Convergence acceleration by self-adjusted time stepsize using Bi-CGSTAB method for turbulent separated flow computation

Poor convergence behavior is usually encountered when numerical computations on turbulent separated flow are performed. A design of self-adjusted stepsize concept both in time span and spatial coordinate systems to achieve faster convergence is demonstrated in this study. The determination of the time stepsize based on the concept of minimization of residuals using the Bi-CGSTAB algorithm is proposed. The numerical results show that the time stepsize adjusted by the proposed method indeed improves the convergence rate for turbulent separated flow computations using advanced turbulence models in low-Reynolds number forms. Copyright © 2002 John Wiley & Sons, Ltd.