基于边光滑三角形壳元和统一计算架构的板料成形仿真并行计算方法  

Parallel Simulation of Sheet Metal Forming Based on EST Element and Compute Unified Device Architecture

作　　者：蔡勇[1] 王琥[1] 李光耀[1] 崔向阳[1] 郑刚[1] 

机构地区：[1]湖南大学汽车车身先进设计制造国家重点实验室,长沙410082

出　　处：《机械工程学报》2012年第6期32-38,共7页Journal of Mechanical Engineering

基　　金：国家重点基础研究发展计划(973计划;2010CB328005);国家自然科学基金(11002053);湖南省自然科学基金(11JJA001)资助项目

摘　　要：针对板料成形过程仿真中计算效率低以及四边形单元几何逼近性差的问题,提出一种基于边光滑三角形壳元(Edge-based smoothed triangular shell element,EST)和图形处理器(Graphics processing unit,GPU)的板料成形并行计算方法。根据EST壳元及板料成形过程显式求解的特点,该方法采用将最小计算单位与线程一一对应的方式进行数组的求解,同时,采用并行缩减的方法进行单值的求解,实现了整个计算过程的细粒度并行。考虑到GPU并行计算系统的特点,采用由CPU进行主控,由GPU进行数值求解的程序架构,并以统一计算架构(Compute unified device architecture,CUDA)作为GPU编程环境,编制相应的程序。通过算例表明,与传统CPU串行计算方法相比,在计算精度一致的情况下,当计算模型单元数超过20 000个时,基于GPU的并行计算方法可以获得35倍以上的计算加速比,显著减少板料成形仿真的计算时间。A parallel algorithm for sheet metal forming simulation based on the edge-based smoothed triangular shell element（EST） and graphics processing unit（GPU） is designed to meet the challenge of computing efficiency and the meshing difficulties during sheet metal forming simulation.Considering the characteristics of EST element and the explicit solution of sheet forming simulation,the smallest computation units are mapped to the threads and the parallel reduction technique is adopted to calculate the single values to realize the fine-grain parallelism of whole computing processes.Compute unified device architecture（CUDA） is used as a GPU programming environment to program our application.The CPU is used to control the program action and the GPU is used to do the numerical solutions in this application to suit to the characteristics of GPU parallel computing system.The numerical examples show that this parallel computing algorithm based on GPU can dramatically reduce the simulation time of sheet forming that it obtained about 35X speedup compare to the traditional serial algorithm based on CPU with the same computing result.

关 键 词：光滑有限元 板料成形 并行计算 图形处理器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]