You are on page 1of 14

DRAFT

Effect of Orifice Plate Manufacturing Variations on Orifice Meter Performance

Ms. Marybeth Nored, Group Leader, Southwest Research Institute

Introduction

In 2004, the Gas Technology Institute and the American Petroleum Institute (API) sponsored research at
Southwest Research Institute to gather a new set of low pressure orifice meter data to support the orifice
plate expansion factor research 1 . The results of this exploratory study were used to justify the equation
for the expansion factor in the orifice meter standard, AGA Report No. 3, Part 1, also known as API
MPMS Chapter 14.3, Part 1 (Reference 3). One requirement for the expansion factor research stipulated
by the API Chapter 14.3, Part 1 Working Group was that Cd values measured under baseline (reference)
conditions should fall within the 95% confidence limit for the Reader-Harris/Gallagher (RG) equation for
the data to be used in the research.

Surprisingly, during the 2004 research the measured discharge coefficient for several orifice plates was
found to lie outside the confidence interval of the RG equation. This effect could produce a flow
metering error for the orifice meter if the same orifice plate was used to measure flow with the RG
regression equation. The cause of the orifice metering error was traced to physical variations in the plates
themselves (predominately the sharpness of the leading bore edge). The plates in question were returned
to the manufacturer, who tested them in a water flow and concurred with SwRI’s findings.

The variation in orifice plate discharge coefficient values raised considerable doubt about the plate
manufacturing tolerances specified in the AGA/API standard and the possible range of plate discharge
coefficients falling outside the RG equation. Based on these results, it is possible that orifice plates used
in the field have Cd values that are not accurately described by the RG equation. Orifice plates with large
discharge coefficient errors could have an adverse impact on the measurement accuracy at natural gas
custody transfer stations.

In 2007-2008, PRCI funded an investigation to flow test and measure 99 orifice plates in 6-inch and 10-
inch diameter sizes at the Southwest Research Institute Metering Research Facility. The research
program determined the following:

• The expected proportion of orifice plates in a given set representing the five major industry
manufacturers that can be expected to lie outside the RG-equation for the discharge coefficient.

• The adequacy of the current set of specifications for assessing the plate’s suitability for flow
measurement and conformance to the 95% confidence limits of the RG-equation.

• The influence of manufacturing variations on orifice plate performance and any dependency
based on orifice plate size and beta ratio.

The PRCI research was divided into two tasks: 1) Flow testing of all plates at three flow rates to
determine the measured discharge coefficient and 2) Physical measurements of the plates with Cd values
which fell outside of the RG equation uncertainty limits, to determine if the plates met the specifications

1
Morrow, T. B. (Southwest Research Institute), Metering Research Facility Program: Orifice Meter Expansion
Factor Tests in 4-Inch and 6-Inch Meter Tubes, Gas Research Institute Topical Report GRI-04/0042, Gas Research
Institute, Des Plaines, Illinois, USA, June 2004.

“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference


M. Nored, SwRI May 19-21, 2009
Page 1
DRAFT

of AGA Report No. 3. If the plates met the standard’s specifications within the allowable tolerance and
still showed discharge coefficients outside of the RG equation uncertainty, the test program would show
that the existing specifications are not sufficient. The test program included assessment of the standard’s
specifications for plate smoothness, bore edge sharpness, bore eccentricity, bore edge thickness, and plate
bevel.

Technical Approach

A large sample of 99 orifice plates was obtained from PRCI member companies (on loan). The set
represented five plate manufacturers, two line sizes (6-inch and 10-inch diameter plates) and two beta
ratios (0.30 and 0.60 ratios). The plates were a collection of new plates ordered by operating companies
directly from the manufacturer and used plates which had already been in service for a period of time.

The five manufacturers represented the primary orifice plate manufacturers commonly used by the
industry: Daniel, Kelley, Red River, Precision Flow Industries and Star Measurement. Approximately 25
plates in each line size / beta ratio were included in the test program. Four of the plates donated to the
program could not be used because the specifications did not match the required bore diameters. The
exact distribution of plates in each line size and beta ratio is summarized in Table 1 below.

The donated orifice plates were ordered by the operating companies from the manufacturers or obtained
from the operating company inventories, in order to provide an objective review of the industry plate
distribution without biasing the results. The plates were also checked by the operating company’s normal
internal process to verify compliance with API 14.3 / AGA-3 prior to use.

The orifice plates were then tested at the SwRI Metering Research Facility to determine the measured
discharge coefficient. The 6-inch and 10-inch diameter plates with beta ratios of 0.3 and 0.6 were
selected to fit the flow rates available in the MRF High Pressure Loop and to allow plate thicknesses of
both 1/4” and 1/8” to be included in the test program. In addition, in order to maximize the resources
available to the project, it was possible to run one of the sets of tests in series to determine discharge
coefficients for both the high beta ratio 6-inch plates and the low beta ratio 10-inch plates at the same
time. The test points for each beta ratio and line size are shown in Table 2 below.

Table 1. Summary of Donated Orifice Plates from PRCI Member Companies

Nominal Plate Nominal Bore Total No. of Plates Manufacturers Represented


Diameter / Line Diameter (inches) / Obtained in Set (No. of plates from
Size (inches) Beta Ratio manufacturer)

6.437 / Kelley (5), Red River (5), Daniel


1.875 / β = 0.30 25
6” schedule 40 (5), Star (5), PFI (5)

6.437 / Kelley (5), Red River (5), Daniel


3.625 / β = 0.60 25
6” schedule 40 (5), Star (5), PFI (5)

10.687 / Kelley (5), Red River (5), Daniel


3.000 / β = 0.30 25
10” schedule 80 (5), Star (5), PFI (5)

10.687 / Kelley (5), Red River (5), Daniel


6.000 / β = 0.60 24
10” schedule 80 (4), Star (5), PFI (5)

“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference


M. Nored, SwRI May 19-21, 2009
Page 2
DRAFT

Table 2. Flow Rate and Test Points Utilized for MRF HPL Discharge Coefficient Tests

Flow Rate
Line Size Beta Pressure (psia) Temp (degF) Orif DP DP (" H20) (acfm) Nozzles Re
9.564 0.3 500 70 3.48 96.40 230.6 HPL 4" 1233088
9.564 0.3 500 70 5.50 152.35 289.5 HPL 4+2" 1548200
9.564 0.3 500 70 7.83 216.89 344.9 HPL 4+3" 1844511
9.564 0.6 500 70 3.38 93.63 979.3 HPL 8+2" 5188344
9.564 0.6 500 70 5.18 143.49 1210.5 HPL 8+4+2" 6414842
9.564 0.6 500 70 6.22 172.29 1325.4 HPL 8+4+3+2" 7024228
Flow Rate
Line Size Beta Pressure (psia) Temp (degF) Orif DP DP (" H20) (acfm) Nozzles Re
6.065 0.3 500 70 1.45 40.17 59.9 HPL 2" 505406
6.065 0.3 500 70 5.35 148.20 114.8 HPL 3" 968401
6.065 0.3 500 70 12.5 346.25 174.7 HPL 3+2" 1473500
6.065 0.6 500 70 1.15 31.86 230.4 HPL 4" 1922162
6.065 0.6 500 70 1.82 50.41 289.7 HPL 4+2" 2416979
6.065 0.6 500 70 2.6 72.02 346.0 HPL 4+3" 2887261

The flow test results were used to determine which plates had significant departures from the RG
equation prediction for discharge coefficient. These so-called “bad” plates could then be inspected and
measured in detail to compare the plate dimensions to the specifications in the orifice meter standards
(both AGA-3 and ISO 5167). The discharge coefficient difference was based on the measured discharge
coefficient in the flow test, calculated as:

qm
Cd meas = (1)
N 1 ⋅ Ev ⋅ Y ⋅ d 2 ⋅ ρ ⋅ ΔP

In this case, the reference sonic nozzle mass flow rate at the SwRI MRF provided the qm value to
determine the measured discharge coefficient. The calculated discharge coefficient based on the RG
equation (per AGA-3):

⎛ ⎛ 19000 ⋅ β ⎞ ⎞⎟ 4 ⎡10 ⎤
0 .7 0.35
⎡10 6 ⋅ β ⎤
0.8 6
Cd RG = C i ( FT ) + 0.000511⎢ ⎥ ⎜
+ 0.0210 + 0.0049 ⋅ ⎜ ⎟ ⋅β ⋅⎢ ⎥ (2)
⎣ Re ⎦ ⎜ ⎝ Re ⎠ ⎟⎠ ⎣ Re ⎦

The difference between the calculated discharge coefficient per the RG equation and the measured
discharge coefficient provides the orifice meter error as:

Cd RG − Cd meas
% Error = (3)
Cd meas

The plates that were found to be outside of the 95% confidence interval of the RG-equation based on the
uncertainty in the regression fit of the discharge coefficient equation. In terms of percent uncertainty, the
95% confidence interval varies based on beta ratio. The relationship between the 95% confidence
interval, the actual percent uncertainty in discharge coefficient and plate beta ratio is illustrated in Figure
1 for an infinite Reynolds number condition (Ref 3). At very low and high beta ratios, the confidence
interval includes a higher percent uncertainty in orifice plate discharge coefficient.

“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference


M. Nored, SwRI May 19-21, 2009
Page 3
DRAFT

0.65
0.60
0.55
0.50

Percent uncertainty (U95)


0.45
0.40
0.35
0.30 Allowable departure in Cd at
0.25 tested beta ratios to fall outside
95% confidence interval
0.20
0.15
Em pirical Coefficient of Discharge:
0.10
Uncertainty at Infinite Reynolds Num ber
0.05
0.00
0.1 0.2 0.3 0.4 0.5 0.6 0.7

Orifice β ratio
Figure 1. RG Equation 95% confidence interval vs. beta ratio

Physical plate dimensions were then recorded for a representative subset of plates, which included one
good plate in each line size and beta ratio and a subset of “bad” plates whose discharge coefficients fell
outside the confidence limits of the RG equation. These orifice plates underwent laboratory
measurements at the Daniel Measurement (Emerson Process) facilities in Houston, Texas. All plate
measurements were witnessed by SwRI.

The dimensional measurements recorded for the plates were as follows:

• Plate roughness (upstream and downstream faces)


• Flatness of upstream and downstream faces
• Bore edge sharpness
• Eccentricity
• Bore edge thickness
• Plate thickness
• Bore deviation from cylindrical
• Plate bevel angle
• Bore diameter
• Overall plate diameter
• Comparison of stamped bore hole diameter to measured bore hole diameter

Flow Testing Program

The flow testing of the four sets of orifice plates (10-inch beta = 0.60, 10-inch beta = 0.30, 6-inch beta =
0.60, and 6-inch beta = 0.30) was performed in three separate test series over the March and April 2008
time period, based on the flow facility schedule. The meter runs were installed in the High Pressure Loop
during the entire time period. Each plate in the set was tested at the same three flow rates. In order to
maximize project resources, the 10-inch low beta ratio tests were tested in series (upstream) of the 6-inch
high beta ratio test run. This test set-up was possible due to the corresponding flow rates for these two
sets of orifice plates.

“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference


M. Nored, SwRI May 19-21, 2009
Page 4
DRAFT

The orifice meter runs were installed in the MRF High Pressure Loop. The meter runs contained ten pipe
diameters of straight upstream length and an upstream flow conditioner. The downstream piping in each
meter run included taps for the temperature and static pressure measurements. Drawings of both the 6-
inch and 10-inch diameter orifice meter runs are provided below. Photographs of the installation are
provided in Appendix 2. Baseline tests of both meter runs using an artifact orifice plate were performed
to verify the senior fittings and flow conditioner in each meter run.

An existing 10-inch orifice meter run from the MRF pipe inventory was installed in between 12x10
reducers upstream of the HPL test section. The meter run was used previously in PRCI and GRI research.
The meter run uses a 19-tube bundle flow conditioner upstream of a standard senior fitting. The
dimensions of the meter run are provided in Figure 1 and the installation photos are provided in Figure 2.

10” Sched. 80 pipe except where noted; all flanges ANSI-600

Flow 19-tube Senior orifice fitting RTD Unused TOL


bundle

12”x10” Sched. 100 Sched. 80 Sched. 80 Sched. 40 12”x10”


reducer 9.6D (91.5”) 8.8D (84”) 3.6D (35”) 49” reducer
10” Sched. 80
UL = 31.4 D RTD 8.0D from plate orifice meter run
UL2 = 13 D
DL = 6.7 D to Sched. 40 spool OAL, including gaskets and tube bundle flange: 456”

Figure 2. Drawing of 10-inch diameter meter run

FLOW Upstream
flange

19-tube
bundle

Figure 3. Installation of 10-inch diameter meter run at SwRI MRF

For the 6-inch orifice plate tests, a new meter run was rented from Pickett Measurement Systems based on
their proximity to Southwest Research Institute. The 6-inch meter run was inspected by SwRI prior to
installation in the MRF High Pressure Loop. The meter run was fabricated per AGA-3 specifications and
“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference
M. Nored, SwRI May 19-21, 2009
Page 5
DRAFT

was in new condition. The meter run contained a CPA 50E plate flow conditioner installed at ten pipe
diameters upstream of the senior fitting and an additional ten diameters of pipe upstream of the flow
conditioner.

Figure 4 provides the dimensional drawing of the 6-inch meter run. The installation photo of the 6-inch
meter run installed in the West Header test section is provided in Figure 5.

Figure 4. Drawing of 6-inch diameter meter run

“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference


M. Nored, SwRI May 19-21, 2009
Page 6
DRAFT

FLOW

FLOW

Figure 5. Installation of 6-inch diameter meter run in west header at SwRI MRF

The differential pressure across the orifice plate was measured using two stacked Rosemount differential
pressure transmitters. The two transmitters included a low range DP transmitter (0-250” of water
column) and a high range DP transmitter (0-1000” of water column.) The higher range transmitter was
primarily used to measure the high flow point in the test series for the 6-inch low beta ratio plates. All
other flow test points relied on the low range differential pressure transmitter.

All orifice plates were tested over three flow rates as required by the test plan. On each flow rate, six 90-
second repeat runs were recorded. The six repeat runs for each flow rate / plate were averaged and
graphed using a probability distribution format. The results for each set of plates may be summarized as:

• Group 1: 6-inch diameter, beta = 0.3: The discharge coefficient error in the lower beta ratio 6-
inch diameter plates ranged from -0.590% to +0.624%. The majority of the plates in this set fell
within the allowable uncertainty of the RG equation. The average error in the 6-inch lower beta
ratio plates was 0.1136%. For this set, three orifice plates fell outside the RG equation
confidence interval - Plate Nos. 1, 15, and 57.

• Group 2: 6-inch diameter, beta = 0.6: The discharge coefficient error in the higher beta ratio 6-
inch diameter plates ranged from -0.547% to +0.113%. The majority of the plates fell within the
allowable uncertainty of the RG equation. The average error in the 6-inch lower beta ratio plates
was -0.208%, as shown in the probability distribution. In this set, two plates fell outside of the
RG-equation because of high measured discharge coefficients: Plate Nos. 16 and 23.

• Group 3: 10-inch diameter, beta = 0.3: The discharge coefficient error in this group ranged from -
0.395% to +0.023%. All of the plates fell within the allowable uncertainty of the RG equation.
The average error in the 6-inch lower beta ratio plates was -0.023%, as shown in the probability
distribution. This finding was an impressive result of the lower beta range.

• Group 4: 10-inch diameter, beta = 0.6 : The final set of 10-inch diameter higher beta ratio plates
showed the largest average error and the highest number of plates outside the 95% uncertainty of
“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference
M. Nored, SwRI May 19-21, 2009
Page 7
DRAFT

the RG-equation. The plates in this set showed errors ranging from +0.190% to +0.895%.
Fifteen plates in the set (out of 24 total) had measured discharge coefficients outside of the RG
equation for at least one of the three flow rates tested, representing 62.5% of the population. The
results demonstrated a clear sensitivity of the larger diameter, larger beta ratio plates to physical
plate variations and manufacturing differences.

Figure 6 emphasizes the representative distribution of plates in each set with high errors outside the 95%
confidence interval of the RG-equation. For 6-inch diameter plates in both beta ratios, approximately 8-
12% of plates can be expected to have a high meter error greater than the confidence interval of the RG-
equation. In this case, the plates with metering errors had a difference in discharge coefficient greater
than 0.45%. For the 10-inch plates, none of the lower beta ratio plates showed a discharge coefficient
with a large difference from the calculated RG equation. However, approximately 60% of the higher beta
ratio (beta=0.6) 10-inch diameter orifice plates had large errors greater than 0.50%. The higher beta ratio
plate population contained 40% of plates with errors greater than 0.75%.
Representative Percent of Outlier Orifice Plates
in 6-inch and 10-inch Diameter Sets
Absolute Error
70.0 Range: 0.48-0.895%
% of Plates with High Error (Outside

60.0

50.0
of RG U-95%)

40.0
Absolute
Error Range:
30.0 Absolute
0.55%-0.62%
Error Range:
20.0 0.46%-0.55%

10.0

0.0
6-inch, 6-inch, 10-inch, 10-inch,
Beta=0.3 Beta=0.6 Beta=0.3 Beta=0.6

Figure 6. Distribution of high error plates in each orifice plate population

Laboratory Physical Plate Measurements

Plates were prepared for the inspection process at the Daniel Measurement laboratory in Houston, Texas.
All plates were transported to the Daniel facilities by Southwest Research Institute engineers. SwRI
witnessed all of the plate measurements as well. The specifications and measurements recorded by the
laboratory are given below. In the case of the edge sharpness measurement, the lead foil test was
performed by Daniel specialized technicians. The measurements were performed over a two week period
from July 1 to July 15, 2008.

• Upstream face flatness and surface roughness


• Downstream face flatness and surface roughness

“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference


M. Nored, SwRI May 19-21, 2009
Page 8
DRAFT

• Thickness, e and E, for bore edge and overall plate thickness


• Bevel angle
• Leading edge sharpness
• Bore diameter
• Overall diameter
• Eccentricity

All edge sharpness measurements and previous Daniel lab physical dimension measurements were
compiled into the final spreadsheet provided in the PRCI final report. In addition, the SwRI engineer’s
report of the detailed lead foil test for edge sharpness is provided. The edge sharpness test followed the
specifications of the ISO-5167 orifice standard (Ref 4).
Most of the plates had common parameters that showed deviation from the standard’s specified
tolerances. These departures were probably a result of the manufacturing process, although some
differences could have been due to standard wear and tear from operational service. It should be noted
that a majority of the plates were new plates, donated to the project upon receipt from the plate
manufacturer.
The observations from the laboratory measurements, the physical parameter variations and causes for
discharge coefficient error in the plates are summarized as follows:

1. The overall diameter was found to be non-compliant on plates 1, 16, 19 and 23. Plates 1 and 16
showed a negative deviation which can be the result of over-grinding. Plates 19 and 23 showed
undercutting. This parameter is not expected to impact the discharge coefficient departure from
the RG equation significantly.

2. None of the bore diameters were found to be out of standard limits, although some of the marked
dr values on the plates were slightly different than the measured bore diameter. The highest
departure in the measured versus stamped bore diameter value was found to be on the order of
0.05%. Although this is a low error, it is a cause of unnecessary uncertainty that can be reduced
through high-quality measurements / stamping procedures by the plate manufacturer.

3. The eccentricity was found to be too high in eight of the eleven outlier plates as well as two of the
representative set of good plates. Eccentricity departures are largely due to defects in the overall
diameter. Plate # 25 (one of the good plates) showed a large deviation and this was caused by a
damaged outer diameter. Since some of the “good” plates had unacceptable eccentricity, it is
debatable whether this parameter influences the measured discharge coefficient significantly and
if it could be relaxed in the orifice standard.

4. A high amount of eccentricity combined with a thick bore edge may cause significant departure
from the RG equation. These two factors were non-compliant in many of the outlier plates. The
eccentricity and bore edge thickness are addressed separately by the present orifice standard.

5. The orifice edge thickness was non-compliant in seven of the eleven outlier plates measured at
Daniel and in six of the additional 10-inch diameter plates measured by SwRI. The amount of
departure from the specified value plus the tolerance varied from 5 to 15% of the nominal bore
thickness. This single characteristic was the primary cause of the large number of 10-inch beta
0.6 plates in the population with high metering errors.

“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference


M. Nored, SwRI May 19-21, 2009
Page 9
DRAFT

6. The deviation from standard values for thickness of plate and bore edge arise mainly from
manufacturing accuracy.

7. The bevel angle for all plates was within specified AGA limits and showed no major deviation.

8. Face flatness was found to be unacceptable in plates 15, 82 and 83. These plates non-standard
flatness was found to be on the upstream side which could be a result of the pressure acting on the
plate face. The upstream flatness is an important parameter and may have also contributed to the
departure from the RG equation for these three plates.

9. The surface roughness of the upstream and downstream faces of the plates was within AGA
standards of 50 RMS in all cases.

10. The edge sharpness measurement did not determine any of the plates to have unacceptably dull
leading edges. However, the selected magnification factor is significantly influential in
determining if a plate is sufficiently sharp, per the ISO 5167 standard. It was noted that plates 4
and 68 were observed to be duller than the other plates in the magnified images.

Results and Discussion

The summary of the eleven plates measured at the Daniel facility and the additional plates characterized
by SwRI is provided in Table 3. The table indicates the plates that did not meet the ISO 5167 or AGA
Report No. 3 specifications – shown as highlighted pink squares. Plate numbers 9, 19, 25, 85 and 97 are
the representative good plates selected for the measurements to compare against the plates with high
discharge coefficient errors.

Table 3 shows that each of the “bad” plates with high discharge coefficient errors had at least one primary
characteristic that failed to meet the standard’s specifications. Many of the plates had more than one
characteristic which did not meet tolerance specifications.

For example, two of the 6-inch diameter plates (plates 1 and 16) had at least three characteristics that
exceeded the defined tolerance limit. It should also be noted that three “good” plates with measured
discharge coefficients within the RG equation uncertainty limits did not meet the standard specifications –
see plate numbers 19, 25, and 85. This suggests that some of the tolerance specifications may be too tight
for some plate sizes and beta ratios. The eccentricity requirements in particular appear to be overly
conservative and not significantly influential to the orifice plate measurement.

The primary causes of plate deviation were linked to:


• Exceeding the tolerance dimension for overall plate diameter
• Non-compliant orifice bore diameter
• Non-compliant plate thickness
• Exceeding orifice bore edge thickness

“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference


M. Nored, SwRI May 19-21, 2009
Page 10
DRAFT

Table 3. Summary of Characteristic Causes for Plate Error – Laboratory Measurements

Plate Plate Line Beta Plate Overall Bore Plate Bore Edge Bevel Face Edge
No. Class Size Ratio Make Diameter Diameter Eccentricity Thickness Thickness Angle Flatness Sharpness
1 BAD 6" 0.3 A X X O X O O O O
4 BAD 10" 0.6 A O O X O X O O O
7 BAD 10" 0.6 A O O X O X O O O
9 GOOD 6" 0.6 B
O O O O O O O O
15 BAD 6" 0.3 C
O O X X O O X O
16 BAD 6" 0.6 C
X X X O X O O O
19 GOOD 6" 0.6 D
X O X O O O O O
23 BAD 6" 0.6 D X O X O O O O O
25 GOOD 6" 0.3 D
O O X O O O O O
57 BAD 6" 0.3 A O O O X O O O O
68 BAD 10" 0.6 A O X O O X O O O
81 BAD 10" 0.6 E O O X O X O O O
82 BAD 10" 0.6 E O O X X X O X O
83 BAD 10" 0.6 E O O X X X O X O
85 GOOD 10" 0.3 E
O O X O X O O O
97 GOOD 10" 0.6 B O O O O O O O O

Plate Plate Line Beta Plate Overall Bore Plate Bore Edge Bevel Face Edge
No. Class Size Ratio Make Diameter Diameter Eccentricity Thickness Thickness Angle Flatness Sharpness
12 BAD 10" 0.6 B O O N/A O O N/A N/A N/A
18 BAD 10" 0.6 C
X X N/A O X N/A N/A N/A
44 BAD 10" 0.6 C
O O N/A O X N/A N/A N/A
45 BAD 10" 0.6 C
X O N/A O O N/A N/A N/A
66 BAD 10" 0.6 A O O N/A X X N/A N/A N/A
67 BAD 10" 0.6 A O O N/A O X N/A N/A N/A
80 BAD 10" 0.6 E O X N/A O X N/A N/A N/A
84 BAD 10" 0.6 E O X N/A O X N/A N/A N/A
99 BAD 10" 0.6 B O X N/A X O N/A N/A N/A

The recommendations related to the existing orifice plate flow standard specifications are summarized
below.

1. Edge Sharpness:
Edge sharpness measurements of the orifice plates at Daniel laboratory revealed that all of the
plates had acceptable edge sharpness. The lead foil method was used to determine edge
sharpness since this is the only method currently available. Per the ISO 5167 standard, the choice
of magnification factor influences the tolerance specification on edge sharpness. The actual
sharpness is measured using the formula:

Sum of Radius Measurements 1


×
64 Magnification
Actual sharpness, rm = (4)
Number of readings

“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference


M. Nored, SwRI May 19-21, 2009
Page 11
DRAFT

High values of the magnification factor can reduce the rm value significantly. This appeared to be
the case in some of the measurements. (In particular, plate 4 and plate 68 had duller edges when
viewed under the microscope, but still had acceptable rm values.)

The standard requires that the actual sharpness must be less than the edge curvature limit rk,
which is based on the orifice diameter:

Edge curvature, rk = 0.0004 × Orifice I .D. (5)

For an orifice plate to have an acceptably sharp leading edge, the actual radius of the leading edge
must be less than the radius of curvature limit:

For acceptance, rm < rk (6)

The existing flow standards do not require a maximum limit to the magnification factor. This
leads to a possibility where dull plates can appear to look sharp because the selected
magnification factor is too large. In addition, non-uniform edge sharpness (around circumference
of plate bore) is not addressed by the present standard. The equation permits an average value for
the entire edge, which overlooks the possibility that certain circumferential regions can be dull.
Higher beta ratio plates appear to be more sensitive to edge sharpness. This beta ratio dependence
is not presently addressed in the standard.

2. Eccentricity:

High values of eccentricity were notable in a majority of the orifice plates but in all cases this was
not the only characteristic which fell outside specified tolerances. Some of the good plates also
showed high eccentricity values. Eccentricity could not be identified as the primary cause of a
high degree of error in the discharge coefficient. It is debatable whether this characteristic’s
tolerance is overly conservative because it clearly did not cause the “good” plates (plates 19, 25,
85) to measure erroneously. This finding was a particularly interesting result pertaining to the
existing orifice flow measurement standard.

3. Marked / Stamped Value of Orifice Bore Diameter:

Of the total collection of 26 plates with high discharge coefficient errors, seven of these plates had
dr values stamped on the plates which were more than 0.01% different than the actual physical
bore diameter. Notably, plate 68 had a dr value of 5.997 inches on its face, when its true value
was measured at exactly 6.000 inches. This physical bore dimension has a direct effect on the
metering accuracy since the squared value of the bore diameter is used in the mass flow rate
equation. In the worst case, plate 68 had a stamped dr bore value with a 0.05% error, which
would lead to a 0.0999% error in flow rate because of the squared relationship.

The disagreement in dr bore values was a finding from the research that affects the plate
manufacturer quality control program. The bore diameter should be a parameter that is carefully
measured and stamped on the plate by the manufacturer. Errors greater than 0.01% should not be
permitted by the operating company.

“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference


M. Nored, SwRI May 19-21, 2009
Page 12
DRAFT

Conclusions

The PRCI orifice plate tolerance specifications research program sought to determine the effect on orifice
meter flow measurement accuracy caused by variations in orifice plate manufacturing. The intent of the
research was to confirm that the specified tolerances in the orifice flow measurement standard are
sufficient and provide orifice plates which meet the uncertainty requirements of the orifice flow
measurement standard. Through flow testing and physical plate measurements, approximately 25% of the
99 orifice plates included in the test program had discharge coefficients outside of the 95% uncertainty
limits of the RG equation. This percentage was not evenly distributed between the subsets of plates.

The orifice plates with high discharge coefficient error (outside the 95% uncertainty limits of the RG
equation) were distributed between the different plate sizes as follows:

• Three (3) of the 6-inch diameter beta 0.3 plates = 12% of the 25 plates in this size
• Two (2) of the 6-inch diameter beta 0.6 plates = 8% of the 25 plates in this size
• None of the 10-inch diameter, beta 0.3 plates
• Fifteen (15) of the 10-inch diameter beat 0.6 plates = 60% of the 24 plates in this size

The research found that the major causes of high discharge coefficient error were traceable to high (out of
tolerance) values of bore edge thickness, plate thickness, plate diameter and bore diameter. Many of the
high error 10-inch beta 0.6 plates had high values of bore edge thickness. This variation could be reduced
through better quality control on the plate manufacturer side. Several orifice plates did not have the
correct bore diameter (dr values) stamped on the plates, which introduced another source of measurement
error that could be eliminated through improved quality standards.

Regarding the present orifice meter standards, the flow testing and physical measurements suggested that
edge sharpness measurement is not sufficiently addressed by the existing flow measurement standard.
The measurement of edge sharpness needs additional clarification to define a maximum allowable
magnification factor. In some cases, it is possible to select an excessively large magnification factor
which effectively makes all edges appear sufficiently sharp. An upper limit to this value would help to
delineate leading edges which are not sufficiently sharp.

“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference


M. Nored, SwRI May 19-21, 2009
Page 13
DRAFT

Acknowledgements

The successful findings and results of the orifice place tolerances specification project is directly tied to
the contributions and assistance of key individuals. These individuals’ support and contributions is
gratefully acknowledged. During the course of the 13 month research project, Jim Witte and Michael
Whelan provided expert review and insight in the review of test data, donation of orifice plates, required
laboratory tolerance measurements and value of the research for the natural gas industry.
The staff at Daniel Measurement and Control (Emerson) in Houston, Texas volunteered to measure the
orifice plates and assess edge sharpness using the lead foil method. The following pipeline companies
provided donated orifice plates to the project for the purpose of obtaining a large set of representative
plates: El Paso Energy (Tennessee Gas Pipeline), Williams / Northwest Pipeline and Pacific Gas and
Electric.
In addition, the staff of the SwRI Metering Research Facility provided the high level of precision and care
needed in the installation of the orifice plates and measurement of the discharge coefficient using the
facility reference flow rate. Their assistance is gratefully acknowledged.

References

1. AGA (2000), “Orifice Metering of Natural Gas and Other Related Hydrocarbon Fluids,” AGA Report
No. 3, Parts 1-4, American Gas Association, Washington, D.C.

2. Morrow, T. B. (Southwest Research Institute), Metering Research Facility Program: Orifice Meter
Expansion Factor Tests in 4-Inch and 6-Inch Meter Tubes, Gas Research Institute Topical Report GRI-
04/0042, Gas Research Institute, Des Plaines, Illinois, USA, June 2004.

3. Southern Gas Association short course, “Factors in Meter Station Design,” Orifice Flow Measurement,
June 2006.

4. ISO (2003), “Measurement of Fluid Flow by Means of Pressure Differential Devices Inserted in
Circular Cross Section Conduits Running Full – Part 2: Orifice Plates,” International Standard ISO 5167-
2, Geneva, Switzerland.

“Effect of Orifice Plate Manufacturing Variations” AGA Operations Conference


M. Nored, SwRI May 19-21, 2009
Page 14

You might also like