Pipeline & Gas Journal
January issue 2000:

Design Consideration

Using Ultrasonic Flow Meters
At Custody Transfer Stations

by Edgar Bowles, Section Manager, Department of Fluids Engineering, and Terry Grimley, Principal Engineer, GRI Metering Facility, Southwest Research Institute, San Antonio, Texas

Historically, the flow meter of choice for natural gas transmission pipeline operators in the United States has been the orifice meter. However, over the past five years, there has been a heightened interest in using ultrasonic flow meters for pipeline custody transfer measurement. Driving this interest are some operational and economic advantages that ultrasonic meters enjoy over conventional orifice meters. For instance, ultrasonic flow meters typically have a broader flow rate range (i.e., a larger turndown ratio) than do orifice meters. Ultrasonic meters also have no flow restriction in the meter element, no moving parts, minimal secondary instrumentation requiring periodic re-calibration, good measurement repeatability, and bi-directional flow capability. However, as with any measurement technology, ultrasonic flow meters have their operational limitations, so good engineering practice must be followed to achieve the desired results for a given flow meter installation. In this article, we will discuss some important aspects to consider when designing a custody transfer meter station that incorporates an ultrasonic flow meter.

Industry Guidelines
Measurement accuracy is a primary consideration when designing a custody transfer meter station on a gas pipeline. To ensure that the highest attainable accuracy is achieved for a given application, meter station designers typically rely on established industry standards or guidelines. The first recommended practice[1] for ultrasonic gas flow meters was published by the American Gas Association (A.G.A.) Transmission Measurement Committee in June 1998. A.G.A. Report No. 9 is fundamentally different from previous gas industry flow meter specifications in that it is a performance-based specification that does not include detailed mechanical installation guidelines. Report No. 9 also states that the meter station piping configuration is to be specified by the meter manufacturer.

A.G.A. Report No. 9 specifies limits on maximum error, maximum peak-to-peak error, repeatability, resolution, and zero-flow reading. Report No. 9 also specifies that the measurement bias error attributed to the effect of the meter station piping configuration must not exceed +0.3%. The performance summary specifications from A.G.A. Report No. 9 are shown on Figure 1. In Figure 1, qmax is the maximum gas flow rate through the meter that can be measured within the error limits, qt is the transitional gas flow rate below which the expanded error limit is applicable, qmin is the minimum gas flow rate through the meter that can be measured within the expanded error limit, and qI is the actual measured gas flow rate passing through the meter under a specific set of operating conditions. Large meters are defined by Report No. 9 as those having a nominal diameter of 12 inches or more. Small meters are those less than 12 inches (nominal) in diameter.
From the information provided in A.G.A. Report No. 9, how does a meter station designer know if a given ultrasonic meter installation meets the requirements of the report? More importantly, how can a designer determine if the greatest measurement accuracy is being achieved with a particular meter station design? There are several means available to help answer these important questions.

Flow Field Effects
First, a meter station designer must have a fundamental understanding of how ultrasonic meters work. Ultrasonic meters measure the gas velocity along various acoustic paths through the cross-section of a pipe. A mathematical algorithm converts these individual velocity measurements into a bulk flow rate. Clearly, ultrasonic meters are sensitive to the velocity profile of the gas stream at the point of measurement. Research has shown that distortions in the flow stream, such as velocity profile asymmetry, swirl, or the combination of swirl and asymmetry, can produce bias errors in the flow rate determination.[2] To only a limited degree have ultrasonic meters demonstrated an ability to compensate or correct for measurement bias errors resulting from flow profile distortions. (Numerous articles describing the workings of an ultrasonic meter have appeared in this journal and others over the past several years and the reader is referred to these for a more in-depth explanation of the measurement of gas by ultrasonic meters).

A meter station designer should, therefore, have a reasonably good idea of the velocity profile that will be produced by a given meter station configuration. Otherwise, the designer runs the risk that a significant measurement bias may be built into the design. We know that the gas velocity profile shape is controlled by the piping geometry through which the gas flows. Elbows, tees, valves, pressure regulators, and other elements in the flow stream can create velocity profile asymmetry, swirl, or a combination of the two.

Determining the velocity profile for a given meter station configuration can be extremely challenging and costly. There are test data in the open literature documenting the velocity profile produced by selected piping configurations.[3] Figure 2 shows two different installation piping configurations under recent test at the GRI Metering Research Facility (MRF). The database of verified piping configurations is relatively small, but growing. Meter manufacturers also have verification test data for a limited number of meter installation configurations. Thus, one option for meter station designers is to choose a meter station piping configuration for which the flow characteristics are known.

Alternatively, the designer can experimentally verify the flow performance of a specific installation configuration. This is usually a costly option. Computational modeling may also be used to determine the flow field produced by a given meter station design. However, computer models tend to be difficult to set up and become rather imprecise when several bends or other obstructions (typical of most meter station configurations) are placed in the flow stream.
If the expected velocity profile cannot be adequately characterized at the design stage, the designer may elect to include a flow conditioner upstream of the meter. The function of a flow conditioner is to eliminate (or at least reduce to an insignificant magnitude) any flow distortions created by the piping and flow control elements upstream of the ultrasonic meter. This helps ensure that the flow measurement is not biased by any flow stream distortions. A number of flow conditioners are currently available for this purpose. Some example flow conditioners are pictured in Figure 3.

As a word of caution when considering the inclusion of a flow conditioner as part of a meter station installation, recent research results from the GRI MRF[2] suggest that a given ultrasonic meter/flow conditioner combination produces unique measurement performance. That is, the total measurement error of a given ultrasonic flow meter may vary, depending on the configuration of the flow conditioner and upstream piping used in combination with the meter. Thus, care should be taken when selecting a meter/flow conditioner combination. Otherwise, measurement bias errors may be unknowingly introduced as a result. Because of the large number of ultrasonic meter/flow conditioner configuration combinations, it is unlikely that a universal installation guideline can ever be established without penalizing (with long upstream pipe length requirements) those meters and flow conditioners capable of performing acceptably well in less stringent installations.

Meter Calibration
During the meter station design phase, attention should also be given to meter calibration requirements. Ultrasonic meter calibration requirements are a function of the application, but for most custody-transfer measurement, meter calibration is considered essential. Ultrasonic meter manufacturers are usually able to deliver ultrasonic meters that perform within the stated error limits specified on Figure 1, without the meters being flow calibrated. However, Figure 1 allows up to +1.0% measurement error, and most custody transfer applications demand a much smaller error limit.

Fortunately, the total measurement error associated with an ultrasonic flow meter can usually be reduced significantly by flow calibrating the meter to eliminate the bias errors. Figures 4 and 5 illustrate how effective flow calibration can be at eliminating bias errors. Figure 4 is a compilation of all “as found” meter calibrations performed on “large meters” (i.e., in this case, meters 12 to 16 inches in diameter) at the GRI MRF over a recent 12-month period. “As found” means that the meters were tested at the GRI MRF without having undergone any prior flow calibration and with the meter setup file configured as received from the meter manufacturer. Note that in Figure 4, very few test points lie outside the error bounds specified in A.G.A. Report No. 9 (i.e., Figure 1). Also note, however, that many of the test points are somewhat removed from the zero error line.
Now observe Figure 5, which is the compilation of “as left” meter calibrations for the same group of meters referred to on Figure 4. “As left” means that each meter was flow calibrated against the flow references at the GRI MRF (i.e., sonic nozzles) and then the final meter calibration factor was corrected. The data shown on Figure 5 are for the confirmation tests that verified meter performance after correction of the meter factor. Note that the data on Figure 5 are much closer to the zero error line, indicating that most all of the measurement bias error was eliminated.
To illustrate the economic implications of calibrating ultrasonic meters, let us study the following hypothetical case. Transmission-grade natural gas at a line pressure of 850 psig is flowing at a steady velocity of 50 feet per second (i.e., at about the mid-range of a typical meter) through a 16-inch diameter ultrasonic meter. The meter was not flow calibrated prior to installation and was delivered from the manufacturer with a -0.7% meter bias (i.e., the specified maximum allowable in Report No. 9). If the value of the gas being transported is approximately $2 per thousand standard cubic feet, the -0.7% meter bias will result in a $1.8 million error in favor of the buyer over one year’s time. The current cost of flow calibrating a meter of this size and capacity at a reputable flow lab is on the order of $15,000 to $30,000. Thus, the cost of a flow calibration to correct for the meter bias could be recovered in approximately three to six operating days.

Two different meter calibration methodologies are referenced in A.G.A. Report No. 9. These are Zero-Flow Verification and Flow Calibration. A Zero-Flow Verification test checks the transit-time measurement system of an ultrasonic meter. In this test, the meter ends are closed off and the meter body is pressurized with a gas of known composition. With no gas flow, the meter measures the speed of sound of the test gas. The measured result is compared to the known (theoretical) value. This test confirms the functionality of the ultrasonic transducers and the electronic timing circuitry. This test does not confirm the measurement performance of the ultrasonic meter under flowing conditions. The Zero-Flow Verification is typically performed before the meter is flow calibrated initially and before the meter is installed in the field. The test can also be performed on a periodic basis, after the meter has been installed at the field site.

Meter performance under flowing conditions can only be verified by a Flow Calibration test. As noted above, a flow calibration test takes into account operational and installation effects that may adversely affect meter performance. Flow calibrations can be performed either in-situ, using a reference test flow meter plumbed in series at the meter station, or off site, at a flow calibration test facility. If the meter is to be calibrated in-situ, then consideration should be given at the design stage as to all valving and piping requirements necessary for a reference test flow meter to be installed onsite. Although a number of pipeline operators have performed in-situ flow meter calibrations in the past, there are no industry standards or guidelines for in-situ or field meter proving. However, two technical references on the subject have been produced by Park, et al.[4] and by Gallagher.[5]

Offsite flow calibration of an ultrasonic meter is, in most cases, the preferred method for verifying meter accuracy. Worldwide, there are a number of flow laboratories capable of providing precision flow calibrations. The calibration test conditions should replicate, as closely as possible, the field service conditions for the meter being calibrated. The general consensus of the user community is that the flow meter calibration should be performed with the same meter tube piping and flow conditioner (if one is to be used) that will be used at the field meter station. This allows measurement biases caused by the installation configuration to be corrected for during the meter calibration.

Selection of a meter calibration laboratory also requires some forethought. Most high-quality test flow labs have a total measurement uncertainty of about +0.2% to 0.3% (for two standard deviations). This measurement uncertainty level has been confirmed experimentally through inter-laboratory testing. However, each lab does have its own individual measurement bias errors. Therefore, if an ultrasonic meter is to be calibrated more than once, it is recommended that all of the calibrations be performed at the same test lab. Otherwise, changes observed in the meter performance over time may be due to variations between the test labs, rather than variability of the meter.

Other Considerations
Two other operational effects may be important and should be considered at the meter station design stage. First, ultrasonic meters can be adversely affected by nearby sources of ultrasonic noise (i.e., energy emitted in the frequency range between approximately 50 to 500 kilohertz). Such noise sources may include “quiet” valves designed to reduce audible flow noise, pressure regulators, and other significant flow restrictions in the pipeline. Meter manufacturers are actively pursuing solutions to this problem and meter station designers should consult the manufacturers for assistance. Field experience has shown that it is best to locate the meter upstream of potential noise sources. In addition, placing pipe bends between the meter and the noise source helps attenuate the extraneous ultrasonic noise.

In addition, the cleanliness of the gas stream can affect ultrasonic meter performance. Buildup (from compressor oil, condensate, or other sources) on the face of an ultrasonic transducer can prevent the unit from transmitting ultrasonic pulses through the gas stream. This can lead to what is called ultrasonic path “dropout.” Path dropout can add to measurement error. Also, dispersed liquids in the gas flow stream (e.g., mists or aerosols) can attenuate the ultrasonic signals and completely incapacitate the meter. Thus, the cleanliness of the gas stream and its effects on the interior of the meter are critical to good meter performance.

Conclusions
Acceptance of ultrasonic flow meters in natural gas transmission pipelines is growing rapidly worldwide. By using good engineering judgment and careful design practices, users of ultrasonic meters can realize the benefits of this new technology. P&GJ