Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
安部 諭; 岡垣 百合亜; 石垣 将宏; 柴本 泰照
Proceedings of OECD/NEA Workshop on Virtual CFD4NRS-8; Computational Fluid Dynamics for Nuclear Reactor Safety (Internet), 11 Pages, 2020/11
The fifth international benchmark exercise (IBE-5), the cold-leg mixing CFD benchmark, was conducted under the support of OECD/NEA. The experiment for IBE-5 was designed to visualize the mixing phenomena of two liquids with different density in a horizontal leg (as a simulant of the cold-leg) and downcomer. This paper shows our CFD result on the open test condition in IBE-5. We selected the Large-eddy simulation (LES) solving the filtered equation of flow and concentration fields. Regarding the eddy-viscosity to model the turbulence flux of the momentum at sub-grid scale (SGS), Wall-adapting locale eddy-viscosity (WALE) model, a modified version from the Smagorinsky model, was applied. The experimental geometry was resolved with three different numerical mesh systems. The CFD analysis predicted the laminar-like flow behavior in the horizontal leg. Due to the large density difference between the two liquids, the turbulence production was suppressed strongly, and the velocity fluctuation in the horizontal leg became very slow and small. In contrast, the strong turbulence mixing in the downcomer was predicted. The plume from the horizontal leg entrained with the surroundings and spread circumferentially in the downcomer. The comparison with the TAMU experimental data reveals the good performance of the WALE model. In addition, we discuss the appearance characteristics of the high concentration of the heavy liquid in the downcomer in the LES. The Probability Density Function (PDF) and Cumulative Distribution Function (CDF) are derived based on the predicted time-series of the heavy liquid concentration. The PDF around the mean concentration in the case with the low mesh resolution is larger than that predicted by the higher resolution due to the excessive homogenization of the heavy fluid concentration. This study reveals the importance to note the required mesh resolution to predict the appearance event of the high concentration.
岡垣 百合亜; 柴本 泰照; 安部 諭
Proceedings of OECD/NEA Workshop on Virtual CFD4NRS-8; Computational Fluid Dynamics for Nuclear Reactor Safety (Internet), 12 Pages, 2020/11
A bubbly flow with a single injection orifice is numerically analyzed for pool scrubbing phenomena using different computational fluid dynamics (CFD) methodologies. The calculation covers the total regime of pool scrubbing from air injection to bubble swarm through the transition region. Such two-phase flow behaviors strongly affect particle removal in a bubble. The experimental bubbles are known to be oblate spherical and exhibit secondary motion, including path instability and shape oscillations. Moreover, bubbles in a swarm are subject to coalescence and breakup. While these may well affect bubble internal heat/mass transfer and particle capture, no established way is available for considering such influences in practical calculations. Pool scrubbing code SPARC-90 uses an oblate spherical bubble model but assumes a steady, rectilinear bubble rise without secondary motion. The 3-D CFD has the potential to capture the bubble interaction in the swarm region in detail. In the present study, the experiment by Abe et al. (Nuclear Engineering and Design 337, 2018) was referred for the calculation, and their data were used to validate if the CFD simulation can predict the flow transition accurately. Two types of solvers based on the volume of fluid (VOF) method and the simple coupled volume of fluid with level set (S-CLSVOF) method are used for the interface capture. The two solvers were validated by comparing with the experimental results. As a result, the void fraction profiles along the vertical central axis were in good agreement with the experimental data, regardless of the solvers, and those along horizontal lines in a central plane slightly improved with the S-CLSVOF method by the more accurate calculation of the surface tension.