Integrated Computational Materials Engineering for … Documents...LM Aero Airframe Goals – Tailoring composite damage tolerance and in-plane strengths to achieve maximum weight

1

Innovation With Purpose

COPYRIGHT 2014 LOCKHEED MARTIN CORPORATION – ALL RIGHTS RESERVED

S. P. Engelstad, R.W. Koon, J.E. Action Lockheed Martin Aeronautics Company J.M. Riga Lockheed Martin Advanced Technology Labs A.M. Waas The University of Michigan D. Robbins, R.W. Dalgarno Autodesk A.R. Arafath, A. Poursartip Convergent Manufacturing Technologies, Inc.

Integrated Computational Materials Engineering for

Airframe Composite Structure Applications

COPYRIGHT 2014 LOCKHEED MARTIN CORPORATION – ALL RIGHTS RESERVED 2

Outline

• Introduction • Model Definition

– Convergent COMPRO – University of Michigan – Micromechanics – Autodesk ASCA

• Current Results – COMPRO Material Characterization – University of Michigan – Micromechanics Modeling of Curing

ply strengths – ASCA Modeling of OHC/FHC, CSAI


Introduction • Integrated Computational Materials Engineering (ICME)

– AFRL and industry vision to reduce the material and process development cycle time and cost

• Integrated Computational Methods for Composite Materials (ICM2) is an AFRL, GE, and LM Aero composites ICME program – GE studying engine applications – LM Aero studying airframe applications

• LM Aero Airframe Goals – Tailoring composite damage tolerance and in-plane strengths to

achieve maximum weight savings – Reduce material qualification costs and time spans

• Today’s BMI systems took 20 years from the lab to its usage on F-22 • The insertion of new materials technologies has become less and less

frequent, and available materials are a constraint on the design


Introduction

• BMI systems have experienced increased usage due to key structural design properties such as

– Open hole compression (OHC) and compression-strength-after-impact (CSAI) – These key properties often “size” the acreage of the aircraft composite skins.

• The ICM2 program attempts to amplify the weight advantage of IM7/M65 BMI – Studying the autoclave cure cycle effects – Goal to optimize these critical design properties for aircraft weight

• GE/LM team has chosen – Cure Process Effects:

Convergent Manufacturing Technologies’ COMPRO

– Multi-scale progressive damage: Autodesk Simulation Composite Analysis (ASCA) software

– Ply level strength effects of cure: University of Michigan (UofM)


Model Integration

• The ICM2 program will utilize a digital framework to integrate the three modeling tools previously introduced

• This framework is based on the model integration interface ModelCenter®, and will be built through a joint effort between GE and LM Aero

• Goal is to show end-to-end integration from materials database through part structure performance to carry out process trade analysis

• This presentation describes the individual components of the digital framework, as well as current results of the airframe portion of this program


COMPRO Analysis

• COMPRO analysis will be performed to analyze the cure cycle of panel/part – Different cure cycles studied to reduce residual stress and

optimize degree of cure – Thermo-mechanical properties and non-mechanical strains at

the end of the curing process are passed to ASCA – Thermo-mechanical properties during the curing process are

passed to UM code


Convergent’s COMPRO Process Analysis

COMPRO: a multi-physics composite processing plug-in for 3rd party FE solvers


UM-Micro Analysis • Micromechanics analysis will consist of progressive damage

analysis at the unit cell level to determine resulting ply level strengths – Inputs

• From COMPRO: Temperature profile, degree of cure, and cure rate vs. time

• Constituent properties • Resin strength test data as a function of cure cycle

– Outputs • Ply level strengths as a function of cure cycle


UofM Micromechanics Based Modeling of Effects of Curing on Ply Strengths • Curing stress can cause micro damage that alters mechanical

properties • Micromechanics analysis performed at the fiber-matrix scale to capture

failure mechanisms – 3D hexagonally packed representative unit cell (RUC) and multi-fiber random

packed unit cells – RUC analyzed to compute the actual strengths of the lamina – Optimize cure cycle to reduce detrimental effects of residual stresses

• UoM approach is based on two steps – Step 1:

• First the temperature profile, degree of cure and cure rate in the matrix are computed using the COMPRO software

• Determine stress evolution in the RUC due to the curing and the possibility of damage during cure

– Step 2: • Apply mechanical loading to the RUC in different directions to determine strengths • Damage progression modeled using the crack band approach • Outcome of this analysis is ply level composite stiffness and strengths, while

accounting for the cure cycle

Hexagonally packed RUC


Autodesk ASCA Analysis

• Autodesk Simulation Composite Analysis (ASCA) Progressive Damage analysis will be performed to determine the effect of cure on critical part strengths – GE will focus on engine duct flange strengths – LM will focus on Open-Hole-Compression (OHC), Filled-Hole-

Compression (FHC), and Compression-Strength-After-Impact (CSAI) allowable strengths

CSAI ASCA Model

Delam 1 Ply 14: 0°

Open Hole Compression


Autodesk ASCA

• ASCA software decomposes composite stress and strain states into constituent average stress and strain states – Stiffness of the damaged composite material is obtained by

homogenization of the microstructure • ASCA features two different methods for degrading the stiffness

of damaged constituent materials – Instantaneous degradation of the stiffness of the damaged

constituent to a user-specified fraction of the original stiffness – Energy-based degradation of the stiffness is gradually reduced to

zero as the strain level increases beyond the damage initiation level


Current Results


Current Results

• First year of ICM2 program Airframe efforts at the lamina level focused on – COMPRO material characterization of IM7/M65 lamina – UofM predictions of cure cycle effects on the lamina elastic

properties and strengths – Physical resin “effect-of-cure-cycle” tests as required by UofM

analysis • At the laminate level, ASCA progressive damage tool being linked

with COMPRO cure residual stress and UofM micromechanics strength property predictions – ASCA baseline cure cycle models developed for

• Open-Hole Compression (OHC) • Filled-Hole Compression (FHC) • Compression-Strength-After-Impact (CSAI)


Convergent Process Analysis Material Characterization • Mechanical constitutive model needed for process-induced residual stress

development – Simplest approach is “Cure Hardening Instantaneously Linear Elastic” (CHILE) model,

where modulus of elasticity changes as a function of the instantaneous temperature and degree of cure (e.g. White and Hahn (1992) and Johnston et al. (2001))

– But polymers show viscoelastic behavior, especially partially cured at high temperatures in a cure cycle

– CHILE models do not accurately capture the residual stress development during free standing post cure of a partially cured composite structure [Zobeiry 2003]

– Differential viscoelastic approach is being implemented in the upcoming next release of COMPRO (Version 3) to represent the viscoelastic behavior of a curing thermoset matrix composite

• An efficient methodology [Thorpe et al. 2013] to characterize the Hexply M65 material for viscoelastic constants was used here

– Testing was performed on neat resin beams and uni-directional (UD) beams – Note that all UD beams were prepared such that testing could be performed in the 2-

direction (in-plane, transverse)

Neat resin and unidirectional beam

samples


Convergent Process Analysis Material Characterization

Model prediction of spring-in angle of an L-shape composite part compared to experimental results

• The COMPRO model was verified by comparing the spring-in angle of an L-shaped part manufactured on an invar tool using the baseline cure cycle

• The model prediction of spring-in angle compared to the experiment is shown below


Convergent Process Analysis Effect of Cure Cycle Residual Stresses

Manufacturer recommended one-hold cure cycle for HexPly M65

simulated by RAVEN

Time

Tem

pera

ture

, Tg

Deg

ree

of C

ure

• Process-induced residual stresses can be sufficiently high to cause damage in the material during the manufacturing process, or accelerate the formation and growth of cracks during the service conditions

• While process-induced residual stresses are unavoidable, it was shown by [Li et al. 2014] that it is possible to reduce them by altering the cure cycle

• In this work, evaluated this effect for Hexply M65 BMI laminates using the manufacture recommended cure cycle (MRCC)

– Convergent’s RAVEN process simulation software was used to predict the development of degree of cure (DoC) and the glass transition temperature (Tg) of the material for the MRCC



Modified three-hold cure cycle for HexPly M65

simulated by RAVEN

Time

Tem

pera

ture

, Tg

Deg

ree

of C

ure

• As shown in other studies [Madhukar et al, 2000, White and Hahn, 1993 and Li et al., 2014], an intermediate hold can be added to the cure cycle to reduce the process-induced residual stresses

• A three hold cure cycle was designed using RAVEN software such that the gelation and vitrification occur during the first hold

• The second ramp rate was decreased to a small value to ensure that Tg always remains higher than the part temperature


• An un-balanced cross ply [02/902] Hexply M65 laminate was analyzed for process-induced residual stress using the three-hold cure cycle

– In both cycles, residual stresses are zero before gelation since resin modulus is negligible – During isothermal hold, cure shrinkage coupled with the resin modulus development

result in the development of residual stress – Upon cooling down, residual stress further increases due to thermal shrinkage – Process induced residual stress reduced by 6% for this preliminary alternate cycle – Although a smaller value than shown by Li et al [2014], this reduction still significant

compared to transverse failure strain of Hexply M65


Temperaturue

Res

idua

l Str

ess

Preliminary results of residual stresses in 900

layer of unbalanced cross-ply laminate along two different cure cycles


UofM: Micromechanics Modeling of Effects of Curing on Ply Strengths

Temperature Profile from COMPRO

Degree of cure vs. time from COMPRO

Cure rate vs. time

from COMPRO

UofM COMPRO model inputs to UofM Micromechanics Analysis


UofM: Micromechanics Modeling of Effects of Curing on Ply Strengths • During curing, the matrix gradually solidifies, its stiffness increases, and the cell

simultaneously contracts (cure shrinkage) due to network formation – Residual stresses develop in the matrix owing to cure shrinkage and thermal stresses – Depending on level of tensile stresses developed, the degree of cure and the rate of cure,

the material may crack locally during curing • If damage occurs, its evolution is tracked until failure is reached • To date, no damage has been observed in this RUC analysis, by applying the cure

cycle discussed • Subsequently, the “cured” RUC is subjected to mechanical load

– Example: Applying a tensile loading in 2-direction

RUC dimensions and mechanical loading applied


UofM: Micromechanics Modeling of Effects of Curing on Ply Strengths

Point D Point C

Point B Point A

Stress-Strain curve mechanical loading

Crack path at Point A

• After the peak strength, a significant reduction in stiffness occurs in RUC, and the crack path starts to be defined

• Along the stress-strain curve the failed elements in the RUC are shown in red for four points (Points A, B, C, D)

– Failure starts around the central fiber in Point A

– Elements start failing around the 2 fibers as shown in Point B

– Failure propagates as shown in Point C

– At Point D the RUC is completely split along the red path • This sequence of events has been verified for different

finite element meshes and mesh objectivity is found to prevail


ASCA OHC/FHC Baseline Cure Analysis

Aperture meshing for filled and open hole specimens

Baseline Cure Cycle Preliminary Analysis



Baseline Cure cycle: Load-stroke plots comparing model and test data • OHC and FHC test panels for the T1 laminate [45/-45/45/-45/90/-45/0/45/-45/45/-45/45]s

-1

0

1

2

3

4

5

6

7

8

0 0.02 0.04 0.06 0.08 0.1 0.12

Com

pres

sion

Loa

d (k

ips)

Actuator Displacement (in)

T1 Laminate, OHC, Room Temperature

Test 1 Load

Test 2 Load

Test 3 Load

Model Load0 0.2 0.4 0.6 0.8 1.0 1.2

Normalized Displacement

Nor

mal

ized

Com

pres

sion

Load

0

0.5

1.0

0

2

4

6

8

10

12

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14

Co

mp

ress

ion

Lo

ad (

kip

s)


T1 Laminate, FHC, Room Temperature

Test 1 LoadTest 2 LoadTest 3 LoadModel Load

0 0.2 0.4 0.6 0.8 1.0 1.2


Nor

mal

ized

Com

pres

sion

Load

0

0.5

1.0

1.5

1.4



0

2

4

6

8

10

12

0 0.02 0.04 0.06 0.08 0.1 0.12

Com

pres

sion

Loa

d (k

ips)




0 0.2 0.4 0.6 0.8 1.0 1.2


Nor

mal

ized

Com

pres

sion

Load

0

0.5

1.0

1.5

-5

0

5

10

15

20

0 0.02 0.04 0.06 0.08 0.1 0.12Com

pres

sion

Loa

d (k

ips)




0 0.2 0.4 0.6 0.8 1.0 1.2


Nor

mal

ized

Com

pres

sion

Load

0

0.5

1.0

Baseline Cure cycle: Load-stroke plots comparing model and test data • OHC and FHC test panels for the T2 laminate [45/90/-45/0]3s



0

5

10

15

20

0 0.02 0.04 0.06 0.08 0.1 0.12

Com

pres

sion

Loa

d (k

ips)




0 0.2 0.4 0.6 0.8 1.0 1.2


Nor

mal

ized

Com

pres

sion

Load

0

0.5

1.0

0

5

10

15

20

25

30

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14

Com

pres

sion

Loa

d (k

ips)




0 0.2 0.4 0.6 0.8 1.0 1.2


Nor

mal

ized

Com

pres

sion

Load

0

0.5

1.0

1.4

Baseline Cure cycle: Load-stroke plots comparing model and test data • OHC and FHC test panels for the T3 laminate [45/0/-45/0/0/45/90/-45/0/0/45/0/-45/0]s



0

2

4

6

8

10

12

0 1000 2000 3000 4000 5000 6000 7000 8000

Com

pres

sion

Loa

d (k

ips)

Strain (µε)

3s

75°F Dry


0 0.2 0.4 0.6 0.8 1.0 1.2

Normalized Strain

0

0.5

1.0

1.4 1.6


Nor

mal

ized

Com

pres

sion

Load

0

2

4

6

8

10

12

14

16

18

0 2000 4000 6000 8000 10000 12000 14000

Com

pres

sion

Loa

d (k

ips)

Strain (µε)

ed o e Co p ess o | [ 5/90/ 5/0]3s

75°F Dry


0 0.2 0.4 0.6 0.8 1.0 1.2

Normalized Strain

0

0.5

1.0

1.4


Nor

mal

ized

Com

pres

sion

Load

Baseline Cure cycle: Load-strain plots comparing model and test data • OHC and FHC test panels for the T2 laminate [45/90/-45/0]3s



• Baseline Cure Cycle: Comparison of ultimate load at failure between model and test, and the % error associated with the model values

Laminate OHC, % Error FHC, % Error

T1 -8.0 -15.5

T2 5.7 -9.8

T3 2.3 -17.7


Compression Strength After Impact Model Baseline Cure Analysis

Outer Section(Single Layer)

ImpactLocation

Delaminationsfrom Impact

Inner Section

(Multiple Layers)

• IM7/M65 Material • 55 ft-lb impact energy • Post impact NDI to determine damage

present – Time of flight scan data shows the depth of

the delamination created during impact • An ABAQUS model was built that matched

what we could see in the NDI data – Multiple delaminations through the thickness – Virtual Crack Closure Technique (VCCT) to

model delam growth during the analysis – ASCA for in-plane damage

Impact LocationDelaminationsfrom Impact

Inner Section (multiple sublaminates through the thickness)

Sublaminate 1

Sublaminate 2

Sublaminate 3

Sublaminate 4

Time of Flight Scan Modeled


Compression Strength After Impact Test Data Overview

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.00 0.20 0.40 0.60 0.80

Nor

mal

ized

Load

Normalized Strain

Far-Field Strain Gauges

Test: SG 1Test: SG 2Test: SG 3Test: SG 4

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.00 0.20 0.40 0.60 0.80 1.00

Nor

mal

ized

Load

Normalized Strain

Near-Field Strain Gauges

Test: SG 5Test: SG 6

Divergence

1

2/3

45/6

Test Fixture(grey)

StrainGauges

Specimen(green)

Applied Load

• Compression strength after impact testing on impacted specimen – 5x7 test fixture – Load and Strain response

measured – Far field gages show a generally

stable response – Divergence of near field gages

indicate a damage propagation during the test


Compression Strength After Impact Initial Correlation to Test Data

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0.00 0.50 1.00 1.50 2.00

Nor

mal

ized

Loa

d

Normalized Strain

Near-Field Back-to-Back

Test: SG 6Model: SG6Test: SG 5Model: SG5

• Initial stiffness correlation to test good

• Ultimate load too high • Back-to-back predicted strains

near the impact do not show divergence when the test does

• Far field predicted strains show similar behavior as the near field predicted strains

1

2/3

4 5/6

CSAI Full Panel

Fixture plate boundary

condition (red)

Test window

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0.00 0.50 1.00 1.50 2.00

Nor

mal

ized

Loa

d

Normalized Strain

Far-Field Back-to-Back

Test: SG 2Model: SG2Test: SG 3Model: SG3


CSAI Model Next Steps

• Need to greatly improve fidelity of NDI and test instrumentation

• Collaborative effort with AFRL/RX – Synergy with existing internal RX program

• AFRL/RX perform detailed Computed Tomography (CT) Scans • Goal to very carefully measure post-impact state • AFRL/RX compression test three specimens and use Digital

Image Correlation (DIC) on both sides of specimen to identify displacement and strain field for improved model correlation

– AFRL/RX to model specimens in BSAM, ICM2 to model in ABAQUS, and compare results


Compression Strength After Impact

• Panel Level Testing will be performed to Validate the Effectiveness of Cure Directed Modeling Efforts

-65°F Dry 70°F Dry 350°F Wet*Short Beam Shear D2344 [0]50 50 1.5"x0.5" 3 3 3

Double Cantilever Beam D5528 [0]26 26 9.0"x1.0" 3 3 3

90°/0° Compression D6641 [90/0]7s 28 5.5"x0.5" 3 3 3

In-Plane Shear D3518 [45/-45]2s 8 9.0"x1.0" 3 3 3

Open Hole Tension D5766 [45/90/-45/0]3s 24 12.0"x1.5" 3 3 3

Filled Hole Tension D5766/D6742 [45/90/-45/0]3s 24 12.0"x1.5" 3 3 3

Open Hole Compression D6484 [45/90/-45/0]3s 24 12.0"x1.5" 3 3 3

Filled Hole Compression D6484/D6742 [45/90/-45/0]3s 24 12.0"x1.5" 3 3 3

Compression Str. After ImpactLMA-PT001

Method 4.30[45/90/-45/0]4s 32 11.0"x13.0" Impact

10.0"x12.0" Comp.3 3 3

Lam

ina

Lam

inat

e

Test ConditionsTest

Test Specification

Layup# of

PliesSpecimen Size


Summary

• Results of the airframe portion of the GE/LM Aero ICM2 program – Studying the IM7/M65 bismaleimide (BMI) system for application to

next generation airframes • Linking general capabilities of three analysis tools

– Convergent COMPRO – Autodesk ASCA – University of Michigan Micromechanics

• Current results include – COMPRO analysis of two cure cycles for the IM7/M65 system – University of Michigan micromechanics analysis of failure strengths

in an RUC for this system – ASCA progressive damage analysis of OHC, FHC, and CSAI

specimens • Integration of these codes will enable:

– Laminate level damage analysis including effects of varying cure cycles


Convergent Process Analysis Material Characterization • Testing was performed using a TA Instruments Q800 Dynamic Mechanical

Analyzer (DMA) using 3-point bend geometry with a 50mm span – During each test, the complex moduli (storage modulus, loss modulus and tan delta) were

measured – Two types of tests were performed; constant frequency temperature ramps and iso-

thermal frequency sweeps. – The constant frequency temperature ramps examine the temperature dependent moduli

and the Iso-thermal frequency sweep tests examine the frequency dependent moduli • A generalized Maxwell model was used to predict the viscoelastic modulus of

HexPly IM7/M65 Material

Temperature

Stor

age

Mod

ulus

Prediction of storage modulus using the developed Maxwell model