UPM 151 In-line Multiphase Flowmeter Field Trial Comparison vs. Three Phase Flow Separator on Naturally Flowing Shale Wells A. Michael Jones, Independent Flow Measurement Consultant, Travis Gaskill, Daniel Rodriguez, and John Lievois, Weatherford International Ltd. Abstract An inline multiphase flow meter (MPFM) was installed upstream of an existing three- phase separator. Flow was routed in series through both devices to test a six well manifold on a pad of six naturally flowing wells. The MPFM was initially commissioned in late June 213. The test was supervised in detail by the operator of the pad and the MPFM vendor s engineers during several days later in the summer (July- August 213). The objective was to assess the performance of the MPFM against the existing separator, with the separator operating under close supervision. The MPFM system was subsequently monitored remotely by vendor s engineers to verify its health, data integrity, and the consistency of its measurements. This monitoring continued until late October 213. The MPFM continued to operate normally during the entire four month period without intervention by its vendor, while the separator required close supervision to maintain consistent measurements. The outcome of the test was positive with 1% uptime on the MPFM, and good measurement performance compared to the existing test separator. Details of the test program including set up, execution, results, and lessons learned, will be discussed. Trial Set Up An inline multiphase flowmeter (MPFM) was installed upstream of an existing three phase separator on a Texas shale asset. The well pad consisted of six naturally flowing wells connected to a three phase separator through a manifold. A field trial was conducted lasting five months. The MPFM instruments and piping were mounted inside of a skid for ease of transport as well as for movement to other well pad sites. Installation, in- situ calibration, and training took a total of three days. The actual trial setup can be seen in Figure 1. Figure 1 MPFM installed on well pad The MPFM used in the field trial was a 2 in. Alpha VSRD inline multiphase meter. This MPFM uses four discrete instruments to measure the three phases. Bulk fluid flow is measured using a Venturi- nozzle differential pressure (DP) element in conjunction with a Sonar based bulk fluid velocity measurement. The DP measurement and Sonar measurement are combined into one spool. Water cut (WC) is measured with a RedEye MP infrared probe. Bulk fluid density is measured using a gamma densitometer. All four of these instruments are polled by the MPFM s flow computer to calculate three volumetric phase flow rates at line and standard conditions, gas volume fraction (GVF), and WC. The overall footprint was approximately 1/3 that of the separator. A breakdown of the different components of the MPFM is shown in Figure 2. Copyright 215, Letton Hall Group. This paper was developed for the UPM Forum, 25 26 February 215, Houston, Texas, U.S.A., and is subject to correction by the author(s). The contents of the paper may not necessarily reflect the views of the UPM Forum sponsors or administrator. Reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Letton Hall Group is prohibited. Non-commercial reproduction or distribution may be permitted, provided conspicuous acknowledgment of the UPM Forum and the author(s) is made. For more information, see www.upmforum.com.
Based upon a sizing calculation conducted by the MPFM vendor, an MPFM with a 2 in. line size was chosen to provide the correct flow rate envelope for the six wells located on the pad. The flow envelope overlaid with data captured during the field trial is shown in Figure 3. Figure 2 Example MPFM skid with labeled components The MPFM vendor s published accuracies are shown in Table 1. The accuracy of the MPFM varies depending on which flow conditions it is exposed to. During the field trial, the average GVF seen by the MPFM was 97.8%. In this regime, the published uncertainty specifications of the MPFM are ±1% on gas rates and ±11% on liquid rates. In reference to MPFM measurement uncertainties, NEL has stated that uncertainties of 5% on liquid rate, 1% on gas flow rate, and 5% on WC are realistic expectations over a wide range of flow conditions 1. Based on Table 1, the MPFM meets these criteria except in the most extreme wet gas regimes. Liquid Loading/ Approx. GVF@ 1 bar Gas Rate Accuracy Liquid Rate Accuracy Water Cut Accuracy (abs.) Table 1 MPFM vendor published accuracy Multiphase 1 < XLM<.5 < GVF < 2% ±5% (rel.) ±5% (rel.) < GVF < 2% 2% < GVF < 9% ±5% (rel.) ±5% (rel.) 2% < GVF < 95% Type II Wet Gas.2 < XLM.4 9% < GVF < 99% Type I Wet Gas < XLM.2 99% < GVF < 1% ±1% (rel.) ±1% (rel.) ±11% (rel.) ±75 bpd (±1 m 3 /hr.) 95% < GVF < 99.5% ±2% ±4% ±1% Figure 3 MPFM flow envelope Pressure, volume, and temperature (PVT) tables were created for use in the MPFM s flow computer calculations at line conditions and conversions to standard conditions. The tables were based on chromatograph data taken on individual well samples. Gas density and compressibility were modeled using a real- time implementation of the American Gas Association s AGA 8 GCMII correlation method. Liquid properties were created from Black- Oil Model correlations. The MPFM was powered using a mobile bank of solar panels plus batteries. The steady state power consumption of the MPFM was approximately 15 watts (W). The MPFM flow computer was connected to a line of sight Wi- Fi connection and was used for Modbus communication as well as File Transfer Protocol (FTP) configuration uploads and datalog downloads. Through Modbus, the flow computer provided real time three phase flow rates as well as well test volumetric totals. Well test start and stop times were controlled remotely via Modbus. Execution The main purpose of the trial was to compare the MPFM s performance to that of the well test separator. The separator used an orifice plate for gas flow rate measurement, a turbine meter for water flow rate measurement, and a Coriolis meter for oil flow rate measurement. Well tests were conducted by the operators on a 24 hour interval. Separator volumetric totals were recorded via the operator s distributed control system. Daily MPFM logs were retrieved via FTP and were processed 2
into graphical format. Comparisons of well test volumetric totals were conducted on a weekly basis. An uncertainty analysis was conducted, determining the accuracy of the MPFM relative to that of the separator. The uncertainty of the MPFM was then compared to its published uncertainty to determine its validity. Results A total of 45 well tests were conducted over a 4 month period. Each well was put into test an average of seven times. Through the MPFM daily logs, individual well characteristics were learned. All six wells were naturally flowing but showed unique characteristics in slugging and liquid loading. One- hour snapshot from two example wells are shown in Figures 4 and Error! Reference source not found.. Well flow characteristics were seen to be affected by the operator s chosen choke setting. Based upon these real- time flow rate results, four of the six wells were classified as long- period slugging wells, with approximately 2-5 events of increased liquid loading per hour. Figure 4 Hour long snapshot of liquid flow rate from Well E After the first 33 tests, comparisons were made between the separator and MPFM volumetric rates to determine measurement accuracy. Initial comparisons showed the MPFM having a negative bias of - 22.9% on oil and a positive bias of +4.2% on gas. This is shown in Figures 6 and 7. Qostd (+/- 11% lines) Figure 5 Hour long snapshot of liquid flow rate from well A At some point during the field trial, the turbine meter on the water leg of the separator began to malfunction and report zero flow rate to the operators. While the MPFM meter continued to report a water rate, there was no data to compare it to and was not included in the analysis. The root cause was eventually found to be paraffin buildup. MPFM [bbl/d] 16 14 12 1 8 6 4 2 5 1 15 Separator [bbl/d] Figure 6 Oil flow rate comparison, MPFM vs. Separator 3
Qgstd (+/- 5% lines) Qostd (+/- 11% lines) MPFM [Mscf/d] 3 25 2 15 1 5 5 1 15 2 25 3 Separator [Mscf/d] MPFM [bbl/d] 16 14 12 1 8 6 4 2 Old PVT New PVT 5 1 15 Separator [bbl/d] Figure 7 Gas flow rate comparison, MPFM, vs. Separator Figure 8 Oil flow rate comparison before and after PVT table update Due to the opposite- sign biases, it was jointly decided to gather new chromatograph data to reduce PVT uncertainty. The updated chromatograph data showed oil API gravity increased from 59.6 to 61.2 and gas specific gravity decreased from.738 to.78 since the last samples were taken. Simulations were run by the MPFM vendor on the 33 test points to determine the effect of changing the PVT on the accuracy of the results. Because the MPFM s flow computer recorded logs of all the real time results from its four instruments, these logs were used to post- process the flow rate data using the new PVT information. Results of the simulations proved that the root cause of the biases was the PVT data. MPFM vendor and operator agreed to update the MPFM s PVT data. The final 12 well tests in the trial were run using the new PVT. The red diamonds shown in Figures 8 and 9 represent the 12 well tests run with the updated PVT. MPFM [Mscf/d] 3 25 2 15 1 5 Qgstd (+/- 5% lines) Old PVT New PVT 5 1 15 2 25 3 Separator [Mscf/d] Figure 9 Gas flow rate comparisons before and after PVT table update An uncertainty analysis was conducted based upon these 12 new well tests and is shown in Table 2. Oil and gas uncertainties were determined based on a 95% confidence interval (C.I.) calculated using Student s t- distribution which is best suited for small data samples. Maximum discrepancies were seen to be +12.16% and - 1.52% on oil and gas rates, respectively. Even with the small sample size of 12 well tests, the accuracy of the MPFM was within 1% of its published accuracy on both the liquid and gas rates. 4
Table 2 Uncertainty specifications based on MPFM vs. Separator tests Phase Mean Bias Confidence Interval* Expected Discrepancy with respect to separator (Min, Max) Oil 3.84% ± 8.32% (- 4.48%, +12.16%) Gas -.76% ±.76% (%, - 1.52%) Confidence Interval Probability 95% *Based on Student s t- distribution After the field trial was completed, the MPFM was moved to a different well pad with a lower manifold pressure (~2 psig). The MPFM was moved and re- comissioned within one day. The MPFM successfully logged flow rates at the new well pad for seven more months before being decommissioned. The new well pad showed the same signs of paraffin buildup on the water turbine meter. Lessons Learned With the wells now characterized by their real time fluctuations in liquid loading, the well test periods were recommended to be reduced to 5 hours on slugging wells and 2 hours on steady wells. This optimized test duration allows every well to be tested daily instead of weekly. This should result in optimized well productivity as well as allow for increased allocation accuracy. Decreased well testing times can also be used to produce higher resolution trending and forecasting for the well monitoring and reservoir management teams. The MPFM showed 1% uptime during the field trial and no fouling issues were seen. The MPFM does not contain any moving parts and the RedEye MP is the only minimally intrusive part on the skid. This allows the MPFM to operate with minimal maintenance relative to the separator, reducing overall operational expense (OPEX) and reducing the life cycle cost of the MPFM. The fouling of the turbine meter on the water leg of the separator was a potential indicator of inadequate liquid- liquid separation. Acquiring up to date oil and gas compositions will decrease the uncertainty for any flow measurement system. An MPFM s accuracy is closely related to the accuracy of the PVT input provided for its calculations. If detailed oil and gas composition is not available, an increase in uncertainty is to be expected. This PVT sensitivity is especially true for the properties of the dominant phase at either extreme of the GVF range. For the field trial, the wells were naturally flowing wells. Under this circumstance, the oil and gas composition is expected to change slowly (if at all) over time. When some artificial lift methods are implemented (e.g. gas lift, CO2 injection), or where comingled streams are measured, compositional changes should be expected and the oil and gas composition will have to be updated on a more frequent basis to maintain a steady level of uncertainty. The accuracy of the MPFM during the field trial is representative of its published accuracies. Based upon these accuracies and its intended use, the MPFM can be designated a Class 3 flowmeter based upon the meter classification 2 presented during a recent SPE webinar. The classes and their description are shown in Table 3. While the MPFM was being used as a Class 3 meter during the field trial, the gas flow rate uncertainty relative to the separator was much better than that of an average Class 3 MPFM. Due to the flow being in the wet gas regime, the accuracy of the MPFM was better for the dominant phase while still maintaining acceptable uncertainty for the disperse phase. Table 3 MPFM classifications Meter Class Typical Use Uncertainty 1 2 3 4 Fiscal measurement systems: Measurements done for monetary exchange Allocation meters: Determines hydrocarbon liquids and natural gas produced from an allocation node. Reservoir management / surveillance: Determines quantity of oil, water, and gas produced from an individual well Environmental and emission trading Gases <= 1% Liquids <=.25% 1% to 5% 1% to 2% 1% to 2% The MPFM was able to be moved to a different well pad on the same asset and brought online within one day. The ability of the MPFM to operate on multiple well pads with differing operating conditions increases its life cycle returned value potential due to having multiple valid areas of use within its expected useful life of at least 15 to 2 years. The ability to re- commission the MPFM in one day decreased downtime and decreased OPEX. Nomenclature DP = Differential Pressure GVF = Gas Volume Fraction FTP = File Transfer Protocol MPFM = Multiphase Flow Meter OPEX = Operational Expenditures 5
Psig = Pounds per Square Inch (gauge) PVT = Pressure / Volume / Temperature W = Watts WC = Water Cut XLM = Lockhart- Martinelli Parameter References 1. Graham, E., MacDonald, M. Webinar Introduction to Multiphase & Wet Gas Flows, Questions & Answers, 214 2. Major, D. Liquid Metering Selection, SPE Web Events, Feb. 215 6