-------
o
DC
2(0
O LU
O O
83
z <
m -J
O O
UJ U.
It
CO
(0 O
CM LU
U,H
CQ
T5
CO
o
3
•o
o
h»
X
O .^
c*
s8
s <-
il
II
+*
o
0)
•+-
03
E
IE
0 —.
N >.
'1
CO O
a 4i
JD
O
r-
r~
JD
O
m
en
CM"
o
o
O5
en"
JD
o
en
00
o
O
*3-
»t
o
oo
t
•o
a.
c
o
o
o
CO
T3
C
to
W
C
o
*4^j
jro
"ro
TO
i
ra
o
5
2-6
-------
($900 to $2,950 per ton). Oxy-firing and electric boost are the most expensive
technologies, with cost-effectiveness values up to $9,900 per ton.
2.4 IMPACTS OF NOX CONTROLS
2.4.1 Environmental impacts
None of the controls shown in Table 2-2 have any solid or wastewater disposal
impacts except for the disposal of spent SCR catalyst. Some catalyst formulations are
potentially toxic and subject to hazardous waste disposal regulations under RCRA and its
amendments. However, recent industry trends have shown that these material are readily
regenerable. In fact, many catalyst vendors recycle this material thus avoiding any
disposal problem for the user. The control technologies do have impacts on other air
pollutants.
2.4.1.1 Combustion Modifications. Combustion modifications in glass furnaces
that decrease NOX may increase emissions of CO and unburned hydrocarbons. For oxy-
firing, Table 2-5 shows an increase in SOX emissions and a decrease in CO and CH4 (a
measure of unburned natural gas) emissions, at least as measured on the basis of Ib (of
SOX, etc.) per ton of glass produced.
2.4.1.2 Process Modifications. Gullet preheat can be done using direct or indirect
contacting devices to carry out the heat transfer. For direct contact systems, in which the
flue gas comes in direct contact with the cullet, there appears to be no net effect on
particulates and some reduction of SOX by adsorption on to the cullet. For indirect control
systems, there are no impacts.
2.4.1.3 Postcombustion Modifications.
Selective catalytic reduction. For SCR, the injection of ammonia into the flue gas
inevitably results in some unreacted ammonia and some byproducts (e.g., NHg, CI2,
(NH^^SO^) in stack emissions. Such emissions generally increase with time as the
catalyst ages. In most SCR applications, unreacted ammonia ("ammonia slip") is kept
below 20 to 40 ppm by controlling the injection rate of ammonia. The injection of
ammonia may increase stack particulate emissions due to the formation of ammonium
sulfate/bisulfate and ammonium chloride, though there is of course a corresponding
stoichiometric reduction in gaseous SOX and HCI emissions.
As with SCR, SNCR generates ammonia slip and byproduct salts from the acidic
components of the flue gas. Ammonia slip in one case is reported as 13 ppm. Tests on
2-7
-------
TABLE 2-5. EFFECT OF OXY-FIRING ON AIR EMISSIONS
Conventional firing Oxy-firing
Parameter (Ib/ton glass pulled) (Ib/ton glass pulled)
Paniculate 1.19 0.884
NOX 5.03 0.812
SOX 0.612 0.968
CO 0.08 0.003
CH4 0.02 0.008
2-8
-------
another process show that SNCR
• has no significant effect on total particulate emissions
• slightly increases CO emissions, and
• slightly decreases SO2 emissions
and ammonia slip (unreacted ammonia emissions) increases with ammonia injection rate.
The same general trend would be expected for SNCR processes using urea.
2.4.2 Energy Impacts
2.4.2.1 Combustion Modifications. Data indicate that LEA operation and changes
in air/fuel contacting do not significantly affect furnace energy usage (MM Btu/ton glass
produced). Based on this, these two combustion modifications are assumed to have
negligible energy impacts. For low NOV burners, the Kortig burner is claimed to result in
s\
energy savings by reducing air infiltration, but no quantitative results are presented. Such a
claim would be difficult to quantify since air infiltration is highly site specific. Such
burners may be more efficient than others and would therefore save energy. However, a
direct comparison cannot be made with the existing data. Oxy-firing results in lower
energy consumption (MM Btu/ton glass produced). This is, in fact, one of the primary
reasons for its use. Fuel savings of 15 percent for oxy-firing on a 75 tons/day end-fired
regenerative furnace are reported. Production during the test was 58 tons/day. Further,
at essentially the same fuel usage rate, glass production increased from 62.7 to 75.8
tons/day (21 percent), as shown below:
Production
(tons/day)
Fuel usage
(MM Btu/hr)
Air-firing
62.7
13.7
Oxy-firing
75.8
13.6
This corresponds to 30 to 40 percent energy savings (Figure 2-1) for regenerative glass
furnaces, but absolute values (MM Btu/ton glass) are not provided. For the Gallo plant,
natural gas usage was 9.5 percent lower than with air-firing (3.74 MM Btu/ton with air-
firing, 3.39 MM Btu/ton for oxy-firing.
2.4.2.2 Process Modifications. Gullet preheaters are designed to recover heat
from the flue gas and therefore will reduce the energy consumption in glass melting. The
2-9
-------
Energy Penalty Energy Savings
-30%
All Electric
Regenerative
Hard Glass
40%
Regenerative
Soda-Lime
30%
Figure 2-1. Energy impact of oxy-firing.
2-10
-------
Teichmann cullet preheater is estimated to account for 8 to 12 percent of the total energy
®
saved by their Low NOX Melter , which also incorporates other energy savings features.
Insufficient information is given to determine absolute energy savings associated with the
cullet preheater alone.
Electric boost simply substitutes electrical energy for fuel in heating the glass
melt. If the efficiency of producing electricity from a fossil fuel and delivering it to the
glass melt is taken into account, electric boost is inherently less efficient than natural gas
firing and would therefore increase, ultimately, the energy requirement associated with
glass melting.
2.4.2.3 Postcombustion Modifications. There is some pressure drop across the
SCR catalyst that will require additional electrical energy for the flue gas fan. Typically,
this pressure drop is of the order of 5 to 10 in. h^O. For a pressure drop of 10 in. H^O,
and using a value of 68 scfm per ton/day of glass (see footnote b of Table 5-8) and a fan
efficiency of 60 percent, the following calculation can be made:
Plant size
(tons/day)
50
250
750
Fan energy
(kW)
6.6
33.2
99.4
If the flue gas temperature at this point is below 350 to 500 °C (660 to 930 °F), the gas
may need to be reheated with gas burners. This highly site-specific energy impact is not
considered further here.
SNCR requires no additional pressure drop for flue gas transport but ammonia or
urea are injected in liquid form at high pressure to ensure efficient droplet atomization and
dispersion. Liquid ammonia or urea must be vaporized with heat mixed with carrier gas(air
or steam) and then injected for adequate mixing.
2-11
-------
CHAPTER 3
GLASS MANUFACTURING
3.1 BACKGROUND
Glass is a material made by cooling certain molten compounds in a way in which
they do not crystalize. Glass viscosity at ambient temperature is so high that for all
practical purposes it is solid. Materials having the ability to cool without crystallizing are
rare, silica compounds being the most common. Essentially all glasses of commercial
importance are based on silica.
This chapter describes the furnaces associated with the melting and fabrication of
container, flat, and pressed/blown glass. Fiberglass is not included. These furnaces carry
out certain chemical reactions at extremely high temperatures in a melting furnace.
Although the furnace geometry, firing pattern, heat recovery techniques, and specific
temperatures vary depending on the type of glass produced, all glass furnaces operate at
temperatures where NOX formation takes place.
3.2 GLASS MAKING
Despite differences in the final products, all glass is manufactured by a process in
which the raw materials are mixed and then melted in a furnace. Glass is produced by
first mixing dry ingredients in what is known as a batch. In most large furnaces this batch
is mixed and fed in a semicontinuous way to one end of the melting furnace. In the
melting furnace chemical reactions take place between the batch ingredients. The main
o
reactions can be summarized as follows :
3-1
-------
N32CO3 + aSiC-2 •* Na2O • aSiC>2 + C02 (3-1)
*
CaCOs + bS\C>2 -» CaO • 6SJ02 + CO2 (3-2)
N32SO3 + cSiO2 -* Na2O • cSi02 + SO2 + CO . (3-3)
The heat for these reactions is usually supplied by natural gas burners that are fired
over the glass melt. Heat is transferred primarily by radiation from the flame to the surface
of the melt. The configuration of the furnace is generally end-port or side-port. These are
o ;
shown in Figures 3-1 and 3-2. In the end-port furnaces, the flames travel in a U-shape
over the melt from one side and flue gases exit the other. These furnaces are generally
used in the container and pressed/blown industries. In the side-port furnaces used in flat
and container glass products, the flames travel from one side of the furnace to the other.
In both cases, refractory-lined flues are used to recover the energy of the hot flue gas.
The high temperature of the flue gas exiting the furnace heats the refractory material
called a checker. After the checker has reached a certain temperature, the gas flow is
reversed and the firing begins on the other side (or end) of the furnace. The combustion
air is then preheated in the hot checker and mixed with the gas to produce the flame. The
combustion air preheat temperatures in flat glass furnaces can reach 1260 °C (2300 °F)
and substantial NOV can be formed in the checkers. Lower preheat temperatures are used
y\
in container glass, and NOX contributions in the checkers are apparently negligible. The
cycle of air flow from one checker to the other is reversed about every 1 5 to 30 minutes in
both the end-port and side-port furnaces. The end-port furnaces are smaller than the side-
port furnaces. End-port furnaces are generally limited to less than 175 tons/day. The
side-port furnaces tend to provide more even heating, which is essential for the high
quality necessary for flat glass. Side-port furnaces are also larger, some over 800
tons/day.
Extensive use is made of cullet (broken glass) in both the container and flat glass
industries. Cullet may consist of internally recycled glass from waste in downstream
operations such as cutting and forming, or it may be externally recycled from glass
returned in recycle operations. Because the chemical reactions necessary to form glass
have already taken place in the cullet, about half the energy is needed to melt the cullet
compared to virgin batch ingredients. Because of the high quality requirements, external
3-2
-------
Glass Surface in Matter
I Natural Draft Stack
Back Wall
Refiner Side Wall.
Melter Side Wall Throat,
Melter BottomV
Rider Arches
Glass Surface in Refiner
Fore hearth
Combustion Air Blowe?X^Duct
Movable Refractory Baffle
Burner
Figure 3-1. Side-port continuous regenerative furnace.
3-3
-------
Glass Surface in Matter
Movable Baffle
Combustion Air Blower
/Matter Side Wall
Induced Draft Fan
Parting Wall
Secondary Checkers
Refiner Side Wall
Glass Surface in Refiner
Forehearth
"Curtain Wall
• Rider Arches
Figure 3-2. End-port continuous regenerative furnace.3
3-4
-------
or "foreign" cullet is not used in flat glass production but is used in container glass
production.
In the melting chamber, the batch components and cullet react to form glass.
Because of heat transfer limitations, a glass melter is generally designed for 0.37 to 0.46
f\ f\ c Ci
m (4 to 5 ft ) of melting area/ton of glass produced in a 24-hour day. The depth of
the glass melt is usually 1 to 2 m (3 to 6 ft) ' and is limited by the need to have proper
heat transfer and melting of the glass batch. Container glass furnaces are usually 6.1 to
9.2 m (20 to 30 ft) wide and 6.1 to 12.2 m (20 to 40 ft) long.4 Flat glass furnaces tend
o
to be longer than those in the container or pressed and blown glass because of the need
to ensure more complete reaction between the batch ingredients and reduce the level of
gas bubbles, evolved in reactions (3-1) through (3-3) above, remaining in the finished
7 QQ
product/ Typical lengths are over 30.5 m (100 ft). As a result, flat glass furnaces
typically have a melting capacity of 500-750 ton/day, compared to that of container and
pressed/blown furnaces, which are no more than about 600 ton/day. The melt becomes
homogeneous and free of bubbles in the "fining" section just downstream of the melting
section. Container and pressed/blown glass furnaces generally have the melting and fining
(or "refining") section separated by a refractory bridge wall or throat through which the
p
molten glass passes. The opening between these sections is beneath the surface of the
glass. This allows only glass that is free of surface contamination [foam or unmelted
batch ingredients, which tend to float or flow to the conditioning section]. Flat glass
c
furnaces do not have a bridge wall. The opening between the furnace and the
downstream refining area is above the surface of the glass in flat glass furnaces.
The production of container, flat, and pressed/blown glass is shown schematically
q
in Figures 3-3 through 3-5. In principle, the three processes are essentially identical
through the melting step, an exception being that pressed/blown glass production does
not, as a general rule, use regenerators to recover heat from the flue gas. [This is
reflected in the higher energy use in pressed/blown glass production, discussed below.]
In container glass production (Figure 3-3), a typical system downstream of the
melter consists of so-called individual section (I-S) machines in which molten glass "gobs"
are fed into molds. The containers are then formed by blowing the molten glass into the
mold to form the final product. The containers are then carefully cooled in the annealing
section to relieve stresses introduced in the molding process. The containers are then
inspected in machines to ensure proper dimension, and packed.
3-5
-------
Raw Materials
Batch
Preparation
NG (Primarily)
E (Boosting)
NG
(2400 °F)
Melting
(2800 °F)
Regenerator
Stack (1200 °F)
Air
(1800°F)
Forehearth
v Glass Gobs (2000 °F)
Forming
NG-
orE
Annealing
(1050°F)
Inspect/
Package
E = Electricity
NG = Natural gas
Figure 3-3. Container glass production/
3-6
-------
Raw Materials
i
Batch
Preparation
NG (Primarily)
orE
(2400 °F)
Melting
(2800 °F)
Molten Glass (1950 °F)
E —
Regenerator
Stack (1200 °F)
Air
(1800°F)
Float
Forming
E or NG
Glass Ribbon (1125 °F)
Annealing
E/NG
Tempering
Furnace (1150 °F)
Laminating
Cooling Air •
Quenching
Autoclave
(285 °F)
E
Hot Air
Inspect/
Package
E = Electricity
NG = Natural gas
Figure 3-4. Flat glass production.5
3-7
-------
In flat glass production (Figure 3-4), the molten glass coming from the fining
section is poured onto a bath of molten tin through the "canal section." As it flows over
this bath, it is gradually cooled from around 1,070 to 610 °C (1,950 to 1,1 30 °F).7 It
then enters an annealing section, after which it is cut, packed, and either sold or further
processed as shown, generally at a separate facility.
In pressed/blown glass production (Figure 3-5), an extremely wide range of
operations can be used downstream of the furnace to produce items such as tableware,
light bulbs, glass tubing, and other products. Each of these operations uses vastly
different machinery and processes, though each shares fhe need for controlled
heating/forming/cooling steps. Further details are given in Reference 11 and elsewhere.
The glass melting industry is a major consumer of energy. A 1977 study showed
that stone, clay, and glass products account for 11 percent of all industrial energy use in
i"}
the United States. Of the total operating costs in the U.S. glass industry, about 15
percent is for energy, essentially all natural gas. The glass industry consumes about 190
o
billion ft of natural gas/year, about 160 billion of which is for the melting furnace. The
theoretical energy requirements for glass can be approximated as follows (per ton of glass
produced) .
106 Btu
Stoichiometric chemical requirements 0.58
Sensible heat of bringing batch to 2,800 °F 1.55
2.13
Because of the inherently low thermal efficiency of gas-fired regenerative furnaces,
about 6x10 Btu is required in practice to produce a ton of glass. Of this total, about 40
percent (or about 2.13 x 10° Btu/ton as shown above) goes to heating the batch and for
the thermodynamic heat of reactions (3-1) through (3-3) above. About 30 percent is lost
through the structure and about 30 percent is lost through the stack.4'14 Electric
"boosting" of gas-fired furnaces is also practiced in the container and pressed/blown
industries, but is not in general use in flat glass furnaces. This consists of placing
electrodes at the end of the melting furnace where the batch is introduced and passing a
current through the melt to resistively heat the melt. About half of all regenerative
furnaces are electrically boosted, with typical boosting being about 10 to 15 percent of the
total melting furnace energy needs. ^'^ Furnace life tends to be shortened by electric
3-8
-------
E or NG
NG-
E
^
E — *
NG — *•
orE
Batch
Preparation
i
r
Melting
(2800 °F)
I \
Forming/
Tableware
i
•
' 1
Forming/
Lighting
1-
Annealing
4 — E
r
Fiber
Drawing
J
Annealing
M Eor
^ NG
^
Sizing
Application
+ 4 1
Finishing
NG — *•
E— ^
Finishing
|
Inspect/
Package
Drying
(280 °F)
NG
T
E = Electricity
NG = Natural gas
Figure 3-5. Pressed and blown glass production.9
3-9
-------
boosting.
Glass can also be melted in all-electric furnaces and electric "boost" can be added
to gas-fired furnaces. The conversion of electrical energy to useful thermal energy in the
glass melt is about 70 to 80 percent, or 2 to 2/4 times higher than for gas-fired furnaces.
However, the production and delivery of electricity from fossil fuel is only about 30
percent efficient, making all electric furnaces generally uncompetitive. There are other
•
factors that limit the use of electric furnaces including limits to the size of electric furnaces
and the electrical conductivity of some batches at high temperature. All electric melters
are used in the container business, though most are found in the pressed/blown business.
Electric boost is common in container furnaces. For flat glass furnaces, electric boost has
not been demonstrated in furnaces larger than 100 ton/day.
Significant progress has been made in reducing the energy consumption per unit of
glass produced in recent years. The increased fuel efficiency has been achieved primarily
through the development of advanced refractory materials which helped lower fuel
consumption per ton of glass produced in the melting operation by 25 percent in the last
15 years.11 In the flat glass industry, energy consumed per unit of glass produced
declined from 23 million Btu/ton to 13 million Btu/ton in the period 1976 to 1986.18
Energy used in the pressed/blown glass segment decreased from 29 million Btu/ton in
1977 to 20 million Btu/ton in 1985. Fuel use for melting operations in the three industries
considered here is as follows :
Total energy consumed
Industry for melting (106 Btu/ton)
Container 8-10
Flat 6-7
Pressed/blown 16
The higher energy consumption in the pressed/blown glass industry reflects the inherent
inefficiencies of the small-scale furnaces characteristic of much of this industry. The high
value-added and the high labor costs due to less automation in this sector make energy
efficiency less important than in the container and flat glass sectors.
3-10
-------
3.3 OVERVIEW OF THE GLASS-MAKING INDUSTRY
or\
A 1984 study reported 800 glass melting furnaces in the United States. u Many
of these are either for fiberglass (not considered here) or are small furnaces for specialty
and art glass. There are a much smaller number of continuous, industrial-scale furnaces
which are of interest here. Figure 3-6 shows the location of container, flat, and major
pressed/blown plants in the United States. Table 3-1 shows the distribution of glass
production among the three industries considered here in 1988.
Despite the general similarities in the glass melting operations in the three segments
of the glass manufacturing business considered here, the three industries are substantially
different. The container glass industry, accounting for over 50 percent of all glass
produced in the United States, generally uses smaller furnaces with lower temperatures
and different raw materials than the flat glass industry. The pressed/blown segment of the
glass business generally uses smaller furnaces than those used for either the container or
oc
flat glass and is generally a more widely dispersed industry. a These three segments of
the glass-making industry are considered separately. The composition of the glass and the
quality specifications are also somewhat different. The flat glass industry has the highest
quality requirements, leading to special care in the melting operation as well as
downstream annealing processes.
TABLE 3-1. GLASS PRODUCTION IN 1988
Industry
Container
Flat
Pressed/blown
Total
Glass production
(10° of tons)
10. 1a
4.1a
4.7b
18.9°
Percent of
production
53
22
25
100
aAsof 1988.
Calculated based on 25% of total production.
cMcGraw-Hill Encyclopedia of Science and Technology reports about 20 million
tons are produced "each year" in the United States.
3-11
-------
CM
CM
00
CO
O>
o
•o
o
T3
C
CO
£,
V)
c
o
"S
JQ
o
I
T3
0)
O
CO
0>
3
*
'a
3-12
-------
3.3.1 Container Glass
Container glass is used primarily for alcoholic and nonalcoholic beverages and food.
The container glass industry has been affected by major restructuring in recent years.
Two companies now account for over 60 percent of the operating capacity, and four
account for over 80 percent (Table 3-2). These four major companies are Anchor Glass,
oc
Ball-lncon Glass Packaging, Owens-Brockway, and Triangle Industries. One projection
showed that total glass production for containers will decrease by about 10 percent by
o~j
1995. This is the result of competition from aluminum and plastic containers in the
beverage business. Figure 3-7 shows the geographic distribution of the 194 furnaces and
op
83 plant locations in the container glass industry in 1988. Melting furnaces are of the
order of 100 to 300 ton/day.
3.3.2 Flat Glass
Flat glass consists almost exclusively of architectural and automotive glass. It is
generally of higher quality than container or pressed/blown glass. Melting is carried out in
large (400 to 800 tons/day) furnaces. Table 3-3 shows the principal U.S. flat glass
companies, which account for essentially all flat glass production.
3.3.3 Pressed/Blown Glass
Pressed/blown glass consists of tableware, lighting/electronic, and scientific
products. A large fraction of this industry consists of owner-managed, small, hand-
operated manufacturing operations with furnace capacities of 5 to 25 tons/day, some of
pq
which are electric. However, some larger operations use gas-fired furnaces on the order
of 100 to 200 tons/day. The production process is shown schematically in Figure 3-5.
The principal U.S. companies are shown in Table 3-4.
3-13
-------
t
EGMEN
CO
cc
UJ
Z
ANIES IN THE CONTAI
a.
^
t>
O
0
cc
o
-J
§
u.
0
CO
UJ
cc
D
UJ
u.
_J
a.
O
Z
cc
a.
J
i
CO
UJ
^J
CD
^^
"O *""•
§ *£
CO ^
C •= s
til'
o. ^ — ~
Other product lines
•o .E **
CD co «r
s si
.i Sg
« « 8
^^ CD (0
•B "
"5 jr
•S°s
ITS S
£
c
CO
a
o
O
LT>
CO
O
O
O)
co"
o
Specialty glasses and plast
containers; forestry
products; health and
financial services
o
o
,
co
^
o
o
m
to
c
o
o
.
un
O
O
CT)
Plastic and metal closures
o
o
CM
O
n
CM
r-~
O
IS
3
CL
m
to
CO
0 fc
? 'S
-C 4-1
0 C
C 0
< 0
CM
o
o
cq
»
to
Plastic and paper container
0
IT)
IX)
O
o
CM
r->"
"ro
m
T3
10 ^
^. p»
f~ .^
1- .2.
to
JO
/n
c w
o .£
0 CD
_C CO
= o
CQ CL
CD
O
O
oq
CO
in
Metal and plastic container
fabricated metal products
o
o
in
0
0
00
CO
^
0
o
O
3
0.
CU
0
CO
^
3-14
-------
CO
CM
">.
CO
T>
(D
S.
i
.1
s
I
*•¥
-8
s
§38
o
»*
o
CO
CO
at
o
"55
o
c
c
o
o
V)
V)
.2
O)
cn
o
CD
Q.
CO
o
•a
Q)
OT
111
CO
0)
D)
Q. if O
3-15
-------
TABLE 3-3. PRINCIPAL U.S. COMPANIES PRODUCING FLAT GLASS11'22
Company
AFG Industries
PPG Industries
Ford Motor
Libby-Owens-Ford
Guardian Industries
Ownership
Public
Public
Public
Pilkington
(U.K.)
Private
Estimated sales
(MM $)
450a
4,687b
2,058a
300a
900
600a
Estimated production
capacity
(MM short tons/year)
0.50
1.40
0.35
0.77
0.50
aGlass sales.
Total sales.
TABLE 3-4. PRINCIPAL U.S. COMPANIES PRODUCING PRESSED AND
BLOWN GLASS TABLEWARE AND KITCHENWARE11
Company
Ownership
Estimated
annual sales
(MM $)
Principal products
Other products
Anchor Hocking Public
758
Table glassware
Corning Glass
Public
1,860
Cosmetic containers at
Carr-Lowrey Div., micro-
waveable ovenware
lighting products at
Phoenix Glass, hardware
and china
"Pyrex" ovenware and Laboratory ware, industrial
dinnerware glass, bulbs, lamps, TV
tubes, etc.
Indiana Glass
Company
Lenox Crystal
Libbey Glassd
Division
St. George
Crystal
Lancaster
Colony Corp.
Lenox, Inc.
Owens-Illinois
Private
a
b
c
10
Hotel and restaurant
glass tableware
Stemware
Glass stemware,
tumblers, tableware
Stemware, tumblers
None
Glass containers, health
and financial services
aSales not known. Employees: ca. 600.
bSales not known. Employees: ca. 300.
cSales not known. Employees: less than 100.
^Libbey Glass is an independent subsidiary of Owens-Illinois.
3-16
-------
3.4 REFERENCES
1. McGraw-Hill Encyclopedia of Science and Technology. 6th ed. New York.
McGraw-Hill, v. 8. p. 125. 1987.
2. Austin, G.T. Shreve's Chemical Process Industries. 5th ed. New York. McGraw
Hill. 1984. p. 198.
3. U.S. Environmental Protection Agency. Compilation of Air Pollutant Emission
Factors, Vol. 1: Stationary Point and Area Sources. Research Triangle Park, NC.
Publication No. AP42 (GPA 055-000-00251-7). 4th ed. (including supplements A,
B, C, and D). September 1991.
4. Glass Packaging Institute, Washington, DC, letter to W.J. Neuffer, EPA/OAQPS,
April 8, 1993.
5. Charles River Associates. Glass Industry: Opportunities for Natural Gas
Technologies. Gas Research Institute. Chicago, IL. Topical Report No. GRI-
88/0266. September 1988. p. 10.
6. Kirk Othmer Encyclopedia of Chemical Technology. John Wiley. 3rd ed. 1981. v.
13. p. 852.
7. Ref. 5, p. 47.
8. Ref. 6, pp. 851-852.
9. Charles River Associates. Adapted from Garrett-Price, B.A., et al. Potential for
Energy Conservation in the Glass Industry. U.S. Department of Energy. Report
PNL-5640/UC95f. June 1986. 1987.
10. Ref. 5, p. 5.
11. Ref. 5, pp. 73-82.
12. Kusik, C.L., J.I. Stevens, R.M. Nadkarni, P.A. Huska, and D.W. Lee. Energy Use
and Air Pollution Control in New Process Technology. Chemical Eng. Prog. p. 36.
August 1977.
13. Ref. 5, p. 13.
14. Ref. 6, p. 856.
15. Memo from Spivey, J.J., RTI to Neuffer, W.J., EPA/OAQPS, August 5, 1993.
Minutes of July 20, 1993, meeting between Primary Glass Manufacturing Council
and EPA/OAQPS.
16. Ref. 6, p. 853.
3-17
-------
17. Rindone, G.E., J.R. Hellman, and R.E. Tressler. An Assessment of Opportunities for
Gas-Fired Boosting of Glassmelting Process. Gas Research Institute. Chicago, IL.
Topical Report No. GRI-89/0254. January 1990. p. 1.
18. Ref. 5, pp. 50-51.
19. Ref. 5, pp. 86, 88, 93.
20. Strumpf, H., D. Kotchick, and M. Coombs. High-Temperature Ceramic Recuperator
and Combustion Air Burner Program. Gas Research Institute. Chicago, IL. Topical
Report No. GRI-83/0039. February 1986.
21. Ref. 5, p. 26.
22. Letter and attachments from Benney, J.C., Primary Glass Manufacturing Council,
Topeka, KS, Neuffer, W.J., EPA/OAQPS, April 22, 1993, 10 pp., comments on
draft ACT report.
23. Ref. 3, p. 8-13-1.
24. Ref. 1, p. 1.
25. Ref. 5, p. 1.
26. Ref. 1, p. 24.
27. Ref. 5, p. 43.
28. Ref. 5, p. 45.
29. Ref. 1, p. 74.
3-18
-------
CHAPTER 4
CHARACTERIZATION OF NOX EMISSIONS
4.1 NOX FORMATION
NOX is formed in glass melting furnaces by:
• the homogeneous gas phase reaction of N2 and O2 in the combustion air,
producing primarily NO,
• the evolution of NO2 from nitrate compounds used in certain glass formulations,
and
• oxidation of fuel-bound nitrogen.
[The term "NO" can refer to any of six nitrogen-oxygen compounds1' , only NO and NO9
xv ^
are of interest and together are referred to as NOX herein.] At conditions of practical
o
interest, about 95 percent of the NOX in the flue gas is NO. The term NOX is thus often
used to refer to only the NO in the flue gas.
4.1.1 Homogeneous NO Formation
The homogeneous gas phase reaction of N2 and O2 in air is generally thought to
proceed through a mechanism first formulated by Zeldovich. This is often called thermal
NOX. The two most important steps in this mechanism are
N2 + O <± NO + N kf = 2 x 1014exp (-76500/RT) (4-1)
N + 02 «=* NO + O kf = 6.3 x 109 exp (-6300/RT) (4-2)
N2 + O2 ^2NO (4-3)
where kf are the forward rate constants for the reactions shown. The high activation
energy of Reaction (4-1), 76.5 kcal/mol, means that this reaction is the most temperature
sensitive.
The equilibrium constant for Reaction (4-3) depends, of course, only on the
temperature. However, the equilibrium concentrations of NOX (NO and NO2) also depend
on the concentrations of N2 and O2 in the gas. Table 4-1 shows the equilibrium
concentrations of NO and N02 (NO2 is generated by reaction of NO with O2) for two
4-1
-------
TABLE 4.1. CALCULATED EQUILIBRIUM CONCENTRATIONS OF NO AND NO2
IN AIR AND FLUE GAS (ppm)5
Temperature
K
300
800
1,400
1,870
°F
80
980
2,060
2,910
Air
NO
3.4(10r10
2.3
800
6,100
NO2
2.mor4
0.7
5.6
12
Flue
NO
i.Kior10
0.8
250
2,000
Gas
NO2
3.3(10)'3
0.1
0.9
1.8
conditions.5 First, the equilibrium NO and NO2 concentrations for N2 and O2
concentrations found in ambient air are shown. These are important for glass melters
because the combustion air is often preheated to temperatures.above 1090 °C
(2,000 °F), which Table 4-1 shows would result in the formation of about 800 ppm NO
and 6 ppm NO2 at equilibrium. Second, Table 4-1 also shows the NO and NO2
concentrations at flue gas conditions, where the O2 and N2 concentrations are defined, for
this table, as 3.3 percent O2, 76 percent N2. In this case, the equilibrium NOX
concentrations are lower because of the lower O2 concentration. For glass melting, this
situation would correspond to the flue gas from the melting furnace, whose temperature
would be around 538 °C (1,000 °F). At this flue gas temperature, the equilibrium NO
concentration is around 1 ppm with NO2 being about 0.1 ppm.
In practice, of course, glass furnace flue gas NOX concentrations are much higher
than this, typically around 1,000 ppm NO. The reason is the high activation energy of
Reaction (4-1), which is generally thought to be rate controlling. After the NO is formed in
the high temperatures of the flame (which can reach well above 1650°C (3,000 °F), the
rate of its decomposition [the reverse of Reactions (4-1) and (4-2)] is kinetically limited at
the lower temperatures and lower 0 and N atom concentrations in the post-combustion
zone of the flame. Thus, although NOV is thermodynamically unstable even at the high
J{
temperatures of the glass furnace flue gas, its decomposition is kinetically limited. The
result is that the NOY concentration in the flue gas is higher than predicted by equilibrium
yv
and depends, to a large degree, on the mixing of the fuel and combustion air in the flame.
Techniques to minimize NOV formation by modification of these conditions are discussed in
s\
Chapter 5. The following empirical expression describes, at least qualitatively, the effects
4-2
-------
Chapter 5. The following empirical expression describes, at least qualitatively, the effects
of temperature, time (of the gases in the flame zone), and N2/O2 concentrations on NO
levels in the outlet gas of a combustion process :
= 5 x 1017[exp (-72,300/7] y^ y t (4-4)
where
CNQ = NO concentration, ppm,
YJ = mole fraction of gas / (/ = N£, O2>,
T = absolute temperature, K, and
t = time, seconds.
Effects of fuel type, flame geometry, and other factors that can significantly affect NO
generation are not accounted for in this expression. Thus, absolute NO concentration from
any specific furnace cannot necessarily be predicted using this expression. The time in the
o
flame zone is about 0.5 seconds. For an adiabatic flame temperature for natural gas at
10 percent excess air of 1,870 °C (3,400 °F), and using yN2 = 0.79 and yo = 0.21
(the N2 and 02 present in ambient air), Equation (4-4) predicts C^Q to be 206 ppm, which
Q
may be an underestimate. Nevertheless, the essential features of this
equation —exponential dependence of NO concentration on temperature, half-order
dependence on 02 concentration, and linear dependence on N£ concentration and
time—provide qualitative guidance on the effect of time, temperature, and excess air on
NO emissions at conditions of practical interest.
Figure 4-1 shows the generation of NOX as a function of excess air. The
importance of this plot for glass melters (and other operations) is that fuel firing rates are
often given in millions of Btu/hr (MM Btu/hr). Knowing the furnace temperature and
excess air, the Ib NOX/MM Btu can be determined (e.g., about 1.5 Ib NOX/MM Btu from
Figure 4-1 for 1370 °C (2,500 °F) and 40 percent excess air). This can then be multiplied
by the firing rate (MM Btu/hr) to give an NOX generation rate (Ib N0x/hr). Thermal NOX
emissions, in turn, vary directly and linearly with fuel firing rate, all other conditions being
equal.
4-3
-------
3000
110
too
90
80
70
60
50
40
30
20
10
0
2750
-40 -20 0 20 40 60 80 100120140160180200
Excess Air (%)
. Figure 4-1. Generation of N0r10
-------
4.1.2 NOX from Nitrates
NO2 is formed when sodium and potassium nitrates, called "niter," are used in
certain glass batch formulations. The purpose of these compounds is to aid in the removal
of bubbles from the melt in the "fining" section of the melting furnace. These materials
react at higher temperatures than needed for melting so that the removal of bubbles
1 ?
continues after the melting reactions are complete. Though some niter is used in flat
glass production, most is used in container and pressed/blown glass.
The evolution of NC>2 from the nitrates is essentially stoichiometric, i.e., all NC>2
present in the nitrate is released in the furnace. Thus the amount of NO? released
depends on the niter content of the batch.
4.1.3 NO from Fuel/Oxidizer
x
NO can also be produced by oxidation of fuel-bound nitrogen, e.g., pyridines or
other organonitrogen compounds. Air infiltration may also be a source of nitrogen.
Natural gas is the fuel used predominantly in glass melters. Though natural gas, as
delivered to the burner from the pipeline, may contain as much as 1 to 3 percent N2, it has
essentially no fuel-bound nitrogen. Many plants have backup fuel capability for
emergencies, which is regarded as essential given the high cost of startup once a fuel
interruption occurs. Typical fuels include LPG, No. 2 fuel oil, and diesel. However, there
are no data at present to assess the proportion of glass melters using fuels other than
natural gas, nor the proportion of time other fuels might be used even in furnaces usually
using natural gas.
Nitrogen is also present even when "oxygen" is used in oxy-firing (Section 5.2.3).
Depending on the source of oxygen, nitrogen levels can be 100 ppm to several percent.
This nitrogen, plus nitrogen from the inevitable air infiltration, is also a potential source of
NOX in oxy-firing.
4.2 FACTORS AFFECTING NOX EMISSIONS
NOV emissions can be measured in two ways. The first is the rate of NOV
J\ S\
generation, e.g., in units of Ib NOx/hr at a given fuel firing rate, or ppm of NOX at a given
flue gas volumetric flow rate, typically corrected to a specific Oo level (e.g., 3% CU). The
second is the amount of NOX produced per ton of production, e.g., Ib N0x/ton glass
produced.
4-5
-------
4.2.1 NOX Generation Rate
Essentially all of the NOX produced in a flame is generated at the peak flame
temperature. The following factors, measured at this temperature, have the greatest
effect on the rate of NOV generation:
/\
• N2 concentration,
• Oo concentration,
• temperature, and
• gas residence time.
If air is used in the combustion process, the nitrogen concentration in the furnace is
essentially constant. The oxygen concentration, however, will decrease as fuel is
consumed. It is the local concentration of oxygen in that part of the flame where the peak
temperature occurs that affects NOX generation. For this reason, many of the low-NO^
burners discussed in Chapter 5 limit NOX generation by staging the combustion, in effect
limiting the oxygen concentration while lowering the peak flame temperature. Note,
however, that Equation (4-1) shows that the NO concentration is only half-order in oxygen
concentration, meaning that decreasing the oxygen concentration by, say, one-half, only
1 /9
decreases the NO concentration by 29 percent (0.5 {'^ = 0.71).
The peak flame temperature is the most important factor affecting NOX generation,
as shown by Equation (4-4). The adiabatic flame temperature, which is the temperature
reached by a given proportion of fuel and combustion gas (e.g., air), can be calculated
from thermodynamic data. This is the maximum temperature that can be achieved in a
flame with that fuel. It is a function of the air/fuel ratio, which is in turn often expressed
as the equivalence ratio of Figure 4-2 [equivalence ratio - 0 -
(air/fuel)actua|/(air/fuel)stojchjometrjc]. For 0 < 1, the combustion mixture is fuel-rich; for
q
0 > 1, the mixture is fuel lean. Figure 4-2 shows such a plot for various fuels. [This;
plot is for an initial pressure of 10 atm and is not, therefore, numerically valid for
combustion at 1 atm. However, adiabatic flame temperature is not a strong function of
pressure (see Reference 14) and the qualitative trends, e.g., adiabatic flame temperature
as a function of equivalence ratio and fuel type, are valid. For natural gas, which contains
mostly methane (with some ethane and propane) the peak flame temperature at the 10 to
20 percent excess air used in glass melters is around 1,820 °C (3,300 °F). In practice,
the peak flame temperature will be somewhat less since heat is transferred (by radiation)
4-6
-------
from the flame to the glass melt. Figure 4-2 shows that the peak flame temperature can
be lowered by either fuel-rich ( 1) conditions. Practical
considerations, such as emissions of unburnt hydrocarbons at fuel-rich conditions and
lower heat generation rate (MM Btu of heat generated from a given quantity of fuel) at
fuel-lean conditions, as well as less than ideal gas/fuel mixing, lead to operation of glass
melters at 0 ~ 1.1 or so. Figure 4-3 shows NOX concentrations measured in the
1 c
combustion zone for glass furnaces as a function of air/fuel ratio. [Air/fuel ratio is
proportional to equivalence ratio; an equivalence ratio of 1.0 corresponds to an air/fuel
ratio of 9.52.]
In some furnaces, the peak flame temperature may vary with furnace position. This
is because multiple firing ports are often used to develop the temperature needed to melt
the glass and react the ingredients at specific points in the furnace. For example, higher
temperatures may be needed at the furnace entrance because raw materials are added
there. This distribution of fuel can cause higher overall NOX emissions than an even
distribution would because of the exponential dependence of NOV emissions on peak flame
)\
temperature.
The final factor affecting the IMOX generation rate is gas residence time, i.e., the
time the fuel/combustion gas mixture remains at the peak flame temperature. As with
oxygen concentration, a great number of burner designs' have been developed to minimize
NOV generation by minimizing this parameter. Because Equation (4-1) suggests that NO
J\
concentration is linear in gas residence time, decreasing it has a numerically greater effect
than decreasing C^ concentration. However, in practice there are narrow limits to gas
residence time within which a stable flame can be produced. Typical gas residence times
at conditions of practical interest are of the order of 0.1 to 0.5 seconds.
The temperatures and residence times required for NOV formation are also present
s\
in the air preheating used on regenerative furnaces (Figures 3-1 and 3-2). Air preheat
temperatures may exceed 1,260 °C (2,300 °F) and residence times are of the order of
seconds. Together, these can lead to formation of NOV in the preheated air.
A
4.2.2 Normalized NOy Emissions
NOX emissions are often expressed by the rate of production of glass; e.g.,
regulations in the South Coast Air Quality Management District (SCAQMD) are written in
units of Ib NOx/ton glass produced. Overall NOX emissions, by this measure, can thus be
decreased by increasing the productivity of the furnace (ton glass produced per hour) even
4-7
-------
5500
5000
4500
0 4000
2
73
® 3500
3000
2500
2000
J I
0.6 0.8 1.0 1.2 1.4 1.6 1.8
Equivalence Ratio
Figure 4-2. Relationship between equivalence ratio
and adiabatic flame temperature.
4-8
-------
2000
Q.
Q.
O* 1500
z
o
I
I 1000
W
J2
O
O
500
9 10 11
Air/Fuel Ratio
12
Figure 4-3. Relationship between air/fuel ratio for natural
gas fuel and NOX concentration normalized to
combustion zone conditions.
4-9
-------
if the rate of NOV generation (Ib N0v/hr) is constant. Factors affecting these normalized
/\. .A
NOX emissions, then, can include better refractory insulation (meaning that less heat is lost
through the refractories) and process changes such as oxy-firing. These control
techniques are discussed in Chapter 5.
4.3 UNCONTROLLED NOV EMISSIONS
yv
Table 4-2 summarizes NOX emissions reported from glass melting furnaces. These
values range from 2.5 to 27.2 Ib N0x/ton of glass produced. This wide range reflects the
variation in site-specific factors that affect uncontrolled NO emissions.
These include furnace size (smaller furnaces tend to have higher normalized NOX emissions
than larger furnaces), furnace age, air infiltration, burner geometry, combustion air preheat,
and other factors. The NOX concentration in the flue gas is also important. As a general
rule, thermal NOX concentrations (i.e., exclusive of NO2 from niter) are in the range of
1,000 to 3,000 ppm, depending on burner design, fuel firing rate, and other
fi 9R
parameters. "^°
For the purpose of calculating the effect of the control technologies on NOX
emissions, uncontrolled NOX emissions are defined as follows:
Uncontrolled NOX
Furnace type emissions,
Ib NOx/ton
Container glass 10.0
Flat glass 15.826
Pressed/blown glass 22.0
9q
These values approximate uncontrolled levels of a wide range of regenerative furnaces.
Based on the information in Table 5-9, NOV emissions reductions are shown in Table
A
5-10. NO reductions based on these uncontrolled levels are used in calculating cost
A
effectiveness in Chapter 6. Assuming a heat input of 6 MM Btu/ton (from Chapter 3),
these values correspond to uncontrolled emissions of 1.67, 2.63, and 3.67 Ib
NOY/MM Btu, respectively, for container, flat, and pressed/blown glass furnaces. It is
A
important to look at both measures of NOX emissions - Ib/ton glass and Ib/MM Btu. These
two measures are, of course, related by the heat input, measured in units of MM Btu/ton
of glass, which is, in turn, a measure of the thermal efficiency of the glass furnace.
Except for oxy-firing, the two measures of NOX controlled emissions in Table 5-9 are
4-10
-------
V)
O
V)
Ul
X
O
0
UJ
_j
O
CC
1-
z
o
o
z
CNJ
UJ
CO
^f
r^
'
1o
^L>
o
£
o
o
cc
o*I
2? ,*^
•D£
«
"* c
t 0
§1
c F
3 a>
•sg
0. C
H.2
x
+"•
1-
"C
g"c
._ o
(ft *^
Q
•o
o
3
Q
O
at
%
o>
"5
CD
>:
t-
^Bfc,
O
I
u
_o
i
Q.
O
O
« s
tO O5
en ?D
*""|
m CN
II
* i
CO LLJ
0 w
to
_CO
JO
o
to
E
to
Q
CO °
CO -
t 0
0 0
JO •<-
J> CD
£ 3
^ t_
> Q
__ Q
O
CO
ntainer
3allo Glass (Modesto,
2arr Lowrey (Baltimor
o
O
If)
CO ^
CM CM
•o x
• r«L
00
t
0
o.
jj-73
z£
to
tnO
^
(0
CO
,—
r contair
CO
JO
cc E
Z <
.atchford Glass
!Los Angeles, CA)
in
CM
4"
CM
CM
r
o
Q.
CO
CO
o
in
CM
to
to
ro
D)
:ontainer
u
c
Diamond Bathurst
ro
C ID
CO
•= o
CO •§
< <
4^4
o
^ *
C TO c
° 0 °
tj *J t)
(0 TO (0
C/)Z CO
O
O5 73
O CM
t — '
LO ^
~; ^
CM
CC CC
z z
O 73
00 »tf
til CM
*t CD
to co
t— i—
to to
co co
c. _c
'ro 'ro
4~> 4->
§^-
0
o o
c
^rnerican National Ca
Q. Q.
O Q- O.
4 44
' s^
1~ ro *~ "~
c 5 c c
o _2 o o
C « 1 '€
CO t- CO CO
co co in co
co
d ^^
CN ' '
ob ^^
CM-
0
Q.
CO
cc ^^ CT"
z coz
CO
§ |S
1^.
*~
to
ir- to to
CO (/) to
c ro ro
TO O) D)
O ro ro
O u- u.
ro
-* — ^"7
0 < ^
° O
"? -0
% 1R
| 0 «+J
O •*-* Q- t- ^.
ul
•p
CD O CD o u u, Uu u f
IJ- LLLL U- OOOOO
<<< , _J_J_J_J_I
4 44 t 44444
*
T""
c cc c ccccc
o oo o ooooo
O (JO O OOOOO
)}to
TO TOTOTO TO TOTOTOTOTO
O) CDO3CO CD CDCDCDCDCD
ro TOTOTO TO TOTOTOTOro
r- CM
-zS- Go
S.!IJ" s § £i- 2-^-^-^Oa:
f^
CM
**- D)
00 CM
CD r^
*~ CM
CDjQ
to )
w to
CO TO
D) D5
±^ ±^
'o 'o
CO CO
Q. Q.
coco
ssed/blown
3TE Products
[Versailles, KY)
#1 furnace
#2 furnace
CO
ct
O)
ro
Q.
! footnotes on following
-------
•o
Q)
O
O
LLI
_l
CQ
E 2 05
Q) fc_ *s -f
4-* m *^ ^
CO >
1 H-
£•_>• ° "o
•' ;
:r 4^
»€ gf § I
11 4=8 « §
^~ »
k. O 4-" '
05 4^ ._ I
•B -s o i
o 3 •;; :
M- E CO 1
O J3 I
* *" P T
co ,- £ 1
o T; o c
Jc " cj t
^~" CO (OCsJ
fll J^ rtrvi
0 °-
31 " <0
• OT ^
E *~ o
-04-*
3 S *7
0 •*
3 " E
? § 5
5 ^ CO
CO CD
0) D)
o w Q."«} o> £
'is <° •£ i 2
4-* — k. <
4_t ^ CD «
CO 05 ."i^ !
S € ro *
— (0 *
4-* C 05 C1
CO — Ok
O5 'ti! CD f
ir >". cr fl
2 "°
: -D c
C CO
S « w
i S t:
c js
q ^o Q_
B ^
03 i? 3 -Sr = =5
CO 4-" *•
X (0 j _,
c C 0) O 4-. TO
O CD 2
w t ^ ^
3 4-* i
'E w - *
j O
T ._ . .
2 t o
: o 05
0 CO ^. t; CD u]
*o
^"~
CD
O
CD
Q.
CO
*~
O
00
C
CO
•o
c
CO
CO
CO
a
"co
CO
•o
CD
CO
3
10
o
o
JD
O
^*
o
CD
LU
4-»
i
CO
u-
>•-
CO
CO
JO
H-
J^
Q
CO
05
i_
CO
CO
05
M
4J
2
Q.
03
CO
CXI
CN
(O
05
^
o
o
^
c
CO
CO
05
k.
CO
co
05
4^
V)
O
I^IZ
a.
CO
05
4-<
05
CO
3
CO
o
CD
to
JO
O)
C—
o
•*i
O
00
.
in
*"~
CO
•^
to
!c
4^
E
o
**—
o
CO
•r;
E
C
CO
05
|
O
c
CO
E
CO
(0
CO
O)
JO
05
_c
>
^2
ni
73
O
Q_
CO
CO
"O
CD
•C
*™^
£
~
5
CO
4-<
co
"*3
05
CO
CD
(~
+-1
H—
O
°. "~ 0 05 LLI Oi * ~
5 5 CD i
"5 'c 52
' . (0
, (0 CD
. CD O
c „_ ~ O O CO
CO ••£ C H- CO C
— w 1-1 C
5 r- k.
S CD '= -D t 5
CO CO Ok
_1 --
a = •§ S - c
. CD U Si c CD
OJ •£ ^- --
X £ CD
CO
0)
O
CO
k.
3
4-<
C
05
05
•SP -o £ E S
4^ *" CO 2
: S: T5
•D
r-
ro
oo
T—
0
*fc_
CD
a_
OJ
c
X
O
"<2.
CD
co
CD
>
TO
CD
(_
C
O
+-*
CJ
S-
-Q
05
E
CO
0)
E § °C -° LO * •£>.*>
o-o o oo co
°S 2« I £
CO >- 2 CX
0 CO t ..
) T3 -0
05 05
•Q. E o "o t t:
>- ,_ Q. o o
"^ •- «n £52 Q. 0.
+j _C tO »- C CD 0)
O 1— t CO C
5 >- k.
— _ ' o '~
c .:= >i_
CD
S
CO
4-J
i^
O
TJ w
CD 05
t .t;
0 C
& - E
•- i2 o
4^ fr
S £ *
^ E •-
co .i to
> 3 10
H- tr co
0 CD o,
<" -^ c
Iff 1
JO 43 ~-x
05 CD O
*• 5
.52 S ^
W *"* °5
£" —
1- += CO
CD H-
05
.ti
C
E
o
k_
_co
CO
CO
CO
O)
c
o
^-«
X
o
Z
£
rv
^
cs
^
O)
1
O
«4-
co
CD
CN
Reference
provided in
CO
CO
Q
£
^
O
4-1
-
-
CD
0
4-»
O
00
CO
o
V-
co
05
4-<
2
c
0
"4^
o
3
k.
Q.
i
b.
ci
_c
^— '
CD"
o\l
Reference
dentified in 1
~
Q
C
responses
•*
o
'•S
05
CO
~
ji.
i^.
4-12
-------
directly proportional once the assumption of 6 MM Btu/ton glass is made. For oxy-firing,
however, much less energy is needed because nitrogen is not present in the combustion
air and energy is not used (and then lost up the stack) to heat it in the furnace. For oxy-
op
firing, a value of 3.4 MM Btu/ton0 is reported, though this varies with different furnaces
(which have different levels of air infiltration) and oxygen sources (which contain different
amounts of nitrogen).
4.4 REFERENCES
1. Classman, I. Combustion. 2nd ed. Academic Press. 1987. p. 328
2. Brunner, C.R. Incineration Systems: Selection and Design. Van Nostrand
Reinhold. 1984. p. 112.
3. Joseph, G.T., and D.S. Beachler. APTI Course 415, Control of Gaseous Emissions.
U.S. Environmental Protection Agency. Air Pollution Training Institute. EPA 450/2-
81-006. December 1981. p. 7-4.
4. Zeldovich, J. The Oxidation of Nitrogen in Combustion and Explosions. Acta.
Physiochem. 21(4). 1946.
5. U.S. Environmental Protection Agency. Compilation of Air Pollutant Emission
Factors, Vol. 1: Stationary Point and Area Sources. Research Triangle Park, NC.
Publication No. AP42 (GPA 055-000-00251-7). 4th ed. (including supplements A,
B, C, and D). September 1991.
6. Fleming, O.K., and F.R. Kurzynske. NOX Control for Glass-melting Tanks. In:
1985 Symposium on Stationary Combustion NOX Control. Vol. 2: Industrial
Processes, Fundamental Studies, and Slagging Combustors. EPRI. January 1986.
p. 55-1.
7. MacKinnon, D.J. Nitric Oxide Formation at High Temperature. J. Air Poll. Control
Assoc. 24(3). 1974.
8. Cooper, D.C. Air Pollution Control: A Design Approach. PWS Engineering, p. 458.
1986.
9. Edwards, J.B. Combustion: The Formation and Emission of Trace Species. Ann
Arbor Science Publishers. 1974. p. 39.
10. American Society of Mechanical Engineers. Combustion Fundamentals for Waste
Incineration. New York, NY. 1974.
4-13
-------
1 1. Spinosa, E.D., D.T. Hooie, and R.B. Bennett. Summary Report on Emissions from
the Glass Manufacturing Industry. U.S. Environmental Protection Agency. EPA-
600/2-79-101. April 1979. p. 32.
12. Kirk-Othmer Encyclopedia of Chemical Technology. John Wiley. 3rd ed. 1981. v.
13. p. 850.
13. Charles River Associates. Glass Industry: Opportunities for Natural Gas
Technologies. Gas Research Institute. Chicago, IL. Topical Report No. GRI-
88/0266. September 1988. p. 3.
14. Ref. 1, p. 26.
*•
15. Ryder, R.J. Use of Electric Boost to Reduce Glass Furnace Emissions. Ceram. Bull.
57(11):1025. November 1 978.
16. Geiger, G. Environmental and Health Issues in the Glass Industry. Ceram. Bull.
71(2):194. 1992.
17. Tuson, G., R. Higdon, and D. Moore. 100% Oxygen Fired Regenerative Container
Glass Melters. Presented at the 52nd Conference on Glass Problems. University of
Illinois. Urbana-Champaign, IL. November 12-13, 1991.
18. Moore, R.D., and J.T. Brown. Conversion of a Large Container Furnace from
Regenerative Firing to Direct Oxy Fuel Combustion. 1991 Glass Problems
Conference. November 1 2-13, 1991. American Ceramic Society. Westerville, OH.
1992. p. 5.
19. Perrine, L.E. Glass Industry, p. 8. February 1 992.
20. Ref. 16, pp. 194-195.
21. Teller, A.J., J.Y. Hsieh, and W. Van Saun. Control of Emissions from a Container
Glass Furnace. Ceram Eng. Sci Proc. 10(3-4):312. 1989.
22. Ref. 21, p. 321.
23. Ref. 21, pp. 312-324.
24. Abbasi, H.A., and O.K. Fleming. Development of NOX Control Methods for Glass
Melting Furnaces. Gas Research Institute. Final Report No. GRI-87/0202. August
1987.
25. Neff, G.C. Reduction of NOX Emissions by Burner Application and Operational
Techniques. Glass Tech. 31(2):37. 1990.
4-14
-------
26. Letter from Benney, J.C., Primary Glass Manufacturing Council, Topeka, KS, to
W.J. Neuffer, EPA/OAQPS, August 3, 1993. Comments on uncontrolled NOX
emissions.
27. Fax message from Weikel, P., GTE Products, Versailles, KY, to Linzel, C., Columbia
Dist. Company, Columbus, OH. August 31, 1992.
28. DeStefano, J.T. Postcombustion NOX Control Technology for Glass Furnaces,
Update. Presented at the 45th Glass Problem Conference. American Ceramic
Society. Columbus, OH. p. 243. November 1984.
29. Abbasi, H.A., and O.K. Fleming. Combustion Modifications for Control of NOX
Emissions from Glass Melting Furnaces. Ceram. Sci. Proc. 9(3-4): 168. 1988.
30. Shelley, S. Chem. Eng. pp. 67-69. December 1 992.
31. Neff, G.C. Reduction of NOX Emissions by Burner Application and Operational
Techniques. Glass Tech. 31(2):39. April 1990.
32. Ref. 18, p. 4.
4-15
-------
CHAPTER 5
CONTROL TECHNIQUES FOR NITROGEN OXIDES FROM GLASS MELTING
5.1 INTRODUCTION
Techniques for controlling NOX emissions from glass melting furnaces can be
divided into three basic types ':
• combustion modifications
- modified burners
- oxy-firing
• process modifications
- modified furnace
- cullet/batch preheat
- electric boost/all-electric melting
• postcombustion modifications
- selective catalytic reduction (SCR)
- selective noncatalytic reduction (SNCR).
Not all of these technologies have been demonstrated on the three types of glass
furnaces considered here. In the following sections, the type of furnace in which these
technologies have been demonstrated will be identified. In cases where the NOV controls
A
have not been demonstrated, technical judgments are made as to whether they could be
applied.
5.2 COMBUSTION MODIFICATIONS
Combustion modifications refer to changes in the burner and flame to reduce NOV
s\
emissions. A wide variety of such modifications have been introduced and studied,
o
particularly on coal-fired industrial and utility boilers. However, conditions in these boilers
o
differ substantially from those found in modern regenerative glass melting furnaces.
Specifically, these differences are as follows:
5-1
-------
Boilers Glass Furnaces
Combustion air preheat Moderate (-500-1000 High (2000-2500
Excess air levels Low High
Combustion chamber "Cold walled" Refractory-lined
(low temperature) (high temperature)
All of these contribute to inherently higher NOY levels in a glass furnace than in a boiler
yv
firing the same fuel at the same rate.
All combustion modifications are designed to minimize NOX formation by reducing
one or all of the following'*:
• peak flame temperature,
• gas residence time in the flame zone, and
• oxygen concentration in the flame zone.
Reducing these three parameters is, of course, suggested by Equation 4-4, which
expresses NOV concentration as a function of these parameters. This equation also shows
yC
that reducing the peak flame temperature has the greatest effect on NOX concentration,
and many combustion modifications have focused on minimizing flame temperature.
In general, combustion modifications to minimize NOV formation in glass furnaces
J\
can be grouped as follows ' .
• Modifications to existing burners and burner part hardware
- low excess air operation
- changing air/fuel contacting
• Modified burners.
Other general combustion modifications have been reported for NOX control on
other combustion processes, including fuel switching (usually from coal or oil to natural
gas), water (steam) injection (used mainly in gas turbines)" reduced air preheat, and
derating. 4/7/° Flue gas recirculation can also be used independently of low NOX burners
(LNBs) on some combustion processes to reduce NOX. 9/ ° However, the limitations of
glass furnace operation (e.g., the need for high furnace temperatures requiring high
combustion air preheat) ' "" make such techniques infeasible. There are also tradeoffs, with
such techniques as derating, between NOV and overall energy efficiency and emissions of
Pv
unburned hydrocarbons and CO.
5-2
-------
5.2.1 Modifications to Existing Burners
5.2.1.1 Low Excess Air (LEA) Operation. As recently as 30 years ago, many
1 ^
industrial furnaces routinely operated with 50 to 100 percent excess air.I0 Increasing
energy costs, requiring higher efficiency, gradually led to decreasing excess air. For utility
boiler and other industrial combustion processes, LEA operation is now considered
routine.14 Because air/fuel mixing is less than perfect in any combustion system, some
excess air is a practical necessity. This ensures complete combustion of the fuel both for
efficiency reasons and to minimize emissions of unburned fuel and hydrocarbons.
LEA is designed to reduce the oxygen concentration in the flame zone and therefore
reduces NOY formation, as shown in Equation 4-4. Figure 5-1 shows the qualitative effect
s\
of excess oxygen level on NOX concentration (% excess oxygen = % excess air). Data
predicted by equilibrium as well as from tests on two glass furnaces are shown. The
trend, showing increase in NOX with increasing excess C^, is clear. Data is also available
on the effect of excess air on NO1^'1^ Tests on a commercial 140 to 1 65 ton/day
yv
Latchford Glass end-port furnace, a 250 ton/day side-port Diamond Bathurst furnace in
Royersford, PA, and pilot scale tests are plotted in Figure 5-2. The data are presented in
normalized terms, i.e., NOX normalized to NOX at 15 percent excess air. Absolute levels
of NO produced at any given excess air level are not shown. However, the same trend is
seen—increasing NOX with increasing excess air.11 Table 5-1 shows data taken on the
1 R 17
two commercial furnaces on NOX reductions as a function of excess air. ' As
expected, lower excess air leads to lower NOV emissions in both furnaces. Reductions of
.X
28 percent were achieved in both cases, though the excess air was much greater in the
side-port furnace. There are, of course, practical limits to the amount of excess oxygen
required to achieve efficient combustion and energy use and to minimize other emissions.
5.2.1.2 Changing Air/Fuel Contacting. As shown in Figure 5-3, regenerative glass
furnaces are generally fired by mixing a horizontal stream of preheated combustion air with
a stream of natural gas fuel injected in a much smaller separate port at an angle. The
natural gas fuel can be injected below (underport firing), beside (sideport firing), or above
(overport firing) the combustion air, though below is apparently the most common.
Typical fuel injection velocities are of the order of 500 to 800 ft/sec. The mixing of the
fuel and air is accomplished by the difference in this high velocity and the much lower
5-3
-------
234
Excess Oxygen (%)
• Source 1 glass furnace
• Source 2 glass furnace
— Equilibrium
Figure 5-1. Effect of excess oxygen on concentration of NOX.11
5-4
-------
(0
UJ 0.8
in
T-
to
ox
-10 0 10 20 30 40
Excess Air (%)
Figure 5-2. Effect of excess air level on NOX (O pilot-scale; • commercial end-port;
• commercial sldeport).15
5-5
-------
TABLE 5-1. EFFECT OF EXCESS AIR ON NOX IN COMMERCIAL FURNACES
Commercial side-
port furnace
(Diamond
Bathurst)
Commercial end-
port furnace
(Latchford Glass)
Excess air
level (%)a
12.5C
18.2
18.4
4.5d
7.4
9.1
Furnace
pull (ton
glass/day)
255
164
NO (Ib/ton
glass)
9.3
13.0 .
12.9
5.2
6.3
7.2
N°* h
reduction
(%)
28
-0-
_c
28
13
_c
NOX cone.
(ppm, 0%
0,)
2430
3240
3100
924
1140
1320
a Calculated from data provided by Abbasi and Fleming. In this work, Tables 3 (p. 41) and 9 (p. 90)
present data for the end-port and side-port two furnaces, respectively, in terms of percent ©2- Table
3 adds the qualifying term "in port." It is assumed here that the oxygen levels reported are directly
comparable and provide a measure of the excess combustion air. There is some difference in the
sample locations used to check the exhaust gas oxygen concentration. Abbasi and Fleming describe
this on p. 33 and p. A-3 for the end-port furnace and on p. 82 for the side-port furnace.
Assuming the fuel is pure methane, the percent excess air (or excess oxygen) can be calculated from
the oxygen concentration in the flue gas, which is reported in some cases by Abbasi and Fleming,
assuming no infiltration of outside air, as follows (x = % 62 in flue gas, expressed as a decimal, i.e.,
2% oxygen in flue gas would be expressed as 0.02):
% Excess air
4.54x
(1-x)
(100%) .
b Percent reduction for each furnace is calculated relative to the highest value of NOX (Ib N0x/ton
glass) reported for each furnace. For example, for the side-port furnace, the percent NOX reduction
for 12.5 percent excess air is (12.9-9.3) Ib N0x/ton glass - 12.9 = 28%.
c All excess air values for this furnace are averages of data taken individually on each of the four firing
ports.
^ All excess air values for this furnace are averages of two data points, one for right-side firing and one
for left-side firing.
5-6
-------
Checkers
Combustion
Air
Crown
Preheated
Combustion Air
• Natural Gas Injection Port
—- *•*—.-
Glass Melt
Flue Gas
Exit
Figure 5-3. Glass furnace burner configuration.
5-7
-------
velocity of the preheated combustion air, typically around 20 to 30 ft/sec. 1^,19
There are several independent variables that can be changed to reduce NOX
formation in such burners. These include the contact angle between the gas and
combustion air, air and gas velocities, and location of the natural gas injection (e.g.,
underport or overport). However, the ability to change these variables in an operating
furnace can be quite limited due to furnace and firing port geometry and the way the
combustion air is introduced into the furnace. As expected, each of these affects the
three primary variables that influence NOX formation—flame temperature, oxygen
concentration, and gas residence time at peak temperatures. A series of studies
investigated the effects of these variables on NOV formation in regenerative glass
s\
furnaces. ' Using data and correlations obtained from a one-quarter scale pilot scale
furnace, tests on two commercial furnaces were carried out (see Section 5.2.1.1 and
Table 5-1).
The tests also examined the effect of underport firing (fuel injected beneath the
combustion air) versus side-port firing (fuel injected beside the combustion air) on the end-
port furnace. Representative test conditions and results are summarized in Table 5-
2 15,17 YJ-,JS table summarizes the range of operating conditions used to determine the
effect of excess air and air/fuel contacting on NOX emissions.
The results generally showed that NOV is minimized by "long, lazy" luminous
X.
flames. This is consistent with reduction of peak flame temperature and gas residence
time at peak temperatures. The effect of excess air from this study is discussed in Section
5.2.1.1. Specifically, NOX was reduced by:
• reduced air velocity,
• reduced fuel velocity,
• reduced contact angle between fuel and air, and
• underport firing (compared to sideport firing; overport firing was not
investigated).
The effect of the first three parameters (air and fuel velocities and contact angle) is
accounted for in a "mixing" factor defined as follows:
Mf = Va sin a + 4.7 Fa v}/2 <5'1;
where Mf = mixing factor
Va = "effective" air velocity, ft/sec
5-8
-------
TABLE 5-2. REPRESENTATIVE TEST CONDITIONS
End-port furnace Side-port furnace
Company
Location
Furnace size, ton/day
Excess air, %
Air preheat, °F
Fuel velocity, ft/sec
Air velocity, ft/sec
Firing rate, 1 MM Btu/ton
Latchford Glass
Huntington Park, CA
140-165
7-10
2200
550-1200
18
5.2
Diamond-Bathurst
Royersford, PA
250
10a
2200-2500
390-610
30
4
Reference 20.
>
Pont reports that end port furnaces typically use lower fuel injection velocities than
side-port furnaces, contrary to the conditions reported here. * This may be due to
the higher than normal air velocity of the Diamond-Bathurst side-port furnace.
5-9
-------
a = air/fuel contact angle
Fa = fraction of air that mixes directly with the fuel, 0
-------
.2
"£
TJ
C
O
O
0>
CO
OB
3500
3000
2500
2000
X 1500
E
Q.
Q.
ox
1000
500
DO Pilot
• • Commercial
30
20
CT
O
-------
o
ox
4000
3500
3000
2500
2000
1500
1000
500
FF = Flat Floor Port
IF = Incline Floor Port
CT = Combustion Tec Burner
W/B = With Air Baffle
90°. CT, IF
0 100 200 300 400 500 600 700 800
Fuel Injection Velocity (ft/s)
Figure 5-5. Effect of fuel Injection velocity on emission of NOX.
23
5-12
-------
the second. This minimizes the peak flame temperature and corresponding oxygen
concentration and thus minimizes NOX formation. Burners have been designed with a
variety of contacting schemes to improve both NOX reduction and fuel efficiency. A
diagram showing the essential features of a three-stage coal-fired LNB is shown in Figure
5-6. NOX reductions of around 30 to 50 percent or higher over older design burners are
possible.6,10,29 Manv currently available burners for glass furnaces include features to
allow adjustment of air/fuel velocities, contact angle, flame shape, and injection orifice.
Each of these can result in NOV reduction (see Section 5.2.1), but do not include all of the
J\
features that characterize what are commonly known as LNBs.
5.2.2.1 Sorq Burner. A 1991 report states that ". . . no LNBs are yet available
"off the shelf" for glass furnaces."30,31 However, a staged burner developed by Sorg
™ "V)
GmbH (Cascade burner) has been tested recently on two container glass furnaces. *-
This staging is the defining feature of what is generically called a low NOX burner. This,
then, apparently represents the recent development of an LNB for glass furnaces. Figure
5-7 shows the staging of the natural gas fuel in a primary and secondary flame in a
regenerative glass furnace. As in other LNBs, this staging reduces the peak flame
temperature, and thus NOX formation.
This burner has been tested on two container glass furnaces, as shown in Table 5-
3. In the test on the end-fired regenerative furnace, NOY emissions were reduced from
yv
6.04 to 2.43 Ib/ton (60 percent) from uncontrolled levels by a combination of furnace and
burner block sealing to limit air infiltration (accounting for a reduction from 6.04 to 4.13
Ib/ton) and use of the Cascade burner (accounting for a further reduction from 4.13 to
2.43 Ib/ton). A second test in which one of five ports in a cross-fired regenerative furnace
was fitted with a Cascade burner resulted in overall furnace NOX emissions reduction
from an uncontrolled level of 9.21 Ib/ton to 5.86 Ib/ton.
Both of these tests were on container glass furnaces with "under-port" firing, in
which the fuel is injected below the port from which the preheated air enters the furnace
as shown in Figure 5-7. Although apparently common in container glass furnaces, under-
port firing is not typically used in flat glass furnaces in the United States, though it is used
in flat glass furnaces elsewhere in the world. Thus, the use of this burner in flat glass
furnaces, has not been demonstrated and may present some difficulties.^ No information
is available on the applicability of this burner to pressed/blown glass furnaces.
5-13
-------
Tertiary Air
Outer
Secondary Air
Inner
Secondary Air'
Coal and
Primary Air
I
Very Fuel-
rich Zone
(Average
Stoichiometry
40%)
Progressive Air
Addition Zone
(Overall
Stoichiometry
70%)
Final Air Addition Zone
for Burnout (Overall
Stoichiometry 120%)
Figure 5-6. Low-nitrogen oxides burner with multistage combustion.7
5-14
-------
TABLE 5-3. RESULTS OF NOX TESTS USING CASCADE
BURNER
NO emissions
Furnace
Uncontrolled
With basic
measures3
Ib/hr
Ib/ton
Ib/hr
Ib/ton
With basic
measures and
Cascade
Burner
Ib/hr Ib/ton
End-fired, regenerative
container glass
70 m2
220 ton/day
oil-fired
6% electric boost
Cross-fired, regenerative0
container glass
94m2
255 ton/day
oil fired w/natural gas
atomization
60.9
6.04
41.6
4.13
23.1
107.7
9.21
basic measures
(not applied)
68.5
c
2.43C
5.86C
a "Basic measures" include the following: furnace and burner block sealing to prevent cold air
infiltration; optimization of furnace pressure; reduction of furnace temperature; optimization of fuel
exit velocity, burner angle, primary air, burner nozzle cooling.
° Allowance has been made for electric boost, i.e., actual emissions measured with 6 percent electric
boost have been increased by a factor of 1/0.94 or 1.06 to show what NOX emissions would be
without electric boost.
c Only one of five ports was equipped with a Cascade burner ; apparently this furnace was not
electrically boosted
5-15
-------
Secondary Flame
Air
Natural Gas
Fuel
Figure 5-7. Sorg Cascade™ burner.32
5-16
-------
5.2.2.2 Kortinq Burner. Korting (Hannover, Germany) has reported the
development of a "reduced NOV burner" that incorporates orifice sealing (to prevent in-
J\.
leakage of air), flue gas recirculation, and a "staged air" system to minimize IMOX.
This "staged air" process injects additional air into the end of the furnace outside of the
burners, and is therefore not the same as the staged air referred to above for LNBs (see
Figure 5-8). Figure 5-9 shows the burner itself. Natural gas enters through a jet nozzle,
creating a vacuum to draw in atmospheric air. Control of this "primary air" can be used to
vary the velocity of the gas/air mixture from the burner tip and provide enough air so that
partial combustion of the gas, at 800 to 1000 °C (1470 to 1830 °F), takes place. This
burner was tested on a 179 tons/day regenerative end-port gas-fired container glass
furnace. No reports of its use on flat or pressed/blown glass furnaces are available.
The uncontrolled NOX concentration was approximately 2,240 ppm. For this test, the
"atmospheric air" of Figure 5-9 was replaced by 280 °C (535 °F) flue gas drawn from the
regenerator and is shown in Figure 5-10. This reduces NOX by minimizing the oxygen
content of the combustion air. The net effect of the orifice sealing, flue gas recirculation,
and staged air was to reduce NOX concentration to 600 to 750 ppm, i.e., by around 65 to
70 percent. Staging of the air had the greatest single effect on NOY reduction, about 50
y\
percent by itself. Table 5-4 summarizes more detailed data on this same furnace. °
From baseline emissions of 2,284 ppm from one group of burners, flue gas recirculation
and staged air reduced NOX emissions by 16 to 44 and 66 percent, respectively.
Combining the two techniques gave no improvement over staged air alone, at least for the
14 percent staged air tests for which direct comparisons can be made. Also note that
decreasing the oxygen concentration from 4 to 3.7 and 2.7 percent using flue gas
recirculation lowered NOX emissions by 24 and 44 percent of the baseline value but
increased CO emissions, as expected (see Figure 5-17 and Section 5.3.1).
5.2.3 Oxvoen Enrichment/Oxv-Firina
Oxygen enrichment refers to the substitution of oxygen for nitrogen in the
combustion air used to burn the fuel in a glass furnace. This enrichment can be anywhere
from its level in ambient air (21%) up to nearly 99 percent. Oxygen enrichment above 90
percent is sometimes called "oxy-firing." Oxy-firing has been demonstrated only in
container^" and pressed/blown^'4^ glass furnaces to date, not in flat glass furnaces.
The conversion of a small (85 ton/day) "flat glass" furnace to oxy-firing is discussed.^
However, this furnace does not produce the high quality glass made by the float process in
5-17
-------
Dog House
Staged Air
Injector Pump
Compressed Air
Figure 5-8. Air staging on a regenerative horseshoe-fired furnace. 1 to 5: sight
hole numbers of the furnace.36
5-18
-------
Natural Gas Atmospheric Air
Figure 5-9. Kflrtlng gas jet.34
5-19
-------
Preheated Air (about 1300 °C)
Flue Gas Recirculation
Figure 5-10. Flue gas reclrculatlon on regenerative glass melting furnaces.36
5-20
-------
CO
CO
CO
z
o
CO
CO
Ul
X
o
z
z
o
z
o
<
o
LU
Q
0
5
CC
Ul
z
cc
3
CO
LU
o
1-
0
LU
LU
U.
Ul
i
U>
Ul
_J
CO
<
p
E
0
* Co
c #
o — *
•4= CO
0 _C
•o "3
CO (0
>- CO
x-°
o
z
.a
c
C
Q.
O
u
.E
n\
w*
o
JM
Q.
a
CO
X
0
z m
.E
^>
E
i?
§
^
CM
O
modification
k
Q)
h.
03
i CO ^
>
CD 0
<^ C^
4-< ;K
co co £
10 «
»_ rv
0 CM
6 /a
s @
ro co
CU c
^: t
> E
ja t
T3 00
CU CD
T3 CD
'> CM
f*
2 -o
Q. C
4-- re
0 i-
C £
cu ro
ro E
T3 CM
0) CD
O CM
T3 (3)
0 ^
Q. ^°°
^ E
to w ^
S2 > co
— '*= p
CO O c-
c X co
o S-m
^ cu •*
x s CM
O _ -
21 ^ lo
.0 ••§ 8
— D. Xro
•*- 1 — *~» •-
°ioi >
CO -*
4-J 4-1 ~ Q
.^ ro co ^
c c •*-
=> O fc c
c O iv o
•- O) 'P
co t O ro
c Q. co 3
° °--D o
io 0 £ -!=
•|S S g
£ o co •-
w cn w
x II cn re
0^ CM co
T»CO . cu
z P ® 5
0> -t 0 ==
ro 0 g ^
•5 O t c
I?-2!
o E "S t
^* o **~
27 : ^^
^^^1
"S"0^"^
® ro - ^
CTO C/J
Tl
cu 0 c 15
is Z re
" f *- "
-c E .c c
.0 Q. CD -5
2 Q.-C °
> ,_ ,_ "•
^ *> o ro
v [V $L 4->
S 0 cu "S
2 H J2 0
Jco > 4?
o E >*-'*-
q= -5 o 0
to O gj g)
ro Z co o)
co _. ro ro
O) ^ u.
CU c CU 0),
3 c > >C
LU «- < <
TO .£2 O T5
H-
O
E
ro
CO
4-1
co
^
O
T3
T!
CU
4^
o
cu
'c
k_
'ro
c
g
'4-*
C/3
3
jQ
E
o
o
_
ro
4-1
O
+-•
0)
_c*
+••
M-
0
+-»
c
cu
o
k-
QJ
a
<*
QJ
b.
'ro
•o
cu
CO
ro
4^
CO
4-J
C
cu
0
cu
a
"Sf
T3
c
3
O
ro
4-<
ro
C
cu
.*:
ro
4-<
cu
^ 1^-
ro co
i
.^ LD
re a,
T3 •-
CU =
CO OO
ro u.
CO CU
I- «
o
-------
much larger furnaces, but rather lower quality, rolled "flat glass." Thus, oxy-firing has not
yet been demonstrated in what is called "float" (or "flat") glass furnaces herein.] Little has
*
been reported on oxygen enrichment in glass furnaces at total ©2 concentration levels of
less than 30 percent. Enrichment to these low levels can be done in two ways41 :
• Oxygen enrichment.
This technique is sometimes called "premix." Oxygen is added directly to the
combustion air to prolong furnace life and increase productivity. It is usually used to
enrich the combustion air up to about 35 percent C>2 and is the most practical for
retrofit situations since most air-fuel burners can be used without major
modification. This usually increases NO consistent with Figure 5-1 1 . Enriching
the combustion air oxygen content from 20.9 percent to 21.7 percent would be
expected to increase the flame temperature by 1 1 °C (20 °F) and to increase NOX
emissions by 10 percent.
• Oxygen lancing.
This technique is sometimes called "undershot." Pure oxygen is injected below an air-
fuel burner to increase productivity. NOX is usually not greatly affected, though at
least one report describing a modified oxygen lancing technique used to combust
around 4 percent of the total fuel at four container glass plants in the UK, showed NOX
increased from 968 ppm to 1073 ppm, about 1 1 percent. ° Field data show that
"improper" lancing of corresponding to 3 percent oxygen enrichment (i.e. from 21 to
24 percent Q£ in the combustion air) actually doubled NOX emissions.
Because only oxy-firing generally results in lower NOX emissions, it is the primary
focus here, though lower levels of oxygen enrichment have been reported on glass
furnaces.
The basic rationale for oxy-firing is improved efficiency, i.e., more of the theoretical
heat of combustion is transferred to the glass melt and is not lost in the flue gas. Many of
the combustion modification techniques discussed (e.g., flue gas recirculation, staged
combustion, and low excess air combustion) reduce NOV formation but also reduce the
xv
combustion efficiency. Oxy-firing was originally developed to improve the combustion
efficiency primarily by eliminating the sensible heat lost in heating the nitrogen present in
air, which is then lost in the flue gas.4^'4^ The equations below compare oxy-firing
combustion of methane with conventional combustion using air:
In air
2O2 + 7.5 A/2 -* CO^ + IH^O + 7.5
5-22
-------
3
S
Energy into Product
30
40
50 60 70
O2 in Oxidizer (%)
80
90
100
Figure 5-11. Adiablatlc equilibrium NO (given In Ib/MM Btu) versus percent oxygen
In the oxidlzer for a methane flame based on gross energy Input
(overall firing rate) and net energy Into the product. Right-hand
scale Is calculated assuming 6 MM Btu/ton glass (see Chapter 4).
5-23
-------
Oxv-firinq
C#4 + 2O2 -» CO2
The difference is that heat is lost as the nitrogen in the combustion air is heated
and then sent up the stack. Also, the volumetric flow rate of the flue gas is 3.5 times
larger when air is used than when oxygen is used. This increases fan, duct, and any gas
treatment (e.g., SNCR) costs.
Nitrogen, which must be present for NOX to form, is introduced in the furnace from
several sources besides the combustion air. Thus, some NOX formation is inevitable even
when using oxy-firing. Nitrogen is invariably present in the natural gas fuel used at glass
plants, usually in concentrations from 0.5 to 3 percent. Nitrogen is also an inevitable
contaminant in the oxygen, even when cryogenically distilled oxygen is used, though the
concentration is very low in this case. Nitrogen concentrations of about 100 ppm are
typical. ^ If pressure swing adsorption is used to produce oxygen, the nitrogen content is
around 2 to 5 percent.51 The largest source of nitrogen is usually air infiltration into the
furnace. This is, of course, highly site specific but experience has shown that even the
best pressure controls on the furnace, usually designed to keep the furnace at slightly
positive pressure, allow at least some air leakage into the furnace. In many cases, air
co
infiltration is the single largest source of nitrogen in the furnace. Practical operating
constraints and furnace degradation with time generally mean that the nitrogen
concentration in a working furnace cannot be reduced below 5 to 10 percent, including
CO
nitrogen from all sources. The source of the nitrogen (from the fuel, oxygen, or air
infiltration) can greatly affect the amount of NOX formed. This is to be expected since,
for different burner types, mixing of the N£ in that part of the flame where NOX is formed
is different depending on how it is introduced into the flame.
Increasing oxygen concentration also causes the temperature of the flame to
increase. Any increase in flame temperature will increase the formation of NOV. Figure 5-
A
12 shows the adiabatic flame temperature for methane as a function of the oxygen
content in the combustion gas. 5 In glass melters, the actual flame temperature will be
somewhat less because heat is transferred from the flame to the glass melt. Nevertheless,
a substantial increase in flame temperature, and therefore NOX formation, with oxygen
content would be expected. The increase in flame temperature with oxygen content
5-24
-------
u_
o
5300
5100
4900
4700
4500
4300
4100
3900
3700
3500
I
20
40
60
O2 in N2
80
3200
3100
3000
2900
2800
2700
2600
2500
2400
2300
2200
100
Figure 5-12. Adlabatic flame temperature versus percent oxyaen In the oxidizing
stream consisting of oxygen and nitrogen.55
5-25
-------
results in a higher rate of heat transfer to the glass for a given rate of fuel being burned.
As shown in Figure 5-13, the effect of oxygen concentration on NOX formation is
not straightforward. Increasing oxygen concentration from the 21 percent in ambient air
to around 60 percent actually increases the equilibrium NO concentration. ° This is a
result of the higher flame temperature and higher 02 concentrations. As shown in Figure
5-12, above 60 percent C^, the equilibrium NO concentration decreases, due to the lower
N2 concentration, even though the adiabatic flame temperature continues to increase.
Another way to look at NO formation for glass melting is to plot the weight of NO formed
per unit weight of glass produced, e.g., Ib NO/ton glass produced. Glass production is
directly proportional to net energy transferred to the glass product, which is in turn directly
proportional to the fuel firing rate. Figure 5-11 shows the equilibrium NO per unit fuel fired
(Ib NO/MM Btu) versus oxygen content. The important difference between Figures 5-11
and 5-13 is that the NO produced, at equilibrium, per unit of glass produced, actually
decreases monotonically above about 30 percent ©2, rather than above 60 percent 02
that might be expected from Figure 5-13.
This trend in equilibrium NO concentration, shown in Figure 5-11, was confirmed in
cc
practice, at least qualitatively, in a series of tests funded by the Department of Energy
and Gas Research Institute. /5' Figure 5-14 shows the actual NQ produced per unit fuel
input as a function of oxygen content, for oxygen concentrations above 90 percent. This
corresponds to the upper end of the theoretical plot given in Figure 5-11. The trend in NO
production at this level of 02 is important since the nitrogen concentration in a working
CO
glass furnace is 5 to 10 percent, corresponding to oxygen concentrations of 90 to 95
percent, as shown in Figure 5-13. The NO produced in these tests is actually somewhat
less than predicted by the equilibrium values given in Figure 5-11, suggesting that the
formation of NO in a working furnace is a rate-controlled process rather than a
thermodynamically controlled one. This is why Equation (4-4), Section 4.1.1, shows NOX
concentrations to be linear with nitrogen concentrations rather than proportional to the
square root of nitrogen concentration, as would be expected at equilibrium. Assuming a
value of 6 MM Btu/ton of glass^ (also see Chapter 3), the right-hand scale of Figure 5-14
shows the Ib NO/ton glass produced in these tests. The important result for these series
of tests is that the NOX emissions for high levels of enrichment (>90% 02) were at least
an order of magnitude lower than for low levels of enrichment «28% O2>- This is
5-26
-------
E
Q.
a.
o
-------
0.12
3
s
o
z
.o
I I I
90
91
92
94 95 96 97
02 in Oxidizer (%)
'assuming 6 MM BtuAon of glass
98
99
100
Figure 5-14. Flue nitric oxide versus percent oxygen In the oxldlzer for an Air
Products' K-Tech burner firing on natural gas.46
5-28
-------
contrary to a widely held perception that the use of oxygen inevitably leads to higher NOX
emissions, regardless of the O2 concentration.^7 Also, unlike air-fuel combustion, typical
oxy-firing produces NO at concentrations that decrease with increasing furnace
temperature. This is because NO concentrations that are above the equilibrium value
calculated at the furnace temperature (due to the very high adiabatic flame temperature)
are produced.^ As the oxy-fired flame cools rapidly to a low furnace temperature, a high
NO concentration, corresponding to that produced at the high oxy-fired flame temperature,
is "frozen." If, however, a less rapid cooling takes place, which happens if the furnace
temperature is higher, the NO formed at the high flame temperature decomposes and
approaches the lower value corresponding to the furnace temperature.
Oxy-firing is especially valuable as a retrofit technology. However, conventional
burners must be replaced. Air Products (Allentown, PA) and Combustion Tec (Orlando, FL)
have developed burners that are designed to minimize furnace temperature variations in
retrofit situations, the benefit being about half the fuel usage for the same temperature
fiO fi 1
profile, ' or higher productivity (ton glass produced per unit of fuel fired) from the
same furnace.
Tests by Union Carbide on oxy-firing of glass melters on a pilot scale furnace
showed large differences in NOV produced by different burner "types," which are not
J\.
further described. However, the qualitative trend shown in Figure 5-11 was confirmed,
i.e., NOV (Ib NOV/MM Btu) decreased with increasing oxygen concentration over the range
J\ X
O
of 35 to 100 percent 02- Larger scale tests were conducted on a 75 tons/day, 300 ft
end-fired regenerative container glass melter fired with pure oxygen. Table 5-5 shows the
results, comparing air-fired with "100 percent oxy-firing." [It was not possible during
these tests to get NOX emission data at identical production rates (ton of glass/day).
Therefore, the data in Table 5-5 provide only qualitative comparison of air versus oxy-
firing.] The higher than expected nitrogen content of the furnace atmosphere in Table 5-5
during the two periods of "100 percent oxy-firing" (38 percent and 30 percent) are due to
large infiltration of air into the furnace. This, of course, contributes to higher levels of NOV
J\
formation than would otherwise be the case. Also, the batch ingredients for this container
glass contain 7.5 Ib niter (as NaNOg) per ton of glass produced. If this were all converted
to NO2» it would yield 2.7 Ib NO2 per ton of glass. Though the actual conversion to N02
is probably less than complete, this accounts for most of the higher than expected NOY
/{
values (2.9 and 2.1 Ib/ton glass) for the two oxy-firing cases in Table 5-5. The high
5-29
-------
TABLE 5-5. NOX EMISSIONS-75 TPD GLASS FURNACE54
Pull (ton/day)
Bridgewall temperature (°F)
Fuel (MM Btu/hr)
Flue gas (scfm)
Furnace atmosphere
N2 (% wet)
H20 (% wet)
CO 2 (% wet)
O2 (% wet)
NOX (Ib/hr)
(Ib/MM Btu)
(Ib/ton)
NOX from niter (@ 1 00%
conversion)
(Ib/hr)
(Ib/ton)
Air
62.7
2676
13.6
200,000
72
14
9
5
56.4
4.28
21.6
7.0
2.7
Oxygen
46.8
2672
8.9
53,000
38b
36
22
4
5.75
0.68
2.9a
5.2
2.7
Oxygen
75.8
2766
13.7
66,000
30b
43
26
1
6.5
0.5
2.1a
8.5
2.7
a Most NOY from niter.
bj\
Thic KlnH r\il-rr\nar* nonoontration \A/ac Hi 10 tn onnciHorahle infiltratinn r\f air inti
furnace.
5-30
-------
nitrogen contents of the furnace atmosphere contributed to NO formation in addition to
A
the niter, though the contribution of this outside air to NOV is not known. Nevertheless,
s\
these tests on an actual operating furnace showed NO reductions of 86 to 90 percent
A
from baseline levels using oxy-firing (from 21.6 to 2.9 and 2.1 Ib/ton, respectively, for the
two oxy-firing tests). A later test at a 100 ton/day container glass furnace with less air
infiltration and which did not contain substantial niter gave NOX emissions of less than 0.2
Ib NOv/ton glass produced «0.05 Ib NOV/MM Btu). This is consistent with values
A A
expected from Figure 5-14.
Corning, working with Linde Division of Union Carbide (now Praxair), has converted
OQ CO
34 of its furnaces to oxy-firing as well as the Gallo plant in California.00'00 "80-plus"
percent NO reduction with oxy-firing, presumably representative of the 34 furnaces
<5Q
installed as of 1991 has been reported.00 The Gallo plant reports 84 percent reduction in
NOV (from 5.03 to 0.81 Ib NOv/ton of glass corresponding to a reduction in NOV from
X A A
1.34 to 0.24 Ib NOX/MM Btu°^ and is the largest oxy-fired glass furnace reported as of
1991 (400 ton/day, 1248 ft2). Related work showed NOX generation as 0.3 Ib NOX/MM
Btu corresponding to around 1.8 Ib NOv/ton glass, assuming 6 MM Btu/ton of glass. A
s\
CC
general value of less than 2 Ib/ton for oxy firing has been estimated. Table 5-6
summarizes the reported NO emissions reductions discussed above.
5.3 PROCESS MODIFICATIONS
Process modifications include changes to the furnace, its combustion system, or its
heat recovery system that have the effect of lowering either the NOV emission rate (Ib
A
N0../hr) or normalized NOY emissions (Ib N0v/ton of glass produced). In many cases, such
A A A
modifications are designed to increase furnace productivity (tons glass produced/hr) with
lower NOX emissions being an unintended benefit. This is the case for the three
process modifications considered here.
5.3.1 Modified Furnace
5.3.1.1 Teichmann System. Teichmann/Sorg Group, Ltd., has developed an
TM
LoNOv furnace that incorporates cullet preheating using furnace exhaust gas into a
A
modified melter design that also uses lower than normal combustion air preheat.
The basic furnace design is shown in Figure 5-15. The combustion air and fuel are
preheated in the convection recuperator section. The combustion takes place in eight
burners, four on each side. The exhaust gas passes over the melt, heating it, and exits
5-31
-------
o
z
E
u.
^
s/
O
2
O
QC
u.
CO
o
CO
CO
i
Ul
X
o
z
<£>
If)
Ul
_J
CO
H-
cu
u
c
£
0)
"S
cc
§S
"4^ *""
U C
3 «
•o «
O .O
i*l
X
O
Z g
•o o
= "*»
"o —
t E
C CD
O
O
3
£
2
o
•t:
•S
aseline NOX
emissions
OQ
2
OQ
S
-
c
o
^
^_
"5 ~
o. >
•fl) ^^
i?
Eo
w
3 —
U.
CD
«••
CD
U
CD
emissions.
X
0
"o
CO
01
"CD
CU
03
y-
4_t
a
•j3
CD
2
^J
CD
U
C
,__
O
\J
CU
c
accoui
j
E
'co
o
CU
3
CO
CD
CU
E
|
CU
^
a>
c
CD
*^
CO
^
^
0
C5
c
CO
£
2
•^
CO
6
"S
CU
_D
CD
>
he measured
1
o
c
E
o
CO
4"1
CO
•o
+•'.
o
Q.
CL
3
C
D
CO
CU
;o
^
O
o.
c
o
Q
O
CO
£.
O
k.
»*~
1
15
0
CU
•°
CO
u
^
^
q
"o
3
5
co
*±
XI
•o
5
o
c
CD
CO
"co
CO
"E
C
C
^5
(0
CO
m
D
c
03
O
CD
c
_ J
0)
>
*"•
CO
^r
g
O)
5-32
-------
-Convection Recuperator
x.^^^^.^^^^^^..^^^^^^^^^^.^^^^^.^^^^^^.^
Radiation Recuperator
Gullet
Preheater
^^^^v^^^^v^^^^^^^^,.^^.,,^.^^^.^^.^^.^..^^
Fuel
Figure 5-15. General arrangement of Telchmann/Sorg LoNOx™ furnace.70-71
5-33
-------
each side. It then passes upward through parallel radiation recuperators, turns downward,
and passes through the convection recuperators. From there, the exhaust gas enters a
crossflow cutlet preheater and finally exhausts through the stack (Figure 5-16). An energy
T)
balance on the preheater itself is shown schematically in Figure 5-17.
The combustion air is preheated to only about 700 °C (1,290 °F), about 550 °C
(990 °F) lower than an efficient regenerative furnace. '0,73 This lower preheat would be
*
expected to require a higher input of fuel to achieve the same furnace temperature,
resulting in higher normalized NOX emissions (Ib NOX per ton of glass produced).
However, this is more than compensated for by the heat recovery in the two recuperators.
This furnace also uses electrical boost (Section 5.3.2), with nine electrodes inserted in the
preheating end to control the glass temperature and viscosity. This electrical boosting
reduces NO emissions since electrical energy is substituted for thermal energy in the fuel.
j\
TU
The initial installation of this LolMOx furnace was a 200 ton/day, natural gas-fired
container glass furnace which began operation at Weigand Glass in Steinbach, Germany, in
1987. A second one, 300 ton/day, has been ordered for the same plant and is under
construction. The first furnace operates with a batch of 80 percent cullet, resulting in
an energy consumption of 3.1 x 10° Btu/ton, about half that shown in Chapter 3 for
virgin batch materials. Design calculations show that at 30 percent cullet, the energy
fi "74-
consumption would be about 3.4 x 10 Btu/ton. Table 5-7 shows the NOV emissions
A
over a 6-month period shortly after startup. These are at a somewhat less than design
glass production rate, 170 ton/day versus 220 ton/day design, and the normalized
emissions, Ib NOV per ton of glass, would presumably be lower at design capacity. The
A
results show emissions of less than 1.45 Ib NOx/ton glass.
5.3.2 Gullet/Batch Preheat
Chapter 3 describes the inherent thermal inefficiency of the glass melting
operation, with roughly one-third of the energy input being lost in the flue gas. This is the
basic reason for the development of cullet preheat systems, which, to date, have been
demonstrated only in container glass production. If some of this energy is recovered, less
fuel is needed to produce a given quantity of glass and the normalized NOX emissions (Ib
NOv/ton glass) are reduced. Reductions in NOV emissions are directly proportional to the
yv A
lower fuel requirements—if a cullet preheater reduces fuel usage by 10 percent, NOV (Ib
s\
NOv/ton glass) should decrease by 10 percent, all else being equal. Two different process
s{
configurations have been developed.
5-34
-------
Gullet
Exhaust to Stack
250 - 300 °C
Exhaust from Furnace
Maximum 550 °C
Gullet
Maximum 450 °C
Figure 5-16. Crossflow cullet preheater.70'71
5-35
-------
Waste Gas Out
66%
Energy to Evaporate
HoO in Gullet
4%
Energy to Heat Gullet
24%
Wall Losses
6%
Figure 5-17. Gullet preheater energy balance.70'71
5-36
-------
TABLE 5-7. NO^ EMISSIONS FOR FURNACE WITH TEICHMANN LoNOj" FURNACE75
Date
Spring 1988
Fall 1988
Summer 1989
Tons/day3
169
178
195
mg/nM3
Corr. 8% O2
400
412
421
kg/hr
3.31
4.9
5.3
NO,
Ib/hr
7.29
10.8
11.7
Ib/ton
1.02
1.45
1.44
a This is reported as "M. tons per day," which is assumed to be metric ton per day. The numbers
reported as such by Moore have been put, above, into English "tons."
5-37
-------
5.3.2.1 Tecoqen System. A cullet preheat system developed by Tecogen, Inc.
(Waltham, MA) operates in a different way from that shown in Figure 5-15. As shown in
Figure 5-15, rather than using the sensible heat of the exhaust gases from the melting
furnace, the cullet preheater itself has small dual natural gas burners (total capacity 2 MM
Btu/hr) to preheat the cullet (Figure 5-18). In effect, this allows some of the fuel that
would otherwise be needed in the melting furnace to be burned at lower temperatures,
resulting in lower NOX emissions for the same energy input. An earlier version of this
system shows a slightly different arrangement of this preheater.^ The principle of
operation is that heat is transferred from the upward flowing natural gas burner exhaust
gases to the downward flowing cullet. The cullet is preheated to 205 to 260 °C (400 to
500 °F).7° Unlike the LoNOx melter described above, this system is not an integral part
of the furnace design and could presumably be more easily retrofit. Figure 5-19 shows the
increase in furnace production as a function of percent cullet in,the batch (these are
calculated numbers, not test results).
This system was installed at the Foster Forbes container glass plant in Milford, MA,
7Q
producing 240 ton/day and was tested over a 5-day period in 1989. The cullet
preheater was designed to preheat 20 to 100 ton/day, but was operated between 1 2 and
78 ton/day for these tests. This corresponds to between 5 and 30 percent of the batch
as cullet (accounting for 10 percent loss from batch to final product; i.e., 264 ton/day of
batch ingredients is needed to produce 240 ton/day of glass). The results of this test
showed that the specific energy use (MM Btu/ton glass produced) declined about 7
percent. All other factors being equal, this would correspond to about a 7 percent
reduction in normalized NOY emissions (Ib NO /ton glass produced). Calculated curves of
/v yv
the expected reduction in normalized NOV emissions as a function of percent cullet in the
/v
batch are shown in Figure 5-20 for a cullet preheat temperature of 480 °C (900 °F).81
As expected, the higher the proportion of cullet, the higher the reduction in NOX
emissions.
Earlier results from a 1987 test of 1670 hrs on a slightly different configuration of
the preheater (compare Figures 5-21 and 5-26) were made using higher cullet preheat
temperatures, around 455 to 516 °C (850 to 960 opj 82,83 |mp0rtant differences in
these two preheaters include the use of natural gas burners, the apparent lack of
mechanical support for the cullet in Figure 5-18, the use of regenerator offgas, and a
moving grate in Figure 5-21. These tests were also done at the Foster Forbes plant. The
5-38
-------
Cold Gullet
Exhaust
Heat Exchanger
Feeder
Air
Natural Gas
Preheated Gullet
Figure 5-18. Gullet preheater concept by Tecogen.
76
5-39
-------
100-
80-
I 60-
§
1
40-
20-
900 °F Gullet Preheat
Cold Gullet
20 40 60
Gullet Fraction (%)
80
100
Figure 5-19. Production increase available with preheated cullet.79
5-40
-------
40
35-
E
§ 30,
&
t§ 25-
20
15
10-
5 -
.1
"8
cc
900 °F Gullet Preheat
20 40 60
Gullet Fraction in Batch (%)
80
100
Figure 5-20. Reduction In specific NOX emissions with cullet preheat.
79
5-41
-------
Exhaust
Exhaust Gases
from Regenerators
Cyclone
Glass Tank
Figure 5-21. Fluidlzed-bed glass batch preheater.
5-42
-------
unit was apparently designed to preheat not only cullet but the entire batch, using exhaust
flue gases from the regenerator rather than independent natural gas burners for preheating
only cullet. Tests were made at preheater throughputs from 90 to 225 ton/day on an end-
port fired, natural gas-fired furnace. This plant has an interruptible gas supply and burns
Q O
heavy fuel oil in the winter months. Figure 5-22 shows the installation. The preheater
design throughput was 165 tons/day, although it achieved a rate of 225 tons/day for one
8-hr period.
The results of the tests showed a 7 to 8 percent less net energy usage rate when
the preheater was operated near its design capacity. Apparently only about 30 percent
(4, 400 scfm) of the flue gas was recycled to the preheater since this was all that was
needed for the preheater to function at design capacity. Measurements of the gases
from the preheater alone showed that the NOX emissions were about 0.58 Ib N0x/ton
glass. This unexpectedly low value was attributed to the reaction of NO in the flue gas
with ingredients in the batch, e.g.,
2FeS + NO-» 1/2N2 + FeO + FeS2
2NO + C-*N2 + CO
N + O
AI,O
'2U3
The first two reactions are simply gas-solid reactions in which NO is reduced to N2
by the FeS and C (carbon) ingredients in the batch. The third is a catalytic reaction in
which alumina (AI2Og) is said to act as the catalyst. There was no decrease in the glass
p O
quality in these tests, suggesting that these reactions do not affect product quality.
However, "furnace dusting problems," not further described, caused the tests to be
R7
discontinued.
Because only 30 percent of the total flue gas from the melting furnace can pass
through the preheater, the overall NOX emissions reduction from the entire furnace is not
as great as if all the flue gas went through the preheater. NOX emissions decreased by 81
percent (from 17.4 to 3.3 Ib NOx/hr) for that part of the overall flue gas passing through
the preheater, corresponding to a 24 percent decrease in the overall NOV emissions (from
XV
58 to 44 Ib N0x/hr) from the furnace. This, in turn, corresponds to a 39 percent decrease
sd M
88
in normalized NOV emissions, from 5.4 to 3.3 Ib N0v/ton of glass produced, from the
f\ X
furnace.'
5-43
-------
Pneumatic
Conveying
System
from
Batch
House
Cyclones
Standby
Batch and
Cullet
Hopper
Screw
Feed
Conveyors
Hot
Screw Feed
Conveyor
Furnace
No. 16
Regenerator
Figure 5-22. The glass batch preheater system Installed at Foster Forbes.
83
5-44
-------
5.3.2.2 Zippe System. A third cullet preheat system by Zippe Industrieanlagen
pQ
GmbH (Germany) is reported by Zippe.00 Units have been installed at two furnaces in
Europe, one (Vetropack) producing 300 tons/day using 100 percent cullet feed. On this
plant, the preheater is used for at least 50 percent of the total cullet throughput. The unit
is a cross-flow countercurrent heat exchanger in which, unlike the Teichmanm and
Tecogen systems, the cullet is heated indirectly. The cullet flow inside the preheater is by
gravity. After passing through the preheater, the cullet is conveyed by a vibrating tray to
the batch charger. The speed of the material through the preheater is about 6 to 12 ft/hr.
Flue gas at around 550 °C (1,020 °F) is used to heat the cullet from ambient to 300 to
350 °C (570 to 660 °F). Apparently, natural gas burners can also be used. No
information is provided on NOX reduction, though calculations shows energy consumption
would be reduced by 12 percent if all the cullet at Vetropack were preheated. Assuming
all other process conditions are constant, this would correspond to a 12 percent decrease
in normalized NOV emissions (Ib NOv/ton of glass produced). A second system has been
j\ x
OA
installed at a 300 ton/day end-fired container glass furnace. The preheater is used for
all melting material, which consists of 70 percent cullet and 30 percent batch.
5.3.2.3 Nienburqer System. A third cullet/batch preheat system (Figure 5-23) has
been demonstrated in Germany by Nienburger Glas GmbH on two container glass
furnaces. The first installation of this system was in 1987 on a 300 ton/day cross-fired
furnace with 80 percent cullet. This furnace operates with 600 to 800 kW electric boost
with a specific heat input of 3.2 MM Btu/ton. No information is provided about the heat
input without the preheater, which would allow an estimate of NOX emission reduction. A
second furnace was equipped with a batch preheater in March 1991. This is a 350
ton/day cross-fired container glass furnace using 30 to 50 percent cullet. The batch is
preheated from ambient temperature to 270 to 290 °C (550 to 590 °F) and the specific
heat input was 3.2 MM Btu/ton with no electric boost. Tests without the preheater
showed a heat input of 3.8 MM Btu/ton, corresponding to a 20 percent decrease in heat
input with the preheater. This corresponds to a 20 percent decrease in NOV emissions.^
xv
An additional decrease in NOV emissions is claimed due to a reduction in the furnace
x\
crown temperature of about 50 to 60 °C (from 1,590 to 1,600 °C to 1,530 to
no
1,550 °C). Actual flue gas NOX concentrations with the preheater are less than 1,490
ppm, corrected to 8 percent O2, dry, but the corresponding gas flow is not given, so that
the calculation of NOX in Ib N0x/ton glass cannot be made.
5-45
-------
i
0)
I
0.
o
*•*
CO
JD
D)
3
_C)
Z
0>
*"*
"o
E
D)
|
JE
CO
in
I
CI
5-46
-------
5.3.3 Electric Boost/Electric Melting
Electric boosting is the use of electrical current, passing between electrodes
submerged in the glass melt, to resistively heat the batch materials. This is done by
placing electrodes, generally made of molybdenum, through the sidewalls or furnace
bottom into the glass melt. Because of differences in quality needs, furnace size, and
temperature-resistivity relationships for different batch materials, electric boosting is
employed only in the container industry. At a given glass production rate, electric boost
allows a reduction in the furnace temperature and therefore in gas-firing rate and NOX
emissions. Reduction in NOX emissions is directly proportional to the percent of the
Q4.
furnace energy supplied electrically.
Electric boost is common in container glass furnaces and in some pressed/blown
furnaces. However, it is not now used in float glass furnaces because of problems related
to productivity, sidewall erosion, glass quality, and furnace campaign life. ° A 1989
survey for GRI of 41 glass melting companies, including some of the largest manufacturers
presented in Chapter 4, showed that 60 percent of these companies use electric boosting
Qfi
in their process. These 41 respondents represent 90 percent of the glass produced per
year in the United States by weight. The reason for electric boosting is often to
increase furnace production (ton glass produced/day) without adding an additional furnace
or otherwise modifying an existing one. There are also certain areas of the country where
business arrangements with gas and electric companies make electric boosting favorable.
The effect of electric boosting on NOX emissions was studied on container glass,
side port furnaces from 400 to 1200 ft in size. ° Figure 5-24 shows the reduction in
NOX emissions (Ib NOx/hr) as a function of furnace production rate (ton glass produced per
day). This figure compares actual (points) and predicted (lines) values for NOV emissions.
X.
Electrical boost appears to lower NOX emissions, as expected (e.g., compare the two data
points at 275 tons/day for 700 kW and 950 kW of boost), though the predictions (lines)
are inaccurate. The increase in the NO emission rate in going from no boost ( — 60 Ib
N0x/hr at 220 tons/day) to 700 kW (-75 Ib IMOx/hr at 280 tons/day) actually corresponds
to a slight decrease in normalized NOV emissions from 6.5 Ib NOv/ton of glass with no
*{ x\
boost to 6.4 Ib NOx/ton with 700 kW boost. Figure 5-24 shows that the use of 950 kW
boost permitted the furnace throughput to increase from 220 tons/day (with no boost) to
280 tons/day with an actual decrease in NOX emissions from 60 Ib/hr to 40 Ib/hr
(corresponding to a reduction from 6.55 Ib NOv/ton at 220 tons/day to 3.43 Ib NO /ton at
^ X
5-47
-------
80-
60-
tn
c
o
V)
'{= 40.
UJ
o
20-
100
No Boost
950 kW
O No Boost N
D 700 kW V Actual
x 950 kW J
200
Load (T/D)
300
Figure 5-24. Rate of NOX emissions versus load for 928 ft2
amber glass furnace.94
5-48
-------
280 tons/day). An equivalence between electric boost and glass production is
estimated to be 25 tons of glass/day per 1000 kVA (or 1 ton glass per 800 kWh).100 As
discussed in Section 5.3.2, electric boost is more efficient than gas firing, i.e., more of the
theoretical energy input to the melt electrically is actually transferred to the melt. This
efficiency value for electric boost is 73 percent compared to about 30 to 35 percent for
gas firing ^ (see Section 3.2). However, the production and distribution of electricity
from fossil fuels are only about 20 to 25 percent efficient, making electricity from fossil
fuels less efficient than gas firing.
Of course, all-electric melting is simply a logical extension of electric boost. All
electric melters, however, are limited by current technology to furnaces that are smaller
(roughly half the size) of conventional gas-fired furnaces for container glass production.
Only 3 percent of respondents in the 1989 GRI study use electric furnaces solely
for their melting.^2 An all-electric melter was installed at the Gallo Glass Company
1 D"3
(Modesto, CA) in 1982. Its design capacity was 162 tons/day. Average energy
consumption was 880 kWh, corresponding to 3 MM Btu/ton. Energy efficiency was
73 percent (i.e., 880 kWh/ton was input as electrical energy to melt a batch formulation
with a theoretical melting energy requirement of 645 kWh/ton). As expected, this energy
1 no
consumption gradually increased with time to maintain a constant production rate.
Glass quality was acceptable and the furnace was operated over a 3-year campaign before
being rebuilt. Furnace campaign life is typically longer than this for gas-fired furnaces,
e.g., 8 to 1 2 years for flat glass furnaces. Of course, there are no NOX emissions directly
from this all-electric melter. NOV would be generated, indirectly, if fossil fuels are used in
s\
the production of electricity.
5.4 POSTCOMBUSTION MODIFICATIONS
5.4.1 Selective Catalytic Reduction (SCR)
SCR is the reaction of ammonia (NHg) with NOX to produce nitrogen (N2) and
water vapor (HoO). The two principal reactions are:
4NH3 + 4NO + O2-*4N2 + 6H2O (5-2)
4NH3 + 2NO2 + O2 -* 3N2 + 6H20 . (5-3)
Reaction (5-2) is the reduction of NO, Reaction (5-3) the reduction of N02. Reaction (5-2)
is by far the most important since 90 to 95 percent of the NOV in the flue gas is NO. To
A
achieve reaction rates of practical interest, a catalyst is used to promote the reaction at
5-49
-------
temperatures of around 300 to 450 °C, (570 to 840 °F) which may be somewhat lower
than those in the flue gas of a glass furnace. Relatively new zeolite-based catalysts can be
used at temperatures more typical of glass furnace' flue gas (500 to 550 °C). °
In practice, an NHg/NO mol ratio of 1.05-1.1/1 is used to obtain NOX conversion of
80 to 90 percent with a "slip" of unreacted ammonia downstream of the catalyst of about
20 ppm.10° The catalyst is typically a mixture of vanadium and titanium oxides
supported on a ceramic monolith, as shown in Figure 5-25.
SCR units have been installed on a number of utility boilers, gas turbines, internal
combustion engines, and process heaters, and SCR is considered commercially
demonstrated. As of late 1992, there are no reported operating SCR installations on glass
furnaces in the United States; however, SCR units have been reported on container glass
plants in Europe. Oberland Glas (Neuberg plant, Germany) reported the installation of an
"SCR-DeNOx" unit on their glass melter flue gas, but few details are provided beyond
problems with fouling of the catalyst by particulates. ** The flue gas is treated in three
consecutive steps:
• Adsorption of acidic compounds by hydrated lime injection,
• Particulate removal, including reacted lime, and
• SCR.
The unit was started up in October 1987 and achieves a reported 80 percent reduction of
NOX, from 1,420 ppm to 283 ppm.109 The flue gas flow rate is 35,300 scfm and the
operating temperature is 350 °C (660 °F).
A higher temperature zeolite-based SCR process called "CER-NOX" is used on a 500
tons/day glass furnace in Germany. '® This catalyst is supplied by EESI (La Mirada, CA),
apparently under license from Steuler (Germany). About 100 of these SCR units are
installed in Europe on processes such as cogeneration and gas turbines. Figure 5-26
shows a schematic of the process, which also includes hydrated lime injection and an
electrostatic precipitator upstream of the SCR unit. The SCR unit treats flue gases from
three glass furnaces using a 25 percent aqueous ammonia injection system (rather than
gaseous anhydrous NHg used in some other SCR units.
The process achieves a reported 80 percent reduction of NOX emissions (from 925
to 195 ppm) at 10 to 30 ppm ammonia slip. The flue gas flow rate is 29,500 scfm and
the inlet temperature to the SCR unit is around 175 °C (350 °F). This temperature is
somewhat lower than other glass furnace flue gas temperatures because of the injection of
5-50
-------
'O€TO €
VYTTT
rodHsc
58*
5-'i ' ~! *
Figure 5-25. Unit cell detail of a monolith SCR catalyst.107
5-51
-------
scyso3
HF/HCI
Reactor
Figure 5.26. Installation of SCR unit on glass furnace.110
5-52
-------
hydrated lime upstream of the SCR unit. Using these values, and a reported furnace
production of 500 ton/day of glass, the NOX emission reduction ("NOX" is calculated by
the authors as NC^) can be calculated as being from 10.1 to 2.1 Ib NOx/ton glass
produced (i.e., from 1.68 to 0.35 Ib NOv/ton glass, assuming 6 MM Btu/ton glass). As
s\.
with the Oberland Glas installation, accumulation of fine dust covered the catalyst shortly
after startup even though there was an electrostatic precipitator upstream of the SCR
catalyst and the SCR NOV reduction decreased. A pulsing blower and steel facings were
s\.
installed in front of the catalyst to minimize dust accumulation. No information is given as
to how successful this was. The dust accumulation is likely to make the application of
SCR to glass furnaces doubtful, although Lurgi (Frankfurt, Germany) reports the
development of a soot blower to remove dust from the SCR catalyst surface.111 A unit
has been installed and tested on a Schott Glaswerk specialty glass furnace in Mainz,
Germany. NOV emissions were reduced by 70 percent. The flue gas flow rate is 29,400
A
scfm and the SCR unit operates at 300 to 400 °C (570 to 750 °F).111
5.4.2 Selective Noncatalvtic Reduction (SNCR)
Selective noncatalytic reduction is the reaction of ammonia or urea with NO, via the
same type of reactions as shown in Section 5.4.1 for SCR, without the use of a catalyst.
These processes do not reduce NC^- In principle, any of a number of nitrogen compounds
can be used to reduce NO to N2 and h^O by similar reactions. These compounds include
cyanuric acid, pyridine, ammonium acetate, and others. However, for reasons of cost,
safety, simplicity, and byproduct formation, ammonia and urea have found the most
widespread application.
Because no catalyst is used to increase the reaction rate, SNCR is carried out at
high temperatures just downstream of the flame. The homogeneous gas phase reaction
of ammonia with NO must take place in a fairly narrow temperature range, roughly 870 to
1090 °C (1600 to 2000 °F). At higher temperatures, the rate of a competing reaction for
the direct oxidation of ammonia, which actually forms NO (2NHg + 5/202 -* 2NO +
Sh^O) becomes significant. At lower temperatures, the rates of the NO reduction
reactions become too slow and unreacted ammonia is present in the flue gas. One
modification of this process incorporates the addition of hydrogen and other
119
compounds ' ' Mo lower (but not widen) the temperature from 870 to 1,090 °C (1,600 to
2,000 °F) to about 705 to 925 °C (1,300 to 1,700 OF) 11 3,114 NH3/NO mo| ratios are
varied —Reactions (5-2) and (5-3) above suggest at 1.5/1 to 2/1 molar ratio, which is
5-53
-------
typical of industrial practice.1 ^5>11 6 There are two commercial SNCR processes —the
Exxon Thermal DeNOv which uses ammonia and the Nalco Fuel Tech NOVOUT" which
A X
uses a urea-based reagant. In addition, PPG has patented its own SNCR design.117
Figure 5-27 shows a schematic of the PPG system, which is similar, at least in
principle, to the other SNCR systems. Ammonia is injected from nozzles into the flowing
gas, as shown in Figure 5-27 for a utility boiler. Because the reaction takes place in the
gas phase, SNCR is particularly suitable to gases from glass furnaces containing particles
that would foul the catalyst in an SCR system.
The Exxon SNCR process has been installed on over 130 combustion processes
worldwide between 1975 and 1993,118-120 jnc|ucjjng at |east four f|at glass furnaces,
one German recuperative glass furnace, and three direct-fired furnaces with Ho addition
capability. Although originally designed to use anhydrous ammonia, concerns about safety
and the need for high-pressure storage has led to the development of a process using
aqueous ammonia. However, this aqueous ammonia process apparently has not been
used in glass furnaces.
An SNCR process using aqueous urea [CO(NH2)2l rather than ammonia was
®
developed by EPRI and is now marketed by Nalco Fuel Tech under the name NOVOUT .
A
The exact reaction mechanism is not understood, but it probably involves the
decomposition of urea, with the subsequent reaction of NH2 groups with NO :
NH2 + NO^N2 + H20 .
Urea is somewhat safer to handle than anhydrous ammonia, though aqueous ammonia can
now be used in the Exxon process. As a more recently developed process, there are
somewhat fewer NOXOUT installations; Nalco claims 70 commercially contracted
1 *)"J i p/i
systems worldwide. None of these are reported as being installed on container,
flat, or pressed/blown glass furnaces. As with ammonia injection, urea injection must
occur in a well-defined temperature window, which is approximately the same as for
ammonia injection, 870 to 1,090 °C (1,600 to 2,000 °F).125 Others state that wider
temperature ranges can be used, presumably due to proprietary additives developed by
Nalco. ' ] NOV reductions are also comparable to Thermal DeNOv , i.e., around
X X
30 to 60 percent with ammonia slip of 5 to 20 ppm, "''29 though reductions of up to
80 percent from uncontrolled levels are reported. One recent modification of the urea-
based SNCR system is the addition of methanol injection downstream of the urea injection
point to improve overall NOX removal. Nalco also recently introduced NOXOUT PLUS ,
5-54
-------
Crown
Burner
Regenerator
Checkers
Ammonia
Injection
GrlcH
^_-=__- Glass Melt I_~ _ __"_ .
Ammonia
Injection
Grid 2
Figure 5-27. PPG SNCR process.
117
5-55
-------
which is said to broaden the operating temperature window and to reduce ammonia slip
and CO and NC^ formation. Nalco is also developing a combined SNCR/SCR system
which reduces SCR costs by decreasing the size of the catalyst and is expected to
achieve NOX reductions similar to SCR alone.
Table 5-8 shows the current SNCR installations on glass melting furnaces for
container, flat, and pressed/blown.' 19-120 ACtua| data on SNCR operating experience for
glass furnaces is limited to the PPG (Fresno, CA), LOF (Lathrop, CA), and AFG {Victorville,
CA) flat glass plants. As expected, NOX reduction is highly dependent on furnace-specific
®
factors. The PPG plant installed an Exxon De-NOx process in 1981 that was later
modified to one of their own design. Though this process uses ammonia injection,
some details are proprietary. NOX reductions for two tests are from 23.6 and 22.3 to
11.7 and 9.2 Ib NOx/ton glass, respectively. This corresponds to actual reductions of 50
and 59 percent for these two tests.
The LOF plant (Lathrop, CA) installed an SNCR system in 1987. The design
1^1
emission reduction was 56 percent. However, LOF intentionally operates the system
to achieve a NOX reduction of 31 percent to achieve emission reduction credits. The
controlled NOV emissions are 9.7 Ib NOY/ton (1991 test) and 12.4 Ib NOv/ton (1992
X. s\ A
test).131
®
The AFG plant installed an Exxon De-NO system in 1987. Two series of tests
s\
have since been made with and without ammonia injection, corresponding to controlled
and uncontrolled NOX emissions. In addition, the ammonia injection rate was also varied.
From uncontrolled levels of 13.1 to 14.6 Ib NOY/ton, NOV emissions were reduced to 8.4
xv A
to 10.7 Ib NOx/ton, respectively, corresponding to 27 to 36 percent reduction. Variation
of the ammonia injection rate, within the range tested, had no major effect on NOX
emissions, as measured in Ib NOx/ton glass. However, ammonia slip increased
monotonically with increasing injection rate, as expected, and NOX concentration (ppm)
generally decreased with ammonia injection rate.
5.5 SUMMARY
Table 5-9 summarizes the reported controlled NOV emission levels for each of the
s\
technologies discussed in Chapter 5.NOX emissions are reported in units of both Ib
N0v/ton glass and Ib NOV/MM Btu. These are related by the heat input, in MM Btu/ton
J\ *\
glass, which is roughly heat input of 6 MM Btu/ton (from Chapter 3), but varies with the
5-56
-------
(O
HI
0
z
cc
u.
O
z
L^
UJ
2
1
C/5
<
^j
o
o
0)
z
o
h-
^
^J
_1
w
z
cc
0
z
H
z
Ul
cc
cc
o
CO
1
If)
TABLE
o*!
Z o
S '
O 0)
•tl ""
£
^"^
CO -D
C 03
.2 "o
w ^
W c
'E °
5J °
0)
X
O -o
Z —
o
c
o
o
c
3
-o
•a ^
• 1
O3
H-
•> 03
co 1
CD O CD ^_
CD LO CO ^
«- r-- r-
!
1
g?
O
LO
A
CN
co oo co ,-„ o
O LO LO .^
LO [^ CN O) 1—
O)
co r
f CO CO
TO "7^ ^ ^
O |g O d
(1
t — «-
;
<- r-
CO
— >•
o 2- CD c
C < £ < D) «< CD
So 50 .§0 i
l^ CD CD
*. _. g 0
CO CO CO
CD CO (D |_
2 t) C3 u- to O
Q- = LL < = X
13 I w
CO
CD
,-
Switzerland
O)
I
CD
O
X
x
LU
C
CD
O>
C
'^
CD
-
CD
S
Q.
O
QL
CD
O
•^
E
CD
CO
CO
_co
+-1
.CD
M-
T3
o
Is
0)
o
c
CO
CO
CD
Q_
CL
CD
CO
CD
CO
C
o
Q.
CO
CD
't
'—
C
o
CU
CO
.c
T3
CD
t
O
0.
CD
CO
CO
CD
*->
. O3
|?
>• c
CO CD
) O
xcn
O O3
Q 0
— CD
™ CO
C CD
QJ CU
£ >
1- <
-Q
CO
CD
CO
c
o
Q.
CO
CD
*j-
^~
«—
C
o
o
CD
CO
_c
•D
CD
t
O
Q.
CD
CO
to
CD
4->
co
O3
T5
C
CD
CD
• co •
22g
H—
CD
DC < DC
0 TJ ~
§ T3
§ £
?!
CD S
CU 4-<
> 0
< Z
M- D)
5-57
-------
CO
LU
a
o
o
z
i
0
LU
H
co
D
O
cc
5
oc
o
u.
.
D)
_O
"o
O
^
CD
CN^OO C75 COC7)'~^'^t -—^n--^..^^!^
CO^r IT5CO CD COCX3CD O5CT) tJlJiJiJiJ^J
CO
*
r— co co c*j i o^ oj ^^ ^J co o*^ ^^ ^^ ^5 r**- to * — *""
*^ co co co ro __ ^J ^J r^1 f^" ii^ LO c^ co ro co
r~ CO CO ^" i C75 f>j CO C^ CO 0^ ^^ CO O} f** ^^ r"~ *""
^- co CTJ co n c*j ^~ r*^ ^" ii^ ufi c*^ co co co
s— ^^
»-^ O^t OO *— tD^J3^^3XlJD
^•OUJCNJ CN rj- CO OLOCO^ftO*—
O ^— O O O OtXQC — — OQCcSr— i—t— <— CN
22 2
u T- H- "3.-^ "* E c c c
CNCOCNJO *— COCCDCCOCO CNlDC*-a)OCOCT3. w1^ .^KMZ-g o0- ^
_Q ^ X -i ^ D ^ O ^ ^~
C ^* ^^ o ^> ^_^ LJJ ^j C^ (/D
o 2 o
00- 0.
c;
a;
D)
TO
a
D)
c
c
c
c
footnotes o
-------
•o
05
3
C
*^3
0
u
9
in
UJ
_j
CO
<
= Not reported.
tr
z
co
(—
4^
4-<
'•6 S
co E
o o
w -*-1
E t3
O CO
=° '5
cr
"o co
c Z-
CO t
0
co
"co
>
CO
CO
to
3
"co
+-'
CO
0
^
"oT
>
CO
CO
00
in
Calculated assuming 6 MM Btu/ton glass (also see Chapter 3
.a
o
co ;=
/- CO
.1— . f.
4^ CO
•K 4J 0
TO ~n
to .E " to
<2 *^ c c
CO .. E Q
-5, >• o .y
c « - 2
Ills
X;F co o
0 o 3 c
Z '5 ° 8
•*-» m
£ to 8 x
_ »_
3 3 m " O
^ +i <° eo 4:
^5 I*
f"° * I fi
c *"
o x « ° •=
£- ££ i
« ^ co ~ 5
t u - o
8 « '- 3 "W
O — - .£ Jl— ' w
5 .^ H 3 §
CO —
•^ ™ • to co
4^ to ^5 "F >
g 8 8 - co
o 3 ra ^:
References on this technology give only N0y concentrations,
Assuming (1) the fuel to be pure methane, (2) 10 percent ex
flue gas flow rate is about 68 scfm per ton/day of glass prod
and Doyle,85 but could vary widely for other furnaces. Usim
600 to 750 ppm given by Barklage-Hilgefort and Sieger.36 T
O
to
CO
3
CO
>
cu
to
cu
.c
4-*
D)
C
'£>
CO
3
0
co
o
not accounted for in
to
to
lc
4-J
•D
C
CO
4-1
3
O
C
>
cp
cu
CO
"co
S
4-1
H-
o
4-J
(U
o
w.
This furnace has electric boost corresponding to about 1 4 pe
TJ
of NOy emissions.
X CO
0 c
Z '5
4^
CO C
N O
•— (J
E
"c S
•^
•o >
c ju
CO
>. 13
& *
'> 0
d to improve product
Based on an uncontr
6 percent reduction.
CO . 00
0) « C
'to co ro
CO — O
•o c °
0) .2 T3
co to c
T tfl O
4? "E Q-
co t to
cu cu co
•*- t
1^3
°ll
•D •g 0
i 8 5
>§l
c «
o j^ to
0 CO .-
a> o E
*-* CO
TO C T3
CO O CO
£ to =
•.— o
This is a new installation that includes the furnace as well as
Because it is not an add-on control, there is no direct compar
glass furnaces of 10 Ib/ton (Table 4-2, Chapter 4), the contn
CD
g
_O
to
JC
4-1
to
3
£
to
c
O
'to
to
E
0)
X
o
cu
£
4-1
w
0)
o
CO
.E
4_i
c
CO
4^
c
o
o
4-*
CO
3
o
* ^
4^ CO
| 0
D) to
•= to
•C CO
cu "I
£^
CO g
o >-
>4- CO
CO j:
CO 0)
.c j:
4-1
?-
CO "O
±1 3
CO O
£ 5
2«
i 3
*; *-
3 C
+^ CO
CO o
C CO
— Q.
>-o
0) 00
CO 4->
c co
« -o
co co
£ 3
to
2= co
CO CO
-c E
3 <«
o 5
a 5
»•=
S .2
* -E
2 5
4-1 «,"
cu ?>
31
«4-
X
o
^
c;
c
o
o
3
TD
CO
4_«
c
CO
u
orrespond to a 1 2 pe
o
•o
0
g
to
!c
E
co
>
'o>
to
3
Q.
c
>-
O)
CO
E
CO
_c
CO
to
No data other than calculations showing a 1 2 percent decrea
O)
05
CO
o
c
CO
Data for the Nienburger preheater is not available. See Refer
.E
•D
CO
o
CO
Q.
at ">
il to
CO
•!2 "Si
co >-
o cu
S -e
§ ™
3 E
O
co u
£ co
o€
co —
5 >
5§
•^ TD
H- CU
o to
4^ =
E to
CO —
0 4^
1°
"• o
0 -°
*~ o
^ '£
o
• CO
•? GJ
4-1 —
3 CO
0. 3
E cr
— eu
c5 CT
? c
CO 'co
£ -°
UJ
0^
4. <"
E = .:
co to v
t\ U
Z -vco
&il
co 2 ^
£ £ £
0 C ^
NOX reductions from electric boost are directly proportional t
by electricity this would correspond to a 10 percent reductioi
industry, typical boosts are 5 to 1 5 percent of the energy to
—
E
O
i; E
5 B
_ +J
*• ox
CD ^
o _
£ J3
^«
8"
CO O
0 ~
CO T3
•D CO
to
^ CO
ill of 220 ton/day) an
NOX emissions decre
5 -n
« =
CO CO
4-1
JS >•
CO
c -o
o >
Ss
^§
CD CM
CO O
>~ 4-*
1'
> CO
4J CO
to co
o co
8 "
•" E
o
C 3
£ 0.
4^
'5 +^
^ to
s|
o -°
'to >
ll
£g
^t m
m,<
CO
!«
CC 0
^^
T:^
eu ^
So
S?
% •=
> 4^
fi §
0 _
0 >
a co
0 |
Is
^g
•s^
H—
to o
CO —
II
O CO
5 £
H -S
—
CO
CO
r-
co
o
E
as reported in Refere
~
Q.
Q.
in
O5
"o
r
CM
O
Z
to
co
E
"co
>
co
X
O
Z
+-J
CO
4-'
3
O
E
CO
T3
E
CO
«4-
o
to
0
O
in
CT5
CM
"o
CO
4-<
CO
O
«*-
to
CO
D)
CO
H^
CO
0
H-
T3
CO
4->
JO
o
CO
O
.^
JO
co
>
_co
•D
JU
"o
+J
0
o
^
o
•D
CD
"o
w
4-1
E
Only the percent NOX reduction is given, not the actual uncc
—
5-59
-------
o
0) o>
1-g
.E TO
•go.
O to
£€
So
o 4=
o
£ E
•C Q.
*- a
cO
TO m
to1"
to o
_TO -M
O) LO
C
o
c to
— TO
TO O
'E 4)
O C
E °
E =
TO
to <6
o £
5 8
4-> C
TO
to DC
- s
S'c
5 >
o
o
.— O k-
O TO Q.
0)
_2
TO
O
4-1
O
3
T3
0)
O)
^
O
a.
o
o
TJ
a>
3
C
O)
in
UJ
_i
CO
ro
CM
CM
T3 H-CM
C °r~
" C W
CD O <«
. -73 TO
co o "g)
CM 3
..TIC
Sfi^
^oxox
TO 2 z
o as
g S TO
£ o °
LL <0 O
- TO ~
O n ~°
n 2 «J
•Hi
O =
O
£2 g
r s ^
TO O TO
" • C
S i.l
^
O T3
Z 0
*• to
(0 CM tO
W TJ
(0 c
'
D.
to
0)
0 0
'
CO X—
TO O C
-D Z £
O.
o r.
«l
H- O
4) £
CC to
5-60
-------
thermal efficiency of the furnace and would be lower for high proportions of cullet. It is
important to look at both measures of NOX emissions —Ib/ton glass and Ib/MM Btu.
Furnace energy input (MM Btu/ton glass) as well as NOV emissions generally increase with
s\
furnace age because the furnace refractory insulation gradually deteriorates. Except for
oxy-firing, the two measures of NOX controlled emissions in Table 5-9 are directly
proportional assuming 6 MM Btu/ton glass is accurate. For oxy-firing, however, much less
energy is needed because nitrogen is not present in the combustion air and energy is not
used (and then lost up the stack) to heat it in the furnace. For oxy-firing, a value of 3.4
MM Btu/ton is reported,1^7 though this varies with different furnaces (which have
different levels of air infiltration) and oxygen sources (which contain different amounts of
nitrogen).
Combustion modifications in Table 5-9 include modified burners and oxy-firing. A
MOV reduction of 66 percent is reported for one low NOV burner. This is the only test data
X X
available, though the NO reduction is somewhat higher than that reported in other
applications.^ Oxy-firing results in NOX reductions of 84 to 90 percent (measured in Ib
N0v/ton glass) and 82 to 88 percent (measured as Ib NOV/MM Btu). These data are from
s\ ){
large-scale container glass melting furnaces.
Process modifications include a modified furnace, cullet/batch preheat, and electric
boost. The modified furnace achieves low levels of NOX, but it is not an add-on control.
Rather, it incorporates a number of heat recovery and design features to achieve NOV
.X
reduction and higher productivity. Insufficient data are available to evaluate cullet/batch
preheat as an NOV control technique. The widely varying values in Table 5-9 are due to
s\
widely varying cullet/batch ratios, proportion of the cullet that is preheated, proportion of
the flue gas used in the preheater, and other variables. In the references cited, there is
insufficient information to compare directly each of the three processes.
Electric boost simply substitutes one form of energy for another. A general
assumption is that NOX emissions from the furnace are lowered in direct proportion to the
proportion of the furnace energy that is input as electricity. A thermal input of 6
MM Btu/ton corresponds roughly to an electrical input of 880 kWh/ton. This value is for a
1 *2 Q
batch containing 10 percent cullet ; of course, the higher the cullet content, the lower
the melting energy needed. [880 kWh = 3 MM Btu, meaning that electrical melting (or
boosting) is about twice as energy efficient as thermal melting.] Dividing these two
values, 147 kWh of electrical energy replaces 1 MM Btu of thermal input. One MM Btu of
5-61
-------
thermal input would, in turn, correspond to one-sixth or 17 percent, of the thermal input
into the furnace, corresponding to a NOX reduction of 17 percent, all else being equal.
»
Postcombustion modifications in Table 5-9 include SCR and SNCR. SCR reduces
NOX emissions in glass furnaces by 70 to 79 percent, SNCR by 27 to 50 percent.
Based on the information in Table 5-9, NOX percent reductions are shown in Table
5-10 for each generic technology. NOX reductions based on these uncontrolled levels are
used in calculating cost effectiveness in Chapter 6. Table 5-11 summarizes the current
status of the technologies shown in Tables 5-9 and 5-10. For flat glass, only SNCR and
electric boost have been demonstrated, though electric boost is no longer used. Oxy-
firing may be applicable for flat glass, but is not yet demonstrated. For container glass,
only SNCR is not demonstrated, though it may be feasible. Cullet preheat has been
demonstrated, but now is not used. For pressed/blown glass furnaces, modified burners,
oxy-firing, and electric boost are the only technologies that have been demonstrated.
5-62
-------
TABLE 5.10. CONTROLLED NOX PERCENT REDUCTION
USED FOR CALCULATING COST EFFECTIVENESS
Technology ^O Reduction (%)
Combustion modifications . 40
Modified
Oxy-firing 85
Process modifications
Modified furnace 75°
Gullet preheat 25
Electric boost 10
Postcombustion modifications
SCR 75
SNCR 40
s See Table 5-9 for a summary of reported NOX reductions reported for these technologies.
3 Based on uncontrolled emissions of 6.0 Ib N0x/ton [calculated assuming 10 Ib/ton for the 20
percent of the batch that is virgin44'°5.139 and 5 Ib/ton for 80 percent of the batch that is
cullet: (10 x 0.2) + (5 x 0.8) = 6 Ib/ton] and controlled emissions of 1.4 Ib/ton as reported
in Reference 69. The resulting value of 77 percent NO reduction is rounded to 75 percent.
5-63
-------
TABLE 5.11. STATUS OF NOX CONTROL TECHNOLOGIES FOR
VARIOUS QLASS FURNACES
Furnace Type
NOX Control Technology
Flat
Container
Pressed/blown
Combustion
modifications
Modified burners
Oxy-firing
Process
modifications
Modified furnace
Gullet preheat
Electric boost
not demonstrated
not demonstrated,
but possibly
feasible39
not demonstrated
not demonstrated
demonstrated, but
not now used95
demonstrated
132
demonstrated69"71
demonstrated, but
not now used76'77'83
demonstrated93'96
demonstrated
140
demonstrated3®'54-62 demonstrated
not demonstrated
not demonstrated
demonstrated
Postcombustion
modifications
SCR
SNCR
not demonstrated
demonstrated 3 •
135,136
demonstrated110'111
not demonstrated,
but possibly
feasible
not demonstrated
not demonstrated
5-64
-------
5.6 REFERENCES
1. Moilanen, G.L., and B. Van Kalsbeek. NOX Control Options for Glass Furnaces.
Ceram. Eng. Sci. Proc. 1 2(3-4):632-649. 1991.
2. Kokkinos, A., J.E. Cichanowicz, R.E. Hall, and C.B. Sedman. Stationary
Combustion NOX Control: A Summary of the 1991 Symposium. J. Air Waste
Manage. Assoc. 1252. 1991.
3. Ref. 1, p. 640.
4. Cooper, D.C. Air Pollution Control: A Design Approach. PWS Engineering. 1986.
5. Sommerlad, R.E. Overview of NOX Control Technologies. NOX Control V Seminar.
Long Beach, CA. Council of Industrial Boiler Owners. Burke, VA. February 10-11,
1992. Pp. 19-29.
6. Joseph, G.T., and D.S. Beachler. Student Manual, APTI Course 415, Control of
Gaseous Emissions. U.S. Environmental Protection Agency, Air Pollution Training
Institute. EPA 450/2-81-006. December 1 981. p. 7-12.
7. Siddiqi, A.A., and J.W. Tenini. NOX Controls in Review. Hydrocarbon Processing.
115-124. October 1981.
8. Campbell, L.M., O.K. Stone, and G.S. Shareef. Sourcebook: NOX Control
Technology Data. U.S. Environmental Protecton Agency, Air and Energy
Engineering Research Laboratory. Contract No. 68-02-4286, EPA-600/2-91-029.
July 1991.
9. Clarke, M.J. Minimizing NOV. Waste Age. pp. 132-138. November 1989.
s\
10. Mclnnes, R., and M.B. von Wormer. Cleaning Up NOX Emissions. Chem. Eng.
p. 131. September 1990.
11. Pont, R.S. Pollution Control by Burner Design and Operation. Glass Tech.
32(3):58. 1991.
12. Burd, V. Squeezing Clean Energy. Chem. Eng. pp. 145-159. March 1992.
13. Ref. 4, p. 465.
14. Ref. 6, p. 7-7.
15. Abbasi, H.A., and D.K. Fleming. Combustion Modifications for Control of NO
Emissions from Glass Melting Furnaces. Ceram. Eng. Sci. Proc. 9(3-4): 168-177.
1988.
1 6. Neff, G.C. Reduction of NOX Emissions by Burner Application and Operational
Techniques. Glass Tech. 31(2):37-41. April 1990.
5-65
-------
17. Abbasi, H.A., and O.K. Fleming. Development of NOX Control Methods for Glass
Melting Furnaces. Gas Research Institute. Final Report No. GRI-87/0202. August
1987.
18. Ref. 17, p. 47.
19. Letter and attachments from Fries, R.R., Glass Packaging Institute, Washington,
DC, to Neuffer, W.J., EPA/OAQPS, Research Triangle Park, NC. April 8, 1993.
Comments on draft ACT.
20. Ref. 15, p. 171.
21. Ref. 11, p. 61.
22. Ref. 17, p. 57.
23. Ref. 17, p. 71.
24. Ref. 17, p. 55.
25. Neff, G.C., M.L. Joshi, and M.E. Tester. Development of a Low NOX Method of
Gas Firing. Ceram. Eng. Sci. Proc. 1 2(3-4):650-660. 1991.
26. Ref. 4, pp. 466-467.
27. Ref. 8, pp. 25-29.
28. Ref. 7, p. 117.
29. Ref. 4, p. 467.
30. Ref. 1, p. 642.
31. Letter and attachments from Newell, P., Guardian Industries, Kingsburg, CA to
Neuffer, W.J., EPA/OAQPS, Research Triangle Park, NC. August 16, 1993.
Information on oil firing and low NOV burners.
s{
32. Letter and attachments from Aker, J.E., Teichmann Sorg Group, Ltd., Pittsburgh,
PA, to Neuffer, W.J., EPA/OAQPS, Research Triangle Park, NC. April 16, 1993.
Comments on draft ACT.
33. Letter from Moore, R.H., Teichmann, Sorg, Inc., McMurray, PA, to Spivey, J.J.,
Research Triangle Institute, Research Triangle Park, NC. Comments on June 11,
1993 meeting minutes.
34. Sieger, W. Development of Reduced NOX Burners. Glass Tech. 31 (1 ):6. 1990.
35. Glass. Reduced NOX with Air Staging System, p. 217. ISSN 001 7-0984. June
1991.
5-66
-------
36. Barklage-Hilgefort, H., and W. Sieger. Primary Measures for the MOX Reduction on
Glass Melting Furnaces. Glastech. 62(5):151. 1989.
37. Scully, P.P. Green Glass Now? Glass International. June 1990.
38. Moore, R.D., and J,T. Brown. Conversion of a Large Container Furnace from
Regenerative Firing to Direct Oxy Fuel Combustion. 1991 Glass Problems
Conference. American Ceramic Society. Westerville, OH. 1992.
39. Summary of telephone conversation, J. Brown, Corning, Inc., Corning, NY, to J.J.
Spivey, Research Triangle Institute, Research Triangle Park, NC, July 22 and
August 17, 1993. Feasibility of oxy-firing for glass furnaces.
40. Tuson, G.B., H. Kobayashi, and E.J. Lauwers, Industrial Experience with Oxy-Fuel
Fired Glass Melters, presented at Classman Europe 93, Lyon, France, April 28,
1993, ®Praxair Inc., Tarrytown, NY.
41. Slavejkov, A.G., P.B. Eleazar, L.G. Mayotte, and M.L. Joshi. Advanced Oxy-Fue!
Burner System for Glass Melting: A Performance Report. Presented at the 90th
Annual Meeting and Convention. Canadian Ceramic Society. Toronto, Canada.
February 16-18, 1992.
42. Baukal, C.E., P.B. Eleazer, and L.K. Farmer. Basis for Enhancing Combustion by
Oxygen Enrichment. Ind. Heating, p. 23. February 1992.
43. Baukal, C.E., and A,I. Dalton. NOX Reduction with Oxygen-Fuel Combustion. In:
1990 Am. Flame Res. Conf. Symposium. October 1990. p. 4.
44. Ref. 16, p. 39.
45. Ref. 35, p. 217.
46. Ref. 43, pp. 1-10.
47. Ref. 42, p. 22.
48. Gupta, P. J. Non-Crystalline Solids. 38(39):761-766. 1980.
49. Joshi, S.V. Oxygen Enriched Air/Natural Gas Burner System Development. Gas
Research Institute. Chicago, IL. 1985.
50. Kobayashi, H., G.B. Tuson, and E.J. Lauwers, NO Emissions from Oxy-Fuel Fired
Glass Melting Furnaces. Paper presented at the European Society of Glass Sciences
and Technology Conference on Fundamentals of the Glass Manufacturing Process.
Sheffield, England. September 9-11, 1991. Union Carbide. Tarrytown, NY. p. 4.
51. Ref. 50, p. 3.
52. Ref. 50, p. 12.
5-67
-------
53. Ref. 50, p. 8,
54. Kobayashi, H., G.B. Tuson, and E.J. Lauwer.s, NOX Emissions from Oxy-Fuel Fired
Glass Melting Furnaces. Paper presented at the European Society of Glass Sciences
and Technology Conference on Fundamentals of the Glass Manufacturing Process.
Sheffield, England. September 9-11, 1991. Union Carbide. Tarrytown, NY.
55. Westbrook, C.K. Computation of Adiabatic Flame Temperatures and Other
Thermodynamic Quantities. Proc. Ind. Comb. Tech. Symp. M.A. Lukasiewicz (ed.).
Chicago, IL. 1986. Pp. 143-150.
56. Kobayashi, H. Oxygen Enriched Combustion System Performance Study. Prepared
for U.S. Department of Energy. Idaho Operations Office. Report DOE/ID/1 2597.
March 1987.
57. Baukal, C.E., and A.I. Dalton. Nitric Oxide Measurements in Oxygen-Enriched Air-
Natural Gas Combustion Systems. In: Proc. Fossil Fuel Combustion Symp.
Petroleum Div. of ASME, Warrendale, PA. New Orleans, PD - Vol. 30, January
1990. Pp. 75-79.
58. Ryder, R.J. Use of Electric Boost to Reduce Glass Furnace Emissions. Am. Ceram.
Soc. Bull. 57(11):1025. November 1978.
59. Ref. 50, p. 9 and Figure 6.
60. American Ceramic Society Bulletin, April 1990. Cleanfire LoNOx Burner. Air
Products Bulletin 337-9104. Allentown, PA. 1991.
61. Air Products. Air Products and Combustion Tec, Providing Environmental Solutions
for the Glass Industry. Air Products Bulletin 336-9102. Allentown, PA. 1991.
62. Ref. 54, pp. 11-12.
63. Brown, J.T. Development —History and Benefits of Oxygen-Fuel Combustion for
Glass Furnaces. Corning Glass. Corning, NY. Presented at Latin American
Technical Symposium on Glass Manufacture. Sao Paulo, Brazil. November 18,
1991.
64. Brown, J.T. 100% Oxygen Fuel Combustion for Glass Furnaces. Ceram. Eng. Sci.
Proc. 12(3-4):594-609. 1991.
65. Shelley, S. Chem. Eng. p. 67. December 1992.
66. Ref. 38, p. 5-6.
67. Ref. 64, p. 598.
68. Ref. 1, pp. 636-637.
5-68
-------
69. Moore, R.H. LoNOx Glass Melting Furnace. Ceram. Eng. Sci. Proc. 11(1-2):89-
101. 1990.
70. Moore, R.H. LoNOx Melter Shows Promise. Glass Indust. p. 14. March 1990.
TM
71. American Glass Review. Satisfied Customer Orders Second LoNOx Melter. pp. 8-
9. October 1991.
72. Ref. 69, p. 100.
73. Ref. 69, p. 90.
74. Ref. 69, p. 94.
75. Ref. 69, p. 101.
76. Cole W.E., F. Becker, L. Donaldson, S. Panahe. Operation of a Cullet Preheating
System. Ceram. Eng. Sci. Proc. 11(1-2):59. 1990.
77. De Saro, R., L.W. Donaldson, and C.W. Hibscher. Fluidized Bed Glass Batch
Preheater, Part II. Ceram. Eng. Sci. Proc. 8(3-4):175. 1987.
78. Ref. 76, p. 57.
79. Ref. 76, pp. 53-68.
80. Ref. 76, pp. 57-58.
81. Ref. 76, p. 60.
82. Ref. 77, pp. 171-180.
83. De Saro, R., and E. Doyle. Glass Batch Preheater Program. Gas Research Institute.
Chicago, IL. Final Report No. GRI-87/0366. September 1987.
84. Ref. 83, p. 100.
85. Ref. 83, pp. 87, 108.
86. Ref. 83, p. 108.
87. Ref. 83, p. 130.
88. Ref. 83, pp. 116-117.
89. Zippe, B.H. Reliable Cullet Preheater for Glass Furnaces. Ceram. Eng. Sci. Proc.
12(3-4):550-555. 1991.
5-69
-------
90. Letter from Zippe Industrieanlagen GmbH, Wertheim, Germany, to U.S.
Environmental Protection Agency/OAQPS, Research Triangle Park, MC. April 5,
1993.
91. Enniga, G., K. Dytrich, H. Barklage-Hilgefort, "Practical Experience with Raw
Materials Preheating on Glass Melting Furnaces," unpublished paper from
Neinburger Glas GmbH, Nienburger, Germany, undated.
92. Ref. 91, p. 9.
93. Ref. 1, p. 638.
94. Ref. 58, pp. 1024-1031.
95. Letter from Horbatch, W., Ford Motor, Dearborn, Ml, to Neuffer, W.J.,
EPA/OAQPS, Research Triangle Park, NC. July 27, 1993. Ford experience with
electric boost on float glass furnaces.
96. Rindone, G.E., J.R. Hellmann, and R.E. Tressler. An Assessment of Opportunities
for Gas-Fired Boosting of Glassmelting Processes. Gas Research Institute.
Chicago, IL. Topical Report No. GRI-89/0254. January 1990. p. 1.
97. Ref. 96, p. 7.
98. Ref. 58, p. 1024.
99. Ref. 58, pp. 1027-1028.
100. Ref. 17, p. 39.
101. Moore, R.D., and R.E. Davis. Electric Furnace Application for Container Glass.
Ceram. Eng. Sci. Proc. 8(3-4):1 88-1 99. 1987.
102. Rindone, G.E., J.R. Hellmann, and R.E. Tressler. An Assessment of Opportunities
for Gas-Fired Boosting of Glassmelting Processes. Gas Research Institute.
Chicago, IL. Topical Report No. GRI-89/0254. January 1990.
103. Ref. 101, p. 192.
104. Ref. 101, p. 191.
105. Letter and attachments from Wax, M.J., Institute of Clean Air Companies,
Washington, DC, to Neuffer, W.J., EPA/OAQPS, Research Triangle Park, NC, May
14, 1992. Response to Section 114 letter on glass manufacturing.
106. Ref. 7, p. 119.
5-70
-------
107. Joseph, G.T., and D.S. Beachler. Student Manual, APTI Course 415 Control of
Gaseous Emissions. U.S. Environmental Protection Agency, Air Pollution Training
Institute. EPA 450/2-81-006. December 1 981.
108. Krause, W. Glass Melting Strategies at Oberland Glas. Glass Intl. pp. 51-52.
June 1990.
109. Letter and attachments from Wax, M.J., Institute of Clean Air Companies,
Washington, DC, to Neuffer, W.J., EPA/OAQPS, Research Triangle Park, NC. April
8, 1993. Comments on draft ACT.
110. Grove, M., and W. Strum. NOX Abatement System: Using Molecular Sieve
Catalyst Modules for a Glass Melting Furnace. Ceram. Eng. Sci. Proc. 10(3-
4):325-337. 1989.
111. Chem. Eng. A DeNOx System That Handles Hot and Dusty Waste Streams, p. 21.
October 1992.
112. U.S. Patent 3,900,554. August 1 9, 1975.
113. Ref. 1, p. 643.
114. Ref. 107, p. 7-13.
115. Ref. 7, p. 120.
116. Ref. 8, p. 38.
117. Hughes, David E. Meeting Glass with Reduced NOX Emissions. U.S. Patent
4,328,020. May 1982.
118. Ref. 1, p. 644.
119. Haas, G.A. Selective Noncatalytic Reduction (SNCR): Experience with the Exxon
Thermal DeNOx Process. Presented at the NOX Control V Conference. Council of
Industrial Boiler Owners. Long Beach, CA. February 10-11, 1992.
120. Letter and attachments from Haas, G.A., Exxon, Florham Park, NJ to Neuffer, W.J.
EPA/OAQPS, Research Triangle Park, NC, April 8, 1993. Comments on draft ACT.
121. Ref. 8, p. 40.
122. Ref. 10, p. 154.
5-71
-------
123. Mincy, J.E. SNCR Technology: The Next Generation. Presented at NOX V
Conference. Long Beach, CA. Council of Industrial Boilers. Burke, VA. February
10-11, 1992. p. 2.
124. Letter and attachments from Pickens, R., Nalco Fuel Tech., Naperville, IL, to
Neuffer, W.J., EPA/OAQPS, Research Triangle Park, NC. March 26, 1993.
Comments on draft ACT.
125. Ref. 123, p. 3.
126. Coal and Synfuels Technology. Pasha Publications. Arlington, VA. February 24,
1992. p. 7.
127. Hofman, J.E., et al. NOX Control in a Brown Coal-Fired Utility Boiler. In: Proc.
1989 Joint Symposium on Stationary Combustion NO Control. San Francisco,
CA. March 6-9, 1989. v. 2. EPA-600/9-89-062b (NTIS P889-220537). June 4,
1989.
128. Ref. 10, p. 134.
129. Ref. 2, p. 1255.
130. Letter and attachments from Benney, J.C., Primary Glass Manufacturers Council,
Topeka, KS to Neuffer, W.J., EPA/OAQPS, Research Triangle Park, NC. April 22,
1993. Comments on draft ACT.
131. Letter and attachments from Keil, J.R., Libbey Owens Ford, Toledo, OH, to Neuffer,
W.J., EPA/OAQPS, Research Triangle Park, NC. June 29, 1993. Response to
Section 114 letter on glass manufacturing.
132. Slavejkov, A.G., C.E. Baukal, M.L. Joshi, and J.K. Nabors. Advanced Oxygen-
Natural Gas Burner for Glass Melting. 1992 Int. Gas Res. Conf., Orlando, FL. Gas
Research Institute. Chicago, IL. November 1 6-1 9, 1 992. p. 319.
133. Ref. 110, p. 330.
134. DeStefano, J.T. Postcombustion NOX Control Technology for Glass Furnaces,
Update. Presented at 45th Glass Problems Conference. Columbus, Ohio. American
Ceramic Society. November 1984. p. 243.
135. Letter and attachments from Osheka, J.W., PPG Industries, Inc., Pittsburgh, PA, to
Jordan, B.C., EPA/OAQPS, Research Triangle Park, NC. September 1, 1992.
Response to Section 114 letter on glass manufacturing.
136. Letter and attachments from Robinson, J.R., AFG Industries, Inc. Kingsport, TN, to
Jordan, B.C., EPA/OAQPS, Research Triangle Park, NC. August 20, 1992.
Response to Section 114 letter on glass manufacturing.
137. Ref. 38, p. 4.
5-72
-------
138. Ref. 101, p. 189.
139. Ref. 15, p. 168.
140. Summary of June 1 1, 1993 meeting, Moore, R.H., Teichmann Sorg, McMurray, PA,
Neuffer, W.J., EPA/OAQPS, Research Triangle Park, NC, Spivey, J.J., Research
Triangle Institute, Research Triangle Park, NC.
5-73
-------
-------
CHAPTER 6
COSTS OF NOX CONTROLS
6.1 INTRODUCTION
Capital and annual costs as well as cost effectiveness ($/ton NOX removed) are
presented for the following NOX control technologies described in Chapter 5:
Combustion modifications
• low NOX I
• oxy-firing
low NOV burners
yv
Process modifications
• cullet preheat
• electric boost
Postcombustion modifications
• selective catalytic reduction (SCR)
• selective noncatalytic reduction (SNCR)
Costs were not available from the vendor or from any installation of the modified furnace.
Thus, costs and cost effectiveness for this control technique are not presented.
The percent NOX reductions for each technology used in making the cost
effectiveness calculations are shown in Table 6-1. The corresponding annual NO
reductions (tons NOV removed/yr) are given for each individual technology in subsequent
J\.
sections.
Costs are developed for the three model plants (50, 250, and 750 tons glass/day)
shown in Table 6-2. These correspond, roughly, to plants in the pressed/blown, container,
and flat glass segments of the glass industry, respectively.
The capital and operating costs were developed using information available in the
literature and from Section 114 requests. In many cases, site-specific details were not
6-1
-------
TABLE 6-1. CONTROLLED IMOX EMISSION LEVELS
USED FOR CALCULATING COST EFFECTIVENESS
Technology
Combustion modifications
Low NOX burners
Oxy-firing
Process modifications
Gullet preheat
Electric Boost
Postcombustion modifications
SCR
SNCR
Controlled NO Emissions
(Ib NOX/ ton glass)
Pressed/Blown
13.2
3.3
16.5
19.8
5.5
13.2
Container
•
6.0
1.5
7.5
9.0
2.5
6.0
Flat
9.5
2.4
NF
14.2
3.9
9.5
NF - Not feasible
TABLE 6-2. MODEL GLASS MELTING FURNACES
Plant size
(tons/day)
50
250
750
Uncontrolled NOV emissions
(Ib N0x/ton glass)
22.0
10.0
15.8
(Ib NOX/MM Btu)a
3.67
1.67
2.63
Flue gas flow rate
(scfm)b
3,400
17,000
51,000
Flue gas NO
concentration^
(ppm)
2,700
1,220
1,930
(mg/m3)
3,610
1,640
2,590
aBased on a heat requirement of 6 MM Btu/ton glass (from Chapter 4).
^Based on 68 scfm per ton/day of glass. See Table 5-8, footnote b.
cThis value is calculated from uncontrolled emissions in column 2 and a value of
68 scfm/ton/day of glass.
6-2
-------
provided by the original references. Such details, including furnace age and outside air
infiltration, can greatly affect both NOX emissions and control costs.
Costs have been updated to January 1994 dollars using the equipment index
component of the Chemical Engineering Plant Cost Index (January 1994 = 397.5).
Capital costs are also scaled, as needed, using the following equation:
Cost for size 1
Cost for size 2
[01
(Q2
0.6
(6-1;
6.2 COMBUSTION MODIFICATIONS
6.2.1 Low NOX Burners
Capital and annual costs were obtained for low NOX burners from North American
*•)
Manufacturing on a glass furnace producing 32 tons/day of glass. This burner differs in
design from the Kortig burner described in Section 5.2.2 in the way the staged air is
introduced. This burner is substantially smaller than those used in larger glass furnaces.
Nevertheless, in the absence of other cost information, these costs are scaled using
Equation (6-1) and are shown in Table 6-3. Capital costs range from $265,000 to $1.34
million and annual costs from $123,000 to $621,000. For the purpose of cost
calculations, a reduction of 40 percent was used. This percent reduction is consistent
with low NOX burner performance in other applications.5 Table 6-3 shows that the cost
effectiveness ranges from $ 790 to $1,680 per ton of NOX removed.
6.2.2 Oxv-Firing
Capital and operating costs for oxy-firing were available for a 250 tons/day
regenerative furnace. Costs have been scaled to provide capital and operating costs for
the other two plant sizes using Equation (6-1). In Table 6-4, Q^ is 250 tons/day and Q.2
is either 50 or 750 tons/day. Table 6-4 shows that capital costs vary from $1.93 to
$9.819 million. Cost effectiveness ranges from $2,150 to $5,300 per ton of NOV
s\
reduced.
6.3 PROCESS MODIFICATIONS
6.3.1 Gullet Preheat
Costs were available for a Tecogen system on a 250 tons/day furnace.** NOY
A.
reduction and costs depend on the fraction of cutlet in the batch. Costs are given in
6-3
-------
TABLE 6-3. COSTS AND COST EFFECTIVENESS OF RETROFIT LOW NOX BURNERS
Plant size
(tons/day)
50
250
750
Capital cost
($103)a
265
695
1,340
Annualized cost
($103/yr)b
123
320
621
NO reduction
(ton NOx/yr)c
73
167
790
Cost effectiveness
($/ton NOX
removed)
1,680
1,920
790
aThese costs are scaled using Equation (6-1) from costs provided by Gilbert for a 32-ton/day
furnace.
It is assumed that there are no operating costs (also, no operating cost savings due to increased
efficiency, if any, of this burner) and that all annual costs (maintenance and indirect costs) are 6
percent of the capital cost and that capital recovery is 40.2 percent, based on 10 percent for the
3-year ("2-4 year") burner life. Annual costs are therefore calculated as 46.2 percent of the
capital cost.
°Based on 40 percent reduction, and 8,000 hr/yr operation, per Table 5-8.
TABLE 6-4. COSTS AND COST EFFECTIVENESS OF OXY-FIRING
Plant size
(tons/day)
50
250
750
Capital cost
($103/yr)
1,930b
5,070
9,810b
Annual cost
($103/yr)
706C
1,860
3,590C
NOx reduction
(ton NOv/yr)a
J\
160
359
1,670
Cost effectiveness
($/ton NOX
removed)
4,400
5,300
2,150
aSee Table 5-8. 85 percent NOX reduction is assumed.
AThese values are scaled from the capital cost of $5 x 103° for a 150-ton/day furnace as follows:
Capital cost =
where Q and Q are the plant sizes in tons/day.
c These values are scaled from "operating costs" of $22/ton for a 250-ton/day furnace as in
footnote a, assuming 333 day/yr (8,000 hr/yr) operation. These "operating costs" account for all
direct, indirect, and capital recovery costs.
6-4
-------
Table 6-5 for 25 percent cullet, more or less representative of container and pressed/blown
glass furnaces, respectively. Some container glass furnaces may operate on essentially
100 percent cullet, but this case is not considered here. Capital costs range from
$188,000 to $492,000. Cost effectiveness range from $ 890 to $1,040 per ton of NOV
A
removed.
TABLE 6-5. COSTS AND COST EFFECTIVENESS FOR CULLET PREHEAT
Plant size
(tons/day)
50
250
Capital cost
($103)a
188
492
Annual cost
($103/yr)b
42
110
NOX reduction
(tons N0x/yr)
46
104
Cost effectiveness
($/ton NOX
removed)
890
1,040
aCapital costs are available only for the Tecogen preheater. Costs given by Becker
have been scaled using Equation (6-1) from 250 tons/day to the 50-tons/day mode!
plant. Control costs are for preheaters using waste heat in the flue gas rather
than separately fired preheaters.
'•'Annual costs are calculated based on a capital recovery of 10 percent/10 yr
(16.275 percent of capital costs) plus 6 percent for maintenance and indirect
operating costs, i.e., annualized costs are 22.3 percent of capital costs and are
scaled using Equation (6-1) from those given for a 250-tons/day plant.^
6.3.2 Electric Boost
Electric boost costs are contained in Reference 10. Technical contraints limit
electric boost to between 5 and 20 percent of the total energy input into the furnace.
Electric boost is used only in the container glass industry. Costs and cost effectiveness
are presented in Table 6-6 for 10 percent electric boost. Because NOX reduction is directly
proportional to the percent of furnace energy supplied electrically [as discussed in Section
5.3.2, i.e., 10 percent electric boost decreases NOX emissions (ib NOx/ton glass) by 10
percent], the cost effectiveness ($/ton NOX removed) is independent of the percent electric
boost. Electric boost is not widely used in furnaces as small as 50 tons/day (possibly due
to electrode placement and cost) nor furnaces as large as 750 ton/day (no furnaces of this
size using electric boost are reported). As shown on Table 6-6, annual costs range from
$178,000 to $525,000. Cost effectiveness range from $2,600 to $9,900/ton. Because
NOX removal is directly proportional to electric boost, the cost effectiveness for any of the
three model plants is independent of the percent boost.
6-5
-------
6.4 POSTCOMBUSTION MODIFICATIONS
6.4.1 Selective Catalytic Reduction
SCR costs depend primarily on the flue gas flow rate (scfm) and NO concentration.
Assuming the SCR unit can be installed at a place in the process where the temperature is
between about 350 and 500 °C (660 and 930 °F), no reheat is needed. The primary
concern for SCR in glass furnaces is dust accumulation. The only cost available that
explicitly accounts for installation of equipment to minimize dust prevention in a glass
furnace is given as $1.9 million for a unit to treat 29,400 scfm. [Assuming 68 scfm per
*•
ton/day of glass, per footnote b of Table 5-8, this would correspond to a 432-tons/day
furnace.] The exact scope of this cost is not provided, but is assumed to include all capital
costs. These capital costs range from $528,000 (50 tons/day) to $2.69 million
(750 tons/day), although somewhat lower capital costs are also reported: from $406,000
(50 tons/day) to $1.38 million (750 tons/day).14 Annual costs are $6/ton glass for a 500-
1 R
tons/day SCR unit. Scaling this value using Equation (6-1), annual costs are shown in
Table 6-7. These costs range from $404,000 to $1.2 million per year. Cost effectiveness
ranges from $800 to $2,950 per ton of NOX removed.
TABLE 6-6. COSTS AND COST EFFECTIVENESS OF ELECTRIC BOOST
Plant size
(tons/day)
50
250
750
Annual cost
($103/yr)a
178
339
525
NOX reduction
(ton NOx/yr)
18
42
200
Cost effectiveness
($/ton NOX removed)
9,900
8,060
2,600
aFor electric boost, separate capital costs are not available. The incremental cost
of electric boost as $40/ton glass compared to $10/ton if gas is used. Approximate
confirmation of this is stated that the operating cost for all electric melters is twice that of a
regenerative natural gas melter. This is assumed to be applicable only to furnaces in the
range given by Reference 10, around 250 tons/day. For the 50- and 750-tons/day cases
above, this cost is scaled using Equation (6-1).
6-6
-------
TABLE 6-7. COSTS AND COST EFFECTIVENESS FOR SCR
Plant size
(tons/day)
50
250
750
Capital cost
($103)a
530
1,390
2,690
Annual cost
($103/yr)c
400
770
1,200
NOX reduction
(ton NOx/yr)b
140
310
1,490
Cost effectiveness
($/ton NOX
removed)
2,950
2,460
810
aCapital costs are scaled from a value of $1.9 million given in Reference 13 for a unit treating
29,000 scfm. Using a value of 68 scfm/ton/day of glass (see Table 5-8, footnote b), this
corresponds to a 432-ton/day furnace. This cost is scaled to the three furnaces shown above using
Equation (6-1). ICAC provided capital costs of $400,000, $720,000, and $1,360,000 for the
three plant sizes above.
14
'•'NO., reduction is taken as 75 percent, based on Table 5-8.
A
cAnnual cost are calculated as $6/ton glass for a 500-ton/day furnace.' ° This is scaled using
Equation (6-1) for the model plant sizes shown here.
6.4.2 Selective Noncatalvtic Reduction
Capital and annual costs were available for two flat glass furnaces that use
ammonia injected SNCR. The averages of these furnaces are 626 TPD, capital cost of $
1,400,000 and an annual cost of $ 589,000.16'17 Capital and annual costs were
obtained from Nalco for their urea based SNCR process for the three model sizes.1^
These costs are much higher than costs for the ammonia-based SNCR. Costs are available
for actual installations using SNCR ammonia and urea based in the ACT documents for
utility boilers and Industrial/Commercial/lnstitutional Boilers. A cost comparison showed
no major difference between the two systems. Thus,in this ACT document, no distinction
is made between costs for the two different SNCR systems. The costs for the ammonia
based SNCR system are assumed to be more accurate as they are based on actual
installations. As shown in Table 5-10, a control efficiency of 40 percent was used. As
shown in Table 6-8, capital costs ranged from $ 310,000 to $ 1,560,000. Cost
effectiveness ranged from $830 to $2,000/ton. Cost and emission data were obtained
from two flat glass installations. 9 Cost effectiveness for these two installations are
$990 and $1700/ton.
6-7
-------
TABLE 6-8. COSTS AND COST EFFECTIVENESS FOR SNCR
Plant size
(tons/day)
50
250
750
Capital cost
($103)
310
810
1,560
Annual cost
($103/yr)
130
340
660
NO reduction
(ton NOv/yr)
J\
*
70
170
790
Cost effectiveness
($/ton NOX
removed)
1,770
2,000
830
(990- 1700)a
a Two actual installations at 40 and 30 percent control, respectively.
6.5
SUMMARY
Table 6-9 summarizes the cost effectiveness of the control technologies considered
here. Cost effectiveness of low NOX burners, cullet preheat and SNCR are similar. Cost
effectiveness of oxy-firing is much higher but low NOV emissions can be achieved. SCR
X.
achieves similar NOX control levels as oxy-firing but cost effectiveness is much lower.
Cost effectiveness for electric boost is also high.
TABLE 6-9. SUMMARY OF COST EFFECTIVENESS FOR NOX CONTROL
TECHNOLOGIES FOR GLASS FURNACES
($/ton NO removed)
Plant size
(tons/day)
50
250
750
Low
NOX
burners
1,680
1,920
790a
Oxy-firing
4,400
5,300
2,150a
Cullet
preheat
890a
1,040
N/F
Electric
boost
9,900
8,060
2,600
SCR
2,950a
2,460
800a
SNCR
1,770a
2,000a
830
(990 - 1700)b
N/F Not feasible
a Not demonstrated
k Two actual installations at 40 and 30 percent control, respectively.
6-8
-------
6.6 REFERENCES
1. Abbasi, H.A., and O.K. Fleming. Combustion Modifications for Control of NOX
Emissions from Glass Melting Furnaces. Ceram. Eng. Sci. Proc. 9(3-4):168.
1988.
2. Letter and attachments from Gilbert, F.C., North American Manufacturing
Company, Cleveland, OH, to Jordan, B.C., EPA/OAQPS, Research Triangle Park,
NC. November 11, 1992. Response to Section 114 letter on glass manufacturing.
3. Barklage-Hilgefort, H., and W. Sieger. Primary Measures for the NOX Control
Methods for Glass Melting Furnaces. Glastech. 62(5):151. 1989.
4. Sieger, W. Development of Reduced NOV Burners. Glass Tech. 31(1):6. 1990.
J\
5. Sommerlad, R.E. Overview of NOX Control Technologies. NOX Control V Seminar.
Long Beach, CA. Council of Industrial Boiler Owners. Burke, VA. February 10-11,
1992. Pp. 19-29.
6. Slavejkov, A.G., P.B. Eleazar, L.G. Mayotte, and M.L. Joshi. Advanced Oxy-Fuel
Burner System for Glass Melting: A Performance Report. Presented at the 90th
Annual Meeting and Convention. Canadian Ceramic Society. Toronto, Canada.
February 16-18, 1992.
7. Ref. 6, p. 3.
8. Fax message from Becker, F., Tecogen, Inc., Waltham, MA, to Spivey, J.J.,
Research Triangle Institute, Research Triangle Park, NC. December 15, 1992.
Gullet preheater costs.
9. Cole, W.E., F. Becker, L. Donaldson, and S. Panahe. Operation of a Gullet
Preheating System. Ceram. Eng. Sci. Proc. 11(1-2):60. 1990.
10. Telecon. Newsome, M., Anchor Glass Company, Tampa, FL, to Spivey, J.J.,
Research Triangle Institute, Research Triangle Park, NC. November 10, 1992.
Costs of electric boost.
11. Brown, J.T. 100% Oxygen Fuel Combustion for Glass Furnaces. Ceram. Eng. Sci.
Proc. 12(3-4):608. 1991.
12. Letter and attachments from Newsome, M., Anchor Glass Company, Tampa, FL, to
Jordan, B.C., EPA/OAQPS, Research Triangle Park, NC. October 2, 1992.
Response to Section 114 letter on glass manufacturing.
13. Chem. Eng. A DeNOx System That Handles Hot and Dusty Waste Streams, p. 21.
October 1992.
6-9
-------
14. Letter from Wax, M.J. Institute of Clean Air Companies, Washington, DC, to
Spivey, J.J., Research Triangle Institute, Research Triangle Park, NC. January 7,
1993.
15. Grove, M., and W. Strum. NOX Abatement System: Using Molecular Sieve
Catalyst Modules for a Glass Melting Furnace. Ceram. Eng. Sci. Proc. 10(3-
4):331. 1989.
16. Letter and attachments from Robinson, J.R., AFG Industries, Inc., Kingsport, TN, to
Jordan, B.C., EPA/OAQPS, Research Triangle Park, NC. August 20, 1992.
17. Letter and attachments from Keil, J.R., Libbey Owens Ford, Toledo, Ohio, to
Neuffer, W.J., EPA/OAQPS, Research Triangle Park, NC. June 29, 1993. Response
to Section 114 letter on glass manufacturing.
18. Letter from Pickens, R.D., Nalco Fuel Tech, Naperville, IL, to Spivey, J.J., Research
Triangle Institute, Research Triangle Park, NC. January 14, 1993.
19. Letter from Keil, John. Libbey-Owens-Ford, Toledo, OH, to Neuffer, W. J., U.S.
EPA/OAQPS, Research Triangle Park, NC, June 30, 1994.
6-10
-------
CHAPTER 7
ENVIRONMENTAL AND ENERGY IMPACTS OF NOX CONTROLS
This chapter presents the energy and environmental impacts of the NOX control
technologies described in Chapter 5. These include low excess air, changing air/fuel
contacting, retrofit low NOY burners, oxy-firmg, cullet preheat, electric boost, selective
s\
catalytic reduction (SCR), and selective noncatalytic reduction (SNCR).
7.1 AIR POLLUTION IMPACTS
7.1.1 NO Emission Reductions
Table 5-8 presents NOX emission reductions for each of the technologies discussed
above with the exception of low excess air (LEA) and changing air/fuel contacting. As
discussed in Chapter 5, these two combustion modifications are assumed to be necessary
to achieve the uncontrolled NOV emissions levels of Table 6-1. Table 5-9 shows that NOY
.X- A
reductions from 1 2 to 98 percent from uncontrolled levels can be achieved. The greatest
reduction (98 percent) is achieved by oxy-ftrmg.
7.1.2 Emissions Tradeoffs
7.1.2.1 Combustion Modifications. Combustion modifications (Section 5.2)
include LEA, changing air/fuel contacting, low NOV burners, and oxy-fmng. These, like
other combustion modifications designed to minimize NOV may affect the emissions of CO
/C
and unburned hydrocarbons.
Low Excess Air. The formation of NO in a glass furnace depends on temperature,
O2/N2 concentration, and residence time, per Equation (4-4) in Chapter 4. LEA operation
will generally decrease NOX emissions but may will increase CO emissions. Figure 7-1
shows this effect for an end-fired regenerative glass furnace producing about 1 65 tons of
glass/day. The lower the oxygen content of the flue gas (i.e., the lower the excess air),
the lower the NOX emissions. However, CO emissions increase rapidly below about 2.2
7-1
-------
4000
1000
2.0 2.5 3.0
O2 Content in Voi %
3.5
Rgure 7-1. NOX and CO concentrations of the flue gas as a function of the oxygen
content from an end-fired regenerative furnace (1 mg NO/m3 = 0.75 ppm NO;
1 mg CO/m3 = 0.80 ppm CO)?
7-2
-------
percent oxygen. For this particular furnace, operation at about 2 percent oxygen in the
flue gas (corresponding to about 13 percent excess air) minimizes both CO and NOX
emissions.
No adverse effect on glass quality is reported for NOX up to 3100 ppm and CO
o o
concentrations above 1000 ppm. ' However, CO concentrations that result in a net
reducing atmosphere in the furnace are known to adversely affect glass quality.
Excess air levels in actual glass furnaces are highly site specific, though levels of 5
«3
to 10 percent are typical of at least two commercial furnaces. Though not reported in
this study, emissions of unburned hydrocarbons (HC) are generally directly proportional to
CO emissions and thus would follow the same qualitative trend as CO emissions shown in
Figure 7-1.
Changing Air/Fuel Contacting. As with LEA operation, any change in the
combustion process that affects NO may affect CO and HC emissions. The effect of the
s\.
mixing factor (a measure of air/fuel contacting defined Equation (5-1) in Section 5.2.1.2)
on NO emissions is reported, though the corresponding effect on CO emissions is not
summarized. '^ However, data are presented showing the same qualitative trend as
Figure 7-1, i.e., changes in air/fuel contacting that decrease NOX cause an increase in
CO. For example, when modifications were made causing NOX to decrease from 2250
ppm to 900 ppm, CO increased from 140 ppm to more than 1000 ppm.
Low NOX Burners. As with LEA and air/fuel contacting, the primary tradeoff in low
NOY burners is between NOV and CO emissions. Tests were made on a regenerative
A A
end-port furnace producing between 154 and 192 tons of glass/day. The effect of
"staged combustion" and flue gas recirculation, which were two of the measures taken to
reduce NOX, are shown in Figure 7-2. The "staged air proportion" in this figure refers to
the proportion of the total combustion air that is taken from the flue gas and introduced
downstream of the burner but within the furnace (see Figure 5-8). The greater the
proportion of staged air, the lower the expected peak flame temperature would be, and,
therefore, the lower the NOV emissions, all else being equal. The oxygen concentration
A
was varied in a series of tests and is shown as a parameter in Figure 7-2. Figure 7-2
shows that NOX emissions decrease and CO emissions remain essentially constant, with
decreasing oxygen concentration.
For a given oxygen concentration, the NOX emissions decrease, and CO emissions
are relatively constant, with increasing proportion of staged air. This suggests negligible
7-3
-------
A 2500
2000 -
1500 -
1000 -
500
O2 in vol %:
a 3.81
o 2.62
A 2.80
o 2.35
6 12 18
• Staged Air Proportion in %
6 12 18
• Staged Air Proportion in % •
(a) NOX concentration.
(b) CO concentration.
Figure 7-2. Concentration of the flue gas as a function of the staged-alr proportion
(left side fired) from an end-fired regenerative furnace.1
7-4
-------
impact on CO emissions, at least for this particular retrofit low NOV burner.
J\
Oxy firing. The impact of oxy-firing on air emissions other than NOX is reported in
Reference 6. The results of stack tests done on a 340 tons/day side port regenerative
furnace before and after conversion to oxy-firing is shown in Table 7-1. In addition to a
substantial decrease in NO particulate, CO, and CH^ emissions decreased. Particulate
emissions decrease because the higher flame temperatures produce fewer unburned
hydrocarbons.7 Only SOX emissions increased. The authors state that SOX emissions
could be reduced to levels achieved before oxy-firing by changes in the batch formulation.
The reduction in CO and CH^ emissions suggests more complete combustion. The
decrease in particulates is possibly a consequence of the greatly reduced gas velocity
across the melt (due to the absence of nitrogen in the combustion air) which carries fewer
fine particles out of the furnace.
7.1.2.2 Process Modifications.
Gullet preheat. Gullet preheaters are designed to increase the overall thermal
efficiency of the glass manufacturing process by transferring heat that would otherwise be
lost in the flue gas to the cullet. The Teichmann and Tecogen systems use direct contact
heat transfer, while the Zippe system uses indirect heating. This affects the air emissions
since direct contact may allow some contaminants in the flue gas to be adsorbed by the
cullet but may increase particulate emissions since fine dust in the cullet can be carried
away by the flue gas.
The Teichmann system has been installed on a 220 tons/day regenerative furnace
O Q
in Weigand, Germany. ' No quantitative results are provided on the impact of the
preheater on emissions other than NOX, though "the cullet preheater is an effective filter
for dust dislodged during on-line cleaning.' Measurements indicated that the preheater
actually removed about half the particulate from the furnace emissions. However, dust in
the cullet itself was entrained back into the exiting flue gas, so that the net effect of the
preheater on particulates leaving the stack is unclear. Data are provided on SO emissions
s\
while the preheater was operating. These averaged about 2.2 Ib SOv/ton glass (around
XV
200 ppm). Though no comparison to operation without the preheater is given, the
1 9
statement is made that ". . . preheater is reducing SOY emissions."
s\
Finally, results on an indirect cullet preheat system at Vetropak AG in Switzerland
show that indirect heating eliminates possible entrainment of dust from the cullet. As
discussed above, this apparently does not occur in the Techmann system.^ it is also
7-5
-------
TABLE 7-1. EFFECT OF OXY-FIRING ON AIR EMISSIONS6
Conventional firing Oxy-firing
Parameter (Ib/ton glass pulled) (Ib/ton glass pulled)
Paniculate 1.19 0.884
NOY 5.03 0.812
J{
SOV 0.612 0.968
A
CO 0.08 0.003
CH 0.020.008
7-6
-------
suggested that HF, HCI, and sulfur can be adsorbed in direct contact systems and that,
while this may be an advantage in eliminating emissions of these compounds, it adversely
affects glass quality,
Electric boost. As a first approximation, it can be assumed that all emissions from
glass melting, including NOV (Section 5.3.2), are reduced in direct proportion to the
J\
percent of the furnace energy supplied electrically. Quantitative estimates of these
emissions, including SOY, acid gases, and particulates, are not available.
s\
In addition, electric boost generates additional emissions and wastes associated
with the production and distribution of electricity if it is generated from the combustion of
fossil fuel. These are not considered here, though they may be large.
7.1.2.3 Postcombustion Modifications.
Selective catalytic reduction. The injection of ammonia into the flue gas from a
glass furnace inevitably results in some unreacted ammonia and some byproducts (e.g.,
, C\2> (NH^) ,80^) in stack emissions. Such emissions generally increase with time
as the catalyst ages. In most SCR applications, unreacted ammonia ("ammonia slip") is
kept below 20 to 40 ppm by controlling the injection rate of ammonia. Much lower
values, of the order of 1 to 5 ppm, are reported for boilers. However, a "maximum"
ammonia slip of 10 to 30 ppm is reported for an SCR unit installed on a glass furnace in
1 fi
Germany. D A value of "below" 30 ppm for an SCR unit on another glass furnace in
Germany was reported. The injection of ammonia may increase stack particulate
emissions due to the formation of ammonium sulfate/bisulfate and ammonium chloride,
though there is of course a corresponding stoichiometric reduction in gaseous SOV and HCI
.A
emissions. There is potential with SCR for a solid waste disposal problem of spent
catalyst, though this can often be returned to the vendor to be reactivated.
Assuming 68 scfm of flue gas per ton of glass produced (see footnote b of Table 5-
8), an ammonia slip of 10 ppm would result in the following emissions from the three
model plants in Table 6-1:
7-7
-------
Plant size Emissions of
(ton/day) ammonia (Ib/day)
50 ' 2.3
250 11.6
750 34.7
Selective noncatalvtic reduction. As with SCR, the SNCR process generates
ammonia slip and byproduct salts from the acidic components of the flue gas. For PPG's
proprietary SNCR process, ammonia slip is reported as 39 ppm. CO emissions are less
than 1 ppm and particulates 0.065 gr/dscf. Values before installation of the system are
not reported.
AFG systematically tested the effect of the ammonia injection rate on NOX, CO,
JO
SOo, particulate, and NHg emissions at their Victorville, CA, plant. Table 7-2 presents
the results, which provide a direct measure of the effect of ammonia injection in this
®
Exxon De-N0x unit on NOX, S02- total particulate, and CO. Two comparisons can be
made to measure this effect. The first is to compare the test done on 2/25/88 with the
series of tests on 2/23/88. The second is to compare the tests done on 6/7/88 with and
without ammonia injection. Fluctuations in firing, glass production, flue gas rates and flue
gas temperatures may be responsible for the wide variation in carbon monoxide and sulfur
dioxide levels. The data indicate that ammonia injection in this SNCR process
• has no significant effect on total particulate emissions,
• slightly increases CO emissions, and
• slightly decreases SO2 emissions
and ammonia slip (unreacted ammonia emissions) increases with ammonia injection rate.
Operating experience, primarily in boilers, has identified several concerns with
both ammonia and urea-based SNCR processes. The most frequently reported is the
buildup of ammonium bisulfate scale, which can also be emitted as a particulate. Because
natural gas, which has very little sulfur, is used in most glass furnaces, such sulfate
formation is negligible in glass furnace flue gas ducts. Even when sulfur-containing fuels
such as fuel oil are used, vendors report that process modifications have been made to
O -I
minimize problems of sulfate scale deposition. SNCR processes also appear to convert
some NO to NoO. The rate of NoO formation is a weak function of both the reactant
and NO concentration (ammonia or urea/NO ratio). However, N2O formation seems to be
2 ^
inherently more prevalent in systems using urea than those using ammonia. ° SNCR
7-8
-------
CN
CM
tc.
O
z
V)
u.
o
V)
w
LU
I-
ai
cc
O
LL
O
CN
r*-
ui
_i
CO
oo
X
2
O
O
to
05
ffl
to
"3
_o
a
a
15
4"^
o
H
cs
O
w
X
O
E
a
a.
C to
O to
.*; co
£ 01
E
a.
Q.
c to
O tO
*"* CO
^ "01
*^
o
CO
^
w
co
C to
S co
£ CO
E
a.
a
£ w
o to
.+; co
^ O)
£
a.
o.
C to
O to
< S.
£1 CO
<
•z.
<
•z.
^
r^
S
o
d
ID
r-*
CO
q
d
CM
CD
d
r>-
d
00
0
CD
d
<
2
<
•z.
r^
06
in
00
'tf
CD
d
CO
CO
q
d
o
r-v
d
CM
in
CN
•*
CN
CD
d
oo
00
en
,_
CO
"
CO
00
f*v
o
00
'
IcD "^1
o g oo S?
i= o co
o '% ^
0-| co
3 c5
CN
1
^
CD
<
Z
<
•z.
co
CN
cn
0)
o
d
in
CM
CO
o
d
,_
in
d
q
CM
*t
PV
00
pv
d
00
o
•-
CD
•^t
^
CO
CO
in
CN
CN
to
c
Q
co
CO
'p
L_
05
•o
05
"o
L_
4->
C
o
u
cc
•z.
cc
~z.
p^
CTJ
00
o
d
CN
-3-
00
q
d
CD
in
d
CD
in
CM
O)
^t-
q
in
o
CD
CM
•^
00
05
•o oo
00 00
S2?5
r^ CM
CD CN
C75
CO
en
CD
•^i-
O
d
CD
in
CO
O
in
CD
d
in
CO
00
o
d1
p**
in
d
CO
d
«*
CD
O
PV
d
•^i-
CM
as
in
,-i
*
i_
£L
00~
4~*
M—
in
p>
r^
•
en
CO
*—
T_
00
o
d
CM
^f
00
00
d
4—
CC
•z.
00
"*
00
p^
CN
CD
d
oo
CM
00
pv
d
"
<
ff
\
X
2
CO
r—
CM
00
i —
* —
d
00
d
CO
co
CM
d
cc
2:
cc
•z.
^
CO
00
0
r~»
d
00
CO
00
P-.
r-^
'
u.
.C
"0~
4-f
H~
O
>o
Jcn
•
q
^
CO
rt
CD
d
CD
CO
CO
*t
«—
d
cc cc
2 "Z.
cc
•z.
CM
CO
CO
O
CO
co
d
r-
CM
oo
p-
d
O
CD
rr^~ ff
" i \ ,
i "" i
"Z. • ~Z.
05
J3
CD
g
"o.
Q.
CD
4-1
O
Z
II
<
reported.
*--
o
•z.
H
cc
were made with the ammonia injection cut off, all other operating parameters were then kept the same in
tfl
easurement!
E
05
co
05
.C
t-
e dates given refer to emission test dates. Emissions are normalized using a glass production of 450 ton/day.
_c
1-
to
4-1
co
05
4-1
4->
C
O5
3
CT
05
CO
-Q
3
CO
ownstream of SNCR unit on these dates.
-D
05
•o
co
E
CD
i_
05
§
CO
4~*
CO
05
+•*
o
•z.
c
X
O
•z.
05
_g
05
jr
4->
1_
O
H-
CD
_c
'*^
3
O
o
o
CD
-C CO
3 W
i z
I 1
CD
•*— '
03
t_
5
o
*4-
co
CD
0)
05
3
4—
05
.c
cf
g
~4->
CD
V«
4-<
C
05
O
C
o
0
i_
o
M—
T5
05
4-t
O
05
"E
CD
'E
o
E
E
<
on this date to test the effect of NHg concentration. The values given are four NHg injection rates in ftj/hr.
ata at 775 ft^/hr ammonia.
05
•o
CD
E
05
h_
05
§
co
co
05
4-t
L_
3
0
LL
T3
om 2/24/88
l_
*+-
c
05
J^
CD
.Q O
TJ 05 H-
7-9
-------
processes may also increase CO concentrations in the flue gas, though the increase for
urea-based systems is apparently much less than that due to combustion modifications
such as overfire air and substoichiometric combustion air.24 One reference states that
ammonia injection has no effect on CO emissions.25 Interestingly, the intentional addition
of CO in the reaction zone of the process broadens the operating temperature for
urea-based systems, even at CO concentrations as low as 500 ppm, although it increases
oc •
N£O emissions. ° However this does not imply that stack emissions of carbon monoxide
increase. Some data on other combustion systems suggest that in some cases the effect
of ammonia injection on CO emissions is negligible and that some data spread is inevitable
due to varying combustion conditions.
7.2 ENERGY IMPACTS
7.2.1 Combustion Modifications
7.2.1.1 Modifications to Existing Burners.
Low Excess Air and Air/Fuel Contacting. Data suggest that LEA operation and
changes in air/fuel contacting do not significantly affect furnace energy usage (MM Btu/ton
77
glass produced). Based on this, these two combustion modifications are assumed to
have negligible energy impacts.
7.2.1.2 Low IMP Burners. The Kortig burner results in energy savings by reducing air
9R
infiltration, but no quantitative results are presented. Such a claim would be difficult to
quantify since air infiltration is highly site specific. Such burners may be more efficient
than others and would therefore save energy. However, a direct comparison cannot be
made with the existing data.
7.2.1.3 Oxv-firing. Oxy-firing results in lower energy consumption (MM Btu/ton glass
produced). This is, in fact, one of the primary reasons for its use. Figure 7-3 shows the
"available heat" as a function of flue gas temperature for various levels of oxygen.
Available heat is defined as the gross heating value of the fuel minus the heat carried away
in the flue gas. Fuel savings of 15 percent for oxy-firing on a 75 tons/day have been
estimated for an end-fired regenerative furnace. Production during the test was 58
tons/day. Further, at essentially the same fuel usage rate, glass production increased
from 62.7 to 75.8 tons/day(21 percent), as shown below:
7-10
-------
2000 2200 2400 2600 2800 3000
Rue Gas Temperature (°F)
Figure 7-3. Available heat as a function of flue gas temperature.7
7-11
-------
Production
(tons/day)
Fuel usage
(MM Btu/hr)
Air-firing
62.7
13.7
Oxy-firing
75.8
13.6
This corresponds to 30 to 40 percent energy savings (Figure 7-4) for regenerative glass
"3D
furnaces, but absolute values (MM Btu/ton glass) are not provided. For the Gallo plant,
natural gas usage was 9.5 percent lower than with air-firing (3.74 MM Btu/ton with air-
09
firing, 3.39 MM Btu/ton for oxy-firing). This energy savings is due to two principal
factors. First, there is reduced radiation from the melting furnace to the regenerator due to
reduced port area. The port area can be reduced because the volumetric flow rate of the
flue gas is reduced. Second, the greatly reduced nitrogen content of the combustion air
means less energy lost to the flue gas. There is also an energy savings due to a lower flue
gas flow rate which requires less electrical energy for the flue gas fan. However, energy
oo
or net utility cost savings are rare when the cost of oxygen is taken into account.00
7.2.2 Process Modifications
7.2.2.1 Gullet Preheat. Gullet preheaters are designed to recover heat from the flue gas
and therefore will reduce the energy consumption in glass melting.
The Teichmann cullet preheater accounts for 8 to 1 2 percent of the total energy
®
saved by their Low NOY Melter , which also incorporates other energy savings
J\
features. ' Insufficient information is given to determine absolute energy savings
associated with the cullet preheater alone.
A 20 percent decrease in energy consumption for the Tecogen preheater (a savings
of 1 MM Btu/ton, from 5 to 4 MM Btu/ton) is estimated. Actual tests showed a slightly
lower energy savings (0.86 instead of 1 MM Btu/ton) at a production rate of 257 tons/day.
An 7 to 10 percent reduction in energy consumption is reported for a 240-tons/day
furnace equipped with a Tecogen cullet preheater processing about 80 tons cullet/day, i.e.
about one-third of the furnace feed.0"5 No absolute values are given.
Energy consumption would decrease by 12 percent on a 300-tons/day furnace
which uses 100 percent cullet feed (no virgin batch ingredients) if all the cullet were
7-12
-------
Energy Penalty
-30%
Energy Savings
All Electric
Regenerative
Hard Glass
Regenerative
Soda-Lime
30%
Figure 7-4. Energy impact of oxy-firing
27
7-13
-------
preheated. ° This calculation is extrapolated from actual results obtained at the Vetropak
plant when 25 percent of the cullet was preheated. No absolute values of energy reduction
(MM Btu/ton) are given.
7.2.2.2 Electric Boost. Figure 7-4 shows the energy penalty associated with electric
boost. The relationship between electric boost and glass production has been estimated to
be 25 tons glass/day per 1000 kVA (or 1 ton glass per 800 kWh).^7 As discussed in
Section 5.3.2, electric boost is more efficient than gas firing, i.e., more of the theoretical
energy input to the melt electrically is actually transferred to the melt. This efficiency
value for electric boost is roughly 70 percent. One refetence states this as 73 percent
compared to about 30 to 35 percent for gas firing (see Section 3.2). However, the
production and distribution of electricity from fossil fuels is only about 20 to 25 percent
efficient, making electricity from fossil fuels less efficient than gas firing. Thus, the energy
impact of electric boost would be to increase the demand for electricity, which is
inherently less efficient in delivering energy to the glass melt from the original fuel than
gas firing.
The electrodes used for electric boosting are made of molybdenum. It is not known
if these pose a solid waste disposal problem.
7.2.3 Postcombustion Modifications
7.2.3.1 Selective Catalytic Reduction. There is some pressure drop across the SCR
catalyst that will require additional electrical energy for the flue gas fan. Typically, this
pressure drop is of the order of 5 to 10 in. h^O. For a pressure drop of 10 in. h^O, and
using a value of 68 scfm per ton/day of glass (see footnote b of Table 5-8) and a fan
efficiency of 60 percent, calculations can be made
using the following equation:
Power (KW) = 1.17 x 10"4 QAP
where
Q = gas flow rate, scfm
AP = pressure drop, in h^O
€ = fan efficiency, 0 < € < 1.
7-14
-------
The results are shown below:
Plant size
(tons/day)
50
250
750
Fan energy
(kW)
6.6
33.2
99.4
Because dust can foul the catalyst, an SCR unit would typically be installed
downstream of a particulate control device, such as an electrostatic precipitator (ESP)
(e.g., Reference 16; see also Figure 5-25 in Section 5.4.2). If the temperature at this
point is below 350 to 500 °C (660 to 930 °F), the gas may need to be reheated with gas
burners. This highly site-specific energy impact is not considered further here.
7.2.3.2 Selective Noncatalytic Reduction. SNCR introduces no additional pressure
drop in flue gas. Energy consumption in the SNCR process is related to the pretreatment
and injection of ammonia-based reagents and their carrier gas or liquids. Liquid ammonia
or urea are injected in liquid form at high pressures to ensure efficient droplet atomization
and dispersion. In some Thermal DeNO installations, anhydrous ammonia is stored in
A
liquid form under pressure. The liquid ammonia must be vaporized with some heat, mixed
with carrier gas (air or steam) and then injected for adequate mixing. The amount of
electricity used depends on whether the process uses air or steam for carrier gas. If steam
is used, less electricity is needed but power consumption must take into consideration the
amount of steam used.
7-15
-------
7.3 REFERENCES
1. Barklage-Hilgefort, H., and W. Sieger. Primary Measures for the NOX Reduction on
Glass Melting Furnaces. Glastech. 62(5):151. 1989.
2. Abbasi, H.A., and D.K. Fleming. Development of NO Control Methods for Glass
Melting Furnaces. Gas Research Institute. Final Report No. GRI-87/0202. August
1987.
3. Abbasi, H.A., and D.K. Fleming. Combustion Modifications for Control of NOX
Emissions from Glass Melting Furnaces. Ceram. Eng. Sci. Proc. 9(3-4):1 68-1 77.
1988.
4. Kircher, U. Gas Warme Int. 35(4):207-212. 1986.
5. Ref. 2, pp. 41, 90.
6. Moore, R.D., and J.T. Brown. Conversion of a Large Container Furnace from
Regenerative Firing to Direct Oxy Fuel Combustion. 1991 Glass Problems
Conference. November 1 2-13, 1991. American Ceramic Society. Westerville, OH.
p. 6. 1992.
7. Baukal, C.E., P.B. Eleazer, and L.K. Farmer. Basis for Enhancing Combustion by
Oxygen Enrichment. Ind. Heating. 22. February 1992.
8. Moore, R.H. LoNOx™ Glass Melting Furnace. Ceram. Eng. Sci. Proc. 11(1-2):89-
101. 1990.
9. Moore, R.H. LoNOx Melter Shows Promise. Glass Indust. p. 14. March 1990.
10. Ref. 9, p. 16.
11. Ref. 9, p. 18.
12. Ref. 9, p. 17.
13. Zippe, B.H. Reliable Gullet Preheater for Glass Furnaces. Ceram. Eng. Sci. Proc.
12(3-4):550-555. 1991.
14. Ref. 15, p. 553.
15. Maier, H., and P. Dahl. Operating Experience with Tail-end and High Dust DeNOx
Techniques at the Power Plant of Heilbronn. Joint EPA/EPRl Symposium on
Stationary Combustion NOV Control. March 1991.
XV
16. Grove, M., and W. Strum. NOX Abatement System: Using Molecular Sieve
Catalyst Modules for a Glass Melting Furnace. Ceram. Eng. Sci. Proc. 10(3-
4):330. 1989.
7-16
-------
17. Letter from Gocht, Lurgi, Frankfurt, Germany, to Spivey, J.J. Research Triangle
Institute, Research Triangle Park, NC. April 29, 1992.
18. Smith, J.C., and M.J. Wax. Selective Catalytic Reduction Controls to Abate NOX
Emissions. Institute of Clean Air Companies. Washington, DC. September 1992.
p. 18.
19. Letter and attachments from Osheka, J.W., PPG Industries, Inc., Pittsburgh, PA, to
Jordan, B.C., EPA/OAQPS, Research Triangle Park, NC. September 1, 1992.
Response to Section 114 letter on glass manufacturing.
20. Letter and attachments from Robinson, J.R., AFG Industries, Inc., Kingsport, TN, to
Jordan, B.C., EPA/OAQPS, Research Triangle Park, NC. August 20, 1992.
Response to Section 114 letter on glass manufacturing.
21. Mincy, J.E. SNCR Technology: The Next Generation: presented at NOX V
Conference. February 10-11, 1992. Long Beach, CA. Council of Industrial Boilers.
Burke, VA.
22. Kokkinos, A., J.E. Cichanowicz, R.E. Hall, and C.B. Sedman. Stationary
Combustion NOX Control: A Sumr
Manage. Assoc. p. 1255. 1991.
Combustion NOV Control: A Summary of the 1991 Symposium. J. Air Waste
s\
23. Muzio, L. NoO Formation in Selective Non-Catalytic Reduction Processes in Proc.
1991 Joint Symposium on Stationary Combustion Nox Control. NTIS. 1991.
24. Moore, R.D., and R.E. Davis. Electric Furnace Application for Container Glass.
Ceram. Eng. Sci. Proc. 8(3-4):1 88-1 99. 1987.
25. Haas, G.A. Selective Noncatalytic Reduction (SNCR): Experience with the Exxon
Thermal DeNOx Process. Presented at the NOX Control V Conference, Council o1
Industrial Boiler Owners. Long Beach, CA. February 10-11, 1992.
26. Teixeira, D. Widening the Urea Temperature Window. In Proc. 1 991 Joint
Symposium on Stationary Combustion NOV Control. NTIS. 1991.
s\
27. Ref. 3, pp. 42, 179, 187.
28. Sieger, W. Development of Reduced NOV Burners. Glass Tech. 31(1):6. 1990.
s\.
29. Kobayashi, H., G.B. Tuson, E.J. Lauwers. NOX Emissions from Oxy-Fuel Fired
Glass Melting Furnaces. Paper presented at the European Society of Glass Sciences
and Technology Conference on Fundamentals of the Glass Manufacturing Process,
Sheffield, England. September 9-11, 1991. Union Carbide. Tarrytown, NY. p.
10.
7-17
-------
30. Brown, J.T. Development —History and Benefits of Oxygen-Fuel Combustion for
Glass Furnaces. Corning Glass. Corning, NY. Presented at Latin American
Technical Symposium on Glass Manufacture. Sao Paulo, Brazil. November 18,
1991. p. 21.
31. Ref. 30, p. 22.
32. Ref. 6, p. 5.
33. Brown, J.T. 100% Oxygen Fuel Combustion for Glass Furnaces. Ceram. Eng. Sci.
Proc. 12(3-41:601. 1991.
34. De Saro, R., L.W. Donaldson and C.W. Hibscher. Fluidized Bed Glass Batch
Preheater, Part II. Ceram. Eng. Sci. Proc. 8(3-4):171-180. 1987
35. Cole, W.E., F. Becker, L. Donaldson, S. Panahe. Operation of a Gullet Preheating
System. Ceram. Eng. Sci. Proc. 11 (1-2), 53-68. 1990.
36. Ref. 13, p. 554.
37. Ref. 2, p. 39.
7-18
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-453/R-94-037
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Alternative Control Techniques Document
Emissions from Glass Manufacturing
NOX
5. REPORT DATE
June 1994
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
William J. Neuffer
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
U. S. Environmental Protection Agency
Emission Standards Division (MD-13)
Office of Air Quality Planning & Standards
Research Triangle Park, North Carolina 27711
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
EPA Project Manager:
William J. Neuffer (919) 541-5435
16. ABSTRACT
This alternative control/techniques (ACT) document describes available
control techniques for reducing NOX emissions from glass furnaces. Control
techniques include low NOX burners^oxy-firing, modified furnace, cullet
preheat, electric boost, selective catalytic reduction, and selective
noncatalytic reduction. Achievable controlled NOX emission levels, costs,
and cost effectiveness and environmental and energy impacts for these
controls are discussed. NOX formation and uncontrolled NOX emission levels
are also discussed.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
Glass Furnaces
Nitrogen Oxide Emissions
NOX Emission Controls
Low NOX Burners
Oxy-firing
Selective Catalytic Reduction
Selective Noncatalytic Reduction
18. DISTRIBUTION STATEMENT
19. SECURITY CLASS (This Report/
Unclassified
21. NO.OF PAGES
-- 154
20 SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
-------
U.S. Environmental Protection Agency
Region 5, Library (PL-12J)
77 West Jackson Boulevard, 12th Floor
Chicago, IL 60604-3590
-------