Thursday, September 29th 2011, 14:00-15:00 in Celestijnenlaan 200A, 200A 05.001, 3001 Leuven-Heverlee
Development of a Core Management Tool for MYRRHA
by David Jaluvka, SCK, MOL (Belgian Nuclear Research Centre); PhD advisors Stefan Vandewalle (dept CS), William D'Haeseleer (dept Mech), Gert Van den Eynde (SCK)
Currently, the advanced nuclear irradiation facility MYRRHA is under development at SCK•CEN in Mol. MYRRHA is intended as a multi-purpose machine for material and fuel irradiations, radioisotopes production, silicon doping, transmutation research, education and training, etc. A special effort is being made within the project to develop an in-core fuel management tool for MYRRHA, addressing the problem of single-cycle and multiple-cycle in-core fuel management optimization.
In this presentation an introduction to the nuclear reactor in-core fuel management is given as well as an overview of the optimization methods being currently used for tackling related optimization problems. Also, a special character of the MYRRHA reactor concept is described followed by the requirements it imposes on the aimed core management tool. Finally, the current version of the core optimization tool is presented together with some preliminary results on a MYRRHA fuel management optimization task.
During the operation of a nuclear reactor the fuel is periodically reshuffled in the core and partially replaced by a fresh fuel. The aim of the in-core fuel management is to find a fuel loading pattern for a single fuel cycle or for a number of successive fuel cycles so that this pattern meets given objective(s), for instance, required irradiation conditions. At the same time all safety margins and other constraints (power peaking factor, fuel stock, etc) must be satisfied. A typical optimization problem to be solved in this context is a non-linear, non-convex, multi-modal, NP-hard, highly dimensional, combinatorial, constrained optimization problem. In the past, both deterministic and stochastic optimization methods, such as mixed-integer non-linear programming, simulated annealing, genetic algorithm, or particle swarm optimization, have been applied to the problem. During the work the decision was taken to implement the meta-heuristic (stochastic) methods in the core management tool.
Taking into account the special character of MYRRHA, various requirements were imposed to the core management tool; the tool shall allow a user to specify different optimization objectives and constraints, the computational effort required should be reasonable, the tool should be applicable also for general material testing reactors, etc. Based on that, a software application has been written in the Python programming language that implements a genetic algorithm in place of an optimization algorithm and the DIF3D reactor analysis code in place of a candidate solution evaluator. The application was then successfully applied to a simple single-cycle optimization problem, proving its ability to solve the in-core fuel management optimization problems.