Résumé : Ce projet de recherche est composé de trois parties principales : la première comprend l'analyse expérimentale et la simulation des performances vibratoires de matériaux avec amortissement viscoélastique, en tant que traitements acoustiques appliqués aux structures du fuselage d'un avion. La deuxième partie comprend l'analyse expérimentale et le calcul de la performance acoustique de ces matériaux amortissant en comparaison avec l'effet d'une masse équivalente. Enfin, la troisième partie est une étude paramétrique sur les effets de localisation, de la densité et de la taille d'un traitement massique. Les systèmes d'isolation phoniques typiquement employés dans la construction des fuselages d'avions sont composés de matériaux poreux, avec ou sans des matériaux amortissant (matériaux viscoélastiques). La performance et donc l'utilité de ces traitements amortissant, en comparaison avec une couche de masse équivalente, reste une question largement ouverte. Dans ce travail on a comparé numériquement et expérimentalement les performances acoustiques d'un traitement amortissant avec celui d'une masse équivalente tous les deux incorporées dans le traitement phonique et ceci pour plusieurs types d'excitations. Deux structures représentant des fuselages, une en aluminium et la seconde en carbone composite, ont été sélectionnées pour cette étude ainsi que deux matériaux poreux couramment utilisés en aéronautique : une laine en fibre de verre et une mousse à cellules ouvertes. Deux types d'excitations ont été étudiés numériquement et expérimentalement. La première est une excitation acoustique (champ diffus) et la seconde mécanique (forces ponctuelles). Une troisième, excitation par couche limite turbulente a été étudié numériquement. Dans tous les cas, la perte par insertion du traitement acoustique est utilisée comme indicateur principal de la performance. D'autres indicateurs comme le coefficient d'absorption, le coefficient de perte par amortissement et la vitesse quadratique moyenne sont aussi utilisés pour mieux cerner et expliquer l'effet du traitement. Il a été démontré que l'utilisation d'une couche de masse équivalente à un traitement viscoélastique conduit systématiquement à la meilleure performance acoustique et ceci pour les trois types d'excitations étudiées. En particulier, dans le cas classique où le traitement amortissant est appliqué directement au fuselage. Dans ce dernier cas, les effets de doubles parois créés par la couche massique, positionnée judicieusement loin du fuselage, augmentent la performance en moyennes et hautes fréquences. Les performances en basses fréquences restent limitées par la fréquence de résonance double parois. Et même l'effet amortissant des traitements viscoélastiques, théoriquement visible aux résonances et coïncidences du système, se trouve limité par l'amortissement ajouté par le montage et le traitement absorbants. Cependant, l'efficacité de la couche massique est compromise par les difficultés d'installations et en particulier pour les fibreux. Les résultats de cette thèse restent toutefois limités par notre choix de structures et de traitements étudiés. / Abstract : The project is made up of three main parts. The first part involves a comprehensive experimental and numerical analysis of viscoelastic damping materials as acoustic treatments to aircraft fuselage structures. The second part involves numerical and experimental acoustic comparison of viscoelastic damping material to equivalent mass. And the third part is a parametric study of equivalent mass for the effects of mass location, density and size. The goal of the project is to identify the vibroacoustic effect of viscoelastic material damping of fuselage skin, and develop possible alternatives to damping. The insulation systems (typically used on aircraft) that are made up of porous materials with or without viscoelastic damping material or equivalent mass were called sound packages throughout this document. The viscoelastic damping material and equivalent mass both incorporated in sound packages were acoustically compared. Fiberglass and open cell foam were used as porous materials. The viscoelastic damping material used in this study is constraining layer damping and abbreviated as CLD. The equivalent mass was an impervious screen. Both representative Aluminum and carbon composite fuselage skin structures were treated with sound packages as part of the comparison. The vibroacoustic performance indicators were used to characterize the sound packages. The indicators were airborne insertion loss (ABIL), structure borne insertion loss (SBIL), turbulent boundary layer insertion loss (TBLIL), average quadratic velocity (AQV), damping loss factor (DLF), absorption coefficient, and radiation efficiency. Diffuse field acoustic excitation was used to obtain the vibroacoustic indicators of ABIL and absorption coefficient. Mechanical excitation was used to obtain SBIL, AQV, DLF, absorption coefficient, and radiation efficiency. Turbulent boundary layer excitation was modeled to obtain TBLIL. The numerical methods of finite element method (FEM) and transfer matrix method (TMM) were used to calculate all of the above vibroacoustic performance indicators. Experimentally, ABIL, SBIL, AQV, DLF and radiation efficiency were measured. Experimental modal analysis was also performed to characterize representative Aluminum and carbon composite fuselage skin structures. Based on the numerical analysis, equivalent mass generated a double or multiple (in case of double wall layer configuration) wall effect and hence became an effective acoustic insulator as part of sound packages at mid to high frequencies. Even at coincidence frequencies (in case of the representative carbon composite fuselage skin), the equivalent mass layer was more effective than viscoelastic damping material. However, the drawback was the occurrence of the double wall resonance at lower frequencies which compromised the effectiveness. Nevertheless, the parametric study of equivalent mass revealed that equivalent mass is superior to viscoelastic damping material at reduced weight in term of vibroacoustic performance indicators of overall ABIL/SBIL/TBLIL in the frequency range of 100 to 6300 Hz and mean ABIL/SBIL/TBLIL in SIL (octave lk, 2k, 4k Hz) frequency range.

<p>Asymptotic Modal Analysis (AMA) is a computationally efficient and accurate method for studying the response of dynamical systems experiencing banded, random harmonic excitation at high frequencies when the number of responding modes is large. In this work, AMA has been extended to systems of coupled continuous components as well as nonlinear systems. Several prototypical cases are considered to advance the technique from the current state-of-the-art. The nonlinear problem is considered in two steps. First, a method for solving problems involving nonlinear continuous multi-mode components, called Iterative Modal Analysis (IMA), is outlined. Secondly, the behavior of a plate carrying a nonlinear spring-mass system is studied, showing how nonlinear effects on system natural frequencies may be accounted for in AMA. The final chapters of this work consider the coupling of continuous systems. For example, two parallel plates coupled at a point are studied. The principal novel element of the two-plate investigation reduces transfer function sums of the coupled system to an analytic form in the AMA approximation. Secondly, a stack of three parallel plates where adjacent plates are coupled at a point are examined. The three-plate investigation refines the reduction of transfer function sums, studies spatial intensification in greater detail, and offers insight into the diminishing response amplitudes in networks of continuous components excited at one location. These chapters open the door for future work in networks of vibrating components responding to banded, high-frequency, random harmonic excitation in the linear and nonlinear regimes.</p> / Dissertation

La méthode SEA (Statistical Energy Analysis) est une approche statistique de la vibroacoustique permettant de décrire les systèmes complexes en termes d'échanges d'énergies vibratoires et acoustiques. En moyennes et hautes fréquences, cette méthode se présente comme une alternative aux méthodes déterministes (coût des calculs dû au grand nombre de modes, de degrés de liberté, unicité de la solution) Néanmoins, son utilisation requiert la connaissance et le respect d'hypothèses fortes qui limitent son domaine d'application. Dans ce mémoire, les fondements de la SEA ont été examinés afin de discuter chaque hypothèse. Le champ diffus, l'équipartition de l’énergie modale, le couplage faible, l'influence des modes non résonants et l'excitation rain-on-the-roof sont les cinq hypothèses qui ont été abordées. Sur la base d'exemples simples (oscillateurs couplés, plaques couplées), les équivalences et leurs influences sur la qualité des résultats ont été étudiées pour contribuer à la clarification des hypothèses nécessaires à l'application de la SEA ct pour borner son domaine de validité SEA. / Statistical energy analysis is a statistical approach of vibroacoustics which allows to describe complex systems in terms of vibrational or acoustical energies. ln the high frequency range, this method constitutes an alternative to bypass the problems which can occur when applying deterministic methods (computation cost due to the large number of modes, the large number of degrees of freedom and the unicity of the solution). But SEA has numerous assumptions which are sometimes forgotten or misunderstood ln this thesis, foundations of SEA have been examined in order to discuss each assumption. Diffuse field, modal energy equipartition, weak coupling, the influence of non-resonant modes and the rain on the roof excitation are the five look up hypotheses. Based on simple examples (coupled oscillators, coupled plates), the possible equivalences and their influence on the quality of the results have been discussed to contribute to the clarification of the useful SEA assumptions and to mark out it's the validity domain.

Este trabalho consiste na aplicação de técnicas de identificação de parâmetros estruturais, especificamente massa, rigidez e amortecimento, à caixa acústica de um violão clássico. A abordagem experimental tradicional é adotada, com medição de aceleração em um ponto da estrutura e excitação proveniente de martelo instrumentado registrada em diferentes pontos. As funções de resposta em frequência são então traçadas para, em conjunto com um modelo analítico representativo dos primeiros graus de liberdade do acoplamento ar/estrutura, realizar-se o ajuste de curvas e posterior identificação dinâmica. Para tanto, esta última etapa foi tratada como um problema matemático de otimização não linear no qual se busca a minimização de uma função objetivo que contabiliza de alguma forma a diferença entre o resultado das medições e os valores previstos pelo modelo utilizado. Várias funções de erro (objetivo) e vários algoritmos foram utilizados, como mínimos quadrados não linear, simplex de Nealder-Mead, algoritmo genético padrão e enxame de partículas (particle swarm), entre outros, sendo este último o de melhor desempenho entre todos, quando aplicado em conjunto com a norma da diferença dos logaritmos das magnitudes ao quadrado. Assim, uma calibração com boa concordância entre dados experimentais e resultados teóricos foi estabelecida para o modelo proposto, sendo realizada também a verificação do ajuste através da comparação de um conjunto independente de dados, trazendo, desta forma, confiabilidade para posteriores cálculos das grandezas associadas ao comportamento dinâmico utilizando-se o modelo ajustado. / This work aims at identify structural parameters of a classical acoustic guitar’s resonant chamber by comparing theoretical and experimental frequency response functions. The quantities used to construct the mass, stiffness and damping matrices of an air/structure analytical model, which is representative of the first few modes of the body, are considered as project variables, and an impact vibration testing approach is used to obtain the measured data, with a roving instrumented hammer and an accelerometer attached to a predefined point of the body. Then, a curve-fit analysis is performed as a mathematical problem of non-linear optimization, wherein the objective function, which is to be minimized, somehow accounts for the difference between the measured data and the theoretically predicted values. Some different error (objective) functions and optimization algorithms, like non-linear least-squares, Nealder-Mead simplex, standard genetic algorithm and particle swarm, among others, were applied, and the latter yielded, together with the squared error norm, the best performance. Then, an updated model was achieved with fair agreement between analytical predictions and experimental results, verified using a validation data set, bringing reliability for further theoretical predictions.

Este trabalho consiste na aplicação de técnicas de identificação de parâmetros estruturais, especificamente massa, rigidez e amortecimento, à caixa acústica de um violão clássico. A abordagem experimental tradicional é adotada, com medição de aceleração em um ponto da estrutura e excitação proveniente de martelo instrumentado registrada em diferentes pontos. As funções de resposta em frequência são então traçadas para, em conjunto com um modelo analítico representativo dos primeiros graus de liberdade do acoplamento ar/estrutura, realizar-se o ajuste de curvas e posterior identificação dinâmica. Para tanto, esta última etapa foi tratada como um problema matemático de otimização não linear no qual se busca a minimização de uma função objetivo que contabiliza de alguma forma a diferença entre o resultado das medições e os valores previstos pelo modelo utilizado. Várias funções de erro (objetivo) e vários algoritmos foram utilizados, como mínimos quadrados não linear, simplex de Nealder-Mead, algoritmo genético padrão e enxame de partículas (particle swarm), entre outros, sendo este último o de melhor desempenho entre todos, quando aplicado em conjunto com a norma da diferença dos logaritmos das magnitudes ao quadrado. Assim, uma calibração com boa concordância entre dados experimentais e resultados teóricos foi estabelecida para o modelo proposto, sendo realizada também a verificação do ajuste através da comparação de um conjunto independente de dados, trazendo, desta forma, confiabilidade para posteriores cálculos das grandezas associadas ao comportamento dinâmico utilizando-se o modelo ajustado. / This work aims at identify structural parameters of a classical acoustic guitar’s resonant chamber by comparing theoretical and experimental frequency response functions. The quantities used to construct the mass, stiffness and damping matrices of an air/structure analytical model, which is representative of the first few modes of the body, are considered as project variables, and an impact vibration testing approach is used to obtain the measured data, with a roving instrumented hammer and an accelerometer attached to a predefined point of the body. Then, a curve-fit analysis is performed as a mathematical problem of non-linear optimization, wherein the objective function, which is to be minimized, somehow accounts for the difference between the measured data and the theoretically predicted values. Some different error (objective) functions and optimization algorithms, like non-linear least-squares, Nealder-Mead simplex, standard genetic algorithm and particle swarm, among others, were applied, and the latter yielded, together with the squared error norm, the best performance. Then, an updated model was achieved with fair agreement between analytical predictions and experimental results, verified using a validation data set, bringing reliability for further theoretical predictions.

Este trabalho consiste na aplicação de técnicas de identificação de parâmetros estruturais, especificamente massa, rigidez e amortecimento, à caixa acústica de um violão clássico. A abordagem experimental tradicional é adotada, com medição de aceleração em um ponto da estrutura e excitação proveniente de martelo instrumentado registrada em diferentes pontos. As funções de resposta em frequência são então traçadas para, em conjunto com um modelo analítico representativo dos primeiros graus de liberdade do acoplamento ar/estrutura, realizar-se o ajuste de curvas e posterior identificação dinâmica. Para tanto, esta última etapa foi tratada como um problema matemático de otimização não linear no qual se busca a minimização de uma função objetivo que contabiliza de alguma forma a diferença entre o resultado das medições e os valores previstos pelo modelo utilizado. Várias funções de erro (objetivo) e vários algoritmos foram utilizados, como mínimos quadrados não linear, simplex de Nealder-Mead, algoritmo genético padrão e enxame de partículas (particle swarm), entre outros, sendo este último o de melhor desempenho entre todos, quando aplicado em conjunto com a norma da diferença dos logaritmos das magnitudes ao quadrado. Assim, uma calibração com boa concordância entre dados experimentais e resultados teóricos foi estabelecida para o modelo proposto, sendo realizada também a verificação do ajuste através da comparação de um conjunto independente de dados, trazendo, desta forma, confiabilidade para posteriores cálculos das grandezas associadas ao comportamento dinâmico utilizando-se o modelo ajustado. / This work aims at identify structural parameters of a classical acoustic guitar’s resonant chamber by comparing theoretical and experimental frequency response functions. The quantities used to construct the mass, stiffness and damping matrices of an air/structure analytical model, which is representative of the first few modes of the body, are considered as project variables, and an impact vibration testing approach is used to obtain the measured data, with a roving instrumented hammer and an accelerometer attached to a predefined point of the body. Then, a curve-fit analysis is performed as a mathematical problem of non-linear optimization, wherein the objective function, which is to be minimized, somehow accounts for the difference between the measured data and the theoretically predicted values. Some different error (objective) functions and optimization algorithms, like non-linear least-squares, Nealder-Mead simplex, standard genetic algorithm and particle swarm, among others, were applied, and the latter yielded, together with the squared error norm, the best performance. Then, an updated model was achieved with fair agreement between analytical predictions and experimental results, verified using a validation data set, bringing reliability for further theoretical predictions.

L’industrie aérospatiale doit faire face à de nouvelles exigences environnementales, tout particulièrement concernant la réduction des coûts de lancement. L’utilisation de matériaux sandwichs composites plus légers permet de répondre à ces besoins. Cependant, l’allégement des matériaux sandwichs favorise une transmission importante du bruit, d’où la nécessité de prendre en compte des critères vibroacoustiques dès la phase de préconception. Dans cette optique, le travail présenté dans ce mémoire a pour objectif de proposer une démarche d’optimisation vibroacoustique des panneaux sandwichs composites légers, sous contraintes de masse et rigidité. Une étude spécifique est consacrée à l’optimisation des variables géométriques de solides cellulaires à périodicité de type nid d’abeille. L’objectif principal est de minimiser la densité modale en s’appuyant sur des modèles homogénéisés fiables. Afin de calculer les propriétés mécaniques macroscopiques des panneaux sandwichs, une méthode numérique d’homogénéisation tridimensionnelle est développée. Cette méthode permet de calculer les propriétés mécaniques équivalentes en utilisant les déformations et contraintes moyennes sur le volume représentatif. Les résultats obtenus sont conformes à ceux calculés par des méthodes classiques basées sur des modèles analytiques. Dans le but d’identifier une fonction objectif riche en informations sur le comportement vibroacoustique de panneau sandwich, on choisit d’étudier la densité modale du panneau. Par la suite, la fréquence de transition, qui sépare la zone de comportement de flexion pure du panneau sandwich du comportement en cisaillement pur de l’âme, est utilisée pour définir la fonction objectif. Après une étude d’analyse de sensibilité sur les paramètres mécaniques et géométriques de la structure sandwich, une démarche globale d’optimisation mono-objectif est mise en oeuvre pour maximiser la fréquence de transition de la structure sandwich composite constituée d’une âme en nid d’abeille hexagonale. Enfin, cette démarche est étendue pour estimer les propriétés géométriques optimales de nouvelles âmes. / The aerospace industry has to adapt to new environmental requirements, especially concerning the reduction of the launch costs. The use of lighter composite sandwich materials can meet part of these requirements. However, their high stiffness-toweight ratio implies that they tend to increase noise transmission, which may be damageable to the payload. Vibroacoustic criterai should hence be taken into account from the early design stages. In this context, the work presented in this thesis aims to provide a vibroacoustic optimization approach of lightweight composite sandwich panels, under mass and stiffness constraints. A specific study is devoted to the optimization of geometric variables of periodic cellular solids such as honeycombs. The main objective is to minimize the modal density based on reliable homogenized models. In order to calculate the macroscopic mechanical properties of the sandwich panel, a numerical method of three-dimensional homogenization is developed. This method allows to calculate the equivalent mechanical properties by applying the average strains and stresses on a unit cell. The results obtained are consistent with those calculated by conventional methods based on analytical approaches. The modal density is chosen as an objective function for optimization, as it is closely related to the vibroacoustic behavior of the structure. The transition frequency, which separates the region of pure panel bending from the pure core shear zone, is further studied and considered as an alternative objective function. After a sensitivity analysis of the mechanical and geometric parameters of the sandwich structure, a mono-objective optimization approach is implemented to maximize the transition frequency of a composite sandwich structure with a hexagonal core. This approach is then extended to estimate the optimal geometric properties of new core shapes.

Vibrações estruturais constituem uma das principais fontes de ruído e de desconforto em diversas aplicações, tais como nos produtos das indústrias automotiva, aeroespacial e naval, que, em grande parte, são compostos por estruturas flexíveis semelhantes a placas. Ao mesmo tempo, a demanda crescente por aumento de eficiência energética e por redução de emissão de gases nocivos ao meio ambiente tem levado os fabricantes de todos os segmentos para a utilização de materiais cada vez mais leves. Um desafio importante relacionado ao uso dessas soluções diz respeito ao seu comportamento vibroacústico, já que estruturas leves tendem a transmitir e emitir ruídos de forma mais intensa, particularmente em baixas frequências. Em decorrência disso, a obtenção de uma boa relação entre peso e desempenho vibroacústico usualmente demanda algumas iterações de projeto. Nesse contexto, este trabalho tem por objetivo estabelecer um procedimento numérico para avaliar as propriedades vibroacústicas de placas flexíveis excitadas por um campo gerado no interior de uma cavidade acústica, bem como investigar a influência de parâmetros estruturais que definem a placa sobre o seu desempenho vibroacústico. O escopo aqui é a faixa de baixas frequências, na qual o comportamento do sistema é descrito de forma determinística através de características modais. Foi desenvolvido um modelo numérico em elementos finitos de uma cavidade acústica retangular de paredes rígidas acoplada a uma placa flexível. Uma fonte acústica posicionada no interior da cavidade produz um campo de pressão sonora que induz a placa à vibração e, consequentemente, a irradiar potência sonora para o campo livre exterior. Para realização das análises de sensibilidade, desenvolveu-se um conjunto de rotinas de modo a permitir o gerenciamento automatizado de simulações. Resultados numéricos são apresentados para avaliar a influência de diversos parâmetros estruturais sobre o desempenho vibroacústico da superfície estrutural. O procedimento proposto pode ser aplicado de forma eficiente para a análise de sensibilidade da transmissão de ruído em sistemas vibroacústicos sujeitos à variação de parâmetros estruturais e pode também ser utilizado para avaliar diversos tipos de materiais, como isotrópicos, ortotrópicos e até metamateriais. / Structural vibrations are the major causes of noise and people discomforts in several applications, such as in the products of the automotive, aerospace and naval industries, which are mainly composed of flexible plate-like structures. At the same time, the ever-increasing demands for energy efficiency and for reduced emissions of harmful gases into the environment lead product manufacturers of all industrial sectors towards an increased use of lighter materials. However, these solutions are most sensitive to issues related to vibroacoustic behavior, since the lightweight structures tend to present a high sound transmission, particularly at low frequencies. Consequently, the optimal design of lightweight and noise insulation properties of a structure usually require some design trade-offs. In this context, this dissertation aims to propose a numerical procedure to analyze the vibroacoustic performance of flexible plates subject to an interior acoustic field, as well as to investigate the influence of structural parameters on the sound insulation properties of the plate. The scope here is the low frequency range, in which the response of the coupled system is described in a deterministic way in terms of a mode set. A finite element model of a rigid rectangular acoustic cavity coupled to a flexible plate was developed. An acoustic source inside the cavity produces an interior sound pressure field that forces the flexible plate to vibrate and, consequently, to radiate sound power to the external free field. In order to enable the automatic management of the variability simulations, a set of routines was also developed. Numerical results are presented in order to evaluate the influence of several structural parameters on the vibroacoustic properties of the plate. The proposed methodology can be largely used in sensitivity analysis of noise transmission in vibroacoustic systems subject to the variation of structural parameters and can be used to evaluate many material types, as isotropic, orthotropic and even metamaterials.

Vibrações estruturais constituem uma das principais fontes de ruído e de desconforto em diversas aplicações, tais como nos produtos das indústrias automotiva, aeroespacial e naval, que, em grande parte, são compostos por estruturas flexíveis semelhantes a placas. Ao mesmo tempo, a demanda crescente por aumento de eficiência energética e por redução de emissão de gases nocivos ao meio ambiente tem levado os fabricantes de todos os segmentos para a utilização de materiais cada vez mais leves. Um desafio importante relacionado ao uso dessas soluções diz respeito ao seu comportamento vibroacústico, já que estruturas leves tendem a transmitir e emitir ruídos de forma mais intensa, particularmente em baixas frequências. Em decorrência disso, a obtenção de uma boa relação entre peso e desempenho vibroacústico usualmente demanda algumas iterações de projeto. Nesse contexto, este trabalho tem por objetivo estabelecer um procedimento numérico para avaliar as propriedades vibroacústicas de placas flexíveis excitadas por um campo gerado no interior de uma cavidade acústica, bem como investigar a influência de parâmetros estruturais que definem a placa sobre o seu desempenho vibroacústico. O escopo aqui é a faixa de baixas frequências, na qual o comportamento do sistema é descrito de forma determinística através de características modais. Foi desenvolvido um modelo numérico em elementos finitos de uma cavidade acústica retangular de paredes rígidas acoplada a uma placa flexível. Uma fonte acústica posicionada no interior da cavidade produz um campo de pressão sonora que induz a placa à vibração e, consequentemente, a irradiar potência sonora para o campo livre exterior. Para realização das análises de sensibilidade, desenvolveu-se um conjunto de rotinas de modo a permitir o gerenciamento automatizado de simulações. Resultados numéricos são apresentados para avaliar a influência de diversos parâmetros estruturais sobre o desempenho vibroacústico da superfície estrutural. O procedimento proposto pode ser aplicado de forma eficiente para a análise de sensibilidade da transmissão de ruído em sistemas vibroacústicos sujeitos à variação de parâmetros estruturais e pode também ser utilizado para avaliar diversos tipos de materiais, como isotrópicos, ortotrópicos e até metamateriais. / Structural vibrations are the major causes of noise and people discomforts in several applications, such as in the products of the automotive, aerospace and naval industries, which are mainly composed of flexible plate-like structures. At the same time, the ever-increasing demands for energy efficiency and for reduced emissions of harmful gases into the environment lead product manufacturers of all industrial sectors towards an increased use of lighter materials. However, these solutions are most sensitive to issues related to vibroacoustic behavior, since the lightweight structures tend to present a high sound transmission, particularly at low frequencies. Consequently, the optimal design of lightweight and noise insulation properties of a structure usually require some design trade-offs. In this context, this dissertation aims to propose a numerical procedure to analyze the vibroacoustic performance of flexible plates subject to an interior acoustic field, as well as to investigate the influence of structural parameters on the sound insulation properties of the plate. The scope here is the low frequency range, in which the response of the coupled system is described in a deterministic way in terms of a mode set. A finite element model of a rigid rectangular acoustic cavity coupled to a flexible plate was developed. An acoustic source inside the cavity produces an interior sound pressure field that forces the flexible plate to vibrate and, consequently, to radiate sound power to the external free field. In order to enable the automatic management of the variability simulations, a set of routines was also developed. Numerical results are presented in order to evaluate the influence of several structural parameters on the vibroacoustic properties of the plate. The proposed methodology can be largely used in sensitivity analysis of noise transmission in vibroacoustic systems subject to the variation of structural parameters and can be used to evaluate many material types, as isotropic, orthotropic and even metamaterials.

L'utilisation de plus en plus fréquente de matériau composite, qui combine une raideur importante pour une faible masse, afin d'alléger les structures aéronautiques entraîne la dégradation des performances d'isolation acoustique aux bruits extérieurs. La plupart du temps, ces nuisances sonores sont réduites par l'installation de matériaux isolants. Ces méthodes, dites passives, deviennent inefficaces aux basses fréquences et il est possible de mettre en place un contrôle actif au moyen de transducteurs électromécaniques. Dans le but de réduire la puissance acoustique transmise à travers la double paroi aéronautique dans la cabine, des unités de contrôle composées d'un actionneur et d'un capteur colocalisé dual sont réparties sur le panneau intérieur afin d'en modifier la vibration. Cette stratégie de contrôle actif vibroacoustique permet, pour des perturbations primaires harmoniques, d'imposer localement une impédance mécanique virtuelle à la structure, au moyen d'un contrôleur décentralisé. Cependant, sans communication entre les unités, le contrôle peut difficilement minimiser un critère global comme la puissance acoustique rayonnée. Afin de calculer les impédances mécaniques virtuelles qui garantissent la minimisation de la puissance acoustique rayonnée par la structure, une approche en deux étapes est considérée : (1) la matrice diagonale des impédances mécaniques virtuelles optimales est calculée à partir de mesures acoustiques ou vibratoires de la perturbation primaire et des transferts avec les actionneurs secondaires, (2) l'objectif exprimé en terme d'impédances mécaniques virtuelles est atteint grâce à un contrôle en temps réel. Une attention particulière est portée à la comparaison de cette approche avec une stratégie classique d'amortissement actif réalisée par un contrôle par rétroaction sur la vitesse de la structure, où l'impédance mécanique virtuelle alors imposée est un réel positif. Le calcul optimal réalisé à l'issue de la première étape se faisant pour une perturbation primaire donnée, la robustesse de la méthode aux variations de la perturbation primaire est un aspect également développé dans cette étude. Des résultats théoriques et expérimentaux sont comparés dans le cas académique d'une plaque mince d'aluminium simplement appuyée et soumise à une onde plane incidente. Enfin, la méthode est appliquée au panneau intérieur d'une double paroi aéronautique, à savoir une structure courbée, en matériau composite, et composée d'un hublot. Contrairement à la majorité des études qui considèrent l'implantation d'impédances virtuelles dissipatives, il apparaît que dans certains cas, le contrôle optimal requiert l'injection d'énergie des unités à la structure. / Composite materials are widely used in the aeronautic industry for their low mass/stiffness ratio. However, this property tends to reduce the acoustic transmission loss, particularly at low frequencies. At these frequencies, active control is an effective mean of controlling sound transmission. Among the various approaches, Active Structural Acoustic Control (ASAC) has received considerable attention because transducers can be integrated to the structure. In order to reduce the acoustic power radiated by a flexible panel, dual colocated actuator sensor pairs are used to modify its vibration. The control strategy implemented for harmonic disturbances leads to locally impose a virtual mechanical impedance to the structure, using a decentralized controller. This virtual mechanical impedance is computed in order to minimise the radiated acoustic power. The challenging problem is then to find the local control to impose on each independent devices that minimizes the global acoustic radiation of the structure. The proposed approach consists in two steps : (1) the matrix of optimal virtual mechanical impedance is calculated by measuring the primary disturbance and the transfer functions between actuators and structural / acoustic sensors, (2) the virtual mechanical impedance objective is achieved using a real-time integral controller. Special focus is put on the discussion about such control approach versus a classical active damping strategy were the virtual mechanical impedance is defined as real positive. Considering that optimal control is computed during the first step for a given primary disturbance, the robustness of the method to variations of the primary disturbance between step 1 and step 2 is discussed. Theoretical and experimental results are compared in the case of a simply supported thin aluminum plate and a primary disturbance under the form of an incident plane wave. Then, the method is implemented on a curved composite aircraft panel comprising a window. Unlike most of previous studies where dissipative virtual mechanical impedance are imposed, it clearly appears that optimal control can require energy injection from the control units into the structure.