Co-simulation of coupled dynamic subsystems: a differential-algebraic approach using singularly perturbed sliding manifolds

An approach is developed for simulation of interacting subsystems described as differential-algebraic equations (DAEs). Multiple simulators of individual subsystems are simultaneously run in order to simulate coupled behavior of interacting subsystems. The dynamic interactions among subsystems are treated as boundary conditions described by algebraic constraints. This leads to a formulation consisting of differential-algebraic equations. An efficient method for solving nonlinear high-index DAEs using singular perturbation theory and sliding mode control is applied. It guarantees computational accuracy of the algebraic constraints. The co-simulation method is implemented on a network-computing environment using a hybrid symbolic and numerical algorithm. The algebraic constraints and the high-index derivatives needed for DAE computation are automatically derived and reduced to C code using Maple V. The interacting subsystem simulators are then connected by the co-simulation coordinator that drives state variables of each subsystem not to deviate from the algebraic constraints. Numerical examples are used to demonstrate the approach.