Flow Control Network
August 2006
Evolution of Ultrasonic Flow Measurement
From Poor Performer to Recognized for Custody Transfer
By Steven S. Yon

Ultrasonic flowmeters of various types have been in use for more than 20 years. Doppler-based technologies were the first ultrasonic systems introduced on the open market, and initial designs showed signs of promise as an alternative to traditional flow measurement technologies. Unfortunately, early-generation Doppler systems ultimately proved to be poor performing instruments, and, as a result, gave ultrasonics a rather lowly reputation among flowmeter users.
   

Figure 1. Doppler systems generate a measurement by sending a single beam into a flow stream and reflecting it off of particles back to the transducer.
Over the years, however, continuing enhancements of Doppler technology, as well as the emergence of the transit-time concept, have dramatically improved the accuracy of ultrasonic systems. In fact, the high accuracy of modern-day ultrasonic meters was acknowledged in January 2005 by the American Petroleum Institute (API, www.api.org) when it approved and published the first edition of Chapter 5.8 “Manual of Petroleum Measurement Standards for Measurement of Liquid Hydrocarbons by Ultrasonic Flow Meters Using Transit Time Technology.” The publication of this document officially cleared the way for custody-transfer ultrasonic flow measurement of oil and natural gas, validating the high performance of this technology for flowmetering.

Doppler Technology

Figure 2. Transit-time systems generate a measurement by calculating the amount of time it takes for a signal to cross a flow stream traveling with and against flow.
Doppler meters work differently than transit-time meters. Most Dopplers transmit a single beam (instead of pulses) at an angle into the flow and reflect the signal back to the transducer to achieve a flow measurement.  
   
The reflection occurs when the signal bounces off of particles in the flow stream (Figure 1). The speed and direction of the particle causes a frequency shift of the returned signal (F1 vs. F2). The average is then taken and the reflected signals approximate the flow velocity.
   
For Doppler technology to perform properly, the flow stream must contain suspended bubbles, solids, or particulates to reflect sound energy. Too many particles though, or particles that are sponge-like, may limit the ability of the signal to reflect. The particles must also be large enough to provide a good reflection.
   

Figure 3. The transit-time principle for ultrasonic flowmetering
Normally the ultrasonic signal extends only into the peripheral flow, and the indication is dependent on the flow profile. Pipe material (Snell’s Law of critical angle), condition of pipe (rust, scale), or any liner can dramatically impact the measurement. Loss or deterioration of coupling compound can influence signal penetration, meter performance, and accuracy.
   
Doppler ultrasonic systems are easy to install on existing pipe, and they are environmentally acceptable in most applications. They also offer a low-cost alternative to some other flow measurement systems and are noninvasive, thus eliminating wear on parts exposed to the flow stream.

Transit-Time Technology
The explanation in Figure 2 has been used many times and is an excellent visual of the principle of transit-time flow measurement. The canoe crossing the river at an angle with the current (flow) needs less time than crossing against the flow. This difference between the time it takes to cross the river with and against the current is dependant on the velocity of the water.
   

Figure 4a. Three or more paths are required to determine the flow velocity profile in one geometrical plan.
To translate this concept to a flow measurement application, consider a pair of transducers, A and B, located upstream and downstream of the flow respectively. The difference of the ultrasonic transit times, TAB and TBA, from A to B and B to A provides the average flow velocity of the liquid VAB between A and B (Figure 3).

The volumetric flowrate, QV, is proportional to the average flow velocity VAB when a hydraulic coefficient K is defined.

Flow Profiles
When considering ultrasonic flow measurement for a given application, there is one important question to consider: Will a single path give you the accuracy you need for your application, or will you require more paths?
   

Figure 4b. Information about the flow velocity profile in three different geometrical planes (three-by-three) is necessary for the correction of asymmetric profiles.
The more paths that pass though the pipe, the more accurate your representation of flow profile will be. In addition, more beams increase the ability to maintain accurate measurement as the velocity and viscosity of the measured fluid change.  
In order to perform an accurate volumetric measurement, the instrument must be able to measure the three-dimensional velocity profile. This is the only way to correctly compensate for swirls, asymmetric flow profiles, and laminar-to-turbulent flow velocity profile changes.
   
API Chapter 5.8 recognizes that more paths represent better accuracy by limiting its recommendations to spool-type, two-or-more path ultrasonic flowmeters with permanently affixed transducers. Three or more paths are required to determine the flow velocity profile in one geometrical plane (Figure 4a). Information about the flow velocity profile in three different geometrical planes (three-by-three) is necessary for the correction of asymmetric profiles (Figure 4b).
   
With swirls, the flow velocity of the liquid is no longer parallel to the axis of the pipe. The flow velocity has a perpendicular component that might generate an inaccurate flowrate measurement if not properly compensated. Swirls are usually generated by single or double bends, partially blocked strainers, or other obstacles located upstream from the meter.
   

Figure 4c. Using crossed-path configuration for two-by-three-by-three ultrasonic paths to compensate for swirls
Figure 4c demonstrates the use of crossed-path configuration for two-by-three-by-three ultrasonic paths to compensate for swirls. In this case you have 18 flow velocity calculations in nine different geometrical planes. If you consider a swirl component, the flow velocity measurements though the ultrasonic paths AB and CD are higher and lower than the actual axial flow velocity respectively. The sum and subtraction of the two measurements provide the percentage of swirl present in the pipe.
   

Figure 4d. A sample from an 18-beam ultrasonic flowmeter, detailing velocity profile changes at different flows
Different manufacturers handle swirls and profiles in different ways, each using proprietary algorithms and numerical calculation techniques to compute the average flow velocity. The end result is the larger the number of measuring paths — whether crossed or chordal — the greater the measuring accuracy at varying Reynolds numbers.
   
Figure 4d is a sample from an 18-beam ultrasonic flowmeter, detailing the velocity profile changes at different flows and the importance of multiple three-dimensional beams to measure and compensate
effectively.

Installation & Proving
Custody transfer for hydrocarbons is one of the most scrutinized and evaluated measurements. With oil and natural gas prices reaching all-time highs, accurately measuring hydrocarbons is paramount to the financial success (or loss) of any company. The International Organization of Legal Metrology (www.oiml.org) and API have established standards to assure this level of measurement accuracy. Figure 6 shows a sample from the API Chapter 5.8 document detailing recommendations for installation.
   

Figure 5. A sample from an 18-beam ultrasonic flowmeter used to analyze and visualize the effects of different obstructions and upstream conditions
As shown in Figure 5, the upstream condition can dramatically impact the velocity profile and swirl. As a result, flow conditioning may be required. The API is currently running tests using an 18-beam ultrasonic meter to analyze and visualize the effects of different obstructions and upstream conditions. (Figure 5 is an example of this data.) Therefore it is important for the end-user to consult the manufacturer for design considerations.
  

Figure 6. API recommendation for ultrasonic installation


Meter proving is required to establish a factor to correct ultrasonic flowmeter performance under actual operating conditions and should be done according to API Chapter 4.8 and 5.8. Turbine meters and positive-displacement meters average out velocity variants. Ultrasonic meters, on the other hand, continuously record snapshots of the flow, resulting in as many as 900 flow snapshots per second depending on the manufacturer.  

All instruments that produce electronic pulse outputs have a slight delay in the output pulses. This delay can lead to a bias error in the calculated meter factor, depending upon the magnitude of the flowrate change during a prove. As such, care should be taken to maintain a consistent flow during a prove.
   
Ultrasonic flowmeters may produce wider repeatability ranges than are typical for mechanical devices. A range exceeding 0.05 percent in five runs does not mean that an ultrasonic flowmeter is defective, or that a meter factor cannot be established within the required uncertainty. Rather, it means the prover did not have enough volume, and proving runs are necessary.
   
Ultrasonic flowmeter performance verification can be ascertained by conventional means with API MPMS Chapter 4.8. The most comprehensive approach to accomplishing this level of repeatability relies on determining the acceptable prover volume. Ultrasonic meters will typically require a larger prover volume to verify an uncertainty of +/-.027 percent. Given the larger prover volume, it follows that more than five proving runs may be required to verify the meter’s performance.  
   
There is no difference in a repeatability range of 0.05 percent in five runs and a range of 0.12 percent in 10 runs — they are the same. The operator is advised to select the appropriate number of runs incrementally until the repeatability range falls within the limits of Table 1.  
   
Table 1 details the number of runs to achieve the uncertainty and Table 2 is the suggested prover volume required for specific line sizes. Experience has shown that when using ball provers, the required meter factor accuracy is typically achieved with fewer than 10-12 runs. This is why at many ultrasonic flowmeter installations a master meter is used to prove the ultrasonic flowmeter. Using a master meter for proving provides several advantages:
• Reduces wear on small volume provers
• Eliminates requirement for many runs
• Eliminates need for on-site ball prover

Steven S. Yon is general manager of the Americas business segment of Faure Herman Meter Inc. Mr. Yon has been in the oil and gas industry for over 30 years. He is a voting active member on several API committees, including Chapter 5.8, Ultrasonic Measurement. He has been a long-time instructor at the Petroleum Extension Division (PETEX) of the University of Texas. And he previously served as president and CEO of Enraf Tank Gauging. Mr. Yon can be reached at syon1@fhmi.com or 713 623-0808.

www.faureherman.com

References
1.    API Chapter 5.8, “Measurement of Liquid Hydrocarbons by Ultrasonic Flow Meters.”
2.    “Ultrasonic Flow Measurement,” Dipling Fredich Hofmann, Krohne GmbH.
3.    “Ultrasonic Flow Measurement,” Caldon.
4.    “Ultrasonic Flowmeters,” Control Engineering.