FLOW MEASUREMENT 2001 – INTERNATIONAL CONFERENCE EFFECT OF PULSATION AND LIQUID CONTAMINANTS ON A MICRO-
MOTION CORIOLIS MASS FLOW METER
Umesh Karnik and John Geerligs?
NOVA Research & Technology Centre
Calgary, Alberta, Canada
ABSTRACT
This paper presents results of tests conducted to evaluate the performance of a 3 inch Micro-Motion Coriolis mass flow meter when subjected to pulsations and liquid contaminants. The testing is performed at the Didsbury Test Facility in natural gas at a pressure of around 5000 kPa.
At pulsation frequencies of up to 70 Hz and amplitudes of up to 70 kPa, the meter performed reasonably well in comparison to its performance in ideal flow conditions within its repeatability of ±0.3%. Literature indicates that large errors result when pulsation frequencies coincide with the meter tube frequency. In the present tests such errors were not detected since the meter tube frequency was around 80Hz. i.e. greater than the test frequency.
In order to test the effects of liquid contaminants, ethylene glycol was used. In comparison to the dry gas flow rate measured by the facility reference, the Micro Motion meter error is linearly proportional to the ratio of liquid to gas flow. The meter measures the total flow (gas plus liquid) reasonably well albeit with a higher uncertainty of ±1.0%.
1.INTRODUCTION
From the perspective of the natural gas industry, Coriolis mass flow meters provide benefits, particularly for line sizes smaller than 6 inches. Apart from the fact that it measures the mass flow rather than volume flow which may be advantageous in some applications, the meter, in comparison to the orifice meter could be beneficial in
(a) eliminating installation effects,
(b) reducing station visits due to plate changes,
(c) reducing station visits by using remote diagnostics (zero offsets, meter settings etc.),
(d) lowering pressure drop coefficient,
(e) increasing flow capacity,
(f) eliminating the need for two differential transmitters the second one sometimes used as
a redundant device,
(g) possible applications to wet gas flows/two phase flows/liquid contaminated gas flows
and
(h) reducing capital costs due to improved meter station design (shorter lengths, reduced
orifice run change outs due to increased capacity demands)
However, before embracing this relatively new technology for natural gas applications, it is important to understand its operating envelope. This would allow the user to make informed decisions on whether the technology is suitable for a desired application.
Mass flow meters have been extensively and successfully used in liquid flow applications. In an attempt to develop a standard for this device, a pattern test was proposed by Jenson (1990). Using water and liquid CO2, these tests evaluated the meter performance when subjected to pulsations, vibrations, swirl and climatic conditions. The meter was found to perform well for
?The authors acknowledge the support of TransCanada Pipelines, Chevron, Southern California Gas, El Paso Energy, GRI and Micro-Motion all participants of the NOVA Met ering Con sortium (METCON)
these conditions in liquid applications. The suitability of this technology to NGL applications was reported by Davis (1990) who noted that the Coriolis meter had excellent repeatability and provided significant capital and operating cost savings in comparison to turbine meter systems.
A good review of the device is presented by Baker (1994).
Recently, manufacturers of mass flow meters have been attempting to develop the technology for gas flow measurement. Several efforts have been made to evaluate the performance of the Coriolis mass flow meter for gas applications. These include the studies of Grenstad et al. (1991) who performed several tests to examine swirl effects and issues related to zero flow offset setting in natural gas, Erdal & Cabrol (1991) who presented tests conducted on natural gas at pressures of up to 100 bar, Nicholson (1994), who tested these types of meters in water and air for installations and swirl, Vetter & Notzon (1994) who studied the effect of pulsation on these meters, Patten and Pawlas (1995a) who studied the effect of swirl on Micro Motion meters and Patten & Pawlas (1995b) who presented several calibrations performed in compressed air. The performance of Coriolis meters has also been evaluated with the use of computational fluid dynamics (Pankratz & Pawlas, 1994).
In most of these previous tests in gas applications, Coriolis meters were found to have demonstrated the promise to become the meter of choice. However, the accuracy and repeatability needed to be improved. Manufacturers have since strived to deliver custody transfer type accuracy (±0.5%) desired by the natural gas industry. The need to reduce pressure drop has resulted in new designs of the Coriolis meters (Hagenmeyer et al., 1994 and Hahn, 1995) consisting of a straight flow through tube as an alternative to the U-tube design. Recently, a 3 inch Micro-Motion (U-tube design) mass flow meter had been tested by NOVA in ideal flow conditions and downstream of installations such as (1) two elbows in plane and (2) two elbows out of plane. The results of these tests are reported by Karnik et al. (1999). For the present testing, a new “out of the box” meter was obtained so that the results for the two meters could be compared.
The present paper contains four sections (a) repeatability and traceability of the facility (b) performance of the meter in ideal flow conditions (c), pulsation effects and (d), liquid contaminants effects.
2.THE DIDSBURY TEST FACILITY
All experiments were conducted at the Didsbury Gas Dynamic Test Facility at line pressures that varied between 4600 kPa and 6200 kPa. Details of the test facility have been presented in the past in several publications (for example Karnik et al., 1996). A sketch of the facility is shown in Figure 1.
Figure 1. Schematic of the Didsbury Test Facility
Before evaluating the meters, it is appropriate that the facility demonstrates its own accuracy and repeatability. The Didsbury Test Facility has been part of round-robin tests since 1995. The first results were published by Karnik et al (1996) who showed that with the use of a sonic nozzle the Didsbury test facility is traceable to other facilities such as NEL, K-Lab, CEESI and the GRI-MRF to within ±0.2%. A similar conclusion was obtained after the testing of the Euromet nozzle in 1997 and recently, the traceability was extended to pigsar, Westerbork, Groningen, Bishop Auckland and TCC using other devices such as a turbine meter and ultrasonic meter (Karnik et al, 2000 and Karnik and Flegel, 2000). In all these tests the facility is well within ±0.2% of the mean curve.
3.PERFORMANCE OF THE MICRO-MOTION METER IN IDEAL FLOW CONDITIONS.
3.1Experimental Conditions
Figure 2b. Photograph of test section for Micro Motion meter tests
For all the present tests, the meter was installed within a vibration-damping cage. The cage was connected to the pipe by two 3 inch plastic clamps and was suspended from the pipe. The meter was supported by means of two pipe stands, each with a 3 inch saddle. One pipe stand was located around 300 mm (3.8D) upstream of the meter whereas the other was located around 400 mm (5D) downstream of the meter.
3.2Results
Results of the baseline tests are shown in Figure 3. The meter was found to have a bias of 0.5% (as reported by Karnik et. Al, 1999) and a repeatability of ±0.3%. Further, at high flow rates the meter exhibited considerable scatter. This occurs at around 10 kg/s. For the present conditions (natural gas at ≈ 5000 kPa), this mass flow translates to a 3 inch pipe velocity of around 40 m/s (some variations due to line pressure and hence density). The scatter in data, at higher flow rates, is consistent with that of Patten & Pawlas (1995) and the tests performed at SwRI (Walker, 1999). Patten and Pawlas (1995) attribute this to the fact that at high flow rates the signal was noisy and resulted in a degradation of repeatability due to a low signal to noise ratio. Erdal & Cabrol (1991) reported that longer averaging times reduced the scatter in the data.
conditions; Karnik et. al (1999)
taken with respect to the calibrated 8 inch-turbine meter (i.e. without the nozzles being choked). This does not appear to change the scatter or the mean in the data.
The bias of 0.5% in the present tests and variable biases in other tests (Walker, 1999) performed by CEESI, SwRI and Ruhrgas for this meter size, suggests a flow calibration in natural gas may be necessary. Grenstad et al. (1991) have reported that the meter needs to be calibrated with the fluid to be used and Myhr (1991) has observed that water calibrations cannot be extended to any fluid. Thus, the current finding is concurrent with previous findings. However, the repeatability of the device has improved considerably since the past tests.
The mass flow meter was also set up to measure gas density. The 4-20 mA output of the meter was calibrated such that density range was from 0-100 kg/m3. Measurements of the density output from the meter are compared to AGA-8 (85) calculations based on pressure, temperature and gas composition. Results presented in Figure 5 indicate that the meter is able to measure the gas density to within 2%.
The pressure loss coefficient, defined as the pressure drop across the meter divided by the dynamic pressure (0.5ρV2) in the 76 mm test section, is shown in Figure 6. The differential pressure was measured by means of two static pressure transmitters. Thus, the accuracy suffers at lower velocities. At higher velocities, with improved accuracy, the pressure loss coefficient is 2.0. This is consistent with testing done with meter A where the pressure loss
coefficient was 2.25. This coefficient is not very different from a typical perforated plate flow
conditioner. For a pipe velocity of 40 m/s (≈10 kg/s) and a typical gas density of 50 kg/m 3
, the pressure loss would be 90 kPa (13 psi).
4.
EFFECT OF PULSATIONS ON MICRO-MOTION METER
This effect has been studied by Vetter and Notzon (1994). Their studies were conducted in water using D6 and D25 type Micro-Motion meters and pulsations were generated by bellows driven by an electrodynamic system, single cylinder piston pump and gear pump. Their conclusions were that if the primary frequency of the pulsation approaches the Coriolis frequency, then large errors could occur. In their case, the Coriolis frequencies (i.e. vibration frequency of the tubes) were between 80 Hz and 200 Hz. Note that there are two tubes in a meter and hence the two frequencies may be slightly different. For example, one of their meters had frequencies of 194 Hz and 208 Hz.
In our case, the meter (Coriolis tube frequency of 80 Hz.) was subjected to a pulsating flow by means of a rotating perforated paddle. McBrien (1997) has described details of the pulsation system. Pulsation tests were conducted for flow rates at which the meter was found to perform adequately in ideal flow (baseline) conditions. Three flow rates were chosen corresponding to 3, 5 and 7 kg/s corresponding to pipe velocities of around 13, 26 and 37 m/s (43, 85 and 121 ft/s). It was established during the baseline runs that the meter performs adequately at pipe velocities of less than 40 m/s (131 ft/s). The tested frequency range was between 10 Hz and 70 Hz. Peak pulsations varied from around 5 kPa to 70 kPa (0.75 psi to 10 psi).
The pressure pulsations were measured by means of 2 PCB transducers with a frequency response of around 20 kHz. Three records were taken for each flow rate and for each record 16384 samples were collected at a sampling frequency of 512 Hz. This resulted in a time window of 32 seconds. This ensured that the data sample was large enough to obtain accurate statistics. The two PCB’s were located as shown in Figure 7 with a separation of 1 m. This separation was chosen such that acoustic wave mapping could be performed for all frequencies of interest. Based on the highest frequency, it is recommended that the separation be less than the quarter wavelength.
In these tests, for the frequency of 70 Hz, the quarter wavelength is approximately
m f
c 43.144==λwhere λ is the wavelength, c is the soun
d velocity in natural gas (≈400 m/s) and f is th
e pulsation frequency. Therefore, the separation o
f 1m should be adequate for mappin
g the
wave. For the lower frequency, we know that a quarter wavelength is 90 degrees. Assuming that a 10 degree resolution is necessary for wave mapping, the lowest frequency that can be resolved by using a separation of 1m can be evaluated by the following
19*49*4==λf
c i.e. a frequency of aroun
d 1 Hz can b
e adequately resolved.4.1
Results
Results of the pulsation tests are shown in Figures 7 and 8. The meter performance is plotted versus the average peak pulsation and also versus the acoustic velocity. It appears that for lower pipe velocities (<27 m/s), the performance of the meter is comparable to the baseline within the 2σ of ± 0.3%. At higher velocities (37 m/s), the data appears to have a bias of around 0.2% with respect the baseline performance. However, apart from 3 data points, the data is within the 2σ of ± 0.3% of the meter.
a function of average peak pulsation.
function of acoustic velocity.
Other correlations have not been presented since, within the repeatability of the meter, there does not appear to be any significant effect of pulsation on the performance of the meter. This is consistent with the findings of Vetter and Notzon (1994). In the present tests, the fundamental freqeuncies do not co-incide with the Corioils frequency and hence we do not see any effect on the measurement.5.
EFFECT OF LIQUID CONTAMINANTS ON MICRO-MOTION METER
Meter B was subjected to liquid contaminants by injecting ethylene glycol using the injection system shown in Figure 9. Details of the injection system have been by McBrien (1999). Once again liquid contaminant tests were conducted for flow rates at which the meter was found to perform adequately in ideal flow (baseline) conditions. Two flow rates were chosen corresponding to 3.6 and 5.4 kg/s corresponding to pipe velocities of around 13 and 27 m/s (43 and 85 ft/s) respectively. The maximum attainable mass fraction (ratio of contaminant mass flow to the gas flow) was around 5.5%
The ethylene glycol is injected upstream of the meter as shown in Figure 9 by means of a spray nozzle. A displacement pump (maximum flow of 17 l/min) is used to draw the glycol from a reservoir and is metered by means of a rotary gear flow meter before being injected into the
Figure 10. The meter error is a linear function of the liquid loading i.e. the ratio of the liquid flow rate to the gas flow rate. However, the meter by definition responds to the total mass flow and not merely the gas flow. Hence, the performance of the meter as a means of measuring the total mass is shown in Figure 11. The reference total mass flow was the sum of the gas flow
since we do not know the hold-up ratio in the tubes, the meter may actually be predicting the true density of the mixture and our estimates may be inaccurate since we do not know the composition of the mixture in the tube. This is further emphasized by the fact that the meter does measure the mass flow accurately to within ±1%.
6.CONCLUSIONS
Testing the Micro-Motion meter has resulted in the following conclusions:
1. Meters provided with a water calibration appear to have a bias of 0.5% in natural gas
applications with a repeatability of ±0.3%.
2. The meter cannot be subjected to high velocities. It appears that for high pressure natural
gas applications, at average pipe velocities greater than 40 m/s the meter begins to cut out resulting in either meter failure or large errors in metering.
3. When subjected to pulsation frequencies of up to 70 Hz and amplitudes of up to 70 kPa
(10 psi) the meter performed reasonably well in comparison to its performance in ideal flow
conditions within its repeatability of ±0.3%. Literature (Vettor & Notzon, 1994) also shows that if one stays away from the Coriolis frequency, measurement errors can be avoided.
4. When subjected to liquid contaminants (ethylene glycol):
?In comparison to the dry gas flow rate measured by the meter prover, the Micro Motion meter error is linearly (nearly directly) proportional to the ratio of liquid to gas
flow.
?The meter measures the total flow (gas plus liquid) reasonably well albeit with a higher uncertainty of ±1.0%.
Details of flow regime and flow patterns were not visualized and could only be inferred from measurements of liquid level along the pipe wall which seemed to indicate that the liquid preferentially flowed along the bottom pipe wall. The liquid level was measured by means of a liquid detector downstream of the meter and upstream of the separator.
7.ACKNOWLEDGEMENTS
The authors would like to thank the NOVA METCON members for their support of the program. In particular, the authors acknowledge Ron Kowch (TransCanada Pipelines), Claire-Becker Castle (Sourthern California Gas), Chuck French (GRI), Frank Ting (Chevron), Hank Poellnitz Jim Witte (El Paso Energy) and Tim Patten (Micro-Motion). Special thanks are due to Ron Kowch and TransCanada for continuous support to flow metering research and time allowed on the high pressure test facility for the presented test. Acknowledgments are also due to Russ Given and Doug Brett for assisting in the data gathering and to Dr. Wojciech Studzinski for his comments.
8.REFERENCES
BAKER, R.C., 1994, “Coriolis Flowmeters : Industrial Practice and Published Information”, Flow Measurement and Instrumentation, Vol. 5, No. 4 pp. 229-246.
DAVIS, T.C.E., 1990, “Fiscal Measurement and Proving Experience with Coriolis Meters”, North Sea Flow Measurement Workshop, NEL, Glasgow.
ERDAL, A., and CABROL, J., 1991, “Comparison of Linearity, Repeatability and Reproducibility for Turbine, Coriolis and Ultrasonic Meters tested at 100bar on Natural Gas”, North Sea Flow Measurement Workshop, Bergen, Norway.
GRENSTAD, J. EIDE, J., and SALVESEN, P., 1991, “Testing of Coriolis Meters for Metering of Oil Condensate and Gas”, North Sea Flow Measurement Workshop, Bergen, Norway.
HAGENMEYER, H., SCHULZ, K-H., WENGER, A., and KEITA, M., 1994, “Design of an Advanced Coriolis Mass Flow Meter Using the Hoop Mode”, Flow Measurement in the mid-90’s, FLOMEKO, NEL, Glasgow.
HAHN, D., 1995, “New Technology Directly Measures Mass Flow of Gas”, 3rd International Symposium on Fluid Flow Measurement, San Antonio, USA.
KARNIK, U., BOWLES, E. BOSIO, J., and CALDWELL, S., 1996, “North American Inter-Laboratory Flow Measurement Test Program”, North Sea Flow Measurement Workshop, Peebles, Scotland.
KARNIK, U., BOWLES, E., and SLOET, G., 2000, “Maintaining Facility Measurement Integrity: Efforts in Canada, USA and The Netherlands”, Proceedings of ASME FEDSM00, ASME Fluids Engineering Division Summer Meeting, Boston, USA.
KARNIK, U., and FLEGEL, D., 2000, “CENTAUR: A Traceability Effort by TransCanada Calibrations”, FLOMEKO 2000, Brazil.
KARNIK, U., GEERLIGS, J., and KOWCH, R., 1999, “Performance Evaluation of a 3 inch Micro-Motion Mass Flow Meter in High Pressure Natural Gas Applications”, 4th International Symposium on Fluid Flow Measurement, Denver, USA.
MANDRUP-JENSON, L., 1990, Testing Coriolis Mass Flow Meters for Pattern Approval”, North Sea Flow Measurement Workshop, NEL, Glasgow.
McBRIEN, R., 1997, High Pressure Pulsation Effects on Orifice Meters, ASME Fluids Engineering Division Summer Meeting, Vancouver, Canada.
McBRIEN, R., 1999, Effect of Liquid on Orifice Meter Accuracy in High Pressure Natural Gas, 4th International Symposium on Fluid Flow Measurement, Denver, Colorado, USA.
MYHR, S., 1991, Field Experience with Coriolis Mass Flow Meter on Hydrocarbon Liquids”, North Sea Flow Measurement Workshop, Bergen, Norway.
NICHOLSON, S., 1994, “Coriolis Mass Flow Measurement”, Flow Measurement in the mid-90’s, FLOMEKO, NEL, Glasgow.
PATTEN, T., and PAWLAS, G., 1995a, “The Effect of Swirl on Coriolis Mass Flowmeters”, North Sea Flow Measurement Workshop, Scotland.
PATTEN, T., and PAWLAS, G., 1995b, “Use of Coriolis Meters in Gas Applications”, 3rd International Symposium on Fluid Flow Measurement, San Antonio, USA.
PAWLAS, G. and PANKRATZ, T., 1994, “Fluid Mechanic Effects in Coriolis Mass Flow Meter”, Flow Measurement in the mid-90’s, FLOMEKO, NEL, Glasgow.
VETTER, G. and NOTZON, S., 1994, “Effect of Pulsating Flow on Coriolis Mass Flow Meters”, Flow Measurement and Instrumentation, Vol. 5, No. 4 pp. 263-273.
WALKER, J., 1999, “Test Results from SwRI, CEESI and Ruhrgas”, Private Communication.