Efficient computational analyses of boiling water reactor fuel assemblies.

Abstract

This research is focused on the development of computationally efficient artificial neural network (ANN) models for the neutronic analysis of accident-tolerant fuel (ATF) candidates in BWR fuel assemblies. Two ATF options: U3Si2 fuel with SiC cladding, and Al2O3 and Cr2O3 doped UO2 fuel (ADOPT fuel) with FeCrAl cladding were considered for 8X8, 9X9 and 10X10 BWR reactor assemblies. Some U3Si2 fuels in the assemblies contained burnable poison gadolinium (Gd) and some ADOPT fuels contained gadolinium oxide (Gd2O3) to account for the excess reactivity at the beginning of life (BOL). The proposed research needed an extensive dataset for training the artificial neural networks (ANNs), which was generated using the Monte Carlo neutronic code, OpenMC. Fuel enrichment, burnable poison concentration, boron concentration, pitch and number of burnable poison rod were considered as the input parameters for the ANN models. 3000 sets of data (k∞) were provided for training the ANN models for each fuel type and each assembly geometry. A three-layer ANN architecture was considered in the study where the number of neurons in the hidden layer were varies in the range 5-20. Results revealed that the average predictive errors of the ANN models developed for U3Si2 fuel with SiC/SiC cladding were in the range 104.71 to 580.17 pcm, respectively. For the ANN models developed for ADOPT fuel with FeCrAl cladding, the average predictive errors were in the range 102.91-567.81 pcm, respectively. A genetic algorithm (GA)-based optimization process was then employed to identify the assembly pitch that corresponds to the maximum value of infinite multiplication factor (k∞) for a specific type of BWR assembly geometry and composition. The obtained results were utilized to derive expressions that defines the boundary separating the under and over-moderated regions. Considering the decrease in U-235 enrichment and burnable poison inventory at the end of cycle (EOL) scenario, it was recommended that pitch should be around 1.40cm to 1.46cm, depending on the fuel assembly type, to ensure design safety from neutronic point of view.