DEVELOPMENT AND ANALYSIS OF COORDINATION ALGORITHMS FOR INTERACTING DYNAMIC SYSTEMS

Modular simulation is believed to be more efficient and flexible than integrated simulation. In modular dynamic simulation, each subsystem is integrated independently with independent error control. Explicit and implicit algorithms are used to integrate non-stiff and stiff subsystems separately. The principle obstacles to running two or more dynamic simulators simultaneously are communication among the simulators and synchronization of the simulators. The synchronization of the simulators in parallel can be maintained by setting a time horizon over which each subsystem is simulated independently. The communication among the simulators is the major subject studied in the development of coordination algorithms.Mathematical descriptions of the communication problem are presented in this study. Three possible algorithms to solve this communication problem are discussed and compared: direct substitution, extrapolation, and linearization. Stability analyses show that only the linearization algorithm is A-stable. Numerical tests of example problems indicate that the linearization method is more efficient than the other two methods.