This section presented the validation of the extended version of DYN3D against the unprotected stage of natural circulation test from the Phenix EOL experiments. The test was simulated with the Serpent-DYN3D codes sequence, i.e. the parametrized cross section library was generated with Serpent and the coupled NK/TH time-de- pendent solution was obtained with DYN3D. The numerical results were compared against the experimental data. The following conclusions can be made from the validation study:
• In general, the DYN3D solutions were in very good agreement with the ex- perimental data indicating the feasibility of using DYN3D in coupled NK/TH transient analyses of SFR cores.
• The deviations from the measurements were only observed when out-of-core thermal expansion effects, such as vessel expansion, start to influence the CR position relatively to the core. In order to account for the correct magnitude and time delay of such relative CR movements, the change in thermal-mechanical conditions of the supporting structures has to be modeled. The development of thermal-mechanical models for the supporting structures is an ongoing research topic at HZDR that will be based on a coupling of DYN3D with a thermal-hydraulic system code capable of sodium flow modeling.
• The study demonstrated that by utilizing the newly implemented fuel rod ther- mal expansion model, the non-uniform thermal expansion effects can be mod- eled in transient simulations. As compared to the layer-uniform fuel rod ex- pansion modeling, the discrepancies between the numerical results and the measurements did not change significantly. This can be attributed to a relatively small variation in sodium heat-up over the course of transient (about 40 K). In more severe cases (e.g. accident scenarios with significantly higher or/and asymmetric sodium heat-up) the selection of one thermal expansion model over the other can lead to much higher deviations in the results.
• This study also demonstrated that the MC code Serpent can be successfully applied to generate few-group XS for transient analyses of SFR cores.
5
Summary and future work
The purpose of the present doctoral research was to extend the capabilities of the Light Water Reactor core simulator DYN3D to perform three-dimensional reactor simula- tions of Sodium cooled Fast Reactor cores. The supplementary methods and models developed in this dissertation make DYN3D feasible to do steady-state and transient calculations on reactor core level. These extensions were verified and validated on both numerical and experimental SFR benchmarks.
The SFR-related DYN3D developments are summarized in the following while the recommendations for further research are provided at the end.
5.1 Thesis summary
By utilizing the Monte Carlo neutron transport code Serpent, a methodology was developed to create homogenized few-group cross sections for more realistic and detailed SFR core configurations. This method is capable of providing group constants to practically any deterministic 3D full core simulator. It is based on lattice-level mod- els, which is favorable in respect to the computation time and memory usage while maintaining an adequate MC statistics. Such is particularly important in branching calculations, when a complete parametrized XS library is prepared for transient simu- lations. This research demonstrated that the Serpent-DYN3D codes sequence can be used for static neutronic as well as time-dependent N/TH analyses of SFR cores.
In order to support the cross section generation method, the Super-homogenization technique was applied for the first-neighbor regions to the fuel assemblies. This method demonstrated a consistent improvement in the nodal diffusion solution
of SFR cores, as compared to full core Monte Carlo results. The highest positive impact was achieved by using the SPH factors on the non-multiplying regions that are surrounded by the fissile core, such as the control and diluant assemblies. The SPH method in combination with the lattice approach of XS generation proved to be an efficient way to improve the nodal diffusion solution of the SFRs.
In order to perform coupled N/TH simulations with DYN3D, the implementation of new thermal expansion models was necessary. In this research, DYN3D was extended with new models to account for time-dependent axial and radial core expansions in the neutronic behavior. The axial expansion model is capable of modeling non- uniform core expansions by using the spatial temperature distribution of the fuel rods. This model allows for an independent treatment of each fuel assembly based on local thermal-hydraulic conditions. The radial diagrid expansion model can account for an uniform radial expansion of the core driven by the average inlet sodium temperature. The verification of these models was carried out in this research based on steady-state analyses of two reference cores, namely the large oxide OECD/NEA benchmark core and the smaller Phenix EOL core.
In the process of verification, and for a better understanding of the reactor behavior during core expansions, a spatial- and reaction-wise reactivity decomposition tech- nique was introduced in this dissertation. The method was especially developed to obtain decomposed thermal expansion feedbacks from the solutions of any nodal diffusion code. This procedure allows in a simple manner to quantify the relation between the change in nodal absorption and leakage rates while locating the regions of significance when the core expands.
Finally, the capability of the extended version of DYN3D to perform steady-state and transient analyses of SFR cores was validated using selected tests from the end-of- life experiments conducted at the Phenix reactor. Firstly, the IAEA benchmark on the control rod withdrawal tests was used to validate the neutronic performance of DYN3D. Then the simulation of the initial stage of the natural convection test was used to validate DYN3D for coupled NK/TH transient calculations. Both tests have also served for the validation of the few-group XS generation methodology for SFR analyses.
In general, all DYN3D benchmark solutions were in good agreement with the experi- mental data indicating the feasibility of using DYN3D in static and time-dependent analyses of SFR cores. Significant discrepancies between DYN3D and the measure-