UC Davis

Residual stresses induced from manufacturing processes produce distortion in aluminum workpieces. It is desirable to control distortion from manufacturing process to prevent part rejection or rework. Both the near surface residual stress from milling processes and the bulk residual stress from manufacturing processes contribute to the distortion. Predicting the distortions using finite element models can help improve the quality of manufactured components and reduce non-conformity costs. The finite element model uses elastic stress analysis which is a mechanics boundary value problem to form the basis of the prediction. The machining is included by introducing an initial stress state near the machined surface. The distortion model is created based off predetermined workpiece geometry, known as the “feature sample”, and uses residual stress measurements on a line milled stress relieved plate AA7050 T7451 as the input to the finite element model. The near surface residual stress measurements are completed using hole-drilling, and the bulk residual stress measurements are completed using slitting. To increase complexity, a model with a spiral path milling pattern and a model with high bulk residual stress from quenched material are created. Models are validated using three methods, by comparing the bottom face of the observed and predicted samples, by comparing the residual stress profiles that are used as inputs to the model to measured residual stress profiles taken in areas of aggressive machining on the sample, and by comparing the bottom surface form of a removed and isolated section of the workpiece. The validation experiments for the stress relieved and line milled samples show agreement between predicted and observed results. For the bottom face comparison, both the surface form and line plots taken along the bottom face are compared. The observed surface forms and the line plots agree with the shape of the predicted surface forms and line plots. The 3 mm low stress sample has better agreement within the uncertainty for the line plots than the 7 mm sample. The residual stress measurements that were taken on the sample are nearly identical with the grand average stress profiles applied to the model. The isolated pocket floors also agree well between the predicted and observed data, with line plots showing good agreement in both the 3 mm and 7 mm samples. Assuming that there is negligible near surface stress in the remaining surfaces of the sample is supported by measurements taken along the walls and bottom of the sample, which demonstrate near zero stress. The validation experiments for the spiral milled and high bulk stress samples add complexity to the distortion prediction method. The spiral milled observed and predicted bottom face line plots and surface form agree, with better agreement in the 3 mm sample than the 7 mm sample. The residual stress measurements completed in the pocket of the spiral stress feature sample have slightly more compressive stress along σxx and σyy, but the profiles do agree with the shape and magnitude of the grand average stress profiles. The isolated pocket floor surface form and line plots for the spiral sample also agree, with better agreement for the 3 mm pocket floor. The high bulk residual stress observed and predicted surface form and line plots agree, with better agreement in the 7 mm sample than the 3 mm sample. Residual stress measurements taken in the 7 mm sample agree with data taken from representative material. Residual stress measurements taken in the 3 mm sample shallower near surface data than the 7 mm measurements. Finite element models still provide a reasonable estimate of distortion in these samples for different milling and material conditions. Future work could analyze other types of machining and other geometries. It might also be useful to create a 3-dimensional model of the distortion for comparison to the observed results. This 3-dimensional model is out of scope for this work but could be useful in other studies. The interaction between the bulk residual stress and machining induced residual stress in samples with thin walls should also be explored. Additional refinement to this method in future work could help improve the method for application to industry.