Object shape error modelling and simulation during early design stage by morphing Gaussian Random Fields

Abstract

3D free-form surfaces with high aesthetical and functional requirements are widely used in automotive and aerospace industry. Geometric and dimensional variations of these free-form surfaces caused by inevitable uncertainties in the manufacturing process often leads to product quality issues. Failing to model the effect of non-ideal parts, i.e., parts with geometric and dimensional errors, during design inhibits the ability to predict such quality issues. Consequently leading to expensive design changes after freezing of design. Therefore, it is vital to model and simulate non-ideal parts that are as similar as possible to true manufactured parts during early design stage and prevent unnecessary design changes during later stages. A major challenge for accurate modelling of non-ideal parts during early design phase is the limited availability of data and the ability to effectively utilise the historical data from similar parts.Overcoming this challenge a novel morphing Gaussian Random Field (mGRF) methodology for shape error modelling of 3D free-form surfaces during the early design stage is presented in this paper. The mGRF methodology works under the constraint of limited data availability and can utilise historical measurement data of similar parts to generate non-ideal parts that exhibit spatial deviation patterns as similar as possible to the true manufactured part and conform to specified form tolerance requirements for the profile of a surface. Additionally, the mGRF methodology provides designers with an intuitive way to explore several 'What if?' scenarios relating to part geometric variations during early design phase. This is achieved by first, modelling the spatial correlation in the deviations of the part from its design nominal using Gaussian processes and then, utilising the modelled spatial correlations to generate non-ideal parts by conditional simulations. The aforementioned capabilities enable to accurately model and predict the effect of part shape variations on product quality during the early design phase. The developed mGRF methodology is demonstrated, compared with state-of-art methodologies, and validated using a sport-utility-vehicle door part.