图形处理器上免组装有限元屈曲分析

发布时间：2018-05-04 23:14

本文选题：屈曲分析 + 免组装有限元　；参考：《西安交通大学学报》2017年05期

【摘要】：为了提高大型三维有限元屈曲分析的速度,克服大规模屈曲拓扑优化的计算速度制约,提出了一种基于免组装有限元的线性屈曲分析算法。针对屈曲分析涉及应力刚度矩阵的特殊性,使用了逆迭代法求解特征值问题;利用体素网格的单元一致性,免组装有限元避免了总体刚度矩阵的组装和存储,减少了计算过程中的内存占用,且有利于并行运算;在图形处理器(GPU)上进行稀疏矩阵与向量乘积的运算,利用并行运算进一步加速了有限元的求解速度。算例结果表明,该算法能有效提高大型三维结构屈曲分析的计算速度,与商用软件Ansys、HyperWorks相比,计算时间可减少60%以上,且随着模型自由度的增加,计算速度提高的程度更加显著。
[Abstract]:In order to improve the speed of buckling analysis and overcome the computational speed constraints of large-scale buckling topology optimization, a linear buckling analysis algorithm based on unassembled finite element is proposed. According to the particularity of stress and stiffness matrix in buckling analysis, the inverse iteration method is used to solve the eigenvalue problem, and the element consistency of voxel mesh is used to avoid the assembly and storage of the total stiffness matrix. It reduces the memory consumption in the computation process and is advantageous to parallel operation. The sparse matrix and vector product are calculated on GPU, and the speed of finite element solution is further accelerated by parallel operation. The numerical results show that the proposed algorithm can effectively improve the computational speed of the buckling analysis of large 3D structures. Compared with the commercial software Ansys HyperWorks, the computational time can be reduced by more than 60%, and with the increase of the degree of freedom of the model, the increase of the computational speed is more significant.
【作者单位】：西北工业大学机电学院;
【基金】：国家自然科学基金资助项目(51375384)
【分类号】：O302

【相似文献】