You are on page 1of 16

REVISION OF ANSI-2530 ORIFICE METER STANDARD

- BASIS OF RECOMMENDED METER RUN LENGTH

W. Studzinski, U. Karnik P. La Nasa


NOVA Research & Technology Centre, Canada CPL & Associates, USA

T. Morrow D. Goodson
Southwest Research Institute, USA Daniel Measurement, USA

Z. Husain J. Gallagher
Texaco, USA Savant Measurement, USA

ABSTRACT

The latest revision of the ANSI/API-2530 orifice meter standard introduces significant
changes in order to improve the performance of the orifice meter when subjected to non-ideal
flow conditions. In the case of bare tube installations, existing recommendations have been
changed and additional recommendations have been provided for new types of pipe fittings. The
major finding was that users of orifice meters have to be particularly careful in selecting meter
run lengths for installations which generate flow swirl, for example, headers or elbows out of
plane.
For applications using flow conditioners several revisions have been made. The use of
the 19 tube bundle flow straightener has been subjected to several new restrictions. To obtain
optimal performance from an orifice meter for a given application, the flow straightener has to be
used in accordance with recommendations for its locations, meter tube length and orifice β-ratio.
Moreover, the new tube bundle flow straightener has to meet defined mechanical specifications
and tolerances to deliver the expected performance. The standard also provides a performance
test procedure, which would allow the use of other flow conditioners, for example, isolating type
flow conditioners.
Specifications for orifice meter run lengths have been revised by using high quality data,
evaluated by well defined rules within the framework of practical considerations. These changes
will result in a metering device, which is more accurate, repeatable and reliable and can be
considered on par with other metering technologies. The present paper describes the process,
which was used to accept data and develop new recommendations for the orifice standard. It
provides the necessary background information, which can help users understand the reasons
behind the proposed changes and demonstrates that the recommendations indeed result in a much
improved orifice meter performance.

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
1.0 INTRODUCTION

Orifice meters are still the most widely used flow meters in North America. Despite the
introduction of new technologies, such as ultrasonic meters, there is still an opportunity to
improve orifice meter performance and use it well into the next century. This improvement
could be brought about by simple and relatively low cost modifications. Users of the orifice
meter have invested several millions dollars to improve the discharge coefficient equation in the
80’s. The most significant areas left for improvement were meter run length for various piping
installations upstream of the orifice plate and the performance of flow conditioners. There was a
desire to have a performance based standard, which would guarantee the users a measurement
uncertainty comparable to the turbine and ultrasonic flow meters. This required the development
of well defined performance criteria for bare tube installations and flow conditioners.
The development of a revised orifice meter standard was driven by several objectives.
The primary factor was to ensure that implementation of the recommendations resulted in no
additional uncertainty to the orifice discharge coefficient equation. It was expected to achieve
the performance objectives at low cost. The users did not want to introduce mechanical
tolerances for elbows and other fittings used upstream of the orifice meter. A significant number
of users wanted to use tube bundle flow straighteners without adding any additional uncertainty
to the discharge coefficient equation. There was also an increasing interest in isolating flow
conditioners. Thus, the expectation was to have a performance based standard, which would
define all conditions necessary for meter tube design, the position of a flow conditioner and the
orifice plate so as to provide the lowest possible uncertainty.
The expectations were high and hence, the recommended meter run length for bare tube
installations is, in many cases, longer than that in the previous orifice standard. Also, the
application of a tube bundle type flow straightener is more restrictive. The user, however,
obtains a device which performs much better than within the old standard specifications. The
orifice meter has been resurrected and with a relatively modest upgrade it can still match the
performance of the newest technologies.
The purpose of the present paper is to provide background technical information which
has led to the revisions of the orifice standard. It will also explain the approach used to convert
the technical conclusions into standard recommendations and show that the obtained solution is
optimal under the given conditions. The revisions of the ANSI/API-2530 were conducted by the
API Chapter 14.3 Part 2 Working Group led by Ron Beaty. The mandate of the White Paper
Working Group was to develop a methodology for the screening of the available data and the
preparation of recommendations for the standard revisions. The White Paper recommendations
were subsequently converted by the Working Group into the revised orifice standard.
It is worthwhile noting that researchers and users of orifice meters had certain
expectations of the behaviour and trends that should be observed with specific installations. Such
preconceptions were not always congruent with the experimental evidence collected later.

2.0 PROCESS USED TO DEVELOP NEW STANDARD

The development of a revised standard resulted from the screening of available data as
well as the generation of new data by planning and conducting the necessary experiments. These
experiments were performed during the process of developing the White Paper and limited
funding did not permit the testing of all possible field installations. Several processes and

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
recommendations were proposed and re-visited as new data became available. This iterative
process was adopted until the proposed recommendations were defensible by data or the
proposed scheme. The main components of the process were as follows:

1. Definition of the revision scope, objectives and necessary experimental program.


2. Execution of the experimental program.
3. Development of data acceptance criteria and performance criteria.
4. Selection of acceptable data according to the criteria.
5. Development of the least conservative recommendations.
6. Aggregation of recommendations into a set of relatively compact standard rules
covering broad range of parameters.
7. Balloting and approval of the standard revisions.

The overall process was conducted by the entire API Chapter 14.3 Part 2, Working Group
and parts of it were delegated to the White Paper Writing Groups as well as research
laboratories. It took over 6 years to get the final product- the revised ANSI/API-2530 orifice
meter standard.

3.0 DATA ACCEPTANCE AND PERFORMANCE CRITERIA

The data acceptance criteria and the procedure to determine the recommended meter run
length were established up front to ensure technical defensibility of the recommendations that
were developed. Several meetings of the White Paper Writing Group were held and various
options were explored until finally an optimum choice was made. The data, which met all the
acceptance criteria, were obtained from twelve independent test facilities. Several thousands of
data points were used, in accordance with the established “rules”, to develop the revised
installation and the 19-tube bundle flow straightener recommendations.

Data Acceptance Criteria


• Only tabulated data would be used as opposed to interpreting the data from figures in
publications.
• In all cases, the original authors were contacted to provide the tabulated data.
• The data would be considered only if the test facility baseline data was within the
uncertainty limits (±2σ) of R-G equation in good flow conditions and/or
demonstrated long term stability of the baseline within ± 0.20%.
• The installation effects ∆ CD were calculated in reference to the facility baseline for
the same Reynolds number ReD and orifice β- ratio, from the following relationship.

∆CD [%] = 100×(CD install – CD baseline)/CD baseline

• For a given installation and orifice β - ratio, at least two data points originating from
one test facility were needed for data acceptance. This condition helped to assess, at
least, a basic data trend.

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
Rules for Determining Meter Run Length
• The recommended meter run length for bare tube installation and the position of the
19 tube bundle flow within the meter run had to be determined such that that it would
not add to the uncertainty of the R-G equation.
• A “tolerance band”, defined as ± σ of the R-G equation (half of the R-G equation
uncertainty), was established for ∆CD values.
• The magnitude of the tolerance band was based on the repeatability of considered test
facilities, bell shape type distribution of measured ∆CD as a function of orifice β -ratio
and an estimation of common uncertainty components of the R-G equation under
ideal flow conditions and in the installation tests.
• A larger tolerance band than the above may result in an additional uncertainty (or
bias) due to pure installation effect whereas a smaller tolerance band would not be
technically achievable.
• Values of ∆CD within this tolerance band were considered as “practical zero”
deviation.
• Some of the accepted test facilities collected limited amount of data manually, quite a
few facilities used data acquisition systems. It was decided that each facility should
have equal impact on the data trend and therefore data from the same pressure tap
location were averaged at each facility.
• Data from at least two independent facilities with at least a partial overlap of an
independent parameter were required to determine installation recommendation.
• The minimum length for the bare tube installations was determined by the first set of
∆CD data points fully within the tolerance band.
• Curve fit of data for bare tube installations was not performed for the following
reasons:
a) This approach did not always lead to a conservative solution
b) Almost all bare tube data were asymptotic to the zero line. Hence, a curve fit
would result in infinite lengths for the meter run.
c) Interpolations were not used for bare tube installations where the accepted
data were relatively sparse because of a few meter run length tested.
• Extrapolation was used when the tested meter run lengths were not long enough to
obtain ∆CD within the tolerance limit ± σ of the R-G discharge coefficient equation.
• The data obtained in the performance tests of 19-tube flow straightener were very
closely spaced. Further, the so called “crossover point” did not have any asymptotic
character with the zero line. Therefore, interpolations between data points were
allowed and used if needed.

4.0 APPLICATION OF THE PROCESS TO DETERMINE METER RUN LENGTH


In most cases, the process could be applied quite easily. However, for some cases,
practical considerations and physical reasoning had to be utilized in conjunction with the
established rules to make the most optimal recommendations. These cases are discussed in the
following.
Bare Tube Installations With Swirl
The recommended revisions for the meter run length are much more significant when
swirl is present. Table 1 summarizes all bare tube installations. It was proven (Mattingly and

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
Yeh, 1991, Williamson et al 1993, Reader-Harris et al 1995, Karnik et al 1995) that at least three
types of pipe fittings generate swirl
i.) two 90° elbows in perpendicular planes with the spacer length up to 15D
Table 1 specifies length for two 90° elbows out of plane with various spacer length.
Installations with spacer length greater than 15D, were grouped with results from a single 90°
elbow. The reasoning was that, ultimately, results with very long spacers should not differ from
those with single 90° elbow.
ii.) various types of headers and
iii.) combination of a 90° elbow with an expander and larger diameter meter run.
The recommendations for headers and the combination of an elbow with an expander
were covered in a “catch all” type column. There was not enough data from two independent
facilities to create separate recommendations for each installation. This column basically
describes the worst case scenario and was developed using header data from Ruhrgas and NEL
as well as high swirl data (using the Chevron swirler) from GRI MRF. It should be noted that
the recommendations for an expander cover only an expander in the straight line and not in a
combination with an elbow.
Recommendation for two 90° elbows in perpendicular planes with a spacer
shorter than 5D:
The recommended meter run length for this installation is significantly longer than in the
previous standard. This recommendation can be explained by examining Figure 1 taken at
β =0.50. If one applies the governing rules i.e. recommend the position at which the first data
set falls within the tolerance bounds, then one would recommend 49D as the meter run length.
The GRI MRF data shows that at 45D, the data exhibits a significant bias in ∆ CD =+0.6%.
Although, GRIMRF do not have a data at 49D, it is improbable that it could be within the
acceptance limit at this location. Extrapolation was the only available choice and indicates the
need for a 95D meter run length at β =0.5.

2.4
GRI MRF[19] Tap 1 (aug 97)
β=0.5 Daniel[16] Tap 3:00
Daniel[16] Tap 9:00 GRI MRF[19] Tap 2 (aug. 97)
2.0
Daniel[16] Tap 12:00 NOVA[12] Tap 3:00
AGA specified location GRI MRF[14] Tap 1
1.6 NOVA[12] Tap 9:00
GRI MRF[14] Tap 2
NIST-G[8]
1.2
∆CD [%]

0.8

0.4

0.0

-0.4

0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.


L2/D

Figure 1. Effect of Two Elbows Out of Plane (Spacer ≤ 2D) (Without Flow Conditioner) on
Orifice Meter; β=0.5; L2 is the distance between the disturbance and the orifice meter.

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
Similar positive biases of ∆ CD around +0.6% were observed by GRI MRF at 45D and
β -ratio 0.60 (Figure 2). The White Paper data showed that for β =0.75, the meter run length
could be 45D. However, data taken at Mannheim (Pecornik & Zimmermann, 1996) for this β-
ratio indicates that the meter run length could be around 62D. Although this data was taken for
corner taps, the data could be used for interpretation of flange tap results since for the same
installation there is good agreement with flange tap data for β =0.6 as seen in Figure 2. For all
these reasons a conservative approach was taken and 95D meter run length was recommended
for all β ≥ 0.50.

2.0
β=0.6 AGA specified location
Mannheim (2D spacer) : corner taps
1.6 Mannheim (1D spacer) : corner taps

1.2
0.8
0.4
∆CD [%]

0.0
-0.4
[19]
GRI MRF Tap 1 (aug 97)
-0.8 GRI MRF
[19]
Tap 2 (aug. 97) Daniel
[16]
Tap 3:00
[16]
[12] Daniel Tap 9:00
-1.2 NOVA Tap 3:00
Daniel
[16]
Tap 12:00
[12]
NOVA Tap 9:00 [14]
GRI MRF Tap 1
-1.6 NEL
[8]
(no spacer) GRI MRF
[14]
Tap 2

-2.0
0.0 20.0 40.0 60.0 80.0 100.0
L2/D

Figure 2. Effect of Two Elbows Out of Plane (Spacer ≤ 2D) (Without Flow Conditioner) on
Orifice Meter; β=0.6.

Another interesting feature of this installation was the fact that the NIST and GRIMRF
data show different trends in comparison to that from NOVA and Daniel. In the case of NIST
and the GRIMRF, the data shows a positive sign for the ∆ CD bias for short meter runs whereas
for the other two facilities the bias is negative. Although not intuitive, these differences can be
explained. It is known (Morrison et al., 1994) that the orifice meter bias caused by elbows in
perpendicular planes results from two separate effects. The positive change of ∆ CD is caused by
an interaction of swirl with the orifice and conservation of angular momentum. This effect is
stronger at smaller orifice openings (lower β -ratio). The negative change of ∆ CD is caused by
the non-symmetry of velocity profile generated by two elbows out of plane and is stronger at
high β -ratio. The net effect in terms of the ∆ CD sign is a result of interplay between both
phenomena. Thus, the sign of the ∆ CD bias depends on the velocity field and the swirl
component.
The intensity of swirl generated by this installation depends on the length of the spacer.
The strongest swirl angle (18°) was measured by Mattingly and Yeh, 1991 for zero spacer
length. The rate of decay of swirl depends on meter run roughness (Mottram and Rawat 1986),
Reynolds number (McManus et al 1985) and velocity distribution itself (Steenbergen and
Voskamp, 1998). The degree of swirl, probably also depends on the geometry/mechanical

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
specifications of the elbows used in the experiments, which were not quantified because of cost
considerations in practical applications.
Thus, it is feasible that the trends for this installation can differ, as observed in the present
experiments. Further, it is also possible that for two similar trends, as in the case of the GRI
MRF and NIST Gaithersburg, the magnitude of the bias in ∆ CD could be different. It also
explains the fact that the worst effect is not seen at a β=0.75 but at a β -ratio of 0.5.
Considering the number of factors that could affect the sign and magnitude of the bias, its
prediction was not obvious by the physical description of the installation. Therefore the most
conservative data trend had to be used for recommendations.

0.8 β=0.5

0.4

0.0
∆CD [%]

-0.4
AGA specified location

-0.8

-1.2

-1.6 GRI MRF[19] Tap 1 (aug 97)


GRI MRF[19] Tap 2 (aug. 97)

-2.0
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0
L2/D

Figure 3. Effect of Two Elbows Out of Plane (Spacer = 5D) (Without Flow Conditioner) on
Orifice Meter; β=0.5.

0.8 β=0.5

0.4

0.0
∆CD [%]

-0.4
AGA specified location
-0.8

NOVA[12] Tap 3:00-10D


-1.2
NOVA[12] Tap 9:00- 10D GRI MRF[19] Tap 1 (aug 97)- 15D
Daniel[16] Tap 3:00- 10D GRI MRF[19] Tap 2 (aug. 97)- 15D
-1.6 Daniel[16] Tap 9:00- 10D GRI MRF[19] Tap 1 (aug 97)- 10D
Daniel[16] Tap 12:00- 10D GRI MRF[19] Tap 2 (aug. 97)- 10D
-2.0
0.0 10.0 20.0 30.0 40.0 50.0
L2/D

Figure 4. Effect of Two Elbows Out of Plane (Spacer ≥ 10D) (Without Flow Conditioner)
On Orifice Meter; β=0.5.

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
For this particular installation, use of the previous standard may lead to biases up to 1%
at β =0.5 and –0.4% at β =0.75. Moreover, the bias at β =0.5 may have any value between 0
and 1%, it depends on the specific make of elbows. Meter run lengths can be reduced from 95D
to 44D by merely increasing the length of the spacer to be ≥5D. The intensity of generated swirl
and the orifice measurement error drops dramatically for longer spacers as seen in Figure 3 and 4
for β =0.5. Although some peculiarities of data trends still resemble the short spacer case, the
magnitude of biases drops much faster with increasing meter run length. Use of meter run length
from the previous standard may result in a bias up to ± 0.3%.
Recommendation for headers and any unspecified installation:
There are a variety of header designs used for meter stations. Some of the designs may
generate swirl angles over 30° (Reader-Harris et al, 1995). The development of a universal
recommendation for any type of header was based on two specific header tests conducted by
Ruhrgas and NEL. Unfortunately the meter run length used in the tests was too short to have the
∆ CD data within the acceptance limits and resulted in a large positive bias. It indicated presence
of strong swirl. To obtain a rational shape of the extrapolation curve, a decision was made to
combine GRI MRF data on the effects of swirl generator with the header data. Additionally, the
data for other installations such as closely coupled two 90° elbows out of plane and a partially
open gate valve were used to develop a recommendation which covers headers and any other
installation which is not specified in the standard. The use of meter run length from the previous
standard (catch all Fig.) may result in a bias up to ± 5%.

Bare Tube Installations Without Swirl


There are several piping installations, which do not generate single swirl but rather two
counteracting and quickly dissipating eddies like from a single 90° elbow or purely non-
symmetrical profile such as from a partially open valve.
In the case of a gate valve (at least 50% open) and expander or reducer, the accepted data
were available only from a single test facility. Since the data did not contradict the previous
standard, it was decided that there is no evidence to suggest the need for new recommendations.
However, the user has to ensure that there is no swirl entering to these fittings.
Single 45° and 90° elbows, two elbows connected by long spacers and 90° Tees were
covered in the previous orifice standard by one recommendation for partially open valves. There
is now sufficient amount of data to develop separate recommendation for each of these piping
installations. Although there are separate recommendations for these fittings, the new standard
requires 44D meter run length at β =0.75. The previous “catch all Figure” (partially open valve)
recommended the same length for this β -ratio. Frequently, meter runs were designed for this
length to maximize the use of the β -ratio range. Therefore the revision to the standard may
have minimal practical impact on the designs of existing meter runs.
The analysis of data in the White Paper (1997) indicates that, if a meter run was designed
for a β =0.6, with the use of the previous recommendation for the single 90° elbow, it may cause
a bias up to –0.4%. However, for a similar situation there would be no bias for a 90° Tee.
The most significant changes of meter run length were needed for two 90° elbows in the
same plane (S-configuration) with a short spacer as seen from Figure 5. The recommended
length almost tripled and the use of the previously recommended meter run length may lead to
bias errors up to –1.0%. On examining Figure 5, one is tempted to recommend a length of 30D,
which is twice the previous recommendation. However, on examining the data from British Gas,

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
it does not appear that the bias would be within the tolerance limits for this meter length.
Hence, the more conservative length of 44D was chosen.

0.8 AGA specified location β=0.6

0.4

0.0
∆CD [%]

-0.4

-0.8 Gaz de France [8] (no spacer)

British Gas [8] (no spacer)


-1.2
NOVA[12] Tap 3:00
NOVA[12] Tap 9:00
-1.6 NOVA[12] Tap 6:00
NOVA[12] Tap 12:00
-2.0
0.0 10.0 20.0 30.0 40.0 50.0
L 2/D

Figure 5. Effect of Two Elbows in Plane (Spacer ≤ 2D) (Without Flow Conditioner) On
Orifice Meter; β=0.6.

The single 45° elbow and two 45° elbows in the same plane require longer meter run
length at lower β -ratio than in the previous standard, however the potential bias with the old
length does not exceed –0.3%.
The weakest installation recommendations are probably for the expander. It cannot be
excluded that in some installation flow separation could occur and swirling flow may generate
larger measurement uncertainty. However, there is no data to support or reject this speculation
and hence, the recommendations for this installation have not been revised.

Installations With 19 Tube Bundle Flow Straightener


The 19 tube bundle flow straightener is commonly used in industrial installations. The
data collected indicate that this device did not perform as well as expected. The performance of
19 tube bundle flow straightener depends on the type of upstream pipe fitting, the location of the
straightener along the meter run, the length of meter run and geometric characteristics and
tolerances of the tube bundle. The 19 tube bundle flow straightener can be used at a relatively
narrow range of locations within the meter run in particular at the higher end of the β -ratio
range. Additionally, the device has to be built according to specified mechanical tolerances and
it is possible that a significant number of the old tube bundles will not meet this specification.
Specifically, the tube bundle has to be concentric, with specified sizes and tolerances on tube ID,
tube wall thickness, open area between individual tubes and gap between tube bundle OD and
the meter tube ID. Therefore, it was decided to stress the fact that it is basically a new flow
straightener and call it the 1998 Uniform Concentric 19 Tube Bundle Flow Straightener.
The new flow straightener can be used downstream of single 90° elbow or two 90°
elbows in perpendicular planes with a spacer 2D and shorter, single 90° Tee and partially open

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
(at least 50%) valves in meter runs 29D and longer over the range of β -ratio up to 0.75. For any
other unspecified fitting it can be used up to β =0.67 in the 29D meter runs and longer. The
main restriction is the location of flow straightener, which depends on the installation type and at
the highest allowed β -ratio the location is somewhere between 11D and 16D (see Table 2 for
details).
Use of shorter meter run length between 17D and 29D restricts the maximum β -ratio
allowed. For example a single 90° elbow can still be used up to β =0.75 ,however, for a single
90° Tee the maximum β -ratio is 0.50, a partially closed valve is restricted up to β =0.47 and
any other installation not covered by the standard can operate only up to β =0.46. The optimal
location of the tube bundle depends again on the type of the upstream fitting and varies between
9.5D and 14.5D from the orifice plate.
The location of the 19 tube bundle flow straightener at 5D to 7.5D from the orifice plate
as recommended in the previous standard may lead to the following biases:
- single 90° elbow, up to –1.0%
- two 90° elbows in perpendicular planes, up to –1.0%
- single 90° Tee, up to –0.8%
- partially closed valves, up to –0.8%
- any fitting, up to –1.0%
All evaluations of the expected bias level are based on the worst case scenario, which for
the installations with the 19 tube bundle flow straightener is usually at β =0.75. This is not the
case for the bare tube installations where the worst case was observed even at β =0.4.

5.0 PERFORMANCE TEST FOR FLOW CONDITIONERS

The development of several isolating type flow conditioners initiated the need for a
general test, which would examine their performance in light of the defined performance criteria.
After several discussions mainly centered around how the general test would meet the “real
world” practical applications, the following 5 tests were developed. It was important that tests
not be too lenient or unrealistically harsh.
a) The Test Facility Baseline Evaluation
This test was necessary to meet the data acceptance criteria as specified in Section 3.
Further, a baseline evaluation performed at the time of the installation test eliminates any “bias”
that may result from using historical baseline data. Deviations of the discharge coefficient (∆CD)
should be with respect to this baseline.
b) Ideal Flow Conditions (Fully Developed Velocity Profile)
This test was to evaluate whether the flow conditioner distorts an ideal velocity profile.
Obviously, if this were a practical situation, one would not use a flow conditioner. However, it
is a necessary test to ensure that if the disturbance is not “strong enough” does the flow
conditioner itself produce an installation effect.
c) 50% Open Gate Valve (Non-Symmetrical Velocity Profile)
This test was to evaluate whether the flow conditioner can condition a velocity profile,
which has mean flow non uniformity i.e. skewness in the mean velocity profile but with the
absence of swirl. A practical situation may be a single elbow or two elbows in plane with a long
spacer between two elbows. There may also be other situations, which have not yet been
documented. Obviously, all these practical situations cannot be tested without incurring a
penalty i.e. cost. Thus, this test was thought to be appropriate.

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
d) Swirl Generator (Strong, 25° Swirl)
This test was to evaluate whether the flow conditioner can condition a velocity profile,
which has a strong swirl component with minimum mean flow non-uniformity. Practically,
results of this test can be applied to situations such as headers. In some cases (Reader-Harris et
al., 1995), swirl angles of up to 25 degrees have been measured.
e) Two 90° Elbows In Perpendicular Planes (Combination Of Weak Swirl And Non-
Symmetry)
In some practical situations, there may exist flow profiles, which consist of a combination
of mean flow non-uniformity and swirl. Typically, in such cases the swirl is weaker. This test
simulates such a case.
The above tests can be used to obtain two types of approvals for use of the flow
conditioner with any upstream installation.
1. Application Test: By conducting the specified 5 tests at one line size and a narrow range
of Re number associated with the tested β-ratio range and differential pressure range used,
approval can be obtained for this limited application.
2. Type Approval Test: This is an approval for any line size and any Reynolds number. To
obtain this approval, tests should be conducted using the ranges for Reynolds numbers and
line sizes as
(2.1) 104 < Re < 106 and Re ≥ 106; the two selected Reynolds numbers must fulfill
certain specified conditions.
(2.2) D ≤ 4” and D ≥ 8”.
The following tests must be conducted to obtain “Type Approval”.
I. Disturbance test : The above 5 tests at one pipe diameter and a narrow Reynolds
number range selected from (2.1) and (2.2). This test is similar to the Application
Test. This test ensures that the flow conditioner can handle any disturbance for a
given pipe size and a narrow range of Reynolds numbers.
II. Scaling Test : Conduct test a) and any one of test b) to e) at two pipe sizes. It is
preferred that one of the pipe sizes be the same as that used in I. The Reynolds
number for this test should be the same for both pipe sizes. The intent of this test is to
ensure that at the same Reynolds number, the flow conditioner is able to perform at
different line sizes.
III. Reynolds Number Sensitivity : Conduct test a) and any one of tests b) to e) at one pipe
size (preferably one of the pipe sizes used in II) and at two Reynolds numbers such
that one is selected from the range Re ≥ 106 and the other is from the range 104 < Re <
106. The ratio of the two selected Reynolds numbers is a function of the friction
factor and the pipe wall roughness. It is preferred that one of the chosen Reynolds
number should be the one used in I or II. The intent of this test is to ensure that, for a
given pipe size, the flow conditioner performs for two distinctly different Reynolds
numbers i.e. there is a definite difference in the velocity profile and hence a
measureable difference in the discharge coefficient. If the application of the flow
conditioner is exclusively for Re ≥ 3x106 then this test need not be performed.
The preferences mentioned in II and III are so that there exists an overlap between the
three tests and if necessary, comparative conclusions can be drawn from the tests.
The flow conditioner can be placed at one or several positions within the meter run
length. The approval for use can be obtained only when the performance results have met
acceptance criteria. In an attempt to reduce experimental costs, specific testing patterns have

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
been suggested. For example, for two elbows in perpendicular planes or strong swirl, test should
be performed first for β=0.40 and β=0.67. If the ∆CD values for both β-ratio are negligible or if
∆CD varies approximately as β 3.5 then swirl is insignificant. In this case it is recommended that
the other flow conditioner performance tests be performed for a single value of β =0.67. If the
flow conditioner passes the test for β =0.67, it can be also tested at higher value of β if desired.
If swirl effects are present at β =0.40 and β =0.67, then both tests will have to be performed for
a complete range of β -ratio values between β =0.20 and β =0.75.

7.0 CALIBRATION TEST OF AN INSTALLATION

In practical applications space and/or cost limitations may prohibit the use of meter run
lengths recommended in the catch all column. This calibration test (performed on the actual or
an appropriately scaled line size) can be used for the specific case in question. It can be used for
a specific installation with or without flow conditioner. The performance of this test will also
generate data for a non-standard installation to be used for future revisions of the standard.
In the case of a installation test performed on a scaled line size, the test facility has to
conduct the baseline calibration at the same line size, Reynolds number and β-ratio range as in
the installation test. The baseline data has to be within the uncertainty of R-G equation. If the
∆ CD data are within one half of the uncertainty ( ± 2 σ ) of the R-G equation, the user can use the
R-G equation in the field measurement. In the opposite case, the calibration curve obtained has
to be applied to field measurement on this installation.

7.0 CONCLUDING REMARKS

Several changes have been made to the existing standard for orifice meters. The changes
to the specifications for orifice meter run lengths have been made by using high quality data and
well defined rules to evaluate them within the framework of practical considerations. It is clear
that the changes will result in a metering device, which is more accurate, repeatable and reliable.
It can be considered to be on par with other technologies such as turbine meters and ultrasonic
meters.
The entire process has taken a while but this has allowed the White Paper Writing Group
to carefully evaluate existing data, plan experiments for additional data and deliberate over the
recommendations to ensure that every possible impact of making these recommendations on
current and future users could be examined.

8.0 ACKNOWLEDGEMENTS

The authors would like to acknowledge the efforts of the various research groups at the
experimental facilities for providing their tabulated data and for executing the experimental
program proposed by the White Paper Writing Group. We also acknowledge API Working
Group for their permission to publish this work. Special thanks to Dr. Michael Reader-Harris of
NEL for his numerous discussions and efforts to pursue a common ISO-API orifice standard.

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
9.0 REFERENCES

1. Manual of Petroleum Measurement Standards, Chapter 14 – Natural Gas Fluids


Measurement, Section 3 – Concentric Square Edge Orifice Meters, Part 2 – Specification
and Installation Measurements (1991, 1999).

2. Studzinski, W., Karnik, U., LaNasa, P., Morrow, T., Goodson, D., Husain, Z., Gallagher,
J., (1997). White Paper on “Orifice Meter Installation Configuration with and without
Flow Conditioners”, API confidential report, to be published.

3. Mattingly, G. E., Yeh, T. T., (1991). Effects of Pipe Elbows and Tube Bundles on
Selected Types of Flowmeters, Flow Meas. Instrum., 2(1): 3-13.

4. Willamson, I.D., Botros, K.K., and Price, G. R., (1993). Flow Characteristics and
Orifice Meter Error Caused by Upstream Headers in Multirun Meter Stations, ASME
Fluids Engineering Conference, Washington, D.C.

5. Reader-Harris, M.J., Woodhead, E., Sattary, J., McEwen, D., (1995). NEL Report on
Flow Conditions Downstream of Headers, report of Header Consortium.

6. Karnik, U., and Williamson, I., (1995). Effect of Expanders on Meter Accuracy, 3rd
International Symposium on Fluid Flow Measurement, San Antonio, TX, U.S.A.

7. Mottram, R.C., Rawat, M. S., (1986). The Swirl Damping Properties of Pipe Roughness
and the Implications for Orifice Meter Installation, Proc. Of Int. Conf. On Flow
Measurement in the mid 80’s, NEL.

8. McManus, S.E., Bateman, B.R., Brennan, J.A., Vasquez, I.P., Mann, D., (1985). The
Decay of Swirling Gas Flow in Long Pipes, Proc. of AGA Operating Section 85-DT-54:
629-633.

9. Steenbergen, W., Voskamp, J., (1998). The Rate of Decay of Swirl in Turbulent Pipe
Flow, Flow Meas. Instrum., 9(2): 67-78.

10. Morrison, G.L., Hall, K.R., Macek, M.L., Ihfe, L, M., DeOtte Jr., R.E., Hanglie, J. E.,
(1994). Upstream Velocity Profile on Orifice Flowmeters, Flow Meas. Instrum., 5(2):
87-92

11. Pecornik, D., Zimmermann,H., (1996). Versuche über den Einfluss einer drallbehafteten
Strömung auf die Durchflusszahl α einer Normblende in Abhängigkeit vom
Durchmesserverhältnis β . Projekt Nr. 06/92, Fachhochschule Mannheim.

4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999
4th International Symposium on Fluid Flow Measurement

Straight unobstructed meter tube length from the upstream side of the orifice plate (in multiples of internal pipe diameters)
Diameter a. Single 90o Two 90o elbows in Two 90o elbows in Two 90o elbows Two 90o elbows Single 90o a. Single 45o Gate Valve Concentric Any other Downstream
ratio elbow. the same plane the same plane, in perpendicular. in perpendicular Tee used elbow. at least a. Reducer configuration Meter Tube
β b. Two 90o elbows “S” configuration “S” configuration planes, planes, as b. Two 45 50% open b. Expander (catch all Length
o
in the same plane spacer 10D < S ≤ 30D S < 5D 5D≤S≤15D an elbow elbows in the category)
with S > 30D. S ≤ 10D same plane “S”
c. Two 90o elbows configuration
in perpendicular. S ≥ 22D
planes with
S > 15D.
A A A A A A A A A A B
<0.20 6 10 10 50 19 9 30 17 6 70 2.8
0.30 11 10 12 50 32 9 30 19 6 108 3.0
0.40 16 10 13 50 44 9 30 21 6 145 3.2
0.50 30 30 18 95 44 19 30 25 7 145 3.5
0.60 44 44 30 95 44 29 30 30 9 145 3.9
0.67 44 44 44 95 44 36 44 35 11 145 4.2
0.75 44 44 44 95 44 44 44 44 13 145 4.5
Recommended 44 44 44 95 44 44 44 145 4.5
length for
maximum
range
β ≤ 0.75

A - Meter tube length upstream of the orifice plate in internal pipe diameters, D.
B - Downstream meter tube length in internal pipe diameters, D.
S - Separation distance between piping elements in internal pipe diameters, D.
Denver, Colorado USA June 27 - 30, 1999

Table 1. Orifice Meter Installation Requirements Without a Flow Conditioner; Copyright API, published with permission.
4th International Symposium on Fluid Flow Measurement

o o o
Diameter Ratio, Single 90 elbow Two 90 elbows out of Single 90 Tee Partially closed valves High swirl and Tee Any fitting Downstream
β R/D = 1.5 plane S ≤ 2D (at least 50% open) combined with single (catch all category) Meter Tube
o
R/D = 1.5 90 Tee Length
A1’ A2’ A1’ A2’ A1’ A2’ A1’ A2’ A1’ A2’ A1’ A2’ B

<0.20 5-14.5 5-25 5-14.5 5-25 5-14.5 5-25 5-11 5-13 5-13 5-23 5-11.5 5-13 2.8

0.30 5-14.5 5-25 5-14.5 5-25 5-14.5 5-25 5-11 5-13 5-13 5-23 5-11.5 5-13 3.0

0.40 5-14.5 5-25 5-14.5 5-25 5-14.5 5-25 5-11 5-13 5-13 5-23 5-11.5 5-13 3.2

0.50 11.5-14.5 11.5-25 9.5-14.5 9-25 11-13 9-23 ** 7.5-15 11-13 9-19.5 *** 11.5-14.5 3.5

0.60 12-13 12-25 13.5-14.5 9-25 * 11-16 - 10-17 * 11-16 - 12-16 3.9

0.67 13 13-16.5 13-14.5 10-16 - 11-13 - 10-13 - 11-13 - 13 4.2

0.75 14 14-16.5 - 12-12.5 - 12-14 - 11-12.5 - 14 - - 4.5

Recommended 13 14-16.5 13.5-14.5 12-12.5 13 12-13 9.5 11-12.5 13 13 9.5 13


tube bundle β≤0.67 β≤0.75 β≤0.67 β≤0.75 β≤0.54 β≤0.75 β≤0.47 β≤0.75 β≤0.54 β≤0.72 β≤0.46 β≤0.67
location for
max range of β

Notes:
Lengths shown under the A1' and A2' columns are the ranges of flow straightener locations allowed expressed as a distance (in pipe diameter multiples) between the outlet of
the flow straightener and the orifice plate.
A1’ = 17D ≤ A' < 29D , Meter tube length in internal pipe diameters
= A' ≥ 29D , Meter tube length in nominal pipe diameters
Denver, Colorado USA June 27 - 30, 1999

A2'
S Separation distance between elbows

* 13D allowed for up to β = 0.54


** 9.5D allowed for up to β = 0.47
*** 9.5D allowed for up to β=0.46

Table 2. Orifice Meter Installation Requirements With 1998 Uniform Concentric 19-Tube Bundle Flow Straightener; Copyright API,
published with permission
4th International Symposium on Fluid Flow Measurement Denver, Colorado USA June 27 - 30, 1999

You might also like