skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Numerical Stochastic Homogenization Method and Multiscale Stochastic Finite Element Method - A Paradigm for Multiscale Computation of Stochastic PDEs

Abstract

Multiscale modeling of stochastic systems, or uncertainty quantization of multiscale modeling is becoming an emerging research frontier, with rapidly growing engineering applications in nanotechnology, biotechnology, advanced materials, and geo-systems, etc. While tremendous efforts have been devoted to either stochastic methods or multiscale methods, little combined work had been done on integration of multiscale and stochastic methods, and there was no method formally available to tackle multiscale problems involving uncertainties. By developing an innovative Multiscale Stochastic Finite Element Method (MSFEM), this research has made a ground-breaking contribution to the emerging field of Multiscale Stochastic Modeling (MSM) (Fig 1). The theory of MSFEM basically decomposes a boundary value problem of random microstructure into a slow scale deterministic problem and a fast scale stochastic one. The slow scale problem corresponds to common engineering modeling practices where fine-scale microstructure is approximated by certain effective constitutive constants, which can be solved by using standard numerical solvers. The fast scale problem evaluates fluctuations of local quantities due to random microstructure, which is important for scale-coupling systems and particularly those involving failure mechanisms. The Green-function-based fast-scale solver developed in this research overcomes the curse-of-dimensionality commonly met in conventional approaches, by proposing a random field-based orthogonal expansion approach.more » The MSFEM formulated in this project paves the way to deliver the first computational tool/software on uncertainty quantification of multiscale systems. The applications of MSFEM on engineering problems will directly enhance our modeling capability on materials science (composite materials, nanostructures), geophysics (porous media, earthquake), biological systems (biological tissues, bones, protein folding). Continuous development of MSFEM will further contribute to the establishment of Multiscale Stochastic Modeling strategy, and thereby potentially to bring paradigm-shifting changes to simulation and modeling of complex systems cutting across multidisciplinary fields.« less

Authors:
Publication Date:
Research Org.:
Stevens Institute of Technology
Sponsoring Org.:
USDOE Office of Science, Chicago Office
OSTI Identifier:
1036255
Report Number(s):
Final Report
TRN: US201217%%321
DOE Contract Number:  
FG02-06ER25732
Resource Type:
Technical Report
Country of Publication:
United States
Language:
English
Subject:
42 ENGINEERING; 58 GEOSCIENCES; 59 BASIC BIOLOGICAL SCIENCES; 77 NANOSCIENCE AND NANOTECHNOLOGY; BIOTECHNOLOGY; FINITE ELEMENT METHOD; FLUCTUATIONS; GEOPHYSICS; HOMOGENIZATION METHODS; MICROSTRUCTURE; NANOSTRUCTURES; PROTEINS; QUANTIZATION; SIMULATION; Multiscale Stochastic Modeling (MSM), Multiscale Stochastic Finite Element Method (MSFEM)

Citation Formats

Xu, X Frank. Numerical Stochastic Homogenization Method and Multiscale Stochastic Finite Element Method - A Paradigm for Multiscale Computation of Stochastic PDEs. United States: N. p., 2010. Web. doi:10.2172/1036255.
Xu, X Frank. Numerical Stochastic Homogenization Method and Multiscale Stochastic Finite Element Method - A Paradigm for Multiscale Computation of Stochastic PDEs. United States. https://doi.org/10.2172/1036255
Xu, X Frank. Tue . "Numerical Stochastic Homogenization Method and Multiscale Stochastic Finite Element Method - A Paradigm for Multiscale Computation of Stochastic PDEs". United States. https://doi.org/10.2172/1036255. https://www.osti.gov/servlets/purl/1036255.
@article{osti_1036255,
title = {Numerical Stochastic Homogenization Method and Multiscale Stochastic Finite Element Method - A Paradigm for Multiscale Computation of Stochastic PDEs},
author = {Xu, X Frank},
abstractNote = {Multiscale modeling of stochastic systems, or uncertainty quantization of multiscale modeling is becoming an emerging research frontier, with rapidly growing engineering applications in nanotechnology, biotechnology, advanced materials, and geo-systems, etc. While tremendous efforts have been devoted to either stochastic methods or multiscale methods, little combined work had been done on integration of multiscale and stochastic methods, and there was no method formally available to tackle multiscale problems involving uncertainties. By developing an innovative Multiscale Stochastic Finite Element Method (MSFEM), this research has made a ground-breaking contribution to the emerging field of Multiscale Stochastic Modeling (MSM) (Fig 1). The theory of MSFEM basically decomposes a boundary value problem of random microstructure into a slow scale deterministic problem and a fast scale stochastic one. The slow scale problem corresponds to common engineering modeling practices where fine-scale microstructure is approximated by certain effective constitutive constants, which can be solved by using standard numerical solvers. The fast scale problem evaluates fluctuations of local quantities due to random microstructure, which is important for scale-coupling systems and particularly those involving failure mechanisms. The Green-function-based fast-scale solver developed in this research overcomes the curse-of-dimensionality commonly met in conventional approaches, by proposing a random field-based orthogonal expansion approach. The MSFEM formulated in this project paves the way to deliver the first computational tool/software on uncertainty quantification of multiscale systems. The applications of MSFEM on engineering problems will directly enhance our modeling capability on materials science (composite materials, nanostructures), geophysics (porous media, earthquake), biological systems (biological tissues, bones, protein folding). Continuous development of MSFEM will further contribute to the establishment of Multiscale Stochastic Modeling strategy, and thereby potentially to bring paradigm-shifting changes to simulation and modeling of complex systems cutting across multidisciplinary fields.},
doi = {10.2172/1036255},
url = {https://www.osti.gov/biblio/1036255}, journal = {},
number = ,
volume = ,
place = {United States},
year = {2010},
month = {3}
}