유동해석 코드 T-FLOW의 계산 효율 개선에 관한 연구

Abstract

Since the internal flow of the turbomachinery is so complex, it is hard to get the accurate result in a short period. Parallelization can improve the computational efficiency also. It is well known that multi-block grid system is good for the turbomachinery simulation with complicated geometry and easy to parallelize the serial code without much modification. T-FLOW, the in-house code of the rotating flow laboratory in POSTECH is one of the robust single-block flow solver which was not considered enough to be of the best quality and time efficiency. T-FLOW mainly utilized H-type grid system which has moderate quality near the leading and trailing edge of the blade and limitation in full parallelization. In this study, the MPI parallel method on multi block grid system is applied to T-FLOW solver using the UTRC compressor case. It is found that multi-block and parallelized computation shows outstanding improvement in computational efficiency and grid quality as well. As the number of process increased, the speed-up of calculation surely improved, but limitation might come out from the number of core of hardware.