Conference Agenda

In a collaborative effort between Argonne National Laboratory and TerraPower, the high-fidelity computational fluid dynamics (CFD) code NekRS is being used to support the Natrium^® demonstration project. The overall aim of this effort is to use high-fidelity results to augment the available experimental data being used to validate the fast-running lower-fidelity tools used for reactor design. As part of this, NekRS has been used to simulate flow in a 37-pin wire-wrapped bundle using an LES turbulence model based on a high-pass filter. This replicates experiments conducted at MIT by Cheng. This study aims to corroborate the Cheng experimental flow split results via independent means, and validate the methodology used in NekRS for predictions of velocity in wire-wrapped assemblies. By validating NekRS for velocity predictions in wire-wrapped bundles, a firm basis for future work using the same methodology is established. Specifically, flow split at a Reynolds number of 16,170 is investigated and agreement is shown between the NekRS and experimental results to within experimental uncertainty. Details of the methodology will be discussed in the paper, including meshing, the turbulence model, convergence criteria and post-processing techniques. Additionally, the advantage of using a high-fidelity approach will be demonstrated by investigating flow phenomena which were not observable in the original experimental data, such as the velocity in the corner subchannels and the velocity skew across the assembly.

Hydrogen management during severe accidents at nuclear power plants has attracted attention as an important issue since the hydrogen explosion at the Fukushima Daiichi nuclear power plant accident in March 2011. In order to improve hydrogen management under severe accident conditions, the propagation of flames and the resulting loads on structures need to be predicted accurately. For this reason, the use of computational fluid dynamics is expected. Various benchmark experiments have been conducted, and turbulence models, turbulent combustion models, and chemical reaction models have been discussed. However, the uncertainties of each model have not been treated independently. Analysis with uncertainty quantification is necessary to promote efficient research activities through uncertainty-based prioritization and to reflect the latest findings in best practice guidelines. This study aims to establish a methodology for quantifying the chemical reaction uncertainty in turbulent premixed combustion CFD and performs the analyses on existing benchmark experiments. The uncertainties in the rate coefficients for the hydrogen combustion reaction were propagated through a one-dimensional flame propagation analysis to estimate the laminar flame speed uncertainty. Furthermore, the laminar flame speed uncertainty was propagated to a Reynolds-Averaged Navier-Stokes simulation using the turbulent flame speed closure (TFC) model to determine the mean and standard deviation of the maximum flame speed and maximum pressure.

Vapor condensation in the presence of a noncondensable gas is an important topic with practical applications in nuclear reactor safety. Passive Containment Cooling Systems (PCCS) involve shell-and-tube heat exchangers where steam is condensed inside tubes that are cooled by water pools.

Analytical models have been developed to estimate the heat removal of a tube condenser in such conditions. These models involve iterative and marching procedures, which is not warranted in fast running system codes. There is thus the need for direct correlations that provide accurate estimates of total condensation rates when the condenser wall temperature results from the interplay between the tube and shell sides.

A CFD model has been developed (Dehbi et al., 2013) and extensively validated under prescribed condenser wall temperature. In this investigation, we extend the validation to address conjugate heat transfer where both the shell and tube heat transfer are considered.

Two experiments are selected as validation databases, namely the Kuhn tube tests (1997), and the CONAN flat plate tests (2008). Both of these experiments involve well instrumented test sections that allow detailed information to be gathered, e.g. local wall/gas temperatures and heat fluxes.

Excellent agreement between the CFD predictions and experimental data is achieved, with heat flux deviations typically less than 5%. Since the CPU requirements are modest, the CFD model can thus be used in a parametric fashion to provide a numerical database from which easily implementable correlations can be developed using machine learning algorithms. This will be the object of a future investigation.

Phenomena involving complex multiphase gas-liquid flows, encompassing elements such as bubbles and free surface flows, are commonly encountered in various industrial processes, including nuclear applications. When it comes to Computational Fluid Dynamics (CFD) simulations, capturing the transition from low to high void fraction conditions presents a formidable challenge, primarily due to the escalating intricacies at the gas-liquid interface. For instance, gas volume fractions within the range where churn-turbulent and slug flows are prevalent are dominated by exceedingly deformable bubbles. In this intricate scenario, a generalized multiphase CFD modeling approach known as GENTOP stands out. GENTOP adopts the concept of a fully-resolved continuous gas phase, wherein this continuous gas phase encompasses all gas structures that are sufficiently large to be resolved within the computational mesh. However, it is important to note that for a typical user, delving into the complexities and technical nuances of setting up multiphase flow simulations can be quite challenging and laborious.

Turbulent mass transfer strongly influences flow-accelerated corrosion (FAC), a critical issue in designing liquid-metal-based nuclear reactors. Accurate simulation of FAC requires modeling scalar transport processes involving species with very low diffusivities, leading to flows characterized by high Schmidt numbers (Sc). Under such conditions, boundary layers become exceptionally thin, making Eulerian computational approaches prohibitively expensive due to the extensive near-wall mesh refinement required. In our previous research, we proposed a two-layer wall model capable of representing the effects of Schmidt and Reynolds numbers on scalar diffusivity. However, the original model relied heavily on numerical integration, thereby increasing computational demands. To address this, we present a surrogate formulation with explicit integration, reducing computational complexity and simplifying integration into computational fluid dynamics (CFD) codes. This study extends validation efforts to challenging high-Sc-number scenarios involving orifice plate and slot flows under strongly non-equilibrium conditions. Simulations were conducted using the Abe-Kondoh-Nagano (AKN) low-Re k–ε turbulence model. Results confirm that our surrogate two-layer model maintains excellent accuracy in predicting peak near-wall mass transfer without the necessity of extensive mesh refinement (with first-wall grid spacing maintained at y⁺ above 1). Moreover, the model demonstrates improved predictions compared to wall-resolved approach from the literature, especially in capturing non-equilibrium effects downstream of flow disturbances. These findings illustrate that the developed surrogate two-layer model provides both computational efficiency and enhanced accuracy, making it highly suitable for engineering applications involving high-Schmidt-number mass transfer phenomena and FAC predictions.

The Compact Steam Generator (CSG) could play a crucial role in the development of Small Modular Reactors, particularly for the Integral-Pressurized Water Reactor (iPWR), which is gaining significant attention due to its potential to provide safe, reliable, and cost-effective nuclear energy. The Printed Circuit Heat Exchanger (PCHE) is a promising candidate technology that could meet the requirements of the CSG. This study examines the capabilities of existing Computational Fluid Dynamics models for the Printed Circuit Heat Exchanger, considering both single-phase and multiphase conditions, with a focus on the mixture-multiphase approach using the Rohsenow boiling model. The steam generator conditions involve boiling heat transfer, transitioning from subcooled liquid to high-quality steam. This results in a high gradient of mixture density and flow acceleration, which may pose challenges for the CFD solver. This study will discuss these challenges and assess the employed methodology. The results are validated against experimental data from the Georgia Institute of Technology, which conducted experiments on a semicircular channel (≈ 2 mm) PCHE under a wide range of conditions. The results demonstrate good agreement between the simulation and experimental data for both single-phase and multiphase flows across a broad range of conditions, despite the Rohsenow model being developed for pool boiling. Furthermore, the Rohsenow model tends to overpredict heat transfer; therefore, additional calibration of the model may lead to slight improvements in predicting a wide range of flow boiling conditions.