PWR [pressurized water reactor] optimal reload configuration with an intelligent workstation

In a previous paper, the implementation of a pressurized water reactor (PWR) refueling expert system that combined object-oriented programming in Smalltalk and a FORTRAN power calculation to evaluate loading patterns was discussed. The expert system applies heuristics and constraints that lead the search toward an optimal configuration. Its rate of improvement depends on the expertise coded for a search and the loading pattern from where the search begins. Due to its complexity, however, the solution normally cannot be served by a rule-based expert system alone. A knowledge base may take years of development before final acceptance. Also, the human pattern-matching capability to view a two-dimensional power profile, recognize an imbalance, and select an appropriate response has not yet been surpassed by a rule-based system. The user should be given the ability to take control of the search if he believes the solution needs a new direction and should be able to configure a loading pattern and resume the search. This paper introduces the workstation features of Shuffle important to aid the user to manipulate the configuration and retain a record of the solution