Research And Implementation On Parallel Computing Technology Of Casting Processes Numerical Simulation

With the rapid development of computer technology, parallel computing has been widely applied to various scientific engineering calculations, such as weather forecast, numerical simulation of manufacture. The emergence of parallel technology had greatly improved the computation speed in a very effective way, and brought a large field of parallel processing urgent needs.As we all know, numerical simulation has a huge amout of computation, and the whole simulation processs would spend a lot of time. Some foreign leading numerical simulaition software such as Flow-3D,ProCAST have been applied parallel computing to reduce simulation time. Therefore, it is very important for domestic casting simulation softwares to carry out parallel computing to improve their market competition ability.Based on the Hyper-Threading hardware platform, Visual Studio.Net and Intel C++, OpenMP technology is adapted to implement the parallel computing of the temperature field, gravity feeding and flow field of the casting numerical simulation software.Firstly, this paper studies the parallel computing development status and current trends of casting numerical simulation, and introduces the parallel computing basic knowledges. Based on the current conditions, OpenMP is choosen for parallel computing and Intel C++ as the ultimate development environment, and studying their concrete application.Secondly, the paper briefly introduces the principle of temperature and gravity feeding, and analyses correlation and optimization of the code, after that carries out parallel computing. A final example is given for complete assessment of the parallel computing in the field of temperature and gravity feeding. In the fluid field, the paper gives two pipelined parallel modes of the same surface and the adjacent surface, and compares their strengths and weakness. The paper also does a lot of work for the do-while mode that the OpenMP is not supported. Finally, the combination of loop unrolling and parallel computing shows an excellent performance, and greatly improves the calculation speed.In addition, the paper has done a lot of work on the entire optimization of the numerical simulation software which include import of the Math Kernel Library and the detailed introduction of compiler optimization. Finally, the development of a special test platform HZ_MFC gives us a more convenient and reliable parallel computing performance testing. Practical test shows that the parallel computing of casting processes numerical simulation can greatly increase the speed of calculation, and it has good practical value.