PB88-126412
Hazardous Waste Combustion in
Industrial Processes
Cement and Lime Kilns
Research Triangle Inst.
Research Triangle Park, NC
Prepared for
Environmental Protection Agency, Cincinnati, OH
Nov 87
U.S. DEPARTMENT OF COMMERCE
National Technical Information Service
NTiS
-------
-------
EPA/600/2-87/095
November 1987
HAZARDOUS WASTE COMBUSTION IN INDUSTRIAL PROCESSES:
CEMENT AND LIME KILNS
by
Robert E. Mournighan
Alternative Technologies Division
Hazardous Waste Engineering Research Laboratory
Marvin Branscome
Research Triangle Institute
Research Triangle Park, North Carolina 27709
EPA Contract 68-03-3149
Work Assignment 11-1
EPA Project Officer:
Robert E. Mournighan
Alternative Technologies Division
Hazardous Waste Engineering Research Laboratory
Cincinnati, Ohio 45268
Hazardous Waste Engineering Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
TECHNICAL REPORT DATA
(hesQ ,e.d Iw.r cncms 0i the Pv ersc before co.n eru g)
I flEPO T NO. 2.
EPA/600/2—87/095
3. ICIPIEPdrS ACCEU O NO.
4. TI1’LE AND SUETITLE
I 1AZARD IS WASTE COMBUSTION IN IPIDI)STRIAL
PROCESSES: CALCINING AND LIME KILNS
I. IPO T DAlE
November 1987
I.PI PO M INGO GA IWJZATIO NCOOE
7 AuTso s
Marvin Branscome and Robert E. Mourn ghan
I. PIRFO M$N DmGANIZATIO’ EPOU’T NO
11—1
I. PE PQ Mlp G O GANtZAT$ON NAME AND ADDRESS
Research Triangle Institute, Research Triangle
Park, MC 27709 and U.S. Environmental Protection
Agency, Cincinnati, Ohio 45268
10. PROGRAM ElEMENT NO.
11CONT RAC TIG RAN TNO.
R—03—3 149
12. SPONSORING AGENCY NAME AND ADDRESS
Hazardous Waste Engineering Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45268
13. TYPE OF REPORT AND PERIOD COVERED
14.SPONSO R ING AGENCV CODE
EPA-600/12
IS. SUPPLEMENTARY NOTES
-—
This report sunsnarizes the results of several studies relating to hazardous
waste combustion in cement and lime kilns. The tests included in this study are
four kilns tested by the 1 1.5. Environmental Protection Agency, four kilns tested
by State agencIes or the kiln operator, two Canadian tests, and one Swedish test.
The predomin&nt types of wastes tested included chlorinated organic compounds,
aromatic compounds, and metal—contaminated waste oil. The kiln types Include lime
kilns and cement kilns, which included the dry, wet, and preheated processes.
Fabric filters and electrostatic precipitators (ESPs) were the pollution control
devices used in these processes, and the primary fuels included coal, coke, coal!
coke, fuel oil, and natural gas/coke.
The parameters examined in this report were flestruction and Removal Efficiency
(ORE) of the °rincipal Organic Hazardous Constituents, particulate and HC1 emissions,
metals, and the effect of burning hazardous waste on SO 2 , N0, , &nd CO emissions. The
primary conclusion of this study is that DRE’s of 99.99% or greater can be obtained
in properly—operating calcining kilns. Particulate matter can increase when chlori-
nated wastes are burned in a kiln equipped with an electrostatic precipitator. Those
kilns equipped with fabric filters showed no change in emissions.
IY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDINT*PIIRS/OPIN ENDED TERMS
C. COSA1i FIeld/Group
IS. D$$TRI$UTION STATEMENT
Ra.EAcE TO P’JRLIC
i, SECURItY C$.AU (Thu *epro.i)
UNCLASSIFIED
t. NO. OF PAGES
7L
3O.UCU R ITYCLAU( T hu ,
UNCLASSIFIED
32.PRICE / hir
tPA P•viis aaa..i (•73)
• 1
-------
NOTICE
This document has been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names
or commercial products does not constitute endorse-
ment or recommendation for use.
11
-------
FOREWORD
Today’s rapidly developing and changing technologies and industrial products
and practices frequently carry with them the increased generation of solid and
hazardous wastes. These materials, if improperly dealt with, can threaten both
public health and the environment. Abandoned waste sites and accidental releases
of toxic and hazardous substances to the environment also have important environ-
mental and public health implications. The Hazardous Waste Engineering Research
Laboratory assists in providing an authoritative and defensible engineering basis
for assessing and solving these problems. Its products support the policies,
programs, and regulations of the Environmental Protection Agency, the permitting
and other responsibilities of State and loca1 governments and the need of both
large and small businesses in handling their wastes responsibly and economically.
This report will be useful to EPA’s Office of Solid Waste and Emergency
Response, U.S. EPA Regional Offices, and State and Local Environmental Permit
Writers. it contains a review and summary of all environmental assessments
of hazardous waste combustion in calcining kilns.
Thomas R. Hauser, Director
Hazardous Waste Engineering Research Laboratory
lii
-------
ABSTRACT
This report summarizes the results of several studies relating to hazardous
waste combustion in cement and lime kilns. The tests included in this study
are four kilns tested by the U.S. Environmental Protection Agency, four kilns
tested by State agencies or the kiln operator, two Canadian tests, and one
Swedish test. The predominant types of wastes tested included chlorinated
organic compounds, aromatic compounds, and metal-contaminated waste oil. The
kiln types include lime kilns and cement kilns, which included the dry, wet,
and preheated processes. Fabric filters and electrostatic precipitators (ESPs)
were the pollution control devices used in these processes, and the primary
fuels included coal, coke, coal/coke, fuel oil, and natural gas/coke.
The parameters examined in this report were Destruction and Removal Effi-
ciency (ORE) of the Principal Organic Hazardous Constituents, particulate and
HCI emissions, metals, and the effect of burning hazardous waste on SO 2 , NOR,
and CO emissions. The primary conclusion of this study is that DREs of 99.99
percent or greater can be obtained in properly—operating calcining kilns.
Particulate matter can increase when chlorinated wastes are burned in a kiln
equipped with an electrostatic precipitator. Those kilns equipped with fabric
filters showed no change in emissions.
iv
-------
_______ Page
in
iv
vi
vii
1
2.0 ConclusIons and Reco ,iendations
2.1 Conclijslons .
2.2 Recoix endations
Cement and Lime Production
31. OvervIew
3.2 Process Description .
3.2.1 Cement Kilns .
3.3.2 Lime Kflrts
T,st Results
4.1 Overview of Tests
4.2 Destrtjction and Removal Efficiencies (ORE)
for Principal Organic Hazardous
Constituents (POHCs)
4.3 POHC Emission Rates
4.4 Products of Incomplete Combustion (PICa)
4.5 Effects on Particulate Matter Emissions
4.6 Fat. of Chlorine
4.7 Effects on Lead Emissions and Removal by
Process Solids
4.8 Effects of Waste Combustion on NO and
302 Emissions
4.9 Emissions of Total Hydrocarbons (TI4C) and
Carbon Monoxide (CD)
Referencas 65
CONTENTS
Foreword
Abstract
Figures
Tables
1.0 Introduction
Section
4
3.0
4.0
5.0
3
3
5
6
6
7
7
13
15
15
• . . . 20
• 28
• 30
34
• . . • 42
• 51
• 56
• 62
V
-------
FIGURES
Number Page
3—1 Wet process cement kiln 10
3—2 Dry process cement kiln 11
3—3 Steady—state temperature and concentration
profiles along kiln length 12
4—] Alkali and chloride cycles 44
vi
-------
Number
3-1
3—2
4-1
4-2
4-3
4-4
4-5
4-6
4—7
4-8
4—9
4-10
4-11
4-12
4-13
4-14
4-15
4-16
4-3.7
-4—18
Page
8
14
16
22
24
27
29
31
Lawrence
33
36
38
40
46
48
50
52
55
59
60
63
in Cement
TABLES
Typical Compositions of Raw Materials Used
Manufacture
Control Devices on Cement Kllns
Summary of Cement Kiln Tests
Summary of DREs
PCB Test Results for Peerless Cement .
OREs of Compounds from f 5 Train at Lone Star
Summary of POHC Emission Rates
PICa from Waste Combustion at San Juan
Potential PICs from Coal Combustion at the
Paulding Test
Emissions of Particulate Matter
Susrunary of Particulate Matter Results at St.
Cement
Particulate Matter Results for Stora Vika
HC1 Emissions
HC1 Results for the Paulding Kiln
HCI Test Results for Lone Star
Boiling Points of Selected Compounds .
Summary 0 f Lead Emissions
Summary of SO 2 Emissions
Summary of MO Emissions
Summary of ThC and CO Concentrations
vii
-------
1.0 INTRODUCTION
Preliminary feasibility studies and tests hive Indicated that some
industrial facilities with high—temperature combustion processes may provide
a cost-effective and environmentally acceptable alternative to the disposal
of hazardous wastes in incinerators or landfills. These industrial processes
have the potential for recovering the heating value of the waste, removing
chlorides, and providing a destruction and removal efficiency (DRE) equiva-
lent to hazardous waste incinerators.
Several Industrial processes can provide temperatures and residence
times similar to those required for hazardous waste incinerators. Examples
include cement, lime, and aggregate kllns, industrial boilers, and blast
furnaces. Cement and lime (calcining) kilis are of particular interest
because of their potential for reducing the environmental impact of disposal
in a cost—affective manner. The promising characteristics of calcining
kilns include:
• Existing high—temperature combustion process at 1,3500 to
1,650° C (2,500° to 3,000° F) with a gas residence time on
the order of seconds.
• Large number of plants scattered throughout the country,
which could potentially handle large quantities of coatbust-
Ible hazardous waste liquids.
• Large-scal. equipment in place, including process control
and pollution control; relatively small capital investment
requf red.
• Instantaneous temperature excursions unlikely because of the
huge thermal Inertia.
• The alkaline environment In the kiln absorbs HC1 from chlor-
inated waste combustion.
• Kilne are operated under draft (slight vacuum); therefore,
there would be little outward leakag. of fumes, mostly
inward leakage of air.
1
-------
• Koncombustible hazardous components may be retained in the
process solids.
• Energy savings from substitution of waste fuel for oil,
coal, or gas.
The combustion of hazardous waste in kilns has been investigated in
Canada, 2 Swedan and Nor .ay 5 as wsll as In the U.S. Papers have been
published that promote calcining kflns as a method of waste disposal that
Is better than current practices, particularly for chlorinated organics. 2
Papers have also been published that warn of actual and potential risks
associated with waste combustion. 1 Several kilns have been tested during
waste combustion to provide more data on the effects of waste combustion.
Some of these tests have been comprehensive In their attempt to determine
effects on emissions, product quality, fat. of metals and chlorine, and
products of incomplete combustion. The test results are used in this
report to:
Present DREs of principal organic hazardous constituents
(POHCs). -
• Compare baseline operation (no waste fuel burned) to the
operation with waste fuel.
• Determine what products of incomplete combustion (PICs) are
formed.
Examine the fate and distribution of metals.
• Examine the effect of waste burning on 1C1 emissions, and
determine the fate and distribution of chlorine.
• Examine the effects of waste fuel burning and process con-
ditions on other pollutants (e.g., particulate matter, NO ,
SO 2 , total hydrocarbons).
2
-------
2.0 CONCLUSIONS AND RECOMMENDATIONS
2.]. CONCLUSIONS
OREs of 99.99 percent are generally achievable in properly operating
calcining kilns. This level of ORE has been achieved at different types of
kllns with a variety of difficult-to-destroy wastes. Baseline, background,
and contaminant lev ls of the target POHCs Imply the OREs are probably
conservative, as are the estimates of POHC emission rates. The high OREs
for spiked POHCs (such as Freon 113), which are unique to the waste fuel
and generally less susceptibl, to background or blank problems, support
this conclusion. The lowest measured emission rates and highest OREs were
found at two kiln tests (Store Vika and Rockwell Lime) that apparently had
the least interference from background levels or contamination.
A small Increase in PlC emissions was observed during the waste burns
at some kilns, and at others no differences were noted between baseline and
waste fuel burns. PICs can Increase at a poorly operating kiln whether it
is fired with waste f ...al or not. PICs from the primary fuel, especially
coal, have been identified and their emission rates appear to be independent
of waste combustion within the accuracy of the emission measurements.
The tests for particulate matter Indicate that particulate matter
emissions can be controlled when chlorinated wastes are burned and do not
necessarily increase. Two tests at kilns equipped with ESPs (thloride
input up to 4 kgCl,’14g clinker) and two tests at kilns equiped with fabric
filters (average chloride input up to 5.5 kgCl/Mg clinker) showed no increase
in particulate matter emissions when chlorinated wastes were burned.
E tensiv. tests at St. Lawrence Cement and Store Vika indicated that control-
led particulate matter emissions increased as the chloride loading Increased.
The increase nay be limited by altering the ESP to compensate for changes
In the dust resistivity, controlling chloride Input, and altering the
chloride cycle In the kiln.
3
-------
HC1 emissions increase as the chloride loading Increases; however,
generally over 90 percent, and in several cases over 99 percent, of the
additional chlorine entering the kiln is retained in the process solids
(wast. dust and clinker). Most of the additional chloride Is removed with
the waste dust, and several plants increased the rate of waste dust removal
to aid In controlling the chloride cycle. Chloride accumulation, although
probably variable from kiln to kiln, appears to start In the range of 6-9
kgCl/Mg clinker with a tendency toward ring formation at the upper •nd of
the range.
Lead emissions and the lead concentration In the waste dust increase
when chlorinated wastes contaminated with lead are burned. However, base-
line emission rates of lead are low to begin with and total lead emissions
during waste combustion are roughly comparable to the lead emissions from
several automobiles burning leaded gasoline. The calclnlng kiln has the
potential to retain over 99 percent of the lead entering the kiln in the
process solids (waste dust and clinker). In comoarison, a boiler burning
waste oil retains less than 50 percent of the lead, and probably retains
even less because of Tosses during soot blowing.
s t emissions are generally decreased when the waste fuel replaces a
sulfur—containing fuel. SO 2 emissions were found to be a function of
several variables, such as sulfur loading, oxygen input, and temperatures.
Cement kilns can be operated to obtain low 502 concentrations. Lime kilns
are operated at higher SO 2 levels deliberately to avoid sulfur in the lime
product.
N0 emissions are not significantly affected by waste combustion.
concentrations vary over time at the same kiln and are affected by oxygen
input, primary to secondary air ratio, and temperatures.. Continuous monitors
for H0 respond rapidly to process changes.
Emissions of carbon monoxide, especially during coal combustion, can
exhibit short-lived spikes and are generally indicative of combustion
stability. Total hydrocarbon emissions increased at several plants during
waste combustion and were not significantly different at others; hydrocarbon
concentrations below 10 ppm were generally observed.
4
-------
2.2 RECOMMENDATIONS
All emission tests should collect an adequate number of background and
blank samples because the required low detection levels are easily biased
high by low level contamination. The testing personnel should be aware of
and account for other sources of the POHCs, such as process water, coal
combustion, and contamination.
As chlorinated waste combustion continues, more information should be
gathered on the chloride and sulfate cycles In the kiln. Guidance may then
be provided to other operators on controlling the cycle, minimizing lid and
502 emissions, and controlling the dust loading on the pollution control
device. General guidance is currently available, but quantitative guidance
could be developed. The operator has additional incentive to obtain the
information to prevent chloride or sulfate ring formation, corrosion of
equipment, and to maintain product quality.
Continuous monitors for N0 and 502 show great promise for kiln control,
optimization of combustion conditions, and emission control. A study of
controlled process changes In conjunction with the continuous monitors
could yield information that is useful to both the operator and regula tory
agency. Improved and more consistent kiln operation could result in lower
emissions at calcining kline whether wastes are burned or not.
Fthally, at some point In the future a comparative analysis of tech-
nologies available for hazardous waste disposal should be conducted. The
study should provide the facts, alternatives, and recommendations for the
benefit of State agencies, the public, waste generators, and waste disposars.
5
-------
3.0 CEMENT AND LIME PRODUCTION
3.1 OVERVIEW
Portland Cement
Portland cement is one of the largest Mineral commodities produced in
the United States with an annual production capacity in 1982 of 79 million Mg/yr
This capacity was owned by 46 companies with the top 10 accounting for 59
percent of the total capacity. A total of 135 plants with 274 kilns are
located throughout 38 states. The kilns are quite variable in capacity and
range in size from less than 45,000 Mg/hr up to 1.7 million Mg/hr. There
is a trend toward larger plants because of better efficiency and lower
capital requirements per ton of product. From 1950 to 1975, the average
plant capacity increased from 305,000 Mg/hr to 487,000 Mg/hr. 9
The cement industry is very capital Intensive and is dependent upon
the construction industry. Er ergy costs normally amount to 33-40 percent
of the total cost of cement manufacture, but may be as high as 65 percent.
Ener ’ efficiency improvements have been made since 1950 when about 7.8
million BIUs were required per ton of product. The estimate for 1979 was
6.1 million Btus per ton, and this figure has probably continued to decline
based on the historical trend. In 1972, only 39 percent of the industry’s
energy was supplied by coal, but conversion to coal was started in the
1970s. Currently, over 90 percent of the installed capacity uses coal as
the primary fuel.
Li me
Lime production in the U.S. was over 18 million Mg/hr in 1979. Lime
is used in various industries, such as steelmaking (36 percent), water
purification (8 percent), alkalies (6.5 percent), and pulp and paper manu-
facture (5.5 percent). The lime is produced by 60 companies at 155 plants
located In 41 states. These plants include a total of 450 kilns. The top
10 of the 60 companies generate about one—half of the total lime
production,
6
-------
The production of lime also requires significant energy input at an
estimated rate of 7 million Btu/ton of lime. The primary sources of energy
are coal and coke (70 percent) and natural gas (23 percent),’° The lime
kline are also obvious candidates for savings in fuel cost from the burning
of wastes as supplemental fuels.
3.2 PROCESS DESCRIPTION
3.2.1 Cement Kilns
Portland cement is produced by the controlled high—temperatures drying
of a carefully proportioned mixture of finely ground limestone, silica,
alumina, and iron oxide in a rotary kiln. The solid discharged from the
kiln is called clinker and is a chemically complex mixture of calcium
silicates, aluminates, and ferritas. This clinker is ground and blended
with 2 to 5 percent calcium sulfate (gypsum) and is sold as Portland cen ent.
The basic raw materials are generally extracted from a quarry at or
near the cement plant and are composed of limestone, cement stone, shale,
clay, sand, and iron ore. The quarry rock is crushed in primary and second-
ary crushers for size reduction and is then stored in silos prior to mixing
with other raw materials (such as clay, silica, alumina, or iron ore). The
kiln feed is prepared by proportioning, grinding and blending these raw
materials to a consistent and homogeneous composition. The composition of
typical raw materials used for cement manufacture are given in Table 3L 1
The three most prevalent types of cement processes are the wet, dry,
and dry with preheater. The oldest of the three processes is the wet
process in which the raw materials are proportioned, ground with water in a
raw mill, and then slurried with 30 to 40 percent water. In the more
recent dry process, the moisture content of the raw materials is reduced to
less than one percent before or during grinding and the dry powder (about
200 mesh) is fed directly to the kiln. Originally, the wet process offered
advantages in eas. of handling, mixing, and blending the raw materials in
slurry form, which aided in maintaining a high quality product. However,
improvements in the handling and blending of dry bulk solids, coupled with
the lower energy requirements of the dry process, have resulted in a trend
sway from the wet process. An even more recent trend toward reducing
energy requirements uses a preheater with a dry raw material feed. The
7
-------
TABLE 3 -1. TYPICAL COMPOSITIONS OF RAW MATERIALS
USED IN CEMENT MANUFACTURE 1
Material
Percent
by weight
Si0 2
A1 2 0 3
F. 3 0 3
CeO
MgO
Limestone
5.7
0.8
0.3
52.0
0.4
High silica
limestone
36.8
LB
0.6
33.6
0.5
Cement stone
18.0
5.0
1.5
40.0
2.0
Shale
53.8
18.9
7.7
3.2
2.2
Sand
70.0
15.0
5.0
0.8
0.2
Clay
61.0
16.9
12.4
0.5
0.4
B
-------
preheater uses the hot kiln exhaust gases In a countercurrent flow system
to heat and precalcine the dry feed before it enters the kiln. The approxi-
mate heat consumptions of the wet, dry, and preheater processes are 5,3,
4.6, and 3 million BTUs/ton, respectively. t
Figures 3- 1 and 3-2 are schematics of wet and dry process cement
kllns. The kiln is an inclined (3 to 6°) cylinder lined with refractory
brick with typical sizes that range from 60-750 ft in length and 6-25 ft
in diameter. 9 The raw material feed enters the upslope end of the kiln and
travels down through the various drying and reaction zones as the kiln
rotates. The first portion of the upsiope end often contains chains to aid
in heat transfer and mixing. The solids then pass through the calcining
zone where CO 2 is dissociated from calcium carbonate. The last section of
the kiln is the clinkering or sintering zone where a strongly exothermic
reaction between CaC and Si0 2 forms dicalcium silicate and where an erido-
thermic reaction between dicalcium silicate and CaO yields tricalcium
silicate, which is the primary hydraulic material In cement.
The clinker is formed at 2500 to 25000 F and then leaves the kiln and
enters the clinker cooler where air is passed through moving grates to cool
the product. The solid’s residence time varies from kiln to kiln and is
controlled by the kiln s revolutions. Typical solids residence times are
on the order of 2 (range of 1 to 4) hours. 9
The kiln is fired at the downslope end and the gases travel upslope
countercurrent to the solid’s flow. Primary air and a portion of the
preheated secondary air from the clinker cooler are used as the combustion
air. The gas residence time Is on the order of seconds and gas temperatures
typically approach a maximum of 3,000° F. Residence time calculations by
Weltzjnan show a gas residence time of over 3 seconds at temperatures of
2000° F or higher and over 8 seconds at temperatures over 15000 F.’ ’
(Estimates of residence tim. are likely to vary from kiln to kiln.)
Figure 3-3 shows the steady—state temperature and concentration pro-
files Ii a kiln based on a mathematical simulation of the process.1 2 Note
that gas temperatures (Tg) are always higher than the solids temperature
(T 5 ). The gas temperature increases (starting from the feed end) fairly
constantly; however, the solld 4 s temperature remains steady in this wet
9
-------
w.’u,
Al .
Lkss.sI.ø.
$$k.
Cl .,
ko.Os.
A l .
tS t)usI
swlv
T1 S
$$11V h4
tD
(i.hausl 0..
- -I
i i .
ILscsv...d Pius
W..Ss Dud
Tailiul
w.I,. Fuui p.oJu $S sois u
ioi...
øsi CaaI$,asd.is
l u
Figure 3-1. Wat p.ocess cement kiln 19 .
-------
sm
M
‘S I’
I—..
I aSI
Usm,usMI
Lr
I.
.êd4 1... sS
Wisis ud
M
Ci . - --
Figure 3-2. Dry process cement kiln 15 .
-------
Figure 3-3. Steady-state temperature and concsrm’ation profiles
along kiln length 12
I -
! W. LV
1
I
I
I
•1
I
- — .
12
-------
process kiln until most of the water is removed. The next significant
Increase in solid’s temperature occurs in the clinkering or sintering zone.
The most conmton types of air pollution control devices that are used
to clean the exhaust gas are electrostatic precipitators and baghouses,
often preceded by a centrifugal separator. The distribution of control
devices for wet and dry kflns are sunm arized in Table 3—2. ESPs appear to
be most coninon at wet process kilns and fabric filters are more conunon at
the dry process kllns.
3.2.2 Lime KVTns
Lime kilns are similar to cement kilns through the calcining zone of
the cement kiln. Lime is manufactured by calcining high quality limestone
(CaC0 ) at temperatures of 2400° F, which volatilizes nearly half of the
limestone’s weight as carbon dioxide (CO ) to form lime (CeO). Lime kilns
are mostly rotary kilns that range up to 500 ft in length and 17 ft in
diameter. ‘
The raw material feed includes carbonates of calcium and magnesium,
which are obtained from quarry deposits of limestone, marble, chalk, dolomite,
or oyster shells. The quarries are chosen to provide low concentrations of
silica, clay, or iron, which are impurities. The stone Is crushed and
sized and often washed to remove contaminants. The crushed and sized
quarry rock is fed to the upsiope end of the kiln and heated to liberate
CO 2 as it travels down the kiln. The lime product is removed at the down—
slope discharge end. The lime kiln is operated at temperatures (2000—2400° F)
that are somewhat lower than cement kiln temperatures.LU The high tempera-
tures and the gas residence times of lime kilns suggest that they are also
candidates for combustion of hazardous wastes.
The kiln gases are exhausted at temperatures of 500—1400° F and general—
iy require cooling before the gas Is cleaned in the control device. Typical
control devices include baghouses, ESPs, venturi scrubbers, and, occasionally,
gravel bed fllters.tU
13
-------
TABLE 3-2. CON1 ROt. OE’IICES ON CEMENT KILNSa
Process
b
Control devf cc
N
of
umber
kilns
P.
of
rcent
total
Wet
ESP
ESP + CS
FF
FF+CS
Other
73
14
27
18
11
51
9.8
18.9
12.5
7.7
Dry
ESP
ESP4CS
FF
FF • CS
Other
39
12.
42
26
13
29.3
8.4
32.].
19.9
9.9
aDerived from EPA’s National Emissions Data Base.
bESP electrostatic precipitator
CS centrifugal separator
FF fabric filter
14
-------
4.0 TEST RESULTS
4.1 OVERVIEW OF TESTS
Tests of waste combustion have been conducted at calcining kilns with
the following features:
Process Fuel Control Device
Wet Coal ESP
Dry Coke Baghouse
Preheater Fuel Oil
The types of waste have also been quite varied and include chlorinated
aliphatics, chlorinated aromatics, PC3s, chloroph.nols, phenoxy acids,
IDetal—contarninatad lubricating oil, toluene, Freon 113, alcohols, and a
wide variety of other organic compounds found in waste clv.nts. Total
chlorine loads up.to 9 kgCl per Mg clinker have been imposed on the kilns
during waste combustion.
This section will first su m iar1ze the tests to provide general back-
ground on each specific site with respect to kiln characteristics and a
general discussion of the tests. The test results are discussed collect-
ively under the various emission categories of intirest in the following
sections. A sunmiary of the tests is given in Table 4-1.
St. Lawrence Cement, Mississauga, Ontario 1 2
Tests were performed at this Canadian plant in 1974 and again in
1975_1976.2 The test tn 1974 burned waste lubricating oils contaminated
with lead and zinc and also included baseline tests without the used oil.
The used oil burn was conducted in the company’s dry process kiln with a
4-stage preheater. The kiln is 17 ft in diameter, 276 ft in length, and
Is fired with No. 6 fuel oil with exhaust gas cleaning in an ESP. A frac-
tion of the exhaust gases is bypassed through a conditioning tower where
water is sprayed Into the gas to.lower the temperature and to condition the
gases for precipitation. The cooling and velocity reduction divide the
particulate matter into two fractions: a high alkali fraction that Is
15
-------
TAut 4- I. SIMIAIY Of C(I(N1 I1I.M 11S15
— - - . ---- - -= --.— --
1! L& or a n
.
PlC. ti
Plant
let.
Pr.c.ss
APCD
Fuel
P14
P04Ks
leec h
luel lypes tested
St. lawreiwe Cement
rn..hsau e. Ontario
1914
49Th-
7
Dry-PH
Wit
ISP
ISP
Fuel oil
Fuel oil
e
a
-
K
-
i
S
K
Iid ricitlii ill contaminated with metals
ChI.rirsat.d aliphatic. (.Lhyl.n. d$chlorkk)
Chlorinated aromatic. (chhor.toiueno)
PC I. ’
Peerless Ceent
Detroit, NI
1916
Wet
ISP
Coal
-
a
-
-
-
PCI’s
Stora VIta
Sweden
3971
V.1.
ISP
Coal
a
a
K
K
-
•
Chlorinated ehiphatics (.elhyi.n. chloe -id.)
Chlorinated aromatic. (PCI 1242)
£hlor.pl.enels and pimnony acids
Irion (trichh.rotri(iuuroathane)
Marquette Cement
Oglesby. IL
*184
0.-p
ISP
Coal
K
-
K
K
Chlorin.I.d aliphatics, WI . t.lu.ne
e-’
Ub
San Juan Cement
Puerto lice
Alpha Cement
tementon. N V
190)-
82
4902
Wet
Vet
Segiseos.
ISP
fuel eli
Coal
K
-
K
-
K
K
K
a
Chlorinated ahiphitics
Waste solvent* (Halogens = 590 pp.)
General Portland
Los Subies, CA
1912
Dry
Saghous.
Coal
-
*
—
e
a
Aromatic. and chlorinated ahiphatics
General P.rtland
PauIdI q , ON
1983
Wet
ISP
Coal
n
i
K
*
K
Chlorinated atIphatIci, W I. tahseue
Ian, Star J d.
O.jl.sby, IL
4903
Dry
ISP
C.alIc. he
a
a
a
a
Chlorinated aliphetics. Hit, tome...
Socicwell Li..
8UCtWO I4, WI
1983
Cl..
Saghous.
Coke
a
-
K
a
Chlorinated eliphutics, WI, loiwen.
-------
collected in the ESP and discarded and conditioning tower solids (lower
alkali content) that is recycled into the raw material feed. Normally, the
fuel oil Is burned using three separate mechanical atomization burners.
During the waste oil burn, one of the burners was used for the waste oil
and the remaining two burners continued to burn No. 6 fuel oil. 1
A second test was started on this kiln In 1975 wIth chlorinated ali-
phatics as the supplemental fuel. A buildup of solids twice occurred in
the bypass duct between the kiln and the conditioning tower. Because of
high particulate emissions, plugging of the bypass system, and the cost of
equipment changes to correct the problems, the test of the dry kiln was
terminated and switched to on. of the wet process kllns at the site. 2
Extensiv, tests wire conducted In 1975-1976 in a wet process kiln that
was 2.23 m (402 ft) In length and 3.5 m (11.5 ft) In diameter with a
capacity of 955 Mg/day. The primary fuel is No. 6 fuel oil which is fired
through a nozzle with mechanical atomization at a flow rate of 76 L/rnin
(20 gal/mm) and a pressure of 120 psi. The kiln exhaust gases are cleaned
by a 6—section electrostatic precipitator (ESP) that collects 320 to
445 Mg/day of dust. Approximately 320 Mg/day of this dust is recycled to
the kiln. The waste types fired included chlorinated aliphatics, chloro—
toluene, and PBCs with a mixture of other chlorinated compounds. The No. 6
fuel oil is normally burned in a single burner at the center of the burner
pipe. During the tests with chlorinated compounds, the waste fuel was fed
through a separat . pipe just above and to one side of the center with
different nozzle sizes for proper atomization at different flow rates. The
waste fuel was supplied with centrifugal pumps at a pressure of 120 psig.
Nozzles plugging from solids was experienced during the PCB burn. On the
last day of the PCB burn, the nozzle was removed to permit uninterrupted
flow • 2
Peerless Cement, Detroit, Michigan
The kiln at Peerless Cement uses the wet process and Is fired by coal
with exhaust gas cleaning In an electrostatic precipitator. This kiln is
91mm (300 ft) In length and 4.9 m (16 ft) in diameter. Baseline (coal
only) tests were conducted before and after two waste burn tests of PCBs at
flow rates of 0.9 to 2.8 L/min. 1 ’
17
-------
Stora Vika, Sweden ‘
This 140—rn kiln uses the wet process and is fired by coal with gas
cleaning in an ESP. The ESP dust is extracted with water and then recycled
to the kiln with the slurry feed. A variety of chlorinated wastes were
burned and included short-term tests with chlorinated aliphatics (100 hr),
PCBs (24 hr), chiorophenols and phenoxy acids (12 hr), and 3 hours with
Freon 113 (Trichioro—trifluoroethane). A wide range of chlorine input
rates (up to 9 kgCl/Mg clinker) were used during the trial burns. Long-
term tests were conducted with chlorinated aliphatics.(1.5 months) and PCBs
(about 1 month). 3
Lone Star Industries, Oglesby, Illinois 14 ‘
Two trial burns have been conducted at this kiln: one when it was
formerly owned by Marquette Cement 14 and another under Lone Star, Inc. ’ 5
The kiln uses the dry process and is tired by a coal/coke mixture. The
kiln is 158 m (520 ft) In length, 4.6 (15 ft) in diameter, and has a
clinker production capacity of 1,320 Mg/day. An ESP collects about 205
Mg/day of dust and about 90 percent of it is recycled back to the kiln with
the raw material feed. The two tests used waste paint solvents composed of
chlorinated aliphatics, methyl ethyl ketone, and toluene In the 1983
test, the waste fuel was spiked with Freon 113 and fired through a separate
pipe under the coal pipe with pressure atomization. Total chlorine loading
ranged from Li to 2.2 kgCl/Mg clinker.’ 5
San Juan. Puerto Ricola
This cement kiln uses the wet process and is fired with No. 6 fuel
oil. The kiln is 137 m (450 ft) in length, 3 m (10 ft) In diameter, with
a clinker production rate of 28—33 Mg/hr. The kiln exhaust gas is cleaned
by a fabric filter. The waste fuel burning was part of an extensive test
program from October 1981 to February 1982 by EPA.’
The POHCs monitored during the tests included the chlorinated aliphatic
series of methylene chloride, trichioromethane, and carbon tetrachloride.
The waste fuel was fired through a separate burner pipe through different
nozzle diameters at a pressure of about 40 psig. The program Included 5
baseline tests and U. waste fuel tests on 7 different batches of waste
fuel. The waste fuel was highly chlorinated (up to 35 percent Cl) and was
burned at an average rate of 5.5 kgCllMg clinker.
18
-------
Alpha Portland Cen ent, Cementon, New York 2
This kiln uses the wet process with coal as the primary fuel and gas
cleaning in an ESP. The kiln clinker production rate is 45—49 Mg/hr. The
tests were conducted under baseline (coal only) test conditions and also
with waste solvents cofired with the coal. The em ssion testing focused on
particulate matter, HC1, 502 and metals emissions for the two test
conditions. ‘
General Portland, Los Robles, Calf fornia 18
The General Portland cement kiln uses the dry process and is normally
fired with coal. The kiln is 171 m (560 ft) In length and has a clinker
production capacity of 1,590 Mg/day. A fabric filter cleans the exhaust
gas and collects 61 Mg/day of dust. About 95 percent of this dust is
recycled back to the kiln. The test program included two baseline tests at
different times and waste burn testing with POHCs that included methylene
chloride, 1,1,1-trichloroethane, 2.,3,5—trimethylbenzene, and xylene.
General Portland, Pauldln , 0hio 1
The cement kiln in Paulding uses the wet process with coal as the
primary fuel. The normal operation of this kiln over the past 3 years has
included cofiring waste solvents as supplemental fuel. The kiln is 228 m
(420 ft) in length and 3.4 tn (U ft) in dia .meter with a clinker production
capacity of about 27 Mg/hr. A 4-field ESP is used for gas cleaning. Dust
is collected at a rate of about 6.4 Mg/hr and is recycled at a rate of 3 to
4 Mg/hr. The EPA test included 4 days of baseline testing and 5 days of
testing during the waste burn. The POHCs ‘In the waste fuel included methy-
lene chloride, 1,1,1-trlchloroethane, methyl ethyl ketorie, toluene, and
Freon 113, which was spiked Into the waste fuel. The waste fuel was fired
through a separate burner pipe inside the coal burner pipe with air atomi-
zation of the solvents. 19
Rockwell Lime, Rockwood, Wisconsin 20
This lime kiln Is 2.4 m (8 ft.) in diameter and 67.1 m (220 ft) in
length and Is fired normally by a mixture of coke and natural gas. Exhaust
gas is cleaned by a fabric filter. The lime product is produced at a rate
of about 7.7 Mg/hr. During the waste burn, the waste fuel replaced natural
gas and was fed through a separate 1-inch pipe above the coke burner pipe
with its nozzle pointed into the flame. The POHCs in this test were chlo-
rinated aiphatics, methyl ethyl ketone, and toluene. 2 °
19
-------
4.2 DESTRUCTION AND REMOVAL EFFICIENCIES (DRE) FOR PRINCIPAL ORGANIC
HAZARDOUS CONSTITUENTS (PCHCs)
POHCs are the hazardous constituents in the waste as defined by EPA ’s
Appendix VIII list and are selected on a case-by-case basis. Two criteria
that are used Include the expected difficulty of thermal degradation of the
compounds and their concentrations in the waste. 2 ORE is a simple efficiency
calculation based on the quantity of a compound entering the kiln with the
waste and the amount emitted through the stack after thermal destruction
and gas treatment in the control device. Generally, no corrections were
made In the test reports summarized here for emissions of the POHC when
waste is not burned (referred to as the baseline condition in this report)
or for POHCs that may enter by Inward leakage of ambient air or from the
process water.
The concentration of the compound in the waste is an important consider-
ation in the selection of POI4Ca because this concentration, the waste and
air flow rate, and the detection limit of the test method determine the
number of nines that can be demonstrated in the ORE calculation. Because
very low (ppb) l*v ls must be measured in the gas, EPA has developed a -
volatile organic sampling train (yOST) to collect nanogram (1O grams)
quantities of volatile organics for analysis. From a 20—liter sample,
roughly one gram—mol of stack gas is sampled (22.4 L/g-mol at standard
conditions). For a total sample of 100 ng of a POHC with a molecular
weight of 100, the VOST should be able to detect levels of 1 ppb by volume.
If collection levels of 10 ng are distinguishable from blanks, then detection
levels on the order of 0.1 ppb ar, possible.
Obviously, the VOST Is designed to provide very low detection levels;
however, contamination is introduced as an important factor that must be
considered in evaluating the test data. Scrupulous care must be exercised
to minimize contamination in this System In which nanograms of contaminants
can create a high bias in test results. The Tenax sorbent that is used is
notorious in its ability to adsorb trace quantities of organics from the
air (e.g. chlorinated organics from new carpet or several types of Freon
from refrigerated storage), even when the sarbent is In a “sealed’ container.
However, d.sorption or loss of Compounds from the Tenax in sealed tubes has
20
-------
not been documented as a problem. These factors imply that the V05T results
may be biased high fro n contaminants, but there is no Indication that the
POHCs are lost by desorptlon In the time from sample collection to analysis.
The collection of blanks and ba3elir e samples In the cement kiln tests have
been a necessary and invaluable tool In sorting cut the results of POHC
emission rates.
Low levels of contamination, POHCs in the ambient air, and POHCs found
during baseline tests do not preclude obtaining useful and defensible
results from the yOST. For example, environmentally conservative estimates
of POHC concentrations can be made by assuming that all detected levels of
POHCs originate from the waste and by ignoring corrections for baseline
but the actual ORE of POHCs entering only with the waste fuel may be higher.
The ORE results are su mnar zed in Table 4-2. One of the first tests
to exainj ne the ORE of hazardous waste in cement ki 1 ns was conducted at the
Canadian plant, St. Lawrence Cement. The wet-process kiln was normally
fired with fuel oil and the exhaust gas was cleaned by an ESP. Three tests
were conducted with batches of waste solvents labeled as chlorinated alipha-
tics, chlorinated aromatics, and PCSs. The chlorinated “aliphatics” were
composed primarily of chloropropane, chioropropene, 1,2—dichioroethane,
1,1,2-trIchioroethan., and chlorobenzene. The chlorinated aromatics batch
was primarily chlorotoluene, and the waste labeled PCBs was about 50 percent
PCB, 28.5 percent chiorotoluene, and a mixture of chlorinated aliphatics.
Destruction efficiencies were calculated conservatively by not subtracting
or correcting for the background levels from the baseline test or Inter-
ferences (contamination) on the control blanks. In addition, a conservative
levels of contamination. If the concentration of POHC in the waste is
sufficiently high and the mass collected on the sample tube Is low, even if
the detectable levels are caused by contamination or some other source,
OREs of 99.99 percent or greater can be demonstrated. Some of the data
suemarized in this report falls into this category in which either low—level
contamination or baseline emissions of the POHC were detected and the
results still showed 99.99 percent ORE. These data will be ldent.ifled and
the reader should keep in mind that for these cases, ORE calculations are
well bounded on the lower end and unbounded on the upper end. In other
words, the OREs for these test results should be at least the measured ORE,
21
-------
TABLE 4-2. SU ARY OF OREs
Plant Waste component Oestructf on efficiency
St. Lawrence Cement. Chlorinated al(phatics >99.990
Chlorinated aromatics >99.989
PCB’s >99.986
Peerless Cement PCB’s 99.9981-99.9986
Stora Vika M.thylene chloride >99.995
In chl oroethyl en. >99.9998
All chlorinated hydrocarbons >99.988
PCB ‘99. 99998
Chlorinated phenols >99. 99999
Phenoxy acIds >99. 99998
Freon 113 >99.99986
San Juan Cement Methylene chloride 93.292 -99.997
Trlchlorom.tharte 92.171-99. 96
Carbon tetrachiorld. 91.043-99.996
Los Robles Methylen. chloride >99.99
1,1,1-Tnichlorotthane 99.99
1,3,5—Tnirnethylbenz.n. >99.95
Xylene >99.99
Paulding Methylene chloride 99.956—99.998
Freon 113 >99. 999
Methyl ethyl ketone 99.978—99.997
1,1]. -T nichloro.thana 99.991-99.999
Toluene 99.940-99.988
Oglesby Methylene chloride 99.94-99.99
Freon 3.13 99.999
Methyl ethyl ketone 99.997—99.999
1 1,1-Tn ich loroethane >99.999
Toluen. 99.986-99.998
Rockwell Lime Methylene chloride 99.9947-99.9995
Methyl ethyl ketone 99.9992-99.9997
1,1,1-Trichioroethane 99. 9955-99. 9982
Tnichloroethylene 99.997—99.9999
Tetrachioroethylene 99.997-99.9999
Tol acne 99.995—99.998
22
-------
collection efficiency of 80 percent was used. The result was a maximum
estimated concentration of 50 ppb for all of the waste fuel tests, and
“because of the high and uncertain background levels, the estimate Is
higher than the actual levels in the emissions.” The OREs were based on
total chlorinated organics In and out instead of analysis of specific
compounds in and out. The reported DREs for wastes with mostly chlorinated
organics was >99.99 percent, >99.989 percent for chlorinated aromatIcs, and
>99.986 percent for PCB mixture. 2
The tests at Peerless Cement used PC8s as the waste fuel at feed rates
of roughly 1 and 3 qt/mln. The rssu ts are suavnarized in Table 4-3 and
show OREs of 99.9983. to 99.9986 percent. These estimates are also conser-
vative because PCB emissions were also present during the baseline test.
The river water used to make the slurry feed to the kiln contained PCBs and
introduced the PCBs into the cooler (feed end) of the kiln. The test
results showed no significant difference between baseline tests and tests
in which PCBs were burned. An Interesting observation in this test is that
some of the PCSs entering the cooler end of the kiln are destroyed or
removed at some point In the process other than with the stack gas. Stack
emissions of PCBs were only 7 to 54 percent of the PCB entering from the
river water alone.’ 3
A test similar to the one conducted at St. Lawrence was conducted in
Sweden at a wet process kiln In Stora Vika. Short-term tests were conducted
with chlorinated aliphatics (100 hr), PCBs (24 hr), chlorophenols and
phenoxy acids (12 hr) and trichlorotrlfluoroethane (3 hr). Long-term
tests were conducted with the chlorinated organIcs (1.5 months) and PCBs
(about 1 month). Results are listed in Table 4—2. During both the short-
and long—term tists with chlorinated aliphatics, none of the waste fuels
major components was detected in the stack gas. Based on the detection
limit, the ORE of methylene chloride exceeded 99. 995 percent and the ORE of
trlchloroethylene exceeded 99. 9998 percent. 3
No PC8s were detected in the stack gas during either the short- or
long—term test and yielded a ORE exceeding 99.99998 percent for PCB.
Similarly, no chloropheno)s or phenoxy acids were detected In the stack gas
with resulting OREs greater than 99.99999 and 99.99998 percent, respectively.
Detectable levels of trlchlorotrifluoroethane (Freon 113) wer, found during
23
-------
aBased only on PCB In fuel
water.
TABLE 4—3. P 8 TEST RESULTS FOR PEERLESS CEMENT
Run number
-
1
3
4.
5
PCB in
fuel (pg/mm)
0
5.56 1O
1.7 x iO
0
PCB out
( igJmin)
5,864
7,763
33,212
44,930
PCS in
feed (pg/mm)
8.3 x 10
6.0 x jO
7.2 x 1O
8.3 x 1O
DREa
- 99.9986 99.9981
with no correction for baseline or for
PCBs In
24
-------
one experiment with this waste fuel and yielded a ORE of 99.99986
percent.’ ‘
The test at Los Robles on a dry process kiln Included dichloromethane,
1,1,1—trichloroethane, 1,3,5—trirnethvlbenzene, and xvlene in the waste fuel
The ORE results are listed in Table 42. The results show a ORE of 99.99
percent or greater for 3 of the 4 compounds, and again the calculations are
conservative. No corrections were made for baseline levels or far the
contribution from ambient air. No trim.thylbenzene was detected; therefore,
the ORE for this compound Is likely to have exceeded 99.95 percint.’
The test at Paulding, Ohio on a wet-process kiln tracked the ORE of
compounds listed in Table 4-2. Again, all of the results are conservative
and in this case, no corrctions were made for blank contamination, baseline
levels, ambient air In leakage, or the contribution from process water.
Mathyleni chloride contamination was a problem and the OREs for this compound
should be viewed as unrellably low because of the contamination. Similarly,
the methyl ethyl ketone results reflect a contamination problem, although
on a scale much smaller than the methylene chloride. No problems with
contaminants were seen with the 1.3,1 -trichioroethane end Freon 113 results
which demonstrated OREs of 99.999 percent of greater.
The toluene emissions were shown to originate from coal combustion.
No blank problems were experienced with toluene, baseline and waste burn
emissions of toluene were the same, and the highest toluene rates occurred
during a kiln upset at the baseline conditions. Benzene emission rates
were also about the same for the baseline and waste fuel burn. These
results were also observed during a baseline test at Los Robles with coal
fuel in which both benz.ne and to7uen. wer, found at concentrations similar
to those found at Paulding. Therefore, the calculated ORE for toluena is
misleading In terms of waste destruction and a more important consideration
Is that toluene emissions did not Increas. from burning wastes that contained
toluene. Even with low but steady baseline level of toluene, a ORE of
99.99 percent was reported on one test day in which the highest level of
toluene entered the kiln with the waste fusi. This does not mean that an
increased input rate increases the ORE; for this test, the higher ORE
derives only from the mathematics of a constant small emission rate and a
significantly increased Input rate.
25
-------
The results of the test burn at Lone Star showed similar contamination
problems with methytene chloride. However, the OREs for other compounds
were all 99.99 percent or greater on an environmentally conservative bisis,
i.e., no correction for blanks, baseline, process water, or outside air.
These results are based on a limited number of VOST runs because one set of
samples was invalidated because of contamination. Several compounds that
were not predesignatad as POHCs were quantified from the Modified Method 5
sampling train. These compounds are listed in Table 4-4 and show OREs
exceeding 99.99 percent. 5
The test at Rockwell Lime burned waste fuel containing nethylene
chloride, methyl ethyl ketone, 1,1,1,-trichloroethane, trlchloroethylene,
tetrachlaroethylene, and toluena. The average ORE results are listed in
Table 4-2 and show an average of 99.99 percent or greater. The reported
OREs are again conservative because no blank corrections were applied.
OREs exceeding 99.99 percent were consistently obtained for all PQHCs
except methylene chloride. Three out of 28 runs averaged less than 99.99
percent for methylene chloride, but the quantity collected suggested contam-
ination problems. The overall average ORE for methylene chloride was
99.997 percent. 2 °
The test at San Juan used methylene chloride, chloroform, and carbon
tetrachioride as the designated POHCs In a waste fuel with a high chlorine
content (5.5- 35.l. percent). This test showed a phenomenon seen at other
cement kiln tests: measurable rates of the POHCs during the baseline test
when no waste was burned. Blank samples showed no contamination problems;
therefore., the source of the POHCs during the baseline was unexplainable.
The reported OREs for this test appear to be low compared with other test
results. The test report concluded that lack of atomization of the waste
fuel and the difficult incinerability of highly chlorinated .onocarbon
compounds contributed to the low OREs. The above normal fret lime content
of the clinker and removal of chloride In the clinker instead of the waste
dust also suggest that operating difficulties were experienced. However,
the detection of POHCs during the baseline mak, the DRE results difficult
to Interpret. If the measured POHCs originated from sources other than
burning waste fuel, the actual OREs nay have been higher than those
measured. lb
26
-------
TABLE 4-4. OREs OF COMPOUNDS FROM 11145 TRAIN
AT LONE STAR
Compound
concentration in
waste (gIL)
Stack emissions
(pig/s)
Percent ORE
Styrene
12.5
<30
>99.999
Ehtylb.nzene
11.4
<30
‘99.999
O—Xylene
15.0
<30
>99.999
Benzaldehyde
18.7
236
99.998
27
-------
The San Juan test, when viewed with results from other cement kiln
tests, suggests that burner design, waste composition, and the kiln’s
operation can have a significant effect on ORE. The extent of their effect
is inconclusive because of baseline POHC emissions.
4.3 POHC EMISSION RATES
POHC emission rates are suimnarized in Table 4-5. Because of baseline,
background, and blank results that show the POKCs, these emission rates
should be viewed with caution. None of the reported wasta burn rates
except the San Juan test were corrected for baseline emissions. The test
at San Juan showed Increases in POHC emission rates over th. baseline for
all three compounds. The test report noted that other baseline tests not
Included in he baseline total, which were conducted on days following a
waste fuel burn, showed levels of POHCs equal to or exceeding the levels
during the actual waste burn.la Note that the baseline emission rates at
San Juan are higher than the POHC emission rates during waste combustion at
other kilns. The San Juan results are difficult to interpret; however, an
environmentally conservative approach would suggest significantly higher
rates during the waste burn at levels exceeding those observed in any other
tests.
The tests at Paulding showed no significant difference between baseline
and waste burn emissions of PCHCs. Note that some contamination problems
were experienced with mathylene chloride and methyl ethyl ketone; therefore,
actual emissions of these compounds were probably lower than those reported.
Coal combustion was shown to contribute to toluene emissions.
The tests at Lone Star also showed methylene chloride contamination
problems and low level contamination by Freon 113. Again, no significant
difference was found between baseline and waste burn emission rates. The
overall emission rates of POHCs are comparable (same order of magnitude) to
those observed at Paulding.IS
During the test at Store Vika, the only compounds detected during any
tests were chloroform (during the chlorinated aliphatics burn and traces
during a baseline test) and Freon during that trial burn. Generally,
emission rates (based on detection limits) are low and comparable to the
28
-------
TABLE 4—5. SU !ARY OF POHC V4ISSIO RATES
Bisalinu 4ista burn
iuions I.lis ion s
PONCi (aq/s) (/s)
San Juan’ 4.thyl•ns ctilari0e 1.4 94 (1..3 to 310)
CT 1or ofor. 1 1 94 (‘9.3 to 400)
Careen tetrachloridi 11 191 (C. 12 to 400)
Paulding Msthyleri* ct loride
Freon 113 <0.022 ‘0.025
1.1,1—TrIeh loroetjt.ne 0.044 0. 3.49
$etJ yl ethyl kitone O.1 O.5
Toluane .r 4.1
Lone Star Msthylsne c ilerida 4.e
Freon 1.13 0.43’ 0.075
1.1.1•Irich)oro.tp,a ns 0.023 (0.01.4
Methyl ethyl ketona 0.1 0.14.. ,
Foluens 0.36
Store Viko Methylsvie chlor ide <7.3
TPiCfllGrOethyl.n. <0.3
Chlareforic • 1.0
Ptt’s <0.04
chisreohenols 0.025
PP4AiXy acids • <0.03
Freon 11.3 0.03
Pee,l. s Caent PC3’s 0.1.0 - 0.75 0.3.3
Los Robles Methylene chlorIde <0. 1. - 0.43 0.003
L,1. l l lchlor o. t han. 0.1.3 - 0.34 0.005
3..3,5—l’rfeetiiyl benzeni <0.1. - <7.5 ‘042
kyle , ,. <0.1. <351
I.nzone 35 - 54 76
Rockwell Li .. Methylene chlorine 0.0006 0.0023
Methyl .VIyl ketons 0.007 0.0105
1 .,1.,3. -Trichlores t han 0.0048 0.0073
Tricnlere.tnyl.n, 0.0027 0.0032
Tetracliloroetnylen, 0.0023 0.0043
lolusni 0.03.57 0. 0332
1*15.1 in. results based on reoort.c average concentration inc an average gas
flow of 1,600 e / .In. WISt. burn results averaged free r 1. 3.3 In lie tist
reoort inc were carrectec for basalt ne
ts are susoact because of eotny ens chi on di contmi nation on b I anti.
CBlInk values for MIX were roughly coiC arlOl. to sae le values; results are
5U8Q t bec*use of est.ntf a! contielnation.
loluen. and enz.ne sI own to ceee fre, coal coe,ust1 .
cOntielnatien of Freon 1.1.3 seon en blinkg.
of the rg coa euMs excest cfllorefere was found during the basiflne
tast. £sisui n estimates are basic en dSts tf on Hilt val ss end eui a
‘wista fee’ rates (exceOt for seasuped vilijis for cnlorcfore and Freon).
luissions could be attr4b ta to PCIs In the slurry watar.
29
-------
two preceding tests. 3 ‘ The PCS burn at Peerless Cement showed tow levels
of PCS emissions during both the baseline and PCB burn. These emissions
wer. attributed to PCBs that entered with the river water in the slurry
feed, which enters the cooler (feed) end of the kiln. 3
The POHCs that were followed at Los Rabies showed no significant
difference between baseline and waste fuel tests. B.nzene was found to be
formed from coal combustion and cannot be fully attributed to waste
combustion. 15
The lowest POHC emission rates of any tests were found at Rockwell
Lime. A fortunat. aspect of this test was the lack of contamination pro-
blems and relatively clean blanks. (The only problems reported were methyl-
ene chloride contamination on 3 of 28 runs.) The methylene chloride and
toluene Increases over the baseline were found to be statistically signif i-
cant. The differences between baseline and waste burn emission rates for
the other POHCs were not significant. The emission rate results for Rockwell
Lime are one to two orders of magnitude lower than the results of other
tests and are several orders of magnitude lower than the San Juan results. 20
4.4 PROOUCTS OF INCOMPLETE COMBUSTION (PICs)
The burning of complex mixtures of organic compounds can obviously
yield emissions of compounds other than the PCHCs. Several tests at kllns
have attempted to identify and quantify both volatile (boiling point<100° C)
and semivolatile organic compounds that are emitted under baseline and
waste fuel test conditions. The baseline results have been particularly
Interesting because of the byproducts formed from coal combustion. As with
the POHCs, the interpretation of the results of waste combustion on PICs
are confounded somewhat by the presence of many of the same compounds
during baseline tests and the potential for high bias from low level contami-
nation or background levels.
Table 4-6 lists compounds found during the waste burn at San Juan
Cement that were not found during the baseline and were attributed to waste
combustion. Samples were also analyzed for polychiarinated dtbenzodloxins
and none were found in any sample. Trace quantities of polychiorinatad
dibenzofurans (Cl 5 , Cl 4 , C l 7 isomers) were detected in the particulate
catch from one run. However, the kiln was in an upset condition during
30
-------
TABLE 4—6. pICsa FROM WASTE COMBUSTION AT SAN JUAN’S
Compound ag/hr
Benzaldehyde 10 - 18
Phenol 2.4 - 12.
C 3 - alkylbenzene 3
1—phenylethanone 6.1
Benzoic acid 3.5 - 48
C 2 -Benzoic acid, methyl ester 6.3
C 2 -Naphthane isomers 10 — 50
C 3 -Naphthane isomers 14 - 46
Anthracan. 1.5 4.0
aNO detected during the baseline and considered attribut-
able to waste combustion..
31
-------
this run and the results Cannot be considered representative of waste
combustion. No polychiorinated dibenzofurans were detected in any other
tests.
The test results from Los Robles provided more insight into PICa from
coal than PICs from the waste fuel. During the baseline tests, detectable
quantities of benzene (120—530 ppb) and toluene (20—70 ppb) were found.
Trace quantities (but above background or blank levels) of methylen. chloride
(0.2 ppb) and trichloroethane (0.4 ppb) wire also found during th. baseline
test.
The test at General Portland’s kiln in Paulding also revealed more
about byproducts from coal combustion than from waste fuel combustion. The
compounds detected and emission rates are given In Table 4—7 for the baseline
tests with coal as the only fuel. During the first baseline test, the kiln
was in an upset condition and coal combustion was far from optimum. The
compounds marked in Table 4-7 were found at concentrations during this
upset condition that wire significantly higher than the concentrations
found during the other three baseline tests. Note that concentrations of
toluane (-30 ppb) and benzene (—290 ppb) are comparable to the concentrations
found during the baseline tests (coal fuel) at Los Robles. The compounds
listed in Table 4-7 were found during both the baseline and waste fuel burn
and no Increase could be attributed to waste fuel combustion. The stack
gas was also analyzed for pclychlorinated dibenzodioxins and dibenzofurar.s.
None were found at a detection limit of about 0.2 ppb by weight. i I
During the baseline test at Lone Star with coal and coke as the primary
fuel, some tolueni (L8 rag/a) and traces of C 3 ’, C 4 -, and C 5 -benzenes were
found. Emission rates of biphenyl, benzaldehyde, naphthalen., and methyl-
rsaphthalenes ranged from 0.02 to 0.3 mg/s during the baseline. A small
Increase In emission rates was noted for a few of these compounds during
the waste burn, but the overall emission rates remained In the range of
0.03 to 0.5 mg/s. No polychiorinated dibenzodioxins or dibenzofurans were
found In the stack gas at a detection limit of about 0.9 ppb by weight. ‘
The majority of chlorinated organic compounds studied in the Store
Vika test were not detected in the stack gas. During an experiment with
chlorinated aliphatics, some chloroform was detected (340 . g/m 3 or about
10 mg/a) at levels above that found in the background.
32
-------
TABLE 4-7. POTENTIAL PICs FROM COAL COMBUSTION
AT THE PAULDING TEST 19
Compound Emissions (mgi ’s)
Methyl ethyl k. OflCa 0.9
Toiu.nea 35
Tetrachioroethylene 0.8
Benz.nea 36
2.9
C 3 -Benzene 5.8
C 4 -Benzene 5.2
C 5 -Benzene 0.7
Biphenyl 5 2.5
Naphthalen.sa 12.3
Methylnaphthalenes’ 8.3
1 Thesa compounds exhibited emission rates during one base-
line test with a kiln upset that were much higher than the
emission rates during three subsequent baseline tests.
33
-------
The lo chloroform content of the waste entering the kiln indicated
that chloroform formation was a possibility. During the burning of chioro-
ph.nols/phSr’OXY acids at Stora Voka, the emissions were analyzed for poly-
chlorinated dlbanzodIoxir%s and dibenzofurans. None of the isomers with four
to six chlorine atoms, which are the most poisonous, were found. In one of
the .xp.rimsnts, a GC peak showed the same retention time as the Cl and
Cl isomers of the dioxin and the Cl 5 isomer of the furan. However, the
concentrations of the substances were too low for GC/MS analysis and the
compounds’ identities could not be confirmed. 3
The test results for PICs have shown som. minor increases from waste
combustion for some tests (several compounds at San Juan and chloroform at
Stora Vika). The results for coal combustion appear to be more conclusive
and indicate that many of the compounds are byproducts of coal combustion.
Polychiorinated dibenzodioxins and dibenzofurans have not been confirmed as
PICs from waste combustion. Trace quantities were found at San Juan during
kiln upset, and trace quantities have been present when chiorophenols
and phenoxyacidS were burned at Stora Vika. Tests at two other kilns t5 19
and most of the analyses at San Juan and Stora Vika revealed no detectable
quantities of these compounds.
4.5 EFFECTS OH PARTICULATE MATTER EMISSIONS
The effect of waste combustion on particulat, matter emissions has
been of interest because the earlier cement kiln tests indicated that
burning chlorinated wastes increased emissions. A review by Weitzman
examined the particulate matter results when chlorinated wastes were burned
at St. Lawrence Cement and Stora Vika. 1 He reported a positive correlation
between chlorine input and particulate matter out of the stack, I.e., an
Increase In chlorine increased particulate emissions. Subsequent tests
Indicate that this conclusion may sometimes be true for kilns equipped with
ESPs and Is not necessarily true for kllns equipped with fabric filters
(baghouses).
An effect on particulat. emissions was predicted because chlorinated
waste combustion forms HC1,which reacts with potassium, sodium, calcium,
and other compounds in the kiln to form the respective chlorides. These
lcom ounds volatilize at the te meratures found at the hotter end of the kiln
34
-------
and are carried to the cooler end by the hot gas. 11 Condensation in the
gas phase can form a fine particulate that remains entrained In the gas.
Alternatively, the compounds may condense on the kiln’s walls or on particles
and form a ring of solids. Ring formation may occur as repeated accumulation
and loss of chlorides, or in the extreme case, as a continuing buildup
which results in a large ring that Impedes or stops production.
The first test at St. Lawrence Cement showed that burning waste oil
contaminated with 6,000 ppm lead actually decreased particulate matter
emissions. Baseline emissions of 1.]. lb/ton were reduced to 0.7 lb/ton
during the oil burn. The test report authors suggested that the increased
amount of water entering with the oil Improved the ESP collection efficiency. 1
The test at Alpha Cement, which burned waste solvents containing 2,000 ppm
lead and very low levels of chlorine, showed similar results. Baseline
emissions of Li lb/ton decreased to 0.8 lb/ton when the solvent was burned. 12
The control devic e for both of these kllns was an ESP. The primary fuel at
St. Lawrence was fuel oil and at Alpha Cement the primary fuel was coal.
The results from these two tests (shown in Table 4-8) suggest that burning
wastes low in chlorine content may actually decrease emissions.
The tests at San Juanla and Rockwell Line 20 included chlorinated
organics in the wastes at average levels of 5.5 and 2.7 kgCl/Mg product,
respectively. The control device for these kilns was a fabric filter
(baghouse). The results in Table 4-8 show no significant difference in
emission rates between the baseline and waste fuel burn. Emissions ware
will controlled with average rates of about 0.65 and 0.25 lb/ton of product
for the cement kiln and lime kiln, respectively. These test results Indicate
that burning chlorinated wastes does not necessarily increase emissions.
No samples wer, taken to determine If the loading on the fabric filters
Increased. If the loading did Increase, the baghouses were evidently
adequate to remove any increased loading of particles for these two cases.
(Note: The test at San Juan may not have imposed the full potential loading
of chloride-laden dust on the baghouse from 5.5 kgCl/Mg clinker. A signifi-
cant portion of the chlorid, was removed with th. clinker and not with the
waste dust.)
The test at Paulding was on a cement kiln with gas cleaning in an ESP.
The waste contained chlorinated organics that entered the kiln at an average
rate of 2.2 kgCl/Mg clinker (a range of 0.8 to 4.7 kg/Mg). The burning of
35
-------
TABLE 4-8. EMISSIONS OF -PARTICULATE MATTER
Particulate
Plant Test condition gr/SCF lb/h
matter
lb/t
Cl (kg/Mg
clinker)
into kiln
St. Lawrence
Chlorinated allphatics
0.21
123 a
3 a
4.0
Chlorinated eromatics
0.086
45
1.2.
5.5
PCB’s
0.078
44
1.1
2.5
Baseline
0.038
21
0.5
-
Lubricating oil
0.064
83
0.7
Baseline
0.107
139
Li
Marquette
Waste solvents
BaselIne
0.104
0.093
58
50
<1
<0.8
1.].
Alpha Cement
Solvents
Baseline
0.041
0.050
44
53
o.s
1.].
San Juan
Wastes
Baseline
0.043
0.041
22.4
21.7
0.66
0.64
5.5
Paulding
Wastes
BaselIne
0.030
0.030
18.9
19.6
0.55
0.64
2.2
0.2
Lone Star
wastes
Baseline
b
0.17
b
116
b
2.0
1.2
0.2
Rockwell Lime
Wastes
Baseline
0.015
0.013
2.2
2.0
0.26
0.24
2.7
Stora Vika
Allphatics
Baseline
PCBs
Baseline
Chiorophenols &
phenoxyacids
Baseline
0.039
0.009
0.024
o.ou
0.058
0.014
21
4.7
12.7
5.9
30.9
7.7
0.88
0.21
0.53
0.25
1.36
0.34
4.4
0
3.6
0
0.95
0
Freon 113
Baseline
0.062
0.022
33.3
11.7
1.39
0.49
1.7
0
1 Rlng formation and ESP difficulties.
bESP sal functioned.
36
-------
waste solvents at this kiln had been routine practice; consequently, the
ESP had been adjusted for chlorinated waste combustion and the kiln’s
operators were experienced In cofiring waste solvents. No significant
difference was found in particulate matter emissions between the baseline
and waste fuel tests (—0.55 lb/ton).’ Similarly, chlorinated wastes were
burned during the Marquette Cement tests with an average chlorine input of
1.1 kg/Mg clinker. This kiln also had an ESP control device, and no signif i-
cant difference was found in particulate matter emissions between the
baseline and waste futi burn. The data from these tests indicate that
chlorinated wastes can be burned in a kiln with a properly operating ESP
without a significant change In particulate matter emissions) 4
More details on the potential effects of chlorinated waste combustion
can be found by examining the test data collected at Store Vika 4 ‘ and St.
Lawrence Cement.S Different types of chlorinated wastes were used and the
total chloride loading was varied among tests. One kiln was fired with
fuel oil, the other was fired with coal, and both hive an electrostatic
precipitator.
The St. Lawrence results summarized in Table 4-9 show baseline emission
rates of about 0.24 kg/Mg (0.48 lb/ton). When chlorinated aliphatics were
burned, problems were experienced with chloride accumulation and release,
which finally resulted in a kiln blockage on the last day of the test.
Emissions Increased to about 1 kg/Mg (2.0 lb/ton) for a level of 34 kgCl/Mg
clinker. The highest rate (6-7 kg/Mg) resulted in a kiln blockage and
upset conditions (4.5 lb/ton) followed by a kiln shutdown to remove the
ring. Th. burning of chlorinated aromatics (primarily chiorotolusne) also
yielded results (1 -1.3 lb/ton) that were higher than the baseline rate. A
ring also formed during this test but eventually broke away. The chloride
input was roughly 4.5 to 6 kg/Mg clinker. During these two tests, a tendency
toward ring formation was found in the range of 6-7 kgCl/Mg clinker. The
authors noted that ring formation compounds th. effect of increased total
loading on the ESP because rings decrease the flow area, which increases
the velocity and in turn increases the quantity of dust entrained by the
gas.
Emissions also Increased over the baseline results when PCBs were
burned, but the data do not show a significant increase In emissions as the
37
-------
TABLE 4-9. SUMMARY OF PARTICULATE MATTER RESULTS AT ST. LAWRENCE CEMENT 2
Test
condition
Date
Particulate
(lb/hr)
Chlorine 1 a
(lb/24 hr)
Clinker out
(tons/day)
kg Cl/Mg
Clinker
kg
Mg
particles!
clinker
Calculateda
b
Reported
Baseline
10/20/75
10/21/75
10/22/75
1/19176
1/20/76
1/21/76
20.8
25.4
19.6
23.9
21.0
14.6
1,516
1,342
1,412
1,361
1,389
1,445
1,071
1,078
1,046
1,020
991
1,016
.71
.62
.67
.67
.70
.71
-
-
-
-
-
-
0.23
0.28
022
0.28
0.25
0.12
Aliphatics
10/28/75
10/29/75
10/30/75
83.9
84.6
200
8,146
8,604
13 999
1,040
1,021
1,032
3.9
4.2
6.8
3.1
3.2
6.3
1.0
1.0
2 . 3 c
Aromatics
12/10/75
12/11/75
12/12/75
40.3
40.5
55.0
11,791
9,062
12,016
994
999
1,024
5.9
4.5
5.9
4.5 — 7.1
3.1 - 5.1
7.9
0.49
0.49
0.64
PCBs
1/7/76
1/8/76
1/9/76
45.3
34.5
52.2
3,999
6,413
12 , 348 d
944
1,045
1,027
2.1
2.6
6.0
0.6 - 1.4
1.3 - 3.3
6.1
0.58
0.40
0.61
bvaiues reported by authors in summary, apparently for chlorine from waste fuel. Ranges reflect variations in
flow rate within the test day.
CRing formation in kiln terminated the test.
dwt was burned 24 hours per day except for this test day which was only 735 minutes. Chlorine in was
adjusted for waste fuel to reflect a 24-hour rate.
cc
aChiorine in and calculated
fuel oil, and waste fuel.
values from the chlorine balance in
These are 24 hour averages.
the test report.
Included chlorine in slurry,
-------
chlorine loading increased from about 2 to 6 kgCl/Mg clinker. The data
show that the combustion of chlorinated aliphatics increases emissions for
chlorine levels of 2-7 kgCl/Mg clinker. The results for chlorotoluane and
PCS show a two—fold increase over baseline for a wide range of chloride
loadings (2.1 to 6 kg/Mg). The aliphatics apparently caused a four—fold
increase with chlorid, loadings of 3-4 kg/Cl/Mg clinker. It is likely that
the Increased emissions are at least partially attributable to a change in
the dust resistivity from an Increased alkali content. The test report
notes that In the design of the precipitator could compensate for this
change. 2
The emission rate5 for St. Lawrence (neglecting the kiln upset) are
not significantly different froni the overall performance of other kilns.
Mote that a baseline test on a dry kiln at St. Lawrence yielded an emission
rate (1.1 lb/ton) that w s about the same as those from the wet kiln burning
chlorinated wastes (1.1, 1.1, and 2 lb/ton). The particulate emission
rates at St. Lawrence did not significantly add to the suspended particulate
matter In the ambient air around the plant.
The Stora Vika tests were also conducted with various types of wastes
and varying levels of chloride input. Short-term tests were conducted with
chlorinated aliphatics (100 hr), PCBs (24 hr), chiorophenols and phenoxy
acids (12 hr) and trichiarotrifluoroethane (3 hr). Long-term tests were
conducted with the chlorinated allphatics (1.5 months) and PC8s (about 1
month). Results for particulate matter are sunrarized In Table 4-10,
The baseline tests at Store Vika yielded results that ranged from 0.10
to 0.24 kg/Mg- (0.2-0.4.8 lb/ton) and indicates a fairly well-controlled kiln
when compared with the emissions dat.a In Table 4-B. The burning of chlor-
inated aliphatics with chloride Inputs of about 4 to 8 kg/Mg clinker Increased
emissions by a factor of 3 to 4 to 0.4-0.5 kg/Mg (0.8 to 1.0 lb/ton). The
burning of PCSs also Increased emissions; however, the increase was not
significant for chloride loadings of about 3 kg/Mg and showed an increase
comparable to the aliphatlcs (1 lb/ton) for a chloride loading of 5.6 kgCl/Mg
clinker. 3
The results for the phenoxy acIds/chlarophenols arid Freon are less
easily explained in terms of chloride. Significant increases (to about
L4 lb/ton) were observed for both types of waste fuel. However, the
39
-------
TABLE 4-10. PARTICIJIATE MATTER RESULTS FOR STORA V1K.A 3 ‘
Rate Chlorine In Clinker out kg Cl per Particulate Ea lss lons
Waste fuel (1/mm) (kg/br) (Hg/br) Mg clinker (kg/br) (kg/Mg)
Chlorinated 395 85 21.6 3.9 9.4 0.44
allpbat lcsd 580 125 21.5 5.8 11.1 0.52
580 125 21.8 5.1 8.1 0.37
Baseline test -- -- - - - - 2.1 0.10
PCBsb 360 68 21.6 3.1 4.0 0.19
640 121 21.6 5.6 10.8 0.50
360 68 21.6 3.1 2.6 0.12
Baseline test 2.7 0.13
Phenoxy ac1ds 980 21.6 20.7 1.0 15.3 0.74
and chioro- 980 21.6 20.7 1.0 14.1 0.68
phenols 980 21.6 20.1 1.0 12.8 0.62
Baseline test - - -- -- - 3.5 0.17
Freon 113(1 85 e 48 218 2.2 - - - -
58 e 33 21.8 1.5 15.2 0.70
59 e 34 21.8 1.6 15.0 0.69
Baseline test - - -- - - -- 5.3 0.24
Chlorinated
aliphatics -- -- -- 4 to 7.9 10.7 0.51
Baseline -- -- - - - - 2.1 0.10
aContalned an average of 24 percent Cl. Assumed specific gravity of 0.9.
t)Contalned an average of 15.7 percent Cl. Assumed specific gravity of 1.2.
Contained an estImated 22 g Cl/I.
dContalned an estimated 51 percent Cl.
elbese rates are iii kg/hr.
t Uesults froa a long term test.
-------
chloride loading was low and ranged from 1 to 2 kg/Mg clinker. Note that
the baseline tests during these two experiments showed emission levels of
about 0.4 lb/ton compared to about 0.2 lb/ton for the other baseline tests.
It Is difficult to conclude If this Is a natural variation in test results
or attributable to some change in the process or ESP.
The tests at Store Vika showed that, for the chlorinated aliphatics
and PCBs, the increased chloride loading Increased emissions. The total
emissions reached a maximum of about 1 lb/ton and are comparable to other
cement kiln test results. The burning of the chlorophenols/phenoxy acids
and Freon also Increased emissions over the baseline by a factor of roughly
3 (to 1.4 lb/ton). Long—term tests with chlorinated aTpihatics and PCBs
confirmed the increase for these types of compounds at chloride loadings of
4.0—7.9 kg/Mg for the aliphatics and 6.2 kg/Mg for the PCBs. 4
The test data collected to date show similar results for pairs of tests:
• The testing of two kilns equipped with baghouses (San Juan
and Rockwell Lime) showed no increase in particulate matter
emissions when chlorinated wastes were burned.
• The testing of two kilns equipped with ESPs (St. Lawrence
Cement’s dry process kiln and Alpha Cement) showed a de-
crease in emissions when wastes were burned that were low in
chlorine content.
The testing of two kilns equipped with ESPs (Paulding and
Marquette) showed no change in particulate matter emissions
when chlorinated wastes were burned at rates of 1.1-4.7 kgCl/Mg
clinker.
• The testing of two kline equipped with ESPs (St. Lawrence’s
wet process kiln and Stora Vika) yielded results for different
wastes and different chlorine loadings. Emissions were
positively correlated with chlorine loading; however, the
extent of the Increas, in emissions was different for different
compounds and different klins. Chloride accumulation, as
evidenced by ring formation and subsequent release or pluggage,
begins to occur in the range of 6—9 kgCl/Mg clinker.
Although increased chlorine loading at the St. Lawrence and Stora Vika
kilns increased particulate matter emissions, there was no known attempt to
compensate foi” changes In the dust’s resistivity. The tendency to form
chloride rings during high chlorine loading provides an Incentive to the
kiln’s operator to limit the chlorine entering the kiln. Limiting the
41
-------
chlorine may avoid plugging and process disruptions as well as limiting the
chloride dust loading on the ESP. The test data suggest that particulate
matter emissions from chlorinated waste combustion are controllable by a
properly—operating control device and a reasonable limit on chloride loading
to avoid ring formation and excessive dust loading. Adjustments may be
required on a site-by-site basis to optimize ESP performance to account for
changes in the dust resistivity.
4.6 FATE OF CHLORINE
Burning chlorinated wastes in cement kilns can significantly increase
the chloride loading in the process. Several studies have focused on the
fate of the chlorine and effects on waste dust composition and generation
rats, emissions of particulate matter, and emissions of HC1.
The effects of additional chloride In cement kilns was known by kiln
operators before the recent test programs of chlorinated waste combustion.
It has been common practice in the cement Industry to add chlorides such as
calcium chloride or waste hydrochloric acid to the raw material feed to
reduce alkalis. 23 The term 11 alkalis 11 refers to potassium oxide (K 2 0) and
sodium oxide (Na 2 0) which are frequently combined and expressed u equivalent
Na 2 0. The concentrations of the two alkalis in the raw meal vary by source
of the raw material. An alkali specification on the cement of 0.6 percent
(as equivalent Na 2 0) is used in North Ajnerica because alkali reactive
aggregates (containing geologically young amorphous silica and some dolomita )
are usid in many areas. These reactive aggregates cause expansion and
disruption of concrete made from cement with a high alkali content. Non-
reactive aggregates have been found to perform well with cement independent
of alkali content. 2 23 One report also states that strength and settling
characteristics are also related to the alkali Concentration in the cement
kiln. 2 Consequently, the alkali content is a specification that must be
dealt with by the operators.
When chlorinated wastes are burned, the temperatures In the kiln favor
complete conversion of chloride in the waste to HC1. The 4C1 is rapidly
absorbed by the calcium, sodium, and potassium oxide in the kilns$ solids
and converted to the respective calcium, sodium, and potassium chlorides.
The resultant potassium and sodium chlorides are vaporized by the temperatures
42
-------
at the hot end of the kiln, travel down the length of the kiln, and begin
to condense on cooler surfaces or in the gas. 11 FIgure 4-1 shows a chloride
cycle that is developed in the kiln and dust collection system. Note that
chlorides are removed with the dust in the cyclones and final gas cleaning
device. It Is a routine practice to recycle the majority of this collected
dust back to the kiln, either with the raw material feed or by additions
near the firing end of the kiln. The second figure shows how the chloride
cycle is shifted and the chloride content of the clinker is decreased by
removing or bypassing some of the collected dust. This dust is removed and
disposed of instead gf being reintroduced into the kiln and provides one
mechanism of control for th. chloride cycle. Note that by increasing the
amount of dust wasted from the cyclone, the chloride dust loading on the
final gas cleaning device can actually be decreased. 23 Disposal of a
portion of the waste dust has been a conron practice, and several kllns
burning chlorinated wastes have increased their dust disposal rate (decreased
their percent recycled) to control the chloride cycle.
These chlorides also condense on the kiln walls and can create rings
of soljd . As the kiln rotates, solids tumble from th. wall and the newly-
exposed wall is cooler than the gas or surface solids. 24 The rotation of
the wall and exposure of cooler wall surfaces as the solids are removed
appears to promote ring formation. The chloride deposites can remain
sticky at the klln 1 s temperatures, and when th. chloride loading remains
too high, the deposits begin to accumulate. Sulfur can be a similar bad
actor in the buildup of solids. Alkali sulfates, because of their higher
melting points, tend to end up In the clinker. Ifowever, when there is art
excess of sulfur to alkali, calcium sulfate forms. These sulfates start
out soft and easy to remove but can become hard and difficult to remove
with time. The hardening has been found to be especially pronounced when
chlorides are present. 23
The chloride cycle Is affected by many factors such as chloride loading,
alkali content, dust recycle rates, and temperatures. In addition, kilns
also exhibit nonequilIbrlum behavior patterns for alkalis and sulfur.
Competition for absorption may also exist between the acidic HC1 and SO 2
gases. These factors tend to confound attempts at describing a precise
behavior pattern of chloride and sulfates that would be applicable to kilns
43
-------
FOUR-STAGE PREHEATER KILN
ALKALI AND O LORIDE CYCLES
K20 CYCLES
NO BY-PA55
5
15% BY-PASS
NA 2 O CYCLES
NO B”-PASS
23
15% BY PASS
CL CYCL!S
NO 5’r—PASS
15% !Y—PASS
KILN ICY CLONES ICOLLECTO
1 CYCLONES CLL!CTO
I I I
I KILN !CYCLONE5ICOLLECTO
I I I
KIt.. ICYCLONESFOLLECTC
55
Figure 4-1. Alkali and chiorids cycIes .
-------
in general. Site specific factors such as chloride loading, alkali content,
temperature control, operating conditions, etc. all effect the chloride and
sulfate cycles. Consequently, the emission test results reveal varying
levels of HC1 emissions, SO 2 emissions, percent chloride in the clinker,
and varying effects on particulate matter emissions. The variability
exists not only among different kilns, but also at the same kiln over time.
Another issue relates to the combustion of chlorinated organics that
are contaminated with metals 5uCh as lead and zinc. The formation of
volatile chlorides such as lead chloride (boiling point of 5010 C) and zinc
chloride (boiling point of 732° C) suggested to many the possibility of
increased metal’s emissions. 8 The fate and distribution of metals entering
with the waste fuels have been tracked in several tests and are discussed
in Section 4.7.
HC1 emission results are listed in Table 4—11. In the test at Alpha
Cement, waste solvents were burned that contained 590 ppm of identified
organic halogens; total chlorine content was not reported. HC1 emissions
during the baseline test averaged 2.4 lb/hr (1.6-3.6 lb/hr) compared to an
average of 5.8 lb/hr (4.2—7.3 lb/hr) during the solvent burn. The chloride
content of the dust was about 1.5 percent for both test conditions.
There are not enough data to estimate the kiln’s chloride removal efficiency.
The data do show an increase in HC1 emissions from burning the waste solvents.
The test at Marquette Cement showed HC1 emissions to be higher during
the baseline than during the waste burn (190 vs. 115 lb/hr). The waste
contained an average of 4.5 percent chlorine and entered the kiln at an
estimated rate of 1.1 kgCl/Mg clinker. These data suggest that the additional
chloride entering the kiln with the waste fuel was removed in the kiln’s
proceSS solids. However, the emission rate for HC1 was much higher for
this kiln during the baseline than the other kilns listed in Table 4-11.
SlinflarlY, SO 2 emissions were higher during the baseline than during the
waste burn (57 vs. 18 lb/hr). The change in SO 2 is more than can be accounted
for by the substitution for 11 to 14 perCent of the sulfur containing coal
by low—sulfur waste fuel. Higher levels of hydrocarbons, $02, NOR, and HC1
wre found during the baseline test but unstable operating conditions
contributed considerable variability within the data sets. At the
45
-------
TABLE 4—Li. HC1 EMISSIONS
Test
COfld Ofla
HC 1
Notes
(lb/h)
(lb/t)
Alpha Cement
W
B
5.8
2.4
0.11
0.05
Organic halogens in fuel =
590 ppm
Marquette
W
B
115
190
1.9
3.2
Avg. 1.1 kgCl/Mg clinker for
waste burn; waste avg.=
4.5% Cl
San Juan
W
B
0.79
<0.19
0.02
<0.06
Avg. 5.5 kgCl/Mg clinker for
waste burn; Waste avg. =
6.5 - 35.1% Cl
Los Robles
W
B
1.03
0.55
0.015
0.007
Waste = Z Cl; Avg. 1.1
kgCl/Mg clinker
Paulding
W
B
4.62
1.25
0.16
0.04
Avg. 2.2 kgCl/Mg clinker;
Waste avg. = Cl
Lone Star
W
B
25.3
2.9
0.46
0.054
Avg. 1.2 kgCl/Mg clinker;
Waste avg. = 1.9% Cl
St. Lawrence
W
B
<1
1
<0.02
<0.02
6.8 kgCl/Mg clinker, max.
0.7 kgCl/Mg clinker for
base! inc
Rockwell Lime
W
B
0.44
0.20
0.05
0.02
Avg. 2.7 kgCl/Mg lime; Waste
avg. = 3.1% Cl
= waste burn
B baseline
46
-------
95-percent confidence level, no significant increase was noted between
waste fuel and baseline emissions of HC1, MO, and SO 2 . ’ 4
During the San Juan test, highly—chlorinated solvents (up to 35.1
percent chlorine) were burned at an average rate of 5.5 kgCl/Mg clinker.
The average chloride input from the waste averaged 351 lb/hr and stack
emissions of HC1 were less than 0.8 lb/hr. HC1 emissions increased during
the solvent burn, but the kiln removed well over 99 percent of the potential
HC1. Most of the additional chloride was removed in the clinker (average
of 308 lb/hr); only marginal and sporadic increases in chloride concentration
were seen in the waste dust. 16 These data indicate that the chloriae cycle
in this kiln was different from those of other kilns tested. Other tests
show most of the chloride is removed with the waste dust.
During the test at Los Rabies, an average of 160 lb/hr of chlorine
•ntered the kiln and stack emissions of i4Cl were about 1 lb/hr. This kilt,
also removed over 99 percent of the potential HC1 emissions in the process
solids. An increase in HC1 emissions over the baseline results was
observed. 18
In the test at Paulding, the average chlorine from the waste as
12.3 lb/hr and the average HC1 emissions were 4.6 lb/hr. The results are
summarized in Table 4-12 and show that ten percent of the chloride escaped
as MCi during the baseline compared to about 5 percent during the waste
burn. HC1 emissions increase as the chloride loading from the fuel increases,
but 95 percent or more of the chloride is retained in the process solids.
Most of the additional chloride was retained by the waste dust, and from
Table 4—12, the chloride concentration of the waste dust increases as the
chlorine input Increases. One exception was on the first day of waste fuel
testing (Test Run 5) when detectable levels of chloride appeared in the
clinker (as in the San Juan test). During this run, about 41 percent of
the chloride was removed with the clinker and only 59 percent with the
dust. The MCi results In Table 4—12 also show for that run the lowest rate
of MCi emissions, which were one—half the average baseline rate. 1 ’
EvidentlY the chloride cycle shifted in this kiln from burning the chlo-
rinated wastes. It is also possible that an equilibrium chloride cycle had
not been established although over 40 hours of waste burning was conducted
before starting the test.
47
-------
TABLE 4 -12. HC1 RESULTS FOR THE PAULDING KILJ1’
Chiorinein HC1 emissions Waste dust
lest run fuel (kg/hr) (kglhr) (percent Cl)
Baselinea 6 0.6 0.15
5 32 0.3 0.70
6 21 1.0 3.70
7 41 1.5 0.67
8 72 2.3 1.0].
9 128 5.4 1.73
aAverage of four runs.
48
-------
The test results for Lone Star are given in Table 4—13. For most
runs, approximately 90-97 percent of the chlorine was removed with the
process solids. The chloride concentration of both the recycle dust and
waste dust increased during the waste fuel burn. The results for Test 5 at
the same chloride input rate as the previous test showed some interesting
results. HC1 emissions increased dramatically (confirmed by 2 HC1 runs)
and the chloride concentration of both the waste dust and recycled dust
continued to increase.’ 5 This also indicates a shift in the chloride
cycle, but in the opposite direction of the shift at San Juan and Test
Run 5 at Paulding. In those cases, the cycle shifted toward the clinker
end of the kiln with chloride removal by the clinker, whereas the shift here
was toward the cooler end of the kiln with an increasing amount of chloride
in the dust and stack gas. This test indicated that a kiln can retain
chloride from the waste fuel in the process solids, but the efficiency can
vary. (There is also a possibility that if any fine chloride particles
escaped through the sampling filter, then they would have been measured as
HC1 in the impinger solution).
During the test at St. Lawrence Cement, highly chlorinated solvents
were burned at rates up to 6.8 kgCl/Mg clinker (571 lb/hr chlorine). HC1
emissions were less than 1 lb/hr during both the baseline ana solvent
burns. The kiln was very efficient at retaining chloride in the process
solids (over 99 percent). The excess chloride was removed with the waste
dust and no detectable levels of chloride were found in the clinker. The
chloride concentration in the waste dust increased, and the dust discard
rate was increased by the kiln’s operator. This test indicates that the
operator’s may control the chloride cycle in the kiln by changing the waste
dust discard rate. This discard rate was increased to an apparent advantage
in this kiln. However, an optimized adjustment In the dust discard rate is
likely to require case by case study because of variations in chlorine
input, alkali content, kiln design, kiln operating conditions, etc.
The test at Rockwell Lime burned chlorinated wastes at only 0.8 lb/hr.
Most of the chlorine entered the kiln with the slurry feed (40—50 lb/hr).
A small but statistically significant increase in HC1 emissions was noted
when waste fuel was burned. However, the chloride material balance showed
that over 95 percent of the total chloride entering the kiln was retained
in the process solids. 2 ’
49
-------
TABLE 4-13. HC1 TEST RESULTS FOR LONE STAR’ 5
Chiorineirl HC1 emissions Recycle dust Waste dust
Test fuel (kglhr) (kg/hr) (X Cl) (% Cl)
Baseline 10 1.3 0.72 3.5
3 60 2.2 1.3 5.7
4 71 5.5 1.4 6.1
5 72 26.7 1.9 8.1
50
-------
In su nary, burning chlorinated wastes appears to increase HC1 emissions.
However, the vast majority of the additional chloride entering the kiln is
retained in the process solids (generally 90 to over 99 percent). The
chloride is usually removed with the waste dust. In some rare cases, there
Is an apparent shift in the chloride cycle and the extra chlorine appears
In the clinker. Increasing the waste discard rate appears to be necessary
when chlorinated organics are burned. These test results indicate that
sore study may be needed, perhaps on a case by case basis, to optimize the
chloride cycle. The advantages to the kiln’s operator to optimize this
cycle include minimizing ring formation, HC1 emissions, chloride in the
clinker, and optimizing the waste dust loading on the final gas cleaning
device.
4.7 EFFECTS ON LEAD EMISSIONS AND REMOVAL BY PROCESS SOLIDS
The distribution of metals has been examined in several cement kiln
tests to track the fate of metals found in the waste fuels at concentration
levels ranging from (100 ppm to 6,000 ppm. The effect of chlorine on
metal’s distribution has been of particular interest because of the potential
for forming volatile metal chlorides, which could increase the loading on
the gas cleaning device. Table 4-14 lists the boiling points of several
compounds that could be present in the kiln’s dust. The lead and zinc
halides boil well below the kiln’s maximum temperatures; consequently,
volatilization of these compounds is likely.
At Saint Lawrence cement, waste lubricating oil was burned in a dry
process (preheated) cement kiln. The oil contained an average of 0.6 percent
lead and 0.15 percent bromine. Lead was added to the kiln at an average
rate of 37 lb/hr during the waste oil burn with a corresponding emission
rate in the stack gas of 0.013 lb/hr. However, two baseline tests yielded
even higher emission rates of 0.015 and 0.023 lb/hr; therefore, the burning
of waste lubricating oil appeared to decrease lead emissions. The lead
entering the kiln was distributed primarily in the clinker (75—85 percent)
with the balance in the conditioning tower solids (9 percent) and bypass
dust (4 percent). Lead concentrations in the clinker ranged from
0.0014—0.0070 percent with a maximun of 0.013 percent. These lead levels
had no adverse effects on cement quality. The authors Concluded that
51
-------
TABLE 4—14. BOILING POINTS OF SELECTED
COMPOUNDS 25
Compound Boiling point (°C)
PbBr 2 918
PbC1 2 954
PbO 1472
Pb 1744
ZnBr 2 650
ZnC1 2 732
Zn 907
KEr 1383
KC1 1407
NaBr 1392
MaCi 1465
52
-------
finding most of the lead In the clinker suggested that the lead must be
either oxidized or taken into solid solution in the clthker.’
A subsequent test on a wet process cement kiln at St. Lawrence burning
chlorinated solvents showed lead concentrations of 0.04 mg/mi in the stack
gas for the baseline and for burning chlorinated aromatics. During the PCB
burn, an average concentration of 0.12 mgJm was reported. The increase in
lead concentration corresponds roughly to the Increase in concentration of
total particulate matter. Baseline emissions of 21 lb/hr of particles
increased to about 44 lb/hr during the PCB burn. Based on an average gas
flow of 2,300 standard m 3 /min, the lead emission rates ranged from 1.5 to
4.6 mg/s. The lead content of the waste solvents was very low and ranged
from 0.3 to 0.8 ppm. 2
The test at Alpha Cement used waste solvents containing about 2,000 ppm
lead. The particulate catch contained about 650 ppm lead during the baselire
compared to 3,000 ppm during th wast burn. These concentrations and the
reported particulate matter rates yield an estimate of about 17 mg/s for
the waste burn and 4 mg/s for the baseline. 11
The test at Marquette Cement burned waste solvents that contained
1,000 to 2,500 ppm lead, 900—1,800 ppm zinc, and 450—725 ppm chromium. A
comparison of average results is given below:
Metal’s Emissions (mg/s)
Pb Zn Cr
Baseline 5.0 0.3 2.5
Waste burn 9.1 2.3 6.7
Although the average results show an increase, the values for individual
runs were quite variable for the two test conditions. 14
The test at Los Robles burned fuel containing 134-739 ppm lead which
was fed at a rate of 0.8 to 4.9 lb/hr. The quantity of particulate matter
captured after gas cleaning in the baghouse was tou small for metal’s
deterfili nation. The lead content of the baghouse dust increased from 750 ppm
during the baseline to about 1,330 pp. during the waste burn.
Waste fuel at the Paulding, Ohio site contained 149 to 814 ppm lead
and 246 to .744 ppm zinc. The total lead entering the kiln averaged about
14 lb/hr during the waste burn compared to about 7 lb/hr (primarily from
the slurry feed) during the baseline. Lead emissions during the baseline
53
-------
averaged 1.5 mg/s compared to 6.9 mg/s during the waste burn. A shift in
the lead distribution was noticed as the precipitator dust increased in
lead concentration from 395 to 1,530 ppm. No increase in the lead concen-
tration of the clinker was noted with values of 66 to 143 ppm lead reported.
At the Lone Star test, the waste fuel contained 600 to 1700 ppm lead.
An ESP malfunction prevented useful emission results from being determined.
During the baseline, the clinker’s average concentration of lead was 5.7 ppm
compared to 35 ppm during the waste burn. The lead concentration in the
waste dust increased from 116 to 2,650 ppin During the baseline, the
clinker removed 77 percent of the lead, whereas the majority of the lead
(58 percent) was removed with the waste dust during the waste burn. ’ 5
The waste fuel burned at the Rockwell Lime test contained low levels
(—130 ppm) of lead. No lead was detected in the lime product, and the
concentration in the baghouse dust increased from below the detection limit
(—28 ppm) to an average of 230 ppm. Stack emissions of lead were not
detectable during the baseline test; an emission rate of about 1.7 g/hr was
reported during the waste fuel burn.
The emission results for lead are listed in Table 4-15. For comparison
purposes and added perspective, the lead emission rates are compared with
the emissions from an automobile that burns lead fuel. The comparison
suggests that lead emissions from the kiln, whether waste fuel is burned or
not, are about the same order of magnitude as the emissions from one to
several automobiles burning leaded fuel.
The first St. Lawrence test indicates that most of the lead is retained
in the process solids with no increase (actually a decrease) in stack
emissions when waste fuel containing lead was burned. This test included a
high concentration of lead and a low concentration of halogens in the waste
fuel. The second test at St. Lawrence included a fuel that was much higher
in halogen content and a very low lead concentration. These data do not
show a significant increase in lead emissions from burning the wastes,
except for the PCB burn in which total particle emissions increased corres-
pondi ngly.
The results for Alpha Cement, Marquette, and Paulding indicate that
when wastes containing 1,000 to 2,000 ppm lead are burned, lead emissions
54
-------
Notes
6,000 ppm In waste; Pb retained In
process solids
03 to 0.8 ppm in waste; Pb in waste
dust. increased
2,000 ppm in waste; Pb In waste dust
Increased
1,000 to 2,600 ppm in waste; Pb in
waste dust Increased
149-814 ppm in waste; Pb In waste
dust Increased
134-139 ppm in waste; Pb In waste
dust Increased
130 ppa In waste dust; Pb In waste
dust increased
Waste fuel
Lubricating oil
Cl aro.atics
PCB’s
Baseline
Waste solvents
Cl solvents
Cl solvents
Cl solvents
Cl solvents
Test
St. Lawrence
St. Lawrence
Alpha Cement
Marquette
Paulding
Los Rolles
Rockwell time
a 0 - baseline
W = waste Fuel
brotai particle
1ABLE 4-15. SUI44ARY OF LEAD EI4ISSIONS
Pb emissions
COfld l t lOfla (mgIs)
B 2-3
W 2
U —1.5
U —4.6
ft —1.5
It —4
U —17
S S
W 9
B 1.5
U 6.9
It (b)
U (b)
B <0.42
U 0.41
burn
emissions too low for determination.
-------
can increase. However, the tests also show that most of the lead enterir g
the kiln is retained in the process solids. The lead in the waste fuel and
perhaps the chlorine content of the fuel appear to shift the distribution
of lead such that more is removed with the waste dust.
In conclusion, the cement kiln has a large potential for retaining
lead from waste fuel in the process solids. The various tests suggest that
lead can be retained in the clinker with no adverse effects on clinker
quality. When chlorinated wastes are burned, the data suggest that the
lead is still retained in the process solids, but more of it appears ‘in the
waste dust. These test results show that, with a high lead loading and
with chlorinated solvents, lead chloride is formed and removed by the ESP
or baghouse. Emission estimates also show that burning lead-containing
fuels increases lead emission rates. Although the percent increase in lead
emissions may appear high for some tests (a factor of roughly 2 to 4), a
more important point is the absolute impact of total emissions. As shown
in Table 4—15, the lead emission rates for all tests but one were belcw
10 mg/s and are roughly comparable to emissions from automobiles burning
leaded gas.
Another perspective is to consider that when waste oil is burned in
boilers, 50 to 60 percent of the lead from the oil is emitted.’ Much
more lead is emitted during soot blowing. In contrast, the cement kiln can
retain well aver 99 percent of the lead in the process solids.
4.8 EFFECTS OF WASTE COMBUSTION ON N0 and SO. EMISSIONS
The tests of several kilns have followed the effects of waste cornDust ’ion
on N0 and SO 2 emissions and the results have been quite variable from kiln
to kiln. The alkaline materials in the kiln scrub SO from the gases at
efficiences that vary because of the sulfur content of the primary fuel,
alkali content of the raw materials, HC1 in the gas, O. input, waste replace-
ment rate, etc.
N0 emissions are also quite variable and range from 200 to 1,500 ppm
within hours at the same kiln. A discussion by Peters and Mournighan (26)
explain that in the kilns, a direct relationship exists between N0 emissions
and secondary air (heated air from the clinker cooler) with little NO
found from the nitrogen in the fuel. Kilns are generally operated as
56
-------
“poor” combustors compared to oilers because of a reduced primary air/fuel
mix, poor secondary air mixing, and generally a confined, narrow flame config-
uration. Excess air and its temperature are the main determinants in kiln
NO emissions. This implies that NO concentrations are very dependent
upon the specific operating conditions of a given kiln and are not likely
to be affected by waste burning.
KVB, Inc. has investigated the effects of different process parameters
on NO emissions in a pilot scale rotary cement kiln fired by natural gas
and coal. Their study found that increases in either kiln exit oxygen or
the secondary air temperature increased NO emissions. The empirical relation-
ships developed for the kiln they studied were:
Gas fuel: NO(ppm) = 87.7 + 45.3(0 %) + 0.16(T 5 °F)
Coal fuel: NO(ppm) = -164.6 + 77.2(02%) + 0.67 (T °F)
These results predict a 4577 ppm decrease in NO emissions from a 1 percent
decrease in the kiln exit °2• The effect of secondary air temperature
appears to be more pronounced for coal. A decrease of 100° F in secondary
air temperature yields a 67 ppm decrease in NO for coal compared to a
16 ppm decrease for gas fuel. 2
The effect of the primary/secondary air ratio was also examined. The
results for natural gas firing showed that NO e.missions peaked at a primary
air concentration of 20 percent and decreased at higher and lower ratios.
A similar result was found for coal fuel where NO emissions peaked at about
23 percent primary air. The report explained that at primary/secondary air
ratios less than the peak (0.20-0.23), flame intensity (mixing) was less
and resulted in lower NO. For the higher ratios, the higher mixing is
overshadowed by the larger amount of cold air introduced into the flame
zone.
NO emissions were also found to increase with increases in secondary
air temperature. For example, increasing the secondary air temperature
from ambient to 900-1100° F doubled NO concentrations. The report concluded
that combustion modifications can be effective in reducing NO emissions.
The report did not discuss the effects on 502 emissions or th. effects on
product quality (although they attempted to make “good” clinker during the
test runs). 4
57
-------
S0 emission test results are summarized in Table 4—16. The results
show that substitution of the sulfur—containing primary fuel with a low
sulfur waste fuel decreased °2 emissions in tests at Alpha Cement 1
Marquette Cement, and General Portland at Paulding. At Paulding, the
kiln’s process solids retained about 60 percent of the total sulfur during
both the waste burn and baseline tests. The San Juan test showed an increase
in SO emissions when waste fuel was burned. The increase was attributed
to a lower O input (as evidenced by lower N0 emissions) and competitive
acid gas scrubbing from HC1 in a relatively low alkaline kiln when burning
the highly chlorinated wastes (average of 5.5 kgCl/Mg clinker).
At Los Robles average 502 concentrations were relatively low (27 ppm)
as was the sulfur content of the coal (0.43 percent). Sulfur removal in
this kiln averaged 89 percent. N0 emissions appeared to decrease when the
waste was burned, but the decrease (see Table 4-17) was not statistically
significant when the variance between runs is considered. At Los Robles, a
steady decrease in NO emissions on one test day from 1054 to 526 ppm was
attributed to a decrease in kiln excess air from 1.3 to 0.5 percent 0.2.
The somewhat lower N0 during the waste burn and one baseline test was
atrributed to additional chains that were installed to improve heat transfer
from the gas to the incoming feed. The more efficient use of heat permitted
the firing end of the kiln to be operated at lower temperatures with a
resultant reduction in NOX.lo
The test at Lone Star showed relatively low S0 concentrations that
ranged from about 2 to 38 ppm. Although the average results indicate an
increase in $02 emissions when wastes are burned, two of three waste fuel
tests results for SO 2 were about the same as the baseline results. The
highest SO 2 concentrations were seen during the third waste burn test which
averaged 38 ppm SO 2 with an exit O. of 6.7 percent. For the two subsequent
test days, the kilns exit oxygen was increased to 7.3 and 7.5 percent,
respectively, with resultant decreases of °2 to 13 and 5 ppm, respectively.
concentrations during the baseline were lower than during the waste
burn (620 vs. 814 ppm) and can probably be attributed to the lower O input
for the baseline (average 6.5 vs. 7.2 percent 02). The variation of NO
with secondary air flow was demonstrated by oscillations in undergrate
pressure. Increases in undergrate pressure yielded increased NO concen-
trations and periodic fluctuations of 100 ppm or more were observed.’ 5
58
-------
in coal = 2.6%; S In waste 0.2%
during waste burn = 2.0%
In waste 0.08%
in coal = ?
in fuel oil = 2.15%
In coal = 0.43%
In coal = 4.3%
in coal/coke 2.1%
In coke = 4.1%
Test
Condition
So 2
(lb/h)
(lb/t)
(ppm)
Alpha Cement
W
8
58.5
138
1.1
2.1
33
18
S
S
Marquette
W
B
11.5
51.1
0.19
0.95
18
93
S
S
San Juan
W
B
264
170
8
5
450
219
S
Los Robles
W
B
21.7
23.1
0.36
0.38
21
27
S
U ’
tD
Paulding
W
B
207
526
6.8
11.2
265
636
S
Lone Star
U
B
14.7
5.6
.12
.10
19
1
S
Rockwell lime
U
8
91
149
11.4
17.5
596
553
S
-------
TABLE 4—17. SUMIkARY OF NO ENISSIONS
Test
Conditiona
NO
(lb/h)
(lb/t)
(ppm)
Marquette
W
B
275
404
4.6
6.7
544
920
San Juan
W
B
31.3
60.4
0.9
1.8
68
136
Los Robles
W
a
304
444
5.3
8.2
4.86
680
Paulding
W
B
174
140
6.0
4.6
478
371
Lone Star
W
8
472
371
8.6
6.9
814
620
Rockwell Lime
W
B
--
134
-—
15.8
446
386
60
-------
The °2 emission results for Rockwell Lime are an exceptional case and
are not at all representative of cement kilns. At the lime plant, SO
eaissions are deliberately not absorbed into the product because the pre-
since of sulfur in the lime is undesirable. No significant difference in
SO emissions was observed between the baseline and waste fuel burns with
average concentrations of 500600 ppm during each. The test at Rockwell
Lime showed that NO and S0 concentrations change simultaneously in oppo-
site directions. 4 NO. emissions increase with increasing °2 input and
degree of preheating while SO 2 emissions decrease under the same conditions.
The same results were observed in the Paulding test during the waste fuel
burn. NO and concentrations tracked together with swings in the
opposite direction. At times the swings were several hundred of ppms in
amplitude for both NO and 502 over 1— to 2-hour periods. “
These test results suggest several conclusions about the effects of
waste combustion on NO and emissions.
• The substitution of low sulfur waste for the sulfur-containing
waste fuel can decrease 502 emissions when the waste fuel is
burned.
• SO 2 emissions are very dependent upon the specific operating
conditions of the kiln. Consequently, variations in SO 2
concentrations are observed at the same kiln over time as
operating conditions change, and the variability increases
from kiln to kiln because of differences in primary fuel,
kiln design, and operating conditions.
• Factors affecting SO 2 emissions apparently include:
• Sulfur loading on the kiln
• Alkali Content of raw materials
• 0.4 input
• Temperature
Competition from HC1
NO emissions are not affected by waste combustion. They
ar affected by 02 input, secondary to primary air ratio,
and temperatures. NO emissions are very sensitive to minor
process changes and snow rapid and sometimes large responses
to these changes. Consequently, NO concentrations have
been quite variable over time at the same kiln.
• An interrelationship exists between NO , S0 , and 02 input.
Continuous monitoring results indicateXthat shifts in NO
concentrations are often accompanied by SO 2 swings in th
opposite direction. An increase in °2 input increases NO
emissions and decreases 502 emissions. X
61
-------
The test data collected on calcining kilns indicate that continuous
monitoring for 502 and N0 at some kilns may offer several advantages.
Combustion conditions may be optimized for minimum fuel consumption.
Emissions may be balanced or optimized within the constraints of current
regulations. Immediate responses of the continuous monitors to process
changes, coupled with accumulation of some historical data, may signifi-
cantly improve process control, temperature control, quality control and
energy usage. The potential exists for predicting upset conditions before
they occur and taking corrective actions.
A careful consideration of costs and potential benefits would be
needed before any recommendation on the wholesale use of the continuous
monitors could be made. Mowever, a cost-sharing research program bet’. een a
government environmental (or even energy) agency and an interested cooTnercial
company on the uses of continuous NO and SO 2 monitors could be to the
mutual benefit of both parties.
4.9 EMISSIONS OF TOTAL HYDROCARBONS (THC) AND CARBON MONOXIDE (CD)
The results for total hydrocarbons and carbon monoxide are summarized
in Table 4-18. Total hydrocarbons increased during waste fuel combustion
at 3 tests, decreased at 2 tests, and remained the same at another. The
consistently low THC concentrations for several different types of compounds
at Stora Vika suggest that kiln operation may affect THC concentrations
more than the waste type does. The levels at Rockwell Lime, which attained
consistently high DRES, averaged 3.5 ppm. Similarly, the Lone Star kiln
maintained consistently low levels of THC between 2. 5 and 5 ppm. These
data for different types of kilns and wastes fuels indicate that a well-
controlled kiln can operate with average ThC concentrations below 10 ppm.
CO emissions are not affected by burning waste fuels. Weitzman noted
that changes in CO can be indicative of flame quenching, improper burner
adjustment, or other imbalances in the flame. 1 Peters and Mourniqhan(26)
found that any process change can create a significant CO excursion, which
Is often a temporary spike. Examples of process changes include changes in
the primary air/fuel ratio, changes in secondary air, irregular solid fuel
feed rates, and changes In the exhaust damper. 4 These effects were observed
during the Paulding test during a kiln upset with coal as the only fuel.
62
-------
TABLE 4-18. SU ?4ARY OF THC AND CO CONCENTRATIONS
Site Test condition mc (ppm)a CO (ppm)
Marquette Waste burn 47O -
Baseline 1,250 —
San Juan Waste burn 12.7 24 -
Baseline 8.3 25 -
Los Robles Waste burn d 25 — 100
Baseline 4 10 - 618
Paulding Waste burn 21 190
Baseline 10 212
Lone Star Waste burn 5 24 -
Baseline 2.5 35 - 40 c
Rockwell Lime Waste burn 3.5
Baseline 8.2 477
Stora Vika Chlorinated aliphatics <10 300 - 1500
Baseline <10 1,500
PCSs <10 100 - 1500
Baseline <10 100
Chlorophenols/phenoxy acids 10 50 - 500
Baseline 10 50
Freon <10 100 - 500
Baseline <10 100
aE,cpressed as ppm methane unless otherwise noted.
rotal nonmethane hydrocarbons.
CRange of test averages.
duct measured.
63
-------
Over the course of the event, several process parameters were changed and
large swings in CO (as well as other monitored gas concentrations) were
observed. The CO results for Stora Vika show a range of 50-1500 ppm for
both the baseline and waste fuel burns. The CO results for Lone Star
appear to be the most consistently low. The kiln was operated with a
higher °2 input (to aid in drying wet coal), which apparently resulted in
very consistently low levels of THC, CO and 502 with increased NO
concentrations. The Los Robles kiln was also very stable during three
waste firing tests in which the maximum CO was 100 ppm.
The CO results indicate that waste fuel combustion does not signifi-
cantly affect CO concentrations. Process stability apparently has a greater
effect on CO concentrations. CO monitors on kllris provide the operator with
a measure of this stability, but only in the extreme because of high detection
limits. More sensitive Continuous monitoring has shown that CO spikes can
be caused by minor process charges, and poor combustion conditions (e.g.,
kiln upset) can lead to longer and higher swings in CO.
64
-------
5.0 REFERENCaS
1. Berry, E. E. , L. P. MacDonald, and 0. J. Skinner. 1975. E cperimental
Burning of Waste Oil as a Fuel in Cement Manufacture. Environment
Canada. Report No. EPS 4-WP-75-j.
2. MacDonald, L. P. , D. J. Skinner, F. J. Hopton, and G. H. Thomas.
1977. Burning Waste Chlorinated Hydrocarbons in a Cement Kiln.
Fisheries and Environment Canada. Report No. EPS 4—WP—77-2.
3. Ahling, Bengt. 1978. Combustion Test with Chlorinated Hydrocarbons
in a Cement Kiln at Stora Vika Test Center. Swedish Water and Air
Pollution Research Institute, Stockholm, Sweden.
4. Ahling, Benqt. 1979. Destruction of Chlorinated Hydrocarbons in a
Cement Kiln. Environ. Sci. Tech . 13(11):1377—1379.
5. Calonius, L. 1984. Morwegian Cement Plant Fires Kilns Using Toxic
Waste Fuel. The Wall Street Journal . December 21, 1984. p. 21.
6. Black, M. W. , and J. R. Swanson. 1983. Destruction of PCEs in Cement
Kilns. Pollution Engineering . June 1983. pp. 50-53.
7. Lauber, J. 0. 1982. Burning Chemical Wastes as Fuels in Cement
Kilns. J. Air Poll . Control Association 32(7):771776.
8. Mix, T. W. , and B. L. Murphy. 1984. Risks Associated with Waste Fuel
Use in Cement Kilns. Env. Progress 3(1):64-70.
9. Hazelwood, D. , at al. 1982. Assessment of Waste Fuel Use in Cement
Kilns. EPA 600/2-82-013, NTIS P8-82-23 6043, U.S. Environmental
Protection Agency, Washington, D.C.
10. Levin, J. E. 1981. Feasibility of Using Lime Kilns to Burn Hazardous
Wastes. (Draft Report) A. 1. Kearney, Inc. EPA Contract No. 68-03-256.
February.
1. Weitzman, L. 1983. Cement Kilns as Hazardous Waste Incinerators.
Env. Progress 2(1):1O—14.
1.2. Dalmia, J. H. 1968. Simulation of a Wet Process Cement Rotary Kiln.
Report R-392. Coordinated Science Laboratory, University of Illinois,
Urbana.
13. Swanson Environmental, Inc. 1976. PCB Emission Study at Peerless
Cement Company, Detroit, Michigan. Department of Natural Resources,
Air Quality Division, Detroit, Michigan.
65
-------
14. Higgins, G. H., and A. J. Helmstetter. 1982. Evaluation of Hazardous
Waste Incineration in a Dry Process Cement Kiln.
15. Branscome, H., J. Boistad, et al. 1984. Evaluation of Waste Combustion
in a Dry—Process Cement Kiln at Lone Star Industries, Oglesby, Illinois.
(Draft Report). EDA Contract No. 68-02-3149. U.S. Environmental
Protection Agency, Cincinnati, Ohio.
16. Peters, J. A., et al. 1983. Evaluation of Hazardous Waste Incineration
In Cement Kilns at San Juan Cement Company. (Draft Report). EPA
Contract Mo. 68—03—3025. U.S. Environmental Protection Agency, Cincinnati
Ohio.
17. Nesseibeck, E. R. 1982. Baseline and Solvent Fuels Stack Emissions
Tests. Prepared for Ener ’ and Resource Recovery Corporation, Albany,
New York. January.
18. Jenkins, A. C., et al. 1982. Supplemental Fuels Project, General
Portland, Inc., Los Robles Cement Plant. State of California Air
Resources Board. Report C-82-080.
19. Branscome, H. , J. Boistad, et al. 1984. Evaluation of Waste Combustion
in a Wet Process Cement Kiln at General Portland Inc., Paulding, Ohio.
(Draft Report). EPA Contract No. 68—02-3149. U.S. Environmental
Protection Agency, Cincinnati, Ohio.
20. Day, D. R., and L. A. Cox. 1983. Evaluation of Hazardous Waste incineration
in Lime Kilns at Rockwell Lime Company. (Draft Report). EPA Contract
No. 68-03-3025. U.S. Environmental Protection Agency, Cincinnati,
Ohio.
21. Hazardous Waste Regulations—Appendix VIII. Code of Federal Regulations,
Part 261, Volume 40.
22. Harris, J. C., et al. 1983. Sampling and Analysis Methods for Hazardous
Waste Combustion. (Draft Report). EPA Contract No. 68-02-3111. U.S.
Environmental Protection Agency, Research Triangle Park, North Carolina.
23. Norbom, H. P. 1973. Application of Suspension Preheater Kilns vs.
Other Kilns in North America. Presented at the I.E.E.E. Cement Industry
Technical Conference, Miami, Florida. May.
24. Brimacombe, J. K., and A. P. Watkinson. 1978. Heat Transfer in a
Direct-Fired Rotary Kiln. Metallurgical Transactions , 9(B):201-208.
25. Perry, R. H. and C. H. Chilton (eds.). 1973. Chemical Engineer’s
Handbook. Fifth Edition. McGraw—Hill, New York.
26. Peters, J. , and P. Mournighan. 1984. Effects of Disposal of Hazardous
Wastes in Cement Kilns on Conventional Pollutant Emissions.
66
-------
27. Fennelly P., et.al. 1984. Environmental Characterization of Waste
Oil Comb’ stion in Small Boilers. Hazardous Waste , 1(4):4.89—505.
28. Carter, W. A., and R. C. Benson. 1982. Application of Combustion
Modification Technology For NO Control to Cement Kilns. Joint S miposium
on Stationary Combustion NO Control, USEPA/EPRI, Dallas, Texas,
November 1-4.
29. Haynes, B. W., and C. W. Kramer. 1982. Characterization of U.S.
Cement Kiln Dust. U.S. Department of the Interior. NTIS P883-122655.
67
------- |