MULTIPHASE FRACTION METER DEVELOPED AND FIELD-TESTED
Eivind Dykesteen, Karl Herman Frantzen
Chr. Michelsen Institute
Bergen, Norway
A nonintrusive multiphase fraction meter, based on the combination of capacitance sensor and a gamma-radiation density meter, has been developed by Chr. Michelsen Institute (CMI) Bergen, Norway.
The project was sponsored by British Petroleum and Saga Petroleum.
Laboratory tests have shown this measurement system capable of determining the composition of a well mixed oil-gas-water flow to an uncertainty of better than -3% for each of the components.
This industrial prototype meter has also successfully completed field testing on a U.K. land-based oil field.
RESEARCH PROGRAMS
In a development scheme in which several wells are completed on a remote satellite and manifolded together to permit transport to a central processing platform, the conventional test separator measurement offers an expensive solution.
In this case, in addition to the requirement for a dedicated test separator on the processing platform, there must be two separate pipelines from the satellite, one for production and one dedicated for the test separator. A valve and manifold must also be installed at the satellite.
Because of the potential savings from development of marginal fields as "minimum facilities" satellites, with multiphase transport of the unprocessed wellstream, many companies are running large R&D programs in multiphase technology.
One such program, "the three-component ratio measurement research program," or "the 3C-project," was started up at Chr. Michelsen Institute in Norway in the spring of 1985, with BP Norway Ltd. and Saga Petroleum as sponsors.
The objective of this 3-year, 15 million Norwegian krone (approximately $2.5 million) project, has been to develop a nonintrusive meter for on-line monitoring of the composition of the multiphase flow in a pipeline. The initial studies leading up to this project were sponsored by the Norwegian Research Council, NTNF.
During the 3C project, three different measurement concepts were evaluated:
- In the "impedance measurement system," a carefully designed surface-plate capacitance sensor is driven at a medium high frequency (500 khz), and both the capacitance of the sensor and the dielectric loss are measured. These two measurements are in turn related to the composition of the flow through mathematical models.
- The "capacitance/gamma measurement system" utilizes a conventional surface-plate capacitance sensor, together with a gamma-radiation density meter to derive the flow composition.
- The third alternative evaluated is a combination of these two ("impedance/gamma measurement system").
These three measurement concepts were analyzed through testing of laboratory prototypes and through sensitivity studies based on detailed simulation models of the measurement systems. On the basis of conclusions from this work, it was decided that an industrial prototype "capacitance/gamma measurement system" should be built and tested.
At this stage a collaboration with Norwegian instrumentation company Fluenta a/s was also initiated.
The time schedules for this prototype development project have been tight: From project kickoff in February 1989, a measurement system ready for laboratory testing was completed by July. Following the lab tests scheduled for August and September, final modifications of the measurement system were completed.
Also, a letter of intent for Baseefa (British Approvals Service for Electrical Equipment in Flammable Atmospheres) approval of the electronics to Exi classification was received, and the measurement section was pressure tested and certified for use as Class 600 equipment.
In December 1989, the measurement system was installed on a BP-operated U.K. land-based site, where it was tested in the pipeline upstream of the first stage separator.
MEASUREMENT PRINCIPLE
The basic concept behind the measurement system developed at CMI is the solution of three equations with three unknown parameters: the fractions of oil, water, and gas, respectively.
The first equation is derived from the fact that the sum of all the fractions will always be equal to 1; i.e., the pipe is always full of oil, gas, or water, or a mixture of two or three of them. This gives Equation 1:
[SEE FORMULA (1)]
where:
[SEE FORMULA]
In order to derive the two remaining equations, only two independent measurements are needed which are correlated to the composition of the flow.
Two such measurements could be given by the complex permittivity of the flow, provided it could be shown that the real and imaginary parts of the complex permittivity were independently correlated to the flow composition.
Indeed, these are the two parameters used in the measurement concept previously denoted as the "impedance measurement system."
This measurement principle is further described in Dykesteen, et al., where it is shown that the complex permittivity actually yields two such independently correlated measurements.1
Because, at the frequencies we could operate the system, the dielectric loss (the imaginary part of the complex permittivity) is an extremely small quantity compared to the real permittivity of the oil/gas/water mixture, it could be advantageous to look for alternatives to this measurement.
This is done in the "capacitance/gamma measurement system," in which a gamma-radiation meter is used to measure the mean density of the multiphase flow. In this system, therefore, the two independently correlated measurements are the mean permittivity and the mean density of the flow.
MEASURING PERMITTIVITY, DENSITY
Nonintrusive measurement of the mean permittivity of a well mixed flow can be done with a capacitance sensor. A simplified electrical equivalent diagram of this sensor is shown in Fig. 1.
While the parallel capacitance (Cp) and the wall capacitance (Cw) ideally are constant, the bulk capacitance (Cb) is proportional to the permittivity of the flow through Equation 2:
[SEE FORMULA (2)]
where:
Co = Sensor constant (the bulk capacitance with empty sensor)
eb = The permittivity of the bulk (relative to vacuum)
The sensor constant (Co) may be established by calibration of the sensors with respect to fluids of known permittivities.
The permittivity of a two-component mixture of oil and water will be related to the fraction of water to oil (f; Fig. 2) which is a feature that has actually been used for development of two-component, water-in-oil monitors.2
For a three-component mixture of oil, gas, and water, there will be a range of such curves dependent on the gas fraction (a; Fig. 3). On the basis of a two-component permittivity model developed by van Beek,3 CMI developed and verified a model for the permittivity of a three-component mixture of oil, gas, and water in which the oil is regarded as the continuous phase (Equation 3).
[SEE FORMULA (3)]
where:
es = The permittivity of the two-component mixture of oil/water
The absorption of gamma radiation in a medium will be a function of the mean density along the path of the gamma particles. This is a well known principle used for industrial measurements such as level measurements and for void fraction measurements in two-phase flows.
The density of a three-component mixture of oil, gas, and water may be expressed as Equation 4:
[SEE FORMULA (4)]
where:
[SEE FORMULA]
Densities
By taking into account that the sum of ail the fractions is 1 (Equation 1), this expression can be reduced to the form shown in Equation 5:
[SEE FORMULA (5)]
Hence, measuring the mean permittivity of the flow (eb) and the mean density (p) and then solving Equations 3 and 5 by iteration will yield the water fraction () and the gas fraction (a). Finally, the fraction of oil in the pipeline (t) is found from Equation 1.
INDUSTRIAL PROTOTYPE
The measurement system is schematically shown in Fig. 4.
It consists of a 4-in. capacitance sensor, a capacitance transmitter, a gamma densitometer, and a control room readout unit.
A cross section of the capacitance sensor is shown in Fig. 5.
It is designed as a spoolpiece with 6-in. flanges, with a 4-in. ID ceramic liner.
The electrodes are metallized on the outside of this ceramic liner. The electrode layout is shown in Fig. 5.
Two rectangular electrodes are situated opposite each other on the outer surface of the ceramic liner.
There also is a guard electrode around the circumference of one of these electrodes in order to minimize the parallel capacitance between the electrode plates.
In addition to acting as an electrical screen, the outside steel housing of the sensor has been designed as a pressure container to Class 1500 specifications. Electrical connections to the electrodes are done via high-pressure glands mounted in the steel wall.
The ceramic liner, which is in direct contact with the fluid, is a zirconia-reinforced alumina tube. This material was chosen because of its good electrical properties in combination with its high strength and extreme toughness against corrosion and erosion.
The volume between the steel housing and the inner ceramic liner is filled with an insulating material.
For accurate measurement of the small capacitances of the sensor and a long-term stability of the measurement system, the electronic circuit design has been very important.
The measuring principle used in this instrument is optimized with respect to temperature stability and has also proved a very high degree of long-term stability. Special care had to be taken to prevent stray capacitances in connectors and the printed circuit board from influencing measurement accuracy.
The capacitance-measurement electronics are based in principle on a resonance circuit in which the sensor capacitance determines the charge or discharge rate of an integrator.
The resulting resonance frequency is thus inversely proportional with the sensor capacitance. With the components chosen for the resonance circuit, the operating frequency band is 10-40 khz. This frequency is divided by 64 before transmission as a current-frequency signal.
The sensor-head electronics are enclosed in a gas-proof polycarbonate box and put on a socket in an IP-65-certified enclosure, together with three temperature transmitters. Two measure the temperature within the sensor (Pt-100 elements), and one measures the temperature within the electronics enclosure.
The sensor electronics are designed to meet Baseefa certification for use in explosive atmospheres (EExia Ilc T4, - 40 to 70 C.), and a letter of intent for such a certificate has been received from Baseefa. The gamma meter is an important part of the system, but there are very few requirements to the meter. The most important need is that there be a fast response time of the meter. For mechanical design reasons, a clamp-on device was preferred.
The unit chosen is a standard Krohne NDM 1000, with a Krohne NDC 2000 readout unit. The radioactive source is Caesium 137. With activity 2.5 mCi, the response time of this meter is 1 sec. This unit is certified for use in hazardous areas.
The two primary sensors, i.e., the capacitance sensor and the gamma-density meter, are connected to a readout unit located in a safe area. The control room readout unit is designed as a 19-in. rack system, which includes a micro-controller-based data acquisition unit, the gamma-meter readout unit, and a rack-mounted PC, in addition to power supplies, safety barriers, and transmitters.
All the field measurements are handled by the micro-controller unit, which in turn communicates with the PC. All the mathematical models and calibration constants are implemented on this PC, which therefore takes care of all calculations, presentation of measurements, and storing of data.
LABORATORY TESTS
Following construction and calibration, the measurement system went through a test period on the 4-in. crude oil flowing at CMI. This rig is designed for low pressures and for a temperature range of 15-40 C.
The rig is a closed loop and generates a well-mixed bubble flow through the vertical test section. Water and oil fractions in the range 0-100% and gas fractions up to 60% can be circulated.
The tests were done by initially filling the closed loop rig completely with a mixture of oil and water with a known water/oil ratio.
Then, while the mixture was being circulated, some of the liquid was drained off and substituted with nitrogen. This was done in steps of 10% up to at least 50% gas/liquid ratio and for a series of different water/oil ratios in the range of 0-75%.
All the tests were done with Statfjord stabilized crude oil and salt water.
Because the CMI 4-in. flowrig is a closed loop configuration with no separation tank and as the gear pump keeps the flow well mixed, the water/oil ratio can be regarded as constant during a test. This assumption has been checked by liquid sampling at different locations and at different intervals and by centrifugal bs&w analysis of these samples.
The water/oil ratio in the flow is therefore known to an uncertainty better than - 1 %.
The total volume of the rig has been calibrated quite accurately. Then, if the amount of liquid actually in the rig at a given time is known, it is easy to calculate the total gas/liquid ratio in the rig.
Under the assumption that there are no liquid or gas hold-ups in the rig, then the gas fraction in the test section will be the same as for the whole rig.
A simple test with two gamma densitometers at different locations in the rig, however, showed that there actually were gas holdups in the rig and that the gas fraction in the test section is generally 2-4% below the total gas fraction of the rig. This is shown in Fig. 6.
With Fig. 6, as a calibration curve, the gas/liquid ratio in the flow is therefore known to an uncertainty better than 2%.
During the tests, in addition to this independent gas-fraction reference, we wanted an on-line reading of the gas fraction in the test section. This was achieved by using the on-line density reading from the gamma meter in the measurement section and applying Equation 6.
[SEE FORMULA (6)]
where:
aref = (On-line gas fraction reference
p = Measured density
ps = Density of the oil and water mixture
By regularly checking this on-line gas-fraction reference with the independent gas reference, this does not add to the stated uncertainty in gas fraction reference. All test results presented here are referred to the on-line gas-fraction reference.
The reference water fraction is then found from the relation shown in Equation 7.
[SEE FORMULA (7)]
where:
f = Water/oil ratio of the liquid phase
TEST RESULTS
Typical samples of the tests are shown in Fig. 7. The solid lines represent the "true" values, i.e., the reference measurements, while the fractions measured by the multiphase fraction meter are plotted as discrete measurement joints. Each of the plots represents one test series, i.e., a fixed water/oil ratio and varying gas/liquid ratio.
In Fig. 8 these measurements are presented in a slightly different way.
Here the difference between the measured component fractions and the reference fractions have been plotted with respect to gas-reference fraction.
These deviation plots contain all data from all of the fraction tests.
Analyses of these data show that for the oil-fraction measurements, more than 95% of all the measurements are within 3% error band of the reference oil fractions. For the water measurements, even better results are achieved in that more than 97% of all the measurements fall within the 3% error band. The gas measurements are exceptionally good in that 100% are within the 3% error band, and that 97% of the measurements are within 1% of the reference.
Also, it is important to note that the deviations from the reference values are mainly randomly distributed. Thus, the systematic deviations are very small. This implies that the accuracy could be further improved by averaging over more spot measurements.
FIELD TEST
In order to gain some operational experience with the multiphase fraction meter, the meter has recently completed a 5-month test period at a land-based BP-operated oil field.
During this period, the meter experienced 4 months of continuous flow through the sensor head, the system was powered up for more than 250 hr (whenever key personnel have been present), and 70 hr worth of test data were logged and analyzed. The multiphase meter was installed in a vertical up-flow section of a 4-in. flowline upstream of the first-stage separator.
Additionally, there was a possibility to switch between two different reservoirs with significantly different characteristics: Reservoir "A" with a medium gas-to-liquid ratio (Q0%) and high water cut (Q0%) and Reservoir B with a high gas fraction (Q0%) and low water cut.
Although it is important to emphasize that the purpose of the field tests was to assess the reliability of the measurement system rather than to the system's accuracy under operational conditions, the particular installation provided some means of reference measurement data. In addition to these single-phase flow measurements downstream of the first-stage separator, well test data have been used for reference.
Fig. 9 shows a sample test run, as measured by the CMI meter, with the measurement system connected to Reservoir A.
In this figure the fractions of the three different components are individually plotted with respect to time. Although the flow appears to be very stable over this 1-hr test period, it may be observed that the gas-fraction curve varies slowly with a time consant of several minutes. An enlarged part of this test (1 min) is shown in Fig. 9b. From this plot we can see that superimposed on the slowly varying gas fraction is also a more rapidly changing flow pattern. This indicates a plug flow through the test section.
In Fig. 9b, we can also observe how the water and oil fractions follow each other quite closely. This means that, at least over shorter periods of time, the water cut is a fairly constant parameter.
Upstream of the 3C meter installation point there was a pig receiver station. Fig. 10 shows how the flow pattern surges during the last hour before the pig is received.
For completeness, Fig. 11 plots a sample test run from Reservoir B. Again we observe that the water cut is very stable but low, while the instantaneous gas fraction varies from 30 to 90%.
EXPERIENCE, CONCLUSIONS
The only serious problem experienced during the fieldtest period was a leaking internal seal (not part of the pressure integrity/safety of the measurement system). This caused an overreading of the water cut but was discovered and repaired.
However, for good performance in future high-pressure applications (above Class 600), we are now redesigning this system. During the field test, the multiphase fraction meter experienced higher gas fractions and flow patterns, which could not be simulated in the CMI flow rig. The measurement system proved able to operate also under these flow conditions.
Still, studying the fast-response measurement data shows that there is a time lag in the gas measurement as a result of response time (1 sec) of the gamma densitometer. Therefore, particularly for flow regimes like plug and churn-flow, improved performance may be achieved by use of a faster response gamma densitometer.
The CMI multiphase fraction meter will in itself provide the offshore oil industry with a useful instrument for well monitoring and reservoir management. In most cases, however, and in particular when the multiphase flows from satellites are to be measured for production-allocation purposes, a composition monitor alone will not solve the measurement problem. The requirement here is for a meter that measures the individual flowrates of oil, gas, and water on-line.
During the field test, the CMI multiphase fraction meter was run in series with a BP-developed positive-displacement (PD) meter. This meter measures the total multiphase flowrate, and thus, in combination with the fraction meter, the individual flowrates of oil, gas, and water may be found.
Because this solution will involve an intrusive flowmeter with moving parts, it will be prone to erosion and mechanical wear. For remote installations, a rugged nonintrusive measurement system will be preferred.
CMI has investigated the possibility of complementing the CMI multiphase fraction meter with a flow-velocity measurement. This measurement is proposed to be done by signal processing of the signals from a multiple-electrode capacitance sensor.
Preliminary results suggest the possibility of nonintrusive measurement of the individual flowrates of oil, gas, and water in the near future.
REFERENCES
- Dykesteen, E., Hallanger, A., Hammer, E.A., Samnoy, E., and Thorn, R., "Non-intrusive three-component ratio measurement using an impendance sensor," J. Phys. E: Sci. Instrum., Vol. 18, 1985.
- Hammer, E.A.. and Frantzen, K.H., "Capacitance Transducers for Non-Intrusive Measurement of Water in Crude Oil," Australian Instrumentation and Measurement Conference, Adelaide, Nov. 14-16, 1989.
- van Beek, L., "Dielectric Behaviour of Heterogenous Systems," Progress in Dielectrics, Vol. 7, 1967.
Copyright 1991 Oil & Gas Journal. All Rights Reserved.