Parallel Software Development On Temperature Field Simulation Of Metal Solidification Process

Numerical simulation of temperature field in solidification process in foundry is the basis on predicting solidification shrinkage, gas porosity, lack of fill, laps, hot cracks, macrosegregation and microsegregatio. However, it needs large computing time when simulating the solidification process's temperature changes of a large-scale and shape-complicated casting, but the numerical simulation based on Von Neuman's model can not boost the computing time further. Therfor, an effective and practical way to enhance computing speed is to apply parallel computing technology based on cluster.The parallel hardware environment this paper adopted is isomeric cluster platform which consist of three common PC with defferent architecture in LAN, each node runs Windows 2000 Professional, and its configuration is as follows: (1) Celeron CPU 2.66GHz, memory 256M; (2) Pentium 4 CPU 2.00GHz, memory 256M; (3) Pentium 4 CPU 1.60GHz ,memory 256M. And the parallel software environment adopted is realized by combining Visual C++6.0 with MPICH2 which is the up-to-date edition of standardized MPI message passing parallel library.As FDM is simple and is applied widely in numerical simulation of solidification process, this paper adopted it to divide three-dimensional grid of computing field and construct difference scheme of differential equation of heat conduction.For mapping solidification simulation problem to the isomeric cluster which has been constructed and getting a parallel computing model to design a good parallel algorithm, this paper adopted domain decomposition method to decompose the entire computing data domain as some subdomains, and each subdomain is distributed to one node as single computing task. And that each node has equal status and almost the same code and function, namely that the method of MPI parallel programming in equal pattern is taken.This paper discussed the selection of MPI communication mode and of parallel I/O access mode in process of parallel programming with MPI..Finally, as based on the parallel environment constructed, this paper tested the efficiency of the MPI parallel algorithm designed about temperature field in solidification process of a T-shape casting with groups of different node's amount and different element's amount.As showed by testing result, the simulative computation which is parallelized with MPI accelerates the simulation speed effectively, and simulation speed is enhanced as the amount of node is increased when the computing scale is the same, and the accelerated ratio's increment of parallel computing from 2 nodes to 3 nodes is increased as computing scale is augmented.On the whole, MPI parallel programming simulation is a effective way to improve the speed of computer simulation in casting simulation process. And also it has practical value from the point of view of software industry's development trend and user's satisfaction.