Application of a Step-Wise Verification and Validation Procedure to the 3D Neutron Kinetics Code DYN3D Within the European NURESIM Project

A generic strategy of core physics codes benchmarking was elaborated within the European NURESIM code platform development. In this paper, the application of this step-wise procedure to benchmarking the 3D neutron kinetics code DYN3D for applications to VVER-type reactors is described. Numerical and experimental benchmark problems were considered for code verification and validation. Examples of these benchmarks including benchmark set-up and results obtained by use of DYN3D in comparison with other codes are given. First, mathematical problems with given cross sections are used for the verification of the mathematical methods applied e.g. in nodal codes against finite difference solutions. Discretization errors were quantified. After minimisation of numerical errors, modelling errors have to be considered. Diffusion approximation and homogenisation errors are due to simplified physical approaches and can be estimated by comparing diffusion solutions with more accurate Monte Carlo or deterministic transport solutions. Methods to reduce these errors are outlined. A series of 2D whole core benchmarks for different core loadings and operational conditions for VVER-1000 reactors was defined for this purpose. Reference transport solutions were calculated by the MARIKO and APOLLO codes based on Method of Characteristics. Homogenised two-group and few-group diffusion parameters were derived from the reference solutions and used as cross section data for the nodal diffusion code DYN3D. The DYN3D solutions were compared to the reference solution. It was shown that the homogenisation error can be significantly reduced by using Assembly Discontinuity Factors (ADF) and Reflector Discontinuity Factors (RDF) which are obtained from the transport solution by applying equivalence theory. A study using the multi-group version of DYN3D has shown that increasing the number of groups in the considered cases has only a small effect in comparison with homogenisation error. After reducing modelling errors by choosing appropriate physical approximations, the code have to be validated against reality. Experimental problems are used for code validation. Experimental data for VVER reactors, which were used for the benchmarking of the DYN3D code within NURESIM, are power distribution measurements at the full-size (VVER-1000) experimental facility V-1000, which have been well documented within the EC project VALCO, and kinetic experiments at the LR-0 zero power reactor in NRI Rˇezˇ. The code DYN3D, being one of the NURESIM platform codes, has proved to be an effective tool for steady-state and kinetics core calculations. The high accuracy of the code is based on the advanced nodal method “HEXNEM2”, multi-group approach, applying discontinuity factors, and intra-nodal flux reconstruction.