An Experimental Study of Surfactant Effect on Gas

Proceedings of the ASME 2017 International Mechanical Engineering Congress and Exposition IMECE2017 November 3-9, 2017, Tampa, Florida, USA

IMECE2017-70165

AN EXPERIMENTAL STUDY OF SURFACTANT EFFECT ON GAS TOLERANCE IN ELECTRICAL SUBMERSIBLE PUMP (ESP) Jianjun Zhu McDougall School of Petroleum Engineering The University of Tulsa Tulsa, OK, 74104 [email protected]

Haiwen Zhu McDougall School of Petroleum Engineering The University of Tulsa Tulsa, OK, 74104 [email protected]

Jiecheng Zhang McDougall School of Petroleum Engineering The University of Tulsa Tulsa, OK, 74104 [email protected]

Hong-Quan Zhang McDougall School of Petroleum Engineering The University of Tulsa Tulsa, OK, 74104 [email protected]

ABSTRACT An experimental study on ESP boosting pressure under airwater flow with/without surfactant injection is presented. The experimental facility comprises of a 3-inch-diameter stainless steel liquid loop and ½-inch-diameter gas loop. A radial-type ESP with 14 stages assembled in series was installed in the testing bench. Pressure ports were drilled at inter-stage to measure the stage-by-stage boosting pressure. Surfactants, isopropanol (IPA) were injected to change interfacial properties of working fluids. Experiments were carried out with mapping and surging test schemes to evaluate pump behaviors at different operational conditions. ESP pressure increment under singlephase water flow agrees well with manufacture curves. For mapping tests without surfactant injection, ESP performance suffers from a severe degradation as gas flow rate increases. High gas entrainment rate causes oscillations of liquid flow rate and pump boosting pressure. A sudden drop of ESP pressure increment, termed as pressure surging, occurs at the critical inlet gas volumetric fraction (GVF). At higher rotational speeds, the critical GVF is higher. With surfactant injection, ESP boosting pressure improves significantly. With different GVFs, only mild degradation was observed. Pressure surging phenomenon disappeared. Further, liquid flow rate and pump boosting pressure are more stable at high GVFs compared to experimental data without surfactant injection.

Np Psep QBEP or Qbep QG Qgd QL T t xv ρ

normalized stage pressure increment separator pressure [psi] liquid flow rate at best efficiency point [bpd] gas flow rate [cfm] dimensionless gas flow rate liquid flow rate [bpd] temperature in K temperature in °C mole fraction density of the fluid [kg/m3]

INTRODUCTION As a widely used artificial lift method in offshore oil production, ESP excels in boosting hydraulic pressure and flow rates of hydrocarbon fluids. Its performance deteriorates with gas entrainment. Previous studies [1-6] show that the presence of gas causes ESP pumping head degradation, pressure surging and gas pocket formation. Surging may result in vibrations and short service life of ESPs, while gas pockets may reduce production rates greatly. Injected or natural surfactants may improve ESP performance under gassy conditions. Surfactants can extend ESP operating envelope, stabilize liquid flow rates, avoid pressure surging or gas locking. Murakami and Minemura [7] reported experimental observations of entrained air bubble behavior in a centrifugal pump using a semi-opened impeller pump with a transparent casing. The decrease of total pump head caused by air admission and the work consumed for air delivery were studied. Since then, many experimental studies on gas-liquid flow in centrifugal

NOMENCLATURE GVF gas volumetric fraction [%] Ma molar mass [kg/mol] N rotational speed [rpm]

1

Copyright © 2017 ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/13/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use

pumps have been conducted [8-13], including effects of gas entrainment rate [8, 11], impeller blade angle [10], and transient pump behavior [9] as well as visualizations of internal flow structures [12, 13], among others. Due to complex pump geometries involved with rotation, most of existing experimental studies focused on the measurement of pump boosting pressure and direct observations of internal flow patterns. Aided with Gamma-ray CT (computed tomography) technology, the distributions of gas phase in a rotating centrifugal pump impeller can be quantified [14]. Surfactants are molecules with a hydrophobic and a hydrophilic part, and therefore they preferentially adsorb at the interface of continuous/dispersed phases. In this process, they reduce the surface tension of the continuous phase [15]. Since surfactants naturally exist in crude oil production, their effects on liquid loading in gas well have been studied [16, 17]. However, studies on surfactant effects on centrifugal pump performance with gas-liquid flow are very few. In this study, an experimental study of surfactant effect on ESP gas-liquid performance is conducted. A 14-stage ESP was

used for experimental testing. The air-water mixture with three different volumetric concentrations of surfactants was flowed through ESP, where the stage-by-stage boosting pressure was measured. To evaluate pump performance under gas-liquid flow conditions, tests were carried out with pressure surging test (constant liquid flow rate) and performance mapping test (constant gas flow rate). The experimental results indicate a significant improvement of ESP gas-handling ability with surfactant injections. EXPERIMENTAL FACILITY The experimental facility for testing ESP performance comprises of a closed liquid flow loop and an open loop for gas flow. The maximum loop capacity is QL = 4,900 bpd (barrels per day) and QG = 186 cfm (cubic feet per minute) at 217 psig. The schematic of the facility layout is shown in Fig. 1, in which the black solid lines represent pipelines for gas/liquid flow, and the grey dashed-lines are sensor signal wires. The arrows marked on lines represent the flow directions of fluids or signals.

Figure 1 Schematic of the two-phase loop of testing ESP at Tulsa University Artificial Lift Project (TUALP) The liquid flow loop made of 3-inch stainless-steel pipe is connected to a horizontal separator. The separator pressure is controlled by a relief valve so that the ESP intake pressure can be regulated. A Coriolis flowmeter (Micro Motion CMF200) is mounted upstream of ESP to measure the liquid mass flow rate and density. Pressurized by an air compressor (Kaeser CSD60), the gas flow rate is regulated by the control valve in gas flow loop. The air mass flow rate is measured by a Coriolis flowmeter (Micro Motion CMF025). The in-situ air density at ESP intake is calculated using correlations for wet air properties based on

the local pressure and temperature [18]. Air is mixed with water inside a T-junction prior to the inlet of the ESP. Before mixing, the chemical surfactants (IPA) can be injected by a metering pump (Iwaki EWN-R) so that the fluid interfacial properties can be changed. After the fluids flow through the ESP testing bench and downstream control valve, the mixture is discharged into the horizontal two-phase separator, where the excessive air is vented to the atmosphere and the liquid continues circulation.

2



(a)

best efficiency point (BEP) is 2700 bpd at 3500 rpm with a stage pressure increment of 22.5 psi. The stage-by-stage absolute pressure (AP) and differential pressure (DP) are measured by AP transducers (Rosemount 2051S) and DP transducers (Rosemount 3051S). The sectional front view of TE-2700 ESP and the mounting positions of pressure sensors are presented in Fig. 2 (a) and (b). Fig. 2(c) shows the schematic of intake zone inside the ESP, where the pressure port is located between the upstream diffuser and downstream impeller. Similarly, the other pressure ports at downstream stages are drilled. On the lateral housing surface of each diffuser, the peripheral notch is grooved to install Teflon O-rings to block fluid communication between stages. ESP is driven by a four-pole 50 hp electric motor (North American H3650) via a variable speed drive (VSD, Hitachi L300P). At the pump discharge, a 3-inch pneumatic valve is mounted to control the liquid flow rate. ESP and electric motor are connected to the thrust chamber which holds thrust force and allows the shaft rotating. The rotational speed and shaft torque are measured by a torque cell (Lebow 1805K). Based on the compact FieldPoint module (National Instrument, NI), the data acquisition system (DAS) is programmed in LabVIEW V2014. The analog signals from field terminals (pressure transducers, temperature transmitters, flowmeters, etc.) are connected to input modules (NI cFP-AI-111), where the electric current signals from sensors can be scaled up into engineering units. The control signals from output module (NI cFP-AO-200) are sent to field terminals (control valves, VSD, etc.) to regulate ESP operational status. All signals are transferred to data process computer using RS-485 portal by Ethernet cable. A computer (DELL Optiplex 9020) serves as the data processing center. Experiment tests are conducted under both liquid and gasliquid flow conditions. In liquid flow tests, the H-Q curve is obtained at a constant separator pressure and pump rotational speed. By reducing the opening of a control valve downstream, the liquid flow rate is regulated. The pressure increment data at each stage are recorded by DAS. In gas-liquid tests, both surging and mapping tests are conducted. For surging tests, the liquid flow rate is kept constant, while the gas flow rate increases continuously from 0 to the maximum flow rate, at which the ESP approaches nil performance. For mapping tests, the gas flow rate is kept constant, while the liquid flow rate reduces from the maximum open flow rate to the minimum value corresponding to the shut-in of downstream control value.

(b)

RESULTS AND DISCUSSIONS This section presents the experimental test results of ESP performance under both single-phase water flow and air-water two-phase flow with and without surfactant injections. The comparison is carried out considering effects of pump intake pressure and the volumetric concentrations of surfactant.

(c) Figure 2 ESP testing bench, (a) sectional front view, (b) top view of pressure measurement ports, (c) schematic of quick connector and mounting location

Single-phase Test Results Tap water and compressed air are used in this study as the working fluids. The surfactant of isopropyl alcohol (IPA) is injected into the flow loop to reduce the interfacial properties

The ESP testing bench consists of the ESP, motor, thrust chamber and other equipment that are needed for ESP operation. The studied ESP is a 14-stage TE-2700 series 538 pump. The

3



between water and air. To verify the facility, single-phase water flow performance obtained by experimental measurements are compared with corresponding catalog curves. As can be seen in Fig. 3, each color corresponds to a different pump rotational speed. The black solid curve at N = 3500 rpm is obtained directly from catalog, and the other catalog curves are calculated with the affinity law. Data points represent experimental measurements. The good agreement with less than ±5% deviation between experiments and catalog curves verifies the testing loop in this study.

Figure 4 Surface tension of air/water versus IPA volumetric concentration Figure 5 presents the solution density as function of the volumetric concentration of IPA. A linear regression with good agreement with the experimental measurements can be observed. Compared to the literature data [21], the measured solution density is slightly higher.

Figure 3 Experimental performance curves compared with catalog curves Surfactant Properties The selected surfactant isopropanol alcohol (IPA) is a harmless and environment-friendly chemical in the alcohol family. With surfactant injection, the surface tension between air and water is reduced significantly before a critical point called CMC (critical micelle concentration) is achieved. In this study, three concentrations of IPA are used to test ESP gas-liquid performance, i.e. 0 vol%, 0.2 vol% and 0.41 vol%, corresponding to 8-hour and 16-hour injection periods, respectively. Fig. 4 shows the comparison of measured surface tension of air/water versus IPA volumetric concentration against the literature data, including model predictions [19] and experiment measurements [20]. The surface tension is measured with a digital tensiometer (Attension Sigma-700). The good agreement with less than ±10% error is found between the empirical correlation and experimental data from Ref. [20]. The deviation of experimental measurements obtained in this study with the model predictions is probably due to the contamination of tap water used in our flow loop compared to distilled water used in Ref. [19].

Figure 5 Solution density versus IPA volumetric concentration Air-water Test Results without IPA Fig. 6 shows the surging test results of air-water flow at stage 3 with different liquid flow rates (0.75QBEP, 1.0QBEP, 1.25QBEP), rotational speeds (1800, 3500 rpm) with Psep = 150 psig. As seen, the pump stage pressure increment decreases with intake GVF increase, even to zero head. However, the decline trend of the measure stage pressure increment versus GVF is not in a linear relationship. The head curve slopes dramatically change at a certain GVF, before which the ESP stage pressure increment has mild deterioration. This phenomenon is termed as

4



pressure surging in Ref. [8]. The critical GVF is called as the initiation of pressure surging, which is an important concept in ESP gas-liquid flow. After the critical GVF, the ESP boosting pressure suffers from severe degradations.

Stage pressure increment (psi)

30

denote the liquid flow rate QL and stage pressure increment, respectively. Two different rotational speeds (1800, 3500 rpm) with different Qgd are tested. Qgd is the dimensionless gas flow rate with respect to the maximum open flow rate of ESP at N = 3500 rpm. As seen in Fig. 8, the mapping curves for water flow are affected by gas and liquid flow rates significantly. The boosting pressure of ESP increases as the liquid flow rate is reduced from the open flow rate to a certain QL. If the liquid flow rate is less than the critical QL, the deviation of the measured pump heads with air-water flow from the single-phase water performance curve increases significantly.

1800rpm_0.75Qbep 1800rpm_1.0Qbep 1800rpm_1.25Qbep 3500rpm_0.75Qbep 3500rpm_1.0Qbep 3500rpm_1.25Qbep

25 20 15 10 5 0 0

4

8

12

16

20

Intake GVF (%) Figure 6 Surging tests of air-water at stage 3, Psep = 150 psig

Figure 8 Mapping tests of air-water at stage 3, Psep = 150 psig Air-water + 0.41 vol% IPA Test Results The surging test results of air-water flow with 0.41 vol% IPA at stage 3 are shown in Fig. 9. As can be seen, within the experimental measurement range of intake GVF, the pressure surging disappears since the measured head curves decline less compared to that without surfactant injection (Fig. 6). A significant improvement of pump head with surfactant presence is observed at relatively high GVFs. Take N = 3500 rpm, QL = 0.75QBEP for example, the stage pressure increment at GVF = 15% is about 24 psi and 11 psi for air-water flow with and without surfactant injections, respectively. The increment ratio of ESP performance under gas-liquid flow is more than 110%. For some cases with severe gas entrainment, the pump performance improvement with surfactant injections will be even higher. The effects of stage number become more evident in gasliquid flow with surfactant presence. As Fig. 7 shows, the head curves of water flow for different stages are close to each other. In contrast, with surfactant presence, they are apart from each other in Fig. 10. Similar to Fig. 9, the measured head curves with surfactant presence render a much more linear relationship against GVF at different stages. The sudden drop of pump head at a certain value of GVF, termed as pressure surging disappears.

Figure 7 Surging tests of air-water at different stages with N =3500 rpm, Psep = 150 psig, QL = QBEP Figure 7 shows the stage effects on ESP boosting pressure under air-water flow conditions. The vertical axis Np stands for the normalized pressure increment, which is defined as the ratio of stage pressure increment under gas-liquid flow to the pressure increment under sing-phase water flow. It shows that the upstream stages have an earlier surging initiation corresponding to lower GVFs. On the contrary, surging initiation at downstream stages is delayed to higher GVFs. Moreover, the downstream stages (6 or 7) possess better pump performance compared with that at upstream stages (3 or 4). The mapping test results of air-water at stage 3, Psep = 150 psig is presented in Fig. 8, where the horizontal and vertical axes

5



Stage pressure increment (psi)

35

1800rpm_0.75Qbep 1800rpm_1.0Qbep 1800rpm_1.25Qbep 3500rpm_0.75Qbep 3500rpm_1.0Qbep 3500rpm_1.25Qbep

30 25 20 15 10 5 0 0

4

8

12

16

20

Intake GVF (%) Figure 9 Surging tests of air-water + 0.41 vol% IPA at stage 3, Psep = 150 psig

Figure 11 Mapping tests of air-water + 0.41 vol% IPA at stage 3, Psep = 150 psig Comparison of Experimental Results This section compares the experimental results obtained under different flow conditions with/without surfactant injections. Figure 12 presents the effect of intake pressure on surging test results. The flow conditions are N = 1800 rpm, QL = QBEP with variable Psep. As it can be seen, a higher Psep corresponding to a higher pump intake pressure, resulting in the slightly better ESP performance for relatively higher GVFs. For instance, at GVF ≈ 6%, the measured Np is about 0.5, 0.6, and 0.7 for Psep = 50, 100, and 150 psig, respectively. The reason for better ESP performance at higher Psep may be due to larger gas density but smaller gas bubble sizes if the intake pressure is higher. Moreover, the pressure surging also initiates at a higher intake GVF with the increase of Psep. For example, the surging initiates at GVF ≈ 3.5% for Psep = 50 psig, but GVF ≈ 5.5% for Psep = 150 psig in Fig. 12.

Figure 10 Surging tests of air-water + 0.41 vol% IPA at different stages with N =3500 rpm, Psep = 150 psig With the injection of surfactant, the two-phase H-Q curves exhibit a distinct behavior compared to that without surfactant present. Fig. 11 shows the mapping test results at stage 3 with 0.41 vol% IPA and Psep = 150 psig. As can be seen, the deflections on two-phase H-Q curves are close to QL = 1100 bpd. Meanwhile, the H-Q curves at different Qgd are closer to the performance curve with single-phase water flow especially at relatively low QL, indicating the significant improvements of ESP gas-handling ability due to surfactant injection. For relatively higher QL, the two-phase H-Q performance curves are just as close to water flow performance without surfactant present (see Fig. 8). This is ascribed to the dispersed bubble flow pattern prevailing if QL is very high. Comparing Fig. 11 and Fig. 8, the injection of surfactant not only improves ESP stage pressure increment under gas-liquid flow but also stabilizes the ESP operations.

Figure 12 Effect of intake pressure on surging tests under air-water flow at N = 1800 rpm without IPA

6



As mentioned above, the surfactant injection causes a significant improvement of ESP gas-handling ability and boosting pressure. The effect of surfactant concentration on the ESP normalized pressure increment (Np) is shown in Fig. 13. Correspondingly, the experimental test conditions for Fig. 13 are at N = 3500 rpm, Psep = 100 psig with variable volumetric concentrations of IPA. As it is shown, the decline trend of N p versus intake GVF becomes linear with surfactant injections, indicating that the pressure surging phenomenon disappears. Meanwhile, Np is much larger for air-water with 0.2 vol% IPA and 0.41 vol% IPA compared to that without IPA presence. The black dashed circle in Fig. 13 depicts the zone of the obvious ESP performance improvement due to the surfactant injections. Another observation in Fig. 13 is that Np increases slightly with the increment of IPA volumetric concentration, especially at relatively high GVFs. However, the discrepancy of Np at low GVFs is not obvious among the test results with different fluids due to the dispersed bubble flow prevailing if the intake GVF is small.

Figure 14 Effect of intake pressure on ESP performance curves under air-water flow at N = 1800 rpm with no IPA Significant improvement of ESP head due to surfactant injection

Significant improvement of ESP head due to surfactant injection

Figure 15 Effect of IPA concentration on ESP performance curves at Psep = 100 psig, N = 3500rpm, Qgd = 0.02

Figure 13 Effect of IPA concentration on surging initiation at Psep = 100 psig, N = 3500rpm

Figure 15 compares ESP H-Q performance curves under airwater flow with different IPA volumetric concentrations. The operation conditions in Fig. 15 are at N = 3500 rpm, Psep = 100 psig, Qgd = 0.02. A clear difference can be seen at low QL in terms of the stage pressure increment between surfactant injection and no surfactant injection cases. Without IPA injection, the pressure increment drops to zero if liquid flow rate becomes very low. However, the ESP performance improve significantly with surfactant presence. The black-dash oval depicts the significant improvement of ESP boosting pressure due to surfactant injections. Within the measurement range, the ESP two-phase H-Q curves with different IPA concentrations are very close to each other. It can be inferred that the significant improvement of ESP performance under air-water flow is not only due to the drag reduction effect of surfactants but also due to the formation of

Figure 14 presents the effect of intake pressure on ESP H-Q performance curves under air-water flow without surfactant injection. As a critical factor that affects gas density, the intake pressure should always be accounted when calculating the in-situ gas properties when gas flowing through the multistage ESP. For comparison, the flow conditions are fixed at Qgd = 0.01, N = 1800 rpm, air-water flow with no IPA injection. As it can be seen, the increase of Psep postpones the deviation trend of gasliquid H-Q curve from the corresponding water performance curve to lower QL. At higher QL before the H-Q performance deviation, the influence of Psep on ESP stage pressure increment is small. Moreover, Psep increase leads to a slightly better ESP performance especially at relatively lower QL.

7



foamy flow [22-23] at relatively low QL and dispersed bubble flow at high QL. It also verifies van Nimwegen et al. study [16] on the surfactant effects on gas-liquid two-phase flow characteristics. The dynamic surface tension, coupled with Marangoni flow effects, is more related to the formation and stability of foam, which reduces gas-liquid slippage and changes the morphology of the gas-liquid interface [24]. Therefore, the ESP air-water two-phase flow behaviors are altered significantly with surfactant injections.

[3]

[4]

[5]

CONCLUSIONS In this paper, the experimental measurements of pump boosting pressure under liquid and gas-liquid flow conditions are conducted on a 3-inch two-phase flow loop with a 14-stage radial-type ESP. The stage-by-stage pressure increment with varying flow conditions is measured. Effects of intake pressure, GVF, rotational speeds, and surfactant presence on the ESP pressure increment are investigated. Two schemes of experimental testing are carried out to evaluate the pump behavior at different operational conditions, including surging tests (constant liquid flow rate) and mapping tests (constant gas flow rate). The conclusions are drawn as below: 1. The measured ESP stage pressure increment with tap water flow matches catalog performance curves well. This validates the experimental setup used in this study. 2. For surging tests, experiments are repeated using air-water mixture with three different surfactant concentrations. As gas flow rate increases, the stage pressure increment suffers from mild degradation to a sudden drop head corresponding to pressure surging. After surfactant is added, the initiations of pressure surging are postponed to higher GVFs. 3. For mapping tests, the gas volumetric flow rates are fixed, but the liquid flow rates vary in a broad range. A sudden drop on the two-phase H-Q curve occurs when liquid flow rate is reduced to a certain value for air-water flow without surfactant. In contrast, the surfactant presence significantly improves ESP two-phase stage pressure increment by postponing the sudden drop to lower liquid flow rate. 4. The small changes in the two-phase H-Q performance curves with different surfactant concentrations indicate that the improvement of ESP boosting pressure is not only due to the reduction of surface tension, but also due to the formation of foam flow, which changes the morphology of the gas-liquid interface significantly.

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

ACKNOWLEDGMENTS The authors appreciate the technical and financial support of the Tulsa University Artificial Lift Projects member companies.

[15] [16]

REFERENCES [1] Barrios, L., 2007. Visualization and modeling of multiphase performance inside an electrical submersible pump. PhD Dissertation, the University of Tulsa, Tulsa, OK. [2] Gamboa, J., 2008. Prediction of the transition in two-phase performance of an electrical submersible pump. PhD dissertation, the University of Tulsa, Tulsa, OK.

[17]

[18]

8

Zhou, D., Sachdeva, R., 2010. Simple model of electric submersible pump in gassy well. Journal of Petroleum Science and Engineering, 70, 204-213. Zhu, J., and Zhang, H.-Q., 2014. CFD simulation of ESP performance and bubble size estimation under gassy conditions. In SPE Annual Technical Conference and Exhibition, Society of Petroleum Engineers, Amsterdam, the Netherlands, October, 2014. Zhu, J., and Zhang, H.-Q., 2016. Mechanistic modeling and numerical simulation of in-situ gas void fraction inside ESP impeller. Journal of Natural Gas Science and Engineering, 36, 144-154. Zhu, J., and Zhang, H.-Q., 2017. Modeling of gas bubble size in electrical submersible pump (ESP) through numerical simulation. SPE Production & Operations, 2017. Murakami, M. and K. Minemura., 1974. Effects of entrained air on the performance of a centrifugal pump (1st report, performance and flow conditions). Bulletin of the JSME, 17(110): 1047-1055. Lea, J. F. and Bearden, J., 1982. Effect of gaseous fluids on submersible pump performance. Journal of Petroleum Technology, 34: 2922-2930. Narabayashi, T., Arai, K., Kubokoya, T., Amano, O., and Gomyo, T., 1986. Centrifugal pump behavior in steady and transient two-phase flow. Journal of Nuclear Science and Technology, 23(2): 136-150. Sato, S., Furukawa, A. and Takamatsu, Y. 1996. Air-water two-phase flow performance of centrifugal pump impellers with various blade angles. JSME International Journal. Ser. B, Fluids and Thermal Engineering, 39: 223-229. Pessoa, R., 2001. Experimental investigation of two-phase flow performance of electrical submersible pump stages. Master thesis, the University of Tulsa, Tulsa, OK. Barrios, L., and Prado, M. G., 2011. Experimental visualization of two-phase flow inside an electrical submersible pump stage. ASME Journal of Energy Resources Technology, 133(4): 042901. Verde, W.M., Biazussi, J.L., Sassim, N.A. and Bannwart, A.C., 2017. Experimental study of gas-liquid two-phase flow patterns within centrifugal pumps impellers. Experimental Thermal and Fluid Science, 85: 37-51. Schäfer, T., Bieberle, A., Neumann, M. and Hampel, U., 2015. Application of gamma-ray computed tomography for the analysis of gas holdup distributions in centrifugal pumps. Flow Measurement and Instrumentation, 46: 262267. DeGennes, P-G., Brochard-Wyart, F., and Quere, D., 2004. Capillarity and Wetting Phenomena. New York: Springer. Van Nimwegen, A.T., Portela, L.M. and Henkes, R.A., 2015. The effect of surfactants on vertical air/water flow for prevention of liquid loading. SPE Journal, 4: 488-500. Ajani, A., Kelkar, M., Sarica, C. and Pereyra, E., 2016. Effect of surfactants on liquid loading in vertical wells. International Journal of Multiphase Flow, 83:183-201. Davis, Richard S., 1992. Equation for the determination of the density of moist air (1981/91), Metrologia, 29(1): 67-70.



[19] Meissner, H.P. and Michaels, A.S., 1949. Surface tensions of pure liquids and liquid mixtures. Industrial & Engineering Chemistry, 41(12): 2782-2787. [20] Hu, B., Nienow, A.W., Stitt, E.H. and Pacek, A.W., 2006. Bubble sizes in agitated solvent/reactant mixtures used in heterogeneous catalytic hydrogenation of 2-butyne-1, 4diol. Chemical Engineering Science, 61(20): 6765-6774. [21] Chu, K.Y. and Thompson, A.R., 1962. Densities and refractive indices of alcohol-water solutions of n-Propyl, Isopropyl, and Methyl Alcohols. Journal of Chemical and Engineering Data, 7(3): 358-360. [22] Van Nimwegen, A.T., Portela, L.M. and Henkes, R.A.W.M., 2015. The effect of surfactants on air-water annular and churn flow in vertical pipes. Part 1: morphology of the airwater interface. International Journal of Multiphase Flow, 71: 133-145. [23] Van Nimwegen, A.T., Portela, L.M. and Henkes, R.A.W.M., 2015. The effect of surfactants on air-water annular and churn flow in vertical pipes. Part 2: Liquid holdup and pressure gradient dynamics. International Journal of Multiphase Flow, 71: 146-158. [24] Khosla, V., 2012. Visual investigation of annular flow and the effect of wall wettability. M.Sc. thesis, Delft University of Technology. [25] Zhu, J., Guo, X., Liang, F. and Zhang, H.-Q., 2017. Experimental study and mechanistic modeling of pressure surging in electrical submersible pump. J Nat Gas Sci Eng (http://dx.doi.org/10.1016/j.jngse.2017.06.027).

9



ANNEX A CALCULATION EQUATIONS FOR AIR PROPERTIES Table A.1 Constants in calculating air properties Parameters Constants Values A 10-5 K-2 1.2811805 B 10-2 K-1 -1.9509874 Saturation vapor pressure pSV C 34.04926034 D 103 K -6.3536311 α 1.00062 Enhancement factor f β 10-8 Pa-1 3.14 γ 10-7 K-2 5.6 a0 10-6 K Pa-1 1.62419 a1 10-8 Pa-1 -2.8969 a2 10-10 K-1 Pa-1 1.0880 b0 10-6 K Pa-1 5.757 Compressibility factor b1 10-8 Pa-1 -2.589 Z c0 10-4 K Pa-1 1.9297 c1 10-6 Pa-1 -2.285 d 10-11 K2 Pa-2 1.73 e 10-8 K2 Pa-2 -1.034 Gas constant R R J mol-1 K-1 8.31441 -3 -1 -1 10 kg K J Ma (𝑥𝐶𝑂2 = 0.0004)/R Ma R 3.48353 The moist air density can be calculated based on Comite International des Poid et Measures (CIPM-81) formulas [18]. The CIPM-81 correlations require air temperature, pressure, relative humidity (or dew-point temperature) and mole fraction of carbon dioxide, as well as a number of constants. The density of moist air can be calculated by, 

pM a ZRT

  M V   1  x v 1  M a   

Z  1



p a0  a1t  a 2 t 2  b0  b1t xV  c0  c1t xV2 T



(A.6) p2 2 d  ex V T2 where a0, a1, a2, b0, b1, c0, c1, d and e are constants in Table A. 





(A.1)

where, p is the pressure, T the temperature, xv the mole fraction of water vapor, Ma the molar mass of dry air, MV the molar mass of water, R the molar gas constant, and Z the compressibility factor. Ma is calculated by an auxiliary equation, M a  28.9635  12.011xCO  0.0004 (A.2) where 𝑥𝐶𝑂2 is the mole fraction of carbon dioxide. xv is calculated by following steps. First the saturation vapor pressure pSV is given by, 2

D  p SV  1  exp  AT 2  BT  C   T 

(A.3)

where, A, B, C, D are constants, as summarized in Table A.1. Next the enhancement factor f is obtained by, f    p  t 2 (A.4) t is the temperature in degrees Celsius. α, β, γ are constants. Thus, xv can be calculated, xV  hf  p, t 

p SV t  p t   f  p, t r  SV r p p

(A.5)

Then the compressibility Z is obtained:

10



ANNEX B ERROR ANALYSIS OF EXPERIMENTAL MEASUREMENT Table B.1 Instrument list and specifications Transducer Model Range Temperature transmitter Emerson Rosemount 3144 -50 ̊C to 85 ̊C Absolute pressure transmitter Emerson Rosemount 2051S 0 to 500 psig Differential pressure transmitter Emerson Rosemount 3051S -10 to 50 psig Coriolis liquid flowmeter Micro Motion CMF200 0 to 1600 lb/min Coriolis gas flowmeter Micro Motion CMF025 0 to 40 lb/min

Accuracy 0.25% 0.1% 0.1% 0.05% 0.05%

In this study, the experimental error is mainly caused by measuring gas/liquid flow rates, temperature, and pressure [25]. Table B.1 above list all the instruments and their measurement accuracies. As Table B.1 shows, the measurement accuracies for liquid/gas flow rates, pressure and temperature are around 0.05%, 0.1% and 0.25%, respectively. Based on the error propagation theory, the error of GVF (λ) at each ESP stage intake is calculated by, 2

            m L     T    P    m G      m  T   P   m G    L where,  M    2 pM a    L m G m L  L m G   G m L  1  xv 1  V  2  T ZRT   M a  2

2

 MV  2 M a     L m G m L  L m G   G m L  1  xv 1  P ZRT   Ma 1   m G  G

 m G m L    G  L

m   G G  L m L

1

 m   G2 G 

 m G m L    G  L

  

 m G m L    G  L

  

  

2

(B.1)

(B.2) (B.3)

2

(B.4)

2

(B.5)

In the equations above, 𝑚̇𝐺 and 𝑚̇𝐿 are mass flow rate of gas and liquid measured by Coriolis flowmeters, respectively. Substitute all the measurement errors listed in Table B.1 into Eq. (B-1), it can found that the final error of calculated GVF is below 5%.

11

