Our study focuses on the development of assessment tools and nondestructive evaluation of porous materials from ultrasonic measurements. These materials are encountered in many industrial applications such as polyurethane foam used for insulation, aluminum foams used in aerospace or cancellous bone for biological applications. Acoustic propagation in these complex heterogeneous materials is governed by the Biot theory [1], involving the propagation of two types of waves: slow and fast wave, whose properties are respectively related to the fluid and solid phases constituting the material. During the propagation, these waves undergo deformations that can be characterized by related propagation models [2], defined by specific frequency-dependent attenuation and dispersion laws. Identification of these waves and of their related propagation parameters then provides a characterization of the material health. This may be a difficult problem in the case of porous materials of low thickness and/or with defects, since the different waves and their echoes may overlap, as shown in the example in Figure 1. Separation of these waveforms should however be possible, by taking into account reliable models describing the propagation of each wave. This paper presents a method for identifying such waves (arrival times and propagation parameters) from signals acquired in transmission or reflection, based on an optimization procedure that minimizes a nonlinear least-squares criterion, which is sufficiently constrained and properly initialized in order to produce robust results. The method is validated with numerical simulations and applied to a laboratory experiment with a porous ceramic plate.

This work is partially supported by the French region “Pays de la Loire”, through the DECIMAP project.