Combined Measurement and Finite Element Analysis to Map Residual Stresses in Welded Components

Abstract

Through thickness measurement of residual stresses is now undertaken routinely for complex welded components. To predict residual stress distributions finite element (FE) simulations of the welding of the component are also carried out using well established codes, with the simulations sometimes validated via measurements. Measurements are usually undertaken at locations where it is judged that the peak residual stresses occur. Therefore comparisons are often confined to limited locations. But this raises the question whether the simulated residual stresses at other locations are correct. To explore this, the work reported in this paper relies on introducing measured residual stresses into an elastic FE model of the welded component. These stresses are mapped into the model at the measurements locations. Then the FE analysis redistributes the initial stresses, with an iterative process introduced to ensure that the measured stresses are retained during redistribution. Examples are shown where agreement between measured and weld model predictions are good at the measurement locations and both the measured and weld model predictions satisfy global equilibrium. However, they do not agree at other locations. It is argued that additional measurements at other locations are required to validate the FE models of welding processes.