Parallel computers are becoming more available. The natural way to improve computational efficiency of multibody simulations seems to be parallel processing. Within this work we are trying to estimate the efficiency of parallel computations performed using one of the commercial multibody solver. First, the short theoretical outline is presented to give the overview of modeling issues in multibody dynamics. Next, the experimental part is demonstrated. The series of dynamics analyses are carried out. The test mechanisms with variable number of bodies are used to gather the performance results of the solver. The obtained data allow for estimating the number of bodies which are sufficient to gain benefits from parallel computing as well as their level. The parallel processing profits are taken into account in the case of contact forces present in the system. The performance benefits are indicated, when the multilink belt chain is simulated, in which contact forces are included in the model.

This paper compares selected optimization-based methods for the analysis of multibody systems with redundant constraints. The following numerical schemes are examined: direct integration method, Udwadia-Kalaba formulation, two types of least-squares block solution method and Udwadia-Phohomsiri formulation. In order to compare efficiency of the algorithms, a series of simulations is performed on two exemplary McPherson struts. In the first variant, the mechanism has no redundant constraints whereas the other is overconstrained. Three constraint stabilization schemes are also compared in terms of integration errors.