三维灰体辐射传递系数蒙特卡罗的GPU计算

Monte Carlo for calculating radiative heat transfer coefficient of 3D gray body based on GPU

为了计算漫反射灰体表面之间的辐射传递系数,建立了三维灰立方体模型,在所建立的三维立体空间内表面选取发射表面,利用蒙特卡罗方法生成发射光线,并对光线进行光线追踪,计算发射表面与其他表面之间的辐射传递系数。针对基于蒙特卡罗法计算三维灰体表面之间辐射传递系数时耗时过长的问题,进行了基于显卡GPU实现蒙特卡罗并行计算的研究。利用统一编程架构CUDA对蒙特卡罗法进行了编程实现,采用CUDA随机数发生算法并行处理了求解辐射传递系数中的光线追踪过程,实现了计算辐射传递系数的并行方案。实验结果表明,相对于CPU串行计算方法,基于CUDA架构的并行蒙特卡罗法求解辐射传递系数可以取得高达80倍的计算加速比。

In order to calculate the radiative heat transfer coefficient between the diffuse surface of the 3D gray body,this paper set up a cell cube model to represent the gray surface,and generated rays emitted from one surface(i.e.emission surface)by using Monte Carlo method.It was tracing the multiple reflections for each ray until the ray was totally absorbed.It recorded the absorbed energy at each ray-facet intersection to calculate the heat transfer coefficient between the emission surface and any other surface.To solve the over-consumption issue of radiative heat transfer coefficient based on Monte Carlo method between the diffuse surface of the 3D gray body,this paper used a Monte Carlo method based on GPU to realize the parallel computation,and used the unified programming architecture CUDA to get the realization of programming the Monte Carlo method.According to the CUDA random number generation algorithm,it parallelly computed the tracing process of the radiative transfer coefficient,and realized a parallel scheme for calculating the radiative transfer coefficient.Experimental results show that,compared with the CPU serial calculation method,the parallel Monte Carlo method based on CUDA architecture can obtain the calculated acceleration ratio of 80 times as high as the radiative heat transfer coefficient.