efficient methods for homogenization of random ... · homogenization determine effective...
TRANSCRIPT
![Page 1: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/1.jpg)
Efficient Methods for Homogenization of Random Heterogeneous Materials
Yufeng NieDepartment of Applied Mathematics
Northwestern Polytechnical University
Department of Applied MathematicsNorthwestern Polytechnical University
Joint with Yatao Wu, Junzhi Cui
Workshop on Modeling and Simulation of Interface-related ProblemsInstitute for Mathematical Sciences :: National University of Singapore
![Page 2: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/2.jpg)
Background information
Improving convergence rate with Richardson extrapolation
Robin boundary condition for high-contrast materials
Contents
Methods for Homogenization of RHMs
Robin boundary condition for high-contrast materials
Conclusions
2
![Page 3: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/3.jpg)
Background information
Improving convergence rate with Richardson extrapolation
Robin boundary condition for high-contrast materials
Contents
Methods for Homogenization of RHMs
Robin boundary condition for high-contrast materials
Conclusions
3
![Page 4: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/4.jpg)
Multiscale problem
Many practical problems have multiple-scale feature.
Composite materials
Porous media
Methods for Homogenization of RHMs
(a) Macroscopic domain
(b) Composites at the microscale
(c) Porous media at the microscale
4
![Page 5: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/5.jpg)
Multiscale problem
BVP with random rapidly oscillating coefficients
, ,
,
da x u x f x D
u x g x D
in
on
Methods for Homogenization of RHMs 5
![Page 6: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/6.jpg)
Multiscale problem
Methods for Homogenization of RHMs 6
Macroscopic behavior of the heterogeneous material may
be approximated by that of a fictitious homogeneous one.
![Page 7: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/7.jpg)
Multiscale problem
Typical multiscale methods for the BVP
Asymptotic homogenization method (Lions 1978)
Variational multiscale method (Hughes 1995)
Methods for Homogenization of RHMs 7
Multiscale finite element method (Hou & Efendiev 1997)
Heterogeneous multiscale method (E & Engquist 2003)
![Page 8: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/8.jpg)
Multiscale problem
Homogenized BVP
,
da u x f x D
u x g x D
0
0
in
on
Methods for Homogenization of RHMs 8
![Page 9: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/9.jpg)
Homogenization
Methods for Homogenization of RHMs 9
, , 0 0,ind
La y u y Y L
![Page 10: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/10.jpg)
Homogenization
Auxiliary problem in volume element
Dirichlet boundary condition (DBC)
, , 0 La y u y Y in
Methods for Homogenization of RHMs 10
Dirichlet boundary condition (DBC)
Neumann boundary condition (NBC)
, , 1, ,i Lu y y Y i d on
, , , 1, ,i La y u y n n Y i d on
![Page 11: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/11.jpg)
Homogenization
Determine effective coefficients
Create microstructure
Solve boundary value problems
Compute relevant volume averages
Methods for Homogenization of RHMs
Figure: The main procedure of multiscale method for random heterogeneous materials
11
Homogenization
Simulations in the macro-scale
Localization
Determine effective coefficients in microstructures of different
sizesCalculate effective coefficients
Repeat above procedure in different realizations, get ensemble average
![Page 12: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/12.jpg)
Background information
Improving convergence rate with Richardson extrapolation
Robin boundary condition for high-contrast materials
Contents
Methods for Homogenization of RHMs
Robin boundary condition for high-contrast materials
Conclusions
12
![Page 13: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/13.jpg)
Motivation
Methods for Homogenization of RHMs 13
![Page 14: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/14.jpg)
Motivation
Sufficiently large volume elements are always needed,
which is time-consuming and high-demanding for computer
memory.
Methods for Homogenization of RHMs 14
![Page 15: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/15.jpg)
Method
Dirichlet boundary condition
It satisfies
, , 1, ,i Lu y y Y i d on
Methods for Homogenization of RHMs
Approximate effective coefficients are calculated by
15
![Page 16: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/16.jpg)
Method
Neumann boundary condition
It satisfies
, , , 1, ,i La y u y n n Y i d on
Methods for Homogenization of RHMs
Approximate effective coefficients are calculated by
16
![Page 17: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/17.jpg)
Method
Theorem (Wu, Nie, Yang 2014)
Methods for Homogenization of RHMs 17
![Page 18: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/18.jpg)
Method
Assume the first-order convergence rates of DBC and NBC approximations are satisfied, then we have
Since
,
1 1, 1whereBC La a C O
L L
Methods for Homogenization of RHMs
We get Richardson extrapolation sequence
18
Since
,2
1 1
2 2BC La a C O
L L
, ,2 ,
12BC L BC L BC LR a a a a O
L
![Page 19: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/19.jpg)
Method
Determine effective coefficients
Create microstructure
Solve boundary value problems
Compute relevant volume averages
Richardson extrapolation
Methods for Homogenization of RHMs
Figure: The main procedure of multiscale method combined with Richardson extrapolation technique
19
Homogenization
Simulations in the macro-scale
Localization
Determine effective coefficients in microstructures of different
sizesCalculate effective coefficients
Repeat above procedure in different realizations, get ensemble average
![Page 20: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/20.jpg)
Numerics
Microstructure (RMDF) random morphology description function,
proposed by Vel et al. in 2010. Closely resemble actual micrographs manufactured by
techniques including plasma spraying and powder processing.
Methods for Homogenization of RHMs 20
processing.
(a) Microstructures generated by computer with different volume fractions of Al
(b) Actual Al/Al2O3
micrographs
![Page 21: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/21.jpg)
Numerics
ExampleWe predict effective thermal conductivity of actual Al/Al2O3
with different volume fractions of Al.
Methods for Homogenization of RHMs 21
Figure: Microstructures with different sizes
![Page 22: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/22.jpg)
Numerics
Material Al Al2O3
k (W/mK) 233.0 30.0
Table: Properties of constituent materials
Methods for Homogenization of RHMs 22
VAl (%) 10 30 50 70 90
DBC 1.08 0.93 0.95 0.99 1.04
NBC 1.02 0.99 0.82 0.95 0.98
Table: Convergence rates of approximate effective coefficients
![Page 23: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/23.jpg)
Numerics
Methods for Homogenization of RHMs 23
Figure: Comparison of approximate effective coefficients by homogenizationand extrapolation with increasing volume element size
(a) VAl=30 % (b) VAl=70 %
![Page 24: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/24.jpg)
Numerics
Methods for Homogenization of RHMs 24
Figure: Comparison of Homogenization, Extrapolation and Variational bounds(Voigt-Reuss bounds and Hashin-Shtrikman bounds)
![Page 25: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/25.jpg)
Numerics
Methods for Homogenization of RHMs 25
Figure: Comparison of Homogenization, Extrapolation and Variational bounds (Voigt-Reuss bounds and Hashin-Shtrikman bounds)
![Page 26: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/26.jpg)
Numerics
Methods for Homogenization of RHMs 26
Figure: Comparison of Homogenization, Extrapolation and Variational bounds (Voigt-Reuss bounds and Hashin-Shtrikman bounds)
![Page 27: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/27.jpg)
Numerics
Methods for Homogenization of RHMs 27
Figure: Comparison of Homogenization, Extrapolation and Variational bounds(Voigt-Reuss bounds and Hashin-Shtrikman bounds)
![Page 28: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/28.jpg)
Numerics
Methods for Homogenization of RHMs 28
Figure: Comparison of Homogenization, Extrapolation and Variational bounds(Voigt-Reuss bounds and Hashin-Shtrikman bounds)
![Page 29: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/29.jpg)
Numerics
Example
We predict effective mechanical properties of actual Al/Al2O3
with different volume fractions of Al.
The contrast ratio between the elastic modulus of Al2O3 and Al
Methods for Homogenization of RHMs 29
is 5.6.
Table: Properties of constituent materials
Material Al Al2O3
E (GPa) 70 393
0.30 0.22
![Page 30: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/30.jpg)
Numerics
Methods for Homogenization of RHMs 30
Figure: Comparison of Homogenization, Extrapolation and Variational bounds(Voigt-Reuss bounds and Hashin-Shtrikman bounds)
![Page 31: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/31.jpg)
Numerics
Methods for Homogenization of RHMs 31
Figure: Comparison of Homogenization, Extrapolation and Variational bounds (Voigt-Reuss bounds and Hashin-Shtrikman bounds)
![Page 32: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/32.jpg)
Numerics
Example
We predict effective mechanical properties of actual stainless
steel/epoxy with different volume fractions of stainless steel.
The contrast ratio between the elastic modulus of stainless
steel and epoxy is high (more than 100).
Methods for Homogenization of RHMs 32
steel and epoxy is high (more than 100).
Table: Properties of constituent materials
Material Stainless Steel (SS) Resin
E (GPa) 193.8 1.31
0.29 0.40
![Page 33: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/33.jpg)
Numerics
Methods for Homogenization of RHMs 33
Figure: Comparison of Homogenization, Extrapolation and Variational bounds(Voigt-Reuss bounds and Hashin-Shtrikman bounds)
![Page 34: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/34.jpg)
Numerics
Methods for Homogenization of RHMs 34
Figure: Comparison of Homogenization, Extrapolation and Variational bounds (Voigt-Reuss bonds and Hashin-Shtrikman bounds)
![Page 35: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/35.jpg)
Background information
Improving convergence rate with Richardson extrapolation
Robin boundary condition for high-contrast materials
Contents
Methods for Homogenization of RHMs
Robin boundary condition for high-contrast materials
Conclusions
35
![Page 36: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/36.jpg)
Motivation
Many random heterogeneous materials have a high
contrast of constituent properties.
a. Composite materials (e.g., CNT-reinforced polymers)
b. Porous media
Methods for Homogenization of RHMs
b. Porous media
The high contrast leads to very broad Dirichlet-Neumann
upper and lower bounds.
36
![Page 37: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/37.jpg)
Motivation
Mixed Dirichlet-Neumann boundary condition (DNBC)
approximations lie within the Dirichlet-Neumann bounds. (Yue
2007)
1, i Lu y y Y on
Methods for Homogenization of RHMs 37
1
2, , 0i L
La y u y n Y on
![Page 38: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/38.jpg)
Method
Methods for Homogenization of RHMs 38
Continuous properties
Figure: Volume elements with different sizes
![Page 39: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/39.jpg)
Method
Robin boundary condition (RBC)
, , , i i La y u y n u y n y Y on
Methods for Homogenization of RHMs 39
Theorem (Convergence)
![Page 40: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/40.jpg)
Method
Determine effective coefficients
Create microstructure
Solve boundary value problems
Compute relevant volume averages
Robin boundary condition
Methods for Homogenization of RHMs
Figure: The main procedure of multiscale method combined with Robin boundary condition
40
Homogenization
Simulations in the macro-scale
Localization
Determine effective coefficients in microstructures of different
sizesCalculate effective coefficients
Repeat above procedure in different realizations, get ensemble average
![Page 41: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/41.jpg)
Numerics
Example
We consider effective coefficients of random checker-board
microstructures. Each unit cell is occupied by matrix material
or reinforcement material with probability p and 1-p,(0<p<1).
Here we set p=0.5.
Methods for Homogenization of RHMs 41
Here we set p=0.5.
Figure: Random checker-board microstructures with different sizes
(a) L=8 (b) L=16 (c) L=24
![Page 42: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/42.jpg)
Numerics
For two-phase random heterogeneous materials,
has a high contrast, i.e.,
Methods for Homogenization of RHMs
Here, I is the identity matrix.
42
has a high contrast, i.e.,
![Page 43: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/43.jpg)
Numerics
Methods for Homogenization of RHMs 43
(a) r = 10 (b) r = 100
![Page 44: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/44.jpg)
Numerics
Methods for Homogenization of RHMs 44
Figure: Effect of adjusting factor in RBC on the accuracy of approximate effective coefficients
![Page 45: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/45.jpg)
Numerics
Methods for Homogenization of RHMs 45
Figure: Approximate effective coefficients with different adjusting factor in RBC
(a) r = 10 (b) r = 100
![Page 46: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/46.jpg)
More Discussion
High contrast leads to large condition number of stiffness
matrix, which may reduce the numerical accuracy of
approximate effective coefficients.
We discuss the effect of condition number on the accuracy
Methods for Homogenization of RHMs 46
of approximate effective coefficients.
Auxiliary problem with DBC
, , 0
, , 1, ,
in
on
L
i L
a y u y Y
u y y Y i d
![Page 47: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/47.jpg)
More Discussion
Methods for Homogenization of RHMs 47
Figure: Effect of contrast ratio and mesh size on the condition number of stiffnessmatrix (constituent with high property in the center)
![Page 48: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/48.jpg)
More Discussion
Methods for Homogenization of RHMs 48
Figure: Effect of contrast ratio and mesh size on the condition number of stiffness
matrix (constituent with low property in the center)
![Page 49: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/49.jpg)
More Discussion
CGMICCG
M
244721168
29325
333061590
36629
431062076
44033
CGM218510 474 289075 612 363644 768
Methods for Homogenization of RHMs 49
Largest condition number is no more that 106, while double-precision floating-point system guarantees more than 15 significant digits of freedom.
Condition number has little effect on the accuracy of approximate effective coefficients.
CGMICCG
M
21851010325
47429
28907513517
61233
36364416902
76837
![Page 50: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/50.jpg)
Background information
Improving convergence rate with Richardson extrapolation
Robin boundary condition for high-contrast materials
Contents
Methods for Homogenization of RHMs
Robin boundary condition for high-contrast materials
Conclusions
50
![Page 51: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/51.jpg)
Conclusion
Richardson extrapolation is an effective technique for
predicting effective coefficients of random heterogeneous
materials.
Methods for Homogenization of RHMs
Since smaller volume elements are used, a lot of
computation time and computer memory could be saved.
51
![Page 52: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/52.jpg)
Conclusion
Robin boundary condition is proposed for predicting effective
coefficients of random heterogeneous materials.
It provides much better approximate effective coefficients
than other boundary conditions.
Methods for Homogenization of RHMs
than other boundary conditions.
It is more flexible than other boundary conditions because
of the adjusting factor.
It is more suitable for random heterogeneous materials
with high contrast.
52
![Page 53: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/53.jpg)
Reference
1. Y.T. Wu, Y.F. Nie and Z.H. Yang, Prediction of effective properties for random heterogeneous materials with extrapolation, Archive of Applied Mechanics, 2014, 84(2): 247-261.
2. Y.T. Wu, J.Z. Cui, Y.F. Nie, et al., Predicting effective elastic moduli and strength of ternary blends with core-shell structure by second-order two-scale method, CMC-Computers, Materials & Continua, 2014, 42(3): 205-225.
3. Y.T. Wu, Y.F. Nie and Y. Zhang, A new boundary condition for homogenization of high-
Methods for Homogenization of RHMs
3. Y.T. Wu, Y.F. Nie and Y. Zhang, A new boundary condition for homogenization of high-contrast random heterogeneous materials, International Journal of Computer Mathematics,2016,93(12):2012-2027.
4. A. Bensoussan, J.L. Lions, and G. Papanicolaou, Asymptotic Analysis for Periodic Structures, Amsterdam: North-Holland Publishing Company, 1978.
5. T.J.R. Hughes, Multiscale phenomena: Green's functions, the Dirichlet-to-Neumann formulation, subgrid scale modles, bubbles and the origins of stabilized methods, Computer Methods in Applied Mechanics and Engineering, 1995, 127: 387-401.
6. T.Y. Hou and X.H. Wu, A multiscale finite element method for elliptic problems in composite materials and porous media, Journal of Computational Physics, 1997, 134(1): 169-189.
53
![Page 54: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/54.jpg)
Reference
7. W.N. E and B. Engquist, The heterogeneous multiscale methods, Communications in Mathematical Sciences, 2003, 1(1): 87-132.
8. A. Bourgeat and A. Piatnitski, Approximations of effective coefficients in stochastic homogenization, Annales De L Institut Henri Poincare-Probabilites Et Statistiques, 2004, 40(2): 153-165.
9. S.S. Vel and A.J. Goupee, Multiscale thermoelastic analysis of random heterogeneous materials i: Microstructure characterization and homogenization of material properties,
Methods for Homogenization of RHMs
materials i: Microstructure characterization and homogenization of material properties, Computational Materials Science, 2010, 48(1): 22-38.
10. R. Ewing, O. Iliev, R. Lazarov, et al., A simplified method for upscaling composite materials with high contrast of the conductivity, Siam Journal on Scientific Computing, 2009, 31(4): 2568-2586.
11. X.Y. Yue and W.N. E, The local microscale problem in the multiscale modeling of strongly heterogeneous media: Effect of boundary conditions and cell size, Journal of Computational Physics, 2007, 222(2): 556-572.
54
![Page 55: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/55.jpg)
![Page 56: Efficient Methods for Homogenization of Random ... · Homogenization Determine effective coefficients Create microstructure Solve boundary value problems Compute relevant volume averages](https://reader035.vdocuments.us/reader035/viewer/2022062918/5edcc511ad6a402d666794e7/html5/thumbnails/56.jpg)
Thank you for your attention !
56