SPE-173945-MS A State-of-the-Art Modeling Technique for Thrust Prediction in Bottom Hole Electrical Submersible Pumps Emanuel Marsis, PhD., Abhay Patil, PhD., David Baillargeon, David McManus, Steven Gary II, Brett Williams, Dario Lana, Jonathan Nichols, Cristian Von Zedtwitz, and Jason Ives, Baker Hughes Inc. Copyright 2015, Society of Petroleum Engineers This paper was prepared for presentation at the SPE Artificial Lift Conference — Latin America and Caribbean held in Salvador, Bahia, Brazil, 27–28 May 2015. 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 have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper 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 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright. Abstract Thrust in Electrical Submersible Pumps (ESPs) is a very important factor that affects pumps’ performance. Thrust forces in pumps can increase the friction losses and reduce the overall pump efficiency and lifetime. Pump designers have to design pumps to handle thrust generated in the operating range whether by adding up-thrust protection, down-thrust protection, or sometimes both kinds of protections if the operating range includes both up and down thrust. Hydraulic thrust mainly depends on the hydrodynamic forces generated inside ESPs. These hydrodynamic forces depend on many factors like blade loading, seal geometries, seal diameters, seal engagement, and balance holes location and sizes. Not only is the thrust magnitude important to predict, but also the shape of the thrust curve plays an important role in defining the operating range. If the thrust curve is flat, then this means a wider operating range for the pump. Currently, no model is available to predict or design for thrust curve. Engineers use an iterative process of manufacturing and testing trying to reach the optimum seal configuration and balance holes geometry for better thrust and pump performance. In this paper, a detailed CFD model of a mixed flow multistage ESP is presented including all seal elements and balance holes to be able to predict hydraulic thrust. Thrust curve is predicted for the first time using CFD analysis. Different seal geometries and balance holes configuration is also modeled to study the effect of geometry change on thrust curve. The model is validated using experimental test results. Also, other design parameters like the shape of the head curve, pump efficiency, and gas handling capability were controlled early in the design phase using CFD analysis. Introduction Electrical submersible pump systems are being implemented extensively in many oil and gas wells for their capability of pumping hydrocarbons from low pressure or depleted reservoirs to the surface. ESP systems are composed of five main items; the bottom hole motor, bottom hole pump, main seal, cable, and a variable frequency drive (VFD). The motor is connected to the VFD through the cable that goes up to the surface and the motor drives the centrifugal pump to pump the hydrocarbons downstream. The ESP is a multistage pump that can be either radial or mixed flow pump. Each stage contains of a rotating 2 SPE-173945-MS impeller and a stationary diffuser that redirects the flow from the previous impeller outlet to the next impeller inlet. Although sand screens are used to prevent sand from going into production, fine sand particles can still make its way through the sand screens and go through the pump causing high erosion rates and leading to degradation in pump performance. Most failures in ESPs are due to sand erosion caused by sand production. Many areas inside the pump are affected by sand erosion, like the diffuser inner wall, impeller and diffuser blades. However, sand erosion at the sealing clearances of the skirt and balance ring seals is the main reason behind pump failures and loss of performance. Small sealing clearances would improve the pump performance by reducing leakage, but if sand particles are trapped in these small clearances, there is a high probability of widening the clearances and losing performance. CFD simulation has been used extensively by many researchers and engineers trying to predict the pump performance. However, due to the complexity of the geometry, the full pump model including the sealing diameters, sealing clearances, bearing clearances, and balance holes was not investigated before. Modeling and simulating the pump and predicting thrust under different seal clearances, and locations can be of great value to design more reliable pumps. Research on Pumps Modeling and Testing Modeling and understanding the behavior of fluids inside centrifugal pumps have been of great interest to many scientists and engineers starting the twentieth century. Researchers started investigating the centrifugal pump performance and came up with semi empirical models that can predict the pump performance such as the work done by Mikielewicz et al. (Mikielewicz 1978). In 1980, Zakem (Zakem 1980) presented a semi empirical formula that predicts a centrifugal pump performance. He based his model on both an analytical formulation coupled with experimental testing. Later in the 1990’s with the evolution of computers, many researchers started modeling pumps using computational fluid dynamics. Pak and Lee (Pak 1998) used CFD modeling to model bubbly flow inside centrifugal pump. It used one set of momentum, continuity and energy equations and used the gas volume fraction as a variable to model the mixture as one flow. Pirouzpanah et al. studied the ESP behavior with sand and two phase flow (Pirouzpanah 2013) & (Pirouzpanah 2014). Morrison et al. studied multiphase flow in twin screw pump both experimentally and theroeticall (Morrison 2012), (Morrison 2013), & (Morrison 2014). In a recent paper by Marsis et al. (Marsis 2013), the authors used CFD to model the single and two phase flow inside a multi-vane ESP. The CFD analysis was used to improve the original pump design and help achieving better gas handling and higher efficiency by changing the diffuser design. In their work, Marsis and Russell (Marsis 2013), presented the first erosion model for ESPs showing the erosion rate prediction inside the hydraulic path. This model utilized the CFD and modeled sand particles using both discrete phase model and Eulerian granular multiphase model to predict erosion rates inside ESPs. Seals Modeling and Testing Annular seals have been extensively used in turbmachineries to minimize gas and liquid leakage from high pressure zones to low pressure zones. Minimizing the leakage across annular seals improves the turbomachinery efficiency. There are mainly two types of seals; contacting and non-contacting seals. In ESPs, most of the annular seals used are non-contacting. Different geometries have been studied and compared to minimize the leakage across annular seals. Many experimental testing and computational fluid dynamics simulations were done to understand the fluid flow across the annular seals. Nelson and Nguyen (Nelson 1987) used a bulk flow model to analyze annular seals performance. Different rotordynamics investigations were performed to better understand the seal performance and stability like the studies shown by (Lucas 1994). Morrison et al. (Morrison 1992) studied the effect of whirling on the flow field inside a labyrinth seal and focused his study on the secondary recirculation zone using Laser Doppler Anemometry. Ustun et al. (Ustun 2013) performed a numerical analysis to understand the effect of surface roughness on convergent and smooth annular seals. Marsis and Morrison (Marsis 2013) performed a SPE-173945-MS 3 numerical study to compare leakage and rotordynamics coefficients of rectangular grooved, circular grooved, and smooth annular seals. Modeling seals together with the whole ESP is crucial as the performance of both the pump and the seal is interdependent. Thrust Effect on ESPs Unconventional wells are highly unpredictable, mainly characterized by gas slugs, frequent changes in flow rate and production of abrasives along with frac sand. These wells require very wide operating range for their ESPs due to the frequent change of production flow rates. An ESP performs optimally within a range from BEP. Any deviation outside this range changes up-thrust/down-thrust condition. Up-thrust or down-thrust is the force exerted by the pressure across impeller in axial direction. At low flow conditions, due to high down-thrust, the pump impeller may rub against the thrust bearing causing undesirable consequences to pump efficiency and reliability. Other causes of down-thrust conditions are wearing out of sealing surfaces and excessive gas in the well fluid. There are different design parameters that affect thrust loading. Thrust balancing is usually achieved by changing skirt diameter (front clearance) and balance ring (back clearance), changing the size, position and number of balance holes, and optimizing the hydraulic design. Thrust is very important parameter that affects pump reliability in production wells (Ye 2015). Hydrodynamic Modeling and Validation The ESP stage is simulated using a 3D model. The model includes the skirt seal which provides a seal between high pressure zone at the impeller exit and low pressure zone at the impeller inlet. The model also includes the balance ring seal which is the sealing element between the high pressure zone at the impeller exit and the balance chamber between the diffuser and the impeller hubs. The balance holes are included in the model. Hybrid mesh was used including mapped hexahedral elements and unstructured polyhedral elements. No tetrahedral elements are used as they are not good for convergence and accuracy criteria. Mesh independence study was performed. A single stage model has 3.5 to 4 million nodes. Standard k-epsilon model was used. The y⫹ value was kept within the acceptable range. Transient analysis was implemented and the impeller motion was simulated as well. The K-epsilon model is an eddy viscosity model that is based on Reynolds Averaged Navier-Stokes equation (RANS). The continuity and momentum equations are given as follows: (1) (2) The working fluid used in both the simulation and testing was water. The model used is the one presented by Marsis et al. (Marsis 2013). The head curve of the pump was generated using the CFD model and was compared with the actual test of the stages that were simulated in these paper. All the cases showed very good agreement between the simulation and the test with 3% maximum error. Two full rotations of the pump were simulated transiently to ensure the full development of the flow inside the pump. Fig. 1 shows a 2D cross-section of the 3D model showing all clearances as well as the main hydraulic path. A 3D model is shown in Fig. 2 showing the impeller, diffuser, and the fluid domain behind the shroud of the impeller. 4 SPE-173945-MS Figure 1—Cross-section of a 3D ESP model showing all clearances and the main hydraulic path Figure 2—3D model showing the diffuser and impeller including the shroud on the back of the impeller Thrust model validation In this paper, thrust of ESP stages is being modeled for the first time. The model was first applied on small 338 series pump stage that is designed and manufactured by Baker Hughes Inc. This stage has a BEP at 610 BPD. In order for this stage to perform as a Flex stage, it had to be designed for a wide operating range to meet the challenging requirements for unconventional wells. In order to have a wide operating range, the head curve has to be continuously rising within the operating range with no dips, flat spots, or droop. Also, thrust must not exceed a certain limit in order to be handled by thrust bearings. Also the minimum the thrust the better as it affects the reliability, efficiency, and the run life of the ESP. Thrust was modeled early in the design phase and the size of the balance ring seal, skirt seal, and balance holes location and size were optimized. Fig. 3 shows the thrust prediction from the CFD model compared to the actual thrust tested for the 338 series pump stage at different balance holes numbers. Results show very good agreement between the actual test and the CFD predictions. In some specific ESP stages, since the first stage performance is significantly different from the second stage performance in order to achieve constant rising head curve, SPE-173945-MS 5 two stages had to be simulated to accurately predict head and thrust for the second stage as seen in Fig. 4. Figure 3—338 series stage thrust prediction vs. test results Figure 4 —2D view from the 3D model showing 2 stages simulation for 338 series pump for thrust prediction Simulation-Based Design Different geometric configurations were studied extensively to understand the thrust behavior and design for a flat low thrust curve throughout the operating range of ESPs. These simulations showed that the diameters of the balance ring and skirt seals affect thrust curve significantly by shifting the whole curve up or down. Also widening the balance holes in an impeller shift the thrust curve down and cause hydraulic efficiency to go down as well due to fluid circulation. However, changing the blade loading on an impeller was used to change the shape of the thrust curve and was very effective to make the thrust curve flat for wider operating range. A 400 series stage was designed for optimum hydraulic performance using CFD-based design techniques. The test matched the predicted performance for head, thrust and efficiency. Fig. 5 shows the predicted thrust curves early in the design phase for this stage with different balance holes sizes and balance ring diameters in order to obtain a flat and low thrust curve. 6 SPE-173945-MS Figure 5—CFD-Based thrust design for 1900 BPD 400 series stage Computational Fluid Dynamics are used not only to predict and control thrust, but also to predict and control other factors that affect the ESP performance. The shape of the head curve is crucial in determining the pump operating range. Having a constant rising head curve for an ESP stage within the operating range is a great advantage as it allows the operator to accurately control the flow rate of the pump and avoid running in unstable conditions. The 338 series-610 BPD pump stage was designed to have high head at shutoff, and yet maintain a constant rising head from 0 BPD up to the maximum flow rate as seen in Fig. 6. Fig. 7 shows another 538 series pump that is designed using CFD simulations and it shows a high head per stage and yet a constant rising head with no dips or flat spots. At 11 KBPD, it is hard to design a stage with no dips as this has been a challenging flow condition in the industry. However, by controlling the blade loading on the impeller and diffuser, a constant rising head was achieved. Figure 6 —338 series tested pump curve SPE-173945-MS 7 Figure 7—538 series pump performance test matching CFD prediction showing constant rising high head Two phase flow simulation was also used as part of simulation-based design in order to ensure the best performance of ESP stages in gassy wells. Two phase flow simulation model presented in Marsis et al. (Marsis 2013), was used to visualize the gas pockets and gas accumulation inside impellers. Blade loading of the impeller was changed and optimized in order to minimize the gas pocket accumulation between the impeller blades as seen in Fig. 8. Figure 8 —Gas volume fraction in two different impellers; (a) Before changing blade loading, (b) After changing blade loading Conclusion This study shows that CFD modeling was used to accurately predict and control thrust of ESPs for the first time. This was based on transient CFD analysis with high quality mesh. This technique is developed to minimize testing time and cost. This model allows pump companies to design their stages, optimize the blade loading, define the right dimensions of seals and balance holes early in the design phase so the pump would meet the expected performance from the first iteration. CFD based design was implemented not only on thrust, but also on other factors that affect the pump performance like shape of the head curve, and gas handling. CFD based design is very useful in predicting and controlling the shape of the head curve to ensure a wide operating range for the ESP stages. Also gas pockets were visualized and eliminated early in the design phase of a new stage development. This new design technique was developed to meet the challenging requirements in many unconventional wells and improve the run life of the ESP systems. 8 SPE-173945-MS Therefore CFD-based design takes pump design to a new level. It allows pump manufacturers to control all aspects of performance early in the design phase. Stage thrust, the shape of the head curve, and gas handling are no more unpredictable or uncontrollable. This technique allows engineers to design stages that will meet the challenging market need and challenging applications in artificial lift industry. References Mikielewicz, J., G. Wilson, D. Chan, and A. Goldfinch, 1978, A Method for Correlating the Characteristics of Centrifugal Pumps in Two-Phase Flow. Journal of Fluid Engineering, 395–409 Zakem, S., 1980, Determination of Gas Accumulation and Two-Phase Slip Velocity Ratio in a Rotating Impeller. Journal of Fluids Engineering, 102, 446 –455 Pak, E. T., and J. C. Lee, 1998, Performance and Pressure Distribution Changes in a Centrifugal Pump under Two Phase Flow. Journal of Power and Energy, 165–171 Pirouzpanah, S., Morrison, G. L., 2014, Predictive Erosion Modeling in an ESP Pump, ASME 2014 4th Joint US-European Fluids Engineering Summer Meeting and 12th International Conference on Nanochannels, Microchannels, and Minichannels, Chicago, Illinois, August 3–7. Pirouzpanah, S., Gudigopuram, S. R., Morrison, G. L., 2014, Flow Characterization in an ESP pump Using Conductivity Measurements, ASME 2014 4th Joint US-European Fluids Engineering Summer Meeting and 12th International Conference on Nanochannels, Microchannels, and Minichannels, Chicago, Illinois, August 3–7. Pirouzpanah, S., Morrison, G. L., 2013, Temporal Gas Volume Fraction and Bubble Velocity Measurement Using an Impedance Needle Probe, ASME 2013 Fluids Engineering Division Summer Meeting, FEDSM2013, Incline Village, Nevada, July 7–11. Morrison, G., Ryan Kroupa, Abhay Patil, Jun Xu, Stuart Scott, Sven Olson, 2014, Experimental Investigation of Wellhead Twin-Screw Pump for Gas-Well Deliquefication, Oil and Gas Facilities Journal, SPE, April 2014. Morrison, G., Pirouzpanah, S., Cevik, M., Patil, A., 2013, Evaluation of a Close Coupled Slotted Orifice, Electrical Impedance, and swirl flow meters for multiphase flow, Proceedings of the ASME FEDSM2013. Morrison, G., Patil, A., Cihak, D., 2012, Evaluation of a Twin Screw Pump for Use in High Gas Volume Fraction Flows, ASME Summer Meeting 2012 Puerto Rico. Morrison, G., Patil, A., Cihak, D., 2012, Efficiency limit for multiphase pumps, 8th North American Conference on Multiphase Technology (BHR Group), Banff, Canada, 20 –22 June. Marsis, E., Pirouzpanah, S., Morrison, G., 2013, CFD-Based Design Improvement for Single-Phase and Two-Phase Flows Inside an Electrical Submersible Pump, Proceedings of the ASME FEDSM2013 Marsis, E., Russell, R., 2013, A State-of-the-Art Computational Fluid Dynamics Simulation for Erosion Rates Prediction in a Bottom Hole Electrical Submersible Pump, SPE Heavy Oil Conference, Calgary, Alberta, Canada. Nelson, C. C., Nguyen, D. T., 1987, Comparison of Hirs’ Equation with Moody’s Equation For Determining Rotordynamic Coefficients of Annular pressure Seals, ASME J. Tribology, Vol. 109, pp. 144 –148. Lucas, V., et alet al., 1994, Roughness Influence on Turbulent Flow through Annular Seals, ASME J. Tribology, 1994, Vol. 116, pp. 321–329. Morrison, G.L., Johnson, M.C. and Tatterson, G.B., 1992, Experimental Verification of a Secondary Recirculation Zone in a Labyrinth Seal, AIAA Journal of Propulsion and Power, Vol. 8, pp. 1064 –1070. SPE-173945-MS 9 Ustun, S., Marsis, E., Morrison, G., Numerical Study of Straight Annular Seals Versus Convergent Annular Seals at Different Geometries, Surface Roughness, and Operating Conditions, ASME 2013 International Mechanical Engineering Congress and Exposition Marsis, E., Morrison, G., 2013, Leakage and Rotordynamics Numerical Study of Circular Grooved and Rectangular Grooved Labyrinth Seals, ASME Turbo Expo 2013 Ye, Z., Semple, R., and Rafie, S., 2015, Approach for Assessing Dormant Reliability in Dual Electrical Submersible Pump (ESP) Systems in High-pressure, High-temperature (HPHT) Environments, Proceedings of the 2015 Offshore Technology Conference, Houston, TX.