Safety systems, like the passive containment air-cooling system, are used to remove heat from a containment during accident conditions in advanced nuclear reactors, including small modular reactors having a steel containment. For the effective performance of the passive containment air cooling system with and without a chimney, it is highly important to understand the natural convection air flow around the containment. The experimental investigation of passive containment air cooling of an experimental test setup with and without a chimney, with a geometry very similar to that of an actual reactor, is presented in this paper. Comparisons between the air-side transient natural convection characteristics with and without a chimney were studied in detail based on the temperature measurement. The effect of a chimney was found to have an adverse effect on natural convection behavior due to possible air recirculation zones and restrictions to air flow due to its present shape.

Depletion simulations of flowing-fuel molten salt reactors do not typically distinguish between the in-core and ex-core regions. Instead, they either irradiate the entire primary loop while scaling the flux, or irradiate the in-core volume alone, which leaves some amount of fuel salt out of the simulation. Either approximation results in some level of inaccuracy in the nuclide concentrations. The method of scaling the flux is more commonly used, but this approach assumes perfect mixing of the fuel salt in the primary loop. This work investigates a spatially resolved method to account for the impact that the flow and chemical removal have on the concentrations of nuclides rapidly and accurately. Specifically, the spatially resolved model uses one-dimensional advection, which assumes no mixing in the primary loop. This model is used to provide the most conservative possible comparison, as any amount of mixing would only move the result closer to the scaled flux method. Comparisons are made in this work between this spatially resolved model, a scaled flux model, other work from the literature, and experimental data. These comparisons show that the scaled flux method is sufficient for modeling most nuclides unless there is a large spatially dependent loss term, near-stationary flow rate, or artificial power oscillations with almost no mixing. In such cases, a spatially resolved method is necessary.

Cognitive style, as a stable individual characteristic, provides critical insights into safety training and personnel selection in nuclear power plants; however, the underlying mechanism by which cognitive style influences situation awareness (SA) remains unclear.This study combines subjective scales, behavioral tasks, eye movements, and electroencephalography data to explore the influence of cognitive styles on the three stages of operators’ SA: perception, understanding, and prediction. It also constructs a SA assessment model based on cognitive styles. The results revealed that field-dependent operators were more susceptible to information interference during the comprehension and projection stages, leading to increased cognitive and mental workloads, accompanied by significant activation in the parietal and temporal regions. These effects subsequently impaired attentional resource allocation and situational comprehension abilities, resulting in lower overall SA compared to field-independent operators. The grey wolf optimizer-support vector machine model constructed using subjective, physiological, and behavioral features achieved an identification accuracy of 83.3%. This research provides theoretical and methodological support for enhancing safety training and personnel selection in nuclear power plant control rooms.

Machine learning (ML) has been leveraged to tackle a diverse range of tasks in almost all branches of nuclear engineering. Many of the successes in ML applications can be attributed to the recent performance breakthroughs in deep learning and the growing availability of computational power, data, and easy-to-use ML libraries. However, these empirical successes have often outpaced our formal understanding of the ML algorithms. An important but underrated area is the uncertainty quantification (UQ) of ML. ML-based models are subject to approximation uncertainty when they are used to make predictions due to sources including but not limited to, data noise, data coverage, extrapolation, imperfect model architecture, and the stochastic training process. The goal of this paper is to clearly explain and illustrate the importance of the UQ of ML. We elucidate the differences in the basic concepts of UQ of physics-based models and data-driven ML models. Various sources of uncertainties in physical modeling and data-driven modeling are discussed, demonstrated, and compared. We also present and demonstrate a few techniques to quantify the ML prediction uncertainties, including Monte Carlo dropout, deep ensemble, Bayesian neural networks, Gaussian processes, and conformal prediction. Finally, we discuss the need for building a verification, validation, and UQ framework to establish ML credibility.

Accidental coolant leaks in Sodium-cooled Fast Reactor (SFR) systems can create sodium fire, which is detrimental to safe operation of the reactor because of thermal damage by the fire event and generation of harmful, corrosive reaction products. Hence, in the safety evaluation of SFRs, validated models for the assessment of sodium fire consequences under various postulated leak events is essential. A leaked sodium jet can be subjected to fragmentation because of interfacial instabilities while falling in the air and can generate a large number of droplets of different sizes, and their burning in a reactor cell creates a spray fire scenario. The thermal consequences of spray fire depend on the droplet size distribution (DSD) formed since it determines the surface area of leaked sodium exposed to surrounding air. In accidental sodium sprays, the DSD can vary widely based on the scenario of spray formation, and hence, assessment of appropriate DSD is essential in numerical modeling of sodium spray fire events. The present study considers the underlying fragmentation mechanism of the liquid streams in various configurations envisaged under accidental leak conditions in SFR systems to evaluate the representative mean size of sodium spray droplets generated, using the numerical models and semi-empirical correlations based on the linear stability theory. The mean droplet sizes predicted by the models developed for various leak scenarios have been validated using experimental results available in the literature for liquid sodium and other Newtonian liquid sprays. Liquid stream breakup models have been incorporated into the in-house one-dimensional sodium spray fire analysis code to predict the leak scenario–specific mean size and define the resultant DSD using suitable mathematical functions to simulate benchmark sodium spray fire tests. Comparison of the thermal transients predicted by the revised code and experimental results revealed the efficacy of liquid stream breakup models in predicting the appropriate DSD and resultant sodium spray fire consequences. The physical mechanism–based, configuration-dependent models for defining the accidental sprays improved the sodium fire analysis required in the safety evaluation of SFRs.

This paper investigates the weight function distribution in the electromagnetic flowmeter with conductive pipe wall by combining the virtual current density method and COMSOL Multiphysics simulation software. It was found that the average weight function values were identical in any arbitrarily small annular domain in the pipe wall or fluid. Additionally, the ratio of the average weight function of the fluid domain to the pipe wall domain was only related to the ratio of their conductivity. Then a correlation for the average weight function ratio of the fluid and pipe wall was derived for the first time. The weight function correlation of the fluid domain was subsequently obtained by combining the physical meaning of the weight function. Finally, a theoretical model was derived based on the weight function correlation and the control equation of the electromagnetic flowmeter. The deviation between this model and the calculation results of the electromagnetic flowmeter model established by COMSOL Multiphysics was less than 4%.

Composite ceramic-metallic (cermet) fuels, such as UO2-Mo, have gained attention as advanced alternatives to traditional UO2 fuel, primarily due to their potential to address the issue of poor thermal conductivity in UO2. This poor thermal conductivity can lead to challenges like elevated fuel temperatures, increased fission gas release (FGR), and heightened clad hoop strain. UO2-Mo composite fuel offers a potential solution by incorporating molybdenum to significantly enhance thermal conductivity and mitigate these issues. In this work we develop a new thermal conductivity model for UO2-Mo fuel and implement it in the ENIGMA fuel performance code to investigate the impact of the increased thermal conductivity on fuel behavior under pressurized water reactor (PWR) operating conditions. We simulate a typical PWR fuel rod, except with UO2-Mo fuel, under both steady-state irradiation and transient scenarios. Predictions of fuel temperature, FGR, and clad hoop strain are compared against conventional UO2 fuel, with results indicating significant advantages for UO2-Mo, including reductions in all three metrics. While not investigating the effects of other differences in material properties and phenomenological behavior of UO2-Mo fuel (in part due to the lack of irradiation data), the results provide an initial quantification of the benefits of the increased thermal conductivity on UO2-Mo fuel performance.

Traditional engineering-based core analyses generally rely on a two-stage approach, involving a transport code and a diffusion (e.g. nodal) solver. The accuracy and fidelity of the core solution strongly depend on the generation of few-group constants, which is a topic that has been extensively studied. However, the practice of calculating group constants with Monte Carlo (MC) codes is still relatively new. The main challenge of this approach is the generation of diffusion coefficients. A common approach adopted by many MC codes is the flux separability approximation (FSA). However, diffusion coefficients generated with FSA have trouble matching the heterogenous solution. This is due to a strong coupling between the energetic, spatial, and angular distributions of neutrons. This is a well-documented challenge, with some literature proposing consistent methods to handle energy-angle dependence and others proposing consistent methods to handle energy-space-angle dependence. Recently, the Cumulative Migration Method has been used to generate consistent diffusion coefficients using MC. However, these diffusion coefficients are valid only for infinite single region problems. As such, it is important to revisit the consistent group constant generation method to better handle the energy-space-angle dependence using MC. This paper will lay out the method to generate these consistent group constants using MC and then verify that they can then be used for a two-stage approach using a steady-state nodal expansion method.

The upward displacement of molten fuel material (or actinides) from the core region due to a reactor-scale fuel bubble expansion–compression, during energetic core disruptive accident (CDA), is essential in evaluating the accident source term for a pool-type sodium fast reactor (SFR). A numerical model is developed to estimate the molten fuel mass distribution in the reactor vessel at the end of the fuel bubble’s expansion–compression phase. The model evaluates the Lagrangian trajectories of representative fuel droplets during the upward displacement from the damaged core. Subsequently, the fuel mass removed from the core region, the fuel mass absorbed by the sodium pool, and the fuel mass that remains in the bubble region after the cessation of the bubble’s expansion–compression cycles are estimated. The fuel droplet diameters are evaluated by considering both mechanical and thermal modes of fragmentation. The initial fuel bubble temperature ranges from 4200 to 4700 K, and the fuel bubble mass is between 1000 and 3000 kg. These values are typical for a medium-sized pool-type SFR under energetic CDA conditions. The fuel droplet trajectories are evaluated by considering drag and gravity as dominant forces under a dilute flow regime. The local fluid (fuel vapor) velocity required to evaluate droplet trajectories is supplied by a fuel bubble model, which accounts for sodium entrainment and the fuel–sodium heat transfer. Analysis results show that the fuel droplet diameter strongly depends on the initial fuel bubble temperature, and it is insensitive to the fuel bubble mass. Also, the sodium entrainment in the bubble region retards the fuel droplet absorption by the sodium pool. Fuel droplets generated by mechanical fragmentation are effectively removed from the bubble region by gravitational settlement. Similarly, inertial impaction effectively removes the thermal fragmentation–generated fuel droplets from the bubble. Results show that the fuel droplets with diameters less than 3 µm (up to 183 kg) remain suspended in the fuel bubble region at the end of the bubble expansion. These suspended fuel droplets can reach the cover gas space during the subsequent buoyancy rise of the fuel/sodium vapor bubble swarm. Results form the basis for the estimation of the accident source term in a reactor containment building due to the fuel bubble phenomenon under energetic CDA in a pool-type SFR.

Analyses of the Advanced Fuels Campaign Fission Accelerated Steady-state Test (AFC-FAST) in the Advanced Test Reactor are presented. A detailed methodology was employed to better account for uncertainties in the planned power and duration of sequential reactor loading cycles. By performing coupled depletion analyses at multiple power levels and durations, the differences in experiment heating outputs can be found. The effects of these uncertainties upon multiple experiment configurations were assessed in an effort to streamline the process of planning for and documenting future irradiations. The data generated from this work have been used to help inform assumptions on subsequent projections to perform only a nominal case depletion.