Premium
Potential Inaccuracies in MODFLOW Simulations Involving the SIP and SSOR Methods for Matrix Solution
Author(s) -
Osiensky James L.,
Williams Roy E.
Publication year - 1997
Publication title -
groundwater
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.84
H-Index - 94
eISSN - 1745-6584
pISSN - 0017-467X
DOI - 10.1111/j.1745-6584.1997.tb00079.x
Subject(s) - modflow , convergence (economics) , conjugate gradient method , matrix (chemical analysis) , mathematics , algorithm , computer science , groundwater flow , groundwater , chemistry , engineering , aquifer , geotechnical engineering , chromatography , economic growth , economics
The ground‐water flow model MODFLOW is the most widely used and accepted model of its type in the world today. Several modeling runs were completed with MODFLOW for relatively simple hydrogeologic conditions to evaluate the accuracy of the strongly implicit procedure (SIP), slice successive overrelaxation (SSOR) and preconditioned conjugate‐gradient 2 (PCG2) methods for matrix solution. A series of simulations was performed with the SIP, SSOR, and PCG2 methods while adjusting the matrix solution parameters sequentially one at a time. A wide range of converged solutions was generated for identical input data. The simulation results show that MODFLOW will produce accurate results for the specific conditions simulated with PCG2, or with SIP if the proper combination of SIP matrix solution parameters is chosen by the user. Accurate results could not be obtained with SSOR for any of the combinations of matrix solution parameters used in this investigation. The simulation results confirm that model convergence for a small head change convergence criterion is not a good indicator of accurate results unless water balance error also is small.