With the development of manufacturing techniques such as the Electron Beam Free Form Fabrication (EBF3), a metal deposition technique which deposits metal in complex shapes on a metallic base plate, it has become easy to manufacture complex shapes such as panels with curvilinear stiffeners. Designing and optimizing stiffened panels with predefined structural and acoustic response is the focus of this dissertation. Researchers have dealt with sizing optimization of panels with straight/curvilinear stiffeners for many years and it has been proven that in some cases the mass of a panel with curvilinear stiffeners is lesser than the mass of a panel with straight stiffeners for a complex loading such as bi-axial compression with shear and transverse pressure. The research work in this dissertation addresses the sizing as well as placement optimization of panel with straight and curvilinear stiffeners for desired structural as well as acoustic response. For acoustic optimization, point-excited stiffened panels are designed for minimal sound radiation given the constraint on total mass of the structure. To reduce the computational expense of structural-acoustic optimization, a new methodology for the objective function evaluation is also proposed and optimal design for minimum radiated acoustic power is discussed. The developed framework, named EBF3PanelOpt for structural acoustic optimization of point excited stiffened panels is extended to multi point excitation to capture the realistic excitations such as turbulent boundary layer (TBL) pressure fluctuations. The Corcos model of representing TBL is used to capture correlation of TBL pressure excitation. Validation of the approach using Corcos TBL model, implemented in EBF3PanelOpt is performed using fast multi-pole boundary element method in FastBEM and a conventional boundary element code, HELM3D. The optimal designs are obtained for a panel with two and four stiffeners, respectively. The minimization of both, the mass and the acoustic response during structural-acoustic optimization is conflicting in nature. Therefore, a multi-objective design optimization using non-dominated sorting genetic algorithm-II (NSGA-II) is performed. The Pareto optimal designs, obtained using multi-objective design optimization approach has reduced the acoustic response significantly with a minor mass penalty of the structure when compared to a baseline design while meeting all the constraints such as buckling eigenvalue, von Mises, and crippling stresses. A multi-objective design optimization framework is also developed for design optimization of diffuse field excited panels with straight and curvilinear stiffeners. The Pareto optimal designs of panel with six stiffeners are obtained using developed framework and a comparative study is performed with a baseline design with six straight stiffeners. The developed framework is also extended to perform multi-objective design optimization of point excited complex structures such as curved panels with straight or curvilinear stiffeners. A fast multi-pole boundary element method is used to calculate the acoustic response of the curved panel with stiffeners and design optimization results of a curved panel with two and four stiffeners are discussed. Experiments are also performed at Sound and Acoustic Load Transmission (SALT) facility of Langley Research Center to measure the sound radiation and transmission loss for two panels with straight and curvilinear stiffeners, respectively. The stiffened test panels with six stiffeners have been designed using multi-objective design optimization framework for TBL excitation. / Ph. D.

Le bruit de roulement est la principale source de bruit des transports ferroviaires pour des vitesses conventionnelles. Il est produit par les vibrations puis le rayonnement acoustique du matériel roulant et de la voie provoqués par le déplacement de la roue et du rail. Ce déplacement est imposé par les défauts microscopiques de surface présents sur les bandes de roulement appelés rugosité.La directive Européenne 2002/49/CE exige la réalisation la cartographies du bruit des infrastructures de transport. Le bruit dans l’environnement peut être prédit par des modèles dont les paramètres d’entrées sont notamment le comportement dynamique de la voie et la rugosité des rails. Ces données sont généralement évaluées par des systèmes de mesures statiques, ce qui limite la caractérisation de la voie à quelques dizaines ou centaines de mètres.Cette thèse a pour objectif de proposer et de valider une méthodologie de mesure de l'état acoustique de la voie sur l’ensemble du réseau. La solution proposée est un système de mesure en situation embarquée nommé LECAV pour « LECture Acoustique de la Voie », qui utilise le bruit rayonné par la voir et par la roue. Leurs gammes de fréquences de rayonnement distinctes permettent de traiter le problème inverse séparément pour la roue et pour le rail, et ceux à partir de l’impédance acoustique.Un modèle vibroacoustique basé sur la méthode des éléments finis et la formulation intégrale de Rayleigh est proposé afin d’estimer l’impédance acoustique de la roue. Pour le rail, les théories des poutres épaisses et des sources ponctuelles élémentaires sont utilisées dans le but de modéliser son impédance acoustique. / The rolling noise is the main source of noise in the railway transport sector at conventional speed. It is produced by the vibration and the acoustic radiation of the vehicle and the railway track, caused by the movement of at the wheel and the rail. These movements are imposed by microscopic irregularities on the rolling surface called roughness.The 2002/49/EC European directive imposes to build noise maps of transport infrastructure. The noise propagating in the environment can be assessed from models whose input parameters are particularly the dynamic behavior of railway tracks and the roughness of rails. These data are generally measured thanks to static systems, which narrow down to the analysis of railway tracks to a few meters.The aim of this Phd is to propose and validate a measurement methodology for the acoustic characterization of tracks for the whole railway network. It is an on-board measurement system named LECAV for "LECture Acoustique de la Voie" in French ("Acoustic reading of railway track") that uses the noise radiated from the wheel and the rail. The respective frequencies ranges of noise emission allow solving the invers problem separately for the wheel and for the rail, and this with the acoustic impedance.