|
ABSTRACTAs the technology of material science progresses day by day,the data of material properties are very important to theapplication of new materials. Material properties may not beeasy to obtain experimentally without the help of propermathematical models. It is feasible to use inverse method inpredicting the thermal properties of material. In this paper, the conjugate gradient method is used to solve inverse heattransfer problems to compute the thermal properties. Conjugate gradient method includes three basic problems:direct problem, sensitivity problem and adjoint problem.In this method, a target function J is defined as the differencebetween measured temperature and computed temperature,which can be obtained from solving the direct problem.The conjugate gradient method predicts the thermal propertyby minimizing the target function. In each iteration, thesensitivity and adjoint problems offer a corrected amount tomodify the property.With assumed thermal properties, three numerical experimentsare used to test the conjugate gradient method. From thecomputational results, it is found that the method can predictthermal conductivity, heat capacity and latent heat correctly.However, in the model of predicting latent heat, there arethree assumptions: (1) Dx must be small, (2) the casting materialis pure metal, (3) the pouring temperature is the meltingtemperature. Finally, the measured temperatures of metal in a(green-sand) casting process is used to predict the thermalconductivity. Compared with the data reported in the literature,the average error of the computed values is 5.6%. From theseresults, it is shown that the conjugate gradient method can beused to predict thermal properties simply and accurately.
|