HTHP Home • Issue Contents • Forthcoming Papers

Efficient inverse estimation tool for radiative and conductive properties of insulating foams based on transient hot-wire measurements
Naouel Daouas and Mohamed-Sassi Radhouani

Heat transfer in insulating foams aroused the interest of many researchers due to their importance in the building insulation industry. In these high porosity materials, heat transfer occurs by combined conduction and radiation through the solid and the gas phases. The analysis and the improvement of the thermal performance of such insulators require an accurate knowledge of their thermal properties. This paper deals with an inverse conduction-radiation analysis for estimating conductive and radiative properties of insulating foams. We propose a new identification approach based on thermograms obtained from classical hot-wire measurements, where the optical thickness, the scattering albedo and the thermal conductivity are estimated simultaneously from a unique experiment using the Levenberg-Marquardt algorithm. A preliminary experimental design analysis is carried out based on sensitivity and correlation analyses. This allows us to evaluate beforehand the accuracy of the identification and the linear dependence of the parameters. The proposed approach is tested on numerically simulated noisy thermograms of real low-density EPS foams. The high correlation between the thermal conductivity and the scattering albedo is overcome using, as initial guess of the thermal conductivity, the value fitted with the classical hot-wire technique.

This identification procedure is an efficient tool for estimating both conductive and radiative properties from simple temperature measurements which also provide an initial guess of the thermal conductivity. This is also an interesting way to evaluate the effect of radiation on the accuracy of the classical hot-wire technique.

Keywords: Conduction-radiation problem, Numerical analysis, Parameter estimation, Insulating foams, Transient hot-wire method, Experiment design, Sensitivity analysis, Correlation analysis.

full text (IP)