Guidelines for Product and System Optimisation

A summary report of the FENet Product and System Optimization (PSO) workshop in Majorca, Spain, 25th March 2004.

By Valentino Pediroda, Alberto Clarich, University of Trieste; Peter Bartholomew, QinetiQ Ltd., UK

Presentations were made covering the intended deliverables from Trieste University in the areas of Design of Experiments (DOE) and Optimisation methods, including the contribution of Game Theory to multi-objective design and of t-Student parameter in the statistical analysis of significance of the design variables.

The researchers of the University of Trieste showed the main topics of the work in the development of the algorithms and in their applications, while a further presentation by Prof. Bartholomew from QinetiQ Ltd. was made covering product design at a multidisciplinary level. The main application of the algorithms developed by Trieste researchers was the multi-point design optimisation of a 3D wing body complete of mobile aerodynamic flaps, while QinetiQ presented a multi-disciplinary Structural Optimisation application (structure, controls, propulsion, etc.) relative to a wing planform; a presentation from University of Naples about application of Matlab and Nastran for structural optimisation of gas turbine components concluded the session.

The presentations from Trieste and QinetiQ included the analysis tools distributed across a heterogeneous network (CAD, mesh generator, CFD code, optimiser, etc.) and the definition of standards for data exchange linking data to the product.

The deliverables will be further developed at Trieste University with additional input from QinetiQ and University of Newcastle to introduce basic terms more carefully and to define the mathematical techniques in the context of design. The deliverables will thus be intended for both the nonexpert users and the users that are instead more confident with the optimisation terminology.

Possibly a qualitative benchmark of features offered by products for optimisation design such as ModelCenter, BossQuattro, mode Frontier, iSight, and Optimus will be planned for the future.

During the meeting, a discussion about the future deliverables has been focused on a new approach for the design optimization characterised by uncertainties and fluctuations of operative conditions (Robust Design).

Multi objective algorithms (in particular Game Theory combined with statistical analysis for the optimum search space decomposition) will be applied to get the best compromise between mean performances and their stability in the range of the operative conditions.