Ultrasonic Meter Diagnostics

John Lansing
 
 .

ABSTRACT
This paper discusses both basic and advanced
diagnostic features of gas ultrasonic meters (USM),
         
identify problems that often may not have been
identified in the past. It primarily discusses fiscalquality, multi-path USMs and does not cover issues
that may be different with non-fiscal meters as they are
often single path designs. Although USMs basically
work the same, the diagnostics for each manufacturer
does vary. All brands provide basic features as
discussed in AGA 9 [Ref 1]. However, some provide
more advanced features that can be used to help
identify issues such as blocked flow conditioners and
gas compositional errors. This paper is based upon
the Westinghouse configuration (also knows as a
chordal design) and the information presented here
may or may not be applicable to other manufacturers.
INTRODUCTION
During the past several years there have been
numerous papers presented which discuss the basic
operation of USMs [Ref 2]. These papers discuss the
meaning of the five basic diagnostic features.
Following is a summary of the five features available
from all USM manufacturers.
x
x
x
x
x

Individual path velocities

Individual path speed of sound
Gains for each transducer
Signal to noise for each transducer
Accepted pulses, in percentage, for each
transducer pair

Although these features are very important, little has

been written on how to interpret them. Part of the
reason is analysis varies by manufacturer.
Some manufacturers provide additional diagnostic
features such as swirl angle, turbulence, AGA 10
[Ref  ! "#  $  !  $
others.
Graphs shown in this paper are from Excel
spreadsheets based on data generated by software
that is used to communicate with the meter. Note that
these graphs were not individually developed but
rather automatically generated from the data collected
during calibration or maintenance procedures.
Obviously it is important for users to collect periodic
$ %&#'%&"*-

+ &  $      $#

Many utilize some of the data for entry into their
company database for tracking over time. However, a
large numb &   &$   /% 
trending of data.
BASIC DESIGNS OF ULTRASONIC METERS
Before discussing diagnostics it might be helpful to
review some of the basic designs that are used today.
Figure 1 shows 5 types of velocity integration
techniques [Ref 4]. The various meter configurations
in Figure 1 provide different velocity responses to
profiles, and are thus analyzed differently. This is
particularly true when trying to perform comparisons
on velocity and SOS. Looking at differences in SOS
between the various paths may require somewhat
different analysis. This is primarily the case when a
meter is operated at very low velocities as thermal
stratification can occur (more on this later). Analysis in
this paper will be applicable to design D in Figure 1.
A

Figure 1 3 Ultrasonic Meter Designs

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.
        *
 /%+  6
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 meters measured
SOS.

Gain
! &  $   &  $  
   & % %   #  '
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 velocity.
Figure 2 shows gains from a 16-inch meter at the time
of calibration. These were taken when the meter was
operating at approximately 20 fps.
Average AGC at 22.7 ft/s
50

Average AGC (dB)

Transducer A

Transducer B

40
37
30

32

37

37

37

32

31

31

20

10
Path 1

Path 2

Path 3

Path 4

Path

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 ultimately be eliminated since they
 &     #     
$&$ &$<==>$%
less.
Figure 4 shows the performance of a 16 meter at a
velocity of about 20 fps.

Figure 2 3 Gain at 20 fps 3 16 inch Meter

Average AGC at 153.6 ft/s

50

Average AGC (dB)

Transducer A

40

41
35

41

40

Transducer B

Average Performance at 22.7 ft/s

110

Average Performance (%)

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.

100
100.0

100.0

Path 1

Path 2

100.0

100.0

Path 3

Path 4

90
80
70
60

40

36

35

35

30

50

Path

Figure 4 3 Transducer Performance at 20 fps

20

10
Path 1

Path 2

Path 3

Path 4

Path

Figure 3 3 Gain at 155 fps 3 16 inch Meter

Figure 3 shows the gains for all pairs have increased.
This is normal when a meter is operating at much

Figure 5 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.

Average Performance at 153.6 ft/s

Average Performance (%)

110
100
92.8

90

91.7

89.6

91.7

80
70
60

Figure 7 show 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.
Average SNR at 153.6 ft/s

50
Path 1

Path 2

Path 3

Path 4

50

Path

Figure 5 3 Transducer Performance at 155 fps

Signal to noise (SNR) provides information that is also

"  "&%  $    
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 indication of
 * /%+ #  '     
same ultrasonic frequency spectrum as that
transmitted from the transducer itself.
The measure of signal strength to the level of
* /%+ %CK
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
&  CK   &  &  $& 
methodology of 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.
Average SNR at 22.7 ft/s
50
Transducer A

Average SNR

40

40

40

40

40

40

40

Transducer B

40

40

30

20

10
Path 1

Path 2

Path 3

Path 4

Path

Figure 6 3 SNR at 20 fps Meter Velocity

Average SNR

Signal-to-Noise Ratio

Transducer A

40

Transducer B

35
30

36
30

32

31

31
27

26

20

10
Path 1

Path 2

Path

Path 3

Path 4

Figure 7 3 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
  &$  ""  *$+ 
signal to the point that the meter becomes inoperative.
' w generation of 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 8 shows a picture of a meter and a
control valve located immediately downstream of the
USM.

In the test shown in Figure 8, the meter was being

operated at 600 psig and the regulator was producing
about 200 psig differential pressure. '$CK
went from a normal of 40 dB to 24 dB. For this meter
when the SNR approaches 13 the meter would begin
to reject waveforms. Figure 9 shows the waveform
during this test. Figure 10 shows the same pair of
transducers when there is no regulator noise.

Figure 8 3 Control Valve near 2-inch USM

Baseline with
Regulator Noise

Figure 9 3 Waveform with Control Valve Noise

Baseline with no
Regulator Noise

Figure 10 3 Waveform with no Control Valve Noise

From Figure 9 we can see there is a little noise on the

baseline preceding the major waveform. The baseline
in front of the received signal is not perfectly flat as it is
in Figure 10. The SNR values are above 24 dB for this
condition on the upstream transducers (one that faces
the source of the noise). The downstream transducer
has a SNR of 30 because it is facing away from the
noise source. Figure 10 shows the waveform when
there is no noise from the regulator.
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.
Average SNR [dB]

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.
Average SNR [dB]

SNR AB

SNR BA

40
35
30
25
20
15

40

SNR AB

SNR BA

10

35
5
30
0
25

Path 1

Path 2

Path 3

Path 4

20

Figure 12 3 Good SNR on all Paths

15

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
$&$ XY>%
normal. Thus, it is possible to have low SNR and all
other diagnostic indicators are normal.

10
5
0
Path 1

Path 2

Path 3

Figure 11 3 Poor SNR on Path 4

Path 4

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
 $ !"%
SOS calculated by the meter. Figure 13 shows a
graph of the SOS of a 10 inch meter at the time of
calibration.
Speed of Sound at 22.7 ft/s
Path 1

Path 2

Path 3

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 15 shows the SOS values of the same 10 inch
meter when it is operated at 1.8 fps at the calibration
lab.

1380
Average

Figure 14 shows the SOS by path in percentage

&& "$!#Zach
&!"=#=\>#'
indicates good correlation between each path and also
no temperature stratification within the meter.

Path 4

1379
1378

0.3

1376

0.2

Path 1

1375

120

110

Time (sec)

Figure 13 3 SOS by Path at Calibration

Figure 14 shows the percent difference of each path

"$ reported average SOS.
SOS Difference from Average at 22.7 ft/s
0.3
Path 2

Path 3

Path 4

0.2

Path 4

0.0
-0.1

120

110

100

90

80

70

60

50

40

30

20

10

-0.3

Time (sec)

Figure 15 3 Thermal Stratification Effects

From Figure 15 it can be seen that 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 the gas is slightly warmer 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 15 shows
upper paths have increased and the ones at the
bottom decreased.

0.1

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 16 shows the results of this
10-inch at the time of calibration.

0.0
-0.1
-0.2

Time (sec)

Figure 14 3 Path Percent Difference in SOS

120

110

100

90

80

70

60

50

40

30

20

10

-0.3

SOS Difference (%)

Path 3

-0.2

This data was taken from the meter operating at 23 fps

and showing a very stable reading. Here we can see
 &  $ ! " re very close. But
perhaps an easier way of looking at the SOS values is
 $% !$"%
value. Doing this is makes it easier to spot problems.

Path 1

Path 2

0.1

100

90

80

70

60

50

40

30

20

10

1374

SOS Difference (%)

SOS

SOS Difference from Average at 1.8 ft/s

1377

10-inch As Found and As Left Results

Flowing Velocity Ratios at 22.7 ft/s

1.00
As Found

As Left

Verification Points

Path 1

0.75

0.917

0.50

Path 2

% Error

0.25

1.018

0.00

Path 3

-0.25

1.021

-0.50

Path 4

0.912

-0.75
-1.00

0.80
0

20

40

60

80

100

0.85

120

0.90

0.95

1.00

1.05

1.10

Average Flowing Velocity (ft/sec)

Meter Velocity (ft/sec)

Figure 16 3 10-inch As-Found Calibration Performance

Figure 17 3 Path Ratios at 23 fps

C*&+^=&"he
$   * &+  <#Y &#  '  
very little impact in the performance of the meter even
though there was some thermal stratification.

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
18   $ $ "  rofile when the
velocity is at 2.8 fps.

Velocity Profile
Monitoring the velocity profile is possibly one of the
most overlooked and under-used diagnostic tools of
 $# "$ 
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 in the
ev$&$ _at a later
date.
Figure 17 shows the velocity ratio of each path relative
$average velocity. This ratio is computed
by taking each path"% 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
$"  %$
operation are easier to detect than by looking at the
actual velocity on each path.

Flowing Velocity Ratios at 2.8 ft/s

Path 1

0.842

Path 2

0.997

Path 3

1.053

Path 4

0.80

0.961

0.85

0.90

0.95

1.00

1.05

1.10

Average Flowing Velocity (ft/sec)

Figure 18 3 Path Ratios at 2.8 fps

When comparing Figure 17 and 18 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, the
$ &$    $ # {% <9
$*&+ from the calibration.

As Found Results for 16-inch Meter

Flowing Velocity Ratios at 1.8 ft/s

1.00
As Found
0.75

Path 1

0.930

% Error

0.50
0.25

1.037

Path 2

0.00
-0.25

Path 3

1.005

-0.50
-0.75

Path 4

-1.00
0

10

20

30

40

50

60

70

0.894

80

Pipeline Velocity (ft/sec)

0.80

0.85

0.90

0.95

1.00

1.05

1.10

Average Flowing Velocity (ft/sec)

Figure 19 3 16 inch As-Found Results

Figure 19 shows  $    %
even though path ratios were different as shown in
Figures 17 and 18. 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.
Figures 20 and 21 show velocity profiles for the 10inch meter discussed earlier at 20 and 1.8 fps
respectively. Figure 20 shows the baseline at about
23 fps and the velocity profile is very symmetrical. In
Figure 21 the profile has become a little distorted. This
is in part due to the minor thermal stratification.
However, as figure 16 shows, there was very little
$ $-found linearity (open circles).
Flowing Velocity Ratios at 22.7 ft/s
Path 1

1.022

Path 3

1.021

0.912

Path 4

0.80

0.85

0.90

0.95

1.00

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 26.
ADVANCED DIAGNOSTIC INDICATORS
The basic diagnostic parameters that are provided by
 | "      "
section. They are gain, performance, signal to noise
ratio (SNR), speed of sound (SOS) and velocity profile.
Of these the most difficult for most to understand is the
Velocity Profile. This is due in part to the various USM
path configurations and different methods of
presenting path velocity information by the
manufacturers.
For the chordal type of meter (see path layout D in
Figure 1), most manufacturers talk about path velocity
ratios. This method of displaying the velocity profile is
       
change significantly over the majority of velocities the
USM is operated at today.

0.908

Path 2

Figure 21 3 10-inch 1.8 fps Path Ratios

1.05

Average Flowing Velocity (ft/sec)

Figure 20 3 10-inch 20 fps Path Ratios

1.10

However, there are easier ways to analyze the various

different profiles that can occur in the field. These
variations occur due to the wide range up upstream
% #  ' &  
minimize the distortion of the gas velocity profile
"%   $    eliminate them.
Even though the USM can handle a wide range of
distortion with minimal impact on accuracy, the ideal
situation is to reduce the * && +
(distortion of the gas velocity profile) to a minimum.
In order to understand if the velocity profile has
changed over time, advanced methods of diagnosing
the USM have been developed. Since the velocity
profile is the most difficult, due to the wide range of
possible scenarios, a simpler method of summarizing
these profiles would be beneficial.

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 bottom, and is called Symmetry.
C$  $  "    "
symmetrical resulting in a value close to 1.000.
Figures 13 and 14 show both the Profile and the
Symmetry in a single graph.

Profile Factor Std. Dev. = 0.0221

1.30

1.15
Profile Factor
Symmetry

1.10

130

120

110

0.85
90

1.00
100

0.90

80

1.05

70

0.95

60

1.10

50

1.00

40

1.15

30

1.05

20

1.20

Symmetry
[(P1+P2)/(P3+P4)]

1.25

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.

Profile Factor and Symmetry at 2.8 ft/s

10

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. One of these methods is called Profile
Factor.

  $ & and this is to be expected at

lower gas velocities.

Profile Factor

Profile Factor and Symmetry

Time (sec)

Figure 23 3 Profile Factor and Symmetry at 3 fps

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

In Figure 22, when the meter was flowing at 23 fps, the

Profile Factor was 1.115 (average of the magenta
colored line). As the velocity dropped to 2.8 fps
(Figure 23) the Profile Factor increased to 1.137. This
is about a 2% change in profile when compared to the
Profile Factor at 23 fps.

Figure 24 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
 /% &  &     $
accuracy. This meter was used to show what happens
not only to the profile, but to quantify the change in
accuracy.

Profile Factor and Symmetry at 22.7 ft/s

Flowing Velocity Ratios at 66.4 ft/s

Profile Factor Std. Dev. = 0.0111

1.15

1.30
Profile Factor
Symmetry

Symmetry
[(P1+P2)/(P3+P4)]

130

120

110

90

0.85
100

1.00
80

0.90

70

1.05

60

0.95

50

1.10

40

1.00

30

1.15

20

1.05

0.919

1.10

1.20

10

Profile Factor

1.25

Path 1

Time (sec)

Path 2

1.020

Path 3

1.017

Path 4

0.915

0.80

0.85

0.90

0.95

1.00

1.05

1.10

Average Flowing Velocity (ft/sec)

Figure 22 3 Profile Factor and Symmetry at 20 fps

Figure 24 312-Inch Meter Profile 3 Normal

The other diagnostic worth reviewing is the Symmetry

value. Figure 23 shows a significant change (on the
order of 10%) in the Symmetry at the lower velocities.
This can be seen by comparing the blue line if Figure
22 to the blue line in Figure 23. However, there was no
significant impact on meter performance (see
Figure 16). These graphs indicate there was a change

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 25
shows the flow conditioner just before it was installed
in the pipeline.

Profile Factor and Symmetry at 66.4 ft/s

Profile Factor Std. Dev. = 0.0086

1.30

1.15
Profile Factor
Symmetry

150

140

130

120

110

90

100

0.85

80

1.00

70

0.90

60

1.05

50

0.95

40

1.10

30

1.00

20

1.15

1.05

Symmetry
[(P1+P2)/(P3+P4)]

1.10

1.20

10

Profile Factor

1.25

Time (sec)

Figure 27 3 Profile Factor and Symmetry at 66 fps

Figure 25 3 40% Blocked Flow Conditioner
Figure 26 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.

From Figure 27 the average Profile Factor is 1.111

and the average Symmetry is 1.003. These are just
about the ideal values for both. Figure 28 shows the
Profile Factor and Symmetry graph with 40%
blockage.
Profile Factor and Symmetry at 67.2 ft/s
Profile Factor Std. Dev. = 0.025

Path Velocity Ratios at 66.5 ft/s

1.30

1.15

Profile Factor

Profile Factor

0.885

1.000

Path 2

1.039

Path 3

1.25

1.10

1.20

1.05

1.15

1.00

1.10

0.95

1.05

0.90

1.00

0.85

0.90

0.95

1.00

1.05

1.10

150

140

130

120

110

100

90

80

70

60

50

40

30

20

0.85

10

0.945

Path 4

0.80

Symmetry
[(P1+P2)/(P3+P4)]

Symmetry

Path 1

Time (sec)

Path Ratios

Figure 26 3 12-inch Meter Profile 3 40% Blocked

The Profile is obviously distorted with higher-thannormal 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 than at the top.
Figure 27 shows the graphical results of the Profile
Factor and Symmetry with no blockage.

Figure 28 3 Profile Factor and Symmetry at 66 fps 3

40% Blocked Flow Conditioner
Figure 28 shows the results of the Profile Factor and
Symmetry with 40% blockage of the flow conditioner.
The average of the Profile Factor is 1.053 and the
Symmetry is 0.936. This is about a -6% change in
Profile Factor and about a -7% change in Symmetry.
Both of these would be considered significant and
should be treated a cause for investigation.
After installation in the field a meter typically will
generate a Profile Factor that is repeatable to 0.02
(or about 2%). 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 29 shows
the result of the three test velocities and the impact on
metering accuracy.

Baseline vs. 40%

Blocked CPA

Flowing Velocity Ratios at 44.3 ft/s

Velocity
(fps)

% Diff. with
Blocked CPA

Path 1

68.4
45.3
22.9

-0.02
-0.12
-0.10

Path 2

0.885

1.000

1.039

Path 3

Figure 29 3 Blocked CPA Results

0.945

Path 4

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
Profile Factor alone would be a good indication if there
was any contamination or flow conditioner blockage.
This may not always be true. Figure 30 shows a
picture of a flow conditioner with 3 holes blocked at the
bottom.

0.80

0.85

0.90

0.95

1.00

1.05

1.10

Path Ratios

Figure 31 3 Path Ratios with 3 Holes Blocked - Bottom

Comparing Figure 31 with Figure 24 (the normal Path
Ratio profile) it is obvious that the two graphs of path
ratios do not look the same. However, when
computing the Profile Factor, the average value for
Figure 31 is 1.114 (1.11 is the ideal factor). This is
almost the perfect number for this meter.
Upon further investigation Figure 32 shows that the
Symmetry has changed significantly.
Profile Factor and Symmetry at 44.3 ft/s
Profile Factor Std. Dev. = 0.0156

1.30

1.15
Profile Factor
Symmetry

150

140

130

120

110

90

100

0.85
80

1.00
70

0.90

60

1.05

50

0.95

40

1.10

30

1.00

20

1.15

1.05

Symmetry
[(P1+P2)/(P3+P4)]

1.10

1.20

10

Profile Factor

1.25

Time (sec)

Figure 30 3 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 24 shows a
graph of the Path Ratios during this test with the 3
blocked holes located at the bottom of the meter
piping.

Figure 32 3 Profile Factor and Symmetry 3 3 Holes

Blocked (at Bottom of Pipe)
In Figure 32 the average for the Profile Factor is 1.114,
and is almost normal, but that the Symmetry value
average is about 0.95, or approximately a -5% shift
from normal (1.00 being normal). Thus, it is possible
for the Profile Factor to be normal even though the
velocity profile in the meter is not. This is the reason
a combination of Profile Factor and Symmetry is
both require to fully analyze the velocity profile
entering the
Figure 33    &&    $
accuracy was with this blockage was located at the
bottom and also rotated 90 degrees so that it was
blocking paths 2 & 3 more significantly (than when it
was at the bottom of the pipe.

12-inch, 4-Path Meter - 3 Holes Blocked Results

Profile Factor and Symmetry at 44.6 ft/s

1.00
Un-Blocked CPA

3 Holes Blocked Results

Profile Factor Std. Dev. = 0.0129

3 Holes Blocked - Rotated 90 Degrees

Profile Factor
Symmetry 2

1.25

1.10

Profile Factor

0.25
0.00
-0.25
-0.50

Average: -0.024%

1.20

1.05

1.15

1.00

1.10

0.95

1.05

0.90

1.00

0.85

-0.75
-1.00

When the blockage was at the bottom of the meter

run, there was very little impact on accuracy (on
average -0.024%). When the blockage was rotated 90
degrees to the side, the meter responded with a shift
of about +0.22%. All other blockage tests to date had
shown the meter responded with a negative shift in
error, but for the first time the meter now measured
fast with this blockage.
The question might be asked as to why this has
   /   " ity profile when this
blockage occurred on the side of the flow conditioner.
Flowing Velocity Ratios at 44.6 ft/s
0.968

Path 1

0.989

Path 2

0.994

Path 3

0.987

Path 4

0.80

0.85

0.90

0.95

1.00

1.05

1.10

Average Flowing Velocity (ft/sec)

Figure 34 3 Path Ratios with 3 Holes Blocked - Side

Figure 34 shows that the velocity profile is much flatter
than normal. That is the center two paths (paths 2 & 3)
are reading much less than normal and are almost the
same as the outer paths (2 & 3). This makes sense
since the blockage was in direct line with the middle
two paths when it was rotated 90 degrees vs. blocking
primarily the bottom path when at the bottom.
When analyzing the Profile Factor and Symmetry a
rather interesting result is apparent. Figure 35 shows a
graph of the Profile Factor and Symmetry for the 3hole blockage on the side.

150

140

130

120

110

Figure 33 3 As-Found Results 3 3-Holes Blocked

90

Meter Velocity (ft/sec)

100

80

80

70

70

60

60

50

50

40

40

30

30

20

20

10

10

Symmetry
[(P1+P2)/(P3+P4)]

Average: +0.22%

0.50

% Error

1.15

1.30

0.75

Time (sec)

Figure 35 3 Profile Factor and Symmetry 3 3-Holes

Blocked (at Bottom and Rotated 90 Degrees)
Figure 35 shows that the average Symmetry is almost
normal (0.985, or about 1.5% change from ideal), but
the average Profile Factor is 1.011, or about a 10%
change. If only Profile Factor alone were monitored, a
shift of 1.5% would not be considered significant. Thus
a problem might be over-looked.
This is why both the Profile Factor and Symmetry need
to be computed, and monitored, in order to state the
$&$ %. If both
are the same, then there is no combination of velocity
profile ratios that can produce a distorted profile and
still provide the same average values of 1.11 for
Profile Factor and 1.00 for Symmetry.
Sometimes, reviewing path ratios as shown by the
USM software, it is difficult to see if everything is
normal. By reviewing a Maintenance Report, where
the average value is generally presented, it is easier to
see. Figures like 26, 28, 31 and 34 show the path
  &|&#
significant changes occur, it is easy to see. However,
what is needed is a simpler way to present this. The
following graph is from the USM software and it shows
all of the typical diagnostics.

Figure 36 3 Summary of All Diagnostics 3 Normal Velocity Profile

Figure 37 3 Summary of All Diagnostics 3 Abnormal Velocity Profile

In Figure 36 we see a very symmetrical velocity profile

(upper left corner). In Figure 37 we see a distorted
velocity profile. The Profile Factor and Symmetry are
shown in the graph called Profile Indication which is
rd
the 3 graph from the left on the second row. In Figure
36 we can see there are two dots inside the red box
and in Figure 37 we see that one of the dots is outside
the red box and the line is yellow. This is telling us the
Symmetry is outside of normal tolerances.
The Profile Indication box is really a summary of both
the Profile Factor and Symmetry. The dot in the middle
of the box is established when commissioning the
meter and is the average value when first started up.
The second dot is the current, or live, value of the
Profile Factor value (represented by the X axis) and
the Symmetry value (represented by the Y axis). When
either of these values causes a change of more than
5% ( an adjustable value but just a baseline for this
example), then the line turns yellow indicating there is

a significant change. Figure 38 shows a close-up of a

normal value, and Figure 39 show a problem.

Figure 38 3 Normal Profile Factor and Symmetry

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 24, now appears
like that in Figure 26, 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 40 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. The average for these is
2.44% and this is considered normal.
Turbulence at 66.4 ft/s

Figure 39 3 Abnormal Profile Factor and Symmetry

100
80

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.

8
6
4

1
0

10

20 30

40 50 60

70 80

90 100 110 120 130 140 150

Time (sec)

Figure 40 3 Normal 12-inch Meter Turbulence

Figure 41 show the Turbulence with a 40% blocked
flow conditioner as shown in Figure 25.
Turbulence at 66.5 ft/s
100
80

Path 1

60

Path 2

Path 3

Path 4

40

20

10

8
6
4

150

140

130

120

110

90

100

80

70

60

50

40

30

20

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.

Path 4

10

During the past several years an additional diagnostic

feature has been studied by Engineering.
This
& *' + ghly in
a previous paper [Ref 6]. Essentially Turbulence is a
measure of the variability of each paths velocity
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.

Path 3

20

10

Turbulence

Path 2

40

Turbulence (%)

When a problem occurs that causes a shift in either

Profile Factor or Symmetry, by more than the
programmed limit in the meter (shown here as 5%),
then the dot will move outside of the box and turn
yellow. Such is the case for Figure 39 where we see
the Symmetry is 0.895, or more than 10% from normal
(1.00 being normal). Thus the technician can very
easily see there is a problem with the profile.

Path 1

60

Turbulence (%)

Figure 38 shows a dot just to the right of the centered

one (which is configured in the meter) and it
represents the current reading of both Profile Factor
and Symmetry. These values are show to the right of
the graph.

Time (sec)

Figure 41 3 High 12-inch Meter Turbulence

It is clear that the turbulence in Figure 41 is about 3
times higher, or an average of 7.03%. Certainly the
velocity profiles for this meter, shown in Figures 24
and 26, 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, or
piece of carpet, 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.
The following figures show the Turbulence with the 3
holes blocked. The first is with the holes located at the
bottom of the piping, and the second with the blocked
holes rotated 90 degrees.
Turbulence at 44.3 ft/s
100
80

Path 1

Path 2

Path 3

Path 4

60

Turbulence (%)

40

20

CONCLUSIONS
During the past several years the industry has learned
a lot about USM operational issues. The traditional 5
diagnostic
features,
gain,
signal-to-noise,
performance, path velocities and SOS have helped the
industry monitor the USM. These 5 features provide a
lot of information about the meter#%
initial baseline on 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
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
"$ment problems.

10
8
6
4

1
0

10

20

30

40

50

60

70

80

90

100 110 120 130 140 150

REFERENCES

Time (sec)

Figure 42 3 Turbulence w/3 Holes Blocked - Bottom

Figure 42 shows the average Turbulence is higher
than normal (the average is 4.02%)#'%
as the 40% blocked, but it is significant. Figure 41
shows the Turbulence when the blocked holes were
rotated 90 degrees.
Turbulence at 44.6 ft/s
100
80

Path 1

Path 2

Path 3

Path 4

60

Turbulence (%)

40

1. AGA Report No. 9, Measurement of Gas by

Multipath Ultrasonic Meters, June 1998, American
Gas Association, 1515 Wilson Boulevard,
Arlington, VA 22209
2. John Lansing, Basics of Ultrasonic Flow Meters,
American
School
of
Gas
Measurement
Technology, 2000, Houston, Texas
3. AGA Report No 10, Speed of Sound in Natural
Gas and Other Related Hydrocarbon Gases, July
2002, American Gas Association, 1515 Wilson
Boulevard, Arlington, VA 22209
4. BSI 7965:2000, Guide to the Selection,
Installation, Operation & Calibration of Transit
Time Ultrasonic Flowmeters for Industrial Gas
Applications

20

10
8
6
4

1
0

10

20

30

40

50

60

70

80

90

100 110 120 130 140 150

Time (sec)

Figure 43 3 Path Ratios w/3 Holes Blocked - Side

Figure 43 shows the average Turbulence is 4.33% and
also higher than the average with no blockage of
2.44%. Both tests with 3 holes blocked indicate about
the same and approximately 60% higher than normal.
Thus Turbulence is generally the best method of
identifying flow conditioner blockage.

5. Klaus Zanker, Diagnostic Ability of the Daniel

Four-Path Ultrasonic Flow Meter, Southeast Asia
Flow Measurement Workshop, 2003, Kuala
Lumpur, Malaysia
6. John Lansing, +RZ 7RGD\V 860 GLDJQRVWLFV
Solve Metering Problems, North Sea Flow
Measurement Workshop, 2005, Tonsberg, Norway