EPA-650/2-75-020
February 1975
Environmental Protection Technology Series
CONTINUOUS MEASUREMENT
OF TOTAL GAS FLOWRATE
FROM STATIONARY SOURCES
OHice o( Research and Development
US Environmental P'oteclion Agency
Washington. DC 20460
-------
EPA-650/2-75-020
CONTINUOUS MEASUREMENT
OF TOTAL GAS FLOWRATE
FROM STATIONARY SOURCES
by
E. F. Brooks, E. C. Beder,
C. A. Flegal, D. J. Luciani, and R. Williams
TRW Systems Group
One Space Park
Redondo Beach, California 90278
Contract No. 68-02-0636
ROAP No. 21ACX-AE
Program Element No. 1AB013
EPA Project Officer: William B. Kuykendal
Control Systems Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
February 1975
-------
EPA REVIEW NOTICE
This report has been reviewed by the National Environmental Research
Center - Research Triangle Park. Office of Research and Development,
EPA, and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency , have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology. Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields. These series are:
1. ENVIRONMENTAL HEALTH EFFECTS RESEARCH
2. ENVIRONMENTAL PROTECTION TECHNOLOGY
3. ECOLOGICAL RESEARCH
4. ENVIRONMENTAL MONITORING
5. SOCIOECONOM1C ENVIRONMENTAL STUDIES
6. SCIENTIFIC AND TECHNICAL ASSESSMENT REPORTS
9. MISCELLANEOUS
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to
develop and demonstrate instrumentation, equipment and methodology
to repair or prevent environmental degradation from point and non-
point sources of pollution. This work provides the new or improved
technology required for the control and treatment of pollution sources
to meet environmental quality standards.
This document is available to the public for sale through the National
Technical Information Service, Springfield, Virginia 22161.
11
-------
ABSTRACT
The program objective was to evaluate hardware and techniques for the
continuous measurement of the total gas flowrate from stationary sources,
specifically in large or complex ducts where total flow metering devices
such as plate orifices are not practical. Work consisted of formulation
of operating specifications, evaluation of commercially available velocity
sensors, development and evaluation of flow mapping techniques, and
field demonstration of both hardware and technique. Results showed that
total volumetric flowrate can be measured with accuracies consistently
better than 10% in either circular or rectangular ducts through proper
placement of from one to eight flow sensors, when standard traversal
techniques would require twenty to fifty traverse points. The rectangular
duct mapping techniques developed during the program were found to have
optimum accuracy immediately downstream of an elbow. Several off-the-
shelf velocity sensors were found acceptable for use in the specified
stack-type environment. The field demonstrations verified the acceptability
of both hardware and techniques.
This report was submitted in fulfillment of Contract No. 68-02-0636,
by TRW Systems Group under the sponsorship of the Environmental Protection
Agency. Work was completed as of March, 1975.
iii
-------
CONTENTS
Page
Abstract iii
List of Figures vi
List of Tables xi
Acknowledgements xiv
Sections
I Conclusions 1
II Recommendations 2
III Introduction 3
IV Task I - Operating Specifications 6
4.1 General 6
4.2 Flow Environment 6
4.3 External Environment 8
4.4 Instrument System Performance Specifications 8
V Task II - Instrument Selection 17
5.1 Literature Bibliography 17
5.2 Instrument Survey 18
5.3 Instrument Selection 18
5.4 Instrument Acquisition 36
5.5 Advanced Instrument Acoustic Velocimeter 36
VI Task III - Mapping Technique Evaluation 50
6.1 General 50
6.2 Description of Facilities 50
6.3 Mapping Techniques 55
6.4 Mapping Test Results 67
6.5 Mapping Test Summary 105
iv
-------
CONTENTS - continued
Sections Page
VII Task IV - Laboratory Assessment 112
7.1 Facility Description and Scope of Testing 112
7.2 Analytical Instrument Description 116
7.3 Basic Instrument Calibration 121
7.4 Stability 142
7.5 Time Response 146
7.6 Sensitivity to Orientation 146
7.7 Environmental Testing 149
7.8 Laboratory Test Summary and Final Evaluation 165
VIII Task V - Field Demonstrations 191
8.1 General 191
8.2 Facility Description 191
8.3 Test Conduct 194
8.4 Flow Data Dorrelation 197
8.5 Test Results 198
8.6 Summary of Results 213
IX Discussion of Results 216
X References 220
XI Glossary 221
XII Appendices 223
-------
FIGURES
No.
1. Flow Measurement Control Volume 9
2. Duct Analogy for Transformation of Test Velocity to
Standard Velocity 12
3. Combined Reversed Pi tot Tube 21
4. Flare Gas Flow Probe 22
5. Flare Gas Probe Specifications 23
6. Drag Meter 26
7. Thermo Systems Hot Film Sensors Tested 28
8. TSI Metal Clad Sensor and Anemometer 29
9. TSI Specification Sheet 30
10. Elementary Fluidic Velocity Sensor 32
11. Annubar 35
12. Acoustic Measurement of Flow Velocity in Duct (Schematic) 38
13. Stack Flue Interior Noise, Mohave Power Plant Unit 1,
March 2, 1973, Operating at 760 ±5 Megawatts. B&K 4136
Microphone. One-third Octave Spectrum 42
14. Stock Flue Interior Noise, Mohave Power Plant Unit 1,
March 2, 1973, Operating at 760 ±5 Megawatts. B&K 4136
Microphone. Narrow,Band Spectrum 43
15. Stack Flue Interior Noise, Mohave Power Plant Unit 1,
March 2, 1973, .Operating at 760 ±5 Megawatts. Statham
PL 80TC-0.3-350 Pressure Transducer. Narrow Band
Spectrum 44
16. Stack Flue Interior Wall Vibration, Mohave Power Plant,
March 2, 1973, Operating at 760 ±5 Megawatts. B&K 4136
Microphone. One-third Octave Spectrum 45
17. Precipitator Inlet Flue Interior Noise, Mohave Power
Plant Unit 1, March 3, 1973, Operating at 775 ±5 Mega-
watts. B&K 4136 Microphone. One-third Octave Spectrum 46
vi.
-------
FIGURES - Continued
No. Page
18. Precipitator Inlet Flue Interior Noise, Mohave Power
Plant Unit 1, March 3, 1973, Operating at 775 ±5 Mega-
watts. B&K 4136 Microphone. Narrow Band Spectrum 47
19. Precipitator Inlet Flue Interior Wall Vibration, Mohave
Power Plant Unit 1, March 3, 1975, Operating at 775 ±5
Megawatts. Narrow Band Power Spectral Density 48
20. Circular Duct Mapping Test Configuration, 1973 51
21. Pitot Traverse Installation -- Circular Duct 52
22. Rectangular Duct Mapping Test Configurations, 1973 53
23. Schematic of Rectangular Duct Reference Traverse Map 54
24. Annubar Location During Duct Mapping Test, 1973 56
25. Schematic -- Top View of Mapping Test Facility, 1974 57
26. Position and Orientation of Reference Probes 58
27. Position and Orientation of Test Probes for Circular
Section 59
28. Rectangular Section Probe Locations 60
29. Probe Locations Relative to Elbows in 1974 Mapping Tests 61
30. Probe Locations -- 1975 Testing 62
31. Four and Five Point Methods for Rectangular Duct Mapping 66
32. Row Average Method 68
33. Typical Flow Profiles from Circular Duct Mapping Test 69
34. Circular Inlet Blockage -- 1974 72
35. Results of Four and Five Point Analysis for a Partially
Developed Flow Run 80
36. Rectangular Inlet Blockage -- 1974 94
37. Probe Placement for General Rectangular Duct Applications 109
38. Probe Placement after a Rectangular Elbow 110
VII
-------
FIGURES - Continued
No. Page
39. Low Speed Wind Tunnel 113
40. TSI Wind Tunnel 114
41. Coal Fired Combustion Flue Gas Simulator 115
42. Manufacturer's Accuracy for Instruments Tested 122
43. S Probe Calibration Factor 124
44. Schematic of Ramapo Mark V Flow Meter 125
45. Ramapo Mark V Calibration: Free Stream Velocity Versus
Ratio of Output Voltage to Input Voltage 127
46. Difference between Factory Calibration and Test Velocity
in Percent Versus Test Velocity for Ramapo Mark V 128
47. Diagrammatic Explanation of Ramapo Output at High Speed
in Wind Tunnel Testing 129
48. Results of Force Calibration of Ramapo Mark V as Percent
Difference in Factory and Test Velocity Versus Equivalent
Test Velocity 130
49. Moment Applied at Strain Gauge Bridge of Ramapo Mark V 132
50. Factory and Test Calibration Curves for Hastings-Raydist
AFI-10K Probe as Probe Output Voltage Versus Standard
Velocity 134
51. Absolute Difference in Percent between Velocity Data Points
and Reference Curve Fit Velocity Versus Standard Velocity
for Hastings Probe . 135
52. Comparison of Calibration Curves for Three Hastings
Probes and Test Data for AFI-10K Probe 136
53. Calibration of TSI Metal Clad Sensor as Power Squared
Versus Velocity 138
54. Calibration of TSI Metal Backed Sensor as Power Squared
Versus Velocity 139
55. Calibration of TSI Wedge Sensor as Power Squared
Versus Velocity 140
56.. Annub'ar Calffiratton Factor 143
57. Bamapo probe Sta&lli;ty Dgta Reduction S&eet 144
vii i
-------
FIGURES - Continued
No. Page
58. S Probe Orientation Angles 148
59. S Pitot Probe Orientation Sensitivity Data 150
60. Ramapo Probe Orientation Sensitivity Data 151
61. Hastings-Raydist Probe Orientation Sensitivity Data 152
62. Annubar Orientation Sensitivity Data 153
63. TSI Metal Clad Sensor Orientation Sensitivity as
Normalized Velocity Versus Orientation Angle 154
64. TSI Metal Backed Sensor Orientation Sensitivity as
Normalized Velocity Versus Orientation Angle 155
65. TSI Wedge Sensor Orientation Sensitivity as Normalized
Velocity Versus Orientation Angle 156
66. Accuracy of Hastings AFI-10K as Percent Difference Between
Test Data Points and Calibration Curve Fit Versus Velocity 160
67. Ramapo Purged Probe 161
68. Pre-and Post-Environmental Calibrations of TSI Metal Clad
Sensor as Sensor Power Squares Versus Sensitivity 162
69. Pre-and Post-Environmental Calibrations of TSI Metal Backed
Sensor as Sensor Power Squared Versus Velocity 163
70. Pre-and Post-Environmental Calibration of TSI Wedge
Sensor as Sensor Power Squared Versus Velocity 164
71. Ramapo Mark VI 170
72. Ramapo Mark VI Specification Sheet 171
73. Hastings Stack Meter 174
74. Hastings Stack Gas Meter Specification Sheet 175
75. TSI Total Vector Acemometer 178
76. TSI Total Vector Anemometer Specifications 179
77. Approximate Pi tot-Static Probe Response at Large Yaw Angles 186
78. Schematic of Moapa Power Plant 192
IX
-------
FIGURES - Continued
Page
79. Field Demonstration Rectangular Duct Geometry 193
80. Schematic of Traversing Mechanism for Ramapo Drag Meter 195
81. Stack Velocity Profiles 203
-------
TABLES
No. Page
1. Key Word Classification for Flow Instruments 19
2. Method for Determining Volumetric Flow from Point
Measurements in a Circular Duct 64
3. Circular Duct Mapping Test Results, 1973 71
4. 1974 Circular Duct Point Sensor Mapping Results 73
5. Annubar Calibration Factors for 1974 Calibration Tests --
Configurations 3 and 4 74
6. Summary of 1974 Circular Duct Mapping Test Results 75
7. Results of Rectangular Duct Analysis by Four and Five
Point Methods . 81
8. Results of Rectangular Duct Analysis by Rows 85
9. Examination of 1973 Row Average Data after a Rectangular
Elbow 88
10. Annubar Duct Measurement Results, 1973 90
11. Summary of Annubar Duct Measurement Results, 1973 92
12. 1974 Row Average Results after a Straight Inlet
(Configuration 1) 95
13. Annubar Calibration Factors for 1974 Calibration Tests —
Configuration 1 96
14. Acceptable Flow Mapping Region for Constant Inlet Geometry
and Variable Flowrate for 1974 Mapping Test — Straight
Rectangular Inlet 98
15. 1974 Row Average Results after a Rectangular Elbow
(Configuration 2) 100
16. Annubar Calibration Factors for 1974 Calibration Tests —
Configuration 2 101
17. Row Average Results for 1975 Testing 103
18. Annubar Results for 1975 Testing 104
XI.
-------
TABLES - Continued
No. Page
19. Recommended Circular Duct Flow Measurement Techniques 107
20. Recommended Rectangular Duct Flow Measurement Techniques 111
21. TSI Hot Film Sensor Calibration Summary 141
22. Stability Test Results 145
23. Response Time Test Results 147
24. Orientation Sensitivity Test Results 157
25. Post Environmental Test Calibration Results 159
26. Ellison Instruments Combined Reversed ("S") Pitot Probe
Calibration Results 166
27. Ramapo Mark V Flow Meter Calibration Results 173
28. Haystings-Raydist AFI-10K Gas Flow Probe Calibration
Results 176
29. Thermo Systems Hot Film Sensor Calibration Results 181
30. Annubar Calibration Results 183
31. Yaw Characteristics of Pitot-Static Probes 185
32 Laboratory Test Summary 188
33. Duct Traverse Summary, 1974 200
34. Duct Traverse Summary, 1975 201
35. Stack Traverse Summary, 1975 201
36. Summary of Annubar Continuous Monitoring Data with Scrubber
Off, 1974 205
37. Computation of Total Flow from Coal Analysis and Measured
C02 Concentration, 1974 206
38. Average Daily Flow through Ducts, 1975 208
39. Average Daily Flow through Stack, 1975 209
xn
-------
TABLES - Continued
No. Page
40. Summary of Annubar and Ramapo Short Term Monitoring Data,
1974 210
41. Field Demonstration Pitot Traverse Data, 1974 211
42. Row and Column Analysis of Field Test Pitot-Static Traverse
Data, 1974 212
43. Summary of Pitot Traverse Row Average Data, 1975 214
-------
ACKNOWLEDGEMENTS
Program personnel wish to thank the instrument
manufacturers whose probes were evaluated for their
help in answering questions which arose during the
testing. Ellison Instruments in particular supplied
several probes and also on-site advice during the
first field demonstration. TRW Systems Group is very
grateful to the Nevada Power Company and the Southern
California Edison Company for their assistance and use
of facilities for the field demonstrations.
xiv
-------
SECTION I
CONCLUSIONS
Desired system accuracy specifications (5% to 10% of full
scale) can be met with hardware and measurement techniques
examined during the program.
Total flow measurement can normally be made with better
accuracy in circular ducts than in rectangular ducts, due
to the way in which the duct itself conditions the flow.
Several point mapping techniques, notably the Log Linear 4
method, are adequate for total flow measurement in cir-
cular ducts, using eight or less sampling points.
The row averaging technique, using eight or less velocity
sensors, is adequate for total flow measurement in rectanq-
ular ducts when supported by in place calibration.
The region immediately downstream of an elbow has been
identified as the most desirable place to determine total
volumetric flow in a rectangular duct when line averaging
techniques are used.
The Ellison Annubar is adequate for measurement in circular
ducts, and also in rectangular ducts when calibrated in
place. A purge capability may be required in moist, par-
ti cul ate laden flows.
The Ramapo Fluid Drag meter had the best characteristics
among the point sensors tested.
The S pi tot probe tested had several undesirable charac-
teristics, including shifts in calibration as a function
of Reynolds number and high angular sensitivity. For
normal manual traversing, it is believed that the
ellipsoidal nosed pitot-static probe can be used to
obtain data several percent more accurate than can be
obtained using an S probe.
The field demonstrations confirmed the basic adequacy
of the techniques and sensors tested.
-------
SECTION II
RECOMMENDATIONS
Results of this program should be coordinated with gas
composition sampling technology to produce a system to
determine both total flow rate and gas composition.
Manufacturers of flow sensors which may be used in the
applications of interest during the program should
consider offering total systems to include measurement,
data processing and readout capabilities. Also, velocity
sensors should have the capability to measure local
pressure and temperature, since these parameters are
always required for data reduction.
Inexpensive, reliable linear traversing mechanisms for
point sensors should be made available as a shelf item.
Consideration should be given to replacing the centroid
of equal areas method by the Log Linear method for
circular duct mapping.
To optimize accuracy in EPA Method 2 (Reference 1), con-
sideration should be given to converting velocity to
standard conditions before averaging, rather than after.
Work should be done to standardize design characteristics
of the S pitot probe to optimize the accuracy.
Ellipsoidal and hemispherical nosed pitot-static probes
were designed to minimize error with respect to the
magnitude of the total velocity vector. For applications
of interest in this program, it would be very desirable
to have a probe which would minimize error with respect
to the axial component of the velocity vector.
-------
SECTION III
INTRODUCTION
The program is concerned with the measurement of total volumetric
flow on a continuous basis. Volumetric flow measurement is important as
an end in itself, as well as being a requirement in support of other
measurements. Examples of the former are total emissions measurement for
regulatory purposes, and monitoring of process stream devices such as
blowers for control and maintenance purposes. Total flow measurement is
required in support of pollutant concentration measurements to translate
a concentration reading into mass output per unit time. Program specifications
are based on continuous monitoring requirements in full scale coal fired
electric utility plants.
The more often a monitoring measurement must be performed, either for
regulatory or control purposes, the more desirable it is to have a reliable,
in place system which can continually perform the measurement without
requiring an operator. In-line flow metering technology for small lines or
long straight pipes is capable of providing reliable, highly accurate
measurements. This program is not concerned with these applications but
rather with ones where the duct size, shape, or configuration makes metering
of the total flow impractical.
The class of sensors considered cons-ists of devices which are inserted
into the flow stream. The total flow rate is then inferred from the readings
of these instruments. Thus the primary program objectives have been to
determine what available instrumentation is applicable to the problem, and
to determine what techniques are required to correctly use the instruments.
The general approach used is similar to that given in the standard
EPA Methods 1 and 2 (Reference 1) -- taking velocity measurements at a
number of discrete points in the flow and averaging the data to obtain a
total flow rate. These standard methods are not generally practical for
continuous monitoring due to the large number of data points usually
required, so one program objective was to determine the minimum number
and location of sampling points required for accurate measurement.
-3-
-------
The program began in October 1972, and the technical effort was
concluded in March 1975. The task breakdown was as follows:
Task I Operating Specifications. This task involved the formulation
of environmental, instrument, and system specifications in order to identify
and bound the problem being considered. The specifications were used in
part to evaluate accuracy, reliability, and survivability of candidate
sensors and measurement systems.
Task II Instrument Selection. This task consisted of a survey of
available instrumentation and subsequent analysis of manufacturers' data
to select those instruments best qualified for further evaluation. A
special effort was devoted to determine the feasibility of using acoustic
flowmeters.
Task III Mapping Technique Evaluation. Task II results clearly
showed that most applicable velocity instruments were point sensors, such
as pitot probes. This meant that development and/or evaluation of flow
mapping techniques in both circular and non-circular ducts was required.
Task III involved development of test facilities and testing to determine
optimum sensor deployment schemes.
Task IV Laboratory Assessment. Instruments selected for further
evaluation in Task II were tested in a low speed wind tunnel to determine
their accuracy characteristics. Environmental tests were also performed
to determine sensor survivability under conditions specified in Task I.
Instruments best suited for field test operation were then selected.
Task V Field Demonstration. Continuous flow monitoring systems
were successfully tested at the Nevada Power Company station at Moapa,
Nevada.
Success was achieved in the development of rectangular duct mapping
techniques which require as few as one velocity sensor regardless of duct
size. Acceptability of mapping techniques such as the Log-Linear method
for use in circular ducts was verified. Acceptable techniques involve
the use of between one and eight velocity sensing devices. Instruments
found acceptable both from the standpoint of accuracy and survivability
include the Ellison Annubar, which was the only non-point sensor tested,
and the Ramapo Fluid Drag Meter.
-4-
-------
EPA Methods 1 and 2 were both examined as reference manual techniques.
It is being recommended that the Log-Linear method be considered as an
improvement over the centroid of equal areas mapping technique for measure-
ment in circular ducts. In addition, the standard pitot-static probe
should be used whenever possible in place of the S type probe in order
to improve system accuracy.
Final results indicate that continuous flow measurement accuracies
of 5% to 10% can be readily attained in non-developed flows in circular
and rectangular ducts using state of the art instrumentation.
-5-
-------
SECTION IV
TASK I - OPERATING SPECIFICATIONS
4.1 GENERAL
The following specifications served as a base for instrument and
technique selection and evaluation. They were obtained in part from the
program RFP and work statement. The flow environment data apply specif-
ically to coal fired electric generating plants.
4.2 FLOW ENVIRONMENT
• Gas Temperature - Range:-18 to 200°Cwith fluctuations of
+17°C in a one hour period.
• Gas Composition
Constituent Concentration Range (Vol)
S02 0-3000 ppm
S03 0-100 ppm
CO 0-50 ppm
C02 12-18%
H20 4-14%
NOX 0-500 ppm
02 1-5%
Hydrocarbon 0-1500 ppm
Np Balance
• Particulate Loading - Range 0-7 grams per standard cubic me-
ter, with particle sizes ranges from submicron to 300y.
Loadings may be exceeded on a temporary basis due to soot
blowing. The particulate material is frequently extremely
adhesive due to the moisture and H2S04 mist content of the
flue gases. Insensitivity or resistance to both fouling
and abrasion are therefore required.
Below 230°C, S03 may be present as H2SO» mist, at a loading of 0-0.1
grams/son.
-6-
-------
• Water Mist Loading - Normally below 0.3 grams/son in process
equipment demister effluent gases. Note: The mists present
in process equipment effluent gases can, depending upon the
abatement process, be contaminated with limestone, calcium
sulfate and sulfite, magnesia, magnesium sulfate and sulfite,
sulfuric acid, sodium sulfate and sulfite, and ammonium
sulfite-bisulfite. Resistance to the effects of these
materials is, therefore, required.
• Pressure - Range 600 - 790 torr, with 27 torr (mm Hg)
vacuum to 36 torr positive pressure departure
from ambient, with pulsations at 30-120 cycles per minute.
• Velocity - Magnitude 1.5-38 meters per second, with fluctua-
tions of j^6 m/s, and flow periodicity of 30-120 cycles
per minute; direction primarily parallel to duct axis
but may be random at any point, including reverse flow.
The following are not specifications but are general application
descriptions.
• Vibration - Maximum frequency 1.0 HZ
Maximum amplitude 1.0cm displacement
• Access - 1-inch pipe nipple to 12-inch x 12-inch inspection
plate
• Disturbances - Eddy currents and static zones due to duct
changes in size and direction both upstream and downstream
of the probe/sensing element zone.
• States of Motion (Flow Types) - Both laminar and turbulent,
with changes between flow types occurring as a function of
changes in plant load.
• Duct Sizes - Pilot scale, range 7-70 cm. diameter. Full
scale 1-12 meter diameter.
• Duct Shapes - Circular, rectangular, and irregular.
• Duct Axis Orientations - Vertical, horizontal, and at any
angle between vertical and horizontal.
-7-
-------
• Length of Straight Run (Ducting) - 0-10 equivalent duct
diameters between direction or cross-sectional shape
changes
4.3 EXTERNAL ENVIRONMENT
The following conditions are applicable to parts of the system not
in the flow.
• Temperature - Range -29°C to 55°C, with direct sunlight on exposed
components
• Pressure - Range 600 to 790 torr, with fluctuations of +25 torr
• Relative Humidity - 0-100%, with conditions such that
condensation forms water or frost
• Moisture - Rainfall, snow, and melting snow
• Vibration - Amplitudes as high as 1.0 cm displacement,
and frequencies as high as 1.0 KHZ
t Electrical Power - 110V or 220V AC, 50-60 HZ, with +20%
fluctuations
4.4 INSTRUMENT SYSTEM PERFORMANCE SPECIFICATIONS
4.4.1 Definition of Parameter to Be Measured
The term "total volumetric flow" must be properly defined in order
to have unambiguous physical significance. From a fluid mechanics
standpoint, it is simpler to work in terms of total mass flow rate, and
then define volumetric flow rate in terms of the mass flow. Consider the
duct shown in Figure 1. The four sides and the entry and exit planes form
a control volume. By definition, the sides are solid, so that all fluid
must enter through the left plane and leave through the right plane. For
simplicity (which does not compromise accuracy), assume that the flow
rate into the control volume is always exactly the same as the flow rate
out of the control volume. This relation becomes true as the size of
the control volume approaches zero. The flow through the control volume
may then be given as the flow through the exit plane of the control volume:
-8-
-------
Mean Flow
Direction
Flow enters from the left and exits
to the right.
Velocity at a point in the exit plane is given
by
tT = u T + v ~j + w k
where
T, J, "k* are unit vectors in the directions
shown, forming an orthogonal coordinate system
and
u, v, w are the scalar components of u in
the ~t, *^ and Tc directions, respectively
The vector~n is the unit vector normal to the
exit plane, so that
—» . .A
n = i
and the net flow component out of the duct at the point
shown is
TT . ~n = u
Figure 1. Flow measurement control volume
-9-
-------
m = //ptT • rT dA = pu A (1)
A
where
•
m = mass flow rate, gm/sec
•3
p = local fluid density, gm/cm (gas phase only)
~u = local velocity (vector), m/sec
"n~ = unit vector normal to exit plane, dimensionless
2
A = exit plane area, m
u = tT • "n", scalar velocity component normal to exit
plane, m/sec
This equation and its application in terms of hardware became the base on
which much of the program was built. The average of the product of the
normal velocity component and the fluid density is exactly defined by
this equation, and the proper definition of the terms "average velocity"
and "volumetric flow" must in turn follow from the definition of p"u~.
It is common practice to evaluate equation (1) in a large pipe or
duct by performing a velocity traverse, for example using EPA methods 1
and 2. The integration in equation 1 is then approximated by the
summation
N
m; !>nVAn (2)
n=l
for N area segments. Usually the segments are of equal area, so that
N
"• = & X>nun
n=l
In terms of commonly measured parameters in a gas, this becomes
n=l
where
p = gas static pressure, torr
-10-
-------
2
R = gas constant, e.g. in
sec °K
T = gas static temperature, °K
The flow through any one area segment is then assumed to be
\ - fT R^ un ' VVA
In order to switch from mass flow rate to volume flow rate, the concept
of "standard conditions" needs to be introduced. This is illustrated in
Figure 2. For this control volume, the gas enters at arbitrary pressure
and temperature. It then undergoes whatever changes are required in
order to emerge uniformly with a static temperature of 20°C and at an
absolute static pressure of 760 torr, which have been defined as standard
conditions for this program. Also by definition, the gas is considered
to be chemically frozen and there are no phase changes. This is important
from a practical measurement standpoint. It means that the mass flow rate
being considered is that of the gaseous components only, since common
velocity sensors for operation in a gas stream respond to gas flow, and
not to liquids or solids which may be entrained in the flow. Therefore
defining a gas flow in terms of standard rather than actual conditions
implies changes in pressure and temperature only -- not composition and
phase. Thus for any one area segment where the velocity and density are
considered to be uniform,
m = pUAA = p u AA (5)
where
( ) = value at standard atmospheric conditions.
Given this background, it is now possible for us to define volumetric
flow rate and average velocity.
For any one area segment, the volumetric flow at standard conditions
is given as
V, = a. AA (6)
sn n
-11-
-------
//
J
X
X
X
X
X
TEMPERATURE
PRESSURE
DENSITY
VELOCITY
AREA
/ /
ml_
1
X
X
X
X
s
,
my
-- ~
7
x
X
X
X
X
STATION STATION
1 2
(TEST) (STANDARD)
Ti TS
Pi Ps
Pl ps
V] Vc
s
A A
.Xi
>
/
/
X
FROM CONSERVATION OF MASS,
= m
SO
AND
PI vi A = PS vs A
PI
Figure 2. Duct analogy for transformation of test
velocity to standard velocity
-12-
-------
where
V = volumetric flow rate at standard conditions
n 3
m /sec
and volumetric flow at actual conditions is given as
V = u M m
n n
where
Vn = volumetric flow rate at actual conditions,
3
m /sec
For the term V to have an unambiguous meaning, the actual flow temperature
and pressure must always be given and must be constant for the region of
interest. In the term Vg, the pressure and temperature are given by
definition. For the entire duct as shown in Figure 1, we have
where V is the total volumetric flow rate at standard conditions. There
^ •
is no directly corresponding equation for V except for the special case
of zero density stratification. This is explored in more detail in
Appendix A. All volumetric flow data presented in this report are given
as volumetric flow rate at standard conditions unless otherwise noted.
In summary, the following relations are those used in the program:
ms = pu~ A Total mass flow (1)
Vs = uj A Total volumetric flow at (8)
standard conditions
and by definition are related as follows:
m = puA = P~~ u~ A = P~~ V (9)
b bo b 5
4.4.2 Preliminary System Error Analysis
Recall from equation (4) that the mass flow rate through any one
area segment during a traverse is given by
• _ A p
01 " N ET u
-13-
-------
The uncertainty in this single point measurement may then be given by
(see Reference 2 and Appendix A):
02 =(Mr 02 + fam r 2 . [am I2 2
mn \9A/
+
where
o = standard deviation of quantity x
/\
If a measurement system is postulated which determines each of the varia-
bles independently of the others, the above equation can be written
222222
O' 0« 0_ On OT 0.,
m_A_£ R T u /,,x
To" ~ o" ";> T 5" "~o ?" VI I /
nr A p R£ r u
Under the best of circumstances, the following assumptions may be made
for normal field measurements:
1. OA = .01 A
This is based upon the assumption of a 2a error of 1% in the duct
linear dimensions.
o o Op aT
-f-T'-T- -005
These are based on 2o accuracy of 1% for each case.
Thus we have
2 2
o- o o a
?5-= (.Ol)2 + 3(.005)2 + —SJ-
m u
a'2 2
= .000175 + a u
.2 — p"
m
-14-
-------
The equation can now be examined in terms of velocity uncertainty,
which is of central importance to the program. For zero uncertainty in
velocity, we get
2am = ± .026 m
i.e. a 2.6% uncertainty for a 95% confidence level. For 2a = .Olu, .05u,
and .10u, respectively, we obtain 2am = .028m , .056m , and .103m . This
shows that under optimum circumstances, a local accuracy of about 3% can
be attained when the accuracy of each individual measurement is 1%. It
also shows that if the velocity accuracy is considerably worse than the
others, the system uncertainty is only slightly greater than the velocity
uncertainty.
The above analysis is for a single point measurement only, and does
not consider errors resulting from differences in point measurement and
true average flow through the area segment, time lag error, and other
error sources, which add to the minimum 3% error. The analysis shows the
parameters which must be involved for any orthodox measurement system,
and an estimate of the best results to be reasonably expected. Within
this scope,the following specifications are recommended for individual
components and the combined system.
4.4.3 Component Specifications
• Range - any component such as a thermocouple, pressure
transducer, etc., must be compatible with the range
given in Section 4.2.
0 Accuracy - Within +3% of full scale reading for velocity
instrument subsystems (such as pitot probe-pressure
transducer combinations) in a uniform flow field; within
+2% of full scale reading for support measurements such
as absolute temperature and pressure.
• Velocity probe angular response - For a velocity probe
aligned with the duct axis, error in measurement of
axial velocity component should be less than +5% for
off-axis flow angles up to 10°.
-15-
-------
4.4.4 System Specifications
• Accuracy - Within +_5% of values obtained by pitot trans-
versal in circular pipes with straight runs greater than
ten pipe diameters; within +_10% of values obtained by
pitot traversal in circular pipes with less than 10 pipe
diameters of straight run or in noncircular ducts.
• Desired repeatability = +_2% of full scale
• Desired zero and span drift - Zero drift <+1.5%/24 hour;
full scale drift <+_1.5%/24 hour (with standard mainten-
ance) ; and <+3% week.
• Desired response time - <30 seconds for 95% of full scale
flow change.
• Calibration - maximum of once per week, by pitot tube
traverse. Note: Mass balance methods may be used as
alternative reference techniques where data of sufficient
accuracy is available.
• Maintainability - Routine maintenance < one-half hour
per eight hour shift (probe installed and available for
use). Refurbishment ^12 man hours per month (probe
installed and available for use).
• Readout - Readout in plant control room with continuous
printout on hard copy as volumetric gas flow per unit
of time versus calendar (or chart) time.
-16-
-------
SECTION V
TASK II - INSTRUMENT SELECTION
5.1 LITERATURE BIBLIOGRAPHY
A literature search was performed to gather data in the following
general areas related to the program:
• Instrumentation (Vendor catalogues)
• Velocity Measurement Techniques
• Instrument Survivability Problems
• Stationary Source Air Pollution
t Gas Sampling Techniques
• Thermodynamic Parameter Measurements
• Power Utility Characteristics
t Error Propogation
Reports were categorized, numbered and cross referenced by computer.
Vendor data was the most used in order to produce the instrument survey
described in the following section.
Three reference books should be singled out due to their importance
during the program. The first is "The Measurement of Air Flow," by
Ower and Pankhurst (Reference 3). The book deals in detail with many
common velocity measurement devices, including the pitot probe and its
variations. It also examines flow in pipes and associated measurement
techniques. It was the best single source of information found during
the program. More technically detailed information for some specific
problems was obtained from "Boundary Layer Theory," by Schlichting
(Reference 4), including Reynolds principle of similarity, turbulent pipe
flow, and flow through pipes of non-circular cross-section. Error
analysis techniques were adapted from "The Analysis of Physical Measure-
ments," by Pugh and Winslow (Reference 2), which provides an excellent
insight into system error propagation.
-17-
-------
5.2 INSTRUMENT SURVEY
A greater variety of instruments is available for flow measurement
than for the measurement of practically any other engineering parameter.
A literature and vendor search turned up the list of "key words" shown
in Table I to categorize various types of flow instruments. The general
industry survey results are presented in Appendix B, which lists manu-
facturers, medium (gas, liquid or solid) for instrument use, general
instrument classification as given in Table I, and the appropriate TRW
file document number. This list was compiled in late 1972 and therefore
does not contain any new instruments produced since then.
5.3 INSTRUMENT SELECTION
Two simple elimination steps were used to narrow the field of
instruments to those most applicable to this program. The first was
procedural, and that was to consider only gas flow measurement instruments,
which is in line with program specifications. The second decision was to
eliminate from consideration instruments which are designed solely for use
in either small or long, straight, circular pipes, such as orifice or
venturi meters or other metering systems which process the total flow.
This program is concerned with applications where the duct may not be
round, straight runs are very short, and characteristic dimensions of the
ducting make an instrument to process the total flow impractical.
The remaining instruments can be characterized most generally as
those instruments which react to localized conditions in the flow, usually
by being immersed in it, in a way which allows estimation of the total
flow by inferring a relationship between the local and total flow. For
example, in a pitot traverse, the total flow is determined by assuming
that the average flow at the discrete sampling points is equivalent to
the total flow. The instrument may make measurements at a single point,
such as a stationary pitot probe; at a number of points, such as in a
pitot traverse; along a line, such as an acoustic velocimeter, or in a
plane, such as a drag meter with a large target.
After the initial list was narrowed to fit the scope of the program,
a final selection of probes for test was performed. Selection was based
largely on user recommendations when available. The instruments chosen
-18-
-------
Table 1. KEY WORD CLASSIFICATION FOR FLOW INSTRUMENTS
Number
Type
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
electromagnetic
flow tubes
laminar
mass
non-obstructing
nozzles
open channel
orifice meter
pi tot-type
positive displacement
turbine
variable area rotometer
Venturis
thermal anemometer
mechanical anemometer
acoustic
vortex shedding
other
•19-
-------
and the reasons for their selection are given below, the reasons based
on information available at the time, late 1972. The state of the art
of acoustic velocimeters was specifically evaluated as a separate sub-
task, and is discussed below in Section 5.5,
5.3.1 S Type (Stauscheibe) Pitot Probe
Since this is the most commonly used field instrument for point
velocity measurement, it was determined that its properties should be
examined. The specific probe used was provided by the Ellison Instrument
Division of the Dieterich Standard Corporation. The probe and its
specifications are shown in Figure 3.
5.3.2 Hastings-Raydist Flare Gas Flow Probe (Figures, 4, 51
The probe is actually a type of pitot probe that is constructed
with two openings at the probe tip. These openings are connected together
by an internal stainless steel tube. A portion of this tube is heated and
thermo-electric sensors measure temperature gradients along the wall of
the tube, external to the flow stream. Purge gas is injected into the
tubing in an arrangement which forms a pneumatic bridge in a manner similar
to that of a gas density balance widely used as a detector in gas chroma-
tography. At zero line velocity, the bridge is balanced and purge gas
exhausts out both openings at the probe tip equally.
As flow across the tip occurs, a differential pressure is developed
as with any pitot type probe, unbalancing the bridge. Purge gas still
exhausts out both openings, but now they are slightly unequal. The thermo-
electric sensors measure the shift in temperature gradients along the
heated portion of the tube which are related to the main gas flow creating
the differential pressure at the tip.
Because the purge gas is continuously exhausted into the flowing gas
whose velocity is being measured, corrosive or particulate laden gas is
prevented from entering the probe to produce fouling. For this reason,
the Hastings Mass Flow Probe has found wide use in refineries for measuring
the velocity of gases in flares contaminated with tar-like material.
The Hastings Mass Flow Probe is normally calibrated for air in a
wind tunnel using either air or nitrogen as the purge gas. A calibration
-20-
-------
2000 SERIES —TECHNICAL DATA ON
COMBINED REVERSED PITOT TUBES
DIGTEHICH STANDARD CORP • DRAWER M • BOULDER COLORADO 3O3DB USA • TEL 3O3 m«7-1OOO • TLX OB-BO3 • CABLB OIBBTAN ODO
COMBINED REVERSED PITOT TUBE
The Ellison Combined Reversed pilot tube is designed to the
Stauscheibe concept. This pilot tube has numerous design fea-
tures. It can be used to measure many liquids, clean or dust-
laden air, and steam. It also produces a higher velocity pressure
or differential gage reading for a given flow rate which allows
measuring of lower flow rates and produces a more accurate
measure of all flow rates. If the large sensing tubes need clean-
ing after use, they can be easily cleaned without special tools.
A special water-cooled version of the Combined Reversed type
is available for use with temperatures up to +2000*F. These im-
portant features have made Ellison's Combined Reversed pitot
tube the choice of many flow specialists around the world.
Combined Reversed Type
INSTALLATION PROCEDURE
As with all pitot tube installations, it is of primary importance
that the impact tube be pointed directly upstream. This means
that the opening of one of the bent tubes on the Combined Re-
versed type points directly upstream, and the other points down-
stream. To insure a uniform flow and accurate measurement, it is
desirable to have as much unobstructed pipe or duct as possible
upstream from a pitot tube as well as a reasonable distance
downstream. Unobstructed pipe or duct means no bends, tees,
valves, or changes in diameter. For air or gas, 15 to 25 diameters
of the pipe or duct, depending on velocity are sufficient for the
upstream side. For liquids and steam 20 to 50 diameters are
desired. Downstream lengths should be Mi to Vz of the upstream
length. If the above suggested lengths are not possible, a com-
plete traverse is recommended. It is very important that the rate
of flow be stabilized during the time a complete manual traverse
is being made. Any change in velocity will greatly affect the
accuracy of the calculated average velocity obtained from the
traverse.
FORMULAS FOR VELOCITY AND VOLUME
Velocity —Combined Reversed Type Only
Gas Flow V = 903.4
Air Flow* V = 3300
Liquids —Wet Seal, using Hg V = 3008
Water" —Dry Seal V = 107.5
V = 395.9
Water'* —Wet Seal, using Hg V .- 381.1
SELECTING A MEASURING
INSTRUMENT
The pitot tube may be used with a wide variety of measuring,
recording, or controlling instruments. It is most commonly used
with an Inclined manometer or a Vertical manometer. Ask for a
copy of Ellison's new General Catalog. To determine the maxi-
mum scale range of the instrument you will need, use the follow-
ing formulas:
hw = (V/3300)1 for Air.
hw = d(V/903.4)' for Gas.
hm = (V/381.1)' for Water-Wet Seal to Mercury (only)
hw = (V/107)' for Water -Dry Seal
hm = (V/395.9)1 for Water —Dry Seal
Terms—
V = velocity in feet per minute
d = density in pounds per cubic foot
hw - differential pressure in inches of water
hm = differential pressure in inches of Mercury
A = inside area of pipe in square feet
a ~ inside area of pipe in square inches
CORRECTIONS FOR NON-STANDARD CONDITIONS
—ALL PITOT TUBES
AIR—For calculating Air Flow at temperatures and pressure other
than at standard conditions, use Gas Flow formula. Air Density or
d = 1 325 x Inches of Mercury (Absolute) in pipe
459.6 + Temperature in *F.
WATER — For calculating Water Flow at temperatures other than
60*F., use Liquid Flow formula. Water densities, d, are shown
below for various temperatures.
Tamp.
+ 32'F.
40
50
60
70
80
90
100
110
120
Deniitxld]
62.42
62.43
62.41
62.37
62.30
62.22
62.11
61.99
61.86
61.71
Temp.
4-130'F
140
150
160
170
160
190
200
210
Oeniity Id)
61.55
61.38
61.20
61.00
60.80
60.58
60.36
60.12
59.88
-At standard conditions (d = .07495). +70'F. 29.92" Barometer, see correc-
tion formulas for other conditions.
"At 60*F. with d =62.37. Dry seal is when manometer is calibrated dry
and water does not contact indicating liquid during use. wet seal is when
manometer is calibrated dry, and water does contact indicating liquid
during use.
Figure 3. Combined reversed pitot tube
--21-
-------
HASTINGS-RAYDIST
Specification Sheet No. 513A
HASTINGS GAS FLOW PROBE MODEL API SERIES
FOR MEASURING VELOCITY OF WET AND DIRTY GASES
v. RANGE: 0-1000 fpm OR 0-6000 fpm of* o-/0,000
A DEPENDABLE, NON-CLOGGING FLOWMETER
FOR CONTAMINATED GAS LINES.
FEATURES
• CONTINUOUS PURGE PRINCIPLE
• NO EXPOSED SENSORS 01 WIRES
•EASILY INSTALLED or MOVED
•EXPLOSION-PROOF TYPE HOUSING
• 0-5 VOLT D-C OUTPUT SIGNAL
• PROVIDES LONG LINE TRANSMISSION
CAPABILITY
•REMOTE: RECORDING, CONTROL, ALARM,
INDICATION
•PURGE WITH AIR, N? OR PROCESS GAS
• CHOICE OF TWO RANGES: 0-1000 or 0-6000 fpm
GENERAL
The Hastings Gas Flow Probe is the result of
nearly two decades of experience in dealing with
difficult-to-measure, corrosive, or inflammable
gases. Using a unique Hastings thermal prin-
ciple with continuous purging, the measured gas
does not come in contact with the internal parts
of the probe. Thus plugging, fouling, con-
densation, corrosion, etc., are no longer prob-
lems.
The Probe is constructed entirely of stainless
steel for all parts (internal and external)
through which any gas flows. Solid-state circuits
are built into the explosion-proof type housing
and require only connection toa 24 volt d-c power
source and remote read out. The output signal of
0-5 volts may be connected to any remote data
logging device, meter, recorder or readout
desired. The purge gas required is quite small
and normally is less than 30 cu.ft. / hr. The probe
may also be used without the purge system in
relatively clean and dry lines where plugging is
not a problem.
PRINCIPLE OF OPERATION
CONTINUOUS PURGE MODE
The probe is constructed with two openings at
the probe tip. These openings are connected
together by an internal stainless steel tube. A
portion of this tube is heated and thermo-electric
sensors measure temperature gradients along
the wall of the tube, external to the flow stream.
Purge gas is injected into the tubing in an
arrangement which forms a pneumatic bridge.
At zero line velocity, the bridge is balanced so
that no flow occurs through the sensing portion of
the tube. The purge gas exhausts out both
openings equally at the probe tip.
As flow across the tip occurs, a differential
pressure is developed, unbalancing the bridge
and causing a small amount of purge gas to flow
through the sensing section. Purge gas still
exhausts out both openings, but now they are
slightly unequal. The thermo-electric sensors
measure the shift in temperature gradients
along the heated portion of the tube which are
related to the main gas flow creating the dif-
ferential pressure at the tip.
Since the purge gas is continuously exhausting
into the main line, it prevents the main line gas
from entering the probe and prevents fouling.
1,1
p= nc\ ="t lS?i -v
s
;
[
tuGlNG
E 14PS
OMPOtSSlO*
"^ *
:
NSOB
P^>
'-^ t
5 ; 1
]
djRCE GAS
'LOW MOC
MAMlFOLO
AT ZERO Fl'OW
AP • &P?
AP. ' fiP4
NO TjO* "^S1 SErrtO^i
APL> ZERO
flT MAIN LINE FLOW
APL IS DEVELOPED. SO
AT ALL T.MES RJ»0£
C4St«HAUSIS iNIO
PWOBE TiPQFlNNGS
0, «NO 02
MAIN |
— LINE /
CONTINUOUS PURGING PRINCIPLE ot OPERATION
( Manufactured under one or more U.S. Patents and Pat. Pending)
Figure 4. Flare gas flow probe
-22-
-------
PURGE GAS REQUIREMENTS
Regulated, clean dry air, nitrogen or process
gas may be used as the purge gas. Consumption
is less than 30 standard cu. ft. / hr. at a pressure
approximately 15 psi above the static pressure of
the line being measured. Since such a small
amount is flowing into relatively large lines, the
percentage of air added to the whole is in-
significant. If inert gases are desired, nitrogen is
recommended.
PURGE MANIFOLD
Included with the probe is a purge gas
balancing manifold built into a weatherproof
conduit type enclosure. This may be mounted
anywhere near the probe. It includes two valves
tor balancing the bridge and zeroing the probe.
POWER CONVERTER
An optional 24 volt d-c converter is offered for
those installations where 24 vdc is not available.
Built into a Grouse Hinds explosion-proof type
housing and rated at Vj amp, it is available for
use from either 115 volt or 230 volt a-c lines.
CALIBRATION
Calibration is related to the gas density and the
velocity profile.
Velocity = KrK?-V1NO
where
K, =Velocity Profile Factor; typically .8
K, = Density Factor; V .075/ gas density
VINO = Velocity from probe calibration curve
Example: What is the full scale (5 volt) range of
an AFI-6K when measuring stack gas having a
density of .092 Ibs./ft.?
Velocity = K, • K?^V|ND
= (.8) ( V .0757 .092 ) (6000)
= 4333 fpm
TYPICAL CALIBRATION CURVE
SELECTION CHART
Model Air Range
AFI-1K 0-1000 fpm
AFI-6K 0-6000 fpm
Above includes: Probe, Purge Minifold, and curve ot
Air Velocity versus 0-5 volts.
ACCESSORIES
Model AOC-2 115 v. i-c/24 v. d-c power converter •
Model ADC-3 230 v. i-c/24 v. d-c power converter
Meter 24-1-419 0-5 volt meter for remote reidout
Meter 24-1-420 0-5 volt meter relay single point control
Meter 24-1-421 fj-5 «olt meter relay double point control
SPECIFICATIONS
POWER : J« V. a-c ( ' 4 V.) (" JIO ma
OUTPUT: 0-S v. d-t. ft 4 ma (m»).
PURGE GAS: API.IK 5 clh
AFI-4K 30 clh
<> IS psig regulated
DIMENSIONS: ?robe—W . «" « «" Overall
with U" • u «" o.O. Wand .
MANIFOLD: «" I »" > IV," Weatherproof Type Bo«
MATERIALS: 104 and lit Stainless Steel lor
all parls in contact with gas
HOUSING: Croine Hindi Type
COMPRESSION SEAL FOR PROBE: Male l»" NPT Threaded
Connection
MANIFOLD CONNECTIONS: Vi" O.O. Tubing to Probe, H" NPT to
Purge Gas Supply
INSTALLATION
The probe is supplied with a compression seal
fitting for easy installation. The fitting has a liv
NPT male thread that will easily connect to a
I'/i" female threaded gate valve. It may be
mounted in any position if used in the purge
mode.
The purge manifold is connected to a regulated
gas supply at 10-30 psig. The manifold should be
connected to the probe by means of V&" O.D.
tubing.
Electrically, 2 wires from the probe are
required for the 24 v. d-c input power, and 2 wires
for connection to the remote read-out.
9
0)4
23
z
r-2
2
3 1
o
0
(
AIF
/
y
>
I VELI
IT
,/
/
/
•^
^.^^—
2 3 4 5 (
DCITY IN THOUSAND OF FPI
HASTINCS-RAYDIST
A TELEDYNE COMPANY
\r1
-.-.- J:
OUTLINE
. .
1 CCMLtT • "' — C
t / , f 1
" "' .'i/" - '"'
DIMENSIONS API SERIES
Literiture lUiilible upon request:
Hjslingi Vicuum Saugn Citaloj No. 3000
Hastings Mcleod Gauge Spec. Sheet No 3408
Hastings Gauge Tube Accessories spec. Sheet No. 35!
Hasting;, Vicuum Gauge Reference Tubes Spec. Sheet No. 351*
Hastings Air-Meiers Calalog No. 4000
Hillings Mass Flcmmeters lor (Uses Calalog No. 5000
Hastings Cilibnted Gis leiis Spec. Sheet No. 9040
SPECIFICATION SHEET NO. 513A PHONE 703-723-6531
HAMPTON, VIRGINIA 23361 TWX: 710-882-0085
PRINTED IN U.S.A. COPYRIGHT O 8-72
Figure 5. Flare gas probe specifications
-23-
-------
curve is obtained for air velocity vs output in voltage. Smaller changes
in voltage will be noted as the velocity increases.
While air is usually the most convenient purge gas to use, other
gases may be used to obtain different sensitivities. Nitrogen is used to
replace air with no major change in calibration. Most other gases such as
methane, propane, or natural gas may also be used but the original air
calibration curve is no longer applicable. It is not possible to multiply
the air purge curve by a constant factor to obtain the purge curve for
some other gas. Changing the purge gas only changes the relationship
between output and voltages and the indicated velocity. The sensitivity
of this relationship is affected by the thermal conductivity of the purge
gas with methane producing a greater sensitivity than air.
The Hastings Probe is capable of measuring the velocity of stack
gases from 0 to 30 m/sec, but the accuracy becomes poorer above about
20 m/sec. The Hastings probe is designed to operate at a temperature
range from -34°C to +315°C with the electrical zero shifting approximately
5% over this range. In addition, as with most flow measuring devices, the
gas density must be determined to correct the indicated velocity measure-
ments. The variations expected in gas composition present no survivability
problems with the Hastings Probe since the probe is constructed of stainless
steel and the purge gas continuously exhausting into the stack prevents
the stack gases from entering the probe and fouling the system. Because
the measured gas does not come into the internal parts of the probe and
the probe is continuously purged, water mists, limestone, calcium sulfate
and sulfite, sulfuric acid mists, sodium sulfate and sulfite, and ammonia
sulfite-bisulfite resistance appears assured. One user reported use in
the presence of zinc oxide dust with oxides of sulfur and moisture being
present in the measured gas stream.
Positive and negative factors identified for these sensors are as
fol1ows:
Positive Factors
0 Probe can be used in extremely wet, corrosive and particulate
laden gas streams; Q-J grams of parttculate per cubtc meter
of gas is acceptable
-24-
-------
• Calibrations should remain constant for six months
• The probe is very sensitive to changes in flow rate at low
flow rates
• Accuracy of full scale voltage readout is reported by the vendor
to be +2% of full scale
Negative Factors
• The velocity vs voltage output curve is non-linear; changes
in velocity at the low end of the scale cause the greatest
change in voltage
• The high sensitivity of the probe at low velocities tends to
accentuate small zero drifts. One user found probe unsatis-
factory because of zero shifts
t The electrical zero will shift 5% over the operating tem-
perature range
• The probe must be isolated from vibrations
t Users contacted indicated initial setup and adjustment
problems
• Gas velocity range between 0 and 20 m/sec can be measured
with good accuracy; between 20 and 30 m/sec measurements
have less accuracy
^
5.3.3 Ramapo Mark V Flowmeter (Figure 6)
Ramapo manufactures the Mark V Flowmeter Probe that measures flow in
terms of dynamic forces acting on a fixed body in the flow stream. Bonded
strain gauges, in a bridge circuit outside the fluid stream and shielded
by stainless steel, translate this force into an electrical output pro-
portional to the flow rate squared. The electronics used are identical
to those used for pressure transducers. Gas velocities of 1.5 to 38 meters
per second can be measured, but the probe has a turn down ratio of 10:1
which means that two different probes will be required to cover the entire
range.
Gas temperatures from 0 to 260°C are acceptable with the Ramapo
probe. Corrections to the indicated velocity must be made for changes in
-------
Figure 6. Drag meter
-26-
-------
density. Particulate loadings of 0 to 15 grains per standard cubic foot,
with particle size ranging from submicron to 300 micron are not expected
to be a problem unless the sensing head becomes excessively loaded.
Positive and negative factors identified for these sensors are
as follows:
Positive Factors
• Particulate loadings of 0-7 gm/scm will be acceptable
if the disc in the probe is coated with Kel-F, or if
provisions for steam cleaning are included
• Resistant to chemical attack (SO^, NO , H^O, etc.) if purge
is used in internal cavity
• Kennecott Copper of Salt Lake City rated probe as their
primary selection for use in smelter stacks
t TRW has a good service record with device in measuring the
velocity of liquids
Negative Factors
t A purge may be required to protect parts from chemical
attack
t One probe will not cover the velocity range of 1.5 to
38 m/sec, as identified in the specifications
• Limited previous applications for stacks
5.3.4 Thermo-Systerns Incorporated Hot Film Sensors (Figures 7-9)
Thermo-Systerns manufactures a velocity transducer that measures
velocity, based on calibration at a given temperature and pressure, by
sensing the cooling effect of a moving flow stream over a heated sensor
surface. The sensor is heated electrically by current from a control
amplifier. The sensor is one leg of a bridge which is continuously bal-
anced. Another leg of the bridge serves as a temperature sensor to com-
pensate for temperature changes. The output signal is proportional to
the square root of velocity. Standard probes are subject to calibration
shifts due to particulate buildup on the probe.
-27-
-------
NEW
METAL BACKED
HOT FILM SENSOR
- LARGE, RUGGED,
FILM SURFACE
- HEAVY QUARTZ
COATING
NEW
METAL CLAD SENSOR
- RUGGED
- METAL ENCASED
SENSOR
A. METAL CLAD SENSOR
B. METAL BACKED SENSOR
WEDGE HOT FILM
OUARTZ R
GOLD FILM ELECTRICAL LEAD*
-OUARTZ COATED PLATINUM
FILM 10.004*1 0040" EACH SIDE)
C. WEDGE SENSOR
Figure 7. Thermo systems hot film sensors tested
-28-
-------
Electronic Control Circuit
" and Bridge Built Into Probe
All Stainless Steel,
High-Reliability Parti
Sensor
VT 161 FEATURES
Self-Contained — Bridge and
Electronic Control Integral
with Probe
Sensitivity from a few ft/min.
to over 200 ft/sec.
New RDM' Bridge System Gives
High Stability - No Controls or
Adjustments Needed
Operates on Batteries or
115/230V AC
Linear Output Available
Temperature Compensated
Calibration Included
Cable Length Not Critical
THE VT 161 IS AVAILABLE
AS FOLLOWS:
1. Probe Package for 12V DC
Power - Non-Linear Output -
No Readout
2. Probe Package for 115/230V AC
Power Using Model 1605 Power
Supply — Non-Linear Output -
No Readout
3. Linear Signal Conditioner with
Probe Package - Analog Readout
4. Linear Signal Conditioner with
Probe Package — Digital Readout
•Patented (Pulse Duration Modulation!
NEW
METAL CLAD SENSOR
- RUGGED
- METAL ENCASED
SENSOR
The first really rugged thermal sensor-
takes advantage of the wide range.
stability and fast response of thermal
techniques without fragility.
WIDE APPLICATION
- TESTING
- RESEARCH
- CONTROL
Ducts, Pipes, etc.
Towers
Field Studies, Remote Points
Mobile Equipment
Multi-Point Studies
VELOCITY
TRANSDUCER
SERIES VT 161
AIR
AND OTHER GASES
0 - 200 ft/sec
THERMO-SYSTEMS Inc.
2500 Cleveland Ave.North
St. Paul, Minnesota 55113
(612) 633-0550
Telex (297—482)
Form No. S161-172
Printed in US.A.
Figure 8. TSI metal clad sensor and anemometer
-29-
-------
MODEL VT 161 VELOCITY TRANSDUCER
PROBE PACKAGE FOR 12V DC POWER SOURCE - NON-LINEAR - NO READOUT
PRINCIPLE OF OPERATION
The VT 161 measures velocity (actually "standard"
velocity based on calibration at 70 F and 1 atmosphere
or mass flow) by sensing the cooling effect of a moving
flowstream over the heated sensor surface. The sensor
is heated electrically by current from the control
amplifier. The sensor is one leg of a bridge which is
continuously balanced by the unique, POM* amplifier
system, requiring no adjustments. Another leg of the
bridge serves as a temperature sensor to compensate
for temperature changes. The output signal is propor-
tional to the electrical power dissipated in the sensor
or about the square root of velocity. If desired a linear
output is provided by a very accurate curve-fitting
amplifier that is part of the Signal Conditioner pack-
ages shown below.
SPECIFICATIONS
Velocity Range: 0 to 200 ft/sec (600 mpsl. (Dete n furnished to
give good accuracy throughout the above range. Any specific
smaller range can be utilized. Calibration is at standard condi-
' tions of temperature and pressure.)
Output: Approx. 0.5 to 3V DC. Impedance - 100 ohms
Accuracy: ±0.1% of FS and ±3% of Reading
Repeatability: leu then 0.1% (repeated conditions)
Response Time: 50 milliseconds
Environmental Conditions (Senior Section): Temperature -20°C to
100°C, Pressure 0.01 to 500 psia
Cable: 15' Standard (specify otherwise)
Power Supply: 12V *?'2 V DC (+15V DC optional). 0.5 Amp. Max.
VCLOCITV (FT/MCI
TYPICAL CALIBRATION
CURVE-VT161
MODEL 1605 POWER SUPPLY
Converts 116/230V DC
power to 12V DC for op-
eration of VT 160 Proem.
Convenient terminals serve
as junction point for signal
leads
MODEL VT 1612 PROBE WITH LINEAR SIGNAL CONDITIONER-ANALOG READOUT
SPECIFICATIONS
(Signal conditioner includes power supplies, linearizing circuit*,
output amplifiers and analog meter.)
Analog Output: Rear Terminals; 50 ohm impedance; 0 — 6V DC
full scale range; 50 millisecond response time; and total linear
accuracy ±5% of reading in 50 to 1 range.
Analog Meter: 0 - 100% and 0 - 10% of full scale. ±2% fS.
Cabinet: Aluminum, portable, with stand — 8.5 x 3.5 x 11
Power Required: 115/230V AC
Probe Specifications: Same as VT 161 except for linear output and
AC power. The VT 161 Probe can be retrofitted later to make a
VT 1612 or VT 1613 system.
Model No.
VT 1612-1
-2
3
-4
•5
Range
0- 1000 fpm
0- 10mpi
0 - lOOfps
0 - 10,000 fpm
0- 100 mps
(other ranges may be specified as needed)
MODEL VT 1613 PROBE WITH LINEAR SIGNAL CONDITIONER-DIGITAL READOUT
SPECIFICATIONS
(Same specifications for Probe, Lmeanzer and Analog output as
above.)
Digital Display: 3'/. digit, neon; 100% over range; 2 second response
time, 0.2 second/reading display, and il digit resolution. BCD
output optional.
Ranges: Modal No. Range (direct reading)
VT 1613-1 0-2000 fpm
•2 0 - 20.00 mps
-3 0 - 200.0 fpi
4 0 - 100.0 mps
(other ranges may be specified as needed)
tsij THERMO-SYSTEMS INC.
3000MOUTH CLEVELAND * VC.
ST. PAUL. MINNESOTA SSI 13
612633 0660
Figure 9. TSI specification sheet
-30-
-------
Positive and negative factors identified for these sensors are
follows:
Positive Factory
• Can measure velocities of 1.5 to 38 m/sec
• Changes in pressure make only small changes in measured
velocity
t TRW has had experience with advanced designed transducers
that operate at high temperatures. The experience indicates
the transducers may operate in a stack environment
• Transducers can operate above 200°C
Negative Factors
• Output in volts vs velocity is not inherently linear, sensi-
tivity decreases at higher velocities
• Sensors are subject to fouling by particulate
• Gas viscosity and thermal conductivity must be known or
sensors must be calibrated in place
5.3.5 FluiDynamics Devices Unlimited Sensors
FluiDynamics manufactures two flowmeters considered for testing -
a cross-flowing sensor and a co-flowing sensor. The fluidic cross-flow
velocity sensor employs a free jet normal to the measured flow, two
receivers or total head tubes, and a differential pressure gauge, trans-
ducer or manometer to indicate the pressure from the receivers, as shown
in Figure 10. The supply gas can be nitrogen, air, natural gas or other
gases at a standard supply pressure of 50 psi. Sensor operation depends
on the jet entraining and mixing with the surrounding fluid. This causes
the jet to spread out with distance from the nozzle, and the jet velocity
to decrease. Consequently the total head profiles becomes flatter with
distance from the nozzle. Entrainment of the cross flow causes jet
deflection.
Characteristics of the total head profiles are used to measure the
jet deflection by means of two total head tubes that are appropriately
-31-
-------
..frfft*
NOZZLE
JfcT
VlfASOftKD
MfAt> TUBES
^r-iuK±.:>jf*^*a
Figure 10. Elementary fluidic velocity sensor
-32-
-------
located with respect to the nozzle. The undeflected total head profile
imposes equal pressure on the two total head tubes that is proportional
to the cross-flow velocity, within the operating range. The relationship
of differential pressure with velocity is linear for the cross-flow
fluidic sensor up to 18 m/sec.
The cross-flow fluidic sensor has been used in flare gas stack appli-
cations by several oil companies in the United States. Since flare stack
gas streams contain adhesive particles which cause fouling of the receiver
ports, these users have found it necessary to back purge the receiver
ports. The users indicate that this purging nearly eliminates fouling
except in extremely dirty flare stacks.
The FluiDynamic co-flowing fluidic velocity sensor is a device some-
what similar in principle to the cross-flow fluidic velocity sensor but
employing a jet of supply fluid parallel to and in the same direction as
the ambient fluid stream whose velocity is being measured. The co-
flowing sensor is complementary to the fluidic cross-flow velocity sensor
in that it can be used to measure much higher velocities, i.e., up to 150
or 180 m/sec in air but at the same time overlaps the velocity of the
cross-flow fluidic velocity sensor although not being capable of measuring
the extremely low velocities that can be measured with the latter.
In order to measure the full velocity range, a cross-flow fluidic
sensor (.3 to 18 m/sec) and a co-flowing sensor (15 to 150 m/sec) must
be used simultaneously. Both instruments have a linear AP readout over
the indicated ranges and should be able to use the same supply gas source.
With the proper differential pressure readout device, the required flow
fluctuations should be detectable. The fluidic sensors measure gas
velocity at one station in the flow stream.
The variations expected in the gas composition and temperature will
present no problems with the FluiDynamic fluidic sensor with respect to
chemical resistance since the fluidic sensor is fabricated from 316
stainless steel. The sensors are reported by users to have very good
resistance to particulate fouling. Positive and negative factors identi-
fied for these sensors are as follows:
-33-
-------
Positive Factors
• Gas temperature range of 0 to 200°C is acceptable
t Sensor and probe are fabricated of stainless steel, therefore,
resistant to chemical corrosion
• Has been used successfully in heavily particulate-laden
flare stack applications
• Has been used successfully in near-saturated and water mist
laden gas streams
• Has been used successfully with trace concentrations of
H2S04
• Excellent maintenance and recalibration record reported by
users
• Rugged construction suitable for field use
t Accuracy and repeatability of 3% of reading is expected
• Device has been used with continuous readouts on strip
charts in control house
• Excellent sensitivity at low flow rates
• Small pressure drop in gas flow caused by instrument
• No flow reference required
Negative Factors
• Two sensors required to cover required velocity range, a
standard Fluidic sensor for .3-18 m/sec and a co-flowing
sensor for 15-150 m/sec
• Receiver ports may require a gas purge to keep particulate
matter from clogging up the ports
5.3.6 Ellison Annubar (Ellison Instrument Division, Dieterich Standard
Corporation) (Figure 11)
The Annubar is distinct from the other sensors considered in that it
is not a point sensor — it obtains a type of average velocity along a
line which spans the pipe or duct in which it is placed. The instrument
-34-
-------
I
CO
U1
I
ASME PAR UW 16 DRILL OR BURN
ONE 1-3/8" DIAMETER HOLE AND
PROVIDE 1/4" MINIMUM WELD BE AD
ALIGN TO PIPE AXIS (TYPICAL! ~
ASTM SPEC. A18VGRADE No. 2
2/3000 LBS. 1"NPT FORGED
STEEL WELD COUPLING IS
SUPPLIED.
4- 3/8" DIAMETER SENSING PORTS
ARE LOCATED IN APPROPRIATE
CENTERS OF CONCENTRIC ANNUL).
FLANGE INCLUDED
TO CUSTOMER'S
SPECIFICATIONS
ARMCO 17-4 PH
HARDENED STAINLESS
COMPRESSION FERRULE
FACTORY TIG WELDED
UNLESS OTHERWISE
SPECIFIED
PERMANENT
METAL TAG
WITH 3" CHAIN
STAINLESS 1/2". 1/4" or
1/8" NPT.or 1/4" TUBE
COMPRESSION. OR
BRASS SHUT-OFF VALVE
WITH 1/4"SAE FLARE
CONNECTION
1/8" NPTF CLEAN-OUT PLUGS
DIETERICH STANDARD CORPORATION
DRAWER M BOULDER COLORADO OO3OB UOA
751 TO 755 ANNUBAR FLOW ELEMENTS
; 9-23-70: ADDED CLEAN-OUTS; WAS 27/64
o o 1 & o o }••-
.^
FLANGE MOUNT OPTION
MATERIAL: 303 STAINLESS. OTHER MATERIALS OPTIONAL.
/»3TO4 PIPE DIAMETERS IS RECOMMENDED FOR DOWNSTREAM SIDE.
V<6OR MORE PIPE DIAMETERS IS RECOMMENDED FOR UPSTREAM SIDE
W AFTER VALVES. ELBOWS AND ETC. SEE CALCULATION REPORT E - 79
©PERMANENT TAG SHOWING MIN.. NORM. & MAX. DESIGNED FLOWS. METER
READINGS FOR DESIGNED FLOWS. TAG NO.. LINE SIZE. SER. NO. S MET6RED FLUID.
»EC.FPV PIPE" D'
«-HFrnil F na 11 „
SCHEDULE OR I.D.
• 'ELEMENT MADE TO TH.S D.MENSION
UNLESS OTHERWISE SPECIFIED. PARTS
|N TH,S REQ|ON g, GASKET NOT SUPPLIED
EXCEPT WELD COUPLING
1968 - OIETERICH STANDARD CORP.. BOULDER. COLO.
Figure 11. Annubar
-------
output is a differential pressure. The high pressure component is
obtained from having four upstream facing holes open into the probe body.
The resultant pressure is then sensed by a single tube in the center of
the probe. The low pressure component is obtained from a single orifice
facing downstream and also located at the center of the probe. The four
forward facing holes are located according to a four point velocity
averaging technique for flow in circular ducts. The probe output must
then be related to the total flowrate by means of an empirical calibration
factor. A probe must be manufactured specifically for a given installation,
since it completely crosses the pipe or duct.
Positive Factors
• The instrument is inherently an averaging one, which is
desirable for this program
• Good results have been reported for stack applications
t Corrosion and abrasion effects are considered to be negligible
Negative Factors
• The probe appears to be suited only for measurements in cir-
cular ducts
• Factory calibration factors may be in error if it is necessary
to use the instrument near a large flow disturbance
5.4 INSTRUMENT ACQUISITION
A Hastings Flare Gas Flow Probe, a Ramapo Fluid Drag Meter, and six
Thermo Systems hot film sensors were purchased for evaluation. An order
was placed with Fluidynamics but eventually had to be cancelled due to
delivery date slippage. Ellison Instruments provided free of charge
an S type pi tot probe and an eventual total of seven Annubars during
the program. Of these, a 150" Annubar and two 65" Annubars were left
permanently installed at the Nevada Power Company Moapa station at the
conclusion of the second field demonstration.
5.5 ADVANCED INSTRUMENTATION - ACOUSTIC VELOCIMETER
The intention at the start of the program for this subtask was to
build (or have built) and test an acoustic velocimeter for the application
-36-
-------
of interest. This type of device takes a line average across the duct,
which laboratory testing proved to be a good way to determine total flow.
Other desirable aspects of such systems, which are currently used to
monitor liquid and high pressure gas flows, are relative ease of instal-
lation and removal, and the fact that no system elements are immersed in
the flow.
Very early in the effort it became apparent that no hardware was
available for applications in a gas at one atmosphere in large ducts.
The main effort was then devoted to obtaining baseline information which
would have to be considered in any system design.
5.5.1 Technical Discussion
Figure 12 shows the basic arrangement of the acoustic flowmeter of
interest. Sound beams are transmitted both up- and downstream. The speed
of propagation is the vector sum of the local sound velocity and the flow
velocity. The parameter measured is the time of flight of a sound pulse,
or the change in phase of a CW beam. In steady flow, an upstream and a
downstream measurement traversing identical paths, combine to yield the
flow speed without an explicit determination of the sound speed.
It is easy to show that the greatest accuracy results when the
sound path and the duct axis are 45° apart.
The nature of the gas flow and the flue place a number of inherent
constraints on the choice of transducer. These constraints are discussed
below.
5.5.1.1 Losses in Acoustic Intensity in the Flue Duct -
Acoustic Impedance Mismatch
A basic constraint is the poor efficiency for acoustic energy
transfer between a solid (transducer face) and a low pressure gas. The
energy transfer for a 1 -dimension normal incidence model would be:
where n is the ratio of specific acoustic impedance of the gas and solid
or (pc) /(pc) ; p is density and c is speed of sound.
-37-
-------
\v
T
I
CO
00
I
r
\
\
\
Figure 12. Acoustic measurement of flow velocity in duct (schematic)
-------
The transmission coefficient is independent of the direction of
sound propagation, i.e., from gas to solid or from solid to gas. Values
of pc's are roughly:
Solids (metals and dielectrics) * 10* to 107 MKS Rayls
Liquids (non-metals) * 106 to 3.106
Gases (0°C. 100 atmos) * 101* to 5.101*
Gases (0°C, 1 atmos) * 102 to 5.102
Therefore energy transmission coefficients and the resulting sound inten-
sity level dB losses are approximately:
Solids •*-» liquids: 10"1 to 1
(-10 to 0 dB)
Solids *-»- gases at 0°C and 100 atmos: 10"3 to 2.10"2
(-30 to -17 dB)
Solids *-+ gases at 0°C and 1 atmos; 10"5 to 2.10"4
(-50 to -37 dB)
Quartz +-+ air at 20°C, 1 atmos: 10"1*
(-40 dB)
The above sound power losses are incurred both at the transmitter and at
the receiver. Therefore, for the subject gas flue, the total loss to be
expected from acoustic impedance mismatch is of the order of 80 dB. For
100 atmos gases, on the other hand, this loss would be between 35 and
60 dB. We have demonstrated a major problem for acoustic flowmeters
when employed in low pressure gas flows. Innovations in transducer design
may reduce these losses but not sufficiently to alter the main conclusion.
Spherical Divergence Loss (Figure 12)
The spherical divergence of the sound beam results in a 6 dB loss
for each doubling of distance from the source.
The source radius can be taken as follows. If the transducer disc
has a radius "a", then an equivalent source sphere is defined to be
ira2 = 4ir r 2, or r = i a. The subject application permits large
transducers; assume a to be 10 cm, and therefore r = 5 cm. With a
path length of L = D/cos 45° = 1420 cm, the number of radial doublings,
-39-
-------
n % 8 is obtained from 2nrQ = L. The resulting intensity loss is
% 8 x 6 = 48 dB. By contrast, a one meter duct would have approximately
5 doublings or a loss of 5 x 6 = 30 dB.
Molecular Absorption of Sound Energy
As the sound beam propagates ^ 14 M across the exhaust flue, it
loses energy to the flue gas by molecular absorption. The energy decay
is exponential I(x) = I exp {-2Ax} where A is the absorption coefficient.
Rewriting we get
ALT = Ll(x) - Lj =10 log]0 I(x)/I0 = -8.7 Ax (dB)
For air at 27°C and 37% relative humidity, Kinsler & Frey (Reference 5)
give the following values for (8.7A) dB/M:
f(KHz) 8.74 (dB/M) 8.7A (14 M) dB
20 0.65 9.1
40 1.1 15.5
100 4.2 59.0
These values are indicative of losses which should occur in the flue gas.
Evidently, it is advisable to use frequencies below about 40 KHz.
5.5.1.2 Summary of Inherent Acoustic Intensity Losses
in the Subject Flues -
From the above three sources of intensity loss, we get the
total:
Acoustic Impedance Mismatch ^ 80 dB
Spherical Divergence (20 cm Transducer) ^ 48 dB
Molecular Absorption Loss (f <40 KHz) * 15 dB
Total ~ 143 dB
These losses do not include additional losses that could be produced
from (1) reflections in a gas flow with temperature, density and
therefore, specific acoustic impedance gradients as encountered in
the subject flues, (2) scattering by turbulent eddies.
-40-
-------
Molecular absorption of sound energy shows a strong dependence
on frequency, and limits the proposed acoustic flowmeter to 40 KHz
or less.
5.5.1.3 Background Noise and Vibration Levels -
Background noise and vibration measurements were made by TRW on
the exhaust flues of the SCE Mojave Power Plant, in Clark County,
Nevada, March 2 and 3, 1973. The objective was to determine acoustic
instrument operating requirements for the subject application. The
power plant has two 750 MW units which were operating near capacity at
the time of the measurements.
Results of the tests are shown in Figures 13 through 16, and
Figures 17 through 19. From the noise tests, the noise sound pressure
level (or intensity level) is quite high (100 to 70 dB) up to about 20 KHz;
at this frequency, the intensity drops very sharply. It appears that
the level should be below 50 dB for frequencies above 30 KHz.
Wall vibration measurements show 3u displacements not exceeding
0.2 cm for frequencies up to ~ 500 Hz. Above 500 Hz, displacements are
much smaller.
These tests demonstrate that the proposed acoustic flowmeter will
have to operate above about 30 KHz to avoid the high background noise
levels at lower frequencies.
5.5.2 Conclusions
A supplier survey has shown that an off-the-shelf (with minor
modifications) acoustic flowmeter for the subject applications is not
available.
A feasible study has shown that two loss factors place large power
demands on the transducers for the subject application. One is the
extreme acoustic impedance mismatch between (solid) transducer and a
one-atmospheric pressure gas; the other is the spherical divergence loss,
caused by the very long sound path length.
The frequency range most suitable for the subject application
should be between 30-40 KHz to avoid the high background noise levels
-41-
-------
Figure 13. Stack flue interior noise, Mohave power plant unit 1, March 2, 1973, operating at
760 ±5 megawatts. B&K 4136 microphone. One-third octave spectrum.
-------
CO
v\ \ v '•
\\\ /> i T~A ,1
\HAft Ui—<
JO
to
5oo
Figure 14. Stack flue interior noise, Mohave power plant unit 1, March 2, 1973, operating at
760 ±5 megawatts. B&K 4136 microphone. Narrow band spectrum.
-------
I
-fc>
-fa
±00
Figure 15. Stack flue interior noise, Mohave power plant unit 1, March 2, 1973, operating at
760 ±5 megawatts. Statham PL 80TC-0.3-350 pressure transducer. Narrow band spectrum.
-------
en
i
500
Figure 16. Stack flue interior wall vibration, Mohave power plant, March 2, 1973, operating at
760 ±5 megawatts. B&K 4136 microphone. One-third octave spectrum
-------
en
i
(00
(ooo
(O Ooo 40 ooo
i
Figure 17. Precipitator inlet flue interior noise, Mohave power plant unit 1, March 3, 1973,
operating at 775 ±5 megawatts. B&K 4136 microphone. One-third octave spectrum
-------
Figure 18. Precipitator inlet flue interior noise, Mohave power plant unit 1, March 3, 1973
operating at 775 ±5 megawatts. B&K 4136 microphone. Narrow band spectrum
-------
00
I
J_
Figure 19. Precipitator inlet flue interior wall vibration, Mohave power plant unit 1,
March 3, 1973, operating at 775 ±5 megawatts. Narrow band power spectral density
-------
in the low ultrasonic and audible ranges, and to avoid high molecular
absorption losses by humid flue gases above * 40 KHz.
For an assumed transducer of 20 cm diameter, the beam major lobe
widths are 3.8° for 30 KHz and 3.0° for 40 KHz; these beam widths appear
suitable for the subject application.
Pulsed sound waves are likely to produce inaccurate measurements
because very short duration pulses (less than a millisec) are needed for
accurate times of arrival; these short duration pulses will have a power
spectrum bandwidth greater than 1 KHz and therefore, a relatively low
signal-to-noise ratio.
Continuous single frequency or narrowband, measuring phase dif-
ferences in the 30-40 KHz range, can operate with all power in a low
noise level region, and may therefore be expected to obtain more accurate
velocity data than with pulsed waves.
The present feasibility study would indicate the acoustic method
to be marginal at best for the subject application. A more refined
assessment of transducer performance is needed, however, to show what
improvements in efficiency are available. In any case, however, this
study has defined the difficulties which have thus far prevented the
development of an acoustic flowmeter by commercial suppliers, for appli-
cations similar to that discussed here.
-49-
-------
SECTION VI
TASK III - MAPPING TECHNIQUE EVALUATION
6.1 GENERAL
With the exception of the Annubar, all instruments chosen for evalua-
tion were point sensors, and the Annubar had factory calibration factors
only for use in circular ducts. This made it clear that sensor evaluation
alone would not accomplish program objectives -- it would also be necessary
to find ways in which to deploy the sensors. It was therefore decided to
perform laboratory flow mapping tests in order determine acceptable sensor
placement techniques. During the course of the program, mapping tests
were performed for many configurations in two different facilities. It
was arbitrarily decided early in the effort that only techniques which
would require not more than eight velocity sensors would be considered,
since use of a large number of sensors would be expensive and complicated.
Two kinds of probes were utilized for mapping work: the hemispherical
nosed pitot-static probe and the Annubar. Pitot-static probes were used
for all point measurements, and the results are considered to apply to
point sensors in general so long as the sensor is small with respect to
the size of the duct. The Annubar had to be evaluated directly since it
is not a point sensor. Most of the effort was devoted to flows in
rectangular ducts, which traditionally have not been as well characterized
as the flows in circular ducts.
6.2 DESCRIPTION OF FACILITIES
Initial mapping tests were performed during 1973. Test configurations
are shown in Figures 20 - 22. For the circular test section configurations,
pitot-static probe data were taken along four diameters, as shown in
Figure 21. For both configurations, the fan inlet was open to the
atmosphere, resulting in negligible temperature and density stratification
in the flow. The purpose of the testing was to investigate velocity
stratification. A sixty-four point traverse was taken in the rectangular
test section, as shown in Figure 23. Resulting data were then analyzed
in accordance with the techniques of interest. For both circular and
rectangular duct data, the reference flow rate was computed from traversal
data in the test plane. An Annubar was installed during most of the rectang-
ular mapping tests, and was located downstream of the test section, as
-50-
-------
INTAKE
BLOWER
HONEYCOMB FLOW
STRAIGHTENERS
4.88
SAMPLING
STATION
1.22
.36 DIAMETER DUCT
CASE A PARTIALLY DEVELOPED FLOW
.30 x .41 DUCT
BLOWER
MOVABLE
VANE
t
.76
3.05
.30 x .41 DUCT
.20
1.52
.SAMPLING
STATION
— .15
•36 DIAMETER -30 * -41
DUCT DUCT
CASE B FLOW AROUND A CORNER
All Dimensions in Meters
Figure 20. Circular duct mapping test configurations, 1973
-51-
-------
TRAVERSING
MECHANISM
DUCT
PITOT-STATIC PROBE
TRAVERSE PANEL
INSERT
DUCT
REMOVABLE PANEL
(3)
Figure 21. Pi tot traverse installation - circular duct
-52-
-------
7
BLOWER
HONEYCOMB FLOW
STRAIGHTENERS
SAMPLING
STATION
.30 x .41 DUCT
CASE A PARTIALLY DEVELOPED FLOW
SAMPLING STATION
MOVABLE
VANE
.20
.30 x .41 DUCT
CASE B TWO STREAM DUCT
.36
t
.76
J_
SAMPLING
STATION
.30 x .41 DUCT
CASE C FLOW AROUND A CORNER
All Dimensions in Meters
Figure 22. Rectangular duct mapping test configurations, 1973
-53-
-------
1 | 9
J -J
1
1
4
1
i j
\- ' ' 1
i
i
i
!
I
i ....
...... 1 . . j
8 1
1
1 1 1 ' ' !
I'll
L. •• ! l
i f - ~ 1 • 7~ "- -
1 1 ' 1
1 : 1 !
' ! ' '
i ' | ! 1
.. - h !
1 ' l
1 1 i
|_ 4 L L i. - -
r " ; ~ | ^ !
1 ' ' 1
---•J- — -+---J- — -[----
1 ' ' 1
! i 1
! ' j. 1 .
' ' I
1 , i '
1 , i 1
WIDTH
8
.' 57
i—
i
. —
— —
r
64
i
\
HEIGH!
8
OBTAIN MEASUREMENT AT CENTER OF EACH SUBSECTION
Figure 23. Schematic of rectangular duct
reference traverse map
-54-
-------
shown in Figure 24. No Annubar was available during the circular
duct mapping tests.
A new mapping facility was constructed specifically for this
program during 1974. The configurations used are shown in Figure 25,
and probe locations for 1974 testing are shown in Figures 26 - 29. Probe
locations for testing during 1975 are shown in Figure 30. The new
facility was built for several reasons. Having the fan at the duct
outlet rather than at the inlet allowed greater freedom of variation
for inlet conditions. The straight, circular reference section down-
stream of a honeycomb flow straightener was used to provide a total
flow measurement in an area free from significant flow angularities.
It also eliminated the need for a full traverse of the test section
for each run, so that the number of runs could be maximized.
6.3 MAPPING TECHNIQUES
6.3.1 Circular Duct Techniques
Many empirical techniques are available for determining volumetric
flow in circular ducts through a number of point measurements. Perhaps
the most traditional method is the centroid of equal areas technique,
as given in EPA Method 1 (Reference 1). This is the best technique to
use if nothing is known about the flow profile shape. It has long been
recognized that flow in a straight circular pipe eventually assumes a
parabolic type profile as a steady state condition. This is called
fully developed flow. This characteristic of pipe flow was used by
Winternitz and Fischl in developing the Log-Linear Mapping Technique
(Reference 6). It is based on the assumption that pipe flow profiles
can be generally characterized by the equation
v(r) = A + B log (12)
where
v = magnitude of velocity, m/sec
-55-
-------
ANNUBAR
i
in
FLOW
DIRECTION
PLANE OF
PITOT TRAVERSE
Dimensions in Meters
-O—
Figure 24. Annubar location during duct mapping test, 1973
-------
I
tn
CONFIGURATION
IN ORDER OF
TEST
MEASUREMENT
PLANE
.69
H
MEASUREMENT
PLANE
FAN
H
.41
.91
FLOW
DIRECTION
c>
Dimensions in Meters
INLETS TEST SECTIONS
H DENOTES LOCATION OF HONEYCOMB PANEL
REFERENCE SECTIONS
Figure 25. Schematic -- top view of mapping test facility, 1974
-------
VERTICAL
01
ca
i
ANNUBAR
— * 6 1n to
d "7
00 . -M.
PROBE
1
j
PROBE
2
.- 20
^
HONEYCOMB
ANNUBAR
HONEYCOMB
Dimensions in Meters
Figure 26. Position and orientation of reference probes
-------
en
-------
Dimensions in Meters
'
.91
1
1
.4
i
6
;
^
^
l
-o-
+
PROBE 5
0 0 0
ANNUBAR
l.i
.04
-is
()
ANNUBAR
Figure 28. Rectangular section probe locations
-------
AIIHUBAR
LOCATION
T1
• 4)
ANNUBAR
LOCATION
PITOT PROBE
LOCATION
.90
1.09
.61
-.41
A. PROBE LOCATIONS AFTER RECTANGULAR ELBOW
PITOT PROBE
LOCATION
i i
.7] DIA
.97
1.22
.76
DUCT HEIGHT
.91
Dimensions in Meters
B. PROBE LOCATIONS AFTER CIRCULAR ELBOW
Figure 29. Probe locations relative to elbows in 1974 mapping tests
-61-
-------
PITOT TRAVERSE /
LOCATIONS
t
W
I
Location
A
B
C
D
E
F
G
H
I
\ B C
I
E
G
Distance From Elbow
Pi tot
Traverse
2.20W
1.36W
.92W
.80M
.61W
.42W
.23W
--
--
Annubar
2.91W
2.06W
1.62W
1.50W
1.31W
1.12W
.94W
.75W
.56W
^
Flow
Direction
W = .406 M
D F H
Annubar
Locations
t
1
A
B C
E
G
:
^
•
Flow
Direction
Figure 30. Probe locations -- 1975 testing
-62-
-------
A, B = constants, m/sec
y .= distance from pipe wall toward center, m
d = pipe diameter, m
As the authors state, "The basis for the proposed integration rules was
the selection of metering position in each annulus such that the exact
mean velocity for a particular cross-section would occur at the selected
gauging points, provided the chosen logarithmic functions are an adequate
representation of the considered pipe-flow velocity profiles."
One of the conclusions in Reference 6 is that for a given number of
measurement points, the Log-Linear technique gives more accurate results
than the centroid of equal areas technique (also known as the tangential
method). This conclusion was accepted as valid and it was decided to use
the Log-Linear technique for reference measurements during the test,
as well as evaluating it for general use with a minimum number of points.
All methods tested are shown in Table 2. Each method involved measurement
of velocity at specific points along two orthogonal diameters.
The Annubar, as described in Section 5, was designed around a four
point per diameter mapping technique. Its used in circular ducts is
straightforward: it is installed according to factory specifications
and a factory calibration factor is used to correct the output. The
computed flow for each run was then compared to the reference flow measure-
ment to determine instrument accuracy.
6.3.2 Rectangular Duct Techniques
A literature search revealed no rectangular duct mapping techniques
involving eight or less sampling points. Consequently, new methods had to
be developed. In order to do so, properties of flows in rectangular
ducts were considered. Some of the words done in this area by Nfkuradse
is documented in Chapter XX of Reference 4, where it is observed that
secondary flows occur "in all straight pipes of non-circular cross-section."
In a rectangular duct, the secondary flows tend to generate high velocities
at the corners. This tendency means that flow angularities with respect
to the duct axis should always be expected regardless of the duct length.
-63-
-------
Table 2. METHOD FOR DETERMINING VOLUMETRIC FLOW FROM POINT MEASUREMENTS IN A CIRCULAR DUCT
RULE
Gauss-
Sherwood
Gauss-
Sherwood
Aichelen
New Empirical
Log-Linear
Log-Linear
Log-Linear
Log-Linear
i
NUMBER OF POINTS
PER DIAMETER
3
Weights*
4
Weights*
2
3
Weights*
2
2 + 4
4
10
DISTANCE FROM WALL
IN PIPE DIAMETERS
.060 .500 .940
585
.036 .208 .792 .964
.1739 .3261 .3361 .1739
.119 .881
.081 .500 .919
2 1 2
.112 .888
.112 .888 (One Dia.) Q .
.043 .290 .710 .957 ( „<)
U i a
.043 .290 .710 .957
.019 .076 .153 .217 .361
.639 .783 .847 .924 .981
TOTAL NUMBER
OF POINTS
(ON TWO ORTHOGONAL DIAMETERS)
5
8
4
5
4
6
8
20
I
cr>
^Multiply measured value by weight, sum, divide by sum of weights.
-------
Another practical problem associated with flow measurements in
rectangular ducts in process streams is that the ducts are normally very
short. In a power plant, for example, it would not be expected to find
any runs of rectangular ducting comparable in length to that of the
circular exhaust stack. Circular duct mapping techniques are derived
from knowledge of fully developed flow characteristics, and then applied
to non-fully developed flows. There has been no directly workable
analogue of this for the rectangular case.
Common circular duct mapping techniques may be termed "universal"
in that they are designed to apply in any circular duct without a need
for in-place calibration. Clearly it would be desirable to have such
techniques available for continuous monitoring in rectangular ducts.
Continuing this process, it is possible to define various types of
techniques in order of acceptability as follows:
1. Universal - sensors can be installed at pre-selected locations
to give accurate total flow measurement without in-place calibra-
tion or prior knowledge of flow conditions.
2. Single calibration - installed system needs to be calibrated
in place at one flow condition only to determine a constant
calibration factor.
3. Full calibration - installed system needs to be calibrated
in place over the full range of flow conditions to produce
a curve of calibration factor vs. flow rate.
Techniques 2 and 3 may also require some prior knowledge of the flow to
determine sensor placement. The goal of producing a Universal technique
was not fully realized, although results indicate that methods developed
are definitely in the Single Calibration category and may be universal
immediately downstream of an elbow. A broader data base is needed to
verify the latter.
The first techniques attempted were the four and five point methods
shown in Figure 31. The objective of the investigation was to determine
if a single location on the diagonals for each point would result in
a correct computation of the total flow rate. If these points existed,
-65-
-------
en
-------
then the methods would be directly analogous to circular duct mapping
techniques. As is shown below, these proposed methods failed.
Existing 1973 test data were then re-analyzed to determine other
possible techniques, and the Row Average Method was discovered. This
method involves determining the average velocity along a line spanning
the duct and assuming that it is the average velocity for the entire
duct. Methodology involves determining which two walls to take the
line average between, and the proper location of the line. This is
illustrated for a fictitious case in Figure 32. By definition, rows
are taken along lines of maximum velocity stratification. Proper location
with respect to the side walls was determined during testing by comparing
local row averages to the total flow average. This method proved to be
successful, especially in the wake of an elbow, and came to be the
recommended method for use with point sensors.
There was initially no expectation that the Annubar would work well
in rectangular ducts, since its hole pattern was based on a circular duct
technique. An Annubar was installed along the center row for most of
the 1973 rectangular duct testing, and the results showed a surprisingly
repeatable calibration factor. As a result, an Annubar was used during all
subsequent rectangular duct mapping tests. It eventually became clear
that the Annubar and Row Average methods are really very similar, since
they both involve line averaging. This led to the further realization
that both techniques work best directly downstream of an elbow.
6.4 MAPPING TEST RESULTS
6.4.1 Circular Duct Techniques
6.4.1.1 1973 Testing -
A total of eight runs was performed for the two configurations shown
in Figure 20. Typical velocity profiles from each configuration are
shown in Figure 33 from which it is clear that the flows were undeveloped,
i.e. non-axisymmetric and not parabolic. For each run, data were taken
along two sets of orthogonal diameters (Figure 21) thus giving two sets
of data for each method per run, or a total of sixteen sets of data for
each method.
-67-
-------
Tabular Entries Are Velocity At Traverse Points
*-Direction Of High Gradients
Direction
Of
Low
Gradients
2
1
0
0
0
1
4
3
4
0
3
3
6
5
4
3
5
7
8
7
6
6
7
9
10
9
9
9
10
10
12
11
10
9
11
12
Row Average
7.00
6.00
5.50
4.50
6.00
7.00
Overall Average =6.00
Rows are taken in the direction of highest velocity gradients.
Row averages for rows 2 and 5 agree with overall average, so
either of these rows would be recommended for sensor placement
Figure 32. Row average method
-68-
-------
DISTANCE FROM WALL
DUCT DIAMETER
10
i
1.00
.50
AIR SPEED, METERS/SEC
25 30 '
\ \
PARTIALLY
DEVELOPED FLOW
Figure 33. Typical flow profiles from circular duct mapping test
-------
Results are shown in Table 3. For each run the average velocity
was determined using the Log-Linear 10 method along all four diameters
for a total of forty points. Each method was then evaluated with respect
to the reference velocity. Since two sets of orthogonal diameters were
used, each method produced two sets of data per run except for the
Log-Linear 2+4 method, which had four possible combinations. The average
error listed is the average of the absolute errors.
The results show that the Aichelen and Gauss-Sherwood 3 methods were
consistently less accurate than the others. The results for the other
five methods are excellent in terms of program requirements—worst case
errors were always under 5 percent, and average errors always under
2 percent.
As mentioned above, the reason for this part of the testing was to
confirm the adequacy of previously documented methods for circular duct
flow determination. The results in Table 3 show that five of these
methods indeed gave very good results in nondeveloped flows.
6.4.1.2 1974 Testing -
As shown in Figure 25, two basic configurations were used -- a
straight inlet and a tiitered elbow inlet. Variation of the inlet flow
was accomplished by blocking off various sections of the inlet with
4" x 8" aluminum plates. Three measurement techniques were evaluated --
the Log-Linear 4 method (8 points) and the New Empirical method (5 points),
for which data were obtained by two pitot static probes, and the Annubar.
Reference data were provided by two pitot static probes in the reference
section, utilizing the Log-Linear 4 method, and another Annubar. The
Log-Linear 4 data in the reference section was taken to be the reference
flow data as a result of the 1973 tests. Accuracy of this reference
measurement was estimated to be not worse than +2%.
Inlet blockage configurations for the test are shown in Figure 34.
These were chosen at random to create a variety of flow patterns at the
test section. For each blockage condition, data were taken at four different
flow rates. A total of 24 runs was performed for the elbow case, and 32
runs were performed for the straight inlet case. Results for the two
configurations are given in Tables 4 and 5 and a summary is given in Table 6.
-70-
-------
Table 3. CIRCULAR DUCT MAPPING TEST RESULTS, 1973
CONFIG. .
A
B
RUN
NO.
1
2
3
1
2
3
4
5
AVERAGE ERROR
WORST CASE ERROR
AVFRflRF
VELOCITY
M/SEC
21.75
17.53
8.11
23.03
17.68
18.87
19.46
14.44
ACCURACY OF METHOD, %
L.L. 2
(4)*
- .5
+ .4
0
-1.4
-2.3
- .8
+ .1
-4.3
-3.3
-2.4
-3.6
-4.0
-1.0
+ 1.9
-1.7
-1.7
1.8
4.3
I
L.L. 4
(8)
+ .4
+ .1
0
- .5
+ .8
- .8
+3.1
+ .1
+ .8
- .3
+ 1.8
+ .5
+ 1.0
+4.6
+ 1.7
+1.9
1.2
4.6
L.L. 2+4
(6)
+ .1, + .6
- .4, + .2
+ .2, - .2
- .7, -1.2
- .8, - .8
-1.1, 0
+1.1, -2.7
+2.1, -1.5
+ .4, -2.3
-2.9, - .4
+ .2, -2.4
-2.0, -1.2
+1.0, +2.4
-1.0, +4.1
+1.2, - .2
1.1, + .4
1.2
4.1
A.
(4)
+ .2
+1.2
- .5
- .7
- .8
+ .4
-3.1
+4.4
-3.2
+7.3
-3.5
+5.7
+2.3
+5.5
-2.7
+7.5
3.1
7.5
N.E.
(5)
- .5
- .1
- .2
+ .2
-1.1
- .8
+ .9
-2.2
+1.0
- .8
+1.5
-1.1
- .7
+2.4
+1.8
- .4
1.0
2.4
G.S. 3
(5)
+ 1.5
+1.8
+ 1.9
+1.9
+ 1.1
+ 1.5
+1.9
+3.7
- .1
+4.7
+2.8
+3.9
+6.3
+1.7
+1.6
+5.4
2.6
5.4
G.S. 4
(8)
+ .1
+ .6
+ .2
- .2
+ .8
0
-1.2
-2.7
-1.2
-1.7
- .6
-4.0
+ .7
+3.4
- .6
-2.5
1.3
4.0
Total number of data points required
-------
Shaded portion of inlet blocked off for runs shown.
Each set of four runs was performed at four different
flow rates.
Runs 89-92,
113-116, 117-120
93-96, 121-124
97-100, 125-128
101-104, 129-132
105-108, 133-136
109-112, 137-140
141-144
Figure 34. Circular inlet blockage - 1974
-72-
-------
Table 4. 1974 CIRCULAR DUCT
Conf. 4: Runs 89-112
POINT SENSOR MAPPING RESULTS
Config. 3: 113-114
Run
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
VREF
rn3/sec
6.2
4.8
4.3
3.1
4.9
4.3
4.0
3.1
5.3
4.4
4.1
3.2
4.6
4.3
3.8
3.1
5.0
4.4
4.0
3.2
5.2
4.4
4.2
3.2
6.8
5.0
4.6
3.3
L.L.
1.020
.995
1.043
1.032
1.052
1.035
1.005
1.084
.954
.994
1.051
1.127
1.034
1.045
1.103
1.065
1.004
.979
1.043
1.037
1.031
1.065
1.105
.960
1.033
1.026
1.011
1.043
N.E.
1.097
1.096
1.140
1.169
1.076
1.086
1.164
1.127
1.016
1.072
1.083
1.108
1.144
1.067
1.213
1.071
1.082
1.149
1.060
1.121
1.141
1.125
1.120
h072
.953
.950
.977
.961
Run
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
126
127
133
139
140
141
142
143
144
VREF
3,
ra /sec
6.7
5.1
4.6
3.4
5.7
4.6
4.3
3.3
5.6
4.6
4.2
3.2
5.2
4.4
4.1
3.0
5.5
4.7
4.6
3.2
5.6
4.7
4.2
3.2
6.1
4.8
4.3
3.2
L.L.
1.022
.978
.992
1.005
1.029
1.075
1.026
1.026
1.052
1.043
1.028
.972
1.032
1.051
1.046
1.058
1.012
1.000
1.031
.984
1.039
1.024
1.059
1.048
1.064
1.025
1.077
1.031
N.E.
.973
.950
.980
1.021
.981
1.099
1.095
1.022
1.001
1.066
1.018
.985
1.061
1.072
1.061
1.093
.810
.898
.868
.809
.982
.990
1.009
1.006
1.029
.950
1.024
1.023
I I - Log-Linear 4
VREF
N.E. =
_ New Empirical
VREF
-------
Table 5. ANNUBAR CALIBRATION FACTORS FOR 1974
CALIBRATION TESTS - CONFIGURATIONS 3 AND 4
Conftg. 4: Runs 89-112
Conftg. 3; Runs 113-144
Run
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
V
m /sec
6.2
4.8
4.3
3.1
4.9
4.3
4.0
3.1
5.3
4.4
4.1
3.2
4.6
4.3
3.8
3.1
5.0
4.4
4.0
3.2
5.2
4.4
4.2
3.2
6.8
5.0
4.6
3.3
%
REF
.738
.691
.692
.665
.705
.673
.684
.658
.742
.672
.693
.658
.700
.723
.694
.729
.720
.719
.681
.702
.732
.772
.712
.703
.684
.683
.688
.675
*r
TEST
.711
.669
.670
.624
.687
.636
.689
.640
.695
.668
.681
.644
.704
.734
.718
.705
.650
.729
.680
.677
.733
.757
.734
.692
.663
.663
.667
.666
Run
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
126
127
133
139
140
141
142
143
144
V
ro' /sec
6.7
5.1
4.6
3.4
5.7
4.6
4.3
3.3
5.6
4.6
4.2
3.2
5.2
4.9
4.1
3.0
5.5
4.7
4.6
3.2
5.6
4.7
4.2
3.2
6.1
4.8
4.3
3.2
%
REF
.667
.695
.689
.709
.696
.698
.694
.694
.687
.680
.673
.680
.714
.690
.698
.672
.746
.713
.689
.670
.727
.745
.726
.729
.677
.672
.672
.683
TEST
.662
.679
.685
.698
.689
.695
.699
.702
.671
.683
.669
.675
.716
.687
.689
.659
.743
.695
.679
.652
.733
.734
.735
.722
.662
.659
.645
.644
- -698
o$ =
S = . 687
+_ .025 (+3. 7%)
+ .030 (+4.4%)
Factory Value:
S = .667
-74-
-------
Table 6. SUMMARY OF 1974 CIRCULAR DUCT MAPPING TEST RESULTS
Method
Annubar(Test
Section)
Annubar
(Reference
Section)
Log-Linear 4
New Empirical
Average Error, Percent, and Standard
Deviation, Percent
Conf. 3
.2.6 +_ 3.9
+2.5 +_ 3.1
+2.9 +_ 2.5
-0.9 + 7.1
+1.2 + 4.5*
Conf. 4
-3.3 + 5.0
+5.3 +_ 4.0
+3.6 + 4.1
+10.8 + 3.8
Total
-3.0 +_4.4
+4.6 + 3.7
+3.2 + 3.3
+4.1 +_8.0
+5.6 +6.2*
Excludes runs 133-136
.75-
-------
Annubar results are presented in terms of the standard calibration
factor, S, as used by the factory:
V S ^s
s bA
where
V = true volumetric flow at standard conditions
S = Annubar calibration factor
S,, = uncorrected volumetric flow at standard conditions
calculated from Annubar output.
For the runs performed, the average S value from the factory was S =.667.
Several important features are shown in the summary. The first is
that the new Empirical method did not provide good results for the elbow
case and for one configuration in the straight inlet case. In both
instances, a systematic shift was observed. For the elbow case, the
average for the New Empirical method was 10.8% above the reference value,
and for runs 133-136 for the straight inlet case, the average was about
15% low. For the other 28 straight inlet runs, the average was 1.2% high,
which is very acceptable. It is significant to note that for the elbow
case, the shift was very systematic -- the standard deviation for those
runs was comparable to that for the other methods.
Annubar results and the Log-Linear 4 results are all comparable,
and the Annubar results are reasonably close to the factory predictions.
It is noteworthy that the Log-Linear results in the test section were
systematically high, and that the Annubar output was systematically low,
which resulted in a higher than predicted calibration coefficient. It is
believed that the shift for the Log-Linear results was due to flow
angularity in the test section. As has been mentioned, the desired
parameter is the axial velocity component. In non-aligned flows, as will
be shown more completely in Section 7, a hemispherical pitot-static
probe tends to indicate too high an axial component. On that basis, it
would be expected that a particular method, such as the Log-Linear, would
-------
indicate a higher velocity in a region of significant flow angularity, such
as the test section, than in a more one-dimensional area, such as the
reference section. This angular response problem was, as previously
mentioned, one of the reasons for constructing the reference section.
It is believed very likely that the Annubar results varied from the
factory calibration due to the presence of the rods used to support the
duct itself and the pitot probes. Two half inch diameter steel rods were
placed upstream of each Annubar to stiffen the duct and hold it in a cir-
cular shape. In addition, the pitot-static probes slid along these rods.
The presence of the rods inevitably created a low pressure (and hence
low velocity) region along the axis of the duct, which is where the
Annubar rear static port was located. This localized low pressure area
would result in lower than normal Annubar output, which in turn resulted
in a higher than predicted calibration factor. This explanation is given
added credence by the fact that the reference Annubar calibration factor
was higher than that for the one in the test section, since the flow was
better aligned axially in the reference section so that the reference
Annubar static port would lie more directly in the wake of the support rods
than would the static port of the test section Annubar.
In summary, it is felt that the Log Linear results would have been
better if an ellipsoidal nosed pitot probe had been used (further explana-
tion is given in Section 7) and the Annubar results would have been better
if the instruments had not been directly in the wake of the support rods.
Even with these discrepancies, however, data for these methods is quite
acceptable. Results for both Annubars and the Log Linear 4 method averaged
better than 5% accuracy and less than 5% standard deviation, which is well
within program objectives. The New Empirical method was accurate only
for some configurations, however, although errors were very systematic
and could be accounted for in actual conditions by an in-place calibration.
The tabulated data show one practical consideration very clearly,
and that is that the Annubar, from which one measurement is required,
gave results comparable in accuracy and consistency to the Log Linear
method, which required eight individual measurements, and gave better
-77-
-------
results than the New Empirical method, which required five individual
measurements. The cost difference between one channel of instrumentation
for the Annubar and eight for a fixed Log Linear array of point sensors is
obvious. From an analytical standpoint, test data reveal no significant
difference between the Log-Linear method and the Annubar. From a hardware
cost and system complexity standpoint, the Annubar is economically
more desirable.
6.4.2 Rectangular Duct Techniques
6.4.2.1 1973 Testing -
A total of forty runs was performed in the configurations shown in
Figure 22. The breakdown was as follows:
Case A - Partially developed flow: 5 runs
Case B - Two stream flow: 19 runs
Case C - Flow around a corner: 16 runs
An Annubar was installed for Cases B and C but was not available for
Case A. In Case A, only flow rate was varied. In Cases B and C, both
flow rate and the position of the moveable vane shown in Figure 22
were varied.
The first techniques examined were the four and five point techniques
discussed in Section 6.3. The analysis proceeded as follows:
1. Compute average velocity by averaging 64 data points.
2. Mathematically curve fit the data points along the diagonals.
3. Specify a desired accuracy (e.g., 1%, 5%).
4. Begin at the center of the duct and proceed in 1 percent length
increments along each half diagonal, computing a four or five
point average (using the center velocity as constant in the
five point method) at each increment until the computed velocity
agrees with the reference velocity within the specified accuracy.
5. Repeat the procedure starting at the corners and moving toward
the middle.
-78-
-------
This analysis was used to determine the tolerance band for the accuracy
specified, so long as the velocity is only increasing or only decreasing
within the band, which would generally be the case. Results for a typical
run are shown in Figure 35 . The results at that point seemed
highly encouraging. They showed a wide bandwidth -- 5 to 10 percent or
more, for the location of the sampling points to obtain a mapping accuracy
of 2 to 5 percent, which is within acceptable limits. The wide bandwidth
was encouraging since it tended to show that a wide variety of flow con-
ditions could have overlapping bandwidths. The overlaps can be tabulated
to determine an optimum sensor location to give acceptable accuracy over
the full range of flow conditions. This in turn means that the four
and/or five point methods would in fact be universal within certain
accuracy limitations.
More complete data analysis quickly dampened the initial enthusiasm
for the four and five point methods. Overall results for the three test
configurations are shown in Table 7 . The five point method worked quite
well for the partially developed flows and for the two stream flows with
vane settings 1 and 2. For these 12 runs, sensors in the region .75 to
.79 would give accuracies of at least +5%. Results for the four point
method were not so good -- there was no region which was acceptable
for all 12 runs. For case C, flow around a corner, the region .42 to
.50 was acceptable for 5% accuracy using the five point method for 13
of the 16 runs, while the four point method was acceptable in the region
.35 to .40 for 13 of the 16 runs. The disappointment here is that while
results were very consistent for this case, the regions were not close
to those for cases A and B. This showed rather conclusively that neither
method will be as universal in scope as the corresponding circular duct
methods such as the Log-Linear technique.
Failure of the four and five point methods led to a search for other
usable techniques, and the Row Average Method was eventually discovered.
The method was described in Section 6.3.2. For each of the forty completed
runs, row averages were determined from the 8x8 reference matrix, and
compared to the total flow average. The row orientation can usually be
-79-
-------
DATE: 3-13-73
RUN NO: 2
DUCT CONFIGURATION: Partially Developed, 1/4 Flow
AVERAGE VELOCITY: 8.56 m/sec
SPECIFIED
ACCURACY
.5
2
5
FRACTIONAL TOLERANCE BAND*
.68 -
.66 -
.62 -
t
.68
.70
.73
r.
.74 -
.71 -
.67 -
>
.74
.76
.80
*
r given in fraction of half diagonal
Figure 35. Results of four and five point analysis for a partially
developed flow run
-80-
-------
Table 7. RESULTS OF RECTANGULAR DUCT ANALYSIS
BY FOUR AND FIVE POINT METHODS
ENTRIES INDICATE PROBE PLACEMENT REGIONS ALONG HALF DIAGONAL LENGTH IN PERCENT TO ACHIEVE
VELOCITY MEASUREMENT ACCURACY AT TOP OF COLUMN. SEE FIGURE 2
CASE A. PARTIALLY DEVELOPED FLOW
INLET '
1
3/4
1/2
1/4
1/8
METHOD:
VELOCITY
ACCURACY:
NET ACCEPTABLE
REGION
5 POINT
0.5%
70 - 71
70 - 71
77 - 77
X
73 - 73
X
2%
67 - 74
66 - 74
75 - 78
71 - 76
70 - 75
X
5%
59 - 80
59 - 80
71 - 81
67 - 80
66 - 79
71 - 79
4 POINT
0.5%
63 - 64
64 - 65
71 - 71
68 - 68
68 - 68
X
2%
60 - 67
61 - 67
69 - 73
66 - 70
66 - 70
X
5%
54 - 73
54 - 73
66 - 75
62 - 73
62 - 73
66 - 73
I
oo
-------
Table 7 (continued).
RESULTS OF RECTANGULAR DUCT ANALYSIS
BY FOUR AND FIVE POINT METHODS
CASE B. TWO STREAM FLOW
INLET
1
1
1
1/16
1
3/4
1/2
VANE
1
1
1
1
2
2
2
METHOD :
VELOCITY
ACCURACY :
NET ACCEPTABLE
REGION
1
3/4
5/8
1/2
1/8
1
3/4
1/2
1
3/4
1/2
V2
2.5
2.5
2.5
2.5
2.5
3
3
3
4
4
4
5
NET ACCEPTABLE
REGION
5 POINT
0.5%
76 - 76
76 - 77
74 - 75
73 - 74
77 - 77
80 - 81
82 - 83
X
45 - 45
49 - 49
X
X
X
86 - 86
86 - 86
89 - 89
49 - 49
53 - 53
51 - 51
X
2%
73 - 80
74 - 79
72 - 77
71 - 76
75 - 78
79 - 82
79 - 85
X
44 - 46
48 - 50
43 - 43
48 - 49
50 - 51
85 - 87
86 - 87
88 - 90
49 - 50
52 - 54
50 - 52
61 - 62
5%
67 - 82
68 - 82
65 - 81
65 - 80
71 - 81
75 - 85
71 - 89
75 - 80
43 - 48
47 - 52
42 - 43
47 - 50
49 - 52
83 - 89
83 - 89
85 - 92
47 - 52
51 - 56
49 - 53
59 - 64
4 POINT
0.5%
78 - 78
79 - 79
78 - 78
76 - 76
80 - 80
84 - 84
87 - 88
X
X
X
37 - 37
41 - 41
X
31 - 31
X
95 - 95
X
43 - 43
41 - 41
52 - 52
2%
77 -880
77 - 81
76 - 79
74 - 78
79 - 81
83 - 85
86 - 89
X
36 - 37
39 - 39
37 - 37
41 - 41
43 - 43
31 - 31
32 - 33
94 - 95
39 - 39
43 - 43
41 - 41
52 - 52
5%
73 - 83
74 - 83
72 - 82
70 - 81
76 - 83
81 - 87
81 - 92
81 - 81
35 - 38
38 - 40
36 - 38
40 - 42
43 - 44
29 - 32
31 - 34
93 - 97
38 - 41
42 - 44
40 - 42
51 - 54
co
ro
i
-------
Table 7 (continued)
CASE C. FLOW AROUND A CORNER
RESULTS OF RECTANGULAR DUCT ANALYSIS
BY FOUR AND FIVE POINT METHODS
T Ml r~T
INLET
1
3/4
5/8
1/2
1
3/4
1/2
A I ft
3/4
1/2
1
3/4
1/2
1
3/4
1/2
\M n it i T~
VANE
0
0
0
0
1R
1R
1R
1L
1L
1L
2R
2R
2R
2L
2L
2L
METHOD:
VELOCITY
ACCURACY :
NET ACCEPTABLE
REGION
5 POINT
0.5%
43 - 44
44 - 45
44 - 44
53 - 55
45 - 45
42 - 42
46 - 46
41 - 42
35 - 35
49 - 50
43 - 44
42 - 43
28 - 28
52 - 53
29 - 32
51 - 53
X
2%
40 - 46
42 - 47
42 - 46
48 - 61
42 - 48
40 - 45
43 - 48
38 - 45
33 - 36
45 - 53
40 - 46
39 - 46
26 - 30
49 - 55
25 - 36
45 - 56
X
5%
35 - 53
37 - 53
39 - 51
42 - 74
37 - 54
35 - 51
39 - 53
33 - 53
30 - 40
40 - 62
35 - 52
34 - 53
25 - 33
41 - 60
25 - 52
25 - 62
42 - 51*
4 POINT
0.5%
36 - 36
37 - 37
38 - 38
38 - 38
37 - 38
35 - 35
39 - 39
34 - 34
31 - 31
40 - 40
36 - 37
35 - 35
X
51 - 52
30 - 32
52 - 53
X
2%
34 - 38
35 - 39
36 - 40
36 - 40
36 - 40
33 - 36
37 - 41
32 - 36
29 - 32
38 - 43
35 - 39
33 - 37
31 - 34
49 - 54
25 - 35
48 - 56
X
5%
30 - 42
35 - 42
35 - 43
35 - 45
35 - 44
30 - 40
34 - 44
29 - 41
27 - 34
33 - 48
35 - 43
30 - 41
28 - 36
43 - 58
25 - 40
25 - 61
35 - 40*
t
co
CO
I
*0mits 3 runs
-------
determined by inspection of the duct work. For example, for Case C
the rows were always being taken parallel to the plane of the elbow
rather than normal to the plane of the elbow. Results are tabulated
in Table 8. Overall results were very encouraging, but those for
Case C, the elbow, were especially good. These results were further
analyzed, and results are given in Table 9. For these 16 runs, a very
consistent pattern of higher row averages toward the walls and lower
row averages toward the center was observed, and the total row average
variation was only about 14%. This shows that for the specific con-
figuration used, a row average taken practically anywhere across the
duct would agree fairly closely with the overall average velocity.
As mentioned above, an Annubar was installed for Cases B and C
as shown in Figure 24. Since it was horizontal, as was the elbow in
Case C, it was in effect installed along a row. A reading from the Annubar
was taken at the conclusion of each pi tot traverse for a total of 35
runs. The data reduction scheme was as follows:
V = average velocity
R = gas constant
T = static temperature
p^ = static pressure
= pitot probe differential pressure
12 R T
V. = velocity from Annubar reading
k = calibration factor
= Annubar differential pressure
-84-
-------
Table 8. RESULTS OF RECTANGULAR DUCT ANALYSIS BY ROWS
X INDICATES .95
row
1.05
CASE A. PARTIALLY DEVELOPED FLOW
INLET
1
3/4
1/2
1/4
V8
AVERAGE (%)
ROW NUMBER
1
X
X
X
60
2
X
X
X
60
3
X
X
X
X
80
4
X
X
X
X
80
5
X
X
X
60
6
X
X
X
60
7
X
X
X
X
X
100
8
X
X
40
-85-
-------
Table 8 (continued). RESULTS OF RECTANGULAR DUCT ANALYSIS BY ROWS
CASE B. TWO STREAM FLOW
INLET
1
.1
1
1/16
1
3/4
1/2
1
3/4
5/8
1/2
1/8
1
3/4
1/2
1
3/4
1/2
1/2
VANE
POSITION
1
1
1
1
2
2
2
2.5
2.5
2.5
2.5
2.5
3
3
3
4
4
4
5
AVERAGE (%)
ROW NUMBER
1
X
X
X
X
21
2
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
79
3
X
X
X
16
4
X
X
X
16
5
X
X
X
X
X
26
6
X
X
X
X
X
X
32
7
X
X
X
X
X
X
X
X
X
X
X
X
X
68
8
X
5
-86-
-------
Table 8 (continued). RESULTS OF RECTANGULAR DUCT ANALYSIS BY ROWS
CASE C. FLOW AROUND A CORNER
INLET
1
3/4
5/8
1/2
1
3/4
1/2
1
3/4
1/2
1
3/4
1/2
1
3/4
1/2
VANE
POSITION
0
0
0
0
1R
1R
1R
1L
1L
1L
2R
2R
2R
2L
2L
2L
AVERAGE (%)
OVERALL
AVERAGE (%)
ROW NUMBER
1
X
X
12
22
2
X
X
X
X
X
X
X
X
X
X
X
69
72
3
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
100
58
4
X
X
X
X
X
X
X
X
X
56
40
5
X
X
X
X
X
X
X
X
X
56
42
6
X
X
X
X
X
X
X
X
X
X
X
X
X
X
. X
94
60
7
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
100
85
8
X
X
X
X
25
18
-87-
-------
Table 9. EXAMINATION OF 1973 ROW AVERAGE DATA AFTER
A RECTANGULAR ELBOW
Date/Run
4-5 1
4-6 2
3
4
5
6
7
8
4-9 1
2
3
4
5
6
7
8
Average
0, %
Location, Fraction of Duct Width
.062
1.051
1.071
1.065
1.055
1.051
1.085
1.102
1.133
1.075
1.076
1.089
1.124
1.095
1.072
1.081
1.095
1.082
+2.1
.188
1.005
.997
.990
1.000
1.065
1.026
1.052
1.086
.996
1.007
1.008
1.048
1.070
.995
1.012
1.014
1.023
+3.0
.312
.991
.948
.964
.967
1.030
.994
1.015
.912
.964
.960
.965
1.016
1.000
.962
.986
.932
.975
+3.3
.438
.969
.929
.956
.967
.935
.923
.956
1.002
.957
.928
.931
.915
.951
.952
.945
.929
.947
+2.4
.438
.963
.951
.962
.968
.927
.963
.944
.980
.953
.933
.940
.922
.892
.967
.950
.929
.946
+2.3
.312
.972
.997
.979
.979
.976
.954
.950
.956
.976
.982
.956
.931
.959
.974
.962
1.008
.969
+1.9
.188
.994
1.031
.1.021
1.002
1.004
.980
.967
.993
.996
1.031
1.010
1.007
.978
1.002
.993
1.041
1.003
+2.0
.062
1.064
1.080
1.065
1.058
1.010
1.074
1.019
.934
1.077
1.088
1.102
1.037
1.057
1.072
1.076
1.049
1.054
+3.7
Tabular values = Row avera9e at Iocat1on noted
Average from complete traverse
-88-
-------
For accurate operation, V = V», so that
2 R T
The Annubar may be defined as acceptable for use in a particular
installation if the calibration factor k is a known function of the total
volumetric flow. Preferably, k should be constant for a particular
installation. The instrument would be considered universal if k were
constant for all installations, i.e., the instrument could be used
without any in-place calibration. The term k is being used to avoid confusion
with the factory calibration factor S for use in circular ducts.
The rectangular duct data were analyzed to determine k for each run
for the two general test configurations. Results are shown in Table 10.
Two runs were omitted from the first group for calculation of mean value
and standard deviation under the assumption of operator error. This left
a net 17 runs in the first group and 16 in the second group for consideration.
For the 17 runs in the first group, the maximum deviation from the mean
value for k was less than 5 percent, and less than 6 percent for the second
group. The largest error based on the mean value for all 33 runs was
8.1 percent. The data summary presented in Table 11 is encouraging. It
shows the mean value of k to be quite constant for each of the two
configurations: for the first group, use of the mean value of k would
result in an average velocity error of less than 4.6 percent, 95 percent
of the time, and an error of less than 6 percent for the second case.
Using the overall mean value would result in velocity errors of less than
7.2 percent, 95 percent of the time.
1973 results were encouraging for the Row Average and Annubar techniques.
The purpose of the 1974 testing was then to broaden the data base to get a
better estimate of the universality of these methods.
-89-
-------
Table 10. ANNUBAR DUCT MEASUREMENT RESULTS, 1973
PUN
r\Uli
NO.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
v
M/sec
2.2
10.3
17.0
17.3
15.6
14.2
12.4
13.3
12.6
16.6
14.5
18.8
20.0
20.0
20.5
14.5
5.5
8.2
10.9
k
.7085
.6554
.6393
.6824
.6809
.6767
.6846
.6896
.6924
.6799
.6395
.6776
.6533
.7192
.6677
.6684
.6685
.6709
.6906
% ERROR FOR k
BASED ON AVERAGE k
FOR TWO-STREAM FLOW
+5.494
-2.412
-4.809
+1.608
+1.385
+0.759
+1.936
+2.680
+3.097
+1.236
-4.780
+0.893
-2.725
+7.087
-0.581
-0.476
-0.462
-0.104
+2.829
BASED ON OVERALL
AVERAGE FOR K
+8.008
-0.087
-2.541
+4.029
+3.800
+3.160
+4.365
+5.127
+5.554
+3.648
-2.511
+3.297
-0.407
+9.639
+1.788
+1.895
+1.910
+2.276
+5.279
A. RESULTS FOR TWO STREAM FLOW CASE
Calibration factor k calculated from
2 R T
-90-
-------
Table 10 (continued). ANNUBAR DUCT MEASUREMENTS RESULTS, 1973
RUN
NO.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
V
M/sec
19.9
20.1
20.6
15.0
16.0
16.5
15.0
12.4
16.3
13.6
15.8
13.3
11.5
12.4
12.9
11.3
k
.6520
.6616
.6722
.6551
.6400
.6365
.6557
.6026
.6431
.6215
.6333
.6510
.6494
.6082
.6301
.6169
% ERROR FOR k
BASED ON AVERAGE k
FOR CORNER FLOW
1.987
3.488
5.146
2.471
0.109
1- .438
2.565
-5.741
.594
-2.784
- .939
1.830
1.580
-4.865
1.439
-3.503
BASED ON OVERALL
AVERAGE FOR k
- .605
.858
2.474
- .133
-2.435
-2.968
- .041
-8.136
-1.962
-5.255
-3.456
- .757
-1.001
-7.282
-3.944
-5.956
B. RESULTS FOR FLOW AROUND A CORNER CASE
Calibration factor k calculated from
V2 R T
-
D
-91-
-------
Table 11. SUMMARY OF ANNUBAR DUCT MEASUREMENT RESULTS, 1973
A
B
OVERALL
ave
.6716
.6393
.6560
STANDARD DEVIATION a
ACTUAL
±.015
±.019
±.024
I
±2.3
±3.0
±3.6
2a
%
±4.6
±6.0
±7.2
Explanation of standard deviation (a):
For a normal distribution, there is a 68 percent
chance that the measurement error will be less than
la, and a 95 percent chance that the error will be
less than 2a.
Runs 1 and 14 from group A were omitted from
calculations of k and a
E
•V?
(k -
n = number of runs
-92-
-------
6.4.2.2 1974 Testing-
Inlet configurations used for the rectangular mapping were analogous
to those for the circular case: a straight inlet and a mitered elbow
inlet as shown in Figures25 and 29. Emphasis was placed on testing
behind an elbow for two reasons: good results were obtained there for
1973 testing, and an elbow is the most common flow disturbance in
rectangular duct systems. The reference total flow measurements were
taken in the reference section in the same manner as for circular duct
testing, so that complete traverses of the test section were not required.
Two techniques were used to obtain data: the row averaging technique
developed in 1973 and the Annubar. The Annubar, as before, was placed
along the center "row." In this case, the rows were parallel to the
short sides unlike in 1973 when they were parallel to the long sides.
Recall that the rows are defined aa being in the plane of the elbow for
that configuration. For the straight inlet, blockage with the 4" x 8"
aluminum plates was generally applied in a manner consistent with the
definition of rows being taken in the plane of maximum velocity strati-
fication. A total of 48 runs was performed for the straight inlet case,
and a total of 40 runs was performed for the rectangular inlet case.
Blockage configurations are shown in Figure 36. In most cases, row
averaging data was taken at two locations: 18.9% and 13.9% of the
duct height from the top of the duct. The 18.9% position was chosen
since the most consistently accurate data in the 1973 testing was
obtained at 18.8% of the duct width.
Data from the straight inlet configuration, as shown in Tables 12
and 13 generally agreed with the 1973 results. The average Annubar cali-
bration factor in the test section was .646 for 46 of the 48 runs, compared
with .656 for the 1973 testing. The standard deviation, however, was up
significantly, from 3.6% to 9.0%. It is apparent that this increase in
the data spread was due to a much greater variety of inlet conditions for
the 1974 tests. The deviation for this case was also significantly higher
than for either circular duct case discussed in the last section. It
would seem apparent, then, that the Annubar cannot be expected to give
as consistent results in a variety of rectangular duct applications as in
a variety of circular duct applications. The data do tend to indicate,
-------
Shaded portion of inlet blocked off. Each set of four runs
1 was performed at four different flow rates .,
2
Runs 1-4,
33-36, 49-52
5-8, 69-72 9-12, 57-60 13-16, 61-64
8
17-20, 77-80 21_24j 73_76 25-28, 81-84 29-32, 65-68
10
12
37-40
41.44 45-48, 85-88
53-56
Figure 36. Rectangular inlet blockage - 1974
-94-
-------
Table 12. 1974 ROW AVERAGE RESULTS AFTER A STRAIGHT INLET
(CONFIGURATION 1)
Run
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
M3/sec
6.72
4.76
3.97
2.62
5.74
4.45
3.94
2.64
6.62
4.80
4.07
2.74
4.71
4.07
3.61
2.62
4.48
4.09
3.73
2.25
4.36
3.90
3.61
2.60
xc
.159
.192
.134
.122
.128
.130
.158
.144
.133
.131
.129
.124
.150
.142
.171
.177
.177
.179
.176
.178
Location,
Fraction of
Duct Height
.189
1.077
1.051
1.039
1.068
.998
1.029
1.111
1.056
1.073
1.077
1.160
1.225
1.089
1.030
1.029
1.263
1.064
1.084
1.060
1.091
.139
.949
.962
1.003
1.028
1.020
1.009
.957
.991
1.017
1.047
1.045
1.106
.940
.995
.948
.909
1.151
.791
.655
.839
.675
Run
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
3 V
.IT/sec
5.35
4.40
4.03
2.79
5.25
4.40
4.10
3.00
6.86
5.05
4.22
2.91
4.90
4.19
3.94
2.84
6.20
4.65
4.11
2.89
5.33
4.60
4.07
2.81
XC
.119
.195
.166
.172
.236
.158
.153
.146
.172
.187
.168
.171
.159
.147
.133
.113
.136
.182
.232
.190
.227
Location,
Fraction of
Duct Height
.189
1.112
1.124
1.095
1.156
.986
1.097
1.066
.921
1.052
1.051
1.071
1.052
1.005
1.052
1.086
1.062
1.147
1.193
1.186
1.278
1.034
.865
.994
.886
.139
1.028
.872
.886
.878
.838
.969
.980
.988
.900
.890
.930
.850
.960
.973
1.020
1.064
1.015
.780
.709
.758
.734
XC = Interpolated location, in fraction of duct height, where row
average equals V
Other tabular data = Row averac*e at 1ocation ™ted
V
V obtained in reference section
-95-
-------
Table 13. ANNUBAR CALIBRATION FACTORS FOR 1974
CALIBRATION TESTS - CONFIGURATION 1
Run
' 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
3 V
MJ/sec
6.7
4.8
8.0
2.6
5.7
4.5
3.9
2.6
6.6
4.8
4.1
2.7
4.7
4.1
3.6
2.6
4.5
4.1
3.7
2.3
4.4
3.9
3.6
2.6
S
REF
.697
.670
.664
.629
.694
.669
.663
.592
.695
.687
.676
.644
.732
.669
.634
.655
.677
.650
.665
.535
.669
.685
.657
.639
k
TEST
.708
.683
.676
.639
.677
.646
.629
.599
.684
.690
.677
.659
.670
.670
.610
.563
.633
.633
.633
.504
.568
.563
.567
.526
Run
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
V
M3/sec
5.4
4.4
4.0
2.8
5.2
4.4
4.1
3.0
6.9
5.0
4.2
' 2.9
4.9
4.2
3.9
2.8
6.2
4.7
4.1
2.9
5.3
4.6
4.1
2.8
S
REF
.709
.700
.719
.676
.675
-
.644
.490
.689
.680
.681
.665
.652
.637
.650
.652
.695
.668
.683
.649
.675
.666
.662
.612
k
TEST
.758
.739
.761
.746
.672
-
.614
.481
.709
.692
.690
.647
.603
.593
.595
.560
.695
.700
.715
.662
.570
.608
.569
.559
S = .668, ac = + .026 (+3.9%)
o —
T = -646 a. = .058 (+9.0%)
* K
Runs 20 and '32 not
included
-------
however, that the data spread as a function of flowrate for a specific
configuration will be acceptably small. Data for the sets of four runs
which correspond to specific inlet conditions show much greater consistency
than is indicated overall. The result would be a suggested value of .65
for the Annubar calibration factor for general use in rectangular ducts,
subject to revision through in place calibration. Another possibility is
that the accumulated test data can be used to predict a calibration factor
for a specific installation. This was done to predict a calibration
coefficient for field test use, and results are described in Section 1 ,
The row mapping data for the straight inlet configuration agreed
reasonably well with 1973 results, but did show some undesirable features.
As mentioned above, the 1973 data showed the best results at 18.8% of the
duct width from ore wall. Since the rows were 12.5% of the duct width part,
however, resolution of the most desirable location was low. The bulk of
the 1974 data was taken at two locations only 5% of the duct height apart,
so that resolution was improved considerably. Data at this interval
allowed interpolation to better locate the exact row location for agree-
ment with the overall average, and this location turned out to be 16.1%
of the duct width from one wall, which agrees with 1973 results.
The disturbing part of the results is that the data scatter for this
location and for the averages obtained for the two test rows is fairly
large, as shown in Table 12 . What is even more disturbing is that there
was an average velocity gradient of 14% between the two rows, which were
only 5% of the duct width apart. Such a high gradient in the vicinity of
the measurement is clearly undesirable, since it would tend to lead to
high amounts of data scatter. For several cases, though, the scatter was
a function of the inlet configuration changes rather than for flowrate
changes for a specific inlet configuration, just as was the case for the
Annubar. As shown in Table 14, for several inlet conditions there was
a net region where the row average agreed with the overall average within
5% for all four flowrates, which is encouraging.
-97-
-------
Table 14. ACCEPTABLE FLOW MAPPING REGION FOR CONSTANT INLET
GEOMETRY AND VARIABLE FLOWRATE FOR 1974 MAPPING
TEST-STRAIGHT RECTANGULAR INLET
RUNS
1-4
5-8
9-12
13-16
17-20
21-24
25-28
29-32
33-36
37-40
41-44
45-48
NET ACCEPTABLE ROW MAPPING REGION
FOR 5% ACCURACY FOR FOUR RUNS, % OF DUCT HEIGHT
Insufficient Data
12.2 - 15.4
13.6 - 15.0
None
None
17.3-18.4
None
None
15.6-17.6
16.5-18.2
13.3-13.5
None
-98-
-------
At first glance, the rectangular elbow data would appear to be
disastrous, but was in fact very instructive. The test section measure-
ments were taken further downstream from the elbow than in 1973 to
evaluate the effect of that parameter. The results revealed that probe
placement relative to an elbow is very important. Results for the 40 runs
are shown in Tables 15 and 16, and indicate virtually no agreement with
1973 results. The greater distance from the elbow is clearly the reason
why, and it gave the following insight into flow around a rectangular elbow.
Flow around a mitered rectangular elbow creates a high velocity
region toward the outside of the elbow and a low velocity recirculation
region toward the inside. The presence of the elbow results in a somewhat
two dimensional flow immediately downstream of the elbow, where "uniform
two dimensional flow" is explained as follows: Consider a coordinate
system in the duct with the x axis parallel to the duct axis, the y axis
in the plane of the elbow, and the z axis normal to the plane of the elbow.
The flow would be two dimensional if the velocity vector at any point had
components only in the x and y directions. It would also be uniform if the
profiles in the x - y plane were the same for all values of z. This
would cl early. 6e tb7e tdeal situation for use of the Row Average Method,
since all row averages would be identically equal to the total average.
1973 results indicated that an elbow tends to produce this ideal uniform,
two-dimensional flow. The 1973 data showed that the highest row averages
occurred toward the outside of the duct, leaving a lower average in the
center. The 1974 data conclusively showed that this condition worsens as
the flow proceeds downstream: the recirculating region recedes from the
ends of the duct, creating higher velocities there, and grows in the
center creating lower velocities. Thus the relatively flat profile
becomes very three dimensional, with high velocities on three sides of
the duct, and a low velocity region in the center, all of which is
reflected in the data. The Annubar readings were consistently low,
resulting in a very high calibration factor, .853, which was 31% higher
than the .65 obtained for other tests. A systematic increase in the
row averages obtained at the same locations as for the straight inlet also
occurred, as well as a change in the direction of the gradient between
the two rows. The one desirable result is that the magnitude of the gradient
decreased by a factor of three.
-99,
-------
Table 15. 1974 ROW AVERAGE RESULTS
(CONFIGURATION 2)
AFTER A RECTANGULAR ELBOW
Run
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
V
m /sec
' 5.6
4.6
4.0
2.9 ,
3.3
3.1
3.1
2.4
5.4
4.5
4.0
3.0
4.7
4.3
4.0
3.1
4.7
4.2
3.9
3.1
Location,
Fraction of
Duct Height
.239
1.006
1.171
1.090
1.012
1.090
1.047
.951
1.056
1.063
1.069
1.037
1.042
1.070
1.133
.189
1.284
1.180
1.181
1.266
1.221
1.265
1.218
1.212
1.061
.999
1.093
1.101
1.060
1.086
1.153
1.151
1.051
1.125
1.080
1.006
.139
1.317
1.426
1.143
1.176
1.222
1.209
1.025
1.105
1.083
1.000
Run
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
V
m /sec
5.3
4.6
4.1
3.2
4.0
3.8
3.4
2.8
4.5
4.3
3.9
3.1
5.1
4.5
4.2
3.2
5.4
4.7
4.3
3.2
Location,
Fraction of
Duct Height
.239
1.060
1.091
1.244
.189
1.190
1.191
1.245
1.229
1.264
1.138
1.177
1.252
1.099
1.055
1.189
1.151
1.050
1.028
1.053
1.074
1.259
1.250
1.140
1.215
.139
1.236
1.198
1.275
1.279
1.210
1.190
1.196
1.192
1.180
1.194
1.225
1.320
1.051
1.068
1.268
1.276
1.214
1.247
Tabular values = Row avera^e at 1ocation noted
V
V obtained in reference section
-100-
-------
Table 16. ANNUBAR CALIBRATION FACTORS FOR 1974
CALIBRATION TESTS - CONFIGURATION 2
Run
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
V
m /sec
5.6
4.6
4.0
2.9
3.3
3.1
3.1
2.4
5.4
4.5
4.0
3.0
4.7
4.3
4.0
3.1
4.7
4.2
3.9
3.1
S
REF
.729
.734
.726
.692
.650
.684
.685
.674
.717
.787
.697
.706
.773
.746
.751
.724
.734
.744
.765
.729
K
TEST
.769
.759
.788
.792
.737
.808
.829
.736
.741
.803
.740
.736
.896
.831
.865
.870
.912
.891
.848
.858
Run
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
V
m /sec
5.3
4.6
4.1
3.2
4.0
3.8
3.4
2.8
4.5
4.3
3.9
3.1
5.1
4.5
4.2
3.2
5.4
4.7
4.3
3.2
S
REF
.779
.750
.731
.714
.746
.729
.705
.755
.766
.738
.752
.722
.704
.717
.719
.687
.763
.751
.764
.801
K
TEST
.824
.906
.936
.906
.974
.959
1.058
1.055
.819
.863
.803
.833
.818
.766
.745
.721
1.046
.970
.920
1.003
S = .7315, os =+ .032 (+4.4%)
~K~= .853, 0(( = + .093 (+10.9%)
-101-
-------
The Annubar in the reference section also apparently showed the effects
of the elbow. For the straight inlet configuration, the average Annubar calibra-
tion factor was .668, lower than the .698 for the circular case. It is
felt that this lower value reflects the transition from a rectangular to
a circular duct at the entrance to the reference section. The narrow
rectangular duct would tend to create a high velocity region along the
duct center, which would affect the Annubar reading more than it would
the Log Linear reference traverse, since the latter uses no center!ine
data. The higher local velocity would increase the Annubar reading, thus
lowering the calibration factor from the circular case. For the elbow
inlet case, the low velocity region created apparently persisted into the
reference section. Again, since the Annubar responds more to the center-
line flow conditions than does the Log Linear method, the local low
velocity region resulted in a relatively low Annubar reading, and thus a
higher calibration factor of .7315.
6,4.2.3 1975 Testing -
This testing was performed solely to determine the optimum location
downstream of a rectangular elbow for Row Average and Annubar measurements.
The mapping facility was set up as in configuration 2 for the 1974 testing
(Figure 25). The test section was initially located as for the 1974 tests,
then moved forward toward the elbow. Locations for the pi tot probe and
Annubar for each case are shown in Figure 30. Note that the locations
are given in terms of the local duct width instead of the effective duct
diameter. The flow properties of interest vary as a function of the duct
width. Six row averages were taken over half the duct for each run, along
with the normal Annubar data. Reference measurements were taken in the
reference section as usual. At each location, runs were performed at
maximum and minimum flow rates for inlet blockage conditions 1, 2 and 6
as shown in Figure 36. A total of 48 runs was performed for row averaging,
and a total of 60 for the Annubar. Results are summarized in Tables 17 and
18.
-1Q2-
-------
Table 17. ROW AVERAGE RESULTS FOR 1975 TESTING
TEST
SECTION
LOCA-
TION
A
Bl
B2
C
D
E
F
G
ROW AVERAGE LOCATION, FRACTION OF
DUCT HEIGHT
.045
1.305
1.106
1.119
1.112
1.099
1.084
1.083
1.067
.136
1.093
1.026
1.022
1.026
1.045
1.032
1.042
1.009
.227
1.005
.987
.976
.966
.968
.957
.946
.963
.318
.908
.965
.970
.957
.964
.979
.960
.989
.409
.824
.958
.949
.963
.962
.977
.981
.985
.500
.730
.914
.927
.942
.928
.942
.978
.974 j
i
Tabular values are the mean ratio of the row average velocity to
the total average velocity for the six runs performed at each test
section location.
-103-
-------
Table 18. ANNUBAR RESULTS FOR 1975 TESTING
TEST
SECTION
LOCATION
i
i
A
Bl
B2
C
D
E
F
G
H
I
. DISTANCE BETWEEN
ANNUBAR AND ELBOW
AS FRACTION OF
DUCT WIDTH
2.906
2.062
2.062
1.625
1.500
1.3125
1.125
.9376
.75
.562
AVERAGE ANNUBAR
CALIBRATION FACTOR
FOR SIX RUNS
.843
.700
.703
.688
.649
.617
.617
.581
.571
.578
-------
To obtain Table 17, row average velocity data at each row location
for each run were divided by the total average velocity for that run.
These six ratios at each row location were then averaged to obtain the
tabular entries shown. Results confirmed the belief that an elbow
tends to produce a uniform, two-dimensional flow in its immediate
vicinity, and that the flow pattern changes radically further downstream.
Consequently, the results showed that for the positions closest to the
elbow, a row average velocity taken between about 14% and 40% of the duct
width for one wall would agree with the overall average velocity within
±5%.
Data in Table 18 for the Annubar show that a "steady state" calibration
factor as a function of distance from the elbow was not reached, as had
been hoped. At location D, which corresponded to the 1973 location, the
calibration factor was .649, compared with .6393 for 1973 testing behind
an elbow. This agreement is good, but the relatively large changes in
calibration factor with distance from the elbow make it appear unlikely
that good repeatability of the calibration factor would be attained in a
large number of different facilities.
6.5 MAPPING TEST SUMMARY
6.5.1 Circular Duct Techniques
EPA Method 1 flow mapping techniques are effectively identical for
both circular and rectangular ducts -- in each case the duct is divided
into a number of equal area segments and velocities are measured at the
center of each segment. No knowledge of flow profiles is assumed. Mapping
techniques such as the Log-Linear method retain the equal area segment aspect,
but go a step further in probe placement by recognizing that fully developed
pipe flows are axisymmetric and that the flow profile is roughly parabolic
in shape. Symmetry is the most important point, since it means that an
average taken along one diameter is the same as that taken along any other
diameter. This reduces the mapping problem to one of determining a proper
technique to obtain a correct line average along any one diameter. Thus
circular duct techniques such as the Log-Linear method are developed
for the special case of fully developed pipe flow, and then their
-105-
-------
accuracy in non-developed flows must be empirically determined.
Design of the Annubar followed this same route.
Testing in 1973 downstream of an elbow, testing in 1974 downstream
of an elbow, and testing in 1974 downstream of a straight inlet with
partial blockage too"k place at 1, 1.4, and 2.2 duct diameters downstream
for a disturbance, respectively, for pitot traverse testing, and slightly
further downstream for 1974 testing with the Annubar. Three techniques
were evaluated extensively: the Log Linear 4, which requires a total
of eight points, the new empirical, which requires five, and the Annubar.
On the basis of the test data, it is being concluded that the Annubar
and Log Linear 4 methods qualify as being "Universal", and should be
expected to have accuracies on the order of + 5% without the need for in
place calibration. The New Empirical method was found to be acceptable
as a "Single Calibration" method -- establishment of a calibration factor
at one flow rate should be adequate to insure accuracy over the entire
range of flows at one location. Testing showed that good accuracy can be
achieved in applications where EPA Method 1 would require in excess of
forty. The Annubar, New Empirical, and Log Linear 4 techniques require
one, five, and eight measurements, respectively. Results are summarized
in Table 19.
6.5.2 Rectangular Duct Techniques
The flow profile symmetry which is the basis for circular duct
mapping techniques does not exist in fully developed flows in rectangular
ducts due to three dimensional effects. Work during the program did,
however, identify a common situation in which flow profiles are very
similar in magnitude and shape across the duct. This situation occurs
immediately downstream of an elbow, which is the most common flow
disturbance in rectangular duct systems in process streams. The row
average and Annubar methods which were successfully used during the
program achieved success by averaging flow along a line between the
duct walls in a consistent manner for a variety of flow conditions.
The critical point is that the flow properties in the duct were such
that the average velocity along the line chosen was representative
of the average velocity for the entire duct. Thus when success was
-106-
-------
Table 19. RECOMMENDED CIRCULAR DUCT FLOW MEASUREMENT TECHNIQUES
A. Single Probe Techniques
Ellison Annubar - Install and use according to factory
instructions
*
B. Point Sensor Array or Traversal Techniques
Preferred:
Log Linear 4
Acceptable1:
New Empirical
Gauss Sherwood 4
Log-Linear 2+4
*
Methods shown in Table 2
In place calibration check considered a requirement; in
place calibration recommended for Log Linear 4 and Annubar
-------
achieved in a rectangular duct, it was for the same reason as in a
circular duct -- similarity of flow profiles.
Testing showed that accurate measurements could be taken downstream
of an elbow with the Row Average and Annubar methods because the elbow
causes a violent disruption of the flow, which accomplishes two important
results: 1. The disruption tends to damp out influences of disturbances
further upstream, resulting in consistent conditions at the test area.
2. The elbow tends to condition the flow to a uniform, two-dimensional
shape immediately downstream of the bend. Testing with the straight
inlet in 1974 showed that both methods work reasonably well for that
general case, but are more consistent downstream of an elbow.
The recommended circular mapping techniques have been in existence
for several years, so a body of data exists for them outside of this
program. It was partly on the strength of this additional work that
the Annubar and Log Linear techniques were recommended as being "Universal".
Since a body of outside data does not exist for Annubar use in rectangular
ducts or the Row Average method, it is felt that it would be premature at
best to turn either technique "Universal". Test data indicate that the
Annubar will probably always need in place calibration at one flow rate,
preferably near the mean flow rate, for use in rectangular ducts, but
that this calibration should result in continuous monitoring accuracies
on the order of +5%. There is good reason to believe that the Row Average
technique is in fact "Universal" downstream of an elbow, but more data
should be accumulated as a final judgement is made. Recommended installations
for the Row Average and Annubar techniques are given in Figures 37 and 38.
Test conclusions are given in Table 20.
-108-
-------
1
o
ROW AVERAGE:
REFINE ESTIMATED CALIBRATION
FACTOR OF 1.00 BY IN-PLACE
CALIBRATION
ANNUBAR:
REFINE ESTIMATED CALIBRATION
FACTOR OF 0.65 BY IN-PLACE
CALIBRATION
MEASUREMENT
PLANE
I
0>
FOR VELOCITY PROFILES AS SHOWN IN Y
AND Z DIRECTIONS, ROWS ARE DEFINED
AS BEING IN Z DIRECTION
VELOCITY
ROW AVERAGE
LINE (CHOOSE ONE)*
ANNUBAR
VELOCITY
,19W
\_
^
t
• • • oi
i
I
t
• JW
t
.19W
*ROW AVERAGE MAY BE TAKEN
ANYWHERE IF SUPPORTED BY
ADEQUATE CALIBRATION DATA.
Figure 37. Probe placement for general rectangular duct applications
-------
CD
I
ROW AVERAGE:
REFINE ESTIMATED CALIBRATION
FACTOR OF 1.00 BY IN-PLACE
CALIBRATION
ANNUBAR:
REFINE ESTIMATED CALIBRATION
FACTOR OF 0.65 BY IN-PLACE
CALIBRATION
MEASUREMENT
PLANE
FLOW
DIRECTION
ANNUBAR: X= 1.50^ (NOMINAL)
ROW AVERAGE: X=
+.40W
'i
ROWS ARE DEFINED TO BE IN
THE PLANE OF THE ELBOW
ANNUBAR
ROW AVERAGE LINE
(CHOOSE ONE)*
-.04 W
*ROW AVERAGE MAY BE TAKEN
ANYWHERE IF SUPPORTED BY
ADEQUATE CALIBRATION DATA.
+.25 W0
-.19W2 2
*• -.04 W2
.5Wo
-.60W
1
Figure 38. Probe placement after a rectangular elbow
-------
Table 20. RECOMMENDED RECTANGULAR DUCT FLOW MEASUREMENT TECHNIQUES
A. Single Probe Techniques
Ellison Annubar - Install as shown in Figures 37-38.
In-place calibration required to modify estimated correction
factor of .65
B. Point Sensor Array or Traversal Techniques
Row Average Method - Install as shown in Figures 33-34. For
array, use of eight sensors recommended. In place calibration
required to minimize systematic error. A row is defined as
being in the direction of maximum velocity stratification.
-Ill-
-------
SECTION VII
TASK IV - LABORATORY ASSESSMENT
7.1 FACILITY DESCRIPTION AND SCOPE OF TESTING
Two types of tests were performed during this task -- calibration
and environmental. The bulk of the calibration work was performed in the
Fluid Mechanics Laboratory Low Speed Wind Tunnel, shown in Figure 39.
The wind tunnel had a closed rectangular test section 28.6 cm square
during the testing period. The nominal speed capability was 3-30 m/sec.
Reference velocity measurements were provided by standard hemispherical
nosed pitot static probes (calibration factor 1.00) connected to a cali-
brated Baratron pressure transducer. Temperature measurement was provided
by a calibrated chromel-alumel thermocouple immersed in the flow. Addi-
tional calibration work was performed on the Ramapo Drag meter using
calibrated weights, and on the TSI hot film sensors using a standard
Thermo Systems Calibration Wind Tunnel (Figure 40)•
Environmental testing was primarily performed in the TRW Process
Simulator (Figure 41) to determine sensor resistance to temperature and
humidity. The facility also had a particulate generation capability,
but use of particulate was prohibited by local Air Pollution Control
District officials unless a scrubber was also operating. Since a scrubbing
capability was not available at the time, it was necessary to evaluate
particulate effects through use of a portable sandblaster, which proved
to be adequate for test purposes.
Calibration tests were selected to characterize test sensors as
fully as possible within the program scope. The primary calibration was
a basic velocity calibration in a uniform laminar flow with the test sensor
directly aligned with the flow. Time response tests were performed to
verify that the sensors could adequately follow flow variations. Stability
testing was performed to determine instrument drift characteristics.
Finally, orientation sensitivity tests were performed to characterize the
ability of the sensors to resolve the required axial component of the flow.
This testing is important because in the applications of interest, it
-------
EXHAUST
LIQOJD I*; SUPPLY
FOR THERMAL CONTROL
DIFFUSER
TRANSVERSWG
MECHANISM
AND PROBE
SUPPLV SECTION FOR
RECfRCULATEDGAS
WATER SPRAY
SECTION
TEST SECTION
Figure 39. Low speed wind tunnel
-113-
-------
THERMAL ANEMOMETER PROBES
CALIBRATORS AND CALIBRATION
CALIBRATION APPARATUS • MANOMETERS • CALIBRATION SERVICE
CALIBRATORS
'Air
Water
• Other Fluids
MODEL 1125 CALIBRATORS FOR AIR
• HIGH ACCURACY
• CALIBRATE ANY PROBES FROM 0.1 FT/SEC TO MACH # 1
• EASY TO USE
• CAN BE USED WITH MOST GASES
The Model 1125 Calibrator is designed to calibrate hot wire and hot film
anemometer probes over a wide velocity range. Reference calibration
data is furnished giving velocity vs. pressure drop for each range.
SPECIFICATIONS:
Velocity Range: Probes can be calibrated from 0.1 ft/sec, to
MACH #1 13 cm/sec to M #1) by appropriate
use of interior chamber, interior nozzle and
exterior nozzle.
Accuracy: t 1% above 10 fps; + 2% 0.5 to 10 fps; ^ 10% below
0.6 fps
Background Turbulence Intensity: <0.1% at 100 fps
Nozzle Sizes: 2.83" (72mm) Dia. Interior Chamber;
0.65 (16.5mm) Dia. Interior Nozzle: 0.15" (3.8mm)
Dia. Exterior Nozzle.
Temperature Limits: 0 - 65°C (32 - 150°F)
Pressure Limit: 7 ATM Internal (20 ATM Max. Into Regulator)
Power Required: 115V AC 60 Hz or 230V AC 50 cos (specify)
Size: 20" High x 18" Wide x 10" deep
Weight: 15* (7 kg)
"•-.*.
ComprMMd Air
supply ix Gn
MODEL 1125 BLOCK DIAGRAM
MODEL 1127 CALIBRATOR FOR WATER
The Model 1127 Calibrator is set up to utilize tap water for calibration of water probes.
SPECIFICATIONS
Range: Three Calibrate Positions:
0.02 to 0.2 fps (.6 to 6 cps)
0.15 to 1.5 fps 14.5 to 45 cpsl
1.5 to 30 fps 1.45 to lOmps)
Line Water Pressure: Must be at Least 10 psig (50 cm Hg) (If tine ,
pressure varies too much a constant head tank and sump pump can be
used pumping from an over flow lank to recirculate water]
Equipment Furnished: Tank, course regulator, fine regulator 1 ^ filter,
complete calibrator, extra larger nozzle, finings, valves and hoses.
(Manometer not furnished - Model 10134 or equivalent recommended)
MODEL 1128 CALIBRATOR (Not Shown)
The Model 1128 Calibrator is a combination of Model 1125 and 1127 i
for those who wish to calibrate in both air and water. Model 1128 comes
as one complete Model 1125 plus the tank and accessories necessary
to convert to the Model 1127 Water Calibrator.
CALIBRATION
CHAMBER
L.ki Mod«i 11 25
40
THERMO-SYSTEMS Inc.
00NORTH CLEVELAND AVE
PAUL. MINNESOTA 55 MJ °
SI7 633 00,0
Figure 40. TSI wind tunnel
-114-
-------
FLOWMETER
FLOW CONTROL VALVE
PRESSURE REGULATOR
" POLLUTANT GAS
MIXER SYSTEM
AIR BLOWER
(1000SCFMTO
.1000 SCFM)
WATER
SPRAY INLET
DRY POLLUTANT
MATERIAL FEEDER
(0.1 TO 10 GRAINS FTJ)
AIR
DAMPER
\
NATURAL
GAS
INLET
L -r
»— [X] - 1
^
NATURAL GAS BURNER
(770FTO400°F)
SCRUBBER
INLET
SECTION A-A
INSTRUMENTATION
PROBE PORTS
O
O
O
O
MATERIAL MIXING
VANES
-PROCESS SIMULATOR SECT ION-
PROCESS
SIMULATOR
MONITORING
SECTION
SCRUBBER
OUTLET
CHARGED
DROPLET
SCRUBBER
VIEWB
CHARGED DROPLET SCRUBBER
EFFLUENT
MONITORIN
SECTION
CONFIGURATION I
(ONE MODULE)
CONFIGURATION II
(THREE MODULES)
.CHARGED DROPLET SCRUBBER.
TEST SECTION
Figure 41. Coal fired combustion flue gas simulator
-------
must be expected that the local flow will not always be parallel to
the duct walls.
Testing of the S type pi tot probe revealed that the probe tested had
a non-constant calibration factor and relatively high sensitivity to small
changes in flow orientation. These findings eventually led to a comparison
between the S type probe and the standard pitot-static probe and the sug-
gestion that the latter be used whenever possible. Pitot-static probe
characteristics have been heavily studied and documented, so no testing
was specifically performed on these probes. Probe characteristics given
were obtained from Reference 3.
7.2 ANALYTICAL INSTRUMENT DESCRIPTION
The instruments tested were described and pictured in Section 5.3. To
understand their calibration characteristics, it was necessary to determine
their analytical properties so that appropriate calibration parameters
could be isolated. The most straightforward way to determine analytical
properties is to construct the desired flow output equation for each
instrument. To do this, consider the flow shown in Figure 2. The flow
cross-sectional area, A, is constant. Recall that the three commonly
desired outputs are volumetric flow at standard conditions, Vg; volumetric
flow at actual conditions, V, and total mass flow, m. Figure 2 illustrates
the relationship among the three as described in Section 4. At each
station, each of the flow parameters is constant. Actual conditions are
given at station 1, where we have
V = uA 04)
and m = puA O5)
At station 2, the static temperature and static pressure have changed so
that they are standard, which has been defined as 20°C and 760 torr. The
flow is considered to be chemically frozen so that no change in molecular
weight occurs. At station 2, we have
vs = USA (is)
and m = p u A (17)
J JO
-116-
-------
Since mass must be conserved in the system, we must have
« •
m = ms
so that PSU$A = puA
or PSUS = pu (20)
also V _ u /0,N
~ 77~ \c-\l
vs us
Instrument properties can be evaluated from these basic relationships.
7.2.1 Pi tot-Static Probe
A properly made pitot-static probe senses the stream dynamic pres-
sure, such that the output, Ap, is. given by
AP = (l/2)Pu2 (22)
so that u = \^ (23)
Since p = ^ (24)
we have u = V-2^ (25)
and
X/2RTAP ,oc.
V p (26)
(27)
so that for the flow quantities we have
V - A (28)
Vs = %; >T (29)
r> /9RTxn . J?nAn
(30)
-117-
-------
7.2.2 "S" Probe and Annubar
Each of these probes' output in a uniform stream can be given by
= k(l/2)Pu2 (31)
where k is an appropriate calibration factor. Performing similar
derivations, we obtain
(33)
(34)
(35)
(36)
The value of k can be deduced from factory data for the "S" probe and is
considered to be constant. The calibration factor for the Annubar in a
uniform stream is not of great importance since the instrument is not a
point sensor. The major calibration evaluation of the Annubar took place
during the mapping tests.
7.2.3 Ramapo Drag Meter
The Ramapo output, fl, is actually a voltage ratio, the ratio of the
strain gage bridge output to the bridge excitation voltage, and is
directly proportional to the drag force on the instrument:
v
(37)
where for the flow range of interest,
D = CD(l/2ku2S (38)
-118-
-------
where Vgout = bridge voltage out
VBin = bnd9e inPut (excitation) voltage
k = calibration constant (known)
D = drag force on sensor
Cpj = disc drag coefficient
p = gas density
u = gas velocity
S = disc frontal area
The factory calibration for the probe lists the factor k as
- c.
" 5>
m f millivolts \
V kg (volt • kilogram )
where the term kilogram is understood to mean the downward force exerted
by a one kilogram mass acted on by the acceleration of gravity at sea
level . Thus we have
E = kCD(l/2)Pu2S (39)
(40)
(41)
(42)
(43)
(44)
The voltage output of the unit tested is a nonlinear function of
velocity. The probe responds to the stream dynamic pressure, so that for
general stream conditions, we have
-119-
u
us
if
Vs
m
Hastings
_ W ^n
' kCp,pS
T
PS ^kCDTS
- A J 2RT*
>kCDpS
_ . Ts J2RpJ?
PS ^kCDTS
_ « J 2pR
A 'RTkCDS
Raydist Flare Gas
J2RT/?
Flow Probe
-------
Vout = kAp = kO/2)pu2 (45)
V2V
—£r 06)
u=
2 RT "out (47)
TCJ
T
us ' j£ 1 kl"" 08)
V= fl * kp («>
VS - * £ Kl^ (»>
s
m= fl >-kRf^ <51>
7.2.5 Thermo Systems Hot Film Sensors
The basic operation of a hot film sensor is as follows: a small
sensing element is electrically heated (i.e., resistance heating) to a
temperature above the ambient stream temperature. The power required to
maintain this temperature overheat is monitored. The power required is
related to the stream velocity by means of a calibration curve. For the
general case, the thermodynamic relationship is given by
Nu - a + b Ren (52)
where
Nu = Musselt Number (heat transfer parameter)
Re = Reynolds number based on sensor diameter
a, b, n = Calibration factors (n^ 0.5)
An important point to note for this type of application is that the
Musselt number and Reynolds number are inversely proportional to the
fluid thermal conductivity and viscosity, respectively. For the type of
anemometer proposed, the output parameter, which is the bridge voltage
VR, may approximately be given as
-120-
-------
1/2
(53)
where
p = ambient pressure
u - freestream velocity
CpC2 = calibration coefficients
so that
m
m -
(55)
(56)
(57)
RT ~~Cr- (58)
7.3 BASIC INSTRUMENT CALIBRATION
A minimum of three calibration runs was performed for each sensor
prior to environmental testing. Estimated accuracies from the manufac-
turers as a function of velocity are shown in Figure 42. Reference
measurement accuracies were as follows: pressure: 0.5% of reading or
better; absolute temperature: 0.3% of reading; accuracy of pitot-static
probe calibration factor: 1% over range of test.
7.3.1 "S" Probe Calibration
For each run, ten data points were taken with the velocity increasing,
and ten with the velocity decreasing. The differential pressures from
the S probe and reference pitot probe were then ratioed to obtain the
calibration factor k$J since
-121-
-------
ro
ro
ACCURACY OF VELOCITY
MEASUREMENT, PERCENT
16 r
14 -
12 -
10 -
6 -
4 -
2 -
10
HASTINGS-RAYDIST
AFI-10K
THERMO-SYSTEMS VT1613
ELLISON "S" PITOT PROBE
ELLISON ANNUBAR (CIRCULAR DUCT ONLY)
RAMAPO MARK V
UNITED SENSOR PITOT-STATIC PROBE
15
20
25
30
35
VELOCITY, METERS/SECOND
Figure 42. Manufacturer's accuracy for instruments tested
-------
upitot-static _ v>/p.s. _' s'p.s. _ v"'Vs._TEST= -852
and
VREF = '823
7.3.2 Ramapo Calibration
The Mark V Drag Meter is shown schematically in Figure 44. The
principle of operation may be recalled as follows: a disc is inserted
into the stream, at the end of a level arm. The drag force on the disc
and on the exposed section of the arm is sensed by a strain gauge bridge
-123-
-------
i
ro
i
l.<
1.42
1.40
1.38
1.36
1.34
1.32
1.30
CALIBRATION FACTOR,
p V
£_ ^00 00
I=ks
-A_
_?~
*.
/v-
_Q
CALIBRATION RUN
A 2
D 3
TAG ON LEFT - V
o
TAG ON RIGHT
^ INCREASING
V DECREASING
MANUFACTURER: ks = 1.477
TEST: kc = 1.376
test
-0
FREE STREAM VELOCITY, V^, METERS/SEC
10 15 20
Figure 43. S probe calibration factor
25
30
-------
DISC
LEVER
ARM
STRAIN
GAGE
BRIDGE
EFFECTIVE
LEVER ARM
LENGTH
±
Figure 44. Schematic of Ramapo Mark V flow meter
-125-
-------
mounted on the enclosed part of the arm. The bridge input and output
voltages are used to determine the force on the arm, and finally the
stream velocity, as given in Section 7.2.3. A typical calibration run
is shown in Figure 45- The line is the reference equation and the points
are test data points. Results are shown more illustratively in Figure 46
for the three standard calibration runs. It should first be mentioned
that the instrument accuracy is better than it would appear in Figure 46,
where it is clear that the Ramapo output became consistently high with
increasing speed. It is believed that the anomaly was due to net down-
ward motion of the disc with increasing speed, as shown in Figure 47.
The probe caused considerable blockage, and thus higher local velocities,
at the disc plane in the wind tunnel. The disc face was initially aligned
with the static ports on the reference pi tot probe. As the wind tunnel
speed increased, the disc moved with respect to the pitot probe, thus
putting the pitot probe into a region of lower and lower relative velocity
with respect to the disc. Consequently, it is important to look at the
data spread in Figure 46 as a key to instrument accuracy. A nominal
accuracy of 1.5% on the reference measurement and a 1% instrument accuracy
are consistent with the observed spread.
Figure 46does show an apparent hysteresis with a magnitude of about
±1/2%—the values for the decreasing side of the velocity calibration are
typically 1% higher than the ascending values. Although hysteresis loops
are not uncommon in cyclic handling of metal rods, it did seem unusual
to see this trend since the disc and arm vibrate quite perceptibly in the
airstream, meaning the loading and unloading are not very smooth, and
since the factory claims that the instrument is hysteresis free. Since the
observed level is ±1/2%, the effect does not degrade the instrument accuracy
of ±1%, and so is not critical in the final evaluation.
Since disc motion as discussed above caused discrepancies in the normal
calibration technique, independent weight testing was done to verify the
calibration accuracy. This testing was done to verify the accuracy of the
calibration factor k. Results are shown in Figure 48. These data also
constitute the post-environmental test for the probe, discussed below.
The factory stated flow range for the instrument is 3.66 to 36.58 m/sec, so the
larger errors for equivalent velocities less than 3.048 m/sec. should not be
-126-
-------
100
z
CO
o
>no
I
ro
i
o
o
FACTORY CALIBRATION
TEST DATA POINTS
FLAG ON LEFT - VELOCITY INCREASING
FLAG ON RIGHT - VELOCITY DECREASING
I
I
I
I I I I
I
I
I
I I I I I I
.01 .1 1
OUTPUT VOLTAGE mv
INPUT VOLTAGE ' V
Figure 45. Ramapo Mark V calibration: Free stream velocity versus ratio of
output voltage to input voltage
10
-------
CO
i
0
o
-1
-2
I
5 /
' '
10
15 20
VELOCITY, METERS/SEC
25
J
30
V V
INCREASING DECEASING
_<>
DATE
6-20
6-20
AVERAGE
Figure 46. Difference between factory calibration and test velocity in
percent versus test velocity for Ramapo Mark V
-------
oo
LOW SPEED:
DISC ALIGNED WITH
STATIC PORTS ON PITOT
PROBE - DIFFERENCE IN
INDICATED VELOCITY
SMALL
oo
HIGH SPEED:
DISC DISPLACED BY
LARGER DRAG FORCE -
NOT ALIGNED WITH
PITOT PROBE - DISC
"SEES" HIGHER WIND TUNNEL
BLOCKAGE AND INDICATES
HIGHER VELOCITY THAN
PITOT PROBE
Figure 47. Diagrammatic explanation of Ramapo output at high speed
in wind tunnel testing
-129-
-------
+4,-
VELOCITY ERROR, %
+2
CO
o
I
-1
-2
-3
-4
O A
EQUIVALENT VELOCITY, METERS/SECOND
J I I
1Q
15
CD
0
-n-
O
A
O
20
25
30
35
CD Q)
O
A
V INCREASING
V DECREASING
Figure 48. Results of force calibration of Ramapo Mark V as percent difference
in factory and test velocity versus equivalent test velocity
-------
regarded as significant. Figure 48 shows clearly acceptable data. The
weights were placed and removed after each reading, so a standard test
for hysteresis was not performed. Hysteresis type effects can be observed
for the lower loadings, however, where the force factor appeared higher
for decreasing loadings than for increasing loadings. This observation
reinforces the previous indications that a small amount of hysteresis
is present.
The drag coefficient of the disc and exposed portion of the lever
arm were calculated to verify the factory supplied value of the assembly
based only on disc area of CD = 1.275. We have for a moment balance
(see Figure 49):
AD *D = CD AD *D + CD AA *A (62)
or
AA £A
AD £D (63)
For the applicable Reynolds number range, "Fluid Dynamic Drag,"
by S. F. Hoerner(Reference 7), published by the author, gives
V1-17
C -,.0
Physical measurements of the instrument give
AA = 6.6 cm2
AD = 33.13 cm2
-13]-
-------
STRAIN
GAGE
BRIDGE
^—'
dp = DISC DIAMETER
dA = ARM DIAMETER
hA = EXPOSED ARM LENGTH
Up = DISC MOMENT ARM
t. = ARM MOMENT ARM
FT
*
I.
I
\
«
*[
A
MOMENT ABOUT
STRAIN
" FD *D + Fa "'A
= I P. U!
WHERE: AQ = } (dp)2
AA = hA dA
Figure 49. Moment applied at strain gauge bridge of Ramapo Mark V
-132-
-------
So we have
Cn =1.17+1.0
UF
= 1.17 + .10
= 1.27
which agrees with the factory experimental data within the accuracy of
the calculation.
7-3-3 Hastings Raydist Flare Gas Flow Probe Calibration
Test calibration and factory calibration data are shown for the
Hastings probe in Figure 50. Accuracy is shown in Figure 51. The solid
line in Figure 51 represents the factory stated accuracy, which is based
on a tolerance of +0.1 volts on the output voltage. The accuracies of all
data points shown are referenced to the experimental data curve fit in
Figure 50. when an additional 1% tolerance is added for the accuracy of
the reference measurement system, all but four of the 104 test data
points are within tolerance, which is very consistent with a 20 accuracy
for the instrument (i.e., 95% or more of the data points fall within the
specified tolerance). It is clear that the factory data does not match
the tolerance below 15 m/sec. The factory was consulted several times
and extra test data were obtained to try to resolve the discrepancies,
which are best illustrated in Figure 52. Here are shown typical factory
calibration curves for these probes: models IK, 6K, and 10K, for maximum
speeds of 304.8 m/sec, 1828.8 m/sec, and 3048 m/sec, respectively. The test
probe was a 10K. Consultation with the factory revealed that a 10K
probe is identical from a hardware standpoint to a 6K probe, the difference
being an electrical span change. The figure shows that the 6K output is
always less than the IK output for a given speed. Considering the physical
similarity of a 6K probe and a 10K probe, it would certainly be feasible
that the 10K output would be correspondingly less than the 6K output
for a given speed. Experimental data confirm this hypothesis, but the
-133-
-------
OUTPUT VOLTAGE, VOLTS
5
U)
^
I
1.5
FACTORY DATA
• DATA POINTS
CURVE FIT
10
15
20
5-7
5-8
5-9
5-15
6-7
6-7
6-8
CURVE FIT
FLAG ON LEFT--V INCREASING
FLAG ON RIGHT--V DECREASING
25
30
Standard Velocity, v$, METERS/SEC
Figure 50. Factory and test calibration curves for Hastings-Raydist AFI-10K
probe as probe output voltage versus standard velocity
-------
00
en
36
32
28
24
20
16
12
V - V
ref
x 100%
"ref
TEST DATA POINTS
FACTORY DATA POINTS
FACTORY STATED ACCURACY
15 20
STANDARD VELOCITY, v_. METERS/SEC
35
Figure 51. Absolute difference in percent between velocity data points and reference
curve fit velocity versus standard velocity for Hastings probe
-------
VOLTAGE OUTPUT, VOLTS
CO
en
i
2 -
1
AFI-6K
STANDARD VELOCITY, vg, METERS/SEC
Figure 52. Comparison of calibration curves for three Hastings probes and
test data for AFI-10K probe
15
-------
factory 10K curve is significantly above the 6K curve at low velocities,
which leads to the possible conclusion that the factory 10K curve is in
error. The factory did agree that the 10K curve looked "unusual" at the
low end. The factory also suggested that the problem may have been an
electrical shift during shipment. Such shifts have been noted by TRW
personnel in items such as pressure transducers.
The Hastings probe tested is not suitable for accurate measurements
regardless of the test discrepancies. This unsuitability is shown by the
factory-claimed accuracy as shown in Figure 42, and confirmed by the
scatter of the experimental data. The loss of accuracy above 15 m/sec
is due to the flattening of the output voltage curve. The factory has
indicated development of a system with a more accurate (velocity)
linearized output, but literature on such a system has not been received
as of this writing.
7.3.4 Thermo Systems Hot Film Sensor Calibration
Test calibrations were performed using the test sensors purchased
for the program and TSI anemometers already owned by TRW. Calibration
data were put in the form which would have been obtained if a VT161 type
anemometer had been used.
Typical calibration curves are shown in Figures 53 to 55. Since
only part of a system was purchased, there are no factory calibration
curves. Figures on accuracy were obtained by curve fitting the test
data points and then determining the accuracy of each point relative to
the curve fit. Results are shown in Table 21. The metal clad sensor
clearly showed the best repeatability, and results for all three sensors
are better than the factory claimed accuracy of 5% for the linearized
system.
7.3.5 Ellison Annubar Calibration
Since the Annubar is inherently an averaging instrument, calibration
in a uniform flow is not a true measure of accuracy. Consequently,
calibration testing could not include determination of accuracy, and had
to be limited to repeatability, shift, hysteresis, etc. The calibration
-137-
-------
.3 r- SENSOR POWER SQUARED,
P2, WATTS2
LO
00
I
JO
15 20
VELOCITY, METERS/SEC
25
30
35
Figure 53. Calibration of TSI metal clad sensor as power squared versus velocity
-------
.04
SENSOR POWER SQUARED,
P2, WATTS2
.03
.02
00
UD
I
SYMBOL
D
O
A
DATE
7-25
7-26
7-27
.01
_L
_L
_L
_L
25
30
5 10 15 20
VELOCITY, METERS/SEC
Figure 54. Calibration of TSI metal backed sensor as power squared versus velocity
35
-------
.04 r SENSOR POWER SQUARED,
P2, WATTS2
.03
.02
o
I
.01
SYMBOL
D
O
A
DATE
7-25
7-26
7-27
1
1
1 1 1 1 1
1Q
15
25
30
35
Figure 55,
VELOCITY, METERS/SEC
Calibration of TSI wedge sensor as power squared versus velocity
-------
Table 21. TSI HOT FILM SENSOR CALIBRATION SUMMARY
V - V
Error = E =
ref
V
x 100%
ref
V = test velocity
V f = reference velocity from curve fit
a =
n = total number of data points
SENSOR
METAL CLAD
METAL BACKED
WEDGE
max
%
2.7
@ V = 3.66 m/sec
5.2
@ V = 3.66 m/sec
9.4
a V = 3.66 m/sec
at
%
0.6
1.6
2.1
2ot
%
1.2
3.2
4.2
tAdjusted to compensate for ±1% uncertainty in reference value.
-141-
-------
consisted of obtaining an effective calibration factor for the instrument
in a uniform flow, as was done with the "S" pi tot probe:
kAl'coV (3D
Ap. = Annubar reading
k^ = Annubar calibration factor
2" p<» ^«2 = dynamic pressure
Test results are shown in Figure 56. Repeatability is shown to be good,
and the pitot factor was reasonably constant for data 6 m/sec or above.
7.4 STABILITY
This test was performed with each test sensor and a reference pitot
probe mounted in the wind tunnel. The pitot probe output from the
Baratron pressure transducer was fed as a voltage to a Hewlett Packard
digital voltmeter and printer. Outputs from each of the test sensors
were handled in the same manner. The wind tunnel was turned on, and
then the printers were turned on at a known sampling rate. These runs were
made for each test sensor at nominal velocities of 6, 17, and 30 m/sec.
Duration was approximately fifteen minutes for each run. Fifteen groups
of fifteen data points were then selected at random for each sensor for
evaluation for each run. Each group of fifteen data points was averaged
and the velocity computed to give fifteen pairs of velocity per run.
The ratios of the test sensor velocity calculation to pitot probe
velocith calculation for the fifteen points were normalized so that the
instrument stability could be evaluated. A typical data reduction sheet
is shown in Figure 57. Instrument stability is shown by the variation in
normalized velocity N. Results are shown in Table 22 for the sensors
tested. Results are considered adequate for all probes. Total system
accuracy is not high enough to allow quantitative comment on individual
readings. All instruments should be considered to have 1% or better
short term stability over the range tested.
-142-
-------
1.60 ^
CALIBRATION FACTOR,
1.58
1.54
1.50
1.46
i
CO
' 1.42
1.38
1.34
1 30
u 0
k - ....
A 1 .. 2
7 P v
f. 00 00 -£
A- -**- ^o - 5.8 m/sec
, . - ' CALIBRATION RUN
(kAj - 1 .531
AVE O 1
MAXIMUM DEVIATIONS IN A 2
TERMS OF VELOCITY:
V 3
-A +1.46%
-1-94% TAG ON LEFT— V INCREASING
4_ TAG ON RIGHT — V DECREASING
-11,1,,
5 10 15 20 25 30
VELOCITY, METERS/SEC
Figure 56. Annubar calibration factor
-------
Figure 57. Ramapo probe stability data reduction sheet
6/4/73 Stability Data Reduction
Ramapo
Speed Control 20
Sample
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Ra
.2238
.2247
.2233
.2253
.2247
.2239
.2251
.2237
.2238
.2227
.2233
.2240
.2237
.2253
.2236
Pa
4.2779
4.2695
4.2877
4.2653
4.2778
4.2824
4.2686
4.2727
4.2754
4.2631
4.2774
4.2879
4.2627
4.2736
4.2748
Pa
.2287
.2294
.2282
.2298
.2292
.2287
.2296
.2268
.2268
.2286
.2285
.2286
.2291
.2296
.2287
N
1.0000
1.0030
.9977 <- Min >
1.0048 «- Max /
1.0020
.9997
1.0040
1.0004
1.0003
.9993
.9989
.9993
1.0016
1.0038
.9999
Total
Variation
.71%
Total run time: 13.8 min
-144-
-------
Table 22. STABILITY TEST RESULTS
trt
PROBE
"S" PITOT
PROBE
ANNUBAR
HASTINGS
RAMAPO
TSI SENSORS
METAL CLAD
METAL
BACKED
WEDGE
TOTAL VARIATION DURING RUN, %
V = 6.10 m/sec
CO
.76
.85
1.54
1.08
0.4
0.4
0.8
V =16.76m/sec
oo
.31
.57
.48
.71
0.5
1-9*
*
4.2
Vm =29.87 m/sec
.22
.47
.62
.77
*
4.0
*
1.3
RUN TIME,
MINUTES
19, 17, 8
13, 12, 13
15, 14, 14
15, 14, 14
22, 19, 19
15, 17, 12
16, 23, -
Jf
Shifts due to uncompensated temperature variation in freestream.
Variation noted for 6.10 m/sec wpuld be representative of variation at
higher speed for standard complete factory unit
-------
7.5 TIME RESPONSE
The setup for each probe was the same as for the stability test. A
typical run proceeded as follows: the wind tunnel was started with the
test probe aligned with the flow. The printer was turned on. The probe
was turned an angle of 90° to the flow. It was then quickly turned back
to its normal position (nominal time required = .5 seconds) and the
printer ran until the reading stabilized at its normal value. This
procedure applied to all probes except the hot film sensors, for which
the response time was determined by covering the sensors and then suddenly
exposing them to the flow. The response time was then calculated from
the known printer speed for all sensors. Results are shown in Table 23.
All probes except the Hastings-Raydtst may be considered to have a one
second or better response time. The Hastings-Raydist time response
acceptability must be determined in individual applications. It may not
be a drawback for long term continuous monitoring, particularly if flow
conditions are steady.
7.6 SEN S FT IV ITY TO OR IOTAT K)N
It was shown in Sectton 4 that the component of velocity parallel to
the duct axis is the most mathematically desirable output for a point
velocity sensor to have. This component can be obtained by aligning
the probe directly with the local velocity vector and computing the
axial component from the measured total velocity and the angle between
the probe and the duct axis. This approach is highly impractical under
most circumstances. A much easier approach is to always align the probe
with respect to the duct axis, and have a probe which responds to the
axial velocity component. The purpose of this test was to determine
how well the probes being evaluated responded to this component of
the flow.
For each case, the probe was rotated with respect to the flow,
rotation being performed about the probe's principal axis since it
was not feasible to rotate the flow itself. Probes such as the S probe
have two independent axes, as shown in Figure 58, but rotation about the
tilt axis would have required wind tunnel modifications beyond the scope
of the program. Runs were made for each sensor at nominal speeds of
-146-
-------
Table 23. RESPONSE TIME TEST RESULTS
PROBE
"S" PITOT PROBE
100% RESPONSE
ANNUBAR
100% RESPONSE
HASTINGS
90% RESPONSE
100% RESPONSE
RAMAPO
100% RESPONSE
TSI PROBES
METAL CLAD
METAL BACKED
WEDGE
RESPONSE TIME, SECONDS
V =36.58 m/sec
CD
1.0
0.5
20.8
56.9
1.3
0.2
0.8
0.4
V =16.76 m/sec
oo
0.7
0.2
12.3
24.1
0.6
^0
0.6
0
V =29.87 m/sec
oo
0.8
0.2
8.5
18.1
0.4
^0
0.2
0.1
Time response for change from V ^ 0
-147-
-------
6, AXIAL ROTATION
ANGLE
\.
\
4, JILT ANGLE
Figure 58. S probe orientation angles
-148-
-------
6, 17, and 30 m/sec. The resulting velocity measurements were
normalized with respect to the measurement at 9 = 0° and plotted.
Results are shown in Figures 59-65. As a reference, cos 0 is also
plotted. A summary of results is given in Table 24. Results indicate
that all probes should be acceptable for use in nonuniform flows with
flow angularities not over 10° in the plane tested.
7.7 ENVIRONMENTAL TESTING
7.7.1 Test Sequence
Environmental testing consisted of three phases:
1. Four hour exposure in duct at 180°C and 10 m/sec, ambient
humidity.
2. Four hour exposure in duct at no°C and 16 m/sec, humidity
100 percent + droplet laden.
3. Exposure to 30kg of quartz sand at 40 m/sec, approximately
equivalent to four days of exposure in a stream at 16 m/sec
with a grain loading of Q.Q5 gm/s,cm.
None of the hot film probes was exposed to the sand due to the large
particle size (average size«300 microns), and one of the hot film probes
was not exposed to the high temperature stream due to temperature limitations
of the sensor supports. Particulate loading was accomplished through use
of a portable sandblaster in a setup outside the process simulator,
since local air pollution control officials have ruled against the use of
particulate material in the process simulator.
7.7.2 Test Results
Each of the probes showed physical corrosion and abrasion, but not at
a high enough level to impair performance except for the hot film sensors.
Corrosion and abrasion over a period of months or years could possibly
affect the performance of all instruments tested. The Hastings and
Ellison instruments showed no calibration change after environmental
testing. The Ramapo probe showed approximately an 80 percent drop in
output. The Ramapo factory was consulted on the problem, and said the
probe was probably clogged internally. The probe was dismantled in
-749-
-------
= velocity @ 0=0C
l.OC
1.04
1.02
1.00
.98
.96
.94
.92
.90
.88
.86
• 6 m/sec
17 m/sec
— n 30 m/sec
.84
-50 -40 -30 -20 -10 0 10 20 30 40
9. DEGREES
Figure 59. S pitot probe orientation sensitivity data
-150-
50
-------
1.02
1.00
VQ = VELOCITY AT 6 = 0°
.78 -
-40
V = 6 m/sec
V =17 m/sec
oo
V =30 m/sec D
-30 -20
Figure 60.
10
20
30 40
-10 0
6, DEGREES
Ramapo probe orientation sensitivity data
50
-151-
-------
VQ = VELOCITY AT e = 0
8 = FLOW ANGULARITY
1.00 -
8, DEGREES
Figure 61. Hastings-Raydist probe orientation sensitivity data
-152-
-------
1.01
1.00
.98
.96
.94
.92
.90
V/Vr
.86
.84
.82
.80
V = 6 m/sec
00
V =17 m/sec
00
V = 30 m/sec
,00-CK
I
-30
-20
-10
20
0 10
9 , DEGREES
Figure 62. Annubar orientation sensitivity data
30
40
-153-
-------
1.00
.96
.92
V = VELOCITY AT
6, $ = 0°
.84
.80
.76
.72
.68
.64
.60
-60
CASE A
SENSOR
NORMALIZED SENSOR
OUTPUT. CASE B
NORMALIZED SENSOR
OUTPUT, CASE A
cose,
COScp
CASE A DATA FROM VIKING HOT FILM SENSOR--GEOMETRICALLY
SIMILAR TO METAL CLAD SENSOR
_L
•O
CASE B
J
-50 -40 -30
-20 -10 0 10
6, <(,, DEGREES
20
30
40
50 60
Figure 63. TSI metal clad sensor orientation sensitivity as
normalized velocity versus orientation angle
-154-
-------
1.00
.98
.96
.94
.92
.90
.88
.86
.84
.82
.80
Vn = VELOCITY AT
0 6=0°
OO
O V^ = 6 m/sec
A V = 17 m/sec
oo
17 V =30 m/sec
-40 -30
30 40
Figure 64. TSI metal backed sensor orientation sensitivity as
normalized velocity versus orientation angle
-155-
-------
.96 -
.94 -
.92 -
.90 -
.88
O O
O V = 6 m/sec
CO
O A v = 17 m/sec
IB
V V = 30 m/sec
O
-30 -20
-10 0
6, DEGREES
10 20 30
40
Figure 65. TSI wedge sensor orientation sensitivity as
normalized velocity versus orientation angle
-156-
-------
Table 24. ORIENTATION SENSITIVITY TEST RESULTS
PROBE
"S" PITOT
PROBE
ANNUBAR
HASTINGS
RAMAPO
TSI SENSORS
METAL CLAD
AXIAL (e)
TILT M
METAL
BACKED
WEDGE
ANGULAR RANGES FOR STATED
MEASUREMENT ACCURACY
ACCURACY:
IX
X
±8°
±8°
±6°
+17°
+8°
+2°
+2°
4%
±14°
±40°
±17°
±12°
+43°
+16°
+ 24°
+8°
10%
±28°
±40°
±26°
±18°
>+60°
+25°
+ 32°
+16°
ANGULAR VALUES ARE FOR AXIAL ROTATION ANGLE 6
UNLESS OTHERWISE NOTED
PROBE AXIS
INTO PAGE
FLOW INTO
PAGE
-157-
-------
accordance with factory instructions and approximately one gram of sand
was removed. The probe was then put back together and functioned
normally. The conclusion is that the Ramapo probe definitely requires
an adequate purge system when used in particle-laden flows. Both the
Annubar and "S" pitot probe had a net accumulation of particulate. In
the Annubar, which was horizontal, an equilibrium point was reached
where the probe became self purging and performance was not affected.
The "S" pitot probe would generally require an occasional purge to prevent
clogging if used for long periods of time.
Results of the post-environmental calibration are shown in Table 25.
The Hastings data, also shown in Figure 66, showed two points slightly
out of spec when a 1% uncertainty in the reference velocity is taken
into consideration. The shifts were in the direction of the original
factory calibration, and so are considered not critical, especially in
view of the small magnitude. The Ramapo post-environmental calibration
was the weight calibration described above and shown in Figure 48. It
is completely acceptable. Recall from the last page that the Ramapo
probe clogged during sand testing, and had to be cleaned before it
could be used again. The factory makes purged probes, as illustrated
in Figure 67, and a probe such as that should have been ordered originally.
No difficulties would be expected with a purged system. The "S" pitot
probe showed no change in calibration factor, and the Annubar pitot
factor change was within normal repeatability limits.
Results for the TSI sensors are shown in Figures 68 to 70. The
test sequence was as follows:
1. Each probe was calibrated before environmental testing.
2. Environmental testing took place.
3. The probes were recalibrated.
4. The probes were cleaned in a dilute nitric acid solution and
a third calibration was performed.
Examination of the curves shows that for each case the power required
to maintain overheat went down for the post-environmental tests. It would
appear that in each case the exposure to high temperature, humidity
-158-
-------
Table 25. POST ENVIRONMENTAL TEST CALIBRATION RESULTS
INSTRUMENT
HASTINGS
AFI-10K
RAMAPO
MARK V
ELLISON
ANNUBAR
"S" PITOT
TSI SENSORS
RESULTS
90% OF DATA POINTS WITHIN SPECIFICATION.
OTHER POINTS NOT MORE THAN 0.6% OUT OF
SPEC (FIGURE 22).
ALL POINTS ABOVE 3.56 m/sec IN SPEC AFTER
PROBE CLEANED. AMER
% CHANGE IN VELOCITY FROM CHANGE IN
PITOT FACTOR
0.7
0.0
UNACCEPTABLE SHIFTS (FIGURES 24 TO 26).
-159-
-------
en
o
i
32
28
24
20
16
12
V - V
ref
x 100%
'ref
O POST ENVIRONMENTAL TEST DATA POINTS
FACTORY STATED ACCURACY
O
O O
o
o o
o
o
o
*-o
0 0
o 8
1 1 . 1 1
10
15
20
25
30
STANDARD VELOCITY, Vs, METERS/SEC
Figure 66.
Accuracy of Hastings AFI-10K as percent difference between
test data points and calibration curve fit versus velocity
35
-------
RAiMAPO INSTRLJMtlNT COMPANY, 11VC.
BLOOMINGDALE NEW JERSEY
JUNCTION BOX %" SIZE
P.N. VTC-2203 OR EQUAL
TERMINAL STRIP ENCLOSED
PURGE
CONNECTION"
'/." TUBE, . .
(FLARELESSll
*SHOWN 90° TO ACTUAL LOCATION
PIPE
8" • 150 ASA
.BLIND FLANGE
BOLT HOLES STRADDLE
C_PIPE
MODEL:
LINE SIZE:
DIM. "A":
FLOW,
Figure 67. Ramapo purged probe
-161-
-------
I
PO
POWER2, WATTS2
O PRE-ENVIORNMENTAL
POST-ENVIRONMENTAL
POST-ENVIRONMENTAL
AFTER CLEANING
28% SHIFT
62% SHIFT
10
15
20
25
30
35
Figure 68.
VELOCITY, METERS/SEC
Pre- and post-environmental calibrations of TSI metal clas sensor
as sensor power squares versus sensitivity
-------
en
CO
i
POWER2, WATTS2
,04
.03
.01
O PRE-ENVIRONMENTAL
A POST-ENVIRONMENTAL
D POST-ENVIRONMENTAL
AFTER CLEANING
80% SHIFT
10
15
20
25
30
35
VELOCITY, METERS/SEC
Figure 69.
Pre- and post-environmental calibrations of TSI metal backed sensor
as sensor power squared versus velocity
-------
.04
POWER2, WATTS2
O PRE-ENVIRONMENTAL
A POST-ENVIRONMENTAL
D POST-ENVIRONMENTAL
AFTER CLEANING
,03
21% SHIFT
31% SHIFT
30
35
VELOCITY, METERS/SEC
Figure 70. Pre- and post-environmental calibration of TSI wedge sensor
as sensor power squared versus velocity
-------
and the amounts of dust present in the process simulator resulted in the
buildup of a film on the sensors, the film acting as insulation so that
less heat transfer to the air took place. Cleaning slightly alleviated
the problem for the wedge sensor, but worsened it for the other two
sensors. Shifts were very small at the lower velocities, which is clearly
not acceptable. Also, the cleaning method used was not adequate, but
even proper methods would have to be used several times a day to keep
the sensors clean. Unless an automatic cleaning system could be devised,
the frequency of maintenance would be unacceptable. It definitely appears
that additional work is needed to make hot film sensors acceptable for
continuous use in stack environments.
7.8 LABORATORY TEST SUMMARY AND FINAL EVALUATION
7.8.1 S Pitot Probe
Calibration test results are summarized in Table 26. Recall that
the laboratory calibration factor disagreed with the manufacturer's
calibration factory by 3.5%. The manufacturer, Ellison Instruments,
stated verbally that the same calibration factor is published for all
S type probes -- they are not calibrated individually. The discrepancy
noted is therefore probably due to errors in the original calibration
and/or variations among the manufactured probes.
The following conclusions about probe performance were drawn from
the accumulated test data, and necessarily apply in the strict sense
only to the probe tested, since not all S probes are alike. It would,
however, be non-conservative to assume that other S probes have better
accuracy characteristics than the probe tested, without having data to
support the assumption. The primary conclusion about the S probe is that
it should be capable of providing point measurement accuracies on the
order of 10% or better for the flow conditions specified in Section 4.
Nominal expected accuracy should be about 5% for a properly operated probe.
Some specific drawbacks related to probe design are as follows:
A. The probe design is not standard.
B. The probe is highly sensitive to flow angularities.
C. Due to the design, the probe is probably subject to significant
Reynolds number and turbulence effects.
-165-
-------
Table 26. ELLISON INSTRUMENTS COMBINED REVERSED ("S") PITOT
PROBE CALIBRATION RESULTS
Manufacturer's
Stated Accuracy*
Manufacturer's
Stated Range
Calibration Test Results
Overall Accuracyt
Test Range
Probe Range Based
on Test Results
±5% of reading
Not specified
±5% of reading
2..74 to 3Q.98 m/sec
2.74 to 45.72 + m/sec
5% accuracy figure includes accuracy, repeatability, short
term zero and full scale shift, and hysteresis. Repeatability,
shift, and hysteresis errors negligible.
Time Response:
Short Term Stability:
Accuracy of Axial
Velocity Component
Measurement:
1 sec or better
Better than 1%
Accuracy, %:
±1 ± 4 ±10
Maximum Angularity, X 14 28
Degrees:
t
Verbal communication—not stated in literature.
Also depends upon accuracy of readout equipment.
-166-
-------
Non-standardization of design leads to variation in many parameters
such as tube curvature, spacing, whether the tubes are connected near
the end, etc. The result is that uniformity of the calibration factor
cannot be assured unless great care is taken in the manufacturing
process to insure repeatability of all critical design parameters, and
it is highly questionable whether all of these parameters are known.
The downstream tube senses a wake pressure, since a detached wake is
formed behind the two tubes. Wake pressure measurement has traditionally
been a difficult fluid mechanical problem in uniform streams, and so
must be considered even more difficult in general industrial applications.
Wake pressures behind many bluff bodies are highly subject to free stream
turbulence, which is responsible for adding energy to the wake and
changing the pressure distribution around the body, and to the Reynolds
number based on the size of the body, since boundary layer characteristics
and consequently the pressure distribution around the body, are Reynolds
number dependent. This dependence is noted in Reference 3, "The
Measurement of for Flow", pn p. 39. inhere it is, stated,
"... such instruments are unfortunately not to be
recommended, because they possess certain undesirable
features. In the first place, experience shows that
for most of those so far proposed the value of k varies
appreciably with Reynolds number, particularly in the
lower ranges for which the instrument is primarily
intended. There is no need to stress the inconven-
ience of a calibration that changes with speed, one
result of which is that at least two calculations
are needed to convert a pressure observation into
velocity: the first has to be made with a guessed
value of k, since the speed is initially unknown;
and the approximate value of the speed thus obtained
has to be used to obtain, from the calibration curve
of the instrument, a nearer approach to the true value
of k to be used in the second, a more exact calculation.
-167-
-------
"Another serious disadvantage of such instruments
is that if k varies appreciably with Reynolds number,
its value at a given Reynolds number is likely to
depend much more than that of a standard pitot-static
tube on the amount of turbulence present in the
airstream, so that the calibration may be appreciably
different, even at the same speed, in different pipe
systems."
The calibration shift which occurred near 6 m/sec should be attributed
to Reynolds number effects. Turbulence effects could not be investi-
gated with the laminar flow wind tunnel facility used during the test.
Since probe to probe design variations are difficult to control
due to ambiguity about the relative importance of the numerous parameters
involved, it is evident that for accurate work, probes must be calibrated
individually over the expected flow range as a function of Reynolds
number based on a standard probe dimension. If the probe must be used
in angular flows, the calibration becomes much more difficult, however.
The problem is that the probe's orientation sensitivity makes such a
calibration highly impractical, because three independent variables
would be involved: Reynolds number and two orientation angles as shown
in Figure 58. Investigation of probe response as a function of the tilt
angle has not been performed, but until it is, there is no reason to assume
that the response would be better than for the investigated axial rotation.
Any calibration of this nature would be prohibitively expensive to
perform.
On the basis of existing data, it must be concluded that use
of the S probe in angular flows will result in systematically high
readings due to the probe's angular response characteristics, which were
worse than those of the other probes tested except for the wedge hot film
sensor. This error could be reduced through use of an artifically low
pitot factor. By coincidence, the factory supplied calibration factor
does in fact produce a low pitot factor which was used during the first
field demonstration.
-168-
-------
As a result of investigation of the S probe, it is being recommended
that a standard pitot static probe be used wherever possible to perform
standard velocity traverses. To justify this, pitot-static probe charac-
teristics, which were not investigated during the program since adequate
characterization has been performed and documented elsewhere, are
presented at the end of this task, in Section 7.8.6.
7.8.2 Ramapo Mark V Drag Meter
This was without question the most accurate point sensor tested. The
one significant problem discovered was susceptibility to clogging by
particulate, as explained in the Environmental Test section. This prob-
lem has apparently since been solved through the introduction of the
Mark VI flowmeter, shown in Figures 71 and 72, with no loss of accuracy.
The instrument is bidirectional, and calibration tests showed it to be
free of Reynolds number effects except in highly angular flows. The
main reason that this instrument is free of such affects while the S
probe is not is that the target disc is flat with relatively sharp edges,
which means that boundary layer separation always occur precisely at the
edge of the disc, and so boundary layer characteristics do not influence
performance in most cases. Page 3-15 in Reference 7 shows that the
drag coefficient is constant above Re £ 3 x 10. For the unit tested,
a velocity of about 3 m/s corresponded to Re £ 1.2 x 104.
The instrument also has attractive calibration features for indus-
trial use. The factory manual gives procedures for resistance calibration,
requiring only an additional decade box, and it is also possible to
calibrate the unit with actual weights, as described in the calibration
section. These techniques eliminate the need for calibration in an
actual air stream.It should be noted that the factory recommends that
the probe be turned 180° from the way ft was tested in this program
for normal use. The orientation used for testing was selected so that
wind tunnel data could be directly correlated with weight testing, and
weight testing could be performed most easily in the direction chosen.
-169-
-------
Xj>^wK_>O<>DOO<->O<>
^^ysXaK^Xa&f^OQwgr^ -. — —— ..
^^^^gggggg^^&^g: ._-_ . ..-
MACOPIN ROAD. BLOOMINGDALE, NEW JERSEY 07403/201 - 838-2300 /TELEX: 13-8892
MARK VI FLOW METER
Developed specifically for the precise measurement of flow in dirty
liquids, slurries and high-melt or crystalizing fluids.
The Ramapo Mark VI Flow Meter incorporates the
basic features of the popular Mark V target type meter
in a flow meter which has no moving parts and no
cavities to trap particles and cause erroneous read-
ings. The Mark VI uses a bonded strain gage bridge to
translate the fluid forces on the target into an electrical
signal proportional to flow rate squared. It has a useful
and practical 10 to 1 range ratio and can be calibrated
for accuracies to 0.5%. Uni-directional or bi-directional
flows of almost any liquid or gas can be measured from
1 GPM to unlimited ranges.
FEATURES:
• No cavities or pressure ports in the flow stream
• No rotating parts, diaphragms, or bearing surfaces
to wear or require servicing and maintenance
• The Mark VI probe mounts directly into a tee,
or thru a branch fabricated on an existing line
• The Mark VI can be mounted in any position
without changing calibration
• Installation i* simple and inexpensive
• Mark VI probes are lightweight and can be easily
installed by one man. (12 kg. for typical 24" unit)
• Range change can be done by simply
replacing the target
* Usable with liquids, gases, slurries, high melting
point liquids, food and dairy products at temperatures
ranging from -200°C to +400"C
• Field calibration is simple;
no flow checks are required
ACCESSORIES AVAILABLE
A complete line of transmitters, analog or digital indi-
cators, recorders, controllers and totalizers can be pro-
vided for most system requirements.
PRINCIPLE OF OPERATION
The Mark V Flow meter measures flow in terms of dy-
namic forces acting on a fixed body in the flow stream:
Force = C«Ai Vf
29
Cj = Drag Coefficient, A = Sensor Area, f = Fluid Den-
sity, V = Velocity. Bonded strain gages in a bridge cir-
cuit outside the fluid stream and shielded by stainless
steel, translate this force into an electrical output pro-
portional to the flow rate squared.
Figure 71. Ramapo Mark VI
-170-
-------
ELECTRICAL SPECIFICATIONS
Input: 5 to 10 volts A.C. or D.C.
Output: 2.0 mv/v or 20 mv maximum at full scale pro-
portional to velocity squared; higher or lower outputs
may be used to meet the requirements of specific
applications
0-10 VDC; 1-5, 4-20, 10-50 made outputs available. See
Bulletin SG-1A
Bridge Resistance: 350 ohm standard (120 ohm optional)
Electrical Connections: Barrier type terminal strip in
junction box with %" conduit connection
GENERAL SPECIFICATIONS
Repeatability: ±0.1% of reading
Accuracy: ±0.5% of full scale How rate with flow cali-
bration
Pressure: to 5000 psi (35.000 KM/m2)—limited by flange
rating
Fluid Temperature: -65 to +350°F (-54° to +177-C) to
-320«F (-195°C) and 4-750-F (400°C) available
Sires: %" to 60"—larger si2es available
Ranges: Any 10101 flow range (maximum recommended
lull scale velocity—15 feet/second water flow equiva-
lenl) [minimum recommended full scale velocity—3
feet/second water How equivalent]
Materials: 17-4 PH and Type 304 stainless element with
carbon steel (large and line housing standard; Type
316 stainless, Hastelloy C, and Inconel available.
V* CONDUIT SIZE
-FLOW-
Norn.
Pipe
Size
In.
tt
1
IVi
m
2
3
4
6
8
to
18
20
and
larger
Flow Ranges >
Minimum
CPU
.8 - 6
1.5 -15
2 - 20
3 - 30
5 - 50
10 - 100
20 - 200
30 - 300
dm'/tec.
.05- .5
.10- 1.0
.13-1.3
.19-1.9
.30 - 3.0
.63 - 6.3
1.3-13
I 1.9-19
Minimum Recommended
Velocity Range
.3-3fl./sec.
(.1-1 m./sec.)
Maximum
GPM
4- 40
8- 80
10-100
12-120
20 - 200
30-300
60-600
120-1200
dm' /tec.
.25 - 2.5
.50 • 5.0
.63-8.3
.76-7.6
1.3-13
1.9-19
3.8-38
7.6-76
Maximum Recommended
Velocity Range
1.5-15 ft, /sec.
(.45-4.5 m./sec.)
Full Seal*
Prest. Lost '
ptl
15
10
8
6
3
2
1
0.5
.4
and
lower
kN/m>
105
70
56
42
21
14
7
4
3
and
lower
f »
If.
o ™.H
Su.cn
V,
1
1V4
1%
2
3
4
4
4
3
Dim. <
A
3Vi
3V2
3V<
4
4Vi
5V4
6
6
1 Customer may select any 10 to 1 range desired between the minimum and maximum ranges shown.
Range in SCFM of air at standard conditions may be estimated at approximately 4 x water flow rate In GPM.
1 Approximate values—lower pressure loss available.
3 Mounting flange size may be varied to suit the requirements of specific applications. ANSI ratings as required.
' Plus gasket thickness.
Figure 72, Ramapo Mark VI specification sheet
-171-
-------
Final test results are summarized in Table 27. The factory modifi-
cation since completion of the test work eliminates the only serious
fault of the Mark V unit.
7.8.3 'Hastings/Raydist Flare Gas Flow Probe
Unfortunately for this probe, the factory accuracy specifications
speak very clearly for themselves when stated in the right way. The
factory tolerance of +. .1 volt on the instrument output, which was for
the most part confirmed by the laboratory tests, translates to a very
unacceptable velocity accuracy, as was shown in Figure 46. There is no
doubt that the instrument has excellent survival capabilities, but the
same can now be said of the Ramapo Mark VI, which is much more accurate
than the Hastings unit.
Conversations with the factory have revealed the intention to
introduce a model with improved electronics, but literature on such a
model has not been received as of this writing. One very noteworthy
Hastings accomplishment has been made since the end of the laboratory
testing, and that is the integrated "suitcase" system shown in Figures 74
and 75. As will be discussed later in the report, no true measurement
system was tested -- the velocity sensors all required significant
amounts of auxiliary support equipment. This integrated approach used
by Hastings is to be commended and should be more widely adopted by
the instrumentation industry.
Results are summarized in Table 28. The model tested was clearly
acceptable from a survivability standpoint, but just as clearly unaccep-
table from an accuracy standpoint.
7.8.4 Thermo-Systems Hot Film Sensors
TRW Systems Group is responsible for the Viking Meteorological
Instrument System which will be taking wind speed and direction measure-
ments on Mars starting July 4, 1976. The system uses hot film sensors
for the measurements. On the basis of that work, it was hoped that
hot film sensor technology would be applicable to this program, since
specifications required instrument survivability in high dust loadings
up to 150 m/sec. Unfortunately, the sensors tested, which reasonably
represent state of the art in terms of ruggedness, failed the
-172-
-------
Table 27. RAMAPO MARK V FLOW METER CALIBRATION RESULTS
Manufacturer's ,~ £ ..
Stated Accuracy ±1% of readin9
Manufacturer's -j cc <- ^c co /
Stated Range 3'66 to 36'58 m/sec
Calibration Test Results
Overall Accuracy* ±1% of reading
Test Range 2.74 to 30.48 m/sec
Probe Range Based 3.66 to 36.58 m/sec
on Test Results
1% accuracy figure includes accuracy, repeatability, short
term zero and full scale shift, and hysteresis. Hysteresis
approximately ±0.5% of reading.
Time Response: ^1 sec or better •
Short Term Stability: Better than 1%
Accuracy of Axial Accuracy, %: ±1 ±4 ±10
Velocity Component
Measurement: Maximum Angularity, +6 ±12 ±18
Degrees:
Also depends upon accuracy of readout equipment.
-173-
-------
-WTELEDYNE
HASTINGS-RAYDIST
Specification Sheet No. 516
HAMPTON. VIRGINIA 23661
(804) 723-6b31
HASTINGS STACK GAS VELOCITY METER
MODEL GSM-1D5K FOR MEASURING VELOCITY OF WET AND DIRTY GASES
FEATURES:
• VELOCITY RANGE: 90-1500 FPM
• PORTABLE ELECTRO-PNEUMATIC VELOCITY
INDICATOR
• CONTINUOUS PURGE PREVENTS ENTRY OF
STACK GAS INTO THE PITOT TUBE AND
INSTRUMENT
• OPERATES WITH A "V" OR "S" TYPE PITOT
TUBE
• PORTABLE UNIT SIMPLIFIES SET-UP FOR
SPOT CHECKS OR LONG-TERM SAMPLING
• SELF-CONTAINED PURGE MANIFOLD AND
PURGE SOURCE
• NO AFTER TEST CLEANING REQUIRED
GENERAL
The Hastings Stack Gas Velocity Meter is the result of
nearly two decades of experience in dealing with difficult-
to-measure. corrosive, wet or dirty gases. A patented
thermal principle with continuous purging is utilized so
that line gas is prevented from entering the pitot tube.
The Hastings Stack Gas Velocity Meter is a portable
test set which can easily be carried to test sites for
periodic checks, or can be left in operation for long-
term monitoring of stack velocities. The instrument
includes solid-state circuitry and a sensitive transducer
for measuring the differential pressure across the pitot
tube, which is related to velocity. An internal pump
supplies a continuous purge of air through the pitot
tube and into the line to prevent line gas from entering
the instrument, and a connection is provided for using
some other purge gas if air is not suitable. A manifold
is also included for ease in balancing the purge. A rugged
meter provides on-site read-out of velocity indications,
and a 0-1 volt d-c output is available to connect to remote
data logging devices, meters, recorders, etc.
A pitot tube with a 36" insertion depth is included with
the instrument and can be interchanged with pitot tubes
already in use in stacks or gas sampling trains.
AIR-/ ^-EXTERNAL
INLET PURGE
CONNECTION
FLEXIBLE TUBING
S'TYPE PITOT TUBE
STACK GAS VELOCITY METER
BLOCK DIAGRAM
Figure 73. Hastings stack meter
-174-
-------
PRINCIPLE OF OPERATION-
CONTINUOUS PURGE MODE
Purge gas is injected into a pneumatic bridge arrange-
ment formed by the velocity transducer, manifold, and
pitot tube. At zero line velocity the bridge is balanced so
that no flow occurs through the velocity transducer and
purge gas exhausts equally through both openings of the
pitot tube.
As flow across the tip occurs, a differential pressure
is developed, unbalancing the bridge and causing a small
amount of purge gas to flow through the transducer. The
transducer measures the flow which is related to the main
gas flow at the tip of the pitot tube. Purge gas still ex-
hausts through both openings, but at slightly unequal
rates.
The purge gas continually exhausts into the main line,
thereby preventing contamination in the main line pitot
tube.
CALIBRATION
Calibration is related to the gas density and the velocity
profile in the stack.
Velocity = kj.k2.Vind where
kj = Velocity Profile Factor; (typically .8 when probe
is inserted to center of stack)
k2 = Density Factor: V075/8as density (Ib/tt1)'
Vind » Velocity from probe calibration curve
Example: What is the full scale (1 volt) range of a GSM-
1D5K when measuring a stack gas having a density of
.092 Ibs/cu. ft.?
Velocity <•= kj..k2.Vind
* (.8) A/.075/.092 ' (1500)
= 1083 FPM
100
D
5
z
HI ^.
u«0
20
O SCO 1000 1500
AM VELOCITY (IfM)
PURGEGASES
The Hastings Stack Gas Velocity Me'ter has a ouilt-in
pump which pulls ambient air in through a filter to use
.as a purge gas. A gauge indicates the pressure at the inlet
to the purge section, and an adjustable "bleed" valve is
used to adjust this pressure. A three-way valve is supplied
to change the purge source from the internal pump to an
external source so purge gases other than ambient air
can be used. A 1/4" NPT fitting is mounted on the front
panel for connecting to an external source. The flow rate
of purge gas is normally 5-10 cfh at 15 psi.
SPECIFICATIONS
INSTRUMENT
Modal: GSM - 1D5K
Rang*: 50-1500 fpm for air
Power: 115 volts. 50-60 Hz. 1.5 amp
Output: 0-1 volt d-C
Cate Dlmentlont: 9" high x 16" wide x 10" deep
Weight: 27 pounds
PITOT TUBE
Type: S
Length: 36* Insertion
Connecting Tubing: Twin Type. 10 ft. Length
FOR CONTINUOUS MONITORING OR
PERMANENT INSTALLATIONS:
HASTINGS GAS FLOW PROBE
MODEL AFI-SERIES
RANGE: 0-1000 fpm or 0-6000 fpm
A dependable, non-clogging flowmeter for contaminated
gas lines.
REQUEST CATALOG 0913A
FEATURES
• CONTINUOUS PURGE PRINCIPLE
• NO EXPOSED SENSORS or WIRES
•EASILY INSTALLED or MOVED
•EXPLOSION-PROOF TYPE HOUSING
• 05 VOLT DC OUTPUT SIGNAL
• PROVIDES LONG LINE TRANSMISSION
CAPABILITY
• REMOTE: RECORDING, CONTROL, ALARM,
INDICATION
•PURGE WITH AIR, N, OR PROCESS GAS
• CHOICE OF TWO RANGES: 0 1000 or 0-6000 fpm
Litinlun tnllibli upon rtqimt:
Hutinp Vicuum Giuf n
Hittinp Mcleod Giu(i
Histinp Giu|t totx tccetsorin
Hntinp Vicuum Giuf e ftetertnct lubn
Hiitinfi Air-Mtten
Hnlinp Mm f Imrimteri for Giiei
Hlllinp Cllibrited Gil Lukl
Cltllot No. 300
Sp
-------
Table 28. HASTINGS-RAYDIST AFI-10K GAS FLOW PROBE CALIBRATION RESULTS
Manufacturer's
Stated Accuracy
Manufacturer's
Stated Range
Calibration Test Results
Overall Accuracy
Test Range
Probe Range Based
on Test Results
See Figure 42--accuracy not better
than 4.8% of reading
0 to 50.6 m/sec
Test data set met factory
tolerance but did not agree
fully with factory calibration
2.74 to 30.48 m/sec
0 to 50.6 m/sec
Accuracy determination includes accuracy, repeatability, short
term zero and full scale shift, and hysteresis. Error due
mainly to repeatability error.
Time Response:
Short Term Stability:
Accuracy of Axial
Velocity Component
Measurement:
57 sec @ V =6 m/sec
24 sec 0 V =17 m/sec
18 sec @ V =30 m/sec
Accuracy, %:
±1 ± 4 ±10
Maximum Angularity, ±8 ±17 ±26
Degrees:
-176-
-------
environmental tests on this program, as shown by the severe calibration
shifts in Figures 68-70. Tfi.e culprit, which did not exist in the Martian
specifications, was high moisture content at high temperature. This
resulted in deposition of a film on the heat transfer surfaces. The
film acted as a thermal insulator, reducing the power required to
maintain overheat at a given wind speed. Efforts used to clean the
probes succeeded only in worsening the condition. Even had the cleaning
been successful, the sensors would not be acceptable for continuous
use due to the need for very frequent cleaning^
Despite the sensors' unacceptability for continuous monitoring, it
is felt that system development as a diagnostic tool should be encouraged.
Thermo-Systems presently offers a total vector anemometer which is
capable of measuring the true velocity vector -- both speed and direction
regardless of orientation. This instrument is the Model 1080 as shown
in Figures 75 and 76 . Frequent cleaning requirements would not be
detrimental for diagnostic applications. Possible applications would
include investigation of blower or other process unit characteristics,
stack swirl properties, etc. to determine possible malfunctions or
general process effectiveness.
A general problem which applies to all hot film sensors should
also be considered, and that is the previously mentioned dependence on
thermal conductivity and viscosity. In general for hot film sensors,
the calculated velocity is proportional to viscosity and inversely
proportional to the square of thermal conductivity. The National
Bureau of Standards is rather reluctant to commit to uncertainties of
less than 3% and 5% for viscosity and thermal conductivity, respectively,
of gases such as air. This means that significant uncertainties must
be expected in the calculated velocity if the measurements are taken in
a gas significantly different from the gas in which the sensors were
calibrated. Thus to insure accuracy, heat transfer devices such as hot
film sensors should be calibrated in the same gas composition in which
they are to be used.
-177-
-------
tsT) SYSTEM 1080 TOTAL VECTOR ANEMOMETER
PROBE TIP DETAIL
Streamlined Supports
\
3 Mutually PerpendicularSplit-Film Sinton
Built-in remotely op«r«ted
shield and calibrator
in each probe
Spatial Resolution: lest
than 0.3" dia. sphere
GENERAL DESCRIPTION
The system 1080 measures the velocity magnitude and direction over the full 360° solid angle in three dimensional
flow fields. It features solid-state electronic circuitry and a small, rugged transducer which allows fast response, wide
dynamic range, excellent spatial resolution and high accuracy in both velocity magnitude and direction. The velocity
probe feature; Thermo-Systems' split-film* sensors which allows the unambiguous determination of magnitude and
direction of the instantaneous velocity vector. Six simultaneous velocity dependent analog voltages and a 0-5 volt
analog temperature signal comprise the system output. The probe includes a pneumatically operated shield with a
built-in calibration feature.
'Patented
PERFORMANCE SPECIFICATIONS
VELOCITY MEASUREMENT:
USABLE RANGE: 0 - 500 fpc (Air, 1.0 ATM)
STANDARD RANGE: 0 - 300 fpt (Air, 1.0 ATM!
0- 100 mps
MAGNITUDE ACCURACY: + 3% of Reading and
+ 0.1% of full scale.O - 150 fps.
DIRECTION ACCURACY: Two independent
directional angles are within 3° over
the complete solid angle I 4 7T
Stand land
RESPONSE: DC - 1 KHz
SPATIAL RESOLUTION: Three orthogonal
sensors fit within 0.3 inch (7.6mm)
diameter sphere.
TEMPERATURE MEASUREMENT:
RANGE: 0°F to 200°F (20°C to 100°C)
ACCURACY: + 2°F (+ 1°C)
TIME CONSTANT: 20 millisecond! at 60 fpt
(Air, 1.0 ATM)
• 64(162) -»
f t t
* t
fl «•
«»•
0 ••
*:
r
1134)
1
» 16 3BI3B1I «
CONTROL CIRCUIT
DATA OUTPUT
Fooooooooou-
1 ooooooooogfc
000060005Jo
SERIES 1080 PROBE
(Dimensions in Brackets an in Millimeters!
Figure 75. TSI total vector anemometer
-178-
-------
SYSTEM 1080 TOTAL VECTOR ANEMOMETER
SPECIFICATIONS
OUTPUT:
Velocity: There are six analog voltage signals available at the
rear output connector as well as the front panel pin jacks.
These voltages constitute the data necessary to calculate the
instantaneous 3D velocity vector. Output range is 0-20 volts
DC, output impedance is 50 ohms. Output signals can be
monitored on an oscilloscope or voltmeter. Data storage can
be via tape recorder or oscillograph. (A 0-5VDC output range
is optional.)
Temperature: A copper-constantan thermocouple - 0.002"
(.05mm) dia.-is mounted on the probe tip. Automatic cold
junction compensation is provided. The output voltage of
0-5V DC corresponds to a temperature range of 0°F to
200°F. Output impedance is 25 OHMS. The temperature
output signal is available at the rear output connector as well
as at a front pin jack.
PROBE SHIELD: A pneumatically actuated shield is standard
for all systems. The shield switch on the front panel can be used
to actuate an external air source (not supplied) to power the
pneumatic cylinder on the probe. A gas supply of 15 psig or
greater is required.
GENERAL:
Weight:
Control Circuit 10 Ibs.
Probe 3 Ibs.
Cable (15 foot) 2 Ibs.
Power Required: 115V/220V AC ± 15%. 50-400 Hz, 1.0
amperes.
Circuitry: All solid state
Size: See drawing, opposite side.
Cabinetry: Bench type, etched aluminum panels, vinyl-clad
aluminum chassis.
Probe Construction: All stainless steel anodized aluminum
materials.
Environmental: Control circuit will operate as specified over
40°F to 120°F, (5°C to 50°C). Probe is built to operate over
temperature extremes of -50°F to 200°F f-45°C to 100°C).
Cable length: Standard length is 15 ft. (4.6m). Cable lengths
up to 50 ft. can be provided. For longer cables, please
consult TSI (Model 10117-15).
CALIBRATION FEATURES: Probe shields have a built-in
orifice to provide velocity calibration. The "calibrate" switch on
the front panel can be used to actuate the calibration pressure
source. The reference pressure can be changed to provide
multiple velocity calibrations.
CONTROLS: Three balance controls are provided on the front
panel of the control circuit. These are used to insure that the
split-film sensors operate at the same temperature.
i
SENSOR ARRAY: Three mutually perpendicular single-ended
sensor rods constitute the velocity vector sensing element.
Significant parameters of the sensors are:
Diameter = 0.006 in. (.15mm)
Sensitive length = 0.08 in. (2mm)
Total length » 0.2 in. (5.1mm)
Substrate - quartz
Conducting film - platinum
Protective coating - quartz*
STANDARD EQUIPMENT SUPPLIED: Each 1080 system is
furnished with a 15' probe cable, one probe with pneumatic
shield actuator and built-in orifice, power supply, control
circuit, bench type cabinet, 0-5V DC analog temperature
output, probe calibration constants, power cord and output
connector, instruction manual and data reduction procedure
manual.
PNEUMATIC SYSTEM: Thermo-Systems' Series 1126 Calibra-
tion systems are used to operate the pneumatic shield and
calibrator. The Model 1126 consists of a filter/trap, pressure
regulator, pressure gauge and solenoid valves which are actuated
by the"CALIBRATE" and "SHIELD" switches located on the
control circuit front panel. Consult TSI for the Muriel 1126 best
suited for your needs.
"Patented
Figure 76. TSI total vector anemometer specifications
-179-
-------
Final results are summarized in Table 29. The sensors were found to
be unacceptable for use in this program due to fouling in a stack environment,
7.8.5 Ellison Annubar
As was previously mentioned, the Annubar was the only averaging
instrument evaluated during the program. As a result of this, it was
the only test sensor directly evaluated during the mapping tests. All
of the point sensors could be, and were, evaluated for mapping purposes
through substitution of a pi tot probe. Throughout the mapping test
phase of the program, the Annubar was found to be comparable in accuracy
to an array of up to eight point sensors, and the hardware cost impli-
cations of this fact are obvious. The mapping tests were in fact
calibrations for both the Annubar and the point sensor arrays. For
the circular duct case, it was a matter of comparing actual results with
predicted results for both systems. For the rectangular case, it was a
matter of producing new calibration data for each type of system, since
none existed previously. This was done for the point sensor arrays by
choosing a desired output and obtaining "calibration" data on the
required array location to achieve that output. For the Annubar, cali-
bration was performed at a fixed location by determining correction
factors based on the instrument output. Success for the point sensor
array was determined by locational repeatability, and for the Annubar
by correction factor repeatability. Each system was found to perform
best in a rectangular duct when the flow was reasonably two dimensional,
such is immediately downstream of an elbow. Test results justified
the attention given to both the Annubar and point sensor arrays.
Other laboratory testing revealed no significant drawbacks for the
Annubar, although field testing did show a significant problem related
to blockage of the rear static orifice, which is discussed in Section 8,
which did not show up in the laboratory tests. Minor errors due to
the static orifice also showed up during the mapping test as discussed
in Section 6, where it was noted that the static orifice "overreacts"
to high or low pressure regions along the duct axis. This feature is
inherent in the probe design.
-180-
-------
Table 29. THERMO SYSTEMS HOT FILM SENSOR CALIBRATION RESULTS
Manufacturer's Stated Accuracy
Manufacturer's Stated Range
Calibration Test Results
Overall Accuracy
Metal Clad Sensor
Metal Backed Sensor
Wedge Sensor
Test Range
Systems Accuracy Based on
Test Results
±3% of reading and ±0.1% of full
scale for nonlinear system;
±5% of reading for linear system
0 to 60.96 m/sec-htgher and lower
ranges available
System Range Based on
Test Results
Manufacturer's Specifications
Time Response:
Short Term Stability:
Accuracy of Axial Velocity
Component Measurement
±1.2%
±3.2%
±4.2%
3.05 to 36.58 m/sec
Meets manufacturer's specs for
metal clad sensor
Meets manufacturer's specs
Given for System Using Metal Clad Sensor
Less than 1 sec
Better than 1%
Accuracy, % ±1 ± 4 ±10
Maximum Angularity
Metal Clad
x-y plane ±17° ±43°
x-z plane ±8° ±16°
Metal Backed ±2° ±24°
Wedge ±2° ±8°
>±60°
±25°
±32°
±16°
-181-
-------
The position of the static orifice makes the Annubar calibration
factor more dependent on the local Reynolds number based on the Annubar
diameter and the free stream velocity ahead of the statis orifice than
on any other parameter. Ellison came to this realization during a
research program performed during the past year, and has revised the
instrument calibration factor on the basis of tests performed in various
facilities in Colorado and elsewhere. The new calibration data applies
only to circular pipes, and was used as reference data for the circular
duct mapping tests and the reference section measurements. Other results
of the research program were a small design change in the static port
to improve instrument-to-instrument repeatability, and development of an
equation to predict Annubar output in any known flow field (i.e. pre-
dictions from mapping data).
Test results are summarized in Table 30. The Annubar has demonstrated
adequate accuracy in both circular and rectangular ducts, although greater
calibration factor variations must be expected for the latter case. Since
the instrument is relatively inexpensive, it would appear that for most
applications a measurement system incorporating the Annubar as the flow
sensor would be less expensive than a point sensor array designed to achieve
a comparable accuracy.
7.8.6 Pitot-Static Probe
This probe was not evaluated during the program since its properties
are rather well known. It is being discussed here primarily to allow
comparison to the S probe, which has been shown to have several undesirable
characteristics. All data in this section was taken from Chapter III of
Reference 3, "The Measurement of Air Flow," unless otherwise stated.
Common pi tot static tubes are generally of two types — either
hemispherical nosed or ellipsoidal nosed, which is descriptive of the
shape near the tip of the probe. There are standard designs for both
shapes, and the designs are simple enough to insure repeatable instrument
to instrument performance. Careful design has led to instruments which
are free of Reynolds number effects for speeds above about 2 m/sec for a
.61 cm diameter probe.
-182-
-------
Table 30. ANNUBAR CALIBRATION RESULTS
Manufacturer's Stated Accuracy
Manufacturer's Stated Range
Calibration Test Results
Overall Accuracy
Duct Mapping Accuracy
Repeatability, shift, hysteresis
in uniform stream
Test Range
Probe Range Based on Test Results
Time Response
Short Term Stability
Accuracy of Axial Velocity
Component Measurement
Circular duct—varies with size:
2.4% for 180" duct
Rectangular duct--no claims for
accuracy
MIN--not stated
MAX--varies—at least 61 m/sec
Determination out of program scope
±7.2% for runs performed
-0.5%
2.7 to 30.5 m/sec
5.8 61 + m/sec depending on size
Better than 1 sec
Better than 1%
Accuracy, %: .
±1 ± 4 ±10
Maximum Angularity, +8 +4Q +4Q
Degrees: ~
-183-
-------
The pitot-static probe is effectively axisymmetric, so that it
responds in the same manner to all flow angularities at a constant angle
to the probe axis, regardless of the actual orientation. As mentioned
previously, the angular response of an S probe is dependent upon two
independent angles, which makes its response more complex. As the yaw
angle of a pitot-static probe increases from zero, both the sensed
impact and static pressures decrease. However, the static pressure
initially decreases faster, so that the differential pressure increases.
At higher yaw angles, the impact pressure begins to drop faster, and the
differential pressure decreases. The same phenomenon occurred during
laboratory testing of the S probe, but in a much more radical manner.
Pitot-static probe yaw characteristics are shown in Table 31. Since
for the current type of application the desired output is V cos e (=u),
the ellipsoidal probe is to be preferred. It would be expected that
the hemispherical probe in particular would tend to give systematically
high output in angular flows. Recall that this was observed during the
1974 mapping tests, although the errors were only around 2%. The
discovery of a preference for ellipsoidal probes was made late in the
program or they would have been used rather than hemispherical probes.
Table 31 shows that both probes more closely follow V than V cos 0.
This was rather intentional from a design standpoint since pitot-static
probes have traditionally been used mainly in wind tunnel and aircraft
applications where the flow is well directed and the desired output is
the total flow speed, rather than a component of it. The flat response
with respect to V is then desirable since it minimizes errors due to
probe misalignment.
Since highly angular flow was occasionally encountered during the
program, it was necessary to extend the calibration data for yaw to
much larger angles. This was performed with one of the test probes, and
the results revealed that response continued to drop when the angle was
increased above 30°. The output went negative around 60°, reached a
negative peak at 90°, and became positive again around 120°. The
symmetry of the curve around 90° led to a decision to assume u = 0
(i.e. V cos e = 0) any time the probe output was negative. Since cos e
is at its minimum absolute value, zero, at 90°, it was felt that any
errors incurred would be minimal. Data are shown in Figure 77.
-184-
-------
Table 31. YAW CHARACTERISTICS OF PITOT-STATIC PROBES
AXIS
V = magnitude of local velocity
e = yaw angle
V cose = axial velocity component
e
degrees
0
5
10
15
20
25
30
Error with Respect
to V cos e, %
El lipsoidal
Nose
0
+ .6
+2.0
+3.3
+2.7
+1.7
-
Hemispherical
Nose
0
+ .7
+2.7
+5.7
+8.4
+10.5
+13.2
Error with Respect
to V, %
Ellipsoidal
Nose
0
+ .2
+ .4
-.2
-3.5
-7.8
-
Hemispherical
Nose
0
+ .4
+1.2
+2.1
+1.8
+ .2
-2.0
-185-
-------
.a
.6-
.4-
Ap = differential
pressure @ Q
APO = differential
pressure @ 9 = 0
C = +1 if Ap >0
C = -1 if Ap > 0
-.4
HEMISPHERICAL
NOSED PITOT-
STATIC PROBE
\
\ COS 9
\
\
\
\
\
\
-1.0
2C
40
60
CO
100
120
140 160
0, DEGREES
Figure 77. Approximate pitot-static probe response
at large yaw angles
180
-186-
-------
It should be noted that the data were obtained in the mapping test
facility, due to unavailability of the laminar flow wind tunnel.
Consequently the accuracy is not as good as would have been attained
in a wind tunnel.
There should be no question that a pitot static probe is preferable
to an S probe from an accuracy standpoint. The S probe is more widely
used for stack sampling since it is easier to handle (i.e. no "hook"
on the end to make insertion and removal difficult), is less subject
to clogging, and gives a higher output than a pitot-static probe.
Electronic pressure transducers are now sufficiently accurate and reliable
to adequately measure the differential pressure from either type of
probe, so that the higher S probe output does not compensate for the
instrument's lack of accuracy. Traverse data with a pitot-static probe
was obtained during the field test with no clogging problems, as is
discussed in Section 8, and the extra care needed to avoid damage during
insertion and removal did not result in significant delays. On the
basis of the accumulated data, it is being concluded that data taken with
an ellipsoidal nosed pitot-static probe will probably be 3 to 4 per cent
more accurate than data taken with an S pitot probe. Consequently the
pitot-static probe should be used for normal manual traversing except
when fouling or handling problems require use of the S type probe. It
must be noted that the expected difference of 3 - 4% does not explain
the 10% difference observed during the field demonstration and discussed
in Section 8. This higher error may be due to flow angularities in the
"tilt" direction.
As a further result of flow angularity investigations, it is being
recommended that a pitot-static probe be developed whose angular
response directly follows the axial velocity component. This could help
to significantly improve system accuracy in angular flows.
7.8.7 General Laboratory Test Summary
Final laboratory results are given in Table 32. The Ramapo Fluid
Drag Meter was definitely the most accurate point sensor. The Annubar
was found acceptable as an averaging instrument for use in both circular
and rectangular ducts, the latter solely as a result of calibration
-187-
-------
Table 32. LABORATORY TEST SUMMARY
ACCEPTABLE AVERAGING SENSOR
SENSOR
Ellison Annubar
ATTRIBUTES
Reasonable accuracy and wide
applicability; good
survivability
LIMITATIONS
In place calibration desirable;
static port may need purge
00
oo
ACCEPTABLE POINT SENSOR
SENSOR
Ramapo Mark VI
ATTRIBUTES
Very good accuracy and
survivability
LIMITATIONS
High cost per channel
-------
Table 32 (continued). LABORATORY TEST SUMMARY
I
CXI
CONDITIONALLY ACCEPTABLE POINT SENSORS
SENSOR
S Probe
Hastings
Raydist
AFI-10K
ATTRIBUTES
Inexpensive, widely used
Very good survivabllity
LIMITATIONS
Relatively poor accuracy
Poor accuracy
UNACCEPTABLE POINT SENSORS
SENSOR
Thermo Systems
VT 161
ATTRIBUTES
Linear output,
simple operation
LIMITATIONS
Poor survivability
-------
work performed during this program. Both the S probe and the Hastings-
Raydist probe were found environmentally acceptable but exhibit un-
desirable accuracy potential, especially the Hastings probe. The Thermo
Systems sensors were found unacceptable due to fouling problems. As
a sidelight of the test, a clear preference for the pitot-static probe
over the S probe was indicated for reference traverse work. As a result
of the laboratory tests, the Ramapo unit and the Annubar were recommended
for field test evaluation, with reference data to be supplied by a pitot-
static probe. The test probes were to be used in accordance with
techniques developed in Task III.
7.4.9 Pre-Field Test Demonstration
The purpose of this demonstration was to verify in the laboratory the
acceptability of the sensors and measurement techniques to be demonstrated
in the field. Environmental acceptability of the Annubar and Ramapo probe
had been demonstrated, so further evaluation of this type was not performed.
The field site selected was the Moapa station of the Nevada Power Company,
Moapa, Nevada, described in Section 8. The mapping facility inlet was
configured to represent expected field conditions as closely as possible,
and the normal test data were taken. The demonstration runs were runs
45-48, using the straight rectangular inlet. Average results for the
four runs showed an Annubar calibration factor of .5765, and a location
for the row averaging technique of 20.8% of the duct width from one side
wall. It was immediately recognized that the Annubar calibration factor
was significantly lower than the normal .65 value, but this was expected
since the geometry of the duct work indicated clearly that a high velocity
region was expected in the center of the duct, which would result in a
high Annubar output and consequently a lower than normal calibration
factor. It was decided that the Ramapo probe would be attached to a
reciprocating traversal unit to obtain the row data for that measurement
technique.
-190-
-------
SECTION VIII
TASK V - FIELD DEMONSTRATION
8.1 GENERAL
Two field demonstrations were performed at the Reid-Gardner Station
of the Nevada Power Company, located at Moapa, Nevada. The plant is about
72 km. northeast of Las Vegas. Each test lasted approximately 18 days.
The first was begun in September, 1974, and the second was begun in
February, 1975.
The primary purpose of the first test was to demonstrate hardware and
techniques for volumetric flow measurement which had been successfully
tested in the laboratory. The primary purpose of the second test was to
demonstrate techniques for gas composition measurement. All composition
data other than that required for flow calculations are to be discussed in
a separate report.
8.2 FACILITY DESCRIPTION
The plant presently consists of two Foster-Wheeler 120 megawatt boilers,
with a third under construction. All work was performed on the #2 unit,
shown schematically in Figure 78. Flow from the boiler is separated into
two streams which pass through Lunjstrom rotary air preheaters. The ducts
at the two preheater outlets go through a shape transition and then rejoin
upon entering a mechanical dust collector. A row of test ports is located
on each duct just ahead of the dust collector. This is shown in Figure 79.
All rectangular duct mapping was done at these locations.
When the plant was first constructed, the dust collector exhausted
directly into the stack. Subsequently, a venturi scrubber and separator
were added by Combustion Equipment Associates. When the scrubber is on,
the flue gas is diverted at the dust collector exit and routed to the
scrubber, where it is processed and fed into the stacks. If the scrubber
is not on, the flue gas goes directly into the stack after leaving the
dust collector. The inlet to the stack is about 13 m above ground level.
Sample ports are located in the stack at the 31.4 m level, or about four
stack diameters above the inlet. All circular duct testing was performed at
this location in the stack.
-191-
-------
PO
I
Plane
A-A
Scrubber
r
Flow routed to and from
scrubber at plane A-A
Stack
Location 2
Preheaters
Dust
Collector
V
Location
I
Boiler
A = Annubar
A
Duct 1 Duct 2
Location 1
Location 2
Figure 78. Schematic of MOAPA power plant
-------
FLOU
DIRECTION
SAMPLE PORTS (ROW NUMBER)
ANNUBAR LOCATION
R = PORT FOR RAMAPO
1 PROBE - PROBE KEPT AT R-4
SAMPLE POINTS
1
TRAVERSE MAP -
49 POINTS
SAMPLING
PLANE
Figure 79. Field demonstration rectangular duct geometry
-193-
-------
8.3 TEST CONDUCT
8.3.1 Test Hardware
Flow instrumentation for the 1974 test included two Annubars, one for
stack measurements and one for measurements in the left hand duct after the
preheaters (Duct 1). The Ramapo Fluid Drag meter which was evaluated in
the laboratory was modified to increase its length and installed in an
automatic traversing mechanism (Figure 80). Thermocouple probes for use
with each of the three test instruments were fabricated to provide temper-
ature data.
Reference flow data were provided by a hemispherical nosed pitot static
probe with an integral thermocouple. This probe was used for reference
traverses in Duct 1. The S type probe evaluated in the laboratory was
modified to increase its length and was used for both velocity traverses
and gas sample traverses. Bulk composition data were provided by a Carle
thermal conductivity detector gas chromatograph which measured concentrations
of N2, 02, C02 and CO.
All pressure sensors were monitored by the same Baratron pressure
transducers used in the laboratory. Instrument outputs were routed through
a scanner to a digital voltmeter and then to a paper tape printer. The gas
chromatograph had its own separate output system. Flow data output inter-
vals varied from ten seconds to one hour. Data reduction was accomplished
by use of a digital computer after the field work was completed. The two
Annubars were left in place upon completion of the test.
Hardware for the 1975 test included the two Annubars left in place
•previously plus a new third Annubar installed in Duct 2. The Ramapo probe
was not used due to malfunction of the traversing mechanism during the
first test. The type S probe was also not used. Reference pitot traverse
data for the 1975 test were obtained in Duct 1, Duct 2, and in the stack.
The Carle gas chromatograph was replaced by a newer model which allowed for
automatic sampling, whereas the older model required manual operation.
1975 hardware also included a purge system to keep the rear orifice on
each Annubar clean. Purging was done for five minutes every half hour.
-194-
-------
PULLEYS
MOTOR
THREADED ROD
Operation: Drag meter is attached
to nut inside square tube. Nut rides
on threaded rod which is fixed to
square tube. When motor is turned
on, nut travels down until it trips
bottom switch, which reverses motor.
Nut travels up until it trips top
switch, again reversing motor.
Cycle continues
ELECTRICAL LEADS
_1 TOP SWITCH
n
"NUT
TRAVERSE TUBE
BOTTOM SWITCH
DUCT WALL
RAMAPO DRAG METER
FLOW
Figure 80. Schematic of traversing mechanism for Ramapo drag meter
-195-
-------
8.3.2 Extent of Testing
The Annubars were monitored continuously unless the pressure transducers
were required to support manual traverses. The duct Annubars were removed
for pitot or gas sample traverses in the ports^where the Annubars were
installed. The Ramapo probe was monitored continuously during the first
test. Complete gas sample traverses were performed at both full and half
load in Duct 1 during the first test. Complete gas sample traverses were
performed at full load in Duct 1 and in Duct 2 during the second test.
Reference pi tot-static traverses were performed as often as the work load
allowed. Twenty-two were performed in the Duct area during both tests,
and fifteen were performed in the stack during the second test.
A total of four basic plant conditions occurred during the first test,
each about equally often. The unit load was held at maximum (~110-120 MW)
during the PM hours and at about half that during the AM hours. The
scrubber was on about 40% of the time. The unit was down about five days
during the test. Full load was maintained during the second test, the
variation being from about 100 to 123 MW. The scrubber was on full time
except for one five hour period. Stack traverses were obtained during
this period (February 28). There was no down time during the second test.
8.3.3 Problems
The automatic traversing mechanism used with the Ramapo probe failed
early during the first test due to the combined effects of particulate
clogging around the mating seal and a too-powerful motor used to drive the
mechanism. The probe was removed from the unit and installed at a fixed
location, except during manual row traverses toward the end of the test.
The Ramapo traverse data had to be discarded due to internally shorted leads.
The leads were modified when the probe was lengthened for the test, and
the insulation around the junctions failed due to heat. These difficulties
were clearly the responsibility of test personnel, and do not reflect
adversely on the instrument itself nor on the manufacturer. No clogging
problems were experienced with the Ramapo probe.
Laboratory testing showed the ability of the Annubar to "self-purge"
the impact holes. This was confirmed during the field test. A new problem
did show up, however, and that was the effect of particulate on the Annubar
-196-
-------
rear orifice. Data during the first test showed that this orifice clogged
repeatedly when the scrubber was operating. Since the particulate loading
in this operation mode was relatively low, it was apparent that the high
moisture content caused the remaining particulate to become much more
adhesive. Purging through the pressure line alleviated the blockage of the
stack Annubar. Accumulation of considerable particulate on the rear side
of the Duct 1 Annubar was observed when the probe was removed during pi tot
traverses. Although complete clogging was never observed, the accumulated
particulate definitely altered the shape of the Annubar's rear orifice.
This showed up as a significant variation in calibration factor during the
first test for both Annubars. The problem was effectively solved during
the second test by addition of the automatic purge system.
The only new problem encountered during the second test was with
installation of the Duct 2 Annubar. The duct itself was deformed in the
immediately area of the test port in which the instrument was placed. This
resulted in the top of the instrument having about an 8° offset in the
vertical plane (the duct axis was horizontal), facing upstream. As the test
got underway, it was immediately apparent that this Annubar had readings
consistently lower than the Duct 1 Annubar. These two Annubars were
switched after the first week of testing, and no change in reading from
either location was noticed, which meant that the Annubars themselves were
identical. Testing showed a large difference in calibration factors between
these two Annubars. It would be impossible to say how much of the difference
was due to the deformation of the duct without getting rid of the deformation
and recalibrating the probe.
8.4 FLOW DATA CORRELATION
8.4.1 1974 Test
The Duct 1 Annubar was calibrated in place by means of the pi tot tra-
verses performed in the duct. Calibration factors were obtained for the
stack Annubar during Duct 1 traverses when the scrubber was off by assuming
that half the stack input came from Duct 1. This was considered valid since
no additional flow sources or sinks occurred ahead of the stack. Correla-
tion could not be obtained when the scrubber was on due to additional water
vapor and reheat air added during the scrubbing process. Failures with
-197-
-------
the Ramapo system precluded its calibration. The final two traverses were
performed simultaneously with the pitot-static and S probes to obtain a
direct comparison between the two.
Coal usage per day and coal analysis data were coupled with test com-
position measurements of CCL in Duct 1 to make independent flow computations.
The coal data were provided by plant personnel.
8.4.2 1975 Test
The Annubars in Ducts 1 and 2 were calibrated in place by means of
pitot traverses in each duct. The stack Annubar was calibrated by pi tot
traverses in the stack. Plant data were used as during the first test.
8.5 TEST RESULTS
8.5.1 Composition Measurements and PI ant Reference Data
8.5.1.1 1974 Test-
Dry gas composition traverses in Duct 1 at full and half loads showed
the average CCL concentration by volume to be 15.83% and 9.078$, respectively.
Plant coal analysis showed that the average usable carbon content of the
coal during the test was 84.0% by weight. Water content was 5.93%. Total
water content of the wet gas Duct 1 was determined to be 4% at full load
and 2% at half load, respectively, reducing the C02 content of the wet
gas in Duct 1 to 15.20% and 8.81% for those conditions. The total flow
per day could then be calculated as follows:
v = Coal used (moles) X .0224SH3/mo1e ,g4j
Average CO- Concentration
where V = Total volume flow, SCM
8.5.1.2 1975 Test-
Average C02 content of the dry gas was 13.85% at full load (the only
run condition) which showed that more excess air was being fed into the
boiler during the second test. Coal composition data during the test period
were not available at the time of this writing, so January, 1975 data
(month prior to test) were used. The January data showed the coal to be
-198-
-------
82.81% C and 7.47% H20. The average wet gas CCL concentration in Ducts 1
and 2 was 12.99%.
8.5.2 Velocity Traverses
8.5.2.1 1974 Test-
Results are shown in Table 33. The estimated flow through both ducts
from plant data is based on average coal usage at full and half load. The
49 point traverses constituted in place calibration of the Annubars. Duct 1
Annubar calibration factor is seen to have varied from .531 to .620. This
wide range is believed to be due to the particulate problem discussed above.
The calibration factor was much more constant during the 1975 test.
The calibration factor for the stack Annubar was initially higher
than the factory suggested value, then seemed to stabilize somewhat at a
relatively low value. An average value of S = .62 was used for continuous
data reduction. Data from the stack Annubar were generally suspect due to
the clogging problems which occurred during the test. Testing during 1975
showed that an average of 49% of the flue gas went through Duct 1 and 51%
went through Duct 2. The stack calibration factors in Table 33 as shown
were modified on the basis of this new information.
8.5.2.2 1975 Test-
In place calibrations of all three Annubars were conducted, and results
are shown in Tables 34 and 35. Forty-nine point duct traverses were per-
formed on four separate occasions. On each occasion, they were performed
in immediate succession, which resulted in a total variation from the mean
of less than 3% for each group. The calibration factors shown were based
on the average Annubar reading during each group of traverses. Duct 1
results were very consistent for the two groups, and were in good agree-
ment with 1974 results. Duct 2 results were also self-consistent, but
the average calibration factor was 14.8% higher than that for the Duct 1
Annubar. Since the ducts are identical in design, equality between the
calibration factors of the two instruments was somewhat expected. The
difference cannot be explainted, but may have to do with the unavoidable
misalignment of the Duct 2 Annubar, discussed in Section 8.3.
-199-
-------
Table 33. DUCT TRAVERSE SUMMARY, 1974
Date
9-25
9-27
9-28
9-28
10-2
10-2
10-2
10-4
10-4
Load
Full
Full
Half
Half
Half
Half
Full
Full
Changing
Pi tot
Traverse
Duct 1
Flow Rate
SCM/Sec
62.8
62.6
42.0
39.8
43.6
41.1
57.4
56.6
52.8
Estimated
Duct Flow From
Plant Data
SCM/Sec
118.5
118.5
85.0
85.0
85.0
85.0
118.5
118.5
*
Average Duct Calibration Factor
Full Load 0.588
Half Load 0.555
Weighted Average 0.575
Annubar
Calibration
Factor
Ductl Stack
0.620 0.717
0.605 +
0.564 +
0.531 +
0.558 0.632
0.566 0.606
0.571 0.619
0.558 0.621
0.584 +
*No data due to rapidly changing load
Scrubber on-no correlation possible
Factory calibration factor for stack Annubar: .661
-200-
-------
Table 34. DUCT TRAVERSE SUMMARY, 1975
Date
2-24
2-28
2-25
2-28
Duct
1
1
2
2
Traverses
2
4
3
4
Average Flow
Rate, SCM/SEC
61.72
61.63
63.99
58.81
Duct Annubar
Calibration Factor
0.586
0.583
0.674
0.668
Duct 1
Duct 2
Average Annubar
Calibration Factor
0.584
0.6705
Table 35. Stack Traverse Summary, 1975
Date
2-17
2-28
Scrubber
ON
OFF
Traverses
10
5
Average Flow Rate
SCM/SEC
154.4 ± 2.2
130.1 ± 0.8
Scrubber ON
Scrubber OFF
Stack Annubar
Calibration Factor
0.723
0.729
Factory calibration factor for stack Annubar = .661.
-201-
-------
Four test ports were available in the stacks at 90° intervals. The
Annubar was installed through two of them, and twenty point traverses were
performed along the diameter through the other two ports. It would have
been preferable to remove the Annubar and perform traverses along both
diameters. Removal of the stack Annubar was considered to be too hazardous
to perform more often than necessary, since the probe is 5 meters long,
weights about 60 kg, and must be handled on a one meter catwalk.
One set of pressure lines of length 60 meters was available from the
stack, so only one instrument could be monitored at a time. As a result,
Annubar readings immediately before and after the pitot traverse were used
to obtain the calibration factor for the instrument. Results are shown
in Table 35, from which it is apparent that flow conditions were quite
steady during the traversal periods. The Annubar calibration factors were
consistent for both operation modes. It should be noted that the first
calibration in 1974 produced a factor close to the 1975 results, further
suggesting that later degradation during the 1974 test was due to par-
ti cul ate accummulation.
Although the 1975 calibration results showed much better consistency,
it must still be noted that poor agreement was obtained with the suggested
factory value. This can be explained on the basis of Figures 81 and 56.
Figure 81 shows very flat velocity profiles except in the immediate vicinity
of the wall. Factory calibration factors are based on fully developed pipe
flow characteristics, for which a parabolic profile would be expected.
The profiles shown in Figure 81 are almost uniform. Since the calibration
data in Figure 56 were taken in a uniform flow in a wind tunnel, the
stack calibration value would be expected to approach the wind tunnel value.
At comparable speed, Figure 56 shows kA = 1.53. The standard calibration
factor S and the factor kA in Figure 56 are related as follows:
s - _J _ (65)
so that for the case of a uniform stream, we would expect S = .81. Con-
sequently, the test calibration factor appears reasonable. The problem
is that the Annubar should not require an in-place calibration for accurate
measurements. Laboratory 'test results in 1974 over a wide variety of
-202-
-------
Velocity at
16
14
12
10
a
L>
P
ro
CD
00
' 6
4
2
0 .
_ 3 lanuar u i-uiiu I L 1 Uilb , U
M/Sec S
"
0 o 0
- nO 0 0 0 0 ^ 0 0 0 -
0° °
- a D a D a a a o o D a
D D
D
O Scrubber On, u =13.62 m/sec
D Scrubber Off , u = 11.42 m/sec
1 1 1 1 1 1 1 |
O
c
D
D D _
C
~
-
-
-
IU°
40 20 0 20 40 60
Stack Radius, Per Cent From Center
Figure 81. Stack velocity profiles
80
IOC
-------
conditions close to disturbances did show S factors this high on occasion
(Table 5) although the average values were much closer to the factory
prediction. Figure 56 and the field test results conclusively show that
an almost uniform flow such as those in Figure 81 will result in a higher
than normal calibration factor. It is an interesting property of the
instrument that it works well in a variety of stratified flows, but shows
a large calibration shift in a uniform flow. The reason is the design of
the rear orifice, which sense a pressure below local free stream static
pressure. This low pressure is the reason why the instrument has a cali-
bration factor less than one. If the instrument were designed so that the
low pressure orifice sensed the static pressure of the stream, the calibra-
tion factor in a uniform stream would be exactly one, and the calibration
in a developed pipe flow would be very close to one.
8.5.3 Comparison of S Type Probe and Pi tot-Static Probe Traverses
The last two traverses during the 1974 test were performed simul-
taneously with a pitot-static probe and the S type probe evaluated in
the laboratory. The pitot factor used for the S probe was .823, based on
the manufacturer's data, rather than the laboratory test value of .852.
This choice was deliberately made to minimize known errors in angular flow
as shown in Figure 59. For these two traverses, the pitot-static data gave
o
flow values of 56.6 and 52.8 m /sec, as shown in Table 33. Corresponding
results with the S probe were 62.4 and 58.4 m /sec, respectively. The
S probe results were 10.3% and 10.6% higher than the corresponding pitot-
static probe data. Results were obviously consistent for both runs. It
is not known why the S probe data showed such a high shift. Laboratory
data showed that high readings would normally be expected in angular flows,
and this is why a low pitot factor was used. As will be shown below, most
traverse point velocities were around 7-8 m /sec, which is where the S
probe tested shows a calibration shift, as shown in Figure 43, which would
result in abnormally high readings, but only by 2-3%. Since angularity
testing was not performed in the "tilt" direction as shown in Figure 58,
the problem may be due to that characteristic. Duct geometry in Figure 79
suggestes that significant flow angularity may be present in that plane.
-204-
-------
8.5.4 Continuous Flow Monitoring
8.5.4.1 1974 Test-
The pi tot-static traverses discussed above established the Annubar
calibration factors to be used during the test. Results are summarized
in Table 36. For proper instrument operation, the Duct 1 flow would be
half the stack flow. Comparison of Duct 1 Annubar data and plant data
was performed for two days and shown in Table 37. The Annubar data
shown assume that 49% of the flow went through Duct 1, as was determined
for the 1975 test. The C02 concentrations shown were obtained by taking
an average of the full load and half load concentrations times the number
of hours each day at each condition. Good agreement between test and plant
data was obtained.
Table 36 SUMMARY OF ANNUBAR CONTINUOUS MONITORING
DATA WITH SCRUBBER OFF, 1974
Load
MW
115-123
49-57
Average Flow Rate, SCM/SEC
Duct 1 Annubar
62.3 ± 4.1
44.6 ± 3.0
Stack Annubar
138.3 ± 8.2
92.1 ± 6.4
Calibration factors used - Stack : 0.729
Duct 1: 0.575
Load
Ratio of Duct 1 Annubar Average Flow to
Stack Annubar Average Flow
115-123
49-57
0.450 ± 0.046
0.484 ± 0.054
-205-
-------
Table 37. COMPUTATION OF TOTAL FLOW FROM COAL ANALYSIS AND
MEASURED C02 CONCENTRATION, 1974
Date
10-2
10-3
Coal Used
«g
3,251,380
3,571,480
Carbon Used
kg gm-moles
2,752,618 4.729x107
3,023,614 5.195x107
Average C09
% *
11.89
12.74
Total Moles
of Dry Gas
3.977xl08
4.078x108
Total Moles
of Wet Gas
4.122xl08
4.226xl08
Date
10-2
10-3
Annubar
Average Flow
Rate, SCMS
104.5
109.4
Annubar
Total Flow
SCM
9.026xl06
9.451xl06
Total Wet Gas
Volume, SCM
9.234xl06
9. 466x1 O6
Difference
%
+ 2.3
+ 1.6
ro
o
01
t
-------
8.5.4.2 1975 Test--
1975 test results are shown in Table 38 on a daily basis. It was
these results on which the conclusion was based that 49% of the flow went
through Duct 1. Plant data are given in terms of average rather than
total flow as in Table 37. Very good agreement is again indicated. Also,
note that the average flow rate in the stack from the pi tot traverses with
the scrubber off was 130.1 SCM/SEC (Table 34). This shows that there was
no apparent loss of accuracy of the stack traverses from taking data along
only one diameter. Flowrates for the 1975 test were higher than for full
load during the 1974 test due to higher amounts of excess air in the system
for the 1975 test period. Absence of data in Table 38 for the Duct 2 Annubar
on some days are solely a result of the pressure transducer being used for
other purposes.
Average daily flow rate through the stack is shown in Table 39, which
shows flow rates higher than those in Table 38. Plant data in Table 39
are based on the coal usage and nominal plant specifications for added
moisture and reheat air during the scrubbing process: 8% and 21% of the
flow through the ducts, respectively. During the test, the reheat fan
was operated at about 60% of full power according to plant personnel.
The fan is throttled to provide a temperature rise of 17°C in the flue
gas leaving the scrubber. The actual flow is not monitored. The measured
flow rate of 152.6 SCM/SEC is 18% higher than the average duct flow, and
so is reasonably compatible with a 60% below operation, which would result
in an average flow rate of 159 SCM/SEC.
8.5.5 Short Term Monitoring - 1974
The Ramapo probe was placed as shown in Figure 80 after the traversing
mechanism failed during the 1974 test. This corresponded to the row
location, for the Row Average Method, selected prior to the test. The probe
was left at the center of the row. Short term stability of the resulting
point measurement was compared with that of the Duct 1 Annubar and results
are shown in Table 40.
-207-
-------
Table 38. AVERAGE DAILY FLOW THROUGH DUCTS, 1975
Date
2-15
2-16
2-17
2-18
2-19
2-20
2-21
2-22
2-23
2-24
2-25
2-26
2-27
2-28
Ave
Average Flow And Standard
Deviation, Test
SCM/SEC
Duct 1
63.4 ± 1.7
63.5 ± 2.8
64.7 ± 2.3
67.3 ± 2.2
65.8 ± 2.3
65.3 ± 1.6
65.1 ± 0.9
63.2 ± 2.6
64.3 ± 5.2
60.1 ± 8.5
62.5 ± 1.7
62.7 ± 1.8
60.8 ± 2.1
60.3 ± 1.6
63.5
Duct 2
X
X
64.6 ± 1.9
64.5 ± 2.4
65.2 ± 3.0
66.7 ± 2.9
64.3 ± 3.9
67.3 ± 8.2
X
67.3 ± 3.0
67.2 ± 0.9
66.7 ± 2.1
X
X
66.0
Total
X
X
129.3
131.8
131.0
132.0
129.4
130.5
X
127.4
129.7
129.4
X
X
129.5
Average Flow In Ducts
From Plant Data
SCM/SEC
132.0
132.6
134.3
135.1
132.8
134.7
133.7
132.9
128.1
130.6
130.1
132.7
130.2
126.3
131.9
^Indicates insufficient data
-208-
-------
Table 39. AVERAGE DAILY FLOW THROUGH STACK, 1975
Date
2-15
2-16
2-17
2-18
2-19
2-20
2-21
2-22
2-23
2-24
2-25
2-26
2-27
*
2-28
Ave
Average Stack Flow And
Standard Deviation,
Test SCM/SEC
145.8 ± 3.8
149.0 ± 4.6
152.6 ± 4.0
159.4 ± 4.4
158.1 ± 9.1
155.7 ± 5.3
155.9 ± 5.0
156.1 ± 4.1
152.1 ± 4.3
154.3 ± 3.1
144.6 ± 3.2
145.9 ± 7.1
150.8 ± 3.8
155.4 ± 3.8
152.6
Average Stack Flow
From Plant Data
SCM/SEC
180.5
181.4
183.7
184.8
181.6
184.2
182.9
181.8
175.2
178.6
177.9
181.5
178.1
172.7
180.4
Assumes scrubber on full time
-209-
-------
Table 40. SUMMARY OF ANNUBAR AND RAMAPO SHORT
TERM MONITORING DATA, 1974
Data
Time
Data Points
Load, MW
9 - 27
14:10 - 14:59
50
118
10 - 3
19:00 - 19:59
60
114
Duct 1 Annubar Flow
Stack Annubar Flow
Ramapo Flow
Stack Annubar Flow
0.589 + 0.022 (3.6%)
0.418 + 0.096 (22.9%)
0.500 + 0.012 (2.4%)
0.414 + 0.071 (17.1%)
Data in Table 40 show two notable things. The first is a variation in
the Annubar calibration coefficients, which has been discussed above.
The second is that the point sensor readings, which on the average main-
tained a constant ratio with respect to the stack Annubar output, showed
a very high degree of scatter with respect to the Duct Annubar readings.
Since the Annubar reacts to the local velocity at five different points,
it is logical that it would show less data scatter than the Ramapo single
point sensor, and the data confirm this. The wide scatter for the Ramapo
probe also suggest that it was located in a region where velocity gradients
were relatively high. This was confirmed, as is shown in the following
paragraph, with resulting implications for use of the Row Average Method.
8.5.6 Row Average Analysis of Pi tot Traverse Data
Due to failure of the Ramapo traversing mechanism, it was necessary to
evaluate the Row Average Method by analysis of the pi tot traverse data, as
was done in the 1973 laboratory testing. Complete velocity maps are shown
for two of the 1974 runs in Table 41. Point velocities equal to zero
indicate a negative pi tot probe differential pressure, which was taken to
correspond to a local flow angularity of 90°, as explained in Section 7.8
and shown in Figure 77. A flow angularity of 90° corresponds to zero
local flow through the measurement plane. 1974 data were analyzed verti-
cally (rows) and horizontally (columns) as shown in Figure 79. Data are
presented in Table 42. Arguments could be made for taking the "Rows" in
either direction according to the definition of a row being in the plane
-210-
-------
Table 41. FIELD DEMONSTRATION PITOT TRAVERSE DATA, 1974
RUN 1 9-25 14:00 LOAD = 106MW
Row
1
2
3
4
5
6
7
Sample Point
1
0.000
0.000
9.516
4.011
6.115
7.182
3.780
2
9.009
0.000
6.928
9.166
8.687
7.663
7.059
3
10.657
7.362
6.674
8.973
7.497
7.011
8.299
4
11.490
8.373
7.747
7.552
6.395
6.966
7.189
5
8.941
6.575
9.267
8.939
6.578
7.089
7.020
6
6.448
8.511
9.700
9.463
6.389
6.299
5.928
7
4.227
7.094
8.600
9.453
5.482
4.630
3.495
RUN 3 9-28 11:00 LOAD = 52MW
Row
1
2
3
4
5
6
7
Sample Point
1
3.147
3.840
2.663
0.000
3.404
3.191
0.000
2
4.200
4.729
8.262
7.043
0.000
5.011
4.768
3
6.649
3.928
3.998
6.047
5.061
4.877
5.013
4
6.546
4.014
6.790
6.269
3.920
3.880
3.785
5
8.473
4.302
6.171
5.274
5.798
5.395
4.788
6
6.439
5.060
6.467
7.682
4.631
4.370
2.431
7
6.479
6.078
7.114
6.661
2.480
1.273
0.000
Tabular values are velocity at standard conditions, M/S
-211-
-------
Table 42. ROW AND COLUMN ANALYSIS OF FIELD
TEST PITOT-STATIC TRAVERSE DATA, 1974
RUN
1
2
3
4
5
6
7
8
9
AVE
a,%
Row
1
1.041
.878
1.285
.961
.952
.929
1.044
.962
.886
.993
11.8
2
.777
.930
.979
1.076
.956
.927
.952
.940
.939
.942
7.6
3
1.193
1.138
1.270
1.136
1.241
1.015
.977
1.072
1.094
1.127
8.3
4
1.180
1.244
1.195
1.184
1.127
1.228
1.228
1.201
1.129
1.191
3.3
5
.967
.953
.775
.872
.923
1.061
.912
1.201
1.013
.943
8.5
6
.960
.939
.858
1.107
.853
1.059
1.034
.996
1.056
.985
8.5
7
.877
.918
.637
.664
.948
.780
.854
.822
.882
.820
12.4
RUN
1
2
3
4
5
6
7
8
9
AVE
a,%
Column
1
.627
.594
.498
.522
.744
.477
.500
.461
.646
.563
16.0
2
.995
1.073
1.043
1.178
1.115
1.196
1.004
1.010
1.293
1.101
8.8
3
1.158
1.077
1.090
.977
1.097
.974
1.157
1.108
1.257
1.009
8.3
4
1.142
1.119
1.079
1.153
1.141
1.246
1.146
1.086
1.092
1.134
4.2
5
1.115
1.107
1.231
1.081
1.097
1.187
1.123
1.170
1.212
1.147
4.4
6
1.081
1.013
1.137
.988
.881
.958
1.072
1.135
1.221
1.054
9.4
7
.881
1.015
.922
1.101
.925
.961
.399
1.029
1.018
.983
6.5
Rows were vertical, columns horizontal
Tabular values are row average divided by overall average
-212-
-------
of the highest gradients, but they-were assumed to be vertical for obvious
hardware installation purposes. Results for the 1975 traverses are shown
in Table 43, the rows being vertical.
1974 results showed higher scatter since data were taken at both full
and half load, and only at full load in 1975. Rows 2 and 6, at 21.4% of
the duct width from the wall, are closest to the nominal recommended
distance of 19%. The average absolute error for these rows was 5.1% which
is considered reasonable, especially in view of the unusual duct geometry,
and the average standard deviation was ± 5.2%. The average error for row
4 was +15.4% and the average standard deviation was ± 3.5%. If a measurement
system using the Row Average Method were to be installed in the future in
one of these ducts, it would be recommended that it be placed in the
number 4 port rather than the number 2 or number 6 port. The systematic
error could then be calibrated out, and the working system would have a
minimal data scatter.
8.5.7 Use of Pitot-Static Probe in Field Applications
Once test personnel got accustomed to the equipment, 49 point pitot-
static probe traverses were conducted in each duct in about 50 minutes.
The probe head was occasionally bent during removal during the first two
traverses, but was not a problem thereafter. No clogging problems were
experienced during duct traverses. Clogging did occur in the stack with
the scrubber on after about one hour of continuous operation. This was
alleviated by wiping off the probe head and purging through the pressure
lines. No other problems were encountered.
8.6 SUMMARY OF RESULTS
There were significant problems with each instrument system during
the first test. The traversing mechanism failed to operate properly, which
has resulted in a general recommendation that such a device be perfected
and made generally available for large ducts. Previous work with smaller
devices in wind tunnels was the basis for construction of the test unit
used, but field operation made it evident that proper design of an adequate
system was outside program scope. The basic adequacy of the Row Average
Method itself was demonstrated through analysis of pitot traverse data.
-213-
-------
Table 43. SUMMARY OF PITOT TRAVERSE ROW AVERAGE DATA, 1975
Run
1
2
3
4
5
6
Ave
a, %
Duct 1 Row
1
0.920
0.913
0.910
0.905
0.961
0.963
0.929
2.6
2
1.090
0.994
1.090
1.108
1.014
1.046
1.057
4.0
3
1.191
1.197
1.123
1.103
1.120
1.070
1.134
4.0
4
1.163
1.235
1.070
1.157
1.142
1.154
1.154
4.2
5
0.856
0.861
0.913
0.929
0.913
0.900
0.895
3.2
6
0.882
0.895
0.969
0.950
0.945
1.035
0.946
5.3
7
0.898
0.904
0.926
0.848
0.904
0.833
0.886
3.8
Run
1
2
3
4
5
6
7
Ave
a, %
Duct 2 Row
1
0.921
0.934
0.906
0.969
1.036
1.074
1.116
0.994
7.6
2
0.892
0.927
0.922
0.892
0.959
0.904
0.894
0.899
2.3
3
1.268
1.236
1.247
1.185
1.198
1.197
1.166
1.214
3.0
4
1.093
1.124
1.094
1.133
1.187
1.078
1.105
1.116
3.0
5
0.949
0.961
0.960
0.942
0.936
0.936
0.883
0.938
2.6
6
0.987
0.995
0.982
0.976
0.898
0.988
1.028
0.979
3.7
7
0.891
0.850
0.890
0.903
0.885
0.823
0.808
0.864
4.0
Tabular values are row average divided by overall average for each
run.
-214-
-------
The Annubars suffered from fouling problems in the static orifice during
the first test. Automatic purging during the second test relieved the pro-
blem completely. The ones left in place after the first test showed no
long term effects from being out of use during the 4^ month interval
between tests other than requiring an initial cleanup, which took about
one hour for each probe. Agreement between continuous Annubar monitoring
data and plant data was excellent.
Pi tot-static probes were used for traverses in both the ducts and the
stack effectively and quickly.
-215-
-------
SECTION IX
DISCUSSION OF RESULTS
9.1 FUNDAMENTAL CONCEPTS
The problem of continuous total volumetric flow measurement in large,
complex ducts has been considered from the standpoint of both hardware and
technique. The basic methodology involved measurement of flow at a number
of discrete points and determining the relationship between these measure-
ments and the total flow rate. The objective of constructing and demonstrat-
ing an accurate, reliable, and economic system with current state of the
art hardware was achieved.
The current practice in flow measurement is to avoid taking data near
a large flow disturbance. The program was deliberately oriented toward
worst case conditions in this regard, since such situations are often
unavoidable in the real world. Most laboratory data were taken at about
two effective duct diameters downstream from large disturbances, and the
rectangular duct data in the field was taken 0.2 diameters downstream and
0.1 diameters upstream from a large disturbance. The general class of flow
considered may be defined as nondeveloped with regions of high angularity
and density stratification.
The general flow environment made it necessary to emphasize basic
physical concepts throughout the effort. The major one is that mass must
be conserved. Application of this principle was stressed in two ways. The
first is that averaging of flow data from a number of discrete points to
obtain total flow must be done at a single pressure and temperature to be
correct. This is why the term "total volumetric flow at standard conditions"
was so carefully defined in Section 4, and illustrated further in Appendix A.
It is also the reason why a recommendation is being made that point velocity
data in EPA Method 2 be averaged after conversion to standard temperature
and pressure rather than before, and also why it is being recommended that
point velocity sensors be manufactured with the capability to measure abso-
lute temperature and pressure.
The other application of the principle of conservation of mass had to
do with flow angularity with respect to the duct axis. In all cases, data
-216-
-------
were taken in a plane normal to the local axis. As was emphasized in
Section 4, only velocity components normal to this plane, i.e., parallel to
the axis, result in a flow through the plane. Consequently, test sensors
were analyzed to determine their ability to measure the correct flow com-
ponent. It was here that the S type probe exhibited its poorest perfor-
mance, and a preference for the ellipsoidal nosed pi tot-static probe became
apparent.
9.2 HARDWARE EVALUATION
This part of the program was very straightforward. Most applicable
devices turned out to be point sensors, and of these the Ramapo Fluid Drag
Meter was found to have superior accuracy. Factory improvements since com-
pletion of laboratory work have also made the probe's survivability charac-
teristics very acceptable. The Ellison Annubar, being an averaging sensor,
required separate evaluation as a technique. It has been found to be a
very good flow measurement device, with the limitation that purging of the
rear orifice is required in particulate laden flows. Results have indicated
that the most cost effective continuous flow measurement system would probably
use an Annubar as the velocity sensor.
Construction of an acoustic flowmeter for this program proved to be
impractical. The analysis work and field test results for that subtask have
been presented in the hope that they may lead to a solution of the problem.
9.3 TECHNIQUE EVALUATION
All methods considered inferred a total flow rate from measurements at
a small number of discrete points. The objective of using not more than
eight points was achieved in both circular and rectangular ducts in situations
where EPA Method 1 would require forty or more points. Perhaps due to the
extreme cases considered, no true "Universal" method which would require no
in-place calibration was found, although the Log-Linear 4 Method for circular
ducts comes extremely close. The next best case was achieved, however, for
all acceptable methods - a single in-place calibration.
Great emphasis was placed on development of the Row Average Method in
rectangular ducts for two reasons. The first is that it works best immedi-
ately downstream of an elbow, which is highly desirable in the practical
-217-
-------
sense due to the wide use of the rectangular elbow in industry. The second
reason is hardware considerations. Use of the Row Average Method requires
no significant modifications of existing ductwork; at most one or perhaps
two ports are required for instrument insertion. Since data are taken along
a straight line, a single reciprocating point sensor could constitute the
entire system.
Early during the program, there was concern that if an acoustic flowmeter
were built, no one knew how to use it in a rectangular duct. This same con-
cern extended to other advanced concepts such as laser velocimeters which
are also line averaging devices. The Row Average Method was clearly the
solution to the problem, and this was another reason for investigating it
so thoroughly. As these advanced types of flow sensor become more readily
available, it is important to know that there is now a method for using them
in rectangular ducts.
The good performance of the Annubar in rectangular ducts was at first
very surprising. As development of the Row Average method progressed, it
became clear that both techniques tended to work best in the same type of
situation and for the same reason. That reason was a consistent relation-
ship between velocity along the line of interest and the total flow rate.
Excellent performance of the Annubar was best demonstrated during the second
field test.
9.4 FIELD DEMONSTRATIONS
The second field demonstration involved correction of several problems
which occurred during the first, and showed uniformly good results. The
major regret is that the Row Average Method could only be evaluated through
analysis of traverse data rather than with an independent hardware system.
Both tests showed clearly that the standard pitot-static probe can be used
very effectively for reference measurements. It is our belief that in-place
calibration should be performed with maximum accuracy, and that this can be
best achieved through proper use of an ellipsoidal nosed pitot-static probe
for reference traverse measurements.
-218-
-------
9.5 FINAL COMMENTS
We have confidence in the methodology that has been developed and hope
to see it put to general use. Since some of the techniques are new, the
data base for them should be expanded. Techniques were developed to utilize
both existing off-the-shelf hardware and more advanced instruments which
are not yet in common use. Program results showed that continuous flow
measurement in large ducts is practical with presently available hardware
and techniques.
-219-
-------
SECTION X
REFERENCES
1. "Standards of Performance for New Stationary Sources",
Environmental Protection Agency; Federal Register, Vol. 36,
No. 159, Part II, 9-17-71.
2. The Analysis of Physical Measurements, E. Pugh and G.
Winslow; Addison-Wesley, 1966.
3. The Measurement of Air Flow, E. Ower and R.C. Pankhurst;
Pergamon Press, 1966.
4. Boundary Layer Theory, H. Schlichting; McGraw-Hill, 1968
5. Fundamentals of Acoustics, I.E. Kinsler and A.R. Frey,
John Wiley & Sons, Inc., 1962.
6. "A Simplified Integration Technique for Pipe Flow Measure-
ment", F.A.L. Winternitz and C.F. Fischl; Water Power, Vol.
9, page 225, 1957.
7. Fluid-Dynamic Drag, S.F. Hoerner; published by the author,
1958.
-220-
-------
SYMBOL
A =
A =
a,b,c, =
CD =
D =
I =
k =
1 =
L =
m =
Nu =
P =
Ap =
R =
R =
r =
Re =
S =
S =
SECTION XI
GLOSSARY
USAGE
area
absorption coefficient
constants
drag coefficient, ,—~-
drag force
energy decay parameter
general calibration constant
length
acoustic path length
mass flow rate
Nusselt number, T—
pressure
differential pressure
gas constant
Ramapo Drag Meter output
radius
Vd
Reynolds number,
H
area
Annubar calibration factor
DIMENSIONS
m
dimensionless
dimensionless
kg-m
sec2
dimensionless
m
m
ks_
sec
dimensionless
kg
m-sec
kg
m-sec
m2
sec2-°K
dimensionless
m
dimensionless
m2
dimensionless
-221-
-------
SYMBOL
V =
VVOUT'VIN
V =
at =
USAGE
absolute temperature
component of velocity vector
parallel to duct axis (axial
velocity component)
magnitude of velocity vector
voltage
volumetric flow rate
energy transfer parameter
acoustic impedance ratio
density
DIMENSIONS
°K
m
sec
m
sec
volts
m3
sec
dimension! ess
dimension! ess
a =
standard deviation
mean value
value at standard atmospheric
conditions
-222-
-------
SECTION XII
APPENDICES
Page
A. Averaging and Error Propagation Analysis Techniques 224
B. Flow Meter Survey 229
-223-
-------
APPENDIX A
AVERAGING AND ERROR PROPAGATION ANALYSIS TECHNIQUES
Error Propagation Analysis
The following technique is taken from Reference 2, "The Analysis of
Physical Measurements", Chapter 11. Assume that it is desired to
calculate parameter G, which is a known function of variables m,, M
—-Mr, given by
G = f(Mr M2, —-, Mr) (66)
where f denotes the functional relationship. Define the error in
the measurement of variable Mr as x^. The standard deviation of the
measurement of M is then given by
or2 = J1m JLlL_ (67)
where
a = standard deviation of
n = number of measurements
By derivation, the standard deviation of G is then given as
The equation in this form is used in Section 2 to perform an error
analysis of a point mass flow measurement.
In the remainder of the report, mean values and standard deviations
are calculated for much of the test data. For this analysis, if N
measurements are made of parameter G, then the mean value of G is
given by ^ G
_ = n=1 n _ (69)
where ~G = mean value
G = value of nth measurement of G
n
N = total number of measurements
-224-
-------
The standard deviation of G is given by
- - 2
(70)
b N
where o~ = standard deviation of G
b
If a known reference value (or assumed value) of G is known, then the
systematic error in the series of measurements is given by
E = G " GREF x 100% (71)
where E = systematic error, %
GREF = reference value of G
and the random error, which is a measure of the data spread, is given
by
-------
For the type of work performed in this program, the system random
error can be at least as important as the system systematic error.
While systematic errors are certainly undesirable, a constant systematic
error can often be removed by adequate system calibration. On the
other hand, large random errors can be much more difficult to correct,
and often imply an unacceptable weakness in the technique and/or
hardware being used.
2. Averaging of Data for Flow Measurement
The general mass flow equation as given in Section 4 is
m = //pudA =
A
The purpose of this section is to explain the type of average appro-
priate to the expression — . The main point to be made is that in
general
— i- — . — (72)
For a multi-point traverse, the mass flow rate is approximated by
N
m ~ n=l pn n r
For equal area segments, this becomes
i . JL?
In a flow field where either the density? or velocity u are constant,
the constant term can be removed from inside the summation sign. This
was done for some cases in laboratory testing when the fluid was air
and the static temperature and static pressure were quite constant,
meaning that the density was constant. These, however, must be
considered special cases. The most desirable practice is to calculate
either mass flow itself or volumetric flow rate at standard conditions.
The latter calculation was performed most often during the program.
Volumetric flow at actual conditions is not really meaningful unless
either the fluid density or velocity is constant and known.
-226-
-------
Consider the case of a duct divided into two equal area segments, with
fluid density and velocity constant in each section. By definition,
let
A = 1
PS = 1 (density at standard conditions)
As shown in Section 4,
Thus for this situation, we have
Pnun = :
J^-n'^hV^)
Several cases are shown in Table A-l. When either p, = p? or u-, = u?,
all calculated results are the same. When both density and velocity
are stratified, however, the calculation . is in error. The
P u
point is that density stratification must be taken into account in
performing traverses in order to minimize error. This can be done by
calculating either total mass flow or average velocity at standard
conditions. Even the latter calculation is in error when converted to
total mass flow if there is significant molecular weight stratification
in the flow field.
-227-
-------
Table A-l. TOTAL FLOW CALCULATIONS FOR FLOWS WITH VELOCITY AND/OR
DENSITY STRATIFICATION
CASE
1
2
3
4
pl
1
1
1
1
P2
1
2
3
2
Ul U2
1 2
1 1
1 3
2 1
P ^
1 1.5
1.5 1
1.5 1.5
1.5 1.5
p • U
1.5
1.5
2.25
2.25
A m
1.5
1.5
2.5
2.0
*.
1.5
1.5
2.5
2.0
no
ro
CD
I
FLOW CONDITIONS
pl
ul
Al
P2
U2
A2
=A
A = A] + A2 = 1
AVERAGES
P2)
m = pu A = ±
= u$A
'(u1 + u2)]A
+ P2u2)A
P, =
-------
APPENDIX B
Flow Meter Survey
-229-
-------
Table B-l . KEY UORD CLASSIFICATION FOR FLOW INSTRUMENTS
Number
Type
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
electromagnetic
flow tubes
laminar
mass
non-obstructing
nozzles
open channel
orifice meter
pi tot-type
positive displacement
turbine
variable area rotometer
Venturis
thermal anemometer
mechanical anemometer
acoustic
vortex shedding
other
FLOW METER SURVEY:
MEDIUM:
TYPE:
G - gas
L - liquid
S - Solid
Numbers correspond to key words
in Table.
FILE ACCESSION NUMBER:
TRW Systems internal retrieval system.
-230-
-------
FLOW METER SURVEY
I
PO
00
MANUFACTURER (No.)
MODEL (No.)
ARC Appareils Precision Control
Aeroflex Laboratories, Inc.
Aeroquip, Barco Div. ,
Barrington, Illinois
Air Products & Chemicals, Inc.
Allentown, Pennsylvania
238-2-05121
238-1-04660
Alnor Instrument Co.
Chicago, Illinois
Velometers
Type 606BP
Type 6070P
American Chain & Cable Co.
Waterbury, Connecticut
American Meter Co.
American Standard,
New Brunswick, New Jersey
SG-1
Astro Dynamics, Inc.
Autotronic Controls
Avien, Inc.
Woods ide, New York
Badger Meter Mfg. Co.
Milwaukee, Wisconsin
Type X701
Type T
MEDIUM
G
G
G
G
G
G
L,G
L,G
L,G
L
L
TYPE
10
11, 14, 15
11 14 15
II) ' " > Iw
12
12
12
9, 14, 15
9
14, 15
8, 10, 11
18
18
4, 10
10
I]
l l
2,6,7,8,10,11,13,
18
20
11
FILE ACCESSION NUMBER
614
612, 614
600
600
600
610, 531, 297, 612
610, 531
610, 531
614
600
600
614
614
613
W i w
87, 600, 614
600
600
-------
I
IN3
CO
(V)
I
MANUFACTURER (No.)
MODEL (No.)
Bailey Meter Co.
Wickliffe, Ohio
Type BY
Type
Type CV
Beckman Instruments, Inc.
Fullerton, California
Bel fort Instruments Co.
Bendix Environmental Science Div.
B&F Instruments, Inc.
BIF General Signal Corp.
BIF Industries, Inc.
Providence, Rhode Island
Current Meters
Shunt Current Meter
Biotronex Laboratory, Inc.
Blaw-Knox Co.
Copes-Vulcan Div.
Lake City, Pennsylvania
Blue White Industries
Bolt Assoc.
Norwalk, Connecticut
Bowles Fluidic Corp.
Silver Spring, Md.
The Bristol Co.
Haterbury, Conn.
Primary Elements
MEDIUM TYPE
L,G,S 1,3,8,12,18
L,G 18
L 12
S 18
15
G 15
G 11
L,G 2,6,7,8,11,13
L 11
G 11
1
9,10
G 18
L,G 8,13,18
L,G 18
FILE ACCESSION NUMBER
87, 600, 613, 614
600
600
600
612
612
612, 614
87, 613, 614
87
87
614
614
613
702, 703
600, 614
600
-------
ro
OJ
OJ
i
MANUFACTURER (No.)
MODEL (No.)
Brooks Instrument Div.
Emerson Electric Co.
Hatfield, Pennsylvania
DS-346
DS-HPB
D8800
3600
3300
7100
9400
Brown Instrument Div, Minneapolis-
Honeywell Regulator Co.
Long Island City, New York
Bubble-0-Meter
Temple City, California
Type 1-10-100
CGS Scientific Datametrics
Clean Room Products, Inc.
Climactronics Corp.
Consolidated Controls Corp.
Cox Instruments Div., Lynch Corp.
Detroit, Michigan
Ind Series
Crane Co.
MEDIUM
G,L
L
L
G,L
G,L
L
L
L
G
G
G,L
G
L
L
TYPE
1,4,10,11,12,18
1
11
12,13
12
11
1
18
10,18
18
4,14
14
15
3
11
11
10
FILE ACCESSION
270, 276,
270
270
270, 276
600
600
600
600
613
600, 614
600
257, 264,
612
612
614
600, 613
600
614
NUMBER
600, 614
612, 614
-------
I
ro
CO
MANUFACTURER (No.)
MODEL (No.)
Daniel Industries Inc.
Houston, Texas
PT Meter
Decker Corp.
Bala Cynwyd, Pennsylvania
Degamo Inc.
Dieterich Standard Corp.
Ellison Instrument Div.
New Buffalo, Michigan
E-100
E-101
Series 700
Primary Element
Dietz Henry 6. Co., Inc.
Disa - S&B, Inc.
Dresser Measurement Div.
Dwyer, F.W. Mfg. Co., Inc.
Michigan City, Indiana
500 Series
Rate-Master
Eastech, Inc.
MEDIUM
L
L
G,L
G
G
G,L
G,L
G
G.L
G.L
G.L
G,L
TYPE
2,11
11
12
8,9,15,18
8,9,15
8,9,15
18
18
8
4,14
10
9,12
12
12
19
FILE ACCESSION NUMBER
600, 613, 614
600
613
612, 614
74, 88, 269, 600
88
88
600
600
614
300, 612, 614
614
301. 614
600
600
611, 613, 614
Plainfield, New Jersey
VS-21
Eclipse Fuel Engineering
Rockford, Illinois
613
-------
I
ro
OJ
en
i
Manufacturer (No.)
MODEL (No.)
Electrosyn Technology Lab. Inc.
Canton, Massachusetts
Bourdon Tube
Ellison Instrument Div. .Dieterich
Standard Corp.
New Buffalo, Michigan
El 00
E101
Series 700
Primary Element
Epic Inc.
Erdco Engineering Corp.
Addison, Illinois
Fischer & Porter Co.
Warminster, Pennsylvania
Swirl meter
Model 10C1505
10A1151
10C1
10D1400
1 OS 1000
Flo-Tech, Inc.
Barrington, Illinois
The Flowcon Co.
Anaheim, California
F+ Series
Flow- Dyne Engineering, Inc.
MEDIUM
G.L
G,L
G,L
G
G
G,L
G,L
G
Gi
>L
G
G.L
G.L
L
L
G
G.L
L
L
L
TYPE FILE ACCESSION NUMBER
18
18
8,9,15,18
8,9,15
8,9,15
18
18
9,14,15
1,2,5,7,8,9,11,12,14
17,18
14,17
11
12
11
1
18
15,11
11
11
11
2,6,8,13
600
600
74, 88, 296, 600, 614
88
88
600
600
612, 614
87,274,532, 600, 611,
613, 614
532, 611
274
600
600
600
600
600, 612, 614
600
600, 614
600
614
-------
ro
GO
en
MANUFACTURER (No.)
MODEL (No.)
Flow Measurement Corp.
Kensington, Maryland
Flow Technology, Inc.
Fluid Data Inc.
Hauppague, New York
UF-100
Flui Dynamic Devices, Ltd.
Ontario, Canada
308
308R
Fluidyne Instrumentation
Foster Engineering Co.
Union, New Jersey
The Fox Valve DEvelopment Co.
The Foxboro Co.
Foxboro, Massachusetts
Model FG
Gelman Instrument Co.
GM Mfg. & Instrument Corp.
Geospace Corp.
Houston, Texas
Gems Co. ,Inc.
Farmington, Connecticut
VFI 700
General Air Drying Div.
MEDIUM
G
L
L
Gl
,L
G
G
G.L
G
G
L
L
TYPE
7,11,15
16
16
1O
o
18
13
10
2,6,8,13
8,11
11
3
15
12
12
9
FILE ACCESSION NUMBER
£ 1 *\
613
612. 614
600
600
534
534
534
614
613
614
365, 600, 613
600
614
612
613
600
600
614
-------
W.UFACTURER (NO.)
TYpE
FILE ACCESSION NUMBER
General Electric Co. ,
Industrial Process Control Div. L,G
West Lynn, Massachusetts
Series 553 L,G
Series 554 G
Gilmont Roger Instruments, Inc. L,G
Greiner Scientific Manostat
Guiton Industries, Inc.
Vibro-Ceramic Div.
Metuchen, New Jersey
W&LE Gurley
A Teledyne Co. L
Troy, New York
622 L
655 L
Hagan Corp.
Pittsburgh, Pennsylvania
Halliburton Co.
Duncan, Oklahoma L
Hallikainen Instruments
Richmond, California
Hamilton-Standard,
Windsor Locks, Connecticut b
Hays Corp.
Michigan City, Indiana
18
18
18
12
12
15,11
11
11
''
lb
600, 613
600
600
600, 612
614
613
600, 612
600
600
6I3
60°
609
-------
ro
CO
oo
MANUFACTURER (No.)
MODEL (No.)
Hastings-Raydist
Hampton, Virginia
Gas Flow Probe
No.AFI Series
LF&HF Series
OF Series
Hersey-Sparling Motor Co.
Dedham, Massachusetts
Hoke Inc.
Honeywell, Inc.
Industrial Div.
Fort Washington, Pennsylvania
Diaphragm
Bellows
Hughes Instrument
Bristol, Pennsylvania
Industrial Measurements Controls
410-3
Industrial Physics & Electronics
Interval Corporation
Agoura, California
Model 44
In-Val-Co. Combustion
Engineering, Inc.
Tulsa, Oklahoma
S-Series
W-Series
MEDIUM
6
G
G
G
L.G
L.G
L,G
L
L
G
G
L
L
L
TYPE
2,3,4,8,9,14
FILE ACCESSION NUMBER
88, 89, 185, 209, 210, 216, 227,
231, 305, 600, 612, 613
88, 89, 185, 209, 210, 216, 227,
231, 612, 614
14
14
2
18
18
18
10
10
5
16
16
11
11
11
600
600
613
614
600
600
600
613
600,
600
614
87,
87,
600,
600
600
614
530
530
614
-------
ro
oo
MANUFACTURER (No.)
MODEL (No.) MEDIUM
ITT Barton Controls & Instr.Div. . r
Monterey Park, California '
F-500 L,G
Ion Exchange Products
Kenics Corp.
King Engineering Corp.
Kingmann-White Inc.
Placentia, California '
515 L,G
509 L,G
Kontes Glass Co. . _
Vine land, New Jersey '
Delta P G
K627900 L,G
K West Corp. p
Westminster, California
Model 325 G
Laboratory Data Control
Larson Aero Development
Leeds & Northrop Co. L,G
North Wales, Pennsylvania
1911 L,G
1913 G
Leopold & Stevens Inc.
Librascope Group, General
Precision Systems, Inc. G
Glendale, California
L60-1 G
TYPE
10,11
10
10
12
8,18
18
18
2,12,18
18
12
16
16
2,10
1,4,9,11,12,13
1
1
1
7
18
18
FILE ACCESSION NUMBER
600,
600
614
614
600,
600
600
600,
600
600
228,
228,
614
614
600,
600
600
614
600
600
614
614
614
229
229
614
-------
o
p
MANUFACTURER (No.)
MODEL (No.)
Linden Labst
State College, Pennsylvania
Turbine
Rotary
Lindsey Meter Co.
Pasadena, California
Turbine
Rotary
Link-Belt, FMC Corp.
Lion Precision Inc.
LKB Instruments Inc.
Liquid Controls Corp.
North Chicago, Illinois
M-3
M-70
Lube Devices Inc.
Mace Corp.
South El Monte, California
960 Series
Manostat Corp.Div. Greiner
Scientific Corp.
New York, New York
36-541-03
36-541-04
36-541-05
36-541-30
Marine-Electro Mechanical Inc.
Marotta Scientific Controls, Inc.
MEDIUM
L.G
L,G
G
L,G
L,G
G
L.G
L
L
L,G
L.G
L,G
L,G
L,G
L,G
L,G
TYPE
11,12
11
12
11,12
11
12
6
3
2,5
10,18
10
10
2,5,8
19
19
12
12
12
12
12
12
13
FILE ACCESSION NUMBERS
600, 613
600
600
600
600
600
614
614
614
87, 600, 614
600
614
600
600
600
600
600
600
600
614
614
-------
I
ro
MANUFACTURER (No.)
MODEL (No.) MEDIUM
Martek Instruments, Inc.
Martin-Decker Corp.
Matheson Gas Products Div.
600 Series L
8112 Series G
8116 Series G
Meco Modern Engineering Co. ,Inc.
St. Louis, Missouri
6401 Argon
Mec-0-Matic Inc. ,
Miami, Florida
251 L
Maxson, W.L. Corp.
New York, New York
Meriam Instruments Co.
Cleveland, Ohio
Laminar Flow
Metrology Research Inc.
Altadena, California
Micron Instruments Inc.
McNeil Corp.
Modern Engineering Co., Inc.
Moore Products Co.
Spring House, Pennsylvania
Narco Bio-Systems Inc.
TYPE
15
4
2,4,12,18
12
18
4
12
12
18
18
3,4,8,9
3
1
10
1
1
FILE ACCESSION NUMBERS
612
614
600, 614
600
600
600
600
600
600, 613, 614
600
613
614
614
614
614
-------
I
PO
PO
MANUFACTURER (No.)
MODEL (No.)
National Instrument Lab, Inc.
General Kinetics Inc.
Rockville, Maryland
10-20-30
20-10-250
10-R-100
New Jersey Meter Co.
Plainfield, NEw Jersey
Neptune Meter Co.
Long Island City, New York
Norrich Plastics Corp.
Nusonic Corporation, Electronics
System Div.
Paramus, New Jersey
6155
Panametrics Inc.
Waltham, Massachusetts
Penn Instrument Div.
Burgess Manning Corp.
Philadelphia, Pennsylvania
Permutit Co.
Potter Aeronautical Corp.
Union, New Jersey
Series 3000
Quantachrome Corp.
MEDIUM
G
G
G
G
L
G
G
L,G
L
L
TYPE
3:18
18
18
18
10,11
6
16
15
4
2,6,7,13
4,11
4
12
FILE ACCESSION NUMBERS
600, 614
600
600
600
613
600, 614
614
612
612
87, 613
614
87, 600
600
614
-------
ro
-P.
CO
MANUFACTURER (No.)
MODEL (No.)
Quantum Dynamics Inc.
Tarzana, California
QL 432 WRG 1C
QL 448 WRC G
Ramapo Instrument Co., Inc.
Bloomingdale, New Jersey
X
Mark V
Republic Flow Meter Co.
Chicago, Illinois
Research Appliance Co.
Allison Park, Pennsylvania
Revere Corp. of America
Wallingford, Connecticut
Robinson Orifice Fitting Co.
Houston, Texas
R
Rockwell Mfg. Co.
Bala Cynwyd, Pennsylvania
Eureka B
Series L
Series M
Rotron Controls Div
Woodstock, New York
P Series
Ross Inc.
Matheson Gas Products Div.
East Rutherford, New Jersey
MEDIUM
L,G
G
L
L,G
L,G
L,G
L,G
L,G
L
L
L
L
L,G
L,G
TYPE
4,11,18
18
18
12,19
12
18
8,9,12
6,13,18
18
11
11
11
11
4,13
13
FILE ACCESSION NUMBERS
600, 614
600
600
600, 613, 614
600
600
C1 *5
613
614
Cl O
0 1 o
600, 614
600
600
600
600
600
600, 613
600
-------
ro
MANUFACTURER (No.) men T MM
MODEL (NO.) MEDIUM
Saratoga Systems Inc. . n
Cupertino, California '
Ultrasonic Flow . r
Ultrameter L'
Scarpa Laboratories Inc. , Q
Metuchen, New Jersey '
Ultrasonic Flowmeter
SFSM-5-RPS L,G
SFM-4-RPS L,G
SFM-3-BP L,G
SFS-2-RPS L,G
Schroeder Brothers Corp. ^
McKeesport, Pennsylvania
Schutte & Koerting Co. . Q
Cornwell Heights, Pennsylvania *
SK-ST Flo-thrutronic G
18000 Series L,G
19000 Series L,G
Selas Corp. of America
Sentry Equipment Corp.
Oconomowoc, Wisconsin
Signet Controls Inc.
Simerl, R.A.
Simmonds Precision Instrument Co.
Tarry town, New York
Simplex Valve & Meter Co.
Lancaster, Pennsylvania
TYPE
16
16
3,4,5,16
16
16
16
16
13
1 ^J
12
12
12
12
11
11
8
15
11
FILE ACCESSION NUMBERS
449,
449,
446,
446,
446,
446,
446,
600
273,
273
600
600
614
600,
614
612
613,
\s t *r f
613
450
450
448, 601-608, 614
448, 601-608
448, 601-608
448, 601-608
448, 601-608
600
614
614
-------
ro
-p»
on
MANUFACTURER (No.)
MODEL (No.)
SK Instruments Div.
SMI Instruments Div.
Smith, A.O. Corp. Meter Systems
Standard Controls
Seattle, Washington
Statham Instruments, Inc.
Oxnard, California
E-3000
SP2200
Taylor Instrument Co. .Process
Control Div.,Sybron Corp.
Series 300
Series 750
Series 737
Technol ogy/Versatroni cs
Yellow Springs, Ohio
MFC
NL
Temp line
Tescom Instruments Div.
Thermal Instruments Co.
MEDIUM TYPE
2,12
2,12
10,11
i
\
L 1
L 1
L,6 1,5,8,9,13,15,18
L,G 18
L.G 18
L 1
4
4
4
9
8
3,4,5,14
FILE ACCESSION NUMBERS
617
614
614
f* T O
613
cnr>
ODD
600
600
226, 612, 613, 614
600
600
600
600, 614
600
600
614
614
612, 614
-------
ro
.£.
cn
MANUFACTURER (No.)
MODEL (No.)
Thermo Systems Inc.
St. Paul , Minnesota
VT-163
VT-161
VT-160
P-2
10
20
1223 & 1229
352 IG
352 IL
Thermoneti cs Corp.
TIP Instruments
Tower Systems Co.
Trans-Sonic Inc.
Lexington, Massacusetts
Trimont Instrument Co.
Tyland Corp.
Torrance, California
FMS 311
RP 781
United Sensor Control Corp.
Wallace & Tiernan Inc.
Belleville, New Jersey
Varea-Meter Bypass
NA-11 & 12
MEDIUM
L,G
G
G
G
G
L,G
L,G
G
G
L
G
L
G
L,G
L,G
UG
TYPE
14
14
14
14
14
14
14
14
14
14
14
15
15
8,9
4,14,18
14
18
2,9,13
12
12
12
FILE ACCESSION NUMBERS
87, 88, 218, 219, 294, 600, 612.
613, 614
219
219
219
218
218
218
294
600
600
612
612
612
/" 1 *1
613
614
600, 614
600
307, 614
249, 600, 614
249
600
Waugh Engineering Co.
Van Nuys, California
613
-------
MANUFACTURER (No.)
MODEL (No.)
urnT,,u
MEDIUM
TYPE
FILE ACCESSION NUMBERS
Waukee Engineering Co.
Milwaukee, Wisconsin
Westberg Mfg. Co.
Westinghouse Computer Instrument
Hagan Computer System Div.
Pittsburgh, Pennsylvania
3000 Series
Wilcoxin Research
Bethesda, Maryland
Wilmad Glass Co. ,Inc.
15
6,7,18
18
600, 614
612
600, 614
600
613
614
PO
-------
TECHNICAL REPORT DATA
(Please read Inslructions on the reverse before completing)
1. REPORT NO.
EPA-650/2-75-020
2.
3. RECIPIENT'S ACCESSION-NO.
4. TITLE ANDSUBTITLE
Continuous Measurement of Total Gas Flowrate
from Stationary Sources
5. REPORT DATE
February 1975
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO
E.F. Brooks, E.C.Beder, C.A.Flegal, D.J.Luciani,
and R.Williams
9. PERFORMING OR8ANIZATION NAME AND ADDRESS
TRW Systems Group
One Space Park
Redondo Beach, CA 90278
10. PROGRAM ELEMENT NO.
1AB013; ROAP 21ACX-AE
11. CONTRACT/GRANT NO.
68-02-0636
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
NERC-RTP, Control Systems Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PI
Final; 10/72-12/74
PERIOD COVERED
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The program objective was to evaluate hardware and techniques for the
continuous measurement of the total gas flowrate from stationary sources, speci-
fically in large or complex ducts where total flow metering devices such as plate
orifices are not practical. Work consisted of formulation of operating specifica-
tions, evaluation of commercially available velocity sensors, development and
evaluation of flow mapping techniques, and field demonstration of both hardware
and technique. Results showed that total volumetric flowrate can be measured
with accuracies consistently better than 10% in either circular or rectangular
ducts through proper placement of from one to eight flow sensors, when standard
traversal techniques would require twenty to fifty traverse points. The rectangu-
lar duct mapping techniques developed during the program were found to have
optimum accuracy immediately downstream of an elbow. Several off-the-shelf
velocity sensors were found acceptable for use in the specified stack-type
environment. The field demonstrations verified the acceptability of both
hardware and techniques.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air Pollution
Velocity Measurement
Gases
Flow Rate
Flowmeters
Flow Distribution
Ducts
Air Pollution Control
Stationary Sources
Continuous Measurement
Velocity Sensors
13B,
14 B
20D
13K
8. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
262
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
-248-
------- |