A gluing algorithm for network-distributed multibody dynamics simulation

The improved performance and capacity of networks has made the combined pro cessing power of workstation clusters a potentially promising avenue for so lving computationally intensive problems across such distributed environmen ts. Moreover, networks provide an ideal platform to employ heterogeneous ha rdware and software to solve multibody dynamics problems. One fundamental d ifficulty with distributed simulation is the requirement to couple and sync hronize the distributed simulations. This paper focuses on the algorithms n ecessary to couple together separately developed multibody dynamics modules so that they can perform integrated system simulation. To identify a usefu l coupling strategy, candidate numerical algorithms in the literature are r eviewed briefly - namely, stiff time integration, local parameterization, w aveform relaxation, stabilized constraint and perturbation. An unobtrusive algorithm that may well serve this `gluing' role is presented. Results from numerical experiments are presented and the performance of the gluing algo rithm is investigated.