-------
coals would exceed even the 1971 Standards of Performance for Large Steam
Generators (NSPS).  Fortunately, only a fraction of the fuel nitrogen is
converted to NO  for both oil and coal firing, as shown in Figure 4-2
               A
(Reference 4-23).  Furthermore, the figure indicates that fuel nitrogen
conversion decreases as nitrogen content increases.  Thus, although fuel
NO  emissions undoubtedly increase with increasing fuel nitrogen content,
  /\
the emissions increase is not proportional.   In fact, recent data indicate
only a small increase in NO  emissions as fuel nitrogen increases
                           A
(Reference 4-24).  From observations such as  these, the effectiveness of
partial fuel denitrification as a NO  control method seems doubtful.
                                    A
       Although  the precise mechanism by which fuel nitrogen is converted to
NO  is not understood, certain  aspects are clear, particularly for coal
  ^
combustion.  In  a large pulverized coal-fired utility boiler, the coal
particles are conveyed by an airstream into  the hot combustion chamber,
where they are heated at a rate in excess of  10  K/s.  Almost immediately
volatile  species, containing some of  the coal bound nitrogen, vaporize  and
burn homogeneously, rapidly  (~10 ms)  and probably  detached  from  the  original
coal particle.   Combustion of  the remaining  solid  char  is heterogeneous and
much slower  (~300 ms).
       Figure 4-3 summarizes what may happen  to fuel nitrogen during this
process  (Reference 4-25).  In  general, nitrogen evolution parallels
evolution of the total volatiles, except during the  initial  10  to  15 percent
volatilization  in which  little nitrogen  is  released  (Reference  4-26).  Both
total mass  volatilized and total nitrogen  volatilized  increase  with  higher
pyrolysis temperature; the nitrogen  volatilization increases more  rapidly
than that of the total mass.   Total  mass  volatilized  appears to be a
stronger function of  coal  composition than  total  nitrogen volatilized
(Reference  4-27).   This  supports the relatively small  dependence of  fuel
N0x  on coal  composition  observed  in  small  scale testing (References  4-18
and  4-28).
       Although  there is  not absolute agreement on how the  volatiles
separate into  species,  it appears  that about half the  total volatiles and 85
percent  of  the  nitrogeneous  species  evolved react to form other reduced
species  before being  oxidized.  Prior to oxidation,  the devolatilized
nitrogen may be converted to a small number of  common, reduced
                                     4-7

-------
                               -  5
00

X
o
z:
f^
° 100-
c
o
W1
1 80-
c
o
o
« 60'
£
*4—
M 40

20

r ,.. .flipVnn, ILiqnite coal 1 1 BUumi
Calif i c «.- 1 1 1 i — j 	 _i
i/ iSuh-hi tumi nous coal


Coal
0 Pereira, et al. (1974)
.£} ^Pershing, et al . (1973)
n A McCann (1970)
* Q O ^ Jonke (1970)
V ^ • Bituminous M.I.T. (1975)
»££)£>. ^ Lignite M.I.T. (1975)
^^^ i^^ ^^
tf j »\ 1*1
£°|a 

0 0.2 0.4 0.6 0.8 1.0 1.2
nous coal ,
I Residual oil ^
fupl oi 1 #4 A R fupl ni 1 _
Data Sources Oil
^ Flagan & Appleton (1974)
A Hazard (1973)
Q Turner & Siegmund (1972
O Fenimore (1972)
O Turner et al. (1972)
V Martin & Berkau (1972)
I> Pershing, et al. (1973)
<«J Martin, et al. (1970)


^^
w ^
^k ^^L
V ^

1.4 1.6 1.8 2.0
                                               Weight % N in fuel
                     Figure 4-2.  Conversion of fuel N in practical combustors (Reference 4-22).

-------
                Volatile fractions
              (Hydrocarbons. RN etc
                                                      Ash
                                                    virtually
                                                    nitrogen
                                                      free
Figure  4-3.
Possible  fate of fuel  nitrogen  contained  in  coal
particles during combustion (Reference 4-25).
                               4-9

-------
 intermediates, such  as HCN  and NH-,  in  the  fuel-rich  regions  of  the flames.
 The existence of a set of common reduced  intermediates  would  explain  the
 observations that the form  of the  original  fuel  nitrogen  compound  does  not
 influence  its conversion to NO (e.g., References 4-20,  4-29).  More recent
 experiments suggest  that HCN is the  predominant  reduced intermediate
 (Reference 4-30).  The reduced intermediates  are then either  oxidized to NO,
 or converted to N~ in the postcombustion  zone.  Although  the  mechanism  for
 these conversions is not presently known, one proposed  mechanism postulates
 a role for NCO (Reference 4-31).
       Nitrogen retained in the char may  also be oxidized  to  NO, or reduced
 to N2 through heterogeneous reactions occurring  in the  postcombustion
 zone.  However, it is clear that the conversion of char nitrogen to NO
 proceeds much more slowly than the conversion of devolatilized nitrogen.  In
 fact, based on a combination of experimental  and empirical modeling studies,
 it is now believed that 60 to 80 percent  of the fuel NO results from
                                                        /\
 volatile nitrogen oxidation (References 4-26, 4-32).  Conversion of the char
 nitrogen to NO is in general lower, by factors of two to three,  than
 conversion of total  coal nitrogen  (Reference 4-29).
       Regardless of the precise mechanism of fuel NO   formation,  several
                                                     /\
 general trends are evident, particularly  for coal combustion.  As  expected,
 fuel nitrogen conversion to NO is  highly  dependent on the  fuel/air ratio for
 the range existing in typical combustion  equipment, as  shown  in  Figure 4-4.
Oxidation of the char nitrogen is  relatively insensitive to fuel/air
changes, but volatile NO formation is strongly affected by fuel/air ratio
changes.
       In contrast to thermal NO , fuel NO  production  is  relatively
                                A         X
 insensitive to small  changes in combustion zone temperature (Reference 4-29).
Char nitrogen oxidation appears to be a very weak function of temperature,
and although the amount of nitrogen volatiles appears to increase  as
temperature increases,  this is believed to be partially offset by  a decrease
 in percentage conversion.  Furthermore, operating restrictions severely
 limit the magnitude of actual temperature changes attainable  in current
systems.
       As described above,  fuel NO  emissions are a strong function of
                                  A
fuel/air mixing.   In general, any change  which increases the mixing between
the fuel and air during coal devolatilization will dramatically increase

                                    4-10

-------
OJ
X
o

c
0)
C7l
o
o


V.
c
o
u
100-


 90


 80


 70


 60-


 SO


 40'

 30-


 20-
    101

     0-
                                                    Wall  temp 1500 K
                                                    Flame temp 1600 K


                                                   A Lignite 75-90 tim
                                                   A Lignite 38-45 nm
                                                   Q Bituminous 75-90 um
                                                   • Bituminous 38-45 nm
A
      0
                                234

                                     Fuel equivalence ratio

                              (Inverse of  stoichiometric  ratio)
                Figure 4-4.   Conversion of nitrogen in coal to NO  (Reference 4-23).

-------
volatile nitrogen conversion and  increase fuel NO  .   In contrast, char NO
                                                 A
formation is only weakly dependent on  initial mixing.
       From the above modifications, it  appears  that,  in  principle, the best
strategy for fuel NO  abatement combines low excess  air (LEA) firing,
                    A
optimum burner design, and two stage combustion.  Assuming suitable stage
separation, low excess air may have little effect on  fuel N0x, but it
increases system efficiency.  Before using LEA firing, the need to get good
carbon burnout and low CO emissions must be considered.
       Optimum burner design ensures locally fuel-rich conditions during
devolatilization, which promotes  reduction of devolatilized nitrogen to
Np.  Two-stage combustion produces overall fuel-rich  conditions during the
first 1 to 2 seconds and promotes the  reduction  of NO to  N2 through
reburning reactions.  High secondary air preheat may  also be desirable,
because it promotes more complete nitrogen devolatilization in the fuel-rich
initial combustion stage.  This leaves less char nitrogen to be subsequently
oxidized in the fuel-lean second  stage.  Unfortunately, it also tends to
favor thermal NO formation, and at present there is  no general agreement on
which effect dominates.
4.1.3  Summary of Process Modification Concepts
       In summary of the above discussion, both  thermal and fuel NO  are
                                                                   A
kinetically or aerodynamically limited in that their  emission rates are far
below the levels which would prevail  at equilibrium.  Thus, the rate of
formation of both thermal and fuel NO  is dominated by combustion
                                     A
conditions and is amenable to suppression through combustion process
modifications.   Although the mechanisms are different, both thermal and fuel
NO  are promoted by rapid mixing  of oxygen with  the fuel.  Additionally,
  A
thermal NOX is  greatly increased  by long residence time at high
temperature.  The modified combustion conditions and  control concepts which
have been tried or suggested to combat the formation  mechanisms are as
follows:
       t   Decrease primary flame zone Op level  by
           —  Decreased overall  0« level
           —  Controlled mixing  of fuel and air
           --  Use of fuel-rich primary flame zone
                                    4-12

-------
       •   Decrease  time  of  exposure  at  high  temperature  by
           —   Decreased  peak  temperature:
               —  Decreased adiabatic flame  temperature  through dilution
               --  Decreased combustion  intensity
               --   Increased flame cooling
               --  Controlled mixing  of  fuel  and air or use of fuel-rich
                   primary flame zone
           —   Decreased primary flame zone residence time
       •   Chemically reduce NO  in postflame region by
                               ^
           ~   Injection of reducing  agent (e.g., MM.,)
       Table 4-2 relates these control concepts to combustion process
modifications  applicable to utility boilers.   The process modifications are
categorized according to their role in the control development  sequence:
operational adjustments, hardware modifications of existing equipment or
through factory installed controls, and major redesigns  of new  equipment.
The controls for decreased 02  are also generally effective for  peak
temperature reduction but have not been repeated.  The following  subsections
briefly review the  status of each of the applicable  control techniques
applied to utility  boilers.
4.2     STATE-OF-THE-ART  CONTROLS
        Based on the general principles  discussed  above for the  suppression
of NO   emissions by process modifications, there  are several  control
     A
techniques that may be used  singly or conjunctively on  utility  boilers.
These  techniques include low  excess  air firing, biased  burner firing,
burners out of service,  overfire  air,  low  NO  burners,  flue gas
                                             A
recirculation, and  reduced  firing rate.   These methods for controlling NOX
may  be used on existing  boilers although  modifications to the units may be
required.  Tables 4-3 through 4-10 give the  average NO  reduction
                                                       A
 achievable with the various control  techniques, compiled from the data base
 of test results and test selection procedures discussed in Section 5.   It
 should be noted that the data base is not complete in that only those  tests
 that were well characterized are included; i.e., such boiler design and
 operating variables as number of burners, burner stoichiometry, direct input
 per active burner,  surface heat release rate, etc., were reported (see
 Section 5.2.2).

                                     4-13

-------
TABLE 4-2.  SUMMARY OF COMBUSTION PROCESS MODIFICATION CONCEPTS
Combustion
Conditions
Decrease
primary
flame zone
0? level
(
Decrease
peak
flame
temperature
Chemically
reduce NOX
in post
flame region
Control
Concept
Decrease overall
02 level
Delayed mixing
of fuel and air
Primary fuel-
rich flame
zone
Decrease
adiabatic flame
temperature
Decrease
combustion
intensity
Increased flame
zone cooling/
reduce residence
time
Inject reducing
agent
Effect on
Thermal NOX
Reduces 0 rich,
high NO pockets
in the flame
Flame cooling and
dilution during
delayed mixing
reduces peak
temperature
Flame cooling in
low 0 , low
temperature primary
zone reduces peak
temperature
Direct suppression
of thermal NOX
mechanism
Increased flame
zone cooling yields
lower peak
temperature
Increased flame
zone cooling yields
lower peak
temperature
Decomposition

Effect on
Fuel NOX
Reduces exposure
of fuel nitrogen
intermediaries
to 02
Volatile fuel N
reduces to N? in
the absence of
oxygen
Volatile fuel N
reduces to N2 in
the absence of
oxygen
Ineffective
Minor direct
effect; indirect
effect on mixing
Ineffective
Decomposition
Primary Applicable Controls
Operational
Adjustments
Low excess air
firing
Burner
adjustments
Burners out of
service; biased
burner firing
Reduced air
preheat
Load reduction
Burner tilt

Hardware
Modification
Flue gas recirculation
(FOR)
Low NOX burners
Overfire air ports
Water injection, F6R


Ammonia injection
possible on some units
Major
Redesign

Optimum burner/
firebox design
Burner/firebox
design for two
stage combustion

Enlarged firebox,
increased burner
spacing
Redesign heat
transfer surfaces,
firebox
aerodynamics
Redesign convective
section for NH3
injection

-------
                                  TABLE  4-3.   AVERAGE  N0x REDUCTION WITH  LOW  EXCESS  AIR  FIRING  (LEA)
Equipment
Type
Tangential
Opposed Wall

Single
Wall


All Boilers


Fuel
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
All
Fuels
Number
of
Boilers
Tested
11
-
1
5
4
6
7
(2)
4
3
(1)
23
8
10
41
Baseline
Stoichionetry
to Active Burners
(percent)
124
~
117
126
120
115
123
(134)b
120
117
(124)
124
120
116
120
HOX Emissions
(ppm dry 9 3» 02)
459
--
340
746
357
717
624
(1338)
409
418
(992)
609
383
492
495
Low Excess Air (LEA)
Stoichiometry
to Active Burners
(percent)
116
-
113
118
113
110
114
(118)
112
108
(112)
116
115
110
114
NO Emissions
(ppm dry * 31 02)
373
-
245
660
290
600
522
(1325)
315
356
(931)
522
302
400
408

Average
NO. Reduction
(percent)
19
-
28
12
19
16
16
(1)
23
15
(6)
16
21
20
19
Maximum NOX
Reduction
Reported
(percent)
42
-
28
23
30
33
25
(3)
26
15
(6)
30
28
25
28
en
            'Boiler load at or above 80 percent MCR.  For individual  tests, corresponding baseline and controlled loads were nearly identical.
            lumbers in parentheses refer to boilers originally designed for coal firing with wet bottom furnaces.

-------
                TABLE  4-4.   AVERAGE NO   REDUCTION  WITH  BURNER OUT  OF  SERVICE  (BOOS)'
Equipment
Type
Tangential
Opposed Wall
Single
Wall
All
Boilers
Fuel
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
All
Fuels

Number
of
Boilers
Tested
7
--
1
4
1
4
8
(1)
3
3
19
4
8
31
Baseline
No. of
Burners
Firing
32 u
(16-56)b
--
8
40 u
(24-54)b
24
26b
(16-36 )b
16
(24 )C
16 .
(12-24)"
16
(12-16)b
28
(16-56)b
20
( 12-24 )b
16 K
(8-36)b
20
(8-56)b
Stoichiometry
to Active
Burners
(percent)
121
—
112
122
107
115
123
(134)C
119
117
122
113
115
117
NOX
Emissions
(ppm dry
9 sx o2)
462
--
146
670
442
674
618
(1196)°
425
418
583
433
412
4/6
Burners Out of Service (BOOS)
Percent
Burners
on Air
Only
17
--
NA
16
33
28
19
(33)C
18
22
17
25
25
22
Stoichiometry
to Active
Burners
(percent)
98
--
86
102
73
84
97
(89)C
95
89
99
84
86
90
NOX
Emissions
(ppm dry
9 3X 02)
293
--
146
522
292
290
412
(577)C
256
214
409
274
217
300
Average NOX
Reduction
(percent)
37
--
0
22
34
57
33
(52)C
40
49
31
37
35
34
Maximum NOX
Reduction
Reported
(percent)
56
-
0
46
34
61
48
(52)C
48
69
50
41
43
45
jjBoiler load  at or above 80 percent MCR.  For individual tests, corresponding baseline and controlled loads were nearly identical.
"Range in number of burners firing
lumbers  in parentheses refer to boilers originally designed for coal firing with wet bottom furnaces.

-------
                                   TABLE 4-5.   AVERAGE N0¥ REDUCTION WITH OVERFIRE AIR (OFA)'
-p.
I
Equipment
Type
Tangential
Opposed Wall

Single
Wall

Fuel
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Number
of
Boilers
Tested
6
-
—
-
S
2
-
-
—
Baseline
Stoichiometry
to Active
Burners
(percent)
129
-
—
—
118
114
~
—
--
NOX
Emissions
(ppm dry
33X02)
454
—
-
—
376
928
—
-
—
Overfire Air (OFA)
Stoichioroetry
to Active
Burners
(percent)
105
-
-
-
96
99
—
-
~
Furnace
Stoi chl erne try
(percent)
122
~
~
~
118
112
-
-
-
NOX Emissions
(ppm dry
9 3X02)
311
~
--
--
287
378
--
--
-
Average NOX
Reduction
(percent)
31
-
--
-
24
59
—
-
~
Maximum NOX
Reduction
Reported
(percent)
41
--
--
--
30
66
-
--
-
              aBo11er load at or above 80 percent NCR.  For individual  tests, corresponding baseline  and controlled loads were nearly identical.

-------
                          TABLE 4-6.   AVERAGE  N0x  REDUCTION WITH FLUE  GAS  RECIRCULATION (FGR)'
Equipment
Type
Tangential
Opposed Wall
Single
Wall
Fuel
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Number
of
Boilers
Tested
-
-
1
1
1
--
--
-
1
Baseline
Stoichiometry
to Active
Burners
(percent)
--
-
117
128
122
-
--
-
106
NOX
Emissions
(ppm dry
3 3* 02)
--
--
340
855
304
-
--
-
470
Overfire Air (OFA)
Stoichiometry
to Active
Burners
(percent)
--
-
115
127
126
-
--
-
107
FGR
( percent )
--
-
23
15
11
--
--
--
11
NOX Emissions
(ppm dry
S 3X 02)
-
-
135
735
263
—
--
-
307
Average NOX
Reduction
(percent)
—
—
60
17
13
—
—
--
35
Maximum NOX
Reduction
Reported
(percent)
-
--
60
17
13
—
—
—
35
oo
            aBoiler load at or above 80 percent MCR.  For individual tests, corresponding baseline and controlled loads were nearly identical.

-------
                        TABLE 4-7.   AVERAGE  N0x REDUCTION  WITH  REDUCED  FIRING RATE
Equipment
Type
Tangential
Opposed Hall

Single
Wall


All
Boilers


Fuel
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
ias
Coal
011
Nat
Gas
All
uels
Nuaber
of
Boilers
Tested
7
-
1
4
4
5
2
(2)
3
2
(1)
13
7
8
28
Baseline (SOX HCR or Above)
Firing
Rate
(percent
MCR)
93
—
100
93
98
98
92
(90)
98
97
(98)
93
98
99
97
Stolchiometry
to Active
Burners
(percent)
112
—
117
131
118
115
125
(133)»
119
118
(115)
126
119
117
120
*>x
Emissions
(ppn dry
8 3*02)
462
~
340
825
362
651
651
(1338)
425
442
(992)
646
393
478
506
Reduced Load
Firing
Rate
(percent
MCR)
64
—
75
70
61
57
67
(54)
53
35
(59)
67
57
55
60
Stoichiooetry
to Active
Burners
(percent)
127
—
135
136
121
115
130
(138)
119
117
(131)
131
120
122
124
NOX
Emissions
(ppm dry
« 3X 0Z)
408
-
332
758
249
269
496
(990)
296
125
(522)
554
272
242
356

Average NOX
Reduction
(percent)
12
--
2
8
31
59
24
(26)
30
72
(47)
14
31
44
30
Maximum NOX
Reduction
Reported
(percent)
25
--
32
18
48
64
25
(33)
45
82
(47)
23
47
59
43
aNumbers In parentheses refer to boilers originally designed for coal firing with wet bottom furnaces.

-------
                              TABLE 4-8.   AVERAGE NOX  REDUCTION  WITH  OFF STOICHIOMETRIC COMBUSTION
                                             AND FLUE GAS RECIRCULATION  (OSC  AND FGR)a
Equipment
Type
Tangential
Opposed Wall
Single
Wall
Fuel
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Number
of
Boilers
Tested
-
-
1
1
1
-
-
2
1
Baseline
Stolen ioroetry
to Active
Burners
(percent)
—
—
117
128
122
~
~
118
106
NO,
Emissions
(ppm dry
9 3X 02)
-
—
340
781
304
--
-
355
470
OSC and FGR
Type of
OSC
-
-
BOOS
BOOS
OFA
—
-
BBF
BOOS
BOOS
Stoichiometry
to Active
Burners
(percent)
-
-
75
99
97
--
--
91
75
FGR
(percent)
--
—
21
19
11
—
—
14
12
NOX
Emissions
(ppm dry
S 3X 02)
-
--
105
453
247
-
-
154
115
Average NOX
Reduction
(percent)
—
—
69
42
19
-
—
57
76
Maximum NOX
Reduction
Reported
(percent)
-
--
69
42
19
—
--
59
76
-p*

o
            aBoiler load at or  above 80 percent MCR.  For individual  tests, corresponding baseline and controlled loads were nearly identical.

-------
            TABLE 4-9.   AVERAGE  NOX REDUCTION WITH  REDUCED FIRING RATE AND OFF  STOICHIOMETRIC COMBUSTION
Equipment
Type
Tangential
Horizontally
Opposed Wall
Single
Wall

All
toilers

Fuels
Fuel
Coal
Oil
Nat
Gas
Coal
011
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
All
Number
of
Boilers
Tested
8
-
-
3
4
6
4
(2)
3
2
(1)
15
7
3
30
Baseline
Firing
Rate
(percent
NCR)
93
—
-
93
99
100
90
98
97
(98)
92
99
99
97
Stoichiometry
to Active
Burners
(percent)
122
-
--
129
118
115
124
(133)«
120
118
(125)
125
119
117
120
NOX
Emissions
(ppm dry
9 3t 02)
453
--
-
820
362
717
663
(1338)
426
442
(992)
645
394
579
539
Low Load and OSC
Firing
Rate
(percent
MCR)
61
-
—
73
64
58
73
(59)
56
35
(H)
69
60
31
1)3
Type of
OSC
BOOS
OFA
-
-
BOOS
BOOS
OFA
BOOS
OFA
BOOS
BBF
BOOS
BBF
BOOS
BOOS
OFA
BBF
BOOS
OFA
BOOS
OFA
BBF
BOOS
OFA
Stolen iometry
to Active
Burners
(percent)
95
—
—
102
117
88
99
(91)
97
93
(102)
99
107
91
99
NOX
Emissions
(ppn dry
9 31 02)
248
—
—
634
177
148
381
(887)
228
78
(641)
421
202
113
245

Average NOX
Reduction
(percent)
45
-
--
23
51
79
43
(34)
46
82
(35)
37
49
80
b5
Maximum
NOX
Reduction
Reported
(percent)
62
--

32
67
89 j
50
(55)
59
87
(35) !
46
!
63
i
88
66 j
!
ro
       aNumbers  in parentheses refer to boilers originally designed for coal firing with wet bottom furnaces.

-------
              TABLE 4-10.   AVERAGE NOX REDUCTION WITH LOAD REDUCTION, OFF STOICHIOMETRIC COMBUSTION

                           AND FLUE GAS RECIRCULATION
Equipment
Type
Tangential
Opposed Wall
Single
Wall
All
Boilers
Fuel
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
All
Fuels
Number
of
Boilers
Tested
-
-
-
-
3
2
-
2
I
-
5
4
9
Baseline
Firing
Rate
(percent
(CR)
-
-
-
-
99
100
-
98
100
--
99
100
100
Stoichiometry
to Active
Burners
(percent)
--
--
--
--
118
113
--
118
110
~
118
112
115
NOX
Emissions
(ppn dry
0 3X 02)
--
--
--
-
398
945
--
355
421
--
376
683
530
Controlled/Low Load and OSC and FGR
Firing
Rate
(percent
MCR)
--
--
--
--
46
43
--
62
65
--
54
54
54
Type of
OSC
--
--
--
--
BOOS
OFA
BOOS
OFA
--
BOOS
BOOS
--
BOOS
OFA
BOOS
OFA
BOOS
OFA
Stoichiometry
to Active
Burners
(percent)
--
--
--
--
87
90
—
92
81
--
90
86
88
FGR
(percent)
--
-
-
-
39
27
--
30
20
--
35
23
29
NOX
Emissions
(ppm dry
0 3* 02)
-
-
-
—
194
130
—
152
171
—
173
150
162
Average NOX
Reduction
(percent)
-
-
—
--
56
87
-
57
59
-
57
73
66
Maximum NOX
Reduction
Reported
(percent)
-
-
--
-
59
90
--
57
83
—
58
87
73
I
ro
ro

-------
       The average NO  reductions reported in Tables 4-3 through 4-10 were
                     y\
calculated in the following manner.  First, reductions obtained from all
tests on each particular boiler were arithmetically averaged.  Only tests
with the same NO  control technique were used.  Next, these average NO
                A                                                     A
reductions were again arithmetically averaged using all the boilers within
the same firing type/fuel classification.  All other numerical table
entries, such as burner stoichiometry, firing rate, etc., were calculated in
a similar manner.
       It should be noted that baseline emission data vary occasionally
between tables for the  same firing type/fuel  classification.  The reason for
this variation is that  only the baseline tests and  the  corresponding
controlled tests with the particular control  technique  under  consideration
were used in the averaging procedure described above.   For example,  if  NOV
                                                                          /\
emissions from boiler "A" firing  coal were controlled  only with  the
technique of low excess  air, then  the baseline data from  that boiler  would
be  used only to  calculate the  average emissions  reported  in  Table  4-3,  and
not used  in  deriving  average baseline emissions  in  Tables 4-4 through 4-10,
for other techniques.   NO  reductions reported  in  these tables represent
                         A
values  typical  of what  can be  expected  when  control techniques are
implemented.   Descriptions of  these  control  techniques follow.
4.2.1   Low Excess Air (LEA)
        Reducing  the  excess  air level  in the  furnace has generally been found
to  be  an  effective method  of NO   control.   In this technique, the
                                /\
 combustion  air is  reduced  to  the minimum amount required for complete
 combustion,  maintaining acceptable furnace cleanliness and steam temperature.
 With less oxygen available in  the flame zone, both thermal and fuel NO
                                                                       A
 formation are  reduced (Reference 4-22).  In addition,  the reduced airflow
 lowers the  quantity  of flue gas released resulting in  an improvement in
 boiler efficiency.
        Low  excess air firing  is usually the first NO   control technique
                                                     A
 applied.   It may be  used with virtually all fuels  and  firing methods.
 However,  furnace slagging and tube wastage considerations may limit  the
 degree of application  (Reference 4-22).  Low excess air may  also be  employed
 in combination with  the other NO  control methods  (Reference 4-33).
                                 X
                                     4-23

-------
       Many units use excess air for control  of  steam  temperature,
especially at lower loads, often as an  alternative to  flue gas
recirculation.  Reducing the excess air  levels on these  units would tend to
lower the outlet steam temperature and  thus reduce cycle efficiency unless
the improvement in boiler efficiency is  enough to compensate for the lower
steam temperature (Reference 4-34).
       In low excess air firing, there  is often  a greater burden on
operating personnel.  The attempt to optimize the excess air level requires
close monitoring of flue gas 0- and CO  analyzers.  In  coal firing, the
operator must also check the furnace periodically for  excessive slag
deposits.  Accurate flue gas analyzers will often need to be purchased if
not already installed.
       As shown in Table 4-3, Average N0¥ Reduction with Low Excess Air
                                         /\
Firing, low excess air firing results in an average NO   reduction of
                                                       /\
16 percent for coal, 21 percent for oil, and  20  percent  for natural gas
firing.
4.2.2  Off Stoichiometric Combustion (OSC)
       Off Stoichiometric, or staged combustion  seeks  to control NO  by
                                                                   A
carrying out initial combustion in a primary, fuel-rich, combustion zone,
then completing combustion, at lower temperatures, in  a  second, fuel lean
zone.  In practice, OSC is implemented through biased  burner firing (BBF),
burners out of service (BOOS), or overfire air injection (OFA).
4.2.2.1  Biased Burner Firing (BBF), Burners Out of Service (BOOS)
       Biased burner firing consists of  firing the lower rows of burners
more fuel rich than the upper rows of burners.   This may be accomplished by
maintaining normal air distribution to the burners while adjusting fuel flow
so that a greater amount of fuel enters  the furnace through the lower rows
of burners than through the upper rows of burners.  Additional air required
for complete combustion enters through the upper rows  of burners which are
firing air rich.
       In the burners out of service mode, individual  burners, or rows of
burners,  admit air only.   This reduces the airflow through the fuel
admitting or active burners.  Thus the burners are firing more fuel rich
than normal, with the remaining air required for combustion being admitted
through the inactive burners.
                                    4-24

-------
       These methods reduce NO  emissions by reducing the excess air
                              /v
available in the firing zone.  This reduces fuel and thermal NO  formation.
                                                               ^
These techniques are applicable to all fuels and are particularly attractive
as control methods for existing units since few, if any, equipment
modfications are required (References 4-33 and 4-35).   In some cases,
however, derating of the unit may be required if there  is too limited extra
firing capability with the active burners.  This is most likely to  be a
problem with pulverized coal units without spare pulverizer capacity.
       Monitoring flue gas composition, especially Op and CO concentrations,
is very important when employing these combustion modifications for N0x
control.  Local reducing atmospheres may  cause  increased tube wastage when
firing coal  and high sulfur  oils.  They may also cause  increased  furnace
slagging when  burning  coal because of the lower  ash  fusion  temperature
associated  with reducing atmospheres  (References 4-34 and 4-36).   In
addition,  it is important  to closely monitor flue  gas,  excess  air,  and  CO to
avoid  reducing  boiler  efficiency through  flue gas  heat  and  unburned
combustible losses,  and  to prevent unsafe operating  conditions  caused  by
incomplete  combustion.   For  these  reasons, accurate  flue gas  monitoring
equipment and  increased  operator monitoring of  furnace  conditions are
required  with  these combustion modifications.
        As shown in  Table 4-4, burners out of service firing results in an
 average NO  reduction  of 31  percent  for  coal,  37 percent for  oil, and  35
           X
 percent for natural gas  firing.  A typical burners out of service  pattern is
 shown  in Figure 4-5(a).
 4.2.2.2    Overfire Air  (OFA)
        The overfire air  technique for NO  control involves firing  the
 burners more fuel rich than normal while  admitting the remaining combustion
 air through overfire air ports ,or an idle top row of burners.
        Overfire air is very effective for NO  reduction and may  be  used
                                             /\
 with all fuels.  However,  there is an increased potential for furnace tube
 wastage due to local reducing conditions  when firing coal  or high  sulfur
 oil.  There is also a greater tendency for slag accumulation  in  the furnace
 when firing coal (References 4-22, 4-35  through 4-37).  In addition, with
 reduced airflow  to the  burners, there may be reduced mixing of the fuel  and
                                      4-25

-------
                                             WINDIOX
                                                                            IICONOAIT
                                                                            All NOZ21IS
                                     SICONDAlr All DAMMIS
                                          IICONOAIT AM

                                        OAMfll DIIV1 UNII
   O Active  burners

   )9C Burners  admitting air only
 a.  Typical  burners  out of
     service  arrangement
     opposed  fired unit
                                                                          — COAL NOZ2HS
    b.  Typical overfire  air system for
        tangential  fired  unit (Reference 4-21)
Burners
                                                                        Air
                                                           Forced draft fan
                Apportioning
                dampers
Flue gas recirculating
fan
                  c.  Typical  flue  gas  recirculation system for NO   control
                                                                   /\
     Figure 4-5.  Typical  arrangements for (b) overfire  air,  (a)  burners
                  out of  service,  and (c) flue gas recirculation.
                                       4-26

-------
air.   Thus,  additional  excess air may be required to ensure complete
combustion.   This may result in a decrease in efficiency (References 4-35
and 4-37).
       Overfire air is  more attractive in original designs than in retrofit
applications for cost considerations.  Additional duct work, furnace
penetrations, and extra fan capacity may be required.  There may be physical
obstructions outside of the boiler setting making installation more costly.
Or, there may also be insufficient height between the top row of burners and
the furnace exit to permit the installation of overfireair ports and the
enlarged combustion zone created by  the staged combustion technique
(Reference 4-35).
       As shown  in Table 4-5, the limited data indicate that with  overfire
air, NO  reductions of about 31  percent for coal, 24  percent for  oil,  and
       A
59 percent for  natural gas  are possible.  A typical  overfire air  system is
shown  in Figure  4-5(b).
4.2.3  Low NOW  Burners (LNB)
             y\
       Several  utility boiler manufacturers  have been active  in  the
development  of  new burners  designed  to  reduce NO  emissions  from
                                                 X
coal-fired  units.  Although the  techniques  of low excess  air  and off
stoichiometric  (staged)  combustion  have been  shown  to be  effective in
reducing NO   levels, there  has  been  some  concern as to potential  increased
            /\
 slagging and corrosion with OSC  operation.   Furnaces fired with certain
Eastern  U.S.  bituminous  coals  with  high sulfur contents may be especially
 susceptible to corrosion attack  under reducing atompsheres.  Local reducing
 atmosphere  pockets may exist under  off  stoichiometric operation.  The
 problem  may be further  aggravated by slagging since  slag generally fuses at
 lower  temperatures under reducing conditions.  The sulfur in the molten slag
may then readily attack  the tube walls.  Faced with  these potential problems
 and stricter N0x NSPS,  manufacturers are developing  and marketing  low
 NOX burners which permit staging at the burners  themselves, away  from  the
 water wells, thus minimizing the potential corrosion  and  slagging  problems
 associated with OSC operation.
        Most low NO  burners designed for utility boilers  control  NO   by
                   ^                                                 X
 reducing flame  turbulence, delaying fuel/air mixing,  and  establishing
 fuel-rich zones where combustion initially takes place.   This represents  a
 departure from  the usual burner design procedures  which  promote  high
                                     4-27

-------
turbulence, high intensity, rapid combustion flames.  The longer, less
intense flames produced with low NO  burners result  in  lower flame
                                   A
temperatures which reduce thermal NO  generation.  Moreover, the reduced
                                    A
availability of oxygen in the initial combustion zone inhibits fuel NO
                                                                      A
conversion.  Thus, both thermal and fuel NO  are controlled by the low
                                           A
NOX burners.
       The Babcock and Wilcox Company is currently installing the Dual
Register Pulverized Coal-Fired Burner in all its new utility boilers  in
order to meet current NSPS (References 4-38 and 4-39).  The limited
turbulence, controlled diffusion flame burner is designed to minimize fuel
and air mixing at the burner to that required to obtain ignition and  sustain
stable combustion of the coal.  A Venturi mixing device, located in the coal
nozzle, provides a uniform coal/primary air mixture at  the burner.
Secondary air is introduced through two concentric zones surrounding  the
coal nozzle, each of which is independently controlled  by inner and outer
air zone registers.  Adjustable spin vanes are located  in the inner air zone
to provide varying degrees of swirl to the inner air to control coal/air
mixing during the combustion process.  In addition, the windbox is
compartmented to provide airflow control on a per pulverizer basis, thus
permitting operation with lower excess air while maintaining an oxidizing
atmosphere around each burner.
       To date seven dual register burner-equipped utility boilers have been
tested for NOX emissions (Reference 4-39).  For the four bituminous
coal-fired units tested, NO  emissions ranged from 194  to 258 ng/J (0.45
to 0.6 lb/106 Btu, 318 to 422 ppm) at or below the current 258 ng/J
(0.6 lb/106 Btu) NSPS for bituminous coal.  The three subbituminous
coal-fired units exhibited NOX emissions in the range of 129 to 151 ng/J
(0.3 to 0.35 lb/106 Btu, 211 to 247 ppm), well below the current 215  ng/J
(0.5 lb/106 Btu) NSPS for subbituminous coal.  It should be noted that
these low NOX burner-equipped boilers came onstream when the original 1971
NSPS of 301 ng/J (0.7 lb/106 Btu) was still in effect.
       Comparisons with NO  emissions from similar units equipped with the
                          /\
high turbulence older burners show reductions in NO  levels from 40 to
                                                   A                   "*
60 percent due to the new burner design.  As explained  above, the majority
of the reduction is attributable to controlled air-coal mixing in the
furnace chamber.  The resulting lower peak flame temperature and the
                                    4-28

-------
decreased availability of oxygen in the primary flame zone tend to suppress
thermal NO  generation and fuel nitrogen conversion.
          J\
       B&W claims that NO  control through its Dual Register Burners is
                         ^
superior to staging as it maintains the furnace in  an oxidizing environment,
hence minimizing slagging and reducing the potential for furnace wall
corrosion when firing high sulfur bituminous coal.  Also, more complete
carbon utilization can be achieved due to better coal-air mixing in the
furnace.  Finally, lower oxygen levels are required with all the combustion
air admitted through the burners rather than having some of  the total  air
injected above the burner zone.
       Although the Dual Register Burners were developed for use in new
boilers, they can also be retrofitted  to older units.   However, the new
boilers  are also designed to  provide airflow control  on a  per  pulverizer
basis.   This may not  be  possible  in some of the  older units, or the cost
involved in retrofitting  a compartmented windbox  and  making  the necessary
changes  in pulverizer burner  piping may be prohibitive.  If  careful  control
of  fuel  and air to each  burner  is  not  feasible,  the burners  will  not  be  as
effective  in reducing NOX emissions.   Nevertheless, the new  burners  should
reduce NO  levels below  those obtained with the  older high turbulence
          A
burners.  They may still  be  considered for  retrofit application,  perhaps in
conjunction with  other NO control  techniques,  but much development work
                          A
remains.
        Foster Wheeler Energy Corporation  has  developed a dual  register coal
burner for installation  in  its new boilers  (References 4-36 and 4-40). The
new burner reduces  turbulence as compared to  the older designs and causes
 controlled,  gradual  mixing  of fuel and air at the  burner.   This is achieved
 using a dual  throat  with two registers which  splits the secondary air into
 two concentric  streams with  independently variable swirl.   The mixing rate
 between the  primary and secondary air streams and  the  rate  of entrainment of
 furnace gases can thus be varied.  The primary air velocity can also  be
 varied by the use of a coal  nozzle in the shape of a tapered  annulus  with an
 axially movable inner sleeve tip.  In addition, a  perforated  plate air hood
 surrounds the burner and is  used to measure airflow, improve  burner
 circumferential air distribution, and provide a discrete means for balancing
 air on  a burner to burner basis.
                                     4-29

-------
       New Foster Wheeler utility boilers  are  equipped with OFA  ports  in
addition to the new low NO  burners.  The  OFA  ports  are  installed  for  use
                          A
in cases where the new burners  alone  cannot  reduce NO  levels  to meet  the
                                                     /\
NSPS requirements.  However, even when  staging  has to be  employed,  it  is
expected that in most cases the total burner fuel/air ratios will  be above
stoichiometric, as part of the NOV reduction burden  is assumed by  the
                                 A
burners.  Reducing atmospheres  are therefore avoided for  the most  part, thus
minimizing associated slagging  and corrosion problems.  The new  burners with
cooler, less intense flames, and the  larger  new furnace designs  with lower
burner zone heat liberation rates also  tend  to  reduce slagging while at the
same time decrease thermal NO  formation.
                             A
       Test results for the new Foster  Wheeler  burners are reported in
References 4-36 and 4-40.  Reductions in NO  emissions of about  40  percent
                                           A
were observed on a four-burner  steam  generator  when  operated at  full load
with the new burners.  Three utility  steam generators, two 265 MW  opposed
fired units and one 75 MW front wall  fired unit, have been retrofitted with
the new burners and tested for NOX emissions.   Controlled NO   emissions were
in the 172 ng/J (0.4 lb/106 Btu, 281  ppm)  to 215 ng/J (0.5 lb/106  Btu,
352 ppm range.  Test results on one of  the 265  MW units are reported and
show a 48 percent drop in NO  emissions due  to  the new burners.  When  the
                            A
boiler was operated with the new burners and overfire air in conjunction,
reductions in NO  levels of 67 percent  were  achieved with the OFA  ports
                A
100 percent open.  Under such conditions,  however, slag began  to accumulate
after about 24 hours of continuous full load operation and unburned carbon
in the flyash increased to 2.4 percent.  Under  normal operating  procedure,
with OFA ports not more than 20 percent open, the NO  reduction  was
                                                     A
approximately 40 to 50 percent over the uncontrolled case.  Carbon monoxide
was maintained below 50 ppm and unburned carbon in the flyash  was  less than
1 percent.
       The general results and trends were found to  be similar for the other
two units tested.  The uncontrolled NO  levels  of all these units were in
                                      A
the range of 367 to 397 ng/J (600 to  650 ppm).  This is atypically low for
older units equipped with high turbulence  burners.   Installation of new
burners and adjusting them for low NOX  operation combined with OFA
operation with ports open up to 20 percent reduced NO  levels down to  183
                                                     /\
to 214 ng/J (300 to 350 ppm).  This represents  normal low NO  operating
                                                            A
                                    4-30

-------
procedure for these units.  When OFA ports are opened 100 percent and the
burners are adjusted for minimum NOV emissions, NO  levels of 122 to
                                   A              A
137 ng/J (200 to 225 ppm) were attained.  Slagging, however, resulted under
these operating conditions.  In all cases a good quality low sulfur coal was
used so that tube wastage problems did not occur.  Since the test units were
not designed for staged combustion, the slagging effect was expected.
However, slagging when overfire air ports were open was significantly less
with the low NO  burners than with the original high turbulence burners.
               A
       Although experience with the new burners, alone and  in combination
with staging, has been successful  it has been  limited to a  few boilers  and  a
particular type of coal.  Minimum  NO  levels obtained with  these fuels  may
                                    ^
not be repeated with a higher nitrogen content, lower heating value  coal.
       In addition to NOV control  in new units, the Foster  Wheeler dual
                        A
register burner is well suited  (technically) for retrofit  application.   The
airflow to the new burners is controlled individually at each burner by
means  of the perforated hood.   Hence, precise  air/fuel control  at each
burner is possible without incurring major  hardware changes besides  burner
replacement.
       In the  tests referred  to  above,  NO   levels  were approximately
                                         A
halved by the  use  of the  new  retrofitted  burners  alone.   The burners can
also be  used with  oil.   In fact  the  original  patent  for  the dual  register
burner was designed for  and tested with  oil.   No  detailed  data  on oil-fired
utility  boilers fitted with the  new  burners have  been  released  to date.
However, Babcock & Wilcox has reported  the  successful  retrofit  of an
oil-fired  dual  register  burner,  reducing  NO  emissions  to below 129 ng/J
 (0.3 lb/106  Btu, 225 ppm) (Reference 4-39).
       Riley Stoker Corporation is currently modifying the burners used in
 its  turbo  furnace  to  lower NO  emissions  (Reference 4-41).  The new
                              A
 burners  are  designed  to be more flexible and to control  fuel/air mixing to
 reduce thermal  and fuel  NO .   With the  new burners and changes  in furnace
                           J\
 design Riley Stoker expects  to meet current NSPS requirements without
 increased  carbon  or unburned  hydrocarbon losses.   The new burners can  be
 used with  coal,  oil,  and gas  fuels but  are not being considered for retrofit
 application.   No test  data are available on the performance of the  new
 burners  at present.
                                     4-31

-------
       In summary,  low NO  burners  appear  very  attractive, with  potential
                         /\
NO  reductions of the order of 50 percent.   Data from  long term, full
  A
scale demonstrations are imminent,  and  commercial  application  is well
underway.  Indeed,  LNB appears to be the preferred combustion  modification
technique for coal-fired utility boilers.
4.2.4  Flue Gas Recirculation (FGR)
       Flue gas recirculation for NO  control consists of extracting  a
                                    /\
portion of the flue gas from the economizer  outlet and returning it to the
furnace, admitting  the flue gas through the  furnace  hopper or  through the
burner windbox or both.  Flue gas recirculation lowers the bulk  furnace gas
temperature and reduces oxygen concentration in the  combustion zone
(References 4-35 and 4-37).
       Flue gas recirculation through the furnace hopper and near the
furnace exit has long been used for steam temperature control.  Flue gas
recirculation through the windbox and,  to a  lesser degree, through the
furnace hopper is very effective for NO  control on  gas- and oil-fired
                                        /\
units (References 4-33 and 4-37).  However,  it has been shown  to be
relatively ineffective on coal fired units (Reference 4-19).
       Flue gas recirculation for NO  control is more attractive for new
                                    /\
designs than as a retrofit application.  Retrofit installation of flue gas
recirculation can be quite costly.  The fan, flues,  dampers, and controls as
well as possibly having to increase existing fan capacity due  to increased
draft loss, can represent a large investment.  In addition, the flue gas
recirculation system itself will require a substantial maintenance program
due to the high temperature environment experienced  and potential erosion
from entrained ash.  Thus the cost-effectiveness of  this method of NO
                                                                     /\
control has to be examined carefully when comparing  it to other control
techniques.
       As a new design feature, the furnace  and convective surfaces can be
sized for the increase in mass flow and change the furnace temperatures.  In
contrast in retrofit applications, the  increased mass flow increases
turbulence and mixing in the burner zone, and alters the convective section
heat absorption.   Erosion and vibration problems may result (References 4-37
and 4-38).   Flame detection can also be difficult with flue gas
recirculation through the windbox.  In  addition, controls must be employed

                                    4-32

-------
to regulate the proportion of flue gas to air so that sufficient
concentration of oxygen is available for combustion (Reference 4-43).
       As shown in Table 4-6, the limited data indicate that with flue gas
recirculation alone, average NO  reductions of about 17 percent for coal,
                               A
13 percent for oil, and 47 percent for gas have been achieved.  It should be
noted that these values are based on very limited data comparing the effects
of flue gas recirculation alone to baseline conditions.  (Additional data
are discussed in Section 6.)  Data comparing the effects of flue gas
recirculation in combination with other control methods to baseline
conditions are more plentiful and briefly discussed below.  A typical flue
gas recirculation system  is shown in Figure 4-5(c).
4.2.5  Reduced Firing Rate
       Thermal NO  formation generally increases as the volumetric heat
                 A
release rate or combustion intensity increases.  Thus, NO  can  be
                                                         A
controlled by reducing combustion intensity through load reduction,  or
derating, in existing units and by enlarging the firebox in new units.   The
reduced heat release rate lowers the bulk gas  temperature which in turn
reduces thermal NO  formation  (Reference 4-44).
       The heat release rate per unit volume  is generally independent of
unit rated power output.  However, the ratio  of primary flame zone  heat
release to heat removal increases as the unit  capacity is increased.  This
causes NO  emissions for  large  units to  be generally  greater  than  for
         X
small  units  of  similar design,  firing characteristics,  and  fuel.
       The increase in NO  emissions with  increased  capacity  is especially
                          X              .
evident for  gas-fired boilers,  since total NO   emissions  are  due  to
                                              A
thermal NO .  However, for coal-fired and  oil-fired  units the effects of
          /\
increased capacity  are less  noticeable,  since the  conversion  of fuel
nitrogen to  NO  for these fuels represent  a  major  component of total NO
              X                                                         X
formation.   Still,  a reduction in firing rate will  affect firebox
aerodynamics which  may, consequently, affect fuel  NOX emissions.   But such
effects on fuel NO  production are  less  significant.
                   A
       Table 4-7  presents a  compilation  of available  data on  NO  reduction
                                                                A
as  a  result  of  reduced firing  rate.   For coal  firing, an  average of 15
percent reduction  in NO   resulted from  a 28  percent reduction in firing
                       /\
rate.  For oil  firing,  an average of  30 percent reduction  in  NO  resulted
                                                                A
from  a 42  percent reduction  in firing rate.   For  gas  firing,  an average of

                                     4-33

-------
44 percent reduction in NO  resulted from a 44 percent reduction in firing
                          A
rate.  Thus, reduction of NO  with lowered firing rate is most evident
                            X
with gas-fired boilers.
       Reduced firing rate often leads to several operating problems.  Aside
from the limiting of capacity, low load operation usually requires higher
levels of excess air to maintain steam temperature and to control smoke and
CO emissions.  The steam temperature control range is also reduced
substantially.  This will reduce the operating flexibility of the unit and
its response to changes in load.  The combined results are reduced operating
efficiency due to higher excess air and reduced load following capability
due to a reduction in control range.
       When the unit is designed for a reduced heat release rate, the
problems associated with derating are largely avoided.  The use of an
enlarged firebox produces NO  reductions similar to load reduction on
                            A
existing units.
4.2.6  Combination of Controls
       To achieve required NO  emission levels, it is often necessary to
                             A
use a combination of control methods.  Low excess air operation is common to
all combined control method strategies.  Other control combinations that are
most effective are primarily fuel dependent.  It is important in this
respect to distinguish retrofit controls from original design controls.
Unfortunately the test data available are primarily from retrofit control
applications.
       Tables 4-8 through 4-10 represent compilations of test data from
employing combinations of controls to reduce NO  emissions.  Detailed data
                                               /\
from combined controls tests are very limited.  Table 4-11 extends the data
presented in previous tables and lists highest NO  reductions attained
                                                 A
through combinations of controls as a function of boiler/fuel classification.
This table represents the best results achieved in specific applications and
should not be interpreted as generally achievable NO  reductions.  In
                                                    A
comparing Tables 4-3 through 4-7 to Tables 4-8 through 4-10, it is seen that
in combining control techniques, results are complementary but not additive
for NOX reduction.
4.3    ADVANCED CONTROLS
       Several other combustion NO  control techniques, which show promise
                                  A
for future application, are in varying stages of development.

                                    4-34

-------
                             TABLE  4-11.   MAXIMUM  REPORTED NOX REDUCTION  ACHIEVED WITH

                                              BOILER LOAD  AT  OR ABOVE 80  PERCENT  MCR*
Equipment
Type

Tangential


Opposed Wall


Single
Wall


Average
All
Boilers


Fuel
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
Coal
Oil
Nat
Gas
All
Fuels
Control Techniques
Implemented
OFA
Reduced Firing Rate (RFR)
+ BOOS + FGR
RFR + FGR
BOOS
BOOS + OFA
BOOS + OFA
BOOS
BOOS + FGR
BOOS
(LEA)b
BOOS + OFA
BOOS + OFA + FGR
BOOS + OFA
BOOS + OFA <• FGR
Firing Rate
(percent NCR)
85
68
50
83
100
100
82
(81)t>
%
98
(100)
83
98
99
94
Stolen iometry
to Active Burners
(percent)
85
110
110
80
73
69
86
(80)
91
88
(123)
84
82
78
81
Furnace
Stoichiometry
(percent)
113
122
no
107
119
111
115
(120)
121
117
(123)
112
120
114
115
FGR
(percent)
--
NA
32
-
--
--
—
40
-
--
20
--
7
NOX Emissions
(ppm dry 9 3X 0?)
196
110
65
334
222
205
225
(386)
145
109
(931)
252
183
194
210
Maximum NOX
Reduction
(percent)
66
55
81
53
53
79
63
(68)
60
71
(6)
61
56
67
61
co
en
          aFor individual  tests, corresponding baseline and controlled  loads were nearly identical.

          bNumbers in parentheses refer to boilers originally designed  for coal firing with wet bottom furnaces.

-------
advanced burner and furnace concepts,  and  non-catalytic  homogeneous  NO
                                                                      A
reduction with ammonia  injection  in  the  boiler's  convective  section.
4.3.1  Advanced Burner/Furnace Designs
       A number of advanced burner designs  are  being  developed  and tested
to reduce NO  emissions from coal- and oil-fired  utility and  industrial
            A
boilers.  Advanced burners, as compared  to  low  NO  burners,  are defined
                                                  A
as those devices still  under experimental  or  pilot scale development for
lowering NO  emissions.  Burner modification  has  the  potential  of
           A
lowering NO  emissions well below levels attainable by conventional
           A
combustion modification techniques.  Burner modification also has the
advantage of requiring minimal changes in  current boiler design and
operation and is suitable for retrofit application.
       TRW, Incorporated is developing an  advanced burner for oil- and
gas-fired commercial and industrial  boilers with  potential application to
utility boilers.  The burner uses shaped fuel injection  ports to control
fuel and air mixing and entrain combustion products into the flame zone
(Reference 4-45).   In addition to reducing thermal NO ,  the  burner is
                                                      A
effective in controlling fuel nitrogen conversion.  In tests with residual
oils in a packaged boiler and a large  industrial  size boiler, the burner
was capable of reducing NO  emissions by about  30 percent to values
                          A
below 200 ppm.  A preliminary timetable for the industrial burner calls
for commercial application at the end of 1979 (Reference 4-46).  An
EPA-sponsored field demonstration is underway and actual operating data
should soon be available (Reference  4-47).
       Some manufacturers of oil-firing equipment are in the process of
developing burners capable of operating at very low levels of excess air.
The low excess air requirements increase boiler efficiency and  reduce fan
power consumption while decreasing NO  emissions.  The low excess air
                                     A
may also reduce SO^ conversion.  The Peabody Engineering Company has
designed the Air Pressure Recovery (APR) burner designed to  operate  at
excess oxygen levels down to 1/2 percent without  increase in particulate
and unburned hydrocarbon emissions.  The Coen Company is developing  the
LEA burner which uses a tip swirler  to operate down to 0.1 percent excess
oxygen (Reference 4-48).  Both burners are currently  undergoing testing
and no data on NO  emissions are available.
                 A

                                     4-36

-------
       For coal-fired utility boilers, Foster Wheeler is currently testing
an advanced dual register split frame burner design.  A device added at the
burner nozzle splits the primary air-coal flow into several distinct
streams.  Coal particles become concentrated within each stream and, hence
diffuse more slowly into the secondary air.  This further inhibits NOX
formation by extending the slow-burning characteristics of the dual register
burner.  Results from an industrial size test boiler are promising with a
NOV level of approximately 129 ng/0 (0.3 lb/106 Btu) for subbituminous
  A
coal (Reference 4-40).  However, the burner tested on a 375 MW electrical
output boiler produced approximately 215 ng/J (0.5 lb/10  Btu).  A further
modification of this burner with a variable velocity split flame nozzle will
be installed, and a NOX level of 151 to 172 ng/J  (0.35 to 0.4  lb/106 Btu)
is expected (Reference 4-49).  The new design permits the  velocity of  the
primary air-coal stream to be optimized for minimum NO  consistent with
                                                      A
flame  stability and minimum CO.
       Babcock & Mil cox and Energy and Environmental Research,  under EPA
sponsorship,  are developing an advanced  utility  coal burner for  low N0x,
the distributed fuel/air mixing  burner,  for  field testing  (Reference 4-50).
The burner  is designed  to control  both thermal and fuel NO  .   It is
                                                           A
estimated  that  in uncontrolled pulverized  coal combustion,  thermal  NOX
represents  approximately 15 percent  of the total  NOX,  the  volatile
component  of  fuel NO  contributes  65  percent,  and the  char component  about
                     A
20 percent  (Reference 4-51).   In the  distributed mixing burner thermal NO
                                                                          A
is reduced  by minimizing peak  flame  temperature.   Volatile NOX is  reduced
by maintaining  fuel-rich conditions  in the flame zone.   NO  formation  can
                                                           A
also  be  reduced by  increasing  residence  times in the rich  zone, thus
promoting  reduction  of  NO   by  hydrocarbon  and char fragments.  For char
                         A
NO  ,  no  effective control measures are  available, but  the  char component
  A
can be reduced  by maximizing  evolution  of  nitrogen with the volatiles.  This
can be accomplished  by  providing for adequate residence times in the rich
flame zone at high  temperature.
        The distributed  fuel-air  mixing  burner design injects coal  and
primary air from the center of the burner  with a moderate  axial component.
This  stream is  surrounded  by a divided  secondary airstream with a swirl
component for stabilization.   Tertiary air for burnout is  added axially
around the periphery of the burner.   The arrangement results in a hot, rich
                                     4-37

-------
recirculation zone  at  the  center  of  the  flame  with  stoichiometric  ratios  as
high as 2 or more.  Adequate  time at high  temperature  is  also  provided  to
maximize evolution  of  nitrogen  from  the  char.   This  time  in  the  rich  zone
helps reduce most of the NO   that may be formed.  Also, axial  addition  of
                           /\
the tertiary air leads  to  a large flame  zone.   Heat  extraction prior  to
completion of burnout  along with  dilution  of the  tertiary air  by combustion
products lowers the peak flame  temperature, thus  reducing thermal  NO  .
                                                                    A
Although experimental  prototypes  have achieved NO emissions below 86 ng/J
          fi
(0.2 lb/10  Btu), actual field  testing is  not  expected to be complete
until late 1982 (Reference 4-50).
       Babcock and Wilcox Company is  developing a primary combustion  furnace
concept for coal-fired  utility  boilers in  a program  sponsored  by the
Electric Power Research Institute (Reference 4-52).  The  fundamental  process
to control NOX in this  concept  is conversion of fuel nitrogen  to N~
through fuel-rich combustion.   Pulverized  coal  is introduced into  an
extended combustor with substoichiometric  air,  so that combustion  occurs
under fuel-rich conditions isolated  from the rest of the  furnace.  The
length of the combustor is sufficient  to provide  the necessary residence
time to partially oxidize the coal and permit  the desirable N? producing
reactions to occur.   Heat is removed  along the combustion  chamber  to  prevent
slagging.  Secondary air is added at  the exit  of  the primary combustion
furnace to bring the combustion products to oxidizing conditions before they
enter the furnace.   Pilot scale testing  of a 1 MW (4 x 10  Btu/hr) heat
input prototype has achieved the  targeted NO   level of below 86  ng/J
          c                                 X
(0.2 lb/10  Btu).  Commercial offering of a full  scale furnace is  not
expected until at least 1983 (Reference  4-52).
       In summary,  advanced burner/furnace concepts though promising, still
require several  years of development.  It remains to be seen whether  these
advanced burners may need to be combined with  other combustion modification
techniques as well.
4.3.2  Ammonia Injection
       The use of ammonia as a  potential homogeneous NO   reducing  agent
                                                       /v
was first reported by Wendt, et al.  (Reference 4-53).  However,  these
authors attributed their results  to  the  pyrolysis of ammonia to  hydrogen
with the hydrogen in turn reacting with  NO.  The  postflame decomposition  of
NO  by reducing agents has more recently shown promise as  a method for
  A
                                    4-38

-------
augmenting combustion modifications if stringent emission limits are to be
met.  Lyon (Reference 4-54) has reported that selective homogeneous
reduction of NO in combustion effluents was possible with direct injection
of ammonia within a specific temperature range.
       The gas phase reaction in the temperature range of 1090K (1500 F)
to 1310K (1900°F) converts nitric oxide, in the presence of oxygen and
ammonia, into nitrogen and water according to the following chain reaction
(Reference 4-55):
                          NH2 + NO   N2 + H + OH                        (4-6)
                          NH2 + NO   N2 + H20                           (4-7)
                          H + 02     OH + 0                             (4-8)
                          0 + NH3    OH + NH2                           (4-9)
                          OH + NH3   H20 + NH2                          (4-10)
                          H + NH3    H2 + NH2                           (4-11)

Oxygen  acts  as a  catalyst  in reducing  ammonia  to  the  intermediate  NH2
compound which in turn reacts selectively with  NO,  reducing  it  to  N2  and
water.  Based on  this discovery a  patent under  the  name  Thermal  De-N0x was
issued  to  Exxon Research  and Engineering for  this NO   reduction  technique.
        Results of lab scale  tests  show that  the level  of NOX reduction
depends on the combustion  product  temperature,  initial  NO  concentration,
                                                          /\
and quantity of  ammonia  injected.   The data  shown in  Figures 4-6 through  4-8
was obtained by Muzio, et  al.  (Reference 4-56)  during pilot-scale tests
using  a 59 kW (200,000 Btu/hr)  heat input  plug flow combustion tunnel
burning natural  gas.  Figure 4-6  shows the  effect of  temperature and
ammonia/nitric oxide ratio on  the reduction  of NO for an initial NO level of
300 ppm and an excess oxygen level of 4 percent.   It  dramatically
 illustrates the  narrow  temperature window  for optimal NO reduction.  This
optimal temperature range is near 1240K (1780°F), where reductions of 30
to 90  percent were  achieved with  ammonia injections of 0.3 to 1.6 times the
 initial concentration  of NO.  Although increasing the ratio of ammonia to
nitric oxide reduced more NO,  Figure 4-7(a)  shows that for NH3/NO ratios
 of greater than  2,  essentially no further  reduction of NO was achieved.  The
 additional ammonia injected at 1240K leaves unreacted.  However, when the
 reaction temperature is greater than 1300K, the ammonia reacts with oxygen
 to form additional  NO,  an undesirable situation.  Thus, at these higher

                                     4-39

-------
1.0
0.8
0.6
        Excess oxygen:  4:

        Initial NO:  300 ppm
0.4
0.2
                                                (NH3)/(NO)
                                                1.6
          1000
   1100        1200       1300

        Temperature,  K
1400
      Figure 4-6.
Effect of temperature on NO reduction with
ammonia injection (Reference 4-56).
                                4-40

-------
               600
           O.
           o.
                      2%  excess  oxygen
                              (NH3)/(NO)
                      a.  Nitric oxide reduction
                              (NH3/(NO)



                         b.   Ammonia carryover



Figure 4-7.  Nitric oxide reductions and ammonia carryover with ammonia

             injection at 2 percent excess oxygen (Reference 4-56).
                                  4-41

-------
 temperatures  essentially  no ammonia leaves unreacted (see Figure  4-7(b)).
 Figure 4-8  shows  that  for a given  (NH3)/(NO)  ratio,  ammonia  injection  is
 more effective  at higher  initial  NO levels.   However,  this trend  is  only
 signficiant at  initial  NO levels  of less  than 400  ppm.   These  factors  are
 important considerations  in assessing  the tradeoffs  between  implementing
 only ammonia  injection  as a NO  control or in combination with combustion
                              J\
 modification  techniques.
       Byproduct  pollutants from  ammonia  injection have  been analyzed  by
 Lyon and Longwell  (Reference 4-57)  who measured  emissions of N20,  CO,  HCN,
 S03 and NH.HSO. from gas- and oil-fired pilot scale  combustion
 facilities.   Emissions  of N20 were  found  to be limited to 2 moles  NpO
 for every 100 moles of  NO reduced.   Ammonia was  not  found to react with
 COp to form CO.   However,  the presence of ammonia  in the combustion
 effluent was  found to  inhibit the conversion  of  CO to COp.  Therefore,
 this technique  becomes  a  problem  only  if  the  concentration of  CO  in  the flue
 gas is significant.  In utility boilers this  may not be  the case,  especially
 for gas- and  oil-fired  units.  However, coal-fired boilers with higher CO
 levels may present a problem.
       Cyanide  can only form if unburned  hydrocarbons are present  in the
 flue gas.  For  normal operation of  utility boilers,  only a few ppm of  HCN
 can be found.
       Careful  laboratory work has  shown  that ammonia injection does not
 produce additional SOo  emissions (Reference 4-58).   In fact, S0?  levels
 remained unchanged during  the injection stage.   The  main  byproduct pollutant
 of concern is ammonium  bisulfate.   The unreacted NH7 leaving De-NO
                                                   «5               X
 reaction was found to combine with  S03 and HpO to form ammonium
 bisulfate.  This substance  forms a  very corrosive liquid  at 480K to 530K
 (400°F to 500°F).  Thus it  could potentially  corrode sections   of  the
 boilers such as the air preheater and  flue gas ducts if  the concentration of
 ammonia bisulfate is significantly  high.   During full scale studies of an
 oil-fired boiler however,  no  evidence  of  additional  corrosion  was  found
 (Reference 4-59).
       Full  scale application of the Exxon Thermal De-NO   process  has been
                                                         A
 reported for six gas- and  oil-fired  combustion sources in  Japan affiliated
with Exxon Corporation.  Figure 4-9  shows  that the average NO   reduction
                                                             A
 reported with the ammonia  injection  technique  is of  the  order  of 50 percent.

                                     4-42

-------
i
-p»
CO
                0.6   -
                                                               Excess oxygen:  2"
                                                               Temperature:  1233K
Initial NO level (ppm)
    100
    200
    400
0  680
   1050
            .=  0.4   -
               0.2   -
                 0
                                                      (NH,)/(NO)
                            Figure  4-8.   Effect  of  initial  nitric oxide concentrations  on  NO
                                         reduction  with ammonia injection  (Reference 4-56).

-------

                     60
:.
                     50
                o    40
                     30
                     20
                     I

                           Unit Description
                         70 t/hr steam boiler
                        150 kbbl/d crude heater
                     A 430 t/hr steam boiler
                     ^430 t/hr steam boiler
                     O 120 t/hr steam boiler
                     O]50 kbbl/d crude heater
1000      1050      1100       1150     1200      1250
        Injection zone flue gas  temperature, K
                                                                                         1300
                   Figure  4-9.   Performance of Thermal  De-NO   systems  in  commercial applications
                                (Reference 4-55).

-------
One other full  scale combustion facility in the United States has been
retrofitted with ammonia injection.  NO  emissions were reduced from
                                       A
270 ppm to 80 to 120 ppm (Reference 4-59).  Details of the process are not
available.
       The Thermal De-NO  process for NO  emission reduction shows a
                        f^               A
promising application for utility boilers, with potential NO  reductions
                                                            A
of 40 to 60 percent.  However, full scale studies have been limited to gas
and oil which are becoming less available in the  utility fuel market.  Pilot
scale tests using ammonia injection on coal-fired furnaces have  been
completed by KVB under EPRI sponsorship (Reference 4-60).  Basically  the
study confirmed the effectiveness  of the  technique as  well as  its  potential
limitations such  as the narrow temperature window and  possible  ammonia
byproduct emissions.
        EPA  has  assessed the applicability and  effectiveness  of  ammonia
injection  in two  recent studies  (References  4-61  and  4-62).   The studies
conclude  that  ammonia  injection  holds  promise  for additional  NOV
                                                                X
reductions,  40  to 60  percent,  in  those air  quality regions where stringent
NO  controls may  be required.   Ammonia injection  could be applied as  an
   A
add-on  technology,  in  combination with conventional  combustion modification
techniques.  However,  a number of limitations  need to be considered and
 evaluated before  the process  is  retrofitted, especially for  coal-fired
 boilers:
        •   Performance is  very sensitive  to flue  gas temperature, and is
            maximized only within a 50K temperature gradient from the optimum
            temperature of about 1240K.  This temperature sensitivity may
            require  special  procedures for load following boilers, such as
            multiple NH3 injection grids.
        •   Performance is very sensitive to flue gas  residence  time  at
            optimum temperatures.  High flue gas  quench rates are  expected  to
            reduce process performance.
        t   Costs of the process can be much higher than for other combustion
            controls
        •   Successful retrofit application  is highly  dependent  on the
            geometry of convective section
        0   Byproduct emissions such as ammonium  bisulfate might cause
            operational problems,  such as  air  preheater fouling, especially
            in  coal-fired boilers
                                     4-45

-------
       •   Ammonia emissions may  be  an  environmental  problem  if  the  process
           is not carefully controlled
       Exxon is currently  investigating  possible  solutions  to  these
potential problems.  Ammonia injection  can  potentially  offer  a near  term
control option for achieving NO   emission levels  not  obtainable  with
                               A
current state-of-the-art controls (Reference 4-62).
4.4    MINOR EMPHASIS CONTROLS
       In this section, controls  which  were given minor emphasis  in  the
present study are briefly  discussed.  These were  treated  in less  detail
because they were considered to have  less promise for widespread  application
than those described above, for such  reasons as energy  penalties, high cost,
or technical difficulties.  However,  flue gas treatment techniques are
included here largely because they are  being studied  in greater  depth in
other efforts (References  4-63 and 4-64).
4.4,1  Reduced Air Preheat
       Thermal NO  production is  strongly influenced  by the effective peak
                 A
temperatures in the combustion zone.  Thus, any modification that lowers
these temperatures, such as reducing  the combustion air temperature, should
lower NO  emissions.  Theory indicates that a 56K (100°F) decrease in
air preheat temperature will result in an approximately 28K (50°F)
reduction in the adiabatic combustion temperature, which  in turn will
decrease thermal NO  formation by 27  percent (References  4-44  and 4-65).
                   A
Since reduced air preheat  does not significantly suppress fuel nitrogen
conversion (Reference 4-66), it is expected that this control  technique
would be most effective on fuels, such as natural gas and distillate oil,
which have low nitrogen content.
       Reduced air preheat is potentially applicable to most utility boilers
because these sources are equipped with regenerative air  heaters which
preheat combustion air.  This method for controlling NO   usually greatly
                                                       A
lowers fuel  economy, however.   New designs to reduce stack gas temperatures,
for example, and redesign of the  convective section of a  boiler for more
heat absorption would be necessary to maintain efficiency.
       Only limited field test data are available on the  effect of reduced
air preheat in utility boilers due to the severe efficiency penalty incurred
                                    4-46

-------
with this method.  Some field test results and discussions on reduced air
preheat for utility boilers are available in References 4-67 through 4-69.
       The data reported for coal firing showed varying trends, although a
maximum reduction of 75 ppm (at zero percent 02) per 56K reduction in air
temperature was reported in one case (Reference 4-66).  In general, NOX
reductions of about 50 percent for gas-fired boilers and 40  percent for
oil-fired boilers can be expected with reduced  air preheat,  in contrast  to
the relatively small reductions in coal-fired boilers  (Reference  4-77).
       In summary, reduced  air preheat reduces  efficiency, and is therefore
not considered a practical  control technique for existing  units.   Design
changes  in new units, such  as  installing or enlarging  an  economizer, would
be required to regain the waste heat which would otherwise be  lost through
the stack.
4.4.2  Water  Injection
       Water  injection  has  been  shown  to  reduce flame temperature and  is
widely used  in gas  turbines.   Only recently  has water injection  been tried
on  utility boilers.
       The Ormond  Beach,  steam generating  units operated by Southern
California Edison  were  tested  with water  injection to reduce NO
                                                                A
 (Reference 4-33).   The  boilers operating  at  75 percent of full load (design
 capacity 800  MW) with 10  percent tertiary air, were emitting 400  ppm of NO
 when  0.6 kg  of water per  kg of oil was injected, the emissions were reduced
 to 228 ppm,  a 43 percent  reduction.   Higher reductions were obtained with
 flue  gas recirculation and water injection combined.  For example, with
 15 percent gas recirculation and injection of 0.2 kg of water/kg  of oil, NO
 reduction of nearly 50 percent was achieved.  Compared to flue gas
 recirculation, water injection imposes a large energy penalty.   Water
 injection increased the minimum 0~ requirement and reduced  boiler
 efficiency by 10 percent in the Ormond Beach case.  The  large efficiency
 loss due to water injection makes this technique  unattractive to the utility
 sector.
         In summary, water injection is not seen as a feasible  NO   reduction
                                                                A
 technique for utility boilers based on the large  energy  penalty  incurred.
 Thus  little  current work with this technique  is being performed  on  large
 steam generators.
                                      4-47

-------
4.4.3  Flue Gas  Treatment
       While combustion modification  techniques  seek  to  lower  NO
                                                                A
emissions by minimizing NO formation,  flue  gas treatment  (FGT)  processes
involve post-combustion NO  removal from  the  flue  gas.  Flue gas treatment
                          A
has potential for use  combined  with combustion modifications when  very  high
removal efficiencies are required  (References 4-63, 4-64,  and  4-76).
       FGT has been applied to  only a  few commercial  oil-  and  gas-fired
boilers in Japan.  No  FGT installation for  NO  control on  utility  boilers
                                              A
exists in the United States as  combustion modifications represent  the most
cost effective approach to achieving moderate NO   reductions.   However,
                                                 A
combustion modifications alone  may not be able to  provide  the  degree of
control necessary to meet future N02 ambient  air quality standards.  Thus
EPA has initiated several demonstration projects to investigate the use of
FGT in the U.S.  (Reference 4-64).
       FGT processes can be divided into  two main  categories:   dry processes
and wet processes.  Dry processes reduce  NO  by  catalytic  reduction and
                                            A
operate at temperatures between 570 to 700K (570 to 800 F).  Wet systems
are generally either oxidation/absorption or absorption/reduction processes,
both operating in the  310K to 320K (100 to  120°F)  range.
       Among the many  dry process variations, selective catalytic reduction
(SCR) using ammonia has been perhaps the  most successful.  Over 50 percent
NO , and often up to 90 percent reductions  have been  claimed using such
  A
processes.  However, plugging of the catalyst bed  and fouling of the
catalyst itself  are major operational  concerns, especially with coal
firing.  Moreover, use of SCR has raised  concerns  in  that  any ammonia left
in the flue gase may combine with existing  SOg/SOp to produce a visible
plume, and byproducts, such as  ammonium bisulfate, which are corrosive to
boiler equipment.
       Wet FGT processes utilize more  complex chemistry than dry processes.
In the oxidation/absorption processes, strong oxidants such as ozone or
chlorine dioxide are used to convert the  relatively inactive NO in the flue
gas to N02 or N205 for subsequent absorption.  In  the absorption/
reduction processes, chelating  compounds, such as  ferrous ethylenediamine-
tetracetic acid are required in the scrubbing solution to trap the NO.
However, because wet processes  rely on absorption, most of them create
troublesome byproducts such as  nitric  acid, potasium  nitrate, ammonium
                                    4-48

-------
sulfate, calcium nitrate, and gypsum which may have little commercial
value.  In addition, the high cost of an absorber and an oxidant or
chelating agent is likely to be prohibitive for flue gases with high NO
                                                                       /\
concentrations.
       In general, the dry F6T techniques used in Japan can  probably be
applied to gas- and oil-fired sources in the U.S.  However,  the
applicability of dry processes to coal-fired boilers remains  to be
demonstrated.  Wet processes are  less well developed and costlier  than dry
FGT processes.  However, wet simultaneous, as well as dry  simultaneous
NO /SO  processes warrant further investigation.  In any case, more
  A   A
field tests  are needed  to determine  the costs, secondary effects,
reliability,  and waste  disposal problems.  Flue  gas  treatment holds  some
promise as a control technique for use when  high NO  removal efficiencies
                                                   A
are necessitated  by  stringent emission  standards.  However,  compared to
combustion modifications FGT is considerably more expensive.
                                      4-49

-------
                          REFERENCES FOR SECTION 4
4-1.   Zeldovich, J., "The Oxidation of Nitrogen  in Combustion and
       Explosions," Acta Physiochem URSS.  (Moscow), Vol. 21, p. 4, 1946.

4-2.   Bowman, C. T. and Seery D. J. , "Investigation of NO Formation
       Kinetics in Combustion Processes:   The Methane-Oxygen-Nitrogen
       Reaction," in Emissions from Continuous Combustion Systems.
       Cornelius, W. and Agnew, W. G., eds., Plenum, 1972.

4-3.   Bartok, W., et al., "Basic Kinetic  Studies and Modeling of NO
       Formation in Combustion Processes," AIChE Symposium Series No. 126.
       Vol. 68, 1972.

4-4.   Halstead, C. J. and Munro, A. J. E., "The Sampling, Analysis, and
       Study of the Nitrogen Oxides Formed in Natural Gas/Air Flames,"
       Company Report, Shell Research, Egham, Surrey, U.K., 1971.

4-5.   Thompson, D., et al., "The Formation of Oxides of Nitrogen in a
       Combustion System," presented at the 70th National AIChE Meeting,
       Atlantic City, 1971.

4-6.   Lange, H. B., "NOX Formation in Premixed Combustion:  A Kinetics
       Model and Experimental Data," presented at the 64th Annual AIChE
       Meeting, San Francisco, 1971.

4-7.   Sarofim, A. F. and Pohl, J. H., "Kinetics of Nitric Oxide Formation
       in Premixed Laminar Flames," 14th Symposium (International) on
       Combustion, The Combustion Institute, Pittsburg, 1973.

4-8.   Iverach, D., et al., "Formations of Nitric Oxide in Fuel-Lean and
       Fuel-Rich Flames," ibid., 1973.

4-9.   Wendt, J. 0. L. and Ekmann, J. M.,  "Effect of Fuel Sulfur Species
       on Nitrogen Oxide Emissions from Premixed Flames," Comb. Flame.
       Vol. 25, 1975.

4-10.  Malte, P. C. and Pratt, D. T., "Measurement of Atomic Oxygen and
       Nitrogen Oxides in Jet-Stirred Combustion," 15th Symposium
       (International) on Combustion, The Combustion Institute,
       Pittsburgh, 1975.

4-11.  Mitchell, R. E. and Sarofim, A. F., "Nitrogen Oxide Formation in
       Laminar Methane Air Diffusion Flames," presented at the Fall
       Meeting, Western States Section, The Combustion Institute, Palo
       Alto, California, 1975.

4-12.  Bowman, C. T., "Non-Equilibrium Radical Concentrations in Shock
       Initiated Methane Oxidation," 15th  Symposium (International) on
       Combustion, The Combustion Institute, Pittsburg, 1975.


                                    4-50

-------
4-13.   Fem'more,  C.  P.,  "Formation of Nitric Oxide in Premixed Hydrocarbon
       Flames,"  13th Symposium (International) on Combustion,  The
       Combustion Institute, Pittsburgh, 1971.

4-14.   MacKinnon, D. J., "Nitric Oxide Formation at High Temperatures,"
       Journal of the Air Pollution Control Association, Vol.  24, No. 3,
       pp. 237 to 239, March 1974.

4-15.   Heap, M.  P.,  et al., "Burner Criteria for NOX Control;  Volume I
       -- Influence of Burner Variables on NOX in Pulverized Coal
       Flames," EPA 600/2-76-061a, NTIS-PB 259 911/AS, March 1976.

4-16.  Bowman, C. T., et al., "Effects of  Interaction Between Fluid
       Dynamics on Chemistry or Pollutant  Formation  in Combustion,"  in
       Proceedings of the  Stationary  Source Combustion Symposium; Volume  I
       — Fundamental Research, EPA 600/2-76-152a, NTIS-PB 256 320/AS,
       June 1976.

4-17.  Shaw,  J. T.  and  Thomas,  A.  C.,  "Oxides of  Nitrogen in  Relation  to
       the Combustion of Coal," presented  at  the  7th International
       Conference on  Coal  Science, Prague,  June  1968.

4-18.  Pershing, D. W.,  et al., "Influence of Design Variables on the
       Production of  Thermal  and  Fuel  NO from Residual  Oil  and Coal
       Combustion," AIChE  Symposium  Series,  No.  148, Vol. 71,  pp. 19 to
       29,  1975.

4-19.  Thompson, R.  E.  and McElroy,  M. W., "Effectiveness  of  Gas
       Recirculation  and Staged Combustion in Reducing NOX  in a  560-MW
        Coal-Fired  Boiler," EPRI FP-257, NTIS-PB 260 582, September 1976.

 4-20.   Sarofim,  A.  F.,  et  al.,  "Mechanisms and  Kinetics of  NOX
        Formation:   Recent Developments,"  presented at the 65th Annual
        AIChE Meeting, Chicago,  November 1976.

 4-21.   Martin.  G.  B.  and Berkau,  E.  E., "An Investigation of the
        Conversion of Various Fuel Nitrogen Compounds to Nitrogen Oxides  in
        Oil Combustion," presented at  the 70th National AIChE Meeting,
        Atlantic City, August 1971.

 4-22.   Habelt,  W.  W. and Howell, B. M., "Control of NOX Formation in
        Tangentially Coal-Fired Steam  Generators," in Proceedings of the  NO
        Control  Technology Seminar. EPRI SR-39, NTIS-PB 253 661,   February
 4-23.  "Air Quality and Stationary Source Emission Control," U.S. Senate,
        Committee on Public Works, Serial No. 94-4, March 1975.

 4-24.  Pohl, J. H. and Sarofim, A. F.,  "Fate of Coal  Nitrogen  During
        Pyrolysis and Oxidation,"  in Proceedings of the  Stationary Source
        Combustion Symposium; Volume i  — hunaamentai  Research,  EHA
        600/2-76-152a,  NTIS-PB  256 320/AS, June  1976.
                                      4-51

-------
4-25.  Heap, M. P., et al., "The Optimization of Burner Design Parameters
       to Control NOX Formation in Pulverized Coal and Heavy Oil
       Flames," in Proceedings of the Stationary Source Combustion
       Symposium; Volume II — Fuels and Process Research and Development.
       EPA 600/2-76-1525, NTIS-PB 256 321/AS, June 1976.

4-26.  Pohl, J. H. and Sarofim, A. F., "Devolatilization and Oxidation of
       Coal Nitrogen," presented at the 16th Symposium (International) on
       Combustion, Cambridge, Massachusetts, August 1976.

4-27.  Blair, D. W., et al., "Devolatilization and Pyrolysis of Fuel
       Nitrogen from Single Coal Particle Combustion," 16th Symposium
       (International) on Combustion, Cambridge, Massachusetts, August
       1976.

4-28.  Brown, R. A., et al., "Investigation of Staging Parameters for
       NOX Control in Both Wall and Tangentailly Coal-fired Boilers," in
       Proceedings of the Second Stationary Source Combustion Symposium:
       Volume III, New Orleans, EPA-600/7-77-073c, NTIS-PB 271 75//AS.
       July 1977.

4-29.  Pershing, D. W., "Nitrogen Oxide Formation in Pulverized Coal
       Flames," Ph.D. Dissertation, University of Arizona, 1976.

4-30.  Axworthy, A. E., Jr., "Chemistry and Kinetics of Fuel Nitrogen
       Conversion to Nitric Oxide," AIChE Symposium Series. No. 148, Vol.
       71, pp. 43 to 50, 1975.

4-31.  Axworthy, A. E., et al., "Chemical Reactions in the Conversion of
       Fuel Nitrogen to NOX," in Proceedings of the Stationary Source
       Combustion Symposium, Volume I, EPA 600/2-76-152a, NTIS-PB 256
       320/AS, June 1976.

4-32.  Pershing, D. W. and Wendt, J. 0. L., "The Effect of Coal Combustion
       on Thermal and Fuel NOX Production from Pulverized Coal
       Combustion," presented at Central  States Section, The Combustion
       Institute, Columbus, Ohio, April 1976.

4-33.  Norton, D. M., et al., "Status of Oil-Fired NOX Control
       Technology," in Proceedings of the NOX Control Technology
       Seminar. EPRI SR-39, NTIS-PB 253 661, February 1976.

4-34.  Durrant, 0. W., "Pulverized Coal — New Requirements and
       Challenges," presented to ISA Power Instrumentation Symposium,
       Houston, Texas, May 1975.

4-35.  Campobenedetto, E. J., Babcock & Mil cock Co.,  letter to Acurex
       Corp., November 15, 1977.

4-36.  Vatsky, J., "Attaining Low NOX Emissions by Combining Low
       Emission Burners and Off-Stoichiometric Firing," presented at the
       70th Annual AIChE Meeting, New York, November 1977.


                                    4-52

-------
4-37.  Rawdon, A.  H.  and Johnson, S.  A.,  "Control  of NOX Emissions from
       Power Boilers," presented at Annual Meeting of the Institute of
       Fuel, Adelaide, Australia, November 1974.

4-38.  Campobenedetto, E. J., "The Dual Register Pulverized Coal Burner --
       Field Test Results," presented to Engineering Foundation Conference
       on Clean Combustion of Coal, New Hampshire, August 1977.

4-39.  Barsin, J. A., "Pulverized Coal Firing NOX Control," in
       Proceedings;  Second NOX Control Technology Seminar, Electric
       Power Research Institute, Report No. FP-1109-SR, Palo Alto,
       California, July 1979.

4-40.  Vatsky, J., "Experience  in Reducing NOX  Emissions on Operating
       Steam Generators,"  in Proceedings;  Second NOX Control  Technology
       Seminar, Electric Power  Research Institute, Report No.  FP-1109-SR,
       Palo Alto, CA, July 1979.

4-41.  "NOX Control Review," Vol. 2,  No.  4, EPA Industrial Environmental
       Research Laboratory,  RTP, North Carolina,  Fall 1977.

4-42.  Krippene,  B. C.,  "Burner and  Boiler Alterations  for NOX Control,"
       presented  to Central  States Section, The Combustion Institute,
       Madison, Wisconsin, March 1974.

4-43.  Barr,  W.  H.,  et  al.,  "Modifying Large  Boilers to Reduce Nitric
       Oxide  Emissions,"  Chemical  Engineering Progress, Vol.  73,  pp.  59  to
       68,  July  1977.

4-44.  Bell,  A.  W.,  et  al.,  "Nitric  Oxide Reduction by  Controlled
       Combustion Processes,"  KVB,  Inc.,  Western  States Section/Combustion
        Institute, April  20-21,  1970.

4-45.  Koppang,  R. R.,  "A Status Report on the Commercialization and
       Recent Development History of the  TRW Low  NOX Burner," TRW Energy
        Systems Group Publication,  Redondo Beach,  CA, 1977.

 4-46.   Personal  communication, Boughton,  M., TRW, Inc., Redondo Beach, CA,
        May 1979.

 4-47.   Matthews, B.  J., TRW, Inc., Redondo Beach, CA,  letter to Peters,
        W., EPA,   IERL-RTP, NC,  March 1979.

 4-48.   Stavern,  D. V.,  "The Coen Low Excess Air Burner," presented at the
        NOX Control Technology Workshop,  Pacific Grove, California,
        October 1977.

 4-49.   Vatsky, J., "Larger  Burners  and Low NOX," Heat  Engineering,
        Vol. 49,  No. 2, pp.  17-25, April-June 1979.
                                     4-53

-------
4-50.  Martin, G. B.,  "Field  Evaluation of Low  NOX Coal  Burners on
       Industrial and  Utility Boilers," in Proceedings of  the Third
       Stationary Source Combustion Symposium,  Volume I,
       EPA-600/7-79-050a, February 1979.

4-51.  Gershman, R., Heap, et a!., "Design and  Scale-Up  of Low Emission
       Burners for Industrial and Utility Boilers,"  in  Proceedings of the
       Second Stationary Source Combustion Symposium, Volume II,
       EPA-600/7-77-073B, NTIS-PB 271 756/9BE,  July  1977.

4-52.  Johnson, S. A., et al., "The Primary Combustion Furnace System —
       An Advanced Low-N0x Concept for Pulverized Coal Combustion," in
       Proceedings:  Second N0y Control Technology Seminar, Electric
       Power Research  Institute, Report No. FP-1109-SR,  Palo Alto, CA,
       July 1979.

4-53.  Wendt, J. 0. L., et al., "Reduction of Sulfur Trioxide and Nitrogen
       Oxides by Secondary Fuel Injection," 14th Symposium (International)
       on Combustion,  The Combustion Institute, 1973.

4-54.  Lyon, R. K., "Method for the Reduction of Concentrations of NO in
       Combustion Effluents using Ammonia," U.S. Patent  No. 3,900,554,
       August 1975.

4-55.  Bartok, W., "Non Catalytic Reduction of  NOX with  NH ," in
       Proceedings of  the Second Stationary Source Combustion Symposium:
       Volume II, EPA-600/7-77-073b. NTIS-PB 271 756/9BE, July 1977.

4-56.  Muzio, L. J., et al.,  "Homogeneous Gas Phase Decomposition of
       Oxides of Nitrogen," EPRI Report FP-253, NTIS-PB  257 555, August
       1976.

4-57.  Lyon, R. R. and Longwell, J. P., "Selective, Non-Catalytic
       Reduction of NOX by NH ," in Proceedings of the N0y Control
       Technology Seminar, EPRI SR-39, NTIS-PB  253 661,  February 1976.

4-58.  Teixeira, D. P., "Status of Utility Application of Homogeneous
       NOX Reduction," in Proceedings of the NOX Control Technology
       Seminar. EPRI SR-39, NTIS-PB 253 661, February 1976.

4-59.  "Exxon Says Stationary NOX Emissions Significantly Reduced at
       Plant," Air/Water Pollution Report, p. 76, February 20, 1978.

4-60.  Muzio, L. J., et al.,  "Noncatalytic NO Removal with Ammonia," EPRI
       Report FP-735, Research Project 835-1, April 1978.

4-61.  Varga, G. M., et al.,  "Applicability of  the Thermal DeNOx Process
       to Coal-Fired Utility Boilers," EPA-600/7-79-079, March 1979.

4-62.  Castaldini, C., et al., "Technical Assessment of  Thermal DeNOx
       Process," EPA-600/7-79-117, May 1979.
                                    4-54

-------
4-63.  Faucett, H.  L.,  et al.,  "Technical  Assessment of NOX Removal
       Process for  Utility Application," EPA 600/7-77-127 (also EPRI
       AF-568), NTIS-PB 276 637/AS, November 1977.
4-64.  Mobley, J. D., "Flue Gas Treatment Technology for NOX Control,"
       in Proceedings of the Third Stationary Source Combustion Sympos
       Volume II. EPA-600/7-79-050b. February. 1979.
4-65.  Cato, G. A., et al., "Field Testing:  Application of Combustion
       Modification to Control Pollutant Emissions from Industrial Boilers
       - Phase II," EPA 600/2-76-086a, NTIS-PB 253 500/AS, April 1976.

4-66.  Armento, W. G. and W. L. Sage, "The effect of Design Operation
       Variables on NOX Formation in Coal-Fired Furnaces," Alliance
       Research Center/B&W Pulverized Coal Combustion Seminar, June 19-20,
       1973.

4-67.  Blakeslee, C. E. and Burbach, H. E., "Controlling NOX Emissions
       from Steam Generators," presented at the 65th Annual Meeting of Air
       Pollution Control Association, June 1972.

4-68.  Blakeslee, C. E. and Burbach, H. E., "NOX  Emissions from
       Tangentially-Fired  Utility Boilers, Part II, Practice," AIChE
       Symposium Series No. 148, Vol. 71,  1975.

4-69.  Bartok, W., et  al.,  "Systematic  Field  Study  of NOX  Emission
       Control Methods for Utility  Boilers,"  ESSO Research  and Engineering
       Co., Report No. GRU.4GNOS.71, December 1971.

4-70.  Breen,  B.  P.,  "Combustion  in Large  Boilers:   Design  and Operating
       Effects on  Efficiency  and  Emissions,"  presented  at  the  16th
       Symposium  (International)  on Combustion, Cambridge,  Massachusetts,
       August  1976.

4-71.  Hunter, S.  C.,  et  al.,  "Application of Combustion  Technology for
       NOX Emissions  Reduction on  Petroleum Process Heaters,"  presented
        at the  83rd National  AIChE Meeting, Houston, Texas,  March 1977.

4-72.   "Standards  Support and Environmental  Impact  Statement Volume 1:
        Proposed  Standards of  Performance for Stationary Gas Turbines,"
        EPA-450/2-77/017a, NTIS-PB 272  422/7BE, September 1977.

 4-73.   Breen,  B.  P.,  "Control of the Nitric Oxide Emissions from Fossil
        Fueled  Boilers,"  The Fourth Westinghouse International  School for
        Environmental  Management, July 1973.

 4-74.   Bell,  A.  W.,  et al., "Nitric Oxide Reduction by Controlled
        Combustion Processes," presented at the Western States Section, the
        Combustion Institute, April 1970.
                                     4-55

-------
4-75.  Jain, L. K., et al., "State of the Art for Controlling NOX
       Emissions, Part I, Utility Boilers," EPA-R2-72-072a, NTIS-PB 213
       297, September 1972.

4-76.  Mobley, J. D. and Stern, R. D., "Status of Flue Gas Treatment
       Technology for Control of NOX and Simultaneous Control of NOX
       and SOX," in Proceedings of the Second Stationary Source
       Combustion Symposium:  Volume III, EPA 600/7-77-073c, NTIS-PB 271
       757/7AS, July 1977.

4-77.  Goodwin, D. R., "Electric Utility Steam Generating Units.
       Background Information for Proposed NOX Emissions Standards,"
       EPA-450/2-78-005a, NTIS-PB 286 155, July 1978.
                                   4-56

-------
                                 SECTION 5
            N0  CONTROL CHARACTERIZATION:  EMISSION CORRELATION
       Based on the general  NO  formation concepts discussed in
                              A
Section 4, it is evident that the basis for combustion modification NO
                                                                      A
controls can be eventually traced back to chemical kinetics, turbulent
mixing, and heat transfer.  However, incomplete understanding of the
combustion phenomena as well as insufficient data have prevented
researchers from fully characterizing NO  formation and control in
                                        X
utility boilers.  Therefore, from a practical control point of view,  it
would be valuable to relate NO  emissions to gross overall furnace
                              A
parameters,  such as surface heat release rates, for which  there are
sufficient  data from full scale utility boiler tests.  These overall
furnace parameters  can,  in  turn, be explained  in  terms of  such  local  or
fundamental  parameters  as temperature  and  combustion  regimes.   This
information can be  used to  evaluate the  basis  and effectiveness of various
NO   control  techniques.   Furthermore,  it can  aid  in  the  cost-effective
   A
design of  new controls  as well  as  in  providing direction for  future
research  efforts  since  key  boiler/burner  design  and  operating  variables
would have been identified.
        However,  it  should be noted that  the NO  emission correlations
                                               A
presented in this study are not intended to be definitive nor predictive.
Rather they are meant only  to present general  trends and highlight
 important burner  and boiler design and operating parameters.   There is
 insufficient data available to warrant in-depth interpretation.
        This section begins  with a review of previous NO  modeling
                                                        A
 efforts Section 5.1 then proceeds to a development of an  interboiler NO
                                                                        /\
 correlation model in Section 5.2.  The available data base of control
 tests, spanning several test programs on many individual  boilers, is  also
 discussed.  Section 5.3 presents the results of  applying  this correlation
                                      5-1

-------
model to some of the major boiler/fuel classifications,  as data
permitted.  The key boiler/burner design  and  operating variables  and fuel
characteristics affecting NO  formation are  identified,  and  the results
                            A
are interpreted in the  light of fundamental  combustion theory and
knowledge of boiler operating practice.   Finally,  Section 5.4 summarizes
the findings.
5.1    PREVIOUS N0¥ MODELING EFFORTS
                  A
       In recent years, numerous investigators have attempted to  predict
NO  formation in utility boilers, with varying degress of success.  In
  A
general, the studies have fallen into one of  two broad categories:  purely
empirical approaches to NO  prediction based  on available field test
                          A
data, and more fundamental approaches relying more heavily on heat and
mass transfer, flowfield, and combustion  fundamentals.   Some of the more
significant efforts are highlighted below.
       Perhaps the most comprehensive of  the  fundamental models is that
developed by Combustion Engineering (CE)  and  reported by Bueters, et al.
(Reference 5-1) and Habelt and Selker (Reference 5-2).   Combustion
Engineering spent several years developing a  performance code to predict
NO  emissions from tangentially fired utility boilers.   Basically, their
  A
model calculates the axial temperature/time history of combustion
products, then introduces NO  generation  via the Zeldovich mechanisms.
                            A
The calculation proceeds by dividing the  furnace into "slices" and solving
the conservation equations for mass (including the combustion reactions),
momentum (flowfields), and energy in each slice.
       Although this model is quite detailed and was designed to be
fundamental, it still relies heavily on actual data to determine the
numerous adjustable parameters incorporated.  For example, although the
model accounts for vertical recirculation of the gaseous products, it
cannot determine the length of the recirculation region  nor the length and
position of the heat release zone.   In addition, the model needs as input
a "gas emissivity operator."  Because these quantities depend upon the
operational mode of the boiler, e.g., load, fuel nozzle  tilt, and excess
air level, data on the length and position of the heat release zone and
gas emissivity still must be correlated empirically against boiler
operational variables.
                                     5-2

-------
       In addition,  since the CE model  incorporates only Zeldovich
kinetics, it has inadequate provisions  for predicting fuel  NO
                                                             /\
generation.  Thus, predictions for units burning high nitrogen content
fuels, such as high nitrogen oil and coal, are uncertain.  The code
predicts NO  to within 10 percent for gas-fired units, but only to
           A
within 15 percent for oil-fired units,  and is inadequate for coal-fired
units.  Perhaps even the good agreement for gas-fired boilers is
fortuitous since the Zeldovich mechanism employed  is still inaccurate in
that  it does not account for superequilibrium N and 0 concentrations.
Nonetheless, the model has proved quite useful as  a predictive and design
tool.
       In  another effort, Quan, et  al., (Reference 5-3)  attempted  to
derive NO   scaling relationships for industrial combustors by developing
         A
dimensionless  groups from  the basic  conservation equations for momentum,
energy,  and  species concentration.   However,  several  simplifying
assumptions  had  to be made for  their model.   First,  Quan,  et al.,  employed
a  single characteristic  length  to represent  the composite  of burner  and
firebox  dimensions important in NO   formation.  Second,  the  radiation
                                  A
heat loss  from the flame in  the NO   formation region was greatly
                                  A
simplified.   Also,  the  kinetic  scaling neglected  the effects of  turbulent
mixing  and hydrocarbon/NO   coupling in NO  formation.   The results
                          J\               J\
predict  NOX emissions  as proportional  to  combustor characteristic
 length,  which is unrealistic.   Thus, such simple  scaling exercises are
 limited  in their abilities to model highly coupled phenomena such as NO
                                                                        A
formation.
        Quan, et al.,  (Reference 5-4) also attempted to predict NOY
                                                                  X
 formation by solving the governing  partial differential equations for  a
 2-D combustor.  2-D elliptic flow code was used with approximate models
 for turbulence, radiation heat transfer,  and kinetics of hydrocarbon
 oxidation and NO  formation.  Unfortunately, the  results were
                 A
 disappointing even for  the  idealized  situation studied  by the author.
 N0¥  predictions were highly sensitive both to the finite difference  grid
   A
 selected and  to the form of the physical/chemical models used.  Also,  the
 grid distribution required  to  compute the near burner events without large
 truncation  errors was impractical.  Efforts  by McDonald, et al.,  also  on  a
 2-D  combustor, encountered  similar  difficulties (Reference  5-5).
                                      5-3

-------
       Subsequent work by Quan  and  other  investigators  has been  limited as
far as application to practical  combustors.  The  phenomena to be modeled
are quite complex.  First,  it is  thought  that  turbulent mixing can
dominate the combustion and NO   formation rates in the  primary flame
                              A
zone.  Second, the hydrocarbon  oxidation  kinetics are strongly coupled to
NO  formation in the near burner  region,  and simple Zeldovich kinetics
  y\
are inadequate.  Also, for  oil  or coal  combustion, the  droplet/particle
combustion rate is strongly coupled to  luminous radiation heat transfer.
Although predicting these phenomena is  possible in idealized lab-scale
situations, it is beyond current  capability for practical combustors where
the mixing, chemistry, and  heat  transfer  are strongly coupled.  Thus,
predicting NO  formation in practical combustors from solutions of basic
             A
conservation equations is currently not possible.
       In a more recent modeling  effort,  Dykema (References 5-6 and 5-7)
adopted a semi empirical approach  employing a furnace model and a simple
Zeldovich mechanism for thermally generated NO , coupled with a largely
                                               A
empirical model for the conversion of fuel bound nitrogen.  Unfortunately,
the result was not very useful  as a predictive tool.  In fact, the model
often predicts wrong phenomena.   For example,  parametric variation of
excess air in Dykema's correlation showed a slight increase in NO  with
                                                                 A
decreased excess air.  Also parametric  variation of the location of
burners out of service rows indicated that middle rows out of service
would give the lowest NO  emissions in gas-fired boilers, while bottom
                        A
rows out of service would be preferable in oil-fired units.  Generally,
these conclusions do not concur with field experience.
       In addition to the more fundamental models attempted above, several
empirical approaches to test data correlation  have been pursued in recent
years.   Bartok, et a!., (Reference 5-8)  and Crawford, et al., (Reference
5-9) used a second order multiple regression analysis to correlate flue
gas NO  concentrations with a limited number of boiler operating
variables.   Because boiler design properties were not considered as
independent variables, Bartok1s and Crawford's analyses were restricted to
the individual boilers and loads  studied.
       Hoi linden, et al.,  (References 5-10 and 5-11) correlated NO
                                                                  A
emissions with the same boiler operating  variables used by Bartok.  Again,
the empirical  analysis was performed on an individual boiler basis, thus
                                     5-4

-------
limiting the applicability of the correlation found.   In addition,  in the
earlier study (Reference 5-10), the author incorporated combustion  staging
only as either "on" or "off" and did not allow continuous variation of the
staging parameters.  In their later study (Reference 5-11), they included
only uncontrolled NO  emissions in their analysis, thus limiting the
                    A
regression equations to predicting NO  levels only under normal firing
                                     A
conditions.
       Cato, et al., (Reference 5-12) have also performed an empirical
study correlating uncontrolled NO  emissions from industrial size
                                 A
boilers with boiler design  and operating variables.  Although  their  data
base was quite diverse, covering more than a single boiler, their  analysis
was limited to uncontrolled  NOX emissions.   In  addition,  industrial,  not
utility, boilers were  studied.
       The  above review was intended to  highlight some of the  difficulties
in  attempting to model NO   emissions on  a fundamental  basis  as well  as
                         A
to  indicate a few  of  the practical  limitations  of reported empirical
models.  The  efforts  of the investigators cited above, though  admirable,
encountered monumental  difficulties  when attempting  to relate  real world
utility  boiler combustion  to fundamental  parameters.   The following
section  presents the  NO   Emission  Correlation  Model,  a multiple
                        A             ,-
regression analysis developed in  this  study.   The model is a crude attempt
to  correlate emissions from a host of  actual  operating utility boilers to
some  common burner and boiler design and operating  variables.   The intent
 is  by no means fundamental, but rather to highlight general trends.
 5.2    N0¥ EMISSION CORRELATION MODEL
          A
        The formation of NOX in utility boilers is a complex and,  at the
 present time, imperfectly understood phenomena.  Thus, although some
 fundamental NO  formation models are available, as discussed  above,
               A
 especially for thermal NO  , these models usually only relate  NO
                          A                                     A
 generation to such fundamental combustion variables as stoichiometry flame
 temperature and residence time of gases in the flame zone.  Because the
 flow in the furnace burner  zone of an actual combustion  source like  a
 utility boiler is  extremely complicated, it is quite  difficult to
 determine  the quantitative  changes in these fundamental  parameters
 resulting from given operational or design changes.   Thus, it is  quite

                                      5-5

-------
difficult to apply these fundamental models to N0x production in actual
combustion systems.
       Consequently, a more empirical approach to NO  correlation was
                                                    A
chosen in this study.  From the point of view of the boiler operator or
designer, it is desirable only to establish which operational and design
variables are important in controlling NO  emissions, and to obtain
                                         A
estimates on how much NO  levels will change with given changes in those
                        A
variables.  Thus, a model based on multiple regression of existing data on
boilers would correlate NO  emissions to specific boiler variables and
                          A
serve these desired ends.  Such a model based on data from boilers under
actual operating conditions would be expected to reproduce the average
response of field boilers, in general, if the sample chosen for study is
representative of the field population.  The following sections outline
the development of the correlation algorithm employed and the data base
used for analysis.  It is noted that since the data base is limited, the
model should only be examined for general trends.  Indeed it should not be
considered predictive, but interpolative.
5.2.1  Procedures
       A large number of operational and design variables may be
postulated to affect NOV formation.  A regression analysis can help in
                       A
screening these variables to determine which ones are most significant in
controlling NO .  Moreover, as many of the variables are highly
              A
intercorrelated, the analysis should incorporate a selection mechanism
whereby only the variables most strongly correlated with NO  emissions
                                                           A
enter the regression and the other intercorrelated variables are
excluded.  The model would, therefore, identify the important independent
variables and quantify the change in NO  emissions due to specified
                                       A
changes in the magnitude of the variables.
       A second order regression model was used to fit the NO  emission
                                                             A
data compiled as discussed below.  A second order model was required
because NO  formation mechanisms are usually nonlinear with the
          A
fundamental parameters, therefore, a first order model does not correlate
the data very accurately.  Also, as there is expected to be some degree of
interaction between the effects of the variables, cross product terms need
to be included in the model.  A second order regression model includes
                                     5-6

-------
quadratic and cross product terms.  The predictive equation in such a
model takes the following form:

                                 2       2
       y = aQ + ajXj + ... + a^ + a22x2 +  ... + &l2*i*2 +  '"

where y is the response variable, in this case the NO  emission level
                                                     A
x-j, X2 ... are the independent boiler variables and  a , a,, ...
are the coefficients to be determined.  The coefficients are  determined by
obtaining the best fit to the  data as defined by minimizing the sum of the
squares of the distance between  the predicted values and the  data  (least
squares fit).
       By choosing a second order model,  it was assumed that  terms of
order higher than  quadratic are  not  important in  the analysis,  and that
exclusion of cross product terms with more  than two  variables does not
significantly  affect the  accuracy of  the  prediction. A third order model
could have been  constructed in a fashion  similar  to  the second order
model, but the number  of  terms required would make the  analysis too  long
and  unwieldy.
       Some  of the assumptions underlying the regression  analysis  are that
the  distribution of  each  variable is  normal,  and  their  joint  distributions
are  also  normal.   Also,  all variables must be homoscedastic;  that  is, the
variance  of  each variable must be uniform over  the sample space.   Other
assumptions  inherent in  the  analysis  are that:   (1)  no important boiler
variables were overlooked by  the investigators, (2) the data selected for
use  in  the regression  were representative of typical boilers and operating
 conditions,  (3) the data reported were accurately measured and reported,
 and  (4)  the  data from different boilers and  different tests  all had
 comparable errors of measurement.
        In one case, a second  order model did not yield the desired degree
 of precision in predicting data.  In that instance, a logarithmic model
 was  used.  The logarithm of NO  emissions are correlated linearly to the
 logarithm of the dependent variables,  and the predictive equation then
 takes the form:
             1 og y = aQ + a^ 1 og x, + a^  1 og X2 +  a~  1 og x., +

                                      5-7

-------
which is equivalent to

                                     al    a2    a3
                           jr — C  X*   Art     4    • • •

       The assumptions underlying a  logarithmic model are similar.
Specifically, the logarithms of the  variables  are  assumed normally
distributed and the variances based  on the logarithms of the variables are
homoscedastic.  It is very unlikely  that these conditions will ever be met
in practice.  But, the procedure can be justified  as useful in screening
variables and providing guidelines on the expected magnitude of changes in
NO  emissions due to changes in boiler variables.
  A
       The multiple regression analysis was performed using the stepwise
regression procedure (References 5-13 and 5-14).   In this procedure
variables are introduced into the correlation one  at a time in order of
most significant correlation.  In the specific procedure employed, a first
order linear multiple regression analysis, using the stepwise procedure,
was first carried out to identify the seven most important variables.
Second order analysis was then performed.
       As each new variable was introduced into the regression, the
multiple correlation coefficient of  the regression up to that point was
calculated.  The stepwise procedure was terminated when the increase in
the multiple correlation coefficient with addition of new variables became
sufficiently small.  As the multiple correlation coefficient is directly
related to the square of the variation explained by regression, this
criterion tends to inhibit variables which do not contribute substantially
to decreasing the standard error of  the estimate from entering the
regression.
       Multiple regression equations obtained  in this manner were examined
to check whether NO  emissions predictions were within desired degrees
                   /\
of accuracy.  If sufficient precision could not be obtained with a second
order model, a logarithmic model was employed.  Thus, in one case the
logarithmic model  yielded better predictive correlations than the second
order model.
5.2.2  Data Base
       For emissions correlations through the procedure described above,
uncontrolled and controlled N0y combustion data were obtained from a
                              A
                                     5-8

-------
total of 61 boiler firing type/fuel combinations.  Table 5-1 breaks out
the test data combinations employed.  Data gathering was limited to
tangential, horizontally opposed and single wall firing types as these
were the most extensively tested.  Fortunately, a representative
population was treated because these firing types represent approximately
87 percent of the current installed utility steam generating capacity.
NO  emissions from turbo furnace, cyclone, vertical, and stoker furnaces
  A
were not analyzed in depth because published emissions data from these
boilers were very limited.
       Emissions data were assembled from emissions field test programs
sponsored by EPA and by several  private utility companies.  In many cases,
previously unreported test data  were incorporated.  References 5-6, 5-8
through 5-11, and 5-15 through 5-19) supplied  the test data assembled.
       Several specific units were tested in more than one program.   In
these instances each individual  program was considered as a separate
unit.  Therefore, the totals in  Table  5-1 include boilers that were tested
more than once during different  programs.  The additional sets  of
emissions data from these units  were considered  valuable  in the  present
analysis because different levels  of NO   control were  usually  achieved
                                       A
from test program to program.   In  addition, baseline  and  controlled boiler
operating conditions varied  slightly between  separate  test  programs on the
same boiler,  thus providing  more representative average  operating
conditions and emission  levels.
        Table  5-2  itemizes the  actual test points incorporated  into the
emissions  data base.  The total  of 563 tests  represents  approximately 25
percent  of the total number  of tests reported in the  various  test
programs.  The table shows that the largest  number  of selected tests,
comprising 54 percent of the data, were  on  coal-fired utility boilers.
Gas-fired  and oil-fired  boilers were  studied  in 24  and 22 percent  of  the
test points,  respectively.   The single most  studied category was
tangential  coal-fired  steam  generators with  147 individual  tests.
        Six single control  techniques were considered  in the analysis:
        •   Low  excess  air (LEA)
        •   Overfire air  (OFA)
        •   Biased burner firing (BBF)
        •   Burners  out  of service (BOOS)
                                      5-9

-------
           TABLE 5-1.  FIELD TEST PROGRAM DATA COMPILED
Fuel
Coal
Oil
Natural Gas
Total
Firing Type
Tangential
13
2
1
16
Opposed Wall
6
7
8
21
Single Wall
10a
7
7b
24
Total
29
16
16
61
alncludes two wet bottom furnaces
blncludes one unit originally designed for coal firing with a
 wet bottom furnace
                               5-10

-------
                  TABLE  5-2.   INDIVIDUAL  TEST POINTS  CORRELATED
Firing Type
Tangential
Opposed
Wall
Single Wall
Tangential
Opposed
Mall
Single Wall
Tangential
Opposed
Wall
Single Wall
All Boilers
Fuel
Coal
Coal
Coal
Oil
Oil
011
Nat gas
Nat gas
Nat gas
All fuels
Baselineb
21
8
18
1
6
4
1
7
5
71
Single Controls
LEAC
29
11
23
--
5
6
1
9
4
88
oscd
46
11
29
1
11
f,
--
18
9
130
FGRe
--
7
--
—
2
4
2
--
2
17
Low
Loadf
24
7
19
1
7
8
2
13
7
88
Combined Controls3
Low load
+ OSC
27
S
19
1
7
6
1
13
7
86
Low Load
+ FGR
-
1
--
1
5
10
5
3
3
28
OSC +
FGR
—
2
--
--
2
10
1
3
4
22
Low Load +
OSC + FGR
—
—
--
1
11
8
-
8
5
33
Total
147
52
108
6
56
61
13
74
46
563
Low excess air  also generally employed
Baseline = no controls applied; boiler load near or  at maximum rating; excess  air at
normal  or above normal settings
LEA =  low excess air setting
OSC =  off stoicMometrlc combustion (Includes:  biased burner firing, burners  out of
service, overflre air)
FGR «  flue gas  recirculatlon; generally includes low excess air setting
Load less than  80 percent of maximum continuous rating (MCR)
                                                  5-11

-------
       •   Flue gas recirculation (FGR)
       •   Load reduction
Data on applying other NO  reduction techniques, such as water
                         /\
injection, mill fineness setting, and reduced air preheat, were
occasionally reported.  However, these were not  included in correlations
because data were limited, and these techniques  are considered of lesser
priority for study in the present analysis.  At  the time this correlation
analysis was performed, little boiler data from  coal-fired units
retrofitted with low NO  burners were available  (References 5-15, 5-20,
                       X
and 5-21).  These data were too limited for statistical treatment.  In
addition, the correlation algorithm derived included no single variable
able to distinguish between conventional circular and dual register burner
designs.
       Virtually all  tests reported in which single controls or
combinations of these controls were applied in various degrees were
included in the data base.  However, test points were excluded from the
analysis if they failed two general selection criteria:
       t   Were NO  reductions representative of the unit tested?
                  A
       •   Were other boiler operating parameters, e.g., register settings
           held nominally constant within normal ranges?
       For example, test points were rejected as failing the first
criterion if the test crew reported inconsistencies between these points
and the remainder of the test program.  Test data describing lowest NO
                                                                      A
levels achieved on a given unit were in general  included.  However, if the
test report noted that operation at these levels was deemed unsafe by
plant personnel, the data were rejected.
       Similarly, test data were excluded if other boiler operating
parameters, not explicitly treated in the correlation algorithm, were not
held to nominally constant values.  For example, several test series
investigated the effects of burner register settings on NO  emissions.
                                                          A
Changing register setting causes variations not  only in burner swirl but
also in airflow through the burner.  However, since register setting was
not treated explicitly in the emissions correlation model, these test
series were excluded from the data base.  Only tests with "normal" and
nominally constant register settings were included.  Similarly, tests on

                                    5-12

-------
tangential  units which varied burner or overfire air port tilt were
excluded.
       Finally, for BOOS tests, only data taken with burners in the top
rows removed from service were included in the correlation data base.  All
other BOOS patterns were disregarded.
       Based on the above, only about 25 percent of the total reported
test data were suitable for inclusion in the assembled correlation data
base.  Of the data excluded, much of it was due to insufficient
information, e.g., the boiler's heat release rate could not  be obtained.
As Table 5-2 shows, a total of 71 baseline tests were chosen to represent
normal, uncontrolled boiler operating conditions.  For comparisons of the
effectiveness of the individual NOY controls studied, 88  LEA firing
                                  ^
tests, 130 off stoichiometric  combustion  (OSC)  tests, 17  F6R tests,  and 88
load reduction (load at less the 80 percent of  unit MCR)  tests were
included.  Off stoichiometric  combustion  in its various  applications  (BBF,
BOOS, and OFA) was by far  the  most  extensively  tested combustion
modification technique.   In contrast,  test data for FGR  as  a single  N0x
control  were insufficient for  a  good  statistical  analysis of the  effects
of  this  technique.
       The  purposes of  performing the  NO   correlation  analyses were
                                         J\
twofold.  Of course,  good statistical  evaluations of  the effectiveness  of
commonly applied  combustion  controls  singly  and in  combination,  were
desired.  But  regression  relationships between NO  emissions and  more
                                                  rt
fundamental  combustion,  boiler design,  and operating  parameters  were also
sought  through the model, to highlight general  trends.
        Thus, specific data on a set of design and operating variables
associated  with each  test point in  the data base were needed to  allow
correlation relationships to be obtained.  These correlation variables
used in the analyses  fell into three categories:
        •   Boiler operating variables
        •   Boiler design variables
        •   Fuel properties
 Boiler Operating Variables
        Correlation parameters in this category are the macroscopic
 combustion variables describing boiler operation, which  are altered when a

                                     5-13

-------
 combustion  control  is  applied.   The  specific  variables  used  in  the  present
 analysis  included:
       •    Overall  furnace  fuel/air  stoichiometry
       t    Stoichiometry  at active burners
       •    Percent  flue gas recirculated
       0    Firing rate  (as  percent MCR)
       t    Percent  burners  firing
       •    Heat  input  per active burner
       For  example, an LEA  application can  be  trivially considered  as  a
 change in overall furnace stoichiometry.  Similarly,  applying OFA alters
 burner stoichiometry and  perhaps also overall  stoichiometry  if  higher
 overall excess air  levels are required.  Burners out  of service firing
 elicits similar  changes while also altering percent burners  firing.  Flue
 gas recirculation is applied by  changing the percent  gas recirculation
 variable.   Load  reduction is obviously accompanied by changes in firing
 rate and oftentimes heat  input per active burner and  overall stoichiometry.
 Boiler Design Variables
       The  boiler design variables considered  in the  regression analysis
 included:
       •    Nameplate maximum continuous rating  (MCR)
       •    Volumetric heat  release rate
       •    Surface  heat release rate
       •    Heat  input per active burner
       •    Number of burners
       •    Number of furnaces
       •    Number of division walls
 Table 5-3 lists  the ranges  and average values  encountered for each  of
 these variables.  Boilers of the same fuel and  firing type were grouped
 together in the  correlation analysis.
       Data on burner zone  surface heat release rate  were generally
 unavailable for  the test reports cited.  This  is unfortunate since  NO
                                                                     A
emission levels  are expected to be stronger functions of this variable
than of the more global overall heat release rate (Reference 5-22).  In
fact, adjusting  this variable alone  allows a significant degree of  NO
 control for gas- and oil-fired boilers.  Still, use of  the overall

                                     5-14

-------
                  TABLE  5-3.   BOILER DESIGN  VARIABLES  CONSIDERED




Equipment
Typ«


Tan-
gential




Opposed
Mill




Single
Hill






All
Boilers









Fuel
CM!

011

Natural
6«
CM!

0<1

Natural
6n
Coil

on
Nituril
Git
Coil

011



Sis

All
FutU
Ma>UH«
Cont 1 nuous
Rating, m*



Range
125-800

(6-320

920

218-820

220-410

220-600

100-340

80-250
80-315

100-820

66-480



80-600

66-820



Avenge
430

193

120

580

320

350

200

190
200

430

234



290

310

Volunetrtc Heit
Release Rite. kU/«3
HO* Btu/ft3-hr)»



Range
116-159
(11-15)
289-310
(28-30)
289
(26)
134-178
(13-17)
255-297
(25-29)
152-287
(15-28)
138-242
(13-23)
198-299
(19-29)
181-282
(18-27)
116-242
(11-23)
198-310
(19-30)


152-289
(15-28)
116-310
(11-30)


Avenge
139
(13)
300
(29)
289
(28)
157
(15)
270
(26)
254
(25)
196
(19)
250
(24)
232
(22)
164

273
(26)


258
(25)
232
(22)
Surfice Heit
Release Rite, ui/m>
(103 Btu/ft'hr)'



Range
78-466
(25-148)
349-541
(111-172)
541
(172)
228-312
172-99)
204-625
(65-199)
204-604
(65-192)
110-455
(35-145)
248-778
(79-247)
248-324
(79-103)
78-466
(25-148)
204-778
(65-247)


204-604
(65-192)
78-778
(25-247)


Average
228
(72)
443
(141)
Stl
(172)
259
(82)
422
(134)
395
(126)
236
(75)
343
(109)
265
(84)
241
(76)
402
(128)


400
(127)
348
(110)
Heat Input per
Active Burner,
W (10b Btu/hr)t>



Range
13-75
(45-261)
11-38
(38-133)
38
(133)
33-86
(115-299)
26-81
(90-282)
2S-63
(87-219)
21-55
(73-191)
22-43
(77-150)
21-43
(73-150)
13-86
(45-299)
11-81
(38-282)


21-63
(73-218)
11-86
(38-299)


Average
40
(139)
24
(85)
38
(133)
49
(170)
44
(153)
42
(146)
30
(105)
35
(122)
33
(115)
40
(138)
34
(118)


38
(132)
37
(129)

Total No.
of Burners



Range
16-64

8-24

24

20-54

12-24

12-36

16-24

12-24
12-24

16-64

8-24



12-36

8-64


fyptcil
Umber
32

16

24

34

20

22

18

14
16

28

28



20

22




Range In
NMber of
Furnaces
1-2

1-2

2

1-2

1

1

1-2

1-2
1-2

1-2

1-2



1-2

1-2



Range In
Nu»ber of
)1«ls1on
Halls
0

0

0

0-1

0-1

0-1

0-1

0-1
0-1

0-1

0-1



0-1

0-1

•Electrical output
"At lutaHX continuous rating
                                             5-15

-------
parameter, for which data were generally  available, did allow reasonably
good NO  correlation.
       X
       Heat input per active burner could  also  be considered a boiler
operating parameter but was grouped here with design  variables for
convenience.  As Table 5-3 shows, heat  input per active burner varied  in
the data base from 11 to 86 MW.  Horizontally opposed boilers in general
recorded the highest values, probably because of the  generally greater
unit size, hence burner size, of these  units.   The value of this variable
can be changed when applying biased burner firing, burners out of service,
and load reduction.  It should be noted, though, that load reduction can
be accomplished by totally removing burners from service, in which case
heat input per active burner could remain  unchanged.
       The number of furnace division walls was introduced as a design
variable as part of a crude attempt to  account  for gross changes in
furnace mixing patterns and burner zone surface heat  release rate.  The
use of dividing water walls allows boiler  designs with smaller surface
heat release, at a relatively constant  volumetric heat release rate.
Ideally, effects of division walls on NO   emissions would be picked up
                                        A
in the regression through these two variables directly.  However, since
these water walls generally separate the furnace only part of the way  up
to the convective passes, they also affect burner zone heat release rate
at constant overall heat release rate.  In addition,  gas mixing patterns
are altered from units of similar size, but of  divided design, with
corresponding effects on NOV emissions.
                           A
       For similar reasons, the number  of  furnaces was included as a
correlation variable.  Twin furnace design is most prevalent in larger
tangentially fired units, though it is  occasionally found with other
firing types.  It should be noted here, though, that  two potentially
important boiler design variables known to affect NO  emission levels
                                                    A
were not included in the correlation analysis because the data were not
available.  These are burner spacing and distance between the top burner
level  and the OFA ports.  Both of these variables can have significant
effects on NO  production in a given unit  by affecting flame
             A
interactions, gas mixing, and heat absorption in the  burner zone.
Furthermore, in BOOS and OFA applications, these variables will affect
first and second stage residence time and  separation.  Unfortunately,
                                    5-16

-------
these data were unavailable for most units tested in the test reports used
in compiling the emissions data.  Thus, they were not included in the
analysis.
Fuel Properties
       The fuel variables considered in the correlation analysis were
nitrogen content, moisture content (coal only), and heating value.
Unfortunately, the information on these fuel properties was not  always
available for each test point.  In cases where fuel analyses were sparse
at the individual run level, they were assumed constant throughout  a
series of tests on a specific boiler.
       Table 5-4 lists the fuel properties  considered  in  the present
analysis  and their average values.  Fuel  nitrogen for  all  coals  tested
varied by a factor of 3 from 0.62 to 1.84 percent by weight.   Even  though
this  represents a significant range in fuel  nitrogen content,  NO
                                                                 A
emissions were  found not  to be  significantly affected  by  fuel  nitrogen
content.
       Moisture content  of coals  also  varied significantly from  1.14 to
36.4 percent.   This  is  to be  expected  since coal  types used in various
tests varied  from the  low moisture  content  Eastern  bituminous to the high
moisture  content Western sub-bituminous  and lignite coals.  Coal moisture
content was also not found to affect  NO   emissions.
5.3    NOY  EMISSION  CORRELATION RESULTS
          rt
        In this section,  the results of applying the correlation model to
the data  base of test  results are discussed.  Key boiler design and
 operating variables, burner characteristics, and fuel  properties which
 affect NO  formation are identified.   The basis and effectiveness  of the
          A
 various NO  control  techniques are reviewed.  These results are further
           A
 discussed in the light of fundamental combustion principles and boiler
 operating practice.
        The major boiler firing types, tangential,  single wall,  and opposed
 wall fired, with the principle fuels, coal, oil, and  gas were treated.
 However, tangential  oil- and gas-fired boilers were not  considered in  the
 correlation study as the data  were insufficient for a statistical  analysis.
 5.3.1  Tangential Coal-Fired Boilers
        A multiple regression analysis was  carried  out on tangential
 coal-fired boilers.  Data were analyzed  for 147 tests carried out  on a
                                      5-17

-------
                                         TABLE 5-4.   PROPERTIES OF FUELS FIRED
tn
i
CD
Equipment Type Fuel
Coal
Tangential
Oil
Coal
Opposed Wall Q11
Coal
Single Wall
Oil
Coal
All Boilers
Oil
Fuel Nitrogen, Percent by Weight
Range Average
0.6-1.6 1.2
0.3-0.6 0.5
1.0-1.8 1.3
0.2-0.4 0.3
0.8-1.5 1.3
0.2-0.3 0.3
0.6-1.8 1.3
0.2-0.6 0.3
Fuel Moisture, Percent by Weight
Range Average
3.4-31.9 12.5
1.1-36.4 7.2
4.5-28.9 8.9
1.1-36.4 9.5
Heating Value3, MJ/kg (103 Btu/lb)
Range Average
19.0-32.3 27.2
(8.19-13.9) (11.7)
NA NA
24.4-31.6 28.3
(10.09-13.57) (12.2)
43.7-45.8 44.2
(18.8-19.7) (19.0)
23.0-32.8 28.8
(9.9-14.1) (12.4)
43.7-45.6 44.7
(18.8-19.6) (19.2)
19.0-32.8 28.1
(8.19-14.1) (12.1)
43.7-45.8 44.4
(18.8-19.7) (19.1)
                 Dry basis

-------
total of 13 boilers.   The data included 21 tests performed under baseline
conditions with the rest conducted under low NO  conditions.  Low NO
                                               ^                    A
techniques tested included LEA, OSC, low load and a combination of low
load and OSC.
       For tangential coal-fired boilers, the following equation
correlates the data with a correlation coefficient of 0.87, i.e.,
75 percent of the variance is explained by the regression:

       y = 184 + 1.09 x 10"7(X1)(x2) - 1.67 x 10"5(/1) +
           2.49 x 10"6(x3)(x4) + 6.54  x lO"14^)2

where
        y   =  NO  emissions  (ppm dry  at  3  percent  09)
               A                                   £
        x,  =  Heat  input  per active burner  (W)
        x«  =  Stoichiometry to  active burners  (percent  stoichiometric air)
                                           2
        *3  =  surface  heat  release  rate  (W/m )
        x.  =  Furnace  Stoichiometry (percent stoichiometric air)
        From  the  regression equation it is seen that burner Stoichiometry
 and heat  release  rate are the most  important parameters governing NO
                                                                     /\
 emissions in these boilers.   This is in agreement with fundamental
 combustion principles as Stoichiometry affects both thermal and fuel NO
                                                                         y\
 while heat release should mainly affect thermal NO .   The equation
                                                   /\
 indicates that,  in general,  NO  emissions will be reduced by  decreasing
                               ^
 both Stoichiometry and heat  release, which is also consistent with theory.
        A graphical representation of how NO  emissions vary with  surface
                                            A
 heat release rate and burner Stoichiometry is shown, in Figure 5-1.  The
 parametric  lines in the figure are generated from the regression  equation
 by  allowing surface heat release to vary while  fixing the  burner
 Stoichiometry at the values shown  beside the curves.  All  other
 variableswere held constant  at their mean values, except for  furnace
 Stoichiometry which was taken equal to burner Stoichiometry for  non-OSC
 operation,  and fixed at  some  lower  limit  (typically 120  percent)  for  OSC
 operation.  The parametric lines are  seen to match reasonably closely with
 the data  points.  The  reduction  in NOX emissions with reduced burner
 Stoichiometry and surface heat release  rate  is  clearly  evident from the
 figure.
                                      5-19

-------
en
i
ro
o
                   7001
                   600J
                CM
               c
               »*
               CO
               §.
               a.
               tn
               c
               o
                   100
                                          Surface heat release rate (kW/m2)
                              25        50      75       100      125      150    ~175"


                                        Surface heat release rate (106 Btu/hr-ft2)
                                                                                       140
                                                                                       640
200
            Stoichiometry to active

            burners  (percent)
         O 140

         CD 120

         X 100

         Z  80
                      Figure  5-1.   Effect  of surface heat release rate and  burner  Stoichiometry

                                      on NOX  from tangential coal-fired  boilers.

-------
       Figure 5-2 shows another graphical  representation of NO
emissions variation with burner stoichiometry and heat input per active
burner.  The parametric curves were generated from the regression equation
in a manner similar to that explained above for Figure 5-1.  Again, the
effect of decreasing NO  with decreasing burner stoichiometry is clearly
                       rt
seen.  Decreasing heat input per active burner, however, seems to have a
mixed effect on NO.  For burner stoichiometry above 120 percent, NO
                  X                                                 X
emissions decrease with reduced heat input, consistent with earlier
discussion.  Note that burner  stoichiometries above 120 percent  generally
preclude OSC operation.  For  burner stoichiometries about  100 percent, the
NO   emissions sometimes actually decrease  with increasing  heat  input  per
  /\
burner.  This can be  explained by  noting that the data  points for  burner
stoichiometries  at  about 100  percent or lower include tests with BOOS
operation.   In such cases,  increasing  heat input per  active burner is
tantamount  to  increasing the  degree  of off stoichiometry,  as fuel  flow to
active burners must be increased  under BOOS operation to maintain  load.
Under these circumstances  the NO   emissions should decrease with
                                 A
 increasing  heat  input per  burner,  and  that is precisely what is observed.
        The  above example points out  the need to  be very careful in
 interpreting the results of the regression analysis.   The equations are
 valid only within the range of conditions of the original data  base,
 sothat any generalizations should be made with caution.   It  should also be
 noted that the independent variables are often related to  each  other
 within certain ranges of operation.   For example, LEA operation, without
 OSC, will influence both burner as well as furnace stoichiometry.  Also,
 low  load, without BOOS, will  affect both  the surface heat  release rate and
 heat  input per  active burner.  And, as pointed  out earlier,  BOOS  operation
 will  affect burner stoichiometry  and  heat  input  per  active burner.
         It  is seen  from the regression  equation  and from Figures 5-1  and
 5-2  that the most  effective  operational technique for  NO   control on
 tangential  coal-fired boilers seems to be reduction  of burner
 stoichiometry.   This  can  be  accomplished  to a certain  extent with LEA and
 to  a greater extent  with  OSC.  Lower  surface  heat release rates also
 result  in  lower NOX  emissions.   In  addition,  lower-heat release per
 burner  also tends  to reduce  NOX emissions, at  least  when not operating
  under BOOS firing.
                                      5-21

-------
                    700
PO
ro
                 o
                 S*
                  £ 500
                 -
                  §.
                  Q.
                  j; 400
                  c
                  o
                  x 300
                 o
                    200
                    100
                       10        20       30       40       50       60       70
                                         Heat input per active burner (MW)
                                                                                       , 120
                                              80      90
                                                                                                   Stoichiometry to active
                                                                                                   burners  (percent)
                                                              (D 140
                                                              m 120
                                                              X 100
                                                              Z  80
                        40
80
120
160
                                                              200
                                     240
                                      Heat input per active burner  (10  Btu/hr)
                                                                                 280
                         Figure 5-2.   Effect of  heat  input and burner stoichiometry on  NOX  from
                                        tangential  coal-fired  boilers.

-------
5.3.2  Horizontally Opposed Coal-Fired Boilers
       The multiple regression analysis was also applied to horizontally
opposed coal-fired boilers.  Fifty-two tests on six boilers were selected
for the analysis.  The data included tests performed under baseline
conditions as well as LEA, OSC, low load, and a combination of low load
and OSC.  In addition, data from a boiler tested with FGR were included.
Some test data on a combination of low load and FGR, and OSC and FGR were
also included.
       The regression analysis yielded the following equation, which has a
correlation coefficient of 0.91, i.e., 83 percent  of the variance  is
explained by the expression:

       y  = -471  + 5.38(x1) +  4.24  x 10"6(x2)  +  7.41(x3)

           -5.84(x4)  - 6.64 x lO1^)  +  2.46  x  lO^Xg)

where
        y  =  NOX  emissions (ppm dry at 3  percent 02)
        x-, =  Stoichiometry to  active  burners (percent stoichiometric air)
        x2 =  Heat input per active  burner (W)
        Xo =  Number of burners firing
        x^ =  Flue gas recirculation (percent)
        Xj- =  Number of division walls
        Xg =  Excess oxygen (percent)
        The regression equation indicates that NO   increases with
                                                 n
 increasing Stoichiometry to burners, heat input to the burners, the number
 of burners firing, and overall excess oxygen, whereas NO  decreases with
                                                         n
 increasing flue gas recirculation and number of division walls.   These
 results  are, in general, in  agreement with past experience and theoretical
 considerations.  Burner  Stoichiometry and overall  excess air  are  known  to
 have a large positive correlation with  NO  formation.   Increased  heat
                                          ^
 input to burners would also  be expected to increase NOX emissions.  The
 positive correlation  of  a number  of  burners  firing with NO   probably
                                                           A
 stems from many factors.  Larger  boilers  produce  significant  NO   and
 usually  have more  burners.   At partial  loads NOX  generation  is  reduced
 and so  are  the  number of active burners.   Finally, with BOOS, the number

                                      5-23

-------
of active burners decreases and so does NOX<  The number of division
walls is negatively correlated with NOX.  This  is most likely due to the
increased surface area available for heat transfer with consequent
lowering of flame temperatures.
       The correlation of NOY emissions with FGR is interesting and is
                            A
shown in Figure 5-3 with burner stoichiometry as a parameter.  Although
the data are relatively sparse, the statistical correlation do point to a
negative trend of NO  emissions with increasing FGR.  It should be noted
                    A
also that the decrease in NOX due to 20 percent FGR is approximately the
same as the decrease in NO  with a 10 percent reduction in excess air at
                          A
the burners.  FGR is known to inhibit thermal NO , whereas OSC controls
                                                A
both thermal and fuel N0y.  OSC is therefore expected to be a more
effective NO  control technique than FGR, in agreement with experience.
            A
       The effect of heat input per active burner on NO  emissions is
                                                       A
shown in Figure 5-4, again with burner stoichiometry as a parameter.  The
data scatter is rather large and very few data points are available for
the substoichiometric region.  Nevertheless, it is seen that, in general,
increasing heat input per active burner increases NO .  It also
                                                    A
indicates the'influence of burner stoichiometry on NO  emissions.  It
                                                     /\
should be reiterated here that the data base is limited, and that data
from different manufacturers were incorporated together.  Hence design
differences between burners are masked in the correlations.  Thus a large
burner (high input) does not necessarily produce high NO .  Indeed the
                                                        A
new burners coming onstream today have designs that limit air/fuel mixing
in the burner zone and hence limit NO  production.  In other words,
                                     A
burner design can overcome the tendency of higher NO  with increasing
                                                    A
heat input (Reference 5-24).
       From the regression analysis, it can be seen that for horizontally
opposed coal-fired boilers, reducing burner stoichiometry is a very
effective means for controlling NOX emissions.  LEA reduces burner
stoichiometry, but OSC must be employed if large reductions up to or below
the stoichiometric level are desired.  FGR also reduces NO  but to a
                                                          A
lesser extent than reduced burner stoichiometry.  The implications for
boiler design from this study are that increased cooling surface and
decreased heat input per burner tend to decrease NO  emissions.  But as
                                                   A
noted above, burner design, though not included in the correlation, is
                                    5-24

-------
en
i
ro
en
                   400
                                       8       12      16      20       24


                                           Fuel gas  recirculatlon (percent)
28
                                                                                                 Stoichiometry to active
                                                                                                 burners (percent)
                                                                                              QJ 140

                                                                                              CD 120

                                                                                              A 100

                                                                                              X  80
                          Figure 5-3.   Effect of FGR  and burner Stoichiometry  on NOX  from
                                         horizontally opposed  coal-fired boilers.

-------
                    1000
PO
CTl
                        20
30      40       50       60       70      80

          Heat input per active burner (MW)
                           80       120        160      200      240       280

                                       Heat input per active burner (10  Btu/hr)
                                                 320
                                                                                         100
360
                                                                                                   Stoichiometry to active
                                                                                                   burners  (percent)
                                                                                                E 140
                                                                                                CD 120

                                                                                                A, 100

                                                                                                X  80
                        Figure 5-4.   Effect of  heat  input and burner Stoichiometry on NOX  from
                                       horizontally opposed coal-fired boilers.

-------
equally if not more important than burner heat input.  The regression
model can be used to illustrate general trends in the change of NOX
emissions with design or operational changes.  One should be very careful,
however, not to extrapolate the equation beyond the range of data on which
it was correlated.
5.3.3  Single Wall Coal-Fired Boilers
       The multiple regression  analysis was  applied  to single wall
coal-fired boilers with data from 86 tests performed on eight boilers.
The data  included tests under baseline and low NO  conditions.  Low
NO  techniques included LEA, OSC, low  load,  and  a  combination of  low
  ^
load  and  OSC.
        The regression  analysis  correlated  the data with  a correlation
coefficient  of 0.896,  i.e.,  80  percent of  the variance was explained  by
the  regression.   The equation which best  correlated  the  data was:

        y = -140  + 1.98 x  10"1(x3)(x2)  +  6.95 x 10"5(x1)(x5)  + 4.5 x 10"6(x1)(x2)

            + 7.57 x 10"8(x4)(x2)  -  1.02  x 10"11(x1)(x4)

 where
        y  =  NO  emissions (ppm at 3 percent 0~)
               x                            ?
        x, =  Surface heat release rate (W/m )
        x2 = Stoichiometry to active burners  (percent stoichiometric air)
        x-j = Number of burners firing
        x4 = Heat input per active burner (W)
        Xj- = Furnace excess oxygen  (percent)
        This regression equation is complex with many variables appearing
 several  times in conjunction with  other variables.  The  Stoichiometry
 tothe active burners  has a marked  large positive  correlation with  N0x
 emissions.  The  number of firing burners, the heat  input per active
 burner,  and the  furnace excess oxygen are all also  positively correlated
 with NO  emissions.   These  positive correlations  are consistent  with
 theoretical considerations  and past experience.   Stoichiometry  to  active
 burners  and overall excess  oxygen  have been  shown to have a marked effect
 on thermal  and  fuel NOY  generation.   The  heat  input per  burner  which  is
                        ^
 related  to  the  flame  intensity and,  hence,  peak temperatures should affect
                                      5-27

-------
thermal NO  emissions.  The number of firing burners  increases with boiler
          A
size, high load, and absence of BOOS firing, all of which tend to  increase
NO  emissions.  The effect of surface heat release rate  is  not
  A
straightforward and is discussed further below.
       Figure 5-5 is a plot of NO  emissions versus surface heat release
                                 A
rate with burner stoichiometry as a parameter.  Here  the trends are
consistent with expectations based on previous correlations.  NO
                                                                A
emissions tend to increase with increasing surface heat  release rate and
burner stoichiometry.  Figure 5-6 shows the variation of NO  emissions
                                                            A
with heat input to active burners and burner stoichiometry.  Again the
trends are consistent with expectations.  NO  emissions  tend to increase
                                            A
with increasing heat release per burner and increasing stoichiometry.  The
data are sparse for higher heat release rates, so that predictions at those
values may not be very accurate.  Nevertheless, the trends  should  be
correctly predicted.
       From the regression analysis, it is seen that  burner stoichiometry
again has the greatest effect on NO  emissions from single wall coal-fired
                                   A
boilers.  LEA can be employed to decrease burner stoichiometry to  a certain
extent.  OSC should be employed if further reduction  is  desired.
Implications for boiler design are that decreasing heat  input per  burner can
reduce NO  emissions.  But, as discussed in Section 5.3.2,  burner  design,
         A
a variable not incorporated here because of data limitations, can
predominate over heat input.  Finally decreasing heat release rate, all
other factors equal, generally does reduce NO .  The  regression equation
                                             A
can be used to estimate trends in NO  emissions due to design or
                                    A
operational changes.  However, as most of the data on which the correlation
is based are confined to a small range, care should be exercised when making
numerical predictions.
5.3.4  Horizontally Opposed Oil-Fired Boilers
       The data base for the multiple regression analysis on horizontally
opposed, residual  oil-fired units was relatively good.  The total of 56
test points from the seven boilers tested gave more than 1  test point for
each control and combination of control methods considered.  The tests
included baseline, low excess air, off stoichiometric combustion,  flue gas
recirculation, load reduction, and combinations of these control methods.

                                    5-28

-------
1000-
   20
40
 60       80      100     120      140     160
Surface heat release rate (106 Btu/hr-ft2)
                                                                 Stoichiometry to active
                                                                 burners  (percent)
                                                                0) 140
                                                                Qj 120
                                                                X 100
                                                                Z  80
        Figure  5-5.  Effect of  surface  heat release rate  and burner  Stoichiometry
                      on  NOX from single wall coal-fired boilers.

-------
                                                           140
en
 i
co
O
                      30
60
90
120
150
180
210
                                   Heat input per active burner (10  Btu/hr)
                                                        MW
240
                                                                                             Stoichiometry to active
                                                                                             burners (percent)


                                                                                            a  140

                                                                                            O  120

                                                                                            A  100

                                                                                            X  80
                         Figure  5-6.   Effect of heat input  per active  burner and burner  stiochiometry
                                       on  N0x from single wall  coal-fired boilers.

-------
       The correlation explained variations in NO  emissions to within
                                                 rt
80 percent.  The significant parameters were found to be firing rate, number
of burners firing, stoichiometry to active burners, number of division
walls,and furnace stoichiometry.  The second order multiple regression
equation which best correlated the data, with a correlation coefficient of
0.90 was:

       y = -228 + 1.05 x 10'1  (xi) (x2) +  7.23 x 10-3(x3)2
           - 1.30 (xj) (x4) +  2.392(x5)
where
       y   = NO  emissions  (ppm dry at 3 percent 0~)
              /\                                  w
       x-i  = Firing  rate  (percent)
       Xp  = Number  of  burners  firing
       X3  = Stoichiometry  to  active burners  (percent  stoichiometric air)
       x,  = Number  of  division walls  plus  one
       x5  = Furnace stoichiometry (percent stoichiometric air)
       The variable dependencies were not  unexpected.  Reducing the firing
 rate lowers  the volumetric heat release rate.   Thus,  reduced heat release
 rate will  lower the bulk gas  temperature in the furnace resulting in reduced
 thermal  NO  formation.  An increase  in the number of active burners for a
 given heat release  rate  should also  increase NO  emissions, as more active
                                                ^
 burners  firing will result in more thorough mixing of fuel and air  in the
 peak flame temperature regions.  The stoichiometry to active burners and the
 overall  furnace stoichiometry influence NO  formation in that higher
 oxygen concentrations in the  peak flame temperature  regions will  increase
 NO  formation.  As shown in Figure 5-7, increasing either boiler  load or
   A
 burner stoichiometry  increases NO  emissions.  Also, as  shown  in
 Figure 5-8, increasing either the number  of burners  firing or  the burner
 stoichiometry increases NO  emissions.  Finally, furnace division walls
 add heat  transfer  surface  to  the furnace.  The increased surface  will  result
 in greater heat  transfer from the furnace gases thus lowering  the furnace
 bulk gas  temperature.  This will reduce thermal NO   formation,  and thereby
                                                   A
 lower NO  emissions.
         /\
                                      5-31