2021-06-15T07:39:39Zhttps://ipsj.ixsq.nii.ac.jp/ej/?action=repository_oaipmhoai:ipsj.ixsq.nii.ac.jp:000802642020-01-24T01:21:59Z06164:06165:06242:06667
Development of Explicit Eulerian Finite Difference Solver for Large-Scale Fluid-Structure Interaction SystemsDevelopment of Explicit Eulerian Finite Difference Solver for Large-Scale Fluid-Structure Interaction Systemsengアプリケーションhttp://id.nii.ac.jp/1001/00080264/Conference Paperhttps://ipsj.ixsq.nii.ac.jp/ej/?action=repository_action_common_download&item_id=80264&item_no=1&attribute_id=1&file_no=1Copyright (c) 2012 by the Information Processing Society of JapanThe University of TokyoFujitsu Nagano Systems Engineering Ltd.RikenRikenThe University of TokyoThe University of Tokyo／RikenThe University of TokyoRikenKazuyasu, SugiyamaYasuhiro, KawashimaHiroshi, KoyamaShigeho, NodaSatoshi, IiShu, TakagiYoichiro, MatsumotoRyutaro, HimenoA scalable numerical algorithm has been reconsidered for massively parallel computations of fluid-structure interaction systems as biological applications. A new Eulerian method using a fixed mesh has been developed to solve the basic equation set for the incompressible Newtonian fluid and hyperelastic material in a finite difference manner. A new artificial compressibility method, corresponding to one of full explicit time-stepping algorithms, with adaptive parameters is proposed. The advocated solver easily attains excellent scalability since it makes the workload on each core equivalent and reduces the amount of node-to-node communication required for the iterative computation. It is applied to wall-bounded flows with biconcave particles, which replicate the shape of red blood cells. The computational performance on a Xeon cluster is presented in terms of a weak scaling and also of a strong scaling with O(109) grid points up to 8,192 cores.A scalable numerical algorithm has been reconsidered for massively parallel computations of fluid-structure interaction systems as biological applications. A new Eulerian method using a fixed mesh has been developed to solve the basic equation set for the incompressible Newtonian fluid and hyperelastic material in a finite difference manner. A new artificial compressibility method, corresponding to one of full explicit time-stepping algorithms, with adaptive parameters is proposed. The advocated solver easily attains excellent scalability since it makes the workload on each core equivalent and reduces the amount of node-to-node communication required for the iterative computation. It is applied to wall-bounded flows with biconcave particles, which replicate the shape of red blood cells. The computational performance on a Xeon cluster is presented in terms of a weak scaling and also of a strong scaling with O(109) grid points up to 8,192 cores.ハイパフォーマンスコンピューティングと計算科学シンポジウム論文集20121531622012-01-172012-01-16