Minimizing the Effect of Geometric Variation in Compliant Assemblies

Sheet metal assemblies are often used as support structures in automobiles, airplanes and appliances. These structures not only provide a frame for other modules to be assembled, but also give the product its aesthetic form. All assembly processes are affected by part and process variation. For this reason, the dimension quality of the assemblies is analyzed in early design phases, to make sure that the product will function as planned and continuously keep the product cost low. Tolerance analysis, as conducted in most industries today, is normally based on the assumption of rigid parts and is thus not always valid for sheet metal assemblies, due to their compliance. The objective of this thesis is variation simulation of sheet metal assemblies. The scientific goal is to analyze, understand and built a mathematical model of how different configurations of input parameter affects variation in the final assembly geometry. To analyze this, there are two fundamental research questions that need to be answered: What parameters influence the final dimension quality of complaint assemblies? and How can the dimension quality be simulated and evaluated in early design phases? The nature of this research work is problem oriented and a problem-oriented scientific method has therefore been used. This method consists of problem identification, observation, modeling, verification, and characteristics & performance steps. The first three steps have been treated in this thesis. The results of this research work so far consist of three appended papers. The first paper identifies important parameters that influence the final geometry and variation of sheet metal assemblies and different criterias for which the assembly designs are evaluated against. Based on this, a suggested early evaluation method is presented. In the second paper, a deeper analysis of how different weld configurations affects the final geometry of the assembly is presented. A simulation model has been built to simulate how different weld gun types and weld sequences affect the final geometry. An important conclusion from the analysis is that it is difficult to find an optimum weld sequence that minimizes its effect on the final geometry, due to the interactions between other input parameters.In the third paper, the simulation model developed in the second paper was extended to include part and positioning variation. The aim of this, was to analyze the interactions between selected input parameters and output measures. Computer experiments and regression analysis were used to analyze the interactions and model building. These models can be used to analyze the effect the input parameters, correlated or uncorrelated, have on the final geometry