Gas-Rise Velocities During Kicks

Abstract

Summary. Experiments to examine gas migration rates in drilling muds were performed in a 15-m [49-ft] -long, 200-mm [7.8-in.] -ID inclinable flow loop where air injection simulates gas entry during a kick. These tests were conducted using a xanthan gum (a common polymer used in drilling fluids) solution to simulate drilling muds as the liquid phase and air as the gas phase. This work represents a significant extension of existing correlations for gas/liquid flows in large pipe diameters with non-Newtonian fluids. Bubbles rise faster in drilling muds than in water despite in the increased viscosity. This surprising result is caused by the change in the flow regime, with large slug-type bubbles forming at lower void fractions. The gas velocity is independent of void fraction, thus simplifying flow modeling. Results show that a gas influx will rise faster in a well than previously believed. This has major implications for kick simulation, with gas arriving at the surface earlier than would be expected and the gas outflow rate being higher than would have been predicted. A model of the two-phase gas flow in drilling mud, including the results of this work, has been incorporated into the joint Schlumberger Cambridge Research (SCR)/BP Intl. kick model partly funded by the U.K. Dept. of Energy. Introduction The rate at which free gas rises up the wellbore has been identified as a key parameter in the development of a gas kick in a well. This paper reports a comprehensive experimental study into the two-phase paper reports a comprehensive experimental study into the two-phase flow characteristics that will occur in a well during drilling. These tests were conducted in a large vertical pipe using fluids with a rheology similar to that of drilling muds. Before this experimental study, some correlations existed for two-phase flow in circular pipes (usually small diameter) and annuli. Only very limited data were available for multiphase flows of non-Newtonian fluids. The experimental tests used a 200-mm [7.8-in.] -diameter, 12-m [39-ft] -long pipe. Because of operational reasons, a xanthan gum solution was used for the liquid phase whose behavior was rheologically similar to that of a genuine drilling mud. Data analysis techniques are also presented. The results of the gas/mud experiments are presented for the pipe geometry. Comparison between the results for gas/water flows (which are the basis of most existing kick simulators) and the gas/mud flows shows some quite striking differences. In this paper we review the existing literature in this field and describe the experimental facility used for the tests and the selection procedure used for the drilling mud analog. We compare the gas-rise velocities between air/water flows and air/mud flows. Finally, we present some results from the SCR kick simulator that demonstrate the importance of having an accurate gas-rise model. Gas/Liquid Flows Much of the two-phase flow literature is experimental and is based mainly on air/water flows in small pipes or annuli, as in the work of Aziz et al. Many authors have made theoretical studies, notably Zuber and Findlay, who derived a model that stated =CovH+vs, ....................................... (1) where = mean gas velocity and vs = gas-bubble slip velocity relative to a stationary fluid. The homogeneous velocity, vH, can be defined as VH = Qg + QL)/A, ............................... (2) where Qg and QL are the volumetric flow rates for the gas and liquid phases, respectively. They proposed that the coefficient Co was related to the distribution of bubbles an d their relative velocities across the pipe. If the gas is concentrated in the center of the pipe, where the liquid has its peak velocity, then the gas will be convected more quickly than the mean flux. By suggesting plausible velocity and void-fraction profiles, they showed that Co would range from 1.0 to 1.5. This observation was later supported with experimental results. Other authors have discussed the rise velocity of isolated bubbles in stationary columns of liquid. Wallis assumed Stokes flow both around and inside the bubble. Govier and Aziz carried this approach further. In air/water flows this approach is limited to bubbles of less than 2-mm [0.078-in.] diameter and is therefore of very little use here. Harmathy developed a correlation for experimental data to describe the rise of single, slightly larger, bubbles as a function of density difference and surface tension. This correlation, which is independent of bubble size, gives Vs = 1.53[g(pL - pG) /pL2] 1/4 ....................(3) where pL and pG=liquid and gas densities, respectively, and a = interfacial tension (IFT). For air and water, at 100-kPa [1-bar] pressure, this predicts a bubble velocity of 0.25 m/s [0.82 ft/sec]. pressure, this predicts a bubble velocity of 0.25 m/s [0.82 ft/sec]. For larger bubbles that almost fill the pipe, the slip velocity is limited by the rate at which the liquid phase will fall past the gas. Davies and Taylors considered inviscid flow around the bubble nose in order to evaluate vs. They derived the equation vs=0.35 g(PL-PG)d/pL,.............................. (4) where d=pipe diameter. This is normally referred to as the "Taylor" bubble velocity. It uses the pipe diameter as the scaling parameter. For a 200-mm [7.8-in.] -diameter pipe, this predicts a parameter. For a 200-mm [7.8-in.] -diameter pipe, this predicts a velocity of 0.5 m/s [1.6 ft/sec]. A number of vs models have been suggested for use in kick simulators that are based on these air/water flows. Most use Eq. 1 to model the flow, with v, calculated from a combination of Eqs. 3 and 4. Typical of these is the model proposed by Nickens, who assumed that Co = 1.0 for bubbly flow and Co = 1.2 for slug flow, with the flow regime determined by the void fraction F, where F=Ag/A......................................... (5) Flows with a void fraction of F less than25% were assumed to have a bubbly flow with vs from Eq. 3. For F>85% the gas transport was assumed to be entirely by Taylor bubbles, with vs calculated from Eq. 4. A linear transitional zone linked the two regimes. More representative geometries and fluids were used by Nakagawa and Bourgoyne, who reported tests made with air and water that used a 15-m [49-ft] -long, 150-mm [6-in.] pipe with a center body. Unfortunately, scatter in the experimental data makes it difficult to draw any realistic conclusions. Rader et al. studied the rise of gas swarms injected at the bottom of a vertical well but had little control on the gas injection conditions. Experimental Faculty The multiphase flow-loop test facility at SCR, Fig. 1, forms a universal multiphase flow test center. In the present test configuration, shown schematically in Fig. 2 and described in detail by Johnson and White, it has been used for gas/liquid flows, although solid/liquid and liquid/liquid flows can also be evaluated. SPEDE P. 257