It is well known that obstructions in two-fluid flow act as mixers, homogenizing the flow towards a more uniform phase distribution, and also causing a significant reduction in relative velocity. Following a recently published study in which a two-dimensional two-fluid computer program was used to demonstrate that these effects could be modelled in concept, this paper extends the analysis to three dimensions and compares results to those obtained .in a recent experiment.
In modelling flow in pipes, one dimensional analysis is adequate to simulate overall or bulk energy transfer, but multi-dimensional analysis is necessary to model detailed local distribution of flows and temperatures. In two-phase flows, the homogeneous or equilibrium model, which considers the flow to be a uniform mixture, is adequate when the phases are well mixed. The homogeneous model cannot be used to model situations in which the phase velocities are significantly different, instead a two-fluid model is required , as this permits the phases to have separate velocities, and therefore to redistribute in response to gravity or centrifugal effects.
Recent experiments reported by Salcudean, et al (1983) studied the effect of flow obstructions and pressure drop in horizontal flow of air and water. These experiments showed that a flow obstruction causes a sequence of inter-related phenomena:
sudden pressure drop subsequent partial recovery (iii) significant mixing of the two fluids in the recovery zone (iv) significant reduction of the air velocity in the recovery zone (v) an eventual return toward stratification.
Similar observations were reported by Gardner and Neller (1970) and Salcudean and Leung (19841, but detailed measurements of void distribution were taken in neither case.
It is of interest to know whether a two-fluid model can incorporate the mechanisms necessary to simulate these phenomena. In a previously published study, Carver and Salcudean (1984) showed that all these phenomena could be simulated in concept using a two-dimensional twofluid computer program. No attempt was then made to simulate specific experimental conditions, as the problem is clearly three-dimensional in nature. However, the twodimensional analysis did reproduce the appropriate trends, and the results were sufficiently successful to suggest that a three-dimensional analysis would reproduce quantitative comparisons. This paper describes the extension of the previous study to three dimensions and compares computed results to those obtained by Salcudean et al (1983).