Implementation of a grid-enabled problem solving environment in Matlab

Abstract

In many areas of design search and optimisation one needs to utilize Computational Fluid Dynamics (CFD) methods in order to obtain numerical solution of the flow field in and/or around a proposed design. From this solution measures of quality for the design may be calculated, which are required by optimisation methods. In large models the processing time for the CFD computations can very well be many orders of magnitude larger than the optimisation methods; and the overall optimisation process usually demands a combination of computational and database resources; therefore this class of problems is well suited to Grid computing. The Geodise toolkit is a suite of tools for Grid-enabled. parametric geometry generation, meshing, CFD analysis, design optimisation and search, database, and notification tools within the Matlab environment. These grid services are presented to the design engineer as Matlab functions that conform to the usual syntax of Matlab. The use of the Geodise toolkit in Matlab introduces a flexible and Grid-enabled problem solving environment (PSE) for design search and optimisation. This PSE is illustrated here with an exemplar problem. © Springer-Verlag Berlin Heidelberg 2003.