Vous êtes sur la page 1sur 9

IADC/SPE 87185 Experimental Assessment of Fluid Velocity to Control Gas Accumulation and Removal While Drilling "Horizontal" Wells

Firat Ustun, SPE, Schlumberger, Dimitris E. Nikitopoulos, Louisiana State University, John R. Smith, SPE, Louisiana State University, and Adam T. Bourgoyne, Jr., SPE, Bourgoyne Enterprises

Copyright 2004, IADC/SPE Drilling Conference This paper was prepared for presentation at the IADC/SPE Drilling Conference held in Dallas, Texas, U.S.A., 24 March 2004. This paper was selected for presentation by an IADC/SPE Program Committee following review of information contained in a proposal submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the International Association of Drilling Contractors or Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the International Association of Drilling Contractors or Society of Petroleum Engineers, their officers, or members. Papers presented at IADC/SPE meetings are subject to publication review by Editorial Committees of the International Association of Drilling Contractors and Society of Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the International Association of Drilling Contractors and Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to a proposal of not more than 300 words; illustrations may not be copied. The proposal must contain conspicuous acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

anced drilling can result in accumulation that ultimately spills into the vertical section of the well. A case history showing excessive surface pressures encountered during underbalanced drilling for gas is presented as an example of this effect. Introduction Effective well control practices for underbalanced, horizontal drilling are becoming increasingly important but are not as well developed as for conventional, overbalanced drilling (Bourgoyne, 1997). Underbalanced drilling is becoming more common in many areas of the world. It can provide a number of potentially important advantages versus conventional overbalanced drilling. These include increased well productivity, faster penetration rates, reduced mud losses from lost circulation, minimized differential sticking, and an indication of well productivity while drilling (Smith, 1999). Horizontal drilling is another technology that is becoming more common around the world. An average of almost 1000 horizontal wells per year were drilled in the U.S. over the past five years (API, 1998, 1999, 2000, 2001 and 2002). In many reservoirs, a horizontal well will provide a longer, higher productivity wellbore than would be possible with a vertical well. Higher production rates due to larger drainage area, less water and gas coning, and the ability to produce from multiple vertical fractures are some benefits of horizontal wells. The combination of these two technologies, horizontal wells and underbalanced drilling, has been particularly successful in some applications. For example, many Austin Chalk prospects, which were uneconomic when using conventional drilling practices, have been commercial successes when developed using underbalanced, horizontal drilling. As a result of its unique advantages, underbalanced, horizontal drilling is being applied to deeper, higher pressure, oil and gas reservoirs (Joseph, 1995) and in some offshore situations (Nessa et al., 1997 and Nakagawa, 1999). The practice of drilling underbalanced and taking a continuous flow, i.e. kick, from the formation has been very successful in low productivity, moderate pressure, oil reservoirs such as the Austin Chalk in Texas. This trend continues to be important with almost one third of all horizontal wells drilled in the U.S. in 2001 in the Austin Chalk, but the wells being drilled to this formation have stretched far beyond South Texas (API, 2002) through Louisiana to eastern end of the

Abstract Horizontal wells often include sections with angles greater than 90 degrees due to steering uncertainties or the well path following the reservoir updip. If gas is present inside the wellbore either because of underbalanced drilling or taking a kick, the gas can accumulate in the resultant high spots and complicate effective well control. A series of experiments were conducted to understand accumulation and removal of gas in a flow tube simulating a fully eccentric well bore annulus at inclinations of 91.5 to 100 degrees. Water, or polymer-viscosified water, and air were circulated in a 6 inch by 2.375 inch annulus at a range of liquid velocities representative of both well control and routine drilling operations. The studies demonstrated that low flow rates favor counter-current flow in either slug or elongated bubble flow patterns that would cause accumulation at high spots. As the liquid flow rate increases, co-current flow develops, and ultimately the counter-current flow disappears. The co-current flow patterns observed were stratified, stratified wavy, and bubble. Flow maps were made to show co-current and countercurrent flow windows and void fractions as a function of superficial gas and liquid velocities, fluid type, and inclination angle. Some counter-current flow was found at superficial liquid velocities below 1.2 ft/sec almost regardless of other conditions. Non-Newtonian fluids with a higher yield point were found to sometimes encourage counter-current flow and accumulation in contrast to conclusions published by other authors. Gas was effectively removed at inclination angles up to 96 degrees by a superficial liquid velocity of at least 3.3 ft/sec. Failure to continuously remove gas during underbal-

www.petroman.ir

IADC/SPE 87185

Gulf of Mexico coast. In addition, the number of horizontal wells drilled in other locations, including West and Northeast Texas, California, the Rockies, Oklahoma, and North Dakota, has become substantial. Adapting underbalanced, horizontal drilling to reservoirs that have higher productivity, higher formation pressure, and/or gas as the dominant reservoir fluid has proven more challenging (Joseph, 1995). Meeting that challenge has also become more important as gas wells are making up a larger percentage of the horizontal wells drilled. The percentage of horizontal wells drilled to develop gas reservoirs has increased from 26% during 1997 to 42% in 2001 (API, 1998 & 2002). Higher pressure rotating heads and high capacity, mud-gas separation equipment have been applied to address this challenge. Nevertheless, maintaining surface pressures and flow rates within the limits of this equipment has required more careful control of formation influx rates than in less critical applications. Wells planned as horizontal usually deviate from being perfectly horizontal due to BHA response, the need for the well to follow bed dip, or simply by collapse of the hole. Horizontal gas wells can present special complications particularly when sections of the well deviate above horizontal. While well is underbalanced and gas is being produced into the wellbore, substantial volumes of gas can accumulate in elevated sections of the horizontal, as shown schematically in Figure 1. This gas can be difficult to remove, as it will tend to migrate away from the vertical portion of the wellbore to the highest spot in the horizontal. Examples of this kind of accumulation have been encountered in horizontal gas wells in the North Sea. Special considerations for removing the accumulated gas were required each time when drilling resumed after a trip. Those considerations included careful well monitoring with the trip tank while tripping in the hole and circulating the horizontal section in stages with the annular preventer closed and returns through the choke. The goal was to remove all of the gas in the section being circulated before continuing to trip further or drilling additional hole. If the gas inside the wellbore cannot be circulated out, the accumulation can continue until the gas fill the available space in the horizontal and spills out into the vertical section of the wellbore. In case of an open wellbore, hydrostatic pressure in the wellbore decreases and results in more influx. The resulting flow rate at the surface may exceed the equipment capacity. If the well is shut-in, high surface pressures that may exceed the pressure rating of the surface equipment may result. In any event, the accumulation of large volumes of gas in a long horizontal offers the possibility of bringing large gas kick volumes and the resulting pressures to the surface. One field well control experience on a horizontal well had high surface pressures (a sevenfold increase) that built rapidly after shut-in was discussed by Baca et al. (1999) and may have resulted from this effect.

The liquid velocity due to mud circulation that would continuously remove a gas influx from the horizontal and prevent accumulation has not been known. Rommetveit et al (1995) performed a series of full scale tests in a simulated annulus at 4 degrees above vertical. He has studied the conditions that removed the trapped fluid. Lage et al (2000) reported on a similar series of tests in a near full scale apparatus. They studied void fractions, flow regimes, and pressure losses in multiphase flows in horizontal and near horizontal annuli but forced all gas to flow co-current. Therefore no conclusions could be drawn about accumulation or actual removal efficiency. Likewise, the liquid velocity that would flush a preexisting accumulation from the horizontal has also been relatively unknown. Rommetveit et al (1995) concluded from their full scale tests that a superficial liquid velocity of .9 m/sec would be required to remove an accumulation. They also concluded that use of more viscous fluids would reduce this limit. Understanding the flow mechanisms and variables, which control these velocities, is one of the objects of ongoing research. This understanding would allow prediction of these velocities for successful control of gas accumulation and avoidance of well control hazards in the field. Some preliminary work addressing only the conditions of gas removal from a typical circulating horizontal well has been conducted by Aas et al (1998). Studies of counter-current gas flow are scarce. Johnson (1998a and b) has examined this type of flow in circular pipes both experimentally and theoretically, covering a wide range of several parameters. He investigated liquid-gas and liquid-liquid combinations, several pipe diameters and two dip-angles over a broad range of flow rates. Johnson (1998) introduced the gas on the downstream end of the downward inclined pipe in his effort to simulate the conditions of a slugcatcher. Consequently, his work does not address flow entering along the flow path as for horizontal drilling. Baca et al (1999, 2000) carried out experiments in a similar full-scale test section as that of Johnson (1998), but with an eccentrically annular cross section and gas injection in the middle of the test section. The inner pipe in this annular configuration simulates the drill string. Thus it is possible to establish counter-current gas flow in one part of the test section and co-current gas flow in the other, as is more relevant to the application of interest. The experiments of Baca et al. (1999, 2000) were at low dip-angles (1.5o, 3o, 6o and 10oabove horizontal, equivalent to 91.5o, 93o, 96o, and 100o inclinations) and relatively low superficial liquid and gas velocities. Initially, these experiments (Baca, 1999) used water and air as the working fluids and therefore, the non-Newtonian nature of actual drilling muds was not taken into account. Subsequently, experiments were conducted using two non-Newtonian waterpolymer-air mixtures (Baca, 2000) more closely resembling drilling fluids. Baca et al (1999, 2000) demonstrated that, even at extremely low dip angles, a considerable operational window exists over which counter-current gas flow can be sustained within the lower end of a range of practical liquid flowrates, and that formation of countercurrent flow is favored by low gas injection rates.

www.petroman.ir

IADC/SPE 87185

The present study (Ustun, 2000) extends the work of Baca et al (1999, 2000) to higher liquid and gas superficial velocities. These conditions were studied to more thoroughly investigate and define co-current versus counter-current flow boundaries. The new experiments were conducted over approximately the same range of dip-angles and liquid rheologies as Bacas experiments. Experimental Facility And Instrumentation A two-phase flow facility has been designed and built for the study of counter-current and co-current, liquid-gas flows in inclined eccentric annular pipes. The experiments have been conducted with air as the working gas and three different working liquid rheologies: water and two different waterpolymer mixtures. The water-polymer mixtures were used in order to simulate the non-Newtonian character of the drilling muds. The first water-polymer mixture, refered to hereafter as Thin Polymer, had a density of 971 kg/m3 (8.1 ppg), a plastic viscosity of 3.54 cpoise and a yield point of 4 Pa (8.23 lbf/100ft2). The second water-polymer mixture, referred to hereafter as Thick Polymer, had a density of 971 kg/m3 (8.1 ppg), a plastic viscosity of 4.5 cpoise and a yield point of 7 Pa (14 lbf/100ft2). Thus the thick polymer solution had almost double the yield point of the thin one. Of the 355 total runs, 208 runs were performed with water, 96 were performed with thin polymer, and the remaining 51 runs used thick polymer as the liquid phase. A schematic of the test section of the facility is shown in Figure 1, in reference to the horizontal well section it is designed to simulate. The test section is 14 m (46 ft) long and is made out of transparent 6.031 ID PVC pipe, in order to facilitate flow observations. The drill-string is simulated by a 2"nominal PVC pipe laid inside the 6" test section. Thus the area available to the flow is an eccentric annulus. The flow loop has been revised since the experiments by Baca so that liquid flows of up to 49 Lit/s (780 gal/min) can be generated and liquid superficial velocities up to 3.16 m/s (10.4 ft/sec) can be achieved. In practice, experiments were conducted at superficial liquid velocities up to 2.5 m/sec (8.17 ft/sec). This represents the upper limit of liquid superficial velocity that is expected during drilling operations. A paddle-wheel flow meter with a 0.5% of full-scale accuracy is used to measure the liquid flow rate. Using high flow rates created some unexpected drawbacks. It was found that the 20 bbl. (840 gal) liquid storage tank was not large enough to provide an adequate residence time for the release of air bubbles trapped inside the non-Newtonian fluids. Consequently, mud weights less than the density of fresh water were observed due to aeration of the polymer viscosified fluids. A deflector plate was placed at the discharge outlet from the test apparatus, which helped to lessen the air entrainment in the fluid. At this stage, foaming of the mud was observed and aluminum stearate was added to the polymer. However the change in the liquid properties due to foaming and air entrainment required that the thick polymer runs be limited to just those required to define a boundary, particularly the co-current flow only boundary.

Gas was supplied to the test section from the inlet placed at the middle of the test section (Figure 1). The maximum gas injection rate is 28,000 Slit/hr (1000 scfh). Gas injection rate is measured by a rotameter accurate to +/-198 Slit/hr (7 scfh). A variety of interchangeable injectors can be used to control gas bubble size and injection conditions. A gas separation tank is used at the upstream end of the test section to vent countercurrent gas which is also measured with one of two rotameters accurate to +/-113 SLit/hr (4 scfh) and +/-425 SLit/hr (15scfh) respectively. Two rotameters are needed as flow rates vary widely. The liquid level in the separator is monitored through a sight glass to within less than 0.5 inch, as is the pressure in the separator, to within 1 psi. During most of the experiments, at steady state, the level of water in the separation tank is held constant by appropriately setting a choke valve at the separator vent exit, and the separator pressure is fixed. Under this condition, the flow rate measured out of the separator is accepted uncorrected. At high liquid flow rates, it is very difficult to keep the separator level constant, so the choke valve is set to maintain constant pressure in the separator while the water level inside it is rising. The rate of the water level rising in the separator is measured using a stopwatch. The rate of gas volume displacement by the water in the separator is then calculated given the separator geometry. The rotameter flow rate is corrected by this displacement rate after adjusting it for the separator pressure. The accuracy of the counter-current gas flow rate measurement has been tested during runs where only counter-current gas flow is observed in the test-section. Under these conditions, the gas flow rate out of the separator agrees with the gas injection rate to within the meter accuracy. When both co-current and counter-current gas flow exists, the cocurrent gas flow rate is taken to be the difference between the gas injection and counter-current gas flow rates. The test section can be isolated from the rest of the system via simultaneously actuated, quick closing valves. This system is used to trap the two-phase mixture in the test section when a flow pattern is developed and to determine the average holdup by measuring the volume of water trapped. The estimates have an uncertainty of approximately 10% on average. When both co-current and counter-current gas flows are present in the test section, a measurement of the void fraction in the co-current or the counter-current section is needed. Counter-current flow was observed in slug, elongated bubble and stratified flow patterns where as in the previous studies (Baca 1999, 2000) slug flow was the only flow pattern seen. Stratified flow, stratified wavy flow, and at high liquid and low gas velocities, bubble flow patterns were the flow patterns observed in co-current flow situations. Some influence of end effects was also observed during the experiments. The flow tube length is only about 20 ft in each direction on either side of the gas injection point. Therefore, it was not possible in some cases to conclusively define the flow pattern. Under some conditions, a bubble was observed to grow in the counter-current flow direction, and before completing its formation, the front end reached the separator outlet and flow turned into stratified flow. Also at the downstream end of the apparatus, the upward path of the hose

www.petroman.ir

IADC/SPE 87185

connecting the flow tube and the discharge outlet was another possible constraint that might have affected the flow patterns. A video record of both the co-current and counter-current regions was taken during the experiments. The void fraction in the counter-current section is estimated based on either the average observed gas-plug velocity or the gas-liquid contact surface of stratified flow, as appropriate for the flow pattern. Gas plug velocity was measured from the video record by timing the progress of gas-plugs between two fixed points along the test section. For slug and elongated bubble flow, the counter-current gas superficial velocity divided by the gas plug velocity (assumed as the average gas phase velocity) provides an estimate of the void fraction in the counter-current section. For stratified flow, the average position of the contact surface enables calculation of the flow area inside the pipe. The void fraction in the co-current section is estimated in essentially the same way. The fraction of a fixed test section volume occupied by gas is estimated from the video record and divided by the eccentric annular total volume. Since, when both flow patterns were seen, the co-current flow is in stratified flow and the geometry of the pipe is known, making this measurement is fairly straightforward. The known outside diameter of the pipe is used for image calibration purposes. In this case where the flow in the annulus is stratified, optical distortion effects on the measurement are negligible. At the high end of the flow matrix, a bubble flow pattern is seen, and it is assumed that both phases are moving at the mixture velocity. Pressure and temperature are measured at both ends of the test section. The average of the two readings is taken as the nominal test-section pressure and temperature. The superficial gas velocities are all calculated at the test-section nominal pressure and temperature. Pressure and temperature are also measured at all the metering locations and used to correct the rotameter readings. The pressure gages used are accurate to 0.6 psig and the temperature probes are accurate to 2o F. Results and Discussion Topological Observations Detailed topological observations of the counter-current, and co-current liquid-gas flows using water (Newtonian) and water-polymer mixtures (non-Newtonian) working liquids at a wide range of superficial velocities (up to 8.17 ft/sec), different dip angles (1.5o, 3o, 6o and 10o) and at various gas (air) flow rates. These observations were recorded in detail by Ustun (2000). In all of the presently examined cases, as in Baca et al. (1999, 2000), air was injected in the middle of the test section length as indicated in Figure 1. It has been observed during these experiments that, for all of the working liquids, counter-current flow was seen at low gas and liquid superficial velocities as shown in Figure 2. For a given liquid flow rate, the counter-current gas flow changes from slug flow to elongated bubble flow and then to stratified flow as the gas flow rate increases. As gas and liquid superficial velocities increase, both co-current and counter- current

flow is observed for all of the four dip angles and three rheologies tested. Finally at high liquid superficial velocities and injected gas flow rates, counter-current flow disappears and only co-current flow is observed as stratified or stratified wavy flow. At the high limit of tested liquid flow rates and low gas rate, the flow pattern becomes bubble flow. These transitions and flow pattern changes are expected, since counter-current flow is viable only if the buoyancy of gas bubbles or plugs overpowers the drag from the liquid. These observations hold for both Newtonian and non-Newtonian rheologies examined. It should be noted that the size of the gas plugs in countercurrent flow varies, depending on the liquid superficial velocity and the injection gas flow rates. These gas plugs were generally longer in the cases where the liquid was nonNewtonian. This is attributed to increased drag because of the yield point and higher plastic viscosity of the water-polymer mixtures. The size of the gas plugs also indirectly depends on the test-section pressure, which varies as the liquid and gas flow rates are changed. Flow Maps Based on the visual observations and the measured gas (Vsg) and liquid (Vsl) superficial velocities, flow maps have been constructed. Since experimental data was mostly on the co-current flow area, the results of Baca (1999, 2000) were also utilized for the completeness of the flow maps. This enabled defining a complete flow map. Three different flow patterns are seen on each flow map: Counter-current gas flow only area, transition zone involving both counter-current and co-current gas flow, and co-current gas flow only area. Each flow direction is defined by two lines for each of the four different angles at which the experiments were run (Figure 2) or for each of the three working liquids used (Figure 3). The boundary lines towards the lower lefthand corner of the figure define the flow boundary between counter-current flow and flow in both directions. The lines on the upper right hand corner represent the boundary between the flow area where flow is in both directions and that where only co-current flow occurs. The derived flow maps are shown in Figure 2 for the three liquid rheologies examined: Water (Figure 2-a), Thin-Polymer (Figure 2-b) and Thick-Polymer (Figure 2-c). The two boundaries for a given angle are given in the same color with different colors used for each angle. Figure 3 is plotted in a similar manner showing the flow direction boundaries depending on the dip angles. There is a separate flow map for each of the four different working angles showing all results from the three liquids used on each map. It was seen that low superficial liquid rates promotes counter current gas flow. For water, at low liquid superficial velocities and high gas injection rates, the boundary of fully counter-current flow generally moves with a negative slope until a certain Vsl value at which the boundary starts to move upward (Figure 2a). Below a thereshold Vsl value of about 1 ft/sec, all the gas moves countercurrent independent of the

www.petroman.ir

IADC/SPE 87185

liquid velocity or dip angle, for the angles studied. Unfortunately, the limited capacity of the apparatus to handle countercurrent gas flow prevented investigation of the response of the boundary to higher gas rates. Therefore the validity of this threshold Vsl value could not be determined for higher superficial gas velocities. A similar trend in the counter-current flow boundary is seen in Figures 2b for thin and 2c for thick polymers. However, it was observed that fully counter-current flow could be prevented at a lower superficial liquid velocity flow rate than the threshold value for Newtonian fluids. The general trends for flow direction behavior are similar for all fluids and angles. For low Vsl values, as gas flow rate increases, both co-current and counter-current flow directions start to coexist in the wellbore (Figure 2). The flow of the liquid limits the amount of the counter-current gas flow in the wellbore and pushes part of the gas out of the wellbore with the liquid flow. For a given liquid flow rate, flow may turn from complete counter-current flow to bi-directional flow and finally to complete co-current flow as gas flow rate increases. These three distinct behaviors can be seen at 1.5 and 3 degrees for all three fluid types (Figures 3-a and 3-b). At higher angles, this transitional behavior of bi-directional flow essentially disappears at high liquid superficial velocities (Figures 3-c and 3-d). Nevertheless, the bi-directional flow can be seen at lower superficial liquid velocities at all angles with all fluids. This transition with increasing gas flow rate can be explained by forces exerted on the gas bubbles. As gas flow rate increases, size of the counter-current moving gas bubbles increases. With increasing bubble volume, the force exerted on the bubble by the liquid also increases and at some point, the drag force due on the bubble overcomes the buoyancy force and counter-current flow stops. It can be seen from the figures that, the liquid rheology has a large effect on where this boundary of fully co-current flow occurs, but that the effect of the liquid yield point is not easily predicted. The flow maps derived for the two non-Newtonian liquids display considerable differences relative to that for water and to each other (Figure 2). When the thin polymer results are compared with water, both show similar responses to angle and liquid flow rate changes at moderate gas flow rates. Both fluids experience bi-directional flow over a much broader area than the thick polymer does. However at lower Vsg values, counter-current flow with the thin polymer occurs at higher Vsl values than with water, especially at higher angles, and the boundary where no counter-current flow occurs was not reached.at the 10 degree angle as seen in Figure 2-b. This behavior is more similar to thick polymer than to water. For all three working liquids, as dip angle increases, the counter-current flow area shifts to higher gas and liquid velocities which is expected due to increasing buoyancy effect (Figures 2 and 3). In this case, water has a distinctive advantage over non-Newtonian fluids for gas removal at superficial gas velocities below 0.6 ft/sec. This effect is evident seeing that a completely co-current flow boundary can be achieved for all values of Vsg with water as seen in Figure 2-a.

In contrast, there is a critical value for gas rate below which some degree of counter-current gas flow will be present for non-Newtonian fluids at most dip angles. Evidently, fluid properties become more important than the dip angle. Similar trends in the fully co-current boundary with both polymers contrast in behavior to water suggesting that this behavior might be related with the yield points of the liquids. For a smaller amount of gas flow, which causes a smaller interface, water apparently is more effective in overcoming the buoyancy forces than the more viscous, non-Newtonian fluids. Outside this range of low superficial gas velocities, the fully co-current boundaries for the thick polymer are very different than for the other two fluids. The boundary occurs at much lower values of Vsl than for the thinner fluids. However, the thick polymer data covers only a limited range of gas flow rates, and the specific boundary location for higher gas velocities is not known. Nevertheless for a given gas flow rate, complete co-current flow can be obtained at lower Vsl values. Consequently, the map area where bi-directional flow occurs is larger for water and the thin polymer than for the thick polymer. For both thin liquids, this area increases with increasing dip angle, and complete co-current flow is seen at higher Vsl values at the higher angles. For the thick polymer, the bi-directional flow area also increases with increasing dip angle but to a much smaller degree. The shape of the boundary for purely co-current gas flow typically has two slopes for the thinner fluids. Therefore the boundary line can be separated into two different sections depending on the Vsg value. The first section applies to lower values of Vsg. Below a Vsg value of about 0.6 ft/sec, water requires a lower Vsl value to achieve fully co-current flow. For low Vsg and low angles, 1.5 and 3 degrees, the thick polymer provides fully co-current flow at slightly lower Vsl than thin polymer (Figures 3-a and 3-b). At 6 and 10 degrees however, the thin polymer boundary occurs at lower Vsl values (Figures 3-c and 3-d) for low Vsg. Apparently the yield value of the liquids has only a small effect on low velocity gas flow. When the Vsg is greater than the critical rate of about 0.6 ft/sec, the thick polymer typically achieves fully co-current flow at a significantly lower superficial liquid velocity than the thinner fluids. At 6 degrees, the co-current flow boundaries for water and thin polymer become stable, at Vsl values of about 2.0 and 2.4 ft/sec respectively, that are independent of the gas velocity. At 10 degrees. the boundary for water moves to a higher Vsl value of about 2.4 ft/sec. Therefore for underbalanced operation giving a superficial gas velocity greater than about 0.6 ft/sec, a thick non-Newtonian fluild will insure co-current gas flow at a lower pump rate than water or a fluid with a low yield point. When co-current boundaries for thin polymer and water are compared for gas velocities greater than the critical value, co-current flow is achieved at the same or lower liquid velocities with water. This result indicates that, although high yield points favor the co-current flow over a certain Vsg value, there is a threshold yield value below which a non-Newtonian fluid with a yield point has no consistent benefit versus a Newtonian fluid.

www.petroman.ir

IADC/SPE 87185

One important conclusion that can be reached from the comparisons between the maps from the three different rheologies of Figure 2 is that, in general, the operational window over which counter-current flow of low gas influx rates can be sustained is expanded as the yield point of the liquid is increased. Specifically, thicker muds favor counter-current flow development for low gas flow rates. Consequently slow gas kicks or low gas production rates during underbalanced drilling are favorable to counter-current gas motion in a viscous mud which leads to gas accumulation at the higher elevations of the "horizontal" well. Conversely, a high yield point mud can achieve fully co-current gas flow at superficial gas velocities above about 0.6 ft/sec at much lower circulating rates than with water or a low yield point mud. It is also important that a relatively high superficial liquid velocity with water can insure fully co-current gas flow, and removal of all gas from the well, regardless of the gas flow rate, which is not necessarily possible with a mud. Horizontal Well Case History Drilling data from a relatively deep, horizontal gas well was used to compare a real operating experience with the flow direction behavior seen in this research. This well had a true vertical depth of approximately 16,000 ft and essentially all operations were conducted with the well underbalanced. The typical casing pressure during routine drilling and circulation was about 1000 psig. This well is relevant for consideration because it encountered excessive surface pressures due to gas in the well annulus on several occasions. The 3000 psi static pressure rating of the rotating head was exceeded in at least four separate instances over a period of about two weeks. The kicks causing excessive surface pressure became more frequent and more severe as the horizontal section was lengthened. The first large kick that was experienced in this well happened after about nine days of relatively routine drilling and tripping operations in the horizontal well section. It occurred during a partial trip out of the hole. Eventually, the well was shut in on the pipe rams, and the well was bullheaded until the casing pressure was reduced to about 1000 psi. Circulation resumed, and the drillstring was tripped back to TD. Drilling resumed with casing pressure averaging 700 psig and was relatively routine until large increases in flow out began to be experienced after about 12 hours of drilling. Drilling continued with a gradual increase in casing pressure and a significant increase in flow out for another 10 hours. At this point, drilling and circulation were discontinued. This kick was ultimately controlled after a maximum casing pressure of 3200 psig was reached. Two more kicks with maximum casing pressures of 4000 and 7000 psi were experienced over the next four days. The available data does not allow rigorous analysis of the cause of these kicks. Operators in the area typically attribute these increases in gas flow to encountering more highly pressured or higher conductivity fractures. Given the increasing frequency and magnitude of the kicks, it also seems plausible

that the alternative explanation of increasing volumes of gas accumulation in the well contributed to the magnitude of the kicks. The second kick apparently developed over several hours after having killed the previous kick by bullheading and then conducting seemingly steady state operations for a half day. It could certainly have resulted from the accumulation of gas in the horizontal until reaching a spill point and then gradually increasing the as envisioned herein. Concluding Remarks Results from a series of experiments conducted in downwards liquid-gas flows in inclined, eccentric annular pipes, with plain water as well as non-Newtonian, polymer/water mixtures and air as the working fluids, have been presented herein. Flow behavior was studied using higher gas and liquid flow rates than in previous work. Air was used as the gas phase and was injected into the system in the middle of the test section length. The operating window, in terms of liquid and gas superficial velocities, within which counter-current gas flow occurs at four dip angles has been determined experimentally for all three rheologies. Pure counter-current flow was seen at every angle and rheology, but this study focused on high velocities that encouraged co-current flows. The coexistence of counter-current and co-current flow was observed at intermediate working velocities. Pure co-current flow was the dominant flow pattern seen in these experiments as expected because of high liquid superficial velocities and high gasinjection rates. The counter-current flows generated were in the slug/plug, elongated bubble and stratified wavy flow patterns, while the co-current flows appeared as stratified, stratified wavy and bubble flow patterns. Our results indicate that generation of co-current flow depends on the working fluid, working angle, and the velocities of gas and liquid phases. Nevertheless, complete counter-current flow is certain for superficial liquid velocities below 1.2 ft/sec and gas superficial velocities below 0.5 ft/sec for the range of fluid rheologies and hole angles studied (Figure 3). Likewise, co-current flow was observed for all fluids and hole angles at liquid superficial velocities over 2.5 ft/sec and superficial gas velocities greater than 0.6 ft/sec. Between these values the flow direction is highly dependent on liquid properties, angles and specific velocities, and flow may exist in both directions. Water gave the most predictable control of gas flow direction. Fully counter-current gas flow is expected only at very low superficial liquid velocities, on the order of 1 ft/sec or less. Fully co-current flow was achieved with water at all angles at superficial velocities of 2.5 ft/sec and higher (Figure 2-a). Between these limits, bi-directional gas flow may occur with some counter-current gas flow almost certain at superficial liquid velocities less than about 1.5 ft/sec. Non-Newtonian fluids gave much less regular control of flow direction. These fluids tended to encourage countercurrent flow at low superficial gas velocities, 0.6 ft/sec or less, and encourage co-current flow at higher superficial gas velocities. If complete removal of gas during underbalanced

www.petroman.ir

IADC/SPE 87185

drilling is necessary at low liquid flow rates, a high yield point fluid is most likely to be successful as shown in Figures 3-a-d. It appears that the relationship between superficial liquid velocity to achieve fully co-current flow and fluid rheology can be divided into two regions based on gas superficial velocity. This result appears to support both Baca et.al. (1999, 2000) indicating that Newtonian fluid behavior is more effective for removing a gas accumulation, at superficial gas velocities less than 0.6 ft/sec, and Rommetveit et als (1995) conclusion that increased liquid viscosity decreases the liquid flow rate required, which is true at higher superficial gas velocities. If gas is believed to have accumulated in the wellbore, a removal process can be defined by using the flow maps and choosing the liquid properties. A general solution is using a superficial liquid velocity of at least 3.5 ft/sec which yields only co-current flow for all fluids at hole angles between 90 and 96 degrees. However if removal has not been continuous, rapid displacement of a large accumulation of gas into the vertical portion of the wellbore could cause dangerously excessive surface pressures. Therefore the operating windows defined by this work are relevant for both the prevention of gas accumulation and the removal of gas that does accumulate. Further research is necessary to determine the more accurate operational boundaries for flow direction and what the effect of the gas entrance mechanism into the wellbore is on these boundaries. Nomenclature cc= Counter-Current Flow coc = Co-Current Flow Vsl = Liquid Superficial Velocity (ft/sec) Vsg = Gas Superficial Velocity (ft/sec) Acknowledgments The support of the U.S. Minerals Management Service for initiating this research and Campanile Charities for continuing it is gratefully acknowledged. The modifications to allow use of higher superficial liquid velocities were designed by Fernando Flores and implemented with the assistance of Ernie Loftice and the staff of student workers at the LSU Petroleum Engineering Research and Technology Laboratory.

References
1. Aas, B., Bach, G., Hauge H. C., and Stern N., Experimental Modeling of Gas Kicks in Horizontal Wells, SPE/IADC 25709. 2. Joint Association Survey on 1997 Drilling Costs, API, 11/98. 3. Joint Association Survey on 1998 Drilling Costs, API, 11/99. 4. 1999 Joint Association Survey on Drilling Costs, API, 11/2000. 5. 2000 Joint Association Survey on Drilling Costs, API, 11/2001. 6. 2001 Joint Association Survey on Drilling Costs, API, 12/2002. 7. Baca, H., Nikitopoulos, D. E., Smith, J. R., and Bourgoyne, A. T., Co-current and Counter Current Migration of Gas Kicks in Horizontal Wells, ASME Journal of Energy Resources Technology, Vol. 121, pp. 96-101, June 1999. 8. Baca, H., Nikitopoulos, D. E., Smith, J. R., and Bourgoyne, A. T., Counter-Current and Co-current Gas Kicks in Horizontal Wells: Non-Newtonian Rheology Effects, ETCE2000/DRILL10117, ASME, February 14-17, 2000, New Orleans, Louisiana. 9. Bourgoyne, Jr., Adam T.: Well Control Considerations for Underbalanced Drilling, SPE #38584, 97 SPE ATCE, October 5-8, 1997, San Antonio, Texas. 10. Johnson, A. J., An Investigation Into Stratified Co- and Counter-current Two-Phase Flows, SPE Production Engineering, August 1998, pp. 393-399. 11. Johnson, A. J., Controlling Effects in Counter-current TwoPhase Flow, SPE Production Engineering, August 1998, pp. 400-404. 12. Joseph, Robert A., Underbalanced Horizontal Drilling in an Abnormally-Pressured Environment, #PS.2DT10.001, Petrotech-95. 13. Lage, A. C. V. M., Rommetveit, R., and Time, R. W., An Experimental and Theoretical Study of Two-Phase Flow in horizontal or Slightly Deviated Fully Eccentric Annuli, SPE/IADC Asia Pacific Drillling Technology, IADC/SPE 62793, Kuala Lumpur, Malaysia, September 11-13, 2000. 14. Nakagawa, Edson Y. et al, Application of aerated-fluid drilling in deep water, World Oil, June, 1999 15. Nessa, D. O. et al, Offshore underbalanced drilling system could revive field developments, World Oil, Part 1 July 1997, Part 2 October 1997. 16. Rommetveit, R. et al, Full Scale Kick Experiments in Horizontal Wells, SPE ATCE, SPE #30525, Dallas,TX, October 22-25, 1995. 17. Smith, J. R., The 1999 LSU/MMS Well Control Workshop: An overview, World Oil, June 1999 18. Ustun, Firat, The Effect of High Liquid Fow Rates on CoCurrent and Counter-Current Gas Kick Migration in High Angle Wells, M.S. Thesis, Louisiana State University, December, 2000, Baton Rouge, Louisiana.

www.petroman.ir

IADC/SPE 87185

Figure 1. Schematic depiction of horizontal well, high angle section of horizontal well where counter-current gas flow may occur, and experimental facility for the simulation of counter-current flow. c) a)

b)

Figure 2: Measured co-current and counter-current flow-map

boundaries for gas-liquid flows in "horizontal" wells at various dip-angles and for three different liquid rheologies. (a) Water/Air, (b) Thin-Polymer/Air, (c) Thick-Polymer/Air.

www.petroman.ir

IADC/SPE 87185

a)

c)

b)

d)

Figure 3: Measured co-current and counter-current flow-map boundaries for gas-liquid flows in "horizontal" wells for three different liquid rheologies at various angles. (a) 1.5 degrees (b) 3 degrees, (c) 6 degrees (d) 10 degrees.

www.petroman.ir

Vous aimerez peut-être aussi