Convergence acceleration by self-adjusted relaxation factor for turbulent separated flow computation

Wen Bin Tsai, Wen-Wei Lin, Ching Chang Chieng

Research output: Contribution to conferencePaperpeer-review

Abstract

The applications of higher-order Turbulent models for complex turbulent flow compulation has been widely recognized and becom es popular, but slow convergence (even divergence) of the turbulent com putations still hinders the use of advanced turbulence m odels. In this study, we propose a method to detenu ine the values of the under relaxation factors so that faster convergence can be achieved. The new m ethod adopts the concept from the Bi-CGSTAB algorithm. The schem e updates the solution vectors such that the 2-norm m inin ized residuals of the governing equations can be obtained. In the present work, we will derive an expression according to BiCGSTAB method so that the under relaxation factors of the computation cellata specific iteration step can be obtained to enforce tie residual of tie conservative properties of tie com pwtatvnal cell team inin urn value, say zero. The turbulen tflow pasta ribbed channel with (or wilhout) periodic boundary conditions is com puted to dem onstrate the effectiveness of the present m ethod. The num erical results indicate that the under relaxation factors selfadjisted by the Bi-CGSTAB method indeed accelerate the convergence rate for the computation of turbulent separated flow.

Original languageEnglish
StatePublished - 1 Dec 2004
EventEuropean Congress on Computational Methods in Applied Sciences and Engineering, ECCOMAS 2004 - Jyvaskyla, Finland
Duration: 24 Jul 200428 Jul 2004

Conference

ConferenceEuropean Congress on Computational Methods in Applied Sciences and Engineering, ECCOMAS 2004
CountryFinland
CityJyvaskyla
Period24/07/0428/07/04

Keywords

  • Bi-CGSTAB method
  • Convergence acceleration
  • Under relaxation factor

Fingerprint Dive into the research topics of 'Convergence acceleration by self-adjusted relaxation factor for turbulent separated flow computation'. Together they form a unique fingerprint.

Cite this