Font Size: 

The hole-drilling method is a widely used technique for measuring near-surface residual stresses. The method is based on drilling a hole into the component, which is a stress concentrator, so local yielding can occur around the drilled hole. Since the method assumes a linear elastic behavior of the measured material, the local yielding affects the measured relaxed deformations and introduces errors into the evaluation process.The authors introduced a procedure capable of correcting the evaluated residual stresses for a wide range of material parameters, hole diameters, strain gauge rosettes, and covering any combination of uniform residual stresses for any orientation of the strain gauge rosette. This procedure uses a neural network that was trained on nearly 8 million states prepared by post-processing the results of finite element simulations.Since all measurements are subject to error and uncertainty, the values of input parameters of correction procedure can slightly differ from their true values. For that reason, a sensitivity analysis of the input parameters was carried out. The basis for the sensitivity analysis was a set of a million randomly generated states that cover the entire range of input parameters. After that, for each testing state, the input parameters were generated in stochastic form and the correction procedure was applied.The analysis results show that even if the input parameters slightly differ from the true value, the correction procedure works well and is capable to correct the residual stresses when plasticity occurs around the drilled hole.

Correction procedure, Sensitivity analysis, Finite element simulations, Hole-drilling method, Residual stress