ULTRASONIC FLOW METERS – PART 1

Terrence A Grimley

Southwest Research Institute

6220 Culebra Road
San Antonio, TX 78238 USA

Introduction

This paper is the first of a two-part series that provides an introduction to flow measurement using transit-time ultrasonic flow
meters. This paper covers the essential knowledge that users should have regarding the basic operation of ultrasonic flow
meters. The second paper will provide more detail on diagnostics and operational effects on ultrasonic flow meters.

Background and Early Development

Early use of ultrasonic signals for flow measurement started in the 1950s, when Doppler ultrasonic flow meters were first
suggested for use in blood-flow monitoring by the National Bureau of Standards (predecessor to the National Institute of
Standards and Technology (NIST)). In the mid-1960s, a multipath transit-time ultrasonic flow meter was developed by
Westinghouse for water flow measurement in the nuclear industry. Westinghouse was issued a patent in 1971 for the first
chordal multipath meter using Gaussian integration. A similar path configuration to that used by Westinghouse was researched
for natural gas applications by British Gas in the 1980s and patented and evolved into the Daniel multipath ultrasonic flow
meter. The availability of digital electronics supported improvements in the stability and accuracy of the ultrasonic flow meter
that allowed the meters to be considered for custody transfer applications in the natural gas industry. The 1998 publication of
American Gas Association (AGA) Report No. 9 (AGA-9) titled “Measurement of Gas by Multipath Ultrasonic Meters”
provided the reference document that the gas industry needed to begin widespread installation of ultrasonic meters.

Fundamental Measurements and Equations

Transit-time ultrasonic meters work by measuring the time required for a pulse of ultrasonic energy to travel from one ultrasonic
transducer to another through the fluid to be measured. The transit time is related to the distance between the transducers, the
speed of sound of the fluid, and the velocity of the fluid that the pulse encounters. The basic geometric arrangement of an
ultrasonic meter is provided in Figure 1. Although this example is shown with a direct path between the ultrasonic transducers,
the same principles and geometry apply when the ultrasonic energy is reflected one or more times between transducers.

UPSTREAM
TRANSDUCER
L

FLOW
 DOWNSTREAM
TRANSDUCER

Figure 1. Basic Geometry of a Transit-Time Ultrasonic Flow Meter

When the fluid is flowing, the transit times are different between pulses propagated in the upstream (tup) direction and ones
propagated in the downstream direction (tdown), and this can be related to the velocity through the equations that follow:

t = Equation 1
 t = Equation 2

where Vpath is the average axial component of gas velocity along the ultrasonic pulse path and C is the average speed of sound.
The transit time equations can be combined to produce the following approximation:

V = − = Equation 3
∙

Where, X = L cos θ

Since the measurements are made in both the upstream and downstream direction, the speed of sound does not need to be
known to solve for the fluid velocity, which allows the speed of sound of the fluid to be computed as a by-product of the
measurement.

C= Equation 4
∙

The velocity (Vpath) represented in the previous equations is the average velocity (in the flowing direction) along the acoustic
path taken by the ultrasonic energy; however, the velocity within the meter is not uniform across the diameter. An
approximation of a fully developed velocity profile is shown in Figure 2. Note also that the shape of the profile changes slightly
as the velocity (Reynolds number) is increased.

𝜌𝑉𝐷
𝑅𝑒𝑦𝑛𝑜𝑙𝑑𝑠 𝑁𝑢𝑚𝑏𝑒𝑟 = 𝑅𝑒 =
𝜇

Where ρ=density
V= average velocity
D=pipe diameter
µ=viscosity

Figure 2. Normalized Velocity Profile at Different Reynolds Numbers

Because the velocity is non-uniform, the measured path velocity must be corrected to the average pipe velocity (V average) by
applying a correction factor (fpath). The value of this factor is dependent on both the path geometry and the velocity profile
present in the meter, which means that it may be a function of the Reynolds number (or velocity)

V =V ∙f Equation 5

To compute the volumetric flow rate (Qf), the meter average velocity is multiplied by the flow area.

Q = D ∙V Equation 6
To compute the mass flow rate, or the standard flow rate, additional information about the gas composition and the pressure
and temperature is needed. For the flow rate at base conditions (Qb), ultrasonic flow meters use the same equation as that used
for any linear volumetric flow meter (e.g., turbine meters):

Q =Q Equation 7

where, Pf, Tf, and Zf are the flowing absolute pressure, absolute temperature and compressibility, respectively, and variables
with the subscript “b” are the base conditions of the same measurements.

Path Configurations

The accuracy of ultrasonic flow meters is improved by sampling the velocity profile at multiple locations through the use of
multiple paths. The individual path measurements are combined to provide an overall average velocity through the meter. The
method for combining the paths depends on the algorithm developed by the manufacturer, but in the simplest sense, the velocity
for each path (Vi) has a weighting factor (wi) applied, and the weighted path values are summed to compute the average as
shown in the equation below.

V =∑ V ∙w Equation 8

The individual path measurements may or may not include corrections based on the Reynolds number, and meters may also
use methods that vary the weighting functions, depending on the characteristics of the measured path velocities, in an effort to
correct for velocity profile distortion that can occur with common pipeline geometries.

There are a variety of path configurations that are used in commercial ultrasonic flow meters. Differing philosophies drive the
path configuration and the methods used for combining the individual path measurements into a bulk flow velocity. Figure 3
shows several path configurations that are used or combined in ultrasonic meters.

Figure 3. Example Path Configurations

The dashed lines shown in Figure 3 may represent paths that are created by reflection of the ultrasonic signal or may represent
paths with different transducer orientations. The path geometry determines how the path will respond to the velocity profile
present in the pipe and specifically the non-axial components of flow (e.g., swirl and cross flow). In some cases, the path
configuration can be used to detect non-symmetric velocity profiles.

Some meters also include additional paths that are specifically intended for diagnostic purposes, such as the presence of liquid
or profile distortion.

Meter Components and Construction

The major components of an ultrasonic meter include the meter body, the ultrasonic transducers, and the electronics.

The meter body is the pressure-containing element through which the gas flows. The body must allow the transducers to be
precisely located relative to one another and to maintain those positions throughout a range of operating conditions. The meter
body is generally attached to the surrounding piping via standard flanges. The meter body includes a pressure tap so that the
pressure within the flow measurement area can also be measured and used to compute the mass (or standard volume) flow rate.
Some meter body designs include the capability to remove and change transducers without removing pressure from the meter
through the use of custom tools.

Ultrasonic transducers have a variety of manufacturer-specific mechanical and electrical configuration details, but all serve the
same purpose: to inject ultrasonic energy pulses into the flow towards another transducer and to receive ultrasonic energy
pulses generated by another transducer. The design of the transducer (shown generically in Figure 4) must be executed in such
a way that the ultrasonic energy overcomes the mismatch in acoustic impedance that is present between the gas and the
piezo-electric material that generates (or receives) the ultrasonic signal. This is typically accomplished through the use of
intermediate materials that are often referred to as a “matching layer.” To ensure that the bulk of the ultrasonic energy enters
the gas stream and not the surrounding metallic pipe, acoustic isolation is needed between the transducer and the meter body.

Electrical Connections

Acoustic Insulating Material

Transducer Metallic Case

Piezoelectric Element
Acoustic Matching Layer
Figure 4. Ultrasonic Transducer for Flow Measurement

The meter electronics control the operation of the ultrasonic transducers by causing them to alternate between sending and
receiving ultrasonic pulses. The electronics control the sequencing of the “firing” of transducers and the frequency at which
the transducers fire. The electronics implement the details of the transit time measurements and the calculations needed to
compute individual path measurements and combine those individual measurements into a bulk flow rate. The meter
electronics also compute and provide diagnostic information that can be used to determine if the meter is functioning properly.

Meter Sizing

The capacity of ultrasonic meters is typically characterized by a velocity limit through the meter; therefore, the meter capacity
can also be computed as an actual volumetric flow limit. The velocity limit is commonly in the range of 100 ft/s, but some
meters may function well above this value, while other meters (typically larger meters) may not be capable of maintaining full
performance on all of the paths at such a velocity. The capacity limits provided by flow meter manufacturers may exceed the
operating company velocity limits that also take into consideration noise, erosion, and structural limits on devices like
thermowells and gas sample probes, so it is important to consider the entire installation when sizing an ultrasonic flow meter.

Although the meter will respond to flow rate values down to zero flow, it should be noted that the practical minimum velocity
(or actual volumetric flow rate) is determined by the allowable error limit of the meter (as set by the manufacturer) and by the
economics associated with the error. Low-flow-rate errors tend to be dominated by unstable velocity profiles and inaccurate
temperature measurements that can be influenced by ambient conditions. The actual low-flow limit for a meter will likely be
site-specific because of the external influences. Most gas companies establish their own best practices for determining the
minimum flow rate based on prior field experience with their installations.

Error! Reference source not found. provides an example of the approximate meter capacity limits for common line sizes.

Table 1. Approximate Flow Capacity Ranges (MMSCF/D) Based on a 5 ft/s to 100 ft/s Velocity Range
Pressure

Size (in) 200 psia 500 psia 1,000 psia

Minimum Maximum Minimum Maximum Minimum Maximum
(MMSCF/D) (MMSCF/D) (MMSCF/D) (MMSCF/D) (MMSCF/D) (MMSCF/D)
4 0.53 10.7 1.40 27.9 3.01 60.1
6 1.21 24.2 3.17 63.4 6.83 136.6
8 2.10 42.0 5.49 109.8 11.83 236.5
10 3.31 66.2 8.65 173.0 18.64 372.8
12 4.70 93.9 12.28 245.6 26.46 529.2

AGA Report No. 9

AGA-9 is the industry document that is commonly referenced in custody transfer agreements and is also used for other
non-custody flow measurement where there is an expectation of high-accuracy flow measurement. Unlike some industry
documents, AGA-9 does not require a specific mechanical configuration. Instead, AGA-9 is a performance-based document
that provides information defining how the flow meter is expected to operate and leaves it to the manufacturer (or end user) to
define the specific piping configuration(s) in which the meter will meet the accuracy requirements specified in the document.
AGA-9 requires flow calibration of ultrasonic flow meters to verify that the meters meet the specification and to provide a
means to adjust the meter to minimize any residual meter error.

Expected Performance

The performance requirements are contained in multiple constraints that include: overall error, speed of sound error,
repeatability, and other performance indicators. Error! Reference source not found. provides the parameter limits that
are independent of meter size, and Table 3. AGA-9 Size-Specific Error Limits
provides a list of the acceptable range of the error limits that are size-specific. The performance limits can be
visualized with the plot shown in Figure 5, with the actual values for the limits coming from Table 3. AGA-9 Size-
Specific Error Limits
.

Table 2. AGA-9 Size-Independent Requirements

Parameter Value
Repeatability: qmin ≤ qi ≤ qt ± 0.2 %
Repeatability: qt ≤ qi ≤ qmax ± 0.4 %
Velocity Sampling interval ≤ 1 second
Resolution 0.003 ft/s
SOS Deviation ± 0.2 %
Maximum SOS Spread 1.5 ft/s

Figure 5. Performance Specification (Error) from AGA-9

 Table 3. AGA-9 Size-Specific Error Limits
Meter Size Less than 4-inch 4-inch to 10-inch 12-inch and Larger
Maximum Error:
qmin ≤ qi ≤ qt ±3.0% ±1.4% ±1.4%
qt ≤ qi ≤ qmax ±2.0% ±1.0% ±0.7%
Maximum Peak-to-Peak Error:
qmin ≤ qi ≤ qt ±1.4% ±1.4% ±1.4%
qt ≤ qi ≤ qmax ±1.0% ±0.7% ±0.7%

Installation

Because of the variation in the configurations of commercially available ultrasonic flow meters, AGA-9 provides limited
guidance on the installation requirements and provides several options for determining the meter tube geometry. The latest
version of AGA-9 (3rd Edition, July 2017) includes descriptions of options that were allowed by the previous version of the
document, but were not necessarily obvious to the reader.

Options 1 and 2 include a flow conditioner located upstream of the meter between the spools UL1 and UL2 shown in Figure 6.
For Option 1, the upstream spool lengths are each a minimum of 10ND (nominal pipe diameters), and the flow conditioner is
specified by the meter manufacturer with the configuration results supported by independent test data. For Option 2, the spool
lengths and the flow conditioner are all specified by the meter manufacturer and again, should be supported by independent
test data.

UL1 UL2 DL

FLOW

Figure 6. Option 1 and Option 2 – Installation with a Flow Conditioner

Figure 7 shows that Option 3 allows for a configuration without a flow conditioner, with the upstream pipe length recommended
by the meter manufacturer and determined through independent test data.

UL1 DL

FLOW

Figure 7. Option 3 – Installation without a Flow Conditioner

Figure 8 shows a bidirectional installation that can be based on Options 1, 2, or 3 with the same piping installation upstream
and downstream of the meter.

UL1 UL2 UL2 UL1

FLOW FLOW

Figure 8. Bidirectional Installation

As an alternative to the previously described installation options, the meter package may be calibrated in a flow facility (or
in-situ) with test piping identical to that installed in the field. Calibrations of this type should include at least two of the
upstream disturbance elements (e.g., end treatments for the meter run and header outlet piping).

The list that follows is a partial summary of other installation requirements:

 The static pressure measurement should be made on the meter body.
  The temperature measurement should be made within 2ND to 5ND of the downstream flange of the meter.
 The meter bore diameter and upstream pipe diameter should be within 1% of each other.
 Meter tube roughness of less than 250 μ-inch for meter tubes less than 16-inch diameter.

Calibration

Flow meter calibration normally starts by first establishing stability in the measuring system by flowing gas at a high rate for
the meter under test. In a closed-loop calibration system, the bulk system pressure is established prior to circulating and then
adjusted as the temperature reaches its set-point condition. For any system, the high flow rate allows the meter and its associated
piping to soak at the temperature of the flowing gas and stabilize prior to starting the meter calibration.

It is common practice to test first at the higher flow rates and gradually decrease the flow to cover the calibration range, since
this typically provides the best stability. The flow rates to be tested can be taken from AGA-9 or established by other suitable
methods that are acceptable to both parties involved, with a measurement that the meter will provide. AGA-9 recommends
calibration flow rates equal to 100%, 75%, 50%, 25%, 10%, 5%, and 2.5% of the maximum flow rate.

At each flow rate, there are a number of discrete test points recorded where each test point consists of an average over a sample
period of 90 seconds to 300 seconds. Different flow labs use different sample periods and different numbers of repeat points,
but all labs should provide a sufficient sampling of the meter performance to establish the meter error relative to the flow
reference of the lab. Data collected include flow rates, pressures, and temperatures for the test meter and reference meter(s),
and gas composition data for the system. Average values for all of the measurements are included in the test report.

During each calibration test point, data are collected not only to determine the measurement accuracy of the meter, but also to
provide a reference for future diagnostic evaluations of the meter. The software from the meter manufacturer is typically used
to log the performance of the meter so that path-by-path values for speed of sound, velocity, ultrasonic transducer gain levels,
and any other diagnostic parameters can be captured. In some cases, this may include capturing the waveforms received by
the transducers.

Another comparison made during the meter calibration is between the speed of sound reported by the meter and that computed
using the AGA Report No. 8 (Thermodynamic Properties of Natural Gas and Related Gases) or AGA Report No. 10 (Speed of
Sound in Natural Gas and Other Related Hydrocarbon Gases), and the pressure, temperature, and gas composition measured at
the meter.

There are various methods that can be used for adjusting the meter after the meter calibration data have been obtained. There
are three common adjustment methods indicated in AGA-9:
1. Adjust based on the Flow-Weighted Mean Error
2. Adjust using a polynomial algorithm
3. Adjust using a multi-point linearization

Other methods are also acceptable, as agreed by the parties involved in the measurement. The selection of the best method
depends on the characteristics of the meter, as well as the application.

Calibration Example

Figure 9 shows the “as-found” results of an example meter calibration including the appropriate AGA-9 limits. The meter is
within the limits that indicate that this meter diameter is in the 4-inch to 10-inch range. The tabular data that the curve represents
is provide in Table 4, along with a calculation of the flow-weighted error.
 Figure 9. As-Found Meter Curve

Table 4. Tabular Data Showing Calculation of the Flow-Weighted Error

FWME
Flow Velocity Error Full-Scale Flow-Weighted
Point Corrected
(acfh) (ft/s) (%) Fraction Error
Error (%)

1 73,023 101.104 0.554 1.011 0.560 0.129

2 53,727 74.388 0.458 0.744 0.341 0.033
3 38,331 53.071 0.360 0.531 0.191 -0.065
4 26,805 37.113 0.357 0.371 0.132 -0.067
5 15,279 21.155 0.166 0.212 0.035 -0.257
6 7,702 10.663 0.090 0.107 0.010 -0.333
7 3,894 5.392 0.314 0.054 0.017 -0.110
Sum = 3.029 1.286
FWME = 0.425
Meter Factor = 0.99577

The flow-weighted error (FWE) is used in determining the Flow-Weighted Mean Error (FWME). These values are computed
using the following equations:

𝐹𝑊𝐸 = ∙ 𝐸𝑟𝑟𝑜𝑟 Equation 9

∑ .
𝐹𝑊𝑀𝐸 = = = 0.425 Equation 10
∑ .

When the FWME method is used, a single correction factor is entered into the meter configuration. The meter factor is the
amount by which the meter output should be multiplied to match the reference flow rate. To convert from the FWME to a
meter factor, the following equation is used:

𝑀𝑒𝑡𝑒𝑟 𝐹𝑎𝑐𝑡𝑜𝑟 = = = 0.99577 Equation 11

.

Figure 10 shows the results that should be expected when a FWME correction is made to the example calibration curve. The
use of a single meter factor uniformly shifts the error curve vertically on the graph. Because of the flow-dependent error for
this example, some error remains at both the high and low flow rates, but since the correction was flow weighted, the residual
error at the high flow rates is significantly less than that for the low flow rates.
 Figure 10. Meter Adjustment Using Flow-Weighted Mean Error

Because of the limitations in a single factor adjustment of the meter, a point-by-point meter linearization is the most common
method of adjusting an ultrasonic flow meter. The average meter factor, or average error, is entered into the electronics along
with the corresponding average test flow rate. Those points are used by the meter electronics to adjust the meter by performing
linear interpolation to establish the correction at any non-tested flow rate. The result of a point-by-point linearization is that
the average error at any tested flow point is expected to be zero as shown in Figure 11.

Figure 11 also includes confirmation points that were collected after the point-by-point correction values were entered into the
meter. The purpose of performing confirmation points is to verify that the meter was properly adjusted to reflect the calibration
values. Often, confirmation points are tested at rates slightly different than those used for the calibration, but in this case, the
points were collected at the same rate as two of the test points.

Figure 11. Meter Adjustment Using Multipoint Linearization

Additional AGA-9 Topics

AGA-9 includes a number of other important topics including commissioning, diagnostics, electronics change performance,
and meter uncertainty. More detail on some of these topics will be provided in Part 2 of this paper.
Conclusions

This paper has provided some basic information to familiarize the reader with the operation of transit-time ultrasonic flow
meters and the application of AGA-9. There is a significant body of literature available that documents the numerous tests and
developments that have taken place since ultrasonic flow meters were introduced to the natural gas industry in the 1990s. Part 2
of this paper provides additional insight into the application of ultrasonic flow meters and some of the operational and
installation effects that should be considered when building a meter station that utilizes an ultrasonic flow meter.

References

1. Fluid Flow Measurement System, U.S. Patent 3564912, February 23, 1971, Westinghouse Electric Corporation,
James T. Malone and Donald K. Whirlow.
2. Ultrasonic Flowmeter, U.S. Patent 4646575, March 3, 1987, British Gas Corporation, John G. O’Hair and Michael E. Nolan.
3. AGA Transmission Measurement Committee Report No. 9, Measurement of Gas by Multipath Ultrasonic Meters, First
Edition, American Gas Association, June 1998, Washington, D.C.
4. AGA Transmission Measurement Committee Report No. 9, Measurement of Gas by Multipath Ultrasonic Meters, Third
Edition, American Gas Association, July 2017, Washington, D.C.
5. AGA Transmission Measurement Committee Report No. 8, Thermodynamic Properties of Natural Gas and Related Gases,
American Gas Association, April 2017, Washington, D.C.
6. AGA Transmission Measurement Committee Report No. 10, Speed of Sound in Natural Gas and Other Related
Hydrocarbon Gases, American Gas Association, January 2003, Washington, D.C.