Parallel Computing Of Thin-walled Tube Crashworthiness Simulation And Optimization

In crashworthiness structure design of the automobile and other modes oftransport, usually use the ductile metal thin-walled elements to absorb impact energyin the axial impact. For the optimization of structural crashworthiness design, theapplication of test methods are time-consuming and higher equipment requirements,therefore, based on the finite element simulation technology to improve the axialcompression energy absorption has been a hot research topic in the field.Structural crashworthiness analysis is a highly non-linear transient dynamicprocess, it leads to very complex relationship between objective function(energyabsorption value) and design variables (wave heigh and angle) without explicitfunction expression to optimize the design of thin-walled tube.The traditional numerical simulation and optimization of the design need to dealwith large amounts of data and very heavy computing. These computation would takemost of total optimization time, which brought a lot of obstacles for furtherimprovement and design optimization of energy absorbing component.This subject by means of commercial finite element software LS-DYNA andMATLAB, with MPI (Message Passing Interface) environment based on Visual C++platform develop response surface method (RSM) parallel programs and multi-wayevolution genetic algorithm parallel programs of structural optimization. Clusterparallel computing technology used in crashworthiness optimization problem with acorrugated thin-walled tube simulation and optimization solve tasks, to savecomputing costs and greatly improve the computational speed and efficiency. Theresults have important practical value for automotive anti-collision performanceimproved to shorten the development cycle and reduce development costs. How to getthe thin-walled tube best crashworthiness design and how to quickly and efficientlyand cheaply deal with a large amount of data of calculate are the purposes of this topic.the main research contents of this subject are as follows:(1)Taken maximum absorption value as objective function, corrugated andcrosssection dimensions as the design variables, an optimization model of pyramidalfolding patterns thin-walled tube with regular polygon cross-section under axialimpact load is establish by RSM to fit the functional relation between the designvariables and objectives.(2)To study the parallelism of response surface method for each test pointstructure re-analysis; Build an high-speed Internet-based distributed computer cluster,parallel compute structure simulation of each sample points, and more than4.8timesacceleration has Obtained by5processor. (3)Research smart structure optimization algorithms—Genetic Algorithms;Improve the simple genetic algorithm based on multi-way evolution for global searchability and convergence speed.(4) Designed parallel genetic algorithm program and used to solve the abovecrashworthiness optimization problems; Greatly improved computational efficiencyand expand the scale of algorithms for solving such problems, the results show thatthe algorithm have near-linear speedup.(5)Parallel simulation and parallel optimization solution of two types ofcorrugated thin-walled structures, obtained a good improve of crashworthinessperformance, reduceed the computational cycle. The energy absorption of optimizedstructure is increased by more than40%and the design time required for completeone product optimization compared to the serial algorithm, saving more than13hours.