STATISTICAL SIMULATION OF MULTIBODY SYSTEMS AND COMPUTER AIDED DESIGN DRAWINGS TAKING SPECIAL CONSIDERATION OF TOLERANCES

摘要

Today the modeling and simulation of multibody systems, and designs created using CAD systems is focused on precise geometric parameters and precise forces. However, in the real world parameters can only be given with a tolerance around a mean value. These tolerances have an important influence on the operation and life cycle of mechanisms, and may not be neglected. In this paper statistical methods are used to compute the resulting tolerance intervals and distributions of kinematic and dynamic values. This gives the designer a powerful tool to reduce only these tolerance intervals, which have a big influence on other tolerances. Any toleranced geometry or force parameter can be used as the input for a statistical tolerance analysis. In the first section of this paper, the structure of multibody systems and computer aided designs will be analyzed. A comparison between these two modeling approaches leads to a unique model description for tolerance analysis. In the second section the Monte Carlo Method as a statistical method for tolerance analysis is presented. Typical distributions such as the continuous uniform distribution and the normal distribution are investigated. Different approaches for optimizing the computation of the tolerance intervals are shown. An estimation of the necessary number of simulations required to gain a specific precision is given. The resulting general algorithm for statistical simulations for all kinds of mechanisms is then presented. With the Monte Carlo Method it is also possible to evaluate the manufacturing process by comparing the tolerances occurring with the simulation results. As an example the modeling of a Geneva mechanism is described. The methods introduced here are applied to this typical mechanism for tolerance analysis. Finally the simulation results are presented and computation times and quality of the tolerance intervals for different input distributions are compared.