The pebble bed gas-cooled reactor is one of the most promising concepts among the Generation III+ and Generation IV reactors. Currently, the pebble bed modular reactor (PBMR) design, both U and Pu and minor actinide fueled, is being developed. Modeling the arrangement of coated particles (CPs) inside a spherical region like a pebble seems to be an important issue in the frame of calculations. To use the (relatively) old Monte Carlo codes without any correction, some approximations are often introduced. Recent Monte Carlo codes like MCNP5 and some new original subroutines that we have developed allow the possibility of obtaining more detailed and more physically correct geometrical descriptions of this kind of system. Some studies on modeling pebbles and pebble bed cores have already been carried out by other researchers, but these works are substantially limited to AVR-type UO2-fueled pebbles. However, the impact of approximated models on fuel mass, reactivity, and reactor life prediction has not yet been investigated for new PBMR-type pebbles.

At the same time, an assessment of introducing a stochastic CP arrangement is not so widespread. Analyzing two PBMR pebbles, one Pu- and the other U-fueled, this paper focuses on quantifying errors due to the different approximations generally used to describe the CP lattice inside a high-temperature reactor pebble bed core, as far as mass of fuel, reactivity, and burnup simulation are concerned. This aim was reached also through a new feature implemented in the MCNP5 code, i.e., capability to treat (pseudo) stochastic geometries. Later, we compared the initial mass of fuel, keff, and isotopic evolution versus burnup of some approximated pebble models with the reference model, built by means of this new MCNP5 feature.