A flexible C++ framework for the partitioned solution of strongly coupled multifield problems. (October 2016)

• Record Type: Journal Article
• Authors: König, Marcel; Radtke, Lars; Düster, Alexander
• Is Part Of: Computers & mathematics with applications. Volume 72:issue 7(2016)
• Page Start: 1764
• Page End: 1789
• Publication Date: 2016-10
• Subjects: Multifield problems -- Partitioned approach -- Convergence acceleration -- C++; Electronic data processing -- Periodicals; Mathematics -- Data processing -- Periodicals; 510.28541
• DOI: 10.1016/j.camwa.2016.07.031
• Languages: English
• ISSNs: 0898-1221

Abstract

Abstract: In this work, we present a flexible and generic C++ framework for the numerical solution of strongly coupled multifield problems, based on a partitioned approach. Coupled problems occur in a wide range of engineering applications, and their numerical treatment has recently gained much attention. We advocate a partitioned solution approach that enables the use of different discretization schemes and different solvers for the individual fields. Highly optimized, existing solvers can thus be reused—which enhances modularity, reusability, and performance. However, depending on the problem at hand, appropriate measures must be taken to stabilize the solution process and accelerate its convergence. In addition, the field quantities of interest need to be transferred between the solvers. To this end, we developed the software framework comana, which facilitates the implementation of different coupling strategies for a vast range of multifield problems. Interaction with the solvers is achieved through a uniform interface to the solvers' databases. Interfaces for solvers for which there is no interface available yet can be implemented with minimum effort.