Residual thermal stress analysis in cylindrical steel bars using finite element method and artificial neural networks

In this study, it was proposed that the residual stresses within steel bars after quenching in water from 600 degreesC could be calculated by using the finite element method (FEM) and an artificial neural network (ANN) algorithm. Three modelled cylindrical specimens of AISI 1020 steel were heated and then quenched in water. Using FEM, temperature distribution with time and thermal residual stress values in the samples were calculated after cooling. The analysis was extended to elastic-plastic deformation during the quenching of steel cylinders of various diameters. The calculated temperature and thermal residual stress values were used in training a multi-layer, feed forward, back propagation ANN algorithm. The results obtained via the ANN algorithm method have been compared with the FEM results. Comparison showed good agreement. (C) 2002 Elsevier Science Ltd. All rights reserved