ULTRASONIC METER CONDITION BASED MONITORING – A FULLY

AUTOMATED SOLUTION

George Kneisley, Transwestern Pipeline, Roswell California, USA

John Lansing, SICK Inc., Houston, Texas USA

Toralf Dietz, SICK AG, Germany

1. INTRODUCTION
The customer requires a fiscal meter that measures with highest reliability within the required
accuracy limits throughout the life time. Whenever this requirement isn’t fulfilled due to changed
process/flow conditions or changes to the meter, the user needs to be warned in real-time. To
ensure such warning, the diagnosis parameters implemented into modern ultrasonic flow meter
can be useful. Since the introduction of the global diagnosis concept major improvements in
diagnosing a USM have been achieved. This requires a thorough understanding of the meter’s
operation and also understanding what normal, and non-normal responses of all diagnostic
parameters are in order to insure proper operation. The automated diagnostics will monitor, and
alarm, on all important parameters such as Profile Factor, Symmetry, SNR, Turbulence, etc.
These warnings are today an important factor for driving the condition based maintenance of the
installation. Additionally, it is very important to have a long-term history of these diagnostics in
order to properly determine if a meter is still operating accurately.

Beside this features inherent to every ultrasonic flow meter with a multiple number of paths
additional concepts to compare measurements directly exist. Two main concepts can be realised -
permanent serial metering with two independent fiscal meters or with a combination of a fiscal
and a check meter, introduced by TransCanada Pipeline (TCPL) several years ago. This concept
involves using a single path USM downstream of the fiscal multipath meter. Papers have shown
that single path USM meters are significantly affected by abnormal measurement conditions such
as flow conditioner blockage, pipeline contamination from oil and mill scale, and any other
change in operation that impacts accuracy. Since the single path meter has significantly more
sensitivity, comparing the uncorrected readings of both meters provides a simple solution for
determining if the fiscal meter is still operating accurately. If both are in agreement, then
measurement must be OK. Should the two meters deviate, then more than likely there is some
condition which might impact the accuracy of the fiscal meter. This paper will discuss the results
and benefits in terms of reliability and economic impact of the TCPL-method installed in various
field applications. Data will be presented on dirty vs. clean meters to show that the single path
meter shifts significantly in a dirty environment while the 4-path custody meter is relatively
insensitive to this.
2. AUTOMATED METER DIAGNOSTICS
A fully automated USM requires having all aspects of the meter’s diagnostics monitored on a
real-time basis within the meter. Traditionally users have collected periodic files and performed
manual analysis of these to determine if the meter is operating correctly. That is no longer needed
with today’s technology.
All USMs have a variety of diagnostics to help the user determine if the meter is operating
correctly. The basics for a typical chordal meter include the following:
• Gain for each transducer
• Performance (percent accepted signals by path)
• Signal to Noise Ratio (SNR)
• SOS for each path
• Velocity profile (path ratios or velocities)
Today these 5 are enhanced by additional diagnostics for the user to better understand if the
meter is operating correctly. These include the following:
• Profile Factor
• Symmetry
• Turbulence
Profile factor and Symmetry are methods of analyzing the velocity profile (or path ratios). They
are values which reflect the shape and amount of distortion in the basic velocity profile. They
have been used for years to help understand the velocity profile which is often considered by
many to be the most difficult diagnostic to understand [Ref 2 & 12]. These two are defined as
follows:
• Profile Factor = (Path 2 + Path 3) / (Path 1 + Path 4)
• Symmetry = (Path 1 + Path 2) / (Path 3 + Path 4)
Many consider a meter’s velocity profile the most difficult to understand because it can vary due
to installation and type of upstream piping components, and to some degree, flow rate. Generally
speaking these two diagnostics (Profile Factor and Symmetry) are very stable above a velocity of
approximately 2-3 m/s.
However, this is only true if the piping design incorporates a flow conditioner. Without a flow
conditioner, the velocity profile will have a wide range of both Profile Factor and Symmetry, and
thus very difficult, if not impractical, to monitor and alarm on these value.
Figure 1 is an example of a profile with a flow conditioner (CPA type plate) and thus has an idea
profile. It was collected at the calibration facility and had a flow conditioner upstream (at 10D)
along with perhaps 50D of straight piping. Figure 2 shows a velocity profile of the same meter
without a flow conditioner when it is subjected to piping that had 3 elbows and a tee upstream.
 Figure 1: Path Ratios with CPA Conditioner Figure 2: Path Ratios w/o CPA Conditioner

Clearly the profile in Figure 2 is significantly distorted. This amount of profile change will
impact the accuracy of the meter. The intent of velocity profile analysis is to identify if there are
any changes from the baseline. If Figure 1 were the baseline profile in the field, and a future
inspection revealed Figure 2, then obviously there is something wrong internally and would
warrant inspection.
Turbulence is a third advanced diagnostic, and is a measure of flow stability, or variability, of
each path’s velocity reading. It is presented in percentage by path. Values are computed from the
variability of all of the transit-time readings (per path) and updated once per second. More
information is available from previously published papers [Ref 2, 11 & 12].
Typically a chordal meter will have a turbulence value of 2-3% for Paths 2 & 3, and 3-4% for
Paths 1 & 4. The Turbulence values are higher for the outer paths (Paths 1 & 4) as they are
located closer to the pipe wall and thus there is some influence due to the surface of the piping.
Even more advanced diagnostics are provided by some meters and include the following:
• High meter velocity (exceeding programmed limit)
• Power supply voltage too low (below a programmed value)
• Logbook(s) warning when full of un-acknowledged entries
These previously identified advanced diagnostics, and more, can now be fully automated in the
meter by having site-specific programmed values for each diagnostic parameter. There are many
benefits to having these programmed in the meter rather than in the User’s software or in a flow
computer. Knowing within minutes that a problem may exist significantly reduces measurement
uncertainty. Some of the many benefits include:
• The meter monitors all parameters/diagnostics on a real-time basis. When a diagnostic is
approaching a limit, the meter then presents an alarm that can be remotely monitored.
• If diagnostic limits established in meter’s software (User interface), the only time the user
will know there is a problem is when they are connected to the meter. As a result, the
meter can operate for weeks with a problem before a technician might identify it.
• Each location, and meter size, will have different limits due to the station design, meter
size, pressure, flow rates, etc. By having these programmed in the meter, they can be
tailored to the site-specific conditions and thus present an alarm more quickly and
accurately.
 • One additional benefit of having the meter do this, rather than the flow computer or RTU,
is that no additional programming is required. Simply monitor a digital output (DO) and
the RTU can then report if there is a problem with a diagnostic. This saves the user time
and money and makes implementing this feature very easy.
To proof the feasibility of the concept to practical installations a special test was performed at
CEESI facilities.
All testing was performed with a significant length of straight piping upstream of the metering
package. This upstream length of straight pipe would present a very symmetrical and non-
swirling profile to the CPA 50E flow conditioner.
Figure 3 shows the 12 inch meter installed at the CEESI facility for the testing.

Figure 3: 12-inch 4+1 CBM Meter CEESI

A CPA was installed 10D upstream for baseline testing. After the baseline testing was complete
the 12-inch CPA was partially covered with duct tape (40% blockage). Figure 4 shows the 40%
blocked flow conditioner prior to testing.

Figure 4: 12-inch CPA with 40% Blockage

Duct tape was used to block the holes as this provides a repeatable method of testing and can
withstand the pressures created by the flow rates used.

The example shows the velocity profile changed from symmetrical to somewhat distorted
towards the bottom of the meter (compare Figure 5 to Figure 6).

Flowing Velocity Ratios at 66.4 ft/s Path Velocity Ratios at 66.5 ft/s

Path 1 0.919
Path 1 0.885

Path 2 1.020
Path 2 1.000

Path 3 1.017
Path 3 1.039

Path 4 0.915
Path 4 0.945

0.80 0.85 0.90 0.95 1.00 1.05 1.10

0.80 0.85 0.90 0.95 1.00 1.05 1.10
Path Ratios
Path Ratios

Figure 5: Path Ratios without Blocked CPA Figure 6: Path Ratios 40% Blocked CPA

Figure 5 shows the velocity profile in normal conditions with no blockage. The Profile Factor
was 1.111 which is normal for this meter design, and the Symmetry was 1.004, or within 0.4% of
a perfectly symmetrical profile. Meter data was collected at three velocities of approximately 7,
14, and 21 m/s. The data here is from the 21 m/s (64 fps) flow rate, but all three velocities
essentially looked the same.
Figure 6 shows the distortion that occurred as a result of the blockage. The Profile Factor
changed to 1.114, or only about 0.4% shift. This amount of change is relatively small compared
to the baseline Profile Factor of 1.111, and if the technician were only monitoring Profile Factor,
they would assume everything is OK. However, it is apparent these two graphs do not look the
same.
This is the reason it is very important to also monitor the second diagnostic called Symmetry. For
the 40% blocked condition the Symmetry is 0.950. This represents about a 5.5% shift in the
velocity profile towards the bottom of the meter (high velocity average). This magnitude of
change is very obvious and easily identified. By monitoring within the meter on a real-time basis,
it can be reported to the RTU or flow computer within minutes as an abnormal profile.
One can argue that 40% blockage is very significant and not likely to occur often. Let’s look at
the profile with only one hole blocked. Figure 7 is a picture of the CPA prior to the testing.
 Figure 7: 12-inch CPA with 1 Hole Blocked
The single blocked hole was located on the bottom of the meter. This location was chosen as the
most likely place for such a blockage to occur. Figure 7 shows the profile with the one hole
blocked at the bottom.

Flowing Velocity Ratios at 66.8 ft/s

Path 1 0.898

Path 2 0.992

Path 3 1.037

Path 4 0.954

0.80 0.85 0.90 0.95 1.00 1.05 1.10

Path Ratios

Figure 8: Path Ratios with One CPA Hole Blocked

Figure 8 shows a similar distortion as with 40% blockage. For this condition the Profile Factor
was 1.096 and the Symmetry was 0.949. In this case the Profile Factor changed about 1.5% and
the Symmetry almost 6%. Thus it is fairly easy to see that these two profiles (blocked vs. not
blocked) create a flow profile (Path Ratios) that can be seen visually, and thus can be alarmed on
via the meter’s firmware.
A typical alarm limit on Profile and Symmetry, based upon these test conditions, and other
testing, would suggest something between 3% and 5% tolerance for both the Profile Factor and
Symmetry values. Thus, when this meter was installed in the field, it would have a baseline
factor of perhaps 1.11 for Profile Factor, and 1.005 for Symmetry, and the Warning alarm limits
would be established at 5% for both. Thus if either the Profile Factor or Symmetry values were
outside of the 5% limit, the meter would then report a Warning alarm.
Another very valuable tool for analyzing flow conditioner blockage is monitoring Turbulence. In
the case of the baseline condition, the Turbulence values are typically around 2-3% for the
middle paths (Paths 2 & 3) and 3-4% for the outer paths (Paths 1 & 4). These numbers are based
upon many tests in the calibration lab and results from field installations.
Figure 9 shows the baseline results at 20 m/s with no CPA flow conditioner blockage, and Figure
10 with 40% blockage of the flow conditioner.
Turbulence at 66.4 ft/s Turbulence at 67.2 ft/s
100
80 100
Path 1 Path 2 Path 3 Path 4 80
60 Path 1 Path 2 Path 3 Path 4
60
40
40
Turbulence (%)

Turbulence (%)
20 20

10 10
8 8
6 6

4 4

2 2

1 1
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150

Time (sec) Time (sec)

Figure 9: Turbulence without Blocked CPA Figure 10: Turbulence 40% Blocked CPA

Figure 9 shows turbulence values that average around 2.8% for the outer paths and around 1.9%
for the inner paths. Figure 10 has average values in the range of 7-10% for the outer paths and 5-
6% for the inner paths. Clearly there is a significant difference in the baseline (Figure 9) and the
blocked condition (Figure 10). Besides blockage, Turbulence can also identify a cyclic flow
control valve or significant pulsation. In these cases Turbulence values can exceed 50%
depending on the frequency and amplitude of the pulsation or flow control valve cyclic
operation.
Based upon testing in the lab and field results, a setting of perhaps 5 or 6% for the field appears
to be a viable limit for identifying problems. In the case of the single blocked flow condition,
Figure 11 shows what the Turbulence would be under this condition.

Turbulence at 66.8 ft/s

100
80
Path 1 Path 2 Path 3 Path 4
60

40
Turbulence (%)

20

10
8
6

1
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
Time (sec)

Figure 11: Turbulence with 1 Hole Blocked CPA

Figure 11 shows the Turbulence doesn’t look much different than Figure 9. Analysis of the data
shows the average for Paths 1 & 4 was about 3.1% and for Paths 2 & 3 the average was about
2.4%. In this example if the technicians were only looking at Turbulence as an indicator for
blockage, it probably would have been over-looked since averages were not that significantly
different. This example shows why it is important to monitor all aspects of the diagnostics.
When a technician connects to the meter using the factory software, it simply polls the meter for
all the diagnostics and presents this in a graphical format. Each manufacturer has developed
software in order to display the diagnostics for the user. In this case the software not only
presents all the diagnostics, but also shows the alarm limits that have been established in the
meter. Figure 12 shows a screen capture of a meter with all diagnostics within normal limits.

Figure 12: Diagnostics for a Meter with No Problems

Figure 12 shows all the diagnostics the user needs to monitor. The Velocity Profile is the graph
that starts at the top left, and from there the graphs, from left to right on the top row, include SOS
difference by path (in percentage), Performance by path (in percentage), AGC (Gains in dB) by
transducer, and on the second row of graphs there is SNR (in dB), Turbulence (in percentage)
and a Profile Indication Graph (more on that later).
The red dotted lines indicate the alarm limits that are programmed in the meter. Once any
diagnostic exceeds (or in the case of Performance and SNR go below) the programmed limit, the
individual graph turns yellow and the meter then reports via Modbus, and a digital output (DO),
that a Warning alarm is activated.
Figure 13 shows an example of a condition where there may be contamination of the Path 4
transducer pair.
 Figure 13: Possible Blockage or Contamination in Front of Path 4
Figure 13 shows the SOS for Path 4 is exceeding the red line (Warning alarm limit), and also the
AGC gains for Path 4 transducers are above the Warning alarm limit (Chart shows warnings in
yellow). In this condition the meter is now reporting a problem, and when the technician
connects to the meter, it is very apparent there is a problem. For this example there may only be a
slight impact on accuracy, but if left un-attended it might mean that Path 4 could eventually fail.
Figure 14 shows a different condition which may be caused by a blocked flow conditioner.

Figure 14: Possible Blocked Flow Conditioner

In Figure 14 Turbulence in the graph is above the alarm point and it is reported as a yellow bar
for Path 4. This graph was developed with the help of a demo model and thus only one path (Path
4) is reporting high turbulence due to the blockage being very close to the meter. In the real
world more than one path would probably be reporting an alarm (many bar graphs would be
yellow). The velocity profile is also distorted as can be seen in the Profile Indication.
Understanding the Path Ratios has perhaps been the most difficult diagnostic to understand.
Previously in this paper two advanced diagnostics were explained. One is Profile Factor and the
second is Symmetry. These two methods of analyzing the Path Ratios greatly simplify
understanding if the meter’s profile is normal. In the example of Figure 14 we can see the line in
the Profile Indication is yellow. The following two graphs show a larger version of a normal
Profile Factor and Symmetry (Figure 15), and an abnormal Profile Factor and Symmetry in
Figure 16 when there is blockage in front of the flow conditioner.

Figure 15: Normal Profile – No Warning Figure 16: Profile Distorted – Warning Reported
For each meter installed in the field the Profile Factor and associated Warning alarm limits can
be programmed for the site-specific conditions. For the example in Figure 15 the normal Profile
factor is 1.11 as shown on the X axis, and 1.00 for Symmetry on the Y axis. This is represented
by the dot in the middle of the red box. This dot stays in this location to indicate what the normal
values are. The size of the box is also programmed in the meter, and in this case a 5% tolerance
(plus and minus for both Profile Factor and Symmetry) from the baseline has been determined as
the limits. Thus the size of the box is ±5% from the baseline values.
The dot at the end of the line (towards the top of the red box in Figure 16) represents the current
average for the Profile Factor and Symmetry. The current value of Profile Factor is 1.133 and the
Symmetry is 1.038 as shown in the dialog boxes to the right of the graph. The dot is near the red
box but inside of it so the meter is not reporting any Warning alarms.
In Figure 16 the dot at the end of the line is outside of the box. This causes the line to turn yellow
and indicates a Warning alarm, and thus is reported via Modbus and the DO. In this example the
Profile Factor is 1.216 or about 10% from normal, and the Symmetry is 1.08 or about 8% from
normal. The velocity profile is perhaps the most important, and most difficult, diagnostic to
understand. This combined graphical representation of Profile Factor and Symmetry (Profile
Indication) greatly simplifies understanding the velocity profile and makes it very easy to
identify if the meter is operating within normal limits or not.
3. DIAGNOSIS ON READINGS OF TWO DIFFERENT METERS CBM 2PLEX
4+1 DESIGN
The first part of the CBM 2Plex 4+1 meter design is a conventional fiscal 4-path chordal
ultrasonic meter. The meter incorporates an additional, independent single-path meter and
associated electronics incorporated into the same body. The purpose of the additional path is for
continuous comparison of volumes to the fiscal 4-path meter’s measurement results.
The transducers for the independent single path are located in such a fashion as to traverse the
meter in the center of the meter body. The transducers for the fiscal 4-path meter are located in
the traditional Westinghouse configuration. The reason for locating the single-path in the middle
(center vertically) is to put it in the most profile-sensitive measurement position of the meter.
This will result in a difference in volumes between the single path and the fiscal 4-path when the
velocity profile changes. That is, the single-path meter, with the sensors located in the center of
the flowing gas, is more sensitive to flow disturbances than the 4-path meter design.
These disturbances (velocity profile changes) can be caused by several external factors including
partially blocked flow conditioners and pipeline contamination. All of these will cause a change
in the velocity profile seen at the meter. This concept works because changes in profiles
significantly impact the reading by the centrally located single path while having very little affect
on the 4-path meter. Figure 17 is an artist drawing of this design. Figure 18 shows an 8-inch 4+1
meter with plastic covers over the transducer mounting area.

Figure 17: CBM 2Plex 4+1 Meter Design

 Figure 18: 8-inch CBM 2Plex 4+1-Path Meter
Note that the plastic is simply for a better view and certainly would not be used in the field. The
single-path transducers on the right of the meter are located in the center of the meter (vertical
axis) and do not bounce off of the meter body (direct path configuration). Figure 1 shows this
direct path uses the traditional angle of 60 degrees. Thus the overall path length of this single-
path pair of transducers is only slightly longer than the longest paths (paths 2 and 3) in a 4-path
meter.
Two independent Signal Processing Units (SPUs) are used, one for the 4-path configuration, and
one for the single-path configuration. Both electronics energize their transducers independently
of each other. There is also no communication between the electronics, and no interaction
between the sound pulses from one “meter” to the other.
The comparison of the output of both meters is not on a real-time basis, but rather performed
once per day, or even once per hour. That is, the uncorrected accumulated volume in the 4 path
meter is compared to the single path meter at the daily (or hourly) level. Daily checks help
eliminate the minor differences that can occur on a real-time basis between the two meters due to
their different velocity sampling techniques. This also permits using a tighter tolerance and
increases the reliability of the comparison.
During operation, conditions can change in the piping system that can impact the accuracy of the
meter, even when using a flow conditioner. These changes include blockage of the flow
conditioner with a foreign object, contamination over time from oil and mill scale, unexpected or
unanticipated pulsation of gas, and potential changes within the 4-path meter electronics and
transducers. By incorporating a second independent electronics with an independent path, this
design essentially provides a real-time flow check against the 4-path meter. But why use a single
path design to check the 4-path meter instead of another 4-path design checking the fiscal 4-path
meter?
During the past several years data has shown that single path meters, with the transducers located
to send sound pulses through the middle of the meter body, are more sensitive to profile changes.
In a paper published in 1998 by Terry Grimley [Ref 3], installation effects were measured on two
multi-path meters, and on two single-path meters. A variety of installation effects were tested
including two elbows in and out of plane upstream of the four meters. The multipath meters
performed relatively well with errors attributed to the installation effects on the order of 0.5% or
less. In the same piping configuration the single path meters had errors that were on the order of
2-5%. Clearly the multipath meters could deal with the asymmetrical and swirling profiles far
better than the single-path meters.
Profile changes also occur when contamination develops on the inside of the piping and meter.
As the buildup occurs, the wall friction increases causing the velocity profile in the center of the
meter to be higher relative to the area along the pipe wall. A paper published at the North Sea
Flow Measurement Workshop (NSFMW) in 2005 [Ref 1] discusses how the profile changed over
time due to internal pipeline contamination. This paper shows examples of the meter’s response
when blockage occurs upstream at the flow conditioner. The velocity profile differences between
the 4 path meter and the independent single path meter resulted in significantly different
measurements between the two designs.
Placing the single-path pair of sensors in the center of the meter body was done intentionally as
this is the most sensitive location for flow measurement. That is the center-line path will shift far
more than if located at any other position within the meter. This makes it an excellent check
against the 4-path which experiences much less shift when the profile changes.
Through the use of meter diagnostics, and the associated manufacturer’s software, many of the
above problems can be identified. The problem with the conventional method of identifying
potential measurement errors is that most users only check the meter’s diagnostics on a monthly
basis, and sometimes less often that that. When a problem occurs, it may be weeks before it is
identified, and thus the impact on billing can be substantial.
By using the CBM 2Plex 4+1 method of comparing the output of a single-path meter to that of
the fiscal 4-path chordal meter, the performance of the two meters is validated every day. This
means if a problem occurs, a potential measurement error can be identified by the system within
one day. Once a problem has been identified, technicians can be dispatched to investigate or the
meter can be monitored more closely for further action. In today’s environment where the price
of gas is ever increasing, errors in transportation, buying and selling of natural gas can lead to
more significant financial risk than ever before. Knowing a meter has a potential problem within
a day (or hours) will help reduce unaccounted for gas (UAF).

4. PROVING THE CONCEPT

Does this technique really work when both meters are incorporated into one meter body? To
answer this question, testing was conducted at the CEESI Iowa high flow calibration facility in
Garner, Iowa. For this test a 12-inch 4+1 meter was installed with a CPA 50E flow conditioner
upstream. This type of flow conditioner has been used in many USM applications around the
world.
One of the issues with using a flow conditioner is that debris can collect in front of the flow
conditioner. When this occurs there can be an affect on the USM accuracy. The effect has been
documented in several presentations [Ref 1, 5 & 6] and is demonstrated in the first part of this
paper.
To quantify the benefits of this design, testing with several blockage scenarios was conducted.
Not only were the 40% blockage, but additional testing was done only 1 hole blocked. Three
velocities were used for all of these tests. These were approximately 7 m/s, 14 m/s and 21 m/s.
Figure 19 shows the results of the 4-path meter after baseline calibration (piecewise linearization)
and subsequent results with 40% blockage of the flow conditioner.
12-inch, 4-Path Meter - 40% Blocked Results

1.00
Un-Blocked CPA 40 Percent Blocked CPA
0.75

0.50

0.25
% Error

0.00

-0.25

-0.50

-0.75

-1.00
0 5 10 15 20 25

Meter Velocity (m/s)

Figure 19: 12-inch, 4-Path Meter Results with 40% Blockage

Figure 19 shows that the 12-inch, 4-path meter shifted on the order of -0.15%, or less, for all
three velocities tested. Figure 20 shows the results of the single-path during this same time.
 12-inch, Single-Path Meter, 40% Blocked Results

1.0
Un-Blocked CPA 40% Blocked CPA
0.5

0.0

-0.5

-1.0
% Error

-1.5

-2.0

-2.5

-3.0

-3.5

-4.0
0 5 10 15 20 25

Meter Velocity (m/s)

Figure 20: 12-inch, single-Path Meter Results with 40% Blockage

The accuracy impact on the single-path meter is on the order of -3.5%. Thus for the same
blockage upstream of the 12-inch meter, the single-path meter shifted more than 20 times as
much as the 4-path meter.
The next step was to test the meter with only one blocked hole. Figure 21 shows the results of the
12-inch, 4-path meter with this blockage and Figure 22 shows the results of the 12-inch, single-
path meter.

12-inch, 4-Path Meter - 1 Hole Blocked Results

1.00
Un-Blocked CPA Single Hole Blocked - Botton CPA
0.75

0.50

0.25
% Error

0.00

-0.25

-0.50

-0.75

-1.00
0 5 10 15 20 25
Meter Velocity (m/s)

Figure 21: 12-inch, 4-Path Meter Results with 1 Hole Blocked

 12-inch, Single-Path Meter, 1 Hole Blocked Results

1.00
Un-Blocked CPA Single Hole Blocked - Bottom CPA
0.75

0.50

0.25
% Error

0.00

-0.25

-0.50

-0.75

-1.00
0 5 10 15 20 25

Meter Velocity (m/s)

Figure 22: 12-inch, Single-Path Meter Results with 1 Hole Blocked

Figure 21 shows there is no impact on the 4-path meter’s accuracy, while the single-path meter,
shown in Figure 22, shifted between -0.6% and -0.85%. Thus, even with one hole blocked, the
single-path meter shift was very significant, and thus the difference between the 4-path and
single-path could be easily identified.
Pipeline contamination, especially over time, is a more challenging problem for the technician.
Many pipelines have some minor amount of oil and mill scale that is being transported down the
pipeline. Although this contamination is generally small, it can accumulate over time and have a
significant impact on a meter’s accuracy.
Several papers have been published over the past 10 years [Ref 4, 7, 8, 9 & 10] which discuss the
impact on the meter’s accuracy. Some meter designs tend to register fast when contamination
coats the meter piping and meter body, while others tend to register slower. The challenge for all
users of USMs is to identify this contamination and then to decide when it is time to clean the
meter run.
The question is: “Can the 4+1 CBM meter design identify contamination in the piping?”. To
answer this question an existing 4-path meter was borrowed from an installation. This particular
meter is an inter-company operational meter, in a bi-directional application installed in the late
1990’s and uses a 19-tube bundle flow conditioner. During the several years of service it had
been cleaned numerous times due to contamination.
For this test the entire meter run (including all piping and flow conditioner) was removed, and
sent to the CEESI Garner calibration facility. As the installed meter was not a 4+1 CBM design,
only the piping was used for this testing to see if the contamination could be identified by the
CBM meter.
Unfortunately the meter run had been cleaned recently and the upstream piping was not as dirty
as was expected. Figure 23 shows the “as-found” condition of the piping between the flow
conditioner (19-tube bundle) and the meter.
 Figure 23: 12-inch Dirty Piping
As Figure 23 shows, there was not a lot of contamination remaining due to the recent cleaning of
the meter piping.
Today most designers do not use this type of flow conditioner but instead select a perforated
plate like the CPA 50E. The customer chose to re-install the meter after testing and replace the
19-tube bundle with the CPA unit. For this reason all testing was conducted with a CPA flow
conditioner. To simulate what the flow conditioner may have looked like it had been subjected to
normal pipeline contamination, “texture” paint was applied to the CPA flow conditioner. Figure
24 shows the flow conditioner just prior to being installed for the testing.

Figure 24: 12-inch CPA with “Texture Paint”

Although this coating might not represent the identical contamination to the piping, it was felt at
the time that some type of contamination was needed to at least simulate surface buildup.
Because it was previously decided to re-install the 4+1 CBM meter, after all testing was
completed, for some long-term testing and wanted the CPA to be used during this time. In order
to save some calibration time, rather than conduct testing with the 19-tube bundle, it was decided
to contaminate the CPA for the “as found” dirty testing.
Figure 25 shows the results for the 4-path meter both dirty and clean.
 Figure 25: 12-inch, 4-path As-Found Dirty and As-Found Clean
The results of the 4-path meter as-found baseline are shown with the blue dots after the piping
was cleaned (meter was brand new and thus clean). The red dots represent the as-found results
with the upstream piping and CPA dirty. The table in Figure 26 shows the difference between the
two at each flow velocity in m/s.
Velocity % Diff.
23.3 -0.12
15.7 -0.10
7.8 -0.05

Figure 26: 4-path Dirty vs. Clean Differences

Figure 26 shows the meter registered slightly slower with the upstream piping being dirty
compared to the clean piping. This is the expected results since a previous paper [Ref 4 & 7] had
demonstrated that the upstream piping tends to cause the chordal meter to register slightly slower
when dirty.
Figure 27 shows the results for the single-path during the same conditions.

Figure 27: 12-inch, 4-path As-Found Dirty and As-Found Clean

In Figure 27 the single path meter registered faster (red dots) when the upstream piping was dirty
compared to the clean upstream piping as shown with the blue dots. This difference is
summarized in Figure 28.
Velocity % Diff.
23.3 0.57
15.7 0.50
7.8 0.35

Figure 28: Single-path Dirty vs. Clean Differences

As this table shows the single path meter registered faster when dirty and is just the opposite of
what the 4-path meter showed. Although this isn’t as thorough of a test as a uniformly dirty
meter, it does show that the single-path meter behaves differently than the 4-path meter. With
very little contamination on the upstream pipe wall, the change in the single path meter’s
response was easily seen.

5. IMPLEMENTING THIS DESIGN

As discussed earlier in this paper, both electronics operate independently. The output of each
meter needs only to be brought into the same flow computer and volumes stored for both as
would normally be done for two separate meters. To take advantage of this feature, the hourly
uncorrected volumes would then be compared and an acceptable tolerance would be determined
based upon some history established during commissioning. The tolerance may vary somewhat
from site to site, and will depend slightly upon the upstream piping conditions and the symmetry
of the profile downstream of the flow conditioner. However, the typical agreement that has been
seen from some field data is on the order of ±0.5%.
The comparison test probably should not be conducted when meter velocities are below perhaps
3 m/s as the profile effects can become more significant. For this reason the flow computer
should accumulate separate totals for comparison testing since the effective cutoff for the
comparison would be perhaps 3 m/s. Thus it may not be practical to use the absolute uncorrected
volumes through each meter if the flow rate is frequently below this velocity.
For most installations meter velocities are usually always above 3 m/s. This is common in
mainline stations where there is always flow. For these cases a direct comparison of hourly (or
daily) uncorrected volumes would suffice. Many users already have this capability built into their
flow computers. They do comparisons of “run ratios” in order to spot potential problems. For
these users they simply have to connect the meter to the flow computer, set the comparison ratio
to a value, and start monitoring for the alarm. Thus full advantage of this meter design can be
incorporated immediately without special flow computer programming.
Figure 29 is an example of a 2Plex 4+1 meter in a bi-directional application that has been
compared at the daily level for more than 19 months.
 2.00% 500,000
Fwd
1.50% Rev 450,000
ACFH
1.00% 400,000
% Diff. 4-path vs 1-path acf

0.50% 350,000

0.00% 300,000

ACFH
-0.50% 250,000

-1.00% 200,000

-1.50% 150,000

-2.00% 100,000

-2.50% 50,000

-3.00% 0
9/18/07 12:00 PM

12/7/07 2:00 PM

1/6/08 10:00 PM
1/18/08 12:00 PM
2/8/08 5:00 PM

3/10/08 6:00 PM

3/30/08 1:00 PM
4/12/08 7:00 PM
4/23/08 12:00 PM

7/12/08 12:00 PM
7/24/08 5:00 PM
8/4/08 2:00 PM
8/14/08 9:00 PM

9/24/08 4:00 PM
10/3/08 11:00 PM
10/13/08 8:00 PM
10/22/08 8:00 PM
10/31/08 1:00 PM

11/18/08 7:00 PM

12/12/08 5:00 PM
12/23/08 1:00 PM
12/31/08 11:00 PM
1/9/09 2:00 PM

2/12/09 11:00 PM

4/5/09 3:00 PM
4/14/09 1:00 PM
4/25/09 1:00 PM
9/30/07 12:00 AM

12/20/07 9:00 AM
12/29/07 2:00 AM

2/20/08 1:00 AM
3/2/08 8:00 AM

3/21/08 3:00 AM

5/5/08 12:00 AM
5/14/08 3:00 AM
5/27/08 7:00 AM
6/10/08 7:00 AM
7/1/08 7:00 AM

8/25/08 4:00 AM
9/6/08 10:00 AM
9/16/08 8:00 AM

11/9/08 5:00 AM

12/1/08 7:00 AM

1/20/09 10:00 AM
2/1/09 12:00 AM

2/24/09 6:00 AM
3/8/09 12:00 AM
3/23/09 5:00 AM
10/11/07 18:00
10/24/07 16:00
11/4/07 6:00
11/17/07 16:00
11/26/07 10:00

Figure 29: 4-Path vs. 1-Path Daily Comparison for 19 Months of Hourly Flow Rates
This graph shows the comparison of the percent difference in dark blue for the forward direction,
and magenta in the reverse with the scale on the left axis. The average flow rate is in orange and
the scale is on the right. This graph is of hourly flow comparisons graphed over 19 months time.
For the most part, all hourly comparisons are all within 0.25%. On a few occasions the difference
exceeds ±0.5%. This is when the flow rate is low and perhaps during this time some minor
contamination occurs. When the flow rate increases, the difference goes back to the normal
(which is approximately -0.15%). This may indicate some clearing out of minor contamination
since the meter difference generally goes positive which is expected if the piping becomes dirty.
Many users today know they have contamination in their metering systems. They periodically
clean the meter in order to minimize the uncertainty effect due to contamination. If the amount of
difference between the 4-path meter and the single-path meter can be used to determine the
cleaning interval, these users will then benefit from extended inspection intervals and thus save
significant O&M expenses.
6. CONCLUSIONS
Today the cost of energy is higher than it was several years ago, and it is not likely this trend will
reverse itself. By implementing ultrasonic metering technology many users have been able to
improve their measurement and reduce their un-accounted for (UAF) gas during the past several
years. One task always remains for the technician and that is to insure the meter is operating
correctly and accurately. This applies to all measurement technologies, not just USMs. The
significant benefit of the USM is the ability to provide diagnostic information for the user to help
determine the meter’s “health.”
Today technicians have software to help understand the operation of their USM. Since each
manufacturer of USMs uses a different velocity integration technique (different path
configurations), it is often difficult for the technician to fully understand whether his USM is
operating correctly or not. Additionally, since most only inspect the meter’s operation once per
month, problems can occur and go undetected for many days or weeks. This can significantly
increase measurement uncertainty during this time.
The CBM 2Plex 4+1 meter design relies on basically two principles. First, the fiscal meter is
chosen to be the least sensitive to any flow profile changes that may occur in normal operation.
And second, the “check” meter design is chosen to be one that is the most sensitive to any flow
profile changes. Ideally any affect from a profile change would not only have a significantly
different impact on accuracy for each path layout, but the affect would be in opposite directions,
making the difference much easier to detect.
The benefit of the CBM 2Plex 4+1 meter design is that the flow computer is used to check the
health of the fiscal 4-path chordal meter by simply comparing it to the single-path meter. If the
velocity profile remains relatively constant, both meters will agree. Should some process
condition upset the normal profile, the single-path meter will respond significantly different than
the 4-path. These upsets can include the following:
• Blockage in front of the flow conditioner
• Contamination due to oil and mill scale buildup over time
• Pulsation in the pipeline due to compressors (sampling rate for the single-path is much
faster than the 4-path and thus less sensitive to pulsation)
• Potential problems with the fiscal meter including transducers and electronics problems
• Full redundancy should there be a failure of the electronics.
When a meter is equipped with automated meter diagnostics, as described in Section 2, it
expands the monitoring of the fiscal meter’s health to an even higher level. The 2Plex 4+1 design
only validates that the velocity profile hasn’t changed. By monitoring all of the remaining
diagnostics on a real-time basis, the meter’s health can be validated on a real-time basis. This is
important should a diagnostic value such as gain for a pair of transducers, or low SNR from a
control valve, approach a value that may cause a path to fail.
By monitoring all aspects of the meter’s diagnostics, both with the 2Plex redundant design, and
with continuous checking of all other diagnostics (for both the 4-path and 1-path meter), the user
can have a much higher degree of confidence that the measurement is accurate.
Today the cost of accuracy has never been more important. Virtually all applications today
require the measurement accuracy be maintained at the highest possible level. The CBM 2Plex
4+1 meter design, combined with automated real-time internal monitoring of all diagnostic
values, provides a complete “health check” on the custody transfer meter (4-path). This can
significantly reduce both operation and maintenance (O&M), and measurement uncertainty, and
thus reduces the cost of doing business.
7. REFERENCES
1. John Lansing, How Today’s USM Diagnostics Solve Metering Problems, North Sea Flow
Measurement Conference, October 2005, Tonsberg, Norway
2. Klaus Zanker, Diagnostic Ability of the Daniel Four-Path Ultrasonic Flow Meter, Southeast
Asia Flow Measurement Workshop, 2003, Kuala Lumpur, Malaysia
3. T. A. Grimley, The Influence of Velocity Profile on ultrasonic Flow Meter Performance,
AGA Operations Conference, May 1998, Seattle, Washington, USA
4. John Lansing, Dirty vs. Clean Ultrasonic Flow Meter Performance, North Sea Flow
Measurement Conference, October 2004, St. Andrews, Scotland
5. Larry Garner & Joel Clancy, Ultrasonic Meter Performance – Flow Calibration Results –
CEESI Iowa – Inspection Tees vs. Elbows, CEESI Ultrasonic Conference, June 2004, Estes
Park, Colorado, USA
6. John Lansing, Features and Benefits of the SICK Maihak USM, CEESI Ultrasonic
Conference, June 2006, Estes Park, Colorado, USA
7. John Lansing, Dirty, vs. Clean Ultrasonic Gas Flow Meter Performance, AGA Operations
Conference, May 2002, Chicago, Illinois, USA
8. L. Coughlan, A. Jamieson, R.A. Colley & J. Trail, Operational Experience of Multipath
Ultrasonic Meters in Fiscal Service, North Sea Flow Measurement Conference, October
1998, St. Andrews, Scotland
9. John Stuart, Rick Wilsack, Re-Calibration of a 3-Year Old, Dirty, Ultrasonic Meter, AGA
Operations Conference, April 2001, Dallas, Texas, USA
10. James N. Witte, Ultrasonic Gas Meters from Flow Lab to Field: A Case Study, AGA
Operations Conference, May 2002, Chicago, Illinois, USA
11. John Lansing, Operation and Maintenance Considerations for Ultrasonic Meters,
Appalachian Gas Measurement Short Course, August 2008, Coraopolis, Pennsylvania, USA
12. John Lansing, Advanced Ultrasonic Meter Diagnostics, Western Gas Measurement Short
Course, May 2007, Seattle, Washington, USA