Sub-Regimes of Vertical Two Phase Annular Flow

Abstract

A convenient method to calculate pressure drop in wells producing natural gas with some water or hydrocarbon liquids is desired in order to design the initial well completion and to consider the technical and economic benefits of a subsequent intervention to extend the operating life of the well. In a laboratory flow loop, air and water have been used as fluid proxies to study two-phase behaviour. Concurrent upward air-water flow has been measured in a 26.1mm internal diameter vertical test section at standard conditions over a range of superficial air and liquid velocities. Several sub-regimes of annular flow were newly observed or refined (pulse/disturbance wave, ripple-wave, partial-wetting, and rivulet) with both still and high speed video images recorded externally. The pressure gradients measured were consistent with previous work under similar conditions. The liquid film within the pipe was examined through Planar Laser Induced Fluorescence (PLIF) imaging, measuring film thickness over a selected range of air and water flow rates. In addition, the onset of droplet entrainment has been observed directly. This data has enabled a new detailed map of sub-regime boundaries to be proposed. Most models for annular flow incorporate a single correlation for interfacial friction without regard to the annular sub-regimes. By observation of computed friction factor and relative roughness data, it is found that the annular region can be represented with three zones of distinct behaviour. In non-entrained flow at high superficial gas Reynolds numbers (Resg >35,000) and laminar superficial liquid Reynolds number (Resl <250) the liquid film exhibits constant relative roughness for a given liquid input. A correlation was derived for superficial gas friction factor as a function of Resl alone. For entrained flow with Resg >35000 and Resl > 250, the film shows relative roughness decreasing as gas rate increases. Another correlation for superficial gas friction factor were derived as a function of both Resl or Resg. Pressure gradients calculated with the new correlations compared well against the experimental database as well as with applicable published data. A third zone (Resg <35,000), close to churn flow regimes, was not amenable to this approach.