Conference Agenda

Thermal radiation plays a crucial role in heat transfer for next-generation nuclear reactors due to the high operating temperatures and reliance on natural convection. In our previous work, we successfully implemented the P1 approximation for thermal radiation in Nek5000/NekRS, a CFD code based on the Spectral Element Method. The implementation was verified against both numerical data and analytical solutions. However, that work focused solely on the fluid domain.

In this paper, we extend the P1 model to the solid domain by integrating it with the Conjugate Heat Transfer model in Nek5000/NekRS. As before, we validate our implementation with reference numerical solution on simple geometries. Then, we apply this approach to the pebble bed case with 1,568 pebbles under salt flow cooling, which was introduced in our previous work, now including the solid domain of pebbles. To ensure accuracy, we conducted parallel simulations using OpenFOAM for comparison.

The presence of a large-scale interface between the two phases presents a modeling challenge related to turbulence in a sense that at a large-scale interface the lighter phase sees the heavier phase like a solid wall. In literature, a widely used strategy is to add a damping source term to the turbulence dissipation equation. These source-term based approaches require that a parameter be tuned based on grid and/or problem. In the present work, a novel approach is proposed where the use of solid wall-function based turbulence treatment is done within the framework of LSI model implemented in Simcenter STAR-CCM+. This is achieved by using a large-scale interface toolkit, which provides the location of the large-scale interface as well as the cell-center to large-scale interface distance; enabling creation of a stencil around the large-interface cell to apply wall-function base damping treatment.

The effectiveness of the new wall-function type interface turbulence damping method is demonstrated in this work though a turbulent air-water co-current stratified flow, studied experimentally by Fabre et al., (1987). The wall-function type treatment shows minimal dependence on grid refinement as well as on the cell aspect ratio when comparison is performed for the full developed velocity profile against the experiment.This is not the case for source term based approach. The observations made using the velocity profile are corroborated for the pressure drop values as well. In addition, the novel approach is shown to not need any problem-based or grid-based tuning.

Two-phase flows are of great interest to the nuclear power community, both in normal operation for coolant systems, and in accident scenarios. Computational multiphase fluid dynamics (CMFD) simulations are a promising tool for detailed analysis of these systems. However, since most CMFD models are based on experiments and analysis of vertical two-phase flows, significant limitations are evident when horizontal bubbly two-phase flows are considered, which have entirely different hydrodynamics arising from the large density difference between phases. In this work, a new experimental database of relative velocity for horizontal bubbly flows is established utilizing the existing 25.4 mm test facility at Purdue University. Local void fraction, gas velocity, and bubble diameter are measured with four-sensor conductivity probes, while local liquid velocity is measured with a Pitot-static probe. This data is used to evaluate the existing CMFD models in ANSYS CFX, demonstrating the problems with the existing models, especially with predicting relative velocity. A recently proposed relative velocity model is implemented in CFX, which improves both the relative velocity and void fraction prediction. Qualitatively, the void fraction location and shape are greatly improved, peaking further away from the wall with an elliptical shape that better agrees with the experimental data. The CMFD predicted area-averaged void fraction matches the experimental data about 10% better with the new model, while the relative velocity is improved by 30%.

Turbulence induced vibration of fuel rods can lead to mechanical wear, which can be responsible for safety issues and significant maintenance costs in Nuclear Power Plants (NPPs). In the context of the “Gathering expertise On Vibration ImpaKt In Nuclear power Generation” (GO-VIKING) European project, fluid structure interaction (FSI) methods are developed and assessed on simpler geometries. Hereby, two-way coupling simulation, of a cantilever rod under a turbulent axial flow have been performed at Reynolds number 21 200. Computational fluid dynamics (CFD) with finite volume method and wall-resolved large eddy simulation (WR-LES) is used for the flow, along with an in-house Euler-Bernoulli beam equation solver for the rod. Free vibrations tests have been performed for the calibration of the mechanical damping of the beam and for validation purposes. The beam motion is accounted for using an Arbitrary Lagrangian Eulerian (ALE) method. The meshes are made of 144 million hexahedra for the flow and 3800 grid points for the beam, respectively. The beam motion and the flow statistics are both investigated; comparisons are drawn with experimental results from the literature. The theoretical and experimental natural frequencies are recovered by the simulation. The R.M.S of the tip displacement of the beam is lower in the simulation than in the experiment. However, this may be due to both the experimental uncertainties at this Reynolds number and a misalignment of the rod in the experiment. Apart from the dissymmetry due to the misalignment, there is quite a good agreement regarding the flow statistics.

Natural convective heat transfer is a prominent research topic in the nuclear energy field and is crucial for the design of passive safety systems. To study the impact of compressibility-induced non-Oberbeck-Boussinesq (NOB-II) effects on natural convection, we conduct lattice Boltzmann simulations of square cavity natural convention in a perfect gas, incorporating the multiple-relaxation-time force model and pseudopotential force. The findings indicate that for a given Rayleigh number (Ra), the Nusselt number (Nu) increases as NOB-II effects strengthen and the Reynolds number (Re) decreases as these effects intensify. This implies that NOB-II effects lead to heat transfer enhancement and convection suppression. The underlying mechanism is as follows (taking the hot fluid as a representative case): under NOB-II conditions, the compression work term absorbs heat from the hot fluid near the central region of the hot wall, resulting in a steeper temperature gradient and a thinner temperature boundary layer near the hot wall. Consequently, the local Nusselt number increases and overall heat transfer is enhanced. Simultaneously, the reduction in the thickness of the temperature boundary layer causes a decrease in the buoyancy difference, ultimately leading to convection suppression. Furthermore, new scaling laws of Nu-Ra and Re-Ra considering NOB-II effects are proposed, with an average error of less than 5%. This study deepens the understanding of natural convection and offers theoretical support for the thermal-hydraulic and safety analyses of advanced reactors.