Towards time stable and high order accurate methods for realistic applications

Ken Mattsson
Center for Turbulence Research
Stanford University
Stanford, California, USA


Abstract:

Strict (or time) stability, which prevents error growth on realistic mesh sizes, is very important for calculations over long times. I will present some recent developments towards an efficient and general technique to accurately handle both complex geometries and transient phenomena. The focus is on a strictly stable hybrid between an unstructured finite volume method and a high-order finite difference method. I will spend some time illustrating typical applications that I am currently involved with at Stanford University. I will also discuss some recent observations and new developments with emphasis on accuracy and stability.