Generation of thermal stress and strain during quenching of low-alloy steel plates

Abstract

A mathematical model has been used to calculate the internal stresses and strains generated during the quenching of infinite plates of a steel of high hardenability in water, oil, or a polymer solution in water. As a preliminary it was necessary to determine experimentally some of the required physical and mechanical property data: they included the mechanical properties of austenite at temperatures between Ms and 800°C, the relationship between specimen length and temperature during cooling, and the effect of surface temperature on the surface heat-transfer coefficient. The results of the calculation indicate that all three quenchants produce plastic flow, which is confined to temperatures above M _s This flow gives rise to a complex residual-stress distribution which, in the case of the water or polymer quenchant, involves a maximum tensile stress just below the surface and a compressive stress towards the centre: the oil quench produces the reverse stress distribution. The predicted residual strains are always positive and are not markedly affected by position within the plate. The variations in the residual stress and strain associated with the experimentally determined fluctuations in the values of the surface heat-transfer coefficient are much smaller than those due to the corresponding fluctuations in the yield stress. A comparison has also been made between the residual stress and strain distributions predicted by the mathematical model and the corresponding results obtained by experiment. The degree of agreement was dependent upon the type of quenchant employed; in the case of water agreement was good, but as the severity of the quench was reduced the discrepancy between the two sets of results became progressively greater. This effect may be due to the influence of time-dependent relationships between thermal stress and strain, which were not considered in the mathematical model, but which may have been a significant factor when the rate of cooling was low. The comparison of these experimental results with those obtained by calculation necessitated consideration of the influence of the plate edges, which affected the experimental data only. Measurements of the variation in the plate thickness gave an indication of the extent of the influence of the edge, and allowed the determination ofa factor that was used to convert the experimental data to the levels appropriate to an infinite-plate solution. The magnitude of this factor was significantly greater than that indicated by the application of Saint-Venant's principle.