Compensating dimensional deviations of car body assemblies by smartly repositioning individual body parts in the joining fixtures

The assembly of highly accurate metal car bodies poses a huge challenge to automotive industry involving diverse thermal and mechanical joining processes. In practice, the complexity of the individual steps, as well as their interaction, hardly admit a precise physical modelling and, frequently, uncertainties regarding the actual geometry of tools, varying part quality and the high sensitivity of the assembly outcome to process parameters render expensive simulation results useless. As the production routines in the shop floor require to deal with errors of the assembled body in an ad-hoc manner, clamping tools are designed to be adjustable, allowing for relative repositioning of body parts and the induction of tension before joining. Usually, a costly optimization procedure of alternating quality measurements and suitable adjustments of the clamping tools is employed by highly experienced staff, a procedure which often is error-prone or suboptimal, though. This article presents, first, a novel approach to virtually place the individual CAD parts on the assembly scan by a weighted and constrained best fit algorithm; thus, delivering compensating rigid body transformations for each individual body part. Second, these transformations are used to automatically determine adjustment settings for the degrees of freedom available in the clamping tools that enforce the desired part repositionings. The approach is investigated in an iterative optimization procedure of measurements and tool adjustments where the real assembly process is replaced by FE simulation. Several practically relevant examples demonstrate a significant quality improvement of the resulting overall assemblies.