SPE (1) Copyright 1999, Society of Petroleum Engineers Inc.

Transcription

1 SPE 5565 Enhancing Liquid Lift From Low Pressure Gas Reservoirs Hiro Yamamoto, SPE, Mobil Exploration and Production, and Richard L. Christiansen, SPE, Colorado School of Mines Copyright 999, Society of Petroleum Engineers Inc. This paper was prepared for presentation at the 999 SPE Rocky Mountain Regional Meeting held in Gillette, Wyoming, 5 8 May 999. This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the 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 Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society of Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. Box 886, Richardson, TX , U.S.A., fax Abstract Liquid loading in low production gas wells is a nuisance for production engineers in the natural gas industry. It is essential to maintain gas wells free of liquid; otherwise, the production will be severely reduced by backpressure of the accumulated liquids, and by reduced gas relative permeability in the surrounding formation. The most fundamental solution for the liquid loading problem is to select tubing diameter for the well such that the natural energy in the reservoir will give a gas velocity sufficient to lift liquids from the sand face of the reservoir to the surface. Unfortunately, the optimum diameter varies for different periods in the life of a well. Here, a new approach to the liquid loading problem is reported. By introducing restrictions, such as orifices, inside the tubing to alter flow mechanisms, liquid may be lifted by gas flow rates below the conventionally accepted critical rate. Extensive laboratory experiments with a -foot-tall flow loop with.5-inch-inside-diameter tubing were conducted to test the effect of five different designs of restrictions. In these experiments, the effects of gas flow rate, of restriction design, of flow loop inclination, and of design of the outlet at the top of the loop were considered. The experiments proved that the restrictions alter two-phase flow behavior and improve liquid lifting rate in the flow loop. Introduction Natural gas that had been flared or re-injected for pressure maintenance and artificial lift in the U. S. and elsewhere has become commercially attractive since the 95s due to the strong domestic and international demand. Because natural gas does not cost as much as oil per unit of energy content, and because its combustion produces less harmful emissions than oil and coal, it is increasingly desired as an alternative fuel for automobiles and electric power plants, as well as for residential heating. When natural gas is produced from a reservoir, species such as water and intermediate to heavy hydrocarbons can condense as liquids in the well bore, depending on the composition of gas produced at the bottom of the well. Condensation is not the only source of liquid in the well bore; free brine or free hydrocarbon liquid can be produced directly from the reservoir. As long as gas flow rate is sufficiently high to maintain annular mist flow, these liquids are lifted from the well. As flow rate declines in a maturing gas field, the flow regime switches from annular mist flow to churning flow, and the liquid lifting capacity of the flowing gas decreases dramatically. The flow rate for this switch in flow regimes is called the critical flow rate. When the liquid cannot be lifted to the surface, it accumulates in the well to create undesired backpressure on the producing formation, restricting gas production rate. If a liquid-lifting method, such as soap sticks, plungers, rod pump, or swabbing is not implemented, the production rate will continue to decline toward zero. Duggan and Turner, Hubbard, and Duckler investigated the critical flow rate in the 96s. Duggan reported on field experience with wells producing condensate; Turner et al. reported on theoretical models and field tests. Turner et al. developed an expression for estimating the critical gas flow rate for lifting liquids from a well: v c ( ρ ρ ) σ L G =.9 ρ G / (Some of the equations for critical velocity and critical flow rate in Turner et al. contain incorrect multiplicative constants. The corresponding equations in Coleman et al. are error free. Equation above is copied from Eq. of Coleman et al.) Equation includes the % upward adjustment to match field data suggested by Turner et al. The THD expression is widely referenced for estimating critical flow rate for continuous, 5, 6, 7, 8, 9, liquid removal. Coleman et al. reinvestigated the THD critical flow rate expression for gas wells that had lower wellhead pressures than the wells in the study of Turner et al. Coleman et al. concluded that the % upward adjustment was not necessary. ()

2 H. YAMAMOTO AND R. L. CHRISTIANSEN SPE 5565 Then, the following expression for critical velocity would apply: v c ( ρ ρ ) σ L G =.59 ρ G / They also concluded that the critical flow rate is insensitive to the gas-liquid ratio below about Bbl/MMcf. About 96% of gas wells in Colorado produce less than Mscf/day. According to the THD method of analysis, all of these wells are susceptible to the ill effects of liquid holdup. Those gas wells contribute 5% of total gas production in the State. These percentages for Colorado probably apply across the U.S. for onshore gas production. There are many alternative solutions to the liquid loading problem. For low-pressure gas reservoirs, the capital investment for the solutions must be small. For example, sucker rod pumps are generally too costly to install for lowpressure gas reservoirs, unless they are very shallow. Of all the solutions, soap sticks require the least capital investment. Plunger lift is a technology widely used in the Denver- Julesberg Basin. Installation costs for automated plunger lift are between $ and $5. Coiled tubing velocity strings have been used to solve some liquid loading problems. Installation costs, labor and capital, of a coiled tubing can range from $, to $,, depending on reservoir depth, condition (new or used) of the coiled tubing, and local labor costs. Each of these technologies has its niche, depending on a collection of factors including composition of reservoir fluid, operating pressures, and of course economics. This paper documents an investigation of a new technology for solving the liquid loading problem in gas wells. The concept of this technology is to introduce baffles, or other flow altering inserts, into the tubing that will enhance the transport of liquids by the flowing stream of gas. This new approach was tested with a flow loop as described below. The results of a portion of the tests are also given below. For more details of the investigation, see Yamamoto. Features and Operation of the Flow Loop A diagram of the flow loop is shown in Figure. The flow loop consists of four -foot lengths of.5-inch-insidediameter transparent PVC pipe, joined together with PVC unions. The PVC pipe is fixed to a metal frame that hangs from an axle attached near the ceiling of our High Bay Laboratory. With little effort, the flow loop can be tilted from the vertical alignment for studies of flow in inclined wells by rotation on the axle. A separator divides the produced liquid from the produced gas. Gas is vented from the top of the separator. The collected liquid is recycled to the bottom of the loop. Recycling of liquid facilitates testing of a wide variety of liquids, including liquid hydrocarbons. As shown in Fig, the flow loop is equipped with several measuring devices: absolute and differential pressure transducers, and a mass flow meter, all connected for data () aquisition by computer. Cumulative lifting of liquids from the flow loop during tests is monitored by changes of water level in the separator, measured with a differential pressure transmitter. An unsteady-state mode of operation was chosen for tests in the flow loop. In these tests, the bottom end of the flow loop was initially filled to a liquid depth of about feet prior to injecting any gas. Then, gas injection was begun. In these unsteady-state experiments, the bottom of the flow loop probably behaves a lot like the bottom of a producing gas well. At the start of this investigation, there was concern for the undesired effects on pressure drop and liquid hold-up caused by the inlet and outlet. Previous investigators of two-phase vertical flow provide some insight for dealing with these,, issues. For the experiments reported here, the liquid was injected into a short mixing section filled with glass marbles at the bottom of the flow loop. To eliminate exit effects, the flow loop was first constructed with a gracefully bending u-tube at the top (Figure ). It was thought that such an exit would provide effective re-direction of the upward flow of gas and liquid back down to the separator. However, early observations showed that large fractions of the drops were impinging on the walls of the u- tube and flowing back into the loop. After this effect was noticed, a 9 elbow and a dead-end-tee were tested for redirecting the flow to the separator. These fittings are also shown in Figure. It was found that both the elbow and the tee transported more of liquids from the top of the loop to the separator. Results for the three well heads will be discussed later in this paper. Also in the early stages of this research, it was recognized that the union couplings were altering the two-phase flow in the loop. Close examination of the couplings revealed a lip at the joint with a slightly smaller inside diameter than the PVC pipe. This slight reduction in diameter was enough to induce a small accumulation of liquid above each coupling. Such behavior became more noticeable as the gas flow rate increased. To eliminate this effect, the couplings were machined to remove the lip. Figure compares liquid collection histories in the separator before and after removal of the lip from the couplings. Of an infinite variety of possible designs for inserts, five designs that were tested are shown in Figure. The first design was a large diameter orifice, termed an open disk. A slitted disk with teeth extending to the center of the disk also was tested. If the teeth are flexible, as they were in this study, they will bend to allow gas to flow through the center of the array of teeth. The toothed disk in Figure has eight teeth, also extending to the center. The motivation for adding teeth to the disks is to bring the liquid closer to the center of the tubing to ensure liquid exposure to the higher gas velocity. In addition, creating droplets near the center of pipe may minimize re-deposition of droplets to the liquid film on the tubing wall. It was found that inserts cut from a sheet of transparency plastic could be placed between couplings of the

3 SPE 5565 ENHANCING LIQUID LIFT FROM LOW PRESSURE GAS RESERVOIRS measurement section without causing leaks. Thicker materials were tested, but leaks resulted. Experimental Results Four factors that contribute to liquid lifting were varied in this study: gas flow rate, type of insert, inclination, and wellhead design. A matrix of experiments was planned and completed to study these factors. The effects of these four factors on liquid lifting are summarized below. Gas Flow Rates. Four flow rates were selected for study:, 5,, and 5 standard cubic feet of air per minute (scf/min). Experiments showed that the critical flow rate for lifting water in the flow loop was about 5 scf/min at mean operating pressure between psig and 7 psig. With Eq., the predicted critical flow rate is between scf/min and 7 scf/min for the range of operating pressures. While with Eq., the predicted critical flow rate is between 5 scf/min and 9 scf/min much closer to the observed 5 scf/min. Although the gas flow rates used in this investigation were generally lower than the critical flow rate, liquid was still lifted from the loop by the flowing gas. There are two main reasons for this. First, the flow loop is just feet tall. If the flow loop were rather than feet tall, there would likely be no lifting of liquids at the rates of this investigation. Second, the ability to lift liquids in these unsteady state experiments strongly correlates to the initial volume of liquid in the loop. With less liquid, lifting would be more difficult. As shown in Figure 5, the liquid production response varies with gas flow rate. As expected, the response increases as the gas flow rate increases. At the start of gas flow in these unsteady-state tests, liquid is often produced by a slugging flow. Slugging is particularly hard to avoid at the start of the higher flow rate tests. With a taller flow loop, the production of slugs at startup might be avoided. Flow regimes, specifically the transition from slug flow to annular-mist flow, are important in liquid lifting. The flow regime map used by Duns and Ros 5 is reproduced in Figure 6. For this figure, the liquid and gas velocity numbers are defined as follows: ρ L Liquid Velocity Number = v sl () gσ ρ G Gas Velocity Number = v sg () gσ In these equations, the superficial velocities for gas and liquid are just the total flow rate of either phase divided by the crosssectional area for flow. (An alternative flow regime map is found in Ansari et al. 6 ) Flow regime maps, like the one in Figure 6, are intended for steady-state flow, while our experiments were all unsteady. However, it is expected that the map should be applicable to snapshots in time of our experiments. / / Using the above definitions, superficial velocities and velocity numbers for gas and liquid were calculated for the flow range of our experiments. Gas velocity numbers varied from to 6, and liquid velocity numbers ranged from. to.. (For comparison with the map of Ansari et al., superficial gas velocities ranged from 8 to m/s, and superficial liquid velocities were less than. m/s.) The domain of our experiments is in the slug flow regime on the map in Figure 6. On the map of Ansari et al. 6, our experiments are in the slug or churn flow regime. It is useful to note that the liquid velocity numbers (and the superficial liquid velocities) for our experiments are below the range usually shown in flow regime maps. Indeed, liquid velocity numbers for gas wells afflicted by liquid holdup are often below those shown in flow regime maps. It is expected that the regime boundaries can be extrapolated vertically downward in these maps to the region of interest without error. Drawings of flow behavior (similar to those in Ansari et al. 6 ) for different flow regimes are shown in Figure 7. The drawing for slug-annular transition is very similar to what is seen in our flow loop for a gas flow rate of 5 scf/min. At this flow rate, upward moving waves of liquid films are observed. As gas flow rate is reduced from 5 scf/min in the flow loop, the behavior of the liquid film alters slowly toward the slug flow regime. At scf/min, greater numbers of waves are observed, and they oscillate up and down slightly. The liquid film at scf/min is thicker than the film at 5 scf/min. Furthermore, fast moving liquid waves that are larger than the others are witnessed irregularly. At and 5 scf/min, oscillation of the waves becomes more apparent. The oscillating waves at the lower flow rates cause the differential pressure to fluctuate severely. Extensive experimentation showed that it is very difficult to define the critical flow rate based on ability to lift liquids in our flow loop. However, fluctuations in the pressure drop may give a reasonable indication of the critical flow rate. One would expect pressure drop fluctuations to be small for gas velocities above the critical flow rate, because the population of large waves in the tubing is small. On the other hand, below the critical flow rate, the population of waves is large, and pressure drop fluctuations are also large. Based on pressure drop fluctuations, the critical flow rate was identified as about 5 scf/min for the operating conditions of these tests. Inserts. Experiments with and without an insert in place were conducted for each gas flow rate, for vertical and 5. inclinations, and for three types of wellhead. Each insert was installed in the union coupling at the middle of the -foot tall pipe. Just a small portion of the results is reproduced here. Other results are available in Yamamoto. The initial intent in testing of the inserts was to witness if if the inserts enhance droplet formation. However, the first discovery from the laboratory experiments was that the open disk inserts completely prevent liquid from falling down the wall of the tube. At the start of an unsteady-state experiment,

4 H. YAMAMOTO AND R. L. CHRISTIANSEN SPE 5565 all the liquid was in the bottom half of the flow loop, with some liquid film extending up to the bottom of the insert. But, as an experiment progressed, the liquid was lifted from the bottom to the region above the insert, about half way up the flow loop, or feet from the bottom. Down to scf/min, no liquid fallback was observed with all inserts when the pipe is in the vertical position. However, at scf/min with the /8-inch open disk insert, when the loop is tilted 5., liquid would pass intermittently down through the insert. Since the inserts prevent liquid fall back, it is no surprise that the volume of liquid lifted is greater with an insert than without an insert as shown in Figure 8. The liquid that normally falls back and accumulates in the lower section of the pipe is lifted above the insert and eventually transported to the separator. Liquid lifted by the flow of gas in the slug-annular transition is mainly in the form of droplets, not a liquid film. From the start of an unsteady-state experiment, the liquid at the bottom gradually rises toward the top of the loop. Liquid droplets are ripped from the liquid film and from bridges of liquid film ruptured by the gas breaking through them. A portion of the droplets seem to be transported by flow of gas through an insert, while the remainder strike the liquid film on the wall, or impinge on the bottom surface of an insert and join the liquid film on the wall below the insert. Droplets that pass through an insert may continue their upward motion or impinge on a liquid film, liquid wave, or bridges. Inserts promote intermixing of liquid and gas. The intermixing is most aggressive with a slitted disk, followed by the perforated-folded-toothed disk, and least with the /8- inch open disk. With the slitted disk, the toothed disk, or the perforated-folded-toothed disk, streams of liquid droplets can be observed moving upward from the insert at the center of the pipe from the tips of the teeth for gas flow rates of scf/min and 5 scf/min. The liquid lifting rates observed in the unsteady-state experiments were generally less than bbl/day except in the first few minutes of a test. Future experiments with steadystate conditions may demonstrate if higher lifting rates are possible. An important concern for field implementation of the inserts is the resulting pressure drop. For low production gas wells, excess pressure drop in the tubing results in low productivity, or ultimately no production. As shown in Figure 9, the pressure drop across the inserts is highest with a slitted disk and lowest with a /8-inch open disk. The pressure drop increases as the opening of an insert decreases. Inclination. Figure shows a comparison of liquid lifting response for a vertical and for an inclined configuration. It is obvious that the response is less with an inclined pipe. In the inclined pipe, liquid film accumulates on the lower side of the pipe; thus, a stream of falling liquid film is visible only on the lower side. It was also observed by shining a flashlight through the pipe (with room lights off) that more droplets were present in the lower portion than in the upper portion of the tube. Wellheads. As mentioned earlier, the wellhead configuration has a strong effect on liquid lifting in our flow loop. Figures and show that liquid production responses are the least with a u-tube wellhead. Although the production with a u-tube wellhead is greater when a slitted disk is inserted as shown in Figure, the response is still much less than that for an elbow or a dead-end tee. There is no significant difference in the production response of an elbow and a dead-end tee as shown in Figures and. In most experiments, the production response was a little higher for the elbow, while in the remaining experiments, the production was higher for the dead-end tee. There is no distinguishable difference the traces of pressure drop for the two wellheads. These experiments with the three wellheads were quite interesting. In the field, the majority of wellheads have either an elbow or a dead-end tee for directing gas flow to surface collection lines. One should wonder if an improvement in liquid production could be obtained by a slight modification of the wellhead design. In addition, severe doglegs in a well may obstruct upward lifting of liquids in much the same way that the u-tube did. Conclusions. In unsteady-state experiments, liquid production response was always greater with inserts than without inserts. It is not yet clear if the critical flow rate is reduced with the inserts.. Below the critical flow rate, liquid fall back can be prevented with inserts such as the open disk and other designs.. Although a specific insert design can inhibit liquid fall and enhance liquid lifting, it is expected that all inserts will eventually fail as gas flow rate decreases.. The slitted disk produces the most droplets, but has the highest pressure drop. 5. The % upward adjustment of the critical flow rate equation (suggested by Turner et al.) is not needed to fit our data. 6. It is difficult to detect critical flow rate from liquid production response of the flow loop. The extent of fluctuations in the pressure drop is a clearer indication of the critical flow rate. 7. The elbow wellhead and the dead-end tee give better liquid production response than the u-tube wellhead. The next step in the design of this new technology is to determine the spacing of the inserts in the tubing. Then, the concept could be brought to a field for tests in real gas wells. Prior to field tests, methods for installation of the inserts, as well as problems with normal well operations with inserts in the tubing should be considered. It would be very helpful if inserts could be installed or removed from the tubing conveniently. In addition, it would be beneficial if the inserts did not interfere with normal wire or slick line operations to collect pressure data or perform other tasks. If the inserts are

5 SPE 5565 ENHANCING LIQUID LIFT FROM LOW PRESSURE GAS RESERVOIRS 5 not properly designed, tools might stick on them or destroy them. In this investigation, methods of installation of inserts were totally ignored, but insert designs were chosen that might not interfere entirely with many well operations. Acknowledgements The data for Colorado gas wells was provided by Dwight s Energy Service; this data was analyzed by Mr. Satya Putra of the CSM Petroleum Engineering Department. The mass flow meter was donated to the department by Micro Motion, Inc. Funding for a portion of this research was provided by the PERFORM consortium. Members of this consortium at the time of this research were Amoco, Arco E&P, BJ Services, BP Exploration Alaska, Carbo Ceramics, Enron Oil & Gas, Marathon Oil, Maxus Energy, Mobil E&P US, Pennzoil E&P. Nomenclature g = acceleration of gravity, L/t, ft/s v c = critical velocity, L/t, ft/s v sl = superficial liquid velocity, L/t, ft/s v sg = superficial gas velocity, L/t, ft/s σ = surface tension, ml/l t, dyne/cm ρ L = liquid density, m/l, lbm/ft ρ G = gas density, m/l, lbm/ft. Yamamoto, H: Reducing Critical Gas Flow Rate for Lifting Liquid from Gas Wells, MS Thesis, Colorado School of Mines, Golden, Fore, L.B. and Duckler, A.E.: Droplet Deposition and Momentum Transfer in Annular Flow, AIChE J. (September 995) 6.. Ros, N.C.J.: Simultaneous Flow of Gas and Liquid As Encountered in Well Tubing, J. Pet. Tech. (October 96) Fore, L.B. and Duckler, A.E.: The Distribution of Drop Size and Velocity in Gas-Liquid Annular Flow, Int. J. Multiphase Flow (995) Duns, H., Jr. and Ros, N.C.J.: Vertical Flow of Gas and Liquid Mixtures in Wells, Proceedings of the 6th World Petroleum Congress (96) Ansari, A.M., Sylvester, N.D., Sarica, C., Shoham, O., and Brill, J.P.: A Comprehensive Mechanistic Model for Upward Two- Phase Flow in Wellbores, SPE Prod. & Fac. (May 99) 5. Vent References. Duggan, J.O.: Estimating Flow Rates Required To Keep Gas Wells Unloaded, Net. Gas Eng. (December 96) Turner, R.G., Hubbard, M.G., and Dukler, A.E.: Analysis and Prediction of Minimum Flow Rate for Continuous Removal of Liquids from Gas Wells, J. Pet. Tech. (November 969) Coleman, S.B., Clay, H.B., McCurdy, D.G., and Lee Norris III, H.: A New Look at Predicting Gas-Well Load-Up, J. Pet. Tech. (March 99) 9.. Smith, R.V.: Practical Natural Gas Engineering, nd Edition, PennWell Books, Tulsa, Hutlas, E.J. and Granberry, W.R.: A Practical Approach to Removing Gas Well Liquids, J. Pet. Tech. (August 97) Libson, T.N. and Henry, J.R.: Case Histories: Identification of and Remedial Action for Liquid Loading in Gas Wells Intermediate Shelf Gas Play, J. Pet. Tech. (April 98) Beggs, H.D.: Gas Production Operations, OGCI Publication, Tulsa, Upchurch, E.R.: Expanding the Range for Predicting Critical Flowrates of Gas Wells Producing From Normally Pressured Water Drive Reservoirs, Paper 696 presented at the 987 SPE Annual Technical Conference and Exhibition, Dallas, September Bizanti, M.S. and Moonesan, A.: How to determine minimum flow rate for liquids removal, World Oil (September 989) Elmer, W.G.: Tubing Flow Rate Controller: Maximize Gas Well Production From Start To Finish, Paper 68 prepared for presentation at the 995 SPE Annual Technical Conference and Exhibition, Dallas, September -5. Air P DelP DelP P Flow Meter Separator Liquid Recycle DelP Figure. Schematic of flow loop. Dashed arrows represent electronic signals for data acquisition.

6 6 H. YAMAMOTO AND R. L. CHRISTIANSEN SPE 5565 U-Tube /8 in /8 in From Flow Loop To Separator (a) Open Disk (b) Open Disk Dead-End Tee From Flow Loop To Separator (c) Slitted Disk (d) Toothed Disk Elbow From Flow Loop To Separator Figure. Three alternative wellheads for the flow loop. (e) Perforated-Folded-Toothed Disk.5.5 Before.5 After.5.5 Figure. Five designs for inserts that were tested. Figure. Water collection in separator Before and After removal of lips from union couplings.

7 SPE 5565 ENHANCING LIQUID LIFT FROM LOW PRESSURE GAS RESERVOIRS 7 5 scf/min scf/min 5 scf/min scf/min Liquid Velocity Number Figure 5. Liquid Figure.. production Liquid response for for different gas flow gas rates flow (no insert) rates (no inserts). Froth Flow Bubble Flow Transition Mist Flow Plug Flow Slug Flow Heading.. 6 Bubble 5 Insert = Slitted Disk Bubbly Slug Slug-Annular Transition Insert = None Annular Annular Mist Figure 7. Drawings of flow behavior for different portions of flow regime maps. Figure 8. Liquid production response with and without an slitted disk insert for scf/min. Response for other inserts was similar. Figure.. Liquid production response, gas flow rate = scf/min Gas Velocity Number Figure 6. Flow regime map of Duns and Ros.

8 8 H. YAMAMOTO AND R. L. CHRISTIANSEN SPE Slitted disk. Perforated-folded-toothed disk Toothed disk. /8 open disk /8 open disk Figure 9. A comparison Figure.8. of Differential differential pressure pressure comparison, traces gas flow for rate the = five scf/min inserts at a flow rate of scf/min. DPT# refers to the lower differential pressure transducer in Figure Figure. Comparison of liquid production response with different wellheads. The gas flow rate is scf/min. There is no insert) 6 5 Tee Elbow Elbow U-Tube Tee U-tube DPT# (psid) Figure. Comparison of liquid production response with different wellheads for a slitted disk insert. The gas flow rate is scf/min. 6 5 Vertical Inclined Figure. Comparison of liquid production response for vertical and inclined flow (5. ) for a toothed disk insert. The gas flow rate is 5 scf/min.