Data-driven multiscale finite element method: From concurrence to separation

• Published in: Computer methods in applied mechanics and engineering Vol. 363; pp. 112893 - 14
• Main Authors: Xu, Rui, Yang, Jie, Yan, Wei, Huang, Qun, Giunta, Gaetano, Belouettar, Salim, Zahrouni, Hamid, Zineb, Tarak Ben, Hu, Heng
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.05.2020; Elsevier BV; Elsevier
• Subjects: Composite materials; Construction materials; Data points; Data search; Data-driven; Engineering Sciences; Finite element analysis; Finite element method; Homogenization; Materials and structures in mechanics; Mechanics; Mechanics of materials; Multiscale analysis; Multiscale finite element; Solid mechanics; Structural analysis; Structural mechanics; Multiscale finite element; Homogenization; Data-driven
• ISSN: 0045-7825; 1879-2138; 1879-2138
• DOI: 10.1016/j.cma.2020.112893

This paper aims to propose a novel data-driven multiscale finite element method (data-driven FE2) for composite materials and structures. The correlated scales in the classical FE2 method are here split to be computed sequentially and separately: the microscopic problems are calculated in advance to construct an offline material genome database, which is later used in the macroscopic data-driven analysis. In this new framework, the difficulties in formulating and solving complicated multiscale system are avoided by dealing with single scale problems. Moreover, the online computation of microscopic problems in classical FE2 method is replaced by searching data points over the offline database resulting in a data-driven FE2 method. Thus, the online computing efficiency of structural analysis is significantly improved. •A data-driven FE2 method is proposed for composite materials and structures.•Complex multiscale problem is split to simple single scale problems.•The online computational efficiency is significantly improved.