Comparison of Various Formulations of the Drift-Flux Model in Calculating Unsteady Pipe Flows

Abstract

The drift-flux models are the fairly popular choice in the development of engineering applications for simulating multiphase flows in pipes. For this mathematical model, several different ways of writing the system of equations are known in the literature; however, there are no convincing comparison of the results that could be used to make a reasoned choice in favor of a particular model for a given application. This paper reviews several formulations of the drift-flux model in the case of one-dimensional cross-section-averaged equations. Three versions of the model are compared on a number of test problems in terms of the accuracy of modeling the results and the volume of required computations. The obtained results demonstrate good agreement between the different models for steady-state flows, but at the same time, they have significant differences in simulating unsteady flows. Moreover, the choice of a more theoretically justified formulation of the drift-flux model does not provide any significant advantage from the standpoint of comparison with experimental data, but significantly increases the computational effort. The obtained results suggest that differences in the formulations of equations contribute less significantly to the error in the simulation results than ambiguities in the empirical correlations used to close the drift-flux model.