CPU parallel-based adaptive parameterized level set method for large-scale structural topology optimization

doi:10.1007/s00158-021-03086-9

2022-01-06

PRE

DOI：10.1007/s00158-021-03086-9

原文链接

原文求助

摘要

En 中文

In this paper, an adaptive parameterized level set topology optimization method (APLSM) is proposed by using a bilinear basis function which is common in the finite element method (FEM) to parameterize the level set function. A CPU parallel computing strategy is applied for large-scale topology optimization problems. In the presented APLSM, the solving of an additional linear system, which is necessary for each iteration in the radial basis functions-based parametric level set method (PLSM), is completely avoided since the coefficient matrix of the system degenerates into an identity matrix when using the basis function to parameterize the level set function. In this way, the computational efficiency could be improved significantly. Furthermore, pattern repetition constraints are also imposed to design the graded hierarchical and cellular structures with high geometric complexity by using the proposed APLSM and parallel computing. Several numerical examples have been carried out for verifying the effectiveness of the proposed method. For two-dimensional examples, the computational efficiency is improved about 20% to 80% by comparison with the traditional compactly supported radial basis functions (CSRBFs)-based PLSM with different support radii. For the large-scale three-dimensional problems, the computational efficiency of the proposed method is also much higher. For a plate model meshed by more than 13 million 8-node hexahedral elements, only approximately 2 s in each iteration are consumed in all the calculation processes except for the solving of the structural static equilibrium equations, including the generating of mesh, the assembling of the element stiffness matrices, the imposing of the boundary conditions, and the updating of the level set function.

Keyword:

Structural topology optimization

Parallel computation

Parameterized level set method

Large-scale

Basis function