Abstract
Abstract
The statistical analysis of geometric tolerances has been carried out, traditionally, by using either the Taylor series method or the Monte Carlo simulation. Although the Taylor series method is fast, it is not capable of analysing many highly non-linear cases of geometric tolerances, especially when several of them are specified for the same feature. Hence, the Monte Carlo simulation remains the most widely used method for the statistical analysis of geometric tolerances. Similarly to other methods, during each step of the Monte Carlo simulation, all features in the assembly are generated including random variations due to the manufacturing processes’ capabilities. If one of the features does not fall within the specified tolerance range, the whole instance of parts (i.e. the whole assembly) is rejected. This simulation is more conservative than a real assembly-inspection process. This paper presents an augmented Monte Carlo simulation in which assemblies are not rejected if one or more parts are rejected while other parts are within specifications. Instead, the in-spec parts are re-grouped with other acceptable parts of other simulations. This emulates the concept of parts interchangeability in real assembly processes. The results obtained by using the proposed approach are compared with those obtained from a standard Monte Carlo simulation tolerance analysis to demonstrate that the latter is unnecessarily conservative.