United States
Environmental Protection
Agency
Industrial Environmental
Research Laboratory
Cincinnati, OH 45268
EPA-600/9-84-022
September 1984
Research and Development
Incineration and
Treatment of
Hazardous Waste
Proceedings of the
Tenth Annual
Research Symposium
-------
-------
EPA-600/9-84-022
September 1984
INCINERATION AND TREATMENT OF HAZARDOUS WASTE
Proceedings of the Tenth Annual Research Symposium
at Ft. Mitchell, Kentucky, April 3-5, 1984
Sponsored by the U.S. EPA, Office of Research & Development
Tndustrial Environmental Research Laboratory
Energy Pollution Control Division
and
Municipal Environmental Research Laboratory
Solid and Hazardous Waste Research Division
Coordinated by:
JACA Corp.
Fort Washington, Pennsylvania 19034
Contract No. 68-03-3131
Project Officer
Harry Freeman
Energy Pollution Control Division
Industrial Environmental Research Laboratory
Cincinnati, Ohio 45268
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
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 endorsement or recommen-
dation for use.
-ii-
-------
FOREWORD
We at the Industrial Environmental Research Laboratory in Cincinnati
have always believed that an important part of our responsibility is to
provide to the technical community the results of our various projects in as
timely a manner as possible. This Proceedings document, containing papers
presented at our 1984 Research Symposium, presents project summaries and
interim results of our current work for this purpose, since much of the work
discussed may not be available in final reports for some time. We hope that
you will find these papers useful. Any questions or comments that you
might have should be directed to the Chief, Incineration Research Group, IERL.
David G. Stephan
Director
Industrial Environmental
Research Laboratory
-iii-
-------
PREFACE
These Proceedings are intended to disseminate up-to-date information on extra-
mural research projects concerning land disposal, incineration, and treatment of
hazardous waste. These projects are funded by the U.S. Environmental Protection
Agency's Office of Research and Development and have been reviewed in accordance with
the requirements of EPA's Peer and Administrative Review Control System.
-IV-
-------
ABSTRACT
The Tenth Annual Research Symposium on Land Disposal, Remedial Action,
Incineration and Treatment of Hazardous Waste was held in Fort Mitchell,
Kentucky, April 3 through 5, 1984. The purpose of the symposium was to
present to persons concerned with hazardous waste management the latest
significant findings of ongoing and recently completed research projects
funded by the Industrial Environmental Research Laboratory's Energy Pollution
Control Division and the Municipal Environmental Research Laboratory's Solid
and Hazardous Waste Research Division.
This volume is a compilation of speakers' papers and poster presenters'
abstracts for Session B, Hazardous Waste Incineration and Treatment.
Subjects include innovative technology, lab-scale research projects, pilot
and full-scale incineration projects, combustion of hazardous waste in
boilers and in industrial processes, and hazardous waste treatment and cost
studies.
This document covers hazardous waste incineration and treatment only. A
separate document for Session A, Hazardous Waste Land Disposal, is available
from the Municipal Environmental Research Laboratory.
-v-
-------
CONTENTS
Page
Environmental Characterization of the Combustion of Waste
Oil in Small Commercial Boilers
Paul F. Fennelly, Mark McCabe, and Joanna Hall
GCA Corp 1
Demonstration of Wet Air Oxidation of Hazardous Waste
William Cop'a, James Heimbeck, and Phillip Schaefer
Zimpro, Inc 9
Evaluation of Pilot-Scale APCDs for Hazardous Waste
Combustion Sources
C.W. Westbrook
Research Triangle Institute 19
Disposal of Hazardous Waste in Aggregate Kilns
James Peters and Duane Day, Monsanto, and
Robert Mournighan, U.S. Environmental Protection Agency . . 24
Effects of Disposal of Hazardous Wastes in Cement Kilns on
Conventional Pollutant Emissions
James Peters, Monsanto, and Robert Mournighan,
U.S. Environmental Protection Agency 38
Evaluation of Hazardous Waste Incineration in a Lime Kiln
Duane Day, Monsanto, and Robert Mournighan,
U.S. Environmental Protection Agency 48
Field Tests of Industrial Boilers Cofiring Hazardous Wastes
Carlo Castaldini, Howard Mason, Robert DeRosier, and
Stefan Unnasch
Acurex Corp 57
Field Tests of Industrial Boilers and Industrial Processes
Disposing of Hazardous Wastes
Radford Adams, Michael Hartman, and Denny Wagoner
Radian Corp 62
Summary of Field Tests for an Industrial Boiler Disposing of
Hazardous Wastes
John Chehaske, Engineering-Sciences, and
Gregory Higgins, SYSTECH Corp 70
Parametric Experimentation with a Pilot-Scale Boiler Burning
Hazardous Compounds
C. Wolbach, Acurex Corp 76
-vi-
-------
CONTENTS (Cont'd)
Products of Incomplete Combustion from Hazardous Waste
Incinerators
Andrew Trenholm and Roger Hathaway,-
Midwest Research Institute, and Don Oberacker,
U.S. Environmental Protection Agency
Preliminary Assessment of Costs and Credits for Hazardous Waste
Co-Firing in Industrial Boilers
R. McCormick and L. Weitzman
Acurex Corp • • •
Hazardous Waste Pretreatment as an Air Pollution Control
Technique
James Spivey, C. Clark Allen, and Robert Stall ings,
Research Triangle Institute, and Benjamin Blaney,
U.S. Environmental Protection Agency
84
96
Programs of the Industrial Waste Combustion Group at the U.S.
EPA Center Hill Facility
George Huffman, U.S. Environmental Protection Agency. . . .
Determination of the Thermal Decomposition Properties of 20
Selected Hazardous Organic Compounds
Barry Bellinger, Juan Torres, Wayne Rubey, Douglas Hall,
and John Graham, University of Dayton, and Richard Carnes,
U.S. Environmental Protection Agency
104
114
116
Hazardous Waste Destruction Using Plasma Arc Technology
Thomas Barton, Pyrolysis Systems Incorporated;
Nicholas Kolak, New York State Department of Environmental
Conservation; and Chun Ching Lee,
U.S. Environmental Protection Agency
Metal Value Recovery from Metal Hydroxide Sludges
L. Twidwell, A. Mehta, and G. Hughes, Montana College of
Mineral Science & Technology. . . •
Operations at the U.S. EPA Combustion Research Facility
Richard Carnes, U.S. Environmental Protection Agency. . .
Sampling Methods for Emissions from Hazardous Waste Combustion
Larry Johnson, U.S. Environmental Protection Agency . . .
Incineration of Hazardous Waste in Power Boilers: Emissions
Performance Study Rationale and Test Site Matrix
Robert Olexsey, U.S. Environmental Protection Agency. . .
124
129
143
155
162
-vii-
-------
CONTENTS (Cont'd)
Posters
Observation of Fluid Dynamic Effects upon High Temperature
Destruction of Organic Compounds
Wayne Rubey, John Graham, Douglas Hall, and Barry Del linger
University of Dayton Research Institute 178
Research Projects at EPA's Louisiana State University Hazardous
Waste Research Center - US EPA Center of Excellence
D.P. Harrison, Louisiana State University
Hazardous Waste Control Technology Data Base
Richard Holberger, The MITRE Corp., and
C.C. Lee, U.S. Environmental Protection Agency.
Evaluation of an Advanced Low NOX Heavy Oil Burner for
Incineration of Nitrated Hazardous Compounds
W.S. Lanier and Charles Sedman
U.S. Environmental Protection Agency
Regulatory Support Research Studies Supported by EPA/Cincinnati
MSD Cooperative Agreement
Boyd T. Riley, RYCON
Flame-Mode Hazardous Waste Destruction Research
W. Randall Seeker, John Kramlich, Michael P. Heap, and
Rachel K. Nihart, Energy and Environmental Research Corp.,
and Gary Samuelson, University of California .
Evaluation and Development of Innovative Incineration Testina
Procedures
Jerome Strauss, Versar, Inc
179
180
181
182
183
184
-viii-
-------
ENVIRONMENTAL CHARACTERIZATION OF THE COMBUSTION
OF WASTE OIL IN SMALL COMMERCIAL BOILERS
Paul F. Fennelly, Mark McCabe, Joanna Hall
GCA Corporation, Technology Division
Bedford, Ma. 01730
ABSTRACT
Tests were conducted on seven boilers in the size range of 0.4 to 20 X 106 British
thermal units per hour (Btu/hr) which were burning a waste oil fuel spiked with a series
of organic chemical compounds commonly found in waste oil. Compounds tested included
chlorinated solvents such as chloroform, trichloroethylene, trichloroethane, tetrachloro-
ethylene and several chlorinated benzenes. Test results are presented here in terms of
stack gas emission concentrations and destruction efficiencies for the various compounds.
INTRODUCTION AND PURPOSE OF THE PROJECT
In recent years the environmental
impact of the disposal of used oils has
been an area of growing concern. Numerous
studies conducted by state and Federal
agencies have documented the presence of
contaminants such as polynuclear aromatics
(PNAs), chlorinated hydrocarbons and heavy
metals in samples of used motor oils. One
of the more common and widespread prac-
tices for disposing of used oils is burning
as a supplemental fuel. In some cases,
waste oil is burned directly, in others, it
is blended with other fuel feedstocks.
The disposal of waste materials in
boilers is of particular interest because
to date, there has been little documenta-
tion of the extent to which chemical con-
taminants- in waste oil are destroyed during
the combustion-process.
In this project, tests were conducted
on boilers in the size range of 0.4 to
20 X 106 Btu/hr. These are commonly clas-
sified as commercial sources, as opposed to
industrial or electric utility sources.
Commercial boilers are of particular inter-
est with regard to waste oil disposal for
several reasons. These units generally
would use untreated or poorly characterized
waste fuels. They could be expected to
provide less efficient combustion because
of the'generally intermittent mode of
operation. In addition, their widespread
distribution and their low stack heights
make their emissions more proximate to the
general population.
APPROACH
Seven boilers were designated for
testing in the program. The units were
selected so as to provide a representative
cross section of the commercial boiler
population. A 4000 gallon lot of used
automotive oil was obtained in order to
maintain a consistent supply of waste fuel
for the program. Portions of. the base
stock oil were spiked with predetermined
amounts of selected organic compounds which
are typically found in waste oil and in
some cases are considered hazardous waste
materials. The selected compounds were as
follows:
e chloroform
c 1,1,1-trichloroethane
e trichloroethylene
-------
These compounds were chosen to simu-
late a typical waste oil, based on data
from a cross section of representative
waste oils.1
Measurements were conducted at each
of the sites to determine the atmospheric
emissions of particulate, inorganic com-
pounds (principally lead an.d hydrogen
chloride (HC1), and volatile organic and
semivolatile organic material. The
destruction and removal efficiencies (ORE)
for each of the spiked components were
also determined.
TEST RESULTS
Destruction and Removal Efficiencies
The flue gas emissions of the organic
compounds of interest correspond to •
destruction and removal -efficiencies of
99.4 to 99.99 percent as indicated in
Table 1. There were no strong correla-
tions between destruction efficiency and
boiler size or firing technique'. One
trend that is apparent- from the data is
that the destruction efficiencies for the
semivolatile compounds are consistently
higher than those of the volatile com-
pounds. The fact that generally higher
ORE'S were achieved for the semivolatile
components, trichlorobenzene, dichloro-
naphthalene and trichlorophenol, is con-
sistent with the ranking.of the spike
compounds on the EPA Hierarchy of Waste
Incinerability.2
Generally the lowest DRE-'s were
found in site A, the only boiler rated at
less than 1 X 106 Btu/hr. This unit nor-
mally fires a No. 2 fuel oil and its
adaption to firing waste oil proved dif-
ficult. Eventually dilution of the waste
oil on a 1:1 basis with No. 2 oil was
required for acceptable operation. Even
with this modification, the combustion
efficiency and destruction efficiencies
were significantly lower than the other
units.
Concentrations of Contaminants in
Combustion Gases '
In Table 2 are shown the concentra-
tion ranges in the stack gas of the com-
pounds studies within the program. In
general, concentrations ranged from 40 to
400 micrograms per cubic meter (ug/m3)
for the volatile compounds and from about
10 to 50 yg/m3for the semivolatile com-
pounds. On a volume/volume (v/v) basis,
these are very low concentrations. As an
example, a range of 40 to 400 ug/m3 for a
compound such as trichloroethylene corre-
sponds to a concentration of 7.4 to 74
parts per billion (ppb) on a v/v basis.
Conducting emission tests at these low
concentrations required extensive refine-
ment of available emission source testing
techniques.
Lead and Other Metal Emissions
The samples of flue gas particulate
collected at each site were analyzed for a
total of 27 metals by Inductively Coupled
Argon Emissions Plasma Spectrometry (ICAP)
techniques. Lead and zinc were present at
concentrations substantially higher than
any other trace metals. The lead concen-
trations in the flue gas samples ranged
from 5,380 ng/m3 to 72,400 yg/m3,corre-
sponding to an average emission rate of
0.12 Ib/hr. Calculations based on simpli-
fied models have shown in some cases, lead
emissions at these levels could cause vio-
lations of ambient air quality standards
for lead. The concentrations of zinc in
the flue gas ranged from 3,100 to 34,000
ug/m3, corresponding to an average emission
rate of 0.06 Ib/hr. The ratio between lead
and zinc emissions was generally 2:1, con-
sistent with their concentration in base
stock oil, which was 1,550 ppm and 760 ppm
by weight, respectively. Lead and zinc
compounds are commonly found in waste auto-
motive oil and result from both gasoline
and oil additives.
Mass flow calculations indicated that
50-60 percent of the lead and chloride
introduced into the boilers exits from the
system in the flue gas. An additional
10-15 percent appears in ash collected in
the firetubes of the boilers.
Other metals which received special
attention were arsenic, cadmium and chro-
mium. These were generally at low enough
concentrations in the stack gas so that
when diluted in the atmosphere they should
not cause major problems, but the situa-
tion is still of some concern as the con-
centration of these metals could be sub-
stantially higher in other waste oil base
stocks.
-------
, The results of metal emissions are
summarized in Table 3.
Particulate and Chloride Emissions
Particulate emission rates at the six
sites ranged from 0.07 to 1.2 Ib/hr with an
average value of 0.73 Ib/hr (0.34 lb/106Btu
heat input). This is significantly higher
than the literature value of 0.09 lb/106Btu
for commercial boilers firing residual
oil,3 but the higher value-is consistent
with the much higher ash content of waste
oil, which can range from 0.15 to 1.5 per-
cent. Particulate sizing measurements con-
ducted at four of the test sites indicated
that 80 to 90 percent of the particulates
containing lead are submicron in nature and
would be readily inhalable.
The flue gas emissions of HC1 from the
six boilers averaged 2.6 Ib/hr. This is a
relatively high emission rate for such
small units, but it is below the 4.0 Ib/hr
air emission standard established for
hazardous waste incinerators, which would
typically burn large quantities of chlori-
nated compounds similar to those used in
this program.
Products of Incomplete Combustion
The flue gas samples from each site
were screened by Gas Chromatograph/Mass
Spectrometer(GC/MS) for additional organic
components, considered to be potential
products of incomplete combustion. The
types of compound which were-identified are
shown in Table 4. In general, the compo-
nents were nonchlorinated in nature and
were representative of the types of com-
pounds that result from the combustion of
traditional fossil fuels.3>4 These com-
pounds were also very typical of contami-
nants sometimes found on the blank sample
adsorbing medium, XAD-2 resin. The extent
to which these compounds, when detected,
resulted from combustion byproducts or from
resin contaminants could not be determined;
hence, the concentrations in Table 4 could
be viewed as upper limits for many of the
nonchlorinated products of incomplete com-
bustion. During the course of this program
there were some baseline runs done on con-
ventional No. 4 fuel oil. As expected, no
chlorinated hydrocarbons were detected in
the stack gas, with detection limits being
8 yg/m3. With conventional No. 4 oil,
combustion products such as naphthalene
and similar polyaromatic hydrocarbon (PAH)
compounds were 100 yg/m3 or less.
Chlorinated dibenzofuran (PCDF) or
chlorinated dioxin (PCDD) species were
detected in 15 of the 25 samples analyzed.
The concentration of these compounds
ranged from 0.07 to 17 yg/m3. On a v/v
basis, this corresponds to a range of 7 to
470 parts per trillion (ppt).
Bulk samples of firetube ash collec-
ted at one of the sites contained parts
per billion levels of 11 PCDF and PCDD.
isomers on a weight by weight basis.
Because chlorinated dioxins and chlorina-
ted dibenzofurans were found in the flue
gas, tests were also conducted on the
waste oil base stock, both spiked and
unspiked to determine the extent to which
these types of compounds might be present
in the oil. No chlorinated dioxins or
chlorinated dibenzofurans were found in
either the spiked or unspiked oil at
detection limits which were 0.2 milligrams
per kilogram (mg/kg) for the tetrachloro-
dibenzodioxin and as low as 0.04 mg/kg for
the monochlorodibenzodioxin. The same
detection limits were achieved for the
chlorinated dibenzofurans.
CONCLUSIONS
Although a sample population of six
boilers is very limited, there do seem to
be several general conclusions which can
be reached regarding the combustion of
waste automotive fuels in boilers in this
size range.
1. It is possible to achieve combustion
efficiencies greater than 99.9 percent for
small commercial boilers firing waste
oils.
2. Destruction and removal efficiencies
of greater than 99.9 percent can be ob-.
tanned for chlorinated organic contami-
nants typically present in waste oils.
For the volatile compounds studies
(Chloroform, trichloroethylene, trichloro-
ethane and perchloroethylene), destruction
and removal efficiencies were on the order
of 99.9 percent. For the semi volatile
compounds, )trichlorobenzene, dichloro-
naphthalene, trichlorophenol), destruction
and removal efficiencies were on the order
of 99.95 percent. '
-------
3. For boilers above 1 X TO6 Btu/hr input,
there were no apparent correlations between
boiler size or firing method and destruc-
tion efficiency of organic contaminants.
4. Inorganic components, as opposed to
organic components of waste oil, have sub-
stantially greater mass emission rates to
the atmosphere, as a result of the combus-
tion of automotive waste oils. The princi-
pal inorganic components of concern are
lead, hydrochloric acid and total particu-
late. 'Also of potential concern are
arsenic, cadmium and chromium. The parti-
al! ate lead emissions from a source may,
during the peak heating season, affect the
compliance with the primary ambient air
standard for lead. A significant percen-
tage of the particulate .lead emissions is
submicron in nature and would be readily
inhalable.
5. Detectable levels of emissions of poly-
chlorinated dibenzofurans and polychlori-
nated dibenzodioxin compounds' were found in
some of the boilers tested. These com-
pounds, when present, were usually at
levels less than 5 yg/m3, which is less
than 0.5 part per billion by volume in the
stack gas. The extent to which these com-
pounds pose a hazard at these low levels is
undetermined.
Tests were done on the base .stock waste
oil, with and without the spiked contami-
nants, to determine the extent to which the
oil may have contained trace levels of
dioxin. No dioxin or dibenzofuran com-
pounds were detected in any of the oil
samples; detection limits were 200 ppb by
weight for TCDD and TCDF. If dioxin com-
pounds were present at or below their
detection limits, such a quantity would
not be large enough to account for the ob-
served levels in the stack gas even with
zero percent destruction. The conclusion
is that dioxin and dibenzofuran found in
the stack gas most probably was formed
during the combustion process.
6. The fly ash deposited in the firetubes
of the boilers may contain percent levels
of lead and parts per billion levels of ,
chlorinated dibenzofuran and dioxin com-
pounds. The ash has the potential for
being classified as hazardous on this
basis, and may be subject to hazardous
waste regulations for disposal.
ACKNOWLEDGEMENTS
This work was done under contract to
the U.S. Environmental Protection Agency
by GCA Corporation under Contract No.
68-02-3168. The authors would-like to
thank our Project Officers, Mr. Harry
Freeman of EPA's Industrial and Environ-
mental Research Laboratory in Cincinnati
and Mr. Mike Petnjska of EPA's Office of
Solid Waste in Washington, DC for their
direction and support. Mr. George
Huffman of EPA's Industrial and Environ-
mental Research Laboratory also provided
important technical direction in the early
stages of the program.
-------
(
h-
p?»
LU
CsC
LU
O_
•* — •*
CO
LU
O
^y
LU
i — i
C_3
i — i
u_
u_
LU
r _ j
o
LU
Q:
o
<
•ZL
o
1— 1
I-
— — 1
o:
1—
LU
O
Q
LU
1 —
3
— i
5
.
•
LU
I
CO
o
I—
to -C
i — O
o
^>
o
CO
CD
CD
CO
CD
CD
CD
CD
CO
CD
CD
n
CD
CD
r-.
01
01
01
LO
CD
CD
CD
00
^
CD
CD
OJ
to
-E
+J
a>
o
S-
o
1 —
CJ
• r—
S-
h-
CM
CO
CD
CD
CO
CD
CD
CM
CD
CD
CD
LO
CO
CD
CD
CD
00
CD
CD
CM
CD
CD
CD
LO
**
CD
CD
0)
E
a>
_c
4-^
O
0
i —
O
•I—
S-
H-
«^J-
co
CD
CD
10
01
01
01
CO
CD
CD
r-.
CD
01
00
CD
01
,_
CD
CD
CD
^.
r^
CD
CD
to
•o
E
CD rs
E O
O) Q.
>5 O
-E O
4J
OJ CU
O i —
S- -r-.
O 4->
r1- tO
u o
S- >
0) .,-
O
cu o. !E
-Q re o
O E 'r-
S- 0 S-
0 S- I—
r— O 1
-El — LO
CJ .E «
•i— O ^d"
1— Q CM
-------
TABLE 2. AVERAGE FLUE GAS CONCENTRATIONS OF THE VOLATILE
AND SEMIVOLATILE SPIKE COMPONENTS
Site
Volatile Components
Chi prof orm
Trichloroethane
Tri chl oroethyl ene
Tetrachl oroethyl ene
Semi -Volatile Components
Trichlorobenzene
Di chl oronaphthal ene
Trichlorophenol
A
no
90
160
160
48
16
10
C
120
88
94
200
20
12
9
D
44
22
66
180
75
22
—
(yg/m3)
E
53
580
62
220
51
23
16
F
88
400
70
170
48
36
6
- G
153
140
200
100
89
89
—
TABLE
3. CONCENTRATIONS
OF METALS IN FLUE GAS
(yg/m3)
Site
Arsenic
Cadmi urn
Chromium
Lead 9
Zinc 5
A
11.2
31.2
62.2
,680
,150
C
655
102
166
72,400
33,700
D
26.
8.
112
5,390
3,134
E
1 106
3 182
230
20,300
12,100
F
251
350
205
49,800
26,800
G
286
81
263
51,000
27,000
-------
TABLE 4. EXAMPLES OF CONCENTRATION RANGES FOR POSSIBLE
PRODUCTS OF INCOMPLETE COMBUSTION
(yg/m3)
Site A
Naphthalene,
cl
Benzaldehyde 10-60
a
Chlorobenzene isomers
Dibenzofuran 70
Ethyl benzene
Chlorotoluene 9,
Benzole acid 70
a
Alkyl benzene, 60
a
C D
30 70
20-70
20
5 2
2-20
70,
20-200
10-100
E
10
30
10
10
—
--
200
100-600
F
40
40
40
20
--
__
300
G
50
20
--
5
--
—
90
300
This compound is a known contaminant of XAD-2 resin and could be a sampling
and analysis artifact as well as a product of incomplete combustion.
, At sites A and D, chlorotoluene was not spiked into the waste oil; hence,
its appearance here is as a product of incomplete combustion.
Note:
The symbol— means the compound was not detected by mass spectrometn'c
computer matching scan. In such cases, the detection limit is generally
100 yg/m3 or less.
-------
REFERENCES
1. Surprenant, Norman F., William H.
Battye, Paul F. Fennelly, 1983.
The Fate of Hazardous and Non-
Hazardous Wastes in Used Oil
Disposal and Recycling, report
prepared by GCA Corporation,
Bedford, MA for U.S. Dept of Energy,
U.S. DOE Report No. DOE/BC/10376-6.
2. EPA Hierarchy of Haste Incinerability-
Organic Hazardous Constituents Under "
Incineration Rules, released June 21,
1982, Environmental Reporter,
June 25, 1983.
3. Surprenant, Norman F., et al., 1980.
Emissions Assessment-of Conventional
Stationary Combustion Systems:
Volume IV: Commercial/Institutional
Combustion Sources, EPA-600/7-79-029e,
U.S. Environmental Protection Agency,
Industrial Environmental Protection
Agency, Research Triangle Park, NC.
4. Hansen, R. L., et al., 1975.
Chemical Characterization of Poly-
nuclear Aromatic Hydrocarbons in
Airborne Effluents from an Experi-
mental Fluidized Combustor, presented
at the Fourth International Symposium
on Polynuclear Aromatic Hydrocarbons,
Battelle Columbus Laboratories.
-------
DEMONSTRATION OF WET AIR OXIDATION OF HAZARDOUS WASTE
William Copa, Ph.D.
James Heimbuch
Phillip Schaefer
Zimpro Incorporated
Rothschild, WI 5W»
ABSTRACT
The purpose of this paper is to suimarize the full scale demonstration of Wet Air
Oxidation of hazardous waste. This work is being done at Casmalia Resources Management,
a commercial waste treater in California.
The report will sunmarize bench scale waste screening tests completed to date, as
well as report the results of full scale oxidation of six specific wastes. Detailed
analysis including Gas Chromatograph/Mass Spec (GCMS) scans of feed and effluent will be
given. Wastes tested include cyanide wastes, phenolic wastes, sulfide wastes, non-
halogenated pesticides, solvent still bottoms and a general organic wastewater.
Preliminary screening tests and the full scale work will be collated such that
predictions might be made for the various groups of compounds.
INTRODUCTION
Wet Air Oxidation is a process
which has been used to oxidize dissolved
or suspended organic substances at
elevated temperatures and pressures.
The process is thermally self-sustaining
with relatively low organic feed
concentrations arid is therefore most
useful for wastes which are too dilute
to incinerate economically yet too toxic
to treat biologically.
The process has been used to treat
various wastes over the last thirty to
forty years. With the recent attention
being focused on hazardous waste, much
interest has been expressed in the use
of Wet Air Oxidation as a means of
destroying and/or detoxifying these
hazardous wastes. During the recent
years much bench scale (5X4) and pilot
plant (2) testing has been performed by
various companies to demonstrate the
applicability of Wet Air Oxidation on
various hazardous organic wastes. Wet
Air Oxidation units currently detoxify
specific waste streams at several
generators (1-).
In this project, a 10 gallon per
minute (gpm) skid mounted Wet Air
Oxidation is used. This unit is located
at Casmalia Resources, a commercial
waste treater in California. Casmalia
Resources is using the unit to process
those acutely hazardous wastes which
have been banned from landfilling by
order of the State of California.
Wastes selected from classified
groups of organic wastes have been
detoxified. These groups include
cyanide wastes, phenolic wastes, sulfide
wastes, non-halogenated pesticides,
solvent still bottoms and general
organic wastewaters. Wastes supplied by
generators in the South Vfest United
States are being tested.
Preliminary autoclave Wet Air
Oxidation tests are run to insure
compatibility and treatability with the
full scale unit. Operating conditions
are also established with these tests.
A continuous test of a minimum of eight
hours is completed with the full scale
unit.
-------
PURPOSE
The purpose of this test is to
demonstrate Wet Air Oxidation of toxic
and hazardous wastes at a full scale
installation. "Data has been developed
on a commercial scale, continuous unit.
Wastes tested were those actually
produced by industrial clients in the
South West portion of the United States.
Results will be collated such that
predictions might be made for all the
compounds of particular classes.
APPROACH
The Zimpro Wet Air Oxidation unit
(see Figure 1), processes aqueous wastes
at a designed reactor temperature of
280 C, a designed reactor pressure of
136 atmosphere (atm) and a liquid waste
flow rate of 2.3 cubic meters per hour
(nr/hr) (10 gpm). Waste is pressurized,
mixed with compressed air, and directed
through the cold, heat-up side of the
heat exchanger. The incoming waste-air
mixture exits from the heat-up side of
the heat exchanger and enters the
reactor v&ere exothermic reactions
increase the temperature of the mixture
to a desired value. The waste-air
mixture exits the reactor and enters the
hot, cool-down side of the heat
exchanger and, after passage through the
system pressure control valves, is
directed to the separator. In the
separator, the spent process vapors
(non-condensible gases) are separated
from the oxidized liquid phase and are
directed into a two-stage water
scrubber-carbon bed adsorber, vapor
treatment system.
In the Vfet Air Oxidation process,
organic substances can be completely
oxidized to yield highly oxygenated
products and water. For exanple,
organic carbon-hydrogen compounds can be
oxidized to carbon dioxide and water,
while reduced oraanic sulfur compounds
(sulfides, mercaptans, etc.J ana
inqrqanic sulfides are easily oxidized.
It'should be noted that oxides of
nitrogen such as NO or N02 are not
formed in Wet Air Oxidation because
reaction temperatures are not suffi-
ciently high to form them.
When incomplete oxidation of
organic substances occurs, the easily
oxidized sulfur and cyanide compounds
are usually still oxidized to sulfate
and carbon dioxide-ammonia provided a
sufficient degree of oxidation is
accomplished. However, incomplete
oxidation of other organic compounds
results in the formation of low mole-
cular weight compounds such as acetal-
dehyde, acetone, and acetic acid. These
low molecular weight compounds are
volatile and are distributed between the
process off-gas phase and the oxidized
liquid phase. The concentration of
these low molecular weight compounds
(measured as total hydrocarbons (THC)
expressed as methane) in the process
off-gas is, dependent on their concen-
tration in the oxidized liquid phase,
which is determined by the degree of
oxidation accomplished; the waste being
oxidized; and the influent organic
concentration of the waste.
Autoclave Screening
In this project, wastes were pre-
screened using autoclave (Photo 1) Wet
Air Oxidation tests. These bench scale
tests generally demonstrate the level of
destruction or detoxification expected.
Corrosion tests were also conducted to
insure compatibility with the full scale
unit.
Raw and oxidized samples were
analyzed for chemical oxygen demand
(COD), biological oxygen demand (BOD),
pH, total solids, ash, soluble chloride,
soluble fluoride and specific component,
e.g. cyanide, phenol, sulfide, pesticide
or solvent. The condensible off-gas
from the autoclave was analyzed for
oxygen, nitrogen, carbon dioxide, total
hydro-carbon and methane. The autoclave
and oxidized liquid effluent was
exanined for any residue which might
cause heat exchanger fouling or plugging
of the full scale unit.
Full Scale Testing
Each waste was tested during a one
(1) day "demonstration period", to
determine the effectiveness of the Wet
Air Oxidation unit. During each
"demonstration period", the sampling and
testing progran outlined as•follows was
used.
10
-------
Upon arrival of a truckload of a
screened and acceptable waste, a sample
of the waste was obtained and analyzed
for COD, pH, and specific component.
During the Wet Air Oxidation of
each class of waste, liquid composites
of the influent raw waste and the
effluent oxidized waste were obtained.
Liquid samples of these two streams were
made on an hourly interval and the
liquid samples were composited over a 24
hour period or throughout the steady-
state operation of the Wet Air Oxidation
unit, if operated less than 24 hours per
day. Split samples were made with one
sample going to the State laboratory and
the other going to Zimpro's laboratory
for analyses. The raw waste and
oxidized waste were each analyzed for
COD, BOD, pH, total solids, ash, soluble
chloride, soluble fluoride and specific
component, e.g., cyanide, phenols, sul-
fide, chlorinated aliphatic compounds,
or non-halogenated pesticide. In
addition, the oxidized waste were also
analyzed for dissolved organic carbon
(DOC) and a GCMS scan was made.
Grab samples of off-gas from the
Wet Air Oxidation unit, sampled after
treatment but prior to discharge to the
atmosphere, were made in Tedlar gas
sampling bags.
The sample of Wet Air Oxidation
process off-gas was analyzed for oxygen,
nitrogen, carbon dioxide, carbon
monoxide, total hydrocarbon, and
methane. Gas sampling was conducted
when the Wet Air Oxidation unit was
operating at steady-state. Two
replicate grab samples were obtained
each day that the waste were processed
in the Wet Air Oxidation unit during the
"demonstration period". The samples
were analyzed by the State.
PROBLEMS ENCOUNTERED
Several problems were encountered
during this project. The most serious
were the delays in obtaining necessary
permits to operate the unit. This
resulted in the obvious delays in
start-up. Once running, it was
necessary to process the available
wastes being brought to the site. The
owner did not wish to run specific
wastes with the added sampling proce-
dures, when wastes were available for
commercial treatment.
A major portion of the wastes pro-
cessed has been plating wastes. With
the destruction of the cyanide and other
organic complexing agents, metal oxides
and hydroxides tend to precipitate.
Adjustments in the operating procedures
and conditions were necessary to prevent
excess fouling and solids deposition.
All classes of wastes to be tested
have been actual wastes except for the
pesticide waste. After searching for
over a year, it has been necessary to
synthesize that waste.
RESULTS
Autoclave Screening Tests
Table 1 lists results of autoclave
screening tests. Corrosion tests showed
that all wastes listed were shown to be
compatible with the materials of
construction of the full scale units.
Most wastes rejected in the prescreening
process were rejected due to a high
corrosion rate. Note that prescreening
of wastes through review of chemical
analysis kept the number of unsuccessful
autoclave screenings to a minimum.
Full Scale Testing (3)
Tables 2, 3, and 4 show the results
for full scale testing. Each table
lists the results of a particular class
of waste. At the time of submission,
complete off-gas data was not available
but Tables 5, 6, and 7 show results for
those tests completed.
Destruction of cyanide was greater
than 99.756. Effluent cyanide was 82
mg/1 with a feed of 25,390 mg/1. The
phenol waste oxidation resulted in
99.77%, destruction with an effluent of
36 mg/1 and a feed of 15,510 mg/1.
Organic sulfur destruction reached 94?
with feed and effluent concentrations of
3010 mg/1 and 180 mg/1. In the general
organics wastewater test dissolved
organic carbon (DOC) destruction was
96.7% with feed and effluent concen-
trations of 20,830 mg/1 and 685 mg/1.
11
-------
While all waste'processing runs
have been completed, complete analytical
data is not available at time of
publication.
REFERENCES
1. Canney, P.J.; and Schaefer, P.T.;
"Detoxification of Industrial
Wastewaters by Wet Air Oxidation"
1983 Spring National AIChE Meeting,
Houston, March 1983
2. Chowdhury, A.K.; Wilhelmi, A.R.;
"Treatment of Spent Caustic Liquors
by Wet Oxidation," 8th Annual
Industrial Pollution Conference,
Houston, June 1980.
3. Dietrich, M.J.; "Commercial
Demonstration of Wet Air Oxidation
of Hazardous Wastes" Interim Reports
1, 2 and 3, October 1983.
1. Randall, T.L.; "Wet Oxidation of
Toxic and Hazardous Compounds",
Mid-Atlantic Industrial Waste
Disposal Conference, University of
Delaware, 1981.
5. Randall, T.L.; Knopp, P.V.;
"Detoxification of Specific Organic
Substances by Wet Oxidation," 51st
Annual Conference; WPCF, Anaheim,
October 1973.
12
-------
TABLE 1. AUTOCLAVE TESTS - CASMALIA RESOURCES
Test Conditions - 280 C, 1 Hr. Reactor Residence Time
Waste
Spent Caustic
Acid Distillate
Alkaline Solvent
Cyanide Plating
Metal Finishing
Cyanide
Cyanide Plating
Pesticide Rinsate
Cadmium Plating
Solvent Still
Bottoms
Rocket Fuel Waste
Mixture
Cyanide Wastewater
Pesticide Wastes
Pesticide
Parameter
COD, g/1
Total Phenols, mg/1
Sulfide, mg/1
COD, g/1
COD, g/1
Total Phenols, mg/1
COD, g/1
Cyanide, mg/1
COD, g/1
Cyanide, mg/1
COD, g/1
Cyanide, mg/1
COD, g/1
Organic N, mg/1
COD, g/1
Cyanide, mg/1
COD, g/1
BOD, 'g/1
BOD/COD
COD, g/1
COD, g/1
Cyanide, mg/1
COD, mg/1
BOD, mg/1
BOD/COD
COD, g/1
BOD, g/1
BOD/COD
Feed
40.0
5350
382
40.1
39.8
840
17.1
6910
40. 4
5090.0
40.2
20960.0
20.0
701.0
11.3
12980.0
43.9
21.3
.49
45.1
29.6
33160.0
1640.0
15.0
0.01
5.38
1.81
Effluent
7.5
66
< 1
5.8
7.5
20
3.2
120
6.0
36.0
8.6
234.0*
4.1
156.0
3.6
52.0
8.6
6.3-
.74
1.8
10.4
185.0
450.0
208.0
.46
.85
.725
% Removal
81.3
98.8
>99.7
85.5
81.2
97.6
81.3
98.3
85.1
99-3
78.6
98.9
79.5
77-7
68.1
99.6
80.4
70.4
96.0
64.9
99.4
72.6
84.2
53.0
* During continuous operation a full scale
7 ppm cyanide or >99.96% destruction.
unit, pxidized product contained less than
13
-------
TABLE 2. WET AIR OXIDATION DEMONSTRATION
OF GULF OIL SPENT CAUSTIC WASTEWATER
Wet Air Oxidation Conditions:
Oxidation Temperature
Nominal Residence Time
Waste Flow Rate
Compressed Air Flow Rate
Reactor Pressure
Residual Oxygen Concentration
515 F (268°C)
113 min.
5.3 GPM
190 SCFM
1610 PSIG
3.7%
Sample Description
COD, g/1
COD Reduction, %
Total Phenols, mg/1
Total Phenols Reduction, %
Total Sulfur, mg/1
Sulfate Sulfur, mg/1
"Organic Sulfur, mg/1
Organic Sulfur Reduction, %
Sulfide Sulfur, mg/1
pH
Total Solids, g/1
Total Ash, g/1
Volatile Solids, g/1
DOC, mg/1
Soluble Chloride, mg/1
Soluble Fluoride, mg/1
Composite Influent
Raw Waste
108.1
15,510
3580
570
3010
13.0
88.6
57.1
31.5
1510
4.4
Composite Effluent
Oxidized Waste
11.6
89.3
36
99.77
3090
2910
180
94.0
<1.0
8.3
59.7
50.2
9.5
3680
550
1.3
"Organic Sulfur - Total Sulfur Minus Sulfate Sulfur
14
-------
TABLE 3. WET AIR OXIDATION DEMONSTRATION .
CYANIDE WASTEWATER CLASS
Wet Air Oxidation Conditions:
Oxidation Temperature
Nominal Residence Time
Waste Flow Rate
Compressed Air Flow Rate
Reactor Pressure
Residual Oxygen Concentration
495 F (275 C)
80 min.
7.5 GPM
190 SCFM
1220 PSIG
7.1%
Sample Description
COD, g/1
COD Reduction, %
Cyanide, mg/1
Cyanide Reduction, %
DOC, mg/1
DOC Reduction, %
PH
Total Solids, g/1
Total Ash, g/1
Volatile Solids, g/1
BOD,-, mg/1
Soluble Chloride, mg/1
Soluble Fluoride, mg/1
Composite Influent
Raw Waste
37.4
25,390
14,710
12.6
' 135.3
112.9
22.4
30
Composite Effluent
Oxidized Waste
4.2
82
99.7
1710
88.4
9.0
91.2
77.4
13-8
603
773
29
15
-------
TABLE 4. WET AIR OXIDATION DEMONSTRATION
GENERAL ORGANIC WASTEWATER CLASS
Wet Air Oxidation Conditions:
Oxidation Temperature
Nominal Residence Time
Waste Flow Rate
Compressed Air Flow Rate
Reactor Pressure
Residual Oxygen Concentration
531 °F (277°O
120 min.
5.0 GPM
190 SCFM
1515 FSIG
M.156
Sample Description
COD, g/1
COD Reduction, %
DOC, mg/1
DOC Reduction, %
PH
Total Solids, g/1
Total Ash, g/1
Volatile Solids, g/1
BOD5> mg/1
BOD5/COD, g/g
Soluble Chloride, mg/1
Soluble Fluoride, mg/1
Composite Influent
Raw Waste
76.0
20,830
1.9
10.0
0.7
9.3
29,680
0.39
212
0.7
Composite Effluent
Oxidized Waste
2.5
96.7
685
96.7
3.3
3.1
1.1
2.0
880
0.35
866
<0.5
16
-------
TABLE 5. WET AIR OXIDATION DEMONSTRATION
SPENT CAUSTIC WASTEWATER (PHENOL & SULFIDES)
Process Off-Gas Analysis
Carbon Dioxide 11.3%
Oxygen H.3%
Nitrogen 83-256
Carbon Monoxide Not Detected
Methane 12.1 ppm
Total Hydrocarbons 85.4 ppm as methane
TABLE 6. WET AIR OXIDATION DEMONSTRATION
CYANIDE WASTEWATER CLASS
Process Off-Gas Analysis
Carbon Dioxide
Oxygen
Nitrogen
Carbon Monoxide
Methane
Total Hydrocarbons
1.55&
8.556
82.8?
Not Detected
9.0 ppm
61.1 ppm as methane
17
-------
TABLE 7. WET AIR OXIDATION DEMONSTRATION
GENERAL ORGANIC WASTEWATER CLASS
Process Off-Gas Analysis
Carbon Dioxide
Oxygen
Nitrogen
Carbon Monixide
Methane
Total Hydrocarbons
12.9%
5.9%
81.2%
0.3%
10.0 ppm
29.1 ppm as methane
FIGURE I. WAO ELEMENTARY FLOW SHEET
FEED HEAT
EXCHANGER
HIGH
PRESSURE
PUMP
REACTOR
-START-UP STEAM
TO SCRUBBER
SEPARATOR
AIR
COMPRESSOR
COOLER
18
-------
EVALUATION OF PILOT-SCALE APCDs FOR HAZARDOUS
WASTE COMBUSTION SOURCES
C.. W. Westbrook
Research Triangle Institute
Research Triangle Park, North Carolina
ABSTRACT
27709
Conventional air pollution control devices used at hazardous waste combustion facili-
ties may not adequately control fine particulate matter and some gaseous species. This
project sought to identify alternate technologies that might provide a higher degree of
control. The performance characteristics of several alternate technolgies is to be
tested at a full-scale incineration facility. This paper is an interim report on the
project.
INTRODUCTION
Combustion of hazardous waste compounds
can potentially generate particulate and
gaseous emissions that may have adverse
health effects on exposed populations.
Conventional control devices are of varying
effectiveness in controlling these emis-
sions. Wet scrubbers are often used to
remove both particulate and gaseous
contaminants in the same device. Wet
scrubbers, however, tend to remove large
particles more efficiently than they
remove small particles, and at high col-
lection efficiencies they require more
energy than do electrostatic precipitators
(ESPs) or fabric filters. These performance
characteristics allow the escape of noxious
fine particulate matter into the atmosphere
from hazardous waste combustion sources.
Advanced devices are expected to be
marketed with higher collection efficiency,
lower energy consumption, or lower cost
than current conventional types of equip-
ment. It is likely that some of these
devices are capable of providing better
emission control than is now available for
hazardous waste combustion products.
In Phase 'I of this project, RTI
assessed the current state of development
of devices that have the potential to
control emissions from hazardous waste
combustion sources and recommended a
number for field testing. Phase II of the
project consists of testing several of
these units on a slip-stream from a
hazardous waste incineration facility.
Currently, Phase I of the project has
been completed. Phase II, Field Testing
of the Units Selected, has just begun.
This paper describes the work completed to
date.
PURPOSE
The purpose of this project is to
identify technologies that are potentially
superior to air pollution control devices
(APCDs) that are in current use for con-
trolling emissions from hazardous waste
combustion facilities and to test the per-
formance of these technologies on
commercial-scale combustion facilities.
APPROACH
The project is being conducted in
several phases. In Phase I, the currently
used technologies and the potentially
superior technologies under development
were identified. The potentially superior
technologies were identified by contacts
with, vendors, research organizations,
literature reviews, and communication with
experts in the field. Both technologies
that have been used on other combustion
sources, but not; widely used at hazardous
waste combustion facilities, and technolo-
gies at various stages of development from
bench-scale to pilot-scale were evaluated.
The primary factor considered in the eval-
uation was potential applicability for
19
-------
controlling submicron particulate matter
and/or HC1 emissions. Secondary factors
considered were the potential for reduced
operating costs and reduced environmental
impacts of the discharge from the devices.
After identification of the technolo-
gies that seemed to have the greatest
potential for controlling emissions from
hazardous waste combustion facilities,
efforts were undertaken, through direct
contact with the developer/marketer of
the technology, to ascertain if pilot
units of suitable size and construction
either existed or could be fabricated.
If the unit existed, discussions were
initiated to determine if the developer/
marketer would be interested in partici-
pating in testing of the unit at a hazard-
ous waste combustion facility.
Concurrent with this latter activity,
discussions were initiated with several
commercial hazardous waste incineration
facilities to solicit their interest and
cooperation in allowing tests of the
selected pilot units at their facility.
Scheduling was coordinated with all
involved parties to pr-ovide sufficient
time for testing, yet not unduly inconven-
ience the host incineration facility.
PROBLEMS ENCOUNTERED
Few problems were encountered in
Phase I of this project. However, RTI
recognizes that some innovative devices
in current commercial use may not have
been identified. A detailed investigation
of all facilities was not undertaken and
some facilities having custom-designed
equipment may have been unwilling to
discuss what might be considered competi-
tive advantages. RTI also recognizes
that it is possible that not all innovative
new technologies were discovered in this
investigation. Technologies that had not
been publicized or that were not reasonably
well-known to researchers in this field
Would have been overlooked.
More significant problems were encoun-
tered in Phase II, Field Testing, that
had an impact on timely completion of the
project. These difficulties arose both
in obtaining units to test and in finding
a suitable host plant at which to conduct
the tests.
In Phase I, we had identified a
number of technologies meriting demonstra-
tion testing. Our study required that
mobile pilot units having capacities of
1,000 to 4,000 cubic feet per minute of
stack gas be available. These were not
available for all technologies. Secondly,
the study would compare the performance
of all tested APCDs on the same source.
Some APCD developers were unwilling to
participate under these constraints,
perhaps because they did not feel their
test unit had yet achieved its potential.
Among the APCD developers who did wish to
participate, a few more mundane problems
arose. These included negotiating the
financial arrangement and coordination of
test schedules.
Delays were also encountered in
locating a suitable host plant for the
study. Our first choice, who had expressed
a strong interest in participating,
encountered difficulties with regulatory
authorities and, after much delay, had
to withdraw. An alternative facility was
sought which met the following criteria:
regulatory authorities will allow the test,
the facility is burning the appropriate
type of waste (contained significant
amounts of ash and chlorine), the facility
design is such that a slip-stream of gas
could safely be taken prior to the plant's
emission control system, the facility
management is willing to allow the testing,
acceptable schedules, and financial
arrangements could be made.
As a result of the difficulties cited
above, a significant schedule delay was
encountered. However, field testing of
the first pilot unit did commence in March
1984.
RESULTS
In this section, we describe some of
the results of this project. However,
since the field testing phase has just
begun, no test results can be provided.
Current Technologies
As might be expected, the principal
generic types of APCDs in current use are
wet scrubbers, baghouses, and electro-
static precipitators. Wet scrubbers,
which have the capability of controlling
both particulate and gaseous components,
predominate. Fine particulate matter is
generally captured poorly, and many
facilities have a difficult time consis-
tently meeting regulatory requirements.
Advanced Devices
A partial listing of the advanced
20
-------
APCDs identified in Phase I is given in
Table 1. The three principal types (ESPs,
wet scrubbers, dry scrubbers) are listed
by the company marketing the device.
Where appropriate, the type of device is
given.
TABLE 1. ADVANCED DEVICES
ESPs
Beltran Inc. .and Micropul, Inc.
(Wet Type)
Wet Scrubbers
Lone Star Steel
(Steam Hydro and Free Jet)
Ceil cote, Inc.
(Ionizing)
University of Washington
(Electrostatic Spray)
Dry Scrubbers
Micropul/Koch
Niro Atomizer, Inc.
ETS, Inc.
Rockwell International
Wet ESPs have been around for sometime
but have not been widely used on hazardous
waste combustion facilities. In these
devices, particulate collection is by
electrostatic forces. Collected particu-
lates are removed with a water spray. \
This reduces re-entrainment problems from
high resistivity.particles.
The Lone Star Steel wet scrubbers
shown in Figure 1 use steam or air to
produce very finely atomized water droplets
which contact the particulate matter,
resulting in an increase in the apparent
diameter. Removal can then be accom-
plished easily.with a cyclone separator.
The Ceil cote ionizing jet scrubber
shown in Figure 2 precharges the particu-
late entering the scrubber. Fine particu-
late adheres to the packing materials by
induced polarization and is removed by the
water spray. The electrostatic spray
scrubber developed at the University of
Washington carries this concept one step
further by also applying a charge to the
water spray.
The discharge from all of the dry
scrubbers is a dry powder. Each differs
in the technology for introducing the gas
scrubbing medium, however. Most of these
units atomize a caustic (lime or sodium1
carbonate) slurry into the gas stream and
use the heat content of the gas to evapo-
rate the moisture. The ETS, Inc. unit
shown in Figure 3, however, injects a dry
powder (either lime or nacholite).
Devices Selected for Testing
At this point, arrangements have been
made to test four pilot APCDs. These are
the two versions of the Lone Star Steel
scrubbers, the Ceilcote, Inc. unit, the
ETS, Inc. unit, and the Hi-Tac unit devel-
oped by Vulcan Engineering Company. Th
Hi-Tac unit, which has not been mentioned
previously, is a specially designed
baghouse that can operate at temperatures
up to 1300°F.
Each of these are mobile pilot units
that have capacities within the range of
1,000 to 4,000 cubic feet per minute of
stack gas.
Host Plant for Testing
ENSCO, Inc. agreed to participate in
testing the selected pilot units at their
El Dorado, Arkansas incinerators complex.
At this facility, oils and capacitors
containing polychlorinated biphenyls
(PCBs) are incinerated. Duct work has
been installed to remove a slip-stream of
gas from the system immediately upstream
of the emission control system. Provision
has been made to cool the gas to the
temperatures required by the various pilot
units.
Test Methods
Simultaneous inlet and outlet testing
is scheduled for each unit. The testing
is comprised of particle size'determina-
tions (by impactor), total particulate
concentration (by EPA Method 5), and HC1
concentration (by adsorption is back half
of Method 5 train). Three Method 5
particulate/HCl tests are conducted each
day. One particle size determination per
day is made. A 2-week test period is
allotted for each unit.
Test Results
Field testing has just been completed
on the first pilot .unit (provided by Lone
Star Steel). Preliminary data indicate
the unit achieved excellent removal of
21
-------
both fine particulate and HC1. The data,
hoy/ever, have not been subjected to a
thorough review at this time. In our
opinion, it is inadvisable to publish data
under these conditions.
ACKNOWLEDGMENTS
This project is being conducted under
EPA Contract No. 68-03-3149, Work Assign-
ments 12-1, 12-4, and 12-9. Harry Freeman,
lERL/Cincinnati, is the EPA Project Monitor.
Engineering Science, Inc., is conducting
the onsite testing under a subcontract to
RTI.
Steam or
Compressed Air
Water
Polluted Gas
Turbulent Mixing
Particulant Wetted
and
Vapors Absorbed
:'-Agglomeration
Supersonic Ejector Nozzle
Figure 1. Ejector Drive for Steam-Hydro Scrubber
Electrode -
Wires
Spray Header Packing
Grounded Plates
>0ut
Figure 2. Ceilcote Ionizing Wet Scrubber
22
-------
Flue Gas
Dry Reactant ,-£
Induced
Draft Fan
Baghouse
Dry Stable
Byproduct
Dry Stable
Byproduct
Controlled
Stack
Figure 3. ETS Dry Reactor
23
-------
DISPOSAL OF HAZARDOUS WASTE IN AGGREGATE KILNS
James A. Peters
Environmental Sciences Center
Monsanto Company
St. Louis, MO 63167
and
Duane R. Day
Monsanto Research Corporation
Dayton, OH 45407
and
Robert E. Mournighan
Incineration Research Branch
U.S. Environmental Protection Agency
Cincinnati, OH 45268
ABSTRACT
Incineration of chlorinated liquid organic waste in an aggregate kiln was investiga-
ted in a one-week program at Florida Solite Company. POHCs (toluene, tetrachloroethylene,
methyl ethyl ketone, and methyl isobutyl ketone) were monitored in the waste and stack
emissions. In addition, stack emissions were monitored for particulate matter, particu-
late trace metals, HC1, S02, and NOX. Process samples were collected and analyzed for
trace metals and chlorine. The destruction and removal efficiency of POHCs and the fate
of trace metals and chlorine in the kiln process were determined.
Consistent achievement of greater than 99.99% DRE was demonstrated for each POHC.
Emissions of other pollutants ranged as follows: particulates - 4.4 to 6.5 kg/hr; HCL -
0.008 to 0.034 kg/hr; S02 - 72.2 to 99.6 kg/hr; NOX - 1.9 to 11.7 kg/hr. Between 60 and
90% of the chlorine is fed to the kiln from the waste fuel and scrubber influent water.
Over 95% of the chlorine is removed from the kiln in the scrubber effluent water.
INTRODUCTION
Cofiring of hazardous wastes in high
temperature industrial processes is an
attractive alternative to hazardous waste
incineration because it utilizes the
waste's heat content. Many cofiring pro-
cesses, which include cement and dolomite
kilns, glass furnaces, steel furnaces, and
some industrial boilers, provide tempera-
tures and residence times similar to those
required for dedicated hazardous waste
incinerators. In addition to the savings
derived from the heat value, the use of
existing industrial equipment does not
require the capital investment required for
a separate or new incinerator to dispose of
a hazardous waste, and it may provide an
environmentally acceptable alternative to
conventional hazardous waste land disposal
options.
Aggregate kilns, because of their high
energy use, are an excellent example of
this concept. Such kilns typically oper-
ate at temperatures over 1093°C (2000°F),
have gas residence times in excess of 1.5
seconds, and have a highly turbulent
24
-------
combustion zone. However, the need exists
for data that shows the environmental
effect of cofiring hazardous waste on the
emissions and effluents from the aggregate
process.
Currently, the State of Florida
Department of Environmental Regulation (DER),
in implementing RCRA regulations, allows
exemptions from hazardous waste incinera-
tor emission standards if the facility
uses the waste as a fuel for the purpose
of recdvering useable energy. The Florida
Solite Corporation located in Green Cove
Springs, Florida, petitioned the Florida
DER for a change in their air pollution
permit to allow the burning of organic
wastes as supplemental kiln fuel, thereby
recovering energy from these wastes and
offsetting their use of pulverized coal.
A sampling and analysis program was con-
ducted at Florida Solite Corporation from
February 20-25, 1983, when burning coal
and hazardous waste as fuel. Program re-
sults will allow the Florida DER to decide
whether the existing regulatory exemptions
are environmentally sound and they will
provide additional hazardous waste incin-
eration data for the EPA Industrial
Environmental Research Laboratory - Cincin-
nati.
The sampling and analysis program in-
cluded evaluation of: (1) the destruction
and removal efficiency (ORE) of principal
organic hazardous constituents (POHCs)
when cofiring coal and waste fuel; (2) the
concentrations of particulate matter, S02,
NOX, HC1, and metals in stack emission, and
(3) the concentration and fate of metals
and chlorine in the process streams.
FACILITY AND PROCESS DESCRIPTION
The Florida Solite Company operates
an aggregate kiln in Green Cove Springs,
Florida, which is located approximately 20
miles south of Jacksonville. Annual pro-
duction of the expanded lightweight inor-
ganic material used as aggregate in a ce-
ment mix is approximately 5.45 x 107 kg
(60,000 tons) per year.
This industrial process involves the
heating of clay to a temperature of 2,000°
F to 2,100°F in a horizontal rotary kiln
to prepare an expanded lightweight inorgan-
ic material used as aggregate in cement mix.
The kiln, with refractory linings, is
2.7 m(9 ft) in diameter and 45.7 m (150 ft)
long. The kiln rotates slowly (90 revolu-
tions per hour) and has a gentle slope
(6.25 cm/m) to allow material to pass
through by gravity. The kiln operates in
a counter current flow pattern; i.e., solid
materials travel in one direction and hot
gases and dust travel in the opposite direc-
tion, as shown in Figure 1. Clay is fed
into the kiln at the upper end at a rate of
approximately 12,260 kg/hr (27,000 Ib/hr).
At the opposite end of the kiln, a mixture
of coal and waste fuel is burned at rates
of approximately 700 kg/hr (1,540 Ib/hr)
and 0.87 m3/hr (230 gal/hr), respectively,
to provide a heat input of approximately
22,000 kw (74 million Btu/hr). As the clay
feed travels down the inclined rotating
kiln, it passes through various temperature
ranges which cause transformation of the
clay into the lightweight aggregate product
The lightweight aggregate is produced at a
rate of approximately 9,080 kg/hr (20,000
Ib/hr). After heating and transformation
in the kiln, the aggregate is graded and
large clumps are crushed for sizing. The
final product is stored in large piles
until sold.
The kiln exhaust gases pass through a
pair of mechanical dust collectors, whose
dust is recycled into the kiln, then into
a horizontal cross-flow water scrubber of
,fiber-reinforced-plastic (FRP) construc-
tion. The series of water sprays clean
the particulate matter and reduces the gas
temperature from about 370° to 70°C (700CF
to 160°F) before the gases reach the knock-
out chamber and fiberglass stack. The
scrubber discharge released from the knock-
out chamber is a mixture of raw steam and
water with the entrapped particulate
matter.' This discharge stream is released
to an open ditch which drains to a pond.
There fs no recycle of the scrubber water.
The fuel used to fire the kiln is an
unblended combination of crushed coal and
waste organic liquids. The liquid wastes,
which are trucked directly from the gener-
ators, consist primarily of solvents,
alcohols, ethers, still bottoms, and a
25
-------
small fraction of chlorinated hydrocarbons.
Any manifested wasteload that contains
pesticides, PCBs, acids, caustics, cyani-
des, sulfides, mercaptans, electroplating
wastes, or metal finishing wastes is re-
jected and returned to the generator. The
organic waste mixture makes up from 50% to
100% of the fuel used. During the test
period, the waste fuel comprised approxi-
mately 54% of the total heat input.
EXPERIMENTAL PROGRAM
The sampling and analytical program
was designed to identify the major pollu-
tants from burning waste fuel in an aggre-
gate kiln, quantify their respective
emission rates, determine the destruction
and removal efficiency (ORE) of the POHCs,
and provide information fora mass balance
around the process for metals and chlorine.
Measured stack pollutants include POHCs
(toluene, tetrachloroethylene, methyl ethyl
ketone, and methyl isobutyl ketone), par-
ti cul ate matter, particulate trace metals,
carbon dioxide, hydrogen chloride, sulfur
dioxide, and nitrogen oxides. In addition,
the distribution of the metals and chlorine
were measured in all of the process input
and output streams; i.e., the coal feed,
waste feed, clay feed, scrubber influent
water, aggregate product, and scrubber
effluent water. Waste fuel and coal sam-
ples were submitted for analyses of sulfur,
ash, and Btu content. Waste fuel and
scrubber effluent water also were analyzed
for principal organics. Table 1 summarizes
the overall test program and lists each
sampling and analytical method use.
Sampling was conducted for five full
test days. Because the kiln rarely oper-
ates on 100% coal fuel feed, the test was
conducted on normal fuel operation (i.e.,
coal and waste fuel) and no baseline (coal
only) testing was performed. A Quality
Assurance Project Plan was reviewed and
approved prior to the test program. A
full description of the QA/QC results
involving replicates, splits, blanks,
spikes, and reference standards is provided
in the final report.
RESULTS AND DISCUSSIONS
WASTE FUEL
A detailed summary of the waste fuel
samples collected as shown in Table II.
Table III and IV show the concentration
of each POHC and other properties for the
five waste fuel samples (one sample per
day, Runs 1-5).
POHC Destruction and Removal Efficiencies
The complex combustion, chemistry for
organic materials becomes perplexing when
a broad range of organic compounds present
in a liquid waste are burned. On a weight
basis, most of the organic carbon in the
waste is oxidized to C02 in the combustion
process, but trace amounts of organic
chemicals survive the oxidation process
and are partially reacted. Accordingly,
the test burn investigated the amount of
destruction of the organic compound in the
hazardous waste.
Destruction and removal efficiency
(DRE) for an incineration/air pollution
control system is defined by the following
equation:
DRE = Uin - Mout (100)
W,
Win
Where DRE = Destruction and removal
w efficiency, %
in = Mass feed rate of principal
organic hazardous constituent
(s) (POHCs) fed to the incin-
erator
™out = Mass emission rate of princi-
pal organic hazardous
constituent(s) (POHCs) to the
atmosphere (as measured in
stack prior to discharge).
DRE calculations are based on combined
efficiencies of the destruction of the
POHC in the incinerator, or the aggregate
kiln in this case, and the removal of the
POHC from the gas stream in the air pol-
lution control system. The presence of
POHCs in liquid/solid discharges from the
air pollution control devices is not
accounted for in the DRE calculation as
currently defined by the EPA. RCRA Part
264, Subpart 0 regulations for hazardous
waste incinerators require a DRE of 99.99%
for all principal organic hazardous
constituents of a waste unless it can be
demonstrated that a higher or lower DRE
is more appropriate based on human health
26
-------
INPUTS
OUTPUTS
Coal
Waste fuel
Scrubber influent water
Clay feed
Aggregate product
Scrubber effluent water
Stack gas
STACK
GASES
COAL
WASTE FUEL
AIR
AGGREGATE
PRODUCT
WATER WATER
IN OUT
Figure 1. Schematic diagram of aggregate kiln process,
criteria. Specification of the POHCs in a
waste is subject to best engineering judg-
ment, considering the toxicity, thermal
stability, and quantity of each organic
waste constituent. For the selected POHCs,
toluene, MEK, and Perc are RCRA Part 261,
Appendix VIII compounds; MIBK is not list-
ed. ORE requirements in the Subpart 0
regulations do not apply to metals or other
noncombustible materials. Currently, ag-
gregate kilns are exempt from trial burn
testing.
Toluene, MEK, and J>1IBK were present
in high concentration for organic compounds
(see Table II). Perc was selected because
the chlorinated hydrocarbons are, in gen-
eral, difficult to thermally destroy and
Perc was the only chlorinated organic
measured in the top twenty constituents of
the wastes fuel.
The four POHCs were sampled in the
exhaust gas by the volatile organic sam-
pling train (VOST) and analyzed by gas
chromatography/mass spectrometry (GC/MS).
The number of acceptable VOST runs made
each day were as follows: day 1 - 0 runs;
day 2-6 runs; day 3 - 6 runs; day 4 - 8
runs; day 5 - 5 runs. The average ORE
and range obtained is shown in Table V.
Figure 2 illustrates the frequency of
obtaining a certain "number of nines" of
ORE for each POHC.
Methyl ethyl ketone was destroyed and
removed to at least 99.99% efficiency.
Only three runs showed DREs less than
99.999%: Runs 4A, 3A3 and 3B3 were side-
by-side runs (with 3A and 3B) that were
split with the EPA QA contractor. Runs
3A3 and 3B3 do not show good comparison
with Runs 3A and 3B for MEK, possibly
because of blank contamination problems
in the QA contractor field blanks. Run
3B3 is an outlier and is not considered a
significant part of the data. The overall
ORE average for MEK for all five days was
99.998% ± 0.006% (95% confidence limits).
DREs for methyl isobutyl ketone (MIBK)
ranged from 99.986% to >99.999%. The
99.986% value was the only ORE less than
99.992%. The overall average for MIBK
was 99.998% ± 0.006% (95% confidence
1 i mi ts).
DREs for tetrachloroethylene (Perc)
ranged from 99.993% to ^99.999%. Excel-
lent consistency was found for each day
damp!ing. Split samples on Day 3 (Runs
3A3, 3B3, and 3D1) all showed low
27
-------
TABLE 1. SUMMARY OF FLORIDA SOLITE AGGREGATE KILN
SAMPLING AND ANALYTICAL PROGRAM
Parameter
Sampling method"
Analytical method
Stack Gas
• POHCs (tetrachloroethylene,
toluene, MEK, MIBK)
Volatile organic sampling
train (VOST)
GC/MS, thermal desorption
and SIM
• Particulate matter
Metals on particulate
• Hydrogen chloride
• C02 and O2
• Nitrogen oxides
• Sulfur dioxide
Waste Fuel
Principal organics
Metals
Chlorine, sulfur
Btu content
Ash content
Scrubber Discharge8
POHCs
Metals
Lead
Hexavalent chromium
Chlorine
Aggregate Product
• Metals
• Chlorine
Clay Feed
• Metals
• Chlorine
Coal
• Metals
• Chlorine, sulfur
• Btu and ash content
Scrubber Influent
• Metals
• Chlorine
EPA 5
EPA 5
Impinger absorption in 0.5 M
NaoAc (back half of EPA 5)
EPA 3
EPA 7
EPA 6
Grab •» composite
Grab •> composite
Grab •» composite
Grab -> composite
Grab •* composite
Grab •* composite
Grab •» composite
Grab •» composite
Grab •» composite
Grab •* composite
Grab •» composite
Grab -> composite
Grab •* composite
Grab •> composite
Grab •* composite
Grab •» composite
Grab •> composite
Grab •» composite
Grab •* composite
EPA 5
ICP
Specific ion electrode
Fyrite
EPA 7
EPA 6
GC/MS
ICP
XRF
ASTM D240-64
ASTM D482-IP4
GC/MS
ICP
AAS
APHA312B
XRF
ICP
XRF
ICP
XRF
ICP
XRF
ASTM D240-64
ICP
XRF
aThe scrubber discharge was split into sludge and supernatant fractions and was
analyzed separately where applicable.
28
-------
TABLE II. RESULTS OF CAPILLARY GC/MS ANALYSIS
OF MAJOR COMPONENTS OF WASTE FUELS
NUMBER 1 AND NUMBER 4
Concentration,
wt %
Waste fuel component
Ethanol
2-Propanol
1-Butanol
Ethyl acetate
Methyl ethyl ketone (POHC)
Methyl isobutyl ketone (POHC)
Toluene (POHC)
Tetrachloroethylene (POHC)
Ethylbenzene
Xylene (isomer No. 1)
Styrene
Xylene (isomer No. 2)
2-Ethoxyethyl acetate ,
Cs-Benzene (isomer No. 1),
Cs-'Benzene (isomer No. 2)
C10-Alkane (isomer),
Cn-Alkane (isomer)
n-Propyl acetate
2-Propanol, l-(2-methoxy-l-methylethoxy)-isomer No. 1
2-Propanol, l-(2-methoxy-l-methylethoxy) -isomer No. 2
2-Cyclohex 4-l-one,trimethyl (isomer)
Number
la
1.55
4.55
1.78
0.68
2.03
1.52
8.40
0.19
1.23
4.47
0.71
1.29
2.03
0.47
0.57
0.83
0.72
1.50
0.46
0.49
1.28
Number
4a
1.83
1.97
0.77
0.72
2.81
1.12
8.06
0.07
2.28
7.89
0.28
2.52
1.20
0.33
0.35
0.76
0.60
1.00
0.14
0.16
0.54
Average of split sample.
Compounds containing three carbons associated with a benzene ring.
cCompounds containing ten carbons associated with an alkane.
Compounds containing eleven carbons associated with an alkane.
29
-------
tJ
M
5
fa
w
EH
VI
f^
15
a
n
w
u
a
CH
•
I-H
t— *
w
LjJ
CQ
2
EH
G
•H
g
\
cn
..
-~^
C
•H
3j_
0)
4J
rei
U)
CO
<0
g
rH
a)
3
<^ in ^^ r*^
in en CM co
CM H i-H
rH
O CO CO CO
O^ C"*1* ^J^ CO
CO CO O
CM
CM in rH P~
co in co en
CO CM CO
^
rH
in o co ^
CM en o in
• • • •
^* CO O !*"•
t-4 CM iO en
CO rH O en
CM rH O r--
CO rH r- rH
CO «* i-H CM
• * • •
rH H O CO
oo o r~-
C-- f~ rH CM
• • • •
rH rH O Cn
en co en co
en in r-j co
H rH O CO
0) OJ
C G
O (U
•P -H
Q) rH >l
££ ^i »G
•P *p
i-H 3 i A 0
£10 (H
-P W O
\
ft >H
W tP
K
f~{
w 5^
f^
^
OJ
rH f.
rd i —
>\
-P -P
ffl PC
0)
fc.
ffl g
o a
P4 a
«.
M
3
4H ^S
3
^
(D
G
-H
M ^^
O
rH
f{
U
M
0)
c
O O O O O
r~ co m <£> r~
vo en oo ^o CM
VO rH CO rH CO
rH CM rH rH rH
>.Q "4* ^ CO CO
in co f- co 10
<* in co CM co
<£> CM CO vO
*«o CM c** r^ ^0
cr» en en o en
O O O rH O
'st* oo r^ co
r- CM •* in rH
• • • • •
r~ c-~ c~- in vo
rH
o o o o o
m in <* co r~-
in ^j< r- in *a
CM rH CM Cf> CM
rH rH rH rH
It)
Q O Q Q O
J? rH Z z 2
rH rH CD VO CM
^ ^ CO CM •*
O O O O O
co oo <* in in
o o o in in
• * • • •
rH rH rH O O
rH CM CO «* in
,
g
ft
rH
II
4J
•H
g
•H
rH
G
O
-H
•P
U
0)
-P
•a
(U
-P
u
-------
Table V. DESTRUCTION AND REMOVAL EFFICIENCIES OF POHCs
DRE, %
Run
number
Day 2
Range
Average
POHC 1
(MEK)
99.999-
>99.999
99.999
POHC 2
(MIBK)
99.999-
>99.999
99.999
POHC 3
(Perc)
99.999-
>99.999
99.999
POHC 4
(Toluene)
99.999-
>99.999
99.999
Day 3
Range'
Average
Day 4
Range
Average
Day 5
Range
Average
99.968-
99.999
99.992
99.998-
99.999
99.999
99.999-
>99.999
99.999
99.998-
99.999
99.999
99.986-
99.998
99.995
99.999-
99.999
99.999
99.998-
99.999
99.999
99.993-
99.998
99.997
99.991-
99.997
99.995
99.999-
99.999
99.999
99.995-
99.999
99.998
99.998-
99.999
99.999
Overall
Average
99.998 99.998 99.997
99.999
relative difference. The overall DRE
average for Perc was 99.997% ± 0.004%
(95% confidence limits).
DREs toluene ranged from 99.995% to
>99.999%. The overall average for toluene
was 99.999% ± 0.002% (95% confidence
limits), making toluene the easiest POHC
to destroy and remove.
VOST S & A Split Results
Because VOST sample tubes are therm-
ally desorbed, analytical splits for QA/
QC purposes are extremely difficult and
typically not attempted. This disadvan-
tage of using a conventional thermal
desorption technique is that once the
sample is desorbed into the analytical
system and is expended, there is no way
of reanalyzing the sample. Therefore,
one cannot run duplicates or adjust for
scaling or instrument sensitivity prob-
lems if the sample is not in an appropri-
ate concentration for the analytical
conditions.
To overcome this analytical deficiency
and provide an estimate of VOST precision,
a test run was specified in the QA Plan in
which two VOSTs were sampled simultaneously
through the'sample port with the probes
tied together. While this was not a true
split, it nevertheless represented a whole
method (S&A) split rather than only an
analytical split.
Only one "within lab" split was speci-
fied; however, while in the field, two
additional "between lab" split runs were
made and these samples were split with the
EPA QA contractor who was conducting a
systems audit. The "within lab" split at
MRC shows excellent duplication and pre-
cision for all POHCs except MEK (relative
difference = 152%). However, even a dif-
ference of 152% for MEK on Runs 3D and 3D1
did not affect the final destruction effi-
ciency of 99.999% for each run, as shown
in Table VI. VOST samples split between
labs (MRC and QA contractor) showed good
duplication for all POHCs except MEK. In
31
-------
30 -
20 -
10 -
Methyl ethyl ketone
12345
Number of Nines ORE
30 -
20 -
10 -
Methyl isobutyl ketone
_L
1 2 3 4.5
Number of Nines DRE
(continued)
Figure 2. Destruction and removal effeciencies.
32
-------
30 -
Tetrachloroethylene
20 -
10 -
12345
Number of Nines DRE
30 -
Toluene
20 -
10 -
12345
Number of Nines DRE
Figure 2 (continued)
33
-------
ca
3
J3
en
•JU
a
u
ex
<
>j»
CQ
^J
CO
i
CO
IK
O
ca
S-|
•J
<
<
p
g
55
M
J
o»
§
03J
ca
u
K
0
,
H-l
"^
C£
§
H
•H
to v
10 3
10
C1-IU4
•"^ fl C
0,0
£1 ^n •
o o
u z
•ml
10
s
•H
3
U
41
0)
Q
to
4J
•H
*o
41 •O —
J4 01 O
2 S3
— 1
S C
•-I 10 CO
a, •-
S-O n
S C 0
CO 10 O
•a co
*M H B
O 10 r-
•a e
• e E
O fl u
ca
u
O 0) »—
M C
••HE
O O.S
Z ca ca
ca
«MJ« 01
O CiH
10 C
••-1 £
OjQ n
Z 0
„
O)
jlative
fferenc
OS-H
•O
•3
1
-O
01 M
!i?M
Se
<*e
D
- CM
31
00
S 1-1
4J }^
p^
a e
10 -H i-l
K tfj
1
04
U
01
§1
"i
CO
o o
01 03
A
Z.H
d
CO
ta
M
a
f-i
en
V.
in
-_-
M
ca
•O
in
M
*
^
O
"*ۥ>
p>|
cn o» in co
cMTfvacM co4*inio co in M co
in co co CM CQ coin co
r- 1-1 u3
«s< vo iH CM m inm
^J«COCOCM CMn^in ^»«oesr^
rt «-i o CM cnmocn t-« CM o «c
•H iH <• ^
CO CO CO
co *H c^ in a» co in
in co ro o -i
O«-IOC4 fHPIOtA CN^Or**
•-• IH
S S S
*.go§ ^gal «^ss
S>-« aj o ut-*ft)o u*-«(uo
•C O* t^ X K CU E^ X X O* t^
•0 TJ TJ
C -• C rt — .^
fO Q (3 < fd CQ
n n fo
a < 03
co co co
*• ~*
? II
egg
•H 41 ft)
£ ?, *,
4* 4J 4J
•H J> t>
•* J3 .Q
«*^ «H^ ^^
1 CM CO
to
a
§
ca
4H
15
i
0)
ca
n
i)
2
5
•o
•0
•3
•H
o
. !
M •>
Bca
M
o c
n n
is 2
I.S-
< *J
01 u
So ca
•H
. c S JS
"So -o e
•-< CO
< 5 -H
g CO (0
4-> A 4J
c 3 e
•HO) 4J U
J? CI O
+3 o -PC
•H CO
» o u
U l-l
^ g IM 0
in -H
co o CM ^
3 » 4->
M 14 -a ia
•H M
t) 10 me
•HO. -03
a* 01 a
a e 4J m
CO O CQ TJ
0) *H 10
•a o . 10 -o
-H J-< 10 C C
ca -H 01 ki <3
i a 1 1 "
I "a n -rt <-i
•O €» J2 p
•H -H CO 43 U
ca IM IH -H «i
to ow * M
3 * O tJ 0
O +•! il
o «-t ^ o
C CO « ^ 33
« oi 4J a> 5
4J 'Q n M ^^
•-* 3 4J
3 r-4 41 M
B 0 14 4 3
I/I 1-1 IM H p4
<0 J3 O tT
34
-------
this case, the DREs for MEK were affected
(I.e., Run 3A ORE = 99.999%, Run 3A3 ORE =
99.985%). Causes for this difference in
MEK results are uncertain but may include
contamination of the tubes with MEK and
analysis performed by different labs.
Stack Samples
Results for particulate, hydrogen
chloride, sulfur dioxide, and nitrogen
oxides are summarized in Table VII. Par-
ticulate emissions of 5.3 kg/hr (11.7 Ibs/
hr) were less than air permit regulations
for this site issued by the Florida Depart-
ment of Environmental Regulations (DER) of
8.82 kg/hr (19.43 Ib/hr). The first S02
test result had a low value of 270 ppm and
is considered an outlier when compared to
the remaining seven S02 test results
which ranged 1,030 to 1,470 ppm. The low
NOX value of 40 ppm was expected as it
occurred on Day 3 during start-up of the
kiln.
Chlorine and Metals Balance
As shown in Table VIII, the amount of
chlorine contributed by the waste fuel is
61-85% of the total chlorine to the kiln.
The scrubber influent water contributes
the remaining 15-39% of the chlorine to the
kiln. Virtually all the chlorine leaving
the kiln does so in the scrubber effluent.
Table VIII shows material balance for
selected metals. The clay feed and waste
fuel contribute the largest percentage of
metals to the kiln process, while the
aggregate product and scrubber effluent
filtrate contain the majority of metals
leaving the kiln. Difficulties in achiev-
ing good material balance closure include
the crude estimates for coal usage, clay
input, and product output rates (not
metered) and the analytical variability
inherent in ICP analyses. The goal of
determining the major contributers of cer-
tain elements and their ultimate fate were
met, however.
TABLE VII. AVERAGE RESULTS FOR STACK GAS, PARTICULATES,
HC1, S025 AND NOX EMISSIONS
Parameter and unit
Stack flow rate, m3/min
Stack moisture, %
Stack velocity, m/sec
Parti culates
mg/dscm
kg/hr
HC1 , ppm
SO 2, ppm
NOX, ppm
Range
623
21.5
16.6
163
4.4
0.15
270
40
- 673
- 28.8
- 17.1
- 273
- 6.5
- 0.68
- 1,470
- 227
Average
652
26.2
16.8
215
5.3
0.46
1,130
162
Standard
deviation
19
3.3
0.2
48
1.0
0.22
380
67
CONCLUSIONS
The results of the program were as
fol1ows:
Constant achievement of at least
99.99% destruction and removal ef-
feciency was demonstrated for each
-POHC (MEK, MIBK, Perc, and toluene)
in the aggregate kiln process.
Emission of pollutants were determined
and ranged as follows: particulates
- 4.4 to 6.5 kg/hr; HC1 - 0.008 to
0.034 kg/hr; S02 - 72.2 to 99.6 kg/hr;
NOX.- 1.9 to 11.7 kg/hr.
Approximately 60-90% of the chlorine
is fed to the kiln from the waste
fuel, while virtually all the chlor-
ine exits the .kiln from the scrubber
effluent water.
The major percentage of any metal is
fed to the kiln from the clay feed
and waste fuel, while the major per-
centage of any metal exits the kiln in
the aggregate product and scrubber
effluent filtrate.
35
-------
W
o
3
i— 1
<
CQ
i_*j
*3«
i_{
&
V}
EH
(rf
g
n
p«*
O
J
8
J-H
^
U
<
EH
-
i-H
(0 (1) (U
•HUM
M C 3
0) <0 W 5-5
4-> rH O
(0 (tf i—H
S XI U
rH -- M
(0 4-> X!
O O X)
EH rH
S-S
rH
•H
K^
S
o
4H
M
•H
OS
•H
(U
(C
S
U CO
fd (d
•p &
CO
M -P
0) C
XI -3
^J rH
M ^H
U s
c
rH
•H
v|
O
4J
U
rH
It!
•H
M
OJ
•P
rd
S
fC Q)
rH (U
u MH
0 C
xi oj
XI 3
SrH
1 1 'j [
u d
CO-H
a)
W F^ ^i
IO VO CT> rH
rH
CM <* O 00 rH
• • • • •
00 **D CM *3* O
rH CM CM rH rH
rH rH rH rH rH
V V V V V
CT» O O CT» CTv
CT> O O CT* CT»
H rH
rH rH rH rH rH
V V V V V
CTI in CTI c*^ oo
• • • • •
r- oo CM CM in
CM CO CM rH rH
rH rH H rH rH
V V V V V
in CT» c^ co CTt
rH rH rH CO CO
in rH CO t~- rH
00 00 00 ^D ȣ>
rH rH rH rH rH
V V V V V
rH CM CO <# in
36
-------
(3
W
O
s
,_q
^
CQ
«
^
n
p*j
w
EH
S
<
EH
W
^
W
U
$
EH
.J
U
I-H
^
1— 1
1 1
W
, 1
PQ
EH
rH
re
4-J
O
EH
1
4-
C/J
C
•H L,
•M <1>
it
S-. Li
W-l O
W
c
u
OJ
CL,
rH
ID
4-J
O
C
i-H
•H
O
*••
4-»
4-J
C
D
•H
tw
4-1
(U
~H ^5*
CM n
o
**H rH r — 1^)
co o OT i— i ^ m
c--cvio<*onr~ocoo
CM t-i 1-1 m
r-H
V V V V V V
(nSincMSfoSScSS
CM i— 1 CO CO ] ^« fo ^« . .^.
V i-t
'
'X>COi-l^}'\Dir)CO O CM CM C— CM
'm
to m c^ r- r-i
0%ino^vDr-aioocnrHn
V V V V V V V
i— | ff) i*-\ r^« CQ ^ | r-,j p^, jrt rQ
V«*VCMCM rH OOUD
•H^ffOCMrH^-l^rHr-li-l
V V V
f— i re c) cr> LI o^ re -H xi C
rtJCQMUOSSSpHN
Ul
§
OJ
V-j
O
tn
DI
re
(U
m
a)
re
in
37
-------
EFFECTS OF DISPOSAL OF HAZARDOUS WASTES IN CEMENT KILNS
ON CONVENTIONAL POLLUTANT EMISSIONS
James A. Peters
Environmental Sciences Center
Monsanto Company
St. Louis, MO 63167
and
Robert E. Mournighan
Incineration Research Branch
U.S. Environmental Protection Agency
Cincinnati, OH 45268
INTRODUCTION
To turn the old adage around,
"one man's poison is another man's
meat" has taken on a more literal
meaning as the RCR_A regulations for
hazardous waste incineration have
become promulgated [1]. Because
many industrial waste products can
be readily used as fuels or proces-
sed into supplemental fuels, a
waste fuel market has been rapidly
developing in the U.S. [2]. if
reprocessed waste liquids do not
contain significant quantities of
toxic metals, halogenated materi-
als, or any publically sensitive
organics (such as PCBs), they can
be economically substituted for
coal, coke, oil or natural gas in
many industrial processes. The
list of high temperature industrial
furnaces/processes is larger than
one may think, and many of these
already burn hazardous waste as
supplemental fuel at some facility:
cement kilns (both wet and dry
process), lime and dolomite kilns,
fuller's earth (kitty litter)
kilns, steel production blast
furnaces, phosphate rock calciners
and dryers, iron one dryers, brick
and tile tunnel kilns, mineral wool
furnaces, and glass melt furnaces.
Cement kilns, in particular,
have become the process heater
leader in using industrial wastes
as supplemental fuel for several
reasons: 1) the product (clinker)
cost is extremely energy intensive
and any fuel cost savings gives an
economic/competitive advantage; 2)
cement clinker quality is relative-
ly insensitive to addition of trace
impurities; 3) cement plants are
typically located near population
and industrial centers; 4) kiln
temperatures and gas residence
times are well in excess of that
seen in most hazardous waste incin-
erators; and 5) regulations enacted
under Subtitle C of RCRA provided
incentives for waste disposal out-
side of conventional hazardous
waste incineration under the aegis
of energy recovery.
The Incineration Research
Branch of EPA's Industrial Environ-
mental Research Laboratory-Cincin-
nati has has been researching haz-
ardous waste incineration in indus-
trial furnaces since 1977. A num-
ber of test programs on waste in-
cineration at cement plants have
been conducted since 1974 in Cana-
da , the U.S., and Western Europe.
The tests documented in the avail-
38
-------
able literature include:
changes in emissions occur.
1. St. Lawrence Cement
Mississauga, Ontario [3,4]
• waste oil and high chlor-
inated organic waste with
No. 6 fuel oil a dry pro-
cess kiln equipped with
suspension preheaters,
air pollution control by
ESP.
2. Stora Vika Test Center,'Stora
Vika, Sweden [5,6]
• highly chlorinated organic
waste, PCBs, and Freon TF
with coal in a wet process
kiln with ESP.
3. Marquette Cement, Oglesby,
Illinois [7]
• low chlorine (<5%) or-
ganic waste with coal in
a dry process kiln with
ESP.
4. San Juan Cement, Dorado,
Puerto Rico [8]
• highly chlorinated organic
waste with Bunker C fuel
oil in a wet process kiln
with fabric filter.
5. Rockwell Lime, Manitowoc,
Wisconsin [9]
• low chlorine organic
waste with coke and
natural gas in dry
process dolomite kiln
with fabric filter.
6. Los Robles Cement, Lebec,
California [10]
• low chlorine organic
waste with coal in
a dry process kiln
with fabric filter.
This paper summarizes the
available information on changes in
conventional pollutant emissions
when burning hazardous wastes. The
example test programs have been
chosen because testing was conduc-
ted during primary fuel firing
(baseline) conditions in addition
to waste combined with primary fuel
firing conditions. This comparison
testing allows the tools of statis-
tics to be used to evaluate whether
STATISTICAL TESTING FOR CHANGE
In a variety of situations,
decisions must be made concerning
whether a new set of test results
represents a significant change
when compared to the old or typi-
cal test results. Sometimes the
results of a change may be obvious.
In other cases, however, it may
appear that a change has occurred,
but it is not clear that the change
is enough to justify the claims or
problems that can accompany such
changes. Decision-making becomes
more difficult in such a situation,
and a decision to endorse or con-
demn a new process, based on data
from small samples, can be diffi-
cult.
In the context of this inves-
tigation, this means that one wants
to know "if the conventional pol-
lutant emissions from a kiln burn-
ing hazardous waste as supplemental
fuel are truly different from nor-
mal emissions." One wants to be
confident that the results are
correct and spend no more money
than is really necessary. Statis-
tical design and interpretation is
the only way to meet this objective
[11] . A fairly simple procedure
called the "one-sided t-test"
enables us to decide, to a subjec-
tively selected degree of certain-"
ty, whether a statistically signi-
ficant difference exists between
one set of data and another (assum-
ing that the sample populations
have approximately normal distri-
butions ). ) '
If one takes stack samples
from the same stack ,on two separ-
ate occasions and analyzes those
samples, the measured concentration
will nearly always be different for
the two samples. There are proba-
bly four principal sources of this
variability:
• variability resulting from
actual changes in pollutant
concentration due to the ef-
fect of two different opera-
ting conditions (normal and
39
-------
waste firing)
• variability observed and inher-
ent in a large-scale combustion
process
• variability introduced as a re-
sult of the sampling procedure
• variability in the analytical
procedure
To determine whether a change
has occurred, one must subtract or
otherwise take into account the
variability from the process and
sampling and analysis, and simply
compare the actual concentrations.
Proper application of the t-test
makes this distinction possible
[11].
All that one needs to use the
t-test is a computation of the mean
and standard deviation of each data
set, one-button features on modern
scientific hand calculators. In
comparing two means, it is not nec-
essary that they be made up of the
same number of terms. A mean of
14 measurements made under one test
condition can be compared with the
mean of 6 measurements made under
another test condition. As long as
the number of measurements, repre-
sented by nA and iig, are correctly
applied in the equations and the
degrees of freedom corresponding
to n. + ng - 2 are used, the re-
sulting t-test will be applicable.
The t-value is calculated us-
ing the following equation:
t =
XA -
"B
where XA and Xg =
n. and
SA and SB
two separate means
number of values
from which XA and
Xj, are, respec-
tively, the means
standard devia-
tions of each
data set,
This calculated t-value is com-
pared to a tabulated value of t at a
known probability level, usually 5%
or 1% (equivalent to a 95% or 99%
confidence level). When the calcu-
lated t-value is less than the crit-
ical tabulated t-value, we say there
is no significant difference between
the means. What we are really say-
ing is that with the observed vari-
ation of the measurements involved
we cannot tell whether the observed
difference between the means is due
to some real cause or is due to
fluctuations in the measurements.
If t-calculated is greater than t-
tabulated, then mean Xn is signifi-
-c\
cantly different than mean Xg.
One must always remember that
statistical tests are based on pro-
bability, consequently there is
some chance that the wrong conclu-
sion is reached. The choice of the
probability level defines the de-
gree of chance that the wrong con-
clusion was reached. For instance,
the use of 0.05 probability level
(95% confidence level) means that
the conclusion reached from using
the t-test has a 5% chance, or 1 in
20, that it is incorrect. Con-
versely, you could give odds of 20
to 1 that the correct conclusion
was reached.
PARTICULATE MATTER
An excellent review and dis-
cussion of the formation of partic-
ulate emissions when chlorinated
wastes are added as fuel to a ce-
ment kiln is provided by Weitzman
[12]. Data from the St. Lawrence
Cement and Stora Vika tests were
re-examined, compared, and plotted
as a function of chlorine feed rate
with respect to clinker production
rate, which clearly demonstrate a
linear increase in particulate
emissions as a function of chlorine
input. The hydrogen chloride com-
bustion product reacts with the po-
tassium, sodium, and calcium com-
pounds in the hot kiln to form the
respective chlorides. These vola-
tized chloride salts are carried by
the hot gas to the cold end of the
kiln where they condense at about
800-900°C to form a fine particu-
40
-------
late. The character of this new
particulate and resultant change in
dust resistivity lowered the elec-
trostatic precipitator (ESP) col-
lection efficiency and produced
significantly increased dust emis-
sions during these two chlorinated
organic waste firing tests. Based
on his presumption of formation of
very fine particles, he prognosti-
cated that a kiln using a fabric
filter rather than an ESP for par-
ticulate ,control would encounter
similar changes in efficiency.
Two other tests conducted at
kilns equipped with an ESP—waste
oil at St. Lawrence Cement and low
.chlorine (<5%) waste at Marquette
Cement—did not demonstrate an in-
crease in particulate emissions.
Comparison between baseline and
waste firing at Marquette Cement
showed no statistically significant
difference, an expected result when
one examines the linear relation-
ship between chlorine input and
particulate emissions. A statis-
tically significant difference was
seen, however, for the nonchlori-
nated waste oil experiments at St.
Lawrence Cement—a decrease in par-
ticulate emissions. This improve-
ment was attributed to the addi-
tional water introduced (in the
waste oil), which been shown to
improve cement kiln ESP efficiency
even at gas temperatures of 350°C
[3].
The particulate emissions from
the kiln tested at San Juan Cement,
equipped with a fabric filter dust
collection system, did not exhibit
an increase with progressively lar-
ger chlorine input as shown in
Figure 1. The linear regression
curves from the St. Lawrence Cement
and Stora Vika tests are also dis-
played for comparison. It can be
seen that, not only was more
organic chlorine fed to the kiln at
San Juan Cement than the previous
highly chlorinated waste tests, the
fabric filter system was able to
maintain a high collection effi-
ciency regardless of the changed
character of dust from the kiln.
An ancillary part of the San Juan
Cement test program was Extraction
too
IS
II
CHLORINE FEED
(KG CHLORINE METRIC TON CLINKER)
Figure 1. Variation of particulate
emissions with chlorine
feed at the three high
chlorine feed to kiln
tests.
Procedure (EP) Toxicity tests on
the baghouse dust collected during
waste firing. Also, the same dust
was extracted and analyzed for
polychlorinated dibenzodioxin and
furan isomeric groups. None of the
compounds tested for were found in
these analyses and the baghouse
dust EP Toxicity tests results
matched the landfill suitability
results obtained in a much larger
survey of cement kiln dusts [13].
Thus, the addition of substan-
tial amounts (>5%) of organic chlo-
rine in waste fuel is the "bad
actor" in lowering ESP collection
efficiency, but no drop in control
effiof fabric filters will result.
An ESP vendor could "fine-tune" the
ESP to overcome the change in dust
resistivity as long as the ESP is
well maintained.
Noteworthy in all the tests
described above was that particu-
late concentrations were well below
the RCRA performance standard of
180 mg/m3 for a hazardous waste
incinerator.
PARTICULATE LEAD
The choicest waste streams for
waste fuel use in kilns are gener-
ated by paint, ink, and coating
41
-------
users because of the high Btu con-
tent of solvents and alcohols.
These waste fuels usually contain
suspended solids, with a high con-
centration of inorganic metals such
as lead. The fate of lead in the
waste incineration process is of
particular interest because the
health effects are recognized, the
concentrations in the waste fuel
are often significant (>1000 ppm
Pb),and implications are that it
may become widely dispersed in the
environment [14].
Comparison measurements of
lead emissions were conducted at
three of the kiln tests described:
St. Lawrence Cement in 1974 burning
a waste oil contaminated with 6,000
ppm (0.6 wt. %) lead; Marquette
Cement burning waste solvents con-
taminated with 1,050 to 2,520 ppm
lead; and Rockwell Lime burning
waste solvents contaminated with
128 to 139 ppm lead. These three
tests provide distinctly different
rates (almost order-of-magnitude)
of lead input to a kiln.
Lead emissions at Rockwell
Lime increased slightly with waste
firing to 1.7 g/hr, and a statis-
tically significant difference was
noted at the 95% confidence level.
The measured emission rate of
1.7 ± 0.25 g/hr (standard deviation
included) is equivalent to an auto-
mobile driven at 30 mph using
leaded fuel. An automobile travel-
ing at 60 mph will emit 3.54 g/hr
and also re-entrain 1.9 g/hr of
fugitive lead dust from a paved
highway [15].
At Marguette Cement the addi-
tion of waste firing reduced lead
emissions by almost one-half, but
this was not a statistically signi-
ficant difference. Apparently, the
substitution of waste fuel for coal
caused the reduction in lead emit-
ted. The lead emissions while burn-
ing waste fuel contaminated with
1,050 to 2,520 ppm lead is equiva-
lent to emissions from ten automo-
biles traveling 30 mph (neglecting
re-entrained dust contributions).
A similar reduction in lead
emissions of almost one-half was
measured at St. Lawrence Cement
when contaminated waste oil was
burned. Because more tests were
run and better precision was
obtained with-in each data set
condition, a statistically signi-
ficant difference was noted for
these data. Again, large energy
substitution for coal may have
caused the lead emission reduc-
tion. The waste oil firing emis-
sion rate of lead was equivalent
to three automobiles traveling at
30 mph.
CARBON MONOXIDE
As part of the regulatory
requirements under RCRA, carbon
monoxide is continuously monitored
in the exhaust gas of hazardous
waste incinerators as an indicator
of combustion efficiency (not as a
surrogate compound for DRE's).
Moreover, the waste feed cutoff sys-
tem is coupled to a maximum CO level
specified in each incinerator's Part
B permit, established from Trial
Burn results.
The term "combustion effi-
ciency" is used as a measure of a
combustion unit's performance. Con-
ceptually, this term defines the
percentage of emissions that are
completely oxidized to CO2, and is
mathematically defined as:
% CE =
COg
CO., + CO
x 100
where CO2 and CO are expressed in
ppm by volume. More rigorous defi-
nitions of combustion efficiency can
also include denominator values for
THC (expressed in ppm volume as
methane) and Cp (expressed in ppm
volume assuming smoke particulate
as gaseous elemental carbon and
ideal gas, 2.03 L/g).
Cement production is a calcina-
tion process in which large amounts
of CO2 are liberated from the feed
to the exhaust gas. In fact, the
CO2 from calcination is more than
double the CO2 from combustion and
confounds a combustion efficiency
determination.
42
-------
CO concentrations in cement
kilns can range from 10 ppm up to
occasional excursions in the percen-
tage range; a smooth kiln operation
is typically indicated by CO concen-
trations of 50 ppm or less. Our ob-
servations of CO from kilns have
convinced us that any process change
creates significant CO excursions,
independent of waste firing. Some
examples which temporairly increase
CO concentrations are changes in the
primary air/fuel ratio, any change
in clinker cooler air (secondary
air), irregular solid fuel feed
rates, and changes in exhaust damp-
er setting. These CO "spikes" are
indistinguishable from the increased
CO observed for waste fuel introduc-
tion and shutoff at San Juan Cement
[16]. Thus, CO is more an indicator
of process stability in calcination
kilns than combustion efficiency.
At two other kiln tests—Los
Robles Cement and Rockwell Lime—
where CO was monitored continuously,
seemingly different results were ob-
served. CO concentrations at Rock-
well Lime fluctuated often during
both baseline and waste firing per-
iods such that the mean concentra-
tions were similar and the standard
deviations large. Naturally, the t-
test conclusion was that no signifi-
cant difference exists.
On the other hand, the Los Ro-
bles Cement kiln demonstrated excel-
lent process stability, i.e. low CO
concentrations. During three waste
firing test periods the maximum CO
concentration was 100 ppm, a value
which is easily exceeded due to any
process change. An earlier series
of baseline tests has a CO maximum
of 618 ppm with daily average CO
concentrations 2-6 times higher than
daily averages during waste firing.
Although the raw data were not
available for statistical testing,
the very narrow ranges of CO concen-
trations reported during waste fir-
ing suggest that a t-test would show
that a significant difference lower.
This would be a misleading use of a
statistical tool given the experi-
ences described above.
TOTAL HYDROCARBONS
Total hydrocarbon analyzers
using a flame ionization detector
have been proposed as possible
regulatory tool for testing and
online evaluation of incinerator
performance [17]. The evaluation
method for hazardous waste inciner-
ation involves determining the
total unburned hydrocarbons in the
combustion gases and calculating
the overall destruction of the
hydrocarbons in the feed. The
theory is that an incinerator pro-
ducing significant amounts of un-
burned hydrocarbons is not approach-
ing thermodynamic equilibrium and
will not pass the overall perform-
ance test of 99.99% DRE of indi-
vidual compounds.
This .proposed monitoring tool
may be workable for incinerators
and even boilers which burn waste
fuels, but testing experience has
shown that kilns do not exhibit
enough change in total hydrocarbon
(THC) emissions. The main reason
is that the primary fuel still
comprises 60% or greater of the
energy input to the kiln. Unburned
carbons, THC emissions, have a
baseline level which exists for
this fuel, typically from 4 to 8
ppm THC (quantified as methane).
The addition of a secondary waste
fuel simply does not cause enough
change in THC emissions, which,
measured continuously, exhibit
very smooth concentration curves
and no wide fluctuations. A kiln
process change which causes a CO
spike goes unnoticed on the THC
analyzer.
Results from the Stora Vika
tests show that THC concentrations
were less than 10 ppm for baseline,
chlorinated aliphatics fired, and
Freon TF fired, but rose to 10 ppm
when chlorinated •aromatics were
burned. The data were not in a
form amenable to statistical analy-
sis. At San Juan Cement a statis-
tically significant difference (and
increase) was observed when chlor-
inated aliphatics were burned to
partially offset fuel oil, presum-
ably because the addition of high
43
-------
levels of chlorine made complete
combustion more difficult. At
Rockwell Lime and Marquette Cement,
however, a significant difference
(and decrease) was observed. At
both of these tests, a low chlorine
waste fuel was burned at high ener-
gy substitution to offset solid
fuel and complete combustion was
probably enhanced.
NITROGEN OXIDES
Nitrogen oxides are formed by
the oxidation of nitrogen (N2) in
air or nitrogen compounds in fuel.
Because nitrogen has oxidation num-
bers ranging from -3 to +5, it is
possible to form many different
oxides. Nitrogen oxides (NOx) con-
sist mainly of NO and N02, with a
typical ratio of NO2 to NO of 2:1.
Most current sampling and analytical
techniques rely on oxidation of NO
to N02, and report concentrations
as NOx.
In industrial kilns there is a
direct functional relationship be-
tween NOx emissions and secondary
air, which is heated air from the
product cooler. There is usually
little to no NOx formed from nitro-
gen in fuel relative to NOx from
combustion air. Kilns are gener-
ally operated as "poor" combustors
when compared to boilers; they have
a reduced primary air/fuel mix,
poor secondary air mixing, and
often require a confined, narrow
flame configuration for product
quality. Typical NOx concentra-
tions in the flue gas can range
from 200 to 1500 ppm, even within
hours at the same kiln. NOx levels
are very operator-dependent; contin-
uous monitoring has shown that at
some kilns NOx " levels vary widely
from hour to hour, and others can
maintain steady NOx concentrations.
The absence or addition of waste
firing has little to do with NOx
variations. The effect of excess
air and its temperature are the
main determinants in kiln NOx
emissions.
At both Marquette Cement and
San Juan Cement a significant dif-
ference (decrease) was observed
with the addition of waste firing.
This may have been the result of
lowered oxygen input more than ad-
dition of waste firing. At Los
Robles Cement the NOx emissions
decreased slightly but it was not
statistically significant. A sig-
nificant increase was observed at
Rockwell Line, but this was due to
operator intent to increase the NOx
emissions during waste firing.
A definite relationship between
O2, NOx, and SO2 in calcination
kilns has been shown [18]. NOx
emissions increase with increasing
O2 input and degree of preheating,
while S02 emissions decrease. For
instance, a 38% reduction in NOx
due to O2 lowering can cause a 47%
increase in SO2. Continuous moni-
toring at Rockwell Lime illustrated
that NOx and SO2 change simultane-
ously, as shown in Figure 2. More-
over, the NOx/SO2 change is followed
within 5-15 minutes by a CO spike,
the kiln's indicator of a momentary
process instability. As shown in
Figure 2, this event was recurring.
ACID GASES
Cement kilns are, by their
alkaline nature, built-in control
devices for acid gases such as SO2
and HC1. In a well operated cement
kiln with medium-to-high feed alka-
linity, SO2 will be absorbed
1000 1030 1100 1130 1ZOO 1300 1!30 1330 1400 KM
Figure 2. Illustration of inter-
relationship between
NOx, SO2, and CO emis-
sions in a lime kiln.
44
-------
(scrubbed) in the kiln environment
and alkaline dust of the APCD to
control efficiencies of greater
than 90%. For HC1 emissions a re-
moval of efficiency greater than
99% was observed at San Juan Cement,
where high chlorine levels were
added to a relatively low alkaline
kiln.
Waste fuels typically do not
contain much sulfur and thus the
partial substitution of waste fuel
for fuel oil or coal will reduce
sulfur input to the kiln. This
will result in even less theoretical
SO2 to be formed. At Marque tte
Cement, where a 13% energy substi-
tution was fired, SO2 stack gas
concentrations were reduced from
93 ppm to 18 ppm. At Los Robles
Cement no significant difference in
SO2 emissions was noted. SO2 emis-
sions increased significantly at
San Juan Cement, from 280 ppm to
450 ppm, probably for two reasons:
the NOx emissions were reduced—
indicative of lower O2 input—and
the high rate of chlorine input may
have caused a competitive acid gas
"scrubbing" situation—HCl was
preferentially absorbed over S02.
The kiln operator's choice of O2
input will have a greater effect
on SO2 emissions than .addition of
a waste fuel. It is possible to
operate a cement kiln with less
than 50 ppm SO2 in the stack gas.
At Rockwell Lime, however, a
different mode of operation was
required; SO2 emissions are inten-
tionally not absorbed because the
presence of sulfur in lime is unde-
sirable. No significant difference
was observed for SO2 emissions,
which remained between 500-600 ppm
regardless of waste firing.
CONCLUSIONS
The foregoing technical exam-
ples of burning liquid hazardous
wastes as a fuel supplement in
cement and lime kilns in Canada,
Sweden, Illinois, Puerto Rico,
California, and Wisconsin have
demonstrated that, often times,
emissions of conventional pollu-
tants are reduced and that process
operation changes have a greater
effect on changes in emissions than
the partial substitution of waste
as fuel. It is a tribute to the
expertise of the operators at each
plant tested that they were able to
make product while burning a new
and different fuel mix. A new
hazardous waste incinerator is per-
mitted 720 hours of "shakedown"
operation before a trial burn is
conducted; these test situations
were not allowed that luxury and
the operator's skill in burning
waste fuels could only improve
with practice.
Before kilns are allowed to
routinely burn hazardous waste
liquids, monitoring devices capable
of assuring the environmental ade-
quacy of the burning operation
should be identified. The amount
of information now available con-
cerning process variables is insuf-
ficient to identify a single ade-
quate monitoring system. Equipment
capable of detecting trace amounts
of unburned hydrocarbons have been
shown to be inappropriate because
of the predominance of .unburned
hydrocarbons from the primary fuel.
The emission tests which have in-
cluded continuous monitoring for
SO2, NOx, O2, and CO have demon-
strated that combustion operation
can be (with training) easily opti-
mized. Addition of these monitors
is an excellent operating tool be-
cause of the increased feedback and
decreased response lag times com-
pared to air flow, temperator, and
undergrate pressure sensors, but
such continuous monitoring (and
subsequent reporting) is excessive
as a regulatory tool. They require
constant attention, calibration,
and maintenance. Unfortunately,
the combination of monitors is val-
uable and no single monitor serves
the purpose.
Further research should be
conducted to identify the simplest
monitoring techniques that can be
used to effectively ensure the
safety of waste firing. The iden-
tification of adequate monitoring
systems entails a thorough under-
standing of the mechanism by which
waste destruction occurs in kilns.
45
-------
REFERENCES
Worthy, W. "Hazardous waste:
treatment technology grows."
Chemical & Engineering News,
60 (10): 10-16, March 1982.
Hazelwood, D. L., F. J. Smith
and E. M. Gartner. "Assess-
ment of Waste Fuel Use in
Cement Kilns." EPA-600/S2-82-
013, U. S. Environmental Pro-
tection 'Agency, Cincinnati,
Ohio. October 1982.
Berry, E. E., L. P. MacDonald,
and D. J. Skinner. "Experi-
mental Burning of Waste Oil as
a Fuel in Cement Manufactur-
ing." Environment Canada, EPS
4-WP-75-1. June 1976.
MacDonald, L. P., D. J.
Skinner, F. J. Hopton, and
G. H. Thomas. "Burning Waste
Chlorinated Hydrocarbons in a
Cement Kiln." Fisheries and
Environment Canada, EPS 4-WP-
77-2 (also EPA/530/SW-147C).
March 1977.
Ahling, B. "Combustion Test
With Chlorinated Hydrocarbons
in a Cement Kiln at Stora Vika
Test Center." Swedish Water
and Air Pollution Research In-
stitute, Stockholm, Sweden.
March 1978.
Ahling, B. "Destruction of
Chlorinated Hydrocarbons in a
Cement Kiln." Environmental
Science & Technology, 13 (11):
1377-1379, November 1979.
Higgins, G. M., and A. J.
Helmstetter. "Evaluation of
hazardous waste incineration
in a dry process cement kiln."
Proceedings of the Eight Annual
Research Symposium: Incinera-
tion and Treatment of Hazardous
Waste. EPA-600/9-83-003. April
1983. pp. 243-252.
Peters, J. A., T. W. Hughes,
J. R. McKendree, L. A. Cox,
and B. M. Hughes. "Industrial
Kilns Processing . Hazardous
Wastes-San Juan Cement Company
Demonstration Program." EPA-
600/S2-84-, U. S. Environmental
Protection Agency, Cincinnati,
Ohio.
9. Day, D. R., and L. A. Cox.
"Evaluation of Hazardous Waste
Incinceration in Lime Kilns at
Rockwell Lime Company." Mon-
santo Research Corporation
draft report on EPA Contract
68-03-3025, Work Directive
SDM-05. October 1983.
10. Jenkins, H. C., G. Murchison,
R. C. Adrian, R. D. Fletcher,
and D. C. Simeroth. "Emissions
from Los Robles Kiln." State
of California Air Resources
Board, Engineering Evaluation
Report C-82-080. August 1983.
11. Duvel, W. A. "Practical inter-
pretation of groundwater moni-
toring results." In: Proceed-
ings of the National Conference
on Management of Uncontrolled
Hazardous Waste Sites. Hazard-
ous Material Control Research
Institute, Silver Spring, Mary-
land, 1982. pp. 86-90.
12. Weitzman, L. "Cement kilns as
hazardous waste incinerators."
Environmental Progress, 2 (1):
10-14, February 1983.
13. Haynes, B. W. and G. W. Kramer.
"Characterization .'of U.S.
Cement Kiln Dust." Bureau of
Mines 1C 8885, Avondale, Mary-
land. 1982.
14. Mix, T. W. and B. L. Murphy.
"Risks associated with waste-
fuel use in cement kilns."
Environmental Progress, 3 (1):
64-70, February 1974.
15. Compilation of Air Pollutant
Emission Factors, "Third Edi-
tion. U. S. Environmental Pro-
tection Agency, Research Tri-
angle Park, N.C. Publication
No. AP-42, July 1979.
16. Peters, J. A., T. W. Hughes,
and R. E. Mournighan "Evalua-
tion of hazardous waste incin-
46
-------
eration in a cement kiln . at
San Juan Cement." Presented
at the Ninth Annual Research
Symposium on Land Disposal,
Incineration and Treatment of
Hazardous Waste, May 2-4, 1983,
Ft. Mitchell, Kentucky.
17. Monroe, E. S. "Quicker, cheaper
testing of incinerator perfor-
mance." Chemical Engineering,
90 (4): 69-71, February 1983.
18. Tidona, R. J. , W. A. Carter,
H. J. Buening, S. S. Cherry,
and M. N. Mansour. "Evaluation
of Combustion Variable Effects
on NOx Emissions from Mineral
Kilns." EPA-600/S7-83-045.
U.S. Environmental Protection
Agency, Research Triangle Park,
N.C. November 1983.
47
-------
EVALUATION OF HAZARDOUS WASTE
INCINERATION IN A LIME KILN
Duane R. Day
Monsanto Company
Dayton, Ohio
and
Robert E. Mournighan
U.S. Environmental Protection Agency
C inc innati, Ohio
ABSTRACT
During a one-week test burn, hazardous waste was used as supple-
mental fuel and co-fired with petroleum coke in a lime kiln in eastern
Wisconsin. Detailed sampling and analysis was conducted on the stack
gas for principal organic hazardous constituents (POHCs), particulates,
particulate metals, HC1, SO2, NOX, CO, and THC and on process
streams for metals and chlorine. POHCs were also analyzed in the waste
fuel. Sampling was conducted during three baseline and five waste fuel
test burn days. The program objectives were to determine the destruc-
tion and removal efficiency (DRE) for each POHC, determine concentration
of stack gas pollutants under baseline and waste fuel test burn condi-
tions, determine the fate of chlorine, sulfur, and trace metals in the
kiln process, and evaluate kiln performance when operating with
hazardous waste as supplemental fuel.
Results show average DRE's greater than 99.99% for each POHC and
little change in pollutant emissions from baseline to waste fuel test
conditions. In addition, material balance results show that 95% of
chlorine enters the process from the limestone feed and the chlorine
exits the kiln in the baghouse dust and lime product at 61% and 38%,
respectively.
INTRODUCTION
Cofiring of hazardous wastes
in high temperature industrial
processes is an attractive alterna-
tive to hazardous waste incinera-
tion. The alternative makes use of
the waste' s heat content and pro-
vides temperatures and residence
times similar to those required for
incinerators dedicated to incinera-
tion of hazardous wastes. In addi-
tion to the savings derived from
the heat value, the use of existing
industrial equipment does not re-
quire the capital required if a
separate incinerator to process a
given amount of hazardous waste is
to be built, and it may provide an
environmentally acceptable alterna-
tive to conventional hazardous
waste disposal.
Lime kilns, because of their
high enengy use, are an excellent
example of this concept. Such
48
-------
kilns typically operate' at tempera-
tures over 1093°C (2000°F), have
gas residence times in excess of
1.5 seconds, and have a highly
turbulent combustion zone. How-
ever, the need exists for data
that shows the effect of cofiring
hazardous waste on the emissions
from the lime process.
The State of Wisconsin Depart-
ment of Natural Resources (DNR) and
EPA Region V, in implementing RCRA
regulations, issued a temporary
permit to Rockwell Lime Company' to
conduct a hazardous waste test
burn. This test would allow the
burning of hazardous liquid waste
as supplemental fuel along with
petroleum coke. The waste fuel
would replace natural gas as a fuel
component.
Monsanto Company (MC) was con-
tracted by the U.S. Environmental
Protection Agency, Office of
Research and Development (ORD-IERL)
in Cincinnati, Ohio to perform
sampling and analysis of stack
gases and. process samples during
the test burn conducted at the
Rockwell Lime Company in Rockwood,
Wisconsin on April 15 and April 29
through May 6, 1983 under baseline
(no waste fuel burned) and waste
fuel conditions (coke and waste
fuel burned).
The primary objectives of the
sampling and analytical program
were to (1) determine the effects
of cofiring petroleum coke and
hazardous waste on the emissions
from the kiln, (2) determine the
fate of the principal organic
hazardous constituents (POHCs) and
determine destruction and removal
effeciencies (DRE), (3) determine
the fate of chlorine and trace
metals in the kiln process,
(4) determine the concentration
of SO2, NO , particulates, HCl,
X
metals, total hydrocarbons, and
carbon monoxide in the stack gas at
baseline and waste fuel test burn
conditions, and (5) evaluate kiln
operation during hazardous waste
fuel burning condition. This test-
ing would provide to the Wisconsin
DNR and the EPA Region V the data
necessary to determine whether a
permit can be issued to Rockewll
Lime Company to burn hazardous
waste. The testing also will pro-
vide the EPA-ORD with additional
data in their research on the in-
cineration of hazardous waste and
the environmental problems associ-
ated with incineration.
FACILITY AND 'PROCESS DESCRIPTION
The Rockwell Lime Company
operates a lime kiln in Rockwood,
Wisconsin, which is located
approximately 10 miles north of
Manitowac in eastern Wisconsin.
Production of lime, which is con-
verted into various forms such as
hydrated and barn lime, is approxi-
mately 1.3 x 106 kg (1,430 tons)
per week, which varies based upon
product demand.
The industrial process in-
volves the heating of limestone to
a 'temperature of approximatly
1,093°C (2,000°F) in a horizontal
rotary kiln to prepare the lime
product. The calcining is achieved
by interfacing the hot gases with
the limestone which drives off the
CO2 from the limestone, leaving the
lime product (CaO).
The kiln, with refractory lin-
ings, is 2.4 m (8 ft) in diameter
and 67.1 m (220 ft) long. The kiln
rotates slowly (approximately one
revolution per minute) and has a
gentle slope (6 cm/m of length) to
allow material to pass through by
gravity. The, kiln also- has a
countercurrent flow pattern, i.e.,
solid materials travel in one
direction and hot gases and dust
emissions travel in the opposite
direction as shown in Figure 1.
The coke and natural gas provide a
heat input of approximately 14,700
kw (50 million Btu/hr) or approxi-
mately 6.5 million Btu/ton of lime
product. As the limestone feed
travels down the inclined rotating
kiln, it passes through various
temperature zones, and the hot
gases calcine the limestone into
the lime product. The product is
produced at a rate of approximate-
ly 7,720 kg/hr (17,000 Ib/hr).
49
-------
After heating and transformation
in the kiln, the lime product is
cooled by air and either stored in
silos or hydrated prior to storage
in silos where . it is held until
sold.
3
e
co
en
-------
TABLE 1. SUMMARY OF ROCKWELL LIME KILN
SAMPLING AND ANALYTICAL PROGRAM
Parameter
Sampling method
Analytical method
STACK GAS
1. POHCs*
2. Particulate matter
Metals on
particulate
3. Hydrogen chloride
4. Carbon dioxide
and oxygen
5. Nitrogen oxides
6. Sulfur dioxide
7. Carbon monoxide
8. Total hydrocarbons
9. Oxygen
10. Waste fuel
POHCs
Metals
Chlorine, sulfur
Btu content
11. Baghouse dust
Metals
Chlorine, sulfur
12. Lime product
Metals
Chlorine, sulfur
13. Dry limestone feed
Metals
Chlorine
14. Primary fuel coke
Metals
Chlorine, sulfur
Btu content
Volatile organic sampling GC/MS, thermal desorption
train (VOST) and SIM
EPA Method 5
EPA Method 5
EPA Method 5
ICP
Impinger absorption in Specific ion electrode
0.5 M NaoAc (back half
of EPA Method 5)
EPA Method 3
Continuous
Continuous
Continuous
Continuous
Continuous
Grab ••»• composite
Grab -»• composite
Grab •* composite
Grab •* composite
Grab.-»• composite
Grab -» composite
Grab
Grab
composite
composite
Grab -» composite
Grab •*• composite
Grab -> composite
Grab -» composite
Grab •* composite
Fyrite
Chemiluminescence photo-
metric analyzer
Pulsed fluorescence
TECO analyzer
Infrared-EPA Method 10
Flame ionization detector
Teledyne's micro-fuel cell
GC/MS
ICP
ASTM D240-64
ASTM D482-IP4
ICP
XRF
ICP
XRF
ICP
XRF
ICP
XRF
ASTM D240-64
Tetrachloroethylene, trichloroethylene, methylene chloride,
1,1,1-trichloroethane, methyl ethyl ketone, and toluene.
51
-------
RESULTS AND DISCUSSION
Waste Fuel
A_ summary of the waste fuel
composition for two samples is
shown in Table 2. Tables 3 and 4
show the concentration of each POHC
and other properties averaged for
the five waste fuel samples (one
sample per day, Runs 4-8).
POHC Destruction and
Removal Effeciencies
The complex combustion chemis-
try for organic materials becomes
perplexing when a broad range of
organic compounds present in a
liquid waste are burned. On a
weight basis, most of the organic
carbon in the waste is oxidized to
CO2 in the combustion process, but
trace amounts of organic chemicals
survive the oxidation process and
are only partially reacted.
Accordingly, the test burn investi-
gated the amount of destruction of
the organic compounds in the haz-
ardous waste.
Destruction and removal effi-
ciency t (DRE) for an incineration/
air pollution control system is de-
fined by the following equation:
DRE = Win " Wout
(100)
(1)
where DRE =
Win =
W
out
destruction and
removal efficiency, %
mass feed rate of
principal organic haz-
ardous constituent(s)
(POHCs) fed to the
incinerator
mass emission rate of
principal organic haz-
ardous constituent^ s)
(POHCs) to the atmos-
phere (as measured in
stack prior to
discharge).
DRE calculations are based on
combined efficiencies of the de-
struction of the POHC in the incin-
erator, or the lime kiln, and the
removal of the POHC from the gas
stream in the air pollution control
system. The presence of 'POHCs in
solid discharges from the air pol-
lution control devices is not
accounted for in the DRE calcula-
tion as currently defined by the
EPA. RCRA Part 264, Subpart 0 reg-
ulations for hazardous waste incin-
erators require a DRE of 99.99% for
all principal organic hazardous
constituents of a waste during
trial burns unless it can be demon-
strated that a higher or lower DRE
is more appropriate based on human
health criteria. Specification of
the POHCs in a waste is subject to
best engineering judgment, consid-
ering the toxicity, thermal stabil-
ity, and quantity of each organic
waste constituent.
Approximately six VOST samp-
ling runs were made each day (Runs
1-8). Destruction and removal
efficiencies, calculated for waste
fuel runs 4-8, are summarized in
Figure 2. In general, DREs ranged
from 99.60% to >99.999% for all
compounds and averaged 99.9989%.
Only four runs had DREs less than
99.99%; three of these were for
methylene chloride, the fourth for
1,1,1,-trichloroethane.
DREs for methylene chloride
ranged from 99.60% to <99.999% and
averaged 99.983%. Three runs at a
DRE less than 99.99% (Run Nos. 5A,
5F, and 4E). In Run No. 5A, the
tube mass collected was 1,066 ng
compared to a maximum of 55 ng for
all of the other runs. This high
mass was most likely caused by lab
contamination. Run No. 5A is not
included in the averages, due to
the lab contamination, thus the
Day 5 average is 99.995% and the
overall methylene chloride DRE
average is 99.997%.
Methyl ethyl ketone had an
average DRE of 99.999% and .-ranged
from 99.998% to greater than
99.999%. These high destruction
efficiencies were consistent
throughout the waste fuel test
runs.
DREs for 1,1,1-trichloroethane
52
-------
TABLE 2. RESULTS OF CAPILLARY GC/MS ANALYSIS
OF MAJOR COMPONENTS OF WASTE FUELS
Concentration,
wt %
Waste fuel component
Acetone
Methyl ethyl ketone (POHC)
1,1, 1-Trichloroethane ( POHC )
1-Butanol
Trichloroethylene (POHC)
2 -Ethoxyethanol
Methyl isobutyl ketone
Toluene (POHC)
Tetrachloroethylene (POHC)
Butyl acetone
Ethylbenzene
Xylene (isomer No. 1)
Xylene (isomer No. 2)
2-Butoxyethyanol
2-Ethoxyethyl acetate
C3 -Benzene (isomer No. 1)
C3 -Benzene (isomer No. 2)
C10 -Alkane
Alkane >C8
Alkane
C1:l- Alkane
2 - Cyc 1 ohexen- 1 - one
Alkane >C6
Number
4s-
0.23
2.48
0.24
0.32
1.73
0.85
1.06
11.0
2.17
0.27
1.42
4.92
1.43
1.99
5.91
0.28
0.46
0.80
0.24
0.14
1.26
0.15
0.27
Number
0.22
3.17
0.22
0.37
2.16
0.92
1.16
12.5
2.49
0.32
1.58
5.58
1.60
2.07
6.37
0.32
0.57
0.94
0.28
0.18
1.48
0.18
0.24
Average of split sample.
TABLE 3. CONCENTRATION OF POHCs
TABLE 4. WASTE FUEL CONDITIONS
Average
concentrati on,
POHC
MeCl2
MEK
CH3CC13
TCE
Perc
Toluene
wt %
0.11
2.69
0.24
1.84
2.26
11.6
Average
Parameter concentration
Chlorine, % vol 3.1
Sulfur, % 0.07
PCB, ppm <1.0
Heat value, Btu/lb 12,870
Specific gravity,
g/cc 1.01
Feed rate, gpm Run 4 0.64
5 1.21
6 2.05
7A 0.78
7B 2.90
8 2.88
53
-------
30-
20H
o 10-
Methylene Chloride
30-
20-
o 10-
12345
Number of Nines ORE
Methylethyl ketone
1 23 4 5
Number of Nines ORE
30-
2CH
0 10H
•
o
Trlchloroethylene
30-
20-
0 ioH
d
1234
Nunber of Nines ORE
Tetrachloroethylene
• • i i
12345
Number of Nines ORE
30-
20-]
« 1
o 10-
1, 1, 1 Trlchloroethane
1 i ^ 4 !;
Number of Nines ORE
30-
20-
. 10-
Toluene
1234
Number of Nines ORE
Fiqure 2. Destruction and removal efficiencies.
54
-------
ranged from 99.989% to 99.999% and
averaged 99.997%. Only Run No. 4E
had a DRE less than 99.99%.
DREs for trichloroethylene
(TCE) were greater than 99.999% for
all runs. TCE was spiked to the
waste fuel prior to the waste fuel
testing to increase its concentra-
tion and allow easier detection of
TCE in the stack gas.
DREs for tetrachloroethylene
(Perc) also were greater than
99.999% for all runs. Like TCE,
Perc was spiked to the waste fuel
to the maximum allowable concentra-
tion ' described in the test burn
permit prior to the test.
Toluene was the POHC of high-
est concentration in the waste fuel
(average 11.6% by weight). DREs
for toluene were above 99.999% for
all runs. Data for toluene was
very consistent during all waste
fuelvtest runs.
Stack Samples
Results for stack conditions
and particulate, HC1, SO2, NO , CO,
and THC emissions for baseline and
waste fuel runs are summarized in
Table 5. The dry stack rate aver-
aged 487 dscm/min (17,210 dscf/min).
As evidenced by the high standard
deviations the CO and, to a lesser
degree, the SO2 fluctuated. Minor
kiln upsets (i.e., coke feed chute
cleaning, clumps of coke falling to
kiln, change in process conditions)
created high CO excurtions. A
trend which occurred quite often
was an increase in SO2 by ~200 ppm,
followed by a reduction in NO by
~50 ppm, following by an increase
in CO by ~500 ppm over a 15 minute
period. These trends are expected
when a lower intensity flame occurs
(or kiln upset). However, as re-
vealed by Figure 2, the kiln upsets
had little or no effect on the DRE
results.
Chlorine, Sulfur and Metals Balance
Chlorine and sulfur macerial
balances are summarized for base-
line and waste fuel conditions in
Table 6. The majority of chlorine
(for either baseline or waste fuel
conditions) enters the kiln in the
limestone feed and exits the kiln
in the lime product and baghouse
dust. Sulfur (for either baseline
or waste fuel conditions) enters
the kiln in the petroleum coke and
TABLE 5. AVERAGE STACK EMISSIONS
Parameter and unit
Stack rate, ms/miri
Stack velocity, m/sec
Range
805
5.0
- 975
- 6.0
Baseline
Average
917
5.7
Waste fuel
Standard
deviation
76
0.4
Range
791
4.9
- 938
- 5.8
Average
847
5.2
Standard
deviation
52
0.3
Particulates
mg/dscm
kg/hr
HCl, ppm
S02, ppm
NO , ppm
A
CO, ppm
THC, ppm
24.0 - 35.0
0.66 - 1.1
0.74 - 3.9
123 - 730
306 - 460
10 - 4,900
6.7 - 12.7
28.7
0.9
2.0
553
386
477
8.2
4.7
0.2
1.4
110
49
966
1.9
24.9 - 48.7
0.68 - 1.4
2.5 - 6.0
183 - 1,924
288 - 552
10 - 4,540
1.5 - 10.0
35.3
1.0
4.4
596
446
599
3.5
8.0
0.3
1.2
240
64
1,409
1.1
55
-------
exits the kiln distributed in the
lime product (~9%), baghouse dust
(~27%), and stack gas (~64%).
An elemental material balance
was performed for 21 elements.
Balance results show no difference
for baseline and waste fuel condi-
tions for distribution of metals in
the kiln process. Also the major-
ity of mass entering the kiln is
contributed by the limestone feed
(typically 80-100%) except for
zinc which has 42% in the lime-
stone and 53% in the waste fuel.
The mass exiting the kiln is dis-
tributed between the lime product
and baghouse dust (approximately
50-90% in product and 10-45% in
dust). Magnesium and calcium had
mass rates (to or from kiln) of
approximately 2,500 kg/hr. Sodium
and iron mass rate were approxi-
mately 20 kg/hr. All other ele-
ments had mass rates less than
5 kg/hr.
Baseline vs. Waste Fuel
and Kiln Operation
Emissions were evaluated
under baseline and waste fuel con-
ditions. For the pollutants
listed in Table 5, HC1, NO and
TEC showed a significant differ-
ence in stack emissions under
baseline and waste fuel condit-
ions . For the POHC's, only methy-
lene chloride and toluene showed a
minor difference (increase) from
baseline to waste fuel conditions.
All remaining POHC's showed no
significant difference in base-
line vs. waste fuel emissions.
As described previously, the
kiln operation fluctuated as indi-
cated by CO and SO2 emission vari-
ations during waste fuel burning.
Kiln fluctuations were caused by
several items including nonconstant
fuel rates, product rushes, clumps
of coke fed to kiln accidentally,
and operator inexperience with
burning waste fuel. The fluctua-
tion resulted in occasional kiln
O2 increases and stack gas SO2 de-
crease causing a poorer quality
lime product most likely due to
excess sulfur in the product. The
following items were identified as
ways to improve operation at this
kiln under waste fuel conditions:
1. Change waste fuel burner con-
figuration such that at low
waste fuel rates the waste
fuel is mixed with the coke
to maintain a flame.
2. Decrease the fan speed (i.e.,
reduce the draft) to lower the
O2 in the kiln resulting in
lower sulfur in product and
higher sulfur in the stack.
TABLE 6. CHLORINE AND SULFUR MATERIAL BALANCE
Percent to kiln
Total
Percent from kiln
Total
Material
balance
Run
number
Chlorine
Baseline3
Waste fuelb
Sulfur
Baseline
Waste fuelb
Coke
4
5
100
99
Waste
fuel
0
2
0
1
Limestone
feed
96
93
NA
NA
mass in,
kq/hr
23
21
59
54
Lime
product
47
34
8
10
Baghouse
dust
52
65
29
25
Stack
gas
1
1
63
65
mass out, closure,
kg/hr %
13 55
20 105
91 73
70 77
Average values of baseline Runs 1-3.
Average values of waste fuel Runs 4-8.
56
-------
FIELD TESTS OF INDUSTRIAL BOILERS COFIRING HAZARDOUS WASTES
Carlo Castaldini, Howard B. Mason, Robert J. DeRosier, Stefan Unnasch
Acurex Corporation
Mountain View, California 94039
ABSTRACT
Thirteen field emission tests have been performed on eight industrial boilers
cofired with conventional fuels and hazardous wastes. The U.S. Environmental Protection
Agency (EPA) sponsored the tests to evaluate the effectiveness of industrial boilers for
hazardous waste thermal destruction while recovering useful energy. Five initial tests
were made on a wide range of waste type/boiler design combinations to quantify emissions
representative of current industry practices. For the second series of tests, boilers
firing waste streams spiked with a mixture of carbon tetrachloride, chlorobenzene, and
trichlorethylene were tested to extend the data base for chlorinated wastes and
facilitate intrasource comparisons. Principal organic hazardous constituents (POHC's) in
the flue gas were quantified using two extractive sampling trains. Volatile compounds
were absorbed on two Tenax traps in the volatile organic sampling train (VOST).
Semi volatile compounds were absorbed on XAD-2 sorbent in a modified Method 5 train. The
sampling protocol also included continuous monitor measurement of stack 0%, CO, C02, NOX
and total hydrocarbons; extractive train samples of HC1; and grab samples of the
conventional and waste fuels and ash streams. Post-test analyses of the inlet fuels and
the outlet stream samples were done with gas chromatography/mass spectrometry to quantify
the destruction and removal efficiency of POHC's. Results for the thirteen tests are
discussed in this paper. Specie destruction and removal efficiency (DRE) ranged from
99.5 to >99.999 percent. The mass weighted destruction efficiency for all RCRA
Appendix VIII compounds was -99.998 percent. Products of incomplete combustion (PIC's)
were observed in concentrations one order of magnitude greater than measured breakthrough
POHC's.
INTRODUCTION
Many hazardous wastes have
sufficiently high heat of combustion to be
candidates for cofiring with conventional
fuels in industrial boilers. Cofiring can
be economically and environmentally
advantageous to other disposal/destruction
options while meeting the Resource
Conservation and Recovery Act (RCRA) goal
of resource recovery. The benefits of
cofiring include partial displacement of
conventional fuels, and use in existing
industrial boilers where incinerators are
not available. Although boiler cofiring
is not currently regulated, the need for
inclusion in RCRA provisions is currently
being studied by EPA. EPA sponsored the
tests discussed in this paper to quantify
the waste destruction potential of
industrial boilers. To obtain results
representative of current or potential
industrial practice, a broad range of
boiler designs and waste types was
selected for testing (1). Eight sites
were tested encompassing thirteen tests of
which the initial three were reported at
the Ninth Symposium(2).
APPROACH
Candidate sites were screened based
on representativeness of the boiler design
and wastes being fired, and the
availability and accessibility'for
cofiring tests. Within this context,
preference was given to sites which were
regarded as more challenging in attaining
57
-------
a high level of waste destruction.
Table 1 summarizes the eight test sites
and thirteen test series. Sites A and H
were solid-fuel-fired and were equipped
with a cyclone and electrostatic
predpitator, respectively. Site G was
fired with chlorinated hydrocarbons
without auxiliary conventional fuel and
was equipped with two scrubber columns for
HC1 recovery and cleanup. The selected
sites spanned a broad range of design and
operating conditions: firetube and
watertube designs; capacities from
1.1 kg/s (8,500 Ib/hr) to 32 kg/s
(250,000 Ib/hr); gas, oil, coal arid wood
firing; loads from 25 to 100 percent of
rated capacity; wastes ranging over an
order of magnitude in heat of combustion;
and residence time and heat release
variations over an order of magnitude. To
extend the range of waste destruction
characteristics tested, the wastes at
site E-H were spiked with carbon
tetrachloride and, in most cases,
monochlorobenzene and trichloroethylene.
The nominal test series involved an
initial conventional fuel baseline test to
characterize unit operation and emissions
in the absence of waste firing, and
VV.isle
Day
T.mK
triplicate cofiring runs. For these runs,
the unit load was held constant to enable
replicate comparisons of results. In most
other respects, routine operational
variations such as excess air levels and
waste flowrates were tolerated to obtain
results representative of normal
operation.
The major inlet and outlet streams
were sampled and analyzed as shown in
Figure 1, for the case of a coal-fired
unit, and boiler operational data was
taken to characterize performance with and
without waste firing. Retails on the
protocol are given in (1). Waste and fuel
grab samples were taken approximately
every hour, composited, and analyzed in
the laboratory for ultimate and proximate
analyses, chloride content, and POHC
concentration. Bottom and hopper ash
composite samples were analyzed for
chlorides, POHC's, and carbon content.
The major sampling was conducted at the
stack where the following samples were
taken:
• Continuous monitor analyses of
02, CO, C02, NOX, and UHC
A — Liquid Waste Grab Samples (Composite) D,F — Stack Emissions
B — Fuel Grab Samples (Composite) E — Particulate Collector
C — Boiler Bottom Ash Grab Samples (Composite) Hopper Ash
Figure 1. Sampling schematic.
58
-------
TABLE 1. BOILER DESIGN AND OPERATING CHARACTERISTICS
Operation
Boiler^ Type
and Capacity
Site kg/s (103 Ib/hr)
A WT Stoker
1.26
(10)
B FT 1 Burner
1.1
(8.5)
C WT 6 Burner
29
(230)
D WT 4 Burner
ii
(90)
E WT 1 Burner
14
(110)
F WT 2 Burner
7.6
'(60)
G FT 1 Burner
4.0
(32)
H WT T-Fired
32
(250)
Waste S
Primary (percent heat input)
Fuel POHC's* (
Wood (40)
waste creosote
Gas Al kyd
wastewater
Gas (38)
phenol /cumene
Oil (26)
xylene, PCE
(48)
toluene, BCEE
Oil (28)
MMA, CC14, Cl, TCE
(42)
MMA, CC14, Cl<|,, TCE
(19)
' MMA, CC14, Cl, TCE
(52)
MMA, toluene
Oil (10)
solvent, CC14,
C1,j>, TCE
NA (100)
CC14, chlorinated
hydrocarbons
Coal (3)
MeA, CC14, Cl$,
1,1,1-TCA
:team Load Residence
kg/s Percent time
103 Ib/hr) 02 (s) (:
1.26 6-16 1.2
(10)
0.26 4-6 0.8
(2)
7.4 10 2
(59)
8.8 4-6 1.1
(70)
5.7 6 1.3
(45)
6.9 7 0.7
(55)
5.0 6 • 1.0
(40)
10 7 0.5
(80)
7.2 6 0.8
(57)
5.6 7 1.1
(44)
4.0 7-11 2
(32)
1.8 8 0.4
(14)
32 6 2
(250)
Heat
release
kW/m3
LO3 Rtu/ft3h)
300
(29)
750
(72)
78
(7.5)
400
(39)
240
(23)
520
(50)
350
(34)
770
(74)
470
(45)
370
(36)
110
(H)
820
(79)
180
(17)
*PCE = tetrachloroethylene; BCEE = bis(2-chloroethyl)ether; MMA = nethylmethacrylate;
Cl<(, = monochlorobenzene; TCE = trichloroethylene; MeA = methyl acetate;
1,1,1-TCA = 1,1,1-trichloroethane
1"WT = watertube; FT = firetube
59
-------
• Volatile organics extractive
samples by the Volatile Organic
Sampling Train (VOST) (3)
• Semi volatile organics and
particulates by the modified
Method 5 extractive sampling
train (4)
t Chlorides by a Method 6
extractive sampling train
• GJ-CS hydrocarbons by a gas bomb
grab sample and gas chromatograph
analyses
Each test required approximately 6 hours
of run time. Post-test analyses of the
volatile and semi volatile samples
collected on resin traps were done by gas
chromatography/mass spectroscopy (GC/MS).
Waste destruction and removal efficiencies
were computed based on feed and stack mass
flowrates quantified in the samples:
ORE = (Mfeed - Mstack)/Mfeed x 100
Constituents appearing in the stack which
were not identified in the feed or other
inlet stream were identified as products
of incomplete combustion (PIC's).
RESULTS
Table 2 summarizes POHC-specific ORE
results. Calculated DRE's are based on
blank corrected emission rates measured
during cofiring. Therefore,, credit is not
given for emission of POHC's measured
during baseline tests. Results indicate a
wide range in DRE's from 99.5 to
>99.999 percent. The mass weighted
average ORE for all POHC's was
99.998 percent. Although the average ORE
for each POHC was generally greater than
99.99 percent, some emission measurements
resulted in DRE's below this level, the
current RCRA incinerator standard. These
low DRE's often coincided with unsteady
boiler operation and burner combustion
instability. For example, low DRE's for
carbon tetrachloride, chlorobenzene, and
trichloroethylene are generally
attributable to site F. Improper burner
settings at this site resulted in coking
of the burner nozzle, fuel impingement on
burner throat, and occasionally high
levels of combustible CO and soot
emissions.
Low DRE's for site E methylmetha-
crylate were measured during few tests
characterized by instability in waste
TABLE 2. SUMMARY OF POHC ORE RESULTS
POHC
Carbon tetrachloride
Trichloroethylene
1,1,1 Trichloroethane
Chlorobenzene
Phenol
Pentachlorophenol
Tetrachloroethylene
B1s(2-chloroethyl )ether
Toluene
Fluorene
Naphthalene
Methyl nethacry late
2-4 Dlraethylphenol
Epichlorohydrin
B1s(2-chloro1sopropyl Jether
Inlet
(ppra) |Hwl x 10-
IR?J
1.0-50
0.78-12
0.82-0.96
0.13-7.5
0.23-21
0.88-2.4
8.0-90
18-20
0.46-450
1.8-3.0
2.2-7.6
7.9-62
0.12-0.52
160-210
400-540
Emission
3 rate
Mg/s
1-460
28-470
<0.3-17
3.6-190
<0. 6-190
<3.2-44
510-1,400
<3-6.8
29-8,800
<0.6-34
25-150
17-8,300
<3.2-13
<2
<2
Mass
average
ORE
percent
99.998
99.998
99.994
99.998
99.999
99.998
99.998
>99.999
99.998
99.996
99.983
99.991
>99.983
>99.999
>99.999
ORE
range
percent
99.97-99.999
99.98-99.999
99.98-99.999
99.97-99.999
99.5-99.999
99.985-99.998
99.994-99.999
>99.999
99.90-99.999
99.986->99.999
99.946-99.988
99.95-99.999
99.96->99.99
>99.999
>99.999
Site
E, F, G,
E, F
H
E, F, H
A, C
A
D
n
B, D, E,
A
A
E
A
G
G
H
F
0.12-540
<0.6-8,800 99.998 99.5->99.999
60
-------
feedrates and insufficient combustion air.
These operating conditions led to several
high CO and smoke emission episodes.
Wood-fired stokers, such as the site A
boiler, are typically high excess air and
high CO emitters. These conditions result
from the physical properties of wood waste
(e.g., wood chip size and high moisture
content), combustion cooling by very high
excess air levels and inefficient
fuel-air/mixing of combustion on a fuel
bed. Site A DRE's were generally below
99.99 percent. Table 3 illustrates this
site dependence more clearly. An attempt
to correlate DRE's with CO emissions was
not successful, however. It is likely
that this was in part due to insufficient
emission measurements during periods of
high CO or smoke emissions.
Figure 2 illustrates an apparent ORE
dependence on POHC concentration in the
waste fuels. The data are based on mass
average DRE's for all POHC's at each site.
Therefore, included are ORE results for
other organics not listed in Table 3. The
data indicate that higher DRE's were
obtained with higher POHC firing rates.
This result suggests that the amount of
POHC breakthrough is independent of the
POHC concentration in the waste fuel. One
factor that contributes to this trend is
the high DRE's for semi volatile organics
(phenol, BCEE, epichlorohydrin, and
bis(2-chloroisopropyl)ether) that were
present at high concentrations in waste
fuels of sites C, D, and G. Results for
other nonhazardous organics present at the
high concentrations in site E waste fuel
also showed high DRE's. A second factor,
is the background contamination from
sampling and analysis techniques which
tends to be more significant at low POHC
emissions. Also important is the
contribution of volatile organics from
fossil fuel combustion.
Quantisation of PIC emissions was
based on the detection of organic priority
pollutants not found in the waste fuel.
Figure 3 illustrates the relationship
between the sum of total PIC's,
halogenated and nonhalogenated organics,
versus the test average ORE for site D
through H. The data suggest that the
lower the quantity of POHC breakthrough
(i.e., higher ORE) the lower the total
quantity of PIC's. Therefore, higher POHC
DRE's are also likely to result in lower
PIC emissions. On the average, PIC
emissions were one to two orders of
magnitude greater than the total quantity
TABLE 3. SUMMARY OF SITE ORE RESULTS
Site
A
B
C
D
E
F
G
H
<
ORE range
99.5-99.999
99.991
99.998->99.999
99.994-99.9996
99.95-99.9995
99.90-99.999
99.995->99.999
99.97-99.999
99.5->99.999
CC14
99.999
99.98
99.998
99.988
99.998
POHC DRE
percent
TCE* Cl<|) PCP1"
99.988
99.998 99.998
99.996 99.98
99.992
99.998 99.998 99.988
PCEt1" Toluene
99.98
99.998 99.9994
99.997
99.95
—
99.998 99.998
Average
DRE
99.98
99.98
99.9990
99.9990
99.995
99.98
99.9996
99.991
99.998
*TCE = trichloroethylene
i"PCP = pentachlorophenol
t1"PCE = tetrachloroethylene
61
-------
FIELD TESTS OF INDUSTRIAL BOILERS AND INDUSTRIAL PROCESSES
DISPOSING OF HAZARDOUS WASTES
Radford C. Adams, Michael W. Hartman and Denny E. Wagoner
Radian Corporation
Research Triangle Park, North Carolina 27709
ABSTRACT
Two industrial boilers were tested to determine their hazardous waste destruction
performance. The tests are part of a research program conducted by the EPA to evaluate a
spectrum of representative boiler types that are currently being used for hazardous waste
disposal. These boiler tests provide destruction efficiency (DE) results as the primary
performance parameter. Waste materials typically burned in the boilers were spiked with
a standard test mixture containing components expected to vary in incinerability. The
standard test mixture provides the EPA with a means of comparing the thermal destruction
performance among several boilers.
The boilers were operated at a single steady state load of 80% of design capacity
with operator settings of excess air and feed rate controlled by normal operating pro-
cedures. A series of three background runs and three waste feed runs were conducted at
the prescribed boiler load. Additional sampling runs were required during Boiler No. 1
testing to evaluate NOX control with and without staged combustion.
The destruction efficiencies of ten POHCs (principal organic hazardous constituents)
were determined at low and high waste to fuel ratios. No correlation was found of
destruction efficiency with potenital incinerability indicators (carbon monoxide and
total hydrocarbons). Only two products of incomplete combustion (PICs) were identified
during the two bciler tests. Numerous organic species not detected in the feed were
detected in the emissions but these are believed degradation products of the sample
collection media. Staged combustion NOX control was used when burning nitrogenous waste
and was found to reduce NOX emissions. Heavy metals concentrations in the emissions when
burning wastes were compared with concentrations when burning the baseline fuel.
INTRODUCTION
Two industrial boilers were tested to
determine their hazardous waste destruction
performance. The tests are part of a
research program conducted by the EPA to
evaluate a spectrum of representative
boiler types that are currently being used
for hazardous waste disposal.
These boiler tests provide destruction
efficiency (DE) results as the primary
performance parameter. Waste materials
typically burned in the boilers were spiked
with a standard test mixture containing
components expected to vary in inciner-
ability. The standard test mixture pro-
vides the EPA with a means of comparing the
thermal destruction performance among
several boilers.
62
-------
The two boilers epitomize considerable
diversity in waste disposal goals and
waste feed characteristics. Wastes are
burned in Boiler No. 1 primarily to dispose
of a hazardous waste and not for apprecia-
ble fuel savings. Waste to fuel ratios
determined on a heat input basis were
0.09 to 1, meaning that only 8% of the
energy was supplied by the waste. There
is no clearly identifiable waste to fuel
ratio to report for Boiler No. 2 because a
substantial amount of the waste burned is
off specification fuel. During the test,
No. 6 fuel oil contributed less than 50%
of the input energy and it is clearly the
intent to obtain substantial fuel savings
by burning either wastes or off specifica-
tion fuels. Table 1 summarizes significant
differences in the design and waste disposal
characteristics of the two boilers.
Boiler No. 1 represents a class of boilers
disposing of a specific waste that is well
characterized with respect to knowledge of
the composition and of the expected volume
of waste. Boiler No. 1 also represents
cofiring of a liquid waste with a gaseous
fuel. Boiler No. 2 represents a class of
boilers that are not limited to a specific
waste stream. Variation .in fuel and waste
composition is the expected norm for
Boiler No. 2. Boiler No. 2 cofires liquid
wastes and off specification fuels with
No. 6 fuel oil. Specific components of
the liquid wastes are not always known.
PURPOSE
The two boilers were tested by similar
procedures. The principal goals were:
the identification of principal
organic hazardous constituents
(POHCs) in the feed and the
extent of their destruction,
the correlation of destruction
efficiency with boiler operating
settings with the boiler
operating'at steady state,
the fate of other feed components
such as chlorine, heavy metals,
and fuel bound nitrogen,
the identification of products
of incomplete combustion (PICs),
and
the correlation of candidate
incinerability indicators with
destruction efficiency.
The objectives of this report are to
(1) report and compare the results of the
two boiler tests and (2) identify and
highlight specific thermal destruction
measurement limitations that were encoun-
tered during the test programs.
TABLE 1. BOILER DESIGN AND OPERATING DATA
Boiler No. 1
Boiler No. 2
Manufacturer FW
Model number AG252
Heat exchanger
Design fuel . Gas
Design excess air 3%
Design steam rate, Ib/hr 68,000
Steam temperature, F 353
Steam pressure, psig 250
Design fuel heating value, Btu/lb 23,000
Waste heating value, Btu/lb 10,600
CE
VU-10
-Water Tube-
No. 6 oil
3%
60,000
406
125
18,500
17,700
FW - Foster Wheeler
CE - Combustion Engineering
63
-------
APPROACH
The boilers were operated at a single
steady state load of 80% of design capacity
with operator settings of excess air and
feed rate controlled by normal operating
procedures. A series of three background
runs and three waste feed runs were con-
ducted at the prescribed boiler load.
Additional sampling runs were required
during Boiler No. 1 testing to evaluate
NQx control with and without staged
combustion.
Characteristics of the boiler feed
streams and procedures for sampling and
analysis are described in the following
sections.
Feed Characteristics
Boiler No. 1 was fired with fuel gas
and a liquid nitrogenous waste. The fuel
gas was predominantly natural gas with
small amounts of a waste gas not substan-
tially different from natural gas. The
nitrogenous waste was predominantly
nitrobenzene with the balance consisting
of benzene, aniline and water. Waste to
fuel ratio on a weight basis averaged
0.2 to 1 at a liquid waste feed rate of
0.6 gpm (362 Ib/hour).
Initial boiler design included two
vertically oriented ring burners. The
burner system had been modified by the
addition of steam atomized liquid burners.
The boiler controls had been modified to
achieve staged combustion during periods
of liquid waste firing by restricting air
flow to the bottom set of liquid and gas
burners. Air not supplied to the lower
burners was directed to the upper burner
to maintain the total air flow necessary
for boiler excess air requirements. Two
series of sampling runs were conducted to
compare NOX emissions with and without
combustion control. To do this, the
automatic controls for restricting air to
the lower burners was disabled. Balanced
air flow to each set of burners was thus
maintained to establish the "unstaged"
combustion effect.
Boiler No. 2 normally burns one of
two fuels that are classified according to
viscosity. Generic designations are
"heavy oil" for the high viscosity liquids
and "light oil" for the low viscosity
liquids. Various blends can be created of
No. 6 fuel oil with various combustible
liquids such as recycled solvents, alco-
hols, mineral spirits and substandard fuel
oils. The blends vary in composition
according to availability. Boiler No. 2
is capable of burning two different blends
simultaneously and this option was chosen
for testing. The boiler has four burners
oriented horizontally on the front face of
the boiler. The heavy oil and light oil
were not substantially different in heating
value. Therefore, approximately equal
amounts of each feed stream could be
burned and a set of two burners each were
set aside for each feed type. Heavy oil
was steam atomized and light oil was air
atomized.
Waste liquids were spiked with a
standard test mixture: carbon tetra-
chloride, 1,1,2-trichloroethylene, mono-
chlorobenzene, and toluene. To protect
the boiler surfaces from chloride corrosion
damage, quantities were limited to one
weight percent as chlorine of the total
fuel and waste. Selection of test mixture
components is an attempt to provide a
broad range of incinerability of hazardous
compounds as postulated from relative
heats of combustion (1). Actual total
chlorine in the boiler feed was 0.7 weight
percent for both boiler tests. The
standard test mixture was added to the
light oil of Boiler No. 2. Table 2 shows
typical feed stream compositions for the
two boilers.
Sampling
Feed and emissions samples were
collected simultaneously. Boiler No. 2
was operated with two liquid feeds during
testing and samples of each were collected,
composited and analyzed separately. One
sample per run was collected of Boiler
No. 1 feed gas.
Emissions were sampled with three
sampling trains:
the volatile organic sampling
train (VOST) for volatile
organics,
the modified EPA Method 5 train
(MM5) for semi-volatile organics
and metals, and
the EPA Method 5 train for HC1
and metals.
64
-------
TABLE 2. FEED COMPOSITIONS
Weight % of fuel + waste
Boiler No. 1
Boiler No. 2
Fuel gas 83.8
Standard test mixture
Carbon tetrachloride 0.3
Trichloroethylene 0.3
Monochlorobenzene 0.5
Toluene 0.5
Liquid waste
Nitrobenzene 13.4
Phenol not detected
Aniline 0.4
Benzene 0.3
m&p-Xylene not detected
o-Xylene not detected
Water 0.5
Small quantities of not used
typical fuel oil
components and solvents
Total POHCs•detected 15.7
not used
0.5
0.4
0.4
2.0
not detected
11.8
not detected
0.2
4.3
0.6
1.1
balance
20.2
The VOST train collects the more
volatile organics from a filtered, pre-
cooled gas stream by adsorption on Tenax
resin. The, gas was passed through two
sorbent traps in series. The first trap
contained Tenax and the second trap con-
tained a layer of Tenax followed by a
layer of charcoal. Protection from sample
contamination by trace organics dictates
strict protocols for cleaning glassware,
for sealing glass joints without the
use of grease, and air tight sealing of
sorbent traps during storage and shipment.
Minimum leakage at glass joints was assured
by strict adherence to leak check proce-
dures and guidelines. The samples were
recovered from the Tenax by thermal
desorption and analyzed by gas chromato-
graphy/mass spectrometry procedures.
The modified Method 5 train collects
organics with boiling points above a
nominal 100 C level by adsorption on
XAD-2 resin. An EPA Method 5 train was
modified by addition of a sorbent trap
ahead of the impinger train. Temperature
of the sorbent traps was controlled so as
not to exceed a 20 C exit temperature.
Organics collected in traps were recovered
by solvent extraction from which the
sample-containing solvent was analyzed by
GC/MS after appropriate concentration or
dilution. Some samples were prescreened
by GC/FID to determine expected calibration
ranges for the GC/MS.
The EPA Method 5 train was used to
collect HC1 samples at a single point on
the duct traverse. The HC1 was absorbed
in 1.0 molar solution of sodium acetate
and the sodium acetate combined with
filter extract was analyzed by ion chro-
matography. Metals were also collected by
this sampling train during Boiler No. 1
testing. Boiler No. 2 testing utilized a
second EPA Method 5 train to collect
particles for metals analysis. Method 5
procedures were employed.
65
-------
PROBLEMS ENCOUNTERED
Benzene Interference
Feed Preparation
Specific quantities of the standard
test mixture (carbon tetrachloride, tri-
chloroethylene, monochlorobenzene, and
toluene) had to be added to the plant
wastes and mixed well enough to attain
consistent feed composition throughout the
test period. A premixed batch was prepared
in enough quantity to provide all of the
feed mixture needed. Examination of
sample mixtures prepared prior to testing
showed that only one liquid phase was
present. However, variation in density at
different tank levels was observed for'the
Boiler No. 2 liquid feed mixture. Hence,
constant circulation with rapid turnover
of the stored liquid was required to
minimize stratification.
Ambient Air Contamination
The VOST sampling train is equipped
with a charcoal filter to trap organics of
the ambient air when the train is pres-
surized for leak testing. It was found
during Boiler No. 1 testing that organics
in the ambient air exceeded organic
emissions. The modified Method 5 train
was not equipped with a similar air
cleaning filter until Boiler No. 2 testing.
Evidently, there were no POHCs or Pics in
the ambient air in detectable quantities
during Boiler No. 1 testing that would
contaminate modified Method 5 samples.
Nevertheless, as a precaution, all ambient
air to the organic sampling trains will be
filtered during future test programs.
Identification of Products of Incomplete
Combustion (PICs)"
Comparisons of analytical results of
all test runs, both waste feed and baseline
runs, consistently show the same compounds
in the extracts from XAD-2 emission
sampling traps. The same compounds are
found in the extracts from field blanks.
These compounds are attributed to degrada-
tion/leaching of XAD-2 resin and not to
PIC formation. The collection method
(XAD-2 resin) followed by fused silica
capillary chromatography with mass spectro-
meter detection allows for excellent
determination of the destruction efficiency
of POHCs, but the identification and
quantisation of PICs must be approached
with caution.
Benzene analysis indicates that
destruction efficiencies were less than
99.99% during both Boiler No. 1 and Boiler
No. 2 testing. Statistical treatment of
Boiler No. 1 test data showed that benzene
emission levels during the waste feed runs
were not significantly different from the
baseline runs. We thus attributed benzene
concentrations in the emissions to a
benzene background in the Tenax sampling
trap. Benzene levels during Boiler No. 2
testing were an order of magnitude greater
in the baseline runs,than in the waste
feed runs despite the fact that no benzene
was detected in the baseline fuel (No. 6
fuel oil). This is a different set of
circumstances than the Boiler No. 1 case
and a phenomena we are unable to explain.
Unfortunately, destruction efficiency of
benzene was not measureable during the two
boiler tests due to the apparent benzene
interferences.
RESULTS
Program accomplishments were as
fol1ows:
Destruction efficiencies of ten
POHCs (principal organic hazard-
ous constitutents) were deter-
mi ned.
Hazardous waste disposal
performance at low and high
waste/fuel ratios were compared.
Data for correlation of destruc-
tion efficiency with operating
parameters were provided.
Very few PICs (products of
incomplete combustion) were
detected and identified.
NOX emissions when burning a
nitrogenous waste was compared
with and without combustion
control.
Heavy metals emission data were
collected.
Chloride mass balance calcula-
tions showed greater than 100%
closures (127% for Boiler No. 1
and 108% for Boiler No. 2).
66
-------
Boiler No. 1, which is used to solve a
specific waste disposal problem without
large fuel savings, was successful in
destroying the plant's wastes at the
concentrations fed to the boiler. Boiler
No. 2, which capitalizes on various waste
disposal opportunities to accomplish
significant fuel savings, realized accept-
able DEs except for xylenes for relatively
high concentrations of POHCs .in the feed.
Destruction Efficiency
Destruction efficiency (DE) was
determined by analyzing for POHCs of
interest in the feed and in the emissions,
calculated as follows:
DE = 100 (POHCin - POHCout)/POHCin
The POHCs include the components of the
standard test mixture and the major
hazardous compounds in the waste feed. DE
requirements of 99.99% were exceeded for
all compounds except xylenes (Table 3).
Better than 99.99% DE was accomplished at
POHC levels in the feed of 16% (Boiler
No. 1) and 20% (Boiler No. 2).
Correlation with Boiler Operation
The test programs provide an
opportunity to look for a correlation of
easily measured boiler parameters with
destruction efficiencies. The purpose is
to develop indicators that could sub-
stitute for the time consuming and costly
destruction efficiency tests. Two of the
indicators often mentioned are carbon
monoxide (CO) and total hydrocarbon con-
centrations in the flue gas. CO and total
hydrocarbon (THC) concentration did not
correlate with destruction efficiency at
the low levels of emissions concentrations
encountered. A possible explanation is
that even the low levels of CO and THC
represent orders of magnitude higher
concentrations than the concentrations of
POHCs in the, emissions. For example; the
highest CO concentration observed, 242 ppmv,
reflects a feed carbon to unoxidized
carbon ratio of 430 to 1 whereas a 99.99%
destruction efficiency of a POHC represents
a 10,000 to 1 ratio of feed POHC to
emissions POHC.
TABLE 3. DESTRUCTION EFFICIENCY
POHC
Boiler No. 1
Staged
combustion
Unstaged
combustion
Boiler No. 2
Standard test mixture
Carbon tetrachloride 99.997
Trichloroethylene 99.999
Monochlorobenzene 99.998
Toluene 99.997
99.997
99.999
99.998
99.998
99.999
99.999
99.999
99.999
Liquid waste
Nitrobenzene
Phenol
Aniline
o-Xylene
m&p-Xylene
99.999
not detected
99.998
not detected
not detected
99.999
not detected
99.998
not detected
not detected
not detected
99.999
not detected
99.958
99.947
Values beyond three decimal places were ignored.
67
-------
Products of Incomplete Combustion (PICs)
PICs were found in the Boiler No. 2
emissions and consisted of substituted
phenols (methyl butyl phenol, dimethyl
phenol). Their presence is probably due
to the high concentration of phenol in the
boiler feed. The substituted phenols are
not the XAD-2 artifacts that were discussed
in the Problems Encountered section.
NOx Control When Burning Nitrogenous Haste
Gas fired Boiler No. 1 had been
modified to accommodate cofiring of a
mixture of nitrobenzene, aniline and
benzene with fuel gas. Included in the
modifications were changes in air distri-
bution to the burner to provide staged
combustion NOx control. Testing confirmed
that staged combustion was effective in
controlling NOx emissions (Figure 1). NOX
emissions were 73% less than the equivalent
organic nitrogen present in the liquid
waste. Organic nitrogen compounds were
not detected in the fuel gas. Corrected
for background emissions, which are all
derived from combustion air, estimated NOx
emissions were 78% less than the organic
nitrogen in the feed. Without staged
combustion, organic nitrogen was lowered
by only 24% or by an estimated 28% if
corrected for baseline emissions.
Fate of Metals
Particle catches were analyzed for
specific heavy metals. Particulate
matter emissions from Boiler No. 1 were
quite low and specific metal concentrations
exceeded minimum detectable limits only
slightly if at all. Also, differences
between liquid waste runs and background
runs v/ere not distinguishable for most
metals. Mercury and lead were the only
exceptions. It was concluded that heavy
metals emissions from Boiler No. 1 were
not a significant problem.
Boiler No. 2 feed streams, both the
No. 6 fuel oil and the waste solvents,
contained significant amounts of several
of the heavy metals. Strainers were used
to remove bulk quantities of solids from
the boiler feed streams. This is usual
practice with fuels of the type used.
Thus, metals concentrations in the filtered
feed streams emulate typical practice.
The two feed streams were examined for
fourteen metals. Nine metals were found
in excess of one ppmw (Table 4). Signif-
icant levels of vanadium, nickel, zinc,
lead and cobalt were found in the baseline
fuel (No. 6 fuel oil). Substantial
quantities of vanadium, zinc and lead in
the light oil added to the potential for
having significant concentrations of these
metals emitted to the atmosphere. Also,
the light oil contained a significant
level of manganese. Since about one third
of the total boiler feed during the waste
disposal runs was baseline No. 6 fuel oil,
it was of interest to know whether there
were significant differences between
baseline emissions and waste firing
emissions. Waste firing resulted in
higher emissions of zinc, lead, cadmium,
copper, manganese, chromium and cobalt.
Vanadium and nickel were the only metals
in significant concentrations in No. 6
fuel oil that did not show an increase.
Of the metals showing an increase, con-
centration ranges during waste firing were
as follows:
zinc:
lead:
cadmium:
chromium
copper:
manganese:
chromium:
cobalt:
ACKNOWLEDGEMENTS
>1000 yg/m3
>1000 yg/m3
100 - 1000 yg/m3
10 - 100 yg/m3
10 - 100 yg/m3
10 - 100 pg/m3
10 - 100 yg/m3
10 - 100 yg/m3
The authors gratefully acknowledge
the cooperation of the management of the
host facilities for providing boilers to
test and for donating employee time to
help coordinate the effort. This project
is sponsored by the U.S. EPA Industrial
Environmental Research Laboratory -
Cincinnati. Technical project monitor for
EPA is Mr. Robert Olexsey.
REFERENCES
1. Olexsey, R.A., 1983. Incineration of
Hazardous Waste in Power Boilers:
Emissions Performance Study Rationale
and Test Site Matrix. U.S. Environ-
mental Protection Agency, Cincinnati,
Ohio.
68
-------
2.0-
.c
o
.a
c
0)
CO
'1- 1.0-
UJ
u
vvwwywww
MB
BwfflsBS
mm
mm
mm*
nstage
^ — Fuel Bound N—^
NOX
Emissions
iJ&J&tefiS*
V* '&**<[ t W
sd Combustion
Figure 1. Boiler No. 1 f
TABLE 4. TRACE [*
mm,
•w
Staged
JO con
A
1ETALS
I
NOX
Emissions
Emissions
Combustion Background
trol performance.
ANALYZED
Greater than Greater than
one p'pm in , one ppm in
feed? feed?
Zinc Yes Chromium Yes
Lead Yes Cobalt Yes
Vanadium Yes Arsenic No
Nickel Yes Mercury No
Cadmium Yes Antimony Mo
Copper Yes Selenium Mo
Manganese Yes Strontium Mo
69
-------
SUMMARY OF FIELD TESTS FOR AN INDUSTRIAL BOILER
DISPOSING OF HAZARDOUS WASTES
John T. Chehaske
Engineering-Science
10521 Rosehaven Street
Fairfax, Virginia 22030
and
Gregory M. Higgins
SYSTECH Corporation
245 North Valley Road
Xenia, Ohio 45385
ABSTRACT
This paper presents the results of field tests of a North American Model 3200X
package fire-tube boiler burning a mixture of toluene and chlorinated solvents. Two
fuel blends were tested. Blend #1 contained 98 wt % toluene, 1 wt % trichloroethylene,
and 0.5 wt % each of chlorobenzene and carbon tetrachloride. Blend #2 differed in that
the trichloroethylene concentration was increased to 2 wt % and the toluene concentration
was reduced to 97 wt %. Flue gas samples were collected with the VOST, Modified Method
5, and Method 23 sampling systems. The VOST and Modified Method 5 samples were analyzed
by GO/MS while the Method 23 samples were analyzed by on-site GC. Continuous analyzers
were used to measure the oxygen, carbon monoxide, and nitrogen oxides content of the
flue gas.
INTRODUCTION
The U.S. Environmental Protection
Agency (EPA) regulates the destruction of
hazardous wastes in incinerators requir-
ing a Destruction and Removal Efficiency
(ORE) of 99.99 percent (%) for the Prin-
cipal Organic Hazardous Constituents
(POHCs) that are present in the waste.
Particulate matter emissions are limited
to 0.08 grains/dry standard cubic foot
(gr/dscf) corrected to 12% carbon diox-
ide (C02), and hydrogen chloride (HC1)
emissions are limited to 4 pounds per
hour (Ib/hr) or 1% of the uncontrolled
HC1 whichever is greater. Processes
which primarily produce energy and coin-
cidentally burn hazardous wastes are cur-
rently exempt from hazardous waste des-
truction regulation. This includes pro-
cesses such as industrial boilers, kilns,
blast furnaces, process heaters, and
ovens.
The EPA is investigating the need for
regulating hazardous waste combustion in
thermal industrial processes. A Regula-
tory Impact Analysis (RIA) is being con-
ducted to determine the need for and ef-
fects of regulating these processes.
Part of the input needed for the RIA is
information on the ability of these pro-
cesses to destroy hazardous wastes.
This paper presents data obtained
from tests of a fire-tube boiler burning
100% hazardous waste. The project was
sponsored by the EPA Office of Solid
Waste (OSW). The OSW Project Manager was
Mr. Marc Turgeon. The project was con-
ducted under contract from the EPA Indus-
trial Environmental Research Laboratory
(IERL) in Cincinnati, Ohio, who provided
the technical direction. Mr. Robert A.
Olexsey was the IERL Technical Manager.
Engineering-Science (ES) was the prime
contractor for the project and was
70
-------
responsible for overall project coordina-
tion; collection of the Volatile Organic
Sampling Train (VOST), Modified Method 5
(MM5), and HC1 samples; and report prepa-
ration.
ES was assisted by SYSTECH Corpora-
tion, who was responsible for obtaining
the test site, coordinating-with the
plant, and conducting EPA Proposed Method
23 (M23) sampling. Analyses of the VOST
and MM5 samples were performed by Re-
search Triangle Institute.
PURPOSE
This test was one of a matrix of
about 14 tests conducted by IERL to ob-
tain DRE data for industrial boilers.
The matrix included a wide range of
boiler sizes and types. This particular
test was to provide DRE data for a small
fire-tube boiler. Most of the other
tests involved co-firing of hazardous
waste with the normal boiler fuel. This
test, however, was designed to determine
DRE for 100% hazardous waste burning, a
practice that is known to be used by some
facilities.
There was a secondary purpose of this
test program: to compare the results ob-
tained with VOST and MM5 to those ob-
tained with M23.
APPROACH
Boiler Description
All field testing was accomplished on
a North American Model 3200X package fire-
tube boiler. This unit is referred to as
an "Ohio Special" because it has only 358
ft2 of flue gas to water heat exchange
surface, and therefore does not require a
full-time licensed operator. "Ohio Spe-
cials" probably represent a worst-case
situation for destroying hazardous wastes
in fire-tube boilers. With a high heat
release of 92,000 Btu/hr ft2 and a high
liberation rate of 164,400 Btu/hr ft3,
it is expected that high combustion tem-
peratures and very short retention times
occur in the furnace of the test unit.
However, ..the nonconservative design of
an "Ohio Special" may also have some
advantages when well controlled combus-
tion is desirable. It was discovered
that the unit used in these tests has a
very narrow range of excess air levels
within which it could operate. Below the
minimum used in the tests (17% excess
air), combustion was incomplete resulting
in smoke formation. Above the maximum
excess air used in the tests (53%) the
fire would blow out the back of the com-
bustion chamber and the system would au-
tomatically shut down. Furthermore,
since this type of boiler is kept at a
positive furnace pressure, a large change
in excess air between minimum firing irate
and maximum firing rate will not occur as
it would in a negative pressure furnace.
The burner of the test unit is de-
signed to fire on either gas or distil-
late oil. For normal operation, natural
gas is fired. Therefore, the fuel oil
firing system was available to be
cleaned, adjusted, and used to fire test
blends of liquid chemicals without inter-
rupting normal operation of the boiler.
The fuel oil system consists of a 1000-
gal cylindrical tank, a pump with mecha-
nical Teflon® seals, an atomizing air
compressor, and a spray nozzle. The
storage tank is plumbed to a nearby tank
farm to permit filling the 1000-gal tank
from long-term storage. Mixing is pro-
vided in the storage tank by a recycle
loop from the boiler.
Test Fuel
To provide control of fuel composi-
tion and assure consistent purity in the
materials burned for all test conditions,
virgin chemicals were purchased and
blended to simulate a hypothetical hazar-
dous waste. Reagent grade carbon tetra-
chloride, technical grade monochloroben-
zene, and technical trichloroethylene
were carefully weighed to the desired
amounts and added into the fuel tank.
The tank was then topped off with nitra-
tion-grade toluene (>99% pure). Fuel was
usually blended the evening before each
test day so that the chemicals could mix
overnight by recirculating through the
fuel pump circuit. Table 1 shows the
composition of the two test fuels fired
in this program.
71
-------
TABLE 1. AVERAGE POHCs COMPOSITION OF
TEST FUEL
Carbon
Fuel tetra- Trichloro- Chloro-
blend Toluene chloride ethylene benzene
no. (wt %) (wt %) (wt %) (wt %)
1
2
98
97
.5
.5
1
2
.5
.5
Test Matrix
Testing was conducted at six differ-
ent boiler operating conditions. Table
2 summarizes the test matrix of fuel
blend and boiler load/excess air (EA).
Triplicate samples were collected with
each test method at each test condition,
except condition 6 for which only dupli-
cate samples were collected.
TABLE 2. TEST MATRIX
Test
condition
no.
1
2
3
4
5
6
Fuel
blend
no.
1
1
2
1
1
2
Boiler
conditions
half load
full load
half load
full load
(high EA)
half load
(high EA)
full load
(load EA)
Procedures
The two volatile organic components
of the fuelr carbon tetrachloride (CC14)
and tricoloroethylene (TCE) were sampled
and analyzed according to the VOST pro-
tocol (1). Each test run consisted of
three pairs (one Tenax®-GC and one
Tenax®-GC/charcoal tube) except test 16
which consisted of six pairs and test 17
which consisted of four pairs. Sampling
was conducted at 0.5 liters per minute
(1/m). The volume sampled by each pair
of tubes was 20 liters (1). VOST ana-
lytical results were subjected to the
following acceptance criteria:
o Both surrogate compounds must have
been detected.
o The fractional recovery of both
surrogates must have been between
0.5 and 1.5.
OR
o The fractional recoveries of both
surrogates must have been less
than 5.0 and the two fractional
recoveries must have agreed within
25%.
(recovery 1 divided by recovery 2
is between .75 and 1.25)
The semi-volatile organic compounds,
toluene and monochlorobenzene (MCB), were
sampled and analyzed by the MM5 procedure
(2). Analysis of the resultant extracts
was by GC/FID.
In addition to the VOST and MM5 sam-
pling, the POHCs in the flue gas were de-
termined by M23 (3). This method, shown
schematically in Figure 1, employs inte-
grated bag sampling followed by gas chro-
ma tography.
FLOW
METER
\
PUMP
Figure 1
Schematic Diagram, Method 23
Sampling Equipment
72
-------
The gas samples were collected in
Tedlar® bags at a flow rate of 0.5 1/min
over a 40-min interval simultaneously
with the first leg of VOST sampling.
Prior to testing, all Tedlar® bags were
assembled, filled with ultra-pure nitro-
gen, leak checked, and tested by GC for
background contamination. Toluene and
chlorobenzene were analyzed by FID, while
BCD was used for the remaining chlori-
nated POHCs. The sampling systems were
also assembled and leak checked. Field
sampling for POHCs consisted of two ba-
sic functions: preparation and analysis
of calibration standards and flue gas
sampling. Both functions were performed
as described in M23. Spikes and repli-
cate analyses were included in the flue
gas analyses.
PROBLEMS ENCOUNTERED
VOST
An unusually large number of the VOST
tubes were broken in shipment from the
laboratory to the field. The breaks oc-
curred at the Swaglok® fittings, indicat-
ing the fittings had been overtightened,
had bottomed out on the tube, or the tube
diameter was too large. Two tubes were
broken on return to the laboratory, re-
sulting in lost samples. There were pro-
blems with the GC/MS that resulted in two
more lost samples, and there were four
tube pairs that were lost by the labora-
tory. These lost samples combined with
the data that were excluded because of
poor surrogate recoveries resulted in a
total of 21 tube pairs that were ex-
cluded out of a total of 55 tube pairs
collected.
enough to verify an average DRE of 95.64%
when MCB was not detected, and MCB was
not detected in any of the 18 test runs.
RESULTS
The boiler performed very well
throughout the testing program. Only a
few minor delays were encountered when
steam production exceeded demand. Table
3 lists the average carbon monoxide (CO)
and nitrogen oxides (NOX) concentrations
for each of the six test conditions.
With the exception of condition 3 and 6
the CO results track boiler conditions as
expected. For condition 3 and 6 the CO
concentrations were lower than expected.
These were the two runs with the highest
fuel chlorine content, but it does not
seem likely that the additional chlorine
would have depressed the CO concentra-
tions. The NOX concentrations tracked
boiler conditions as expected, although
the NOX concentration for condition 6
was lower than expected.
TABLE 3. CO AND NOX RESULTS
Condi-
tion
no.
1
5
3
2
4
6
Boiler
load/EA (%)
half/38
half/50
half/31
full/38
full/41
f ull/1 7
Chlori-
nated
Solvent
(%)
2
2
3
2
2
3
CO
(ppm)
108
89
10
142
86
20
NOX
(ppm)
172
130
158
82
75
84
Modified Method 5
The analytical detection limit for
MCB was four times higher than had been
expected by the laboratory. As a result,
for those runs where MCB was not de-
tected, the maximum calculable DRE was
less than 99.99%. There were 11 of 17
runs when MCB was not detected.
Method 23
Similar detection limit problems were
encountered for MCB with the M23 proce-
dure. The detection limit was only low
Destruction Removal Efficiencies
DREs were calculated from measured
fuel feed rates and compositions, flue
gas flow rates, flue gas molecular
weights, POHC flue gas concentrations,
and POHC molecular weights.
Tables 4 and 5 show the calculated
DREs obtained from the VOST and MM5 sam-
pling, respectively. The values are the
mass weighted averages for each, boiler
test condition. For toluene, carbon te-
trachloride, and trichloroethylene the
73
-------
DREs ranged from a low of 99.9976 to a
high of 99.9999. The MCB results were
lower, primarily because of the high
analytical limit of detectability men-
tioned previously. The reported MCB
DREs were calculated by inserting the
minimum detectable quantity for the 11
runs where no MCB was detected. There-
fore, the reported MCB DREs are lower
than the actual DREs.
TABLE 4. DREs BY VOST
Condition
no.
1
3
5
2
4
6
Boiler
load/EA (%)
half/38
half/31
half/50
full/38
full/41
f ull/1 7
CC14
DRE (%)
99.9979
99.9990
99.9993
99.9989
99.9998
99.9992
TCE
DRE (%)
99.9999
99.9995
99.9976
99.9998
99.9999
99.9996
TABLE 5. DRES BY VOST
Condition
no.
1
3
5
2
4
6
Boiler
load/EA (%)
half/38
half/31
half/50
full/38
full/41
f ull/1 7
Toluene
DRE (%)
99.9997
99.9992
99.9991
99.9991
99.9994
99.9991
MCB
DRE (%)
99.952
99.978
99.972
99.946
99.948
99.973
For the M23 results, there were only
four test runs in which POHCs were found
in the flue gases. In the cases where no
POHCs were detected, the analytical detec-
tion limit was used to calculate the DRE.
Therefore, the actual DREs occurring in
the boiler were at least as great or
greater than the calculated DREs. With
the exception of MCB, the M23 analytical
procedures were sensitive enough to sub-
stantiate at least 99.99% DRE. Because
most of the DREs were based on detection
limits, the individual values by test
condition are not given. Instead, Table
6 shows the number of runs for which DREs
greater than and less than 99.99% were
verified.
TABLE 6. DREs BY M23
POHC
Toluene
CC14
TCE
MCB
No. of
cases DRE
>99.99%
17
17
12
0
No. of
cases DRE
>99.99%
•l(a)
0
3(b)
18
measured DRE v 99.98
average measured DRE = 99.46
Table 6 illustrates the average DREs
calculated for each of the four POHCs by
the M23 procedure. The DREs reported in
Table 6 illustrate that the EPA Method
23, coupled with GC/FID-ECD analysis, was
sufficient to subtantiate DREs of at
least "four nines" occurring in the test
boiler for toluene, CC14, and TCE. Be-
cause of the low instrument sensitivity
to MCB, however, this method could not
provide adequate sensitivity for DRE mea-
surement of this POHC under the given
test conditions. The best DRE calculable
for MCB under these conditions averaged
95.64.
Methods Comparison
A direct comparison of the POHC con-
centrations measured by the various me-
thods is not possible because only four
of the 72 determinations by M23 yielded
detectable POHC quantities. It is pos-
sible, however, to compare the DREs ob-
tained with the three methods as shown
in Table 7. The test results for all six
test conditions was combined to give a
mass weighted average DRE for each POHC.
The VOST and MM5 test procedures yielded
higher DRE values for each POHC than did
the M23 procedure. The reason that the
DRE values are higher for VOST and MM5
than for M23 is because VOST and MM5 had
lower limits of detectability. With the
exception of MCB, VOST and MM5 were able
to measure actual flue gas concentrations
and thus actual DREs could be calculated.
74
-------
Method 23 could only provide information
that the ORE was at least as great as the
calculated value, based on the higher li-
mit of detectability inherent with M23.
TABLE 7. DREs COMPARISON
Compound
Toluene
CC14
TCE
MCB
DRE by
VOST or MM5
99.999
99.999
99.999
>99.961
DRE by
Method 23
>99.998
>99.954
>99.995
>95.642
It is important to emphasize that the M23
DREs reported in Table 7 are more a re-
flection of limitations in analytical
methods sensitivity than of the ability
of the test boiler to destroy or remove
the test POHCs. The test boiler likely
destroyed or removed a larger percentage
of the input POHC load than is reported.
REFERENCES
1 . Protocol for the Collection and An-
alysis of Volatile POHCs Using VOST,
Envirodyne Engineers, Inc.,
St. Louis, Missouri, August 1983.
2. Sampling and Analysis Methods for
Hazardous Waste Combustion (First
Edition, Arthur D. Little, Inc.
Cambridge, Massachusetts, February
1983, pg 36.
3. Federal Register, 45 FR 39766,
October 3, 1980.
ACKNOWLEDGEMENTS
This project was conducted under Con-
tract No. 68-03-3040, Work Assignment No.
SBE04 for the U.S. Environmental Protec-
tion Agency, Industrial Research Labora-
tory, Incineration Research Branch,
Cincinnati, Ohio.
The authors wish to thank Mr. Jon N.
Bolstad of Engineering-Science and Mr.
Ned J. Kleinhenz of SYSTECH Corporation
for their diligent efforts as field team
leaders for the two test crews.
75
-------
PARAMETRIC EXPERIMENTATION WITH A PILOT-SCALE
BOILER BURNING HAZARDOUS COMPOUNDS
C. D. Wolbach
Acurex Corporation
Mountain View, California 94042
ABSTRACT
Thermal destruction of wastes by direct incineration or by cofiring with
conventional fuels in boilers, furnaces, or kilns is one of the most effective methods
currently available for disposal of hazardous organic material. While direct
incineration of hazardous wastes is regulated by Part 264 of the Resource Conservation
and Recovery Act (RCRA) as adopted in January 1981, boiler cofiring is currently exempt
from RCRA provisions. However, the potential for boiler cofiring regulations has been
evaluated by the Environmental Protection Agency (EPA) and they are now in the process
of preparing regulatory positions in drafting regulations for promulgation. To support
this effort, EPA's Incineration Research Branch (IRB) in conjunction with the Office of
Solid Waste is conducting research and development programs on incineration
effectiveness and regulatory impacts.
The global purpose of this study was to gather data to aid the EPA in selecting a
strategy or set of strategies for regulating the combustion of hazardous wastes in
boilers. The specific objectives were twofold: to identify which of several boiler
operational parameters have a major impact on boiler destruction and removal efficiency;
and, to evaluate and if practical, establish a mathematical model for predicting an
upper limit on the amount of cofired waste that could be emitted. In particular, those
parameters that could be easily changed by an operator or might represent major
differences between boiler types were studied. A secondary objective of the study was
to gain sufficient information to allow judgements of what particular parameters to
monitor closely during full-scale testing.
GENERAL APPROACH AND TEST MATRIX
The experimental goals of the
program, in sequence, were to:
• Determine the furnace operating
parameters which affect the
furnace three-dimensional
time-temperature environment
• Quantify these effects
• Identify the relationships
between operating parameters,
time-temperature history, and
destruction of hazardous waste
components in the fuel
The test plan divided the tests
into three subsets: a baseline study
burning only distillate oil, a single
compound study of one hazardous waste
compound cofired with distillate oil,
and a multiple compound study burning
more than one compound simultaneously.
The baseline study was designed to
define the thermal environment within
the facility under the various operating
conditions of choice, to define the
magnitude of the effects of changes in
variables on the thermal environment,
and to demonstrate the ability of a
simulation to predict the thermal
environment resulting from a specific
set of variable settings. The single
76
-------
compound studies had two goals: to
define the magnitude of variable effects
on destruction efficiency; and to define
the effects of cofiring wastes on the
expected thermal environment. The final,
multiple compound studies were to
determine the destruction efficiencies of
several compounds with known thermal
environments. Chlorinated wastes were
selected for sensitivity and ease of
analysis. Chlorobenzene was selected for
the single compound tests because of its
high heat content, compatibility with
distillate oil, and simple structure.
For the multiple compound tests
chlorobenzene was fired with carbon
tetrachloride, chloroform, methylene
chloride, and dichloroethane.
Within each task a set of
independent variables were identified
that are relevant to the operation of
incinerators and boilers.
Table 1 shows the test matrix of
nominal operating conditions used in the
program. Firing rate and stoichiometry
parameter values are listed at nominal
conditions based on analysis of the pure
distillate oil fuel. As shown, two
nominal firing rates (0.2 and 0.38
megawatts) and three nominal
stoichiometries (10, 25, and 50 percent
excess air) were used.
The highest concentration of waste
to fuel was achieved with chlorobenzene
(10 percent v/v). Other materials were
injected at levels of approximately 1.5
to 3.5 percent (v/v).
A total of 40 tests were run
generating 99 individual data points.
Fourteen firing conditions were studied
and the destruction and removal
efficiency relative to four compounds was
examined. A computer simulation was
prepared that predicted the thermal
profile within 25°C and the destruction
and removal efficiency within two orders
of'magnitude.
FACILITY DESCRIPTION
All tests were performed in the
Acurex Pi lot-Scale Furnace. The furnace
has a single burner, front wall fired,
into a horizontally oriented firebox.
The premixed fuel oil/waste mixture is
pumped out of drums through a
pressure-atomizing nozzle and stabilized
at the front wall. The fuel flow is
monitored by rotometer. Combustion air
is preheated and injected in the annular
region around the fuel delivery tube.
The IFRF/Acurex burner design allows
swirl adjustment by rotating swirl
blocks. Air flow is monitored by hot
wire anemometer.
The walls of the radiant section are
refractory lined with optionally mounted
waterwalls. Ports along the top and
sides of the furnace accommodate ceramic
type R thermocouple probes for gas
temperature'profiles'. Type K probes are
TABLE 1. TEST MATRIX OF VARIABLES AND NOMINAL VALUES
Variable
Values
Firing rate (load)
0.23, 0.35 _+ 0.03 megawatts (0.8, 1.2 +_ 0.10
million Btu/hr)
Flame shape (swirl) 7.5, 5.0, 2.0
Excess air rate (stoichiometry) 10, 25, 50 percent, +2 percent
Waste to fuel ratio
Waterwall area
Compounds
10, 3, 1.5 percent
0, 8 percent
4
77
-------
embedded in the refractory for wall
temperature determinations.
Thermocouples are also placed at the
manifold inlet to and the outlet of each
cooling panel, and to and from the burner
quarl. Cooling fluid flow measurements
are by rotometer. Downstream of the
convective a stainless steel probe
extracts gas samples for continuous
emissions analysis. Electro-optical
analyses are used to define levels of CO,
C02, 02, NOX, and total unburnt
hydrocarbon (THC). Samples for the
Volatile Organics Sampling Train (VOST)
are extracted with a stainless steel
probe from the end of an 8-in. diameter
15-ft long duct. The gases are then
exhausted to a 45-ft high stack.
RESULTS AND DISCUSSION
Table 2 provides a complete summary
of the program results. The roman
numeral in the first column is an
identifier for the run condition. The
run condition is summarized in the second
column. The first item in parentheses
is the nominal firing rate in million
Btu/hr. The second and third items are
the percent excess air and the swirl
setting, respectively. The final item is
the percent of the radiative section
surface area covered by water walls. The
number in the compound columns represent
the average penetration value, C/C0,
(mass of waste out per unit, time/mass
rate of waste in), the value of one
standard deviation, and (in parentheses)
the number of data points. The
relationship to ORE is: ORE = 1-C/C0.
The effects of the various firing
conditions of DRE's are discussed in the
following sections. A semiquantitative
summary on the variable effects is given
in Table 3. These values are rough
estimates of the difference between
either temperature or ORE for two runs
where only the specified parameter was
TABLE 2. AVERAGE FRACTIONAL BREAKTHROUGH C/Cn FOR EACH WASTE AND FIRING CONDITION
Condition
Chlorobenzene*,t
Carbon
tetrachloride
Chloroform
1,2 Dichloroethane
I* (0.8, 10, 7.5, 8) 1.4 ± 0.1 x 10"* (4) 1.6 ± 0^2 x 10'4 (2) 1.6 ± 0.2 x 10"5 (3) 5 ± 2 x
Ia§ (0.8, 10, 7.5, 0) 4.2 x 10~6 (1) —i
II (0.8, 25, 7.5, 8) 1.4 ± 0.3 x 10-5 (4) 5 ± 3 x HT5 (5)
Ha (0.8, 25, 7.5, 0) 4 ± 3 x 10-7 (6)
III (0.8, 50, 7.5, 8) 8.9 ± 6 x 10-5 (5) 2>1 x 10-4 (1)
Ilia (0.8, 50, 7.5. 0) 2.5 ± 2 x 10-7 (3)
IV (1.2, 10, 7.5, 8) 6 ± 8 x 10-5 (5)
IVa (1.2, 10, 7.5, 0) 2.2 ± 0.5 x 10~7 (3)
V (1.2, 25, 7.5, 8) 4.5 ± 2 x 10-6 (5)
Va (1.2, 25, 7.5, 0) 0.5 ± 0.3 x 10-7 (3)
VI (1.2, 50, 7.5, 8) 5.5 ± 2 x 10-5 (4)
Via (1.2, 50, 7.5, 0) 3.4 ± 0.7 x 10-7 (3)
VII (0.8, 25, 5.0, 8) - 6.3 ± 3 x 10-5 (3)
IX (1.2, 25, 5.0, 8) 2.0 ± 0.5 x 10-6 (3) 1>5 ± 0>3 x 10-4 (3)
ID"6 (3)
1 ± 0.5 x lO-6 (2)
1 ± 0.2 x HT6 (3)
2.6 ± 2 x ID'5 (5)
2.0 ± 0.8 x ID'5 (3)
1.2 ± 0.3 x 10-* (2) 3.2 ± 0.5 x 10'5 (3)
2.5 ± 1 x ID'5 (2) 8.8 ± 2 x 10'5 (3)
7.9 ± 6 x lO-5 (3) 1.1 ± 0.02 x 10'4 (2) 8 ± 4 x 1Q-6 (2)
* t represents 1 standard deviation from average.
t ( ) Indicates number of valid data points.
+ Firing rate 1n million Btu/hr, percent excess air, swirl, percent water-wall.
§ Run conditions suffixed with an "a" are without waterwalls. All other conditions are with
water-walls 1n place.
I Not tested under that condition.
78
-------
TABLE 3. SEMIQUANTITATIVE ESTIMATION
OF THE EFFECTS OF THE
INDEPENDENT VARIABLES ON
TEMPERATURE PROFILE AND DRE
Variable
Waterwalls
Excess air rate
Firing rate
Burner swirl
Compound
Range of
Temperature
Changes
°C (±°F)
275 (500)
165 (300)
110 (200)
30 (50)
30 (50)
Range
of DRE
Changes
(Factor)*
100
10
3
1
20
*C/C0 is mass rate of waste out
divided by mass rate of waste in.
varied. For examplej a change in firing
rate between 0.8 million Btu/hr and
1.2 million Btu/hr changed the
temperature profile by about 110°C
(200°F). The corresponding change in DRE
was about a factor of three.
Thermal profiles were presented in
detail in a previous paper (1) and will
be discussed only incidentally to DRE
discussions. The temperature profiles
clearly show a thermal boundary layer
within four inches of the furnace wall.
This boundary layer is significant, since
it represents greter than 40 percent of
the volume or thermal environment of the
furnace. Because .temperature and
velocity are the dominant factors in
volume flowrate and destruction, this
zone will be significant when considering
destruction efficiencies in this furnace.
Effect of Waterwalls on Thermal
Environments and DRE's
Addition of waterwall panels
replaced about '8 percent of the
refractory surface area but extracted 50
to 60 percent of the heat input. With
the waterwalls in place the thermal
environment of the first six feet of the
firebox was comparable to a D-type
packaged watertube boiler. However, the
pilot-scale waterwall surface temperature
is much lower than for industrial boilers
(600° to'800°F). The effect will be to
lower DRE in the pilot unit with respect
to an industial boiler through wall
quenching.. This is offset by the larger
industrial boiler waterwall surface area
(>50 percent) compared to the pilot
unit.
Waterwalls have a dominant effect on
DRE. This is shown in Table 4. For
chlorobenzene DRE varied over two orders
of magnitude with the introduction of
wall cooling. Qualitatively, this was to
be expected' since the waterwalls
decreased the corresponding temperature
profiles 220 to 280°C (400° to 500°F).
Not surprisingly, the magnitude of the
effect is within the range predicted by
theory. Using the Arrenius expression
for a rate constant, k = Aexp (-Ea/RT),
and algebraically manipulating the
kinetic expressions to present the ratio
of breakthrough fractions for a given
waste under two different temperatures,
one obtains
= exp [E
- T2)/RT1T2]
Cj and ,C2 are the breakthrough quantities
at two different run temperatures
(assuming the same feedrate in), Ea is
the empirical pseudo first order energy
of activation for the disappearance of
the waste, R is the appropriate gas
constant, and Tj and T2 are the effective
temperatures. Using effective
temperatures of 1,100 and 1,380K, an
activation energy of 50 kilocalories per
gram mole, and a gas constant of
1.982 calories per gram mole °K, the
calculated ratio of breakthrough
quantities is 100.
TABLE 4. AVERAGE MEASURED PENETRATION
C/C0 OF CHLOROBENZENE
With Water Without Water
Condition Walls (x 105) Walls (xlO7)
(0.8,10,7.5)
(0.8,25,7.5)
(0.8,50,7.5)
(1.2,10,7.5)
(1.2,25,7.5)
(1.2,50,7.5)
14 + 1
1.4 + 3
8.9 + 6
6 +8
0.5 + 0.2
5.5 T 2
(4)
(4)
(5)
(5)
(5)
(4)
40
4 + 3
3 + 2
2 + 0.5
0.5 + 0.3
3 +_ 0.7
(1)
(6)
(3)
(3)
(3)
(3)
79
-------
Effect of Excess Air on Temperature
Profiles
The effects of excess ai r on
temperature profiles followed
expectation. A comparison of Figures 1
and 2 shows effects of excess ai r on
centerline axial temperature profiles for
different firing rates. An alternate
demonstration is given in Figure 3.
Here, radial temperature profiles at
two positions are shown for three excess
air levels at a nominal 0.8 million
Btu/hr firing rate. All three conditions
are without waterwalls. On the
centerline, gas temperatures vary as much
as 140°C (250°F) at cross section C-3
(approximately 80 in. from the burner
face) but only approximately 40°C (70°F)
at position E-2 (114 in. from the face).
The equivalent temperature difference at
the higher firing rate of 1.2 million
Btu/hr are on the order of 80°C and 25°C
(150° and 50°)F), respectively. The
lower difference at the higher firing
rate is caused by radiative effects from
the hotter flame zone radiating into the
cooler postflame zone.
Qualitatively these excess air
results will have opposing effects on
destruction efficiencies. At the higher
excess air level the increased oxygen
content will promote destruction while
the decreased residence time and
temperature will suppress destruction.
In fact, there appears to be a peak
destruction efficiency between excess air
rates of 10 and 50 percent. This is
shown graphically in Figure 4.
Effect of Swirl on Temperature and ORE
From the observed temperature
profiles, the effect of increased swirl
is to increase the temperature gradient.
However, the temperature changes only
about 15°C (25°F) at the centerline.
This is comparable to the measured
temperature variance between runs (see
Table 2), and is probably not
significant. There may be a significant
variation in DRE with swirl, but the data
is difficult to interpret. A significant
difference is seen only in the long, thin
flame (swirl setting 2.0), but it is in
the opposite direction for chlorobenzene
and carbon tetrachloride.
Products of Incomplete Combustion
In order to maximize the number of
variable data points, the analytical
design for this program was simplified by
not performing GC/MS analyses. This
reduced the ability to identify and
quantitate PIC's. Because the program
was not directed toward PIC's, much raw
.data (such as the numerous
chromatographic strips) that contains PIC
information has not been reduced.
However, even with limited examination of
the data some important information is
available. For example, in every run
methylene chloride was-observed at levels
equal to or greater than the POHC of
interest, even when it was not present as
a POHC in the waste (Figure 5 displays
typical chromatograms from the analysis).
Although some of this methylene chloride
can be attributed to contamination, the
levels were consistently higher than
corresponding blank values. Peaks in the
retention time range for methylene
chloride were routinely, off scale.
Several other chlorinated organics not
fired in the fuel have been identified as
being in the emissions of one or
chlorinated organics not fired in the
fuel have been identified as being in the
emissions of one or more runs. These
include such species as •
trichloroethylene, the dichloroethylenes,
chloroform, and possibly chloromethane.
An example is given in the summary
Table 5. The significance of this
finding is the following — although
DRE's for both POHC's (chlorobenzene and
carbon tetrachloride) are greater than
99.99 percent, the DRE of total
chlorinated organics as fired is only
99.985 percent.
80
-------
2,500
2,400
2,300
2,200
2,100
2,000
• A-7- (1.2, 10, 7.5, 0)
• A-6 (V.2, 25, 7.5, 0)
AB-4 (0.8, 10, 7.5, 0)
TB-2 (0.8, 25, 7.5, 0)
1.40T
1,300
1,200
1,100
1.68
2.90 3.70
Axial Distance (m)
4.47
Figure 1. Center-line Axial Temperature Profiles for Four Firing
Conditions Without Waterwalls.
2,000 -
1,800
§• 1,400
1,200
1,000
0.8 million 1.2 million Excess
Btu/hr Btu/hr Air -
Condition I - O IV - • (10%)
Condition II - D V - • (25%)
Condition III - A VI - A (50%)
1,100
i.ono
900
800
2.29 2.90 ' 3.50
Axial Distance (meters)
4.47
Figure 2. Center-line Axial Temperature Profiles for Six Firing
Conditions with Waterwalls.
81
-------
2,500
2,400
2,300
g 2,200
3
2,100
2,000
H
£
1,900J_
Cross Section C-3
rr^
— o o Q
O Run B-5 Condition la 10% Excess air
D Run B-3 Condition lln 252 Excess air
A Run B-l Condition Ilia 50% Excess all
Cross Section E2
% II
10.16 20.32 30.48 40.64 30.48 20.32 10.16
Cross sectional distance (cm)
Figure 3. Effects of Excess Air on Temperature for Three Firing
Conditions Without Waterwalls. Normal Firing Rate is
0.8 million Btu/hr.
,-3
10'
I in"1
5 IS
10'
.-G
10
20
H1ll1on Btu/hr
(firing rate)
0.8 1.2
Chlorobenzene o O
Carbon Tetrachlorlde • •
30 40
Excess air rate
50
60
70
Figure 4. Effect of Excess Air on Fractional Breakthrough for
Carbon Tetrachloride and Chlorobenzene.
82
-------
TABLE 5. SUMMARY OF RUN C-7
•
Mass flow
inn
Mass flow
out**
Out/in
Average
ORE
Chloro-
benzene
17.34
1.48 x 10~5
8.54 x lO-7
99.9999
+0.00016tt
Carbon
tetrachloride
9.72
8.32 x 10-4
8.56 x 10-5
99.9914
+0.0035
Chloroform*
0
1.22 x 10-3
N/A
N/A
.Subtotal1"
27.05
2.06 x 10-3
7.62 x 10-5
99.9924
jKJ.004
Other* Total §
0 27.05
2 x 10-3 4 x ID'3
N/A 1.48 x lO-4
N/A 99.985
+0.020
Major identified PIC.
"" Includes POHC's plus chloroform.
* Estimate of unidentified and/or unquantified PIC's.
§ -f-stimate of total chlorinated organic species.
11 Estimate of total g/min.
Average of three samples (g/min).
"ft Two standard deviation. . ,
,-Start
^__ — 10.32 CH2BrCl
r
f
f~
Lstop
. 13.13 CHCI3
' 13.94 1.2 OCA
Standard 1,000 ng each on
blank trap
E
Start
9.96 CH2 BrCI
12.85 CHCI3
Stop
Sample run C-7B carbon
tetrachloride
Figure 5. Typical Chromatograms Demonstrating Presence of PIC's,
83
-------
PRODUCTS OF INCOMPLETE COMBUSTION FROM HAZARDOUS WASTE INCINERATORS
Andrew Trenholm
Roger Hathaway
Midwest Research Institute
Kansas City, Missouri 64110
Don Oberacker
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
ABSTRACT
The Environmental Protection Agency (EPA) is preparing a Regulatory Impact Analysis
(RIA) of hazardous waste incineration. Midwest Research Institute (MRI) provided input
data to the RIA from tests conducted at eight incineration facilities. This paper pre-
sents information gathered during these tests related to products of incomplete combus-
tion (PICs). PICs were defined as any hazardous organic constituent detected in the
stack gas but not present in the waste feed at a concentration of 100 |Jg/g or higher.
Benzene and chloroform were the compounds found most often among approximately 30 iden-
tified PICs. PIC emissions were comparable to principal organic hazardous constituents
(POHC) emissions in concentration and total mass output. Three mechanisms are proposed
to account for the presence of PICs: (a) POHCs in the waste feed below detection lim-
its, subject to low destruction and removal efficiency (ORE), and therefore detected in
the stack gas; (b) introduction into the system from some source other than the waste
feed (e.g., auxiliary fuel, in-leak air, scrubber water); and (c) actual PICs (compounds
occurring as a result of combustion reactions).
INTRODUCTION
Incineration is a popular option for
ultimate disposal of combustible hazard-
ous wastes. While incineration provides
a high degree of destruction for com-
pounds in the waste feed streams, ques-
tions have been raised about possible by-
products of combustion reactions. This
paper addresses some of these concerns
based on results of tests at full-scale
operating incinerators.
Midwest Research Institute (MRI)
performed a series of tests at operating
hazardous waste incineration facilities
and analyzed the collected samples for
products of incomplete combustion (PICs).
These, tests were performed as part of
technical support to EPA's preparation of
a regulatory impact analysis (RIA) for
hazardous waste incinerators.
Prior to this study by MRI, the ac-
cumulated data base on products of incom-
plete combustion of Appendix VIII com-
pounds, principal organic hazardous con-
stituents (POHCs), was limited primarily
to laboratory scale studies. In those
studies, certain specific compounds were
thermally destroyed in high concentra-
tions to test for possible PICs. Com-
pounds tested include kepone1 and poly-
chlorinated biphenyls.2 These studies
are illuminating for specific cases, but
the data base for PICs from combustion of
the complex matrices of organic constitu-
ents which currently are being fed to
84
-------
hazardous waste incinerators was virtur
ally nonexistent before MRI' s study.
This paper first describes briefly
the MRI study and identifies some of the
limitations of the data base. Then, re-
sults of the tests conducted are dis-
cussed in terms of three possible mecha-
nisms to explain the presence of PICs in
the stack gas, and the relative contribu-
tion of each mechanism. The conclusion
section contains some general observa-
tions . This paper is intended to be an
overview of the PIC phenomenon. It does
not provide final definition of mecha-
nisms for PIC formation.
MRI STUDY OF HAZARDOUS WASTE INCINERATION
FACILITIES
PICs were defined for this study as
any Appendix VIII compound detected in
the stack gas which was not present in
the waste feed in concentrations greater
than 100 (Jg/8- Tne study included mea-
surement of PICs at eight full-scale op-
erating incinerators with a variety of
key design and operating parameters and
waste feed characteristics. Each incin-
erator had a liquid injection burner, and
some facilities also included a rotary
kiln or hearth. Three incinerators had
no air pollution control devices. The
remaining five had wet.scrubbers for HC1
control, and four of these had other par-
ticulate control devices.
Combustion chamber designs were var-
ied. Several facilities had multiple
combustion chambers. The incinerators
ranged in waste feed capacity from 50 to
6,000 kg/hr (110 to 13,000 Ib/hr), with
corresponding heat inputs from 1 to 78
GJ/hr (1 to 74 million Btu/hr). Operat-
ing conditions for three key parameters
exhibited wide ranges: combustion tem-
perature, 820° to 1450°C (1400° to
2650°F); residence times, 0.07 to 6.5
sec; and percent excess air, 60 to 130%.
The facilities included some exam-
ples of incineration of specific process
wastes generated onsite as well as varied
mixtures of wastes disposed offsite. The
wastes fired at different facilities var-
ied also with respect to waste type (liq-
uid, solid, and gas), heating value,
chloride concentration, and ash and
moisture content. The organic liquids
and many of the solids had a heating
value high enough to sustain combustion
without auxiliary fuel. Percentages of
chloride ranged up to 25%, the highest
values occurring for organic liquids.
Ash values ranged up to 9% for organic
liquids and 29% for solids. Water con-
tent ranged up to 50% for organic liquid
and solid wastes, and to nearly 100% for
aqueous wastes.
Samples were collected of all inflow
and outflow streams at the incineration
facilities. All samples were analyzed
for Appendix VIII constituents by gas
chromatography/mass spectrometry. Lower
detection limits for compounds in the
waste feed ranged from 1 |Jg/g up to 100
pg/g, depending on the variety and com-
plexity of the sample matrix. The sam-
pling and analysis procedures used were
capable of quantifying a wide variety of
compounds present in the stack gas with
lower detection limits of 1-2 ng/L.
Certain compounds which are sus-
pected of being PICs were not amenable to
measurement by the methods used in these
tests, e.g., formaldehyde and phosgene.
Also, at seven of the eight facilities,
carbon tetrachloride and trichloroethyl-
ene were present in the waste feed, lim-
iting their identification as common
PICs.
PRESENCE OF PICs IN THE STACK GAS—AN
OVERVIEW
Twenty-nine compounds were classi-
fied as PICs from the eight incinerator
tests. These compounds are shown in
Table 1 along with the number of sites
where they were detected as PICs and the
range of concentrations measured in the
stack effluents. Figure 1 shows an al-
ternate display of the data categorized
as volatiles, semivolatiles, and trihalo-
methanes (THMs). The total concentra-
tions of all PICs, and of the three cate-
gories of PICs are shown for each plant.
THMs are not included in the totals be-
cause they may have originated in the
liquid influent to the control systems at
some plants.
Total stack output of Appendix VIII
compounds for the eight plants is shown
85
-------
TABLE 1. PICs FOUND IN STACK EFFLUENTS
PIC
Benzene
Chloroform
Bromodichlorome thane
Dibromochloromethane
Naphthalene
Bromoform
Chlo r ob enzene
Tetrachloroethylene
1,1, 1-Trichloroethane
Toluene
o-Nitrophenol
Methylene chloride
Phenol
2,4, 6-Trichlorophenol
Carbon disulfide
o-Chlorophenol
2,4 Dimethyl phenol
Methylene bromide
Bromo chlo r omethane
Trichlorobenzene
Hexachlorobenzene
Diethyl phthalate
Pentachlorophenol
Dichlorobenzene
Chloromethane
Methyl ethyl ketone
Bromomethane
Pyrene
Fluoranthene
Number
of sites
6
5
4
4
3
3
3
3
3
2
2
2
2
1
1
1
1
1
1
.1
1
1
1
1
1
1
1
1
1
Concentrations
(ng/L)
12-670
1-1,330
3-32
1-12
5-100
0.2-24
1-10
0.1-2.5
0.1-1.5
2-75
25-50
2-27
4-22
110
32
2-22
1-21
18
14
7
7
7
6
2-4
3
3
1
1
1
86
-------
11 *
Fll 1 —g ^ r
£11" v\ >
\\
51
CN
O
to aU9|Dl|4l|dD|s] . ».
(—
O
•of | [OUeildOJOIlpIJl-p'^Z
SI
*
o
1
* <»l
o
SI |u«S
§1
1 1 1. 1 1 1 1 1 1 1 1 1 1 1 1 1 1
s
to
>
H-
X
2;
s
r—
t/)
>
S
>
S
5
X
LO
S
to
S
3 4 5 678
V\ = Trihalomethanes - chloroform,
romethane (not included in other totals).
1 2
evice used.
, V = Volatile, SV = Semivolatile, TH/
n, bromodichloromethane, dibromochio
1-8 identify the plants.
— ° ° m
2 .S'S J
! 1 "1 §
* U h- -D Z
* * *
* *
*
to
B
0
•ri
4J
rt
4J
fi
0)
O
fi
0
O
O
M
t-)
0
M
CO
H
01
INCMO^tNCM — — — ^Sl
87
-------
as a comparison between PICs and POHCs in
Figure 2. In general, PIC concentrations
are slightly higher than POHC concentra-
tions, although this ratio varies widely
from plant to plant. The PIC output was
also compared to the total POHC input in
the waste feed. As Table 2 shows, only
one of the eight plants exceeded 0.01% of
POHC input, leading to the conclusion
that the PIC output rate only infre-
quently exceeds 0.01% of the POHC input
rate.
Three possible mechanisms to explain
the presence of compounds in the stack
gas classified as PICs were assessed.
These mechanisms and the relative contri-
bution of each are discussed in the fol-
lowing sections.
Compounds Originally Present in the Waste
Feed Below 100 |Jg/g with low DRE
For this study, a compound present
in the waste feed below 100 (Jg/g but de-
tected in the stack would be classified
as a PIC. The approximate stack concen-
tration for a compound present just below
100 |Jg/g, if it were subject to a DRE of
99.9%, would be 10 ng/L. Since this con-
centration is substantially above the de-
tection limit of the stack sampling
method, the compound is identified as a
PIC. Data gathered on the program con-
firmed that this situation might be rea-
sonably expected.
In certain cases, the waste feed
samples were relatively "clean" and de-
tection limits were on the order of
1 M8/8- These waste feed samples fre-
quently contained compounds at concentra-
tions below 100 |Jg/g. When DREs were
calculated for these compounds, they gen-
erally fell below 99.9%. Although this
mechanism would not account for a large
percentage of the total mass output of
Appendix VIII compounds, it could explain
the presence of several of the PIC com-
pounds detected in the stack effluents.
Compounds Introduced from Some Source
Other Than Waste Feed
PIC compounds may also be introduced
to the incineration process by streams
other than the waste feed. Three sources
are: (1) scrubbing liquid used in con-
trol devices; (2) inleak of ambient air
through spaces in ductwork; and (3) aux-
iliary fuel streams. Instances of all
three possible sources of introduction to
the incineration process occurred during
the course of the testing program.
Municipal water is often used as
makeup to scrubbers. Appendix VIII com-
pounds commonly are found in measurable
concentrations in municipal water sup-
plies. These compounds are primarily the
THMs mentioned earlier, and they are in-
troduced into the water via chlorination.
The THMs are in low concentrations in the
water inflow streams, but can be stripped
out of the water with high efficiency by
the hot stack gases. In a number of
cases the scrubber water effluents con-
tained substantially less of these or-
ganic compounds than did the makeup wa-
ter. This phenomenon could account
for the presence of THMs which appear as
PICs in the stack effluent of plants with
wet control devices using chlorinated
makeup water. In addition to THMs con-
tributed by treated water supplies, other
trace organics can sometimes be found in
well water or lagoon water used as makeup
to scrubbers.
Typically, a plant exhibited a
scrubber makeup flow rate of 70 L/min (18
gpm) and a stack flow rate of 100 Nm3/min
(3,500 dscfm). THMs were measured in
makeup water at concentrations of 100
|jg/L for chloroform and 10 (JgA each for
bromodichloromethane, dibromochlorometh-
ane, and bromoform. The scrubber water
was stripped clean of all THMs (the ef-
fluent contained < 1 Mg/L of all com~
pounds). This stripping resulted in-a
contribution to the stack effluent of
0.007 g/min (70 ng/L) for chloroform and
0.0007 g/min (7 ng/L) for each of the
other THMs. These concentrations were
easily detectable by the sampling methods
used. In one case, well water used for a
quench makeup was found to contain four
Appendix VIII compounds in quantities
greater than 100 |Jg/L.
A second possible source of Appendix
VIII compounds in the stack effluent is
ambient air that may enter the negative
pressure system through spaces in duct-
work or other openings downstream from
the combustion chamber. Heat balance
calculations performed upon the eight in-
cinerators indicated that from 0 to 50%
88
-------
J
p.
H
PH
O
4J
O
O
4J
3
O,
4J
3
O
I
§
8 8
UOJ4DJ4U33UO3
89
-------
TABLE 2. CALCULATED VALUES FOR PIC EMISSIONS AS A
PERCENTAGE OF POHC INPUT
Calculated values for PIC emissions as a percentage of POHC
input:
01 - PIC output (g/min)
'" POHC input (g/min) X
Plant PIC output (g/min)
POHC input (g/min)
1
2
3
4
5
6
7
8
0.090
0.0046
0.092
0.00062
0.011
0.00090
0.26
0.0076
9,800
100
780
27
1,200
310
8,300
1,200
0.00092
0.0044
0.012
0.00023
0.00091
0.00039
0.0031
0.00061
90
-------
of the measured stack flow rate might
have entered the system downstream from
the combustion chamber. It is not un-
usual for ambient air to contain higher
concentrations of some Appendix VIII com-
pounds than the concentrations measured
in the incinerator stack.3 The ambient
air concentrations may be higher in areas
of a facility where hazardous wastes are
handled.
Ambient air samples collected at one
facility contained measurable concentra-
tions (between 2 and 30 ng/L) of eight
Appendix VIII compounds, as shown in
Table 3. The inleak rate at this plant
was calculated to be 25% of the total
stack flow. This inleak rate was used to
calculate the figures shown for contribu-
tion to stack gas concentrations. All of
these compounds were detected at low lev-
els in the stack, so this source is not
suspected of contributing a large mass
flow rate of compounds; but it still can
result in the identification of addi-
tional compounds as PICs.
A third potential source of Appendix
VIII compounds is the auxiliary fuel
stream, which is not usually classified
as a waste feed. Uncombusted fractions
of hazardous compounds may be detected in
the stack gas, and identified as PICs.
One of the facilities tested fired
auxiliary fuel oil along with liquid and
solid waste. A sample of the fuel oil
was found to contain eight Appendix VIII
compounds including five at concentra-
tions exceeding 100 Hg/g, as shown in
Table 4. All of these compounds had been
identified as POHCs in other waste feeds
at this facility, but the auxiliary fuel
was the primary contributor of two of the
compounds. While this particular example
does not influence the PIC emissions of
the facility, had the same fuel oil been
used at a facility where the compounds
were not listed as POHCs in the waste
feed, the undestroyed fraction would be
detected in the stack as a PIC.
Compounds Occurring as Actual Combustion
By-Products
The third mechanism to explain the
presence of compounds identified as PICs
is their formation as actual combustion
by-products. Due to the extremely
complex nature of waste feed and stack
effluent, it is impossible to arrive at
specific reactions or possible kinetics
based simply on the results of this
study. However, a review of the informa-
tion collected is.useful. The discussion
which follows is presented in three
parts, according to the category of com-
bustion by-product.
Identifiable fragments of feed
constituents which result from partial
oxidation or simple substitution reac-
tions .
An example in this category would be
studies reporting the appearance of hexa-
chlorobenzene as a by-product of PCB com-
bustion.2 Several of the plants MRI
studied exhibited PICs which were unique
to the type of waste being burned and
were identifiable products of reactions
of waste constituents. Two facilities
which burned large quantities of aniline
both exhibited high concentrations of
nitrophenols in the stack. Only one
plant showed measurable concentrations of
polychlorinated benzenes in the waste,
and this plant exhibited di-, tri-, and
hexachlorobenzene in the stack. Like-
wise, chlorophenols were detected in the
waste at only the one plant where ortho-,
tri-, and pentachlorophenol were detected
in the stack. These cases suggest that
the compounds identified as PICs were
formed as a result of specific constitu-
ents in the waste feed.
Reaction products which are the
re'sult of complex recombination or substi-
tution reactions.
This category typically includes
compounds of high molecular weight. High
molecular weight compounds identified as'
PICs which might have resulted from com-
plex reactions included naphthalene,
fluoranthene, and pyrene.
Simple fragments which cannot
be readily traced to any' single reactant,
but rather seem to be universal by-
products of the combustion of organic
compounds.
This category, which includes many
low molecular weight compounds, was the
most broadly represented among the plants
studied. Each plant exhibited one or
91
-------
TABLE 3. APPENDIX VIII COMPOUNDS FROM AMBIENT AIR
Compound
Benzene
Chlorobenzene
Toluene
Chloroform
1,1, 1-Trichloroethane
Tetrachloroethylene
Trichloroethylene
Carbon tetrachloride
PIC
PIC
POHC
POHC
POHC
POHC
POHC
POHC
Concentration
in stack gas
(ng/L)
130
3.9
14
20
0.3
4.2
2.5
8.4
Potential contribution
to stack gas concentration
(%)
4
11
50
4
100
17
21
5
TABLE 4. APPENDIX VIII COMPOUNDS FOUND
IN AUXILIARY FUEL OIL
SAMPLE
Compound
Concentration
(M8/8)
Toluene
Methylene chloride
Tetrachloroethylene
1,1,1-Trichloroethane
Methylene bromide
Benzene
Chloroform
Trichloroethylene
1,500
440
380
220
210
91
23
20
92
-------
more fragments which might have resulted
from the combustion of organic compounds.
These included chlorinated methanes
(chloromethane, methylene chloride,
chloroform, and carbon tetrachloride),
and chlorinated ethanes and ethylenes
(1,1,1,-trichloroethane, trichloroethyl-
ene and tetrachloroethylene). In addi-
tion, bromoform was detected in high con-
centrations at one facility which was
burning brominated wastes. Other frag-
ments which were found as PICs could have
resulted from the combustion of aromat-
ics, and included benzene, phenol, tolu-
ene, and chlorobenzene.
Relative Contribution of Different PIC
Mechanisms
Figure 3 summarizes, for each of the
three PIC formation mechanisms discussed,
the relative potential contribution of
each to the total mass output of PICs.
Low concentration POHCs in the waste feed
potentially accounted for 0 to 68% of the
total mass of PICs detected at each fa-
cility. The higher percentages were for
the plants with the lowest total mass of
PIC emissions. The stripping of THMs •
from the scrubber waters potentially ac-
counted for up to 14% of the total mass
of PICs detected. The rest of the com-
pounds were most likely actual products
of combustion reactions. Figure 3 also
includes a bar which represents the per-
centage of the PICs which occurred most
frequently and in the highest concentra-
tions. The six compounds in this cate-
gory accounted for a substantial per-
centage of the total mass of PIC output
at most of the facilities tested. Ta-
ble 5 lists these compounds, the concen-
tration range measured, and how often
they were detected.
CONCLUSIONS
This program did not initially focus
on identification of PIC mechanisms;
thus, definite limits were imposed on in-
terpretation of the PIC data. The dis-
cussion regarding appearance of PICs in
the stack is fairly well documented for
the case of the compounds stripped from
the scrubber water, but otherwise the
discussion of PIC formation mechanisms
is somewhat conjectural. Possible
TABLE 5. SIX MOST COMMON PICs
Compound
Concentration
range (|Jg/L)
No. of
plants '
as a PIC
Chloroform
Benzene
Naphthalene
o-Nitrophenol
Bromoform
Phenol
1.5
13
4
25
0.21
5
- 1,330
670
120
35
23
19
mechanisms for formation of reaction by-
products, in particular, would require
more information before being confirmed
as explanations. More thorough investi-
gation is also needed for the non-POHC
fraction of the waste feed which, for the
most part, remained unidentified in this
study. Since the non-POHC fraction typi-
cally constitutes 30 to 90% of the waste
feed, by-products from non-POHC combus-
tion might prove very important in ex-
plaining the PIC phenomenon!.
In summary, general observations
from the study are:
Stack gas concentrations of
PICs were typically as high as
or higher than those for POHC
compounds.
PIC output rate infrequently
exceeded 0.01% of POHC input
rate.
The three likely mechanisms
that explain the presence of
most PICs are:
Low DREs for Appendix VIII
compounds present at low
concentrations (< 100
M8/g) in the waste feed;
Input of Appendix VIII
compounds to the system
from sources other than
waste feed; and
93
-------
U
O
oo
00
Os
CM
col
00
U
to
U
to
I I
I I I I I
O
>o
.000000
si
§[
=[=
N.
CO
col
o-l
CN
ON
£>[
R
ol
£|
"1
"I
I 1 1 1 1 1 1 1 1 1 1
U
to
Q.
^
U
oo
VI
•*-
n
a. ^
D-
(U
"c
0)
CO
52
_Q
U E
to £
V)
a. 1
D
O
Q.
^ °
•^ u
to
u- II
U
to
05
U Ou
to „
o.
o. C
^ II
T3
1 1 (y
•£
u
u_
Figure 3. Relative Contribution of Different PIC Mechanisms
o ooo ooo oooo
o ON co t'N. o «•> •*}• co CM •—
4U3DJ3J
94
-------
Actual products of incom-
plete combustion, or prod-
ucts of complex side reac-
tions .
Benzene is the most likely com-
pound to appear as a PIC, based
on both frequency of occurrence
and presence at higher concen-
trations .
Measurement of Appendix VIII
compounds at low concentrations
in the waste feed, auxiliary
fuel, and inflow streams to
control systems is often neces-
sary to explain the presence of
PICs.
REFERENCES
Duvall, D. S. and Rubey, W. A.,
Laboratory Evaluation of High Tem-
perature Destruction of Kepone and
Related Pesticides. Cincinnati, OH;
U.S. Environmental Protection
Agency; 1976 December, EPA-600/
2-76-299.
Ahling, B. and Lindskog, A., Thermal
Destruction of PCB and Hexachloro-
benzene, The Science of the Total .
Environment, Vol. 10 (1978), 51-59.
Singh, H. B. , et al., "Distribution
of Selected Gaseous Organic Mutagens
and Suspect Carcinogens in Ambient
Air," Environmental Science and
Technology, Vol. 16 (1982), pp. 872-
880.
95
-------
PRELIMINARY ASSESSMENT OF COSTS AND CREDITS
FOR HAZARDOUS WASTE CO-FIRING IN INDUSTRIAL BOILERS
R. McCormick and L. Weitzman
Acurex Corporation
Cincinnati, OH
ABSTRACT
This paper outlines the results of an IRB-sponsored study to develop preliminary
cost vs. credit information for hazardous waste co-firing in industrial boilers. An
overview of the cost/credit estimating methodology is presented, along with pertinent
background information on hazardous waste co-firing practices. Details of all cost
estimates are provided in the project report, Reference 1.
BACKGROUND
The practice of burning hazardous
wastes in industrial boilers has become
increasingly popular over the past decade.
The reasons for this trend are twofold:
(1) Increased prices for conventional
fuels, favoring waste substitution or
co-firing as a means of reducing fuel
expenditures; and
(2) Stricter environmental regula-
tions and higher costs for hazardous
waste disposal by conventional
methods, promoting waste co-firing as
a means to eliminate or reduce waste
disposal costs.
Two regulatory actions under the
Resource Conservation and Recovery Act
(RCRA) in particular have favored boiler
co-firing as a hazardous waste disposal
alternative. The first such action was the
virtual ban on liquid waste landfill ing
which promoted thermal destruction (in
Incinerators, boilers, cement kilns, etc.)
as the favored disposal method for hazard-
ous organic liquids. The second regula-
tory action favoring boiler co-firing was
the promulgation of emissions limitations
for incinerators burning hazardous wastes.
Since 1980, hazardous waste incinerators
have been subject to Destruction and
Removal Efficiency (ORE) requirements for
hazardous organic constituents of the
waste, particulate emission limitations,
and HC1 removal requirements. No such
regulations have been imposed on boilers
burning hazardous wastes.
This regulatory dichotomy has come
under increasing attack over the last four
years. From the standpoint of environ-
mental safety and health, there is no
conceptual difference between hazardous
waste combustion in incinerators and
boilers (although actual emission char-
acteristics may vary). Boilers were
originally exempted from'regulation
because the? implied objective of waste
.co-firing was energy recovery, rather than
waste disposal. However, this distinction
has been clouded by the recent trend
toward waste heat recovery in incineration
facilities.
In order to assess the need for and
impacts of regulations governing hazardous
waste co-firing in boilers, EPA is cur-
rently conducting a Regulatory Impact
Analysis (RIA) study. One of the major
elements of this program is an economic
assessment of the impacts of boiler
96
-------
performance standards and associated air
pollution control requirements. The study
addressed herein is intended to provide
preliminary cost/ credit information for
current boiler co-firing practices. As
such, it serves as a precursor to the eco-
nomic impact assessment phase of the RIA.
OBJECTIVES AND SCOPE
The underlying objective of the study
is to provide a preliminary evaluation of
the costs and credits associated with
hazardous waste co-firing in industrial
boilers. Specific objectives are as
follows:
(1) Identify the major waste/boiler
co-firing scenarios that need to
be addressed for the purposes of
the RIA.
(2) Identify the major costs and credits
associated with boiler conversion to
waste co-firing.
(3) Develop preliminary cost data
for the major and most probable
boiler retrofit activities.
(4) Develop a parametric approach to
estimating retrofit costs, incre-
mental operation and maintenance
(O&M) costs, and credits so that
cost/credit tradeoffs can be pro-
jected as a function of waste type,
waste:fuel co-firing ratio, boiler
design and capacity, and potential
air pollution control requirements.
Due to the preliminary nature of the
study, none of the objectives are ad-
dressed in a completely thorough or
rigorous manner. The goal is simply to
provide a basis for initial decision-mak-
ing and more detailed future study.
CO-FIRING SCENARIOS EVALUATED
The range of hazardous waste co-fir-
ing scenarios addressed in the study is
necessarily limited in terms of waste
type, boiler design characteristics,
capacity, and waste:fuel co-firing ratios.
The evaluation is limited to those scenar-
ios which best represent current practice
and probable future practice.
Waste Characteristics
As a starting point, the evaluation
is limited to hazardous organic liquid
waste co-firing. This limitation is
imposed because organic liquids are the
prime candidates for boiler co-firing in
terms of volume, handling characteristics,
primary fuel firing compatibility, and
compatibility with standard industrial
boiler designs.
Second, the economic evaluation is
based on in-house generation of the
organic liquid wastes being co-fired. The
assumption here is that industrial boiler
operators do not routinely accept wastes
from outside sources, thus functioning as
commercial waste disposers.
The third major assumption is that
the majority of wastes co-fired in indus-
trial boilers possess desirable fuel
properties. Heating values are assumed to
be greater than or equal to 8,000-10,000
Btu/lb, such that the waste will support
combustion. Water, ash, and halogen
contents are also assumed to be reasonably
low, so as to preclude flame quenching,
excessive ash buildup on the boiler tubes,
excessive bottom ash handling problems,
and/or cold-end corrosion problems.
Finally, reasonable mid-range waste:
primary fuel co-firing ratios are con-
sidered. The range is 10-50% waste with
primary fuel, with the percentage based on
gross heat input.
Boiler Designs and Capacities
The basic boiler designs considered
in the study are those characteristic of
the industrial size range (up to 250,000
Ib/hr steam) originally designed to burn
natural gas, distillate oil, residual oil,
or a combination of these fuels. The
three most common design types, addressed
in the study, are as follows:
1. Scotch firetube, N-pass design,
packaged boilers for natural gas,
distillate oil, or combination
gas/distillate oil firing. Single
burner design, with no economizer or
air heater. Typical capacities are
10,000 to 30,000 Ib/hr steam, up to
50,000 Ib/hr steam. Saturated steam
up to 150 psig.
97
-------
2. Watertube design for natural gas,
distillate oil, gas/distillate,
or gas/distillate/residual oil
firing (with oil preheat equip-
ment). Single or multiple burner
design. Steam capacities of 20,000 to
100,000 Ib/hr saturated at 125 to 250
psig design rating. Economizers not
atypical at more than 50,000 Ib/hr
steam, but air heaters rare.
3. Watertube design with multiple
burners for gas/oil firing. Steam
capacities of 100,000 to 250,000 Ib/hr
and up, with turbogenerator steam
pressures and superheat in the larger
size ranges. Economizers typical, and
possibly air heaters.
These boiler designs are most common
(and compatible) for liquid waste co-fir-
ing, although a limited number of coal-
fired boilers are also being used to
co-fire hazardous liquids. Both stoker
and pulverized coal-fired boiler conver-
sion for waste co-firing are addressed to
a minor extent.
CAPITAL REQUIREMENTS FOR BOILER SYSTEM
RETROFIT
As indicated in the preceding sec-
tion, one of the baseline assumptions of
this study is that boilers co-firing
hazardous wastes were originally designed
to burn one or more conventional fossil
fuels — natural gas, oil, or coal.
Therefore, an initial capital investment
is required to retrofit a boiler for waste
co-firing. This capital investment can be
divided into three components:
(1) Addition of waste storage and feeding
equipment,
(2) Boiler modification to accommodate
the waste fuel, and
(3) Air pollution control device (APCD)
addition for parti oil ate/HC1 removal,
if required by regulation.
The total investment for retrofit is
usually modest, unless air pollution
control device addition is mandated.
Costs are summarized below; details may be
found in Reference 1.
Waste Handling and Feeding Equipment
Liquid wastes are usually transported
from the process area to the boiler
location by tank truck and pumped into a
settling tank, or holding tank. Typical
installed costs for individual 5,000,
10,000, and 20,000 gal tanks designed for
hazardous liquid waste storage are approx-
imately $34,000, $46,000 and $61,000
(December 1982), respectively.
From the storage tank, wastes are
usually pumped directly to the boiler
through self-cleaning filters or dual
strainers to remove any suspended solids.
Dual feed pumps, costing $1,000 to $1,500
each, are desirable to provide redundancy
in the event of equipment failure. Piping
will cost $2,000 to $5,000. Costs for
in-line filters can range from several
hundred to several thousand dollars,
depending on the flowrate, pressure,
solids loading, and solids size distribu-
tion.
Boiler Retrofit
In any co-firing situations, retrofit
at the boiler is a relatively minor
consideration. High costs for boiler
retrofit are usually incurred only when a
completely new burner assembly, or burner
assembly plus blower, damper, and control
system is required. This may be the case
when stoker or pulverized coal-boilers are
retrofitted to burn liquid waste, or even
when gas-oil fired boilers are converted
over for waste co-firing.
Equipment cost vs. capacity data for
three potential retrofit requirements are
presented in Figure 1. The lower curve
applies to one of the simplest retrofit
cases: liquid waste nozzle addition/re-
placement. The costs shown in Figure 1
are the nozzles only; total installed
costs including pressure and flow control
valves, plus labor, can easily run two to
three times the costs for the nozzles
alone.
-------
Figure 1.
60 r
Burner Component and
System Costs
• Liquid waste nozzle
A Complete burner assembly
Burner assembly plus
blower and controls
100
200 300 400
GPH CAPACITY
500
600
The middle curve is for complete
liquid fuel burner assemblies with natural
gas firing capability built-in. The costs
include pilot, spark ignition assembly,
and damper assembly. The upper curve is
for complete burner, systems, including
forced-draft blowers and master controls
for flame safeguard, fuel modulation, air
pressure and flow regulation.
Installation costs for burners and
complete burner systems are typically 50%
of the equipment cost. If structural
modifications are required to install the
burner, this installation cost can easily
double.
Air Pollution Control Device Addition
Industrial boilers designed for
natural gas or fuel oil firing are gener-
ally no.t equipped with any APCD's, includ-
ing boilers co-firing hazardous liquid
wastes. If Federal standards similar to
the RCRA performance standards for incin-
erators were to be promulgated for boilers
co-firing hazardous waste, it is likely
that at least some boiler facilities would
need to provide one or more APCD's for.
particulate and/or HC1 removal. Obviously,
this would depend on the ash content of
the waste, chlorine content of the waste,
and the regulatory requirements imposed.
It is impossible to predict industry
trends relative to APCD addition at this
point, should,performance standards for
particulate and/or HC1 be promulgated.
However, six scenarios can be envisioned:
(1) The waste being co-fired at the site
in question contains negligible ash
and chlorine, so APCD addition is
unnecessary to comply with regulatory
requirements.
(2) Chlorinated wastes are not being
burned at the site, but particulate
emissions exceed the standard as a
result of waste co-firing. The
mandated particulate removal effi-
ciency is in the 95 to 99% range. In
this case, a "cold-side," dry elec-
trostatic precipitator (ESP) is the
likely choice for particulate control.
(3) This scenario is identical to the
previous one, except that the man-
dated particulate removal efficiency
is more than 99.5%. In this case, a
baghouse would probably be more
economical than an ESP for particu-
late control due to the excessive
plate area required for an ESP to
achieve this efficiency. This is a
generalization, however, which could
be reversed by site-specific factors.
(4) The co-fired wastes do not contain
significant ash, but chlorine is
present in quantities such that the
waste/fuel mix contains more than
0.5% organically-bound chlorine. In
this case, a packed bed absorption
system may be required for HC1
removal. This is not considered
a typical waste co-firing scenario,
due to the adverse boiler corrosion
impacts associated with burning
chlorinated materials.
(5) The co-fired wastes contain signifi-
cant ash and chlorine, such that
particulate and HC1 emission stand-
ards would both be violated. In this
case, a venturi scrubber/packed bed
absorption system would be the
likely choice for air pollution
control. Again, this is not
considered a typical waste co-firing
scenario for the reason stated
above.
(6) Either particulate or HC1 emission
standards are exceeded, but the
99
-------
boiler facility owner/operator
chooses to eliminate waste co-firing
rather than install the necessary air
pollution control equipment.
These scenarios apply specifically to
natural gas/fuel oil-fired boilers, but
also to coal stoker-fired boilers in prin-
ciple because stokers are usually equipped
with no more than cyclonic separators for
particulate control.
Electrostatic Precipitators
Electrostatic precipitators are one
of the most common particulate control
devices for large-scale utility boilers,
and are well suited for industrial boiler
retrofit to control particulate emissions
unless (a) the gas is particularly wet, or
(b) the gas contains significant HC1 such
that corrosion is a major potential prob-
lem. ESP advantages are high particul ate
collection efficiency at neglibile pres-
sure drop.
Basic ESP costs are determined by two
primary factors: insulation requirements
and collection plate area requirements.
Insulation requirements are a function of
the cold-end temperature and gas dew
point, i.e., condensation is undesirable
due to the associated corrosion problem.
For standard, uninsulated ESP's, cost is
related to plate area by
P * 129,400 + 4.42ft (1)
where P = purchased cost in June 1983
dollars
A = collection plate area, ft2
If insulation is required,
P * 201,000 + 6.62A (2)
The basic equation for plate area
is
A * Q In (1- n)/w
(3)
where Q * gas flow rate, cfs
n * fractional collection efficiency
w = drift velocity ft/s
For the purpose of this study, a drift
velocity range of 0.1-0.3 ft/s is assumed.
The total installed cost for a
basic ESP system is roughly double the
purchased equipment cost.
Baghouses
Baghouses, or fabric filters, are
also viable candidates for retrofit
control of particulate emissions from
boilers co-firing hazardous wastes,
especially if extremely efficient par-
ticul ate removal (^99.5%) is needed.
However, baghouses are subject to the same
condensation/corrosion problems that can
plague ESP operation; thus, the same
limitations on gas moisture content and
acidic constituents apply.
The purchased cost for a fabric
filter system is a complex function of gas
temperature, composition, particulate .
loading and size distribution, and system
design features. Some of the major
factors impacting costs are as follows:
(1) Type of system used:
- intermittent pressure, mechanical
shaker cleaning
- continuous pressure, pulse jet
cleaning
- continuous pressure, mechanical
shaker cleaning
- continuous pressure, reverse air
cleaning
(2) Type of bags (filter medium) used.
The nominal operating temperatures of
various fabrics are:
Cotton
Ifylon
Acrylic, polyester
Nomex
Tefl on
Fiberglass
180°F
200° F
275°F
425°F
400°F
550°F
(3) The need for insulation to prevent
condensation
(4) The cloth area required (function of
capacity and air-to-cloth ratio)
(5) Whether the system is custom built or
standard.
100
-------
The approach used to estimate a
purchased equipment cost for a baghouse
system, incorporating the design consid-
erations listed above, does not lend
itself to the simplification needed for
this paper. Therefore, the reader should
refer to Reference 1 for a procedure to
estimate purchased costs for fabric filter
systems. Installation costs may then be
estimated as 95-100% of the purchased
equipment cost.
Venturi Scrubber/Acid Gas Absorber Systems
If the gas is extremely wet or
contains significant HC1 as a result of
co-firing chlorinated wastes, a venturi
scrubber/acid gas absorber system would be
the probable choice for particulate and/or
HC1 removal. Such a system would likely
include a small nickel alloy quench to
reduce the temperature of the boiler exit
gas to saturation, a high nickel alloy
venturi scrubber, a cyclonic separator and
integral packed tower absorber of fiber-
glass construction, polypropylene tower
packing, a caustic recycle system, a
corrosion-resistant fan, and an acid-
resistant stack. This type of system
would be preferred in cases where wastes
containing significant quantities of ash
and/or chlorine are to be co-fired.
Total equipment cost, C, for this
generic system is approximately given by
the relationship derived in Reference 2:
log C = 0.81 log Q + 1.7
where Q is the inlet gas flowrate, ranging
from 1,000 to 50,000 acfm. These costs •
are based on the assumption of a 30-in WC
venturi pressure drop, which would be
typical for hazardous liquid waste co-fir-
ing. If no venturi scrubber is needed,
(i.e., little or no particulate control
problem), the total equipment cost for the
system is reduced by approximately 15%.
Installation costs for this type of
scrubbing system are typically about 50%
of the equipment cost.
Indirect Costs
Indirect engineering and construction
costs will be associated with any major
retrofit activity such as burner system
replacement/addition or APCD addition.
These indirect costs are usually estimated
as percentages of the total direct cost
for equipment purchase and installation,
as shown in Table 1.
TABLE 1. INDIRECT COST MULTIPLIERS
Indirect Cost
Percentage of Direct Cost
Engineering and supervision
Construction and field expenses
Construction Fee
Startup
Contingency
10 to 20a
10 to 20b
7.5 to 10
1 to 3
10
TOTAL
38.5 to 63
a 20% applies to ESP addition
b 20% applies to ESP or baghouse addition
101
-------
These percentages are somewhat higher
than would be expected for new facility
construction because they apply to retro-
fit situations.
INCREMENTAL O&M COSTS
Incremental O&M costs for boiler
operation are likely whenever waste,
rather than fuel, is burned. These incre-
mental costs can include increased con-
sumption of electric power and water,
operating labor, higher costs for residue
disposal and maintenance, add-on costs
for scrubbing chemicals, liquid nitrogen,
and waste analysis, capital recovery
charges, and taxes and insurance. Residue
disposal and maintenance may be particu-
larly significant and are discussed below.*
The other O&M costs are discussed in
Reference 1.
Ash Disposal
Co-firing hazardous waste with
natural gas or fuel oil will in all
likelihood increase the quantities of ash
deposited on the furnace floor and the
hopper fallout. Addition of baghouses or
ESP's will also increase the quantities of
ash that must be disposed. However, the
major potential cost impact of hazardous
waste co-firing is in ash characteristics,
rather than the quantities of ash gen-
erated .
Under current RCRA regulations for
incinerators and other treatment pro-
cesses, any residue (e.g., ash) from the
processing of a hazardous waste is, by
definition, a hazardous waste as well. If
this criterion was expanded to include ash
from boilers co-firing hazardous waste,
the cost impact could be significant
(depending, of course, on the quantity of
ash handled).
Conventional landfill ing costs in
major industrial areas throughout the U.S.
range from about 0.1 to l.Ojd/lb for dry
*This assumes that there are no air
pollution controls added. If controls are
added, then capital recovery costs are
likely to be significant, and if the
control is a scrubber, then the cost of
caustics is important to consider.
solid waste. Secure landfilling costs in
the same areas range from 1.2 to 4.8£/lb
for the same type of waste. Typically,
the secure landfilling costs are an order
of magnitude higher. Total annual cost
for ash disposal can be estimated using
the general relationship:
Annual cost, $ =
(Ash content of waste, lb/1b)x
(Annual waste throughput, lb)x
(Landfilling cost, $/lb) (4)
Maintenance
Maintenance is potentially the most
significant annual cost associated
with co-firing hazardous wastes in boilers
and also the most difficult to predict.
Within the boiler itself, severe convec-
tive pass fouling and acid-gas corrosion
are the two major concerns. The major
conclusion pertinent to this study is that
incremental boiler maintenance due to
waste co-firing can range from virtually
nil to the other extreme of catastrophic
tube bundle failure. This type of corro-
sion-related problem is most likely when
chlorinated, other halogenated, or phos-
phorous-bearing materials are burned and
economizers or air heaters are incorpor-
ated in the boiler design. Likewise,
fouling problems are most likely to occur
when high ash content and/or alkali metal
contaminated waste are burned in boilers
with narrow, staggered tube spacings in
the convective passes.
A considerably larger data base is
available on APCD maintenance costs.
Assuming that the proper type of equipment
is specified for a given,application, and
that suitable materials of construction
are used, the annual maintenance cost for
ESP's and wet scrubbing systems is typic-
ally 5% of the .equipment cost. A 5%
multiplier can also be used to project
hardware maintenance for baghouse systems,
but the cost for bag replacement must
also be considered.
CREDITS
The credits associated with co-firing
hazardous wastes include the obvious fuel
102
-------
savings, plus elimination of certain
on-site and/or contractor costs for
dispoal of the waste.
Fuel Savings
Unless the full capacity of the
boiler is not needed to satisfy plant
steam requirements, one of the basic goals
of waste co-firing is to maintain the same
stream flow and quality. Therefore, the
net fuel savings as a result of waste
co-firing can be estimated by
Of = Qw - Qs (1/H2 - 1/ni) ,(5)
where Qf = Net fuel savings, Btu/hr
Qw = Heat input with waste, Btu/hr
Qs = Desired (constant) heat output
with steam, Btu/hr
ni = Boiler efficiency during
baseline operation as measured
by the ASME heat loss method
\\2_ = Efficiency during co-firing
From equation (5) it is obvious that the ,
fractional heat input with the waste must
be greater than the efficiency loss
associated with co-firing in order to
realize a net fuel savings. Assuming that
this is the case, net fuel savings is
given by:
Ff = Qf/HVf
(6)
where Ff = mass or volumetric fuel con-
servation rate
HVf = fuel heating value
Based on these estimates, the boi.ler
facility operating schedule, and local
fuel prices, the annual fuel conservation
credit can be calculated directly.
Waste Disposal Elimination
The savings associated with elimina-
tion of the waste disposal problem are
highly variable, and not always easy to
estimate. Because the vast majority of
waste candidates for boiler co-firing are
combustible liquids, it is likely that the
alternative form of waste disposal would
be incineration, either on-site or at a
commercial waste disposal facility.
If the waste was burned in an on-site
incinerator prior to boiler co-firing, the
variable incineration costs such as
auxiliary fuel, power, water, chemicals,
residue disposal, labor, and maintenance
would be eliminated by conversion to
boiler co-firing. Fixed costs for the
incineration.facility would not be elim-
inated, however. Thus, the analysis must
be done on a case-specific basis.
If the waste was shipped to an
off-site incineration facility prior to
boiler co-firing, the cost analysis
is more straightforward. However,
estimating an across-the-board cost
savings is still difficult. Commercial
waste disposal firms typically charge
whatever the local market will bear,
rather than a fixed percentage above cost.
Moreover, most firms charge a premium for
chlorine and ash content, and may offer a
discount for volume and/or fuel value.
Typical costs are 4£ to 5£/lb,
although special wastes such as PCB oils
may demand a price of 30£/lb or even
higher.
REFERENCES
1. McCormick, R.J. and L. Weitzman.
Preliminary Assessment of Costs and
Credits for Hazardous Waste Co-Firing
in Industrial Boilers (Draft), EPA
Contract No. 68-02-3176. U.S. Envi-
ronmental Protection Agency, Cincin-
nati, Ohio. October 1983.
2. McCormick, R.J. and R.J. DeRosier.
Capital and O&M Cost Relationships for
Hazardous Waste Incineration Contract
No. 68-02-3176.- U.S. Environmental
Protection Agency, Cincinnati, Ohio.
July 1983. p 199.
103
-------
HAZARDOUS WASTE PRETREATMENT AS AN
AIR POLLUTION CONTROL TECHNIQUE
James J. Spivey, C. Clark Allen, Robert L. Stallings
Research Triangle Institute
Research Triangle Park, NC 27709
Benjamin L. Blaney
U.S. Environmental Protection Agency
Cincinnati, OH 45268
ABSTRACT
There are many identified hazardous waste streams that contain volatile compounds
that can be emitted to the atmosphere from landfills, surface impoundments, storage
tanks, or other disposal facilities. One way to minimize or eliminate these emissions
is to pretreat hazardous waste streams to remove volatile compounds.
This project examines specific hazardous and potentially hazardous v/aste streams
(e.g., phenolic sludge from plastics production) and the technical applicability and
economic feasibility of specific pretreatment techniques (e.g., steam stripping) for
removing volatile compounds from them. In the absence of competitive removal mechanisms
(such as biological decay), all the volatile constituents will ultimately be released to
the atmosphere. The results show that air stripping, steam stripping, or evaporation
(coupled with carbon adsorption of the offgases); steam stripping; and batch distilla-
tion are the most widely applicable pretreatment techniques. An example of air strip-
ping coupled with carbon adsorption for removal of volatile constituents from hazardous
waste streams containing aromatic solvents is given. The cost-effectiveness of pre-
treatment varies widely with waste stream characteristics and type of pretreatment, with
typical values being between $0.099 and $2.00 per kilogram (kg) of volatile material
removed. This paper is a summary of a more comprehensive report prepared by the Re-
search Triangle Institute, which is unpublished as of May 1984. The reader is referred
to the final report for Task 12-5 of EPA Contract 68-02-3149 for further information.
INTRODUCTION
The purpose of this study is to
conduct a preliminary assessment of the
technical and economic feasibility of
various pretreatment techniques for the
removal of volatile constituents from
hazardous waste streams. This study was
conducted in response to increasing
concern over the potential adverse health
and environmental consequences associated
with emissions of volatile substances.
Such emissions from hazardous waste
treatment, storage, and disposal facili-
ties (TSDFs) are of particular interest.
According to preliminary unpublished
results from an Environmental Protection
Agency (EPA) national survey of TSDFs
conducted in 1981 (U.S. EPA 1983a), there
are about 4,820 TSDFs in this country
managing a total of about 151 billion
liters, or approximately 100 million
megagrams (Mg), of hazardous waste annu-
al ly. Although no comprehensive estimate
of the total volatile emissions from all
domestic TSDFs has been reported, Breton
et al. (1983) have estimated that nation-
wide emissions of 54 selected volatile
compounds regulated under the Resource
Conservation and Recovery Act (RCRA) may
be as high as 1.6 million Mg per year.
These emissions from TSDFs may represent
as much as 19.5 percent of the total
nationwide emissions of volatile organic
compounds. These results, although based
on limited data, suggest that emissions
104
-------
of volatile compounds from TSDFs are
probably significant.
There are a number of sources within
TSDFs from which volatile emissions can
be emitted. These sources include
Aerated impoundments
Landfills
Surface impoundments
Cooling towers
Storage tanks
Waste handling and transfer
operations.
Reasonably detailed data on stream
composition and physical properties are
required to evaluate pretreatment tech-
nology for removing volatile constit-
uents. Available compilations of hazard-
ous and potentially hazardous waste
stream composition and generation rate
information were evaluated to find the
one most suitable for this study. The
Waste Environmental Treatment (WET) Model
(U.S. EPA 1983b) was judged to have the
most pertinent information for an engi-
neering assessment of this type. It is
also the most comprehensive and, in spite
of some limitations (e.g., volatile
constituents are not identified for some
streams and waste streams can vary in
composition), the WET Model data were
useful for assessing potential pretreat-
ment techniques.
A thorough screening of current
technology was conducted to determine
which pretreatment techniques could be
used for volatile constituent removal/
recovery. Nine broad engineering tech-
niques were selected:
Steam stripping
Chemical oxidation
Adsorption (liquid or gas
phase)
Biological treatment
Ozonation/radiolysis
Distillation
Wet oxidation
Solvent extraction
Physical, separation.
Techniques that are used primarily for
ultimate destruction, such as incinera-
tion and pyrolysis, were excluded. The
individual compatibility of these tech-
niques and associated economics with all
WET model waste streams was evaluated.
PRETREATMENT TECHNIQUES
For each pretreatment technique,
there is a set of hazardous waste stream
characteristics (or criteria) that deter-
mine if the technique is applicable to
that stream. For example, one criterion
for using liquid phase carbon adsorption
pretreatment to remove volatiles is that
the waste stream in contact with the
adsorbent must not contain excessive
concentrations of metallic ions or sol-
ids. Using this and other criteria, an
appropriate WET Model stream has been
selected, and a detailed example has been
prepared to show how carbon adsorption
pretreatment might be used. A similar
analysis and an example have been pre-
pared for each pretreatment technique
listed above and are presented in the
final report for this work, which will be
available in the near future.
A distinction can be made between
pretreatment processes that are applic-
able at the site of generation and proc-
esses that could be used at a TSDF that
accepts waste materials from a variety of
sources. At a typical TSDF, the waste
streams are generally not segregated by
source. Thus, pretreatment using carbon
adsorption, steam stripping, and batch
distillation, which has the capability of
handling a variety of waste types, is
likely to be most applicable at TSDFs.
EXAMPLE CASE—CARBON ADSORPTION
The adsorption of organic compounds
from both liquid and gaseous phases onto
activated carbon is a mature process
technology with widespread use as an
integral unit operation in such indus-
trial manufacturing processes as corn
syrup and Pharmaceuticals production and
sugar refining, in industrial and munici-
'pal wastewater treatment, in drinking
water purification, in the separation and
recovery of organic compounds from vapor
streams, and in pollution control of
atmospheric emissions. Although acti-
vated carbon has been and continues to be
the dominant adsorbent used, other adsorb-
ents such as resin or polymeric materials
and zeolite molecular sieves have found
increasing use for a number of special
applications in recent years.
Since the selection of a pretreat-
ment process depends on the nature of the
105
-------
hazardous waste stream, a number of
different process configurations incor-
porating adsorption methods may be de-
vised to accommodate a broader spectrum
of hazardous waste streams. For this
discussion, adsorption processes may be
broadly classified into two groups:
liquid phase and gas phase adsorption
processes. Liquid phase adsorption
usually involves the addition of rela-
tively inexpensive powdered activated
carbon (PAC) directly to the waste
stream. The volatile material is ad-
sorbed and the waste is disposed of
without removal of the PAC; or, if de-
sired, the PAC can be removed and regen-
erated. Gas phase adsorption systems
(for pretreatment of hazardous wastes)
include processes where volatile constit-
uents are separated from the hazardous
waste stream into a gaseous stream for
subsequent removal by adsorption. This
allows adsorption to be used on slurries
and sludges that cannot be treated easily
by liquid phase adsorption. The phase
separation processes and adsorption
process that were considered in this
study are air and steam stripping and
evaporation. For these gas phase sys-
tems, granular activated carbon (GAC) is
the predominant adsorbent.
Process Description
The liquid phase activated carbon
adsorption process involves two basic
steps as shown in Figure 1. In, Step 1
(adsorption), the waste stream contacts
the carbon, which selectively adsorbs the
hazardous material(s) and allows the
purified stream to pass through. Step 2
(disposition of contaminated or spent
carbon) represents a number of process
options. When the carbon reaches its
maximum capacity or when the effluent is
unacceptable for discharge, the carbon is
removed from the adsorber for disposal,
destruction, or regeneration as estab-
lished by the option selected under Step
2. In some cases, the carbon can be
regenerated in such a way that the ad-
sorbate is recovered. This may be impor-
tant in pretreatment of hazardous wastes
since the recovered volatile material may
have some economic value (e.g., as a
solvent).
Process Operation
The technical suitability of a waste
Hazardous
Purified
Waste Stream
Waste Stream
Regeneration with
Destruction of
Hazardous Waste,
e.g.. Thermal
Reactivation
Regenerated with
Reclamation of
Hazardous Material
Figure 1. Steps in carbon adsorption.
stream for carbon adsorption pretreatment
depends mainly on its physical form and
the type and relative concentration of
constituents. However, other factors
that affect the treatment economics often
dictate which streams are actually feas-
ible for carbon treatment; such factors
include the required degree of solute
removal, waste throughput rate, and
carbon utilization. The discussion below
is intended to provide guidance in pre-
screening waste streams for carbon treat-
ment. Selection should be based on
laboratory adsorption and regeneration
studies with the actual waste stream.
In general, carbon adsorption is
applicable only to single-phase fluid
streams; specifically, liquid solutions
and gas mixtures. Although both aqueous
and nonaqueous streams may be treated
with carbon, liquid phase waste treatment
applications to date have been confined
to aqueous streams.
The suspended solids level in the
waste influent to the carbon contactors
(adsorbers) generally should be less than
50 ppm for optimum process operation.
Although suspended solids levels up to
106
-------
2,000 ppm may be handled in special
equipment, stream prefiltering is usually
employed with these streams.
The amount and type of nonvolatile
organics and dissolved inorganics must
also be considered. Oil and grease in
the waste stream should be less than
10 ppm to avoid carbon' fouling. High
levels of dissolved inorganics may cause
such problems as scaling (corrosion) and
serious loss of carbon activity with
thermal carbon reactivation. In some
cases, however, these problems may be
mitigated through pH control and soften-
ing of the waste influent or by acid
washing the carbon before reactivation.
Technically, a high concentration of
solute(s) in the waste influent does not
limit carbon adsorption treatment.
However, in actual practice the most
concentrated influent to be treated on a
.continuous basis contained approximately
10,000 ppm total organic carbon (TOC).
Nevertheless, higher solute concentra-
tions up to 15 percent may possibly be
carbon treated. This limit of 15 percent
solute concentration is strictly a prac-
tical and economic one. Technically,
there is no real limit to solute concen-
trations that are treatable by carbon
adsorption. In addition, when the carbon
is regenerated, the higher solute influ-
ent concentrations result fn higher
loading of the carbon (mass solute
adsorbed/mass of carbon) and thus less
frequent regeneration may be required.
The flow rate of the waste stream
does not affect the technical aspects of
carbon treatment, but it does have a
major impact on treatment economics.
Carbon treatment benefits significantly
from economies of scale, especially where
thermal1 reactivation is used. If carbon
usage is less than 500 kg/day, then the
economics generally favor disposal of the
spent carbon. At carbon usage over
4,000 kg/day, onsite thermal reactivation
is usually feasible. Where carbon rates
are between 500 and 4,000 kg/day, offsite
reactivation should be considered.
The identification of waste streams
that are suitable for carbon treatment is
more difficult when multicomponent sol-
utes are present. Some "secondary"
constituent may interfere with the ad-
sorption of the solute(s) of interest.
For example, long-chain organic soaps
were found to cause poor carbon adsorp-
tion performance on a wastewater stream
from a polyvinyl chloride production
plant (U.S. EPA 1971). Thus, preliminary
laboratory studies on the actual waste
stream are essential in prescreening
candidate waste streams for carbon treat-
ment.
In summary, the following charac-
teristics may be used as guidelines to
identify wastes streams that are likely
candidates for carbon treament:
Aqueous waste streams with
organic solute concentrations
that are less than 15 percent.
Waste streams where the aggre-
gate concentration of high
molecular weight nonvolatile
organics is substantially lower
than the concentration of the
volatile organics.
Waste streams where suspended
solids are less than 50 ppm if
the stream is not prefiltered
and less than 2.5 percent if
prefiltering is used.
Waste streams where oil and
grease concentrations are less
than 10 ppm.
Waste streams where the concen-
tration of dissolved inorganics
is low (less than 100 ppm)
unless waste stream precondi-
tioning and spent carbon wash-
ing before reactivation opera-
tions are included.
The removal efficiency of carbon
treatment can be controlled to prac-
tically any level through the design of
the carbon contactor. Typical carbon
treatment efficiencies are better than
99 percent with influent concentrations
below 1,000 ppm. At higher influent
concentrations, removal efficiencies can
exceed 99.99 percent removal to yield
effluent concentrations at several ppm.
Of course, as with most alternative
treatment processes, carbon treatment
removal efficiencies must be compared to
capital and operating costs, which in-
crease dramatically as efficiencies
approach 100 percent.
107
-------
Process Economics
Several variables and/or alterna-
tives in the design and operation of a
carbon treatment system can have a major
impact on the economics of the process.
These factors include:
Type of carbon (GAC or PAC)
Flow rate
Contact time
Process configuration (series,
parallel, or moving bed)
Number of stages
Flow direction (packed or
expanded; upflow or downflow).
WET Model Example
Waste stream 02.02.14 from the WET
Model was selected as an example to show
a typical carbon adsorption system de-
sign, associated material balances, and
treatment economics.
This stream, with a nominal rate of
426,000 kg/day (17,750 kg/h, based on 365
days per year operation) contains ben-
zene, toluene, and phenol at concentra-
tions ranging from 3,000 to 5,000 ppm.
Although these concentrations are on the
upper range of the concentrations cur-
rently being treated in commercial prac-
tice, GAC has been shown to be effective
for these constituents at these levels.
The composition of WET stream 02.02.14 is
given in Table 1.
The process flowsheet for carbon
adsorption pretreatment of the above
waste stream is shown in Figure 2. The
process design and configuration shown
includes waste stream preconditioning,
moving-bed adsorption, thermal reactiva-
tion, and furnace offgas cleaning units.
The preconditioning system is designed to
remove suspended solids to below 50 ppm
and to allow pH adjustment of the waste
influent for maximum adsorption effi-
ciency of the carbon.
The material balance for the pre-
treatment example is shown in the tabular
portion of Figure 2. Although the mate-
rial balance indicates essentially 100
percent removal of the three organics, in
actual practice the waste effluent
(stream No. 8) may contain these constit-
uents at concentrations of several ppm.
The major process uncertainties in
the design of this carbon adsorption are:
(1) the equilibrium capacity of the
carbon for the three organics in a multi-
component aqueous solution of this par-
ticular composition, (2) the adsorber
residence (retention) time, and (3) the
carbon recirculation rate. For this
case, a carbon loading of 0.3 kg adsor-
bate/kg carbon and a minimum adsorber
residence time of 30 minutes was assumed.
Also, to size the reactivation furnace, a
furnace residence time of 30 minutes was
assumed. These assumptions are consist-
ent with the ranges used in current
practice.
Table 1. Composition of WET Stream
02.02.14, Quench Blowdown from
Ethylene Production by Thermal
Cracking of Heavy Liquids
Component
Benzene
Phenol
Toluene
Solids
Water
Mass
fraction
0.005
0.003
0.004
0.01
0.98
Flow rate,
kg/h
88.8
53.3
71.0
177.5
17,359.5
Total
1.002
17,750.1
Economics
The capital and operating costs for
the above example case were based on
24-h/day, 330-day/yr operation, an ad-
sorber design capacity of 200,000 gal/
day (126 percent of waste stream rate
including recycle streams to the ad-
sorber), and reactivation furnace through-
put rate of 25,500 kg carbon/day. The
capital costs of the major components of
the carbon treatment system including
support equipment, installation, engi-
neering, legal, financing, and adminis-
trative costs are presented in Table 2.
The annual operating costs for the
system are also included in Table 2. The
major operating costs include: labor,
electricity, fuel (natural gas), mainte-
nance materials, and carbon makeup.
108
-------
~l
c
OJ
S JS
k. en
"tt <0
s.*
II
<-• S:
D. to
t N
O ea
•8-=
ro £
si
"
o 92
s §
a- E
O)
iZ
109
-------
CM
S
1
o
&
1
J£
S
&
g
1
S
&
^
r
1=
•S
<
ei
c -
|f
i~
c
i
o
c
§>
g
2
c e
^
« C
15
13
w
«
id
3
M
S
S
g
01
N
i
1
1
o
c
S
3
£
to -5
S E
«I
f^cor^mr^mcOf'^coLQ*— o oo r*-
rC rC irT T- «— ^f" CN~ ^r* co* CN co* co »- ^-"
*— CN CN CN
r*
^
i i i i i i i i i i i i i i
i i i i i i i i i i i i i ">
o
1 1 1 1 1 1 I I I i i i i CO
1 1 1 1 1 1 1 1 1 1 1 1 1 in
CM
8
o ooo
^ t^ r- r~
cocor««mu5ri*«cor**'COco*— r*o)co
tP*S'~'~CN'"SCD'~<0<>' *~
CO CO t- r- ^.
C: i t ' C: i v i " i i i i
g i g i i g i i i § i g g i
» i » i i » i i iSiSSi
*™ *~ t™ ^-
r. " - ' SP '
T— CMco^incor^cooo^'CMco *^-
(ocNcnr--«— r-nc:
ci «- r~-
"~
1 1 1 1 S CM co |
t- r- CO
co co in
1 g 5 1 1 1 1 "*
CM
S £2 !2
, CO CO i . i i 00
1 in •». 1 1 1 1 •»
CM" o>" TJ-"
CO
r.
CD
O T-
1 1 1 £• I 1 1 «°
11111111
1 1- 1 1 1 1 1 1
11111111
'
ncois*coojo«r-cN
i ,_ I 1 1 1 1
1 1 1 1 1 1 1
in
10 1 1 1 1 1 1
in
00 ...
*
§, , ,', , ,
00
ST}" O O O
| O) O O | O
1
1 1 1 1 1 1 1
1111111
1111111
CO ^t" U3 CO r*^ CO C3>
CN CN C>J CN CN CM C*J
no
-------
TABLE 2. CAPITAL AND OPERATING COSTS FOR CARBON ADSORPTION
PRETREATMENT WITH THERMAL REACTIVATION OF THE CARBON FOR
WET MODEL STREAM 02.02.14
Capital Costs
Influent pump station
Carbon adsorption system (2 pulsed-bed contactors)
Carbon regeneration system (fluidized-bed furnace)
Carbon inventory (150,000 Ib @ $0.85/1b)
Construction costs
Total Installed Cost
Annualized Operating Cost
Operating labor (36,000 man-hours @ $15/m-h)
Maintenance (5% of capital cost)
Electricity (824,000 kWh @ $0.05/kWh)
Steam (13,680,000 Ib @ $4/1,000 Ib)
Fuel (1,430,000 therms @ $0.59/therm)
Water (20,800,000 gal @ $0.40/1,000 gal)
Carbon makeup (829,000 Ib @ $0.85/lb)
Taxes, insurance, administration (4% of capital cost)
Capital recovery (16.3%; 10% over 10 years)
Total Operating Costs
Product Recovery Credit
Net.Operating Costs
Waste treated (kg/yr)
Total volatiles removed (kg/yr)
Unit treatment, cost ($/kg waste treated)
($/kg volatiles removed)
21,000
181,000
1,925,000
128,000
925,000
$3,180,000
540,000
159,000
41,200
54,700
840,000
8,300
704,700
127,200
518,300
$2,993,400
0
$2,993,400
155,490,000
1,865,880
0.019 $/kg
1.60 $/kg
^Construction fee (10%), contingency (15%), engineering (15%), startup (1%).
Advantages and Disadvantages
The major advantages of carbon
pretreatment are:
It is a mature technology in
commercial use for waste treat-
ment applications.
Carbon adsorption can handle a
broad range of organic constit-
uents and concentrations.
The disadvantages of carbon pretreatment
include:
Carbon adsorption treatment,
especially with thermal reacti-
vation, is a complex and labor-
intensive operation.
Carbon adsorption has substan-
tial operating costs.
Table 3 shows the results of the
analysis of other types of pretreatment
processes. Applicable pretreatment
processes are shown for some typical
waste types that may contain volatile
constituents. This table shows that air
stripping, steam stripping, or evapora-
tion (coupled with carbon adsorption of
the offgases); steam stripping; and
distillation are the most widely applic-
able techniques for volatile removal.
Insofar as the hazardous waste
streams are typical of the application,
the pretreatment technique, the volatile
removal efficiency (kg volatile removed/
kg volatile in the stream), unit cost
111
-------
TABLE 3. APPROPRIATE PRETREATMENT
PROCESS BY WASTE TYPE
Applicable
Waste type pretreatment process(es)
Organic liquids Distillation
Aqueous, up to
20% organic
Aqueous, less
than 2%
organic
Sludge with
organics
Some sludge in
organic or
aqueous stream
Steam stripping
Solvent extraction
Steam stripping
Steam stripping with
carbon adsorption
Resin adsorption
Air stripping with
carbon adsorption
Ozonati on/radi olys i s
Wet oxidation
Biological treatment
Air stripping with
carbon adsorption
Evaporation with carbon
adsorption
Ozonation/radiolysis
Wet oxidation
Chemical oxidation
Evaporation with carbon
adsorption
Physical separation
($/kg stream treated), and cost-effec-
tiveness ($/kg of volatile removed) are
typical of what may be expected if the
pretreatment technique were used on other
streams of a similar nature.
One further important result of this
study is an engineering judgment regard-
ing the most applicable pretreatment
technique(s) for all WET Model streams.
This is presented (in the final report
for this project) in a matrix that
matches the 12 pretreatment techniques
considered in this report with all WET
Model streams that contain volatile con-
stituents. Based on this matrix, the
following pretreatment techniques are
considered most applicable for removing
volatile constituents from the WET Model
waste streams.
Air stripping, steam stripping,
or evaporation/carbon adsorp-
tion
Steam stripping
Batch distillation.
SUMMARY AND CONCLUSIONS
In summary, this paper is a prelim-
inary examination of various pretreatment
techniques for the removal of volatile
compounds from hazardous waste streams
identified in one data base. Although
the conclusions and analyses herein may
not be regarded as final, this paper does
provide insight into the potential applic-
ability of pretreatment to reduce emis-
sions of volatile compounds from TSDFs.
The conclusions of this investiga-
tion are:
An analysis of potential air
emissions "from treatment,
storage, and disposal of 72
hazardous waste streams indi-
cated that a substantial frac-
tion of volatile hazardous
components could be emitted to
the atmosphere from landfills,
land treatment facilities,
surface impoundments, and
cooling towers. It is reason-
able to assume that in many
cases, removal of organics by
pretreatment would cause a
comparable reduction in air
emissions.
Pretreatment of these hazardous
waste streams could remove most
(90 to 99 percent) of the
volatile materials. A number
of alternative processes are
available for pretreatment for
most of the waste streams.
The cost-effectiveness varied
from $70 to over $2000/Mg of
volatiles removed. The cost-
effectiveness of pretreating
specific waste streams is not
necessarily in the same range
as the example cases. The
actual cost-effectiveness would
depend on the chemical and
physical characteristics of the
waste stream(s), the design
capacity of the pretreatment
system, and the degree of
volatile removal required.
112
-------
Pretreatment techniques using
carbon adsorption, steam strip-
ping, or batch distillation are
the most applicable ones for
the WET model streams evalu-
ated. This judgment considers
cost, the range of applicabil-
ity, and the extent to which
the technology has been demon-
strated.
REFERENCES
Breton, M., T. Nunno, P. Spawn, W. Farino,
R. Mclnnes. 1983. Assessment of
Air Emissions From Hazardous Waste
Treatment Storage and Disposal
Facilities (TSDFs), Preliminary
National Emissions Estimates, Draft
Final Report, EPA Contract 68-02-
3168, August 1983.
U.S. Environmental Protection Agency.
1971. Wastewater Treatment Facil-
ities for a Polyvinyl Chloride Pro-
duction Plant. U.S. EPA Project
No. 12020 DJI.
U.S. Environmental Protection Agency.
1983a. National Survey of Hazardous
Waste Generation and Treatment,
Storage, and Disposal Facilities
Regulated under RCRA in 1981.
Preliminary Highlights of Findings.
August 30, 1983.
U.S. Environmental Protection Agency.
1983b. W-E-T Model Hazardous Waste
Data Base, 2nd draft. Prepared by
SCS Engineers for Office of Solid
Waste. July 21, 1983. This data
base was incorporated in slightly
revised form in the "RCRA Risk Cost
Analysis Model Phase III Report,"
Office of Solid Waste. March 1984.
113
-------
PROGRAMS OF THE INDUSTRIAL WASTE COMBUSTION GROUP
AT THE U.S. EPA CENTER HILL FACILITY
George L. Huffman
Chief, Industrial Waste Combustion Group
Energy Pollution Control Division
U. S. Environmental Protection Agency
Cincinnati, Ohio 45268
ABSTRACT
The primary focus of EPA's in-
house Thermal Destruction research
being conducted at their Center Hill
Facility (part of their Cincinnati-
based Industrial Environmental
Research Laboratory) is to:
* Establish how combustion
variables affect failure of
a simulated hazardous waste
incinerator or a simulated
industrial boiler that
co-fires hazardous waste
with conventional fuel to
achieve 99.99% ORE (Destruc-
tion and Removal Efficiency);
* Determine how, when and why
PIC's (Products of Incomplete
Combustion) are formed
(e.g., too low an oxygen con-
centration or too low an
operating temperature) and
to determine how or whether
they can be subsequently
destroyed or removed (e.g.,
by secondary combustion or
scrubbing/adsorption tech-
niques); and
• Determine which organic
chemicals are the hardest
to burn [this will assist
in making better selections
of POHCs (Principal Organic
Hazardous Constituents)].
Key research being done at Center
Hill features:
Conducting parametric testing, in
bench-scale and mini-pilot combus-
tors on simulated hazardous waste
mixtures to determine the range
of operating conditions under which
efficient hazardous waste destruc-
tion can be expected and where
PICs can be minimized. The combus-
tion equipment located there
consists of a bench-scale non-flame
"TDU-GC" (Thermal Destruction Unit-
Gas Chromatograph), a "Microspray
Reactor" (a laminar-flow inter-
mediate-size unit having a quiescent
flame), a "Turbulent Flame Reactor"
(a larger, mini-pilot, water-jacketed
.unit having the ability to induce
turbulence into a flame), and a
"Controlled Temperature Tower" (a
refractory-lined, 4-foot outside
diameter by 7-foot high reactor
which provides for turbulence,
secondary combustion and simulated
cold walls). The operation of each
of these units will be supported
by an onsite GC/MS to provide
comprehensive analytical capability.
Examining key combustion variables
to determine their impact on POHC
DRE's and on PIC formation and des-
truction. Key variables will
include residence time, operating
1T4
-------
temperature, degree of turbulence,
degree of atomization, pressure drop
across the burner nozzle, oxygen con-
centration, air-to-fuel ratio, waste-
to- fuel ratio, waste mixture constitu-
ent impacts, secondary combustion
impacts, and the impacts resulting
from quenched destruction react-
ions due to cold walls in combustors.
* Carrying-out parametric testing on key
process variables attendant to the
operation of a two-stage air pollution
control system (consisting initially
of a packed-bed scrubber for HC1 con-
trol and an activated carbon bed for
unburned hydrocarbon removal) to define
the amounts of pollutant removal that
is possible (i.e., to provide data
on the "R" part of the ORE).
Specific hypotheses that will be
addressed in this In-House Thermal Destruc-
tion Research Program include:
(1) The ORE performance of a hazardous
waste combustor can be predicted
by CO/C02 or THC (total hydrocar-
bon) in the exhaust gas; conse-
quently, continuous monitoring
based on one or more of these
parameters may be suitable as a
compliance monitoring technique.
(2) Flame cooling (due to flame
proximity to cooler heat trans-
fur surfaces) and post-flame heat
heat loss can have a marked effect
on overal 1 ORE performance and on
PIC formation.
(3) The standardized POHC test mixture
used in EPA's ongoing full-scale
test burns is a representative mix-
ture for characterizing the ORE
and PIC formation/destruction per-
formance of a given combustor
across a range of operating con-
ditions.
(4) High levels of PIC's can be fed
into the post-flame zone during
the combustion of hazardous waste,
and they can be significantly
lessened there due to the large
gas residence times that exist
in that zone, or the opportunity
for secondary combustion there.
(5) A two-stage scrubber/carbon bed
system can effectively remove HC1
and organics formed in or passed
through thermal destruction
devices burning hazardous wastes.
The first three hypotheses will be
tested in the flame-mode Turbulent Flame
Reactor/Controlled Temperature Tower,
the fourth hypothesis in the non-flame
TDU-GC system and the fifth, obviously,
in the scrubber system discussed above.
115
-------
DETERMINATION OF THE THERMAL DECOMPOSITION
PROPERTIES OF 20 SELECTED HAZARDOUS ORGANIC COMPOUNDS
Barry Del linger
Juan L. Torres
Wayne A. Rubey
Douglas L. Hall
John L. Graham
University of Dayton
Research Institute
Dayton, Ohio 45469
Richard A. Carnes
U.S. Environmental Protection Agency
Combustion Research Facility/NCTR
Jefferson, AR 72079
ABSTRACT
Laboratory determined thermal decomposition profiles and kinetic data for a list of
20 selected hazardous organic compounds are reported. All data were obtained in flowing
air at mean gas-phase, high-temperature zone residence times ranging from one to six
seconds. The extrapolated temperatures required for 99.99% destruction of the parent
compound at two seconds mean residence time, Tgg.g9(2), ranged from 600°C for 1,1,1-
trichloroethane to 950°C for acetonitrile. The processes and parameters potentially con-
trolling incineration efficiency are discussed, and four previously proposed methods of
ranking compound incinerability are reviewed.
The possible chemical mechanisms for destruction of hazardous organic compounds are
examined and used to explain trends in the experimentally determined thermal d.ecomposition
data. It is proposed, through proper application of the principles of organic chemistry,
kinetics, and physics, that laboratory, gas-phase thermal decomposition data generated
under controlled conditions can be incorporated into models of full-scale incineration,
serve as a viable ranking of waste incinerability, and used to predict the formation of
products of incomplete combustion.
INTRODUCTION
The ultimate goal of hazardous waste
incineration is to destroy the waste mat-
erial with as high a destruction efficiency
(OE) as possible. Under the Resource
Conservation and Recovery Act (RCRA) of
1976, an incinerator operator must show
that the facility can adequately destroy
those hazardous waste constituents which
are most difficult to incinerate. In
theory, the permit writer will select com-
pounds within the mixture which are of
sufficient toxicity, concentration, and
thermal stability so as to be designated
as principal organic hazardous constitu-
ents (POHCs). It must then be shown,
possibly by trial burn, that the designated
POHCs can be destroyed or removed by the
particular incineration system to a
destruction and removal efficiency (DRE)
of 99.99 percent.
The development of a ranking of the
incinerability for compounds which are
candidates for POHC selection is of obvious
utility. The US-EPA currently is using a
ranking based on the heat of combustion per
gram of pure compounds. This method has
received considerable criticism, and the
development of an alternative ranking
scheme is of a very high priority to the
EPA.
Experimentally determined gas phase
thermal stability under controlled
116
-------
laboratory conditions has been proposed as
an alternative ranking method. This
report presents the results of the labora-
tory determination of the gas phase thermal
decomposition properties of twenty (20)
hazardous organic compounds. The com*)
pounds were selected by the EPA based on
their frequency of occurrence in hazardous
waste streams, apparent prevalence in the
stack effluent, and representativeness of
the spectrum of hazardous organic waste
materials.
EXPERIMENTAL PROCEDURES
All of the experimental data pre-
sented in this study were generated using
the thermal decomposition unit-gas
chromatographic (TDU-GC) system which was
designed and built with funding provided
by the US-EPA (Cooperative Agreement No.
807815-01-0) [!]. Samples of the 20 com-
pounds were prepared and introduced to the
system by several procedures depending
upon their physical state and vapor
pressure.
To initiate a test, the sample is
introduced into the system and gradually
vaporized in a flowing gas stream (i.e.,
nitrogen, air, or nitrogen/oxygen mix-
tures). The vaporized sample passes
through a controlled, high-temperature
tubular reactor where it undergoes thermal
decomposition. The products evolving from
the thermal decomposition of the compound
and the remaining parent compound are
swept into a Varian VISTA 4600 high-
resolution gas chromatograph for analysis.
The sample insertion chamber, the reactor,
and the entire transport system are fabri-
cated of fused quartz to minimize inter-
action with the sample.
In the TDU-GC system, the sample is
deposited in a sample insertion chamber
which is packed with quartz wool. The
chamber is slowly heated to 250°C to 300°C
by applying a linear temperature program
(10-20°C/min). The sample molecules are
thermally desorbed and gradually swept into
the thermal reactor.
The sample emerging from the reactor
is trapped at the head of the chromato-
graphic column by maintaining the GC oven
at a cryogenic temperature ('Mninus 30°C).
A 15 meter dimethyl silicone chemically
bonded stationary phase fused silica capil-
lary column in conjunction with a flame
ionization detector was used in the major-
ity of the investigation.
For the ethane analyses, Tedlar bags
were used to capture the reactor effluent
at the splitter. The captured gas was sub-
jected to analyses using a Varian Aerograph
Series 1800 gas chromatograph equipped with
one meter x 4.0 mm ID packed column (5 K
molecular sieve, 45/60 mesh) operated
isothermally at 150°C. This system was
utilized to ensure resolution of the three
C2 hydrocarbons.
RESULTS
For each of the twenty (20) test com-
pounds, the fraction of the feed material
undestroyed at a given set of temperatures
and mean reactor residence times (tr) was
determined. This resulted in the genera-
tion of what may be termed thermal decom-
position profiles, i.e., a semilogarithmic
plot of fraction remaining (fr)
vs. the reactor temperature (°C) at con-
stant residence time. Thermal decomposi-
tion profiles were generated in flowing
air at four mean residence times, ip = 1.0,
2.0, 4.0, and 6.0 seconds. An example of
the determination of this family of pro-
files is given in Figure 1 for chloroform.
The thermal-decomposition profile for
chloroform is representative of the great
majority of the compounds tested and serves
to illustrate several features.
The data for the test compounds have
been summarized in Table 1 with entries
for the temperature for the onset of decom-
position, T0nset (2) (°C), the inter-
polated temperature for 99% destruction,
Tgg (2) (°C), and the extrapolated tempera-
ture for 99.99% destruction, Tgg gg (2)
(°'C). All these values are for tr = 2.0
seconds in flowing air. Using only the
data presented in this table, the thermal
decomposition profile for the compounds
may be approximately reconstructed. The
table lists the compounds in order of
decreasing temperature required for 99%
destruction efficiency. A slight reorder-
ing occurs if Tgg^g(2) is used for rank-
ing. However, the numerical differences
for the reordered compounds are small.
For the conditions possibly
encountered during gas-phase thermal decom-
position in an incinerator (600°C to
1,400°C and oxygen levels of 0.1 to 21%),
two possible global decomposition pathways
117
-------
cC
O
«££
o
H- «
^4
t/>
2
IE
O
0
LU
a
t
S*
S
LU
U.
O
^_
cc
n
l/>
r—
Ul
CO
&
1—
"o
E
LO r—
re
u
^-^
i
•— •
"*
^^
CM
*"cn^~«
encj
ov— •
en
I-
^ i»
CM ^"^
^-^CJ
cno
ov — •
CM
| \ —
'('"* '
cucj
LOO
H?~
i—
re re
O •—
•r- 3
J- S
•r- i.
0. 0
E Lu
LU
"2
o
o.
O
O
f*^
O
X
r-
<3-
o
in
en
o
o
en
o
to
t-.
2£
CO
IE
CM
°
0)
•r-
S.
4J
•r—
C
O
+J
a)
u
>
QJ OJ
H—
P— co ^~ CD *d~ "^i* ^o cn en en *=i* co f~~ " co "d* en en CM
CO-=t--*CO CMVDCM LO CO COCMCOI~-.CMlOCM-^-CO
CO CM LO LO CM
locncoio mi — in i — co oo LO co i— •=)• i — r--i — co
OOOO OOO O CD OOOOOOOOO
XXXX XXX X X XXXXXXXXX
coLOLoen i — oco co *cncncri
i — COCMi — i — COCM IO CM COi — CMCncO" — i — CMi —
OOCDO OOO CD CD OLOOOOOOOO
usr^ooin ^J-COCM co cn coco^DLooocD«d-CMCD
co co co co cot^co t-~ r~- r-~r--i — p-.t~.t-~u3%j=re •>•»-> -r- jr xi i — too) «c i roNi — oi-reo «.c
i-4-> XCMCU S- O i--C X-r-CMCUCM^: C-^ C+J Xi — i P
o cu cu "I— >,••!- re o cu "a ».a «4->cuco-i-cu^: «cu
eCSICf" Q.QCJ 1C i — » — ; LUCCl«=CSs:inCJi —
118
-------
100
10
8-
iu
O.I
0.01
CHLOROFORM
O tr= 1.0
n fr = a.o
A fr =4.0
o tr=e.o
100
400
5OO
EXPOSURE TEMPERATURE, °C -
600
700
Figure 1.
Thermal Decomposition Profiles for Chloroform in Flowing
Air at Mean Residence Times of 1.0, 2.0, 4.0, and 6.0
seconds.
predominate, pyrolysis and oxidation.[2]
The global expression for these two
reaction schemes are
dt
= k][A]a+ k2[A]a[02]b.
(1)
large excess of molecular oxygen relative
to the concentration of the waste material,
the decomposition equation could be simpli-
fied to an expression that is first order
in the concentration of the sample. This
resulted in the integrated pseudo-first
order rate expression:
where: k-i and k2 are the global rate con-
stants for pyrolysis and oxidation,
repsectively, and a, a, and b, are
the reaction order for the decom-
position of species A with respect
to A and 02.
The time dependence is included in this
expression. The temperature dependence is
included'in the rate constants for the two
processes which may be expressed by the
Arrhenius equation:
fr = exp (-k2' tr)
(3)
k = A exp (-Ea/RT)
(2)
where: fr is the fraction of the parent
species remaining, and
k2' = k2[02l is the pseudo-first
order rate constant
This expression may be combined with the
Arrhenius Equation to yield an expression
for the required temperature for a given
level of destruction in an atmosphere of
flowing air:
-1
where:
Ea is the activation energy for the
process, cal mole"'
A is the Arrhenius coefficient, s""1
R is the universal gas constant,
1.99 cal mole"1 °K-1.
For each of the test compounds, it was
found that when .the thermal decomposition
reaction occurred in an atmosphere with a
= BOS E
a
In
In
(4)
where:
is the temperature required for
a given DE, °K,
has units of kcal mole" and
the other variables are as pre-
viously defined.;
A plot of In fr vs. tr for the four
119
-------
residence times will yield the rate con-
stant for the reaction at a given tempera-
ture. A plot of In k vs. 1/T for the four
experimental temperatures should then
yield a straight line with the slope equal
to -Ea/R and an intercept of In A.
Regression analyses of this type have
been performed on each of the twenty (20)
test compounds. In all cases, the first
order kinetic plots yielded far better
fits than zeroth or second order plots.
The measured kinetic parameters, A and Ea,
are included in Table 1.
DISCUSSION
Several methods have been proposed
for ranking the relative incinerability of
hazardous organic compounds. These
include heat of combustion per gram mole-
cular weight (He/gram of the pure compound)
autoignition temperatures (AIT) of pure
compounds, theoretical approach based on
the kinetics of flame mode thermal decom-
position, and gas-phase thermal decomposi-
tion of pure organic compounds in flowing
air [3-7]. Other parameters can be envi-
sioned which might be appropriate as a
basis for such a scale. An attempt was
made to correlate laboratory generated gas-
phase thermal decomposition data reported
in the previous section with the previously
proposed rankings scales. This comparison
was complicated by lack of overlap of the
compounds investigated.
For the heat of combustion scale, the
only agreement that could readily be dis-
cerned was the general increase in thermal
stability with decreasing heat of combus-
tion for the chlorinated benzenes.
There appeared to be a general posi-
tive correlation for both Tgg(2) and
Tgg.ggU) with AIT for those compounds
with an AIT below 550°C. Above this tem-
perature the gas phase thermal stability
appeared to vary little with AIT.
Direct comparison of thermal decom-
position data with the theoretical ranking
was hampered by the fact that only four of
the 20 compounds were ranked by the group
at NBS who developed the approach. How-
ever, the thermal stability of the chloro-
benzenes ranked by NBS fell in the same
order as suggested by the Tggsgg(2i) based
ranking. Chloroform was predicted to be
rather unstable due to the relatively low
energy carbon chlorine bond. This thermal
instability is evident from the gas-phase
thermal decomposition results, although it
appears to be even more fragile than pre-
dicted by the NBS ranking.
Only five of the test compounds were
studied at Union Carbide. The most signi-
ficant disagreement was for acrylonitrile
which was the least stable of the five
compounds based on the Union Carbide cal-
culated Tgg.gg(2), while the data from
this study indicates its thermal stability
rivals that of methane. Ethane is slightly
less stable than monochlorobenzene, as
measured by Union Carbide, while this trend
is reversed in the present data, although
the difference is small. With the excep-
tion of acrylonitrile, the two rankings are
similar although the data from this study
predicts the compounds to be typically more
stable by 40°C than that predicted by
Union Carbide.
No single proposed ranking scheme or
molecular parameter was identified which
correlated with all our thermal decomposi-
tion data, although trends were observable
in homologous subgroups. By careful consi-
deration and utilization of the principles
of chemical reactions, it is possible to
explain the behavior of each of the twenty
compounds on a relative basis and identify
possible mechanisms that can be used to
extrapolate this limited data to other com-
pounds not studied in the laboratory. The
20 test compounds may be divided into five
subclasses; these classes are discussed in
the following paragraphs.
Methane, Dichloromethane, Chloroform, and
Carbon Tetrachloride
The observed trend in this group is
decreasing thermal stability with increas-
ing chlorine substitution, except for car-
bon tetrachloride which is intermediate in
thermal stability between methane and
dichloromethane
The data are in agreement with a mech-
anism based on abstraction of a hydrogen
probably by an electrophile. Since the
carbon-hydrogen bond dissociation energy
(BDE) decreases with increasing chlorine
substitution up to chloroform, one would
predict decreasing thermal stability,
which is observed. However, carbon
tetrachloride contains no hydrogens and
thus would not be susceptible to this mode
120
-------
of attack. It would instead be expected to
decompose by bond rupture.
Benzene, Monochlorobenzene, 1,2-
Dichlopobenzene, 1,2,4-Trichlorobenzene,
1,2,3,4-Tetrachlorobenzene, Hexachloro-
benzene, Pyridine, Aniline, Nitrobenzene
1 The observed trend is increasing ther-
mal stability with increasing chlorine
substitution. Pyridine is more stable,
nitrobenzene is less stable, and aniline
has about the same thermal stability as
benzene. All of the bonds in benzene, the
chlorinated benzenes, and pyridine are
probably in excess of 90 kcal/mole and one
would expect electrophilic addition to be
the predominant reaction path. The chlor-
ines and the nitrogen in pyridine are more
electronegative than hydrogen or carbon
which leads to a destabilization of the
electron deficient intermediate resulting
from electrophilic addition, thus reducing
the rate of decomposition and greater ther-
mal stability than benzene.
In cases of aniline and nitrobenzene,
their stability relative to benzene can
also be explained by electrophilic attack.
A resonance stabilized intermediate may be
envisioned for nitrobenzene, the nitrogen
carbon bond dissociation energy (BDE) is
70 kcal/mole which may be easily broken and
therefore, represents an alternative mode
of decomposition.
Ethane 1,1,1-Trichloroethane,
Hexachloroethane
,0ne might expect ethane to be
destroyed by unimolecular decomposition
through bond rupture at the weakest bond.
Due to the 88 kcal/mole carbon-carbon bond
one would predict significantly less sta-
bility than methane but still moderate
thermal stability for ethane.
Although similar in structure, one
would expect the pathway for decomposition
for 1,1,1-trichloroethane to be concerted
elimination of HC1 which is a very low
energy process [10]. Based on this, one
would expect this compound to be one of the
least stable compounds studied, which is
in fact, the case. Hexachloroethane would
have to eliminate Cl2 to proceed by a
concerted pathway. This process is more
endothermic and the decomposition of
hexachloroethane would instead be expected
to proceed through carbon-chlorine or
carbon-hydrogen bond rupture, both BDE's
being approximately 73 kcal/mole. Based on
these considerations, one would expect
hexachloroethane to be intermediate in
stability between ethane and 1,1,1-
trichloroethane, which is the observed
trend.
Tetrachloroethylene and Hexachlorobutadiene
Both of these compounds are quite
stable. This is probably due to the large
BDEs caused by sp2 hybridization of all
carbon atoms and the lack of hydrogen atoms
available for abstraction. Electrophilic
addition however, may be the predominant
mode of attack under incineration condi-
tions. One would then expect hexachloro-
butadiene to be thermally less stable
since it can form an ally! radical inter-
mediate which is known to be quite stable.
Acetorn'trile and Acrylonitrile
These compounds are very stable and
all BDEs are 93 kcal/mole or greater.
Carbon nitrogen triple bonds are expected
to be much less reactive towards electro-
philic addition than double bonds.
Apparently, the only mode of decomposition
for acetonitrile is loss of a hydrogen
through bond rupture or abstraction.
These are both high energy processes which
account for its stability. Acrylonitrile
may, on the other hand, be susceptible to
addition at the carbon carbon double bond.
CONCLUDING REMARKS
Preliminary calculations indicate
that "fault" or "failure" modes of incin-
erator operation, or equivalently the
"extremes" of operational parameter dis-
tribution functions, may well control
measured incineration efficiency for full-
scale units [11,12]. Following this line
-of reasoning, one would conclude that the
incinerator effluent would only contain
undecomposed feed material and products
of incomplete combustion (PICs) which were
formed under these conditions of failure.
Thus, "fault" modes are essentially worst
case conditions and appear to have a
dominating effect on the composition of
the incinerator effluent.
A point to be made along this vein is
that the temperature at which this study
was conducted is probably representative
of fault modes. The experimental
121
-------
laboratory temperature range covered 0 to
99.935 destruction of the feed material which
is typically several hundred degrees below
mean temperatures quoted for hazardous
waste incineration. If a given incinerator
does not meet the 99.99% destruction effic-
iency requirement, yet has a high mean
operating temperature, then a likely possi-
bility for its failure is that a fraction
of the waste feed experiences temperatures
somewhat lower than the mean (where destruc-
tion efficiency is low) i.e., the destruc-
tion efficiency and temperature range
measured in laboratory studies.
Thus, one might expect the actual PICs
emitted from the incinerator to be the same
as those formed under the conditions
studied in the laboratory. Furthermore,
this reasoning suggests that the relative
thermal stability of hazardous wastes
should be compared at fault mode tempera-
tures since only this fraction of the waste
is escaping incineration. Since selecting
a suitable temperature for comparison of
every compound is difficult and still some-
what arbitrary, a ranking based on the tem-
perature required for 99% destruction at
2 seconds mean residence time wouldvseem to
be a reasonable measure of POHC destructa-
bility. One could just as well select 90%
or 99.9%, but examination of the data shows
that the rankings over this range are
essentially identical.
Products of incomplete combustion have
not been determined in this study. PIC
determination would remove much of the
speculation in the discussion of reaction
mechanisms. In previous research, the
formation of numerous PICs from a wide
variety of organic compounds has been
observed [13]. These PICs have, in some
cases, been produced in as much as 30%
yields and been as hazardous or more haz-
ardous than the parent compound. The det-
ermination of PICs should be an integral
part of future research programs and consi-
dered for inclusion in an ultimate ranking
of waste incinerability.
The reported research has addressed
non-flame, high-temperature, gas-phase
reaction chemistry. Extension of this work*
to include so-called flame mode studies
represents a special challenge due to the
difficulties in scaling results. Of all
incineration processes which must be
modeled to full-scale, gas-phase chemical
kinetics is the easiest and most success-
fully performed. The temporal and spatial
distributions present in small laboratory
or bench-scale flames are not easily scaled
to the turbulent poorly-defined flames pre-
sent in full-scale systems. Consequently,
an elementary chemical kinetic approach to
determining flame mode destruction
efficiencies might prove most effective.
The importance of hydroxyl radicals in
flames is well documented [14]. Thus, an
experimental program to determine the rate
of attack of hydroxyl radicals on hazardous
wastes would produce kinetic results which
is easily scaled. Kinetic data of this
type in combination with measurement or
estimation of hydroxyl radical concentra-
tions in full-scale systems would allow
simple scaling of laboratory results to
full-scale. These data over different
temperature ranges would be applicable
to both flame and non-flame modes of
destruction.
REFERENCES
1. Rubey, W. A., Fiscus, I. B., Torres,
J. L., Description and Operation of a
Thermal Decomposition Unit-Gas
Chromatographic System, prepared for
US-EPA under Cooperative Agreement
CR-807815, 1984.
2. K. J. Laidler, Reaction Kinetics,
Vol. 1., Homogeneous Gas Reactions
eds., W. C. Agosta and R. S. Nyholm,
Appleton-Century-Crofts, New York, NY,
1971.
3. E. P. Grumpier, E. J. Martin, and
G. Vogel, Best Engineering Judgement
For Permitting Hazardous Waste
Incinerators, Presented at ASME/EPA
Hazardous Waste Incineration
Conference, Williamsburg, Virginia,
May 27, 1981 .
4. J. J. Cudahy, L. Sroka, and W. TroxTer,
Incineration Characteristics of RCRA
Listed Hazardous Hastes, IT
Environscience Draft Report Submitted
under EPA Contract 68-03-2568, July,
1981.
5. W. Tsang and W. Shaub, Chemical
Processes in the Incineration of
Hazardous Materials, presented at
American Chemical Society Symposium
on Detoxification of Hazardous Wastes,
New York, NY, August, 1981.
122
-------
6. B. Dellinger, J. L. Torres, and R. A. 11.
Carnes, The High-Temperature Gas-Phase
Oxidation of Selected Chlorinated
Aliphatic and Aromatic Hydrocarbons,
presented before the Division of
Environmental Chemistry.of the 185th
American Chemical Society Meeting,
Seattle, WA, March, 1983.
7. K. C. Lee, N. Morgan, J. L. Hansen, 12.
and G. M. Whipple, Revised Model for
the Prediction of the Time-Temperature
Requirements for Thermal Destruction
of Dilute Organic Vapors and its Usage
for Predicting Compound Destructabil-
ity, Presented at 75th Annual Meeting
of the Air Pollution Control
Association, New Orleans, June, 1982. 13.
8. J. A. Kerr and A. F. Trotman Dickinson,
Bond Strength of Polyatomic Molecules,
IN: CRC Handbook of Chemistry and
Physics, Chemical Rubber, Co.,
Cleveland, OH 59th Ed., 1979.
9. S. W. Benson, Bond Energies*, Journal 14.
of Chemical Education, Vol. 42, No. 9,
pp. 502-518, September, 1965.
10. S. W. Benson, and H. E. O'Neal,
Kinetics on Gas Phase Unimolecular
Reactions, NSRDS Report NBS-21, U.S.
Government Printing Office (1970).
B. Dellinger, J. L. Torres, W. A.
Rufoey, D. L. Hall, J, L. Graham,
Determination of the Thermal
Decomposition Properties of 20
Selected Hazardous Organic Compounds,
Final Draft Report prepared for
US-EPA under Cooperative Agreement
CR-807815,, 1984.
J. C. Kramlich and W. R. Seeker,
Laboratory-Scale Flame Mode Study of
Hazardous Waste Incineration,
Presented at Ninth Annual Research
Symposium on-Solid and Hazardous
Waste Disposal, Ft. Mitchell, KY,
May, 1983.
D. L. Hall, B. Dellinger, and W. A.
Rubey, Considerations for the Thermal
Degradation of Hazardous Haste,
Prepared for Presentation at 1983
(4th) International Symposium on
Environmental Pollution, Miami Beach,
Florida, October, 1983.
The Mechanisms of Pyrolysis, Oxidation,
and Burning of Organic Materials,
L. A. Wall, ed., Proceedings of the
Fourth Materials Research Symposium
held by NBS, Gaitherburg, MD, NBS
Special Publication 357 CODEN:XNBSAV
(1972).
123
-------
HAZARDOUS WASTE DESTRUCTION USING PLASMA ARC TECHNOLOGY
Thomas G. Barton
Pyrolysis Systems Incorporated
Well and, Ontario L3B 5P1
Nicholas P. Kolak
New York State Department of Environmental Conservation
Division of Solid and Hazardous Waste
Albany, New York 12233
Chun Cheng Lee
U.S. Environmental Protection Agency
Industrial Environmental Research Laboratory
Cincinnati, Ohio 45268
ABSTRACT
A thermal plasma properly applied to toxic and hazardous waste destruction pro-
vides a pyrolytic environment with distinct advantages over other competing technolo-
gies. Complete atomization of organic fluids have been demonstrated to occur in less
than one third of a millisecond. Kinetic recombinations of atomic entities are predict-
able and generally form acid gas, fuel gas and graphitic soot. Since the 'process is
pyrolytic, the scale of the equipment is small for high throughput rates. Energy
requirements for the destruction of organic fluids are typically less than one kilowatt-
hour per kilogram of waste. A mobile prototype is being constructed for a throughput
rate of four kilograms per minute of liquid organic waste. The mobility and efficiency
of such systems may have significant advantages over fixed facilities.
An assessment of approximately
forty emerging waste destruction techno-
logies was conducted by the Environmental
Protection Agency (EPA). Five were
selected for demonstration since they
showed a high potential for successful
application to toxic and hazardous waste
destruction. Under a co-operative agree-
ment between the New York State Depart-
ment of Environmental' Conservation
(NYSDEC) and the EPA Industrial Environ-
mental Research Laboratory (IERL) 1n
Cincinnati a demonstration program was
established and jointly funded to further
Investigate mobile plasma pyrolysis. A
contract was awarded by NYSDEC • to
Pyrolysis Systems Incorporated of
Hell and, Ontario in 1983 for the fabrica-
tion and demonstration of the mobile
prototype.
The plasma arc device forms the
heart of this technology. A co-linear
electrode arrangement creates an electric
arc which is stabilized by field coil
magnets. Low oressure air is used as a
medium through which an electric
current 1s passed. In passing through
the air, the electrical energy is
converted to thermal energy by absorption
by the air molecules which are activated
into ionized atomic states, losing elec-
trons in the process. Ultraviolet radia-
tion is emitted when the molecules or
, atoms relax from their highly activated
states to lower energy levels.
Waste fluids can be injected into
'this plasma zone to interact with decay-
ing plasma species to become rapidly
atomized prior to exiting this electrode
space. An integral part of this process
124
-------
ATOMIZATION m EQUILIBRATION P QUENCHING
Figure 1. Process Schematic.
technology is the pyrolytic simulation
computer model . New compounds which are
created from the recombination of atomic
species are predictable based on kinetic
equilibrium. The minimization of Gibb's
free energy is used to determine the
equilibrium concentration of product
species over a wide range of selected
temperatures and pressures. Thus, appro-
priate operating conditions can be pre-
dicted, for any organic waste fluid prior
to its destruction. Similarly, undesir-
able products can be minimized or elimi-
nated through altering the character of
the fluid (blending) or the operating
conditions. Changes in enthalpy between
feedstocks and predicted output products
are used to determine the plasma energy
required adiabatically for the pyrolysis
of the waste being treated.
A simplified process schematic for
the mobile prototype 1s shown in Figure
1. A five hundred kilowatt plasma device
is fitted' to one end of a stainless steel
reaction chamber and mated to a hollow
graphite core to form an atomization
zone. Residence time in this atomization
zone is approximately five hundred micro-
seconds. The reaction chamber serves as
the equilibration zone where the atomized
species recombine to form new products.
This zone is equilibrated at a temperature
range of 120G-1800 Kelvin and the resi-
dence time in this zone is approximately
one second.
A three stage spray ring is located
at the producer gas outlet from the plasma
reaction chamber. The spray ring is
injected with water and appropriate
amounts of liquid caustic soda to quench
125
-------
the product gas, neutralize acidic
products and trap partlculates. Salt
water and partlculates are pumped from
the scrubber sump and sampled prior to
approved discharge. An induction fan
continuously draws on the reaction cham-
ber and scrubbing system to maintain a
slightly negative pressure in the sys-
tem. Producer gas from the induction fan
passes through a three-way valve. During
normal operations, this gas flows to the
flare stack where it is electrically
ignited. Since this gas is mainly hydro-
gen and carbon monoxide, it burns with a
clean flame at a temperature of 2000-2300
Kelvin. This flare stack serves as an
air pollution control device to prevent
the release of fuel gas to the environ-
ment. In the evant of a power failure,
the three-way valve directs the product
tgas_ through an activated carbon filter to
'block' the potential release of un-
destroyed toxic material to the environ-
ment. Provision is made for sampling the
product gas prior to the flare.
During normal operation, the plasma
pyrolysis unit requires approximately
three minutes of warmup time prior to com-
mencing injection of toxic or hazardous
wastes. During warmup, the system is
Injected with a flushing fluid such as
ethanol to maintain a reducing atmosphere
inside the atomizatlon zone. This unit
also requires approximately twenty minutes
to cool key components in order to conduct
post run maintenance. These short cycle
times make this unit basically an on/off
system which can respond rapidly to
adverse conditions. The system hardware
is arranged to provide failsafe operation
even in the event of total power failure.
Normal scheduled maintenance of pre-engi-
neered components such as fans, motors,
valves., meters and sensors should safe-
guard against unforeseen malfunctions.
Thus, consideration for 'failsafe is
limited to monitoring essential conditions
and providing protection in the event of
an adverse situation.
A HACSYM 350 process control com-
puter is used to control the total opera-
tion of the system. The computer provides
operator warnings as well as taking posi-
tive action in the occurrence of adverse
conditions. Under total loss of power,
the arrangement of hardware takes over
this latter function. Waste feed shutdown
signals are triggered by the computer if
it senses any loss of the plasma arc,
induction system and cooling systems or
notes the presence of unacceptable pro-
ducts in the stack product gas. If the
arc is lost during toxic waste injection,
the response time of the shut off valve
may allow as little as five micrograms of
undestroyed waste to pass through the
pyrolysis system to be trapped on the
activated carbon filter. Unacceptable
products directed to the carbon filter may
include trace levels of principle organic
hazardous constituents or excessive
release of hydrogen chloride.
In addition to providing waste feed
shutdown signals and taking appropriate
action, the computer system, through total
system monitoring, will also provide oper-
ator warning signals which, if not reme-
died in a preset time period, will auto-
matically shut the system down. Such
warning signals stem from monitoring all
temperatures, pressures, flows, fluid
reserves and standard operating para-
meters. Failsafe criteria are instituted
to safeguard the environment. Obviously,
the degree to which the environment is
protected will bear directly on the
accuracy of monitoring activities.
All hardware is designed to be
located within a forty-five foot long
moving van type trailer. The hitch end
houses the six-pulse, water cooled
thyristor unit which is fed with primary
480 volt, three phase power. The rear
section of the trailer houses the skid
mounted hardware and access is made for
the connection of domestic water and
sanitary sewer services. The central
area of the trailer contains a laboratory
1n which the control and monitoring
equipment are located. In the mobile
unit, trace analysis will be performed by
a Hewlett-Packard 5792A gas chromotograph
coupled to a Hewlett-Packard 5970A mass
selective detector (MSD). The MSD has a
detection level of one nanogram while in
the spectral scan mode and 10 picograms
while in the selected ion mode. It is
proposed at this time to monitor the pro-
duct gas before it is flared. Detection
of trace chemicals 1s assisted by their
being at their highest molar concentra-
tion at that point, where they have yet
to be altered through combustion in the
f1 are.
Bulk gas analysis of the scrubbed
product gas will also be performed at a
126
-------
sampling point downstream of the induc-
tion fan. A Hewlett-Packard 5880A gas
chromatograph complete with both flame
ionization and thermal conductivity
detectors will provide analysis of
hydrogen, water, nitrogen, methane,
carbon monoxide, carbon dioxide,
ethylene, ethane, 'acetylene, propane,
propylene, 1-butene and hydrogen
chloride. A detection of 200 parts per
million (ppm) of hydrogen chloride in the
producer gas will signal a ninety-nine
percent removal efficiency of hydrogen
chloride in the scrubber during the
destruction of wastes containing
thirty-five percent chlorine by weight.
It is proposed that this unit be
demonstrated in Canada during the summer
of 1984. A test program is presently
under review by the Canadian Federal and
Provincial agencies. Subsequent testing
within New York State is contingent upon
Federal and State Agency permits.
PROBLEMS ENCOUNTERED
During this project, problem areas
have been encountered involving design
changes which stemmed from a continuous
indepth- review 'process. The as-built
schematic for the mobile unit is signifi-
cantly altered from the proposal original-
ly put forward in the contract. Many of
the changes resulted from the selection of
alternative hardware. - Others were re-
quired to improve the failsafe perform-
ance. Examples of additional equipment
added or equipment modified are the de-
ionized water system, activated carbon
filter, air dryer, scrubber water tank,
scrubber water pump, plasma field current
supply, redundancy in instrumentation and
sensors as well as a large selection of
valves, piping and wiring to accomodate
the requested modifications.
Modifications to the Westinghouse
plasma arc system were necessary to ensure
that it would provide a similar process
performance at an equivalent plasma effi-
ciency. Several weeks of design discus-
sion and testing on a Marc II prototype
device and its, modified version were con-
ducted to verify the final design. This
design was finalized in February, 1984 and
the unit is scheduled for delivery in
May. The prototype unit achieved destruc-
tion efficiencies of 99.9999 percent for
askarel solution fed at a rate of 3.6
kilograms per minute on the laboratory
reactor at the Royal Military College in
Kingston, Ontario. Similarly, parametric
data for the modified device show a pre-
dictable performance envelop for a rated
performance between 200-500 kilowatts.
Plasma enthalpies .in excess of ten kilo-
watt-hours per normalized cubic meter were
also achieved.
Present guidelines call for moni-
toring air emissions from the unit. How-
ever, it has been shown that the parti-
tioning of toxic compounds passing through
the scrubber leaves better than ninety
percent of them in the scrubber water.
Calibration of the laboratory scale scrub-
bing unit shows that better than ninety-
six percent of low volatile non-soluble
(LVNS) compounds are trapped by the scrub-
ber fluid. During typical operation the
producer gas and scrubber fluid flowrates
are fifteen cubic meters per minute and
forty liters per minute respectively.
Calculation will show that the LVNS oro-
qucts are more than 69,000 times more
concentrated in the scrubber fluid than
in the producer gas prior to flaring.
These calculations have been verified
with parafin oil which is characteristic
of most LVNS products of interest
(dioxin, etc.). Thus the question of how
to measure the performance of this new
technology has yet to be resolved. It
may be more practical to exercise control
over the performance of the system by
monitoring for toxic effluents in the
scrubber water since it functions on a
single pass basis.
RESULTS
The fabrication of the mobile unit
should be completed in May 1984. Demon-
stration testing of the unit is forecast
to start in June 1984 and will continue
until all regulatory and monitoring con-
cerns have been addressed-. Guidelines to
assess this technology and the protocols
127
-------
for this assessment are currently under
development and should be completed by
June.
ACKNOWLEDGEMENTS
Funding for this project has been
made possible from the New York State
Department of Environmental Conservation
and the Environmental Protection Agency
Industrial Environmental Research Labora-
tory. Special recognition is given to
Mr. John Grimaud '(NASA) for valuable
advice in the course of this effort.
REFERENCES
(1) Barton, Thomas G., Unpublished test
data from the Department of Civil
Engineering, Royal Military College
of Canada, February 1984.
(2) Gordon, Sandford, Computer Program
for Calculation of Complex Chemical
Equilibrium Compositions, NASA N78-
17724, 1971.
(3) Lee, Chun Cheng, A Comparison of
Innovative Technology for the Des-
truction of Hazardous Wastes.
USEPA/IERL Cincinnati, Ohio 45268,
1984.
128
-------
METAL VALUE RECOVERY FROM METAL HYDROXIDE SLUDGES
L. G. Twidwell, Professor
A. K. Mehta, Adjunct Professor
G. Hughes, Research Associate
Department of Metallurgy and Mineral Processing Engineering
Montana College of Mineral Science and Technology
Butte, MT 59701
ABSTRACT
Metal bearing hydroxide sludge material is generated by the metal finishing industry.
This product has traditionally been disposed of in hazardous landfill sites. Long-term
maintenance of such sites is required and metal values are unnecessarily lost. The pre-
sent research effort was initiated to investigate the application of well established hy-
drometallurgical unit operations to mixed metal bearing hydroxide sludge materials. Test
equipment capable of treating 75-100 Ibs. per day of mixed metal sludge has been assembled.
The test system has been used on a variety of hydroxide sludge materials to demonstrate
successful dissolution and selective and effective metal value separation.
INTRODUCTION
In recent years increased emphasis has
been placed on preventing the introduction
of heavy metal containing industrial waste-
waters in publicly owned treatment works
and the environment. Legislation has
established regulatory authority for con-
trolling the discharge of heavy metals into
the environment. It also has mandated
resource recovery whenever economically
feasible. Many treatment and control tech-
nologies have come into existence to remove
metals from these wastewaters, i.e., a
sludge, concentrate, or regenerate form is
created and is, in most cases, disposed,of
in a landfill. Metals are recoverable, but
are not recovered significantly because of
a lack of proven technologies.
Process wastewaters from the metal fin-
ishing and electroplating industry contain
cyanides and heavy metals. These waste-
waters have a detrimental effect on the en-
vironment if discharged directly. Such dis-
charges are regulated by Federal, State,
County, or City ordinances, and require in-
stallation of treatment .technology. One of
the treatment technologies presently in use
is oxidation, neutralization and precipita-
tion, which destroys cyanide and removes
heavy metals as a hydroxide sludge. This
product has traditionally been disposed of
in hazardous landfill sites.
Disposal of sludges in landfills has
certain inherent disadvantages:
1. Perpetual maintenance of the disposal
site is required,
2. Dilution in metal content by mixing
with other types of waste materials,
and,
3. Permanent loss of non-renewable metals.
If heavy metals are recovered from metal
finishing sludges, it will alleviate or re-
duce the disposal.problem and provide for
conservation of energy and metal resources.
The present study outlines a technical
methodology to treat metal bearing sludges
by hydrometal1urgical techniques.
The treatment of hydroxide sludges for
metal value recovery will produce several
beneficial results, i.e.., economic benefits
from the metal values recovered will help
offset the cost of recovery/treatment; non-
renewable resource metals will be recycled
for use by society, and there will be
129
-------
significantly less hazardous material to be
disposed of in landfills.
PURPOSE
The purpose of the present study was
to investigate at an advanced laboratory
scale the potential for application of well
established hydrometallurgical techniques
to a mixed metal sludge. The design, de-
velopment, fabrication, acquisition, and
assembly of a treatment system has been
conducted at the Montana Tech Foundation
Mineral Research Center in Butte, MT.
The objectives of the study included:
•Develop a flowsheet to separate and
recover metal values from metal fin-
ishing hydroxide sludge materials,
•Develop a test assembly of unit oper-
ations to accomplish the separation
of metal values on a scale of 75-100
Ibs. of sludge per day, and
•Verify that the large scale unit
operations accomplish the separation
of metal recovery efficiency for each
unit operation; delineate process and
materials handling problems when
treating complex mixed metal sludge
materials.
APPROACH
A laboratory test program has been
conducted to support the development of an
appropriate flowsheet made up of unit oper-
ations designed to accomplish the stated
objective of separation'and metal value re-
covery from mixed metal hydroxide sludge
materials. The laboratory test program was
based on a comprehensive review of current
literature; discussions with consultative
experts in the field of extractive metal-
lurgy; and previous experimental research
conducted at Montana Tech Foundation. A
flowsheet was designed based on the inputs
gained from the above sources and modified
as dictated by the laboratory test program.
The flowsheet that resulted from the
Phase I study was based on the treatment of
a mixed metal sludge material containing
significant concentrations of metal values:
copper, nickel, zinc, chromium, and vari-
able concentrations of other elements that
were not considered recoverable as mar-
ketable products but which required removal
in order not to contaminate the metal value
products, e.g., iron, aluminum, and calcium.
The developed flowsheet is not an unalter-
able sequence of operations; alternatives
do exist and will be discussed in the body
of this report. As is often the case, unit
operations may be accomplished by several
different technical approaches, e.g., the
unit operation of chromium oxidation may be
accomplished by use of chemical oxidation
reagents or by use of electrochemical cells.
The choice is usually based on efficiency
and economical considerations.
Another important point concerning the
development of the flowsheet is that only
commercial process unit operations and com-
monly used reagents were considered. That
is, new developments in the separation of
metals from complex solutions have been re-
ported but are not yet adopted commercially,
e.g., iron removal from a mixed metal solu-
tion by solvent extraction techniques.
These new developments are reviewed in the
final project report and the consequences
of th.e adoption into the present flowsheet
are discussed.
The flowsheet developed for mixed
metal sludges is presented in Figure 1.
The treatment sequence consists of the
following unit operations:
•Sulfuric acid dissolution of the
metal hydroxides.
•Selective precipitation of iron from
the solution as a potassium or
sodium jarosite ((KFe3(OH)g(S04)2).
•Solid-liquid separation of the leach
residue and precipitated jarosite
solid.
•Selective extraction of copper from
the leach, solution (with subsequent
copper recovery as electrodeposited
copper or as crystallized copper
sulfate by solvent extraction.
•Extraction of zinc (and residual
iron) in preference to chromium and
nickel from the leach solution (with
subsequent zinc recovery by crystal-
lizing as zinc sulfate) by solvent
extraction.
•Selective oxidation of chromium ions
to form dichromate.anions.
130
-------
•Selective precipitation of chromium
from solution (using lead sulfate) as
lead chromate. The lead chromate can
subsequently be redissolved to pro-
duce a concentrated chromic acid sol-
ution with the regeneration of lead
sulfate.
•Solid-liquid separation of the pre-
cipitated lead chromate.
•Selective precipitation of nickel
ions from solution as nickel sulfide
or crystallization as Mi SO..
•Solid-liquid separation of the nickel
sulfide or NiSO..
•Recycle of the resulting purified
leach solution to the original leach
as make-up water. The unrecycled
final leach solution may be pre-
treated by ion exchange before
discharge.
Laboratory testwork supports the con-
cept that metal values can be separated
and recovered effectively and efficiently
from complex mixed metal sludge materials.
RESULTS
The experimental approach and philos-
ophy for the laboratory verification
studies include preliminary test of con-
cept by screening experiments; development
of a two-level factorial design matrix for
the experimental bench-scale studies; ex-
ecution of the studies in the design matrix
to establish which variables are most im-
portant and what the relative effect of
each particular variable is on the measur-
ed result; and subsequent use of the de-
sign matrix effects (by using the Box-
Wilson "steepest ascent" approach) to
optimize the selection of experimental
variables for further larger scale
testwork.
Large Scale Testwork
The objectives of the large scale test
program were: to size the unit operation
so that 45.4 kg (100 Ibs.) of sludge could
be treated per day; to test the unit oper-
ations to ascertain if effective and effi-
cient metal value extraction and recovery
could be'achieved; and'to determine what
chemical and mechanical problems might be
associated with treating approximately 200
liters of leach solution per day.
The test system consists of leach
vessels, settlers, a filter press, solvent
extraction mixer-settlers, chlorine or
electrochemical oxidizer, pH monitors and
controllers, precipitating vessels,
crystal!izers, and an ion exchange column.
A summary of the experimental program
and conclusions drawn from the testwork
are presented below. The discussion
follows the sequence of operations pre-
sented in the flowsheet in Figure 1.
The sulfuric acid leach operation is
effective in redissolving the metal values
(example starting sludge compositions are
presented in Table 1). The dissolution is
rapid and without control problems. The
leach is carried out in a single two hun-
dred seventy liter vessel. The conditions
required are well characterized, and
rather mild, i.e., one-half hour, 40-50°C,
sludge/liquid ratio of 0.8, acid content
to control pH in the range 0.5-1.5, and
agitation sufficient to suspend the par-
ti cul ate in the solution phase. Example
data are presented in Table 2.
The sludge dissolution is essentially
complete in less than one-half hour.
Therefore, the leach operation is not the
controlling step in the overall treatment
sequence. The leach unit operation is
capable of treating over a ton of sludge
per eight hour day. The filterability of
the Teach residue product is difficult, •
but may be overcome with the use of filter
a'ids. The filterability of a mixed leach
residue-jarosite product is rapid and
effective. Therefore, in most testwork
the jarosite precipitation process was
performed in-situ with the leach residue
solids.
The iron removal unit operation is via
the precipitation of potassium jarosite.
The precipitation process requires ele-
vated temperatures and relatively long
reaction times. Two hundred liters of
leach solution slurry can be treated in
six-eight hours. The jarosite process
allows iron to be removed from an acid
solution. The product is a crystalline
compound that has excellent settling and
filtering properties (see Table 3). The
iron removal process has been demonstrated
on .high iron sludge materials, i.e.,
131
-------
* 15-20% iron in the starting sludge solids.
This means that for these particular
sludges a significant quantity of leach
residue-jarosite solids is formed, e.g.,
11.6 kg of solids or 17.8 kg of wet mater-
ial. Therefore, the disposal of 17.8 kg
would be required instead of 45.4 kg or
approximately forty percent of the original
sludge weight. A significant quantity of
sludge material exists that has iron con-
tents much lower than the above values.
The jarosite process is also effective for
treating the low iron containing sludges,
e.g., two-four percent iron. The quantity
of leach residue-jarosite solids produced
from such sludge material would be rather
small, i.e., a two percent iron bearing
sludge would yield 5.6 kg of leach residue-
jarosite solid. This quantity of solids
translates, at 65% solids, to 8.6 kg of
disposable material instead of 45.4 kg or
approximately one-fifth the original sludge
weight. Jarosites are widely produced in
the zinc industry. They are deposited in
lined storage ponds. It is difficult to
state whether their heavy metal content
means that the jarosite should be consider-
ed a hazardous material, but even if that
is the case at least a significantly
smaller weight of material must be
considered for disposal.
High iron sludges do (low iron sludges
do not) present a problem for chromium re-
covery. Significant amounts of chromium
are lost when the jarosite precipitation is
performed. It is believed that the loss
can be minimized by maintaining conditions
such that chromium is not oxidized and the ,
pH is maintained below 2.5.
Mechanical control of the system is no
problem. Chemical control must be exer-
cised to ensure that the pH is maintained
in the range 1.8-2.5 and that the iron is
in the ferric form. Solid-liquid separa-
tion is effectively accomplished by allow-
ing the leach residue-jarosite to settle;
decanting the solution from the solids;
and pumping the small volume of remaining
slurry to a filter press.
The removal of copper is accomplished
by solvent extraction (SX) using LIX 622.
The extraction of copper from zinc, chrom-
ium, nickel, and aluminum is selective and
effective (>96% extraction per contact
stage). Copper contents of a few mg/liter
are achievable in two stages of contact and
one stage of strip. The pH of the solution
exiting the jarosite precipitation unit
operation need not be adjusted prior to
copper extraction.
The SX testrack is designed to treat
up to 200 liters of solution per day. The
design throughput is 500 cc/min. for each
phase. Most tests to date, have been per-
formed at 250 cc/min. Ten contact mixer-
settler units are available for copper SX.
Therefore, this unit operation is not the
slow step in treatment sequence, i.e.,
three concurrent streams could be treated
(each in three cells) at one time; at 250
cc/min., 360 liters could be treated per
day.
Large scale testwork has been conduc-
ted for up to six hours. Control of flow-
rate and interface levels is easily achiev-
ed and requires constant attention only
during initial loading of the system. Once
the system interfaces have been established
little operator attention is required. A
summary of experimental testwork for large
scale continuous testing is presented in
Table 4.
The removal of zinc is accomplished by
solvent extraction using D£EHPA. The ex-
traction of zinc from chromium and nickel
is selective. Ferric iron, aluminum, and
calcium are partially coextracted with the
zinc.
The extraction of iron (only between
0.2-0.6 gpl present) with zinc is desirable
because it provides a way of removing re-
sidual iron from the solution. The iron
once extracted into the organic phase is
not stripped by ^$04 but is stripped by
HC1 acid (4-6N). Zinc is stripped by H2S04
(200 gpl). Therefore, a means of bleeding
iron from the process stream is to load
iron and zinc into the organic phase, strip
the zinc by contacting with H2S04 (200 gpl)
followed by stripping the iron from the
organic by contacting with HC1 (4N). Both
strip solutions can be recycled until the
metal content is appropriate for recovery
of zinc as zinc sulfate monohydrate and for
disposal of iron as ferric chloride
solution.
Calcium is coextracted with zinc but
poses no problem because it precipitates as
gypsum in the H2S04 strip circuit. It can
be effectively filtered continuously from
the solution during solution recovery.
132
-------
Aluminum is co-extracted with zinc and
partially stripped,in the t^SO^. strip cir-
cuit. Its presence must be considered dur-
ing the crystallization of zinc sulfate.
The zinc SX testrack is the same de-
sign as used for copper removal. Ten SX
cells are available. The removal of 5 gpl
zinc requires four stages of extraction,
three stages of zinc stripping, and one
stage of iron stripping. Therefore, the
removal of zinc is the limiting step in
the present treatment process. Two hundred
liters can be treated at a flow rate of 400.
cc/min. for each phase in an eight hour
period. Some flexibility does, however,
exist by control of the extracting reagent
concentration in the organic and by chang-
ing the organic to aqueous ratio in the
system.
Control of flowrate and interface
levels is easily achieved and does not re-
quire constant attention once the initial
loading and interface levels have been es-
tablished; i.e., operator attention is
minimal. The system can be shut off and
restarted without difficulty. Chemical
control of pH is required in zinc extrac-
tion to achieve effective zinc removal.
Solution pH control is exercised by adjust-
ing pH after the first two stages of con-
tact. Temperatures in the range of 40-55°C
are desirable for rapid phase disengagement.
A summary of experimental testwork for
large scale continuous testing is presented
in Table 5.
Chromium removal is accomplished by
first oxidizing the chromium with chlorine
gas, then precipitating the dichromate ion
as lead chromate. Oxidation may alterna-
tively be accomplished by use of electro-
chemical cells. Both systems have been
shown to be effective in laboratory scale
test reactors. Large scale oxidation test-
work using chlorine has been performed suc-
cessfully; Table 6. Both chlorine oxida-
tion and electrochemical oxidation reactors
will be further investigated during Phase
II testwork.
A recycle system for stripping the
oxidized chromium from the leach solution
has been operated successfully: the solu-
tion is exposed to lead sulfate in an agi-
tated reactor; lead chromate precipitates;
the lead chromate product is crystalline
and dense and settles rapidly; the solution
essentially free of lead chromate solid is
pumped from the solids for further treat-
ment for nickel removal; the lead sulfate
solid is separated from the chromic acid
and is recycled to the lead chromate
precipitation reactor.
Nickel can be removed by sulfide pre-
cipitation. The reaction is rapid and near
quantitative. The pH is maintained in the
range 4-5 so hydrogen sulfide is not re-
leased. The solid product is readily fil-
terable. Quantitative removal of nickel
is not necessary because practically all
the final solution can be recycled to the
leach-jarosite precipitation unit operation.
Therefore, the addition of a deficiency of
sulfide (less than the stoichiometric re-
quirement for complete nickel removal) is
desirable so that all the added sulfide
ions are consumed. Then when the solution
is recycled to the acid leach step,
hydrogen sulfide gas will*not be formed.
An alternative nickel recovery unit opera-
tion is nickel sulfate crystallization.
The sequence of unit operations des-
cribed above results in appropriate metal
value recovery from mixed metal sludge
materials. Not only are metal values
recovered but there is a significant
decrease in the quantity of waste material
that must be treated for permanent disposal.
ACKNOWLEDGMENTS
This research was supported by a
Cooperative Agreement (R 809 305 010) be-
tween EPA-IERL and Montana Tech Foundation.
The financial support is gratefully
acknowledged.
133
-------
TABLE 1. STARTING SLUDGE MATERIAL BLENDED SAMPLE REPRODUCIBILITY
Composition
Cu
Fe
Zn
Cr
Barrel 5 (used
2.41
2.41
2.48
2.4310.05
8.26
6.57
6.03
4.41
4.05
5.8612.36
11
11
11
.33
.88
.65
1 1 .6210.29
19
18
17
17
16
.05
.06
.37
.16
.91
17.7111.34
8
8
8
8.
6
8
9
9
6
.40
.45
.75
53+0.22
Barrel
.15
.61
.99
.83
.96
8.3112.16
1.36
1.35
1.35
1.35+0.01 4.
in Solid (%)
Ni
Al
Cd
Cu
Pb
in kettle test)
5.08
4.08
5.08
99+0.91
18 (used in large
8.52
7.10
6.24
4.46
4.13
1.91
2.23
2.28
2.45
2.47
6.0912.43 2.27+0.35
4.05
4.15
4.55
4.25+0.20 0.
scale test)
2.66
2.81
3..13
2.77
2.58
2.7910.34 0.
0.39
0.41
0.41
40+0.01
0.04
0.08
0.12
0.11
09+0.03
1
1
1
.08
.00
.10
1.06+0.04
0
0
0
0
0
.31
.45
.44
.64
.64
0.5010.19
0.09
0.10
0.09
0.09+0.01
0.08
0.08
0.07
0.05
0.11
0.08+0.03
134
-------
TABLE 2. TYPICAL SULFURIC ACID LEACH OF MIXED METAL HYDROXIDE SLUDGE:
STANDARD CONDITIONS
Condition*
100
650
1,000
50,600
gm
gm
gm
gm
sludge
sludge
sludge
sludge
Fe
92
95
55
92
.0
.4
.8
.0
Cu
93.7
94.9
94.3
93.7
Metal
Zn
95
90
94
95
.9
.5
.2
.1
Extracted
Ni
95.9
97.8
85.0
95.9
(«
Cr
96.5
96.7
96.7
96.5
Cd
93.0
100.0
97.0
93.0
Al
89.9
95.7
96.0
96.9
*Standard Conditions: one-half hour leach; ambient temperature; sludge/liquid ratio =
0.8; acid content equivalent to weight of solids in sludge.
135
-------
TABLE 3. JAROSITE FILTRATION RATES
Feed Solids
Solids Loading
Rate (kg/m2/Hr.)
Cake Solids (%)
Present
Study
40-50
25-55
66-71
Japan* Canada*
50-55 20-30
80-100 40-50
78-80
75-78
*Data supplied by
Ingersoll-Rand
136
-------
TABLE 4. SUMMARY OF LARGE SCALE TESTS ON
SOLVENT EXTRACTION OF COPPER WITH
LIX 622
Condition
Copper extraction from
leach solution
Per-
cent
Sequential Series
One
Raffinate from 98.9
contact
(40 lit.)
Sequential Series
Two .
Raffinate from
contact
(60 lit.)
Sequential Series
Three
Raffinate from 98.0
contact
(20 lit.)
Sequential Series
Four
Raffinate from 96.9
contact
(90 lit.)
Sequential Series
Five
Raffinate from 99.0
contact
(160 lit.)
Copper content
in solution
Initial
1.37
94.4 0.39
2.32
3.89
3.05
Fi nal
(gpi)
0.017
0.022
0.047
0.120
0.030
137
-------
TABLE 5. SUMMARY OF LARGE SCALE TESTS ON
SOLVENT EXTRACTION OF ZINC WITH
D2EHPA
Condition
Sequential Series
Three
Raffinate from
contact
(20 lit.)
-Sequential Series
Four
Raffinate from
contact
(50 lit.)
Raffinate from
contact
(90 lit.)
Sequential Series
Five
Raffinate from
contact
(160 lit.)
Zinc extraction from
leach solution
Zinc content
in solution
Per- Initial Final
cent (gpl) (gpl)
97.4 5.70
97.8 5.89
98.8' 4.94
98.9 6.20
0.15
0.13
0.060
0.070
138
-------
TABLE 6. SUMMARY OF LARGE SCALE TESTS ON
CHROMIUM PRECIPITATION
Condition
Chromium removed
from solution
Percent chromium
content in
solution
Per- Initial Final
Cent (gpl) (gpl)
Sequential Series
Four
Starting solution
(10 liters)
Fi nal fi1trate
after 30 min.
exposure
Sequential Series
Five
Starting solution
(42 liters)
Final filtrate
after 30 min.
exposure
99.5
99.7
1.65
0.008
2.34
0'.007
139
-------
Recycle
Solution
Acid
SIudgo
Slurry
Decant
Settler
Wash .
11
Storage
FIItrate
COPPER SX
Loaded Organic
Aq
Strip Organic
Jaroslte
Cake
IRON REMOVAL
Organic
Pregnant Aqueous
Organic
Cu
EW
Strip Acid
Raffinate
To Storage then to Zn SX
COPPER REMOVAL
Figure 1. Flowsheet for Treatment of Mixed Metal Hydroxide Sludges.
140
-------
Copper
Raffinate
Storage
ZINC SX
Loaded Organic
Organic
Strip
Organic
Organic
Aq
Organic
Aq
Organic
Organic
Strip Acid
Raffinate
pH Adjust
(Bleed Stream
to Crystallization)
To Storage
then to Oxidation
ZINC REMOVAL.
Figure 1. (Continued).
141
-------
Zinc
Na2S
C12 PbSO.,
4 Solution
Raff I note
1
Storaga
Decant Decant |
•(• lit Recycle
1 1 Makeup
1 1 * Water
PreclpI- Preclpl-
tat Ion tat ion
Oxidation "\^ ^/^ \^ ^S~
Wash | Wash
Restdual
Solids
rn
i i
1 Filter Filter
Press | Press "~|
to
Initial
Loach
PbCrO, Cake NIS Cake
1
IX
CHROMIUM REMOVAL
Smal I
VoIume
Discharge
NICKEL. REMOVAL.
Figure 1. (Continued).
142
-------
OPERATIONS AT THE USEPA COMBUSTION RESEARCH FACILITY
Richard A. Carnes
U.S. EPA Combustion Research Facility
Jefferson, AR
INTRODUCTION
The safe disposal of this Nation's
industrial waste is the major goal estab-
lished by Congress in the Resource Conser-
vation and Recovery Act (RCRA) of 1976.
More specifically the USEPA has determined
that incineration of those wastes compat-
ible with thermal destruction is the
preferred disposal option. Although
incineration is the method of choice, its
proper implementation and continued safe
operation are required before it will be
accepted by local citizens or heavily
invested in by industrial generators or
waste disposal firms. To these ends the
USEPA Combustion Research Facility (CRF)
has established its mission and goals.
PURPOSE
The CRF has a multi purpose mission
in that it will assess the performance
capabilities of the rotary kiln/air pol-
lution control device presently installed
and will follow these capability analyses
to the planned liquid injection unit and
its appurtenances when they are fabricated.
It will be through performance testing
that the CRF will assist in forming the
framework of the USEPA policies and
regulations regarding thermal destruction
of industrial wastes as the disposal route
of choice.
The CRF will provide a physical plant
for the Agency to test the selection of
appropriate Principal Organic Hazardous
Constituents (POHCs) for specification in
permits and it will directly support the
identification of those critical and
perhaps lesser known operating variables
that contribute to the production of
Products of Incomplete Combustion (PICs).
The CRF has as an operating purpose the
integration of data generated at both the
laboratory and bench scale so as to
provide the Agency with a potential
"incinerability" matrix and identify those
waste components associated with thermal
environments which should be avoided to
prevent the formation of toxic products.
The daily operations at the CRF will
provide the Agency and the technical com-
munity as a whole a better understanding
of the complex thermal processes occurring
in an incinerator and the capability to
describe the thermal process so as to be
able to characterize and assess performance
in full-scale systems and to extrapolate
performance data from one waste to another
or from one technology design scale to
another, or even another design itself.
Finally, the CRF will serve as the
focal point for the USEPA for definition
of compliance monitoring and will provide
valuable insights on what must be moni-
tored so proper assurances are given that
143
-------
the incinerator is performing within pre-
selected ranges. The CRF will provide
data on sub-system evaluation leading to a
total system Destruction and Removal
Efficiency (ORE). In this way the CRF
will make an invaluable contribution to
the national permitting process, the
enforcement process, and to citizen
acceptance of incineration as an accept-
able neighbor.
As can be seen in Table 1, the
objectives of the research at the CRF are
varied.
TABLE 1. RESEARCH OBJECTIVES OF THE COMBUSTION RESEARCH FACILITY
Specific Objective
1 2
Process Seal ing
Operating and
Variable Design
Reliability
& Control-
abil ity
4
Destruction
of Various
Wastes
5
Spec ial
Test
Burns
Components
and
Subsystem
A - Determine Destruction
Mechanisms & Kinetics,
Identify PICs
(Unregulated
Substances)
B - Engineering Character-
ization of System
Operat ion
C - Effects of Waste Feed
Constituents &
Physical Properties &
of Feed System
Configuration
D - System Response to
Abnormal Operational
Modes
E - Permit Compliance
Testing
F - Determine Destruction
Efficiencies for
Regulated Waste
Materials &
Substances
X
X
X
X
144
-------
.PROBLEMS ENCOUNTERED
The conduct of research at the pilot
scale is a most difficult task. Being
involved in hazardous waste disposal at
this scale compounds the difficulty. In
the course of events that has transpired
at the CRF in getting operations ready
to conduct hazardous waste research there
are few problems that have not been
encountered.
Some problems that caused the largest
cost and time increments are
(1) Identification of a properly
tested and approved waste
container.
(2) Design and procurement of air
pollution control equipment
standard to the field.
(3) Installation and operation of
the on-line combustion gas
analysis system.
(4) Preparation of documents for
obtaining the necessary EPA
permits under RCRA.
FACILITY. STATUS
At the present time, the technology
that is available at the CRF is a pilot
scale rotary kiln with afterburner and an
air pollution control system consisting of
a variable throat venturi wetted elbow and
packed tower scrubber.
The present configuration of the kiln
is such that the feed waste is countercur-
rent to the primary burner. The burner is
fitted with an in-line Validyne Engineer-
ing Model 11-1-1000 propane meter. The
primary combustion air is monitored by an
in-line pitot. .The kiln temperature is
controlled using a shielded thermocouple
at the feed face. Additional monitoring
equipment on the kiln includes the combus
tion gases for both composition and flow
rate. Combustion gases are monitored for
CO, C02 and 02 using an on-line Bendix
Combustion Gas Analyzer. The organic
components can be determined by extractive
sampling using either liquid techniques
(EPA Method 5) or solid adsorber sampling
systems, such as the Volatile Organic
Sampling Train (VOST).
The afterburner temperature is
controlled using a shielded thermocouple
located at the center point of the chamber.
Additional monitoring facilities consist
of surface temperature, exit gas tempera-
ture, 'and provisions for determining
combustion gas composition and flow rate
(essentially the same as are available for
the kiln gases).
An Anderson 2000 air pollution
control system is fitted to the exit of
the afterburner chamber by a refractory
line elbow. Here the high temperature
gases are contacted with a variable throat
wetted venturi operating in the range of
30 inches of water gravity. The gases are
cooled and large particulate (if any) is
removed. The gases are then passed
through a packed spray column and a two
stage mist eliminator. Thus far, results
show the system performing very well
regarding halogen acid neutralization and
particulate removal.
The Bendix Combustion Gas Analysis
system is utilized to continuously monitor
CO, C02, and 02 in the kiln transfer
duct, the afterburner and the stack.
Temperature is monitored at more than
35 locations throughout the system. These
range from routine maintenance tempera-
tures to those in the afterburner and in
the stack. Velocity measurements are
taken at the head of the kiln and the head
'of the afterburner. Combustion gas flow
measurements ate taken in the kiln exit,
afterburner exit and 'stack as is moisture
content and the chlorine and HC1 content
in each of the three places.
145
-------
The CRF has in place an approved
quality assurance (QA) program that is in
complete agreement with all aspects of the
required program units. The quality
control (QC) aspects of operation also
conform to EPA requirements and are an
integral part of all analytical operations.
A detailed health and safety program has
been established at the CRF and is part of
daily operations. The general require-
ments that have been placed on this pro-
gram are: (a) protection of all personnel
who will be responsible for the operation
of the facility, (b) protection of the
surrounding area, including the community
and environment, and (c) protection of the
facility and its equipment.
RESULTS
Thus far experiments have been con-
ducted and samples collected during the
combustion of propane (background blank),
toluene (feedstock for chlorobenzenes),
hexachlorobenzene (HCB), and trichloroben-
zene (TCB). Table 2 presents the experi-
mental conditions for the experiments.
TABLE 2. HCB AND TCB EXPERIMENTAL TEST SERIES
EXP.
NO.
1.
2.
3.
4.
5.
POHC(l)
IDENTITY
Propane
Propane
Toluene
Propane
Toluene
Propane
Toluene
Propane
Toluene
TAB
1093
1093
982
871
871
(°C)
879
871
871
760
760
FEED RATE SAMPLING ,„.
(kg/hr) POSITIONS^'
-59 (a),(b)
2,43
2.43 (a)!(b)
~ 2.43 (a) lib)
2.43
REMARKS
Background
Background
Background
Background
Failure Mode-AB
burner shut off
6.
7.
8.
9.
Propane
Toluene
HCB
Propane
Toluene
HCB
Propane
Toluene
HCB
Prdpane
Toluene
1,2,4-TCB
1093
982
810
1093
871
760
760
871
2.43
0.10
2.43
0.10
^-46
2.43
0.10
2.43
0.10
(c)
(c)
(c)
(c)
2 min. every 15 min.
HCB fed as solution
in Toluene
HCB fed as solution
in Toluene
HCB fed as solution
in Toluene
1,2,4-TCB fed as
solution in toluene
(continued)
146
-------
TABLE 2 (continued)
EXP.
NO.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
(1)
(2)
POHC(1)
IDENTITY
Propane
Toluene
1,2,4-TCB
Propane
Toluene
1,2,4-TCB
Propane
Toluene
1,2,4-TCB
Propane
Toluene
1,2,4-TCB
Propane
Toluene
1,2,4-TCB
Propane
Toluene
1,2,4-TCB
Propane
Toluene
1,2,4-TCB
Propane
1,2,4-TCB
Propane
1,2,4-TCB
Propane
1,2,4-TCB
Chemical(s)
Position (a)
TAB
CO
1093
1093
1093
1093
982
871
871
1121
1121
1121
fed to
= kiln
-------
Selected DRE's for the experiments using
the chlorobenzenes are presented in Table 3.
TABLE 3. RESULTS OF SELECTED CHLOROBENZENE EXPERIMENTS
Exp.
NO:
8
7
6
9
10
11
17
18
19
TAB
CO
810
982
1093
1093
1093
1093
1121
1121
1121
rko
760
760
871
871
871
871
788-760
732-704
677-649
Feed Rate
(kg/hr)
0.1
0.1
0.1
0.1
0.25
0.5
1.0
*
*
*
Emission Rate
(qm/hr)
2. 23xlO-3
2.03x10-3
4.41xlO-3
3.47x10-3'
1.66xlO-3
0.50x10-3
0.31x10-3
0.25x10-3
0.67x10-3
0.86x10-3
1.90x10-3
2.74x10-3
TOTAL EMISSION
104.6 MQ
342.3 Mg
0.305 grams
ORE
%
99.998
99.998
99.996
99.997
99.998
99.999
99.9997
99.9999
99.9998
99.9998
99.9998
99.9997
DE%+
99.9991
99.992
97.2
+ DE = Destruction Efficiency as opposed to ORE.
* The 1,2,4-trichlorobenzene was fed at a rate of 66.7 gm per minute for an elapsed time
of 30 seconds. Thus a total of 33.35 gms was fed in each of the experiments. Assuming
that all feed was volatilized and removed from the kiln during this time, and taking the
appropriate (measured) flow rate for each of the conditions, we compute DE as follows:
DE = 1 - £-
C0
33 35
where C0 = '- and would be the concentration in the kiln flue
kiln flow
gas If DE=0.
Kfln flow for experiment 17=5.70 (DSCMM), 18=6.28 (DSCMM), 19=6.02 (DSCMM)
148
-------
In addition to the aforementioned
sequence of experiments, the ability to
accurately determine flow rates and spe-
cific concentrations of combustion pro-
ducts allows an estimate of the heat
release from the stoichiometric equations
for the oxidation of propane:
Researchers at the CRF haye'compared
the total heat release against four (4)
methods of measurement. The results of
these measurements are presented in
Table 4.
TABLE 4. COMPARISON OF HEAT RELEASE (MMBtu/hr.)
Heat Release From
Stack Gas Analysis
Heat Release Based
On Inline Gas Meters
Heat Release Based on Combustion
• Gases and Surfaces Losses
C02
1.94
H20
1.87
Propane Flow
1.97
1.97
The excellent agreement between the
various estimates of the heat release
lends credence to the various measurement
techniques used.
FUTURE OF CRF
Presently ,the CRF is anticipating
receipt of a final Part B RCRA incinerator
permit to be issued by the State of
Arkansas under Phase II authority granted
by the USEPA. Until such a permit is
received operations at the CRF shall be
limited to commercially available chemi-
cals and synthetic mixes. Once permission
is received the CRF is geared to secure
several candidate waste streams for active
investigation.
Research shall be directed toward a
better understanding of residence time.
It has often been said that the operation
of an incinerator is governed by the
three T's - time, temperature and _turbu-
lence. The Time spent (its residence
time) in an afterburner is normally deter-
mined by the prescription, Equation (1).
' (1)
where V is the chamber volume and Q the
throughput of combustion gases measured at
the exit temperature of the chamber.
There are several problems with this. The
gas temperature within the chamber can be
expected to be significantly above that at
the exit and therefore to move the same
mass of gas through the unit area requires
a higher velocity. Also, the turbulence
within the chamber must result in not all
elements of the combustion gas remaining
within the chamber for the same time
period. For these reasons the CRF is
presently undertaking a detailed investi-
gation to define tr as some sort of
average time within the high temperature
chamber. It was recognized that helium
dflution could be used for an actual
measurement of residence time in addition
to the above mentioned utility for high
temperature flow measurements. Instead of
a steady input flow as required by the
former measurement, a measure of the time
149
-------
of arrival of hel inn introduced as a sharp
pulse will then allow a measurement of the
residence time within the chamber. Pre-
liminary results of the residence time
distribution measurements have been very
exciting and it is anticipated that these
experiments will be completed in the near
future. Figure 1 is a schematic of where
helium will be introduced in the system
and where analysis will occur.
FIGURE 1
PORT
GO <— WATER TRAP •*
WPUTFUEL
INPUT FUEL
OUTPUT GAS
WPUT FEED
He
PORT
The concept of hot zone sampling has
been investigated by others but not to the
extent as called for at the CRF. At the
present time there is a stationary probe
located in the transfer line from the kiln
to the afterburner and in the duct from
the afterburner to the scrubber. These
probes are in zones that are non-laminar
in flow so that mixing might be induced
and thus concentration uniformity of
chemical species would occur. There is no
attempt to sample iso-kinetically. At the
present time samples are being collected
using the solid absorbents XAD-2 and Tenax
in which further sample manipulation
follows in the laboratory. Results of
these efforts will be forthcoming in
separate CRF documents, some of which will
be submitted to referred journals.
Other areas in which the CRF is
conducting investigations are:
o Chlorine balance in the scrubber
water
o Fixed vs_ Travers ing sampling in
the stack
o Flame out failure mode experiments
o Sub-system destruction/removal
efficiency calculations
o Product(s) of incomplete combus-
tion analysis and prediction of
species
o Correlation of pilot scale data to
lab and bench scale results
o Evaluation of modified solid
absorbent clean-up procedures.
In the future several reports and
journal publications will transmit the
results and significance of these and
other studies at the CRF.
In an attempt to assure that there
exists a high degree of coordination
between the several IERL incineration
research programs it has been determined
that each should study several of the
synthetic waste streams referred to as EPA
"Soups." This approach has been taken to
obviate the variations that can be ex-
pected in actual industrial wastes and is
expected to yield data that will allow
direct intercomparison among the several
programs. This procedure, in addition to
generating information of direct value
itself, should also assist in understand-
ing the means by which laboratory data
should be extrapolated to pilot and full
scale technology. The protocol which is
herein presented is designed to extract
150
-------
the maximum of information from a series
of incineration tests at the CRF on the
first of these soups.
The experiments described below are
anticipated to yield data directly applic-
able to the following:
o A determination of the effects of
afterburner temperature, residence
time and excess air on the thermal
stability of the compounds that
make up the EPA soup #1
o The effect of kiln conditions on
the OE of the afterburner
o The effect of kiln conditions on
the production of PICs from the
soup feed,
o The effect of feed rate of the
soup on both ORE for the compo-
nents of the soup and on the
production and stability of PICs
o The effect of variations of RTD on
the observed ORE for the after-
burner
o Development of thermal stability
data that will allow comparison to
other technologies
o Investigation of scrubber holdup .
effects
o Obtain operational experience on
the use of'the VOST system for
sampling and analysis of POHCs
(and PICs) which are suitable for
purge and trap analytical methods.
The philosophy that has underlain the
CRF operations from the beginning of the
program has been that all experiments
should be conducted from a technically
conservative viewpoint. That is to say,
every effort should be expended to design
experiments that would yield useful data
while at the same time conducting the
tests so as to have minimal impact on the
facility and the immediate environment.
This attitude will be considerably im-
pacted when the HEPA/carbon filtration
system is in place since use of the system
will always ensure that emissions are
within permissible-1imits. The experi-
mental protocol suggested for the soup
experiments is presently dictated by this
conservative attitude which requires
beginning with low feed concentrations and
maximum thermal conditions within the
incinerator. As it is shown that no more
than permissible emissions are found under
these severe conditions the conservative
restrictions are gradually relaxed. At
each step the emission levels determine
the advisability of proceeding to the next
less severe operations set. This process
continues until either the entire set of
experimental conditions has been explored
or until a set of operational parameters
has been attained for which non-permitted
conditions are found.
The proposed set of test conditions
will be carried out using the EPA soup as
indicated below:
Soup components:
carbon tetrachloride (CC14);
trichloroethylene (02^013);
monochlorobenzene (C5Hr;Cl); and
Freon 113 (C2Cl3F3) in a
toluene matrix.
It is proposed that several solutions be
made up of this combination as follows:
151
-------
Feed Mix BI (in the proportions as follows)
Ce^Cl 93.5 ml corresponding to 31.5 gm
C2H3C13 68-° ml corresponding to 58.0 gm
CC14 62.7 ml corresponding to 92.2 Cl2
C2C13F3 66-° ml corresponding to 37.4 gm Cl2 + 30.4 gm
C7H8 2500.0 ml
TOTALS 2.79 liters containing 219.1 gm Cl2 + 30.4 gm
Feed Mix 62 (in the proportions as follows)
467.5 ml corresponding to 157.5 gm
340.0 ml corresponding to 290.0 gm
CC14 313.5 ml corresponding to 461.0 gm
F3 330.0 ml corresponding to 187.0 gm Cl2 + 152 gm F2
1349.0 ml
C7H8
TOTALS
2.80 liters containing 1095.5 gm Cl2 + 152 gm F2
It is noted that Mix BI has a
heating value of 114 MBtu/gal whereas Mix
83 has a heating value of approximately
62 MBtu/gal.
The feed rates to be used for both
Mixes BI and 62 will be 2.84 1/hr
(0.75 gal/hr).
The actual proposed experimental matrix is
shown in Table 5.
TABLE 5. EPA SOUP - EXPERIMENTAL MATRIX
Experiment
A!
A2
AS
A4
AS
AS
A?
AS
Ag
AID
TV
N
1600
1400
1200
1200
1600
1400
1200
1200
1400
1400
&
2100
1800
1600
1400
2100
1800
1600
1400
2000
2000
Feed
Ql
D*|
Bl
BI
B2
B2
B2
B2
B2
B2
Sampl ing*
K
4
4
4
4
4
4
4
4
2
4
A
2
2
2
2
2
2
2
2
2
2
Stack
2
2
2
2
2
2
2
2
2
2.
Slowdown
1
1
1
1
1
1
1
1
1
1
*Integer refers to the number of samples to be taken.
152
-------
Experiments Ag and AIQ will be fail-
ure mode experiments. In experiment Ag
the afterburner propane feed will be shut
off for 2 minutes out of every 15 minutes.
AIQ W1'H involve the same procedure for
the kiln burner.
In the event that the high concentra-
tion feed stock, 62, shows compliance
under all conditions, it will be appropri-
ate to conduct an additional series of
experiments using Mix 62 but at a feed
rate of 5.6 1/hr (,f.5 ga/hr) at conditions
s imilar to AS and AS-
In all cases (sets of parameters in
Table 1) the kiln residual pressure will
be adjusted to -0.05 inches of water,
gauge, "in order to keep the excess air
approximately constant.
CONCLUSIONS
The following conclusions, may be
drawn from recent studies on the incinera-
tion characteristics of selected chlor-
inated benzenes in toluene solution:
o The CRF incinerator can produce
the required ORE of greater than
99.99% for the selected POHCs.
o The procedures for hot zone sampl-
ing that have been implemented
appear to yield consistent results.
o Comparison of absolute quantities
of several chlorinated benzenes to
soot weight in kiln samples shows
ri£ correlation.
o The behavior of the afterburner at
low injection rates is apparently
dependent on the presence of PICs
that are derived from the POHC.
o There is apparently a direct
relationship between DE/\ and the
injection rate at low injection
rates.
As expected, the existence of a
residence time distribution in the
afterburner results in the failure
of first order kinetics to describe
the destruction of hexachloroben-
zene and 1,2,4-trichlorobenzene.
The proposed residence time dis-
tribution and the method of its
measurement apply quite well to
the CRF incinerator.
The use of helium dilution for the
measurement of combustion gas flow
yields consistent results.
There apparently exists a strong
interrelationship between 1,2,4-TCB
pentachlorobenzene and HCB, a
relationship suggesting that any
two will be PICs derived from the
third.
The proposed scrubber blowdown
model appears to be directly
applicable to the CRF incinerator.
The poor turndown of the Iron
Fireman burners 1imits the feed
rate that can be used, indicating
the need to replace these burners.
The castable refractory has stood
up remarkably well even under the
severe thermal stress of intermit-
tent operation. Engineering
tests conducted without POHC feed
always show the POHC and PIC
from previous analytical tests
which were conducted with POHC
feed.
RECOMMENDATIONS
The successful demonstration of 3
method of residence time distribu-
tion measurement in the CRF after-
burner opens the door to an on-
1ine measurement of retention time
which could then become a monitor-
ing and control method.
153
-------
The successful application of
simple models to such aspects of
the CRF incinerator as the after-
burner retention time and the
blowdown discharge suggest that
modeling of other aspects of the
system can lead to enhanced under-
standing of the incineration
process and to better control
thereof.
Attempts should be made to allow
the measurement of mixing in the
afterburner and the effects of
such mixing on the temperature
distribution as well as on the
retention time distribution and
thereby on the performance of the
incinerator.
The fact that a new feed face
configuration has been designed
with the capability of pumpable
fluids, solids and containerized
solids feed indicates that by the
time the RCRA permit has been
issued, the CRF will be in posi-
tion to study complex or problem
wastes.
ACKNOWLEDGEMENTS
The author wishes to acknowledge the
on-site contractor, Versar, Inc., for
their diligent performance under very
trying conditions; in particular,
Dr. Frank C. Whitmore, Program Manager;
Mr. Robert Ross, Senior Chemist; Mr. Tony
Reynolds, Senior Engineer; and Dr. Fred
Fowler, Health and Safety Officer.
154
-------
SAMPLING METHODS FOR EMISSIONS FROM HAZARDOUS WASTE COMBUSTION
Larry D. Johnson
Industrial Environmental Research Laboratory
US Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Incineration as a means of hazardous waste disposal has received increasing
attention during the last several years. Since it is necessary to characterize
the stack emissions as well as the waste in order to determine how well the unit
is operating, research into adequate methods of sampling and analysis has been
necessary. This paper presents an overview of the sampling methods which are
resulting from current and previous research programs and which are being recommended
to EPA regulatory programs to EPA engineering research and development projects,
and to interested parties in the industrial community. The methods discussed
are generally applicable to incineration and also to processes closely related
to incineration, such as co-firing of waste in industrial boilers, and burning
of contaminated heating oil.
Although methods for inorganic hazardous compounds are very briefly outlined, the
primary emphasis of the paper is on organic compounds which are likely to be chosen
as principal organic hazardous constituents (POHCs) for a trial burn. Methods
receiving major attention in this paper include: the Modified Method Five Train (MM5)
which includes an XAD-2 sorbent module, the Source Assessment Sampling System (SASS),
the recently developed Volatile Organic Sampling Train (VOST), and assorted containers
such as glass bulbs and plastic bags.
INTRODUCTION
Incineration as a means of hazardous
waste disposal, especially for organic
materials, has received increasing atten-
tion during the last several years.
Although there are many measurements that
may be made in order to characterize the
operation of an incinerator, virtually all
of them require sampling and analysis of
the flue gas a's well as the waste feed
material. Although sampling and analysis
of incinerator emissions have been accom-
plished in the past, a number of advances
in the state of the art have been made
within the last few years. The range of
chemical compounds likely to be of interest
has also widened considerably.
A program to'provide and document
adequate methods for characterization of
hazardous waste incinerators is being
carried out by EPA's Office of Research
and Development (ORD). A reference
document(l) for use by ORD incineration
engineering R&D programs, EPA's Office
of Solid Waste and Regional Offices,
and the regulated community has been
released, following peer and policy
review. The methods in Reference 1 are
state of the art, and are scientifically
sound. Extensive programs of careful
and thorough validation, which are under-
way at present, will greatly benefit the
users of these methods. In the meantime,
the methods to be described are best used
by those with considerable experience in
stack testing and chemical analysis: as
much expert advice as possible should be
sought when preparing or carrying out an
incineration characterization project.
This paper presents an overview of
the methods proposed in Reference 1 with
special emphasis on stack sampling of
organic compounds likely to be chosen as
principal organic hazardous constituents
(POHCs).
155
-------
Stack Sampling Methods
Although numerous sampling methods
might be applied to a given incinerator,
this discussion will concentrate on four
approaches that are being utilized to the
greatest extent, not only because of their
generally high performance, but because of
their flexibility and usefulness for a
wide range of materials.
The Source Assessment Sampling System
(SASS), Figure 1, was developed for envi-
ronmental assessment programs and is still
the train of choice when large amounts of
samples are necessary for extended chem-
ical analysis or biotesting. The SASS
includes: cyclones for particle sizing, a
glass or quartz fiber filter for fine
particle collection, a sorbent module for
collection of semivolatile organics, and
impingers for collection of volatile
metals. The SASS operates at 110 to 140
Lpm (4 to 5 cfm) and is usually operated
long enough to collect 30 m3 of flue gas.
It is possible, but not convenient, to
traverse a stack with an SASS, so it is
usually operated at a single point in the
stack under pseudo-isokinetic conditions.
For further details as to the meaning and
potential ramifications of this mode of
sampling, see Reference 2.
Potential corrosion of stainless steel
in the sorbent module of the SASS has
prompted development of glass sorbent
modules which appear to perform adequately.
One of these is described in Reference 3.
The Modified Method Five Train (MM5) is
conceptually very similar to the SASS but
operates at a lower flow rate, usually 14
to 28 Lpm (0.5 to 1 cfm). The MM5, shown
in Figure 2, does not include particle siz-
ing cyclones, and is usually constructed of
glass rather than stainless steel. The
MM5 results from a very simple modification
of any of the commercial sampling trains
available which conform to the requirements
of EPA Method 5. The sorbent module with
cooling capability is simply inserted
between the filter and the first impinger.
The sorbent module must be positioned
vertically so the gas and any condensed
liquids flow downward through it.
The sorbent of choice for most sampling
jobs for both the SASS and MM5 is XAD-2.
For further discussion of the reasons be-
hind this choice, as well as sorbent
module placement in the train see
Reference 2.
Because of its more convenient
size and ready availability,
the MM5 is usually chosen over
the SASS for incinerator sampling
unless larger samples are needed
for lower detection limits or
extensive analysis requirements.
Either the SASS or MM5
provides collection ability for
particulate material, acid
gases such as HC1, gaseous
metal compounds (if appropriate
collection liquids are chosen),
medium boiling organics (b.p.
100°C to 300°C), and high boiling
organics (b.p. greater than
300°C). Organics with boiling
points between 100°C and 120°C
require individual attention
during the sampling planning
stage and may require decreased
sampling times to prevent volumetric
breakthrough. Volumetric break-
through is related to the migration
rate of sorbed material through
unsaturated sorbent beds. For
further discussion of this
important concept, see Reference
2.
Considerable field experience
has been gained with the SASS
and MM5 trains even though
formal validation procedures
have not been completed. Much
of the confidence in these
trains' ability to collect
organics rests on knowledge of
the behavior of sorbents with
respect to qollection and recovery.
Reference 2 discusses this
point in more detail. A particularly
effective demonstration of
these trains' performance on a
dioxin injected into combustion
gas is given in Reference 4.
Similar research with other
compounds is underway, and a
validation project has begun.
The MM5 and SASS are not
generally quantitative collection
trains for organics with b.p.
less than 100°C. For these low
boiling compounds, the recommended
methods are plastic sampling
156
-------
bags, glass sampling bulbs, or the newly
developed volatile organic sampling train
(VOST). The ambient air at many inciner-
ator sites exhibits relatively high levels
of volatile organics. This greatly
increases the difficulty of obtaining an
uncontaminated sample of low concentration
and requires a great deal of care as well
as adequate blanks. All of the above
methods have shortcomings, but they are
much less severe than the .shortcomings and
limitations of alternate approaches <2) .
The VOST is shown in Figure 3. This
train and the analysis approach applied to
the samples resulting from its use were
developed in order to address stack con-
centrations as low as 0.1 ng/L. A sorbent
tube containing 1.6 g of Tenax is position-
ed early in the train in order to remove
organics from the gas and liquid stream as
soon as possible. A second sorbent tube
containing 1 g of Tenax and 1 g of charcoal
follows the condensate collector in order
to act as a backup in case of breakthrough.
The charcoal provides added stopping power
for the very low boilers such as vinyl
chloride. The train was designed to use
six pairs of sorbent tubes sequentially,
each operating for 20 minutes at 1 L/min.
The ability to concentrate the organics
from all six sets of tubes onto one ana-
lytical tube and subsequently heat desorb
into a GC or GC/MS makes it possible to
achieve very low levels relative to the
stack gas concentration. For higher stack
gas concentrations, it appears to be pos-
sible (perhaps even necessary) to operate
the VOST at lower flow rates and longer
sampling times. There is no fundamental
reason why this piece of equipment should
not be useful in a number of operation
modes as long as excessive volumes of gas
are not pulled through a single tube,
since low boilers break through the sorbent
after relatively low volumes.
The development and use of the VOST has
been discussed in several publications(5,6,
7). Reference 7 is particularly useful to
users of the train. Preliminary evidence
that the train is effective is given in
these references, and an EPA validation
.project for the VOST is underway.
Various types of plastic sampling bags
have been used in the past, with very
mixed results. It is quite possible to get
good results with this approach, but it is
essential that the sampling and storage
characteristics of the specific POHC,
relative to the specific types of bags to
be used, be well known. For example,
organics such as alcohols usually exhibit
poor storage characteristics in bags (°> .
It is also essential that field blanks be
included in the sampling strategy, since
all known methods for volatile organics
from incinerators are subject to potenti-
ally severe contamination in the field and
during transit.
When sampling relatively high con-
centrations of volatiles, glass sampling
bulbs with secure seals may be the best
choice. Although somewhat inconvenient
and lacking in sample concentrating ability,
the glass bulbs do show better sample stor-
age characteristics than plastic bags (8>.
In addition to the methods discussed,
special procedures may be necessary for
certain pollutants. Formaldehyde is an
example of compounds that require special
handling C*). Reference 1 gives guidance
in such cases. A number of other less
satisfactory general sampling approaches
exist, but have not been recommended. For
further discussion of these techniques and
their shortcomings, see Reference 2.
Summary
Sampling methods have been discussed which
are generally applicable to incineration
and to processes closely related to incin-
eration, such as cofiring of waste in
industrial boilers, and burning of contami-
nated heating oil. Although some of the
methods are relatively new and all require
a great deal of care and attention, it is
possible to produce excellent results
through their application.
157
-------
REFERENCES
1. Harris, J.C., Larsen, D.J., Rechsteiner,
C.E.- and Thrun, K.E., "Sampling and
Analysis Methods for Hazardous Waste
Combustion," EPA-600/8-84-002, PB84-
155845, February 1984.
2. Johnson, L.D., and Merrill, R.G., "Stack
Sampling for Organic Emissions,"
Toxicological and Environmental Chemistry,
6, 109 (1983)
3. Berg, S., Williamson, A.D., and Miller,
B.C., "Design and Fabricate a Corrosion
Resistant Organic Sorbent Module,"
Appendix D, in "Modified Method 5 Train
and Source Assessment Sampling System
Operators Manual" Draft Final Report,
Task 16, Contract 68-02-3627, August 1983.
4. Cooke, M., DeRoos, F., Rising, B.,
Jackson, M.D., Johnson, L.D., and Merrill,
R.G., "Dioxin Collection from Hot Stack
Gas Using Source Assessment Sampling
System and Modified Method 5 Trains - An
Evaluation," Presented at Ninth Annual
Research Symposium on Land Disposal,
Incineration, and Treatment of Hazardous
Waste, Ft. Mitchell, KY, May 1983.
5. Jungclaus, G.A., Gorman, P.G., Vaughn,
G., Scheil, G.W., Bergman, F.J., Johnson,
L.D., and Friedman, D., "Development of
Volatile Organic Sampling Train (VOST),"
Presented at Ninth Annual Research
Symposium on Land Disposal, Incineration,
and Treatment of Hazardous Waste.
Ft. Mitchell, KY May 1983.
6. Jungclaus, G.A., Gorman, P.G., and
Bergman, F.J., "Sampling and Analysis
of Incineration Effluents with the
Volatile Organic Sampling Train (VOST),"
In Proceedings: National Symposium on
Recent Advances in Pollutant Monitoring
of Ambient Air and Stationary Sources,
Raleigh, NC, May 1983, EPA-600/9-84-001,
January 1984.
7. Hansen, E.M., "Protocol for the Collect-
ion and Analysis of Volatile POHCs
Using VOST," EPA-600/8-84-007, PB84-
170042, March 1984.
8. Thrun, K.E., Harris, J.C., and Beltis,
K., "Gas Sample Storage," EPA-600/7-
79-095, PB 298-350, April 1979.
Beltis, K.J., DeMarco, A.J., Grady, V.A.,
and Harris, J.C., "Stack Sampling and
Analysis of Formaldehyde," Presented at
Ninth Annual Research Symposium on Land
Disposal, Incineration, and Treatment
of Hazardous Waste, Ft. Mitchell, KY.,
May 1983.
158
-------
1
OT.
o>
v>
$
o
co
159
-------
c
2
•*-•
tn
•a
o
•g
CM'
£
160
-------
Ifl
3
to
O
c
'(0
O)
(O
S"
O
_a>
1
O
n
£
§>
u.
161
-------
INCINERATION OF HAZARDOUS WASTE IN POWER BOILERS:
EMISSIONS PERFORMANCE STUDY RATIONALE AND TEST SITE MATRIX
Robert A. Olexsey
Incineration Research Group
Energy Pollution Control Division
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45261
INTRODUCTION
Purpose
The purpose of this document is to
summarize the background, rationale, and
current study approach for the U.S.
Environmental Protection Agency (EPA)
research and assessment program relating
to the incineration of hazardous waste in
Industrial boilers. The overall objective
of the boiler program is to evaluate the
technical and environmental acceptability
of the use of hazardous wastes as fuels in
industrial boilers, to aid in an Agency
determination on the need for the nature
of regulations of this practice under the
Resource Conservation and Recovery Act
(RCRA).
Background
Under RCRA, regulations and specific
process performance standards have been
developed and promulgated governing the
issuance of permits for the operation of
incinerator facilities for hazardous waste
control (January 23, 1981). These
standards, however, do not currently apply
to the use of combustible hazardous wastes
as fuels in,energy recovery operations
such as power boilers, process heaters,
and high temperature industrial processes
(cement kilns, steel furnaces, etc.).
The use of hazardous waste as fuels
in power boilers is known to be a wide-
spread practice. As much as 19 million
metric tons of hazardous waste may be
disposed of this way. EPA, however, has
chosen to defer a regulatory decision on
this practice because specific data on the
extent of the practice were not available
(i.e., number of sites, type of wastes,
operating conditions), nor was there data
on the destruction efficiency or the
emissions performance of boilers employing
hazardous waste as a fuel supplement.
Other factors in this decision are docu-
mented in the Federal Register, Volume
45, No. 98, May 19, 1980.
Problem Definition
While an initial decision was made to
not regulate hazardous waste incineration
in boilers, it was also recognized that a
potentially significant loophole existed
in the regulatory structure. The same
combustible hazardous waste which could be
incinerated only under strict RCRA regula-
tory standards at the 300+ incineration
facilities in the United States could also
be incinerated as a fuel supplement in any
of the over 400,000 boilers in the United
States with virtually no environmental
controls, save those involving the storage
of the wastes and their transport to the
boiler. The EPA, through the Office of
Solid Waste (OSW) and the Office of
Research and Development (ORD), initiated
a program of research and performance
testing to evaluate the engineering and
environmental acceptability of hazardous
waste incineration in boilers and to
determine the need for and nature of
regulatory standards to control the
practice.
162
-------
Because of the large number of
facilities, boiler types and operating
conditions, and wastes which potentially
need to be evaluated with limited resources,
a considerable amount of thought is
necessary to carefully select a research
and performance testing approach which
optimizes the usefulness of the informa-
tion collected.
The purpose of the balance of this
document is to summarize the stepwise
process, important assumptions, and
rationale which form the basis for
the boiler testing, program now underway.
Limiting Assumptions
With limited funding available for
field testing, any such test program must
be directed toward obtaining only essen-
tial information in an economical manner.
A test program that set out to attack the
450+ listed RCRA hazardous wastes across
the full spectrum of boiler types in an
unordered fashion would lack credibility
and accomplish little in resolving the
issue of whether boiler co-disposal of
hazardous waste can be an acceptable
practice. In 1975, there were about
400,000 industrial boilers of all sizes
and types in operation. In order to be
realistic, the test program must be
directed toward boilers that- are actually
used to burn wastes, and combinations of
wastes and boilers that truly represent
test cases and will allow generalizing
extrapolations to be made across the
entire waste/boiler matrix.
Some simplifying assumptions can be
made:
1. Each owner/operator of an industrial
boiler will use his boiler only for
disposal of in-plant generated wastes and
will not likely accept waste from any
other plant.. The logic behind this
assumption is that the primary advantage
of boiler disposal (aside from energy
conservation) is that the practice is
exempt from RCRA regulations if the waste
is not transported, stored, or disposed of
off-site. Any facility which accepts
wastes that are not produced onsite must
be permitted as a Treatment, Storage and
Disposal Facility (TSDF) under RCRA. To
accept "imported" wastes could jeopardize
the "exempt" status of the boiler facility.
2. Most hazardous wastes are generated
by industry. This will exclude the large
number of "industrial" size boilers that
,are located at commercial and institutional
facilities such as shopping centers and
hospitals.
3. It is unlikely that wastes will be
burned in boilers smaller than 10 MM
Btu/hr. This size (7,000 Ib/hr of steam,
1.0 megawatts of electricity) is probably
the minimum size at which combustion units
can be retrofitted to accept significant
quantities of wastes without negatively
impacting the thermal operating charac-
teristics of the units. Boilers below
this size range are generally natural-gas
or oil-fired firetube or cast iron units
used for space heating. The 10 MM Btu/hr
cutoff point is also convenient in the
most boiler inventory reports (NEDS, EADS,
etc.) do not report on boilers below that
capacity. There are roughly 43,000
industrial boilers in the United States
with capacities larger than 10 MM Btu/hr
(1).
4. Most hazardous wastes that will be
burned in boilers are generated by the
organic chemicals industry. Table 1
describes the number of boilers with
capacities larger than 10 MM Btu/hr for
the top 10 hazardous waste generating SIC
codes (2). As can be seen from Table 1,
the largest hazardous waste generation is
from the chemicals industry. The chemicals
industry also has a large number of
boilers. Therefore, the opportunity for
disposal of hazardous wastes in boilers is
greatest in the chemicals industry. The
primary metal industries generate about 10
percent of the hazardous wastes, but
wastes from this industry are less likely
to be combustible than are wastes in the
chemicals industry. The largest sources
of organic hazardous wastes that are
available to boilers for disposal are the
chemicals, petroleum refining, and paper
industries, in that order.
Through the application of these
limiting assumptions, the universe of
industrial boilers of concern can be
reduced to less than 5,500. However,
these 5,500 boilers are accessible to
about 70 percent of total hazardous waste
generation and over 95 percent of all
potentially combustible hazardous wastes
generated in the United States.
163
-------
WASTE MATRIX DEVELOPMENT
Haste Property Consideration
Limiting consideration (for the
purpose of scoping the proposed field
testing program) to the three target
industries greatly reduces the boiler
population of concern. However, these
three industries are known to generate.
over 400 of the 456 RCRA listed wastes' in
the F, K, P, and U Series. Of course,
funding provisions do not allow for
testing all 400 waste compounds against
all 5,500 individual boilers. The wastes
and boiler types must be ordered in such a
manner that the limited test program
answers the crucial question of which
waste and boiler combinations are techni-
cally suitable and which are to be avoided.
As a first cut, wastes can be sorted
into Good, Potential, and Poor candidates
for combustion in any operation. This
process was conducted in development of
the Engineering Handbook for Hazardous
Waste Incineration (3). The categoriza-
tion process was approached as follows:
Waste Containing
o Carbon, hydrogen, and/or
oxygen
o Carbon, hydrogen, £30
percent by weight,
chlorine and/or oxygen
o Carbon hydrogen, and/or
oxygen, <30 percent by
weight chlorine, phos-
phorous, sulfur, bromine,
iodine, or nitrogen
o Unknown percent of
chlorine
o Inorganic compounds
o Compounds containing
metals
Combustion
Category
Good
Good
Potential
Potential
Poor
Poor
TABLE 1. BOILER POPULATION AND HAZARDOUS WASTE
GENERATION BY SIC CODE (Reference 2)
Hazardous Waste Generation
103 Metric TPY
28
33
29
34
40
26
37
36
31
35
Chemicals and Allied Products
Primary Metal Industries
Petroleum Refining
Fabricated Metals Products
Non-Manufacturing Industries
Paper and Allied Products
Transportation Equipment
Electrical and Electronic Equipment
Leather and Leather Products
Machinery
25510
4061
2118
1997
1971
1296
1241
1092
474
323
Number of Boilers
% of Total w/Capacity >10? Btu/hr
62.0
9.8
5.1
4.8
4.8
3.1
3.0
2.6
1.1
0.8
3346
1417
1177
399
No Data
1584
775
318
No Data
432
164
-------
Of course, mixed wastes can contain
compounds that represent each category.
An appropriate testing matrix would
encompass wastes from each category.
Incinerability
The wastes of concern can be sorted
into good, potential, and poor candidates
for incineration as a function of gross
physical and/or chemical parameters as
described above. However, once inorganic
and metallic compounds are dismissed as
incineration candidates, the remaining
organic materials must be further categor-
ized in some hierarchy of thermal stabil-
ity to ensure that a variety of thermal,
as well as physical conditions are exam-
ined. Each of the numerous approaches
that have been taken to define "inciner-
ability" to date can be best described as
representing either one of two basic
premises: kinetics or equilibrium. The
kinetics premise assumes that the energy
required to ignite a given molecule is the
primary determinant of how difficult it is
to "destroy" the molecule in combustion.
The equilibrium premise holds that the
energy required to complete the combustion
process (to produce carbon dioxide and
water) is the dominant concern. The
kinetics approach is best represented by
^99.99- The equilibrium approach is
best represented by the heat of combustion
AHc/MW. For reasons presented in several
forums, neither method is an unquestion-
ably accurate description of "incinerabil-
ity." However, EPA has chosen to utilize
heat of combustion (in Kcal/gram) as the
currently recommended Principal Organic
Hazardous Constituents (POHC) selection
criteria. The AHc approach will be the
primary incinerability parameter used to
select wastes for test firing in this
boiler test program (4).
The main advantages of using the
AHc method for boiler waste test firing
are that the method is comprehensive
(there exists at least a computed AHc for
each waste in Appendix VIII) and that the
rfaste/boiler selection can be fitted into
the waste/incinerator permitting network.
The AHc method is, as described earlier,
an inexact incinerability measure. To
subdivide the AHc listing too finely would
be imputing more accuracy to the inciner-
ability index than can be supported.
Therefore, it is proposed here to divide
the AHc list into three groups in describ-
ing difficulty to incinerate:
POHC Group
II
III
AHc Range (Kcal/g)
0 - 3.99
4.00 - 6.99
7.00 - 10.00+
This split results in three subgroups
of roughly equal size. Examination
of the POHC constituents within the
categories lends some empirical credibil-
ity to the system. Materials ip Group I
tend to be highly halogenated ethylenes,
ben£enes, and ethanes. Group II contains
lesser halogenated materials and process
production bottoms. Group III contains
the nonhalogenated solvents and solvent
recovery bottoms. In essence, Group I
contains materials that would not normally
be considered for burning as fuels in the
absence of RCRA. Group III contains
materials that have historically been
considered as candidate fuels for combus-
tion operations.
Waste Classification Matrix
Three characteristics of a given
waste material are important in mapping
out a combustion testing program:
1. Thermal stability of the waste
(incinerability)
2. Volume of waste .generated in
the-United States
3. Physical state of the waste
at 25°C.
The thermal stability of the waste is
important in aiding waste selection that
will provide information on the relative
ease of destroying various waste types.
The volume of waste produced is important
in that a goal of the test program is to
test realistic waste/boiler combinations.
A low volume waste is less likely to be
burned in a boiler than is a high volume
waste. The physical state of the waste
will impact the probability of boiler type
selection. A waste that is a solid or a
sludge will not likely be burned directly
in a gas- or oil-fired boiler.
Table 2 provides an example matrix of
waste POHC, waste identification code for
some wastes that contain that POHC, waste
description, annual waste production rate,
and physical state of the waste at 25°C.
165
-------
Table 3 summarizes the information in
Table 2 in terms of EPA waste identifica-
tion numbers. Implicit in this analysis
is the assumption that wastes in mixtures
retain their discrete properties.
TABLE 2. WASTE CLASSIFICATION MATRIX
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
1.
2.
3.
4.
POHC Group I:
Trichl orof 1 uoromethane
Tetrachl oromethane
(Carbon Tetrachl on' de)
Hexachloroethane
Chloroform
Tetrachl oroethyl ene
(Perch! oroethylene)
Methyl ene Chloride
(Dichl oromethane)
Trichl oroethyl ene
Hexachl orobenzene
Tr,ichloroethane
Hexachl orobutadiene
Pentachlorophenol
Dichl oroethane
(Ethylene Dichloride)
Trichl orobenzene
2,4 Dichl orophenol
POHC Group II:
«
Vinyl idine Chloride
Dichl orobenzene
(1,2 Dichl orobenzene)
Phthalic Anhydride
Nitrobenzene
AHC
(Kcal/g)
0.1JL
0.24
0.46
0.75
1.19
1.70
1.74
1.79
1.99
2.12
2.19
3.00
3.40
3.81
4.45
4.57
5.29
5.50
Representative
Waste I.D.
F002
K073
K085
K009
F002
F002
F002
K030
F002
K016
- K001
K020
K015
K001
K020
F002
K022
K025
Generation
Rate
MT/YR
9,072
12,500
12,500
100,000
255,800
213,000
181,000
40,000
181,000
*
*
30,000
1,800
*
*
11,800
*
1,000
State at
25°C
Li qu i d
Liquid
Solid
Liquid
Liquid
Liquid
Liquid
Solid
Liquid
Solid
Solid
Liquid
Solid
Solid
Liquid
Liquid
Solid
Liquid
166
-------
Table 2 (continued)
5
6
1
2
3
4
5
6
7
8
9
10
*
POHC Group II (cont.)
. Toluene Diisocyanate
. Chlorobenzene
POHC Group III:
. Acetonitrile
. Phenol
Pyridine
. Acrylonitrile
. Methyl Ethyl Ketone
. Cresylic Acid
. Cresols
Naphthalene
Benzene
. To! uene
Undetermined Quantity
AHC
(Kcal/g)
5.92
6.60
*
7.37
7.78
7.83
7.93
8.07
8.09
8.18
9.62
10.03
10.14
Representative
Waste I.D.
K027
F002
K014
K022
K026
KOI 3
F005
F004
F004
K060
K001
K027
Generation
Rate State at
MT/YR 25° C
6,000 Solid
77,000 Liquid
6,000 Liquid
2,000 Solid
*
*
* Liquid
* Liquid
* Liquid
* Sludge
* Solid
* Liquid
167
-------
TABLE 3. WASTE MATRIX BY RCRA WASTE I.D. NUMBER
WASTE I.D.
POHC GROUP
PHYSICAL FORM
GENERATION RATE (MT/YR)
F001
F002
F004
F005
K001
K009
K011
K013
K014
K015
K016
K018
K019
K020
K021
K022
K023
K025
K026
K027
K029
K030
K036
K037
K042
K073
K085
K095/96
I
I, II
III
III
I, III
I
III
III
III
I
I
I
I
I, II
I
II, III
II
II
III
III
I
I, II
.III
III
III
I
I
I
Organic, Sludge, Solid
Liquid
Liquid
Liquid
Solid
Aqueous Liquid
Aqueous Liquid
Aqueous Liquid
Organic Liquid
Organic Liquid
Organic Liquid
Organic Liquid
Organic Liquid
Organic Liquid
Aqueous Metallic Sludge
Organic Sludge .
Liquid
Organic Sludge
Aqueous Liquid
Organic Sludge
Aqueous Liquid
Organic Liquid
Organic Sludge
Aqueous Sludge
Organic Solid
Organic Liquid
Organic Sludge
Organic Liquid
100,000*
100,000
-
-
65,000
-
-
6,000
47,000
50
3,200
35,000
8,000
125,000
100,000
-
1,000
1,000
6,000
1,200
61,000
-
-
-
125,000
6,600
34,000
* Estimated 10 percent of Production.
- Undetermined Quantity
Aside from thermal stability, selected
waste characteristics may be desirable
from a testing perspective. Two charac-
teristics that may be complicating factors
from a combustion perspective are the
presence of significant amounts of water
and metal contaminants in the wastes. A
hierarchy of wastes in descending order of
difficulty to destroy in boilers can be
envisioned as follows in Table 4:
168
-------
TABLE 4. WASTE CLASSIFICATION MATRIX
CHARACTERISTICS
WASTE TYPE EXAMPLE(s)
cc
o
Group I, Aqueous
Group I, Metals
Group I, Nonaqueous
Group II
Group III, Aqueous
Group III, Metals
Group III, Nonaqueous
K001, K009, K029
K021
F001, F002, K015, K016, K018, K073, K085
F002, K020, K022, K023, K025, K036
K011, K013, K026, K037
F003, F004, F005, K001, K014, K022, K027, K036, K042
BOILER MATRIX DEVELOPMENT
Boilers In the Chemical Industry
The purpose of this testing program
is to demonstrate to what degree and which
boilers are capable of destroying which
waste. Since the focus of the study has
been determined to be the chemicals
industry, we are interested in testing
boilers that are actually used in that
industry.
Table 5 describes the types and
capacities of boilers used in the chem-
icals industry. By far the largest number
of boilers are oil (46.2 percent) or gas
(42.0 percent) fired units. Coal (10.3
percent) and waste (1.5 percent) fired
boilers are far less represented in the
chemicals industry. Coal boilers do make
up 13.8 percent of total boiler capacity,
indicating that the average coal boiler is
larger than either the average gas- or
oil-fired unit.
TABLE 5. BOILERS IN THE CHEMICALS INDUSTRY
(SIC 28) (Reference 2)
FUEL
TYPE
Gas
Oil
Waste
Coal
NUMBER OF
10-50
542
679
13
41
50-100
276
351
13
74
BOILERS IN SIZE RANGE (106
100-200
261
299
12
128
200-400
206.
150
10
77
400-700
109
61
1
21
Btu)
700-1250
11
8
1
2
TOTAL NO.
BOILERS
1405
1548
50
343
TOTAL
CAPACITY
(109 Btu)
196.78
168.12
7.25
59.55
% TOTAL
CAPACITY
45.6
38.9
1.7
13.8
TOTALS 1,275 714
700
443
192
22
3346
431.70
100.0
169
-------
Existing Data and Previous Studies
Substantial data exists on emissions
of regulated criteria pollutants from
power boilers. However, there exists no
data base on destruction of hazardous
chemical compounds in industrial boilers.
Source emissions data for regulated
pollutants may be useful in characterizing
a given boiler's ability to destroy
hazardous chemical compounds. These will
be discussed in more detail in a subse-
quent section of this document. Some
attempts have been made to directly assess
the capabilities for boiler destruction of
hazardous wastes through modelling and
engineering analysis. These attempts will
be discussed in this section.
In two separate studies, conducted in
1980 and 1981, Acurex Corporation examined
thermal conditions for distinct boiler
types (1, 5). Time above temperature
predictive plots were developed for each
boiler type. Fuel particle path analyses
were conducted at full and half loads.
The environment that each boiler presented
to a given fuel was described in terms of
conditions of time and temperature re-
quired for heating, volatilization,
thermal cracking, and oxidation of th,e
fuel. Zonal heat balances were applied to
estimate conditions at each point in the
boiler.
The Acurex analysis produced some
general conclusions. It appeared that the
larger the boiler the greater the likeli-
hood that conditions for waste destruction
will exist. The medium-sized watertube
units present mean conditions that are
conducive to destruction of most wastes
but have significant potential for excur-
sion into conditions not favorable for
waste destruction at the margins of
operating conditions. The firetube
boilers present high temperatures and good
flame exposure but the fuel residence time
appears to be of insufficient duration to
allow for volatilization and oxidation of
the waste fuel particles.
Haste Destruction Surrogates
In the absence of specific waste
Destruction and Removal Efficiency (ORE)
data for boiler firing of hazardous
wastes, examination of some boiler opera-
tional parameters and stack emissions data
could prove helpful in assessing a boiler's
ability to destroy hazardous wastes.
None of these candidate ORE "surrogates"
have been verified in experimentation or
testing but discussion of the various
available parameters is of-some value.
ORE is primarily a function of
temperature. Flame exposure, residence
time, and mixing are also factors that
affect the degree of oxidation of a
particular molecule. For a boiler with
all other factors held constant, waste
destruction will more than likely occur as
described below:
ORE will increase as the following
items increase:
Boiler load
Mean gas temperature
Mean gas residence time
C02 emissions
ORE will increase as the following
items decrease:
Excess air
02 emissions
Boiler surface-to-volume ratio
CO emissions
Total HC emissions
Examination of the above listing leads to
a winnowing out process. While it is true
that boiler combustion efficiency in-
creases with load tip to optimum (generally
nameplate) output conditions, it is also
true that the same boiler can operate at
different load conditions. Boiler load is
not a machine characteristic but an
operating characteristic and cannot be
used to distinguish between boiler types.
The same statement can be made concerning
excess air, C02 emissions, and QZ
emissions. Excess air is largely a
function of load conditions. C02 and
02 emissions are excess air- and load-
dependent.
Table 6 describes pertinent machine
specific parameters and emissions that can
be used to classify boilers with respect
to their estimated ability to destroy
hazardous compounds (1, 6, 7, 8). Each of
the parameters presented in Table 6 will
be described in detail.
170
-------
Mean gas temperature and residence
times are significant from a waste -kine-
tics perspective and were utilized exten-
sively in the Acurex analysis. However,
at temperatures above the kinetics operat-
ing zone and in the presence of a flame,
time-temperature paths are not sufficient
to fully describe compound destruction and
require detailed knowledge of the kinetic
behavior of the target compounds to be
effective.
Table 6 does show that while the
temperatures in the firetube boilers
are very high, the bulk gas residence
times are so short that there is less
certainty that all particles of a given
material will "experience" the complete
kinetics process (e.g., volatilization,
pyrolysis, ignition).
TABLE 6. BOILER CLASSIFICATION PARAMETERS (References 1, 6, 7, 8)
BOILER
TYPE
Firetube
Fi retube
Watertube
Wate'rtube
Watertube
Watertube
Watertube
Watertube
Watertube
BOILER
FUEL
Natural Gas
Oil
Natural Gas
Oil
Stoker Coal
Stoker Coal
Natural Gas
Oil
Pulverized
Coal
BOILER
SIZE
(MM Btu/hr)
10-50
10-50
10-100
10-100
10-100
100+
100+
100+
150+
EXIT
T (°F)
2400
2400
2300
2300
2060
2000
2200
2000
2000
RESIDENCE
t (sec)
0.23
0.22
0.50
0.50
1.50
1.90
0.82
0.77
1.82
S/V )
RATIO
1.60
1.60
0.80
0.80
0.30
0.22
0.52
0.52
0.27
CO
ng/J
8
15
15
14
146
86
17
14
10
HC
ng/J
3.4
1.7
2.4
3.7
3.6
1.7
0.7
3.0
1.7
NOX
ng/J
42
59
156
156
155
240
232
190
361
Surface-to-volume ratio (S/V) is
important as a descriptor for a potential
ORE failure mechanism -- gas quenching.
The larger the boiler tube surface with
respect to the total furnace volume, the
more likely that a particle of waste
material will experience something less
than mean furnace temperature. As ex-
pected, S/V generally decreases with
boiler size. The principal exception to
this rule is the case of coal fired
boilers. For coal boilers, the fuel
temperature must be kept below the ash
fusion temperature to avoid slagging and
fouling. This lower temperature operation
requires that, in order to produce a given
pound of steam, a given coal boiler must
be much larger than an equivalent oil- or
gas-fired boiler. Therefore, furnace heat
liberation rates for coal-fired boilers
are much lower than for other types of
boilers (by a factor of from 2 to 10).
This factor largely explains why coal
boilers tend to be much larger (capacity
and size-wise) than other boilers. The
net effect is that while larger boilers
have lower S/V ratios than smaller boil-
ers, coal boilers have even lower S/V
ratios than either oil or gas boilers in a
similar size range.
171
-------
CO and HC emissions may be machine
specific and are a measure of the com-
pleteness of combustion. Unfortunately,
these emissions are also fuel- and opera-
tional-dependent. Two coals or oils may
produce different CO and HC emissions due
to different fuel and operational condi-
tions. Two different machines burning the
same fuel under "optimum" operating
conditions may yield meaningful CO and HC
data. In a similar manner, HC and CO
emissions from the same boiler may provide
useful surrogate data on compound destruc-
tion. However, the AP-42 data presented
in Table 6 is compiled in an a posteriori
manner and is not useful in characterizing
ORE.
Data on NOx emissions, on the other
hand, can be very useful in describing a
given boiler's potential to destroy waste.
The importance of NOX emissions as a
boiler classification tool is discussed in
the next section.
Nitrogen as a POHC Surrogate
Production of oxides of nitrogen
(NOX) from fuelborne nitrogen occurs
relatively easily in combustion operations.
However, production of NOX from airborne
nitrogen is an extremely energy intensive
operation that occurs ony at elevated
temperature (assuming atmospheric pres-
sure). In the temperature regions in
which boilers operate far more NOX is
produced from air than from fuel. There-
fore, in a boiler, the most likely process
for production of NOX is the cracking of
the N£ molecule to N* and the oxidation
of the N+ to form N02, Nos, etc.
This highly energy intensive process is
thermodynamically similar to that re-
quired to break a carbon = carbon double
bond or a carbon - chlorine bond and
produce CO and C02 from fuel. Therefore,
all other things being equal, the tendency
of a given boiler to produce NOX should
be a rough measure of that boiler's
ability to destroy a waste compound. Of
course, all things are never equal but
this sort of macroapproach is well suited
toward use of the AP-42 data presented in
Table 6.
Rearranging the presentation of the
data in Table 6 so that boilers are
displayed in terms of ascending NOX
emissions and deleting the CO and HC daa,
we obtain Table 7. Examination of Table 7
yields boilers ranked in ascending order
of size (capacity). Also, it should be
noted that within boiler fuel types, the
ranking by size is rigidly adhered to
(i.e., the smallest oil boiler listed has
the lowest characteristic NOX emissions
of all oil boilers; likewise, the largest
oil boiler has the highest characteristic
NOX emissions). Similar boiler types
(e.g., firetubes) have similar NOX
emissions.
172
-------
TABLE 7. BOILERS RANKED IN'TERMS OF HISTORICAL NOX EMISSIONS
BOILER
TYPE
Fi retube
Fi retube
Watertube
Watertube
Watertube
Watertube
Watertube
Watertube
Watertube
BOILER
FUEL
Natural Gas
Oil
Stoker Coal
Natural Gas
Oil
Oil
Natural Gas
Stoker Coal
Pulverized Coal
BOILER
SIZE
(MM Btu/hr)
10-50
10-50
10-100
10-100
10-100
100+
100+
100+
150+
Boiler Classification Matrix
NOX
Ng/J
42
59
155
156
156
190
232
240
361
cut makes
S/V
RATIO
1.60
1.60
0.30
0.80 '
0.80
0.52
0.52
0.22
0.27
sense when
EXIT T
("F)
2400
2400
2060
2300
2300
2000
2200
2060
2060
we consider
RESIDENCE
(sec)
0.22
0.23
1.50
0.50
0.50
0.77
0.82
1.90
1.82
that, at
Through utilization of Tables 5, 6
and 7, a boiler matrix can be developed.
The size and type breakdown in Table 7
conforms to AP-42 structure and need not
impose a constraint on matrix development
but should be used as a guide to boiler
selection. However, function as well as
convenience is served by generally adher-
ing to these categories.
From Table 7, firetube boilers
represent a suspected worst case for
waste destruction. From Table 5, over
one-third of the boilers found in the
chemicals industry are in the size range
of 10 to 50 MM Btu/hr. Not all of those
small boilers are firetubes, but all
firetube boilers are found in this size
range. From Table 7, regardless of fuel
type, the firetube boilers perform and
appear to be identical. Therefore, one
boiler matrix constituent will be "Fire-
tube Boilers."
The oil- and gas-fired watertube
boilers in the 10 to 100 MM Btu/hr size
range are twins and should be grouped
together (Table 7). The size range spread
may appear to be large. However, such a
tice is to go from single or multiple
burner configurations. The larger and
more uniform flame zone should provide for
more constant and harsher combustion
conditions with the boiler.
The oil^ and gas-fired boilers in the
greater than 100 MM Btu/hr size range
should also be grouped together. From
Table 5, those boilers account for approxi-
mately 30 percent of the population of
boilers in the chemicals industry. Too
fine a breakdown would be of little merit,
particularly since this size range appears
to be biased toward a best case with
increasing boiler size.
Coal boilers warrant little attention
due to their small presence in the
chemicals industry (Table 5). However,
coal boilers have two characteristics
which mandate interest: 1) the ability to
accommodate waste solids and sludges and,2)
low S/V ratios, long residence times and
relatively high NOX emissions. The
ability to handle solids is unique to coal
boilers. The latter point makes coal
boilers a potential best case. Examina-
tion of Table 5 yields that the critical
173
-------
mass of coal boilers in the chemicals
industry is in the larger size range (100
to 400 MM Btu/hr), little difference
between pulverized coal and stoker boilers
with respect to physical and operating
characteristics. The higher NOX emis-
sions for the pulverized coal boilers
would be a function of size rather than
type of boiler. A single coal-fired
boiler entry on the matrix will be suffi-
cient.
Table 8 lists the boiler categories
on the test matrix in ascending order
of assumed ability to destroy wastes.
WASTE/BOILER TEST MATRIX
Waste/Boiler Matrix Format
By overlaying the boiler test matrix
from Table 8 onto the waste classification
matrix from Table 4, one can obtain a
waste/boiler test matrix format. This
format is described in Table 9.
TABLE 8. BOILERS RANKED IN ASCENDING ORDER OF ASSUMED
ABILITY TO DESTROY WASTES
CASE I.D.
BOILER TYPE
FUEL TYPE
A
LU
O
£
k B
c
GC
8 D
Fi retube
Watertube
Watertube
Watertube
Oil, Gas
Oi 1 , Gas
Oil, Gas
Coal
10-50
10-100
100+
s
100+
174
-------
TABLE 9. ECONOMY WASTE/BOILER TEST PROGRAM
(12 Tests)
WORST CASE
BOILER DESCRIPTIONS
WASTE
POHC
GROUP
DESCRIPTORS
PHYSICAL
QUALIFIERS
• CASE I.D.
BOILER TYPE
FUEL TYPE
CAPACITY (106 Btu/hr)
A"
FIRETUBE
OIL, GAS
10-50
- &
WATERTUBE
OIL,' GAS
10-100,
C
WATERTUBE
OIL, .GAS
100+
0
WATERTUBE
COAL
100+
AQUEOUS
METALS
I
SOLIDS
i
UI
CO
o
t/0
o
NON-AQUEOUS
II
. AQUEOUS
METALS
III
SOLIDS
NON-AQUEOUS
X X
N/A N/A N/A
X X X
X
X
N/A N/A N/A
XXX
X
X
NA - Not Applicable
Utilizing Table 9, we can structure
a test program that comprehensively
covers the spectrum of waste and boiler
types or one that encompasses only boun-
dary conditions within a proscribed
budget.
The wastes are described by POHC
Group and Physical Qualifiers that may
impact boiler type selection and waste
destruction. "Aqueous" implies signifi-
cant water in the waste. "Metals" refers
to the presence of inorganic or metallic
contaminants in the waste which could -
present problems for the boiler. "Solids"
refers to the waste containing solid
materials or sludges. "Non-aqueous"
refers to.a waste with no significant
water, solids, or contaminants. Of
course, a-given waste could contain a
combination of conditions. All permuta-
tions are not shown but can be accommo-
dated within the matrix. It may be argued
that a highly aqueous Group III waste can
be a worse case than a non-aqueous Group I
waste. The wastes are arrayed from top to
bottom in order of assumed worst to
assumed best case. Group II wastes are
not broken down further since these are
not assumed to be limiting conditions.
The boilers are described by firebox
type, fuel type, and capacity. The boil-
ers are arrayed from left to right in
order of assumed worst to assumed best
case. . Certain physically constraining
cases can be assumed to exist. For ex-
ample, it is unlikely that a solid-contain-
ing waste can be burned in an bil- or gas-
fired boiler unless that boiler has been
converted to a liquid fuel from a coal
operation or the waste fuel has been vola-
tilized in some preprocessing application.
175
-------
Additionally, air pollution control
devices (APCDs) are not factored into the
matrix. Smaller industrial boilers (10 to
50 MM Btu) generally have no APCDs.
Larger coal-fired boilers may have multi-
clones (stokers) or electrostatic precipi-
tators (pulverized coal), but large oil-
and gas-fired industrial boilers generally
have no APCDs. It is felt that the
addition of factors for presence, absence,
or types of APCD would overly complicate
the matrix discussion.
Test Program Selection
Ideally, of course, a comprehensive
test program would involve at least one
test of each waste/boiler combination and
permutations of each combination described
in Table 9. Such a test matrix would
require completion of at least 60 tests.
However, EPA currently has sufficient
funds for only an "economy" 12-test
program. Such a truncated program must
focus on "worst case" conditions. Limited
testing would be conducted on "best case"
wastes and only one Group II verification
test could be tested. Comprehensive
testing of the most representative boiler
type (B) would be emphasized. Little
exploration of qualifying waste conditions
(metals, aqueous) would be conducted.
Hopefully, this program, supported by
pilot-scale parametric testing, would
provide an "envelope" around acceptable
waste/boiler combinations.
Utility and Flexibility of the Matrix
Almost every conceivable waste and
standard boiler combination will fit into
the waste/boiler matrix. Ideally, access
to all types of wastes and boilers can be
secured. In the "real" world of testing
targets of opportunity, the matrix can
serve as a guide to test selection. For
instance, if eight test candidates are
available, the matrix would help ensure
that five III B tests are not selected.
Waste volume is also a concern. For
example, if there exists a choice between
two IB combinations, the waste that has
the highest national generation rate would
be selected.
Cautions On the Use of the Matrix
The matrix is a tool for ordering a
testing program. Implicit in its use are
certain assumptions. The first assumption
is that all boilers are operated equally
expertly. It is conceivable that all
boilers, if well designed and perfectly
operated, can achieve DREs of 99.99
percent. Within boiler types the over-
riding factor concerning waste destruction
is boiler operation. It is entirely
probable that a well operated "worst case"
boiler will outperform a poorly operated
"best case" boiler.
The second assumption is that addi-
tion of the waste material to the boiler
does not alter boiler operating character-
istics. Any waste added in abundant
quantities will indeed have an effect on
boiler operating conditions.
The third assumption is that all
wastes are equally likely to be burned in
all boilers. In fact, due to their
potentially damaging impact on boiler
hardware, certain wastes (particularly
Group I highly chlorinated materials) may
not be burned in boilers at all.
REFERENCES
1. Castaldim", et al., "A Technical
Overview of the Concept of Disposing
of Hazardous Wastes in Industrial
Boilers," Acurex Corporation for
USEPA, lERL-Ci, October 1981.
2. National Environmental Data System,
1982.
3. Bonner, et al., Engineering Handbook
for Hazardous Waste Incineration,
Monsanto Research Corporation for
USEPA, lERL-Ci, September 1981.
4. Guidance Manual for Hazardous Waste
Incinerator Permits, USEPA, QSW,
and MITRE-'Corporation, September
1982.
5. McCormick, et. al., "Engineering
Analysis of the Practice of Disposing
of Hazardous Wastes in Industrial
176
-------
Boilers," Acurex Corporation for
USEPA, lERL-Ci, January 1982.
6. Personal Communication from Howard
Mason, Acurex Corporation, to R.A.
Olexsey, USEPA, December 1982.
7. "Compilation of Air Pollutant Emis-
sions Factors," AP-42, USEPA, 1982.
8. Estimates and Interpolations by R.A.
Olexsey, December 1982.
177
-------
OBSERVATIONS OF FLUID DYNAMIC EFFECTS UPON
HIGH-TEMPERATURE DESTRUCTION OF ORGANIC COMPOUNDS
Wayne A. Rubey, John L. Graham,
Douglas L. Hall, and Barry Dellinger
University of Dayton Research Institute
Dayton, OH
ABSTRACT
In the controlled high-temperature
Incineration of industrial organic wastes,
events that occur in the gas-phase are of
vital importance. Within a rotary kiln-
afterburner incinerator the large volume
afterburner chamber is responsible for the
complete destruction of organic materials
that survive, or are emanated from the
previous exposures.
The fluid dynamic behavior of the
high-temperature gas phase has been exam-
ined with special attention to the various
aspects of exposure time and fluid interac-
tion of gas-phase substances. These
observations with respect to destruction
efficiency of organic substances have been
made on the laboratory-scale, and several
implications can be drawn with respect to
the gas-phase decompositions of both
hazardous organic compounds and their
thermal reaction products.
178
-------
RESEARCH PROJECTS AT ERA'S LOUISIANA STATE UNIVERSITY
HAZARDOUS WASTE RESEARCH CENTER - USEPA CENTER OF EXCELLENCE
D. P. Harrison
Hazardous Waste Research Center
Louisiana State University
Baton Rouge, Louisiana 70803
ABSTRACT
Now beginning its third year of opera-
tion, the Hazardous Waste Research Center
at Louisiana State University is conducting
fundamental and exploratory research in
three general areas: chemical/material
interactions, incineration-, and alternate
methods of treatment/destruction. The re-
search projects range from studies on the
rates of supercritical extraction of or-
ganics from solids to the destructibility
of pure chlorinated hydrocarbons in labora-
tory flames to an investigation of bonding
mechanisms associated with cement - stabi-
lized orgarncs. Although administered
through LSU's College of Engineering, indi-
vidual research projects are being conduc-
ted by multidisciplinary groups represent-
ing a number of academic departments.
In addition to its fundamental and
exploratory research efforts, the Center
has adtive and expanding programs in the
areas of applied research and technology
transfer. The applied research component
will encompass projects of more immediate
interest to federal and state agencies and
to private industry. The objective of the
technology transfer activities is to dis-
seminate state-of-the-art technology and
research advances to the professionals in
government and industry and also to the
nonprofessional public. These latter ef-
forts are beyond the scope of the EPA-
funded exploratory research center and are
supported by the university, other govern-
mental agencies, and private industry.
179
-------
HAZARDOUS WASTE CONTROL TECHNOLOGY DATA BASE
Richard L. Holberger
The MITRE Corporation
McLean, VA
C.C. Lee
U.S. Environmental Protection Agency
Cincinnati, OH
ABSTRACT
The Hazardous Waste Control Technology
Data Base (HWCTDB) is an automated data
base sponsored by EPA's Incineration
Research Branch to store detailed technical
data relevant to the thermal destruction of
hazardous wastes. The data base stores
design, operating, and performance data
obtained from incineration facility permit
applications, trial burn reports, research
and development projects, and literature
sources. The data base went on-line in
late 1982.
780
-------
EVALUATION OF AN ADVANCED LOW NOX HEAVY OIL BURNER
FOR INCINERATION OF NITRATED HAZARDOUS COMPOUNDS
W.S. Lanier and Charles B. Sedman
U.S. Environmental Protection Agency
Research Triangle Park, NC
ABSTRACT
Among the RCRA-listed hazardous waste
compounds is a variety of chemical species
which are highly nitrated. During the
combustion of these compounds the bound
nitrogen atom is readily oxidized to form
nitrogen oxide (NO) in the incinerator or
boiler exhaust. Limited test data has
shown that co-firing nitrobenzene with
natural gas in an industrial boiler in-
creased NOx emission by a factor of 6 to 9
when the nitrobenzene feed rate represented
only 6 percent of the input heating value.
Accordingly, a concern exists that class-
ical incineration of such materials may
exchange destruction of the waste for a
potential NOX emission problem.
A significant body of data on the
pyrolysis of nitrated compounds is avail-
able from the EPA NOX control program.
Species such as nitrobenzene, acryloni-
trile, and aniline were previously studied
as possible model compounds for nitrogen
bound within the structure of coal and
heavy residual fuel oils. Those pyrolysis
studies clearly indicated that hydrogen
cyanide was the dominant nitrogen species
evolved. Accordingly, incineration of
this class of waste raises concern that
HCN would be a potential product of incom-
plete combustion (PIC) and that emission
of that PIC would be of equal or greater
concern than emission of the principal
organic hazardous compound (POHC).
Based upon considerations such as
those identified above, a need exists for
combustion technology which could assure
high destruction efficiency for the POHC
and probable PIC and simultaneously mini-
mize the emission of NOX. To accomplish
that objective requires a combustor which
establishes an extremely high temperature,
fuel-rich primary combustion zone followed
by an excess air burnout region. A fuel-
rich primary zone would shift nitrogen
equilibrium almost totally to N2 whereas
fuel-lean equilibrium is substantially
shifted toward NO formation. The high
temperature promotes initial destruction
of the POHC but also accelerates chemical
kinetic processes driving fixed nitrogen
species (such as any HCN PIC) to N2-
As part of the EPA NOX control pro-
gram, combustion techniques such as that
described above have been previously
established. A three million Btu/hr
version of this burner is available at the
in-house research facilities of IERL/RTP
and a 60 x 106 Btu/hr version is cur-
rently being field tested in Kern County,
California, on an enhanced oil recovery
steam generator. The current paper re-
ports on an evaluation of this low HO^
-burner technology for incineration of
three highly nitrated hazardous compounds.
Test results are presented on the destruc-
tion efficiency of the POHC, emission of
HCN and NOX while burning nitrobenzene,
aniline, and acrylonitrile. All waste
compounds were doped into distillate fuel
oil at concentrations of 5 and 25 percent
by weight. Test results are being evalu-
ated prior to proceeding with a full-scale
field verification of this technology.
181
-------
REGULATORY SUPPORT RESEARCH STUDIES SUPPORTED BY
EPA/CINCINNATI MSD COOPERATIVE AGREEMENT
Boyd T. Riley
RYCON
Cincinnati, OH
ABSTRACT
A number of research studies have been
proposed and carried out to produce infor-
mation on various aspects of a publicly
owned, full scale hazardous waste incinera-
tor. These studies are intended to produce
data to assist in the development, support
and revision of EPA's regulatory responsi-
bilities addressing hazardous waste incin-
erators. Typical studies that have been
carried out include: the preparation of a
design and economic history of the inciner-
ator complex, evaluation of options for
acid neutralization in combustion gases,
evaluation of DRE capabilities and poten-
tial monitors, and testing of alternative
modifications to particulate emission
control devices. Anticipated studies
include: developing a detailed understand-
ing of ash characteristics with regard to
hazardous properties, testing of various
pilot scale particulate control devices,
and preliminary evaluation of a colorimetric
DRE monitor.
182
-------
FLAME-MODE HAZARDOUS WASTE DESTRUCTION RESEARCH
Wm. Randall Seeker, John C. Kramlich, Michael P. Heap, Rachel K. Nihart
Energy and Environmental Research Corporation
Irvine, California 92714
Gary S. Samuel sen
Department of Mechanical Engineering
University of California
Irvine, California 92717
ABSTRACT
The results of a laboratory-scale
study of flame-mode hazardous waste in-
cineration are reported. The primary goal
of the work was to compare the various
incinerability ranking procedures with
rankings generated under flame conditions
typical of liquid-injection incinerators.
A secondary objective was the comparison
of flame and non-flame thermal destruction
behavior. We previously reported (Ninth
SHRD Symposium) that, when operated under
efficient combustion conditions, flames
nearly quantitatively destroy organic
waste compounds. For incomplete
destruction to take place, the flame must
be perturbed into inefficient operation.
The compound rankings resulting from
these perturbed states (termed failure
conditions) were dependent on the cause of
the perturbation. The results presented
here represent an extension of the pre-
vious work.
The results were generated using a
65 Kilowatt (kW) turbulent liquid spray
flame. The flame enclosure was cooled to
concentrate on flame-zone processes by
quenching post-flame reactions. In the
present study the flame was fired on
No. 2 fuel oil doped with four model waste
compounds (3% by weight): Benzene,
chlorobenzene, chloroform and acryloni-
trile. The flame was operated under high
efficiency conditions by high and low
excess air, by degraded fuel spray atomi-
zation quality, and by impingement of the
flame on a cold surface. The'destruction
efficiency was measured by Tenax trapping
of the stack effluent followed by thermal
desorption and gas chromatographic/flame
ionization detector analysis. Simul-
taneous measurements of CO, C02, 02, CH4
and total unburned hydrocarbons were
obtained.
The results sustained the previously
reported conclusions that flames are
capable of high DE, and that loss of this '.
high DE is necessarily due to perturbation
of the flame into a failure condition.
Waste DE rankings were found to be
dependent of flame condition. The new
results also indicated that a linear
relationship exists between DE and un-
burned hydrocarbon emissions from the
flame zone. Correlation of DE and CO
measurements showed that CO emissions
must increase to substantial levels
before DE begins to degrade. It should
be emphasized that these flame-zone
results may be modified by realistic
post-flame time-temperature histories.
These results suggest a performance
monitoring and control methodology in
which the difference between a correctly
operating incinerator and a failure
mode can be detected. The CO measure-
ment could be used as an early indication
of the approach of a failure condition.
It could also be used as a means of
controlling operating parameters. Total
hydrocarbon measurement could be used
to indicate the presence of a failure
condition. Significant hydrocarbon
emissions could be used to shut down waste
flow.
183
-------
EVALUATION AND DEVELOPMENT OF INNOVATIVE INCINERATION TESTING PROCEDURES
Jerome B. Strauss
Versar, Inc.
6850 Versar Center
Springfield, VA 22151
ABSTRACT
This study Investigated Innovative
real-time monitoring techniques for haz-
ardous waste Incineration. The objec-
tives were to survey and select monitor-
Ing systems that measured (either
directly or Indirectly) destruction and
removal efficiency (ORE), and to demon-
strate the effectiveness of the selected
system 1n a test program. Seven poten-
tial monitoring systems were evaluated
using the criteria of operating charac-
teristics, limitations, applications,
degree of development, and cost. The
selected system was an 1n-l1ne flame
1on1zat1on detector (FID) that measures
total organic carbon In the stack.
Tests were performed at the EPA's Com-
bustion Research Facility 1n P1ne Bluff,
Arkansas. Six test runs were conducted
using hexachlorobenzene and/or toluene
as surrogate hazardous wastes. Total
organic carbon concentrations corres-
ponded 1n the same direction and
magnitude as the ORE values as measured
by Method 5 stack tests. The results
from these tests Indicate that the
selected approach 1s a promising system
for on-Hne, real-time monitoring of
Incinerator performance.
From the results of these tests 1t
1s recommended that further study be
performed to quantify the relationship
between TOC concentration and POHC
destruction and removal. Appropriate
relationships need to be rigorously
challenged and tested 1f TOC concen-
tration 1s to become an accepted method
for monitoring ORE values.
184
* U.S. GOVERNMENT PRINTING OFFICE: 1984 - 759-102/10696
-------
-------
o
o
^
VJD
oo
-C-
O
to
1X3
2 n c
-?• a.
•-
X w o
3 o -^
TO O (B
^ «r n
= -o
-
= 2
H.5 3
" 2.
-
s
0)
§.
§
.
D 0> I
01 rt |
NJ
C3>
00
------- |