首页 | 本学科首页   官方微博 | 高级检索  
     检索      


A method of finding unstable steady solutions by forward time integration: relaxation to a running mean
Institution:1. Professor, Graduate school of Engineering, Hiroshima University, 1-4-1 Kagamiyama, Higashi-hiroshima city, 739-8527, Japan;1. Visualization Research Center, University of Stuttgart, Allmandring 19, D-70569 Stuttgart, Germany;2. Institute for Numerical Simulation, University of Bonn, Wegelerstr. 6, D-53115 Bonn, Germany;3. Interdisciplinary Center for Scientific Computing, Heidelberg University, Im Neuenheimer Feld 205, D-69120 Heidelberg, Germany;4. IPVS/SGS, University of Stuttgart, Universitätsstraße 38, D-70569 Stuttgart, Germany;5. Multiscale Meshfree Methods, Fraunhofer SCAI, Schloss Birlinghoven, 53754 Sankt Augustin, Germany
Abstract:Proposed is a method of finding unstable steady solutions based on relaxation to a running mean. It works for moderate supercriticality, when the growth rates of the instabilities do not exceed their frequencies. To calculate the running mean one should use the decaying exponential function as a kernel in the convolution formula, it offers a combination of necessary stabilization properties with the efficiency of numerical implementation. The method requires specification of the relaxation TR and averaging TA time scales, such that in the transformed eigenspectrum frequencies above the cutoff 1/TA experience additional damping with 1/TR rate. All originally decaying modes remain decaying. The zero frequency remains unchanged, which implies that the steady states are invariant under the transformation. The acceptable values for TR and TA can be obtained from an estimate of the frequencies and growth rates of the modes leading to instability. For faster convergence to the steady state TR should be set as large as possible, provided it causes sufficient damping of the instability. The numerical implementation of this method requires only very simple modifications to the program used for time integration of the original system.
Keywords:
本文献已被 ScienceDirect 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号