Void fraction covariance has been introduced into the interfacial drag calculation used to close the one-dimensional two-fluid model. A model for void fraction covariance has been developed for large diameter pipes. The newly developed model has been compared with two previously developed models in terms of void fraction prediction accuracy. The effects of these additions on the void fraction prediction uncertainty have been evaluated utilizing a computational tool developed in MATLAB. The results indicate that there are small differences in the void fraction prediction between the models evaluated and the two-fluid model without void fraction covariance. Higher void fractions above 0.7 show the most significant changes. However, the differences in the uncertainty are not significant when compared to the uncertainty in the data used for the comparison. The results highlight a need for additional data for higher void fractions, collected with steam–water systems in large diameter pipes.

This study presented the state-of-the-art computational fluid dynamics (CFD) validation and scaling of the condensation heat transfer (CHT) models for passive containment cooling system (PCCS) of the small modular reactor (SMR). The STAR-CCM+ software with real 3D computational domains was used to validate the condensation models with a preliminary assessment of pure steam scaling performance. The boundary and appropriate physics conditions from the test data were applied. The condensation was modeled using the condensation-seed parameter as a source term for mass, momentum, and energy conservation equations. A small percentage of air (within 1%) was considered in the test section; hence, multi-component gas models were used. The implicit-unsteady numerical solver was applied to improve numerical stability. Mesh size, run time (duration), and time step sensitivity analyses were applied to obtain optimized simulation results. The test fluid parameters—temperature (at bulk steam-mixture, bulk coolant, inner and outer tube walls), condensation film thickness, mass fraction, and heat flux—were utilized to validate the CFD simulations. Finally, Nusselt number (Nu), as the dimensionless number heat transfer, was calculated for diameter scaled-up and scaled-down geometries. The heat transfer coefficient and Nu values were compared to evaluate the scalability performance of CHT models.

This paper presents the computational fluid dynamics (CFD) validation and scaling assessment of the condensation heat transfer (CHT) models in the presence of non-condensable gas for the passive containment cooling system (PCCS) of the small modular reactor (SMR). The STAR-CCM+ software with 3D scaled-down SMR containment geometries was used in CFD simulations with steam and non-condensable gas (NCG). The limitations and approximations of the previous studies were resolved to avoid scaling distortion and uncertainties. Air was used as the NCG gas with steam. The multi-component gas model was used to define the steam–NCG mixture, and the condensation-seed parameter was used as the source term for the fluid film model. Three different turbulence models were used to check the heat flux performances and temperature distributions on the coolant side. The heat flux was estimated from the axial coolant bulk temperature, which was identical to the test data reduction method. An implicit-unsteady numerical solver was applied to the conjugate heat transfer models between the gas, liquid, and solid regions. Detailed simulations were performed, and simulation results were validated with the measured parameters experimentally. The condensation heat transfer performance was quantified using non-dimensional numbers and compared for different scaled geometries to identify the scaling distortions.

Void fraction prediction of the one-dimensional two-fluid model has been evaluated utilizing a computational tool developed in MATLAB. Various drift-flux correlations, specifically those of Ishii, Kataoka and Ishii, Chexal-Lellouche, Hibiki and Ishii, Shen et al., and Schlegel et al., have been used in the calculation of interfacial drag. The uncertainty in the void fraction prediction by the two-fluid model has been evaluated for each of the specified models under one-group conditions for the gas phase. In addition, several two-group drift-flux models have been implemented for the prediction of interfacial drag, and the accuracy has been compared to the accuracy of the one-group predictions. The results indicate that drift-flux correlations in the same "family" show minor improvements when using updated models. The models of Hibiki and Ishii and Schlegel et al. show the best results of the models included in this study. Some shortcomings were observed for the Chexal-Lellouche correlation. The two-group approach has shown significant error reduction over one-group models. The results highlight the need for properly formulated drift-flux correlations based on physical principles rather than curve fits, as those curve fits can hide compensating errors. Additional data in large diameter pipes, collected with steam-water systems at various pressures, would be extremely valuable for further analysis.

In commercial nuclear reactors, heat exchangers, and bubble column reactors, two-phase flows are present. When predicting the safety and process efficiency of these systems, it is necessary to model the behavior found in them. The most common model used in two-phase flows is the two-fluid model due to its practicality. In the two-fluid model, two key parameters are the void fraction (VF, also known as the gas fraction or gas holdup) and interfacial area concentration (IAC, also known as interfacial area density). In order to produce accurate results, the bubbles are separated into groups based on the transport properties. When benchmarking models, experimental data are required. In many cases the experimental data are produced with the use of intrusive conductivity or optical probes. Recently a new data processing algorithm was developed to improve bubble interface identification and implement a method to group bubbles based on diameter rather than chord length. In this paper, the new data processing algorithm is evaluated by comparing the results when using both conductivity and optical probes. At a data acquisition frequency of 22 kHz, the optical probe collected more bubbles than the conductivity probe using the old algorithm. The new algorithm results in similar bubble counts for both instruments. There is a shift in bubbles from Group 1 to Group 2 in both the optical and conductivity probes. The new bubble size calculation means that several bubbles, which were previously classified as "spherical/distorted" , are now classified as "cap/slug/churn" bubbles for both the optical and conductivity probes. However due to low sample rates used in this research, the IAC is larger for the conductivity probe when compared to the optical probe by 10% to 60%. While some of these changes were expected, the increase in the IAC was larger than the reported uncertainty of the instruments.

Many industrial systems make use of two-phase flows for processing or safety applications. Modeling these flows is essential for ensuring safety and engineering optimization. In general, these flows are modeled using the two-fluid model. Two key parameters in the two-fluid model are void fraction and interfacial area concentration. To improve model accuracy, the bubbles are typically broken up into groups based on transport properties and modeled separately. Such models must be validated using experimental data, which is often collected using intrusive probes such as electrical conductivity or optical void probes. Current algorithms for converting the signals from these instruments into void and interfacial area measurements struggled with missing interfaces due to signal rise and fall time. These types of instruments also use the chord length to classify the "group" of a bubble, which can lead to incorrect grouping behavior. In this paper, a new dynamic signal processing method and a grouping algorithm based on calculating bubble diameter have been introduced in an attempt to correct these inaccuracies. The ability of the new algorithm is to correctly identify smaller bubbles and to more accurately identify bubble signals is demonstrated by comparison of the output logic pulse for both the old and new algorithms with the same input signal. The new bubble size calculation means that a number of bubbles that were previously classified as "spherical/distorted" are now classified as "cap/slug/churn" bubbles. This leads to changes in average bubble properties. While these changes were expected in several cases, the increase was larger than the reported uncertainty of the instruments. This may indicate significant shortcomings in data analyzed using the previous algorithms. Additional data collection and analysis is required in order to evaluate this possibility. However, the new algorithm has a significant weakness: the bubble diameter calculation increases the computational time by an order of magnitude.