Validation and Benchmarking of Westinghouse BWR lattice physics methods

Abstract: A lattice physics code is a vital tool, forming a base of reactor coreanalysis. It enables the neutronic properties of the fuel assembly to becalculated and generates a proper set of data to be used by a 3-D full coresimulator. Due to advancement and complexity of modern Boiling WaterReactor assembly designs, a new deterministic lattice physics codeis being developed at Westinghouse Sweden AB, namely PHOENIX5.Each time a new code is written, its methodology of solving the neutrontransport equation, has to be validated to make sure it providesreliable output. In a wake of preparation for PHOENIX5 release andconsecutive validation efforts, a set of reference Monte Carlo calculationswas prepared, using the code Serpent. A depletion calculation with achosen type of branch cases was conducted. Methods implemented inPHOENIX5 are based on the Current Coupling Collision Probabilitymethod used in older versions of the code HELIOS. Therefore, a comparisonbetween reference Monte Carlo simulations and HELIOS 1.8.1is made, in order to discover problems inherent to the said method ofsolving the neutron transport equation. A special care should be givenduring PHOENIX5 validation, to issues highlighted in this work.Discrepancies in results of Serpent and HELIOS are attributed mostlyto disparities in the basic nuclear data used by the codes, as well as arange of approximations and corrections adopted by the deterministiccode.Serpent and HELIOS showed a good agreement in a typical voidrange (up to 90 % void) and ‘less’ challenging branches (coolant void,fuel temperature and spacer grid branches). More significant discrepanciesappeared for extreme cases with a very high void and control rodpresence (k1 differences as high as 1000 pcm) and rather pronouncedconcentrations of the natural boron dissolved in coolant (absolute differencesroughly at a level of 900 pcm). The issues do not seem to stemsolely from discrepancies in the nuclear data libraries used by Serpentand HELIOS.Moreover, a coolant void bias was consistently found in the resultsof branch calculation at changing coolant void. This confirms the analogousphenomenon found in previous studies of the CCCP based deterministiccodes. It most probably stems from the assumptions used bythe method while tackling the neutron transport equation, such as theflat source approximation, the isotropic scattering assumption and thetransport correction. An alternative transport correction approximationis proposed to alleviate this issue.