Boriding response of AISI W1 steel and use of artificial neural network for prediction of borided layer properties

In the present study, boriding response of AISI W1 steel and prediction of boride layer properties were investigated by using artificial neural network (ANN). Boronizing heat treatment was carried out in a solid medium consisting of Ekabor-I powders at 850–1050 °C at 50 °C intervals for 1–8 h. The substrate used in this study was AISI W1. The presence of borides FeB and Fe2B formed on the surface of steel substrate was confirmed by optical microscope and X-ray diffraction analysis. The hardness of the boride layer formed on the surface of the steel substrate was over 1500 VHN. Experimental results indicated that there is a nearly parabolic relationship between boride layer and process time for higher temperatures. Optical microscope cross-sectional observation of the borided layer revealed columnar and compact morphology. Moreover, an attempt was made to investigate possibility of predicting the hardness and depth of boride layer variation and establish some empirical relationship between process parameter of boriding and boride layer, and hardness changes using back-propagation learning algorithm in ANN. Modelling results have shown that hardness and depth of boride layer were predicted with high accuracy by ANN.