Benchmarking of MCNP modelling of HTR cores against experimental data from the astra critical facility

Abstract

The subject of this dissertation is to validate a developed MCNP model of the ASTRA critical facility, through performing comparisons with experimental reactor physics parameters. This validation effort, along with others found in the literature that are focused both on the physics models embedded in MCNP and on the MCNP models of experiments, will help provide the basis for confidence in the use of the code. At PBMR, MCNP, along with other extensively used nuclear engineering computational tools help in the support of the design and eventually the definition of passive safety case for a High Temperature Reactor (HTR). The ASTRA critical facility was chosen as the basic analysis system for this work; with experimental results made available through an Eskom-Kurchatov Institute contract aimed at investigating some PBMR-neutronic characteristics. The availability of the ASTRA experimental set-up information, executed experimental results and some code comparisons presents a very good opportunity for PBMR to validate its own computational tools as per the outcome of the contract collaboration between the two. Some of the experiments performed in support of the investigation of PBMR neutronic characteristics included the study of critical parameters, control rod worths, neutron and power spatial distributions (axial and radial) and reactivity effects. The Monte Carlo n-particle transport code MCNP5 was used to perform all the analyses reported in this work. The findings of this thesis indicate that considering the experimental tasks analysed for the ASTRA critical facility Configuration No. 1 using our MCNP5 consideration (code, modelling approach and used cross section set), there is a relatively good prediction of experimental results (nuclear physics parameters), with control rod reactivity results in particular very well predicted, despite an overestimation in criticality of the modelled experimental configuration. However, there are areas of concern, both experimentally and in our MCNP5 consideration (both for reactivity and reaction rate results). Concerning the experimental uncertainty, the MCNP5 results for the last side reflector block seem to consistently lack agreement with their experimental counterparts (something that is also seen from the Kurchatov Institute’s computational tools used to calculate the same results), leading us to consider a lack of precise experimental information to be behind our models not being representative enough of the experimental set-up. On the model uncertainty, the arrangement of moderator, fuel and absorber spheres in the reactor cavity (particularly the core region) and the neutron flux spectrum and profiles throughout the assembly need to be further investigated in future. This MCNP model validation effort for the ASTRA critical facility reports promising results, albeit not complete, as indicated above, and also a need to study further ASTRA critical facility configurations in order to make a final decision about MCNP5’s suitability in modelling and performing nuclear engineering analysis on HTR cores.