Font Size: a A A

Realization Of High Performance Computing Method For Dynamic Analysis Of Isolated Structure

Posted on:2020-10-13Degree:MasterType:Thesis
Country:ChinaCandidate:S YangFull Text:PDF
GTID:2392330596977680Subject:Architecture and civil engineering
Abstract/Summary:PDF Full Text Request
With the enhancement of people's safety consciousness and the improvement of living standard,more and more buildings are applied to the isolation technology,but the design of isolation structure is mainly time-history analysis,when the upper structure is complex and the number of supports is high,the existing commercial software is slow in nonlinear time-history analysis,and the isolation design needs multiple trial and error to complete.It greatly affects the design efficiency and design quality.Therefore,this paper takes OpenSees as the platform and uses GPU high performance calculation method to realize the dynamic analysis and isolation structure design of isolation structure based on high performance GPU calculation.The main research contents and conclusions of this paper are as follows:(1)Two kinds of nonlinear time-history analysis methods,fast nonlinear analysis method(FNA method)and direct integration method,which are commonly used at present,are selected,and the calculation accuracy and calculation speed of the two algorithms are compared.Taking a typical multi-layer isolation structure as an example,it is found that the exact solution method of nonlinear time history analysis is direct integration method,but the calculation speed is slow,which is not convenient for practical design.FNA method is fast,but If the number of selected Ritz modes is insufficient,the result obtained by the FNA method has a large error.Explain that the calculation software based on a single CPU platform can not improve the calculation efficiency while ensuring the calculation accuracy,and further select the suitable CPU/GPU isolation structure design platform.(2)In order to improve the calculation speed of the dynamic analysis of the isolated structure,the finite element software OpenSees is used as the GPU accelerated computing platform,and the CUDA heterogeneous CPU and GPU collaborative computing platform is used to build the GPU acceleration environment of OpenSees,which describes how to perform between the CPU and the GPU.Data interaction completes the accelerated solution of linear equations.Since the OpenSees model is created in Tcl script mode,this paper gives the method flow and precautions for converting ETABS model into Tcl script file for the convenience of designers.Tested by engineering examples,the results show that the reliability of the built-in GPU-based sparse matrix solver,and the direct integration method,compared to ETABS software,OpenSees isolation structure analysis using GPU acceleration can save nearly half of the time.(3)In order to improve the design efficiency of the isolated structure and ensure the quality of the support arrangement.The realization process of high-performance computing isolation structure design is introduced.The two-stage optimization method is used to arrange the isolation structure of the isolation structure.The optimization model of each stage is established respectively.The first stage optimization combines the multi-group genetic algorithm with the GPU-based OpenSees finite element software.The second stage optimization uses the analysis results of the previous stage,and the linear programming gives the actual branch Plan layout.(4)An actual isolated structure is selected.Through the design method of isolated structure based on high performance calculation,the optimized support arrangement is given,and the isolation effect is compared with the original traditional method.In addition to better economics,the new support layout method is more assured in terms of accuracy and efficiency.
Keywords/Search Tags:isolated structure, high performance computing, Genetic algorithm, OpenSees, GPU
PDF Full Text Request
Related items