摘要：In this study, a new method is developed to predict an accurate convective heat transfer coefficient (HTC) during quenching of large size steel blocks, using a combination of 3D Finite Element (FEM) simulations and a progressive artificial neural network (ANN). The HTC profile of the first inputs used for FEM simulations were acquired from the literature to calculate the cooling temperature profiles at specific locations. The training of the ANN was set up between HTCs and their corresponding FEM-calculated temperature. Experimental validation was carried out by instrumenting a large size forged steel block during the quench process. The experimental cooling curves were used for validation of the FEM simulation, as well as for the prediction of new HTCs by simulating the ANN. Results show that the proposed method provides progressively more accurate predictions than the existing ones reported in the literature. A mean absolute percentage error ( MAPE ) of 1.47% was found between experimental and calculated cooling curves for the predicted HTC, further demonstrating a better prediction ability of the proposed method.
关键词：Heat transfer coefficient of steel quenching large forged blocks ; FEM simulation ; Artificial Neural Network ; TTT ; Phase transformation
