Interfacial Area Transport Equation Models and Validation against High Resolution Experimental Data for Small and Large Diameter Vertical Pipes.

Abstract

For analyses of Nuclear Power Plants, the current state-of-the-art model for predicting the behavior of two-phase flows is the two-fluid model. In the two-fluid model, balance equations are coupled together through transfer terms that depend on the area of the interface between liquid and gas. Efforts in the past have been focused on the development of an interfacial area transport equation model (IATE) in order to eliminate the drawbacks of static flow regime maps currently used in best-estimate thermal-hydraulic system codes. The IATE attempts to model the dynamic evolution of the gas/liquid interface by accounting for the different interaction mechanisms (i.e. bubble break-up and coalescence).

The further development and validation of IATE models has been hindered by the lack of adequate experimental databases in regions beyond the bubbly flow regime. At the TOPFLOW test facility, experiments utilizing wire-mesh sensors have been performed over a wide range of flow conditions, establishing a database of high resolution (in space and time) data. The objective of the dissertation is to evaluate and improve current IATE models using the TOPFLOW database and to assess the uncertainty in the reconstructed interfacial area measured using wire-mesh sensors.

The small-diameter Fu-Ishii model was assessed against the TOPFLOW 52 mm data. The model was found to perform well (within the experimental uncertainty of ±10%) for low void fractions. At high void fractions, the bubble interaction mechanism responsible for poor performance of the model was identified. A genetic algorithm was then used to quantify the correct incidence of this mechanism on the overall evolution of the interfacial area concentration along the pipe vertical axis.

The large-diameter Smith-Schlegel model was assessed against the TOPFLOW 198 mm data. This model was also found to perform well at low void fractions. At high void fractions, the good agreement between the model predictions and the experimental data were found to be due to a compensation of errors. Studies using the genetic algorithm indicated significant performance improvement for the DN200 data. However, the improvement in prediction capabilities could not be reproduced when the model was assessed against independent large-diameter databases available in the literature.