Escolar Documentos
Profissional Documentos
Cultura Documentos
Blaine D. Sawchuk
Dale P. Sawchuk
Danny A. Sawchuk
AGMSC 2010
Page 1
Flow Conditioning and Effects on Accuracy for Fluid Flow Measurement
- American School of Gas Measurement Technology 2010
results are those for E = 0.67. The results for Test 2 are -0.60%
-0.80%
shown in Figure 2. Note that when the flow conditioner is -1.00%
very close to the orifice plate the reading is outside the 1 3 5 7 9 11 13
x/D
limits set forth. However, after the flow conditioner is
moved 5D or so from the plate, the measurement is within
the specification.
Figure 2 - Nova 50E Performance in Good Flow Conditions
GRI MRF Data - Beta 0.67 Test 5 - High Swirl The results are shown in Figure 5.
The flow conditioner meets the requirements 7D upstream
1.00%
of the orifice plate.
Deviation from Baseline Cd
0.80%
0.60%
0.40%
0.20% Tap 1 Figure 5 - Nova 50E Performance - High Swirl
0.00%
-0.20%
Tap 2 GRI MRF Data - Beta 0.67
-0.40%
-0.60%
Deviation from Baseline Cd
-0.80% 1.00%
-1.00%
0.80%
0.60%
0 5 10 15 20 25 30 35 40 45 50 0.40%
x/D 0.20% Tap 1
0.00%
-0.20% Tap 2
-0.40%
-0.60%
Test 3 Two 9(OERZVLQ3HUSHQGLFXODU3ODQHV -0.80%
This test demonstrates the performance of the flow -1.00%
Perpendicular Planes, was repeated. The results are done by Southwest Research at the GRI MRF (3) shows
reported in detail in Reference 5. Results are shown in that this configuration results in much higher uncertainties
Figure 6. than described in the previous version of the standard
Figure 6 - Nova 50E Scalibility Test
(almost 2%).
NRTC Data
Under all of these installations using tube bundles and the
minimum lengths specified in API 14.3, the situation can
1.00
be corrected by installing a CPA 50E flow conditioner in
Deviation from Baseline Cd
B 0.31
0.75
0.50 B 0.41
0.25 B 0.50
place of the tube bundle.
0.00
-0.25 B 0.60
-0.50 B 0.67 Another example is a meter with no straightening vane or
-0.75
-1.00 B 0.75 flow conditioner, downstream of two in-plane elbows. In
0 5 10 15 20 25 this configuration, the previous version of the standard
x/D
requires 17D between the last elbow and the orifice plate.
Again, testing done by SwRI shows that this length of
upstream pipe is not adequate to avoid high uncertainty.
A flow conditioner such as the CPA 50E can also be used
Reynolds Number Sensitivity Results reported by in this case, instead of substantially increasing the
SwRI (3) for testing on 0.67E plates are at Re 2.7 x 106. upstream length required.
Testing performed by NRTC (6) on air at 0.68 E plates at
Re 1.7 x 105 and 1.0 x 105 give results as follows: These examples show how a CPA 50E flow conditioner
can be used to retrofit existing meter runs to meet the new
NTRC Testing on Two Elbows our of Plane on Air at standard.
Low Pressure
New Installations When new meters are being
Beta Ratio Reynolds Deviation from Baseline installed, using a CPA 50E flow conditioner can reduce
Number Cd the amount of upstream pipe required to keep uncertainty
as low as possible. Of course, flow conditioning is only
0.68 170,000 -0.24
one of many factors which affect overall meter
0.68 100,000 -0.08 performance.
8.0 Sensitivity to Velocity Profile Since the transit time ultrasonic meters are measuring the
time difference of a sound pulse traveling on a
Ultrasonic meters are an alternative for turbine meters,
predetermined path in the gas stream, the meter only has
especially for high flow (high capital cost) applications.
one measurement per path. This measurement can only
The meters are similar in how the flow rate is presented to
determine the average velocity along the path. There are
the flow computer, and both have a relatively high
a number of factors which affect this measurement:
turndown ratio. Ultrasonic meters have the additional
advantages of being bi-directional, requiring less 1. The velocity profile of the fully developed flow
maintenance, having less pressure loss across the meter stream, which depends on the Reynolds number,
and being more tolerant of foreign material in the gas roughness, imperfections in the pipe such as welds and
stream. flanges, and other factors.
There are significant differences, however. The turbine 2. How much the velocity profile is distorted
meter is an intrusive device, as such, the turbine meter (including the effects of swirl) resulting from the
(Figure 7) forces the gas into
AGMSC 2010
Page 5
Flow Conditioning and Effects on Accuracy for Fluid Flow Measurement
- American School of Gas Measurement Technology 2010
1.3
Calculations using these slightly varying axial velocity
1.2 profiles show sensitivity to the profile. For these two
1.1 profiles the transit time difference is 0.42% for the same
1
0.9
flow rate.
0.8
0.7
The reason for this difference in transit time is that the
0.6 velocity in the centre of the pipe, is treated the same as
0 0.5 1 1.5 2 the velocity closer to the wall, even though the bulk flow
Radial Coordinate (r=1)
is closer to the wall. Over 60% of the bulk flow is
traveling in the outer 50% of the diameter.
Figure 9 Fully Developed Profile
This result is not sensitive to the model used to calculate
There are several models for the fully developed velocity velocity profile, because all the models for velocity
profile. These are empirical approximations, and are profile have similar characteristics, being empirical curve
somewhat interchangeable. When measurements are taken fits to the same data. Even though this simple model
on flow that is fully developed, there is some difference breaks down at the centre line, other models will produce
between the theoretical profile and the measured profile. very similar results.
This is normal, and reflects some measurement
uncertainty as well as theoretical assumptions. If the
measured profiles are within one or two percent of the 8.2 Asymmetric Velocity Profile
theoretical profiles, they can be considered fully
developed. In ISO 5167 [4], the profile is considered to Yeh and Mattingly have conducted CFD studies on
be fully developed if the measured profile is within of five ultrasonic meter performance in ideal and non-ideal pipe
percent of the calculated profile. flows [5]. They concluded that single path meters have a
high degree of flow profile sensitivity and that
The Power Law is one of these empirical profiles. The performance is strongly dependant upon path orientation
equation is shown as equation 1. to a disturbance (as high as 35% close to an elbow).
(1) These results lead to the further conclusions that
1 installation location and orientation are critical, and that
U y n multi-path meters are desirable for high accuracy.
Ub R Certainly, a 35% shift in meter reading is not acceptable
in current meter applications. The issue for custody
Where: transfer meters is whether the specifications in AGA-9 [1]
are appropriate, and whether a particular installation
U = local axial velocity
meets this specification.
Ub = Bulk or average axial velocity
In order to reduce the effect of this asymmetry, Yeh and
y = distance from the pipe wall Mattingly recommend using multi-path meters. These
meters will significantly improve the response to
R = pipe radius asymmetric profiles, but may not totally compensate for
n = Empirical exponent the profile. Results from meter testing by Grimley show
AGMSC 2010
Page 6
Flow Conditioning and Effects on Accuracy for Fluid Flow Measurement
- American School of Gas Measurement Technology 2010
that some conditions will result in shifts of over one The testing was done to evaluate the effect of the
percent when compared to good flow conditions. disturbance, not the effect of a rotated flow conditioner.
The normal procedure that companies now employ, is to
calibrate the flow conditioner, spool piece and meter
8.3 Changing Velocity Profile
together at the calibration facility. The parts are then
If the meters are sensitive to small changes in velocity disassembled for shipping to the site. At the site, care
profile, they are probably also sensitive to the profile needs to be taken to re-assemble the meter, flow
itself changing within the meter. This is an issue with conditioner and spool piece(s), to ensure that it is the
ultrasonic meters since they are from one to three same as at the calibration facility. Even if this care is
diameters long, and numerous tests have shown that the taken, the reassembly may not be exact, since there is no
velocity profile can change significantly within this comprehensive testing to show how exact the reassembly
distance. needs to be.
In a straight pipe, a distorted velocity profile will
transform itself, moviQJWRZDUGDIXOO\GHYHORSHG
9.0 Testing at Southwest Research
profile. If a profile is close to the fully developed state,
the change will be gradual. An example of a rapidly Grimley [2] tested the meters with a variety of upstream
changing profile is the profile at the exit of a perforated flow conditions, including single elbow, elbows out of
plate flow conditioner (Figure 10). The profile at the exit plane, and elbows in plane. These conditions are similar
of the flow conditioner will be a number of jets, to those found in field locations, but of course, are not
corresponding to the holes in the plate. These jets quickly representative of everything found in the field.
form into one profile. Testing has shown the distance to
Testing done on several ultrasonic meters with different
form one profile is about five pipe diameters.
flow conditioners shows that the user must be careful in
how the meters are installed. Some combinations of
CPA 50E Flow
meter, flow conditioner, and piping configuration were
Conditioner clearly shown to be better at repeating baseline
measurements than other combinations.
Grimley concluded that use of a high performance flow
conditioner was preferable due to the difficulty in
WUDQVODWLQJEDUHWXEHUHVXOWVZLWKRXWIORZFRQGLWLRQHUV
to a field location. These high performance flow
conditioners isolate the meter, to a certain extent, from the
upstream conditions.
~5D
Testing using high pressure gas on these large meters is
expensive, and the results of the testing complete to date
Figure 10 - Profile Downstream of a Perforated Plate
are not conclusive. There are other questions yet to be
addressed, such as: Are the results repeatable? What is
the effect when other variables are altered? How much
Perforated plate flow conditioners accelerate the gas flow
depends on the spool pieces, the exact position and
through holes in the plate. These individual gas streams
orientation of the flow conditioner, and other specific
then recombine to a single stream, which will then move
installation conditions? Are the results the same at higher
toward full development given enough straight, uniform
gas velocity?
pipe.
These results show some variation in performance of the
As the jets recombine, the profile will be changing
ultrasonic meters tested which is not fully understood.
quickly. If the meter is located too close to the flow
While these deviations are generally small in percentage
conditioner, the indicated flow rate could be affected by
terms, they are of concern due to the high volume of gas
these jets, and the bulk velocity may not be calculated
that is generally measured by these meters.
correctly. While it may be possible to calibrate this effect
out of the system, a slight change could significantly alter It should be noted that in the testing carried out at
the indicated flow rate. Southwest Research Institute, the objective was to
determine the effect of changing the upstream profile, and
Meters are currently being installed with flow
the meter remained bolted to the flow conditioner and
conditioners as close as three pipe diameters. There is
spool piece throughout the test. At custody transfer meter
some support for this, however in testing by Grimley, the
stations, the meter is generally unbolted from the flow
meter was not unbolted from the flow conditioner when
conditioner after calibration, and re-assembled on site. It
the assembly was placed in front of the flow disturbance.
AGMSC 2010
Page 7
Flow Conditioning and Effects on Accuracy for Fluid Flow Measurement
- American School of Gas Measurement Technology 2010
will be difficult to re-assemble the meter exactly as it was should take this into account when designing facilities.
assembled at the calibration facility. The following is recommended:
The CPA 50E flow conditioner is designed to recombine 1. High performance flow conditioners should
into a fully developed profile within about five diameters, be used, at a distance from the meter which
in order to minimize this effect. Even if the flow will allow the jets to fully re-combine, so
conditioner, spool piece, and meter is disassembled after that the velocity profile at the meter will be
calibration, the shift should not be expected to be large. If stable.
the opportunity to keep the meter run bolted together is
2. Meters should be calibrated with the flow
available, it may be best to leave it this way until
conditioners and spool pieces which will be
installation. This option is of course not possible with a
used in the meter station.
136RUPHWHUUXQ
3. Manufacturers minimums for upstream
10.0 Verification Techniques
lengths should be maintained or exceeded.
There are techniques which will improve the users A minimum of 8 diameters should be
confidence that the metering installation is working between the flow conditioner and the meter.
properly. These include meter diagnostics and check
4. Spool pieces should be finished with welds
measurement.
ground out and perhaps internally coated.
10.1 Meter Diagnostics Diameter of the spool piece should closely
match the meter diameter.
When the meter is calibrated, a log file from the meter is
available which shows such things as the individual path 5. Flow conditioner orientation at the
velocities, speed of sound, gain, and other information. calibration facility should be marked and re-
Users should acquire this data from the calibration established.
facility, and gather some similar information when the
6. Piping upstream of the meter run should be
meter is installed and measuring gas. Differences,
designed not to produce extreme velocity
especially in the path velocity ratio, indicate that the
profiles, swirl or noise.
velocity profile is not the same as when calibrated. This
likely means that there is a shift in meter performance. The installation might me slightly more expensive, but
there will be more confidence in the measurement.
10.2 Check Measurement
11.1 Diagnostics
Conservative design suggests that there be a mechanism
to verify or trend the measurement. Ultrasonic meters can Diagnostics should be employed when the metering
provide some redundancy if there is more than one path, installation is commissioned. These include log file
and there is sometimes other measurement to trend evaluation and check measurement (if available).
against the meter. However, in some cases operators may
want some other verification.
While installing an additional meter may not be
considered cost effective, it is possible to use the flow
conditioner as a check meter. Testing has been done with
the CPA 50E flow conditioner with differential taps
drilled on the upstream face for stagnation pressure, and
in the wall of one of the holes for downstream pressure.
This measurement has been shown to be very stable and -
repeatable.
The data can be collected using a standard flow computer
setup. In order to reduce data handling requirements, the
data can be trended and saved. It only need be retrieved
and analyzed if there is a change compared to the
ultrasonic meter.
12.0 Conclusions
References
The most critical components of the meter station are the
1. Measurement of Gas by Multipath
meter, flow conditioner and upstream piping
Ultrasonic Meters, AGA Transmission
configuration. In terms of the overall capital cost of the
Measurement Committee Report No. 9,
facility these components do not necessarily get the
American Gas Association, June 1998
attention they deserve.
2. *ULPOH\7$8OWUDVRQLF0HWHU
Small percentage cost savings efforts aimed at the meter,
,QVWDOODWLRQ&RQILJXUDWLRQ7HVWLQJ$*$
meter, flow conditioner, meter run and upstream piping
2000 Operations Conference, May 2000,
configuration are dangerous. Especially considering the
Denver, Colorado
life span of a typical meter station is 20 years.
3. .DUQLN80HDVXUHPHQWVRIWKH
Measurement bias cost over this capital life span can be
Turbulence Structure Downstream of a Tube
substantially larger than the initial savings obtained.
%XQGOHDW+LJK5H\QROGV1XPEHUV
If there is no conclusive testing data to substantiate cost Transactions of the ASME, Vol. 116,
saving efforts in these critical areas it is best not to chance December 1994
it.
4. ISO 5167, Measurement of Fluid Flow in
Circular Cross Section Conduits Running
Full
1. Put thought into the upstream meter station
piping to minimize loading of the flow 5. <HK77DQG0DWWLQJO\*(&RPSXWHU
conditioner and meter by minimizing flow Simulations of Ultrasonic Flow Meter
disturbing designs. Performance in Ideal and Non-Ideal Pipe
IORZV, 1997 ASME Fluids Engineering
2. Ensure that the meter run is not allowed to
Division Summer Meeting, June, 1997
become the last item of interest in the bill of
6. API 14.3/AGA 3 Part 2, 1991
materials. Roughness, roundness, length and
overall fabrication quality is important until 7. API 14.3/AGA 3 Part 2, 1999
testing proves otherwise.
8. Morrow, T . 1997 Gas Research
3. Work closely with your meter and flow Institute/Southwest Research Institute
conditioner vendors. 7HFKQLFDO0HPRUDQGXP'HYHORSPHQWRID
)ORZ&RQGLWLRQHU3HUIRUPDQFH7HVW
4. If at all possible leave the meter run assembly
bolted together. 9. Studzinski, W., Karnik, U., LaNasa, P.,
Morrow, T., Goodson, D., Hussain, Z. and
5. 'RQWFKDQJHSLSHURXJKQHVVWKURXJKWKHPHWHU
*DOODJKHU-*5,UHSRUW:KLWH3Dper
run and meter.
on Orifice Meter Installation Configurations
6. Use the diagnostics the meter manufacturers ZLWKDQGZLWKRXW)ORZ&RQGLWLRQHUV
supply.
10. .DUQLN8DQG.RZFK56FDOH8S
7. Use the CPA 50 E flow conditioner as a Tests for the Nova Flow Conditioner for
secondary measurement device. It will 2ULILFH0HWHU$SSOLFDWLRQVth International
drastically increase the confidence level of the Symposium of Fluid Flow Measurement,
primary measurement. Denver, Colorado
11. .DUQLN8$&RPSDFW2ULILFH
0HWHU)ORZ&RQGLWLRQHU3DFNDJHrd
International Symposium of Fluid Flow
Measurement, San Antonio, Texas
AGMSC 2010
Page 9