Coupling CTF with SERPENT for the thermal hydraulic and neutronic simulation of the NuScale SMR

Abstract

South Africa's severe energy crisis has prompted a shift towards diversifying the country's energy mix, with the Integrated Resource Plan specifying that a nuclear new-build programme to the extent of 5200 MW will be implemented to add to the energy mix. Although conventional PWRs will be suitable for this deployment, light-water small modular reactors such as the NuScale SMR can also be deployed. Significant benefits over larger nuclear reactors would be cost-effectiveness and advanced safety features. To enable the deployment of such a reactor design, a safety analysis report must be developed for licensing purposes. Multiple reactor analyses including neutronic and thermal hydraulic analyses are required to complete this report. Since neutron behaviour directly influences heat generation and thermal-hydraulic conditions affect neutron moderation and reactivity, neutronic and thermal hydraulic calculations are strongly interdependent. Hence for best-estimate calculations, coupled neutronic thermal hydraulic calculations are used. This study presents a coupled neutronic and thermal hydraulic simulation of the NuScale SMR using the Monte Carlo neutron transport code Serpent and the thermal hydraulic code CTF together with the North-West University Reactor Code Suite (NWURCS) Version 3 as the code coupler. Initially, a benchmark CTF input model by Avramova was compiled and run successfully with no errors. The Avramova reactor core was modelled as a quarter core with subchannels. The model was then geometrically translated-first to an intermediate 16x16 fuel assembly layout, and finally to the specific 17x17 NuScale fuel assembly geometry. Thorough verification of geometric sub-channel quantities was done, ensuring that flow areas, wetted perimeters and gap lengths accurately reflected the NuScale core design as per Chapter 4 of the NuScale Safety Analysis Report (SAR). The rest of the NuScale CTF base model was built using values obtained from Chapter 4 of the NuScale SAR and input into their specific card groups in CTF. Steady-state conditions were assumed. The guide tubes were modelled as a solid rod. The CTF base model was built and run successfully with no errors. To ensure reliable power profile data, source convergence was verified by monitoring Shannon entropy until it stabilized during inactive cycles. Consequently, only active cycles reflecting a converged fission source were utilized to generate the power profiles required for the SerpentCTF coupling. NWURCS served the purpose of input generation for Serpent and CTF and the coupling. System level coupling was done by NWURCS where Serpent power profiles were mapped to CTF, and the resulting fuel and moderator temperatures were returned to Serpent to update the neutronic state. This process was repeated through several iterations until convergence was achieved. Convergence was monitored visually since stability was also monitored. Establishing convergence using a computer algorithm could lead to false convergence, since the oscillations in the solutions had bounding values larger than 5 oC. To ensure accuracy and reliability, postcoupling verification was performed on key parameters including the effective multiplication factor, detector data, fuel and moderator temperatures and power fields, confirming that the coupled simulation produced stable and consistent results before the base case couplingcalculation was conducted. The base case coupling calculation was initiated using well guessed values for boundary conditions to establish numerical stability over 30 iterations. To ensure reliable results, the boundary conditions were then adjusted step by step to match NuScale specifications. Inlet temperature, mass flow rate, pressure, and power were updated in separate steps. These progressive steps allowed for the isolation of any errors and confirmed that convergence was obtained as it approached the final base NuScale model.The final base case coupling calculations were executed for 42 iterations, and convergence was visually confirmed through the uniform distribution of data points between boundary lines. Statistical analysis over the iterations revealed that the percentage difference for maximum heat structure and moderator temperatures remained below 1%, while maximum power was within an acceptable 5%. The heat structure (fuel) temperature standard deviation was at 1.5 K. Moreover, the standard deviation of the multiplication factor was consistent with the statistical uncertainty of the Monte Carlo Serpent simulation, confirming that the coupled results fall within the expected stochastic range. Finally, the heat structure temperatures were verified by confirming that the axial profiles correctly reflected the sinusoidal like power distribution, with peak temperatures aligning with the maximum neutron flux as dictated by reactor physics. The effect of increasing active cycles on accuracy was evaluated. The results demonstrated that while higher cycle counts reduced the standard deviation (implying improved statistical precision) the impact on the mean values was negligible, with differences remaining below 0.035%. Upon visual inspection no significant effect had been noticed from changing the active cycles.A comprehensive sensitivity analysis further tested the model's reliability, examining its response to changes in inlet enthalpy, mass flow rate, power, and pressure within a range of ±10%. The results indicated that the model is most sensitive to changes in power and inlet enthalpy, both of which produced statistically significant deviations exceeding the coupling standard deviation. Theoretical analysis using fundamental heat transfer equations confirms this behaviour, as power is the primary source term directly influencing fuel and cladding temperatures, while inlet enthalpy sets the starting moderator temperature. Conversely, variations in mass flow rate and pressure resulted in differences that fell within the statistical standard deviation of the coupling, meaningtheir individual effects could not be distinguished from the statistical noise.A difference was noticed when comparing the model's results to the NuScale specifications. This difference could be attributed to the modelling of guide tubes as solid rods (without water flowing within the guide tubes). Other possible causes for the differences were the exclusion of detailed pressure losses and a potential difference in density of water used in CTF and Serpent. Despite this difference, the study successfully verifies the coupling methodology itself. The results prove that the coupling between CTF and Serpent was successful using NWURCS.