U.S. Environmental Protection Agency Industrial Environmental Research EPA~600/7~77-094fl
Office of Research and Development Laboratory
Research Triangle Park. North Carolina 27711 AUQUSl 1977
BURNER DESIGN CRITERIA FOR
NOX CONTROL FROM LOW-BTU
GAS COMBUSTION; VOLUME I.
AMBIENT FUEL TEMPERATURE
Interagency
Energy-Environment
Research and Development
Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S.
Environmental Protection Agency, have been grouped into seven series.
These seven 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 seven series
are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from
the effort funded under the 17-agehcy Federal Energy/Environment
Research and Development Program. These studies relate to EPA's
mission to protect the public health and welfare from adverse effects
of pollutants associated with energy systems. The goal of the Program
is to assure the rapid development of domestic energy supplies in an
environmentally—compatible manner by providing the necessary
environmental data and control technology. Investigations include
analyses of the transport of energy-related pollutants and their health
and ecological effects; assessments of, and development of, control
technologies for energy systems; and integrated assessments of a wide
range of energy-related environmental issues.
REVIEW NOTICE
This report has been reviewed by the participating Federal
Agencies, and approved for publication. Approval does not
signify that the contents necessarily reflect the views and
policies of the Government, nor does mention of trade names
or commercial products constitute endorsement or recommen-
dation for use.
This document is available to the public through the National Technical
Information Service, Springfield, Virginia 22161.
-------
EPA-600/7-77-094a
August 1977
BURNER DESIGN CRITERIA FOR NOX
CONTROL FROM LOW-BTU GAS
COMBUSTION; VOLUME I.
AMBIENT FUEL TEMPERATURE
by
Donald R. Shoffstal! .
Applied Combustion Research
Institute of Gas Technology
I IT Center, 3424 South State Street
Chicago, Illinois 60616
Contract No. 68-02-1360
Program Element No. EHE624a
EPA Project Officer: David G. Lachapelle
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, N.C. 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
-------
ABSTRACT
This research program was initiated to characterize problems associated with
retrofitting existing utility boilers with low- and medium-Btu gases manufactured
from commercially available coal conversion processes. All the experimental re-
sults were gathered from a pilot-scale furnace fired with a movable-vane boiler
burner at a heat input of 0.66 MW (2,250,000 Btu/h). The low- and medium-Btu
gases tested ranged in heating value from 3.7 to 11.2 MJ/m3 (100 to 300 Btu/SCF).
They were synthetically produced with a natural gas reformer system.
Data were collected to permit a comparison between natural gas and low-Btu
gases in the areas of flame stability, flame length, flame emissivity, furnace effi-
ciency, and NO emissions.
X
Flame stability was found to be very sensitive to fuel jet velocity. An injection
velocity of 30.5 m/s (100 ft/s) was found to be optimum. Flame length decreased
with increasing movable vane angle (swirl of the combustion air), and the low- and
medium-Btu gases tested were generally shorter than those of natural gas. Good
agreement was obtained between measured and calculated flame emissivities.
Some boiler modifications would be necessary to maintain rating when burning
gases of less than 7.5 MJ/m3 (200 Btu/SCF) heating value. NO emissions were
ordered by adiabatic flame temperature. The NO emission data yielded an activa-
tion energy of 153 k-cal/mole compared to kinetic model predictions of 135
k-cal/mole. The use of adiabatic flame temperature provided a good empirical
method of predicting NO emissions for the fuels tested.
11
-------
TABLE OF CONTENTS
Page
INTRODUCTION 1
SUMMARY 2
PROCESS GASES SELECTED FOR STUDY 5
DESCRIPTION OF FURNACE TEST FACILITY 9
NO and NO£ Instrumentation 18
CO, CH^, and CO£ Measurements 18
Oxygen Measurements 19
Chromatographic Measurements 19
EXPERIMENTAL PLAN 25
RESULTS 29
Flame Lengths 29
Gas Temperature Measurements 29
Flame Emissivity Determination 57
Nitric Oxide Emissions 77
DATA CORRELATION 88
NOx Emissions 88
Correlating Furnace Performance 88
Mathematical Model of Furnace Performance 98
In-the-Flame Analysis 104
111
-------
FIGURES
No. Page
1 Schematic Diagram of IGT Gas-Generating System 7
2 Gas-Generating System 8
3 Rectangular Test Furnace 10
4 Removable Sidewall Furnace Panels for Interior 11
Flame Probing
5 Overall System Schematic Diagram of Rectangular 12
Test Furnace System
6 Radiant Tube Preheater for Main Furnace 13
Combustion Air
7 Flue-Gas Cooler 15
8 Control Room Facility and Analytical 17
Instrumentation
9 Gas-Sampling Probe Head for Nonparticulate Flue Gases 20
10 Modified IFRF Temperature Probe 21
11 General Probe Holder 22
12 Pyroelectric Radiometer (Used for Emissivity 24
Measurements)
13 Guide-Vane Boiler Burner 26
14 Method of Measuring Movable-Vane Angle for 26
Boiler Burner
15 Natural Gas Flame Using a 0.5-Inch Axial Nozzle 30
16 Lurgi Oxygen Gas Flame Using a 2-Inch Axial Nozzle 30
17 Winkler Oxygen Gas Flame Using a 2-Inch Axial Nozzle 31
18 Koppers-Totzek Oxygen Gas Flame Using a 2-Inch 31
Axial Nozzle
19 Wellman-Galusha Air Gas Flame Using a 3-Inch Axial 32
Nozzle
20 Winkler Air Gas Flame Using a 3-Inch Axial Nozzle 32
IV
-------
FIGURES, Cont.
No. Page
21 Average Gas Temperature Profile for Natural Gas 35
With a 15-Degree Vane Rotation
22 Average Gas Temperature Profile for Natural Gas 36
With a 30-Degree Vane Rotation
23 Average Gas Temperature Profile for Natural Gas 37
With a 45-Degree Vane Rotation
24 Average Gas Temperature Profile for Natural Gas 38
With a 60-Degree Vane Rotation
25 Average Gas Temperature Profile for Lurgi Oxygen Gas 41
With a 15-Degree Vane Rotation
26 Average Gas Temperature Profile for Lurgi Oxygen Gas 42
With a 30-Degree Vane Rotation
27 Average Gas Temperature Profile for Lurgi Oxygen Gas 43
With a 45-Degree Vane Rotation
28 Average Gas Temperature Profile for Lurgi Oxygen Gas 44
With a 60-Degree Vane Rotation
29 Average Gas Temperature Profile for Winkler Oxygen Gas 45
With a 15-Degree Vane Rotation
30 Average Gas Temperature Profile for Winkler Oxygen Gas 46
With a 30-Degree Vane Rotation
31 Average Gas Temperature Profile for Winkler Oxygen Gas 47
With a 45-Degree Vane Rotation
32 Average Gas Temperature Profile for Winkler Oxygen Gas 48
With a 60-Degree Vane Rotation
33 Average Gas Temperature Profile for Koppers-Totzek 49
Oxygen Gas With a 15-Degree Vane Rotation
34 Average Gas Temperature Profile for Koppers-Totzek 50
Oxygen Gas With a 30-Degree Vane Rotation
35 Average Gas Temperature Profile for Koppers-Totzek 51
Oxygen Gas With a 45-Degree Vane Rotation
36 Average Gas Temperature Profile for Koppers-Totzek 52
Oxygen Gas With a 60-Degree Vane Rotation
-------
FIGURES, Cont.
No. Page
37 Average Gas Temperature Profile for Wellman-Galusha 53
Air Gas With a 15-Degree Vane Rotation
38 Average Gas Temperature Profile for Wellman-Galusha 54
Air Gas With a 30-Degree Vane Rotation
39 Average Gas Temperature Profile for Wellman-Galusha 55
Air Gas With a 45-Degree Vane Rotation
40 Average Gas Temperature Profile for Wellman- Galusha 56
Air Gas With a 60-Degree Vane Rotation
41 Average Gas Temperature Profile for Winkler Air Gas 58
With a 15-Degree Vane Rotation
42 Average Gas Temperature Profile for Winkler Air Gas 59
With a 30-Degree Vane Rotation
43 Average Gas Temperature Profile for Winkler Air Gas 60
With a 45-Degree Vane Rotation
44 Average Gas Temperature Profile for Winkler Air Gas 61
With a 60-Degree Vane Rotation
45 Refractory Emissivity Versus Refractory Temperature 65
46 Gas Emissivity Versus Axial Position for Natural Gas 66
47 Gas Emissivity Along Furnace Length for Lurgi Oxygen Gas 67
48 Gas Emissivity Along Furnace Length for Koppers-Totzek 68
Oxygen Gas
49 Gas Emissivity Along Furnace Length for Winkler Oxygen 69
Gas
50 Gas Emissivity Along Furnace Length for Wellman-Galusha 70
Air
51 Gas Emissivity Along Furnace Length for Winkler Air 71
52 Flame Flow Pattern Tested 73
53 Emissivity Profile for Coke-Oven Gas 77
VI
-------
FIGURES, Cont.
No. Page
54 NO Versus Secondary Air Preheat With Natural Gas 78
55 NO Versus Secondary Air Preheat With Lurgi Oxygen Gas 80
56 NO Versus Secondary Air Preheat With Koppers-Totzek 81
Oxygen Gas
57 NO Versus Secondary Air Preheat With Winkler 82
Oxygen Gas
58 NO Versus Secondary Air Preheat With Wellman-Galusha 84
Air Gas
59 NO Versus Secondary Air Preheat With Winkler 85
Air Gas
60 Arrhenius Plot of Experimental NO Emissions 87
61 NO Flue Concentration Versus Firing Density 89
62 Emissivity Versus Square Root of Path Length for Natural 95
Gas
63 Emissivity Versus Square Root of Path Length for Winkler 96
Air
64 Emissivity Versus Square Root of Path Length for Winkler 97
Oxygen
65 Comparison of IGT Experimental and B&W Calculated 101
Furnace Efficiencies
66 Comparison of Experimental and Well Stirred 103
Speckled-Wall Model Calculated Efficiencies
67 Variation of Well Stirred Approximation With 105
Adiabatic Radiance
68 Comparison of Experimental and Non-Well Stirred 106
Speckled-Wall Model Calculated Efficiencies
vn
-------
TABLES
No. Page
1 Synopsis of Furnace Operating Conditions and Test 3
Results for a 45-Degree Vane Angle
2 Species Concentrations, Adiabatic Flame Temperatures, and 6
Gross Heating Value (Wet) for Medium- and Low-Btu Gases
Tested
3 Flame Length as a Function of Fuel Type and Vane Rotation 33
4 Summary of Average Gas Temperature Data 39
5 Calculated and Measured Emissivities 75
6 Projected NOX Emission Levels for a Utility Boiler 90
7 Test Furnace Efficiencies and Combustion Product 90
Properties at the Flue
8 B&W Calculated Boiler Efficiencies Using IGT Data 92
9 Calculated and Measured Efficiencies and Gas Emissivities 93
10 Comparison of Measured and Calculated Furnace Efficiencies 107
11 Comparison of Calculated Emissivities 107
viu
-------
INTRODUCTION
The electric power industry is caught between the Federal Government's concern
for cleaning up the environment and for decreasing (or eliminating) our depen-
dence on foreign sources for oil. The environmental concerns make the use of coal
prohibitively expensive, if not impossible, because of the necessity of decreasing
sulfur emissions. Low-sulfur oil could replace coal, but this oil must be imported,
which is against long-range government policy. Coal appears to be the only fossil
fuel available in sufficient quantities to eliminate the necessity of importing oil;
however, an environmentally acceptable and practical way must be found for its
use.
One concept that could provide a practical method for using coal is to convert it
to low-sulfur, ashless, low-Btu gas for use in boilers. This would alleviate both the
utility industry's fuel supply problem and the air pollution problem. Of particular
concern, however, are the prospects of operating problems and loss of boiler out-
put (downrating), which can occur when retrofitting a unit originally designed to
use another fuel. What are needed are accurate combustion data to define the po-
tential magnitude of these problems and to indicate whether practical solutions
are possible.
This report presents results from an experimental program designed to character-
ize the problems associated with retrofitting existing boilers for low- and
medium-Btu gases made from commercially available coal conversion processes.
Data were collected to permit a comparison of natural gas and low- and medium-
Btu gases in the areas of :
• Flame stability
• Flame length
• Flame emissivity
• Flue-gas pollutant emissions
• Detailed temperature profiles, and
• Calculated changes in heat transfer in the radiant and
convective sections of a boiler.
-------
SUMMARY
All of the experimental results presented in this report were taken from a pilot-
scale furnace fired with a movable-vane boiler burner (MVBB) at a heat input of
2.25 million Btu/hr. In addition to natural gas, five medium- and low-Btu gases
were tested: Lurgi oxygen, Koppers-Totzek oxygen, Winkler oxygen, Wellman-
Galusha air, and Winkler air. A synopsis of the test results for a 45-degree vane
angle is presented in Table 1. This table includes furnace operating conditions,
burner operating conditions, fuel combustion properties, and experimental data for
the MVBB with a 45-degree vane angle. The information includes gross heating
value of the fuel; adiabatic flame temperature with 3% excess oxygen and
combustion air preheat of 325°F; input velocity of fuel and combustion air at
60°F; hydrogen-to-carbon monoxide ratio of fuel; gross input enthalpy, including
combustion air preheat; chemical species analysis of flue gases; volume of flue
gases at 60°F; measured flue-gas temperature; measured furnace rear-wall tem-
perature; calculated and measured emissivity; and concentration of NO measured
in the flue at 3% excess oxygen.
Based on a detailed analysis of the experimental data, the following conclusions
were made:
1. Flame stability was established for each test fuel under the most severe oper-
ating conditions, that is, with the furnace walls, burner block, and combustion
air at ambient temperature. The flame stability was very sensitive to fuel jet
velocity leaving the fuel nozzle. An injection velocity of 100 ft/s was found to
give the best operation.
2. For all fuel gases tested, the flame length decreased with increasing movable-
vane angle (swirl of the combustion air). Medium- and low-Btu fuel gas flames
were shorter than those for natural gas with the exception of Winkler air at a
45-degree vane angle, which produced a flame 10% longer than the natural gas
flame.
3. Babcock and Wilcox used the pilot-scale test furnace data as input to their
utility boiler design calculation to evaluate overall unit output when retrofit
with medium- and low-Btu gases. The calculation technique used divided the
boiler into a radiant (furnace) section and a convective section. The IGT ex-
perimental data were used as input conditions to the convective section. The
It is EPA policy to use metric units; however, in this report
English units are occasionally used for convenience. See attached
conversion table.
-------
Table 1. SYNOPSIS OF FURNACE OPERATING CONDITIONS AND TEST RESULTS
FOR A 45-DEGREE VANE ANGLE
(Part 1)
Fuel
Natural gas
Lurgi oxygen
Winkler oxygen
Koppers-Totzek oxygen
Wellman-Galusha air
Winkler air
Gross heating
value.
Btu/SCF
1035.0
285.4
270.2
285.0
159.4
116.3
Adiabatic
flame
temp, °F
3337
3164
3329
3578
2948
2579
H2/CO
ratio
_
2.2
1.3
0.7
0.5
0.6
Fuel input
vel, ft/s
100.6
100.0
105.6
97.6
136.4
109.6
Combustion
air inlet
vel, ft/s
53.3
51.2
52.3
50.4
54.6
74.7
Table 1. SYNPOSIS OF FURNACE OPERATING CONDITIONS AND TEST RESULTS
FOR A 45-DEGREE VANE ANGLE
(Part 2)
Fuel
Natural gas
Lurgi oxygen
Winkler oxygen
Koppers-Totzek oxygen
Wellman-Galusha air
Winkler air
Gross heat
input,
Btu/hr
2,371,265
2,356,066
2,350,940
2,348,367
2,358,036
2.358,940
Flue products
N2
C02 | H20
02
%
72.5
62.9
63.1
65.0
72.3
74.5
8.3
16.8
18.3
20.2
15.8
14.6
16.7
17.8
16.1
12.2
9.2
8.2
2.5
2.5
2.5
2.6
2.7
2.8
Vol of flue
products,
CFH at 60T
27,357
27,565
26,169
24,356
32,772
38.339
SCFair
SCF fuel
11.23
2.71
2.43
2.51
,1.54
1.13
SCF flue products
SCF fuel
12.26
3.41
3.06
3.07
2.33
1.96
Table 1. SYNPOSIS OF FURNACE OPERATING CONDITIONS AND TEST RESULTS
FOR A 45-DEGREE VANE ANGLE ,
(Part 3)
Fuel
Natural gas
Lurgi oxygen
Winkler oxygen
Koppers-Totzek oxygen
Wellman-Galusha air
Winkler air
Flue-gas
temp, °F
2553
2442
2523
2554
2434
2335
Rear-wall
temp, °F
2309
2300
2309
2327
2066
2012
Emissivity
Calc
0.159
0.194
0.189
0.174
0.154
0.150
Meas
0.177
0.185
0.218
0.190
0.190
0.170
Flue NO,
ppm
65
32
73
104
15
4
-------
design of the computer program did not allow information from the IGT tests to
be inputs for the radiant section, although a comparison between the IGT ex-
periment results and Babcock and Wilcox's predictions was made. The IGT data
were used as input conditions to the convective section. It was found that all
fuel gases tested were capable of maintaining boiler output; however, those
with a heating value below 200 Btu/SCF may require retubing in the convective
sections.
4. A second evaluation of furnace efficiency — defined for the purposes of this
report as the fraction of the total enthalpy input to the furnace that is trans-
ferred to the furnace load — was made by Babcock and Wilcox using their stan-
dard design calculation technique. The inputs required for this calculation
were: 1) fuel composition and 2) desired temperature of combustion products
leaving the air heater. The values calculated for the combustion products of the
fuel gases tested leaving the radiant section of the boiler agreed remarkably
well with experimental values.
5. An iterative calculation technique using a non-well stirred reactor model pro-
duced efficiencies for each test fuel in good agreement with the experimental
results.
6. The furnace efficiency (fraction of heat transferred within the furnace) cal-
culated with the well stirred model was found to be rather insensitive to emis-
sivity, with a 10% change in emissivity producing only a 3% change in
efficiency.
7. NO concentration levels from the medium- and low-Btu gases were ordered by
adiabatic flame temperature. Only test fuels with heating values below 200
Btu/SCF had emission levels that conformed to the New Source Performance
Standard.
8. The NO emission data yielded an activation energy of 153 kcal/mol compared
to kinetic models which predict 135 kcal/mol. This good agreement suggests
that peak temperatures approaching the adiabatic flame temperature are con-
trolling the rate of NO formation. Also, the use of the adiabatic flame tem-
perature provides a good empirical method of predicting NO emissions.
-------
PROCESS GASES SELECTED FOR STUDY
To ensure the immediate relevance of the program's results, only gas compositions
from commercially available coal conversion processes (Lurgi, Koppers-Totzek,
Winkler, and Wellman-Galusha) were considered. The gas properties of interest
were heating value and hydrogen-to-carbon monoxide ratio (r^/CO). To permit
the study of a complete range of heating values and hydrogen-to-carbon monoxide
ratios five gases were selected for testing. These five gases are listed below, and
a detailed gas composition analysis for each is given in Table 2.
Heating value,
Test gas Btu/SCF (wet) H7/CO
Lurgi oxygen 285 2.2
Winkler oxygen 269 1.3
Koppers-Totzek oxygen 284 0.6
Wellman-Galusha air 160 0.5
Winkler air 117 0.6
These representative medium- and low-Btu gases are produced using the gas-
generating and fuel-preparation facility shown schematically in Figure 1. The
critical item in the gas supply system is the reformer, shown in Figure 2. This is a
special gas generator that produces varying ratios of I^/CO. Natural gas and car-
bon dioxide are fed to the reformer through a specially designed mixing tee.
Additionally, steam may also be mixed, depending on the desired EU/CO ratio.
The resulting mixture is supplied to four reaction retorts contained in a vertical
cylindrical furnace. The catalyst-filled retorts are heated by the furnace, and the
input gases undergo endothermic chemical reactions at a temperature of 2100°F
as they pass through the retort tubes. Globally, the reaction scheme is the water-
gas shift. The gas generated within the reaction tubes passes through water-
jacketed coolers, where it is quenched to prevent deterioration of the rLj/CO
ratio.
The reformed gas is compressed to 30 psig and pushed through an MEA (methyl-
ethyl-amine)-CO2 (carbon dioxide) absorbing tower. This tower is used to remove
-------
from the reformed gas if its concentration is above what is needed to synthe-
size the desired low-Btu gas. The reformed gas then enters a mixing chamber
where it is blended with nitrogen, carbon dioxide, steam, and natural gas to reach
the desired mixture of gas to be modeled. The product gas is analyzed by a gas
chromatograph to ensure that the correct composition is attained. The
synthesized gas is then fed to the pilot-scale furnace for combustion testing.
Table 2. SPECIES CONCENTRATIONS. ADIABATIC FLAME TEMPERATURES.
AND GROSS HEATING VALUE (Wet) FOR MEDIUM- AND LOW-Btu GASES TESTED
Fuel
Lurgi oxygen
Wjnkler oxygen
Koppers-Totzek oxygen
Wellman-Galusha air
Winkler air
CO
18.5
32.9
52.1
26.3
21.1
H2
40.2
41.2
34.5
14.3
13.0
C02
29.4
20.0
9.2
7.4
6.9
CH4
9.4
3.0
0.5
2.6
0.6
N2
0.6
1.0
1.0
46.9
56.5
H2O
1.9
1.9
1.9
1.9
1.9
Adiabatic
flame
temp, °F
3156
3327
3578
2948
2579
Heating
value,
Btu/SCF
285
269
284
160
116
-------
STORAGE
EXISTING
FURNACE
STACK
LOW-Btu
NATURAL
GAS
DEMINERALIZER''
CONDENSER
CO,
J
FURNACE
FLUE
MEA SOLUTION
Figure 1. Schematic diagram of IGT gas-generating system
-------
ex
2b. Reformer controls and
synthetic gas mixing panel
2a. Natural gas reforming system
Figure 2. Gas-generating system
-------
DESCRIPTION OF FURNACE TEST FACILITY
The experimental work was conducted on a rectangular furnace with a 25-sq ft
cross-sectional area and a 13-foot length. This furnace can be end- or sidewall-
fired at a rate of 4 million Btu/hr. The furnace is equipped for in-the-flame sam-
pling, preheated air, and flue-gas recirculation. (See Figure 3.) This furnace is
capable of operating at temperatures up to 3000°F or as low as 1600°F at a con-
stant (maximum) gas input of 4 million Btu/hr and up to 40% excess air. Cooling
is achieved with cooling coils embedded in the refractory walls. The furnace is
constructed completely of 9-inch thick cast refractories, with removable panels in
one sidewall to permit insertion of sampling probes (Figure 4). The overall fur-
nace system is shown schematically in Figure 5. The system is flexible enough
that the following operating parameters can be independently varied:
• Heat input, up to 4 million Btu/hr (8.0 million for certain burners)
• Air input, up to 40% excess
• Heat losses to the furnace walls by changing flow in water-cooling
tubes cast into the refractories
• Combustion air temperature, up to 1000°F
• Flue-gas recirculation capability, up to 35% of combustion air
• Furnace pressure, up to +0.05 inch of water.
The combustion air for the main furnace can be preheated up to a temperature of
1000°F with a separately fired radiant tube air preheater. The radiant tube fur-
nace (Figure 6) consists of an insulated airtight steel chamber 4 feet high, 4 feet
wide, and 16 feet long. As the combustion air to be preheated passes through this
chamber, it is heated by convection from three 6-inch diameter "hairpin" gas-fired
radiant tubes.
The radiant tubes and refractory flow passages inside the preheater are arranged
to provide an S-shaped flow pattern, which maximizes residence time for heating
at the maximum allowable pressure drop (20 ounces) for which the flow pattern
will provide the necessary air flow of 75,000 SCF/hr.
-------
Figure 3. Rectangular test furnace
-------
Figure 4. Removable sidewall furnace panels
for interior flame probing
-------
SAFETY SHUTOFF-
HRCHEATER COMBUSTION ; I—(P|J)
AIR BLOWER-F3 ~ • '
-^-, METERING ORIFICE-O8
l «?
tS)
SECONDARY
COMBUSTION
BLOWER-F2
MAIN GAS
SMUTOFF - V
(TI8-T48)
WALL-COOLING TUBES,
| || Z — METERING ORIFICE-OI
1— '^.(PS)
MANUAL SHUT-
FLOW CONTROL OFF-V7V8
SAFETY
SHUTOFF-V2
EXPERIMENTAL FURNACE
SECOND-STAGE PRESSURE
REGULATORS V9.V6
JR FLOW CONTROL-VII
WALL-COOLING
SWITCHING VALVES
VI2.VI3
FILTERED
AIR INTAKE
WALL-COOLING
BLOWER-F3
£: METERING ORIFICE -05
WALL- COOLING WATER
r — -» --- ! IHEAT EXCHANGER
5nFLUE-GAS COOLER RIVER-WATER
'HEAT EXCHANGER FLOW CONTROL
VBLVES-VI4.VI5
FIRST-STAGE PRESSURE
REGULATORS V3,V4
y/RIVER WATER
WALL-COOLING
PUMP-P3
LEGEND
£ zrr::.,
ui swi'i i«u'C" M.>I
t^l -...-^-o..^
-H- -«.,.-,.
SS^ UTOt >.C*«I1>
»*OLLOW[D HW-Mtl
lu*l j
Figure 5. Overall system schematic diagram of rectangular test furnace
-------
~4ft
AIR
INLET-
(8 in.)
—*- ^^^^^^^^^K^j^^^^^^^^^K^S^^^^
_ l}***** v v v v- * ./v v ^ Jf^Jjt? s >V /v ^ v v f_f_JJTr *• yv- ^-JJ1 -/^r vyf ^ v >v -^ ^ v > v * * IJ J*J * f* rrm-s&A
INTERNAL INSULATION
STEEL SHELL
RADIANT
BURNER
FLUE
COLLECTOR
FLUE
COLLECTOR
Figure 6. Radiant tube preheater for main furnace combustion air
-------
The temperature of the air can be regulated by changing the heat input to the
radiant tubes. Ambient temperature air can be supplied by completely shutting
down the preheater or by directing the air through the preheater bypass pipe. The
bypass pipe was installed to allow working on the preheater without shutting down
the air supply to the main furnace. Air bypass is achieved by selective switching
of valves.
Flue products for recirculation back to the burner and main furnace are obtained
from the furnace itself. Flue products can be withdrawn from the furnace flue
passage just prior to the main furnace flue damper. Up to 14,000 SCF/hr of flue
products can be withdrawn from the flue, which provides a 30% recirculation fac-
tor when the furnace is fired at 3.5 million Btu/hr with Z0% excess air.
The main furnace flue products are actually pulled from the flue (Figure 5) by the
suction in the inlet side of the main furnace combustion air fan (F4). The flue
products enter the recirculation withdrawal and treatment system at about
2800°F through a short length of internally insulated steel duct. These hot gases
are cooled to about 1Z5°F in a packed-bed water cooler (Figure 7). Cooled city
water (about 70°F) is sprayed down on a bed of refractory packings as the hot
gases pass up through the packed bed. This cooling system lowers the water con-
tent of the flue gas from about 0.008 to about 0.007 Ib/CF, which is the dew point
of the gases at about 1Z5°F. (The lost water content can be added again later in
the system if experimental conditions require this treatment.) The cooled gases
then pass through a flow-control shutoff valve (V25 in Figure 5). This valve con-
trols the flue-gas flow rate, which regulates the percentage of recirculated prod-
ucts. This valve is interlocked to an outlet temperature sensor on the gas cooler.
If the outlet temperature of the gases exceeds 150°F, which would damage the
combustion air fan, the control valve (V25) shuts down. This stops the flow of flue '
gases. Beyond the flue-gas control valve, the flue gases are mixed with the re-
quired amount of air for combustion. A control valve (V24) regulates the amount
of air pulled in by the fan. The total amount of air for combustion and flue prod-
ucts is metered with an orifice plate (O7) at the outlet of the fan. The flue prod-
ucts and air then pass into the air preheater or preheater bypass pipe.
The water used in the flue-gas cooler is clean city water, which is continuously re-
cycled. A water flow of 60 gpm is supplied at 150 psig by a turbine pump to a
14
-------
COOLING WATER SYSTEM
60 gpm AT 150 psig
SPRAY MANIFOLD
AND HEADS
COOLED
FLUE
GASES
WIRE MESH LIQUID
DEMISTTER
PERFORATED
PLATE
2-in. WATER LEVEL
>.'.•?> /,TI //p /(^ /ii> /'ft
PACKED REFRACTORY
BED-80% VOIDS
HOT (2800 °F)
FLUE GASES
/
s
y
g
J
T
HOT-WATER
COLLECTION TANK
-« 4CI in to-
T
-16 in.
I
t
~20in.
1
TO DRAIN
Figure 7. Flue-gas cooler
15
-------
series of spray heads in the gas cooler. The hot (200°F) spent water flows out of
the cooler into an atmospheric holding tank. This tank is equipped with a
constant-level overflow to the building drain. In this way, any condensed water
from the combustion is removed and disposed of in the sewer. The water in the
holding tank is periodically treated with sodium hydroxide to prevent acid buildup
in the water due to condensing flue-gas components. One such component
removed by the flue-gas cooler is NO2. The hot water from the water holding
tank is cycled through an American Standard heat exchanger capable of removing
1.5 million Btu/hr of heat from the water. Cooling in the heat exchanger is pro-
vided by a flow of river water at a rate of about 150 gpm at 80 psig. The river
water is supplied by a river adjacent to the test facility through a service pump
(PI) maintained by IGT.
Figure 8 is an overall view of furnace controls and the analytical instrumentation
package. The equipment used for concentration measurements of chemical
species during this program is listed below; these analyzers included the following
items:
1. Beckman 742 Polarographic Oxygen (O,)
2. Beckman Paramagnetic Oxygen (O2)
3. Beckman NDIR Methane (CH4)
4. Beckman NDIR Carbon Monoxide (CO)
5. Beckman NDIR Carbon Dioxide (CO2)
6. Varian 1200 Flame lonization Chromatograph (Total HC and C2 to Cg)
7. Beckman NDIR Nitric Oxide (NO)
8. Beckman UV-NO2
9- Hewlett-Packard Thermoconductivity Chromatography, Hydrogen (H),
Nitrogen (N2), Argon (A2), CO, CO2, Cj to C5, Oxygen (O2)
10. Beckman Chemiluminescent NO-NO2
11. Data Integration System.
This instrumentation package allowed concentration measurements of the follow-
ing major components: 1) measurement of hydrocarbon compounds Cj to C^;
16
-------
Figure 8. Control room facility and analytical instrumentation
17
-------
2) independent check of NO-NO2 chemiluminescent with NDIR-NO and NDUV-
NO2; 3) independent check of paramagnetic O2, polarographic O2, NDIR-CH4,
NDIR-CO, and NDIR-CO^ with the respective chromatographic species concentra-
tion; and 4) measurement of hydrogen (H2), argon (A2), and nitrogen (N2).
The following sections give a general description of the measurement system used
for this program.
NO AND NO2 INSTRUMENTATION
The chemiluminescent NOX and NDUV-NO2 system was mounted in a roll-around
cabinet that could be placed out at the furnace. This was important in minimizing
sampling distances, which can affect accuracy. The chemiluminescent unit was
equipped with a carbon converter. Test work by IGT and others has demonstrated
that in a reducing environment the carbon converter maintains a better conversion
efficiency than converters made of stainless steel, quartz, or molybdenum.
The instrumentation was calibrated by using both a permeation tube with a con-
trolled known release of NOX and certified prepared cylinders of NO and NO2
gases.
The sample gas was drawn from the furnace through a special alumina probe by a
(R)
Dia-Pump Model 08-800-73 all stainless-steel and Teflon pump delivering
approximately 0.4 CF/min- (This sample delivery rate was dictated by the re-
quirements of the measuring instruments.) The sample is immediately passed
through a stainless-steel large-particle filter. Both the pump and filter were kept
above 100°C to prevent condensation of the water vapor inherent to combustion
products.
CO, CH4, AND C02 MEASUREMENTS
Nondispersive infrared analyzers were used for carbon monoxide, methane, and
carbon dioxide measurements. These analyzers do not affect the sample gas and
can be operated in series. They were calibrated by using certified gases with
known concentrations of the species being determined. The infrared analyzers
require a completely dry sample. Therefore, the sample was first passed through
a water trap and a 3 A molecular drying sieve. A small in-line filter was placed
immediately after the drying tube to trap particles of sieve that may be carried
over by the gas stream.
18
-------
OXYGEN MEASUREMENTS
A portion of the "conditioned" sample gas is diverted from the NDIR units to a
Beckman Model 600 paramagnetic analyzer. A second oxygen analyzer, a Beck-
man Model 742 polarographic, was used as a cross-check on the oxygen concentra-
tion. The Model 742 analyzer has an advantage over the paramagnetic in time
response.
CHROMATOGRAPmC MEASUREMENTS
As a detailed gas analysis was required, the sample was fed to a Hewlett-Packard
7620-A thermal conductivity chromatograph, which permitted concentration eval-
uations of hydrogen, nitrogen, argon, oxygen, carbon monoxide, carbon dioxide,
and hydrocarbons Cj to Ce« To achieve separation of these species, a helium
carrier gas was used in conjunction with a Porapak Q column. Three temperature
program rates were also required, ranging from —100° to 300°C. A sample loop
volume of 100 ml was used to ensure linearity in the hydrogen response for con-
centrations up to 60%.
For total hydrocarbon analysis, a Beckman hydrocarbon analyzer was used. A
detailed hydrocarbon analysis could be made using a Varian 1200 flame-ionization
chromatograph. All chromatographic readings were electronically integrated and
printed out as a function of resolution time.
In addition to flue-gas analysis, a major task of this program was to map profiles
of temperature, chemical species, and flame emissivity. Modified designs of the
International Flame Research Foundation (IFRF) were used to construct probes,
which enabled this type of data collection.
Figure 9 shows the assembly drawing of the gas-sampling probe used in both the
flame and the flue. To minimize NO^ reduction in the probe, an alumina tube was
inserted for the first 18 inches and was joined to Teflon tubing to carry the gas
sample to the analyzers.
Temperature data were collected using a suction pyrometer; the design is illus-
trated in Figure 10. A Pt-Pt Rh 10% thermocouple was used. The efficiency of
the pyrometer was measured at 96% with a 25-second response time.
The gas-sampling probe, suction pyrometer, and radiation cooling target were all
designed to be installed in the general probe holder shown in Figure 11.
19
-------
O
l-l/2-in. 304 SS TUBING
-l/2-in. 304 SS TUBING
l-3/4-in. 304 SS TUBING
-2-in. 304 SS TUBING
-2-1/2-in. 304 SS TUBING
3/4-in. 304 SS
TUBING
ALUMINA TUBING IN\
l/8-in. 304 SS TUBING
3/4-in. 304 SS TUBING
Figure 9. Gas-sampling probe head for nonparticulate flue gases
-------
WATER
GAS
EXTRACTION N
•GAS
•GAS
WATER
i
L
COOLING
JACKET
THERMOCOUPLE
HOT JUNCTION"
ALUMINA y SILLIMANITE J/ REFRACTORY
SHEATH SHIELDS CEMENT PLUG'
SUCTION TIP FOR MEASUREMENTS IN
NATURAL GAS AND OIL FLAMES
SUCTION TIP FOR MEASUREMENTS IN
PULVERIZED-COAL FLAMES
Figure 10. Modified IFRF temperature probe
-------
ro
— 0.75 in.-H
0.5 in. [-•- I
•5ft 6in.-
2.50 in. x II go
2.00 in. x 16 go
1.75 in. x II go
-O.75in. x 12 go
TUBING
5ft-
5 ft 3 in. -
\ .
/ V
/ \
\ .. , \
V
^
II ^£ — ""
I
\ ^
3/4 in. x 13 go
TUBING
Figure 11. General probe holder
-------
To evaluate radiation intensity, which is needed for a determination of flame
emissivity, a PR 200 Pyroelectric Radiometer, manufactured by Molectron Corp.
in Sunnyvale, California, was used. Figure 12 shows the pyroelectric radiometer
plus radiation shield. This radiometer uses a permanently poled lithium tantalate
detector that is capable of resolving radiant power into the nanowatt range while
maintaining a continuous spectral response from the vacuum UV to 500 Jim. A
built-in optical calibration system in the form of a highly stable LED (light-
emitting diode) that is calibrated against an NBS traceable standard of total irra-
diance permits a direct correlation of experimental data from different trials.
23
-------
Figure 12. Pyroelectric radiometer
(used for emissivity measurements)
24
-------
EXPERIMENTAL PLAN
All test data were collected using a single-register movable-vane boiler burner
with an axial fuel injector. An assembly drawing of the MVBB tested is shown in
Figure 13. The combustion air enters perpendicular to the axes of the burner and
passes through a register of guide vanes that impart a degree of spin to the air,
dependent on the vane orientation. Figure 14 illustrates how the angle of the
movable vane is measured. The ratio of the average tangential and radial velocity
components at the exit of the movable-vane register depends only on the geo-
metric dimensions of the vanes in the axis perpendicular cross section (assuming a
negligible Reynolds number influence).
A straight pipe was used as the fuel injector, guaranteeing that the fuel had only
an axial velocity. This type of injector was used for two reasons. First, this type
of nozzle would be the most stringent test for flame stability in the cold-wall
stability tests; and, second, medium- and low-Btu gases would normally be avail-
able at relatively low pressures, dictating the use of a low-pressure injector. The
diameter of the fuel injector was varied for fuel type to maintain an injection
velocity of 100 ft/s.
The burner block used during this program had a 30-degree divergent angle with a
15.2-cm (diameter) entrance and a 48.2-cm (diameter) exit to the furnace. To
maximize flame stability the throat nozzle position was used throughout this pro-
gram. The throat nozzle position is located 2.5 cm from the burner block
entrance within the 15.2-cm (diameter) refractory duct connecting the burner
with the block.
Burner operating conditions used throughout the program had the level of flue
oxygen fixed at 3% with the combustion air preheated to 325°F (unless specified
differently). These conditions are considered typical for utility boiler burners.
Based on the overall program objective of evaluating boiler performance changes
when retrofitting from natural gas to medium- and/or low-Btu gas, it was neces-
sary to select an operating variable to model the pilot-scale test furnace against a
field service boiler. Several parameters were considered, including operating wall
temperature, heat absorption profile, and firing density. It was decided to match
the temperature of the combustion products from natural gas at the pilot-scale
25
-------
Figure 13. Diagram of movable-vane boiler burner
AIR FLOW
Figure 14. Method of measuring movable-vane angle for boiler burner
26
-------
furnace exit to the temperature measured between the boiler and the secondary
superheater section of a utility boiler. This decision was based on the complexi-
ties of multiburner interactions within a utility boiler versus the single-burner test
furnace, where several flames would not view the boiler water walls. The data of
most interest in evaluating performance changes were gas volume, temperature,
and gas emissivity entering the secondary superheater. Thus, baseline data were
collected for natural gas after the furnace load had been adjusted to give an exit-
gas temperature of Z550°F.
The first task to be conducted after the baseline furnace operating conditions had
been fixed was to determine flame stability. The stability trials were to be con-
ducted under the most stringent operating conditions, that is, with the furnace
walls, burner block, and combustion air at ambient temperature.
The second problem addressed was the relative flame length of the medium- and
low-Btu gases compared with natural gas. It was anticipated that flame length
changes could affect the rate of heat transfer from the flame to the boiler tubes
through a change in flame emissivity. If the flame length change resulted in a
lower heat removal within the boiler, the combustion products entering the sec-
ondary superheater would be at a higher than normal temperature. This condition
could result in tube damage or a loss in efficiency, because the convective section
would be undersized. Using the burner and furnace operating conditions outlined
for the baseline tests, flame length measurements were made as a function of
vane angle for each test gas. In each case, measurements were taken only after
the exit-gas and furnace-wall temperatures had stabilized. As stated above, a
straight nozzle was used with a diameter giving a fuel velocity of 100 ft/s. The
criterion for evaluating the end of the flame was when 1% of the total gas-sample
volume was combustibles.
The third problem to be answered by this program was what the changes in effi-
ciency — that is, the fraction of total enthalpy input transferred to the load —
would be when retrofitting a unit designed for natural gas with a medium- or low-
Btu gas. Detailed temperature and emissivity data were collected for each test
gas. These measurements provide not only the information needed to evaluate
changes in efficiency within the boiler but also the data that can be used by
manufacturers to calculate any redesign that may be needed in the convective
27
-------
passes. These data are presented and discussed, and calculation methods are
presented in the "RESULTS" section of this report.
The final question addressed by the program was the level of NO emissions from
each test fuel gas and how it compares with emissions from natural gas combus-
tion. These data are presented in detail and analyzed in the "RESULTS" section.
28
-------
RESULTS
During the flame stability trials, fuel-gas velocities from 25 ft/s up to 500 ft/s
were investigated. The optimum fuel-gas injection velocity range was 75 to 200
ft/s. Above 200 ft/s the flame began to lift (detach) from the injector, and below
75 ft/s the flame became lazy and buoyed badly (moved toward the furnace roof).
A velocity of 100 ft/s was selected for use throughout the program because the
velocity could be maintained for the spectrum of fuel gases being studied using
standard pipe sizes.
Figures 15 through 20 show photographs of the flame for each test gas.
FLAME LENGTHS
Table 3 presents the flame length data determined by radial scans at various
points down the furnace axis. The end of the flame was defined as the point
where 99% of the fuel was consumed. In general, the flame length decreased with
increasing burner vane angle, except for the 60-degree angle natural gas and
Wellman-Galusha air, which exhibited longer flames than the corresponding 45-
degree burner vane angle. All medium- and low-Btu fuel-gas flames tested were
shorter than the natural gas flames except Winkler air at a 45-degree vane angle,
which was 21 cm longer than the natural gas flame produced with a 45-degree
vane angle.
All gas temperature measurements reported were gathered using a suction pyro-
meter. The head of the pyrometer consisted of a thermocouple (Pt-Pt Rh 10%)
protected by refractory radiation shields. These shields are fitted to the end of a
water-cooled probe through which gases are sucked and the thermocouple wires
passed. The efficiency of the pyrometer was determined to be 94%, using the
technique presented by Land.*
GAS TEMPERATURE MEASUREMENTS
In order to have a standard of comparison for the medium- and low-Btu gases,
baseline data were collected for natural gas at seven axial positions for vane-
1. Land, T., Instrum. Autom. 29, (No. 7): 1956.
29
-------
Figure 15. Natural gas flame using a 0. 5-inch axial nozzle
Figure 16. Lurgi oxygen gas flame using a 2-inch axial nozzle
-------
Figure 17. Winkler oxygen gas flame using a 2-inch axial nozzle
Figure 18. Koppers-Totzek oxygen gas flame
using a 2-inch axial nozzle
'.I
-------
Figure 19. Wellman-Galusha air gas flame
using a 3-inch axial nozzle
Figure 20. Winkler air gas flame using a 3-inch axial nozzle
-------
Table 3. FLAME LENGTH (cm) AS A FUNCTION OF
FUEL TYPE AND VANE ROTATION
Fuel
Natural gas
Lurgi oxygen
Winkler oxygen
Koppers-Totzek oxygen
Wellman-Galusha air
Winkler air
Vane angle
15°
290
250
250
28!
250
250
30°
250
227
208
208
226
250
45°
219
208
188
188
188
240
60°
226
157
1 46
156
198
208
33
-------
angle orientations of 15, 30, 45, and 60 degrees. These data are plotted in Figures
21 through 24. The axis of the flame was located at the 15.2-cm axial position by
making several radial scans at different vertical positions. The flame center line
was then confirmed by gas analysis. Selected profiles were measured several
times to establish their integrity. Table 4 gives a list of rear-wall temperatures
measured with an optical pyrometer and average flue-gas temperatures measured
with the suction pyrometer at the 395.5-cm axial position as a function of vane-
angle rotation for each of the gases studied. Table 4 also presents the maximum
flame temperature measured and its axial position.
Lurgi oxygen gas temperature data are presented in Figures 25 through 28. Gen-
erally, the maximum average temperature is 112° to 363°F lower than those mea-
sured for natural gas. This difference reflects the 173°F lower adiabatic tem-
perature of the Lurgi oxygen flame. Because of the complex wall-gas radiant flux
equilibrium, this difference reduces to 21° to 110°F at the flue.
Winkler oxygen temperature profiles are shown in Figures 29 through 32, the
adiabatic flame temperature of Winkler oxygen being up to 8°F lower than for
natural gas. For the 15, 45, and 60-degree vane-angle rotations, Winkler oxygen
has a lower maximum average temperature, between 14° and 113°F. The 30-
degree vane rotation produces a maximum temperature 56°F higher than does
natural gas. The flue temperatures, however, are higher than those measured for
natural gas for the 15, 30, and 60-degree vane angles, with temperature differ-
ences in the range of 30° to 56°F.
Koppers-Totzek oxygen has a 241°F higher adiabatic flame temperature than
natural gas. For all vane rotations, the Koppers-Totzek oxygen yielded higher
maximum average temperatures than natural gas. The differences ranged be-
tween 70° and 322°F. These differences have been reduced to 1° to.60°F at the
flue because of wall-gas equilibrium. These temperature profiles are shown in
Figures 33 through 36.
The Wellman-Galusha air temperature profiles are presented in Figures 37 through
40. The maximum average gas temperature is 123° to 225°F lower than that mea-
sured for natural gas. This reflects the 389°F lower adiabatic flame temperature.
34
-------
3000-
2800 —
2600-
2400-
2200-
2000
w
w
0,
w
H
en
<
O
W
o
3
u
> 2700-
2900—1
2500-
2300
6'o
•LEGENI
AXIAL POSITION, cm
O 15.2 O 146. 1
V 57. 2 <] 219.1
A 105. 2 D 310. 8
<>395.6
Q—q
I I I I I I
40 20 £
RADIAL POSITION, cm
20
40
Figure 21. Average gas temperature profile for
natural gas with a 15-degree vane rotation
35
-------
3100-
2900-
2700'-
2500-
5 TEMPERATURE, °F
u> r
O L
0 C
0 C
1 1
< 2800-
O
U
O
W
> 2600-
2400-
o ^V''^"^'" \ ""^^-^A
'A . o ° O-
°"°-o'
r r*f- TTMr\_
AXIAL POSITION, cm
O 15.2 O 146. 1
V 57. 2 < 219. 1 ' 4 ' 20 ' 40 '
A105.2 D 310. 8
O395.6
^o^-o— o--o-o— o—o
o'^,^• ^ o— o- ^
^o— — v ^
1 1 1 1 1 1 1 1 1 1 1 1
60 40 20
-------
.51UU —
2900-
2700-
^
2500-
2300 —
0A
/ ° /°\
i \ ^3 *^^^\
I \ ^^ ^^ ^™^^ ^/V ^^
_ ^v^X:>V' V0/ ^
^ ^y ^^V^ 1
°\ /
o'
T TP^- T7-XTr\
J_i Jl* O ^ IN 1^ '• -~
AXIAL POSITION, cm
O 15.2 O 146. 1
^ 60 V 57.2 < 219.1 ' ^ ' ' 20 ' 40 '
W" A 105. 2 D 310. 8
§ O395.6
H
3 3000-1
06
w
o,
2
W
H
W
< 2800-
0
W
O
2
W
> 2600 -
<;
2400-
2200-
l~\^"^ " O"~~^~ y\.^
/^— ^^ O "O-.
— ^ m Q— ^O '-' Vy~
^Q-^ vXl. <0 -^ ^ ' ^ .krl
_ Q . —Q ** *^^ «^»-— "^J ^ *^J ""•••• ^
^1-— -< <^^*~
^^J
xD'^ " ° D— D " — D-
I 1 1 1 1 i i i 1 i i i
60 40 20 C£ 20 40
RADIAL POSITION, cm
Figure 23. Average gas temperature profile for
natural gas with a 45-degree vane rotation
37
-------
3100 -i
2900 -
2700 -
2500 -
2300
w
D
H
^ 3000 -
W
0,
w
H
< 2800'
O
u
O
<
cd
a
> 2600-
-LEGENI
AXIAL POSITION, cm
O 15.2 O 146.1
V 57. 2 <] 219.1
A 105.2 D 310.8
O395.6
I
-------
Table 4. SUMMARY OF AVERAGE GAS TEMPERATURE DATA
(Part 1)
Fuel
Natural gas
Lurgi oxygen
Winkler oxygen
Koppers-Totzek oxygen
Wellman-Galushaair
Winkler air
Adiabatic
flame
temp
1 5°vane-angle rotation
Wall
Flue
Max
°F
3337
3164
3329
3578
2948
2579
2300
2238
229I
2377
2I29
1 886
2554
2533
2610
2606
2497
2245
2918
2723
2904
2988
2795
2515
Axial
position,
cm
57.2
146.1
1 46. 1
1 46. 1
146.1
146.1
Table 4. SUMMARY OF AVERAGE GAS TEMPERATURE DATA
(Part 2)
Fuel
Natural gas
Lurgi oxygen
Winkler oxygen
Koppers-Totzek oxygen
Wellman-Galusha air
Winkler air
30°vane-angle rotation
Wall
Flue
Max
°F
237I
2318
2318
2372
2129
1 994
263 1
2454
2567
2554
2456
2345
2918
2806
2974
3240
2737
2556
Axial
position.
cm
57.2
146.1
1 46. 1
146.1
1 46. 1
146.1
39
-------
Table 4. SUMMARY OF AVERAGE GAS TEMPERATURE DATA
(Part 3)
Fuel
Natural gas
Lurgi oxygen
Winkler oxygen
Koppers-Totzek oxygen
Wellman-Galusha air
Winkler air
45°vane-angle rotation
Wall
Flue
Max
°F
2309
2300
2309
2327
2066
2012
2553
2442
2523
2554
2434
2335
3045
2682
2932
3115
2820
2556
Axial
position,
cm
15.2
57.2
57.2
57.2
146.1
146.1
Table 4. SUMMARY OF AVERAGE GAS TEMPERATURE DATA
(Part 4)
Fuel
Natural gas
Lurgi oxygen
Winkler oxygen
Koppers-Totzek oxygen
Wellman-Galusha air
Winkler air
60°vane-angle rotation
Wall
Flue
Max
°F
2345
2309
2336
2372
2147
2012
2534
2460
2564
2594
2424
2306
3087
2780
3016
3254
2877
2515
Axial
position,
cm
15.2
105.2
57.2
105.2
105.2
146.1
40
-------
2700-1
2500-
2300-
2100^
1900-
W
oJ
H
1700-
W
W
H
60
O
W
O
3
u
2800-
2600-
2400-
2200-
60
-o—o—o
LEGENI
AXIAL POSITION, cm
O 15.2 O 146. 1
V 57.2 < 219.1
A 105. 2 D310.8
<>395.6
20
1335
T
40
40 20 (
RADIAL POSITION, cm
I
20
I
40
Figure 25. Average gas temperature profile for
Lurgi oxygen gas with a 15-degree vane rotation
41
-------
2900-
2700-
2500-
2300_
2100-
W
ai
D
H
<
ai
W
CX
2
w
H
1900-
O
W
O
2800—1
u
2600-
2400-
2200-
60
-LEGENI
AXIAL POSITION, cm
O 15.2 O 146. 1
V 57. 2 < 219.1
AlOS.2 D310.8
<>395.6
40
20
-------
2900-1
2700-
2500-
2300-
2100-
u
at
D
H
1900-
W
0,
AXIAL POSITION, cm
O 15.2 O U6. 1
V 57. 2 <] 219.1
A105.2 D310.8
<>395.6
T
i I I I
20 40
O
W
O
W
2800H
2400-
2200-1
60
40 20 4,
RADIAL POSITION, cm
40
Figure Z7. Average gas temperature profile for
Lurgi oxygen gas.with a 45-degree vane rotation
43
-------
2900-1
2700-
2500-
2300-
2100-
W
OJ
D
H
3 1900-
'~^s
°-~-0-
V>-0
J
U
a
2
u
H
w
60
40
1*
LEGENI
AXIAL POSITION, cm
O 15.2 O 146.1
V 57. 2 < 219.1
A 105. 2 D 310.8
O395.6
< 2800>-
W
0
3
u
^ 2600-
2400-
2200
x-o—°^-o^
yv^<-^.
^^^ <_^«:^_a___
w v/— — w—— w •— ' 'S/ • — ~"O-— ^
66
I
40
20 (J, 20
RADIAL POSITION, cm
40
Figure 28. Average gas temperature profile for
Lurgi oxygen gas with a 60-degree vane rotation
44
-------
3100-1
2900-
2700-
2500-
2300
W
a!
3000~
60
H
2800-
O
o
H
> 2600-
<
2400-
2200.
D
AXIAL POSITION, cm
O 15.2 O146.1
V 57. 2 <] 219.1
A 105.2 D 310. 8
O395.6
60
50
!
40
I
30
I
20
I
10
I
10 20
RADIAL POSITION, cm
30 40
I
50
Figure 29. Average gas temperature profile for
Winkler oxygen gas with a 15-degree vane rotation
45
-------
3100-
2900H
2700-
2500-
2300
U
Qj
D
H
tf
U
3000-1
u
H
to
< 2800'
O
W
O
U
> 2600-
2400-
2200-
i r
10 20 30 40 50
AXIAL POSITION, cm
O 15.2 O U6.1
V 57. 2 < 219.1
A105.2 D310.8
<>395.6
50
40
30 20 10 £ 10
RADIAL POSITION, cm
20
30 40
50
Figure 30. Average gas temperature profile for
Winkle r oxygen gas with a 30-degree vane rotation
46
-------
3100-1
2900-
2700-
2500 _
2300
p
H
£j 3000-]
W
0,
W
H
< 2800-
W
O
u
> 2600-
LEGEN
AXIAL POSITION, cm
O 15.2 O 146. 1
V 57. 2 <] 219.1
A 105.2 D 310.8
<>395.6
I I
10 20
I
30
40
I
50
2400-
2200-
I
60
50 40
I
30
20
I
10
10 20
I
30
I
40
I
50
RADIAL POSITION, cm
Figure 31. Average gas temperature profile for
Winkler oxygen gas with a 45-degree vane rotation
47
-------
3100-
2700-
2500-
2300"
60
W
at
< 3000-1
W
ex
2
W
H
< 2800-
O
W
O
W
> 2600-
I
-LEGENI
1
20
10
10 20 30 40 50
2400—
2200.
AXIAL POSITION, cm
O 15.2 O 146.1
V 57. 2 <] 219.1
A105.2 D310.8
O395.6
o—o
I
60
I
50
I
40
30
I
20
10
I
10
20 30
40
I
50
RADIAL POSITION, cm
Figure 32. Average gas temperature profile for
Winkler oxygen gas with a 60-degree vane rotation
48
-------
3000-1
2800-
LEGEN
AXIAL POSITION, cm
O 15.3 O 146. 1
V 57. I
A 105. 2
< 219.1
D 310.8
O395.6
2200
40
RADIAL POSITION, cm
Figure 33. Average gas temperature profile for
Koppers-Totzek oxygen gas with a 15-degree vane rotation
49
-------
3300-1
3100-
2900-
2700-
2500
W
0{
EH
3200-1
0,
S
w
H
3000^
O
w
O
2
w
2800-
2600-
2400-
AXIAL POSITION, cm
O 15.2 O146. 1
V 57. 2 < 219.1
A105.2 D310.8
<>395.6
60
I
40
I
20
I
-------
3300-
3100-
2900-
2700-
2500-1—
w-
3200-1
W
H
< 3000-
O
W
O
w
> 2800-
AXIAL POSITION, cm
O 15.2 O 146.1
V 57. 2 <] 219.1
A105.2 Q 310. 8
<>395.6
2600^
2400-
^
2467
i i i T I r
40 20 £
RADIAL POSITION, cm
20
40
Figure 35. Average gas temperature profile for
Koppers-Totzek oxygen gas with a 45-degree vane rotation
51
-------
3300-,
3100-
LEGEN
AXIAL POSITION, cm
O 15.2 O 146. 1
V 57. 2 <] 219. 1
AlOS.2 D 310.8
<>395.6
2400
60
RADIAL POSITION, cm
Figure 36. Average gas temperature profile for
Koppers-Totzek oxygen gas with a 60-degree vane rotation
52
-------
2800n
2600-
2400-
2200-
W
Oi
D
H
<
oj
u
cu
2
u
H
w
<
O
U
o
3
u
<
2000
2800-1
2600-
2400-
2200-
20
40
RADIAL POSITION, cm
Figure 37. Average gas temperature profile for Wellman-Galusha
air gas with a 15 -degree vane rotation
53
-------
2800-1
2600-
2400-
2200-
w
0{
D
H
<
BJ
H
U
H
O
W
O
^
w
2000'
LEG EN
AXIAL POSITION, cm
O 15.2 O 146.1
V 57. 2 <] 219.1
A105.2 D310.8
<>395.6
2800-1
2600 -
2400 -
2200-
60
i
40
I
20
I
20
I
40
RADIAL POSITION, cm
Figure 38. Average gas temperature profile for Wellman-Galusha
air gas with a 30-degree vane rotation
54
-------
2800-
2600-
2400-
2200-
w
QJ
D
H
< 2000'
K
W
0,
W
H
^ 2800-
O
W
O
(4
> 2600-
-LEGEN
AXIAL POSITION, cm
O 15.2 O 146.1
V 57. 2 < 219.1
A 105.2 D 310.8
<>395.6
2400-
2200-
60
i
40
l
20
o—-ex
20
RADIAL POSITION, cm
40
Figure 39. Average gas temperature profile for Wellman-Galusha
air gas with a 45-degree vane rotation
55
-------
uf
a!
Di
W
(X
u
2800-1
2600-
2400-
2200-
2000'
28001
O
u
u
^
w
2600-
2400-
2200-
I
60
-LEGENI
AXIAL POSITION, cm
O 15. 2 O 146. 1
V 57.2 < 219. 1
A105.2 D 310. 8
<>395.6
I I
-------
The Winkler air temperature profiles are presented in Figures 41 through 44. This
116 Btu/CF gas's maximum average temperature is 362° to 572°F lower than that
of natural gas, reflecting its 758°F lower adiabatic flame temperature. Both
Wellman-Galusha air and Winkler air appear able to approach their adiabatic
flame temperatures in conventional hot-wall combustion systems.
To maintain similar mixing between fuel and air, the inlet velocity gradient be-
tween the fuel and air jets was fixed. The fuel inlet velocity was approximately
100 ft/s, while the combustion air inlet velocity was 50 ft/s. This constant velo-
city ratio was maintained by increasing the diameter of the fuel injector. Increas-
ing the nozzle diameter naturally increased the nozzle circumference and there-
fore the interaction area between the fuel and air. The interaction surface for
natural gas is initially 3.14 inches, compared with 6.24 inches for the medium-Btu
gases. The air/fuel ratio for stoichiometric combustion is also lower for the
medium- and low-Btu gases than for natural gas. This, coupled with the larger
interaction surface and the high flame speeds of the medium-Btu gases, would
indicate a higher rate of combustion. However, for the 12 cases studied, although
the flame lengths were generally shorter, the maximum average temperature was
measured at larger axial positions when compared with natural gas for all
medium- and low-Btu test conditions. For the 15, 30, and 45-degree vane angles,
the axial position at which the maximum average temperature was measured was
independent of medium-Btu fuel-gas type. This axial position was 146.1 cm for
the 15 and 30-degree vane-angle rotation and decreased to 57.2 cm for a 45-
degree rotation.
The interaction surface for Wellman-Galusha air gas was 6.87 inches, while for
Winkler air it was 9-62 inches. These interaction surfaces are factors of 2.69 and
3.06, respectively, greater than that for natural gas. The resulting flame lengths
were shorter than those measured for natural gas, except for Winkler air at the
30-degree vane angle, where they were equal, and at the 45-degree vane angle,
where the Winkler air flame length was greater. The position of the maximum
temperature for these gases was longer than for natural gas.
FLAME EMISSIYTTY DETERMINATION
Using the pyroelectric radiometer with a 16-inch diameter cooling target, data
were collected for evaluating gas absorptivity. The Schmidt method was
57
-------
2600-
2400-
2200-
2000-
U
Oj
W
IX
W
H
o
u
o
2
u
1800-
1600-
-LEGEN
AXIAL POSITION, cm
O 15.2 O 146. 1
V 57. 2 < 219. 1
A 105. 2 n 310. 8
<>395.6
60
T
40
T
20
-------
2600-1
2400-
2200-
2000-
W
Bj
D
H
W
W
h
O
W
1800-
1600'
LEGEN
AXIAL POSITION, cm
O 15.2 O 146. 1
V 57.2 < 219.1
AlOS.2 D310.8
O395.6
60
I
40
I
20
1579
W
2600-1
2400-
2200-
-o
I
60
I I 1 I I I
40 20 £
RADIAL POSITION, cm
I
20
I
40
Figure 42. Average gas temperature profile for Winkler
air gas with a 30-degree vane rotation
59
-------
2600 _
Z400_
2200 J
W
oJ
P
H
2000-
1800-
-LEG EN
AXIAL POSITION, cm
O 15.2 O 146. 1
V 57. 2 < 219.1
C*
w
(X
2 J
w
H
w 1600-
A105.2 D310.8 U
<>395.6
I 1 1 1 1 1 I
O 60 " 4"0 " 20 £
W
O
1 1 1 1 1
20 40
2600-1
2400-
2200-
60
40
i
20
i
20
40
RADIAL POSITION, cm
Figure 43. Average gas temperature profile for Winkler
air gas with a 45-degree vane rotation
60
-------
2600-1
2400-
2200-
2000-
h
O
w"
H
1800-
w
w
H
w
o
1600-
.A-
-0— 5—S
60
LEGEN
AXIAL POSITION, cm
O 15.2 O U6.1
V 57. 2 <] 219.1
AlOS.2 D310.8
<>395.6
40
20
1572
> 2600-1
2400-
2200-
60
40
20
-------
employed to reduce the measured data into absorptivities. The following
measurements were made for each data point:
R! = radiation intensity of flame backed by a cold black target
= efEf
RT = radiation intensity of a hot black target
= Er
R3 = radiation intensity of a flame backed by a hot black target
= ef£f + (1 - af) Er
where e^ is the flame emissivity, af is the flame absorptivity for radiation
originating at the hot target, and Ef and Er are the black body emissive powers
at the flame and hot target temperatures. It follows that:
R,
As the target temperature Tr is usually colder than the flame temperature Tf, and
given the properties of nonluminous flame radiation, it is expected that a£ >ej.
The relationship between af and 6f can be found in Hottel and Sarofim^ as:
T a+b~c
— = (—-)
e v T '
f L£
Values of a+b-c can be estimated from Hottel and Sarofim's plots of CO, and
H20.
R£ is measured from a hot refractory surface that is not black; thus the measured
radiation intensity contains reflected as well as emitted radiation. When siting on
a refractory of reflectivity pr, one measures a leaving flux density Wf (the appar-
ent emissive power) equal to the sum of the emitted radiation e E_, and the re-
flected radiation prHr is the flux density incident on the refractory in question.
When there is no flame, the value of Hr is simply the flux density Wr leaving the
refractory surfaces viewed from the reference spot. When there is a flame, the
2. Hottel, H. C. and Sarofim, A. F., Radiative Transfer, 300, New York:
McGraw-Hill, 1967.
62
-------
value of Hf is equal to the sum of the flame radiation CfEf and the radiation
received from other refractory surfaces after attenuation by the flame (1 — €f)Wr.
The three measurements for the Schmidt are then:
1. Flame with cold black background: R . = ff Ef
2. Hot refractory: R, = W = e E +pH = f E + p W = E
1 i. f rr rr rr rr t\
3. Flame backed by hot refractory: R = efE. + ( 1 — cf) W^
efEf + (i - €f) {CrEr + pr [€fEf
f + (i _€f) [€fEr +Pr [CfEf
_
Applying the Schmidt method as before:
R -R
a, = l - —z-=
the following expression for absorptivity is derived:
To quantify the refractory reflectivity, the radiometer was sited on a hot refrac-
tory furnace sidewall through the general probe holder (Figure 11) and then with
the probe holder removed. All precautions were taken to ensure that the radi-
ometer view angle was identical both with and without the probe holder and that
the cold probe holder did not change the refractory temperature during the
63
-------
measurements. Figure 45 presents data collected during these reflectance trials
plotted against the refractory emissivity data provided by Babcock and Wilcox
(solid line).
To evaluate the gas composition emissivity using Equation 1, the following experi-
mental data must be collected: 1) Rj, R£, and Rg are measured using the pyro-
electric radiometer and water-cooled black target; 2) a£ is then calculated using
the Schmidt technique; and 3) the furnace wall temperature is measured using an
optical pyrometer. These measured data are used in conjunction with Figure 45 to
evaluate ^ and pr (refractory emissivity and reflectivity). The final item of in-
formation needed to use Equation 1 is the flame temperature (Tr). These data
were presented in Figures 21 through 43. A review of these temperature profiles
points out the large radial temperature gradients that occur near the furnace
burner wall. Because the analysis technique is built around a uniform radial tem-
perature, the reh'ability of the emissivities at the front of the furnace is very
small. However, as the gases approach the furnace back wall (flue), these gradi-
ents decrease until, at the last axial position where data are collected (395.6 cm
from the furnace burner wall), the temperature profile is flat. The flame tem-
perature used in evaluating emissivity is an average of the radial temperature
measurements weighted by their annular area. These values (average flame tem-
perature Tj, refractory temperature Tr, refractory emissivity €r, and refractory
reflectivity Pr) are plugged into Equation 1, leaving two values to be determined:
&f and €f. A value for the flame emissivity is assumed, which allows the flame
absorptivity to be estimated. This estimated value for a^ is compared with the
measured value. If the two values do not correspond, an iterative procedure is
started and continued until satisfactory agreement has been reached.
The emissivity profiles evaluated for natural gas, Lurgi oxygen, Koppers-Totzek
oxygen, Winkler oxygen, Wellman-Galusha air, and Winkler air are presented in
Figures 46 through 51.
The variation of emissivity along the furnace length was observed for all fuel
gases tested. Because gas emissivity is a function of composition, temperature,
and path length, the variation of emissivity along the furnace axis can best be
explained as an aerodynamic effect.
64
-------
-------
0.2—I
15
H
HH
>
w
o
o.i—
NATURAL CAS, 2174 SCF/hr
GAS NOZZLE THROAT POSITION
O2 IN FLUE, 3°~,
BURNER BLOCK ANGLE, 30°
VANE ROTATION AS LABELED
I
50
I
100
I I 1
150 200 250
AXIAL POSITION, cm
300
I
350
400
Figure 46. Gas emissivity versus axial position for natural gas
-------
0.2—1
o.i—
O
LURGI OXYGEN, 7840 SCF/hr
GAS NOZZLE THROAT POSITION
O2 IN FLUE. 3%
BURNER BLOCK ANGLE. 30°
VANE ROTATION AS LABELED
I
50
I
100
I I I
150 200 250
AXIAL POSITION, cm
I
300
I
350
I
400
Figure 47. Gas emissivity along furnace length for Lurgi oxygen gas
-------
0.2—1
H
" o.i-
o
oo
I
50
KOPPERS-TOTZEK OXYGEN, 7679 SCF/hr
GAS NOZZLE THROAT POSITION
O2 IN FLUE, 3%
BURNER BLOCK ANGLE, 30°
VANE ROTATION AS LABELED
I
100
150
I
200
250
I
300
350
400
AXIAL POSITION, cm
Figure 48. Gas emissivity along furnace length for Koppers-Totzek oxygen gas
-------
0.2 —
o>
1
W 0.1
(0
O
i!5°
WINKLER OXYGEN, 8303 SCF/hr
GAS NOZZLE THROAT POSITION
O2 IN FLUE, 3%
BURNER BLOCK ANGLE, 30°
VANE ROTATION AS LABELED
I
50
100
I
150
I
200
250
I
300
I
350
I
400
AXIAL POSITION, cm
Figure 49. Gas emissivity along furnace length for Winkler oxygen gas
-------
0.2-,
15°
H
KH
-J
o
w
o
0.1-
WELLMAN-GALUSHA AIR, 13,554 SCF/hr
GAS NOZZLE THROAT POSITION
O2 IN FLUE, 3%
BURNER BLOCK ANGLE, 30°
VANE ROTATION AS LABELED
I
50
100
150
I
200
250
I
300
350
I
400
AXIAL POSITION, cm
Figure 50. Gas emissivity along furnace length for Wellman-Galusha air
-------
0.2-.
w o.H
o
WINKLER AIR, 19,396 SCF/hr
GAS NOZZLE THROAT POSITION
O2 IN FLUE, 3%
BURNER BLOCK ANGLE. 30°
VANE ROTATION AS LABELED
50
100
I I I
150 200 250
AXIAL POSITION, cm
I
300
350
400
Figure 51. Gas emissivity along furnace length for Winkler air
-------
A general aerodynamic characterization of a flame can be made by determining
the different types of flow patterns that exist within a combustion chamber. A
detailed flow analysis of a confined flame reveals that the front section of a com-
bustion chamber can be divided into four zones: primary jet, primary recircula-
tion, secondary jet, and secondary recirculation. The primary and secondary jets
contain only particles with a forward flow direction (away from the burner),
whereas the recirculation zones contain gas particles moving in the reverse flow
direction (back toward the burner). The size, shape, and particle density of the
recirculation zones are determined by the velocity, the ratio of gas to air, the spin
intensity of the secondary jet, the burner-block angle, and, for the secondary
recirculation zone, the size and shape of the combustion chamber. Figure 52
shows the type of flow pattern that was observed during this investigation. This
flow pattern is conventionally labeled Type II.
A Type II flow pattern is generated when the secondary jet has a tangential velo-
city component large enough to cause the particles to adhere to and pack tightly
against the burner block. This packing creates a low or negative pressure region
in the center of the burner block. The pressure differential between the furnace
and the central region of the burner block causes gas molecules to be pulled into
this region and back toward the burner, thus creating the primary recirculation
zone. When the velocity of the primary jet is greater than the velocity of the
recirculating gases in the primary recirculation zone, the primary jet penetrates
this reverse flow region, and a recirculation lobe occurs on each side of the burner
axis.
A gas analysis of the secondary recirculation zone reveals species concentrations
similar to those measured at the furnace flue. Thus, an emissivity measured at
the furnace front wall would be an addition of the flame plus secondary recircula-
tion zones. The emissivity from the recirculation Zone is approximately that mea-
sured at the rear of the furnace. As the flame spreads, the optical path of the
secondary recirculation zone is decreased, reducing the concentration of radiating
sources. As the combustion continues, the concentrations of CC>2 and I^O along
the optical path are increased as products of combustion. At the end of the fur-
nace, the gas temperatures and species concentrations across the width of the fur-
nace are nearly uniform, and the emissivity is as if from a static mixture at a uni-
form temperature and composition. This is confirmed by the emissivities
72
-------
OJ
TVPE n
LOW SWIRL INTENS\TY
PRIMARY JET VELOCITY >
SECONDARY JET VELOCITY
Figure 52. Flame flow pattern tested
-------
measured for natural gas combustion (Figure 46), where the measured emissivity
at the front of the furnace is different for each vane angle tested; toward the end
of the combustion chamber, however, all the values converged to approximately
0.177.
The variation of emissivity near the furnace front wall as a function of vane angle
occurs due to the tangential velocity component of the secondary air - that is,
how fast the vortex of combustion air will spread upon leaving the burner block.
The larger the tangential velocity component, the higher the rate of post burner-
block expansion and the smaller the contribution of the secondary recirculation
zone to the measured emissivity.
Another possibility is that a^ (which is the value measured) increases for some
distance into the furnace as a consequence of the presence of either soot or cold
pockets of absorbing gas.
Emissivities were calculated using the measured flue-gas temperatures and com-
positions for a 45-degree vane-angle rotation. Two methods of calculating emis-
sivities were used. Emissivities were calculated using the method given by Hottel
and Sarofim^ in Radiative Transfer and that developed by Leckner^ using NASA
data. For all the test gases except one, the measured emissivities fell between
the two calculated values. The exception is the Lurgi oxygen medium-Btu gas,
where the measured value was lower than both the calculated emissivities. These
calculated values are listed in Table 5. They range from a minimum of 0.150 for
Winkler air to a maximum of 0.194 for Lurgi oxygen. These calculated values are
independent of burner operating conditions, and the furnace geometry is con-
sidered only when evaluating the mean beam length. Because the radiometer
viewed a collimated beam, the furnace width corresponds to the beam length.
Lurgi oxygen measured emissivities are illustrated in Figure 47. The emissivity
measured at the furnace flue was 0.185, which is 5% lower than the value
2. Hottel, H. C. and Sarofim, A. F., Radiative Transfer, 300, New York:
McGraw-Hill, 1967.
3. Leckner, B., Combust. Flame 19:33, 1972.
74
-------
calculated by the Hottel method and 25% lower than the emissivity calculated by
the Leckner method.
Table 5. CALCULATED AND MEASURED EMISSIVITIES
£caic = calculated emissivity
a'meas = measured absorptivity
emeas = measured emissivity
Difference =
| ecaic-Cmeasl
«calc
Fuel
Natural gas
Lurgi oxygen
Koppers-Toizck
oxygen
Winkler oxygen
Wellman-Galushu air
Winkler air
Winkler oxygen plus
15% FGR
PCOZ
0.083
0.169
0.204
0.183
0.159
0.147
0.183
"HP
0.167
0.179
0.121
0.161
0.092
0.081
0.161
Gas temp.
°R
3013
2902
3014
2983
2894
2795
2955
ameas
0.131
0.158
0.145
0.153
0.115
0.112
0.182
emeas
0.177
0.185
0.190
0.218
0.190
0.170
0.238
Hottel-Sarofim
Emissivity
0.159
0.194
0.174
0.189
0.154
0.150
0.193
%A«
+ 11
-5
+ 9
-US
+ 23
+ 13
+ 23
Leckner
Emissivity
0.215
0.246
0.211
0.232
0.195
0.193
0.237
%A«
-18
-25
-10
-6
-3
-12
0
The average emissivity measured at the flue for Koppers-Totzek oxygen was
0.190, compared with 0.174 from the Hottel calculation and 0.211 from the
Leckner calculation. Thus the measured value was 9% high or 10% low, respec-
tively. The profiles of the measured emissivities for Koppers-Totzek oxygen are
presented in Figure 48.
The Winkler oxygen fuel measured emissivity was 15% higher than the Hottel and
6% lower than the Leckner emissivity. The emissivity profiles measured down the
furnace axis are presented in Figure 49*
Emissivity profiles measured for Wellman-Galusha air are presented in Figure 50.
These measured emissivities converged to a value of 0.190, which is 23% higher
than the Hottel calculated value, but only 3% lower than the Leckner value.
75
-------
Winkler air measured emissivities are presented in Figure 51. The post-flame
emissivity value was measured at 0.170, which is 13% greater than the Hottel and
12% less than the Leckner calculated values.
The agreement between the calculated and measured emissivities is amazingly
good considering the fact that the measurements were made on an operating fur-
nace with nonuniform temperatures and compositions. From reviewing the synop-
sis of calculated and measured emissivities presented in Table 5, the following
observations can be made:
• The measured emissivities, with the exception of Lurgi oxygen, were
higher than the Hottel calculated values and lower than the Leckner,
indicating a bias in the measurements or calculations rather than random
experimental error.
• The average deviation of the measured emissivities was 11% with re-
spect to the Hottel calculated emissivities and -12% with respect to the
Leckner calculated emissivities.
These experimental data thus provide a basis for recommending the use of calcu-
lated emissivities for extrapolating to the dimensions of commercial boilers. As
Table 5 shows, the Hottel calculation method gives the more conservative esti-
mate, while the Leckner calculation method is more optimistic.
At least part of the 11% error between measured emissivity and that calculated
from the Hottel and Sarofim method occurs in the correction due to spectral over-
lap of CO2 and ^O in a mixture. The highest temperature at which correction
data are currently available is 1700°F. However, it can be inferred from emis-
sivity charts of CC^ and ^O that the overlap correction will decrease with tem-
peratures above 1700°F. Because the values were calculated using the 1700°F
overlap correction, they are smaller in magnitude than if they had been calculated
with a 2500°F overlap correction. It is estimated that this correction could lead
to an increase of 5% to 10% in the calculated emissivities.
Figure 53 is a plot of emissivity versus furnace length for a furnace fired with
coke-oven gas.* These data help to quantify the differences in emissivity between
the "clean" medium- and low-Btu gases tested in this program and those contain-
1. Land, T., Instrum. Autom. 29, (No. 7): 1956.
76
-------
ing tars, oils, and particulates, which could be expected from a coal gasifier. The
data in Figure 53 were collected on the International Flame Research Foundation
furnace, and the firing density was higher than that used in our experiments
(10,000 Btu/CF-m versus 6,000 Btu/CF-m). The coke-oven gas gave a calculated
gas emissivity of 0.2. The measured emissivity, however, was 0.38. The large
difference between the calculated and the measured values is due mainly to the
tars and particulates carried by the coke-oven gas.
0.6-,
H
I-H
>
W
w
<
a
0.4
0.2
COKE-OVEN GAS
M. RIVIERE
IJMUIDEN 1953
100 200 300
AXIAL POSITION, cm
400
500
Figure 53. Emissivity profile for coke-oven gas
NITRIC OXIDE EMISSIONS
To determine the environmental impact of retrofitting utility boilers with
medium- and low-Btu gases, NO emissions levels were measured as a function of
secondary air-preheat temperature for each of the test fuel gases. Baseline NO
emission data were collected using natural gas. The combustion-air temperature
was varied from ambient to 800°F. Graphs of the experimental results are pre-
sented in Figure 54. For the vane orientations studied, the measured NO levels
remained relatively constant up to 400°F, above which dramatic increases
occurred for the 45-degree and 60-degree angles.
77
-------
oo
240-1
200-
160-
a
120-
80-
NATURAL GAS, 2174SCF/hr
GAS NOZZLE THROAT POSITION
02 IN FLUE, 3%
BURNER BLOCK ANGLE, 30°
VANE ROTATION AS LABELED
-0.3
4->
B
•0.2 o
^E
5
-0.1
40-
100
300 500
SECONDARY AIR PREHEAT, °F
700
900
Figure 54. NO versus secondary air preheat with natural gas
-------
Figure 55 presents NO versus secondary air-preheat temperature test results using
Lurgi oxygen gas. The flue concentrations of NO are approximately 50% less than
with natural gas for all test conditions. For the detailed burner operating condi-
tions (15-degree vane angle, 325°F air-preheat temperature), the NO level was re-
duced from 65 ppm (natural gas) to 32 ppm (Lurgi oxygen gas). These emission
reductions were expected based on an adiabatic flame temperature comparison,
because Lurgi oxygen has a 173°F lower temperature than natural gas.
Flue levels of NO measured as a function of combustion-air temperature with
Koppers-Totzek oxygen gas are shown in Figure 56. For all conditions tested, the
Koppers-Totzek gas resulted in higher NO emission levels than did natural gas.
Again, by comparing the adiabatic flame temperatures listed in Table 2, this re-
sult was anticipated because Koppers-Totzek gas has a 241°F higher adiabatic
flame temperature than natural gas. For the 45-degree vane angle with a 325°F
secondary air-preheat temperature, this is translated into a 60% increase in the
NO emission level.
Comparing the adiabatic flame temperatures of Winkler oxygen gas with natural
gas shows that the former is 9°F lower. Thus, based only on adiabatic flame tem-
perature considerations, the NO emission levels should be comparable. The NO
test results with Winkler oxygen gas are presented in Figure 57. These measured
levels were lower than those measured for natural gas, except for the 60-degree
vane rotation with secondary air-preheat temperatures below 560°F and the 45-
degree vane rotation with preheat temperatures below 400°F. These results indi-
cate that NO formation strongly depends on combustion aerodynamics because the
peak flame temperature and the time that the gas is subjected to this temperature
are controlled by aerodynamic turbulence. Although average temperatures may
be the same in two cases, the NOX formation can be different if the turbulent
temperature fluctuations are different. The temperature dependence of the
chemical reaction rate constant can be expressed by the Arrhenius rate law:
k(T)=ko exp [-E/RT(t)]
The time-averaged value would be:
^ (t +T)
— J ° exp[-E/RT(t)] dt
t
79
-------
120-,
oo
o
100-
80-
a
O
2
60-
40-
LURCI OXYGEN, 7840SCF/hr
GAS NOZZLE THROAT POSITION
O, IN FLUE, 3%
BURNER BLOCK ANGLE, 30°
VANE ROTATION AS LABELED
20-
-0.15
-0.10
n
c
o
d1
2
JD
-0.05
100
T~
300
500
SECONDARY AIR PREHEAT, °F
700
900
Figure 55. NO versus secondary air preheat with Lurgi oxygen gas
-------
00
400-
300-
cx
a
o"
200-
100-
KOPPERS-TOTZEK OXYGEN, 7679 SCF/hr
GAS NOZZLE THROAT POSITION
O2 IN FLUE, 3%
BURNER BLOCK ANGLE, 30°
VANE ROTATION AS LABELED
—O
,_0.6
-0.5
4-0.4
-0.3
.-0.2
-0.1
§
100
300
500
SECONDARY AIR PREHEAT, °F
I
700
900
Figure 56. NO versus secondary air preheat with Koppers-Totzek oxygen gas
-------
oo
200-
160-
g
a
120-
O
2
80-
40"
WINKLER OXYGEN, 8303 SCF/hr
GAS NOZZLE THROAT POSITION
02 IN FLUE, 3%
BURNER BLOCK ANGLE, 30°
VANE ROTATION AS LABELED
60°
100
300
500
700
SECONDARY AIR PREHEAT, °F
-0.3
•0.2 c
o
O*
2
-0.1
900
Figure 57. NO versus secondary air preheat with Winkler oxygen gas
-------
for an interval of time, T, while the rate constant given by the time-averaged
temperature would be:
_ k _
k(T) =-~ exp [-E/RTJ.
The ratio -
K = -^D.
k(T)
This ratio is always greater than 1 because the exponential nature of the tempera-
ture dependence means that the rate constant increase for an increase in tem-
perature above the average exceeds considerably the decrease for a similar reduc-
tion in temperature. The value of the ratio, K, depends on the extent and the
nature of the temperature fluctuations. Thus, observing that the average tem-
perature (the temperature given by a suction pyrometer) is similar in two cases
does not necessarily mean that the NOX emissions will be similar if the turbulent
fluctuations are different. Unfortunately, the technology is not sufficiently
advanced to allow measurement of the extreme turbulent temperature pulsations.
An additional indication of combustion aerodynamic differences is the reordering
of the NO concentration levels as a function of vane angle, which can be seen by
comparing Figure 56 with Figure 57.
Figure 58 shows the NO emission data collected for Wellman-Galusha air. The
flue concentrations of NO are about 70% lower than those measured for natural
gas at the lower air-preheat temperatures. They show an almost linear rise with
air temperature. The lack of a dramatic increase in emissions with temperature
results in up to a 90% lower NO concentration at elevated combustion-air tem-
peratures for Wellman-Galusha air than for natural gas. Some reduction of NO
emissions could be anticipated because of the lower adiabatic flame temperature
(2948° versus 3337°F).
The NO emission data for Winkler air are presented in Figure 59« As the low adia-
batic flame temperature would predict, the NO concentrations in the flue are the
lowest of all the gases. The emissions are 82% to 97% lower than those measured
83
-------
oo
30-,
25-
20-
O
10-
WELLMAN-GALUSHA AIR, 13,554SCF/hr
GAS NOZZLE THROAT POSITION
5-| O2 IN FLUE, 3%
BURNER BLOCK ANGLE, 30°
VANE ROTATION AS LABELED
100
300
500
SECONDARY AIR PREHEAT, °F
I
700
15°
-0.050
-0.025
a
o
(51
2
XI
900
Figure 58. NO versus secondary air preheat with Wellman-Galusha air gas
-------
00
C.O -
20-
15-
E
a
a
O
10-
5-
0
WINKLER AIR, 19, 396 SCF/hr
GAS NOZZLE THROAT POSITION
O2 IN FLUE. 3%
BURNER BLOCK ANGLE, 30°
VANE ROTATION AS LABELED
A, 60"
A A A A
A _--^-v--~~v 45°
_-^— -^ 'v '^^L=Q^==o=^=^^
?\"ZZ— rO" D "
^=L> D
« 1 1 1 1 1 1 1
-0.050
3
+>
CO
o
• p4
i—4
-0.025-E
n
u
z
5
100 300 500 700 900
Figure 59. NO versus secondary air preheat with Winkler air gas
-------
for natural gas and show an even more gradual, nearly linear increase with
combustion-air temperature.
The NO data for Wellman-Galusha air show a systematic increase in emissions
with increased vane angle. This same relationship is not seen with Winkler air.
The 15, 45, and 60-degree angle data are all within 2 to 3 ppm of each other but
are 5 to 6 ppm lower than the 30-degree angle data.
Figure 60 is an Arrhenius plot for the fuel gases tested with the utility burner at a
45-degree vane rotation. The plot, jfln(NO) versus reciprocal of the adiabatic
flame, yields a linear relationship. Correlating these data to a relationship sug-
gested by Thompson, Brown, and Beer^ on the formation rate of NO which is of
the form:
d(NO)
= A exp(-134. 7/RT)
Vi V-
yields an activation energy of 153 kcal/mol compared with their 134.7. This good
agreement suggests that peak temperatures approaching the adiabatic flame tem-
peratures are controlling the rate of NO formation and suggests that the use of
adiabatic flame temperature is a good empirical method of predicting NO emis-
sions.
4. Thompson, Brown and Beer, Combust. Flame, 19, 69 (1972).
86
-------
6.0
5.0-
4.0
E = 153 kcal
o
I—J
p-
3.0'
2.0
1.0
0.25
0.30
IOOO/T
AF,
0.35
Figure 60. Arrhenius plot of NO versus
inverse "pseudo-adiabatic" flame temperature
87
-------
DATA CORRELATION
NOX EMISSIONS
Due to the size of the gas-generating system for synthesizing the medium- and
low-Btu test gases, the firing density of the pilot-scale test furnace could not
exceed 6923 Btu/ft^'hr. To permit a better correlation as to the level of emis-
sions that could be anticipated from a utility boiler, natural gas was fired with
inputs up to 3.75 million Btu/hr. This corresponds to a firing density of 11,538
Btu/ft^-hr, which produced a NO emission level of 225 ppm compared with the
New Source Performance Standard for natural gas at 1000 Btu/SCF of 168 ppm
(0.2 pounds NO2/million Btu heat input). These data are graphed in Figure 61.
A 300-MW utility boiler with a radiant chamber 30 x 60 feet in cross section and
80 feet high is fired with 2.86 X 10^ Btu/hr. The firing density for a boiler of this
size at this firing rate is 19,860 Btu/ft^-hr. Extrapolating the data of Figure 61
yields a NO emission level for the 19,860 Btu/ft^-hr firing density of 458 ppm. To
quantify the anticipated NO emission levels from this 300-MW boiler for each of
the fuel gases tested, it is assumed that an identical firing density-to-NO emission
level relationship would occur for each medium- and low-Btu gas as is represented
for natural gas in Figure 61. From this assumption, the NO emission levels pre-
sented in Table 6 were projected.
Thus, there were only two fuel gases tested that could comply with the 168-ppm
performance standard with the burner system operating normally. Both of the
fuel gases (Wellman-Galusha air and Winkler air) that comply with the standard do
so because of their low adiabatic flame temperatures, 389° and 758°F, respec-
tively, below the natural gas adiabatic flame temperature. All medium-Btu gases
that have adiabatic flame temperatures near that of natural gas would require
modifications to the burner/boiler system in order to comply with the perfor-
mance standards.
CORRELATING FURNACE PERFORMANCE
The efficiencies evaluated on the pilot-scale furnace for each of the fuels tested
are listed in Table 7. These efficiencies (defined the same as previously for the
boiler, that is, the fraction of the total input enthalpy given up in the furnace by
the combustion products) can be evaluated using the flue gas temperatures in
88
-------
400-,
300 -
g
a
a
Z
o
HH
H
H
Z
W
u
z
O 200
U
W
O
Z
100
- 0.3
- 0.2
2
a
c
o
(51
z
-0.1
FUEL: NATURAL CAS
4.0 8.0 12.0 16.0
FIRING DENSITY, Btu/ft3 x 103
20.0
24.0
Figure 61. NO flue concentration versus firing density
89
-------
Table 6. PROJECTED NOX EMISSION
LEVELS FOR A UTILITY BOILER
Fuel
Natural gas
Lurgi oxygen
Koppers-Totzek oxygen
Winkler oxygen
Wellman-Galushaair
Winkler air
NOX, ppm
458
225
733
514
106
28
lbN02/106Btu
0.714
0.353
1.019
0.766
0.198
0.061
Table 7. TEST FURNACE EFFICIENCIES AND COMBUSTION
PRODUCT PROPERTIES AT THE FLUE
Fuel
Natural gas
Lurgi oxygen
Winkler oxygen
Koppers-Totzek oxygen
Wellman-Galusha air
Winkler air
na
28.6
25.2
26.4
33.1
19.0
10.8
Tfl0e. °Fb
2553
2442
2523
2554
2434
2335
rh,SCFc
26,639
26,883
25,481
24,236
32,889
37.919
Q.Btu/hrd
,693,083
,762,337
,730,291
,570,836
,910,009
2.104.174
a Efficiency measured Tor pilot-scale test Furnace.
b Temperature of combustion products at the flue.
c Volume of combustion products at 60°F.
d Heat content of combustion products at listed temperature.
90
-------
Table 7 and the combustion products in Table 1. Subtracting the flue gas enthalpy
from the total enthalpy input gives the amount of heat transferred in the furnace,
and dividing by the total enthalpy input gives the efficiency. The efficiencies
evaluated for natural gas and the oxygen produced low-Btu test gases (Winkler
oxygen, Koppers-Totzek oxygen, and Lurgi oxygen) are comparable within a range
of 25% to 33%. For Wellman-Galusha air with a 160 Btu/ft^ heating value, the ef-
ficiency decreased to 19*0%. The lowest efficiency measured was 10.8%, for
Winkler air with a 116 Btu/ft^ heating value.
In addition to changes in the furnace efficiency when natural gas is replaced with
a low-Btu gas, the temperature and volume of combustion gas products entering
the convective section will determine the amount of heat absorbed. Using Table 7
to compare the volume of combustion products for the test fuels reveals that
Koppers-Totzek oxygen has a 9% smaller volume of combustion products than
natural gas, while Winkler air produces a 42% greater volume of combustion pro-
ducts than natural gas. These changes in combustion-product volume will result in
changes in gas velocities and will shift the heat absorption patterns within the
convective section.
To evaluate a shift in heat absorption when retrofitting a boiler with low-Btu gas
requires a method of calculation. Babcock and Wilcox Co. (B&W) agreed to take
our experimental data and use them for the input conditions to the convective
section to predict differences in overall boiler efficiency (radiant plus convective)
between natural gas and low-Btu gases. The method of calculation employed by
B&W was to use their design computer program to forecast changes in boiler per-
formance when retrofit with a low-Btu fuel gas.
The particular boiler selected for their calculations was a standard design with a
maximum rated steam flow of 2,430,500 Ib/hr at 2620 psig/1005°F at the super-
heater outlet when fired with natural gas. The unit (B&W Contract RB-455) is
installed at the Teche Station of the Central Louisiana Electric Co. and supplies
steam to a Westinghouse turbine having a maximum capacity of 361 MW. This
same unit was reported on by B&W for EPRI Project 265-2, entitled "Low Btu Gas
Study," a program designed to look at retrofitting boilers using only existing de-
sign information.
91
-------
To update these results, the flue gas temperatures and gas compositions measured
on the IGT furnace were used as input data. These temperatures and gas composi-
tions were assumed to be identical to those entering the convective section of the
boiler. Using these experimental results, B&W calculated unit efficiencies for
natural gas, Winkler oxygen, and Winkler air fuel gases. The total required output
of Unit RB-455 based on a 2,198,000 Ib/hr main steam flow is 2652.5 X 106 Btu/hr.
Based on the experimental data and fuel analysis presented in Table 1, B&W gen-
erated the information in Table 8. The first line shows natural gas firing with the
amount of excess air that the unit was designed for at this load. The second line
shows natural gas firing with input conditions (325°F air preheat and 3% excess
oxygen) identical to those used for the pilot-scale test furnace. The actual output
from firing natural gas with the increased level of excess air is in excess of 2652.5
X 10^ Btu/hr. This is because higher gas weights result from the increased volume
of combustion air creating an overabsorption in the reheater. Similar reheat
overabsorption occurs in the output for Winkler air, which is also greater than
2652.5 X 106 Btu.
Table 8. B&W CALCULATED BOILER EFFICIENCIES USING IGT DATA
Fuel
Natural gas
Natural gas
Winkler oxygen
Winkler air
IGT
flue-gas
temp, °F
__
2553
2523
2335
Actual
unit
output,
1 06 Btu/hr
2652.5
>2652.5
2652.5
>2652.5
Unit
efficiency,
%
85.1
84.8
84.4
80.4
62 in
flue gas,
% by vol
1.1
3.0
3.0
3.0
B&W
furnace
exit-gas
temp, °F
2702
2633
2583
2397
It should be noted that these calculations in no way imply that this, or any other
unit, could economically be retrofitted to burn these test fuels. Several other
factors in addition to unit efficiency must be considered in an actual retrofit
study. Two of these important factors are: 1) redistribution of heat absorption
between boiler, primary and secondary superheater, and reheat and economizer
surface (a significant redistribution of which will affect metal temperatures,
92
-------
spray flows, and circulation); and 2) changes in combustion-air and flue gas weight
from the original design quantities, which will affect fan and air heater perfor-
mance. Both of these factors can have a significant influence on the feasibility of
retrofitting existing units designed for conventional fossil fuel firing.
To aid in determining the value of the experimental data, B&W reran their design
calculation without using the experimental results as input. The new inputs re-
quired for the computer program were the fuel composition and the desired tem-
perature of combustion products leaving the air heater. The results of these cal-
culations for the gas temperature and emissivity at the exit of the radiant furnace
section are listed in Table 9. The combustion-products temperature leaving the
radiant section show excellent agreement with the experimental values. This pro-
vides substantiation on how realistically the pilot-scale test furnace was able to
model the radiant section of a utility boiler.
Table 9. CALCULATED AND MEASURED
EFFICIENCIES AND GAS EMISSIVITIES
Fuel
Natural gas
Lurgi oxygen
Winkler oxygen
Koppers-Tolzek oxygen
Wellman-Galusha air
Winkler air
IGTa
T8
2553
2442
2523
2554
2434
2335
«9
0.18
0.19
0.22
0.19
0.19
0.17
B&Wb
T9
2633
—
2583
—
—
2397
«9
0.29
—
0.35
—
—
0.31
a Experimental data collected on pilot-scale test furnace.
b Calculated values based on normal design technique.
Comparing the gas emissivities of the combustion products at the radiant section
exit as measured by IGT to those calculated by B&W reveals a large difference.
This difference can be understood by looking at the emissivities dependent vari-
ables. These variables are the partial pressure of the radiating gas, the tempera-
ture of the gas mixture, and the distance across (beam length) the radiating gas.
The temperature and partial pressure of the radiating gases are similar for the
93
-------
IGT test data and the B&W calculated values. Therefore the difference between
the pilot-scale furnace and the 360-MW boiler is due to the different beam
lengths, 4 feet and 22 feet, respectively. To directly compare the experimental to
the calculated emissivity would require knowing the dependence of emissivity on
path length. Adjusting the B&W calculated emissivity to the beam length of the
test furnace was not practical because B&W does not use a publicized calculation
technique for evaluating emissivities but has developed a semi-empirical method
based on years of building and measuring heat absorption rates within the radiant
boiler section. On the other hand, the width of the test furnace was fixed, thus
negating the possibility of experimentally quantifying the dependence of emis-
sivity on beam length. However, since the only difference between the IGT mea-
sured and B&W calculated emissivities is the beam length (products of combustion
and furnace exit temperatures are similar), a graphical development of the rela-
tionship between emissivity and beam length is possible. This relationship is gen-
erated using the zero-zero data point (zero emissivity at zero path length), the
IGT measured emissivity, and the semi-empirical emissivity of B&W. The result-
ing curve for natural gas is presented in Figure 62 and is labeled experimental.
Also shown in this figure are relationships between the calculation technique of
Hottel and Sarofim and the technique of Leckner.
As the plot shows, the calculated values are approximately proportional to the
square root of the mean beam length. Leckner developed a statistical model
based on existing spectral data to evaluate total emissivities of carbon dioxide and
water vapor in homogeneous gases. For carbon dioxide emissivity, there is good
agreement with Hottel's emissivities; however, for water vapor, Leckner's emis-
sivities at temperatures above 1650°F are consistently higher and show that the
partial pressure correction factor is temperature-dependent. The measured emis-
sivity lies between the two theoretical curves presented. At the larger path
lengths comparable to utility boilers, the experimental curve drops below both
theoretical curves.
Similar graphs have been developed showing the relationship of emissivity and the
square root of mean beam path length for Winkler oxygen and Winkler air fuel
gases. These graphs are presented in Figures 63 and 64. Conclusions similar to
those with natural gas can be drawn; the measured emissivity lies between the
theoretical curves, while the semi-empirical data point lies below both theoretical
94
-------
0.6-1
0.5-
0.4-
H
K-t
>—)
O
CO
t-H
s
w
0.2-
0.1—
LECKNER
HOTTEL
EXPERIMENTAL
FUEL: NATURAL GAS
TEMPERATURE =2553°F
10
I
20
I
30
40
SQUARE ROOT OF PATH LENGTH, cm
1/2
(
50
60
Figure 62. Emissivity versus square root
of path length for natural gas
95
-------
0.6-1
0.5-
0.4'
W
0.2-
0.1-
BXW
LECKNER
HOTTEL
EXPERIMENTAL
FUEL: WINKLER AIR
TEMPERATURE =2335°F
10 20 30
SQUARE ROOT OF PATH LENGTH, cm
40 50
1/2
60
Figure 63. Emissivity versus square root
of path length for Winkle r air
96
-------
LECKNER
HOTTEL
EXPERIMENTAL
FUEL: WINKLER OXYGEN
TEMPERATURE =2523°F
30
40
SQUARE ROOT OF PATH LENGTH, cm
1/2
50
60
Figure 64. Emissivity versus square root
of path length for Winkler oxygen
97
-------
curves. These curves provide a means of estimating the emissivity of nonluminous
combustion products with CO^/r^O ratios between 0.5 and 1.8. However, these
results are very restrictive because they apply only to the temperatures for which
the calculations were made. However, as stated previously, these temperatures,
approximately 2500°F, are comparable with those found at the exit of the furnace
(radiant section) in a utility boiler.
MATHEMATICAL MODEL OF FURNACE PERFORMANCE
To allow the experimental data to be accurately extrapolated, a mathematical
model was sought that would provide estimates of temperature and emissivity in
agreement with the experimental results. In addition, the model had to be easy to
use (not requiring a high-speed computer) yet provide reliable estimates. The
method selected to compare the experimental results with those calculated by
B&W is the "speckled-wall" model (SWM) developed by Hottel.5 This model makes
use of the following assumptions:
1. The combustion products and flame hi the combustion chamber are assigned a
single mean temperature; this means that the furnace is treated as a well
stirred reactor,
2. The gas is gray,
3. The heat sink surface area is gray and can be assigned a single temperature,
4. External losses through the furnace walls are negligible, and internal
convection to the refractory furnace walls can be neglected, and
5. The arrangement of heat sink and refractory surfaces is such that any point on
the furnace walls has the same view-factor to sink surfaces as any other point.
This occurs only when the sink and refractory surfaces are intimately mixed.
The walls are speckled.
The items of information needed to use the model applying the iterative method
described below are the adiabatic flame temperature, the total enthalpy, the total
surface area of the furnace, the ratio of sink surface area to refractory surface
area, and the emissivity of the sink.
5. Hottel, H. C., J. Inst. Fuel 34; 220, 1961.
98
-------
The model has the capability to:
1. Account for convective heat flux to the sink,
2. Allow for heat losses through the furnace walls, and
3. Allow for the difference in flue gas temperature and effective gas radiating
temperatures, because an industrial furnace does not have perfect stirring.
The model gives substantially correct predictions of furnace performance for a
wide variety of furnaces and can be expected to give a good comparison between
fuels in a particular furnace.
A flame-to-sink energy exchange factor, (GS)R, for radiation heat transfer can be
defined by: A
for the case where the heat sink is well distributed. Here A^. is the total furnace
area, AS is the heat sink area, and €g and €g are the sink and flame emissivities,
respectively. Then a total energy exchange factor can be defined to include
radiative and convective heat transfer by:
(3)
where h is the convective heat transfer coefficient, O is the Stefan-Boltzman
b
constant, and T is the mean temperature between the sink and the flame.
A "reduced firing density" can be defined by:
where Hp is the total enthalpy input to the furnace and T^y is a "pseudo-
adiabatic" flame temperature defined as the temperature that the combustion
products would attain if the total enthalpy input were used to heat the combustion
products from ambient, TQ, at the mean heat capacity between TQ and the flame
temperature. Combining the equation above with an energy balance and denoting
the fraction of the input enthalpy transferred to the sink by TJ, we obtain:
99
-------
TJ'D'+T4= (1 + D'-T?')4 (5)
where T,' = T,(TAF - TQ)/TAF
T = Ts/TAF
and A' = A/T (6)
where A accounts for departures of the furnace from the well stirred approxima-
tion. The effective radiating temperature of the gas will differ from the mean
enthalpy temperature of the gases at the flue of the furnace by an amount A
which depends on the firing density and furnace geometry.
The first attempt at using the well stirred speckled-wall model ( A* = 0) was to
evaluate how it would agree with the IGT experimental data and the values calcu-
lated by B&W. The measured and calculated efficiencies are plotted in Figure 65
versus adiabatic radiance. Adiabatic radiance is defined as the total radiant
energy that would be emitted from the combustion products of a test fuel at the
test fuel's adiabatic temperature. Figure 65 again illustrates how accurately the
pilot-scale test furnace was able to model the radiant section of a utility boiler.
Thus to get efficiencies calculated with the well stirred speckled-wall model to
agree with the IGT test data would demonstrate the model's ability to predict
performance of a utility boiler radiant section.
To allow the speckled-wall model to be used from a completely theoretical posi-
tion with no experimental data needed, an iterative method was devised. A
combustion-products gas temperature is estimated at the exit of the radiant sec-
tion. From this temperature an emissivity is calculated using the Hottel-Sarofim
charts. (Any technique for calculating emissivities can be used. The Hottel-
Sarofim method was chosen based on the ease of calculation and the universal
availability of the gas emissivity charts.) This emissivity is plugged into Equation
2 to evaluate the flame-to-sink energy exchange factor. This permits the total
energy exchange factor, Equation 3, and the reduced firing density, Equation 4, to
be calculated. A reduced furnace efficiency can then be determined by Equation
5. This reduced efficiency can be turned into a true furnace efficiency using
100
-------
35-
25-
O
2
a
u
t—I
CK
Cn
W
15-
IGT EXPERIMENTAL DATA
B &W DESIGN CALCULATION
567
ADIABATIC RADIANCE,
10
(104)
11
Figure 65. Comparison of IGT experimental and B &W calculated furnace efficiencies
-------
Equation 6. Having the true furnace efficiency, the exit-gas temperature can be
determined using the heat content of the combustion-product gases. If the exit-
gas temperature derived from the speckled-wall model agrees with the tempera-
ture used to calculate the emissivity, then the solution is finalized. If the tem-
perature does not agree, then the estimated temperature is incremented, a new
emissivity is calculated and plugged into the speckled-wall model, and the itera-
tion continues. This procedure is followed until a satisfactory agreement between
estimated and calculated temperatures is achieved, that is, until the calculated
and estimated temperatures are sufficiently close that the emissivity is constant
over the temperature range between them.
Figure 66 compares the efficiencies calculated with the well stirred speckled-wall
model determined using the iterative method discussed above to the IGT measured
efficiencies as a function of adiabatic radiance. These calculated efficiencies are
proportional to the adiabatic radiance. The slope of this linear relationship of cal-
culated efficiencies to adiabatic radiance is smaller than the slope of the tine re-
lating measured efficiency to adiabatic radiance, indicating that the pilot-scale
test furnace deviated from a well stirred reactor. If the slopes of the experimen-
tal and theoretical lines were equal, then the deviation from a well stirred reactor
would be independent of adiabatic flame temperature and emissivity. However,
the slopes of the relationships illustrated in Figure 66 are not equal, meaning that
as the adiabatic radiance of the fuel gas becomes larger the deviation of the
furnace being modeled will get farther from a well stirred reactor. The speckled-
wall model can make allowances for this deviation by assigning A an appropriate
value. (Refer to Equation 5.) From reviewing the extreme left (Winkler air) and
right (Koppers-Totzek preheated to 4Z5°C) data points of the theoretical curve of
Figure 66, it can be concluded that no connection need be made for Winkler air (A
= 0); however, a value for A must be determined that will raise the 4Z5°C
preheated Koppers-Totzek oxygen fuel gas efficiency from 22.9% to 34.2%. The
value of A needed to get the calculated and measured efficiencies to agree can be
determined by evaluating the exit-gas temperature that would result in a furnace
efficiency of 34.2%. Having determined this temperature, 2592°F, it is used to
find the combustion-product gas emissivity, 0.359. The speckled-wall model is
then solved for A , 605°R. A linear relationship can now be developed between A
102
-------
35-i
25-
u
2
a
u
E
fn
u
15-
ITERATIVE CALCULATION WITH
WELL STIRRED SPECKLED-WALL MODEL
IGT EXPERIMENTAL DATA
\
4
567
ADIABATIC RADIANCE,
I
10
1
11
Figure 66. Comparison of experimental and well stirred
speckled-wall model calculated efficiencies
-------
and adiabatic radiance. This relationship is shown in Figure 67. This figure
illustrates that the departure of the furnace from a well stirred reactor is
dependent on the adiabatic radiance of the fuel gas. Using Figure 67, can be
determined for each of the fuel gases tested. This allows the efficiency for each
fuel gas to be reevaluated using this non-well stirred speckled-wall model. As
with the well stirred model, the iterative procedure is followed. The results of
this calculation are shown in Figure 68 and are listed in Table 10. The figure
illustrates that the iterative speckled-wall calculation can give results in good
agreement with experiment when properly modified to allow for deviation of the
furnace from perfect stirring. Table 11 lists the iteratively evaluated emissivities
as well as those calculated by B&W. To evaluate the effect emissivity has on
efficiency, the emissivity calculated by B&W for Winkler oxygen fuel gas was
substituted in the non-well stirred speckled-wall model to determine an exit-gas
temperature. The non-well stirred speckled-wall model using the B&W computed
emissivity yields an exit-gas temperature of 2463°F, or 33°F above that
determined using the iterative technique. Thus a 7.9% change in gas emissivity
will produce only a 1.4% change in exit-gas temperature.
The non-well stirred speckled-wall model provides a calculation tool that yields
good agreement with the more sophisticated techniques employed by B&W. Using
the iterative technique, accurate predictions of furnace performance can be made
for nonluminous fuels.
IN-THE-FLAME ANALYSIS
Detailed in-the-flame data were collected for natural gas and Winkler oxygen gas.
These data are to aid in quantitative modeling of large-scale turbulent diffusion
flames and to provide a qualitative guide in understanding and controlling NO
emission levels. The measurements included gas species concentrations and tem-
perature profiles. The detailed in-the-flame data are available upon request from
the EPA Project Officer (919)549-8411, extension 2236.
104
-------
o
Ul
600-i
500-
. 300-
200-
100-
SLOPE =92.65
10 ADIABATIC RADIANCE
I
8
ADIABATIC RADIANCE, e
. ^
A JD
I
9
10
I
11
Figure 67. Variation of well stirred approximation with adiabatic radiance
-------
35 n
25-
u
1
o
t-H
Cn
b
W
15-
IGT EXPERIMENTAL DATA
/\ NON-WELL STIRRED ITERATIVE
^ SPECKLED-WALL MODEL CALCULATION
Ox
X
,0'
6 7
ADIABATIC RADIANCE,
I
10
I
11
Figure 68. • Comparison of experimental and non-well stirred
speckled-wall model calculated efficiencies
-------
Table 10. COMPARISON OF MEASURED AND CALCULATED
FURNACE EFFICIENCIES
Fuel
Natural gas
Lurgi oxygen
Winkler oxygen
Koppers-Totzek oxygen
Wellman-Galushaair
Winkler air
Koppers-Totzek oxygen,
425°C
Koppers-Totzek oxygen,
425°C plus 15% FGR
6AcrTA4(104)
8.98
8.72
9.61
10.12
6.31
4.47
11.00
9.10
IGTa
28.6
25.2
26.4
33.1
19.0
10.8
34.2
28.2
B&Wb
26.3
—
25.1
—
—
7.8
—
—
SWMC
19.8
19.8
19.9
22.4
16.3
12.2
22.9
20.4
NSWMd
27.0
26.6
29.6
31.4
18.9
12.2
34.2
27.4
" Measured on IGT pilot-scale lesl furnace.
b Calculated by B&W for Teche Station Boiler (RB-455) using design calculation technique.
1 Calculated using the well stirred iterative speckled-wall model (A = 0°R).
^ Calculated using the non-well stirred iterative speckled-wall model (A evaluated using Figure 67).
Table 11. COMPARISON OF CALCULATED EMISSIVITIES
Fuel
Natural gas
Winkler oxygen
Winkler air
B&Wa
0.29
0.35
0.31
SWMb
0.30
0.35
0.34
NSWMC
0.33
0.38
0.34
11 Calculated by B&W for Teche Station Boiler (RB-455) using design calculation tech-
nique.
b Calculated using the well stirred iterative speckled-wall model technique
(A = 0°R).
c Calculated using the non-well stirred iterative speckled-wall model technique
(A evaluated using Figure 67).
107
-------
CONVERSION TABLE
ENGLISH TO SI METRIC CONVERSION FACTORS
To Convert From
lb/106 Btu
106 Btu/hr
Btu
psi
SCFH
ft/s
Inch
Feet
Feet2
Inches of water (pressure)
lb/ft3
gpm
Inch2
°F
°R
To
ng/J
MWt
J
Pa
m3/s
m/s
m
m
m2
Pa
kg/m3
m3/s
m2
°C
°C
Multiply By
4.299 E + 02
2.928751 E - 01
1.055 E E + 03
6.894757 E + 03
7.865790 E - 06
3.048000 E - 01
2.540000 E-02
3.048000 E - 01
9.290304 E-02
2.4884 E +01
1.601846 E + 01
6.309020 E - 05
6.451600 E - 04
t°C= (t°F-32)/1.8
t°C = [(5/9)t°R]-273.15
J = Joule
g = gram
s = second
MWt = megawatts thermal
gpm = gallons (U.S. liquid)/minute
Pa = Pascal
m = metre
k =kilo(103)
n = nano (10"9)
M = mega (106)
C = Celsius
F = Fahrenheit
R = Rankine
psi = pounds per square inch
SCFH = standard cubic feet per hour
108
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/7-77-094a
2.
3. RECIPIENTS ACCESSION-NO.
4. TITLE AND SUBTITLE
Burner Design Criteria for NOx Control from Low-Btu
Gas Combustion; Volume I. Ambient Fuel Tempera-
ture
5. REPORT DATE
August 1977
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Donald R. Shoffstall
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Applied Combustion Research
Institute of Gas Technology
IIT Center, 3424 South State Street
Chicago, niinois 60616
10. PROGRAM ELEMENT NO.
EHE624a
11. CONTRACT/GRANT NO.
68-02-1360
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
3. TYPE OF REPORT AND F
Final; 9/75-10/76
PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/13
is. SUPPLEMENTARY NOTES IERL-RTP project officer for this report is David G. Lachapelle,
Mail Drop 65, 919/514-2236.
is. ABSTRACT
gives results of a research program initiated to characterize
problems associated with retrofitting existing utility boilers with low- and medium-Btu
gases produced using commercially available coal conversion processes. All experi-
mental results were gathered from a pilot-scale furnace fired with a movable -vane
boiler burner at a heat input of 0. 66 MW (2. 25 million Btu/h). The synthetic gases
tested, ranging in heating value from 3.7 to 11.2 MJ/cu m (100 to 300 Btu/SCF), were
produced using a natural gas reformer system. Data were collected to permit a
comparison between natural gas and the synthetic gases in the areas of flame stability,
flame length, flame emissivity, furnace efficiency, and NOx emissions. Flame sta-
bility was found to be very sensitive to fuel jet velocity. An injection velocity of 30. 5
m/s (100 ft/s) was found to be optimum. Flame length decreased with increasing
movable-vane angle (swirl of the combustion air): flames of the synthetic gases tested
generally were shorter than those of natural gas. Good agreement was obtained be-
tween measured and calculated flame emissivities. Some boiler modifications would be
necessary to maintain rating when burning gases of less than 7. 5 MJ/cu m (200 Btu/
SCF) heating value. NO emissions were ordered by adiabatic flame temperature. The
NO emissions data yielded an activation energy of 153 kcal/mole compared to kinetic
model predictions of 135 kcal/mole. _
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air Pollution
Nitrogen Oxides
Aerodynamics
Manufactured Gas
Combustion Control
Boilers
Burners
Design
Temperature
Coal Gasification
Natural Gas
Flames
Swirling
Emissivitv
Air Pollution Control
Stationary Sources
Low-Btu Gases
Medium-Btu Gases
13B
07B
20D
21D
21B
13A
07A
18. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
119
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
109
------- |