Two-scale analysis of composite materials using surrogate computational homogenization

Prof Kenjiro Terada

Department of Civil and Environmental Engineering, Tohoku University


Abstract: A class of surrogate computational homogenization (SCH) using radial basis function (RBF) interpolation has recently been proposed to perform two scale analyses of inelastic composite materials. The idea behind the RBF-based SCH is to replace the microscopic analysis in the FE^2 computation with a surrogate model created by radial-basis interpolation approximation using the dataset composed of results obtained obtained from a series of numerical material tests. The method has been applied to composite materials consisting of rate-independent elastoplastic materials within the small and finite strain frameworks, and rate- and temperature-dependent viscoelastic materials, and is guaranteed to be sufficiently accurate. However, the computational cost of the process of obtaining the weights of RBFs by solving linear equations with the kernel matrix as coefficients is high, and this approach has been applied only to two-dimensional (2D) problems. In this study, various sampling methods and interpolation methods will be investigated and their application to 3D problems will be realized. If it can be applied to 3D problems, it will be available to practitioners through implementation in general-purpose FEM software, and it is expected to expand the range of applications, such as topology optimization for anisotropic inelastic materials.


Bio: Dr. Terada is Professor at The Department of Civil and Environmental Engineering. He has published more than 360 peer-reviewed papers in domestic and international journals, four Japanese textbooks, and many other papers in the field of computational mechanics and CAE. He has received numerous awards, including The IACM Fellows Awards from the IACM (2014) and the Kawai Medal from the Japan Society for Computational Engineering and Science (2014). Dr. Terada is currently the President of IACM (2022-Present), and is the former President of JSCES (2016-2017FY). In addition to these various activities in academia, he renders a service on adult education for CAE engineers and practicians as the President of the Japan Association for Nonlinear CAE. 

The main thrust of Dr. Terada's research is computational homogenization methods and related mathematical modeling of heterogeneous materials. Dr. Terada's major contribution in this context is the proposal of a general class of algorithms for nonlinear homogenization, which has been further extended theoretically and applied to various engineering problems. The last decade has also seen a renewed interest in hyper-complex disaster simulation for disaster hazard estimation using various computational methods such as particle methods, data science, etc. At the same time, practical applications of computational homogenization methods to enable practitioners to perform various multi-scale finite element analyses in their CAE work. More recently, research interest has shifted to computational mechanics accelerated by quantum computers.