On the computation of heat-transfer coefficients from imperfect temperature measurements

The main difficulty in obtaining accurate heat-transfer coefficients from measured temperatures is that the experimental data are unlikely to be perfect. In this paper, an approximate theoretical model is developed to estimate the effects of errors in temperature measurement on the computation of heat transfer coefficients from the numerical solution of Fourier's equations. Three types of data imperfection are considered: noise, bias, and sparseness. The physical model chosen for analysis is a plane disk with axisymmetric boundary conditions. Two case studies are included: (1) the 'quenched disk', where the disk is initially at a uniform temperature and is subsequently cooled by air with a constant heat-transfer coefficient; and (2) the 'free disk', where the disk is cooled by air with a radially varying heat-transfer coefficient. A theoretical estimate for the effects of noise and bias on computer heat-transfer coefficients is presented. 'Experimental results', which are generated numerically by Monte Carlo methods, are compared with theoretical values for the quenched-disk and the free-disk cases.