Adaptive Trust Region MDO

 

Title: Adaptive Trust Region Multi-Disciplinary Optimisation

Investigators: Alex Orr, Philip Hewitt

Sponsor: DEL

 

Project Backgroud

The development of improved aircraft designs is a significant challenge due to the complex interactions of several disciplines (e.g.: aircraft structures, aerodynamics, propulsion, etc.). As a result of this complexity designers can struggle to provide optimised solutions readily. The development of efficient design tools is critical to the UK aerospace industry as it must meet increasingly demanding environmental and commercial requirements. This project aims to establish a method that could be integrated into the regular design framework and provide a practical means of utilising optimisation techniques for improved designs. The widespread use of optimisation within the design process would contribute to the UK aerospace industry to remain as one of the world leaders in terms of aircraft design.

The goal of the project is to assist the designer in overcoming technical and scientific challenges by developing a robust framework that can automatically produce unintuitive, robust and optimal designs. Aircraft designers are becoming increasingly dependent on CFD analysis rather than wind tunnel testing due to the reduction of costs and the rapid change and update of designs. With this demand models are being analysed with ever increasing fidelity and complex-
ity. When considering high-fidelity models for multidisciplinary optimisation (MDO) costs can become unacceptable, inhibiting its use as a regular design tool. In order to facilitate the widespread use of high-fidelity MDO solutions to reduce the high computational expense need to be implemented. There are several methods by which this may be performed, one robust method is known as the trust region method (TRM). This method involves the coupling together of high and low fidelity model in each discipline. Such methods attempt to minimise the use of high-fidelity model without compromising the accuracy of the simulation.

This project will investigate new algorithms based on the TRMs philosophy and develop methods capable of solving realistic, industrial size optimisation challenges, efficiently and robustly. This new framework will be demonstrated via a series of test cases to evaluate its effectiveness and robustness.

 

References:

Marques, S., Orr, A., Hewitt, P. & Robinson, T., MULTI-DISCIPLINARY WING OPTIMIZATION USING A DISCIPLINE-SPECIFIC TRUST-REGION METHOD, IFASD, Bristol, U.K., 2013, Trust_Region_MDO_IFASD_2013.pdf