The scale of engineering problems has sharply increased over the last twenty years. Larger coupled systems, increasing complexity, and limited resources create a need for methods that automatically decompose problems into manageable sub-problems by discovering and leveraging problem structure. The ability to learn the coupling (inter-dependence) structure and reorganize the original problem could lead to large reductions in the time to analyze complex problems. Such decomposition methods could also provide engineering insight on the fundamental physics driving problem solution. This work forwards the current state of the art in engineering decomposition through the application of techniques originally developed within computer science and information theory. The work describes the current state of automatic problem decomposition in engineering and utilizes several promising ideas to advance the state of the practice. Mutual information is a novel metric for data dependence and works on both continuous and discrete data. Mutual information can measure both the linear and non-linear dependence between variables without the limitations of linear dependence measured through covariance. Mutual information is also able to handle data that does not have derivative information, unlike other metrics that require it. The value of mutual information to engineering design work is demonstrated on a planetary entry problem. This study utilizes a novel tool developed in this work for planetary entry system synthesis. A graphical method, force-based clustering, is used to discover related sub-graph structure as a function of problem structure and links ranked by their mutual information. This method does not require the stochastic use of neural networks and could be used with any link ranking method currently utilized in the field. Application of this method is demonstrated on a large, coupled low-thrust trajectory problem. Mutual information also serves as the basis for an alternative global optimizer, called MIMIC, which is unrelated to Genetic Algorithms. Advancement to the current practice demonstrates the use of MIMIC as a global method that explicitly models problem structure with mutual information, providing an alternate method for globally searching multi-modal domains. By leveraging discovered problem inter-dependencies, MIMIC may be appropriate for highly coupled problems or those with large function evaluation cost. This work introduces a useful addition to the MIMIC algorithm that enables its use on continuous input variables. By leveraging automatic decision tree generation methods from Machine Learning and a set of randomly generated test problems, decision trees for which method to apply are also created, quantifying decomposition performance over a large region of the design space.

Traditionally, the design of complex aerospace systems requires iteration among segregated disciplines such as aerodynamic modeling and trajectory optimization. Multidisciplinary design optimization algorithms have been developed to efficiently orchestrate the interaction among these disciplines during the design process. For example, vehicle capability is generally obtained through sequential iteration among vehicle shape, aerodynamic performance, and trajectory optimization routines in which aerodynamic performance is obtained from large pre-computed tables that are a function of angle of attack, sideslip, and flight conditions. This numerical approach segregates advancements in vehicle shape design from advancements in trajectory optimization. This investigation advances the state-of-the-art in conceptual hypersonic aerodynamic analysis and trajectory optimization by removing the source of iteration between aerodynamic and trajectory analyses and capitalizing on fundamental linkages across hypersonic solutions. Analytic aerodynamic relations, like those derived in this investigation, are possible in any flow regime in which the flowfield can be accurately described analytically. These relations eliminate the large aerodynamic tables that contribute to the segregation of disciplinary advancements. Within the limits of Newtonian flow theory, many of the analytic expressions derived in this investigation provide exact solutions that eliminate the computational error of approximate methods widely used today while simultaneously improving computational performance. To address the mathematical limit of analytic solutions, additional relations are developed that fundamentally alter the manner in which Newtonian aerodynamics are calculated. The resulting aerodynamic expressions provide an analytic mapping of vehicle shape to trajectory performance. This analytic mapping collapses the traditional, segregated design environment into a single, unified, mathematical framework which enables fast, specialized trajectory optimization methods to be extended to also include vehicle shape. A rapid trajectory optimization methodology suitable for this new, mathematically integrated design environment is also developed by relying on the continuation of solutions found via indirect methods. Examples demonstrate that families of optimal hypersonic trajectories can be quickly constructed for varying trajectory parameters, vehicle shapes, atmospheric properties, and gravity models to support design space exploration, trade studies, and vehicle requirements definition. These results validate the hypothesis that many hypersonic trajectory solutions are connected through fast indirect optimization methods. The extension of this trajectory optimization methodology to include vehicle shape through the development of analytic hypersonic aerodynamic relations enables the construction of a unified mathematical framework to perform rapid, simultaneous hypersonic aerodynamic and trajectory optimization. Performance comparisons relative to state-of-the-art methodologies illustrate the computational advantages of this new, unified design environment.

Engineering design may be viewed as a decision making process that supports design tradeoffs. The designer makes decisions based on information available and engineering judgment. The designer determines the direction in which the design must proceed, the procedures that need to be adopted, and develops a strategy to perform successive decisions. The design is only as good as the decisions made, which is in turn dependent on the information available. Information is time and process dependent. This thesis work focuses on developing a coherent bottom-up framework and methodology to improve information transfer and decision making while designing complex systems. The rotorcraft drive system is used as a test system for this methodology. The traditional serial design approach required the information from one discipline and/or process in order to proceed with the subsequent design phase. The Systems Engineering (SE) implementation of Concurrent Engineering (CE) and Integrated Product and Process Development (IPPD) processes tries to alleviate this problem by allowing design processes to be performed in parallel and collaboratively. The biggest challenge in implementing Concurrent Engineering is the availability of information when dealing with complex systems such as aerospace systems. The information is often incomplete, with large amounts of uncertainties around the requirements, constraints and system objectives. As complexity increases, the design process starts trending back towards a serial design approach. The gap in information can be overcome by either “softening” the requirements to be adaptable to variation in information or to delay the decision. Delayed decisions lead to expensive modifications and longer product design lifecycle. Digitization of IPPD tools for complex system enables the system to be more adaptable to changing requirements. Design can proceed with “soft” information and decisions adapted as information becomes available even at early stages. The advent of modern day computing has made digitization and automation possible and feasible in engineering. Automation has demonstrated superior capability in design cycle efficiency [1]. When a digitized framework is enhanced through automation, design can be made adaptable without the requirement for human interaction. This can increase productivity, and reduce design time and associated cost. An important aspect in making digitization feasible is having the availability of parameterized Computer Aided Design (CAD) geometry [2]. The CAD geometry gives the design a physical form that can interact with other disciplines and geometries. Central common CAD database allows other disciplines to access information and extract requirements; this feature is of immense importance while performing systems syntheses. Through database management using a Product Lifecycle Management (PLM) system, Integrated Product Teams (IPTs) can exchange information between disciplines and develop new designs more efficiently by collaborating more and from far [3]. This thesis focuses on the challenges associated with automation and digitization of design. Making more information available earlier goes jointly with making the design adaptable to new information. Using digitized sizing, synthesis, cost analysis and integration, the drive system design is brought in to early design. With modularity as the objective, information transfer is made streamlined through the use of a software integration suite. Using parametric CAD tools, a novel ‘Fully-Relational Design’ framework is developed where geometry and design are adaptable to related geometry and requirement changes. During conceptual and preliminary design stages, the airframe goes through many stages of modifications and refinement; these changes affect the sub-system requirements and its design optimum. A fully-relational design framework takes this into account to create interfaces between disciplines. A novel aspect of the fully-relational design methodology is to include geometry, spacing and volume requirements in the system design process. Enabling fully-relational design has certain challenges, requiring suitable optimization and analysis automation. Also it is important to ensure that the process does not get overly complicated. So the method is required to possess the capability to intelligently propagate change. There is a need for suitable optimization techniques to approach gear train type design problems, where the design variables are discrete in nature and the values a variables can assume is a result of cascading effects of other variables. A heuristic optimization method is developed to analyze this multimodal problem. Experiments are setup to study constraint dependencies, constraint-handling penalty methods, algorithm tuning factors and innovative techniques to improve the performance of the algorithm. Inclusion of higher fidelity analysis in early design is an important element of this research. Higher fidelity analyses such as nonlinear contact Finite Element Analysis (FEA) are useful in defining true implied stresses and developing rating modification factors. The use of Topology Optimization (TO) using Finite Element Methods (FEM) is proposed here to study excess material removal in the gear web region.

Aluminum Cross-Car Beam is significantly lighter than the conventional steel counterpart and presents superior energy absorption characteristics. The challenge is however, its considerably higher cost, rendering it difficult for the aluminum one to compete in the automotive market. In this work, using material distribution techniques and stochastic optimization, a Multidisciplinary Design Optimization procedure is developed to optimize an existing Cross-Car Beam model with respect to the cost. Topology, Topography, and gauge optimizations are employed in the development of the optimization disciplines. Based on a qualitative cost assessment, penalty functions are designed to penalize costly designs. Noise-Vibration-Harshness (NVH) performance is the key constraint of the optimization. To fulfill this requirement, natural frequencies are obtained using modal analysis. Undesirable designs with respect to the NVH criteria are gradually eliminated from the optimization cycles. The new design is verified by static loading scenario and evaluated in terms of the cost saving it offers.

Aluminum Cross-Car Beam is significantly lighter than the conventional steel counterpart and presents superior energy absorption characteristics. The challenge is however, its considerably higher cost, rendering it difficult for the aluminum one to compete in the automotive market. In this work, using material distribution techniques and stochastic optimization, a Multidisciplinary Design Optimization procedure is developed to optimize an existing Cross-Car Beam model with respect to the cost. Topology, Topography, and gauge optimizations are employed in the development of the optimization disciplines. Based on a qualitative cost assessment, penalty functions are designed to penalize costly designs. Noise-Vibration-Harshness (NVH) performance is the key constraint of the optimization. To fulfill this requirement, natural frequencies are obtained using modal analysis. Undesirable designs with respect to the NVH criteria are gradually eliminated from the optimization cycles. The new design is verified by static loading scenario and evaluated in terms of the cost saving it offers.

Improvements in performance, reliability and durability as well as reductions in production costs, remain critical prerequisites for the commercialization of proton exchange membrane fuel cells. In this thesis, a computational framework for fuel cell analysis and optimization is presented as an innovative alternative to the time consuming trial-and-error process currently used for fuel cell design. The framework is based on a two-dimensional through-the-channel isothermal, isobaric and single phase membrane electrode assembly (MEA) model. The model input parameters are the manufacturing parameters used to build the MEA: platinum loading, platinum to carbon ratio, electrolyte content and gas diffusion layer porosity. The governing equations of the fuel cell model are solved using Netwon's algorithm and an adaptive finite element method in order to achieve quadratic convergence and a mesh independent solution respectively. The analysis module is used to solve two optimization problems: i) maximize performance; and, ii) maximize performance while minimizing the production cost of the MEA. To solve these problems a gradient-based optimization algorithm is used in conjunction with analytical sensitivities. The presented computational framework is the first attempt in the literature to combine highly efficient analysis and optimization methods to perform optimization in order to tackle large-scale problems. The framework presented is capable of solving a complete MEA optimization problem with state-of-the-art electrode models in approximately 30 minutes. The optimization results show that it is possible to achieve Pt-specific power density for the optimized MEAs of 0.422 $g_{Pt}/kW$. This value is extremely close to the target of 0.4 $g_{Pt}/kW$ for large-scale implementation and demonstrate the potential of using numerical optimization for fuel cell design.

Computational models and simulations are essential system design tools that allow for improved decision making and cost reductions during all phases of the design process. However, the most accurate models are often computationally expensive and can therefore only be used sporadically. Consequently, designers are often forced to choose between exploring many design alternatives with less accurate, inexpensive models and evaluating fewer alternatives with the most accurate models. To achieve both broad exploration of the alternatives and accurate determination of the best alternative with reasonable costs incurred, surrogate modeling and variable accuracy modeling are used widely. A surrogate model is a mathematically tractable approximation of a more expensive model based on a limited sampling of that model, while variable accuracy modeling involves a collection of different models of the same system with different accuracies and computational costs. As compared to using only very accurate and expensive models, designers can determine the best solutions more efficiently using surrogate and variable accuracy models because obviously poor solutions can be eliminated inexpensively using only the less expensive, less accurate models. The most accurate models are then reserved for discerning the best solution from the set of good solutions. In this thesis, a Value-Based Global Optimization (VGO) algorithm is introduced. The algorithm uses kriging-like surrogate models and a sequential sampling strategy based on Value of Information (VoI) to optimize an objective characterized by multiple analysis models with different accuracies. It builds on two primary research contributions. The first is a novel surrogate modeling method that accommodates data from any number of analysis models with different accuracies and costs. The second contribution is the use of Value of Information (VoI) as a new metric for guiding the sequential sampling process for global optimization. In this manner, the cost of further analysis is explicitly taken into account during the optimization process. Results characterizing the algorithm show that VGO outperforms Efficient Global Optimization (EGO), a similar global optimization algorithm that is considered to be the current state of the art. It is shown that when cost is taken into account in the final utility, VGO achieves a higher utility than EGO with statistical significance. In further experiments, it is shown that VGO can be successfully applied to higher dimensional problems as well as practical engineering design examples.

Improvements in performance, reliability and durability as well as reductions in production costs, remain critical prerequisites for the commercialization of proton exchange membrane fuel cells. In this thesis, a computational framework for fuel cell analysis and optimization is presented as an innovative alternative to the time consuming trial-and-error process currently used for fuel cell design. The framework is based on a two-dimensional through-the-channel isothermal, isobaric and single phase membrane electrode assembly (MEA) model. The model input parameters are the manufacturing parameters used to build the MEA: platinum loading, platinum to carbon ratio, electrolyte content and gas diffusion layer porosity. The governing equations of the fuel cell model are solved using Netwon's algorithm and an adaptive finite element method in order to achieve quadratic convergence and a mesh independent solution respectively. The analysis module is used to solve two optimization problems: i) maximize performance; and, ii) maximize performance while minimizing the production cost of the MEA. To solve these problems a gradient-based optimization algorithm is used in conjunction with analytical sensitivities. The presented computational framework is the first attempt in the literature to combine highly efficient analysis and optimization methods to perform optimization in order to tackle large-scale problems. The framework presented is capable of solving a complete MEA optimization problem with state-of-the-art electrode models in approximately 30 minutes. The optimization results show that it is possible to achieve Pt-specific power density for the optimized MEAs of 0.422 $g_{Pt}/kW$. This value is extremely close to the target of 0.4 $g_{Pt}/kW$ for large-scale implementation and demonstrate the potential of using numerical optimization for fuel cell design.

A indústria aeronáutica vem promovendo avanços tecnológicos em velocidades crescentes, para sobreviver em mercados extremamente competitivos. Neste cenário, torna-se imprescindível o uso de ferramentas de projeto que agilizem o desenvolvimento de novas aeronaves. Os atuais recursos computacionais permitiram um grande aumento no número de ferramentas que auxiliam o trabalho de projetistas e engenheiros. O projeto de uma aeronave é uma tarefa multidisciplinar por essência, o que logo incentivou o desenvolvimento de ferramentas computacionais que trabalhem com várias áreas ao mesmo tempo. Entre elas se destaca a otimização multidisciplinar em projeto, que une métodos de otimização à modelos matemáticos de áreas distintas de um projeto para encontrar soluções de compromisso. O presente trabalho introduz a otimização multidisciplinar em projeto (Multidisciplinary Design Optimization - MDO) e discorre sobre algumas aplicações possíveis desta metodologia. Foi realizada a implementação de um sistema de MDO para o projeto de asas flexíveis, considerando restrições de aeroelasticidade dinâmica e massa estrutural. Como meta, deseja-se encontrar distribuições ideais de rigidezes flexional e torcional da estrutura da asa, para maximizar a velocidade crítica de flutter e minimizar a massa estrutural. Para tanto, foram utilizados um modelo dinâmico-estrutural baseado no método dos elementos finitos, um modelo aerodinâmico não-estacionário baseado na teoria das faixas e nas soluções bidimensionais de Theodorsen, um modelo de previsão de flutter que utiliza o método K e, por fim, um otimizador baseado no método de algoritmos genéticos (AGs). São apresentados os detalhes empregados em cada modelo, as restrições aplicadas e a maneira como eles interagem ao longo da otimização. É feita uma análise para a escolha dos parâmetros de otimização por AG e em seguida a avaliação de dois casos, para verificação da funcionalidade do sistema implementado. Os resultados obtidos demonstram uma metodologia eficiente, que é capaz de buscar soluções ótimas para problemas propostos, que com devidos ajustes pode ter enorme valor para acelerar o desenvolvimento de novas aeronaves. / The aeronautical industry is always trying to speed up technological advances in order to survive in extremely competitive markets. In this scenario, the use of design tools to accelerate the development of new aircraft becomes essential. Current computational resources allow greater increase in the number of design tools to assist the work of aeronautical engineers. In essence, the design of an aircraft is a multidisciplinary task, which stimulates the development of computational tools that work with different areas at the same time. Among them, the multidisciplinary design optimization (MDO) can be distinguished, which combines optimization methods to mathematical models of distinct areas of a design to find compromise solutions. The present work introduces MDO and discourses on some possible applications of this methodology. The implementation of a MDO system for the design of flexible wings, considering dynamic aeroelasticity restrictions and the structural mass, was carried out. As goal, it is desired to find ideal flexional and torsional stiffness distributions of the wing structure, that maximize the critical flutter speed and minimize the structural mass. To do so, it was employed a structural dynamics model based on the finite element method, a nonstationary aerodynamic model based on the strip theory and Theodorsen’s two-dimensional solutions, a flutter prediction model based on the K method and a genetic algorithm (GA). Details on the model, restrictions applied and the way the models interact to each other through the optimization are presented. It is made an analysis for choosing the GA optimization parameters and then, the evaluation of two cases to verify the functionality of the implemented system. The results obtained illustrate an efficient methodology, capable of searching optimal solutions for proposed problems, that with the right adjustments can be of great value to accelerate the development of new aircraft.

La conception de lanceurs est un problème d’optimisation multidisciplinaire dont l’objectif est de trouverl’architecture du lanceur qui garantit une performance optimale tout en assurant un niveau de fiabilité requis.En vue de l’obtention de la solution optimale, les phases d’avant-projet sont cruciales pour le processus deconception et se caractérisent par la présence d’incertitudes dues aux phénomènes physiques impliqués etaux méconnaissances existantes sur les modèles employés. Cette thèse s’intéresse aux méthodes d’analyse et d’optimisation multidisciplinaire en présence d’incertitudes afin d’améliorer le processus de conception de lanceurs. Trois sujets complémentaires sont abordés. Tout d’abord, deux nouvelles formulations du problème de conception ont été proposées afin d’améliorer la prise en compte des interactions disciplinaires. Ensuite, deux nouvelles méthodes d’analyse de fiabilité, permettant de tenir compte d’incertitudes de natures variées, ont été proposées, impliquant des techniques d’échantillonnage préférentiel et des modèles de substitution. Enfin, une nouvelle technique de gestion des contraintes pour l’algorithme d’optimisation ”Covariance Matrix Adaptation - Evolutionary Strategy” a été développée, visant à assurer la faisabilité de la solution optimale. Les approches développées ont été comparées aux techniques proposées dans la littérature sur des cas tests d’analyse et de conception de lanceurs. Les résultats montrent que les approches proposées permettent d’améliorer l’efficacité du processus d’optimisation et la fiabilité de la solution obtenue. / Launch vehicle design is a Multidisciplinary Design Optimization problem whose objective is to find the launch vehicle architecture providing the optimal performance while ensuring the required reliability. In order to obtain an optimal solution, the early design phases are essential for the design process and are characterized by the presence of uncertainty due to the involved physical phenomena and the lack of knowledge on the used models. This thesis is focused on methodologies for multidisciplinary analysis and optimization under uncertainty for launch vehicle design. Three complementary topics are tackled. First, two new formulations have been developed in order to ensure adequate interdisciplinary coupling handling. Then, two new reliability techniques have been proposed in order to take into account the various natures of uncertainty, involving surrogate models and efficient sampling methods. Eventually, a new approach of constraint handling for optimization algorithm ”Covariance Matrix Adaptation - Evolutionary Strategy” has been developed to ensure the feasibility of the optimal solution. All the proposed methods have been compared to existing techniques in literature on analysis and design test cases of launch vehicles. The results illustrate that the proposed approaches allow the improvement of the efficiency of the design process and of the reliability of the found solution.