Research On The Refueling Algorithm Of The CAMDU Reactor

CANDU reactor uses the refueling methods without stopping the reactor power. The refueling methods not only needs to maintain the reaction efficiency and improves the economical benefit, but also considers the influence of the control and protection of the reactor. So, the problem of refueling of the methods becomes the most challenge tasks of the fuel exchange engineer. Refueling is not only the main method to compensate the use of the fuel, but also to maintain the reactor power to exist in an idea condition. The refueling method's quality not only affect the economical benefit of the reactor's function, but also have a deep relationship with the safety of the reactor. So, the refueling method is the most important problems that the engineer needs to solve. Combining with the features of the CANDU reactor, the optimized object of the reactor is changed from time to time. At the first stage, the remaining of the reactor is adequate, so the main task is to optimize the burn up of the fuel. At the last stage, because of the aging, the running remain is decreased. So in this time the main task is to maintain the distribution of the power in order to maintain the running of the reactor. The changing features of the object makes the refueling becomes more and more complex. This thesis based on my knowledge and the experience of the foreign CANDU reactor.This thesis proposed an algorithm which is called the refueling of the pipes. It is based on the combination of expert system, genetic algorithm (GA) and thesearch based on the neighbors of the pipe. The process of selection of the pipes consists of two phases. Phase one, firstly, evaluate every pipe by the evaluation rule, and then choose 20-40 outstanding pipes from the results. Phase two, choose about 7 candidates and provide their refueling sequence. The arithmetic in this thesis is an integration of the genetic algorithm and searching on the neighbors, and adding searching on the neighbors to control the quantity of the gene, after that, the fitness is concluded by the result of the evaluation function. The sequence of refueling(consist of the sequence of 7 pipes) can be generated by the optimized of the genetic algorithm. Because of the aging, in order to get the better sequence, the candidate of the pipes must be increased. Then the feasible solution is totally increased, the traditional algorithm to solve this problem needs more time. As to the time to the CANDU reactor, it is not benefit. But using my algorithm can solve this complex problem. Generally speaking, to select 7 pipes from 40 pipes, using genetic algorithm needs integrating 30 times in order to get convergence. Through experiment the algorithm based on genetic algorithm and searching can get the better solution. Checked by the program FMPHWR demonstrated that the selected sequence can fit the required of the reactor, and can select the sequence of pipes quickly.