A new method for the characterization of temperature dependent thermo-physical properties

Abstract

The proposed method is based on the laser flash technique. Radially distributed thermograms are calculated via a finite element model and used in an inverse method by optimizing either specific heat or thermal conductivity of a material. These properties are evaluated as a function of radius and respective temperature. Two approximations are introduced inferring the dependence of each property as a function of radius – a polynomial (PNOM) approximation and an iterative gradient (IG) approximation. The method was tested using synthetic thermograms and both approximations were capable of yielding excellent results. The IG approximation was more universal and less sensitive to initial fitting parameters. The PNOM approximation was less computationally expensive but was prone to artefacts (such as un-physical minima or maxima) and more dependent on initial fitting parameters. Both approximations were successfully used on experimental data from UO2 and isostatically pressed graphite. Thermal conductivity was within 5% of the reference empirical correlation for UO2 and within 7% of the reference curve for graphite.