Evaluation Approach for Residual Stress in Drilling of Aluminum Alloy

Abstract

Residual stress is one of the critical evaluation factors as well as machining accuracy and surface finish in cutting. From the perspective of the fastening strength of parts, residual stress in drilling has an influence on the product or part life. Especially in the aircraft manufacturing, the tensile residual stress is crucial as it can cause major accidents. However, few studies have focused on residual stress in drilling so far. Most of them investigated the characteristics of residual stress based on the experimental data. This paper discusses residual stress in drilling in terms of its mechanical effect. The residual stresses in the inner surfaces of holes were measured in the depth direction from the top of the plate. The change in the residual stress has good correlation with the chip flow direction and the load applied to the inner surface, which acts as a counter force to the cutting force near the outermost ends of the lips. The cutting force is determined using an analytical model based on the minimum cutting energy. The correlation between the residual stress and the direction of the load applied to the inner surface depends on the margin width and back taper on the peripheral sides of a drill. The effect of the margin contact on the residual stress is characterized by linear regression analysis. Finally, an analysis-neural network hybrid system was developed to estimate the residual stresses in drilling. In the system, the mechanical effects of the drilling process, which are the magnitude and direction of the load applied to the inner surface, are obtained through analytical simulation. Meanwhile, the margin effects are expressed as the coefficients in the linear functions, which are associated with the margin width and back taper angle by a neural network. Then, another neural network, working as a post process, estimates the residual stresses using the information of mechanical and margin effects. The developed system is validated by presenting a comparison between the estimated and the actual residual stresses.