RESULTS OF THE PRIMO SYSTEM ON THE BASIS OF EVOLUTIONARY ALGORITHMS WITH HEURISTICS FOR NUCLEAR-FUEL MANAGEMENT OPTIMIZATIONS

In close interaction with experts for the PWR loading pattern design a program package for the fuel management automatic optimization is bei ng developed with the final objective of taking into account most prac tical needs. The PRIMO code systems combines the RELOPAT optimization code of the Technical University of Braunschweig and the new SAV95 cor e management system of SIEMENS with the fast running nodal flux calcul ation module PRISM for stationary reactor states. By use of adaptive E volutionary Algorithms in combination with expert knowledge encoded in heuristics a parallel numerical optimization method has been establis hed to calculate improved PWR's fuel loading pattern for user-defined optimization goals. Simple parallel algorithmic structures coupled wit h a low amount of communications between computer processors make it p ossible for workstation clusters to be employed efficiently, offering short turn-around cycles in the design process. Highly promising resul ts have been found by the recalculation of known reload plans for all types of PWRs.