Phase Separation Using a Simple T-junction

Abstract

Abstract Gas/liquid flow dividing at a T-junction can exhibit significant phase maldistribution. This has potential as a simple, compact (partial) phase separator. Though there is by now a reasonable body of knowledge on the performance of such junctions, the data is predominantly for air/water at near atmospheric pressures in small diameter equipment. Moreover, there is as yet no agreed criterion to identify at what conditions and for with what geometry a T-junction is a good (partial) phase separator. This paper presents data from a field trial in which a T-junction was placed on the outlet line from an onshore well. Data were obtained on the phase split at the junction for a number of inlet flow rates and for different resistances in the two outlet legs of the junction. The trends in the data are compared to those for other published work. In addition, a criterion is suggested for good phase separation. The present data, together with that from the literature are then examined and a method for specifying conditions for good phase separation is presented. Introduction In spite of the apparent simplicity of their geometry, T-junctions can have very complex flows, which affect the division of the phases when the junction is used with one inlet and two outlets. However, the simplicity is only apparent, as for gas/liquid applications the junction requires eight variables to define it. These are the diameters of the three pipes making up the junction, the orientation of the inlet and one outlet pipe, the angles between the and the two outlet pipes and the radius of curvature of the corner. There is by now a significant literature on the topic of phase split at junctions. This has been reviewed recently by Azzopardi1. Workers in the field recognise that the effect of dividing a two-phase gas/liquid flow at a junction is, almost inevitably, to produce a maldistribution of the phases. That is, it tends to produce a stream richer in liquid than the feed and a corresponding stream richer in gas. This can have both negative and positive consequences. On the negative side, the maldistribution can result in a fall in efficiency in downstream equipment. There is anecdotal evidence to support this. A bank of four air-cooled heat exchangers (fin-fan coolers) was operated in parallel as condensers. The feed arrived partially condensed at the inlet manifold connecting the exchangers. This consisted of a series of side pipes off a main header, i.e., a series of dividing junctions. In operation the fourth exchanger significantly underperformed. Simple laboratory trials using four pipes connecting the inlet and outlet headers quantified the routes taken by the two phases and showed that the first three side arms each took approximately 30% of the gas and ~10% of the liquid2. The final pipe received most of the liquid. It then became obvious why the fourth condenser was underperforming; it was overwhelmed with liquid. Another example of phase maldistribution has been reported from offshore platforms in the UK North Sea. Here, it had been decided to install two main (phase) vessel separators in parallel. This would enable production to continue albeit at a reduced level if there was a need for maintenance or modification of a separator. To ensure an even split of the phases, an impacting T-junction was employed, i.e., one in which both outlet pipes were at right angles to the inlet. When the system was started up it was found that one separator received most of the gas whilst the other got most of the liquid. Inspection of the pipework upstream of the junction showed that a bend had been positioned at the worst possible place. This bend was centrifuging the phases and presenting each outlet with substantially one phase.