Experimental determination of the thermal properties of multi-layered surfaces

Abstract

This paper outlines a project which aims to use Certified Reduced Basis and General Empirical Interpolation Methods to conduct rapid, inexpensive, computationally simple thermal property estimation for the purpose of material identification. In this specific case, thermal conductivity and diffusivity were the parameters of interest. Towards this end, an experimental apparatus was constructed which applied a thermal load to various materials and observed their thermal responses. Bugs in the experimental apparatus were compensated for by way of a MATLAB script, until the data produced by individual tests became highly repeatable. Software was developed which simulated these thermal responses for given thermal loads and "true" parameter values. The materials were put through multiple tests (Laser Flash Test, Transient Plane Source) to independently identify possible values for these thermal properties. The "true" values were then chosen from these possible values based on how well they allowed the simulated response to fit the measured response. It was found that implementation of the CRB and GEIM allowed for an accurate estimate of these "true values," and did so without exhaustively carrying out a finite element analysis for every possible combination of parameters, creating an exponential increase in performance.