Vibration reduction of a turbojet fan blade with piezoelectric patches connected to a passive or semipassive electrical circuit, commonly called "shunt", is addressed in this study. The purpose of this work is to present a method for maximizing the performance of piezoelectric shunts. To validate the model, 2 experiments on a CFM56-7b fan blade are then done. To improve the damping level, a key issue is the optimization of the whole system, in terms of location and size of the piezoelectric patches and electric circuit components choice. It was shown these two optimizations, mechanical and electrical, can be realized separately. Moreover, it is proved the only parameters to maximize are the modal electromechanical coupling factors, which characterize the energy exchanges between the mechanical structure and the piezoelectric patches for a given mode. Since the optimal value of the electric circuit parameters are known as functions of the coupling factors and the system structural characteristics, they can be evaluated in a second step. Thus, the mechanical optimization consists in maximizing the coupling factors by optimizing the patches positions and dimensions, i.e. finding the best design. To fulfill this requirement and in order to manage a complex geometry, a 3D finite element formulation of the coupled electromechanical problem is derived from the one developed by Julien Ducarne during his Ph.D. thesis. A reduced order model of the discretized problem is then obtained by expanding the mechanical displacement unknowns vector onto the short-circuit eigenmodes to get the modal electromechanical coupling factors. However, when the optimization aims to reduce the vibration level with several patches, the main concern arises from the huge number of possible designs to test. For that reason, a method is proposed to cut back simulations time as well as to cope with the many local minima. This method consists in splitting up the optimization procedure in two steps. In the first one, the influence of patches on the structural eigenmodes is neglected. Therefore, an analytic coupling indicator, based on the eigenmodes of the naked host structure, can be defined and gives rise to a first approximate optimization using a simulated annealing algorithm. Then, the solutions of the first step are used as a starting point for a second optimization, working with the tabu search algorithm and where eigenmodes are computed for each new tested design.
