determining height, thus bend angles and

surface profiles could be acquired through the use of custom written software

(discussed later).

Affixed to the top of the X & Y stages is a steel work bed with a number of

drilled and taped holes. This allows for a number of sample holding methods and

clamping arrangements for processing which include, centre clamp (figure 3.1.7),

edge clamp (figure 3.1.8), corner clamp (figure 3.1.9) and unclamped with guides to

prevent sliding as the tables accelerate (figure 3.1.10).

Figure 3.1.6: MEL M5 Laser Range Finder

Figure 3.1.7: Centre Clamp Figure 3.1.8: Edge Clamp



The studies on laser forming presented in this thesis required the selection of

various de-focused beam diameters. As mentioned the de-focused beam diameter is a

function of the lens to workpiece stand-off (Z position), the lens focal length, the M2

of the laser, the wavelength and the diameter of the beam before the lens. Beam

diameters were determined using standard beam propagation equations and burn

prints in wood at various focal positions. The mathematical method of determining

beam diameter at a known focal position is given in Appendix 3. An example of burn

prints in ply-wood at 5mm Z axis steps, working below the focus, is given in figure

3.1.11.

The actual effective beam diameter is taken as the inner ring of higher

intensity and not the overall diameter. This corresponds to the ‘top hat’ profile with a

slight halo of lower intensity observed earlier (figure 3.1.3). Only a large diameter

nozzle (figure 3.1.8) and a small amount of Argon (~40 l/min, 3bar) delivered co-

Figure 3.1.9: Corner Clamp Figure 3.1.10: Un-clamped with guides

Figure 3.1.11: Burn prints in wood at 5mm Z steps, 127mmFL lens 130mm – 220mm stand-off



axially for lens cooling and protection against debris was used for this test and

throughout the laser forming studies.

The laser power is selected manually on the hand held control box, the laser

reports this power level based on a calibrated thermocouple measurement taken from

the back mirror in the cavity (totally reflecting). This reading is the laser power

leaving the cavity at the output window, however due to the number of mirrors used

to guide the beam to the workstation there is some power loss each time the beam is

turned, the mirrors and lens absorb some of the incident energy (the optics heat up

and hence have to be water cooled). To get an accurate value for the laser power

arriving at the work surface it is necessary to perform a power puck test. A power

puck is a calibrated device whereby the temperature rise in a coated black metal

block (puck like, hence the name) exposed to a laser beam is directly proportional to

the incident laser power. A number of these tests are performed at various powers

and at regular intervals (every month) to build up a calibration chart, such that for a

required incident power the power level on the laser can be found. An example of a

calibration graph for the Electrox 1.5kW workstation 2 is given in figure 3.1.12.

Another piece of hardware of note used throughout the studies was an

Agilent 34970A data acquisition and switch unit (figure 3.1.13), this unit combined

with the 34902A 16 channel high speed card and the 34901 20 channel multi-

function interface card, can record data at up to 250 channels per second (250 Hz

maximum bandwidth) in the form of AC and DC Voltage, Current and Resistance. In

addition the unit can log thermocouple output of any type (software selectable), it

Figure 3.1.12: Power offset calibration graph



has on-board a calibrated reference junction, such that no manual calibration of the

thermocouple is required. The unit can run as standalone or can be set-up by and data

sent to a PC via a RS232 serial cable.

3.1.2 Software The CNC tables are controlled by a Galil DMC 1730 ISA pc based controller.

This control card has as standard some basic software for terminal based command

line control and for part program creation and downloading, using the Galil CNC

language, an example of which (with annotations) is given in Appendix 6. Using this

language, as with any CNC language (e.g. G-code), it is possible to define movement

in both absolute (with respect to the table origin) and relative co-ordinates (with

respect to the last position) for independent and co-ordinated axis movement. In

addition it is possible to define values such as: traverse speed in each axis and vector

speed (co-ordinated movement), acceleration and de-acceleration in each axis and

for combined movement, storage of vector arrays and the setting of output ports high

or low for hardware interfaces such as shutter control. A point to note about the Galil

CNC language is that values for movement are specified in encoder counts (encoder

wheels and counters are attached to the ends of the motor shafts to control

movement), such that for the combination of encoder resolution and pitch of lead

screw on the tables the conversion factor for the X and Y stages is 1mm = 400

counts, for the Z stage 1mm = 1000 counts.

In addition to the OEM software there are dynamic link libraries (.dll) and

ActiveX drivers for custom software authoring in Microsoft Visual Studio

Figure 3.1.13: Agilent 34970A Data Acquisition unit



applications such as Visual Basic. Using Visual Basic, it is possible to create forms

or user interfaces (UI) that contain (mouse) clickable buttons, behind which can be

hidden commands (or series of commands) for the operation of the CNC system. It is

possible to create applications (.exe) to automate processes and functions that would

otherwise be impossible or may take many lines of typing using the command line

interface such as simply jogging the tables to a desired start position. One the first

applications developed (and still developing) was a control interface to simplify the

interaction with the controller and to display reported information such as current

position. The user interface to this application can be seen in figure 3.1.14

It can be seen in figure 3.1.14 that there are a number of features in this

control program added to automate and simplify the interface with the controller,

these include: single click jogging at a selectable speed of each axis (Z axis brake

needs to be released before movement of that axis, again single click on the UI), axis

homing, origin reset, position reporting (real-time), laser range finder output (real-

time), I/O status, terminal for optional command line control e.g. part-program

execution and access to Galil editor for the creation and downloading of part-

programs.

A number of other applications were developed for this work in Visual Basic,

two are outlined here. The first application is a tool for the 2D laser forming (using

the TGM, +ve bending angle) of 80x80mm coupons along a centre line (@40mm)

using the edge clamp mentioned earlier (figure 3.1.8), it employs the MEL M5 laser

Figure 3.1.14: Control Application User Interface



range finder (described earlier) to measure the bend angle per pass and output this to

file for later analysis. The basic concept of bend angle measurement (αb) using two

height readings either side of the scan line can be seen in figure 3.1.15.

A development on this basic method was the ability to take account of any

initial angle in the coupon e.g. not mounted horizontally in the clamp. As can be seen

in figure 3.1.16, this was achieved by taking a bend angle measurement (α0) prior to

A

H2 H1

αb

Tan αb = (H1-H2)/A

Figure 3.1.15: Basic bend angle measurement using two height readings

Figure 3.1.16: Improved bend angle measurement accounting for any initial angle

A

H2 H1

α0

Tan α0 = (H1-H2) / (A+A0)

A0

AH2 H1

αb

αb = C - α0 Tan C = (H1-H2-H3)/AH3 = A0 Tan α0

α0

A0

C H3

α0

1

2



processing and subtracting this from the measured angle, C, using the indicated

method. The height readings are taken by positioning the workpiece under the range

finder at each location (X&Y axes movements plus Z movement to bring the device

in range). For the 80x80mm coupons distances A = 15mm and A0 = 20mm were

used, this gives required resolutions of 1° = 0.262mm, 0.5° = 0.131mm, 0.25° =

0.0654mm, this is approaching the quoted resolution of the sensor (30µm on a mat

white surface) and likely to be the smallest angle change the system could measure

accurately. As mentioned, using the method described above, a Visual Basic

application was developed to automate the laser forming of these coupons with

online bend angle measurement and output file generation, the user interface to this

application is shown in figure 3.1.17.

As can be seen in figure 3.1.17 the application allows the user to select the

process speed, time delay between passes (minimum of 24 seconds due to the length

of time for the two point measurement cycle), number of passes, focal position

(function of beam diameter), material thickness, lens focal length (again required for

spot size selection) and the output file name. The laser power is set manually and

needs to be monitored by the user throughout the process. During the process the

program displays the current bend angle and pass count. The file generated is a

comer delimited text file (.csv) containing the pass count and corresponding bend

angle (already compensated for any initial bend angle), this file can be opened by

Excel for post processing (ploting of graphs etc.), this program uses the comers as a

distinction between each column and the carriage return for each row.

Figure 3.1.17: User Interface for the automated 2D laser forming of 80x80mm coupons



The second application of note that was developed was a tool for the surface

profiling of large formed sheets using a Co-ordinate Measuring Machine (CMM)

technique. This application again employs the laser range finder to take single point

height readings over a surface at known steps or spacing in order to build up a grid

or an array of data points with which to plot contour maps to verify geometries. The

user interface can be seen in figure 3.1.18.

The application allows the user to select a scan area (maximum of

435x435mm, the tables’ movement limit) and a scan resolution or step size in each

axis. The smaller the step size the more data points taken and hence the longer the

scan time, for a 400x200mm scan area using a 10x10mm grid (figure 3.1.18) the

scan takes approximately 45 minutes. In the interest of saving time as coarse a grid

as possible is used, for a scan area as large as 400x200mm providing the geometry is

not too abrupt, a step size of about 20mm can be used successfully, this cuts the scan

time down to 10 minutes. As with the previous application a comer delimited file is

generated containing the grid of Z height data and the corresponding X and Y co-

ordinate. This file can be again opened by Excel for post-processing, and example of

the output from the CMM is given in figure 3.1.19.

Figure 3.1.18: Co-ordinate Measuring Machine (CMM) User Interface

Figure 3.1.19: Example CMM output



3.1.3 Absorptive Coatings

Crucial to the laser forming process is the absorption by the workpiece of the

incident laser radiation, in particular by a relatively large diameter and low intensity

beam when compared to other laser processes such as laser cutting. Presented in this

section is some background on absorptivity in metals which demonstrates the need

for a coating when using infra-red laser wavelengths. Also given is some data on

known coatings and surface treatments that improve surface absorption and then the

type of coating and procedures used throughout the investigations in this thesis.

For metals laser radiation is predominantly absorbed by the free electrons in

an “electron gas”. These free electrons are free to oscillate and reradiate without

disturbing the solid atomic structure. Thus the reflectivity of metals is very high in

the waveband from visible to infrared; this can be seen in figure 3.1.20.

As a wave front arrives at a surface all of the free electrons in the surface

vibrate in phase generating an electric field 180˚ out of phase with the incoming

beam. The sum of this field will be a beam whose angle of reflection equals the

angle of incidence. This “electron gas” within the metal surface means that the

radiation is unable to penetrate metals to any significant depth, only one or two

atomic diameters. Metals are thus opaque although they appear shiny.

Figure 3.1.20: Reflectivity of various metals as a function of wavelength11



The reflection coefficient, R, for normal angles of incidence from a dielectric

or metal surface in air (n=1) may be calculated from the refractive index, n, and the

extinction coefficient, k, for that material 120:

( )( ) 22

22

11

knknR

+++−

= (3.2.1)

For an opaque material such as a metal the absorptivity, A is:

( ) 2214

1

knnA

RA

++=

−=

(3.2.2)

Typical absorption values for various metals are given in figure 3.1.21.

Reflection and Absorption of metals to laser radiation are influenced by a number of

factors:

Temperature R ∝ 1/T A ∝ T

Wavelength R ∝ λ A ∝ 1/λ

Conductivity R ∝ σ A ∝ 1/σ

Figure 3.1.21: Absorptivity of various metals as a function of wavelength at room temperature 121



Surface Roughness R ∝ 1/Ra A ∝ Ra

Intensity R ∝ 1/I0 A ∝ I0

Angle R ∝ (Complex) A ∝ (Complex)

Polarisation R ∝ (Complex) A ∝ (Complex)

Because of the high reflectivity of metals to 10.6µm CO2 laser radiation, and

the fact that reflectivity increases at the relatively low power densities involved in

laser forming (<106 W/m2), absorptive coatings are usually required when using this

laser type. A range of surface treatments and coatings can be applied. Data on a

number of coatings is available in the literature and is summarised in table 3.1.1.

Surface Type Reflectivity % Direct Diffuse Total Sand Paper roughened (1µm) 90.0 2.7 92.7 Sandblasted (19µm) 17.3 14.5 31.8 Sandblasted (50µm) 1.8 20 21.8 Oxidised 1.4 9.1 10.5 Graphite 19.1 3.6 22.7 Molybdenum Sulphide 5.5 4.5 10.0 Dispersion paint 0.9 0.9 1.8 Plaka paint 0.9 1.8 2.7

Absorptive coatings are the most common means of increasing the absorption

of 10.6µm CO2 laser radiation and are widely used in industry. The laser power is

absorbed in these thin layers and transferred to the substrate, absorption may be

increased to rates of 70-80% depending on the coating and substrate (figure 3.1.22).

Table 3.1.1: Typical values of reflectivity of various surfaces to 10.6µm radiation at normal angles of incidence 120

Figure 3.1.22: Absorption of CO2 laser light on steel at room temperature dependent on surface condition11



The coatings are sprayed or printed onto the surface to be processed. Very

often, they are manually applied because an automatic application is difficult to

perform. Consequently, their thickness may vary. Coatings, however, not only

absorb the laser energy, they also have to transport the released heat to the metal

surface, hence the thermal conductivity of the substrate will influence the overall

absorption coefficient. If the layers vary in thickness or the heat transmission to the

metal substrate is not uniform the process efficiency will be reduced. In addition,

coatings tend to burn off during processing at higher intensities, this can be seen in

figure 3.1.23.

The degradation of the coating is dependent on the interaction time and the

intensity of the laser beam. The resulting variation of absorption and heat

transmission to the workpiece might reduce the process reliability. In addition to this

drawback, the application and removal of coatings represents an additional working

step raising the costs of the process.11

The coating used throughout the studies on a number of materials presented

in this thesis was Graphite; this has a quoted absorptivity of 77.3% (table 3.1.1),

however this can vary depending on the substrate between 75-90% (figure 3.1.22).

The graphite was sprayed onto the surface manually via a hand held spray can;

uniform coverage was possible with smaller samples however this became more

difficult with the larger samples. Samples to be sprayed were first cleaned with

Acetone to remove any surface contaminants e.g. grease in order to give a good and

even adhesion of the graphite. The typical coating thickness was measured using a

Figure 3.1.23: Dependence of coupling rate of coated surfaces on interaction time and incident intensity 11



Scantron Proscan 1000 laser range finder based surface profiling device, typically

used to give surface roughness values. A sample was masked using tape and sprayed

normally, the tape was then removed and a surface height measurement was taken on

the uncoated and coated surfaces, this gave a resultant thickness of 6µm. The typical

surface appearance is given in figure 3.1.24.

As mentioned earlier the coating can degrade or burn off with increased laser

beam interaction time or multiple passes over the same track, an extreme example of

the optical effect of coating degradation can be seen on the 0.9mm Ti6Al4V sample

after 20 passes in figure 3.1.25. An empirical study into the effect of coating

degradation on achievable bend angle was conducted as part of this research and is

presented in section 4.1.

Figure 3.1.24: Graphite Coated Sample Figure 3.1.25: Example of coating degradation, Ti6Al4V, 20 Passes, 740W, 5.5mm∅, 45mm/s



3.2 2D Laser Forming

This section outlines the experiments performed and the procedures used for

investigations into the 2D laser forming of metallic components. There are a number

of self-contained investigations that provide individual results and conclusions, some

of which feed into the set-up for further studies. It is hoped that the investigations are

presented in a particular order so as to allow the reader to have all relevant and

referred to information to hand before reviewing a new section.

3.2.1 Empirical Study - Characterisation of the Laser Forming

Process

The first study presented in this thesis is an empirical 2D laser forming investigation

on a number of materials using the TGM, characterising the 2D laser forming

process. Variables investigated included; beam spot size, laser power, traverse speed,

multiple and single pass strategies, time delay between passes, bend angle rate and

coating degradation. The materials investigated were sheet mild steel CR4, Ti6Al4V,

AA1050 and AA6061 (in 3 heat treatments O, T4, T6) of various gauges. The

investigations into these materials were tailored for each one, as such details of the

experiments carried out and technical data on each material is given in the following

sections. The Electrox CO2 laser system described earlier was used throughout. The

sample size used was 80x80mm, these dimensions were used for historical reasons 29

and the coupons are large enough for the process to be considered macro scale. The

coupons were held on the work bed using an edge clamp as mentioned earlier. They

were clamped 10mm in from one edge and processed along a half way line at 40mm

from the edge (figure 3.2.1). The TGM was thought to be active throughout these

studies such that the coupons always bent towards the laser.

Figure 3.2.1: Experimental set-up for 2D laser forming characterisation



All of the coupons were cleaned with acetone and manually sprayed with

graphite (nominal thickness 6µm) prior to clamping and processing. The coupons

were processed in alternating directions (for multi-pass) using the software tool and

laser range finder described earlier (figure 3.1.17), this allowed the semi-automated

investigation of many of the variables in laser forming, more detail on the

investigations into each material is given in the following sections. The results of the

investigations are presented in chapter 4.1

3.2.1.1 Mild Steel CR4

The first investigation presented is the 2D LF of 1.5mm thick mild steel CR4 (AISI

1010), a cold rolled low carbon steel sheet. This material was used in a number of

studies throughout this thesis due to its cost, availability and the fact that it is a

common material found in virtually every manufacturing sector. This material was

purchased with a bright surface finish and laser cut to the required dimensions from a

local laser cutting job-shop. Having the material laser cut meant that any additional

residual stresses due to cutting would be minimal as large residual stresses could

influence the geometry of a formed part. Technical data on this material is given in

the following tables:

Designation According to UK

BS 1449-1 (1983)

EN 10027-1 (1999)

EN 10130 (1991)

Germany DIN 1623-T1

(1983) SAE Grade

CR4 DC01 FeP01 St 12 AISI 1010 Table 3.2.1: Material designation according to different international standards.122

Element Fe C P Mn S Wt. % 99.19 0.12 0.045 0.045 0.6

Table 3.2.2: Material composition by weight percentage of Mild Steel CR4.122

Density [kg/m3]

Young’s Modulus

[GPa]

Tensile Strength

[MPa]

Yield Strength

[MPa]

Shear Modulus

[GPa]

Bulk Modulus

[GPa]

Hardness [Vickers]

7870 205 365 305 80 140 108

Table 3.2.3: Mechanical properties of Mild Steel CR4.122



Melting Point [°C] 1515 Thermal Conductivity [W/m K] 49.8

Coefficient of Thermal Expansion [10-6/K] 20°C 12.2



Specific Heat Capacity [J/kg K] 50-100°C 448




Table 3.2.4: Thermal Properties of Mild Steel CR4.122

The first investigation performed on this material was to determine a process

map or window such that for a given incident power, laser beam spot size and

traverse speed an expected bend angle for a single pass could be known. Due to the

large amount of variable combinations from just these three, it was decided to

investigate only three beam diameters, 3mm, 5.5mm and 8mm, three power levels

per beam diameter (from 500W to 1200W at the surface) and a traverse speed range

of 10mm/s to 90mm/s (comfortable speed range for the X,Y CNC tables). It was

found from initial trails that this range of processing parameters did produce some

forming with no obvious surface damage (smaller beam diameters did however), for

this thickness, type of material and laser beam mode even the larger beam diameters

produced positive bending such that a significant thermal gradient must still be

present (hence TGM). It can be noted that a more complete range of traverse speed

data was collected, this is due to the fact that it was easier to control the traverse

speed via the Galil controller than the other variables. This is significant when

considering closed loop control of the LF process and will be discussed in a later

section.

Once a process map had been determined for each of the beam diameters

investigated, three ideal processing parameter combinations were selected (one for

each beam diameter) for a study into multiple pass LF (over the same irradiation

track; alternating direction). The bend angle and bend angle rate per pass were

analysed up to 30 passes. A further study was conducted into the effect of the time



delay in between each pass on the bend angle achieved. It was not known as to

whether the heat remaining in the sheet after each pass aided subsequent passes by

reducing the yield stress of the material or if the increase in bulk material

temperature reduced the available thermal gradient on the next pass.

3.2.1.2 Ti6Al4V

An investigation was conducted into the 2D LF of Ti-6Al-4V (Ti64) mill annealed

sheet, an aerospace alloy of titanium. Sheet thicknesses of 0.9mm, 1.4mm, 1.6mm,

2mm and 3.2mm were used. Ti64 is widely used in the aerospace sector due to its

high strength yet low density, excellent resistance to fatigue and crack propagation

and outstanding resistance to corrosion. It is the most widely used titanium alloy.

The atomic structure of titanium undergoes a transformation from a close

packed hexagonal arrangement (alpha or α phase) to a body centred cubic

arrangement (beta or β phase) at 882°C. This transformation can be considerably

modified by the addition of alloying elements to produce alloys that have all α, all β

or α + β structures. Ti64 is an alpha + beta alloy containing 6% aluminium and 4%

vanadium. The aluminium stabilises and strengthens the alpha phase, so raising the

beta-transus temperature (~980°C), as well as reducing the density of the alloy. The

vanadium is a beta stabiliser, and provides a greater amount of the more ductile beta

phase during hot working. On solution treatment high in the alpha + beta field,

followed by rapid cooling to room temperature, the beta phase transforms to a

structure which can be subsequently tempered to a fine dispersion of beta in an alpha

matrix, with consequent strengthening of the alloy. Temperatures up to 700°C are

commonly used in warm-working or forming this alloy conventionally.

This material, supplied by the industrial partners in the larger work

programme which this work formed part of (discussed earlier), was guillotined to the

desired size. Ideally a cutting method that induced less residual stresses would have

been preferred such as laser cutting or EDM, however these were not available or not

possible without additional equipment and expertise. It was felt that this material

would likely be guillotined in a manufacturing environment anyway and that LF

should be studied on the as received material and material conditions. Another factor

in this was the cost and availability of the material, in that there was no room for

error. Titanium, despite being the fourth most abundant structural metals in the



Earth’s crust and possessing excellent mechanical properties, is rarely used in

engineering due to its cost and availability. This is due to the cost and difficulty of

removing titanium from its ore, the current Kroll batch process (invented 1940) is

slow and low volume. The world’s steel industry matches the annual titanium output

in less that one hour. A new large volume, low cost production method invented at

Cambridge University and backed by Qinetiq will lead the way for the wider cost-

effective use of titanium in the future. Technical data on this material is given in the

following tables:

Designation According to UK

BS 2TA10 (1974)

IMI Titanium ASTM UNS

TA10 IMI 318 Grade 5 Titanium R65400

Table 3.2.5: Material designation according to different international standards.122

Element Ti Al V O + N Fe H Wt. % 89.44 6 4 0.25 0.3 0.0125

Table 3.2.6: Material composition by weight percentage of Ti-6Al-4V.122

Density [kg/m3]

Young’s Modulus

[GPa]

Ultimate Tensile

Strength [MPa]

Tensile Yield

Strength [MPa]

Compressive Yield

Strength [MPa]

Shear Modulus

[GPa]

Hardness [Vickers]

4430 113.8 960 880 970 44 349 Table 3.2.7: Mechanical properties of Ti-6Al-4V.122

Melting Range [°C] 1604-1660 Thermal Conductivity [W/m K] 6.7




Specific Heat Capacity [J/kg K] 526.3

Beta-Transus [°C] 980

Table 3.2.8: Thermal Properties of Ti-6Al-4V.122

As with the previous material, the first study was to determine a process map

or window such that for a given incident power, laser beam spot size, traverse speed

and sheet thickness an expected bend angle for a single pass could be known. For



each of the five material thicknesses investigated (0.9, 1.4, 1.6, 2 & 3.2mm), 3 beam

diameters (3, 5.5 & 8mm), three incident laser powers (in the range 500-1200W) and

a range of traverse speeds (10-90mm/s) were used. Once a process map had been

produced for single passes, a study was conducted into multi-pass LF on all five

gauges using selected ideal processing parameters for each beam size; selected on

the basis of significant forming per pass (~1° and above) and no obvious surface

damage. An additional study into the effect of coating degradation on bend angle rate

per pass was also conducted. A comparison was made with a sample processed with

a single sprayed graphite coating (initial coating before clamping) and samples with

a coating re-spray every 5 and 15 passes up to 30 passes. It was thought this study

would reveal the dependence on absorptive coating integrity on the fall off of bend

angle at high numbers of passes. For the thicker samples, 3.2mm gauge, parameters

that produce forming are limited, due to this a new double pass technique was

investigated; more detail on this is given in the results section (chapter 4.1).

3.2.1.3 AA 1050

An investigation was conducted into the 2D LF of 0.9mm gauge Aluminium 1050-

H14, a 1000 series 99.5% pure aluminium sheet in a H14 temper. This is a non-heat

treatable metal and so increased strength is acquired through cold rolling, the degree

of cold working is indicated by the 4 of H14, e.g. H16 would involve more cold

working than H14. This material was chosen due to its cost and availability and that

it is a common use engineering metal. The material could also be readily laser cut to

the desired dimensions. In addition the Al 1050-H14 provides data on the effect of

LF on materials with high thermal conductivities. Technical data on this material is

given in the following tables:

Designation According to UK BS UNS DIN ASTM

B491 France

BS 1B A91050 Al99.5 AA1050-H14 NF A5 Table 3.2.9: Material designation according to different international standards.122

Element Al Cu Zn Si Mn Ti Wt. % min 99.5 max 0.05 max 0.05 max 0.25 max 0.05 max 0.03

Element Fe Mg V Wt. % max 0.4 max 0.05 max 0.05

Table 3.2.10: Material composition by weight percentage of

Aluminium 1050-H14. 122



Density [kg/m3]

Young’s Modulus

[GPa]

Ultimate Tensile

Strength [MPa]

Tensile Yield

Strength [MPa]

Shear Strength

[MPa]

Shear Modulus

[GPa]

Hardness [Vickers]

2705 69 110 103 69 26 35 Table 3.2.11: Mechanical properties of Aluminium 1050-H14.122

Melting Range [°C] 646-657 Thermal Conductivity [W/m K] 227



Specific Heat Capacity [J/kg K] 900

Table 3.2.12: Thermal Properties of Aluminium 1050-H14.122

From initial LF trials on this material it was found that the TGM was only

active using smaller beam diameters (e.g. guaranteed positive bend), possibly due to

the high thermal conductivity. For this reason a 2D LF single pass bend angle study

was only conducted using a 3mm beam diameter, a larger range and number of laser

power levels (200-800W, seven powers) and a traverse speed range of 10 to 90mm/s.

This gave a useful process map from which it was possible to select ideal or useable

processing parameters. Two processing parameters were selected at two power levels

for a study into the multi-pass 2D LF of this material, again the bend angle and bend

angle rate per pass were analysed. An additional repeatability study was also

conducted.

3.2.1.4 AA 6061

The final material investigated was 1.6mm gauge AA 6061, a non-ferrous wrought

and age hardenable 6000 series aluminium alloy whose major alloying elements are

Magnesium and Silicon. Compared with ferrous alloys, for instance, stainless steels

and cast irons, outstanding specific strengths, which is defined as strength-to-weight

ratio, can be obtained due to its relatively high tensile strength and low density.

Furthermore, the corrosion resistance and workability of AA 6061 is also excellent.

For this reason, heavy-duty structures requiring good corrosion resistance can be

made of this alloy, such as street furniture, windows, automotive (e.g. brake pistons)

and marine applications (valve and valve parts). However, the amount of



precipitation, which has a significant effect on mechanical properties in different

tempers, which can be formed, is limited. This material is available in a number of

different tempers, three different tempers were investigated here, O, T4 and T6. O

denotes that the alloy is annealed and T indicates that the alloy is thermally treated to

produce stable tempers other than F (as fabricated, hot-worked, forged, cast, etc.), O

(annealed), or H (strain-hardened, cold-work). In addition, T is always followed by

one or two numbers which shows the exact type of heat treatment, and more details

of the processing of the alloy. The lowest strength temper for wrought products is

obtained by the O temper; AA 6061-O is in the softest possible condition because

the strain hardening form cold working is reduced by annealing. AA 6061 in both of

the T4 and T6 temper are solution heat treated and cold worked. The main difference

between these two tempers is that the T4 temper is naturally aged to a substantially

stable condition after solution heat treatment, but the T6 temper is artificially aged.

Generally speaking, an alloy in the T4 temper owns higher ductility and lower

strength then the same alloy in the T6 temper. The reasons for investigating the three

tempers was to determine the effect of heat treatment condition of the same alloy on

the laser forming characteristics, in addition an insight into the possible factors

influencing bend angle rate fall-off with increasing numbers of passes may be gained.

In that each of the three tempers has a different rate of strain hardening for a given

working regime, an identified factor in the process, and that the degree of fall-off

may echo or confirm this. The material was guillotined to the desired size for

processing (80x80mm). Technical data on this material is given in the following

tables:

Designation According to

ISO UNS ASTM Russia

AlMg1SiCu A96061 AA6061 AD 33 Table 3.2.13: Material designation according to different international standards.122

Element Al Cr Cu Fe Mg Mn Wt. % 98 max 0.35 max 0.4 max 0.7 max 1.2 max 0.15

Element Si Ti Zn Wt. % max 0.8 max 0.15 max 0.25

Table 3.2.14: Material composition by weight percentage of AA6061122



Temper Density [kg/m3]

Young’s Modulus

[GPa]

Ultimate Tensile

Strength [MPa]

Tensile Yield

Strength [MPa]

Shear Strength

[MPa]

Hardness [Vickers]

O 2700 69 125 55 80 40

T4 2700 69 240 145 165 75

T6 2700 69 310 275 205 107 Table 3.2.15: Mechanical properties of AA6061 in three different tempers.122

O T4 T6 Melting Range [°C] 582-652 582-652 582-652

Thermal Conductivity [W/m K] 180 154 166.9 Coefficient of Thermal Expansion

[10-6/K] 20°C 23.6 23.6 23.6 Coefficient of Thermal Expansion

[10-6/K] 250°C 25.2 25.2 25.2 Specific Heat Capacity

[J/kg K] 896 896 896

Table 3.2.16: Thermal Properties of AA 6061 in three different tempers.122

It can be noted from the above tables that there are significant differences

between the three tempers of the alloy in terms of material strength and thermal

conductivity. As these are factors identified as influencing the bend angle during LF

there is likely to be significant differences between the LF characteristics of the three

tempers.

As with the previous materials the first study was to build up a process map

of the single pass LF of the three tempers. As with the AA1050-H14 only one beam

diameter was used, 3mm, and a more complete range of laser powers (8 power levels,

200-900W) over the speed range 10-90mm/s. From the three process maps

parameters for multi-pass 2D LF were selected, such that it would be possible to

vary one of the following parameters whilst holding the others constant; laser power,

traverse speed, time delay in-between each pass, sample re-coating interval (graphite)

and material temper. It was hoped that this approach would yield an insight into the

effect of each of the variables above on the 2D LF process.



3.2.2 Thermal Analysis A number of studies are grouped into this section, which involved the analysis of the

thermal behaviour within a sample during 2D laser forming. The first is a study that

used a thermocouple technique to determine the temporal temperature cycle at a

number of single points during multi-pass LF. The second employed an Infra Red

(IR) camera to determine a whole field thermal image of the LF process during a

single pass. The final study in this section was an investigation into the effect of

forced cooling both in-process and during pass intervals on the LF process.

All of these studies were performed on 1.5mm mild steel CR4 using process

parameters determined from the empirical study. The CO2 laser system described

earlier was used throughout (section 3.1). The results are presented in chapter 4.2

3.2.2.1 Thermocouple Study

A study was conducted using a thermocouple technique into the temporal

temperatures cycles at single locations on the upper and lower surfaces of 1.5mm

mild steel CR4 during multi-pass 2D LF. Three processing parameter sets were

investigated, chosen from the empirical study; 3mm beam diameter, 760W, 55mm/s;

5.5mm beam diameter, 760W, 30mm/s; 8mm beam diameter, 760W, 20mm/s. The

thermocouple technique of temperature measurement relies on the voltage drop

across two dissimilar metals when they are placed into contact, this voltage is a

function of temperature. As only the potential difference between the two metals

need be measured and no input is required, these sensors are considered passive. The

thermocouples used in this study were welded tip ‘K Type’; this indicates that the

junction is made up of Nickel

Chromium (positive side, green wire)

and Nickel Aluminium (negative side,

white wire) and that the junction has

been welded in an inert atmosphere

to form a small bead at the end of the

sensor wire; this junction has a

reliable linear temperature range of -

270°C to 1372°C. Figure 3.2.2: Thermocouple locations

used on the 80x200mm sample



Thermocouple measurements were taken at distances of 10, 22, 34, 46, and

58mm from the scan line along the centre of 80x200mm coupons (figure 3.2.2), the

longer coupons were used to provide additional working space to attach the sensors.

The thermocouple tips or beads were located using a small punch mark and held in

place using thermo-pads (figure 3.2.3); these thermo-pads are heat resistant to

temperatures up to approximately 250°C and provide a method of re-using the

thermocouples without cutting the welded tip off after the use of adhesives. However,

this does limit how close the thermocouples can be placed to the scan line and how

close each thermocouple can be placed to each other; the thermo-pads are 12mm

wide, hence the distribution in figure 3.2.2. It was thought that the temperature data

at these locations should still reveal significant results. In addition placing the

thermocouples closer to the scan line would be difficult even with an adhesive

without destroying the lead wire or the junction during laser processing.

The 80x200mm 1.5mm gauge coupons were clamped 30mm in from one end

and processed 65mm in from the same end; the thermocouples were attached to the

free end of the plate (figure 3.2.4). The graphite coating was applied after clamping.

Figure 3.2.3: Thermocouple attachment using Thermo-pads.

Figure 3.2.4: Thermocouple study experimental set-up.



The data was recorded using the Agilent 34970A data logger described

earlier (figure 3.1.13); this device has a reference junction built in for auto-

calibration of the voltage response from the thermocouple and can record at a

maximum rate of 250 channels per second. For each of the three process parameter

sets six alternating passes at 60 second and 24 second intervals on separate plates

were recorded plus the subsequent cooling time. In addition a further study was

conducted up to 10 passes for one of the identified processing parameter sets at a

single location to ascertain the effect on peak temperature as the number of passes

increases still further.

3.2.2.2 Thermal Imaging (IR) Study

A Thermovision® 880 infrared detection system was utilised for the thermal

imaging of a sample during laser forming. The laser forming of edge clamped

graphite coated 80x80mm coupons of 1.5mm mild steel CR4 at a number of process

input parameters (CO2 laser) were investigated; selected from the empirical study.

The detector was positioned 0.5m away from the centre of the samples, at 50º to

their plane. It was mounted on a triangular frame structure that was bolted to the X-Y

work bed (Figure 3.2.5).

50º

0.5 m

LN2 Cooled Infrared Detector

Sample

Triangular Support Structure

X-Y Work Bed

20º Field of focus Lens

Figure 3.2.5: The Thermovision® 880 Infrared Detector Set-up



This system (loaned from the EPSRC), produced by Agema Infrared Systems

AB, consisted of a temperature read-out computer (TRC), a thermal image computer

(TIC–8000) using a 64K IBM PC (fairly antiquated), and a Liquid Nitrogen (LN2)

cooled infrared detector (scanner). Supplied were optional infrared filters (selected

externally) and three detachable lenses that gave a 7º, 12º or 20º field of view. Some

technical data is shown in Tables 3.2.17 & 3.2.18.

Infrared Detector

MCT, Liquid Nitrogen (LN2)

Cooled

Temperature Measurement

Range -20º to 1500º C

Apertures 3 externally selectable Lenses 7º , 12º and 20º

Field of view

Spectral Response

Broadband antireflective

coating 8-12µm

Sensitivity NETD

0.07ºC at a 30ºC object temperature

Resolution (elements/line) 175

System Operational

Ambient Temperature

-15ºC to 55ºC

MCT – Mercury Cadmium Telluride

Lens Minimum Focus

Focal Distance Geo. Res.*

7º 1.2 m 110 mm 0.7 mrad 12º 0.8 m 65 mm 1.2 mrad 20º 0.5 m 38 mm 2.0 mrad

*Geometric Resolution measured at slit response at 50% contrast

For this investigation the 20º field-of-view lens was used. A part program

was written to move the X-Y work bed back-and-forth along a straight line. This set-

up enabled the infrared detector to move with the laser beam across its scanning path.

Typical data output from the thermal imager can be seen in figure 3.2.6.

Table 3.2.17 Technical Data for the Thermovision® 880 Infrared Detector

Table 3.2.18 Lens Specifications for the Infrared Detector

Figure 3.2.6: Optical & IR images of a graphite coated sample



The parameter settings chosen for the infrared detection system were (had to be

specified):

(i) Temperature measurement range between 40º C and 1500ºC

(ii) Aperture of 2.0 chosen on a 20º field-of-view lens

(iii) Data Capture frequency of 25 Hz.

(iv) Manual start/ stop

(v) Emissivity set at 0.95 (typical for air)

The thermal scanning was started one second prior to beginning the laser processing

of the sample and stopped approximately ten seconds after the laser scan.

The collected data was transferred to the software package Irwin OLE V2,

for thermal analysis and image editing to be carried out. This involved selecting start,

middle and end thermal images taken during the laser scanning process across the

mild steel coupons, and three images showing the cooling of the coupons, one

directly after cessation of laser processing and two at later time intervals. An

emissivity of 0.6, and temperature range from 40ºC to 1500ºC, were chosen for the

thermal images in the post-processor, these parameter values being the emissivity of

graphite, and working range for the thermal scanner, used in this project. An issue

arose from initial tests with the system due to the large differences in emissivity

between the graphite and the metal substrate, a test using a coupon sprayed with

graphite along the scan line only (shown in figure 3.2.7) revealed the need to start

with a homogenous surface. This identifies a possible problem with using a thermal

imaging technique in conjunction with an absorptive coating, in that as the coating

degrades the emissivity may change in the irradiated zone.

Figure 3.2.7: Optical & IR images of a masked graphite coated sample showing differences in emissivity causing false readings



Data for the temperature field across the mild steel coupons, during the laser

processing and upon cooling, were obtained from the thermal images using data

acquisition lines; a technique used in IRWIN to extract tabular data. This data was

transferred to the spreadsheet software Microsoft™ Excel to enable graphical

analysis to be undertaken.

3.2.2.3 Forced Cooling Study

A study was conducted into the use of forced cooling in combination with the laser

forming process. Forced cooling should lead to improvements in processing time by

reducing the time interval required to prevent material damage in between each pass

in a multi-pass strategy. However, the effect of forced cooling on process efficiency

is less understood. A study was conducted on 80x80mm 1.5mm mild steel coupons

using three process parameter combinations; 3mm beam diameter, 760W, 55mm/s;

5.5mm beam diameter, 760W, 30mm/s; 8mm beam diameter, 760W, 20mm/s; the

time interval between passes was 40 seconds. The forced cooling was provided by a

compressed air jet (3 bar) on the underside of the coupons (figure 3.2.8), the air jet

was left on throughout processing and during cooling.

The first part of the study was to ascertain the effectiveness of the air jet at

cooling the plate. This was achieved using a thermocouple technique; thermocouples

were attached to the upper surface of the coupons at 10mm and 22mm from the scan

line, they were then processed with 4 passes with and without cooling and the

temperature data recorded. A study was then performed on the effect of cooling on

Figure 3.2.8: Forced cooling experimental set-up



the achievable bend angle for the three processing parameter combinations using the

laser range finder system to record the bend angle after each pass up to 30 passes.

3.2.3 Displacement / Time Analysis

An investigation was conducted into the displacement or bend angle development of

a coupon with respect to time. The laser forming of 80x200x1.5mm mild steel CR4

coupons were investigated using the process parameter combinations, outlined

earlier, of; 3mm beam diameter, 760W, 55mm/s; 5.5mm beam diameter, 760W,

30mm/s; 8mm beam diameter, 760W, 20mm/s; the time interval between passes was

60 seconds up to 6 passes. The displacement of the coupons was measured with a

second laser range finder (MEL M1L/100) positioned over the free end of the

clamped coupon, 20mm in from the edge (figure 3.2.9), the longer coupons act as an

amplifier of the displacement. The laser displacement sensor had a range of 100mm

and an output rate of 100Hz; more detail on this sensor is given in appendix 5. The

0-10 Volt output from the sensor was recorded using an Agilent data logger; the

output range corresponds to the 0-100mm sensor range. Graphite coating was used

on the sensor measurement position to reduce surface reflections impairing the result.

Although some early work by Vollertsen23 had included bend angle / time

analysis in order to confirm the TGM theory (figure 2.6.2), investigations into the

more subtle aspects of 2D LF have not been conducted using this technique. In

Figure 3.2.9: Displacement / Time analysis experimental set-up



particular the effect of using larger beam diameters and a multiple pass strategy on

the bend angle development with respect to time was a new development.

3.2.4 Strain Gauge Analysis 123, 124

This investigation aims to complement the understanding of two-dimensional laser

forming, offering an insight into the mechanical behaviour of a part during the

process using a strain gauge analysis technique. The investigation was performed on

80x200mm 1.5mm gauge Mild Steel CR4 (AISI 1010) coupons using a high power

CO2 laser source and 3 axis beam manipulation, as described in section 3.1.1, using

the temperature gradient mechanism (TGM)11 throughout. The processing parameters

used were selected from the results of an empirical study into this material, (data

presented in chapter 4.1.1); 760W, 5.5mm beam diameter and a processing speed of

30mm/s. Included in the investigation is strain gauge analysis of the transverse and

longitudinal localised strains at a number of locations on the top and bottom surfaces

of the 80x200mm coupons during alternating direction multi-pass laser forming. The

longer coupons used provided sufficient area to attach the strain gauges. The

Graphite coated coupons were clamped 30mm from the scan line along one edge

during processing using an aluminium clamp.

The strain gauge analysis was performed using polyimide backed 5mm long

uniaxial foil gauges, which have a temperature range of -30 to 180ºC, in this range

the gauge factor K is constant. The gauges were attached to the keyed and cleaned

surface of the coupons using a cyanoacrylate adhesive (CN). Thermocouple analysis

was performed in order to determine ideal gauge location so as to remain within the

Figure 3.2.10: Quarter Bridge Circuit



operating range; this is presented in chapter 4.2.1. The gauges selected were

temperature compensated for the thermal expansion of Mild Steel. The gauges were

each incorporated into a balanced and zeroed quarter Wheatstone bridge circuit

(Figure 3.2.10) the output of which was amplified by a four channel bridge amplifier

and recorded using the Agilent data logger described earlier, thus allowing four

single strain gauges to be monitored at any one time (Figure 3.2.11).

Although the gauges could be arranged to give the net bending strains or

temperature compensation in half or full bridge configurations, it was thought that

due to the asymmetry of the laser forming process the determination of the average

localised strains would yield more meaningful results, hence a quarter bridge

configuration.

3.2.4.1 Transverse Strain

The transverse component of strain with respect to the scan line is orthogonal or at

90° to the scan direction, therefore the orientation of the uni-axial strain gauges was

also orthogonal to the scan line for this part of the investigation. The thermocouple

analysis (section 4.3.1) at the chosen processing parameters (760W, 5.5mm ∅

30mm/s, 60s interval) after six passes revealed that the gauges could be placed as

close as 10mm from the scan line and still be within operating temperature range. It

was decided that locations near and far from the scan line would yield a better

picture of the transient strain behaviour of the component, thus locations at 10 and

Figure 3.2.11: Experimental Set-Up for Strain Gauge Analysis



46mm from the scan line, close to the centreline and 10mm from the edges of the

upper and lower surfaces, were used. This can be seen in figure 3.2.12.

As the output from only four gauges could be logged at any one time, a

separate coupon was used for each configuration. An assumption was made that the

process conditions were identical for each sample as each coupon was laser cut from

the same mild steel sheet. A prepared sample with four gauges attached can be seen

in figure 3.2.13.

Figure 3.2.12: Schematic Showing Strain Gauge Locations for the Transverse Strain Study

Figure 3.2.13: Strain Gauges Attached to a Coupon



3.2.4.2 Longitudinal Strain

The longitudinal component of strain with respect to the scan line is in the same

direction as the scan direction (parallel), therefore the orientation of the uni-axial

strain gauges was also parallel to the scan line for this part of the investigation. All

of the experimental arrangements were identical for this study to the previous

transverse strain study. A schematic of the locations and orientations of the strain

gauges for this part of the study is shown in figure 3.2.14.

The results from both these investigations on the transverse and longitudinal

localised strains are presented in chapter 4.4.

3.2.6 Finite Element Analysis

A finite element method was used to investigate the 2D LF process. Finite Element

Analysis (FEA) is a numerical method which provides solutions to problems that

would otherwise be difficult to obtain. It was first developed in 1943 by R. Courant,

who utilized the Ritz method of numerical analysis and minimization of variational

calculus to obtain approximate solutions to vibration systems.

Figure 3.2.14: Schematic Showing Strain Gauge Locations for the Longitudinal Strain Study



By the early 70's, FEA was limited to expensive mainframe computers

generally owned by the aeronautics, automotive, defence, and nuclear industries.

Since the rapid decline in the cost of computers and the phenomenal increase in

computing power, FEA has been developed to an incredible precision. Present day

computers are now able to produce accurate results for all kinds of parameters.

FEA consists of a computer model of a material or design that is stressed and

analyzed for specific results. It is used in new product design, and existing product

refinement.

FEA uses a complex system of points called nodes which make a grid called

a mesh. This mesh is programmed to contain the material and structural properties

which define how the structure will react to certain loading conditions. Nodes are

assigned at a certain density throughout the material depending on the anticipated

stress levels of a particular area. Regions which will receive large amounts of stress

usually have a higher node density than those which experience little or no stress.

Points of interest may consist of: fracture point of previously tested material, fillets,

corners, complex detail, and high stress areas. The mesh acts like a spider web in that

from each node, there extends a mesh element to each of the adjacent nodes. This

web of vectors is what carries the material properties to the object via the resultant

stiffness matrix, creating many elements.

There are multiple loading conditions which may be applied to an FE model for

analysis:

• Point, pressure, thermal, gravity, and centrifugal static loads

• Thermal loads from the solution of a heat transfer analysis

• Enforced displacements

• Heat flux and convection

• Point, pressure and gravity dynamic loads

There are numerous FEA software packages available ranging from low cost basic

capability up to very flexible and powerful suites of software with a large price tag.

The software used in this investigation was ABAQUS ver.5.8 from Abaqus, Inc., this

is an example of the later type.

ABAQUS© is the world's leading software for advanced finite element

analysis. The ABAQUS software suite has an unsurpassed reputation for technology,

quality and reliability. It has been adopted by many major corporations across all

engineering disciplines as an integral part of their design process. ABAQUS,



provides superior solutions for linear, non-linear, explicit and multi-body dynamics

problems to deliver a unified finite element analysis environment. The ABAQUS

suite consists of three core products - ABAQUS/Standard, ABAQUS/Explicit and

ABAQUS/CAE.

ABAQUS/Standard provides ABAQUS solver technology to solve traditional

implicit finite element analyses, such as static, dynamic and thermal, all powered

with the widest range of contact and nonlinear material options.

ABAQUS/Explicit provides ABAQUS solver technology focused on

transient dynamics and quasi-static analyses using an explicit approach appropriate

in many applications such as drop test, crushing and many manufacturing processes.

ABAQUS/CAE provides a complete modelling and visualization

environment for ABAQUS solvers, with direct access to CAD models, advanced

meshing and visualization. This is a recent addition to Abaqus and is only available

in versions 6.2 and above.

The program used (Version 5.8 and below) is executed via the command line,

calling a pre-written text file (.inp) which contains the model data, boundary

conditions, loading conditions and step/increment (analysis) data, all defined in the

Abaqus specific syntax. An example of one of the input files used in this

investigation is given in appendix 2.

Reported in this thesis is the development of an FEA model for the single

pass laser forming of 80x80x1.5mm Mild Steel CR4 coupons using a CO2 laser

source and edge clamped boundary conditions. The process parameters investigated

were those obtained from the empirical study; 3mm beam diameter 760W, 55mm/s;

5.5mm beam diameter, 760W, 30mm/s; 8mm beam diameter, 760W, 20mm/s. The

model was developed to ascertain peak temperatures, thermal behaviour, transient

stress/strain conditions, residual stress/strains and displacements during and after

laser forming. The models developed were run in two parts, a purely thermal

analysis which then fed into a coupled thermo-mechanical analysis. The thermal

analysis ran in approximately 2 hours, however the coupled thermo-mechanical

analysis ran in approximately 48 hours. The run time of an FEA model is dependent

on the numbers of nodes and elements used and the complexity of the problem, key

to a usable model is a balance between acceptable run times and complexity of the

mesh in order to give accurate data.



3.2.6 Metallurgical Study

A metallurgical investigation was conducted on laser formed 1.5mm mild steel CR4

and 1.6mm AA6061 in three different tempers, O, T4 and T6, to ascertain the effect

of LF on the structure and mechanical properties of the materials. Optical

microscopy, Vickers micro-hardness testing and section thickening (only for the

AA6061) were investigated. Samples generated by the empirical multi-pass

investigation at ideal processing parameters were used for this study. Samples as

received (0 passes), and with 1, 2, 5, 10, 20 & 30 passes were investigated.

For the mild steel CR4 the following process parameters were considered:

3mm beam diameter 760W, 55mm/s; 5.5mm beam diameter, 760W, 30mm/s; 8mm

beam diameter, 760W, 20mm/s; 60 second intervals between passes. For the

AA6061 the process parameters considered can be seen in table 3.2.19.

Sample No. Number ofScans (N)

Power (W)

Velocity (mm/s)

Process Specifications

10, 20, 30 0 500 55 11, 21, 31 2 500 55 12, 22, 32 5 500 55 13, 23, 33 10 500 55 14, 24, 34 20 500 55 15, 25, 35 30 500 55 16, 26, 36 30 400 55 17, 27, 37 30 600 55 18, 28, 38 30 500 45 19, 29, 39 30 500 65

Beam Diameter = 3mm

Graphite Coating

Cooled by

Convection Cooling to the Surrounding

Air

10 ~ 19: AA 6061 O, 20 ~ 29: AA 6061 T4, 30 ~ 39: AA 6061 T6

After processing all the samples were cleaned with acetone solution in order

to remove the graphite coating and then they were cut into strips about 10 mm in

width. The cutting direction was perpendicular to the bent line. The laser processed

area of the samples was kept in the centre of the strips with the lateral sides

approximately 10 mm in length each.

The samples were marked with their own number by using an electrical

engraver and then they were hot mounted with a metal clip in a thermosetting

polymer (Phenolic Mounting Compound) using a Buehler/Mataserv Pneumet

Table 3.2.19: AA6061 Samples considered in study



mounting press. The hot mounting process took about 15 minutes for each sample

and the operation temperature was approximately 150ºC.

After mounting, the specimens were ground using a Struers LabPol-21

grinding machine with water-lubricated silicon carbide (SiC) grinding papers. A 600

grit SiC grinding paper was used as the first stage of grinding in order to remove all

polymer left on the samples fast. However, the quality of the specimen surface was

not satisfactory for photomicrography, particularly on the AA6061 samples, due to

too many scratches left on the surface. Therefore, in order to obtain a fine quality of

specimen, the 1200 and 2500 grinding papers on a Metaserv 2000 Grinder/ Polisher

machine with plenty of water were used. In addition, a polishing cloth with an oil-

based lubricant with 6 and 1 micro diamond particles was used to remove the

grinding scratches. Liquid soap and plenty of water were used to wash the specimens

between the two stages of polishing. The grinding and polishing were done manually;

the specimens were ground and polished for approximately 2 minutes at each stage.

Finally, the specimens were washed well, cleaned with ethanol and dried in hot air.

The samples were then etched. For the mild steel CR4 the etchant used was

“NITAL” which contains 2% HNO3 (nitric acid) and 98% ethanol, the etching time

was approximately 1 minute, until the surface visibly tarnished. For the AA6061

Keller’s reagent was the suggested etchant in order to obtain a clear grain size and

shape through grain contrast. This Keller’s reagent comprises a mixture of 2 ml

hydrofluoric acid (HF, 48%), 3ml of hydrochloric acid (HCL, concentrated), 5 ml of

nitric acid (HNO3) and 190 ml of water. Extra safety precautions must be taken when

using HF acid, this mixture must be mixed in a fume extraction chamber and in a

plastic beaker, HF acid will etch glass (and destroy bone). During etching double

rubber gloves, rubber apron, facial shielding and full face mask were used. The

etching time was 8 to 15 seconds. After etching the samples were washed in stream

of warm water and dried in hot air.

A Leitz Wetzlar optical microscope was used to observe the microstructure

of the etched samples. The magnification lens, including eyepiece and objective,

used were 40, 128, 256, 500, 640, and 1280 times. Both the Leitz Wetzlar optical

microscope and a JVC colour video camera were used with GrandCAM, an image

capture software package, to capture the images of the samples. A comparison of the

section thicknesses was also made from the optical microscopy images and a

measurement using a digital Vernier calliper.



For the hardness tests a Vickers Micro-Hardness Tester HVS-1000 was used.

For this system the load control, including loading duration and releasing were fully

automatic; in addition the Vickers hardness value is calculated automatically from

user selection of the indent dimensions through an inbuilt microscope. For the

AA6061, eighteen hardness values were obtained per sample in the top, middle and

bottom of the laser HAZ cross-section (figure 3.2.15, mounting clip included in the

schematic) and an average value obtained. The load on the diamond indenter used

for measurements was 9.806 N (1kgf) and the load duration was 5 seconds.

For the mild steel CR4, five hardness values were obtained in the top, middle

and bottom of the HAZ cross-section (figure 3.16). The load force was set to be

9.806 N and the loading time 10 seconds.

The results of the metallurgical study are presented in chapter 4.6.

Figure 3.2.15: Locations for Hardness tests in the AA6061 study

Figure 3.2.16: Locations for Hardness tests in the mild steel CR4 study



3.2.7 Closed Loop Control 125

Due to the inherent variability in the mechanical properties of metallic components,

such as an unknown residual stress history and non-linear bend angle rate fall off

with increasing number of passes, there is variability in the laser forming

characteristics between any two identical samples. There is no guarantee of

repeatability for given process parameters for an open loop set-up. In order to

demonstrate that laser forming can be used to produce repeatable accurate bends a

system is presented in this thesis for the closed loop controlled 2D laser forming of

80x80mm 0.9mm aluminium 1050-H14 and 1.5mm mild steel CR4 coupons. The

system uses the laser range finder system described earlier to provide bend angle

feedback per pass to custom written control software, the user interface of which can

be seen in figure 3.2.17. In order to set up a control loop for any material the process

window has to be found, this was determined for these materials in the empirical

study (chapter 4.1), bend angle data was found at a number of beam diameters at

various laser powers and processing speeds.

From a control point of view it is necessary not only to monitor the current

bend angle but to also control the bend angle rate or how many degrees per pass are

added, this is to ensure that an overshoot doesn’t occur for a given set of energy

parameters. It was found from the empirical study that by the selection of a

processing speed for a given power and spot size it could be possible to reduce the

bend angle rate as the required forming target approaches. For this purpose it is

necessary to select processing parameters that give a large range of bend angles for a

series of speeds within the range of the CNC tables.

Figure 3.2.17: Software User Interface for closed loop 2D laser forming



The control software was developed in Visual Basic to process a coupon

initially at optimum parameters, monitor the bend angle, compare it to the desired

angle and make a prediction of the processing speed for the next pass. It was found

that the traverse speed is the more straightforward and responsive variable to control

as opposed to laser power or beam spot size. As the desired angle approached the

software increases the processing speed per pass according to the determined

calibration graph, so as to reduce the bend angle rate in order to slowly move

towards the desired angle and not overshoot. The user interface for the control

software can be seen in figure 3.2.17; the user can select a target bend angle, the

ideal forming processing parameters and the output file, a maximum number of

passes is also given in order for the control loop to have a way of exiting if the target

angle cannot be achieved. The results of initial trails and final output for the

development of a reliable closed loop control system for these two materials are

presented in chapter 4.7.

3.2.8 Thick Section and Large Area 2D Forming for Ship Building 22

To be relevant to the ship building industry, particularly for the fabrication of hull

components, the laser forming process must be capable of thick section large scale

processing. A number of studies have been made to this end in both 2D and 3D laser

forming of thick section materials 4, 103-106, these studies and others have

demonstrated the potential of the LF process to produce accurately repeatable

geometries.

In this thesis a study is presented on thick section 2D laser forming of mild

steel in order to investigate the factors influencing a scaling of known scan strategies

for thinner section materials, the results of this study are reported here. The study

was conducted on 5 mm thick mild steel using three different laser systems. An

initial study was conducted on an Electrox 1.5kW CO2 laser system, wavelength

10.6µm, run in continuous wave mode, described earlier (figure 3.2.18). A second

study was conducted on an 8kW Ferranti CO2 laser and a 0.9x1.5m Wadkin CNC

table (figure 3.2.19). A third study was conducted at the Lairdside Laser engineering

Centre (LLEC) on a large 5 Axis Laserdyne 890 beam delivery system, employing a

3kW PRC CO2 laser (figure 3.2.20).



The sample dimensions used for the

initial study were 360x190x5mm, the samples

were sprayed with graphite in order to increase

the absorption of the 10.6µm radiation, not as

necessary for shorter wavelengths. For the other

studies the sample dimensions were

800x400x5mm. An attempt was made to

reproduce the work on part-cylinders using

thinner section material 29. For a part-cylinder

the scan strategy is relatively simple, a series of

straight line multi-pass bends across the longer

axis will produce the desired geometry (figure

3.2.18). For previous studies using 1 to 1.5mm

mild steel, titanium alloy and aluminium alloy sheet 29 (and chapter 4.1) a single pass

strategy was used per line, i.e. after each single direction pass the bend angle was

measured and the plate was allowed to cool before the next pass in the opposite

direction was made. However as the section thickness and hence material strength

increases, more energy input is required to achieve the same forming result and if

power availability is limited then thick section forming can be difficult. In order to

address this issue a ‘Double Pass’ technique was developed initially for the laser

forming of thick section high strength Titanium alloy (presented in chapter 4.1.2),

however it was found to be very effective for thick section Steels. The technique

involves a scan strategy of a pass in one direction followed immediately by a return

Figure 3.2.18: Initial study Set-up

Figure 3.2.19: Ferranti 8kW CO2 laser, 0.9x1.5m table, 800x400mm sample

Figure 3.2.20: 5 Axis beam delivery system



pass in the opposite direction; the plate is allowed to cool after each double pass

(forced cooled by air jet to decrease process time). The concept behind this strategy

is that, providing the material surface isn’t damaged on the second pass, the

additional energy input per pass is essentially akin to processing with a much higher

laser power (factor increase dependent on overlapping interaction times); this was

confirmed by thermocouple data. Another factor in this technique is that on the

second pass the heat retained in the irradiated area from the first pass could serve to

produce additional forming by reducing the temperature dependent flow or yield

stress of the material, in that a hot plate is easier to form than a cold one. To produce

the part-cylinder one end of the plate was fixed to the work bed by a bar (figures

3.2.18 – 3.2.20), the laser scans were started at the free end and worked towards the

clamped end, this ensured that the plate would be flat and at the same height for the

next line. The plates were forced cooled by a compressed air jet on the bottom

surface.

For the smaller samples it was possible to confirm the geometry formed

using the CMM software system described earlier. For the larger samples on the

larger workstations (figures 3.2.19 & 3.2.20), a CMM system was not available so

the formed geometries were confirmed by manually taking Z height measurements at

10 and 20mm steps along the edges of the plates and inferring the geometry in the

centre.

Studies were conducted into the 2D LF of part-cylinders along the shorter

axis and the longer axis of the larger plates using a number of laser powers, pass

numbers and step distances. Bends along the longer axis, 800mm long, would

demonstrate the potential of LF for larger scale applications, bends of this length

have not been reported in the literature. A thermocouple study was also conducted on

one of the plates to confirm the double pass strategy and to ascertain the effect of

laser line heating on the rest of the plate. The results from this study on thick section

forming are presented in chapter 4.8.



3.2.9 Laser Forming of Novel Materials – Metal Laminate

Composite (MLC) Materials 125, 126

The application reported in this work demonstrates how the laser forming process

can be used to form recently developed high strength metal laminate composite

materials. These materials due to their construction and high strength are difficult to

form once constructed using conventional techniques. Metal Laminate Composite

materials (MLC), sometimes referred to as Fibre Metal Laminates (FML), are of

particular interest to the aerospace industry, were the high strength yet lightweight

construction of parts made with these materials offers significant weight reductions

and hence a reduction in operational costs of new large commercial aircraft such as

the Airbus A380 and the proposed ultra-efficient Boeing 7E7. Military applications

are also being considered such as the Joint Strike Fighter (JSF) program.

In the A380 an FML called GLARE (Glass Reinforced Aluminium

Laminate), supplied by Fokker Aerostructures, is to be used in the construction of

the outer skin panels for the upper fuselage at the front and rear of the plane. As

mentioned, it is much lighter than the fully solid monolithic conventional materials

and represents a weight saving of about 500 - 800kg in the construction. The metallic

layers of the panels are pre-formed prior to lay-up and curing with the composite. In

a move away from the traditional rivet-based fixturing, the GLARE panels are

spliced and bonded seamlessly with no break in the fibre reinforcements and the

stringers are adhesively bonded to the panel surface. The required openings for

windows and doors are milled out of the formed panel; a fusion process is not

possible due to the presence of the composite.

Other industries where these materials are of interest include automotive, in

particular the high performance sports car and racing sectors such as formula one. A

more recent application under investigation for these materials is in the construction

of street furniture (e.g. litter bins) and airline cargo containers utilising their

excellent blast resistance capabilities to save lives in the event of terrorism.



3.2.9.1 Materials

The materials investigated in this study are composite laminates or layered structures.

MLC or FML consist of thin sheets of aluminium bonded and alternated with thin

sheets of traditional composite (figure 3.2.21). The first FML was ARALL (Arimid

Reinforced ALuminium Laminates) developed at the Delft University of

Technology, a combination of aluminium and aramid/epoxy. Although the material

showed promise, adoption by the aerospace industry for which it was developed was

slow. With the development of GLARE (GLAss REinforced), an aluminium glass

fibre laminate, a commercial break through came when Airbus decided to use the

material on its 650 seat A380. GLARE was intended to be an alternative to

aluminium in aircraft structures. Research has shown it has benefits over both

aluminium and fibreglass composites, especially in fatigue and impact. By the

selection of different types of laminate components together with the possibility to

vary the volume fraction of the composite and fibre orientation, a wide range of

material properties of the resultant product can be produced.126 The MLC materials

used in this investigation have been developed in the Materials Science Division of

the University of Liverpool, work is ongoing in this department to develop materials

(or material combinations) that require much shorter manufacturing time and have

superior impact and blast resistance. Four types of materials were investigated of

different lay-ups or construction, a schematic of the construction, lay-ups and

nomenclature for the MLC used is shown in figure 3.2.21.

1/1 2/1

3/2 4/3

0.3mm Al2024/2025

Fibre Reinforced Composite

Figure 3.2.21: Schematic of the MLC lay-ups used in the investigation



The first material used was a laminate of 0.3mm Aluminium 2024 alloy and a

glass reinforced Polyamide. This material is a thermoplastic which has a melting

point of approximately 280ºC. The second material was a laminate of 0.3 mm

Aluminium 2024 and a self-reinforced Polypropylene, this material is also a

thermoplastic which has a lower melting point of approximately 165ºC. The third

material was a laminate of 0.3mm Aluminium 2024 and a glass fibre reinforced

Polypropylene, this has a similar melting point to the previous material. Unlike the

other materials used where the fibre orientations are orthogonal and bi-directional (as

supplied by the manufacturer), it was possible with this last material to set the fibre

orientations as bi-directional (standard) or in a single direction so as to investigate

the affect of material anisotropy.

A fourth GLARE thermosetting type material was also investigated after

work was completed on the previous three material combinations. This material was

a 2/1 lay-up combination of 0.9mm Al2024 and glass fibre reinforced epoxy. The

reasons for investigating this lay up combination will be discussed in the results

section.

The materials are manufactured using Teflon-coated steel moulds where the

laminates are laid-up, using a polypropylene interlayer to adhere the pre-preg

composite material to the metallic layers. The moulds are then heated and a pressure

applied to the upper surface to melt the bonding layers. Larger panels can be

processed in large autoclaves. A mounted and polished section of a 4/3 Polyamide

based FML is shown in figure 3.2.22.

MLC materials can be formed conventionally into components, however due

to their high strength and laminated construction difficulties can arise such as a

Figure 3.2.22: 4/3 Polyamide based FML as Received Section



limited minimum bend radius and the need for a metal layer within the laminate in

order to deform it. The materials have considerable anisotropy and its axes change

direction as any forming operation proceeds also in laminated parts the layers can

slip over one another. Another factor is the large residual stresses that remain

between each layer after manufacturing, this can produce considerable distortion in a

formed part.

It is hoped that this study will demonstrate the potential of laser forming as a

manufacturing tool for MLC materials, either as a means of direct manufacture or a

means of alignment and distortion removal.

3.2.9.2 Experimental

The investigation consists of an initial feasibility study to determine whether or not

laser forming could be used to bend MLC materials, a more detailed study of the

laser forming characteristics and an investigation into the implications of laser

forming on the material’s structure, including thermocouple analysis. A coupon size

of 80x40mm was used throughout the characterisation studies and the bend line was

always across the shortest width (i.e. a 40mm long bend) in the middle of the edge

clamped coupon. Energy parameters consistent with the temperature gradient

mechanism (TGM) were used throughout this study, this mechanism produces a

bend towards the laser.11 An additional investigation was also performed to

demonstrate the capability of the process to form larger more complex structures

from 200x100mm coupons of MLCs.

The laser forming rig discussed earlier in section 3.1 was used for the

forming procedure, employing an Electrox 1.5 kW CO2 laser, wavelength 10.6µm,

run in continuous wave mode. The MLC coupons were guillotined cut to the correct

dimensions and the upper surfaces were cleaned with acetone. They were then

sprayed with graphite in order to increase the absorption of the 10.6µm radiation.

The coupons were clamped 30mm from the scan line along one edge during

processing using an aluminium clamp as can be seen in figure 3.2.23. The

thermocouple analysis of a multi-pass strategy using a single Al foil was performed

using K type thermocouples, which have a range of -200 to 1370ºC. The

thermocouples were attached to the bottom surface of the foil on the centreline of the



laser irradiation line using adhesive pads. An Agilent 34970A data acquisition unit,

discussed earlier, was used to record the temperature data from the thermocouples. In

order to verify the material integrity the laser formed samples were band sawn,

mounted, polished and photographed using an optical microscope. The materials

could not be etched due to the presence of the composite.

3.2.10 Application Example – Aero Engine Strut

In order to prove the manufacturing capabilities of the LF process at attempt was

made to replicate an actual aerospace component. Discussed earlier (chapter 2.7.2),

Roll-Royce, an industrial partner to the work programme this thesis forms part of,

identified an ‘A’ frame strut component from their Trent 700 Aero engine as an ideal

candidate for laser forming (figure 3.2.24). The strut is made from 3.2mm gauge

Ti6Al4V, 574mm long and 106mm wide (when formed), and is currently

manufactured using a hot creep forming process (discussed earlier also). Although

the part is 3D it was proposed that it could be formed using a 2D LF straight line

approach. It can be seen in figures 3.2.24 & 3.2.25, CAD drawings of the part, that

the component is a U channel, two of these channels are welded together to form a

hollow strut. The strut is welded onto another strut via an end component to form the

‘A’ frame; the completed frame can be seen in figure 2.7.3. This component is a

structural component in the engine.

Figure 3.2.23: MLC Experimental Set-up



Figure 3.2.24: CAD drawing of an RR Trent 700 Aero Engine ‘A’ frame strut component.

Figure 3.2.25: CAD drawing of an RR Trent 700 Aero Engine ‘A’ frame strut component (magnified).

An initial attempt to produce the part was made without the aid of the above

drawings. A flat sheet of Mild Steel CR4, 400x200x1.5mm, was used to demonstrate

that a part of similar length could be formed and that the whole enclosure could be

produced needing only one welded seam. The CO2 laser system described earlier

was used for the demonstration, a graphite coating was used.



A second attempt was made to produce the section geometry of the strut from

200x100x1.6mm Ti6Al4V sheet clamped at the centre (figure 3.2.26). A double

pass strategy (parameters selected from the empirical study) and forced cooling were

employed to speed up the process. More detail on the development of a LF strategy

to produce the section is given in the results section.

A third attempt was to produce a full scale part from 3.2mm mild steel CR4.

This was performed at the LLEC using a 4kW Nd:YAG laser and a six axis Reis

robot beam delivery system (figure 3.2.27). The same centre clamping arrangements

as the previous study were used and a double pass strategy employed. Despite the

shorter wavelength of laser light (1.06µm) it was found to be still necessary to coat

the surface with graphite, this not only improved the absorption but prevented back

reflection damaging the fibre delivery system. A strategy of tightly spaced irradiation

lines were used to form the tight corners of the section and more spaced out bend

lines to form the gentle curvature of the middle of the section. More detail is given in

the results section.

Figure 3.2.26: Set-up for the laser forming of the strut section from 200x100mm 1.6mm Ti64 sheet.

Figure 3.2.27: Set-up for the full scale laser forming of the strut section from 3.2mm Mild Steel sheet.



3.3 3D Laser Forming

As discussed earlier, 3D laser forming encompasses laser forming operations that

can utilise combinations of multi-axis two dimensional out-of plane bends and in-

plane localised shortening to produce three dimensional spatially formed parts e.g. a

dome.

There has been a considerable amount of work carried out on two-

dimensional laser forming, using multi-pass straight line scan strategies to produce a

reasonably controlled bend angles in a number of materials, including aerospace

alloys 1, 19, 57-72 and some 3D work 93-106. However in order to advance the process

further for realistic forming applications and for straightening and aligning

operations in a manufacturing industry it is necessary to consider larger scale 3D

laser forming. The initial approach taken to develop this idea was to investigate scan

strategies for the production of the continuous 3D primitive shapes saddle, pillow

and twisted shape from rectangular 400x200mm 1.5mm Mild Steel CR4 sheet and

Ti6Al 4V (Ti64 or TA10) sheet of various gauges (Figures 3.3.1, 3.3.2 & 3.3.3).

Figure 3.3.2: 3D Primitive, ‘The Pillow Shape’

Figure 3.3.1: 3D Primitive, ‘The Saddle Shape’



A study into 3D laser forming using successful strategies on much larger

800x400x5mm Mild Steel CR4 was also completed, with the aim of proving the

validity of the process as a tool for the ship building industry.

The primitive shapes themselves, suggested by Magee, 29 were chosen due to

their relatively simple 3D geometry and the ability to produce a large number of

more complex shapes through combinations of these three. It was thought that these

shapes would provide useful case studies with which to build up the design rules for

other more complex 3D shapes and surfaces. No absolute dimensions were set for

these shapes as they were considered merely conceptual surfaces.

The material of interest to the aerospace industry was the Ti64, however due

to cost and availability, the substitute Mild Steel CR4 material was used for the

majority of the study and only successful strategies were tested on the Ti64.

The initial investigation was based around a purely empirical approach to

establish rules for the positioning and sequencing of the irradiation lines required for

the controlled 3D-laser forming of symmetrical/uniform saddle, pillow and twisted

shapes from the rectangular sheet material, plus an investigation as to how well the

strategies scaled up to larger thicker sheets and sheets of different length to width

ratios. This was followed up by a more analytical approach to determine scan

patterns and the development of a geometry based model in Matlab. Further

development of this model formed the basis for a system of closed loop feedback

control of 3D laser forming. In addition, as one of the final goals for the larger

research programme for which the results in this thesis formed part of, was the

creation of a demonstrator system for the controlled 3D laser forming of one of the

Figure 3.3.3: 3D Primitive, ‘The Twisted Shape’



primitive shapes, the subsequent evolution of the Matlab based model coupled with

hardware and software developments formed the basis for this system. The results

from this work are presented in chapter 5.3.

The larger research programme was an EPSRC funded consortium of BAE

SYSTEMS, Rolls-Royce, The University of Liverpool, Heriot Watt University and

The University of Cambridge. All of the contributions to this thesis by any or all of

the consortium partners have been fully acknowledged.

3.3.1 Empirical Study

This initial investigation was based around a purely empirical approach to establish

rules for the positioning and sequencing of the irradiation lines required for the

controlled and repeatable 3D-laser forming of symmetrical/uniform saddle, pillow

and twisted shapes from rectangular 400x200x1.5mm sheet Mild Steel CR4 (AISI

1010) material with tests using successful strategies on the more commercially

interesting and expensive 0.9-1.6mm Ti64 sheet.

The Electrox 1.5 kW CO2 laser system, wavelength 10.6µm, run in

continuous wave mode, outlined earlier was used for the forming process. The

400mm x 200mm samples were held in place by a simple pair of stops that held the

plate in position in the X-Y directions but allowed for Z-axis deformation as the part

was processed (Figure 3.3.4). The steel work bed acted as an effective heat sink for

the plate being formed, no additional cooling was used.

Figure 3.3.4: Experimental Set-up for the 3D Laser Forming empirical study



Figure 3.3.5: Set-up for thick section 3D LF

The samples were laser cut to the correct dimensions in order reduce any pre-

stressing that might influence the forming results. To prepare them for forming they

were first cleaned with acetone in order to remove the oil that protected them from

oxidation (Mild Steel plates) plus any grease. They were then spayed with graphite

in order to increase absorption. The graphite was sprayed using a hand held can,

making it quite difficult to achieve an even coverage over such a large plate. If too

thick the graphite would simply burn off and if too thin the absorption would be poor.

The scan strategies were developed via a purely empirical method. By simply

taking an initial concept it was possible to develop it on a scan-by-scan basis. As

there are many process variables in laser forming it was decided to hold as many as

constant as possible and simply vary the scan speed and scan pattern to achieve the

desired result, a distortion free and smooth contoured symmetrical shape (Figures

3.3.1 – 3.3.3) with the minimum heat input.

The laser power was held at 800W at the output window of the laser,

however a power puck test at the processing head revealed a 5% loss from the

turning mirrors and ZnSe lens, the actual power at the work surface was 760W. For

the first study on the saddle shape using 1.5mm Mild Steel a beam diameter of 8mm

was used, a beam of this size produces a large radii un-faceted or non-humped bend

which was desirable in this case for a smooth contour. For the second study on the

pillow shape a smaller beam diameter of 5.5mm was used, the reasons for this will

be discussed in the results section. For the third study on the twisted shape a 5.5mm

beam diameter was also used.

An investigation into the scaling up of the

empirically found scan strategies to larger thicker

sheets in order to demonstrate the validity of the

process for shipbuilding was performed initially

on the 1.5kW Electrox workstation 2 (figure 3.3.5),

using 360x190x5mm Mild Steel CR4 plates,

1200W, an 8mm beam diameter and a traverse

speed of 10mm/s, with the attempt to produce a

saddle shape.

A further study on 800x400mm 5mm Mild

Steel CR4 was performed at the Lairdside Laser



engineering Centre (LLEC) on a large 5 Axis Laserdyne 890 beam delivery system,

employing a 3kW PRC CO2 laser (Figure 3.3.6). This study used 1.8kW (power

limited by fault in the laser), an 6mm beam diameter, 83mm/s processing speed and

multiple pass strategy in an attempt to produce a saddle shape based on the data from

the smaller sheets.

The post forming geometries of the 400x200mm sheets and smaller were

verified online using the laser range finder based co-ordinate measuring machine

system (CMM) integrated into the XYZ beam manipulation hardware and software,

this system was outlined earlier in this chapter. This system produces Z or height

data at known X & Y locations, from this data it is possible to produce a contour

map of the surface.

As the larger sheets could not be accommodated on the 440x440mm stages

of the Electrox Workstation 2 for measurement by CMM, a cruder less accurate

method of shape measurement was employed. The Z height of the plate edge was

measured by hand using a ruler at 40mm intervals; it was thought that by monitoring

the edges it was possible to infer the geometry of the rest of the plate surface.

Figure 3.3.6: 3D Laser Forming using a Laser Dyne 890 5 Axis Gantry



3.3.2 Development of a Geometry based Model for 3D Laser

Forming using Matlab

This work was produced as part of a collaboration with Andrew Moore of Heriot-

Watt University Edinburgh, the author acknowledges this input without which the

work could not be produced.

It was realised from the empirical study that in order to control the process of

3D laser forming it was necessary to have the ability to define the surface to be

formed. In addition by defining the surface and analysing properties such as gradient

and curvature, it was thought this may lead to a method of scan strategy prediction.

To this aim, a method of surface creation and analysis was devised using Matlab.

This study concentrated initially on the pillow shape, as this was the more likely

candidate for use in the 3D laser forming demonstrator system, other shapes were

investigated once the model was shown to produce useful results.

Matlab, the PC based computer program, is an integrated technical

computing environment that combines numeric computation, advanced graphics and

visualization and powerful high-level programming capabilities. It is a tool for doing

numerical computations with matrices and vectors. It can also display information

graphically. The program can be operated as a stand-alone or can be augmented with

powerful toolboxes such as Simulink to enable process modelling, online control,

sensor integration, image analysis and much more. Matlab is operated via a

command line interface, where individual Matlab specific routines and operations

can be run. In order to improve process time sequences of commands and operations

can be grouped together in a text file (.m) to be executed as a whole on the command

line. An example of a .m file used for this study is given in appendix 1.

The development of the model and subsequent results form the ‘results and

discussion’ section of this study (Section 5.2). The output from the model was

verified using the graphite coated 400x200x1.5mm Mild Steel CR4 plates using

energy parameters consistent with the TGM. An improved set-up was used for this

verification and subsequent model developments, employing a centre clamp, bolt and

pre-drilled plate to fix the workpiece in space, with all deformation relative to the

centre-bolt (figure 3.3.7). Although not ideal to drill plates (additional pre-stressing),

fixing the plates was essential to avoid any unwanted movement, additionally by



raising the plate off the bed this alleviated any problems with the weight of the plate

limiting the amount of forming available. An alternative method of clamping that

was investigated was to use four corner clamps mounted on universal joints so as to

allow the plate to deform (figure 3.1.9), this was successful, however there was an

upper limit to the amount of forming achievable.

In order to take account of the movement in Z, which is a function of the spot

size, a long focal length lens was used (190mm FL ZnSe) which has a long depth of

focus and hence little change in spot size is produced for small variations in Z. In

addition it was decided to work above the focus (normally below), in this region a

movement towards the lens would result in the beam becoming more defocused and

hence a reduction in intensity, thus the plate will under-form in a this region not over

form. From a control point of view this is essential as over-shoot past the required

shape is not desirable. Ideally the beam parameters e.g. spot size and shape (elliptical

beams during oblique angles of incidence) should be kept constant. Possible methods

of achieving this are (although not used) beam collimation, capacitive focus control

and auto-normal hardware (may require more than 3 axes).

The output from the Matlab study was initially manually converted to CNC

code for the Galil DMC 1730 controller using the linear interpolation routine of the

Galil programming language. The controller can take a fairly coarse set of X and Y

data points and draw a smooth line through them. Matlab can generate a table of X

and Y location points along a predicted possible scan line. The additional operational

commands would then have to be added such as shutter control. A later development

of the code allowed for the production of the full CNC code without user

intervention, this will be discussed later.

Figure 3.3.7: Improved 3D Laser Forming Set-up



The geometries of the plates were verified using the laser range finder based

CMM integrated with the XYZ workstation, an improvement to the control software

was made to take account of the new starting location (addition of once touch start

location find buttons) and at a later development the increase in step size (reduce

number of data points taken) to reduce measurement cycle time to 10 minutes from

45 minutes.

The concept and theory behind an in-plane strain approach rather than a

geometry approach to the energy distribution realised on a surface is also given the

in results section for this study (chapter 5.2).

3.3.3 3D Laser Forming Demonstrator System

The 3D laser forming demonstrator system was a development from the work on the

geometry based model using Matlab. Its development and subsequent testing and

output forms the results and discussion section for this study, due to this, only a brief

overview of the procedures and concepts used are given here.

It was found that, although a prediction of a scan pattern could be obtained

either by either empirical or analytical means leading to a ‘single shot’ scan strategy,

from a control point of view this was not desirable. Due to inherent material

variability and unknown residual stress conditions an open loop method of 3D laser

forming based on a look up table of known results would ultimately produce

significant errors. Instead a combined approach was sort, using the predictive

capability of the matlab model to produce an initial scan strategy to be realised, and

then use an adaptive/iterative approach to produce further corrective scan strategies

based on the current geometry when compared to a target geometry, incrementing

slowly towards the desired result scan after scan providing the plate is under-formed

on the first pass. As with closed loop 2D LF the processing speed was used to

control the energy input and hence the rate and magnitude of forming. The difference

in closed loop 3D LF was that the output data from the matlab model called for a

distribution of required distortion (based on geometry and required in-plane strain),

such that the energy input at any point depended on its location on the sheet. This

was realised by varying the traverse speed along any scan line according to the

required deformation at that point (an example of the CNC code through which this



was achieved can be seen in appendix 6), the required distortion was related to the

traverse speed by calibration data obtained from the 2D LF empirical study, such that

for a given power, spot size and traverse speed a bend angle could be predicted.

After each scan was realised a measurement of the surface geometry was

taken, a comparison was then made to the target geometry and based on the error a

further scan strategy was predicted. In order to control the forming rate to avoid

overshoot, the scan speed range was increased automatically to compensate. The

demonstrator system was set up for the closed loop forming of the pillow shape, an

ideal mathematical surface, an elliptic paraboloid, was used as the target shape, the

pillow shape falls into this class of surfaces. Energy parameters consistent with the

TGM were used throughout (power and spot size held constant), the experimental

set-up remained the same as the previous study with the addition of an air jet

controlled by the Galil system on the underside of the plate to speed cooling after

processing prior to measurement, the air jet was not used during processing.

As the control software was written in Visual Basic and the

predictive/adaptive model run through Matlab, a manual intervention was required to

transfer the surface height data from the CMM system to the Matlab script after each

pass. The script was then executed in order to generate the CNC code for the next

pass. Scans were continued until the error reduced to a desired level. The results

from this study are presented in chapter 5.3.

Chapter 4 2D Laser Forming- Results & Discussion


Chapter 4

2D Laser Forming –

Results and Discussion This chapter contains the results and discussion of experimental and numerical

studies into the 2D laser forming of a number of materials, including mild steel,

aluminium and aluminium alloys, titanium alloy and newly developed Fibre Metal

Laminate materials (FML).

4.1 Empirical Study - Characterisation of the Laser Forming

Process

The first results presented in this thesis are from an empirical 2D laser forming

investigation on a number of materials using the TGM, characterising the 2D laser

forming process. Variables investigated include; beam spot size, laser power,

traverse speed, multiple and single pass strategies, time delay between passes, bend

angle rate and coating degradation. The results for each material investigated are

presented in the following sections.



4.1.1 1.5mm Mild Steel CR4

The first part of the study into this material was to determine a process map or

window such that for a given incident power, laser beam spot size and traverse speed

an expected bend angle for a single pass could be known. For the three beam

diameters investigated, 3, 5.5 and 8mm a process map was found for a number of

incident laser powers and traverse speeds. A sample size of 80x80mm was used. The

results can be seen in figures 4.1.1 to 4.1.3.

Figure 4.1.1: 2D LF process map for 1.5mm mild steel CR4, 3mm Beam Dia.

Figure 4.1.2: 2D LF process map for 1.5mm mild steel CR4, 5.5 mm Beam Dia.

Figure 4.1.3: 2D LF process map for 1.5mm mild steel CR4, 8mm Beam Dia.



For the above results, due to the large amount of data, only a single datum

point was taken at each traverse speed and a line of best fit drawn through them

(using a polynomial fit of various orders). It was thought that this would be sufficient

to obtain processing regions were laser forming was possible and that an

approximate speed \ bend angle relationship for a given power and spot size could be

found. In addition due to material and process variability a reliance on obtaining an

exact bend angle only from given process parameters is flawed. This will be

discussed later when considering closed loop control.

It can be seen in figures 4.1.1 to 4.1.3 that for this material there is an

approximately (negative) linear relationship between the bend angle and the traverse

speed for a mid portion of the data for a given power and spot size. At higher

traverse speeds the amount of forming is small and very similar. At a point, as the

speed slows, an activation energy intensity level appears to be reached whereby any

further decrease in speed results in a significant increase in bend angle achieved.

This is consistent with the increasing energy input causing a higher thermal gradient

through the thickness and thus increased forming. This linear increase is arrested at

lower traverse speeds, particularly for the larger 5.5 and 8mm beam diameters,

where a maximum forming level occurs and any further decrease in speed results in a

decrease in bend angle. This could signify a point where the increasing energy input

is detrimental to forming a sufficiently high enough thermal gradient through the

section and thus process efficiency is lost. This could also be a transition point to

another mechanism, in that the bending strains developed are being outweighed by

the in-plane strains as the section becomes uniformly heated and so the shortening

and buckling mechanisms become dominant. For the 3mm beam diameter result

(figure 4.1.1) the fall off at lower traverse speeds is not as apparent. However some

surface melting occurred on samples processed at 20mm/s and below. Thus it is

likely that for the smaller beam diameter at higher intensities a sufficiently high

thermal gradient through the thickness is still maintained.

The effect of the incident laser power on the bend angle response can be seen

in the above figures. As the power increases for a given spot size, the data is shifted

to the right since to achieve the same bend angle the traverse speed must be

increased to compensate for the increased power. This is consistent with obtaining a

similar energy fluence on the sample (energy fluence being the intensity multiplied

by the interaction time, recorded in J/cm2). However, as can be seen in figures 4.1.2



and 4.1.3 for a 5.5mm and 8mm beam diameter, the increase in laser power from

1000W to 1200W has resulted in little or no increase in bend angle for the same

traverse speed. This could indicate a saturation point whereby due to the relatively

low thermal conductivity of the mild steel, a further increase in the thermal gradient

without surface melting and hence process efficiency is not possible. This may also

be the case in figure 4.1.1 for a 3mm beam diameter, where an increase in laser

power to 1000W from 760W has resulted in an anomalous bend angle response.

Another possible explanation is the coating interaction since at higher intensities the

absorptive graphite coating may degrade or burn off more readily without sufficient

heat transfer to the substrate and so the process efficiency is decreased.

The effect of increasing the laser beam diameter can also be seen in the above

figures. At 760W the data is shifted further to the left as the beam diameter is

increased, such that to achieve the same bend angle the traverse speed must decrease,

again consistent with maintaining a similar energy fluence. This highlights the

differences between the experimental process parameters used and the temperature

gradient mechanism (TGM) theory 11, in particular the laser beam diameters selected.

The TGM theory states that the beam diameter should be of the order of the sheet

thickness (e.g. 1.5mm for this material) for the mechanism to be active, yet an

assumption was still made here that the TGM is the active mechanism throughout.

The initial reasons for using beam diameters of 3mm and greater was to prevent

surface melting that was present when using smaller beam diameters at powers

greater than 400W. In addition an assumption was made that due to the relatively

low thermal conductivity of the mild steel a thermal gradient through the thickness

would always be present in the material even at larger beam diameters causing a

positive bend towards the laser. This is backed up by the experimental data. It is

likely, however, that a combination of distinct mechanisms are active as the beam

diameter increases, in that the net plastic bending strains are decreased as the section

becomes more uniformly heated and the in-plane strains that cause lateral shrinkage

(shortening) are increased. However providing an asymmetry exists in the plasticized

zone through the section a positive out of plane bend is produced without the

development of a buckle or instability characteristic of the buckling mechanism.

For the 1.5mm mild steel it was found that the laser beam diameter governed

the radius of curvature in the bent zone, in that a smaller beam diameter produced a

sharp bend and a larger beam diameter produced a smoother contoured larger radii



bend. This could be useful when considering LF as a manufacturing tool, both for

practical and aesthetic considerations.

The next part of the study on this material was to select process parameters

for each beam diameter from figures 4.1.1 to 4.1.3 for multi-pass LF up to 30 passes.

The parameters selected were: 3mm beam diameter, 760W, 55mm/s; 5.5mm beam

diameter, 760W, 30mm/s; 8mm

beam diameter, 760W, 20mm/s;

the time interval between each

pass was 60 seconds. The

criteria for selecting these

parameters were that they

should be part of the linear

section of the speed\bend angle

calibration graph and that they

were predicted to give a bend

angle of approximately 1° per

pass, thus consistent with the

TGM theory. The results can

be seen in figures 4.1.4, 4.1.5

and 4.1.6.

For these results the

cumulative bend angle and

bend angle rate per pass are

presented up to 30 passes, the

bend angle is on the primary Y

axis (left) and the bend angle

rate is on the secondary Y axis

(right).

For each of the three

process parameter

combinations used (figures

4.1.4 to 4.1.6) it can be seen

that a considerable bend angle

Figure 4.1.4: 1.5mm mild steel CR4, 3mm Beam Dia., 760W, 55mm/s, 30 pass

Figure 4.1.5: 1.5mm mild steel CR4, 5.5mm Beam Dia., 760W, 30mm/s, 30 pass

Figure 4.1.6: 1.5mm mild steel CR4, 8mm Beam Dia., 760W, 20mm/s, 30 pass



has been produced in each coupon after 30 passes. It can be noted that the bend angle

increase or bend angle rate per pass was not constant for each of the energy

parameters investigated. It can also be noted that the rate falls off at higher numbers

of passes. This is consistent with previously observed data and is attributed to a

number of factors including strain hardening in the HAZ, section thickening and

absorptive coating burn off, discussed earlier. For the data obtained for a 3mm beam

diameter (figure 4.1.4), the expected bend angle rate of ~1° (figure 4.1.1) was

achieved on only the first pass and for each subsequent pass the bend angle rate per

pass generally decreased (somewhat variably) down to a level of ~0.25° per pass.

For the data obtained using 5.5mm and 8mm beam diameters (figures 4.1.5 and 4.1.6)

the bend angle response was similar, this may be due to the similar energy fluence

realised on each coupon. It can be observed in figures 4.1.5 and 4.1.6 that the

expected bend angle rates per pass of ~1.4° and ~1.6° respectively (figures 4.1.2 and

4.1.3) were obtained in the first few passes. It can also be observed that the bend

angle rate increased up to a maximum over the first four passes and then as with the

3mm beam data a general decrease was observed, however in contrast there was still

a considerable forming rate available after 30 passes for these last two forming

parameters. These results demonstrate that the reliance on a given set of processing

parameters to produce a certain bend angle is not practical in an open loop set-up due

to the variable nature of the bend angle rate per pass. They demonstrate that a closed

loop system would need to be considered to increase the reliability and repeatability

of the LF process; the practical implementation of this concept will be demonstrated

in a later section.

Whilst there are a number of theories for the bend angle rate fall off at higher

number of passes (discussed earlier), the distribution of bend angle rates over a

number of passes has not been commented on before in the literature. This is mainly

due the fact that bend angles measurements are rarely taken less than every five

passes (equipment limitations) or that the data was not analysed in the same way. A

possible explanation for the variable nature of the bend angle rate is that it could be

related to the absorptive coating burn off, in that if regions of the coating within the

irradiated track are degraded more than others (possibly due to the manual nature of

coating application causing variation in thickness) the localised absorption

coefficient and hence heat input locally will be different as the laser is traversed

across the coupon. In addition, as an alternating strategy was used throughout (scan



direction alternated for each pass), the same heat distribution would not be realised

across the coupon for each pass (each subsequent pass producing more coating burn

off), thus producing the variable bend angle rate. A possible explanation for the

distribution of the bend angle rates over the first few passes may also be found in the

interaction of the laser beam with the absorptive coating. It is likely that there is an

optimum thickness of graphite coating for the transmission of heat into the substrate

without significant reflection of the incident laser beam (coating too thin) and

excessive burn off (coating too thick). For the data obtained from the larger beam

diameters, 5.5mm (figure 4.1.5) and 8mm (figure 4.1.6), the initial bend angle rate

increase may be a result of the coating burn off per pass achieving an optimum

coating thickness before further loss of the coating becomes detrimental to the

process efficiency. Further work could be undertaken to confirm this by processing

at a shorter laser wavelength without a coating.

Another possible explanation for the variation in bend angle rate during the

initial few passes could be the heat distribution in the coupon, in that as the bulk

material temperature increases with increasing number of passes, the additional heat

may aid the process but the achievable thermal gradient through the thickness may

decrease. A balance point may be reached over the first few passes where thermal

equilibrium is achieved and hence optimum forming. The results of a study into the

heat distribution within a coupon during LF are presented in a later section.

Related to the temperature distribution is the delay or interval time between

each pass. A study was conducted using 1.5mm mild steel into the effect of interval

time on the bend angle over 30 passes. The results are presented in figures 4.1.7 to

4.1.12.

Figure 4.1.7: Laser forming of 1.5mm mild steel CR4, 3mm Beam

Dia., at various inter-pass time delays



Figure 4.1.8: Effect of inter-pass time delay on the laser forming of 1.5mm mild steel CR4, 3mm Beam Dia.

Figure 4.1.9: Laser forming of 1.5mm mild steel CR4, 5.5mm Beam Dia., at various inter-pass time delays

Figure 4.1.10: Effect of inter-pass time delay on the laser forming of 1.5mm mild steel CR4, 5.5mm Beam Dia.

Figure 4.1.11: Laser forming of 1.5mm mild steel CR4, 8mm Beam Dia., at various inter-pass time delays



The reason for using a time delay between each pass is to allow the material

to cool so as to prevent surface damage, however it can be seen in the above figures

that the length of the time delay in-between each pass does have an effect on the LF

process. In figures 4.1.7, 4.1.9 and 4.1.11 the bend angle after each pass up to 30

passes are presented for 24, 40, 60 and 80 second time delays for the three process

parameters investigated. In figures 4.1.8, 4.1.10 and 4.1.12 the bend angle is

presented against the time delay at every five passes, this was found to be a useful

way of determining the effect of time delay on process efficiency. The shortest 24

second time delay was arrived at from the time taken to measure the bend angle in

the 80x80mm coupon using the laser range finder system.

It can be seen in figures 4.1.7 and 4.1.8, for the smaller 3mm beam diameter

result, that the shorter inter-pass delay times produce more forming over the 30

passes. A possible reason for this may be the heat remaining in the coupon from the

previous pass aids subsequent passes by reducing the flow or yield stress of the

material, in that a hot plate is easier to form than a cold one. As the time delay

between passes reduces the more heat is retained in the plate for the next pass. For

the data obtained using a 5.5mm beam diameter (figures 4.1.9 and 4.1.10) the effect

of the inter-pass delay time is less obvious and more subtle. A slight increase in the

achievable forming was observed at the 60 second time delay (figure 4.1.10). For the

data using an 8mm beam diameter (figures 4.1.11 and 4.1.12) a peak in the bend

angle achieved at 60 second intervals can be observed in figure 4.1.12. This may

indicate the existence of a balance point or a trade off between the heat retained in

the coupon aiding the process by reducing the flow stress and the increased bulk

material temperature reducing the available temperature gradient through the

Figure 4.1.12: Effect of inter-pass time delay on the laser forming of 1.5mm mild steel CR4, 8mm Beam Dia.



thickness as the laser beam is passed over the surface. This optimum inter-pass delay

time may only be apparent when using larger beam diameters for forming due to the

lower intensities and hence lower peak surface temperatures realised since the

achievable thermal gradient through the thickness is more sensitive to the bulk

material temperature.

4.1.2 0.9-3.2mm Ti6Al4V

This empirical study was conducted on 0.9, 1.4, 1.6, 2 and 3.2 mm thick Ti6Al4V

sheet (Ti64). For clarity the results for each thickness investigated are presented

separately.

0.9mm Gauge Ti64 Sheet

As with the previous material the first study conducted was to ascertain a laser

forming process map. The bend angle results at various traverse speeds, using three

beam diameters (3, 5.5 and 8mm) and three laser powers for each, are shown in

figures 4.1.13 to 4.1.15.

Figure 4.1.13: 2D LF process map for 0.9mm Ti64, 3mm Beam Diameter.

Figure 4.1.14: 2D LF process map for 0.9mm Ti64, 5.5mm Beam Diameter.



It can be seen in the above figures that the bend angle response across the

same range of traverse speeds of 0.9mm Ti64 is greatly different to that of mild steel

presented earlier. It can be observed that for each of the beam diameters investigated

and at each of the laser powers used, a peak in the amount of forming exists at a

certain traverse speed (only observed at very low traverse speeds in mild steel). As

the speed reduces from 90mm/s (or the highest data point taken) the bend angle

produced steadily increases up to a maximum, after this a further reduction of the

traverse speed leads to a decrease in the bend angle produced. A similar response to

this, in this material, has been observed before by Magee 29. It was attributed in this

work to a point were the additional energy input from lowering the traverse speed,

initially beneficial, becomes detrimental due to a loss in the extent of the thermal

gradient achievable through the section. A possible reason for the large difference in

bend angle response between mild steel and Ti64 is the very low thermal

conductivity (table 3.2.8). If the heat is retained within the area of the irradiated track

for longer then the effect of the lower traverse speed reducing the thermal gradient

will be amplified.

For the data obtained using a 3mm beam diameter, figure 4.1.13, it can be

seen that the distribution of the bend angles produced for a given laser power are

relatively uniform across the speed range, with a peak of ~2°. The effect of

increasing the laser power can be observed however. Whilst there is no significant

increase in the peak bend angle, the traverse speed at which it occurs and the rest of

the data are shifted more to the right of the graph with the increasing power. This

may indicate that a maximum achievable bend angle for a single pass exists for a

given beam diameter, in that the maximum depth and width of the plasticized zone

must be governed by the beam diameter. (conduction limited hemispherical

Figure 4.1.15: 2D LF process map for 0.9mm Ti64, 8mm Beam Diameter



temperature distribution) and that for the TGM the magnitude of the moment

generated about the scan line and hence the bend angle is governed by the size of the

plasticized zone. It can be observed in the above figures that an increase in the

maximum bend angle per pass is possible with a larger beam diameter.

For the larger beam diameters (figures 4.1.14 and 4.1.15) the bend angle

response over the range of traverse speeds investigated for the 0.9mm Ti64 is less

evenly distributed. As with the 3mm beam data a maximum distortion for a certain

speed can be observed, in addition at lower traverse speeds after an initial decrease

(below the peak bend angle/traverse speed) a slight increase in the bend angle

produced can be observed giving a second peak. This could indicate a change in

mechanism to the BM (buckling mechanism) or could be related to the absorption

coefficient increasing due to some surface melting at low traverse speeds. As with

the 3mm beam data the effect of the increasing power results in a shift of the data to

the right of the graph, consistent with the idea that at a higher power a higher speed

would be required to give the same amount of forming.

Three process parameter combinations were chosen from the above process

maps for a multi-pass study. These were: 3mm beam diameter, 500W, 40mm/s;

5.5mm beam diameter, 500W, 30mm/s; 8mm beam diameter, 900W, 40mm/s; an

inter-pass delay of 30 seconds was used throughout (considered long enough for the

thin material). These parameters were chosen due to their close proximity to the

identified maximum forming point (peaks of 2°, 2.3° and 3° respectively). The

results of this study can be seen in figures 4.1.16 to 4.1.18. As with the previous

material the cumulative bend angle and bend angle rate per pass are presented.

Figure 4.1.16: 0.9mm Ti64, 3mm Beam Dia., 500W, 40mm/s, 30 pass



For the data obtained using the 3mm beam diameter (figure 4.1.17) it can be

seen that a considerable bend angle has been produced in the 80x80mm coupon after

30 passes. As with previously observed phenomena the bend angle rate per pass is

not constant, after an initial increase (possible reasons discussed previously for mild

steel) the rate declines with increasing number of passes (discussed earlier also). The

variable nature of the rate observed with the mild steel can be seen here also. It can

be noted that at higher numbers of passes a darkening and some sintering of the

surface on the irradiation line was observed. This may affect the absorption

coefficient of the coupon. The reason for mentioning this becomes clearer when

studying figures 4.1.17 and 4.1.18. It can be seen that using the larger beam

diameters, forming is possible at a good bend angle rate for 8 to 10 passes. For

subsequent passes the rate falls to effectively zero with no more forming possible for

increasing numbers of passes. It is unlikely that material factors such as strain

hardening and section thickening (identified as factors that reduce bend angle rate

per pass) are to blame for such a dramatic fall off after only 10 passes. On inspection

of the coupons after 20 passes it was realised that a significant amount of the

Figure 4.1.17: 0.9mm Ti64, 5.5mm Beam Dia., 500W, 30mm/s, 20 pass




graphite coating had been burnt away or degraded. This could be observed optically

and can be seen in figure 4.1.19.

It can be assumed that if the reflectivity of the irradiated area has increased

significantly visibly, then the reflectivity to the 10.6µm laser radiation must have

also increased to approach that of the substrate (~98%, from figure 3.1.21). The high

degree of coating degradation when compared to that of mild steel may be related to

the very low thermal conductivity of the Ti64 since if the heat is being retained

locally for longer and the heat transfer rate from the coating to the substrate is low,

the coating itself reaches and possibly remains at a higher temperature for a longer

period of time than with mild steel, thus it degrades faster. This is backed up by

observation of the laser beam surface interaction on the graphite coated mild steel

and Ti64 coupons. For the mild steel there is a dull glow from the graphite surface as

the beam passes over (brightness dependent on the incident intensity). For the Ti64

an extremely bright interaction can be observed during the first few passes (reducing

for the above cases after 10 passes). To confirm this theory it was decided to re-spray

(over the top of the original degraded coating) the last two coupons with graphite at

20 passes and continue processing. The results can be seen in figures 4.1.20 and

4.1.21.

Figure 4.1.19: Graphite coating condition after 20 passes, 0.9mm Ti64, 5.5mm Beam Dia., 500W, 30mm/s

Figure 4.1.20: 0.9mm Ti64, 5.5mm Beam Dia., 500W, 30mm/s, 30 pass, Coating re-spray at 20 passes



It can clearly be seen in the above figures that the condition of the absorptive

coating is a large factor in the laser forming of this material. After re-spraying at 20

passes it was possible to continue forming using the two energy parameters

investigated. It can be seen that the bend angle rate immediately increases from zero

at pass 21. It can be noted that the rate does not however approach the initial peak

rate obtained during the first few passes but does appear to fall off again as with the

first 20 passes. This difference could be due to other material factors affecting the

bend angle rate or to how the coating is applied since the re-sprayed coating is

applied on top of the degraded coating rather than a clean flat surface.


The process maps for this sheet thickness can be seen in figures 4.1.22 to 4.1.24.

Figure 4.1.21: 0.9mm Ti64, 8mm Beam Dia., 900W, 40mm/s, 40 pass, Coating re-spray at 20 passes

Figure 4.1.22: 2D LF process map for 1.4mm Ti64, 3mm Beam Dia.



It can be seen in the above figures that there is a more limited number of

usable forming parameters for this thickness of material when compared to the

0.9mm data. It is reasonable to assume that this is due to a significant increase in

bending strength with the increase in thickness of this high strength material. It is

still however possible to identify similarities with the data for the previous thickness.

For each of the three beam diameters investigated it can be seen that an increase in

the laser power results in the data shifting to the right. It can also be observed that

there are a number of peaks in the forming data. However, the effect of the loss of a

high thermal gradient through the thickness at lower traverse speeds does not appear

to be as significant as with the 0.9mm gauge Ti64. This may be due to the increase in

section thickness and low thermal conductivity ensuring a significant thermal

gradient exists even at low traverse speeds.

For the multi-pass study three process parameter combinations were used,

these were: 3mm beam diameter, 900W, 50mm/s; 5.5mm beam diameter, 900W,

45mm/s; 8mm beam diameter, 900W, 30mm/s; an inter-pass delay of 50 seconds

was used throughout. These parameters were chosen on the basis that they were

Figure 4.1.23: 2D LF process map for 1.4mm Ti64, 5.5mm Beam Dia.




positioned within a region where significant forming was possible and no obvious

surface damage took place. The results can be seen in figures 4.1.25 to 4.1.27.

As can be seen in the above figures a similar bend angle response to the

0.9mm gauge Ti64 has been produced (although much less forming due to the

increased thickness). As with the previous thickness, the higher intensity 3mm beam

diameter caused some surface sintering resulting in a darkening along the track

despite the coating degradation. This may account for the ability to continue forming

of the coupon past 10 passes with only a slight fall off in the bend angle rate (figure


Figure 4.1.26: 1.4mm Ti64, 5.5mm Beam Dia., 900W, 45mm/s, 20 pass




4.1.25). For the larger beam diameters (figures 4.1.26 and 4.1.27) the bend angle rate

again fell dramatically after 7 passes (more so for the 8mm beam data). As with the

previous thickness the coating was observed to be severely degraded after only 10

passes. In order to confirm the idea of coating degradation the last two coupons were

re-sprayed with graphite (at 20 passes) and processing continued, the results can be

seen in figures 4.1.28 and 4.1.29.

It can be seen in the above figures that by topping up the absorptive coating it

was possible to continue forming of this thickness of titanium alloy. It can be seen

that although the bend angle rate was increased by re-spraying it did not reach the

same level achieved during the first few passes suggesting that there are other more

subtle factors involved in the bend angle rate fall off. As with the first cycle the bend

angle rate rapidly falls off after only 7 to 8 more passes after re-spraying.

Figure 4.1.28: 1.4mm Ti64, 5.5mm Beam Dia., 900W, 45mm/s, 30 pass, Coating re-spray at 20 passes

Figure 4.1.29: 1.4mm Ti64, 8mm Beam Dia., 900W, 30mm/s, 30 pass, Coating re-spray at 20 passes




The process maps for this sheet thickness can be seen in figures 4.1.30 to 4.1.32.

It can be seen in the above figures that the bend angle\speed response for the

beam diameters and powers investigated are very similar to the data obtained for the

1.4mm gauge Ti64. This is reasonable due to the similar thickness and it being

unlikely that the 0.2mm difference would add a significant increase in section

strength (although some differences exist). For the multi-pass study the following


Figure 4.1.31: 2D LF process map for 1.6mm Ti64, 5.5mm Beam Dia.




process parameter combinations were selected: 3mm beam diameter, 740W, 40mm/s;

5.5mm beam diameter, 740W, 30mm/s; 8mm beam diameter, 740W, 20mm/s; an

inter-pass delay of 50 seconds was used throughout. These parameters were chosen

on the basis that they were positioned within a region where significant forming was

possible and no obvious surface damage took place. In addition a different power

level curve to the previous thickness was selected to prove the validity of the process

map method. The results can be seen in figures 4.1.33 to 4.1.35.

Figure 4.1.33: 1.6mm Ti64, 3mm Beam Dia., 740W, 40mm/s, 20 passes

Figure 4.1.34: 1.6mm Ti64, 5.5mm Beam Dia., 740W, 30mm/s, 20 passes

Figure 4.1.35: 1.6mm Ti64, 8mm Beam Dia., 740W, 20mm/s, 20 passes



It can be seen in the above figures that a very similar result to the 1.4mm

gauge material has been produced. The predicted bend angle rates for each of the

beam diameters 3, 5.5 and 8mm were approximately 1.5°, 1.6° and 1.5° respectively

(from figures 4.1.30 to 4.1.32). It can be observed in figures 4.1.33 to 4.1.35 that the

predicted rates are achieved (approximately) within the first few passes

demonstrating the usefulness of the process maps for the selection of usable process

parameters. However, as with the previous thicknesses investigated, the bend angle

rates after only a few passes fall dramatically (even more so for the 8mm beam

diameter data), attributable to the rapid loss of absorptive coating. Only for the data

acquired using a 3mm beam diameter (figure 4.1.33) is the bend angle rate sustained

above zero up to 20 passes. This, as in the thinner Ti64, may be attributable to some

surface melting or sintering that produces a darker rougher surface that my improve

the absorption of the 10.6µm radiation. Possibly due to the low thermal conductivity

of the material, the heat input from the higher intensity laser beam (3mm beam

diameter) may be retained for longer within the scanned area rather than be

conducted into the bulk material. This would raise the peak temperature within the

scan line possibly above the melting point (1604°C) within the upper surface layers.

The surface melting was not apparent within the first few passes but developed as the

more passes were realised on the coupon. This could indicate an excessive increase

in the bulk material temperature, increasing the peak temperature within the scan line.

These results indicate the use of larger beam diameters for a more evenly distributed

energy input is necessary to prevent surface melting particularly in the thicker

materials.

As with the previous Ti64 gauges investigated, a study into the effect of re-

spraying the graphite coating at 20 passes for the 5.5 and 8mm beam diameter data

was performed. The results can be seen in figures 4.1.36 and 4.1.37.

Figure 4.1.36: 1.6mm Ti64, 5.5mm Beam Dia., 740W, 30mm/s, 30 passes, Coating re-spray at 20 passes



It can be seen in figures 4.1.36 and 4.1.37 that by re-spraying the graphite

coating at 20 passes more forming has been possible. The forming rate immediately

increases at pass 21; however, the rate does not achieve the same level as in the first

few passes and falls off rapidly within the same number of passes. This suggests that

other more subtle factors are influencing the fall off such as the section thickening

and work hardening phenomena. Another possible reason for the drop in the peak

bend angle rate after re-coat is the less than ideal partially degraded surface onto

which the coating is sprayed. Normally for optimum adhesion a surface would be

cleaned with acetone first, however, this was thought to be adequate and for this set-

up. The coupon was left clamped and re-sprayed on the workbed to ensure the

alignment of the laser for the next pass.

2mm Gauge Ti64 Sheet

From the LF results of the 1.4 and 1.6mm Ti64 coupons it was decided to only

investigate the beam diameters

5.5mm and 8mm as usable

processing parameters without

melting for the 3mm beam

diameter were difficult to find.

The process maps for this

material can be seen in figures

4.1.38 and 4.1.39.

Figure 4.1.37: 1.6mm Ti64, 8mm Beam Dia., 740W, 20mm/s, 30 passes, Coating re-spray at 20 passes

Figure 4.1.38: 2D LF process map for 2mm Ti64, 5.5mm Beam Dia.



It can be seen in the above figures that as the sheet thickness has increased

the distribution of the bend angle data at different laser powers obtained using a

5.5mm beam diameter (figure 4.1.38) now resembles the data obtained using a 3mm

beam diameter in the thinner Ti64. It can be seen that it is possible to obtain a

significant single pass bend angle even in this 2mm thick high strength material.

However, as would be expected, higher laser powers are required to achieve the

larger bend angles.

The bend angle distributions across the speed and power ranges are similar to

previous gauges since it is possible to identify parameter combinations that give a

useful bend angle rate. The first parameter combination selected and tested for multi-

pass forming from the 5.5mm beam diameter data set were 1200W and 25mm/s.

This corresponds to a peak bend angle of approximately 2°. The result can be seen in

figure 4.1.40.

It can be seen in the above figure that an unexpected result has been produced,

in that the bend angle rate has peaked at the predicted level of approximately 2° but

the rate has not fallen off to the same extent as with the thinner samples. On closer

Figure 4.1.39: 2D LF process map for 2mm Ti64, 8mm Beam Dia.

Figure 4.1.40: 2mm Ti64, 5.5mm Beam Dia., 1200W, 25mm/s, 20 passes



inspection of the coupon it was found that some surface melting or sintering had

occurred and that, although the coating had degraded along the scan line, this

possibly aided absorption through a darkening of the surface or through a surface

roughening. The surface condition after 20 passes can be seen in figure 4.1.41; this

can be compared to figure 4.1.19.

It was decided to select a second set of parameters at a lower laser power for

the multi-pass study so as to avoid any surface melting; namely 5.5mm beam

diameter, 900W and a speed of

30mm/s. This was predicted to

give a bend angle of ~1° (figure

4.1.38). For the 8mm beam data

parameters at 1200W and

25mm/s were used; this was also

predicted to give a bend angle of

~1° (figure 4.1.39). An inter-pass

delay of 50 seconds was used

throughout. The results up to 15

passes can be seen in figures

4.1.42 and 4.1.43.

It can be seen in these

figures that a result more akin to

the data obtained from the

thinner TI64 has been produced.

With no surface melting the

Figure 4.1.41: 2mm Ti64, 5.5mm Beam Dia., 1200W, 25mm/s, Surface condition after 20 passes

Figure 4.1.42: 2mm Ti64, 5.5mm Beam Dia., 900W, 30mm/s, 15 passes

Figure 4.1.43: 2mm Ti64, 8mm Beam Dia., 1200W, 25mm/s, 15 passes



effect of the coating burn off is now more apparent in the 5.5mm beam diameter data.

For both of the results the predicted bend angle rate is reached within the first few

passes and then the rate falls, with no more forming possible after 8 to 10 passes. It

can be seen that the bend angle rate per pass does not fall off as quickly as in the

1.4mm and 1.6mm coupons. However, it can be seen that the overall amount of

forming is similar despite the increased available number of useful passes.

In order to confirm the coating burn-off theory the samples were re-sprayed

and forming continued up to 25 passes. In addition a second study was conducted re-

spraying the coupons every 5 passes up to 30 passes to ascertain whether a semi-

constant forming rate could be achieved. The results of the studies can be seen in

figures 4.1.44 to 4.1.74.

Figure 4.1.44: 2mm Ti64, 5.5mm Beam Dia., 900W, 30mm/s, Re-spray at pass 15

Figure 4.1.45: 2mm Ti64, 5.5mm Beam Dia., 900W, 30mm/s, Re-spray every 5 passes

Figure 4.1.46: 2mm Ti64, 8mm Beam Dia., 1200W, 25mm/s, Re-spray at pass 15



It can be seen in the above figures that as with the other thicknesses of Ti64

investigated the effect of coating burn-off on the bend angle rate is large for this

material. It can be seen that by re-spraying at 15 passes the bend angle rate can be

increased (but not to the same level as the first few passes) to produce more forming

(figures 4.1.44 and 4.1.46). It can be seen in figures 4.1.45 and 4.1.47 that by re-

spraying the absorptive graphite coating every 5 passes continued forming can occur

with an approximately linear bend angle increase (straight line). The individual bend

angle rates for each new surface coating are presented in the above figures. It can be

seen that the rates fluctuate considerably. For the 5.5mm beam diameter data (figure

4.1.45) there is a significant amount of forming after 30 passes (~18°). The bend

angle rate although rising and falling after each re-coat shows a general downward

trend consistent with other more subtle factors outlined earlier. For the 8mm beam

diameter data (figure 4.1.47) it can be seen that the bend angle rate fluctuates

considerably; an initial peak after each re-spray quickly falls off during subsequent

passes. It can be noted though that the peak values after each re-spray are reasonably

consistent (~1.5°) and as such a large overall bend angle has been produced (~25°).

It is difficult to determine whether there is a downward trend in the data here due to

the large fluctuations in the bend angle rate.

3.2mm Gauge Ti64

This material is considered to be a thick section material due the significant increase

in the section modulus coupled with the high material strength of the alloy (table

3.2.7). At 3.2mm this was the thickest Ti64 sheet investigated. Due to the cost and

availability of this gauge of Ti64, an insufficient quantity was available for a full

Figure 4.1.47: 2mm Ti64, 8mm Beam Dia., 1200W, 25mm/s, Re-spray every 5 passes



process map investigation. In addition it was likely that the process window for this

material would be very small. A small trial and error study was conducted to

determine usable forming process parameters. It was determined that the following

would produce some forming without surface damage; 800W, 7mm beam diameter

and 15mm/s speed. In order to improve on this a new ‘double pass’ technique was

developed. The technique involves a scan strategy of a pass in one direction followed

immediately by a return pass in the opposite direction; the plate is allowed to cool

after each double pass (forced cooled by air jet to decrease process time). The

concept behind this strategy is that, providing the material surface is not damaged on

the second pass, the additional energy input per pass is essentially akin to processing

with a much higher laser power (factor increase dependent on overlapping

interaction times). Another factor in this technique is that on the second pass the heat

retained in the irradiated area from the first pass could serve to produce additional

forming by reducing the temperature dependent flow stress of the material, in that a

hot plate is easier to form than a cold one. The results of a direct comparison of

single and double pass techniques is shown in figure 4.1.48, using 3.2mm Ti64,

800W (Electrox 1.5kW CO2 laser), 7mm beam diameter and a processing speed of

15mm/s.

The results show an increase in achievable bend angle using a double pass

technique when compared to the same number of overall passes using a single pass

strategy. It can be seen that after only a few passes using the single pass technique

the bend angle increase per pass falls off akin to the thinner materials. However,

using the double pass technique forming is continued. This could be due to the

elevated temperatures along the scan line after the initial pass improving absorption

for the second pass as the absorption is proportional to temperature.

Figure 4.1.48: Single & Double Pass Comparison, 3.2mm Ti64 Sheet



A study using thermocouples attached to the surface of a 3.2mm Ti64 sheet

revealed the thermal profiles of the two techniques. Using adhesive thermopads the

K type thermocouples were attached to the top surface of an 80x80mm plate at 7.5

and 9.5mm from the scan line centre. The temperature data was recorded using an

Agilent 34970A data acquisition unit (described earlier), acquiring at up to 250

samples a second. Figure 4.1.49 shows the thermocouple output for a single pass

using the processing parameters outlined earlier for figure 4.1.48.

It can be seen that the maximum temperature recorded at 7.5mm is 120ºC. It

can be assumed that the peak temperature at the centre of the beam will be

considerably higher. Figure 4.1.50 shows the thermocouple output from a Double

Pass, it can be seen that the distinction between the two passes is not that apparent

and that the peak temperature for the second pass is built on the first pass

temperature; the peak temperature at 7.5mm is now 200ºC. This confirms the idea

that the second immediate pass is akin to processing with considerably more power,

in this case approximately double, and that there would be sufficient heat remaining

on the second pass to aid the forming process. Both plates in this study were forced

cooled after processing using a compressed air jet on the bottom surface (~4 bar).

Figure 4.1.49: Thermocouple Analysis Single Pass, 3.2mm Ti64 Sheet

Figure 4.1.50: Thermocouple Analysis Double Pass, 3.2mm Ti64 Sheet



4.1.3 0.9mm AA 1050

The first part of the 2D LF study into this thin section pure aluminium sheet (AA

1050-H14) was to determine a process map. Due to the high thermal conductivity of

the aluminium (table 3.2.12) only a 3mm beam diameter was considered so as to

ensure that the TGM would be active. Several laser power levels were investigated in

the range 200W to 800W. The results can be seen in figure 4.1.51.

It can be seen in figure 4.1.51 that some forming of this relatively weak

material is possible virtually across the whole range of speeds investigated at each

power level. It can be seen that as the laser power is increased the data is shifted

from the bottom left to the top right of figure 4.1.51. This suggests that as more laser

power is realised on the surface a higher traverse speed is required to maintain a

similar thermal gradient profile through the sheet thickness. It can be observed that

that at lower powers (200W) there is an activation traverse speed were the incident

energy fluence reaches a sufficiently high enough level to cause a plastic

compression due to the thermal expansion of the aluminium. As the traverse speed

slows further, the depth and width of the plasticized zone increases with the

increased energy input, thus producing more forming due to an increase in the

induced bending moment. Similarly as the laser power is increased the activation

energy level has already been reached at 90mm/s and a minimum achievable bend

angle per pass can be observed (300W and above) for the speed range of the tables.

Figure 4.1.51: 2D LF process map for 0.9mm AA1050, 3mm Beam Dia.



This could have implications when considering closed loop control of the process

(this will be discussed in a later section).

It can be observed that at the higher powers (400W and above) there is a

maximum single pass forming limit at around 2.75°, such that no further forming can

be achieved by reducing the speed. This could indicate a point where, due to the high

thermal conductivity of the aluminium and small thickness, an optimum heating of

the section occurs and an increasingly higher traverse speed is required, as the laser

power increases, to reduce the magnitude of the thermal gradient through the

thickness and so reduce the bend angle produced. Another possibility is that a change

in mechanism occurs in the transition from the linear increase in bend angle with

decreasing traverse speed and the point where no more forming occurs with further

decrease in speed since as the section becomes uniformly heated (e.g. zero thermal

gradient from top to bottom through the thickness) with the speed reduction, the

buckling mechanism (BM) may be active. This mechanism relies on the

development of an elastic-plastic buckle which can be fed across the sheet from edge

to edge; the amount of forming per pass is not as governed by the traverse speed (or

energy input) as with the TGM, providing the section is uniformly heated. A larger

factor in the BM is the beam size which governs the maximum width of the

plasticized zone and hence the size of the buckle and subsequent bend. If the BM is

active at the lower traverse speeds for the higher powers (figure 4.1.51), this would

explain why no further forming is possible for any further reduction in traverse speed.

In terms of parameters that give a large range of available bend angles per

pass within in the speed range of the tables, the data obtained at 300W is ideal. This

is significant when considering closed loop control of 2D LF and will be discussed

in a later section. For a study on multi-pass LF two laser forming parameter

combinations were selected from figure 4.1.51, these were; 3mm beam diameter,

300W and a speed of 35mm/s; 3mm beam diameter, 800W and a speed of 85mm/s.

A repeatability test was performed using this last parameter set. An inter-pass delay

of 30 seconds was used throughout. This was thought to be adequate due to the high

thermal conductivity of the aluminium. These two parameter combinations were

predicted to give a bend angle per pass of approximately 2° and 2.5° respectively;

they were selected as they are at the extremes of the usable forming parameters in

terms of laser power. The results up to 30 passes can be seen in figures 4.1.52 to

4.1.54.



It can be seen in figures 4.1.52 and 4.1.53 that considerable forming has been

possible at the two energy parameter combinations investigated. It can be seen that a

significant bending rate has been maintained up to 30 passes with no dramatic fall

off as observed in the Ti64 material. For the data at 300W (figure 4.1.52) the

predicted bend angle rate (~2°) is reached on the second pass and is maintained up

Figure 4.1.52: 0.9mm AA1050, 3mm Beam Dia., 300W, 35mm/s, 30 passes

Figure 4.1.53: 0.9mm AA1050, 3mm Beam Dia., 800W, 85mm/s, 30 passes

Figure 4.1.54: 0.9mm AA1050, 3mm Beam Dia., 800W, 85mm/s, 30 passes, Repeatability test



until pass 10. Here a gradual fall off is observed up to 30 passes, however, the rate is

still above 1° per pass at pass 30. For the data at 800W (figure 4.1.53) a similar

distribution of the bend angle rates is observed; however, the fall off past 10 passes

differs slightly in terms of the slope or gradient of the line of best fit. For the coupon

processed at the higher laser power (figure 4.1.53) the extent of the fall off in bend

angle rate is not as large as for the coupon processed at the lower power despite the

similar energy fluence realised. A possible reason for this is that for the higher power

a higher surface temperature maybe realised; the energy fluence is a function of the

intensity and the interaction time so a similar fluence does not necessarily mean the

same temperature is realised. If a higher temperature is realised then the extent of the

factors that have been identified to influence the bend angle rate fall off with

increasing number of passes may differ. These factors include strain or work

hardening, section thickening and coating burn-off. At higher material temperatures

the rate of work hardening will certainly differ and so this may account for the

differences.

It can be seen in the above figures that the effect of coating burn-off is not as

significant as in the Ti64. This was backed up by an inspection of the scan line after

30 passes which showed the graphite coating still intact with little obvious

degradation. It can be assumed however that some burn-off must be taking place and

that this must influence the bend angle rate per pass at higher number of passes. It is

not easily possible to separate out all of the factors. However, as the coating integrity

was such a large factor in the previous material it is likely to be a large factor here.

The only true way determine this and to investigate the other factors is to use a laser

wavelength that may not require an absorptive coating e.g. high power Nd:YAG at

1.06µm, unfortunately access to this laser type with bend angle measurement

capabilities was not available at the time of this investigation. This will be

investigated, however, as part of future planned work by the author.

Presented in figure 4.1.54 are the results of a repeatability test over three

samples at one of the process parameter combinations investigated. It can be seen

that a good repeatability is possible for this material. This may be aided by the fact

that the samples were laser cut from the same sheet of aluminium and thus they

would have a similar residual stress history. It has been reported 118 that material

factors such as differences in the residual stress history of a component can influence



the repeatability of the LF process and as such closed loop control becomes a

requirement.

4.1.4 1.6mm AA 6061 O/T4/T6

A study was conducted into the 2D LF of this 1.6mm gauge 6061 aluminium alloy.

Three heat treatment conditions were considered O (annealed), T4 and T6 (solution

heat treated, cold worked and aged). For this study (as with the previous materials) a

process map was built up for each of the material conditions. Then by selecting

usable parameters an investigation into the factors influencing the multi-pass LF

process was conducted. These factors included laser power, traverse speed, inter-

pass time delay, absorptive coating condition (re-spray) and heat treatment condition.

Tied in with the heat treatment condition of this alloy are variations in material

strength, thermal conductivity and hardness (tables 3.2.15 and 3.2.16).

The process maps for

each of the material conditions

determined from this study are

shown in figures 4.1.55 to

4.1.57. As with the pure

aluminium only a 3mm beam

diameter at a range of laser

powers was considered. This

was to ensure that the TGM

would be active (due to high

thermal conductivity) and to

reduce the number of variables.

Figure 4.1.55: 2D LF process map for 1.6mm AA6061 O, 3mm Beam Dia.

Figure 4.1.56: 2D LF process map for 1.6mm AA6061 T4, 3mm Beam Dia.



It can be seen in the above figures that there are considerable differences in

the laser forming characteristics of the three heat treatment conditions of the

AA6061 alloy. This can be illustrated be considering the bend angle response at

300W. In the O condition data (figure 4.1.55) some forming is possible even at

90mm/s and the bend angle produced increases significantly below 45mm/s. For the

T4 and T6 condition data (figures 4.1.56 and 4.1.57) the data at 300W is shifted

more to the bottom left of the figure. This is perhaps consistent with the increase in

material strength hence a decrease in formability for a given set of energy parameters.

The T6 condition possesses the highest material strength and hence an even slower

traverse speed is required to give some forming. As the power increases a similar

variation can be observed, this makes the selection of useable forming parameters for

a comparison between the material conditions difficult. It can also be noted that at

higher powers a peak in the bend angle response for a given traverse speed can be

observed. This is similar to the response observed in the thinner section Ti64, where

this was attributed to a balance point between efficient energy coupling and loss of

high thermal gradient through the section due to overheating. It is likely that at

higher powers and slower traverse speeds (coupled with high thermal conductivity) a

change in mechanism to the buckling mechanism is occurring.

A nominal set of energy parameters were selected from the above figures in

order to investigate the factors listed earlier during multi-pass LF up to 30 passes.

These energy parameters were: 3mm beam diameter, 500W, 55mm/s and an inter-

pass delay of 30 seconds. It can be observed in figures 4.1.55 to 4.1.57 that these

parameters give a bend angle of approximately 1.5° in each of the material

conditions. A study was then conducted into multi pass LF by varying separately the

power, speed, inter-pass delay and coating re-spray interval whilst keeping all of the

Figure 4.1.57: 2D LF process map for 1.6mm AA6061 T6, 3mm Beam Dia.



other variables constant. It was thought that this approach would yield the influence

of each variable on the process. The results are presented in the following sections:

Effect of Heat Treatment Condition

The effect of the heat treatment condition of the AA6061 alloy on its laser forming

characteristics over 30 passes can be seen in figure 4.1.58.

The first thing of note from the above figure is the dramatic bend angle rate

fall off after 10 passes on the T4 and T6 samples. This is similar to the effect noted

in the study on Ti64, where this was attributed to the coating degradation caused by

localised overheating due to the low thermal conductivity of the material. In this case

the thermal conductivity is high therefore a different factor must be present. Whilst

the coating has not degraded to the same extent as in the Ti64 study, some loss is

present and given that the reflectivity of the substrate (AA6061) is very high to the

incident 10.6µm radiation, this small loss may result in a large drop in absorption

and hence the energy coupled in and bend angle produced. For the O condition the

reflectivity of the surface may be different and so may influence the result somewhat

although not significantly. A more likely explanation for the difference between the

T4, T6 and the O condition in terms of bend angle rate fall off is in the large

difference in material strength, in particular between the relatively weak O condition

and both the higher strength T4 and T6 conditions. This could mean that the

reduction in coupled energy after approximately 10 passes may have less impact on

the weaker material since the level of coupled energy is still above a threshold to

Figure 4.1.58: Effect of heat treatment condition, AA6061, 3mm Beam Dia. 55mm/s, 500W, 30s interval, 30 pass



produce a plastic compression within the scan line. This would also explain why this

effect was not observed in the thinner section pure aluminium presented earlier.

Due to the large bend angle rate fall off at higher numbers of passes for two

of the conditions (figure 4.1.58) the results in terms of stable parameters can only be

used for comparison up to 10 passes. It can be seen that up to 10 passes for the given

forming parameters, the highest bend angle is achieved in the T6 condition and the

lowest in the O condition. This is perhaps not what would be expected as the T6

condition sample has the highest material strength and the O condition the lowest. A

possible reason for this maybe due to the difference in thermal conductivity between

the conditions. The O condition has the highest thermal conductivity of the three

(table 3.2.16) and therefore for the same energy input the heat will be transferred

more quickly into the section therefore reducing the thermal gradient between the

upper and lower surfaces and hence the bend angle. This demonstrates that the

thermal conductivity of a material is a large factor in the laser forming process.

Effect of Laser Power

The effect of incident laser power on the laser forming characteristics of each of the

heat treatment conditions during multi-pass laser forming up to 30 passes are shown

in figures 4.1.59 to 4.1.61.

Figure 4.1.59: Effect of incident laser power, AA6061 O, 3mm Beam Dia. 55mm/s, 30s interval, 30 passes



It can be seen in the above figures that a bend rate fall off after approximately

8 to 10 passes is present to some degree in all of the parameter combinations tested.

Due to this only the first 8 passes can be considered for comparison. It can clearly be

observed that the bend angle increases significantly with increasing laser power, it

can also be seen that above a certain power level this increase in achievable bend

angle levels off. This is backed up by the process map data (figures 4.1.55 to 4.1.57)

where at higher laser powers for a given speed a plateau is reached in achievable

bend angle per pass. This could signify a point above which optimum heating of the

section has occurred to give a maximum bend angle per pass for a given beam size;

further increase in energy input may lead to a loss in this efficiency.

For the O condition (figure 4.1.59) continued forming up to 30 passes has

been possible. It can be noted, however, that the bend angle increase per pass does

fall off to some degree with increasing number of passes after 10 passes, a major

Figure 4.1.60: Effect of incident laser power, AA6061 T4, 3mm Beam Dia. 55mm/s, 30s interval, 30 passes

Figure 4.1.61: Effect of incident laser power, AA6061 T6, 3mm Beam Dia. 55mm/s, 30s interval, 30 passes



factor in this is likely to be the coating degradation (given the effect in the other

conditions). The unpredictable nature of this can also be observed in figure 4.1.59,

where the data obtained at 500W shows a very different fall off rate when compared

to the other laser powers tested.

For the data obtained from the T4 condition material (figure 4.1.60), it can be

observed that at 400W a relatively low forming rate can be maintained up to 30

passes with little fall of in bend angle increase per pass. As the power increases a

considerable fall off occurs, this may indicate a threshold energy level above which

significant burn-off of the coating occurs. It can also be noted that, the pass number

the fall off begins at, increases with increasing laser power. This suggests that a

similar amount of coating is lost after approximately 10 passes (i.e. a similar

resultant absorption coefficient) such that a higher laser power would give a higher

coupled energy level and hence continued forming for several passes more.

For the T6 condition (figure 4.1.61) a similar distribution to the T4 condition

can be observed. However, the bend angle rate fall off is more acute. It can be noted

that no power level has been found (of those tested) to give the same small but

consistent bend angle rate as observed in the T4 data for 400W. This could be a

result of the higher material strength of the T6 condition, in that the reduction in the

coupled energy has a significant effect on the bend angle produced.

Comparing the data for all three of the heat treatment conditions it can be

seen that a similar distribution to figure 4.1.58 occurs across the power levels

investigated. Considerably more forming is possible in the higher strength T6 and T4

conditions than in the lower strength O condition using the same energy parameters.

This can be attributed to the considerable differences in the forming characteristics

between the materials (figures 4.1.55 to 4.1.57), metallurgical change or the

difference in thermal conductivity outlined earlier.

Effect of Processing Speed

The effect of processing speed on the laser forming characteristics of each of the heat

treatment conditions during multi-pass laser forming up to 30 passes are shown in

figures 4.1.62 to 4.1.64.



It was clear from the process maps presented earlier that the processing speed

for a given spot size and laser power has a significant effect on the bend angle

produced. This is backed up by the above figures, where, over the first 10 passes

(without significant coating loss adding to the problem), the decreasing processing

speed results in an increase in the bend angle produced consistent with the increase

in energy coupled into the surface. For the O condition (figure 4.1.62), for the first

Figure 4.1.62: Effect of processing speed, AA6061 O, 3mm Beam Dia. 500W, 30s interval, 30 passes

Figure 4.1.63: Effect of processing speed, AA6061 T4, 3mm Beam Dia. 500W, 30s interval, 30 passes

Figure 4.1.64: Effect of processing speed, AA6061 T6, 3mm Beam Dia. 500W, 30s interval, 30 passes



10 passes, the difference between the samples process at the speeds investigated is

small, this is consistent with the process map data presented in figure 4.1.55 where,

for 500W, the speed range investigated corresponds to a plateau or maximum in the

bend angle produced. This bend angle response maybe due to the higher thermal

conductivity of the O condition material since for the given speed range little

difference in the thermal profile through the thickness is realised. This also

corresponds to a balance point between the energy coupled into the surface and a

loss of high thermal gradient due to over heating of the section.

Effect of Inter-Pass Time Delay

The effect of inter-pass time delay on the laser forming characteristics of each of the

heat treatment conditions during multi-pass laser forming up to 30 passes are shown

in figures 4.1.65 to 4.1.67.

Figure 4.1.65: Effect of inter-pass time delay, AA6061 O, 3mm Beam Dia. 500W, 55mm/s, 30 pass

Figure 4.1.66: Effect of inter-pass time delay, AA6061 T4, 3mm Beam Dia. 500W, 55mm/s, 30 pass



The reason for having a delay in between each alternating direction pass is to

allow the coupon to cool somewhat (if no additional cooling is used) so as to not

melt the surface or adversely alter the microstructure on subsequent passes. A

balance may be reached, however, between too short a delay whereby the material is

damaged, a delay whereby the bulk material temperature increases significantly thus

the high thermal gradient through the thickness is lost, and a delay whereby the

remaining heat within the plate aids the process by reducing the temperature

dependent flow or yield stress, in that a hot plate is easier to form than a cold one. It

can be seen in the above figures that the inter-pass time delay does have an effect on

the bend angle produced. Once again, however, the dramatic fall off in bend angle

rate per pass after 8 to 10 passes (attributed to some coating loss and high reflectivity

of the alloy surface) influences the final outcome after 30 passes considerably, such

that only the first few passes can be considered for a true comparison.

For the T6 condition data (figure 4.1.67) it can be observed that a peak

forming rate occurs using a 50 second interval. It can also be noted that the coating

degrades much faster than at the other time delays investigated. This is unlikely to be

anything to do with the time delay selection but more likely to be a variation in the

coating thickness due to the manual method of application. This demonstrates

another significant problem with using absorptive coatings with the LF process; if

the condition of the coating becomes critical, as with this material and the Ti64, then

the process becomes more sensitive to small variations in coating thickness.

For the T4 condition (figure 4.1.66) it can be seen that there is little

difference in the forming result (up to 10 passes) between the 50, 70 and 90 second

data, with the 50 second delay producing slightly more forming (although not as

pronounced as in the T6 data). The 30 second delay produces less forming than the

Figure 4.1.67: Effect of inter-pass time delay, AA6061 T6, 3mm Beam Dia. 500W, 55mm/s, 30 pass



other intervals investigated. This may indicate that at this short time delay in-

between each pass a reduction in the thermal gradient through the thickness is

occurring and hence a reduced bend angle for the same energy parameters is

produced.

For the O condition (figure 4.1.65) little difference is observed between all of

the time intervals investigated, with the 30 second delay producing only a slightly

smaller bend angle after 8 to 10 passes. This could be due to the higher thermal

conductivity of this heat treatment condition since the faster rate of heating at the

lower time intervals is compensated by the higher heat transfer rate into the bulk

material and the aluminium edge clamp and hence no significant difference can be

observed.

Effect of Absorptive Coating Condition

The effect of absorptive coating condition (graphite spray) on the laser forming

characteristics of each of the heat treatment conditions during multi-pass laser

forming up to 30 passes is shown in figures 4.1.68 to 4.1.70. The coating was re-

sprayed at different intervals to ascertain the significance of the coating condition on

the process.

Figure 4.1.68: Effect of coating re-spray interval, AA6061 O, 3mm Beam Dia. 500W, 55mm/s, 30 passes, 30s interval

Figure 4.1.69: Effect of coating re-spray interval, AA6061 T4, 3mm Beam Dia. 500W, 55mm/s, 30 passes, 30s interval



It can clearly be seen in the above figures that the absorptive graphite coating

condition has a significant effect on the forming process when using multiple passes

over the same track. This data also confirms the explanation stated earlier for the

drop in bend angle rate in this alloy after 8 to 10 passes, in that the loss or burn off of

the coating is responsible for this significant fall off. It can be observed in all of the

heat treatment conditions (figures 4.1.68 to 4.1.70) that a consistent bend angle

increase can be maintained (past 10 passes) by re-spraying the irradiated scan line

with the graphite absorptive coating. The more frequent the re-spray the more

consistent the bend angle increase per pass; every 5 passes has produced the more

consistent result, even in the O condition where the fall off was not that acute. It can

be noted that in the samples re-sprayed at 15 passes the bend angle rate has fallen off

again some 8 to 10 passes later thus re-confirming the coating degradation theory.

In the samples re-sprayed every 5 passes it can be seen that there is still a

more subtle fall off in bend angle increase at higher numbers of passes. This more

subtle effect is attributable to the other metallurgical factors identified as influencing

the process, these are section thickening (a thicker material is harder to form) and

strain or work hardening (reducing the ductility of the material).

Figure 4.1.70: Effect of coating re-spray interval, AA6061 T4, 3mm Beam Dia. 500W, 55mm/s, 30 passes, 30s interval



4.2 Thermal Analysis

Thermocouple and thermal imaging techniques were used in the investigations

presented in this section to experimentally determine the transient temperature field

in a component during the laser forming process and subsequent cooling. A study

was also conducted into the effectiveness of using forced cooling in the LF process

and its effect on forming efficiency.

4.2.1 Thermocouple Analysis

A study was conducted using a thermocouple technique into the temporal

temperatures cycles at single locations on the upper and lower surfaces of 1.5mm

mild steel CR4 during single and multi-pass 2D LF (details given in chapter 3.2.2.1).

Three processing parameter sets were investigated, chosen from the empirical study;

3mm beam diameter, 760W, 55mm/s; 5.5mm beam diameter, 760W, 30mm/s; 8mm

beam diameter, 760W, 20mm/s, inter-pass delays of 24 seconds and 60 seconds were

investigated. A study using thermocouples placed at distances of 10, 22, 34, 46, and

58mm from the scan line along the centre of 80x200mm coupons (figure 3.2.2). The

thermocouple output from a single pass at the three energy parameter combinations

stated are given in figures 4.2.1 to 4.2.3.

Figure 4.2.1: Thermocouple output at various locations, 3mm Beam Dia. 55mm/s, 760W, 1 pass



It can be seen in the above figures that the temperature observed at the

nearest location to the centre of the beam, 10mm, are relatively small. It can be noted

that some data for the measured locations is missing, this is due to a failure in the

adhesive pad holding the thermocouple to the plate during a pass and so the data was

unusable. As the nominal measurement distances from the centre of the scan line are

the same for each of the beam diameters investigated, as would be expected, the

temperature recorded on the upper and lower surfaces increases with increasing

beam size. This does not necessarily indicate that a higher peak temperature has been

realised with the larger beam diameter energy parameters, but that the edge of the

beam is closer to the first measurement point and the heat has not dissipated into the

bulk material over this distance as efficiently as in the smaller beam diameter

parameters. In addition for the 3mm beam diameter the significant heating would be

concentrated more in the upper surface area and this would be quenched rapidly by

the cold material below and surrounding the heated zone. The overall energy input,

however, using the larger beam diameter maybe higher (increased time to cool

Figure 4.2.2: Thermocouple output at various locations, 5.5mm Beam Dia. 30mm/s, 760W, 1 pass

Figure 4.2.3: Thermocouple output at various locations, 8mm Beam Dia. 20mm/s, 760W, 1 pass



indicates this), this being more distributed over the surface hence a lower peak

temperature but more significant heating into the section, examination of the relative

size of the heat affected zones (observable darkening of the surface) on the upper and

lower surfaces on the processed samples confirms this.

The thermocouple output over 6 passes using the same energy parameters

given above at time intervals of 60 and 24 seconds are given in figures 4.2.4 to 4.2.8.

Figure 4.2.4: Thermocouple output at various locations, 3mm Beam Dia. 55mm/s, 760W, 6 passes, 60 second intervals


Figure 4.2.6: Thermocouple output at various locations, 5.5mm Beam Dia. 30mm/s, 760W, 6 passes, 60 second intervals



It can be noted in the above figures that some of the data was unusable after a

number of passes, this was due to the adhesive thermo-pad failing (peeling back)

above a temperature around 100°C. Although not ideal, the use of thermo-pads

meant that the welded tip of the thermocouple could be used again, this would not be

the case if a higher temperature resistant adhesive was used to fix the tip directly. It

can also be noted that only the data for 60 second intervals is presented for the

5.5mm beam diameter processing parameters, the data for passes at 24 second

intervals was not available for inclusion.

The main observation from figures 4.2.4 to 4.2.8 is that the temperatures

recorded are increasing with increasing numbers of passes for all the energy

parameters investigated. The peak temperature observed during each pass at each

location increases also, however, the temperature increase is roughly the same for

each pass (same amount of energy added each time), it is the bulk material

temperature this increase is added onto which is increasing. This can be seen in





figure 4.2.4 where the data recorded at 58 mm from the scan line, which can be

considered the bulk material temperature (as far as that which influences the scan

line area), is increasing with increasing numbers of passes. This effect may have

implications on the efficiency of the process for subsequent passes, in that if the bulk

material temperature is increasing there maybe a reduction in magnitude of the

thermal gradient through the section directly under the beam (consistent with TGM).

Another factor is the elevated temperatures remaining in the heated area (bulk

material temperature increase) aiding the process by reducing the temperature

dependent flow or yield stress of the material thus making it easier to plastically

deform.

Comparing between the results from the three different beam diameters (60

second intervals) it can be seen that a similar distribution is present here to that

observed earlier for a single pass. The lowest peak temperatures at the locations

investigated are observed in the 3mm beam data (figure 4.2.4) and the highest in the

8mm beam diameter data. As was discussed earlier this does not necessarily mean

that the peak temperatures within the scan line are higher for the larger beam, it is

more likely that this is a factor of the relative beam size to the measurement

locations. The smaller beam diameter is likely to cause a higher peak temperature in

the coupon as the intensity is higher, this should be confirmed by the Finite Element

Analysis (FEA) work presented in a later section.

It can be seen in figures 4.1.5 and 4.1.8 that by decreasing the scan interval to

24 seconds (from 60 seconds) has a significant effect on the temperatures recorded

on the surfaces of the coupons. This significant increase in temperature observed at

the locations investigated must also occur within the scan line. This backs up the

results and discussion presented earlier in the empirical section on this material at

various inter-pass time delays (figures 4.1.7 to 4.1.12), in that it was observed that

the time delay has a significant effect on the final bend angle produced in a multi-

pass strategy. For the 3mm beam diameter processing parameters it was found that a

shorter inter-pass time delay lead to an increase in the amount of forming (24

seconds produced the highest). This is consistent with the thermocouple data

presented in this section (figures 4.2.4 and 4.2.5), where an increase in the

temperature realised in the coupon is apparent when comparing the 60 second and 24

second interval data. If the plate was significantly warmer within the scan line for

subsequent passes at 24 second intervals then the reduction in the flow stress should



lead to more forming for the same energy parameters. For the 8mm diameter beam

(figure 4.2.7 and 4.2.8) the effect of the time delay was less acute on the bend angle,

however it was found that a slight increase occurred at a 60 second interval. This

maybe due to the excessive heating observed in figure 4.2.8 after only 3 passes (no

more data available), if the section under the beam and surrounding material is over

heated during each pass because there is no time to cool sufficiently, then the process

efficiency must decrease as a high thermal gradient cannot be maintained.

As the peak temperature increases per pass and appears that it would increase

still further after 6 passes, a study was conducted to determine the temporal thermal

output of a 10 pass strategy. This can be seen in figure 4.2.9, an 8mm beam diameter,

760W, 20mm/s traverse speed and an inter-pass delay of 40 seconds was used.

Perhaps as would be expected, it can be seen in figure 4.2.9 that a plateau is

reached after a number of passes whereby there is no significant increase in the peak

temperature recorded per pass when compared to the previous pass. This is likely to

be a point where thermal equilibrium is reached, in that the bulk material

temperature of the whole plate (not just the area surrounding the scan line) has

increased and the heat losses due to conduction into the clamp, convection to the air

and radiation to the surroundings are balanced with the heat input per pass. The point

at which this equilibrium or stabilisation occurs must be governed by the energy

input parameters and the inter-pass delay, in that the greater the heat input and the

rate of heat input the more passes required before a stable thermal cycle occurs. This

effect could be an explanation for the initial increase in bend angle rate per pass

observed during the first five or six passes of a number of materials at various energy

parameters (section 4.1). If the peak temperature per pass within the scan line is

Figure 4.2.9: Thermocouple output, 8mm Beam Dia. 20mm/s, 760W, 10 passes, 40 second intervals



increasing for the first few passes until an equilibrium point is reached then the

amount of forming (governed by the energy input) must also increase for the first

few passes. Once a stable thermal cycle has been established a reasonably consistent

bend angle rate is observed which then drops due to the other factors influencing the

bend angle rate discussed earlier.

4.2.2 Thermal (IR) Imaging Study

The parameter values utilised for this section are given in tables 3.2.17 and 3.2.18.

The images were produced and analysed using the software package Irwin OLE V2.

The emissivity was set at 0.6 (typical for a graphite surface) and the temperature

range given as 40ºC to 1100ºC.

Two-dimensional thermal images of the start, middle and end of the laser

processing and post-processing cooling of the 1.5mm CR4 mild steel coupons, for

each of the beam diameters 3mm, 5.5mm and 8mm, are given in figures 4.2.10,

4.2.11 and 4.2.12, respectively.

The temperature data for analysis was obtained from the laser processing

end-scan image for each beam diameter, using the software post-processor.

Comparisons between the temperature distributions, during laser processing, for the

three process parameter combinations are given in figure 4.2.13. Temperature

distributions for the 3mm, 5.5mm and 8mm diameter laser beams when incident

upon the mild steel coupons are given in figures 4.2.14, 4.2.15 and 4.2.16,

respectively.

The thermal images obtained using the infrared detector all exhibited a

Gaussian-type form and this can be attributed to the distribution of energy within the

beam mode of the CO2 laser utilised in this investigation (figure 3.1.3).

Data acquired using the infrared detector was limited to defined ranges

between -20ºC and 1500ºC. A range of 40ºC to 1100ºC was selected so as to obtain

reasonable heating and cooling temperatures during the laser processing of the

samples. The imaging of the temperature field on the surface of the coupons could

only be considered accurate at positions surrounding the incident laser spot where

the temperatures fell within the detection range.



1. Start of Laser Scanning 2. Middle of Scan

3. End of Scan 4. Start of Cooling

5. Cooling (5 seconds) 6. Cooling (20 seconds)

Figure 4.2.10. 2D Thermal Images Obtained for the 3mm Beam Diameter with Laser Power 760W and Scan Velocity 55mm/s [Centre red spot implies temperatures > 1100ºC]

Reflections from MEL sensor and clamp



1. Start Laser Scanning 2. Middle of Scan

3. End Scan 4. Start of Cooling


Figure 4.2.11. 2D Thermal Images Obtained for the 5.5mm Beam Diameter with Laser Power 760W and Scan Velocity 30mm/s [Centre red spot implies temperatures > 1100ºC]

Reflections from MEL sensor and clamp



1. Start Laser Scanning 2. Middle of Scan

3. End Scan 4. Start of Cooling


Figure 4.2.12. 2D Thermal Images Obtained for the 8mm Beam Diameter with

Laser Power 760W and Scan Velocity 20mm/s [Centre red spot implies temperatures > 1100ºC]



Figure 4.2.13: Comparison of the Temperature Distributions for the 3mm, 5.5mm and 8mm Diameter Laser Beams

Figure 4.2.14: Temperature Distribution for the 3mm Incident Beam

Comparison of Temperature Distributions for the 3mm , 5.5mm and 8mm Diameter Beams

0

50

100

150

200

250

300

350

400

450

500

550

600

650

700

750

800

850

900

950

1000

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55

Thermal Data Point

Tem

pera

ture

(Deg

. C)

5.5 mm BeamDiameter3 mm Beam Diameter

8 mm Beam Diameter

Temperature Profile for 3mm Beam Diameter

0

200

400

600

800

1000

1200

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59

Thermal Data Pointa Along Beam Scan Path

Tem

pera

ture

(Deg

. C)

Laser Direction

Heated Material Heat

Retention Increasing with Beam Spot Size

Unprocessed Section of Mild Steel Coupon

Laser Processed Section of Mild Steel Coupon

Laser Beam Direction

Heating Cooling

Tail Zone

A1 Temp.



Figure 4.2.15: Temperature Distribution of the 5.5mm Incident Beam

Figure 4.2.16: Temperature Distribution of the 8mm Incident Beam

Temperature Profile For 5.5mm Beam Diameter

0

200

400

600

800

1000

1200

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59

Thermal Data Points Along Beam Scan Path

Tem

pera

ture

(Deg

. C)

Temperature Profile for 8mm Beam Diameter

0

200

400

600

800

1000

1200

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59

Thermal Data Points Along Beam Scan Path

Tem

pera

ture

(Deg

. C)


Unprocessed Section

Laser Processed Section of Mild Steel Coupon

Unprocessed Section

Laser Processed Section of the Mild Steel Coupon


Heating

Cooling

Cooling

Heating

Tail Zone

Tail Zone



At the point of incidence between the laser beam and the surface of the

coupons the infrared detector gave measurements that were in excess of 1100ºC

(shown as a centre red spot in the thermal images). This could be due to several

factors:

(i) Correct measurement of a Gaussian-type temperature distribution, with the

temperature being measured at the point on the surface where the energy

distribution of the laser beam was at its maximum.

(ii) Ionisation of the surrounding air above the surface of the mild steel coupons.

(iii) System limitations of the peripheral hardware and software.

(iv) Imaging of the temperature field generated by the graphite coating during

absorption of the laser beam energy.

(v) Incandescence of the graphite coating.

These discussion points are shown, graphically, in figure 4.2.13, which

indicates the temperature distribution, and the temperature measurement overshoot,

for each of the beam diameters. It is likely that the incandescence of the graphite due

to the interaction with the incident laser beam is the cause of the measurement

overshoot, in that it is unlikely that the peak surface temperature in the mild steel

will reach the levels indicated by the IR analysis. This is backed up by observations

during the process where a bright burning can be seen on the graphite surface under

the laser beam. Although the graphite heats up to temperatures in excess of 1100ºC

(possibly much higher), it is the overall heat transfer to the mild steel beneath that

determines the absorption coefficient for the laser type and material. This

demonstrates another problem with using this thermal imaging technique when using

absorptive surface coatings, in that it is not possible to record the peak surface

temperatures realised in the mild steel substrate through the graphite coating when

there are significant differences in temperatures between them.

The time period and surface area over which heat was retained in the mild

steel sheet increased for larger laser beam spot diameters. This can be seen in the

cooling images for each of the beam diameters shown in figures 4.2.10 to 4.2.12.

This was observed in the thermocouple data presented earlier and may have resulted

in the energy absorption being enhanced for larger beam diameters during multiple-

pass scanning regimes due to the increase in the temperature dependent absorption

coefficient. This maybe an additional factor in the initial increase per pass in the bulk



material and peak scan line temperatures and hence bend angle observed during the

first few passes.

Figure 4.2.13 highlights the differences between the temperature distributions

for each of the three beam diameters utilised in this investigation. The mid scan plots

in figures 4.2.14 to 4.2.16 show the effect of heat transfer, with each case showing

that there is a pre-heated zone in front of the laser beam and a tail zone following the

laser beam. From figure 4.2.13 it can be seen that the temperature in the tail zone

increased as the beam diameter was increased, figures 4.2.14 to 4.2.16 also imply

this distribution for tail zone temperatures. This is consistent with the discussion

point raised in the last section from the thermocouple analysis study, whereby the

larger beam diameters showed a higher temperature at the distances recorded and a

longer time to cool, however, it was argued that this does not necessarily indicate

that a higher peak temperature has been realised. It is more likely that for the smaller

beam diameters (3mm) the significant heating (with higher peak temperatures)

would be concentrated more in the upper surface area and this would be quenched

rapidly by the cold material below and surrounding the heated zone thus the heat

retention behind the beam would be less. For the larger beam diameters (8mm) the

overall heat input maybe higher but this is more distributed over the surface and

through the section thus the peak temperature would be less.

The heat distribution on cooling in figures 4.2.10 to 4.2.12 reveals (based on

heat retention after the beam has passed) that a higher temperature is realised at the

end of the scan line when compared to the beginning. This is backed up by

observations of the HAZ at the end of a scan line, where a widening or flaring can be

seen. A possible explanation for this is that the heat from the incident laser beam and

the heat retained behind the beam is flowing into the cold region ahead of the beam,

as the beam reaches the second edge the heat flowing ahead of the beam cannot

travel any further and so a heat build up occurs, hence the increase in temperature at

the second edge. This demonstrates the need for an alternating direction strategy to

even up this temperature distribution along the scan line. In addition the fact that a

higher temperature maybe realised at the end of the scan line means that more

forming maybe realised there, this could be a source of unwanted distortion in the

process (e.g. edge effects) and suggests a need to reduce the energy input near the

edge to account for this heat build up. This could be achieved by varying the traverse

speed along the scan line (e.g. speed up towards the edge).



4.2.3 Forced Cooling Study

It was shown in the empirical section that the dwell time in between each pass had a

significant effect on the bend produced in a multi-pass strategy. This was

emphasised by the thermocouple data which showed that there was a significant rise

in the temperature realised in a component as the dwell time was reduced, this could

be beneficial or detrimental to the bending efficiency depending on the beam

diameter used. As this adds an extra complication to the process it was decided to

investigate the use of forced cooling. An additional potential benefit to the use of

cooling is a decreased overall processing time, in that the relatively long inter-pass

delay could be reduced significantly.

A study was conducted with and without the compressed air cooling jet on

the 1.5mm mild steel coupons (figure 3.2.8) using three process parameter

combinations; 3mm beam diameter, 760W, 55mm/s; 5.5mm beam diameter, 760W,

30mm/s; 8mm beam diameter, 760W, 20mm/s; the time interval between passes was

40 seconds. The results of the first study to ascertain the effectiveness of the

compressed air cooling jet using a thermocouple method are shown in figures 4.2.17

to 4.2.22.

Figure 4.2.17: Thermocouple Output, 3mm Beam Dia. 55mm/s, 760W, 4 passes, 40 second intervals, no cooling

Figure 4.2.18: Thermocouple output, 3mm Beam Dia. 55mm/s, 760W, 4 passes, 40 second intervals, With cooling



Figure 4.2.19: Thermocouple output, 5.5mm Beam Dia. 30mm/s, 760W, 4 passes, 40 second intervals, no cooling

Figure 4.2.20: Thermocouple output, 5.5mm Beam Dia. 30mm/s, 760W, 4 passes, 40 second intervals, With cooling

Figure 4.2.21: Thermocouple output, 8mm Beam Dia. 20mm/s, 760W, 4 passes, 40 second intervals, no cooling

Figure 4.2.22: Thermocouple output, 8mm Beam Dia. 20mm/s, 760W, 4 passes, 40 second intervals, With cooling



It can clearly be seen in the above figures that the addition of a basic cooling

regime influences the thermal cycle in the coupons considerably. Without cooling

using a 40 second inter-pass delay the temperature ramps up significantly over the 4

passes investigated even at the acquisition point distances from the scan line. With

the continuous addition of a cooling air jet on the under surface of the coupon during

and post processing the temperature cycle stabilised within 1 pass and very little

increase in peak temperatures were observed for subsequent passes at all three of the

energy parameters investigated. This can be attributed to the efficient reduction in

the bulk material temperature after each pass, such that although the same

temperature increase can be seen per pass (similar energy input per pass), this is not

now added to an elevated bulk material temperature as is the case without cooling. It

can also be noted from the above figures that by having the cooling present during

processing the temperature increase per pass has reduced slightly, this may influence

the process efficiency an indicate that only a post-processing cooling regime should

be used. The more effective the cooling solution used the shorter the inter-pass delay

required and thus the overall processing time can be reduced.

In order to ascertain the

effect of the cooling regime

used on the process efficient

the bend angle per pass was

recorded for all the processing

parameters investigated, the

results are shown in figures

4.2.23 to 4.2.25.

Figure 4.2.24: 5.5mm Beam Dia. 30mm/s, 760W, 30 passes, 40

second intervals, with and without cooling

Figure 4.2.23: 3mm Beam Dia. 55mm/s, 760W, 30 passes, 40 second intervals, with and without cooling



It can be seen in figure 4.2.23 to 4.2.25 that there several effects on the bend

angle produced when forming with the cooling regime depending on the processing

conditions used. For the 3mm beam diameter conditions (figure 4.2.23) the use of

the forced cooling produces an increase in the amount of forming over 30 passes. A

possible reason for this could be an increased thermal gradient through the section

due to the lower bulk material temperature and the fact that the cooling is realised on

the lower surface of the coupon. It can be noted, however, that the bend angle rate

per pass is reasonably similar for the two processing conditions after 6 to 8 passes.

This could be related to the effect noted in the thermocouple study earlier whereby it

takes several passes before a stable thermal cycle occurs through the establishment

of thermal equilibrium, thus at higher numbers of passes similar bending rates to the

regulated thermal cycle in the cooled samples can be observed.

For the 5.5mm beam diameter conditions (figure 4.2.24) little difference

between the processing conditions can be observed. For the 8mm beam diameter

conditions (figure 4.2.25) the opposite effect to the 3mm beam data can be observed,

where the sample produced with no cooling has formed more than the sample with

cooling over 30 passes. This is consistent with the idea that although more energy is

transferred into the coupon using the 8mm beam diameter conditions the peak

temperature maybe less than for the smaller beam diameters, thus by forming with

cooling the peak temperature may drop such that this would be akin to forming with

lower laser power and hence the bend angle would also drop.

Although the effect of forced cooling on the LF process in terms of bend

angle produced is subtle, the reduction in processing time gained makes its use

essential. In addition reducing the thermal input into a component must be beneficial

both for the reduction in unwanted distortion and any adverse effects on metallurgy.

Figure 4.2.25: 8mm Beam Dia. 20mm/s, 760W, 30 passes, 40 second intervals, with and without cooling



4.3 Displacement / Time Analysis

An investigation was conducted into the displacement (or bend angle development)

of 80x200x1.5mm mild steel CR4 coupons with respect to time. A laser range finder

was used at a single location to record the displacement during LF at three different

processing parameters (details in chapter 3.2.3). The results recorded during 6 passes

are shown in figures 4.3.1 to 4.3.9; given is the data for all 6 passes plus the 1st and

6th pass isolated and expanded for further analysis. Also indicated is the time the

laser beam was on the coupon for (based on the speed and point of first movement).

Figure 4.3.1: Displacement/Time, 3mm Beam Dia. 760W, 55mm/s, All 6 passes, 60s int.

Figure 4.3.2: Displacement/Time, 3mm Beam Dia. 760W, 55mm/s, pass 1, 60s int.




Figure 4.3.4: Displacement/Time, 5.5mm Beam Dia. 760W, 30mm/s, All 6 passes, 60s int.

Figure 4.3.5: Displacement/Time, 5.5mm Beam Dia. 760W, 30mm/s, pass 1, 60s int.

Figure 4.3.6: Displacement/Time, 5.5mm Beam Dia. 760W, 30mm/s, pass 6, 60s int.

Figure 4.3.7: Displacement/Time, 8mm Beam Dia. 760W, 20mm/s, All 6 passes, 60s int.



It can be seen in the above figures that the temporal displacement

characteristics of the coupons during LF depend greatly on the energy parameters

used and the number of passes realised. For the 3mm beam diameter data (figures

4.3.1 to 4.3.3) on the first pass (figure 4.3.2) the major part of the bend angle

development can be seen to occur whilst the beam is still on the plate surface, very

little additional movement was recorded after the beam has left the coupon surface.

Additionally it can be noted that the counter-bend effect or initial negative bending

due to thermal expansion (consistent with TGM theory), is extremely small in terms

of magnitude and time taken when compared to the overall deformation, this was

still present for all six of the passes recorded. By the 6th pass (figure 4.3.3) it can be

observed that, although the majority of the bend angle occurs whilst the laser beam is

on the surface of the coupon, the final deformation or bend angle isn’t reached until

some 20 seconds after processing. This effect becomes more prevalent with

increasing numbers of passes.





For the 5.5mm beam data (figures 4.3.4 to 4.3.6) on pass 1 (figure 4.3.5) it

can be seen that a similar deformation output to the first pass of the 3mm beam data

was recorded. As before the small counter-bend effect is observed at the start of the

scan and the majority of the deformation occurs whilst the beam is on the surface,

only a small increase in bend angle is observed after processing. The rate of

deformation is lower (more drawn out) than with the 3mm beam due to the lower

traverse speed used, hence it takes longer to reach the other side of the plate to

complete the bend angle. For the 6th pass (figure 4.3.6) the counter-bend is barely

observable (appears to get smaller with increasing numbers of passes) and the

temporal bend angle development has taken on an ‘S’ curve formation, this also

becomes more prevalent with increasing numbers of passes. As with the 3mm beam

data increasingly more deformation occurs after the laser has left the sheet with

increasing numbers of passes, by pass 6 it is taking some 12 seconds after processing

for the final bend angle to be achieved.

For the 8mm beam data (figures 4.3.7 to 4.3.9) on the first pass (figure 4.3.8),

as with the other process parameters investigated, an initial counter bend can be

observed followed by the majority of the deformation with the laser beam on the

coupon. Significantly more deformation after processing occurs, however, for these

energy parameters on the first pass when compared to the others investigated. The

slight ‘S’ curve bend angle development observed after several passes in the 5.5mm

beam data is present in the 8mm beam data during pass 1. On pass 6 (figure 4.3.9) it

can be seen that there is a significant change in the temporal bend angle development,

with the slight ‘S’ curve formation observed at pass 1 has become more extreme and

has in fact become two points of inflection. Here during the laser pass the initial

positive bend angle development has been arrested and has been negated before the

bend angle development continues. An additional observation is that increasingly

more of the deformation occurs (over increasing numbers of passes) after the laser

has left the coupon during cooling, after pass 6 it takes some 27 seconds before the

final bend angle is achieved.

It was observed in the above results that the counter-bend effect is very small

using the energy parameters investigated and that the effect diminishes with

increasing numbers of passes and with increased beam diameter (and lower traverse

speed). This can be compared to the work published by Vollertsen23 (figure 2.6.2)

where the measured counter-bend was well defined and took up approximately one



third of the deformation cycle time. The data recorded in this study would suggest

that the counter-bend is not as significant an event as this during the LF process

using the TGM. A possible reason for this difference is the beam diameter selection

used in this investigation. The TGM theory as proposed by Vollertsen 23 suggests a

beam diameter equal to or of the order of the sheet thickness, i.e. 1.5mm in this case,

it was found in this work, however, that a beam diameter smaller than 3mm caused

excessive surface damage at the power levels used. It was argued (earlier) that the

mechanism works by the setting up of a high thermal gradient through the thickness

such that the generated plastic compression would be asymmetric though the section

and hence a bend towards the laser work occur. This should be still be possible to set

up with larger beam diameters providing the energy input is high and the thermal

conductivity of the material is relatively low, the results presented demonstrate that

this is the case. By using larger beam diameters in conjunction with the TGM

appears to effect the temporal bend angle development, this could be due to the

increased depth of heating present causing a more lateral or in-plane displacement

during the initial thermal expansion rather than significant out of plane displacement

present where the heating is very localised to the upper surface area.

As the number of passes and beam diameter increases the heat retained in the

sample and time to cool increases also. This could be an explanation for the

increased time taken for the final bend angle to be reached observed in the above

results, in that as the purely elastic stresses are relieved during cooling the full extent

of the plastic compression or shrinkage in the upper surface along the scan line is

realised, hence the bend angle increases. Another possible factor in this is the

development of edge effects whereby the bend angle maybe different depending on

the location measured, the results from this study may indicate that the edge effects

develop after the beam has passed during cooling, this is backed up by other work on

the shape measurement of the process 129. If the plate geometry changes during

cooling then this may influence the displacement measurement recorded, multiple

measurement locations would confirm this.

The unusual temporal displacement or ‘S’ curve data observed using the

larger beam diameters at higher numbers of passes could be attributed to a change in

mechanism to the buckling mechanism (BM) or the increased in-plane movement

discussed earlier due to the hybrid TGM conditions, in that as the plate flexes as the

beam is drawn across the scan line there is a mechanical effect ahead and to the rear



of the beam that influences the displacement realised at the measurement point at the

centre. An alternative or development on this could be a delayed counter-bend effect

due to the slow traverse speed and larger beam diameter. In that, for the results

shown in figure 4.3.9 especially, there is an initial bend angle recorded at the plate

centre from the bend angle development at the first edge, as the beam reached the

centre the thermal expansion there causes a counter-bend or arrests the bend angle

development momentarily before the bend angle is complete once the beam has left

the sheet. As the beam size increases the amount of initial thermal expansion must

increase also and so the effect is magnified. This discussion point is shown

schematically in figure 4.3.10.

These results emphasise the asymmetry of the process when using a single

point laser beam to achieved a symmetrical solution. Also demonstrated are the

subtle differences in the bend angle development depending on the process

conditions, namely the beam diameter and the number of passes realised.

Time

Dis

plac

emen

t

Time

Dis

plac

emen

t

Time

Dis

plac

emen

t

Figure 4.3.10: Schematic of possible reasons for ‘S’ curve bend angle development

1

2

3

Measurement location

Bend angle development at first edge causes initial positive displacement

Localised thermal expansion causes counter-bend or arrests bend angle development

Bend angle is complete once the beam leaves the sheet (apart from additional displacement on cooling)



4.4 Strain Gauge Analysis 123, 124

This investigation aims to complement the understanding of two-dimensional laser

forming, offering an insight into the mechanical behaviour of a part during the

process using a strain gauge analysis technique. The results of the investigation of

the transverse and longitudinal localised strains close to and far from the scan line

during the LF of 200x80x1.5mm mild steel coupons are presented here. The

processing parameters used were 5.5mm beam diameter, 760W and a traverse speed

of 30mm/s. 6 alternating passes were realised at 60 second intervals.

4.4.1 Transverse Strain

The transverse component of strain with respect to the scan line is orthogonal or at

90° to the scan direction.

The output from the gauges after six passes at 46mm from the scan line on

the top surface is shown in figures 4.4.1 to 4.4.4 and the output from the bottom

surface is shown in figures 4.4.5 to 4.4.8. The gauge locations are given as distances

from the first edge of the 80mm wide plates on the first pass of six using an

alternating direction strategy (refer to figure 3.2.12 in section 3.2.4 for clarification

of gauge locations).

Figure 4.4.1: Strain gauge output at 46mm top surface, 10mm from 1st edge






Figure 4.4.5: Strain gauge output at 46mm bottom surface, 10mm from 1st edge



Figures 4.4.2, 4.4.3, 4.4.6 & 4.4.7 show the output from the strain gauges

around the centre of the plate. Figures 4.4.1, 4.4.4, 4.4.5 & 4.4.8 show the output

from the two edges on the top and bottom surfaces at 46mm from the scan line. A

positive strain value indicates a tensile component and a negative value a

compressive component. It can be seen that even at this relatively large distance

from the irradiation line a small but significant strain measurement can be made, the

peak range being in the region of 18 microstrain (18x10-6 strain). It can also be seen






that there is a considerable difference in the strain output at the centre of the plate

compared to the edges. At the centre of the plate on the top surface (figures 4.4.2 &

4.4.3) a tensile component (perhaps due to thermal expansion) is seen during each

pass that recovers initially to a small residual tensile strain. Then, as the number of

passes increases an increasing residual compressive strain is seen, which recovers

somewhat several minutes after processing. At the plate edges on the top surface

(figures 4.4.1 & 4.4.4) an initial tensile component changes to a compressive

component during each pass and a tensile residual strain develops sometime after

processing. On the bottom surface similar strain outputs to the top surface from the

centre (figures 4.4.6 & 4.4.7) to the edges (figures 4.4.5 & 4.4.8) of the plate are seen.

However at the centre the magnitudes of the induced tensile strains are less than the

top surface but there is still a residual compressive strain component after processing.

At the plate edges on the bottom surface the initial tensile component during

processing is less than that at the top surface and the recovery after processing is to

an increasing residual compressive strain. The residual strains on the bottom surface

appear to recover several minutes after processing upon cooling. It can also be noted

in all of the results at a distance of 46mm that the effect of the alternating processing

direction is a variation in the peak values depending on the direction.

To summarise the results at 46mm from the scan line for the 200x80mm

plates, it can be seen that compressive strains are generated near the centre of the

plate and tensile strains at the edges on the top and bottom surfaces during

processing. The residual strain components appear to recover upon cooling, however

a residual tensile component is observed at the edges on the top surface.

These results appear consistent with the observed edge effect 60 or

longitudinal bowing phenomena where it is thought a change in boundary conditions

from the centre to the edge of the plate results in a variation in bend angle from edge

to edge. If a different in-process strain cycle occurs and hence a different residual

transverse strain state exists between the centre and the edge of the plate and top and

bottom surfaces, then this could possibly explain this edge effect distortion. Further

investigation is necessary to determine the longitudinal strains during laser forming,

as this should aid the explanation of the edge effect distortion still further.

The output from the gauges at 10mm from the irradiation line on the top

surface is shown in figures 4.4.9 to 4.4.11 and the output from the bottom surface is

shown in figures 4.4.12 to 4.4.14. As there was a very similar output from the



centrally located strain gauges at 46mm, it was decided to use a single strain gauge

on the centreline and two further gauges 10mm from each edge at 10mm from the

scan line (Figure 3.2.12).

As with the output at 46mm, figures 4.4.9 to 4.4.14 show a difference in

strain output from the centre to the edge and between top and bottom surfaces at

10mm from the scan line. However there was a significant difference in output


Figure 4.4.10: Strain gauge output at 10mm top surface, 40mm from 1st edge (Centreline)




between the two locations. An initial observation was the expected increase in strain

values recorded closer to the scan line. The range is now in the region of 100

microstrain. Figure 4.4.9 shows the transverse strain data 10mm from the first edge

on the top surface in the alternating direction irradiation strategy. It can be seen that

on the first pass and subsequent odd numbered passes there is a large tensile

component consistent with thermal expansion as the beam passes that point followed

by a recovery sometime after processing. On the return second pass and subsequent

even numbered passes, however, there is an initial compressive strain that switches

to a tensile strain as the beam reaches the other side of the plate followed again by a

recovery to an increasing residual tensile strain component. On the opposing side of

the plate (figure 4.4.11) the reverse occurs. On the first pass and subsequent odd

numbered passes the output from the gauge furthest way from the laser beam starting

position shows initially a compressive strain that switches to a tensile strain as the

beam reaches that point, followed by recovery to a tensile residual strain component.

On the second and subsequent even numbered passes a large tensile component is

seen as the beam passes followed by a recovery. It can be seen (Figures 4.4.9 &

4.4.11) that the residual tensile strain component present in both edges after six

passes appears to be decreasing some time after processing. As the plate is cooling

the purely elastic strains are relieved. At the centre of the plate, figure 4.4.10, it can

be seen that there is a compressive or less tensile strain component during each pass

which recovers to an increasing residual tensile strain that appears constant

sometime after processing. The effect of the alternating direction strategy appears

not to occur at the centreline. The residual tensile strain observed in the top surface

close to the scan line may be due to the plastic compression and hence transverse

shortening in the irradiated area consistent with the TGM.




Figures 4.4.12, 4.4.13 & 4.4.14 show the output from the gauges positioned

on the bottom surface 10mm from the irradiation line. In figures 4.4.12 & 4.4.14 it

can be seen that as with the top surface the edges on the bottom are affected by the

asymmetry of the process and the traverse direction. At the edge closest to the start

point of the laser (Fig. 4.4.12) on the first pass and subsequent odd numbered passes

a compressive strain is seen initially, consistent with the upper surface expansion and

counter bend. This rapidly reverts to a tensile component as the beam moves to the

other side of the plate. On the reverse second pass and subsequent even numbered

passes a tensile strain component is seen that reverts rapidly to a compressive or less

tensile strain and then recovers to a higher residual tensile strain. The tensile

component remaining in the sheet after processing appears constant sometime after

processing. On the opposing side of the sheet (Fig. 4.4.14) the sequence is mirrored;

the edge furthest away from the laser start position experiences a tensile strain

initially that rapidly changes to a less tensile state as the laser reaches that location

Figure 4.4.13: Strain gauge output at 10mm bottom surface, 40mm from 1st edge (Centreline)




followed by a recovery to a residual higher tensile strain. The edge closest to the

laser start position experiences a compressive strain initially that rapidly reverts to a

tensile strain as the beam traverses to the other side of the plate. There is a residual

tensile strain on this edge as well. It can be noted though from figures 4.4.12 &

4.4.14 that the magnitude of this strain depends on the direction of the final pass. At

the centre location on the bottom surface (Fig. 4.4.13), the gauge records evenly the

effects at both edges, in that a tensile strain is seen as the beam moves across the

plate and this reverts to a compressive strain as the beam passes the centre and

moves to the other side of the plate. This effect has been noted in other studies 4.

Again a recovery to a residual tensile strain occurs after processing due perhaps to

the development of the bend towards the laser which is consistent with the TGM

theory 11.

The strain gauge results demonstrate the complexity of the laser forming

process even during a simple straight line 2D bend. A large factor in this is the

inherent asymmetry of the process when using a single point laser source to achieve

a symmetrical solution. Whilst absolute readings of strain are difficult at such high

thermal gradients the general trends in transverse strains due to thermal and

mechanical influences have been revealed. It has been shown that along an

irradiation line depending on where the beam is and its direction, there is a

mechanical effect in the plate ahead and to the rear.

Figure 4.4.15: Visualisation of the transverse strain output close to the scan line at the start of a pass



Figures 4.4.15 & 4.4.16 show a visualisation of the results obtained from the

strain gauges at either edge, top and bottom surfaces, 10mm from the scan line at the

start and end of a pass. The beam at the first edge causes a thermal expansion of the

upper surface and hence a compression of the lower surface consistent with a counter

bend effect. As one side of the plate expands a compression of the top surface of the

other side of the plate is seen perhaps due to a moment generated in the upper

surface (4.4.15). An opposing moment may be present in the lower surface, as one

side is under compression a tensile strain is seen in the other side. This would

suggest a torsion force is present in the plate between the top and bottom surfaces.

As the beam reaches the other side of the plate (Fig. 4.4.15) the effect of a reversal in

this moment may be evident by the sudden reduction in the tensile strain component

of the first edge before a recovery of the elastic strains during cooling (Fig. 4.4.9).

A residual tensile strain was observed at 10mm from the scan line on the

bottom surface and along the centreline on the top surface. The tensile component

left in the bottom surface is consistent with the mechanical bend in the plate. The

component in the top surface may be due to a transverse shortening of the upper

surface along the scan line consistent with TGM theory.

Figure 4.4.16: Visualisation of the transverse strain output close to the scan line at the end of a pass



4.4.2 Longitudinal Strain

The longitudinal component of strain with respect to the scan line is parallel to the

scan direction. Refer to figure 3.2.14 for clarification of the strain gauge locations on

the 200x80mm 1.5mm mild steel coupons.

The results from the three gauges at 46mm from the scan line on the top and

bottom surfaces during six laser passes at 60 second intervals are presented in figures

4.4.17 and 4.4.18. As the magnitude of the data at this distance was small and

somewhat noisy, the output for the three gauges on each surface has been combined

for presentation; this allows a comparison of the relative magnitudes at each location.

The locations refer to the distance from the first edge to be processed.

Figure 4.4.17: Output from gauges on the top surface at 46mm from the scan line, longitudinal strain

Figure 4.4.18: Output from gauges on the Bottom surface at 46mm from the scan line, longitudinal strain



It can be seen in figures 4.4.17 and 4.4.18 that due to the noise level within

the measurement circuit (possibly due to a shielding failure) compared to the

magnitude of the output from the gauges at 46mm, the more subtle strain cycle detail

pass by pass is difficult to observe. What can be seen, however, is the large

difference in longitudinal strain cycle measured at the centre of the plate when

compared to the edges on both the upper and lower surfaces. On the top surface

(figure 4.4.17) the strain cycles per pass at all three of the locations are similar, with

an initial tensile component which is quickly reversed to a compressive component.

This then recovers to somewhat after the laser has passed. The main difference

between the locations is the residual strain level that is reached after each pass. It can

be seen that the data recorded at either edge is similar with a small residual

longitudinal strain remaining after the six passes. At the centre of the plate the strain

cycle is much larger and after pass three, and for subsequent passes, an increasing

residual tensile longitudinal strain component remains in the coupon when compared

to the edges. On the bottom surface (figure 4.4.18) the strain cycles at each location

also appear similar with an initial large tensile component which is quickly reduced

and followed by a small tensile increase which recovers to some lower level. As with

the upper surface the strain cycles and residual longitudinal strain levels are very

similar at either edge on the lower surface after the six passes. At the centre,

however, the strain cycle recovers to an increasing compressive residual longitudinal

strain when compared to the edges and the upper surface. These results show that

after six passes there is a residual tensile longitudinal strain component in the upper

surface (higher at the centre) and a residual compressive longitudinal strain

component (also higher at the centre). These results appear consistent with observed

edge effect phenomena 60 where, as can be seen in figure 4.4.19, a longitudinal

bowing of a laser formed coupon can be present. If a part is held in the form shown

in figure 4.4.19, the upper surface would be in tension and the lower surface would

be in compression and the relative magnitudes of each would be highest at the centre.

Figure 4.4.19: Exaggerated view of edge effects 60



The strain gauge output at 46mm from the scan line confirms this edge effect

phenomenon and shows why the distortion would occur. The development of the

differences in strain levels between the upper and lower surfaces and between the

edges and centreline can be seen in figures 4.4.17 and 4.1.18. It can be seen that a

compressive longitudinal residual strain is developed (at the centre) after the first

pass in the bottom surface. However, a significant tensile component and hence a

significant net difference between the top and bottom surfaces, is not developed until

after pass 3. This is consistent with observations made in work related to this

study129 whereby a small concave (or positive camber) distortion was observed

during the first two passes which changed to a convex (or negative camber)

distortion for subsequent passes (as in figure 4.4.19). This highlights the complexity

of the process even during 2D laser forming.

The data recorded at 10mm from the scan line on the upper and lower

surfaces can be seen in figures 4.4.20 to 4.4.25. The magnitude of the data at this

distance was much larger than that at 46mm and so the unwanted noise level became

less significant. Therefore the data for each location is presented in a separate figure

so as to observe the more subtle strain cycles.

Figure 4.4.20: Output from gauge at 10mm from 1st edge on the top surface at 10mm from the scan line, longitudinal strain

Figure 4.4.21: Output from gauge on the centreline on the top surface at 10mm from the scan line, longitudinal strain



Figure 4.4.22: Output from gauge at 70mm from 1st edge on the top surface at 10mm from the scan line, longitudinal strain

Figure 4.4.23: Output from gauge at 10mm from 1st edge on the lower surface at 10mm from the scan line, longitudinal strain

Figure 4.4.24: Output from gauge on the centreline on the lower surface at 10mm from the scan line, longitudinal strain

Figure 4.4.25: Output from gauge at 70mm from 1st edge on the lower surface at 10mm from the scan line, longitudinal strain



As with the data obtained at 46mm the data presented in the above figures

shows that at 10mm from the scan line there is a significant difference in the

longitudinal strain cycle between each of the edges and the centre and between the

upper and lower surfaces of the mild steel coupon during LF.

On the top surface (figures 4.4.20 to 4.4.22) it can be observed that the strain

cycles per pass are similar, in that an initial tensile component occurs (possibly due

to local thermal expansion as the beam passes) which is rapidly reversed or negated.

This then recovers during cooling to a higher residual strain level. There is a

significant difference in the residual longitudinal strain level between the edges and

the centre on the upper surface. At both the edges (figures 4.4.20 and 4.4.22) there is

a similar residual strain after each pass which appears to level off after pass 3 at

approximately 30 microstrain. At the centre of the coupon (figure 4.4.21) the initial

tensile component is not reversed to the same extent as at the edges and the

subsequent recovery during cooling is to a larger tensile longitudinal residual strain

(~45 microstrain after 6 passes). This also appears to level off after pass 3.

On the bottom surface (figures 4.4.23 to 4.4.25) the data shows that the strain

cycle is more subtle. It can be seen in figures 4.4.23 and 4.4.25 that using an

alternating direction strategy has an effect on the longitudinal strain cycle (effect also

observed in the transverse strain study presented earlier). It can be seen that there are

fluctuations in the peak strain values depending on the direction of the scan (these

were not as well defined but still observed in the upper surface data, figure 4.4.22).

At the first edge on the first pass (figure 4.4.23) there is an initial large tensile

component which is rapidly negated to a compressive component. This then recovers

during cooling to a residual tensile level (~30 microstrain). At the opposite edge on

the first pass (figure 4.4.25) there is an initial compressive component which rapidly

becomes a tensile component which is in turn then negated. As with the first edge the

strain level recovers during cooling to a similar residual tensile longitudinal strain.

For the second pass in the opposite direction and subsequent even number passes the

opposite output can be observed. At the centre of the plate (figure 4.4.24), as would

be expected, the effect of laser scanning direction is not as apparent. The strain cycle

here is more akin to the second edge (furthest away from the start point of a new

pass), in that an initial compression can be observed which becomes a large tensile

component. This is in turn negated or reversed and rapidly recovers to a consistent

residual tensile strain level of approximately 40 microstrain. No further increase or



decrease occurs in this level for subsequent passes. This is also the case at the edges.

After pass 1 the lower surface close to the scan line is in tension parallel to the scan

line (longitudinal).

A difference in the residual longitudinal strains can be observed in the gauge

output data at 10mm between the centre of the coupon and the edges similar to that

observed at 46mm from the scan line. Although there is a tensile residual component

in both the upper and lower surfaces, there is a net difference between them of

approximately 5 microstrain, in that the upper surface is more in tension than the

lower surface parallel to the scan line. This is consistent with the data observed at

46mm from the scan line (discussed earlier) which provides an additional

explanation of the edge effect phenomena.

As with the study on the transverse strain development an attempt to

visualise the strain cycles near the scan line at the start and near the end of a scan

was produced. These schematics can be seen in figures 4.4.26 and 4.4.27.

Figure 4.4.26: Visualisation of the longitudinal strain output close to the scan line at the start of a pass



The data near to the scan line (10mm) suggests that a tensile component of

longitudinal strain occurs due to thermal expansion local to the laser beam (figure

4.4.26). Ahead of the beam the thermal expansion causes a mechanical compression.

This was observed in the data on the bottom surface (figures 4.4.23 to 4.4.25) at the

centreline and the second edge. This initial effect was less obvious on the top surface

and may be due to the close proximity of the laser beam to the gauge overriding the

more subtle mechanical rather than thermal output. As the laser beam is traversed

across the sheet a localized thermal expansion occurs at each of the gauge locations

which may correspond to the high observed tensile strain output. Once the beam has

passed (figure 4.4.27) a rapid cooling occurs thus a contraction and compression is

observed. In addition the mechanical effect that may influence the surface strain

ahead of the beam must also affect the strains observed to the rear, in that an

additional compression acts to rapidly reduce the peak tensile strain and at some of

the gauge locations produce a net compressive strain (figures 4.4.20 and 4.4.22).

More detail on the three dimensional strain field development and further

explanation and confirmation of this data may be obtained through finite element

methods. The development of an FEA combined thermo-mechanical model is

presented in this thesis.

Figure 4.4.27: Visualisation of the longitudinal strain output close to the scan line at the end of a pass



4.5 Finite Element Analysis

Reported here is the development of a Finite Element Analysis (FEA) model for the

single pass laser forming of graphite coated 80x80x1.5mm Mild Steel CR4 coupons

using a CO2 laser source and edge clamped boundary conditions (as in figure 3.2.1).

The process parameters investigated were those obtained from the empirical study;

3mm beam diameter 760W, 55mm/s; 5.5mm beam diameter, 760W, 30mm/s; 8mm

beam diameter, 760W, 20mm/s. The model was developed to ascertain peak

temperatures, thermal behaviour, transient stress/strain conditions, residual

stress/strains and displacements during and after laser forming.

4.5.1 Development of a Graded Mesh Model

For the Abaqus FEA software program the model had to be developed to run in two

parts. The first part was a purely thermal model to determine the temperature field

realised in the coupon as a laser is traversed across it over a known time period. The

second part was a coupled thermo-mechanical model using the temperature history

data results from the first part to calculate the thermal stress/strain field and hence

the distortion or bend angle induced. An initial model was developed around a dense

1200 element mesh (figure 4.5.1), using 20 node 3D elements. This represented only

a small section of the plate.

The physical properties of 1.5mm mild steel were used (as given in tables

3.2.3 and 3.2.4) to describe the mechanical and thermal behaviour of the sheet. One

Figure 4.5.1: Initial 1200 element FEA model developed



edge of the sheet was fully constrained to simulate a clamped edge. The laser beam

was simulated by using a non uniform heat flux on the upper surface. This heat flux

was described by a Fortran user sub-routine. This allowed for a Gaussian energy

distribution for a given laser power level and absorption coefficient, with the beam

diameter governed by the lens focal length, input beam diameter, M2 of the laser,

workpiece focal position and wavelength. In order to simulate the cooling conditions

a convection heat transfer to air on all exposed element surfaces was realised.

The output from this initial model was promising and demonstrated that the

process could be modelled this way (figure 4.5.1). However, the run times and data

volume generated were unacceptable. The run time for the first thermal part was

approximately 24 hours. The second thermo-mechanical part ran for 4 days on an

IRIX mainframe and generated a great deal of data (>10Gb) before it was terminated

due to lack of disk space. This was unacceptable, especially when the model cannot

be guaranteed to run without error. It was realised that the main reason for the

excessive run time was the number of elements used. It was decided to develop a full

sized graded mesh model to describe the mild steel coupon. This model type has the

benefits of using less elements by using a fine mesh for areas where the rate of

change of output data is likely to be high (therefore more data points area required to

describe the event) and a graduated coarser mesh were the rate is low and the data is

not as important. The graded mesh model of the 80x80x1.5mm coupon can be seen

in figure 4.5.2; the total number of elements has been brought down to 580 (20 node

3D elements used again). It can be seen that the irradiated track has a very dense

mesh which gradually becomes coarser further away from the centre. It was decided

to reduce the number of elements into the thickness to one as this would reduce the

element number further (testing of the model revealed that this was not detrimental

as there were sufficient data points into the surface due to the 20 node elements).

Figure 4.5.2: 580 element graded mesh model

Clamped Edge Edge 1 (Laser start point)

Edge 2 Centreline (Data acquired along this line)



All of the boundary and loading conditions and physical properties developed

for the first model (described earlier) were used for this model as well. The Abaqus

input file that describes this model is given in appendix 2.

The run times for this model were much better, for the first thermal part the

run time was approximately 1 hour and for the more computer intense coupled

thermo-mechanical analysis the run time was approximately 36 hours (depending on

output requirements). A SunOS 5.8 based mainframe was used to run the model.

The output from this FEA model is presented in the following sections; this is

divided into the purely thermal output for the three process parameter combinations

(3, 5.5 and 8mm beam diameters) and the output from the thermo-mechanical model

for the 5.5mm beam diameter process combination.

It is felt that the output from this FEA model will be useful, and whilst some

calibration with real data occurs (described later), the output data cannot be regarded

as absolute (although checks are made to see that the data is realistic). However, the

FEA technique does allow the visualisation of events at any point on the coupon in

real time (such as 3D strain field development) which may aid the further

understanding of the 2D laser forming process.

4.5.2 Thermal Analysis

Key to a usable FEA model is the production of realistic results. A method of

ensuring this is to tune the model to real measured data. This was achieved here by

the use of thermocouple data presented earlier in section 4.2.1. This gave

temperature development with respect to time for locations from 10 to 58mm from

the scan line on the upper and lower surfaces (the closer to the scan line the better).

The absorption coefficient, A, is key to tuning the model to this real data, in that by

varying A, observing the temperature output at 10 and 22mm from the scan line and

then comparing this to the measured data it was possible to achieve a good

agreement in terms of heating rate and peak temperature at these locations. The

cooling rate was tuned to the measured data by varying the convection coefficient to

the surrounding air.

The variation in peak temperature from the model at the centre of the scan

line (on the centreline, upper surface, figure 4.5.2) with absorption coefficient for a



5.5mm beam diameter, 760W and a speed of 30mm/s can be seen in figure 4.5.3.

The work piece to lens stand-off to give a 5.5mm beam diameter (focal position) was

selected using standard beam propagation equations given in appendix 3.

The above figure demonstrates the importance of the absorption coefficient

on the laser forming process. This also shows why the bend angle rate per pass can

fall so dramatically when the coating degrades somewhat as observed in section 4.1.

The peak temperature drops from 600 °C to 425°C for A=0.85 to A=0.6.

After analysing the data obtained for all of the absorption coefficients shown

above (for this beam size), it was found that the best agreement at 10 and 22mm

from the scan line was obtained for A=0.85. This can be seen in figures 4.5.4 and

4.5.5.

Figure 4.5.3: Variation in peak upper surface temperature with absorption coefficient (model output).

Figure 4.5.4: Temperature output from the FEA model at 10 and 22mm from the scan line for a) Upper surface b) Lower surface 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

a) b)



It can be seen in the above figures that there is a good agreement between the

heating and cooling curves and the peak temperatures of the model output and the

measured thermocouple data at 10 and 22mm from the scan line. Ideally measured

data closer to the scan line would be better for tuning purposes as there is a

considerable difference in temperature between the centre of the scan line and at

10mm away. However, as this data was not available it was assumed that the

agreement at 10mm would be sufficient to give a reasonably accurate simulation of

the LF process. Agreement was also found when comparing thermocouple and

model output data for the 3mm and 8mm beam diameter process combinations

(given earlier) using A=0.85. This suggests that the absorption coefficient of graphite

is quite high for 10.6µm wavelengths; this is consistent with the data presented in

table 3.1.1 and figure 3.1.22 (data available in the literature). It is likely that the

overall absorption coefficient for graphite, however, is dependent on the material it is

sprayed onto, in that it is the efficiency of the heat transfer of the energy absorbed by

the graphite into the substrate material, such that materials with different thermal

properties will have different overall absorption coefficients to 10.6µm radiation

when sprayed with graphite. In addition the model output is for the first single pass

where A is likely to be at its highest level.

Using A=0.85 for the graphite coated mild steel the model was run using the

following energy parameters; 3mm beam diameter 760W, 55mm/s; 5.5mm beam

diameter, 760W, 30mm/s; 8mm beam diameter, 760W, 20mm/s. The thermal output

at these process parameters is presented below. It can be noted that the scales of the

3D temperature contour plots have a variable range i.e. highest temperature is always

in red and lowest in blue at any given time.

Figure 4.5.5: Thermocouple measurements at 10 and 22mm from the scan line for a) Upper surface b) Lower surface 5.5mm beam dia. 760W, 30mm/s, single pass

a) b)



Figure 4.5.6: Model Output, 3D contour plot of temperature at; a) Mid-pass b) End of pass c) t=4.2s d) t=46.2s 3mm beam dia. 760W, 55mm/s, single pass, A=0.85

Figure 4.5.7: Temperature output at various distances from the scan line along the centreline of the plate, Upper Surface 3mm beam dia. 760W, 55mm/s, single pass, A=0.85

Figure 4.5.8: Temperature output at various distances from the scan line along the centreline of the plate, Lower Surface 3mm beam dia. 760W, 55mm/s, single pass, A=0.85

20mm10mm8mm4mm

2.4mmCentre

20mm10mm8mm4mm

2.4mmCentre

Peak Temperature = 914°C


b) a)

c) d)



Figure 4.5.9: Model Output, 3D contour plot of temperature at; a) Mid-pass b) End of pass c) t=4.5s d) t=28.6s 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

c)

Figure 4.5.11: Temperature output at various distances from the scan line along the centreline of the plate, Lower Surface 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

Figure 4.5.10: Temperature output at various distances from the scan line along the centreline of the plate, Upper Surface 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85



20mm10mm8mm4mm

2.4mmCentre

20mm10mm8mm4mm

2.4mmCentre

a) b)

d)



Figure 4.5.12: Model Output, 3D contour plot of temperature at; a) Mid-pass b) End of pass c) t=5.4s d) t=34.4s 8mm beam dia. 760W, 20mm/s, single pass, A=0.85

Figure 4.5.13: Temperature output at various distances from the scan line along the centreline of the plate, Upper Surface 8mm beam dia. 760W, 20mm/s, single pass, A=0.85

Figure 4.5.14: Temperature output at various distances from the scan line along the centreline of the plate, Lower Surface 8mm beam dia. 760W, 20mm/s, single pass, A=0.85

20mm10mm8mm4mm

2.4mmCentre

20mm10mm8mm4mm

2.4mmCentre



d) c)

a) b)



It can seen in the above figures that the output from the Abaqus FEA model

can be presented in both X-Y plot form (from many points) and as a 3D contour plot

at any point in the time cycle. In addition an animation of the thermal cycle can be

created. It should be noted that these are the thermal outputs only and so no

deformation was recorded. For the 3D contour plots (figures 4.5.6, 4.5.9 and 4.5.12)

the variable scale can be somewhat confusing; however, this allows the presentation

of the lower temperature data as the plate cools. This would be lost if the scale was

fixed (i.e. 0 to 600°C), instead the highest temperature in the plate at a given time is

assigned the red colour and the lowest blue.

It can be observed in figures 4.5.6, 4.5.9 and 4.5.12 that the modelling of the

incident laser beam on the coupon appears to work well, with the beam size

increasing as the workpiece stand-off is increased. The energy distribution within the

beam can also be observed; although this is not the exact beam mode used, this was

thought to be a reasonable approximation. A problem with the mesh generation can

also be observed in these figures. It was realised that at the interface between the

different densities of mesh a discrepancy between the numbers of nodes has led to

the generation of the unsmooth contours or ripples observed. Although not ideal, it

was felt that in the region of interest along the scan line the data is unaffected and

therefore the analysis should continue. It is intended that this problem will be

rectified in ongoing future work in this field.

In figures 4.5.7, 4.5.10 and 4.5.13 it can be seen that the heating and cooling

curves recorded on the upper surface on the scan line show very high heating and

cooling rates. It can be seen that peak temperatures are only achieved for fractions of

a second before being rapidly quenched by conduction into the surrounding (cold)

bulk material. As would be expected the further away from the centre of the scan line

the lower the peak temperature observed. It can be seen that the peak temperature

increases considerably with decreasing beam diameter, consistent with the increase

in intensity for the same power.

A comparison has been made between the temperatures realised at the same

locations on the upper and lower surfaces. For the 3mm beam diameter data (figures

4.5.7 and 4.5.8) a high peak temperature of 914°C has been predicted at the centre of

the plate on the upper surface. On the lower surface at the same point a peak

temperature of only 242°C has been observed. This is consistent with the TGM

where a high thermal gradient through the thickness is necessary to give the



differential thermal expansion and subsequent plastic compression through the

section to generate a bending moment. As the beam diameter increases the

temperature difference between the upper and lower surfaces becomes less. It can

also be seen that the peak temperature observed on the lower surface increases with

increasing beam diameter, consistent with the larger beam and lower traverse speed

heating the section more uniformly. It can be noted, however, that even for the 8mm

beam diameter data (figures 4.5.13 and 4.5.14) a significant difference in peak

temperature can still be seen (141°C difference), such that the TGM must still be

active to some degree generating a sufficient bending moment to give a positive

bend (confirmed by experimental data). It can be concluded from these results that

providing a positive bend can be ensured (if TGM is required) a larger beam

diameter (much greater than the sheet thickness) will induce lower peak

temperatures in a material with no loss in forming efficiency and therefore less

unwanted metallurgical changes associated with high temperatures. This will be

investigated in the metallurgical study presented later.

Another issue that was observed in these results was the temperature

difference from edge to edge along the scan line during forming due to the

asymmetric nature of the process. It can be observed in figures 4.5.6, 4.5.9 and

4.5.12 that during cooling the edge at the end of the scan line remains at a more

elevated temperature than the first edge as the heat is dissipated into the bulk of the

plate. To demonstrate this further the temperature profiles at edge 1 and edge 2 (refer

to figure 4.5.2) were isolated for the 5.5mm beam diameter data and are presented in

figures 4.5.15 and 4.5.16.

Figure 4.5.15: Temperature output at various distances from the scan

line along at Edge 1, Upper Surface 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

Peak Temperature = 378°C 10mm

8mm4mm

2.4mmCentre



The above figures demonstrate that a different temperature can be realised on

the surface of a component during LF along the scan line dependent on location and

processing direction. At the first edge a low peak temperature of 378°C was

observed and at the second edge this raises to 658°C. These can also be compared to

the value recorded at the centre of the plate of 599°C. A possible explanation for this

phenomenon is in the inherent asymmetry of the process, in that the heat imparted

into the plate as the laser beam traverses across it is continually flowing into the cold

region ahead of the beam, thus increasing the temperature realised. At the centre of

the plate equilibrium conditions are realised (stable peak temperature). However, as

the beam reaches the second edge the heat flowing ahead of the beam cannot travel

any further and so a heat build up occurs. Hence the increase in temperature at the

second edge. This heat build up can be observed as a widening of the HAZ at the

second edge and can be seen in figure 4.5.17. This effect was also noted in the

thermal imaging study presented earlier.


Figure 4.5.16: Temperature output at various distances from the scan line along at Edge 2, Upper Surface 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

10mm8mm4mm

2.4mmCentre

Figure 4.5.17: Widening of the HAZ near the edge in mild steel 5.5mm beam dia. 760W, 30mm/s



This asymmetry is likely to cause unwanted distortion particularly in samples

processed at a relatively low traverse speed and demonstrates the requirement to

utilise an alternating direction strategy to even out the temperature distribution when

using a multiple pass strategy. Even using a multiple pass strategy the heat build up

on both edges will still occur. This may be a source for the edge effect phenomena

described earlier, in that more forming may occur at these points than in the centre.

A possible solution to the edge effect problem suggested by Magee 29 was to vary the

traverse speed along the scan line, such that a higher speed should be used towards

the edges so as to reduce the heat input. The work presented here would suggest that

a slower traverse speed at the first edge may be necessary as well to produce an even

temperature distribution along the scan line.

The next step in this study was to use the temperature data generated from

this first model run to calculate the thermally induced stresses and strains and hence

any resultant deformation. As the second thermo-mechanical part of the model was

computer intensive only one of the process parameter combinations was analysed.

The data generated using a 5.5mm beam diameter, 760W and a traverse speed of

30mm/s was used. The results are presented in the following sections and are divided

into displacement, transverse strain (E11), longitudinal strain (E22), transverse stress

(S11) and longitudinal stress (S22).

4.5.3 Displacement

This section contains the results from the coupled thermo-mechanical analysis of the

laser forming of an 80x80x1.5mm mild steel coupon in terms of displacement or

resultant distortion, using a 5.5mm beam diameter, 760W and a traverse speed of

30mm/s.

Figure 4.5.18: Final displacement output, magnification factor =30 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85



The final output from the model can be seen in figure 4.5.18; a displacement

magnification factor of 30 has been used to aid the analysis. The clamped boundary

conditions along one edge are also represented in this view. It can be seen that the

model does predict a bend angle about the scan line for the energy parameters used.

By overlaying the temperature data it is possible to observe the development of the

bend angle over time with the beam location, this can be seen in figure 4.5.19. Once

again the ranges for the contour plots are automatically selected.

It can be seen in the above figure that as the beam is traversed across the

plate a longitudinal U deformation forms around the beam and this is drawn (zipped

up almost) across the sheet. The bend angle appears to form local to the beam but

influences can be observed in other parts of the sheet. It can be seen that the bend

angle appears fully formed as the beam leaves the sheet with little movement after

the beam has passed on cooling.

The displacement of the end of the plate (on the centreline) with respect to

time can be seen in figure 4.5.20. This confirms that the majority of the distortion

occurs during the pass and that only a slight increase in bend angle occurs during

cooling. It can also be noted that virtually no counter bend is present, this could be a

Figure 4.5.19: Model Output, 3D contour plot of temperature and displacement at; a)Start of pass b) Mid-pass c) End of pass d) t=49s 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

a) b)

c) d)



result of the coarse data output missing a fast event (output to file every 10

increments to preserve disk space). This is, however, consistent with the

displacement/time experimental data presented earlier where as the beam diameter

increased the counter bend effect became less significant. This is perhaps due to the

more uniform heating of the section causing an in-plane rather than out of plane

movement due to the initial thermal expansion. The same bend angle development

with time was also observed in this study.

Analysis of the height contours of the formed plate reveals a longitudinal

bowing of the free end consistent with the edge effect phenomenon. This can be seen

in figure 4.5.21

The above figures show that a U shaped longitudinal distortion has been

formed after the first pass. This indicates that the two corners of the free end of the

plate are higher than the centre. This can also be seen in figure 4.5.22 which shows

Figure 4.5.20: Displacement/time output, free end of the plate on the centreline5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

[m]

Figure 4.5.21: Height contour plots of the formed plate, magnification factor =305.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

Centre



the displacement / time development of the two corners when compared to the centre

of the free end.

This figure shows that the bend angle development is dependent on the

location on the sheet. It can be seen that slightly less distortion occurs at the centre of

the plate when compared to the corners and that corner 2 on edge 2 (end of the scan

line) is slightly higher than corner 1. These results outline the problems of the

asymmetry of the process when trying to produce a symmetrical part.

The maximum distortion of 0.354mm at the edge of the free end of the plate

corresponds to a bend angle of approximately 0.5°. This can be compared to the

empirical result of 1.2° for the same energy parameters (figure 4.1.5). Although

slightly different the result from the model is a realistic value. The model could be

tuned further towards the experimental value by varying the physical properties of

the material and acquiring more of the temperature dependent mechanical and

thermal properties such as yield stress and coefficient of thermal expansion

(interpolation between a few known values was used here), however, as the model

was intended only to give an insight into the LF process and not produce absolute

values, it was felt that the accuracy attained was sufficient.

The displacement output from the model is consistent with observations

made in work related to this study on real time dynamic shape measurement 129 of

the LF process, whereby a small concave (or positive U camber) distortion was

observed during the first two passes which changed to a convex (or negative camber)

distortion for subsequent passes. The ‘zipping up’ nature of the beam passing over

the surface (clearer when observed as an animation) and the relative movements of

Corner 1 CentreCorner 2

Figure 4.5.22: Displacement/time output, free end of the plate 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

[m]



each corner observed in the output from the model were also observed in the

dynamic shape measurement work in the LF of 80x80x2mm Ti64.

4.5.4 Transverse Strain E11

The transverse or lateral strain is perpendicular or 90° to the scan line direction. This

is referred to here as E11 or the direct strain in the 1 or x direction. 3D contour plots

of the E11 output at various points during the process can be seen in figure 4.5.23.

As with the previous 3D contour plots presented, the scale has been varied

automatically (by Abaqus viewer) according to current highest and lowest values.

Here the highest tensile or positive strain is represented by red and the lowest

compressive or negative strain is represented by blue.

It can be seen in figure 4.5.23 that directly under the beam a tensile

transverse strain can be observed in the upper surface of the plate due to the localised

Figure 4.5.23: 3D contour plots of E11 at; a) Start b) Near Start c) Near End of pass d) t=29.5s 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

d) c)

b) a)

5

5 5

5

5

4

4 4

4

4

4 4

4

4

444

5

6

5

44

44

44

44

445

5

5

5

44

44

44

44

4

4 4

5

6

5

4 4

4

4

4 4

4

4



thermal expansion. Surrounding this tensile area are regions of compressive

transverse strain. This is likely due to the thermal expansion against the cold bulk

material. In figure 4.5.23 b) it can be seen that the thermal expansion at the first edge

has an effect on the opposite edge where a small compressive strain can be observed.

This is consistent with the results of the strain gauge analysis presented earlier

(section 4.4.1) where a mechanical effect ahead of the beam was observed. As the

beam traverses across the sheet, figure 4.5.23c), it can be seen that large areas or

lobes of compressive transverse strain form around the high tensile strain under the

beam. This demonstrates the complexity of the strain field within the plate during the

process. A reaction against the clamp can also be seen in this figure. On cooling,

figure 4.5.23 d), a residual compressive transverse strain can be observed along the

centre of the scan line on the upper surface, a residual tensile strain can be seen in

the lower surface. On the upper surface the compressive strain within the scan line is

surrounded by a region of (small) tensile residual strain. This effect was also

observed in the strain gauge study presented earlier and was attributed to the

transverse shrinkage within the scan line on cooling setting up a tensile strain in the

surrounding upper surface.

In order to compare the model output with the strain gauge data the

transverse strain development over the scan time for single locations were isolated.

Data is presented in the following figures from close to the scan line (~10mm) near

edge 1 (figure 4.5.24), at the centre of the plate (figure 4.5.25) and near edge 2

(figure 4.5.26) on the upper surface. The transverse strain gauge output for a single

pass on the upper surface is also presented.

Figure 4.5.24: Transverse strain E11 at ~10mm from scan line near edge 1 Upper surface: a) Model output b) Strain gauge output 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

Edge 1

4

a) b)



It can be seen in the above figures that at this distance from the scan line the

transverse strain predictions from the model over time are reasonably similar, in

terms of strain cycle, to the measured strain gauge data. The values of microstrain

are also consistent to those observed during the first pass in the strain gauge study

(isolated in the figures above from the 6 pass data from figures 4.4.9 to 4.4.11). It

can also be observed that the localised transverse strain cycle varies with position on

the sheet at the same distance from the scan line. At edge 1 it can be seen that a large

tensile component occurs as the beam passes which reduces on cooling. It can be

observed that the initial tensile sequence at this first edge recorded by the strain

gauge measurement is missing in the model output. It can be noted, however, that the

data starts at a compressive value (figure 4.5.24) and so this maybe a facet of the

Figure 4.5.25: Transverse strain E11 at ~10mm from scan line at the centreUpper Surface: a) Model output b) Strain gauge output 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

Figure 4.5.26: Transverse strain E11 at ~10mm from scan line near edge 2 Upper Surface: a) Model output b) Strain gauge output 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

Centre

4

Edge 2

4

b)a)

a) b)



coarse data capture rate from the model incorporated to preserve disk space, in that a

rapid initial sequence at the first edge may have been missed.

At the centre of the plate and at edge 2 (figures 4.5.25 and 4.5.26) similar

output predictions can be observed. It can be seen that there is a small initial tensile

component which reverts to a compressive one before a large tensile component

occurs which reduces to a lower level on cooling. The initial tensile component (also

observed in the strain gauge output) could be a flexing of the plate similar to the

counter bending effect, in that a mechanical effect may be occurring as the localised

thermal expansion occurs at edge 1. The compressive component could be attributed

to the areas of compression generated around the large tensile area under the beam

due to thermal expansion (figure 4.5.23 c); the data taken at 10mm from the scan line

is within these compressive areas. As the beam passes a location (at 10mm from scan

line) a compressive component is observed. However, just behind the beam there is a

tensile region (figure 4.5.23 c) due to thermal expansion below a level so as not to

cause a compression of the surrounding area (hence a lower temperature); this

corresponds to the larger tensile component in figures 4.5.25 and 4.5.26. The

mechanical compressive effect observed at edge 2 due to the thermal expansion at

edge 1 at the start of a scan (concluded from the strain gauge analysis section) can be

seen in figure 4.5.26 a). This effect, although present (and maybe larger nearer the

scan line) may be less significant than originally thought as it is masked or

outweighed by the compressive region generated by the localised thermal expansion

under the laser beam.

Differences in residual transverse strain levels can also be noted in figures

4.5.24 to 4.5.26 along the scan line at this distance after a single pass. For each of the

locations, the centre and edges 1 and 2, a tensile residual strain can be seen and this

is different at each. The highest level is at the centre and edges 1 and 2 are at

different lower residual transverse strain levels. A similar distribution was also

observed in the strain gauge study after 1 pass and demonstrates the effect of the

asymmetry of the process, the need for an alternating direction strategy and a

variable traverse speed to realise an even strain cycle at all locations along the scan

line to reduce any unwanted distortion (such as edge effects).



4.5.5 Longitudinal Strain E22

The longitudinal strain is parallel to the scan line direction. This is referred to here as

E22 or the direct strain in the 2 or y direction. 3D contour plots of the E22 output at

various points during the process can be seen in figure 4.5.27. Again the highest

tensile or positive longitudinal strain is represented by red and the lowest

compressive or negative longitudinal strain is represented by blue (scale variable).

It can be seen in figure 4.5.27 that directly under the laser been a high

longitudinal tensile strain area occurs consistent with a localised thermal expansion

in the scan direction. In parts a and b of figure 4.5.27 it can be seen that this thermal

expansion causes a compression of the cold material ahead of the beam as it

traverses across the sheet; this effect was also observed in the strain gauge analysis.

The tensile component continues into the sheet surrounding the beam as the heat is

dissipated into the bulk material (figure 4.5.27 b). On cooling it can be seen in part c

that near the edges there is a larger tensile residual region extending into the plate

than at the centre; this persists on further cooling (part d). This is an indication of the

Figure 4.5.27: 3D contour plots of E22 at; a) Start b) Mid Pass c) End of pass d) t=19.5s 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

a) b)

c) d)

4 5

5

5

8

5 5

5

4 4

4

4

4

4

4 5

5

5

5

5 4

4

4 4

4

4

4

4

445

5

5

5

54

44

44

44

445

5

5

544

44

44

44



change of boundary conditions at the edges of the plate when compared to the centre

and so a different strain field is realised there. As mentioned earlier this is likely to

be a large factor in the edge effect phenomenon. It was also observed that a residual

compressive longitudinal strain remains at edge 2 when compared to the rest of the

scan line. This is likely to be a facet of the asymmetry of the process and may be due

to the heat build up and the second edge noted earlier.

Although some similarities were found between the strain gauge

measurements on the larger 200x80mm coupon (figures 4.4.20 to 4.4.22) and the

model prediction at similar locations (on a smaller coupon) in terms of strain cycle,

residual strains and magnitude, it was thought that the strain output along the centre

of the scan line gave a better insight into the longitudinal strain behaviour during the

LF process. The isolated output over time at locations on edge 1, edge 2 and the

centre of the plate along the centre of the scan line on the upper surface are given in

figures 4.5.28 to 4.5.30.


5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

Figure 4.5.29: Longitudinal strain E22, centre of the scan line, plate centre Upper Surface


Edge 1

Centre

4

4



It can be seen in the above figures that the longitudinal strain cycle varies

considerably depending on the location along the scan line. At edge 1 (figure 4.5.28)

it can be seen that a large tensile component occurs as the beam passes. This reduces

to virtually zero on cooling. At the centre on the plate there is an initial compression

observed due to the thermal expansion at edge 1 against the cold material ahead of

the beam. As the beam reaches the centre a reversal to a tensile component occurs;

this rapidly reduces somewhat after the beam passes which may indicate that the

effect ahead of the beam also occurs behind. A small tensile residual longitudinal

strain component remains in the upper surface along the scan line at the centre of the

plate. At edge 2 (figure 4.5.30) a similar effect to the centre can be observed.

However, a much larger tensile component occurs as the beam reaches the edge.

This is consistent with the heat build up at the end of the scan line noted earlier. It

can be seen that a residual compressive longitudinal strain remains in the sheet at

edge 2; this could be due to the increase in thermal expansion against the cooling

surrounding material at this location (due to the heat build up) causing a plastic

compression along the scan line. This asymmetry in the residual longitudinal strains

along the scan line also demonstrates the need for an alternating direction strategy

and a variable traverse speed to realise an even strain cycle at all locations along the

scan line to reduce any unwanted distortion.



Edge 2 4



4.5.6 Transverse Stress S11

As with the strain data the transverse or lateral stress is perpendicular or 90° to the

scan line direction, it is referred to here as S11 or the direct stress in the 1 or x

direction. 3D contour plots of the S11 output at various points during the process can

be seen in figure 4.5.31, the output is in MPa.

The transverse stress development can be seen in figure 4.5.31, as with

previous conventions the largest compressive or negative stress is represented by

blue and the largest tensile stress is red, the scales are once again variable.

It can be observed in the above figure that directly under the beam on the

upper surface there is a compressive transverse stress realised (blue area). Unlike the

tensile transverse strain which gives an indication of the surface movement during

the localised thermal expansion. The compressive stress arises from the thermal

expansion against the cold bulk material consistent with the TGM mechanism.

It can also be observed (figure 4.5.31 b) that a tensile transverse stress occurs

ahead of the beam and to some degree to the rear. A possible explanation for this is

Figure 4.5.31: 3D contour plots of S11 at; a) Start b) Mid Pass c) End of pass d) t=49.5s 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

a) b)

c) d)



given in figure 4.5.32, where the localised compressive stress or pinch around the

beam leads to a tensile stress ahead and to the rear along the scan line.

It can also be noted that on cooling (figure 4.5.31 d) there are residual

compressive transverse stresses remaining at each end of the scan line. This could be

related to the difference in boundary conditions from the centre of the plate to the

edge, since less mechanical restraint exists at the edges and there is less limiting

surrounding bulk material when compared to the centre. This, coupled with the

asymmetric nature of the process, appears to lead to a variation in stress distribution

along the scan line during and after processing.

In order to illustrate the transverse stress distribution over time further, the

S11 output from the centre of the scan line at edge 1, the plate centre and edge 2

were isolated and are presented in figures 4.5.33 to 4.5.35.

CompressionTension

Figure 4.5.32: Schematic of the stress distribution around the laser beam during laser forming

Edge 1



[MPa

]



The above figures back up the observations from the 3D contour plots. At

edge 1 (figure 4.5.33) an initial tensile stress is observed. This rapidly switches to a

compressive stress, which increases on cooling to a large residual compressive

transverse stress (~180MPa). At the centre the positive tensile stress ahead of the

beam can also be observed. This rapidly changes to a compressive stress as the beam

reaches the centre location and this recovers rapidly possibly due to a tensile

component behind the beam. Virtually no residual transverse stress remains in the

plate at the centre after cooling. At edge 2 a similar stress cycle occurs to the centre

with an initial tensile stress ahead of the beam rapidly changing to a compressive

stress as the beam reaches the location. As with edge 1, however, a large

compressive residual stress develops on cooling at the second edge (slightly higher ~

220MPa). These residual compressive stresses at either end of the scan line must

influence the geometry of the formed part and could be an origin of the edge effect

phenomenon.

Centre

Edge 2

Figure 4.5.34: Transverse Stress S11, centre of the scan line, plate centre Upper Surface




[MPa

] [M

Pa]



4.5.7 Longitudinal Stress S22

The longitudinal stress is parallel to the scan line direction; here it is referred to as

S22 or the direct stress in the 2 or y direction. 3D contour plots of the S22 output at

various points during the process can be seen in figure 4.5.36, the output is in MPa.

Once again the largest compressive or negative stress is represented by blue and the

largest tensile stress is red and, as before, the scales are variable.

It can be seen in figure 4.5.36 that, as with the transverse stress, a

compressive longitudinal stress occurs directly under the laser beam consistent with

a thermal expansion against the cold bulk material ahead of the beam (and to some

degree to the rear). Similar tensile areas can be observed either side of the beam to

those observed in the S11 output. These are consistent with the localised

compression causing a mechanical effect around the beam as in figure 4.5.32.

The compressive region under the laser beam recovers immediately to the

rear on the scan line (figure 4.5.36 b) to a relatively large residual tensile stress

surrounded by a compressive region. This persists on cooling (figure 4.5.36 d) along

Figure 4.5.36: 3D contour plots of S22 at; a) Start b) Mid Pass c) End of pass d) t=29.5s 5.5mm beam dia. 760W, 30mm/s, single pass, A=0.85

a) b)

c) d)



the entire length of the scan line apart from the edges. A possible explanation for this

could be due to the longitudinal compressive stress generated by the laser beam, in

that only the transverse compressive stresses that cause a plastic compression are

necessary to produce a bend about the scan line. The addition of a plastic

compression in the longitudinal plane (the plastic compression would be 3D anyway)

may mean that a second bend or moment is generated about the laser beam at 90° to

the scan line in the xy plane along the width of the beam. This moment is attempting

to bend the plate towards the laser beam at 90° to the bend angle produced. As the

moment is limited by the plate stiffness and is outweighed by the moment generated

along the length of the scan line (transverse) the plate cannot bend along the scan

line in the y direction. However, the edge effect deformation may be attributed to

this moment. If the plate cannot bend about the x or 1 axis and the longitudinal

plastic compression has occurred then this would be akin to holding a deformed plate

flat (e.g. a part-cylinder shape). Hence a tensile stress would exist in the upper

surface. At the edges of the plate the local stiffness is less and so sufficient distortion

may have occurred to relieve the residual stresses. The residual compressive region

surrounding the scan line could be a mechanical reaction in the un-deformed part of

the plate to the residual tensile stress along the scan line.

The longitudinal stress cycle over time during the process can be illustrated

further by considering the S22 output from the centre of the scan line at edge 1, the

plate centre and edge 2; these are presented in figures 4.5.37 to 4.5.39.



Edge 1



It can be seen in the above figures that at edges 1 and 2 the S22 output from

the centre of the scan line, although complex, is relatively small and the residual

stresses are effectively zero. At the centre (figure 4.5.38) it can be seen that the

initial compression generated by the laser beam switches rapidly to a large tensile

residual longitudinal stress as observed in the 3D contour plots. This predicted

residual stress of ~200MPa is over half of the tensile yield stress for the mild steel

(table 3.2.3). If correct, tensile stresses this high remaining in a formed component

would certainly be detrimental to its strength in the longitudinal direction. Further

study would be required to confirm this. However, factors such as the effect of

multiple passes, component size and post forming heat treatments may reduce this

unwanted residual stress. Another possibility to reduce this would be in the use of

scanning optics to distribute the energy input rather than along an asymmetric line 89.



Figure 4.5.38: Transverse Stress S22, centre of the scan line, plate centre Upper Surface


Centre

Edge 2



4.6 Metallurgical Study


and 1.6mm AA6061 in three different tempers, O, T4 and T6, to ascertain some of

the effects of LF on the structure and mechanical properties of the materials. Optical

microscopy, Vickers micro-hardness testing and section thickening (for AA6061

only) were investigated; the results are presented in this section. Refer to chapter

3.2.6 for the details of each investigation.

4.6.1 1.5mm Mild Steel CR4

A portion of the iron-carbon phase diagram is presented in figure 4.6.1. Pure iron,

upon heating, experiences two changes in crystal structure before it melts. At room

temperature the stable form (ferrite, or ά iron) has a BCC (Body Centered Cubic)

crystal structure.

Ferrite experiences a polymorphic transformation to FCC (Face Centered

Cubic) austenite, (or γ iron), at 912oC. This austenite persists to 1394ºC, at which

temperature the FCC austenite reverts back to a BCC phase known as δ ferrite,

Figure 4.6.1: Iron-Carbon Equilibrium Phase Diagram with some typical microstructures 122



which finally melts at 1538ºC. All these changes are apparent along the left vertical

axis of the phase diagram.

The material investigated here is a mild steel which contains approximately

0.12wt% carbon (table 3.2.2). This carbon steel has been normalised (held at high

temperature for some time then slowly cooled in air). Above about 900 °C (A3

Temperature), the microstructure consists of austenite. This transforms to ferrite as

the steel cools. The amount of ferrite increases as the temperature decreases, while

the amount of austenite decreases. The solubility of carbon is much lower in the

ferrite than in the austenite, so the carbon concentration of the austenite increases as

the temperature decreases. At 723°C (A1 temperature) the remaining austenite,

which now has a carbon concentration of about 0.8wt%, transforms into pearlite.

This is the eutectoid mixture of ferrite and iron carbide, Fe3C. The iron carbide is

also known as cementite. At low magnifications, the pearlite is the dark phase, and

the light phase is the ferrite. The amount of pearlite in this steel will be quite low due

to the low carbon content.

At higher magnifications, the lamellar eutectoid structure of the pearlite

should be observed. This structure is due to the periodic formation of ferrite and

Fe3C from the austenite in the form of alternating lamella. The spacing of the

lamellae increases with slower cooling rates.

This material is commonly called Mild Steel. The carbon content of mild

steels is typically 0.1 to 0.2wt% carbon. They have moderate strength and high

ductility. They are easily machined, formed and welded. The surface hardness of low

carbon steels can be increased by carburisation.

Although the data available on the microstructure of mild steel cooled from

elevated temperatures is useful, the data is generated at very low heating and cooling

rates thus allowing the microstructures to fully develop. In the case of LF the heating

and cooling rates are very high and so either unique microstructures may form or

little effect may be observed due to the lack of time at elevated temperatures.

Although austenitic temperatures maybe rapidly reached (< 723°C) these are equally

rapidly quenched (into the bulk material) and may either prevent a phase

transformation or lock a phase transformation in place. Some data is available on

rapid quenched steels during heat treatments to improve mechanical properties such

as hardness and ductility. Quenching rapidly from 900 °C will transform the

austenite to martensite instead of pearlite. Martensite is a non- equilibrium,



metastable body centred tetragonal (BCT) phase, this forms by a shear

transformation which takes place at the speed of sound. As there is no time for

diffusion of the carbon to occur, this remains in interstitial sites within the martensite

(a supersaturated solid-solution of C in a BCT lattice). These transformations require

time at elevated temperatures (above A3) in order for the complete diffusion of the

carbon into the austenite to occur. As the time at high temperatures during LF is

short (in seconds) it is unlikely that complete transformation in the heated area will

not occur, however, there maybe regions in the upper surface of partial martensitic

transformation.

The martensite should be seen at high magnification as fine plate-like laths.

The martensite is supersaturated with carbon and has very high hardness and low

toughness. As a consequence, the quenched steel is extremely brittle and weak, it

would normally then be tempered, this decreases the hardness of the martensite and

improves the toughness.

The results from the optical microscopy of laser formed mild steel samples

can be seen in figures 4.6.2 to 4.6.11. Three process parameter combinations were

investigated; 3mm beam diameter 760W, 55mm/s; 5.5mm beam diameter, 760W,

30mm/s; 8mm beam diameter, 760W, 20mm/s. samples were produced at 1, 10 and

30 passes at 60 second intervals between each pass. The as-received microstructure

was also recorded and can be seen in figure 4.6.2. It can be seen that an equiaxed

ferrite structure is present with a small volume fraction of pearlite (dark areas).

Figure 4.6.2: Microstructure of the ‘as-received’ coupon (x500 magnifications)

W N

W N



Figure 4.6.3: 1.5mm Mild Steel, 3mm beam diameter 760W, 55mm/s, 1 pass, Top Middle and Bottom of the HAZ section (x500 magnifications)

W N

W N

W N

Top Middle

Bottom

Figure 4.6.4: 1.5mm Mild Steel, 3mm beam diameter 760W, 55mm/s, 10 passes, Top Middle and Bottom of the HAZ section (x500 magnifications)

Top Middle

Bottom

W N

W N

W N




Bottom

MiddleTop

W N

W N

W N

Figure 4.6.6: 1.5mm Mild Steel, 5.5mm beam diameter 760W, 30mm/s, 1 pass, Top Middle and Bottom of the HAZ section (x500 magnifications)

Bottom

MiddleTop W N

W N

W N



Figure 4.6.7: 1.5mm Mild Steel, 5.5mm beam diameter 760W, 30mm/s, 10 passes, Top Middle and Bottom of the HAZ section (x500 magnifications)

Figure 4.6.8: 1.5mm Mild Steel, 5.5mm beam diameter 760W, 30mm/s, 30 passes, Top Middle and Bottom of the HAZ section (x500 magnifications)

Top

Top Middle

Middle

Bottom

Bottom W N

W N

W N

W N

W N

W N



Figure 4.6.9: 1.5mm Mild Steel, 8mm beam diameter 760W, 20mm/s, 1 pass, Top Middle and Bottom of the HAZ section (x500 magnifications)


Top Middle

Bottom W N

W N

W N

Top Middle

Bottom

W N

W N

W N



It can be seen in the above figures that the effects if any of LF on the

metallurgical structure of mild steel are subtle. With no obvious melting the process

maybe more akin to a rapid quenching, although the quench rates may be much

higher (indicated by FEA presented earlier) and the time at high temperature,

however, is not as long as would be in a quenched steel (only seconds in LF). An

additional complication is that over a number of passes further heating may lead to a

tempering effect or allow for other structures to form. Tempering usually means

heating a steel and holding it above 200°C for some time (~1 hour), the heat allows

any trapped carbon to diffuse and many small carbide precipitates can develop. The

time above this temperature within the heated region during multiple pass laser

forming is significantly more than at the higher temperatures. Tempering leads to a

decrease in hardness when compared to quenched steel (martensitic), however,

tempered steels are usually harder and stronger than normalised steel but possess less

ductility.

For the samples processed using a 3mm beam diameter (figures 4.6.3 to

4.6.5), after 1 pass (figure 4.6.3), the top, middle and bottom of the material cross-

section directly underneath the laser appear unaffected with the ferrite structure and


Top Middle

Bottom W N

W N

W N



some pearlite still present. The FEA study presented earlier (figure 4.5.7) predicted

that a temperature above A1 had been reached (723°C, figure 4.6.1) in the upper

surface area during a single pass such that austenite could form, however, the time at

this temperature is possibly too short for any significant change in microstructure to

occur, this is confirmed by the micrographs of the heated area. After 10 and 30

passes (figures 4.6.4 and 4.6.5) the middle and bottom of the section still remain

relatively unaffected (optically). In the upper surface region after 30 passes a

reduction in the average grain size can be observed, this could be consistent with an

ongoing tempering effect by each subsequent laser pass.

For the samples processed using a 5.5mm beam diameter (figures 4.6.6 to

4.6.8), as with the 3mm beam samples after 1 pass, (for the magnifications used)

there appears to be little observable effect in the top, middle and bottom of the

processed section. After 10 and 30 passes (figures 4.6.7 and 4.6.8) the main

observable effect of the LF process is a grain size reduction which again could be

due to a tempering operation, this effect is present to some degree into the depth of

the material probably due to the larger beam diameter causing a larger depth of HAZ.

For the 8mm beam diameter samples (figures 4.6.9 to 4.6.11) the only

observable effect, as seen in the 5.5mm beam samples, is an ongoing grain size

reduction perhaps due to the persistent elevated temperatures (above 200°C) causing

a tempering effect. This effect, however, extends into the depth of the section

implying that a more uniform heating has occurred. Inspection of the visible HAZ on

the upper and lower surfaces confirms that, whilst there is a thermal gradient present,

the larger beam diameter and low traverse speed produce significant heating into the

section.

Whilst the data obtained in this study is useful to give an insight in to the

metallurgy of laser formed parts, observations at higher magnifications and using

different techniques such as SEM or TEM may reveal more about the specific

metallurgy formed. It is encouraging to note that, optically at least, there is little

observable effect on the microstructure of this mild steel from the laser forming

process using the energy parameters investigated.

Another method of determining the effect of LF on the metallurgy and

mechanical properties of the mild steel is through hardness testing. Each of the

samples presented above were tested for hardness using a Vickers micro hardness

testing device (details presented in chapter 3.2.6). Five locations were tested at the



top (or near to), middle and bottom (or near to) of the section directly under the scan

path and two further locations either side of the mid-section location (figure 3.2.16).

A load force of 9.806N and a dwell time of 10 seconds were used. The results are

presented below; for clarity the Vickers hardness (HV) results are presented

schematically in the orientation and location on each sample that they were taken.

109

108

106

115

117

121

115 117

136

145

129

141 139

155

165

141

162 169

116

118

125

119 117

Table 4.6.2: 1.5mm Mild Steel, 3mm beam diameter 760W, 55mm/s, 1 pass, Vickers hardness

Table 4.6.1: 1.5mm Mild Steel, ‘As received’ Vickers hardness

Table 4.6.3: 1.5mm Mild Steel, 3mm beam diameter 760W, 55mm/s,10 passes, Vickers hardness


Table 4.6.5: 1.5mm Mild Steel, 5.5mm beam diameter 760W, 30mm/s, 1 pass, Vickers hardness



It can be seen in table 4.6.1 that the ‘as received’ hardness values through the

section are comparable with the value quoted in table 3.2.3 for this material.

138

146

125

137 144

138

155

145

155 160

113

120

120

120 119

119

130

119

125 131

126

151

147

155 156

Table 4.6.6: 1.5mm Mild Steel, 5.5mm beam diameter 760W, 30mm/s, 10 passes, Vickers hardness

Table 4.6.7: 1.5mm Mild Steel, 5.5mm beam diameter 760W, 30mm/s, 30 passes, Vickers hardness

Table 4.6.8: 1.5mm Mild Steel, 8mm beam diameter 760W, 20mm/s,1 pass, Vickers hardness





For the data obtained from the samples processed using a 3mm beam diameter,

tables 4.6.2 to 4.6.4, it can be seen that the hardness levels through the section have

increased from the nominal un-processed values. After 1 pass a hardness increase is

even observed near the lower surface. As the number of passes increases the

hardness level increases still further, this could be an indication of the development

of a (harder) plastically deformed zone (through compression) in the upper part of

the section (consistent with the TGM mechanism). It can be observed that there is a

distribution of the hardness values through the section, the highest levels appear in

the mid-section of the HAZ after 30 passes. Near the top surface the hardness value

is lower and could be an indication of the possible tempering effect or grain size

reduction by repeated scans of the laser beam observed earlier (figure 4.6.5). Near

the bottom surface after 30 passes (table 4.6.4) there is a comparable increase in

hardness also. As the optical microscopy of this region revealed little or no

microstructural changes in this area indicating that a lower temperature was reached,

the increase in hardness near the bottom surface of the plate could be due to a cold

working or strain hardening effect.

For the data obtained from the samples processed using a 5.5mm beam

diameter (tables 4.6.5 to 4.6.7) a similar result to the previous can be observed. The

hardness values do increase with increasing numbers of passes and the largest

increases can be observed near the mid-section of the plate thickness. Near the upper

surface the HV level has not changed from pass 10 to pass 30, this could also

indicate the microstructural change due to tempering discussed earlier. The

magnitude of the hardness value near the bottom surface after 30 passes is

comparable with the 3mm beam data, this could indicate some cold working or an

increase in the depth of the plastically deformed zone due to the beam diameter

increase.

For the data obtained from the samples processed using an 8mm beam

diameter (tables 4.6.8 to 4.6.10), once again a similar result can be observed in terms

of hardness increase with increasing number of passes and the distribution of

hardness values. Below the softer upper region a more uniform hardness distribution

can be observed, possibly due to the larger beam diameter creating a more uniform

heating of the section and hence increased depth of the plastically deformed zone or

again the result of strain hardening.



The strain hardening phenomena is attributed to the entanglement of

dislocations. Plastic deformation in metals proceeds atomic step by atomic step by

the generation and movement (by external force) of dislocations (or 1D defects or

holes) within the crystal lattice. The area swept by the movement defines a plane, the

glide plane, the movement of a dislocation moves the whole crystal on one side of

the glide plane relative to the other side. During plastic deformation multiple

dislocations created within the lattice interact during movement, as deformation

continues the dislocation density increases and entanglement occurs. This manifest

itself as an increase in hardness and material strength in the region as further plastic

deformation becomes more difficult. Another possible explanation for the lower

hardness in the upper surface compared to the rest of the section after 30 passes

could be the mechanism of dislocation climb. Once a dislocation movement has

taken place (hence plastic deformation) the dislocations are immobile and trapped by

the lattice. The mechanism of dislocation climb, which is strongly influenced by

temperature, makes immobile dislocations mobile again (by circumventing

obstacles), albeit they may move very slowly, thus a reduction in hardness and

material strength would occur. The elevated temperatures in the upper region of the

section during laser forming may allow this phenomenon to take place and hence

provide an explanation as to why the possible strain hardening effect observed in the

mid and lower parts of the heated section are not as apparent in the upper section.

The concept of strain hardening in the heated area of a laser formed

component has been identified in other research 58 as a significant factor in the fall

off of bend angle increase per pass over multiple passes. This is consistent with the

results and discussion above, whereby if significantly more strain hardening occurs

in the lower part of the section of a bend over increasing number of passes, the

bending strength of the section would increase (for a positive bend), and hence the

moment generated by the plastic compression in the upper surface would be less

effective for each subsequent pass.



4.6.2 1.6mm AA 6061 O/T4/T6

This 1.6mm gauge aluminium alloy AA 6061 is a non-ferrous wrought and age

hardenable 6000 series ternary aluminium alloy whose major alloying elements are

magnesium and silicon. The number of alloying elements creates a complex

relationship between phase transformation data, temperature and weight percentage

of each. Because of this, several coherent precipitates form before the final

equilibrium phase is produced in AA 6061.

The 6000 series alloys use precipitation of an intermetallic phase containing

Mg and Si as a strengthening mechanism. In simple ternary Al-Mg-Si alloys (6061)

the precipitation hardening is based on Mg2Si. Achievement of peak strength entails

two stages of heat treatment. First, the material is soaked at a high temperature close

to the melting temperature (the solution heat treatment) to dissolve solutes.

Quenching from the solution temperature creates a supersaturated solution of

alloying ingredients in the Al lattice, and in this condition (the W condition) the

material is soft. Subsequent heat treatment at an intermediate elevated temperature

(the aging heat treatment) facilitates clustering of solute atoms and eventually

formation of ultra-fine semi-coherent precipitates. The maximum hardening is found

for very small precipitates (~4 nm) of non-equilibrium phases. A sequence of

different phases can occur, depending on the composition and aging temperature.

Aging to peak strength is identified as the T6 condition. Excessive aging time or

temperature causes precipitate coarsening with attendant strength loss, commonly

known as over-aging or T7. If the material is left to age naturally at room

temperature after solution heat treatment some incipient solute clustering and

strength increase will occur - this condition is denoted T4. Copper is added to 6000

series alloys to increase the peak aged strength above that achievable with Mg and Si

alone. Sufficient copper changes the main precipitating species from Mg2Si to a

quaternary intermetallic, i.e. AlxMgySizCuw. 122

Due to the complexity of the metallurgy and limited analysis techniques

(optical microscopy only), only a limited study of the microstructure itself was

possible for this work. A typical micrograph of the AA 6061 F (as fabricated, no heat

treatment) microstructure is given in figure 4.6.12.



It can be seen in the above figure that the detail that can be obtained optically

of the microstructure of this material is limited. What can be observed are the large

grey particles of Fe3SiAl12 and the large black particles of Mg2Si.

Three tempers of this alloy were considered O, T4 and T6 (details of each of

these tempers are given in chapter 3.2.1). The data presented here was acquired from

the samples used in the empirical study presented earlier. Each of the samples shown

in this first study have been processed using the following energy parameters; 3mm

beam diameter, 500W, 55mm/s traverse speed. Various number of passes up to 30

have been investigated (details given in chapter 3.2.6). It was seen in section 4.1.4

that the forming of this material over a number of passes was largely influenced by

the integrity of the absorptive coating, such that providing the coating was not re-

sprayed the bend angle rate per pass could fall off dramatically after approximately

10 passes. It is likely that the results presented here will be influenced by this

phenomenon. The optical micrographs of the polished and etched upper surfaces of

the laser formed samples can be seen in figures 4.6.13 to 4.6.20.

Figure 4.6.12: Typical microstructure of AA 6061 F (x250 optical) 122

Figure 4.6.13: AA 6061 O ‘As Received’

Figure 4.6.14: AA 6061 O After 5 passes



Figure 4.6.16: AA 6061 T4 ‘As Received’

Figure 4.6.19: AA 6061 T6 ‘As Received’

Figure 4.6.15: AA 6061 O After 30 passes

Figure 4.6.17: AA 6061 T4 After 5 passes






It can be seen in the above figures that a number of effects of LF on this

material can be observed. It can be noted firstly that the ‘as received’ microstructure

of the O or annealed condition (figure 4.6.13) appears quite coarse when compared

to the more refined heat treated, cold worked and aged T4 (figure 4.6.16) and T6

(figure 4.6.19) microstructures. Little difference can be observed optically between

the T4 and T6 tempers (T4 naturally aged, T6 artificially aged) the main difference

should be observed in terms of tensile strength and hardness.

After 5 passes a dendritic or coarser structure can be observed near the top

surface in the O condition (figure 4.6.14) and extending deeper into the material in

the T4 (figure 4.6.17) and T6 (figure 4.6.20) samples. The difference between the

samples could be related to the large differences in thermal conductivities between

them (table 3.2.15); in that the O condition has the lowest thermal conductivity and

therefore the elevated temperatures would not extend as far into the material. It can

also be observed that the individual precipitate particles have reduced within this

coarse dendritic formation; this may be consistent with a coarsening of the

precipitates due to excessive heating which may lead to a loss of strength in the heat

affected area. It can be noted that the first 5 passes correspond to a region were a

high bend angle rate was possible and so an efficient coupling of the laser energy

into the surface was possible (section 4.1.4).

After 30 passes in all three of the tempers (figures 4.6.15, 4.6.18, and 4.6.21)

it can be observed that the microstructure appears to be returning to the as received

state, with the coarse dendritic structure observed after 5 passes confined to just the

very upper surface. A possible reason for this could be the coating degradation and

hence bend angle rate fall off after 10 passes observed in the empirical study. This

would mean that the peak temperature experienced by the material will fall off also

for a higher number of passes, and although very little forming occurs due to the

coating loss, there may be sufficient energy coupled into the surface to produce a

heat treatment effect. The elevated temperatures at higher number of passes may

allow for a refinement of the coarse precipitates formed earlier, akin to a post-

forming heat treatment.



The micro hardness testing results can be seen in table 4.6.11. The results

presented are average values for the HAZ taken from 18 tests points (figure 3.2.15).

The data presented are for samples processed at 500W, 55mm/s and a 3mm beam

diameter at various numbers of passes.

Hardness Testing at Different Number of Passes (HV) No. of Passes 0 2 5 10 20 30 AA 6061-O 43.39 (10) 44.18 (11) 42.58 (12) 43.18 (13) 44.98 (14) 45.97 (15)AA 6061-T4 94.03 (20) 89.88 (21) 90.82 (22) 91.43 (23) 94.01 (24) 95.99 (25)AA 6061-T6 108.3 (30) 101.6 (31) 84.40 (32) 85.08 (33) 87.97(34) 89.57 (35)

The number in brackets is the number of the sample from table 3.2.19

It can be seen in table 4.6.11 that the ‘as received’ hardness values are

comparable to those quoted from the literature in table 3.2.15 (AA 6061 T4 slightly

higher though).

For the AA 6061 O it can be seen that there is little change in the hardness

values for increasing number of passes. This is perhaps consistent with the optical

microscopy presented earlier, where any noticeable effect on the material was

confined to the very upper surface (figure 4.6.14) and likely to be missed by the

hardness test. This material is quite soft to start with and any additional hardness

gained from precipitation hardening has not occurred, such that excessive heating

has little effect on the already coarse microstructure.

For the T4 and T6 tempers in can be seen that up to 10 passes there is a

decrease in the average hardness within the heated area. From 10 passes up to 30

passes there is a recovery somewhat in the hardness values. For the T4 temper this

has reached the pre-forming levels after 30 passes. This again is consistent with the

optical microscopy results earlier, in that the possible precipitate coarsening

observed after 5 passes would reduce the hardness within the heated area and the

subsequent heat treatment effects of the poorly coupled laser beam act to restore the

original microstructure and hence the hardness to some degree.

A study into the irradiated section thickening phenomenon was conducted on

these materials. Optical micrographs and measurements of the sheet thickness at the

scan line were taken at various numbers of passes (samples processed at 500W,

55mm/s and a 3mm beam diameter). The results are presented below.

Table 4.6.11: Hardness results for AA 6061 O/T4/T6



Figure 4.6.22: AA 6061 O - a) 0 pass, b) 5 passes c) 30 passes

Figure 4.6.23: AA 6061 T4 - a) 0 pass, b) 5 passes c) 30 passes

Figure 4.6.24: AA 6061 T6 - a) 0 pass, b) 5 passes c) 30 passes

a) b) c)

a)

a)

b)

b) c)

c)



Thickness Measurements At Different Numbers of Passes (mm) No. of Passes 0 2 5 10 20 30 AA 6061-O 1.61 (10) 1.68 (11) 1.73 (12) 1.80 (13) 1.85 (14) 1.84 (15)AA 6061-T4 1.53 (20) 1.58 (21) 1.66 (22) 1.70 (23) 1.73 (24) 1.73 (25)AA 6061-T6 1.55 (30) 1.62 (31) 1.68 (32) 1.71 (33) 1.71 (34) 1.73 (35)

The number in brackets is the number of the sample from table 3.2.19

It can be seen in figures 4.6.22 to 4.6.24 and table 4.6.12 that there is a

significant increase in section thickness with increasing number of passes. It can be

seen in table 4.6.12 that for all three of the tempers the section increase occurs up to

10 passes and then falls off; this is consistent with the dramatic bend angle rate fall

off around 10 passes for these materials. Some further section thickness increase

occurs in the O condition but as can be seen in figure 4.1.58 some additional forming

was possible in this material after 10 passes at these energy parameters.

The fact that the section thickness does increase with increasing number of

passes must influence the bend angle rate at higher number of passes (providing the

absorption can be maintained). This effect has been noted before in other materials 29

and is considered a major factor in the bend angle rate fall off phenomenon. If the

section is thicker then the material will be harder to bend due to the increase in the

section modulus, or alternatively the moment generated about the section for a given

set of energy parameters is less effective. Given the simple beam theory equation for

bending:

zRE

IM σ

== (4.6.1)

Such that:

REIM = (4.6.2)

And:

12

30bsI = (4.6.3)

For a given radius of curvature R the required moment M is governed by the section

thickness S0 cubed. Hence for any increase in thickness the generated moment by LF

must increase by a factor cubed in order to maintain the same radius of curvature.

Table 4.6.12: Irradiated section thickness measurements for AA 6061 O/T4/T6



4.7 Closed Loop Control 125

As discussed earlier due to the inherent variability in the mechanical properties of

metallic components, such as an unknown residual stress history, there is variability

in the laser forming characteristics between any two identical samples. In addition,

as demonstrated earlier in this chapter, there is no guarantee of repeatability for

given process parameters for an open loop set-up due to variation in forming

efficiencies depending largely on the processing parameters, number of passes

realised and in conjunction with this, the condition of the absorptive coating (if used).

In order to demonstrate that laser forming can be used to produce repeatable accurate

bends a system is presented in this section for the closed loop controlled 2D laser

forming of 80x80mm coupons of two materials, 1.5mm mild steel and 0.9mm

AA1050 - H14.

The factors considered essential for control of the process were:

1. The current bend angle.

2. The difference between current and desired bend angle.

3. The current bend angle rate or bend angle increase per pass.

4. Selection of a bend angle rate per pass so as to avoid overshoot (when the

bend angle difference between current and desired angle is small, i.e. bend

angle rate should be less than or equal to the required deformation).

As observed in section 4.1 the bend angle rate per pass in a given material can

be selected by varying the laser power, beam diameter and traverse speed. For the

laser system used in this investigation the process speed is the easiest variable to

control and so it was decided to base the 2D LF control software around the selection

of process speed based on the above criteria. Key to the control of LF, therefore, is

the selection of process parameters that give the largest possible range of bend angle

rates for the speed range of the CNC tables (i.e. 5 to 95mm/s). For the first material

considered, 1.5mm mild steel CR4, the process map data using a 5.5mm beam

diameter and a laser power of 760W was used, this is shown in figure 4.7.1. This

data range gave a bend angle rate selection per pass between 0.5° and 2° in the speed

range 15 to 70 mm/s. These values can be used as a basis for control, they cannot be



taken as absolute as the bend angle rate per pass for a multi-pass strategy is variable

for any given material and constant processing parameters.

The concept of the system is that forming would initially start using the ideal

forming parameters for the material or ones that give a good bend angle rate per pass

(in the case of the mild steel, 30mm/s). The bend angle per pass would then be

monitored. The first part of the control loop that was created was a simple check to

compare the current bend angle to the desired bend angle, providing that this was not

equal to or greater than the desired angle the process could continue on the next pass

if this was the case the process would be terminated. This certainly provides a

method of closed loop control, however, there is a possibility of considerable over

shoot if the current bend angle rate is more than the bend angle difference between

the current bend angle and the desired angle (i.e. if 20° is required, the current bend

angle is 19.5 and the current bend angle rate is 2° per pass an over shoot will occur).

Thus the bend angle rate must be controlled based on the difference between the

current bend angle and the desired angle. This is controlled via the process speed and

selected by the control program per pass for the mild steel based on the calibration

data presented in figure 4.7.1, such that as the desired bend angle approaches the

process speed increases in order to decrease the bend angle rate. Another factor to be

taken account of is the decrease in bend angle rate per pass with increasing number

of passes over the same track observed earlier. This means that at higher number of

passes the calibration data is less reliable thus the speed selection per pass as the

desired bend angle is approached had to be tuned empirically. The results of the first

attempt at the closed loop laser forming up to a target angle of 20° are shown in

Figure 4.7.1: Laser forming of 1.5mm mild steel CR4, 3mm beam dia. 760W 10 – 70mm/s process map



figure 4.7.2. Shown on the secondary axis (right hand side) is the speed selection

from the control program for each pass.

It can be seen in figure 4.7.2 that for up to 11 passes the control software

keeps the processing speed at the initial high forming rate of 30mm/s. Once the bend

angle is within 5° of the target the traverse speed is increased for the next pass up to

40mm/s, the fall off in bend angle increase per pass (or rate) can clearly be seen

(pass 12 to 17) because of this. Once the current bend angle is within 1° of the target

the speed has been increased to 50mm/s, unfortunately at this speed and number of

passes realised (due to bend angle rate fall off) no additional forming was possible

such that the target angle could not be achieved within a reasonable amount of passes.

The basic concept of closed loop control was proven however. An improvement to

the speed selection code was made in order to achieve the target angle; this second

attempt result is shown in figure 4.7.3.

Figure 4.7.2: Closed loop laser forming of 1.5mm mild steel CR4, 3mm beam dia. 760W, 20° target, attempt 1

Figure 4.7.3: Closed loop laser forming of 1.5mm mild steel CR4, 3mm beam dia. 760W, 20° target, attempt 2



It can be seen in the above figure that, as for attempt 1 (figure 4.7.2), the high

bend angle rate speed of 30mm/s was used for the initial forming parameters. Once

the bend angle is within 5° of the target the speed increases to only 35mm/s this time.

It can be noted that for this coupon 14 passes were required to achieve a bend angle

over 15° as opposed to 11 in the previous attempt, this highlights the variable nature

of the process (due to factors such as different residual stress conditions in the

samples) and demonstrates the need for closed loop control. The drop off in bend

angle rate is more subtle for this speed increase at 5° to go. At 2° from the target

angle (as opposed to 1° previously) the speed was increased again to 40mm/s for the

last few passes. It can be seen the target angle has been achieved this time to within

0.2°, at this point, as the target angle has been exceeded, the control program exits

and the component is complete. As discussed in the experimental set-up section

earlier (chapter 3.1), the resolution of the range finder used coupled with the

measurement distances for the triangulation of the bend angle results in a ~0.25°

reliable accuracy for the system. Due to this factor the result above can be

considered accurate to within the resolution of the system. For subsequent trails it

was decided to implement an additional control parameter based on these result such

that the program would exit (i.e. no more passes realised) if the current bend angle

was within 0.25° of the target angle so as to avoid any overshoot however small.

In order to demonstrate that this method of closed loop control of LF is

possible in other materials and other thicknesses, the system was reconfigured for the

LF of 0.9mm AA1050-H14. As with the mild steel, the start point of the control loop

is the selection of process parameters that give the largest possible range of bend

angle rates for the speed range of the CNC tables.

Figure 4.7.4: Laser forming of 0.9mm AA1050, 3mm beam dia. 300W 10 – 90mm/s process map



For this material the ideal parameters are a 3mm beam diameter, 300W and the

speed range 10 to 90mm/s, these can be seen in figure 4.7.4 (taken from the

empirical study presented earlier). These parameters give a range of bend angles

from 0.25° to 3° in this material. A much larger range is available here when

compared to the mild steel, possible due to the thinner and weaker material, this may

mean that more (subtle) control of the process is possible here. For the first attempt a

similar set-up was used to the successful mild steel study (two speed increases), the

results are given in figure 4.7.5. The initial ideal forming parameters were taken

from empirical study in which a traverse speed of 35mm/s was used for a multi-pass

study.

It can be seen in figure 4.7.5 that, although the target angle has been found

(reasonably accurately) and the control program has terminated after pass 10, the

forming rate per pass leading up to the target angle is very high and it is only

coincidence (i.e. just the right amount of forming on the last pass) that has led to the

accurate bend angle without significant overshoot. This is due to the use of only two

speed increases, the second (within 2° of target) only allowing one pass before the

target was exceeded (due to excessive bend angle rate at 70mm/s). The solution to

this was to introduce more speed increases when approaching the target angle and to

employ very high speeds (85mm/s, correspond to a 0.25° bend in figure 4.7.4) for

the final few passes so as to reduce the bend angle rate considerably and coalesce

with the target over more laser passes. The result of the second attempt using closed

loop control for LF of the AA1050 sheet is shown in figure 4.7.6.

Figure 4.7.5: Closed loop laser forming of 0.9mm AA1050, 3mm beam dia. 300W, 20° target, attempt 1



It can be observed that, by the introduction of more speed steps (at within 10°,

5° and 2° of the target angle) with much higher traverse speeds as the target angle

approaches, the bend angle can be controlled with a great deal of accuracy.

Developed from the work on mild steel the control loop is allowed to exit if the

current bend angle is within 0.25° of the target angle, this is to take account of the

resolution of the measurement system. In order to demonstrate that the system can be

used to form the material to any positive bend angle in a controlled way, the result

using a target angle of 30° is shown in figure 4.7.7.

The results in this section demonstrate that it is possible to produce

controlled accurate repeatable 2D bends using laser forming independent of the

residual stress history and non-uniformity of a material. By using the easily

controlled process speed the bend angle rate can be selected so as to avoid overshoot.

Accuracy in the process is only limited by the resolution of the sensor used for

feedback control. The higher the resolution of the sensor or sensing method the more

accurate the bend produced in a component.

Figure 4.7.6: Closed loop laser forming of 0.9mm AA1050, 3mm beam dia. 300W, 20° target, attempt 2

Figure 4.7.7: Closed loop laser forming of 0.9mm AA1050, 3mm beam dia. 300W, 30° target



4.8 Thick Section and Large Area 2D Laser Forming for

Ship Building 22

As discussed in the literature review section earlier the ship building industry is an

industry sector where the use of laser forming has a great deal of potential. To be

relevant to the ship building industry, however, the process must be capable of thick

section large scale processing. A study was conducted on thick section 2D laser

forming of mild steel in order to investigate the factors influencing a scaling of

known scan strategies for thinner section materials; the results of this study are

reported here. An attempt was made to reproduce the work on part-cylinders using

thinner section material 29. Details of the set-up and experimental procedure used are

given in chapter 3.2.8. The study was conducted on 5 mm thick mild steel using

three different laser systems (all high power CO2 lasers) and two plate sizes

(360x190mm and 800x400mm).

For a part-cylinder the scan strategy is relatively simple, a series of straight

line multi-pass bends stepped across the longer or shorter axis should produce the

desired geometry. As the section thickness increases more energy input is required to

achieve the same forming result and if power availability is limited then thick section

forming can be difficult. In order to address this issue with the systems used in this

investigation a ‘Double Pass’ technique that was developed initially for thick section

Ti64 (presented earlier) was adapted for the thick section mild steel. The technique

involves a scan strategy of a pass in one direction followed immediately by a return

pass in the opposite direction; the plate is allowed to cool after each double pass

(forced cooled by compressed air jet to decrease process time). The concept behind

this strategy is that, providing the material surface isn’t damaged on the second pass,

the additional energy input per pass is essentially akin to processing with a much

higher laser power (factor increase dependent on overlapping interaction times). In

addition the elevated temperature remaining in the plate after the first pass may

improve the forming efficiency as a reduction in the temperature dependent yield

stress will occur for the second pass. This implies that the technique could be used

even where the available laser power is not a concern in order to improve the

available amount of forming and reduce the processing time.



The results of the initial study on 360x190x5mm mild steel sheet using the

1.5kW Electrox workstation 2 (described in chapter 3.1) are shown in figures 4.8.1

and 4.8.2. The processing parameters used were; 900W, 6.5mm beam diameter,

20mm/s traverse speed, 30 lines, 10mm step and 6 double pass per line. As the sheet

was small enough to be processed on this workstation the formed sheet geometry

could be verified using the in-built CMM (Co-ordinate Measuring Machine) system.

It can be seen in the above figures that it has been possible to produce a

considerable bend in this thick section material with relatively low laser power using

the ‘double pass’ technique. It can also be seen in figure 4.8.2 that a reasonably

uniform part-cylinder has been produced in this size of plate, showing that the

additional weight of the thicker section sample appears to have little effect on the

outcome. For the next investigation this successful scan strategy was scaled up for

the use on an 800x400x5mm plate using a higher power laser (8kW CO2) and larger

translation stages (0.9x1.5m). As more power was available the processing

parameters were effectively doubled (except beam diameter) for the doubled sheet

size, these were; 1.8kW, 6mm beam diameter, 40mm/s traverse speed (doubled

traverse speed means that the overlap time is the same as the smaller sample for the

Figure 4.8.1: Part-cylinder formed from 390x180x5mm mild steel plate

Figure 4.8.2: CMM 3D contour plot of part-cylinder geometry formed from 390x180x5mm mild steel plate



double pass), 35 lines, 20mm step, 6 double pass per line at 60 second intervals. This

strategy is shown schematically in figure 4.8.3.

The results of the above strategy can be seen in figures 4.8.4 to 4.8.6. As the

plates were too large to be measured by the CMM system the formed geometries

were confirmed by manually taking Z height measurements at 10 and 20mm steps

along the two longer X and two shorter Y edges of the plates (as the plate sits

unclamped on the work bed) and inferring the geometry in the centre (backed up by

observations). These results are shown in figures 4.8.5 and 4.8.6.

400mm

800mm

60mm

35 lines at 20mm steps

Figure 4.8.3: Schematic of the LF strategy used to form a part-cylinder along the Y axis in an 800x400mm sheet.

Figure 4.8.4: 800x400x5mm mild steel sheet formed into a part-cylinder.

x y



It can be seen in the above figures that a considerable amount of forming has

been possible in the 800x400mm sheet (~40mm max). From the edge height

measurements it can be seen that the shape is very uniform with a ~3mm difference

in height between the shorter edges (figure 4.8.6). It can also be seen in this figure

that there is very little evidence of edge effects on the shorter edges. This may

indicate that as the bending line increases in length the factors that cause the edge

effect phenomenon are somehow reduced, this will require further study to fully

understand this. For the sample size and when considering the larger tolerances used

in the shipbuilding industry the observed non-uniformity (~3mm) in the plate above

after forming is extremely promising.

A third study was conducted using the 800x400x5mm mild steel plates

forming longitudinally along the X or longer 800mm axis. This was to ascertain

whether forming was possible over such a long scan track and if any significant

Figure 4.8.5: Height measurements along the two longer X axis edges of an 800x400x5mm mild steel sheet formed into a part-cylinder.

Figure 4.8.6: Height measurements along the two shorter Y axis edges of an 800x400x5mm mild steel sheet formed into a part-cylinder.



distortion would occur. A third CO2 laser system with limited power capability had

to be employed here due to a failure in the previous system during the investigation

(details chapter 3.2.8). Due to the system change the forming parameters were re-

tuned, the following were used; 1.8kW, 6mm beam diameter (different energy

distribution to before), 83.3mm/s speed, 35 lines, 10mm step, 3 double pass per line

at 60 second intervals. The higher traverse speed does allow for a better overlap of

the passes along the 800mm track. The schematic of the scan strategy used here can

be seen in figure 4.8.7. The results using this scan strategy are shown in figures 4.8.8

to 4.8.10.

400mm

800mm

30mm

35 lines at 10mm step

Figure 4.8.7: Schematic of the LF strategy used to form a part-cylinder along the longitudinal X axis 800x400x5mm mild steel sheet.

x y

Figure 4.8.8: Laser forming a part-cylinder along the longitudinal X axis from 800x400x5mm mild steel sheet.



It can be seen in the above figures that it has been possible to laser form

along the 800mm axis to produce a reasonably uniform part-cylinder shape. The

amount of forming has only been limited by the power level used, ideally because of

the increased scan path length an increase in power coupled with the increased

traverse speed would allow for the same energy fluence whilst maintaining the same

overlap time for the double pass. In figure 4.8.9 it can be seen that over the 800mm

length the plate remains relatively straight with only a slight deviation near the edges

of only ~2mm. On the shorter edges (figure 4.8.10) it can be seen that, whilst the

same curvature is present at either end, there is an offset between the two of

approximately 2mm which may indicate a slight longitudinal bowing of the plate. As

mentioned the distortion level recorded in this sample when compared to

shipbuilding tolerances are promising.

Figure 4.8.9: Height measurements along the two longer X axis edges of an 800x400x5mm mild steel sheet formed into a part-cylinder along the X axis.

Figure 4.8.10: Height measurements along the two shorter Y axis edges of an 800x400x5mm mild steel sheet formed into a part-cylinder along the X axis.



The results here demonstrate that bends of this length and maybe longer are

possible using LF which shows promise for large scale component manufacture.

Providing a sufficiently high enough power level can be maintained a high traverse

speed can be realised, this is to avoid any temporal or asymmetric factors that may

occur when forming a large component at slow traverse speeds. Ideally a scan

strategy should be realised as fast as possible on a component.

A fourth study was undertaken using a thermocouple technique to confirm

the double pass strategy used on the larger plates (800x400x5mm) and to ascertain

the effect of laser line heating on the rest of the plate. The measurement locations

and scan line investigated are shown in figure 4.8.11. The thermocouples were

placed on the top surface apart from location 1 (on the scan line), where a hole was

drilled in the bottom surface (to a depth of ~2.5mm) and the thermocouple tip

inserted and held in place with adhesive. The results of this study are shown in

figures 4.8.12 and 4.8.13. Given is the temperature response from the three locations

during one double pass and three successive double passes at 60 second intervals

using; 1.8kW, 6mm beam diameter, 83.3mm/s.

40mm

40mm

60mm 40mm

40mm

200mm

400mm

800mm

‘K’ Type Thermocouple Locations

1

2

3

Figure 4.8.11: Thermocouple measurement locations



It can be seen from the thermocouple data above that the double pass effect

has been confirmed. Although not as overlapped as the data recorded using 3.2mm

gauge Ti64 (figure 4.1.50), the temperature rise for the second pass (figure 4.8.12)

per double pass is built on the temperature remaining in the scan line from the first

pass. This is therefore akin to forming with greater power plus the heat remaining in

the plate aids the process by reducing the heat dependent flow stress. The overlap per

pass would be greater if the laser power and hence traverse speed were larger. Over

the three double passes recorded (figure 4.8.13) it can be seen that the peak

temperature per pass is increasing, this effect was noted in the thermocouple study

using single pass strategies presented earlier. The temperature increase per pass is

built on the bulk material temperature which is steadily increasing. This effect may

be beneficial to the process as an increase in the bend angle rate per double pass

Figure 4.8.12: Thermocouple output, 1 double pass, 800x400x5mm mild steel sheet

Figure 4.8.13: Thermocouple output, 1 double pass, 800x400x5mm mild steel sheet



could be occurring over the three passes as the plate is heated up. It can be seen that

at the other locations monitored little effect of the laser is experienced, this is likely

due to the size of the component and the relatively low thermal conductivity of the

mild steel, this emphasises the localised nature of the LF process

It can be seen in all of the above figures that the plate dimensions and amount

of forming prove to be a useful demonstration of the laser forming process in terms

of ship component manufacture. In particular for the manufacture of hull

components that require high accuracy such as the bulbous bow (discussed earlier in

the literature review, chapter 2.7.1). It is unlikely that the whole component could be

laser formed in one, more likely is that the shape could be split into manageable sub-

components or surfaces and (laser) welded together after laser forming to make the

final shape. There may well be an upper limit to the thickness of material that can be

laser formed (thicker material than 5mm is widely used in shipbuilding). The

limitation may largely be down to available laser power and usable larger beam

diameters (to match the section thickness increase). Whilst future work is planned

utilising the full capability of the 8kW Ferranti laser system, it may be necessary to

find alternative heating sources for the process for thicker section materials, such as

induction coils and stadium lights. These may provide a cheaper cost-effective

method of forming thicker materials where the unique capabilities of the laser are not

necessarily required. A point that was argued earlier, however, was that a laser

system can provide a user with a versatile cutting, welding, marking, surface heat

treatment, etc capability as well as laser forming. If a system was purchased for the

other applications it could well be used for laser forming as a bonus application thus

making it more cost-effective.



4.9 Laser Forming of Novel Materials – Metal Laminate

Composite (MLC) Materials 125, 126

The application reported here demonstrates how the laser forming process can be

used to form recently developed high strength metal laminate composite materials

(MLC) or fibre metal laminates (FML). These materials due to their construction and

high strength, are difficult to form once manufactured using conventional techniques.

The aim of this study was to demonstrate the potential of laser forming as a

manufacturing tool for MLC materials, either as a means of direct fabrication or a

means of alignment and distortion removal.

4.9.1 Feasibility Study

The initial concept for the use laser forming with MLC materials was specifically

based around the development of new thermoplastic based fibre reinforced

composites with high melting temperatures. It was considered that it may be possible

to form the material as if it were the equivalent thickness of a metallic solid,

generating a semi-uniform thermal gradient through the thickness (TGM 11) (Figure

4.9.1). Initial tests on 1.38mm thick 2/1 glass reinforced polyamide based MLC

using 800W, a 5mm beam diameter and 80mm/s processing speed produced

excessive melting of the upper 0.3mm Al 2024 layer, with little or no heat transfer

into the lower layers. This led to the perhaps expected conclusion that due to the

extreme differences in thermal properties between each of the layers in a MLC lay-

up it is not possible to set-up a uniform thermal gradient through the thickness.

However, just as the TGM causes a plastic compression of just the upper surface

layers of a solid metallic section, it was thought that by forming by TGM the upper

aluminium layer alone a moment could be generated sufficient to bend the material

section (Figure 4.9.2). Initial tests of this theory found that it was difficult to set up

the TGM across the 0.3mm thickness, the material tended to buckle (Buckling

Mechanism 11) and as the upper layer was constrained the buckle tended to be away

from the laser and hence delamination occurred. Figure 4.9.1 shows the result for a

2/1 polyamide based MLC processed at 300W, 3mm beam diameter and 90mm/s.



By tuning the processing parameters for a high thermal gradient across the

thickness of the upper layer it was possible to produce a significant bend in the

1.38mm 2/1 polyamide based MLC without any melting or damage (Figure 4.9.2).

Figure 4.9.3 shows the results of increasing numbers of passes using two

different processing parameters, hence demonstrating the feasibility of using laser

forming to bend MLC materials. Additionally what can be noted from this is the

relatively small energy input required to bend the material.

Figure 4.9.1: Treating the section as a metallic solid results in a buckling of the upper layer due to non-TGM parameters and excessive heating

Figure 4.9.3: Laser forming of 1.38mm 2/1 glass reinforced polyamide based MLC

Figure 4.9.2: Laser forming the upper layer alone results in a positive bend, no melting and no obvious damage

Laser forming the upper layer alone produced significant bending

2/1 Glass Fibre Reinforced Polyamide FML



4.9.2 Laser Forming Characteristics of MLC Materials A number of studies were conducted in order to determine the laser forming

characteristics of laminated materials. The first study was a repeatability test in order

to confirm the initial results. Figure 4.9.4 shows the results of laser forming three 2/1

polyamide based MLC coupons at 200W, 90mm/s and a 2.5mm beam diameter. The

results show a good repeatability of the process, in addition it can be seen that there

is a consistent linear increase in bend angle with increasing number of passes up to

20 passes.

A study was also conducted to determine the effect of increasing numbers of

layers for each of the thermosetting materials tested, using the same energy input

parameters for each lay-up. In order to determine the effect of the bottom layer on

achievable bend angle when forming 2/1 structures it was possible to manufacture

1/1 MLCs (figure 3.2.21) without a bottom aluminium layer. Figure 4.9.5 shows the

Figure 4.9.4: Repeatability Test, 1.38mm 2/1 Polyamide based MLC

Figure 4.9.5: The effect of increasing number of layers on the laser forming of polyamide based MLC



results for polyamide based MLC. A comparison is also made with the laser forming

of a single 0.3mm Al 2025 foil at the same energy input parameters. The thickness of

the 3/2 and 4/3 MLCs are 2.35mm and 3.1mm respectively. It can be seen in figure

4.9.5 that as would be expected the achievable bend angle falls with increasing

numbers of layers. For the 3/2 MLC the maximum bend angle after 10 passes is 2º.

This is consistent with the increase in material strength and increasing ratio of depth

of material to available depth of plasticized zone. Hence, the moment generated in

the upper surface generates less overall bend. The limiting effect of the lower layers

can clearly be seen when comparing the 1/1 and 2/1 forming results. The results of

this study for the second material type, a self-reinforced polypropylene based MLC

are shown in figure 4.9.6.

For the study shown in figure 4.9.6 on self-reinforced polypropylene a

slightly higher energy fluence was used; 150W, 1.5mm beam diameter and 90mm/s.

It can be seen that in this material using these energy input parameters a considerable

bend angle can be formed after 10 passes in the 2/1 lay-up, of 20.6º, and in the 3/2

Figure 4.9.6: The effect of increasing number of layers on the laser forming of self-reinforced polypropylene based MLC

Figure 4.9.7: The effect of increasing number of layers on the laser forming of glass-reinforced polypropylene based MLC



and 4/3 lay-ups more forming is seen when compared to the results in figure 4.9.5.

This bend angle increase may be due to the optimisation of the forming parameters

and/or an indication of a difference in strength between the materials and hence

formability. Figure 4.9.7 shows the result for the study on glass-reinforced

polypropylene. The fibre orientations for this study were as standard, bi-directional

and orthogonal. As can be seen a comparison was made with the single 0.3mm Al

2024 foil, a 1/1 MLC and the other standard lay-ups. As with figures 4.10.5 & 4.10.6

the results shown in figure 4.9.7 show the decrease in achievable bend angle with

increasing numbers of layers used. For the same energy parameters used in figure

4.9.6 there is less overall forming indicating an increase in material strength between

the self-reinforced polypropylene and the glass fibre reinforced polypropylene based

MLCs. It can also be seen in figures 4.9.5 to 4.9.7 that the bend angle rate per pass

falls off with increasing number of passes. This may be due to a combination of the

factors, discussed earlier, that influence the laser forming of solid metallic

components such as strain hardening, plus an indication of a mechanical limit were

the non-uniformity of the mechanical properties through the material thickness

allows for a certain amount of distortion, before the bending strength of the material

increases as the lower layers are placed under increasing tensile load.

Figure 4.9.8 shows the results of a study on the effect of fibre reinforcement

orientation on achievable bend angle using a 2/1 glass fibre reinforced polypropylene

based MLC.

The fibre orientations are reported relative to the scan line direction. As can

be seen in figure 4.9.8 the orientation of the reinforcement fibres has a large effect

on the achievable bend angle and hence the material strength. The largest bend angle

21.2º is formed after 10 passes with the fibres parallel to the bending line thus

Figure 4.9.8: The effect of fibre orientation on the laser forming of glass-reinforced polypropylene based MLC



offering little or no additional bending strength to the material. This study gives an

insight into the effect of material anisotropy on the laser forming process. This effect

could be used to improve the formability of a material in a particular orientation.

The results of all these studies show that the effectiveness of laser forming to

produce sharp single bends in these materials decreases with increasing numbers of

layers. However there is sufficient available distortion per scan line even in 4/3 lay-

ups for multiple scan line large radii bends and even the capability to use the process

to align and remove distortion post conventional forming. It can be seen that the 2/1

MLC system shows the best potential for the use of laser forming as a direct

manufacturing tool.

4.9.3 Implications of Laser Forming on Material Integrity It has been shown that laser forming can be used to produce significant bends in

MLC materials, in particular 2/1 MLC lay-ups. It is necessary to determine what

effect this process has on the material integrity, in particular the effect on the

thermoplastic composite material which has a relatively low melting temperature. It

has been described earlier that the approach taken to laser form these laminated

structures relies on forming the top thin layer and as such a very small energy input

is used. In order to determine how much heat is transmitted through the upper

aluminium layer to the composite material a thermocouple study was performed. As

described earlier, a K type thermocouple was mounted on the bottom surface of a

0.3mm aluminium foil under the scan line. The foil was then processed using the

empirically determined energy parameters. Figure 4.9.9 shows the typical output

from this study. Shown is the thermocouple result for 6 passes at 200W, 90mm/s and

a 2.5mm beam diameter.

Figure 4.9.9: Thermocouple output for a 0.3mm Al 2024 foil, centreline bottom surface



As can be seen in figure 4.9.9 the peak temperature seen at the bottom

surface during forming is approximately 65ºC. Due to the thin section and high

thermal conductivity of the aluminium the heat is rapidly quenched after each pass,

and it can be seen that thermal equilibrium is reached after the second pass with no

further increase in peak temperature for subsequent passes. At the temperatures

recorded on the bottom surface and hence the temperature seen by the first

composite layer, there should be little or no effect on the structure of the composite.

This is backed up by optical microscopy of the irradiated area, an example of which

is seen in figure 4.9.10. This shows the irradiated zone of a 2/1 polyamide based

MLC after 5 passes at 200W, 90mm/s and a beam diameter of 2.5mm. It can be seen

that the composite layer appears undamaged, with no de-lamination and no reduction

in the distance between the upper and lower aluminium layers. All other samples

processed at optimum parameters are consistent with this result shown in figure

4.9.10. It can be also noted from this figure that an obvious bump is formed in the

irradiated zone of the upper layer, this is perhaps consistent with the TGM theory 11

in that as the material is shortened laterally in the upper surface layers, to account for

the volume of material, there is a thickening of the section. The effect observed in

the MLC samples is very pronounced when compared to laser formed solid metallic

components, this could be due to the thin section of the upper layer or a unique effect

due to the constraints of the lower layers. Further study would reveal this.

An effect on the MLC structure when processing with non-optimum (TGM)

parameters is shown in figure 4.9.11, this 4/3 polyamide based MLC was processed

using 5 passes at 300W, 90mm/s and a 3mm beam diameter. It can be seen that the

upper layer is cracked. This is thought to be due to excessive heating through the

section leading to a sufficient reduction in yield stress and hence ultimate tensile

Figure 4.9.10: 2/1 polyamide based MLC after 5 passes, 200W, 90mm/s, 2.5mmØ



strength, such that due to the limiting strength of the lower layers, the generated

compression of the upper layer in the irradiated zone is more than can be carried by

the Al 2025 at that temperature and thus a crack forms. Figure 4.9.12 shows how the

laser forming process on laminated materials relies on the ability to transmit the

generated moment through to the lower layers. In other words the process relies on

the strength of the bonds between the layers. In figure 4.9.12 it can be seen that the

bonding between the upper layer and the composite layer has failed and de-

lamination has occurred. On closer inspection it was discovered that the

polypropylene bonding layer had folded back on itself in the mould prior to curing

for this sample and thus an incomplete bond was formed.

4.9.4 Laser Forming of More Complex MLC Components

It has been demonstrated in the previous sections that it is possible to laser form

Metal Laminate Composite materials. For 2/1 lay-ups a considerable single bend

angle is possible, for 3/2 and 4/3 MLCs bend angles of only a few degrees are

possible in a reasonable number of passes. This limits the manufacturing capability

Figure 4.9.11: Upper layer cracked due to non-optimum excessive heating.

Figure 4.9.12: Delamination due to failure in bonding layer.



of the laser forming of MLC components. However, it is possible to form large radii

bends using a series of stepped single bends of no more than a few degrees each.

Examples of this strategy are shown in figure 4.9.13 and 4.9.14; For the part-cylinder

in figure 4.9.13 formed from a 200x100mm 2/1 polyamide based MLC coupon, 12

scan lines were used at 10mm intervals, using just 2 passes per line at 150W,

90mm/s and a 1.5mm beam diameter. From figure 4.9.5, 2 passes at these energy

parameters resulted in a ~2.8º bend angle. Considerably more forming has been

achieved in figure 4.9.14 using the same strategy on a self reinforced polypropylene

MLC, it can be seen in figure 4.9.6 that these parameters give ~5º after 2 passes per

line.

Figure 4.9.13: 200x100mm part-cylinder formed from 2/1 polyamide based MLC

Figure 4.9.14: 240x80mm part-cylinder formed from 2/1 polypropylene based MLC



As discussed earlier there is sufficient available distortion per scan line even

in 4/3 lay-ups for this multi-line strategy and even the capability to use the process to

align and remove distortion post-conventional forming. It can be seen however that

the 2/1 MLC system shows the best potential for the use of laser forming as a direct

manufacturing tool. As it has been reported 127 that for conventional forming a metal

layer needs to be within the material to form the material successfully (i.e. a

minimum of a 3/2 lay-up MLC), laser forming offers a useful tool to produce bends

in 2/1 lay-up materials. It may also be possible to use a 3D laser forming approach to

form MLC materials. However, as can be seen in figure 4.10.8 the inherent effect of

material anisotropy may add an unwanted additional complication to a 3D problem.

Work is presented in this thesis on the development of a system that uses predictive

and adaptive approaches to 3D laser form independent of residual stress history and

non-uniform material behaviour, this work is discussed in a later chapter.

4.9.5 Laser Forming Thermosetting MLC Materials

After the success of laser forming the thermoplastic based MLC’s it was decided to

verify the results using a GLARE type material (GLARE 3 127). This material is a

laminate of 2024 aluminium and glass fibre reinforced epoxy, a thermosetting

material.

As shown earlier, the laser forming process when applied to laminate

structures relies on the bending of the upper layer alone. Therefore in order to aid the

process for this study, it was decided to increase the thickness of the Al2024 layers

to 0.9mm, improving the laser formability of the material as it were by increasing the

achievable moment. The material investigated was a 2/1 glass fibre reinforced epoxy

and 0.9mm Al2024 laminate. The fibre directions were bi-directional and orthogonal.

The samples were again 40x80mm.

As with the work presented earlier on the thermoplastic MLC’s an initial

feasibility study was performed. The first energy parameters investigated were taken

from the successful work on the other materials, 150W, 1.5mm beam diameter and a

speed of 90mm/s. As may be expected there was little or no forming, perhaps due to

the increased thickness and hence strength of the upper layer. It was therefore

decided to increase the laser power to 300W and leave the other parameters the same,



this produced a usable result. This result and the result of a repeatability test can be

seen in figure 4.9.15.

It can be seen in figure 4.9.15 that it is possible to laser form this

thermosetting GLARE type material to the same extent as the thermoplastic MLC’s.

A limit can be seen in the data at approximately 6° which corresponds to a point

were some delamination occurs. This is consistent with the problem of a minimum

achievable bend radius for these materials. In addition the energy parameters used

caused some surface melting, therefore it was decided to increase the beam diameter

to 3mm for a further study to establish optimum processing parameters where de-

lamination does not occur. The result of this study at various processing speeds can

be seen in figure 4.9.16.

Figure 4.9.15: Laser forming 2/1 GLARE type materials, initial feasibility test

Figure 4.9.16: Laser forming 2/1 GLARE type materials at various processing speeds



It can be seen in figure 4.9.16 that the rate of forming per pass is governed by

the energy input. The slower the traverse speed for the same power and spot size, the

higher the energy input and hence the higher the bend angle per pass. At 40mm/s it

can be seen that the bend angle data is reasonably linear until approximately 5°. At

this point (pass 5) there is a significant increase in bend angle rate per pass. As with

the previous study (figure 4.9.15) this also corresponds to a point where some

delamination of the upper layer away from the lower layers can be observed in the

sample. In this case the delamination appears to reduce the bending stiffness of the

section as seen by the upper layer, thus allowing more deformation to occur for the

given energy parameters. At approximately 13° (pass 7) the sample fails. The upper

layer was completely delaminated on the free end of the plate (the other end was still

attached and in the edge clamp). The lower layers sprung back flat, as they were only

elastically constrained and the upper Al2024 layer remained bent. Although this

result demonstrates that it is possible to damage the material using laser forming, it

can also be seen that as with the previous sections there is sufficient available

distortion per scan line for multiple scan line large radii bends and even the

capability to use the process to align and remove distortion post-conventional

forming. This was observed at 60mm/s, that providing the plate is not bent to more

than 5° in a single location, no delamination or damage occurs.

Figure 4.9.17: Laser forming a multiple scan line large radii bend, 2/1 GLARE type material, 240x80mm



In order to demonstrate the concept of multiple scan line large radii bends in

this material, a demonstration part was produced. As with the previous section a

240x80mm coupon was used and processed using 300W, 3mm beam diameter, a

processing speed of 40mm/s, 21 lines, 10mm step between the lines and 2 pass per

line. From figure 4.9.16 it can be seen that these energy parameters would give

approximately 2.5° per line, well below the 5° damage threshold. The completed

demonstration part is shown in figures 4.9.17 and 4.9.18. It can be seen that this

technique employing a number of smaller bends can produce a large overall

distortion. One possible drawback with this technique however, is the fact that only

the upper metal layer is plastically deformed. The lower layers are merely elastically

constrained at such a large bending radius, thus adding possible un-wanted residual

stresses between the layers. A solution to this issue and to the problem of a minimum

bend radius maybe the (laser) pre-forming of each metal layer to a near required

shape prior to bonding, then final alignment and adjustment with a laser forming

technique.

Figure 4.9.18: Laser forming a multiple scan line large radii bend, 2/1 GLARE type material, 240x80mm (reverse angle)



4.10 Application Example – Aero Engine Strut


made to replicate an actual aerospace component. Described earlier in chapter 3.2.10,

an ‘A’ frame strut component from a Rolls-Royce Trent 700 Aero engine (figures

3.2.24 and 3.2.25) was identified as an ideal candidate for laser forming.

An initial attempt to produce the part was made without the aid of accurate

drawings (figure 3.2.24). A flat sheet of mild steel CR4, 400x200x1.5mm, was used

to demonstrate that a part of similar length could be formed and that the whole

enclosure could be produced needing only one welded seam (as opposed to forming

the two separate halves and welding along two seams). The method used and results

of this initial investigation can be seen in figures 4.10.1 and 4.10.2. The energy

parameters used were taken from the empirical study presented earlier (CO2 laser).

Laser form along the 400mm long edges to produce a U channel

1

Laser form along the centre of the plate to fold the sheet over

2

3

800W 5.5mm ∅30mm/s 5 scan lines 5mm spacing 14 pass

800W 5.5mm ∅30mm/s 7 scan lines 5mm spacing 20 pass

As the gap closed it became more difficult for the laser beam to gain access to the inner surface

Figure 4.10.1: Initial method to produce the ‘A’ frame strut from 400x200mm mild steel sheet.



As can be seen in figure 4.10.2 the part produced shows little distortion over

its length, demonstrating that laser forming can produce relatively long bending

edges. It was also found that closing the gap along the open edge was very difficult

as access for the laser beam on the internal surface became limited. It was thought

that the buckling mechanism could be employed on the external surface (negative

bending) to close the gap if necessary.

With access to the technical drawings for the strut component (figure 3.2.24)

it was decided to use the two halves method (as used for the conventional hot creep

forming production method) rather than the more ambitious approach outlined above

(although the above method would be more cost-effective). A second attempt to

produce the component was conducted on 200x100x1.6mm Ti64 sheet (clamped at

the centre, figure 3.2.26). This was to demonstrate that the section geometry could be

laser formed accurately from the correct material (although slightly thinner and

100mm long instead of 574mm). It can be seen in figure 3.2.25 that the section

geometry consist of two relatively sharp bends (near 90°) at the edges and a more

gradual bend in the surface at the centre on the strut. The method used to form this

geometry can be seen in figure 4.10.3 (centre clamping arrangement also shown)

Figure 4.10.2: Result of initial attempt to produce the ‘A’ frame strut from 400x200mm mild steel sheet.

7 Lines, 3mm step 12˚ per line = 84˚ (10 Double Pass) 5.5mm ∅, 740W, 30mm/s

6 Lines, 10mm step 1˚ per line = 6˚ (2 Single Pass) 5.5mm ∅, 740W, 30mm/s

Figure 4.10.3: Method used to produce the ‘A’ frame strut section from 200x100x1.6mm Ti64 sheet.



The energy parameters were selected from the empirical study presented

earlier (section 4.1.2). Although closed loop control of the bend angle produced was

possible (presented earlier) it was decided to use an open loop method of control for

this demonstration part, i.e. an approximate bend angle was known for given energy

parameters and number of passes from the empirical study. The scan strategy was

devised to give an overall 90° bend to the flat areas at the very edges of the plate, i.e.

84° from the sharp bend and a further 6° from the centre section. For this centre part

of the section only a gradual bend is required thus only 6 lines are necessary at a

large step and only two passes each. For the sharper bend (still a relatively large

bend radius i.e. not an 84° bend from a single point) it can be seen in figure 4.10.3

that a double pass strategy was used for the overlapping scan lines. It was found that

due to a combination of the coating degradation per pass and high strength of the

material the required (high) bend angles per scan line could not be achieved easily

using a single pass strategy. Thus by performing trails on 80x80mm coupons of this

material it was possible to determine the number of double passes required to give

approximately 12° per line (10 double pass). The graphite coating was re-sprayed

after each forming line was realised, this served to ensure a high forming rate (as

observed in the empirical study) and distribute the incident energy i.e. flatten out the

near gaussian energy profile so as to avoid excessive heating (and melting) along the

centre of the scan line. The results of this study are shown in figures 4.10.4 and

4.10.5. Forced cooling (compressed air jet) was used throughout. The sharp bends at

the edges were formed first (figure 4.10.4), forming from the outside towards the

centre. The more gradual bend at the centre was then formed from the outer edge to

the centre on either side of the centre clamping bolt (figure 4.10.5). This ensured that

a flat section of plate would be available for each new scan line (constant spot size).

Figure 4.10.4: ‘A’ frame strut section production from 200x100x1.6mm Ti64

sheet. Forming the sharp bends at the edges first



It can be seen in figure 4.10.5 that the formed part does resemble the strut

section shown in figure 3.2.25. This demonstrates that the geometry of this aerospace

component can be laser formed. In addition it shows that this high strength material

can be laser formed into discontinuous surfaces (sharp bends). Providing the graphite

coating is re-sprayed at regular intervals forming can continue to some degree, it is

unlikely, however, that high angle bends (e.g. 90°) would be possible from a single

scan line in this material (and thicker gauges) due to the high strength and sheet

thickness. This is consistent with the concept of a minimum bend radius for a

material during conventional forming based on the sheet thickness and material

strength, such that for increasing thickness it becomes more difficult to produce

sharp internal corners without material damage. This has been avoided here by using

a series of smaller bends in close proximity to produce a relatively large radius near

90° bend.

The final study conducted was an attempt to produce a full sized accurate

laser formed prototype of the strut halve from 574x175x3.2mm mild steel CR4 sheet.

This study was conducted in an industrial environment at the lairdside laser

engineering centre (LLEC) using a 4kW Nd:YAG CW laser and 7 axis robot beam

delivery system (figure 3.2.27). This component was formed using a similar strategy

to the Ti64 section shown above; the strategy can be seen in figure 4.10.6. As the

material thickness and laser type were previously untested a small study was

conducted to determine approximate bend angles for given processing conditions and

numbers of passes, this data could then be used to form the strut section in an open

loop manner. The mild steel still had to be coated with graphite mainly to reduce

back-reflection so as to avoid damage to the fibre delivery system.

Figure 4.10.5: ‘A’ frame strut section production from 200x100x1.6mm Ti64 sheet. Forming the gradual large radii bend at the centre to complete the geometry.



The results of using the above strategy can be seen in figures 4.10.7 and

4.10.8. The sample was clamped using a bolt, centre clamp and drilled hole at the

centre of the plate, this was simply to hold the plate in place and provided little or no

effect on the process (non-contact process). It was not ideal to have to drill an

unnecessary hole in the component (may add localised unwanted residual stresses).

Other clamping arrangements under consideration include magnetic clamps (only

useful for mild steel), vacuum clamps, edge clamps and clamping to a sacrificial un-

formed part of the sample that could be (laser) cut out post-forming.

It can be seen in the above figures that a good approximation of the required

shape has been formed. It can be noted that the reverse side of the strut prototype has

8 Lines, 2mm step 10.5˚ per line = 84˚ (8 Double Pass) 4 mm ∅, 800W, 20mm/s

6 Lines, 7mm step 1˚ per line = 6˚ (2 Single Pass) 4 mm ∅, 800W, 20mm/s

Figure 4.10.6: Method used to produce the full sized ‘A’ frame strut from 574x175x3.2mm mild steel sheet.

Figure 4.10.7: U channel formed first in 574x175x3.2mm mild steel sheet.

Figure 4.10.8: Large radii bend at the centre added to complete the geometry



been painted white to prevent oxidation. The dimensions of the cross-section are

accurate to within 4mm of the component data given in figure 3.2.25, more accuracy

should be gained through the use of online closed loop control. The longitudinal

accuracy over this long a bend (574mm) can be a concern due to the edge effect

phenomenon. Here the initial formation of a U channel (figure 4.10.7) in the

component acts to stabilise the shape by adding considerable bending stiffness in the

longitudinal direction, thus little distortion was observed. An effect that was apparent

was a flaring of the plate towards the end of the scan tracks, this is consistent with

the heat build up phenomena (due to heat flow into the cold region ahead of the

beam) noted in the thermal and FEA analysis presented earlier. This causes a

localised distortion of the plate at the end of the scan line, in particular on the heavily

worked sharp bends. This compromises the dimensional accuracy of a laser formed

part, a possible way around this would be to increase the length of the part and

simply trim (laser cut) the distorted section off the end of the strut (~15mm). An

alternative to this is to employ a variable speed along the scan line, such that towards

the component edge the speed would increase so as to reduce the heat input and

minimise the distortion due to heat build up. Further investigation would be required

to ascertain the increase in speed required and distance over which it takes place for

a given material (size and thickness dependent) and processing conditions. This may

also eliminate the edge effect phenomenon in components that do not possess the

stabilising U channel feature seen here.

The study presented in this section does prove the potential manufacturing

capabilities of the LF process. A reasonably accurate full scale prototype of an

aerospace component has been produced albeit in the wrong material. The next step

in the ongoing work in this area will be the production of a full scale prototype of the

part in 3.2mm Ti64. For this material tight controls on the entrapment of O2 into the

surface have to be taken (discussed earlier in chapter 2.6.6), to this end investigations

are underway (in addition to this thesis) on the use of atmospheric control chambers

and specially designed shrouding nozzles for the laser forming of this material.

Although this study has replicated an existing component, laser forming

offers the capability to alter the dimensions of the component easily (CAD enabled)

without the need to produce another die or former, this is a major advantage of the

process over conventional forming technologies.



Chapter 5

3D Laser Forming –

Results and Discussion This chapter contains the results and discussion of experimental and analytical

studies into the 3D laser forming of a number of materials, including mild steel, and

titanium alloy.

5.1 Empirical Study

Presented in this section are the results and discussions of investigations into the 3D

laser forming of the primitive shapes discussed earlier (the saddle, the pillow and the

twisted shapes) using an empirical approach to determine the scan strategies.

Additional work and discussion are also presented on the use of 3D laser forming on

thick sections, specifically for the ship building industry.

5.1.1 The Saddle Shape 18, 128 The scan patterns used to laser form the plates into saddles were arrived at after

considering how the various mechanisms at work during laser forming may be used

to form a saddle shape. The essential characteristics of forming a saddle from a flat

rectangular piece of material are a shortening of both the diagonals and the length

and width of the rectangle to give rise to the contours of a saddle. Due to this all the

patterns must be symmetrical along the length and width of the plate and have their



centre at the centre point of the rectangular plates. Therefore the initial concept then

was to shorten the centre of the plate using a relatively slow processing speed to

employ the upsetting mechanism and a conservation of volume and an attempt to

assume a zero stress state in the plate itself would produce the saddle shape.

Strategy 1

The concept behind this strategy (Figure 5.1.1) was to shorten the plate along both

its length and diagonals to form the saddle without using a basic ‘X’ shape that

might give rise to faceting or folding effects on the curved surface. The centre line

was irradiated first followed by the arcs in opposite directions. A not to scale

schematic of the strategy is given in Figure 5.1.1.

All of the given contour plots are in the ‘as formed’ orientation, with the laser

direction being vertically down in the Z plane.

Figure 5.1.2 shows that a saddle shape has been formed. However figures

5.1.3 and 5.1.4 show that the saddle is distorted with little or no forming along the

shorter edges. A visual inspection of the saddle showed faceting or folding effects

around the centre of the sample and minimal curving of the short edges. This

strategy was a success as an attempt to shorten the length and width of the plate with

Figure 5.1.1: Scan Strategy 1,

Speed 15mm/s Figure 5.1.2: 3D Contour Plot Strategy 1

Figure5.1.3: Contour Plot Strategy 1 Figure 5.1.4: Contour Plot Strategy 1 (end view)

2

1

3



the minimum of heat input to avoid distortion of the shape. Smooth curvature of the

sample is prevented however due to the fact that the three irradiation lines pass too

close to each other in the centre of the plate, giving rise to the faceting effects in this

area. These effects only became noticeable when the sample had been measured by

the CMM. In Figure 5.1.4 the crease down the centre of the plate is evident. It was

thought a faster traverse speed might aid in avoiding this effect as opposed to

changing the placement of the arcs. Also the plate suffers from a lack of curvature

along its short edges, which is why scan strategy 2 was developed based on strategy

1.

Strategy 2

This strategy was a development of strategy 1. This aimed to achieve the same effect

as the previous but with additional forming of the shorter edges (Figure 5.1.5). An

attempt was made to avoid the faceting effects by increasing the speed. As with the

previous strategy the centre line was irradiated first, then the longer arcs and then the

shorter arcs both in opposite directions.

Figures 5.1.6 and 5.1.7 show that this strategy has successfully produced a

reasonably symmetrical saddle shape, if a little shallow. Forming of the shorter sides

can also be seen. There was still some evidence of faceting and a lack of smooth

Figure 5.1.5: Strategy 2, Speed 20mm/s Figure 5.1.6: 3D Contour plot Strategy 2

Figure 5.1.7: Contour Plot Strategy 2 Figure 5.1.8: Contour Plot Strategy 2 (end view)

1

2

3

4

5



contours at the centre of the plate however, and the saddle was somewhat twisted

(Figure 5.1.8). There was some success in gaining more curvature along the shorter

edges of the plate (Figure 5.1.8). The additional two arcs near the ends of the plate

were added and the centre line shortened in an attempt to achieve this. The increase

in speed did reduce the faceting effects but the drop in heat input to the plate reduced

the overall amount of forming achieved. It was thought that a further increase in

speed would eliminate the faceting but more irradiation lines would be required in

order to maintain the amount of forming and to even out the heat input in order to

avoid distortion. This led to the development of strategy 3.

Strategy 3

The concept behind this strategy was to shorten the plate along its width and length.

Curved lines were used in an attempt to avoid forming a ‘box’ shape. This strategy

used a concentric square circular pattern, the inner square was irradiated clockwise,

the middle anticlockwise and the outer clockwise again to even out any distortion

(Figure 5.1.9).

Figures 5.1.10 and 5.1.11 show that this strategy has also produced a

reasonably symmetrical if shallow saddle shape. Figure 5.1.12, however, shows that

Figure 5.1.9: Strategy 3, Speed 30 mm/sFigure 5.1.10: 3D Contour Plot Strategy 3


1

23



the shorter edges have little or no forming but the contours leading up to the edge do

suggest the formation of the saddle shape. The faceting effects were eliminated with

the increase in speed but the further reduction in heat input into the plate resulted in

less forming and a loss of the smooth contours. However an attempt to increase the

magnitude of forming by slowing the speed down using this strategy resulted in a

distorted sample. It was concluded that a more subtle approach was required. It was

thought that more irradiation lines in the areas where the magnitude of forming was a

problem, namely the shorter sides, were required and this led to the development of

strategy 4.

Strategy 4

This strategy was developed to increase the depth of curvature along the shorter

edges (Figure 5.1.13). The concentric arcs at both of the shorter ends were designed

to accentuate the contours required in those areas. Straight lines were used along the

longer edges as it was thought arced lines were not influencing the final geometry.

This strategy demonstrated the influence of the sequence of irradiation lines within a

pattern. The pattern is concentric and circular and was initially executed processing

from the centre to the periphery. This produced a shape with its highest point at the

centre, not a saddle shape. However reversing the sequence, by irradiating firstly the

arcs at alternating ends and then processing towards the centre using clockwise and

anticlockwise concentric squares, produced the results below.

Figure 5.1.13: Strategy 4, Speed 30mm/s Figure 5.1.14: 3D Contour Plot Strategy 4


1

2 3

4 56

7

8



Figures 5.1.14 and 5.1.15 show that a saddle shape has been formed. The

influence of the concentric arcs is clear along the shorter sides with the contour lines

echoing the irradiation strategy there. The use of straight lines in the X direction

produced the same if not more depth of curvature along the longest side as the arced

lines. Figure 5.1.16 shows that one side of the saddle is higher than the other but

only by approximately 1mm. However, as can be seen on the right hand side of

Figure 5.1.15, this seems to pull the rest of the geometry out of shape. This distortion

could be due to not centring the plate correctly or a pre-stressing of the plate. This

second point can have a large influence on the repeatability of laser forming in that it

is not always possible to know the stress history of a sample. Also, symmetry is

hindered due to the asymmetric nature of the laser forming process itself since it is

impossible to form the whole plate at once.

To develop this strategy further it was thought that the sharp corners should

be avoided, as these can cause distortion due to a ‘hot spot’ where the laser dwells as

the table changes direction. Also working too close the edge of the plate should be

avoided as this appears to influence the generation of a smooth curve along the edge.

These points were taken into account when developing strategy 5.

Strategy 5

This strategy was a development of all the previous attempts. It was designed to

shorten the plate across its length and width in order to give a smooth contoured

saddle. The concentric circular pattern or ‘race track’ strategy was found to work

best when processing from the centre to the periphery. The inner circle was

processed clockwise then each subsequent outer loop in the opposing direction

(Figure 5.1.17). The start points of the loops were spread evenly around the plate as

dwell points occur due to a mechanical delay between the shutter opening and

closing and table movement. It was found that this strategy allowed slowing of the

processing speed in order to increase the magnitude of forming.

Figure 5.1.17: Strategy 5, Speed 20mm/s Figure 5.1.18: 3D Contour Plot Strategy 5

1 2

3

4



This ‘race track’ strategy successfully produced a very symmetrical saddle

shape. Figs. 5.1.18 to 5.1.20 confirm this. The concentric pattern appears to stabilise

the saddle shape, even when processing at slower speeds for additional forming. The

repeatability of this strategy was also very good. Samples processed at the same

parameters are within a 2mm tolerance.

It was also found that further forming could be achieved with this strategy for

a given speed if the plate was supported centrally above the base plate and allowed

to form freely without being hindered by its own weight.

In strategy 5 the straight irradiation lines in the X-axis provide a longitudinal

shrinkage of the plate that causes the longer sides to curve downwards (negative

curvature) and as a result of this and the transverse shrinkage due to the semi-circular

irradiation lines, the shorter sides curve upwards (positive curvature). The direction

and relative magnitude of the forming of each side is dependent on the length to

width ratio of the sample used. Providing the scan strategy is resized accordingly and

beam parameters tuned it is thought a saddle shape could be produced in any size or

type of sheet material. In order to demonstrate this concept a 1.5mm mild steel sheet

with square dimensions was formed (200x200mm). Due to the 1:1 length to width

ratio strategy 5 becomes a concentric circular scan pattern, this can be seen in figure

5.1.21.

Figure 5.1.19: Contour Plot Strategy 5

Figure 5.1.20: Contour Plot Strategy 5 (end view)

1 2 34

Figure 5.1.21: Strategy 5: square plate 20mm/s

Figure 5.1.22: 3D Contour Plot Strategy 5 (square plate)



It can be seen in figures 5.1.22 to 5.1.24 that the modified strategy 5 does

produce a reasonably symmetrical saddle shape. Unfortunately, due to the 1:1 length

to width ratio there is now no guarantee as to which two sides will produce the

negative and positive curvatures. Other factors such as rolling direction, residual

stress condition and perhaps even scan strategy starting location must influence this

result.

In order to demonstrate the use of strategy 5 in other materials a study was

conducted using strategy 5 to laser form 1.6mm gauge 400x200mm sheet Ti6Al4V

(Ti64), a high strength aerospace alloy of titanium. The results are presented in

figures 5.1.25 to 5.1.27. The processing parameters were tuned for the Ti64 and

selected from the empirical 2D LF study for this material (figure 4.1.31), namely

740W, 5.5mm beam diameter and 20mm/s traverse speed.

Figure 5.1.23: 3D Contour Plot (side) Strategy 5 (square) Figure 5.1.24: Contour Plot Strategy 5

(square)

Figure 5.1.25: 1.6mm Ti64. Strategy 5 Figure 5.1.26: 1.6mm Ti64. Strategy 5

Figure 5.1.27: 1.6mm Ti64. Strategy 5 contour plot



The above results in 1.6mm Ti64 demonstrate that strategy 5 (‘race track’

strategy) does produce a consistent reasonable symmetrical saddle shape when used

in other materials and that a considerable amount of forming is available even in this

high strength material.

Strategy 6

In order to demonstrate that there may in fact be multiple solutions to any 3D

laser forming problem a sixth strategy was developed for the saddle shape using

1.5mm mild steel. Whilst, as in strategy 5, the concept of producing a shortening in x

and y was used, a different execution was developed. A cross hatch or X scan

strategy was used (figure 5.1.26), this time using energy parameters consistent with

the TGM; namely 760W 5.5mm beam diameter and 40mm/s traverse speed. This

strategy, rather than forming along the x and y axes individually takes the resultant

vector direction to give the diagonal strategy seen in figure 5.1.28. Unlike the

previous strategies a number of complete passes were realised (all 6 scan lines) and

the plate was measured after each. The results after each pass are presented below.

After pass 1 (6 scan lines) it can be seen in figures 5.1.28 to 5.1.31 that a

shallow reasonable symmetrical saddle shape has been formed. As the amount of

1

2

3

4

5

6

Figure 5.1.28: Strategy 6: 5.5mm beam dia. 40mm/s 400x200x1.5mm Mild Steel

Figure 5.1.29: 3D Contour Plot Strategy 6 (pass 1)

Figure 5.1.31: Contour Plot Strategy 6 (pass 1)




forming was small, perhaps due to a reduction in the energy input (40mm/s instead

of 20mm/s previously), a second and third pass was realised to determine if more

forming could be achieved and symmetry could be maintained.

It can be seen in figures 5.1.32 to 5.1.37 that with additional passes it is

possible to produce additional forming (max ~8mm) thus demonstrating the

possibility of incrementing towards a final shape as opposed to a ‘single shot’

strategy. After three passes the shape is still symmetrical and a saddle shape.

However, the effect of the scan lines crossing over at the centre of the plate results in

a flat region (figure 5.1.36). Although this is exaggerated by the contour plot this

effect would not be desirable when forming a continuous surface and may be a result









of using the same scan lines for each pass since a fold of the material may be present

after three passes due to TGM conditions. A solution to this may be to avoid using

the same scan lines for each pass by offsetting the line for each new pass.

The results of these investigations show that the problem of 3D laser forming

is extremely complex. The active forming mechanisms used in all of the strategies

attempted are a combination of the upsetting and the temperature gradient

mechanism (TGM). Inspection of the heat affected zone on samples processed at

speeds as low as 20mm/s (5.5 and 8mm beam diameters) still show evidence of a

steep thermal gradient through the thickness of the material more consistent with the

TGM than the upsetting mechanism. It may be the case that for materials with low

thermal conductivities such as mild steel the upsetting or shortening mechanism

cannot be entirely active without some TGM being present. Therefore it may also be

the case that 3D LF scan strategies will have to be devised that incorporate or are

tolerant to the out of plane plastic strains developed by the TGM as well as using the

in-plane plastic strains generated by the shortening mechanism.

5.1.2 The Pillow Shape

The pillow shape (figure 3.3.2) is a rectangular formed bowl or dome. A scan

strategy to form the pillow shape from 400x200x1.5mm mild steel CR4 was

developed from work on the saddle shape presented in the previous section. It was

discovered that for strategy 4 (figure 5.1.13) forming from the centre to the periphery

of the plate produced a surface with its highest point (or lowest depending on which

side was measured) at the centre akin to the pillow shape. It was thought that the

inner concentric rectangular scan lines were responsible for this phenomenon. This

theory was backed up by earlier work on the laser forming of dish shapes 64 from

circular blanks since a successful strategy for the LF of a dish was to use concentric

circular irradiation lines. It was thought that by the use of concentric rectangular scan

lines of the same length to width ratio as the 400x200mm plate (figure 5.1.38) a

pillow shape could be formed. In addition it was thought that as the required

distortion was in a positive direction (upwards), energy parameters consistent with

the TGM should be used (positive bend), namely 760W 5.5mm beam diameter and a

40mm/s traverse speed (taken from figure 4.1.2). 8 concentric rectangular scan lines



were used forming from the centre to the periphery in alternating directions to reduce

unwanted distortion. As with the strategies developed for the saddle shape the start

locations were evenly distributed across the plate to avoid heat build up and dwell

points. The results of this strategy can be seen in figures 5.1.38 to 5.1.41.

It can be seen in the above figures that this strategy has produced a

symmetrical pillow or bowl shape with a considerable amount of forming (max

~13mm). This demonstrates that it is possible to form this shape and that the concept

of using a concentric scan pattern of the same shape and length to width ratio as the

blank to be formed is valid. A possible improvement to this strategy, however,

would be the introduction of small radii or fillets at the sharp corners of the

rectangles to avoid dwell points causes by the CNC tables slowing to change

direction.

A possible limitation to this strategy was identified when forming at lower

speeds to increase forming. It was found that if the plate was formed too much or

worked too hard the longer edge tended to buckle and the pillow shape was lost

(figure 5.1.42). This could indicate a forming limit (particularly with the TGM)

where the amount of material within the plate and/or the increased plate stiffness

Figure 5.1.39: 3D Contour Plot Pillow Shape Strategy

Figure 5.1.40: 3D Contour Plot Pillow Shape Strategy

Figure 5.1.38: Pillow Shape Strategy: 5.5mm beam dia. 40mm/s

Figure 5.1.41: Contour Plot Pillow Shape Strategy

1

8



once deformed hinders further symmetrical forming and the plate assumes a

(buckled) zero stress state. This would indicate the need for more in-plane shrinkage

to account for the unwanted material around the edges.

A factor that was unknown when investigating this strategy was that no

account was taken of the beam parameters changing as the sample was distorting in

the Z-axis. As the plate distorted toward the laser on the workbed, when working

below focus of the lens, the intensity realised along the scan lines effectively

increased moving towards the outer edge of the plate. Without the presence of online

beam control, essential when considering 3D LF, a long focal length lens or a

collimated beam should be adequate to take account of small movements in Z of the

workpiece.

5.1.3 The Twisted Shape

The twisted shape (figure 3.3.3) could be considered the most simple of the three

shapes investigated. It was initially thought that a modification of the strategy

developed for the LF of a part-cylinder 29 could be used to create the shape from the

400x200x1.5mm mild steel CR4. By taking the straight parallel line (2D LF)

strategy of the part-cylinder (figure 2.6.8) and setting the scan lines at an angle

(figure 5.1.43) it was thought that a twist in the sheet would be produced. Energy

parameters consistent with the TGM were used, namely: 760W, 5.5mm beam

diameter and a 45mm/s traverse speed. The sample was pinned at one end and

processed from the free end towards the fixed end using a 10mm step in between the

single pass scan lines (figure 5.1.43). This set-up ensured that the plate was flat to

Figure 5.1.42: Distorted pillow shape due to over forming.



the work bed for each new scan line and thus the beam parameters were constant.

The results can be seen in figures 5.1.44 to 5.1.46.

It can be seen in the above figures that the 400x200mm plate had twisted

reasonably uniformly using the strategy in figure 5.1.43. On closer inspection,

however, it was realised that the shape produced is in fact a combination of the

twisted shape and the part-cylinder. From figure 3.3.3 it can be seen that the desired

twisted shape only has a rotation about a single axis (y axis) and that the longer sheet

edges are in fact straight. A strategy therefore was required to remove the part-

cylinder distortion from the twisted shape in the above geometry. A concept of un-

forming the part-cylinder on the reverse side of the sheet whilst leaving the twisted

shape in place was devised, this can be seen in figure 5.1.47.

Figure 5.1.44: 3D Contour Plot Twisted Shape Strategy 1

Figure 5.1.43: Twisted Shape Strategy 1

Figure 5.1.45: 3D Contour Plot Twisted Shape Strategy 1

Figure 5.1.46: Contour Plot Twisted Shape Strategy 1

Fixed End

1

40mm

20mm

19 Scan lines per side

Upper

Lower

Figure 5.1.47: Twisted Shape Strategy 2, 760W, 50mm/s, 5.5mm beam diameter

1



Similar energy parameters to strategy 1 were used. As forming was required

on both sides on the plate it was decided to clamp the plate at its centre (centre hole

drilled and centre clamp used, figure 3.1.7). As can be seen in figure 5.1.43 the

initial scan strategy is similar to strategy 1 on the upper side moving from left to

right in sequence, the results of processing the upper surface (single pass per line, 19

scan lines) can be seen in figures 5.1.48 to 5.1.50.

It can be seen in the above figures that, as with strategy 1, a twisted part-

cylinder has been produced. Slightly less forming is present due to the increase in

traverse speed from strategy 1. It can be seen that the left hand side of the plate

(figure 5.1.50) is slightly higher than the right, this may be due to material variability

or boundary conditions changing from edge to edge and is a facet of forming using

an open loop set-up, since, if for whatever reason less forming than expected is

produced, it is difficult to account for without feed back. A solution to the lack of

symmetry for this set-up could be to alternate from end to end the order of the scan

lines working towards the centre rather than from one edge to the other.

The next step for this strategy was to turn the plate over and process the

lower side (lower side from the initial orientation) in order to remove the unwanted

Figure 5.1.48: 3D Contour Plot Twisted Shape Strategy 2 (upper surface, pass 1)

Figure 5.1.49: 3D Contour Plot (side) Twisted Shape Strategy 2 (upper surface, pass 1)

Figure 5.1.50: Contour Plot Twisted Shape Strategy 2 (upper surface, pass 1)



part-cylinder from the twist. It can be seen in figure 5.1.47 that the scan lines have

the same step size and energy parameters but they have been mirrored about the X

axis in orientation when compared to the upper surface. This should act to negate the

part-cylinder from the plate and accentuate the twist geometry. By turning the plate

over the same CNC part program as the upper surface can be used. The results of the

first pass on the lower surface can be seen in figures 5.1.51 and 5.1.52.

It can be seen in the above figures that although some of the part-cylinder

distortion has been removed the longer edges of the plate still show some curvature

(figure 5.1.47). It was therefore decided to make another pass of the laser over the

surface (another 19 scan lines) in order to attempt to remove more of the part-

cylinder. The results can be seen in figures 5.1.53 and 5.1.54.

It can be seen in the above figures that the twisted shape is becoming more

apparent in the plate’s geometry; two of the diagonally opposing corners are either

high or low. It was decided to process the plate a third time to ascertain whether a

more definitive twisted shape could be produced; a slight curvature of the longer

edges could still be observed (figure 5.1.49) and more of a twist could still be formed.

The results can be seen in figures 5.1.55 to 5.1.57.

Figure 5.1.51: 3D Contour Plot Twisted Shape Strategy 2 (lower surface, pass 1)

Figure 5.1.52: Contour Plot Twisted Shape Strategy 2 (lower surface, pass 1)

Figure 5.1.53: 3D Contour Plot Twisted Shape Strategy 2 (lower surface, pass 2)

Figure 5.1.54: Contour Plot Twisted Shape Strategy 2 (lower surface, pass 2)



It can be seen in figures 5.1.55 to 5.1.57 that after the third pass the geometry

of the plate now resembles the desired twisted shape, thus proving the usefulness of

the double-sided forming strategy. A concern with this strategy is that it is possible

to begin forming a part-cylinder in the opposite direction if the forming for the initial

part-cylinder is not exactly matched on the underside and over forming occurs. This

again is a result of forming in an open loop set up without feed back. Some control,

akin to the 2D LF closed loop control, would be necessary to accurately produce a

desired geometry. It can be noted from the above figures, however, that the final

shape produced is quite uniform and that the initial distortion from the first pass on

the upper surface (figure 5.1.50) has been over written. It can also be noted from this

study that although the same energy parameters were used on the lower surface as on

the upper, it took three passes to produce the desired reversed forming. This could be

due to the additional stiffness of the deformed plate or the fact that no account of the

movement in the z axis was taken, in that when the plate was turned over the lens to

workpiece distance increased and hence less energy input (lower intensity) was

being realised.

Figure 5.1.55: 3D Contour Plot Twisted Shape Strategy 2 (lower surface, pass3)

Figure 5.1.56: 3D Contour Plot (side) Twisted Shape Strategy 2 (lower surface, pass3)

Figure 5.1.57: Contour Plot Twisted Shape Strategy 2 (lower surface, pass3)



5.1.4 Thick Section 3D Laser Forming for Ship Building 22

In order to demonstrate the validity of the 3D LF process for the shipbuilding

industry, a key industry identified earlier as an ideal route for LF to develop into a

manufacturing process, an investigation was performed into the scaling up of the

empirically found scan strategies to larger thicker sheets. In particular, the ‘race

track’ strategy developed for the saddle shape was used to form 5mm thick mild steel

plate of a similar length to width ratio as the samples used in the original work

(section 5.1.1). The first study was conducted on 360x190x5mm mild steel CR4

plates using the Electrox workstation 2 described earlier. As the plates were thicker

than in the original work (1.5mm) the energy parameters had to be scaled

accordingly, namely 1200W, an 8mm beam diameter and a traverse speed of 10mm/s,

5 passes over the same line sequence were used at 1 minute intervals to improve the

amount of forming at this relatively low power level. As the plates were small

enough to be processed on the Electrox workstation 2, they could be measured using

the in-built CMM system. The results are presented in figures 5.1.58 to 5.1.61.

Figure 5.1.58: 3D Contour Plot Saddle Shape, 5mm Mild Steel

Figure 5.1.59: 3D Contour Plot Saddle Shape, 5mm Mild Steel

Figure 5.1.60: Contour Plot Saddle Shape, 5mm Mild Steel

Figure 5.1.61: Saddle Shape, 5mm Mild Steel, image of longer edge



It can be seen in the above figures that a reasonably symmetrical saddle

shape has been formed in the 5mm mild steel sheet, proving that it is possible to 3D

laser form sheets of this thickness by using a higher energy fluence and that scan

strategies developed using thinner materials can be scaled to thicker sheets. It can be

seen in bottom right of figure 5.1.60 that there is a slight twist or distortion in the

sheet. This could be due to an offset in the scan strategy or a temporal effect

resulting from the very low traverse speed due to the low maximum power. The

asymmetry of using a single laser spot to form a large continuous surface is

magnified at low traverse speeds, such that as one portion of the sheet is being

formed the geometry of the rest of the sheet is being influenced and possibly

stiffened, hence, this would be detrimental to the process once the laser reaches other

areas. It is thought that the faster a scan strategy can be realised over the surface of a

component the less the asymmetrical nature of the process will influence the final

result. At a higher laser power a higher traverse speed can be used whilst still

maintaining the energy fluence, and hence the scan strategy can be realised faster.

A second study on much larger sheets of the 5mm mild steel was performed

on a large 5 Axis Laserdyne 890 beam delivery system, employing a 3kW PRC CO2

laser. The sample size used was 800x400mm, the same length to width ratio as the

original work. Due to a fault with the laser at the time of this study a maximum

power of only 1.8kW was available, a slightly smaller beam diameter of 6mm was

used, a traverse speed of 83.3mm/s (able to be increased due to higher power and

smaller spot size) and 25 passes at 1 minute intervals over the same track were used.

r = 40mm r = 80mm

r = 120mm

80mm

120mm

240mm

480mm

800mm

400mm R = 20mm

Direction of laser movement

Figure 5.1.62: Scaled ‘race track’ strategy for the 800x400x5mm mild steel plates



The ‘race track’ strategy had to be scaled for the larger sheets, as can be seen

in figure 5.1.62. It was decided to use the same numbers of lines as with the smaller

sheets so as to investigate how well the strategy scaled up. As the samples were so

large the geometry could only be verified by measurement along the plate edges after

each pass. The results after 25 passes can be seen in figures 5.1.63 to 5.1.65. It can

be noted that the graphite coating was re-sprayed after each pass.

Figure 5.1.63: 800x400x5mm mild steel, height measurements along shorter edges

Figure 5.1.64: 800x400x5mm mild steel, height measurements along longer edges

Figure 5.1.65: 800x400x5mm mild steel plate after processing with ‘race track’



It can be seen in the above figures that a saddle shape has been formed.

Along the shorter edges (400mm, figure 5.1.63) there is a positive curvature and

along the longer edges (figure 5.1.64) there is a negative curvature It can be assumed

from this that a saddle has been formed, this is backed up from observations. It can

be seen that although a saddle has been formed the amount of distortion is very small

compared to the size of the plate (~3mm max). This demonstrates that although the

processing parameters have been scaled (more power could be used however) the

effectiveness of the scaled ‘race track’ pattern has reduced (possibly due to the

increased weight of the plate as well). A solution to this could be to introduce more

scan lines to increase the ratio of the amount of surface area processed to plate

dimensions, to be equal to that used in the 400x200mm plates. As the beam diameter

cannot be scaled (e.g. 8mm to 16mm) and still ensure forming when moving to the

larger plates, more scan lines are needed. Further study is required to confirm this.

However, the potential of the process for shipbuilding applications (outlined earlier)

with the addition of the 2D thick section work (chapter 4.8) has been shown.

As discussed earlier in the literature review the development of an online

monitoring system with predictive distortion correction abilities is a requirement if

any 3D laser forming operation is to be used in a manufacturing environment. The

results of all the empirical studies presented demonstrate the need for such a system

due to the unknowns that can be present when forming in an open loop set-up, such

as residual stresses and variability in the absorption of the incident laser radiation (a

large factor identified from the 2D empirical work). A foreseeable problem with a

system which makes online distortion correction during processing is that, as with

strategies used in this investigation, the final geometry of the part is not reached until

sometime after processing has stopped, when the plate has cooled somewhat and the

elastic stresses have been released leaving a plastically formed part. This suggests

that a strategy of a one off single pass to produce a required geometry would be

extremely difficult to predict and control. A more sensible method of producing a

required geometry would be to increment towards it over a number of passes, taking

surface measurements after each pass so as to have the ability to take account of any

unwanted distortion. The development of a closed loop system for 3D laser forming

based on the recommendations and knowledge gained from the empirical studies is

outlined in the following sections.



5.2 Development of a Geometry based Model for 3D Laser


It was concluded from the empirical study that in order to develop control of the

process of 3D laser forming it was necessary to have the ability to define the surface

to be formed. In addition by defining the surface and analysing properties such as

gradient and curvature, it was thought this may lead to a method of scan strategy

prediction. To this aim, a method of surface creation and analysis was devised using

Matlab. This study concentrated initially on the pillow shape (figure 3.3.2), as this

was the more likely candidate for use in the 3D laser forming demonstrator system,

other shapes were investigated once the model was shown to produce useful results.

The key model developments and the results from it are given in the following

sections. For experimental procedures and set-up refer to chapter 3.3.2.

5.2.1 Initial Predictions and Results of Scan Paths for the Pillow

Shape

In order to further understand and control the 3D laser forming process it was

considered essential to be able to accurately define and analyse a desired surface.

There are a number of methods of defining a surface available (e.g. by equation,

z=fn(x,y) ), one of the more flexible methods is the ‘Bezier Surface Patch’. The

Bezier surface patch is the surface extension of the Bezier curve and is widely used

in surface definition and graphic rendering for computing applications such as CAD

and 3D gaming. Whereas a curve is a function of one variable and takes a sequence

of control points, the patch is a function of two variables with an array of control

points. Most of the methods for the patch are direct extensions of those for the curves.

Figure 5.2.1: The Bezier surface patch130



Bezier patches are defined by a 4×4 grid of evenly spaced control points that

form a surface made up of nine rectangular sub-patches (figure 5.2.1). These control

points (in x, y & z) can be thought of as specifying the desired shape of the patch; it

will attain this shape within the limits imposed by smoothness and continuity. The

Bezier patch is generated "above" the control point grid and interprets the shape of

the grid to create a surface that is smooth and continuous. A Bezier patch does not

necessarily pass through all of its control points - only the four corner points of the

control grid are guaranteed to lie on the surface of the patch. Mathematically, a

Bezier patch is defined by a 4×4 matrix P that contains the heights of the sixteen

control points. The patch is generated by the function P(u, v) for values of u and v

that are between 0 and 1. The parameter u corresponds to the distance along one side

of the patch while v corresponds to the distance along the perpendicular side (figure

5.2.1). A general point on the surface is then given by:

(5.2.1)

Where Bi(u) and Bj(v) are the vectors of the Bezier basis functions and Pij is a

4x4 matrix of the control points 130. As a surface definition using this method

involves evaluating a great many polynomials Matlab was used to calculate and

display the results. The initial 16 control points were chosen arbitrarily (in unit form,

in the range 0 to 1), the dimensions chosen reflected the 400x200mm plate size used

for the 3D LF investigation presented earlier (i.e. 2:1 length to width ratio). Matlab

has the Bezier surface function built in and can be called via the ‘interp2’ function.

The surface definition output from Matlab for the pillow shape can be seen in figure

5.2.2.

Figure 5.2.2: Matlab output showing a Bezier surface patch for a pillow shape



It can be seen in figure 5.2.2 that the raw surface defined by the 16 control

points is given in upper left of the figure. By applying a Bezier surface patch to these

points it is possible to produce a smooth continuous surface densely populated with

data points (the three other images shown in figure 5.2.2). Using this defined surface

it was then possible to analyse attributes of the geometry for possible laser forming

scan strategy prediction. The first attributes analysed were the localised gradients in

X and Y (dz/dx and dz/dy respectively), these were calculated and isolated from the

dense surface matrix data defined earlier. The ‘gradient’ function in Matlab was used

to perform this operation (example Matlab code with this command usage is given in

Appendix 1). The matrices produced are displayed in the form of contour plots of

constant gradient values over the surface in figure 5.2.3.

It was felt that the contour plots above may indicate a possible scan strategy

for the pillow (or dome) shape. If the two contour plots were overlaid and forming

(using the TGM due to positive required deformation, UM maybe necessary as well

but this will be discussed later) was realised along the lines of constant gradient

values a usable scan strategy may be found. It was essential to test this possible scan

strategy on the 400x200x1.5mm mild steel plates. However, rather than attempting

to approximate the shape of the irradiation contour lines, it was found that Matlab

could output (through code manipulation) a table of X and Y locations for a series of

points along each line. The Galil PC based motion controller for the Electrox

workstation 2 (chapter 3.1.2), as with most CNC controllers, can take a table of X

and Y points and linearly interpolate through them to produce a smooth continuous

line for motion. The text file generated had to then be formatted manually (time

intensive) into the Galil CNC language, adding commands for shutter and execution

sequence control (example file given in Appendix 6). Automation of this operation

by Matlab was possible in a later version of the code; this will be discussed in a later

Figure 5.2.3: Contour plots of constant gradient values in X and Y for the Bezier interpolated surface of the pillow shape



section. As the Matlab program outputs data points in terms of X and Y co-ordinates

the subsequent programmed motion had to be in absolute dimensions relative to an

origin on the bottom left corner of the plate (i.e. 0,0).

The gradient contour output (in terms of data points) from Matlab can be

seen in figure 5.2.4 (9 contour levels selected). This could be realised on the plate in

a number of ways, shown in figure 5.2.4 as well are the results using a 5.5mm beam

diameter, 760W and 50mm/s traverse speed (selected from chapter 4.1 to give TGM

at a low bend angle rate so as to avoid distortion due to over-forming). The surfaces

were verified using the in-built CMM system described earlier. For the results

presented all of the ∂z/∂y contour lines (gradients in Y) and then all of the ∂z/∂x

contour lines were realised in alternating directions (to even out distortion) on an

unclamped (free on the workbed) 400x200x1.5mm mild steel plate.

It can be seen in the above figure that the desired pillow or dome shape has

not been formed with this strategy. A number of variations on this gradient

magnitude based strategy were attempted in addition to the one presented above.

These included single direction approaches, realising the ∂z/∂x lines before the ∂z/∂y

line, alternating the line types (ie one ∂z/∂x then one ∂z/∂y etc.), starting from the

centre to the edge and vice versa, however, the desired pillow shape could not be

Figure 5.2.4: Matlab data point output of the (overlaid) gradient based scan strategy for the pillow shape and forming results. ∂z/∂y then ∂z/∂x, alternating directions, 5.5mm beam diameter, 760W and 50mm/s



produced. This lead to the conclusion that it may not be possible to form in one axis

without influencing or distorting the geometry in the other axis (particularly with the

curved lines here) and that the method of isolating the gradients in X and Y was not

valid because of this. A method of combining these gradients was then sought; this

was achieved by considering the resultant gradient vector of the data presented in

figure 5.2.3. By combining the ∂z/∂x and ∂z/∂y data a resultant gradient vector and

magnitude can be found. This was achieved in Matlab using its ‘quiver’ plot

capability (figure 5.2.5). It can be seen in the quiver plot below that the arrow

directions and lengths are a representative of the resultant gradient vector and

magnitude in X and Y. The density of arrows in the displayed output can be selected

in the Matlab program (governed by the x and y grid density for the surface patch).

From the resultant gradient vector quiver plot a second possible scan strategy

was observed. By producing contours along values of constant gradient vector angles

(or directions) a radial scan pattern emerged (figure 5.2.5). This data was output to a

spreadsheet initially to remove the discontinuity at the centre (caused by the arc Tan

function near zero degrees) before production of the CNC file. The scan strategy can

be seen in figure 5.2.6.

Figure 5.2.5: Quiver plot and contour plot of the resultant gradient vector and magnitude in X and Y for the pillow shape

Figure 5.2.6: Constant gradient vector direction based scan strategy for the pillow shape

Figure 5.2.7: Constant gradient vector direction based scan strategy forming result



The laser forming result of the above scan strategy can be seen in figure 5.2.7.

This was produced using a 5.5mm beam diameter, 760W and a constant speed of

50mm/s (a variable speed along the scan lines may be necessary due to the variable

vector magnitude, this will be discussed later). As can be observed the desired pillow

shape has not been formed. A number of variations were also attempted for this

strategy such as varying the order and direction of each of the scan lines, however,

very similar results to figure 5.2.7 were produced. It can be seen in this figure that

the surface is a reasonably symmetrical saddle shape thus demonstrating yet another

method of producing this surface in addition to those investigated in the 3D LF

empirical study.

By considering the concept of a gradient vector direction in relation to 2D

laser forming using the TGM it was possible to produce a third scan strategy to

investigate. If the resultant gradient vector is in the direction of the bend then in

order to laser form a bend in this direction a scan line at 90° to it would have to be

realised. This concept is illustrated in figure 5.2.8.

Applying the above concept to the resultant gradient vector quiver plot in

figure 5.2.5 involves the rotation of all the gradient vectors by 90° (π/2). The

resultant quiver plot can be seen in figure 5.2.9.

αb

Scan direction (orientation) required to produce indicated gradient vector

direction (at 90° to it)

Gradient vector direction from a simple 2D bend

Figure 5.2.8: Illustration of required forming direction for a given gradient vector

Figure 5.2.9: Quiver plot of resultant gradient vector rotated by 90° for pillow shape



It can be seen in the above figure that a concentric pattern has emerged by

rotating the gradient vectors through 90°. On closer inspection it was realised that

this concentric pattern corresponds to the contour lines of constant height for the

defined pillow surface, this is shown in figure 5.2.9 (slight offset observed due to

quiver density and number of contour lines selected). A test was performed using the

contour lines of constant height as a basis for a scan strategy; the results can be seen

in figure 5.2.10. A variable in this method of scan line output was the number of

contour levels selected, a limit of 9 contour lines was used throughout in order to

reduce the amount of manual formatting required to produce the CNC file. The same

energy conditions as before were used, namely; 5.5mm beam diameter, 760W and a

constant speed of 50mm/s. The scan line start locations and directions were as output

from Matlab and so were reasonably random (should aid the reduction of unwanted

distortion); the lines were irradiated from the inside to the periphery.

As can be seen in figure 5.2.10 a smooth contoured reasonably symmetrical

pillow shape has been formed using the scan strategy based on lines of constant

height of a surface. This could potentially be a relatively straightforward method of

predicting a 3D LF scan strategy for a given surface, providing the surface can be

Figure 5.2.10: Matlab data point output of contour lines of constant height for the pillow shape and forming results. 9 contours, 5.5mm beam diameter, 760W and 50mm/s



defined in the Matlab programming environment. Although a straightforward

solution this is perhaps not an obvious one, an attempt to explain why lines of

constant height give a usable scan pattern for a surface is given in figure 5.2.11.

In the above figure for the forming of a bowl shape from a circular blank the

forming or bend lines would also correspond to contour lines of constant height and

it is thought that this should be the case for other shapes. On a continuous non-

faceted surface the contour levels can be arbitrarily chosen (by Matlab in this case)

and the localised bend angles along the scan lines are considered small enough so as

to not facet the surface significantly. A scan line or a location where a bend takes

place about should always correspond to a line of constant height and vice versa

since the point about which a moment is generated should be stationary in space and

bending legs either side will move instead. The problem does become more complex

due to the asymmetric and temporal nature of the process since a scan strategy

cannot be realised at all points on a plate instantaneously with the current energy

delivery system. Forming one part of the plate will always influence another part (i.e.

adding additional stiffness to the geometry), however, providing the scan speed is

reasonably high distortion should be minimised, the result in figure 5.2.10 indicates

this as well.

Now that a usable method of scan line prediction had been discovered it was

realised that a more subtle approach to the energy input per scan line was necessary.

Observations of the quiver data presented in figure 5.2.9 indicates that a higher

gradient magnitude is required near the edges of the plate as opposed to the centre.

Hence increasingly more forming or energy input is required nearer to the edges.

This is illustrated further in figure 5.2.12, where the resultant gradient vector

magnitude has been calculated for points along the contour lines of constant height

Z

X

Z4 Z3 Z2 Z1

Y

X

Bend Lines

Simple Bowl Shape

Z1

Z4

Figure 5.2.11: Schematic of possible reason why lines of constant height give a usable scan pattern for a surface



for the pillow shape. The vector magnitude is represented by the size of the blue dot

at locations on the lines.

It can be noted in the above figure that the gradient vector magnitude not

only varies between each contour line but there is also a subtle variation along the

same contour line as well. These variations in required gradient vector magnitude

should correspond to variations in required energy input and hence bend angle or

forming requirement to produce the geometry. As was concluded earlier in the

closed loop 2D laser forming study the most straightforward method of varying the

energy input is through the process speed. Changing the process speed for each

contour line is relatively simple, however, variations in speed along a scan line are

more difficult to achieve in CNC terms. The Galil motion controller does have the

capability to vary the table speed dynamically during a program execution, however,

this will be discussed in a later section.

The ability to define and subsequently analyse a surface to be formed is

essential, especially for controlling the 3D LF process. The initial trails presented in

this section have revealed a method of predicting a scan strategy based on contour

lines of constant height of a surface. More subtlety the required energy input varies

for each individual contour line and within each contour line dependent on the

location on the surface and the localised gradient vector magnitude. This work is

expanded upon in the following sections in order to potentially develop a method of

closed loop control for the process.

Figure 5.2.12: Height contour plot of pillow surface with an indication of the required gradient vector magnitude at points along the contour lines



5.2.2 Application of the Model to the Saddle Shape

In order to demonstrate that the Matlab based geometrical analysis model discussed

in the previous section is valid for other shapes, it was applied to the saddle shape

(figure 3.3.1). The control points to define the saddle shape were based on shape data

in the literature that employed the Bezier surface technique130. The control points

used initially (figure 5.2.13) were for a rotated version of the saddle shape when

compared the desired shape in figure 3.3.1. To make this resemble the desired shape

the data was rotated through 90° and a polynomial fit or interpolation was used to fill

the missing data in the four corners. The result can be seen in figure 5.2.13

As before the code produces a number of possible scan strategies. Although

not successful for the pillow shape the prediction based on the lines of constant

gradient in X and Y is noteworthy.

Figure 5.2.13: Matlab output showing a Bezier surface patch for a saddle shape, based on rotated and interpolated control point data

Figure 5.2.14: Contour plots of constant gradient values in X and Y for the Bezier interpolated surface of the saddle shape



The suggested scan lines above are similar to a strategy mentioned in the

literature used in traditional flame forming in shipyards for the production of a

saddle shape7,8. The difference being that the transverse parallel lines would be on

one side of the plate and the orthogonal longitudinal lines would be on the reverse

side. This suggests that there may be scan strategy prediction methods that work in

some shapes and not in others. This is further emphasised by the number of forming

solutions found in the empirical study earlier for the saddle shape, where a number

of distinctly different scan strategies produces similar results (in terms of general

shape).

The resultant gradient vector rotated by 90° overlaid on the contour lines of

constant height is shown in figure 5.2.15.

The above strategy prediction method successfully produced the desired

geometry for the pillow shape earlier. Here, for the saddle shape, the rotated resultant

gradient vector quiver plot again corresponds to the contour lines of constant height

(figure 5.2.15). It can be seen that the scan strategy predicted involves semi-circular

lines divided into four quadrants along each of the edges. It is encouraging to note

that this is similar to the ‘bow tie’ strategy developed in the empirical study (figure

5.1.9). In this strategy a series of concentric bow tie shaped scan lines did produce a

saddle shape to some degree (figures 5.1.10 to 5.1.12). However, the strategy was

discounted as considerable forming along the longer axis was possible but

insufficient forming in the opposite direction was present in the shorter axis (positive

and negative curvature of orthogonal edges being an attribute of the desired saddle

shape). A possible reason for this was gained by analysing the displacement

direction in the z axis (positive or negative) for each resultant gradient magnitude

Figure 5.2.15: Quiver plot of resultant gradient vector rotated by 90° for saddle shape



enquiry location. This can be seen in figure 5.2.16, positive z displacements are

shown in blue and negative in red.

It can be seen in the above figure that by analysing the required forming

direction there is a requirement in this strategy prediction for a positive and a

negative forming direction. Thus the blue areas in figure 5.2.16 indicate that the

irradiation lines should be placed on the upper surface of the plate (positive bending

through the TGM). Similarly the red areas indicated that the irradiation lines should

be on the lower surface of the plate (negative deflection required, for the TGM this

means reversing the plate). This certainly would be an explanation as to why the

‘bow tie’ strategy developed in the empirical study failed to produce significant

forming along the shorter sides (figures 5.1.9 to 5.1.12). According to the results

here, the semi-circular arc lines on the shorter axis should be placed on the reverse

side of the plate.

Unfortunately there was no direct way to isolate the lines for the reverse side

with the current method of CNC file generation from the Matlab output, and so

confirmation of this saddle shape strategy prediction is limited to the comparison to

the empirical study data discussed above. The results in this section do, however,

demonstrate that the Matlab scan strategy prediction method is potentially a

powerful tool for the control of the 3D laser forming process. Improvements to the

CNC file generation method are presented in the following sections, however, the

development of the Matlab code for the output of scan lines on the reverse surface of

Figure 5.2.16: Height contour plot of saddle surface with an indication of the required gradient vector magnitude at points along the contour lines. Blue indicates positive deflection, Red indicates negative deflection



a component was ongoing but was still incomplete for inclusion in this work. This

must therefore become a recommendation for further work. Another factor to

investigate for the two sided scan strategy indicated by this work is the scan line

sequence to give the required shape, i.e. upper or lower surface first or perhaps an

alternating strategy.

5.2.3 Developable and Non-Developable Surfaces – Bending Strain

and In-Plane Strain Requirements for 3D Laser Forming

It was observed in the previous sections that the required energy fluence realised on

a sheet during LF will vary depending on the location on a surface. This can be

attributed to the fact that for a given surface there will be areas that require more

forming than others. A method of determining the distribution of the energy fluence

is to simply use the magnitude of the resultant gradient vector as a scaling factor for

known energy parameters selected by the traverse speed (this will be demonstrated in

the next section). Another method of energy distribution over a surface was proposed

after considering the concept of developable and non-developable surfaces.

Developable Surface (singly curved) Bending strain required only (TGM) TGM simulated by V groves cut into sheet

Non-developable Surface (doubly curved) In-plane strain now required (Shortening, UM) UM simulated by removal of part of the sheet to allow deformation

Figure 5.2.17: Developable and non-developable surfaces, analogous to the 3D laser forming of continuous surfaces.131



The term developable surface comes from mathematics. In mathematics, a

surface is called ‘developable’ if it can be flattened, and it is termed a non-

developable surface if it cannot be flattened. Developable surfaces are special ruled

surfaces which can be unfolded or developed into a plane without stretching or

tearing (or alternatively formed into a surface from a plane without stretching,

tearing or compression). Because of this property, they are of considerable

importance to sheet-metal and plate-metal based industries. Applications include

windshield design, binder surfaces for sheet metal forming processes, aircraft skins,

ship hulls and others 132. The concept of these surfaces is shown schematically in

figure 5.2.17. Given is an example of a developable surface, a part-cylinder, and a

non-developable surface, a dome or pillow shape. An analogy is drawn in figure

5.2.17 to how these surfaces have been formed and what LF mechanism would be

required to form a flat sheet (whole sheet, no cuts) into the two shapes. For a singly

curved developable surface the TGM should be the dominant mechanism used to

produce plastic bending strains and out of plane deformation. For a doubly curved

non-developable surface, material needs to be removed (in-plane) in order to allow

the deformation to take place. This suggests that the shortening mechanism should

be the dominant mechanism when forming this type of surface, the in-plane plastic

shrinkage accounting for the limiting material near the edges (at the expense of the

section thickening). This can be further emphasised by considering the mathematical

analysis of plates during deformation available in the literature 133. From the analysis

of thin plates with small deflections the strain component within a sheet can be

expressed in terms of the deflection of a plate, w. For a developable surface the

strain components at a given location in x and y and the shear strain in the xy plane

(for a plate thickness z) during deformation are given by:

(5.2.1)

(5.2.2)

(5.2.2)

yx

wz

ywz

xwz

xy

y

x

∂∂∂

−=

∂∂

−=

∂∂

−=

2

2

2

2

2

2γ

ε

ε



These terms are for bending strain components only. For a non-developable

surface the in-plane strain must be included in the total strain calculation, this is

given by the addition of in-plane strain terms to the above equations:

(5.2.4)

(5.2.5)

(5.2.6)

In the above equations it was found that the in-plane strain component is the

largest factor in the calculation of the total strain requirement to form a given non-

developable surface (or strain induced by forming the surface). This suggests that in

order to laser form a non-developable surface such as the pillow and saddle shapes

(assumed to be non-developable) significantly more in-plane plastic strain must be

induced than out of plane bending strain. This effectively means that the upsetting or

shortening mechanism (UM) should be employed rather than the temperature

gradient mechanism. It may even be possible to calculate a required strain field (and

hence a scan strategy) in the whole plate for a given shape from the above equations

providing the amount of induced in-plane strain for a given set of UM energy

parameters is known. Liu and Yao et al 101 have presented work (based on work by

Ueda et al 131) using a similar approach to this by using the principle of planar

development or flattening of a final surface (mathematically) to determine the

required strain field to form the shape, a principle of forming normal to principle

curvature directions was also used. A foreseeable problem with this is that the single

pass implementation would be computationally intense (many hours for a small grid

of points) and cannot take into account the residual stress history of a plate. In

addition it is not realistically possible to get exclusively in-plane strains using a laser

forming method. A thermal gradient can be present through the thickness even for

large diameter beams and low traverse speeds, particularly for materials with low

yw

xw

yxwz

yw

ywz

xw

xwz

xy

y

x

∂∂

∂∂

+∂∂

∂−=

∂∂

+∂∂

−=

∂∂

+∂∂

−=

2

2

2

2

2

2

2

2

21

21

γ

ε

ε

Bending Strain (Developable Surface)

In-Plane Strain (Non-Developable Surface)



thermal conductivity such as mild steel; this was shown to be the case in the FEA

study earlier. If a thermal gradient is present there will be an asymmetry in the in-

plane plastic strains generated through the thickness, and hence net bending strains

will develop which may cause out of plane deformation depending on the constraints

of surrounding material. Similarly for the TGM there are in-plane plastic strains

developed in the heated section as well and it is the fact that these are asymmetric

through the section that a net bending strain develops and hence causes out of plane

bending (i.e. significantly larger plastic in-plane strains near the top surface

compared to the bottom surface). The depth of heating and hence the depth of the

plasticized zone (or the extent to which the in-plane strains were present through the

thickness) were found to increase for the larger beam diameter TGM conditions e.g.

5.5mm; presented in the FEA study earlier. A compromise may therefore be found

by forming along lines orthogonal to the principle gradient i.e. contours of zero

gradient or constant height (presented earlier), these are the only paths that are

acceptable for the development of bending strains and in-plane strains at the same

time. The localised energy input along a contour line (controlled by the scan speed)

can therefore be scaled to the bending and in-plane strains from the equations above

resolved in the direction of the principle gradient. The resultant strain components at

angle α to the x axis are given by:

(5.2.7)

(5.2.8)

(5.2.9)

The scaling factor between known energy parameters and the induced strain

can either be found numerically through FEA or empirically by the measurement of

the induced deflection for given energy parameters and desired final shape. In

addition, if an incremental approach was used to form a surface rather than a single

pass implementation, the data obtained from the first pass (providing the final shape

has not been achieved) would provide a strain scaling factor for the subsequent

passes. This would be based on the current plate being formed and so should take

account of residual stress history and material non-uniformity.

The concepts discussed here and the previous sections are implemented in the

next section in the development towards controlled 3D laser forming in the form of a

demonstrator system based on the non-developable pillow shape.

ααεεααγγ

ααγαεαεε

ααγαεαεε

α

α

α

cossin)(2)sin(cos

cossincossin

cossinsincos

22

2290

22

xyxy

xyyx

xyyx

−+−=

−+=

++=

+



5.3 3D Laser Forming Demonstrator System

In order to demonstrate the manufacturing capabilities of the 3D laser forming

process, one of the final goals of the EPSRC funded work programme (for the work

at Liverpool, which the research in this thesis forms part of), was the production of a

3D LF demonstrator system for the controlled LF of one of the primitive shapes

(figures 3.3.1 to 3.3.3) from a 400x200x1.5mm mild steel sheet. As discussed in the

literature review (chapter 2.6.5) either predictive or adaptive approaches could be

taken to achieve this (figure 2.6.12).

Intelligent predictive systems, perhaps based on Knowledge-Based Systems

(KBS), neural networks or thermo-mechanical models can achieve predictability

through a knowledge of the material (including its stress history) combined with a

developed, highly tuned process model / control algorithm.

In an adaptive system the use of sensors to provided accurate controlled

feedback coupled with the development of intelligent control software e.g. neural

network, provides an incremental or even real time closed loop method of accurate

3D laser forming, based on the current part characteristics independent of material

variability e.g. residual stress.

For the system developed in this work aspects of both approaches were used

based on the data presented in the previous sections. A potential method of scan

strategy prediction has been developed based on lines of constant height. The energy

distribution within the scan strategy can be given either by the gradient vector

magnitude or the sum of the bending and in-plane strains resolved in the direction of

the principle gradient. This gives a potential predictive capability to a system.

However, as discussed earlier, the single pass implementation of this would be

computationally intense and cannot take into account material non-uniformity and

residual stresses. The system developed here uses the predictive Matlab model to

give an initial scan strategy based on a required geometry. When the geometry is not

formed within one pass (or over formed), an incremental adaptive approach can then

be used for subsequent passes, utilising the error between the current and desired

geometry to give a new scan strategy. Thus any unwanted distortion due to material

variability can be accounted for. The forming rate and distribution of the magnitude

of forming across the surface can be controlled by the process speed based on the



factors above and the amount of forming required so as to avoid overshoot (similar

to the closed loop 2D LF presented earlier). A strategy of monitoring and controlling

the process during a scan was considered, however, this was not possible with the

current monitoring hardware. In addition it was observed that the final formed

surface was not realised until sometime after forming making online monitoring

ineffective. A strategy of per pass monitoring and control was therefore used.

A number of developments to the control software, both the Visual Basic

motion control software and the Matlab code, contributed to the development of the

demonstrator system presented. A major development was the ability to create the

Galil CNC file directly from Matlab. It was realised that Matlab could easily

generate the text based file in the correct format (including shutter control commands)

thus automating the manual method used so far (figure 5.3.1). This was achieved by

creating tables or arrays of x and y locations points (in absolute co-ordinates) along

the predicted contour or scan lines and using the Galil controller’s ability to linearly

interpolate through these points to create a smooth line of motion. A filter had to be

introduced to reduce the number of data points generated in the CNC file as the

controller memory was limited to 1000 lines of code. This was not detrimental to the

smoothness or accuracy of motion as data points as far apart as 10mm could be

linearly interpolated between successfully. Another important development was the

ability to vary the scan speed, not only for each contour line but dynamically along

each contour as well. This exploits the Gallil controller’s ability to accept for every x

and y location data point (used for linear interpolation) a starting and end speed

between each point. This dynamic control of the process speed allows for the

implementation of a variable energy distribution (controlled by the speed) scaled

dependent on either the gradient vector magnitude or total strain requirements (figure

5.3.1).

#Z1 SB1 SB3 SP16000,16000,30000 AC450000,450000,900000 DC450000,450000,900000 PA54667,32496,0 BG AM CB3 WT200 LM XY VA100000 VD100000 LIX,Y <Start Speed >End SpeedLI-410,2667 <20102 >20130 LI-103,1333 <20130 >20137 LI0,2667 <20137 >20137 LI103,1333 <20137 >20130

CNC file generation by Matlab

Includes variable speed along contour line

Figure 5.3.1: CNC File Generation by Matlab



The version of the Matlab code developed to date did not allow for the

generation of CNC data for scan lines predicted to be on the reverse side of the plate.

Any negative or red indicated scan lines were simply ignored by the subroutine for

CNC file generation. Due to this the 3D LF demonstrator system was based around

the pillow shape (figure 3.3.2) as the required deformation is in a single (positive)

direction. Work was ongoing on the modification of the code for the output of a

separate CNC file for the reverse surface, however, this was not available for

inclusion in this thesis and must remain a recommendation for further work.

A change to how the desired surface was described was also implemented.

The Bezier surface route, although extremely flexible, does not guarantee the defined

surface passes through the specified control points (due to constraints on the

smoothness of the surface produced). As a demonstration of the potential accuracy of

the process was the intention of the system, a more accurate surface definition was

used. The pillow shape falls into the mathematical surface category of an ‘elliptic

paraboloid’ similarly the saddle shape is a ‘hyperbolic paraboloid’. Both these

surfaces can be described by a mathematical equation, these are given below.

(5.3.1)

(5.3.2)

+=

2

2

2

2 by

axZ

−=

2

2

2

2 ax

byZ

Figure 5.3.2: Elliptic paraboloid or pillow shape130

Figure 5.3.3: Hyperbolic paraboloid or saddle shape130

Where a & b define the limits of the surface in x and y respectively.

Where a & b define the limits of the surface in x and y respectively.



The representation of the two surfaces in Matlab using the above equations

can be seen in figures 5.3.4 and 5.3.5. The magnitude of the deflection can be

selected by a scaling factor on the above equations and the surface distribution can

be altered by changing the factors a and b for the same 400x200mm grid.

A test using the new surface definition for the pillow shape was conducted in

order to confirm that the general shape could be formed using the contour lines of

constant height approach. The energy (speed) distribution for this test was scaled

between a manually selected range according to the gradient vector magnitude at

each data point along the contour lines. A beam diameter of 5.5mm, laser power

760W and a speed range 45-55mm/s (selected from the empirical study, uncalibrated)

was used. The 400x200x1.5mm mild steel plate was clamped at the centre (figure

3.3.7). The formed plate geometry was verified using the laser range finder based

CMM system. The scan path prediction and speed distribution output from Matlab

for a pillow shape with 15mm maximum deflection can be seen in figure 5.3.6; the

size of blue dot now represents the magnitude of the speed at that location.

Figure 5.3.4: Matlab output using an elliptic paraboloid definition for the pillow shape

Figure 5.3.5: Matlab output using a hyperbolic paraboloid definition for the saddle shape

Figure 5.3.6: Predicted scan strategy and speed distribution for the pillow shape



The Matlab code has been setup to output the scan or contour lines in a

specific order. The areas requiring the smallest deflection are irradiated first moving

through to the areas requiring the largest deformation last. An alternating direction

strategy is used and the line sequence is distributed over the plate so as to even out

the thermal input and asymmetry of the process. The results of the first test can be

seen in figure 5.3.7. A repeatability test was also performed using this strategy on

two additional plates; the results of these can be seen in figure 5.3.8. A standard

deviation between the three samples can be seen in figure 5.3.9.

Figure 5.3.9: Standard deviation between each of the repeatability tests

Figure 5.3.7: Laser formed elliptic paraboloid based pillow shape, 5.5mm beam diameter, 760W, 45-55mm/s

Figure 5.3.8: Repeatability tests 2 and 3



It can be seen in figure 5.3.7 that a pillow shape has been formed and that the

more circular contour lines of the elliptic paraboloid based pillow shape are present

in the formed sample (as opposed to the elliptical contour lines resulting from the

Bezier surface earlier, figure 5.2.10). The results of the repeatability tests (figure

5.3.8) show that a reasonable repeatability can be achieved using the same scan

strategy, the maximum difference between the samples is 1.5mm. The standard

deviation (figure 5.3.9) between the samples reveals that the largest variation occurs

towards the shorter edges, this suggest that the larger the displacement induced by

LF the larger the deviation and hence reduction in repeatability. This demonstrates

the need for closed loop control for accurate repeatable forming independent of

material and process variability.

Improvements to the control software were made in order to set up the

incremental predictive/adaptive forming approach for the demonstrator system.

Firstly the ability to subtract the current surface away from the desired surface was

included. An assumption was made that the error contour plot between the current

formed surface and the desired surface should give a usable scan strategy for the next

pass. This incremental error based approach should take account of unwanted

distortion due to the process and material variability.

The speed distribution over the plate based on the sum of the bending and in-

plane strains resolved in the direction of the principle gradient was also implemented.

The initial speed distribution was taken from the calibration data available from the

2D empirical study using a 5.5mm beam diameter and 760W (figure 4.1.2). Such

that for given speed an approximate induced displacement and hence strain could be

known. After pass 1 a calibration can be made between the predicted deformation

and the actual deformation to ascertain a strain scaling factor for subsequent passes.

This takes account of the fact that the 2D LF data is taken from a single bend

location from edge to edge (developable surface), and that the same energy

parameters would not necessarily produce the same amount of forming in the centre

of a plate due to the additional stiffness and constraints.

In order to avoid overshoot the minimum process speed (selected initially by

the user) was monitored and varied according to the magnitude of the error between

the current and desired surfaces, such that as the desired deformation approached the

minimum speed distributed over the plate per pass would increase and the forming



rate would decrease. This technique proved extremely useful for the closed loop 2D

LF earlier.

An improvement to the CMM system was also made. A reduction in the

number of data points taken was implemented, this was reduced from 40x20 to

20x10, this reduced the measurement time to 10 minutes from 45 minutes. The

course data set was found to be sufficient to describe the formed surface. For

comparison with the desired surface, however, a more complete data set was

required. A Bezier surface patch was therefore applied to this data. It was found that

by using more control points a better representation than before of the surface can be

gained using the straightforward Bezier 16 point surface definition method.

A forming sequence from the 3D LF demonstrator system is given below,

based on the forming of a pillow shape with 20mm maximum deflection.

Figure 5.3.10: Desired 20mm max deflection pillow shape and error plot between it and the flat unformed sheet

Figure 5.3.11: Predicted scan strategy and speed distribution for pass 1



Figure 5.3.12: Speed selection based on 2D LF data for a 5.5mm beam diameter and 760W. 50mm/s selected as a minimum speed. All other speeds distributed in the range 50 to 85mm/s

Figure 5.3.13: Pass 1 forming result, 5.5mm beam diameter and 760W. Maximum forming ~8mm

Figure 5.3.14: Comparison between formed surface after pass 1 and desired shape, ~12mm difference. Error plot gives a prediction for the next pass



Figure 5.3.15: Scan strategy prediction for pass 2. Calibration with pass 1 data gives a strain calibration scaling factor for the speed based on the current plate’s forming characteristics

Figure 5.3.16: Speed distribution used for pass 2. As there is less required forming the minimum process speed has automatically increased to 67mm/s so as to avoid overshoot. The predicted induced strain has also been adjusted according to the pass 1 data

Figure 5.3.17: Pass 2 results, 5.5mm beam diameter and 760W. ~17mm maximum deflection



Figure 5.3.18: Comparison between formed surface after pass 2 and desired shape, ~4.5mm difference. Error plot now gives a prediction for the next pass. More forming along the longer edges now is required.

Figure 5.3.19: Scan strategy prediction for pass 3. No further calibration is performed after the pass 1 data. The Galil controller can easily reproduce smooth motion based on the complex scan prediction

Figure 5.3.20: Speed distribution used for pass 3. As only fine adjustments are required a speed range of 73.13 to 85.4mm/s is predicted.



Figure 5.3.21: Pass 3 results, 5.5mm beam diameter and 760W. ~21mm maximum deflection (slight overshoot)

Figure 5.3.22: Comparison between formed surface after pass 3 and desired shape, +/- ~2.5mm error. Small overshoot has occurred

Figure 5.3.23: Predicted scan strategy for pass 4 suggests forming on the reverse side of the plate (red dots) to account for the overshoot. Forming had to be ended here as this capability was not yet included in the system



The laser forming sequence shown in figures 5.3.10 to 5.3.24 demonstrates

the potential for closed loop repeatable 3D LF of continuous surfaces. Using an

incremental approach based on the error between the current and desired surfaces it

has been possible to produce a component to within +/- 2.5mm of the target shape.

LF using scan patterns based initially on contours of constant height and then error

difference plots have been shown to produce useful results.

Providing over-forming has not occurred on the first pass it has been possible

to iterate towards the final shape increasing the traverse speed to reduce the bend

angle rate and calibrating for the current plate’s forming characteristics. This is a

much faster route than a single pass implementation by calculation of the required

strain field. It has the potential to produce a final component independent of residual

stress history and material non-uniformity and take account of unwanted distortion,

perhaps brought about by these two factors or process variability.

The energy distribution based on the sum of the bending and in-plane strains

resolved in the direction of the principle gradient has been shown to be of merit.

However, the differences between this method and the gradient vector magnitude

distribution are subtle as the possible speed range is limited to between the manually

selected minimum speed, and the maximum speed where no forming occurs (inferred

from the 2D laser forming empirical data). Further work is required to confirm one

method over the other in the forming of other surfaces and materials.

A number of limitations of the demonstrator system have been identified

from the process trials on the pillow shape. Firstly it is currently possible to

overshoot the target shape by a small degree. Refinements to how the speed is scaled

as the target shape is approached may avert this problem. As the Matlab code can

Figure 5.3.24: Image of a laser formed 400x200x1.5mm mild steel plate showing the complex scan patterns realised over the surface.



currently only produce the CNC data for the upper surface the red negative bending

requirements on the lower surface are ignored. This can lead to additional problems

as the speed is then scaled from the next blue or positive bending requirement and

further over-forming can therefore occur. This must be taken account of in further

developments of the code. On reflection it may be beneficial to over-form to some

degree to reduce the error near the centre of the plate and then turn the plate over to

bend the outer edges back to the required deformation. This highlights another

limitation of the system as no account is taken of the influence on the rest of the

plate of each forming line. The forming lines at the centre of the plate will cause a

deflection of the outer edges and so the amount of forming required near the edges

must reduce. This stems from the use of the large beam diameter TGM conditions

(bending strain) and the assumption that a significant amount of in-plane strain is

still present. It may be the case that there is a limit to the amount of (required)

distortion that can be produced in a non-developable surface using these hybrid

TGM conditions. If a large amount of distortion is required in say a dome or pillow

shape then a shortening regime (large beam and low traverse speed) may be required

to account for the additional material preventing the formation of the desired shape.

A combination between the two mechanisms may in fact be necessary, the TGM for

accurate shape definition and the shortening mechanism to selectively account for

the additional material. The system as it stands, however, should be ideally suited to

the laser forming of developable surfaces such as the part-cylinder and possibly the

twisted shape; this should be shown by further research.

The system presented in this section does demonstrate the potential of the

laser forming process to produce accurate repeatable 3D surfaces in a controlled way.

This suggests that laser forming could be utilised as a direct manufacturing tool or as

a means of distortion removal in an industrial environment (many potential

applications discussed earlier in the literature review). Providing the desired and the

current surfaces can be realised in a virtual way (e.g. CMM data with a Bezier

surface patch), a scan strategy can be predicted to give the final shape. Ongoing

improvements to the Matlab code will hopefully lead to a full realisation of this

concept. Further work recommended on this system includes investigations into the

use of other materials of different dimensions and forming of non-symmetrical

shapes and irregular shapes (non rectangular) so as to ascertain the robustness of this

3D laser forming method.

Chapter 6 Conclusions and Future Work


Chapter 6

Conclusions and Future Work

6.1Conclusions

The conclusions from each of the investigations in this thesis are given in the

following sections.

6.1.1 2D Empirical Study

An empirical 2D laser forming investigation on a number of materials using the

TGM, characterising the 2D laser forming process, was conducted. The materials

were sheet 1.5mm mild steel CR4, 0.9mm AA 1050 H14, 0.9mm to 3.2mm Ti6Al4V

and 1.6mm AA 6061 O/T4/T6. Variables investigated included; beam spot size, laser

power, traverse speed, multiple and single pass strategies, time delay between passes,

bend angle rate and coating degradation. The main conclusions from this study were:

a) Process maps built up at various beam diameters, laser powers and process

speeds revealed the unique 2D laser forming characteristics of each of the

materials and material thicknesses investigated for a single pass. It was

concluded that the thermal conductivity, material strength and section thickness

are the major factors for the differences between each of the process maps

obtained at similar energy parameters. The process maps were found to be

invaluable for the selection of forming parameters for many other studies

throughout this thesis and to further the under standing of the process.



b) It was found that beam diameters larger than the sheet thickness had to be used to

induce temperature gradient mechanism conditions. TGM theory would suggest

that a beam diameter of the sheet thickness should be used, however, for realistic

forming operations it was found that beam diameters of this size would cause

significant surface damage.

c) Multiple pass studies using energy parameters selected from the process map

data revealed the 2D laser forming characteristics of each of the materials

investigated during the development of large bend angles of many degrees. By

monitoring the bend angle per pass with custom built hardware and software it

was possible gain further insight into the process. Analysing the bend angle rate

per pass for all the materials and processing conditions used revealed the

subtleties of the process; most notable is that the bend angle rate varies

considerably during a multi-pass strategy. It was found that during the first few

passes there is an initial increase in the bend angle rate per pass thought to be a

result of the absorptive coating burn-off per pass achieving an optimum coating

thickness. An optimum level is reached after a few passes which then begins to

decline after 8 to 10 passes. The extent of this decline was found to depend on

the material and processing conditions used. For the mild steel and pure

aluminium AA 1050 H14, the rate declines slowly over the 30 passes

investigated. This was attributed to the previously determined factors of strain or

work hardening, section thickening and absorptive coating burn-off. For the

Ti6Al4V and AA 6061 the bend angle rate fell dramatically. The major factor for

this was revealed to be the absorptive coating burn-off which was found to be a

larger factor than previously thought. In the case of the Ti6Al4V the high coating

burn off rate was attributed to the low thermal conductivity of the material

preventing adequate conduction of the absorbed heat into the bulk material, thus

overheating the graphite. For the AA6061 the high reflectivity of the substrate

after some graphite burn-off was concluded to be the reason for the dramatic fall

off. This was confirmed by re-spraying the graphite coating on the samples

which gave an immediate increase in the bend angle rate. The other factors were

still present at higher numbers of passes but were overridden by the graphite

coating burn-off. The use of coatings with laser forming, it was concluded from

this study, is not reliable for certain materials. It is therefore recommended that a



shorter laser wavelength should be employed if the process is to be used in an

industrial environment so as to negate the use of absorptive coatings.

d) An investigation to determine the effect of inter-pass time delay on process

efficiency during a multi-pass strategy revealed that there is an optimum delay

for a given set of process parameters. It was concluded that this is due to a

balance point or a trade off between the heat retained in the coupon aiding the

process by reducing the flow stress and the increased bulk material temperature

reducing the available temperature gradient through the thickness as the laser

beam is passed over the surface.

e) A ‘double pass’ technique was developed for thick section materials. The

technique involves a scan strategy of a pass in one direction followed

immediately by a return pass in the opposite direction; the plate is allowed to

cool after each double pass. The concept behind this strategy is that, providing

the material surface is not damaged on the second pass, the additional energy

input per pass is essentially akin to processing with a much higher laser power,

this was confirmed by thermocouple data. Another factor in this technique is that

on the second pass the heat retained in the irradiated area from the first pass

could serve to produce additional forming by reducing the temperature dependent

flow stress of the material, since a hot plate is easier to form than a cold one.

f) A study was conducted into the 2D LF of 1.6mm gauge AA6061 in three heat

treatment conditions O (annealed), T4 and T6 (solution heat treated, cold worked

and aged). It was found that there were considerable differences in the laser

forming characteristics of the three heat treatments of the same alloy. This was

attributed to, apart from the coating burn-off, the considerable differences in

material strength and thermal conductivity between the materials.

6.1.2 Thermal Analysis

Thermocouple and thermal imaging techniques were used in this investigation to

experimentally determine the transient temperature field in 1.5mm gauge mild steel

during the laser forming process and subsequent cooling. A study was also

conducted into the effectiveness of using forced cooling in the LF process and its

effect on forming efficiency. The main conclusions of this study were:



a) The thermocouple study revealed the temporal temperatures cycles at single

locations on the upper and lower surfaces of 1.5mm mild steel CR4 during single

and multi-pass 2D LF. For the multi-pass study the temperatures recorded were

found to increase with increasing numbers of passes for all the energy parameters

investigated. The peak temperature observed during each pass at each location

increased also. However, the temperature increase was roughly the same for each

pass as the same amount of energy is added each time, it was realised that it is

the bulk material temperature this temperature increase per pass is added onto

which is in fact increasing. This effect may have implications on the efficiency of

the process for subsequent passes, since if the bulk material temperature is

increasing there maybe a reduction in magnitude of the thermal gradient through

the section directly under the beam (consistent with TGM). Another factor is that

the elevated temperatures remaining in the heated area may aid the process by

reducing the temperature dependent flow or yield stress of the material thus

making it easier to plastically deform. By reducing the inter-pass time delay it

was found that the peak and bulk material temperatures per pass increase

significantly with increasing numbers of passes. It was found that a plateau is

reached after a number of passes whereby there is no significant increase in the

peak temperature recorded per pass when compared to the previous pass. It was

concluded that this is likely to be a point where thermal equilibrium is reached.

Where the bulk material temperature of the whole plate has stabilised and the

heat losses due to conduction into the clamp, convection to the air and radiation

to the surroundings are balanced with the heat input per pass.

b) A thermal imaging camera revealed the real time heat distribution in a coupon

during a scan using various processing parameters. The graphite was found to be

problematic when used with this measurement technique. It was found that it was

not possible to measure the sample temperature directly within the beam due to

the incandescence of the graphite. The data surrounding the beam revealed that

the time period and surface area over which heat was retained in the mild steel

sheet increased for larger laser beam spot diameters. In addition it was revealed

that a higher temperature is realised at the end of the scan line when compared to

the beginning. This is backed up by observations of the HAZ at the end of a scan

line, where a widening or flaring can be seen. A possible explanation for this is

that the heat from the incident laser beam and the heat retained behind the beam



is flowing into the cold region ahead of the beam, as the beam reaches the second

edge the heat flowing ahead of the beam cannot travel any further and so a heat

build up occurs, hence the increase in temperature at the second edge, this could

be a source of unwanted distortion in the process such as edge effects.

c) In the forced cooling study it was shown that the addition of a basic cooling

regime influences the thermal cycle in the coupons considerably. With the

addition of a cooling air jet on the under surface of a coupon continuously during

and post processing the temperature cycle stabilised within one pass with very

little increase in peak temperature observed for subsequent passes at all three of

the energy parameters investigated. In addition the effect on the LF process in

terms of bend angle produced was shown to be subtle. The use of forced cooling,

it was concluded, has a potential benefit of decreased overall processing time,

since the relatively long inter-pass delay could be reduced significantly; this

makes its use essential. In addition reducing the thermal input into a component

must be beneficial both for the reduction in unwanted distortion and any adverse

effects on metallurgy.

6.1.3 Displacement / Time Analysis

An investigation was conducted into the displacement (or bend angle development)

of 80x200x1.5mm mild steel CR4 coupons with respect to time during LF using

various energy parameters. The conclusions from this study are:

a) It was found that the temporal displacement characteristics of the coupons during

LF depended greatly on the energy parameters used and the number of passes

realised.

b) For the 3mm beam diameter data on the first pass the major part of the bend

angle development was seen to occur whilst the beam is still on the plate surface,

very little additional movement was recorded after the beam has left the coupon

surface. By the sixth pass it was observed that, although the majority of the bend

angle occurs whilst the laser beam is on the surface of the coupon, the final

deformation or bend angle isn’t reached until some 20 seconds after processing.

This effect became more prevalent with increasing numbers of passes and was

attributed to relieving of the purely elastic stresses during cooling.



c) For the larger beam diameters investigated the first pass is similar to the smaller

beam diameter data. For increasing numbers of passes an ‘S’ curve in the

temporal displacement became more apparent. For the 8mm beam diameter this

became two points of inflection where the bend angle was arrested as the beam

was mid way though the plate. This was attributed to a number of factors, a

change in mechanism to the buckling mechanism, increased in-plane movement

due to the large beam TGM conditions and a delayed counter-bend effect due to

the slow traverse speed and larger beam diameter.

d) It was observed that the counter-bend effect was very small using the energy

parameters investigated and that the effect diminishes with increasing numbers of

passes and with increased beam diameter. This suggests that the counter-bend is

not as significant an event during the LF process as given in the TGM theory. A

possible reason for this was thought to be in the use of larger beam diameters

than what was stated in the TGM theory.

6.1.4 Strain Gauge Analysis

An investigation was conducted using a strain gauge technique to determine the

transverse and longitudinal localised strains close to and far from the scan line

during multi-pass LF of 200x80x1.5mm mild steel coupons at various energy

parameters. The conclusions from this study were:

a) The results of the strain gauge analysis investigation demonstrated the

complexity of the laser forming process even during a simple straight line 2D

bend, a large factor in this is the inherent asymmetry of the process when using a

single point laser source to achieve a symmetrical solution.

b) It has been shown that along an irradiation line depending on where the beam is

and its direction, there is a mechanical effect in the plate ahead and to the rear.

c) Whilst absolute readings of strain are difficult at such high thermal gradients the

general trends in transverse and longitudinal localised strains due to thermal and

mechanical influences have been revealed.

d) A significant difference in mid pass and residual strain output was recorded at

the plate edges when compared to the centre on the top and bottom surfaces close

to and distant from the scan line consistent with edge effect phenomenon.



6.1.5 Finite Element Analysis

A Finite Element Analysis (FEA) model for the single pass laser forming of graphite

coated 80x80x1.5mm Mild Steel CR4 coupons using a CO2 laser source a graded

mesh and edge clamped boundary conditions was developed to improve the

understanding of the process. The model was verified with thermocouple and strain

gauge data. The conclusions from the subsequent analysis are given here:

a) The thermal analysis revealed the temporal and peak temperatures realised in a

mild steel coupon during LF using various energy parameters. The peak

temperature was found to be critically dependent on the absorption coefficient

and was cited as a reason as to why the bend angle rate can fall dramatically with

some loss of coating. The heating and cooling rates in LF were found to be

extremely high; these increased with the smaller the beam diameter.

b) It was shown that as the beam diameter increases the temperature difference

between the upper and lower surfaces becomes less. It was also shown that the

peak temperature observed on the lower surface increases with increasing beam

diameter, consistent with a larger beam and lower traverse speed heating the

section more uniformly. It was also observed that there was a temperature

difference from edge to edge along the scan line during forming due to the

asymmetric nature of the process. A higher temperature is realised at the edge at

the end of the scan line than the first edge.

c) Temporal displacement data similar to that recorded in the displacement/time

analysis study was observed in the model. The majority of the bend angle is

produced whilst the beam is still on the plate surface and little or no counter bend

was observed. An edge effect or longitudinal bowing has been recorded

consistent with observations of experimental results.

d) An insight into the temporal 3D strain field development was gained from the

model. The transverse strain was found to be similar to that recorded in the strain

gauge analysis, the results revealing a residual compressive transverse strain

within the irradiated track. Analysis of the longitudinal strains revealed little or

no residual strain in this direction, however a considerable difference in strain

cycle was observed between the first edge, the centre and the second edge

demonstrating the asymmetry of the process.



e) An insight into the temporal 3D stress field development was gained from the

model also. Analysis of the transverse stress reveals that near the centre of the

plate these is no residual stress, however near the edges there is a large

compressive transverse residual stress, this being largest at the second edge.

Analysis of the longitudinal stress reveals that there is a large tensile stress

residual stress on cooling along the scan line surrounded by a compressive zone.

This tensile stress is predicted to be ~200MPa which is over half the yield stress

of the mild steel. If correct, tensile stresses this high remaining in a laser formed

component would certainly be detrimental to its strength in the longitudinal

direction. Further study is necessary to confirm this and to seek methods of

reducing this value such as post forming heat treatments.

6.1.6 Metallurgical Study


and 1.6mm AA6061 in three different tempers, O, T4 and T6, to ascertain some of

the effects of LF on the structure and mechanical properties of the materials. Optical

microscopy, Vickers micro-hardness testing and section thickening were investigated.

The main conclusions were:

a) The limited optical microscopy of laser formed mild steel at various energy

parameters and numbers of passes revealed that the effects if any of LF on the

microstructure of mild steel are subtle. With no obvious melting the process

maybe more akin to a rapid quenching, although the quench rates may be much

higher and the time at high temperature is not as long as would be used for a

quenched steel.

b) The hardness tests on the mild steel samples revealed that LF does have an effect

on the metallurgical properties of this material. The hardness values do increase

with increasing numbers of passes and the largest increases were observed near

the mid and lower sections of the plate thickness consistent with a cold working

or strain hardening effect. A lower hardness level was present in the upper

surface of the heated section and was attributed to a possible tempering effect by

subsequent passes of the laser.



c) Similarly to the mild steel data, the optical microscopy of the laser formed AA

6061 in three different tempers revealed only subtle changes to microstructure. A

possible precipitate coarsening was observed in the upper surface area after 5

passes in all of the heat treatments. These reverted back to the original

microstructure after 30 passes. This was attributed to the fact that, due to coating

loss, the amount of energy coupled into the surface reduced significantly after 10

passes. The small amount of heating present after this point may allow for a

refinement of the coarse precipitates formed earlier, akin to a post-forming heat

treatment.

d) Hardness tests on the laser formed AA 6061 samples revealed additional effects

on the metallurgy. For the O condition little effects on the hardness were

observed for increasing numbers of passes, the additional heating has little effect

on the already coarse microstructure. For the T4 and T6 tempers it was observed

that up to 10 passes there was a decrease in the average hardness within the

heated area. From 10 passes up to 30 passes there was a recovery somewhat in

the hardness values. This was consistent with the optical microscopy results,

since the possible precipitate coarsening observed after 5 passes would reduce

the hardness within the heated area, and the subsequent heat treatment effects of

the poorly coupled laser beam act to restore the original microstructure and hence

the hardness to some degree.

6.1.7 2D Closed Loop Control

In order to demonstrate that laser forming can be used to produce repeatable accurate

bends a system was presented for the closed loop controlled 2D laser forming of

80x80mm coupons of two materials, 1.5mm mild steel and 0.9mm AA1050 - H14.

a) The factors considered essential for control of the process were:

1) The current bend angle.

2) The difference between current and desired bend angle.

3) The current bend angle rate or bend angle increase per pass.

4) Selection of a bend angle rate per pass so as to avoid overshoot (when the

bend angle difference between current and desired angle is small, i.e.

bend angle rate should be less than or equal to the required deformation).



b) The bend angle rate per pass was controlled by the easily adjusted process speed

and feedback was given via a laser range finder coupled with control software.

c) Providing a selection of process parameters could be produced that give the

largest possible range of bend angle rates for the speed range of the CNC tables,

closed loop control can be setup for the laser forming of any metallic material

using this method.

d) The controlled laser forming of mild steel and AA 1050 to a number of preset

bend angles was successfully demonstrated

e) The only limit on the accuracy of the system was the resolution of the sensor

used. The higher the resolution of sensor for feedback the more control over the

process there is.

6.1.8 Thick Section and Large Area 2D Forming for Ship Building

A study was conducted on thick section 2D laser forming of mild steel in order to

investigate the factors influencing a scaling of known scan strategies for thinner

section materials, in particular for application in the ship building industry. The

conclusions from this study are:

a) The forming of thick section large area materials has been demonstrated through

the production of part-cylinders from 800x400x5mm thick mild steel on a

number of laser systems.

b) A double pass strategy has been employed throughout to improve the amount of

forming possible with limited laser power.

c) 800mm long bends were successfully produced in this study.

d) Measurement of the formed surfaces revealed a high degree of uniformity for the

size of the component. Little or no effect of the additional weight of the plate was

observed. This is extremely promising for application in the ship building

industry.

e) Thermocouple analysis confirmed the double pass strategy and emphasised the

localised effect of the laser forming process.



6.1.9 Laser Forming of Metal Laminate Composite Materials

An investigation was conducted to demonstrate how the laser forming process can be

used to form recently developed high strength metal laminate composite materials

(MLC) or fibre metal laminates (FML). These materials due to their construction and

high strength are difficult to form once manufactured using conventional techniques.

The conclusions made from this investigation were:

a) It has been shown that it is possible to laser form Fibre Metal Laminate materials

without damage to the material or structure. The process is realised by laser

forming by the TGM the upper aluminium layer alone.

b) The results have shown that the effectiveness of laser forming to produce sharp

single bends in these materials decreases with increasing number of layers.

However, it was shown that there is sufficient available distortion per scan line

even in 4/3 lay-ups for multiple scan line large radii bends and even the

capability to use the process to align and remove distortion post-conventional

forming.

c) It has been shown that the 2/1 lay-up shows the best potential for the use of laser

forming as a direct manufacturing tool. As it is a requirement that a metal layer

needs to be within the material to conventionally form the material successfully

(i.e. a minimum of a 3/2 laminate), laser forming offers a useful tool to produce

bends in 2/1 FML materials.

d) An insight into the effect of material anisotropy on the laser forming process was

also presented. This effect could be used to improve the formability of a material

in a particular orientation.

e) It was also shown that the technique could be used to form thermosetting

GLARE type materials. A large part-cylinder was formed using a series of small

bends. An improvement to the laser formability of the material was made by

increasing the thickness of the metal layers, so as to increase the moment

generated by laser forming the upper layer alone.



6.1.10 Application Example – Aero Engine Strut


made to replicate an actual aerospace component. An ‘A’ frame strut component

from a Rolls-Royce Trent 700 Aero engine was identified as an ideal candidate for

laser forming. The conclusions from this work are:

a) An initial attempt to reproduce the strut demonstrated that it may be possible to

form the two halves in one. However, it became difficult to close the gap up as

access to the inner surface became limited. It may be possible to employ the

buckling mechanism on the outer surface for future attempts.

b) An accurate cross-section of the strut was produced in 1.6mm Ti6Al4V sheet

demonstrating that the geometry of this aerospace component can be laser

formed.

c) A full sized accurate laser formed prototype of the strut halve from

574x175x3.2mm mild steel CR4 sheet was also produced. This component was

produce in an industrial environment and demonstrates the manufacturing

capability of the laser forming process to produce real part, whether for actual

use or prototype evaluation. Laser forming also offers the capability to alter the

dimensions of the component easily (CAD enabled) without the need to produce

another die or former, this is a major advantage of the process over conventional

forming technologies.

6.1.11 3D Laser Forming Empirical Study

An investigation was conducted into the 3D laser forming of the primitive shapes,

the saddle, the pillow and the twisted shape using an empirical approach to determine

the scan strategies. Also investigated was the use of 3D laser forming on thick

sections, specifically for the ship building industry.

a) The results of these investigations showed that the problem of 3D laser forming

is extremely complex.

b) For the study on the saddle shape it was found that it was possible to produce a

saddle shape from rectangular sheet Mild Steel CR4 using a concentric ‘race

track’ strategy. This strategy was also found to work in 1.6mm Ti64 and in



square length to width ratio sheets. Another successful strategy was presented

based on a cross-hatch pattern and an incremental route. This demonstrates that

there may be multiple solutions to any 3D laser forming problem.

c) For the pillow shape a concentric rectangular forming strategy was developed for

TGM conditions using rectangles of the same length to width ratio as the sheet to

be formed. A limitation to the amount of symmetrical forming possible with this

strategy was found.

d) For the Twisted shape a strategy of the production of a combination of a twisted

shape and a part-cylinder was developed. The part-cylinder was then un-formed

out of the shape by processing the reverse side of the plate to leave the desired

twisted shape in the sheet.

e) The ‘race track’ strategy for the saddle shape was found to scale up to larger

thicker materials to some degree, however, it was concluded that more forming

lines were required to account for the increased surface area.

f) Any pre-stressing of a work piece was considered a large factor in the magnitude

of forming and any distortion of the final part.

g) Symmetrical laser forming is hindered due to the asymmetric nature of the laser

forming process itself, in that it is not possible to form the whole plate at once. A

solution to this may be scanning optics.

h) Due to material and process variability development of an online monitoring

system with predictive distortion correction capabilities is a requirement if any

3D laser forming operation is required to be used reliably in a manufacturing

environment.

6.1.12 Development of a Geometry based Model for 3D Laser


It was realised from the empirical study that in order to develop control of the

process of 3D laser forming it was necessary to have the ability to define the surface

to be formed. In addition by defining the surface and analysing properties such as

gradient and curvature, it was thought this may lead to a method of scan strategy

prediction. To this aim, a method of surface creation and analysis was devised using

Matlab. The conclusions from this work are given here:



a) A surface definition method using a Bezier surface patch technique proved

crucial in the development of a predictive model for 3D LF.

b) Attributes such as contour lines of constant surface gradient and resultant surface

gradient vector were investigated as possible scan prediction routes. It was

discovered for the pillow shape that by forming orthogonal to the resultant

gradient vector a successful scan strategy was produced. This also corresponded

to the contour lines of constant height of the desired surface.

c) The required amount of forming and hence the localised energy input

requirement was found to vary across the sheet dependent on location. This

observation was based on the resultant gradient vector magnitude across the

sheet. The energy input requirement was found to vary not only per contour line

but within each contour line as well.

d) Application of the model to the saddle shape revealed that forming was required

on the reverse side of the plate to take account of the positive and negative

curvature of the surface. The ability to isolate the contour lines for forming the

reverse side of the plate was not available to test this prediction. It was

encouraging to note however, that a similar scan strategy was developed in the

empirical study that gave a promising result.

e) By considering the concept of developable and non-developable surfaces another

method of energy distribution over a surface was proposed. For a singly curved

developable surface the TGM should be the dominant mechanism used to

produce plastic bending strains and out of plane deformation. For a doubly

curved non-developable surface, material needs to be removed (in-plane) in order

to allow the deformation to take place. This suggests that the shortening

mechanism should be the dominant mechanism when forming this type of

surface, the in-plane plastic shrinkage accounting for the limiting material near

the edges. From the analysis of thin plates with small deflections it was found

that the strain component within a sheet can be expressed in terms of the

deflection. It was found that the in-plane strain component is the largest factor in

the calculation of the total strain requirement to form a given non-developable

surface. A limitation with this finding was that it is not realistically possible to

get exclusively in-plane strains without some bending strains using a laser

forming method. In addition it was argued that a significant amount of in-plane

strain is present anyway in the large beam TGM processing conditions used to



date. As the mechanisms could not be separated effectively, a compromise was

proposed by forming along lines orthogonal to the principle gradient i.e. contours

of zero gradient or constant height. These are the only paths that are acceptable

for the development of bending strains and in-plane strains at the same time. The

energy distribution over the plate could then be given by the sum of bending and

in-plane strains resolved in the direction of the principle gradient.

6.1.13 3D Laser Forming Demonstrator System

In order to demonstrate the manufacturing capabilities of the 3D laser forming

process, one of the final goals of this research was the production of a 3D LF

demonstrator system for the controlled LF of one of the three primitive shapes from

a 400x200x1.5mm mild steel sheet. The conclusions from this work were:

a) A demonstrator system based around the 3D LF of the pillow shape was

presented. Improvements to the Matlab code allowed an automated production of

the CNC code to describe a predicted scan strategy. As the Matlab code could

currently only produce CNC code for a positive forming direction the pillow

shape was the only useable shape. The ability to vary the scan speed per scan line

and within each scan line was included. This allowed the implementation of a

variable energy distribution realised on the plate surface based on either the

gradient vector magnitude or total strain requirements

b) The desired pillow surface was defined in terms of the mathematical equation for

an elliptic paraboloid. The Bezier surface route, although extremely flexible,

does not guarantee the defined surface passes through the specified control points

due to constraints on the smoothness of the surface produced. As a demonstration

of the potential accuracy of the process was the intention of the system, the more

accurate surface definition was used.

c) Using an incremental approach based on the error between the current and

desired surfaces it was possible to produce a component to within +/- 2.5mm of

the target shape.

d) Providing over-forming has not occurred on the first pass it is possible to iterate

towards the final shape increasing the traverse speed to reduce the bend angle

rate and calibrating for the current plate’s forming characteristics. This is a much



faster route than a single pass implementation by calculation of the required

strain field. It has the potential to produce a final component independent of

residual stress history and material non-uniformity and take account of unwanted

distortion, perhaps brought about by these two factors or process variability.

e) The energy distribution based on the sum of the bending and in-plane strains

resolved in the direction of the principle gradient was shown to be of merit.

However, the differences between this method and the gradient vector magnitude

distribution were subtle as the possible speed range is limited to between the

manually selected minimum speed and the maximum speed where no forming

occurs.

f) A number of limitations of the demonstrator system were identified. Firstly it

was currently possible to overshoot the target shape by a small degree. In

addition, as the Matlab code can currently only produce the CNC data for the

upper surface, the red negative bending requirements on the lower surface are

ignored. This can lead to additional problems as the speed is then scaled from the

next blue or positive bending requirement and further over-forming can therefore

occur. No account is taken of the influence on the rest of the plate of each

forming line since the forming lines at the centre of the plate will cause a

deflection of the outer edges and so the amount of forming required near the

edges should be reduced.

g) The system presented did demonstrate the potential of the laser forming process

to produce accurate repeatable 3D surfaces in a controlled way. This suggests

that laser forming could be utilised as a direct manufacturing tool or as a means

of distortion removal in an industrial environment. Providing the desired and the

current surfaces can be realised in a virtual way, a scan strategy can be predicted

to give the final shape.

h) The system as it stands should be ideally suited to the laser forming of

developable surfaces such as the part-cylinder and possibly the twisted shape.

Further development is necessary to the code for non-developable surfaces with

the inclusion of more in-plane strain to account for the additional limiting

material in these surface types.



6.2 Future Work

A number of recommendations for further research have arisen from the work in this

thesis, these are:

1. The use of 2D laser forming closed loop control for the forming of other

materials – In order to demonstrate the robustness of the control method use on

mild steel and pure aluminium in this thesis, it would be beneficial to confirm

this success on other materials such as Ti6Al4V and FMLs. Providing a range of

bend angle rates per pass can be selected via the process speed for a given

material then the system developed here can be used for controlled 2D laser

forming.

2. Closed loop manufacture of a complete component using 2D LF – As a

follow on from the previous recommendation the closed loop control system

could be further developed for the closed loop manufacture of an actual

component such as the aerospace strut section presented in this thesis. This

would truly demonstrate the manufacturing capability of the process and would

certainly interest manufacturing industries.

3. Investigate the use of laser wavelengths that require no absorptive coatings

– The variability of absorptive coatings used in the work in this thesis

demonstrates that they should not be used if possible. In addition the application

and removal of the coatings constitutes additional process steps in an industrial

process and an environmental hazard, by not using them an improvement in

efficiency can be achieved. Research is recommended into the use of shorter

laser wavelengths that do not require absorptive coatings in order to improve the

industrial viability of the process.

4. Investigate the use of a variable scan speed for 2D laser forming – A concern

that was raised in the work in this thesis was the considerable asymmetry of the

LF process even during a simple straight line scan strategy. Differences in the

temporal thermal, stress, strain and displacement characteristics were observed

along a straight scan line. These differences are thought to be responsible for the

edge effect phenomenon observed in laser formed samples. In order to reduce

this asymmetry a variable speed strategy has been suggested to even out the



thermal input along the scan line. Little or no research has been performed,

however, to ascertain the ideal speed distribution along a line to negate the edge

effects.

5. Use of scanning optics to remove the asymmetric nature of the process – As

a follow on from the previous recommendation the use of scanning optics has the

potential to realise a scan strategy in a rapid segmented fashion, offering the

ability to evenly distribute the incident energy rather than a single point source.

This has great potential for 3D laser forming process, particularly for large area

forming were the temporal effects of using a single point source are magnified.

6. Development of faster 3D surface measurement techniques – Key to the

improvement of process efficiency is the development of faster surface

measurement techniques. Single snap shot methods are available, such as fringe

project, to give instantaneous surface profiles. The ability to measure a large

surface mid-process would also be invaluable for the further understanding of LF

and potentially dynamic control could be introduced.

7. 3D laser forming of non-symmetrical surfaces – In order to acertain the limits

and robustness of the Matlab based scan strategy prediction method the laser

forming of non-symmetrical surfaces is recommended e.g a saddle at one end

and a pillow at the other.

8. 3D laser forming of a real component – Similarly to the aero engine strut

section presented in this thesis the controlled 3D laser forming of an actual

component would generate considerable interest in the manufacturing industries.

9. Removal of Distortion – There is considerable interest in the removal of

unwanted distortion from processed components. Processes such as welding,

chemical etching and mechanical forming can produce unwanted distortion in a

component. Research is recommended into the use of laser forming to correct

this distortion. The error based prediction method employed in this research has a

great deal of potential for application in this field.

10. 3D laser forming commercial system development – there is sufficient interest

in laser forming by manufacturing industries as a manufacturing tool. However,

to date there is no commercially available system for closed loop controlled 3D

laser forming. Due to this, production of such a system (or even a part of) would

be lucrative.

References


References

1. J. Magee, K.G. Watkins & W.M. Steen, ‘Advances in Laser Forming’,

Journal of Laser Applications, 10, 6, 1998.

2. K. Wegener & M. Adelhardt, ‘Shipbuilding experiences a revolution’,

Industrial Laser Solutions, Volume 17, Number 12, pp.9-12, Pennwell

Corporation, Tulsa, December 2002 (ISSN 1523-4266)

3. M. Geiger, P. Hoffman, J. Hutfless. ‘Laser Material Processing in Synergy to

Metal Forming’, Production Engineering (Annals of the German Academic

Society for Production Engineering), v.1, p.37, 1993.

4. K. Scully. “Laser Line Heating”, Journal of Ship Production, v.3/4, p.237,

1987.

5. K. Masubuchi, I. Maumoto. “Study on Metal Working by High Power

Lasers”, Proceedings of the 2nd International Conference on BEAM

Technology, v.2, p. 20, 1985.

6. T. Hashimoto, Y. Fujshiro. “An experiment of Line Heating Designed with

the Table Orthogonal Array L32 (281)”, Journal of The Society of Naval

Architects of Japan, v.104, p.201, 1959.

7. M. Araki, N. Inoue, M. Horioka, M. Ando. “On Angular Distortion of Hull

Steel Plates by Line Heating Methods”, Journal of the Society of Naval

Architects of Japan, v.133, p.343, 1973.

8. A. Moshaiov, W. Vorus. The Mechanics of the Flame Bending Process,

Theory and Applications, Journal of Ship Research, Vol. 31 No. 4: December,

1987: pp. 269-281.

9. R.L. Rothman. “Flame Straightening Quenched-and-Tempered Steels in Ship

Construction”, SSC247, Ship Structure Committee, 1974.

10. F. Vollertson. “Mechanisms and Models for Laser Forming”, Proceedings of

Laser Assisted Net shape Engineering Conference (LANE 94), pp. 345-359,

1994.

11. F. Vollertsen. “Forming, Sintering and Rapid Prototyping.” Handbook of the

Eurolaser Academy, Vol. 2. Schuöcker, D (Editor), Chapman & Hall, 1998:

pp. 357-453.

References


12. A.T. Male, Y.W. Chen, C. Pan, Y.M. Zhang. “Rapid prototyping of sheet

metal components by plasma-jet forming” Journal of Materials Processing

Technology 135 (2003) pp. 340–346.

13. W. Hoving, ‘Accurate Manipulation Using Laser Technology’, Proc. of

SPIE, 3097, pp.284-295, 1997.

14. W. Maher, K. Tong, C. Bampton, M. Bright, J. Wooten, C. Rhodes. Laser

Forming of Titanium and Other Metals is Useable Within Metallurgical

Constraints. Proceedings of ICALEO’98, Section E 1998; pp. 121-130.

15. J. Shackel, J. Sidhu, P.B. Prangnell, “The Metallurgical Implications of Laser

Forming Ti-6AL-4V Sheet.” Proceedings of ICALEO’2001, Section D, 2001

16. J. Magee, K.G. Watkins, W.M. Steen. “Laser Bending of High Strength

Alloys.” Journal of Laser Applications, Vol.10 No. 4: 1998; pp. 149-155.

17. J. Magee, K.G. Watkins, W.M. Steen, R.L. Cooke, J. Sidhu. “Development

of an Integrated Laser Forming Demonstrator System for the Aerospace

Industry.” Proceedings of ICALEO’98, Section E, 1998; pp. 141-150.

18. S.P. Edwardson, K.G. Watkins, G. Dearden, J. Magee. “Generation of 3D

Shapes Using a Laser Forming Technique.” Proceedings of ICALEO’2001,

Section D, 2001.

19. K.G. Watkins, S.P. Edwardson, J. Magee, G. Dearden, P. French, R.L. Cooke,

J. Sidhu, N. Calder, “Laser Forming of Aerospace Alloys.” Proceedings of

the SAE Aerospace Manufacturing Technology Conference, 2001, Aerospace

Congress, Seattle, 2001: Paper No. 2001-01-2610

20. A.C. Tam, C.C. Poon & L. Crawforth, ‘Laser Bending of Ceramics and

Application to Manufacture Magnetic Head Sliders in Disk Drives’,

Analytical Sciences, Vol.17 (Special Issue), pp.419-421, The Japan Society

for Analytical Chemistry, April 2001.

21. R. E. Holt, “Flame Straightening Basics,” Welding Engineer, 49-53 (June

1960).

22. G. Dearden, S. P. Edwardson ‘Laser Assisted Forming for Shipbuilding’

Proceeding of the SAIL Conference, 3rd - 5th June 2003, Williamsburg,

Virginia.

23. F. Vollertsen, “Laser forming, Mechanisms, Models, Applications” in LFT

Erlangen monograph (1995).

References


24. F. Vollertsen, “An analytical model for laser bending,” Lasers Eng. 2, 261-

276 (1994).

25. F. Vollertsen, M. Rodle, “Model for the Temperature Gradient Mechanism of

laser Bending,” in Laser Assisted Net Shape Engineering, Proceedings of the

LANE’94, edited by M Geiger and F. Vollertsen, (Meisenbach Bambërg,

Germany, 1994), Vol. I, pp. 371 - 378.

26. Z. Mucha, J. Hoffman, W. Kalita, and S. Mucha, “Laser Forming of Thick

Free Plates”. Laser Assisted Net shape Engineering 2. Proceedings of the

LANE’97, edited by M. Geiger and F. Vollertsen, (Meisenbach Bamberg,

Germany, 1997), Vol. 2, pp. 383-392.

27. M. Geiger, F. Vollertsen, “The Mechanisms of Laser Forming,” CIRP

ANNALS Vol. 42, 1, 301 - 304 (1993).

28. C. L. Yau, K. C Chan, W. B. Lee, “A New Analytical Model for Laser

Bending,” in Laser Assisted Net Shape Engineering 2, Proceedings of the


Ger many, 1997), Vol. 2, pp. 357-366.

29. J. Magee. “Laser Forming of Aerospace Alloys.” PhD Thesis, University of

Liverpool, 1999.

30. F.Vollertsen, I. Komel, and R. Kals, “The laser bending of steel foils for

microparts by the buckling mechanism - A model,” Model. Simul. Mater. Sci.

Eng. 3, 107 - 119 (1995).

31. W. W. Duley, Laser Processing and Analysis of Materials. Plenum Press

(1983).

32. F. Vollertsen, I. Komel, and R. Kals, “The laser bending of steel foils for

microparts by the buckling mechanism - A model,” Model. Simul. Mater. Sci.

Eng. 3, 107 - 119 (1995).

33. J. Kraus, “Basic processes in laser bending of extrusions Using the Upset ting

Mechanism,” in Laser Assisted Net Shape Engineering 2, Proceedings of the


Germany .1997) Vol. 2, pp.43l

34. A. Moshaiov, J. G. Shin, “Modified Strip Model for Analysing the Line

Heating Method (Part 2): Thermo-Elastic-Plastic Plates,” Journal of Ship Re

search, Vol. 35, No. 3, 266-275 (Sept. 1991).

References


35. Y. Iwamura, E. F. Rybicki, “A transient Elastic -Plastic thermal Stress

analysis of flame forming,” Transactions of ASME, Journal of Engineering

for Industry, 163-171 (February 1973).

36. T. Suhara, “Study on Thermo Plastic working: Bending of Beam of

Rectangular Cross Section,” Journal of Zosen Kyokai, Vol. 103, 233-243

(1958).

37. M. Araki, N. Inoue, M. Horioka, M. Ando, “On Angular Distortion of Hull

Steel Plates by Line Heating Methods,” Journal of the Society of Naval

Architects of Japan, Vol. 133, 343-348 (1973).

38. F. Vollertsen, M. Geiger, W. M. Li, “FDM- and FEM- simulation of laser

forming: a comparative study,” Advanced Technology of Plasticity, edited by

Z. R. Wang, Y. He, III, 1793-1798 (1993).

39. M. Geiger, S. Holzer, F. Vollertsen, “Laserstrahlbiegen - Simulation eines 3-

dimensionalen, thermomechanischen Prozesses,” in Metal Forming Process

Simulation in Industry, edited by B. Kropli and, E. Luckey,

Monchengladbach, Germany, 1994, pp. 335 - 352.

40. N. Alberti, L. Fratini, F. Micari, “Numerical simulation of the laser bending

process by a coupled thermal mechanical analysis,” in Laser Assisted Net

Shape Engineering, Proceedings of the LANE’94, edited by M. Geiger and F.

Vollertsen (Me Germany, 1994); Vol. I, pp.: 327 - 336.

41. N. Alberti, L. Fratini, F. Micari, M. Cantello, G. Savant, “Computer Aided

Engineering of a laser assisted bending processes,” in Laser Assisted

NetShape Engineering 2, Proceedings of the LANE’97, edited by M. Geiger

and F. Vollertsen, (Meisenbach Bamberg, Germany, 1997), Vol. 2, pp.375-

382

42. Y-C. Hsiao, H. Shimizu, L. Firth, W. Maher, K. Masabuchi, “Finite Element

Modelling Of Laser Forming,” in Proceedings of the International Congress

On Applications of Lasers and Electro-Optics, (ICALEO97), (San Diego,

U.S.A, 1997), Section A. pp.31-40.

43. H. Holzer, M. Arnet, M. Geiger, “Physical and Numerical modelling of the

Buckling Mechanism,” in Laser Assisted Net Shape Engineering,

Proceedings of the LANE’94, edited by M. Geiger and F. Vollertsen,

( Bamberg, Germany, 1994), Vol. I, pp. 379 - 386.

References


44. G. Yu, K. Masubuchi, T. Maekawa, N. M. Patrikalakis, “Thermomechanics

of laser forming of metal plates” Massachusetts Institute of Technology (MIT)

Fabrication Memorandum 99-1. November 1, 1999.

45. W. Li, Y. L. Yao, “ Effects of Strain Rate in Laser Forming” Proceedings of

ICALEO’99, San Diego CA, Section F, 1998: pp. 107-116

46. W. Li, Y. L. Yao, “Laser Forming with Constant Line Energy” International

Journal of Advanced Manufacturing Technology, Vol. 17 (2001): pp 196-203

47. J. Bao & Y.L. Yao, ‘Analysis and Prediction of Edge Effects in Laser

Bending’, Journal of Manufacturing Science & Engineering, Vol. 123,

ASME, pp.53-61, 2001

48. K. O. Tong, “Analyses on Laser Forming of 0.063” Thick Titainium Sheets”

Proceedings of ICALEO’99, San Diego CA, Section F, 1998: pp. 97-106

49. P. J. Cheng, S. C. Lin, “An Analytical Model for the Temperature Field in

Laser Forming of Sheet Metal” Journal of Materials Processing Technology,

vol. 101 (2000): pp 260-267

50. W. Li, Y. L. Yao, “Numerical and Experimental Investigation of Laser

Induced Tube Bending” Proceedings of ICALEO 2000, Detriot, Section D,

2000: pp. 53-62

51. Z. Hu, M. Labudovic, H. Wang, R. Kovacevic, “Computer Simulation and

Experimental Investigation of Sheet Metal Bending using Laser Beam

Scanning” International Journal of Machine Tools & Manufacture vol. 41

(2001): pp 589-607

52. C. Cosenza, L. Fratini, F. Micari, M. Cantello, M. Penasa, “Explicit Thermo-

Mechanical Analysis of Laser Forming Processes” Laser Assisted Net Shape

Engineering 3, Proceedings of the LANE’2001 Erlangen, Germany, 2001

53. Z. Hu, R. Kovacevic, M. Labudovic, “Experimental and Numerical Modeling

of Buckling Instability of Laser Sheet Forming” International Journal of

Machine Tools & Manufacture vol. 42 (2002): pp 1427-1439

54. W. Shichun, J. Zhong, “FEM Simulation of the Deformation Field during the

Laser Forming of Sheet Metal” Journal of Materials Processing Technology,

vol. 121 (2002): pp 269-275

55. K. C. Lee, J. Lin, “Transient Deformation of Thin Metal Sheets During

Pulsed Laser Forming” Optics & Laser Technology, vol. 34 (2002): pp 639-

648.

References


56. Y. Namba, “Laser Forming in Space”. Proceedings of the International Con

ference on Lasers’85, edited by C. P. Wang, (STS Press, McLean 1986),

pp.403-407.

57. K. Masabuchi. “Studies at M.I.T related to Applications of Laser

Technologies to Metal Fabrication,” in Proceedings of Laser Advanced

Materials Processing (LAMP’92) (Niigata, Japan, 1992), pp. 939 - 946.

58. A. Sprenger, F. Vollertsen, W. M. Steen, and K. G. Watkins, “Influence of

strain hardening on laser bending” Manuf Syst. 24, 215-221 (1995).

59. J. Magee, K. G. Watkins, W. M. Steen, N. Calder, J. Sidhu, J. Kirby, “Laser

Forming of Aerospace Alloys,” Proceedings of ICALEO 97, Vol. 83, Part 2,

section e-156 (1997), pp.156-165.

60. J. Magee, K. G. Watkins, W. M. Steen, N. Calder, J. Sidhu, and J. Kirby,

“Edge effects in laser forming,” in Laser Assisted Net shape Engineering 2.

Proceedings of the LANE’97, edited by M. Geiger and F. Vollertsen,

(Meisenbach Bamberg, Germany, 1997), Vol. 2, pp.399-408.

61. J. Ramos, J. Magee, K.G. Watkins, W.M. Steen, F. Noble. “Microstructure of

laser bent aluminium alloy Alclad 2024-T3.” Proceedings of ICALEO’98,

Section E, 1997: pp. 178-185.

62. J. Ramos, J. Magee, K.G. Watkins. “Microstructure and Microhardness

Study of Laser Bent Al-2024-T3”

63. J. Magee, K.G. Watkins, T. Hennige. “Symmetrical Laser Forming.”

Proceedings of ICALEO’99, San Diego CA, Section F, 1998: pp. 77-86.

64. S.P. Edwardson. “Laser Forming Dish Shapes – A 3D Case Study.”

M.Sc.(Eng.) Thesis, The University of Liverpool, 1999.

65. G. Dearden, S. P. Edwardson "Some Recent Developments in Two- and

Three-Dimensional Laser Forming for ‘Macro’ and ‘Micro’ Applications"

Journal of Optics A: Pure and Applied Optics Vol. 5 No. 4: July 2003; pp.

S8-S15

66. K. C. Chan, C. L. Yau, W. B. Lee, “Laser Bending of Thin Stainless Steel

Sheets” Journal of Laser Applications, vol. 12, No. 6, 2000

67. Z. Mucha, M. Cabaj, R. Gradoń, M. Pawlowski, J. Widlaszewski, “Laser

Forming of Plates by the use of Beam with Circular and Rectangular Cross-

section” Laser Assisted Net Shape Engineering 3, Proceedings of the

LANE’2001 Erlangen, Germany, 2001

References


68. X. F. Wang, J. Takacs, G. Krallics, A. Szilagyi, T Markovits, “ Research on

the Thermo-Physical Process of Laser Bending” Journal of Materials

Processing Technology, vol. 127 (2002): pp 338-391

69. L. Liqun, C. Yanbin, L. Shangyang, G. Dayong, “Characterization of Laser

Bending under Different Cooling Conditions” Proceedings of ICALEO’2002,

Scottsdale, Arizona, 2002.

70. L. Lin, Y. Chen, S. Lin, “ The Characteristics of Laser Forming of

Aluminium Sheet Under Consecutive Irradiations” Proceedings of

ICALEO’2003, Paper 109, Jacksonville, Florida, 2003

71. K.C. Chan, Y. Harada, J. Liang & F. Yoshida, ‘Deformation Behaviour of

Chromium Sheets in Mechanical and Laser Bending’, Journal of Materials

Processing Technology, Vol.122, pp.272-277, 2002.

72. N. Abe, N. Nakagawa, M. Tsukamoto, K. Nakacho, M. Sogabe, S. Miyake,

“Laser Forming of Thick Steel Plates with a High Power Diode Laser”

Proceedings of ICALEO’2001, Jacksonville FL, Section D, 2001

73. J. Magee, L.J. De Vin, ‘Process planning for laser-assisted forming’, Journal

of Materials Processing Technology, Vol.120 (2002), pp.322-326.

74. J. Magee & L.J. De Vin, ‘Combining Forming Tools and Laser Beams’, The

Industrial Laser User, Issue 21, pp.32-34, The UK Association of Industrial

Laser Users (AILU), November 2000.

75. G. Thompson & M. Pridham, ‘A Feedback Control System for Laser

Forming’, Mechatronics, Vol.7, No.5, pp.429-441, 1997.

76. M. Pridham, G. Thompson, “Laser forming: a force for the future,” Materials

World, Vol. 2, (Nov 1994).

77. M. Pridhain, G. Thompson, “Laser forming,” Manufacturing Engineer, p.24

(June 1995).

78. P. J. Cheng, S. C. Lin, “Using Neural Networks to Predict Bending Angle of

Sheet Metal” International Journal of Machine Tools & Manufacture vol. 40

(2000): pp 1185-1197

79. D. E. Peck, G. Jones, “Line Induced Thermal Forming” Proceedings of

ICALEO’2002, Scottsdale, Arizona, 2002.

80. P. O’Hara, “Lasershot Peening of Metals”, The Industrial Laser User, Issue

19, pp.27-29, The UK Association of Industrial Laser Users (AILU), May

2000.

References


81. J. Z. Zhou, J. C. Yang, Y. K. Zhang, M. Zhou, “A Study on Super-Speed

Forming of Metal Sheet by Laser Induced Shockwaves” Journal of Materials

Processing Technology, vol. 129 (2002): pp 241-244

82. W. Hoving, ‘Accurate Manipulation Using Laser Technology’, Proc. of the

3rd International Conference on Laser Assisted Net-shape Engineering

(LANE 2001), Erlangen, Germany, 28-31 August 2001, Eds. M. Geiger & A.

Otto, Meisenbach, Bamberg, Germany, pp.113-124, 2001.

83. B. Müller, “Laser Adjustment - Models for the Design of Actuators” Laser

Assisted Net Shape Engineering 3, Proceedings of the LANE’2001 Erlangen,

Germany, 2001

84. J. Widlaszewski, “Precise laser bending,” in Laser Assisted Net Shape Engi

neering 2, Proceedings of the LANE’97, edited by M. Geiger and F.

Vollertsen, (Meisenbach Bamberg, Germany, 1997), Vol. 2, pp.393

85. J. Widlaszewski, “Thermal Deformation of Multilevel Spatial Structures

Induced by Laser Pulses” Laser Assisted Net Shape Engineering 3,

Proceedings of the LANE’2001 Erlangen, Germany, 2001

86. E. Gaertner, J. Fruehauf & E. Jaensch, ‘Plastic reshaping of silicon

microstructures: process, characterisation and application’, Microsystems

Technologies, Vol.7, pp.155-160, 2001.

87. E. Gaertner, J. Fruehauf, U. Loeschner & H. Exner, ‘Laser bending of etched

silicon microstructures’, Microsystems Technologies, Vol.7, pp.23-26, 2001.

88. Y. Okamoto, Y. Uno, T. Shibata & K-I. Ohta, ‘Precision Laser Forming of

Plastic with YAG Laser’, Proc. 5th International Conference on Progress of

Machine Technology, pp.221-226, 2000.

89. G. Dearden, C. Taylor, K. Bartkowiak, S.P. Edwardson, K.G. Watkins, “An

experimental study of laser micro-forming using a pulsed Nd:YAG laser and

scanning optics” Proceedings of ICALEO’2003, Paper M409, Jacksonville,

Florida, 2003

90. T, Miyazaki, T. Misu, S. Yoshioka, T. Tokunaga, M. Saito, “Forming

Characteristics of Thin Metal Plate with Diode Laser Beam” Proceedings of


91. R. Oba, S. Yoshioka, T. Miyazaki, K. Nakamura, “ Laser Forming of Metal

Foil by the Upsetting Mechanism” Proceedings of ICALEO’2002, Scottsdale,

Arizona, 2002.

References


92. S. Yoshioka, T. Miyazki, T. Misu, R. Oba, M. Saito, “ Laser Forming of Thin

Foil by a Newly Developed Sample Holding Method” Journal of Laser

Applications, Vol. 15, No. 2, 2003

93. S. Silve, B. Podschies, W.M. Steen & K.G. Watkins, ‘Laser Forming – A

New Vocabulary for Objects’, Proceedings of the 18th Intl. Congress on

Applications of Lasers & Electro-Optics (ICALEO’99), San Diego, Vol.87,

Section F, pp.87-96, 1999.

94. T. Hennige, “Laser Forming of Spatially Curved Parts”, Laser Assisted Net

Shape Engi neering 2, Proceedings of the LANE’97, edited by M. Geiger and

F. Vollertsen, (Meisenbach Bamberg, Germany, 1997), Vol. 2

95. T. Hennige, “Development of irradiation strategies for 3D-laser forming”

Journal of Materials Processing Technology, Vol. 103 n. 1 (2000): pp 102-

108.

96. M. Otsu, M. Fujii, K. Osakada, “Three-Dimensional Laser Bending of Sheet

Metal” Advanced Technology of Plasticity Vol. II, Proceedings of the 6th

ICTP Sept. 19-24, 1999.

97. G. Casalino, A. D. Ludovico, A. Ancona, P. M. Lugará, “Stainless Steel 3D

Laser Forming for Rapid Prototyping” Proceedings of the 20th International

Congress on Applications of Lasers & Electro-Optics (ICALEO 2001),

Jacksonville, FL, October 2-5, 2001.

98. Vollertsen, F. and Holzer, S., “3D-Thermomechanical Simulation of Laser

Forming”, Simulation of Materials Processing: Theory, Methods and

Applications, Balkema, Rotterdam, p.785, 1995.

99. M. Otsu, K. Osakada & M. Fujii, ‘Controlled Laser Forming of Sheet Metal

with Shape Measurement and Using Database’, Proc. Metal Forming 2000,

Balkerma, pp.433-438, 2000.

100. J. Cheng, Y. L. Yao, “Process Synthesis of Laser Forming by Genetic

Algorithm” Proceedings of the 20th International Congress on Applications of

Lasers & Electro-Optics (ICALEO 2001), Jacksonville, FL, October 2-5,

2001.

101. C. Liu, Y. L. Yao, V. Srinivasan, “Optimal Process Planning for Laser

Forming of Doubly Curved Shapes” Proceedings of ICALEO’2002,

Scottsdale, Arizona, 2002.

References


102. J. Kim, S. Na, “A study on the Laser Forming for Free Curve Generation”

Proceedings of ICALEO’2003, Paper 104, Jacksonville, Florida, 2003

103. R. P. Martukanitz, E. W. Reutzel, R. W. Erskine, S. Nasla, “ Adaptive and

Heuristic Techniques for Laser Assisted Forming of Plate” Proceedings of

ICALEO 2000, Detriot, Section D, 2000: pp. 47-52

104. E.W. Reutzel, R.P. Martukanitz, P. Michaleris, L. Zhang, A.J. Savitz, J.P.

Magnusen, J.U. Aburdene & K.J. Gombotz, ‘Development of a System for

the Laser Assisted Forming of Plate’, Proceedings of the 20th International


Jacksonville, FL, October 2-5, 2001.

105. E.W. Reutzel, J.U. Aburdene, K.J. Gombotz, , J. P. Magnusen, R.P.

Martukanitz, P. Michaleris, L. Zhang, “Continuing Development of a Laser

Line Forming System” Proceedings of ICALEO’2002, Scottsdale, Arizona,

2002.

106. E.W. Reutzel, K.J. Gombotz, R.P. Martukanitz, P Michaleris, “Path

Planning Strategies for Laser Line Forming” Proceedings of ICALEO’2003,

Paper 106, Jacksonville, Florida, 2003

107. J.E. Jones, V.L. Rhoades, J.C. Jones, A.C. Beck, P. Oberly, P. Sewell, D.D.

Schwemmer, D. Stompro & T. Whipple, ‘Use of the Flexible Laser

Automated Intelligent Real-Time (FLAIR) System for Laser Thermal

Forming’, Proceedings of the International Conference on the Industrial

Applications of Lasers in Manufacturing, Minneapolis, April 1998.

108. G. Thomson, M. Pridham, “Material Property Changes Associated with

Laser Forming of Mild Steel Components” Journal of Materials Processing

Technology, Vol. 118 (2001): pp 40-44.

109. M. Marya, G. R. Edwards, “A Study on the Laser Forming of Near-Alpha

and Meta-stable Beta Titanium Alloy Sheets” Journal of Materials Processing

Technology, Vol. 108 (2001): pp 376-383.

110. M. Marya & G.R. Edwards, ‘Factors Affecting the Laser Bending of Ti-

6Al-2Sn-4Zr-2Mo’, Journal of Laser Applications, Vol.12, Number 4, Laser

Institute of America, pp.149-159, 2000.

111. M. Merklein & M. Geiger, ‘A Comparative Study of Two Different Laser

Forming Mechanisms Regarding the Mechanical Properties of Aluminium

References


Alloys’ Laser Assisted Net Shape Engineering 3, Proceedings of the

LANE’2001 Erlangen, Germany, 2001

112. M. Merklein, T. Hennige, M. Geiger, “Laser Forming of Aluminium and

Aluminium Alloys – Microstructural Investigation” Journal of Materials

Processing Technology, Vol. 115 (2001): pp 159-165.

113. K.C. Chan & J. Liang, ‘Thermal Expansion and Deformation Behaviour of

Aluminium-Matrix Composites in Laser Forming’, Composites Science &

Technology, Vol.61, pp.1265-1270, 2001.

114. K.C. Chan, Y. Harada, J. Liang & F. Yoshida, ‘Deformation Behaviour of

Chromium Sheets in Mechanical and Laser Bending’, Journal of Materials

Processing Technology, Vol.122, pp.272-277, 2002.

115. J, Zhang, D. Pirzada, L. V. Smith, C. Chin, J. G. Cheng, “Fatigue Life

Prediction after Laser Forming” Proceedings of ICALEO’2003, Paper 105,

Jacksonville, Florida, 2003

116. J. Zhang, J. G. Cheng, “Experimental Study and Computer Simulation on

Fracture Toughness of Sheet Metal after Laser Forming” Proceedings of

ICALEO’2003, Paper 108, Jacksonville, Florida, 2003

117. J. A. Ramos, “Creep Behaviour of Thin Metal Sheets during Laser

Bending” Proceedings of the 20th International Congress on Applications of

Lasers & Electro-Optics (ICALEO 2001), Jacksonville, FL, October 2-5,

2001.

118. P. Cheng, Y. L. Yao, “The Influence of Sheet Metal Anisotropy on Laser

Forming Process” Proceedings of ICALEO’2003, Paper 101, Jacksonville,

Florida, 2003

119. J.R. Dydo, H.R. Castner & K. Koppenhoefer, ‘Guidelines for Control of

Distortion in Thin Ship Structures’, EWI Project No. 42372-GDE, Edison

Welding Institute, Columbus, Ohio, October 1999.

120. W.M. Steen, “Laser Materials Processing”, 2nd Edition, Sprenger-Verlag, 1998

121. W. O’Neil, “Laser Cutting Lecture Notes” MSc (Eng) Advanced

Manufacturing with Lasers, The University of Liverpool, 1999.

122. ASM Metals Handbook, ASM International, 10th Ed. 1990

123. S. P. Edwardson, K. G. Watkins, G. Dearden, P. French, J. Magee “Strain

Gauge Analysis of Laser Forming”, Proceedings of the 21st International

References



Scottsdale, Arizona, October 14-17, 2002.

124. S. P. Edwardson, K. G. Watkins, G. Dearden, P. French, J. Magee "Strain

Gauge Analysis of Laser Forming" Journal of Laser Applications -

November 2003 Volume 15, Issue 4, pp. 225-232

125. S. P. Edwardson, G. Dearden, P. French, K. G. Watkins, W.J. Cantwell,

“Laser Forming of Metal Laminate Composite Materials”, Proceedings of the

22nd International Congress on Applications of Lasers & Electro-Optics

(ICALEO 2003), Jacksonville, Florida, October 13-16, 2003.

126. S. P. Edwardson, G. Dearden, K. G. Watkins, W. J. Cantwell, “A new

forming process for a new material” Laser Industrial Solutions, March 2004.

127. Vlot, A., Gunnink, J. W. Fibre Metal Laminates: An Introduction. Kluwer

Academic Publishers, 2001, ISBN 14020-0038-3.

128. S.P. Edwardson, K.G. Watkins & G. Dearden, “3-D Laser Forming of

Saddle Shapes” Proc. of the 3rd International Conference on Laser Assisted

Net-shape Engineering (LANE 2001), Erlangen, Germany, 28-31 August

2001, Eds. M. Geiger & A. Otto, Meisenbach, Bamberg, Germany, pp.559-

568, 2001, ISBN 3-87525-154-7.

129. M. Reeves, M. D. Stoikou, A. J. Moore, D. P. Hand, J. R. Cho, S. P.

Edwardson, K. G. Watkins, G. Dearden, P. French, J. D.C. Jones. "A system

for Dynamic Shape Measurements During Laser Processing" Proceedings of


130. M.E. Mortenson, “Geometric Modelling” Second Edition, Wiley, 1997.

131. K. Ueda, H. Murakawa, A.M. Rashwan, Y. Okumoto, R. Kamichika, 1994,

“Development of computer-aided process planning system for plate bending

by line heating (report 1) – relation between final form of plate and inherent

strain”, Journal of Ship Production, Vol.10 No.1, pp.59-67.

132. M.J. Mancewicz, and Frey, W.H., “Developable surfaces: properties,

representations and methods of design” General Motors R&D Publication

7637 (1992).

133. D. McFarland, B. L. Smith, W. D. Bernhart, “Analysis of plates” Spartan

Books, New York, 1972



Publications to Date by the Author

S. P. Edwardson, K. G. Watkins, G. Dearden, J. Magee

“3D Laser Forming of Saddle Shapes”

Laser Assisted Net Shape Engineering, Proceedings of the LANE’2001 Erlangen,

Germany, 2001

S. P. Edwardson, K. G. Watkins, G. Dearden, J. Magee

“Generation of 3D Shapes Using a Laser Forming Technique”

Proceedings of ICALEO’2001, Section D, Jacksonville, Florida, 2001.

K. G. Watkins, S. P. Edwardson, J. Magee, G. Dearden, P. French,

R. L. Cooke, J. Sidhu, N. Calder

“Laser Forming of Aerospace Alloys”

Proceedings of the SAE Aerospace Manufacturing Technology Conference, 2001,

Aerospace Congress, Seattle, 2001: Paper No. 2001-01-2610

S. P. Edwardson, K. G. Watkins, G. Dearden, P. French, J. Magee

"Strain Gauge Analysis of Laser Forming"

Proceedings of ICALEO’2002, Scottsdale, Arizona, 2002.

"Received 3rd Prize in the ICALEO 2002 Student Paper Award"

M. Reeves, M. D. Stoikou, A. J. Moore, D. P. Hand, J. R. Cho,

S. P. Edwardson, K. G. Watkins, G. Dearden, P. French, J. D.C. Jones.

"A system for Dynamic Shape Measurements During Laser Processing"

Proceedings of ICALEO’2002, Scottsdale, Arizona, 2002.

Stuart Edwardson

"High Powered Laser Forming of Metallic Components"

Society of Manufacturing Engineers, Dies and Stamping News Letter, April 2003

Online Publication: http://www.sme.org/dies&stamping/

G. Dearden, S. P. Edwardson

"Laser Assisted Forming for Shipbuilding"

Proceeding of the SAIL Conference, 3rd - 5th June 2003, Williamsburg, Virginia



G. Dearden, S. P. Edwardson

"Some Recent Developments in Two- and Three-Dimensional Laser Forming for

'Macro' and 'Micro' Applications"

Journal of Optics A: Pure and Applied Optics Vol. 5 No. 4: July 2003; pp. S8-S15

S. P. Edwardson, G. Dearden, P. French, K. G. Watkins, W. J. Cantwell

“Laser Forming of Metal Laminate Composite Materials”

Proceedings of ICALEO’2003, Paper 107, Jacksonville, Florida, 2003

G. Dearden, C. Taylor, K. Bartkowiak, S.P. Edwardson, K.G. Watkins

“An experimental study of laser micro-forming using a pulsed Nd:YAG laser and

scanning optics”

Proceedings of ICALEO’2003, Paper M409, Jacksonville, Florida, 2003

S. P. Edwardson, K. G. Watkins, G. Dearden, P. French, J. Magee

"Strain Gauge Analysis of Laser Forming"

Journal of Laser Applications - November 2003

Volume 15, Issue 4, pp. 225-232

M. Reeves, A. J. Moore, D. P. Hand, J. D.C. Jones, J. R. Cho, R. C. Reed,

S. P. Edwardson, G. Dearden, P. French, K. G. Watkins

"Dynamic Distortion Measurements during Laser Forming of Ti-6Al-4V and their

comparison with a finite element model"

Proceedings of the Institution of Mechanical Engineers, Part B: Journal of

Engineering Manufacture, December, v 217, n 12, 2003, pp 1685-1696

S. P. Edwardson, P. French, G. Dearden, K. G. Watkins, W. J. Cantwell

“Laser Forming of Fibre Metal Laminates”

Lasers in Engineering (Journal) Submitted January 2004

S. P. Edwardson, G. Dearden, K. G. Watkins, W.J. Cantwell

“Forming a New Material” Industrial Laser Solutions, March 2004

Magazine Article

Appendix


Appendix

A1. Matlab Code I A2. Abaqus Input Files IV A3. Beam Diameter Prediction X A4. Safety Interlocks & System Layout for the Electrox Workstation No. 2 XIII A5. MEL M5 & M1 Laser Range Finder Specifications XV A6. Example Galil CNC code XVIII

Appendix 1

Stuart P. Edwardson PhD Thesis I

Appendix 1 – Matlab Code Simple Code example in order to generate a pillow shape and analyse its

geometry. % Interpolate pillow from control points % Stuart Edwardson / Andrew Moore 9/12/02 % Comments in Green and % % Figure Titles in Red clear all; close all; aspect=2; num=30; num_contour=9; num_quiver=15; %Given data points [x,y]=meshgrid(0:aspect/3:aspect,0:1/3:1); z=-[ 0, 0.025, 0.025, 0; 0.042, 0.08, 0.08, 0.042; 0.042, 0.08, 0.08, 0.042; 0, 0.025, 0.025, 0; ]; [x1,y1]=meshgrid(0:aspect/num:aspect,0:1/num:1); z1=interp2(x,y,z,x1,y1,'cubic'); %* signifies x and y equally spaced %Trim image edges to avoid exaggerated errors in gradient and cutvature z1(1,:)=NaN; z1(end,:)=NaN; z1(:,1)=NaN; z1(:,end)=NaN; %Calculate grid spacing for ~num_quiver quivers per plot in x and y direction [ysize,xsize]=size(z1); quiverx=1:round(xsize/num_quiver):xsize; quivery=1:round(ysize/num_quiver):ysize; x1_max=max(x1(:)); x1_min=min(x1(:)); xscale=xsize/(x1_max-x1_min); y1_max=max(y1(:)); y1_min=min(y1(:)); yscale=ysize/(y1_max-y1_min); figure subplot(2,2,1) colormap(jet(256)); surf(x,y,z); title('16 Control Points'); subplot(2,2,2) colormap(jet(256)); surf(x1,y1,z1); title('Bezier Bicubic Surface Patch'); subplot(2,2,3) contour(x1,y1,z1,10);

Appendix 1

Stuart P. Edwardson PhD Thesis II

title('Contour Plot of Interpolated Surface'); figure colormap(jet(256)); surf(x1,y1,z1); colorbar title('Interpolated Pillow Surface'); figure [C,h]=contour(x1,y1,z1,10); clabel(C,h); title('Contour Plot of Interpolated Surface'); %Gradient: Use default spacing to get agreement with del2 [z1x,z1y]=gradient(z1); %Single pixel spacing in each dimension figure subplot(2,2,1); [C,h]=contour(x1,y1,z1x,10); clabel(C,h); title('Gradient in x (dz/dx)'); subplot(2,2,2); [C,h]=contour(x1,y1,z1y,10); clabel(C,h); title('Gradient in y (dz/dy)'); subplot(2,2,3); temp=sqrt(z1x.^2 + z1y.^2); [C,h]=contour(x1,y1,temp,num_contour); title('sqrt[(dz/dx)^2 + (dz/dy)^2]'); subplot(2,2,4); quiver(x1(1:2:end,1:2:end),y1(1:2:end,1:2:end),z1x(1:2:end,1:2:end),z1y(1:2:end,1:2:end),0.5) title('Quiver Plot of Gradients in x&y') figure; subplot(2,2,1); %Repeat last plot quiver(x1(1:2:end,1:2:end),y1(1:2:end,1:2:end),z1x(1:2:end,1:2:end),z1y(1:2:end,1:2:end),0.5) title('Quiver Plot of Gradients in x&y') subplot(2,2,2); %Rotate quiver by pi/2 quiver(x1(1:2:end,1:2:end),y1(1:2:end,1:2:end),-z1y(1:2:end,1:2:end),z1x(1:2:end,1:2:end),0.5) title('Quiver Vector Rotated by pi/2'); subplot(2,2,3); %Contours of constant angle theta=atan2(z1y,z1x); theta_pos=sqrt(theta.^2); contour(x1,y1,theta_pos,20); title('Contours of Constant Gradient Vector Angle'); subplot(2,2,4); %Contours of constant angle [C,h]=contour(x1,y1,z1,15); hold on quiver(x1(quivery,quiverx),y1(quivery,quiverx),-z1y(quivery,quiverx),z1x(quivery,quiverx),0.5) hold off

Appendix 1

Stuart P. Edwardson PhD Thesis III

title('Contour Plot of Pillow Surface'); figure hold on [C,h]=contour(x1,y1,z1,15); axis([x1_min,x1_max,y1_min,y1_max]); temp_max=max(temp(:)); temp_min=min(temp(:)); %temp contains sqrt((dz/dx)^2 + (dz/dy)^2) temprange=(temp_max-temp_min); cur_pos=1; %Position in contour matrix C (for structure see help for contourc) for contour=1:size(h,1) %Loop over each contour for point=1:C(2,cur_pos)-1 %Count through each point on contour xi=round( C(1,cur_pos+point)*xscale ); %x values in C(1,:) y in C(2,:) yi=round( C(2,cur_pos+point)*yscale ); %Convert to array pixels sizei=round( 20*( (temp(yi,xi)-temp_min)/temprange ) + 0.5 ); %Scale to size of quiver arrow %+0.5 to avoid 0 (else plot fails) if ~isnan(sizei) if z1(yi,xi)<0 %Scan on other side if deflection negative plot(C(1,cur_pos+point),C(2,cur_pos+point),'.','MarkerSize',sizei,'MarkerEdgeColor','r'); else plot(C(1,cur_pos+point),C(2,cur_pos+point),'.','MarkerSize',sizei,'MarkerEdgeColor','b'); end end end cur_pos=cur_pos+C(2,cur_pos)+1; %Position of next contour end title('Contour Plot of Z1 with Representation of Vector Magnitude Along Contour Line'); hold off

Appendix 2

Stuart P. Edwardson PhD Thesis IV

Appendix 2 – Abaqus Input File Example input file for forming6aa, thermal model only, graded mesh,

80x80x1.5mm Mild Steel. Written for Abaqus Version 5.8: Split into Columns

for display only, normally one continuous .inp text file *HEADING

forming6aa

lASER FORMING A STEEL PLATE

CLAMPED AT ONE END

L80mm, W80mm, D1.5mm

20 node 3D elements

combined fine and coarse mesh

580 elements

HEAT TRANSFER

*NODE

1,0,0,0

161,0,0.08,0

806,0.020,0,0

966,0.020,0.080,0

967,0.024,0,0

1127,0.024,0.080,0

1772,0.034,0,0

1932,0.034,0.080,0

1933,0.036,0,0

2093,0.036,0.080,0

3543,0.044,0,0

3703,0.044,0.080,0

3704,0.046,0,0

3864,0.046,0.080,0

4509,0.056,0,0

4669,0.056,0.080,0

4670,0.060,0,0

4830,0.060,0.080,0

5475,0.080,0,0

5635,0.080,0.080,0

*NGEN,NSET=A

1,161,8

*NGEN,NSET=B

806,966,8

*NGEN,NSET=C

1 ,806 ,161

9 ,814 ,161

17 ,822 ,161

25 ,830 ,161

33 ,838 ,161

41 ,846 ,161

49 ,854 ,161

57 ,862 ,161

65 ,870 ,161

73 ,878 ,161

81 ,886 ,161

89 ,894 ,161

97 ,902 ,161

105 ,910 ,161

113 ,918 ,161

121 ,926 ,161

129 ,934 ,161

137 ,942 ,161

145 ,950 ,161

153 ,958 ,161

161 ,966 ,161

*NGEN,NSET=D

967,1127,4

*NGEN,NSET=E

1772,1932,4

*NGEN,NSET=F

967 ,1772 ,161

971 ,1776 ,161

975 ,1780 ,161

979 ,1784 ,161

983 ,1788 ,161

987 ,1792 ,161

991 ,1796 ,161

995 ,1800 ,161

999 ,1804 ,161

1003 ,1808 ,161

1007 ,1812 ,161

1011 ,1816 ,161

1015 ,1820 ,161

1019 ,1824 ,161

1023 ,1828 ,161

1027 ,1832 ,161

1031 ,1836 ,161

1035 ,1840 ,161

1039 ,1844 ,161

1043 ,1848 ,161

1047 ,1852 ,161

1051 ,1856 ,161

1055 ,1860 ,161

1059 ,1864 ,161

1063 ,1868 ,161

1067 ,1872 ,161

1071 ,1876 ,161

1075 ,1880 ,161

1079 ,1884 ,161

1083 ,1888 ,161

1087 ,1892 ,161

1091 ,1896 ,161

1095 ,1900 ,161

1099 ,1904 ,161

1103 ,1908 ,161

1107 ,1912 ,161

1111 ,1916 ,161

1115 ,1920 ,161

1119 ,1924 ,161

1123 ,1928 ,161

1127 ,1932 ,161

*NGEN,NSET=G

1933,2093,1

*NGEN,NSET=H

3543,3703,1

*NGEN,NSET=I

1933 ,3543 ,161

1934 ,3544 ,161

1935 ,3545 ,161

1936 ,3546 ,161

1937 ,3547 ,161

1938 ,3548 ,161

1939 ,3549 ,161

1940 ,3550 ,161

1941 ,3551 ,161

1942 ,3552 ,161

1943 ,3553 ,161

Appendix 2

Stuart P. Edwardson PhD Thesis V

1944 ,3554 ,161

1945 ,3555 ,161

1946 ,3556 ,161

1947 ,3557 ,161

1948 ,3558 ,161

1949 ,3559 ,161

1950 ,3560 ,161

1951 ,3561 ,161

1952 ,3562 ,161

1953 ,3563 ,161

1954 ,3564 ,161

1955 ,3565 ,161

1956 ,3566 ,161

1957 ,3567 ,161

1958 ,3568 ,161

1959 ,3569 ,161

1960 ,3570 ,161

1961 ,3571 ,161

1962 ,3572 ,161

1963 ,3573 ,161

1964 ,3574 ,161

1965 ,3575 ,161

1966 ,3576 ,161

1967 ,3577 ,161

1968 ,3578 ,161

1969 ,3579 ,161

1970 ,3580 ,161

1971 ,3581 ,161

1972 ,3582 ,161

1973 ,3583 ,161

1974 ,3584 ,161

1975 ,3585 ,161

1976 ,3586 ,161

1977 ,3587 ,161

1978 ,3588 ,161

1979 ,3589 ,161

1980 ,3590 ,161

1981 ,3591 ,161

1982 ,3592 ,161

1983 ,3593 ,161

1984 ,3594 ,161

1985 ,3595 ,161

1986 ,3596 ,161

1987 ,3597 ,161

1988 ,3598 ,161

1989 ,3599 ,161

1990 ,3600 ,161

1991 ,3601 ,161

1992 ,3602 ,161

1993 ,3603 ,161

1994 ,3604 ,161

1995 ,3605 ,161

1996 ,3606 ,161

1997 ,3607 ,161

1998 ,3608 ,161

1999 ,3609 ,161

2000 ,3610 ,161

2001 ,3611 ,161

2002 ,3612 ,161

2003 ,3613 ,161

2004 ,3614 ,161

2005 ,3615 ,161

2006 ,3616 ,161

2007 ,3617 ,161

2008 ,3618 ,161

2009 ,3619 ,161

2010 ,3620 ,161

2011 ,3621 ,161

2012 ,3622 ,161

2013 ,3623 ,161

2014 ,3624 ,161

2015 ,3625 ,161

2016 ,3626 ,161

2017 ,3627 ,161

2018 ,3628 ,161

2019 ,3629 ,161

2020 ,3630 ,161

2021 ,3631 ,161

2022 ,3632 ,161

2023 ,3633 ,161

2024 ,3634 ,161

2025 ,3635 ,161

2026 ,3636 ,161

2027 ,3637 ,161

2028 ,3638 ,161

2029 ,3639 ,161

2030 ,3640 ,161

2031 ,3641 ,161

2032 ,3642 ,161

2033 ,3643 ,161

2034 ,3644 ,161

2035 ,3645 ,161

2036 ,3646 ,161

2037 ,3647 ,161

2038 ,3648 ,161

2039 ,3649 ,161

2040 ,3650 ,161

2041 ,3651 ,161

2042 ,3652 ,161

2043 ,3653 ,161

2044 ,3654 ,161

2045 ,3655 ,161

2046 ,3656 ,161

2047 ,3657 ,161

2048 ,3658 ,161

2049 ,3659 ,161

2050 ,3660 ,161

2051 ,3661 ,161

2052 ,3662 ,161

2053 ,3663 ,161

2054 ,3664 ,161

2055 ,3665 ,161

2056 ,3666 ,161

2057 ,3667 ,161

2058 ,3668 ,161

2059 ,3669 ,161

2060 ,3670 ,161

2061 ,3671 ,161

2062 ,3672 ,161

2063 ,3673 ,161

2064 ,3674 ,161

2065 ,3675 ,161

2066 ,3676 ,161

2067 ,3677 ,161

2068 ,3678 ,161

2069 ,3679 ,161

2070 ,3680 ,161

2071 ,3681 ,161

2072 ,3682 ,161

2073 ,3683 ,161

2074 ,3684 ,161

2075 ,3685 ,161

2076 ,3686 ,161

2077 ,3687 ,161

Appendix 2

Stuart P. Edwardson PhD Thesis VI

2078 ,3688 ,161

2079 ,3689 ,161

2080 ,3690 ,161

2081 ,3691 ,161

2082 ,3692 ,161

2083 ,3693 ,161

2084 ,3694 ,161

2085 ,3695 ,161

2086 ,3696 ,161

2087 ,3697 ,161

2088 ,3698 ,161

2089 ,3699 ,161

2090 ,3700 ,161

2091 ,3701 ,161

2092 ,3702 ,161

2093 ,3703 ,161

*NGEN,NSET=J

3704,3864,4

*NGEN,NSET=K

4509,4669,4

*NGEN,NSET=L

3704 ,4509 ,161

3708 ,4513 ,161

3712 ,4517 ,161

3716 ,4521 ,161

3720 ,4525 ,161

3724 ,4529 ,161

3728 ,4533 ,161

3732 ,4537 ,161

3736 ,4541 ,161

3740 ,4545 ,161

3744 ,4549 ,161

3748 ,4553 ,161

3752 ,4557 ,161

3756 ,4561 ,161

3760 ,4565 ,161

3764 ,4569 ,161

3768 ,4573 ,161

3772 ,4577 ,161

3776 ,4581 ,161

3780 ,4585 ,161

3784 ,4589 ,161

3788 ,4593 ,161

3792 ,4597 ,161

3796 ,4601 ,161

3800 ,4605 ,161

3804 ,4609 ,161

3808 ,4613 ,161

3812 ,4617 ,161

3816 ,4621 ,161

3820 ,4625 ,161

3824 ,4629 ,161

3828 ,4633 ,161

3832 ,4637 ,161

3836 ,4641 ,161

3840 ,4645 ,161

3844 ,4649 ,161

3848 ,4653 ,161

3852 ,4657 ,161

3856 ,4661 ,161

3860 ,4665 ,161

3864 ,4669 ,161

*NGEN,NSET=M

4670,4830,8

*NGEN,NSET=N

5475,5635,8

*NGEN,NSET=O

4670 ,5475 ,161

4678 ,5483 ,161

4686 ,5491 ,161

4694 ,5499 ,161

4702 ,5507 ,161

4710 ,5515 ,161

4718 ,5523 ,161

4726 ,5531 ,161

4734 ,5539 ,161

4742 ,5547 ,161

4750 ,5555 ,161

4758 ,5563 ,161

4766 ,5571 ,161

4774 ,5579 ,161

4782 ,5587 ,161

4790 ,5595 ,161

4798 ,5603 ,161

4806 ,5611 ,161

4814 ,5619 ,161

4822 ,5627 ,161

4830 ,5635 ,161

*NSET,NSET=BOT

A,B,C,D,E,F,G,H,I,J,K,L,M,N,O

*NCOPY,SHIFT,CHANGE NUMBER=10000,OLD SET=BOT,NEW

SET=MID

0.,0.,0.00075

0.,0.,0.,0.,0.,1.,0.

*NCOPY,SHIFT,CHANGE NUMBER=20000,OLD SET=BOT,NEW

SET=TOP

0.,0.,0.0015

0.,0.,0.,0.,0.,1.,0.

*NSET,NSET=ALL

BOT,MID,TOP

*NSET,NSET=ENDA,GENERATE

20001,20161,8

*NSET,NSET=END

ENDA,A

*ELEMENT, TYPE=DC3D20

1,1,323,339,17,20001,20323,20339,20017,162,331,

178,9,20162,20331,20178,20009,

10001,10323,10339,10017


31,967,1289,1297,975,20967,21289,21297,

20975,1128,1293,1136,971,21128,21293,

21136,20971,10967,11289,11297,10975


91,1933,2255,2257,1935,21933,22255,22257,21935,

2094,2256,2096,1934,22094,22256,

22096,21934,11933,12255,12257,11935


491,3543,3865,3873,3551,23543,23865,23873,23551,

3704,3869,3712,3547,23704,23869,23712,

23547,13543,13865,13873,13551


551,4509,4831,4847,4525,24509,24831,24847,24525,

4670,4839,4686,4517,24670,24839,24686,

24517,14509,14831,14847,14525

*ELGEN,ELSET=L1

1,10,16,1,3,322,10

*ELGEN,ELSET=L2

31,20,8,1,3,322,20

*ELGEN,ELSET=CENTRE

91,80,2,1,5,322,80

*ELGEN,ELSET=R2

491,20,8,1,3,322,20

*ELGEN,ELSET=R1

551,10,16,1,3,322,10

Appendix 2

Stuart P. Edwardson PhD Thesis VII

*ELSET,ELSET=PLATE

L1,L2,CENTRE,R1,R2

*ELSET, ELSET=END1, GENERATE

1,10,1

*ELSET, ELSET=END2, GENERATE

571,580,1

*ELSET, ELSET=SIDE1A, GENERATE

1,21,10

*ELSET, ELSET=SIDE1B, GENERATE

31,71,20

*ELSET, ELSET=SIDE1C, GENERATE

91,411,80

*ELSET, ELSET=SIDE1D, GENERATE

491,531,20

*ELSET, ELSET=SIDE1E, GENERATE

551,571,10

*ELSET,ELSET=SIDE1

SIDE1A,SIDE1B,SIDE1C,SIDE1D,SIDE1E

*ELSET, ELSET=SIDE2A, GENERATE

10,30,10

*ELSET, ELSET=SIDE2B, GENERATE

50,90,20

*ELSET, ELSET=SIDE2C, GENERATE

170,490,80

*ELSET, ELSET=SIDE2D, GENERATE

510,550,20

*ELSET, ELSET=SIDE2E, GENERATE

560,580,10

*ELSET,ELSET=SIDE2

SIDE2A,SIDE2B,SIDE2C,SIDE2D,SIDE2E

*ELSET, ELSET=LASER1

91 ,171 ,251 ,331 ,411


92 ,172 ,252 ,332 ,412


93 ,173 ,253 ,333 ,413


94 ,174 ,254 ,334 ,414


95 ,175 ,255 ,335 ,415


96 ,176 ,256 ,336 ,416


97 ,177 ,257 ,337 ,417


98 ,178 ,258 ,338 ,418


99 ,179 ,259 ,339 ,419


100 ,180 ,260 ,340 ,420


101 ,181 ,261 ,341 ,421


102 ,182 ,262 ,342 ,422


103 ,183 ,263 ,343 ,423


104 ,184 ,264 ,344 ,424


105 ,185 ,265 ,345 ,425


106 ,186 ,266 ,346 ,426


107 ,187 ,267 ,347 ,427


108 ,188 ,268 ,348 ,428


109 ,189 ,269 ,349 ,429


110 ,190 ,270 ,350 ,430


111 ,191 ,271 ,351 ,431


112 ,192 ,272 ,352 ,432


113 ,193 ,273 ,353 ,433


114 ,194 ,274 ,354 ,434


115 ,195 ,275 ,355 ,435


116 ,196 ,276 ,356 ,436


117 ,197 ,277 ,357 ,437


118 ,198 ,278 ,358 ,438


119 ,199 ,279 ,359 ,439


120 ,200 ,280 ,360 ,440


121 ,201 ,281 ,361 ,441


122 ,202 ,282 ,362 ,442


123 ,203 ,283 ,363 ,443


124 ,204 ,284 ,364 ,444


125 ,205 ,285 ,365 ,445


126 ,206 ,286 ,366 ,446


127 ,207 ,287 ,367 ,447


128 ,208 ,288 ,368 ,448


129 ,209 ,289 ,369 ,449


130 ,210 ,290 ,370 ,450


131 ,211 ,291 ,371 ,451


132 ,212 ,292 ,372 ,452


133 ,213 ,293 ,373 ,453


134 ,214 ,294 ,374 ,454


135 ,215 ,295 ,375 ,455


136 ,216 ,296 ,376 ,456


137 ,217 ,297 ,377 ,457


138 ,218 ,298 ,378 ,458


139 ,219 ,299 ,379 ,459


140 ,220 ,300 ,380 ,460


141 ,221 ,301 ,381 ,461


142 ,222 ,302 ,382 ,462

Appendix 2

Stuart P. Edwardson PhD Thesis VIII


143 ,223 ,303 ,383 ,463


144 ,224 ,304 ,384 ,464


145 ,225 ,305 ,385 ,465


146 ,226 ,306 ,386 ,466


147 ,227 ,307 ,387 ,467


148 ,228 ,308 ,388 ,468


149 ,229 ,309 ,389 ,469


150 ,230 ,310 ,390 ,470


151 ,231 ,311 ,391 ,471


152 ,232 ,312 ,392 ,472


153 ,233 ,313 ,393 ,473


154 ,234 ,314 ,394 ,474


155 ,235 ,315 ,395 ,475


156 ,236 ,316 ,396 ,476


157 ,237 ,317 ,397 ,477


158 ,238 ,318 ,398 ,478


159 ,239 ,319 ,399 ,479


160 ,240 ,320 ,400 ,480


161 ,241 ,321 ,401 ,481


162 ,242 ,322 ,402 ,482


163 ,243 ,323 ,403 ,483


164 ,244 ,324 ,404 ,484


165 ,245 ,325 ,405 ,485


166 ,246 ,326 ,406 ,486


167 ,247 ,327 ,407 ,487


168 ,248 ,328 ,408 ,488


169 ,249 ,329 ,409 ,489


170 ,250 ,330 ,410 ,490

*SOLID SECTION,ELSET=PLATE,MATERIAL=STEEL

*MATERIAL,NAME=STEEL

*DENSITY

7.8E3

*CONDUCTIVITY, TYPE=ISO

46.1,20

46.1,100

44.8,200

39.8,400

34.3,600

26.4,800

27.2,1000

28.5,1100

29.7,1200

30.0,1300

34.0,1465

72.0,1500

100.0,1520

120.0,1544

120.0,3000

*LATENT HEAT

2.7379E5,1465.,1544.

*SPECIFIC HEAT

477.,100

511.,200

590.,400

741.,600

821.,800

821.,1000

821.,3000

*PHYSICAL CONSTANTS,ABSOLUTE ZERO = -273.16

STEPHAN BOLTZANN = 5.669E-9

*INITIAL CONDITIONS,TYPE=TEMPERATURE

ALL, 20.

*RESTART,WRITE,FREQUENCY=1

**********************************************************************

** USER SUBROUTINES

*

**********************************************************************

**

** Dflux User Subroutine

*********************************************************************

**

*USER SUBROUTINE

SUBROUTINE

DFLUX(FLUX,SOL,KSTEP,KINC,TIME,NOEL,NPT,COORDS,

&JLTYP)

include 'ABA_PARAM.INC'

C

DIMENSION FLUX(2), TIME(2), COORDS(3)

REAL thermcon, alpha,vel,radius1,radius2,pi,spot_radius_zero,

&x,y,z,power,lamda,dist_z,focal_length,spot_radius_z,

&boiling_temp,melting_temp,absorp

C

power = 760

boiling_temp = 2750.

melting_temp = 1400.

vel = 0.03

C ********************

alpha = 6.4103E-6

thermcon = 30.

pi = 3.141592654

dist_z = 0.047

focal_length = 0.127

lamda = 10.6E-6

absorp= 0.7

M_sq = 2.0

C

x = COORDS(1) - 0.04

y = COORDS(2) - (vel * TIME(2))

z = COORDS(3) - 0.0015

C

spot_radius_zero = 2*M_sq*focal_length*lamda/(pi*0.012)

C

spot_radius_z = sqrt(spot_radius_zero*spot_radius_zero

&+(lamda*dist_z*lamda*dist_z/(pi*spot_radius_zero*

&pi*spot_radius_zero)))

C

Appendix 2

Stuart P. Edwardson PhD Thesis IX

cenint = (power/(pi*spot_radius_z*spot_radius_z))

C

radius1 = sqrt(x*x + y*y + z*z)

C

radius2 = sqrt(x*x + y*y)

C

FLUX(1) = absorp * cenint * exp(-2*(radius2*radius2)/

&(spot_radius_z*spot_radius_z))

C

FLUX(2) = 0.

C

C

C *** ALTERNATIVE FLUXES ***

C FLUX(2) = 2*pi*thermcon*radius1 * exp(vel*(radius1-x)

C &/(2*alpha))

C FLUX(1) = thermcon*SOL

C FLUX(2) = 4*pi*alpha*radius1

C

return

end

**

**Central Intensity is calculated from Power and approximate spot

**radius

**at appropriate focal distance

*STEP, INC=1000

*HEAT TRANSFER,DELTMX=20000,END=SS

0.01,60,1.E-5,0.1,0.01

*FILM

SIDE1,F3,25,10

SIDE2,F5,25,10

END1,F6,25,10

END2,F4,25,10

PLATE,F2,25,10

PLATE,F1,25,10

*DFLUX

LASER1,S2NU

LASER2,S2NU

LASER3,S2NU

LASER4,S2NU

LASER5,S2NU

LASER6,S2NU

LASER7,S2NU

LASER8,S2NU

LASER9,S2NU

LASER10,S2NU

LASER11,S2NU

LASER12,S2NU

LASER13,S2NU

LASER14,S2NU

LASER15,S2NU

LASER16,S2NU

LASER17,S2NU

LASER18,S2NU

LASER19,S2NU

LASER20,S2NU

LASER21,S2NU

LASER22,S2NU

LASER23,S2NU

LASER24,S2NU

LASER25,S2NU

LASER26,S2NU

LASER27,S2NU

LASER28,S2NU

LASER29,S2NU

LASER30,S2NU

LASER31,S2NU

LASER32,S2NU

LASER33,S2NU

LASER34,S2NU

LASER35,S2NU

LASER36,S2NU

LASER37,S2NU

LASER38,S2NU

LASER39,S2NU

LASER40,S2NU

LASER41,S2NU

LASER42,S2NU

LASER43,S2NU

LASER44,S2NU

LASER45,S2NU

LASER46,S2NU

LASER47,S2NU

LASER48,S2NU

LASER49,S2NU

LASER50,S2NU

LASER51,S2NU

LASER52,S2NU

LASER53,S2NU

LASER54,S2NU

LASER55,S2NU

LASER56,S2NU

LASER57,S2NU

LASER58,S2NU

LASER59,S2NU

LASER60,S2NU

LASER61,S2NU

LASER62,S2NU

LASER63,S2NU

LASER64,S2NU

LASER65,S2NU

LASER66,S2NU

LASER67,S2NU

LASER68,S2NU

LASER69,S2NU

LASER70,S2NU

LASER71,S2NU

LASER72,S2NU

LASER73,S2NU

LASER74,S2NU

LASER75,S2NU

LASER76,S2NU

LASER77,S2NU

LASER78,S2NU

LASER79,S2NU

LASER80,S2NU

*NODE FILE,FREQ=1,GLOBAL=YES

NT

*END STEP

Appendix 3

Stuart P. Edwardson PhD Thesis X

Appendix 3 – Beam Diameter Prediction The laser spot diameters used in the investigations reported in this thesis were verified by burn prints, discussed earlier, and via the following method. The minimum beam waist, W0, was derived using equation A1. (A1)

Were:

mm 12 =Lensat Radius Beam =rmm 0.0106 =length Laser Wave =

190mm & mm 127 =Length Focal Lens =fFactorQuality Beam= M2

λ

Using this result and the lens focal length, calculation of the incident beam spot radius, W(z), at a lens-to-workpiece distance z, can be found using equation A2. (A2) Were:

A spread sheet was used to determine an approximation for the spot diameter at a

given lens to workpiece standoff, employing the above equations.

An example using a 127mmFL lens and an M2 of 2 is given here:

Parameters used:

Laser Wavelength [m] 10.6E-06Lens Focal Length [m] 127.0E-03M^2 (Approx.) 2Beam Radius at Lens [m] 6.00E-03Beam Waist Wo [m] (A1) 1.43E-04

rfMW

2

0 πλ

=

21

2

20

2

0 1)(W

+=

WzMWz

πλ

FactorQuality Beam Mm)(10.6lenght Laser Wave=

Waist Beam MinimumW Surface W/PoPosition t Waist Beam Minimum from Distance

2

0

=

==

µλ

z

Appendix 3

Stuart P. Edwardson PhD Thesis XI

Output:

Lens to Workpiece Distance [mm] z [m]

Beam Radius at z [m]

Beam Diameter at z [m]

127 000.0E+00 1.43E-04 2.86E-04 128 1.0E-03 1.50E-04 3.01E-04 129 2.0E-03 1.71E-04 3.43E-04 130 3.0E-03 2.01E-04 4.02E-04 131 4.0E-03 2.37E-04 4.74E-04 132 5.0E-03 2.76E-04 5.52E-04 133 6.0E-03 3.17E-04 6.35E-04 134 7.0E-03 3.60E-04 7.20E-04 135 8.0E-03 4.04E-04 8.08E-04 136 9.0E-03 4.49E-04 8.97E-04 137 10.0E-03 4.94E-04 9.87E-04 138 11.0E-03 5.39E-04 1.08E-03 139 12.0E-03 5.85E-04 1.17E-03 140 13.0E-03 6.31E-04 1.26E-03 141 14.0E-03 6.77E-04 1.35E-03 142 15.0E-03 7.23E-04 1.45E-03 143 16.0E-03 7.69E-04 1.54E-03 144 17.0E-03 8.16E-04 1.63E-03 145 18.0E-03 8.62E-04 1.72E-03 146 19.0E-03 9.09E-04 1.82E-03 147 20.0E-03 9.56E-04 1.91E-03 148 21.0E-03 1.00E-03 2.00E-03 149 22.0E-03 1.05E-03 2.10E-03 150 23.0E-03 1.10E-03 2.19E-03 151 24.0E-03 1.14E-03 2.29E-03 152 25.0E-03 1.19E-03 2.38E-03 153 26.0E-03 1.24E-03 2.47E-03 154 27.0E-03 1.28E-03 2.57E-03 155 28.0E-03 1.33E-03 2.66E-03 156 29.0E-03 1.38E-03 2.76E-03 157 30.0E-03 1.42E-03 2.85E-03 158 31.0E-03 1.47E-03 2.94E-03 159 32.0E-03 1.52E-03 3.04E-03 160 33.0E-03 1.57E-03 3.13E-03 161 34.0E-03 1.61E-03 3.23E-03 162 35.0E-03 1.66E-03 3.32E-03 163 36.0E-03 1.71E-03 3.41E-03 164 37.0E-03 1.75E-03 3.51E-03 165 38.0E-03 1.80E-03 3.60E-03 166 39.0E-03 1.85E-03 3.70E-03 167 40.0E-03 1.90E-03 3.79E-03 168 41.0E-03 1.94E-03 3.88E-03 169 42.0E-03 1.99E-03 3.98E-03 170 43.0E-03 2.04E-03 4.07E-03 171 44.0E-03 2.08E-03 4.17E-03 172 45.0E-03 2.13E-03 4.26E-03

Appendix 3

Stuart P. Edwardson PhD Thesis XII

173 46.0E-03 2.18E-03 4.36E-03 174 47.0E-03 2.23E-03 4.45E-03 175 48.0E-03 2.27E-03 4.54E-03 176 49.0E-03 2.32E-03 4.64E-03 177 50.0E-03 2.37E-03 4.73E-03 178 51.0E-03 2.41E-03 4.83E-03 179 52.0E-03 2.46E-03 4.92E-03 180 53.0E-03 2.51E-03 5.02E-03 181 54.0E-03 2.56E-03 5.11E-03 182 55.0E-03 2.60E-03 5.20E-03 183 56.0E-03 2.65E-03 5.30E-03 184 57.0E-03 2.70E-03 5.39E-03 185 58.0E-03 2.74E-03 5.49E-03 186 59.0E-03 2.79E-03 5.58E-03 187 60.0E-03 2.84E-03 5.68E-03 188 61.0E-03 2.89E-03 5.77E-03 189 62.0E-03 2.93E-03 5.87E-03 190 63.0E-03 2.98E-03 5.96E-03 191 64.0E-03 3.03E-03 6.05E-03 192 65.0E-03 3.07E-03 6.15E-03 193 66.0E-03 3.12E-03 6.24E-03 194 67.0E-03 3.17E-03 6.34E-03 195 68.0E-03 3.22E-03 6.43E-03 196 69.0E-03 3.26E-03 6.53E-03 197 70.0E-03 3.31E-03 6.62E-03 198 71.0E-03 3.36E-03 6.71E-03 199 72.0E-03 3.40E-03 6.81E-03 200 73.0E-03 3.45E-03 6.90E-03 201 74.0E-03 3.50E-03 7.00E-03 202 75.0E-03 3.55E-03 7.09E-03 203 76.0E-03 3.59E-03 7.19E-03 204 77.0E-03 3.64E-03 7.28E-03 205 78.0E-03 3.69E-03 7.38E-03 206 79.0E-03 3.74E-03 7.47E-03 207 80.0E-03 3.78E-03 7.56E-03 208 81.0E-03 3.83E-03 7.66E-03 209 82.0E-03 3.88E-03 7.75E-03 210 83.0E-03 3.92E-03 7.85E-03

Appendix 4

Stuart P. Edwardson PhD Thesis XIII

Appendix 4 – Safety Interlocks & System Layout for the

Electrox Workstation No. 2

Workstation 2 Layout – R/H side of laser

PC Based Controller, Servo Amplifier Housing and Handheld Manual Shutter Control

Processing Head & Laser Range finder

Appendix 4

Stuart P. Edwardson PhD Thesis XIV

Manual Shutter Release

(Key Lockable Closed to give software Control)

Software Controlled Relay (Normally

Closed)

Shutter Control

Hand Held Control

White Relay & I/O Control Box

Control circuit for Shutter Control

Galil DMC 1730 ISA Card and Naples Coombe Servo Amplifier and Integration housing

System Layout Schematic

Control PC & Galil DMC 1730

NC Servostep 1700 Integration Housing

X,Y,Z Tables Electrox 1.5kW CO2 Laser

Manual Handheld Shutter Control

Relay Control Box & E/stop

Appendix 5

Stuart P. Edwardson PhD Thesis XV

Appendix 5 – MEL M5 & M1 Laser Range Finder

Specifications

www.melsensor.de

M5 Laser/100

Appendix 5

Stuart P. Edwardson PhD Thesis XVI

Appendix 5

Stuart P. Edwardson PhD Thesis XVII

M1 L/100

Appendix 6

Stuart P. Edwardson PhD Thesis XVIII

Appendix 6 – Example Galil CNC Code Program to linearly interpolate along three lines that are defined by a series of X & Y points Starting speed and end speed between each point is also specified Values are given as ‘Encoded Counts’ 1mm = 400 counts, Comments are not part of the CNC file #Z1 Internal program name SB1 Unlock Z Axis SB3 Arm Shutter (closed) SP16000,16000,30000 Set general speed X,Y,Z AC450000,450000,900000 Set general acceleration X,Y,Z DC450000,450000,900000 Set general de-acceleration X,Y,Z PA63841,36667,0 Specify an absolute position BG Begin Movement to position AM Wait until movement complete CB3 Open Shutter WT200 Wait 200ms for shutter to open LM XY Specify Linear Interpolation Mode in X&Y VA100000 Vector Acceleration VD100000 Vector De-Acceleration LI2826,-2292 <34162 >34162 Linear Interpolation through these points LI3333,-2470 <34162 >34162 XCo-ordinate,Yco-ordinate,<Starting Speed >End speed LI4000,-1576 <34162 >34162 for each segment LI3333,-700 <34162 >34162 LI4000,-77 <34162 >34162 LI4000,806 <34162 >34162 LI4000,1205 <34162 >34162 LI3333,1793 <34162 >34162 LI3333,2750 <34162 >34162 LI1335,3228 <34162 >34162 LI-1435,4000 <34162 >34162 LI-3234,2059 <34162 >34162 LI-3868,1941 <34162 >34162 LI-3465,1675 <34162 >34162 LI-4000,943 <34162 >34162 LI-4000,-303 <34162 >34162 LI-3759,-982 <34162 >34162 LI-3574,-1550 <34162 >34162 LI-3503,-2450 <34162 >34162 LI-2497,-2059 <34162 >34162 LI-1333,-4280 <34162 >34162 LE Linear Interpolation Mode End BGS Begin Specified Sequence AM SB3 Close Shutter WT500 Wait 500ms for Shutter to close PA18254,76667,0 Specify next start location BG AM CB3 WT200 LM XY VA100000 VD100000 LI2307,-4000 <28457 >29104 LI2105,-3114 <29104 >29971 LI2644,-2886 <29971 >31021 LI3036,-2667 <31021 >31624 LI2986,-2562 <31624 >31891 LI3333,-2445 <31891 >32732 LI3333,-1895 <32732 >33461 LI4000,-1578 <33461 >33803 LI4000,-246 <33803 >33454 LI4000,-246 <33454 >33607 LI4000,-808 <33607 >34162

Appendix 6

Stuart P. Edwardson PhD Thesis XIX

LI3333,-194 <34162 >34162 LI4000,616 <34162 >34123 LI4000,870 <34123 >33060 LI4000,1013 <33060 >33088 LI3665,808 <33088 >33179 LI4335,878 <33179 >32681 LI4000,438 <32681 >32689 LI4000,87 <32689 >32517 LI4000,-155 <32517 >32770 LI4000,-757 <32770 >32959 LI3333,-817 <32959 >33126 LI4000,-100 <33126 >33194 LI4000,374 <33194 >33834 LI4000,909 <33834 >33430 LI4000,1379 <33430 >32632 LI4000,1567 <32632 >33701 LI3333,2053 <33701 >33716 LI3059,2145 <33716 >32693 LI2941,2415 <32693 >31369 LI2667,3134 <31369 >30741 LI2000,3205 <30741 >30364 LE BGS AM SB3 WT500 PA114823,74667,0 BG AM CB3 WT200 LM XY VA100000 VD100000 LI-3490,-1703 <32610 >32790 LI-4000,647 <32790 >32752 LI-4000,718 <32752 >32579 LI-3333,-1976 <32579 >32609 LI-2667,-2616 <32609 >33389 LI-4000,-859 <33389 >33477 LI-4000,311 <33477 >32731 LI-4000,-570 <32731 >31934 LI-4000,-853 <31934 >31322 LI-3333,-358 <31322 >31461 LI-4000,-300 <31461 >32013 LI-4000,-497 <32013 >32347 LI-4548,-610 <32347 >32569 LI-3452,-379 <32569 >32351 LI-4000,-358 <32351 >32078 LI-4000,-285 <32078 >32267 LI-4000,-182 <32267 >32725 LI-4000,360 <32725 >32925 LI-4000,1069 <32925 >32132 LI-4000,1220 <32132 >31681 LI-3849,1222 <31681 >31382 LI-3484,2059 <31382 >30896 LI-2930,2607 <30896 >30220 LI-2462,3333 <30220 >29120 LE BGS AM SB3 WT500 PA0,0,0 Specify Home Position BG Move to home (Begin Specified Movement) AM EN Program End

Documents

PhD Thesis by Stuart P Edwardson