16
Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel Trauth, Fritz Klocke, Patrick Mattfeld, Andreas Klink Laboratory for Machine Tools and Production Engineering (WZL) of RWTH Aachen University Director: Prof. Dr.-Ing. Dr.-Ing. E.h. Dr. h.c. Dr. h.c. Fritz Klocke

Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

Time-Efficient Prediction of the Surface Layer State after Deep Rolling using

Similarity Mechanics Approach

Daniel Trauth, Fritz Klocke, Patrick Mattfeld, Andreas Klink

Laboratory for Machine Tools and Production Engineering (WZL) of RWTH Aachen University

Director: Prof. Dr.-Ing. Dr.-Ing. E.h. Dr. h.c. Dr. h.c. Fritz Klocke

Page 2: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

Agenda

Conclusion and outlook 5

Prediction of the surface layer state using similarity mechanics approach 4

FE-Modeling of the deep rolling process 3

Experimental examination of deep rolling and identification of significant process parameters 2

Motivation, objective and approach 1

Page 3: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

Motivation and objective

Processing highly stressed technical

components using deep rolling is currently

based on individual know how of

experts, elaborate experiments and

subsequent time- and cost-intensive

measurements

Process modeling using finite element

method requires a high expertise of the

users as well as high computation times

For these reasons, a new method is

required, which enables an efficient and

also quantitive process design regarding

Residual stresses and

Strain hardening

in the surface layer

Motivation Objective

Development of an innovative method to

predict the residual stresses and the

strain hardening by

Developing FE-models to create a

comprehensive dataset of residual stresses

and strain hardening for different input

parameters

Deriving correlations between input and output

parameters using similarity mechanics

approach

Input Output

Page 4: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

Analysis

Experiments

Simulation and

Verification

Interpretation of

the exp. and

num. results

Literature

review State of the

art

Time-efficient

prediction of the

surface layer state

Modeling

Strain

hardening

Abstract

Verification

= ?

Real. Sim.

Simulation

F, x...

Similarity

mechanics

Surface

roughness Residual

stresses Deep rolling

Evaluation

- +

Significance

analysis

A

B

A

B

𝜋 = 𝐿1𝑀−2𝑇1

Database

1s

Analyze

F v

Approach

Page 5: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

Conclusion and outlook 5

Prediction of the surface layer state using similarity mechanics approach 4

FE-Modeling of the deep rolling process 3

Experimental examination of deep rolling and identification of significant process parameters 2

Motivation, objective and approach 1

Agenda

Page 6: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

Test parameters

Deep rolling tools (Ecoroll AG, Germany)

– Hydraulic tool types: HG6 and HG13

Hydraulic power unit (Ecoroll AG)

– Type: HGP 400

Material to be tested

– Heat treated steel

42CrMo4 (ASTM: A322-4140)

– Ductile cast iron

GGG60 (ASTM: A536-80-55-06)

– Nickel-based alloy

IN718 (ASTM: B637)

Sample geometry

– Lateral surface area

– Outer radius

– Borehole

Experimental set-up

Three jaw chuck

Sample

Sleeve

Deep rolling tool

GGG60 IN718 42CrMo4

Borehole

(hidden) Outer radius Lateral surface

area

Lathe

Experimental examination of deep rolling

Page 7: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

Significance analysis for maximum strain hardening at the lateral surface area acc. to DOE

Str

ain

ha

rde

nin

g

[H

RC

]

20

30

40

50

60

100 400

Walzdruck Rolling velocity v Rolling overlap o Ball diameter d Rolling pressure p

mm mm/s bar %

42CrMo4 42CrMo4

GGG60

IN718 IN718

-1400

-1200

-1000

-800

100 400

Rolling pressure p

-1400

-1200

-1000

-800

HG6 HG13

Ball diameter d

-1400

-1200

-1000

-800

0.3 0.8

Rolling overlap o

Re

sid

ua

l str

esse

s σ

RS [M

Pa

]

-1400

-1200

-1000

-800

70 150

Rolling velocity v

Significance analysis for maximum residual stresses at the lateral surface area acc. to DOE

mm mm/s bar %

better

better

Identification of significant process parameters

20

30

40

50

60

HG6 HG13

20

30

40

50

60

0,3 0,8

20

30

40

50

60

70 150

Page 8: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

Conclusion and outlook 5

Prediction of the surface layer state using similarity mechanics approach 4

FE-Modeling of the deep rolling process 3

Experimental examination of deep rolling and identification of significant process parameters 2

Motivation, objective and approach 1

Agenda

Page 9: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

FE-Modeling of material characteristics

-1200

-800

-400

0

400

800

1200

-3% -2% -1% 0% 1% 2% 3%

Sp

an

nu

ng

[M

Pa]

Dehnung [%]

Bauschinger-Versuch bei 42CrMo4

-1200

-800

-400

0

400

800

1200

-3% -2% -1% 0% 1% 2% 3%

Sp

an

nu

ng

[M

Pa]

Dehnung [%]

Bauschinger-Versuch bei GGG60

-1200

-800

-400

0

400

800

1200

0 20 40 60 80S

tres

s σ

[M

Pa

]

Time t [s]

GGG60 [2%]

Simulation and validation

Material modeling

Simulation Experiment

0

Strain ε [%]

GGG60

Str

es

s σ

[M

Pa

]

0

+2 -2

+400

+800

-800

-400

Tensile

load

Compressive

load Tensile

load

F F FE-Model

F

Experimental cyclic

compression-tension-test with

2%, 4% and 6% total strain

Tests were performed for all

three materials to determine the

parameters of the combined

cyclic nonlinear isotropic and

kinematic plasticity model acc.

to Lemaitre-Chaboche (LCP-

Model)

FE-Setup of the

compression-tension-test

Fitting of the Parameters

𝑄∞, 𝑏, 𝐶 and 𝛾 of the LCP-

Model using Curve Fitting

Toolbox in Matlab

Validation of the results

LCP-Model:

Isotropic behavior:

𝜎0 = 𝜎0 + 𝑄∞ (1 − 𝑒−𝑏𝜀−𝑝𝑙)

Kinematic behavior:

𝛼 = 𝐶𝜀 𝑝𝑙(𝜎 − 𝛼)

𝜎0− 𝛾𝛼𝜀 𝑝𝑙

Bauschinger experiment

Page 10: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

Boundary conditions

ωB

RC

rB

Workpiece

Rolling tool

Connector

FE-Modeling and simulation

Five rolling tools

Workpiece

(C3D8R, LC-Plasticity)

Averaged

evaluation path

σRS [MPa]

+1431 0

Rigid

y

x

z

ωC 𝑐

Modeling of process kinematics using

a special connector assembly

This makes the use of the Abaqus

mass scaling option usefull

Tests have shown, that a mass

scaling factor of 250 has no influence

on the results despite 15-times

acceleration of the computation time

Setup of 9 evaluation paths in the

model center to minimize numerical

instabilities

Evaluating residual stresses by

means of the Mises-stress-tensor and

strain hardening by the plastic

equivalent strain

FE-Modeling of the deep rolling process

Page 11: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

-1500.00

-1000.00

-500.00

0.00

Res

idu

al s

tres

se

s σ

RS

,i

[MP

a]

Surface layer depth t [µm]

Exp. σx Exp. σy

Sim. σx Sim. σy

Exp. σRS,x

Sim. σRS,x

Exp. σRS,y

Sim. σRS,y

Surface layer depth t [µm]

Res

idu

al S

tres

s P

rofi

le

σR

S,i [

MP

a]

Verification Characterization

Maximum deviation of 1.1% for σRS,x

for the highest compressive residual

stresses

Good qualitative accordance of the

numerical residual stresses with the

experimentally measured stress

Thus, the FE-model is verified and

qualifies for further investigations

using similarity mechanics

This involves the characterization of a

typical residual stress depth profile by

significant values (𝜎0, 𝜎𝑚𝑖𝑛 , 𝜎𝑚𝑎𝑥)

𝜎𝑚𝑎𝑥

𝜎0

𝜎𝑚𝑖𝑛 𝑧𝑚𝑖𝑛 𝑧0 𝑧𝑚𝑎𝑥

Verification of the numerical deep rolling process

Page 12: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

Conclusion and outlook 5

Prediction of the surface layer state using similarity mechanics approach 4

FE-Modeling of the deep rolling process 3

Experimental examination of deep rolling and identification of significant process parameters 2

Motivation, objective and approach 1

Agenda

Page 13: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

-1200

-700

-200

300

0.0 1.0 2.0

Surface layer depth t [mm]

s_parallel

s_parallel_AM

s_senkrecht_AM

s_senkrecht

Res

idu

al s

tres

se

s σ

RS

[M

Pa

]

Sim. σRS,x

Est. σRS,x

Est. σRS,y

Sim. σRS,y

42CrMo4, 250 bar, HG6, 30%

-1200

-960

-720

-480

-240

0

-5.80E-03

-4.64E-03

-3.48E-03

-2.32E-03

-1.16E-03

0.00E+00

sm

in/E

[ -

]

Intensity number π8 [ - ]

σmin,x

σmin,y

42CrMo4

Analytic functions Estimation and validation

Significant dimensionless numbers

acc. to DOE and Buckingham

𝛱1= 𝑓

𝐷 𝛱2=

𝑣

𝐸𝜌

𝛱3= 𝑑

𝐷 𝛱8=

𝑝

𝐸

𝜎𝑚𝑖𝑛,𝑦

𝐸= −191721𝜋8

2 + 36,649𝜋8 − 0,0035

𝜎𝑚𝑖𝑛,𝑥𝐸

= −199054𝜋82 + 20,999𝜋8 − 0,0042

The significant values of a residual

stress profile can be well estimated

by the presented method (maximum

deviation of 5%)

In our paper, the authors propose a

function to connect the estimated

values to a representative profile

Prediction of the surface layer state

Page 14: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

Conclusion and outlook 5

Prediction of the surface layer state using similarity mechanics approach 4

FE-Modeling of the deep rolling process 3

Experimental examination of deep rolling and identification of significant process parameters 2

Motivation, objective and approach 1

Agenda

Page 15: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

Conclusion Outlook

This work presents significant process

input parameters acc. to DOE to

maximize residual stresses and strain

hardening for 42CrMo4, IN718 and

GGG60

Furthermore a verified FE-Model based

on a special connector is presented

which allows a comprehensive

investigation of the correlation between

input and output parameters

Hereby a very effective and simple

method could be developed, which

allows an estimation of process results

like residual stresses or strain

hardening using similarity mechanics

approach

Future work aims at the increase of

the amount of characteristic values of

a depth profile to make the application

of a connecting-function as presented

in the paper obsolete

Besides, the prediction of the fatigue

life using similarity mechanics will be

investigated. Therefore dimensionless

numbers relating the surface layer

state to a S/N curve are determined

Using experiments and FE-analyses,

suitable analytic functions allow the

prediction of the number of stress

cycles or the critical fatigue stress de-

pending on residual stresses, strain

hardening and the surface quality

Conclusion and outlook

Page 16: Time-Efficient Prediction of the Surface Layer State after ... · Time-Efficient Prediction of the Surface Layer State after Deep Rolling using Similarity Mechanics Approach Daniel

The authors would like to thank the German Federal Ministry of Economics and Technology (BMWi)

for supporting this research project through the Central Innovation Programme for SME (ZIM).

Further, we express our graditute to K. Röttger and S. Fricke of ECOROLL AG Werkzeugtechnik for

their support in conducting the experiments cited in this paper.

Thank your very much for your attention

Acknowledgement