Prediction of tribological properties of structural steels using artificial neural networks

Abstract

The effect of quenching temperature on wear resistance of 40Kh steel after tempering has been investigated. It was found that compared to standard heat treatment, quenching from 1050 °С and high temperature tempering increase its tribological characteristics. The character of fracture of the contacting surfaces was studied. It was shown that in the specimens quenched from 860 °С and tempered, the fracture of the contact surface occurs by the mechanisms of smooth splitting and delamination with plastic deformation. Increasing the quenching temperature to 1050 °С along with high temperature tempering changes the character of the contact surface destruction. The areas with a distinctive microstructure appear on the surface exhibiting substantially higher wear resistance during friction as compared to the surrounding volume. The structuralgeometrical parameters characterizing the roughness and bearing capacity of the contact interaction surface were analyzed. It was found that increasing the quenching temperature to 1050 °С allows to reduce the surface roughness and increase the bearing capacity. Using the methods of optical and transmission electron microscopy, the peculiarities of forming the microstructure of the investigated steel were studied, depending on the temperature conditions of the thermal treatment. It was shown that raising the quenching temperature to 1050 °С increases the austenitic grain size, enhances nonuniformity of carbon distribution, which leads to the formation of large needle-shaped crystals of lath martensite with microtwin boundaries inside. This, in turn, promotes the formation at high tempering of non-uniformly distributed aggregates of coarse carbides at these microtwin boundaries. The aggregates form areas of microstructure with increased resistance to plastic deformation processes. That is, the morphology of the carbide phase is one of the main factors that determine the tribological characteristics of steel, namely roughness, structural-geometrical parameters and bearing capacity of the surface. The expediency of using artificial neural networks for prediction of tribological properties of structural steels was shown. According to the results of modeling the structural-geometrical parameters of the surface and the roughness characteristics, the bearing capacity of the 40Kh steel surface during friction was predicted.