Abstract: Dynamic simulation based on virtual prototype technology is an important issue in functional design of production development and innovation. However, dynamic simulation of a multi-contact mechanical system may be unavailable due to its inefficient and instable computational model. Generally, the higher dimension of the Jacobian matrix is, the bigger condition number will be, which makes the solving more inefficient. Contact constraints are always pre-defined in modelling, but some of them don’t work at current iteration step of solving. The Jacobian matrix of the solving model, which includes the constraint definitions, maintains its dimension even if some of the contacts don’t occur. That is to say the complexity of computation is the same at every iteration step. To improve efficiency, the intrinsic properties of the Jacobian matrix for the solving model are analyzed. A contact constraint deactivation or activation method is proposed to reduce the dimension of Jacobian matrix when iterating. Meanwhile, the conditions for robust dimension reduction are given, as well as its algorithm. The experiment reveals that the given method can improve computational efficiency greatly without losing solving precision for dynamics simulation

Key words: Dynamics modelling, Multi-contact constraint, Simulation efficiency