A novel approach in optimization problem for research reactors fuel plate using a synergy between cellular automata and quasi-simulated annealing methods

• An innovative optimization technique for multi-objective optimization is presented.
• The technique utilizes combination of CA and quasi-simulated annealing.
• Mass and deformation of fuel plate are considered as objective functions.
• Computational burden is significantly reduced compared to classic tools.

This paper presents a new and innovative optimization technique utilizing combination of cellular automata (CA) and quasi-simulated annealing (QSA) as solver concerning conceptual design optimization which is indeed a multi-objective optimization problem. Integrating CA and QSA into a unified optimizer tool has a great potential for solving multi-objective optimization problems. Simulating neighborhood effects while taking local information into account from CA and accepting transitions based on decreasing of objective function and Boltzmann distribution from QSA as transition rule make this tool effective in multi-objective optimization. Optimization of fuel plate safety design while taking into account major goals of conceptual design such as improving reliability and life-time – which are the most significant elements during shutdown – is a major multi-objective optimization problem. Due to hugeness of search space in fuel plate optimization problem, finding optimum solution in classical methods requires a huge amount of calculation and CPU time. The CA models, utilizing local information, require considerably less computation. In this study, minimizing both mass and deformation of fuel plate of a multipurpose research reactor (MPRR) are considered as objective functions. Results, speed, and qualification of proposed method are comparable with those of genetic algorithm and neural network methods applied to this problem before.