Ce mémoire s’intéresse à la conception robuste de systèmes complexes dans le cadre de l’ingénierie système et de la méthode First Design. Ces travaux s’appliquent plus particulièrement aux prestations en choc frontal de véhicules de la gamme Renault. L’objectif principal de ces travaux est de proposer une méthode de conception robuste basée sur la modélisation numérique des prestations crash du véhicule. Cette stratégie vise à assurer la robustesse du produit dès la phase de conception, afin d’éviter des modifications de conception tardives et coûteuses, conséquences d’apparition de problèmes durant le cycle de validation ou la vie série du véhicule. Les spécificités du crash sont le coût important des simulations, la forte non linéarité du phénomène, ainsi que les bifurcations de comportement. Ces particularités rendent les méthodes classiques de conception robuste peu efficaces ou très couteuses. Afin de répondre à ce problème, nous développons une méthode originale, baptisée détection de défaillances, permettant d’identifier les problèmes de robustesse en crash, afin de les corriger dès le cycle de conception. Cette méthode est basée sur l’utilisation des techniques d’optimisation par les plans d’expériences. La méthode développée vise aussi à intégrer l’expertise des concepteurs crash afin de localiser rapidement les défaillances, ce qui permet de limiter le nombre de simulations nécessaires. La contrepartie d’une méthode de conception robuste reposant sur la simulation numérique est la nécessité d’avoir un bon niveau de confiance dans les résultats du modèle. On propose donc dans ce mémoire des améliorations des modèles éléments finis des véhicules Renault, afin d’améliorer la qualité de la simulation. Ces travaux vont dans le sens d’un remplacement des prototypes physiques par des prototypes numériques dans l’industrie, enjeu majeur permettant la réduction des coûts et des délais de développement. Cet enjeu est particulièrement important dans un secteur automobile très concurrentiel, où la survie d’un constructeur dépend de ses coûts et de sa réactivité face au marché. / This PhD thesis deals with robust design of complex products, within the framework of system engineering methods, such as First Design. This work focuses on frontal crashworthiness of Renault vehicles. The main goal of this PhD is to develop a robust design method based on crashworthiness numerical simulation. This method aims at ensuring the robustness of a vehicle crashworthiness right from the design stage of the product, in order to avoid costly design modifications, necessary when problems are found during the validation cycle or the life cycle of the product. Characteristics of crashworthiness phenomena are a high cost of numerical simulation, highly non-linear and bifurcative behaviour. Due to this behaviour, classic robust design methods would be unefficient or very expensive to use. In order to face this problem, we develop an original robust design method, based on optimization using design of experiments method. The goal of this method is to identify crash failures as soon as possible in the design stage, in order to correct them. This method also aims at integrating knowledge from the crash engineers, in order to find crash failures quickly, using as few simulations as possible. A challenge we meet when using numerical simulation of the crashworthiness is the need to trust the results of the model. This thesis also deals with improvements in the crash models at Renault. This work is well suited for a very competitive industry such as the automotive, where car manufacturers need to replace physical prototypes with numerical ones, in order to reduce design costs and be more reactive.

An investigation was conducted to assess the usefulness of anti-optimization in the process of validating of aerodynamic codes. Anti-optimization is defined here as the intentional search for regions where the computational and experimental results disagree. Maximizing such disagreements can be a useful tool in uncovering errors and/or weaknesses in both analyses and experiments. The codes chosen for this investigation were an airfoil code and a lifting line code used together as an analysis to predict three-dimensional wing aerodynamic coefficients. The parameter of interest was the maximum lift coefficient of the three-dimensional wing, CL max. The test domain encompassed Mach numbers from 0.3 to 0.8, and Reynolds numbers from 25,000 to 250,000. A simple rectangular wing was designed for the experiment. A wind tunnel model of this wing was built and tested in the NASA Langley Transonic Dynamics Tunnel. Selection of the test conditions (i.e., Mach and Reynolds numbers) were made by applying the techniques of response surface methodology and considerations involving the predicted experimental uncertainty. The test was planned and executed in two phases. In the first phase runs were conducted at the pre-planned test conditions. Based on these results additional runs were conducted in areas where significant differences in CL max were observed between the computational results and the experiment — in essence applying the concept of anti-optimization. These additional runs were used to verify the differences in CL max and assess the extent of the region where these differences occurred. The results of the experiment showed that the analysis was capable of predicting CL max to within 0.05 over most of the test domain. The application of anti-optimization succeeded in identifying a region where the computational and experimental values of CL max differed by more than 0.05, demonstrating the usefulness of anti-optimization in process of validating aerodynamic codes. This region was centered at a Mach number of 0.55 and a Reynolds number of 34,000. Including considerations of the uncertainties in the computational and experimental results confirmed that the disagreement was real and not an artifact of the uncertainties. / Ph. D.

Hoje em dia, é amplamente reconhecido que o projeto de estruturas otimizadas deve ser robusto em relação a incertezas nas forças, geometria e propriedades do material. Entretanto, existem diversas alternativas para considerar tais incertezas em problemas de otimização estrutural. Esta tese apresenta quatro formulações para lidar com incertezas no problema de otimização topológica com restrição de tensão. As três primeiras são desenvolvidas para lidar com incertezas na intensidade e direção das forças aplicadas: 1) formulação robusta probabilística, onde substituem-se as restrições de tensão originais por uma soma ponderada entre os seus valores esperados e desvios padrão, obtidos por meio do método de perturbação de primeira ordem; 2) formulação baseada em confiabilidade, onde consideram-se restrições de tensão probabilísticas; o problema é formulado por meio de uma abordagem acoplada de primeira ordem; 3) formulação robusta não probabilística, onde considera-se o pior cenário possível para as restrições de tensão; o problema é formulado com uma abordagem acoplada de otimização com anti-otimização. A quarta formulação não segue o padrão das três primeiras; diferente das demais, esta é desenvolvida para lidar com incerteza uniforme de manufatura: 4) formulação robusta de três campos, onde três topologias são consideradas de forma simultânea durante o processo de otimização, de forma a simular possíveis imperfeições que possam ocorrer devido a erros de manufatura. As quatro abordagens são bastante diferentes na forma de lidar com as incertezas; no entanto, o procedimento de solução é o mesmo: a abordagem baseada em densidade é empregada na parametrização material, enquanto que o método do Lagrangiano aumentado é empregado para solucionar o problema resultante, de forma a lidar com o elevado número de restrições de tensão. Diversos exemplos são solucionados para mostrar a aplicabilidade das formulações propostas. Os exemplos são posteriormente verificados através da Simulação de Monte Carlo e comparados com os resultados determinísticos. Os resultados mostram que as estruturas obtidas com a abordagem tradicional determinística são extremamente sensíveis a incertezas. As formulações desenvolvidas nesta tese, por outro lado, mostraram-se alternativas válidas a formulação determinística, fornecendo resultados robustos e confiáveis na presença de incertezas. / It is nowadays widely acknowledged that optimal structural design should be robust with respect to the uncertainties in loads, geometry and material parameters. However, there are several alternatives to consider such uncertainties in structural optimization problems. This thesis addresses four formulations to handle uncertainties in topology optimization with stress constraint. The first three are developed to handle uncertainties in magnitude and direction of applied loads: 1) probabilistic robust formulation, where the original stress constraints are replaced by a weighted sum between their expectations and standard deviations; these are obtained by first-order perturbation approach; 2) reliability-based formulation, where probabilistic stress constraints are considered; the problem is formulated by a coupled first order approach; 3) non-probabilistic robust formulation, where the worstcase scenario for the stress constraints is considered; the problem is formulated by a coupled approach called optimization with anti-optimization. The fourth formulation is quite different from the first three; it is developed to handle uniform boundary variation: 4) three-field robust approach, where three topologies are simultaneously considered during the optimization process, in order to simulate imperfections which may occur due to manufacturing errors. These four formulations are quite different in handling with uncertainties; however, the solution rocedure is the same: the density approach is employed to material parameterization, while the augmented Lagrangian method is employed to solve the resulting problem, in order to handle the large number of stress constraints. Several examples are solved to demonstrate applicability of proposed formulations. Numerical examples are further verified via Monte Carlo Simulation and compared to deterministic results. The results show that the structures obtained with raditional deterministic formulation are extremely sensitive to uncertainties. On the other hand, the formulations developed in this thesis are shown to be valid alternatives to the deterministic formulation, providing robust and reliable results in the presence of uncertainties.