COUPLED EMERGENCY DIESEL GENERATORS - COMPONENT COOLANT WATER SYSTEM MAINTENANCE SCHEDULING OPTIMIZATION BY GENETIC ALGORITHMS

摘要

Previous work has demonstrated that genetic algorithm (GA) together with probabilistic safety analysis (PSA) could successfully be applied to maintenance scheduling optimization problems. The main objective of this work is to apply the recently developed GA-PSA based maintenance scheduling optimization method, which allows non-constant time intervals between maintenance interventions, to a more complex and practical problem. In order to accomplish that, the coupled Emergency Diesel Generators - Component Coolant Water system of Angra-I Brazilian nuclear power plant has been chosen for application. Such system has fundamental importance for the plant safety, moreover its structural and aperational complexity makes the maintenance policy optimization a difficult task. The consideration of both Emergency Diesel Generators and Component Coolant Water System is due to their interdependence in case of a blackout, where the diesel generators must start to feed the coolant water system, but it must also be cooled by this system. This situation creates a severe scenario from the reliability point of view. Obtained results ratify the feasibility of the proposed methodology that could successfully be applied to a practical and more complex problem, increasing the average reliability of the system when compared to those obtained by traditional standard policies. Furthermore, the scheduling configurations suggested by the GA reveal the expert knowledge acquired during the optimization process.