A Cosimulation T-T Procedure Gluing Subsystems in Multibody Dynamics Simulations

In the case of models with numerous and eventually multidisciplinary components, cosimulation technique can be employed to link independently developed submodels. A T-T gluing algorithm is presented and validated for rigid multi-body systems. The coupling of the different submodels, resulting from exploding the all-in-one model at its mechanical joints, consists in imposing holonomic constraints for each joint separating two submodels. At each simulation step, kinematic quantities are computed and sent from the submodels to the central part of the algorithm. There a global Newton-Raphson iterative procedure permits, by means of a so-called gluing matrix, to upgrade the interface efforts which are then returned to the submodels. Two concrete examples are “cosimulated” with minimal coordinates and by considering the constraints at velocity level. The illustrated results and errors demonstrate the accuracy of the method. The algorithm can be generalized for the other kinematic approaches and for finite element and computational fluid dynamics codes.

Integration of Finite Element and Multibody Dynamics Models Using Gluing Algorithm

This paper provides a method of the coupling multibody dynamics (MBD) and finite element (FE) codes to solve flexible multibody dynamics problems across a distributed simulation platform. The aim is to minimize the information exchange across the different platforms yet still obtain results of engineering accuracy. Many approaches for solving flexible multibody systems consisting of elastic and rigid components have focused on how the equations of motion for the flexible components can be formulated and combined with other component equations. We present a Partitioned Iteration Method (PIM), which can decouple the elastic deformation and rigid body motion of the flexible body by employing a CG-following floating reference frame. In the PIM, the global motion of a flexible body is expressed as a summation of the linearized elastic deformation measured in the CG-following reference frame and the motion of the reference frame and be solved in separated finite element and multibody dynamics solvers. The PIM is combined with the gluing algorithm and is used to solve multiple flexible bodies in a distributed simulation environment. It enables the use of independent simulation servers, where each server can run commercially available or research-based MBD and/or FE solvers. Examples are provided to demonstrate the performance of the new method and to show how to decouple and integrate a general flexible multi-body dynamics system.

A Partitioned Iteration Method Employing CG-Following Reference Frames for Distributed Simulation of Flexible Multibody Systems

A distributed simulation platform, denoted as D-Sim, has been developed previously by our research group, which comprises three essential attributes: a general XML description for models suitable for both leaf and integrated models, a gluing algorithm, which only relies on the interface information to integrate subsystem models, and a logical distributed simulation architecture that can be realized using any connection-oriented distributed technology. The overarching research focus is to integrate heterogeneous subsystem models, e.g., multibody dynamics subsystems models and finite element subsystems models and to conduct seamlessly integrated simulation and design tasks in a distributed computing environment. A Partitioned Iteration Method (PIM) is proposed in this paper, which decouples the rigid body motion from elastic deformation of the simulated system using an iteration scheme. The method employs a CG-following reference frame for each deformable body in the distributed simulation of flexible multibody systems. The resultant simulation system can be used to integrate distributed deformable bodies D-Sim, while allowing large rigid body motions among the bodies or subsystems. It also enables using independent simulation servers; where each server can run commercially available or research-based MBD and/or FEM codes. Examples are provided that demonstrate the performance of the method and also how to decouple and integrate rigid body motion and elastic deformation using the developed gluing algorithm.

A Distributed Mechanical System Simulation Platform Based on a “Gluing Algorithm”

Three key concepts are presented in this paper, which comprise the foundation of a distributed simulation platform for design and virtual prototyping of general mechanical systems that have their subsystems distributed amongst dispersed development units in multilayered supply chains. First, a general and efficient model description for simulation is defined using XML. Each model is described with an XML file and stored in model database. A complete model can then be assembled based on these model descriptions. Simulation of a model is started simply by sending the model description to a simulation server and running it through a web-based graphics user interface. Second, a new gluing algorithm, denoted as the T-T method, is developed, which enables distributed simulations (both the component models and simulation of the components) to be coupled while maintaining the independence of the separate component simulations. Third, a logical distributed simulation architecture is laid out that can be implemented with one of the existing technologies for distributed computing. Interfaces between different network components have been standardized to enable extensibility of the architecture. These concepts have been incorporated into a prototype web-based distributed simulation system that demonstrates the potential of the new techniques for solving real engineering design problems.

Distributed Mechanical System Simulation Based on a General “Gluing Algorithm”

Three key concepts are presented in this paper, which comprise the foundation of a distributed simulation platform for design and virtual prototyping of general mechanical systems that have their subsystems distributed amongst dispersed development units in multi-layered supply chains. First, a new gluing algorithm, denoted as the T-T method, is developed, which enables distributed simulations (both the component model and simulation of the component) to be coupled while maintaining the independence of the separate component simulations. Second, a general and efficient model description for simulation is defined using XML. Each model is described with an XML file and stored in model database. A complete model then can be assembled based on these model descriptions. Simulation of a model is started simply by sending the model description to a simulation server. Third, a logical distributed architecture is laid out that can be implemented with one of the existing technologies for distributed computing. Interfaces between different network components have been standardized to enable extensibility of the architecture. These concepts have been incorporated into a prototype distributed simulation system that demonstrates the potential of the new techniques for solving real engineering design problems.

A Gluing Algorithm for Distributed Simulation of Multibody Systems

A new gluing algorithm is presented that can be used to couple subsystem models for dynamics simulation of mechanical systems. The gluing algorithm developed relies only on information available at the subsystem interfaces. This strategy not only improves the efficiency of the algorithm, but also engenders model security by limiting model access only to the exposed interface information. These features make the algorithm suitable for a real and practical distributed simulation environment.