Sources of Conservatism in an EPR Monte Carlo Model

Sammanfattning

During the EPR design phase, radiation protection analyses were performed using a Monte Carlo model which was developed at Framatome and relies on conservative assumptions. Recent radiation measurement feedback from operating EPR reactors allows for the evaluation of the model constructed using Monte Carlo N-Particle (MCNP) radiation transport code and AutomateD VAriaNce reduction Generator (ADVANTG) code for variance reduction calculations. The numerical model transports neutrons from the core, and N16 gamma radiation from the primary circuit, throughout the reactor building up to the areas of interest. The present thesis focuses on the identification and assessment of several conservative factors of this numerical model against experimental data. The influence on model-measurement discrepancy of source term modeling, material modeling, civil engineering elements modeling, and choice of nuclear data library is studied in the form of sensitivity studies. Comparison between the variations of the numerical model is carried out by evaluating the relative deviation between computed dose rates from the EPR Monte Carlo model and measured dose rates. Source term is demonstrated to be a substantial source of conservatism for the part of the reactor building located in close proximity to the core. The modeling choices that influence particle propagation in the reactor building have a significant impact on the model’s deviation from reality. In addition to accurate modeling of radiation protection elements, such as the radiation shielding block in the melting discharge channel, isotopic composition of the materials is also a source of penalties in the EPR numerical model. Increasing the model’s hydrogen concentration in ordinary concrete, an element which thermalizes fast neutrons, decreases numerical dose rates. Neutron propagation is subject to variation in iron-containing materials depending on whether ENDF/B-VII.1 and JEFF-3.1.1 cross section library is used, which impacts some dose rates calculations. This work provides results concerning the impact of different conservatism factors on neutron doses representativeness in the EPR reactor building, but little evidence on gamma doses representativeness due to a lack of experimental gamma data.