AUTOMATIC LOADING PATTERN OPTIMIZATION TOOL FOR LOVIISA VVER-440 REACTORS
22nd Symposium of AER on VVER Reactor Physics and Reactor Safety (2012, Průhonice, Czech Republic)
CORE DESIGN, OPERATION AND FUEL MANAGEMENT
Abstract
An automatic loading pattern optimization tool called ALPOT was developed for Loviisa
VVER-440 reactors as a master’s thesis work. This paper discusses the methodology and the
performance of this tool.
The ALPOT code utilizes three different optimization methods. The first method is the
imitation of the equilibrium pattern that is the optimized pattern in case the cycle length and
the operation conditions are constant and the same shuffling pattern is repeated from cycle to
cycle. The imitation utilizes heuristic rules but is also stochastic in nature. The function of the
imitation algorithm is to provide initial patterns quickly for the next optimization phase,
which is performed either with the stochastic guided binary search algorithm or the
deterministic burnup kernel method depending on the choice of the user. The former is a
modified version of the standard binary search. The search is guided so that the swapping of
fuel assemblies always starts at or near the most restrictive assembly. The kernel method
creates burnup kernel functions to estimate burnup variations that are required to achieve
desired changes in the power distribution of the reactor.
The neutronic analysis is always performed with the HEXBU-ELSI code package, which is
the standard 3D reactor core simulator at Fortum. Although, the tool was especially designed
for Loviisa reactors the basic ideas of the search algorithms could be utilized in any VVERreactors.
The performance of ALPOT was tested and validated extensively already in the development
phase. Furthermore, the tool has been used to search loading patterns for upcoming cycles of
Loviisa plants with good results. Depending on the strictness of the safety margins, finding a
feasible loading pattern takes from few minutes to couple of hours. The quality of the found
patterns is equal or better compared to the manually designed loading patterns.