The modular simulation idea is to represent the components of a physical system by subroutines whose interaction mimics the physical interaction. This approach facilitates validation and also the introduction of parallel processing. In this paper a modular scheme is introduced in which the component subroutines iterate to an approximate steady state in between each time step. The iteration is supposed to account for interactions between components which are fast relative to the time step. Although the current implementation of this scheme is sequential, it is intended for parallel processing. The notion of convergence is discussed for the iterative parallel process which occurs in between time steps. A convergence criterion is found which is akin to diagonal dominance for linear systems. Parallel computation of potential functions is also discussed.
