APPLICATIONS OF GENETIC ALGORITHMS TO OPTIMIZATION PROBLEMS IN THE SOLVENT-EXTRACTION PROCESS FOR SPENT NUCLEAR-FUEL

Applications of genetic algorithms (GAs) to optimization problems in t he solvent extraction process for spent nuclear fuel are described. Ge netic algorithms have been considered a promising tool for use in solv ing optimization problems in complicated and nonlinear systems because they require no derivatives of the objective function. In addition, t hey have the ability to treat a set of many possible solutions and con sider multiple objectives simultaneously, so they can calculate many p areto optimal points on the trade-off curve between the competing obje ctives in a single iteration, which leads to small computing time. Gen etic algorithms were applied to two optimization problems. First, proc ess variables in the partitioning process were optimized using a weigh ted objective function. It was observed that the average fitness of a generation increased steadily as the generation proceeded and satisfac tory solutions were obtained in all cases, which means that GAs are an appropriate method to obtain such an optimization. Secondly, GAs were applied to a multiobjective optimization problem in the co-decontamin ation process, and the trade-off curve between the loss of uranium and the solvent flow rate was successfully obtained. For both optimizatio n problems, CPU time with the present method was estimated to be sever al tens of times smaller than with the random search method.