;PA-600/2-76-170
'une 1976
Environmental Protection Technology Series
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research arid Development, U S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
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 through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
PARTICULATE SAMPLING STRATEGIES
FOR LARGE POWER PLANTS
INCLUDING NONUNIFORM FLOW
by
H. A. Hanson, R. J. Davini, J. K. Morgan
and A. A. Iversen
FluiDyne Engineering Corporation
5900 Olson Memorial Highway
Minneapolis, Minnesota 55422
Contract No. 68-02-1244
Project Officer
Thomas E. Ward
Emissions Measurement and Characterization Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
U, S, ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
/x/
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences
Laboratory, U. S. Environmental Protection Agency, and approved
for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the U. S. En-
vironmental Protection Agency, nor does mention of trade names
or commercial products constitute endorsement or recommendation
for use.
-------
ABSTRACT
This report describes the results of a study to de-
termine the effects that various geometric ducting config-
urations have on the flow profiles and the distribution of
particulate in ducting systems of large (>100 MW) power
plants. The program included both laboratory model studies
and field testing at large power plants. The measurement
of total volumetric flow and particulate emissions at less
than full operating capacity was also investigated. The
results of flow angularity measurements in large stacks at
typical sample port locations, including downstream of in-
duced draft fans, are similarly discussed. Special atten-
tion was given to the aerodynamic effects of S-tube/sampling
probe interference on velocity measurements with an S-tube
in EPA Stack Emissions Measurement Reference Methods 2 and 5.
A computerized technique was used to determine the
effectiveness of various equal-area sampling strategies in
providing accurate measurements of three emission parameters:
average particulate concentration, total volumetric flow
rate, and total emissions. Numerous typical and atypical
velocity and particulate concentration profiles were studied.
Sampling strategy recommendations are presented.
111
-------
CONTENTS
ABSTRACT
FIGURES viii
TABLES xx
ACKNOWLEDGEMENTS xxi
I INTRODUCTION 1
II. CONCLUSIONS AND RECOMMENDATIONS 4
III. LITERATURE AND PERSONAL CONTACT SURVEY 7
A. INTERNAL DUCT FLOW 7
B. SURVEY OF POWER PLANT DUCTING 9
1. TVA Data 9
2. File Drawings 10
3, Design Surveys 10
4. Manufacturers Literature 11
C. FLOW ANGULARITY IN LARGE STACKS 23
1. Tall Chimneys 23
2. Cold-Flow Model Studies 23
3. Preliminary Survey 29
4. Preliminary Conclusions 30
D. SAMPLING METHODOLOGIES 31
1. EPA Test Methods 31
2. ASME Power Test Codes 31
3. ASTM Methods 34
4. British Standards 34
5. Miscellaneous Techniques 34
IV, MODELING 39
A. SIMILARITY PARAMETERS 39
B. EFFECT OF MODEL SCALE ON SIMULATION 40
C. PROTOTYPE TEST SITE SELECTION 41
D, MODEL TEST FACILITY 43
E. INSTRUMENTATION 45
-------
CONTENTS (Cont.)
F, TEST PARTICULATE 46
G, LOCAL PARTICULATE. CONCENTRATION
MEASUREMENTS 4 8
H, KONITEST DISTRIBUTION MEASUREMENTS 57
I. PRECIPITATOR PLATES 60
J. MODEL TEST RESULTS WITH SKEWED PARTI-
CULATE DISTRIBUTION 65
K, MODEL TEST RESULTS WITH AND WITHOUT
TURNING VANES 70
V. FIELD TESTING 78
A. FIELD TESTS AT ALLEN S. KING POWER
STATION 78
1. Inspection of Ductwork Interior 78
2. Particulate Concentration Measure-
ments Downstream of Prototype
Test Elbow 78
3. Particulate Concentration Measure-
ments Upstream of Prototype
Test Elbow 80
B. FIELD TESTING AT BLACK DOG POWER
STATION 98
C. ADDITIONAL FIELD TEST DATA 106
D. FIELD TESTS AT PART LOAD OPERATION 107
E. FLOW ANGULARITY IN LARGE POWER PLANT
STACKS 117
1. Field Observations 117
2. Stack Sampling 117
3. Double Vortex Phenomenon 120
4. Field Test Probes 121
5. Field Test Program at Allen S.
King Plant 122
6. Field Tests at TVA 123
7. Final Conclusions and Recom-
mendations 125
VI
-------
CONTENTS (Cont.)
F, FLOW ANGULARITY IN LARGE POWER
PLANT BREECHINGS 138
VI, S-TUBE AERODYNAMIC INTERFERENCE STUDY 139
VII. SAMPLING STRATEGY 140
A. BACKGROUND 140
B, COMPUTERIZED EVALUATION 140
C. RECTANGULAR DUCTS 143
D. ROUND DUCT SAMPLING STRATEGY 150
REFERENCES 157
APPENDICES
A. DATA FROM THE EVALUATION OF VARIOUS
EQUAL AREA SAMPLING STRATEGIES 161
B. AERODYNAMIC EFFECTS ON VELOCITY
MEASUREMENTS WITH AN S-TUBE IN
EPA METHODS 2 AND 5 308
Vll
-------
FIGURES
1 Johnsonville units 1-6 (Tennessee Valley
Authority) 13
2 Kingston units 1-10 (Tennessee Valley Authority) 14
3 John Sevier units 1-4 (TVA) 15
4 Gallatin units 1-4 (TVA) 16
5 Colbert unit 5 (TVA) 17
6 Paradise units 1-3 (TVA) 18
7 Karn units 1 and 2 (Consumers Power Company) 19
8 Campbell unit 2 (Consumers Power Company) 20
9 High Bridge unit 3 (Northern States Power Company) 21
10 Roxboro unit 3 (Carolina Power and Light) 22
11 Multiples breeching arrangement in Colbert 24
12 Triple inlet breeching arrangement in the
Cumberland Steam Plant 25
13 Multiple breeching arrangement at the Allen Steam
Plant 26
14 Multiple breeching arrangement at the Bull Run
unit number 1 Steam Plant 28
15 Minimum number of traverse points (EPA Method 1) 32
16 EPA sampling point locations 33
17 Error in 4-point averaging of some arbitrary
axially-symmetric velocity distributions 36
18 Minimum number of measurements for rectangular
sampling sites (ASTM D-3154) 37
19 British Standards traverse plan 38
20 Circular ring analog plan 38
21 Effect of Model Scale with Model Froude number
and Stokes number equal to prototype values 42
viii
-------
FIGURES (Cont.)
22A The 1/10 scale model of the Allen S, King
Power Plant breeching and electrostatic
precipitator 44
22B The 1/10 scale model of the Allen S, King Power
Plant breeching showing upstream and downstream
sampling sites of the test elbow 44
23 Particulate sampling instrumentation for model
tests 47
24 Size distributions of Fly Ash in 1/10 scale
model (upstream of test elbow) 49
25 Model ductwork layout 52
26 Velocity distribution measured upstream of the
test elbow in the 1/10 scale model 53
27 Velocity distribution measured downstream of
the test elbow in the 1/10 scale model 54
28 Particulate concentration distribution measured
upstream of the test elbow in the 1/10 scale model 55
29 Particulate concentration distribution measured
downstream of the test elbow in the 1/10 scale
model 56
30 Konitest sample dilution system 61
31 Particulate concentration distribution measured
downstream of the test elbow in the 1/10 scale
model with the Konitest (1/4 inch probe - 0,64 cm) 62
32 Particulate concentration distribution measured
downstream of the test elbow in the 1/10 scale
model with the Konitest (1/2 inch probe - 1.27 cm) 63
33 Konitest signal current vs lateral position
(Looking downstream)-sampling site located
downstream of the test elbow in model 64
34 Precipitator plates installed in the laboratory
model ductwork 66
35 Laboratory model ductwork layout 66
ix
-------
FIGURES (Cont.)
Number Page
36 Comparison of velocity profiles before and after
addition of precipitator plates 67
37 Lateral filter sample survey taken upstream and
downstream of the test elbow in the 1/10 scale
model 68
38 Details of model internal cross-bracing 69
39 Particulate concentration distribution measured
upstream of the test elbow in the 1/10 scale
model (upstream particulate distribution
artificially skewed) 71
40 Particulate concentration distribution measured
downstream of the test elbow in the 1/10 scale
model (upstream distribution artificially skewed) 72
41 Velocity distribution measured upstream of test
elbow in 1/10 scale model (No turning vanes) 74
42 Velocity distribution measured downstream of the
test elbow in the 1/10 scale model (no turning
vanes) 75
43 Particulate concentration distribution measured
upstream of the test elbow in the 1/10 scale
model (no turning vanes) 76
44 Particulate concentration distribution measured
downstream of the test elbow in the 1/10 scale
model (no turning vanes) 77
45 Twin electrostatic precipitators and horizontal
breeching of the Allen S. King Power Plant 84
46 Internal view of the Allen S, King Power Plant
ducting 85
47 Internal view of the Allen S. King Power Plant
ducting 86
48 Internal view of the Allen S. King Power Plant
ducting 87
49 A side view of the horizontal breeching at the
Allen S, King Power Station 88
x
-------
FIGURES (Cont.)
Number Page
50 Photographs of the Field Test Site 89
51 Particulate concentration and velocity across
width of duct downstream of prototype test
elbow in the Allen S, King plant 90
52 Particulate concentration across width of duct
downstream of prototype test elbow (Allen S,
King Plant) 91
53 Mass median diameter (microns) as a function of
cross-sectional position 92
54 Velocity distribution upstream of test elbow at
the Allen S, King Plant 93
55 Particulate concentration measured upstream of
the prototype test elbow (Allen S, King Plant) 94
56 Particulate concentration measured downstream
of the prototype test elbow (Allen S. King Plant) 95
57 Velocity distribution across width of duct
upstream of precipitator at the Allen S. King
Power Station 96
58 Particulate concentration across width of duct
upstream of precipitator at the Allen S, King plant 97
59 Side view of north breeching (Unit No.4) at
the Black Dog Power Station 99
60 Velocity distribution measured 1,5 equivalent duct
diameters downstream of I. D. Fan at the Black
Dog Power Station 100
61 Velocity distribution in north breeching of a
midwestern power plant (250 MW load) - sampling
station located downstream of I. D, Fan, 101
62 Velocity distribution in north breeching of a
midwestern power plant (330 MW load) *• sampling
station located downstream of I, D. Fan, 102
63 Velocity distribution in south breeching of a
midwestern power plant (250 MW load) - sampling
station located downstream of I, D, Fan 103
xi
-------
FIGURES (Cont.)
Number Page
64 Velocity distribution in south breeching of a
midwestern power plant (330 MW) sampling station
located downstream of I. D, Fan 104
65 South breeching of a midwestern power station 105
66 Particulate concentration downstream of precipi-
tator in south breeching of the Allen S. King
Plant - Half load operation 111
67 Particulate concentration profile downstream of
precipitator in south breeching of the Allen S.
King Plant during full load operation 112
68 Power plant load, velocity and particulate
concentration during the transition period
from low load to full load downstream of pre-
cipitator in south breeching of the Allen S.
King Power Plant 113
69 Particulate concentration upstream of precipitator
at the Allen S. King Plant - Half load 114
70 Particulate concentration across width of duct
upstream of precipitator at the Allen S. King
Plant - full load 115
71 Power plant load, velocity and particulate con-
centration during the transition period from
low load to full load above the precipitator
in south breeching of the Allen S. King Plant 116
72 Twin spiraling vortices observed at the Allen
S. King Plant 127
73 Single spiral flow 128
74 Double helical flow 129
75 Complex three-dimensional flow field observed
when cooling air is ejected from holes on
turbine blades into a cross-flowing main
flow stream 130
76 Fecheimer type pitot probe 131
77 Visual flow direction indicator 131
xii
-------
FIGURES (Cont.)
Number Page
78 Fecheimer pitot probe sensitivity (Yaw angle) 132
79 Cone-type five-hole pitot probe 133
80 Five-hole pitot probe sensitivity (Yaw angle) 134
81 Velocity profile in Allen S. King Power Station
stack at the 61 meter level 135
82 Flow Angularity in Colbert Unit No. 1 Stack 136
83 Flue gas turning vanes at Colbert Steam Plant 137
84 Error in volumetric flow rate versus percentage 146
of the 21 velocity profiles studied with the
given error level for three different sampling
schemes
85 Error in volumetric flow rate versus percentage
of the 21 velocity profiles studied with the
given error level for four different sampling
schemes, all with 12 total traverse points. 147
86 Sampling error versus aspect parameter for
velocity distribution shown in Figure A-10 148
87 Sampling error versus aspect parameter for
velocity distribution shown in Figure A-l 149
88 Sampling error versus round duct aspect parameters
(Ratio of the number of sampling rays to the
number of sampling points per ray) for three
different 16-point sampling surveys for the
velocity distribution shown in Figure A-50 156
A-l Velocity distribution simulating profile measured
upstream of test elbow (80° mitered bend) - Allen
S. King Plant 165
A-2 Particulate concentration distribution simulating
profile measured downstream of the electrostatic
precipitator at the Allen S, King Plant 169
A-3 Velocity distribution simulating profile measured
across width of duct upstream of the electrostatic
precipitator at the Allen S, King plant 173
xiii
-------
FIGURES (Cont.)
Number Page
A-4 Particle Concentration distribution simulating
profile measured across width of duct upstream
of the electrostatic precipitator at the Allen S.
King Plant 177
A-5 Theoretical distribution (Possible distribution
downstream of an electrostatic precipitator
where hopper sweepage is significant ) 180
A-6 Velocity distribution simulating profile measured
upstream of test elbow (without turning vanes) in
1/10 scale model 184
A-7 Particle concentration distribution simulating
profile measured upstream of test elbow (without
turning vanes) in 1/10 scale model 188
A-8 Velocity distribution simulating profile measured
downstream of test elbow (without turning vanes)
in 1/10 scale model 192
A-9 Particle concentration distribution simulating
profile measured downstream of test elbow
(without turning vanes) in 1/10 scale model 196
A-10 Velocity distribution simulating profile measured
upstream of test elbow (with turning vanes) in
1/10 scale model 200
A-ll Velocity distribution simulating profile measured
downstream of test elbow (with turning vanes) in
1/10 scale model 204
A-12 Velocity distribution simulating profile observed
in rectangular duct between I, D. fan and stack in
a 120 MW oil-fired electrical generating station 208
A-13 Velocity distribution simulating profile measured
downstream of I, D. Fan at Black Dog Power Station 212
A-14 Velocity distribution simulating profile at
Allen Steam Plant Unit No, 1 precipitator inlet
ports 1-3 216
A-15 Velocity distribution simulating profile at Allen
Steam Plant Unit No, 1 precipitator inlet ports
4-6 220
xiv
-------
FIGURES (Cont.)
N mnber Page
A-16 Velocity distribution simulating profile at
Allen Steam Plant Unit No, 1 precipitator
inlet ports 7-9 224
A-17 Velocity distribution simulating profile at
Allen Steam Plant Unit No. 1 precipitator inlet
ports 1-12 228
A-18 Velocity distribution simulating profile at Allen
Steam Plant Unit No, 1 precipitator inlet ports
13-15 232
A-19 Velocity distribution simulating profile at Allen
Steam Plant Unit No. 1 precipitator inlet ports
16-18 236
A-20 Velocity distribution simulating profile at
Colbert Steam Plant Unit No, 4 Precipitator
inlet ports 13-24 240
A-21 Velocity distributions simulating profile at
Colbert Steam Plant Unit No. 1 precipitator
inlet ports . 1-12 244
A-22 Velocity distribution simulating profile at Black
Dog Power Station, I. D, Fan Inlet Station No. 1 248
A-23 Velocity distribution simulating profile measured
at Black Dog Power Station, I, D. Fan Inlet Station
No. 2 252
A-24 Velocity distribution simulating profile measured
at Black Dog Power Station, I. D. Fan Inlet
Station No. 3 256
A-25 Velocity distribution simulating profile measured
at Black Dog Power Station, I. D. Fan Inlet
Station No. 4 260
A-26 Velocity distribution simulating profile measured
at Black Dog Power Station, I. D. Fan Inlet
Station No. 5 264
A-27 Duct layout at Black Dog Power Station 275
A-28 Duct layout at Black Dog Power Station 276
A-29 Duct layout at Colbert Steam Plant units 1-4 277
xv
-------
FIGURES (Cont.)
Number Page
A-30 Velocity distribution simulating that measured
at the 23 meter level (about four stack diameters)
above the stack inlet in a 3,35 meter diameter
foundary stack 278
A-31 The effect of angular orientation on the accuracy
of four different 2 ray sampling strategies for
the velocity profile shown in Figure A-30 280
A-32 The effect of angular orientation on the accuracy
of four different 4 ray sampling strategies for the
velocity profile shown in Figure A-30 281
A-33 The effect of angular orientation on the accuracy
of four different 8 ray sampling strategies for
the velocity profiles shown in Figure A-30 282
A-34 Velocity distribution simulating that measured
at the 91 meter (300 ft) level in a 9,1 meter
(30 ft) diameter stack at a large coal-fired
electrical generating station 283
A-35 The effect of angular orientation on the accuracy
of four different 2 ray sampling strategies for
the velocity profiles shown in Figure A-34 285
A-36 The effect of angular orientation on the accuracy
of four different 4 ray sampling strategies for
the velocity profile shown in Figure A-34 286
A-37 The effect of angular orientation on the accuracy of
four different 8 ray sampling strategies for the
velocity profile shown in Figure A-34 287
A-38 Velocity distribution simulating that measured
2.2 stack diameters above the breeching plane inlet
in the stack (2,4 meter diameter) from an oil-fired
steam generator 288
A-39 The effect of angular orientation on the accuracy
of four different 2 ray sampling strategies for the
velocity profile shown in Figure A-38 290
A-40 The effect of angular orientation on the accuracy
of four different 4 ray sampling strategies for
the velocity profiles shown in Figure A-38 291
A-41 The effect of angular orientation on the accuracy of
four different 8 ray sampling strategies for the
velocity profiles shown in A-38 292
xvi
-------
FIGURES (Cont.)
Number
A-42 Velocity distribution simulating that measured at
the 85.3 meter level (approximately 13 stack
diameters above the breeching plane inlet) in the
stack at the Allen Steam Plant Unit No. 1 293
A-43 The effect of angular orientation on the accuracy
of four different 2 ray sampling strategies for
the velocity profiles shown in Figure A-42 295
A-44 The effect of angular orientation on the accuracy
of four different 4 ray sampling strategies for
the velocity profiles shown in Figure A~42 296
A-45 The effect of angular orientation on the accuracy
of four different 8 ray sampling strategies for
the velocity profile shown in Figure A-42 297
A-46 Velocity distribution simulating that measured
at the 85.3 meter level (approximately 13 stack
diameters above the breeching plane inlet) in
the stack at the Allen Steam Plant Unit No. 2 298
A-47 The effect of angular orientation on the accuracy
of four different 2 ray sampling strategies for
the velocity profile shown in Figure A-46 300
A-48 The effect of angular orientation on the accuracy
of four different 4 ray sampling strategies for
the velocity profile shown in Figure A-46 301
A-49 The effect of angular orientation on the accuracy
of four different 8 ray sampling strategies for
the velocity profile shown in Figure A-46 302
A-50 Velocity distribution simulating that measured at
the 30.5 meter (100 ft) level (approximately four
stack diameters above the breeching inlet plane)
in the stack (dia=5.3 m) at the Colbert Steam
Plant Unit No. 1 303
A-51 The effect of angular orientation on the accuracy
of four different 2 ray sampling strategies for
the velocity profile shown in Figure A-50 305
xvii
-------
FIGURES (Cont.)
Number page
A-52 The effect of angular orientation on the accuracy
of four different 4 ray sampling strategies for
the velocity profile shown in Figure A-50 306
A-53 The effect of angular orientation on the accuracy
of.four different 8 ray sampling strategies for
the velocity profile shown in Figure A-50 307
B-l Isokinetic source-sampling probes furnished
by EPA 328
B-2 Commercially available isokinetic source-sampling
probes 329
B-3 Commercially available isokinetic source-sampling
probes 330
B-4 Pitch and yaw probe misalignment 331
B-5 Photographs of s-tubes tested 332
B-6 Medicine Lake probe calibration setup 333
B-7 Typical velocity profile at probe test station
(Channel 8) 334
B-8 EPA model ductwork at the FluiDyne Energy Laboratory 335
B-9 Velocity distribution with no screens at the
probe calibration test station 336
B-10 Velocity error with yaw angle, (3/16" s-tube) 337
B-ll Velocity error with yaw angle, (3/8" s-tube) 338
B-12 Velocity error with pitch angle (3/16" s-tube) 339
B-13 Velocity error with pitch angle, (3/8" s-tube) 340
B-14 Photographs of probe interference test setup 341
B-15 Isokinetic source sampling probe geometry 342
B-16 Sample probe tip/s-tube orifice longitudinal offset 343
B-17 S-tube (alone) coefficient vs. true velocity 344
B-18 S-tube coefficient vs. true velocity for five
different s-tube/sample probe tip spacings 345
xviii
-------
FIGURES (Cont.)
Number Page
B-19 Velocity error vs. spacer thickness 346
B-20 Velocity error versus sample probe tip/s-tube
orifice longitudinal offset 347
xix
-------
TABLES
Number Page
1 Description of TVA power installations and ducting
configurations 12
2 Location of traverse points in circular stacks 33
3 Station locations and weights for averaging 35
4 Calculated values of the effect of model scale on
simulation 41
5 Particulate concentrations, total emission, velocity
and total flow rate for Runs 1 and 2 of the 1/10
scale model tests 52
6 Effectiveness of various 2-ray sampling strategies
in round ducts 152
7 Effectiveness of various 4-ray sampling strategies
in round ducts 153
8 Effectiveness of various 8-ray sampling strategies
in round ducts 154
9 Comparison of various 16-point sampling strategies
in round ducts 155
xx
-------
ACKNOWLEDGEMENTS
The authors wish to express appreciation to the
project officers, Mr. Charles Rodes, Dr. Fred Jaye, Dr.
Kenneth Knapp, and Mr. Thomas E. Ward for their guidance
and expertise.
Special thanks are also given to Northern States
Power Company and the Tennessee Valley Authority for
allowing measurements programs to be conducted at their
plants.
The assistance and cooperation of numerous Mid-
western utilities and stack sampling organizations who
responded to several surveys conducted during the course
of this study are gratefully acknowledged.
xxi
-------
I. INTRODUCTION
The basic purpose of this program was to determine optimum
sampling strategies for the measurement of particulate emissions
from particle-laden gas streams in the ducting of large C >1QQ
MW) combustion operations. This is equivalent to minimizing the
effort required to obtain the particulate flow rate within a
given accuracy. The most numerous and most important applica-
tions of the results of this study will be to large electric
power plant emission measurements. The need for this program
arose because of the nonuniform and non-steady character of the
flow fields in which power plant emission measurements can be
most easily made due to accessibility of potential sampling
locations. Flow disturbances are introduced by I. D. fans,
internal bracing, non-straight duct components, such as bends,
expansions, contractions, stack entries, and combinations of
these components. The flow fields in these large installations
are especially complicated because the flow field at a given
survey station is usually influenced by more than one upstream
disturbance source. Bends are closely coupled, hence in many
situations, the only reasonable sampling locations are only a
few equivalent duct diameters or less from a flow disturbance.
Large scale turbulence is generally present. This adds to
the unsteadiness in the flow, but it also contributes to the
rapid decay of spatial nonuniformities.
The prime objectives of this work included:
(1) Determination of the effects of duct and stack geometries
on velocity profiles and particulate distributions.
(2) Establishment of sampling strategy guidelines which des-
cribe the expected flow profiles and the magnitude of
errors caused by proximity to mechanical disturbances. This
includes a study of the effects that skewed velocity and
particulate concentration profiles have on the required
sampling strategy.
(3) Identification of geometric flow configurations in large
power plants which generate cyclonic flow.
(4) Studying the effect of determining emission levels at part-
load operation and scaling such measurements to the full-
load case.
(5) Determination of the impact of the results of this work on
EPA Methods 1 and 5.
The program began with a review of pertinent literature related
to:
- 1 -
-------
(1) Similarity problems with gas-particle mixtures.
(2) Internal flow in ducts.
(3) Existing emission survey data.
(4) Large modern day power plant ducting arrangements.
(5) Emission measurements at part-load operation.
(6) Flow angularity in large power plant ducting.
The literature survey was followed by a laboratory scale
modeling, verified by full-scale field testing. The laboratory
test model was a 1/10 scale replica of the precipitator and stack
breeching of the Northern States Power Company's Allen S. King
Station. As the program proceeded, emphasis was shifted away from
the laboratory modeling and towards more full-scale field testing.
As part of the study, attention was given to identifying
geometric configurations in large power plants which might
generate cyclonic flow. Special emphasis was given to flow
angularity in large power plant stacks, flow profiles downstream
of induced draft fans, and aerodynamic interference effects
on velocity measurements with an S-tube in EPA Methods 2 and 5.
This investigation also included a brief field test program
to study the effect of measuring volumetric flow rates, particulate
concentrations, and total emission levels under part-load oper-
ation and scaling such measurements to the full-load case.
Particular attention was given to the transient behavior during
the transition from low-load to full-load and vice versa.
A least-squares curve fitting routine was devised which
permitted field test data, both velocity and particulate concen-
tration profiles, to be fitted by polynomial equations. Given
sampling schemes were tested by comparing the overall emission
level determined from calculated values (using the defining
polynomials) of the particulate concentration and velocity
at the selected sample sites with the actual emission level
determined by integrating the polynomials over the cross-sectional
area of the duct. This computerized evaluation technique allowed
us to quickly and accurately test large numbers of sampling
strategies on the particulate and velocity distributions
obtained from the model tests, field test programs, from other
investigations, and from the literature. The search for optimum
sampling strategies was focused on correlating the total number
and location of the test points with sampling performance. With
rectangular ducts, various equal-area sampling matrices were
tested. For round ducts, the effects of the number of rays (test
ports in stack), the number of sample points per ray, and the
- 2 -
-------
angular orientation of the sampling rays with respect to the
velocity profile were analyzed. The purpose of this detailed
analysis was to reveal methods and techniques for improving
sampling accuracy in stratified flow fields, while simultaneously
reducing the required sampling time and effort.
- 3 -
-------
II. CONCLUSIONS AND RECOMMENDATIONS
The results and findings from this investigation lead to
the following conclusions:
(1) Duct components (contractions, bends, stack entries,
turning vanes, etc.) in large modern day power plants are
designed and constructed to maintain relatively uniform
flow in order to minimize costly draft losses and maximize
particulate control device performance. These components
tend to preserve flow and particulate gradients generated
upstream.
(2) Today's high efficiency emission control devices, required
on large power plants to assure compliance with emission
standards, preferentially remove the larger particles,
leaving the smaller particles which more closely follow
the streamlines.
(3) As a result of conclusions 1 and 2, flow in these large
installations is relatively uniform (spatially) and sus-
pended particulate trajectories typically coincide with
the flow streamlines. Gravitational and inertial force
effects are small because of the large dimensions, limited
residence time, and fine aerosol.
(4) Field measurements have shown that control devices and fans
can and often do generate large nonuniformities in the
velocity and particulate concentration distributions and
these nonuniformities persist as the flow passes through
the downstream ducting.
(5) As a result, moving the sampling station further downstream
from a supposed disturbance (bend, contraction, etc.),
within the limited distances available in modern power
plant breeching arrangements (maximum length of straight
sections are typically less than 8 equivalent duct diameters)
will not necessarily improve the flow and particulate
concentration uniformity. This is a change from the condi-
tions that existed in earlier plants, where control
devices were less efficient and less use was made of flow
correction devices, such as turning vanes. In those cases,
the particulate size was large, hence gravitational and
inertial effects were significant. Separation of flow from
the inner walls of bends was common. Moving the sampling
ports downstream, and thereby allowing sufficient time
for turbulent mixing to restore uniformity, was an effective
and acceptable procedure for improving sampling accuracy.
- 4 '
-------
(6) A computerized sampling strategy was applied to 32 velocity
and particulate profiles that include some of the extremely
irregular distributions in large power plants.. Sampling
points are at the centroids of equal matrix areas. Based on
the results of the study, it was found tnat:
a. With certain exceptions, equal-area rectangular
duct sampling arrays with 12 or less traverse
points were found to have large erratic and very
unpredictable error levels. The exceptions are
that all matrices with 9 or greater points
which contained three or more rows of sampling
points in each direction had errors less than 5%.
b. Equal-area rectangular duct sampling matrices
with 12-24 points had errors that were generally
less than 2%. Sampling matrices with the most
balanced number of sampling points vertically and
horizontally (aspect parameter = 1.0) were found
to provide the maximum accuracy for a given
number of sample points. All sampling matrices
with at least three sampling points in each
direction had errors less than 5%. In some
sampling matrices of up to 16 points with only
two points in one direction, 2x6, 2x7, and
2x8, however, the error exceeded 5%.
c. Generally, the error decreased with an increase
in the number of points in the range of 24 to 36
points. However, in many cases, a 50% increase
in sample size C32 to 48 points) yielded only a
small decrease in error (1% to 0.5%).
d. Six sets of stack profile measurements which
provide a good representation of extreme condi-
tions existing in large power plant stacks, were
analyzed. Velocity gradients were found to be
more dependent on radius than on angular
orientation. Sampling schemes with an equal num-
ber of rays and points per ray gave the lowest
error. For example, with a four ra.y-16 point
survey, the largest error in total volumetric
flow was only 1.5%.
(7) Severe cyclonic flow was not found to exist in large power
plants ( >100 MW). Field test data revealed little varia-
tion in flow angularity downstream of elbows, contractions,
and stack entries, even those with diametrically opposed
and other complicated multiple-inlet breeching arrange-
ments. - 5 _
-------
Significant dual or multiple cyclonic flow was not found,
even in stacks which had "apparent" dual visible plumes.
The dual plumes from a single stack were determined
to he due to the effect of cross winds on the plumes.
(8) Field test data from the part load measurements at a 600 MW
power plant revealed that even throughout a transient load
period a close correlation exists between volumetric flow
rate and plant electrical power output. The flow rate
was proportional to electrical power within a few percent.
However, the particulate concentration, and therefore,
emission level, was not proportional to the electrical
output during load changes and for a time interval follow-
ing load changes.
The particulate concentration and emission level fluctu-
ated during the load change. After a return to full
plant output, the particulate concentration and emission
level remained above the eventual steady state value for
over one hour. This phenomenon needs to be investigated
further.
RECOMMENDATION:
When measuring total volumetric flow rate and/or average
particulate concentration in large (>100MW) rectangular power
plant ducting, sampling at the centroidsof equal area elements
is recommended. Based on the results of this study, for N,
by N2 sampling matrices, NI and N2 should be greater than or
equal to 3 to assure an accuracy of 5% or greater than or equal
to 4 to assure an accuracy of 2%. NI and N2 are the number of
rows of sampling points across the width and length of the duct
respectively.
-------
III. LITERATURE AND PERSONAL CONTACT SURVEY
A. Internal Duct Flow (Velocity and Particulate Concentration
Profiles )~~
Over the years internal gas flow technology has received
particular attention because of its relevance to turbomachines,
fluid handling devices, and other mechanical and aeronautical
engineering applications. Boundary-layer behavior, flow separa-
tion and reattachment, secondary flow, turbulence, eddy form-
ation and other centrifugal and frictional effects have all
been studied and reported in detail (see References 1-8).
Historically, in prototype industrial gas ducting, spatial,
structural, and other economic related design considerations
frequently took precedence over the quality of the gas flow.
As a result, poor temperature and gas flow distributions often
induced unnecessarily large pressure losses, unsatisfactory
control device performance, and excessive dust fallout and corro-
sion problems. As power plant systems have become larger and
emission control regulations more stringent, design of efficient
and effective duct work layouts which minimize flow stratifica-
tion has become an important design factor. Very uniform and
steady temperature, velocity, and particulate concentration
profiles must be maintained at the inlet of electrostatic pre-
cipitators and other expensive emission control devices, if
optimum collection efficiency is to be obtained. Furthermore,
excessive draft losses in these large systems can be extremely
expensive. For these reasons, velocity distributions and the
necessary corrective devices required to minimize stratification
in large industrial ducting have been (recently) studied exten-
sively. References 8-23 describe typical flow patterns in
elbows and other transitions and design recommendations for eff-
ective use of guide vanes and diffuser elements, such as perfor-
ated plates and grids. These papers provide qualitative indi-
cations of the effectiveness of various duct work arrangements.
Unfortunately, as the findings in Section III-B indicate, the
duct work arrangements in today's large power plants are often
too complicated to be analyzed solely on a component by com-
ponent basis.
To assist in the recognition and elimination of possible flow
separation, large-scale turbulence, high velocity jets, and flow
pulsations, designers are frequently turning to three-dimensional
cold flow model studies of the flow patterns in proposed plant
flue systems. As References 9-23 indicate, this approach, espec-
ially for precipitators, stack entrance breeching arrangements,
and other very high cost installations, has become very popular
in recent years. Flow visualization studies involving smoke trace
-------
and tuft observations, cork erosion and dust dropout tests,
and direct measurements with miniature velocity transducers
(S-tubes and thermal anemometers} are employed to determine
velocity profiles and patterns. In general, relatively good
correlation between model studies and full-scale results have
been reported.
Considerably less evident in the literature is data on
particulate concentration (emission) distributions. Quite
possibly this is because velocity data is much easier to
accumulate, but also because, for many years, some investi-
gators believed that the particulate matter was distributed in
much the same manner as the flue gas. Furthermore, for large
continuous industrial processes, such as coal fired power
plants, where particulate sampling is essentially conducted
for compliance purposes, only composite samples are typically
extracted. In the composite sample procedure, the particulate
matter obtained at individual traverse points are all collec-
ted in a single filter, hence this procedure provides little
information on concentration profiles. Measurements of local
particulate concentration levels, using individual filters at
each point, are extremely expensive and time consuming.
Hawksley, et al, (Ref. 4) stress the fact that the amount
by which the velocity and emission distributions differ is
dependent on an inertia parameter which includes particle
size, gas velocity, and bend radius. They further stress the
necessity of employing statistical analysis to show the exist-
ence of a pattern, particularly from raw data taken in full-
scale gas ducting where mass flow is variable with large
random fluctuations. Reported test data indicates that the
emisson distributions are skewed as much as, or more than, the
velocity distributions in the duct. In traversing 90° elbows
solids were found to be centrifuged towards the outside with
uniformity gradually restored by turbulent mixing in the
downstream ducting. According to these findings, "about three
to five diameters is sufficient to establish a tolerable
uniformity, although this may not be the case when the quant-
ity of grit (>76 microns) is high. When there is little grit,
e to three diameters is adequate". It is important to
ze that these measurements were made in a power plant
:m without efficient emission control devices or guide
vane baffling. As a result of the larger mean particle size
and the lack of flow correction devices, stratification would
be expected to be much more severe than in the exhaust breech-
ings of modern day installations.
Sansone CRef. 24} studied emission distributions down-
stream of a 90° bend very extensively. This study was differ-
ent from the other work reported in that it was a laboratory
setup (23.76 cm diameter round duct) with a single mechanical
disturbance (the elbow) and monodispersed particles C25 micron
.- 8 -
-------
glass beads). Sansone discovered that the emission distribu-
tion remained significantly skewed far downstream Cat least 16
diameters] of the bend, a point far beyond where the velocity
distribution recovered its uniformity. Higher emissions at
the periphery of the duct and at the far side in the plane of
the bend were reported, although these were not graphically
represented. Sansone also used statistical techniques to test
for significance from a limited set of data points.
Other references (25-27) give some graphical representa-
tion of emission distributions which reflect similar results.
Brown's data (Ref. 25) shows centrifuging of the particulate
matter towards the corners of a vertical square duct. A
cyclone probe was used for sampling, possibly biasing the data
towards the large particles. No corresponding velocity distrib-
ution was given. Reference 26 shows a severely skewed dust
loading profile accompanied by a relatively uniform velocity
distribution downstream of a right angle bend.
B. SURVEY OF POWER PLANT DUCTING
An extensive survey was made of the types of ducting con-
figurations found in modern day power plants. The sources of
information included:
a. engineering data supplied by TVA,
b. file drawings from three private power companies,
c. power station design surveys published yearly in
industry magazines,
d. advertisements and literature available from air
pollution control device manufacturers.
1. TVA Data
Study of TVA power installations of the past twenty years
shows some obvious trends in ducting configurations. Figures
1-6 are simple sketches of typical installations in a rough
chronological order. Table 1 gives additional information.
As the size of units became larger and thus more gas ducts
were needed, a plenum-manifold arrangement became popular
(Figures 3 and 4). The sampling ports are usually located on
the top of each individual duct somewhere upstream of the
entry to the plenum. This style persisted through the 196Q's
when the forced draft (pressurized furnace) design became
popular and induced draft fans disappeared (Figure 5). As
particulate emission standards became tighter, the size of
precipitator installations increased, resulting in the -utiliza-
tion of the Chevron arrangements, two versions of which are
- 9 -
-------
sketched in Figure 6. This arrangement features a sharp
contracting bend immediately downstream of the precipitator
with only one (at most} additional bend before stack entry.
In still more recent plant designs, the two unit, 2600 MW
Cumberland Station, increased precipitator capacity has result-
ed in double-decking the precipitators to save plant area.
2. File Drawings
The ductwork configurations of fifteen other units were
studied. These units represented eight different power sta-
tions and three power companies. These units have all been
put in service during the last twenty yearsf and the original
ductwork layouts generally exhibit characteristics similar to
the TVA designs of the day.
Drawings of four units at two stations of Consumers Power
(Michigan) showed very similar layouts to TVA units of the
same time periods. These units were installed between 1958
and 1965. Sketches of two typical units are shown in Figures
7 and 8.
Northern States Power Company (Minnesota) has recently
added additional precipitator capacity to older units at three
stations, inserting the new precipitators between the existing
induced draft fans and a new tall stack. These retrofit flow
paths are highly unique and feature combinations of abrupt
flow disturbances, i.e., a contracting elbow off the face of a
precipitator followed by an additional contraction close-
coupled with a mitre bend (Figure 9). The abrupt combinations
allow relatively long and simple runs of straight ducting
entering the stack.
Drawings of recently installed units at two stations of
Carolina Power and Light (North Carolina) indicate a continued
use of balanced draft designs. This means that sampling ports
are usually located somewhere between an induced draft fan
and the stack breeching (Figure 10).
3. Design Surveys
Power plant design surveys are published yearly by industry
magazines such as "Power" and "Electrical World". A review of
these surveys for the last few years reveals that the forced
draft, pressurized furnace design (with no induced draft fan in
the flow path between the dust collector and the stack entry)
was preferred for several years. The most recent trend has been
back toward balanced draft designs.
- 10 -
-------
4. Manufacturers' Literature
The trend in the last few years has heen for the jmanufact-
urers of the dust collectors to furnish more elements of the
system, including the inlet and outlet flues and associated gas
ducting. Prior to 1968, the TVA system designed, model tested
and fabricated their own flues and ductwork but since then has
made this the responsibility of the dust collector manufacturer.
Therefore, an inspection of brochures of various manufacturers
provides enough information to make some general comments:
a. The necessity of higher collection efficiencies
has resulted in fabrication of larger precipitators,
many with Chevron arrangements. Even larger precipita-
tor installations, in some cases hot precipitators,
have lead to double deck arrangements.
b. Shallow or gentle contractions downstream of dust
collectors have given way to very abrupt ones, in many
cases, combined with right angle bends, additional
contractions, and turning vanes.
c. In the newer designs, internal flow distribution aids
designed on the basis of extensive model studies are
used to overcome the flow problems created by abrupt
disturbances.
- 11 -
-------
W
t
(_3
flj
1
w
M
EH
rt?
Q
ta
CO
EH
H
2
D
2
O
H
EH
(X
H
U
CO
W
Q
D
0
fa
1"
in
i-H
H CN
in in
en en
rHrH
. «•«• ~-»
VO VO
1 1
rH rH
U
G
O W
tQ £
C O
J3 T3
O-H
O
4J
U (0
n
in ^
in rH
en vo
rH CM
••-* ,—, ,_)
y »—<
<* O
I id r^
H 4J
co A:
JH \ (U
(U (0 CU
•H -P )H
> -H U
Q) C
to D to
]j
C CN O
_r* •— rQ
O -H
h) S
U)
•P rV?
O U
3 fd
Q 4J
•P CO
MH ^* "V.
id
O fa CN
f<
Id Q rl
rH H O
id
CO ^3* r™l
n
in vo
in •«*
CN O
rH (N
^-^
m
in
en ^-*
^H *^
— O co
n)
1 CO 0)
rH X, 0) rH
w in M
C -P O C
•H -rl 4J 3
-PC rl 0) «
id D CU 5
rH XI OH
rH CN H tJ H
id ^ O -H 3
O U 12 «
PO
m w
CU -P
rl 0
3 3
Cd Q
•H
fa "3*
>M 4J 1
O MH
C
•P i id
id en in
rl CJ <
Q id
fa ^
T3 O
(U Q H
O H >
rl CU
O O J3
fa !3 O
vo
- 12 -
-------
C>vl
H
«
o
a
E-i
w
i
w
CO
CO
w
vo
I
w
Kl
H
o
CO
O
- 13 -
-------
r~
r_
n
V
FIGURE 2 KINGSTON UNITS 1-10 (TENNESSEE VALLEY AUTHORITY)
- 14 -
-------
>//
FIGURE 3
JOHN SEVIER UNITS 1-4 (TENNESSEE VALLEY AUTHORITY)
- 15 -
-------
E-i
H
«
O
ffi
EH
§
W
W
w
2
JS
w
EH
H
§
A
D
O
H
- 16 -
-------
FIGURE 5 COLBERT UNIT 5 (TENNESSEE VALLEY AUTHORITY)
- 17 -
-------
tN
CO
EH
H
25
W
w
1-1
- 18 -
-------
FIGURE 7 KARN UNITS 1 and 2 (CONSUMERS POWER CO.)
- 19
-------
FIGURE 8 CAMPBELL UNIT 2 (CONSUMERS POWER COMPANY)
- 20 -
-------
PKETCIPITATOR.
OUTLCT FLUE:
J
FLOW/
I
FIGURE 9 HIGH BRIDGE UNIT 3 (NORTHERN STATES POWER CO.)
- 21 -
-------
•etc
FIGURE 10 ROXBORO UNIT 3 (CAROLINA POWER & LIGHT)
- 22 -
-------
C. FLOW ANGULARITY IN LARGE STACKS
1. Tall Chimneys
As power plant sizes get bigger and pollution regulations
more stringent, chimneys are correspondingly growing taller.
The trend today is to design new plants and retrofit older
plants with a single 183-366 meter (600-1200 ft) stack. Accord-
ing to Reference 28, about 50 new tall chimneys are being built
each year. Tall chimneys greatly reduce ground level concentra-
tion by high altitude dispersions.
With all the flue gas from a given plant being channeled
into a single stack, the exhaust breechings and inlet duct
configurations at the base of the stack can become very compli-
cated. The diametrically opposed breeching arrangements shown
in Figures 11 and 45 are very common. The popularity of this
arrangement arises from the fact that modern power plants are
typically composed of several similarly shaped and sized parallel
system legs. Design criteria, including cost, space requirements,
and incorporation of equipment from various manufacturers, are
often responsible for other breeching configurations, however,
such as the triple inlet shown on Figure 13 and the double inlet
design shown in Figure 14, which is not diametrically opposed.
In older systems, where new precipitators have been installed
and the outlet gas streams from several different boilers (often
of various sizes and design) have been coupled to a single new
stack, the inlet ducting can become even more complicated.
Retrofitting around existing systems can necessitate feeding
ducts of various sizes and varying flow rates into the chimney
base from several different levels and directions. It's not
always possible to manifold the effluents from several boilers
into a common duct. Since flow rates are not necessarily
balanced and design of effective baffling in the chimney base
is more difficult, predicting flow field behavior becomes more
complex.
2. Cold-Flow Model Studies
Although very little data is available defining flow angu-
larity in large power plant stacks, the literature reveals an
awareness of poor flow conditions occurring in the lower sections
of chimneys as the "result of abrupt expansion and interaction
between the flows from opposing jets" (see Reference 29). Flow
separation, with the accompanying secondary flow vortex forma-
tion, jet interactions, and other undesirable flow patterns in
the stack entrance section can produce erratic and unpredictable
- 23 -
-------
45'
-W = 3m
= 9m
H = 10.3m
FIGUEIE 11 Multiple Breeching Arrangement in the Colbert
Unit No. 5 Steam Plant
- 24 -
-------
FIGURE 12 TRIPLE INLET BREECHING ARRANGEMENT
IN THE CUMBERLAND STEAM PLANT
- 25 -
-------
Vertical Deflector Baffle
Precipitator
Precipitator
Precipitator Inlet Duct
Precipitator Inlet Duct
•x%>'^
-,
-------
flow in stacks, large draft loss, fly ash deposition and
pressure fluctuations. The adverse effects of large stack
entrance losses on plant efficiency and performance (added fan
requirements, I. D. fan vibrations, etc. ) have forced designers
to develop effective guide vane structures positioned in the base
of the stack and/or inlet ducting. The lack of data on which
to base designs has forced designers to cold flow modeling.
TVA has used cold-flow air model studies (References 21-23)
to evaluate flow conditions in the lower sections of several
of their chimneys. These models were used in designing vaned
deflectors to be positioned in the chimney bases to improve
flow conditions. Model studies, using smoke film and other flow
visualization techniques, of the Colbert Unit No. 5 stack with its
diametrically opposed breeching arrangement (see Figure 11)
without any baffling showed a four-cell vortex pattern. The energy
dissipation from these vortices was responsible for the large
observed draft loss. Model testing of the Bull Run chimney,
see Figure 14, showed a spinning action (single vortex)
accompanied by random pressure pulsations. Very different flow
profiles were observed in the Bull Run Model (spinning type flow)
and the Colbert Model (4-vertical eddies) in spite of the fact
that each had similarly shaped and diametrically opposed breeching
ducts. Although the velocity and momentum of the entering gas
streams, the chimney, and the "breeching duct geometric para-
meters" were found to influence the existing flow patterns, the
widths and heights of the breeching duct relative to the chimney
diameter were proposed to be the most dominant factor.
In all of the four multiple duct arrangements described
in References 21-23 and 29, including two with diametrically
opposed breeching ducts, one with three ducts entering at 60°
to each other, one with two ducts entering at 60° to each
other and one with two ducts entering 90° apart, guide vane
structures were designed which allegedly eliminated the cyclonic
action and, thus, reduced the chimney draft losses.
Other cold flow model studies of the lower sections of chim-
neys have indicated similar spiral action. According to Refer-
ence 30, smoke tracers, in an unbaffled double inlet stack
showed the flow profile oscillating between the 4-vortex and
single vortex patterns. Accompanying these changes were severe
flow and pressure surges.
In Reference 20, Gilbert describes the use of flow modeling
techniques to obtain a minimum loss design for the stack entrance
section of a 213 meter (700 ft) power plant chimney. In this
specific applicatipn, two ducts entering the chimney base 149°
apart and at different elevations carry the flue gas from eight
boiler units. A 90° six-sided stack bottom configuration was
designed which reportedly"eliminated the sharp upper bend and
- 27 -
-------
W = 3.6m
I
-V
l~
-v^
D = 15m
FIGURE 14 MULTIPLE BREECHING ARRANGEMENT IN BULL
RUN UNIT NO. 1 STEAM PLANT
- 28 -
-------
separated regions at the side of the stack where the vortex
flows were observed with the smoke probe."
As the studies described above indicate, guide vane
structures in the stack base have become an integral part of
today's large power plant installations. Proper and efficient
plant operation dictates good aerodynamic design and construc-
tion of the stack entrance region.
3. Preliminary Survey
In an attempt to establish flow angularity and the fre-
quency of existence of cyclonic flow in large ( >100 MW) power
plant chimneys, the literature was reviewed and telephone contacts
were made with several individuals actively engaged in stack
sampling. Those contacted were asked if they frequently encount-
ered cyclonic flow in large power plant stacks and, if so, under
what circumstances, i.e., breeching arrangements, including
single or multiple tangential connected to the stack ,unbalanced
ducts, part-load operation, etc. Qualitative estimates of the
flow angularity and a description of the test procedures used
in non-streamlined flow fields were also sought.
In general, the results of this search indicated that little
qualitative data is available defining flow angularity in stacks.
The lack of test ports in many stacks is responsible, in part,
for the lack of factual information.
More than twenty engineers concerned either with stack
sampling or efficient effluent gas flow in tall stacks were
contacted by telephone in an informal survey about cyclonic flow.
The majority considered cyclonic flow to be an important concern.
Almost invariably, reference was made to observed double vortex
plume profiles. Various opinions as to the causes of these double
plumes or vortices issuing from the tall stacks included multiple
breeching feeds, I. D. fans followed by short breechings, insuf-
ficient guide vane design and construction, and other reasons.
Calculating total volumetric flow from plant input data rather
than S-tube measurements was indicated to be a preferred procedure .
However, this may be due more to the cost of the S-tube determin-
ation procedure than to the accuracies of the calculated total flow
versus the S-tube experimental total flow.
In contrast to the above-described observations, several
individuals indicated that cyclonic action in large power plants
did not seem particularly serious. Specific reference was made
to an extensive test program in a large power plant with a mul-
tiple breeching arrangement in which the total volumetric flow
calculated from boiler input data, from a 22-point survey taken
upstream of the precipitator inlet, and from a survey taken in
the stack at the 73.1 meter (240 ft) level, all agreed within 5%.
- 29 -
-------
Invariably, foundry cupolas and small stacks were cited as
sources where cyclonic flow was very troublesome.
When confronted with these angular or cyclonic flow poss-
ibilities, the most commonly mentioned test procedures included
aligning the S-tube in the direction midway between the
two S-tube maximum pressure differential directions (see Figure
B-ll) or the two S-tube zeroA P directions. An alternate proced-
ure, when visibility is permitted through the test ports, involves
positioning a probe with a fine string attached to the end at the
desired test point. The S-tube is then aligned in the direction
assumed by the string.
4. Preliminary Conclusions
The preliminary investigaton of cyclonic flow in chimneys
suggested that:
a. Little qualitative data is available defining
flow angularity in large ( > 100 MW) power plant stacks.
b. Data is available to support contentions that flow
is frequently very disturbed in small stacks and
cupolas. (Regions with reverse flow often observed.)
c. Cyclonic flow is thought by many stack sampling engineers
to be especially prevalent in large power plant stacks
having multiple inlet arrangements. Most of these
judgments were based on visual plume observations.
- 30 -
-------
D. SAMPLING METHODOLOGIES
To assure maximum efficiency in determining the total
volumetric flow rate and average particulate concentration in
large combustion source ducting, careful attention must be paid
to the number of traverse points and the location of the sampling
site. Over the years, various guidelines for extracting a
representative sample for emission level determination have
been established. For example, Hawksley, et al (1961), in
addition to describing duct configurations and other conditions
which frequently lead to stratification, discuss the accuracy
associated with emission measurements in these nonuniform flow
fields.
1. EPA Test Methods
EPA Method 1 recommends that sampling sites be selected at
least eight equivalent duct diameters downstream and two diameters
upstream from any flow disturbance. Satisfying this requirement
permits the minimum number of twelve (12) sample points. When
this is impractical, a convenient sample location is to be
selected, and Figure 15 used to determine the minimum number of
traverse points required.
For rectangular ducts, the cross section is to be divided
into as many equal rectangular areas as traverse points, such
that the ratio of the length to the width of the elemental
areas is between one and two. With this equal-area method,
sometimes referred to as the tangential method, traverse points
are located at the centroid of each subdivision (see Figure
16) .
The test procedure also specifies that round ducts
be divided into zones of equal area, and the measurements be
made at radii which further divide the zones into two equal
area parts. Traverse points are to be located on at least two
diameters according to Figure 16 and Table 2. The traverse
axes are to divide the stack cross section into equal parts.
2. ASME POWER TEST CODES
ASME PTC-27 (Ref. 26) again recommends the equal area
sampling approach, with traverse points located at the center
of zone areas. It is further recommended that where the flow
is uniform, i.e., where the range of velocities does not exceed
2 to lf from twelve to twenty points can be used for large
ducts (exceeding 2.3 square meters in area) and from eight to
twelve points for small ducts. Where high velocity differentials
or extreme turbulence of stream flow are encountered, this
procedure proposes that it is necessary to double and sometimes
to triple the number of points required in order to establish
true conditions.
- 31 -
-------
Q>
O
C
(0
X)
M
a
•P
w
•H
Q
(0
a)
M
-p
w
s-
(Q
M
0)
-P
8
«j
•H
Q
+J
U
U-i
O
Q)
•i
3
Q
O
W
cn
EH
S
H
O
w
en
B
O
§
D
5
«~(
H
D
O
n
W
u
w
w
EM «
jo aaquinN
- 32 -
-------
CN
Si
III
S "I1
m g t'
5 i !li I
ill
ej »n r- u» «o en
«^ c\ eo «r (o n
r— ou»o*^i»JOif>oeo«— t*i n .-^ <*> ^.o f» *o c 5
*— p-r>4r-i<*>c*)*0u>r-*i^ r j c~ tn o> o» «\
*— in o»
»«•» o •* r>
- C3l
•::
O
0-
o
o
e
c
0
_ .
0
0
o
o
._____.
o
^3
'> w
w
o J=
« y
*j ro
M Q>
»_ *•»
« O
g,H
= &
ra •-•
•^ c
o a
2 o
"5 «
c "
.2 c
o S
0) 0.
C/) Q)
w i?
" <5
2 >
o 3
O4 O
•"" tn
0 .
O
H
U. fO
.?
-------
Additional sampling schemes for measuring total volumetric
flow rate are described in ASME PTC-19.5.3. These include the
Newton-Cotes, Chebyshev, and Gauss methods presented in Table
3. The relative effectiveness of these different procedures
in evaluating bulk velocity when applied to three different
axisymmetric linear-velocity profiles are shown'in Figure 17.
3. ASTM Methods
The American Society for Testing and Materials (Reference
33) also recommends the centroid-of-equal-area method for
determining the average velocity in stacks. Figure 18 is to be
used to determine the minimum number of traverse points when
sampling at least eight equivalent duct diameters downstream
and two equivalent diameters upstream of the nearest flow
disturbance. When the sampling station is only four to six
diameters (straight duct) downstream of the last disturbance,
the number of sampling points is to be doubled. When even
still closer to the last disturbance, it is recommended that
"each case will have to be determined on its own merits in the
field".
4. British Standards
The British Standards (Ref. 34) traverse plan for rectangular
ducts is illustrated in Figure 19. This procedure requires
that the test cross section be divided into a minimum of 16
areas/ with multiple sampling in the corner and wall zones.
5. Miscellaneous Techniques
Numerous other sampling procedures, including among
others the log-linear, Aichelen's, and the equal area circular
ring analog (see Figure 20) methods have also been proposed.
Unfortunately, practical considerations such as cost of
installing ports and associated scaffolding, comfort and safety
of the test crew, availability of electrical and other services,
internal obstacles (bracing, etc), adequate room for probe
insertion and extraction, and still other details such as flue
gas velocity, temperature and ducting configurations frequently
take precedence over other theoretical requirements. The only
convenient sampling sites are often close to mechanical disturb-
ances, in regions where the flow might be assumed to be non-
uniform.
- 34 -
-------
TABLE -3
STATIOM LOCATIONS AND WZICHTS ?0« AVE8ACIN6
Averaging for linear Interval 0 _ 11 I
Averaging In a circular duct. In Interval 0 £ r i 1
CF
3TXTIOI3
n
2
3
L
S
6
7
S
9
10
n
hiTHOD
CBnnoili OK 1\IV.L A/ii Ao
X
0.2500
.7500
9.1667
!S)33
0.1250
.3750
.6250
.8750
,?rrv>
0.»833
.250O
.1)167
.'7500
.9167
0.071U
!3571
.5OOO
.61)29
.7R57
0.0' P)
.1375
.3125
!^125
.9375
'.27 78
.3889
.5000
.6111
.7222
.8333
O,n5
.15
.25
.35
.1.5
.55
.'?
.75
X
r
[066o
o.toez
.7071
.912?
.'6121,
\X''AL /liFjUr,
NEWTotl-CWES
X
(b)
0
1
(o)
0
o.S
1
0
0.3333
'l
0
0.25
.50
.75
1
0
O.J
!s
i
0
0.1667
.3333
!6V,7
.3333
1
0
0.11,29
.2857
.'?H3
.0571
1
0
0.1250
.2500
.3750
.5000
.6250
.7500
.6751
1
0
0.1111
.222?
.3333
. 1)1.1, Ij
.5556
.'?77fl
.BH69
1
'
r
0
x
(c)
0
0.7071
1
(d)
0
0.577fl
.8165
1
0
.7"71
1
0
O.U:72
.6325
1
0
O.liOS?
17171
.6165
' 1
0
0.3780
.53a5
17559
.SL52
.9258
1
0
0.3534
.5000
.61214
.7071
.7906
.£6*0
.9351)
1
0
0.3333
!577H
!3165
.8t'19
' 1
r
w
(b)
1/2
(c)
0.1667
.fVV-7
(d)
0.1250
.37.'0
.3750
.1250
CKntsia^
X
(e)
o.au
.735T
0.1U6),
.iCCC
,£536
0.10J7
.JtOVZ
.59^3
0.0778 (0.0838
.1333
.3t56
.0/73
0.11.N1
.0321
.OJ21
O.OI..-5
.2070
.(1766
.1730
. 1730
.07^6
.2070
.01,35
0.0369
.2077
-.0327
.3702
-.1601
.3702
-.0327
.2077
.031.9
0.0319
.V,57
.?159
.'o61)S
,?159
.0171
.17S.7
.0319
w
KM ro-. ana
.5000
19162
i2r'07
17113
.9331
O.Codl
!3331
.5000
.'?6I,8
(O
0.0513
.i'036
!5513
,7036
O.OUu
.lyj'j
.2356
.5000
[9553
(r)
o.oia?
.1561)
.2500
!li?81
!6?61,
.7500
X
r
O.LS97
.8831
0.3327
.7971
,9239
0.3203
.6362
.769?
.91)71
0.28-/1
!yS7Z
!S373
ieicb
!7j/i
..'13*
.87-5
.9705
(I)
il.513
!7L25
.8383
.892li
.971)0
0.2103
.61)50
.7071
.7642
!9776
(r)
0.101,6
isooo
.5862
.6763
.7361
.3102
.ttffl
'.WW
r
"
IX
V3
lA
VS
1/6
I//
r»t?rit» of
t^Uil Ufiiftlt
(b) Trapc told*! relt
(c) P«rls^•clllc rule
(biapoon'a ivlo)
(d) Ih«c-.«ighth« j-ul*
(o) 0.2 - 0.8 rul«
(f) Two lj-at.ii.lcn
Inters *1*
(.<;} Two S-»t«tton
lnt»rr«d»
REFERENCE 35
- 35 -
-------
0
1 1 1 1 N
.2 .4 .6 .B
RADIUS, r
(o) VELOCITY OliffCBUTlONS
CEfJlROID
OF rQUAL AftEAS
HELTON-COTE5
CHEBYSHEF *"
GAUSS
1 0
OlSTHiSUTION
I
,-0.26
-ocbTl
-000,
-000,
II
4-1.16
-020
•4002
-dob,
I III
40 n
-ooa
+001
'-000,
It; LRfOR. %. IN t-POINT
FIGURE 17 ERROR IN 4-POINT AVERAGING
OF SOME ARBITRARY AXIALLY-SYMMETRIC
VELOCITY DISTRIBUTIONS .
(REFERENCE 35)
- 36 -
-------
MINIMUM NUMBER OF MEASUREMENTS
FOR RECTANGULAR SAMPUNG-SITES
Cross sectional area
of sampHng-site, ft^
Less than 2
.2 to 25
Greater than 25
Number of
measurements
4
12
20
MINIMUM NUMBER AND LOCATION OF MEASUREMENTS
BASED UPON SAMPLING ALONG TWO PERPENDICULAR
DIAMETERS OF A CIRCULAR DUCT
SftmpUni
• it*
d.AinMr r
lnci.»»
1
>2
14
14
IB
IB
I!}
21
26
I!
30
3?
36
41
54
GO
£6
n
a
9
12
131
"^"l™
• r*A
Minimum
number a
mea»
urfrnrnts
i
i
i
i
i
i
i
i
i
2
2
2
2
2
2
j
2
, ,
No I
/J
/
/
/
/
/
/
/
/
/
('8
14
It
n
it
M
n
/t
.-.
N".l
1
1 J/4
2
2 !/S
? 1/fl
2 VS
2 l/«
2 3/8
2 3/4
1
2 1/2
2 j '8
3 l/l
3 3/4
4 i/a
4 i/)
4 3 ,'4
b 5/d
6 3/d
a
8 ife
No. 3
4
3 in
4 1/3
4 III
1 3ft
* 1/4
3 7/1 _j
4 1/4
5
i J/8
4 3/8
4 ;/8
S 5/8
C 3/8
7 i/a
7 3/4
S 1/2
9 7/B
0 5/S
1 3/8
4 I/B
S 3/4
DUUnc
No. 4
7 lit
i in
j 7 /a
10 l/i
n :,<
U 3/4
e 3/a
7 i/a
8 3/1
9
6 3/4
7 \)\
t 1/2
9 1/2
10 S/8
ii s/a
12 3/4
14 T/S
17
21 1/4
33 3/0
(l urn i
No. ^
0 1/4
2
2 3/a
35/S
5 3/3
3 V3
4 7/3
7 5/S
9
0 1/8
1
r~
2
3 I/I
S
6 1/2
a
i
2<
30
33
.mplin^ ,
No 0
U '/
13 3/
14 31
15 )/
17 \i
it. i ;
'7 3/*
21
22 5/S
t» i/S
2 1
J7
iy 1/2
£1 l/
25 J/
2S 1,
28 3/
1*
42 W»
urt to si
Hft, 1
17 7/8
H V«
23 t/4
25
i, 1/4
24 */i
31
34 t/2
38 3/1
•
62
77 J/8
idling |:c
No. B
13 3/E.
?l 1/4
'S 1/8
:: j/i
?i 4/a
^7 3/C
30 3/4
3«
40 1/2
IS
-19 1/2
S4
*i3
7J
5'J
I'll, tlichf
NO e
.7 1/2
29 3/6
JT
J> J/2
44 III
40 J/S
54 V#
ifl 1/4
'^ l/»
79
d8 3/4
•
No. 10
;:) 1/4
11 1/4
3S 1/8
42 J/D
4i s/e
S2 7 /«
ia 1/1
«3 Ml
73 1/8
«4 i,'3
lu 7/3
No. 11
<* 3/4
10 1/4
55 7 fk
Cl 1/2
fl7 1/4
It 3/8
t» •>/«
112
N->. ta
47
12 7/1
!.B 3/4
64 i/»
70 1/J
5 3/«
S
11 1/2
FIGURE 18
MINIMUM NUMBER OF MEASUREMENTS FOR
RECTANGULAR SAMPLING SITES (ASTM D-3154)
REFERENCE 33
- 37 -
-------
J
FIGURE 19 British Standards Traverse Plan (Reference 34 ).
FIGURE 20 Circular Ring Analog Plan (Reference 34 )
- A2 m A3 = V
- 38 -
-------
IV. MODELING
Flow model studies have been used for years to design effi-
cient ductwork systems and minimize dust fall-out. For the most
part, these studies have concentrated on modeling the gas flow
without considering two-phase gas/particle flows. However, there
are treatments of the parameters affecting the similarity of such
a two-phase flow in the literature. One TVA paper (Ref. 36) pro-
posed a trial-and-error method of achieving kinematic similitude
in two-phase flow in order to study dust accumulations in duct-
work.
A. Similarity Parameters
For the motion of two aerosol systems to be similar, the
flow boundary geometry, the fluid flow, and the particle trajec-
tories must all be similar. Geometrical similarity can be
obtained by proportionally scaling all model dimensions. For
satisfying the motion related similitude, equality of the Reynolds
number, Stokes number, and Froude number is required. These
parameters are defined as:
Reynolds number =
UL
Inertia force
Viscous force
where:
U
L
fluid density
characteristic fluid velocity for
system
characteristic system dimension
(diameter of duct, etc.)
fluid viscosity
Stokes Number
Stop distance of the particle
Characteristic dimension
The Stokes number is a valid criterion for similarity only
where Stokes law of resistance applies, i.e.,
'D
24
R
TJ
where e is the particle Reynolds number based on the relative
particlepto fluid velocity. For Stokes flow and pf «p
_ p p _ Particle resistance
Stokes number
18 yf L
- 39
Inertia
-------
where: D = particle diameter
p = particle density
CD = drag coefficient
Froude number = -£ = Inertia force
Lg Gravity force
where: g = acceleration of gravity.
The Froude number is important when gravity is being considered.
These parameters, along with a treatise on similarity in the
mechanics of aerosols, are developed and discussed by Fuchs (Ref
37).
B. Effect of Model Scale on Simulation
Modeling of particle-laden gas flows constitutes a very
difficult problem in fluid dynamic similarity. Satisfaction of
all of the similarity parameters requires a full-scale model.
Reduced model sizes inevitably mean compromise with one or more
of these parameters. Furthermore, models may not adequately
simulate the level and scale of turbulence that exists in the
prototype.
For many modeling purposes, although attainment of a high
degree of geometric similarity is important, operation at reduced
Reynolds number is often acceptable when a flow system is turbu-
lent in nature, since large changes in Reynolds number exert much
smaller changes on the flow fields.
The original modeling goal in this work was to try for exact
particle parameter similarity (i.e., match prototype Stokes and
Froude number values). The largest practical model size con-
sistent with reasonable costs was used to maximize Reynolds number
and hence to simulate flow fields as realistically as possible.
To examine the effect of model scale on simulation, values for
several cases were calculated (see Table 4 and Figure 21). The
tabulated parameters assume prototype and model gas temperatures
of 350°F and 80°F, respectively. In order to match prototype
Froude number, the velocity in the model must be reduced, i.e.,
— =
U
P
Then, in order to match the prototype Stokes number, the model
particle diameters must also be reduced in accordance with:
- 40 -
-------
(DP>P VUM LP VM M
(pf)p
1.503
1.503
1.503
1.503
(Re)p
(Re)M
44.2
15.7
5.54
1.40
C. Prototype Test Site Selection
Power plant ducting is characterized by an extremely large
number of variations of a relatively few geometric internal flow
shapes. These shapes are bends, contractions, and expansions.
Bends vary primarily in cross-sectional shape, radius of curva-
ture, and turning angle. Contractions vary in cross-sectional
shape, contraction angle, and contraction ratio or contraction
length. Expansions vary in cross-sectional shape, expansion
angle, and expansion ratio or expansion length. Finally, power
plant ducting is frequently made up of combinations of the above,
e.g., diffusing bends.
It is quite evident that the more that is known about the
flow distribution within a particular duct shape, the more effi-
ciently can the optimum particle concentration measurement probe
location and orientation be selected. To this end, it is neces-
sary that the flow field of the test shapes be accurately de-
lineated. The subscale tests were thus directed at determining
the effects that the above-mentioned mechanical disturbances have
on the velocity profile and distribution of particulates. The
results were to be applicable to combustion sources larger than
100 MW and to the ducting that exists downstream of the pollution
control equipment.
- 41 -
-------
1.0
50
U
U,
M
VM
DP)P
VLM
FIGURE 21 Effect of Model Scale with Model Froude Number
and Stokes Number Equal to Prototype Values
- 42 -
-------
The literature survey, although providing some general des-
criptions of the flow fields to be expected, provided little
data on particulate distributions as affected by mechanical
disturbances, especially downstream of power plant precipitators.
Very little data is available concerning the effects of internal
flow aids on aerosol behavior, in spite of the fact that these
are very popular in modern installations.
The precipitator and stack breeching at the Northern States
Power Company's Allen S. King power station at Oak Park Heights,
Minnesota, was selected as the test configuration to be modeled.
Selection of this test site was based, in part, on the results of
the power plant duct configuration survey. This arrangement
contains three basic test configurations, an 80° miter bend, a
contracting elbow, and a rectangular duct stack entry. This 600
MW forced draft coal-fired plant, typical of many built during
the 1960's, has undergone compliance testing with EPA equipment,
and has been the site for EPA sponsored programs to evaluate the
repeatability and reproducibility of the current EPA methods
(Ref 38). The configuration represents a typical modern ducting
layout in the region where emission sampling is done.
D. Model Test Facility
The 1/10 linear scale model, shown in Figure 22, was con-
structed of 3/4 inch particle board with inside surfaces finished
to a smooth surface by applying several coats of polyester resin
and alternately sanding. Turning vanes were fabricated out of
20-gauge galvanized sheet metal and the stack region was made with
commercial spiral ducting. One-hundred and four Masonite plates
were positioned in the model precipitator box and scaled-down
internal cross braces were mounted irj the ducting to make the
flow model complete. The duct connecting the model to an exist-
ing flow channel, driven by an adjustable forced draft centrifugal
blower, was also equipped with flow baffles.
A dry chemical feeder capable of maintaining a feed rate
accuracy of +_ 4% over successive 3-minute periods was used for
particulate injection. The feed rate accuracy improves signi-
ficantly with the period length, so even better accuracy was
obtained over the 20-minute sampling periods. The particulate
matter was fed into an ejector, powered by an auxiliary blower,
and pumped into the main model flow lines. Mixing was accom-
plished by means of a central baffle followed by six diameters
of straight ducting. Desired concentrations were obtained by
adjusting the feed rate which was determined by hopper weight changes,
- 43 -
-------
FIGURE 22A.
THE 1/10 SCALE MODEL OF THE ALLEN S. KING
POWER PLANT BREECHING AND ELECTROSTATIC PRE-
CIPITATOR
FIGURE 22B.
THE 1/10 SCALE MODEL OF THE ALLEN S. KING POWER
PLANT BREECHING SHOWING SAMPLING SITES UPSTREAM
AND DOWNSTREAM OF THE TEST ELBOW
- 44 -
-------
E. Instrumentation
The procedures specified in EPA Method 5 were used as
design criteria for the multi-point particulate sampling system.
However, specific instrumentation and methods were selected to
take advantage of the model size and the gas flow and particu-
late properties. To afford more flexibility in sampling with-
out slowing down the rate of data collection, separate particu-
late sampling probes rather than the originally proposed rake
arrangements were used.
Specially constructed .635 cm (1/4") I.D. sampling probes
were used in extracting the filter sample. Filtration of the
individual particulate samples took place outside the duct on
preweighed 47 mm glass fiber filters. Probe blockage effects on
the flow prohibited in-duet filter sampling. The particulate
concentrations in the model corresponded to that downstream of
the precipitator. Furthermore, since the carrier gas was ambient
air, there was no need to heat the filter section. Prior
knowledge of the ambient moisture content permitted omission
of the impinger section.
To allow calculation of isokinetic sampling rates, pitot
probe velocity determinations were made at each probe location
immediately before inserting and positioning the sampling probe, ,
In an attempt to minimize probe size and hence blockage effects,
velocity measurements were made before rather than during the
sampling period. The low concentrations and predominantly fine
particulate matter permitted a standard pitot tube (United Sensor
Model PAE-24-M-W) to be used for test instrumentation. The dif-
ferential pressure from the probe was sensed by a 2.54 cm (I'1)
H-O pressure transducer (Validyne Model DP 45). The output of
tnis transducer was displayed on a Honeywell Model 17 pen re-
corder. Barometric pressure was recorded from a mercurial
barometer and duct air temperature was measured using a mercury
thermometer inserted in the flow upstream of the test section.
Duct pressure was measured from a static port in the duct wall
connected to a U-tube water manometer.
Flow metering was different from that most commonly used
with Method 5, in that it was done by choking a calibrated ori-
fice (see Figure 23), This approach was considered to be more
desirable for the multi-probe arrangement because it is fast,
accurate, and takes advantage of existing equipment. These
orifices were calibrated to assure critical flow over a known
range of isokinetic velocities. A throttling valve upstream
of each orifice was used to maintain constant flow at the orifice
in the face of changing pressure drops through the filter. Pres-
sure instrumentation (mercury tube manometers, etc.) was used
- 45 -
-------
to monitor and set flow rates and confirm critical flow condi-
tions at each orifice. A large (8.4m /minute free air) vacuum
source was used to assure critical orifice flows under the most-
demanding flow rate and pressure conditions.
Filter weighing before and after sample collection and
conditioning was done on an analytical balance. Monitoring of
particulate loading was maintained continuously upstream of the
test section with a Konitest, a German instrument which operates
on the tribo-electric charge transfer principle. Published
Konitest calibrations indicate a direct linear relationship
between Konitest signal current and particulate concentration
of the sample.
F. Test Particulate
The full-scale prototype modeled in this study is the gas
ducting associated with a cyclone-fired utility boiler. The
literature (Ref. 39) indicates that the emissions from such
boilers are usually quite fine, with a mass-median diameter of
between 3 and 11 microns. Furthermore, since the sampling
station of interest is downstream of an electrostatic precipi-
tator, the particle size will be lower,, perhaps by as much as
a factor of two. Finally, the similarity parameters indicate
the desirability of using a model test particle which is roughly
half the size of the prototype particle. This means using a
particle with a mass-median diameter of at most a few microns.
Several types of materials are available in such a size range,
e.g., clays, alumina, paint pigments; however, the extremely
small size introduces several problems.,
A great deal of consideration was given to selection of an
optimum model test particulate. The problem involved getting the
proper particle size and density, maintaining a uniform flow rate
of the material, monitoring the particle concentration success-
fully with the Konitest, and recovery of the test particulate
that settles out in the sampling probes.
Particulate sampling was undertaken with Kaolin, an air-
floated form of Georgia red clay. The manufacturer's literature
lists the mean particle size at about one micron, however, actual
measurements made in the model test section with the Andersen
inertial impactor indicated a mass-median diameter of approxi-
mately 2.8 microns. The particle size distribution of such
materials is sometimes measured in a deflocculant-treated liquid,
but no indication of the method used for the existing Kaolin data
could be obtained. Agglomeration, both in the dry chemical
feeder and the gas stream are thought to be responsible for the
larger measured particle size. Although the specific gravity of
Kaolin is similar to that of fly ash (2.5) and yet the average
- 46 -
-------
Q
g HH
» 0
i— l r,.
t— ' kH
O H
< 2
t> g
/\ /
w
EH
co
D
< •
a
X /I
w
w
^
s
giJ
I—1 *£*
U D
< PH
>
V
1
OH
» W
OH EH Q
P W OH
OH O O
W 2 OH
O W
U 2 U
H H H
2 04 fa
O W H
CO EH OH
H O
1 ^ s. -
OH
W Q
OH 2
1 fa
1 OS
r^ co W
•* CO EH
2 H
O fa
OS
w
Q
!_J|
O
/\
EH EH
CO O
W D
EH Q
^»
^
W CT1
K^ m
PJ Q
s «
rt, PH
CO
s
CO
N
o
2
1
CO
S W fa
OWH
OH EH OH
fa O O
CO
B
CO
W
EH
W
Q
i
OS
O
fa
2
O
H
EH
CO
2
H
O
CO
U
M
EH
Pi
<:
OH
ro
CN
§
D
O
H
- 47 -
-------
particle size substantially smaller (both important in satisfy-
ing similarity parameters), the highly attractive electrostatic
forces exhibited by the clay particles introduced several new
problems. Increased particle buildup on duct walls, sample
probe tips, and the Konitest exciter tube were encountered. This
latter accumulation was thought to be responsible for the erratic
Konitest output. As a result of the agglomerative characteris-
tics of Kaolin inhibiting the dry chemical feeder performance,
additional difficulties were encountered in establishing uniform
aerosol generation. In general, the above characteristics made
Kaolin clay an undesirable choice for the present study.
Unfused alumina (Norton E266), commonly employed in super-
fine polishing, was similarly studied. Although the specific
density ( 3.5) is higher than fly ash and the "platey" crystal
shape different, the material is available in mean particle size
down to one micron. Again, alumina was found to possess some of
the same undesirable agglomerative characteristics experienced
with Kaolin.
Still further testing was done with fly ash provided by
Northern States Power Company. This ash, commercially marketed
as a concrete additive, was originally collected by electro-
static precipitators from the exhaust gases of a pulverized
coal-fired utility boiler and recovered from the hoppers. An
eight-stage inertial impactor, Andersen 2000 Incorporated Stack
Sampler, was used to classify four samples of fly ash drawn from
the test ductwork. The results of these tests, conducted at
the desired test location shown in Figure 25, are presented in
Figure 24. There is a slightly larger particle consist on the
bottom of the duct. Also, the mean particle size measured for
the four samples taken, approximately 6 microns, is small in
comparison with particle size expected in the actual prototype
plant. Some of the larger ash particles dropped out before
reaching the test section, the prime dropout area being the
large precipitator box section where velocities were about one
meter/second (3 ft./sec.). In spite of the fact that the
average particle size is a bit larger than desired, the fly
ash was thought to be the most viable test aerosol for the
model simulation. Unlike Kaolin and alumina, fly ash with its
less cohesive appearing properties was found to be very com-
patible with the dry chemical feeder and the Konitest sensor.
G. Local Particulate Concentration Measurements
Two sets of data (Runs 1 and 2) were taken at the test
station upstream and downstream of the 80° mitered test elbow
(see Figure 25). A 5 x 10 matrix of points was individually
sampled across both 36.6 cm x 83.9 cm stations. Model flow was
adjusted to approximately 6.1 meters/sec, in accordance with
- 48 -
-------
Sampling Locations 83,_9
(Dimensions are in cm from the
bottom of the duct)
• 7.6
- 5,0 3
36.6cm
100.0
50.0 0
N
•H
CO
(1)
20.0 o
•H
4J
10.0
O
rH
0)
PQ
C
•rH
2.0
1.0
0.5 -H
a)
0.2 *
0.1
0,2 0,5 1.0 2.0 5.0 10,0 20,0 50.0 100
Aerodynamic Particle Diameter (Microns)
FIGURE 24.SIZE DISTRIBUTIONS OF FLY ASH IN 1/10 SCALE MODEL
(UPSTREAM OF TEST ELBOW)
- 49 -
-------
similarity requirements in the test sections and the dry chemical
feeder was set to meter approximately 4.08 kg/hr of fly ash into
the flow stream. The chemical feeder itself was positioned on a
scale and the weight was recorded periodically throughout the
sampling day. Similarly, other facility variables, such as duct
pressure and temperature, were recorded.
The Konitest instrument was used to monitor the fly ash in-
jection (see Figure 25). The Konitest has a fast enough response
to show even the variation of particulate concentration due to
the rotation of the dry chemical feeder helix at 36 cycles/min.
Samples ranged in duration from 15 to 20 minutes. Individ-
ual variations in duration were caused by the mechanics of col-
lecting two or three samples simultaneously, measuring the velo-
city of the next sampling point, and recovering samples from
sample probes which had just been removed. These variations
were not considered significant, as long as adequate sample was
obtained for gravimetric analysis and as long as the sampling
duration was correctly recorded. Samples ranged in weight from
30 to 170 mg and were weighed out to 0.1 mg on an analytical
balance.
The raw data (filter weights, transducer readings, times,
orifice settings and facility parameters) were reduced by a
computer program. This program computed mass concentration,
velocity, sampling rate and isokinetic variation for each test
point. The results were further corrected by referring to the
trace produced by the Konitest instrument and correcting for
small timewise variations in the fly ash feed rate. Most of
these corrections were within 5 percent, although a few during one
run had to be corrected more, due to a momentary chemical
feeder malfunction.
It was discovered that a calibration error resulted in an
isokinetic variation during Run 1 that was biased by about 10
percent. Due to the small size of the particulate, however, the
effect on the data of this isokinetic error was considered to be
small and corrections for it were not made. The calibration
error was corrected before Run 2, where the isokinetic variation
averaged about two percent.
Table 5 presents the data averaged over the upstream and
downstream stations for Runs 1 and 2. The particulate concen-
tration, total emission, velocity and total flow rate show good
agreement over all and even better within the context of the
individual runs. It was noted that the average total emission
measured amounted to only about 2/3 of what the chemical feeder
was supposed to have fed into the system. Most of the missing
fly ash was located in the precipitator box section of the model,
- 50 -
-------
where the velocity was quite low ( 1 meter/second ). The
largest particles settled out causing the smaller particle
sizes measured by the Andersen sampler. Also, the average
particulate concentration and total emission results are
slightly lower at the test station downstream of the test
elbow than upstream, indicating that small amounts of fly
ash might have been lost in the duct between these stations.
Not much ash was visible on the duct floor through the plexi-
glass upper surface, however.
Figures 28 and 29 show the local particulate concentration
data for the stations upstream and downstream of the test elbow,
respectively. The data for the two runs was averaged point by
point to get the plotted values. The complex set of isopleths
are a result of some random scatter in individual point measure-
ments of particulate concentration. Graphical interpolation was
used to draw the isopleths. There is a noticeable shift toward
the bottom and outside surface of the duct downstream of the test
elbow, although it is not nearly as pronounced as that reported
by Hawksley, et al (1961), in some of the cases they studied
(Ref. 4). This is, in part, due to the fact that the mass
median diameter of the test aerosol (approximately 5 microns) is
much smaller, hence less centrifuging of the solids occurs, than
in the older (without modern control devices or flow baffling)
full-scale power plant ductwork that Hawksley, et al, studied.
Although no particle size is listed for the Hawksley data, the
heavy particulate concentrations and lack of efficient collec-
tion devices suggest a large population of coarse solids.
Furthermore, the internal guide vanes in the model test elbow
are effective in minimizing the gas flow stratification. The
implications of the models internal vaning on the particulate
and gas flow profiles is discussed in detail in Section K.
Since the particulate size in the model and the prototype
unit were similar, the Stokes number in the model will be about
four times larger. Since the centrifuging effects will be
exaggerated by a similar amount, the stratification realized
in the full-scale prototype plant will be much less than that
observed in the test program.
One interesting item to note about the raw data is that the
isopleths denoting average concentration (1.0) run through the
central portion of the duct, indicating that for this model a
representative sample could be obtained from a reduced number of
points distributed around the duct centerline.
The velocities measured in conjunction with the isokinetic
point samples are shown in a similar fashion in Figures 26 and 27.
The raw velocity data is subject to much less random variation
than the particulate concentration data. Furthermore, the overall
- 51 -
-------
O
§
in
W
A
I
i
m
CM
H
fe,
CN
C
3
«
rH
C
^
P4
CO
4J
•H
fl
D
VI
cu
4J
(U
g
n3
M
fCJ
CX
g
03
CU
4-1
CO
c
J5
a
g
03
CD
>_l
4J
CO
CX
g
03
CU
4J
CQ
C
^
o
Q
g
03
to
CX
D
00
CO
•
in
in
en
r
m
Vfl
vo
•
m
rH
CO
•
in
O
CU
CQ
*^
g
Jx,
4J
•r)
U
O
H
CU
*
tn
^
«C
^*
•
^3*
^
CN
*
rH
04
•
m
H
cr*
•
^s*
i-H
U
0
•
CX
g
EH'
•
cn
j>
<
o
ro
t~«.
0
fl
r^
•^
n
r~-
T
ro
r~
03
tn
EC
g
g
CU
M
3
CO
to
CU
M
I
4J
U
3
Q
CO
0
H
CTi
0
rH
^1*
O
rH
r~
o
rH
C
•fH
g
ro
g
O
4->
n3
P4
jj
O
H
fe
m
r^- co
n
1* OJ
o
o cr\
""i* •
*r O
<3 EH
- 52 -
-------
10
11
12
13
14
i-i
0)
55
O
ft
16
17
18..
19
H
W
H = 2.75 ft.
(0.84m)
W = 1.2 ft.
(0.37m)
Data Normalized
U = 5.88 m/sec.
Looking Downstream
FIGURE 26. VELOCITY DISTRIBUTION MEASURED UPSTREAM
OF THE TEST ELBOW IN 1/10 SCALE MODEL
- 53 -
-------
20 .
21
-0 OS O-8
H
X' W -V
H = 2.75 ft.
(0.84m)
W = 1.20 ft.
(0.37m)
Data Normalized
U = 5.77 m/sec.
Looking
Downstream
FIGURE 27.VELOCITY DISTRIBUTION MEASURED DOWNSTREAM
OF THE TEST ELBOW IN 1/10 SCALE MODEL
- 54 -
-------
19
H
<—W —>
H = 2.75 ft.
(0.84m)
W = 1.20 ft.
(0.37m)
Data Normalized
C = 447 mg/Nm3
Looking
Downstream
FIGURE 28. PARTICULATE CONCENTRATION DISTRIBUTION
MEASURED UPSTREAM OF THE TEST ELBOW IN
1/10 SCALE MODEL
- 55 -
-------
20
21
22
23
24
QJ
25
o
PU
26
27
28
29
0-7
H
-. W —
H =
W =
2.75 ft.
(0.84m)
1.20 ft,
(0.37m)
Data
Normalized
C = 436 mg/Nm3
Looking
Downstream
FIGURE 29. PARTICULATE CONCENTRATION PROFILE
MEASURED DOWNSTREAM OF THE TEST ELBOW
IN THE 1/10 SCALE MODEL
-------
variation of velocity values is considerably less than the vari-
ation in particulate concentration values/ and there is no
obviously significant effect of the test elbow on the velocity
distribution.
H. Konitest Distribution Measurements
It was recognized that simplifying the effort required to
obtain a particulate concentration distribution would be helpful
not only in full-scale unit measurements, but also in the model
data gathering. It was thought that the Konitest sensor might
provide this simplification. Its instantaneous readout of
particulate throughput eliminates the time-consuming sampling
and gravimetric analysis. Therefore, measurements of the con-
centration distribution were made with the Konitest at the test
stations upstream and downstream of the test elbow. The results
of these measurements were unsatisf actorv because the flow rate
recommended for the Konitest, .424 m3/min. (15 cfm) required
a large sampling nozzle, 4.1 cm, in order to achieve isokinetic
sampling at about 6.1 m/sec. This large diameter sampling probe
is not compatible with the model ductwork, since it would tend to
miss the local variations in particulate concentration, which
would be measured more sensitively by the smaller sampling
nozzles used in the gravimetric measurements (0.635 cm dia.).
Furthermore, local variations in velocity required large adjust-
ments in the Konitest flow rate and, thereby, large corrections
in the data due to its dependence on flow rate.
To reduce the required probe size, a sample dilution system
shown in Figure 30 was devised. This system was used with a
6.35 mm (1/4") and a 12.7 mm (1/2") sample probe. The sample
flow rate was metered by calibrating ,the pressure drop through
the sample probe, and total Konitest flow rate was set by
observing the pressure drop across the instrument. The dilution
air (ambient air) had to be cleaned by a 20 cm x 25 cm sheet of
glass-fiber filter media to prevent any background signal from
ambient dust.
There was some problem with loss of sample in the sample
probe and in the plastic hose which connected the probe to the
dilution point. This created a situation where the response
time of the Konitest was seemingly affected. Rapping on the
sample probe or line would loosen accumulated particulate
causing a large signal jump which would gradually decay awav
Some trace amounts of particulate could be seen clinqina to fh,/ '*
clear plastic sample hose. ^J-J-nging to the
with theSsamn?f L?^10Cal PJrticulate concentration were made
sJrea^ of ?he L^ ^°n system,at the test ports located down-
stream of the test elbow (see Figure 25) , The cross-section at
-
- 57 -
-------
this station was again divided into 50 equal areas (a 5 x 10
matrix) , with the measurement points being at the centers of these
areas. Each point was sampled three times during the test
period, and the order in which various rows of points were
sampled was determined in a random fashion. The mean of the
three test results for each point was used and the point values
so obtained were curve-fitted using a stepwise multiple regres-
sion computer program (Ref. 40 and 41). This program fit a
least-squares equation to the sets of data. The independent
variables in this equation were the two duct coordinates plus
second and third order combinations of these two. The computer
program selectively entered or removed these variables from the
regression equation, depending on the relative influence on the
goodness-of-f it of the regression equation. The program finally
re-calculated the values of the dependent variable according to
the regression equation and output them. The net effect of
such curve-fitting is to reduce the influence of obviously
spurious data and provide a simplified pattern which more
closely corresponds with reality.
The results of this curve-fitting are shown in Figure 31 for
the 6.35 mm (1/4") probe and Figure 32 for the 12.7 mm (1/2")
probe. These results are shown as Konitest signal levels (in
units of electrical current) normalized to the average signal
level, and displayed as a series of isopleths. Published Konitest
calibrations indicate a direct linear relationship between
Konitest signal current and particulate concentration of the
sample. These results show a distinct similarity to the con-
centration profile obtained from filter samples taken at the
same test position with 6.35 mm (1/4") probes (see Figure 29).
No attempt has been made to relate the average Konitest
signal level to a specific particulate concentration level due
to two reasons. First, there was uncertainty as to how much of
the particulate reached the instrument due to sample loss in the
probe and hose used in the sample dilution system. Secondly, no
quick and effective method was developed to calibrate the
Konitest periodically, and the applicability of previously
published Konitest calibrations is questionable.
believed that a pattern of concentration gradient may
of the test ^w due the presenc^o^a^et^f
c/ri? cmnr3.6"?,eqa radius o, curvature
— 5 8 —
-------
-------
-------
H
CO
D
(
£
H
EH
1
2
»
•
•
d
1
J
<
IFFERENT]
PRESSURE
pi
W
H &
w S
3 co
fa
/
w
W D
^
01
MANOMETE
EH EH
CO 2
W W
H p
§g
M CO
2
H
/
H
. N
\
a to
3
i_i
3§S
CO H
&j
Q
H4 W
s o
js p^
"
en
\
[
rxj O1
X FJ v«
§E
o fr
CM
EH
H
1 O
1 (X)
1
v
H
«
W
04 EH Q
H « OS
Oj rij O
EH W CJ
CO O g
w
H
CO
*
CO
§
H
Q
CO
EH
CO
H
EH
H
§
O
H
- 61 -
-------
20
21
22
23
0)
S25
o
CM
26
27
28
29
or 0-8
0-9
1.0 1-10
O.B
Konitest Siqnal
Current
H A
H
H =
W =
2.75 ft.
(0.84m)
1.20 ft.
(0.37 ra)
Calculated
Data
Normalized
Averaqe Siqnal
0.404 x 10~9 Amps,
. Lookinq
Downstream
FIGURE 31. PARTJCULATE CONCENTRATION DISTRIBUTION
MEASURED DOWNSTREAM OF THE TEST ELBOW IN
THE 1/10 MODEL WITH KONITEST (1/4 INCH
PROBE - 0.64 CM)
- 62 -
-------
28
29
H
-<— W ->
Konitest
Signal
Current
H =
W =
2,75 ft,
(0.84m)
1.20 ft
(0.37m)
Calculated
Data
Normalized
Average Signal
4.55 x 10"9 amps,
Looking
Downstream
FIGURE 32. PARTICULATE CONCENTRATION DISTRIBUTION
MEASURED DOWNSTREAM OF TEST ELBOW IN 1/10
SCALE MODEL WITH KONITEST (1/2 INCH PROBE
- 1.27 CM) - 63 -
-------
0.8
0.7
r-t
0)
o>
c
0)
p
o
'-i 0.5
DI
•H
w
(fl
Qj 0.4
-H
I
QJ
'£ °'3
I—)
0)
Q
• 00 Q a
Q
- • in O EE 3 Q
Q 3Q Q
•• •» •* Q 3CQ OQ . £Q
- ••• *» * * «*« «» « 'ti B
* ' A' ^ ^ *
A A^3
,i • • U
^ 'ci '&' • • <^
^ ^ • » •• 03
* *
• •
* * ^
it *
X^^r^---- — '^" **
Dashed Lines Indicate Relative Lateral *
Position of Turning Vanes Upstream *
0.2
0.1
24.6 in. 20.9 in. 17.3 in.
[~ (62.5 cm) (53.1 cm) t (43.9 cm)
Relative Lateral Positions of
s^ Filter Samfcle Points \,
- -^ S X Xx
1 1 1 t I 1 1 1 I | | (
55 50 45 40 35 30 25 20
Sampling Increments
15
10
FIGURE 33. Konitest Signal Current vs. Lateral Position (Looking Downstream)
- Sampling Site Located Downstream of Test Elbow in Model -
Symbol
•
B
&
Probe
Size, in. (cm)
1/2 (1.27)
1/2 (1.27)
1/4 (.635)
Full Scale
Signal, Amps
3 x 10~9
3 x 10~9
1 x 10"9
Test Particulate
Fly Ash — Michigan (Eastern)
Fly Ash — Michigan
Fly Ash—NTS,P, (Illinois and
Western)
-------
The results from the above measurements are shown in
Figure 37. The support bracing upstream of the test elbow,
see Figure 38, is thought to be in major part, responsible for
the dip in the upstream velocity distribution near the duct
center line. Flow disturbance, due to the internal bracing,
would be expected to be largest downstream of the center gusset.
Note the uniform velocity distribution existing downstream of
the test elbow where no internal bracing is present.
In spanning the test elbow, the shape of the particle con-
centration profile appears to be conserved, except near the
duct center. The gradient between the inside wall and the
first turning vane is obviously generated upstream and not by
the elbow or turning vanes, as previously speculated. The dip
in the upstream distribution, similar to that found in the
velocity profile, is again presumably due to brace interference.
The above results again suggest, as might be expected for an
aerosol with a mass median diameter of approximately 5.0 microns,
that gravitation and impaction effects are minimal.
J. Model Test Results withSkeP Distribution
The aerosol generation system in the model was originally
designed to provide a uniform particulate concentration upstream
of the test section. The aerosol was injected far upstream and
with a mixing baffle, turning vanes, etc., a fairly uniform
profile was produced. In order to test the effect of the elbow
on a nonuniform particulate concentration distribution, the aero-
sol injection location was moved further downstream.
The injection point was placed upstream of the modeled
precipitator . The duct at this location was partially baffled
and fly ash was injected into the lower corner of the duct. The
turning vanes and the precipitator plates in the model assisted
in the preservation of the distribution up the test elbow inlet.
Although simple in principle, the new system was found effective.
The artificially generated distribution measured upstream of the
test elbow (see Figure 39) simulates closely that measured in the
King Plant ducting (see Figure 55) . By varying the point or
points of injection and the degree of baffling, almost any con-
centration profile should be attainable.
With a particulate concentration profile similar to that
observed in the prototype System, 25 point local mass concentra-
tion surveys were taken concurrently upstream and downstream of
the test elbow. The results of these measurements are presented
in Figures 39 and 40. These isopleths indicate, as did previous
measurements, with a uniform inlet particulate distribution and
in the prototype installation, that the test elbow has little
- 65 -
-------
FIGURE 34. Precipitator Plates Installed in the Laboratory
Model Ductwork
Scale
1 m
Flow
r"r7
r r r~r
r r ->---„
\
^~>n
•JJ
! I i
V s J i
1. Test Elbow
2. Upstream Test Ports (midway between the top and tne
bottom of the duct)
3. Downstream Test Ports (midway between the top and bot-
tom of the duct) ,
4. Internal Brace Positioning - for Design Details, see
Figure 38
FIGURE 35.
Laboratory Model Ductwork Layout
- 66 -
-------
u
0)
CO
-H
° «
0)
O
o
o
o
o
Q Q Q
00 0
Si
Outside of -^ Duct Position
Bend Looking Downstream
Inside of
Bend
Upstream
Downstream
O
u
Before Plates Added
After Plates Added
(Test Ports Located Midway between the Top and Bottom of the
Duct - See Figure 35 ).
FIGURE 36. Comparison of Velocity Profiles Before and After Addition
of Precipitator Plates
- 67 -
-------
u. /
0.6
ro
|
\
§> 0.5
c
0
•H
1)
TJ
4J
c 0.4
(U
0
c
0
u
0)
M 0.3
o
••H
-M
m
0,
0.2
0.1
0
«
to 00 o ft •
® o • °
© ® ^~\ jgfc
il^B • BffliBBB0
n B Q
B™
ia CM •"
Hi gj
a-
UPSTREAM DOWNSTREAM
• O PART. CONCENTRATION
gg g VELOCITY
^.^-^
Dashed Lines Indicate Relative Lateral
Position of Turning Vanes in Elbow
6.0
5.0
4.0
*
u
0)
to
'e
3-0 £
•H
U
0
t-H
0)
2.0
1.0
n
Outside of f<- mint- Position ^ Inside of
Bend (Looking Downstream) "*"" Bend
FIGURE 37. Lateral Filter Sample Survey Taken Upstream and Downstream
of the Test Elbow in the 1/10 Scale Model.
- 68 -
-------
36.6 cm
Test Port
83.9 cm
West Side
of Duct
Looking
Upstream
D
Test Port
I)
East Side
of Duct
FIGURE 38. Details of Model Internal Cross-Bracing
- 69 -
-------
effect on the aerosol distribution. Only in the extreme lower
inside corner of the ducting is there a significant difference
in the concentration levels measured downstream. Considering
the dust dropout observed in the prototype elbow (see Figure 48)
and the low velocity levels measured in this region, these
results are not unexpected.
K. Model Test Results with and without Turning Vanes
Additional tests were conducted in the 1/10 scale model to
determine the influence that the turning vanes in the test elbow
have on the velocity and particulate concentration profiles.
Twenty-five point local velocity and particulate concentration
surveys were taken concurrently upstream and downstream of the
unvaned 80° mitred elbow. The results of these measurements
are presented in Figures 41 to 44.
Without turning vanes to pick off equal volumes of gas
and project them toward equal areas of downstream ducting, a
very nonuniform velocity profile is established (see Figure 42 -
Sample site approximately one equivalent duct diameter downstream)
Additional pitot tube measurements near the downstream inside
wall revealed the apparent existence of reverse flow, indicating
a region where the flow has separated from the wall. The higher
velocities measured near the lower inside corner further suggest
the possible presence of large eddies or secondary swirl.
Except for the upper inside half of the duct (near and
including the separated flow region), the local particulate con-
centration levels downstream are not 'substantially different
from those observed upstream of the test elbow. These results
that for the given test aerosol (HMD ~ 5 u M) even when
iveraing an unvaned elbow, the particles follow the gas flow
stre^mines closely, again confirming previous findings and
theoretical predictions. The apparent discrepancy between the
above results and those obtained by other investigators, such
as Hawksley, et al (Reference 4), result from the large dif-
ferences in the size distributions of the test aerosols.
In the low velocity region, which is the upper inside
portion of the duct, the emission level (particulate weight
rather than concentration) will be very low.
These tests emphasize the importance of internal guiding
devices. Previous measurements with three turning vanes (curved
- 70 -
-------
10
11
12
13
14
-p
M
o
16
17
18
19
Particle Concentration: gr/scfd
gm/Nm-5
0.05
Q.\0
°*10 ^a
o.u
_P-£I
l-foo
o- 0.2.
O.92 "S^fc9 O-Afe
Q.1O
0-E5
H =
W =
W
2.75 ft,
(0.84m)
1.20 ft.
(0.37m)
C = 0.47 gm/Nm3
Looking Downstrea
FIGURE 39 PARTICULATE CONCENTRATION DISTRIBUTION
MEASURED UPSTREAM OF TEST ELBOW IN 1/10
SCALE MODEL (UPSTREAM PARTICULATE DISTRIBUTION
ARTIFICIALLY SKEWED)
- 71 -
-------
20
21
22
23
24
0)
325
o
26
27
28
29
Particulate Concentration:
gr/scfd
gm/Nm3
H
H =
W =
W —>
2,75 ft,
(0,84 m)
1,20 ft.
(0.37m)
Looking
Downstream
C = 0.47 gm/Nm3
FIGURE 40. PARTICULATE CONCENTRATION DISTRIBUTION
MEASURED DOWNSTREAM OF THE TEST ELBOW IN
1/10 SCALE MODEL (UPSTREAM DISTRIBUTION
ARTIFICIALLY SKEWED)
- 72 -
-------
with long trailing edges) in the test elbow showed little strati-
fication. The vane-equipped elbow was very effective in pre-
serving the upstream velocity and particulate distribution.
Fortunately, although the numbers and geometric shapes may vary,
turning vanes are standard equipment in .modern power plant
installations.
- 73 -
-------
18
19 .
H
-. W-
H =
2.75 ft.
(0.84m)
W = 1.2 ft.
(0.37m)
Looking
Downstream
FIGURE 41. VELOCITY DISTRIBUTION MEASURED UPSTREAM
OF TEST ELBOW IN 1/10 SCALE MODEL
(NO TURNING VANES)
- 74 -
-------
20
21
22
23
24
(1)
O
OH
26
27
28.
Gas Velocity:
3Z.5
31.5
„ "° 3,
29h 3i-s -;^-
ft/sec
m/sec
H
i
H =
W =
W—v
2.75 ft
(0.84m)
1.2 ft
(0.37m)
Looking
Downstream
FIGURE 42 VELOCITY DISTRIBUTION
DOWNSTREAM OF THE TEST ELBOW IN THE
1/10 SCALE MODEL (NO TURNING VANES)
-------
10
11
12
13
14
3
2
15
in
O
16
17
18
19
Particle Concentration: gr/scfd
gm/Nm3
o.-so a. 32.
0.4|
o.i6 °-'fe
5*T o^ °'31
0-4|
O.lfc,
0.37
O-2.0
0.-50
H
H
2,75 ft,
(0.84m)
W = 1,2 ft.
(0,37m)
Looking
Downstream
FIGURE 43 PARTICULATE CONCENTRATION DISTRIBUTION
MEASURED UPSTREAM OF THE TEST ELBOW IN
THE 1/10 SCALE MODEL (NO TURNING VANES)
- 76 -
-------
Particulate Concentration gr/scfd
cT5o~
H
H =
W =
W ->
2.75 ft.
(0.84 m)
1.20 ft.
(0.37 m)
Looking
Downstream
Looking Downstream
FTCURK44 .PARTICULATE CONCENTRATION DISTRIBUTION
MEASURED DOWNSTREAM OF TEST ELBOW IN 1/10
SCALE MODEL (NO TURNING VANES)
-------
V. FIELD TESTING
A. Field Tests at Allen S. King Power Station
To provide a basis for evaluating the effectiveness of
the model in simulating full-scale behavior, a comprehensive
field test program was conducted in the full-scale prototype
unit.
1. Inspection of. Ductwork Interior
During a scheduled maintenance outage, an inspection of
the Allen S. King Plant breeching and precipitator interior
was made. The purpose of this inspection was to note dust
accumulations and variations from construction drawings of
internal braces and turning vanes and simultaneously, to
insure that additional test ports could be installed without
interference from internal structures.
Significant dust accumulations were found on the floor
of the ducts. The minimum depth of the dust was 45-60 cm,
with larger buildups found in the corners and other possible
void regions. Immediately downstream of the precipitator,
in front of the turning vanes, a large ridge of fly ash,
ranging in height from 1.2 to 1.8 meters, spanned the entire
width of the duct (see Figures 46A and 46fi). It's important
to note, however, that these and other accumulations in the
breeching (see Figures 47 and 48) are the net result of six
years of plant operation.
Since only minor variations from the construc-
tion drawings of the internal bracing and vanes were observed,
the model adequately simulated the prototype structure.
2. Particulate Concentration Measurements Downstream
of Prototype Test Elbow
Local particulate concentration measurements
were made downstream of the full-scale prototype test elbow
(see Figure 45). Seven test ports (see Figure 49), accessible
from four different stages of the scaffolding, were installed
along the 8.39 meter duct height less than 1 equivalent duct
diameter downstream of the 80° miter bend. Sampling ports
were placed on both sides of the duct in order to reduce
probe length requirements. Photos of the test elbow from
outside the concrete stack and the test platform and ports
on the duct wall inside the concrete stack are shown in
Figure 50.
Two complete and identical sampling probes,
designed in accordance with EPA Method 5 principles, were
- 78 -
-------
assembled for the test program. Since the probes were designed
specifically for individual point sampling, where sampling
periods would not exceed 15-2Q minutes, a 47 mm stainless steel
filter holder close-coupled to the probe tip, with a double
thickness of 47 mm MSA 1106-BH or equivalent glass fiber filter
media, was used in lieu of the larger normal out-of-stack fil-
ter. This arrangement improved measurement accuracy by reducing
the probe length from the sample nozzle to the filter, and
thereby reducing the amount of dust collected in the probe
tubing. By reducing the time required to clean the probe, the
total time period needed for a complete survey was reduced, and
hence, the probability of having steady plant conditions during
the survey was enhanced. The entire probe assembly was calibra-
ted in the EPA model ductwork prior to delivery to the field.
Aerodynamic interference effects caused by the probe tip and
filter holder assembly were all accounted for in the S-tube
calibration coefficient.
Initially, a 16-point lateral survey of
local particulate concentration was made approximately halfway
down the 8.39 meter depth of the duct. The results of this
survey, presented in Figure 51, showed a much larger gradient
near the outside of the duct than was anticipated. The velocity
profile at this port level, also shown in Figure 51, appeared to
be relatively uniform.
Additional, less detailed surveys of particulate
concentration were taken in the top port (Port 1} and the bottom
port (Port 7} to check for the presence of a similar gradient at
these levels. The results of these tests along with the local
particulate concentration curve from Figure 51 are shown in
Figure 52. The local concentration profiles measured in the
center port and bottom port showed similarity in shape and
level, while the top port profile was significantly lighter and
flatter.
Still another limited lateral survey of local
particulate concentration was made in the opposite (north) duct.
This was done in an attempt to confirm that the gradient measur-
ed in the south duct was related to the ductwork geometry and
not caused by a local condition unique to that duct, i.e.,
deteriorated precipitator performance near the outside wall.
The extent of this survey was limited by the existence of a test
port on only one side of the north duct (again approximately
halfway down the 8.39 m depth) and the limited probe length.
If it is assumed that the gradients in particu-
late concentration downstream of the test elbow are caused
by the existence of a population of aerodynamically large
- 79 -
-------
particles, say greater than 20 microns/ such, a size distribution
should be evident from samples taken with an inertial impactor.
Accordingly, particle* size distribution measurements were taken
with an eight-stage (Andersenl in-stack cascade impactor in the
duct cross section downstream of the test elbow. It was subse-
quently learned, however, that insufficient desiccating of the
moisture pickup on the backup filter mounted behind the last
impactor plate probably biased the mass median diameter in the
low direction. By eliminating the backup filter data from the
analysis, this error was overcompensated for and a conserva-
tively high mass median diameter was computed. Results derived
by this elimination are presented in Figure 53. The small
average particle size is apparent from these measurements.
Following refinement of the desiccation technique, only
two additional samples were taken due to plant outages.
These data are presented on the same figure and consist of
two mass median diameters, one calculated on the basis of
deleting the backup filter accumulation and one with it
included. Data including backup filter accumulations is
circled. The difference in the two values indicates the
effect that the inclusion of the fines collected on the
backup filter has on the mass median diameter.
The particle size distribution measurements
in the prototype plant indicate a particle size similar to that
of the test aerosol being used in the 1/10 scale model. These
distributions might be anticipated downstream of an electro-
static precipitator and a cyclone-fired boiler. Such small
particles would not be expected to deviate a significant amount
from the gas streamlines proceeding through the test elbow.
This fact has been demonstrated in previously reported model
data where only a small elbow effect was observed. In fact, the
lateral (impaction) effect of the test elbow in the model would
be expected to be exaggerated because the test aerosol used in
the model was at least twice the size of that specified by
similarity parameters. These facts suggested that the concen-
tration gradient measured downstream of the prototype test elbow
was not caused by the elbow itself, but rather was a consequence
of something upstream, perhaps even further upstream than the
precipitator. Confirmation of this possibility required measure-
ments upstream of the elbow.
3. Particulate Concentration Measurements Upstream
of Prototype Test Elbow
Four sample ports, equally spaced across the
top of the 8.39 meter high duct (visible in the upper photo
of Figure 50} were found to provide the only existing access
to the upstream duct interior. The duct location and outer
- 80 -
-------
I s
insulated housing prohibited easy installation of additional
new ports. This port arrangement necessitated the need for
a much longer (8.39m) probe than required in concentration
surveys taken downstream of the test elbow where probe
insertion was from the side rather than the top of the duct.
After positioning the required scaffolding
and after completing the assembly of a rigid 8.39m isokinetic
sampling probe designed specifically for vertical traversing,
a local particulate concentration survey was made upstream
of the prototype test elbow. Local particulate measurements
were made at 1.52m vertical intervals in each of the four
upstream sample ports, equally spaced across the top of the
8.39m high and 3.65m wide breeching (see Figures 45 and 49).
The results of this survey, presented as isopleths in
Figure 55, show a large particulate concentration gradient
existing near the extreme lower outside corner of the duct.
In spite of the relatively uniform gas velocity (see Figure
54), local particulate concentration levels in this lower
region were an order-of-magnitude larger than those existing
in the upper portion of the ducting. In light of the similar
skewed, distribution previously measured downstream of the
test elbow (see Figure 52), the above findings were not
unexpected.
As a result of a plant outage, a delay during
the contract renewal period, and a subsequent delay during
the assembly of an 8.39m vertical traversing probe, four
months elapsed between the testing upstream and downstream
of the prototype test elbow. To confirm that the downstream
particulate concentration profile had not changed significant-
ly as a result of possible changes in coal mixture, precipi-
tator performance, etc., during this period, an additional
25 point local particulate concentration survey was taken.
The results of this survey, presented in Figure 56, show
essentially the same skewed distribution as that measured
earlier. The lack of a lower test port prevented a more
complete determination of the particulate concentration
gradient near the very bottom of the duct downstream of the
test elbow.
The existence of the skewed distribution both upstream
and downstream of the prototype test elbow suggested, as
previously speculated on the basis of model testing and
theoretical predictions, that the effects of the turning
vane-equipped 80° mitered test elbow on the given flue gas
aerosol (MMD ^ 5 yM) are very limited even when only one
equivalent duct diameter downstream of the bend.
Recognizing that severe concentration gradients impose
rigid restraints on sampling strategy, attention was focused
- 81 -
-------
on isolating the source qf this irregularity. In an attempt
to establish whether the [measured skewed distribution was
the result of improper; precipitator performance or a product of
disturbances further upstream, a twenty-four point local partic-
ulate concentration survey was made in eight existing sampling
ports, equally spaced across the top of the 9.14m x 3.66m flue
gas ducting immediately upstream of the'expanding section feed-
ing the precipitator. (See Figure 45 for the location of the 8
equally spaced test ports. The results of the tests are shown
in Figures 57 and 58 ,) In spite of the U-shaped air preheater
and the contracting section immediately upstream of this test
location, both the velocity and the particulate concentration
appear relatively uniform, especially in comparison to the
distributions found downstream of the precipitator. A lead
shot system is employed to periodically cleanse the air pre-
heater tubing of fly ash deposits. The results of several
separate concentration measurements made at the same point, both
with and without the shot system energized, revealed no signifi-
cant effect, due to the shot system. The uniformity of the
distribution presented in Figure 58 shows that the effects are
small.
The above results show that the skewed distribution is
produced by the precipitator. A check of precipitator
system performance data showed that the cable, which energizes
the outside half of the precipitator Section 1 (see Figure
45), was open during the test period. Since there are three
additional precipitator sections downstream of Section 1 and
similarly, since earlier test results showed a similar
gradient even though Section 1 was then at full power, this
malfunctioning section is thought to play only a minor role
in the formation of the downstream nonuniformity. Using
precipitator section voltages recorded in the daily test
logs and typical experimental values of migration velocity
versus voltage determined in other precipitator performance
studies at the FluiDyne Energy Conversion Laboratory, theoreti-
cal values of the overall precipitator performance were
calculated at several positions across the width of the
precipitator. The results of these calculations indicate
that even though the outside half of the precipitator Section 1
was open during the test periods, the particulate concentra-
tion across the width of the duct should vary by no more
than a factor of 2. This might explain the horizontal
gradients observed in the upper portion of the ducting, but
fails to account for the near order of magnitude increase in
particulate concentration measured across the bottom.
Hopper sweepage and rapping losses are thought to be primarily
responsible for the large vertical particulate concentration
gradient. The existence of high concentrations near the
bottom of the ducting will enhance the probability of particle
dust dropout, perhaps in part explaining the large accumula-
tions observed at the outlet.
- 82 -
-------
During a scheduled maintenance outage, a second inspec-
tion of the Allen S. King plant breeching and precipitator
interior was made. An examination of the-precipitator
revealed no apparent explanation for the large horizontal
particulate concentration gradient observed downstream, but not
upstream of the precipitator. Dust accumulations on the collec-
tor plates (less than .15 cm} and the charging wires were not
excessive. No unusual dust accumulations could be found up-
stream of the precipitator. Similarly, visual observations
revealed no evidence of high velocity scrubbing or other indic-
ations of flow imbalance.
In conclusion, this initial field test program substantiat-
ed the model test elbow results. The test program also revealed
that very nonuniform and unpredictable particulate concentration
profiles, generated by sources other than the ductwork itself,
can exist and even more importantly persist in the exhaust gas
breechings of large power plants. Since other system components
such as I. D. fans, air heaters, and other control devices,
which are not simulated at least performance wise in the model,
were speculated to also be possible potential producers of
severe irregularities which might overwhelm the much smaller
gradients induced by elbows, expansions, and contractions, the
program direction was shifted, with primary emphasis focused on
full-scale field measurements rather than additional model
testing as originally planned. The intent of this approach is to
assemble a data bank incorporating the severely skewed velocities
and particulate profiles that exist in large power plants.
- 83 -
-------
M-l
O
tn
c
u
^ c
o -H
•P ^
(0
-p •
•H C/3
&
-H C
O 0)
OJ ^
O 0)
•H .C
-P H
10
.p
w
O
O
(I)
i-H
W
C
•H
in
- 84 -
-------
46a View of Turning Vanes Positioned at
the Precipitator Outlet Looking
Upstream (See Figure 45 for
Camera Location)
46b Close-Up View of Turning Vanes Shown Above
FIGURE 46. INTERNAL VIEW OF ALLEN S. KING POWER PLANT DUCTING
- 85 -
-------
47a View of Ducting Looking Downstream Toward
The Prototype Test Elbow
(See Figure 45for Camera Location)
47b View of Ducting Looking Downstream Toward
The Prototype Test Elbow
FIGURE 47. INTERNAL VIEW OF ALLEN S . KING POWER PLANT DUCTING
- 86 -
-------
48a View of Ducting Looking Upstream Toward
The Prototype Test Elbow
(See Figure 45 for Camera Location)
48b Close-Up View Looking Downstream Into
The Prototype Test Elbow
FIGURE 48. INTERNAL VIEW OF ALLEN S. KING POWER PLANT DUCTING
- 87 -
-------
I
TOTAL STACK HEIGHT =765' (239.27
OPEN
ANMULAR
SPACE
SAMPLE PORTS
(DOWNSTREAM OF TEST
•STEEL CHIMNEY
SUPPORT
STACK
PORTS
(UPSTREAM OF
TEST EL.60W)
(8.39m)
FIGURE 49.
A SIDE VIEW OF THE HORIZONTAL BREECHING
AT THE ALLEN S. KING POWER STATION
- 88 -
-------
FIGURE 50. PHOTOGRAPHS OF FIELD TEST SITE
- 89 -
-------
•H OJ
U U5
O \
o
CM
I
Q>
M
-P
(fl
C
1
q
•iH
T) q
•H 0)
CO PQ
G
C
O
•H
-P
•H
tn
o
en
u
3
Q
o
0) -H
-P 4J
ra nj
rH U
3 -P
u q
•H a)
-p o
M q
i 01
4-t 4J
O 0)
se
1^ (N
dj p
•p Q)
W Q
a
4J
U
O M
O
c] pt
-P
-a -P
•H ro
q
o m
>H EH
O
£>i fl3
•P Q
•H
U I
O
ns
rH
"0 fU
f(3 tT1
q
q -H
o ^5
-H
-P •
(0 W
SH
-p q
q
-------
h
X
•
4-1
O
4-J
O
3
Q
O
CM (N (M CTi
in r^ co ^
O <\T VO co
•-I in r~ ro
-c a x: x:
4J 4J 4-1 4J
3 3 3 M
O O O O
W W W 2
o <] a
T
s
±
M-l
O
<0
T3 T3
•H C
W 0)
d «
0)
n
4-1
U
Q
o x:
•H 4-1
4-> 3
•rt 0 4J
U! W U
O 3
ft n Q
-H
V
e 4->
OJ O
M s
4-1
U)
O
Q
•H
O
5
•H C
W Q)
-p m
3
O
4J
Ul
0)
EH
O
4->
O
H
ft
4-1
O
S
(0
(U
Vi
4J
Ul
4-1
O
U-l
0
TI
•rH
W)
w
o
C
o
•rH
-p
(0
M
4-1
C
0)
U
C
o
o
0)
-p
3
U
•H
+J
^
(0
ft
CO
I
§.
CN
in
H
CM
- 91 -
-------
TOP OF DUCT
OUTSIDE
OF
BEND
2.0 2,1 1.4
5.5 3.3 2,7
(1.6)
5.2 4,1 2,5
(3.2)
INSIDE
OF
BEND
SOUTH DUCT
LOOKING
DOWNSTREAM
BOTTOM OF DUCT
FIGURE 53.
MASS MEDIAN DIAMETER (MICRONS) AS A
FUNCTION OF CROSS-SECTIONAL POSITION
(NOTE: VALUES IN PARENTHESES INCLUDE
DATA FROM BACKUP FILTER)
- 92 -
-------
60
50
40
U -DO
rn -JU
14-1
4-1
•H
U
O
§20
10
0
Symbol
o
X
D
A
O
y/H
.181
.364
.545
.727
.908
(r
T
H
H=27,5 ft
(8.39 m)
W= 12.0 ft
(3.66 m)
W
O
o
A
O
o
a
x
A
O
12.0
9.0
u
0)
W
6.0
3.0
Outside of
Bend
S Duct Position —
Looking Downstream
Inside of
Bend
FIGURE 54. Velocity Distribution Upstream of Test Elbow
Allen S. King
-------
Particle Concentration
25
Outside
of
bend
Duct position
looking downstream
Inside
of
bend
H
H = 27,5 ft,
(8.39m)
W = 12.0 ft.
(3.66m)
FIGURE 55. PARTICLE CONCENTRATION MEASURED UPSTREAM
OF PROTOTYPE TEST ELBOW (ALLEN S. KING PLANT)
-------
10
15
Depth,Feet
20
25
O.O2
Concentration
Q.QI8
O.O41
Outside of
Bend
Q-'S Q-12. o.io
0-4 \ Q.Z.1 O.Z»>
0,02.
o.o4fe
H
•<— W —>
H=27.5 ft
(8.39m)
W=12.0 ft
(3.66m)
3.0
Depth, meters
6,0
Duct Position
Looking Downstream
Inside of
Bend
FIGURE 56. PARTICULATE CONCENTRATION MEASURED DOWNSTREAM
OF PROTOTYPE TEST ELBOW (ALLEN S. KING PLANT)
-------
(oas/ui)
-p
u-t
E o
V£> CN
• H •
00 —• CTi
4-4
E
o
rH O
II
S3
II
oo
in
o
O O D
S3
3
o
EH
U
D
Q
53
[H
D
O
H !?,
EH H
O
W
U W
D IS
OF:
O
Q
H
«
O
O
EH
D
O
CO
cq o
S H
EH EH
co ri!
ft £H
D CO
EH «
U W
P O
tn
O O
S3 H
cn 2
co w
O
o
ro
o
CN
2 S3
O EH
H
« o
EH EH
W rt!
H EH
P H
CM
>H H
EH U
H M
U K
O PM
J
W fo
> O
r-
m
W
tf
P
O
- 96 -
-------
o
•H
-P
•H
M
O
eu
-P
u
3
Q
o
o
o
00
o
CN
P
o
w
3
0
C
•H
O
Q
•H
- 97 -
-------
B. Field Testing at Black Dog Power Station
A velocity survey, using a Fecheimer type pitot probe
(see Figure 76}, capable of measuring yaw angle along with
total and static pressure, was made immediately downstream of
a double inlet-single outlet induced draft fan at Northern
States Power Company's Black Dog Power Station.
The in-duct testing procedure involved rotating the
probe until the yaw pressures were equal, and then measuring
the angle of rotation (yaw angle). Velocity was then calculated
from the total pressure, indicated static pressure (average
of the two outside orifice pressures), and a predetermined
calibration constant. As illustrated in Figure 78, .the
directional sensitivity of this two-dimensional null-type
pressure probe was very good.
A 105-point velocity survey was made in the 2.33 meter
by 3.50 meter vertical duct shown in Figure 59. Measurements
were made at 15.2 cm intervals in each of the seven sample
ports located 1.5 equivalent duct diameters downstream of the
fan outlet. The results of this survey, displayed as a
series of isopleths in Figure 60 show a very nonuniform
velocity profile. Local velocity levels varied from 0-33
m/sec (0 to 110 ft/sec) while the flow direction deviated by
no more than 13 degrees from the vertical direction. The flow
angularity is directly attributable to bending of flow as it
begins traversing the elbow. As shown in Figure 59, the
sampling ports are in the elbow inlet. The low velocity
region near the centroid of the duct is thought to result
from starving of the central part of the fan rotor. This
profile illustrates the kind of skewed velocity distributions.
that can exist in breeching ducts downstream of I.D. fans.
- 98 -
-------
N
Sample Ports
X
(Duct Depth = 2.33m)
Damper
O O O O O O O
3.5m
I
(Depth = 2.33m)
I. D. Fan
Fan
Inlet
FIGURE 59. SIDE VIEW OF NORTH BREECHING (UNIT NO. 4)
AT THE BLACK DOG POWER STATION
- 99 -
-------
D
O
H
- 100 -
-------
Test Ports
Symbol
O
D
0
A
x, m
.54
1.63
2.71
3.80
TT TT TT
.7m
4.3m
100
80
-P
•^
o
rH
0)
60
40
20
•=130
250 MW Load
0
East «<-
Wall
Duct Position
20
10
Looking Downstream
->- West
Wall
u
0)
-M
•H
O
O
FIGURE 61. Velocity Distribution in North Breeching of a Midwestern
Power Plant (250 MW Load) - Sampling Station Located
Downstream of I.D. Fan.
- 101 -
-------
Symbol
O
D
O
A
x, m
,54
1,63
2.71
3,80
•3.7m-
4.3m
u
0)
in
U
0
100
80
60
40
20
0
330 MW Load
East
Wall
Duct Position
Looking Downstream
30
20 8
10
-p
•H
U
O
0)
10 >
West
Wall
FIGURE 62. Velocity Distribution in North Breeching of a Midwestern Power
Plant (330 MW Load) - Sampling Station Located Downstream
of I.D. Fan.
- 102 -
-------
Test Ports
TT
Symbol
O
n
O
A
x, m
.54
1.63
2.71
3.80
X
3.7m
4.3m
JL_
100
80
u
0)
w
4-J
-H
O
0
60
40
20
250 MW Load
0
West
Wall
A .. __
30
20
10
Duct Position
Looking Downstream
East
Wall
o
0)
-p
•H
O
0
FIGURE 63. Velocity Distribution in South Breeching of a Midwestern
Power Plant (250 MW Load) - Sampling Station Located
Downstream of I.D, Fan (See Figure 65 ).
- 103 -
-------
Test Ports
e
•
9
•
•
•
*
9
0
t
«
•
Symbol
o
D
O
A
x. m
1.63
2.71
3.80
T^
>
•«-
II IT
f
~ 3.7m =*"
Jj
4
N
\
3 m
t
u
0)
ft
4-4
•H
U
o
100
80
60
40
20
0
West
Wall
330 MW Load
Duct Position
Looking Downstream
30
20
10
•East
Wall
u
0)
(0
-p
•rH
o
o
FIGURE 64. Velocity Distribution in South Breeching of a Midwestern Power
Plant (330 MW) - Sampling Station Located Downstream of
I.D. Fan (See Figure 65 )•
- 104 -
-------
f
k
58
?
s
1 X
L
b
"*,
Csi
cv
>:
U
V4>
Oil.'
I *
k
fc
^
^
^
i
»
j
O
H
w
&
o
EH
Cfl
Q
H
a
o
H
w
u
w
in
O
M
CM
- 105 -
-------
C. Additional Field Test Data
Several additional velocity distributions, again measured
downstream of I. D. fans, were provided by a Midwestern power
cooperative. These surveys (see Figures 61 to 64) taken under
both full and part load conditions, provide additional examples
of the nonuniform character of the flow fields in which power
plant emission measurements must often be made.
Field tests were also conducted at the Allen (Memphis,
Tennessee) and the Colbert (Florence, Alabama) Power Stations.
A Fecheimer probe, an S-type pitot tube, and a flow visualization
indicator were used to evaluate the flow fields in these two
TVA power plant stacks. The results from these and other
test programs are described in detail in Section V-E.
Numerous other typical and atypical velocity and par-
ticulate concentration profiles are included in the 33 test
distributions presented in Appendix A.
- 106 -
-------
D. Field Tests at Part Load Operation
A field test program was conducted at the Allen S. King
power station with the explicit purpose of studying the
effects of part load operation on plant emission characteristics.
More specifically, the program was designed to provide a
better understanding of the effect of measuring volumetric
flow rates, particulate concentrations and total emission
levels under part load operation and scaling such measurements
to the full load case. Particular attention was given to the
transient behavior during the transition from low load to
full load and vice versa.
During the late summer and early fall season, the Allen
King plant typically operates near full load (550 MW) but is
forced during the early morning hours (2:00 - 5:00 a.m.), as
a result of diminishing demand, to trim plant output to the
240 - 300 MW level. This provided an unusually convenient
opportunity for investigating part load emission performance
on a large base-load fossil fuel fired power plant. Previous
testing experience and data along with conveniently accessible
and strategically positioned existing test ports made this an
especially attractive site.
Velocity and local particulate concentration profile
measurements were made upstream and downstream of the precipitator
both at near full load and half load operation along with
fixed point transient measurements at both sites. Upstream
surveys (8 vertical sampling ports shown in Figure 45) were
made to determine the effects of load range on boiler emission
output. The added effect of decreased velocity on precipitator
collection efficiency was included in the measurements made.
in the breeching (see Figure 49 - Test Ports Downstream of
Elbow) downstream of the precipitator.
Two specially designed light versatile sampling probes
with accompanying monorails were assembled to permit easy
horizontal traversing of the entire 3.65 meter width of the
downstream breechings. These probes eliminated time losses
associated with transporting sampling equipment from one side
of the duct to the other. The limited steady state half load
operating period (2:00 - 5:00 a.m.) necessitated the need for
very efficient sampling procedures. These same probes, with
a special condensation trap attachment, were also used for
the vertical sampling in the 3.65 m duct upstream of the
precipitator.
The results of these profile measurements, both above
and below the precipitator, are shown in Figures 66 to 71.
Accompanying each data point in parenthesis is the corresponding
plant load level in megawatts. The load information was
-------
obtained from control room output data recorded at fifteen
minute intervals. Figure 66 shows a very skewed particulate
concentration similar in shape, but with concentration levels
considerably lower than those measured in earlier full load
test programs (see Figure 52). During the part load cycle,
the particulate concentration was again .observed to be highest
in the lower outside (west) corner of the breeching. The
corresponding new full load test results downstream of the
precipitator are shown in Figure 67. Although vertical
gradients resemble those observed at half load and in previous
full load measurements, the horizontal gradients, especially
near the bottom of the duct (Port 7), are not so similar. The
transient data measurements, presented in Figure 68, reveal a
possible explanation for these dissimilarities. These measure-
ments were made at a fixed sampling point, 1.37 meters from
the west wall in Port 7, during the half load to full load
turn-up period. Throughout this transient period, a near 1:1
correlation is maintained between the ratio of gas stream
velocities and the corresponding megawatt loadings. The
particulate concentration, however, varies in a more pulsating
fashion, with instabilities observed to persist for at least
one hour after achieving full load operation. It's interesting
to note the continually decreasing particulate concentration
levels during the 8:00-9:00 a.m. time period. In-stack
transmissometer readings further confirm the existence of
this transient behavior. Additional test points should have
been taken in order to determine exactly how much time is
required to establish steady-state conditions. Our original
assumption of one hour is obviously invalid. These findings
suggest that much of the full-load data (Figure 67) was taken
during a potentially unstable period. Those data points taken
at a load level below 530 MW are especially suspect. This
uncertainty is thought to be at least, in part, responsible
for the unexpected observed profiles.
As a result of the uncertainty associated with the
stability of the system during the period in which the full
load test data was taken, the possibility of comparing part
load data with full load data from previous test programs
(Summer, 1974) was investigated. As a result of changes
incorporated in the system, these earlier profiles may not be
representative of those existing at the time of this latter
testing. Prior to July 1975, fly ash from the precipitator
hoppers was being re-injected with coal into the cyclone
burners. At present, without ash re-injection, particulate
concentrations might be anticipated to be significantly
lower, however, distribution shapes, especially downstream of
the precipitator, would not be expected to be as drastically
affected. Slight changes in the coal mixture might also have
an effect on the quantity of ash being generated.
- 108 -
-------
Test results from the surveys, made upstream of the pre-
cipitator are presented in Figures 69 to 71, The rather uniform
profiles are similar in shape to the full load profiles observed
in the previous test program.. The irregularities in the
emission concentrations during the transient load period (see
Figure 71) appear to be less pronounced:than those measured
below the precipitator,* however, a similar continuous decline
in grain loading is observed during the first hour after reaching
full load production. Again, there ±s insufficient data to es-
tablish the exact steady state full load local particulate emis-
sion level. As a result of the high grain loadings upstream of
the precipitator, permitting short 2 to 4 minute sampling periods,
differences in the megawatt loading between adjacent sample points
were minimized. This procedure does not, however, lessen the un-
certainty about whether true steady state plant conditions had
been established. Again,'the authenticity of the full load data
is in question,
A 48% decline in average plant loading produced a 49.6%
drop in the upstream total volumetric gas flow rate. Down-
stream of the precipitator, a volumetric flow rate 40.6% be-
low the measured full load value accompanied a 44.6% decrease
in megawatt loading. The data presented in Figures 63 and 64
indicates a similar linear correlation between total volumetric
flow rate and system loading. These findings suggest that volu-
metric flow rate is a scalable function.
A 62% drop in total measured particulate emission (average
particulate concentration times total volumetric flow rate)
upstream of the precipitator was produced by the same 48%
drop in megawatt output. Downstream of the precipitator, the
particulate reduction was even more drastic, A 72% drop in .
total measured particulate accompanied a 44.6% drop in the
plant power output. Increased settling losses associated
with the formation of new low velocity regions is thought to
be primarily responsible for the disproportionate decrease in
particulate loading upstream of the precipitator, since the
boiler cyclones when operating at part load levels, perhaps
well below their designed range of optimum efficiency, might
be expected to produce a disproportionately high amount of
particulate per unit of heat output. At the King Plant, 12
individual cyclones each capable of burning up to 18,160
kilograms of coal per hour, fire into a single boiler. To
reduce system output, individual units are typically throttled
to 11,350 kilograms of coal per hour before being de*-
energized. During the given part load test periods, only 7
of the 12 cyclones were in operation. Although uncertainties
associated with the full load particulate concentration
- 109 -
-------
measurements prevent precise conclusions about particulate
emission scalability, these test results do reveal important
overall trends. They explicitly illustrate the persisting
instabilities in particulate loading associated with system
load changes and underline the need for further investigation.
- 110 -
-------
c
0
-P
3
o
c
-H C
O
M -H
0 -P
•P (0
rH C
3 -H
u x;
•H U
-P 0)
n
-------
U
C
o
U
13
m
u
w
,c
o
0)
(U
M
CQ
-P
0
C
-H
(0
-p
•H
ft •
•H C
O O
rH
(0 H
rH rtj
3
U I)
•H ,C
4-1 4-1
M
(d <+-!
eu o
vo
8
H
- 112 -
-------
-P U
to o
ft O
T!
O
CO
0
O V£) CN
™ rH r-t
0 00
' T • oas/ui
IT) O
H rH
CO ^f
O 0
0 0
4 AH\T DO-[9A '
lO
' ' i /^
o
o
O 0
O"l
1
U H
H K
EH U
H W
W P
ac o
EH CO
O 53
S3 H
3
Z
O
H
CTl
ro O
i £ N
^— ffi— --
x ts
' t t'
d v.
1
-P
H
Symbol
Load
Velocity
Concentration
o H
EH
H
EH S
IS
CO
H
(n
- 113 -
-------
o
m
4J 0
%4 C
a o
ft U
o
o
g 4J E
O li) O
CO O (Ti
^- n —
33
CN iH O
O <\ D
•O
3
K
I
c
(fl
iH
ft
c
•H
C
0)
o
C W
0
•H C
w 6
O fO
Oj OJ
-p -t-i
o w
2 S
Q 5
O
VD
4-> O
M C
(0 O
ft O
O
CN
o
Qj
M
n.
4-)
rfl
4-1
C
0)
U
0
o
Q)
4-)
3
U
•H
4-1
cn
- 114 -
8
-------
U
. . ui
4-> U \
(0 0 tn
ft O
•p B 4 g
±4 M-l
IN ro O CTi
rH ^ CO •-'
II II
w
M kO O
r-- v£ \^
CN rH O
o
.n
4-1 U r*
(0 O 01
ft O 6
O
fN
4-J
4-)
U
3
Q
-P
C
O C
•H -H
-P
•H B
W (0
4-1
4j m
u c
a
•H
O
V
o
4-1 T3
OJ (0
4-1 rH
U i-H
Q S
4H
O
4->
C
4-1
T3
•H
W C
O W
U •
< W
C C
O 0)
•H rH
C
U
O
O
-------
8
S
- 116 -
-------
E* Flow Angularity in Large Power Plant Stacks
A brief investigation of gas flow behavior in large power
plant chimneys was undertaken. The lack of existing data de-
fining flow angularity in large stacks, the twin spiraling
vortices frequently seen issuing from large stacks, and the
concern, contrary opinions, and general confusion over the
existence or nonexistence of cyclonic action in stacks among
the stack sampling engineers interviewed in a preliminary sur-
vey suggested the need for further studies.
1. Field Observations
During the field test programs at the Allen S. King Power
Station (600 MW), two separate plumes were often seen issuing
from the 244 meter (800 foot) stack, each one having a cyclonic
motion opposite the other (see Figure 72). A close-up view
from an observation catwalk near the top of the stack showed
two very strong spirals emerging from the chimney mouth. The
most obvious explanation seemed to be that the cyclonic action
was generated by the interaction of opposing gas streams from
the diametrically opposed breechings below. Since large-scale
turbulence is inherently self-preserving in round ducts, it
was reasonable to assume that such cyclonic motion could per-
sist all the way to the top of the stack, which is at least 27
stack diameters above the inlet breechings. Initially, little
consideration was given to the draft loss that would be accrued,
The effect that such a nonuniform velocity profile would have
on present stack sampling methods became the immediate concern.
2. Stack Sampling
In spite of the difficulties associated with raising
test equipment to the test port levels, often several hundred
feet above the ground, and other difficulties associated with
handling 3-4.5 meter (10-15 ft) sampling probes at these ele-
vations, there are numerous reasons for sampling in the stack
rather than the accompanying breechings. Where several ducts
feed into a single chimney (frequently occurring in today's
large power plants, especially with the new emphasis on tall
stacks), a single survey can replace several individual duct
tests. In addition to the many inherent advantages of sampling
in a vertical rather than a horizontal duct, especially with
Method 5 equipment, the stack is often the only location where
a sampling site can be found eight or more equivalent duct
diameters downstream and two or more diameters upstream from
a mechanical disturbance, permitting the minimum 12 point
traverse.
-------
Before pursuing in more detail the existence or non-
existence of cyclonic flow in large power plant stacks, it is
imperative that consideration be given to'the relative effect
that such flow might have on present stack sampling strategies
(EPA Methods 2 and 5). Should the existence of cyclonic flow
be found to have an insignificant influence, there would be
little justification in carrying this discussion any further.
EPA Methods 2 and 5 specify the S-tube as the velocity
measuring device. The S-tube must provide an accurate measure-
ment of the velocity distribution to permit the determination
of the volumetric flow and the maintaining of isokinetic sam-
pling conditions. Its ability to perform this function in flow
fields with cyclonic action is not well established. A brief
theoretical investigation.is in order.
The S-tube velocity measurement probe is known to be sen-
sitive to both yaw and pitch angle (see Figures B-ll and B-13).
The effect of yaw is symmetrical. Note that a yaw of 20° in-
creases the indicated velocity by about 4%.
Pitch produces larger errors than yaw and is asymmetrical
as would be expected. A negative pitch angle causes the probe
to indicate a reduced velocity, corresponding to a rise in
pressure at the downstream facing pressure tap. Ten degrees
gives an error of 4% to 6%, while 20 degrees gives an error
of 12% to 13%. This is the angular orientation that produces
the largest errors. Positive pitch angle causes the probe to
indicate an increased velocity, corresponding to a drop in
the pressure at the downstream facing pressure tap. The errors
are less than with negative pitch angles. Ten degrees gives an
error of 2% to 4%, and 20 degrees gives an error of 4% to 7%.
If the S-tube is aligned so that neither the yaw angle
nor the pitch angle are zero, the resulting errors are approxi-
mately the sum of the two individual errors.
(Single spiral flow)
Consideration will first be given to the case of
single spiral flow (superimposing vortex motion on uniform
axial motion) up the stack (see Figure 73). Note that the
tangential velocity (Ufl ) increases with the radial distance
from the duct center. Sampling in accordance with Method 2
(assuming no rotational flow), the S-tube is traversed along a
horizontal line passing through the stack center with the
pressure taps aligned in the stack center line (axial) direc-
tion. The actual duct velocity vector (U) will be at some
other direction, depending on the magnitude of the tangential
- 118 -
-------
velocity component (see Figure 73). The actual velocity vector
(U) at any given point will, however, lie in a plane perpendicu-
lar to the S-tube strut axis. This means that the probe will
be yawed but not pitched with respect to the actual velocity.
At a given sample point, the angle of yaw can be calculated
from the relation: :
Tanp =
ue
U
Z
From Figure B-ll it is apparent that the S-tube can be yawed
up to 45° in either direction without introducing more than a
8^ error. This means that the tangential velocity component
(UQ ) can be as large as the axial velocity (Uz) without the
measured velocity (magnitude) deviating by more than 8% from
the actual velocity magnitude JTJJ . For cases where the tan-
gential velocity component is larger compared to the axial
velocity component (yaw angle > 45°) the errors become much
greater.
This analysis suggests that when the yaw angle is less
than 45°, the S-tube will provide a relatively accurate measure-
ment of the magnitude but not the direction of the actual duct
velocity \J . For emission testing, however, the direction is
required for the maintaining of isokinetic sampling and for the
determination _of the velocity component in the axial direction
of the stack ~UZ = u COS p . If the S-tube reading is assumed
to give the magnitude of the axial velocity | U I (merely be-
cause it is aligned in this direction) an additional error
proportional to the COS p must be added to that described above.
(Double Helical Flow)
For double helical flow, and more complicated
cyclonic profiles, the resultant velocity vectors in the plane
normal to the stack center line axis, here referred to as
the circular velocity component (U ), has both a radial
and a tangential component - see Figure 74. For this case, the
orientation of the vortices with respect to the sample ports
becomes very important. If sampling from Port #1 (Figure 74)
the circular velocity vector (Uc ) at any point is essentially
perpendicular to the S-tube strut axis so the probe is^again
only yawed with respect to the actual duct velocity (U).
The error analysis discussed in the previous section is thus
applicable.
When sampling from other ports, #2 for example,
the circular velocity component (Uc ) is no longer perpendi-
- 119 -
-------
cular to the S-tube strut and hence the actual stack velocity
(U) is oriented with respect to the S-tube with both pitch
and yaw. As illustrated in Figure 74, the yaw angle can be
found from the projection of (U) on the plane perpendicular
to the S-tube strut axis, and the pitch from the projection
on the plane containing the stack center line axis and the S-
tube strut axis. Unfortunately, pitch produces even larger
errors (-45° pitch = 30% error). For non-streamlined flow
large errors can occur in the measurement of the magnitude of
the actual velocity (0*) , not to mention the direction. Errors
in the measurement of the face velocity (velocity vector
parallel to the stack walls Uz ) will be a combination of the
above error and the cosine of the angle between the flow and
the S-tube. Based on the preceding discussion, cyclonic flow could
have a very significant effect on velocity determination by EPA
Test Methods 2 and 5..
3. Double Vortex Phenomenon
Several factors suggest that the frequently observed
twin-spiralling vortices are the result of secondary flow effects
generated by the bending of the vertically ejected stack gas
by the prevailing crosswind and are thus confined to the dis-
charge or outlet region of the stack. These include:
(a). Similar profiles have been observed when cooling
air is ejected from holes on turbine blades into
a cross flowing main flow stream. Reference 42
(see Figure 75) describes the "complex three-
dimensional flow field downstream of the ejection
holes which is characterized by a pair of strong
streamwise vortices downstream of each ejection
hole."
(b). The orientation of the King Plant stack gas vortices
were found to vary with existing wind direction,
with the cord line between the vortex centers
always staying perpendicular to the wind direction.
(c). Double helical flow has been observed in numerous
stacks, especially small stacks, without multiple
inlet ducting arrangements.
(d). The King Plant is equipped with a pyramidal
shaped guide vane structure in the chimney base,
similar in design to that reported in Reference 29
to be effective in preventing flow irregularities.
(e). None of the cold-flow model studies in the litera-
ture (see Section III-C-2), even with unbaffled
multiple stack entries, report a double vortex
type velocity profile. Only 4-vortex or single
spiral type profiles have been observed in the
models.
- 120 -
-------
(f). From visual observations of the Allen S. King
Stack Gas Plume, from a catwalk near the top of
the stack, the circular tangential velocity com-
ponent is estimated to be several times that of
the axial component. A large pressure drop would
be required to provide the energy necessary for
internal formation of these large-scale eddies.
The fact that draft losses of this magnitude
are not present in the King breeching, further
rules out the possibility of vortex formation
within the confines of the stack and associated
inlet breechings.
4. Field Test Probes
The brief treatise describing the effect of cyclonic flow
on present stack sampling strategies, presented in Section 2,
suggests that when using the S-tube probe, specified by EPA
Methods 2 and 5 as the velocity measuring device, in nonstream-
lined flow fields, large uncertainties in the orientation and
magnitude of the flow field can result. The yaw and pitch
sensitivity curves (see Figures B-ll and B-13) indicate how
critical it is that the probe be positioned parallel to the
flow stream. Unlike the standard pitot tube, the S-tube can,
for two-dimensional flow, be roughly aligned with the flow by
rotating the probe to produce a zero null reading and rotating
the probe back 90°. An alternate procedure involves positioning
the probe halfway between the maximum AP readings (see yaw
angle curve - Figure B-ll).
Several directional sensitive probes were assembled for
use in the field test programs. The first, a simple flow
direction indicator with a string attached to a fine rod (see
Figure 77) permitted visual flow observations. Vortex flows
are particularly sensitive to the introduction of foreign
objects which may trigger instabilities leading to vortex
breakdown. Probe dimensions were thus kept small with respect
to the anticipated vortex core sizes in an attempt to minimize
disturbing effects.
A Fecheimer type pitot probe (see Figure 76) capable of
measuring yaw angle in addition to total and static pressure
was also assembled and calibrated. The three-orifice yaw head,
similar in design to that described in Reference 22, consists
of a long circular cylinder with three piezometers located 39°
apart near the smoothed end of the cylinder. The in-duct
testing procedure involves rotating the probe until the yaw
pressures are equal, and then measuring the angle of rotation
(yaw angle). Velocity is then calculated from the total
- 121 -
-------
pressure, indicated static pressure (average of the two out-
side orifice pressures), and a predetermined calibration
constant. As illustrated in Figure 78, the directional sensi-
tivity of this two-dimensional null-type pressure probe is
very good. The long tubular shaped probe head and strut per-
mit easy insertion into 7.62 - 10.16 cm' (3 and 4 inch) sampling
ports.
A cone-type five-hole pitot probe (see Figure 79) , capable
of measuring both the yaw and pitch angles, along with total
and static pressure, was also constructed. Again, as illustra-
ted in Figure 80, the directional sensitivity of the three-
dimensional null-type pressure probe is very good. The sampling
procedure involves rotating the probe until the yaw pressures
are equal, measuring the angle of probe rotation (yaw angle),
and finally determining the pitch angle from the pressure differ-
ential measured from the pitch pressure orifices (see Figure 79)
At this pitch angle, with preestablished velocity pressure co-
efficient and total pressure coefficient curves, it is possible
to determine both total and static pressure from the observed
probe pressures. Velocity components are then calculated from
total pressure, static pressure and yaw and pitch angle measure-
ments .
5. Field Test Program at Allen S. King Plant
The lack of available data defining flow angularity in
large power plant stacks prompted the planning of the initial
field test program at Northern States Power Company's Allen S.
King plant. This site, with its diametrically opposed breech-
ings (see Figure 45), was specifically selected because two
very strong vortices can often be seen issuing from the stack.
Although the 244 meter (800 ft) stack in this 600 MW coal-fired
power plant is not specifically equipped with sampling ports,
it does have 30.48 cm (12 inch) hatches at 61 meter (200 ft)
intervals to provide access to the stack interior. During an
outage, the hatch doors were replaced with cover plates con-
taining sampling ports.
Visual surveys made with the flow direction indicator
shown in Figure 77 at both the 61 meter (200 ft) and the 122
meter (400 ft) levels revealed no significant radial or tan-
gential velocity components. Light entering from the stack
outlet was sufficient to illuminate the string, making it
easily visible at all times. An additional survey, made with
the three-dimensional five-port pitot probe at the 61 meter
(200 ft) level, approximately five duct diameters above the
inlet breeching, again showed relatively uniform rectilinear
flow. With the probe tip aligned with the vertical axis of
- 122 -
-------
the stack, no measurable yaw or pitch angle was observed at
any of the test points (see Figure 81). The existence of
a single test port at each vertical level and the limited
length of the probe prohibited spanning the entire duct dia-
meter. More complete surveys were possible, especially at the
122 meter (400 ft) level, where the stack diameter has shrunk
to about 6.7 meters (22 ft), with the lighter, longer, and
more versatile flow direction indicator.
The results of these initial field measurements, indi-
cating the absence of vortex type flow fields, suggest that
the triangular shaped vaned deflector positioned in the base
of the chimney is effective in minimizing helical flow from
the interaction between the flows from the opposing breechings.
The findings are in agreement with TVA cold-flow model study
results on similar configurations (Reference 29).
6. Field Tests at TVA
In addition to the opposed breeching effects, induced
draft fans, especially when close coupled to the stack, were
cited by many engineers contacted in the preliminary survey
as a possible precursor in the formation of cyclonic action
in stacks. It should be recognized that the King Plant is
a positive pressure system and, thus, has no I.D. fans. Before
discounting the existence of severe cyclonic action in large
power plant stacks, it became apparent that additional field
tests should be conducted in systems with I.D. fans, and pos-
sibly with different breeching configurations and/or different
deflection vane structures.
Field tests were subsequently conducted at the Allen
(Memphis, Tennessee) and the Colbert (Florence, Alabama)
power stations. The Fecheimer probe, an S-type pitot tube,
and the flow direction indicater were used to evaluate the
flow fields in these two TVA Power Plant stacks.
At the Allen Plant Unit No. 2 (300 MW), again a coal-fired
positive pressure system with the multiple breeching arrange-
ment shown in Figure 13, data was taken at the 85.3 meter
(270 ft) level (13 duct diameters above the breechings). A
single vertical baffle, positioned in the base of the stack
midway between the breeching inlets, separates the opposing
gas streams. Four existing test ports allowed probe traversing
along two perpendicular stack diameters. A survey made with
the Fecheimer probe, showed the gas flow direction at all test
points to be within +_ 2 degrees of the vertical stack center-
line axis, approximately the accuracy of the test procedure.
- 123 -
-------
The 2.54 cm (1 inch) of water positive pressure in the stack
prohibited flow visualization studies at this site. The results
of a 76 point velocity survey, made with an S-tube type pitot
probe, are shown in Figure A-46. As the isopleth indicates, the
velocity near the center of the stack was found to be about
30% larger than that measured near the wall.
In the Colbert Unit No. 1, induced draft fans, close-
coupled to the stack, feed the effluent gas from the precipi-
tator through diametrically opposed breechings into the 91.5
meter (300 ft) stack - see Figure 82. Four turning vanes, shown
in Figure 83, positioned in the base of the 5.33 meter (17.5 ft)
diameter stack direct the flue gas in the stack axial direction,
while minimizing interaction from the opposing gas streams. Tests
were conducted at the 30.5'meter (100 ft) level, approximately four
duct diameters above the breeching inlet plane. This site was
specifically chosen because of reported difficulty in obtaining
agreement between total volumetric flow from pitot tube surveys
and boiler input data.
Radial traverses with the Fecheimer probe were made in
each of the four test ports (see Figure 82). Only in the
northeastern quadrant of the stack (Port #1), where yaw angles
from -9° to +5° were measured, was significant flow irregularity
observed. No measurable deviations from the axial direction were
found in the southern half of the stack (Ports #2 and #3) and
only very slight, again barely detectable, deviations were found
in the northwestern quadrant. These findings were further con-
firmed with visual observations using the flow direction indica-
tor shown in Figure 77.
The above findings suggest the existence of one of the
four secondary flow vortices shown in Figure 11. It's not
immediately apparent why this spiraling is confined to only
one quadrant of the stack. The I.D. fans and the 90° elbow
immediately upstream of the stack entrance might be expected
to induce a skewed velocity profile, with the highest velocity
near the north wall of the inlet breeching. This might, in
part, explain why flow irregularity was greater in the northern
than the southern half of the stack. Possible imbalance in the
flow rates from the two breechings might further explain the
nonsymmetry in the eastern versus western sides of the stack.
The velocity distribution, again measured with an S-tube
type pitot tube, is presented in Figure A-50. Sampling in
accordance with Method 2, assuming no rotational flow, the
S-tube was traversed along two perpendicular lines passing
through the stack center with the pressure taps aligned in
the axial stack direction. Assuming that the probe is yawed,
but not pitched with respect to the actual velocity vector,
- 124 -
-------
perhaps a valid assumption, since the Port #1 traverse appears
to pass through the core of the observed vortex, from
Figure B-ll, it is apparent that even for a yaw angle of +9°,
the measured velocity would be only about 2% higher than the
actual velocity magnitude. Errors in the measurement of the
face velocity will be a combination of the above error and the
cosine of the angle between the flow and the S-tube (see
Section V-E-2). Applying this factor, for a yaw angle of + 9°,
the net total error in face velocity would thus be reduced to
the 1% level.
7. Conclusions and Recommendations
Although incomplete,'the results of this investigation of
cyclonic motion in large power plant stacks suggest that:
(a). Little qualitative data is available defining flow
angularity in large ( > 100MW) power plant stacks.
(b). Visual observations of the plume provide little
information about the in-stack gas flow behavior.
Field tests confirm that the twin-spiraling
vortices often seen issuing from stacks are the
result of secondary flow effects generated by
the bending of the gas stream by the prevailing
crosswind.
(c). In spite of the fact that very complicated exhaust
breeching and stack inlet duct configurations are
common, severe cyclonic motion has not been ob-
served in large power plant stacks. Since exces-
sive draft losses and other detrimental effects
associated with irregular flow fields can not be
tolerated, guide vane structures (which minimize
effects of jet interactions, flow separation,
etc.) have become an integral part of modern day
power plant installations. Cold flow modeling of
the stack and breeching inlet configurations is
frequently used in plant design.
(d). For a stack having diametrically opposed inlet
breechings, the best source sampling port orienta-
tion would appear to be at 45° with respect to the
inlet ducting centerline (see Figure 82). In the
event that either the single spiral (see Figure 14)
or the 4 vortex cell arrangement (see Figure 11) were
present, possibly in older systems with inadequate
deflector baffling, the probe would always be
traversing through the core of the vortices and
- 125 -
-------
thus the probe would be yawed but not pitched with
respect to the flow. With the S-tube type pitot
probe, yaw errors are smaller and more predictable.
- 126 -
-------
FIGURE 72. TWIN SPIRALING VORTICES OBSERVED AT
ALLEN S. KING PLANT
- 127 -
-------
Probe Traverse Axis
u
3 = yaw angle
S-Tube
Strut
Axis
Plane_L S-Tube
Strut Axis
U =Radial Velocity Component
U. =Tangential Velocity Component
tf = Axial Velocity Component
z
U" = Actual Duct Velocity
FIGURE 73. Single Spiral Flow
- 128 -
-------
Port #1 I
/Port #2
u
U = Circular Velocity Component
U = Axial Velocity Component
S-Tube Strut Axis
a = Pitch Angle
3 = Yaw Angle
FIGURE 74. Double Helical Flow
- 129 -
U
-------
c
-H
tn tn
G C
•H -H
H 5
o o
O rH
U fe
c w
0) ra
X! O
M O
<1) -P
cn c
cn
OJ
QJ
•H
P3
o c
rH -H
CM £1
J-i
rH 3 .
«J tt ^
§cQ!
•HO
W
-------
5.5m
r
YAW
PRESS
TAPS
o
7.6 on
0.8 cm dia
~T~
3.8 on
J
Section A-A _, .
String
FIGURE 76. FECHEIMER TYPE PITOT PROBE
.32c
Probe Strut
3.2 cm
6.4 m
131 -
FIGURE 77. VISUAL FLOW DIRECTION INDICATOR
-------
AP
AP
(Inches of
-T
-—Dynamic Pressure
35 -30 -25 -20 -15 -10 -5
5 10 15 20 25 30 35
FIGURE 73 Fecheimer Pitot Probe Sensitivity (Yaw Angle)
- 132 -
-------
PITCH
PRESS _
TAPS
YAW
PRESS
TAPS
3.2 en
25.4 cm
6m
- 133 -
FIGURE 79. CONE-TYPE FIVE-HOLE PITOT PROBE
-------
r~
UNIFORM
FLOW
<^>
te
•+*
0 00
AP
I 1
A
DYNAMIC PRESSURE
-20 ~ "-Is — -io ' -
B, degrees
Ap (INCHES OF H 0)
2
-- + .04
.--.04
FIGURE 80. FIVE-HOLE PITOT PROBE SENSITIVITY (YAW ANGLE)
- 134 -
-------
g
Hi
u
<
EH
10
H
EH
<
EH
CO
iH
« 0)
W >
O 4J
J3 O
H 0
• o
CO O
-------
4J
C
•H
O
-M
tn
0)
.c
O
fO
w
-P
m
c
(U
>
-H
O
•P
O
rH -"
<^
1
X
V
x1
'o
x X
x-*-V>
x^
•o
'o x
0 x
0 ^
°* *o ^
•* • o ^,
X yp O X
x°.
-P
u
U
fO
O
2
4J
•H
C
-P
^
0)
O
U
C
•H
-P
•rH
"H
O
rH
Pn
CN
00
- 136 -
-------
EH
12
w
H
C/3
W
CQ
(J
O
u
M
§
D
EH
CO
3
ro
oo
H
- 137 -
-------
F. Flow Angularity in Large Power Plant Breechings
In general, significant cyclonic flow in large power plants
has not been found. Therefore, the cyclonic flow in large power
plants is not considered a problem or obstacle for emission
measurements. Measurements made downstream of I. D. fans have
shown large gradients in the magnitude but not direction of
the velocity vectors at given duct cross-sections. In spite
of the fact that ducting is frequently very complicated,
internal guide vane structures (used in modern systems to
minimize costly draft losses) tend to inhibit separation,
secondary flow vortices, and other effects which contribute
to flow stratification. Field and model tests have shown
these devices to be very effective in keeping the flow parallel
to the axis of the flue. Furthermore, square ducts, which
are popular in large power plants, are much more effective
than round flues in damping out helical flow patterns.
- 138 -
-------
VI S-TUBE AERODYNAMIC INTERFERENCE STUDY
A special treatise titled "Aerodynamic Effects on Velocity
Measurements with an S-Tube in EPA Methods 2 and 5", included
in the Appendix Section B, describes in detail the S-tube
calibration studies conducted as a part of the present contract.
- 139 -
-------
VII SAMPLING STRATEGY
A. Background
The existence of severe velocity and/or particulate
concentration gradients such as those observed downstream of
the King Plant precipitator and Black Dog I. D. fan were
speculated to have a substantial effect on the sampling
strategies required for extracting a representative sample.
This concern prompted a study to determine methods of assuring
accurate measurement of the average particulate concentration
level and the total volumetric flow rate in these nonuniform
flow fields which would simultaneously minimize the cost and
effort required to obtain these measurements. Attention was
focused on correlating total number and location of the test
points with sampling performance. Particular consideration
was given to the errors that might be introduced in the eval-
uation of the emission levels using present EPA Method 1
procedures.
Before an effective sampling strategy can be formulated,
severe nonuniform or irregular flow fields in large power
plants must be defined. The 32 test profiles presented in
Appendix A obtained from field tests, model studies, a liter-
ature search, and from telephone and personal contacts provide
a good representation of the extreme conditions that might be
encountered downstream of the control devices in large power
plant ducting. A computerized technique was developed which
allowed us to quickly and accurately analyze large numbers of
sampling strategies on these 32 typical distributions. Using
these computer and empirical techniques, some general trends
were discovered which allowed confidence levels to be esti-
mated for various sampling schemes.r
B. Computerized Evaluation
A computerized multiple-regression curve-fitting routine,
which evolved from two commercially available programs (see
References 40 and 41) was used to fit a polynomial equation
to field test data. The independent variables in this equation
were the two duct coordinates plus higher order combinations
of these two. The program recalculated the values of the
dependent variable according to the regression equation and
outputed them. These theoretical distributions, with defining
polynomial, become a very useful tool in evaluating the
performance of various sampling strategies. A given sampling
scheme can be tested by comparing the overall emission level
determined from the calculated values (using the defining
polynomials) of the particulate concentration and velocity at
- 140 -
-------
the selected sample sites with the actual emission level
determined by integrating the polynomial over the cross-
sectional area of the duct.
Our initial step-wise multiple regression program, a
modified version of that described in Reference 41, used up
to 100 bi-variate data points and calculated (least—squares
equation) the coefficients of a polynomial of nine terms (up
to third order in each of two independent variables). Unfortun-
ately, this program tended to "smooth" the gradients found in
mechanically disturbed flow fields because the polynomial
calculated was only third order, and not all combinations of
the two independent variables were used.
Because the above method appeared to show promise, an
improved program was written which calculated the coefficients
of an up to 49 term polynomial in two independent variables
(all combinations up to sixth order in each independent
variable). The new program, an adaptation of a commercially
available multiple-regression curve-fitting program (Reference
43), is a least squares curve fitting routine which fits the
dependent variable (z) and the first independent variable (x)
for each value of the second independent variable (y). This
yields a series of polynomials of the form:
o c
z, = a, + b,x + c,x + + g»x
*\ C
Z2 = a2 + k?X + cpx + + 92X
2 6
z, = a, + b,x + c.,x + + g,-x
b b b b b
The constants from these equations are then made functions of
the second independent variable (y), as shown below:
2 6
a = A + A, y + A2y + . . . + Agy
b = BQ + BlY + B2y2 + . . . + B6y6
g = GQ + Gxy + G2y6
- 141 -
-------
The final polynomial is determined by substituting the coef
ficients from the above equations into the general equation
form;
Z = A "H A V *f* A V 4" A V
o 1* 2* 6
+ (B + B,y + . . . B^y ) :X
O X O
With up to 84 data inputs and 49 total terms used in the
polynomial fitting process, actual distributions were simulat-
ed very closely.
The approximation polynomials were inputted to a second
computer program which generated a matrix of any chosen
dimensions, computed the concentration as determined by the
polynomial at the centroid of each of the matrix areas, com-
pared the average of these values to the integral of the
polynomial over the entire duct cross section, and printed the
percent error. This technique allowed rapid and accurate
evaluation of any sampling plan which consisted of a matrix of
equal areas. A sample of the output from this program is
shown in Table A-l.
A polynomial root solving computer program was also
devised to permit more accurate presentation of the velocity
and particulate distributions being studied. For a given
value of the first independent duct coordinate (Y), the
program determined all values of the second independent
coordinate (X) for specified values of the independent variable
(velocity or particulate concentration). By specifying a
large number of (Y) values and a large number of values for
the dependent variable, the program provides a complete
package of data for easy and accurate isopleth construction.
This procedure greatly reduces the amount of required interpo-
lation and extrapolation and, thus, reduces the overall
uncertainty.
The analysis of the data obtained from the computer
model was broken down into two categories: rectangular ducts
and round ducts. Though most of the methods used were the
same, the geometric difference required some variation in
the analysis.
- 142 -.
-------
C. Rectangular Ducts
According to present EPA Method 1 requirements, an
acceptable sampling scheme for rectangular ducts consists of
dividing the test duct cross section into as many equal
rectangular areas as traverse points, such that the ratio of
the length to the width of each elemental area is between one
and two. According to this sampling criterion, the traverse
points are to be located at the centroid of each equal area.
To test the proficiency of the above test procedure, 48
different equal area sampling schemes with from 1 sample
point (1x1 sample matrix) to 48 sample points (6x8 sample
matrix) were tested on the 26 rectangular duct distributions
presented in Figures A-l to A-26 by the computerized methods
described above. Results from these test cases are presented
in tables A-l to A-26. Accompanying each table is a listing
of the polynomial and an isopleth of the simulated
distribution. Since the polynomials are fit to a limited
number of data points (up to 84), the resulting profiles are
referred to as simulated distributions. With the exception
of one or two theoretical distributions, the velocity and
particulate concentration distributions being studied
represent real profiles observed downstream of precipitators,
air heaters, induced draft fans, internal bracing, ducts
containing composite and very closely connected components
(expansions, contractions, miter elbows) with and without
turning aides, and various other mechanical disturbances.
In analyzing the error levels associated with various
sampling matrices and test profiles, extreme care must be
exercised in separating general trends, common in a majority
of the distributions, from the specific idiosyncrasies.
In large power plant breechings, rectangular ducts with
aspect ratios greater than 2.0 are frequently encountered.
Recognizing that normalized velocity and particulate
concentration gradients are often equally as large across the
X direction of the duct as across the Y direction, it is
reasonable to anticipate that optimum sampling performance may
be obtained with a near equal number of divisions in each
direction. This hypothesis is generally supported by the data
presented in Tables A-l to A-30. For a given sample size,
sampling arrangements having the most balanced number of
sampling points in each direction typically provided the most
accurate results.
The above findings prompted the defining of aspect parameters;
- 143 -
-------
c = Elemental Sample Area X dimension/Y dimension
Duct Area X dimension/Y dimension
where the X and Y dimensions of the elemental sample cross-
sectional area and the duct cross-sectional area are measured
in the same directions, respectively. ,
This parameter allows the discrimination between sampling
strategies containing the same number of points (i.e., 2 x 6,
3x4, 4x3 and 6x2 sample matrices contain 12 points each,
but their aspect parameters are .33, .75, 1.33 and 3 respect-
ively) . The aspect parameters (S) for each of the 48 dif-
ferent sample matrices under investigation are presented in
Table A-31.
Figures 86 and 87 show the error level as a function of
aspect parameter and sampling scheme for the distributions
shown in Figure A-l and A-10. These and other plots of
aspect parameter vs. the absolute value of the percent error
indicated three general regimes of accuracy. With two major
exceptions (3x3 and 4 ±z; 3 sample matrices consistently
yielded errors less than 5%) sampling schemes with 12 or less
total traverse points generally have large erratic and un-
predictable error levels (both negative and positive) with
little observable correlation between the accuracy of measure-
ment and the number of sample points. Only when the gradients
are small or linear would these small sampling grids be ex-
pected to produce consistently accurate results.
Sampling arrays with 15-24 points and aspect parameters
0.375 ^ S < 2.67 have smaller more uniformly distributed error
levels; many strategies have errors less than 2% with an
apparent trend for the maximum accuracy, for a given number of
sampling points at S = 1.0. As the histogram in Figure 84
illustrates, in twenty of the twenty-one velocity profiles
tested, a 16 point (4x4) sampling matrix provided a volu-
metric flow rate within 1% of the actual. Applying the same
16 point sampling scheme for evaluating the average particu-
late concentration for the five test profiles, only for the
very skewed theoretical distribution presented in Figure A-5,
was the error level greater than 2%. Even with the skewed
distribution observed downstream of the Allen S. King plant
(see Figure A-2) where there is an order-of-magnitude increase
in particulate concentration from the top to the bottom of
the duct, the error associated with a 16-point survey was less
than 1%. This behavior supports the theory based on this work
- 144 -
-------
that errors in measurement of flow in flow fields with large-
scale gradients are equally dependent on each equal area ele-
mental sampling and duct dimension. A high level of confidence
for a given traverse scheme can be held in flow and particulate
samples taken according to this regime.
The third regime (above 24 points with at least 3 rows of
sampling points in each direction) contains samples with
errors typically less than 1%. There is good correlation
between the number of sample points and the accuracy of the
measurement, though in many cases, a 50% increase in sample
size (32 to 48 points) yields only a small increase in
accuracy (1% to 0.5%). Small-scale gradients, which can be
accounted for only with a large number of sampling ports in
the one direction and not in the other direction, might be
one explanation for this behavior. Due to the increased cost
of large numbers of sampling points and the presently required
5 minutes minimum time at each sampling point, the law of
diminishing returns may preclude sampling in the regime of
greater than 24 points.
These findings suggest that sampling schemes with
elemental areas having aspect ratios similar to that of the
test duct provide more accurate results. The histogram shown
in Figure 85, listing error in volumetric flow rate versus
percentage of the 21 velocity profiles studied with the given
error level for four different sampling schemes, all with 12
total traverse points, again shows the minimum error with the
most balanced matrices. These comparisons suggest that even
for very skewed arrangements, 15-24 point surveys with at least
3 rows of sampling points in each direction provide a surprisingly
accurate evaluation of the true emission level. Present EPA test
methods allow sampling regimes with one or more rows of sampling
points in one direction. Many regimes with only one or two rows
of sampling points in one direction and up to eight in the oppo-
site direction which were evaluated in this work yielded errors
in excess of 5%. However, all of the sampling regimes with
three or more sampling points in each direction yielded errors
less than 5%, and all regimes with 4 or more in each direction
yielded errors less than 1.56%. Based on these data, the de-
scribed nine (3 x 3) point sampling regime should assure a
representation of the particulate matter within 5% accuracy.
The performance of various sampling strategies in simul-
taneously providing an accurate determination of the average
particulate concentration level and the total volumetric flow
rate were also studied. In four of the test cases, velocity
and particulate concentration data were obtained simultaneously.
The two distributions were combined and the total emission levels
analyzed in the same manner as the individual velocity and parti-
culate concentration distributions above. These results, presented
in Tables A-27 to A-30 lend further support to the findings pre-
viously presented.
- 145 -
-------
ioo-
80-
60.
40_
w
•H
O
O
CM
m
O
20_
80-
4J 60 H
w
QJ
Ei
40
20-
0-
CnlOO
g 80
o
m 60
40-
20-
3x3 Sample Matrix (S=1.0)
4x4 Sample Matrix (S=1.0)
T
1
6x8 Sample Matrix (S=.75)
^
I1 2 3 5 5
Percent error (Absolute Value) in measured
total volumetric flow rate.
FIGURE 84 ERROR IN VOLUMETRIC FLOW RATE VERSUS
PERCENTAGE OF THE 21 VELOCITY PROFILES STUDIED
WITH THE GIVEN ERROR LEVEL FOR THREE DIFFERENT
SAMPLING SCHEMES
- 146 -
-------
40 _
20 _
0 -
w
c
0
3 60 -
I-J
•H i40 ^
H 1
.p
w
-H 20 -
t>, n
t (Velocit
' CTi
0 C
UJ
0
o
2x6 Sample Matrix (S=.33)
i 1 I 1
0 ]l 2 1 4 5 (75)
3x4 Sample Matrix (S=«75)
i a
0 1 2 3 4 5 (75)
4x3 Sample Matrix (S=1.33)
i i r ^
012345 (75)
6x2 Sample Matrix (S=3.00)
1 1
1 , - - -- ,
0 1 i i 4 5 (75)
Percent Error (Absolute Value) in Total Volumetric
Flow Rate
FIGURE 85 ERROR IN VOLUMETRIC FLOW RATE VERSUS PERCENTAGE
OF THE 21 VELOCITY PROFILES STUDIED WITH THE
GIVEN ERROR LEVEL FOR FOUR DIFFERENT SAMPLING
SCHEMES, ALL WITH 12 TOTAL TRAVERSE POINTS.
- 147 -
-------
0.1
.48
.42
.15
/
J2. * •
20
• '2
9
Present EPA
Method 1 Range
Aspect Ratio
of Elemental
Area=l.0-2,0
(either direc-
tion)
I I
i
.16
8
10
• „ \
Total Number
of
. Sampling Ports
z
0.1
0.3
0.5
i.o
3.0
5.0
10,0
0)
3
iH
(0
-P
P
rH
0
tn
30.0
-50.0
o
j-i
M
W
C
0)
u
M
D
0.3 0.5 1.0
Aspect Parameter (s)
3,0 5,0
10.0
FIGURE 86. SAMPLING ERROR VERSUS ASPECT PARAMETER FOR
VELOCITY DISTRIBUTION SHOWN IN FIGURE A-10
- 148 -
-------
40 4ft
32 •
260 3*0
•BO
rf
\
\
• -18
10
.4
8
-------
D. Round Duct Sampling Strategy
Round duct data (see Figures A-30 to A-50) was analyzed
in much the same manner as the rectangular duct data. Using
the conversions;
2
y = r
x = 0/360°
and the curve fitting computer program used in previous
rectangular duct analyses, polynomials were obtained for the
six test velocity profiles.
These conversions allow the round duct data to be
treated as if it had come from a rectangular duct with a line
y = 0 corresponding to the center of the duct, the line y = 1
corresponding to the wall, the line x = 0 corresponding to 0
= 0° and the line x = 1 corresponding to 0 = 360°. To check
the feasibilty of this treatment of the data, the polynomials
were evaluated at several values of r, at 0 = 0°,
0 = 360° and at r = 0 for several values of 0. This check
shows that there were no discontinuities due to this
treatment, the velocity profiles as described by the
polynomials were the same at 0 = 0° and 0 = 360°. There were
occasional small variations in velocity (less than .5%) at r
= 0 for various values of 0. But since at r = 0 the area is
equal to 0, so there is no appreciable error in the value of
the integral of the polynomial, and, because none of the
sampling strategies studied included a traverse point at the
center of the duct, this variance was disregarded.
The computer program developed for testing equal area
sampling strategies was modified to include a routine for
evaluating the performance of sampling matrices with 2, 4,
and 8 rays symmetrically arranged in the duct, and with 1 to
8 sample points per ray. The average velocity for each
strategy was calculated and compared to that determined by
integration of the polynomial, and percent error was printed.
With this computer model, the effects of the number of rays
(test ports), the number of sample points per ray, and the
angular orientation of the sampling rays with respect to the
velocity profile can be analyzed.
In Figures A-30 through A-50, the effect of angular
orientation with respect to the velocity profile is shown for
sampling strategies with 2, 4, and 8 rays, having 2, 4, 6 and
8 sampling points per ray, for the six velocity profiles
studied. The six specifically selected test profiles are
thought to provide a good representation of the extreme
- 150 -
-------
conditions existing in large power plants ( >100 MW
capacity). The error levels (range associated with various
orientations - errors calculated at 15° intervals) for each
of the above sampling schemes are tabulated in Tables 6 to 9.
Since several of the test distributions have large angular
gradients (see Figure A-50), in addition to radial gradients,
two ray sampling matrices, even with 16 sample points,
frequently produced error levels of 3-7%. On the other
extreme, with 8 rays-16 point surveys, large errors (3-4%)
were frequently produced by the limited radial exposure (2
points per ray). The more balanced sampling arrangements
were found to be surprisingly efficient for all six profiles
and all orientations, a better compromise than had been
anticipated. For example, with a four ray-16 point survey,
the largest error observed in volumetric flow rate was only
2.7%. The apparent significance of the aspect parameter
defined for rectangular ducts prompted the definition of an
analogous round duct parameter;
_ Number of Rays
~ Number of Points per Ray
This parameter again permits discrimination between sampling
matrices with equal numbers of sample points. When percent
error (absolute value) is plotted vs. T, the greatest
accuracy for a given number of points was obtained by those
strategies for which T < 1, suggesting that gradients in
round ducts are more dependent on radius than on angle. This
finding is supported by plotted isopleths of the
distributions studied, which appear in general to be series
of skewed concentric circles . The data presented in
Figure 88, a plot of sampling error versus T for three
different 16-point sampling surveys for the velocity
distribution shown in Figure A-50 indicates optimum accuracy
for T=l.
In actual field testing, the value of one of the
parameters which define the aspect parameter or the round
duct parameter will most likely be fixed by the number and
location of the test ports. The analysis of sampling strategy
accuracy has shown that, in any case, the number of sampling
points measured through each test port should be not less
than the number of test ports. Though the cost of sampling
at additional points in a given test port is generally not
large, the improvement in accuracy with additional points
above four points per ray is often very slight.
- 151 -
-------
TABLE 6
Effectiveness of Various 2-Ray Sampling Strategies
in Round Ducts
Test
Profile
(Figure)
# A-30
# A-34
# A-38
# A-42
# A-46
# A-50
Maximum
Error
% Error (Absolute) in Total Volumetric Flow
Rate - Range for Different Port Orientations
2 Points
Ray
1.3-6.0
2.8-3.8
.10-5.0
.30-2.0
.52-4.7
.15-5.8
6.0
4 Points
Ray
.68-2.6
.78-2.2
.10-5.5
.10-1.6
.30-2.8
.29-6.8
6.8
6 Points
Ray
.25-2.2
.16-1.5
.10-5.0
.10-1.5
.10-3.1
.47-7.0
7.0
8 Points
Ray
.12-2.1
.10-1.3
.20-4.7
.10-1.5
.10-3.2
.55-7.0
7.0
Data Presented in Figures A-30 to A-50
- 152 -
-------
TABLE 7 '
Effectiveness of Various 4-Ray Sampling Strategies
in Round Ducts
Test
Profile
(Figure)
v A- 30
# A-34
# A-38
S A-42
* A- 4 6
« A- 50
Maximum
Error
% Error (Absolute) in Total Volumetric Flow
Rate - Range for Different Port Orientations
2 Points
Ray
3.4-4.1
3.0-3.3
.10-2.6
.10-. 57
1.8-3.3
.78-2.1
4.1
. 4 Points
Ray
.32-. 86
1.1-1.5
.10-2.7
.10-. 27
.10-. 79
.10-1.8
2.7
6 Points
Ray
.10-. 49
.44-. 86
.12-2.3
.10-. 26
.10-. 88
.15-1.9
1.9
8 Points
Ray
.10-. 38
.20-. 62
.30-2.2
.10-. 25
.10-. 94
.21-2.0
2.2
Data Presented in Figures A-^30 to A~50,
- 153 -
-------
TABLE 8
Effectiveness of Various 8-Ray Sampling Strategies
in Round Ducts
Test
Profile
(Figure)
# A-30
# A-34
# A-38
# A-42
# A-46
# A-50
Maximum
Error
% Error (Absolute) in Total Volumetric Flow
Rate - Range for Different Port Orientations
2 Points
Ray
3.7-3.9
3.1-3.2
.70-1.4
.27-. 33
2.6-2.9
.61-1.3
3.9
4 Points
Ray
.53-. 63
1.2-1.3
.71-1.5
.10-. 10
.10-. 41
.20-. 53
1.5
6 Points
Ray
.17-. 27
.55-. 66
.29-. 98
.10-. 10
.15-. 21
.33-. 39
.98
8 Points
Ray
.10-. 16
.31-. 41
.10-. 79
.10-. 10
.12-. 21
.27-. 45
.79
Data Presented in Figures A-30 to A-50.
- 154 -
-------
TABLE 9 ! ,
Comparison of Various 16-Point Sampling Strategies
in Round Ducts
Test
Profile
(Figure)
# A-30
#A-34
# A-38
# A-42
* A-46
# A-50
Maximum
Error
% Error (Absolute) in Total Volumetric Flow
Rate - Range for Different Port Orientations
2 Ray
(16 pts)
.12-2.1
.10-1.3
.20-4.7
.10-1.5
.10-3.2
.55-7.0
7.0
4 Ray
(16 pts)
.32-. 86
1.1-1.5
.10-2.7
.10-. 27
.10-. 79
.10-1.8
2.7
8 Ray
(16 pts)
3.7-3.9
3.1-3.2
.70-1.4
.27-. 33
2.6-2.9
.61-1.3
3.9
Data Presented in Figures A-30 to A-50.
- 155 -
-------
NOTE: FLAGS INDICATE ERRORS LESS THAN 0.1%
0.1
0.2
0.4
0.6
0.8
1.0
ERPO
2.0
4.0
6.0
8.0
10.0
—
0
o
&
&
ft A A 0
o
A
3
1
9
o
1 1 1 ' 1
s
Y
M
B
0
L
O
D
o
A
A
C\
Q
0
0
&
A
O
0.2 0.4 0.6 1.0 2.0 3.0 4.0
ROUND DUCT ASPECT PARAMETER (T)
R
O
T A
A N
T G
I L
0 E
N
0
15
30
45
60
75
90
105
120
135
150
165
FIGURE 88.SAMPLING ERROR VERSUS ROUND DUCT ASPECT PARAMETERS
(RATIO OF THE NUMBER OF SAMPLING RAYS TO THE NUMBER
OF SAMPLING POINTS PER RAY) FOR THREE DIFFERENT
16-POINT SAMPLING SURVEYS FOR THE VELOCITY
DISTRIBUTION SHOWN IN FIGURE A-50
- 156 -
-------
VIII. REFERENCES
1. Sovran, G., Fluid Mechanics of Internal Flow, Elsevier Publishing
Company, New York (1967)
2. Zakak, A., et al, "Procedures for 'Measurement in Stratified Gases
Volume I", EPA-650/2-74-086 September 1974
3. Zakak, A., et al, "Procedures for Measurement in Stratified Gases
Volume II", EPA-650/2-74-086b, September 1974
4. Hawksley, P. G. W., Badzioch, S., and Blackett, J. H.,
Measurement of Solids in Flue Gases, British Coal Utilization
Research Association, Leatherhead, England, 214pp. (1961)
5. Olson, R. M., Essentials of Engineering Fluid Mechanics,
International Textbook Company, Seranton, Pennsylvania, pp 200-
256 (1964)
6. Hwang, C. C., Singer, J. M., and Hartz, T. N., "Dispersion
of Dust in a Channel By a Turbulent Gas Stream", U. S. Bureau of
Mines RI-7854 (1974)
7. Scorer, R. S., Natural Aerodynamics, Pergamon Press, New York
pp 78-82 (1958)
8. Ahmed, S., and Brundrett, E., "Performance of Turning Vanes in a
Square Conduit Elbow", ASME 69-FE-32 (1969)
9. Burton, C. L., and Willison, R. E., "Application of Model
Studies to Industrial Gas Flow Systems", ASME Annual Meeting,
Atlantic City, New Jersey, November-December 1959, Paper 59-A-280
9 pages (1959)
10. Dimmock, N. A., "Cascade Corners for Ducts of Circular Cross-
Section", British Chemical Engineering, pp 302-307, June 1967
11. Archbold, M. J. "A Visual Qualitative Approach to Duct Design for
Power Plants" Combustion, pp 34-40, April 1958
12. Wirt, L., "New Data for the Design of Elbows in Duct Systems",
General Electric Review, Volume 30 (6), June 1927
13. Opfell, J. B. and Sproull, W. T., "Limitations of Model Studies
in Predicting Gas Velocity Distributions in Cottrell Precipitators",
I & EC Process Design and Development 4, pp 173-177, April 1965
- 157 -
-------
14. Sproull, W. T., "Laboratory Wind Tunnel and Model Studies to
Improve Gas Velocity Distribution in Cottrell Precipitators",
Journal of the Air Pollution Control Association 10, pp 307-313,
August 1960
15. Zarfoss, J. R., "Ductwork Arrangement'Criteria for Electrostatic
Precipitators Without Model Study", Journal of the Air Pollution
Control Association 20 (9), September 1970.
16. Bragg, L. G., "Gas Flow Model Studies of Flues", Canadian
Mining and Metallurgical, pp 707-712, October 1962
17. "Methods for Producing Uniform Gas Flow in Processing Equipment",
British Chemical Engineering, pp 359-363, July 1957
18. Burton, C. L. and Smith, P. A., "Precipitator Gas Flow Distribution"
EPA-650/2-75-016, pp 191-217, January 1975
19. Preszler, D. L. and Lajos, D. T., "Uniformity of the Velocity
Distribution, Upon Entry into an Electrostatic Precipitator, of a
Flowing Gas", Staub 32 (11), pp 1-7, November 1972 '
20. Gilbert, G. B., "The Use of Flow Modeling Techniques to Obtain
a Minimum Loss Design for the Stack Entrance Section of a 700-ft.
Power Plant Chimney", ASME 70-WA/Pwr-l, January 1971
21. TVA Engineering Laboratory-Research 1965-66, Norris, Tennessee,
May 1967, 61-72
22. TVA Engineering Laboratory-Research 1967-68, Norris, Tennessee,
April 1969, 130-133
23. TVA Engineering Laboratory-Research 1969-70, Norris, Tennessee,
August 1971, 83-86
24. Sansone, E. B., "Sampling Airborne Solids in Ducts Following a 90°
Bend", American Industrial Hygiene Association Journal 30, (5),
pp 487-493 (1969)
25. Brown, R. L., "Some Coal Research Problems and Their Industrial
Implications," Journal Institute of Fuel, Volume 29, pp218-236 (1956)
26. American Society of Mechanical Engineers, "Determining Dust Concen-
tration in a Gas Stream," ASME Power Test Code No. 27, 25 pages,
New York, New York, 1957
27. Achinger, W. C. and Shigehara, R.- T., "A Guide to Selecting Sampling
Methods for Different Source Conditions," Presented at 60th Annual
Meeting of the APCA, Cleveland, Ohio, June 11-16, 1967, Journal of
Air Pollution Control Association, 18 (9), pp 605-609 (1968)S
- 158 -
-------
28. Miller, S., "The Building of Tall (and Not So Tall) Stacks",
Environmental Science and Technology, June 1975, pp 522-527
29. Price, J. T., "Chimney Flow Improvement", Power Engineering,
September 1967, pp 52-55
30, Burton, C. L., "Quantitation of Stack Gas Flow", Journal of Air
Pollution Control Association, 22 (8) August 1972, pp 631-635
31. EPA Standards of Performance for New Stationary Sources, Federal
Register, Volume 36, No. 247, December 23, 1971, 24376-24895
32. Smith, W. A., Wheeler, D., Olson, R. W., and Coy, D. W.,
"The Use of a Flow Model in the Design of an Electrostatic
Precipitator", Blast Furnace and Steel Plant, pp 1097-1102-,
December 1967
33. ASTM D-3154 "Standard Method of Test for Average Velocity in
a Duct (Pitot Tube Method)" 1972
34. "Flow Measurement", B.S. 1042, British Standards Institution,
London (1951)
35. "Fluid Velocity Measurement", PTC 19.5.3, American Society of
Mechanical Engineers 1965
36. Anon., "Gas/Dust Flow Studies: Experiences wtth Fly-Ash and
Simulation of Gas/Dust Flows." Advance Report No. 1, TVA
Report No. 0-6684, Norris, Tennessee, September 1967
37. Fuchs, N. A., The Mechanics of Aerosols, MacMillan Company,
New York, New York~;pp 353-367 (1964)
38. Hamil, Henry F., et al, "Collaborative Study of Method for the
Determination of Paniculate Matter Emissions from Stationary
Sources (Fossil Fuel-Fired Steam Generators)", EPA 650/4-74-021
36 pages June 1974
39. Walker, A. B., "Emission Characteristics from Industrial Boilers",
Air Engineering 9 (8), pp 17-19 (1967)
40. File Name-BSTEP/, System Library Manual, International Timesharing
Corporation, Chaska, Minnesota, 1973
41. Efroymsen, M. A., "Multiple Regression Analysis", Mathematical
Methods for Digital Computers , Part V, (17), Edited by A. Rolston
and H. S. Eilf, Wiley (1970)
- 159 -
-------
42. Liess, C., "Experimental Investigation of Film Cooling! with
Ejection from a Row of Holes for the Application to Gas Turbine
Blades", Journal of Engineering for Power, January 197*5
43. Sommerlad, R. E., Zoldak, F. D., McMillan, R. E.-and Karg,J.S.,
"The Experience of EPA Standards of Performance Tests with
Gaseous and Liquid Fuels on a Steam Generator", ASME 74-WA/APC-l
November 1974
- 160 -
-------
APPENDIX A
DATA FROM THE
EVALUATION OF VARIOUS EQUAL AREA
SAMPLING STRATEGIES
- 161 -
-------
TABLE A-l
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-l
Number of traverse points along x axis
Number of traverse points along y axis
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1x1
1x2
2x1
1x3
3x1
1x4
2x2
4x1
1x5
5x1
1x6
2x3
3x2
6x1
1x7
1x8
2x4
4x2
3x3
2x5
5x2
2x6
3x4
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
- 162 -
-4.89
2.75
-3.58
3.44
-3.56
3.48
.49
-3.56
3.45
-3.57
3.43
.45
.37
-3.57
3.41
3.40
.38
.35
.26
.33
.35
.30
.18
-------
TABLE A-l (Cont,)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
4x3 12 .22
6x2 12 .35
2x7 14 .29
3x5 15 .14
5x3 -15 .20
2x8 16 .28
4x4 16 .14
3x6 18 .11
6x3 18 .20
4x5 20 .10
5x4 20 .12
3x7 21 .10
3x8 24 .09
4x6 24 .08
6x4 24 .12
5x5 25 .09
4x7 28 .06
5x6 30 .06
6x5 30 .08
4x8 32 .05
5x7 35 .05
6x6 36 .06
5x8 40 .04
6x7 42 .04
6x8 48 .04
- 163 -
-------
<(<((((
>>>>>>>
*********
C C ( C < < C C C ( (
»>»>»>»
***********
M M M M M
**-*"M
COCMrj-OVO'*'HMCMinMff»in^NNCO^-OOtGJ
ftMftCMCMNCMCaCSCJ^Q^HOOCMVOCJMCOM
Q^IS.^-QftrHOCOtfMODNinOCMMinMVON
CJoSMTHCOrr^^^MHCMVOlririVDlricdNHlri
MMODMHCOCnM^CMHCTtrHCSinN^-rrCafflVO
CMM^-MO(MOOOCOlf)
HMCSOONrHtOHNOOHCSOTH I H N t 00 ft O
I CM M ft m VO 00 OJNCMVO^-m riTrNVO^-CM
IM'S-ftVOOO MN^-T*CS> I^VOMCM^1
HCMCM ,HMMV .^H
XI
(0
CO
c
CO
4J Ci
-C 0)
CO 73
•H C
O H
-H
d) M Cij
O -H X
U CM W
0)
rH
XI
4J
c
Q)
fi
Q)
CD
T!
C
H 4->
c
o
o
cu
CO
w
P
o
H
H
o
C/l
o
H
P
H
tf
O H CM
m
Q
M
C C C C C ( C
*******
VO CO H CM M t in
(D-HrlrlTl'HTlTfCMCMCM
M O
N H
M M
..voovovoincMN
I ^ CM 00 00 N O M
I VO N m ® N
CM in in H
i i
C C C ( ( C
» » > > -
********
i-i-
CM
C C C C C C C C C C C C C < C C C C C ( C
xxxxxxxxxxxxxxxxxxxxx
*********************
QVOCMfMCOVOCMCainN-THMftVOftOftHftNxJ-
CaWMMQCOmCMCMCOVDftCJCSM-dftCQ^-MfM
S>QHCO^'r-l'
c
0)
T3
c
O)
m 4J
c
Q)
c
X
c
o
U
CO
Q)
c
o
X
H
P
w
o
CM
H
a
o
- 164 -
-------
Gas Velocity
["ft/sec
|_m/sec
FIGURE A~l Velocity Distribution Simulating Profile Measured
Upstream of Test Elbow (80° Mitered Bend) -
Allen S. King Plant
- 165 -
-------
TABLE A-2
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE PARTICULATE DISTRIBUTION'PRESENTED IN FIGURE
A-2
— Number of traverse points along x axis
j-—Number of traverse points along y axis
4x3
% Error in Measured
Total Number of Average Particulate
Matrix Traverse Points Concentration
1 X
1 X
2 x
1 x
3 x
1 x
2 x
4 x
1 x
5 x
1 x
2 x
3 x
6 x
1 x
1 x
2 x
4 x
3 x
2 x
5 x
2 x
3 x
1
2
1
3
1
4
2
1
5
1
6
3
2
1
7
8
4
2
3
5
2
6
4
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
- 166 -
-16.
-5.
-14.
*
-14
1.
-6.
-15.
1.
-15.
1.
1.
-8.
-15.
1.
1.
2.
-7.
•
2.
-7.
1.
•
02
20
74
80
.9
39
91
0
39
0
23
93
04
06
12
07
52
93
05
25
77
23
86
-------
TABLE A-2 (Cont)
% Error in Measured
Total Number of Average Particulate
Matrix Traverse Points Concentration
4x3 12 -.11
6x2 12 -7.66
2x7 14 1.71
3x5 15 .75
5x3 15 -.11
2x8 16 1.50
4x4 16 .70
3x6 "18 .54
6x3 18 -.03
4x5 20 .59
5x4 20 .70
3x7 21 .38
3x8 24 .27
4x6 24 .43
6x4 24 .75
5x5 25 .64
4x7 28 .32
5x6 30 .48
6x5 30 .70
4x8 32 .16
5x7 35 .32
6x6 36 .54
5x8 40 .21
6x7 42 .38
6x8 48 .27
- 167 -
-------
>»»»>»»>»»»>>>»
************************
xxxxxxxxxxxxxxxxxxxxxxxx -^
************?***********
N 00
in N
^E- ,
*•-
(t)
l-i
Q
(0
•H
"^ ' !-l
(0
.p
G
•<- G)
•a
c
0)
•p PI
C
G
0)
n3
c
CD
O-i
Q)
rfl
Q) 'C G
•H C -P H 4-
-er— O H G f
•H Q) rd C
"4-4 -P G G f
m to o o c
Q) H CX. U £
O -H X 0) >
U fe W W P
G
0)
G
a
a
0)
•a
U Pn W Ul PI
H
S
O
o
- 160 -
-------
Particle Concentration
fgr/scfd)
Igm/Nm3 J
FIGURE A-2 Particulate Concentration Distribution Simulating
Profile Measured Downstream of the Electrostatic
Precipitator at the Allen S. King Plant
- 169 -
-------
TABLE A-3
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-3
r—Number of traverse points along x axis
I Number of traverse points along y axis
4 x
r
3
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1 X
1 X
2 x
1 x
3 x
1 x
2 x
4 x
1 x
5 x
1 x
2 x
3 x
6 x
1 x
1 x
2 x
4 x
3 x
2 x
5 x
2 x
3 x
1
2
1
3
1
4
2
1
5
1
6
3
2
1
7
8
4
2
3
5
2
6
4
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9 '
10
10
12
12
- 170 -
10
3
5
4
5
4
5
3
5
3
5
3
3
.54
.92
.86
.14
.82
.04
.79
.87
.95
.90
.88 '
.78
.62
.92
.84
.81
.62
.58
.57
.52
.57
.45
.41
-------
TABLE A-3 (Cont)
% Error in Measured
Total Number of Total Volumetric
Matrix . Traverse Points Flow Rate
4x3 12 .52
6x2 12 .57
2x7 14 .41
3x5 15 .31
5x3 15 .51
2x8 16 .38
4x4 16 .37
3x6 "18 .24
6x3 18 .50
4x5 20 .27
5x4 20 .35
3x7 21 .20
3x8 24 .17
4x6 24 .20
6x4 24 .35
5x5 25 .25
4x7 28 .16
5x6 30 .19
6x5 30 .25 -
4x8 32 .13
5x7 35 .14
6x6 36 .18
5x8 40 .12
6x7 42 .14
6x8 48 .11
171 -
-------
in v> N oo
* * * *
*#***************£#*
WNNNNMMMMMMMMMMxr-*-*'***1***-*
(
((((((((((((((((((((<((.
XXXXXXXXXXXXXXXXXXXXXXXX.-
*******#******#**********
V) IO ® O CO 0^
^* y) O O O CPk
f^ O CD ® O 00
V) O (9 ® O 00
8&«»*S
83
N I
Q
Q
G)
vo m
rim
Vf 1
1 (9
o» oo
in l
m w
O M
00 U)
N PI
U)-HlOinCOC3®(3MNCOCS'tSOCSQC9CSninCpcnS)
N N VO y) *J
in N u> in c
ON in
8
Cfi
M
CO
N CM • ® rH
^rooNN
II
^i 00 G9
. CO 00 N
i CD oo m
— i ut U)
4«lf>
QJ
r-i
XI
•rH
-p
d
QJ
•H
o
•H
«w
H-l
QJ
o
u
d
QJ
a
QJ
QJ
rH
X!
(0
•M
m
-p
d
QJ
QJ
a,
QJ
d -P
H d
QJ
4J d
Ul O
t-i CX
d
H -P
d
T3 QJ
d d
o o
O £X
QJ X
en w
EH
en
H
Q
A
o
fa
g
o
-------
Gas Velocity
FIGURE A-3 Velocity Distribution Simulating Profile Measured
Across Width of Duct Upstream of the Electrostatic
Precipitator at the Allen S. King Plant
- 173 -
-------
TABLE A-4
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE PARTICULATE DISTRIBUTION PRESENTED IN FIGURE A-4
x
Number of Traverse points along x axis
y—Number of traverse points along y axis
3
Matrix
Total Number of
Traverse Points
1
1
2
1
3
1
2
4
1
5
1
2
3
6
1
1
2
4
3
2
5
2
3
x
x
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
1
2
1
3
1
4
2
1
5
1
6
3
2
1
7
8
4
2
3
5
2
6
4
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
- 174 -
% Error in Measured
Average Particulate
Concentration
13.5
1.33
-.61
3.63
-1.28
3.90
1.16
-1.37
3.90
-1.39
3.86 '
.31
1.07
-1.39
3.82
3.79
.25
1.03
.05
.28
1.01
.31
-.03
-------
TABLE A-4 (Cont)
% Error in Measured
Total Number of Average Particulate
Matrix Traverse Points Concentration
4x3 12 -.02
6x2 12 1.00
2x7 14 .33
3x5 15 -.01
5x3 15 -.04
2x8 16 .35
4x4 16 -.10
3x6 '18 .02
6x3 18 -.05
4x5 20 -.08
5x4 20 -.13
3x7 21 .05
3x8 24 .07
4x6 24 -.05
6x4 24 -.14
5x5 25 -.11
4x7 28 -.02
5x6 30 -.08
6x5 30 -.12
4x8 32 0
5x7 35 -.05
6x6 36 -.09
5x8 40 -.03
6x7 42 -.07
6x8 48 -.05
- 175 -
-------
* ID \0
M*-W
M
(M
( (
M
* *
< C C < <
» » >
* * * * *
MMM**^****
XXX XX X X
*******
(( C (((((((
xxxxxxxxxx
**********
tf)COMtM<9iny>NTUf>KT«NNM1t
N*CMN.<9U)OOTHO^CMCM<COCM&QCJ'iV)Ci)V)CArlVOy)NMN
ChCOQTlOSyjO'iMCMIN.
CON-H-HNTrOM^-H
<9 CM Ok I « CM in CM O N
in (9 U) I <9 T« ^f M <9
in « H T< in <9 en M
II I Tl I
CM N
4->
C
0)
rd
_ r,
^ a)
4J CX
C 0)
C
r-
( _
J
H
1
••*•*
(y
-H
H
(0
4->
f
C
H
1
L)
-d
c
c.
i
c
U
T3
e
d
l|-l 4J
m M
<1) M
O -H
U fo
C
(U
C
O O
ft
X
o
H
o
ffi
en
s
o
O T«C4 «*• «
M^t-IOWOTtCMM
*?*
w « TT
gf\ Nn jv%
W G9 O
( ( (
XV ">^
X X
* * *
»jt
TT
Ojft
Oil
( (
Jt J^m
* *
m oo
•«• vo
t ^t
^f 00
00 (9
COM
H M
N *
1
* *
K TT
CO TW
( (
rfV J^
* *
•H CT>
tl- it
CTi CM
* tf)
HO!
t y>
00 (M
MTl
1
**
If »
• 1
Tl Tl
( (
* *
coo
* 00
Tl 00
M V)
*M
*
«x.
M
.-• ^J
Tl Tl
( (
* *
CM M
Q T<
CT» 00
N Tl
M T<
00 M
m o
1 (M
> J-
* *f
H--.1
Tl
( (
X%p
.A.
* *
CT> kD
y> •*
CT» N
CTt U)
S!G
S3
i
j- j-
— —
i ^i
Tl Ol
( (
XV'
Jt
* *
CTkOO
VO Q
in •*
m Q
Q t
!?
* *
M (M
ill *M
( C
Xy
^s.
* *
in u)
00 M
N CM
CM VO
IS
"8
1
*
if
rj
Ol •
(
*
CM-=£ — i
00
a *
^ s
* •
y> c
S ••"
CM q.
M-
a
c
L
OJ
rH
&
(d
-H
M
rr(
>u
4J
C
0)
1
0)
> ex
1 (U
) T3
< C 4-
) H f
< a
< 4J C
< W C
) H C
> -H >
> fa pi
c
P"
s.
a
•r
J.
n
K.
*-•
4-
c
0
T:
c
a
c
Q
T:
c
J h
^
3 T
S C
) C
3. C
< a
3 c/
i)
•1
3
H
-1
J
J
4
;
5
•t
")
i>
|
^
H 4-
C
5 a
J C
) C
) C
) X
3 fr
CQ
H
&
EH
w
H
Q
M
a
YNOMIA
176 -
-------
Particle Concentration
gr/scfd
gm/Nm-
FIGURE A-4 Particle Concentration Distribution Simulating
Profile Measured Across Width of Duct Upstream of the Electro-
static Precipitator at the Allen S. King Plant
- 177 -
-------
TABLE A-5
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE PARTICULATE DISTRIBUTION PRESENTED IN FIGURE A-5
\
jNumoer or traverse pom
jNunujer or traverse
' Y
rs aj.ong x axis
points along y axis
4x3
Matrix
1x1
2x1
3x1
4x1
5x1
6x1
Total Number of
Traverse Points
1
2
3
4
5
6
% Error in
Measured Average
Particulate Concentration
-62.8
-15.72
- 6,97
- 3.90
- 2,48
- 1.77
- 178 -
-------
c
0)
•o
c
a;
a
4-1
c
(U
• -a
c
-=
O
H
fo
is
S-a a> B
c c c ^
&§Sgs
C/3 W PL,
K4
Pi
<(((((((C((((C C
nmmmmt-
o
H
EH
CO
M
Q
W
a
PS
o
s
o
S3
(S)G)NO
-------
X
Q.QS
(0.02.2.)
(o.
Q.Q5
(0-022.)
O.>0
"(0.044)
o-eo
Particle Concentration gr/scfd
FIGURE A-5 Theoretical Distribution (Possible Distribution
Downstream of an Electrostatic Precipitator
Where Hopper Sweepage is Significant).
- 180 -
-------
TABLE A-6
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-6
i Number of traverse points along x axis
y ,r—Number of traverse points along y axis
4x3
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1x1 1 5.69
1x2 2 5.94
2x1 2 3.22
1x3 3 5.68
3x1 3 2.16
1x4 4 5.61
2x2 4 .14
4x1 4 1.97
1x5 5 5.58
5x1 5 1.93
1x6 6 5.57
2x3 6 1.07
3x2 6 -.62
6x1 6 1.91
1x7 7 5.57
1x8 8 5.56
2x4 8 1.14
4x2 8 -.65
3x3 9 . .08
2x5 10 1.11
5x2 10 -.61
2x6 12 1.07
3x4 12 .14
- 181 -
-------
TABLE A-6 (Cont)
% Error in Measured
Total Number of . Total Volumetric
Matrix Traverse Points Flow Rate
4x3 12 -.03
6x2 12 -.57
2x7 14 1.04
3x5 15 .13
5x3 15 -.03
2x8 16 1.02
4x4 16 .03
3x6 '18 .11
6x3 18 -.01
4x5 20 .02
5x4 20 .02
3x7 21 .10
3x8 24 .09
4x6 24 0
6x4 24 .03
5x5 25 .01
4x7 28 -.01
5x6 30 -.01
6x5 30 .02
4x8 32 -.03
5x7 35 -.02
6x6 36 .01
5x8 40 -.03
6x7 42 -.01
6x8 48 -.02
- 182 -
-------
nu CD ri
» »
* * * *
MM
XXXXXX XXX ..
**********
00 «fl M 10
CD N CM 00
t 00 M U)
rl rl 00 IO
M
yj
xxxxxxxxxxxxxxx
-**************
ft N 00 CD
•* rl* N
M CD N CM
8*9*
Mr,
r (D (V
. . . lOrlOO CD CD
ON fMCM rl I M ' '
M 00 CM CO CM I
I rl I
rl ^ ffl Cft N
M ON M
4J
C
OJ
T3
G
0)
ft
a)
13
G
_^__-H G -P H
U H C
-H Q) rd
IH -P G C
<
W to
o
H
H
O
sc
co
§
CQ
H
M
CM
HHHH^
(9CDC9C9C9CDCgrlrirlrlrlrlr1CMCMCMCMCMCMCMMMMM
xxxxxxxxxxxxxxxxxxxxxxxxx
*************************
T^ (A M M r^ ^o r^» CD oo CD vS 10 r^
f^» CO T4 ID 04 ^ M) O G^ ^ ® T^ (7^
S5!
CO
IO
r
Q)
^ 1 £
G
CU
Tl
G
0)
DH
Q)
•a
4-)
[/)
4J
G
0)
G
o
W
no
G
O
o
0)
CU
G
O
ft
X
w
CO
H
Q
M
ffi
EH
O
CM
S
O
13
^
O
103 -
-------
Gas Velocity
FIGURE A-6 Velocity Distribution Simulating Profile Measured
Upstream of Test Elbow (Without Turning Vanes) in
1/10 Scale Model
- 184 -
-------
TABLE A-7
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE PARTICULATE DISTRIBUTION PRESENTED IN FIGURE A-7
>
i
iNuinoer ui traverse points cm.
' f — Number of traverse points
1 x 3
Total Number of
Matrix Traverse Points
1x1 1
1x2 2
2x1 2
1x3 3
3x1 3
1x4 4
2x2 4
4x1 4
1x5 5
5x1 5
1x6 6
2x3 6
3x2 6
6x1 6
1x7 7
1x8 8
2x4 8
4x2 8
3x3 9
2x5 10
5x2 10
2x6 12
3x4 12
- 185 -
jng x O.X.1.S
along y axis
% Error in Measured
Average Particulate
Concentration
-6.39
-2.30
-12.29
5.15
-11.47
6.36
-4.87
-11.17
6.67
-11.03
6.65 -
-6.92
-3.08
-10.96
6.64
6.62
-6.04
-2.97
-3.00
-5.51
-3.04
-5.14
-2.29
-------
TABLE A-7 (Cont)
% Error in Measured
Total Number of Average Particulate
Matrix Traverse Points Concentration
4x3 12 , -2.33
6x2 12 -3.10
2x7 14 -4.89
3x5 15 -1.81
5x3 15 -2.16
2x8 16 -4.72
4x4 16 -1.58
3x6 18 , -1.51
6x3 18 -2.11
4x5 20 • -1.12
5x4 20 -1.40
3x7 21 -1.31
3x8 24 -1.18
4x6 24 -.84
6x4 24 -1.34
5x5 25 -.95
4x7 28 -.66
5x6 30 -.67
6x5 30 -.90
4x8 32 -.53
5x7 35 -.50
6x6 36 -.63
5x8 40 -.38
6x7 42 -.46
6x8 48 -.34
- 186 -
-------
IO<0<9r4CMM*IO
* * * * *
M M M
C C C C C
3f v 5c nf v
* ¥?¥??? * * * * ¥ ¥¥ * *
MCMOOODriaaBCMN*MIOCMaCOCMIOU>0'tOOOONrtrlMCMNVOrlU>CMCMOt
ONVDONOO^-tNIO^OO^rl^COrlNTt-MeBONNONOON
N IO M
M CM N
M^M
§
(M
*• rl ON QCM
<0 VO ON CM
00 M
-P
QJ
G
-P QJ
G QJ
(y
XI
(0
•H
£
-P
G
T3
G
Q)
0)
•a
—g—0) t5 G
O H G
•H QJ T3
4-1 -P G G
m cn O o
QJ M QJ 0
O -H X QJ
O ft) M CO
4-
C
a
C
C
C
X
CL
w
D
a
H
Cn
o
EC
I
D
OQ
E-<
cn
O rl CM M ^t 10
* *
*
( C (
C C ( ( C ( (
QQaSOQCBrtHrlrlrlrlrlCMCMCMCMCMCMMMMMM-^
C C C C C C ( C ( C
v y ^^ *y y x* ^^ ^* *& ^
«««*««*««*
OS'tf-MMONIOlOCOONCOIONONON'OVOOONrllOHOMO
lONCvlCMONCOVDH^in-HOOWONrHMCMONNO^inON
at0NOIONrt'tf-IO-Hrs-U>rtCMIO(Mr1*>r''l<0<8rN-U>
rl 10
XT CM rl <0 CM 00 CM ON N * ON U> 10
N -e-i
* 1
<0
ON
(9
-------
X
Particle Concentration
gr/scfd
gm/Nm-3
FIGURE A-7 Particle Concentration Distribution Simulating
Profile Measured Upstream of Test Elbow
(Without Turning Vanes) in 1/10 Scale Model.
- 188 -
-------
TABLE A-8
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-8
Matrix
Number of traverse points along x axis
Number of traverse points along y axis
4x3
Total Number of
Traverse Points
% Error in
Measured Total
Volumetric Flow Rate
1x1
1x2
2x1
1x3
3x1
1x4
2x2
4x1
1x5
5x1
1x6
2x3
3x2
6x1
1x7
1x8
2x4
4x2
3x3
2x5
5x2
2x6
3x4
4x3
6x2
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
12
- 189 -
21.26
20.35
-.40
21.12
3.59
21.40
-13.26
3.81
21.53
3.63
21.60
-8.27
-4.72
3.46
21.64
21.66
-6.92
-3.55
-1.78
-6.39
-3.45
-6.14
- ,90
- .98
-3.52
-------
TABLE A- 8 (Con t,)
Of
Traverse Points Volumetric Flow Rate
2x7 14 -6.00
3x5 15 - .53
5x3 15 - .96
2x8 16 -5.91
4x4 16 - ,19
3x6 -18 - .35
6x3 18 -1.05
4x5 20 .14
5x4 20 - .20
3x7 21 - .24
3x8 24 - ,17
4x6 24 .31
6x4 24 - .30
5x5 25 .13
4x7 28 .41
5x6 30 .29
6x5 30 .02
4x8 32 .47
5x7 35 .39
6x6 36 .18
5x8 40 .45
6x7 42 .28
6x8 48 ,34
- 190 -
-------
X X XX
* * * *
( c
vfl y>
(9 IS
+ +
UJ
cn
01
Q
N
00
in M
O Q
+ +
111 UJ
o y>
M oo
in
Q
ui ui
CM ^
00
CM
X XX X XX XXX X X
***********
UJ
O
CM
in
Q S>
+ +
Ul Ul
in in
s» co
in cs
< c
^ in
ui ui
ON Q
CM CM
in ON
in H
C
Q
y) M ^t in y>
O O O O Q GD
+ + -f + + + + + +
lilUIUIUIUIUIUIUIUI
•HN'I'NONCMCOy)^'
^ y) CM ON in y> yj ^ N-
co •* CM y> o in
in
O Q
-I- +
Ul Ul
oo ri
M CO
CM TT y> CM
[
1
J
4J
— G
•H
U
-H
(4-1
14-1
(U
O
o
, HNinrvMTHIiTON^-NinOfv-OONinCMONCOM
O CM H CM N rl CO N CO 00 CM «) M in TJ- CO CO •* ON N i1^ M t •* N
I I I I I I I I
- 191 -
1
J
-<
•H
U
-J
T
J
H
-{
H
U
D
J
|
Q)
t-H
pQ
(0
•H
H
5s
-p
c
Q)
T3
c •
(U
a
Q)
(U
rH
XI
(0
•H
M
"3
|>
4J
C
(U
'O
C
Q)
QJ
Q)
r-«
H
Q
f4
IH
O
jj)
f-C
H
s
o
J2
X
J
o
p<
C 4J H 4J
H C C
Q) *"0 Q)
-P G G C
W O O O
n cx u a
•H X Q) X
tn w tn w
-------
Gas Velocity
FIGURE A-8 Velocity Distribution Simulating Profile Measured
Downstream of Test Elbow (Without Turning Vanes)
in 1/10 Scale Model.
- 192 -
-------
TABLE A-9
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE PARTICULATE DISTRIBUTION PRESENTED IN FIGURE A-9
-Number of traverse points along x axis
^—Number of traverse points along y axis
4x3
% Error in
Total Number of Measured Average
Matrix Traverse Points Particulate Concentration
1x1
1x2
2x1
1x3
3x1
1x4
2x2
4x1
1x5
5x1
1x6
2x3
3x2
6x1
1x7
1x8
2x4
4x2
3x3
2x5
5x2
2x6
3x4
4x3
- 193 -
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
-14.75
-24.58
-12,35
-11.75
-16,76
- 9.40
- 7.85
-17.40
- 8.80
-17.48
- 8.63
2.54
-16.02
-17.46
- 8.57
- 8.56
4.65
-17.17
- 2.96
5.26
-17.30
5.49
- .46
- 3.76
-------
TABLE A-9 (Cont.)
% Error in
Total Number of Measured Average
Matrix Traverse Points Particulate Concentration
6x2
2x7
3x5
5x3
2x8
4x4
3x6
6x3
4x5
5x4
3x7
3x8
4x6
6x4
5x5
4x7
5x6
6x5
4x8
5x7
6x6
5x8
6x7
6x8
12
14
15
15
16
16
18
18
20
20
21
24
24
24
25
28
30
30
32
35
36
40
42
48
-17.26
5,60
,22
- 3.87
- 5.65
- 1.21
,44
- 3.85
- .52
- 1.31
.53
.56
- .30
- 1.30
- .62
- .21
- .39
- .60
- ,18
- .31
- .37
- .27
- .29
- .25
- 194 -
-------
************************
xxxxxxxxxxxxxxxxxxxxxxxx-
************************
o o o o
UJ UJ UJ UJ
M •* M N. oo
CD CO V) ^"4
in CM M (r»
(M M
O OJ
•*• +
UJ Ul
(A 00
00 rt
UJ Ul
oo en
N ~
UJ UJ
N
Ul rl CO OS CM ^1-
UJ
GO
CM
in
CM M
G) O
UJ UJ
fM N
a> CM
CM CM
t C C (<(<<((<
(3 (3 (S CS) CJ CD O ^S O CS) O
•f-f+-f + + + -f + + +
UJUIUJUJUJUJUJUJUJUJUJ
Jj
•r*
C
0)
C
<0
Pi
(1)
C -P
H C
0)
-P C
W O
>-l &
-H X
k, H
1
ft
CTt
I
O
H
H
O
O
H
» »» >
*******
*•*"*
OOOBOOOrirlTHTtrlH^CMCMCMCMCMCMCMMMMM
^f *^£ v ^c *v "^ ^r *y v *s^ ^^ *v ^* *^ v *^ *s^ ^^ v "s^1 *v *v^ v* v^ %^
*************************
<
H
O
Ul
H
m
CM
N
(
H
O
UJ
in
fw
M
N
<((<((<(<<(<<(<(((<
HrHHHCMfMCMCMOHCilCMCMCM^HrlM^
ooooooooooooooooooo
UJUIUJUJUJUJUJUJUIUJUJUJUJUJUJUJUJUIUJ
CM in 00 T4 CM ^ QD O CM CA CM CM f^ ^3h M ^ CM ^) O
VD vi O 00 *?4 ^»D 1^ M fv Q) Is*- in M T*4 ^0 CM O 1^ ^f
T4 oo T^ T^ CM yi oo r^ CM ^ co r^- CM ^o TH co ^t* ^ ^r
in CM in in CM
i i i
II
II
M H ^ *• CM -H
i i i i i
PQ
H
(X
cn
H
a
w
EH
§
o
CM
- 195 -
-------
Particle Concentration
FIGURE A-9 Particle Concentration Distribution Simulating
Profile Measured Downstream of Test Elbow
(Without Turning Vanes) in 1/10 Scale Model.
- 196 -
-------
TABLE A-10
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-10
Number of traverse points along x axis
T—Number of traverse points along y axis
x 3
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1x1 1 -2.13
1x2 2 11.0
2x1 2 6.14
1x3 3 6.09
3x1 3 1.47
1x4 4 5.10
2x2 4 -4.19
4x1 4 .65
1x5 5 4.82
5x1 5 .46
1x6 6 4.72
2x3 6 1.16
3x2 6 .28
6x1 6 .41
1x7 7 4.68
1x8 8 4.66
2x4 8 1.66
4x2 8 .95
3x3 9 .86
2x5 10 1.58
5x2 10 1.05
2x6 12 1.44
3x4 12 .76
4x3 12 .74
6x2 12 1.04
- 197 -
-------
TABLE A-10 (Cont)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
2x7 14 1.33
3x5 15 % .65
5x3 15 .68
2x8 16 1.24
4x4 16 .54
3x6 18 .57
6x3 18 .64
4x5 -20 .42
5x4 20 .46
3x7 21 , .51
3x8 24 .47
4x6 24 .35
6x4 24 .42
5x5 25 . .34
4x7 28 .31
5x6 30 .27
6x5 30 .31
4x8 32 .28
5x7 35 .23
6x6 36 .24
5x8 40 .20
6x7 42 .20
6x8 48 .17
- 198 -
-------
((( C (<(( C C (<((( C (<((((((
»»»>»»»>»>»>»>3
***********************i
MMM*-*-*** t •«• in m in
m
vo
xxxxxxxxxxxxxxxxxxxxxx
************************
/
x-^
<7tC9^NHinO'kHNC90T»«-
cooicr»oicovocf»s>Hinoiinu>HoiM
M CO tH CM U> <9 in CM * I H IO G> CO Y«n
rS M IO H Ol ri Ch Ol Ol H O H 00 M IO
H O M OIT4O1HN I CM CD Cft U>
HT< IrlOIOll H 01 H
CO oo 0110 © H
<£> •* O Ol \0 CM CO
05 •
M
M
4->
C
GJ
13
c:
&4
4J
C
0)
rQ
C
0)
0)
t3
T3 C
C 4J H 4-
M C f.
O T3 C
4-1 C C C
in O O C
M O4 U C
•H X (1) >
[j-i W CO t
o
I
W
D
O
H
fa
3
H
O
W
CO
EH
D
ffl
(9 r
U) ^ 10 00
M ri VD M
Ol
£Tt
CO * IO
M CO NO1
« N
I
i
m M in in y> r
TlN^-MH^
M vx> 10 ON co M
IOHQNM
MNYJOI
i i
COMMCf»0S*-Cr>00*
(OQOlOMMCftNCd
.
-H •* «) g CO
01 co H Svp
1 r*OI 01 T
SHOI
O OS
I *
c
0)
T)
c
CD
a
0)
n3
^ G
C 4J M 4
H C f
0) T3 C
4J C C f
U O O C
n a o c
-H X 0) >
fa fa" CO PI
O
fa
H
a
o
3
- 199 -
-------
Gas velocity
FIGURE A-10velocity Distribution Simulating Profile Measured
Upstream of Test Elbow (With Turning Vanes) in
1/10 Scale Model.
- 200 -
-------
TABLE A-11
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-ll
Number of traverse points along x axis
xc—Number of traverse points along y axis
x 3
% Error in Measured
Total Number of . Total Volumetric
Matrix Traverse Points Flow Rate
1x1 1 -4.64
1x2 2 9.56
2x1 2 2.67
1x3 3 10.10
3x1 3 1.17
1x4 4 10.05
2x2 4 -.72
4x1 4 .84
1x5 5 9.98
5x1 5 .73
1x6 6 9.93
2x3 6 -.28
3x2 6 -.03
6x1 6 .69
1x7 7 9.89
1x8 8 9.86
2x4 8 -.11
4x2 8 -.02
3x3 9 . -.28
2x5 10 -.03
5x2 10 -.07
2x6 12 .01
3x4 12 .42
4x3 12 .42
6x2 12 -.11
- 201 -
-------
TABLE A-11 (Cont)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
2x7 14 .03
3x5 15 .40
5x3 15 .37
2x8 16 .05
4x4 16 .41
3x6 18 .38
6x3 18 -.33
4x5 -20 .37
5x4 20 .36
3x7 21 . .36
3x8 24 .35
4x6 24 .34
6x4 24 .32
5x5' 25 .32
4x7 28 .32
5x6 30 .28
6x5 30 .27
4x8 32 .30
5x7 35 .26 -
6x6 36 .24
5x8 40 .24
6x7 42 .21
6x8 48 .20
- 202 -
-------
tnovo ® riw M *• io
N M t IO W>
M * 10 y>
(((((((((<(((((<(<(((<((
>»>»»»»>»>>>>>>>>>
************************
MMMt-*1**-* Xh •* IOIOIOtOIOIOIO«JO<0VD
u>
xxxxxxxxx
*********
xxxxxxxxxxxxxxx
*************
* *
rl rl M N OS N
y> ^r ri y» ri ri
CJ N OS 00 VO M
OS M rl M M 0)
rlVOQVDOSWNCM
M N y) IO N 03 OS
N 0) 00 M OS,N th
(M rl OS N OS N M
10 (NJ rl O 00 O 10
N N OS M «> rl IO
IO y> M NN N N
Q M rl N rl M IO
OSVO (SJ I rIOS Q
IO n (Si I rl N
I I I
M & f*1 (9 f^» IO ^f* IO O CO
CSIrlOOOJO(MN^OSIO
NM 05 rl 00 N^" rl 00
M(SJ"t rlUXSJOJ*-
rlrl I rl rl I
I I
CD (M 00 VD rl
O OS rl ^ O
G) M y> (M N
MN N Nt
U)M t MN
I •
T
-«- 1
rH
XI
"*i (0
•H
^
>
-P
C
0)
ro
' G
«S 1 Q)
-P O4
G QJ
0) fd
vu
rH
XI
•H
J_l
(0
4J
G
QJ
T3
G
QJ
&
QJ
•a
G
-H C -P H
U H G
•H Q) TJ
m jj G c
<+H CO O O
QJ rl Dj U
0 -H X QJ
CJ fe W CO
-P
G
QJ
G
O
a
X
H
O
O
E
55
O
H
EH
D
CQ
H
PI
EH
HuHHHHHHHHHH
<9 O Q
xxxxxxxxxxxxxxxxxxxxxxxx
****»**»**«««•********«**
rlVONOSIOyXMM^OOrlOSIOrlNQNNIONNOSO
ooHos^yJosoDoiri
rl*M(SJNO«CSIM
M OS O IO T
rl
1
N M N *
I rl rl
OS M
n\
QJ
rH
X)
J<; (Tl
-H
5-1
>
-e c
QJ
'O
G
-t-, QJ
-P ' 0<
, G QJ
rH
X)
(0
-H
S-l
>
-P
C
T)
G
QJ
a
QJ
QJ rQ C
' -H G -P H
, U H C
•H Q) T3
, «M -P G C
m co o o
QJ M CL, u
O -H X QJ
U Pn H CO
-P
G
QJ
G
O
D
X
W
Q
W
K
EH
O
PH
O
53
- 203 -
-------
Gas Velocity . ,
_m/sec
FIGURE A-llVelocity Distribution Simulating Profile Measured
Downstream of Test Elbow (With Turning Vanes) in
1/10 Scale Model.
- 204 -
-------
TABLE A-12
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-12
r Number of traverse points along x axis
.Number of traverse points along y axis
4x3
Matrix
1
1
2
1
3
1
2
4
1
5
1
2
3
6
1
1
2
4
3
2
5
2
3
4
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
1
2
1
3
1
4
2
1
5
1
6
3
2
1
7
8
4
2
3
5
2
6
4
3
Total Number of
Traverse Points
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
% Error
Total
Flow
-9.
-9.
12.
-9.
1.
-9.
13.
•
-9.
•
-9.
13.
1.
^ •
_Q
-9.
13.
•
1.
12.
•
12.
1.
•
in Measured
Volumetric
Rate
74
75
65
79
73
81
20
06
82
20
83
07
92
20
83
83
01
18
88
98
14
96
85
10
- 205 -
-------
TABLE A-12 (Cont)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
6x2 12 -.16
2x7 14 12.95
3x5 15 1.83
5x3 .15 -.22
2x8 16 12.94
4x4 16 .07
3x6 18 1.82
6x3 18 -.25
4x5 20 .05
5x4 20 -.26
3x7 21 1.81
3x8 24 1.81
4x6 24 .04
6x4 24 -.29
5x5 25 -.27
4x7 28 .03
5x6 30 -.28
6x5 30 -.30
4x8 32 .03
5x7 35 -.29
6x6 36 -.31
5x8 40 -.29
6x7 42 -.32
6x8 48 '-.33
- 206 -
-------
MMM^*t^^^iDioinir)ioiny)U)\oy)w-
x x x x x x x x x x x x x x x x x x x x x-
*********************
n(T>NCO^-COCOMMNNNMininCh(MChNW)N --s-i fl
in'*(\jN(v1Jty!>fy|^fN (Tt N CO M rl «) I I CO CD CO tf) N CD M W o3 S K
rlr< M^T1 I M M H HrJI O-HX
?*Y
^~
0)
rH
*•" .Q
(0
•H
H
(0
^
_^
4J
C
QJ
•d
-s-i c
I cu
•p DJ
C 0)
Q) 13
1
a;
5
•H
^_l
nJ
j>
-P
C
OJ
•a
c
0)
ex
w w o o o o
0) H (X U CU K
O -H X 0) X in
o fo w en w
eq
**********
CDCDOCDCDOtHrlTHrlHrtNNOJMNNnMM
( ( ( ( < C ( ( ( ( ( (
i *%
(((((((((,,,,>
^^^f??^^^?^^^^^^?^?^^
U)MW3(MCOOOIOCOVOn-M^OCDHMCOHCO(r>T M N M ^t t N M r»vo CD yj m cr> co W) CO H M * H «) CO CO TO •* CD M ON CD H CO in CD r4
C*CDCD<9HmmHIOCDU>NCDCrt*NNNU>V0-*
M
M
•H N CM CO
N CM rH rH
OCDCTiNinWM'*
CDO^-CMC^MH'H
«) CD
WM
in in
CM N
CM CO M
VD tf) *
N M rl NM M CD
-------
Gas Velocity
FIGURE A-12Velocity Distribution Simulating Profile Observed
in Rectangular Duct Between I.D. Fan and Stack in
a 120 MW Oil-Fired Electrical Generating Station
(Reference 33).
- 208 -
-------
TABLE A-13
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-13
— Number of traverse points along x axis
^— Number of traverse points along y axis
4x3
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1x1 1 ' -78.90
1x2 2 -21.42
2x1 2 -55.12
1x3 3 -26.28
3x1 3 -42.29
1x4 4 -27.19
2x2 4 -8.48
4x1 4 -40.33
1x5 5 -27.39
5x1 5 -39.99
1x6 6 -27.44
2x3 6 -5.71
3x2 6 -.95
6x1 6 -39.97
1x7 7 -27.44
1x8 8 -27.43
2x4 8 -5.48
4x2 8 ,40
3x3 9 -.95
2x5 10 -5,51
5x2 10 .74
2x6 12 -5.56
3x4 12 -1,06
4x3 12 -.05
- 209 -
-------
TABLE A-13 (Cotit)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
6x2 12 .84
2x7 14 -5.61
3x5 15 -1.11
5x3 15 .20
2x8 -16 -5.65
4x4 16 -.22
3x6 18 -1.13
6x3 18 .29
4x5 20 -.28
5x4 20 .01
3x7 21 -1.15
3x8 24 -1.16
4x6 24 -.31
6x4 24 .09
5x5 25 -.05
4x7 28 -.32
5x6 30 -.08
6x5 30 .03
4x8 32 -.32
5x7 35 -.09
6x6 36 0
5x8 40 -,10
6x7 42 -.01
6x8 48 -.02
- 210 -
-------
>>»>>>>>>>>>>»>>»-
********************
M M M
xx(
* *
XX
* * *
(<(
<(((((
1 ^^ ^f ^^ ^^ ^^ *V* *^^ *^^ *^r^ ^^* *^
***********¥¥¥¥¥
:Je' '
( C ((
COOirlOOOOM*ID^OOy)^'OOU)r«MOOOOM
QNIvON'iOIVNOOf'IIDCOrlQfirllVOlGJCMQMN.
MIOMN^OO^NrltlDtrlrlMrllDO'kNlS
id
00
MIDWOOOO^CMIDIDONrlNIDrlNNCvlthMOOrlrlO^S
M>MVOfftVO(nNMU}OOrN.|DC4rlrllDeO&(MN(nU>OOID
^ ^ ^f* ^ 00 T^ M ON (M 0^ O O O O V0 M (M w T^ M M M ID ^J"
MO>>WN.^:N^^pf11005r<>lONrl
N WW) M ID 00
MrlOO MM
I rl rl rl I
ID *JO ID
NID<£rl*00<9*NOOT4rlrl'tlDU>
8&M ff»(D rl OOOrl I OD M rl rl O
..0100 W?0»Mlf> I 0? N 0) *
M C4
* (DM _.
rl f N ID rl
I I I
rl MID
O
U
OJ
rH
XI
(0
•H
-p
a
cu
0)
rH
XI
•rH
-P
C
(U
cu
ft
0)
t3
fi
4J H -P
C C
» >
* * *
SQOOOHrlHrlrlrlrlNNNOJOrfOJNMMMM
((((<(
XX X XX X
******
Q)
ID rl N N N -^rCiiQrl(>ir>OdO10SNOS(9S6C9000S
rlO^rtNJOIDIN.WIDNOIOSOrlrlCO^U)
ID « rl N ID N M
00 rl rl ^ ^t ^ *t
n vo vo ID cu
I I I
rl N M M t <0 COO N O 00 rl N ID 00
riw^^riMS^oS^rairi^-
N rlrIN HOM rl OSN M^tON
TON N
Irl
^tfXSOO
-*NU)rl
1 T
r,«D
1
M
<•>
o
U
C
o
X
w
(U
XI
(d
•H
(0
-P
C
(U
0)
(U
C
H -P
C
T3 0)
C fl
o o
O ft
(U X
en w
D
CQ
H
05
en
H
Q
w
w
05
o
H
g
o
a
a
o
PM
- 211 -
-------
Gas Velocity
FIGURE A-13Velocity Distribution Simulating Profile Measured
Downstream of I.D. Fan at Black Dog Power
Station.
- 212 -
-------
TABLE A-14
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-14
Number of traverse points along x axis
r—Number of traverse points along y axis
4x3
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1x1 1 -15.15
1x2 2 • -13,70
2x1 2 ,21
1x3 3 -13,77
3x1 3 ,72
1x4 4 -13.82
2x2 4 .47
4x1 4 .73
1x5 5 -13,85
5x1 5 .70
1x6 6 -13,86
2x3 6 ,19
3x2 6 ,81
6x1 6 ,68
1x7 7 -13,87
1x8 8 -13.88
2x4 8 .07
4x2 8 .77
3x3 9 ,53
2x5 10 ,01
5x2 10 .71
2x6 12 - ,02
3x4 12 ,41
4x3 12 ,48
- 213 -
-------
TABLE A-14 (Cont)
% Error in Measured
Total Number of • Total Volumetric
Matrix Traverse Points Flow Rate
6x2
2x7
3x5
5x3
2x8
4x4
3x6
6x3
4x5
5x4
3x7
3x8
4x6
6x4
5x5
4x7
5x6
6x5
4x8
5x7
6x6
5x8
6x7
6x8
12
14
15
15
16
16
18
18
20
20
21
24
24
24
25
28
30
30
32
35
36
40
42
48
.68
-.04
.35
.43
-.06
.36
• 31
.39
.30
.31
.29
.28
.26
.27
.25
.24
.21
.21
.23
.19
.18
.18
.16
.14
- 214 -
-------
J3
<0
0)
(U
a
»»»>»»>»»»»»»
a************************
x**>
xxxxxxxxxxx
((<
X X
X X X X
( ( ( (
X X X
* * * * *
Ch N If) H M (M ®
o1* co in in oo in CM
«t
I I
M
00
00 rl yj
th If) IT)
^O rt VD IO
**
r4
I
o
H
PL,
H
O
ffi
§
H
EH
D
CQ
H
CO
H
Q
w
ffi
H
5
a
- 215 -
-------
X
A
•>-Y
Gas Velocity
f t/-see
m/sec
FIGURE A-14Velocity Distribution Simulating Profile
at Allen Steam Plant Unit No. 1
Precipitator Inlet Ports 1-3 (See
Figure 13 )-
- 216 -
-------
TABLE A-15
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-15
Number of traverse points along x axis
,r—Number of traverse points along y axis
4x3
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1x1 1 -13.44
1x2 2 , -5.26
2x1 2 -6.03
1x3 3 -4.64
3x1 3 -5.31
1x4 4 -4.48
2x2 4 -1.23
4x1 4 -5.11
1x5 5 -4.42
5x1 5 -5.03
1x6 6 -4.40
2x3 6 -.93
3x2 6 -.72
6x1 6 -4.98
1x7 7 -4.38
1x8 8 -4.37
2x4 8 -.86
4x2 8 -.56
3x3 9 . -.44
2x5 10 -.84
5x2 10 -.49
2x6 12 -.83
3x4 12 -.38
4x3 12 -.28
- 217 -
-------
TABLE A-15 (Cont)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points ' Flow Rate
6x2
2x7
3x5
5x3
2x8
4x4
3x6
6x3
4x5
5x4
3x7
3x8
4x6
6x4
5x5
4x7
5x6
6x5
4x8
5x7
6x6
5x8
6x7
6x8
12
14
15
15
16
16
18
18
20
20
21
24
24
24
25
28
30
30
32
35
36
40
42
48
-.45
-.83
-.36
-.21
-.82
-.23
-.35
-.17
-.21
-.16
-.35
-.35
-.20
-.12
-.14
-.20
-.13
-.10
-.19
-.13
-.09
-.12
-.09
-.09
- 218 -
-------
0)
I— I
XI
ro
(1)
i— i
J3
(0
C
QJ
TJ
0)
a
a>
•a
TJ
s~
X -3 "
CO W p.. O
c o
"
#*****************#******
(((((((((<((((((<((((((((
xxxxxxxxxxxxxxxxxxxxxxxxx-
*******#*****************
CD CO t*- <& &
(5> VO *X) 00 ^
G» N M
(VQ I
M T*
I
N Q M N H in r< Ch
N^
rH H
I I
(M y> t
r4 -H I
"jDf^OO^rl^CxJkOlO
(M M T! T< H I
I I
m
H
<;
w
ci
p
o
H
Pn
s
H
I
O
K
CO
D
CQ
H
Q
W
H
leH
^-4
O
5S
o
At
- 219 -
-------
Gas Velocity
["ft/secl
[m/'sec J
FIGURE A-15Velocity Distribution Simulating Profile at
Allen Steam Plant Unit No. 1 Precipitator
Inlet Ports 4-6 (See Figure 13 ).
- 220 -
-------
TABLE A-16
EVALUATION OF VARIOUS EQUAL AREA SAMPLE STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-16
Number of traverse points along x axis
,r—Number of traverse points along y axis
4x3
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1x1
1x2
2x1
1x3
3x1
1x4
2x2
4x1
1x5
5x1
1x6
2x3
3x2
6x1
1x7
1x8
2x4
4x2
3x3
2x5
5x2
2x6
3x4
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
-221 T-
8.95
6.81
4.20
6.24
4.21
6.03
.07
4.28
5.93
4.32
5.87
-.11
.05
4.35
5.84
5.82
•-.12
.13
-.15
-.13
.18
-.12
-.18
-------
TABLE A-16 (Cont)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points • Flow Rate
4x3
6x2
2x7
3x5
5x3
2x8
4x4
3x6
6x3
4x5
5x4
3x7
3x8
4x6
6x4
5x5
4x7
5x6
6x5
4x8
5x7
6x6
5x8
6x7
6x8
12
12
14
15
15
16
16
18
18
20
20
21
24
24
24
25
28
30
30
32
35
36
40
42
48
-.08
.22
-.12
-.18
-.04
-.12
-.11
-.18
-.01
-.12
-.07
-.18
-.18
-.13
-.04
-.08
-.13
-.09
-.06
-.13
-.09
-.06
-.09
-.06
-.06
- 222 -
-------
-o c
C (U
<1> "O
a c
0) 0)
vo
H
I
R
o o
o a
-------
Gas Velocity
ft/ sec
m/ sec
FIGURE A-16 Velocity Distribution Simulating Profile at
Allen Steam Plant Unit No. 1 Precipitator
Inlet Ports 7-9 (See Figure 13 ) .
- 224 -
-------
TABLE A-17
EVALUATION OP VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-17
Number of traverse points along x axis
JTI—Number of traverse points along y axis
4x3
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1x1 1 1.88
1x2 2 -8.97
2x1 2 10.13
1x3 3 -9.08
3x1 3 9.43
1x4 4 -8.98
2x2 4 1.85
4x1 4 9.02
1x5 5 -8.91
5x1 5 8.81
1x6 6 -8.87
2x3 6 1.33
3x2 6 1.42
6x1 6 8.68
1x7 7 -8.84
1x8 8 -8.82
2x4 8 1.23
4x2 8 1.09
3x3 9 ' .94
2x5 10 1.19
5x2 10 .90
2x6 12 1.18
3x4 12 .85
- 225 -
-------
TABLE A-17 (Cont)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
4x3
6x2
2x7
3x5
5x3
2x8
4x4
3x6
6x3
4x5
5x4
3x7
3x8
4x6
6x4
5x5
4x7
5x6
6x5
4x8
5x7
6x6
5x8
6x7
6x8
12
12
14
15
15
16
16
18
18
20
20
21
24
24
24
25
28
30
30
32
35
36
40
42
48
.63
.79
1.17
.82
.46
1.17
.56
.81
.36
.52
.38
.80
.80
.51
.28
.36
.51
.35
.26
,50
.34
.25
.34
.24
.24
- 226 -
-------
QJ
r—I
,Q
(0
QJ
•—H
,Q
(0
0 4J
•o c
0)
a
0)
•a
c
I*
• g
D
C +j !.(>(>>>>>>
*********
^
a
o
o
CO
0)
*.xxxicxxxxxxxxxxxxxxxxxxxx-
*************************
(Tt M (T> O 00 rl N
os Q H co in in N
Ol
I
ON
(9 VO C4 N
I rlHNCMN
I IT-"
o
ffi
H
EH
D
CQ
H
CO
H
Q
EH
«
o
M
s
- 227 -
-------
FIGURE A-17Velocity Distribution Simulating Profile at
Allen Steam Plant Unit No. 1 Precipitator
Inlet Ports 10-12 (See Figure 13 ) .
- 228 -
-------
TABLE A-18
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-18
Number of traverse points along x axis
Number of traverse points along y axis
4x3
% Error in
Total Number of Measured Total
Matrix Trave-rse Points Volumetric Flow Rate
1x1 1 -14.30
1x2 2 - 5.13
2x1 2 -14.14
1x3 3 - 4.02
3x1 3 -14.05
1x4 4 - 3.68
2x2 4 - 1.16
4x1 4 -14.01
1x5 5 - 3.53
5x1 5 -13.99
1x6 6 - 3.45
2x3 6 - .41
3x2 6 - 1.02
6x1 6 -13.98
1x7 7 - 3.40
1x8 8 - 3.37
2x4 8 - .27
4x2 8 - 1.01
3x3 9 - .26
2x5 10 - ,23
5x2 10 - 1.02
2x6 12 - .21
3x4 12 - .12
4x3 12 - .25
6x2 12 - 1.02
- 229 -
-------
TABLE A-18 (Cont)
% Error in
Total Number of Measured Total
Matrix Traverse Points Volumetrie Flow Rate
2x7 14 - .20
3x5 15 - .07
5x3 15 - .25
2x8 16 - .20
4x4 16 - ,10
3x6 18 - .06
6x3 18 - .25
4x5 20 - .06
5x4 20 ' - .10
3x7 21 - .05
3x8 24 » .04
4x6 24 - .03
6x4 24 - .10
5x5 25 - .05
4x7 28 - .03
5x6 30 - .03
6x5 30 - .05
4x8 32 - .02
5x7 35 - .02
6x6 36 - .03
5x8 40 - .02
6x7 42 - .02
6x8 48 - .01
- 230 -
-------
£ 4-J
c c
x
CO
c
a;
TJ
c
X ~
fi b
t (
*********
**#******
( C ((((((((((<(((
sm^^^m^^^m?*
X
a
N
M
N in a*
^ T< 00
JD -H (M
m H i
y> M
od in
Cg CJ T< I
I I
H os N r r< oj 10 w oo
CM H
-------
Gas Velocity
ft/sec
sec
FIGURE A-18Velocity Distribution Simulating Profile at
Allen Steam Plant Unit No. 1 Precipitator
Inlet Ports 13-15 (See Figure 13 ) .
- 232 -
-------
TABLE A-19
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-19
\
Number of traverse points along x axis
Number of traverse points along y axis
Y
4x3
% Error in
Total Number of Measured Total
Matrix Traverse Points Volumetric Flow Rate
1x1
1x2
2x1
1x3
3x1
1x4
2x2
4x1
1x5
5x1
1x6
2x3
3x2
6x1
1x7
1x8
2x4
4x2
3x3
2x5
5x2
2x6
3x4
4x3
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
- 233 -
-22.91
-15.33
-12.80
-13.57
-12.36
-12.93
- 2.20
-12.32
-12.63
-12.31
-12.46
- 1.06
- 1.77
-12.32
-12.36
-12.30
- .73
- 1.77
- .65
- .58
- 1,80
- .50
- .31
- .64
-------
TABLE A-19 (Cont)
% Error in
Total Number of Measured Total
Matrix Traverse Points Volumetric Flow Rate
6x2 12 - 1.82
2x7 14 - ,46
3x5 15 - ,17
5x3 15 - ,66
2x8 16 - .43
4x4 '16 - .30
3x6 18 - .09
6x3 18 - - .68
4x5 20 - ,16
5x4 20 - .32
3x7 21 - .05
3x8 24 - .02
4x6 26 - .08
6x4 24 - .34
5x5 25 - .18
4x7 28 - .04
5x6 30 - .10
6x5 30 - .19
4x8 32 - .01
5x7 35 - .06
6x6 36 - .12
5x8 40 - .03
6x7 42 - .07
6x8 48 - .04
- 234 -
-------
(U
-" 01
10 —<
"C -0
nn 'O
0)
TJ
CJ
0
a
c -1 e>2 -rt
s ^ % £ "
f3 C C N
NC^rl
I I
en
H
<
W
H
fc.
H
I
O
a
O
M
C-t
D
03
H
Q
w
EH
S
O
2
- 235 -
-------
t
->- X
Gas Velocity
ft/sec
m/ sec
FIGURE A-19Velocity Distribution Simulating Profile at
Allen Steam Plant Unit No. 1 Precipitator
Inlet Ports 15-18 (See Figure 13 ) .
- 236 -
-------
TABLE A- 20
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A- 20
Number of traverse points along x axis
Number of traverse points along y axis
4x3
T°tal Measured Ttal
Matrix Traverse Points Volumetric Flow Rate
1x1
1x2
2x1
1x3
3x1
1x4
2x2
4x1
1x5
5x1
1x6
2x3
3x2
6x1
1x7
1x8
2x4
4x2
3x3
2x5
5x2
2x6
3x4
4x3
6x2
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
12
- 237 -
1.42
-1.36
.12
-1.87
2.75
-2.05
-1,02
2.93
-2.14
2.84
-2.18
-1.23
.77
2.75
-2.21
-2.22
-1.30
.91
.41
-1.33
.87
-1,35
,28
,54
,81
-------
TABLE A-20 (Cont)
% Error in
Total Number of Measured Total
Matrix Traverse Points Volumetric Flow Rate
2x7 14 -1.36
3x5 15 .22
5x3 15 .50
2x8 16 -1.37
4x4 16 . .41
3x6 '18 .19
6x3 18 .46
4x5 20 - .35
5x4 20 .35
3x7 21 .17
3x8 24 .16
4x6 24 .32
6x4 24 .33
5x5 25 .32
4x7 28 .30
5x6 30 .28
6x5 30 .27
4x8 32 .28
5x7 35 .27
6x6 36 .24
5x8 40 .25
6x7 42 .22
6x8 48 .21
- 238
-------
r«M MG> H (M
0)
O
CO pj PL, C,
o
CM
I
<
O
3
H
CD O O H rl rl
((( C ((((((((((((
((
XXXXXXXXXXXXXXXXXXXXXXXXXXX.-
«****««*«**«««««********«#**
CO M Q CO N OU CO
oo co oo iv. CD m
ON y) M N O1"* ^
-------
Gas Velocity
[~ft/secl
[_ m/secj
FIGUREA-20Velocity Distribution Simulating Profile at Colbert
Steam Plant Unit No. 4 Precipitator Inlet Ports
13-24 (See Figure A-29 )
- 240 -
-------
TABLE A-21
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-21
Number of traverse points along x axis
— Number of traverse points along y axis
4x3
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1x1
1x2
2x1
1x3
3x1
1x4
2x2
4x1
1x5
5x1
1x6
2x3
3x2
6x1
1x7
1x8
2x4
4x2
3x3
2x5
5x2
2x6
3x4
4x3
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
- 241 -
-3.94
-5.11
9.10
-5.32
- .03
-5.40
8.59
- .59
-5.43
r .25
-5.45
8.50
-1.60
.10
-5.46
-5.47
8.46
-2.12
-1.89
8.45
-1.67
8.44
-1.99
-2.40
-------
TABLE A-21(Cont.)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points ' Flow Rate
6x2 12 -1.22
2x7 14 8.44
3x5 15 -2.04
5x3 15 -1.93
2x8 16 8.43
4x4 '16 -2.50
3x6 18 -2.07
6x3 18 . -1.46
4x5 20 -2.55
5x4 20 -2.02
3x7 21 -2.08
3x8 24 -2.09
4x6 24 -2.57
6x4 24 -1.55
5x5 25 -2.06
4x7 28 -2.59
5x6 30 -2.08
6x5 30 -1.59
4x8 32 -2.59
5x7 35 -2.10
6x6 36 -1.61
5x8 40 -2.11
6x7 42 -1.62
6x8 48 -1.63
- 242 -
-------
—i
•2 <"
(0 .-I
"£ -2
£ 5
s 5
OJ +J
T! C
C 0)
OJ T3
a c
>»>
**********#*****#*****#**
(( f (((((
**
O O H rl rt rH
. xxxxxxxxxxxxxxxxxxxxxxxxxxx-
****************************
«j ^t oj y> r
^ N N tC f
ID M (M 00 f
otNOtnwat^co
n
0) N IO M O
•<• H (M
-------
Gas Velocity
ft/sec]
m/secj
FIGURE A-21. Velocity Distributions Simulating Profile at
Colbert Steam Plant Unit No.IPrecipitator Inlet
Ports 1-12 (See Figure A-2$ .
- 244 -
-------
TABLE A-2 2
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-22
I Number of traverse points along x axis
I _ Number of traverse points along y axis
4x3
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1x1
1x2
2x1
1x3
3x1
1x4
2x2
4x1
1x5
5x1
1x6
2x3
3x2
6x1
1x7
1x8
2x4
4x2
3x3
2x5
5x2
2x6
3x4
4x3
6x2
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
12
- 245 -
11.54
6.09
1.81
8.88
- .05
9.38
2.28
- .71
9.50
-1.01
9.53
2.51
1.17
-1.18
9.54
9.53
2.52
.75
1.17
2.51
.55
2.50
1.15
.70
.44
-------
TABLE A- 2 2( Cont.)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points • F.low Rate
2x7 14 2.49
3x5 15 1.14
5x3 15 .47
2x8 16 2.49
4x4 16 .67
3x6 -18 1.13
6x3 18 .35
4x5 20 .65
5x4 20 .44
3x7 21 1.12
3x8 24 1.12
4x6 24 .64
6x4 24 .32
5x5 25 .42
4x7 28 .64
5x6 30 .42
6x5 30 .30
4x8 32 .63
5x7 35 .41
6x6 36 .29
5x8 40 .41
6x7 42 .29
6x8 48 .29
- 246 -
-------
M + r) y>
m u>
( < (
M M MM
C ( ( C < < (
urnmm-
10 10
(((((((((
xxxxxxxxx
*********
G>G5NrHI-CM*-CO
NN*-inCOO>*COrl
ooNCA«-cMCMy>y>*
TriOinU>«>CANCA
N rl CA M M t O
H M * <§> N 00 N
ID CO "
§CM M N <0 rlU>
M M 00 rl CACA
cowcAQM-a-io^-Mttioyjco
OONC9MCACA rl M CA 01 CM CO
00 CO IO rl CO rl r! CA M N M
I M* M I
M
IT*
( ( <
XXX-
* * *
00 CD CA
M CA CM
UM rl
<0 CM *
N cs> in
MWOO
lOrl VO
CO (M U)
OH N T
el - 11
Q)
a)
iH
«*
J3
03
•H
_^ 1 ^|
03
4J
C
r»-<
TJ
C
4J
C
0)
'O
0)
o.
KM
0)
•a
m -P
H-l (0
Q) M
C9G)(S>G>C9rlrlrl
****?**£*
cf,oQi£ivcTt-in(Min
CACArlCOTrCOCa*CO
CACACOCACOO^rlCA
CACMCAMCAiflMrllO
CM M (M
( ( (
xxx
oo vo cf>
* s> o
in in co
N ® M
1 rl V)
N M
r4 M
in m
1 -H ^
tf) CD * Cf» O
1 -H
— n "v
(U
r-H
£1
(0
^e — 1 -H
M
id
4J
(U
•O
C
(U
-P ft
C Q) .
Q) 13
w
r-t
^3
03
•H
M
(0
>
4J
C
(U
t3
C
0)
O4
O
T3
fi
-*-- H C -P H 4.
U H C C
-H 0) T) C
m 4J G C £
W 01 O O C
0) M DJ O S
o -H x Q) :
en
H
Q
w
ffi
£-•
O
Pn
O
a
o
PM
C
W
247 -
-------
X
Gas Velocity
ft/sec
m/sec
FIGUREA-22 Velocity Distribution Simulating Profile at
Black Dog Power Station.
I.D. Fan Inlet Station No. 1 (See Figure A-27 )
- 248 -
-------
TABLE A-2 3
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-2 3
Number of traverse points along x axis
Number of traverse points along y axis
Y
4x3
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1x1
1x2
2x1
1x3
3x1
1x4
2x2
4x1
1 x 5
5x1
1x6
2x3
3x2
6x1
1x7
1x8
2x4
4x2
3x3
2x5
5x2
2x6
3x4
4x3
6x2
- 249 -
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
12
-25.28
4.27
-43.47
4.28
-38.78
5.30
-5.47
-37.55
5.26
-37.09
5.22
-4.14
-1.98
-36.88
5.19
5.16
-4.09
-1.25
- .81
-4.14
-1.04
-4.19
- .82
- .10
- .95
-------
TABLE A-23(Cont.)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
2x7 14 -4.22
3x5 15 - .89
5x3 15 .11
2x8 16 -4.25
4x4 16 - .11
3x6 '18 - .95
6x3 18 - .20
4x5 20 - - .19
5x4 20 - .11
3x7 21 -1.00
3x8 24 -1.03
4x6 24 - .25
6x4 24 .19
5x5 25 .03
4x7 28 - .29
5x6 30 - .03
6x5 30 .12
4x8 32 - .32
5x7 35 - .07
6x6 36 .06
5x8 40 - .10
6x7 42 .02
6x8 48 - .01
- 250 -
-------
(<<((<((<<((((((<(<<(
»»>>»»»»»»»>
*********************
M
xxxxxxxxxxxxxxxxxxxxx
*********************
00 0 N M
CO M N N OS
w in
IONNNdSQON.*-NNNinU)y)
(M M N ^*y>IG>My>IOH
M (ft IO vl <* tf Q H N U) N N
MMH OH0NM Ir-IIOWOJ
W O <* N W
M n
-^ 4J
G
CU
T3
G
-«e-| cu
-P Q<
C CU
CU T3
1
cu
H
XI
ns
•H
M
(d
>
4->
G
CU
ro
G
CU
Ci
CU
'O
C
-H G -P M 4-
O H G f
•H cu T) a
m -P G G C
M-t W O O C
CU H D, 0 C
O -H X CU >
U fri W CO PI.
n
CN
I
<
O
H
&4
H
3
IT*
5
w
co
o
H
(<(<((<<»>>»>»>»>»>»>>
*********************
CD
XXXXXXXXXXXXXXXXXXXXX
*********************
00
00
c CM
y> H N
(MOpO co
HIOO'IJNI
(M N
NM
1 Q
^ rt)
CU
r-l
X)
frt
^- ID
-^ ,H
i i
M
m
t t
-M
C
CD
r?-v
O
G
-*: , CU
-P a
G CU
rH
XI
m
*«H
S-l
/rt
UJ
>
4J
,»!
c
cu
tJ
f-»
}~t
Q)
Dw
. cu
ts
CU ^ G
•H G 4-> H 4-
0 M C C
•H CU rO Q
^ 4J G G E
m w o o c
CU M a 0 C
O -H X CU >•
U fa W CO t
CQ
H
tf
E-i
CO
M
a
w
w
EH
s
o
o
04
- 251 -
-------
Gas Velocity
FIGURE A-23Velocity Distribution Simulating Profile Measured
at Black Dog Power Station
I.D. Fan Inlet Station No. 2 (See Figure A-27 )•
- 252 -
-------
TABLE A-24
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-24
Number of traverse points along x axis
Number of traverse points along y axis
t
4x3
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
1x1 1 15.71
1x2 2 , 11.40
2x1 2 6.89
1x3 3 9.51
3x1 3 11.44
1x4 4 8.76
2x2 4 1.04
4x1 4 12.23
1x5 5 8.41
5x1 5 12.40
1x6 6 8.21
2x3 6 - .51
3x2 6 2.61
6x1 6 12.44
1x7 7 8.09
1x8 8 8.01
2x4 8 -1.08
4x2 8 2.86
3x3 9 . 1.02
2x5 10 -1.35
5x2 10 2.91
2x6 12 -1.50
3x4 12 .46
4x3 12 1.26
6x2 12 2.91
- 253 -
-------
TABLE A-2 4 (Cont)
% Error in Measured
Total Number of , Total Volumetric
Matrix Traverse Points Flow Rate
2x7 14 -1.59
3x5 15 .21
5x3 15 1.30
2x8 16 -1.65
4x4 16 . .71
3x6 18 .07
6x3 18 1.30
4x5 20 - .46
5x4 20 .75
3x7 21 - .02
3x8 24 - .07
4x6 24 .32
6x4 24 .75
5x5 25 .50
4' x 7 28 .24
5x6 30 .36
6x5 30 .50
4x8 32 .18
5x7 35 .28
6x6 36 .37
5x8 40 .23
6x7 42 .28
6x8 48 .23
- 254 -
-------
< < ( < < S.>>£.&.i.i.i.c * > * >>>' "*"
****************
M
*•**•*•<*<»*>«
u>«>
•-«-
xxxxxxxxxxxxxxxxxxxxx--
*********************
(TtOOHNCMNUOSOOOOOOOSMOOinNinQOinOOO
OinCMQSTfNt\ATlMU>Oy)CM'a-CO M 00 ^ rl
CMCMy)H'HHy)M^'OOOinMOMOV)OH ~
inM^-^HCMooN^t^-inintoincMHvooo
iricMrtNOSOU)OSNNOOinHr1inU>T4CMOOOinOS
QO*MOOMCMCMTlMV0QOinCMHU>M)NO *• CM M
HH CMM -H
OS CM t
N M in
CM
vo in M +
I CMM rl
o oo in M>-
tf) * U) •*
CM CJS CM m
1 H I
•H
IW
4-1
a)
o
0)
xt
•H
(0
c
0)
•o
c
Q)
C
H
M
( C C C (
» > »
*****
CM
m
(>(>
* *
C C ( ( C C
>>>>>>
******
o o
OOT>>>>>
******
CM CM CM M M M
((((((((( C ((((((( t ((<
xxxxxxxxxxxxxxxxxxxxx
*********************
Q)
r-l
XI
•H
(d
M
oo in
o *
OSVDHCMVOCMMOin^TtONVOCMVOtJjyj
inososinNoinoHMNOscMMCMiny)Ti
^•My)HOS'Hy)itooososy)y)*M'H'riy)
HMOOOSOSM-*NMCMMOOOSCMy)inCM
CMOOOOfMOOOONK
C4 00 ^ U) O Cft ^ v^
I M CM * N <0 OS
1
M 00
B*
CM
CMOOOHOtMOO
N os oo in oo u>
M 00 3; «> M «)
M
«> N
T H
0) T)
c c
o o
cu o
XI V
W CO
(U
X!
(d
•H
(d
4-1
c
cu
tJ
G
0)
c
CD T3
fi fi
o o
o
X Q)
W CO
4J
c
cu
c
o
Oi
X
4-1 H 4J
c
(U
c
o
CU
X
w
-------
Gas Velocity
FIGURE A-24Velocity Distribution Simulating Profile Measured
at Black Dog Power Station
I.D. Fan Tnlet Station No. 3 (See Figure A-27 )•
- 256 -
-------
TABLE A-2 5
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-25
Number of traverse points along x axis
x Number of traverse points along y axis
4x3
% Error in Measured
Total Number of - Total Volumetric
Matrix Traverse Points Flow Rate
1x1
1x2
2x1
1x3
3x1
1x4
2x2
4x1
1x5
5x1
1x6
2x3
3x2
6x1
1x7
1x8
2x4
4x2
3x3
2x5
5x2
2x6
3x4
4x3
6x2
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
12
- 257 -
-35.60
15.70
-24.91
.96
-22.85
- 2.78
8.39
-22.13
- 4.13
-21.79
- 4.75
2.26
6.79
-21.61
- 5.09
- 5.29
.83
6.21
2.15
.37
5.94
.17
.98
2.09
5.79
-------
TABLE A-25 (Cont)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points • Flow Rate
2x7 14 .07
3x5 15 .56
5x3 15 2.05
2x8 16 .02
4x4 16 .99
3x6 -18 .38
6x3 18 2.03
4x5 20 .58
5x4 20 .99
3x7 21 .28
3x8 24 .22
4x6 24 .40
6x4 24 .98
5x5 25 .59
4x7 28 .29
5x6 30 .39
6x5 30 .58
4x8 32 .23
5x7 35 .29
6x6 36 .39
5x8 40 .23
6x7 42 .29
6x8 48 .22
- 258 -
-------
M «• in
M * m vo o H
>>»»»>>»>»>»>
*******************
MMMMMMM^^Tl-trrtftfinin
( C ( ( ( C ( C ( C C
xxxxxxxxxxx
***********
^ CO ©
ft CM Q
ft O VO
CM CM *
C C ( ( C
X XX X X
* * * * *
00 N
ft
0^ (Tk CD
in H in
fv CM UD 00
M CM « N
I O *• N
H MM
-
1
1 CM
CO
CM CM .
tf H in
(ft M w OD 00
^ ^ CO CM Irt
in t * in ft
*J5 00 N CM «)'
00 CM N ft M
ft -H ino N
Q th U> ft U)
r« M ^t CM I
O
•H
*M 4->
U-l
OJ
O
U
WOOD
H O4 U &
•H X 0) X
fe W CO W
m
CN
I
O
H
O
K
W
2
o
D
m
H
M
yj
in
CM M ^ m
c c c c c c c c c c c c c c
**********
********
(U
QCDCOCSOCOOtH^THH-H-HrlCMCM
CM CM CM
C C ((<((( C ((((( C C C (( C
xxxxxxxxxxxxxxxxxxxx
********************
ft CD
ft Gd
ft ft
ft CD
oo vo ft
<0 N •*
00 ^ ft CM *
00 N in M ft
CM M N 00 N
CO CM CM in O
Q)
^ o M
*«"H
II
^ 00 M) (M 10 00 ^ IO 10 Q\ M ^ N»
rH O N- ^Or^OOIs^^^f*O(MN>
IO ^JD VO ^0 VD ^ 00 VD CD f*l M CM ^»
(MCSrj-ftOOin'HCOinN^ftCM
IM^tftNftl CM G5 00 CM M
H CM CM I rl CM M H
II II
X
CO W
w
H
Q
H
B
O
S3
3
o
0H
- 259 -
-------
FIGURE A-25Velocity Distribution Simulating Profile Measured
at Black Dog Power Station. I.D. Fan Inlet
Station No. 4 (See Figure A-28 ).
- 260 -
-------
TABLE A-2 6
EVALUATION OF VARIOUS EQUAL AREA SAMPLING STRATEGIES
FOR THE VELOCITY DISTRIBUTION PRESENTED IN FIGURE A-26
Number of traverse points along x axis
j; Number of traverse points along y axis
4x3
% Error in Measured
Total Number of . Total Volumetric
Matrix Traverse Points Flow Rate
1x1 1 5.07
1x2 2 • -4.50
2x1 2 - .87
1x3 3 3.28
3x1 3 -1.81
1x4 4 5.26
2x2 4 -9.18
4x1 4 -2.13
1x5 5 5.99
5x1 . 5 -2.27
1x6 6 6.34
2x3 6 -1.74
3x2 6 -9.44
6x1 6 -2.35
1x7 7 6.52
1x8 8 6.63
2x4 8 - .03
4x2 8 -9.49
3x3 9 -2.34
2x5 10 .54
5x2 10 -9.51
2x6 12 .79
3x4 12 - .70
4x3 12 -2.53
6x2 12 -9.51
- 261 -
-------
TABLE A-26 (Cont)
% Error in Measured
Total Number of Total Volumetric
Matrix Traverse Points Flow Rate
2x7 14 .91
3x5 15 - .14
5x3 15 -2.61
2x8 16 .98
4x4 16 - ,91
3x6 18 .10
6x3 18 -2.65
4x5 20 - .36
5x4 20 -1.00
3x7 21 .22
3x8 24 .29
4x6 24 - .12
6x4 24 -1.05
5x5 25 - .46
4x7 28 0
5x6 30 - .22
6x5 30 - .51
4x8 32 .08
5x7 35 - .09
6x6 36 - .27
5x8 40 - .02
6x7 42 - .15
6x8 48 - .07
- 262 -
-------
WOONCOOSptTJOS
ID O l^» N» N* CD M CD
tt CM : :
rH
CN
a
D
O
O
S3
O
H
EH
D
CQ
* *
> >
* *
( (
((( C ((
((
CD O O O CD O
CM-
xxxxxxxxxxxxxxxxxxxxx-
*********************
oifMN-inmNT
oy)^'^j'THoo(vicft
(sinkocft^tNCMCd
QC9MCMNOOHCM
MCS^fCMinMCfiMVOCSO^
VO OSrlO'HinH'iO^tin
*
4->
C
0)
H3
C
Q)
4-1 CX
0)
rd
O H
-H
4-1
d
0)
c
(0
•H
M
rti
4-1
d
0)
T>
d
0)
cx
o
t3
a u
X OJ
W
ffi
EH
O
En
H
o
C 4J H 4-»
d
0)
d
w o o o
ex
x
01 w
- 263 -
-------
X
Gas Velocity
ft/sec
m/sec
FIGURE A-26Velocity Distribution Simulating Profile Measured
at Black Dog Power Station I.D. Fan Inlet
Station No. 5 (See Figure A-28 ).
- 264 -
-------
.TABLE A-27
MEASUREMENTS TAKEN DOWNSTREAM
OF ELECTROSTATIC PRECIPITATOR AT ALLEN S, KING PLANT
Matrix
Total Number of
Traverse Points
% Error in
Volumetric
Flow Rate
% Error in
Particulate
Coiicen trat ion
% Total
Emission
Level Error
1
1
2
1
3
1
2
4
1
5
1
2
3
6
1
1
2
4
3
2
5
2
3
4
6
2
3
5
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
1
2
1
3
1
4
2
1
5
1
6
3
2
1
7
8
4
2
3
5
2
6
4
3
2
7
5
3
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
12
14
15
15
-4
2
. -3
3
-3
3
-3
3
-3
3
-3
3
3
.89
.75
.58
.44
.56
.48
.49
.56
.45
.57
.43
.45
.37
.57
.41
.40
.38
.35
.26
.33
.35
.36
.18
.22
,35
,29
.14
,20
-16.
- 5.
-14.
•
-14,
1.
- 6,
-15.
1.
-15.
1.
1.
-8.
-15.
1.
1,
2.
- 7.
*
2.
- 7.
1.
•
^ •
- 7.
1.
.
- .
02
20
74
80
90
39
91
00
39
00
23
93
04
06
12
07
52
93
05
25
77
23
86
11
66
71
75
11
-20
- 2
-17
4
-17
4
- 6
-18
4
-18
4
2
-7
-18
4
4
2
- 7
2
- 7
1
1
- 7
2
.13
.59
.79
.27
.93
.92
.45
.03
.89
.03
.70
.38
.70
.09
.57
.50
.90
.61
.31
.59
.45
.59
.04
.11
.34
.00
.89
.09
- 265 -
-------
TABLE A-27 (Cont)
% Error in % Error in % Total
Total Number of Volumetric Particulate Emission
Matrix Traverse Points Flow Rate Concentration- Level Error
2
4
3
6
4
5
3
3
4
6
5
4
5
6
4
5
6
5
6
6
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
8
4
6
3
5
4
7
8
6
4
5
7
6
5
8 •
7
6
8
7
8
16
16
18
18
20
20
21
24
24
24
25
28
30
30
32
35
36
40
42
48
.28
.14
.11
.20
,10
.12
.10
.09
.08
,12
.09
.06
.06
.08
.05
,05
.06
.04
.04
.04
1.50
,70
.54
- ,03
.59
.70
,38
.27
.43
,75
.64
.32
.48
.70
,16
,32
.54
.21
.38
.27
1.78
.84
.65
.17
,69
.82
.48
.36
.51
.87
.73
.38
.54
.78
.21
.37
.60
.25
.42
.31
- 266 -
-------
TABLE A-28
MEASUREMENTS TAKEN UPSTREAM OF PRECIPITATOR
AT ALLEN S. KING PLANT
Matrix
Total Number
of
Traverse Points
% Error in
Volumetric
Flow Rate
% Error in
Particulate
Concentration
% Total
Emission
Level Error
1
1
2
1
3
1
2
4
1
5
1
2
3
6
1
1
2
4
3
2
5
2
3
4
6
2
3
5
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
1
2
1
3
1
4
2
1
5
1
6
3
2
1
7
8
4
2
3
5
2
6
4
3
2
7
5
3
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
12
14
15
15
10.54
3,92
5.86
4.14
5.82
4.04
.79
5.87
3.95
5,90
3.88
.78
.62
5.92
3.84
3.81
.62
.58
.57
.52
.57
.45
.41
.52
.57
.41
,31
.51
- 267 -
13
1
-
3
-1
3
1
-1
3
— 1
3
1
-1
3
3
1
1
-
-
1
-
-
.50
,33
,61
.63
.28
.90
.16
.37
.90
.39
.86
.31
.07
.39
.82
.79
,25
.03
.05
.28
.01
.31
.03
.02
.00
,33
,01
.04
25.
5.
5.
7.
4.
8.
1.
4.
8.
4.
7.
1.
1.
4.
7.
7.
•
1.
•
.
1.
•
•
•
1.
*
•
t
46
30
21
92
47
10
96
42
00
43
89
09
70
45.
81
74
87
62
62
80
59
76
38
50
58
74
30
47
-------
TABLE A-28 (Cont)
% Error in % Error in % Total
Total Number of Volumetric Particulate Emission
Matrix Traverse Points Flow Rate Concentration Level Error
2 x
4 x
3 x
6 x
4 x
5 x
3 x
3 x
4 x
6 x
5 x
4 x
5 x
6 x
4 x
5 x
6 x
5 x
6 x
6 x
8
4
6
3
5
4
7
8
6
4
5
7
6
5
8
7
6
8
7
8
16
16
18
18
20
20
21
24
24
24
25
28
30
30
32
35
36
40
42
48
.38
.37
.24
.50
,27
.35
.20
.17
.20
.35
.25
,16
.19
.25
.13
.14
.18
.12
.14
.11
.35
-.10
.02
-.05
-.08
-,13
,05
.. ,07
-.05
-.14
-,11
-,02
-.08
-.12
0
-.05
-,09
-.03
-.07
-.05
.73
.27
.26
.45
,19
,22
.25
.24
.15
,21
.14
,14
.11
.13
.13
,09
.09
.09
.07
.06
- 268 -
-------
TABLE A-29
Matrix
MEASUREMENTS TAKEN DOWNSTREAM (NO TURNING VANES)
OF TEST ELBOW IN 1/10 SCALE MODEL
Total Number of
Traverse Points
% Error in
Volumetric
Flow Rate
% Error in
Particulate
Concentration
% Total
Emission
Level Error
1
1
2
1
3
1
2
4
1
5
1
2
3
6
1
1
2
4
3
2
5
2
3
4
6
2
3
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
1
2
1
3
1
4
2
1
5
1
6
3
2
1
7
8
4
2
3
5
2
6
4
3
2
7
5
1
2
2
3
3
4
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
12
14
15
21.
20,
-0.
21.
3.
21.
-13.
3,
21.
3.
21.
-8.
-4.
3.
21.
21.
-6.
-3.
-1.
-6.
""* J •
-6.
-0,
-0.
-3.
-6.
-0.
26
35
40
12
59
40
26
81
53
63
60
27
72
46
64
66
92
55
78
39
45
14
90
98
52
00
53
269 -
-14.
-24.
-12.
-11.
-16.
- 9.
- 7.
-17,
— 8 «
-17.
- 8.
2.
-16.
-17.
- 8.
- 8.
4.
-17.
- 2.
5.
-17.
5.
- 0.
- 3.
-17.
5.
0.
75
58
35
75
76
40
85
40
80
48
63
54
02
46
57
56
65
17
96
26
30
49
46
76
26
60
22
3.
-9,
-12,
6.
-13.
9.
-20.
-14.
10.
-14.
11.
- 5.
-19.
-14.
11.
11.
-2.
-20.
-4.
— 1 ,
-20.
^ •
-1.
-4.
-20.
-0.
-0.
37
32
70
89
77
99
07
25
84
49
11
94
98
60.
22
25
59
11
69
47
15
99
36
70
17
74
31
-------
TABLE A-2 9 (Cont)
Matrix
Total Number of
Traverse Points
% Error in
Volumetric
Flow Rate
% Error in
Particulate
Concentration
% Total
Emission
Level Error
5x3
2x8
4x4
3x6
6x3
4x5
5x4
3x7
3x8
4x6
6x4
5x5
4x7
5x6
6x5
4x8
5x7
6x6
5x8
6x7
6x8
15
16
16
18
18
20
20
21
24
24
24
25
28
30
30
32
35
36
40
42
48
-0.96
-5.91
-0.19
-0,35
-1.05
-0,14
-0.20
-0.24
-0.17
0.31
-0.30
0.13
0.41
0,29
-.02
0.47
0.39
0.18
0.45
0.28
0.34
-3.87
-5,65
-1.21
0,44
r3.85
-0.52
-1,31
0.53
0,56
-0.30
-1.30
-0.62
-0.21
-0.39
-0.60
-0.18
-0.31
-0.37
-0.27
-0.29
-0.25
-4.79
-11.23
-1.40
-0.09
-4.86
-0.66
-1.51
0.29
0.39
0.01
-1.60
-0.49
0.20
-0.10
-0.58
0.29
0.08
-0.19
0.18
0.01
0.09
- 270 -
-------
TABLE A-30
MEASUREMENTS TAKEN UPSTREAM' (No Turning Vanes)
OF TEST ELBOW IN 1/10 SCALE MODEL
Total Number of
Matrix Traverse Points
1
1
2
1
3
1
2
1
5
1
2
3
6
1
1
2
4
3
2
5
2
3
4
6
2
3
5
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
1
2
1
3
1
4
2
5
1
6
3
2
1
7
8
4
2
3
5
2
6
4
3
2
7
5
3
1
2
2
3
3
4
4
5
5
6
6
6
6
7
8
8
8
9
10
10
12
12
12
12
14
15
15
% Error in
Volumetric
Flow Rate
5.69
5.94
3.22
5.68
2.16
5.61
1.97
5.58
1.93
5.57
1.07
-.62
1.91
5.57
5.56
1.14
-.65
.08
1.11
-.61
1.07
.14
-.03
-.57
1.04
.13
-.03
- 271 -
% Error in
Particulate
Concentration
-6.39
-2.30
-12.29
5.15
-11.47
6.36
-11.17
6.67
-11.03
6.65
-6.92
-3.08
-10.96
6.64
6.62
-6.04
-2.97
-3.00
-5.51
-3.04
-5.14
-2.29
-2.33
-3.10
-4.89
-1.81
-2.16
% Total
Emission
Level Error
-1.06
3.50
-9.46
11.12
-9.56
12.33
-9.42
12.62
-9.31
12.59
-5.92
-3.68
-9.26~
12.58
12.55
-4.97
-3.60
-2.92
14.46
-3.63
-4.12
-2.15
-2.36
-3.65
-3.90
-1.68
-2.19
-------
TABLE A-30(Cont.)
Matrix
Total Number of
Traverse Points
% Error in
Volumetric
Flow Rate
% Error in
particulate
Concentration
% Total
Emission
Level Error
2x8
4x4
3x6
6x3
4x5
5x4
3x7
3x8
4x6
6x4
5x5
4x7
5x6
6x5
4x8
5x7
6x6
5x8
6x7
6x8
16
16
18
18
20
20
21
24
24
,24
25
28
30
30
32
35
36
40
42
48
1.02
.03
0.11
-.01
.02
.02
.10
.09
0
.03
.01
-.01
-.01
.02
-.03
0.02
.01
-.03
-.01
-.02
-4.72
-1.58
-1.51
-2.11
-1.12
-1.40
-1.31
• -1.18
-.84
-1.34
-.95
-.66
-.67
-.90
-.53
-.50
-.63
-.38
-.46
-.34
-3.75
-1.55
-1.40
-2.12
-1.10
-1.38
-1.21
-1.09
-.84
-1.31
-.94
-.67
-.68
-.88
-.56
-.52
-.64
-.41
-.47
-.36
- 272 -
-------
TABLE A-31
Matrix Total Number of Points
1 x 1 1 i'00
1x2 2 0.50
2x1 2 2.00
1x3 3 0.33
3x1 .3 - 3.00
1x4 4 0.25
2x2 4 1.00
4x1 4 4.00
1x5 5 0.20
5x1 5 5.00
1x6 6 0.167
2x3 6 0.667
3x2 6 1-50
6x1 6 6.00
1x7 7 0.143
1x8 8 0.125
2x4 8 0.50
4x2 8 2.00
3x3 9 1-00
2x5 10 0.40
5x2 10 2.50
2x6 12 0.33
3x4 12 0-75
4x3 12 1-33
6x2 .12 ' 3.00
2 x 7 14 0.286
3x5 15 0.60
5x3 15 I-67
2x8 16 0.25
4x4 16 1.00
- 273 -
-------
TABLE A-31 (Cont)
Matrix Total Number of Points
3x6 18 0.50
6x3 18 2.00
4x5 20 0.80
5x4 20, 1.25
3x7 21 0.429
3x8 24 0.375
4x6 24 0.667
6x4 24 1.50
5x5 25 1.00
4x7 28 0.571
5x6 30 0.833
6x5 30 1.20
4x8 32 0.50
5x7 35 0.714
6x6 36 1.00
5x8 40 0.625
6x7 42 0.857
6x8 48 0.75
- 274 -
-------
Inlet
Grate
50
o o o o o
T~\
2,44m
e
o
T
1
+ 4- + 4- 4-
4 4 + -f- 4
t- 4 4- t t
-t- t + + +
U U U U LT
Sampling
Station 1
VD
cn
h-
n
4-
4-
t-
4-
4-
4-
4-
~ 2,
n
-U
4-
4-
+
4-
-»-
+
44m
n
+
4-
4-
4-
+-
4-
+
n
f
4
•f-
4
•4-
4-
4
Sampling
Station 2
p
4- 4-
4- t-
+ -1-
r . t
4 1-
4- 4
Sampling
Station 3
FIGURE A-27Duct Layout at Black Dog Power Station
- 275 -
-------
Station 5
Station 4
FIGURE A-28Duct Layout at Black Dog Power Station
(Sampling Stations 4-5)
- 276 -
-------
- 277 -
-------
GAS VELOCITY
c
METER/SEC "1
FT/SEC J
FIGURE A-30
'SIMULATION OF THE VELOCITY DISTRIBUTION
ABOUT FOUR STACK DIAMETERS ABOVE
THE STACK'INLET IN-A 3.35__
METER DIAMETER FOUNDRY STACK
(REFERENCE 33)
- 218 -
-------
nUXDrldJM^-insOSrlOJM^-invO
( ( (
>»
(MNWMMMMMMM
xxxxxxxx
********
m
* *
X XX X'
* * * *
ftHftVOM^NtMin
w^vokominijDN^ty)
M®ftomftinftcoM
CdCDftCdHTlHftCOft
N W •* t M
in ft ft co *• N N in N
o in ft * N m co
4-3
C
— . CU
~~* 1 I"Q
c
' 0)
c a
"J 41 i
r—l
ro
-rH
fO
••*
c
CD
-a
c
cu
a
CD
-, 1 -
U C 0 rj-j Q
£ *-• c c c
H-l 4-> O O C
cu w a o c
o % x 0 >
O PL, W CO P.
o
ro
I
PJ
O
I—I
P-.
s
1—I
g
H
CQ
II II
kmvu.
o cs) © o o o CD
(((((<
< t
i-c>5
'x^x^x^x^^xk^x^x^
NNOOMftinNOOCMCSCMMVDft
cBMvoftco^'f^inO'^ft'Tftvo
rHOO(MftMtQCON*ininrjJ;VO
ftC4r1NVOftN>NNrlOOOOM09
M M tf)
voiniyvoft^ftin gfocp
N * ft N
M M r< M
VI I v ^^ ^~i • • > ^ * •• i**r ** • •» *
ICJMH rlVDC9MMC4
I I H I
n vo
C
cu
-rH
O
QJ
O
O
O)
ro
ra
a>
T3
C
(D
a.
P-,
c
o
a
x
cu
I—I
(D
-rH
ro
CU
T3
0
a
0
c
t-H
T3
C
O
O
0
W
C
0
C
o
a
EH
ra
i — i
Q
w
Hrt
HH
H
05
O
P-,
i-4
-------
10.0
6.0 -
4,0 -
2.0 -
Error
1.0
0.8
0.6
0.4
0.2
0.1
B
Qj
i.1)
El
•'v >
A
0
a
30 60 90 120
Rotation Angle (a ) - Degrees
150
Symbol
O
D
O
A
No. of Points
in Sample
4
8
12
16
180
'FIGURE A-3lThe Effect of Angular Orientation on the Accuracy
of Four Different 2 Ray Sampling Strategies for
the Velocity Profile Shown in Figure A-30
- 280 -
-------
10.0
6.0
4.0
2.0
%
Error
1.0
0.7
0.4
0.2
0.1
0
O
o
0
H
E3
A
a
A"'
o
30 60
Rotation Angle (a ) - Degrees
90
Synfibol
O
n
O
A
No. of Points
8
16
24
32
in Sample
a
Note; Flags indicate errors less than 0.1%
FIGURE A-32 THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 4 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-30
- 281 -
-------
10.0
7.0
4.0
O
o
2.0
1.0
0.7 I—
D
D
0.4
0.2
O
O
o.i
15
ROTATION ANGLE ( a ) - DEGREES
A
1,
30
45
Symbol
O
D
O
A
Number of Points in Sample
16
32
48
64
Note: Flag indicates error less than 0.1%
FIGURE A-33THE EFFECT OF ANGULAR ORIENTATION ON THE ACCURACY OF FOUR
DIFFERENT 8 RAY SAMPLING STRATEGIES FOR THE VELOCITY
PROFILE SHOWN IN FIGURE A-30
- 282 -
-------
GAS VELOCITY
METERS/SEC
FT/SEC
L
]
FIGURE A-34
VELOCITY DISTRIBUTION SIMULATING THAT MEASURED
AT THE 91 METER (300 FT) LEVEL IN A 9.1 METER
(30 FT) DIAMETER STACK AT A LARGE COAL-FIRED
ELECTRICAL GENERATING STATION
(REFERENCE 33)
- 283 -
-------
m
M * m
n *«
NMMMMMMM***'*'*1*"*'-
(((((<(((((((<(((
X^^ *^^' ^^ ^^ "^^ *^^ *^T *^r* ^^ *V^ *V ^^ *^^ ^^ *W^ *W
^^ ^T» «rN ^^> ^^ ^V J^ Jv ^^ ^*W rfV J>» ^N> ^X ^% ^
*****************
m^*Hoocou)inoc3in®inT<*-ineMco
HMCMOOSOHinHNininNflMMN
00(v1(MCON»(M^'(r>VDMCB(M(MfM'4"0^f^-
ojMy)^''H<*r 0^ M
M in r-i
«nw
r4Nff>M-HOONV£)MMNOOinOO(M
o^iOHo^Nini" —
O M (M 00 M M
't N 00 M Q N
I N OS NU) U)
M Ch M
I I I* %W M/ W \V
N t in m ^ N-
o vo M in d
"* -H C -P
O H C
•H Q)
MH -P fi
>4H W O
0) >-l DJ
O -H X
U fa H
m
u/
rH
£1
(d
•H
^
(0
>
4J
d
0)
rrj
V^
c
0)
a<
QJ
T3
G
H
•d
c
o
u
0)
CO
— 1
4-
c
Q
f.
c
c
>
ft
m
I
ID
0
H
o
H
EH
IT)
ffi
O H (M M -«r-
****************
O
CM (M
XXXXXXXXXXXXXXXXXX
******************
CM (M
u) in
0"» N
rl (M
W) M
n N
00 N
* M
in H
^ t
•H CD
in M
•4g —
-p
0)
•H
U
•rH
HH
m
0)
o
0)
rH
(0
•H
J_l
(td
_ ^
~ -p
a
c
0)
a
•d
C -P
H C
QJ
-P C
w o
M &,
•H X
vu
XI
•iH
J_j
rt
>
-P
^J
(U
ti
0)
D.
-ri
d
H
T>
C
0
o
0)
QJ
C
O
X
EH
CO
H
Q
W
W
EH
O
fa
O
o
£4
U fa
CO
- 284 -
-------
10.0
6.0
4.0
2.0
%
Error
1.0
0,8
0.6
0.4
0,2
0.1
0 0 0 0
13
As
LJ
/\s
0
0
<•>
-Symbol
O
n
O
A
30 60 90 120
Rotation Angle ( a) - Degrees
150
180
— -Mo. of Point-ln-Sample-.
4
8
12
16
Note; Flags indicate errors le'ss"" tTTan 0.1%
FIGURE A-35THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 2 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-34
- 285 -
-------
10.0
7.0
4.0
2.0
%
Error
1.0 -
0.7
0.4
0.2
0.1
0
0
L'l
O
0
30 60
Rotation Angle (a ) - Degrees
No_... of Points in Sample
8
16
24
32
90
Note; Flags indicate errors less than 0.1%
FIGURE A-35 THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 4 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-34
- 286 -
-------
10.0
7.0
4.0
2.0
Error
1.0
0.7
0.4 ^
0.2
0.1
I
I
15 30
Rotation Angle (a ) - Degrees
Symbol
O
D
O
A
No. of Points in Sample
16
32
48
64
FIGURE A3 7THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 8 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-34
- 287 -
-------
GAS VELOCITY
DV1ETERS/SEC I
FT/SEC J
FIGURE A-38 VELOCITY DISTRIBUTION SIMULATING THAT MEASURED
2.2 STACK DIAMETERS ABOVE THE BREECHING PLANE
INLET IN THE STACK (2.4 METER DIAMETER) FROM
AN OIL-FIRED STEAM GENERATOR
(REFERENCE 43)
- 288 -
-------
v in
rxi 01
( (
88
rl CM
M CM
-785916
642342
MJ
1
>
CO
(M
CM
-204320
OS r<
>
ri
X
Q U)
y> ID
N CM
-49
-36188
01 m
pi
X
COM
M O
248823
-846366
v «
>
n
(v v
X X
Ch ID
00 ID
ID *
VOID
1459780
-1226379
w
CM
ID
(T>
398963
Q* rl CM
* * *
N UJ CM
M CM N
03 tffc 0S
rl N. tf)
22
17543
-122091
m v
>
LL
X X
00 CM
00 CM
CM ID
424530
-745549
u
i
634843
Ml —«-" •
C
^' "*
Q)
(d
•H
(0
- •*
C
QJ
M -* 1 C
M 1 Q)
M C (U
Q) T3
M -H C -I
IS.. U H
ID -H
CO 4-1 -P
Q M-l CO
1 O -H
U fa P
1
Q)
H
C
Q
C
C
X
fc
00
m
I
D
O
H
3
O
H
u
*»
O
II II
o
niHHHVH^5
CD CD GJ O ^d CD
**
CM CM CM
( ( C ( C C C C C ( C C C C C C ( (
xxxxxxxxxxxxxxxxxx
******************
NIDMNQrlCMONOOOOMMOSMOONSO
0>IDO>fMCT»CSOOMnNOOIO
C
(U
'd
c;
Q)
p.
Q)
'd
1
0)
i-H
XI
(0
•H
(0
>
4J
£j
0)
•d
C
0)
cu
(U
T3
d
C 4-> H
H C
Q) T3
•P G C
WOO
M O4 O
-H X OJ
fa W U3
4-
C
a
c
C
D
X
K
H
Q
M
ffl
EH
H
a
o
- 289 -
-------
10.0
7.0 -
4.0 -
2.0
%
Error
1.0
0.8
0.6
0.4
0.2
0.1
H
I
Q
0
A
I
El
CD
A A A
A
L3
I
Q
A A
0
Q
A
..Symbol
O
n
30 60 90 120
Rotation Angle ( a) - Degrees
No. of Points in SamglLe
4 ""
8
12
16
150
Note: Flags indicate errors less than 0.1%
FIGURE A-39 THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 2 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-38
- 290 -
180
-------
10.0
7,0
4.0
2.0
%
Error
1.0
0.7
0.4
0.2
0.1
A
H
I
Symbol
O
D
c-
30 60
Rotation Angle ( a) ~ Degrees
of Points in Sample_
" 8
16
24
32
90
Note; Flag indicates error less than 0.1%
FIGURE A-40 THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 4 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-38
- 291 -
-------
10.0
7.0
4.0
2.0
Error
C.
1.0
0.7
a
0.4
0.2
0.1
/N
15
A
45
Rotation Angle ( a) - Degrees
. _ . . . , .^
Symbol No. of Points in Sample
O 16
D 32 ^kl L~
I O 48
I £. I 64
Note: Flags indicate errors less than 0.1% '
FIGURE A-4ITHE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 8 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-38
- 292 -
-------
GAS VELOCITY
METERS/SEC I
FT/SECJ
FIGURE A-42 VELOCITY DISTRIBUTION SIMULATING THAT MEASURED
AT THE 85.3 METER LEVEL (APPROXIMATELY 13 STACK
DIAMETERS ABOVE THE BREECHING PLANE INLET) IN
THE STACK AT THE ALLEN STEAM PLANT UNIT NO, 1
- 293 -
-------
< C C C t < C C < < C C C C ( C C C < C C
MMMMMMM^^^^-^^tinioininininin-
< c c ( c c c
X X X X X X X
*****#*
rl
3
®
c c c c c < < c c c c c c c
xxxxxxxxxxxxxx-
**************
y> N N co
00 CD U) CM
VD CM M CO
M rl VO VO
CD
00 tf) rl VO * *
m CM «• oo ft M
rl Tf 00 VO rl rl
M ft CO U) ft
+ ri co in m ca
rl ft W N. N 00
I CM M CM
I I
rl * N Y>
M N CD CM
CO M ft M
CO ft CM CM
N rlrl U>
IO MIO CO
f*^ 00 O
I rl
rl +
rl N
co *•-
CM CM
W) M
CO N
y> rl
I
m
•H
•H
«
>
«* -P
c
Q
T3
C
-« r O
-P QJ
C 0
H
m
I
•H
fd
-P
£
0)
^0
c
Q)
P,
0)
0) T3 fi
•H C -P H
0 H C
•H (1) T3
MH ^-) C C
U-) 0) O O
QJ M 0-: D
O -H X CU
U fa W CO
-i-
c
([
c
0
0
X
W
CN
I
u
H
H
O
CQ
M ^ «
C C C < ( C C C C ( C C ( C C C C C C C (
***************#***»*
CDOQCDCgQrlrlrlrlrlrlrlCMCMCMCM
XXXXXXXXXXXXXXXXXXXXX
CD
Q
O
O
ft VD m •*
N in O M
CM VO ft N
ft ft CO N
n
oo
in
co CM M M y>
rl M U) M 00
ft CM N « CM
"cIS?
rl CD
Irl
CMinCOCDTl-NTNNCMCMCO
MNQUrl ftin^tCOtf
N rl CM
CM CM
M CM
CM
rl O O
NCMO
CM
n
1
CD CM V
CM CM CM
CM CM
1- CM
TO rl *• CO
M Q N •*
Hrlrl
CM -*
fa W
(U
rH
•H
rd
4-)
C
QJ
13
C
o,
Q)
H
C
0
O
0)
Q)
fi
O
DJ
X
CO
H
Q
W
a
EH
ff,
o
fa
o
3
J
O
- 294 -
-------
10.0
6.0
4.0
2.0
%
Error
1.0
0.8
0.6
0.4
0.2
0
r-
V
I
I
Symbol
O
n
O
A
30 60 90 120
Rotation Angle ( a) - Degrees
Qf_JPoint_SL_ jn Sample
'"4 '
8
12
16
150
180
Note: Flags indicate" errors less than 0.1%
FIGUREA-43THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 2 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-42
- 295 -
-------
10.0
7.0
4.0
2.0
Erroi
1.0
0.7
0.4
0.2
0.1
I j: ,+
30 60
Rotation Angle ( a ) - Degrees
Symbol _________ No- of _. Pain ts in S
O
8
16
24
90
a
Note: Flags indicate errors less than 0.1%
FIGURE A-44THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 4 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-42
- 296 -
-------
10.0
7.0
4.0
2.0
%
Error
1.0
0.7
0.4
0.2
0.1 --
G
15 30
Rotation Angle ( ct ) ~ Degrees
45
.—Syrobo.L-
, of - Points- -in -Saiupl-e-
16
32
48
64
Note; Flags indicate errors less than 0,1%
FIGUREA-45THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 8 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-42
- 297 -
-------
GAS VELOCITY
METERS/SEC ~]
FT/SEC J
FIGURE A-46
VELOCITY DISTRIBUTION SIMULATING THAT MEASURED
AT THE 85.3 METER LEVEL (APPROXIMATELY 13 STACK
DIAMETERS ABOVE THE BREECHING PLANE INLET) IN
THE STACK AT THE ALLEN STEAM PLANT UNIT NO.2
- 298 -
-------
<«• m
H CM M <* in v)
M *• m
< c
CM CM
X X
* *
CM "4"
* CO
N In
-258314.
220912
> >
CM M
X X
* *
u) 00
N 00
tf CM
-73554
— 1
> >
M M
X X
* *
VD
^t CM
N H
m o
-1651
34770
>• j-
M M
X X
* *
Oi n
CM N
00 00
-137343
254675
j- j-
M
X
* *
00 N-
H M
ijO fs.
O -t
-229698
80531
^ >
X
* *
Al |rt
M 0
ff> in
N (5
TS
N
1
T f
X
* *
IM
y) N
(Tt M
-6289
33924
^ ^
X
* *
fS. (^
CT) 00
-73952
75175
- •<
%
00
CO-
00
CM
1
(D
rH
J3
Varia
-P
G)
•r)
"I d)
-!-> a.
Ci QJ
QJ rO
H C 4
O H f
•H (
^1 4-> (
4H W (
0) M (
O -H >
u OH r.
1
Q)
rH
.Q
id
JJ
enden
Q/
fi
-> H
-4
U T3
5 o
X O
< Q)
.1 W
4-
C
a
c
c
><
P!
H
O
H
EH
D
m
^
O
X
o
'ro
I!
CD
EH
Cfl
H
Q
W
H
g
O
- 299 -
-------
10.0
6.0
4.0
2.0
%
Error
1.0
0.8
0.6
0.4
0.2
0.1
1
I'l
0
30 60 90 120"
Rotation Angle (a ) - Degrees
0
0
1
THTT
Symbol
No. of Points in Sample
4 ~
8
12
16
Note; Flags indicate errors less than 0.1%
FIGURE A-47THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 2 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-46
- 300 -
-------
10.0
7.0
4.0
0.4 _
0.2 _
0
0
0
0
2.0
Error
G
1.0 _
0.7 _
LJ
[£]
7
0.1
30 60
Rotation Angle ( a) - Degrees
r- -Symbol--
i. of Pqints in Sample ,'
8 I
16 !
24
32
Note : Flags Indicate errors' less "than 0.1%
FIGURE A-48THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 4 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-46
- 301 -
y o
-------
10.0
7.0
4.0
2.0
%
Error
1.0
0.7
0.4
0.2
0.1
A
15 '30
Rotation Angle (a ) ~ Degrees
45
O
u
O
.....A
Note:
No«.. e.f_ .Points in . Sample...
16
32
48
64
Flags indicate errors less than 0.1%
FIGURE A49THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 8 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-46
- 302 -
-------
73
FIGURE A-50
GAS VELOCITY
[~ METERS/SEC"]
L FT/SEC _J
VELOCITY DISTRIBUTION SIMULATING THAT MEASURED
AT THE 30.5 METER (100 FT.) LEVEL' (APPROXIMATELY
FOUR STACK DIAMETERS ABOVE THE BREECHING INLET
PLANE) IN THE STACK (DIA=5.3 M) AT THE COLBERT
STEAM PLANT UNIT NO.l
- 303 -
-------
((<(((((((((((((<
>>»>»»»»>>»--^
*****************
N
M M M M M M
kxkxx
* ** **
ON IO O 00 N N
M O N M ON ^
M (M H ON ON ©
<0 CO Ol ON ON if
( ( (
XXX
* * ¥
( ( c <
X XX X-
* * * *
IO rl ON
N IO N M ON ON
IO CNI -fr y> CO
COMinNy>!fr<\INCO
ON ON IO
N ON 00 N
N(OMG)ON<9*IOG)
-M
C
_-. 0>
— 1 "0
c
i QJ
c °<
(i, 0 .
cu
1 — 1
,-Q
ro
-•-H
1-H
ro
>
-M
, — *
c*
a;
T3
C
CU
a
CD
-a
, ^ ^
-^ "a c •"• I
H £ cu -a a
CM a c c
^ +-• o o c
X w a o t
o -^ x cu >
U PL, pq co PI
o
m
<
w
pi
P
O
i—i
r-M
S
l—I
S
O
I—I
H
t>
PQ
H
CO
I - 1
P
w
K
H
o
X
I! II
>- o
((((((((((<(((((((
>>>»»>>»»>»»-*-
******************
*• oo
£H
IOH
00 H
N (M
ON H
N ON
ON \0 N *
ON N N 00
N «> * *
N 10 ON vl
ON 10 CD
* 00 tf «JO 00
M «) M (M 00
N CO IN- * W
a» 10 «• CD w>
«)ION*NN.T<-Hr«T<
IONON T rHMMHrlN
HOOrl «) t «) COM O
rH I IO H kO ON ON
I M M(NJ I
10 M
03 ON
M M
H 00
M T<
rl
C
cu
-rH
U
CU
o
O
c
cu
T3
C
CU
a
c
cu
w
0)
„'-;
ro
C
CU
T5
C
CU
a
cu
T3
13
O
O
CU
CO
C
CU
C
o
a
x
O
PH
o
o
Cu
304 -
-------
10.0
6.0-
4.0
2.0
%
Error
1.0
0.6
0.4
0.2
0.1
O
O
,-J
30 60 90 120 150
Rotation Angle (a ) - Degrees
180
jBymbol
O
LJ
O
A
jNo. of Points in Sample
8
12
16
FIGURE A51 The Effect of Angular Orientation on the
Accuracy of Four Different 2 Ray Sampling Strategies
for the Velocity Profile Shown in Figure A-50
-305 -
-------
10.0
7.0
4.0
2.0
Erro:
1.0
0.7
QJ
0
Q
G
0.4
0.2
0.1
JZC
30 60
Rotation Angle ( a) - Degrees
Symbol
O
LI
No. of Points,
8
-Sample -
24
32
\
Flags i
errors less than 0.1
90
a
Figure A-52THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 4 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-50
- 306 -
-------
10.0
7.0
4.0
2.0
Error
1.0
0.7
0.4
0.2 .
0.1
Fh-
Q
A
Symbol
O
15 30
Rotation Angle (a ) - Degrees
No, of PQintS--in~Sampie-
16
32
48
64 _ _ ....
FIGURE A-53 THE EFFECT OF ANGULAR ORIENTATION ON THE
ACCURACY OF FOUR DIFFERENT 8 RAY SAMPLING
STRATEGIES FOR THE VELOCITY PROFILE SHOWN
IN FIGURE A-50
- 307 -
45
-------
APPENDIX B
AERODYNAMIC EFFECTS ON VELOCITY
MEASUREMENTS WITH AN S-TUBE
IN EPA METHODS 2 AND 5
. by
D. G. DeCoursin
H. A. Hanson
R. J. Davini
I INTRODUCTION
II AERODYNAMIC BASIS
III TURBULENCE
IV PITCH AND YAW
A. S-Type Pitot Tubes Tested
B. Test Setup
C. Forward and Backward Calibrations
D. Blockage Effects
E. Effect of Pitch and Yaw
V AERODYNAMIC INTERFERENCE
A. Test Setup
B. Results
VI CONCLUSIONS
- 308 -
-------
INTRODUCTION
EPA Methods 2 and 5 specify the S-tube as the velocity
measuring device. In Method 2, the S-tube is used for measure-
ment of the velocity distribution and volumetric flow in a duct
or stack. In Method 5, the S-tube is attached to a particulate
sampling probe and is used to establish and maintain isokinetic
sampling conditions.
The S-tube is calibrated (in accordance with Method 2) by
comparing its pressure differential with that of a standard
pitot tube when both are separately placed in the same gas
stream. The S-tube calibration factor is dependent upon Reynolds
number and, therefore, the .calibration is performed over the
range of velocities to be measured.
Examination of the above methodology raises the following
considerations:
1, Unlike the pitot tube (see References B-2 and B-3,
Reference B-2 lists 130 papers dealing with pitot tube perfor-
mance characteristics) , the S^-tube has not been thoroughly
studied; nor have its dimensions been standardized to give a
specified calibration factor. Thus, the factors that signifi-
cantly affect the velocity measurement accuracy of the S-tube
have not been established,
2. The flow in most emission measurement situations is
highly turbulent. Turbulence is known to affect velocity mea-
surements with pitot tubes; indicated velocities four percent
higher than the true values have been reported. Similar effects
on S-tube measurements can be expected,
3. Emission measurements are often, by necessity, made at
locations where the flow is not uniformly parallel to the duct
walls. Thus, the flow impinges on the S-tube at an angle (which
fluctuates with time because of turbulence, but here the mean
flow is considered). The effect of pitch and yaw has been re-
ported in Reference B-4 for one S-tube configuration. The velo-
city error ranged from 4% to 12% for a 20* flow angle, depending
on orientation in pitch arid yaw.
4. The calibration factor of S-tubes (and pitot tubes) is
influenced by the size and shape of the support strut. If, in
Method 2, the S-tube that was calibrated is mounted in a larger
strut (for additional structural support or extension of length),
an error is introduced.
5. In Method 5 the S-tube is attached to the sample probe
and is smaller than the sample probe. In this arrangement the
S-tube is in the velocity field created by the sample probe and
the support strut, and is in a region where the velocity is
- 309 -
-------
higher than the approaching stream velocity. Under these con-
ditions of aerodynamic interference even an exact measurement
of velocity would yield a value larger than the desired stream
velocity,
This last mentioned source of inaccuracy is not due to a
deficiency of the S-tube. It simply means the S-tube is placed
in the flow field of a nearby object and is exposed to the local
velocity at its particular location. Aerodynamic interference
of this kind would happen if the S-tube were placed near any
object, such as a structural beam within the duct. Unless the
S-tube is in the wake behind the object, it will "see" a velocity
higher than the freestream velocity.
At the present time there is no uniformity in the way the
manufacturers and users combine the S-tube and the particulate
sampling probe. Examples of commercial probes are shown in
Figures B-l through B-3, There are variations in the size of
the S-tube relative to the sample probe, the relative positions
of the two probes, the presence and location of a thermocouple,
and the size and shape of the support strut. If aerodynamic
interference effects are significant, the magnitude would prob-
ably be different for each of these configurations.
The user is interested in the measurement accuracy of the
system, which consists of the S-tube, pressure-connecting tubing,
pressure differential transducer, and transducer output measure-
ment device (the latter two items may be one instrument such as
an inclined manometer or bourdon gage), Further, the question
of system accuracy must be considered in relation to the desired
accuracy.
This report discusses only the behavior of the S-tube and,
specifically, presents information and data on the considerations
outlined above.
- 310 -
-------
II AERODYNAMIC BASIS
The measurement of low velocities with a pitot tube or S-
tube involves application of the Bernoulli equation for incom-
pressible flow, and consideration of the dynamic pressure de-
noted by q.
where
P = static pressure
f = density
u = velocity
P = stagnation pressure.
The stagnation pressure is constant along streamlines and
is the pressure at the stagnation point on the front of a body
facing the stream. Thus, it is easily measured with a pressure
tap placed at the stagnation point.
Direct application of the above equation also requires
measurement of the static pressure, which is the freestream
value, i.e, the value in the undisturbed flow. Accurate measure-
ment of the freestream static pressure is not as easy as measuring
the stagnation pressure. With a pitot-static tube, this is done
by finding a location along the tube where the pressure equals
the freestream value and placing pressure taps at this location.
The pressure distribution along the tube is fixed by two opposing
effects. The head of the instrument is attached to a stem at
right angles. The flow approaching the stem decelerates and so
increases the pressure at the static orifices. Simultaneously,
the flow accelerates around the front of the head (nose) , thus
dropping the pressure below the freestream value. The taps are
located where these two effects offset one another.
Based on these considerations, several standard pitot tube
configurations have been developed (see, for example, References
B-2 and B-3). Use of these configurations allows direct appli-
cation. (The earliest calibrations, before 1920, were done with
the whirling arm method. )
In comparison, the S-tube has the advantages of a more com-
pact shape and larger pressure taps (less susceptible to plugging) .
From an aerodynamic viewpoint, the major difference is the use of
a downstream facing tap that measures base pressure, rather than
- 311 -
-------
freestream static pressure. The base pressure is less than the
freestream static pressure and, therefore, the measured pressure
differential for a given velocity is larger than for a pitot-sta-
tic tube, which gives increased sensitivity at low velocities.
For application to the S-tube, the Bernoulli equation is modified
by introducing a coefficient, Cs.
P u
Pbase +J = P°
In accordance with Method 2, the coefficient is obtained by mea-
suring the pressure differential in a gas stream with both the
S-tube and a pitot tube of known coefficient, C
'std.
cs ~ cstd.
PO -'
static
p -
1/2
The value of the coefficient, C , , is equal to the number one
for any of the standard design s * pitot tubes.
Since the characteristic feature of an S-tube is that it
measures base pressure, some discussion of base pressure is
appropriate. Much theoretical and experimental work has been
done on base pressure because of its importance to the aerody-
namic drag of bodies with blunt bases, A valuable summary for
present purposes is given in Reference B-5.
The flow past a blunt base separates leaving a "dead water"
or separated flow region behind the base. The flow at the
boundaries of this separated region acts somewhat as a jet pump.
This action decreases the pressure in the separated region (equal
to the base pressure) to a value below the freestream static
pressure, (.In aerodynamics this low pressure constitutes an
undesirable drag force.) The effect is greater for two-dimen-
sional bodies (such as a long cylinder in crossflow - like the
strut of an S-tube) than for three-dimensional bodies (such as
a bullet, a wing tip, or the sensing head of an S-tube.)
Base pressure is dependent upon Reynolds number, which
covers the range 1000 to 100,000 for S-tubes and probes in
emission testing. The base pressure coefficient is defined as
= pbase " ^static
where the static and dynamic pressures are the freestream
(undisturbed) values. This coefficient is related to the S
tube coefficient through the equation
- 312 -
-------
cs - ^ 1 ^
""" T~t
I - !
The strut of an S-tube is a cylinder in crossflow. The base
pressure coefficient over the rear part of the strut (two-dimen-
sional regime far from the end) is about -1.2 (Reference B-5,
p. 3-3), This corresponds to a value Cs = 0.7.
Treating the head of the S-tube as a bullet-shaped object
(or cylinder in axial flow) gives a base pressure coefficient
of about -0.24 (Reference B-5 p. 3-19), This corresponds to
Cs = 0.9.
The S-tube coefficient should lie between these values and
closer to the 0.9, which it does, being typically 0.85, Note
also that the pressure at the base of the strut (away from the
end) is lower, by about one dynamic pressure head, than that
at the base pressure tap. These results suggest that the shape
of the head of the S-tube will effect its coefficient and its
response to yaw and pitch. In the absence of standardized con-
figurations, individual calibrations as specified by Method 2
are certainly appropriate.
- 313 -
-------
Ill TURBULENCE
Turbulence may affect S-tube and pitot tube measurements in
two ways. If the turbulence is of small scale, i.e., the eddy
sizes are smaller than the probe dimensions, the effect of the
velocity fluctuations is to increase the pressure sensed by
pressure taps. This effect is not large, probably rarely exceed-
ing 2.5% error in velocity (Reference B-3, p. 46).
If the turbulence scale is large, i.e., the eddy sizes are
much larger than the probe dimensions, the effect of the eddies
is to vary the angle with which the flow impinges on the probe
(Reference B-3 and B-6). In this case, the response of the S-
tube or pitot tube to pitch and yaw determines the effect of
turbulence. This is the case that exists in large size ducting
that is of interest here. Reference B-7 p, 75 cites German work
of 1921 where pitot-static tube velocity measurement errors of
4% were attributed to this effect.
Very little work has been done on the effect of turbulence,
especially large scale turbulence, and we do not know of any such
measurements with S-tubes. However, the yaw and pitch sensitivi-
ty is at least as great as for a pitot tube, therefore, errors
of similar size can be expected.
Another factor to consider is that the S-tube will be
calibrated under laboratory conditions where the intensity and
scale of turbulence will almost certainly be much lower than in
the field applications. Thus, the error is not eliminated by
the calibration process.
- 314 -
-------
IV PITCH AND YAW
The effect of pitch and yaw (defined in Figure B-4) on an
S-tube coefficient has been reported in Reference B-4. Additional
measurements with two S-tubes (.95 cm and .48 cm) have been made
under the present contract. The results are generally in agree-
ment.
A. S-Type Pitot Tubes Tested
Two different S-type pitot tubes were tested, The larger
(.95 cm diameter tubing) was a probe purchased from Research
Appliance Company, while the smaller probe (,48 cm diameter
tubing) was fabricated in our shop. Figure B-5 shows comparative
photographs of the two probes. It is obvious that the smaller
probe is not just a scaled-down version of the bigger probe.
There are three significant differences; (1) the large probe
tips extend proportionately farther beyond the probe stem, (2)
the large probe has proportionately thinner tube walls, and (3)
the larger probe is welded at three locations along the stem
length whereas the smaller probe is silver-soldered along the
entire length.
B. Test Setup
The probes were calibrated using an existing flow channel
(Channel 8) at the FluiDyne Medicine Lake Laboratory and the EPA
Model ductwork at the FluiDyne Energy Laboratory,
The Channel 8 setup, shown in Figure B-6, is supplied with
500 psi air through an ASME flow nozzle which provides an accur-
ate measurement of air mass flow rate. Different size nozzles
are installed to accommodate a wide range of flow rates. The
flow nozzle was operated with choked flow for the S-tube cali-
bration.
Channel 8 has a 20.3 cm (8 in.) diameter stilling chamber
about 1.82 meters (6 feet) long. The normal test section was
removed and replaced with a 20,3 cm (8 in,) diameter extension.
The S-tube calibrations were made inside of (closed jet) and
just downstream of (open jet) this extension.
As noted above, the 20.3 cm (8 in.) pipe normally forms the
stilling chamber for smaller diameter nozzles which are attached
to its exit flange. The 20,3 cm (8 inch) pipe is about 9 dia-
meters long and, therefore, has a relatively thick boundary
layer. This does not significantly affect the flow uniformity
- 315 -
-------
in the smaller diameter nozzles (velocity variations decrease with
the square of the nozzle area ratio). But it does not provide a
sufficiently uniform distribution of velocity across the 20.3 cm
(8 in.) diameter.
The velocity uniformity was increased by placing two screens
at the connecting flanges, as shown in Figure B-6. A typical
velocity distribution at the calibration station is shown in
Figure B-7, These measurements were made using a standard pitot
tube (United Sensor Model PAE-24-M-W) . The probe was introduced
into the duct such that the tip would be at the S-tube probe
insertion station, The flow was surveyed across two diameters,
90° apart, at radii corresponding to equal areas. Of particular
interest was the flow quality within a radius of 5.1 cm (2 in.)
from the duct centerline, since the S^tubes were kept within this
region for all calibration -data.
Surveys were made at seven different velocities and the
results are shown in Table B-l ,V"C is the velocity calculated by
a mass balance between the upstream flow metering nozzle and the
20.3 cm (8 in.) diameter duct downstream (assuming plug flow with
no boundary layer). vAvr is the average measured velocity of
18 points within 5,1 cm {2 in) of the duct centerline and Vc is
the average of two velocity measurements made on the duct ^ center-
line. The percentage difference between the meter derived velo-
city and the other two is also noted. It can be seen that there
is less than one percent average difference between the calcula-
ted and average measured velocity values with no apparent de-
pendence on velocity level. The difference between the calcula-
ted and measured centerline velocity values is somewhat larger
and biased toward a higher calculated value. It was concluded
that the upstream flow meter data could be used to calculate the
average duct velocity with an accuracy of + 1% within the area
of radius 5.1 cm (2 in.) from the duct centerline. This pro-
cedure was henceforth used in the calibration of the two S-type
pitot tubes.
Details of the calculations used to evaluate the velocity at
the test station from the upstream flow metering nozzle are given
in Appendix B-l.
The laboratory model of large power plant ductwork at the
FluiDyne Energy Conversion Laboratory was also used for the probe
calibration and interference testing. The probes were mounted
horizontally near the center in a section of the model ductwork
that is 36.6 cm (14.4 in) wide and 83,9 cm (33 in) high *- see
Figure B-8. The calibration level was established by measuring
the velocity at precisely the same point with a standard Prandtl
type pitot-static tube (United Sensors Model PAE-24-M-W, co-
efficient = 1.00).
- 316 -
-------
The differential pressures from the two probes were sensed
by a 1" H20 pressure transducer (Validyne Model DP45). The output
of this transducer was displayed on a Honeywell Model 17 Pen
Recorder- Barometric pressure was recorded from a mercurial
barometer, and duct air temperature was measured using a mercury
thermometer inserted in the flow upstream of the test section.
Duct pressure was measured from a static port in the duct wall
connected to a U-tube water manometer. The experimental accuracy
of the S-tube coefficient was estimated at + 1/2%.
Velocity surveys were taken to confirm a uniform velocity
field across the plane of the test section and for a reasonable
axial length, say, one duct diameter. The results of these
surveys are shown in Figure B-9, Because the test point was
downstream of a contracting elbow (contraction ratio is 6.5:1),
the flow field was reasonably flat, even without putting screens
in the duct upstream. In fact, the addition of screens did not
improve the flow field and at some points caused some deteriora-
tion in the profiles' flatness. Therefore, the test point was
established in the middle of the flat portion of the flow field
without upstream screens,
C. Forward and Backward Calibrations
The two S-tubes were calibrated in the forward (yaw angle,
0=0) and the backward (6 = 180Q) orientations over the velocity
range, (15 to 60 fps) 4,5 - 18,3 m/sec. Both probes met the EPA
Method 2 requirement that the forward and backward coefficients
must agree within 0.01.
D. Blockage Effects
The insertion of a probe into an air stream changes the
velocity of the stream at the probe from the value that would
exist without the probe being present. This effect is encoun-
tered in wind tunnel practice and much analysis has been done to
determine the magnitude of the corrections that must be applied
to the raw data. The effect can be made negligibly small by
keeping the probe (or model) size much smaller than the duct
cross-section. The effect is greater for blunt bodies that
produce large wakes than for smooth streamline bodies. The
presentation in Reference B-8 has been used to estimate the effect
as it applies to the calibration of S-tubes in either closed or
open jets.
The stream lines must flow around the probe, but this out-
ward "bulging" is constrained at solid walls. Thus, the
velocity in the vicinity of the probe is increased. The effect
is less than would be predicted by the one-dimensional incom-
pressible continuity equation.
- 317 -
-------
U1A1 = U2A2
in which case, a probe with a frontal area equal to one percent
of the test section area would increase the velocity by one
percent.
Both the body itself and its wake contribute to the effect;
thus, the analytical calculations treat the total effect in two
parts: solid blockage and wake blockage.
A model in an open jet produces an opposite effect, causing
the velocity to decrease. This results from the constant pres-
sure condition at the jet boundaries which allows the stream
lines to "bulge" outward further than they would otherwise. The
velocity changes are less in an open jet than in a closed jet.
The wake blockage is usually negligible, and the solid blockage
is about one-fourth the closed jet value (and of opposite sign.)
The appropriate equation for the solid blockage effect in a
closed jet is (Reference B-8):
fsb
0.9 wing volume
C1'5
where
= velocity change
V = undisturbed velocity
C = test section area (duct cross-section) .
The probe is treated as a wing model that is mounted on the duct
wall .
The wake blockage effect is (Reference B-8) :
AV \ _ S C,
wb
4C
where:
S = model area upon which C, is based
C,= drag coefficient.
Applying these equations to the .95 cm (3/8 in.) probe
installed in the 36.6 cm x 83.9 cm duct gives the following
results :
- 318 -
-------
3
Probe Volume 3/8 (3/4) (4) « 18.44 cm (1.125 in)
C = 0.307 m2 (462.0 in2)
S = 3/8 (4) - 9.68 cm2 (1.5 in2)
Cd = 1,2
A_V
V
closed jet
AV
V
sb
AV
V
wb
= O.Q001 + 0.0009
= O.QOl
For comparison, the simple one*-dimensional continuity
aives
approach gives
probe frontal area
duct area
= 0.003
for which
AV
V
= 0.003
max
Thus, the blockaqe analysis yields a velocity increase of 33%
the area ratio value-
With the same probe in a 36.6 cm x 83.9 cm open jet,
the wake blockage is negligible and the solid blockage is approx-
imately one-fourth the closed jet value and of opposite sign,
A V
V
open jet ^ 2.5x10
-4
which is negligible.
- 319 -
-------
E. Effect of Pitch and Yaw
Each probe was mounted in a special holder which allowed a
range of angles of pitch (oc) or yaw (0) to be obtained during a
data run. The probe entry into the test duct was sealed to the
probe with wax and a rubber gland, which allowed flexibility.
Reference marks on the probes were used to maintain the probe
head position on the duct centerline, regardless of pitch angle.
The data are presented in Figures B-10 through B-13. The
reference curves in each figure are taken from the data of Grove
and Smith (Reference B-4). Their measurements were made in a
30.5 cm (12 in.) diameter duct at approximately 6.1 m/sec. (20fps)
Error Due To Yaw
The effect of yaw is symmetrical, as would be expected. As
illustrated in Figure B-ll, the yaw angle sensitivity for the
.95 cm (.3/8 in.) probe showed very little dependence on the
velocity level, especially for yaw angles below 40°, The smaller
probe, which exhibited similar yaw characteristics in the (20 ft/
sec) 6.0 m/sec velocity range, produced much larger indicated
velocities at the (50 ft/sec) 15,2 m/sec velocity level. Indi-
cated velocities up to 13% larger than the true velocity were
measured with a yaw angle of 40°, In the upper velocity ranges,
the indicated velocity appears to depend upon detailed probe
geometry. It is most likely that the upstream facing taps of the
two probes respond the same to yaw, and that the difference is
due to the smaller probe having lower downstream tap pressures
(base pressures) than the larger probe at the higher velocity
levels. The effect of S-tube head and strut dimensions on mea-
sured base pressure is discussed briefly in Section II.
The test results for both probes show a maximum indicated
velocity at a yaw angle of 40° rather than the 20° indicated by
Grove and Smith data. These differences are thought to be pri-
marily attributable to differences in the probe geometry.
Error Due to Pitch
All of the pitch data (Figures B-12 and B-13) was taken in
Channel 8 (closed jet) and shows reasonable agreement with
Reference B-4. Pitch produces larger errors than yaw and is
asymmetrical, as would be expected. There is no significant
effect of velocity over the range tested -- 3 to 10 meters/
sec.
- 320 -
-------
The largest error occurs at negative pitch, corresponding
to probe being bent in the downstream direction, and this is the
only orientation that causes a negative error. Ten degrees gives
an error of 3% to 6% and 20 degrees gives an error of 12% to 15%.
A positive pitch angle occurs when a horizontal probe droops
at the tip. This causes the probe to indicate an increased velo-
city, corresponding to a drop in the pressure at the downstream
lacing tap. The errors are less than with negative pitch angles.
Ten degrees gives an error of 1% to 4%, and 20 degrees gives an
error of 4% to 7%.
- 321 -
-------
V AERODYNAMIC INTERFERENCE
As discussed in the Introduction, the problem of aerodynamic
interference arises in application of Method 5, where the S-tube
is attached to the sample probe. The sample probe and mounting
strut create a velocity field which interferes with the desired
measurement of the freestream velocity.
A, Test Setup
Aerodynamic interference tests were made using two commercial
probes supplied by EPA and shown in Figure B-l. The laboratory
model of large power plant ductwork (Figure B-8) was also used
for these tests. The probes were mounted horizontally near the
center of the 36.6 cm (14 in.) wide by 83,9 cm (33 in.) high test
section.
The insertion of a probe into an air stream changes the
velocity of the stream at the probe from the value that would
exist without the probe being present. This blockage effect
is discussed in Section IV-D and it is shown that the change in
velocity is negligible for the cases discussed.
Provisions were made to position the isokinetic source
sampling probe assemblies from one side of the duct and a Prandtl
type pitot-static tube (United Sensors Model PAE-24-M-W, co-
efficient = 1.00) from the other side of the duct. The top
surface of the duct over the test point was fabricated from
plexiglass in order to visually confirm the position and align-
ment of all test probes. The duct pressure was essentially
ambient under all flow conditions so this lid could be removed
while running, for manually positioning the probes and checking
alignment with a spirit level. Photos of the test setup are
shown in Figure B-14. These show a typical probe assembly and
Prandtl pitot-static tube inserted into the test duct, but both
somewhat short of the test point. Both probes are shown in the
photographs, but only one probe was in the stream during the
tests. First, the Prandtl tube was inserted and the duct velocity
was established. The Prandtl tube was then removed from the stream
and the S-tube/probe assembly was inserted to the same location.
The S-tube pressure differential was measured and was correlated
with the duct velocity (corrected for blockage as noted above).
The differential pressures from the two probes were sensed
by a 2.54 cm (1 in.) H20 pressure transducer (Validyne Model
- 322 -
-------
DP45), The output of this transducer was displayed on a Honey-
well Model 17 Pen Recorder. Additional test instrumentation is
described in Section IV*-R, Velocity profiles at the test section
are also presented.
A table and sketch showing the spatial relationships of the
elements comprising the two isokinetic source-sampling test
probes is shown in Figure B^-15, Entries are made for the dif-
ferent spacers and sample probe tips used in the test program.
The longer EPA-furnished probe (see Figure B-l) was not
tested extensively. The S-tube of this probe was welded to the
sample probe and, therefore, could not be removed to provide a
baseline calibration. In addition, inspection of the probe re-
vealed a small nick in the.very tip of one leg of the S-tube,
and the width of the probe tip as fabricated (dimension o in Figure
B-15) was more than generally used in practice. For these rea-
sons, it was felt that the shorter adjustable probe would provide
more accurate and representative data.
The short EPA probe, shown in Figure B-l, with its more
flexible hose clamp secured arrangement, allowed measurements to
be made as a function of sample probe tip size and location.
As illustrated in Figure B-l, two probe tip sizes and five
different spacers were provided. Similarly, as shown in Figure
16, the S-tube can be moved inward or outward such the orifices
extend past the entrance orifice of the sample probe (outboard)
or vice versa (inboard),
B, Results
The S-tube from the short EiPA probe was calibrated while
removed from the entire probe assembly. The results, shown in
Figure B-17, are in close agreement with coefficients cited in
the literature. The coefficient is plotted against duct velo-
city as measured by the Prandtl type pitot-static tube. A
straight line was drawn through the calibration data for the
forward (Yaw angle - 0°) orientation of the S-tube, and this
line was used as a baseline on subsequent graphs. For comparison
purposes, coefficients were also measured for the backward (yaw
angle = 180°) orientation, EPA Method 2 specifically stipulates
that the S-tube be used only if the two coefficients differ by
no more than 0.01. The tube meets this criterion.
Once the coefficient for the S-tube alone was established,
the isokinetic source-sampling probe was re-assembled and the
S-tube pressure differentials were measured as a function of
sample probe tip size, spacing between probe tip and S-tube, and
duct velocity. Effective S-tube coefficients were calculated
- 323 -
-------
from these measured pressure differences and the pitot tube pres-
sure difference used to establish the flow velocity, The effect-
ive coefficients are plotted against duct velocity in Figure B-18
for five different S-tube/sample probe tip spacings. Pertinent
dimensions are presented in Figure B-15, It should be noted that
initially, measurements were made with and without the probe tips
aspiring isokinetically, However, there was no measureable
change in the coefficient due to this factor. All of the data
was taken at zero angle of pitch and yaw.
The effective S-tube coefficients (Figure B-18) relate the
S-tube pressure difference to the duct velocity and apply
exactly, only to the probe/S-tube assembly. The flow accelerates
in passing around the probe, and, therefore, the velocity at the
S-tube is higher than the duct velocity. The effective coeffi-
cient is less than the true value which corresponds to a bare S-
tube (not near another object), The difference between the two
coefficients is a measure of the velocity error introduced by
calibrating the S-tube as a bare tube and then using is strapped
to a sample probe.
The data in Figures B-18 and B-19 generally show an increase
in velocity error as the S-tube and probe are moved closer to-
gether. Also, with the probe tip removed, the presence of the
sample tube and fitting introduce errors up to 3%.
A practical need is to determine if there is an S*-tube/
sample probe tip spacing that will reduce the error to a negli-
gible value over the velocity range of interest. The data are
plotted against probe separation in Figure B-19. It is clear
from this figure that the complicated flow field, and possible
Reynolds number effects, cannot be described simply by a sepa-
ration distance. The trend of smaller error at larger separation
is clear, but the errors do not reduce rapidly. The effects of
boundary layer thickness and wake formation (boundary layer
separation) will be a complex function of velocity, probe tip
size and shape, S-tube size and shape, and the proximity of the
sample probe tip. These effects have not been fully examined.
As illustrated in Figures B-2 and B-3, many different configur-
ations are available commercially. It appears from this data,
that convenient separation distances, 1.27 cm (1/2 in.) for
example, will introduce velocity errors up to 3 or 4%. However,
it must be noted that these results are based only on data
taken at zero angle of pitch and yaw. Pitch and yaw of a probe/
S/tube assembly will affect the velocity field around the probe,
thus changing the velocity and flow direction at the S-tube.
These effects were not measured.
Another configuration parameter to be considered is S-tube/
sample probe tip offset, as shown in Figure B-16, Reference
- 324 -
-------
to Figures B-2 and B-3 shows that some commercial probes have
this offset arrangement.
Tests were conducted with the S-tube orifice extending
2.54 cm (1 in,} beyond the sample probe tip (outboard), with zero
offset, and with the sample probe tip extending 2.54lcm (.1 in.)
beyond the S-tube orifice (inboard), As shown in Figure B-20,
the indicate'd velocity is as much as 8% above the freestream
velocity.
The largest effect occurs when the S-tube is inboard of the
sample probe. This is also the arrangement of three of the
probes shown in Figures B-2 and B-3.
- 325 -
-------
VI CONCLUSIONS
As stated in the Introduction, there is very little informa-
tion on the behavior of S-tubes, The information and data pre-
sented here are incomplete, but do lead to the following
conclusions.
1, Large scale turbulence, as very likely exists in ducts
and stacks where emission tests are made, will probably
cause the indicated S^tube velocity to be high by 4 or
5 percent, based on very limited data with pitot-static
tubes.
2, The effects of pitch and yaw produce velocity measure-
ment errors of several percent. At 20° yaw, the indi-
cated velocity will be from 3 to 7% high. Still larger
indicated velocities, up to 13% larger than the true
velocity, were measured with a yaw angle of 40°. At
20° pitch, the error may be +7 or -12%, depending on
the direction of pitch.
3, Fastening the S-tube to a sample probe, as in Method 5,
places the S-tube in the velocity field of the probe
and strut. The S-tube then measures the local velocity
which is higher than the approach stream velocity.
With the S-tube and sample probe arranged side-by-side,
and with zero angles of pitch and yaw, the velocity
error will be up to 4% for spacings up to 1.27 cm (1/2
in,) and does not decrease rapidly as the spacing in-
creases. (The effect of pitch and yaw on the S-tube/
single probe assembly was not measured.)
4. Placing the S-tube inboard or outboard of the sample
probe affects the magnitude of the error described
under Conclusion 3. An inboard displacement of one inch
(2,5 cm) causes the S-tube to read a velocity up to 7%
above the approach stream velocity. This arrangement
is used in some commercial probes.
- 326 -
-------
VII REFERENCES
B-l EPA Standards of Performance for New Stationary Sources,
Federal Register, Volume 36, No. 247, December 23, 1971,
24376-24895.
B-2 Folsom, R. G., "Review of the Pitot Tube," Transactions
ASME, October 1956, 1447-1460.
B--3 Ower, E. and Pankhurst, R. C., The Measurement of Air Flow,
4th Ed,, Pergamon Press, 1966.
B-4 Grove, D. J., and Smith, W. S., "Pitot Tube Errors Due to
Misalignment and Nonstreamlined Flow,", Stack Sampling News,
November 1973, 7-11.
B-5 Hoerner, S. F. , Fluids-Dynamic Drag, Published by the author,
1958.
B-6 Barat, M., "Pressure Measurements in Highly Turbulent Flows,
Paper in Heat and Mass Transfer in Flow with Separated
Regions, Pergamon Press, 1970.
B-7 Dean, R. C., Jr., Aerodynamic Measurements, MIT Press, 1953.
B-8 Pope, Alan and Harper, J» J,, Low^peed Wind Tunnel Testing,
Wiley & Sons, 1966,
- 327 -
-------
Short Probe Assembly with Extra Spacers and Probe Tip
Long Probe Assembly
FIGURE Bl.ISOKINETIC SOURCE-SAMPLING PROBES FURNISHED
BY EPA
- 328 -
-------
THERMOCOUPLE
^SAMPLE NOZZLE [TIP]
(a) Glass Innovations - 209 N Series
(b) Scientific Glass and Instruments
(c) Western Precipitation - (A 2040)
FIGURES-2.COMMERCIALLY AVAILABLE ISOKINETIC SOURCE-SAMPLING PROBES
- 329 -
-------
(a) Scott Environmental Technology - Model 100
(b) Lear Siegler - PM 100
(c)
Aerotherm- Sniffer
FIGURE B-3. COMMERCIALLY AVAILABLE ISOKINETIC SOURCE-SAMPLING PROBES
- 330 -
-------
Flow
Yaw Angle
Flow
Pitch Angle
FIGURE B---1 PITCH AND "AW PROBE MISALIGNMENT
- 331 -
-------
3/16" x .028" Wall 3/8" x .020" Wall
FIGURE B5.PHOTOGRAPHS OF S-TUBES TESTED
- 332 -
-------
STATION
RUN
PSI Air Supply
Flow Straightening Section
r— Additional Screens
^\ Probe Calibration Station
FIGURE B-6 MEDICINE LAKE PROBE CALIBRATION SETUP
TABLE B-l
RESULTS OF FLOW SURVEYS
v CALC .
m/sec
16
17
18
19
20
23
24
25
26
3
6
9
12
15
18
20
18
6
.18
.12
.21
.34
.38
.10
.99
.07
.13
AVG '
m/sec %
3
6
9
12
15
17
21
18
6
.18
.35
.17
.43
.44
.96
.01
.17
.06
0
-3
+ 0
-0
-0
+ 0
-0
-0
+ 1
.0
.7
.5
.8
.4
.8
.1
.6
.2
V _ LJ -1-.L X
m/sec %
3
6
9
12
15
15
20
18
5
.13
.31
.07
.23
.26
.60
.78
.05
.91
+ 1
-3
+ 1
+ 0
+ 0
+ 2
+ 1
+ 0
+ 3
.4
.0
.6
.9
.8
.8
.0
.1
.6
Average Difference
-.34
+ 1.0
- 333 -
-------
10.0
7.5
5.0
2.5
0
Radius
(cm)
2.5
5.0
7.5
in n
1 1 1 A1 0
A 0
A O
A o
A 0
A0
A 0
A O
A 0
A©
f\
Lj DucL a
0 A
0 A
Symbol Traverse ° A
o E to w 0A
0 A
^ Pop to Bot . 0 A
0 A
0 A
©A
II 1 ©
0 5 10 15 20
-
-
2
ft/se
I i i
036
X»— i
jr~
m/sec
FIGURE B-7 TYPICAL VELOCITY PROFILE AT PROBE
TEST STATION (CHANNEL 8)
- 334 -
-------
CO
i
D
O
1-4
- 335 -
-------
80
70 _
60 _
50 -
VELOCITY
40
FT/SEC
30 -,
20 -
10 -
West Side
of Duct ^
.Duct Position
East Side
"7 of Duct
M/SEC
FIGURE B-9 VELOCITY DISTRIBUTION WITH NO SCREENS
AT THE PROBE CALIBRATION TEST STATION
- 336 -
-------
w
CQ
D
VD
r-t
O
EH
M
!2
O
w
u
o
J
w
I
CQ
O
H
En
- 337 -
-------
G
0
•H
4-)
O
CU
CO
-P
in
cu
EH
CD
O
s
1
^^
u
CD
CO
-p
4-1
in
H
0
OJ
CO
e
r^
m
t
I
O
£
O
•H
•P
CJ
CU
CO
4J
CO
cu
EH
(U
O
s
1
'u
cu
CO
\^
-P
4-1
O
m
^-"
u
cu
CO
"v^
e
-------
^ -- ~~~
X
i
1
\
\
r7TT\
L_ 1L_^
_ \
l-J CJ ~~
\
r*i co \
o — \
£ \
^ \GD
'3 \
0 X 0-
r~\ o^P
0) o \
> r-i N
t \
'
_0
vo
"*"
_o
+
w
0)
(U
Cn
Q)
"CM Q
' W
8
0 0
rH CN
1 1
x T i
OH
X
"7 X xGQ
^^
\\
\
° (Q T
,0) ™
1 0) =tt=
Cri M-(
Q) Q)
Q K
8
— o
1
I 1 I \
0 0
3 P
Q Q
gg
CO O
• co
0 •
CM O
1 CN
— 1
U '-•»
Q) O
V) *
0 H
• t
CO CT\
1 1
O U
3 O
\
^
0
co
— t1
r"
Q>. O
"^^ ,-
Q^- ^
w
m
D
EH
1
en
vo
cH
co
W
O
IS
ffi
U
H
f\
P-I
EH
H
w
EH
H
U
O
g
-------
M
M
W
-P
•H
O
0)
y
V
0
- VD
O W
- ^ Q)
+ Q)
Cn
QJ
T!
3
O
-
0 0
rH CN
1 1
1 . . . . 1 ..
\ ' 1
\
GBXX
\
•s ^
1 \
i OQ1
-P
o 8
5 Q
n i ro
o o
1 P^
/
CO
CO
QJ
H
I
-H
8
CO
I—I
03
- 340 -
-------
FIGURE B-14 PHOTOGRAPHS OF PROBE INTERFERENCE TEST SETUP
- 341 -
-------
All Dimensions in Inches
I cm)
Short
Probe
Long
Probe
Probe
O.D.
.375
(.952)
.625
(1.59)
.375
(.952)
.625
(1.59)
i Tip
I.D.
.250
(.635)
.500
(1.27)
.250
(.635)
.500
(1.27)
Spacer
.250
(.635)
.313
(.794)
.375
(.952)
.500
(1.27)
.625
(1.59)
.250
t^.635)
.313
(.794)
.375
(.952)
.500
(1.27)
.625.
(1.59)
. 188
I A -J-J \
Vi477'-i
[fixedj
t
J
2.
(6.?
%
k
r5
39)
r
.875
(2.22)
.938
(2.381
1.00
(2.- 54)
1.13
(2.86)
1.25
(3.18)
.875
(2.22)
.938
(2.38)
1.00
(2.54)
1. 13
(2.86)
1.25
(3.18)
.75
.11.91}
.500
11.27)
.563
(1.43)
.625
(1.59)
.750
(1.91)
.875
(2.22)
.375
(.952)
.438
(1.11)
.500
(1.27)
.625
(1.59).
.750
(1.91).
.375
(.952)
.250
(.€35)
0
4
.93
(2.2
1.2
(3.1
8
81)
5
8)
[
E
t
C
\
)
i
FIGUREB-15ISOKINETIC SOURCE SAMPLING PROBE GEOMETRY
- 342 -
-------
ZERO OFFSET
c
L_rH_K.
S-TUBE INBOARD
^~\
l—U-VJ
S-TUBE
FIGURE B-16
SAMPLE PROBE TIP/S-TUBE ORIFICE
LONGITUDINAL OFFSET
- 343 -
-------
.9
.8
.7
¥
~4TT
_L
-8-
7
ft/sec
—-ri
m/sec
9 12
TRUE VELOCITY
18
-*• FLOW
Yaw Angle
Symbol
O
a
0 , degrees
0
180
FIGURE B-17 S-TUBE (ALONE) COEFFICIENT VS. TRUE VELOCITY
- 344 -
-------
•H
O
W
U
O
90
80
70
O
|
S-Tube alone
1.27 cm (1/2") sample
probe
__J ! |_
-P
u
4-1
W
O
90
80
70
S-Tube alone
^
.635 cm (1/4") sample
probe tip
i
90;-
-., '2> {"^
~> ' No sample
i i i i
70 10 20 30 40
,_ i i . 1 ..... i .
03 6 9 12
probe
1
50
1 _
15
,T<<3 rv>
iN s
tip
1 . .. ..
60 7C
1 ft/seip,
18
m/sec
Symbql
V* '
13
A
Oj
<•>
X, In. (cm)
.250 (.635)
.313 (.794)
.375 (.953)
.500 (1.27)
.625 (1.59)
Spacer
FIGURE B-18
S-TUBE COEFFICIENT VS. TRUE VELOCITY FOR FIVE
DIFFERENT S-TUBE/SAMPLE PROBE TIP SPACINGS
- 345 -
-------
J-l
Q)
•H
n
w
2.0
1.0
0.0
0
1.27 cm (1/2 Inch)
Sample Probe Tip
Spacer
0.2
0.4
0.6
0.8
1.0
2.0
Inc
-5s-
hes
cm
Spacer X
>, 5.0
-p
•rH
U
° 4"°
>
a 3.0
•H
o 2.0
M
W
OP 1-0
o.n
i I i I
_
^\^^ .635 cm (1/4 Inch)
l'} ^^~~-^^^ Sample Probe Tip
" ~~"""—— — -_
O ' 1
<-•) '^
;\ , .A
(v ^
'-.-/
Q /-;•
"v»<^_-'
~--LV_ , ,
-
—
-
-
— ~
1 1 1 1 ...
0.2
0.4
0.6
0.8 Inohes
1.0 2.0 cm
Spacer, X
FIGURE B-19 VELOCITY ERROR VS. SPACER THICKNESS
- 346 -
-------
8
Velocity = 6.1 m/sec (20 ft/sec)
A,._ /— 0.635cm Probe Tip
+4.0
r
c
•H
1-1 +2.0 - , r.
o
M
W 0.0
<*>
-&JT
1.21 cm Probe Tdp
o Sample Probe^Tip
f
_o o L ^— S-Tube alone
•
-^^
> +4.0 |-
'* +2.0 L
o
w °-°
*
-2.0^-
Velocity = 12.2 cm/sec (40 ft/sec)
i
0.635 cm Probe Tip~l
1.27 cm Probe Tip
No Sample Probe
i-Tube alone
Velocity =18.3 m/sec (60 ft/sec)
0.635 cm Probe Tip
1.27 cm Probe T3p
£y—
Sample Probe Tip'
^
w
0.0
Us-
-2.0
2.54 cm
(1 inch)
Inboard
Tube alone
-L
" 0
2.54 cm
(1 inch)
Outboard
S-Tube/Sample Probe Offset
FIGURE B-20 VELOCITY ERROR VS. SAMPLE PROBE TIP/S-TUBE
ORIFICE LONGITUDINAL OFFSET
- 347 -
-------
APPENDIX B-l
Calculation of velocity from upstream metering nozzle..
Metering nozzle mass flow:
_ D M
(Ib./sec)
where: K = .53117
P = meter pressure, PSIA
T = meter temperature, °R
A* = (7T/4) d 2
d = 0.504 inch
m
K is subject to real-gas correction amounting to + 1.0%
at P = 400 PSIA, T = 530°R.
M M
C is subject to a Reynolds number correction amounting to
+ 0.2% at PM = 400 PSIA, T., = 530°R.
M ' M
Referencing this mass flow to velocity at the 8 inch duct
test station and making corrections for duct temperature and
pressure
(ft/secl
t
duct duct
where; YaTo = density of air at STP = 0.0749 Ibm/ft
£\ JL t\
A = _ 1L- d 2 ft 2
cluct 4 x 144 QD ' IT1'
d =8 inches
- 348 -
-------
P = duct pressure (essentially ambient) f "Hg
T = duct temperature, °R
valuating the expression:
Op rn
,53117 x .99 x 144 ..504. M 29,92 1D
VCALC .0749 ( 8 J P_ X 530 X
VCALC « «2268 -
M
ote: T and T usually agree to within 4°F.
- 349 -
-------
TECHNICAL REPORT DATA
/Please read Instructions on the reverse before completing)
1 REPORT NO. 2.
EPA-600/2-76-170
4. TITLE ANDSUBTITLE
PARTICULATE SAMPLING STRATEGIES FOR
LARGE POWER PLANTS INCLUDING NONUNIFORM FLOW
7. AUTHOH(S)
H.A. Hanson, R.J. Davini, O.K. Morgan, A. A. Iversen
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Fluidyne Engineering Corporation
5900 OVson Memorial Highway
Minneapolis, Minnesota 55422
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
3. RECIPIENT'S ACCESSION-NO.
5. REPORT DATE
June 1976
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
1AA010
11. CONTRACT/GRANT NO.
68-02-1244
13. TYPE OF REPORT AND PERIOD COVERED
Final - 6/73-12/75
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report describes the results of a study to determine the effects that
various geometric ducting configurations have on the flow profiles and the dis-
tribution of particulate in ducting systems of large (>100 MW) power plants.
The program included both laboratory model studies and field testing at large
power plants. The measurement of total volumetric flow and particulate emissions
at less than full operating capacity was also investigated. The results of flow
angularity measurements in large stacks at typical sample port locations, includ-
ing downstream of induced draft fans, are similarly discussed. Special attention
was given to the aerodynamic effects of S-tube/sampling probe interference on
velocity measurements with an S-tube in EPA Stack Emissions Measurement Reference
Methods 2 and 5.
A computerized technique was used to determine the effectiveness of various
equal area sampling strategies in providing accurate measurements of three emis-
sion parameters: average particulate concentration, total volumetric flow rate,
and total emissions. Numerous typical and atypical velocity and particulate
concentration profiles were studied. Sampling strategy recommendations are
presented.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
*Air pollution Tests
*Electric power plants Field tests
*Ducts
^Sampling
*Profiles
*Particles
*Flow rate
13 DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
b.lDENTIFIERS/OPEN ENDED TERMS
19. SECURITY CLASS (This Report)
UNCLASSIFIED
20. SECURITY CLASS {This page)
UNCLASSIFIED
c. COSATI Field/Group
13B
10B
13K
14B
20D
14B
21. NO. OF PAGES
371
22. PRICE
EPA Form 2220-1 (9-73)
------- |