Real time simulation based predictive models using parallel computers for advanced operator aids and control systems

This paper describes the authors' efforts in the application of parallel computer architectures to the real time thermal-hydraulic simulation of nuclear power systems as part of a larger research program to develop predictive expert system models to assist in reactor control and operation. applications of this work include providing advice to the operator on planned power maneuvers, advance warning of potential technical specification violations, and identification of plant upset conditions, including detector failure or malfunction. A mixture drift flux model is used to describe the thermal-hydraulic conditions in the reactor fluid system. Numerical techniques which are amenable to parallel solution are presented along with typical run time estimates on a mini-supercomputer class Alliant FX/4