GASUNIE SELECTS TURBINE METERS FOR RENOVATED EXPERT METERING STATIONS
P. M. A. van der Kam
A. M. Dam
N.V. Nederlandse Gasunie
Groningen, The Netherlands
K. van Dellen
Daniel Industries Inc.
Houston
Results of extensive research in preparation for renovation of N.V. Nederlandse Gasunie's gas-export metering stations indicated the selection of the turbine meter as the primary flow-measurement device.
The decision was made after extensive flow-measurement research whose most significant results are given here.
Research was also carried out on calorific-value measurement.
EXISTING SYSTEM
In 1989, Gasunie sold 72 billion cu m (2.5 tcf) of natural gas. About 32 billion cu m (bcm) of this was exported to other Western European countries as a result of long-term contracts.
The export of gas started in the late 1960s, and the volumes are expected to remain more or less constant until the first decade in the next century.
The gas is measured at the border in 13 export stations. Among these are six large ones, operating at 50 or 60 bar (725-870 psi). The numbers characterizing these stations are given in Table 1. They were built between 1968 and 1973.
At the time of construction, orifice plates were the only available proven instruments to measure large gas flows. The turbine meter was still in its infancy and no experience of using large meters was available.
The present layout of the stations, the straight lengths upstream and downstream of the orifice plate, and the geometry of the plate itself are in accordance with the ISO Standard 5167.1
The values are typically 0.5-0.7. The _P is measured with differential pressure balances of Desgranges and Huot, the gas density with densitometers.
In case of power failure, battery-powered pressure-temperature (P-T) recorders are available as backup. With billing based on delivered energy, the calorific value is measured with a Cutler Hammer calorimeter.
All signals are fed into a station computer which calculates the energy flow with ISO 5167.
The instrumentation, which although at that time was of the highest quality available, is now becoming outdated. Maintenance is getting more difficult each year and for some instruments spare parts will no longer be available in the foreseeable future.
Therefore, it is in Gasunie's interest to redesign the stations with modern instruments.
REQUIREMENTS
According to both the export contracts and the policy of Gasunie, the measurements of the sold energy (gas) shall be taken with state-of-the-art techniques.
With this in mind, in the autumn of 1988 Gasunie formulated the requirements for the renovated stations.
- No blockage of gas flow, even in the case of failures in instrumentation
- Reliable instrumentation with availability near 100%
- The highest accuracy possible, given that even the smallest system errors have enormous financial consequences (e.g., 0.1% of 32 bcm at 10/cu m = $3.2 million)
- Proven technology
- Double measurement, with a complete backup system based upon different measurement principles which can be used to check the primary measurement and can take over in case of failure
- Unmanned operation, with telemetry to the central dispatching.
First choices for the type of instrumentation were made on the basis of data existing in published literature as well as on the basis of Gasunie's experience. For some choices, extra supporting research programs have been set up.
A major decision had to be made relating to the flow-metering principle. Two options were available: retain the orifice plate or change to turbine meters.
It has been shown on several occasions that the uncertainty of a metering system based on orifice plates or on turbine meters is generally equal from a theoretical point of view.
That is true if the installation effects are ignored, and the uncertainties in the calibration of the various components of the secondary equipment, the uncertainty in the turbine meter calibration (at high pressure), and the uncertainty specified in ISO 5167 for the orifice plate are taken into account.
The resulting "theoretical uncertainty" varies between 0.6% and 0.8%, depending on the input uncertainties.
This theoretical uncertainty is not decisive and other parameters must be considered. Other major aspects which Gasunie has considered are installation effects, working range, and long-term performance.4
Additional research was necessary in support of some of these aspects.
INSTALLATION EFFECTS
In the export stations upstream of the meter runs, there is a header with a tee, a straight length of about four pipe diameters (D), and a 90 bend into the meter run. This "double elbow out of plane" configuration is known to produce swirl.5 6
Recent research has confirmed the possibility of swirl in this layout.7 It has also become clear6 that the straight lengths specified in ISO 5167 are not enough for decay of swirl and that straighteners will improve the situation but also have an unpredictable influence on the orifice meter performance.8 9
For turbine meters, perturbation tests have been carried out according to the new ISO draft proposal ISO DP9951.10 11 Among these are tests featuring a double elbow located close to and in front of the meter.
Typical results are shown in Fig. 1. It can be seen that even with a straight length of only 5D in front of the meter, a simple tube-bundle straightener reduces the swirl influence to less than 0.1%.
Thus, in the relatively ideal situation at the export stations, the flow profile will have no influence on the turbine meter.
Another installation parameter that is known to influence the orifice plate performance is pipe-wall roughness. The results from Gasunie's project on 600-mm orifice plates have already raised some doubts on the requirements specified in ISO 5167 regarding wall roughness.12 Further research on this was considered necessary.
WORKING RANGE, PERFORMANCE
The working range is a feature which clearly favors the turbine meter. The typical orifice meter-run working range is 1:3, and with application of modern "smart" transmitters, possibly 1:6. High-pressure turbine meters easily achieve a working range of 1:50 with a stable error curve within 0.5%.
The use of turbine meters thus results in an easier operation of the stations: No line switching is necessary; all meter runs are kept open and are operated in parallel.
Also, for the same maximum capacity, the total number of meter runs can be reduced from 33 to 24, thus resulting in a considerable reduction of investment and maintenance costs.
Orifice plates require periodic inspection, which can usually be done simply by withdrawing the plate from the line. No wear will occur because the plate can be kept in good condition.
The turbine meter, on the other hand, has moving parts and thus is ultimately subject to wear. Gasunie, with about 20 years' experience with some 3,000 turbine meters ranging in size from 100 mm to 600 mm, has a good insight into the long-term performance of these meters.
Mechanical problems are rare and can be detected by the flow computer through, for example, double pulse counting and pulse intercomparison techniques.
After completion of the first high-pressure calibration of all meters in operation, Gasunie commenced a high-pressure recalibration program in 1985. The first sample of 60 meters1 3 showed that the turbine meter is stable, reliable, and accurate.1 3
Moreover, in the renovated stations the backup meter must possess such quality that any shift in error curve is detected.
ROUGHNESS INFLUENCE
For the contractual flow through the export station at Hilvarenbeek, six 500-mm meter runs are sufficient. Therefore, two of the meter runs (Nos. 5 and 6) could be dismounted and taken to Gasunie's Westerbork test site for the performance of a number of lab tests on the influence of pipe-wall roughness.
When the pipes were inspected internally, a thin (
In the latter condition the wall was clearly rougher than before. The roughness (Rz) was then measured to be Rz-140 mm, well within the ISO limit of 200 mm.
With this wall roughness, the orifice-plate meter run was calibrated with the available values, ranging from 0.35 to 0.70. The results are shown in Fig. 2. All discharge-coefficient (Cd) values found are larger than the ISO prediction, and for the larger values they are even outside the ISO uncertainty range.
The =0.65 plate was recalibrated twice in this "rough" pipe condition. In between, the plate was withdrawn and reinstalled. Then the upstream pipe of 26D length was removed, cleaned, and coated.
The resulting pipe-wall roughness was Rz-30 mm, which is not extraordinarily smooth. With this coated upstream pipe the 0.65 plate was calibrated again.
The results are shown in Fig. 3. The difference in Cd between both wall conditions is about 1.7%. Note that all other parameters were kept constant and that the change can thus be attributed only to the alteration in wall roughness.
Flow profiles, measured with a Pitot tube in both roughness situations, are well developed (Figure 4). The absence of swirl has been proven before.12 The power-law description of the profiles leads to an exponent value n = 8.2 in the rough situation and n=12.5 in the smooth situation.
This downward shift of Cd with smoother pipe is in line with other work:
- Brennan has reported the same order of magnitude.14
- The Gasunie 600-mm project was performed with coated pipes resulting in Cd values well below the ISO prediction.12
- The computational work by Reader-Harris and Keegans predicts a shift in Cd of about 1% for the roughness changes presented here.15
EXPORT-STATION TESTS
It is interesting to look at the performance of the orifice-plate meter runs in the actual situation at the export stations, given the strong influence found from wall roughness. The wall deposit is still there, resulting in a roughness somewhere between the two situations described.
The layout of the Hilvarenbeek station is such that between meter runs Nos. 5 and 6 a considerable pipe length is available, and the header between the two runs contains a shut-off valve. This made it possible to put a 600-mm turbine meter in series with four meter runs (Fig. 5), enough to measure the maximum gas flow during autumn and spring.
The flow through the station was monitored constantly using a data-acquisition system, collecting data from the turbine meter and from the orifice meter runs.
The turbine meter was calibrated before the tests started. This meter has a long record of stable performance. After the autumn 1989 tests, the meter was recalibrated and its error curve reproduced within an average of 0.05%.
A configuration with a flow straightener installed after the upstream bend, followed by a straight length of 14D, ensured the proper performance of the turbine meter.
Fig. 6 shows the hourly mean differences between the total flow through the station as measured by the orifice plates and the turbine meter. On average, the difference is in the range of -0.1% to -0.3%, which means that the actual Cd is this amount higher than the ISO prediction.
As the wall roughness is most probably between the two laboratory roughnesses, the level of Cd found is as expected. The flow rate has varied in a random way between 3,000 cu m/hr and 18,000 cu m/hr and thus the difference shown will have to be correlated with flowrate and selected meter runs.
This might give more insight into the nature of the variations shown in Fig. 6.
Long experience has shown that the turbine meter is a stable, accurate means to measure gas flow. From the point of view of working range, the turbine meter is favored.
On the other hand, the orifice plate in a practical application introduces far more additional uncertainties due to installation effects than the turbine meter.
Specifically, the influence of pipe-wall roughness is not well quantified, leading to too large a systemic uncertainty.
This has lead Gasunie to conclude that the orifice plate is unsuitable for applications in which accuracy counts, and that the turbine meter is at present the best solution for measuring the gas flow through the company's export stations.
Given the preference of Gasunie to have a back-up meter with a different metering principle from the primary turbine meter, two options were open: the vortex meter or a multipath ultrasonic (US) meter.
With both types of meters, several tests have been performed because there were few data available from the literature on field experience, especially on the large meters involved.
FIELD ENDURANCE TESTS
At the export station at Oude Statenzijl, a 300-mm pipe loop was installed upstream of the station. In this loop several meters were installed in two configurations, lasting each for about 3 months.
Originally, two turbine meters served as local references to the other meters; in the second configuration (Fig. 7), only one turbine meter was installed.
The loop also contained: a 300-mm vortex meter, a 300-mm four-path US meter (Fig. 8), an auto-adjust turbine meter, a direct-energy meter, and a reflection-type ultrasonic meter. Only the first two, the vortex meter and the four-path US meter, are relevant to the renovation project and are reported on here.
Through correct switching, this loop could run in series with the station's orifice-plate meter runs. Thus, with the two reference turbine meters, the performance of the export station could also be monitored as at Hilvarenbeek.
All meters in the loop were calibrated in the loop configuration at the Westerbork test site before and after the endurance test.
The loop has been in operation from July 1989 until January 1990. The performance of the loop was monitored constantly with a data-acquisition system. For each meter the totalized flow in a period of 10 min was obtained and converted to mass to allow intercomparison.
The aim of the tests was to collect information on:
- The stability of the various meter types over a long period of time
- Problems with meters in a field situation
- Short-term stability of the backup candidates so that a protocol for the intercomparison between primary and backup meter can be formulated later
- The difference between the reference meters and the orifice-plate meter run, which again gives performance figures of the orifice plates in their actual situation.
Fig. 9 shows the difference of the US and the second turbine meter from the first turbine meter, without correlation with flow rate, for a period of 1 month.
The two turbine meters, corrected for their calibration curve, operated to within 0.1% of each other. The meters were calibrated for a zero difference between them.
Recalibration will have to s ow whether the difference found results from installation effects or a systemic shift in one of the meters.
In the period shown, the US meter operates to within 0.3%, although it seems to shift a little upwards. Further analysis of all data is needed to obtain more insight into the US meter performance.
In Fig. 10, the comparison between the total station flow measured by the station meter runs and the first turbine meter is given. It shows differences ranging from +0.4% to -0.4% (the +1.7% on July 13 is probably due to an extremely low flow rate). This is comparable to what was found at Hilvarenbeek.
LAB TESTS
In the field tests at Oude Statenzijl, 300-mm meters were used. Extension to 500 mm is rather new to the manufacturers of the meters involved. Thus, at the Westerbork test site, a 500-mm vortex meter and a 500-mm, four-path US meter were calibrated.
These tests were aimed specifically at the shape and stability of the error curves. With the vortex meter, some tests were done in combination with a 500-mm turbine meter.
For a vortex meter, the optimum location with respect to the turbine meter is not as obvious as with the US meter.
The latter should, of course, be installed upstream of the turbine meter.
The vortex meter was run in combination with a 500-mm turbine meter over a period of 10 days. The flow rate was adjusted again and again in a random way. Readings every 15 min were taken and compared with the flow standards. Thus a "long-term error curve" emerged (Fig. 11).
The vortex meter curve has a width of about 0.4% that does not decrease significantly when longer measuring times are taken. The curve also shows a reproducible oscillating structure, which is not yet explained.
Fig. 12 shows error curves for the US meter and another 500-mm turbine meter. Between 4,000 cu m/hr and 6,000 cu m/hr, the curves result from a continuous measurement during 14 hr with 5 min readings.
At greater than 6,000 cu m/hr, a standard calibration was carried out with 30-sec readings. The downward slope of the curve above 6,000 cu m/hr is subject to further research.
It is expected that the repeatability of the US meter, now about 0.2%, will improve somewhat if longer readings are taken. A long-term performance curve, like with the vortex meter, will be obtained later.
The two turbine meters tested show an excellent reproducibility.
On the basis of the first results presented and the analysis of all other data gathered, Gasunie has decided to use the four-path ultrasonic meter as backup flow meter. Details on this analysis and the considerations will be published later.
From the experience collected in recent years by Gasunie in the 600-mm orifice-plate project and in the dedicated research projects, Gasunie decided to replace the orifice plate by full-bore turbine meters as primary volume metering devices in the export stations.
In each meter run, a four-path ultrasonic meter will be installed as secondary meter continuously to check the performance of the primary instrument.
Separate research, not reported here, indicated that the gas chromatograph is suitable for calorific-value measurement in the field. A second gas chromatograph is foreseen as a backup instrument.
The proposed system presents a new concept for reliable metering of very large-scale, gas-energy flow. Through the back-up systems, a very high availability is anticipated.
The uncertainty in the final energy account will be the smallest attainable with present technology.
ACKNOWLEDGMENTS
The authors would like to thank R. Breukers, H. Ferwerda, G. J. Gerrits, J. Huizinga, M.T. Nieboer, and H. de Vries for their contributions to this article.
REFERENCES
- "Measurements of fluid flow by means of orifice plates, nozzles and venturi tubes inserted in circular cross section conduits running full," ISO international Standard 5167.
- Kinghorn, F., "The estimation of uncertainty in practical situations with orifice meters," Int. conf. on metering of natural gas and liquefied hydrocarbon gases, London, February 1984.
- "Measurement of large gas quantities," Report of subcommittee C2, 16th IGU World Gas Conference, 1985.
- Van der Kam, P.M.A., "Gas flow measurement in the renovated export stations," Gasunie Internal Report TP/T 89.R.2069, 1989.
- Murakami, M., and Kito, O., "Effect of swirling flow component on discharge coefficients on construction meters (in case of orifice)," Bull. J.S.M.E., No. 21, p. 120, 1978.
- Mattingly, G.E., and Norman, R.S., "The decay of flow in pipes," McFaddin, S.E., 1989 International Gas Research Conference.
- Harbrink, B., Zirnig, W., Hassenpflug, H.U., Kerber, W., and Zimmermann, H., "The disturbance of flow through an orifice plate meter run by the upstream header," 5th Flomeko Conference, 1989.
- Brennan, J.A.. McFaddin, S.E., and Sindt, C.F., "Optimum location of flow conditioners in a 4-inch orifice meter," NIST Technical Note 1330, 1989.
- Smith, D.J. M., "The effects of flow straighteners on orifice plates in good flow conditions," Int. Conference Flow Measurement in the Mid 80's, 1986.
- The measurement of gas volumes by turbine meters, ISO DP 9951, 1988.
- Report on perturbation tests, ISO TC30/WG15, 1985.
- Gorter, J., "Investigation of the flow coefficient of a 600 mm orifice plate," 5th Flomeko Conference, 1989.
- Essen, G.J., Koning, H., and J. Smid, "Time behaviour of turbine meters-Statistical analysis of (re)calibration results of turbine meters," 5th Flomeko Conference, 1989.
- Brennan, J.A., McFaddin, S.E., Sindt, C.F., and Wilson, R.R., "Effect of pipe roughness on orifice flow measurement," NIST Technical Note 1329, 1989.
- Keegans, W., and Reader-Harris, M.J., "Comparisons of computation and LDV measurement of flow through orifice and perforated plates, and computation of the effect of rough pipework on orifice plates," Procs. Int. Symp. Fluid Flow Measurement, American Gas Association, 1986.
Copyright 1990 Oil & Gas Journal. All Rights Reserved.