A three dimensional, simultaneous and direct estimation technique of orthotropic thermal diffusivity tensor of composite medium embedded in a two-layers material is presented in this paper.
The identification method is based on an inverse heat conduction problem that consists in fitting the outputs of an analytical model inspired from the thermal quadrupoles method, with the evolution of the temperature on the front or rear face of the sample. This evolution is the consequence of a short and localized thermal excitation applied on one of the sample faces. The model developed, as well as the choice of the observables, are consistent with the well-known experimental flash method. Excitation and measurement faces combination leading to four possible experimental protocol, the main objective of this paper is to prioritize these cases depending on the liner type.
The estimation strategy, investigated in this work, is applied on an orthotropic carbon fiber reinforced polymer (CFRP) composite combined with an isotropic layer, to form a two-layers material commonly used in many industries.
Estimation is performed using a Particles Swarm Optimization algorithm minimizing the least-squares criterion between the 3D analytical model outputs and synthetic noisy data. The stochastic optimization algorithm is found to be appropriate to investigate for such non-linear and multidimensional problem as it succeeds in retrieving the original data parameters even for high addition of noise level.
A sensitivity analysis is conducted in order to test and compare the feasibility of the estimation for each configuration. Then, the estimation results are compared in terms of accuracy and robustness, with respect to actual experimental constraints.