| Distributed simulation is close to reaching its potential to fulfill the demands of industrial CAE by harnessing nearly unlimited computing power across network environments and by efficiently reusing and integrating already constructed simulation models. A distributed simulation platform, denoted as D-Sim, has been under development in our research group since 2001. The present work focuses on the integration of heterogeneous subsystem models, including multibody dynamics (MBD) and finite element (FEM) subsystem models, and conducting seamlessly integrated simulation for design tasks in a distributed computing environment.;Under the guise of a gluing algorithm, the Partitioned Iteration Method (PIM) was developed, which can be used to integrate distributed deformable bodies while allowing large rigid body motions among the bodies or subsystems. The PIM is based upon a floating frame of reference, in which the global motion of the flexible body can be expressed with linearized elastic deformations by assumption of infinitesimal strains and reference frame as large overall motion. When embedded in D-Sim, it also enables using independent simulation servers, in which each server can run commercially available or research-based MBD and/or FEM codes to minimize the information exchange across the different platforms yet still obtain results within engineering accuracy. Examples are provided which integrate FEM and MBD models and which demonstrate the performance of the PIM. The examples also highlight how to decouple and integrate rigid body motion and elastic deformation using the enhanced gluing algorithm.;A gluing algorithm plays a critical role in integrating the distributed subsystems and components. It is one of the research objectives to apply the gluing algorithm to general simulation models, which may be assembled by diverse connecting methods, including spot welds, bolts, bushings, and other physical connections. The gluing algorithm concept has been extended by creating flexible gluing joints, which can deal with various connections between subsystems, and can account for linear and non-linear flexibility at these connections. This not only improves the accuracy of the simulation to represent the real physical system, but also can improve the convergence of multibody dynamics simulation. |
