From vector-chain-based to geometrical Tolerance Analyses of Systems in Motion – the Use of Meta- models and their Evaluation

Statistical tolerance analyses are usually performed to ensure the functional capabilities of a product, which underlies deviations. Therefore, several tolerance analysis techniques are available, whereas the vector-chain-based tolerance analysis is well-known and widely used. However, this technique is limited concerning the consideration of geometrical deviations, like the deformation of parts. This paper presents a methodology, which overcomes this limitation by the use of meta-models. These meta-models are used to represent the deviations as well as the effects of other deviations towards those (interactions). Since meta-models are approximations, a certain error appears. Hence, the accuracy of a tolerance analysis depends largely on the meta-model’s prediction quality. Consequently, three different case studies are performed, based on a non-ideal one-way clutch, to answer the question on “the prediction quality the meta-models should have”.