Parallel methodologies for large-scale simulation

System simulation involves the construction of a computational analogue of a physical system-a process which has traditionally been associated with intemperate demands on processing time. This is more true today than ever before. Complex large-scale plant simulators and process mimics are almost commonplace in some high-technology industries. In such settings, simulation may be used in both the design and validation of novel control strategies, as well as forming part of an operator training programme, or in cause-consequence diagnostics. The problem arises that as physical systems evolve in complexity, the dimensionality of the underlying dynamical equations can rise dramatically, creating an immense computational burden at simulation time. Often this may result in the violation of any real-time constraints, and the effects may become so acute as to render computer times that are excessive to the point of impracticality. Consequently, the authors are concerned with the development of parallel methodologies for such problems. Attention is focused on a necessarily restricted set of large-scale systems (LSS): those that are adequately described by high-dimensional sets of ordinary differential equations (ODE´s) in state-space form. It should be noted however, that many of the general concepts relating to parallelism exploitation, extend to a much wider class of problems