Many experiments on density stratification breakup in several flow conditions have been performed with the large- and small-scale experimental facilities to understand the mechanism underlying hydrogen behavior in a nuclear containment vessel during a severe accident. To improve the predictability of the RANS (Reynolds-averaged Navier Stokes) approach, we implemented the dynamic modeling for turbulent Schmidt Sc t and Prandtl Pr t numbers. In this paper, the capability of the RANS analysis with dynamic Sc t modeling is assessed with several experimental data obtained by using the MISTRA (Commissariat à l’énergie atomique et aux énergies alternatives, CEA, France), CIGMA and VIMES (Japan Atomic Energy Agency, Japan). For the quantitative assessment, the completion time of the stratification breakup, defined as when helium concentration in the upper region decreases to the same value in the lower region, is focused. The comparison study shows the good performance of the dynamic modeling for Sc t and Pr t . Besides, in the case with the low jet Froude number, the CFD accuracy declines significantly, because the jet upward bending is over-estimated.

During a severe accident of a nuclear reactor, radioactive aerosols may be released from degraded nuclear fuels. Pool scrubbing is one of the efficient filters with a high aerosol removal efficiency, in other words a high decontamination factor (DF). Because of its high performance, many pool scrubbing experiments have been performed and several pool scrubbing models have been proposed. In the existing pool scrubbing experiments, an experimental condition of aerosol number concentration was seldom taken into account. It is probably because DF is assumed to be independent of aerosol number concentration, at least, in the concentration where aerosol coagulation is limited. The existing pool scrubbing models also follow this assumption. In order to verify this assumption, we performed a pool scrubbing experiment with different aerosol number concentrations under the same boundary conditions. The test section is a transparent polycarbonate pipe with an inner diameter of 0.2 m. 0.5 μm SiO 2 particles were used as aerosols. As a result, DF was increasing as decreasing the aerosol number concentration. In order to ensure a reliability of this result, three validation tests were performed with meticulous care. According to the results of these validation tests, it was indicated that DF dependence on the aerosol concentration was not because of our experimental system error including measurement instruments but a real phenomenon of the pool scrubbing.

Dry-out and rewetting phenomenon may occur on a fuel rod surface during anticipated operational occurrences (AOOs) for a boiling water reactor (BWR). The conventional rewetting model included in the current system code tends to underestimate the rewetting propagation velocity due to the absence of an appropriate precursory cooling model. The present research aims at the development of a mechanistic model for the precursory cooling in the annular mist flow regime typical of AOO and anticipated transition without scrum (ATWS). Rewetting experiments were carried out using a single heater rod in the circular glass pipe with air-water flow at atmospheric pressure to visualize the rewetting behavior and obtain mechanistic understanding on the phenomena. This paper summarizes the experimental results and discusses the liquid film flow characteristics including roll wave formation and spattering at the rewetting front.

There has been an extensive reorientation of the light water reactor (LWR) research in Japan since the Fukushima Dai-ichi nuclear power station (NPS) accident, which focuses on severe accidents and accident managements. The Japan Atomic Energy Agency (JAEA) initiated the ROSA-SA project in 2013 for the purpose of studying thermal hydraulics relevant to over-temperature containment damage, hydrogen risk, and fission product transport. For this purpose, the JAEA newly constructed the Containment InteGral Measurement Apparatus (CIGMA) in 2015 for the experiments addressing containment responses, separate effects, and accident managements. Recently, we successfully conducted first experiments using CIGMA to characterize the facility under typical experimental conditions investigating basic phenomena such as buildup of pressure by steam injection, containment cooling and depressurization by internal or external cooling, and density stratified layer mixing by impinging jet. This paper provides an overview of the research programs, the brief description of the facility specification and the outcomes obtained from the first experiments.

Pool scrubbing is one of the effective mechanisms to filter out radioactive aerosols in a severe accident of a nuclear reactor. A lot of work has been done on the pool scrubbing models and experiments. However, large discrepancies still exist between the simulation and experimental results. To improve the pool scrubbing model, an accurate decontamination factor (DF) evaluation by an aerosol measurement and a detailed two-phase flow structure measurement is inevitable. A pool scrubbing experimental apparatus was constructed by the thermohydraulic safety research group in Japan Atomic Energy Agency. The test section is a transparent pipe with the inner diameter of 0.2 m and the length of about 4.5 m. The aerosol laden air flow was injected upwardly into the pool water. The aerosol particle diameter distribution was measured by a light scattering aerosol spectrometer. White polydisperse BaSO 4 particles were used as the aerosol test particles. In the first step, we focused on investigating and reducing the error of DF experimentally. Several problems resulting in the error and their solutions for the error reduction were summarized in this paper. Based on the error reduction methods, the DFs of pool scrubbing were measured in two water submergences. The results showed that the DFs for the aerosol with small diameter were independent of the injecting air velocity in the submergence of 0.3 m. In addition, it was found that the DFs increased with increasing the air flow rate in the submergence of 2.9 m. It was presumed that the increase of DF was dominated by the increase of bubble surface area and/or turbulence intensity with the air flow rate increase, while the effect of the reduced bubble traveling time in the water, which may reduce the DF, was smaller than the increasing effect.

An experiment on a PWR station blackout transient with the TMLB’ scenario and accident management (AM) measures was conducted using the ROSA/large scale test facility (LSTF) at Japan Atomic Energy Agency under an assumption of non-condensable gas inflow to the primary system from accumulator (ACC) tanks. The AM measures proposed in this study are steam generator (SG) secondary-side depressurization by fully opening the safety valves in both SGs with the start of core uncovery and coolant injection into the secondary-side of both SGs at low pressures. The LSTF test revealed the primary pressure started to decrease when the SG primary-to-secondary heat removal resumed soon after the coolant injection into the SG secondary-side. The primary depressurization worsened due to the gas accumulation in the SG U-tubes after the completion of ACC coolant injection. The RELAP5 code predicted well the overall trend of the major phenomena observed in the LSTF test, and indicated remaining problems in the predictions of SG U-tube collapsed liquid level and primary mass flow rate after the gas ingress. The SG coolant injection flow rate was found to affect significantly the peak cladding temperature and the ACC actuation time through the RELAP5 sensitivity analyses.

The analysis on a density stratified layer consisting of multiple gases in the reactor containment vessel is important for the safety assessment of sever accidents. Computational Fluid Dynamics (CFD) code has a potential to clarify detailed stratification phenomena in the containment vessel. In this paper, CFD analyses were carried out in order to investigate the erosion of the stratified layer by a vertical buoyant jet injected from the bottom of a small vessel. Although the Reynolds-Averaged Navier-Stokes (RANS) model is commonly used in industrial applications, it is known that the RANS analyses tend to overpredict effects of turbulent mixing and stratification erosion for these phenomena. This study carried out the RANS and Large-Eddy simulations (LES) in order to understand the detailed phenomena of the stratification erosion in a containment vessel, and clarify the problems of the RANS analysis from the comparison. As a result, although both the RANS and LES models calculated the erosion, the erosion rates calculated by the RANS models were faster than that by the LES model. The calculated erosion behavior was qualitatively different: the LES analyses showed the vertical helium turbulent transport was enhanced only in the radial region directly affected by the impinging jet, while the RANS analyses indicated the occurrences of such transportation at all the radial locations. Although more detailed validation is required using appropriate experimental data, this difference among the calculated cases suggests the importance of the improvement of the turbulence models in order to accurately predict turbulence damping in the stratification layer.