Evaluation of the Systematic Error in the Fuel Mass in the Modeling of Fuel Pebbles with MCNP Vladimir Raduloviˇ c 1 , Andrej Trkov 1 , Igor Lengar 1 1 Joˇ zef Stefan Institute Jamova cesta 39, SI-1000 Ljubljana, Slovenia Vladimir.Radulovic@ijs.si, Andrej.Trkov@ijs.si, Igor.Lengar@ijs.si ABSTRACT The large number of randomly distributed coated fuel particles (TRISO particles) inside the fuel pebbles in a high temperature reactor (HTR) makes their accurate modelling in MCNP very complicated and time-consuming. A simpler method is to create an infinite cubic lattice of fuel particles and filling the pebble cells with said lattice. This technique introduces a systematic error in the fuel mass, due to the truncation of the fuel particles on the surface which defines the fuel region boundary in a pebble. This error, depending on the lattice pitch, can reach a few percent, and may thus introduce inaccuracies in the calculation of k eff . The exact dependence of the fuel mass on the fuel kernel lattice pitch has been established for three lattice types. With these relationships the lattice pitch may be adjusted and the error in the fuel mass can be corrected. 1 INTRODUCTION The error in the fuel mass due to the truncation of the fuel kernels on the fuel region bound- ary in a pebble has been studied in [1]. A solution to avoid the error has, however, not been presented yet. A slight change in the fuel mass, depending on the conditions, can lead to an increase or a decrease in the value of k eff of a reactor [2]; under certain conditions this effect can be quite considerable. In some criticality benchmark evaluations, [3] the fuel kernels are modelled accurately, i.e. there are no truncated kernels and their spacing inside the fuel region is random. This type of model is expected to give the most accurate results, on account of the fuel mass in the model being known exactly. In most cases, the calculation of k eff of a pebble bed reactor is performed with models in which the fuel kernels are placed in a regular lattice and are truncated at the fuel region boundary [4]. In such models the fuel mass is not known exactly. So far this problem has not been given enough or any attention at all. Even some benchmarks have been published, in which the MCNP code was used, and the problem was not taken into account. The fuel can also be smeared with all the other materials contained in the fuel region [5], thus preserving the fuel mass; the results published in [6] suggest, however, that this technique is best to be avoided, due to the inaccurate treatment of self-shielding effects. 211.1