Analysis of uncertainties affecting the Monte Carlo simulation of a neutron multiplicity counter

When applying numerical simulation to physical problems it is important to be able to clearly understand the quality that can be reached and the uncertainty associated with the modelling. In this paper we will deal with the Monte Carlo simulation of neutron coincidence and multiplicity counters. These instruments are intensively used in nuclear safeguards for the measurement and verification of the mass of fissile materials in nuclear installations. Their numerical simulation, either for detector design or for numerical calibration, is now used more and more frequently as support to inspections. Several benchmark exercises have been organised in order to assess the capabilities of Monte Carlo simulation to correctly predict realistic practical cases in neutron coincidence and multiplicity counting. Nevertheless most of them allowed reaching conclusions only on the overall uncertainties associated to the modelling through the comparison between calculated and measured integral data. In this paper we aim to analyse more deeply the sources of these uncertainties, separating the different possible causes and analysing parametrically the individual factors. We will include influencing factors deriving from the physical model of the detector, of the acquisition electronics, of the sample and from the nuclear data. The final total uncertainties are consistent with the results of the integral benchmarks.