Predicting deformation of compliant assemblies using covariant statistical tolerance analysis

Abstract

In assemblies with compliant parts, dimensional variation causes misalignment between mating parts. To correct the misalignment, the compliant parts are deformed before being fastened. The resultant springback and residual stress can hinder performance. A new method uses statistical tolerance analysis and stochastic finite element analysis to predict the probable range of deformation caused by dimensional variation. To account for surface variation, a hybrid method models the surface covariance in which Legendre polynomials are used to model long wavelengths and the frequency spectrum is used to model shorter wavelengths. The hybrid covariance model accurately predicts the covariance of simulated parts and the covariance calculated from sheet-metal part data. The hybrid covariance method is an important part of an effective system for statistical analysis of variation in compliant assemblies. © 2007 Springer.