Degree Type


Date of Award


Degree Name

Doctor of Philosophy


Nuclear Engineering

First Advisor

R. A. Danofsky

Second Advisor

B. I. Spinrad


Most commercial pressurized water reactors are unstable to xenon oscillations that occur during load-follow operation;Control of the axial xenon oscillations is a knowledge- and experience-intensive activity for reactor operators. To aid reactor operators in the control of axial xenon oscillations, an advisory expert system was developed;A rule-based expert system shell, INSIGHT2+, was used to build the expert system which was interfaced with a microcomputer-based core control model of a pressurized water reactor, graphic engine, and data base;A core control model described by one-group diffusion theory with moderator temperature and xenon feedbacks was used to develop heuristic control rules and to test the system. Full- and part-length control rods, boron concentration, and coolant inlet temperature were considered as control variables of the core control model;This expert system consists of a search space: the set of possible power level and power shape patterns. The search space was made by combining the following core state variables: the sign of relative power and axial offset (AO) error, sign of the rate of change of power level and AO, and magnitude of relative power and AO error. Control inputs corresponding to an existing pattern were determined by a rule learning process employing two performance indexes. One performance index is the square of relative deviation of power from the desired power. The other relates to the number of evaluations performed over total simulation time;Possible patterns and the corresponding control inputs were represented in the knowledge base, using IF-THEN production rules. The knowledge base was used to obtain a suitable control action at a given time step by a forward-chaining inference mechanism. The control variables were adjusted by the control inputs from a rule corresponding to an existing pattern until required power level and AO were achieved;The results obtained from the study show that the control strategy developed by the expert system was capable of adjusting the reactor power to meet the scheduled load demand while maintaining an AO in the desired range.



Digital Repository @ Iowa State University,

Copyright Owner

Sun-Kyo Chung



Proquest ID


File Format


File Size

279 pages