Karl Merkley, Edgix
Presented June 16, 2000 at the ADCATS 2000 Conference

Sheet metal or plastic assemblies are subject to misalignment between mating parts due to warped or distorted surfaces, deflection under gravity or handling loads, and residual stresses from welding or bonding processes. Statistical tolerance analysis is a popular tool for predicting variation in assemblies, but is limited to assemblies of rigid bodies. While the dimensional variation and warping of parts may be characterized statistically, the compliance of sheet metal or plastic parts is not a random process variable. More sophisticated analytical tools are required for compliant assemblies, which combine statistical tolerance analysis (STA) with finite element structural analysis (FEA) to predict not only tolerance stackup, but also the probable range of assembly forces, deformation and residual assembly stress.

Two types of statistical covariance are required to obtain statistical results from FEA. The first is called “material covariance”, which is the result of stiffness coupling between points in a compliant part. The second is called “geometric covariance”, which is the result of continuity requirements on the mating surface variations. That is, surface distortions at any point on a flexible surface are not independently random, since they must always describe a smooth surface. Methods for computing and combining both sources of covariance have been developed and verified by Monte Carlo simulation.

Bio
BS, MS ME BYU ’87, PhD ME BYU ’98. ADCATS alumnus. Currently working on a several software applications for engineering, manufacturing and the internet.