Towards improvement of geometrical quality for manual assembly parts

Licentiate thesis, 2017

Geometrical variation affects all mass-produced products. This variation will lead to deviations from the nominal design of the product both in terms of aesthetical and functional properties.  Geometrical variation originates either from the manufacturing of the parts or from the assembly process. In order to minimize the effect of variation robust design principles are often used. In early product development the majority of the properties in the system solutions are fixed and to change these later in the product development will be costly. In order to verify the system solution (locating scheme and tolerances), different simulation techniques are used to predict the behavior of the product. This is done using virtual tools, for example Computer Aided Tolerancing (CAT). In order to gain confidence for such tools it is very important that the simulation results are accurate and that they capture all factors that influence the product. In this thesis the focus has been on geometry assurance and CAT simulations for products that are manually assembled. Although many things can be automated, in the automotive industry most of the final assembly is performed by humans and nothing suggests that this will change. Since humans are quite different from robots’ other factors need to be taken into consideration when designing products that are to be manually assembled. The research presented in this thesis reports current issues and problems when performing geometry assurance, robust design and CAT simulations during product development of manual assembly products. In the thesis it is shown that the level of manual assembly complexity affects costs of poor quality, failure rate and geometrical quality. A simulation tool, is developed that simulates the robustness of an assembly both with consideration to sensitivity to variation and level of manual assembly complexity. The tool is implemented in a CAT system, RD&T. Finally, a number of existing research gaps are identified for further research.