A NEW METHOD FOR THE EVALUATION OF THERMAL-CONDUCTIVITY AND THERMAL-DIFFUSIVITY FROM TRANSIENT HOT STRIP MEASUREMENTS

The standard straight-line fit to data of a transient hot strip (THS) experiment to determine the thermal conductivity lambda and thermal diffusivity a suffers from two major drawbacks: First, due to the statistical nature of the estimation procedure, there is no relation between the uncertainty of the measured value on one hand and the transport properties obtained on the other. Second, in order to account for the heat capacity of the strip and outer boundary conditions, two intervals of the plot must be rejected before analyzing it. So far, these intervals are selected arbitrarily. We now treat the THS working equation as a function of the four parameters concerned, lambda, a, U-0 (initial voltage), and t(0) (time delay). Chi-square fittings, following the Levenberg-Marquardt algorithm, are performed separately for several overlapping time intervals of the entire plot to find lambda and a with minimal standard deviation. In the course of subsequent iterations an individual weighting Factor is applied to each point to account for systematic errors. This procedure yields the ''best'' values of lambda and a along with their individual errors, comprising the systematic and the statistical errors. Experimental results on Pyrex glass 7740 were taken to verify the new procedure.