Clamp-on transit time flowmeters for liquids have been around since the 1960s. Now the technology extends to gases.
Larry Lynnworth, Panametrics, Inc.
This article was first planned as a sequel to “High-Temperature Flow Measurement with Wetted and Clamp-On Ultrasonic Sensors” , bringing ultrasonic flowmeter buffers up to date for liquids and gases. Buffers are used to extend the temperature range of ordinary ultrasonic flowmeters to very high or very low temperatures (see the sidebars, “What’s a Fiberacoustic Bundle Good For?” , and “Hockey Sticks for Clamp-On Measurement of Hot or Cold Liquids”). However, a breakthrough in ultrasonic clamp-on gas flowmetering occurred in the interim [2,3]. Because we believe this breakthrough to be of more interest to many readers than hearing about buffer developments and improvements, the present article focuses on clamp-on meters, generalized now to fluids, i.e., liquids and gases. We hope that even those readers who do not have an immediate application for clamp-on flow measurement of compressed air, steam, natural gas, hydrogen, or other gas will find that the following discussion will help clarify certain aspects of liquid clamp-on technology.
We’ll begin with ultrasonic clamp-on flowmetering for liquids. We’ll then ask, and try to answer the question, Why not gases too? Our comments are restricted to transit time or contrapropagation flowmeters for industrial applications, i.e., measurement of flow in pipes. This article emphasizes clamp-on but it also includes some information on spoolpieces and wetted transducers.
As early as the 1930s, noninvasive ultrasonic measurements had their industrial start in the field of NDT (nondestructive testing). Applicability to solving flow and medical problems was recognized by the late 1940s and 1950s, respectively, or perhaps earlier. The angle beam transducers, coupling methods, alignment, and timing necessary in a clamp-on flowmeter can be recognized in counterpart items in NDT. (The transit time flowmeter’s need for rapid upstream-downstream switching was already known by the late 1940s, as reviewed in [4-6]).
The earliest paper we’ve been able to find on clamp-on ultrasonic flowmeters appeared in 1954 , but this was for a plastic pipe and the article, while claiming 2% linearity from 1–100 cm/s, did not include any flow calibration test data. In1957, a second “clamp-on” paper appeared  that suggested ways of modifying the wall of a metal pipe to encourage propagation along a desired oblique path. By 1964 , practical clamp-on ultrasonic flowmetering of water in large steel pipes was established in Japan, and by the 1970s the technology was available, both imported  and independently, in the U.S. [4-6, 11]. Today, liquid flow is measurable by clamp-on equipment for industrial conduits as small as 10 mm, and if the conduit is plastic, special techniques are available to interrogate the full cross-section axially  or by means of a vee bounce .
The technical difficulties faced by early flowmeter investigators trying to measure liquid flow in steel pipes by clamp-on included:
Aiming the beam along a desirable path
Getting a strong received signal
Avoiding interference from crosstalk around the pipe
Today, if we try to measure air flowing in a steel pipe, the same sort of difficulties arise. Table 1 compares the acoustical properties of some liquids, gases, and solids (pipe materials) that govern the refraction and energy transmission coefficients.
Acoustical properties of some liquids, gases, and solid materials
used for pipes, at 20°C and atmospheric pressure (except steam).*
Sound speed c, m/s
Z = c, Mrayls
Steam at 200ºC & 200 psig
Gasoline (87-93 octane)
* Most values are rounded off.
Energy transmission between two media, for example, depends on their impedance ratio. This means it is >3000 times more difficult to transmit from a steel pipe into ordinary air than into water. As air pressure increases, or if the pipe were plastic instead of steel, the problem gets easier to solve. However, the refracted angle in air will still be small compared to that in water, so spacing in small pipes remains a problem. Figure 1 compares several aspects of liquid vs. gas clamp-on.
Figure 1. Commercial contrapropagation clamp-on flowmeters have been available since the early 1990s for measuring the flow of liquids. In water, the vee path usually works, and 30is ~25° at room temperature (A, B, C). In air, 3.AIR is only about 6° and the transducers usually need to be placed on opposite sides of the pipe (D). For liquid clamp-on, the vee path shown in (A) tends to cancel crossflow as well as double the sensitivity to flow compared to a single traverse. For gases, odd numbers of traverses are preferred in order to reduce crosstalk. This means that if crossflow is significant, crossed paths are recommended. The velocities measured along the legs of the X should be averaged (E). Best solution: find a long straight run far from disturbances and joints. For gases (D and E), the flowmeter instrument introduced in 2001 is the GC868 (F).
In addition to the weak signal, there is a relatively strong signal (crosstalk) traveling around the steel pipe. Considering just these two factors, we see that for ordinary air in a steel pipe, the clamp-on signal-to-noise ratio (SNR) will be very small. That is the short answer to two questions: “Why not air?” and “Why not gases too?”
If we imagine slicing a steel pipe lengthwise to eliminate crosstalk between diagonally opposed transducers, there would still remain the problem of achieving adequate signal strength. With wetted transducers, signals can be increased by acoustic impedance matching, achieved by incorporating an impedance-matching material of quarter-wavelength thickness. With clamp-on, this isn’t possible. Three of the keys that underlie the clamp-on steam or gas examples below are:
Finding a way to eliminate crosstalk (other than slicing the pipe)
Finding a way to increase signal strength, other than impedance matching
Improving the SNR by signal processing
In other words, to solve the clamp-on gas flowmeter problem, the entire transit time ultrasonic flowmeter system had to be revisited, modified where necessary, and optimized to deal with the gases of interest [2,3].
Industrial Applications—Liquids, Gases
Shortly after our previous article went to press, bundle waveguide technology (BWT) transducers  were patented. Their main use continues to be in wetted sensors for gases. A secondary use is for liquids that are intermittently two-phase or highly attenuating. A tertiary use, currently under development, is for a temporary clamp-on measurement of sound speed in the liquid pipe, to check the feasibility of a questionable application and to more accurately preset the spacing of OKS “hockey stick” clamp-on transducers. These devices, also patented after our previous article appeared, have developed into standardized versions for temperature extremes. They are now routinely coupled with indium foil for cryogenic applications down to –200ºC and zinc + gold sandwiches to 360ºC. Occasionally, gold is used as well, e.g., to couple at 400ºC. Once installed, the coupling is stable and maintenance free for years. Nuclear power plant installations on 75 mm dia. pipe at 260ºC–282ºC have been running without any maintenance of the foil couplant for more than two years in the U.S. and more than four years in Belgium, for example. Weld-on and weld-in spoolpiece versions also exist. Welded transducers provide a different (nonremovable) solution for permanent coupling. With welded transducers, pipe diameters have been extended downward to ~25 mm (1 in.). A 75 mm application of clamped-on OKS transducers running at 282ºC at a nuclear plant in Connecticut since 1999 is reported elsewhere in detail .
One potential application, still at the R&D stage, occurs at the sea bottom, where the OKS transducer’s solid weldable construction offers potential advantages of small size and ruggedness to withstand seabottom pressure without requiring an enclosure. Perhaps of more general interest is the OKS transducer’s potential role in liquid custody transfer. Here, the thinness of its clamped-on waveguiding blade offers opportunities for multipath (quadrature) spoolpiece construction while permitting transducer removability at almost any time, without valves and without dewatering/emptying the pipe. The temperature of the liquid can be ambient, cryogenic, or very hot (see Figure 2).
Figure 2. OKS "hockey stick" clamp-on transducers first appeared in 1996. The long handle acts as a buffer, protecting the piezo material from extremes of heat or cold (400°C to -200°C) for pipe diameters 50 mm (A). A weld-in or weld-on transducer eliminates both couplant and clamp and extends the diameter down to 25 mm; a triple traverse triples the sensitivity to low flow velocities (B). A quadrature clamp-on spoolpiece currently under development is intended for high-accuracy (0.5% or better) liquid custody transfer applications (C). In this multipath true clamp-on can be seen the equiangularly spaced arrangement in four diametral planes (D). The swirlsensor shown here is clamped onto a pipe for R&D or other purposes (E).
Gas applications of the type of clamp-on flowmeter reported here began in 1998 in collaboration with plant maintenance personnel at two universities. They assisted us by providing several steam pipes in which flow was to be measured by clamp-on; the smallest was 100 mm (4 in.) and the largest, 300 mm (12 in.). One early steam application where clamp-on worked relatively easily was on a good-quality 6 in. dia. steel pipe in which the steam was at ~232ºC and 205 psig. Wet steam or pipes of rough, corroded, or scaled condition can be more problematical.
As we worked on lower pressure steam, we recognized that the method might be extendable down toward atmospheric pressure. This encouraged us to continue research on eliminating crosstalk, strengthening the signal, and dealing with other practical aspects of clamp-on gas flowmetering in the field. The system currently available (see Figure 3) is designed for a variety of clamp-on gas applications, including pressurized natural gas, hydrogen, or air, but not steam.
Figure 3. Clamp-on applications using the Model GC878 prototype or related predecessor instruments are shown for steam (A) in a photo courtesy of Ted Borer. Clamp-on gas flowmeter details are shown in (B); methane calibration data at 720 psig in (C); and air calibration data at a pressure of ~4 bar (60 psig) in (D). If the pipe is plastic, clamp-on works with ordinary air down to atmospheric pressure (E). The methane calibration data shown here were obtained at Southwest Research Institute in May 2001 on schedule 40 steel pipe of diameter 12 in. (300 mm i.d.)
In some cal lab situations, clamp-on transit time measurements in liquids have achieved 0.5% agreement with a gravimetric reference. As required for industrial process control applications, this agreement is obtained with a response time of a few seconds or less. However, if the flow is unsteady and/or not developed (meaning the profile is changing as it travels down the pipe), then one sample of flow along a tilted diameter generally is not sufficient to yield 0.5% accuracy, and response time may need to be as short as 1 s or less.
There are several solutions to the geometric part of the flow-sampling problem. One is to average two or four diameter-path interrogations, but this is not always sufficient. Another is to send the sound wave down the axis and examine 100% of the cross section.
For ultrapure semiconductor liquids and some other liquids in small-diameter (<25 mm) plastic tubes, a guided mode solution  has recently become available (see Figure 4).
Figure 4. A guided-mode clamp-on for ultrapure liquids in plastic tubing with a diameter of 25 mm  is shown in cross section and in its entirety (A). The Cerberus flow measurement system used for PTB certification testing includes a pair of transducers per path and two paths per electronics module (B). A quadrature method samples the flow in three planes of the spoolpiece (C).
For the unsteady, transient, or oscillatory flow part of the sampling problem, the standard solutions have been to launch upstream and downstream interrogations as nearly simultaneously as practical.
For liquids or gases in large-diameter pipes, e.g., >300 mm (but possibly smaller in the future—goal=100 mm), quadrature and other off-diameter sampling solutions have existed for some years. The Cerberus flow measurement system combines an industrial gas meter, a flow transmitter, a T11 transducer, and a precision-machined spoolpiece. Quadrature and other off-diameter interrogation systems generally are less sensitive to the influence of profile details . The illustrated quadrature flowmeter system was developed jointly by Panametrics and RMG of Butzbach, Germany, to measure critical natural gas flows at custody transfer stations. The Cerberus system shown in Figure 4 was used in PTB (Physikalische Technische Bundesanstalt) tests, where it achieved preliminary approval for custody transfer of natural gas in Germany. The ability of the industrial gas meter flow transmitter to be calibrated in air at atmospheric pressure greatly accelerated the approval process.
The first industrial cryogenic application of OKS transducers occurred in Korea and was reported in . Installed in January 1999, the first pair worked well for six months. Then they were transferred to a second pump line.
The customer (KOGAS) later purchased two DF868s with OKS transducers, based on satisfactory test results for a year. One of these was installed on 23 March 2001 as follows:
Pipe material: stainless steel
Pipe OD: 114.3 mm
Figure 5. Cryogenic (approximately -200°C) liquid flow clamp-on transducers were used in Japan on pipes of 50-200 mm dia. In this image, the insulation has been partly removed to show the clamps and the indium-coupled OKS transducers. (Photos courtesy of NASDA.)
Wall thickness: 6.02 mm
Couplant: Molycote anti-seize grease (initially), indium later
Liquid temperature: –160 C (normal: pump on)
Flow rate: normal 12–8 ton/hr.
The second cryogenic customer for OKS transducers was NASDA, the Japanese Space Agency (see Figure 5). The smallest pipe in their applications was ~50 mm dia.
Clamp-on ultrasonic transit time flowmeters, once restricted to liquids, have now been demonstrated on a variety of industrial gases. These applications range from air at atmospheric pressure (if the conduit is plastic) to pressurized gases such as steam, methane, or nasty gases typically pressurized to >4 bar in steel pipe, and preferably >10 bar, depending on pipe conditions, gas attenuation, Mach number, and gas molecular weight. If the application is appropriate, clamp-on ultrasonic transit time flowmeters can be counted on for measuring the flow rate of fluids (liquids and gases) in steel and other industrial pipes.
The work reported here includes important contributions from the authors’ colleagues Dr. Xiaolei Shirley Ao, Jim Bradshaw, Chris Brolin, Jim Hill, Jim Hurd, Saul Jacobson, Oleg Khrakovsky, Peter Kucmas, Scott Li, Bernard Lupien, Chris Smart, and others. Panametrics’s Bill J. Park and Yoshiteru Urata provided the cryogenic application information in Korea and Japan, respectively.
The authors acknowledge Panametrics’s permission to reproduce passages, one table, and illustrations from its copyrighted report UR-252. Thanks too to KOGAS and NASDA.
Hans J. Kastner, Andreas Weber, and others at RMG contributed to the equipment shown in Figure 4 C. Cindy Lynn Rountree’s 1995 gasoline data are the basis of the 87-93 octane approximations in Table 1.
Edward T. (Ted) Borer, P.E., Princeton University, provided much assistance on steam beta sites. Arjan Stehouwer provided useful feedback from test sites in Holland. The manuscript was prepared by Lin L. Leeming.
BWT (Bundle Waveguide Technology) and OKS (hockey stick transducer) are trademarks of Panametrics, Inc.
2. X. Ao. April 1, 1999. Clamp-On Steam/Gas Flow Meter, U.S. Patent Pending.
3. X. Ao et al. “Ultrasonic Clamp-On Flow Measurement of Natural Gas, Steam and Compressed Air,” abstract submitted for presentation at, and paper, if accepted, to be published in Proc 5th ISFFM (5th International Symposium, Fluid Flow Measurement), to be held in Arlington VA, 7-10 April 2002.
4. A.E. Brown. 1991. “Ultrasonic Flowmeters,” Flow Measurement—Practical Guides for Measurement Control, D.W. Spitzer, ed. ISA:415-442; with L. Lynnworth. 2001. Ibid., 2nd Ed.:515-573.
5. L.C. Lynnworth. 1979. “Ultrasonic Flowmeters,” Physical Acoustics—Principles and Methods, W.P Mason and R.N. Thurston, eds., 14, Academic Press:407-525.
6. L.C. Lynnworth. 1989. Ultrasonic Measurements for Process Control, Theory, Techniques, Applications, Academic Press.
7. H.P. Kalmus. 1954. Rev Sci Instrum 25:201-206.
8. V.A. Del Grosso and E.M. Spurlock. 1957. “The Feasibility of Using Wholly External Ultrasonics to Measure Fluid Flow Within Thickwalled Metal Pipes,” NRL 4967, AD No. 149409.
9. T. Baba. 1964. Jpn J Waterworks Assoc No. 4:21-31.
10. W.K. Genthe and M. Yamamoto. 1974. “A New Ultrasonic Flowmeter for Flows in Large Conduits and Open Channels,” Flow—Its Measurement and Control in Science and Industry, 1 (Part 2), R.B. Dowdell, ed., ISA:947-955.
11. L.C. Lynnworth and V. Mágori. 1999. “Industrial Process Control Sensors and Systems,” Ultrasonic Instruments and Devices: Reference for Modern Instrumentation, Techniques, and Technology, 23, E.P. Papadakis, ed., Physical Acoustics Series, Academic Press:275-470.
12. J.A. Hill and J.P. Pell. 23 May 2000. Flow Measurement System with Guided Signal Launched in Lowest Mode, U.S. Patent No. 6,065,350.
14. Y. Liu, L.C. Lynnworth, and M.A. Zimmerman. Feb. 1998. “Buffer Waveguides for Flow Measurement in Hot Fluids,” Ultrasonics, 36 (1-5):305-315.
15. R.N. Burnham et al. 23 May 2001. “Measurement of the Flow of Superheated Water, in Blowdown Pipes at MP2 Using an Ultrasonic Clamp-On Method,” on Panametrics Web site, PCI R&D pages.
What's a Fiberacoustic Bundle Good For?
Small-diameter wires, or a bundle of several hundred of them, can convey ultrasonic pulse distances to at least 1 m with little dispersion. This means the pulse shape is retained and doesn’t get smeared. Timing the arrival of such well-defined pulses is much easier than it would be if the pulse were to become smeared and weakened along its journey in the waveguide. But why use a waveguide in the first place? Answer: the waveguide  allows the piezoelectric transducer element to remain rather distant fýom the fluid, whose temperature may be far from ambient, say 500°C or –200°C. Since about 1995, Panametrics has been bundling stainless steel or titanium wires within pipes of the same material. The ends of these BWT (Bundled Waveguide Technology) tranýducers are welded shut. Applications have ranged from gases (e.g., steam at 340°C) to hot or cold liquids, and include two-phase fluids and some that are only intermittently two-phase. The accompanying figure conveys the general idea of the BWT transducer’s fiberacoustic bundle.
Hockey Sticks for Clamp-On Measurement of Hot or Cold Liquids
Another form of buffer waveguide evolved for clamp-on applications, first at high temperature (now 400°C) and later down to cryo levels (now –200°C). The clamp-on buffer takes the form of a small hockey stick . Instead of thin wires, a thin blade acts as an efficient waveguide for shear waves, buffering the shear-wave piezo from the temperature extremes of a hot or cold process and facilitating clamp-on measurements when the fluid is a liquid. Termed the OKS transducer, its length is <100 mm in short versions and can be 300 mm when the handle needs to be long enough to place the piezoelectric element away from the heat of a very hot insulated pipe. The OKS transducer shown here can be interpreted with the handle length adjusted to suit the application (see also Figure 2).
It should also be mentioned that whether the fluid is a liquid or a gas, its sound speed, c, often conveys useful information. In a gas of known composition, c can reveal the temperature ; in flare gas of known temperature, it can reveal the average molecular weight. In a pure liquid, c can be interpreted in terms of density. By combining c with the average flow velocity, V, and multiplying by the duct area, A, we arrive at a self-contained mass flowmeter. If c and V have been obtained by clamp-on, the result is a clamp-on mass flowmeter.
For further reading on this and related topics, see these Sensors articles.