FUNDAMENTALS OF MULTIPATH ULTRASONIC FLOW METERS FOR GAS

MEASUREMENT
Eric Thompson
Regional Sales Manager
 
 
15415 International Plaza Drive, Suite 100
Houston, Texas 77032
Summary
This paper outlines the operating principles and
application of ultrasonic gas flow metering for
custody transfer. Basic principles and underlying
equations are discussed, as are considerations for
applying ultrasonic flow meter technology to station
design, installation, and operation. These
applications are illustrated based on operating
experience with the SICK ultrasonic flow meter,
however many of these issues can be generalized to
other meter manufacturers.
Introduction
The use of ultrasonic meters for custody (fiscal)
applications has grown substantially over the past
several years. This is due in part to the release of
AGA Report No. 9, Measurement of Gas by
Multipath Ultrasonic Meters [Ref 1], Measurement
 -G-E-06
Provisional
Ultrasonic
Specification [Ref 2], and the confidence users have
gained in the performance and reliability of
ultrasonic meters as primary measurement devices.
Just like any metering technology, there are design
and operational considerations that need to be
addressed
in
order
to
achieve
optimum
performance. The best technology will not provide
the expected results if it is not installed correctly, or
maintained properly. This paper addresses several
issues that the engineer should consider when
designing ultrasonic meter installations.
Why Use Ultrasonic Meters?
Before discussing installation issues associated with
ultrasonic meters (USMs), it might be good to review
what the benefits of using USMs are. Since the mid1990s the installed base of USMs has grown
steadily each year. It is estimated that more than
$65 million was spent worldwide on purchasing
USMs in 2006. There are many reasons why
ultrasonic metering is enjoying such healthy sales.
Some of the many benefits of this technology
include the following:
x
x
x
x

Accuracy: Can be calibrated to <0.1%.

Large Turndown: Typically >50:1.
Naturally Bi-directional: Measures volumes in
both directions with comparable performance.
Tolerant of Wet Gas: Important for production

x
x
x
x

applications.
Non-Intrusive: Minimal pressure drop.
Low Maintenance: No moving parts means
reduced maintenance.
Fault Tolerance: Meters remain relatively
accurate even if sensor(s) should fail.
Integral Diagnostics: Data for determining a


It is clear that there are many important benefits to

using USMs. The most significant, however, is the
       ! 
question users want to know is whether the meter is
still accurate after having been in service for some
period of time in the field.
Other primary measurement devices such as orifice
and turbine meters offer little insight into whether
they are still operating accurately after some period
of time. Issues such as contamination from pipeline
oil and mill scale can impact the accuracy of any
meter. Visual inspection is often required to validate
proper operation for traditional primary measurement
devices. Ultrasonic meters, on the other hand, offer
electronic diagnostics that can help validate proper
operation, and thus reduce the internal inspection
requirements often required of other devices. These
internal diagnostics can also be used to help identify
whether the other components at the measurement
station, such as temperature measurement and gas
composition, are also operating correctly. For these
reasons many designers are now specifying the use
of ultrasonic meters more today than ever before.
Ultrasonic Meter Basics
Ultrasonic gas flow meters operate on the transittime measurement principle. The basic operation is
relatively simple. Even though there are numerous
designs on the market today, the principle of
operation remains the same (Figure 1).

Figure 1
Ultrasonic meters are velocity meters by nature.
This means, they measure the velocity of the gas

within the meter body. Knowing the velocity and the

cross-sectional area, uncorrected volume can be
computed. In ultrasonic flow measurement, sound
waves or acoustic pulses are transmitted and
received by a pair of piezoelectric transducers
(Figure 2).

flow the transit time is equal in both directions and

the measured time of flight difference them (T ab /
Tba) is zero. Thus with flow the transit time of the
signal traveling with the flow will be less than that of
the signal traveling against the flow due to the
velocity of gas within the meter body (Figure 4).

Operating Principle
Transducer B

tBA
tAB

Transducer A

Travel Time Difference

t AB

L
c  v cos D

t BA

L
c  v cos D

Path velocity

v Pfad

L 1
1


2 t AB tBA

Figure 4

Figure 2
This figure shows the simple geometry of two
transducers, A and B, at an angle M with respect to
the axis of the straight cylindrical pipe.

Sound Velocity

1
L
1


2 cos D t AB t BA

Reviewing the basic equation needed to compute

volume, the transit times in the upstream and
downstream directions may be calculated as shown
in Figure 4
L
.
t AB

c  v cos D

"#&' *'+#

L
c  v cos D

t BA

Where L is the length of the path, M is its angle with

respect to the aspect of the pipe and c is the speed
of sound (SOS) in the gas (e.g. about 1,000 to 1,500
ft/sec. at typical natural gas transmission pressures).
If the speed of sound is constant during both transit
times the two equations can be combined, See
Figure 4 Path Velocity

v Pfad
Figure 3
In Figure 3, L denotes the path length, D denotes
the diameter, and V is the velocity vector. The
transit time (Tab) of an ultrasonic signal traveling with
the flow is measured from Transducer A to
Transducer B. When this measurement is
completed, the transit time (Tba) of an ultrasonic
signal traveling against the flow is measured from
Transducer B to Transducer A. When there is no

L
2 cos D

1
1


t AB t BA

where v denotes the flow velocity (positive in the

downstream direction). The history of this method
dates back to Rtten (1928), who filed the first
patent on the application of ultrasound in flow
measurement. From the above equations, the
speed of sound can be calculated;

L 1
1


2 t AB t BA

Since the speed of sound is related to density of the

medium in the transport system, it can be used to
calculate mass flow. Further, it is noted that the
cancellation of c from the average velocity equation
(gas properties, such as density, affect both T ab and
Tba equally) means that absolute velocity
measurement is not dependent on gas density.
This means, pressure, temperature, and gas
composition have no effect on the velocity
calculation from pulse transit time.
The average transit time, with no gas flowing, is a
function of meter size and the speed of sound
through the gas (pressure, temperature, and gas
composition). Consider a 12-inch meter for this
example. Typical transit times, in each direction, are
on the order of one millisecond (and equal) when
there is no flow. The difference in transit time during
periods of flow, however, is significantly less, and is
on the order of several nanoseconds (at low flow
rates). Thus, accurate measurement of the transit
times is critical if an ultrasonic meter is to meet
performance requirements established in AGA
Report Number 9.
These equations look relatively simple and they are.
The primary difference between computing gas
velocity and speed of sound is the difference in
transit times is used for computing velocity, where
the sum of the transit times is used for computing
speed of sound.
Determining the correct flow rate within the meter is
a bit more comprehensive than it appears. The Path
Velocity equation shown refers to the velocity of
each individual path. The velocity needed for
computing volume flow rate, also known as bulk
mean velocity, is the average gas velocity across the
** *+
are not always uniform and often there is some swirl
and asymmetrical flow profile within the meter. This
makes computing the average velocity a bit more
challenging.
Meter manufacturers have differing methodologies
for computing this average velocity. Some derive
the answer by using proprietary algorithms. Others
&'?@
computations. Regardless of how the meter
determines the bulk average velocity, the following
equation is used to compute the uncorrected flow
rate;
Q=V*A

This output (Q) is actually a flow rate based on

volume-per-hour, and is used to provide input to the
flow computer. A is the cross-sectional area of the
meter.
Summarizing keep points to keep in mind about
ultrasonic meter operation are:
x The measurement of transit time, both
upstream and downstream, is the primary
function of the electronics.
x All path velocities are averaged to provide a
?'L@ ' compute
'*'QQ).
x Because the electronics can be determined
which transit time is longer (T ab or Tba), the
meter can determine flow direction.
x Speed of sound is computed from the same
measurements as gas velocity.
Transducers Pulse Generation
Specially designed transducers are used for the
generation of ultrasonic pulses that both transmit
and receive these pulses. The main component
within a transducer performing these functions is a
Piezoceramic element. In the transmitting mode,
these Piezoceramic elements are excited with a
characteristic voltage that results in the emission of
?#-  X@'*'Y'
receiver, the incoming sound pulse generates a
small voltage, that is processed after amplification.
The frequency and directivity pattern of a transducer
depends, for the most part, on the dimensions and
characteristics for the piezoceramic element.
Transducer Pulse Detection
Before pulse detections and recognition takes place,
the received sound wave is pre-processed using an
Automatic Gain Control and filtering circuitry to
ensure pulse discrimination. After the preprocessing stage (detection) takes place, the signal
X *#?+*@+
reference pulse. This method provides the unique
ability to check the quality of every single pulse
against preset standards before processing for
velocity measurement purposes. The pulses are
either accepted when the signal completely meets
the preset quality standards or rejected when a
deviation from these quality standards is detected.
Only when both pulses are accepted are their travel
times used to calculate flow velocity and speed of
sound. This method results in the highest precision.
During signal detection and processing, built-in
diagnostics supply real-time information to the user
about the performance of the system, and may be

used to set alarm limits on meter performance.

These parameters will be discussed in more detail
later.

velocity. Figure 5 shows gains from a 16-inch meter

at the time of calibration. These were taken when
the meter was operating at approximately 20 fps.

Timing Characteristics
The accuracy required in the travel time
measurement can be found from the equations. For
example, when a velocity of 3 ft/s is measured with
0.5% accuracy along a 3 ft. path length in a gas with
sound velocity of 1,300 ft/s, both travel times are of
order of 2.5 milli-seconds, and their difference is
about 6 micro-seconds, which must be measured
with an error no greater than 30 NANO-seconds!
This minute travel time difference requires highspeed, high-accuracy digital electronics. The travel
times of only a few milli-seconds enable individual
ultrasonic flow velocity measurements at high
repetition rates. Typical rates are 20 to 50 Hz,
depending on pipe diameter. The need for high
repetition rates is evident in cases such as surge
control applications, where the flow may drop from
its set point to its minimum in less than 0.05
seconds.
BASIC DIAGNOSTIC INDICATORS
One of the principal attributes of modern ultrasonic
meters is the ability to monitor their own health, and
to diagnose any problems that may occur. Multipath
meters are unique in this regard, as they can
compare certain measurements between different
paths, as well as checking each path individually.
' '?
 L@  
external diagnostics. Internal diagnostics are those
indicators derived only from internal measurements
of the meter. External diagnostics are those
methods in which measurements from the meter are
combined with parameters derived from independent
sources to detect and identify fault conditions. An
example of this would be to compute the gas SOS,
based upon composition, and compare to the
'\
Gain
\+* +
the presence of strong signals on all *!
multipath USMs have automatic gain control on all
receiver channels. Transducers typically generate
the same level of ultrasonic signal time after time.
The increase in gain on any path indicates a weaker
signal at the receiving transducer. This can be
caused by a variety of problems such as transducer
deterioration, fouling of the transducer ports, or
liquids in the line. However, other factors that affect
signal strength include metering pressure and flow

Figure 5 / Gain at 20 fps / 16 inch Meter

Note that the gains on each of the pairs are very
similar, and the gains by path are higher in the
middle two paths. This is due to the increased path
length requiring additional amplification. Figure 3
shows the same meter at 155 fps.

Figure 6 / Gain at 155 fps / 16 inch Meter

Figure 6 shows the gains for all pairs have
increased. This is normal when a meter is operating
at much higher velocities due to signal attenuation.
However, notice both graphs have the same look in
that the center pairs have higher gains than the
outer ones. Again, this is due to the longer path
length.
Signal Quality Transducer Performance
This expression is often referred to as performance

(but should not be confused with meter accuracy).

All ultrasonic meter designs send multiple pulses
across the meter body to the opposing transducer in
the pair, before updating the output. Ideally all the
pulses sent would be received and used. However,
in the real world, sometimes the signal is distorted,
too weak, or the received pulse does not meet
certain criteria established by the manufacturer.
When this happens the electronics rejects the pulse
rather than use something of questionable quality
that might distort the results.
The level of acceptance (or rejection) for each path
is generally considered as a measure of
performance, and is often referred to as signal
quality. Unless there are other influencing factors,
the meter will normally operate at 100%
performance until it reaches the upper limit of the
velocity rating. Here the transducer signal becomes
more distorted and some of the waveforms will
' +*'
   *
performance will drop from 100% to something
less. Figure 7 shows the performance of a 16 meter
at a velocity of about 20 fps.

Figure 8 / Transducer Performance at 155 fps

Signal-to-Noise Ratio
Signal to noise (SNR) provides information that is
'+
the user of possible impending problems. Each
transducer is capable of receiving noise information
from extraneous sources (rather than its opposite
transducer). In the interval between receiving
pulses, meters monitor this noise to provide an
 +? L'@This noise can
be in the same ultrasonic frequency spectrum as
that transmitted from the transducer itself.
The measure of signal strength to the level of
? L'@ _
Ratio, or SNR for short. Typically this is not
monitored nearly as often as gains and
performance. SNR is generally not an issue unless
there is a control valve or other noise generating
piping component present. When that occurs, the
SNR values will drop. The magnitude of the SNR is
a function of the manuf '+
expressing the value. Figure 6 shows the SNR from
a 16-inch meter flowing 20 fps at the time of
calibration. As can be seen the SNR is about 40 dB.

Figure 7 / Transducer Performance at 20 fps

Figure 8 shows the same meter operating at 155
fps. As we can see the performance has fallen from
100 percent on all paths to the 90+% range. This is
normal for high velocities as signal distortion will
have some impact on waveforms at these high
velocities.

Figure 9 / SNR at 20 fps Meter Velocity

Figure 10 shows the same meter at about 155 fps.

The SNR values have decreased by about 5-15 dB,
depending upon whether they are upstream or
downstream. This is due to ultrasonic noise being
generated inside the piping. As the downstream
transducers face the upstream direction, the
increased level of noise has more impact on the
downstream transducers. Also note that the SNR for
the middle pairs has decreased more than the outer
pairs. This is due to the path length being longer and
thus attenuating the signal more.

Figure 11 / Poor SNR on Path 4

Here we can see that the SNR from upstream to
downstream is not consistent. All of the SNR values
of AB are lower than BA. This is due to a problem
with the electronics. Figure 12 shows the results of
the same meter after the electronics was replaced.

Figure 10 / SNR at 155 fps Meter Velocity

Noise levels can become excessive if a control valve
is placed too close to the meter and the pressure
differential is too high. When this happens the meter
may have difficulty in differentiating the signal from
the noise. By monitoring the level of noise, when no
pulse is anticipated, the meter can provide
information to the user, via the SNR, warning that
meter performance (signal quality) may become
reduced. In extreme cases, noise from control valves
?#*@*t that the meter
 *!#+
transducers can handle significant levels of control
valve noise. By using transducers that have a higher
frequency, combined with higher efficiency and
stronger sound pressure levels, the affects of control
valve noise have been significantly reduced as
compared to past generations of USMs. Figure 11
shows a picture of a meter and a control valve
located immediately downstream of the USM.
SNR can also be low if the electronics has a
problem or there is a poor connection between the
transducer and the electronics. Figure 11 shows the
SNR graphed when there is a problem with the
electronics.

Figure 12 / Good SNR on all Paths

In Figure 12 we can see that all the SNR values are
now close to 40 dB. This is the normal for this meter.
Even though the SNR was poor in Figure 11, the
*+ #`{|
were normal. Thus, it is possible to have low SNR
and all other diagnostic indicators are normal.
Velocity Profile
Monitoring the velocity profile is possibly one of the
most overlooked and under-used diagnostic tools of
'  * '
as to the condition of the metering system, as well
as the meter. AGA Report No. 9 requires a multipath
meter provide individual path velocities. Once the
USM is placed in service, it is important to
collect a baseline (log file) of the meter. That is,

record the path velocities over some reasonable

operating range, if possible. These baseline logs can
also be obtained at the time of calibration. However,
as the piping in the field will likely be different than
that at the calibration facility, there could be some
minor changes in profile. Good meter station
designs produce a relatively uniform velocity profile
within the meter. The baseline log file may be helpful
*+ &'
at a later date. Figure 13 shows the velocity ratio of
 * 
This ra *'L *
velocity during a period of time and dividing it by the
average velocity as reported by the meter over the
same period of time. Since the ratio for each path
remains essentially constant at all meter velocities,
* 
than by looking at the actual velocity on each path.

Figure 14 / Path Ratios at 2.8 fps

When comparing Figure 13 and 14 it is very clear
that the velocity profiles are very different. Both of
these were taken from a 16-inch meter at the time of
calibration. Even with the difference in path ratios,
*+ #* ~'
15 shows th?+'@+
calibration.

Figure 13 / Path Ratios at 23 fps

Typically the ratio for a chordal design meter is
about 91% (ratio = 0.91) for paths 1 and 4, and
about 102% (ratio = 1.02) for paths 2 and 3. The
difference in ratios is due to the fact that the outer
paths are closer to the pipe wall, and thus the
velocity of the gas there is less than the gas that is
closer to the center of the pipe. When the velocity
falls below approximately 3 feet per second,
depending upon meter size and station design, the
velocity profile may change. Figure 14 shows the
 *+# 
2.8 fps.

Figure 15 / 16 inch Performance

~'# 
even though path ratios were different as shown in
Figures 13 and 14. This is the same meter
discussed earlier that was calibrated to 155 fps, but
the x-axis has been adjusted to better show the low
end performance. Looking at four path ratios takes
understanding why the velocities are different. Since
these can change by small amounts, a simpler
method of identifying changes in profile is desired. A
single value would be much easier to understand,
and also easier to quickly analyze. This value is
called Profile Factor. The Profile Factor is computed
by adding the velocity ratios of paths 2 and 3
together and dividing by the sum of the ratio of paths
1 and 4. The equation looks like this: Profile Factor =
(2 + 3)/(1 + 4). Assuming that paths 1 and 4 are 0.91
and the path 2 & 3 values are 1.02 the Profile Factor

is about 1.12. This value does vary a little from

meter to meter due to piping installation effects and
to some degree the type of flow conditioner and its
distance from the meter. Another method used to
analyze path velocities is to compare the sum of
paths 1 & 2 to the sum of paths 3 & 4. This provides
a look at the symmetry of the profile from top to
_* #
very symmetrical resulting in a value close to 1.000.
Figures 13 and 14 show both the Profile and the
Symmetry In Figure 16, when the meter was flowing
at 20 fps, the Profile Factor was 1.115. As the
velocity dropped to 2.8 fps (Figure 17) the Profile
Factor changed to 1.137. This is about a 2% change
in profile when comparing the middle paths to the
outer paths.

showed what happens to the Profile Factor and

Symmetry due to low velocity operation. This profile
change is typical when the meter is operated at
lower velocities. Figure 18 shows an ideal profile
from a 12-inch meter. This was based on the log file
collected at the time of calibration. Users have often
asked what impact partial blockage of a flow
 ' !
was used to show what happens not only to the
profile, but to quantify the change in accuracy.

Figure 18 /12-Inch Meter Profile / Normal

Figure 16 / Profile Factor and Symmetry at 20 fps

The Profile Factor for this meter is 1.118. For the

second test, the flow conditioner was modified to
have about 40% of the holes blocked with duct tape.
Duct tape was used to ensure repeatability. Figure
19 shows the flow conditioner just before it was
installed in the pipeline.

The other diagnostic worth reviewing is the

Symmetry value. Figure 14 shows a significant
change (on the order of 10%) in the Symmetry at the
lower velocities, but again there was no significant
impact on meter performance. This just indicates
there was a chan*+

Figure 19 / 40% Blocked Flow Conditioner

Figure 17 / Profile Factor and Symmetry at 3 fps

The Profile Factor can be a valuable indicator of
abnormal flow conditions. The previous discussion

Figure 20 shows the velocity ratios during the time

the flow conditioner was blocked. This was taken at
a velocity of 66 fps. The profile at two other
velocities, 22 and 45 fps, looked the same.

Profile Factor alone would be a good indication if

there was any contamination or flow conditioner
blockage. This may not always be true. Figure 22
shows a picture of a flow conditioner with 3 holes
blocked at the bottom.

Figure 20 / 12-inch Meter Profile / Blocked

The Profile is obviously distorted with higherthannormal readings on path 3 and 4, and lower
than normal on paths 1 and 2. The flow conditioner
was installed with the blockage at the bottom of the
pipe. As the gas flowed through the open holes,
there was a low-pressure created just downstream
of the blocked area causing the gas to then
accelerate downward, thus causing the higher
velocity at the bottom of the meter. The Profile
Factor for this 12-inch meter, as determined from
Figure 20, is 1.0!++ 
like much, but it certainly indicates a significant
change in profile. After installation in the field a
meter typically will generate a Profile Factor that
is repeatable to 0.02. However, this does depend
upon the piping, and makes the assumption that
there are no other changes like flow conditioner
blockage. The next question is what was the impact
on accuracy with this distorted velocity profile?
Figure 21 shows the result of the three test velocities
and the impact on metering accuracy.

Figure 22 / 3 Holes Blocked Flow Conditioner

In this test the three blocked holes in the flow
conditioner were located at the bottom of the meter
run. This is the same 12-inch meter and testing as
discussed with the 40% blockage. Figure 23 shows
a graph of the Path Ratios during this test.

Figure 23 / Path Ratios with 3 Holes Blocked

Figure 21 / Blocked CPA Results

As can be seen the meter was affected by an
average of about 0.15% for all flow rates. In this
case the meter slightly under-registered with this
distorted profile. Later in this paper a more
advanced diagnostic feature will also show the
meter has blockage, but for now one can see the
Profile Factor has indicated a significant change.
In the past many have thought that looking at the

When we compare Figure 23 with Figure 18 (the

normal Path Ratio profile) it is obvious that the two
do not look the same. However, when computing the
Profile Factor, the average value for Figure 23 is
1.114. This is almost the perfect number for this
meter. Upon further inspection we see in Figure 24
that the Symmetry is not correct.

This data was taken from the meter operating at 23

fps and showing a very stable reading. Here we can
see all of the me\' '
perhaps an easier way of looking at the SOS values
 * *\
average value. Doing this is makes it easier to spot
problems. Figure 26 shows the percent difference of
each path *
SOS.

Figure 24 / Profile Factor and Symmetry at 40 fps

In Figure 24 we see the average for the Profile
Factor is 1.114 and almost normal, but that the
Symmetry value average is 0.95, or about a 5% shift
from normal. Thus it is possible for the Profile Factor
to be normal, when the velocity profile in the meter is
not. The combination of Profile Factor and Symmetry
are needed to fully analyze the velocity profile
entering the meter.
Speed of Sound
Probably the most discussed and used diagnostic
tool of an ultrasonic meter is the speed of sound
(SOS). The reader may recall that speed of sound
on an individual path is basically the sum of the
transit times divided by their product, all then
multiplied by one half of the path length. A more
detailed discussion on this can be found in a
previously presented paper [Ref 5]. There are at
least 2 ways of looking at SOS. The first would be to
* *\\
calculated by the meter. Figure 25 shows a graph of
the SOS of a 10 inch meter at the time of calibration.

Figure 25 / SOS by Path at Calibration

Figure 26 / Path Percent Difference in SOS

Here the difference, in percentage, of each of the
*\'#'|!
indicates good correlation between each path and
also no temperature stratification within the meter.
When a meter is operated at lower velocities,
typically less than 3 fps, and there is a large
difference between the gas and atmospheric
temperature, heat transfer can occur. As the heat
transfer occurs, internal temperature gradients can
develop. When this happens the hotter gas inside
the pipe rises to the top of the meter. Since the
speed of sound in the gas is relatively sensitive to
temperature, this will be seen as a SOS difference
between the paths. This is often called thermal
stratification. Figure 27 shows the SOS values of the
same 10 inch meter when it is operated at 1.8 fps at
the calibration lab.

Figure 27 / Thermal Stratification Effects

Figure 29 / 10-inch 20 fps Path Ratios

From Figure 27 we can see the average percent

difference in SOS compared to the meter has
increased a little. This is due to a slight thermal
gradient within the meter. That is the gas at the top
of the meter is hotter than that at the bottom. Path 1
color is blue, path 2 is red, path 3 is green and path
4 is gold in color. Figure 27 shows upper paths have
increased and the ones at the bottom decreased.
This difference in SOS may be thought to impact the
accuracy of the meter. In extreme cases this can be
the case. However, for this example the impact is
virtually on-existent. Figure 28 shows the results of
this 10-inch at the time of calibration.
Figure 30 / 10-inch 1.8 fps Path Ratios
Even though this meter had some thermal
+  *+ 
significantly. There is a difference, but once again it
is not as significant as the blocked flow conditioner
example in Figure 20.
ADVANCED DIAGNOSTIC INDICATORS

Figure 28 / 10-inch Low-Flow Error

_?+'@+*'
?+'@{+*!'#
very little impact in the performance of the meter
even though there was some thermal stratification.
Figures 29 and 30 show profiles for this same 10inch meter at 20 and 1.8 fps respectively.

During the past several years an additional

diagnostic feature has been studied by Engineering.
!+' ?!'' @ '
thoroughly in a previous paper [Ref 6]. Essentially
Turbulence is a measure of the variability of each
*city readings during the time the meter
was sampling, and is provided each time it updates
the velocity information. This gives the technician an
idea of the steadiness of the flow as seen by the
meter. Typically the level of turbulence on a
Westinghouse design shows paths 1 and 4 to have
around 4% turbulence, and paths 2 and 3 around
2%. This is based upon the history of many meters.
The outer paths 1 and 4, being closer to the pipe
wall, always exhibit higher turbulence because they
are more affected by the surface friction of the
upstream piping.
Turbulence can be computed from the maintenance

log file for older meters. With the advent of more

advanced electronics, it is now computed real-time
in the meter and reported on the maintenance log
files. This greatly reduces the time for analysis since
it is not only stored in the log file, it is graphed out
automatically for quick review. Recently viewing
Turbulence has solved several metering problems.
Distorted velocity profiles often cause concern about
metering accuracy. If the velocity profile, as shown in
Figure 18, now appears like that in Figure 20, the
cause needs to be determined. Some might feel this
is just due to upstream affects and may not believe
there is any object blocking the flow conditioner.
The 12-inch meter in Figure 31 shows a very
consistent level of Turbulence during the period of
the test. It was collected at the time of calibration
and the velocity was about 66 fps.

Figure 31 / Normal 12-inch Meter Turbulence

Figure 32 is the same meter with a blocked flow
conditioner as shown in Figure 16.

Figure 32 / High 12-inch Meter Turbulence

It is clear that the turbulence in Figure 32 is about 3
times higher. Certainly the velocity profiles for this
meter, shown in Figures 18 and 20, look different.

Anyone looking at the blocked profile would

immediately recognize there is a problem. It is
possible, however, to have a complete blockage of a
flow conditioner with something like a porous bag
and have a relatively symmetrical profile. In this
situation the turbulence would be excessive,
indicating there is a problem with blockage. This has
been observed in the field and without Turbulence it
would have gone un-detected.

Sizing of USMs
Traditional measurement devices have been limited
to flow rates that were equivalent to 50-60 feet per
second (fps) maximum. Although high-capacity
turbines operate in the 80-95 fps range, the majority
of installations are still designed for the standard
capacity meter. One of the significant advantages of
ultrasonic meters is their ability to operate accurately
in excess of 100 fps with no damage to the meter.
* + ?'-of-'@
the capacity of an ultrasonic meter is at least 1
times that of an orifice meter.
Many users specify velocity limits in piping to
minimize the potential for erosion in fittings (bends,
elbows and tees) that can cause failure of the piping
should the wall thickness become too thin. There
have been reports of severe erosion of elbows
downstream of meters that were operated above
100 fps for significant periods of time. However,
there are several other issues that limit high velocity
operation.
Higher velocity operation increases the stress on the
thermowell(s). There have been reported cases of
thermowells cracking when operated at high
velocities. Some work has been performed to
determine how accurate the RTD element remains
when the thermowell is subjected to very high
velocities. Studies indicate that vibration of the
thermowell may cause the RTD to register higher
'  ?+-@   ' #  !
element rubs against the thermowell.
Higher velocities also create more differential
*'    -performance flow
conditioners.
The magnitude of this loss is
somewhat dependent on the type of conditioner
installed. However, typically the pressure drop at 60
fps is on the order of 2-3 psi differential (psid). If the
gas velocity through the meter is increased to 120
fps, the differential becomes 4-9 psid.
The
increased velocity and subsequent differential
pressure also can create higher noise, both audible
and ultrasonic. The ultrasonic noise at these high
     + #  
operation.
The audible noise may necessitate

additional costs related to noise abatement or

possible facility relocation to minimize impact on
surrounding environment.
Primarily these reasons, most designers limit normal
operation to either 70 or 80 fps. It is tempting to
install a smaller meter and operate at higher
velocities. However, the initial cost savings of going
to these higher velocities can quite possibly be offset
with increased maintenance and reliability problems
that will occur later. The real benefit of the higher
velocity capability is the ability to have accurate
measurement at the higher flow rates which may
occur during upsets or other abnormal conditions. In
addition, the high velocity will not damage the meter
as may be the case with orifice or turbine meters.
Low Flow Performance Issues
Traditionally USMs have been sized to operate so
that the lowest velocity expected is between 5 and
10 fps. Using 5 fps as the lowest velocity provides a
rangeability of 14-1 if the maximum velocity
expected is 70 fps. Comparing this to traditional
devices such as orifice and turbine this rangeability
is generally better. However, larger turbine meters
can approach this rangeability (and in some cases
exceed this). Recently users have been asking how
to increase the rangeability in order to help reduce
CAPEX (capital expense) cost and operation
maintenance cost (O&M, or OPEX) of the meter
station. Since most users limit their maximum
velocity to 70-80 fps, the only area left to expand
rangeability is to operate the meter at lower
velocities.
During the past several years many users have
come to see the value in calibrating meters to
velocities as low as 1 fps (and sometimes even
lower). Calibrating a meter to below 5 fps is often
referred to as a ?#-+# @-8
data points are used for calibrations with a 5-70 fps
operation, and 2 additional data points, often 1 and 3
fps, are added for the low flow calibration. By
lowering the minimum velocity to 1 fps the user has
essentially multiplied the rangeability by 5!
A typical installation for a gas powered electric
generation plant might include a 12-inch meter for
the main flow rates, and a 6-inch USM for the lower
flow rates. Assuming the lowest the 6-inch is
operated is 5 fps, this would equate to about 1 fps
for the 12-inch meter.
This translates into a
rangeability of approximately 80-1 by using both
meters. Suppose a single 12-inch meter could be
operated at this low velocity with a reasonable
degree of accuracy. What would be the capital cost
savings?

If the designer uses a low flow meter (say a 6-inch

USM) with a 12-inch meter, in lieu of installing a
single 12-inch meter, it would probably cost on the
order of $60-90K more. By purchasing only one
meter, and not having to do run switching with all the
associated piping and control systems, the
measurement station is also much easier to
maintain.
The capital cost saving is only the
beginning. The benefit of less equipment translates
into lower O&M costs for the life of the station. One
hidden benefit is with no run switching, the
possibility of the main run not activating is eliminated
since there are no automated actuators required to
open a valve. When a power plant goes online, and
tural gas they need, this
generally causes a loss in revenue, and the end
user is not happy with the gas provider.
There are probably two major concerns about
operating a USM at these low velocities. One is
accuracy and the second is repeatability. The
accuracy concern is from the absolute performance
of the meter. That is, how accurate can the meter
be after flow calibration at very low velocities?
Another accuracy concern is how accurate will the
meter be if there is thermal stratification (gas hotter
at the top of the meter than at the bottom) within the
pipeline. That is, if the gas temperature at the top of
the pipe is hotter than the bottom, the measured
temperature from the RTD will read higher, and thus
under-registration is likely. Also, this can contribute
to flow profile problems (distortion of the velocity
profile) which also may impact metering accuracy,
although this is generally not considered to be
significant.
     
# # 
technology, these can be solved very easily. The
L &'   '   ?  
provide a method of correction to reduce uncertainty
at these very low flow rates, and can thermal
+ +~L#
improve accuracy at very low flow rates.
Regarding the accuracy of a meter, a calibration
technique called piece-wise linearization (PWL),
commonly used for optimizing turbine meter
performance, can be applied within the USM. This
technique permits substantial reduction of the
uncertainty at low flow rates. That it, assuming the
meter is calibrated at the very low flow rate
(velocity), the resultant uncertainty will become very
small since any errors at these lower velocities have
  +!' '  
be exceptional once calibration is complete, and the
PWL technique has been implemented. Following is
an example of a 12-inch meter that was calibrated
from 2 fps to 118 fps.

12-inch As Found and As Left Results

1.4

As Found

As Left

Verification Points

1.2
1.0
0.8
0.6

% Error

0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
-1.2
-1.4
0

20

40

60
80
Meter Velocity (ft/sec)

100

120

140

Figure 33 / 12-Inch Meter Calibration Results

As can be seen from this calibration result in Figure
   *+    +* 
approximately 0.3% from the average of all the other
flow rates prior to adjustment. It is important to note
+' easily
provide accuracy values much greater than
previously thought possible. This is due, in part, to
the improvements in manufacturing and dry
calibration that have been achieved during the past
several years, and also to technology improvements
in electronics.
After implementation of PWL meter calibration
factors, all errors are driven to zero, so the
uncertainty is essentially reduced to that of the lab.
There is an additional uncertainty when operating
between calibration points. This is because the
PWL technique assumes the error between
calibration points is a straight line. However, due to
the typical linearity of the un-calibrated meter by
using an appropriate number of data points at the
low velocities, this added uncertainty should not
exceed 0.1%.
The second accuracy concern comes from the
possibility of thermal stratification at lower velocities.
A previously published paper discussed a 10-inch
meter that had thermal stratification on the order of 2
degrees F [Ref 4]. This stratification occurred
because the gas in the upstream piping was flowing
at velocities below 3 fps, and more than 100 feet of
the upstream piping was subjected to a bright sun.
Also, there was a significant difference in gas-toambient temperature. One might say this is typical,
but perhaps this is not a typical installation.

      '*

piping below grade (underground). Piping emerges
above ground and then makes a 90-degree direction
change into the meter. With a 10-inch meter, this
typically means 20-25 feet of piping is exposed to
the sun and ambient temperature. In this condition,
even when operating at 2 fps, the gas takes less
than 15 seconds from the time it exits the elbow or
tee to enter the meter run before it reaches the
meter. In 15 seconds there probably is not sufficient
time for any significant heat transfer to occur and
cause stratification within the flowing gas stream.
Thus, the possibility of thermal stratification of the
gas at these low velocities is very minimal.
If significantly more upstream piping is exposed to
ambient and solar effects than in the previous 10inch meter design example, there is a possibility of
thermal stratification. However, this can be easily
detected by looking at the speed of sound reported
by a meter that samples transit times horizontally.
That is, if the gas temperature is hotter at the top,
the SOS on the top path will read higher than the
#  \ # *   
gas velocity of 5-10 fps. By being able to see there
is no deviation in SOS at the lower velocities, the
user can have confidence that there is no thermal
stratification, and thus temperature measurement is
accurate. The caveat is that not all USM designs
can provide this type of information.
T    +  +   L
 *\
of calibration velocities. Figure 6 below shows the
results of a 10-inch meter calibrated from velocities
of about 1.8 fps to 118 fps.

10-inch As Found and As Left Results

1.00
As Found

As Left

Verification Points

0.75
0.50

% Error

0.25
0.00
-0.25
-0.50
-0.75
-1.00
0

20

40

60

80

100

120

Meter Velocity (ft/sec)

Figure 34 / 10-Inch Meter Calibration from 1.8 to 118 FPS

  +   " L   *
SOS values at two velocities to see how much
stratification may have occurred.

After meter adjustment the uncertainty of the meter

would be that of the lab (about 0.2%) and the
repeatability of the meter (probably on the order of
0.1% or less). At the lower velocities there could be
Speed of Sound at 22.7 ft/s

1374.0

1379.0

1373.0

1378.0

1372.0

Figure 35 / Path SOS at 23 fps

Figure 35 shows the SOS difference of each path
*\!*
a good look at how consistent the individual SOS
values are relative to the average. This figure
shows that the spread of path SOS values is on the
order of about 0.05%, or less than 0.7 fps. Figure
36 shows the same meter at 1.8 fps and indicates
the spread in SOS has increased slightly. Now the
maximum deviation in SOS is on the order of 0.1%,
or about 1.4 fps. Thus, we can see there is a little
thermal stratification occurring since the path at the
bottom of the meter (indicated by the gold colored
line) is reading lower than shown in Figure 3,
(relative to the average / purple line) and the path at

Path 3

Path 4

120

110

90

70

50

120

110

100

90

80

70

60

50

1367.0

40

1368.0

1373.0
30

1369.0

1374.0
20

1370.0

1375.0

10

1376.0

Time (sec)

Path 2

1371.0

40

1377.0

30

SOS

1380.0

Path 1

100

Average

1375.0

80

Path 4

60

Speed of Sound at 1.8 ft/s

Path 3

20

Path 2

10

Path 1

1381.0

SOS

Average

Time (sec)

Figure 36 / Path SOS at 1.8 fps

the top of the meter (shown by the blue line) is now
reading higher than in Figure 35 (relative to the
average). Since higher temperatures increase the
SOS of the gas, the path at the top now increases
compared to the path at the bottom. This is an
indication of thermal stratification, but the magnitude
is very small, and thus there is no significant impact
' 
When a meter is operated at lower velocities, and
thermal stratification is possible, there may also be a
change in the velocity profile of the meter. Figures
37 and 10 show path velocity ratios from this meter
at the same two velocities.

Flowing Velocity Ratios at 22.7 ft/s

Flowing Velocity Ratios at 1.8 ft/s

0.908

Path 1

Path 2

1.022

Path 2

Path 3

1.021

Path 3

0.912

Path 4

0.80

0.85

0.90

0.95

0.930

Path 1

1.037

1.005

0.894

Path 4

1.00

1.05

1.10

Path Ratios

Figure 37 / Path Ratios at 23 fps

When this meter was operating at 23 fps, the path
velocity ratios are symmetrical and normal. As the
    '  #   
change in the ratios. The values for paths 1 and 4
are no longer very close (about a 4% change), and
the ratios for paths 2 and 3 also show some minor
change (about a 3% change). This is normal at
lower velocities, and does not have any significant
impact on accuracy. This can be validated by
#?-+'@ #
in Figure 34.
"'-inch meter as an example for cost
savings based upon low-flow performance.
Suppose a site had an operational range of 225
MMSCFD to 5 MMSCFD at 750 psig. If the user
selected an operational range of 5-80 fps, this would
require a 6-inch meter to handle the lower flow rate
at 5 fps. At 5 fps the 10-inch meter could handle
approximately 14 MMSCFD. However, the engineer
might decide to install a smaller meter since lowest
flow rate is 5 MMSCFD. If the 5 fps minimum
operation is followed, the designer would select a 6
inch meter to handle the low flow rate. The 10-inch
meter would operate at 2 fps if no low flow meter
were used.
This is well within the accurate
operation of the meter as shown in Figure 34.
Looking at the economics of adding the additional
smaller 6-inch meter in this scenario, the added cost
could easily exceed $75,000 when including the
headers, valving, meter runs, RTU and actuators
needed to sequence the 6 and 10-inch meters. A
single 10-inch meter could be used for this example
by operating it at 2 fps (instead of the typically lower
limit of 5 fps), substantially reducing CAPEX, and
certainly reducing the long-term O&M (OPEX) costs.
Perhaps there might be some increased uncertainty.
However, even if the uncertainty were assumed to
be 0.1%, this would translate about $1.15 per hour
based on $5 per thousand cubic foot gas price.
Thus, there is probably not enough potential

0.80

0.85

0.90

0.95

1.00

1.05

1.10

Path Ratios

Figure 38 / Path Ratios at 1.8 fps

reduction in measurement uncertainty to justify
spending more $75,000.
Dual or Multiple Meter Runs
For larger applications, many designers choose to
use two similar sized meters in parallel. The break
point often comes when the flow rate dictates a
meter size of 16-inch and above. There are many
variables to consider before choosing whether one
meter is a better solution, or if two might be the best
choice.
Two meters provide some redundancy. This might
be important if company policy dictates removal and
periodic re-calibration. Also, if maintenance policy
requires periodic internal inspection, having two runs
makes it much easier to perform this task without
having to bypass the meter and estimate volumes.
Additionally, should one meter develop an
operational problem, having a second meter may
help identify the problem(s) Of course the meters
must be sized to be capable of handling the entire
capacity should one be taken out of service, or this
benefit may not be available.
The disadvantage of using two meters is obviously
the capital cost and additional maintenance
required. Two smaller meters will require more
capital expense in the beginning, and most likely
more long-term O&M cost. Which design is best for
any given application somewhat depends upon the
operational requirements, space limitations, flow
rangeability, and other factors.
Generally speaking, designers that choose to use
two or more mete  *    +
&'!*
least 30-1 rangeability, and many are flow calibrating
to achieve in excess of 80-1 rangeability. The
primary reason is for redundancy and to permit
removal of a meter for maintenance while measuring
all station flow through the other meter(s).

Basic Piping Issues

As with any other technology, ultrasonic meters
require adherence to basic installation guidelines.
Recommendations related to the installation of
primary metering elements, such as the orifice and
turbine, have been in place for a long time. These
are provided through a variety of standards (API,
AGA, ISO, etc.) to insure accurate performance
(within some uncertainty guidelines) when installed.
The reason for th '    
accuracy can be affected by profile distortions
caused by upstream piping and proximity to noise
generating control valves. One of the benefits of
         +
upstream piping designs with less impact on
accuracy than other primary devices.
Installation effects on measurement devices have
been studied in much more detail than ever before.
This is due in part to the available technology
needed for such evaluation. Much of this research
has been focused on ultrasonic meters. Reducing
uncertainty has also become a higher priority for
pipeline companies today due to the increasing cost
of natural gas. Designing an ultrasonic meter station
that provides the same installed accuracy as that at
the time of calibration is very important.
According to research work performed at Southwest
Research Institute (SwRI) by Terrance Grimley, it
would take a minimum of 100D of straight pipe for
the profile to return to a fully symmetrical, fully
developed, non-swirling velocity profile [Ref 6].
Complex upstream piping, such as two elbows out of
plane, create even more non-symmetry and swirl
###!'
handle profile distortion and swirl in order to be
accurate and cost-effective. However, just as with
orifice and turbine meters, installation guidelines
must be followed to achieve a predictable accuracy.
In 1998 AGA released AGA Report No. 9 /
Measurement of Gas by Multipath Ultrasonic Meters.
The report was updated and revised in April 2007.
This document discusses many aspects and
requirements for installation and use of ultrasonic
meters. Section 7.2.2 specifically discusses the
USMs required performance relative to a flow
calibration. It states that the manufacturer must
?  '*  # **
configuration in minimum length / one without a flow
conditioner and one with a flow conditioner - that will
not create an additional flow rate measurement error
of more than r0.3% due to the installation
+'@    # '  

were calibrated with ideal flow profile conditions, the

manufacturer must then be able to recommend an
# # ' ' 
to deviate more than r0.3% from the calibration
once the meter is installed in the field.
During the past several years a significant amount of
testing was conducted at SwRI in San Antonio,
Texas to determine installation affects on USMs.
Funding for these tests has come from the Gas
Technology Institute, formally knows as the Gas
Research Institute (GRI). Much of the testing was
directed at determining how much error is introduced
in a USM when a variety of upstream installation
conditions are present. The results of the testing
were presented in a report at the 2000 AGA
Operations Conference in Denver, Colorado [Ref 6].
From this data one can conclude that upstream
piping (elbows, tees, etc.) does have an effect on
  *+   \  * 
installation affects test with no flow conditioner when
located 20D from the effect. Nevertheless, most
users still choose to use a flow conditioner in order
to reduce potential upstream effects.
Which brand of flow conditioner to choose often
* '*  ' *+   ~ e
SwRI report the results indicate that several brands
and styles of flow conditioners work well. The only
exception is that 19-tube bundle conditioners are not
recommended by USM manufacturers as test results
are inconsistent and generally not as good as other
flow conditioners. This is due to the fact that these
tube bundles are good for straightening the flow
(eliminating swirl) but are ineffective in modifying
and generating a consistent flow profile.
Each manufacturer has conducted extensive testing
on their meter with and without a flow conditioner.
From this data it is apparent that ultrasonic meters
have reduced uncertainty when a flow conditioner is
installed. With the price of gas expected to continue
increasing for the foreseeable future, virtually all
companies in North America install USMs with a flow
conditioner. Not only does the flow conditioner
reduce uncertainty, but it makes the velocity ratio
diagnostics much easier to understand.
Figure 39 shows an ideal velocity profile for a 12inch, 4-path meter with significant diameters of
straight pipe upstream and a flow conditioner
installed 10D upstream of the meter. Figure 8 shows
the same meter with an installation effect of 3
elbows and a tee upstream of the piping (10D
upstream of the flow conditioner).

Meter A
Meter A

Straight - CPA - Flowing Velocity Ratios at 75.8 ft/s

Path 2

1.020

Path 2

Path 3

1.020

Path 3

0.80

0.85

0.90

1.014

1.027

0.935

Path 4

0.913

Path 4

0.892

Path 1

0.915

Path 1

Elbows - Tees - CPA - Flowing Velocity Ratios at 75.7 ft/s

0.95

1.00

0.80

1.05

0.85

0.90

0.95

1.00

1.05

Average Flowing Velocity (ft/sec)

Average Flowing Velocity (ft/sec)

Figure 39 / 12-inch Path Ratios / Ideal

Figure 40 / Path Ratios with Installation Effect

CPA Flow
Conditioner
10D
4 + 4 Meter (8 Path)
10D

>5D

Figure 41 / 12-Inch Meter Upstream Piping Disturbance

Meter A

Elbows - Tees - No FC - 20 - Flowing Velocity Ratios at 75.7 ft/s

SICK Maihak Electronics "A" Installation Effects Results

0.8

1.067

Path 1

0.6
0.4

% Error

1.061

Path 2

0.2
0.0
-0.2

0.960

Path 3

-0.4
-0.6

0.804

Path 4

-0.8
0

10

20

30
40
Velocity (ft/sec)

Straight Pipe with CPA 50E 10D

50

60

"T"s with Elbows CPA 50E 10D

Figure 42 / Accuracy Impact on 12-inch Meter

70

0.6

0.7

0.8

0.9

1.0

1.1

1.2

Path Ratios

Figure 43 / Path Ratios with No Flow Conditioner

Figure 41 shows a pictorial drawing of the upstream

piping disturbance that was used to create the lessthan-ideal profile in shown Figure 40. Figure 42
shows the impact on accuracy the three elbows and
one tee had on 4-path meter. The table inside of
Figure 42 shows the effect on accuracy was less
than 0.1%. Figure 15 shows what the profile looks
like with no flow conditioner. Clearly this is a very
distorted profile.
The effect on accuracy for the profile in Figure 43
was approximately 0.15%. Even though the error
was small, having a profile that is this distorted
makes understanding whether the meter is still
operating correctly very difficult.
Thus, most
customers prefer to install a flow conditioner in order
to have a profile that is much more like the baseline
one shown in Figure 39.
Piping Recommendations
When AGA 9 was released in 1998, it did not include
any recommendations for upstream or downstream
piping lengths. At that time the GTI installation
affects data was not available. Rather than specify
any minimum requirements, it was decided to allow
the manufacturer to state their recommendation

 '*     *+ 

With the data that is available today, most users
have now adopted their own company standards.
Although they vary from company to company, most
are using a more conservative approach to insure
the best possible performance in the field.
AGA 9 has been revised and the Second revision
was released in April 2007. In Section 7, Installation
Requirements, two recommended default piping
designs, one for uni-directional and one for bidirectional applications, are included. For the unidirectional design there is a recommendation of two
10D upstream spools with a flow conditioner in the
middle (10D from the meter). For the bi-directional
design,
both
upstream
and
downstream
recommendation would be two 10D spools with flow
conditioners again located 10D from the meter.
Figure 44 shows an example of the piping design
recommendations. These are only recommendations
and the use of the optional Tee or Elbow is strictly
optional. The optional tee may be installed to help
attenuate ultrasonic noise from control valves (see
section entitle Control Valve Noise), or for ease of
internal inspection of the meter run.

Figure 44 / AGA 9 Recommended Default Piping

The location of thermowells is discussed in Section
7.2.5 of AGA 9. It states that the thermowell should
be located between 2 and 5 nominal diameters
downstream of the meter. The reason for locating it
close to the meter is for the same reason as other
primary devices / to make sure the temperature at
the RTD is the same as in the meter. For bidirectional applications, a potential issue arises.
AGA 9 states the RTD should be 3-5D from the
meter. This is to minimize the effect turbulence
(caused by the protrusion at the inlet to the meter)
 ' 
SwRI conducted tests with thermowells located a
distances ranging from 1D to 5D upstream on more
than one manufacturer of 8-inch ultrasonic meters.
Results showed no measurable effect when the
thermowell was at 3D, with only a slight influence at
2D.
From this testing AGA 9 adopted the
recommendation of a minimum of 3D upstream.
Installing thermowells outside of the flow conditioner
area (much further than 5D from the meter) may
cause inaccurate temperature measurement.
Some designers include a tee upstream and
downstream of the meter run. The purpose of the
tee is to permit easy inspection inside the meter run.
The addition of a tee, rather than an elbow, has long
been used on orifice meters. However, some
designers feel the introduction of a tee upstream of
the meter creates a more distorted velocity profile.
Although flow conditioners do a very good job of
minimizing asymmetrical and swirling flow profiles,
they are not perfect.
At present there is published data to show the
impact a tee has on the accuracy of a meter.
Testing presented at the CEESI USM Conference in
2004 and 2006 [Ref 7 & 8] showed the impact on
installing an upstream tee, along with three elbows,
and using 10D + CPA + 10D between the tee and
the meter, to be on the order of 0.2%, or less, on the
 ' !*' ' 
profile distortion than simply installing one elbow and
Flow Calibration Basics
The primary use for USMs today is custody
measurement applications. As discussed earlier,
the introduction of AGA Report No. 9 has helped
spur this growth. Section 5 of AGA 9 discusses
performance
requirements,
including
flow
calibration. Initially the 1998 release of AGA 9 did
not require meters be calibrated for fiscal use.
However, with the publication of Revision 2, it is now
a requirement for all fiscal (custody) applications.

tee, and thus confirmed that using a tee upstream

does not cause the USM to deviate outside the 0.3%
requirement of AGA 9
Other Piping Issues
Using filters upstream of ultrasonic installations is
often a subject of discussion. Many designers use
filters to remove debris that may be traveling down
the pipeline. However, due to the non-intrusive
design of an ultrasonic meter, small particles
generally pass through the meter with little or no
damage.
Also, some ultrasonic designs have
'    *'   
wall, reducing the chance that flying debris will erode
or damage them.
Another concern relates to the impact any potential
buildup on the inside of the meter and associated
piping will have on measurement accuracy. Using a
filter or strainer may seem like a proper method of
eliminating any potential coating or debris
contamination that may occur. However, they may
become a high maintenance item with the potential
of becoming clogged if there is any significant
amount of oil or particulate present. Assuming there
is no oil or grease, particulates will just pass through
the meter (if no filter is used).
Other concerns are that the filter or strainer will
create a pressure loss in addition to possibly
becoming clogged and subsequently breaking apart.
Part of the element may lodge against the flow
  ! # *   
operation, and mos L ++   
accuracy. This can be generally identified at the
time of maintenance by observing the path velocity
information.
However, this will now require
disassembly of the meter run to inspect and clean
the foreign material. For these reasons, and the
issue of cost and additional space requirements,
using filters for ultrasonic metering applications is
not usually beneficial.
The basic accuracy requirement, prior to any
adjustment, is that 12-inch and larger meters must
be within r0.7%, and 10-inch and smaller meters to
be within r1.0% for the flow range of qt to qmax (low
flow rate to maximum flow rate). Below qt, the error
limit is r1.4%. If the meter is outside these values,
and is calibrated with upstream piping and a flow
conditioner (considered the package), then the user
*| ?+'@+'
be outside of this, the User now has the option to
 * +  #  !'  ?' +  @

performance has less meaning today than in the

previous AGA 9 version.
Initially USMs were installed without a flow
conditioner. However, customers are using flow
conditioners in the majority of applications today.
 +  '  ? *+ @ +#
conditioner (not a 19-tube bundle) provides more
consistent performance (reduced uncertainty). Even
though data exists to support the supposition that
some USMs perform quite well without flow
conditioners, the added pressure drop and cost is
often justified by the reduction in uncertainty. One
thing that most everyone does agree upon is that if a
flow conditioner is used with a meter, the entire
system should be calibrated together.
It should be noted that one of the benefits of the
ultrasonic meter is that it does not create significant
pressure loss. The pressure drop resulting from the
flow conditioner is offset by the lack of pressure drop
across the meter (when compared to orifice or
turbine meters). As such, total pressure loss across
the metering facility as a result of using flow
conditioner is probably no greater than with other
primary devices for the same given flow rate.
Most companies have standard designs for their
meters that typically specify piping upstream and
downstream of the flow conditioner(s) and meter.
Usually these USMs are calibrated as a unit with the
customers upstream and downstream piping spools.
Calibrating as a unit helps insure that the accuracy
of the meter, once installed in the field, is as close
as possible to the results provided by the lab.
In the past most customers felt their applications
deserved, and required, less uncertainty than the
minimum requirements of the original AGA 9 (where
flow calibration was not required). To ensure these
higher standards were met, virtually all users began
flow calibrating their USM meters used in custody
+ **      ``   
2002 AGA Operations Conference a paper was
presented that discussed the benefits of flow
calibrating ultrasonic meters [Ref 12]. Summarizing
from that paper, there are three main reasons users
are calibrating meters:
x
x
x

Reduce uncertainty
Verify performance
Improve rangeability

Calibration Labs
There are several flow labs in North America that
provide calibration services. Each will calibrate to
any number of points the designer feels are
necessary. Typically most designers are requesting
  {  *  \    ?-+'@ 
points have been determined, an adjustment factor

(or factors) is (are) computed. Facility personnel

enter the value(s) into the meter. Usually one or two
verification points are used to validate the predicted
?-+@*+ !# 
or two flow rates and verify the meter error is zero
(or very close to zero). Generally the USM will
repeat within r0.1% of the predicted value, with
more recent results showing verifications typically on
the order of r0.05%.
'``#-capacity calibration
facilities were commissioned in North America.
These two laboratories permit users to costeffectively calibrate larger USMs (greater than 10inch) to full capacity, thus reducing the uncertainty
related to measurement error when installed in the
field. In 1998 when AGA 9 was first published these
two facilities were not in operation. Since there was
no large capacity calibration facility in North America
in 1998, AGA 9 did not require flow calibration.
However, the new release of AGA 9 requires flow
calibration for fiscal (custody) measurement
applications.
Even though there is a substantial amount of data
showing USMs to be linear to better than r0.2%,
many want to reduce this error further. Because of
this, some users have implemented multi-point
linearization (PWL), within their flow computers, to
further reduce the uncertainty of their USM
calibration results. This is not a new technique, and
has often been used for turbine meters in the past.
When using this multi-point linearization technique,
external to the USM, at least two issues always
surface. First, the output signals from the USM
(serial, frequency and analog) would then be
adjusted in the flow computer to correct all errors. If
the output signal (serial or frequency) from the meter
is sent to two separate flow computers, care must be
taken to ensure both computers are utilizing the
same algorithm, or their results will differ. Second,
since the linearization is not taking place in the
meter, there is no audit trail within the meter to verify
 '*'   '   + Q?
left@    ** '  ! #
discussed in more detail in an AGA 2002 paper titled
?+ + ~#     +
12].
To overcome the above problem, some USM
manufacturers have provided the user with the
ability t * Y"      
This allows the linearization to be tested and verified
at a test facility, and the meter output does not
require external correction which has its inherent
problems.

Re-Calibration
AGA 9 does not require an ultrasonic meters to be
re-calibrated, and will not require this in the
upcoming revision. As USMs have no moving parts,
and provide a wide range of diagnostic information,
many feel the performance of the meter can be field
verified. That is, if the meter is operating correctly,
its accuracy should not change, and if it does
change, it can be detected. This, however, remains
to be proven with a significant number of meters.
The use of USMs for custody transfer applications
began increasing rapidly in 1998. Now, with more
than 9 years of installed base, there is significant
information to conclude USMs may not require recalibration. Many companies are not certain as to
whether or not they will retest their meters in the
future. They are waiting for additional data to
support their decision. Manufacturers are also trying
to show the technology may not require recalibration.
During the past several years, many meters have
been re-calibrated in Canada. Their governmental
agency, Measurement Canada, requires USMs to be
re-tested every 6 years. The data obtained from
these meter re-calibrations, from random re-testing
by customers, and long-term data from meters at
calibration labs typically shows the meter to be
within 0.3% of the original calibration assuming the
meter is clean and operating correctly.
Several users in the US have removed meters in the
past year and returned them to the calibration facility
for a quick verification. If the meter is clean, the
performance on these has typically been within 0.10.3%. Unfortunately, there is very limited published

information to date. Over the next several years, the

' L#  # *    term accuracy of the ultrasonic meter will continue to
grow.
Some designers have chosen to incorporate a
* ?+ @     
[Ref 13]. The purpose of this meter is to provide an
in-situ verification against all the other fiscal meters
at that location. The idea is to route all the gas from
a given operational meter periodically through the
?+ @   L  ' 
upon the difference. In many ways this is exactly
what the calibration facility is doing. However, this
technique has several issues that must be
addressed.
First, if there are any installation effects on the
reference meter, a bias could be introduced in the
results. The installation effect could come from the
upstream piping or pipeline contamination. Second,
the addition of a reference meter adds significantly
to the cost of the station. Not only is the designer
paying for the additional reference meter, there is an
additional cost for each meter run as a separate ball
valve must be included to permit diverting the gas
through the reference meter. Additionally, the extra
reference run requires more space on the skid that
adds to the cost. Also, using a reference meter on
location somewhat limits the ability to verify
performance over the entire range of operation.
Finally, removing a meter and having its
performance verified at a calibration facility provides
*+*+ 
This would most likely be required in the event of a
dispute by the purchaser of the gas.

CONCLUSIONS
During the past several years the industry has
learned a lot about USM operational issues. The
traditional 5 diagnostic features, gain, signal-tonoise, performance, path velocities and SOS have
helped the industry monitor the USM. These 5
features provide a lot of information about the

meter at the time of installation, and monitoring
these features on a routine basis can generally
identify metering problems in advance of failure.
More advanced diagnostic indicators, such as
References:
1. AGA Report No. 9, Measurement of Gas by
Multipath Ultrasonic Meters, June 1998.
2. Provisional
Verification

Specification for
and Installation

the Approval,
and use of

Turbulence, are paving the way to allow the meter to

become virtually maintenance-free. In the future it is
likely that a meter will have enough power and
intelligence to quickly identify potential measurement
problems on a real-time basis. As the industry learns
more about not only the USM, and the operation of
their own measurement system, the true value of the
ultrasonic meter will be recognized. The USM
industry is still relatively young and technology will
continue to provide more tools to he*
measurement problems.
Ultrasonic Gas Meters, Measurement Canada
PS-G-E-06, 1998-03-05 (Rev 1, 9-26-2002).
3. John Lansing, Basics of Ultrasonic Flow Meters,
AGMSC 2003, Coraopolis, PA.

4. John Lansing, Smart Monitoring and Diagnostics

for
Ultrasonic
Meters,
NSFMW
2000,
Gleneagles, Scotland.

19. Jim Micklos, Ultrasonic Gas Flow Meters for

Custody Transfer Measurement, ASGMT 2008,
Houston, TX

5. John Lansing, +RZ 7RGD\V 860 'LDJQRVWLFV

Solve Metering Problems, NSFMW 2005,
Tonsberg, Norway

20. Bird R.B., Stewart W.E., and Lightfoot E.N.,

Transport Phenomena, 1960, Wiley, New York

6. T. A. Grimley, Ultrasonic Meter Installation

Configuration
Testing,
AGA
Operations
Conference, 2000, Denver, CO.

22. Schlichting, H., Boundary Layer Theory, 6

edition, 1968, McGraw-Hill, New York

7. Larry Garner & Joel Clancy, Ultrasonic Meter

Performance / Flow Calibration Results /
CEESI Iowa / Inspection Tees vs. Elbows,
CEESI Ultrasonic Conference, June 2004, Estes
Park, CO
8. John Lansing, Features and Benefits of the
SICK
Maihak
USM,
CEESI
Ultrasonic
Conference, June 2006, Estes Park, CO
9. John Lansing, Advanced Ultrasonic Meter
Diagnostics, Western Gas Measurement Short
Course, 2007, Seattle, WA
10. John Lansing, Ultrasonic Meter Station Design
Considerations, Western Gas Measurement
Short Course, 2003, Victoria, BC, Canada
11. Klaus Zanker, Diagnostic Ability of the Daniel
Four-Path Ultrasonic Flow Meter, Southeast
Asia Flow Measurement Workshop, 2003, Kuala
Lumpur, Malaysia
12. John Lansing, Benefits of Flow Calibrating
Ultrasonic Meters, AGA Operations Conference,
2002, Chicago, IL
13. James N. Witte, Ultrasonic Gas Meters from
Flow Lab to Field: A Case Study, AGA
Operations Conference, 2002, Chicago, IL.
14. T. A. Grimley, Performance Testing of Ultrasonic
Flow Meters, AGA Operations Conference,
1997, Nashville, TN
15. Dr. Jim Hall, William Freund, Klaus Zanker &
Dale Goodson, Operation of Ultrasonic Flow
Meters at Conditions Different Than Their
Calibration, NSFMW 2002, St. Andrews,
Scotland.
16. John Lansing, Dirty vs. Clean Ultrasonic Flow
Meter
Performance,
AGA
Operations
Conference, 2002, Chicago, IL.
17. John Stuart, Rick Wilsack, Re-Calibration of a 3Year Old, Dirty, Ultrasonic Meter, AGA
Operations Conference, 2001, Dallas, TX
18. John Lansing, Using Software to Simplify USM
Maintenance, AGA Operations Conference,
2003, Kissimmee, FL

21. Rtten, O., Deutsches Patent No 520484, 1928

th

23. Steenbergen, W., Turbulent Pipe Flow with

Swirl, Ph.D Thesis, Eindhoven University,
Eindhoven, The Netherlands