Eric J. Whitney

Optimum Multidisciplinary and Multi-Objective Wing Design in CFD Using Evolutionary Techniques

This paper details some current extensions and applications of hierarchical asynchronous parallel evolutionary algorithms (HAPEA) for multidisciplinary and multi-objective wing design and optimisation problems. In this work the search for the solution takes place in separate hierarchical layers comprising different CFD solvers or resolutions. The performance and advantages of the algorithm are compared to that of a classical EA which would normally use only a single complex model and involve larger computational expense. The formulation and implementation of the algorithm are described and a test case for a multidisciplinary transonic wing design in structures and aerodynamics is presented. The trade-off between the objective functions produced a set of compromise designs represented in an optimal Pareto front. Results indicate that the algorithm is fast and robust for multi-objective and multidisciplinary optimisation problems and as designed produces classical as well as alternative wing configurations.

Multidisciplinary Aircraft Conceptual Design Optimisation Using a Hierarchical Asynchronous Parallel Evolutionary Algorithm (HAPEA)

In this paper we present some results of continuing research into improving robustness speed and application of Hierarchical Parallel Asynchronous Evolution Algorithms (HAPEA) to multidisciplinary design optimisation (MDO) and aircraft conceptual design problems. The formulation and implementation of the HAPEA-MDO algorithm is described and can be regarded as an architecture that is applicable to either integrated or distributed system optimisation design for complex, non-linear and non-differentiable problems. In this paper the formulation for HAPEA-MDO will be described and applied to single and multi objective MDO problems. Two cases related to aircraft design are analysed. We compute the Nash and Pareto optimal configurations satisfying the specified criteria in both cases and show that the HAPEA approach provides very efficient solutions to the stated design problems.

A Generic Framework for the Design Optimisation of Multidisciplinary UAV Intelligent Systems Using Evolutionary Computing

This paper describes the formulation and application of a design framework that supports the complex task of multidisciplinary design optimisation of Unmanned Aerial Vehicles (UAVs). The framework includes a Graphical User Interface (GUI), a robust Evolutionary Algorithm optimiser, several design modules, mesh generators and post-processing capabilities in an integrated platform. Traditional deterministic optimisation techniques for MDO are effective when applied to specific problems and within a specified range. A new class of optimisation techniques named Hierarchical Asynchronous Parallel Evolutionary Algorithms (HAPEAs) have shown to be robust as they require no derivatives or gradients of the objective function, have the capability of finding globally optimum solutions amongst many local optima, can be executed asynchronously in parallel and adapted easily to arbitrary solver codes without major modifications. The application of the methodology is illustrated on multi-criteria and multidisciplinary design problems. Results indicate the practicality and robustness of the method in finding optimal solutions and Pareto trade-offs between the disciplinary analyses and producing a set of non dominated individuals.

Aerodynamic Optimisation using a Robust Evolutionary Algorithm and Grid-free Flowsolver

It is well known that Evolutionary Algorithms (EAs) can provide solutions to problems that are difficult to solve with conventional deterministic optimisers. In this paper, we present continuing research on the application of a modern Evolutionary Algorithm (EA) for aerodynamic shape optimisation coupled with a grid-free or meshless flowsolver based on Kinetic schemes. The evolutionary method is based upon traditional evolution strategy with the incorporation of an asynchronous function evaluation for the solution and uses a hierarchical topology where the search for the best individual takes place successively in separate hierarchical layers comprising different fidelity models/resolutions or number of points. The grid-free formulation requires the domain discretisation to have very little topological information. A simple random distribution of points along with local connectivity information is sufficient. The connectivity which contains a set of neighbouring points is used to evaluate the special derivatives appearing in the conservation law. The derivatives are evaluated using Least Square (LS) approximation. The application of the methodology is then illustrated on two-dimensional inverse aerofoil optimisation problems. Results indicate that the method is robust and efficient on its application to real world problems.

Evolutionary Algorithms, Game Theory and Hierarchical Models in CFD

This paper details some current extensions to Evolutionary Algorithms (EAs) using concepts from game theory, as well as multiple-model and asynchronous computing. Some optimization case-studies are presented, which indicate the robustness and broad applicability of the method. These cases are 2D multi-element aerofoil reconstruction using Nash games, multiple ID nozzle reconstruction, 2D nozzle reconstruction using multiple models and asynchronous evaluation, and 2D multiobjective nozzle reconstruction using Pareto techniques. In each case it is shown that application of the extensions provide often significant gains in speed when compared to the traditional EA framework.

Evolutionary Algorithms for Multi-Objective Design Optimization

This article presents the general principles of Evolutionary Algorithms (EAs), along with a series of applications in the field of aeronautics. Classical EAs are good enough for problems based on simple mathematical models (i.e. linear models) . However, as the applications evolve in complexity, we had to develop new algorithms with better capabilities: among these, we will mostly focus on algorithms combining EAs and Game Theory (hence enabling the algorithm to deal with multi-criteria problems) as well as EAs with a hierarchical structure (which speeds up the convergence by using models of increasing complexity) . These concepts are then illustrated via experiments on several applications: minimization of the Radar Cross Section (RCS) around a multi-element airfoil in CEM, reconstruction of a 2D nozzle using multiple CFD models, and a coupled minimization (CEM + CFD) of the drag and RCS for an airfoil. These examples open the way for future applications of EAs in multi-disciplinary design optimization.

Multi-Criteria Aerodynamic Shape Design Problems in CFD Using a Modern Evolutionary Algorithm on Distributed Computers

A modern Evolutionary Algorithm (EA) for multi-criteria aerodynamic shape design optimisation is presented. This algorithm attempts to overcome some of the drawbacks associated with earlier evolutionary algorithms, the most important being that of large computational time.

Chapter 18 – CFD Design in Aeronautics Using a Robust Multilevel Parallel Evolutionary Optimiser

Publisher Summary This chapter explores the potential merit of an innovative parallel evolutionary algorithm (EA) coupled with current computational fluid dynamics (CFD) solvers. The chapter outlines the hierarchical asynchronous parallel evolution algorithm (HAPEA), and its important differences to a more conventional evolutionary method. A multi-objective test case involving the reconstruction of a set of two dimensional aerofoil geometries from scratch that are found by considering multiple prescribed pressure distributions at two different flow states is described. The chapter presents conclusions on the increased speed and robustness of the HAPEA algorithm operating in various parallel states, as compared to a more traditional method. Numerical experiments presented in this chapter provide the designer a gateway for practical applicability of parallel EAs in 3D industrial environments and MDO approaches using Navier–Stokes flow analysis solvers coupled with complex turbulence models. The increased speed and robustness obtained from the results underscore that the proper application of sound engineering judgment in conjunction with evolutionary techniques and parallel computing architectures can lead to optimal design solutions and significant computational savings when applied to real world problems.