| ANSYS, a famous commercial finite element software, can perform finite element analysis for most engineering problems. Parallel Virtual Machine (PVM) is a parallel computing software system, which can form a virtual parallel computer with network PCs or other computers and implement network parallel computing. Multilevel decomposition optimization is such a kind of method that according to disciplines or physical components, a large engineering system optimization problem is divided into several independent subsystem optimization problems and one coordinating system optimization problem, and the couplings between subsystems are considered in the coordinating system. A PVM-based subtask dispatching algorithm is designed to effectively solve the load balancing in parallel computing and implement the parallel computing of the two-level optimization. By using VC++ and APDL (ANSYS Parameter Design Language) languages, ANSYS has been integrated into the decomposition optimization procedure in which parallel substructuring analysis is involved. The feasibility and effectiveness of the algorithm are verified by 10-bar and 52-bar truss optimization examples. Also the speedup performance and parallel computing efficiency have been investigated using different number of computers. Parallel implementation of multilevel decomposition optimization and integrated analysis of ANSYS in the optimization shorten the optimization computing time and make the finite element programming unnecessary, thus greatly improve optimization efficiency. A parallel distributed computing platform for decomposition optimization of engineering systems has been developed and described in detail in this paper. In the last part of the paper, a substructure-based multilevel decomposition optimization model is proposed for the truss optimization problem with coordinate variables.
|
PDF Full Text Request