N. Zhang (PRC)
SIP solver, TDMA solver, Verification, Convergence
This paper compares three different approaches: SIP (Strongly Implicit Procedure), TDMA-SOR (Tri Diagonal Matrix Algorithm, combined with Successive Over Relaxation) and ICCG (Conjugate Gradient method, preconditioned using Incomplete Cholesky decomposition), by using three typical fluid flows: closed surface, partial opening and full opening. The results showed that SIP has better stability and convergence for the flow with Dirichlet momentum boundary conditions of outlet than that with Neumann momentum boundary conditions of outlet. The reason is that SIP is designed specially for algebraic equations, which are discretizated from partial differential equations, and it may not be used for generic system of equations. Generally, the types of momentum boundary conditions of outlet change the mathematical characteristics of SIP in the inner iteration. The often used iteration factor (0< <1) is only optimized for Dirichlet boundary condition, while is un-optimized for Neumann boundary. Optimization scheme of the former is presented in this paper.
Important Links:
Go Back
