Extraction of Cohesive Properties of ElastoPlastic material using Inverse Analysis
-
Upload
independent -
Category
Documents
-
view
0 -
download
0
Transcript of Extraction of Cohesive Properties of ElastoPlastic material using Inverse Analysis
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio©2009 Board of Trustees of the University of Illinois
Extraction of Cohesive Properties of Elasto-Plastic material using Inverse
Analysis
Arun Lal Gain, Jay Carroll, Glaucio H. Paulino, John Lambros
University of Illinois at Urbana Champaign
07/18/2009
10th U.S. National Congress on Computational Mechanics
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Contents
Introduction
– Cohesive Zone Modeling
– Elasto-Plastic Forward v/s Inverse Problem
Modeling Approaches
– Shape Regularization
– PPR model
Numerical Simulations
Summary & Conclusions
Future Work
2
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Introduction: Cohesive Zone Model
Cohesive Zone Modeling - fracture seen as phenomenon
of gradual separation taking place across cohesive zone
(path of crack) and resisted by cohesive tractions
Four staged failure
– Stage 1: Material homogeneous
– Stage 2: Crack Initiation
• Criterion: Stress reaching tensile
strength (simplified)
– Stage 3: Crack propagation based on
traction v/s separation curve
– Stage 4: Complete failure
• Criterion: e.g. Crack width reaches
certain predefined value
3
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Introduction: Cohesive Zone Model
4
Various approaches to obtain cohesive zone model are
available in literature
– Obtain through experiments
• Direct tension test
– van Mier, van Vliet, Uniaxial tension test for determination of
fracture parameters of concrete, Fracture of Concrete & Rock,
2002
– Assume the shape
• CZM shape significantly affects fracture analysis results –
should be chosen carefully
– Song S.H, Paulino G.H, Buttlar G.H, Influence of cohesive zone
model shape parameter on asphalt concrete fracture behavior,
American Institute of Physics, 2008
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Introduction: Cohesive Zone Model
5
– Indirect method: Inverse analysis
• Development in experimental stress analysis techniques like
photo-elasticity, DIC have made Inverse Analysis attractive
– van Mier, Fracture processes of concrete : assessment of material
parameters for fracture models, CRC Press, 1997
– Hanson J. H. , An experimental - computational evaluation of the
accuracy of the fracture toughness tests on concrete, PhD Thesis,
Cornell university, 2000
– Hanson J.H., Ingraffea A.R. Using numerical simulations to compare
the fracture toughness values for concrete from the size-effect, two-
parameter and fictitious crack models, Engineering Fracture
Mechanics, 2003
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Introduction: Forward v/s Inverse Problem
6
Forward Problem
Inverse Problem
P P
0 50 100 1500
2
4
6
CMOD(m)
Tn (
Mp
a)
0 0.1 0.2 0.30
200
400
600
CMOD (mm)
Lo
ad
(N
)
Global Response
P P
,x yu
P P
,x yu
DIC / Synthetic Data from forward problem
Optimization
Nelder-Mead Scheme0 50 100 150
0
2
4
6
CMOD(m)
Tn (
Mp
a)
Constitutive Response
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Elasto-Plastic Forward Problem
7
1
1
12
2
element elementcoh coh=K u T
n xf t ld
d N N
1
coh coh+ ,j j j extb i i i i
K u K u u F
1
1
elementcoh
Tc nk t ld
K NN
1 2,
T
x x xd d d
Shen, B., 2009, “Functionally Graded Fiber – Reinforced Cementitious Composites : Manufacturing andExtraction of Cohesive Properties using Finite Elements and Digital Image Correlation”, PhD Thesis, UIUC
Plane Stress J2 Plasticityb K
s
l
2xd
1xd
21
34
Cohesive Element
FF
0 50 100 1500
2
4
6
n (m)
n (
Mp
a)
ck
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Modeling Approaches: Shape Regularization
8
Elasto-Plastic Inverse Problem
1 21 1 2
ext int M
λmin , : R Rf fw w f w f λ F F Y X
1 2 1 2
Cohesive Parameters
X ,X , . .,X ,Y ,Y , . .,Y
coh
n n
λ
Yn c
n
Xn ncn
i
X i - 1 Xi Xi+ 1
Yi X , Yi i coh coh+ , extb K u K u u F
Nelder-Mead Optimization
1
1 1 iψ γ Y10 ,
i
f
Y 2 2
2iψ γ
10i
f
X
1 2
0
0
X X . . Xn
Constraints
1 1
1 1
2
2
i i
X X Xwhere,
X X
γ <<1, ψ >>1
i i i
i
i i
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Modeling Approaches: PPR model
9
Unified potential based model: PPR (Park-Paulino-Roesler)
Park K., Paulino G.H., Roesler J.R., 2009, A unified potential-based cohesive model of mixed-mode fracture,Journal of the Mechanics and Physics of Solids, 57, 891
1
1
ψ , min ,
m
n nn t n t n n t
n n
n
t t
t t n
t t
m
n
ψ
,n n t
n
T
ψ
,t n t
t
T
0 100 200 3000
2
4
6
n (m)
Tn (
Mp
a)
= 1.3
= 2.0
= 4.0
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Modeling Approaches: PPR model
10
Elasto-Plastic Forward Problem
1
1
12
2coh coh=K uelement element T
xf t ld
d N N
1
coh coh+ ,j j j extb i i i i
K u K u λ u F
1
1
elementcoh
Tc nk t ld
K NN
1 2,
T
x x xd d d
s
l
2xd
1xd
21
34
Cohesive Element
FF
0 50 100 1500
2
4
6
n (m)
n (
Mp
a)
ck
coh
max
Cohesive Parameters
, ,n
λ
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Elasto-Plastic Inverse Problem
11
11 1
ext int 3
λmin w , : R Rfw f λ F F
max
Cohesive Parameters
, ,n
λ 1 Constraints:
1
ψ 110 ,f
Barrier Function:
Modeling Approaches: PPR model
coh coh+ , extb K u K u u F
Nelder-Mead Optimization
ψ >> 1
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Problem Details
Numerical Simulations
12
• 5782 Nodes
• 5346 Q4 Elements
• 304 Cohesive Elements
• Displacement Ctrl: 100 Steps
70 0 25
0 14
20 100
GPa, . ,
.
Isotroplic Hardening
MPa, MPa
app
y
E
D mm
K
P P
38.4
150.4
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Forward Problem
Numerical Simulations: Shape Regularization
13
0 0.05 0.1 0.15 0.2 0.250
100
200
300
400
500
600
CMOD (mm)
Lo
ad
(N
)
Elastic bulk material
Elasto-plastic bulk material
0 0.1 0.20
2.5
5
CMOD (mm)
(
MP
a)
Linear softening CZM
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Inverse Problem: Different Loading Points
Numerical Simulations: Shape Regularization
14
0 0.1 0.20
2.5
5
CMOD (mm)
(
MP
a)
Linear softening CZM
• 6 Control Points
• Piecewise Cubic Hermite interpolation
• Synthetic data without any noiseForward Problem Plot
0 0.2 0.4 0.6 0.80
200
400
600
X: 0.3247Y: 170.5
CMOD (mm)L
oa
d (
N)
X: 0.3606Y: 143
X: 0.3964Y: 122.1
X: 0.4322Y: 105.7
X: 0.468Y: 92.52
X: 0.1808Y: 365.5
A
BC DE F
0 0.05 0.1 0.15 0.20
1
2
3
4
5
6
CMOD (mm)
(
MP
a)
Point APoint BPoint CPoint DPoint EPoint F
Initial Guess
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Inverse Problem: Various Control Points
Numerical Simulations: Shape Regularization
15
• Displacement field taken from loading point C = 122.1 N
• Piecewise Cubic Hermite interpolation
• Synthetic data without any noiseForward Problem Plot
0 0.2 0.4 0.6 0.80
200
400
600
X: 0.3964Y: 122.1
CMOD (mm)L
oa
d (
N)
C
0 0.05 0.1 0.15 0.20
1
2
3
4
5
6
CMOD (mm)
(
MP
a)
3 - Control Points4 - Control Points5 - Control Points6 - Control Points
Initial Guess
0 0.1 0.20
2.5
5
CMOD (mm)
(
MP
a)
Linear softening CZM
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
0 0.1 0.2 0.3 0.4 0.50
2
4
6
8
10
CMOD (mm)
(
MP
a)
Inverse Problem: Different Initial Guess
Numerical Simulations: Shape Regularization
16
• Displacement field taken from loading point C = 122.1 N
• 6 Control Points
• Piecewise Cubic Hermite interpolation
• Synthetic data without any noise
Initial Guess 1
Initial Guess 2
0 0.1 0.20
2.5
5
CMOD (mm)
(
MP
a)
Linear softening CZM
Forward Problem Plot
0 0.2 0.4 0.6 0.80
200
400
600
X: 0.3964Y: 122.1
CMOD (mm)L
oa
d (
N)
C
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Inverse Problem: Noise in Synthetic Data
Numerical Simulations: Shape Regularization
17
• Displacement field taken from loading point C = 122.1 N
• 6 Control Points
• Piecewise Cubic Hermite interpolation
• Synthetic data with varying amount of noise
0 0.05 0.1 0.15 0.20
1
2
3
4
5
6
CMOD (mm)
(
MP
a)
Max Noise – 0.00 %
Max Noise – 0.02%
Max Noise – 0.20 %
Max Noise – 2.00 %
Initial Guess
0 0.1 0.20
2.5
5
CMOD (mm)
(
MP
a)
Linear softening CZM
Forward Problem Plot
0 0.2 0.4 0.6 0.80
200
400
600
X: 0.3964Y: 122.1
CMOD (mm)L
oa
d (
N)
C
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Forward Problem
Numerical Simulations: PPR model
18
500
5max
N / m
MPa
= 3
n
0 0.05 0.1 0.15 0.2 0.25 0.3 0.350
100
200
300
400
500
600
CMOD (mm)
Lo
ad
(N
)
Elastic bulk material
Elasto-plastic bulk material
0 0.1 0.2 0.30
2
4
6
CMOD (mm)
(
Mp
a)
= 3.0
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Inverse Problem: Different Loading Points
19
• Synthetic data without any noise
Forward Problem Plot
0 0.1 0.2 0.3 0.40
200
400
600
X: 0.1245Y: 414.2
CMOD (mm)
Lo
ad
(N
)
X: 0.1549Y: 371.6X: 0.1824
Y: 334.9X: 0.2131Y: 296.4X: 0.2439
Y: 260.4X: 0.2749Y: 227.4
AB
CD
EF
0 0.1 0.2 0.3 0.40
1
2
3
4
5
6
CMOD (mm)
(
MP
a)
Initial Guess
Point A
Point B
Point C
Point D
Point E
Point F
Initial Guess
0 0.1 0.2 0.30
2
4
6
CMOD (mm)
(
Mp
a)
= 3.0
Numerical Simulations: PPR model
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Inverse Problem: Various Initial Guesses
Numerical Simulations: PPR model
20
• Displacement field taken from loading point D = 296.4 N
• Synthetic data without any noise
0 0.1 0.2 0.3 0.4 0.50
2
4
6
8
10
CMOD (mm)
(
MP
a)
Initial Guess 2
Initial Guess 1
Initial Guess 3
Forward Problem Plot
0 0.1 0.2 0.3 0.40
200
400
600
X: 0.2131Y: 296.4
CMOD (mm)
Lo
ad
(N
)
D
0 0.1 0.2 0.30
2
4
6
CMOD (mm)
(
Mp
a)
= 3.0
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Inverse Problem: Noise in Synthetic Data
Numerical Simulations: PPR model
21
• Displacement field taken from loading point D = 296.4 N
• Synthetic data with varying amount of noise
0 0.1 0.2 0.3 0.40
1
2
3
4
5
6
CMOD (mm)
(
MP
a)
Initial Guess
Max Noise = 0.00 %
Max Noise = 0.05 %
Max Noise = 0.50 %
Max Noise = 5.00 %
Initial Guess
Forward Problem Plot
0 0.1 0.2 0.3 0.40
200
400
600
X: 0.2131Y: 296.4
CMOD (mm)
Lo
ad
(N
)
D
0 0.1 0.2 0.30
2
4
6
CMOD (mm)
(
Mp
a)
= 3.0
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Load v/s CMOD from experiment
P P
Preliminary results using
PMMA
Inverse Analysis using DIC
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0 200 400 600 800 1000 1200 1400 1600 1800
Load
(kN
)
Crack Opening Displacement (um)
Load vs. COD
Crack Propagation Observed in Images,Image 16
Image 17 DIC Data used for simulation
Load Line Displacement (mm)
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Extracted Cohesive Relation using Inverse Analysis
23
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.040
2
4
6
8
10
12
14
16
18
n, mm
n, M
Pa
Shape of Cohesive Relation
• Results from different
simulation runs
• Shape similar to the
one used in reference
below
Elices M., Guinea G.V., Gomez J. & Planas J., 2002, The Cohesive Zone Model: Advantages, Limitations and Challenges, EFM, 69, 137
Preliminary results using
PMMA
Inverse Analysis using DIC
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Traction Separation relation for PMMA used by Elices et al.
24Elices M., Guinea G.V., Gomez J. & Planas J., 2002, The Cohesive Zone Model: Advantages, Limitations and Challenges, EFM, 69, 137
Inverse Analysis using DIC
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Summary & Conclusions
Developed inverse analysis techniques to extract
cohesive fracture properties of elasto-plastic
materials
– Shape regularization
– PPR model
Verified implementation for various conditions
Ongoing collaborative work: Hybrid technique
(Experimental DIC + Inverse analysis) for
polymers and metal/metal composites such as
Ti/Ti composites
25
07/18/2009 10th U.S. National Congress on Computational Mechanics, Columbus, Ohio
Future Work
Inverse analysis for fatigue loading
Validation of elasto-plastic inverse analysis using
DIC experiments
Extension of elasto-plastic inverse analysis to
plates and shells
26