{"title":"FAST METHODS FOR VORTEX INFLUENCE COMPUTATION IN MESHLESS LAGRANGIAN VORTEX METHODS FOR 2D INCOMPRESSIBLE FLOWS SIMULATION","authors":"D. Leonova, I. Marchevsky, E. Ryatina","doi":"10.2495/BE420231","DOIUrl":null,"url":null,"abstract":"Vortex methods are a powerful tool for solving engineering problems of incompressible flow simulation around airfoils. The vorticity is considered as a primary computed variable. According to the Navier–Stokes equations written down in Helmholtz-type form (for 2D case), new vorticity is generated only on the surface line of an airfoil. Its intensity is unknown and can be found from the solution of the boundary integral equation resulting from the no-slip boundary condition satisfaction. The right-hand side of the integral equation in the simplest case depends on the incident flow velocity and vorticity distribution in the flow domain. For the velocity field computation, it is necessary to take into account the influence of all the vortices, which simulate the vorticity distribution in the flow. These vortices are moving in the flow with velocities calculated as sums of point-to-point vortex influences, so the computation complexity of such operation is proportional to 2 N where N is the number of vortices. In practice, N can have the order of tens or hundreds of thousands, up to a million, so the application of the “direct” method for velocities calculation becomes impossible. In this paper, two fast methods having logarithmic (proportional to log N N ) computational complexity are implemented and investigated. The first method is an analogue of the Barnes–Hut fast method for the N-body problem. The second one is based on fast solution of the Poisson’s equation with respect to the stream function on rather coarse mesh by using the fast Fourier transform technique with some special procedure for the results correction. Numerical complexity estimations for both methods are derived. Their sequential and parallel implementations are developed. Numerical experiments show that the FFT-based method is more efficient. Total acceleration compared with the “direct” method is over 1000 times for 500,000 vortex elements.","PeriodicalId":429597,"journal":{"name":"Boundary Elements and other Mesh Reduction Methods XLII","volume":"23 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2019-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Boundary Elements and other Mesh Reduction Methods XLII","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.2495/BE420231","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1
Abstract
Vortex methods are a powerful tool for solving engineering problems of incompressible flow simulation around airfoils. The vorticity is considered as a primary computed variable. According to the Navier–Stokes equations written down in Helmholtz-type form (for 2D case), new vorticity is generated only on the surface line of an airfoil. Its intensity is unknown and can be found from the solution of the boundary integral equation resulting from the no-slip boundary condition satisfaction. The right-hand side of the integral equation in the simplest case depends on the incident flow velocity and vorticity distribution in the flow domain. For the velocity field computation, it is necessary to take into account the influence of all the vortices, which simulate the vorticity distribution in the flow. These vortices are moving in the flow with velocities calculated as sums of point-to-point vortex influences, so the computation complexity of such operation is proportional to 2 N where N is the number of vortices. In practice, N can have the order of tens or hundreds of thousands, up to a million, so the application of the “direct” method for velocities calculation becomes impossible. In this paper, two fast methods having logarithmic (proportional to log N N ) computational complexity are implemented and investigated. The first method is an analogue of the Barnes–Hut fast method for the N-body problem. The second one is based on fast solution of the Poisson’s equation with respect to the stream function on rather coarse mesh by using the fast Fourier transform technique with some special procedure for the results correction. Numerical complexity estimations for both methods are derived. Their sequential and parallel implementations are developed. Numerical experiments show that the FFT-based method is more efficient. Total acceleration compared with the “direct” method is over 1000 times for 500,000 vortex elements.
涡旋方法是解决翼型周围不可压缩流动模拟工程问题的有力工具。涡度被认为是一个主要的计算变量。根据以亥姆霍兹型形式(二维情况下)写下的Navier-Stokes方程,新的涡度仅在翼型的表面线上产生。它的强度是未知的,可以从满足无滑移边界条件的边界积分方程的解中得到。在最简单的情况下,积分方程的右边取决于流域中的入射流速和涡度分布。在速度场计算中,需要考虑所有涡的影响,以模拟流动中的涡度分布。这些涡旋在流动中以点对点涡旋影响的总和计算速度,因此此类操作的计算复杂度与2n成正比,其中N为涡旋的数量。在实际应用中,N的数量级可以是几万、几十万,甚至一百万,所以用“直接”法计算速度就变得不可能了。本文研究了两种计算复杂度为对数(与log N N成正比)的快速算法。第一种方法与n体问题的Barnes-Hut快速方法类似。第二种方法是利用快速傅立叶变换技术对粗网格流函数的泊松方程进行快速求解,并对结果进行特殊的校正。给出了两种方法的数值复杂度估计。开发了它们的顺序和并行实现。数值实验表明,基于fft的方法更有效。50万个涡元的总加速度比“直接”法高1000倍以上。