A combined genetic algorithm and finite element method for the determination of a practical elasto-electric set for 1-3 piezocomposite phases.

1-3 piezocomposites are widely used in ultrasonic transducers, particularly for imaging applications. The fabrication process is often based on the dice and fill method, leading to a periodic structure. This process can modify the initial properties of the two phases due to the machining of the piezoelectric bulk ceramic and setting of the polymer. A method is proposed to directly determine a practical set for 1-3 piezocomposite properties and all the elastic, dielectric and piezoelectric parameters of the two piezoelectric (11 constants) and inert phases (3 constants). This method is based on a fitting process of the electrical impedance as a function of frequency (one thickness and two lateral modes). For this purpose, a genetic algorithm coupled with a finite element method (GA/FEM) was used in an iterative process to deduce all these parameters. This method was first performed on a numerical phantom (Pz21/epoxy resin). Comparisons showed that the GA/FEM obtained a good set of the 14 parameters, and the accuracy of several parameters was discussed. Finally, the GA/FEM algorithm was applied to a fabricated 1-3 piezocomposite (dice and fill method). The results showed that the fabrication process introduced several changes in the properties of the two phases (in particular, the dielectric constants of the ceramic and one elastic constant of the polymer) compared to the initial commercial data, while keeping the identical thickness coupling factor at 64%.