Computational & Technology Resources
an online resource for computational,
engineering & technology publications
ADVANCES IN COMPUTATIONAL MECHANICS WITH HIGH PERFORMANCE COMPUTING
Edited by: B.H.V. Topping
An Open Software Framework for Multiphysics Applications in Computational Mechanics on Parallel and Distributed Computers
P. Chow*, C. Bailey**, K. McManus**, M. Cross** and C. Addison*
*Fujitsu European Centre for Information Technology Limited, Uxbridge, United Kingdom
P. Chow, C. Bailey, K. McManus, M. Cross, C. Addison, "An Open Software Framework for Multiphysics Applications in Computational Mechanics on Parallel and Distributed Computers", in B.H.V. Topping, (Editor), "Advances in Computational Mechanics with High Performance Computing", Civil-Comp Press, Edinburgh, UK, pp 127-133, 1998. doi:10.4203/ccp.52.6.4
An open software framework for the computation and model developments of Multiphysics and multidiscipline applications in Computational Mechanics is presented. In order to run coupled problems involving Computational Fluid Dynamics (CFD), Computational Solid Mechanics (CSM), Computational Electromagnetics (CE), and the interactions between the physical phenomena, of reasonable size and in practical timeframes, parallel computation is required. Bringing together the various models (physical, numerical, and computational) needed to construct such an "Analysis Engine" is no simple task. A single software framework could help in such engine constructions. With the present framework, both new and existing elements such as libraries, discretization methods, algorithms and numerical models can all be accommodated comfortably via a component paradigm that is open to interchanges.
We illustrate the single-code software strategy behind such an analysis engine for these types of Multiphysics and coupled applications using a casting example case (lead ingots solidify in cast-iron moulds). This example includes analysis of thermal convection, solidification, stress, and interaction behavior, and run on a distributed memory parallel computer, the Fujitsu AP3000.
purchase the full-text of this paper (price £20)