In dynamic substructuring, a complex structure is divided into multiple substructures that can be analysed individually and these individual component responses are coupled together to obtain the global response of the whole structure. Dynamic substructuring can be performed on substructure models that are identified either experimentally or analytically. For dynamic substructuring to be successful, it is very essential to have the precise information of the connection points or the interfaces between the substructures. The method has been extensively used with analytical models in most of the available standard finite element software packages where the information about all degrees of freedom is known. However, it is difficult to get the information about all connection degrees of freedom from the measurements and experimental substructuring is thus limited in its use compared to analytical substructuring. In order to overcome these difficulties, the Transmission Simulator method commonly also known as Modal Constraints for Fixture and Subsystem method can be used. In this method, an additional fixture called Transmission Simulator which is available both physically and analytically, is attached to the substructures at the interfaces and their respective responses are measured. The substructures could be analytical as well as experimental. The coupling is done by constraining the transmission simulator on the substructures to have the same motion and the effect of the transmission simulator is later removed from the coupled structure by subtracting the analytical transmission simulator model. This method has been successfully implemented for Component Mode Synthesis and Frequency Based Substructuring for structures with multiple connection points at a single location. In this thesis work, frequency response function based experimental–analytical dynamic substructuring using the transmission simulator is performed on a rear subframe and rear differential unit assembly of a Volvo XC90 car where the differential unit is connected to the subframe at three locations. The aim of this work is to verify the Transmission Simulator Method for multiple location connection points using the frequency response functions and build confidence on the methodology in order to be used for future work at Volvo Car Corporation.

Dynamic Substructuring is a powerful tool for simplification of the analysis of complex structures and it has been well established along the years in analytical calculations by means of the Craig-Bampton technique. Recently, a new branch of substructuring, the Experimental Dynamic Substrucuring, appeared as a promising field of research for the engineering community. This area presents several intrinsic difficulties, evincing a need to develop the traditional substructuring methods towards obtaining better results using the experimental approach. In this scenery, the Transmission Simulator technique emerges as an instrument for potential improvement of the achieved results. This work represents a study on the use of the Transmission Simulator technique in the analysis of an Ampair A600 wind turbine blade subjected to loads at the interface to the hub, and it is a part of the benchmarking studies of SEM (Society of Experimental Mechanics). The work consisted of collecting experimental data via vibration tests of a single blade connected to different sizes of transmission simulators. After that, a mathematical representation of the blade was obtained via subtraction of the effect of the transmission simulators via substructuring technique. The computed model was subsequently coupled to a model of the remainder of the wind turbine (the hub plus two blades), and the results were compared to data acquired in tests of the whole assembly. The final findings did not reflect the theory prospects and further investigation is necessary to evaluate the effectiveness of the used methodology.

Dans cette thèse, nous nous sommes intéressés au comportement vibratoire d’un véhicule soumis à une excitation moteur dans deux plages de régimes différentes : basses fréquences (0 – 50 Hz) et moyennes fréquences (200 – 800 Hz). Le but était de fournir des méthodologies numériques permettant de prendre en compte les phénomènes de couplage vibratoires existant entre les différents sous-systèmes constitutifs d’une caisse automobile.En basses fréquences, nous avons adopté une approche globale où chaque sous-système a été caractérisé séparément. Tout d’abord, le comportement de la caisse a été caractérisé expérimentalement et numériquement par une méthode jusqu’alors réservée au domaine aéronautique dite d’appropriation modale. Les résultats numériques ont alors été confrontés aux résultats expérimentaux. Par ailleurs, le comportement non-linéaire en amplitude et en fréquence des pièces de filtration moteur (SMO) a été déterminé sur banc de mesure. Un fort comportement non-linéaire a pu être constaté et ces caractérisations ont été exploitées en construisant des nappes raideur-fréquence-amplitude. Dans un second temps, des méthodes numériques permettant de réaliser l’assemblage non - linéaire de la caisse et du groupe moto – propulseur (GMP) via les pièces de filtration non linéaires ont été mises en place. Pour ce faire, nous avons développé une méthode dite de Balance Harmonique (HBM) qui permet de prédire la dynamique non-linéaire de systèmes complexes. Afin d’appliquer cette méthode à une structure industrielle, nous avons utilisé une méthode de condensation sur les degrés de liberté non-linéaires, technique bien adaptée aux cas de structures linéaires reliées localement par des éléments de liaison non-linéaires. Cette méthode a tout d’abord été validée sur un périmètre restreint comprenant un GMP relié à un bâti rigide par ses pièces de filtration. A cette occasion, des phénomènes non-linéaires importants ont été constatés expérimentalement. Un modèle numérique de GMP a été construit et l’utilisation de la méthode HBM a permis de retrouver ces constats. Enfin, après avoir réalisé l’assemblage non-linéaire des trois sous-systèmes GMP - SMO - Caisse, la structure a été excitée de plusieurs manières différentes : appropriation numérique non-linéaire et excitation réelle d’un GMP. En moyennes fréquences, nous avons présenté dans ce mémoire une étude importante pour le groupe Renault concernant la caractérisation des pièces de filtration en moyennes fréquences. Au cours de la thèse, une méthodologie numérique basée sur la méthode FBS permettant de déconfiner (ou découpler) une suspension moteur initialement reliée à un banc de mesure a été proposée. La faisabilité numérique du déconfinement a ainsi été démontrée. Cette méthode permet donc, en dépit de phénomènes de couplage avec le banc de mesure, d’obtenir le comportement vibratoire intrinsèque de la pièce. / In this thesis we studied the vibratory behaviour of a whole vehicle under engine excitation at low frequencies (0 – 50 Hz) and medium frequencies (200 – 800 Hz). The aim of the thesis was to provide numerical methodologies to take into account coupling effects between all the sub-systems constituting a whole car. In low frequencies, we used a global approach where each subsystem was characterized separately before coupling. First the car body was characterised both experimentally and numerically using a modal appropriation method that is commonly used in the aeronautic field. Numerical shapes of the modes were correlated to experimental shapes. In addition, the amplitude and frequency non linear behaviour of the engine mounts was measured on a test bench. A strong non linear behaviour was observed and stiffness – frequency – amplitude layers were constructed based on those data.Secondly, numerical methods were developed in order to calculate the coupled non linear response between the engine, the engine mounts and the car body. We used a harmonic balance method that allows calculating the non linear dynamics of complex mechanical systems. In order to apply this method to large industrial finite element models, a condensation method on non linear degrees of freedom was developed. This technique is well adapted to problems of linear structures linked together with localnon linear joints. This method was validated on the isolated engine linked to a bench by the engine mounts. Strong non linear phenomena on the rigid body modes of the engine were observed experimentally.A numerical model of the engine was developed and the HBM method allowed reproducing these non linear phenomena. Eventually, the non linear model of the whole vehicle was coupled and excited by different efforts. First we calculated the response of the assembly using the appropriation method. Then, the structure was excited by a real four – cylinder engine excitation.In medium frequencies, we presented an important study for the group Renault concerning the stiffness measurement of the engine mounts. A numerical methodology based on the FRF Based Substructuring(FBS) method was developed. This method was applied to uncouple an engine mount initially coupled to a test bench. The numerical feasibility of the method was proved and allowed to get the own vibratory behaviour of the engine mount despite coupling phenomena with the test bench.