Analysis of Two-Phase Tests in Large-Diameter Flow Lines in Prudhoe Bay Field

Abstract

Abstract A total of 29 two-phase flow tests was conducted in two 3-mile-long flow lines in the Prudhoe Bay field of Alaska. Of these, 11 were for a 12-in.-diameter line and 18 were for a 16-in. line. Nine of the tests were in slug flow, and 20 were in froth flow. Flow rates, inlet and outlet pressures, and temperatures were measured for each test. Gamma densitometers were used to monitor flow pattern and to determine mixture densities and slug characteristics. It was found that a modified Beggs-Brill1 pressure-loss correlation predicted culled data to within − 1.5% on the average compared with + 11.4% for a modified Dukler-Eaton2,3 correlation. Very little scatter was observed with either method. Analysis of flow-pattern observations showed that none of the slug-flow tests were in the Schmidt4 severe slug region characterized by extremely long slugs. It also was found that the slug/froth (dispersed) flow-pattern boundary existed at a much lower liquid flow rate than predicted by either Mandhane et al.5 or Taitel and Dukler.6 Four of the slug-flow tests in 16-in. lines lasted for a sufficient time to permit statistical analysis of slug-length distributions. Sixteen additional tests on 4- and 7-in.-diameter pipe reported by Brainerd and Hedquist* were analyzed statistically. It was found that slug lengths could be represented by a log-normal distribution. A regression analysis approach was successful for estimating the mean slug length for stabilized flow as a function of superficial mixture velocity and pipe diameter. The extreme percentiles of the slug-length distribution then can be computed using standard probability tables, making possible probability statements about expected maximum slug length. A mechanistic analysis of the slug-flow tests resulted in equations for predicting slug velocities, liquid holdup in both the liquid slug and the gas bubble, and the volumes of liquid that are produced and overrun. These parameters are important for predicting liquid-slug effects on separator performance.