Effective Inelastic Response of Polymer Composites by Direct Numerical Simulations A. Amine Benzerga Aerospace Engineering, Texas A&M University R. Talreja, K. Chowdhury, X. Poulain, A. DeCastro and B. Burgess
Feb 24, 2016
Effective Inelastic Response of Polymer Composites by Direct Numerical Simulations
A. Amine Benzerga
Aerospace Engineering, Texas A&M University
With: R. Talreja, K. Chowdhury, X. Poulain, A. DeCastro and B. Burgess
Background & Motivation
2
Example: Composite blade containment casing for jet engines
Wide range of temperatures (service conditions)
Wide range of strain-rates (design for impact applications)
Ideal for implementing a multiscale modeling strategy:
(i) the material is heterogeneous at various scales;
(ii) the physical processes of damage occur at various scales
Li et al. (JAE, July 2009)
Goal: Develop a strategy aimed at predicting durability of structural components
Basic ingredient: Reliable physics-based inelastic constitutive models
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
July 23rd 2009
Background & Motivation
3
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
July 23rd 2009
Typical Response of a Polymer
4
elastic
hardening
softening
rehardening
T=298K
Compression
510 / s
Epon 862Littel et al (2008)
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
July 23rd 2009
Temperature & Rate sensitivity
5
Effect of Temperature (Epon 862)
The behavior of polymers is temperature and strain-rate dependent
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
July 23rd 2009
Tension -3ε=10 /s
298K
323K
353K
Littel et al (2008)CompressionLittel et al (2008)
- 5ε=10 /s - 3ε=10 /s
- 1ε=10 /s
ε=700/sε=1600/s
Strain-rate effects (Epon 862)
Specification of plastic flow:
e pD D DAssume additive decomposition
5/6
0 exp 1kk e
kk
A sT s
2
:3
p pD D
3
2 de
p
3
:2e d d d b
where and
1 :eD L pD p
Pointwise tensor of elastic moduli Jaumann rate of Cauchy stress
Effective strain rate:
(define direction of plastic flow)
Flow rule:
3
2p
de
D
Effective stress: Deviatoric part of driving stress:
Back stress tensor
Strain rate effects
Material parameters
Describe pressure sensitivity
Internal variable
6
Polymer model
July 2009
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
Modified Macromolecular Model (Chowdhury et al. CMAME 2008)
7
1-ss
ss hs
Nota Bene: Original law(Boyce et al. 1988 )
p
3 ch 8 ch:
1
b R DR R R
Evolution of back stress:
1 21 2
( ) 1 ( ) 1
s ss h hs s
Evolution of athermal shear
strength s :
Polymer Model
July 2009
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
8
2 12
13 ( )
1. , ( ), , 3 1 csch
i j klijkl R ik jl jl ikc c cch
c c
T c cc c c
c c
B BR C N g B g Btr BN
B F F tr BN
L
Material parameter identification
8
• Material parameters :
Elastic constants : ,E
Temperaturesensitivity
Strain-ratesensitivity
Pressuresensitivity
Small strainsoftening
Large strain hardening,
cyclic response
Pre-peak hardening
1 0 , 00 2 3,, , , ,, ,,, , , pR s s hA C Nm s f h Related to inelasticity :
E, n
s0
pεs1
s2
f
h0
CR
N
A, 0ε
h3
Littell et al. (2008)
Reverse flow stress
Forward flow stress
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
July 18th 2009
9
1- Uniaxial tension, compression and torsion tests at fixed strain-rate :
2- Tensile data at various temperatures and strain-rates :
3- s0 is determined from :
4- s1 is determined from : (at lowest temperature at given strain-rate)
5- s2 is determined from : (at lowest temperature at given strain-rate)
6- Large strain compressive response and/or unloading response at fixed strain-rate and temperature :
7- Specific shape of stress-strain curve around peak :
Material parameter identification
0,A
( )( )2(1 )E TT
0 0.077
(1 )s
log
( )ref
ref
ET T
E T
1
0
( )( )
p p
y y
ss
1
2
( )( )
p p
d d
ss
,RC N
0 , ,h f
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
July 18th 2009
10
Model validation
Tension at T=323K
10-1/s
10-3/s
620/s
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
July 18th 2009
11
Model validation
Tension at 10-1/s
T=298KT=323K
T=353K
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
July 18th 2009
12
Model validation
Compression at T=298K
700/s
10-1/s
10-3/s
10-5/s
1600/s
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
July 18th 2009
Numerical Homogenization
13
• Principles of Numerical Simulations :
Unit cell composed of Epon 862 matrix (not optimized set), interface of fixed thickness and carbon fiber
Plane strain conditionsDamage not included
• Objectives :
Investigate evolution of mechanical fields (strains, stresses) in unit-cells
Relate micro/macroscopic behaviors Input for understanding of
onset/propagation of fracture
x1
x2
a
bEpon 862
C fiber
interface
02 1 1
220
220
1 ( , )
ln
a
T x b dxabEb
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
July 18th 2009
14
• Geometries :
Height: b= 100Cell aspect ratio: Ac= 2Fiber volume ratio: Vw =0.1Fiber aspect ratio: Aw=variable
Numerical HomogenizationModel Validation Damage ProgressionNumerical
HomogenizationMaterial Parameter
IdentificationPolymer ModelExperimentsBackground/
Motivation
July 18th 2009
Numerical Homogenization
15
2
2
UM Ft
• Numerical implementation :
Convective representation of finite deformations (Needleman, 1989)
Dynamic principle of Virtual Work:
FEM : Linear displacement triangular elts arranged in quadrilaterals of 4 crossed triangles.
Equations of Motions :
They are integrated numerically by Newmark-B method (Belytshko,1976) in an explicit FE code.
Constitutive updating is based on the rate tangent modulus method of Pierce et al (1984)
2 i
2
ud dS - dt
ij iij i
V S V
V T u V
Kirchhoff stress
Green-Lagrange strain
Surface traction
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
July 18th 2009
16
• Calculations at E22=0.10:
TensionFiber : AS4 (sim. To T700)• Et= 14 GPa• t=0.25
• Geometries :
Height: b= 100Cell aspect ratio: Ac= 2Fiber volume ratio: Vw =0.2 Fiber aspect ratio: Aw=1 (cyl.)
• Dramatic effect of fiber volume ratio on strengthening at all fiber aspect ratios
Numerical HomogenizationModel Validation Damage ProgressionNumerical
HomogenizationMaterial Parameter
IdentificationPolymer ModelExperimentsBackground/
Motivation
July 18th 2009
17
• Calculations at E22=0.10:
CompressionFiber : AS4 (sim. To T700)• Et= 14 GPa• t=0.25
• Geometries :
Height: b= 100Cell aspect ratio: Ac= 2Fiber volume ratio: Vw =0.2 Fiber aspect ratio: Aw=1 (cyl.)
• Plastic strains: Localization and maxima : same as in tension
• Hydrostatic stresses : Building-up in thin ligament between fiber and
edge Aw=6 : proximity of fiber to top surface where
stresses are computed may explain strengthening?
Numerical HomogenizationModel Validation Damage ProgressionNumerical
HomogenizationMaterial Parameter
IdentificationPolymer ModelExperimentsBackground/
Motivation
July 18th 2009
18
Damage ProgressionModel Validation Damage ProgressionNumerical
HomogenizationMaterial Parameter
IdentificationPolymer ModelExperimentsBackground/
Motivation
July 18th 2009
Objective: Develop an experimentally-valided matrix cracking model for use in mesoscale analyses
19
Damage ProgressionModel Validation Damage ProgressionNumerical
HomogenizationMaterial Parameter
IdentificationPolymer ModelExperimentsBackground/
Motivation
July 18th 2009
Finding: Irrespective of the microscopic damage mechanisms, the fracture locus of the polymer matrix is pressure dependent and is temperature-dependent
-5 0 5 10 15 20 25 30 35
-20
0
20
40
60
80
100
120
StrainRate_10e-1eng
StrainRate_10e-3eng
StrainRate_10e-5eng
Maximum local Strain (%)
F/So
(MPa
)
0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.70
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
f(x) = NaN x + NaN
Room Temperature; strain rate 10e-3/s
Notched Bars
Linear (Notched Bars)
Smooth Bars
Stress Triaxiality Ratio
20
TENSION (PMMA)
Benzerga et al. (JAE, 2009)
DEBONDING : 0crit kv v kU U
Asp et al., 1996
Damage ProgressionModel Validation Damage ProgressionNumerical
HomogenizationMaterial Parameter
IdentificationPolymer ModelExperimentsBackground/
Motivation
July 18th 2009
21
COMPRESSION (PMMA) DEBONDING : 0crit kv v kU U
Asp et al., 1996
No debonding :
0kk
Damage ProgressionModel Validation Damage ProgressionNumerical
HomogenizationMaterial Parameter
IdentificationPolymer ModelExperimentsBackground/
Motivation
July 18th 2009
Polymer Fracture Model
22
2 1
, 0
,
c k kI k k
c kk k
k
T
c TT c T
Sternstein et al, 1979Gearing et Anand, 2004
Initiation:micro-void nucleation
fC
Propagation:Drawing of new polymer from active zone
1/
02(1 ( ) )
m
p cr II I
crc
D e es
Gearing et Anand, 2004
1f
Breakdown:Chain scission and disentanglement
c
Element Vanish Tech. of Tvergaard, 1981
Model Validation Damage ProgressionNumerical Homogenization
Material Parameter Identification
Polymer ModelExperimentsBackground/Motivation
July 18th 2009