Margot Gerritsen
Petroleum Research Institute
School of Earth Sciences
Stanford University
Stanford, California, USA
In many applications, including reservoir flows, explicit methods are desirable for accuracy, but prohibitively expensive because of the time step constraints imposed by numerical stability. Often, however, the constraints are local and far too much work is done in large parts of the domain. A straightforward approach seems to be to allow cells to step forward on their own terms and somehow couple the solutions at the many time levels thus introduced. Such asynchronous methods are however difficult to design in a conservative and consistent manner. In this talk I would like to discuss with you recent ideas we have had to address this rather interesting problem.