Most problems in aerodynamic design are multi-objective in nature. For example, in compressors there are the contrasting requirements for large stable operating range and high performance, while in pump applications designers have to cope with the requirements of good cavitation performance and high efficiency or good performance at the design point and certain shut off head, and in fans there are the contrasting requirements of acoustic performance and efficiency. The most difficult issue facing designers is the choice of design parameters in the blade design process to meet these contrasting requirements in a time efficient manner. In many circumstances there is no general know-how in dealing with these multi-objective problems and the process of trial and error is rather time consuming.

An approach which has been shown to be effective in these types of multi-objective problems is to couple a 3D inverse design method with the Design of Experiments Method (DoE) and Response Surface Models (RSM). The DoE approach is a statistical model, which for a set of design parameters provides an experimental table of different design configurations that should be evaluated in order to cover the design space in an efficient manner. Each configuration corresponds to a different geometry for which all performance parameters need to be evaluated. Data provided by the DoE experimental table and the results of CFD evaluations are used to construct the Response Surface Model, which is a polynomial expression that relates the output performance parameters to the input design parameters. Once the RSM is obtained, one can use it to carry out sensitivity analysis or run a Multi-Objective Genetic Algorithm (MOGA) on the response surface. Since MOGA will be using the response surface, the evaluation costs of different performance parameters will be negligible as it only involves the evaluation of a polynomial given a set of input design parameters. By employing the response surface, one can then use MOGA to create the Pareto front for different performance parameters. The Pareto front provides a set of optimum designs that illustrates the nature of the tradeoffs between multiple objectives.

The success of such an approach is of course directly dependent on the accuracy of the RSM. The RSM accuracy depends in particular on the complexity of the response function which is here approximated by a polynomial. On the basis of this consideration, the use of an inverse design method to parameterize the blade geometry leads to major advantages. In contrast to conventional parameterization based on purely geometrical design parameters, an inverse method allows the use of aerodynamic parameters, such as the blade loading, which are much more closely related to the flow field and hence to the aerodynamic performance. This helps to make the shape of the response function relatively simple and as a result, it is much easier to obtain a high accuracy RSM based on a simple polynomial regression for a given number of data points and given region of the design space considered for the study.

ADT, a global leader in the development of advanced turbomachinery design software based on an inverse design approach, is joining forces with Engineous Japan Inc., a global leader in the development of engineering process integration, automation, and design optimization software, to establish a multi-sponsor consortium, initially open only to companies in the ASIA/Pacific region. The purpose of the consortium is to:

1. Develop a Design of Experiments (DoE) module in ADT’s TURBOdesign-1 (TD1) code to allow for multi-objective/multi-point design by using Response Surface Models (RSM) and Multi-Objective Genetic Algorithms (MOGA). This module will use specific modules such as DoE, Approximations, and Optimization from EJI’s iSIGHT-FD code. The module will be called the TD1-iFD module.

2. Develop a component in EJI’s iSIGHT-FD code for TURBOdesign-1 that allows TURBOdesign-1 to be inserted in an iSIGHT-FD workflow by dragging and dropping a TURBOdesign-1 icon from the component tab onto the Design Gateway workflow palette. The module will be called the iFD-TD1 module.

3. Demonstrate the considerable potential of inverse design based multi-objective/multi-point design by applying the methodology to each consortium member’s specific test case.

Further information on this exciting consortium plan can be obtained at the EJI CAO Frontier Seminar in Yokohama, Japan on September 12th and 13th of 2006. A seminar outlining the detailed objectives of the consortium is planned for Thursday 14th of September at 2pm in Yokohama. To obtain a flyer outlining the details of the consortium plan, or to register for the seminar planned on 14th of September, please contact:

ADT: sales@adtechnology.co.uk EJI: suzuki@engineous.co.jp

.