Geometry Extension and Assemblywise Domain Decomposition of nTRACER for Direct Whole-Core Calculation of VVERs

The capability and performance of the hexagonal version of the nTRACER direct whole-core calculation code are enhanced for VVER applications by extending the geometry-handling features and also by implementing assemblywise parallelization of the planar method of characteristics (MOC) calculation with higher-order scattering. The geometry-handling methods for the VVER hexagonal geometry having various special constituents are presented with detailed illustrations. The assemblywise domain decomposition (ADD) scheme is established under the hexagonal coarse-mesh finite difference formulation, which is exploited to update the incoming angular flux needed for the ADD parallelization. The solution accuracy and parallel performance are assessed for various hexagonal core problems, including the VVER benchmarks. It is shown that the hexagonal geometry solutions of nTRACER match with the reference Monte Carlo solutions within about 50 pcm in reactivity and 1% in pin power distribution and that the hexagonal ADD can reduce the computing time of the planar MOC calculation by up to 53% when compared to the anglewise parallelization.