Turbulence

Calibration of ultrasonic
meters using small
volume provers
Dr Gregor Brown
Research Director
Caldon Ultrasonics
Introduction
• Background
• Challenges associated with calibrating ultrasonic
meters using small volume provers
• API Chapter 5.8 guidelines
• Improvements that can be achieved
• Summary and conclusions
Why use small volume provers?
• Traceable calibration of ultrasonic meters can be
performed in a calibration laboratory or in the field
• Reliance on laboratory calibration requires great
care at each stage
– Meter selection
– Specification of calibration conditions
– Avoidance of ‘installation effects’
Advantages of field calibration (proving)
• Calibrating a meter in the field removes some of
the uncertainty related to fluid properties and
installation conditions
• No need to periodically remove and return the
meter to a calibration laboratory
Disadvantages of proving
• Installation of a prover can be more costly both in
terms of capital and maintenance
• In practice, a requirement for in-situ proving could
limit the size of ultrasonic flow meter that can be
used
Small Volume Provers
• By small volume prover
we mean a piston style
prover, often called small
volume, compact or
portable provers as they
can be smaller in size
than their ball prover
equivalents
Small Volume Prover
Invar Rod
Prover Body
Spring Plenum
Start and End Detector
Standby
Switches
Hydraulic
System
Inlet
From Meter
Outlet
Small Volume Provers
• The essential
difference between a
piston prover and a
ball prover is the
enhanced resolution
of the detector
switches, hence
allowing the same
uncertainty to be
achieved with a
smaller volume
Challenges for ultrasonic meters
• The small volume of piston provers results in two
specific challenges for ultrasonic meters
– A short time for each calibration run
– Only a small volume of flowing fluid is passed though
the meter for each run
• These might sound like the same thing, but they
relate to two different aspects of meter
performance
Short time per calibration run
• Ultrasonic meters measure the flow electronically
and hence sample the flow, perform calculations
and then update an electronic output
• Too low a sampling rate, or a significant delay
between sampling and output, will result in errors
when the flowrate changes during proving
Short time per calibration run
• This is recognised in API Chapter 5.8, Appendix
C – Manufactured Flow Pulses and Their Impact
on the Proving Process
Sampling and update rate
• Caldon meters perform approximately 400 velocity
measurements a second and update the flowrate
output at a rate between 50 and 100 Hz depending
on meter model
• If the flow rate variations during proving are not too
extreme, this is sufficient to minimise
‘synchronisation’ uncertainties
What about the second issue?
Performance of modern electronics
• Modern transit time ultrasonic meter electronics
allow measurement of transit time differences with
sub nanosecond accuracy
• This can be verified at zero flow conditions
• Yet when we use the meters in flowing conditions
the variability is greater
50
2000
45
1800
40
1600
Standard deviation
35
1400
Flowrate
30
1200
25
1000
20
800
15
600
10
400
5
200
0
0
09:21:36
09:28:48
09:36:00
09:43:12
Time
09:50:24
09:57:36
10:04:48
Flowrate (m3/hr)
Standard deviation of transit time difference (ns)
Flow effect on variability of measurements
Variability in percentage terms
7
6
Standard deviation (%)
5
4
1 m/s
3
3 m/s
5 m/s
2
7 m/s
9 m/s
1
10 m/s
0
1
2
3
Path number
4
This instability of reading is caused by
turbulence
Turbulence
Leonardo da Vinci (1452 – 1519)
First to attempt scientific study of turbulence
(turbolenza): placed obstructions in water and
observed the result:
“Observe the motion of the surface of the water,
which resembles that of hair, which has two
motions, of which one is caused by the weight of
the hair, the other by the direction of the curls;
thus the water has eddying motions, one part of
which is due to the principal current, the other to
random and reverse motion.”
So what are the practical implications of
the effects of turbulence?
Transit Time
Transit times with flow
tup
tdown
Flow with turbulence
Averaging the random turbulence
Average velocity with superimposed turbulence
Average velocity
Ultrasonic velocity sample
Averaging the random turbulence
Average velocity with superimposed turbulence
Average velocity
Ultrasonic velocity sample
Averaging the random turbulence
Average velocity with superimposed turbulence
Average velocity
Ultrasonic velocity sample
Averaging the random turbulence
Average velocity with superimposed turbulence
Average velocity
Ultrasonic velocity sample
Averaging the random turbulence
Larger volumes and more runs
• The effects of turbulence are random and
are reduced by averaging
• It is not the sample rate of the meter that is
the limiting factor in terms of repeatability, it
is the frequency of the turbulence
• This fact is recognised in API Chapter 5.8,
which allows for more runs to be used, and
acknowledges that owing to the effects of
turbulence, larger prover volumes are
normally required for ultrasonic meters
API recommended volumes
600
5 runs, equation 5
Suggested Volume (barrels)
500
8 runs, equation 5
10 runs, equation 5
400
5 runs, API B-2
8 runs, API B-2
10 runs, API B-2
300
200
100
0
2
4
6
8
10
12
Meter diameter (inches)
14
16
18
How can we apply this to SVPs?
• First we have to choose the size a small volume
prover that we are going to use
• We do this by considering the flow capacity of the
prover and also with due consideration of the
resulting prove time compared to update rate of
the meter
• Once we know our prover size we have to use a
sufficient number passes per run to accumulate
the volume suggested by API
Prover volume vs flowrate
0.700
0.600
Daniel Brooks
Calibron SyncoTrak
Displaced volume (m3)
0.500
Flow Management Devices
y = 1.379E-04x
0.400
8-inch ultrasonic meter
16-inch ultrasonic meter
0.300
0.200
0.100
0.000
0
500
1000
1500
2000
2500
3000
Maximum flowrate (m3/hr)
3500
4000
4500
5000
So how many passes are required?
• Even if the prover is sized to cover the full range of
the meter, then to accumulate the API suggested
volumes (for a meter factor uncertainty of 0.027%),
typically more than 300 passes in total are
required
• Ouch!
Can improvements be made?
• A number of things can potentially be done to
improve performance
• Increasing the sample rate used in the meter has
limited effect on its own but can help when
combined with flow conditioning measures
• A test programmed carried out in the Cameron
laboratories has demonstrated that significant
levels of improvement can be achieved
Test programme
• Tests carried out using a 6-inch meter with a
reducing nozzle (4-inch throat)
• The meter was tested directly against a 120 litre
Brooks small volume prover
• The meter body had eight paths and could be
tested as either a four or eight path meter
• Tests were conducted with and without a Cameron
designed ‘turbulence conditioner’
Test programme
• Tests carried out using a 6-inch meter with a
reducing nozzle (4-inch throat)
• Tested directly against a 120 litre Brooks
small volume prover
• The meter body had eight paths and could
be tested as either a four or eight path meter
• Tests were conducted with an without a
proprietary ‘turbulence conditioner’
Example data sets
1520
1520
4-path meter 25 Hz sample frequency
1518
1516
1516
1514
1514
1512
1512
Pulse count
Pulse count
1518
1510
1508
1510
1508
1506
1506
1504
1504
1502
1502
1500
1500
0
20
40
60
Pass number
80
100
8-path meter 50 Hz sample frequency
0
20
40
60
Pass number
80
100
Caldon meter with reducing nozzle
• Reducing nozzle shaped inlet
• Substantial diameter/area reduction
• Beta < 0.64, area ratio < 0.41
• Downstream pressure recovery cone
Diagnostic data indicating
improved repeatability
8%
Meter with reducing nozzle
Meter without reducing nozzle
Path velocity standard deivation
7%
6%
5%
4%
3%
2%
1%
0%
0
1
2
3
Path number
4
5
Presentation of proving data
• Results are presented in the form of a ‘provability
factor’
• The basis of this factor is described in the paper
presented at South East Asia flow measurement
workshop in March this year
• It is used to describe the proving performance of a
meter design independent of the size of the meter,
or the size of the prover used to perform the tests
Provability factor
• The fluid volume required to achieve a given
proving performance is proportional to the
provability factor
• Therefore the smaller the provability factor
the better
• The API Chapter 5.8 suggested volumes
correspond to a provability factor equal to
2.734 x10-5
4-path meter with reducing nozzle
2.5E-05
2.0E-05
185 m3/hr, 35 psi plenum pressure
Provability factor, p
192 m3/hr, 75 psi plenum pressure
320 m3/hr, 35 psi plenum pressure
1.5E-05
320 m3/hr, 75 psi plenum pressure
P corresponding to API table b-2
Caldon standard 4-path meter vs SVP
1.0E-05
5.0E-06
0.0E+00
0
25
50
75
Flow sampling and update rate (Hz)
100
125
Eight paths vs four paths
3.5E-06
3.0E-06
Provability factor, p
2.5E-06
2.0E-06
1.5E-06
4-path configuration, 185 m3/hr
1.0E-06
4-path configuration, 320 m3/hr
5.0E-07
8-path configuration, 330 m3/hr
0.0E+00
0
25
50
75
Flow sampling and update rate (Hz)
100
125
Turbulence conditioning
• It has been shown already that the repeatability of
the measurements is affected by natural
turbulence
• This creates the opportunity to improve
performance by adding a turbulence conditioning
device upstream of the meter
• The job the conditioner has to do is different to that
of a ‘normal’ flow conditioner
Turbulence conditioning
• It divides the flow into smaller channels thus
restricting the size of the turbulent eddies
– Reduces variability in the samples by averaging
along/within the path
– Increases the turbulence frequency
• It should be placed close to the meter so that
natural turbulence does not redevelop between the
conditioner and the meter
Turbulence conditioner
Turbulence conditioner
• As the conditioner has
small openings, in
practice it may be
necessary to use a
strainer upstream
Diagnostic data indicating
improved repeatability
8%
Meter with reducing nozzle
Path velocity standard deivation
7%
Meter without reducing nozzle
Meter with reducing nozzle and
turbulence conditioner
6%
5%
4%
3%
2%
1%
0%
0
1
2
3
Path number
4
5
4 and 8 paths with turbulence conditioner
3.0E-06
4-path configuration, no conditioner
4-path configuration, with conditioner at 0D
8-path configuration, no condtioner
2.5E-06
8-path configuration, conditoner at 0D
Provability factor, p
8-path configuration, conditioner at 4D
2.0E-06
1.5E-06
1.0E-06
5.0E-07
0.0E+00
150
170
190
210
230
250
270
Flowrate (m3/hr)
290
310
330
350
Summary comparison
• Green = SVP with < 10 passes per run
Prover volume vs meter size for 5 runs with 0.05% spread
4-path
8-path
4-path
8-path
meter with meter with
API 5.8
meter
meter
reducing
reducing
Table
with
with
nozzle and nozzle and
B-2
reducing reducing
turbulence turbulence
Pipe
nozzle
nozzle
conditioner conditioner
diameter
Prover size (barrels)
(inches)
6
73
7.4
3.2
2.0
0.8
8
130
13.2
5.7
3.5
1.5
10
204
20.6
8.8
5.4
2.2
12
294
29.6
12.7
7.8
3.2
14
400
40.2
17.2
10.6
4.3
16
522
52.5
22.5
13.8
5.7
CONCLUSIONS
• Ultrasonic meters are fundamentally different
from turbine and PD meters and have to be
treated differently
• Sample rate and calculation delays can result in
errors if the meter’s electronics are not of a
sufficiently fast design
CONCLUSIONS
• API chapter 5.8 gives recommendations in terms
of proving volume and number of runs based on
an assumed meter performance
• The API suggested volumes result in a large
number of passes per run when using a SVP
• Combined with some ultrasonic meters having
lower sampling rates and/or longer calculation
delays this has resulted in a widely held
perception that ultrasonic meters can not be
calibrated directly using a SVP
CONCLUSIONS
• The test data presented in this paper shows that
dramatic reductions in proving volume
requirements can be achieved by changes to
meter design or by use of specialised flow
conditioning
• Combined with fast sampling and update rates,
this means that ultrasonic meters can be
calibrated directly using small volume provers
using only a small number of passes per run
THANK YOU
QUESTIONS?