Calibration of Ultrasonic Flowmeters

G M - 4055
William Johansen and Joel Clancy
Colorado Engineering Experiment Station, Inc.

long data points or many data points at a single

flowrate can be taken. This type of system can hit
very high flowrates allowing calibration of the largest
ultra-sonic meter sizes. The high velocities allow
observation of the meter and meter components
under actual pipeline conditions.
Some meter
components such as temperature wells and flow
conditioners fail or generate significant amounts of
noise at high velocity. The effect of noise can be
immediately determined at a calibration facility with
the ability to flow at high meter velocities.

Introduction
Ultra-sonic flowmeters are currently being
put into service in large numbers. When used in
custody transfer applications, ultrasonic flowmeter
calibration is required.
There is currently no
standard covering the calibration of ultra-sonic
flowmeters, although AGA Report No. 9 is used as a
guide. The process of calibrating an ultra-sonic
meter will be discussed as well as the calculation of
a flow weighted mean error (FWME) and a
calibration factor in accordance with AGA Report No.
9.

The disadvantages of this type of system

vary. The pressure drop across the calibration
facility will be determined by the pipeline operator.
This should not affect ultra-sonic calibrations but can
be a consideration when calibrating meters like
orifice meters that need a pressure drop to measure
flow. The calibration system has to operate at the
pressure in the pipeline. This means that a meter to
be used in a low-pressure system may have to have
a flange with a higher pressure rating installed
temporarily for the calibration.

General Description of Calibration

Calibrating an ultrasonic flowmeter is
performed by placing a flow standard in line with the
ultra-sonic flowmeter being calibrated. The flow
standard is used to accurately measure flow and has
been calibrated using standards that are traceable to
NIST. As long as there are no leaks in the system
between the flow standard and the meter being
calibrated it can be assumed that the two meters are
passing the same amount of flow. There may only
be one standard or there may be many that can be
placed in parallel in the flowstream to produce a
wide flowrate range.

The second type of natural gas calibration

system used is shown in Figure 2. This type of
system is referred to as a pressurized loop. Prior to
flowing, the loop is pressurized with gas to the
desired flowing pressure. Flow through the system
is created by a compressor that must run
continuously while calibrating. Flow control valves
can be placed in the system for flow control. Heat
exchangers in the system allow some temperature
control.

There are two basic types of calibration

systems. Figure 1 shows a calibration system on an
existing natural gas pipeline. When the large valve
on the pipeline is closed slightly a differential
pressure across the valve is produced.
The
differential pressure across the pipeline valve
provides a motive force to push flow through the
calibration system. As the main pipeline valve is
closed further, more flow is pushed through the
calibration system. In this manner a wide flowrate
range can be passed through the calibration system
allowing the calibration of a wide range of meter
sizes. At very low flowrates, fine flow control can be
accomplished by throttling with a smaller valve inline
with the meter being calibrated.

A pressurized loop system also has certain

advantages. The flow in pressurized loop systems
can be controlled very precisely; this allows the
system to produce high quality data.
The
temperature in a pressurized loop can be varied
over a limited temperature range allowing the effects
of different flowing temperatures to be investigated.
The composition of the flowing gas can be varied by
injecting different components into the loop.
The disadvantages of a pressurized loop
calibration system include high operating expenses.
This is because the compressor must be operating

There are advantages to this type of system.

Because the pipeline is passing flow constantly, very

415

disassembled for shipping.

Inspect the cables
carefully ensuring all cables are with the meter and
that no damage has occurred. Inspect the inside
surfaces of the meters. There is often a build-up of
contaminants. Ensure the pressure taps are clear.
The customer may want the meter calibrated in the
condition it arrives in, referred to as an "As Found"
calibration, then cleaned and recalibrated clean,
referred to as an "As Left" calibration. There is not
much data available yet about how well ultra-sonic
flowmeters perform with thick layers of contaminants
on the transducers and pipe walls. Any difference in
performance of the meter between the "As Found"
and "As Left" calibration may be very useful to the
customer.

continuously.
The suction pressure on the
compressor must be maintained above some
minimum value, which places a limitation on the
differential pressure across the loop. This limitation
on differential pressure places a flowrate limit on
pressurized loop systems that is typically much
lower than that of pipeline based systems.

Pre-Calibration Inspection and Meter Installation

Upon receiving the ultra-sonic flowmeter at
the calibration facility, a thorough inspection is
started. Ultra-sonic meters are often very large with
attached electronic instruments so the inspection of
the ultra-sonic flowmeter begins before it comes out
of the box. A damaged shipping container indicates
that the meter may have visible damage or damage
to electronic components that will be harder to find.
Open the crate and inspect the electronics. Ensure
that all electronic boards are securely fastened to
the junction box that houses the electronics. Look
for any signs of damage or a n y loose parts or
fittings. Inspect the meter body. Ensure that the
transducers are not damaged. Ensure all cables are
securely fastened.

Once the meter and spool pieces have

been installed in the test section the instrumentation
can be installed and the meter can be powered up.
Install pressure and temperature transmitters. Dual
instrumentation
is preferred.
When
dual
instrumentation is used any differences in readings
can be identified quickly allowing the calibration to
proceed smoothly. Connect the power supply to the
ultra-sonic meter. They typically require a 24-volt
DC power source, but sometimes require 110 VAC.
Connect the RS-485 communication lines to the
meter. One RS-485 output is used to communicate
flow and meter status information to a computer
running software provided by the manufacturer.
Another output is also connected to the meter. This
output is a second flow signal from the meter. The
second output may be an RS-485 output or the
meter may produce a frequency output, which is
proportional to flow passing through the meter. If an
older meter is received from the field for recalibration
it may require some communication switch changes
to allow communication with the calibration facility.
These changes must be accomplished with the
support of the manufacturer.

When installing the meter in the piping

system check the holes where the pressure taps
penetrate the meter body on the inside surface. Any
burrs or protrusions on the pressure taps can create
pressure-reading errors and must be removed prior
to calibration. Ultra-sonic flowmeters are often sold
with upstream and downstream spool pieces. There
may be identification stamps on the meter and
accompanying spool pieces, make sure the
identification numbers match. The meter and spool
pieces may have alignment pins.
Check the
alignment, it is not unusual to find the pins do not
provide good alignment.
Spool pieces may come
separately from a different supplier. In this case
make sure that the internal diameters match well.
Drawings may accompany the meter and spool
pieces. Assemble the meter parts as shown in the
drawings. Check the drawings for alignment pins
that may not have been installed. Flow conditioners
sometimes fail when first used. Inspect the flow
conditioner to ensure the manufacturing and
assembly is complete. Some flow conditioners need
to be pinned as they can move around inside the
pipe when installed. It is important that all upstream
components that can affect the flow conditions at the
meter remain exactly the same in use as they were
during the calibration.
Meters that have been in use in the field are
often recalibrated.
These meters do not have
original shipping containers and are often partially

The location of the thermal wells should be

noted. AGA Report No. 9 discusses the appropriate
placement of a thermal well stating 2 to 5 pipe
diameters downstream of the ultra-sonic meter. In
the case of a bi-directional meter, AGA 9 calls for a
thermal well placement of 3 diameters from either
ultra-sonic meter flange face. Although thermal well
placement is defined in AGA 9, some users elect to
choose a different location for their thermal well(s).
Caution should be taken here. Often thermal wells
are placed upstream of a flow conditioner. The
pressure drop created by the flow conditioner also
creates a corresponding temperature drop know as
the JT effect.
You have, therefore, a different
temperature at the meter than is being recorded by
the temperature transmitter.

416

monitor the system carefully when increasing flow to

the highest flowrate. If any components like flow
conditioners are going to fail then this is the time
when failure is most likely to happen. Any unusual
noises or large changes in noise may indicate that a
system component is experiencing failure.

Meter Calibration

Test section pressurization, leak check, and

pre-flow are now performed. As the meter body
pressurizes, dual pressure instrumentation is
checked for good agreement. Pre-flow is generally
conducted at 60 to 80% of the meter capacity. Preflow will last for 30 to 60 minutes. The pre-flow
allows the meter and test section piping to warm up.
Dual temperature instrumentation is checked for
good agreement. During pre-flow, several piping
and instrumentation conditions are checked. Flow
conditioners are often a source of flow noise. The
amount of noise being generated by the flow
conditioner is monitored during pre-flow.
Any
unusual mechanical noises may be an indication
that the flow conditioner is coming apart or vibrating
violently. Installing a thermal well too close to a flow
conditioner can cause thermal well vibration. This
installation can produce several problems. This
vibration can cause problems for the ultra-sonic
meter. The introduction of noise inhibits the meter's
ability to function properly.
The thermal well
vibration also creates a heating effect that will
produce a temperature measurement error at the
ultra-sonic. Pre-flow is very important for multi-path
ultra-sonic flowmeters.
Multi-path ultra-sonic
flowmeters have an internal memory they draw upon
when they experience chord failure.
When
performing pre-flow, the performance of all the flow
transducers is monitored to ensure there are no
chord failures. Unusual signals can be produced
from a variety of problems to include a bad set of
transducers, improper characterization of the
transducers, or improper internal wiring to name a
few. If all the chords are performing well, the flow
memory stored in the meter is that of good solid
meter performance.

When flow at the highest flowrate has been

established, the calibration system is allowed to
stabilize. Ultra-sonic meter calibration systems may
be composed of large piping systems with a
considerable amount of volume between the
standards used to accurately measure flow during
the calibration and the ultra-sonic meter being
calibrated.
It is important that any pressure
fluctuations that may be present in the system due
to changes in flowrate are allowed to dissipate.
When stable flow conditions have been observed for
an adequate length of time, calibration data can be
taken from the ultra-sonic meter being calibrated
and the calibration system. Several data points may
be taken at a single flowrate. The number of data
points may be specified by the customer, or it may
be left to the judgement of the calibration system
operator. In general, fewer data points are taken at
higher flowrates than at the lower flowrates.
Data is typically taken at a minimum of six
flowrates. AGA Report No. 9 specifies flowrates of
qmin, 0.10qmax, 0.25qmex, 0.40qmax, 0.70qmax, and qmax.
Additional data points may be requested at specific
flowrates by the customer if the meter is to be used
in a specific flowrate range.
Data may be acquired using two separate
computer systems. One ~system will be running
software supplied by the manufacturer that will
interrogate the meter while a data point is taken and
another system will acquire data from the calibration
system. It is important that the two systems acquire
data during the same time period.

Once pre-flow is finished, the calibration

begins. The flow is taken up to the highest flow rate
requested by the customer. If no flow rates have
been specified by the customer, the flow rate is
taken to the maximum flowrate suggested by the
manufacturer. At the high flow rate there may be
enough flow noise to cause chord failure. That is,
the flow noise is of a sufficient level to weaken the
signal received by the meter. The flow noise is often
produced by system components like pipe reducers
and flow conditioners. When chord failure occurs it
is very important that the memory of flow stored in
the meter is good. This flowrate may not be a
flowrate the meter will normally operate at but some
indication of the performance of the meter at the
high flowrate may be desired. It is important to

Obtaining the calibration .log or .asc from

the meter's software data logs can prove to be an
important tool once the meter is put into service.
Logs collected at the time of the calibration can
provide information such as speed of sound, or gain
level to limit ratios on a chord by chord basis to
name a few. This information can be used as a
baseline. When collecting logs throughout the life of
the meter, the baseline logs can be used as a
reference. Any deviations from the ratios observed
at the calibration can be used as a way to
troubleshoot
potential
problems
with
meter
performance.

417

Calibration Report

2.0

The results of a calibration will include the

individual data points taken at each flowrate, the
average values from the individual data points, and
the calibration factor. These values may not all
appear on the calibration report but they should be
available to the customer. Meter and electronics
identification numbers should also be recorded and
maintained in records. There are also software
settings, which need to be recorded.
These
software settings include the meter factor and other
settings, which directly affect the performance of the
meter.

0.5
uJ
"~

-1.0
-1.5
-2.0
0

35,000
21,875
8,750
3,750

3. ft/sec

60000

80000

100000

37,801
21,980
6,890
3,475

37,681.9
21,910.1
6,915.9
3,508.1

-0.315
-0.318
0.376
0.953

The plot shown in Figure 3 below shows the

results of the calibration. It appears that the meter
consistently under-measured flow through most of
the range and then appeared to begin overmeasuring flow at the low flowrates.
A flow weighted mean error (FWME)is now
calculated.
The FWME is a single value that
describes the performance of the meter just
calibrated. The FWME is found by first expressing
each flowrate as a percent of full-scale flow. The
percent full-scale calculation for the high flowrate
values is shown below.

Percent Full Scale =

Table 1. Calibration Test Plan

qmax

40000

The calibration now proceeds with pre-flow and

recording data at the 6 flowrates listed above. After
all the individual data points have been averaged the
data would appear as shown in Table 2.

An 8-inch ultra-sonic flowmeter is to be

calibrated. The maximum flowrate of this meter is
87,500 ACFH. The customer has requested that
calibration data be taken in compliance with AGA
Report No. 9 with a minimum flowrate at a meter
velocity of about 3 ft/sec. The calibration facility
operator sets up the calibration plan shown in Table
1.

0.75qma~
0.40qm~
0.25qr~
0.10qn~
qmi,

20000

Flowrate (ACFH)

AGA Report No. 9 Calibration Factor Calculation

Example

Maximum
Meter Flowrate
of 87500
ACFH

0.0

no -0.5

The meter factor or calibration factor is the

final product of the calibration. This factor is input
into the software of the ultra-sonic flowmeter to
introduce a constant offset in meter performance.
The calculation of the calibration factor is shown in
an example in Appendix A of AGA Report No. 9.
This example is covered in detail in the following
paragraphs.

AGA Report No. 9

Flowrates
(ACFH)

...............................................................................................................................................................

1.5

Calibration
Flowrates
(ACFH)
87,500
61,250
35,000
21,875
8,750
3,750

Indicated Flowrate
Maximum Flowrate

xlO0.

Table 2. Calibration Results

Desired
Flowrate
(ACFH)
87,500
61,250

Actual
Flowrate
(ACFH)
86,500
60,415

Flowrate
Indicated by
Meter being
Calibrated
(ACFH)
86,183.4
60,190.3

Error

(%)
The percent full-scale values are then multiplied by
the percent error values. The values in the percent
full-scale column and the percent full-scale times

-0.366
-0.372

418

percent error column are summed. This is shown in

Table 3 below.

Actual
Flowrate
(ACFH)
86,500
60,415
37,801
21,980
6,890
3,475

Table 3. S u m m e d Values
Percent
Percent F.S.x.
Error (%)
Full Scale
Error

0.5
w
o

0.0

#. -0.s

-0.366
-0.372
-0.315
-0.318
0.376
0.953

0.9886
0.6905
0.4320
0.2512
0.0787
0.0397
Sum = 2.4807

-0.3618
-0.2568
-0.1361
-0.0799
0.0296
0.0378
Sum =-0.7672

-1.0
-1.5
-2.0

- 0.7672

80000

100000

Summary
Ultra-sonic flowmeters are being installed in natural
gas pipelines in large numbers. The calibration of
ultra-sonic flowmeters is performed in accordance
with AGA Report No. 9. The inspection, installation,
and calibration of ultra-sonic flowmeters were
discussed. The FWME calculation example in AGA
Report No. 9 was covered in detail.

= - 0.3093

The FWME value is not the value entered into the

ultra-sonic flowmeter software to correct metering
performance. The calibration factor is the value
used to correct meter performance and is calculated
in the following manner.
100

60000

Figure 4. Adjusted Calibration Results

2.4807

Calibration Factor =

40000

Flowrate (ACFH)

The FWME is then found by dividing the summed

percent full scale times percent error values by the
summed percent error values as shown in the
following equation.

FWME -

20000

= 1.0031

(100 + F W M E )

Each of the indicated flowrate values can

now be multiplied by the calibration factor. The
adjusted flowrates are shown in the Figure 4.
If the meter just calibrated is to be operated
primarily in the 20,000 to 30,000 ACFH range should
the low flowrate range data be used in the
calculation of the FWME and calibration factor? If
the 2 lowest flowrates were deleted from the
calculation of the FWME the new calibration factor
would be 1.0035, a difference of less than 0.05%.
The small change is due to the flow weighting.

Figure 3. Calibration Results

419

Direction of Flow through Pipeh'ne

Natural Gas t~peh~e

Pipeh'ne disserential Pressure Control Valve

Flow Control Valve

Ultra-sonicMeter

Flow Standard

Compressor

Heat Exchanger

Esl
Flow Control Valve

U/tra-sonic Meter

Flow Standard

Direction of FlowAroundLoop

Figure 1. Pipeline Based Calibration System

Figure 2. Pressurized Loop Calibration System

420

,)