On modeling of the Ae3 temperature of steel

Abstract

Ae3 temperature of steel is required for many purposes for mechanical processing, thermal treatment and design of steels. However, the models available in the literature are not adequate. An attempt has been made to develop a relation for satisfactory prediction of the Ae3 temper- ature of steel. Ae3 temperature was expressed as a funct-ion of concentration of seven components of steels. The terms upto fourth order in concentration were retained. Total number of terms in the expression was 260. From a careful search of the literature 173 steels for which accurate values of the Ae3 temperature were known were selected. Step wise multiple regression analysis was used to isolate the terms which made significant contribution to the Ae3 temperature. Only nineteen of the 260 terms were found to be significant. These terms predicted the Ae3 temperature with an accuracy which is much more than that offered by the other models. Although other invest-igators have tried to modify the model in various ways, the improvements have not been much. In all these treat-ments the exponents of the concentration parameters have been integral. It is possible to optimize allowing expo- nents as fractions with the help of genetic algorithm. However this technique also does not improve the over all quality of predictions appreciably.