Custody Transfer Implementation of Multipath Ultrasonic Meters
MICHAEL BROWN, NOVA Gas Transmission Ltd, Calgary, Alberta, Canada.

ABSTRACT
Multipath ultrasonic meters are growing in popularity throughout North America as a costeffective means of custody transfer measurement for high pressure natural gas. NOVA Gas Transmission (NGT) has been evaluating multipath ultrasonic meters for custody transfer measurement since 1995, and is now realizing the benefits of this. new technology.

NGT recently implemented multipath ultrasonic flow meters as the primary measurement device at a custody transfer meter station. The January Creek meter station (MIS) is a bi-directional facility, consisting of a NPS 20 multipath meter in parallel with a NPS 8 multipath meter measuring gas for a new natural gas storage facility. This paper will outline the benefits and' decisions involved in the implementation of multipath ultrasonic meters for custody transfer measurement, including a design comparison with a multi-run turbine
facility.

Additionally, the use of a data acquisition system (DAS) to build a reliable, historical performance database for the meters will be discussed. The implementation of a comprehensive data acquisition and monitoring system allows NGT to monitor the meter performance and collect long-term performance information. This information is then used to characterize each . meter's performance, and assist in ensuring continued system integrity for NOT customers.


INTRODUCTION
NGT's 13,500-mile system transports natural gas for use within Alberta and to provincial boundary points for connection with pipelines serving markets elsewhere in Canada and the United States. The system moves over 18 percent (4.4 trillion cubic feet) of the natural gas produced annually in North America and more than 80 percent of Canadian natural gas production. NGT's system consists of 49 compressor stations, 938 receipt metering points, and 166 major delivery points.

NGT has gained significant knowledge and learnings about the benefits and concerns in using single-path and multipath ultrasonic meters through it's evaluation work over the past few years (Rogi et all, 1996; Karnik et all, 1997; and Rogi et all, 1997). As a result, NGT is moving forward with a controlled implementation of this new metering technology for custody transfer meter stations. The first opportunity to implement custody transfer ultrasonic meters was the January Creek WS. The January Creek MIS meters the gas for a new natural gas storage facility in central Alberta.

The meter station design was based on NGT's' learnings and those of others in the industry (Lygre et all, 1995; Grimley, 1996). A conservative approach was taken in areas of the facility design that NGT felt had not yet been thoroughly researched, while still achieving significant capital and operational savings.

The facility has been operational since April 1997, and has been a success story for NGT. NGT is now continuing to evaluate the use of this technology at new large volume receipt and delivery meter stations (typically NPS 10 and larger) on a facility-by-facility basis. This controlled implementation is based on maintaining the measurement integrity of NGT's pipeline system, while providing quality measurement facilities and achieving capital and operational savings.

Figure 1 Meter Station Layout (Click on the pic to enlarge)

STATION DESIGN DETAILS
The measurement capacity for the January Creek WS is 500 mmscf/d of natural gas bidirectionally. One NPS 20 ultrasonic meter would be able to handle this volume of gas, but could not accurately measure the lower flow rates expected from the producer. Therefore, a NPS 8 ultrasonic meter was placed in parallel with the NPS 20 and run-switching was provided to accommodate for the changes in flow rate.

Meter Runs
Without having substantial research work completed on installation effects at the time of the station design, a conservative approach to upstream piping was used. The upstream and downstream NPS 20 meter run lengths were set at 30 diameters (D) (Figure 1) with a perforated plate flow conditioner (NOVA 50E; Measurement Canada, 1997). The NPS 20 meter run was designed to allow the replacement of the meter tube and meter with a NPS 24 meter and meter run. This added very little cost to the station, and allows for cost effective station expansion in the future.

The NPS 8 meter run was made the same length as the NPS 20 meter run, and therefore had several diameters (approx. 130 D) of straight upstream and downstream piping, and did not require any flow conditioning.

Being a bi-directional facility, the ultrasonic meters would have the thermowells located upstream in one of the two flowing directions. Not having significant research data on how thermowells impact the performance of ultrasonic meters, NGT decided to locate the thermowells 10D away from the meters. This would help reduce any thermowell effects from influencing the meters performance, yet provide proper temperature measurement for the meters.

Instrumentation
The ultrasonic flow meters were connected to the flow computers via a frequency output. The meters also communicated, via a serial link, with a remote telemetry unit (RTU) to provide diagnostic information about the meter performance.

NGT's standard static pressure (smart transmitters) and temperature equipment (platinum RTDs) was used at this facility. Due to the harsh climatic conditions that can occur at this location, the meters and all related instrumentation, including pressure and temperature transmitters, were located in the temperature controlled meter run building.

COMPARISON TO TYPICAL DESIGN
The January Creek MJS measurement requirements were similar to those of another NGT meter station built for a storage facility in 1994. The 1994 meter station design consisted of four NPS 12 turbine meters and the appropriate yard valves to allow for the changing of flow direction through the station.

The major differences between the design of these two facilities are shown in Table 1. The estimated total capital savings for the January Creek M/S is $300k CAD. The operation and maintenance costs are also a significant factor for the design of this facility. Firstly, the number of transmitter calibrations is reduced.

Also, automated monitoring of ultrasonic meter diagnostics eliminates the need for routine meter inspections such as the turbine meter spin test.

METER INFORMATION
In addition to the NPS 8 and NPS 20 ultrasonic flow meters at this location, the producer operating the storage facility is also operating a NPS 24 multipath ultrasonic meter in series with the NGT January Creek WS. The information from all of the meters is shared between NGT and the producer.

NPS 20 Meter
The NPS 20 meter was flow calibrated at the NM Westerbork facility in January, 1997. The meter was flow calibrated from I m/s to 25 mls in both directions. The calibration results were (see A.G.A. Report No. 9 for definitions):

• Repeatability: ± 0.1 % for qmin to qmax
• Maximum Error: 0.6%
• Peak-to-Peak Error: ± 0.18% for qt to qmax
• Flow Weighted Mean Error(FWME): -0.3%

The calibration data was entered into the flow computer in the form of a multiple K-factor table. This table helped to remove the ± 0.2% non-linearity,

NPS 8 Meter
The NPS 8 meter was initially flow calibrated at the Rurhgas PIGSAR facility in January, 1997. The results of this testing did not meet NGT's performance requirements, and therefore the meter was rejected. The transducers were replaced, liquid drain-holes were plugged, and the meter was then recalibrated at both the Gas Research Institute Metering Research Facility (GRI MRF) and PIGSAR. The results from these two facilities were similar, and indicated that the meter still did not meet NGTs linearity specification of t 0.2%. A decision was made to use this meter temporarily, correcting for the non-linearities in the flow computer. A replacement meter was to be provided at a later date.

To date, the NPS 8 meter has operated very little due to the manner the facility is operated. The producer at this facility is usually flowing above the range of the NPS 8 meter, and therefore the meter is usually only operated when the storage facility is starting up or shutting down.

DATA ACQUISITION AND TRENDING
The January Creek WS data acquisition system (DAS) consists of a dedicated personal computer running a human-machine-interface (HMI) software package, an RTU, and flow computers. The HMG system collects meter station information from the RTU, monitors and logs data. The primary function is to collect supplemental diagnostic information not captured through the custody transfer flow computer system.

Meter flow and diagnostic information is polled from the RTU and logged to a file once every fifteen seconds. Gas composition data is polled every two minutes and written to the file once every four minutes. Data is also displayed for the field operators to monitor station information. The information logged to the file includes: total volumes, total energies, pressures, temperatures, frequencies, flow velocity for the meter and each individual chord, meter and chord status codes, upstream and downstream transit times, velocity of sound for the meter and each chord, and information from the producer's meter.

The files generated from this acquisition system are remotely retrieved, and postprocessed. The post-processing performs three main functions:

1. Validate meter calculations for chordal velocity (Eq.l) and velocity of sound (Eq. 2), as well as the meter velocity and velocity of sound.

2. Average the data into daily averages. Data is also averaged into separate 'buckets' depending on the average meter velocity for each record. These buckets are recorded for velocities between 0 and 25 m/s, in 1 m/s increments. One bucket average is calculated for each day, and for each velocity range encountered. For example, if the meter flows between 4 and 6 m/s during the period of one day, one bucket average would be created for the 4-5 m/s data, one for the 5-6 nits data, and one for the entire day. This data is stored in an Oracle database for future analysis.

3. Suspect meter diagnostic information is also flagged during the processing. This helps to catch diagnostic codes which may occur periodically, that are not significant to the daily operation of the meter, but may be in the long term.

Chordal Velocity Analysis
One of the key parameters monitored is the individual chord velocities. The ratio of chord velocity to meter velocity is monitored over time (Figure 2.a), and for various velocity ranges (Figure 2.b).
From the daily average values shown in Figure 2.a, the chordal velocity ratios are stable. In Figure 2.b, the ratios for chords B and D are somewhat dependant on velocity. From this trend, the long-term performance can be monitored. Each velocity bucket can also be displayed over time to better determine if a drift is occurring (Figure 2.d). This information will also be valuable should the replacement of transducers be necessary.
The positive flow direction chordal velocity ratios are shown in Figure 2.c. In this flow direction the meter was operated over a wider velocity range than in the reverse flow direction (Figure 2.b). From this data the dependance on flow velocity becomes more apparent. The lower velocity characteristics for forward and reverse flow directions are opposite for each of the chords. This may indicate that these characteristics may be caused by a low velocity differences between each of the chords.

Figure 2.a Meter #1 (NPS 20) DailyChordal Velocity Ratios(Reverse Flow Direction)

Velocity of Sound Analysis
The velocity of sound (VOS) is calculated within the ultrasonic meter for the each of the chords of the meter and averaged. The VOS is also determined in the post-processing program using the A.G.A. Report No. 8 Equation of State, gas composition, pressure and temperature. These values for VOS can be monitored, and used as a means of monitoring the performance of the meter prior to and after the replacement of electronic components.

Figure 2.b Meter #1 Bucket Chordal Velocity Ratios (Reverse Flow Direction)
Figure 2.c Meter#1 Bucket Chordal Velocity Ratios (Forward Flow Direction)
Figure 2.d Meter #1 Bucket #4 Chordal Velocity Ratios
Figure 3.a Meter #1 VOS% Difference Compared to A.G.A. VOS

Figure 3.a shows the percent difference between the velocity of sound calculated by the meter and the velocity of sound calculated using the A.G.A. calculation within the post processing program.

Figure 3.b compares each of the chordal velocity of sound measurements with the meters mean velocity of sound. Chords D and C appear stable over time, where chords A and B fluctuate (± 0.05%).
The VOS data shown in Figure 3.b can also be plotted against other variables such as: flowing temperature, pressure, gas velocity, and meter VOS. This feature allows flexibility in data analysis, and allows the user to look for trends.

Volume Comparison
The total standard volume measured by the NOT meter station can be compared to the volume measured by the producer's meter. The results of this comparison are shown in Figure 4. The comparison shows a difference between the systems of approximately +1.5% to +2.0%. A large portion of this discrepancy can be attributed to differences in static pressure measurement. The producers static pressure transmitter is reading approximately 0.8% higher. A portion may also be due to the producer's meter not being flow calibrated.

Figure 3.b Meter #1 Chord VOS % Difference Compared to Meter VOS

Another feature of the meter station design is the ability to place the NPS 8 meter in series with the NPS 20 meter. This feature added one NPS 8 valve and 50 meters of NPS 8 pipe to the cost of the facility, but allows for comparison between these two meters. This comparison has been performed periodically, but is intended as another check prior to and after the replacement of electronics or transducers. The results of a couple of these tests are shown in Table 2.

Figure 4 Volume Comparison Between NGT and Producer Meter

DISCUSSION
Based on the operational experience to date, some changes to the design of a similar future facility can be recommended. Firstly, the NPS 8 meter run has not been utilized consistently in a range that the NPS 20 meter could not meter accurately. Especially considering that the NPS 20 meter was intended to be operated down to 1 rule. If this lower limit were reduced to 0.5 m/s, then the turndown ratio of this single meter could handle the flow ranges seen to date.

Secondly, the validation routines currently performed in the post-processing application could be moved to the RTU, and the averaging functions moved to the HMI. This would help provide more real-time analysis, and eliminate the need for post-processing.

Due to some further investigation into installation effects (Grimley, 1997; Karnik et all, 1997; and Karnik et all, 1998), and some computational modelling of multipath meters and installation effects (Studzinski et all, 1998) the upstream meter run length could now be reduced to approximately 11 diameters from the perforated plate flow conditioner.

A review of the horizontal separator design indicated that gas could be flown through the separator in both directions. The bypass loop around the separator could be removed in future designs, and this would eliminated three NPS 24 valves and NPS 24 piping.

Beyond these changes, NGT is very satisfied with the meter station design and performance, and will be looking at this station for guidance for future custody transfer ultrasonic meter stations.

CONCLUSIONS
Many industry users are finding, capital and operational savings are being realized through the implementation of rnultipath ultrasonic meters. NGT has approached this technology methodically, investing in thorough evaluation activities, and is now able to realize both shortterm and long-term benefits.
The utilization of a comprehensive data acquisition and monitoring has added great value to this implementation, and helps ensure system integrity. Monitoring detailed ultrasonic meter diagnostics, such as chordal VOS and velocity, is an excellent operational tool.
NGT will continue the implementation of this technology and will continue looking for opportunities to improve and learn from other users in the industry. NGT's next opportunity to use this technology is a 2.2 bcf facility consisting of three NPS 30 multipath meters in parallel. The success of this implementation is expected to result in a capital savings of approximately $5MM CAD.

ACKNOWLEDGMENT
The authour would like to acknowledge the efforts of the January Creek design team, including Gordon Pruden, Jaye Selin, Wayne Snyder, and Ron Wong. Also, the maintenance personnel, including Dennis Klemp and Stan Pierog, for their input and support throughout this project.

REFERENCES
A.G.A Report No. 9, "Measurement of Gas by Multipath Ultrasonic Meters", American Gas Association, Arlington, Virginia.
Grimley, T., 1996, "Multipath Ultrasonic Flowmeters Performance", A.G.A. Operating Sections Operations Conference, Montreal, Quebec, Canada.
Grimley, T., 1997, "Performance Testing of Ultrasonic Flow Meters", North Sea Flow Measurement Workshop, Kristiansand, Norway.
Karnik, U., Studzinski, W. and Rogi, M., 1997, "Performance evaluation of flinch multi-path ultrasonic meters", A.G.A. Operating Sections Operations Conference, Nashville, USA.
Karnik, U., Studzinski, W., Gerrligs, J. and Rogi, M., 1998, "Effect of Flow Conditioners and Pulsation on the Performance of flinch Multi-PathUltrasonic Meters", ASME International Pipeline Conference, Calgary, Alberta, Canada.
Lygre, A., Lunde, P.,. Froysa, K.E., 1995 "Present Status and Future Research on Multipath Ultrasonic Gas Flow Meters", GERG Technical Monograph.
Measurement Canada, 1997, "Provisional Specification, PS-G-05-E", Measurement Canada, Ottawa, Ontario, Canada.
Rogi, M., Shen, J. and Karnik, U. 1997, "Field Performance of flinch Multi-Path Ultrasonic Meters", ASME Fluids Engineering Conference, Forum on Fluid Flow Metering, Vancouver, British Columbia.
Studzinski, W., Jelen, J. and Brown, M., 1998, "Application of CFD to the Design of Multirun Station with Ultrasonic Meters", ASME International Pipeline Conference, Calgary, Alberta, Canada.

Click on the pic to enlarge