Block diagonal scaling for iterative methods in thermal simulation

Autor: T. Steihaug, W. R. Jines, J. C. Díaz, A. E. McDonald
Rok vydání: 1985
Předmět:
Zdroj: Communications in Applied Numerical Methods. 1:263-267
ISSN: 1555-2047
0748-8025
DOI: 10.1002/cnm.1630010602
Popis: Linear equations arising from reservoir simulation are closely coupled at the local-cell level, and the variables and equations tend to be poorly equilibrated. Scaling by the inverse of the diagonal block has an equilibrating effect. It usually improves the condition of the matrix for solution by linear iteration. The effects of block diagonal scaling are illustrated for a thermal reservoir simulator, where the original matrix may be indefinite. These results illustrate that for difficult problems block diagonal scaling can significantly enhance the convergence rate of the iterative method. It is also illustrated that preconditioned generated by block incomplete LU decomposition (ILU) and block symmetric Gauss Seidel (BSGS) methods are effective preconditioned for use in thermal simulation.
Databáze: OpenAIRE
Externí odkaz: