Electrical Submersible Pumping of Gassy Oil Wells Literature Study Project work by Bendik Huflåtten Trondheim 17. December 2007 Preface This project has been carried out from September to December 2007. The topic of the project was chosen with inspiration from Gunder Holmstvedt at Aker Kværer Subsea. I would like to thank my supervisor Jon Steinar Gudmundsson, Professor at the Department of Petroleum Engineering and Applied Geophysics for good ideas and helpfulness during the work. --------------------Bendik Huflåtten Trondheim 17. December 2007 ii Abstract Artificial lift is widely used by the oil industry to both accelerate and increase oil production. The electrical submersible pump (ESP) is one of the most commonly used artificial lift systems in oil wells. The ESP is a multistage centrifugal pump placed down in the wellbore, capable of delivering high pressure increments for large oil rates. Reservoir oil is always saturated with natural gas, which will separate from the oil as the pressure is decreased below the bubble point. Since oil production is based on pressure drawdown, ESPs are likely to experience free gas in the suction fluid. Free gas is known to have a detrimental effect on ESP performance. This project is a literature study of how ESPs are affected by free gas and of how any such effects can be reduced or removed in ESP operations. ESP performance starts to reduce when the free gas volume fraction exceed three percent. Total loss of pressure increment over the pump, known as gas locking, can occur for gas fraction as low as 10 %. Severe pressure fluctuation, referred to as surging, is always observed just before the gas lock condition. The negative effect from the gas can be removed by separating oil and gas before pumping, leaving only the liquid phase to the pump. Another solution is to use a gas handling device to alter the two-phase flow before pumping, making it easier for the ESP to handle. iii Table of Contents Preface ........................................................................................................................................ ii Abstract ..................................................................................................................................... iii Table of Contents ...................................................................................................................... iv List of Tables .............................................................................................................................. v List of Figures ............................................................................................................................ v 1. Introduction ............................................................................................................................ 1 2. Literature review .................................................................................................................... 3 3. Electrical Submersible Pumps ................................................................................................ 8 3.1 Introduction ...................................................................................................................... 8 3.2 Designing an ESP system ................................................................................................. 9 3.3 ESP inflow/outflow characteristics .................................................................................. 9 3.5 Limitations and Service envelope .................................................................................. 11 4. Effect of gaseous fluid on ESP performance ....................................................................... 12 4.1 Basic concepts ................................................................................................................ 12 4.2 Two-phase considerations .............................................................................................. 13 4.3 Experimental data ........................................................................................................... 14 5. ESP operations in gassy oil wells ......................................................................................... 16 5.1 Prediction of downhole gas volumes.............................................................................. 16 5.2 Operating ESPs below the perforations.......................................................................... 17 5.3 Downhole gas handling equipment ................................................................................ 18 5.4 Radial or mixed-flow design .......................................................................................... 19 6. Discussion ............................................................................................................................ 20 7. Conclusions .......................................................................................................................... 22 8. Nomenclature ....................................................................................................................... 23 9. References ............................................................................................................................ 24 10. Figures ................................................................................................................................ 25 iv List of Tables Table 1: Service envelope for operating ESPs per 2003. ......................................................... 11 List of Figures Figure 3.1: Typical ESP configuration. .................................................................................... 25 Figure 3.2: Typical plot of pump performance curves. ............................................................ 25 Figure 3.3: Well flows naturally............................................................................................... 26 Figure 3.4: Well needs artificial lift. ........................................................................................ 26 Figure 3.5: Inflow & Outflow performance with and without ESP. ........................................ 26 Figure 4.1: Pump performance with water only. ...................................................................... 27 Figure 4.2: Pump performance with 3.1 % free gas at intake. ................................................. 27 Figure 4.3: Pump performance with 4.5 % free gas at intake. ................................................. 27 Figure 4.4: Pump performance with 7.0 % free gas at intake. ................................................. 27 Figure 4.5: Pump performance with 11 % free gas at intake. .................................................. 27 Figure 4.6: Pump performance with 14 % free gas at intake. .................................................. 27 Figure 4.7: Pump performance with 17 % free gas at intake. .................................................. 28 Figure 4.8: Limitation of “black box” measurement. ............................................................... 28 Figure 4.9: Average and stage-wise pump performance for water, 250 psig intake pressure.. 29 Figure 4.10: ESP average pressure increment as a function of gas and liquid flow rate. ........ 29 Figure 4.11: Pressure increment for all stages at 5 Mscf/D. .................................................... 30 Figure 4.12: Pressure increment for all stages at 7.5 Mscf/D. ................................................. 30 Figure 4.13: Pressure increment for all stages at 15 Mscf/D. .................................................. 30 Figure 4.14: Approximated volumetric gas fraction for Figures 4.10 through 4.13. ............... 31 Figure 5.1: ESP operations below the perforations. ................................................................. 31 Figure 5.2: Excluding gas separators. ...................................................................................... 32 Figure 5.3: Expelling gas separators. ....................................................................................... 32 Figure 5.4: Gas handling devices. ............................................................................................ 32 Figure 5.5: Axial (K-70) and radial (C-72) impeller designs. .................................................. 33 Figure 5.6: Axial impeller design (K-70), 60 psig intake pressure, 9.92 % gas. ..................... 33 Figure 5.7: Radial impeller design (C-72), 55 psig intake pressure, 9.92 % gas. .................... 33 v 1. Introduction Artificial lift is used when reservoirs do not have enough energy to naturally produce oil to the surface or at the desired economic rate. An artificial lift system adds energy to the fluid column in the wellbore, thus reducing the backpressure on the reservoir. More than 90% of producing oil wells require some form of artificial lift. The electrical submersible pump (ESP) was introduced as an artificial lift system in 1927. Today ESPs are the second most common artificial lift method, with over 100,000 pumps operating globally. The ESP consists of a centrifugal pump mounted on an electrical motor. Centrifugal pumps are dynamic devices that use kinetic energy to increase fluid pressure. The centrifugal pumps used in ESP applications are mostly multistage pumps with small diameters and many stages. This design makes them fit inside small wellbores, still being able to deliver high pressure increments. ESPs are successful when handling liquids, ranging from low to medium viscosities. Their performance is characterized by the relationship between the pressure increment over the pump and the flow rate through the pump, for a certain rotational speed. This relationship is usually shown in characteristic pump performance curve plots. The performance curves are based on tests done by pump manufacturers using water as test fluid. Generally the production of oil is associated with the production of free gas. Reservoir oil is always saturated with gas. If the pressure or temperature is decreased below the oil’s bubble point, free gas will start to separate from the oil. Therefore, the pressure drawdown required to produce oil from a reservoir will result in formation of free gas somewhere in the production system. Usually, in mature fields it is preferable to reduce the reservoir pressure below the bubblepoint, making use of solution gas drive to maintain the somewhat lower pressure. In any case, ESPs are likely to experience free gas during their lifetime. Centrifugal pumps are highly affected by the presence of free gas in the pump fluid. Depending on the amount of free gas, the effects may vary from slight gas interference to complete gas locking. For small gas fractions, gas interference shows up as minor performance degradation. As the amount of free gas increases, rapid pressure fluctuations, known as surging, will take place. Eventually, if the pump ingests too much gas, the result will be a complete loss of pressure increment over the pump, known as gas locking. 1 This project is a literature study regarding the use of ESPs in gassy wells. The work intends to identify how ESP performance is affected by free gas in the suction fluid. Experimental data in the literature will be the basis for the study. How different ESP setup and design can reduce gas interference is also discussed in the project. The latter will use standard ESP operations as basis. 2 2. Literature review ESP performance under two-phase flow conditions Before the petroleum industry became interested in the effect of gas on ESP performance, very few studies were carried out on this topic. Although the nuclear industry had conducted several experiments on the effect of free gas on centrifugal pump performance, the models they had developed could not be used to predict ESP performance for the corresponding flow conditions. This was due to significant difference in pump design, usually single stage pumps with large diameter. The petroleum industry is mainly concerned with multistage pumps with small diameter. The petroleum industry started to perform their own experiments more than twenty years ago, and several studies have been carried out since then. These studies have been fundamental to understand the basic effects and trends regarding ESP performance under two-phase flow conditions. Recent studies have also been important to derive theoretical models and correlations to predict ESP performance and behavior when handling two-phase flow. Some of the most important works done by the petroleum industry will be presented in the next paragraphs. Lea and Bearden (1982) was among the first to perform laboratory testing of ESP performance under gassy conditions. Their research has been fundamental for understanding the basics, and essential for further studies done on the topic. They conducted two different tests. The first test used air and water as working fluids. A multistage centrifugal pump containing five Centrilift I-42 stages were used in the experiment. The gas fractions were varied from 0 to 17 %, and the intake pressures from 25 to 30 psig. The second test used CO2 and diesel as working fluids. Three different pump types were tested. The first of them was an eight-stage I-42 pump, the same as used in the air-water test. It was tested with gas fractions ranging from 0 to 40 %, and with intake pressures at 50, 100 and 400 psig. The two other pumps tested were a C-72 radial pump and a K-80 mixed flow pump, with optimal flow rate of 72 and 80 gal/min respectively. These two last pumps provided a comparison of two different designs with the approximate same design rate. These pumps were tested with 10 and 15 % free gas, at intake pressures of 50, 100 and 200 psig. 3 The work by Lea and Bearden was essentially experimental, and no models to account for the observations were presented. They concluded that it exist critical limits of free gas, at which the pump performance is severely reduced. Further they concluded that the pump performance under two-phase conditions is somewhat affected by both intake pressure and pump design. Sachdeva (1989) presented the first two-phase model for ESP radial stages developed by the petroleum industry. This model was based on the two-phase models made by the nuclear industry, adapted to fit the multistage pumps used in ESP installations. The model was calibrated with data from the Lea and Bearden experiments. To include axial design in the model it was extended three years later by Sachdeva (1992). Through the next years several experiments and analysis were executed, and a few theoretical models were developed. Unfortunately, further testing showed that these models were valid only for pumps with the same number of stages as the pump used to obtain the data from which the models were made. Pessoa (2000) introduced a new way to perform two-phase tests on multistage pumps to solve the model validity problem. Rather than measuring the pressure increment over the pump, the pressure increment was measured over each stage. The experiment was conducted at a facility provided by Tulsa University Artificial Lift Projects (TUALP). A 22-stages centrifugal pump (GC6100, 513 series) was used in the experiment. Air and water were used as working fluids, and the pump intake pressure was kept constant at 100 psig. The test procedure consisted of varying the liquid flow rate for a constant gas rate. Then increase the gas rate and repeat the experiment. Liquid flow rates from 900 to 8,200 B/D and gas flow rates from 5,000 to 39,000 scf/D were used. All tests were done at a constant pump speed of 3,208 RPM (55 HZ). This was the first time stage wise two-phase performance were evaluated and presented. Pessoa concluded that the average performance of the pump is significantly different from that observed for each stage. He also concluded that current knowledge was not sufficient to develop a general and accurate model for predicting head degradation, gas locking, and surging conditions. 4 Beltur1 (2003) performed a new two-phase experiment with the 22-stage pump used by Pessoa in 2000. Automatic control of the intake pressure, and the gas and liquid rate, were obtained by making a few minor modifications to the pump. The tests were done with constant pump speed at 2,916 RPM (50 HZ), and air and water as test fluids. Data were collected at intake pressures from 50 to 250 psig, in steps of 50 psig increments. The rest of the test procedure was performed much as in the Pessoa experiment. Maintaining constant pump intake pressure and gas flow rate, only liquid flow rate was varied through a single test. Beltur concluded that the stage position has an important effect on performance. Downstream stages experience better intake conditions, such as higher intake pressure, smaller gas void fraction and more homogeneous mixture than upstream stages. Because of this, downstream stages have better performance than the stages closer to the intake. The author also concluded that the average peak performance point in terms of liquid flow rate increases as gas flow rate increases. In addition to this Beltur agreed with Pessoa (2000) that current knowledge is not sufficient to develop a general and accurate model for predicting head degradation, gas locking, and surging conditions. Duran and Prado (2003) continued to test the 22-stages pump previously tested by Pessoa (2000) and Beltur (2003) at the TUALP facility. For the first time two-phase performance of one stage of a commercial ESP pump was completely mapped. In this test the 10th stage pressure was varied from 50 to 350 psi, in steps of 50 psi. Gas rates were varied from 5,000 to 90,000 scf/d and liquid rates were varied from 2,000 to 6,950 B/D. Air and water were used as work fluids. The pump speed was constant at 2,450 RPM (42 HZ). This work was both experimental and theoretical. Analyses and experimental data enabled the development of models and correlations to predict ESP performance under two-phase flow conditions. A model to predict the stage pressure increment under bubble flow conditions was developed. Additionally, correlations to predict the stage pressure increment under elongated bubble conditions, and to predict the flow pattern boundaries, were obtained. Duran and Prado concluded that three different two-phase flow regimes can exist inside an ESP. The first region corresponds with standard single-phase performance, i.e. pressure increment increase with decreasing liquid flow rate. The third region is the elongated bubble flow condition. In 1 Raghavan Beltur and Mauricio Prado, SPE, U. of Tulsa, Javier Duran, SPE, Ecopetrol, and Rui Pessoa, SPE, PDVSA-Intevep.: “Analysis of Experimental Data of ESP Performance Under Two-Phase Flow Conditions” (2003) – not used in this project. 5 this region severe degradation of ESP performance is observed. The second region is located between region one and three, and is characterized by pressure fluctuation and surging. They also concluded that the effects of intake pressure can be neglected for intake pressures below approximately 350 psi. For higher intake pressures a positive effect on pump two-phase performance is expected. In most of the experiments listed above, air and water are used as test fluids. This is easy and convenient and the physical properties for both air and water are well known. The two phases separates very fast, and so the air-water mixture can be used as worst case scenario for real applications using oil and natural gas. ESP operations in gassy oil wells ESPs are widely used in oil wells, and a number of technical reports are available on ESP operations in gassy oil wells. However, as most ESP operations are case dependent, so are the papers written on the subject. Only a few studies summarize the basics of, and the knowledge obtained from, this type of operations. A general study of downhole gas handling equipment used in connection with ESPs is presented by Wilson (1994). This study is based mainly on field experience with ESPs operating with or without some sort of gas separation device. The paper focuses on gas separators affect on ESP run life, but it covers several other aspects of ESP operations in gassy oil wells just as thoroughly. Different types of gas handling equipment are well presented and described in the paper. Different ways to operate ESPs in gassy environments without any gas handling equipment are also discussed. Finally the effects and basics of free gas in ESPs are to some extent covered in the paper. The advantages and disadvantages of ESP operations below the perforations are presented by Wilson (1998). Several equipment options for this type of operations are presented. Solutions to the problem with insufficient motor cooling are discussed thoroughly. A case study of a field where below perforation operations have been successful is also presented. Wilson concluded that the natural annular separation of gas, got in below perforation operations, will minimize gas interference in the pump. Further, he concluded that as long as the production rate is limited to keep the reservoir saturated, there will be no advantages in locating the ESP 6 below the perforations. This type of operation should only be considered in cases where the production can be increased. Artificial lift and basic ESP operations The increasing use of artificial lift in the petroleum industry has resulted in a lot of documentation on the field. The reasons to use artificial lift, and the benefits of using it, are well covered by technical papers. Various articles are also written on each of the different artificial lift systems, making it easy to find general information about ESP operations. On the other hand, only a few of the articles provide overview and comparison of all the artificial lift systems. The work of Vachon (2005) presents some of the latest aspects of ESP operations. The paper discusses optimization of ESP operations using downhole chokes and variable speed drives. Intelligent well technology, i.e. the ability to monitor downhole parameters and remotely control the system, is emphasized in the paper. Some pump thrust considerations are also mentioned. Additionally, a theoretical comparison between different ESP operations is provided. Vachon concludes that success in ESP applications depends on operating within the operation range of the selected pump. It is also concluded that intelligent completion system allows for optimal production rates, keeping the operation within the operating range, thus reduce the risk of pump failure. A more detailed presentation of variable speed drives is given in an article by Powers (1987). The paper focuses on how speed variation affects performance, thrust and longevity of ESPs. Critical pump speeds, vibration and cavitation are also discussed. Brown (1982) provides an overview of artificial lift systems and gives guidelines indicating which system to select. This paper covers the concepts of inflow and outflow relationships. It also lists relative advantages and disadvantages of each of the artificial lift systems. This is a good paper on the subject, although some parts of it are outdated. A newer article on how to select the right artificial lift method is presented by Heinze (1995). Mainly selection criterions are discussed in this article, but a brief overview of the different systems is also included. 7 3. Electrical Submersible Pumps 3.1 Introduction The basic concept of artificial lift is to add energy to the fluid column in a wellbore in order to create a predetermined bottomhole pressure so the reservoir may produce the objective flow rate [1]. Artificial lift is used for two purposes: Induce flow in naturally dead wells and, more commonly, increase/maintain flow rate in already producing wells. As artificial lift can start, accelerate and increase recovery from a reservoir it is widely used. By 2004 more than 90% of producing oil wells needed some form of artificial lift to reach the desired performance. Depending on the fluid properties and the well geometry there are several ways to artificially lift the production stream from an oil well. The most commonly used artificial lift systems are: Sucker rod (beam) pumps. Electrical submersible pumps. Gas Lift. Progressive cavity pumps. Hydraulic piston pumps. Hydraulic jet pumps. The electrical Submersible Pump (ESP) is today the second most used artificial lift device with over 100,000 operating pumps worldwide. The ESP system consists of a multistage centrifugal pump, an electrical motor, a motor controller, power cables, a topside power unit and some sort of gas separation component. The ESP is fully submerged and placed at a chosen dept in the wellbore. Power is transmitted to the motor through electrical cables fastened to the outside of the tubing. Figure 3.1 shows a typical ESP system configuration. [2] The ESP performance is described in characteristic pump performance plots. These plots show the relationship between the pressure rise (head) and flow rate through the pump, the horsepower requirements and the pump efficiency, all for a certain rotational speed. The plots are created by the pump manufacturer based on tests with water as test fluid. Usually for a multistage pump only one pump stage is displayed in such a plot. A typical multistage pump performance plot for one stage is shown in Figure 3.2. Any centrifugal pump will have a 8 design rate which will optimize the pump efficiency and run life. The pump design rate is identified by the vertex of the efficiency curve in Figure 3.2. [2] 3.2 Designing an ESP system The desired fluid flow rate, and the dynamic head required to produce this rate, will always be the two first considerations when designing an ESP system. The third consideration is the well casing size. As the flow rate is determined, a pump of the largest series that will fit inside the well casing, and that has a design rate as close as possible to the desired flow rate is selected. The total number of pump stages is calculated by dividing the total dynamic head needed by the head per stage. The required motor horsepower will then be calculated as the product of the horsepower per stage, the number of stages, and the specific gravity of the fluid to be produced. Another factor that must be considered when designing an ESP system is the use of a variable-speed drive (VSD). Since ESPs are powered by two-pole induction motors, their speed depends on the power supply frequency. A VSD makes it possible to control the power supply frequency, and thus the pump’s rotational speed. The VSP expands the pump’s rangeability, making it adjustable, and thus, capable to deal with a number of situations. [4] 3.3 ESP inflow/outflow characteristics Whether a well needs artificial lift is analyzed based on the predicted well performance. This predicted well performance compares to the well production flow rate and is calculated based on two factors referred to as the inflow and outflow relationships. According to Nodal analysis1 these relationships can be mathematically expressed as in the next sections. [5] Inflow performance relationship How changes in bottomhole pressure will affect the reservoirs ability to produce fluids is known as the inflow performance relationship (IPR). For ESP operations the inflow curve is given by: 1 Nodal analysis, defined as a systems approach to the optimization of oil and gas wells, is used to evaluate a complete producing system. 9 Inflow pwf ph p f ESP reservoir pc (1) Here pwf, Δph, Δpf, and, Δpc are the wellbore flowing pressure, the hydrostatic pressure drop, the frictional pressure drop, and the pressure drop across the downhole choke. Introducing the productivity index, PI: PI q (2) pres pwf or pwf pres q (3) PI The PI can be derived from the average reservoir pressure, pres and a well production test. When substituting equation 3 into equation 1 the expression becomes: q Inflow pres ph p f PI ESP reservoir pc (4) Outflow performance relationship This factor refers to the total pressure loss through the production system for any given flow rate. The outflow, or tubing intake, relationship depends on a number of factors. These factors are all related to the geometry and design of the production system or to the rate, properties and temperature of the well stream. In order to get the characteristic curve the outflow relationship is calculated for a wide range of flow rates. The outflow curve is given by: Outflow pwh ph p f wellhead ESP p pump (5) Here pwh, and Δppump are the wellhead pressure, and the pressure gain generated by the pump. Note that except for the hydrostatic pressure drop, all the pressure parameters used in the above equations vary directly with production flow rate. [5] If the IPR and the outflow relationship are plotted together in a common plot the intersection between the two lines will mark the well’s predicted production rate. Figure 3.3 shows a naturally flowing well. If the two lines do not overlap the well will not flow naturally. Figure 3.4 shows a well which needs artificial lift to flow. If artificial lift is utilized in the latter case the outflow curve is sifted downwards on the plot and production starts as it overlaps the IPR curve. [2] 10 The ESP adds energy to the fluid column by increasing pressure somewhere in the wellbore. This creates a characteristic outflow pressure curve from the bottomhole to the well head. A sudden pressure increment will appear in the wellbore at the ESP placing point. Figure 3.5 shows the inflow and outflow curves with and without ESP. It is seen in Figure 3.5 that the bottomhole pressure decreases while the well head pressure increases when the ESP is used. Ideally, provided sufficient well head pressure, the bottomhole pressure could be decreased without any increase in well head pressure. [6] 3.5 Limitations and Service envelope The advance in ESP technology has through the years made it applicable in increasingly harsh environments. However there is still a list of challenges concerning the use of ESPs in oil wells. This list includes the presence of high gas volumes, high temperatures, abrasive solids in the production stream and corrosive environments. The effects of these factors are to some extent connected to each other. ESPs are used in a large number of artificial lift operations worldwide. Table 1 gives a brief summary of the operation range of ESP systems deployed before 2003. The data listed in Table 1 is only meant to illustrate the wide ESP operation range, thus slight variations in the data may exist. Table 1: Service envelope for operating ESPs per 2003. ESP sevice envelope Maximum production rate: 15900 m³/d 100000 Minimum production rate: 24 m³/d 150 Maximum pressure head* increment: 3650 m 12000 Maximum operation depth: 4570 m 15000 Maximum operation temperature: 232 °C 450 Minimum surrounding casing diameter: 0,114 m 4 1/2 Maximum surrounding casing diameter: 0,340 m 13 3/8 *the pressure head increment depends on the number of pump stages B/D B/D ft ft °F in in 11 4. Effect of gaseous fluid on ESP performance 4.1 Basic concepts The presence of free gas in the pump fluid is known to affect the ESP performance. Generally the effects of free gas show up as a deterioration of the ESP single-phase performance curve. The extent of the deterioration depends on the amount of entrained gas and varies from slight gas interference to complete gas locking. [7] Between these two effects a phenomenon referred to as surging takes place. [8] Gas interference occurs for minor amounts of free gas and is characterized by a reduction in the overall pump performance. Slight gas interference is often indicated by rapid variations of the motor loading. As the gas interference increases it will show up as a reduction in pump performance. [7] Surging, or pressure instability, is recognized by a cyclic fluctuation of the system’s pressure. Although surging is a system phenomenon, rather than an exclusive pump phenomenon, it is always observed in ESP experiments with two-phase flow. This is because pump head and flow rate variations can influence the occurrence or the severity of surging. For two-phase flow surging tends to appear in regions close to gas locking. [8] Gas locking occurs if the pump ingests too much gas, resulting in a total loss of pump head. The control system will usually shut down the motor to protect it from the excessive underload before the pump head is completely lost. In any case the pump looses its ability to deliver any pressure head. [7] It is important not to confuse gas locking with gas blocking. Gas blocking is characterized by gas accumulation in the low pressure side of the impeller vanes. Although gas blocking may cause significant reduction in pump performance, the pump will still be able to deliver some pressure head and, thus, pump fluids. The term gas blocking refers to the static gas pockets partially blocking the flow area. [8] 12 4.2 Two-phase considerations To understand why free gas affects the performance of a centrifugal pump, it is necessary to analyze the forces acting on the fluid flow through the pump. As the fluid flow enters the impeller, the radial velocity of the flow is increased by the centrifugal force acting on it. Then the radial velocity is slowed down again as the flow approaches the stationary diffuser. This creates a radial pressure gradient. Generally the lowest pressure is found in regions near the rotational axis. When a two-phase flow enters a centrifugal pump, the centrifugal force will act differently on the two phases, depending on the phase density. Since the density of the gas is much smaller than the density of the liquid, the centrifugal force acting on the gas bubbles will be small compared to the one acting on the liquid. As a result, the force acting on the bubbles will be dominated by the radial pressure gradient, and the drag force which is a function of the slip ratio (the difference between gas and liquid velocity). These two forces work in opposite directions, with the force from the radial pressure gradient working towards the low pressure regions near the center of the impeller. A result of the situation described in the previous section could be accumulation of free gas in the pump. As the force from the pressure gradient tries to move the gas bubbles towards the impeller centre, the gas velocity decreases relative to the liquid velocity. This increases the slip velocity, until equilibrium is established between the forces due to the pressure gradient and the slip. The slip velocity is connected to the gas void fraction. When the slip velocity increases so will the gas void fraction. If the slip is small the gas will continue to move together with the liquid, but at a higher void fraction. For higher slip values the coalescence of small bubbles will occur at the impeller entrance. As the bubbles grow larger it gets more difficult for the liquid phase to drag them through the pump, leading to further coalescence and eventually to the formation of a stationary elongated bubble. This simple description of two-phase flow through a centrifugal pump reflects the basic concepts listed in chapter 5.1. Small slip values give a minor performance reduction, referred to as gas interference. Bubble coalescence and the formation of a stationary bubble results in severe performance reduction and surging. As the stationary bubble increases in size it will occupy most of the flow area, leading to gas locking. [9] 13 4.3 Experimental data To quantitatively connect ESP performance and free gas, experimental data has to be presented. Generally, the results of these types of experiments are difficult to present due to the large number of variables. [8] The air-water experiment, conducted by Lea and Bearden (1982), is maybe the most illustrative presentation of how pump performance and free gas are connected. This presentation is showed in Figure 4.1 through 4.7. The free gas volume at pump intake is the only parameter varied throughout the experiment, thus it is easy to see how this parameter influences the pump performance. For each gas rate, the experimental data obtained are compared to the manufacturer’s single-phase performance curve. As expected, in the test with water only (Fig. 4.1) the test results are found close to the published performance curve. For 3.1 %, 4.5 %, and 7 % free gas at intake (Fig. 4.2 through 4.4) the test results are shifted downwards, away from the single-phase curve. Still the test results form a curve, shaped almost like the single-phase curve. This shows a general reduction in pump performance. For 7 % free gas however, the test results show the beginning of serious departure from the single-phase curve. Figure 4.5 and 4.6 show the results with 11 % and 14 % free gas at pump intake. At 17 % free gas virtually no pressure increment is delivered by the pump, see Figure 4.7. [7] Stage wise pressure measurements were introduced by Pessoa in 2003. All previously conducted experiments were based on the “black box” concept. See Figure 4.8. The problem with this concept was that the data acquired from the experiments were valid only for a certain number of stages, thus no general theoretical models could be developed. Pessoa found that the average pump performance was significantly different from the individual performance of each stage. Figure 4.9 shows a comparison between average pump performance and the performance per stage for single phase water at 250 psig intake pressure. The dimensionless head used in Figure 4.9 is obtained by dividing the head values with the maximum head value (at zero flow). Similarly, the flow rate values are divided by the maximum flow rate (at zero head). It is seen in the figure that each stage has a different behavior. Particularly noticeable is the poor performance corresponding to the first pump stage. Pessoa suggests that this may be caused by some form of intake effect, or because this pump stage is old and exhibits high wear. [8] 14 The air-water tests done by Pessoa were carried out by keeping the pump intake pressure constant at 100 psig and increase the gas flow rate in steps of 2,500 scf/D. For each gas rate the liquid flow rate was varied from the maximum delivered by the pump to the minimum achieved just before the gas-locking condition occurred (ca. 900 to 8,200 B/D). Figure 4.10 shows the overall ESP pressure increment as a function of gas and liquid flow rates. The gas rate is varied from 5 to 35 Mscf/D. For reference the single-phase water curve is also included in the plot. The overall negative effect from free gas is seen in the plot. The stage-wise air-water performance corresponding to the 5, 7.5 and 15 Mscf/D free gas curves (in Fig. 4.10) is shown in figure 4.11 through 4.13. For 5 Mscf/D gas flow rate (Fig. 4.11) the pressure behavior per stage match the average for the pump. However, as the liquid flow rate becomes small, it is seen that the performance of the lower stages drops off somewhat before the higher stages. For 7.5 Mscf/D gas flow rate (Fig. 4.12) the average pump performance is no longer matched by all the stages. Stages one and two depart a lot from the pump average, and to some extent also stage three. As the gas rate is increased to 15 Mscf/D (Fig. 4.13) the same effect as seen for stages one and two at 7.5 Mscf/D is now seen for the eight lowest stages. Major variations in stage performance between the various stages are observed. Altogether, the three last plots show that the variation in stage performance increases as the gas fraction is increased. For comparison, the pumps average performance is illustrated by the thick dashed line in the plots. Again, there is some uncertainty connected to the first stage. The dimensionless rates used in the plots were calculated as for the single-phase experiment. [8] For comparison Figure 4.14 provides an approximation of the volumetric gas fraction (as a function of dimensionless liquid rate) for some of the gas rates used in Figure 4.10 through 4.13. 15 5. ESP operations in gassy oil wells In most oil wells, oil production is associated with the production of gas. As a consequence, ESPs are likely to experience gas. Since ESPs are highly affected by gas, it is critical to predict both present and future downhole gas volumes, in order to create the right design of the ESP system. Although the ESP system design will be case dependent, the measures taken to deal with free gas can t some degree be generalized. 5.1 Prediction of downhole gas volumes The oil in a reservoir contains dissolved gas due to high pressure. The oil can be either saturated or undersaturated with gas. Saturated oil is completely filled with dissolved gas, thus no more gas can dissolve in the oil. In the undersaturated oil, on the other hand, more gas can be dissolved in the oil. Whether the oil is saturated or not depends on the pressure and temperature, and the oil’s bubble point. The bubble point is where free gas starts to come out of the oil. As long as the pressure and temperature are kept above the oil’s bubble point no gas will form in the oil. The production of oil from a reservoir requires a pressure drawdown. In addition to this, the reservoir pressure depletion results in further pressure reduction as production continues. Therefore it will seldom be possible to produce the reservoir at the desired rate, while keeping the bottomhole pressure above the bubble point throughout the well’s lifetime. It can also be preferable to reduce the reservoir pressure below the bubble point, resulting in accumulation of free gas, eventually forming a gas cap. As the pressure is decreased further, the gas cap will expand, maintaining the pressure. This is known as gas solution drive. The amount of gas coming out of the oil depends on the oil and gas properties as well as the pressure and temperature. When the gas volumes are predicted, several measures can be taken to reduce or remove the negative effects from the gas on the ESP operation. [10] 16 5.2 Operating ESPs below the perforations ESP operations in gassy conditions can be optimized by placing the pump intake below the bottom-most casing perforation. The fluids will then have to flow downwards from the perforations to the pump intake, resulting in a passive, but efficient, gas-liquid gravity separation in the casing annulus. This separation depends on the liquid flow velocity (i.e. the liquid flow rate), and becomes less effective as the velocity increases. Nevertheless, due to this separation, most of the gas can be produced through the casing, leaving the ESP with only minor gas fractions. [11] The ESP’s intake pressure is also an important factor when handling gas. As the intake pressure increases, the gas interference is reduced. If the ESP is set below the perforations the pump intake pressure can to some extent be increased without increasing the pressure at the perforation. This makes it possible to apply the maximum drawdown and still keep the intake pressure somewhat higher than the perforation pressure, thus reduce gas interference without reducing the production. [12] The main problem with ESP operations below the perforations is the cooling of the motor. In standard ESP operations the motor is cooled by the well stream constantly flowing around it. If the ESP motor is located in a rathole below the perforations, the well stream will not naturally flow past the motor. This will cause the motor to overheat. There are several ways to solve this problem. The ESP can be placed inside a small casing (shrouded), open only in the lower end, forcing the fluids to flow past the motor before entering the pump. Alternatively, fluid can be circulated around the motor, either by using a separate pump, or by divert some of the fluid flow from the main pump. There is also possible to make a motor that is able to tolerate extreme temperatures, but this will increase the motor costs. Another problem with this type of ESP setup is the production of sand. The rathole could be filled with sand due to extensive sand production. [12] Figure 5.1 shows three configurations of ESPs operating below the perforations. [13] In any case, if the well production rate is limited in order to keep the reservoir saturated, there will be no advantages in setting the ESP below the perforations. The extra expenses related to this type of operations will only be justifiable in situations where the production can be increased. [12] 17 5.3 Downhole gas handling equipment If it is not preferred to place the ESP below the perforation, operations in gassy environments can be optimized by using some sort of downhole gas handling equipment. Downhole gas handling equipment can be divided into three categories depending on the method used to deal with the gas. The two first methods, the excluding gas separator and the expelling gas separator, both intend to separate the gas and liquid phases before pumping. The third method, the gas handling device, intends to change the flow path or the fluid properties to make the flow easier for the pump to handle. [13] The excluding gas separator is a passive separation system. It uses gas buoyancy and gravity to separate the two phases. Since gas generally will move upwards in a liquid, by forcing the flow to flow downwards before entering the pump intake, the gas is separated from the liquid. Figure 5.2 shows some of the configurations commonly used in this type of separation. The excluding gas separation method is very much the same as locating the pump intake below the perforations. Due to somewhat smaller downward flow area however, the downward flow velocity will be higher, thus the separation less efficient, compared to the below perforation separation. The excluding gas separator is claimed to be functional for gas fractions under approximately 15%. [13] The expelling gas separator is a dynamic, or rotary, separation system. By inducing rotation to the fluid flow the two phases will separate, due to centrifugal force. Although a number of different designs are available, they all shear some general features. This includes an impeller or inducer at the intake, to force the fluids through the device, a separation chamber, where the fluid is subjected to rotation, and a crossover to lead the gassy flow out into the casing annulus. The different types can be named by the way they apply rotation to the fluid flow. Figure 5.3 shows three different types, the paddle wheel, the rotating chamber and the vortex respectively. The total gas handling capacity of these separators can be increased by installing several of them in series. [13] The gas handling device changes the properties of the flow, or the pump intake condition, making either of them more compatible for the pump to handle. These devices are all dynamic, and three types are used: Chargers, mixers and the inverted impeller type. See Figure 5.4. Chargers use oversized impellers to increase the fluid pressure. This can reduce the bubble size, and possibly the free gas volume, before the fluid enters the pump. Mixers 18 break up gas bubbles, making the fluid stream more homogeneous, which will reduce separation in the pump. The inverted impeller type intends to removes the possibility of gas lock in the first impeller of the pump. This is done by invert the first impeller so that the impeller eye faces upwards. The fluid will then “fall” into, rather than being sucked into the first impeller. This should result in a permanent fluid column over the eye of the first impeller, keeping it from gas locking. [13] 5.4 Radial or mixed-flow design The impeller design can also affect the pump’s gas handling capacity. The axial (or mixed flow) impeller design does in general suffer less head degradation than the radial design. This is mainly because the gas-liquid velocity lag will be lower in an axial pump. The liquid-phase will be accelerated less in an axial pump than in a radial pump, and less head will be lost as velocity head. The axial design also tends to keep a bubbly flow regime for higher gas fractions than the radial design, which gives somewhat better gas handling capacity. The two impeller designs can bee seen in Figure 5.5. The Lea and Bearden diesel-CO2 experiment provides a comparative test between the two. The results from this test were used by Sachdeva (1992) to adjust a theoretical model for both axial and radial two-phase pump performance. The differences between the two types for can be seen if Figures 5.6 and 5.7 are compared. [14] 19 6. Discussion That free gas has a detrimental effect on centrifugal pump performance is well known and clearly seen in the experimental data available on the subject. Nevertheless, as the volumetric free gas fractions are very low only a slight reduction in pump performance is observed. No major reduction in pump performance seems to occur as long as the gas is carried through the pump by the liquid phase. As soon as free gas starts to accumulate in the pump however, pump performance is likely to drop rapidly. This is probably caused by the self intensifying effect of gas accumulation, leading to more gas accumulation. This effect tends to make the performance reductions sudden and to some extent unpredictable. Results from the Lea and Bearden air-water experiment show severe reduction in pump performance for a volumetric free gas fraction somewhere between 7 % and 11 %. This indicates that gas accumulation starts in this region. This is a wide gap, and the numbers should be used only to describe general trends. The test procedure used by Pessoa provides continuous data, but the results are not given in terms of volumetric gas fraction, which makes it difficult to compare them with the Lea and Bearden results. However, by using the approximated gas volume fractions some general trends can still be recognized and used for comparison. Free gas fractions below 3 % are seen to have virtually no effect on the average pump performance. As the free gas fraction is increased above this level the pump performance gradually separates from the single-phase curve, still increasing with decreasing flow rate. A peak in the two-phase curves is observed for gas fractions between 11 % and 15 %, followed by severe surging and rapid performance reduction. It must be noticed that when these peaks occur the pump performance is already way lower than the single-phase performance. The stage-wise pressure measurements done by Pessoa indicate that more pump stages could make the pump more tolerant to gas. This last indication agrees with the connection between pump performance and both gas fraction and flow regime. The use of air and water as test fluids can be considered a worst case scenario for pumping of oil and natural gas. In some cases this may prove useful, adding tolerance to the performance estimates. Nevertheless, more tests with real reservoir fluids would increase the understanding of the subject, making the predictions more accurate. 20 ESP operations below the perforations depend on sufficient motor cooling. Provided a costefficient and reliable motor cooling system this type of operation have a few advantages. The setup utilizes the passive gravitational separation, thus includes no moving parts, making it almost fail safe. At the same time the gas separation capacity is good, and can be adjusted by varying the distance between the ESP and the perforations. An excluding gas separator provides the same type of separation as below perforation operations, however at a somewhat reduced capacity. In both cases a problem with these operations could be that sand may separate out and accumulate in the region where the flow starts to flow upwards again. Active separation devices have proved to have good gas separation capacity, but add complexity to the system at the same time. To make use of the natural gas lift provided by the gas-liquid flow, downhole separation should be avoided if possible. A gas handling device could be able to keep the ESP performance high without the increased static backpressure from the pure liquid column. A gas handling device will add complexity to the operation, so the longevity must also be considered. 21 7. Conclusions 1. Gas-liquid fluids with free gas fractions above three percent affect ESP performance. 2. The effects of gas on ESP performance may vary from slight gas interference to gas locking. 3. Correct downhole gas fraction estimation is imperative when designing an ESP operation. 22 8. Nomenclature B/D Barrels per Day ESP Electrical Submersible Pump PI Productivity Index pwf Wellbore Flowing Pressure pwh Wellhead Pressure RPM Revolutions per Minute Mscf/D Thousand Standard Cubic Feet per Day (at 60°F and 14.7 psia) TUALP Tulsa University Artificial Lift Projects VSD Variable Speed Drive Δpc Pressure Loss across Downhole Choke Δpf Frictional Pressure Drop Δph Hydrostatic Pressure Drop Δppump Pressure Increase over the Pump 23 9. References 1. Kermit E. Brown: “Overview of Artificial Lift Systems” (1982) 2. Schlumberger webpage: http://www.slb.com/media/services/resources/oilfieldreview/ors04/sum04/03_esp_surveillanc e.pdf 3. Lloyd R. Heinze, SPE and Herald W. Winkler, SPE, Texas Tech University, and James F. Lea, SPE, AMOCO Productions Company: “Decision Tree for Selection of Artificial Lift Method” (1995) 4. Maston L. Powers, SPE, Conoco Inc.: “Effects of Speed Variation on the Performance and Longevity of Electrical Submersible Pumps” (1987) 5. G. Vachon, and K. Furui, Baker Oil Tools: “Production Optimization in ESP Completions with Intelligent Well Technology by Using Downhole Chokes to Optimize ESP Performance” (2005) 6. Professor Jon Steinar Gudmundsson webpage: http://www.ipt.ntnu.no/~jsg/undervisning/petroleumsteknologiGK/PetroleumsteknologiJSG.p df 7. James. F. Lea, Amoco Production, and J. L. Bearden, Centrilift Hughes Inc.: “Effect of Gaseous Fluid on Submersible Pump Performance” (1982) 8. Rui Pessoa, SPE, PDVSA-Intevep, and Mauricio Prado, SPE, U. of Tulsa: “Two-Phase Flow Performance for Electrical Submersible Pump Stages” (2003) 9. Javier Duran, SPE, Ecopetrol, and Mauricio Prado, SPE, U. of Tulsa: “ESP Stages AirWater Two-Phase Performance – Modeling and Experimental Data” (2003) 10. Zolotukhin & Ursin: “Introduction to Petroleum Reservoir Engineering” (2000) 11. R. Bohorquez, V. Ananaba and A.L Podio, University of Texas at Austin; O. Lisigurski, Oxy; and M. Guzman, Shell International E&P: “Laboratory Testing of Downhole Gas Separators” (2007) 12. B. L. Wilson, SPE, Oil Dynamics Inc., John Mack, Centrilift, and Danny Foster, AMOCO Production Company: “Operating Electrical Submersible Pumps Below the Perforations” (1998) 13. B.L. Wilson, Oil Dynamics Inc.: “ESP Gas Separator’s Affect on Run Life” (1994) 14. Rajesh Sachdeva, SPE, and D.R. Doty and Zelimir Schmidt, SPE, U. of Tulsa: “Performance of Axial Electrical Submersible Pumps in a Gassy Well” (1992) 24 10. Figures Figure 3.1: Typical ESP configuration. Figure 3.2: Typical plot of pump performance curves. 25 Figure 3.3: Well flows naturally. Figure 3.4: Well needs artificial lift. Figure 3.5: Inflow & Outflow performance with and without ESP. 26 Figure 4.1: Pump performance with water Figure 4.4: Pump performance with 7.0 % only. free gas at intake. Figure 4.2: Pump performance with 3.1 % Figure 4.5: Pump performance with 11 % free gas at intake. free gas at intake. Figure 4.3: Pump performance with 4.5 % Figure 4.6: Pump performance with 14 % free gas at intake. free gas at intake. 27 Figure 4.7: Pump performance with 17 % free gas at intake. Source: Lea and Bearden [7] (Figures 4.1 through 4.7) Figure 4.8: Limitation of “black box” measurement. 28 Figure 4.9: Average and stage-wise pump performance for water, 250 psig intake pressure. Figure 4.10: ESP average pressure increment as a function of gas and liquid flow rate. 29 Figure 4.11: Pressure increment for all stages at 5 Mscf/D. Figure 4.12: Pressure increment for all stages at 7.5 Mscf/D. Figure 4.13: Pressure increment for all stages at 15 Mscf/D. Source: Pessoa [8] (Figures 4.8 through 4.13) 30 Volumetric Gas Fraction 60 5 Mscf/D 50 7,5 Mscf/D 15 Mscf/D 20 Mscf/D Gas fraction % 40 30 Mscf/D 30 20 10 0 0,00 0,10 0,20 0,30 0,40 0,50 0,60 0,70 0,80 0,90 1,00 Dimensionless liquid flow rate (based on 8200 B/D) Figure 4.14: Approximated volumetric gas fraction for Figures 4.10 through 4.13. Figure 5.1: ESP operations below the perforations. 31 Figure 5.2: Excluding gas separators. Figure 5.3: Expelling gas separators. Figure 5.4: Gas handling devices. Source: Wilson [13] (figures 5.1 through 5.4) 32 Figure 5.5: Axial (K-70) and radial (C-72) impeller designs. Figure 5.6: Axial impeller design (K-70), 60 psig intake pressure, 9.92 % gas. Figure 5.7: Radial impeller design (C-72), 55 psig intake pressure, 9.92 % gas. Source: Sachdeva [14] (figures 5.5 through 5.7) 33