The modelling of LEU heterogeneous systems as tetrahedral arrays in MONK^®, SCALE and MCNP^® and the impact of heterogeneity on runtime

For LEU systems, up to about 5% enrichment with a fixed mass but unlimited hydrogenous moderator, heterogeneous systems are the more reactive. To ensure models of such systems are conservative a reflected, spherical system of a tetrahedral array of LEU spherical particles in a moderating medium is represented and the particle radius and pitch are optimised. This paper presents how such an array of particles can be modelled within three well known Monte Carlo codes MONK^®, SCALE and MCNP^®. MONK and SCALE have built-in features which make modelling tetrahedral arrays relatively easy. Conversely, MCNP does not and the user is required to construct a unit cell which can be arrayed together. This paper sets out the alternative unit cells that could be utilised and presents details of modelling a hexagonal close packed unit cell containing only 14 particles, the smallest number. An 'MCNP_-pstudy' parameterised input is provided to ensure ease of modelling such systems in the future. For most systems such models will contain several thousands of particles, and this can impact significantly on runtime due to the code checking the surfaces of the particles against neutron tracks. This paper compares the k-eff results of all three codes which show good agreement. The relative increase in runtime for each code for the heterogeneous system over the homogeneous system is also explored. Increases in runtime were by a factor of ∼2.3 for MCNP, ∼2.0 for SCALE and only ∼1.3 for MONK, which employs Woodcock tracking within such repeated structures. (author)