Multiobjective Optimization in Engineering Design

THIS THESIS FOCUSES on how to improve design and development of complex engineering systems by employing simulation and optimization techniques. Within the thesis, methods are developed and applied to systems that combine mechanical, hydraulical and electrical subsystems, so-called multi-domain systems. Studied systems include a landing gear system for a civil aircraft, electro-hydrostatic actuation systems for aircraft applications as well as hydraulic actuation systems. The usage of simulation and optimization in engineering design is gaining wider acceptance in all fields of industry as the computational capabilities of computers increase. Therefore, the applications for numerical optimization have increased dramatically. A great part of the design process is and will always be intuitive. Analytical techniques as well as numerical optimization could however be of great value and can permit vast improvements in design. Within the thesis, a framework is presented in which modeling and simulation are employed to predict the performance of a design. Additionally, non-gradient optimization techniques are coupled to the simulation models to automate the search for the best design. Engineering design problems often consist of several conflicting objectives. In many cases, the multiple objectives are aggregated into one single objective function. Optimization is then conducted with one optimal design as the result. The result is then strongly dependent on how the objectives are aggregated. Here a method is presented in which the Design Structure Matrix and the relationship matrix from the House of Quality method are applied to support the formulation of the objective function. Another approach to tackle multiobjective design problems is to employ the concept of Pareto optimality. Within this thesis a new multiobjective genetic algorithm is proposed and applied to support the design of a hydraulic actuation system. The outcome from such a multiobjective optimization is a set of Pareto optimal solutions that visualize the trade-off between the competing objectives. The proposed method is capable of handling a mix of continuous design variables and discrete selections of individual components from catalogs or databases. In real-world situations, system parameters will always include variations to some extent, and this fact is likely to influence the performance of the system. Therefore we need to answer not only the question “What is best?”, but also “What is sufficiently robust?” Within this thesis, several approaches to handle these two different questions are presented.