Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Computational Science, Engineering & Technology Series
ISSN 1759-3158
Edited by: B.H.V. Topping, G. Montero, R. Montenegro
Chapter 15

Using Mixed Discretisation Schemes in Multi-Physics Simulation

M. Cross, T.N. Croft, D. McBride, A.K. Slone and A.J. Williams

School of Engineering, University of Wales Swansea, United Kingdom

Full Bibliographic Reference for this chapter
M. Cross, T.N. Croft, D. McBride, A.K. Slone, A.J. Williams, "Using Mixed Discretisation Schemes in Multi-Physics Simulation", in B.H.V. Topping, G. Montero, R. Montenegro, (Editors), "Innovation in Engineering Computational Technology", Saxe-Coburg Publications, Stirlingshire, UK, Chapter 15, pp 309-324, 2006. doi:10.4203/csets.15.15
Keywords: multi-physics, CFD, finite volume, unstructured mesh, finite elements, solid mechanics, parallel computation.

The practicalities of computational modelling of physical systems are always limited by a combination of computer hardware capability (essentially compute memory and speed), simulation data handling software (CAD geometry capture, mesh generation and visualisation technologies) and the simulation solver technologies. In the past decade or so great strides have been made in PC based hardware and low cost parallel cluster technologies exploiting powerful user environments such as Linux. Concomitant with this has been the development and deployment of very reasonable simulation handling software. In the case of the simulation analysis solver software the progress has not been quite so rapid.

Traditionally, CAE simulation analysis technologies are predominantly based upon the demands of one of the continuum phenomena:

  • computational fluid dynamics (CFD) : fluid flow, heat transfer, combustion
  • computational structural mechanics (CSM): solid mechanics (static and dynamic), contact, heat transfer.
  • computational electro-mechanic (CEM): electric and electro-magnetic field calculations
  • computational acoustics (CA)

This computational heritage means that the simulation of the closely coupled interactions between the various disciplines, that is, multi-physics analysis, has often either been accomplished by concentrating on one phenomenon and treating the others in a more simplistic fashion, or by simply ignoring key aspects of the multi-physics nature of the problem. A key reason for this limited approach to multi-physics simulation is that the, essentially, phenomena-specific CAE software employs numerical solution approaches which are specific and optimised for that particular physics component (for example, solids or fluids). Thus some years ago, we would typically have CFD tools using finite volume (FV) techniques with segregated iterative solution procedures over a block-structured mesh, whereas CSM tools use finite element (FE) techniques, with direct or iterative solvers, but on an unstructured mesh. In addition, CFD procedures use low order approximations, but CSM frequently uses higher order element approximations. Both CA & CEM codes tend to use either FE or FV techniques.

Increasingly, over the past few years, the CFD simulation tools have incorporated unstructured mesh technologies. This has certainly made multi-disciplinary analysis more practical because a single mesh structure may then be used for a range of distinct analyses. However, for close multi-physics analysis, in addition to being able to employ the same or, at least, a highly compatible mesh, it is important to ensure accurate filtering and mapping of data from one solver, for example, for the fluid, and then using it to generate volume source and boundary data for another solver, for example, for the structures, in the closely coupled framework. It is this issue that has inhibited the exploitation of existing CAE technologies in closely coupled multi-physics simulation.

There is no generic approach to the modelling of complex physical interactions, as the level and nature of physical coupling required to accurately represent such processes is problem-dependant. This may range from relatively straightforward multi-disciplinary processes requiring very low coupling, achieved by means of simple data transfer between codes, to strong two-way coupling, requiring mesh compatibility and subject to separate time-step constraints for each of the phenomena involved, which renders the implementation of a coherent solution strategy more difficult.

Most commercial CAE analysis software claims multi-physics capabilities but, in reality, what they actually offer at the present time, is multi-disciplinary, that is, data generated by one code is used as input into another, either as boundary data or as a volume source, where the data transfer is one way. This is distinguishable from full multi-physics analysis, which involves the two-way exchange of information, possibly requiring implicit convergence within a time-step, for example, thermo- mechanical. Even here there can be additional levels of sophistication, whereby both a time and space-accurate exchange of data is required; this is classed as closely coupled multi-physics. Hence, the type of problem under consideration will influence the level of coupling that is required for multi-physics analysis.

In this contribution we discuss the various options for addressing the challenge of closely coupled multi-physics simulation from where the CAE analysis industry is today. This discussion will focus upon two main options:

filter structures to enable phenomena-specific codes to exchange information directly from each other's databases without opening and closing files, that is, a technology which effectively enables full interoperability of separate codes,
emerging solver technologies which are specifically developed to enable closely coupled interaction

purchase the full-text of this chapter (price £20)

go to the previous chapter
go to the next chapter
return to the table of contents
return to the book description
purchase this book (price £90 +P&P)