Industry-strength Numerical Optimization: Application to Aerodynamic Shape Design

Boris Epstein and Sergey Pelgin (OPTIMENGA Ltd. and The Academic College of Tel-Aviv-Jaffa)

Wednesday, March 28th, 2012, CERFACS Conference Room - 11h00

Abstract.

Applying numerical optimization techniques to the design of new industrial products may not only significantly reduce the overall cost of a design cycle, but also to improve the quality of design. This presents a challenging problem whose solution requires the concurrent progress in at least three key areas: development of high-accuracy and robust tools of analysis, use of low-dimensional, yet representative parameterization of the design object, and development of computationally efficient techniques for global constrained optimization in high-dimensional search spaces.

In this work, we propose a new approach for multipoint multi-constrained design of essentially three-dimensional aerodynamic shapes in a wide range of flight conditions. The objective is to minimize total drag at fixed lift subject to numerous geometrical and aerodynamic constraints. The method and the corresponding code OPTIMENGA employ Genetic Algorithms (GAs) as an optimization engine in combination with a Reduced-Order Models (ROM) method, based on linked local data bases obtained by full Navier-Stokes computations.

The important features of the method include:

Robust handling of over 100 independent and interrelated design constraints
Use of a high-accuracy ENO-based full Navier-Stokes solver as a driver of optimization stream
PVM driven multilevel parallelization on multiprocessor cluster
Robust fault-tolerance treatment
High level of overall computational efficiency, which allows to access industrial solutions on a daily basis

Applied to a wide range of complex aerodynamic shapes, the method has allowed to significantly improve aerodynamic performance at design flight conditions, to maintain good off-design behavior and to essentially reduce the design cost. We believe that the proposed method gives opportunity to switch the design process in aircraft industry from the conventional "trial and error" approach (based on expertise and intuition of experienced engineers) to an almost fully automated process.

Cerfacs' Conferences Home Page