Determination of Control Rod Position by Global Optimization Methods

The method is developed to find the implicit position of the control rod using the flux value shown by the neutron detectors. When using the 3D core physics analysis code CPACT, the detector response of a set of predicted rod positions can be calculated. On a double duel problem, functions can be built to describe the difference between the prediction and the target. Under the frame of the inverse problem, combinatorial optimization methods are operated to renew the prediction of rod positions by iteration to a result with convergence. By such a method, the results are guaranteed to achieve the self-consistent with the physical and thermal field while reaching stable convergence when the forward calculation software has been fully verified to simulate the real case. In the meanwhile, the appropriate computational cost is needed and can be adjusted with time tolerance. So, we have compared several global search algorithms (Genetic Algorithm, Particle Swarm Optimization, and Characteristic Statistic Algorithm) with their performance. We can also get the conclusion that hidden features and intermediate calculation results are important to improve the search efficiency. However, the responses of the detectors to some remote rods are not obvious, which are not easy to determine. All the methods were tested on the numerical calculation examples of a pressurized water reactor case.