Computationally Efficient Concurrent Multiscale Framework for the Nonlinear Analysis of Composite Structures

Abstract

This paper presents a computationally efficient concurrent multiscale platform to undertake the nonlinear analysis of composite structures. The framework exploits refined one-dimensional models developed within the scheme of the Carrera unified formulation (CUF), which is a generalized hierarchical formulation that generates refined structural theories via a variable kinematic description. The CUF operates at the macro- and microscales, and the macroscale interfaces with a nonlinear micromechanical toolbox. The computational efficiency derives from the ability of the CUF to obtain accurate three-dimensional (3-D)-like stress fields with a reduced computational cost. The nonlinearity is at the matrix level within the microscale, and its effect scales up to the macroscale through homogenization. The macrotangent matrix adopts a perturbation-based method to have meliorated performances. The numerical results demonstrate that the framework requires some 50% of the computational time and 10% of memory usage of traditional 3-D finite elements. Very detailed local effects at the microscale are detectable, and there are no restrictions concerning the complexity of the geometry.