Modelisation et optimisation d'isolateurs non-lineaires en elastomeres

The main purpose of this research is to develop a numeric tool to optimise engine mounting systems. The design method proposed is intented to be the first step of the design of new engine mounts and reduce the number of iterations and prototypes.;A selected engine mount was caracterized using a harmonic traction machine. The engine mount was tested over a wide range of deformations, frequencies an orientations according observations on the vehicle. Depending on the frequencies and the displacements, significant differences in the stiffness (20-30%) and damping (90%) are observed. The experimental data is fitted with mathematical expressions with good correlation..;A new way to obtain the forces exerted by the engine on his mounts was proposed by M.Lavoie (2009). The engine mounts are replaced by force sensors and the attachment points of the engine are instrumented with accelerometers. The measured forces are corrected by the computed inertial pseudo-forces. This method was improved in the present thesis to have a better handle the comutation of moments. The engine motions are determined using a least squares formulation and using the inertial properties of the engine.;An analytical modeling of the engine is carried out using the dynamics of a rigid body supported by flexible elements. The non-linear rubber mounts properties are included in model using the experimental curves for stiffness and damping. The frequency domain model can handle time domain observed phenomenons such as combinatory effects of multiples input forces and jump effect at resonance. It is shown that the non linear model evaluates the natural frequencies and the coupling between modes with better precision than a linear model.;Some optimisations are carried out using the non linear model. The hypothesis that the non linear terms are proportional to the linear terms is used. This hypothesis was observed experimentaly. Three optimisation criteria are used : modes decoupling, limitation of the correlation between natural modes and engine forces and limitation of the correlation with frequency avoidance. The results and the performance level are close to the solution that was implemented using the proposed optimisation method. Some more validations for different cases are needed to fully qualify the process.;A sensitivity study on the possible solutions is conducted. The algorithm tends to converge to robust areas which are not necessarily the optimal solutions. In reality, it is not posssible to obtain precise stiffness values with the manufacturing process. Then it is mandatory to avoid surrounding values that will lead to a much less effective engine mounting design.