Modeling of realistic pebble bed reactor geometries using the Serpent Monte Carlo code

Highlights: • The explicit stochastic geometry model in Serpent is documented. • A pebble bed criticality benchmark was calculated demonstrating the geometry model. • Stochastic pebble configurations were obtained from discrete element simulations. • Results deviate from experiments but are in line with example calculations. - Abstract: This paper documents the models available in Serpent for high temperature reactor (HTR) calculations. It is supplemented by a calculation example of ASTRA critical pebble bed experiments. In the pebble bed reactor modeling, different methods have been used to model the double heterogeneity problem occurring in pebble bed reactor calculations. A solution was sought to avoid unphysical simplifications in the pebble bed modeling and the stochastic geometry modeling features available in the Monte Carlo code Serpent were applied for exact placement of pebbles and fuel particles. Randomly packed pebble beds were produced in discrete element method (DEM) simulations and fuel particles were positioned randomly inside the pebbles. Pebbles and particles are located using a Cartesian search mesh, which provides necessary computational efficiency. Serpent uses Woodcock delta-tracking which provides efficient neutron tracking in the complicated geometries. This detailed pebble bed modeling approach was tested by calculating the ASTRA criticality benchmark experiment done at the Kurchatov Institute in 2004. The calculation results are in line with the sample calculations provided with the benchmark documentation. The material library selected for the calculations has a major effect on the results. The difference in graphite absorption cross section is considered the cause of this result. The model added in Serpent is very efficient with a calculation time slightly higher than with a regular lattice approximation. It is demonstrated that Serpent can be used for pebble bed reactor calculations with minimal geometric approximations as it allows exact pebble bed modeling with randomly positioned fuel particles and locations of pebbles produced by DEM. Due to the added stochastic geometry features and efficient neutron tracking, it is concluded that Serpent is well suited for the calculations of HTRs