浸没边界-格子玻尔兹曼方法的GPU并行加速  

GPU accelerated immersed boundary-lattice Boltzmann coupling method

作　　者：吴家阳[1] 熊智勇 杨峰 程永光[1] WU Jiayang 1, XIONG Zhiyong 1,2 , YANG Feng 3, CHENG Yongguang 1(1 . State Key Laboratory of Water Resources and Hydropower Engineering Science, Wuhan University,Wuhan 430072, China; 2. China Railway Siyuan Survey and Design Group Co., Ltd., Wuhan 430063, China;3.Construction Management Company for Chushandian Reservoir Project of Henan Province, Zhengzhou 450003, Chin)

机构地区：[1]武汉大学水资源与水电工程科学国家重点实验室,武汉430072 [2]中铁第四勘察设计院集团有限公司,武汉430063 [3]河南省出山店水库建设管理局,郑州450003

出　　处：《中国科技论文》2018年第5期517-521,共5页China Sciencepaper

基　　金：国家自然科学基金资助项目(11172219);高等学校博士学科点专项科研基金资助项目(20130141110013)

摘　　要：为了高效地模拟大规模三维流固耦合问题,提出了一种在GPU平台上并行加速浸没边界-格子玻尔兹曼(immersed boundary-lattice Boltzmann,IB-LB)耦合方法的策略。为了验证准确性和加速效果,计算了不同攻角下的三维平板绕流问题,得到了与实验结果吻合良好的升力和阻力系数随攻角的变化规律,并发现GPU内核函数的性能均受制于带宽,且其带宽利用率均达到70%。为了展示计算能力,模拟了中等雷诺数下三维拍动平板的绕流问题,证明所提方法和加速策略在模拟大规模复杂流固耦合问题上的有效性。To efficiently simulate large-scale three-dimensional（3 D）fluid-structure-interaction（FSI）problems,a GPU parallel implementation of the immersed boundary-lattice Boltzmann coupling method is proposed in this paper.To verify the accuracy and speedup of the proposed method,the 3 Dflow past a flat plate is simulated,and both the lift and drag coefficients for different attack angles are found to be in good agreement with the experimental results.Also,all GPU kernels are found to be bandwidth bound and the achieved memory efficiencies are all above 70%.Finally,to demonstrate the capability of the method,the 3 Dflow past a flapping flat plate under moderate Reynolds number is presented,and the results suggest that the proposed method is effective in simulating large-scale complicated FSI problems.

关 键 词：中央处理器 流固耦合 浸没边界法 格子玻尔兹曼方法 平板绕流 

分 类 号：TV131[水利工程—水力学及河流动力学]