Two-Phase Flow Splitting at a Pipe Tee

Abstract

When a two-phase (gas-liquid) fluid splits at a pipe tee, the gas-liquid ratio usually changes because the phases tend to split in different proportions. This paper discusses experimental verification of the proportions. This paper discusses experimental verification of the flow-splitting phenomenon and presents a method for predicting downstream gas-liquid ratios, with sample calculations for steam distribution networks. Introduction Pipe tees are used widely in pipeline networks, oilfield Pipe tees are used widely in pipeline networks, oilfield flowlines, and refinery streams to split or to combine flows. The flowing medium often is a high-quality two-phase steam or a low liquid-content, two-phase (gas-liquid) fluid, such as gas containing a small amount of condensate. The gas-liquid ratio (GLR) of a two-phase fluid usually changes during splitting because the liquid and gas phases tend to split in different proportions. Some segments of such a flow network may have greatly different GLR'S, operate in unintended flow regimes because of liquid buildup, or become inoperative or bypassed. The flow-splitting problem first was brought to the petroleum industry's attention by Oranje, who reported petroleum industry's attention by Oranje, who reported an observation he made one winter in a natural gas transmission system in The Netherlands. He noticed that condensate forming in the pipeline at low ground temperatures appeared only at certain delivery stations, while gas arriving at other stations remained dry. His subsequent laboratory study showed that this behavior resulted from the gas and liquid phases tending to split at tee junctures in different proportions. Similar observations made in 1975 by Bergman et al. appeared in a study conducted for the AGA. Analyzing data from two gas pipeline networks in Michigan, the study highlighted problems caused by dissimilar splitting of gas and liquid phases. The flow-splitting phenomenon appeared to have an important effect on oilfield steam-injection projects in California. Many operators there split two-phase steam from one header for simultaneous injection into two or more wells. Because this usually involves pipe tees, it was suspected that individual wells receive higher or lower quality steam than that entering the header. Laboratory and field studies, followed by extensive evaluation of the parameters involved, verified this suspicion. Flow parameters studied include gas velocity, liquid content, liquid viscosity, tee attitude (angle), and side-arm entry. Experiments The apparatus used, shown schematically in Fig. 1, had network lines of 3/8-in.-ID Lucite pipe and a 3/8-in.-ID flow-splitting tee made from a Lucite block. Air and plain or viscous water were used as two-phase mixtures. Air from a constant-pressure source passed through an aspirator, where water was injected at rates controlled by a metering valve and a rotameter (Fig. 1). The dry-air flow rate varied between 2 and 10 scf/min and the water rate was between 10 (0.00264 gal/min) and 200 cc/min (0.0528 gal/min). Liquid viscosities were 1 cp for plain water and 5 and 10 cp for water plus hydroxyethyl cellulose. Upstream air pressure measured by Pressure Gauge 3 (Fig. 1) was limited by restrictions in downstream metering equipment, but was kept constant for these air rates: 21.5 psia at 2 scf/min, 17.7 psia at 5 scf/min, and 23.2 psia at 10scf/min. These values gave respective upstream psia at 10scf/min. These values gave respective upstream linear velocities of 30, 90, and 140 ft/sec. JPT P. 290