| Based on Linux operation system and message passing interface, builded PCcluster system, developed site seismic response analysis program PSAP, packed PSAPand ANSYS, researched calibrating method of seismic response spectrum based ongenetic algorithm, analysed some site seismic response examples. The main researchcontents are listed as follow:①Based on Linux operation system and message passing interface, builded PCcluster system, had a performance test with Linpack for the PC cluster. The result ofperformance test has showed that, parallel computing accelerate ratio of the PC clusteris 1.475, efficiency is 49.2%. The result has indicated that this system could meet therequirements of FEM parallel computing.②Analysed the parallel computing function of ANSYS. Based on PC clustersystem, had a perfomance test of ANSYS parallel computong function with a factdynamic finite elememt example. The test result has indicated that ANSYS in the PCcluster system could not achieve satisfactory performance parallel computing.③Developed dynamic finite element parallel computing platform based onmessage passing environment. In the PC cluster, used the TCP/IP network protocol,based on message passing interface, programmed the site seismic response proceduresPSAP for parallel computing with Fortran and C language. This procedure connectedwith the ANSYS program before processing functions.④Researched calibrating method of seismic response spectrum based on geneticalgorithm. Used the method to calibrate the spectra of Tongji University teaching andresearch building site seismic safety evaluation project.⑤In view of the four common conditions of soil linear and nonlinear seismicresponse analysis: uniform excitation of layered sites; multi-supported excitations oflayered sites; uniform excitation of canyon topography; multi-supported excitations ofcanyon topography. Verificated the result of PSAP through the fact examples, Analysedthe advantages and efficiency to solve large-scale problems in PC cluster.
|
PDF Full Text Request