&EPA
United States
Environmental Protection
Agency
Hazardous Waste Engineering
Research Laboratory
Cincinnati, OH 45268
EPA/600/9-85/028
September 1985
Research and Development
Incineration and
Treatment of
Hazardous Waste
Proceedings of the
Eleventh Annual
Research Symposium
-------
-------
EPA/600/9-85/028
September 1985
INCINERATION AND TREATMENT OF HAZARDOUS WASTE
Proceedings of the Eleventh Annual Research Symposium
at Cincinnati, Ohio, April 29-May 1, 1985
Sponsored by the U.S. EPA, Office of Research & Development
Hazardous Waste Engineering Research Laboratory
Alternative Technologies Division
Thermal Destruction Branch
and
Land Pollution Control Division
Containment Branch
Coordinated by:
JACA Corp.
Fort Washington, Pennsylvania 19034
Contract No. 68-03-3131
Project Officer:
Harry Freeman
Alternative Technologies Division
Cincinnati, Ohio 45268
HAZARDOUS WASTE ENGINEERING RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
These proceedings have been reviewed in accordance with the U.S.
Environmental Protection Agency's peer and administrative review policies
and approved for presentation and publication. Mention of trade names or
commercial products does not constitute endorsement or recommendation for
use.
-------
FOREWORD
Today's rapidly developing and changing technologies and industrial
products and practices frequently carry with them the increased generation of
solid and hazardous wastes. These materials, if improperly dealt with, can
threaten both public health and the environment. Abandoned waste sites and
accidental releases of toxic and hazardous substances to the environment also
have important environmental and public health implications. The Hazardous
Waste Engineering Research Laboratory assists in providing an authoritative
and defensible engineering basis for assessing and solving hazardous waste
problems. Its products support the policies, programs, and regulations of
the Environmental Protection Agency, the permitting and other responsibilities
of State and local governments, and the needs of both large and small business-
es in handling their wastes responsibly and economically.
These Proceedings present the results of completed and ongoing incinera-
tion and treatment research projects. Those wishing additional information
on any of the projects discussed in these proceedings should contact the
Alternative Technologies Division.
David G. Stephan, Director
Hazardous Waste Engineering Research Laboratory
-iii-
-------
PREFACE
These Proceedings are intended to disseminate up-to-date information
on research projects concerning land disposal, remedial action 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 Eleventh Annual Research Symposium on Land Disposal, Remedial
Action, Incineration and Treatment of Hazardous Waste was held in Cincinnati,
•Ohio, April 29-May 1, 1985. 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
Hazardous Waste Engineering Research Laboratory's Alternative Technologies
Division and Land Pollution Control Division.
This volume is a compilation of speakers' papers for Session B concern-
ing hazardous waste incineration and treatment. Subjects include thermal
treatment and destruction, air pollution control, biological and chemical
treatment, bench and pilot scale research, industrial processes and boilers,
and economics and institutional studies.
This document covers Hazardous Waste Incineration and Treatment only.
A separate document for Session A, Hazardous Waste Land Disposal, is avail-
able from the Hazardous Waste Engineering Research Laboratory.
-v-
-------
CONTENTS
Page
EPA's Hazardous Waste Research Programs: 1985
Carl Gerber, U.S. Environmental Protection Agency 1
Emission and Control of By-Products from Hazardous Waste
Combustion Processes
Robert Olexsey, George Huffman, and Gordon Evans
U.S. Environmental Protection Agency 8
Practical Guide to Trial Burns at Hazardous Waste Incinerators
Paul Gorman, Midwest Research Institute
Donald Oberacker, U.S. Environmental Protection Agency 16
Carbon Monoxide and ORE: How Well Do They Correlate?
Laurel Staley, U.S; Environmental Protection Agency 23
Summary of Testing Program at Hazardous Waste Incinerators
Andrew Trenholm, Midwest Research Institute
Donald Oberacker, U.S. Environmental Protection Agency 36
Tier 4 Dioxin Test Program Status
A.J. Miles, R.M. Parks, and J. Southerland
Radian Corporation
Donald Oberacker, U.S. Environmental Protection Agency 44
Powdered Activated Carbon Treatment (PACT) of Leachate from
the Stringfellow Quarry
William Copa, Marvin Dietrich, and Tipton Randall
Zimpro, Inc.
Patrick Canney, Casmalia Resources 52
Field Testing of Pilot-Scale APCDs at a Hazardous Waste Incinerator
Wayne Westbrook and Eugene Tatsch
Research Triangle Institute
Lawrence Cottone, Engineering Science, Inc.
Harry Freeman, U.S.' Environmental Protection Agency 66
Case Studies of Waste Treatment at Hazardous Waste Facilities
C. Clark Allen, Research Triangle Institute
Benjamin Blaney, U.S. Environmental Protection Agency 76
A Case Study of Direct Control of Emissions from a Surface Impoundment
R.G. Wetherold, B.M. Eklund, and T.P. Nelson
Radian Corporation 85
-vi-
-------
CONTENTS (continued)
Page
Products of Incomplete Combustion - Analytical Methods
M.M. Thompson, R.H. James, and R.E. Adams
Southern Research Institute
L.D. Johnson, U.S. Environmental Protection Agency 93
EPA Research to Recover Toxic Heavy Metals from Waste Streams
S. Garry Howe11, U.S. Environmental Protection Agency . 100
Chemical Destruction/Detoxification of Chlorinated Dioxins in Soils
Robert Peterson and Edwina Milicic
Gal son Research Corporation
Charles J. Rogers, U.S. Environmental Protection Agency 106
Gene Engineering of Yeasts for the Biodegradation of Hazardous Wastes
John Loper, Chien Chen, and Chitta Dey
University of Cincinnati 112
Biodegradation of Environmental Pollutants by the White Rot
Fungus Phanerochaete Chrysosporium
John Bumpus, Ming Tien, David Wright, and Steven Aust
Michigan State University 12°
Bacterial Degradation of Chlorinated Compounds
Paul Tomasek and A.M. Chakr.abarty
University of Illinois at Chicago . . . .
Techniques for Microscopic Studies of Solidification Techniques
H.C. Eaton, M.E. Tittlebaum, and F.K. Cartledge
Louisiana State University • «
127
135
USEPA Combustion Research Facility Permit Compliance Test Burn
Richard Carnes, U.S. Environmental Protection Agency. ....... 143
Engineering Analysis of Hazardous Waste Incineration: Failure
Mode Analysis for Two Pilot Scale Incinerators
W.D. Clark, J.F. LaFond, O.K. Moyeda, W.F. Richter, and
W.R. Seeker
Energy and Environmental Research Corp.
C.C. Lee, U.S. Environmental Protection Agency 150
Examination of Fundamental Incinerability Indices for Hazardous
Waste Destruction
Barry Dellinger, John Graham, Douglas Hall, and Wayne Rubey
University of Dayton Research Institute . 160
An Overview of Laboratory- and Bench-Scale Research in Hazardous
Waste Thermal Destruction
George Huffman and Chun Cheng Lee
U.S. Environmental Protection Agency. .... .
171
-vii-
-------
CONTENTS (continued)
Page
A Laboratory Study on the Effect of Atomization on Destruction
and Removal Efficiency for Liquid Hazardous Wastes
John Kramlich, Elizabeth Poncelet, W. Randall Seeker,
and Gary Samuel sen
Energy and Environmental Research Corp
Evaluation of a Pilot-Scale Circulating Bed Combustor with
a Surrogate Hazardous Waste Mixture
Daniel Chang and Nelson Sorbo
University of California at Davis
182
191
Summary of Testing at Cement Kilns Cofiring Hazardous Waste
Marvin Branscome and Wayne Westbrook
Research Triangle Institute
Robert Mournighan, U.S. Environmental Protection Agency
Jon Bolstad and John Chehaske
Engineering-Science
Evaluation of Hazardous Waste Destruction in a Blast Furnace
Radford Adams, Thomas Buedel, Carol McCarthy, and
Michael Palazzolo
Radian Corporation
Field Evaluation of Surfuric Acid Regeneration Unit Burning
Hazardous Waste as Fuel
R.C. Adrian and P.K. Ouchida
California Air Resources Board
Nonsteady Industrial Boiler Waste Cofiring Tests
Robert DeRosier, Howard Mason, Ursula Spannagel, and
C. Dean Wo1 bach
Acurex Corporation
MOUSE - A Computerized Uncertainty System for Environmental
Engineering Analyses
Albert Klee, U.S. Environmental Protection Agency .' . .
Uncertainties and Incineration Costs: Estimating the Margin of Error
Gordon Evans, U.S. Environmental Protection Agency
Update on California Program to Restrict Hazardous Waste Land
Di sposal
Jan Radimsky, California Department of Health Services. .
The Thermal Decomposition Characteristics of a Simple Organic Mixture
John Graham, Douglas Hall, and Barry Dellinger
University of Dayton Research Institute'
VOST Applications at the USEPA Combustion Research Facility
Robert W. Ross, II, F. C. Whitmore, R. H. Vocque,
T. H. Backhouse, and B. M. Cottingham, Versar, Inc. . .
199
206
213
217
227
234
244
251
252
-vm-
-------
ERA'S HAZARDOUS WASTE RESEARCH PROGRAMS: 1985
(Keynote Address)
Carl R'. Gerber
Office of Environmental Engineering & Technology
U.S. Environmental Protection Agency
Washington, DC
This morning I would like to discuss
EPA's hazardous waste research programs —
in light of the recently passed Resource
Recovery and Conservation Act (RCRA)
Amendments — to highlight what I believe
are some of our successes; to touch
briefly on our ongoing and new programs;
and to share with you where I think the
Agency's hazardous waste research is going
in general terms.
First, however, I would like to
acknowledge the work of Dave Stephan and
the staff of the Hazardous Waste Engineer-
ing Research Laboratory -- HWERL, as we
call it— in putting this llth Annual
Research Symposium together. Francis Mayo
also deserves credit for his work in
previous symposia and for the early stages
of this one. As many of you probably
know, a reorganization of EPA's engineering
research program late last year basically
moved all the hazardous waste and Superfund
engineering research into one laboratory.
This meeting has become one of the
major national meetings for hazardous
waste research professionals, and certainly
reflects well on the Agency in general and
our Cincinnati-based operations in particu-
lar. I understand that although the date
has not yet been set for next spring's
meeting, the commitment has already been
made to hold the meeting. I am pleased to
hear this and hope the meetings will
continue to happen and grow in the future.
Exchange or dissemination of informa-
tion should be the goal of all government
research programs, but is particularly
important, if not critical, in a field
like hazardous waste where developments
are occurring rapidly and there is a great
need for effective means of dealing with
the problems.
You need go no further than a town
newspaper to document environmental
problems, or potential problems, caused by
the disposal of hazardous waste, and to
appreciate that this is a major national
problem. In fact, with so many stories,
it is all too easy to conclude that
nothing significant is being done, or that
there have been no technological solutions
developed to mitigate these problems.
While such feeling is certainly great for
keeping us motivated, it can lead at times
to counterproductive feelings of frustra-
tion or helplessness.
Fortunately, progress is being made.
First, the enactment into law of the RCRA
Amendments last November signaled a
whole new national commitment to preventing
the occurrence of hazardous waste problems.
This new law not only extends the existing
law, but provides a whole new impetus to
the Agency in regulating the storage,
treatment, and disposal of hazardous
wastes.
In addition to specifically directing
the Agency to address such problems
as, underground storage tanks, the use
of hazardous waste as fuels, small gener-
ator exclusions, management of used oil,
and mining and other special wastes, the
law directs the Agency to put into place
regulations that would prohibit the
disposal of certain substances in land-
fills. Such direction reflects a differ-
ent approach to waste management; one that
directs the country towards alternative
-------
hazardous waste treatment and disposal
processes, rather than land disposal, as
the ultimate answer to many waste disposal
problems. This is having a significant
effect on our research programs, which I
will address later in this talk.
But the revised law aside, what have
we learned, or what technology has been
developed, that can be viewed as success-
ful? We have, through the combined
efforts of the Agency's Office of Solid
Waste and Emergency Response (OSWER) and
the Office of Research and Development
(ORD), determined that high temperature
incineration is an acceptable disposal
option for the overwhelming majority of
liquid organic hazardous wastes. Based on
this determination, the Agency has issued
final regulations for incineration. These
regulations are intended to ensure that
disposal practices for organic hazardous
wastes are upgraded throughout the country.
We have also determined what types of
boilers are suitable for using hazardous
waste as fuels; and determined what
industrial processes are suitable for
waste disposal. Papers presenting findings
in these areas are being given at this
symposium.
We have developed, and are currently
field testing, a mobile incinerator
that appears to be a very promising
technology for treating soils, as well as
liquids, contaminated with hard-to-destroy
substances such as dloxins. We are also
actively involved in furthering the
development of biological organisms for
decontaminating soils in place. This
work, which is also the subject of papers
at the symposium, could lead to very
attractive technological options for
clean-up operations.
As all of you know, landfills and
surface impoundments have been used for
years as an inexpensive way of storing or
disposing of wastes. Through either a
lack of information or foresight, design
of many of these sites was not based on
presently accepted or even uniform
criteria. Consequently, many of today's
hazardous waste problems are related to
land disposal operations; most importantly,
contamination of groundwater.
Today, because of the research
programs carried out over the past 10
years by the Agency, we know much more
about disposal operations and what works
and what does not. We have learned and
documented that putting liquids in land-
fills is extremely risky, even in the best
designed landfills, and consequently this
is no longer allowed by the Agency. We
have also learned and documented that it
is highly probable that certain highly
mobile, toxic, or persistent chemicals
should not be placed untreated into the
land. Although work is continuing to
determine more about the effects of
placing these substances in landfills, I
believe we can safely say that many of
these chemicals will be banned from land
disposal, and that land disposal operations
in this country in five years will be
markedly different from those of previous
years. A significant reason for this will
be the result — as summarized in the
Technical Resource Documents — of work at
HWERL.
We have also witnessed significant
advances in the development of the knowl-
edge base and technology related to
clean-up of Superfund sites. HWERL has
produced a series of guidance documents
related to clean-up operations that have
been used extensively as textbooks in
major universities around the country.
These documents have also been widely used
by the remedial action industry in struc-
turing clean-up programs for Superfund and
other abandoned disposal sites.
While we have realized successes in
our programs we, of course, cannot
rest on our laurels. Given the high
visibility and magnitude of hazardous
waste problems and the substantial risks
to the environment posed, we must press
forward on many fronts.
In the time remaining, I would like
to highlight the activities in six of
our programs: the Alternative Technology
Research Program, the Land Disposal
Research Program, the Remedial Action
Technology Program, the High Temperature
Incinerator Research Program, the Large
Volume Waste Management Program, and the
Leaking Underground Storage Tank, or the
LUST, Program. This last program, tech-
nical considerations aside, will undoubt-
edly be a contender for best acronym of
the year award.
-------
ALTERNATIVE TECHNOLOGY RESEARCH PROGRAM
The 1984 amendments to RCRA require
that within 66 months — or by May 1990 —
EPA either ban the disposal of hazardous
wastes to the land or ascertain that
certain wastes or forms of wastes are
acceptable for land disposal. In order to
implement provisions of the legislation,
the Office of Solid Waste (OSW) must
determine that acceptable alternatives to
land disposal of hazardous wastes exist.
Through its alternative technology research
program, HWERL is providing support to OSW
to identify, evaluate, and develop alter-
native technologies to treat hazardous
wastes. Chief among the wastes being
studied are compounds that are very
difficult to degrade without subjecting
them to some extraordinary process. They
may be not only resistant to degradation,
but also may be inaccessible to reagents,
bacteria, etc., by virtue of being bound
in some way to the medium, or matrix in
which they exist. It therefore becomes
apparent that effective innovative treat-
ment processes must be developed that will
not only destroy these materials in bulk,
(or concentrate them so that they may be
more efficiently destroyed) but will also
treat contaminated soils, buildings, etc.,
as well.
Since, up to this point, the Agency's
regulatory process has been focused
primarily on thermal destruction and land
disposal of hazardous wastes, much less is
known about other processes that may be
used to treat these wastes. Information
is needed on what treatment processes
exist, what types of wastes can be handled
by different processes, what residuals and
environmental discharges are produced by
these processes, and what costs will be
incurred. Assessments of alternative
treatment processes are required so that
OSW can be certain that wastes that will
be banned from land disposal car> be
adequately treated.
This is a relatively new program area
for OSWER and for ORD. In Fiscal Year
1984, a program to conduct performance
evaluations of existing alternative
treatment systems was initiated. That
program's first year of major activity is
the current fiscal year. Several technolo-
gies have already been evaluated for
removal of metals from hydroxide sludges
and electrochemical and electrodischarge
machining wastes. We are also in the
initial stages of a large field assessment
project to gather envi ronmental data on
existing commercial scale treatment
facilities.
Some of our current objectives for
the alternative technology research
program are:
0 To provide performance evaluations
on a sufficient number of alterna-
tive hazardous waste treatment
systems to allow OSW t'o direct
that these systems be used instead
of, or as pretreatments for, land
disposal for selected hazardous
waste streams;
• To develop and refine physical/
chemical treatment techniques for
dilute aqueous waste streams and
those solid wastes containing
water, such as soils and harbor or
river sediments; and
t To assure the demonstration of
promising alternative hazardous
waste treatment technologies, so
that these technologies may be
recommended to and utilized by
regulatory and industrial organ-
izations.
In a related area, hazardous waste
minimization, the Agency is supporting a
small program to identify and encourage
the adoption of process and treatment
technology for reducing the generation of
hazardous waste. As regulations limiting
the characteristics and quantities of
waste come into effect, I expect this area
to become more important and the Agency's
efforts in recycling and source reduction
to increase.
LAND DISPOSAL RESEARCH PROGRAM
Our Land Disposal Research Program
has been, and continues to be, a major
part of our total program. The purpose of
this program is to produce guidance on
design, operation, maintenance, and
closure of hazardous waste treatment,
storage, and disposal facilities. An
exciting aspect of this program is the
development of user friendly, artificial
intelligence systems that will standardize
-------
the review of applications submitted to
the Agency. These systems, which are in
very early stages of development, will be
based on the experience of experts and
field proven techniques.
The Land Disposal Program currently
addresses all aspects of land disposal
operations, including such areas as cover
systems, waste modifications, waste
leaching, and liners. Some of current
objectives for this program are:
t To develop and evaluate the
effectiveness of various cover
systems in relation to their
functional requirements, for
actual field application.
• To continue to evaluate the
effectiveness of chemical stabil-
ization and encapsulatory pro-
cesses; and
• To evaluate the effectiveness of
various types of clay and membrane
liners in containing leachates.
A new objective authorized by the
RCRA Amendments is investigation into
municipal solid waste disposal, an area
where we have considerable prior experience.
Each objective mentioned above
presents a challenge that has knowledge
gaps and will present difficulties. The
biggest hurdle to overcome is field
validation, as data from field sites
are often difficult, if not impossible to
obtain. However, such data are necessary
if predictive models are to be verified
and performance proved with time. Tradi-
tionally, our longer-term research has
been measured in terms of two to three
years, but needs to be revised to an eight
to 10 year time frame if field results are
to be meaningful. Retrospective studies,
unfortunately, are either impossible to
conduct or yield data of uncertain quality.
The main thrust of this research
program to date has been with landfills.
While much of these data are transferable
to surface impoundments, surface impound-
ments offer unique problems that have not
been fully addressed. Consequently,
existing database for surface impoundments
is limited and needs to be expanded to
develop control and design technology.
Regardless of the obvious problems
created by past land disposal operations,
as long as waste is generated, there will
be a need for final disposal of materials,
even if the material is residuals from
treatment operations. For that reason we
envision a continuing highly visible land
disposal program.
LEAKING UNDERGROUND STORAGE TANK PROGRAM
Billions of gallons of petroleum
products and other hazardous materials
are stored in underground tanks. The
basic problems associated with these
storage systems are leaks, and the costs,
dangers, and potential environmental
threats associated with those leaks. The
primary causes of leaks from underground
storage tanks .are corrosion, poor fabrica-
tion and installation, and poor operating
practices. The RCRA legislation of 1984
requires EPA to promulgate regulations to
control underground storage tanks contain-
ing "regulated substances."
EPA must set final standards for
existing tanks that cover leak detection
and tank testing, record-keeping and
reporting, corrective action, financial
responsibility, and closure. Regulations
for new tanks must include design, con-
struction, installation, release detection,
and compatibility standards. Research is
required to establish a strong technical
foundation not only to meet mandated
regulatory deadlines, but also to reduce
the rate at which the environment is being
threatened. We in ORD are jointly working
with OSW in this effort. The initial
venture was a state-of-the-art document on
leak detection that is now being finalized.
Importantly, it revealed that the data
presently available and the accuracies of
various detection methods and devices are
not acceptable, leading to a need to
improve our ability to detect existing
leaks. A leak detection method evalua-
tion facility comprised of actual under-
ground storage tanks is planned for
construction in the very near future at
our Releases Control Branch located in
Edison, New Jersey.
INCINERATION AND HIGH TEMPERATURE PROCESSES
We are continuing our research
studies of existing and new thermal
destruction technologies to develop a high
-------
degree of technical understanding and
confidence in this kind of treatment
process. The program continues to investi-
gate laboratory, pilot, and full-scale
environmental performance issues on a
broad range of incineration and alternative
thermal treatment devices. A combination
of theoretical, hypotheses-based studies;
studies on pilot and full scale operations;
and overall environmental emissions work
results in a coordinated approach to the
overall technology. While the Agency's
thermal destruction program began with
early f i-el d tests in the mid-1970's, it
became identified as a formal effort in
1980, and now exists as a major continuing
segment of our hazardous waste research
operations.
Over $20 million in thermal destruc-
tion research has been conducted since
1980. A major portion of this work has
concentrated upon the assessment of the
ability of existing thermal destruction
facilities (incinerators and high tempera-
ture industrial processes) to meet RCRA
performance standards for destruction and
removal efficiencies for hazardous mater-
ials. These studies have produced an
extensive performance database, which
indicates that these facilities typically
meet RCRA standards under ideal operating
conditions.
In addition to studying existing
systems, it is necessary to conduct more
controlled studies if we are to truly
understand the thermal destruction phenom-
ena and to be able to apply technology to
a wide range of wastes in a cost-effective
and safe manner.
To this end, we have constructed or
adapted several unique research facilities
for thermal destruction studies. The
Combustion Research Facility (CRF) in
Jefferson, Arkansas provides highly
instrumented, pilot-scale incinerators for
the evaluation of the causes and conse-
quences of process failure, while keeping
actual net environmental emissions of
hazardous constituents to within RCRA
requirements. Experimental facilities at
the Center Hill Facility here in Cincin-
nati — which you can visit as a part of '
this meeting — and at the Air and Energy
Engineering Research Laboratory in Research
Triangle Park, enable the conduct of
studies essential to understanding the
basic physical and thermochemical processes
involved in effective waste destruction
and the minimization of hazardous PIC
formation.
Extensive technical support has been
provided to regulatory program offices in
the development of regulations and stand-
ards and to EPA's Regional Offices in the
issuance of permits. Handbooks and
guidance documents have been prepared on
incineration, permits, and sampling and
analytical methods for stack emissions
measurement of trace organic compounds.
In addition we have developed, as I
mentioned earlier, a mobile incinerator
at our Edison, New Jersey facility which
should enable the Agency to carry out
research and demonstration projects while
cleaning up contaminated sites.
Among the current objectives in
incineration research are:
a To assess the performance capabil-
ities (destruction and removal
efficiencies) of existing thermal
destruction facilities which have
not been evaluated (e.g., asphalt
plants and carbon furnaces) to
form the technical foundation for
Agency policies and regulations
with respect to these thermal
destruction options;
• To define easily monitored incin-
erator facility operating parame-
ters that correlate with incinera-
tor performance;
• To project sampling and analytical
methods for assessment of thermal
destruction process performance;
and
• To evaluate and develop innovative
thermal treatment and destruction
technologies that may represent
improvements to the conventional
waste incineration process or
may be applicable to wastes that
resist incineration.
LARGE VOLUME WASTE MANAGEMENT
We have recently initiated a program
that addresses environmental problems
related to the "large volume" wastes from
mining, drilling, fossil fuel combustion,
-------
cement kilns, and extracting and processing
minerals. This work, which is specifically
called for by the RCRA Amendments, will
result 1n various special reports to the
Congress. Although much work has been
done by the EPA and others to date on
mining wastes, much remains to be learned
about the other "large volume" waste
streams. Work in this area should continue
at least through 1990.
REMEDIAL ACTION RESEARCH PROGRAMS
As a part of the Agency's Superfund
Programs to clean up abandoned hazardous
waste sites, HWERL is responsible for a
research program to develop and evaluate
various remedial action technologies.
To date, EPA has identified some
18,000 potentially hazardous waste
sites. This number is expected to rise to
22,000 by the time the Agency completes
Its survey. Of those sites, 552 have been
placed on the National Priority List.
A problem that has been found at
almost all sites is contaminated soils.
Frequently, large volumes of soils with
low levels of contaminants are present;
however, in several cases the contamina-
tion is at a high concentration.
Remedial actions for contaminated
soils have usually been removal and burial
in a secure landfill; or isolation on-site,
e.g., covering with clean soil. Both of
these methods have shortcomings. -Removal
and burial has transportation-related
problems, high cost, and long-term liabil-
ities, i.e., the landfill of today may be
the Superfund site of tomorrow. Isolation
techniques are frequently only, partially
effective, and the waste is not destroyed.
Alternative methods are needed'to solve
this national problem.
In-situ 'technologies such as surfac-
tant washing, grouting, thermal fusion,
and microbial application have had very
little acceptance at uncontrolled hazardous
waste sites, because the technologies are
still in the developmental stages and/or
have not had field verification. From an
institutional standpoint, we have problems
of conducting tests on existing hazardous
waste sites because of the inaccessibility
of sites, the cost and time associated
with permitting the studies, the risks
associated with working with hazardous
waste sites, and our inability to predict
precisely the results of such field work.
As a result of all these factors, EPA, the
states, and industry are reluctant to use
cleanup methods that are unproven. Thus,
the role of ORD is to encourage and assist
the development of new and innovative
methods, and to assist in the verification
of the cost/effectiveness of these methods,
to assure their acceptability for remedial
actions at Superfund sites.
Emphasis in future years will be on
the development and/or field evaluation
of primary technologies such as thermal,
vegetative, extraction, exchange and
chemical and biological degradation.
Another need identified by the Agency
for uncontrolled sites is improved contain-
ment technologies for minimizing releases
from these sites. We are currently pur-
suing a program to evaluate the effective-
ness of containment technologies. The
immediate objective is to screen all
applicable technologies that might show
promise. The longer term objective is to
pilot- and field-scale demonstrate the most
promising technologies, or assist in such
demonstrations. The end objective is to
provide the user community with improved,
more economical, long-lasting, and low
operation and maintenance cost recommenda-
tions for containment technologies.
Clearly, for all our Superfund related
work we will draw on information and
technologies being developed to address
RCRA related problems.
CONCLUSION
Of course there are many other
programs, any one of which could be a
suitable subject for a talk such as this
one. However, I have included these six
to give you a flavor of our current and
future programs. As I indicated, papers
at this symposium will amplify and provide
details on all these areas and many others.
So where do we go from here? Without
a doubt I think public and government
interest will continue in this area. Lee
Thomas, our new Administrator, has stated
repeatedly his support of hazardous waste
programs. Recently before the-Senate
Committee on Environment and Public Works
he stated that the reauthorization of
Superfund was his top legislative goal.
-------
With such clear direction, I believe we
will continue our effort to provide the
improved databases and technology necessary
to solve our nation's hazardous waste
problems. It will not be ea'sy: 'technol-
ogy-related tests, particularly full scale,
field tests, are expensive and subject to
major failures. A better understand-
ing of the basic chemical, physical, and
biological characteristics of hazardous
wastes and how they might be rendered
"non-hazardous" is complicated by the very
complexity of these wastes and the media
in which they exist and will not come
cheaply or even quickly. Challenges await
all of us involved in hazardous waste
related research and development, not only
to provide data and technology to assure
the safe handling and disposal of these
wastes, which are an integral part of our
society, but also to do it in a cost-effec-
tive manner, and in a way that instills
public confidence. I leave that challenge
with you.
-------
EMISSION AND CONTROL OF BY-PRODUCTS FROM HAZARDOUS WASTE COMBUSTION PROCESSES
Robert A. Olexsey, George L. Huffman, and Gordon M. Evans
U.S. Environmental Protection Agency
Hazardous Waste Engineering Research Laboratory
Cincinnati, Ohio 45268
ABSTRACT
Data on emissions of products of incomplete combustion (PICs) from full scale incin-
erators and boilers burning hazardous wastes are presented. Emissions of volatile PICs
from incinerators do not exceed the limitation of 0.01 percent of input of principal or-
ganic hazardous constituents (POHCs) that has been proposed by the U.S. Environmental
Protection Agency. Volatile PIC emissions from three of the boilers tested would exceed
the proposed standard. However, for both boilers and incinerators there does appear to
be a trend toward reduced PIC emissions with increased POHC destruction and removal effi-
ciency (DRE). Combustion conditions and control technologies which may result in reduc-
tion in PIC emissions are discussed.
INTRODUCTION
Combustion is an effective method for
disposal of hazardous waste compounds.
Hazardous waste incinerators, industrial
boilers, and many industrial thermal proc-
esses have demonstrated the ability to
destroy principal organic hazardous con-
stituents (POHCs) in compliance with EPA's
required detruction and removal efficiency
(DRE) of 99.99 percent. However, it is
theoretically possible for a given combus-
tion device to eliminate POHCs but, at the
same time, produce hazardous by-products,
or products of incomplete combustion
(PICs). While emissions of such PICs are
not currently subject to regulations, they
are of concern because these compounds
would be POHCs if they were present in the
feed to the combustion device.
This paper addresses the issue of
emissions of PICs from hazardous waste
combustion processes. Data on PIC emis-
sions are presented from full-scale tests
of hazardous waste incinerators and indus-
trial boilers burning hazardous wastes.
Combustion conditions and control tech-
nologies which may result in reduction in
PIC emissions are discussed.
FORMATION OF PRODUCTS OF INCOMPLETE
COMBUSTION
The objective of incineration is to
convert organic compounds to carbon dio-
xide (C02) and water. In actual practice,
combustion is never "complete," that is
combustion by-products are produced in the
reaction. The most prevalent combustion
by-products are partial combustion prod-
ucts such as carbon monoxide (CO). In
fact, the most widely accepted measure-
ment of combustion completeness, or effi-
ciency is the ratio of C02 to C02 + CO in
the exhaust gas from a combustion reac-
tion.
In the context of incineration of haz-
ardous wastes, as regulated under the Re-
source Conservation Recovery Act (RCRA),
the focus of concern with hazardous com-
bustion by-products is limited to product
compounds that are, of themselves, hazard-
ous compounds under RCRA. Therefore, the
PICs that we are concerned with are those
combustion by-products that are listed as
hazardous wastes in Appendix VIII of 40
CFR Part 261 (1). Therefore, for purposes
of our discussion and, in a regulatory
sense, CO is not a PIC. Total unburned
hydrocarbon (THC) is not a PIC. Only
-------
fractions of THC that are listed as a RCRA
hazardous waste are PICs.
Under EPA's RCRA incinerator regula-
tions, an incineration unit must achieve a
destruction and removal efficiency (ORE)
of 99.99 percent for each principal organ-
ic hazardous constituent (POHC) in the
waste material that is fed into the incin-
erator. PICs, therefore, are compounds
present in the incinerator exhaust gas
which are not present in. the waste fed
into the incinerator, but which would be
classified as POHCs if they were present
i n the waste.
Emissions of PICs from hazardous
waste incinerators are not currently regu-
lated by the USEPA. However, in 1981, EPA
did propose a regulation on "hazardous
combustion by-products" that would limit
PIC emissions to 0.01 percent of total
POHC input to an incinerator (2). To
date, this regulation has not gone past
the proposal stage.
The mechanisms of PIC formation are
complex and not particularly well under-
stood. PICs are suspected to occur
through any one or all of three reaction
mechanisms (3):
1. Breakdown Of individual POHCs Into
PIC reaction products. This mechanism has
been documented through laboratory experi-
mentation which has found that some mate-
rials, when subjected to temperature, con-
sistently produce specific by-products. A
case in point is the production of hexa-
chloro-benzene (HCB) as a breakdown re-
action product of kepone (4).
2. Complex recombination or substitu-
tion reactions. Through this mechanism,
PICs are formed through chemical reactions
among constituents in the feed. The
reactions are enhanced through exposure of
the reactants to high temperature and to
catalytic materials which may be present
in the reaction mixture. PICs produced
through this mechanism will normally be
high molecular weight compounds.
3. Universal combustion by-products.
AIT fuels, including fossil fuels, can
produce hydrocarbon by-products. Most
fuels, including fossil fuels, contain
trace quantities of many compounds, in-
cluding halogenated materials. In any
combustion process, hazardous PICs can be
produced. Most often, these PICs are
aromatics such as benzene, but chlorin-
ated materials, particularly chlorinated
methanes, such as chloroform, can also be
found. .
Formation of PICs is the result of
some "failure" in-the combustion process.
The predominate mode of failure is most
likely inadequate exposure of the molecule
of waste material to a sufficiently high
temperature to break the molecule and its
derivatives down and then drive the com-
bustion reaction through to completion
(carbon dioxide and water). Even though
nominal combustion temperatures may appear
to be adequate, the range of temperatures
that the molecules may experience may be
such that small quantities of material may
not be completely destroyed.
Inadequate mixing is a prime cause qf
insufficient temperature and inadequate
oxygen and flame exposure for molecul.es
of waste material. Material traversing a
combustion zone in a plug flow configura-
tion is more likely to produce regimes of
low temperature and low oxygen exposure
than will occur in a well mixed reactor
with proper atomization of the waste mate-
rial.
Short-circuiting and rapid exit of
the combustion zone by waste material can
lead to a situation where the POHC expe-
riences insufficient residence time at
temperature to destroy any PICs that may
be produced. In addition, "quenching" of
the flame through cold wall effect or im-
proper feed point location for fuels can
produce a situation where reaction prod-
ucts are "frozen" into the exhaust gas
stream.
FULL-SCALE PIC .EMISSIONS DATA
EPA recently completed and reported on
a series of full-scale tests of incinera-
tors and industrial boilers burning haz-
ardous wastes (5, 6). The testing con-
ducted under these full-scale programs
was designed primarily to obtain data on
POHC ORE to support EPA's regulation de-
velopment. In addition to the POHC data,
considerable data was obtained on emis-
sions of PICs.
Table 1 describes the POHCs that were
most frequently burned in the incinerator
and boiler test programs. While eight in-
cinerators and eleven boilers were actual-
ly tested under these programs, PIC data
-------
TABLE 1
POHCs MOST FREQUENTLY BURNED IN COMBUSTORS
SEVEN INCINERATORS
Toluene
Tetrachloroethyl ene
Trichloroethylene
Carbon Tetrachl oride
Naphtha!ene
Chloroform
Methylene Chloride
•Methyl Ethyl Ketone
Phenol
FIVE BOILERS
Carbon Tetrachloride
Trichloroethylene
Chlorobenzene
Tetrachloroethylene
Toluene
1,1,1-Trichloroethane
Methyl methacryl ate
Epichlorohydri n
Bis (2-Chloroethyl) Ether
from seven incinerators and five boilers
is considered to be of sufficiently rig-
orous detail to be evaluated for analyti-
cal and comparative purposes.
Tables 2 and 3 list the volatile PICs
which were most frequently found in the
incinerator and boiler stack gases. All
PIC concentration values in Tables 2 and 3
were obtained through analysis of Tenax
resin samples of stack gas for Appendix
VIII constituents through gas chromato-
graphy/mass spectrometry. Volatile PICs
(boiling point
-------
TABLE 2
VOLATILE PICs MOST FREQUENTLY PRESENT IN BOILER STACK GASES
PICs
Chloroform
Tetrachl oroethyl ene
Chloromethane
Methyl ene Chloride
Benzene
1,1,1-Trichloroethane
1,2-Dichloroethane
NUMBER OF FACILITIES
5
5
4
4
3
3
3
CONCENTRATION
4.2 -
0.3 -
4.6 -
RANGE (ng/L)
1900
760
410
83 - 2000
9.4 -
5.9 -
1.3 -
270
270
1200
TABLE 3
VOLATILE PICs MOST FREQUENTLY PRESENT IN INCINERATOR STACK GASES
PIC
Benzene
Chloroform
Tetrachl oroethyl ene
1,1,1-Trichloroethane
Toluene
Methyl ene Chloride
NUMBER OF FACILITIES
6
5
3
3
2
2
CONCENTRATION
12 -
1 -
0.1
0.1
2
2
RANGE
670
1330
- 2.5
- 1.5
- 75
- 27
(ng/L)
which was oil for sites D, E, and F and
pulverized coal for site H. Emissions of
PICs were not characterized for the base-
line tests in which only fossil fuels were
burned. Finally, for both incinerators
and boilers, compounds that were present
in the feed in concentrations less than
100 parts per million (ppm) were not con-
sidered to be POHCs. It is possible that
the higher PIC emissions from the boilers
could be attributed to compounds that were
present in the feed in less than 100 ppm
concentrations. It is possible, although
it cannot be ascertained, that the boiler
wastes contained more of these low concen-
tration feed compounds than did incinera-
tors.
Figure 1 represents the normalized PIC
output data with respect to POHC ORE. The
abscissa axis of the plot in Figure 1 dis-
plays total POHC ORE for the incinerators
and the boilers. The ordinate axis dis-
plays the ratio of volatile PIC emissions
to total POHC input for all the sites.
The numerical designations and the symbol
e refer to incinerator sites and the
alphabetical designations and the symbol A
refer to boiler sites. The vertical dash-
ed line represents the ORE requirement for
POHC destruction of 99.99 percent from the
existing RCRA incinerator regulations.
The horizontal dashed line indicates the
proposed PIC limitation of 0.01 percent of
total POHC input.
From Figure 1, a rough trend can be
perceived of reduced PIC to POHC emissions
11
-------
TABLE 4
INCINERATOR SUMMARY OF AVERAGE PIC AND POHC VALUES
SITE
1
3
4
5
6
7
8
TOTAL VOLATILE
PIC EMISSIONS
(mg/min)
68.00
9.00
0.57
1.36
0.23
228.00
0.71
TOTAL POHC
INPUT
(g/min)
9818
780
31
1500
323
8287
1098
TOTAL POHC
EMISSIONS
(mg/min)
76.22
48.00
1.09
0.59
6.60
58.61
56.40
MASS WEIGHTED
POHC ORE
(Percent)
99.9992
99.9938
99.9964
99.99996
99.998
99.9993
99.9949
TABLE 5
BOILER SUMMARY OF AVERAGE PIC AND POHC VALUES
SITE
D
E
F
G
H
TOTAL VOLATILE
PIC EMISSIONS
(mg/min)
205.70
567.36
546.20
298.40
594.00
TOTAL POHC
INPUT '
(g/min)
5945
2605
137
9436
1020
TOTAL POHC
EMISSIONS
(mg/min)
37.80
98.37
25.60
13.26
96.00
MASS WEIGHTED
POHC DRE,
(Percent)
99.9994
. . 99.996
99,981
99.9999
99.9"91
TABLE 6 j
INCINERATOR PIC RATIOS
SITE
1
3
4
5
6
7
8
100 x VOLATILE PIC EMISSIONS
TOTAL POHC INPUT
(PERCENT)
0.0007
0.0011
0.0018
0.0009
0.00007
0.0028
0.00007
VOLATILE PIC EMISSIONS
TOTAL POHC EMISSIONS
0.89
.0.19
0.52
2.31.,
0.03
'' 3.89
0.01
12
-------
TABLE 7
BOILER PIC RATIOS
SITE
D
E
F
G
H
100 x VOLATILE PIC EMISSIONS
TOTAL POHC INPUT
(PERCENT)
0.0035,
0.0218
0.3987
0.0032
0.0582
VOLATILE PIC EMISSIONS
TOTAL POHC EMISSIONS
5.44
5.77
21.33
22.50
6.19
ratio with increased POHC ORE. In order
to test the existence of such a trend, a
computer was employed to conduct a simple
ordinary least squares regression analysis
with the POHC ORE as the predictor vari-
able (X) and the PIC emission/POHC input
ratio as the response variable (y). In
order to avoid problems with the computer
algorithm, the ORE data points were trans-
formed by subtracting the value 99.0 from
each ORE, thus running the regression
analysis utilizing just the decimal infor-
mation. The .resulting regression equation
was y = 0.18862 -'0.18902x .with an R?
(adjusted for degrees of'Freedom) of 79
percent, confirming this trend. The trend
is more evident for the boilers than for
incinerators. Point F is the only site
that exhibits both a POHC ORE below the
99.99 percent and a PIC/POHC input ratio
above 0.01 percent. Points E and H repre-
sent sites where ORE was i n excess of
99.99 percent but the PIC/POHC input ratio
was above 0.01 percent for volatile pICs,
All the incinerator sites were above the
required ORE and below .the proposed PIC
limitation. Points 6 and 8 represent
sites where absolute values of volatile
PIC emissions Were extremely low, 3 ng/L
and 5 ng/L, respectively.
CONTROL OF PRODUCTS OF INCOMPLETE
COMBUSTION
Much of the above deals with the
amounts and types of,PICs that are formed
in various hazardous waste thermal de-
struction operations. But, perhaps a more
pertinent question is "Given that PICs do
form, what do we do about them:—i.e.,
how do we control them?" Little is known
about how to answer this question. Data
relative to how to control these poten-
tially hazardous air pollutants is severe-
ly lacking. A significant amount of
research is needed to fill this void.
Nevertheless, some general statements can
certainly be made.
Methods that have been proposed as
potentially worthwhile, efficient, control
techniques for PIC removal include: (1)
Optimization of the combustion process
itself to minimize the formation of PICs
(2) Minimization of the amount of "cold"
heat transfer surfaces that the exhaust
gases see upstream of the "PIC Destruction
Section" in the combustor; (3) Afterburn-
ing; (4) Enriched oxygen utilization; (5)
Catalytic oxidation; (6) Scrubbing of the
organic (gaseous, liquid or solid) PICs
using organic scrubbing media and effi-
cient contactors; and (7) Sorption of the
organic PICs onto beds of activated carbon
or other effective sorptive material. The
first five of these are, of course, high
temperature methods, whereas the last two
are (generally) low temperature processes.
Regarding the optimization of the com-
bustion process, suffice it to say that
operation in the air-to-fuel regime where
resultant PIC levels are low is a very
reasonable objective as long as that re-
gime is also the place where optimum POHC
destruction occurs. A significant amount
of research is needed here (on pilot and
larger scales) to define the bounds of
this regime as a function of combustor
type.
Regarding the limiting of the amount
13
-------
FIGURE 1
10-2
10-3
10-4
Q.
in
o
o.
UJ
10-6
10-7
10-8
99.0
99.9
9
A
Incinerators - Numerical Site Designations
Boilers - Alphabetical Site Designations
A
F
A
|H
.!_
PROPOSED PIC
LIMITATION" "
0A
0 0 7D
3 4
0
1
0 0
8 6
A
G
0
5
o
a;
a cz.
i—i
£|g-
O UJ
a. cc.
99.99
99.999
99.9999 99.99999
POHC ORE, PERCENT
PLOT OF POHC DRE VS. RATIO OF VOLATILE PICS TO POHC INPUT
of "cold" surfaces that an exhaust gas
sees upstream of the PIC destruction sec-
tion, this is of course relevant primarily
to new combustor designs; it would have
limited applicability to retrofit designs.
Regarding the use of afterburning as a
PIC control technique, this established
practice should be quite useful1 in de-
stroying some of (perhaps most of) the
PICs that are formed in the primary com-
bustor. However, this needs to be veri-
fied through additional research aimed at
determining just how effective the various
conventional afterburner des.igns are at
destroying PICs.
Relative to the use of enriched oxy-
gen as a PIC co.ntrol technique, the
applicability and effectiveness of this
14
-------
method needs to be determined. Conceptual-
ly, the use of higher concentrations of Oa
in tearing apart complex organic molecules
during combustion does seem to make sense.
Nonetheless, experimentation is needed to
prove whether this conjecturing is valid.
The authors feel that the catalytic
oxidation option is particularly intrigu-
ing. This technique has enjoyed wide-
spread success in many other applications.
Directing combustor off-gases through a
catalyst bed especially designed for gross
PIC oxidation makes a considerable amount
of sense, particularly for new combustor
designs. Now, of course, "all" that is
needed is for some process developer to
find or develop an extremely versatile,
low-cost, poison-resistant, sturdy, highly
effective PIC oxidation catalyst — the
rest will be easy.
Regarding the (probable) low tempera-
ture option of scrubbing with organics,
this technique can surely be developed if
enough research is devoted to it. How-
ever, it does have the inherent draw-back
of having to treat the scrubber liquors to
remove the absorbed PICs prior to their
final disposition.
Finally, relative to the dry sorption
onto carbon option, this should be readily
implemented and reasonably effective re-
garding PIC removal. Once again, however,
this option does need additional research.
CONCLUSIONS
From the data presented here, it is
apparent that the boilers tested had high-
er'PIC emission levels and ratios than did
the incinerators tested. For all combus-
tors tested, there appears to be a trend
toward reduced PIC emissions with in-
creased POHC ORE.
Very little is known about the mech-
anisms of PIC formation in combustors.
Similarly, little is known about the re-
gimes of operation that tend to either
produce or minimize PIC emissions from
burners.
There is a need for more research into
PIC generation mechanisms and PIC control
technologies. Near-term research should
be conducted at laboratory scale under
ordered experimental protocols with con-
trolled variables. In the longer term,
additional full-scale testing that is
directed toward proving out the small-
scale hypotheses should be conducted.
REFERENCES
1. U.S. Federal Register, 40 CFR, Part
261, Volume 45, No. 98, May 19, 1980.
2. U.S. Federal Register, 40 CFR, Parts
264 and 265, Volume 46, No. 15, Janu-
ary 23, 1981.
3. Trenholm, A., et al.', "Products of In-
complete Combustion from Hazardous
Waste Incinerators," in Incineration
and Treatment of Hazardous Waste,
Proceedings of the Tenth Annual Re-
search Symposium, September 1984.
4. Duvall, D. S., et al., "Laboratory
Evaluation of High Temperature De-
struction of Kepone and Related Pes-
ticides," Report to USEPA, EPA-600/2-
76-299.
5. Trenholm, A., et al., "Performance
Evaluation of Full-Seale Hazardous
Waste Incinerators," Report to USEPA
under Contract No. 68-02-3177, 1984.
6. Castaldini, C., et al., "Engineering
Assessment Report Hazardous Waste Co-
firing in Industrial Boilers," Report
to USEPA under Contract No. 68-02-
3188, June 1984.
15
-------
PRACTICAL GUIDE TO TRIAL BURNS AT
HAZARDOUS WASTE INCINERATORS
Paul Gorman
Midwest Research Institute
425 Volker Boulevard
Kansas City, Missouri 64110
Donald Oberacker
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268
ABSTRACT
The U.S. Environmental Protection Agency contracted with Midwest Research Institute
(MRI) to prepare a manual, "Practical Guide to Trial Burns for Hazardous Waste Inciner-
ators," based on MRI's experience in conducting trial burns. Directed mainly to incin-
erator operators, the guide addresses the trial burn process, including planning and prep-
aration, sampling and analysis, process monitoring, and data reduction and reporting.
The guide concentrates on the most important, as well as the more troublesome, aspects
of a trial burn. This paper is a condensation of some of the contents of the guide rela-
tive to planning for a trial burn and to problems that may be encountered when actually
conducting the trial burn.
INTRODUCTION
The Resource Conservation and Recovery
Act requires that hazardous waste inciner-
ators effectively destroy hazardous organic
compounds and maintain acceptable levels
of particulate and chloride emissions.
Owners and operators of these incinerators
must demonstrate the performance of their
facility by means of a trial burn. Conse-
quently, industry and control agency per-
sonnel have become involved in planning
for, conducting, and interpreting the re-
sults from trial burns.
Midwest Research Institute (MRI) has
prepared a document, under contract to the
U.S. Environmental Protection Agency (EPA),
titled "Practical Guide - Trial Burns for
Hazardous Waste Incinerators." This paper
explains the purpose of the guide, de-
scribes the approach used in its prepara-
tion, and discusses selected areas and po-
tential problems covered in the guide.
Planning for and conducting a trial burn
are emphasized in the paper. Several exam-
ples are included.
PURPOSE
The purpose of the "Practical Guide"
is to assist incinerator operators and reg-
ulatory personnel who will be involved in
trial burns but who may not have had ex-
perience in planning and executing a trial
burn. The guide's intent is to convey in-
formation about the many complex and var-
ied aspects of a trial burn, primarily
those that are considered most important
and those that frequently cause problems.
APPROACH
Experienced personnel at MRI first
prepared a list of the most important parts
of a trial burn, from planning through final
reporting of results, and the most common
problems that have occurred. Narrative
descriptions of each identified part or
problem were then written to describe how
each could be handled and how problems
could be avoided or minimized. The de-
scriptions were kept brief in order to en-
hance clarity and conciseness, and tables
16
-------
and figures were used whenever possible.
RESULTS
Anyone who is familiar with trial
burns knows that there are many complex
and varied steps involved, and numerous
potential difficulties. The guide, conse-
quently, was directed to the more impor-
tant aspects of a trial burn and the prob-
lem areas that have been experienced.
The guide has a question-and-answer
format to make it easier to use and more
directly address the concerns of those who
will be involved in trial burns. The ques-
tions asked and answered in the guide are
listed below.
Planning for a Trial Burn
What equipment or instrumenta-
tion is the incinerator required
to have?
How should trial burn operating
conditions be selected?
What types and quantities of
waste are needed?
How many runs are necessary?
How many workers will be needed?
How are stack sampling methods
selected?
What detection limits are re-
quired for the sampling and anal-
ysis methods?
What quality assurance/quality
control (QA/QC) measures need to
be taken?
What if the results do not meet
Resource Conservation and Re-
covery Act (RCRA) requirements?
in preparing
Conducting Trial Burns
What is involved
for the tests?
What is involved in actual sam-
pling?
What is involved in analysis of
samples?
How are data converted to final
results?
How are the results usually re-
ported?
Since all these questions cannot be
answered in this paper, the most important
areas and problems will be summarized in
the following paragraphs, along with exam-
ples and some items of advice. Those re-
lated to planning will be discussed first,
followed by those related to conducting
the actual trial burn.
Planning for a Trial Burn
Much is involved in planning for a
trial burn. Many aspects of the planning
may be known, but their magnitude or com-
plexity may not be realized. In addition,
other aspects that need to be included in
the planning activity are sometimes over-
looked. Experience has shown that some
of the most important or troublesome as-
pects in planning for a trial burn are:
Selection of operating conditions,
Pretests,
Number of runs and quantity of waste
required,
Time requirements,
Sampling and analysis methods, and
Cost of a trial burn.
These aspects, which are discussed in de-
tail in the guide, are summarized below.
Selection of operating conditions:
The objective of a trial burn is to demon-
strate compliance with the following RCRA
requirements:
Destruction and removal effi-
ciency (ORE) ^ 99.99% for the
selected principal organic haz-
ardous constituents (POHC)
Particulate emission concentra-
tion ^ 180 milligrams per dry
. standard cubic meter (mg/dscm)
HC1 emissions ^1.8 kilograms
per hour (kg/hr), or removal
efficiency ^ 99%
A major problem that is often slow
to be realized is that these RCRA require-
ments need to be demonstrated under the
worst practical operating conditions
(worst case) because the operating condi-
tions and waste characteristics during
the trial burn usually become the limits
specified in any subsequent operating per-
mit. Some worst case conditions are listed
in Table 1. It is difficult to achieve
all the criteria shown in Table 1 because
some are contrary to others (e.g., maximum
air flowrate with minimum 02 content).,
17
-------
TABLE 1. TRIAL BURN CONDITIONS
Waste Characteristics
- Maximum concentration of selected
POHCs
- Maximum Cl content
- Maximum ash content
- Minimum heating value (HHV) of
waste feed
Operating Conditions
- Maximum heat input rate
-. Minimum combustion temperature
- Maximum waste feed rate
- Minimum 02 concentration in stack
gas
- Maximum air input rate (maximum gas
flowrate to yield minimum resi-
dence time)
- Maximimum CO content in stack gas
When all, or most, of the worst case
scenario is achieved, it will often repre-
sent an operating situation considerably
different from normal operation. Burning
a low heating value waste with high Cl and
ash content at maximum feed rate, while
trying to maintain all the other worst case
conditions, has led to trial burn problems,
if not inadvertent shutdowns. A common
problem when burning high ash content li-
quid wastes has been frequent plugging of
strainers. This has caused problems in
maintaining maximum waste feed rate and
even caused shutdowns during the'trial burn.
In another case, an incinerator operator
burning a specific waste wanted to conduct
one trial burn at maximum waste feed rate
and minimum temperature. However, the op-
erator realized that the incinerator always
operated at lower temperature for lower
feed rates. Revised operating procedures
and a minor design change had to be made
in order to. operate at the higher tempera-
ture when burning waste at lower feed rates.
Realization of what must be done and
good planning help minimize problems caused
by the worst case situation represented in
a trial burn. Wherever possible, the plant
should be operated at the test conditions
sometime prior to the actual trial burn.
Pretests: The major problem with op
erating at worst case conditions is that
it maximizes the chance of failure (not
meeting RCRA requirements). Since the
plant wants to meet RCRA requirements,
test conditions must be carefully se-
lected, and plant operating experience is
very important in making those selections.
Pretests or miniburns are also highly de-
sirable before the actual trial burn be-
cause they identify problem areas, one of
the more common being failure to achieve
the particulate emission limit, rather
than failure to meet ORE. Mist carryover
from alkaline scrubbing systems has often
been an important factor in failure to
meet the particulate limit. It is much
better to discover such problems during a
pretest, than to fail the official trial
burn.
Number of runs and quantity of waste
required:Other elements in the trial
burn that have caused problems have to do
with the number of runs and the quantities
of waste required. EPA recommends three
runs (at each set of operating conditions)
for a trial burn. What is not usually
recognized is that the operator should
plan one run per day and that each run
may require 8 hr of operating time. The
quantities of waste feed necessary for.
three 8-hr runs are often large. If a
plant is burning liquid waste at a rate
of 5 gallons per minute (gpm), then the
total amount required would approach
7,200 gallons. If, in addition, solids
are being burned at a rate of one drum
every 5 minutes, then a total of 288 drums
may be needed. All these wastes must
possess the characteristics required for
the trial burn, and must be acquired and
stored well in advance of the tests.
Time requirements: Many parts of
the trial burn process require good plan-
ning and sufficient time for carrying out
those plans. A list of the time factors
is given in Table 2. Preparations for
the actual sampling require considerable
time, and analysis of samples after the
tests usually requires at least 1 to
1-1/2 jnonths, due mainly to complexities
of POHC analysis. The analysis time is
even longer when polychlorinated biphenyls
(PCB) or dioxins must be analyzed in a
PCB trial burn, under Toxic Substances
Control Act regulations.
18
-------
TABLE 2. TIME FACTORS INVOLVED IN A TRIAL BURN
Receive notification to submit Part B application.
Evaluate all conditions at which plant desires to be permitted (1 month).
Prepare trial burn plan and submit to EPA (required 6 months after notification).
Prepare responses to EPA on any questions or deficiencies in the trial burn plan
(1 month).
Make any additions or modifications to the incinerator that may be necessary (1 to
3 months).
Prepare for trial burn.
* Prepare for all sampling and analysis (S&A) (2 to 3 months).
* Select date for trial burn, in concert with S&A staff or contractor '(completed
1 month prior to test).
* Notify all appropriate regulatory agencies (1 month).
* Obtain required quantities of waste having specified characteristics.
* Calibrate all critical incinerator instrumentation (2 weeks).
Conduct trial burn sampling (1 week).
Conduct sample analysis (1 to 1-1/2 months).
Calculate trial burn results (1/2 month).
Prepare results for .submittal to EPA (1/2 to 1 month). Include requested permit
operating conditions.
Obtain operating permit.
. Sampling and analysis methods: Sam-
pling and analysis results are the basis
on which it is decided whether the incin-
erator did or did not achieve the RCRA re-
quirements. It is for this reason that
part of the "Practical Guide" is devoted .
to POHC sampling and analysis methods and
procedures for estimating the detection
limits necessary to demonstrate a ORE of
99.99%. The sampling and analysis that
are required in a trial burn, and the asso-
ciated QA/QC procedures, are among the most
complex parts of the trial burn. Much of
the guide is devoted to sampling and anal-
ysis, and covers many of the important
aspects thereof, in a manner that, we hope,
is relatively easy to understand, even
though it must necessarily be technically
oriented. However, because of their com-
plexity, they are only briefly discussed
in this paper.
Selection of sampling and analysis'
methods depends, of course, on the se-
lected POHCs. One factor that is often
not recognized is that the POHCs fall into
the following three groups, that involve
different sampling and analysis methods:
19
-------
Volatile POHCs
Semivolatile
POHCs
Other POHCs
Sampling and
Analysis Method
Volatile organic sam-
pling train (VOST) and
gas chromatography/mass
spectrometry (GC/MS) or
gas bag and GC/MS
EPA Modified Method 5
(MM5) and GC/MS
Special methods
When selecting POHCs, it is advanta-
geous if all are in either Group A or
Group B, since this will require only one
sampling method and therefore mean reduced
analytical costs. It is advantageous in
most cases to avoid POHCs in Group C (e.g.,
formaldehyde).
After the POHCs have been selected,
the detection limits required must be esti-
mated to assure quantisation at the ORE
level of 99.99%. A useful rule of thumb
discussed in the guide is as follows:
100 ppm in waste feed =
1 ug/m3 in stack gas,
at 99.9% ORE
Higher concentrations in waste feeds
yield proportionally higher concentrations
in the stack gas. This rule of thumb is
useful for estimating concentrations of
POHCs in incinerator stack gas and for se-
lecting the sampling and analysis methods
with appropriate detection limits.
The VOST and MM5 methods usually can
achieve the required detection limit, as-
suming no special interferences. However,
the upper detection limit for VOST can be
exceeded when the POHC concentration in
the waste is high, even if the incinerator
is achieving 99.99% ORE. When this happens,
the ORE may have to be reported as a value
like "< 99.995%." Clearly such a result
is of no use. The solution to this poten-
tial problem is to estimate the stack con-
centration assuming a 99.99% ORE. Then,
if the estimated concentration exceeds the
VOST capability, gas bags will need to be
used.
In general, gas bags should be used
whenever the estimated stack concentration
(at 99.99% ORE) exceeds 500 (jg/m3. Refer-
ring to the rule of thumb just described,
this could occur whenever the POHC concen-
tration in the waste feed is 50,000 ppm
(5%) or higher. Consequently, gas bags
are often needed and are frequently used
in addition to VOST.
Cost of a trial burn: One of the
most surprising aspects of a trial burn,
regardless of whether or not one uses a
contractor, is the cost of sampling and
analysis. In general, the cost ranges
from $30,000 to $150,000 depending on the
number of runs, number of samples taken
in each run, and analysis required on each
sample (i.e., the specific POHCs involved).
The number of samples that are taken dur-
ing a trial burn, including all replicates
and blanks, is surprisingly large (in the
range of 100 to 300). Each must then be
multiplied by the respective number of
analyses required. For example, each
waste feed sample may be analyzed for
heating value (HHV), Cl, ash, viscosity,
and all POHCs.
The large number of samples and the
number of analyses of each constitute the
major parts of the cost for a trial burn.
Cost is also significantly affected by
the QA/QC that is specified.
Quality assurance/quality control:
QA/QC must be adequate, but not excessive
because of its potential impact on cost.
More important, the QA/QC must be speci-
fic, clearly identifying:
Number and types of samples to be
analyzed in replicate;
Number and types of samples to be
spiked with POHCs or surrogates
to assess recovery efficiency
(accuracy);
Number and types of blanks;
Number and types of calibration stan-
dards; and
Number and types of audit samples.
Conducting a Trial Burn
Some of the planning aspects for a
trial burn discussed previously were di-
rected toward avoidance of problems that
20
-------
can occur in actually conducting the trial
burn:.,, First, of course, was having all
the-waste available with the desired char-
acteristics,. Second, was being able to
operate the plant at the desired worst
case conditions for the trial burn period.
Other significant problems that can develop
during actual trial burns have to do with
the following:
Stack sampling facilities,
Cyclonic flow,
Recording of process data,
Documenting samples and analysis, and
Calculating and reporting of results.
These aspects, which are discussed in the
"Practical Guide," are summarized below.
Stack sampling facilities: Stack sam-
pling facilities must be adequate for all
the test methods and equipment. Usually
the degree of adequacy is determined during
pretests or a pretest site survey by a sam-
pling and analysis contractor. Even then,
inadequate facilities often causes several
hours of delay in starting the actual trial
burn sampling. Lack of adequate electrical
outlets and sufficient power for all the
stack sampling equipment is the most common
problem encountered.
Another problem the incinerator opera-
tors need to be aware of is that the trial
burn requires a great deal of sampling
equipment, which can have mechanical prob-
lems that cause test delays, especially if
provisions have" not been made for backup
equipment and spare parts. More critical
is the fact that the stack sampling equip-
ment (Method 5) must pass a final leak
check after a run has been completed. If
a substantial leak is detected the run may
be invalid and have to be repeated. This
will mean an additional run, requiring addi-
tional quantities of the waste feed and
additional time.
Cyclonic flow: One of the first checks
done just prior to the actual trial burn
sampling is a check for cyclonic flow in
the stack. Cyclonic flow is not commonly
encountered, but when it is, then installa-
tion of flow straighteners in the stack
may be necessary. This is not a simple
task and may require several days, if not
weeks, to do. That possibility is another
reason for a pretest, well in advance of
the actual trial burn tests.
Recording of-process data; Careful
planning for all. .process data to be re- _ .
corded, and by whom, "is another seemingly •
minor but, important part of the actual
trial burn. Usually these recorded data:
form a basis for ORE calculations (e.g.,
waste feed rates) and for subsequently
specified permit operating limits. Close
attention to the desired process operat-
ing conditions versus actual conditions
during the tests is essential, along with
actions to be taken if actual operating
conditions are not within preselected op-
erating limits. The advice here is to
consider the "what if" situations and de-
velop clear action plans. For example,
if the operating temperature is to be held
at 2000° ± 100°F, is sampling to be inter-
rupted whenever the temperature goes out- •
side that range (e.g., 1890°F)? Another
example: if there is a momentary flameoutj
what action is to be taken? Examples like
these may seem minor but many such inci-
dents have taken place, and have caused
difficulties during actual tests because
there is then little time available for
discussion.
Documenting samples and analysis:
Many samples are taken during a trial burn
and most of these samples must undergo
multiple analyses and QA/QC. Therefore,
an important requirement after the actual
sampling is listing every sample and spe-
cifying exactly what analyses are to be
performed on each sample and by what
methods. That list and its requirements
must be distributed to all those who have
responsibility for analyses and final re-
sults. Without such a comprehensive list,
it is easy to overlook some required anal-
ysis or step.
Calculating and reporting of results:
The last part of the trial burn process .
is converting analytical results into final
results for the test report. This involves
several calculations which can cause prob-
lems or questions related to:
Blank correction procedures,
Rounding and significant figures,
Use of less-than or greater-
values, and ..
Formatting of reported results.
The last section of the guide dis-
cusses all the above, together with recom-
mended methods, and presents several sample
tables of data to show how results are
calculated, and how they may be presented.
21
-------
CONCLUSION
We hope that the conclusion to be
drawn is that the Practical Guide is worth
reading, because it identifies problems
that have been encountered in actual trial
burns and the advice given in the manual
may help minimize or avoid those problems
in future trial burns. The Guide, which
is in draft form and nearing completion of
EPA review, should be available in a few
months.
22
-------
Carbon Monoxide and ORE: How Well Do They Correlate?
Laurel J. Staley
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
ABSTRACT
A series of six experiments were conducted in which five different organic compounds
diluted in heptane were burned in a water-jacketed, bench-scale combustor. Excess air
levels were varied to produce changes in the carbon monoxide (CO) concentration in the
incinerator exhaust. Tenax trap samples were taken at various CO levels for purposes
of determining accompanying Destruction and Removal Efficiencies (DREs) and levels of
Products of Incomplete Combustion (PICs).
Results indicate that CO levels have only a tenuous relationship with ORE and vary
with the compound being burned as well as with excess air and other combustion conditions.
The highest levels of PICs and unburned POHCs (Principal Organic Hazardous Constituents)
occurred at very high excess air levels and were accompanied by high CO levels.
INTRODUCTION
The Resource Conservation and
Recovery Act (RCRA) regulations require
that hazardous waste incinerator opera-
tors achieve 99.99% ORE (Destruction and
Removal Efficiency) of the hazardous
waste fed to the incinerator. Compli-
ance with these regulations is generally
determined by a trial burn in which the
incinerator burns a hazardous waste
stream representative of what the in-
cinerator is intended to burn. Inlet
and outlet mass flow rates for several
selected chemicals, POHCs (Principal
Organic Hazardous Constituents)., chosen
because they are thought to be relative-
ly hard to burn, are determined. From
these mass flow rates, the DREs of the
POHCs are determined.
The trial burn method for determin-
ing compliance with RCRA has several
problems. In spite of the fact that
trial burns are expensive and time con-
suming, they provide only a "snapshot"
of how well the incinerator is operating
during the trial burn. No information
is obtained about how the incinerator's
performance might fluctuate with future
changes in operating conditions or waste
feed characteristics. Further, trial
burns do not provide the incinerator
operator with up-to-the-minute informa-
tion on incinerator performance which
would enable corrective action to be
taken if incinerator performance deteri-
orates .
Several research groups are trying
to correct this deficiency by determining
whether a continuous indicator of perfor-
mance, which would signal the onset of
incinerator performance deterioration,
can be found. These potential indicators
are combustion intermediates or tracers
spiked into the waste feed. The emission
rates of these chemical intermediates
and/or tracers can all be monitored.
Theoretically, fluctuations in emission
rates could then be used to signal
changes in incinerator performance.
Research is ongoing to determine the
nature of the relationship between the
emission rates of these chemicals and
ORE, if, indeed, such a relationship
exists.
23
-------
One combustion intermediate under
consideration as a ORE indicator is
carbon monoxide (CO). Emission levels
of carbon monoxide have long been used
by operators of combustion devices to
indicate performance levels in a very
general sense. High CO levels are
generally considered to indicate poor
combustion and general operating prob-
lems with the incinerator. Conversely,
low CO levels generally indicate that
the combustor is operating well. In
addition, inexpensive readily available
and reliable monitors already exist for
monitoring CO. Thus, because it seems
economically and technically feasible,
research is ongoing to attempt to tie
ORE directly to CO levels.
Thus far, results indicate that
the correlation between CO and ORE
exists in the laboratory and in small-
scale devices but not at full-scale.
Researchers at the Energy and Environ-
mental Research Corporation (EERC) in
California have found that CO does not
indicate minor changes in ORE, but does
indicate general trends. In predicting
general trends, EERC found that CO is a
conservative indicator of ORE. That
is, that CO increases long before ORE
drops (2). However, full-scale tests
at both hazardous waste incinerators
and industrial boilers cofiring hazard-
ous waste indicate that there is no
apparent correlation between CO and ORE
0, 3).
Consequently, a set of experi-
ments was conducted at the EPA-
Cincinnati Thermal Destruction Labora-
tory (located at their Center Hill
Facility) to further explore not only
how POHC destruction varied with CO
levels, but how the formation of Products
of Incomplete Combustion (PICs) varied
as well. We chose to study PIC formation
in addition to POHC destruction because
excessive PIC formation in hazardous
waste incinerators can perhaps be a
significant source of air pollution from
these devices.
EXPERIMENTAL EQUIPMENT
The combustor used for the in-
house EPA work is nearly identical to
that used in the laboratory-scale
research mentioned above which showed a
correlation between CO and ORE. Figure
1 is a diagram of the Turbulent Flame
Reactor (TFR) used in the in-house study.
The TFR is water-jacketed and made of
stainless steel. A mixture of test com-
pound and heptane enters the TFR through
a Delavan pressure atomizing nozzle.
Nitrogen pressurization from bottled gas
is the driving force for fuel flow. Com-
bustion air supplied by a compressor
enters the TFR through an International
Flame Research Foundation (IFRF) windbox
located at the bottom of the TFR. Adjust-
able swirl vanes allow for the control of
angular and linear momentum imparted to
the combustion air and afford some con-
trol over flame shape and stability.
Tenax-GC sorbent traps are used to
collect POHC and PIC emissions from the
TFR. The tenax sampling system used is
shown in Figure 2. Sample gases enter
through a 1/4" outside diameter (O.D.)
stainless steel probe which penetrates
the TFR exhaust duct. They then pass
through a heat-traced length of 1/4" O.D.
teflon tubing where the temperature is
kept at 130°C. Prior to leaving the heat-
traced length of tubing, sample gases
pass through a particulate filter. Heat
tracing ends roughly six inches upstream
of the Tenax trap in order to allow the
gases to air-cool from 130°C to 20°C.
The gases then pass through a Tenax trap
containing 1.5 grams of Tenax-GC resin
which adsorbs all of the volatile organ-
ic compounds in the exhaust gas sample.
After that, the gases pass through a
pump, a rotameter and a dry gas meter
prior to being exhausted to the atmos-
phere. The Tenax trap samples collected
in this way were then analyzed by a gas
chromatograph equipped with a Hall
Detector.
To obtain continuous CO level read-
ings, a Beckman Non-Dispersive Infra-Red
(NDIR) monitor is used. A Beckman NDIR
monitor is also used to continuously
monitor C02 concentrations in the ex-
haust gas. Oxygen in the exhaust gas
is monitored by a Beckman Paramagnetic
oxygen monitor. Total Unburned Hydro-
carbons (TUHC) are continuously moni-
tored using a Beckman Flame lonization
Detector (FID) Total Hydrocarbon Moni-
tor.
Figure 3 shows the continuous moni-
toring system employed for the experi-
ments. Sample gas for CO, C02 and 02
24
-------
TURBULENT FLAME REACTOR
toCOCO2 + O2
monitors
to TEN AX
(heated line)
H2O—El
AIR
(FROM —
COMPRESSOR)
SWIRL VANE
ADJUSTMENT
IT
u
to THC monitor
T.C. for exhaust
j__ gas temp
heptane &
waste compound
FIGURE 1
25
-------
TENAX SAMPLING TRAIN
T.C. 130°C
I
particulate filter
/T.C. < 24°C
heat traced,
insulated
sampling line
TENAX trap
TFR
FIGURE 2
pump
exhaust
gas meter
rotameter
26
-------
CONTINUOUS MONITORS
particulate filter
t
laboratory
gas drier
I cat)
NDIR
CO
Monitor
NDIR
C02
Monitor
Paramagnetic
02
Monitor
FID
Total Hydrocarbon
Detector
FIGURE 3
-------
analysis Is extracted from the TFR
through a common sample line. Prior to
entering the exhaust gas monitors, the
sample gases pass through a particulate
filter, and a laboratory gas drier filled
with calcium sulfate.
Finally, three Type K thermocouples
are used to measure TFR exhaust gas tem-
perature and sample gas temperatures in
the heated and unheated portions of the
sample line upstream of the tenax trap.
EXPERIMENTAL DESIGN
The purpose of these experiments was
to get a general idea of POHC and PIC
levels in the exhaust gas of the TFR when
that unit is operated under conditions
that give rise to varying CO levels.
This was primarily a "scoping" study
intended to define areas in which future
research might be most relevant and fruit-
ful. Particular areas of concern in-
cluded the following:
1. How well does CO level in the
exhaust gas correlate with ORE?
2. How do POHC and PIC levels
vary with changes in CO level?
3. How do POHC and PIC levels for
long-duration failure modes
such as improperly set combus-
tion air compare to POHC and
PIC levels for short-duration
failure modes such as flame-
outs, both of which are indi-
cated by high CO levels?
4. Are there mixture effects
which influence the relation-
ship between CO level and ORE?
To address the above questions, a
series of five compounds: trichloro-
ethylene (TCE), carbon tetrachloride,
tetrachloroethane, chlorobenzene and
Freon-113 (1,1,2 trichloro-1,2,2 tri-
fluoroethane) were each burned in the TFR
separately in 2.0 mole percent solutions
in heptane. We felt it would un-
necessarily complicate the experiments
to routinely use more than one POHC per
test run at this early stage of our
testing.
To test for mixture effects however,
a test was conducted in which both Freon-
113 and chlorobenzene were mixed
together in heptane and burned. The
same volumes of chlorobenzene and
heptane were used in this test as in
the previous tests where these com-
pounds were burned separately
(i.e., 2 mole percent each of chloro-
benzene and Freon-113 resulting in 4
mole percent of total POHCs in heptane).
For each test, fuel flowrate was
held at a constant level while the com-
bustion air flowrate was varied to pro-
duce carbon monoxide levels that span-
ned the entire range of observable
values. At five or sometimes six sets
of operating conditions and observed
carbon monoxide levels, five minute
Tenax trap samples were taken of the
TFR exhaust gas in order to determine
unburned POHC.levels, DREs and the
levels of any (PICs) that may have been
produced at those CO levels. In addi-
tion, carbon monoxide (CO), total
unburned hydrocarbons (TUHC), carbon
dioxide (C02) and oxygen (02) levels
and exhaust gas temperature were con-
tinuously monitored.
At some point during each test,
the flame was turned off during a
sampling period. Sampling continued.
The purpose of this was to see if the
POHC levels and PIC species were
significantly different for this
sudden deterioration in incinerator
performance (indicated by high CO)
than they were for other more long-
term failure conditions such as those
arising from improperly set combus-
tion air.
RESULTS
DRE Findings
For the most part, 99.99% DRE
was achieved for the test POHCs in
each of the six tests. In 62.5% of
the cases DRE exceeded 99.99%. The
rest of the time, DRE was at least
99.9%. Table 1 summarizes the DREs
achieved.
Correlation Between CO and DRE
The correlation between CO and
DRE is loose at best. DRE seems to be
large ly determined by what POHC is
being burned. Some POHCs burn easily;
28
-------
trt
CU
cu
_J
c.
r—
O
r
<
J
3
1
1
§
U
C
«l
>
g
p
I
>
i
1
i
0
CO
i
ex
1
§
5
s
3.
10
L.
01
<
vo ot
Ol Ot
Ot Ot
Ot Ol
Ot Ol
CO Ot
a\ a\
as:
(^
cn cn
cn c<
ct cn
*
« ot
1** C*
cn o
0)
c
01
N
C c:
0 £
c. u.
o
•— -a
<-> «
S
cn
o>
cn
Ol
e\j
cr
^
oc
cn
c
§
*
CM
cn
ot
cn
on
CO
cn
cn
cn
cn
^.
ot
01
cr
cn
01
c
a
c
01
-a
o
c
t
*
ct r^.
cnen
cn cn
cncn
Lf)
at
cn
en
ot
cn
cn
cn
cn
cn
A*
CM
Ot
a>
cn
cn
cn
cn
cn
ot
01
c
an
cn
g
8
Tf
cn
ot
on
cn
u
I—
V
o
CM
O
CM
O
cn
0
00
s-
o
• 10
^~
o
• in
o
0
CO
a
• CM
• 2
«c
S
2
01
L.
O
iS
>*
* g
-------
some do not. Results from this test
series seemed to follow those obtained
by EERC in their earlier studies.
Namely, for hard to burn POHCs, CO
levels increase without a concommitant
decrease in DRE. Presumably, CO levels
are increasing in advance of a decline
in DRE (2). Figure 4 shows CO levels,
excess air levels and accompanying DRE
levels for all six tests. At excess air
levels above 170% theoretical air, DRE
for some POHCs does seem to decrease.
For other conditions, however, DRE seems to
remain above 99.99% even though CO
levels again increase as excess air
levels drop below 140% theoretical air.
Above 170% theoretical air, CO levels,
although elevated, are not at their
highest levels even though POHC and PIC
emissions are at their highest. Con-
versely, CO levels are at their highest
below 130% theoretical air even though
in this region DREs are consistently
above 99.99% and PIC emissions are low.
Another indication that CO levels
cannot closely predict DRE is that the
high DREs in this test series were
achieved at CO levels that are high in
comparison to those found at full-scale
incinerators. Minimum CO levels for
these six tests ranged from 100 ppm to
300 ppm as seen in Figure 4. Full-scale
incinerators typically operate at CO
levels lower than 100 ppm (3).
Variation in POHC Destruction and PIC
formation With Changes in CU Tevel
Fiqure 5 shows a graph of POHC and
PIC emissions for our tetrachloroethane
test burn. Emissions for both the POHC
for the test (tetrachloroethane) and the
main PIC (trichloroethylene) increased
with increasing excess air.reaching their
highest levels at 195% theoretical air.
The average CO level under these con-
ditions was about 800 ppm. Conversely,
at 130% theoretical air, where the
average CO level exceeded 1000 ppm, POHC
and PIC emissions were at their lowest.
This is contrary to what would be ex-
pected if the CO/DRE correlation were
quite close.
Figure 5 also indicates that, even
under conditions that result in low
POHC emissions, it is still possible to
produce comparatively large quantities
of PICs when burning certain chemicals.
Even though the emissions of tetra-
chloroethane are low, relatively high
levels of TCE form. Even though no
TCE was fed to the incinerator during
this test, levels of TCE are always
much higher than those of tetrachloro-
ethane in the exhaust. Instead of
being completely combusted to C02 and
HoO, a small portion of the tetra-
cnloroethane seems to have been merely
converted to trichloroethylene through
the liberation of HC1.
Comparison of Flameout and Non-Flameout
conditions
Fiaure 5 shows that the levels of
POHCs emitted at high excess air levels
did not differ significantly from those
emitted when the flame was extinguished.
Nor did the types of PICs. This could
be serious. When the flame goes out,
there is usually only a momentary surge
of soot and emission of exhaust gas
laden with PICs. At high excess air
levels though, the flame can be quite
stable. Thus under high excess air con-
ditions, large volumes of POHCs and
PICs can be emitted over long periods
of time resulting in a worse air
pollution problem than a momentary
upset like a flameout.
Mixing Effects
Table 2 which compares the results
from Test 4 using chlorobenzene, Test 5
using Freon 113, and Test 6 using both
chlorobenzene and Freon 113 shows that
there are no major mixing effects which
change the DREs achieved. Although the
emissions of chlorobenzene (Test 4) and
Freon (Test 5) differ slightly from the
emissions in Test 6 (where both were
burned together), the DRE for Freon was
consistently higher than that for
chlorobenzene. No mixing effects on CO
levels were noted.
CONCLUSIONS
Any conclusions resulting from
this work will need to be confirmed by
future research. Nevertheless, we did
identify some tentative conclusions.
They are:
1. CO/DRE Correlation: Based on
our tests, levejs of CO do
30
-------
DC
<
LLJ
O
X
LU
LU
>
LJJ
LJJ
O
X
O
S5
u.
CQ
CE
Ill
(N3OAXO%0 Ol Q3103UaOO)
31
13A31 3QIXONOW
-------
CO
D
O
LU
O
X
CM
o
< LU
Jf >
o
o
o
CO
O
X
CM
O
o
O
LL
If) mUrn
LU CO
cc CO
1 §
LL QJ
o <
£ w
O V)
LU X
>
111
w
DC
0)
O)
0>
O)
0)
0)
E
a
a
O
O
SNOISSfW3 Old
t- p
OHOd
32
-------
—
s
,
-
10
£J _>>
a cu
±j
"obenzene
id Separa-
o re
o eu
i_
T3 =
re x
•r—
C S '
O
CU C
f «f—
1 1
M- O
O -r-
C U)
O 13
^
tl O
re cj
Q.
E c_
o o
0 1-
9
CM
eu
jz
re
t_
0
CM O
0
o
O i —
o
CM
0
en
o
o
oo
o 1
"
i
o
un
170
-400
0
to
o
0
op
o
IT)
^—
o
o
cyi
o
^—
o
o
rO
0
^~
£
C
0 C
i — C
c
c:
ri
1
o 1
^_
£_
•f—
^_
re
o
"cu
o :
CU Q
-C —
I—
fc^ c.
*
CM
en
•
en
en
10
r--
en
CO
en
•
en
en
en
en
en
en
en
en
en
en
eu
c:
cu
N
CU
-Q
o
c_
o
o'
•* 1
I—
t/1 • —
eu
to
en
en
en
en
^t-
en
en
en
en
CM
en
en
en
en
2
OO
en
en
en
en
c
o
eu
U_ I
ir> i
^~
t/) ^^-
CU
1—
en
en
en
en
en
en
en
oo
en
en
en
en
en
en
en
*
r-.
cn
en
X— *
T
(
1 X
O •!-
i. E
o *•—
1 —
ej i
<£> 1
'
to J
CU
I—
en
en
•
en
en
!£>
en
en
en
en
en
en
en
en
en
en
en
en
~
CO
en
a
en
en
^""*
n
c: (
0 X
CU-r;
£- 1
CU
jQ
re
^™
4_J 'r™
3 re
o >
ey. ^c
M jj
•— o
1 * 1
*
-------
3.
not seem to correlate closely
with ORE although they may
indicate general trends in
incinerator performance. High
DREs (>99.99%) can be achieved
in the TFR at CO levels con-
sidered high for hazardous
waste and incinerators. The
lowest DREs achieved occurred at
the highest excess air levels
and at elevated, although not
the highest, CO levels.
POHC Destruction and PIC
Formation; The highest PIC
and unburned POHC levels of
any of the conditions tested
occurred at the highest excess
air levels tested. We expected
the highest emissions of POHCs
and PICs to occur at low
excess air levels. At low
excess air levels and corres-
pondingly high CO levels,
POHC and PIC emissions did
not increase even though high
CO levels normally indicate
incomplete combustion.
Therefore, even though DREs
remained high under these
conditions, the POHCs may not
have been completely combus-
ted to C02 and H20. Instead
the POHCs may have formed
semi-volatile and non-volatile
intermediates that were not
detected using the available
sampling and analysis methods.
(The chlorinated species
detecting GC/Hall Detector,
or they may have formed
semi-volatile PICs that we
were unable to trap on Tenax-
GC resin).
Flame-out Emissions vs. High
Excess Air Emission?;Sudden
deterioration in incinerator
performance caused by extin-
guishing the flame (shutting
off the fuel) did not result
in significantly higher levels
of unburned POHC or signifi-
cantly different types of PICs
than those produced under
conditions of high excess
air.
4. Mixture Effects: For the two
chemicals tested together
(chlorobenzene and Freon-113)
there did not seem to be a
significant mixture effect on
ORE or CO level.
FUTURE WORK
The tentative conclusions out-
lined above will have to be supported
by further experimentation. In doing
these future experiments, we plan to
take a closer look at PICs that may be
formed under conditions of low excess
air/high CO level. Further, we plan to
examine the effect of post-flame heat-
ing and cooling on CO levels, POHC
destruction and PIC formation. Both our
studies and those of EERC (each using
the TFR), show a tenuous relationship
between CO and DRE (2). Field tests at
full-scale boilers and incinerators
show no correlation between CO and DRE
(1, 3). One of the main differences
between the two sets of tests is that
the laborabory-scale work was conducted
in a water-jacketed stainless steel
device while the full-scale tests were
conducted in refractory-lined vessels.
The EPA Cincinnati has a refractory-
lined device which is similar in firing
rate to the TFR. Experimentation on
this device should help resolve the
discrepancy in results obtained from
the laboratory- and full-scale work.
Hopefully, data gathered in this
effort can be usefully integrated with
data from other researchers both in the
field and laboratory to finally deter-
mine the usefulness of carbon monoxide
as a continuous indicator of Destruc-
tion and Removal Efficiency.
34
-------
REFERENCES
2.
DeRosier, H., U. Mason, C. Span-
nagel, D. Wolbach. Emissions
Testing of Industrial Boilers
Cofiring Hazardous Wastes -~STte L.
EPA Contract 68-02-31/b
Niehart, Rachel K., John C. Kram-
lich, Gary S. Samuel son and William
Randall Seeker. "Continuous Per-
formance Monitoring Techniques for
Hazardous Waste Incinerators."
Draft report July 1983.
3. Trenholm, Andrew, Benjamin Smith
and Donald Oberacker. "Emission
Test Results for a Hazardous Waste
Incinerator RIA." EPA 600-9-4-84-
015. Proceedings: Ninth Annual
Research Symposium: Incineration
and Treatment of Hazardous Wastes.
Acknowledgments
The author would like to thank Louis H. Garcia, James Horton and Robert E.
Mournighan for their help with the experimental portion of the program.
35
-------
SUMMARY OF TESTING PROGRAM AT
HAZARDOUS WASTE INCINERATORS
Andrew Trenholm
Midwest Research Institute
425 Volker Boulevard
Kansas City, Missouri 64110
Donald Oberacker
U.S. Environmental Protection Agency
26 West St. Clair Street
Cincinnati, Ohio 45268
ABSTRACT
The Environmental Protection Agency's (EPA) regulatory impact analysis (RIA) on
hazardous waste incinerator regulations included definition of the baseline performance
of incinerators operating under normal conditions. As input to the RIA, a study was
conducted to establish current baseline levels of performance through measurements of
pollutant emissions (including over 40 Appendix VIII compounds) at eight hazardous waste
incinerators and to analyze the data generated from these measurements to identify per-
formance trends. The results of that study are described in this paper.
Tests were conducted at eight full-scale facilities to characterize all feed and
effluent streams. Pollutants measured included EPA Appendix VIII hazardous compounds,
particulates, hydrogen chloride, carbon monoxide, and total hydrocarbons. The data were
analyzed to address questions on destruction and removal efficiency (ORE), products of
incomplete combustion (PIC), pollutant emission levels, performance of air pollution
control systems, and relationships of process parameters to destruction efficiency and
emission levels. A large base of data on performance of hazardous waste incinerators
was accumulated, and many possible relationships between parameters were explored Con-
clusions included: DREs were generally above 99.99%; measured DREs tended to be higher
when concentration of the hazardous compound in the waste was higher; the most fre-
quently observed PICs were benzene, toluene, chloroform, tetrachloroethylene, and naph-
thalene; the particulate EPA regulation of 180 milligrams per normal cubic meter
(mg/Nm3) was not routinely met; and the hydrogen chloride EPA regulation almost always
met.
INTRODUCTION
The EPA's Office of Solid Waste
(OSW) conducted an RIA as part of EPA's
hazardous waste incinerator regulatory
development. The RIA includes definition
of the baseline performance of incinera-
tors operating under normal conditions
and evaluation of the costs and benefits
of alternative regulatory approaches.
These analyses require knowledge of in-
cinerator performance for different
incinerator designs and operating condi-
tions, a wide range of waste feed charac-
teristics, and various combinations of
•air pollution control equipment. This
study of eight hazardous waste incinera-
.tors was conducted to provide input to
the data base that will be used to char-
acterize performance.
Midwest Research Institute (MRI)
conducted sampling and analyses for the
eight sites, analyzed the data, and
36
-------
reported the results. The results are
available in five volumes from the Na-
tional Technical Information Service
(NTIS); Volume I is an executive summary;
Volume II contains the program descrip-
tion and discussion of results; and Vol-
umes III through V provide documentation
of the methods used and data collected on
the program (NTIS Numbers PB85 129500,
18, 26, 34, and 42, respectively). This
paper summarizes the program and the re-
sults obtained.
PURPOSE
The purpose of this study was to es-
tablish current baseline levels of per-
formance through measurements of pollut-
ant emissions at the eight incinerators
and to analyze the data generated to
identify performance trends. The primary
measures of performance during this pro-
gram were DREs for principal organic haz-
ardous constituents (POHC) and removal of
HC1 and particulates from the stack
gases.
APPROACH
The eight hazardous waste incinera-
tors tested had been selected by EPA to
provide a broad representation of the in-
cinerators in the country. These incin-
erators span a wide array of incinerator
design and operating parameters, particu-
late and HC1 control equipment, and waste
feed characteristics.
The distribution of incinerator
types and control devices is shown in
Table 1. The facilities exhibited a wide
range in capacity of 1 to 78 gigajoules
per hour (GJ/hr) (1 to 74 million British
thermal units per hour (Btu/hr)) heat in-
put. Values for three key parameters,
combustion temperature, residence time,
and percent excess air or percent oxygen,
also varied widely. Operating tempera-
tures ranged from 650° to 1450°C and cal-
culated residence time varied from 0.07
to 6.5 seconds. Excess air values fell
within 60 to 130%. Corresponding oxygen
content ranged from 8 to 12%.
Waste feeds with a wide range of
characteristics were encountered at the
eight sites. The heating values were be-
low 4,600 kilojoules per kilogram (kJ/kg)
(2,000 British thermal units per pound
(Btu/lb)) for aqueous liquids and ranged
from 14,000 to 37,000 kJ/kg (6,000 to
16,000 Btu/lb) for organic liquids, and
from 0 to 29,000 kJ/kg (0 to 12,500 Btu/
Ib) for solid wastes. The organic liq-
uids and many of the solids had a high
enough heating value to sustain combus-
tion without auxiliary fuel. Percent
chloride in the wastes ranged up to 25%,
with the highest values for organic liq-
uids (chlorinated solvents). Little ash
was found in aqueous liquids; values for
organic liquids ranged up to 9%; solids
had the most ash with values from 17 to
29%. Water content is obviously high for
the aqueous wastes but ranged up to 50 to
60% for the organic liquids and solids.
TABLE 1. DISTRIBUTION OF INCINERATOR
TYPES AND CONTROL DEVICES
Type or device Number of facilities
Incinerator type
Liquid injection
Rotary kiln -
secondary chamber
Hearth
Gas injection
Control device
None
HC1 scrubber
Particulate control
8
2
2
1
3
5
4
This program encompassed a wide ar-
ray of activities centered on testing at
the eight hazardous waste incinerator
facilities. These activities included
evaluation and, in some cases, modifica-
tion of sampling and analysis methods;
multimedia sampling and analysis (S&A),
and extensive data evaluation to assess
the performance of hazardous waste incin-
erators.
The S&A activities characterized
each input and output stream to the
greatest degree possible. Typical input
streams included waste feed, auxiliary
fuel, and control system makeup and re-
cycle waters. Output streams included
stack gases, control device effluent
(solid or liquid), and incinerator bottom
ash. Each stream sampled was generally
37
-------
analyzed for organic constituents,
chlorides, particulate or ash, and, in
some cases, metals. The S&A activities
included full quality assurance/quality
control.
The S&A activities generated exten-
sive data of potential use to evaluate
incinerator performance. Procedures were
developed to reduce these data to a for-
mat usable for subsequent analyses. Ac-
tivities included evaluating and defining
procedures for calculating chloride and
POHC input rates, blank correction proce-
dures for POHC effluent rates, and calcu-
lating the most representative residence
time and excess air for each of the eight
facilities.
Specific technical issues related to
incinerator performance were identified
throughout the program. Engineering and
statistical analysis techniques appropri-
ate to each problem were selected and, to
the degree possible within the available
data base, conclusions were drawn. Pat-
tern recognition data analysis was used
to identify relationships between param-
eters that were not apparent from the
overall examination of the data. These
techniques were used primarily to iden-
tify the incinerator and waste charac-
teristics that contributed most signifi-
cantly to the achievement or nonachieve-
ment of 99.99% ORE.
PROBLEMS ENCOUNTERED
The sites and the specific'incinera-
tor operating conditions were limited by
program constraints. Sites were those
where access could be obtained, and op-
erating conditions during the tests were
those selected by the plants as their
normal conditions. Thus, this study did
not provide a complete characterization
of incinerator performance for specific
POHCs under varied operating conditions.
A rigorous experimental matrix of incin-
erator parameters was not used nor were
detailed facility characterizations pre-
pared.
A number of S&A problems were en-
countered which are too detailed to pre-
sent in this paper. The reader is re-
ferred to the full report for discussion
of these problems and the solutions used.
RESULTS
This section presents the more sig-
nificant results of the program. The
discussion is divided into six subsec-
tions: POHCs, PICs, CO/THC monitoring,
HC1 control, particulate control, and
metals emissions.
POHCs
One key element of this study was
determination of DREs for any POHC (Ap-
pendix VIII compound) present at 100
micrograms per gram (pg/g) (approximate
detection limit) or greater in any waste
feed stream. The DREs span a wide range
for both volatile and semi volatile com-
pounds. The 134 volatile and 106 semi-
volatile data points (a data point is one
test run for one compound) are distrib-
uted as shown in Table 2.
TABLE 2. ORE RANGE
DRE range
Type of POHC
Volatiles Semivolatiles
< 99
99-99.9
99.9-99.99
99.99-99.999
> 99.999
2
12
30
51
39
3
5
20
46
32
The average DRE for the eight fa-
cilities was 99.992% for all volatile
compounds. An average for semivolatile
compounds could not be calculated because
over half the stack measurements were be-
low detection limits; however, most semi-
volatile compounds measured consistently
achieved at least 99.99% DRE.
DREs < 99.99% tend to occur .when one
or both of two factors are present. The
first factor is-low concentration of the
POHC in the waste feed (i.e., < 1,000
Mg/g). The second factor involves com-
pounds commonly identified as PICs (espe-
cially chloroform, methylene chloride.,
benzene, and naphthalene),. The formation
.38
-------
of these compounds during the incinera-
tion of chlorinated organics increases
their concentration in the stack gas, re-
sulting in a lower ORE. Eighty percent
of the volatile points and 96% of the
semivolatile points below 99.99% ORE oc-
curred in cases where one or both of
these factors were present.
Three types of relationships between
ORE and waste feed and incinerator oper-
ating parameters were addressed--the ef-
fect of waste feed concentration on ORE;
the relationship between ORE and heats of
combustion for specific POHC, and the
impact on ORE of such incinerator parame-
ters as temperature, residence time, and
stack 02 concentration.
The relationship between ORE and
waste feed concentration was examined us-
ing linear regression analyses. Separate
regression analyses were conducted for
volatile and semivolatile compounds using
the log transforms for both penetration
(1-DRE) and waste feed concentration.
For volatile compounds, the data were
best fit by a line with a slope of -0.79.
Figure 1 shows that no compounds below
200 ijg/g in the waste feed achieved a ORE
greater than 99.99% and no points above
12,000 ng/g failed to achieve 99.99% ORE.
The correlation coefficient of -0.84 for
the regression line is highly significant.
A similar comparison using the data
for semivolatile POHCs showed the same
basic relationship (Figure 2). The slope
of -0.81 is comparable to that obtained
for volatile compounds. A statistically
significant correlation coefficient of
-0.76 was obtained. Phenomena that might
cause such a relationship were not iden-
tified.
The relationship between ORE and the
heat of combustion of a POHC was examined
using rank order statistics. The analy-
sis was based on data from the four fa-
cilities which had sufficient quantifi-
able points to provide a meaningful anal-
ysis. If the incinerability of a com-
pound were dependent on its heat of com-
bustion, the DREs for compounds subject
to the same incinerator conditions should
be positively correlated with the heats
of combustion of those compounds. None
of the data for the four plants show a
significant positive correlation.
Multivariate .statistical analyses
were used to examine the impact of four
incinerator operating parameters—resi-
dence time, chamber temperature, heat in-
put rate, and stack 02 concentration—on
ORE. The results of two types of multi-
variate analyses, factor analysis and
discriminant analysis, indicate that the
only operating variable strongly related
to ORE is chamber temperature. However,
the data were not sufficient to define a
quantitative relationship between temper-
ature and ORE. For the sites tested,
factors such as waste feed concentra-
tion, specific compound kinetics, and PIC
formation tend to mask relationships be-
tween incinerator operating conditions
and ORE. These results indicate rela-
tionships which warrant further investi-
gation; they do not establish.definitive
relationships between the operating pa-
rameters and ORE.
Measured DREs may also be affected
by removal of POHCs from the gas stream
by absorption in a scrubber. At one
plant naphthalene was detected at 35
micrograms per liter ((jg/L) in the scrub-
ber effluent. Otherwise toluene, phenol,
and naphthalene were present in effluents
at several sites at concentrations less
than 20 yg/L. These low levels suggest
that scrubbers generally do not remove
significant quantities of POHCs from the
stack gas.
PICs
For this program, a PIC was defined
as any Appendix VIII organic hazardous
constituent that was not present in the
waste feed at concentrations greater than
100 |jg/g. Some 30 compounds were classi-
fied as PICs from the eight tests (Table
3). On the average, the PIC levels were
slightly higher than the POHC levels, al-
though this relationship varies widely
from facility to facility. At only one
of the eight tests were PIC output rates
greater than 0.01% of POHC input rates.
Three explanations for the pres-
ence of PICs are: (a) the compounds
were actually present at low concentra-
tions (< 100 (jg/g) in the feed and were
destroyed at a relatively low ORE;
(b) the compounds were introduced to
the system from a source other than
the waste; and (c) the compounds are •
39
-------
Ill III
II I I I t t lull
8
|
2
§
J
8
b
CO
b
O
[) uoi4DJ43ua(j
b.
•o
b
I
t
en
8
•1°
8
s
^ J
11
b
- l) UOI4DJ43U3J
40
-------
TABLE 3. PICs FOUND IN STACK EFFLUENTS
PIC
Benzene
Chloroform
Bromodi chl oromethane
Dibromochloromethane
Bromoform
Naphthalene
Chlorobenzene
Tetrachl oroethyl ene
1,1,1-Trichloroethane
Hexachl orobenzene
Methylene chloride
o-Nitrophenol
Phenol
Toluene
Bromochl oromethane
Carbon disulfide
Methylene bromide
2,4,6-Trichlorophenol
Bromomethane
Chl oromethane
Pyrene
Fluoranthene
Di chl orobenzene
Tri chl orobenzene
Methyl ethyl ketone
Di ethyl phthalate
o-Chlorophenol
Pentachlorophenol
2,4 Dimethyl phenol
No. of
sites
6
5
4
4
3
3
3
3
3
2
2
2
2
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Cone.
Cng/L)
12-670
1-1,330
3-32
1-12
0.2-24
5-100
1-10
0.1-2.5
0.1-1.5
0.5-7
2-27
25-50
4-22
2-75
14
32
18
110
1
3
1
1
2-4
7
3
7
2-22
6
1-21
formed as products of combustion reac-
tions.
The estimated stack concentration
for a compound present in the waste feed
just below the POHC limit of 100 |jg/g and
subject to a ORE of 99.9% is 10 nanograms
per liter (ng/L). This low ORE becomes
plausible when the trend of ORE versus
concentration in waste feed is examined.
The trend of decreasing DREs shows that
many of the PICs detected could be ex-
plained by low concentrations in the feed.
Other potential sources of organic
constituents in the stack gas are feed
waters to pollution control systems,
inleak air, and auxiliary fuel. In sev-
eral cases, the control system added to
the concentration of POHCs in the stack
gas. The scrubber influent at three fa-
cilities contained about 100 ng/L of
chloroform, while the effluent contained
no measurable amounts. In all three
cases, the quantity of chloroform lost
from scrubber waters accounts for all
chloroform detected in the stack. Scrub-
ber waters also appear to contribute
small quantities of other trihalomethanes
to stack gases. Inleakage of air down-
stream from the combustion chamber may
contribute to PIC emissions. Analysis of
data from one facility where ambient con-
centrations of organic constituents were
available and estimates of inleak air
could be developed, indicated that inleak
air could account for up to 10% of some
organic constituents detected in the
stack. Finally, analysis of auxiliary
fuel oil at one plant indicated that this
source could be a major contributor for
some POHCs. Fuel oil streams can contain
Appendix VIII compounds that are not
present in the waste feeds.
The third mechanism, the actual for-
mation of combustion reaction products,
is the least understood of the three pos-
sibilities. The occurrence of simple
stable compounds such as chlorinated
methanes, ethanes, and benzene compounds
in most stacks supports the hypothesis
that PICs may be formed during combustion
reactions. Primary candidates for this
list on the basis of these tests include
compounds which were present in all stack
samples regardless of their presence in
the waste feed: methylene chloride;
chloroform; carbon tetrachloride; tri-
chloroethylene; tetrachloroethylene;
1,1,1-trichloroethane; benzene; toluene;
chlorobenzene; naphthalene; and phenol.
Either benzene or toluene accounted for
the highest measured stack concentrations
of any compound at seven of the eight
facilities.
CO and THC were monitored continu-
ously during the testing at each of the
eight sites. These data were subse-
quently reduced and compared to both POHC
concentrations and DRE. The results in-
dicated that continuous monitors for THC
and CO may provide some indication of in-
cinerator performance but that they are
not good predictors of POHC concentra-
tions either at specific plants or across
plants. The results should be inter-
preted with caution in that the tests
were not conducted in a parametric
41
-------
fashion specifically designed to examine
such correlations.
The results showed that CO is pos-ir "
tively correlated with POHC concentra-
tions when CO concentrations are above
50 ppm, but high POHC concentrations are
also measured when CO concentrations are
below 50 ppm. For THC, concentrations of
POHCs are measurably higher when THC con-
centrations are above 10 ppm than when
they are below 10 ppm. However, no cor-
relations between THC concentrations and
POHC .concentrations were exhibited when
THC concentrations were above 10 ppm.
Nonparametric methods were also used
to compare CO concentrations to ORE. The
analysis compared the average CO concen-
tration for a run to the penetration
(1-DRE) of a specific POHC for the same
run using the data for carbon tetrachlo-
ride, toluene, trichloroethylene, and
tetrachloroethylene. The concentrations
of CO and penetration were not positively
correlated for any of the four compounds.
HC1 Control
Most hazardous waste incinerators
that require add-on HC1 control equipment
use some type of countercurrent wet
scrubber for HC1 removal. Both water and
water containing caustic for pH control
are used as a scrubbing liquid. Five of
the eight facilities had HC1 scrubbers.
Three of these systems were single-stage
and two were multistage. Two of the sys-
tems used water only as a scrubbing
media; the other three added caustic for
pH control. While the systems did vary
significantly, all five appear to be
within normal ranges with respect to key
design parameters.
The five facilities with HC1 control
complied with the standard except for a
single run at one facility. Three of the
facilities with packed towers 'achieved
greater than 99% HC1 removal and emitted
less than 1 kilogram per hour (kg/hr) on
all runs. One of the facilities which
had only a venturi scrubber with mist
eliminator consistently achieved 97 to
98% HC1 removal and emitted less than
0.3 kg/hr of HC1. The one facility which
failed to achieve the standard had a
single packed bed scrubber. This facil-
ity achieved about 99% HC1 removal effi-
ciency and emitted less than 1 kg/hr of
HC1 on two runs. On the third run, the
removal efficiency was 96% and the HC1
emission rate was 2.0 kg/hr. No reason
for the decreased performance on the
third run was identified.
Particulate Control
The data collected during this study
indicate that achieving the particulate
emissions limitation of 180 mg/Nm3 (cor-
rected to 7% 02) may be more difficult
than achieving either the HC1 or ORE
standards for hazardous waste incinera-
tors. This conclusion holds true for
incinerators with and without particulate
control devices.
The data from three facilities with
no particulate control device suggest
that any facility firing wastes with ash
content greater than 0.5% will clearly
have difficulty achieving the particulate
limit without add-on particulate con-
trols. At ash contents of 0.2 to 0.5%,
facilities may still have difficulty
meeting the limit without controls. Each
of these three facilities fired liquid
organic wastes. One facility with ex-
tremely low ash (0.1% or less) easily
complied with the standard. A second fa-
cility with wide ranges in ash content
(< 0.005% to 0.5%) had mixed results. At
< 0.05% ash, emissions easily complied
with the limit; at 0.2% ash, emissions
marginally complied with the limit; and
at 0.5%, ash emissions exceeded the limit
by a factor of 2. The third facility had
ash contents ranging from 0.7 to 0.8%.
Emissions from this facility exceeded the
180 mg/Nm3 limit by factors of 4 to 6.
Two of the five facilities with par-
ticulate control systems failed to comply
with the emission limit and two achieved
marginal compliance. The one facility
with a venturi scrubber with an extremely
high pressure drop (~ 30 kilopascals or
lOOjnches H20) easily'complied with the
limit. The results for four systems are
summarized below:
Control type
Stack particulate
cone. (mg/Nm3)
Range Avg.
Venturi scrubber 10,4-33.7 23.0
Ionizing wet scrubber 139-176 151
Venturi scrubber 102-290 169
Packed bed scrubber 183-217 200
42
-------
Metals Emissions
Metals contained in the waste feed
to incinerators eventually emit from the
incinerator by one of three pathways—as
part of the participate emissions or as
vapor from the stack, in liquid or solid
effluents from the control device, or in
the incinerator ash. Metals were ana-
lyzed in selected samples of these efflu-
ent streams from five of the eight fa-
cilities tested. The analysis focused on
the 12 metals identified in Appendix VIII
as hazardous constituents are shown in
Table 4.
The data from these analyses were
not sufficient to complete an overall
materials balance for metals. A summary
of the air emissions data for the five
tests is presented in Table 4. These
data are for particulate emissions only;
they do not include metal vapor emissions.
ACKNOWLEDGMENTS
This project was jointly sponsored
by the EPA's Hazardous Waste Engineering
Research Laboratory and the Office of
Solid Waste. The valuable assistance
and guidance of Don Oberacker, Timothy
Oppelt, Edward Martin, Benjamin Smith,
and Gene Grumpier is greatly acknowledged.
The cooperation of all test sites and the
efforts of many individuals in MRI's En-
vironmental Systems and Analytical Chem-
istry departments are also acknowledged.
TABLE 4. RANGES OF METALS EMISSIONS FOR FIVE HAZARDOUS
WASTE INCINERATOR TESTS
Metals
__^^ Emission range
|jg/g of particulate
g/mi n
Antimony
Arsenic
Barium
Beryllium
Cadmium
Chromi urn
Lead
Mercury
Nickel
Selenium
Silver
Thai 1 i urn
< 1,200
41
< 0.32
140
668
3,100
230
< 500
7.6
- 15,200
ND*
- 3,090
6
- 4,300
- 47,500
- 100,000
56#
- 49,000
- 61,600
1,880
**
< 0.012 -
ND*
0.0032 -
< 0.0001 -
0.0033 -
0.0066 -
< 0.016 -
0.0043
< 0.004 -
< 0.046 -
0.001 -
< 0.004 -
1.2
0.12
0.002
0.087
0.47
7.3
0.57
4.5
0.0064
0.012
* ND - not detected in all runs.
# Detected during only one run; others not detected.
**Range could not be developed because of high < values.
43
-------
TIER 4 DIOXIN TEST PROGRAM STATUS
Miles, A. J., Parks, R.M., Oberacker,- D., Southerland, J.
ABSTRACT
The paper discusses the status of Tier 4 of the National Dioxin Study and focuses
primarily on the identification and selection of candidate combustion sources for the
test program. No laboratory results will be presented.
The Tier 4 program addresses the broad questions do combustion sources emit
significant quantities of dioxin? If so, how much and what are the health risks
associated with these emissions?
The approach taken to address these questions was divided into several steps.
First, an extensive literature search was conducted, and all available dioxin emissions
data were summarized. The information was then analyzed to attempt to identify factors
which would affect dioxin emissions from combustion sources. The analysis indicated that
the following factors could have an effect on dioxin emissions.
— Presence of dioxins in the feed;
- Presence of dioxin precursors in the feed;
- Chlorine in the feed;
— Combustion conditions (temperature, oxygen level, etc.); and
- Feed processing.
The analysis also indicated that combustion of waste products as opposed to fossil
fuels was more likely to form dioxins. Using the factors listed above, in conjunction
with a. list of major combustion source categories and the available dioxin emissions data
base, a ranked list of combustion source categories with the potential to emit dioxin was
developed.
In conjunction with the development of the list, a source test program was devised.
Ten to twelve sources are to be tested for dioxin emissions using a Modified Method 5
sampling train including an XAD-2 resin trap. The test program also includes sampling of
feed materials and collected particulate; continuous monitoring of CO, C02, 0-, and THC;
and acquisition of operating data over the period of each test. In addition, some
40 sites were selected for ash sampling and analysis to broaden the dioxin data base.
Site selection for the test program began in August 1984, and the first test was
conducted in October. Two tests were conducted in November, and subsequent tests were
conducted in December and January. All testing will be complete by June of 1985. No
results from the test program are available.
INTRODUCTION AND PURPOSE
The Air Management Technology Branch
(AMTB) within the EPA's Office of Air
Quality Planning and Standards (OAQPS) is
responsible for the development and imple-
mentation of a source testing program for
Tier 4 of the National Dioxin Study.
Technical guidance for the source test
program is provided by IERL/HWERL in
Cincinnati. The purpose of Tier 4 is to
assess combustion source emissions of
polychlorinated dibenzo-p-dioxin (PCDD),
with a focus on the most toxic isomer -
2,3,7,8-tetrachlorodibenzo-p-dioxin
(2,3,7,8-TCDD). Combustion source
emissions of polychlorinated dibenzofuran
(PCDF) will also be addressed in this
study.
Radian Corporation, under task order
contract, is providing support to the
AMTB by collecting and reviewing available
literature data on emissions of PCDD's
and PCDF's from combustion sources. In
addition, Radian will perform PCDD
44
-------
emissions tests oti twelve combustion '
sources. The samples will be analyzed
by EPA laboratories for PCDDs and PCDF's. '
The source test plan addresses the
following questions:
1. Which combustion source catego-
ries emit PCDD's (and PCDF's) to the
atmosphere?
2. What range of concentrations
and quantities of PCDD's (and PCDF's) are
emitted from these source categories?
3. What are the estimated health
risks associated with these emissions?
This paper describes development of
the source test program including selection
of candidate combustion sources. In
addition, the status of the program as of
April 1985 is discussed. Further details
of the program can be found in the project
plan (1) , the initial literature review
document (2)} and the sampling guidance
manuals (3, 4) developed for the program.
Dioxins are members of a family of
organic compounds known chemically as
dibenzo-p-dioxins. All dioxin compounds
have a three ring nucleus containing two
benzene rings interconnected by a pair of
oxygen atoms. The structural formula of
the dioxin nucleus and the convention
used in numbering its substituent posi-
tions are shown in Figure la. Chlori-
nated dioxins can contain up to eight
chlorine atoms and 75 different chlori-
nated dioxins are possible. Each congene.
has its own physical, chemical, and
health effects properties.
with varying physical, chemical, and
health effects properties.
• 4
DIOXIN CONFIGURATION
Figure la.
Dibenzofurans are a group of organic
compounds that have a similar structure
to the dibenzo-p-dioxins, except that the
two benzene rings in the nucleus are
interconnected with a five member ring
containing only one oxygen atom (Figure
Ib). Theoretically, the chlorinated
dibenzo furan group can contain up to
135 different structural congeners, each
8 4
FURAN CONFIGURATION
Figure Ib.
Of all the PCDD's and PCDF's 2,3,7,8-
TCDD has received the most attention.
However, in general, 2,3,7,8-TCDD repre-
a small fraction of the total PCDD found
in combustion source emissions. In
addition, PCDF emissions can exceed
2,3,7,8-TCDD emissions by two to three
orders of magnitude. For some sources
PCDF's contribute significantly (over
80 percent) to the health risk associated
with combined PCDD and PCDF emissions.
APPROACH AND SCOPE OF THE TIER 4 STUDY
The Tier 4 program was divided into
three phases: (1) the literature evalua-
tion and design of a test program and
(2) the source test program including ash
sampling, and (3) data summary and
analysis.
The first phase of the Tier 4 study
was accomplished in three general -tasks.
The first task was to obtain and review
the available literature on PCDD emissions
from combustion sources. Over 100
published and unpublished reports were
obtained. In addition, contacts were
made with key individuals to identify
recently performed or ongoing studies
that could be used to supplement the data
base. The second task was to develop a
ranked list of source categories with
potential to emit PCDD's. The criteria
used to develop the ranked list was based
on information obtained from the litera-
ture review. The final task was to
develop a testing program to assess PCDD
(and PCDF) emissions from combustion
sources. The results of these tasks are
summarized below.
PCDD Emissions Data
A review of the literature has
produced a list of 12 broadly defined
source categories for which some dioxin
45
-------
emissions data has been collected.
Table 1 lists sources for which published
dioxin data could be found and summaizes
the measured TCDD concentrations.
Although the National Dioxin Study is .
focusing on 2,3,7,8-TCDD, most of the
data found in the literature addressed
TCDD and total PCDD. Since 2,3,7,8-TCDD
data are limited, TCDD was used during
the evaluation of the data base as the
bast indicator of 2,3,7,8-TCDD emissions.
with over 100 ppm chlorine and normally
some chlorinated phenol content. The
highest TCDD emissions are generally
associated with solid feed fuels and low
combustion temperatures. Combustion
sources burning fossil fuels tended to
emit less TCDD's than those burning waste
products.
TABLE 1.
COMBUSTION SOURCES IDENTIFIED IN THE 1984 LITERATURE SURVEY (2)
Number of
Facilities
Source Category Tested
Hunlcipal Waste Cowbustors
USA
European
Hazardous Waste Incinerators .
Incinerator Ship
„ Land Based Incinerators
Sewage Sludge Incinerators
Utility Coal Boilers
Cowwrcial Boilers (Waste Fired)
Industrial Boilers (Waste Fired)
Activated Carbon Regeneration
Residential Wood Combustion
Mobile Sources
Wire Reclamation Incinerators
time/Cesent Kilns (Waste Fired)
Accidental Electrical
Equipment Fires
6
13
2
10(7)d
1
7
6(6)d
6(l)d
1
4
9(4)d
1
Kl)d
2
Sample
Stack
Stack
Stack
Stack
Stack
Stack
Stack
Stack
Stack
Scrapings
Exhaust
Scrapings
Stack
Wall Swipes
TCDDa
Mean
3.5 ng/m3
25.6 ng/m3
NDC
0.56 ng/m3
f
ND
g
10.13 ng/m3
0.013 ppt
329 ppt
4.0 ppt
234 ppt
g
44 ppm1
Range
ND-240 ng/m3
ND-128 ng/m3
ND
ND-2.5 ng/m3
f
ND
g
ND-40.5 ng/m3
ND-0.050 ppt
ND-777 ppt
ND-20 ppt
58-410 ppt
g
ND-ig5 ppm
2,3,7,8-TCDD
Mean Range
3.5 ng/m3 0.30-9.1 ng/m3
b
ND e
18,000 ppth ND-55,000 ppt
0.019 ppt ND-0.083 ppt
242 ppt 26-600 ppt
3.0 ppt *
0.059 ppm, *
aTCDO * Tetrachlorod1benzo-p-d1ox1n.
b0ash « Ho Data.
QM « Hone detected (Detection limits vary).
Kuwoer of tests have been performed, but the results have not been officially reported.
« One datum, no range available.
PCDO scan only. PCDO concentrations ranged from 483 ng/m3 to 1,140 ng/m3 with a mean of 739 ng/m3
•'Results have not yet been officially reported.
%>t « Parts per trillion by weight.
Fuels include wood, wood/oil mixture, and natural gas.
Jc
Includes PCS transformers and capacitor batteries.
ppoi * parts per million by weight.
A general characterization of each
of the 12 source categories identified in
the initial literature survey was made to
identify similarities and differences
that may affect the magnitude of PCDD
emissions from each source. The following
broad characteristics emerged. The
source categories with the highest TCDD
emissions were burning waste materials
"ther
Experimental Studies
In addition to gathering PCDD emis-
sions data, all available experimental
studies concerning PCDD formation
mechanisms for combustion sources were
obtained and reviewed in order to identify
more specific factors that may contribute
to PCDD formation.
46
-------
There are several unproven hypotheses
concerning PCDD emissions from combustion
processes. Dow Chemical's "Chemistries
of Fire" theory proposes that PCDD's are
a natural byproduct of fire and will be
formed at some quantities in all combus-
tion processes (5). However, experimental
results by Buser and Rappe (6) and an
evaluation of data from the literature
suggests that PCDD's are emitted only
under limited conditions. The most
prevalent theories, including Esposito's
formation mechanism (7), involve the
incomplete combustion of PCDD's or PCDD
precursors. Although there is some
disagreement of the definition of PCDD
precursors, they are defined in this
paper as chlorinated aromatics that can
product PCDD's through bimolecular
reactions and thermal rearrangements.
Examples include chlorinated phenols and
chlorinated benzenes. PCDD precursors
may be thermally rearranged during
incomplete combustion to form PCDD's.
Also, when PCDD's are present in the feed
to a combustion source, they can escape
with the fine particulate if the destruc-
tion efficiency is low.
Neither the Dow hypothesis nor the
precursor hypothesis is conclusively
supported or refuted by the available
data. Recent studies involving pyrolysis
of wood with and without chlorination, in
conjunction with studies of pyrolysis of
chlorinated coal, suggest that any
organic material combusted in the presence
of high levels of inorganic chlorine may
lead to PCDD formation under certain
conditions.
Based on the literature review, the
following factors are believed to affect
dioxin emissions:
- Waste composition
- PCDD In feed,
- Precursors in feed,
- Chlorine in feed,
- Combustion conditions
- Residence time,
- Oxygen availability,
- Waste characteristics
- Feed processing, and
- Supplemental fuel
(variability and Btu value)
The interaction of these factors in PCDD
formation mechanisms is not well under-
stood. In addition, quantitative data
concerning waste compositions is generally
not available for many waste materials
that are combusted and combustion condi-
tions are not well defined for many
combustion devices and are largely
determined by site specific operating
practices.
Source Category Prioritization
Despite these limitations the
factors listed above in conjunction with
available TCDD emissions data were used
to subjectively rank all combustion
sources for the purpose of the Tier 4
source tested program. A short list of
combustion sources was developed by
excluding those sources burning or using
a relatively clean feedstock or fired
with fossil fuels, such as coal, oil or
gas, and those source categories which
are fairly small or intermittent in
nature, examples include incinerator ships
and coffee roasting. The remaining
sources were then divided into 4 groups
ranked A-D using the rationale outlined in
Figure 2. The ranks are defined as
follows:
Preliminary Souro Ll»t
1.
2.
3.
Potential to
Emit TCDD
TCDD Detected
Precursor Level
Combustion Condition
Low
High
Size of
Source Category
Small
Large
A
Figure 2. Ranking Flow Chart.
47
-------
Rank A are large source categories -
(greater than 1 million tons of fuel
and/or waste burned annually) with
elevated dioxin precursor contamination
of feed/fuel. These categories are
judged to have a high potential to emit
TCDD. Rank B are small source categories
(less than 1 million tons of fuel and/or
waste burned annually) or source catego-
ries with limited dioxin precursor
contamination of feed/fuel. These
categories have some potential to emit
TCDD. Rank C are source categories less
likely to emit TCDD. Rank D are source
categories which have already been tested
three or more times.
The ranked list was then used as the
focus of the source test program. Pre-
liminary cost estimates indicated that
only 10-12 source tests could be performed
with the available budget. With this
limitation in mind, a decision was made
to test three facilities for each of the
two rank A source categories, and a single
facility in each of the rank B categories.
Through supplementary coordination with
other in-house programs, samples were
also planned for mobile sources and
woodstoves.
Test site selection began in August
1984. This effort was initially focused
on the rank A categories, sewage sludge
incinerators and black liquor boilers.
The site selection process involved
identifying candidate sources from lists
of sites provided by State and EPA
Regional offices, trade associations and
previous EPA studies. The lists were
narrowed to three or four candidate sites
using data on facility size, age, type of
combustion device, etc. Each of the
candidate sites on the short list was
then contacted by telephone to explain
the Tier 4 program, to gather further
site specific information, and to ascer-
tain if they were interested in partici-
pating in the program. Pretest survey
visits were conducted at least two
candidate sites per source category. For
the rank A source categories, an attempt
was made to pick at least one average
source within the source category and one
worst case candidate. For the rank B
categories, attempts were made to select
worst case sites with respect to PCDD
emissions.
As the program progressed it became
evident that it was very difficult to
—-define the conditions which constitute
, . worst case with respect to potential PCDD
--emissions. Very little if any PCDD or
precursor information was available for
the selected source categories, and the
relative combustion conditions within the
source category were not always known.
For these reasons, a great deal of
reliance was placed upon the total
chlorine content of the primary feed
materials to the combustion device as an
indication of worst case conditions. For
example, further information was found
concerning the sources and levels of
chlorides in black liquor circuits at
Kraft pulp mills. During the initial
source category selection process, black
liquor boilers associated with the
pulping of salt laden wood were suspected
of having the highest chlorine content.
After visits to numerous mills a brief
literature survey and receipt of chlorine
analyses for several mills it was
discovered that very little wood is now
stored in salt water prior to pulping.
In addition black liquor with chlorine
contents of 0.1 percent to 2 percent have
been identified and a study was found
with discussed the potential for chloro-
benzene formation from combustion of
black liquor (8). The highest chlorine
content black liquor was associated with
a mill that uses spent acid from the
chlorine dioxide generator as a source of
salt cake make up.
As the site selection process
continued, additional source category
specific data became available, and as a
result, some changes were made to the
ranked list. Table 2 presents the initial
and final ranked list. The changes to the
list are briefly discussed below.
Commercial boilers firing waste oils were
dropped from the rank A category because
the proposed RCRA amendments would pre-
clude the burning of waste oils blended
with chlorinated solvents as other
hazardous wastes. Combustion of wood
treatment plant sludges containing penta-
chlorophenol and/or creosote (K001 waste)
in boilers was identified as a rank B
candidate in the initial list. Following
contacts with the major wood treating
companies and with various regulatory
agencies, this category was dropped'from
rank B to rank C. All of the wood treat-
ment companies claimed to be either land-
filling the sludge or incinerating the
sludge in a hazardous waste incinerator.
48
-------
TABLE 2. RANKED SOURCE CATEGORY LISTS FOR PCDD TESTING
Initial List - March 1984
Rank A
Sewage Sludge Incinerators
Black Liquor Boilers
Commercial Boilers
Rank B
PCP Sludge
Carbon Regeneration
Charcoal Manufacture
Wire Reclamation
Current List - March 1985
Sewage Sludge Incinerators
Black Liquor Boilers
Carbon Regeneration
Wire Reclamation
Industrial Incinerators
Salt-Laden Wood Fired Boiler
Secondary Metals Blast Furnace
Drum & Barrel Furnace
Rank C
Mobile Sources
Wood Stoves
Wood Fired Boilers
Small Spreader Stoker
Hazardous Waste Incinerators
Lime/Cement Boilers
Rank D
Municipal Waste
Industrial Boilers Firing Hazardous Waste
Hazardous Waste
Mobile Sources
Wood Stoves
Small Spreader Stoker
Commercial Boiler
PCP Sludge
Lime/Cement Boilers
.Municipal Waste
Industrial Boilers Firing Hazardous Waste
Charcoal manufacturing facilities
were dropped from rank B to rank C
because all facilities contacted were
processing untreated forest scraps and
not sawmill slabs that might have been
pretreated with chlorophenols for
Sapstain control.
In addition to these changes in the
source category ranking, three categories
were added to the source test program;
these are industrial incinerators,
secondary metals blast furnaces, and drum
and barrel reclamation furnace.
Several industrial incinerators were
identified during the course of the site
selection process. Further investigation
showed this to be a large category in
terms of numbers of units nationwide (9).
Solid waste materials burned in these
units often contain appreciable levels of
chlorine. In addition, most incinerators
are fairly small and batch fed resulting
in poor combustion conditions. One site
selected for Tier 4 testing burns poly-
vinyl chloride (PVC) coated wood scraps
that have been treated with pentachlor-
phenol. The secondary metals blast
furnace source category was also added to
the source test list. Some blast furnaces
in the secondary metals industry process
metal bearing scrap that contains plastics
including PVC (10). PCDD has previously
been detected in the baghouse dust from
one such facility.
Drum and barrel reclamation furnaces
were added to the list because of the
large number of facilities, the diverse
nature of wate materials combusted
during the drum burning process and
because of the relatively poor combustion
conditions encountered (11).
The Source Test Program
A total of twelve complete source
tests will be conducted as part of the
source test sampling program. The
anticipated schedule is shown in Table 3.
In addition, one woodstove will be
sampled and two mobile source samples
will be analyzed. The test program is
49
-------
complex and involves characterization of
combustion device conditions using
continuous emissions monitors in addition
to Modified Method 5 sampling for PCDD's
following the draft ASME protocol (12) .
Samples of the feed materials to the
combustion device will be sampled and
analyzed for chlorine and precursor
content. Ash and soil samples will be
collected and analyzed for PCDD's. The
sample matrices for the first 5 tests
are shown in Table 4.
.TABLE 3. TIER 4 SOURCE TEST SCHEDULE
testing is expected to be
completed by July 1, 1985. Only
limited PCDD analytical results are
available at this time. All results
will be presented in the final Tier 4
report which is scheduled for completion
in late 1985. Results of the study will
also be included in a report to Congress
scheduled for December 1985.
Ash Sampling
A total of 40 ash samples will be
collected for combustion sources selected
by EPA regional offices. Analysis of
these samples will supplement the source
test data.
Test
Number
1
2
3
4
5
6
7
8
9
10
11
12
Schedule
October 1984
November 1984
November 1984
December 1984
February 1985
March 1985
March 1985
April 1985
April 1985
May 1985
May 1985
June 1985
June 1985
Source Category
Sewage Sludge Incinerator #1
Industrial Incinerator
(Wood/Plastic)
Sewage Sludge Incinerator #2
Black Liquor Boiler #1
Black Liquor Boiler #2
Wire Reclamation Incinerator
Wood Stove*
Wood Fired Boiler
Black Liquor Boiler #3
Industrial Carbon
Regeneration Furnace
Secondary Metals Blast
Furnace
Sewage Sludge Incinerator #3
Drum and Barrel Reclamation
Furnace or Coal Fired
Spreader/Stoker Boiler
*This test is being conducted jointly with an AA EPA program,
the integrated air cancer project.
TABLE 4. SUMMARY OF TIER 4 SAMPLE MATRICES
Sample Stream
Primary Feed
Materials
Auxiliary Fuels
Combustion Air
Combustion Device Outlet
Control Device Outlet
Combustion Device
Bottom Ash
Emission Control
Device Ash
Other Plant Materials
Soils
Site 01
SSI-A
Sewage
Sludge
No. 2
Yes
GEM, MM5
MM5
• Yes
Scrubber
Slowdown
No
Yes
Site 02
ISW-A
Wooden
Wastes
No. 2
Yes
GEM, MM5,
NA
Yes
NA
No
Yes
Site 03 Site 04
SSI-B BLB-A
Sewage
Sludge
None
Fuel Oil
No
HC1 GEM
MM5
Yes
Filtered Scrubber
Slowdown
No
Yes
Black
Liquor
None
Fuel Oil
No
GEM, MM5
MM5, HC1
NA
No
Yes
Yes
Site 05
BLB-B
Black
Liquor
None
No
GEM, MM5
MM5, HC1
NA ,
No
Yes
Yes
50
-------
National Dioxin Study Tier 4 -
Combustion Sources: Project Plan.
EPA-450/4-84-014a, Monitoring and
Data Analysis Dioxin. U.S.
Environmental Protection Agency.
Research Triangle Park, N.C.
February 1985.
National Dioxin Study Tier 4 -
Combustion Sources: Initial Litera-
ture Review and Testing Options.
EPA-450/4-84-014b. U.S. Environ-
mental Protection Agency. Research
Triangle Park, N.C.
National Dioxin Study Tier 4 -
Combustion Sources: Sampling
Procedures. EPA-450/4-84-014c.
U.S. Environmental Protection
Agency. Research Triangle Park, N.C.
National Dioxin Study Tier 4 -
Combustion Sources: Ash Sampling
Program. EPA-450/4-84-014d. U.S.
Environmental Protection Agency.
Research Triangle Park, N.C.
Dow Chemical. The Trace Chemistries
of Fire - A Source of and Routes for
the Entry of Chlorinated Dioxins
into the Environment. Dow Chemical
U.S.A., 1978. 46 pp.
Buser, H. R. and C. Rappe. Formation
of Polychlorinated Dibenzofurans
(PCDFs) from the Pyrolysis of
Individual PCB Isomers. Chemosphere,
8^(3): 157-174, 1979.
Esposito, M. P., T. 0. Tiernan and
F. E. Dryden. Dioxins: Volume 1:
Sources, Exposure, Transport, and
Control. EPA-600/2-80-156,
U.S. Environmental Protection
Agency, Cincinnati, Ohio, June 1980.
Ahling, B. and A. Lindskog. Emission
of Chlorinated Organic Substances
from Combustion. In: Pe'rgamon
Series on Environmental Science,
Volume 5, 1982. pp. 215-225.
Source Category Survey - Industrial
Incinerators. EPA-450/3-80-13.
U.S. Environmental Protection
Agency. Research Triangle Park,
N.C. May 1980.
10. Pretest survey report to V.S. Metals
Facility, Cateret, N.J.
11. Tonhill, C. J. "Barrel and Drum
Reconditioning Industry Status
Profile." Tonhill, Shuckrow, and
Associates. EPA Contract
No. 68-03-2905. 1980.
12. Draft ASME Sampling Protocol
"Sampling for the Determinations
of Chlorinated Organic Compounds
in Stack Emissions". October 1984.
51
-------
TM
POWDERED ACTIVATED CARBON TREATMENT (PACT )
OF LEACHATE FROM THE STRINGFELLOW QUARRY
William M. Copa, Ph.D.^1-*
Marvin J. Dietrich (!)
Patrick J. CanneyC2)
Tipton L. Randall^
(1) Zimpro Inc., Rothschild, WI 54474
(2) Casmalia Resources, Santa Barbara, CA 93108
ABSTRACT
JIM
The PACT " process, which incorporates the use of powdered activated carbon in a
conventional activated sludge system, was used to treat leachate from the Stringfellow
Quarry. Wastewater characterization of this leachate indicated that a high proportion
of the organic content was biologically refractive but adsorbable on carbon. Laboratory
screening tests indicated that this particular leachate was ammenable to treatment by
Che PACT™ process. Demonstration scale PACT™ treatment of the leachate was conducted
at the Casmalia Resources landfill site using a 1000-5000 GPD (gallon per day) trailer
mounted PACT™ unit. Stringfellow leachate was pretreated by lime precipitation for
heavy metals removal prior to PACT™ treatment. Chemical Oxygen Demand (COD) reductions
in excess of 72 percent, along with removal of organic priority pollutants, were
achieved in this PACT™ demonstration study.
INTRODUCTION
The State of California has embark-
ed upon an investigation of the feasi-
bility of using alternative technologies
to safely reduce, recover, treat, and
destroy hazardous wastes, and thereby,
reduce its dependency on land disposal
of toxic wastes. The intent of the in-
vestigation is to document the avail-
ability of alternative waste management
technologies and to help redirect the
State's hazardous waste management pro-
gram toward the development and imple-
mentation of these technologies. The
alternative waste treatment technologies
were categorized as follows:
Physical Treatment
Chemical Treatment
Biological Treatment
Incineration
Stabilization/Solidification
Treatment
TM
The PACT wastewater treatment
process, which combines powdered acti-
vated carbon with the biological acti-
vated sludge process, is one of the
alternative technologies investigated
by the State of California. In treating
hazardous wastewaters, the PACT™ pro-
cess offers a combination of physical
treatment, biological treatment, and
solidification treatment. The PACT™
process would appear to be ideally suit-
ed to the treatment of high-priority
wastewaters containing pesticides, halo-
genated organics, or non-halogenated
volatile organics.
The present study reports on the
demonstration of the PACT™ process as
an alternative technology for the treat-
ment of leachate from the Stringfellow
Quarry. The Stringfellow Quarry leach-
ate can be classified as a high-priority
wastewater containing halogenated and
non-halogenated volatile organics 'as
well as heavy metals. The PACT™ demon-
stration study consisted of a prelimin-
ary laboratory screening and waste
52
-------
characterization followed by a continuous,
large scale PACT™ treatment of the
Stringfellow Quarry leachate at the
Casmalia Resources landfill site in
Santa Barbara, County, California.
TM
THE PACT PROCESS
TM
The PACT wastewater treatment
process consists of the addition of
powdered activated carbon to the
aeration basin of a biological activated
sludge system. The combination of
physical adsorption with biological
oxidation and assimilation, which is
accomplished in PACT™ wastewater treatment,
has been shown to be particularly
effective in treating wastewaters which
are variable in concentration and
composition, highly colored,.or contain
materials which are refractive or
potentially toxic to biological growth.
The characteristic advantages of the
PACT™ wastewater treatment process over
conventional activated sludge are:
1. High BOD and COD removals.
2. Stability of operation with
variability in influent concentra-
tion and composition.
3. Enhanced removal of toxic sub-
stances and priority pollutants.
4. Effective color removal.
5. Improved solids settling.
6. Supression of volatilization of
organlcs.
TTVT
The PACT wastewater treatment
system operates in accordance with the
flow diagram shown in Figure 1. Influent
wastewater flows to the aeration tank
where the wastewater is mixed with
powdered activated carbon and biological
solids. The wastewater-carbon-biological
solids mixture is aerated for a suffici-
ent period of time (Hydraulic Detention
Time - HOT) to effect biological oxid-
ation and assimilation of the biodegrad-
able content of the wastewater. After
aeration, the mixture flows to a clarif-
ier where the powdered carbon and bio-
logical solids are settled and separated
from the treated wastewater. The clarif-
ier overflow (treated wastewater) is
discharged from the PACT™ system. The
clarifier underflow solids are returned
to the aeration tank on a continuous
basis to maintain a desired concentration
of powdered activated carbon and biologi-
cal solids.
A portion of the biological solids
and spent powdered activated carbon is
wasted daily from the PACTTM system.
The wasted solids are directed to an
aerobic digester for solids storage and
stabilization. Solids remaining after
aerobic digestion would be periodically
wasted to a solids disposal system.
Specifically for this PACT™ demonstra-
tion study, the wasted solids were de-
watered in a filter press. The dewatered
cakes were placed in drums and ultimately
disposed of in the Casmalia Resources
landfill.
Because powdered activated carbon
is wasted with the excess biological
solids, virgin powdered activated carbon
addition is required, on a daily basis,
to maintain the desired aeration basin
carbon concentration. The addition rate
of virgin carbon is determined by treat-
ment requirements, waste load and waste-
water characteristics.
LEACHATE CHARACTERIZATION
A composite sample of leachate from
the Stringfellow Quarry was analyzed
using standard wastewater examination
methods(1,2). The composition of the
leachate is shown in Table 1. The
composite sample of leachate was acidic,
brown colored, and contained some sus-
pended solids.
The Stringfellow Quarry leachate was
also analyzed for fifteen heavy metals.
Both total and soluble heavy metal
concentrations were determined on the
initial composite sample (see Table 2).
The precipitation of the heavy metals
as their hydroxides was evaluated by
• the addition of sodium hydroxide to ad-
just the pH of the leachate to 7.0 and
9.0. The concentration of the heavy
metals remaining in solution in the
leachate at pH 7.0 and 9.0 is also re-
ported in Table 2.
53
-------
TM
LABORATORY PACT SCREENING TESTS
TM
Laboratory PACT screening tests
consisted of generating physical and
biophysical adsorption isotherms on the
Stringfellow Quarry leachate. A physical
adsorption isotherm was generated by ad-
justing the pH of the composite leachate
to 7.0 and contacting this full strength
material with varying concentrations of
powdered activated carbon (ICI Hydrodarco
H) for two hours. After the two hour
contact period, the mixtures were filtered
and the filtrates were analyzed for COD,
Color, etc. A biophysical adsorption
isotherm was obtained in a similar manner
except that a small portion of activated
sludge biomass, sufficient to effect a
biomass concentration of approximately
2000 to 4000 mg/1, was added to each
portion of leachate along with the vary-
ing amounts of powdered activated carbon.
Also, the leachate samples were aerated
during the two hour contact time. The
filtrates were analyzed for COD, color,
etc., just as in the generation of the
physical adsorption isotherm. Data
obtained in this manner are shown in Table.
3. The adsorption isotherm data were
plotted according to the Freundlich
equation (log x/m - log k + 1/n log C)
to determine the isotherm constants (see
Figure 2). Inspection of the
physical and biophysical isotherms,
Figure 2, shows that the physical isotherm
is linear over the wide range of effluent
COD values but that the biophysical
isotherm deviates from the Freundlich
equation at high loading rates (low carbon
doses). This may imply that biological
removal, by adsorption on the biomass or
by biological assimilation, is the
principle mechanism for removal of COD
at low carbon doses but that physical
adsorption is the principle mechanism
at high carbon doses. The isotherms also
show that,a higher COD removal can be
achieved in the biophysical system com-
pared to the physical system with similar
carbon doses.
The biophysical isotherm can be used,
in conjunction with other waste treatment
principles, to select operating parameters
for the PACT1^ process. Consider the
biophysical isotherm shown in Figure 2.
If a treated Stringfellow leachate efflu-
ent with a COD of 500 mg/1 is desired, a
powdered activated loading rate of
approximately 115 rag COD/gram carbon is
required. For an influent COD of 2000
mg/1, a mixed liquor carbon concentration
of approximately 13 g/1 would be requir-
ed. With the choice of a nominal solids
retention time (10 days) and a net
biological cell yield (0.15 Ibs. biomass/
Ib. COD removed), a practical food to
microorganism (F/M) ratio can be target-
ed (3). The F/M ratio can then be used to
determine the required wastewater
hydraulic detention time for a given
microorganism concentration. In this
manner, all of the initial PACT™ process
parameters can be established.
TM
PACT DEMONSTRATION STUDY
TM
PACT Process Equipment
TM
The PACT demonstration unit,
supplied by ZIMPRO INC., is a complete
PACT™ wastewater treatment system
mounted on a flatbed trailer (see
Figure 3). The demonstration unit is
capable of treating 1000 to 5000 gallons
per day (GPD) of industrial or municipal
wastewater, depending upon the strength
of the wastewater and the degree of
treatment required.
The demonstration unit was equipped
with a six foot diameter (2580 gal.)
aeration tank containing eight Sanitaire
coarse bubble air diffusers. The air
diffusers were fed by a variable speed
Roots blower which has an output
capacity of 40 to 100 SCFM of air. The
demonstration unit also had an eight
foot diameter clarifier, equipped with
a collector rake mechanism. A 500
gallon tank is provided for collection
of wasted carbon and biological solids.
The demonstration unit is also equipped
with a pH control system, three chemical
feed tanks, and peristaltic pumps to
feed chemicals, e.g. nutrients, polymers,
etc. The PACT™ demonstration unit was
operated according to the process flow
diagram shown in Figure 1 throughout
this demonstration test.
54
-------
Two 20,000 gallon Baker portable
storage tanks were installed to store the
influent waste prior to treatment in the
PACT™ demonstration unit. For the
duration of the demonstration test,
leachate from the Stringfellow landfill
was received at the Casmalia Resources
waste disposal facility. The leachate
was stored in an open pond. On a batch
basis, leachate was mixed with lime
sludge (pH adjusted to 9.0) to precipit-
ate heavy metals. After settling, the
clear supernatant was then pumped to the
Baker storage tanks for subsequent use in
the PACT™ demonstration unit.
TM
The PACT demonstration unit was
equipped with a one square foot pilot
filter press. The filter press was
equipped with ten recessed, cast iron
plates capable of producing 1 1/4 inch
thick dewatered filter cakes. In oper
ation, waste sludge was initially pumped
to the empty press. The filtrate was
discharged into the Casmalia Resources
site evaporation ponds. The filter press
run was terminated when the filtrate flow
decreased to a small trickle. After
pressure release, the moveable end of
the filter press was retracted and the
dewatered filter cakes were allowed to
drop into a collection hopper. The de-
watered cakes were drummed and deposited
in the Casmalia Resources landfill, for
ultimate disposal.
PACT™ Startup and Operation
TM
The PACT demonstration study was
initiated on July 18, 1984. The PACT™
demonstration unit was started by adding
approximately 2300 gallons of mixed
liquor obtained from the Lompoc, Califor-
nia wastewater treatment plant, to the
aeration tank. Four 50 Ib.bags of ICI
Hydrodarco H powdered activated carbon
were added to the mixed liquor and the
entire mixture was aerated to sustain
biological activity and provide uniform
mixing. Acclimation of the PACT™ system
to the Stringfellow Quarry leachate was
initiated on July 19, 1984. The feed
rate of Stringfellow leachate was initi-
ally set at a low rate, approximately
0.4 gallons per minute, and was gradu-
ally increased throughout the acclimation
period.
The start-up and acclimation period
required approximately forty five days
and was confounded by reoccurring aeration
blower malfunctions. Ultimately the
aeration blower was replaced and, except
for periodic plugging of the feed line,
the remainder of the operation of the
demonstration study proceeded smoothly.
When the PACT™ system was acclimated
to the Stringfellow leachate waste, as
indicated by a substantial oxygen uptake
of approximately twenty mg per liter per
hour or greater, the operation was __.
guided by controlling the selected PACT
process parameters:
Hydraulic Detention Time
(HOT) =16 hours
Solids Retention Time
(SRT) = 10 days
Mixed Liquor Suspended Carbon
(MLSC) = 10,000 mg/1
Carbon Dose (CD) = 667 mg/1
The HOT was gradually reduced as
the study progressed, to the desired
value of 16 hours which corresponds to
the maximum feed rate of 2.6 gallons
per minute. The SRT was varied only
when the wasting rate was interrupted
because of decreased oxygen uptake.
Throughout the PACT™ demonstration
study, sodium hydroxide solution was
added to the aeration basin to maintain
a target pH range of 6.5 to 7.5. An
automatic pH controller was used to
facilitate this pH control.
55
-------
Ammonia and phosphorus nutrients were
added to the waste stream as ammonium
chloride and di-ammonium phosphate.
Levels of nutrient addition were adjusted
in an attempt to obtain, at steady state
operation, effluent ammonia nitrogen and
phosphorus concentrations in the range of
1 to 5 mg/1. In the screening of String-
fellow leachate, both ammonia and
phosphorus nutrient deficiencies were
indicated but trace nutrient levels were
assessed to be adequate.
TM
PACT Operational Testing
During the acclimation and operation-
al testing period, the PACT™ unit
operation was monitored by implementation
of a routine sampling and analysis
schedule. Analyses performed three times
per week were COD, BOD, suspended carbon,
suspended biomass, total Kjeldahl nitro-
gen, ammonia nitrogen, and total phosphor-
us. The three time per week analysis
schedule, showing sample streams and
respective analyses performed, is outlined
in Table 4. In addition, as shown on the
analysis schedule, biweekly samples of
influent and effluent were analyzed for
specific toxic components, i.e. volatile
organic chemicals, associated with the
Stringfellow leachate.
TM
PACT operational parameters that
were determined on a daily basis are as
follows:
1. Dissolved Oxygen (DO) in the
aeration basin and clarifier.
2. Temperature of the aeration
basin and clarifier.
3. pH of mixed liquor in the
aeration basin.
4. Total hydrocarbon concentration
in the aeration basin offgas -
monitored only until the Santa
Barbara County APCD was assured
that hydrocarbon air emissions were
not significant.
5. Influent and recycle flow rates.
6. Weight of powdered activated
carbon added to the system.
7. Volume of mixed liquor wasted
from the system.
8. Oxygen uptake of the mixed
liquor biological solids.
Results and Discussion
TM
The PACT demonstration study was
conducted between July 18, 1984 and
October 17, 1984 at the Casmalia
Resources waste disposal facility.
The wastewater that was used throughout
this study was leachate from the String-
fellow Quarry. A sample of Stringfellow
leachate was screened by laboratory
testing and was shown to be treatable
by the PACT™ process. The Freundlich
physical and biophysical adsorption
isotherms, which indicated carbon load-
ing rates, organic removal efficiencies,
and biological uptakes, were used to
assess PACT™ treatability.
TM
A performance summary for the PACT
demonstration study is shown in Table 5.
This summary shows operational and
analytical data averaged for the entire
study (7/18/84 through 10/17/84) and for
the steady state period (9/10/84 through
10/17/84). This summary shows that, on
the average, a 72 percent COD reduction,
an 82 percent BOD reduction, and a 72
percent DOC reduction were achieved
throughout the entire study. Similar
average COD, BOD, and DOC reduction were
achieved during the steady state oper-
ation, i.e., 74, 89, and 71 percent. The
COD parameter- of this particular waste-
water, which had a low BOD:COD ratio, was
the chosen parameter for monitoring over-
all treatment performance. The average
COD removal for the entire PACT™ study
period was 72 percent. A plot of influent
and effluent COD, for the entire length
of the study, is shown in Figure 4. As
would be expected with a waste feed pre-
pared on a batch basis, the influent COD
was relatively constant with only small
variations. Effluent COD values indi-
cated consistently good removal effici-
encies and varied between 345 and 733
mg/1. The plot of influent and effluent
COD also indicates a consistent removal
efficiency with increasingly stable
values as the PACT™ process became fully
acclimated in the latter half of the
study. The COD removal data also indi-
56
-------
cate that the mechanical problems (i.e.
aeration blower malfunction) and the
interrupted wasting of solids did not
appreciably effect the COD removal
efficiency. This overall result attests
to the reliability of the PACT™ process,
but more importantly, may indicate that
the basic mechanism for organic removal
may indeed be physical adsorption.
The influent BOD values were con-
sistently low throughout the entire study.
The influent BOD values ranged between 15
and 127 mg/1. These low BOD values re-
flect the difficulty in obtaining a
culture of biological organisms (BOD seed)
that can readily assimilate the complex
organic compounds present.in this waste-
water. The effluent BOD values are quite
variable, with values ranging between
<1.0 and 73 mg/1.
The suspended solids in the effluent
averaged 17.2 mg/1 for the entire study.
This concentration of suspended solids is
comparable to that obtained from an
efficiently operating municipal waste-
water treatment plant. Improved effluent
quality, i.e. lower suspended solids,
could be obtained by tertiary filtration
of the effluent in a sand filter, prior
to discharge.
TM
Throughout this PACT demonstratxon
study, color removal was relatively low.
The PACT™ process is normally very
effective in removing color from waste-
waters. The lack of color removal in
the PACT demonstration study is in direct
conflict with the laboratory PACT™
screening test which indicated nearly
complete color removal. The explanation
of this conflicting data is not evident
at the present time.
Organic priority pollutant analyses
were obtained during this PACT™ demon-
stration study. The base neutral, acid
fraction, and pesticide analyses of
influent and effluent samples indicate
that essentially all.organic priority
pollutants were removed by the PACT™
process. The effluent base neutral
fraction did however contain certain
phthalates which are believed to have
been leached from the plastic pipe and
tank coatings of the PACT™ demonstration
unit.
Volatile organic chemicals (VOC's)
were known to be present in the String-
fellow leachate and were designated as
the specific toxic component of this
wastewater. Analyses of the influent
wastewater for VOC's indicates the
presence of benzene, chlorobenzene,
dichloromethane, chloroform, 1,2-dichloro-
ethylene, trichloroethylene, tetrachloro-
ethylene, ethyl benzene, and toluene in
nearly all of the influent samples at
detectable levels. Dichloromethane is
the only VOC detected in the effluent
samples and may be due to laboratory
contamination. Dichloromethane is used
in the extraction of base neutral and
acid priority pollutant fractions and is
commonly detected due to cross contamin-
ation in the laboratory. All other VOC's,
present in the feed wastewaters, were
completely removed (below limits of
detection) by the PACT™ process.
The offgas from the aeration basin
was monitored for total hydrocarbon (THC)
concentration, to assess stripping of
organics from the aeration basin. THC
levels were shown to range between 2.5
and 11 parts per million (volume/volume)
expressed as methane. This THC level is
comparable to the background level of
THC at the Casmalia Resources disposal
facility, and verifies that the PACT™
process does indeed limit the volatiliz-
ation of organics from the aeration basin.
A complete report on this study,
which contains all of the tabulated oper-
ational parameters, analytical data, and
the Quality Assurance Plan, has been sub-
mitted to the State of California Depart-
ment of Health Services™'.
REFERENCES
1. Standard Methods for the Examination
of Water'arid'Wastewater, 15th Ed.;
APHA, AWWA. WPCF, 1980. ;
2. Methods for Chemical Analysis of Water
and Wastes. U.S. EPA, EPA-600/4-79-020,
March 1979.
3. Jenkins, D., and Garrison, W.E.,
"Control of Activated Sludge by Mean
Cell Residence Time", J.W.P.C.F.,
Vol. 40, No. 11, Part 1, Nov. 1968,
pp 1905-1919.
4. W.M. Copa, et. al., "Demonstration of
Unit-Scale Powdered Activated Carbon
Treatment of Hazardous Wastes",
Contract No. 83-82053 OPR-54, A-l,
State of California Department of
Health Services, Nov. 1984.
57
-------
Table 1. Composition of Stringfallow Quarry Leachate
(Values in mg/1 except pH and Color)
Constituent
Composite Leachate
(used for waste
characterization)
COD
DOC
SS
VSS
pH
Color (APHA)
Total Solids
Total Ash
TKN
NH3-N
N03-N
N02-N
Total Phosphate as P
Soluble Ca
Soluble Mg
Soluble Cl
3550
<60
856
6300
700
3.27
860
31,200
24,200
32.6
14.4
14.0
<0.5
2.5
407
540
303
58
-------
Table 2. Metal Content of Stringfellow Quarry Leachate
Before and After pH Adjustment (Values in mg/1)
Metal
Cu
Cr
Zn
Hg
Pb
Cd
As
Sb
Co
Fe
Tl
Ni
Ag
Ba
Se
Composite
PH =
Total
15.4
98.2
536
<0.002
7.1
2.4
0.1
2.4
3.6
851
0.14
15.8
3.0
1.6
0.036
Leachate
3.27
Soluble
12.5
77.5
465
<0.002
1.6
1.8
0.007
0.65
3.2
363
0.9
15.8
0.06
0.07
<0.002
pH Adjusted
pH = 7.0
Soluble
0.52
0.09
24.4
<0.002
0.4
1.0
<0.004
0.6
0.2
162
0.8
0.5
0.4
0.3
0.022
Leachate
pH = 9.0
Soluble
0.13
0.11
0.09
<0.002
0.27
0.04
<0.004
0.5
0.17
2.1
0.7
0.002
0.05
0.08
<0.002
59
-------
Table 3. Data for Physical and Biophysical Isotherms
COD and Color.Removal.from.Stringfellow Quarry Leachate
Physical Isotherm Biophysical Isotherm
Carbon
Dose, g/1
0
0.8
1.6
3.2
6.4
12.8
25.6
COD mg COD Color
rag/1 gc APHA
2310 - 223
2150 200 142
2020 181 106
1790 163 76
1370 147 50
870 112 16
410 74 13
APHA COD mg COD Color
gc mg/1 gc APHA
1930 - 148
102 1870 550 94
73 1690 388 75
46 1563 233 64
27 1520 123 49
16 617 132 20
8 279 79 8
APHA
gc
-
161
92
50
27
16
8
TM
Table 4. PACT Demonstration Study
Analysis
COD
BOD
DOC
TKN
NH3-N
Phosphorus
Suspended
Solids/Ash
pH
VOC's
Priority
Pollutants
Biomass/Carbon
THC
Sampling
TM
PACT* Influent '
3 times /week - C*
3 times /week - C
3 times /week - C
3 times /week - C
3 times /week - C
3 times /week - C
3 times /week - C
3 times /week - C
Bi-Weekly - G
Once/Study - C
Oxygen Uptake Rate
and Analysis Schedule
TM
PACT Effluent Mixed Liquor
3 times /week - C
3 times/week - C
3 times /week - C
3 times /week - C
3 times /week - C
3 times /week - C
3 times /week - C
3 times /week - C 3 times /week
Bi-Weekly - G
Once /Study - C
3 times/week
Daily - G
Aeration
Tank Offgas
- G
- G
Daily r G**
Dissolved Ojcygen
Daily - G
* C « Composite, G = Grab ** Daily until Santa Barbara County APCD advises otherwise.
60
-------
Table 5. Performance Summary*
PACT™ Demonstration Study
Study Dates
7/18/84 - 9/10/84 -
10/17/84 10/17/84
Pilot Plant Feed Rate, GPM
Hydraulic Residence Time, Days
Solids Residence Time, Days
Mixed Liquor, mg/1:
Suspended Solids
Volatile Carbon
Volatile Biomass
COD:
Influent, mg/1
Effluent, mg/1
Removal , %
BOD:
Influent, mg/1
Effluent, mg/1
Removal , %
DOC:
Influent, mg/1
Effluent, mg/1
Removal , %
Suspended Solids:
Influent, mg/1
Effluent, mg/1
Color (APHA Units)
Influent
Effluent
1.74
1.03
29
19850
9150
3590
1786
498
72
55.6
10.1
82
556
154
72
.92.5
17.2
174
128
2.43
0.74
15
20780
9700
3240
1788
467
74
50.3
5.5
89
535
154
71
103
17.9
177
154
.* All values are averages of data acquired throughout the indicated time periods.
61
-------
-TM
FIGURE l- PACT WASTE WATER TREATMENT SYSTEM
FLOW DIAGRAM
I NUTRIENTS]
CARBON |
INFLUENT-
AERATION
RECYCLE
•^•EFFLUENT
SOLIDS WASTING
AEROBIC
DIGESTER
J
SOLIDS TO
DISPOSAL
62
-------
1000
800
600
400
300
200
mg COD
<3c
100
90
80
70
60
50
40
30
20
100
FIGURE 2-PHYSICAL AND BIOPHYSICAL ISOTHERM
COD REMOVAL FROM STRINGFELLOW QUARRY LEACHATE
BIOPHYSICAL ISOTHERM
PHYSICAL
ISOTHERM
000
800
600
400
300
200
_L
200
400 6OO 800IOOO
2000
I I I I I I I IIQ
100
90
80
70
60
50
40
30
20
4000 6000800O
EFFLUENT COD, mg/L
63
-------
-TM
FIGURES- PACTIM DEMONSTRATION UNIT
64
-------
2000
o>
Q
O
O
1800 -
1600
1400
1200
1000
800
600
400
20O
FIGURE 4 - PACT™ DEMONSTRATION STUDY
COD DATA
i - \ - 1 - 1 - 1 - 1 - 1
' I I T
_L
A
\!
v v-\
EFFLUENT
2000
i . i
I - 1 — L — i
J - 1 - 1
800
600
400
200
000
800
600
400
200
7/27/84 8/6 8/16 8/26
9/5 9/15
DATE
9/25 10/5 10/15 10/25
65
-------
FIELD TESTING OF PILOT-SCALE APCDs AT A HAZARDOUS WASTE INCINERATOR
Wayne Westbrook
Eugene Tatsch
Research Triangle Institute
P. 0. Box 12194
Research Triangle Park, North Carolina 27709
and
Lawrence Cottone
Engineering Science, Inc.
10521 Rosehaven Street
Fairfax, Virginia 22030
and
Harry Freeman
U.S. Environmental Protection Agency
Hazardous Waste Engineering Research Laboratory
Cincinnati, Ohio 45268
ABSTRACT
Pilot scale air pollution control devices supplied by Hydro-Sonic® Systems, ETS,
Inc., and Vulcan Engineering Company were installed at the ENSCO, Inc. Incinerator in El
Dorado, Arkansas in the spring of 1984. Each of these units treated an uncontrolled
slipstream of the incinerator exhaust gas. Simultaneous measurement of the total
particulate and HC1 in the gas streams were made at the inlet to and exit from the units
using an EPA Method 5 sampling train. Particle sizing at both locations using Andersen
impactors was also done. The units supplied by Hydro-Sonics® Systems and ETS, Inc.
exhibited a high degree of HC1 and particulate matter control. The Hydro-Sonic® Tandem
Nozzle SuperSub Model 100 gave the best overall performance for HC1 and particulate
control and ability to accommodate the variable composition of the exhaust gas.
INTRODUCTION
Much hazardous waste generated has
characteristics that make incineration the
disposal method of choice. Incineration
of these wastes must be performed accord-
ing to the applicable regulations of the
Resource Conservation and Recovery Act
(RCRA) and State and local regulations.
The RCRA regulations specify the destruc-
tion and removal efficiency (DRE) that
must be achieved for principal waste
components and set limits for the emission
rates of particulate matter and hydrogen
chloride (HC1).
Advancements in air pollution con-
trol technologies is a continuing process.
Improvements in existing' technologies as
well as innovative approaches are occur-
ring. It is to be expected that advanced
devices claiming better collection effi-
ciency, lower energy consumption, or lower
cost than current equipment will be
marketed. It is likely that some of these
devices are capable of providing better
control of emissions from hazardous waste
combustion than is now available.
66
-------
PURPOSE
The purpose of this project is to
evaluate innovative air pollution control
devices and to test the performance of
these technologies on commercial-scale
facilities. The specific goals are to
examine the cleaning capabilities of each
device under specified operating condi-
tions. The pollutants to be monitored are
(a) particulates; total mass, and as a
function of particle size, and (b)
hydrogen chloride (HC1).
The data developed in the project
will be used by EPA and others in the
waste management technical community to
assist in optimizing the control of air
emissions from hazardous waste combustion.
APPROACH
Three vendors with pilot units
meeting the project criteria agreed to
participate. They are HydroSonics®
System, Inc.; ETS, Inc.; and Vulcan
Engineering. The Hydro-Sonic® pilot unit
is a wet scrubber that operates by fine
atomization of water into the gas stream
resulting in HC1 capture and particle
growth with final removal by cyclonic
action. The ETS, Inc. unit uses dry lime
injection for particulate capture. The
unit provided by Vulcan Engineering is a
high temperature (>550° C, 1000° F)
metallic weave filtration system. It is
not currently designed for acid gas
control. The Ceil cote unit applies an
electric charge to the incoming
particulate which then adheres to the
packing material and is removed by water
spray.
Several commercial hazardous waste
incineration facilities were contacted to
select the host facility for the tests.
Of those companies willing to participate,
ENSCO, Inc.'s incinerator in El Dorado,
Arkansas best met the project criteria.
After a pilot unit was connected to
the installed duct work, a slip stream of
particulate-laden gas from the emergency
bypass stack was drawn through the unit.
Inlet and outlet gas streams were sampled
simultaneously by EPA Method 5 for the
concentration and total mass of the
particulate matter and HC1. Particle size
distributions were also determined at both
locations.
Each APCD vendor operated his own
equipment and was allowed to vary condi-
tions to test the performance of the unit.
ENSCO operated their facility in a routine
manner during these tests. No speci'a?
fuel blends on operating procedures were
used to accommodate the units.
TEST SOURCE DESCRIPTION
The host plant for these tests was
ENSCO, Inc.'s incineration complex in El
Dorado, Arkansas. This permitted facility
primarily incinerates polychlorinated
biphenyl (PCB) contaminated oils and
capacitors.
The plant is designed to receive and
incinerate whole.capacitors. The
shredded parts are transported with into a
rotary kiln. The kiln exit gas passes
through a cyclone which removes much of
the large suspended particulate matter.
The gases then pass into a two-chambered
afterburner referred to as the thermal
oxidation unit (TOU). The upstream side
of the afterburner can be fired with
PCB-containing oils or other high Btu
liquids.
Gases leaving the TOU are drawn into
the custom-designed wet scrubber. This
unit consists of two circulating water
loops. The first, prequench, loop removes
the bulk of the particulate and HC1. Lime
slurry is added for pH control. Slowdown
is routed to a pond for solids settling.
The clarified water is recycled. The
second loop is comprised of a jet eductor,
knockout vessel, and demister. Fresh
makeup water is added to maintain inven-
tory. All blowdown from this loop is used
as a makeup to the first loop. No water
is discharged from this system except for
that evaporated and carried out with the
stack gas.
The incinerator typically operates 24
hours per day, 7 days per week. In
general, wastes of a specific class are
accumulated onsite until sufficient
quantity is available for burns of at
least one day. This permits achieving and
maintaining steady-state incineration
conditions. During these tests, wastes
accumulated were incinerated on the
schedule determined by plant management.
No special wastes were burned or excluded.
Close contact was maintained with plant
personnel so that the APCD testing spanned
only one operating condition, insofar as
possible.
DESCRIPTION OF APCD CONNECTION TO PLANT
Connection was made to the emergency
bypass stack. The stack, located between
the TOU and the scrubber, is refractory
67
-------
lined. Openings around the stack cap were
plugged to prevent air infiltration into
the stack. A stainless steel duct is
connected to a flanged opening in the
stack. The duct connects to a vertical
section leading to a Hastalloy cooling
section. Gas cooling is by direct water
spray. A temperature control device
mounted at the exit of the cooling section
modulates the water flow. The cooled gas
was then ducted to the APCD connection via
a 12 inch ID 17 feet horizontal run of
insulated carbon steel pipe. The APCD gas
inlet sample ports are located near the
midpoint of this duct. The two ports were
on the vertical and horizontal axes
perpendicular to the gas flow.
TEST METHOD DESCRIPTION
All emission measurements field work
was performed by Engineering Science,
Inc., Fairfax, VA, under subcontract to
RTI. An RTI project coordinator was
onsite to direct the test.
The test program was designed to
withdraw a slipstream of incinerator
exhaust gas and to test the APCDs
performance in removing particulate and
HC1. It was necessary, therefore, to
measure these pollutants at the inlet to
and exit from the air pollution control
device (APCD). Sufficient equipment and
personnel were available to conduct test-
ing simultaneously at these two locations.
Both the concentration of total par-
ticulate in the gas stream and particulate
mass per unit time were determined using
EPA Method 5. The impinger solutions in
the back half of the train were analyzed
to obtain equivalent information for HC1.
The particle size distribution at both the
inlet and outlet test locations was deter-
mined using Anderson cascade impactors.
Cascade impactors classify particles on
the basis of their aerodynamic properties.
Due to the large difference in parti-
cle loadings between the inlet and outlet
sites, it was not possible to obtain
simultaneous impactor runs spanning the
same time interval. Typically 5 to 7
minutes operation at the inlet site
product optimum stage loadings. Operation
for 60 to 90 minutes was required at the
outlet location for collection of adequate
sample. Impactor runs were arranged to
coincide with one or more Method 5 runs.
An onsite laboratory was set up to
reduce as much of the test data as
possible. Facilities were available to
properly clean all of the test equipment,
recover samples, and to desiccate the
particulate catches. A certified accurate
balance was used to weigh all of the
samples. A chloride ion-specific
electrode and supporting electronics were
available to measure chloride
concentrations in the Method 5 train
back-half impingers. Chloride audit
solutions were onsite to audit this
procedure.
The impactor data were processed at
the conclusion of the test using the
Particulate Data Reductionl (PADRE) data
program.
HYDRO-SONICS AIR POLLUTION CONTROL DEVICE
General Description
Lone Star Steel Co. originally devel-
oped this wet scrubber to control particu-
late emissions from various iron and
steelmaking operations. It has been used
on such sources as electric arc furnaces,
coke oven emissions, open hearth steel
furnaces, and sintering plants. The
scrubbers have also been used on exhaust
streams containing uranium hexafluoride
and its hydrolysis products with particu-
late removal efficiency consistently
exceeding 99 percent.
There are three versions of the
scrubber: The Steam-Hydro, the Tandem
Nozzle, and the SuperSub. All versions
have the same basic concept. Water is
atomized into the waste gas stream forming
water droplets of about the same size as
the particulate. The gas stream then
enters a turbulent contact zone in which
the particles are wetted and vapors are
absorbed. Particle growth then takes
place in an agglomeration zone. Because
of the design, a single waste droplet may
contain hundreds of micronic and sub-
mi cronic dust particles. As a result of
the growth of droplets containing partic-
ulate into increasingly larger size, the
initial size of the particulate has only a
small effect on its removal. Actual
removal of the agglomerated particle is
accomplished in a specially designed
low-pressure-drop cyclone. Water and
p.articulate are gravity drained from the
cyclone bottom, and the cleaned gases exit
through the top. Demisters are not
required.
, The Steam-Hydro employs a supersonic
ejector drive to provide the energy for
pumping and cleaning the polluted gas.
Steam or compressed air is commonly
68
-------
employed as the working fluid. The gas is
drawn into the unit by the ejector nozzle
which is fitted with a water injection
ring at the exit of the nozzle. The '-
expanding jet causes violent shattering of
the water droplets and turbulent mixing of
gas and water. The steam version is most
attractive when there is a source of waste
heat available to generate the
high-pressure steam.
The Tandem nozzle scrubber uses a fan
drive to pump the polluted gas and to pro-
vide the energy to generate the fine water
droplets. The system used two subsonic
nozzles and agglomeration sections in
series. Both nozzles are equipped with
water spray rings. The first section
serves to condense vapors, remove the
larger particulates, and initiate growth
of the fine particulate. Additional water
is atomized at the exit of the second
nozzle and is turbulently'mixed to
continue the agglomeration of the
particles.
The SuperSub version of the system is
a combination of the above two system
concepts. A small supersonic ejector
(steam or air) is located upstream of the
subsonic nozzle. The systems main driving
force is fan power as in the Tandem Nozzle
version. This arrangement provides good
water atomization for fine particle
control coupled with the lower energy
requirements of the fan drive. Water con-
sumption is about the same as a venturi
scrubber.
Test Results
Test data for these units were col-
lected between March 15, 1984, and March
23, 1984. Given in Table 1 are the
removal efficiency data for total particu-
late and chloride organized by the APCD
operating version. With the exception of
two tandem nozzle runs (10 and 11),
chloride removal for all combinations was
greater than 98 percent. Runs made using
recycle water from the ENSCO scrubber show
higher removal of chloride than those
using freshwater. Since the recycle water
contained some alkalinity, this is not
surprising. In a commercial application,
alkalinity would be added to the scrubbing
water. Thus, chloride removals of 99
percent or better should be expected for
any version of this unit.
Particulate removal efficiency ranges
from about 82 to 88 percent for the
Steam-Hydro and Tandem Nozzle versions.
The SuperSub version achieved a particu-
late removal efficiency of about 95
percent. It should be noted that these
removal efficiencies refer only to the gas
stream brought into the APCD and not the
efficiency that might be obtained on the
entire gas stream from the ENSCO
incinerator. Since the APCD connection to
the main gas duct should have resulted in
a bias toward the smaller particles, we
would expect that substantially higher
removal efficiencies would have been
obtained if the APCD had been treating the
unbiased gas stream.
Shown in Table 2 are the percent mass
less than 1 micron at the inlet and outlet
of the unit. Much of the particulate
matter exists below 1 micrometer, whether
at the inlet or outlet of the control
device. It must be noted that design
philosophy of the unit is to agglomerate
fine particles. Thus, it is to be
expected that some the submicron particles
entering the device exit as particles
greater than 1 micron in size. These data
indicate that the SuperSub configuration
is the most effective of the versions
tested for control of the submicron
particulate matter from this source.
ETS, INC. DRY SCRUBBER
General Description
The unit has two major components:
the dry reactor and a particulate
collection device. The patented dry
reaction has a number of unique com-
ponents. The basic operating principles
are as follows. Flue gas is directed
cyclonically into the reactor. The
rotating slinger unit (driven by a
hydraulic motor) delivers the dry reactant
(usually 200 mesh hydrated lime) perpen-
dicular to the flue gas flow. This
creates maximum mixing and intimate con-
tact of the reactant and pollutants. An
internal recirculator, with no moving
parts, is located above the slinger. This
device increases the contact time and
enhances removal of the acid gases. The
slinger then directs the dry reaction
products down and into an expansion
section where the larger particles are
removed. The finer particulate matter is
carried into the particulate collection
device.
The particulate collector can be a
conventional baghouse, an electrostatic
fabric filter (ESFF) baghouse, or a
Reduced Entrainment Precipitator (REP)
69
-------
developed by ETS. The lime dust entrained
1n the flue gas continues to react with
acid gas components throughout the trans-
port ductwork. In addition, the lime dust
aids in building a reactive filter cake on
the fabric filters. This serves two
purposes. First, reaction with acid gas
components continues until final particu-
particulate removal occurs. Secondly, the
precoat assists in removing the fine par-
ticulate without excessive pressure drops
across the filters.
This system is totally dry. The
reactants are delivered as dry powder, not
as a s-lurry required for the spray dryer
type scrubbers. The reacted product is
also a dry powder and can be handled much
as one handles dust collected in a
baghouse. Since no water is used at any
point in the system, there are no mist
carryover problem, little or no corrosion
in the exhaust stack which remains dry and
minimum loss of stack gas buoyancy since
this energy is not used to evaporate
water.
Unit Tested
The unit tested has a rated capacity
of 2000 ACFM.
The dry scrubber was connected to a
pulse jet baghouse which used Nomex®
fabric cartridges. An inducted draft fan
was located at the outlet of the baghouse
forcing the cleaned gas into the 12-inch
I.D. vertical exhaust stack.
Test Results
The ETS system was installed during
the last week of March 1984. The system
was operated and tested during two
separate periods, April 1-10 and April
23-26, 1984. Two reactant materials,
hydrated lime and nacholite were evaluated
for HC1 removal effectiveness.
Given in Table 3 are the particulate
and HC1 removal efficiency data for the
two reactants at the various stoichiome-
tries. Six of the nine tests using lime
as the reactant indicate an HC1 removal
efficiency of over 98 percent. Only one
of the tests using nacholite achieved over
90 percent HC1 removal. The stoichio-
metric ratio (SR) ranges from 2 to 9 for
the tests in which lime was the reactant.
The data suggest that the SR for lime must
be nearly 3:1 to insure scrubbing
efficiencies of 99 percent.
Due to limitation in the feed equip-
ment, nacholite was injected at much lower
stoichiometric ratios. The HC1 removal
efficiency does not appear to correlate
with the nacholite/HCl stoichiometric
ratio up to a ratio of 1.7. Therefore,
from these tests, it is not possible to
determine the nacholite/HCl ratio required
to achieve 99 percent HC1 removal.
The particulate removal efficiencies
are calculated strictly from the Method 5
data. No allowance has been made for the
reactant materials added in the dry'
reactor. For most of the runs, the weight
of reactant added ranged from about 40
percent to over 100 percent of the-weight
of particulate entering the system from
the incinerator. The reactants added were
100 percent less than 200 mesh (74
micrometers). Obviously some of the
material was much smaller and may have
passed through the baghouse.
, The calculated efficiencies for the
first 10 runs were all greater than 90
percent with most in the 95 to 98 percent
range. It should be noted that, due to
the bias toward smaller particles caused
by the slipstream sample withdrawal, this
is the removal efficiency for the finer
particles in the ENSCO incinerator exhaust
gas. Since the unit would be expected to
remove the larger particles more easily,
the removal efficiency for the total gas
stream would be expected to be even
higher.
It is suspected that the low effi-
ciencies measured for Runs 11 through 16
were due to persistent difficulties in
sealing the cartridges in the baghouse.
In addition, a tear was discovered in one
cartridge after Run 11. We do not
believe, therefore, that the particulate
removal efficiencies reported for Runs 11
through 16 accurately reflect the partic-
ulate removal capabilities of the unit.
The particulate control capability of
the ETS unit as a function of particle
size was also determined.
Shown in Table 4 are the percentages
of particulate matter less than I microm-
eter in size in the inlet and outlet
samples. This implies better removal of
fine particulate than coarse particulate.
This is counter to the expected
performance of a fabric filtration system.
In our opinion, these data imply that a
significant fraction of the particulate
matter in the unit exhaust originated ,with
the lime feed.
70
-------
VULCAN ENGINEERING COMPANY HI-TAC FILTER
General Description
The high temperature air cleaning
(HI-TAC) device was developed by Vulcan
Engineering Company for the removal of
particulate matter from gas streams at
higher collector inlet temperatures than
previously possible. The unit is not
designed nor equipped to control HC1
emissions. Although the HI-TAC unit looks
much like a conventional baghouse, there
are substantial differences.
The filter media is entirely
metallic. This results in greater
resistance to moisture, temperature,
corrosion, abrasion, and pressure than is
provided by conventional fabric media.
Although temperatures below 500°F do not
decrease the collection efficiency, the
media is designed to operate in the 600°F
to 1350°F range. Temperature excursions
to 2000°F can be tolerated. The filter
elements will even tolerate flame
temperatures.
The unit was developed for use in the
iron and steel industry to control partic-
ulate from the many high temperature
sources. The initial tests on the unit
were performed on exhaust gas from an iron
cupola. The gas entering the unit was
750°F and contained 1.39 grains of
particulate per dry standard cubic foot
(dscf). The unit operated with an
air-to-cloth ratio of 16:1 and collected
99.2 percent of the particulate.
Unit Tested
The unit tested is completely porta-
ble and self-contained. Connection to an
electrical power source and the exhaust
duct are the only setup requirements. The
unit is mounted on a Talbert 47-ft double
drop Tandem axle trailer.
Each of the four cleaning modules
mounted on the trailer houses four removal
cartridge elements. Each element consists
of a metallic reinforcing framework
covered with stainless steel mesh. The
weave may be varied to determine appro-
priate design for a particular source.
Test Results
During this test program, there were
three (3) types of weaves in the unit.
These were:
Module I - 150 x 105 plain weave
Module II - 325 x 325 twilled weave
Module III and IV - 50 x 250 dutch
weave
The total removal efficiencies for partic-
ulate matter and chloride are shown fn
Table 5. As can be seen, removal effi-
ciency for particulate matter is quite
erratic, varying from 34 percent to 98
percent. Excluding run number 1, the
average removal efficiency is 89.7
percent. Inspections of the cartridges
before and after test runs indicated that
cake formation on the filter fabric was
not reliable. During several inspections
following cartridge cleaning, it was
observed that a very large portion (about
70 percent) of the fabric surface was
still covered by the filter cake, with
about 20 percent of the surface clean down
to the fabric. The sticky particulate
from the incinerator interfered with
proper cake formation and filter cleaning.
It is possible that fabric weave changes
or precoating of the fabric might correct
this difficulty. The test program was not
designed to adequately explore these
options.
Particulate removal versus gas flow
rate through the APCD and particulate
removal versus air-to-cloth ratio showed
no clear trends. It is suspected that the
peculiar cake formation properties and
resultant filter cleaning problems made
these standard plots of little value.
Recognizing the difficulties cited
above for particulate control, we
attempted to assess, with much trepi-
dation, the removal efficiency at various
particulate size ranges. These data
indicate 90 percent or better control of
particles of less than 1 micrometer but
decreasing efficiency as the particle size
increases.
These apparently strange results may
be due to the method used, determining
small differences between highly variable
numbers, or to agglomeration of fine,
sticky particles passing through the
filters.
In summary, the test results indi-
cate that effective control of the fine
particulates generated by hazardous waste
combustion may be achievable by the Hi-Tac
unit. Additional development work to
address the problem of sticky particulate
will be required.
CONCLUSIONS
1. All versions of the Hydro-Sonic units
tested achieved excellent HC1
control. Ninety-nine percent of
HC1 was obtained without adding
71
-------
additional alkalinity to the ENSCO
scrubber recycle water. With addi-
tional alkalinity, any of these units
should be capable of well over 99
percent HC1 removal.
2. The Tandem Nozzle SuperSub Model 100
achieved the best particulate removal
of the three Hydro-Sonic units
tested.
3. The ETS dry scrubber achieved both
high removal efficiencies for HC1 and
particulate.
4. The ETS scrubber reagent consumption,
hydrated lime, appears to be high (3
moles lime per mole HC1) but cannot
be stated with confidence since no
attempts were made to improve
utilization through reagent recycle.
5. The ETS unit does not presently have
capability to adjust reagent feed
rate to accommodate rapidly varying
HC1 content in the gas to be treated.
6. The Vulcan Hi-Tac unit is not
designed to, and cannot presently,
remove acid gases from the exhaust
gas being treated. Therefore, it is
not applicable to incinerators
burning significant amounts of
halogenated hazardous wastes.
7. Further development work and/or emis-
sions testing will be required before
the Hi-Tac unit demonstrates the
capability to reliably control
sources which might product sticky
particulate as encountered in the
exhaust of the ENSCO incinerator.
8. Considerable variability was encoun-
tered in the ENSCO exhaust gas for
particulate concentration,'
particulate size distribution, HC1
concentration, flue gas moisture con-
tent, and exhaust gas temperature.
It is necessary that any air pollu-
tion control device installed on a
hazardous waste incinerator have
design features that allow
compensation for this variability.
9. Of the APCDs tested in this project,
the Hydro-Sonic Tandem Nozzle
SuperSub gave the best overall
performance in terms of HC1 and
particulate removal and ability to
accommodate variability in the gas
stream being treated.
ACKNOWLEDGMENTS
This report was prepared under EPA
Contract No. 68-03-3149. Research
Triangle Institute (RTI) is the prime
contractor. The field testing described
was subcontracted to Engineering-Science,
Inc. (ES) and was carried out under RTI
direction. Mr. Harry Freeman of EPA's
Thermal Destruction Branch, Hazardous
Waste Engineering Research Laboratory,
Cincinnati, Ohio, was the Project Monitor.
RTI and ES wish to acknowledge the
technical contribution of ENSCO, Inc.
Hydro-Sonic® Systems, ETS, Inc., and
Vulcan Engineering and to express
appreciation for their participation in
this project.
REFERENCES
1. Tatsch, C.E., W. M. Yeager, and G. L.
Johnson, 1984. PADRE: A computer-
ized data reduction system for cas-
cade impactor measurements. Journal
of the Air Pollution Control
Association, Volume 34. No. 6. pp.
655-660.
72
-------
TABLE 1. PARTICULATE AND CHLORIDE REMOVAL EFFICIENCIES: HYDRO-SONIC SCRUBBER
Configuration3
SH/Lo/R
SH/Lo/F
SH/Hi/R
TN/-/R
TN/-/F
SS/-/F
Parti cu late
efficiency
(Percent)
81.6
87.4
88.2
92.1
86.3
95.4
Chloride
efficiency
(Percent)
99.1
98.7
98.8
99.8
96.0
98.3
Configuration codes are: SH = Steam-Hydro; TN = Tandem Nozzle;
S = Supersub; Hi= High energy input; Lo = Low energy input; - = not applicable
to this version; Fresh = Freshwater used in APCD: R = Recycle water from ENSCO
scrubber used in APCD.
TABLE 2. COMPARISON OF FINE .PARTICLE (<1 MICRON) ENRICHMENT FACTORS
(OUT/IN) FOR VARIOUS OPERATING CONDITIONS
Operating
Condition
SH/Lo/F
SH/Lo/F
TN/- /F
SS/- /F
SS/- /F
Inlet
Impactor
Run No.
2
3
4
5
6
Percent
of Mass
<1 Micron
43
57
38
64
70
Outlet
Impactor
Run No.
3
4
5
6
7
Percent
of Mass
<1 Micron
69
67
60
36
26
Out/In
1.6
1.2
1.6
0.6
0.4
73
-------
TABLE 3. SUMMARY RESULTS: ETS UNIT
Reactant
used
Lime
Lime
Lime
Lime
Lime
Lime
Lime
Lime
Lime
Nahcolite
Nahcolite
Nahcolite
Nahcolite
Nahcolite
Nahcolite
Nahcolite
Stoichiometric
ratio
2.80
4.53
4.50
6.61
8.71
8.07
5.29
1.94
2.42
0.70
0.53
0.87
-
0.56
1.73
0.98
HC1
Efficiency
percent
98.7
89.0
98.0
99.2
98.9
99.6
99.8
56.3
79.1
93.2
82.0
54.7
(2)
53.8
68.1
33.1
Participates
- efficiency
percent
91
97
96
95
95
98
90
98
98
96
(1)
76
75
85
83
38
1) Torn baghouse cartridge.
2) Sample lost.
TABLE 4. COMPARISON OF FINE PARTICLE (
-------
TABLE 5. VULCAN HIGH TEMPERATURE BAGHOUSE EFFICIENCY SUMMARY
Run
number
System
efficiency
(Percent)
1
2
3
4
5
6
7
8
9
10
34
93
90
92
94
84
89
74.8
90.0
98.4
75
-------
CASE STUDIES OF WASTE TREATMENT AT HAZARDOUS WASTE FACILITIES
C. Clark Allen
Research Triangle Institute
P. 0. Box 12194
Research Triangle Park, North Carolina 27709
and
Benjamin L. Blaney
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
ABSTRACT
Commonly practiced commercial waste treatment processes were investigated for the
removal of volatile organic compounds from waste. The effectiveness of thin film
evaporators, steam stripping, and distillation to treat waste was evaluated by field
sampling and analysis. The major sources of treatment process residuals associated with
these treatment processes were also investigated.
INTRODUCTION
The EPA Office of Air Quality Plan-
ning and Standards (OAQPS) is developing
regulations to control emissions from
hazardous waste treatment, storage, and
disposal facilities (TSDFs). The purpose
of the OAQPS air emissions regulations is
to protect human health and the
environment from emissions of volatile
organic compounds (VOCs), particulates,
and aerosols.
The sources of TSDF emissions include
storage tanks, treatment processes, sur-
face lagoons, landfills, land treatment,
and drum storage and handling facilities.
There are approximately 5,000 locations in
the United States where one or more of
these activities are in progress at a
TSDF.
To date, research has concentrated
upon characterization of uncontrolled
emissions from these sources through field
measurements and upon determining the
reliability of emissions models. Options
identified for controlling emissions from
TSDFs include banning the handling of
wastes from sources where emissions rates
are high, treatment of wastes to remove
volatiles, and the use of in-situ (add-on)
control techniques. In addition, changes
in waste management practices (such as
using holding tanks instead of ponds) may
be a cost-effective control option.
Treatment of waste to remove the VOCs
appears to be a viable emissions control
option and is being used at several TSDFs.
In general, it is attractive because it
can be used by the waste generator or TSDF
operator to remove volatiles from the
waste before there is much opportunity for
release of VOCs to the air. Treatment may
be the most cost-effective control tech-
nique for sources with large surface
areas, such as land treatment facilities
and waste treatment and storage lagoons.
Waste treatment may also serve to reduce
ground water impacts from waste disposal
in landfills or by treatment lagoons, etc,
as well as to reduce the amount of wastes
disposed.
This paper presents the results from
field studies of three treatment tech-
niques: thin film evaporation, steam
stripping, and distillation. The
processes investigated were located at
waste recycling facilities. The data
collected included volatile removal effec-
tiveness, the disposal of residuals, and
limitations of the technology.
VOC TREATMENT PROCESSES
Steam stripping will remove low
levels (less than 1 percent) of volatile
76
-------
organic compounds from oils and
wastewater. Steam stripping can operate
at a lower temperature than a thin film
evaporator and is therefore more appro-
priate for some reactive materials. The
volatile organic compounds can be
recovered from the process by decanting if
the organic phase separates from the
condensed steam.
Thin film evaporators are used in
many solvent recovery operations.
Typically, 80 percent of the volatile
materials can be recovered, with a sludge
containing residual VOC obtained at the
bottom of the evaporator.
Thin Film Evaporators
'Samples were taken from the over-
head, bottoms, and feed of wastes treated
at three thin film evaporators (1,2,3).
The waste treated at Plant A repre-
sents a class of waste oils containing a
small amount of solids and approximately 5
percent VOCs. During the visit to Plant
A, a "batch" consisting of mixed chlori-
nated xylenes was being processed through
the thin film evaporator (1). This
material was relatively clean, and
therefore 95 percent was being taken
-overhead, with the bottoms or treatment
process product being acceptable for fuel.
The data obtained from samples of the
feed, the bottoms, and the product are
presented in Table 1. The headspace
analysis demonstrated that the volatile
organic material was removed by the thin
film evaporator from the bottoms. The
following removal effectiveness was
estimated using a material balance from
the information presented in Table 1
(based upon the headspace analysis and 95
percent product recovery):
Methylene chloride 99.1
Chloroform >99.99
1,1,1-Trichloroethane >99.5
Toluene <85.0
Freon TF 80.0
The concentrations of the volatiles are
reduced in the bottoms of the thin film
evaporators. The vapor pressures of the
more volatile compounds are more than 90
percent reduced by the treatment.
Sampling and analysis was carried
out on another thin film evaporator at
Plant B (2). The composition of the
material evaluated was a volatile organic
liquid composed of alcohol and aromatics.
The composition of liquid samples is
presented in Table 2. The bottoms
materials from the thin film evaporator
had a similar volatility characteristic as
the feed, although the volume of the waste
was reduced by about 50 percent. In
general, the more volatile material was
concentrated in the product stream and the
less volatile materials tend to be more
concentrated in the bottoms.
The waste being processed in the thin
film evaporator at Plant C was acetone
containing xylene, low levels of chlori-
inated solvents, and dissolved polymeric
resins (3). The acetone and other low
boiling point compounds were somewhat
more concentrated in the distillate and
xylene was enriched in the bottoms. The
analysis is presented in Table 3. Because
of the requirements to maintain the resins
in solution, the VOCs in the bottoms at
the end of the run was not substantially
different than the VOCs in the feed and
the concentrations in the vapor phase did
not change significantly (a reduction of
acetone concentration in the vapor from
378 to 308 mg/L) although the volume of
waste was reduced by 10 percent.
Steam Stripping
Plant D stripped waste material by
direct injection of live steam into a
waste batch. The process of stripping
continued until the desired concentrations
were achieved in the waste (4). Four
batches of liquid wastes were evaluated:
[1] an aqueous xylene, [2] a chlorinated
organic-oil mixture, [3] a chlorinated
organic-water mixture, and [4] a mixed
solvent-water mixture.
The test results for the aqueous
xylene batch are presented in Table 4 and
Figure 1. The rates of removal of all the
compounds appeared to be roughly the same,
with the heavier materials being somewhat
more slowly removed. There was removal of
all the VOCs to less than 300 ppm in the
final treated waste.
In the chlorinated organic-oil mix-
ture, there were two major components
which were removed from the organic waste.
These were the trichloroethane and
chloroform. Chloroform was quickly
removed from the process with only 23 ppm
remaining after 80 minutes. The trichlor-
oethane was somewhat more slowly removed
with 4,000 ppm or 0.4 percent residual
trichloroethane in the oil material at the
end of the batch. Since the volume of the
organic batch was changing as the more
volatile component was removed, the
77
-------
concentration of the trichloroethane did
not drop initially but appeared to
decrease much more rapidly in the later
stripping as the batch temperatures
increased. The average stripping rate of
the chloroform and the trichloroethane was
approximately the same.
When the batch of waste trichloro-
ethane had been processed, a residual tank
of condensed steam containing some of the
volatile organics was collected in the
MST. This chlorinated organic-water
mixture was stripped and the concentra-
tions of the volatile materials monitored.
The results are presented in Table 5. In
this batch of aqueous waste (which
differed from the aqueous xylene batch in
that there was no organic phase of xylene
present during stripping), the stripping
rates varied for the different compounds,
with the more volatile materials such as
chloroform and acetone stripped at a
relatively faster rate. As seen from Fig-
ure 2, an approximately logarithmic
decrease in the concentrations of the
major volatiles was observed. The major
component present in the aqueous waste was
trichloroethane, and the waste was
stripped down to 1 percent trichloroethane
in 60 minutes. Approximately another 40
minutes of stripping would possibly be
required to drop the trichloroethane level
to .1 percent based the extrapolated curve
in Figure 2.
The fourth batch was a mixed-solvents
mixture produced by the project team to
simulate a dilute aqueous waste stream
containing miscible and immiscible
solvents in the following approximate
concentration: toluene (0.1 percent) and
acetone (0.6 percent), trichloroethane
(0.2 percent), and xylene (0.2 percent).
The concentrations of each of the volatile
components in this batch were reduced to
less than .03 percent after 33 minutes of
stripping. The toluene, xylene, and
trichloroethane stripped at approximately
the same rate, with acetone approximately
twice as fast. The initial concentrations
of toluene and xylene were not the same as
initially charged to the batch, apparently
because of inadequate mixing in the batch
while the sample was taken.
For all four waste streams, the equi-
librium concentration of the volatile
organic compounds at equilibrium in the
cool waste decreases as their concentra-
tions decrease due to steam stripping.
Based on a headspace analysis of the
treated waste in the laboratory,it is
apparent that the volatility character-
istics of the wastes are dramatically
altered by the steam stripping process.
These volatility characteristics of the
cooled wastes are of concern because of
the relationship between the vapor
pressure of the volatile components and
its release into the atmosphere upon
disposal. The waste material generally
showed at least an order of magnitude
decrease in the vapor concentrations at
equilibrium with the waste.
Table 6 presents a summary of the
data analysis of the four batches evalua-
ted at Plant D. The correlation coeffi-
cient was relatively high for many of the
rates of stripping. The steam rate
influences the rate of stripping, so the
rate constants are presented in a
dimension!ess form to account for the
steam rate and batch size.
The values of the dimensionless rate
constant were much smaller than would have
been expected for dilute organic materials
in water (Henry's Law). Table 7 demon-
strates that the equilibrium vapor
pressures were lower in the mixed waste
system than in dilute concentrations. The
rate constants were roughly the same for
chloroform and 1,1,1-trichloroethane for
each of the aqueous batches.
AIR EMISSIONS
Besides solid residuals produced by
these treatment processes the significance
of treatment process air emissions is also
of interest. The treatment after disposal
would be of little use if substantial
quantities of contaminated residuals were
produced. For these processes considered
here, the principal residuals are in the
form of either organic materials which
could be incinerated or aqueous streams
with the VOCs removed. A small amount of
the VOCs were lost to the .atmosphere.
Thin Film Evaporators
Air emissions were evaluated from
three thin film evaporators and two steam
stripping units. The major air emission
source identified at the thin film
evaporator process was the vent from the
vacuum pump. It'should be noted in the
processes investigated that in all cases
any air emissions were estimated as only a
small fraction of the amount of the
organics recovered; however,- it is
78
-------
possible for the system to have' a rela-
tively significant loss of some of the
particularly volatile materials from the
vacuum pump vent in a mixed volatile
system.
The volatile materials, methylene
chloride and chloroform, were present in
the feed stream at Plant A in somewhat
higher concentrations than in the product,
and were not present to a significant
extent in the bottoms. This suggests a
possible loss of some of the more volatile
materials to the atmosphere through the
vacuum pump vent. The following list
presents the estimated VOC losses (as
determined by mass balance) to the
atmosphere as determined by a material
balance, according to boiling point:
Methylene chloride 57%, 40.7°C
Chloroform 97%, 61.3°C
1,1,1-Trichloroethane -20%, 74°C
Toluene -17%, 110°C
Negative values are attributed to uncer-
tainty in the data (see below).
The exhaust of the vacuum pump did
contain some volatile materials during
startup. Carbon adsorption tubes were
analyzed for VOCs present in air samples
from the vacuum pump vent. Five tubes
were analyzed, one 10-minute sample (85
mL/min)j two 5-minute samples, and two
field blanks.
The analysis of the gas samples of
the vacuum pump discharge vent is
presented in Table 8. Chloroform and
1,1,1-trichloroethane were present in the
vent discharge with statistically signifi-
cant quantities (relative to the field
blank) of the higher boiling VOCs
(toluene, ethyl benzene, xylene). The
absence of methylene chloride captured on
the carbon tube and the apparent insensi-
tivity of the quantity of chloroform
captured to the length of sampling time
suggest that the quantities of the more
volatile chlorinated compounds may be
greater than reported in Table 8.
Although there is an expected error
of 15 percent or less in the pollutant
concentration data, and even greater error
in differences and estimates based on gas
partitioning, the data do suggest that the
lower boiling VOCs may be subject to much
higher process losses than higher boiling
VOCs. These data suggest that thin film
evaporators operating under a vacuum can
have potentially significant VOC emis-
sions. The emission rate would depend on
the operating conditions of the still.
The data reported in Table 8 was taken
during startup and the concentrations
observed were expected to be lower after
the transitional period of startup. Some
of the waste treatment plants visited
indicated that control vent emissions
could be reduced by multiple passes
through the evaporator (?.}, or by using a
second vent condenser (5).
At Plant B, theJargest source of air
emissions observed from the thin film
evaporator was from the vacuum condenser
vent. The concentrations of VOCs in the
vent were roughly one-half the
concentrations observed in the laboratory
from equilibrium headspace analysis of the
condensate. The concentration of air
emissions would depend on the flow rate of
noncondensibles out of the vent. This
flow would depend on the concentration of
noncondensibles in the waste and the
magnitude of any potential system leaks
(into the process).
The concentrations of VOC in the
process vent of the thin film evaporator
was evaluated for Plant C for comparison
to the vacuum pump vent at Plants A and B.
(The thin film evaporator at Plant C was
operating at atmospheric pressure.) The
analysis of the air samples from the
process vent indicated that no significant
(relative to the field blank) air emis-
sions were observed from the process vent.
The vent pipe was located perpendicular to
the wind flow on the building exterior,
and due to the wind gusts air flowed
alternatively in and out of the vent (as
measured by the Alnor velometer). No
odors were detected at the vent.
At Plant C, the largest source of air
emissions which was identified was the
product storage tank. The product storage
tank had a loose fitting steel top with a
gap of approximately 8-3 cm (3-1 inch) and
was 1 raeter (4 ft) wide with an area of
620 cm (0.67 ft ). Acetone was the major
component lost from the storage tank at
0.56 and 0.70 g/sec- Based on an emission
factor of 2.0 x 10 g/mol/cm sec and a
vapor concentration of 383 mg/1 (measured
at 25°C in the laboratory), the estimated
emissions for a fully exposed surface is
•1.7 g/sec, greater than the field rate by
a factor of 3. (At the time of the test
of the produce storage tank, the plant was
planning to replace the current receiver
tanks with a more enclosed tank system.)
79
-------
Steam Stripper
In contrast to the open product stor-
age tank at Plant C, the product storage
vessels at Plant D were enclosed inside a
building with a vent to the roof from each
tank. The air emissions from the process
vents at Plant D were estimated on two
batches, the xylene aqueous batch and the
production waste trichloroethane batch.
Air concentrations were measured, but no
flow from the process storage tanks was
detected. The concentrations of the vapor
at equilibrium over the distillate
obtained midway through the process
corresponded with concentrations obtained
approximately concurrently from the air in
the vents. The volume emitted from the
receiver vent was estimated on the basis
of the data obtained from the dry gas
meter readings. The emission factors were
estimated by obtaining the ratio of the
estimated grams of emissions to the total
amount of wastes. The total VOC emission
factog (g/g waste) was estimated to be 2.7
x 10 for the 1,1,1-trichloroethane
production run and 3.8 x 10 for the
aqueous xylene run. In both cases, the
emissions were only a small fraction of
the VOCs recovered from the process. The
rate of emission of VOCs from the storage
tanks was estimated a 0.006 g/sec.
The condensate receiver vent concen-
trations of the methylene chloride,
chloroform, and trichloroethane from the
1,1,1-trichloroethane batch were much
greater at the beginning of the process
than at the middle of the process. This
is not unexpected since the concentrations
of these components in the condensate are
much greater at the beginning of the cycle
than midcycle, as determined from the
waste analysis. The concentrations of the
volatiles in the vent from the product
storage and the MST tank vent were
substantially lower than the concentra-
tions obtained at equilibrium within the
storage tank. This suggests that the air
emissions are substantially lower than
predicted from the equilibrium values of
the tank contents.
The air emissions from the condenser
vent were only a small fraction of the
recovered VOCs. In addition, the air
emissions from the storage tanks at Plant
D were much lower than from the storage
tank surface exposed to the flow of air at
Plant C.
CONCLUSIONS
1. Thin film evaporators are useful
for waste volume reduction and
VOC removal in systems with high
boiling point organics.
2. There are potential emissions
from the condenser vent when
volatiles are removed with high
boiling organics in a thin film
evaporator.
3. Thin film evaporators may not
substantially alter the
volatility characteristics of
wastes containing primarily
VOCs.
4. Steam stripping is effective for
removing VOCs from wastes: [1]
aqueous, [2] liquid organic/
aqueous mixtures, and [3] liquid
organic.
5. The air emissions from thin film
evaporation and steam stripping
were only a small fraction of
the waste treated.
REFERENCES
1. Allen, C. C., and 6. Brant, 1984a.
Hazardous waste pretreatment for
emissions control: field evaluations
at Plant A. EPA Contract No.
68-03-3149, Work Assignment 25-1.
2. Allen, C. C., and G. Brant, 1984b.
Hazardous waste pretreatment for
emissions control: field evaluations
at Plant B. EPA Contract No.
68-03-3149, Work Assignment 25-1.
3. Allen, C. C., and G. Brant, 1984c.
Hazardous waste pretreatment for
emissions control: field evaluations
at Plant F. EPA Contract No.
68-03-3149, Work Assignment 25-1.
4. Allen, C. C., and G. Brant, 1984d.
Hazardous waste pretreatment for
emissions control: field evaluations
at Plant C. EPA Contract No.
68-03-3149, Work Assignment 25-1.
5. Allen, C. C., and G. Brant, 1984e.
Hazardous waste pretreatment for
emissions control: field evaluations
at Plant E. EPA Contract No.
68-03-3149, Work Assignment 25-1.
80
-------
TABLE 1. THIN FILM EVAPORATOR WASTE COMPOSITIONS, PLANT A
Bottoms* Feed Product
mg/L Liquid mg/L Liquid mg/L
(in vapor) (by vol) (in vapor) (by vol) (in vapor)
Methyl ene chloride
Chloroform
1,1, 1-Tri chl oroethane
Toluene
Mixture of High
Boiling Hydrocarbons
Freon TF
0.03
<0.01
<0.01
0.03
-
0.24
2.0%
1.5%
0.7%
1.3%
94.4%
1.7
5.1
0.11
<0.01
-
0.06
0.9%
ND
ND
1.6%
93.9%
1.8
0.97
0.14
0.14
0.04
-
1.5
ND indicates compound not detected.
*Bottoms solid upon cooling.
TABLE 2. ANALYSIS OF LIQUID SAMPLES, THIN FILM EVAPORATOR, PLANT B
Feed Product Bottoms*
Liquid mg/L Liquid mg/L mg/L
(by vol) (in vapor) (by vol) (in vapor) (in vapor)
Isopropyl
alcohol
Freon TF
Toluene
Ethyl benzene
Xylenes
38.2%
0.6%
0.4%
11.4%
49.2%
0.75
38.
0.58
5.5
22.
53.8%
0.7%
0.4%
8.4%
34.0%
1.1
62.
0.94
5.3
19.
1.6
5.3
0.32
9.0
39.0
'''Bottoms solid upon cooling.
TABLE 3. ANALYSIS OF LIQUID SAMPLES, THIN FILM EVAPORATOR, PLANT C
Compound
Feed
Liquid Gas
(%) (ppm)
Distillate
Liquid Gas
(%) (ppm)
Bottoms
Liquid Gas
(%) (ppm)
Acetone
Freon TF
1,1, 1-Tri chl oroethane
Tri chl oroethyl ene
Toluene
Ethyl benzene
Xylene
Tetrachl oroethyl ene
74.3
0.1
1.5
0.2
0.5
5.9
0.6
378.0
2.0
17.9
0.1
0.3
0.1
2.1
2.4
82.2
<0.1
2.2
0.3
0.9
0.3
2.0
0.5
383.0
2.0
19.1
0.1
0.2
<0.1
0.2
1.6
60.6
0.1
0.9
<0.1
0.9
0.3
<0.1
13.6
308.0
1.5
9.2
0.1
4.1
0.4
<0.1
5.0
81
-------
TABLE 4. WASTE VOC CONCENTRATION DURING STEAM STRIPPING:
AQUEOUS XYLENE BATCH
Process Sample
time (nrin) number
Concentration (mg/L)
Chloro-
Acetone IPA form TCEA TCE EB Toluene Xylene
0
15
64
86
45-3
45-5
45-6
45-7
39
10
<6
<6
960
640
47
<6
5,200
2,300
350
. 17°
170
99
33
20
290
230
72
<20
360
560
56
100
86
32
17
42
2,000
480
410
270
IPA s isopropanol
TCEA = 1,1,1-trichloroethane
TCE * trichloroethylene
EB - ethyl benzene
TABLE 5. WASTE VOC CONCENTRATIONS DURING STRIPPING:
1,1,1-TRICHLOROETHANE MST BATCH
Process time
(minutes)
0
22
43
57
Sample
number
45-29
45-31
45-32
45-33
Concentration (mg/L)
Acetone
290
71
9
<6
Chloroform
1,600
250
<34
<34
TCE
180,000
71,000
28,000
12,000
EB
44
30
24
12
IPA
37
<6
<6
<6
TABLE 6. LINEAR CORRELATION OF THE LOGARITHM OF THE WASTE CONCENTRATION
WITH THE STRIPPING TIME
Batch Component
1
1
1
1
2
2
3
3
3
4
4
4
4
Isopropyl alcohol
Chloroform
1 , 1 , 1-Tri chl oroethane
Tetrachloroethane
Chloroform
1,1, 1-Tri chl oroethane
Acetone
Chloroform
1,1, 1-Tri chl oroethane
Acetone
Toluene
Xy 1 ene
1,1, 1-Tri chl oroethane
Correlation
coefficient
-.974
-.9977
-.9971
-.9975
-.9974
-.9593
-.9925
-.9971
-.9966
-.9377
-.9875
-.990
Stripping
(min"1)
.0507
.0393
.0242
.0122
.0998
.078
.0806
.0844
.0468
.28
.0657
.0859
.0684
rate constants
(dimension! ess)
14.7
11.4
7.0
6.47
19.7
15.3
10.0
10.5
5.82
25.4
5.97
7.8
6.21
*Dimensionless rate obtained by dividing the rate constant (min ) bythe
ratio of the steam rate (L/min) to the amount of waste (L).
S
82
-------
TABLE 7. A COMPARISON OF THE RATE CONSTANTS OF STRIPPING
TO THE HENRY'S LAW CONSTANT
Compound
Isopropyl alcohol
Chloroform
1,1, 1-Tri chl oroethane
Tetrachl oroethane
Acetone
Toluene
Xylene
H
(atm)
186
273
24
360
340
K
(Y/X)
2.4
2,0
3.4
1.6
1.90
2.8
3.6
Batch 1
14.7
11.4
7.0
6.5
Batch 3
10.5
5.82
10
Batch 4
6.21
25.4
5.97
7.8
TABLE 8. ANALYSIS OF VENT GAS FROM VACUUM PUMP
ON THIN FILM EVAPORATOR
Process vent
10 minutes
(mg/L)
Process vent
5 minutes
(mg/L)
Duplicate
process vent Field
5 minutes Blank 1,2
(mg/L) (5 min basis)
Chloroform 2.4
1,1,1-Trichloroethane 3.1
Trichloroethylene 0.87
Tetrachloroethylene 1.0
Toluene 27.1
Ethyl benzene 0.31
Xylenes 1.41
17.8
3.05
0.59
1.04
23.5
0.52
2.16
6.1
2.35
0.61
1.01
25.8
0.45
1.9
0.1
83
-------
o
•H
4J
•H
,5
I
CJ
.3
4J
n)
M
4J
d
0)
o
0
o
o
0.6
0.3
0.1
0.06
0.03
0.001
0.0006
Isopropylalcohol
Chloroform
1,1,1, Trichloroethane\
Tetrachloroethene
Xylene
20
100
Figure 1.
40 60 80
(Time (minutes)
Concentrations in the batch of waste
as a function of time: xylene MST.
o
o
•t
r-t
5
4J
•H
a)
1-1
j-i
g
o
o
Acetone
Chloroform
1,1,1 Trichloroethane
Ethyl Benzene•
0.06
0.03 .
0.01
Figure 2.
20 30 40
Time (minuses)
Concentrations in the batch of waste as
function of time: trichloroethane MST.
84
-------
A CASE STUDY OF DIRECT CONTROL OF EMISSIONS
FROM A SURFACE IMPOUNDMENT
R.G. Wetherold, B.M. Eklund and T.P. Nelson
Radian Corporation
P.O. Box 9948
Austin, TX 78766
ABSTRACT
Field measurements were taken to determine the effectiveness of a cover/enclosure in
controlling VOC emissions from an aerated wastewater treatment lagoon. The control
system included a carbon adsorption system for the air which was vented, under controlled
conditions, from the enclosed system.
INTRODUCTON
The Hazardous Waste Engineering
Research Laboratory (HWERL) of EPA has the
responsibility of providing technical sup-
port to OAQPS in the area of atmospheric
emissions from hazardous waste management.
In this study, testing was performed at a
specialty chemicals plant in the northeast.
The aerated wastewater lagoon at this faci-
lity has been enclosed with an inflated
flexible dome structure to reduce odors and
to keep the lagoon warm during winter. A
stream of air from the enclosure is vented
through a regenerable carbon adsorption
system. The effectiveness of the enclosure
and carbon adsorption system in containing
and controlling VOC emissions was evaluated
in this study. Additionally, the fate of
the VOC entering the aerated lagoon was
estimated.
PURPOSE
The primary objective of this study
was to determine, through field testing,
the efficiency of a cover and associated
carbon adsorption system in controlling the
emissions of VOC from an aerated wastewater
treatment lagoon.
APPROACH
Liquid and slurry samples were collec-
ted at various locations in and around the
wastewater treatment facility at the chemi-
cal plant. The objective of the sampling
was to gather enough data to perform a
material balance around the system and,
thus, determine the fate of the VOC enter-
ing the system.
Process Description--
A schematic flow diagram of the waste-
water treatment system at the plant is
shown in Figure 1. Specialty chemicals are
produced in a number of separate batch pro-
cessing areas within the plant. The waste-
water entering the wastewater treatment
system can come from a variety of sources,
including reaction products, process wash-
water, and washwater used in cleanup oper-
ations. Waste solids are also associated
with the wastewater.
Wastewater from the batch processing
is quite variable in both flow rates and
compositon. As shown in Figure 1, the
wastewater from the plant first flows into
two neutralizer tanks for pH adjustment.
85
-------
CENTRIFUGES
CLARIFIER
UNDERFLOW
• LIQUID SAMPLE LOCATION
A SOLIOIUQUIO SAMPLE LOCATION
• OAS SAMPLE LOCATION
TO MUNICIPAL
WASTEWATER SYSTEM
(NOT OPERATING)
Figure 1. Wastewater Treatment Facility Sample Locations.
Do
AIRLOCK
//////,
l[
|
f
1
/////
1
IZ
i
BLOWER MAIN BLOWER
WITH PROPANE
HEATER
NE CORNER
OF DOME
DOME
FOOTING
CARBON ADSORBER
FAN
(OPEN) (CLOSED)
OUTSIDE AIR
INTAKE DAMPERS
Figure 2. Dome Structure Fan Inflation and Exhaust System.
86
-------
From the neutralfzer tanks, the wastewater
(including the waste solids) is pumped to
the primary clarifier. The overflow from
the clarifier is sent to a trim basin for
additional pH adjustment and then to the
aerated lagoon.
About one-third of the primary clan'-,
fier underflow is recycled back to the
clarifier, and the remainder is directed to
a belt filter for solids dewatering. The
solids on the belt filter are washed with
additional process water before being,sent
to a solids drying area. The recovered
process and washwater are sent to the
aerated lagoon through a sump. Hydrobac®,'
a freeze-dried bacteria designed to effec-
tively attack aromatic/phenolic materials,
is added in this sump to promote bacterial
action in the lagoon.
From the aerated lagoon, the treated
wastewater and resulting solids are pumped
through a sump to two open-tank secondary
clarifiers. Floe and powdered activated
carbon are added in the sump.
The treated overflow from the secon-
dary clarifiers is currently discharged to
a nearby river. The major portion of the
underflow from one of the secondary clari-
fiers is recycled to the aerated lagoon.
The remainder of the underflow is sent to
the solids centrifuge for dewatering. The
recovered water is returned to the lagoon,
and the solids are wasted to a solids dry-
ing area.
The aerated lagoon is approximately
46 x 130 meters. Aeration is performed
with two 75-hp' aerators and 25 smaller 7.5
hp aerators. Residence time in the lagoon
is approximately 5 days at the normal depth
of 1.5 meters. However, the lagoon level
had been lowered to approximately 0.5-0.6
meters about two weeks before the testing.
They residence time during the testing was
approximately 1.5-2 days.
The entire aerated lagoon is covered
by an inflated dome made of a PVC-coated
polyester fabric. The structure is pres-
surized by a main blower with a capacity of
20,000 CFM. The integrity of the dome is
maintained by steel cables strung over the
fabric. The fabric is lined with Tedlar to
minimize chemical attack and to reduce the
permeability of the dome to the internal
gases.
A purge stream of air from inside the
enclosure is continuously vented to the
atmosphere through a fixed-bed carbon
adsorption system. A schematic diagram of
the dome inflation and purge system is
shown in Figure 2. The purge rate of ap-
proximately 1.4 m3/sec (3000 cfm) pro-
vides a gas-phase residence time of 9.4
hours under the dome. The carbon adsorp-
tion system is a two-bed unit. During nor-
mal operation, one bed is adsorbing while
the other is being regenerated with steam.
The adsorption cycle lasts 24 hours, while
the regeneration cycle consists of 12 hours
of steaming and 12 hours of cooling. The
recovered hydrocarbons and steam condensate
produced during the regeneration cycle are
returned to the aeration lagoon.
Liquid, slurry and gas samples were
collected at various locations-around the
wastewater treatment facility. The sam-
pling points are shown in Figure 1. The
inlets to the system were considered to be
the underflow and overflow from the primary
clarifier. The outlet streams were the
treated wastewater, the sludges (from the
filters and centrifuges) and the exhaust
from the carbon adsorption system. Key re-
cycle streams were also sampled. The two
secondary clarifiers were identical in size
and operation. Only the outlet stream from
one of the clarifiers was sampled.
The flow rates of the major process
liquid and slurry streams were metered by
the plant using flow meters as shown in
Figure 1. Flow rates of some internal
liquid streams were not measured, and these
rates were estimated by plant personnel.
The total gas flow rate at the outlet of
the carbon adsorption system was measured
by Radian twice each day. EPA Reference
Method 2 Procedures were followed in making
these measurements.
Sample Collection--
Liquid and sludge samples were col-
lected from 2 to 8 times per day during the
test period. These samples were collected
in glass containers with Teflon-lined caps
and stored at approximately 4°C until ana-
lyzed. Gas samples were continuously col-
lected from the inlet and outlet air
streams of the carbon adsorption system.
The sampling system consisted of a rake-
type in-stack probe, a heated pump and
87
-------
heat-traced lines leading to the analytical
instruments. The gas samples were analyzed
on-site using two Byron Model 401 THC ana-
lyzers and a HNU Systems, Inc., Model 301
Gas Chromatograph.
Gas samples were collected for de-
tailed off-site analyses to determine
Individual species. These samples were
collected in evacuated stainless steel
canisters 2-3 times per day from the inlet
and outlet of the carbon adsorption sys-
tem.
Sample Analysis—
The on-site analyses were limited to
gas-phase samples. The Byron 401 Analyzers
were used to monitor the total hydrocarbon
(THC) concentrations. Analyses were per-
formed on one-minute cycles. All data from
the Byron Analyzers were automatically col-
lected, stored and reduced using a micro-
computer.
The HNU Systems, Inc., Model 301 GC
was used to provide rudimentary on-site
speciation as well as total VOC analyses of
gas samples which were collected periodi-
cally from the carbon adsorption system air
streams.
The solid, liquid and air canister
samples were analyzed for Cg-Cio or-
ganic compounds at Radian's Austin labora-
tories. The air samples from the gas
canisters were passed through a Perma-Pure
drying tube and then through a trap cooled
with liquid oxygen. A known amount of
sample was then desorbed from the trap onto
the analytical GC columns.
All analyses were performed on a
Varian 3700 GC equipped with fused silica
capillary columns. A flame ionization de-
tector (FID) was used to detect and quan-
titate hydrocarbon species, while a photo-
ionization detector (PID) was used to pro-
vide additional qualitative information. In
addition, a second sample was resolved on a
separate column and analyzed with a Hall
Electrolytic Conductivity Detector (HECD)
operated in the halogen mode. The FID/PID
output from the GC was processed with a
Varian 401 Chromatographic Data System
(CDS) and an Apple 11+ microcomputer.
The liquid samples were analyzed with
a purge-and-trap technique. A stream of
ultra-high purity nitrogen was passed
through aliquots of the liquid samples.
The resultant gas was passed through and
trapped in the cryogenic traps of the
analytical system. The analytical proce-
dure followed after the cryogenic trapping
was identical to that used in the analyses
of air samples.
Solid samples (sludge and activated
carbon) were extracted with tetraglyme,
purged in 5 ml of deionized water to a
cryogenic trap and then analyzed in the
same manner as the gas samples.
RESULTS
Sampling and Analytical Results--
The results of the analyses of selec-
ted liquid/solid samples are summarized in
Tables 1 and 2. The concentrations of com-
pound classes and selected individual com-
ponents in the primary clarifier overflow
(aeration pond inlet) and the secondary
clarifier feed (aeration pond outlet) are
shown. Samples from several different days
were analyzed.
Tables 3 and 4 contain summaries of
the analyses of the inlet and outlet air
streams of the carbon adsorption system.
The concentrations of major compound clas-
ses and a few selected organic compounds
are presented.
Emission Rates and Control Effectiveness—
A mass balance about the system was
performed using average flow rates and
compositions of the measured streams. The
results of the mass balance for total non-
methane hydrocarbons (NMHC) are summarized
in Table 5. The major emission point was
the exhaust of the carbon adsorption sys-
tem, accounting for 59% of the incoming
NMHC. Fifteen percent of the incoming NMHC
left the aeration pond in the effluent
water stream. An additional 7 percent of
the NMHC present in the feed to the aera-
tion pond was carried from the system in
the centrifuge sludge.
Biological activity in the aerated
p'ond was assumed to be responsible for the
difference between the measured inlet'and
outlet NMHC rates. The obvious assumption
in this determination is that all signifi-
cant sources of VOC emissions were
88
-------
c
o_
E
o
o
Li_
c
CM i—
1
cn
11 c
c o
ID J3
J= £-
"S S
CU tf)
4_j U
ra -i-
01 en
o c.
r— O
i— r-. CM
CM f— r-
i— CM in
in CM co co
CO <— OJ •— •—
CO CO CO
en cn cn
cn C7> cn
in in LO
cn cn cn
CO CO CO
cn cn cn
to
1
CD
O
E£
CD
o:
0
c
LiJ
H-
0
UJ
_J
UJ
to
U_
CD
to
z:
o
h-
ct
h~
UJ
O
O
o
UJ
to
>
*o
c
o
u
OJ
to
—
^
OJ
0
!
11
1 CO
o c
t_ CU
O N
f— C
J= OJ
CU
c
CU
f3-
0
1—
CU
M
OJ
CO
1
0
O OJ
i — c:
O -E
Q CU
CU
c cu
CU "O
-C*0
eu -£=
y (_)
cn
c
•r— QJ
O. (O
ro
to
1 1 I C
iii*
0
cc
1 t | CM
1 1 1 *
0
f 1 f CO
1 I 1 *
1 t | C
t I 1 •
0
1 1 1 0
1 ' ' d
CO CO «3- "JD
p- in «a- •—
in CM CM
co vo in •—
CO «5t" CO <*>
P-. CO
co CD cn in
P-. O CD CD
cn CD o CD
«a-
-------
TABLE 3. CARBON ADSORPTION SYSTEM: INLET AND OUTLET GAS
CONCENTRATIONS OF MAJOR.CLASSES OF ORGANIC COMPOUNDS
Gas Phase Concentration, ppmv
Sampling Date
8/17/84 8/177848/18/84 8/19/84
Carbon Adsorber Inlet
Paraffins
Aromatics
Halogenated Organics
Nonmethane Hyrocarbons
Carbon Adsorber Outlet
Paraffins
Aromatics
Halogenated Organics
Nonmethane Hydrocarbons
63.2
33.1
251
348
49.8
32.2
251
334
10.4
11.2
200
222
13.0
38.6
264
317
153
117
331
607
167
89.8
409
698
__
__
~
63.4
43.5
46.1
158
TABLE 4. CARBON ADSORPTION SYSTEM: INLET AND OUTLET GAS
CONCENTRATIONS OF SELECTED ORGANIC COMPOUNDS
Gas Phase Concentration, ppmv
Sampling Date
8/17/84 8/17/84 8/18/84 8/19/84
Carbon Adsorber Inlet
Methyl ene chloride
Dichloroethane
Benzene
Toluene
Chlorobenzene
Dichlorobenzene
Carbon Adsorber Outlet
Methyl ene chloride
Dichloroethane
Benzene
To! uene
Chlorobenzene
Dichlorobenzene
4.0
204
26.0
5.7
13.2
0.6
4.2
205
27.8
7.5
13.3
0.8
5.1
172
4.5
5.1
3.6
0.5
5.1
231
15.1
19.6
6.6
0.2
4.3
240
21.2
92.1
0.4
1.2
2.1
355
24.8
54.1
8.8
0.1
__
—
__
__
__
--
0.1
15.3
9.5
24.0
26.3
2.7
90
-------
TABLE 5. SUMMARY OF MASS BALANCE RESULTS
(Based on Averages of the,Data)
Source or Effluent
Influent
Primary Clarifier Feed
Effluent
*By difference.
NMHC Flow
(Kg/min)
0.212
'% of
Influent
100
Dome exhaust
Effluent wastewater
Effluent centrifuge sludge
Biological oxidation*
0.125
0.032
0.014
0.041
59
15
7
19
TABLE 6. ULTIMATE FATE OF ORGANIC GROUPS
Paraffins Aromatics Halogen Total
Percent of Total Influent
Primary Clarifier Feed
Percent of Total Effluent
6.9
55.4
*By difference.
37.7
100
Dome Exhaust
Effluent Wastewater
Effluent Sludge
Biological Oxidation*
11.6
8.2
4.1
-(17.0)
8.2
4.4
1.9
40.9
39.2
2.4
1.0
-(4.9)
59
15
7
19
Too
91
-------
included in the mass balance. As estimated
by difference, biological activity in the
pond appears to have accounted for approx-
imately 19 percent of the incoming NMHC.
The large majority of the compounds
detected in the streams around the waste-
water treatment system fall into the paraf-
fin, aromatic or halogenated hydrocarbon
classification. Table 6 contains a summary
of the fates of these three classes of com-
pounds. The estimated loss by biological
oxidation was estimated by difference.
Almost all of the halogenated com-
pounds appear to be lost in the carbon
adsorber exhaust. The small amount (4.9%)
indicated by difference as actually being
produced is probably due to measurement and
analytical uncertainties. On the other
hand, more than 70 percent of the incoming
aromatic compounds are apparently destroyed
through biological activity in the pond.
The influent paraffinic compounds seem
to be relatively unaffected by the waste-
water treatment. Indeed, there appears to
be an increase in the amount of paraffins
through the system. The relatively large
amount produced (17.0%) appears to be sig-
nificant, even when normal experimental and
analytical uncertainties are considered.
Some paraffin compounds may actually have
been produced as a result of biological
degradation of aromatic and halogenated
compounds.
The effectiveness of the pond cover in
containing VOC emissions could not be quan-
titatively assessed in this study. How-
ever, plant personnel indicated that the
dome had a good seal around the base, and
they estimated that the total leakage was
about 0.14 m3/sec (300 cfm). A crude leak
check of the dome by Radian personnel found
only very small leaks around the dome base.
Permeation of gas through the dome fabric
is negligible according to information sup-
plied by the manufacturer. Thus, it ap-
pears that VOC containment by the cover is
virtually complete.
The results of the gas phase analyses
around the carbon adsorption system show
that this device is not removing any sig-
nificant amount of VOC in the dome exhaust.
In some cases, higher levels of VOC were
observed in the outlet air than were de-
tected in the inlet. These differences are
presumed to be due to the variability of
the inlet concentrations and to the opera-
tion of the carbon adsorption system. This
result is not unexpected for several rea-
sons. The carbon adsorption system was not
designed for bulk removal of organic com-
pounds from the air stream. It was in-
stalled for odor control and specifically
for the removal of orthochlorophenol. The
system appears to be effective in this
application.
The air stream entering the carbon
adsorption system is saturated with water
at the temperature under the dome, which is
generally warmer than the ambient condi-
tions. Very high humidities drastically
reduce the adsorptive capabilities of the
activated carbon. There is probably actual
condensation of water within the beds.
Activated carbon is generally effective
only on gas streams having relative humi-
dities of 50 percent or less.
The results of this study support the
following conclusions:
- The combination of the pond cover
and carbon adsorption system is
not an effective method for con-
trolling VOC emissions because of
the poor efficiency of the carbon
adsorbers, as operated at this
site: ,
- The carbon adsorption system is
almost completely ineffective in
removing significant quantities
of VOC from the dome exhaust
stream;
- Leakage around and/or through
the cover over the aeration pond
appears to be quite small:
- The aromatic compounds are effec-
tively destroyed through biological
activity in the aerated pond;
- The halogenated compounds are
vaporized and lost in the purge
stream through the carbon adsorp-
tion system; and
- The paraffinic compounds are vir-
tually unaffected in wastewater
treatment system and leave the
system in the effluent wastewater.
92
-------
PRODUCTS OF INCOMPLETE COMBUSTION -
ANALYTICAL METHODS
M.M. Thomason, R.H. James, R.E. Adams
Southern Research Institute
P.O. Box 55305
Birmingham, Alabama 35255-5305
L.D. Johnson
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
Products of incomplete combustion (PICs) may be formed during the incineration of
hazardous wastes. The monitoring of principal organic hazardous constituents (POHCs)
may be confused by the formation of PICs which are also POHCs. The total amount of
hazardous PICs formed can easily exceed the total amount of the indicator POHCs detected
in the exhaust gases of an incinerator. The compounds may be fragments of incineration
feed constituents, products of complex recombinations, products of substitution reac-
tions in the flame or post-flame zone, or compounds that enter the incineration process
through other sources.
We have concentrated our efforts on applying previously developed general analysis
methods for POHCs to the analysis of PICs. A large part of the current literature on
PICs has resulted from finding polychlorinated dibenzo-jr-dioxins (PCDDs) and^polychlor
inated dibenzofurans (PCDFs) in incinerator effluents. Thus, we have specifically
evaluated generalized GC/FID and GC/MS methods for PCDDs and PCDFs. Substituted poly-
nuclear aromatic hydrocarbons (PAHs) are also frequently formed as PICs. We have also
evaluated generalized analysis methods for a number of amino-, nitro-, alkyl , oxy , and
halogenated PAHs. The compounds chosen are not yet included in Appendix VIII of 40 CFR,
Part 261.
The results presented here give representative chromatograms, mass spectral condi-
tions, and detection limits for these groups of compounds. Future work includes the
identification of PICs in incinerator effluent samples with the extension of the general
analysis procedures to high-priority compounds.
93
-------
INTRODUCTION
During the incineration of hazardous
wastes, compounds not identified in the
waste feed may be formed. These compounds
are known as products of incomplete com-
bustion (PICs). The aim of incineration
is to destroy all of the principal organic
hazardous constituents (POHCs) in the
waste and to make sure that harmful levels
of PICs, which may be formed during com-
bustion, do not escape into the environ-
ment. The total amount of hazardous PICs
formed can easily exceed the total amount
of the indicator POHCs detected in the
exhaust gases of an incinerator.
Trenholm (1) has defined PICs as any
hazardous organic constituent detected in
the stack gas of an incinerator but not
present in the waste feed at a concentra-
tion of 100 pg/g or higher. PICs may be
fragments of incineration feed constitu-
ents, products of complex recombinations,
products of substitution reactions in the
flame or post-flame zone, or compounds
that enter the incineration process
through other sources.
The combustion of chlorinated hydro-
carbons has been shown to form PICs (2).
In chlorinated hydrocarbon combustion,
pyrolysis of the parent molecules forms
dechlorinated intermediates because of the
weak C-C1 bond. The dechlorinated inter-
mediates may then undergo a variety of
reactions: recombination to form other
chlorinated hydrocarbons, further pyroly-
sis, oxidation, or polymerization. As the
chlorine content of the hydrocarbons
increases, the formation of soot is
favored because of the weak C-C1 bond
strength. Soot formation is undesirable
because it is often related to the produc-
tion of toxic compounds such as polycyclic
aromatic hydrocarbons (PAHs). The combus-
tion of chlorinated hydrocarbons has also
been reported to produce polychlorinated
dibenzo-j>-dioxins (PCDDs) and polychlor-
inated dibenzofurans (PCDFs) C3-,4). PCDDs
and PCDFs have also been found in inciner-
ator effluents and fly-ash samples by
several research groups (5_-_7) •
Tiernan and co-workers (8) sampled
and analyzed the effluents from an incin-
erator used to heat a steam boiler
designed for energy recovery from the
combustion of municipal refuse. Chloro-
phenols, chlorobenzenes, PCDDs, PCDFs, and
traces of PCBs were found in the incinera-
tor effluent. Karasek and co-workers (_9)
recently reported a detailed analysis of
fly-ash extracts of municipal incinerators
in Canada and Norway. They identified
more than 200 organic components in the
extracts. Compounds found included hydro-
carbons, phthalate esters, PAHs, PCDDs,
PCDFs, and other polychlorinated organic
compounds.
The formation of PICs in an incinera-
tor environment is probably a very complex
phenomenon. The two classes of PICs which
are under very close scrutiny at the
present time are PCDDs and PCDFs. PCDDs
and PCDFs are apparently formed from pre-
cursors such as PCBs, polychlorinated
naphthalenes, polychlorobenzenes, and
other chlorinated organic compounds found
in the hazardous-waste load. Incinerator
conditions also play a major role in the
extent of PIC formation.
PURPOSE
We have concentrated our efforts on
applying previously developed general
analysis methods for POHCs to the analysis
of PICs. A large part of the current
literature on PICs has resulted from
finding PCDDs and PCDFs in municipal
incinerator effluents. Thus we have
specifically evaluated generalized GC/FID
and GC/MS methods for 10 PCDD isomers and
3 PCDF isomers which are frequently formed
as PICs. We have also evaluated the
generalized analysis methods for a number
of amino-, nitro-, alkyl-, oxy-, and
halogenated PAHs. The compounds chosen
are not yet included in Appendix VIII of
40 CFR, Part 261.
EXPERIMENTAL APPROACH
• Preparation of Stock Standard
Solutions of PICs
Stock standard solutions of the can-
didate compounds were prepared in appro-
priate solvents at a concentration of
about 0.05 mg/mL. These stock solutions
were serially diluted as required to pre-
pare working standards. Methylene chlor-
ide, toluene, and isooctane were used as
solvents.
94
-------
• Description of General Operating
Conditions
We developed the GC/MS generalized
test method on a Hewlett-Packard Model
5985 GC/MS data system. The supplemental
GC/FID work was performed on a Hewlett-
Packard Model 5840 GC that was equipped
for use with capillary columns.
The generalized GC/MS, GC/FID proce-
dure is:
Instruments: HP 5985A for GC/MS and
HP 5840 for GC/FID
Columns: Two SE-54 bonded-fused-silica
capillary columns, 0.32-mm ID,
25-m length
Carrier gas: Helium, 2 mL/rain
Injection: 2 ^L splitless
Column temperature program: 60 °C for
3 min, then programmed at
20 °C/min to 300 °C and held
at 300 "C for 15 min
Injection temperature: 290 °C
FID temperature: 300 °C
MS parameters: Scan 41 to 500 amu/sec,
electron ionization at 70 eV,
source temperature set to
200 °C, capillary column
plumbed directly into the ion
source.
The initial operating conditions chosen
were a synthesis of conditions that would
allow the separation and detection of as
many compounds as possible. Modifications
of temperatures and rates may be necessary
or desirable to optimize the analysis for
specific compounds.
PCDDs and PCDFs were also analyzed
using selected-ion monitoring (SIM) GC/MS
to improve the detection limits obtain-
able. Table 1 summarizes the ions moni-
tored in the SIM studies. A 70 msec/ion
dwell time was used.
• Optimization of the GC/FID and
GC/MS Procedures
GC operating conditions were opti-
mized by analyzing solutions containing
selected PICs by the GC/FID technique.
The column-head pressure was adjusted
appropriately to allow adequate separation
of the compounds of interest in less than
30 min.
Having established GC operating con-
ditions by the GC/FID procedure, we then
applied the method to the determination of
the candidate PICs by GC/MS. The mass
spectrometer was operated in a full mass
scanning range (41 to 500 amu) in the El
mode. The scan time was maintained at
<1 sec to enable the collection of enough
scans to characterize each capillary GC
peak. During SIM, the mass spectrometer
dwelled on each selected mass for
70 msec/scan.
• Quality Control Procedures
We calibrated the GC/FID and the
GC/MS procedures with standard solutions
of the candidate PICs using anthracene-d1Q
as the internal standard. Four- to six-
point calibration curves were prepared for
each candidate PIC determined by gas
chromatography. Each curve was a plot of
the FID or the MS response (relative to
anthracene-d1Q) as a function of the quan-
tity of the particular PIC injected on the
GC column. Appendix A of reference 10 was
used as a guideline for estimating detec-
tion limits for each candidate POHC.
The precision of determinations by
GC/FID and GC/MS was assessed by tripli-
cate injections of at least one standard
solution of each PIC investigated by GC.
Relative response factors (RRF) were
calculated for each PIC relative to
anthracene-d....
Recovery data were not applicable for
this phase of the project because our
analyses were done with standard solutions
of PICs. We did not collect, prepare, or
analyze field samples that required
spiking of the matrix with surrogate
standards of specific PICs.
RESULTS AND DISCUSSION
The generalized GC/FID and GC/MS
methods for POHCs (11) were adapted for
the analysis of 36 candidate PICs.
Table 2 presents a summary of the GC/FID
determinations of candidate PICs. The
relative retention time (RRT) and on-
column detection limit are given for each
compound. The compounds are listed,in the
order of their elution from the GC column.
95
-------
TABLE 1. IONS USED IN SELECTED-ION MONITORING OF SELECTED PICs
Compounds/Classes
PCDD Ions
Internal standard
ions
PCDF Ions
Internal standard
ion
Mono-substituted
PCDDs and PCDFs
Di-substituted
PCDDs and PCDFs
Tri-substituted
PCDDs and PCDFs
Tetra-substituted
PCDDs and PCDFs
Hexa-substituted
PCDDs and PCDFs
Octa-substituted
PCDDs and PCDFs
218
252
286
320
388
458
220
254
288
322
390
460
202
236
270
304
372
442
204
238
272
306
374
444
332
332
332
332
332
332
334
334
334
334
334
334
10
188
188
188
188
188
188
Retention times are relative to that
observed for the internal standard,
anthracene-djg. The on-column detection
limit was estimated using experimentally
determined calibration curves as suggested
in Appendix A of Reference 10. Typical
values were 2 to 40 ng.
The precision of determinations was
assessed by triplicate injections of at
least one standard solution of each PIC
investigated by GC. RRFs were calculated
for each PIC relative to anthracene-d .
These results are also summarized in
Table 2. The calculated values of the
standard deviations (SDs) and relative
standard deviations (RSDs) in Table 2
indicate that most GC/FID determinations
gave acceptable precision. The percent
RSDs range from 0.2 to 9.3%.
Table 3 summarizes the GC/MS deter-
mination of the candidate PICs. The five
most abundant mass fragments of each com-
pound, along with their relative
abundance, were used for the establishment
of detection limits and for the generation
of calibration curves. RRTs and RRFs were
determined for each compound. The RRTs
for GC/FID and GC/MS were similar. The
precision of the GC/MS determinations was
also evaluated by triplicate injections.
In general, the precision of the GC/MS
determinations was poorer than the GC/FID
determinations. The RSDs range from 3 to
fc //» •
Figure 1 is a representative total-
ion chromatogram (TIC) obtained for
several dioxin isomers. The candidate
dioxin and furan isomers studied were also
evaluated using SIM. The estimated limits
of detection were 40 to 200 times lower
using SIM instead of full mass
acquisition.
a'a '
* ^ s 4 10 n iai'3 '4 ts ;
RETENTION TIM!, mm
Figure 1. TIC of selected dioxin
isomers .
96
-------
TABLE 2. SUMMARY OF GC/FID DETERMINATIONS OF CANDIDATE PICs
Relative
Precision of
RRFb
- reten- On-column Relative
tion detection Standard standard
Compound timea limit, ng Mean deviation deviation, %
1-Fluoronaphthalene
2-Bromonaphthalene
3-Br omoquino 1 ine
l-Bromo-2-methylnaphthalene
1-Iodonaphthalene
l-Bromo-4-methylnaphthalene
2- ( Bromomethyl) naphthalene
l-Amino-4-chlorona*phthalene
2-Chlorodibenzo-p_-dioxin
l-Chlorodibenzo-p_-dioxin
9-Bromofluorene
5-Bromoacenaphthene
l-Amino-4-bromonaphthalene
l-Chloro-8-nitronaphthalene
2-Chloroanthracene
3,6-Dichlorodibenzofuran
2-Bromof luorene
2,7-Dichlorodibenzo-£-dioxin
2 , 3-Dichlorodibenzo-_p_-dioxin
1-Chloroanthracene
9-Chloroanthracene
9-(Chloromethyl) anthracene
1 ,2 ,4-Trichlorodibenzo-_p_-dioxin
9-Bromophenanthrene
9-Bromoanthracene
9 , 10-Dichloroanthracene
2,3,7 ,8-Tetrachlorodibenzofuran
1,2,3 ,4-Tetrachlorodibenzo-p_-dioxin
6 ,9-Dichloro-2-methoxyacridine
9 , 10-Dibromoanthracene
3 ,6-Dibromocarbazole
1 ,2 ,3 ,4 ,7-Pentachlorodibenzo-j>-dioxin
1,2,3,4,7, 8-Hexachlorodibenzo-p_-dioxin
1,2,3,4,6,7,8-Heptachlorodibenzo-
f-dioxin
,3,4,6,7,8,9-Octachlorodibenzo-
p-dioxin
l,T,3,4,6,7,8,9-Octachlorodibenzofuran
0.619
0.820
0.825
0.896
0.899
0.907
0.929
0.972
0.980
0.983
0.991
1.01
1.02
1.03
1.05
1.06
1.06
1.08
1.08
1.11
1.11
1.11
1.16
1.16
1.17
1.20
1.23
1.28
1.28
1.30
1.37
1.40
1.57
1.83
2.15
2.17
4
5
2
3
2
3
9
40
10
2
10
5
4
4
7
3
10
2
10
4
4
4
5
4
5
4
2
2
5
9
8
10
8
40
20
20,
0.928
0.733
0.543
0.635
0.540
0.664
0.431
0.028
0.659
0.474
0.379
0.554
0.548
0.560
0.356
0.626
0.595
0.388
0.553
0.743
0.678
0.945
0.352
0.528
0.529
0.638
0.356
0.277
0.318
0.318
0.178
0.379
0.190
0.118
0.090
0.104
0.012
0.008
0.0071
0.0050
0.0022
0.0070
0.0095
0.00012
0.0014
0.0013
0.0050
0.0010
0.0055
0.0058
0.0015
0.0044
0.0039
0.0020
0.0023
0.0017
0.0083
0.0034
0.0043
0.0011
0.0053
0.0017
0.0005
0.0077
0.0012
0.0009
0.0017
0.0094
0.0011
0.0023
0.0077
0.0073
1.3
1 . 1
1.3
0.79
0 .42
1.1
2 .2
0.42
0.22
0.28
1.3
0.2
0.99
1.0
01 r\
.42
0.71
0.65
0.52
0.42
2.3
1.2
0.36
1.2
2.0
In
.0
2.6
1.4
2.8
3.8
2.8
9.3
2.5
6.0
2.0
8.5
7.0
aRelative to the retention time of anthracene-d1Q.
bRelative response factor (RRF) = (AsCis/AisCs).
97
-------
TABLE 3. SUMMARY OF GC/MS DETERMINATIONS OF CANDIDATE PICs
Compound
1-Fluoronaph thalene
2-Bromonaphthalene
3-Bromoquino line
l-Brorao-2-methylnaphthalene
1— lodonaph thalene
l-Bromo-4-methylnaph thalene
2- (Broraorae thyl ) naph thalene
l-Chlarodibenzo-£-dioxin
2-Chlorodibenzo-j>-dioxin
l-Amino-4-chloronaphthalene
5-Bromoacenaphthene
9-Bromofluorene
l-Chloro-8-nitronaphthalene
l-Amino-4-bromonaphthalene
3,6-Dichlorodibenzofuran
2-Bromofluorene
2 , 7-Dichlorodi.benzo-jv-dioxin
2 ,3-Dichlorodibenzo-jr-dioxin
1-Chloroanthracene
2-Chloroanthracene
9-Chloroandhracene
9-Bromophenanthrene
9-Bromoanthracene
1 ,2 ,4-Trichlorodibenzo-jgr-dioxin
9- ( Chlorome thy 1 ) anthracene
9 , 10-Dichloroanthracene
2,3, 7 ,8-Tetrachlorodibenzof uran
1 ,2 ,3 ,4-Tetrachlorodibenzo-
_p_-dioxin
6 , 9~Dichloro-2-methoxyacr idine
9 , 10-Dibromoanthracene
1,2,3,4, 7-Pentachlorodibenzo-
j>-dioxin
3 ,6-Dibromocarbazole
1,2, 3,4,7, 8-Hexachlorodibenzo-
_p_-dioxin
1,2,3,4,6,7,8-Heptachlorodi-
benzo-jr-dioxin
1,2,3,4,6,7,8,9-Octachlorodi-
benzo-_p_-dioxin
1,2,3,4,6,7,8,9-Octachlorodi-
benzofuran
Mol
wt
146
206
207
220
254
220
220
218
218
177
232
244
207
221
236
244
252
252
212
212
212
256
256
286
226
246
304
320
277
334
"354
323
388
422
456
440
Masses of characteristic El ions On-column
(relative abundance) detection
1
146(100)
208(100)
207(100)
141(100)
127(100)
141(100)
141(100)
218(100)
218(100)
177(100)
153(100)
165(100)
172(100)
223(100)
236(100)
165(100)
252(100)
252(100)
212(100)
212(100)
212(100)
256(100)
256(100)
288(100)
191(100)
246(100)
306(100)
322(100)
277(100)
176(100)
356(100)
325(100)
390(100)
424(100)
460(100)
442(100)
2
73(10)
206(90)
209(94)
220(89)
254(98)
220(97)
115(17)
220(34)
220(34)
179(32)
152(83)
166(15)
126(72)
221(94)
238(70)
246(63)
254(66)
254(63)
214(32)
214(37)
214(33)
258(99)
258(98)
286(89)
189(28)
248(73)
304(77)
320(78)
234(78)
88(84)
354(71)
323(54)
392(86)
426(99)
458(100)
444(98)
3
147(10)
127(55)
128(89)
222(86)
126(19)
222(93)
139(14)
219(14)
155(30)
115(28)
76(67)
163(14)
207(40)
115(51)
173(20)
244(59)
189(28)
126(22)
176(26)
176(29)
176(30)
176(57)
88(61)
290(34)
226(21)
176(33)
308(45)
324(49)
279(69)
336(52)
358(63)
327)52)
388(50)
428(50)
142(77)
446(72)
4
145(8)
126(14)
101(33)
139(34)
255(12)
139(34)
70(13)
155(12)
127(23)
178(13)
232(39)
82(13)
161(37)
70(24)
237(15)
163(24)
126(25)
189(22)
88(20)
177(18)
88(18)
88(48)
176(54)
223(23)
192(18)
247(18)
307(11)
323(12)
236(51)
87(37)
293(26)
246(24)
394(36)
422(41)
462(74)
440(41)
5 limit, ng
125(8)
207(12)
75(22)
115(32)
128(10)
115(26)
142(12)
221(5)
109(18)
149(11)
234(38)
164(10)
149(25)
114(16)
240(12)
166(18)
253(13)
253(14)
213(15)
213(17)
213(15)
177(34)
177(29)
225(20)
94(17)
87(15)
310(10)
257(11)
164(23)
334(26)
291(22)
244(24)
327(30)
361(28)
167(57)
154(40)
9
5
4
50
5
5
4
9
10
9
10
10
9
40
3
5
9
3
4
7
8
8
10
10
7
4
9
10
10
4
10
4
8
10
40
30
98
-------
ACKNOWLEDGMENT S
The authors are especially grateful
for the individual efforts of the profes-
sional staff of Southern Research Insti-
tute who participated in the performance
of this work.
DISCLAIMER
The research described in this
article has been funded by the U.S.
Environmental Protection Agency through
Contract No. 68-02-3696 to Southern
Research Institute. It has been subjected
to the Agency's required peer and policy
review. Approval does not signify that
the contents necessarily reflect the views
or policies of the agency nor does mention
of trade names or commercial products
constitute endorsement or recommendation
for use.
REFERENCES
1. Trenholm, A.; Hathaway, R.;
Oberackers D. Products of Incomplete
Combustion From Hazardous Waste
Incinerators; Incineration and treat-
ment of hazardous waste—Proceedings
of the tenth annual research
symposium. NTIS PB 85-116291,
EPA-600/9-84-022. September 1984.
2. Senkan, S.M. Combustion character-
istics of chlorinated hydrocarbons.
In: Exner, J.H., ed. Detoxification
of hazardous waste. Ann Arbor: Ann
Arbor Sci.; 61-92: 1982.
3. Bumb, R.R.; Crummett, W.B.; Cutie,
S.; Gledhill, J.R.; Hummel, R.H.;
Kagel, R.O.; Lamparski, L.L.; Luoma,
E.V.; Miller, D.L.; Nestrick, T.J.;
Shadoff, L.A.; Stehl, R.H.; Woods,
J.S. Trace chemistries of fire: A
source of chlorinated dioxins.
Science 210: 385-390: 1980.
4. Study on state-of-the-art of dioxin
from combustion sources. New York,
American Society of Mechanical
Engineers; 78: 1981.
5. Olie, K.; Vermeulen, P.L.; Hutzingerj
H. Chlorodibenzo-p_-dioxins and
dibenzofurans as trace components of
fly ash. Chemosphere 6: 455; 1977.
6. Buser, H.R.; Bosshardt, H.P.; Rappe,
C. Identification of polychlorinated
dibenzo-jj-dioxin isomers found in fly
ash. Chemosphere 7: 165; 1978.
7. Eiceman, G.A.; Clement, R.E.;
Karasek, F.W. Analysis of fly ash
from municipal incinerators for trace
organic compounds. Anal. Chem. 52:
1942; 1980.
8. Tiernan, T.O.; Taylor, M.L. ; Garrett,
J.H.; Van Ness, G.F.; Solch, J.G.;
Deis, D.A.; Wagel, D.J. Chlorodi-
benzodioxins, chlorodibenzofurans and
related compounds in the effluents
from combustion processes. Chemos-
phere 12: 595-606; 1983.
9. Tong, H.Y.; Sohre, D.L.; Karasek,
F.W.; Helland, P.; Jellum, E.
Identification of organic compounds
obtained from incineration of munici-
pal waste by high-performance liquid
chromatographic fractionation and gas
chromatography/mass spectrometry. J.
Chromatogr. 285: 423-441; 1984.
10. Longbottom, J.E.; Lichtenberg, J.J.,
eds. Methods for organic chemical'
analysis of municipal and industrial
waste water, EPA-600/4-82-057. NTIS
PB 83-201-798. July 1982.
11. James, R.H.; Adams, R.E.; Finkel,
J.M.; Miller, H.C.; Johnson, L.D.
Evaluation of analytical methods for
the determination of POHC in combus-
tion products. Proceedings of the
77th annual meeting of the Air Pollu-
tion Control Association. San
Francisco; June 1984.
99
-------
EPA RESEARCH TO RECOVER TOXIC HEAVY METALS FROM WASTE STREAMS
S. Garry Howell
Hazardous Waste Engineering Research Laboratory
U. S. Environmental Protection Agency
Cincinnati , Ohio 45268
ABSTRACT
The Environmental Protection Agency (EPA) has sponsored several research programs
with the goal of reducing heavy metal pollution of land, surface and groundwater. Past
efforts were aimed at reducing discharges of heavy metals from electroplating operations,
and investigating the possibilities of establishing centralized treatment and recovery
systems for large metropolitan areas such as Cleveland, Los Angeles, and Milwaukee. EPA
studies indicate that such facilities would be feasible, and that a large part, if not
ail, of their cost of operation could be met by sale of the metals recovered.
INTRODUCTION
Heavy metal wastes such as those pro-
duced by metal plating operations, have
traditionally been discarded rather than
treated for reclamation of metal values.
Sludges and mill scales were usually land-
filled, and plating wastewaters often dis-
charged to the sewer. Various regulations
have greatly restricted these practices,
and the recent escalation of both metal and
energy costs have made metal recovery not
only a desirable environmental goal but the
time is rapidly approaching when the value
of recovered metals will cover the cost of
pollution abatement, and in some instances
yield a modest profit. The Environmental
Protection Agency has sponsored research
into metals recovery from several sectors
of the metals (i.e., mill scale, grinding
swarf) and metal finishing industry (plat-
ing wastes). This paper will cover only
plating and the chemically closely related
electrochemical machining wastes.
History and Background
There are an estimated 20,000 metal
finishing shops in the U.S., about 80% of
which plate copper, nickel, chromium, and
zinc. In years past spent plating wastes
were hauled to a landfill, or even diluted
.and poured into a sewer. Toxic effects of
heavy metals on marine life, and pollution
of drinking water has led to a number of
laws and regulations restricting landfill-
ing of sludges and liquids containing heavy
metals. In addition to the ultimate cost
to society in health effects, the economic
loss suffered by discarding potentially
valuable minerals is tremendous. As
several of these minerals are mostly im-
ported (nickel and chromium, and- to a
lesser degree copper) they are also of
strategic value. Table 1, from a report by
the Comptroller General to Congress, shows
an estimate of metal values lost in 1980.
Using the same amounts of metal discarded,
and adjusting for 1985 prices, the total
value is almost 58 million dollars.
It is obviously impractical to attempt
to recover all the heavy metals shown in
Table 1 (iron and aluminum are not econom-
ically recoverable in small quantities).
The report from which this table was taken
suggests that it would be technically feas-
ible to recover 80 to 90 percent of the
copper, 30 to 40 percent of the zinc, 90 to
95 percent of the nickel and 70 to 75
percent of the chromium. Economic feasi-
bility is the prime obstacle; quite likely
only 40 to 50 percent recovery could be
obtained even under the best conditions.
100
-------
TABLE 1. ANNUAL LOSS OF MINERALS IN ELECTROPLATING WASTES
Mineral
Amount
Value/Ton
Total Value
Iron
Copper
Zi nc
Nickel
Aluminum
Chromium
Lead
Total
(metric tons)
3,460
2,560
3,718
5,000
1,651
6,357
756
23,502
$ 77
791
748
4,000
1,060
2,000
462
$ 266,420
2,024,960
2,781,064
20,000,000
1,750,060
12,714,000
349,272
$39,885,776
RESULTS AND CONCLUSIONS
Operation of a pilot scale metal re-
covery unit has demonstrated that metal
recovery from electroplating and electro-
chemical machining wastes is technically
feasible, and in most cases, economical.
If transport costs from generating sites
to the recovery facilities are very low, as
could be the case in some metropolitan
areas, the value of metals recovered will
often lower operating expenses, and in many
cases yield a profit.
Profitability of a metals recovery
operation is primarily dependent on the
following factors:
• Metal type and concentration in the
sludge; for instance, high concen-
trations of high value metals such
as nickel and chromium can be sold,
yielding enough profit to support
the rest of a recovery operation.
f Availability of large quantities of
high metal content wastes within a
reasonably short range. Hauling
distances that averge over 25 miles
can greatly increase recovery costs
to the point where recovery is no
longer economically feasible.
• An economical and reasonably flex-
ible recovery process; some varia-
bility in composition is to be
expected in metal wastes fed to the
operation. This cannot always be
offset by blending to achieve homo-
geneity of feed material.
• The actual viability of a metal
recovery operation in a specific
area can only be determined by
.careful weighing of all the above
factors plus the availability of
capital and special tax considera
tions often given to resource re-
covery facilities.
• The probable minimum economical
size of a control recovery operation
is 25-50 tons of 25% solids sludge
per day.
• The most practical and economical
overall process appears to be a
series of unit operations tying
together more or less conventional
hydrometallurgical processes; pyro-
metallurgical approaches do not
appear to be economical for small
volume recovery operations.
• While sulfide precipitation yields
effluent water with a very low metal
content, the difficulty of and
energy costs for separating and re-
cycling sulfides makes it a poor
second choice to hydroxide pre-
cipitation for metal -recovery. In
addition, EDTA must be added to
101
-------
selectively precipitate metal
sulfides, and organic coagulants
must be used to produce filterable
precipitates.
RESEARCH PROJECTS
The Environmental Protection Agency
has sponsored research into metal removal
and/or recovery from wastewaters and
sludges for over ten years. Early efforts
concentrated on water purification (10) but
the wisdom of reclaiming metal was soon
recognized as a means of reducing water
pollutfon conserving metal/resources. Work
done for EPA at Battelle Columbus in 1975
Included a survey of plating facilities and
the volume and types of metals discarded in
sludges. As understanding of the problem
began to develop, a series of investiga-
tions were funded, each adding to the
overall bank of information. These pro-
jects are listed in reverse chronology,
below.
Metal Value Recovery from Metal Hydroxide
Sludges (F)
The latest EPA projects were under-
taken at the Montana College of Mineral
Science and Technology (Montana Tech), and
tied together much of the previous work.
The first phase of the investigation was a
carefully executed series of laboratory ex-
periments to establish the processes needed
to leach hydroxide sludges (the most
plentiful type produced by electroplating
(EP Shops) with sulfuric acid, and the
various recovery methods
required for copper, zinc, chromium, and
nickel. Flowsheets were constructed and
the unit operations required to make a
complete recovery process were developed.
In Phase Two, a pilot assembly capable
of processing 75-100 Ibs. of'sludge per day
was built and operated at a large electro-
plating facility. While not entirely
troublefree, the operation was smooth
enough to allow confirmation of the possi-
ble commercial viability of the process,
and development of a computerized mass
balance.
The steps of the process were as
fol1ows:
• The sludge is leached with sulfuric
acid.
• Iron, which is frequently present in
EP sludges, was removed by precipi-
tating as a potassium jarosite
[KFe3(S04)2(OH)6]. This must be
done first, as iron interferes in
many of the subsequent recovery
steps. This step is in need of im-
provement, because in the process of
precipitating the jarosite some
anions such as chromate are removed,
along with some heavy metal cations.
Even though the volume of the
jarosite sludge is much less than
that of the starting material, the
waste would be classed as hazardous,
requiring special precautions and
extra costs when landfill ing.
Because of these faults, it is pro-
posed that a new program be insti-
tuted to attempt to remove iron
without the inclusion of large
amounts of other heavy metals.
• Conventional techniques were used to
extract copper from the acid solu-
tion using an oxime in petroleum
solvent. The copper was then strip-
ped from the organic phase with
stronger acid and electrowon or
crystallized as sulfate.
• Zinc and cadmium were extracted with
diethyl hexyl phosphoric acid
(D2EHPA), leaving chromium and
nickel behind. Again, a sulfuric
acid strip was used for zinc removal
from the organic phase. Small
amounts of residual iron are removed
in early stages of extraction; ten
extraction stages are required to
remove iron-free zinc.
• Chromium was oxidized to Cr+f> with
chlorine, then selectively precipi-
tated as lead chromate which can be
sold as a pigment or regenerated to
chromic acid with sulfuric acid.
The lead sulfate precipitate formed
when chromic acid is regenerated may
itself be recycled to precipitate
more lead chromate. In some in-
stances S02/02 oxidation of the
chromium-may be more desirable.
• Nickel is precipitated as sulfide,
the only heavy metal in the process
which is not directly recyclable
back to a plating operation. How-
ever, several alternative techniques
102
-------
are presented to yield a nickel
ammine, or nickel sulfate may be
solvent extracted with a mixture of
chelating agents.-
Recognizing that some improvements
could be made in the various separation
steps, first order cost estimates indicate
that a 50 ton per day recovery plant could
not only reduce landfill volume consider-
ably, with consequent less chance of pollu-
tion, but in addition could yield a profit
as shown in Table 2 below.
Metal Recovery from Electrochemical Machin-
ing and Electrodischarge Machining (ECM/EDM)
Hastes (2)
These relatively new machining pro-
cesses produce heavy metal wastes; ECM
yields hydroxide sludges heavily contami-
nated with sodium chloride, and EDM finely
divided particles of the alloy being
machined. Both ECM and EDM are primrily
used on hard alloys, often high in nickel,
chromium, cobalt, molybdenum, etc. This
project involved a survey of ECM/EDM shops,
the volumes of sludge produced, metal con-
tent, and disposal practices and costs.
Several recovery routes were studied,
including the procedures outlined in
Reference (1) above.
Cadmium Recovery from Electroplating
Wastes (3)
This project was primarily concerned
with methods and costs for recovering cad-
mium from plating shops which are able to
segregate Cd from other sludges, or perhaps
those whose only business is cadmium plat-
ing. The authors give very little thought
to the fact that many platers use cadmium
at least occasionally, and do not segregate
their sludges. The recovery technique
(melt and/or distill cadmium electro-
motively displaced with zinc metal) is most
effective on low chromium and nickel
sludges. While chromium is not displaced
by zinc, nickel would have to'be removed by
distilling the zinc.
Coupled Transport Systems for Control of
Heavy Metal Pollutants (4)
Coupled transport membranes are made
by immersing a microporous membrane into a
liquid complexing agent such as those used
TABLE 2. PROCESS COST: FIRST ORDER ESTIMATE (2)
UNIT OPERATION
COST ($)*
Factored CapitalAnnualizedOperation CostTotal Cost
1. Leach, jarosite
precipitati.on
2. Jarosite storage
3. Copper solvent extrac-
tion, electro-winning
4. Zinc, residual iron
solvent extraction, zinc
sulfate crystallization
5. Chromium oxid., chromic
acid production
6. Nickel recovery
TOTAL COST
Profit (Gross)
Cost Estimate Capital Cost
430,800
390,500
336,100
661,600
1,818,200
231,600
3,868,800
119,500
108,200
93,100
183,300
503,600
64,200
1,071,900
Per Year Per Year
223,500 343,000
25,400 133,600
205,900 299,000
269,700 453,000
407,700 911,300
230,000 294,200
1,362,200 2,434,100
1,071,900
*Assuming a $1.00 per gallon credit for sludge from producers.
103
-------
for liquid/liquid extraction of metals from
solution. The liquid (often called a liquid
1on exchanger) then acts as a transfer
agent to move metal ions through the capil-
laries, using the driving force of the pH
gradient across a membrane separating an
acidic from a basic solution. Data are
presented which suggest that a membrane
unit having 450 to 500 square feet of sur-
face could separate chromium from the rinse
water of a medium sized plating line. The
separation unit would cost about $5,000,
compared to over $50,000 for an evaporative-
ion exchange system as noted in Reference
(5) below.
Evaporative Recovery of Chromium Plating
Rinse Maters (5)"
A demonstration of this process was
given by Advance Plating Company,
Cleveland, Ohio on the first of four
counterflow rinse tanks. After first
passing through a cation exchanger to
remove metallic impurities, water from the
first rinse tank is drawn into a vacuum
evaporator, whose condensate is returned
to the third rinse tank. Concentrated acid
produced in the evaporator bottoms is re-
turned to the plating bath. The value of
recovered chrome has enabled the operator
to pay off the equipment in two years or
less, in addition, the chromic acid in
dragout water is reduced by 99.98%.
Removal of Toxic Metals from Metal Finishing
Wastewater by Solvent Extraction (6)
This short investigation of solvent
extraction was primarily limited to
one extraction agent, Alamine 336 (General
Hills). By increasing acidity stepwise
with HC1, selective removal of chromium,
cadmium, and zinc from wastewaters was
demonstrated. Copper and nickel could not
be extracted with this reagent; several
other high molecular weight amines were
tried without success.
Reverse Osmosis Field Test: Treatment of
Matts Nickel Rinse Waters (7)
Rinse waters from a Watts type (pri-
marily nickel sulfate) nickel plating oper-
ation were passed through a polyamide
membrane, effectively separating a concen-
trated nickel salt solution and recycling
the rinse water. Results were very encour-
aging, indicating that the capital cost of
the R.O. unit could be recovered in about
a year, based on the reduction in water
usage plus metal values returned to the
plating bath. This process could be more
accurately described as recycling or con-
servation since there is no chemical change
of metals recycled to the bath.
Treatment of Metal Finishing Wastes by
Sulfide Precipitation (8,9/10.11)
Although the process of sulfide pre-
cipitation has been known for many years,
its use as a means of pollution control
appears to be of fairly recent origin, and,
is mentioned here primarily to illustrate
an effective method of metal removal. A
major disadvantage of sulfide precipitation
is the lack of selectivity; it is apparent-
ly impossible to precipitate a single metal
without major contamination by other metal
sulfides. Use of a polymeric flocculant
(preferably anionic) increased the settl-
ing rate, and adding a chelating agent such
as EDTA (ethylenediamine tetraacetic) acid
at the proper pH greatly aided selectivity.
Heavy metal sulfides, even if effect-
ively separated, are extremely difficult to
reduce to their corresponding metals by
hydrometallurgical methods, and are
generally recovered by smelting. The large
volumes of raw material required to support
such smelters makes plating waste metal
recovery by this method impractical.
A Reclamation of Metal Values from Metal
Finishing Waste Treatment Sludges (12)
A survey of plating opertions involv-
ing over 600 questionnaires (150 replies)
revealed the extent of sludges generated by
the industry, and even developed a "typical"
analysis of metal content; the copper,
nickel, and chromium contents were very
similar to those found in subsequent
studies. Some preliminary laboratory
leaching and extraction studies led to the
conclusion that metal recovery could be
economically feasible only if the process
could be carried out on a large, and pre-
ferably continuous scale. Based on then
current (1971) metal prices, it was con-
cluded that chromium and zinc recovery were
uneconomic; the writers apparently did not
foresee the ban on landfill ing of heavy
metal wastes now being implemented.' (10)
The only laboratory work done by these
investigators were leaching experiments
with ammonium carbonate and sulfuric acid
which indicated that acid leaches were
104
-------
adequate for most hydroxide sludges while
ammonium carbonate would only dissolve
copper and nickel. A literature study of
several potential recovery methods for
copper and nickel was made, but no work
was done to verify their efficacy; a'gain,
chromium recovery was not considered.
REFERENCES
1. Twidwell, L. G. Metal Value Recovery
from Metal Hydroxide Sludges, report
in preparation.
2. Hoi combe, L. J., et al. Metal Re-
covery from Electrochemical Machining
and Electrical Discharge Machining
Hastes, report in preparation.
3. Lloyd, T. B. and K. J. Wise. Cadmiurn
Recovery from Electroplating Hastes,
unpublished EPA report.
4. Babcock, H. C., et al. Coupled
Transport Systems for Removal of Heavy
Metal Pollutants, EPA 600/2-79-181.
5. Evaporative Recovery of Chromium Plat-
ing Rinse Haters, prepared by Advance
Plating Co. for EPA 600/2-78-127.
6. McDonald, C. H. Removal of Toxic
Metals from Metal Finishing Hastewater
by Solvent Extraction, EPA 600/2-78-
011.
7. McNutty, K. J., et al. Reverse
Osmosis Field Test: Treatment of
Hatts Nickel Rinse Haters, EPA 600/2-
77-039.
8. Schlauch, R. M., and A. C. Epstein.
Treatment of Metal Finishing Wastes by
Sulfide Precipitation, EPA 600/2-77-
Industry: Sulfide Precipitation,
625/8-80-003.
EPA
049.
Control and Treatment
the Mptal Finishinn
10. Robinson, A. K. and J. C. Sum.
Sulfide Precipitation of Heavy Metals,
EPA 600/ 2-80-139.
11. Bhattacharyya, D. and L. F. Chen.
Sulfide Precipitation of Nickel and
Other Heavy Metals, 3 Volume Un-
published EPA Report.
Tripler, A. B., et al. Reclamation of
Metal Values from Metal-Finishing Waste
Treatment Sludges, EPA 670/2-75-018.
12.
Wilson, D. L. Control ai...
Technology for the Metal Finishing
Additional References
Environmental Pollution Control Alterna-
tives: Economics of Wastewater Treatment
Alternatives for the Electroplating In-
dustry, June 1979, EPA 625/5-79-016.
Environmental Pollution Control Alterna-
tives: Centralized Waste Treatment
Alternatives for the Electroplating
Industry, June 1981, GPO# 757-064/0322.
Summary Report - Control and Treatment
Technology for the Metal Finishing
Industry - In-Plant Changes, January 1982,
GPO# 560-565(R).
Summary Report - Control and Treatment
Technology for the Metal Finishing
Industry - Ion Exchange, June 1981, GPO#
757-064/0321.
Summary Report - Control and Treatment
Technology for the Metal Finishing
Industry - Sulfide Precipitation, April
1981, EPA 625/8-80-003.
Environmental Regulations and Technology -
The Electroplating Industry, August 1980,
GPO# 660-868 8/80.
105
-------
CHEMICAL DESTRUCTION/DETOXIFICATION OF CHLORINATED DIOXINS IN SOILS
Robert L. Peterson
Edwina Milicic
Gal son Research Corporation
East Syracuse, New York 13057
Charles J. Rogers,-
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
ABSTRACT
Laboratory experiments using 1,2,3,4 tetrachlorodibenzo-p-dioxin show that
chlorinated dioxins in soil may be chemically reduced to levels below one part per
billion. The decontamination processes involve the addition to the soil of a mixture
of alkali metal hydroxide, polyethylene glycols or capped polyethylene glycols,
dimethyl sulfoxide and water. The reagent is either added directly to the soil with
mixing (in-situ process) or mixed 1:1 by volume with soil in an external reactor
(slurry process). The dioxin is dechlorinated to a water soluble form, which may be
then removed from the soil (slurry process) or allowed to biodegrade (in-situ process),
INTRODUCTION
The contamination of large areas of
soil with dioxin have created a need for a
cleanup method which is capable of
handling large volumes of contaminated
soil in a cost effective manner. Propos-
ed treatment methods include incinera-
tion, solvent extraction and direct
chemical dechlorination. This paper will
discuss the successful use of direct
chemical dechlorination in decontamina-
tion of dioxin contaminated soils on a
laboratory scale and the implications of
these data for large scale soil treatment.
The basis for use of chemical
dechlorination as a decontamination
method lies in the relationship between
the toxicity of chlorinated dioxins and
the number of chlorine atoms on the
dioxin molecule. In order for a dioxin
isomer to exhibit high toxicity a mini-
mum of three chlorine atoms are requir-
ed, and these must be in the 2,3, and 7
positions (1). In addition, the lipophilic
nature of many chlorinated dioxins
contributes to their bioaccumulation
activity. The dechlorination processes
used in this study reduce the number of
chlorine atoms on the dioxin molecule
and produce a hydrophilic material which
is more easily removed from the soil
than the parent dioxin.
PURPOSE
The purpose of this project is to
identify and evaluate at the laboratory
level an effective chemical process for
destruction/detoxification of chlorin-
ated dioxins in soil.
APPROACH
Direct chemical decontamination of
soil can be considered to be a two step
process;
1. Application of the reagent to the
dioxin contaminated soil
2. Reaction of the dioxin and reagent
These two aspects of the process
interact. For example, a reagent with
high mobility in soil requires a less,
rigorous application method than a
reagent which is highly viscous.
106
-------
However at a minimum the reagent must
be capable of reducing dioxin
concentrations to <1 ppb, or favorable
mobility and ease of application become
irrelevant. For this reason, the reaction
system was selected first, with the
method of application designed around
the reagent system.
Application methods for direct
decontamination of soil must be
compatible with the reaction system
selected. In order for a direct chemical
reaction process to be effective, the
contaminated soil and reagent must be
brought into close contact. The degree
of contact between soil and reagent may
be increased by mechanical mixing, by
addition of diluents and co-solvents to
increase the reagent mobility, and by
heating the soil/reagent mixture to
reduce the viscosity of the reagent.
After selection of the reaction system, a
variety of application methods were
tested under laboratory conditions for
use in dioxin soil decontamination.
In order to reduce the costs of
handling and analyzing soil samples, a
low toxicity dioxin isomer, 1,2,3,4
tetrachloro-p-dibenzo dioxin (TCDD), was
used in place of 2,3,7,8 TCDD for all
testing. Laboratory tests of 2,3,7,8
TCDD and 1,2,3,4 TCDD indicated that
rates of reaction for the two isomers
were sufficiently similar to allow
generalization of results between
isomers.
To insure sample uniformity, all
samples for treatment used uncontamin-
ated soil which was spiked with a known
amount of dioxin. Each soil sample was
spiked individually, with the entire
sample used for analysis. Uncontam-
inated soil samples were obtained from
the vicinity of a dioxin site in
Mississippi and spiked with 1,2,3,4 TCDD
before processing. Soil from an authen-
tic test site was used to avoid the wide
variations seen in previous tests of
treatability of different PCB contam-
inated soils. It was anticipated that
dioxin contaminated soils would show
similar variations.
PROBLEMS ENCOUNTERED
Major problems encountered during
this project involved the analytical
procedure. Analysis of soil samples by
gas chromatography was frustrated by
the large number of interferences found
in the test site soils. The interference
problem was exacerbated by the ability
of the reagents used to extract materials
from the soil which were not extracted
by the analytical solvents used. While
this/high extraction efficiency for the
treatment solvents contributed to the
success of the dioxin destruction, it
complicated the analysis. The failure of
cleanup procedures to produce an
acceptable sample necessitated a move
to gas chromatography/mass
spectroscopy as .the primary method of
analysis.
An additional problem was caused by
the tendency of the test site soil to
solidify if exposed to high temperatures
(100°C or above) under alkaline condi-
tions. This problem was solved by
moving to lower temperatures, which
proved adequate for purposes of
treatment.
RESULTS AND DISCUSSION
Selection of Reagent System - Of the
available reactions, including ultraviolet
dechlorination, ruthenium tetroxide
oxidation and nucleophilic substitution,
nucleophilic substitution using alkali
metal hydroxides has given the best
results in both literature studies and
laboratory trials. The preferred
reactions of this type for use with
halogenated aromatics involve combination
of potassium hydroxide and polyethylene
glycols to form an alkoxide, which is
the reactive species. The addition
of a sulf oxide catalyst/co-solvent,
usually dimethyl sulfoxide, greatly
enhances the rate and degree of
reaction, although it is not known if
this is due to the effectiveness of the
sulfoxide as an extractant for
aromatics or to the catalytic effect
of the sulfoxide for substitution
reactions. The probable mechanism for
this class of reactions follows;
RON + KOH --------- > ROK + HOH
ArCl
Sulfoxide
ROK — ----
ArCln_1OR
KCL
107
-------
The partially dechlorinated, water soluble
reaction product may continue to undergo
dechlorination, depending on the reaction
conditions. Several reactions of this
type have demonstrated reduction of
dioxin concentrations in liquids to <1 ppb.
Selection of Application Method - A wide
variety of hydroxide/alcohol/sulfoxide
reaction systems will effectively
dechlorinate dioxins in liquid solution.
However, application of the reagent to
the soil in such a manner as to allow
these reactions to occur is a significant
problem. The necessary degree of
contact has been achieved using two
different approaches; direct addition of
reagent or a reagent/water mixture to
the soil with mixing in place (in-situ
process) and excavation of the soil
followed by mixing equal volumes of soil
and reagent in an external reactor (slurry
process).
Both the slurry and in-situ process-
es may be used at elevated temperatures.
Heating methods for an in-situ system
would use radio frequency (RF) or
microwave heating(2). The simplest
method of heating for the slurry process
1s to preheat the reagent prior to mixing
of the soil and reagent. Both of these
heating methods are applicable to large
scale processing.
The two application methods have
different areas of application. The
in-situ process is limited to areas of
shallow contamination and the soil and
reagent may be mixed with conventional
agricultural equipment. In addition, the
degree of contamination must be
relatively uniform. If small areas of
high concentration or "hot spots", are
present, the high local concentration of
contaminant might exhaust the amount of
reagent which may be economically
applied. While the concentrations of
dioxin generally observed in the field are
not high enough to cause a problem,
dioxins are often found in combination
with other chlorinated wastes, which
may be present in high concentrations.
However, in suitable cases, large areas
of shallow soil contamination could be
treated in a fairly short period of time
using the in-situ process. In cases
where dioxin has penetrated to depths >
1-2 feet, or where significant areas of
high concentrations occur, the slurry
process is more suitable. The slurry
process, which uses large volumes of
reagent followed by reagent recovery, is
more suited to treatment of "hot spots"
than is the in-situ process.
Results of Combined Application/
Reaction Process - The slurry and in-situ
process results will be discussed
separately.
Results of In-situ Processing - Two
different reagents have been used for
in-situ processing in this study;
potassium hydroxide/polyethylene glycol
400/ dimethyl sulfoxide (KOH/PEG/DMSO)
and potassium hydroxide/2-(2-methoxy
ethoxy ethanol)/dimethyl sulfoxide
(KOH/MEE/DMSO), Water has been added
in some cases as a co-solvent in an
attempt to give readier penetration of
small amounts of reagent into the soil.
Each set of samples had an associated
set of spikes and blanks. The spikes used
reagent without the addition of KOH and
were handled in the same manner as the
samples. The blanks were untreated
soil. Samples treated at above ambient
temperatures were maintained uncovered
in a water bath unless otherwise noted.
The results of the in-situ processing are
summarized in table 1.
Discussion of In-situ Processing
Spike Recovery - The recovery of TCDD
from spikes in the in-situ process was
extremely variable, ranging from 120% to
<4%. In general, the higher the
temperature and longer the hold time, the
lower the spike recovery. Sealing the top
of the spike reduced the loss of TCDD for
the 2 day KOH/MEE/DMSO run to less
than that for the 1 day run, indicating
that the lower spike recoveries are
probably due to losses from
volatilization and/or extraction into the
spike solvent (PEG/DMSO or MEE/DMSO),
which is not analyzed. In the slurry
tests, where volatilization is not a
factor, analysis of the spike reagent
showed that some 77% of the original
TCDD was present in the spike reagent'.
The highest spike recovery was from the
20 C run, which is consistent with the
volatilization hypothesis. Extraction of
the TCDD into a polar solvent would tend
108
-------
Table 1 - Summary of Results of In-situ Processing - All soils initially at 2000 ppb
1:1:1 KOH/PEG/DMSO
1:1:1 KOH/PEG/DMSO
1:1:1 KOH/PEG/DMSO
2:2:2:1 KOH/MEE/DMSO/WATER
2:2:2:1 KOH/MEE/DMSO/WATER
2:2:2:1 KOH/MEE/DMSO/WATER
2:2:2:1 KOH/MEE/DMSO/WATER
wt%
in soil
20
20
20
20
20
20
20
2:2:2:6 KOH/MEE/DMSO/WATER 20
2:2:2:30 KOH/MEE/DMSO/WATER 50
2:2:2:30 KOH/MEE/DMSO/WATER 20
BLANKS - ALL
* SPIKES SEALED
temp,
°C
20
70
70
70
70
70
70
70
70
70
time,
days
7
7
1
1
2
4
7
7
7
7
avg
spl
980
<1
5.3
3.3
2.8
2.1
1.2
2.1
18
50
ppb
TCDD
spike
2500
740
730
500
870*
210
190
140
170
70
to enhance volatilization, similar to the
effects of water on the volatilization of
PCBs in soil (3). It is interesting to
note that the worst spike recoveries were
in the MEE reagents which used water as a
co-solvent.
Effects of Temperature - Only two reaction
temperatures have been tested to date using
the in-situ process. The improvement in
reaction efficiency in going from 20 C to
70 C was dramatic, improving reaction ef-
ficiency from 50% to > 90%.
Effects of Reagent Formulation - PEG vs.
MEE - The test results for the 70 C reac-
tions were slightly better at 1 day for the
MEE reagent and slightly better at 7 days
for the PEG reagent. This difference does
not appear to be significant.
Effects of Water as a Co-Solvent - Four
sets of tests were run for 7 days at 70 C
using KOH/MEE/DMSO with water as a co-
solvent. These data show a good correla-
tion between wt% reagent and TCDD concen-
tration after 7 days, as shown in figure 1.
The plot of % active reagent vs. ppb TCDD
shows the expected first order relationship.
This demonstrates that dilution of the rea-
gent with water to provide more contact,
followed by evaporation of the water to en-
courage reaction, was not effective in
reducing the amount of reagent required.
The plot of % reagent vs. ppb TCDD
shows the expected first order relation-
ship. This demonstrates that dilution
of the reagent with water to provide more
contact, followed by evaporation of the
water to encourage reaction, was not
effective in reducing the amount of
reagent required.
Results of Slurry Processing - The slurry
process conditions tested and the results
of GC/MS analysis are summarized in table
2.
As little as 2 hours at 70°C were ade-
quate to reduce TCDD levels from 2000 ppb
to < 1 ppb, for a removal efficiency of
> 99.95%. The bulk of this removal
occurred in the first 30 minutes, when
< 99% of the TCDD had been reacted. The
reaction at 25 C was slower, but did remove
98% of the original dioxin after two hours.
Reagent Recovery Efficiency - Reagent re-
covery of the PEG reagents by distillation
was only partially successful (about 50%
recovery), due to the poor heat stability
and low vapor pressure of the PEG. Rea-
gent recovery by washing has been more
successful, with 94-99+% recovery of rea-
gent. The degree of recovery is important
109
-------
100
PPB
TCDD
10
4 6 8 10 12 14
WP/oREAGENTT
Figure 1 - wt% active reagent vs. ppb TCDD after 7 days
16
18
Table 2 - Results of Slurry Processing
Reagent
1:1:1 KOH/PEG/DMSO
1:1:1 KOH/PEG/DMSO
1:1:1 KOH/MEE/DMSO
1:1:1 KOH/MEE/DMSO
1:1:1 KOH/MEE/DMSO
1:1:1 KOH/MEE/DMSO
Temp, °C
180-260
180
150
70
70
25
Rxn time, hrs
4
2
2
2
0.5
2
ppb TCDD
< 1
< 1
< 1
< 1
15
36
Blanks - all < 1 ppb TCDD
Spikes - % recovery in soil - 0.1-5.9
- % recovery in decanted solvent - 77
to the overall economics of-the process.
Dioxin Recovery Efficiency - Analysis of
the PEG/DMSO reagent used in the slurry
process spikes gave a high recovery of
dioxin (77%). As in the case of the in-
sltu samples, slurry spikes were treated
with an alcohol/sulfoxide mixture without
the addition of alkali. Additional
analysis of both the solvents and the soil
(soil extracted with acetone/hexane after
decantation of the alcohol/sulfoxide) from
the slurry spikes showed the presence of a
large variety of halogenated materials
which were not originally added to the soil
or removed from the soil using the analyti-
cal reagents. The amount of additional
chlorinated material removed from the
untreated soil by the PEG/DMSO was on the
order of 140 ppm, calculated as lindane.
These materials may be pesticide residues
or naturally occurring chlorinated species.
Analysis of treated slurry samples also
shows these materials, although at much
lower levels, indicating that these are
some form of aromatic halide. Results
of analyses for DDT, DDE and chlorinated
dioxin were negative.
These data indicate that conven-
tional methods of extraction, may under-
estimate concentrations of halogenated
110
-------
organics in soil due to an inability to
remove halogenated species from the humics
present in the soil, while reagents
containing PEG/DMSO are capable of remov-
ing these halogenated organices from
treated soi1.
PRELIMINARY ECONOMIC EVALUATION
In order to provide a rough estimate
of relative costs, two scenarios were
constructed. In the.first case, a 1 acre
site 3 feet deep was to be treated using
the in-situ process with radio frequency
heating. Capital costs for this option
are estimated at $2,970,000 for a capacity
of 27,600 tons/year of soil (4). In the
second case, -soil was to be excavated and
placed in a 3 reactor slurry process system.
Capital costs for this option are
$2,350,000 for a 40,000 ton/year capacity.
Capital recovery costs for both processes
'assumed an interest rate of 14% over 5
years. Cost estimates for the two cases
are shown in table 3.
Table 3 - Preliminary Economic Analysis of
In-situ and Slurry Processes
Cost, $/ton soil
Cost item in-situ slurry
Capital recovery 31
setup and operatin 65
reagent 200
Total costs 296
17
54
20
91
The major difference in costs between the
two processes is in the cost of reagent.
In the in-situ process, where reagent is
not recovered, this cost is 67% of the
total cost. For the slurry process, the
operating costs assume a site that is
reasonably easy to excavate. For cases
where excavation is required to levels
below the water table or in very rocky
soil, this cost could increase greatly,
although this would also be the case for
landfill or incineration. In cases where
excavation is difficult, the overall costs
for the in-situ process may be lower than
for the slurry process.
CONCLUSIONS
1. Dioxin concentrations in soil can
be reduced from 2000 ppb to <1 ppb by
mixing the soil with a combination of
alkali metal hydroxide, alcohol and
sulfoxide.
2. Mixing of the soil and reagent
can be done effectively in two different
ways; direct addition of reagent to in-
place soil with one time mixing (in-situ
process) or combination of soil and
reagent in a reaction vessel with
continuous mixing (slurry process).
3. Estimated costs for processing
are in the range of $100-$300/ton soil.
ACKNOWLEDGEMENTS
This work was supported by the
United States Environmental Protection
Agency under the direction of Mr. C.
Rogers and the United States Air Force
Headquarters Engineering and Service
Center with the assistance of Lt. E. Heyse
under EPA contract 68-03-3219.
REFERENCES
1. Esposito et al, "Dioxins" EPA-600/
2-80-197, p 187
2. Dev et al, "Decontamination of
Hazardous Waste Substances from Spills
and Uncontrolled Waste Sites by Radio
Frequency In-situ Heating", 1984
Hazardous Material Spills Conference
3. Mackay and Wolkoff "Rate of
Evaporation of Low Solubility
Contaminants from Water Bodies to
Atmosphere" Environ. Sci. Tech. 7 (7),
611-614
4. Dev, H., personal communication
4/3/85
111
-------
GENE ENGINEERING OP YEASTS FOR THE BIODEGRADATION OP HAZARDOUS WASTES
John C. Lopera'b, Chien Chena and Chitta R. Deya
Department of Microbiology and Molecular Genetics
bDepartment of Environmental Health, College of Medicine
University of Cincinnati
Cincinnati, OH 45267-0524
ABSTRACT
Yeasts are eukaryotic microorganisms whose cytochrome P-.450 monooxygenase systems
may be amenable to genetic engineering for the hydroxylation and detoxication of
polychlorinated aromatic hydrocarbons. We are examining the molecular genetic
properties of strains of baker's yeast, Saccharomyces cerevisiae, and an n-alkane
utilizing yeast, Candida tropicalis ATCC750. Standard methods were used to purify
cytochrome P-450 and NADPH-cytochrome c (P-450) reductase proteins from cells cultured
by serai-anaerobic glucose fermentation (S_. cerevisiae, C. tropicalis) and by growth on
tetradecane (C. tropicalis). Polyvalent antisera prepared in rabbits to some of these
proteins were used in tests of immunological relatedness among the purified proteins
using sodium dodecyl sulfate-polyacrylamide gel electrophoresis and nitrocellulose
filter immunoblots. The results provide evidence for gene relationships which should
prove useful in gene isolation and subsequent engineering of P-450 enzyme systems in
yeast.
INTRODUCTION
Chemicals entering the environment
can be classified into three categories
with respect to biotransformation. Those
subject to complete biodegradation or
mineralization are consumed as organic
nutrients and energy sources by the
bacteria, yeasts, molds and other
microorganisms in the biosphere; the
great majority of natural and man-made
chemicals are in this category. Certain
other compounds undergo limited
biotransformation reactions without being
degraded for appreciable utilization.
Typically these cases involve co-
oxidation reactions in which the compound
is modified by enzymes normally active
with other substrates. Given the
diversity of microorganisms in the
environment, it is not surprising that
some of these altered compounds may then
be metabolized further by additional
members of the raicrobial mixtures. In
this manner a subset of compounds in the
category of co-oxidation occasionally
also may become completely mineralized.
The remaining category, relatively
small in number, is comprised of
environmentally stable compounds.
Persistence of. these compounds is
ascribed to a variety of reasons. Some
may be inherently toxic for the available
microorganisms. For some the genes for
the necessary enzymes might not exist
among the members of the available
microbial population. For microbes
capable of only limited co-oxidative
transformation, failure to derive useful
energy or building blocks from the
compound offers no selective advantage
and such organisms may not compete.
Alternatively the necessary enzymes may
not be induced by the amounts of the
compound which enter the cell.
PURPOSE AND APPROACH
Unfortunately among these
environmentally stable compounds are
several known to be hazardous to human
112
-------
The isolation or development of
microbes capable of degrading such stable
organic wastes could yield a safe and
cost-effective treatment of industrial
waste streams and toxic environmental
sites. Development of such
microorganisms is the purpose of this
research.
We anticipate that several of the
barriers to biodegradation can be
overcome in yeasts using recombinant DNA
technology (9). The compounds of
interest are all polychlorinated aromatic
hydrocarbons having no unsubstituted
adjacent carbons, e.g. hexachlorobenzene/
and highly chlorinated biphenyls,
dibenzodioxins and dibenzofurans.
Because of their structure, the initial
step in any microbial transformation of
these compounds is likely to require
hydroxylation by a monooxygenase or a
dechlorination step, or both. Fungi and
yeasts are eukaryotic microorganisms
which are known to catalyze a broad range
of monooxygenase reactions using
cytochrome P-450 enzymes, (15) and
mammalian cytochrome P-450 monooxygenase
systems have been implicated in the
hydroxylation/dechlorination of such
compounds as 2,3,7,8-tetrachlorodibenzo-
p-dioxin (3,12,14), 2,3,7,8-
tetrachlorodibenzofuran (13), and
hexachlorobenzene (19).
Our strategy in using yeasts is
based upon the common features of
cytochrome P-450 systems among higher and
lower eukaryotes (17). These are
particulate electron transport systems
whose chief components are membrane
lipid, NADPH-cytochrome c (P-450)
reductase (reductase) and one of several
cytochromes P-450. According to the
substrate specificity of each cytochrome
P-450, the overall reaction results in a
mixed distribution of the atoms of 02 to
yield a monooxygenated product and water.
By our strategy, we first would isolate a
structural cytochrome P-450 gene and its
essential regulatory sequences from the
yeast Saccharomyces cerevisiae.
Molecular genetic techniques are
particularly well developed for this
eukaryotic microorganism. Next,
exogenous cytochrome P-450 genes which
encode desired biotransformations would
be introduced into this yeast under the
control of appropriate regulatory
sequences. There is evidence that
functional cytochrome P-450 systems can
be assembled from yeast and mammalian
protein components in vitro (2) and in
vivo (H. Ohkawa, personal
communication). Thus we expect that the
P-450 genes could come from any
appropriate source: mammalian, yeast or
fungal. Further laboratory experiments
would be conducted to express such a
system in a yeast variant which could
persist in a contaminated medium. A
yeast such as a Candida sp. which
utilizes n-alkanes in petroleum may be
useful for this purpose. The intended
product organisms would be able to take
up the hazardous organic compound and
detoxicate it through co-oxidation, even
though the compound were present in low
concentrations in the environmental site-.
Our implementation of this strategy
has included the isolation of major
cytochrome P-450 system proteins from S_.
cerevisiae and Candida tropicalis and the
production in rabbits of immune sera to
these components. This paper presents
our observations of immunological cross-
reactivity among these protein isolates.
MATERIALS AND METHODS
Chemicals
Procedures of Guengerich (4) were
followed for the recrystallization of
cholic acid for use as sodium cholate,
and for the synthesis of n-octylamino
sepharose 4B. Cytochrome c was horse
heart type III obtained from Sigma
Chemical Co., St. Louis, MO. Protein
molecular weight standards were the low
molecular weight standard kit of Bio-Rad
Laboratories, Richmond, CA, supplemented
by preparations obtained from Sigma of
both bovine liver catalase and bovine
liver glutamic dehydrogenase, used as
molecular weight markers of 58 kd and 53
kd respectively (20). Nitrocellulose
filter paper was supplied by Schleicher
and Schuell, Keene, NH; hydroxylapatite
was Hypatite C of Clarkson Chemical Co.,
Williamsport, PA. Tetradecane was
obtained from Fisher Chemical Co., .
Cincinnati; all other chemicals were
acquired as reagent grade from commercial
suppliers.
Cell Culture and Protein Purifications
Sources and relevant properties of
113
-------
the strains of S. cerevisiae and C.
tropicalis used in this study are
presented in Table 1. All cultures of £3.
cerevisiae were carried out at 30°C and
C. tropicalis was grown at 24°C. Except
where tetradecane was used, all strains
were grown semi-anaerobically for glucose
fermentation (glucose grown cells) under
conditions described by Yoshida et al.
(22) using one of two media. Medium for
strains JL20 and JL21 consisted of 2%
casamino acids, 0.67% yeast nitrogen base
without added araino acids, 6% glucose and
0.004% each of L-histidine HC1 and
adenine HC1. The second medium, used for
all other strains, consisted of 1%
bactopeptone, 1.7% yeast extract, 0.09%
NaCl and 6% glucose, pH 6.7. All glucose
grown cells were harvested at raid-log
phase.
Growth of C. tropicalis on
tetradecane was based upon the media and
procedures described by Duppel et al.
(2). A New Brunswick fermentor FM-75 was
used to maintain 40 liter cultures under
aeration of 3 ft3 rain"1 with mechanical
agitation at 200 rpra. Cells were
cultured initially using 0.5% glucose and
were then induced for 16 hr in 0.5%
tetradecane. Cells were harvested and
stored at -80°C in phosphate buffered
saline made 20% with respect to glycerol.
Cell disruption employed a Biospec
Products "Bead-Beater" (Bartlesville, OK)
under conditions recommended by the
manufacturer. Microsomes were isolated
and washed as described by Yoshida et al.
(23). For solubilization, microsomes
from S_. cerevisiae strains were
suspended at 15 mg total protein ml"1 in
solubilization buffer of Yoshida et al.
(24) and sodium cholate was added to be
1%. For £. tropicalis, microsomes were
suspended at 19 mg total protein ml"1 in
buffer C of Guengerich (4) and sodium
cholate was added to be 1.6%. All
preparations were clarified by
centrifugation at 100,000 xg for 60 rain.
Cytochroraes P-450 were partially
purified by column chromatography on n-
octylaraino Sepharose 4B and DEAE Sephacel
based upon a procedure of Guengerich (4).
Subsequent purification steps utilized
CM-Sephadex based upon a method described
by Yoshida and Aoyama (25).
Partially purified reductase
preparations, obtained by elution from
the same columns by Guengerich's
procedure (4), were purified by further
chromatography. For S_. cerevisiae this
involved separation on hydroxylapatite
according to the method of Bertrand et
al., using the detergent Lubrol PX in
place of Mulgofen BC-720 (1). This was
followed by chromatography on Sephedex G-
150. For C. tropicalis, reductase
obtained from tetradecane grown cells was
chromatographed through a column of
Sephadex G-150; reductase from glucose
grown cells was passed through Sephadex
G-150 and then received a final
chromatography on hydroxylapatite as
described by Bertrand et al. (1).
Cytochrome P-450 concentrations were
determined as described by Guengerich
(10) using a molar extinction coefficient
of 91 mM~l cnT1 (11). Activity of
reductase was determined based upon the
assay using horse heart cytochrome c as
described by Guengerich (4). Protein
determinations were according to Lowry et
al. (10) or used the methods and reagent
kit of Bio-Rad Laboratories.
Gel Electrophoresis, Transfer and
Detection of Antigen Immunoblots
Procedures for sodium dodecyl
sulfate polyacrylamide gel
electrophoresis (PAGE) were based upon
those of Laemmli (8). Protein bands in
gels were visualized by the silver
staining method of Wray et al. (21).
Transfer of proteins from the
polyacrylamide gel to nitrocellulose
filters was done according to Bio-Rad
Laboratories using their product "Trans-
blot" cell. Immunological screening of
the filters employed the BLOTTO procedure
of Johnson et al. (5). Bound antibody
was determined by the horse radish
peroxidase (HRP) color reaction of Towbin
et al. (18) using the reagents obtained
from Bio-Rad Laboratories.
Antibody Preparations
Polyclonal antisera to the isolated
proteins were prepared in rabbits
according to the method described by
Kaminsky et al. (6). Control sera were
collected from.the rabbits prior to their
inoculation.
114
-------
RESULTS
Pioneering work in the laboratories
of Yoshida and Aoyama (25), of Wiseman
(7), and of Coon (2) among others has
defined methods for the isolation from
yeasts of the major cytochrome P-450
system proteins. Using minor
modifications of their published
procedures with our strains, we were able
to purify several such proteins to
homogeneity as determined by silver
staining of proteins bands formed in
PAGE. In those cases where impurities
appeared likely to be a problem in
antibody production, the enzymes were
recovered from slices of the
polyacrylamide gel.
Five different preparations of
antigens were used to produce antisera in
rabbits. For £3. cerevisiae cytochrome P-
450, the antigen was a pair of proteins
isolated from glucose grown cells of
diploid strain JLD15. These two proteins
overlapped extensively through several
separation procedures and both gave
characteristic CO binding spectra (data
not shown). Antisera to reductase of S_.
cerevisiae was prepared against enzyme
purified from the same strain. For C_.
tropicalis, two cytochrome P-450 proteins
were isolated and used separately for
antisera production. One of these, of
apparent m.w. 59 kd, was purified from
glucose grown cells and the other, of
apparent m.w. 54 kd, was purified from
cells grown on tetradecane. Antisera to
the reductase of this organism was
prepared against enzyme purified from the
tetradecane grown cells.
These and additional isolated
cytochrome P-450 system proteins were
then characterized as to apparent m.w.
and immunological relatedness using PAGE
plus immunoblots. Comparable amounts of
the proteins being compared were
electrophoresed from each well, and the
visual intensity of each HRP color
reaction was recorded on a scale of 0 to
4+. Immunoblots run as controls using
nonimmune sera were always negative (data
not shown). All the data, excluding
those using control sera, are summarized
for cytochrome P-450 proteins in Table 2A
and for reductase in Table 2B.
All the £3. cerevisiae cytochrome P-
450 proteins tested reacted strongly with
the antisera prepared to the two protein
mixture isolated from the diploid strain
JLD15. This diploid had been formed by
crossing JL11 and JL12, two haploid
strains which had been shown in previous
studies to have relatively high P-450
content among haploid strains (1). More
recently we have demonstrated that a 56
kd P-450 protein is formed by one of
these haploids and a 57 kd one is formed
by the other. Thus the two P-450
proteins isolated from JLD15 appear to
have resulted from the expression of two
genes provided singly by the haploid
progenitors.
This strong immunoreactivity among
the S_. cerevisiae cytochrome P-450"s is
of particular interest for the P-450
proteins from strains JL10, JL20 and
JL21. Strain JL10 was used in
experiments to be described elsewhere for
the isolation of plasmids pVKl and pVK2,
plasmids which code for a functional
cytochrome P-450 gene (Kalb et al.,
manuscript in preparation). Examination
by PAGE of cytochrome P-450 protein
partially purified from JL10 revealed a
major band of apparent m.w. 56 kd.
Strain JL10 variants containing pVKl, or
pVK2, were designated as JL20 or JL21
respectively. Thse strains expressed
elevated levels of cellular P-450
(manuscript in preparation); microsomal
extracts of these strains also showed
elevated P-450 content. For strain JL20
this appeared as an increase in a
cytochrome P-450 of apparent m.w. 56 kd,
not shown to differ from the protein
detected from JL10. Strain JL21 however
developed high levels of a cytochrome P-
450 with an apparent m.w. 58 kd, a form
not detected in our other strains.
By contrast one of the cytochrome P-
450 proteins purified from C. tropicalis
was immunologically unrelated to two
others. Antisera prepared against the
form purified from glucose grown cells
did not react with either of the two
immunologically related forms purified
from tetradecane grown cells. Similarly
antisera prepared against one of the two
latter proteins, the 54 kd form, did not
bind to the cytochrome P-450 from glucose
grown cells.
However, as recorded in Table 2A,
evidence of low level cross-reactivity
was observed for cytochrome P-450
115
-------
proteins isolated from glucose grown
cells of S. cerevisiae and C.
tropicalis.
Data for the reductases isolated
from these yeasts are summarized in Table
2B. For C. tropicalis, reductase
purified from glucose grown cells was
identical by m.w. and immunological
reactivity to reductase purified from
cells grown on tetradecane. Although
this reductase did differ in m.w. from
the form purified from £3. cerevisiae,
the reductases of these two yeasts were
distinctly immuno-cross-reactive.
DISCUSSION
Although the purification of protein
components of yeast cytochrome P-450
systems has been published, the isolation
of yeast genes for cytochrome P-450
expression has not been described.
Elsewhere we present results of a
parallel study which has yielded plasmid
clones of cytochrome P-450 genes of £3.
cerevisiae (Kalb et al., manuscript in
preparation). In this phase of the work
we were interested in these proteins
primarily in order to obtain antibodies
prepared against them for use in
procedures of gene characterization.
Access to yeast strains engineered for
high level expression of such genes will
facilitate the purification of these
enzymes in quantity. The enzymes can
then be examined as to their possible
substrate specificity for compounds of
environmental concern and in comparison
to activities of yeast cytochrome Pr450
system enzymes described in the
literature (1,2,7,16,25).
The data in Table 2 provide useful
insights as to the probable yeast genes
for these systems. For C. tropicalis,
identical properties were seen for the
NADPH-cytochrome c (P-450) reductase when
that organism was cultured by glucose
fermentation or by tetradecane oxidation.
Thus this enzyme is expressed from a
single gene or from nearly identical
genes under these two growth conditions.
Comparison to the reductase data for £[.
cerevisiae in Table 2 shows an apparent
m.w. difference of 2 kd, but these
enzymes of both species react to
immunosera prepared using either
reductase. This is a strong indication
that the corresponding genes will prove
to share regions of sequence homology.
By. contrast, culture of C.
tropicalis on tetradecane resulted in
elevated levels of two cytochrome P-450
proteins not detected in extracts of this
yeast grown on glucose. These two
proteins showed apparent m.w. values of
53 and 54 kd, compared with the 59 kd
value obtained for the cytochrome P-450
isolated from glucose grown cells. Both
the 53 kd and the 54 kd proteins reacted
with antisera elicited to the 54 kd form.
However, when these two P-450 proteins
were tested in immunoblots using antisera
prepared to the 59 kd form, no cross-
reactivity was observed. Thus these data
suggest that unique P-450 genes are
induced under these alternate growth
conditions, a conclusion reached earlier
by Sanglard et. al. on the basis of other
criteria (16).
For £5. cerevisiae, the data for
cytochrome P-450 proteins in Table 2 are
particularly interesting as they provide
evidence for three proteins, all
immunologically related, but of different
genetically determined size. Cytochrome
P-450 antiserum was prepared using a
mixture of two P-450 proteins obtained
from a diploid strain. These two
proteins, with apparent molecular weights
of 56 kd and 57 kd, subsequently were
shown to be coded separately, one each in
the haploids which were used to derive
the diploid. Of the two plasmid clones
isolated with cytochrome P-450 structural
genes (Kalb et al., in preparation),
presence in the cell of one of these,
pVK2, results in the accumulation of a
protein with an apparent molecular weight
of 58 kd, a form not detected in our
other j3. cerevisiae strains. Gene
sequence and enzyme specificity studies
will be required to determine whether
these three proteins are expressed from
alleles of one or more P-450 genes. The
gene for the glucose grown P-450 in C_.
tropicalis may have some sequence
homo logy with these S_. cerevisiae
.gene(s), based upon the low level immuno-
cross-reactivity observed for these
proteins.
Thus use of these antisera
preparations has provided evidence as to
possible gene relationships among several
cytochrome P-450 monooxygenase systems of
these two organisms. The antisera will
116
-------
be useful in the isolation and
characterization of these genes for the
subsequent gene engineering of P-450
enzyme systems in yeast.
ACKNOWLEDGEMENTS
This work was supported by
Cooperative Agreement No. CR810605, U.S.
Environmental Protection Agency, Office
of Research and Development, Hazardous
Waste Engineering Research Laboratory,
Cincinnati, P.R. Sferra, Project Officer.
REFERENCES
1. Bertrand, J.C., Gilewicz, M., Bazin,
H. and Azoulay, E. 1980. Purified
detergent-solubilized NADPH-
cytochrome c (P-450) reductase from
Candida tropicalis grown on alkanes.
Biochem. Biophys. Res. Commun.
94,889-893.
2. Duppel, W., Lebeault, J-M. and Coon,
M.J. 1973. Properties of a yeast
cytochrome P-450-containing enzyme
system which catalyzes the
hydroxylation of fatty acids,
alkanes, and drugs. Eur. J.
Biochem. 36,583-592.
3. Gasiewicz, T.A., Geiger, L.E.,
Rucci, G., and Neal, R.A. 1983.
Distribution, excretion, and
metabolism of 2,3,7,8-
tetrachlorodibenzo-p-dioxin in
C57B1/6J, DBA/2J, and B6D2Fi/J mice.
Drug Metab and Disposition 11,397-
403.
4. Guengerich, F.P. Microsomal enzymes
involved in toxicology - analysis
and separation, in Principles and
Methods of Toxicology, A.W. Hayes,
ed., Raven Press, NY, 1982, pp. 609-
634.
5. Johnson, D.A., Gautsch, J.W.,
Sportsman, J.R. and Elder, J.H.
1984. Improved technique utilizing
nonfat dry milk for analysis of
proteins and nucleic acids
transferred to nitrocellulose.
1984. Gene. Anal. Techn. 1^,3-8.
6. Kaminsky, L.S., Fasco, M.J. and F.P.
Guengerich. 1981. Production and
application of antibodies to rat
liver cytochrome P-450, in Methods
in Enzymology, Vol. 74, 262-272.
7. King, D.J., Azari, M.R. and Wiseman,
A. 1984. Studies on the properties
of highly purified cytochrome P-448
and its dependent activity
benzo(a)pyrene hydroxylase, from
Saccharomyces cerevisiae.
Xenobiotica 14,187-206.
8. Laemmli, U.K. 1970. Cleavage of
structural proteins during the
assembly of the head of
bacteriophage T4. Nature 227,680-
685.
9. Loper, J.C., Lingrel, J.B. and Kalb,
V.F. Engineering genes in yeast for
biodegradation, in Incineration and
Treatment of Hazardous Waste,
Proceedings, Ninth Annual Research
Symposium, USEPA, IERL, EPA-600/9-
84-015, July, 1984, pp. 274-281.
10. Lowry, O.H., Rosebrough, N.J., Farr,
A.L. and Randall, R.J. 1951.
Protein measurement with the folin
phenol reagent. J. Biol. Chem.
193,265-275.
11. Omura, T. and Sato, R. 1964. The
carbon monoxide-binding pigment of
liver microsomes. J. Biol. Chem.
239,2370-2378.
12. Poiger, H., Buser, H.-R., Weber, H.,
Zweifiz, U., and Schlatter, Ch.
1982. Structure elucidation of
mammalian TCDD-metabolites.
Experientia 38,484-486.
13. Poiger, H., Buser, H.-R. and
Schlatter, Ch. 1984. The
metabolism of 2,3,7,8-
tetrachlorodebenzofuran in the rat.
Chemosphere 13,351-357.
14. Poiger, H. and Buser, H.-R. The
metabolism of TCDD in the dog and
rat in Biological Mechanisms of
Dioxin Action, Bambruy Report 18, A.
Poland and R. Dikinbrough, eds.,
Cold Spring Harbor Laboratory. 1985.
15. Rosazza, J.P. and Smith, R.V. 1979.
Microbial models for drug
metabolism, Advances in Applied
Microbiology 25,169-208.
117
-------
16. Sanglard, D., Kappeli, O. and 21.
Fiechter, A. 1984. Metabolic
conditions determining the
composition and catalytic activity
of cytochrome P-450 monooxygenases
in Candida tropicalis. J.
Bacteriol. 157,297-302. 22.
17. Schunck, W-H., Riege, P. and Kuhl,
R. 1978. Cytochrome P-450 of
eukaryotic microorganisms.
Pharmazie 33,412-414.
18. Xowbin, H., Staehelin, T. and
Gordon, J. 1979. Electrophoretic 23.
transfer of proteins from
polyacrylaimide gels to
nitrocellulose sheets: procedure
and some applications. Proc. Natl.
Acad. Sci. USA 76,4350-4354.
19. van Ommen, B., van Bladeren, P.J.,
Temmink, J.H.M., and Muller, P. 24.
1985. Formation of
pentachlorophenol as the major
product of microsomal oxidation of
hexachlorobenzene. Biochem.
Biophys. Res. Commun. 126,25-32.
20. Weber, K., Pringle, J.R. and Osborn, 25.
M. 1972. Measurement of molecular
weights by electrophoresis on SDS-
acrylamide gel, in Methods in
Enzymology. 26c,3-27.
Wray, W., Boulikas, T., Wray, V.P.,
and Hancock, R. 1981. Silver
staining of proteins in
polyacrylamide gels. Anal. Biochem.
118,197-203.
Yoshida, Y., Kumaoka, H. and Sato,
R. 1974. Studies on the microsomal
electron-transport system of
anaerobically grown yeast. I.
Intracellular localization and
characterization. J. Biochem.
75_, 1201-1210.
Yoshida, Y., Kumaoka, H. and Sato,
R. 1974. Studies on the microsomal
electron transport system of
anaerobically grown yeast, II
Purification and characterization of
cytochrome b5. J. Biochem. 75,1211-
1219.
Yoshida, Y., Aoyama, Y., Kumaoka, H.
and Kubota, S. 1977. A highly
purified preparation of cytochrome
P-450 from microsomes of
anaerobically grown yeast. Biochem.
Biophys. Res. Commun. 78,1005-1010.
Yoshida, Y. and Aoyama, Y. 1984.
Yeast cytochrome P-450 catalyzing
lanosterol 14@-demethylation. I.
Purification and spectral
properties. J. Biol. Chem.
259,1655-1660.
Table 1. Strains
Name
£>. cerevisiae;
JL10
JL11
JL12
JLD15
JL20
JL21
C_. tropicalis;
ATCC750
Description
local name for BWG2-9A
strain D5-1C
a, ade2-40 leu
strain D5-3C
a, ade2-40 trpl
diploid isolate from mating of
JL11 and JL12
JL10 containing pVKl
JL10 containing pVK2
prototroph
Source
L. Guarente,
Massachusetts Inst.
of Technology
this laboratory
this laboratory
this laboratory
tnis laboratory
this laboratory
American Type Culture
Collection
118
-------
Table 2. Molecular Weights and Iiranunological Relatedness of Cytochrome P-450
System Proteins Purified from Strains of Saccharomyces cerevisiae and from
Candida tropicalis ATCC 750.
A. Cytochromes P-450
Organism and
culture conditions
Apparent m.w. (kd)
determined
using PAGE3
Cross reaction in immunoblots
with rabbit antisera prepared
to protein (kd) from strain
JLD15
ATCC?50
56,57
59
54
S. cerevisiae, glucose
fermentation
JLD15
JC11
JC12
56
57
56
57
4+
4+
4+
4+
JL10
JL20
JL21
56
56
56
58
4+
4+
4+
4+
0
0
0
0
C. tropicalis, glucose
fermentation
tetradecane oxidation
59
53
54
4+
0
0
2+
4+
B. NADPH-cytochrome c (P-450) reductase
S. cerevisiae JLD15,
glucose fermentation
C. tropicalis,
glucose fermentation
tetradecane oxidation
72
74
74
ATCC 750
JLD15 grown on tetradecane
72
4+
3+
3+
74
2+
4+
4+
aPAGE: polyacrylamide gel electrophoresis.
119
-------
BIODEGRADATION OF ENVIRONMENTAL POLLUTANTS BY THE WHITE ROT FUNGUS
PHANEROCHAETE CHRYSOSPORIUM
John A. Bumpus, Ming Tien/ David S. Wright, Steven D. Aust
Department of Biochemistry
and
Center for the study of Active Oxygen in Biology and Medicine
Michigan State University
East Lansing, MI 48824-1319
ABSTRACT
The white rot fungus, Phanerochaete chryj;os£oriji>m secretes a unique
hydrogen peroxide-dependent oxidase capable of degrading lignin, a highly
complex/ chemically resistant, non-repeating heteropolymer . Due to its
ability to generate carbon-centered radicals, this enzyme is able to non-
specifically catalyze numerous cleavage reactions producing smaller
lignin-derived compounds which may then be metabolized by more conven-
tional enzyme systems. We have proposed that the lignin degrading
system of this fungus may also have the ability to degrade environmen-
tally persistent organopollutants. In this study we have shown that P^_
.!B is able to degrade carbon-14 labeled 1, 1 ' -Bis( 4-chloro-
phenyl)-2,2,2-trichloroethane (DDT) , 3,4,3' , 4 ' -tetrachlorobiphenyl ,
.2/4, 5/2', 4', 5 '-hexachlorobiphenyl, 2, 3,7, 8-tetrachlorodibenzo-p-dioxin
(TCDD), the gamma isomer of 1 , 2, 3,4 , 5,6-hexachlorocyclohexane (Lindane)
9S well as the non-halogenated pollutant benzo[a Ipyrene to 14C-carbon
dioxide.
INTRODUCTION
Biotreatment systems using
microorganisms for the degradation
of toxic recalcitrant organopollu-
tants hold the promise of being an
efficient and economical means of
detoxifying vast quantities of
contaminated water, soils and
sediments. Unfortunately, few
microorganisms possess the ability
to degrade recalcitrant organo-
pollutants. Generally these com-
pounds are poorly soluble in water
and are adsorbed to particulate
matter thus making them even less
accessible to microbial attack.
Some pollutants of interest are
often present in the parts per
million range or less. Microbial
enzymes which may possess the
innate ability to degrade organo-
pollutants rarely possess the very
high affinity (low Km) required in
order for significant degradation
to take place at such low sub-
strate concentrations. Degrada-
tion of organopollutants present
in low concentrations is also
difficult even for microbial
strains which have been adapted to
grow on these chemicals. When the
selective pressures imparted by
high concentrations of the chemi-
cal are removed, these strains may
not be able to compete effectively
in biotreatment systems containing
other microorganisms. similar
120
-------
problems face nvicrobial strains
developed using recombinant DNA
technology.
Our studies focus on the use
of the lignin-degrading white rot
fungi as potentially useful
organisms in the aerobic biologi-
cal treatment of hazardous organic
wastes. The use of lignin de-
grading fungi may overcome many of
the problems generally associated
with the biological degradation of
insoluble recalcitrant organic
compounds.
In the initial stages of
lignin degradation by P_^
Sh.r-Y.§2s.22!Liy.m.' numerous carbon-
carbon and carbon-oxygen bonds in
the insoluble lignin polymer are
cleaved by a unique non-specific
and non—stereoselective tl2(^2~
dependent/ extracellular oxygenase
(1,2). The enzyme catalyzes the
formation of carbon-centered radi-
cals which react with oxygen to
initiate oxidation (3). This free
radical mechanism may be responsi-
ble for the lack of specificity
and stereoselectivity that is
characteristic of this system.
Synthesis of this enzyme is pro-
moted by nitrogen starvation (4)/
not substrate availability, as is
the case with many biodegradative
systems- The mixture of soluble
low molecular weight aromatic com-
pounds thus formed may then under-
go further modification or ring
cleavage and metabolism via the
B -ketoadipate pathway to Krebs
cycle intermediates and/ hence/ to
carbon dioxide.
PURPOSE
It was the purpose of this
study to determine if a white rot
fungus/ P^ £h£^s o sjoor ium / nad tne
ability to degrade recalcitrant
organopollutants.
APPROACH
Cultures of
(ME—£46) were maintained on 2%
malt/agar slants at room tempera-
ture. Experiments were performed
in liquid culture at 39°C. For
the first 3 days cultures were
incubated under air after which
the culture atmosphere was changed
to 100% oxygen. Cultures (10 ml)
were incubated in 250 ml Wheaton
media bottles equipped with gas
exchange manifolds similar to
those described by Marinucci and
Bartha (5). The culture medium/
consisting of glucose (56 mM)/
ammonium tartrate (1.2 mM)/ and
dimethylsuccinate buffer/ pH 4.2
(100 mM)/ supplemented with
thiamine and trace metals/ has
been previously described (4).
The 14CO2 evolved during degrada-
tion of *4C-labeled organopollu-
tants was trapped by using oxygen
to flush the atmosphere of the
incubation flasks through an
ethanolamine-containing scintilla-
tion cocktail as described (4).
Radioactivity was quantitated
using a Packard Tri-Carb Liquid
Scintillation Spectrometer (Model
3310) .
Metabolism of l/l'-Bis(4-
chlorophenyl)-2/2/2-trichloro-
ethane (DDT) was studied by incu-
bating DDT with the fungus and
assessed by extracting the cul-
tures with 25 ml of acetonitrile/
2 ml of saturated NaCl and two 50
ml portions of hexane. The hexane
extracts were combined and the
amount of DDT and metabolites were
quantitated using a Varian GLC
(Model 3700) equipped with an
electron capture detector and a
Hewlett Packard Digital Integrator
(Model 3390A). The identity of
metabolites was confirmed by GC-MS
and by comigration of metabolites
or their derivatives with authen-
tic standards during GLC.
Experiments were conducted in
triplicate or quadruplicate. Data
points represent the mean + 1
standard deviation.
121
-------
RESULTS
Initial studies using DDT as
a model recalcitrant substrate
demonstrated that P_-_ chr^soggori^um
degraded DDT- During a thirty day
incubation period approximately
53% of the DDT was metabolized
(Fig. 1). Synthesis of DDT meta-
5.0
X 40
ui
c
o
z
I 2.0
UI
-------
TABLE 1- DEGRADATION OF DDT BY P^ CHRYSOSPORIUM^
EFFECT OF SUPPLEMENTAL GLUCOSE
Days of
Incubation
Glucose
Addition
(56 mM)
DDT
Concentration
(uM)
Percent
Degradation
O
30
31
48
61
75
4.8
2.3
0-4
<0-048
0
48
92
>99
accounted for approximately 25% of
the DDT initially present. These
studies suggest that degradation
will continue as long as a
suitable carbon source is present.
When the organism was grown
in the presence of 14C-DDT/ 4% of
the original DDT was evolved as
14CO2 after 30 days (Fig. 3) when
roughly 50% of the DDT was de-
graded (Fig. I)/ indicating that
the remaining carbon atoms are
either incorporated into the orga-
nism or are present as inter-
mediates in the pathway between
DDT and CO2« Unlike DDT dis-
appearance studies (Fig. I)/ which
demonstrate that substantial
degradation occurs during the
first three days of incubation/
14CO2 evolution did not occur
until after 3 days of incubation
(Fig- 3). This is the same type
of lag as that which is observed
for lignin degradation (4) and is
consistent with the hypothesis
that lignin degradation and DDT
degradation are mediated by the
same enzyme system.
Table 2 lists six struc-
turally diverse toxic organopollu-
tants that were found to be de-
graded by P. chry_sc>sp_ori.u_m as
assayed by ^4CO2 evolution. Like
DDT, 1/2,3,4/5/6-hexachlorocyclo-
hexane (Lindane), 2,3,7/8-tetra-
chlorodibenzo-p-dioxin (TCDD)/
Benzo[a]pyrene/ 3/4,3',4'-tetra-
chlorobiphenyl (3,4,3',4'-TCB) and
2/4/5,2',4'/5'-hexachlorobiphenyl
(2,4,5,2',4',5'-HCB) are all
degraded to CO2 by this microorga-
nism- In all cases radiolabeled
substrates were added to cultures
when they were inoculated with
fungi at day 0- By day 3 substan-
tial growth was apparent as
evidenced by the appearance of a
mycelial mat- However/ in no case
did *-4CO2 evolution occur.
Between day 3 and day 6, 14CO2
evolution began in all cultures.
It was maximal between day 3 and
day 18 and continued at decreasing
rates until day 30- After 30 days
of incubation/ glucose was not
detectable in the incubation
media. Fortification of the media
with additional glucose (56 mM)
and continued incubation resulted
123
-------
in an increased rate of 14CO
evolution for the duration of
another 30 days at which time the
experiment was terminated.
CM
O
2500
2000
O
O
u ^-.
t 2 1500
8
KX)0
500
6 12 18 24
TIME (DAYS)
30
Figure 3. Conversion of 14C-DDT to
H2 in cultures of P^
chrysosporium. Each incubation
flask contained 50 uCi of C-DDT-
Th© initial concentration of DDT
was 4-8 uM- These data represent
the minimum amount of DDT
converted to CO2 - Some CC>2 may
have been incorporated into the
organism* in carboxylation
reactions/ for example.
The lignin degrading system of P^
chrysospor i um may prove to be
ideally suited for use in bio-
treatment processes for the degra-
dation of recalcitrant organo-
pollutants. First, the enzyme
system normally attacks an insol-
uble recalcitrant substrate. Thus
organopollutants which are
adsorbed to sediments may actually
mimic the lignin molecule. The
analogy is even more striking when
one considers that many sediments
and soils to which organopollu-
tants are adsorbed have high
lignin contents. Second/ problems
associated with substrate
specificity appear to be obviated
by the non-specific and non-
stereoselective carbon-centered
free radical mechanism that is
characteristic of this system.
This lack of specificity has the
advantage of allowing the organism
to attack and degrade a broad
spectrum of structurally diverse
recalcitrant compounds as shown in
Table 2. Third/ P_^ chrvjsosjDorjjjm
is a highly successful competitor
in nature/ especially when the
carbon source consists of wood/
wood by-products or other lignin
containing materials. Thus/ if
wood chips or sawdust/ for
example, are used as the carbon
source in biotreatment systems/
competition by non-1ignin
degrading organisms is likely to
be minimal. Fourth/ because de-
gradation is promoted by nitrogen
starvation rather than by the
presence of substrate/ low levels
of organopollutants do not repress
the biosynthesis of enzymes re-
quired for their degradation.
Fifth/ typically/ degradative
enzymes must possess very high
affinities (low Km) in order for
significant degradation to occur.
The carbon-centered radical mecha-
nism appears to provide an alter-
native mechanism which allows de-
gradation of lignin/ and possibly
other recalcitrant compounds/ to
proceed to completion.
Of special significance for
the degradation of organohalides
is the fact that P^ chr^sosjaoriym
has the demonstrated ability to
dehalogenate and degrade chloro-
benzene derivatives (6) and to
cleave aromatic rings (7).
Included among the compounds
degraded by £_-_ chr^sosgori^um are
4/5/6-trichlorophenol/ 3/4/5/6-
tetrachloroguaiacol and 2/4,6-
trichlorophenol (6). In our
studies we have shown that ODD 'is
an intermediary metabolite formed
during DDT degradation. We have
124
-------
TABLE 2. DEGRADATION OP 14C-RADIOLABELED ORGANOPOLLUTANTS TO 14CO2 BY P_^ CHRYSOSPORIUM
Initial Rate of
Degradation to
(pmoles/day)
Radiolabeled Substrate
evolved as 14CO2
(pmoles)
% of
Radiolabeled
substrates evolved
as 14CO2 in 60 days
30 days
60 days
Lindane
Benzo[a ]pyrene
DDT
TCDD
3,4/3' , 4 ' -TCB
2,4,5,2' ,4' ,5'-HCB*
11-
7.
2.
1.
0.
2.
3
5
7
2
7
4
190
117
48
27
13
44
.8
.2
.0
.9
.8
.2
267
171
116
49
25
86
-6
.9
.4
.5
.1
.0
21
13
9
4
2
1
.4
.8
.3
.0
.0
.7
^Substrate concentration was 1.25 nmoles/10 ml for all ^C-radiolabeled compounds except
2,4,5,2',4',5'-HCB. Because of its low specific radioactivity a concentration of 5.0 nmoles/10 ml
was used for 2,4,5,2',4',5'-HCB.
also shown that Lindane/ a compound
that is chlorinated at every carbon
atom/ is converted to CC>2 • These
results demonstrate that P^_
£llEY.s.2.sp_orium niay also dehaloge—
nate alkyl halides.
SUMMARY
The lignin degrading system
of P^_ £!lEy_s.o.sJ2o.]Liy_m. appears to be
ideally suited for the degradation
of other recalcitrant organic com-
pounds- The initial stages of
lignin degradation are extracellu-
lar/ non-specific and non-stereo-
selective. Degradation proceeds
all the way to 2- Additionally/
nas a const itui-
as well as a system which can
dehalogenate many organohalides -
Under conditions which pro-
mote lignin degradation/ P^_
EQ also degraded
tive aromatic ring cleavage system
Lindane/ benzo[a Jpyrene/ DDT/
TCDD, 3,4,3* /4 '-TCB and
2, 4, 5, 2', 4', 5' -HCB to CO2 - These
studies suggest that P^
£!lEY.s.o.sJ2o.]Liy.!B "iay prove to be an
extremely useful microorganism in
the biological treatment of hazar-
dous organic waste.
ACKNOWLEDGEMENTS
This work was supported by
Cooperative Agreement #CR811464
U.S. Environmental Protection
125
-------
Agency/ Office of Research and
Development/ Hazardous Waste
Engineering Research Laboratory,
Cincinnati/ P.R. sferra, Project
Officer. The authors wish to
thank Ms. Cathy M. Custer for
secretarial assistance in the
preparation of this manuscript.
REFERENCES
1. Tien/ M- and T-K. Kirk, 1983.
Lignin-degrading enzyme from
the hymenomycete £hanero=
SllSS^S £hJ£2S2s.E2£iy.!S Burds-
Science, Vol. 221, pp. 661-
662.
2. Tien, M. and T-K- Kirk, 1984.
Lignin-degrading enzyme from
£ha_nerochae^e chry.sosgor i.um:
purification, characterize-"
tion, and catalytic proper-
ties of a unique H2O2-
requiring oxygenase. Proc_^
Natl. Acad^ Sc^, USA, Vol.
81/ pp. 2280-2284.
3. Kersten, P.J., M. Tien, B-
Kalyanaraman and T.K- Kirk,
1985. The ligninase of
Pha_n_e.rochaete 2h£Y.^2§.E21Ziy.E5
generates cation radicals
from methoxybenzenes- j^
Biol. Chem., Vol. 260, No. 5,
pp. 2609-2612.
4- Kirk, T.K- , E- Schultz, W-Cf.
Connors/ L-F. Lorenz and J.G.
Zeikus, 1978. Influence of
culture parameters on lignin
metabolism by Pha_nerochaete
chry sospor ijjm - Arch^
MjcrobiolT/ Vol. 117, pp.
277-285.
5. Marinucci/ A.C. and R.
Bartha/ 1979. Apparatus for
monitoring the mineralization
of volatile 14C-labeled
compounds. Apjol^ BnyjLron^
Microbiol^, Vol. 38, No. 5,
pp. 1020-1022.
6. Chang/ H--m./ T-W- Joyce/
A.G. Campbell/ E-D. Gerrard,
Van-Ba Huynh and T*K. Kirk,
1983- Fungal decolorization
of bleach plant effluents. In:
(T. Hicuchi, H.-m.
Chang and T.K. Kirk, eds.), uni
Publishers Co., Ltd., Tokyo, pp.
257-268.
7. Leatham, G.F., R.L- Crawford
and T.K. Kirk, 1983. Degrada-
tion of phenolic compounds and
ring cleavage of catechol by
.
. Environ^
No. 1, pp. 191-197-
Vol.
126
-------
BACTERIAL DEGRADATION OF CHLORINATED COMPOUNDS
Paul H. Tomasek and A.M. Chakrabarty
Department of Microbiology and Immunology
University of Illinois at Chicago; Chicago, IL 60680
ABSTRACT
Plasmid gene sequences of the 2,4,5-trichlorophenoxyacetate (2,4,5-T) degrading
Pseudomonas cepacia AC1100 strain show strong homology to a cluster of genes present on
the 2,4-dichlorophenoxyacetate (2,4-D) degradative plasmid pJP4. There is little AC1100
deoxyribonucleic acid (DNA) homology to the chlorocatechol genes present on the 3-chloro-
benzoate (3CBA) degradative plasmid pAC27. The homology to the 2,4-D degradative plasnnd
pOP4 may therefore be attributed to other genes involved in the degradation of chlorophen-
oxyacetates. 2,4,5-T grown AC1100 cells seem to lack ring-fission dioxygenase activity
for chlorocatechol intermediates of the 3CBA and 2,4-D pathways. These results suggest
that a significantly different catabolic pathway may have evolved in £. cepacia AC1100.
Transposon mutagenesis has shown that at least some of the 2,4,5-T catabolic genes are
chromosomal. A short chromosomal DNA sequence associated with a putative 2,4,5-T gene is
repeated many times on both the chromosome and plasmid of strain AC1100. Such a repeated
element may have been important in the evolution of the degradative plasmids as well as
genetic rearrangements necessary to efficiently degrade novel substrates.
INTRODUCTION
It is generally believed that the com-
plete biodegradation of chlorinated com-
pounds in water and soil is accomplished
primarily by microorganisms (1). Failure
of microbial populations to degrade some
of these compounds can be attributed to
the lack of genetic information coding for
appropriate degradative enzymes (2). This
reflects the relatively short timespan
such compounds have been present in the
biosphere (several decades) compared to
the long evolutionary history of microorga-
nisms in contact with naturally occuring
products (3).
It is clear that the biodegradative
abilities and limitations of microorga-
nisms must be thoroughly examined if we
want to utilize them for degradation of
persistent toxic chemicals. It is of
equal importance that we understand the
genetic organization and regulation of
existing biodegradative genes in a variety
of organisms if genetic engineering is to
play a role in the construction of new
microbial strains with enhanced degrada-
tive capabilities. The study of gene
structure may shed some light on those
particular evolutionary mechanisms which
allow microorganisms to adapt to novel
compounds. This may, in turn, allow for
more efficient gene engineering. Studies
of the biochemistry and genetic organiza-
tion of 3-chlorobenzoate (3CBA) and 2,4-di-
chlorophenoxyacetate (2,4-D) degrading bac-
teria have contributed most extensively to
our knowledge of chlorinated aromatic
chemical biodegradation (4). The biochem-
ical pathways of both compounds is shown
in Fig. 1.
Genetics of 3CBA-degrading bacteria
The complete degradation of 3CBA has
been reported in Pseudomonas putida strain
AC858, Pseudomonas sp. B13, and Alcali-
genes eutrophus JMP134 (5,6,7). All,these
independently isolated strains carry plas-
mids (pAC27, pWRI, and pJP4 respectively)
127
-------
2,4-Dichlorophenoxy-
acelic acid
2.4-Dichtorophenol
3,5-Dtchlorocalechol
.',4'DichIcTomuconfc
acid
2-Chtoro-dlene
lac lone
OH
, ,.„
2-cWoroniil.ytac.tte
2-ChlofO-3-oxoad(pic
acid
Fig. 1. Proposed pathways for the degrada-
tion of 3CBA and 2,4-D. Solid arrows in-
dicate plasmid-encoded enzymes while open
arrows represent chromosome-encoded
enzymes in an appropriate host. The plas-
mid-encoded pyrocatechase II, and cyclo-
isomerase II have broader substrate specif-
icities than their chromosomal iso-func-
tional counterparts and thus catalyze
reactions with chlorinated substrates.
Hydro!ase II specifically recognizes
chlorinated substrates. Dechlorination of
chloromuconates is thought to occur
spontaneously as the result of lactoniza-
tion by cycloisomerase II. A-Benzoate Diox-
ygenase, B-Dehydrogenase, C-Pyrocatechase
II, D-Cycloisomerase II, E-Hydrolase II,
F-Chloromaleylacetate Reductase, G-Maleyl-
acetate Reductase, X-2,4-D Monooxygenase,
Y-2,4-Dichlorophenol Monooxygenase. (After
K.N. Timmis et al., 4).
which code for many of the essential
degradative enzymes: pyrocatechase II,
cycloisomerase II and hydro!ase II genes
(C,D, E; Fig.l) (8). The initial degrada-
tive enzymes are coded by the host chromo-
some (9). The plasmids pAC27 and pWRl are
essentially identical (10) whereas pJP4
also carries genes for mercury resistance
and additional enzymes necessary for 2,4-D
degradation. The structural genes for
pyrocatechase II, cycloisomerase II, and
hydrolase II have recently been cloned
from both pAC27 and pJP4 (8). In both
plasmids the genes are clustered: pAC27
genes are localized to the 4.2 kilobase
pair (kb) Bgl II E fragment; pJP4 genes
are located on the 15 kb Eco R1 B fragment
(Fig. 2). These gene clusters show
extensive DMA homology with each other.
In addition, potential regulatory gene(s)
necessary for efficient expression of the
3CBA pathway are located near the structur-
al gene clusters: pAC27 Bgl II C fragment
and pJP4 Eco R1 E fragment (Fig. 2).
Genetics of 2,4-D degrading bacteria
A number of different 2,4-D degrading
bacterial strains have been isolated (7,
11,12). The genetically best character-
ized is Alcaligenes eutrophus strain JMP
134, carrying the plasmid pJP4. In addi-
tion to carrying the pyrocatechase II,
cycloisomerase II, and hydrolase II genes
(C,D,E; Fig.l) which allow chlorocatechol
metabolism, the pJP4 plasmid also encodes
the two initial pathway enzymes, 2,4-D
monooxygenase and 2,4-dichlorophenol mono-
oxygenase (X,Y; Fig.l) (9) as well as 2-
chloromaleylacetate reductase (F; Fig.1)
(13). With the possible exception of the
Fig. 2. Physical maps of the plasmid
pAC27 with Eco R1 and Bgl II and pJP4. with
Eco R1. Fragments B1-B3 were generated by
Bam HI digestion of the pJP4 Eco R1 E frag-
ment. Regions of homology are indicated
by shaded areas. (After D. Ghosal et al.,
8). '
128
-------
latter snzyme, all 2,4-D catabolic enzymes
are coded by the 15 kb Eco R1 B gene clus-
ter of pJP4. This fragment has been
cloned (8,9) and transfer of the recombi-
nant plasmid to a plastnid-free A. eutro-
phus JM134 derivative allowed growth on
2,4-D and 3CBA, provided selection pres-
sure was maintained (4,9). There have
been recent attempts to further localize
genes within this-clus-ter; Liu and
Chapman (13) recently localized the pyro-
catechase II (dichlorocatechol dioxygen-
ase) gene to a 4.9 kb Pst I restriction
fragment of pJP4. They also reported that
the hydrolase II (chloro-diene lactone
hydrolase) gene and the chloromaleyl-
acetate reductase were both present on a
3.4 kb PstI fragment of pJP4. The rela-
tionship of these fragments to the
established physical map (8,14) of pJP4
remains to be reported.
Degradation of 2,4,5-trichlorophenoxy-
acetic acid
Pseudomonas cepacia AC1100, original-
ly isolated from a chemostat enrichment
system, has the ability to completely de-
grade 2,4,5-trichlorophenoxyacetic acid
(2,4,5-T) as it sole source of carbon and
energy (15). Although 2,4-D and 2,4,5-T
are chemically analogous, it is not known
whether the biodegradative pathway of £.
cepacia AC1100 is similar to that found in
2,4-D degrading strains. Rosenberg and
Alexander (16) proposed a 2,4,5-T catabol-
ic pathway based on gas chromatography-
mass spectrometry .of accumulated metabol-
ites from a mixed population of soil
bacteria . Similar to 2,4-D degradation,
they proposed that 3,5-dichlorocatechol
was the ring-fission substrate. A varie-
ty of metabolites including 3,5-dichloro-
catechol are produced when ACT 100 cells
are grown on 2,4,5-T. Mass spectral evi-
dence indicates that hydroxylation, reduc-
tive dechlorination of the aromatic ring,
dehalogenation/ hydroxylation of the ring,
and cleavage of the acetate side chain are
all possible transformations (17,18).
Genetics of the 2,4,5-T degrading strain
Unlike its 3CBA and 2,4-D degrading
counterparts, little is known concerning
the genetic organization of strain AC1100
for 2,4,5-T biodegradation. Determination
of the role of plasmid genes encoding
2,4,5-T biodegradative enzymes has been
hampered by the presence of several differ-
ent plasmids and their low yield from
AC1100. Through electron microscopic
measurements it has'been determined that a
170 kb plasmid (pDG3) comprises about 85%
of the tot-al plasmid population. A 40 kb
plasmid (pDG4) accounts for 10% of the
plasmid population while the remainder is
made up of a heterogeneous mixture of
plasmids ranging from 3 - 30 kb (18).
In order to learn more about the gene-
tic organization of the 2,4,5-T degrading
P. cepacia AC1100, we compared AC1100
chromosomal and plasmid DNAs to sequences
known to carry clusters of 3CBA and 2,4-D
catabolic genes through DNA-DNA hybridiza-
tion. To establish whether AC1100 util-
ized a ring-fission dioxygenase such as
pyrocatechase II, we synthesized several
chlorocatechols and used them as sub-
strates for dioxygenase assays. We also
used transposon mutagenesis to introduce
genetic markers for mapping the 2,4,5-T
genes and to generate mutants blocked in
the catabolic pathway. In the course of
these and other studies we obtained evi-
dence of genetic rearrangements affecting
the expression of degradative genes. We
also identified a DNA sequence from the
chromosome of ACT 100 which is repeated
more than 25 times throughout the chromo-
some and plasmids of this strain.
METHODS:
Growth of strain AC1100 (15), isola-
tion of plasmid DNA (6), restriction diges-
tion, gel electrophoresis and Southern
blotting of DNA fragments to nitrocellu-
lose (19), and DNA electron microscopy (8)
have been documented elsewhere. When £.
cepacia ACT 100 was grown for plasmid DNA
isolation, the 2,4,5-T medium was supple-
mented with 2 g/1 glucose. Isolation of
individual DNA fragments was done with
NA45 DEAE paper (Schleicher and Schuell,
Keene, NH) using procedures recommended by
the manufacturer. DNA fragment sizes were
determined by electrophoresis of standard
DNA fragments (1 KB Ladder, BRL, Gaithers-
burg, MD). Radiolabeling of DNA probes
was accomplished using the nick-transla-
tion kit and protocols of BRL, and alpha-
32P-nuc1eotides from Amersham, Arlington
Heights, IL. In vitro packaging of DNA
cloned in the cosmid vector pHC79 and
subsequent transfection into _E. coli AC80
were performed using reagents and proto-
129
-------
cols of Amersham. Transposon mutagenesis
was performed in a tri-parental mating
(20) using E. coli strains carrying
ColEl::Tn5 fdonor) and pRK2013 (mobilizing
functions) and 2,4,5-T-grown AC!100 cells
(recipient). Selection of kanamycin-
resistant Tn5 mutants was done by plating
mating mixtures on Pseudomonas Isolation
Agar (Difco,, Detroit, MI) supplemented
with 100 ug/ml kanamycin. 2,4,5-T~
mutants were identified by failure to grow
on 2,4,5-T agar plates containing 15 ug/ml
kanamycin. Preparation of cell extracts
and measurement of oxygen consumption in
enzymatic analyses were performed essen-
tially as described previously (21).
Chlorocatechols were synthesized according
to Knuutinen (22).
RESULTS:
oxidized.
Isolation of 2,4,5-T mutants generated by
transposon mutagenesis
AC1100 DNA homology to chlorocatechol
degradative genes
To establish whether there was signif-
icant homology between the genes of the
2,4,5-T degrading AC1100 strain the 2,4-D
plasmid pJP4 originally isolated from A.
eutrophus. Eco R1 digested AC1100 chromo-
some and plasmid DNA were probed with
individual radiolabeled pJP4 Eco R1 frag-
ments. There was strong homology between
a fragment (about 4 kb) from the AC!100
plasmid digest and the 15 kb EcoRI B frag-
ment of pJP4 which harbors the structural
genes for both 3CBA and 2,4-D metabolism.
A similar hybridization experiment with
the chlorocatechol genes of the 3CBA
plasmid pAC27 (4.2 kb Bglll E fragment)
recently failed to show any homology to
AC1100 plasmid or chromosomal DNA under
stringent conditions.
Testing chlorocatechol dioxygenase
activity in AC1100
Despite similarities in chemical
structure between 2,4-D and 2,4,5-T, £.
cepacia AC1100 cannot grow on 2,4-D, un-
less these genes are deliberately intro-
duced into AC1100 (18). We have been
unable to demonstrate any ring-fission
dioxygenase activity on 3,5-dichloro-
catechol or 4-chlorocatechol in cell
extracts of 2,4,5-T grown £. cepacia
AC1100. These compounds are substrates
for pyrocatechase II described earlier.
4, 5-Dichlorocatechol was also not
Transposon insertion mutagenesis with Tn5
was used to generate blocked mutants and
introduce convenient genetic markers into
the 2,4,5-T degrading £. cepacia strain
AC1100. About 100 mutants deficient in
2,4,5-T metabolism were obtained from 1700
putative Tn5 insertion mutants, expressing
kanamycin resistance of Tn5. A number of
amino acid auxotrophs were also identi-
fied. Plasmids and chromosome from five
Tn5 mutants (four 2,4,5-T~, one 2,4,5-
T+ leu~) were digested with Eco R1
restriction endonuclease. These DNA frag-
ments were immobilized on nitrocellulose
and hybridized with radioactively labeled
Tn5 DNA. The probe hybridized only to the
chromosomal DNAs in all cases. Thus the
Tn5 insertion site for the 5 mutants test-
ed is chromosomal. The 2,4,5-T mutants
varied in phenotype upon growth on glucose
and exposure to 2,4,5-T: different amounts
of 2,4,5-trichlorophenol accumulated, vary-
ing amounts of chloride were released to
the culture medium, and after 2-3 days of
culturing the color ranged from colorless
to a very dark brown color. One mutant,
PT88, was selected for further genetic
analysis since it produced a dark brown
color when exposed to 2,4,5-T. PT88 was a
putative chlorocatechol dioxygenase mutant
since such a color is often associated
with chlorocatechol formation. The chlor-
inated metabolite produced by this mutant
remains to be adequately characterized.
Identification of a chromosomal repeated
DNA sequence specific to strain AC1100
The kanamycin resistance marker of the
PT88 Tn5 insertion was used to clone chrom-
osomal DNA sequences flanking the site of
insertion. Total genomic DNA was partial-
ly digested with Sau3A restriction endonuc-
lease. The large (20 - 40 kb) fragments
generated were cloned into the vector
pHC79, and the recombinant plasmids intro-
duced into _E. coli. Kanamycin-resistant
colonies were used to identify clones con-
taining the Tn5 and the flanking chromo-
somal DNA. Using radiolabeled Tn5 DNA as
a probe, two clones were identified which
shared a common 6 kb Sail restriction frag-
ment with the original PT88 chromosome.'
This 6 kb DNA fragment contained approx-
imately half of'the Tn5 sequence including
130 .
-------
the kanamycin resistance gene. The 6 kb
Sail fragment was isolated from one of the
clones and used to probe native AC1100
chromosomal digests in an effort to local-
ize the putative 2,4,5-T structural gene.
As shown in Fig. 3A, the 6kb probe identi-
fied several homologous regions on both
the £. cepacia AC!100 chromosomal (lanes
1,4,7) and plasmid (lane 6) digests. It
did not hybridize with P. cepacia 383
chromosomal DNA (lane 37 included as a
control. The highly repeated sequence
present on the 6 kb Sail fragment has been
further localized to a 1.4 kb Sail, PvuII
restriction fragment. When this fragment
was used as a radioactive probe against £„
cepacia AC1100, £. aeruginosa, £. putida,
and £. mendocina digested chromosomes,
only the AC1100 showed homology (Fig.SB).
Genetic rearrangements in pJP4 affecting
degradation
Transfer of the intact pJP4 plasmid from
A. eutrophus to £. putida and prolonged
selection on 3CBA plates (4-5 weeks) even-
tually allowed the recipients to grow on
3CBA. The restriction profile of the
resident plasmid was different from .native
pJP4 (18). The resulting plasmid (repre-
sented by pYG2) was further characterized
by Southern hybridization and electron
microscopy of heteroduplex structures ob-
tained by DNA denaturation/fast renatura-
tion (Fig.4). These experiments demonstrat-
ed that pYG2 was derived from pJP4 which
had undergone genetic rearrangements: dele-
tion of about 15 kb, inverted duplication
of the chlorocatechol structural gene clus-
ter (Eco Rl B fragment and other frag-
ments), and gene fusion. A. eutrophus
carrying pJP4 grows slowly on 3CBA and re-
striction digestion profiles of plasmids
isolated from cells maintained on 3CBA
showed a mixture of native pJP4 and altei—
ed (pY62) plasmids. Such 3CBA selected A.
eutrophus strains do not readily express
2,4-D phenotype. Transformation of £.
putida with intact pJP4 and selection with
2,4-D has not been successful. The trans-
formation did occur using the plasmid-free
£. cepacia strain 383 as a recipient and
there were no apparent genetic alterations
of pJP4 in such 2,4-D degrading strains.
DISCUSSION
Unlike 3CBA and 2,4-D degradative genes
which are located on plasmids, the loca-
tion of the 2,4,5-T biodegradative genes
in £. cepacia AC!100 plasmids is still un-
known. Jhe chromosomal location of sever-
al Tn5 insertion mutations which block
2,4,5-T metabolism suggests that chromo-
somal genes are essential for catabolism
of this substrate. The role of plasmid
gene sequences is suggested by the report-
ed homology between a 4 kb region of plas-
mid DNA (a mixture of at least two plas-
mids) and the cluster of 2,4-D catabolic
genes located on the 15 kb EcoRI B frag-
23*5678 1234
Fig. 3. Hybridization
of various chromosome H
and plasmid restriction
fragments probed with
radiolabeled DNA fragments containing a
repeated sequence from £. cepacia AC1100
chromosome. A._ Lanes 1-3: AC1100, PT88,
and £. cepacia 383 chromosomal DNAs,
respe~ctively, digested with Eco Rl; Lanes
4-6: AC1100 and PT88 chromosomal DNAs,
and AC1100 plasmid DNA, respectively,
digested with Pvu II; Lanes 7-8: AC1100
and PT88 chromosomal DNAs, respectively,
digested with Sal I. The hybridization
probe was a 6 kb Sal I fragment isolated
from the Tn5-generated 2,4,5-T mutant
PT88. The 6 kb fragment contains part of
Tn5 and flanking chromosomal sequences at
the site of Tn5 insertion, B._ Lanes 1-4:
£. cepacia AC1100, £. aeruginosa, £.
putida, and £. mendocina chromosomal DNAs,
respectively, digested with Eco R1. The
hybridization probe was a 1.4 kb Sal 'I-Pvu
II fragment from the 6 kb Sal I fragment ,
described above.
131
-------
. i, :':; i,":,-. I i'!!!!! ',•
• • * «<
ment of pJP4 (17). However, the apparent
lack of DMA homology to pAC27 sequences
coding-for pyrocatechase II, cycloisomer-
ase II, and hydrolase II genes indicates
that the homology between AC1100 plasmids
and pJP4 may be due to other genes in-
volved in the degradation of chlorophenoxy-
acetates. Although the putative chloro-
catechol dioxygenase genes may be unstable
in this strain, the inability of Pseudo-
monas cepacia AC1100 to utilize 2,4-D or
3CBA as well as chlorocatechol inter-
mediates of these pathways suggests that
AC1100 may have evolved a significantly
different pathway for 2,4,5-T degradation.
Further mapping of Tn5 mutants and addi-
tional DMA hybridization studies are
required to localize the 2,4,5-T struc-
tural and regulatory genes.
Fig. 4. Electron micrograph and corre-
sponding drawing of an altered pJP4
plasmid (pYG2 form) upon denaturation and
fast renaturation. The presence of a 25
kb inverted duplication of pJP4 genes is
Indicated by the double-stranded DNA
"stem" with 1.3 kb and 43 kb single-strand-
ed loops on the ends. Restriction mapping
and Southern hybridization analysis reveal
that about 15 kb of pJP4 DNA have been
deleted from pY62. pYG2 forms are generat-
ed when either £. putida or A. eutrophus
hosts carrying pJP4 are subjected to 3CBA
selection pressure. W74 single stranded
and double stranded DNA standards are indi-
cated. (From Ghosal et al., 18).
Gene rearrangements play an important
role in the ability of microorganisms to
develop new catabolic pathways (23). This
is illustrated by the amplification, dele-
tion, and fusion which occur when pJP4-
harboring strains are exposed to strong
3CBA selection pressure. In this case the
genetic' alterations retard the ability of
A. eutrophus JMP134 to degrade 2,4-D,
normally a readily metabolized substrate
in this strain (8,17). Through such rear-
rangements the cell is able to express pre-
viously existing biodegradative genes with
broad substrate specificity for catabolic
(3CBA) functions which are normally
silent. It is interesting that biodegrada-
tive gene sequences from two different
catabolic plasmids (pAC27 and pJP4) are
amplified 7-10 times (8) when cloned in
isolation from the normal regulatory genes
and subjected to strong selection pres-
sure. Apparently such an increase in gene
dosage is required to produce sufficient
quantities of 3CBA degradative enzymes in
the absence of genetic regulatory systems
that allow efficient enzyme formation.
Repeated sequences in tandem or
inverted orientation are often associated
with gene amplification and deletion (24,
25). These sequences may be partially or
completely homologous. When the plasmid
pJP4 is denatured and rapidly reannealed
and examined by electron microscopy,
Ghosal et al. (17,18) observed many small
stem-loop structures (indicating inverted
repeats) up to 1.8 kb in length. Several
such structures were present in EcoRl frag-
132
-------
merits B and E, which contain the structur-
al and putative regulatory gene(s) for
chlorocatechol metabolism.
A highly repeated DNA sequence pres-
ent in both the plasmid and the chromosome
of £. cepacia AC1100 has been identified.
The repeated element was isolated by its
close proximity to a chromosomal Tn5 inser-
tion affecting 2,4,,5-T biodegradation.
It is likely that this element is associat-
ed with a 2,4,5-T biodegradative or regula-
tory gene. It is tempting to consider that
this or a similar repeated sequence led to
the recruitment and rearrangement of
various biodegradative genes into an effic-
iently regulated 2,4,5-T catabolic system.
Further efforts are currently underway to
establish whether the repeated sequence
present in a 1.4 kb segment will enhance
genetic recombination given the multiple
chromosomal and plasmid sites showing
strong homology to the repeat.
ACKNOWLEDGEMENTS
The authors express their thanks to Drs.
I.S. You, D. Ghosal, and O.K. Chatterjee
for their invaluable contributions to
these studies. We also thank Drs. V;
Deretic and J. Gill for helpful sugges-
tions. This work was supported in part by
Cooperative Agreement, CR809666, U.S.
Environmental Protection Agency, Office of
Research and Development, Hazardous Waste
Engineering Research Laboratory, P. R.
Sferra, Project Officer. P.H.T. acknowl-
edges the support of a National Research
Service Award (PHS ES 5264-02) from the .
National Institute of Environmental Health
Sciences.
LITERATURE CITED
1. Dagley, S., 1975. A biochemical .•
approach to some problems of environ-
mental pollution. Essays in Biochemis-
try 11: 81-138.
2. Atlas, R.M. and R. Bartha, 1981.
Microorganisms and some novel
pollution problems. Microbial
Ecology; Fundamentals and Applica-
tions. Addison Wesley, Reading, MA.,
3. Chapman, P.J., 1976. Microbial
degradation of halogenated compounds.
Biochemical .Society Transactions 4:
463-4&6.
4. Timmis, K.N., P.R. Lehrbach, S.
Harayama, R.H. Don, N. Mermod, S. Bas,
R. Leppik, A.J. Weightman, W. Reineke,
H.-J. Knackmuss, 1985. Analysis and
manipulations of plasmid-encoded path-
ways for the catabolism of aromatic
compounds by soil bacteria. Plasmids
in Bacteria. D.R. Helinski, S.N.
Cohen, D.B. Clewell, D.A. Jackson, and
A. Hollaender (eds). Plenum Press,
New York, pp719-739.
5. Dorn, E., M. Hellwig, W. Reineke, and
H.-J. Knackmuss, 1974. Isolation and
characterization of a 3-chlorobenzoate
degrading pseudomonad. Archives of
Microbiology 99: 61-70.
6. Chatterjee, O.K., S.T. Kellog, S.
Hamada, and A.M. Chakrabarty, 1981.
Plasmid specifying total degradation
of 3-chlorobenzoate by a modified
ortho pathway. Journal of Bacteriol-
ogy 146: 639-646.
7. Don, R.H., and J.M. Pemberton, 1981.
Properties of six pesticide degrada-
tion plasmids isolated from Alcali-
genes paradoxus and Alcaligenes
eutrophus. Journal of_ Bacteriology
145: 681-686.
8. Ghosal, D., I.S. You, O.K. Chatterjee,
and A.M. Chakrabarty, 1985. Genes
specifying degradation of 3-chloro-
benzoic acid in plasmids pAC27 and
pJP4; Proceedings £f the National
Academy of Sciences 82; 1638-1642.
9. Don, R.H., A.J. Weightman, H.-J.
Knackmuss, and K.N. Timmis, 1985.
Transposon mutagenesis and cloning
analysis of the pathways for degrada-
tion of 2,4-dichlorophenoxyacetic acid
and 3-chlorobenzoate in Alcaligenes
eutrophus JMP134 (pJP4). Journal of
Bacteriology 161: 85-90.
10. Chatterjee, O.K. and A.M. Chakrabarty,
1983. Genetic homology between inde-
pendently isolated chlorobenzoate-de-
gradative plasmids. Journal of Bacteri-
ology 153: 532-534.
11. Bollag, J.-M., G.G. Briggs, J.E.
133
-------
Dawson, and M. Alexander, 1968. 2,4-D
metabolism: enzymatic degradation of
chlorocatechols. Journal of
Agricultural and Food Chemistry 16:
829-833.
12. Evans, W.C., B.S.W. Smith, N.H.
Fernley and J.I. Davles, 1971.
Bacterial metabolism of 2,4-dichloro-
phenoxyacetate. Biochemical Journal
122: 543-551.
13. Liu, T., and P.J. Chapman, 1985.
Cloning and expression of 2,4-dichloro-
phenoxyacetate-degradative genes in
Eschericia coli and Alcaligenes
eutrophus. Abstracts of the Annual
Neeting-T985. American Society for
Microbiology, Washington, D.C., #
H-154.
14.
15.
Don, R.H. and J.M. Pemberton, 1985.
Genetic and physical map of the
2, 4-dichlorophenoxyacetic acid-degra-
dative plasmid pJP4. Journal of
Bacteriology 161: 466-468.
Kilbane, J.J., O.K. Chatterjee, J.S.
Karns, S.T. Kellog, and A.M. Chakra-
barty, 1982. Biodegradation of 2,4,5
trichlorophenoxyacetic acid by a pure
culture of Pseudomonas cepacia. Ap-
lied and Environmental Microbiology
16. Rosenberg, A., and M. Alexander,
1980. Microbial metabolism of
2,4, 5-trichlorophenoxyacetic acid in
soil, soil suspensions, and axenic
culture. Journal of Agricultural and
Food Chemistry 28: 297-302.
17. Ghosal, D., I.S. You, O.K. Chatterjee,
and A.M. Chakrabarty, 1985. Microbial
degradation of halogenated compounds.
Science 228: 135-142
18. Ghosal, D., I.S. You, O.K. Chatterjee,
and A.M. Chakrabarty, 1985. Plasmids
In Bacteria, D. Helinski, S.N. Cohen,
D. Clewell, D. Jackson, and A.
Hollaender (eds). Plenum Press, New
York, pp667-686.
19. Darzins, A., and A.M. Chakrabarty,
1984. Cloning of genes controlling
alginate biosynthesis from a mucoid
cystic fibrosis isolated of Pseudo-
monas aeruginosa. Journal of_
Bacteriology 159: 9-18.
20. Ruvkun, G.B., and F.M. Ausubel, 1981.
A general method for site-directed
mutagenesis in prokaryotes. Nature
(London) 289: 85-88.
21. Crawford, R.L., P.E. Olson, and T.D.
Frick, 1979. Catabolism of
5-chlorosalicylate by a Bacillus
isolated from the Mississippi River.
Applied and Environmental Microbiology
38: 379-384':
22. Knuutinen, J., 1984. Synthsis,
Structure Verification and Gas
Chromatographlc Determination of
Chlorinated Catechols and Guaiacols
Occuring in Spent Bleach Liquours of
Kraft Pulp Mills.Ph.D. Thesis,
University of Jyvaskyla, Jyvaskyla,
Finland, pp9-12.
23. Chatterjee, O.K., and A.M. Chakra-
, barty, 1982. Genetic Rearrangements
in plasmids specifying total degrada-
tion of chlorinated benzole acids.
Molecular and General Genetics 188:
279-285.
24. Albertini, A.M., M. Hofer, M.P. Galas,
and J.H. Miller, 1982. On the forma-
tion of spontaneous deletions: the
importance of short sequence homolo-
gies in the generation of large dele-
tions. Cell 29: 319-328.
25. Anderson, R.P., and Roth, J.R., 1977.
Tandem genetic duplications in phage
and bacteria. Annual Reviews of
Microbiology
134
-------
TECHNIQUES FOR MICROSCOPIC STUDIES OF SOLIDIFICATION TECHNOLOGIES
H. C. Eaton, M. E. Tittlebaum and F. K. Cartledge
Colleges of Engineering and Basic Sciences
Louisiana State University
Baton Rouge, LA 70803
ABSTRACT
The mechanism(s) whereby organic hazardous wastes interact with cement matrices is
important for characterizing existing and future solidification technologies. Because in
the complex chemistry of cement setting reactions there are many microscopic phases
formed, it is necessary to use microscopy and microanalytical tools in studies-of these
systems. Several methods, currently in use by the authors, are discussed. They include
scanning electron microscopy, transmission electron microscopy, selected area diffrac-
tion, energy dispersive x-ray analysis and electron probe microanalysis. Specimen pre-
paration techniques are discussed. Results from the application of the techniques to
studies of cements containing ethylene glycol and para-bromophenol are given.
INTRODUCTION
Although landfill ing'is not the
method of choice for the disposal of some
hazardous chemical wastes, it remains an
important technology today. Moreover, it
is expected to retain its place in the
overall waste management scheme of the
future. This is true for a number of
reasons. First, there are, and will be,
equally hazardous residues from environ-
mentally and socially more acceptable
processes, such as incineration. These
residues cannot be ignored. Secondly,
there are a number of hazardous wastes
which are not amenable to such disposal
techniques as incineration, physical/
chemical or biological treatment. The
disposal of these wastes may still require
land disposal techniques. Thirdly, land
treatment and disposal will remain an
important HW management technique with
regard to abandoned sites where economics
prohibit other alternatives.
For these reasons, it is important to
increase the scientific base of informa-
tion about the nature of the solidified/
stabilized complexes that are currently
being produced. It must be understood,
for example, precisely how the waste is
contained in the solidified product.
It has not been the accepted practice
for the waste disposal vendors to provide
information about the mechanism(s) whereby
their product entraps the waste. Instead,
they are required to show that there is an
acceptably low level of leaching of the
waste from the landfill site. Although
this provides a practical database for
"prediction by extrapolation" of the long-
term stability of the site, it does not
provide a sound scientific basis for this
kind of prediction. The real need is for
a complete elucidation of the mechanisms
of chemical waste entrapment and a des-
cription of the kinetic processes which
govern the leaching behavior.
PURPOSE
The present communication describes
the approach used in the authors' labora-
tory where there is an attempt to answer
these basic and important questions about
a variety of solidified waste systems. To
date, the primary focus has been on
systems which utilize Type I Portland
cement as the matrix. This commonly used
base material has been widely assumed to
provide a product with good longterm
135
-------
stability. It is important to recognize,
however, that the chemical processes
involved in the setting of hydrated Port-
land cement are many and complex. There
is a multiplicity of mineral phases formed
and the formation of this composite of
minerals can be further complicated by the
presence of organic and inorganic
additions.
APPROACH
In the sections which follow, several
microscopic and microanalytical techniques
are described. Each of these are
currently being used to determine the way
in which two different model wastes are
entrapped. The specific model wastes were
chosen because (!) they are chemically
different, (2) one of them is on the list
of hazardous materials, and (3) one of
them is amenable to analysis by one of the
important microanalytical tools to be
described in the text. Organic chemicals
were chosen because they represent a
special problem to the solidification
Industry relating to the well documented
"interference" with the hydration
processes occurring in the setting of
Portland cement. The two organics were
ethylene glycol (EG) and para-bromophenol
(PBP).
RESULTS
Material Preparation
As discussed, the objective of the
study is to identify the kinds of bonding
present in stabilized organics. An addi-
tional goal is to quantify the bonding
potential of stabilizing agents towards
organics using extraction procedures.
Type I Portland cement was solidified with
pure EG, a water soluble organic, and
water. Samples were prepared by weighing
EG into a 20 mL borosilicate glass, screw
cap scintillatcr vial. Portland cement
and water were added and the mixture was
stirred to homogeniety with a glass
stirring rod. Samples were prepared by
mixing 0.4, 1 or 2 gm of EG with 10 gm of
cement and 4 mL of water. They were
allowed to cure for variable lengths of
time. Extractions were carried out with
solvents of varying polarity. They
included dichloromethane (DCM) with a
polarity index of 3.1, dimethyl sulfoxide
(DMSO) with a polarity index of 7.2, and
water with a polarity index of 10.2.
After the appropriate curing time had
elapsed, the sample vials were broken and
the contents were ground to pass a 100
mesh sieve. The powder was transferred to
a 125 mL Erlenmeyer flask, mixed with 120
mL of solvent and shaken for 0.5 hour.
The mixture was filtered under suction and
analyzed by gas chromotography using an HP
5790A Gas Chromatograph coupled to an HP
3390A Integrator. The results of a one
month old EG sample is shown in Table 1.
A similar method was used to prepare
samples containing PBP and to extract
using the same solvents.
The extractions provide information
about the percent recovery of EG as a
Table 1. THE CONCENTRATION DEPENDENCE IN THE
EXTRACTION OF ETHYLENE GLYCOL
Grams of EG per 10.0
Grams of Cement
Percent Recovery in
a Single Extraction
0.4
1.0
2.0
DCM
DMSO
(pH=9.9)
3.4
6.8
7.1
11.0
15.3
43.4
78.9
83.3
83.4
136
-------
function of solvent polarity, hydrogen
bonding ability, concentration of organic
in the curing cement and the characteri-
zation of the entrapment mechanics once
the particular phases are identified. In
addition, the data provides a macroscopic
clue to the concentrations where important
chemical processes may be occurring. This
is a valuable aid to selecting a time of
cure. The data is very important in
the selection of specimens which should be
examined by the microscopic and micro-
analytical methods.
Scanning Electron Microscopy
The scanning electron microscope has
become one of the most valuable micro-
scopic tools for the characterization of
engineering materials. It reveals the
microscopic details of the specimen
surface with magnifications easily up to
several tens of thousands of diameters.
In addition, the large depth of field
allows an image of the surface which
displays the large excursions that are
often possible in fractured surfaces. The
electron beam that is rastered over the
specimen surface is usually produced by
thermionic emission of electrons from a
heated tungsten filament.. These electrons
are focused and then rastered by electro-
magnetic lenses that are central to the
image formation system. When these
primary electrons strike the specimen,
secondary electrons are emitted and then
detected by a scintillator within the
vacuum chamber. The light from the
scintillator is detected by a photomulti-
plier tube mounted on the air side of the
chamber but in communication with the
scintillator by means of a light tube.
These signals are used to modulate the
raster of a cathode ray tube monitor.
Since the raster of the monitor is
synchronized with the raster of the
primary electron beam, the image displayed
on the monitor contains information about
the microscopic point-to-point differences
in the secondary electron emission from
the specimen surface. These differences
are a function of the topography largely
and to a minor extent on other things,
e.g. atomic number. Therefore, the image
reveals the desired information about the
microscopic variations in the surface
topography.
Fracture Surfaces
Samples of the solidified waste
systems were fractured at liquid nitrogen
temperatures in order to reveal the nature
of the interior. Liquid nitrogen tempera-
tures were chosen because of the plastic
deformation and tearing that has been
observed to occur in these materials at
room temperature (2).
A small cube of material, approxi-
mately 1-3 mm on edge, was selected and
then mounted on an aluminum specimen
holder with double-sided conducting tape.
The entire assembly was then coated with
AuPd in a sputter coater or with carbon in
a vacuum evaporator. The specimen was
then examined in an ISI 60-A Scanning
Electron Microscope operated at no more
than 15 keV. These low operating voltages
were preferred in order to avoid extensive
surface penetration. Previous studies of
cement structure (1) have failed to
realize the importance of this parameter,
however. Figures 1 and 2 illustrate well
that at the lower voltages more surface
detail is revealed.
An important observation of the
authors' studies is that the structure of
the cement/waste system remains similar to
that of the pure cement (2). This dis-
covery greatly aids our investigation
since there is a large amount of litera-
ture about the structure of hydrated
Portland cement. It should be noted,
however, that micrographs should be
interpreted with extreme caution because
of the widespread controversy over the
structure of the pure cement itself.
Polished Surfaces
The examination of polished surfaces
allows yet another type of information
about the nature of the microstructure to
be obtained. Studies of polished surfaces
allow the potential use of energy disper-
sive x-ray analysis for the determination
of the microchemistry of the waste con-
taining solids. In addition, the polished
surfaces appeared to provide views of the
material that are much easier to interpret
than the fractured surfaces.
Each specimen was impregnated with a
low viscosity, hard grade acrylic resin in
order to increase the mechanical integrity
and stability of the sample prior to
137
-------
grinding and polishing. Polymerization of
the resin was carried out in vacuo. When
polishing and grinding was not preceeded by
acrylic impregnation, a grossly distorted
specimen surface resulted, therefore it
was concluded that there is some degree of
acrylic penetration into the specimen pore
spaces. Grinding was performed with sand-
paper using grit sizes 250, 340, 400, and
600, followed by polishing on a rotating
wheel which was impregnated with 6 micron
diamond paste. This was followed by a 0.5
micron diamond wheel, and finally on a
wheel containing 0.05 micron alumina. In
all of the grinding and polishing opera-
tions isopropanol was used, rather than
water, in order to cool and lubricate
without the removal of microphases.
Previously, methanol and dichloromethane
were used but with less desirable results.
A carbon coating was applied to the
polished surfaces and they were then
examined in the SEM and some analyzed by
EDX. Figure 3 is an example of a
micrograph of a polished specimen surface.
Our preliminary analyses suggest that the
matrix phase consists of CSH (calcium
silicate hydrate) gel phase and that the
large inclusions (indicated by the arrow)
are calcium aluminosilicates.
Electron Microprobe Microanalysis
The cement samples were analyzed
using an ARL-EMX electron probe micro-
analyzer (EPMA). The probe is equipped
with three wavelength dispersive spectro-
meters and a Tracer-Northern Energy
Dispersive Spectrometer system. It is
partially automated and simultaneous
analysis of several major and trace
elements is possible. The electron gun is
operated at 15 keV at 150 microamperes
curgent. The vacuum is maintained in the
10~ torr range. Detection-and measure-
ment of x-rays follows Bragg's law. The
sample surface is highly polished and
carbon coated to make it conductive. The
take-off angle of the x-rays is 52.5°,
where take-off angle is defined as the
angle that the detector crystal makes with
the horizontal surface of the specimen.
Different crystals having different
d-spacings allow the analysis of elements
ranging from C to U and most elements can
be analyzed with concentrations in the
parts per million (ppm) level.
The EPMA method is a comparative
method, like many other spectroscopies,
therefore its accuracy is only as good as
the purity of the standards used. The
ratio of the x-ray.intensity generated for
any element in the unknown sample compared
to the intensity of the element in the
known is called the k-ratio and, to a
first approximation, is equivalent to the
concentration of the element in the
unknown. It is not exactly equal to the
concentration because the x-ray intensity
is modified by its interaction with the
atoms in the sample. The theory of micro-
probe analysis is quite well developed now
and basically two methods of correction of
data are used to obtain accurate concen-
tration information. One is an empirical
method and the other theoretical. The
theoretical procedure requires the calcu-
lation of concentration from first princi-
ples and is known as the ZAP method ("Z"
is from atomic number, "A" from adsorp-
tion, and "F" from fluorescence).
The specimen current, the current
which flows during analysis when the
specimen is earthed, for the analyses was
set to 10 nA for Bentonite. All analyses,
including standardization, were carried
out under identical conditions. Seven
elements were analyzed using mixed
standards, e.g. Si02 for Si. Oxygen was
calculated by difference assuming that the
components should sum to 100%. Figure 4
is an example of the form of the data
obtained from EDX and Table II illustrates
the quantitative information that can be
obtained using EPMA.
Energy dispersive x-ray analysis can
also be performed in the SEM. The advan-
tage of this type analysis is the ability
to image very small objects prior to their
analysis by EDX. The EPMA, on the other
hand, is not a high spatial resolution
instrument (an optical microscope is used
to locate the region on the specimen
surface which is to be analyzed). There-
fore, there are times when very small
phases cannot be chemically analyzed in
the EPMA but can be analyzed in the SEM.
Transmission Electron Microscopy
The transmission electron microscope
(TEM) has been an effective tool in. the
study of the microstructure of ceramic and
composite materials. The main advantage,of
the TEM is the high spatial- resolution, that
is possible in modern instruments ( *> 0.5
nm). In addition, the selected area
138
-------
Table 2. Cement Analysis by EPMA
Element
Si
Ca
Al
Fe
K
Mg
S
0
Standard
Quartz
(Si02)
Diopside
(CaMgSi206)
Corundum
(A1203)
Fayalite
((Fe,Mg)2Si04)
Microcline
((K.Na)AlSi, OJ
-------
SUMMARY
Several techniques have been
described. These advanced techniques can
be used to characterize the microcrystal-
line and amorphous phases that are formed
during the processes attendant to the
solidification/stabilization of organic
hazardous wastes. The methods can be used
for a host of other environmental studies
where it is important to understand the
mechanisms whereby complex materials
systems are formed, e.g. in the formation
of incineration residues.
The application of these methods to
the solidification of organic wastes has
indicated that the waste is not contained
in the solidified mass in a homogeneous
way, even when the constituents are mixed
thoroughly during the preparation process.
The inhomogenities consist of microscopi-
cally concentrated regions of waste.
Since cements, and other setting
materials, undergo complex chemical
changes over rather long periods of time
(sometimes months) it will be important to
determine the time dependent transport
paths of the wastes as the system ages.
ACKNOWLEDGEMENTS
The authors would like to acknowledge
support for this work from the United
States Environmental Protection Agency
through the Hazardous Waste Research
Center at Louisiana State University,
Environment Canada and Ontario Environ-
ment. In addition, many individuals have
contributed to the work and our under-
standing of cement and waste solidifica-
tion. They include M. B. Walsh, D.
Chalasani, D. Skipper, A. C. Chou, A. Roy,
P. Malone, T. Bridle and C. Wiles.
REFERENCES
1. Diamond, S., 1972. Indentification of
Hydrated Cement Constituents Using a
Scanning Electron Microscopy Energy
Dispersive X-Ray Spectroscopy, Cement
and Concrete Research 2, pp617-632.
2. Eaton, H. C., M. B. Walsh, M. E.
Tittlebaum, F. K. Cart!edge and D.
Chalasani, 1985. Microscopic Char-
acterization of the Solidification/
Stabilization of Organic Hazardous
Wastes, Energy-Sources and Technology
Conference and Exhibition, American
Society of Mechanical Engineers,
Dallas, Texas, (ASME Paper No. 85-
Pet-4).
3. Thomas, G., 1962. Transmission
Electron Microscopy of Metals. John
Wiley and Sons, Inc., New York,
pp!38-139.
Figure 1. A scanning electron micrograph
of a pure cement sample aged for six
months. The electron accelerating
voltage was 15 keV.
Figure 2. A scanning electron micrograph
of the same specimen and region shown
in Figure 1. The accelerating voltage
was 5 keV and less penetration is evident.
140
-------
},,iV-%
•,4fe
«.f£3Ks?
<4^*
&%&»'*.>•;*
g ^73^3%5,*§^;|JS
«>—- "^ '
^/^' -,v>»4^^«^-rf3^%K«l«^sa&ja=^,--«i'--v,..^a~ ,-i-T
a»Jg^?*^v..>....?.gKfe. a?
- '"*'*•""-»• i3»,i'!Sftrfit^«^ C ,J JKQ5ry>gfj#s ^&j&&»f;&gfyfaim^ •* 4»^" "* * * * i'viiiS*-"^- ,*'™3-«*' *w "^vis^S *JB
5
Figure 3. A scanning electron micrograph of a
six month old, pure cement sample.
to
IOK
5000
o
O.O 1.0 2.0 3.O 4.O 5.0 6.O 7.0 8.O 9.O IO.O
ENERGY(keV)
Figure 4. An EDX spectrum from an ethylene glycol specimen.
141
-------
Specimen
Plastic Film
/Hiastic Him .Carbon Film
'Metal Shadow
Replica
Figure 5. A schematic representation of the procedure
used to produce TEM replicas.
(b)
Figure 6. (a) A transmission electron micrograph of a replica containing
material removed from a polished surface of a cement containing PBP.
(b) The diffraction pattern produced by the material shown in (a).
142
-------
USEPA COMBUSTION RESEARCH FACILITY
PERMIT COMPLIANCE TEST BURN
Richard A. Carnes
USEPA CombustlonResearch Facility
Jefferson, Arkansas 72079
INTRODUCTION
In August of 1982, when the USEPA
officially accepted title to the
Combustion Research Facility (CRF) 1t
was unclear as to the exact permit
structure that would be required. At
that time 1t was proposed that the
existing air permit, Issued by the State
of Arkansas Pollution Control
Commission, would be sufficient for
research purposes. The concept being
that the CRF was not a commercial
operation, 1t was not a 24 hour a day
operation, 1t would conduct experiments
on controlled volumes of waste material.
USEPA Region VI officials finally
decided that 1n order to conduct
research on Resource Conservation and
Recovery Act (RCRA) Appendix VIII listed
material the CRF would have to have a
complete Part B Incinerator permit. The
final decision was made 1n early 1983
and set 1n motion the preparation of the
necessary documents for a State
Hazardous Waste Permit as authorized by
the USEPA under RCRA regulations.
The completed application was
officially received by the Arkansas
Department of Pollution Control and
Ecology on August 8, 1983, and set 1n
motion the following prescribed schedule
for processing 1t:
TASK
Completeness Determination
Technical Review
Preparation of Draft Permit
Public Notice Issuance
Public Hearing
End of Public Comment Period
Final Permit Prepared
Final Permit Presented to
Commission
DATE
9-8-83
2-7-84
2-7-84
2-14-84
3-29-84
4-11-84
5-11-84
5-25-84
1.
2.
3.
4.
5.
6.
7.
8.
Following several short
the final Dart B permit was officially
issued on. July 27, 1984, by the
State of Arkansas Commission on
Pollution Control and Ecology. A
specified requirement of the permit 1s
that a compliance test burn, using
prescribed chemicals, be conducted to
determine system efficiency.
COMPLIANCE TEST BURN PROTOCOL
A major element of the Combustion
Research Facility (CRF) Incineration
system Is the newly Installed carbon
bed/HEPA filter system. The carbon
(some 4500 Ibs) 1s reclaimed charcoal
and so must be described as "dirty"
from the point of view of the sampling
and analytical procedures to be
employed In the compliance tests. In
addition, the Installation now
Includes a considerable run of new
fiber reinforced plastic pipe (FRP)
that 1s also expected to be a copious
source of background organlcs (for
example, the methyl ethyl ketone used
143
-------
as a solvent 1n cementing sections of
the FRP). These reasons make the
cleanup or removal of these background
materials a first requirement of the
compliance test.
The synthetic "waste" that was
used to challenge the Incinerator 1s a
mixture of carbon tetrachlorlde,
monochlorobenzene, trlchloroethylene
and toluene. This mixture,
characterized as EPA Soup One, has
been the subject of a rather extensive
series of Incineration tests at the
CRF during the period
1 February, 1984 through 15 June,
1984. These tests were conducted
using very small feed rates and a
conservative research approach so that
no discharge to the environment
occurred. To that end the experiments
with "Soup 1" were a success, but as
we shall see the compliance test burn
used much larger feed rates and was
designed to study breakthrough of the
existing scrubber but not the
redundant device.
The compounds that make up the
Soup are all low-bo1Hng point
materials which are therefore suitable
candidates for the use of the volatile
organic sampling train (VOST) and for
subsequent analysis by
thermal-desportlon/purge and trap
analysis. The procedures are 1n place
and the QA/QC Program at the CRF has
been reviewed and accepted by
Independent audit groups.
Test Materials
The composition of Soup One
chemicals are tabulated 1n Table 1.
Sampling and Analysis
The sampling procedures, using VOST
techniques, are essentially those used
by others In the field. Analyses of the
exposed Tenax Cartridges was carried
out using established purge and trap
techniques. Each cartridge was spiked
prior to sampling; failure to recover
the spike at better than 70% resulted 1n
rejection of the sample. The analyses
was conducted under the QA/QC procedures
established for the CRF.
Sampling of the blowdown water and
discharge consideration was conducted
for POHC analysis.
Partlculate emission, using the full
EPA method five train with 1sok1net1c
sampling was conducted during experiment
#2 (Table 2).
A sheet outlining a summary of the
compliance test burn follows and shows
plctorlally the sampling points and the
parameters that were analyzed.
Source of POHC
The chemicals for the feed stock
were obtained from the Aldrlch Chemical
Company from the same batches that were
used 1n the work previously conducted at
the CRF. These compounds have been
shown to be void of Impurities that
Interfere with quantltatlon of the
primary compounds.
TABLE 1. SOUP COMPOSITION
COMPOUND
FEED RATE
(kq/hr)
FEED RATE
(gm/hr)
CHLORINE FEED
(qm/hr)
CC14
C2HC13
C7H8
TOTALS
3.84
3.49
1.05
1.69
10.07
64.0
58.2
17.5
28.2
167.8
3.54X103
2.83x1O3
5.33x162
6.90xl03
144
-------
Feed Stocks
Since the purpose of the trial burn
was to determine that the carbon bed can
Insure compliance even when the
Incinerator might Itself be seriously
out of compliance, 1t was proposed that
the test burn be limited to a test at a
high feed rate and at minimal
temperature. Hence, the feed stock will
be made up as follows: to one liter of
C7H8 will be added 1.56 1 of
C6H5CL plus 2.75 1 of CC14 plus
2.95 1 of C2HC13. The resulting
solutions should have a heating value of
14.7 KBtu/1, a density of 1.37 kg/1 and
contain approximately 71% of chlorine.
This solution was feed at the rate of 10
kg/hr resulting 1n a heat release of 107
KBtu/hr.
conducted as Indicated 1n the Protocol
with several operational changes. The
preburn, background, and the post burn
samples were taken at the same
temperatures as for the actual Soup
challenge. In place of the proposed
three sets of VOST samples, a total of
six sets , one 1n the prefeed Interval,
were taken for the Input/output of the
carbon bed filter during the trial
burn.
One additional change was made 1n
the Protocol: namely, the THC
measurements were not made. The fact
that the combustion gases were saturated
(relative humidity of 100%) at the
required sampling sites required
excessive pre-conditioning. This has
not yet been successfully accomplished.
Test Schedule
The experimental schedule 1s
outlined 1n Table 2.
The results of this test series
showed that the CRF rotary kiln
Incinerator meets or exceeds the permit
requirements on all counts:
TABLE 2. TRIAL BURN SEQUENCE
Experiment
1
2
3
Feed Stock
Blank
Soup
Blank
Feed Rate
(kg/hr)
0
10.07
0
TK
(°F)
995
1150
995
TA
(°F)
1400
1400
1400
Sampling Point*
2+3
2+3
2+3
Note
(a)
(b)
Notes from Table 2 are designed to give considerably more Information as
follows:
Note (a): sample background run to establish baseline.
Note (b): feed to begin one (1) hour after temperature steady state 1s
established. One (1) hour after feed start, VOST samples
at 2 and 3. This will be repeated for a total of six (6)
complete samples. After samples taken - shutdown.
RESULTS AND DISCUSSION
During the week of 7 January, 1985
the test burn required by RCRA was
conducted at the USEPA Combustion
Research Facility. The tests were
the ORE exceeded 99.99% for all
POHC' s
the HC1 removal resulted 1n
emissions of less than 0.5 kg/hr.
the residual partlculate matter
was significantly below that
required by the regulations.
145
-------
The only unsatisfactory aspect of
the Trial Burn was the behavior of the
pH control system. The design was
supposed to be capable of sustained
control at chlorine levels up to 100
lb/hr. Actually, the system experienced
wild swings from high to very low pH
within periods of a few seconds even
though the chlorine feed was of the
order of only 20% of the design
maximum. Careful checks of the
SUMMARY OF EXPERIMENTS
EXPERIMENT! RCRA Trial Burn
OBJECTIVE; To Demonstrate Compliance
of CRF Incinerator System
DURATION: One Week
EXPERIMENTAL MATPTX-
Feed: CC14; C6H5C1;
C2HCl3and C7H
[70X chlorine
Feed Rate: 10 kg/hr
Tk1ln: 1000°F
1400°F
EXPERIMENTAL SYSTEM
r
CD
Combustor
APC1
elements of the control system was
operating properly. With the
given chlorine feed, there 1s no way for
the pH of no less than 120 gallons of
water to experience a change of from 10
to less than 1 1n a few seconds. The
clear Indication of these observations
was a failure of the mixing action
Examination of the recycle system showed
that the downer fromthe tower extended
Into the recycle reservoir to a position
near the Intake port for the recycle
pump. This arrangement, shown
schematically 1n Figure 1, apparently
resulted 1n short circuiting the
Intended mixing action. The water
returned to the reservoir was, at least
partly, derived from the ventuM
return. This latter solution can be
expected to be strongly add so that the
resulting short circuited fluid would,
itself, be of very low pH. Since the pH
sensor was Installed 1n a by-pass
Immediately adjacent to the recycle pump
output, the wild swings of pH observed
simply reflected those of the short
circuited solution.
This situation has been corrected by
shortening that portion of the downer to
approximately 8 Inches below the cover
of the reservoir. The open end has been
fitted with a dlffuser plate to markedly
/ Carbon \
X Filter/
T
(2)
HEPA
Filter
SAMPLING & ANALYSIS
Parameter
Sampling
.Point
(l)
(2)
(3)
POHC
Feed
(CONC1
•
/
^
Fuel
Feed
s
Air Temp.
Feed
•
V
,/
02
/
./•
CO
C02 Vol
flow
HC1 THC p* RH+
* p a pressure
+ RH * relative humidity
caustic feed pump, the pH sensor and the
controller Indicated that each of the
Increase mixing within the reservoir.
The modified system has been challenged
with the Injection of HC1 solution Into
146
-------
the kAIn with the chambers operating at
the temperatures used during the test
burn. The changes have been shown to
markedly Increase the pH control so that
the operation 1s now acceptable and a
Removal Efficiency (RE) 1n excess of the
required 99.5% was achieved.
Tower
Caustic
tiL
'QL
N
21
—
^
y
Recycle
Reservoir
Recycle Pump
FIGURE 1. SCHEMATIC OF RECYCLE RESERVOIR
As will be shown below, 1n spite of
the control problems, the HC1 emissions
were, 1n fact, adequately controlled.
Two measurements were made of the
HC1 emission rate at the Input to the
carbon bed; one of 60 minutes duration,
the other of 326 minutes duration. The
resulting data are shown 1n Table 3.
At the feed rate Indicated in Table
1, the chlorine feed was 6.90xl03
gm/hr which, on conversion to HC1,
represented an HC1 feed rate to the
scrubber of 7.094xl03 gm/hr. Thus,
the removal efficiency was, 1n the two
cases found to be
RE = 99.45%
RE = 99.34%
and the total emission rate was found to
be well below the permissible 500
gm/hr.
Partlculate Loading at the Stack
The results of three Method. 5 scans
of the stack during the Test Burn are
summarized 1n Table 4.
The measured partrlculate loading
Is well within the compliance
requirements, the latter being a
particle loading of not more than 180
mg/OSCM.
POHC Emission Levels and ORE
The measured emission rates for the
four POHCs along with the corresponding
TABLE 3. IONIC CHLORINE OUTPUT RATE
SAMPLE
EOT 0911 44®
E01091254
Sample
Flow Rate
n/m1n)
1.2
0.96
Sample
Time
(m1n)
60
326
Sample
Total
M)
72
313
Chloride
Total
(mq)
2.04
10.55
Gas
Cone.
(ma/1)
0.028
0.0337
Duct Flow
Rate
(DSCMM)
23.3
23.3
HC1
Emission Rate
(am/hr)
39.1
47.1
TABLE 4. PARTICULATE LOADING OF STACK
Sample Degree of
Isok1net1c1ty
m
S01091221#
SOI 091 532
SOI 091 737
84.39
102.28
89.54
Total Partlculate Sample Vol
Collected
(mq) DSCM
25.9
29.9
13.7
1.127
1.042
0.894
Loading
(mq/DSCM)
50.14
62.61
33.44
faralns/DSCF)
0.022
0.027
0.046
* corrected to 1254 C02. The regulatory maximum for particle loading 1s
180 mg/DSCM = 0.08 gralns/DSCF
© Sample taken at entrace to carbon bed, point number 2 on Summary
Sheet.
# Sample taken at the stack, point number 3 on Summary Sheet.
147
-------
ORE'S are tabulated 1n Tables 5 through
8 which follow. It 1s noted that 1n all
cases the required ORE of 99.99% was
attained.
TABLE 5. EXPERIMENTAL MEASUREMENTS OF POHC's - CARBON TETRACHLORIOE
[Feed Rate of 64.0 gm/m1n]
E01091102
SOI 0911 02
EOT 091 203
SOI 091 205
E01091300
SOI 091 300
EOT 091 400
SOI 091 400
EOT 091 500
SOI 091 500
Sample
Vol
(DSL)
21.57
20.83
22.91
21.89
22.29
20.93
20.48
21.09
23.96
21.95
Sample
Collected
-------
TABLE 1. EXPERIMENTAL MEASUREMENTS OF POHC's - TOLUENE
[Feed Rate =17.5 gm/m1n]
Sample Number
EOT 091 102
SOI 091 102
E01091203
SOI 091 205
£01 091 300
SOI 091 300
E01091400
SOI 090400
E01091500
SOI 091 500
Sample Vol.
(DSL)
21.57
20.83
22.91
21.89
22.29
20.93
20.48
21.09
23.96
21.95
Sample
Collected
(ua)
0.108
0.409
0.081
0.328
0.078
0.337
0.074
0.332
0.056
0.327
Duct Flow
Rate
(DSMM)
25.75
25.06
23.34
24.19
23.34
23.85
24.16
23.85
23.79
24.81
Emission Rate
(ug/m1n)
128
494.5
82.5
362.2
81.6
384.5
87.2
377
55.5
369
ORE
(%)
99.999
99.997
99.999
99.998
99.999
99.997
99.999
99.997
99.999
99.997
TABLE 8. EXPERIMENTAL MEASUREMENTS OF POHC's
[Feed Rate =28.2 gm/mln]
- MONOCHLOROBENZENE
Sample Number
E01091102
SOI 0911 02
EOT 091 203
SOI 091 205
E01091300
SOI 091 300
EOT 091 400
SOI 091 400
EOT 091 500
SOI 091 500
Sample Vol.
(DSL)
21.57
20.83
22.91
21.89
22.29
20.93
20.48
21.09
23.96
21.95
Sample
Collected
(ua)
.002
0.011
0.010
0.010
0.017
0.017
0.0009
0.013
0.010
0.008
Duct Flow
Rate
( DSMM)
25.75
25.06
23.34
24.19
23.34
23.85
24.16
23.85
23.79
24.81
Emission Rate
(uq/m1n)
2.4
13.3
10.2
11.0
17.8
19.4
10.6
14.8
9.9
9.0
ORE
(%)
99.99999
99.9999
99.9999
99.9999
99.9999
99.9999
99.9999
99.9999
99.9999
99.9999
Discussion of Effects of Redundant
Scrubber
A comparison of the E and S samples
shows that, except for CC14, the S
emission rates are either the same as
the E rates or represent greater rates.
It appears that the bed has only minimal
effect on this group of compounds. It
1s known that high temperature and.water
adversely affect the carbon adsorption
Isotherms so that perhaps this effect 1s
the origin of the observation. In*
addition the differences are very small
and might simply be the result of
statistical sampling and analytical
errors.
In conclusion the results show that
based on an Input/output model the CRF
rotary kiln system achieved the required
99.99% ORE for all of the target POHCs.
After minor modifications to the pH
control recycle reservoir system the
removal 1n excess of the required 99.5%
HC1 was achieved.
The detailed technical report,
Including appendices, 1s In the final
stages of review In preparation for the
required submission to Arkansas permit
officials as part of the compliance test
requirements.
ACKNOWLEDGEMENT
The author wishes to acknowledge the
dedicated engineers, chemists,
technicians and secretaries of Versar
for their contributions during the trial
burn.
149
-------
ENGINEERING ANALYSIS OF HAZARDOUS WASTE INCINERATION-
FAILURE MODE ANALYSIS FOR TWO PILOT SCALE INCINERATORS
W. D. Clark, J. F. La Fond, D. K. Moyeda, W. F. Richter, W. R. Seeker
Energy and Environmental Research Corporation
Irvine, California 92718
and C.C. Lee
EPA Hazardous Waste Engineering Research Laboratory
Cincinnati, Ohio 45268
ABSTRACT
An engineering analysis procedure for prediction of hazardous waste destruction has
been applied to two pilot scale incinerators: the Control Temperature Tower (CTT) and the
Combustion Research Facility (CRF). The CTT was analyzed in backheated, insulated, and
cooled configurations using two dimensional heat transfer and flow models under plug flow
conditions with a turbulent core and boundary layers along the wall. First order global
kinetic rates were utilized to calculate waste destruction efficiencies. For the CRF kiln
and afterburner, residence time distributions were obtained from tracer gas measurements
taken on an isothermal flow model; and temperatures were obtained from a single zone heat
transfer model of each component. Potential failure modes were investigated including:
cooling, change of waste type, poor atomization, low load, high excess air and flameout.
INTRODUCTION
There is increased public awareness
and concern associated with the disposal
of hazardous waste materials. When
operated properly, incineration is the
quickest and most effective method for
destroying hazardous wastes (7).
However, there are a number of risks
associated with incineration which must
be addressed before an incine'rator can be
put into operation. The pubTic must be
assured that the facility can be operated
safely. The incinerator must be designed
and built to meet government regulations
requiring that hazardous compounds be
destroyed and removed with an efficiency
greater than 99.99 percent and must prove
that these regulations can be met based
on costly trial burns or other appro-
priate data. An envelope of safe
operating conditions must be defined, and
the impact of equipment failure or
operator error must be assessed to deter-
mine what precautions are necessary to
avoid incinerator failure.
A reliable assessment methodology
allowing scaling and extrapolation of
pilot scale and trial burn data could
substantially decrease the expense and
risk of defining the limits of safe
operation for a particular incinerator.
Such a methodology is under development
at Energy and Environmental Research
(EER). The state of the art of combus-
tion modeling with respect to hazardous
waste incineration has been reviewed (1)
and six critical modeling areas have been
identified: flow, heat transfer, mixing,
injection, tracking, and- kinetics. A
composite engineering analysis procedure,
described in detail by Clark, et al. (1),
has been developed incorporating sub-
models for each of the six critical
areas, although some of the submodels are
quite simplistic in their current, pre-
liminary form. The procedure has been
demonstrated to successfully predict
trends of hazardous waste destruction in
a pilot scale incinerator (2)'.
150
-------
This paper describes the application
of the engineering analysis procedure to
assist in the planning and interpretation
of experiments and to identify potential
failure modes for two EPA pilot scale
incinerators, the Control Temperature
Tower (CTT) and the Combustion Research
Facility (CRF).
CTT
The CTT, built by EER, described by
Overmoe, et al. (4), and operated by the
Hazardous Waste Engineering Research
Laboratory of EPA, is shown schematically
in Figure 1. The downfired pilot scale
incinerator is designed to operate in
backfired, insulated, or cooled modes or
in combinations of the three. The
incinerator has an inside diameter (ID)
of 0.254 tn and a total length of 2.08 m.
It is insulated to an outside diameter
(OD) of 1.02 m with layers of refractory
designed to withstand high temperatures
and provide maximum insulation. The CTT
is divided into three sections. The top
section includes a 45 degree transition
from an ID of 0.048 m at"the burner to
0.254 m. The middle section has a
slightly enlarged ID to make room for
cooling coils which are inserted in the
cooled mode of operation. The bottom
section has channels in the refractory to
allow auxiliary backheating without
contact between the main flow and the
auxiliary flame.
The engineering analysis procedure
was applied to the CTT to analyze and
extrapolate the results of the first test
burn in late 1983. Data and operating
conditions were reported by L. Staley of
EPA in personal communication with the
authors. The main burner was operated
with 16,500 J/s (56,300 Btu/hr) propane
at 140 percent theoretical air. The
flame swirl number of 1.13 resulted in a
short, squat flame which completely
filled the cavity. Each of the two back-
fired channels was operated with a heat
input of 10,100 J/s (34,500 Btu/hr). Gas
temperature was measured with a
thermocouple at a single axial position
near the end of the backheated .zone.
Plug Flow Core
Swirl
Burner
— Primary Flame
Zone
- Cooling
- Backheating
-I
0)
sz.
i-
O)
1
j\ Boundary
u v j \Layer
•4
••
If
t T
- -4
lf
*•
»•
•9.
<4
Overlayed
Turbulent
Exchange
»•
Cooling
»-
03
•3.
^^^
Backheating
Figure 1. Schematics of CTT reactor and model.
151
-------
Centerllne temperature was measured to be
1161°K and the temperature did not change
with radial position until within the
boundary layer beginning 0.0254 to 0.0508
m (1 to 2 in) from the wall where the
temperature dropped from 1161°K at the
edge of the boundary layer to 1126°K at
the wall. The true center!ine temper-
ature was estimated to be 1307°K from a
heat balance on the thermocouple equating
heat gained from convection between the
gas and the thermocouple to heat lost
from radiation between the thermocouple
and the wall. This temperature correc-
tion was approximate at best due to
uncertainties in the thermocouple
emissivity and in the wall temperature.
To simplify the modeling effort, the
incinerator was approximated as a cylin-
der with a constant ID of 0.254 m. Ten
axial by five radial zones were used in
the two dimensional model as shown in
Figure 1. Plug flow with overlayed
turbulence was assumed for the 0.203 m
diameter core of the incinerator, modeled
as three radial zones. The outer two
zones were modeled as a laminar boundary
layer with lower velocities and lower
interzonal exchange based on diffusion
rather than turbulence. Recirculation
was not included because the flame was
observed to fill the entire burner cone.
Heat release was calculated to coincide
with visual estimates of the flame
length. Gas temperatures in each zone
were calculated using a two-dimensional
version of Richter's semi stochastic zone
model described in detail in references 2
and 5. For backheated sections, effec-
tive wall temperatures were calculated
based on heat balance averages of the
insulated wall temperature and the hot
wall temperature. The conductivity of
the insulation was calculated from
manufacturer specifications.
Results of the heat transfer calcu-
lations are shown in Figure 2. No
adjustments from the design specifi-
cations were necessary to achieve
excellent agreement between predicted and
measured temperature for the backheated
case; however, since corroboration data
were available at only a single axial
position, predicted temperatures
(especially flame temperatures) should be
considered only approximate. Neverthe-
less, estimation of the relative effects
Measured
-r- Insulated-
Cooled
800
Axial Distance (m)
Figure 2. CTT centerline temperature
predictions for three modes
of operation.
of changing operating conditions should
be accurate.
The heat transfer model was used to
estimate the effects of changing the the
heating/cooling configuration of the CTT.
Three modes of operation were investi-
gated: the previously described
backheated case, an insulated case under
the same operating conditions except that
no backheating was used, and a cooled
case where under the same operating
conditions except that no backheating was
used and 0.55 m2 of the inside wall was
covered with cooling coils removing about
1180 J/s (4020 Btu/hr) or seven percent
of the heat input. Only the backheated
case was run in the actual experiment.
Due to the short flame, temperatures were
predicted to peak close to the burner and
decline rapidly early in the post-flame
zone. For the backheated case, the
predicted temperature leveled off around
1300°K as heat loss to the walls was
balanced by heat addition from the
auxiliary flame. Predictions for the
insulated case were similar to the
backheated case close to the burner (far
from the backheating) where there was a
difference of only about 30°K in the peak
temperatures. However, unlike the
backheated case, predicted temperatures
for the insulated case continued to
decline throughout the incinerator,
exiting almost 350°K cooler than the
backheated exit temperature. Temperature
predictions for the cooled case fairly
closely paralleled those for the
152
-------
fnsufated case: peak temperature was
75°K cooler and exit temperature was
115°K cooler.
Waste destruction was calculated
using first order global post flame
kinetic rates reported by the University
of Dayton Research Institute (UDRI) (6)
according the the equation:
d[C]/dt = -[C]A exp (-E/RT)
where [C] is the concentration of the
waste compound, t is time, A and E are
the frequency factor and the activation
energy, R is the universal gas constant,
and T is temperature. The kinetic rates
were applied to a statistically large
number of representative temperature
histories obtained from a Lagrangian
Monte Carlo tracking model. Temperature
histories were the result of particular
paths through the incinerator, chosen at
random, but weighted by probabilities
associated with the flow field. Figure 3
shows predictions of the undestroyed
waste fraction (one minus the destruction
efficiency) for a number of compounds
ranked according to the isothermal
temperature required to achieve 99.99
percent destruction in 1 second (Tg9.9gj
I sec) as calculated from the UDRI rates.
Waste was not burned in the actual exper-
iment, so no measurements are available
for comparison. For the backheated case,
better than 99.9999 percent destruction
efficiency (10~6 fraction undestroyed)
.I.W
1
10"1
•o
-------
-a
I
4-»
CO
-------
thermocouple sheathed in 0.95 cm diameter
mullite; however, due to uncertainties in
the emissivity of the partially dirty
sheath, the wall temperatures and the
view factors, the actual gas temperature
was unknown. A heat balance on the
thermocouple sheath indicated that the
gas could easily have been 300°K hotter
than the thermocouple. An additional
290,000 J/sec propane was fired in the
afterburner with enough additional air
added to bring the overall stoichiometry
to 201 percent theoretical air. After-
burner exit temperature was measured at
1088°K in the exit duct, also with a 0.95
cm diameter mullite sheathed
thermocouple.
An isothermal flow model of the CRF
kiln/afterburner system was built to a
1:3 linear scale at EER to investigate
the effects of the recent burner modifi-
cation on kiln flow patterns. La Fond
and Moyeda (3) observed that the modified
kiln exhibited a markedly different flow
pattern from the original. Both had
highly three-dimensional swirling flows
with large recirculating eddies. But in
the original kiln, waste was injected
into a dead zone on the opposite end from
the burner and took a relatively long
time to exit; whereas, in the modified
kiln, waste was injected on the same side
of the kiln as the burner and was quickly
entrained into the main flow leading to a
relatively rapid exit.
Complete analysis of the CRF
requires utilization of the isothermal
flow model results with three dimensional
versions of the heat transfer and track-
ing models described in the previous
section. However, for the preliminary
analysis reported in this paper, the kiln
and the afterburner were each modeled
using Richter's semi stochastic heat
transfer model (5) with a single zone.
This simplistic approach, ignoring
temperature variations within the kiln or
afterburner, is certainly inadequate to
predict absolute levels of waste des-
truction, but should give fairly accurate
predictions of relative destruction,
allowing trends to be investigated and
failure modes to be identified.
Residence time distributions were ob-
tained from tracer gas measurements on
the isothermal flow model. To allow
scaling and extrapolation, the residence
times were nondimensionalized, dividing
by the ratio of the reactor volume to
total volumetric flow. Waste destruction
was calculated by averaging the results
obtained from applying UDRI rates (6) at
the temperature predicted by the heat
transfer model for a number of residence
times representative of the residence
time distribution.
Residence time distributions of
tracer gas injected into the waste end of
the kiln for the original and the modi-
fied burner configurations are compared
in Figure 7. The curves represent the
best fits to isothermal model measure-
ments, assuming the kiln could be
represented by the combination of a plug
flow reactor and a stirred reactor in
series which would respond to a step
change in tracer gas concentration
according to the following equation:
— = 1-exp
Co
ft-TpTugl
|_rstirredj
where C/C0 is the ratio of the.transient
to the steady state tracer gas concen-
tration, t is time, and is the plug
flow or stirred reactor time constant.
Response times for the injection system
and the sampling system were taken into
account using the same assumptions, and a
correction factor was applied to the
resulting time constants to account for
the slight difference in the volumetric
flow to volume ratio between the model
and the incinerator as operated on
1/26/84. For. the identical operating
1.0
0.0
02 46 8 10 12 14 16
Time (sec)
Figure 7. Kiln residence time distribu-
tions for original and modified
CRF configurations.
155
-------
O
2
u.
10
io-2
J.U
io-6
io-8
io-10
i *•!
n Modified
13 Original
99.99% .
**» •» WM» M 4MI QJ
-------
Tower excess air ratio in the after-
burner, the better insulating refractory
used in the afterburner, and the high
preheat temperature of the kiln exit gas.
Predicted temperatures decreased with
decreasing load resulting in decreasing
predicted hexachlorobenzene destruction
efficiency (increasing fraction unde-
stroyed) as shown in Figure 10. Failure
conditions (99.99 percent destruction or
10~4 fraction undestroyed) were predicted
at loads less than 300 KJ/sec (1.02 MM
Btu/hr) for the kiln alone, and 120 KJ/
sec for the afterburner alone, and at 100
KJ/sec for the system comprising the kiln
and the afterburner in series.
Effects of excess air on predicted
temperatures are shown in Figure 11.
Afterburner air flow was maintained
proportional to kiln air flow. Increas-
ing excess air resulted in decreasing
temperatures due to the dilution effect.
These decreasing temperatures coupled
1200
O)
1000-
rd
OJ
Q-
800
Afterburner
50
Figure 11.
100 150 200
Kiln Excess Air (%)
250
Effect of excess air on pre-
dicted temperatures in the CRF.
with decreasing residence times due to
increased total flow resulted in decreas-
ing predicted destruction efficiencies as
shown in Figure 12. Failure conditions
for hexachlorobenzene destruction were
10
,0
10
-2
-a
cu
OJ
-a
O
(O
i-
v-6
10-8
10'
,-10
Kiln
10
100 200 300
Kiln Load (KJ/sec)
400
Figure 10. Effect of load on predicted
hexachlorobenzene destruc-
tion in the CRF.
10-10
50 100 150 200
Kiln Excess Air (%)
250
Figure 12. Effect of excess air on
predicted hexachlorobenzene
destruction in the CRF.
157
-------
predicted to occur for kiln excess air
levels above 113 percent for the kiln
alone, 154 percent for the afterburner
alone and 163 percent for the system as a
whole.
Effects of momentary flameout were
investigated by turning off the fuel to
the kiln or the afterburner, but main-
taining the wall temperatures unchanged
from the 1/26/84 data point. Such a
condition would only last for a short
time before the v/all began to lose its
residual heat and the wall temperatures
began to fall. Figure 13 shows that kiln
flameout resulted in a sharp decrease in
predicted kiln temperature, but a
relatively small decrease in the after-
burner temperature. Kiln temperature
was, of course, unaffected by afterburner
flameout, and afterburner temperature
dropped sharply. Figure 14 shows that
kiln flameout was predicted to cause a
total failure in hexachlorobenzene
destruction; however, the afterburner
alone v/as predicted to be sufficient to
cause the system to pass the 99.99
percent destruction efficiency criterion.
For afterburner flameout, the afterburner
alone was predicted to fail, but the
combination of the kiln and the after-
burner together was sufficient to
marginally pass. From these predictions,
it is evident that afterburner flameout
has the greater effect on the performance
of the system as a whole, but afterburner
flameout alone is not sufficient to cause
the system to fail at the 1/26/84
operating point.
1200
1000
2
rtJ
03
*,
-
a
o
-------
In the future it is planned to apply
Richter's three dimensional heat transfer
model (5) to the CRF. This, coupled with
the isothermal flow model, should give a
more accurate estimation of possible
temperature histories within the inciner-
ator resulting in better predictions of
waste destruction. Model development
efforts are presently concentrating on
the areas of mixing, to identify flame
and post flame zones allowing the
appropriate kinetic rates (if available)
to be applied to each, and liquid
injection to determine where droplets
evaporate and waste enters the gas
stream.
REFERENCES
1. Clark, W. D., J. C. Kramlich, J. La
Fond, R. 8. Myers, W. R. Seeker, and
W. Richter, 1984. Hazardous Waste
Incineration Engineering Analysis,
Work Assignment #2 - Engineering
Analysis Definition. Draft Final
Report for EPA Contract 68-02-3313,
JRB Subcontract 2-850002-70.
2. Clark, W. D., M. P. Heap, W. .
Richter, and W. R. Seeker, 1984.
The Prediction of Liquid Injection
Hazardous Waste Incinerator
Performance. ASME Paper 84-HT-13.
3. La Fond, J. F., and D. K. Moyeda,
1985. Engineering Analysis of
Hazardous Waste Incineration;
Isothermal Flow Modeling of a Rotary
Kiln Waste Incinerator. Poster
Presentation at the llth Annual
Research Symposium on Land Disposal,
Remedial Action, Incineration and
Treatment of Hazardous Waste.
4. Overmoe, B. J., S. L. Chen, and W.
R. Seeker, 1983. Development of
Laboratory Scale Reactors for
Hazardous Waste Incineration.
Topical Report, EPA Contract No.
68-02-3633.
5. Richter, W., and M.- P. Heap, 1981.
A Semi stochastic Method for the
Prediction of Radiative Heat
Transfer in Combustion Chambers.
Western States Section, The
Combustion Institute, Spring
Meeting, Paper 81-17.
6. Rubey, W. A., J. Torres, D. Hall, J.
L. Graham, and B. Dellinger, 1983.
Determination of the Thermal
Decomposition Properties of 20
Selected Hazardous Organic
Compounds. University of Dayton
Research Institute, Draft Report,
Cooperative Agreement
CR-807815-01-0.
7. Sittig, M. 1979. Incineration of
Industrial Hazardous Wastes and
Sludges. Noyes Data Corporation,
Park Ridge, N.J.
8. Versar Inc. Southern Operations,
March 1984. Operation, Performance,
and Test Results to Date for the
USEPA Combustion Research Facility
Rotary Kiln Incineration System.
Interim Technical Report, EPA
Contract No. 68-02-3128.
ACKNOWLEDGEMENT
The authors are grateful for the
support of C. C. Lee of EPA, the task
officer for Hazardous Waste Engineering
Analysis, for the cooperation of L. J.
Staley of EPA for information on the CTT,
for the cooperation of R. A. Carnes of
EPA and F. C. Whitmore of Versar for
information on the CRF, and for the
assistance of R. Zimperman and D. Endsley
of EER and C. Weatherston now at the
University of Utah for their work with
the isothermal model.
159
-------
EXAMINATION OF FUNDAMENTAL INCINERABILITY
INDICES FOR HAZARDOUS WASTE DESTRUCTION
Barry Dellinger, John L. Graham, Douglas L. Hall, and Wayne A. Rubey
University of Dayton Research Institute
300 College Park
Dayton, Ohio 45469
ABSTRACT
Rankings of POHCs by the various proposed indices of hazardous waste "incinerability"
have been compared to observed incinerability for ten pilot or full scale hazardous waste
thermal destruction devices. Each index failed to predict field results except for the
method based on laboratory determined thermal stability for hazardous waste mixtures under
oxygen deficient conditions. Most importantly, it was found that thermal reaction
products (which were also POHCs), could be formed in sufficient yield to dominate the ap-
parent POHC ORE. It is concluded that product formation is the most important factor in
determining observed DREs for all but the most stable or difficult to form POHCs.
Laboratory thermal decomposition testing of actual waste streams is useful to predict
results of full scale incineration.
Current incinerator performance stan-
dards require that the principal hazardous
organic constituents (POHCs) of each waste
be destroyed or removed to an efficiency
of 99.99%. Compliance with this standard
1s usually established through a trial
burn. Since the demonstration of 99.99%
destruction and removal efficiency (ORE)
for every listed organic compound in any
waste that may be fed to a given in-
cinerator is prohibitively expensive and
difficult, only selected compounds are
tested during a trial burn. In order to
be certain that the trial burn ensures
that all listed compounds are efficiently
incinerated, various hierarchies of
"incinerabilty" of hazardous organics have
been developed. The selected compounds
should in general, be more difficult to
incinerate (higher of the incinerability
hierarchy) than other listed compounds to
be fed to the incinerator. Clearly, an
accurate prediction of the relative ORE is
of critical importance to the current
regulatory approach.
In this manuscript, we compare the
results of field tests at various
facilities with the results predicted by
the proposed hierarchies. It appears that
sufficient laboratory and field results
are now available to make some generaliza-
tions concerning predicted field results
and to establish guidelines for further
trial burn testing.
INCINERABILITY INDICES
Six methods of ranking the relative
incinerability of hazardous organic com-
pounds have been previously proposed
Li-9J.
Heat of Combustion
Auto-Ignition Temperature
Theoretical Flame Mode Kinetics
Experimental Flame Failure Modes
Ignition Delay Time
Gas Phase Thermal Stability
The heat of combustion of a substance
is defined as the enthalpy change for a
160
-------
reaction in which one mole is completely
reacted with oxygen [1]. The hypothesis
behind this hierarchy is that those com-
pounds with a large heat of combustion per
gram molecular weight (AHc/g) will produce
a higher flame temperature due to the ex-
othermicity of combustion reaction.
Presumably, the higher the flame tempera-
ture, the greater the destruction
efficiency of the compound. Conversely,
those compounds with low AHc/g will be
poorly destroyed due to low flame
temperature. A ranking of£Hc/g for listed
compounds is presented in the US-EPA
Guidance Manual for Hazardous Waste
Incineration Permits.
Autoignition temperature (AIT) is the
lowest temperature at which a combustible
material in the presence of air begins to
self-heat at sufficient rate to produce
combustion without any other source of ig-
nition [2]. Laboratory studies have shown
a correlation between gas phase thermal
stability and autoignition temperature
[7]. As a result AIT has been suggested
as a possible ranking scheme. The lower
the AIT, the easier the substance is to
decompose. The basis for its ap-
plicability would appear to be related to
the self heating properties of the waste,
and its to sustain radical chain
reactions.
Theoretical flame mode kinetics
(TFMK) focuses on estimation and ex-
trapolation of elementary reaction rate
data that is available from experiment and
theory [3]. Only a small number of com-
pounds may be ranked using this approach
due to limited data. The approach is .
based on calculations that predict ther-
modynamically complete oxidation of most
POHCs below 500C. In contrast, field and
laboratory results show incomplete oxida-
tion at temperatures greater than 700C for
most substances. This implies that
kinetics and not thermodynamics is con-
trolling the rate of destruction of the
compounds.
The experimental flame failure mode
(EFFM) approach is generally based on ex-
perimental determination of destruction
efficiencies in bench scale flame systems
[4]. However, under this approach as
originally proposed, the compound ranking
may vary depending on the "failure mode"
or upset conditions of the flame. Four
failure modes have been identified: poor
atomization of the waste feed, poor mixing
of waste and air, low flame temperature,
and quenching of the reactant waste by
contact with cool surfaces or makeup air.
Only five compounds were originally
ranked, but recent laboratory studies have
generated additional data and rankings
based on flame speed in a flat flame
burner [5].
The ignition delay time (IDT) of a
hazardous organic compound or mixture is
defined as the interval between an initial
exposure to a step function change in tem-
perature and the principal exothermicity
of the reaction as indicated by a rapid
increase in temperature and pressure of
the mixture [6]. These times may be
measured in shock tube experiments. The
basis of this approach is that ignition
delay is controlled by, and inversely
proportional to, the reaction kinetic
rate. Thus, the smaller the IDT of a sub-
stance, the greater the ease with which it
may be incinerated.
A ranking has been previously
proposed based on laboratory determined
thermal stability specified by the tem-
perature required for 99% or 99.99%
destruction at 2.0 seconds reactor
residence time in an atmosphere of flowing
air [Tqq(2) and T qg(2)]> [7.8]. The
basis for this approach is that any un-
destroyed material escaping the flame,
must eventually be dealt with by thermal
oxidation in the post-flame zone. It is
proposed that the destruction of POHCs in
the fraction of the waste feed experienc-
ing the flame environment is essentially
the same for all organic compounds, i.e.,
greater than 99.999%. Thus, the dif-
ferences in their measured ORE must be due
to differences in their rate of destruc-
tion for the fraction of the waste
escaping the rigors of the flame.
This scale was originally developed
for pure compounds in flowing air.
However, recently generated data has shown
that relative stability varies as a func-
tion of the composition of the waste feed
and oxygen concentration [9]. This has
led to modification of the rankings to ac-
count for the thermal stability of
individual POHCs fed as a mixture in both
an oxygen rich (TSHiOJ and an oxygen
deficient (TSLo02) environment.. These
hierarchies have been applied to predict-
ing the results of studies described in
the following paragraphs.
161
-------
COMPARISON OF FIELD RESULTS AND PREDICTIVE
METHODS
Intel-comparison of field and
laboratory data should be conducted with
extreme caution. While lab studies are
usually conducted under precisely control-
led well-defined conditions, field studies
are generally not [4,7,9,10]. Upon ex-
amination of field study reports, it is
obvious that the quantitative intercom-
parison of the performance of the
facilities with respect to operational
parameters is not viable. However, rela-
tive DRE data for POHCs within a waste
feed at a given facility can be analyzed
with proper data validation guidelines.
To ensure a valid comparison of predicted
and observed results, the following data
validation and reduction criteria were
used:
t only compare POHC DREs for a given
incinerator
• only compare POHC DREs when they
are fed to the system at a common
point
• use averages of DREs when no sig-
nificant run to run variation in
relative POHC DRE is observed
I only use data where the majority
of the POHC DREs are less than
99.995%
• include data from non-concurrently
fed POHCs if other key parameters
are held constant
• conduct the correlation of ob-
served field vs. predicted results
on a rank/order basis with a mini-
mum of four data points
The observed incinerability ranking
of the test compounds at each source were
compared with the prediction of each
proposed hierarchy using a rank/order cor-
relation approach [11]. This method was
judged to be superior to a linear regres-
sion analysis since the latter judges the
agreement of the data with a best-fit
straight line while the former simply
determines if a statistically significant
relationship exists between the observed
and predicted rankings. The rank-
correlation coefficient, r , was used to
judge if a correlation exilted at the 90%
confidence level for a number of test com-
pounds, N.
Results of this analysis are sum-
marized in Table I for the ten studies
judged to meet the data validation
criteria [8-10,12-17]. Of the eight
proposed ranking methods, only Hc/g, AIT,
TQQ(2), TSHi02, and TSLoOp had a suffi-
cient data base to make predictions for a
significant number of sources. Of these,
only the experimentally predicted order
under low oxygen conditions, TSLoO? met
with a reasonable success, i.e., 70%. The
other four methods only correlated with
field observations 10-20% of the time.
More importantly, it was apparent after
' detailed examination of the individual
data plots that certain trends were occur-
ring that could not be explained by simple
application of the ranking methods. In
particular, the compounds that deviated in
stability from predictions of the TSLoO
hierarchy were often the same for the
various studies. In many cases, this
deviation could be explained using other
available information.
The paragraphs that follow discuss
the data from the specific sources in a
manner that demonstrates how the field
scale observations can be reliably
predicted with modifications of the TSLoO
hierarchy. ^
Study A. The test compounds follow
the order of stability toluene methyl
ethyl ketone 1,1,1-trichloroethane
Freon 113. The observed order was the
same as predicted by TSLoO~ except for
reversal of 1,1,1-trichloroethane and
Freon 113. In actuality, both of these
compounds are predicted to be relatively
very fragile under low 0? conditions, and
the predicted rankings could have been
easily reversed. The predicted rankings
as pure compounds in flowing air or in a
mixture with high 0 were quite different
and did not correlate with the
observations. This is consistent with the
low 0 levels noted in the field study
reports.
Study B. The predictions of the
TSLoO method and the observed stabilities
agree quite well with only a few
^exceptions. Chlorobenzene-and
dichlorobenzene were observed to be
reversed from the predicted order. This
is readily explained by the observation
that significant levels of chlorobenzene
were detected in the scrubber makeup waste
and could be stripped out and into the
stack gases. This would result in an ap-
parent chlorobenzene DRE lower than that
actually achieved by thermal destruc-
tionand account for the disparity with the
TSLo02.
162
-------
CM
CD
O
CD cn CD
CD CM CD
cn LO 10
cn oo i-i «*•
oo cn co CD
cn CM ^ o
^~ - *si~ *"^ cn
*
* t—I
00 r-l
LO VO
LO OO
<£> LO
O O
LO
o CD
CM
CD
£
OO
h-
iO
CD
O
o
o
fe
oo
CO
LO
o
LO
CD
0
O
cn
CO
CO
o
CO
LO
CM
o
LO
r-t
I— 1
o
o
*
"5f
CD
o
oo
o
- I—
_i
=C Q oo
Z LiJ C£S
CJ UJ =C
^H i— i cc o:
I— Qu
UJ OO OO
_1 i— l CO UJ
CQ t— =3 •— i
1-1
U_ CO
o a «a:
UJ I—
oo > oo
I— D;
_l UJ
ID oo
00 OQ
LU CD
o;
O
CD
O
(O
eu
IE
o
o
o
o
oo
o o
I I
o
o
o
I
o o
I O O
I IO
cu
CU
o
e
cu
T3
•r-
<*-
o
o
o
cn
eu
+J
c:
o
o o
CD CD I
CM CM I
• *
O O
I
^o ^«
O CO rH
IO CM f-
O <* LO
IO OO CO
r— CM
LO VO
•* CM
f-.
•Si-
co
(O
O
LO CO
o
CD
CO
O O O
cn CD CD
i—I LO t—I
*
l-~.
,— •* r-.
cn co
oo «*•
LO CO
o
o
oo r~~
oo r~-
co CD
cn
CM
oooooooooo
I I I I III
cfl
cu
o
0
oo
^
CD
=»=
3
r—
•r—
CO
U_
t| ^
O
=*fe
•X
I/I
-------
A major deviation was observed for
bis-2-ethyl-hexylphthalate which appeared
more stable th'an predicted. Although the
predicted stability for phthalate is ques-
tionable due to lack of laboratory data,
phthalates are ubiquitous and detected
levels may be due to outgassing of plas-
tics in the system and not from
undecomposed feed. High levels of phtha-
lates are commonly found in ambient
environments and for this reason, should
probably be excluded from all data sets
[18], Bis-2-ethyl hexyl phthalate was
found at high levels in the scrubber
water. Stripping from the water by the
effluent gas could account for its ob-
served emissions.
Two other major outliers were aniline
and trichloroethylene. These compounds
were significantly more fragile than
predicted. Neither aniline nor
trichloroethylene would be expected to be
a major thermal reaction product from this
test sample. This is in contrast to
chloroform, carbon tetrachloride, and
phosgene which unexpectedly surpassed
aniline and trichloroethylene in apparent
stability. The apparent thermal stability
of carbon tetrachloride, chloroform, and
phosgene may be due to their formation as
products from other components of the
waste as opposed to their stability as
POHCs. Furthermore, these compounds are
quite volatile and could be present in the
ambient air in the form of fugitive
emissions. Either formation as a product
or as an ambient air contaminant could ex-
plain the unexpected reversal in thermal
stability.
Study C. The waste was spiked with
theoretically stable POHCs which had an
observed order of stability: acetonitrile
benzene trichloroethylene chloroben-
zene carbon tetachloride. This was as
expected except for benzene which was con-
siderably more stable than predicted based
purely on thermal stability. It is pos-
sible that benzene was formed as a product
from chlorobenzene (or the auxiliary
fuel). This hypothesis is supported by
two independent observations. First, a
simulated waste stream very close in com-
position to the actual waste was subjected
to thermal decomposition in the
laboratory. Under low CL conditions, ben-
zene would actually have been predicted as
a reaction product resulting in a low ap-
parent ORE for benzene as a POHC.
Furthermore, the waste stream was fed to
the full scale incinerator without benzene
in the feed. Roughly equivalent levels of
benzene were found in the stack effluent
confirming the hypothesis that its emis-
sion was due to sources other than
residual POHC from the waste feed.
Study D. Field test results were in
basic agreement with prediction for low
oxygen conditions. The exceptions were
phthalates, which were previously dis-
cussed, and tetrachlorethylene which was
predicted to be the most stable component
but was observed to be less stable than
benzene, toluene, naphthalene, carbon
tetrachloride, and methyl ethyl ketone.
Laboratory studies have demonstrated or
strongly suggested that each of these com-
pounds can be a significant reaction
product from various precursors [9,12,19].
Dichloromethane and chloroform were also
found in the source emissions, suggesting
the formation of chlorinated methanes as
thermal reaction products. Thus, the ap-
parently greater stability of these
compounds than tetrachloroethylene may be
due to their formation as products in the
incineration process.
Study E. A correlation was observed
between predicted and observed rankings
but there was significant scatter. The
fragile nature of 1,1,2-trichloroethane,
1,1,1-trichloroethane, and methyl ethyl
ketone were correctly predicted (DREs all
at 99.999% or greater). The observed
stability of these three compounds were
permuted from their predicted value con-
tributing to the poor correlation
coefficient.
Methylene chloride and to some ex-
tent, carbon tetrachloride appeared more
stable than predicted. It should be noted
that high levels of other halogenated
methanes were found in the stack effluent
indicating a source of carbon
tetrachloride and methylene chloride emis-
sions other than residual POHC (i.e.,
either incomplete combustion products or a
result of stripping of these volatiles
from the scrubber water).
The most unexpected behavior was ex-
hibited by tetrachloroethylene, which was
predicted to be the most stable POHC but
was observed to be very fragile.
Study F. Although this facility ex-
hibited the lowest correlation of
predicted and observed emissions, the
164
-------
results are extremely informative. Two
cffstfnct groups were evident, one consist-
ing of primarily chlorinated aromatics and
olefins; and a second consisting of
primarily halogenated aliphatics along
with bis-2-ethyl hexyl phthalate and
hexachlorocyclobutadiene.
Methylene chloride and chloroform
were found in the scrubber makeup water
which could readily account for their ob-
served emission levels. The other
halogenated compounds (in the second
group) are also very volatile and have
been found in the ambient air surrounding
such facilities (presumably due to fugi-
tive emissions) [10]. As previously
discussed, phthalate emissions are consis-
tently high at most sources. Finally,
there is some question concerning the ac-
curacy of the predicted ranking for
hexachlorocyclopentadiene due to lack of
lab data. Its low stability prediction
was based on possible strain of the five
membered ring structure, but could well be
in error. If the six compounds in ques-
tion are eliminated from the data set and
a correlation is performed with the
remaining nine compounds, a statistically
significant rank correlation coefficient
of 0.89 is obtained.
Study G. The observed stability is
as predicted under low 0? conditions ex-
cept for carbon tetrachloride which
appeared more stable than chlorobenzene.
This is not surprising since chloroform
was also present in the mixture and carbon
tetrachloride has been established as a
thermal reaction product of chloroform by
lab studies.
Study H. The POHCs in this test es-
sentially fbl lowed the predicted order
except for tetrachloroethylene and
trichloroethylene which appeared less
stable than benzene and toluene, contrary
to predictions. This type of result has
been observed in other studies and is
ascribed to the propensity for formation
of toluene and benzene as reaction
products. It is also interesting to note
that carbon tetrachloride emissions were
also quite high (.average of 173 g/s)
which, tends to confirm its prevalence as a
reaction product from incineration of
chlorinated wastes.
Study I. The observed POHC
stabilities followed predicted trends ex-
cept for benzene, carbon tetrachloride,
and 1,2-dichloroethane. Benzene and
carbon tetrachloride are again expected to
be products of thermal degradation
(primarily from chlorobenzene/toluene and
methylene chloride respectively). The
1,2-dichloroethane is a volatile compound
that is commonly found in scrubber water
or in the ambient air as a fugitive emis-
sion, factors which could account for its
elevated emissions level [10]. The emis-
sion level of 1,1,1-trichloroethane, also
sometimes found as a fugitive emission or
in scrubber makeup water, was also
slightly elevated.
Study J. The observed deviations
from the predicted rankings were similar
to those observed for the previous nine
cases. Benzene, toluene, and carbon
tetrachloride emissions were higher than
expected, an observation which is at-
tributed primarily to product formation.
DISCUSSION
The degree of success, as indicated
by the results reported in Table 1 and the
subsequent discussions of predicting the
relative thermal stabilities of hazardous
organics through laboratory flow reactor
studies may appear somewhat surprising
considering the complexity of the in-
cineration process. However, when one
breaks down the overall process into more
fundamental phenomena, the degree of suc-
cess of the predictions is understandable.
In determining the destruction ef-
ficiency of hazardous organic materials by
incineration, chemical reactions occurring
in condensed phases may effectively be
neglected. This is true due to mass and
heat transfer considerations. Thus, we
may primarily concern ourselves with gas-
phase chemistry although the nature of the
passage of material from condensed phase
into the gas-phase by physical processes
may be important.
Once in the gas phase, there exist
more than one mode of destruction of the
material and it is necessary to address
the factors affecting these destruction
modes. Two modes are clearly evident and
they may be designated as direct flame and
thermal (non-flame).
Either flame mode or thermal destruc-
tion studies indicate that any known
organic waste can be destroyed in an
165
-------
incinerator to greater than 99.99% DE if
it is operating under theoretically op-
timum conditions [4,8,20]. Thermal
destruction can be expected at less than
1000C in flowing air at a mean residence
time of 2.0 seconds. Flame destruction of
waste droplets may occur in flames operat-
ing in excess of 850C. The fact that
these theoretical optimum conditions
roughly correspond to the mean conditions
experienced in an incinerator has caused
much confusion. The observation of or-
ganic emissions from incinerators
(sometimes in large quantities) is proof
that frequent excursions from the optimum
or even the mean conditions are occurring.
Excursions, or fault modes, are prob-
ably the controlling phenomena for
incineration efficiency. Four parameters;
atomization in efficiency, mixing in ef-
ficiency, thermal failure, and quenching,
have been identified as failure modes in
flames [4]. Laboratory studies have shown
that relatively small excursions from
ideality for these parameters can easily
drop measured flame destruction ef-
ficiencies from greater than 99.99% to 99%
or even less than 90% (three orders of
magnitude).
Non-flame upset parameters can be
conveniently classified in terms of dis-
tributions of oxygen, residence time, and
temperature [8,9],
The key to understanding the sig-
nificance of upset conditions is that only
a very small fraction of the total volume
of the waste needs to experience these
less than optimum conditions to result in
significant deviations from the targeted
destruction efficiencies.
POHC Destruction Model
To Illustrate how laboratory thermal
decomposition testing relates to upset
modes and can be used to predict observed
emissions from full scale facilities, let
us examine a specific example.
Previous research has shown that the
destruction kinetics of typical hazardous
organic compounds can be described satis-
factorily using simple pseudo-first order
kinetics [9], Although different or more
complex models may be used, the actual
model used is not important for the scope
of this discussion.
We will first examine the case of a
simple one stage combustor where a waste
feed mixture is fed directly into a tur-
bulent flame and the hot gases evolving
from the flame pass on through a rela-
tively long, high temperature holdup zone
prior to exiting the system.
Representative reaction conditions
for the flame can be chosen as an average
residence time of O.ls and a bulk flame
temperature of 1700K. For the post-flame
zone, we may choose a mean residence time
of 2.0 s and a bulk gas phase temperature
of HOOK. Although a range of residence
times and temperatures are actually ex-
perienced by the individual molecules, the
values chosen are typical effective
residence times and temperatures.
As discussed in the previous sec-
tions, several destruction failure modes
have been identified for the flame. In
this model, we will assume that only 1% of
the waste feed avoids experiencing the
bulk reaction conditions in the flame.
This might be caused by a reduced gas
phase residence time from an improperly
operating nozzle or from experiencing a
reduced temperature as a result of being
sealed in particulate matter. A third
cause might be reduced time at temperature
from quenching by cold gases or poor
mixing with oxygen.
This one percent of the waste feed
must then enters the post-flame zone. The
overall measured destruction efficiency at
the stack is the weighted average of the
destruction efficiencies of the flame and
post-flame zones. The results of these
calculations for hazardous waste of a
range of thermal stabilities are shown in
Table 2. From examination of the table,
it is apparent that each of the compounds
are destroyed to essentially the same ef-
ficiency in the flame, i.e., greater than
99.99%. In the post-flame region, sig-
nificant differences in thermal stability
are observed.
From examination of the last column
of the table, it is apparent that the
overall destruction efficiency parallels
the destruction efficiency in the post-
flame region. The principle value of the
overall DE is 99% in all cases, with the
variations in DE occurring to. the right of
the decimal. The destruction achieved In
the flame determines the principle value,
while the non-flame destruction efficiency
determines the approach to four nines.
166
-------
TABLE 2
CALCULATED DESTRUCTION EFFICIENCIES FOR-REPRESENTATIVE
HAZARDOUS ORGANICS :.. >
COMPOUND
CALCULATED DESTRUCTION EFFICIENCIES
A1 E DE DE DE
(S ) (kcal/mole) (Flame) (Post-Flame) (Overall)
Acetonitrile 4.7x10
Benzene 2.8x10*
Chloroform 2.9x10
Tetrachlorobenzene 1.9x10
.12
Tetrachloroethylene 2.6x10
Trichlorobenzene 2.2x10
40
38
,49
30
33
38
99.999+
99.999+
99.999+
99.999+
99.999+
99.999+
66.357
99.999+
99.999+
98.566
77.127
99.968
99.664
99.999+
99.999+
99.986
99.771
99.999+
The overall destruction efficiencies
quoted in the table are typical of
preliminary results reported for studies
on full-scale incinerators. The measured
destruction efficiencies for essentially
all full scale systems have exceeded or
approached 99.99% for most compounds.
Variations have been in the third, second,
or in some cases, the first decimal place.
A further observation has been that
most incinerators can achieve a DE of
99.99% for essentially all waste feeds
when operating optimally. However, op-
timum operation cannot be attained on a
continuous basi.s. If an incinerator could
be sampled on a continuous basis, one
would probably find that at least 90% of
the hazardous organic emissions occur in
the fraction of time when the incinerator
experiences an upset. Such upsets could
be loss of flame, an overload of waste
feed, or a failure of a spray nozzle. It'
is during these system upsets that a large
percentage of the feed material can escape
flame mode destruction and the reaction
conditions in the post-flame zones can be
degraded from their steady state operating
values. Under upset conditions, the dif-
ferences in waste incinerability may be
magnified, the non-flame zone destruction
comes to even greater prominence, and the
performance of the incinerator fails to
achieve four nines for greater number of
components of the waste feed.
As indicated in Table 1, laboratory
results obtained under high oxygen condi-
tions were not successful in predicting
relative POHC stabilities while predic-
tions under low oxygen were very
successful. The reason for this relates
again to the concept of failure modes.
Poor mixing of waste and oxygen in
the afterburner gives rise to a certain
fraction of the waste being subjected only
to low oxygen conditions. Numerous
laboratory studies have shown that
destruction of the feed material is much
slower under these conditions and product
formation is enhanced. We again have the
case where although most of the waste ex-
periences oxidizing conditions and is
destroyed, the small fraction of the feed
experiencing the pyrolytic conditions may
be responsible for the emission. The ob-
servation in field and lab studies that
most reaction products are pyrolysis type
products (e.g., benzene, toluene,
naphthalene) tends to confirm this
hypothesis.
Although the conclusion that a sub-
fraction of a fraction of the waste feed
is responsible for most hazardous organic
emissions may be surprising at first, the
same process is generally responsible for
emission of most air pollutants. One is
not really concerned with the major -
chemistry (for example, in a power plant)
which forms carbon dioxide and water, but
instead the minor reaction pathways which
167
-------
form sulfur dioxide, sulfuric acid, and
nitrogen oxides. These pathways are
responsible for less than 0.1-1% of the
stack emissions but are the reactions of
interest in pollutant formation.
Impact of Thermal Reaction Products
Products of incomplete combustion
(PICs) resulting from the incineration of
hazardous waste are not currently regu-
lated by the US-EPA. However, the
previously discussed field data and
results of other laboratory, pilot, and
full-scale testing programs have shown
that toxic products can be formed and are
emitted from incinerators [9,10,12-21].
The previously presented in-
cinerability hierarchies do not directly
address the issue of PIC emissions as they
are only concerned with thermal stability
of the POHCs in the feed material. Many
observed PICs are also potential POHCs,
consequently, it is entirely possible that
a PIC may also be a POHC in the original
mixture. Three documented examples are:
the formation of carbon tetrachloride from
chloroform, hexachlorobenzene from pen-
tachloronitrobenzene, and benzene from
chlorobenzene or toluene [9,12,19],
In the previous discussion of field
results many such cases were identified.
This gives rise to a low apparent ORE for
the POHC. Since this effect would be more
important when the input concentration of
the POHC is low, the result would be an
apparent dependence of ORE on input POHC
concentration (i.e., the higher the input
concentration, the greater the apparent
ORE). The true effect, however, is that
the emission concentration is constant,
since the emissions are probably due to
product formation from other waste
components.
The observation of an apparent ORE
dependence on concentration has been made
for hazardous waste incinerators and at-
tributed to greater than first order
kinetics for individual POHCs [10]. While
such an effect could be possible for com-
bustion of a pure compound, it is highly
improbable when the POHC is only a small
portion of a complex waste. The reaction
chemistry is determined by the overall
waste and fuel composition as opposed to
pure compound kinetics.
Volatile POHCs in the ambient air as a
result of fugitive emissions, volatile
POHCs stripped from scrubber waters, and
outgassing of phthalate containing
materials would also give rise to apparent
concentration dependencies since their
emissions levels would be constant while
the POHC input rate varies. Specifically,
it has been shown in the results section
that most of the observed deviations from
laboratory predicted rankings of in-
cinerability may be attributed to product
formation or "contamination" of the stack
effluent by volatile POHCs that did not
pass through the destruction zones of the
incinerator.
As if predicting POHC stability were
not difficult enough, we must now predict
product formation. This is best ac-
complished by laboratory thermal
decomposition testing of the actual waste
stream to be incinerated, or a very close
simulation. As indicated by the agreement
of lab predictions based on low 0? condi-
tions, these studies should be conducted
under pyrolytic conditions.
An excellent example of this approach
is study C. The incinerability ranking
based purely on POHC ORE was successful
for four out of the five constituents of
the waste, only benzene being apparently
more stable than the other components.
However, lab testing was performed on a
very similar waste stream and under
pyrolytic conditions, significant levels
of benzene were observed. Thus, when
product formation is included, lab testing
of a simulated waste stream would cor-
rectly predict the observed field results.
SUMMARY AND CONCLUSIONS
The results of comparison of ten
field studies with thermal stability
predictions indicates that no ranking
based on pure compound properties can
provide an appropriate scale of
incinerability. However, a ranking based
on predicted POHC stability in complex
mixtures under low oxygen conditions gave
a statistically significant correlation
with field results in seven of ten cases.
More importantly, analysis of results .
gives strong reason to believe that forma-
tion of POHCs in the incineration process,
may be responsible for their observed
DREs.
168
-------
Pending further confirmatory com-
parisons with field results, the following
conclusions are proposed.
• Measured POHC DREs and relative
stabilities of all but the most stable
compounds are due to formation as
products from other components of the
waste or fuel feed.
I Only DREs for very stable POHCs or
POHCs difficult to form as reaction
products (e.g., acetonitrile) are ex-
pected to be unaffected by PIC
formation and these stabilities are
predictable from pure compound thermal
decomposition kinetics.
I The stack emissions and observed DREs
of very volatile compounds (e.g.,
methylene chloride, chloroform, di-and
trichloroethanes) may be dominated by
fugitive emissions in the ambient air
or stripping of these compounds from
contaminated scrubber water.
• Thermal destruction, not in-flame
destruction determines relative POHC
DREs and the identity and yield of
products of incomplete combustion.
• Pyrolytic conditions in the incinerator
are responsible for most emissons and
control the relative DREs of POHCs and
the formation of products.
I Predictions of laboratory thermal
decomposition testing of pure compounds
and mixtures can be effectively used to
predict relative POHC DREs.
I Laboratory testing under pyrolytic con-
ditions on actual waste streams or
closely simulated waste streams is the
most effective and reliable method for
predictng relative POHC stabilities and
PIC emissions.
ACKNOWLEDGMENTS
We gratefully acknowledge the support
and inputs of our EPA colleagues Mr.
Robert E. Mournighan and Mr. Richard A.
Carnes. We also acknowledge the dedicated
efforts of Mr. Michael Graham and Ms.
Debra Tirey who performed the majority of
the data reduction for this manuscript.
CREDIT
This work was performed under the
partial sponsorship of the US-EPA
Hazardous Waste Engineering Research
Laboratory under Cooperative Agreement CR-
80783.
REFERENCES
1. 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, 1981.
2. J. J. Cudahy, and W. L. Troxler,
"Autoignition Temperature as an
Indicator of Thermal Oxidation
Stability," Journal of Hazardous
Materials, 8, 1983.
3. W. Tsang, and W. Shaub,
Detoxification of Hazardous Waste
Chapter 2, "Chemical Processes in the
Incineration of Hazardous Materials,"
Exner, J. H., ed; (Ann Arbor Science
Publishers, Ann Arbor, MI, 1982, pp.
41-60).
4. J. C. Kramlich, et al., "Laboratory
Scale Flame-Mode Hazardous Waste
Thermal Destruction Research,"
Revised Draft Final Report by EERC to
EPA Prime Contract Number, 68-03-3113
under Subcontract Task 24-1, 1983.
5. R. D. Vandell and L. A. Shadoff,
Chemosphere, Vol. 13, No. 11, 1984.
6. D. L. Miller, V. A. Cundy, and R. A.
Matula, "Incinerability
Characteristics of Selected
Chiori nated Hydrocarbons,"
Proceedings of the Ninth Annual
Research Symposium on Solid and
Hazardous Waste Disposal, Cincinnati ,
OH, May, 1983.
7. K. C. Lee, N. Morgan, J. L. Hansen,
and G. M. WhippTe, "Revised Model for
the Prediction of the Time-
Temperature Requirements for Thermal
Destruction of Dilute Organic Vapors
and its Usage for Predicting Compound
Destructibility," presented at 75th
Annual Meeting of the Air Pollution
Control Association, New Orleans,
June, 1982.
8. B. Dellinger, J. L. Torres, W. A.
Rubey, D. L. Hall, J. L. Graham, and
R. A. Carnes, "Determination of the
Thermal Stability of Selected
Hazardous Organic Compounds,"
HAZARDOUS WASTE, Vol. 1, No. 2, 1984.
169
-------
9. J. L. Graham, D. L. Hall, and B.
Del linger, "Laboratory Investigation
of the Thermal Degradation of a
Mixture of Hazardous Organic
Compounds - I," Submitted to ENV.
SCI. & TECH.
10. Performance Evaluation of Full Scale
Hazardous Waste Incinerators. Final
Report, MRI report submitted to the
U.S. Environmental Protection Agency,
Contract 68-02-3177 August, 1984.
11. W. 0. Dickson and F. J. Massy, Jr.,
Introduction to Statistical Analysis,
2nd ed., McGraw-Hill, New York, 1957.
12. B. Dellinger, D. L. Hall, J. L.
Graham, and W. A. Rubey, "Destruction
Efficiency Testing of Selected
Compounds and Wastes," Final Report
to Eastman Kodak Company, September,
1984.
13. C. D. Wolbach, and A. R. Gorman,
"Destruction of Hazardous Wastes
Cofired in Industrial Boilers: Pilot
Scale Parameters Testing," Acurex
draft final report FR-84-46/EE,
February, 1984.
14. A. W. Wyss, C. Castaldini, and M. M.
Murray, "Field Evaluation of Resource
Recovery of Hazardous Wastes, Acurex
Technical report TR-84-160/EE,
August, 1984.
15. "Evaluation of Waste Combustion in
Cement Kilns at General Paul ding,
Inc., Paulding, Ohio, draft final
report to EPA, prepared by Research
Triangle Institute and Engineering-
Science, March, 1984.
16. Trial Burn Report for Kodak Park
Division Chemical Waste Incinerator
US-EPA ID. No. NYD980592497.
17. Private Communication, Ron Bastian.
18. M. Cooke, R. E. Hall, and W. H.
Axtman, "PNA Emissions in Industrial
Coal-Fired Stoker Boilers," presented
at the 185th National ACS Meeting in
Seattle, WA, March, 1983.
19. D. L. Hall, B. Dellinger, and W. A.
Rubey, "Considerations for the
Thermal Degradation of Hazardous
Waste," Presentated at the 1983 (4th)
International Symposium on
Environmental Pollution, Miami Beach,
Florida, October, 1983.
20. "The Mechanisms of Pyrolysis,
Oxidation, and Burning of Organic
Materials," L. A. Wall, ed.,
Proceedings of the Fourth Materials
Research Symposium held by NBS,
Gaithersburg, MD, NBS Special
Publication 357, CODEN;XNBSAV, 1972.
170
-------
AN OVERVIEW OF LABORATORY- AND BENCH-SCALE RESEARCH IN HAZARDOUS WASTE
THERMAL DESTRUCTION
George L. Huffman, Chun Cheng Lee, Ph.D. U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
ABSTRACT
This paper describes some of the past and current efforts in the area of small-
scale research that has come about due to the passage of RCRA (the Resource Conservation
and Recovery Act). The purpose of this paper is to indicate who is doing what in
the area of hazardous waste thermal destruction research. This paper covers the
following research activities in brief outline form (these activities are synops.ized
from Reference 9) (9):
A Past Research Activities Including:
• Non-flame Thermal Decomposition Research by the University of Dayton Research
Institute (UDRI) [though some of this research is still continuing —- see
below]
• Non-flame Thermal Decomposition Research by the Union Carbide Corporation
(UCC)
• Laminar Flame Combustion of Chlorinated Hydrocarbons by the Illinois Institute
of Technology (IIT)
• Flame-Mode Hazardous Waste Thermal Destruction Research by the Energy and
Environmental Research Corporation (EERC)
• Thermal Destruction of Chlorophenol Residues by Environment-Canada
J^Current Research Activities Including:
• Hazardous Waste Incineration Engineering Analysis (by EERC)
• EPA In-House Research (by the Hazardous Waste Engineering Research Laboratory,
HWERL)
• Investigation of Gas-Phase Thermal Decomposition Properties of Hazardous
Organic Compounds by UDRI
• The Incineration Characteristics of Selected Chlorinated Hydrocarbons by the
Louisiana State University (LSU)
• Non-Flame Waste Decomposition of Hazardous Waste by the Midwest Research
Institute (MRI)
• Heterogeneous Catalytic Oxidation of Model Chlorinated Hydrocarbons by the
Massachusetts Institute of Technology (MIT)
• Oxidation of Model Waste Components in Supercritical Water by the Massachusetts
Institute of Technology (MIT)
• Molecular Beam Mass Spectroscopic Study of Chlorinated Hydrocarbon Flames by the
Illinois Institute of Technology (IIT).
171
-------
PAST RESEARCH ACTIVITIES
Non-flame Thermal Decomposition Research
(by UDRI)'~
Under a series of contracts and
grants with EPA's Hazardous Waste Engin-
eering Research Laboratory, the University
of Dayton Research Institute (UDRI) has
been performing laboratory-scale studies
of waste thermal oxidation characteristics
in a non-flame environment since 1974
(4, 15, 5, 6, 2, 7, 3).
During their studies for EPA from
1974 to 1984, UDRI developed four thermal
instrumentation systems and tested numer-
ous compounds. The four instrumentation
systems are:
* Discontinuous Thermal Systems
(DTS)
• Thermal Decomposition Analytical
System (TDAS)
• Thermal Decomposition Unit-Gas
Chromatographic (TDU-GC) System
• Packaged Thermal Reactor System
(PTRS)
Compounds tested on these four systems
have the following general attributes
in common:
* Microgram quantities of test
compounds investigated
* Non-flame environment
* Pure compounds premixed with
air and some actual waste
streams and mixtures tested
• Plug flow assumption.
For purposes of illustration, Figure
1 represents a typical thermal decom-
position profile (here for chloroform
and its associated Products of Incom-
plete Combustion, PICs) routinely
available from the TDU-GC system.
100
t
m 10
DC
£
LU
0.1
0.01
CHLOROFORM
tr = 2.0 SEC
O CHCI3
D PIC#1
A PIC #2
O PIC #3
0
800
900
100 200 300 400 500 600 700
EXPOSURE TEMPERATURE. °C - -
FIGURE 1. THERMAL DECOMPOSITION PROFILE OF CHLOROFORM
1000
172
-------
Non-flame Thermal Decomposition Research
by Union Carbide Corporation (UCC)
This research featured a modifica- ,
tion of the system used by the University
of Dayton Research Institute (4). The
major modification was the manner in
which the mixtures of experimental gases
(test compounds) were inserted. UCC
pre-mixed the test compound with air and
contained the mixtures in a cylinder,
while UDRI mixed the test compound with
air when air was used as carrier gas to
bring the test compound.into the reactor.
UCC claimed that their test set-up had
better mixing than that of UDRI's.
In general, the test set-up used a
0.9mm bore quartz tube. The quartz tube
was twice folded at twelve-inch intervals
so that 75 percent of its length of 130
centimeters could be located within the
central portion of the furnace and within
the limits of the flat temperature pro-
file. The radial and longitudinal velo-
city and temperature gradients were mini-
mized by the narrow bore, which allowed
conditions approaching a plug flow.
As a result of the research, UCC
published two papers. The first paper
(10) discussed thermal oxidation kinetics
of four chemicals: vinyl chloride, ben-
zene, ethyl acrylate, and acrolein. The
thermal oxidation data for all four
chemicals was shown to fit a first-order
kinetic rate model. The kinetic-rate
model was then used to develop predic-
tive equations for oxidation temperature
and residence time requirements for a
99.9 percent destruction efficiency under
different flow conditions. During the
experiments, all compounds were run at
1000 ppmv. Some of the compounds were
run at 100 and 500 ppmv and these runs
gave a good agreement with the 1000 ppmv
runs. The temperatures studied ranged
from 800 to 1500°F and residence times
from 0.1 to 2 seconds. Analyses were
done for both specific chemicals and
total hydrocarbons.
The second paper (11) discussed
thermal oxidation kinetic data for 15
chemicals and also presented equations
for predicting thermal oxidation des-
truction efficiencies based on molecular
structure, residence time and autoigni-
tion temperature. The two most impor-
tant variables in the predictive
equations were found to be autoignition
temperature and residence time.
Laminar Flame Combustion of Chlorinated
" Hydrocarbons by MIT/IIT "~~"
Under an EPA Grant, Professor S.M.
Senkan studied this subject first at
the Massachusetts Institute of Technol-
ogy and then at the Illinois Institute
of Technology (22, 16, 18).
Flame velocities were measured by
using a quartz Bunsen burner with a
diameter of 1.0 centimeters. Liquid
chlorinated hydrocarbons (all of which
were with at least 99.9% purity) were
injected into metered and heated air/
methane mixtures. When the desired
flow-rates were set, and a stable,
laminar flame was obtained, pictures
were taken using an Olympus OM I camera
and high-speed black and white film.
Flame-front areas were then determined
from the enlarged picture of the Bunsen
cone using the cone-angle method (22).
Burning (flame) velocity is important
because the square of it is proportion-
al tjo the overall rate of combustion of
the mixture.
Figure 2 shows that the flame
velocities of chlorinated hydrocarbon
systems decrease as the chlorine subs-
titution of the molecule is increased.
20
1.4
Equivalence Ratio
FISURE 2. FUME VELOCITIES OF CHLORINATED METHANE-METHANE-AIR MIXTURES
173
-------
Dr. Senkan also carried out a Carbon
Formation Study (17). Carbon formation
limits were determined visually, and were
identified as the combustible mixture
equivalence ratio at which the carbon
luminosity was just suppressed. Dr. Sen-
kan found that, in actual flames, carbon
formation occurs more readily than might
be expected and varies from fuel to fuel.
Furthermore, Senkan's data sugges-
ted that the measurements of the CO
concentration levels in the incinerator
flue gases may be quite suitable as an
indicator of PIC and CMC emissions.
That is, if significant CO emissions are
not present, the presence of the other
carbon-based pollutants would be highly
unlikely. Conversely, the presence of
significant levels of CO in the combus-
tion products would indicate that the
conditions in the incinerator are im-
proper and jnay result in CHC and other
PIC emissions.
Flame-Mode Hazardous Waste Thermal Des-
truction Research by EERC
Under an EPA contract, the Energy
and Environmental Research Corporation
(EERC) conducted a laboratory-scale
study of flame-mode hazardous waste
thermal destruction during 1981-1982
(8).
A Microspray Reactor was employed
by EERC to investigate single droplet
thermal reactions without limitations
associated with atomization, mixing,
quenching or turbulent mixing. In this
flame reactor, particles or droplets of
the material to be studied were injected
through a laminar, premixed, hydrocarbon
flat flame and thermal decomposition of
the material took place in a- flame
environment.
Also, -a Turbulent Flame Reactor was
used to provide a turbulent liquid spray
flame, including swirl, recirculation,
broad droplet size distribution, and high
variation in droplet number density. The
reactor was capable of simulating the
compound escape mechanisms that could
occur in flame zones of liquid injection
incinerators. Very high heat removal
rates were utilized to quench post-flame
reactions. Thus, the destruction which
occurred in the turbulent diffusion flame
was exaggerated over any non-flame
decomposition which could occur in the
post-flame region.
Some of the more important con-
clusions of this study were:
(1) Under optimum conditions,
flames are capable of destroy-
ing hazardous waste compounds
with very high efficiencies
(greater than 99.995 percent)
without the need for long
residence time, high-tempera-
ture post-flame zones or
after-burners;
(2) Reduced flame destruction
efficiencies are the result of
operation under some failure
mode such as poor atomization,
poor mixing, or flame quench-
ing; and
(3) Optimum conditions for destruc-
tion of hazardous waste com-
pounds in turbulent diffusion
spray flames correspond to
minimal concentrations of CO
and total hydrocarbons in the
exh au st.
Thermal Destruction of Chlorophenol
Residues (by Environment-CanacTa)
Environment-Canada conducted this
research (20) to study the thermal de-
composition phenomena of chlorophenol
which the Canadian wood industry is
using for wood protection/preservation.
The experimental reactor used was
similar to what was used earlier by
UDRI and the Union Carbide Corporation.
The compounds were evaluated in order
of increasing complexity of thermal
destruction, beginning with pure com-
pounds and then continuing with wood
preserving mixtures and contaminated
sludges. During the work, with pure
compounds, the thermal destruction
characteristics including destruction
efficiency and PICs were determined as
a function of temperature and retention
time. Testing was done in a non-flame
mode oxidative environment in two
steps: the compound was gradually
vaporized and then passed through a
high temperature zone with an excess of
oxygen in order to avoid the possibil-
ity of pyrolytic reactions.
174
-------
Pure compounds of pentachlorophenol
(PCP), 2,3,4,5 tetrachlorophenol, 2,4,6
trichlorophenol and 2,4,5 trichlorophenol
(TCP) were tested and the decomposition
thermograms of PCP and 2,4,5 TCP are
given in Figure 3. Based on the pure
chlorophenol destruction data, Environ-
ment Canada estimated that at 900°C a gas
residence time not exceeding 2.9 seconds
is needed to ensure complete (>99.99
percent) destruction of both tetrachloro-
phenol and PCP whereas extrapolated data
for 2,4,5 TCP show a residence time in
excess of 5 seconds would be required.
CURRENT RESEARCH ACTIVITIES
Hazardous Waste Incineration
Engineering Analysis (by EERC)
Under contract to EPA, the Energy
and Environmental Research Corporation
(EERC) is carrying out a "paper study"
to: (1) Predict the impact of changes
in waste compositions and operating
conditions on incinerator performance
parameters such as destruction effi-
ciency and PIC formation; (2) Extrapo-
late performance data from trial burns
or field tests to predict the perform-
ance of similarly designed units; (3)
Extrapolate performance data from
smaller scale to larger scale thermal
destruction devices; and (4) Analyze
the susceptibility of a design to
different failure modes and to predict
their impact on incinerator performance.
The research approach is to inte-
grate both theoretical analysis and
experimental results into predictive
models so that the performance of incin-
erators can be estimated. The approach
for the theoretical analysis part is to
use existing models (kinetic, fluid
dynamics, heat transfer, etc.) to
develop predictive equations that govern
incinerator performance. These models/
equations will also be useful in gener-
ating hypotheses for performing future
experimental work. The experimental
part is to monitor the conduct of
laboratory-, pilot- and full-scale
testing to verify the applicability of
the predictive hypotheses assumed.
Results of all of EPA experimental
projects will be analyzed for achieving
this project goal.
EPA In-House Thermal Destruction
Research (by HWERL)
To supplement its extramural
research, EPA has constructed both
bench- and pilot-scale combustors at
the Center Hill Facility of EPA's
Hazardous Waste Engineering Research
Laboratory in Cincinnati, Ohio (HWERL).
The primary focus of this in-house
research is to: (1) Establish how
combustion parameters and variables
affect failure of a simulated hazardous
waste incinerator or an industrial
boiler that co-fires hazardous waste
with conventional fuel to achieve
99.99% ORE (Destruction and Removal
Efficiency); and (2) Determine how,
when and why PICs are formed (e.g., too
low an oxygen concentration or too low
an operating temperature) and-to deter-
mine how or whether they can be subse-
quently destroyed or removed (e.g., by
secondary combustion or scrubbing/
adsorption techniques).
Four experimental systems will be
utilized; they are:
• Thermal Destruction Unit-Gas Chroma-
tographic (TDU-GC) System --- A
non-flame unit akin to what UDRI
used previously.
• Microspray Reactor A flame-
mode reactor similar to what EERC
used in their previous study.
• Turbulent Flame Reactor — A flame-
mode reactor similar to what EERC
used in their previous study.
• Controlled Temperature Tower (CTT)
-— A pilot-scale (150,000 Btu/hr),
flame-mode, refractory-lined unit.
All four experimental systems described
above and a scrubber system have been
installed and all but the Microspray
Reactor have undergone shakedown test-
ing.
Investigation of Gas-Phase Thermal
Decomposition Properties of Hazard-
ous Organic Compounds (by UDRI)
Under EPA's support, the Univer-
sity of Dayton Research Institute
(UDRI) is continuing their investiga-
tion of non-flame thermal decomposition
175
-------
SAMPLE: PGP
SIZE: 24.19 mg
RATE: 10°C/min
0 4*0 ED 12.0 16.0 20.0 24.0 28.0 32.0 380 '4o!o '44!o
TIME (mln]
PCP DECOMPOSITION THERMOGRAM
1
320
280
240
200
160
120
80
40'
SAMPLE: 245TCP
SIZE: 17.70 mg
RATE: 10°C/mln
28
24
20
16
12
8
4
4J) 8.0 12.0 16.0 20.0 24.0 28:0 32'.0 '36'.0 4o!o *44!o
TIME |mln)
2,4,5 TCP DECOMPOSITION THERM06RAM
FIGURE 3 . PCP DECOMPOSITION THERMOGRAM AMD
2,4,5 TCP DECOMPOSITION THERMOGRAM
176
_
-------
phenomena. Their previous studies are
described herein under Past Research
Activities. UDRI's current research
objectives are to study the: (1)
Kinetics of Radical Attack; (2) Factors
Affecting High Temperature Gas-Phase
Thermal Decomposition; and (3) Measure-
ment of the Thermal Stability of Princi-
pal Organic Hazardous Constituents
(POHCs) and the Formation of Products of
Incomplete Combustion (PICs).
The Incineration Characteristics of
Selected Chlorinated Hydrocarbons
(by LSU)
As an EPA Center of Excellence, the
Louisiana State University (LSU) at Baton
Rouge has been conducting fundamental
studies relative to the incineration
characteristics of selected chlorinated
hydrocarbons (CHC) since 1982 (14, 19).
A stainless steel shock tube was
employed to study the ignition delay time
for measuring the incinerability of CHC.
The ignition delay time was defined as
the time interval from the arrival of the
shock at the end of the tube until the
sudden rise in pressure due to the onset
of the principal reaction exothermicity.
In the operation of a shock tube, a
one-dimensional shock wave is caused to
propagate within a tube filled with a
potentially reactive gas sample. This
shock wave compresses the gas sample,
thereby, heating the sample to a tempera-
ture high enough to initiate reactions.
Compounds/mixtures tested in the
shock tube were: (1) methane and its
chlorinated derivatives; (2) ethane,
1,1,1-trichloroethane, and 1,2-dichloro-
ethane; (3) ethane and trichloroethane,
and (4) benzene and monochlorobenzene.
Current efforts are focussed on the
spectroscopic study of the apparent inhi-
bition of the carbon monoxide conversion
to carbon dioxide during the oxidation of
CHCs. Future research direction will
include the use of a single pulse shock
tube to study the product distributions
from the pyrolysis, oxidation and reduc-
tion of selected CHCs.
In a companion study at LSU, a
"laminar-flow flat-flame" burner is also
being tested. It is similar to the EERC
and HWERL units described earlier.
The burner has a 6.0 cm diameter,
sintered, porous stainless steel plug
with heating/cooling coils embedded in
it. The flames are shrouded from
possible entrainment effect by passing
a concentric nitrogen (Ng) shield .gas
around the flame holder through a
bronze porous plug of 6.6 cm diameter.
A 75 micro mesh screen is placed 6.2 cm
above the burner to improve flame
stability. The burner assembly which
is housed within a 15.2 cm Pyrex tube
is supported by a 28.6 cm diameter
Teflon® rod.
Compounds tested on this unit were
CH3C1, CH2C12 and CC14. Both O^Clg
and CC14 are liquid at standard condi-
tions. Major results, so far, are the
findings of stable, hazardous inter-
mediate compounds such as vinyl chlor-
ide (C2H3C1 ) and dichloroethene
(C2H2C12T during the flame-mode de-
composition of CHsCl , CH2C12 and CC14.
Non-Flame Waste Decomposition of
Hazardous Waste (by MRI)
The Midwest Research Institute
(MRI) has built a laboratory-scale unit
to test gram quantities of hazardous
waste in liquid, semi-liquid, or solid
forms. The MRI system consists of a
volatilizing/pyrolysis heater contain-
ing the liquid or solid sample, and a
second-stage incineration heater. In
this system, a sample of the waste
material is inserted into a sample
"boat" (in a hood) and pushed into the
sample heater: Alternately, liquid can
be continuously fed into the sample
heater. The sample in that heater is
volatilized or pyrolyzed, with the
gaseous products being transported by a
heated carrier gas (e.g., N2). This
gas then combines with heated flue gas,
containing excess oxygen, after which
it enters an incineration section
(heater). Gas exiting the incineration
heater passes through a sampling train
for full-flow collection of samples for
analysis.
Hexachlorobenzene (HCB) tests have
been conducted because HCB is a solid
and also a surrogate for PCBs. A
liquid waste (containing trichloro-
ethane, tetrachloroethane, bromochloro-
methane, pentachloroethane, hexachloro-
177
-------
ethane and dichlorobenzene) is being used
because MRI feels that its constituents
may be difficult to incinerate since they
have low heats of combustion.
Heterogeneous Catalytic Oxidation of
Model Chlorinated Hydrocarbons (by
HIT)
Under an EPA Grant, Professor
Michael P. Manning of the Massachusetts
Institute of Technology is currently
conducting research to investigate the
catalytic oxidation of selected chlorin-
ated hydrocarbons.
The first catalyst tested was a
commercial supported v^Os sulfuric
acid oxidation catalyst. Preliminary
experiments were carried out at tempera-
tures of 360 and 412°C using inlet con-
centrations of 5% (mol) CHsCl in air.
CH$C] and air were passed over the fresh
catalyst for several hours before data
was taken to allow the reaction and
catalyst activity to reach steady state.
Visual inspection of the used
catalyst showed a pronounced color
change: the yellow-brown fresh catalyst
had turned blue-green after exposure to
CH3C1.
Preliminary study results included:
(1) A commercially available, promoted
V205 oxidation catalyst has been found
to be inactive for the oxidation of
methyl chloride below 420°C [this in-
activity is most probably due to the
temperature involved rather than halogen
poisoning (12)]; and (2) Chromia-based
catalysts, CrgOa supported on A^OS, have
been found to be effective for the oxi-
dation of several chlorocarbons such as
di, tri, and perchloroethylene (13).
Oxidation of Model Waste Components In
Supercritical Mater (by MfTl
Supported by NASA, Professor Jeffer-
son W. Tester of the Massachusetts Insti-
tute of Technology is conducting research
relative to the reaction rates and mech-
anisms of waste destruction in the super-
critical water environment.
The overall goal of this research
program is to understand the oxidation
kinetics of model waste components in
supercritical water. Reaction mechanisms
in supercritical water are different
from those in either low-pressure flame
oxidation (combustion) or wet oxidation.
Rate expressions from these other pro-
cesses cannot be extrapolated a priori
to conditions in supercritical water,
because the supercritical water environ-
ment is much different than combustion
or wet oxidation environments.
Oxidation in supercritical water
is a new process for wastewater treat-
ment, capable of almost complete con-
version of toxic organics to carbon
dioxide and water. Supercritical
water, which is water above both its
critical temperature of 374°C and
critical pressure of 27.6 MPa (3200
psia), has much different physical
properties than room temperature and
pressure water and is an excellent
media for oxidative reactions for
several reasons. The high temperature
of supercritical water promotes rapid
reaction rates. Supercritical water
also forms a miscible solution with air
or oxygen, which eliminates the inter-
phase mass-transport limitations that
exist in a two-phase oxidation process
(21). The high solubility of organics
in supercritical water promotes com-
plete oxidation by preventing char
formation from oxidation or decomposi-
tion by-products (1).
Currently, this research is inves-
tigating the oxidation of carbon mon-
oxide and ammonia in supercritical
water, as these steps may be the last
and rate-limiting steps in the con-
version of organic carbon to carbon
dioxide and of organic nitrogen to
molecular nitrogen. The reactor util-
ized is operated isothermally in plug
flow, both of which are important to
simplify the rate expression calcula-
tion. The reactor is constructed of
Inconel 625, which can withstand the
high temperatures and pressures of the
experiments. The reactor can be oper-
ated over a wide range of conditions,
between 400 and 600°C and between 24
and 34 MPa (3500-5000 psia).
By the end of the current phase of
the project, the experimental investi-
gation of the oxidation of ammonia and
carbon monoxide in supercritical water
will be completed. Consequently, the
orders of reaction for carbon monoxide,
178
-------
ammonia and oxygen, along with the acti-
vation energy and preexponential constant
will then be able to be determined.
Molecular Beam Mass Spectroscopic Study
of Chlorinated Hydrocarbon Flames
(by IIT)
Under an EPA grant, Professor S. Sen-
kan of the Illinois Institute of Technolo-
gy is continuing his studies on the lami-
nar flames of chlorinated hydrocarbons
CMC). Present research is aimed at
developing a more detailed understanding
of the chemistry of such flames, more
comprehensive models describing the CMC
oxidation process, and at assessing the
formation and emission of pollutants from
practical combustion incineration systems.
Experiments are being carried out in
a laminar flat flame burner similar in
design to the system's described earlier.
The flames are probed for the determin-
ation of species profiles using molecular
beam sampling coupled with on-line mass
spectroscopy (MBMS). The MBMS is a
highly versatile method for studying
reactive gaseous mixtures, because of
its ability to furnish information on
the presence of all stable and radical
species in flames.
In parallel with the experiments,
detailed chemical kinetic modeling of
the flames is also being pursued.
RECOMMENDATIONS
JL Information on PICs (Products of
Incomplete Combustion) is very
limited. Because PICs could be
more hazardous than compounds in
the original waste, studying PIC
formation and control should be
one of the most important and
focal research areas for EPA.
A Much of limited amount of data -
shown in this paper were obtained
from non-flame and microgram-quan-
tity testing environments. The
credibility of these data for
actual thermal destruction under
flame conditions remains to be
demonstrated. Therefore, research
is needed to determine the appli-
cability of non-flame, and micro-
gram-testing, results to the
full-scale realities of actual
incinerators and other types of
combustors.
A Although there is a significant
amount of experimental data now
available in the literature, no
calculational methods have been
developed to predict what ORE or
PICs would result if incinerator
conditions change. Research is
needed to fill that void.
179
-------
REFERENCES
1. Connolly, J.F., 1966. Solubility of
Hydrocarbons in Water Near the Criti-
cal Solution Temperature. J. Chem.
Eng. Data. 11. 13. ' - "
2. Dellinger, B., D.S. Duvall, D.L.
Hall, W.A. Rubey and R.A. Cannes,
1982. Laboratory Determinations of
High-Temperature Decomposition Be-
havior of Industrial Organic
Materials. Presented at the 75th
Annual Meeting of Air Pollution
Control Association, New Orleans.
-
3. Dellinger, B., et. al . Determina-
tion of the Thermal Decomposition
Properties of 20 Selected Hazardous
Organic Compounds. Draft Report to
EPA to be published.
4. Duvall, D.S. and W.A. Rubey, 1976.
Laboratory Evaluation of High-
Temperature Destruction of Kepone
and Related Pesticides. EPA600/2-76-
299.
5. Duvall, D.S. and W.A. Rubey, 1977.
Laboratory Evaluation of High-Temper-
ature Destruction of Polychlorinated
Biphenyls and Related Compounds.
EPA600/2-77-228.
6. Duvall, D.S. et. al . , 1980. High
Temperature Decomposition of Organic
Hazardous Wastes. Proceedings of
the Sixth Annual Research SymposTum:
Treatment and Disposal of Hazardous
Waste, U.S. EPA. Municipal Environ-
mental Research Laboratory.
EPA-600/9-80-011. '~
7. Graham, J.L., et. al . , 1984. Design
and Evaluation of the Prototype
Packaged Thermal Reactor System.
Draft Report to EPA.
8. Laboratory Scale Flame-Mode Hazardous
Waste Thermal Destruction Research,
1984. U.S. Environmental Protection
Agency. Report published by NTIS
(Report # PB-84-1 84902).
9. Lee, C.C. and 6.L. Huffman, 1984.
An Overview of "Who Is Doing What"
in Laboratory- and Bench -Scale
Hazardous Waste Incineration Research.
Presented at the National Conference
on Management of Uncontrolled
Hazardous waste Sites'. Washington,
~ -
10. Lee, K.C., H.J. Jahnes and D.C.
Macauley, 1978. Thermal Oxidation
Kinetics of Selected Organic Com-
pounds. Proceedings of 71st Annual
Meeting or the Air Pollution -
^Control Association! Houston, TX.
11. Lee, K.C., J.L. Hansen and D.C.
Macauley, 1979. Predictive Model
of the Time-Temperature Requirements
for Thermal Destruction of Dilute
Organic Vapors. Proceedings of the
72nd Annual Meeting of the Air ~*
Pollution Control AssoclationT
Cincinnati- OH.
12. Manning, M.P., 1981. Heterogeneous
Catalytic Oxidation of Model Chlor-
inated Hydrocarbons. Presented at
|he Environmental Control Process"
State-of -the-Art Semi na r .
Cincinnati. OH.
13. Manning, M.P. Fluid Bed 'Catalytic
Oxidation: An Underdeveloped
Hazardous Waste Disposal Technology.
To be published in Hazardous Waste
Journal . Tufts University, Medford,
"~
14. Miller, D. et. a., 1983. Inciner-
ability Characteristics of Selected
Chlorinated Hydrocarbons. Presented
at the Ninth Annual EPA Research
lymposium on Land Disposal, fncTner-
ation and Treatment of Hazardous
Waste. Ft. Mitchell, KY.
15. Rubey, W.A. , 1980. Design Consider-
ations for a Thermal Decomposition
Analytical System. EPA-600/2-80-098.
16. Senkan, S.M. , et. al., 1983a. On
the Combustion of Chlorinated Hydro-
carbons, Part I: Trichloroethylene.
Combustion Science and Technology,
Vol. U, pp. 187-202. -
17. Senkan, S.M., J.M. Robinson and A.K.
Gupta, 1983b. Sooting Limits
of Chlorinated Hydrocarbon - Methane
- Air Pre-mixed Flames. Combustion
and Flame, Vol. 49, p. 30TTI
18. Senkan, Selim M. , 1984. On the
Combustion of Chlorinated Hydro-
ISO
-------
carbons, Part II: Detailed Chemical
Kinetic Modeling of Intermediate
Zone of the Two-Stage Trichloro-
ethlyene-Oxygen-Nitrogen Flame.
Combustion Science and Technology,
Vol. 38, p. 197.
19. Senser, D. and V. Cundy, 1984.
The Incineration Characteristics
of Selected Chlorinated Methanes.
Presented at the 22nd ASME/AIChE
Heat Transfer Conference"! Niagara
Falls, N.Y.
20. Thermal Destruction of Chlorophenol
Residues, 1983. Technical Services
Branch, Environmental Protection
Service. Environment Canada.
21. Timber-lake, S.H., 6.T. Hong, M.
Simson and M. Model!, 1982. Super-
critical Water Oxidation for
Wastewater Treatment: Preliminary
Study of Urea Destruction.
SAE Tech. Pap. Ser. Number 820872.
22. Valeiras, H., A.K. Gupta and S.M.
Senkan, 1984. Laminar Burning
Velocities of Chlorinated Hydro-
carbon - Methane - Air Mixtures.
Combustion Science and Technology,
Vol. 36, p. 123.
181
-------
A LABORATORY STUDY ON THE EFFECT OF
ATOMIZATION ON DESTRUCTION AND
REMOVAL EFFICIENCY FOR LIQUID HAZARDOUS WASTES
John C. Kramlich
Elizabeth M. Ponce!et
Wm. Randall Seeker
Gary S. Samuel sen
Energy and Environmental Research Corporation
18 Mason
Irvine, California 92718-2798
ABSTRACT
The results of a laboratory-scale experimental program on the effects of spray
atomization quality on waste destruction efficiency in incineration processes are
presented. The hypothesis considered is that spray atomization quality can, under some
circumstances, be the dominant cause of a unit's failure to achieve high efficiency
destruction. The v/aste destruction efficiency measurements were performed in a
laboratory scale turbulent spray flame reactor. The measurements were obtained under
correct atomizer operation and under degraded operation. These measurements were
compared with direct measurements of droplet size distribution from the nozzles by
laser diffraction. The results indicated that the penetration of oversized droplets
through the flame zone or to the wall was the principal cause of reduction in waste
destruction efficiency. In addition, the potential of a phenomena known as "secondary
atomizatlon as a means of improving efficiency was examined. (Secondary atomization
1s a phenomena in which a volatile component is introduced into the fuel; during
heating the volatile component vaporizes and fractures the droplet, thereby improving
atomization quality. The occurrence of secondary atomization due to the inclusion of
volatile wastes in No. 2 fuel oil was demonstrated and was shown to markedly improve
efficiency.
INTRODUCTION
Incineration is an attractive
alternative native for the disposal of
organic hazardous wastes. As opposed to
landfilling or deep well injection, it
effects a permanent solution. However,
Incineration is attractive only if the
waste is destroyed to an acceptable
efficiency and if harmful emissions of
hazardous byproducts are avoided. The
Federal government has recognized that the
public welfare requires government
regulation of waste disposal through the
Resources Conservation and Recovery Act
(RCRA)(10). Through RCRA Congress has
182
-------
cftarged the Erm ronmental Protection
Agency (EPA) with the development of
regulations and the enforcement of these
regulations. The EPA has. identified over
300 compounds as hazardous (2,3) and has
established licensing and operating
regulations for devices destroying these
compounds (4). These regulations rec-
ognize the fact that thermal destruction
devices cannot operate to 100 percent
efficiency. Therefore, some emission
level must be defined as a minimum
standard for safety. Presently, 99.99
percent destruction and removal efficiency
(DRE) of the principal organic hazardous
constituents (POHCs) is the standard.
Field testing of full-scale waste
destruction facilities (11) and testing of
subscale flames (8) has shown that well
designed systems have little trouble
meeting the performance standard. Indeed,
the evidence suggests that a substantial
perturbation of design or operational
parameters are necessary for substantial
emissions to occur (8). These perturba-
tions have been termed "failure modes"
because the perturbation has caused some
fundamental rate limiting step to fail to
completely destroy the waste (7). Thus,
the key questions with respect to' DRE are:
1. What are the mechanisms that per-
mit the small amount of waste to
escape during high efficiency
operation?
2. What different mechanisms are re-
sponsible for waste release dur-
ing a failure mode?
In this paper we address failure
modes associated with the atomization of
liquid fuel or waste.
Considerable work has been directed
toward characterizing the effect of atomi-
zation quality on the combustion effi-
ciency of liquid fuels. Edwards (1)
describes two ways in which atomization
influences efficiency. First, the spray
must be sufficiently fine to allow
complete evaporation within the flame.
Secondly, the spray must be injected into
the correct portion of the flow field to
ensure stability. Organic hazardous waste
can be viewed as simply another fuel
constituent. A high DRE of the hazardous
component can be viewed as its high
"combustion" efficiency. Thus, the
same atomization factors that influence
fuel consumption efficiency would also be
expected to influence waste DRE.
In practical units, atomization fail-
ure modes can be associated with worn or
plugged nozzles. Previous results from
our laboratory (8) have shown that atom-
ization characteristics representative of
worn or plugged nozzles can result in DRE
failure. The objective of the work re-
ported here was to characterize the link
between spray fineness and DRE, and to
examine the potential of "secondary atom-
ization" to improve DRE through in-flame
reduction in droplet size.
The approach was to characterize the
droplet size distribution produced by a
series of sub-scale nozzles. This was
done under cold-flow conditions by laser
diffraction. This droplet size data was
directly compared with DRE results from a
small-scale reactor (8) to evaluate the
influence of droplet size on DRE.
In a second portion of the study, the
potential of "secondary atomization" as a
means of improving DRE was investigated.
Secondary atomization is the term used to
describe the in-flame fragmentation of '
droplets with broad boiler point distri-
butions (9). The fuel is blended with a
volatile compound; upon introduction into
the flame the volatile constituent is
evaporated from the surface of the dropr
let. The surface temperature approaches
the boiling point of the less volatile
component.
Heat from the surface conducts to the
droplet interior, which can cause homogen-
eous vaporization of the volatile constit-
uent. This internal vapor generation
causes the droplet to expand into a
bubble, which eventually ruptures. This
fragments the droplet and effectively
reduces the mean droplet diameter.
Secondary atomization has been explored as
a means of improving spray fineness for
liquids with poor primary atomization
qualities (e.g., highly viscous fluids and
slurries).
The hypothesis investigated here is
that the volatile wastes present in a
multicomponent waste stream can, in high
concentrations, induce secondary atom-
ization and improve overall DRE. The
approach was to screen a serves of wastes
for secondary atomization potential and .
compare the DRE in the small-scale reactor
for conditions where secondary atomization
183
-------
was present against conditions for which
1t did not occur. These tests were
performed under a previously characterized
atomtzation failure mode.
EXPERIMENTAL SYSTEMS
Three experimental rigs were used
during this study. The slip-flow reac-
tor was used to screen mixtures of No. 2
fuel oil and waste compounds for secondary
atomization intensity. A cold flow spray
chamber with laser diagnostics was used to
characterize atomization quality from the
test nozzles. A turbulent flame reactor
(TFR) was used to obtain ORE measurements.
Slip-Flow Reactor
The reactor was originally designed
to study the thermal decomposition
characte-isties of synthetic fuel oils; it
has proven useful for the examination of
physical processes accompanying the
thermal decomposition and combustion of
all liquid fuels. The reactor consists of
a 5 x 28 cm flat-flame burner downfired
Into a chimney of similar dimensions. The
flat flame is supported on a water-cooled
sintered stainless steel plate. The
chimney is fitted with four 156 x 28 cm
Vycor windows for optical access. As
shown I'n Figure 1, the fuel droplets are
injected ball istically normal to the hot
gas tlow. The key attributes of the rig
for the present study are 1) the
controlled generation of droplets of known
diameter by a vibrating orifice technique,
2) the exposure of these droplets to a
high temperature gas stream and 3) the
ease of visual access for determination of
secondary atomization intensity. The
final point is the principal reason the
reactor configuration is used. The
intensity of the secondary atomization
reaction can be visually gauged through
the large windows by the disruption of the
droplet stream. When no secondary
atomization is present the droplets move
smoothly through the laminar gas stream.
Even a small amount of activity can be
easily distinguished against this time
steady background. The intensity of the
secondary atomization is subjectively
graded by a scale similar to that
developed at Princeton (9).
Spray Characterization Rig
The spray characterization rig was
developed to determine dropsize
distributions of small seale- atomizers
under cold flow conditions. The rig,
illustrated in Figure 2, consists of a
plexiglass cylinder in which the nozzle is
mounted on centerline downfired. Air is
co-flowed axially around the nozzle to
simulate the combustion air field and to
prevent recirculation of drop-
lets into the optical path.
Figure 1. Slip-flow reactor.
Figure 2. Spray characterization system.
184
-------
Two ports at opposite sides of the
chamber provide access for the Malvern
2600 HSD particle size analyzer. The
Malver measures dropsize distribution by
measuring the diffraction of a laser beam
as it passes through the spray field. The
diffraction pattern is collected by a
Fourier transform lens and is focused onto
a detector array. This can
two parameter size fit (e.g.,
or Rosin-Rammler) or a model
fit capable of resolving
distributions.
Turbulent Flame Reactor
SAMPLE POINT
MIXING BAFFLES
be either a
, log-normal
independent
multi model
The Turbulent Flame Reactor (TFR) was
designed to simulate a number of aspects
of the flame zone performance of liquid
injection incinerators. These include
swirl,recirculation , broad drop-size
distribution, and high variation in
droplet number density. It is parti-
cularly important that the reactor be
capable of simulating the compound escape
mechanisms that can occur for flame zones
of liquid injection incinerators. Very
high heat removal rates are utilized to
quench post-flame reactions. Thus, the
destruction which occurs in the turbulent
diffusion flame is emphasized over non-
flame decomposition which occurs in the
post-flame region.
The design of the reactor is detailed
elsewhere (7). The reactor is shown in
Figure 3; it consists of a 30 cm diameter
by 90 cm long stainless steel cylindrical
enclosure with water-cooled walls. The
burner consists of a pressure-atomized
nozzle (Delavan WDA series) located level
with the bottom plate of the reactor as
shown in Figure 3. The nozzles have a 60°
angle hollow coned spray pattern and are
used in sizes corresponding to 1.9, 2.85,
3.8, 5.7 liters/hour (0.5, 0.75, 1.0, and
1.5 gallons/hr). The fuel, here No. 2
fuel oil, is doped with waste compounds
and supplied from a pressurized tank.
The main burner air is introduced
through the annular space around the
nozzle. A research-type variable swirl
block burner is used to introduce the
burner air. The burner air is supplied
from the compressed air supply and metered
by a venturi meter prior to introduction
into the swirl burner. Gas samples are
obtained downstream of a series of mixing
baffles at the reactor exit by an uncooled
stainless steel probe.
SIGHT-GLASS
COOLING
WATER
SWIRL VANES
BURNER AIR FLOW
Figure 3. The turbulent flame reactor.
Volatile Organic Analysis
The ORE of the waste compounds was
measured in the exhaust of the turbulent
flame reactor by use of a Nutech Volatile
Organic Sampling Train (VOST). This train
is described in detail elsewhere (6). In
brief, gas samples are drawn through chil-
led cartridges within which the volatile
organic compounds are absorbed onto Tenax-
GC. After sampling the compounds are
released by thermal desorption and anal-
yzed on a gas chromatograph equipped with
a flame iomzatlon detector (7).
RESULTS AND DISCUSSION
Atomizer Characterization and ORE
Performance~
The purpose of the tests reported in
this section was to quantitatively char-
acterize the droplet size distributions
from the test nozzles and to compare these
characterizations with ORE performance.
The nozzles were operated at both the de-
sign points and under off-design condi-
tions to obtain atomization qualities
characteristic of failure modes. These
results were compared with the DRE ob-
tained using these nozzles in the TFR.
The Delavan WDA series nozzles used
in this study generate a .6J)° hollow spray
185
-------
pattern. Liquid is forced through small
slots under pressure into a swirl chamber.
The swirling liquid leaves the chamber
through an endport and establishes the
characteristic hollow-coned spray pattern.
Thus, the energy needed to overcome
surface tension and form droplets Is
supplied by the fuel pressure drop across
the nozzle. The nozzles are supplied for
fixed flow rates; these are set to yield
the required pressure drop across the
nozzle for design operation. Substitution
of a larger capacity nozzle at the same
flow rate will reduce fluid pressure and
nozzle performance. This is illustration
1n Figure 4 where the area-mean diameter
is plotted against fuel pressure for four
nozzles. The design pressure is approx-
imately 200 psi. As flow and pressure are
reduced the mean diameter increases from
30 to over 70 microns. However, changes
1n mean size do not fully represent the
situation. The change in the size dis-
tribution is key because evaporation time
limited behavior will tend to be dominated
by the largest droplets present.
130
100 150
Nozzle Pressure (psl)
Figure 4.
Droplet diameter as a function
of nozzle pressure
Figure 5 illustrates the droplet size
distributions obtained at 3.8 liters/hour
(1.0 gallons/hour). The data set labeled
"Design Operation" was for a 3.8 liter/
hour capacity nozzle, and thus represents
the size distribution resulting from
correct operation. The "Off-Design" data
set is for the identical flow rate, but an
oversized (5.7 liters/hour) nozzle. As
described above, use of oversized pressure
atomized nozzles result in low fluid pres-
sure, and low atomization energy. Thus,
the off-design results show the dropsize
600 400 200 100 80 60 40 20 10
Diameter, microns
Design Operation
Off-Oestcn
600 400 200 100 80 60 40
Diameter, microns
Figure 5. Droplet size distributions
and estimated evaporation
time as a function of diameter
is shifted toward larger values. The key
to interpreting the effect of the shift in
droplet size is found in the evaporation
time plot of Figure 5. This plot shows
evaporation time as a function of droplet
diameter for No. 2 fuel oil, based on the
"d^ law" (5). The on-design data shows
approximately 10 percent of the mass is
above 170 microns. According to the
evaporation rate plot, this 10 percent
will require more than 50 msec to
evaporate. The off design data indicate
that fully 46 percent of the mass is
greater than 170 microns. Also note that
the largest size class (250-560 microns)'
has increased from 2 to 16 percent of the
total mass. Since the evaporation times
for this category range from 100-700 msec,
it is evident that the effect of moving
from on-to-off design operation is a sub-
stantial increase in the evaporation time
of a significant fraction of the fuel.
The on-design and off-design atomizer
conditions were used in the TFR to deter-
mine the influence of atomization quality
on ORE. The No. 2 fuel oil was doped to
3.0 weight percent with an equimolar
mixture of test compounds. These test
186
-------
compounds were selected to represent the
various classes of organic hazardous
wastes; they were acry1 onitri1e ,
chloroform, chlorobenzene, and benzene.
These compounds were used in our previous
subscale testing, and a discussion of
their incineration related properties is
available elsewhere (7).
The TFR was operated under both the
on-design and off-design conditions as
defined above, and theoretical air was the
independent variable. Figure 6 shows the
fraction of each of the compounds that
escaped-destruction for each of the nozzle
conditions.
0.005
O
U.
0.004
0.003
0.002
O 0.001
O
0.005
i 0.004
0.003
"g 0.002
i 0.001
o
o
o '
DESIGN OPERATION
60 ~100 140 180 220 ~2gO
THEORETICAL AIR, percent
Figure 6. Waste destruction for the nozzle
conditions identified in Figure 5.
The on-design nozzle results show be-
havior similar to those documented in our
previous work (7,8):
• A range of high ORE between 100-200
percent theoretical air.
• At low theoretical air the increased
waste emissions indicate a failure mode
due to fuel-rich pockets breaking
through the flame.
• At high theoretical air the increased
waste emissions indicates a quenching
failure mode in which the high air flow
is quenching portions of the flame
prior to complete reaction.
Comparison of the on-design and off-
design plots shows that the emissions at
the rich and lean failure modes are not
significantly different. However, the ORE
in the region between 100 and 200 percent
theoretical air has degraded markedly from
the previous high efficiency. Thus, the
change in atomization quality that
accompanied the use of the oversized
nozzle induced an atomization failure
mode; the ORE, which was much greater than
99.99 percent was reduced to the order of
99.9 percent.
Two general mechanisms can be identi-
fied by which poor atomization can influ-
ence ORE. In the first, droplets which
are too large to evaporate in the
available time penetrate to the reactor
wall. The liquid evaporates and exits the
reactor along the cold boundary layer at
the wall. In the second mode, the
droplets penetrate through the flame-zone
without fully evaporating until well into
the post-flame region. Here, mixing or
temperature may not be sufficient to
ensure complete destruction.
Estimation of the maximum droplet
diameter for which the droplets avoid
striking the wall involves 1) determin-
ation of fraction of the hydrodynamic
energy released by the nozzle that is
converted into droplet velocity, and 2)
determination of the aerodynamic drag on
the droplets as they simultaneously evap-
orate and burn. While such calculations
cannot be performed to a great degree of
accuracy, the estimation indicates that
the threshold diameter fpr striking the
wall is approximately 200-300 microns.
This is consistent with the shift in ORE
behavior associated with the spray degra-
dation and it indicates the following
methodology for evaluating the atomization
adequacy of full-scale nozzles:
• Evaluate atomization quality in cold
flow on either the actual waste stream
or on a surrogate stream of Identical
properties. If possible, both dropsize
and droplet velocity information should
be obtained.
187
-------
• Use the spray information to evaluate
the adequacy of the match between the
combustion chamber and the spray
pattern.
The actual evaluation of a full scale
setup by this methodology cannot presently
be done to a great degree of accuracy.
However, the approach does indicate future
research directions.
Influence of Secondary Atomization on ORE
Large-scale atomizers generally fail
to provide acceptable atomization quality
for two reasons:
1. The liquid is unusually viscous or it
contains solids (i.e., slurry).
2. Portions of the nozzle have degraded
during use such that design operation
cannot be obtained.
Since the liquid hazardous waste desig-
nation covers a wide variety of fluids and
slurries, there is little doubt that a
certain portion of these will be difficult
to atomize. Also, many wastes have, cor-
rosive and abrasive properties that will
accelerate nozzle wear during service.
Thus, under certain conditions primary
atomization quality can be expected to be
a limiting factor in overall ORE.
As discussed in the first section of
this paper, the secondary atomization
phenomena has been investigated as a means
of Improving combustion efficiency through
the in-flame reduction of droplet diameter
via fragmentation. Reviews of the exper-
imental data (9) and theory (12) are
available in the literature. Most wastes
are multicomponent mixtures of varying
volatility. Since waste atomization can
be a critical deficiency in the incin-
eration process, secondary atomization can
occur either naturally or be induced by
addition of a volatile component as a
means of improving performance.
The two questions addressed on secon-
dary atomization in the present work were
1) whether secondary atomization can be
induced by the presence of hazardous com-
pounds 1n fuel oil, and 2) whether this
secondary atomization has the capability
of improving ORE. Five compounds were
selected for doping into No. 2 fuel oil
for secondary atomization screening.
These were selected to represent a broad
range of boiling points: dichloromethane,
39°C; acrylonitnle, 79°C; benzene, 80°C;
isopropanol, 82°C; and benzal chloride,
205°C. These were also selected to re-
present a wide range of volatility dif-
ferences with respect to No. 2 fuel oil
(boiling point: 210-260°C).
Each of the compounds were screened
in the slip reactor at 0.5, 2, 5, 10, 20,
and 40 weight percent in the No. 2 fuel
oil. The results are presented graphi-
cally in Figure 7 as a plot of secondary
atomization intensity vs. concentration
for each of the compounds. These results
indicate that secondary atomization is
active only for compound concentrations
above two percent (except) for isopro-
panol, which was active above 0.5 per-
cent). Also, for secondary atom.ization to
occur there must be some difference
Violent —.
•g Readily
o Apparent
Regular —
None
Figure 7.
10 100
Waste Concentration in Fuel
Effect of waste concentration on
secondary atomization intensity.
between the boiling points of the base
fuel and the additive. For example, ben-
zal chloride, which has a boiling point
comparable with that of No. 2 fuel oil,
showed no activity at any concentration.
The results indicate that intensity is not
entirely a function of boiling point dif-
ferential. For example, isopropanol has a
boiler point of 82°C, but it induced a
substantially more active reaction ,than
dichlormethane (39°C). Thus, other
factors than boiling point differential
188
-------
(e.g., compound polarity) are related to
intensity.
The screening tests indicated that
isopropanol and benzal chloride represent
the limits of secondary atomization in-
tensity. These two compounds were
selected for ORE testing in the TFR. The
compounds were tested at 0.5, 2.0, and
10.0 weight percent in No. 2 fuel oil.
The experiments were designed to determine
the effect of compound concentration ORE
for 1) a compound for which no. secondary
atomization occurs across the entire
concentration range, and 2) a compound for
which no secondary atomization occurs at
low concentrations, but a strong response
is obtained at high concentrations. Thus,
benzal chloride yields the concentration
dependence in the absence of secondary
atomization. Any strong additional con-
centration dependence for isopropanol can
be attributed to an increase in secondary
atomization intensity with concentration.
The test condition corresponded to
the off-design atomization condition
illustrated in Figure 5. In all other
respects, the TFR was set for high effic-
iency operation (120 percent theoretical
air, 0.8 swirl number). Thus, the only
variables were test compound type and con-
centration.
The results for ORE of the test com-
pounds are shown in Figure 8. Waste
penetration (fraction of original waste
escaping the reactor) is plotted against
the percent waste in the fuel for the two
test compounds. Benzal chloride shows an
approximately one order of magnitude de-
crease in penetration between 0.5 and 10
percent waste concentration. Since no
secondary atomization takes place for this
compound, the concentration effect on
penetration must be due to other factors.
For isopropanol, however, the effect of
concentration is much more pronounced.
Between 0.5 and 10 percent concentration
ORE improves from less than 99.9 percent
to greater than 99.9999 percent. Sig-
nificantly, this increase in ORE occurs
concurrently with an increase in secondary
atomization intensity from none to vio-
lent. Thus, at least a substantial
portion of the difference in behavior
between benzal chloride and isopropanol
can be attributed to the secondary
atomization behavior of isopropanol.
Figure 8. Comparison of compound pene-
tration for benzal chloride and
isopropanol as a function of com-
pound concentration in the auxil-
iary fuel. Results are for the
turbulent flame reactor operating
under an atomization failure
condition.
This work suggests that the ORE of
liquid injection incinerators operating
under atomizer limited conditions can be
improved by the blending of small amounts
of high volatility liquids into the waste
stream. The blending agent may be a
second waste stream of markedly different
volatility rather than a pure organic
liquid. These blending agents may be
particularly appropriate for slurry
atomization, whose primary atomization
quality is usually limited.
CONCLUSIONS
The results of this study have shown
that atomization quality and ORE are
strongly correlated. The specific
conclusions are:
• Degradation of atomization quality
appears to influence ORE primarily
through the penetration of droplets
through the flame zone or to the
wal1s.
• A potential methodology for
evaluating nozzle performance in
relation to a particular incinerator
geometry was identified. This
consisted of evaluation of the nozzle
for spray angle, droplet size and
droplet velocity. This information
189
-------
is matched with the incinerator
geometry to yield the approximate
maximum tolerable droplet size to
penetration.
avoid wal1 or firebal1
• Secondary atomization has been
indicated to be a mechanism for
improving ORE in situations in which
atomizer performance limits ORE. A
means of obtaining this effect and
improving ORE could be the selective
blending of waste streams of varying
volatility.
ACKNOWLEDGEMENTS
The authors wish to thank Mr. Howard
0. Crura for his contribution to the
experimental portion of the program. This
work was supported by the U.S.
Environmental Protection Agency—Air and
Energy Engineering Research Laboratory,
Research Triangle Park, NC--through
Contract 68-02-3633.
REFERENCES
1.
2.
3.
4.
Edwards, J.
Formation
species,
Arbor, MI.
B., 1974. Combustion;
and Emission
THfn
of Trace
Arbor Science. Ann
5.
Environmental Protection Agency, July
16, 1980. Hazardous Waste and
Consolidated Permit Regulations,
Federal Register, 45:138.
Environmental Protection Agency, July
16, 1980. Incineration Standards for
Owners and Operators of Hazardous
Haste Management Facilities, Federal
Register, 46:138.
Environmental Protection Agency,
January 2, 1981. Incineration Stand-
ards for Owners and Operators of
Hazardous Waste Management Facili-
ties, Federal Register, 46:264.
Glassman, I., 1977. Combustion Aca-
demic Press, New York.
Jungclaus, G. A., P. G. Gorman, G.
Vaughn, L. D. Johnson, and D.
Friedman, 1984. Development of a
Volatile Organic Sampling Train
(VOST), Incineration and Treatment of
Hazardous Waste: proceedings of "tHe~
Ninth Annual Research Symposium, EPA-
600/2-84-086, U.S. EPA.
Kramlich, J.C., M.P. Heap, J.H. Pohl,
E.M. Poncelet, G.S. Samuelsen, and
W.R. Seeker, 1984. Laboratory Scale
Flame-Mode Hazardous waste Thermal
Destructi on Research ,
L-PA-600/2-84-086, U.S. EPA.
Kramlich, J.C., M.P. Heap, W.R.
Seeker, and G.S. Samuelsen, 1985.
Flame-Mode Destruction of Hazardous
Waste Compounds, 20th Int. Symp.
Combust., The Combustion
(in press).
The
Pittsburgh, PA
Institute,
9. Lasheras, J.C., A.C. Fernadez-Pel 1 o,
and F.L. Dryer, 1981. On the Dis-
ruptive Burning of Free Droplets of
Alcohol In-Paraffin Solutions and
Emulsions, 18th Int. Symp. Combust.,
The Combustion institute, Pittsburg,
PA, p. 293.
10. Resource Conservation and Recovery
Act, 1976. Public Law 94-580.
11. Trenholm, A, P. Gorman, B. Smith, and
D. Oberacker, 1984. Emissions Test
Results for a Hazardous Waste
Incineration RIA, Proceedings of the
Ninth Annual Research Symposium on
Tncineration and Treatment of
Hazardous Waste, EPA-600/9-84-015.
U.S. EPA, p. 160.
12. Wang, C.H., X.Q. Liu, and C.K. Law,
1984. Combustion and Microexplosion
of Freely Falling Multicomponent
Droplets, Combust. Flame 56, 175.
190
-------
EVALUATION OF A PILOT SCALE CIRCULATING BED COMBUSTOR
WITH A SURROGATE HAZARDOUS WASTE MIXTURE
Daniel P.Y. Chang and Nelson W. Sorbo
Department of Civil Engineering
University of California, Davis
ABSTRACT
Circulating bed combustors (CBC) appear to be an emerging technology for the destruc-
tion of hazardous wastes. A cooperative study among the CBC manufacturer, the California
Air Resources Board, the California Department of Health Services and the Environmental
Protection Agency was carried out on a pilot-scale CBC. A surrogate waste mixture having a
heating value of about 19 MJAg (8000 Btu/lb) and comprised of water, xylene, ethylbenzene,
toluene, hexachlorobenzene, 1,2,4 trichlorobenzene, Freon 113, and carbon tetrachloride was
fed to a CBC at a heat release rate of about 530 MJ/hr (0.5 MMBtu/hr). The destruction and
removal efficiency (ORE) and formation of products of incomplete combustion (PIC) were eva-
luated. Samples were drawn throughout an 18 hour test period for volatile and semi-volatile
compounds. Combustion parameters such as oxygen, total hydrocarbons, carbon monoxide, car-
bon dioxide, oxides of nitrogen, CBC bed temperature and air flowrate were also measured
continuously. In addition, a refractory tracer SF, was injected into the combustion zone
and monitored on a continuous basis as well as by grab samples. Results of those tests
indicate that the CBC was being operated in a near "failure mode" condition throughout much
of the test. Nevertheless, 'high ORE was observed (>99.99%) for the majority of samples.
Formation of volatile PIC chlorine compounds appeared to be correlated with carbon monoxide
and total hydrocarbons.
INTRODUCTION
The California Air Resources Board
(CARS) staff evaluated emissions from a
circulating bed combustor (CBC) owned and
operated by GA Technologies, Inc., (GA) San
Diego, California. This test.was the first
in a series of three, funded by the
Environmental Protection Agency (EPA) and
the California Department of Health
Services (DOHS), to evaluate the perfor-
mance of different types of hazardous waste
incinerators. The goals of this study were
to evaluate specific combustion units, to
increase the database on emissions of non-
criteria pollutants, and to evaluate
methods for permitting and monitoring such
units in the future. The CBC was included
in this study because of its potential
application to on-site waste destruction by
generators of small quantities of waste.
The overall objective of the GA test
was to evaluate the operational charac-
teristics of the CBC when fueled with a
simulated liquid waste mixture. This
included defining an envelope of operation
of the CBC using typical parameters such as
CO and THC in the effluent gases, and
measuring the destruction and removal effi-
ciencies (ORE) of a number of refractory
compounds under various combustion con-
ditions. A complete description of all
aspects of this project is found in
Reference 1.
191
-------
DESCRIPTION OF CIRCULATING BED COMBUSTOR
The CBC is an outgrowth of coal com-
bustion technology involving fluidized bed
combustion. The bed solids in a CBC are
continually carried out of the top of the
main combustor, collected by a cyclone, and
continuously returned to the bottom of the
combustor, while the bed solids in a con-
ventional fluidized-bed combustor maintain
a relatively stationary position.
Operation in a circulating mode offers the
advantage of a greater range of turndown
ratios and improved mixing and temperature
uniformity over that of conventional com-
bustors (2). The CBC is operated at
substantially lower temperatures than con-
ventional incinerators resulting in relati-
vely low NO concentrations. The mean gas
temperature!? in a CBC are maintained below
870 C (1600UF) to prevent slagging of the
bed material. Complete oxidation of the
fuel is enhanced by mixing of bed solids
with excess oxygen and by a substantial
residence time afforded by recirculation of
solids. The presence of reactive bed
solids offers the potential for capture of
halogen and sulfur reactants in the bed
materials. GA has demonstrated S00 and HC1
removal ranging from 88 to >99% (37.
The nominal thermal rating of the CBC
tested was 2100 Md/hr (2 MMBtu/hr). it had
an inner diameter of 40 cm (15.5 in), and a
height of 7.6 m C25 ft). Bed solids were
stored externally and transported pneumati-
cally to the bed as needed to maintain the
quantity of material in the bed. The test
was designed to determine the response of
the CBC to an aqueous-organic fuel mixture.
Therefore no solid fuel was used during the
test except during the initial warm-up
period. Natural gas was used to augment
the heat release rate as needed during the
test to maintain the combustion bed tem-
perature. A forced draft fan supplied pri-
mary and secondary combustion air to the
combustor. Solids carried out of the bed
were collected in a cyclone and returned to
the bed through a proprietary return seal.
Gases exiting the combustor were cooled by
a liquid water heat exchanger.
Supplemental cooling and dilution air were
permitted to enter between the gas cooler
and the inlet to the baghouse. Gas flows
were drawn into the baghouse by two induced
draft fans, one of which failed during the
test. The CBC was operated so that
atmospheric pressure was maintained at the
exit of the combustor.
GA gas sampling instruments (C00, CO,
THC, 0 , NO , HC1, S09) were Itcated
downstream of the gas cooler and upstream
of the dilution air inlet. In addition, GA
continuously recorded pressure drop,
average bed temperature, and total com-
bustion air. The volumetric flowrate of
the dilution air stream was estimated by
chemical mass balances of N9, C00, and
0, - concentrations. The dllutidn air
flowrate was found to be about 10% of the
combustion air flowrate and thus this value
was assumed throughout the test period.
Fuel and water flowrates were also recorded
periodically. CARB instruments (C00, CO,
THC, CH , Q NO , S0?) were focated
downstream of th£ bagfiouse. Tracer
Technologies injected sulfur hexafl'uoride
(SF-) into the CBC and monitored the SF-
with a semi-continuous GC/ECD system
located at the CARB sampling point.
Integrated samples were obtained by CARB
for volatile organic compounds (Tedlar bag
samples) and semi-volatile organic com-
pounds (XAD-2 resin) for the ORE and PIC
determinations. Bag sample analyses were
carried out by CARB's Haagen-Smit
Laboratories in El Monte, California.
Analysis of the XAD-2 resin samples were
performed by the California Air Industrial
Hygiene Laboratory in Berkeley.
California.
The fuel mixture used for this test
was selected to simulate a •refractory
hazardous waste stream and to provide com-
parability to other EPA tests. The com-
position and analysis (% total mass) of the
fuel used for this test was: xylene
(74.38*), ethylbenzene (20.975K), Freon 113
(1.00%), toluene (0.35*), hexachlorobenzene
(0.26*). The heating value of this fuel
was about 39 MJ/kg (17,000 Btu/lb).
Because the test was designed to simulate
the combustion of an aqueous mixture of
organo-chlorine fuels, water was dynami-
cally mixed with the surrogate fuel to
achieve an effective heating value of about
19 MJAg (8000 Btu/lb).
RESULTS AND DISCUSSION
The operation of the CBC. was charac-
terized by combustion parameters and by
analysis of bag samples, resin samples, and
SF6 ORE data.
192
-------
Combustion Parameters
A graph of CO, THC, CU, and bed tem-
perature as a function of run time is pre-
sented in Figure 1.
gas
The unsteady nature of the combustion
composition is evident. A careful
examination of this figure would suggest
that there was a relationship between CU,
CO, and THC concentrations. To evaluate
this relationship the data were stratified
for 02 concentration into 1% intervals and
the averages of CO and THC of all points in
each interval were calculated. The results
of this analysis are shown in Figure 2.
60 120 180 240 300 360 420 480 540 600 660 720 780 840 900 960 102010801140 1200
RUN TIME (MINUTES)
I 1 1 n 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
21 22 23 24 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16
CLOCK TIME (HOURS)
Figure 1. 09 concentration, THC concentration, CO concentration, and bed temperature
versus run time and clock time for the first 1200 minutes of the test.
193
-------
High CO and THC emissions occurred at con-
ditions of low or very high
0? concentrations. An analysis of the tem-
peratures that occurred during high CO and
THC concentrations indicated that low
Ou was responsible for more of the CO and
mC excursions than was low temperature.
The frequency of CO and THC excursions
increased with decreasing 09 levels, par-
ticularly below about 5 to 5% 09. The GA
Technologies THC and CO analyzer^ went off
scale at about 400 and 5000 ppm, respec-
tively, thus the apparent leveling off of
THC and CO readings did not occur in
reality. This range of 0, concentration
was somewhat higher than anticipated in
order to achieve high combustion effi-
ciency. GA Technologies staff suggested
that a possible reason for the deteriora-
tion of combustion efficiency was the size
of the positive displacement pump used to
deliver fuel to the CBC. It had a nominal
rated capacity of about 110 liters per
minute (jppm) (30 gallons per minute), where
as fuel was being fed to the bed at about
0.44 - O.SOJipm (7-8 gallons per hour).
Thus there may have been "surges" of fuel
that resulted in fuel rich pockets of gas
passing through the bed in spite of the
fact that the overall stoichiometry was
slightly fuel-lean. In a sense, the CBC
was being operated in a "failure mode" or
" non-optimum combustion" condition
throughout the test.
10.000
1,000
100
10
T T
o _
00
000
o = co
• = THC
1 I 1 1 I i
1 I
0 2 4
8 10 12 14 16 18 20
Figure 2. Comparison of average THC and CO
concentrations as a function of
O concentration.
A minimum in THC can be discerned from
Figure 2. The considerable scatter in the
data at high 02 concentrations can be
explained simply as a result of the small
number of data points in this range. It is
reasonable to hypothesize "quenching reac-
tions" at extreme 09 concentrations. This
form of "quenching^' could result from
pockets of fuel lean mixtures where reac-
tion propagation may have been hindered.
Based on a careful analysis of Figure
1 it was founds a) increases in CO and THC
concentrations were correlated with
decreases in 0^ concentrations, b) rapid
fluctations of CO and THC concentrations
were virtually independent of the average
Oy concentration, and c) the response of
tne 0~ analyzer lagged the responses of the
THC arid CO analyzers by about 3 minutes
(1).
Low combustion temperature appears to
have resulted in excessive CO and THC
emissions on three occasions. Two were
momentary periods when the bed temperature
fell below 700°C (1300°F) and were accom-
panied by CO and THC spikes at 825 and 1012
minutes (Figure 1). The other period was
during the start-up of the CBC.
Bag Samples
Bag samples were drawn over nine
sampling periods. These were analyzed for
the POHCs Freon 113 and carbon tetrach-
loride, and for a variety of PICs including
chloroethane, chloromethane, dich-
lorodifluoromethane, dichloromethane,
trichlorofluoromethane, trichloromethane,
1,2-dichloroethane, 1,1,1-trichloroethane,
1,2-dichloropropane, trichloroethylene,
tetrachloroethylene, and benzene.
Concentrations were expressed as parts per
billion (ppb) on a dry basis, and average
values of fuel flowrate, air flowrate,
total gas flowrate, and effluent gas con-
centrations are presented in Table 1.
Concentrations of CO and THC normalized to
the C02 concentration also appear in Table
1. Tff provide an indication of volatile
organo-chlorine PIC formation, a parameter
^Volatile PIC-Cl/Cl-in is also presented
in Table 1. This parameter is defined as
the ratio of the mass of chlorine in
chlorinated PICs to the total chlorine
input with the fuel.
Based on an examination of the bag
sample analyses shown in Table 1,
194
-------
1/3
D
LL]
py
UJ
a.
UJ ">
-> O
to r.
^
u.
O
>•
"^
s
2
D
1/3
O
O
o
i
o
o
s
o
JQ
CN
O
a-
CN
A
O
CN
a>
§
£
ON
O
V -
>5 ^3
H 3
en
(N
ve
•*£
^jj
oo
a-
2
a
^H
•»4
|
i-H
IA
O
3
en
ON
O
c
.2 /
43 /
E /
•y /
S /
u /
S /!
.2 /z
I/I
°0/ 1
<«/ S
«Y U
3
a
JU)
u
1
IA IN* r^ a- oo vo
•a- ON O • en o
O ON 1 ON 1
._; ON uj ON UJ
° ON >A ON en
ON IA ON CN
°> CN . °J NO
0 0
IA r^ r^ CN ON VD
CN ON O ON O O
0 ON 1 ,-i ON 1
• ON UJ ° ON UJ
° ON f^ ON VD
ON •— « ON O
d "^ d °^
en VD 1^ f*^ c^ VD
VD ON O «N 00 O
o ON i ,.5 ON i
• ON UJ ° ON UJ
° ON CN ON CN
ON ON ON en
ON ^! ON *
d d
CN a° t^ CN VD VD
a ON O CN r^ O
O ON 1 ^ ON 1
ON oo ON en
d d
VD VO l^ ~~* ON NO
3
a
t
V)
o
cu
_s oo a VN,
• en M °
Cm oo i
O O uj
vo
j oo ON a-
CN CN ^
**J
r^ r^ a-
" -5 IA °
c ^ ^ £
S
(*^
O OO 00 C«N.
§ oo en °
ft" § 1 %
"H
Cd °i ^ g
c r2 i2 '
VD VO (ij
a-
jf
j a- oo en
CO • • O
-! 00 VD |
s s s
— H
• O ch f*>
(D • • f*^
e ^" ON i
^*" ill
*5 S t^
•a- *r oo
O -^ ON C«N.
^ r^ *^ LU
° ° S
j oo oo ch
nj • • o
gS »^ ON |
r— i
a „ -s
i U -H
11 tl
ro CQ " cu
§ !l 5 ^
I I1 g I
ITN cn fN O f1^ f\ ij- O ch
"• r««v _^ »o ON ^^ ^? **^ ^*
fx. ^ o UJ UJ
f*A vo — *
tn» (N
CN f^l
-:^^^2°^5»
1^ ^ 0 — i ^ UJ UJ
r*j -sj"
cNa-a-t^a-enen .j „»
—Ii **^ r«^ t*^ C^ '^ ^? * •
\o ^^ w^ ™^ uj
-------
generalizations can be drawn regarding the
performance of the CBC during this series
of tests. Although the average CO con-
centrations were high (greater than 500 ppm
in seven of nine samples), the DREs of the
volatile POHCs (Freon 113 and carbon
tetrachloride) were also high in most
cases. However, the penetration of organo-
chlorine compounds ( volatile PlC-Cl/Cl-in)
was greater than 1 x 10" . The predominant
halogenated PICs identified were chloro-
methane and chloroethane. The 0934 bag
sample yielded low ORE for Freon 113,
appreciable amounts of fluorinated PICs,
and the lowest SF. ORE. An examination of
the combustion parameters during this
period (Figure 1) suggests that low bed
temperature was the probable reason for
reduced ORE of fluorinated compounds.
A plot of the normalized PIC chlorine
content against both CO/CXU and THC/C09 is
presented in Figure 3. i\ linear least-
squares regression analysis was performed
on the logarithms of the respective quan-
tities yielding the dashed line correla-
tions shown in Figure 3. The correlation
of THC/CXU with PIC-chlorine formation is
high with"% regression coefficient of about
0.94. The
PIC-chlorine
correlation of
formation was
CO/CXU with
IXKJ"
:xio
o
v.
O
IXIO
o CO/COZ
a THC/COZ
• Oofs Not Used in
CO/COZ Regression
Analysis
a
logY»7.70+1.38 loqX~yf
r«OS4 /
/
~tog Y = -0.64* 1.22 logX
r»054
IXK)
IXIO ixio
THC/COZ or C0/C02
Figure 3. Correlation of volatile PIC
chlorine normalized by total fuel
chlorine input versus THC and CO
concentration normalized by the
C02 concentration.
not as high if all points were considered.
The correlation of CO/CXU with PIC-chlorine
formation was low as a result of two points
(marked as • in Figure 3). One of the
points which exhibits high CO but low PIC
chlorine can readily be explained since it
occurred during a period (0934 - 1035,
Table 1) when natural gas was being
injected into the bed to supplement the
fuel heating value. Thus it was eliminated
from subsequent regression analyses. The
second questionable sample (low PIC and
very low CO) was taken during a period
(0030 - 0100, Table 1) when water was not
being mixed with the fuel. All other bag
samples were taken during periods when an
approximately constant amount of water was
mixed with the fuel. During the 0030 -
0100 period, both the fuel 'heating value
and the bed temperature were higher than
those occuring in other bag sample periods.
Thus the point corresponding to 0030 - 0100
period was eliminated from the regressioon
analysis. The resulting correlation coef-
ficient of CO/CXU with PIC-chlorine for-
mation was alsa 0.94. The results
presented in Figure 3 are qualitatively
similar to those presented by Waterland
(4).
Resin Samples
XAD-2 resin samples were collected
over five distinct time periods during this
test. The POHCs analyzed were toluene,
hexachlorobenzene, ethylbenzene, xylene,
Freon 113, and trichlorobenzene.
Concentrations, ORE, and penetration values
were calculated for each POHC. The only
PIC recovered by the resin analysis was
benzene, although toluene was probably both
a POHC and a PIC.
Based on the resin analyses it was
determined that the DREs of each compound
detected were fairly uniform throughout the
series of samples. Toluene had the lowest
ORE of all POHCs reported with ORE values
exceeding 99.9% in all samples except one
where the ORE exceeded 99.99%. All other
POHCs showed excellent DREs. For example,
trichlorobenzene had DREs exceeding 99.999%
for all samples while hexachlorobenzene,
ethylbenzene and xylene exhibited DREs
greater than 99.9999% for all samples. PIC
benzene concentrations ranged from about
5500 ppb to almost 20,000 ppb.
Correlations between the benzene con-
centration, normalized to the fuel rate,
with both C0/C02 and THC/IXU proved to be
insignificant.
196
-------
fl comparison of the results from the
bag and resin samples showed that the only
compound in common was the PIC benzene.
During the time period common between the
resin and bag samples, the benzene con-
centrations were 20,000 and 37,000 ppb
respectively. Although bag and resin
samples exhibited reasonable agreement
during similar time periods, CARB had only
validated the bag sampling technique.
There was evidence of benzene breakthrough
in the resin samples.
SF6 Test
Gaseous sulfur hexafluoride was
injected into the CBC and monitored
throughout the run. The average ORE of the
SF, for the bag and resin sample periods is
shbwn in Table 2. No correlation between
SF, penetration and "volatile PIC-Cl/Cl-in
(Table 1) was found. The only other
apparent feature in these data was a gra-
dual increase of SF, ORE as the test
progressed, corresponding to an increase in
the average bed temperature. Assuming the
CBC to be plug-flow reactor with tem-
perature as the only significant variable
throughout the duration of sampling, an
elementary model assuming an Arrhenius tem-
perature dependence of rate and equilibrium
constants. predicted that the
SF, penetration would have the form*
6
ln(-ln(Pt)) = A + BA
(1)
where Pt represents SF^ penetration and A
and B are constants, eased on this model
and data in Table 2 a regression equation
was developed*
ln(-ln(Pt)) = 9.68 - (9.26E + 3)A (2)
Table 2
Summary of Average
SF, ORE Values
o
for Bag Samples
TIME
ORE
0934-1035
1100-1124
1345-1418
1443-1623
1630-1705
1954-2054
2102-2140
2315-0020
0030-0100
.8342
.9140
.9789
.9786
.9799
.9786
.9826 '
.9917
.9916
This correlation is remarkable in that
the total range of average bed temperatures
for the data sets was only slightly greater
than 100°C, yet about 60% of the variance
appeared to be explained by temperature,
in spite of the fact that fueling rates
and combustion air flowrate were not held
constant. Thus, considering both the bag
sample data on the ORE of Freon 113 and the
SF, data, it appears that temperature is
the key parameter in the thermal decom-
position of fluorocarbons in the CBC.
CONCLUSIONS
Based on an analysis of the data pre-
sented in this paper, the following conclu-
sions can be drawn.
1. The ORE of volatile and semi-volatile
POHCs under less than optimum com-
bustion conditions met RCRA require-
ments (.99.99% ORE).
2. Total volatile PIC formation was found
to correlate well with CO and THC,
normalized to fuel flowrate (C02).
Penetration of volatile chlorinated
PICs (based on total chlorine content
of the fuel) exceeded 1 x 10" . PIC
benzene appeared in substantial con-
centrations in several samples and was
not correlated with any conventional
combustion parameters.
3. The ORE dropped sharply when the bed
temperature fell below 700°C.
Temperature appeared to be a major
factor in the destruction of the
fluorinated compounds and a moderate
correlation between SF, ORE and tem-
perature was observed.
4. An analysis of rapid changes of CO and
THC concentrations compared with
changes in 09 suggests that a minimum
09 level wdtild not be adequate to
guarantee low PIC formation.
5. The CBC seemed to behave as a plug-
flow reactor, susceptible to pockets
of non-stoichiometric air/fuel mix-
tures passing through the bed causing
increased PIC formation. This obser-
vation suggests the importance of the
fuel feed system on CBC performance
and should be evaluated carefully 'by
permitting authorities.
197
-------
ACKNOWLEDGEMENTS
REFERENCES
This project was a collaborative
effort involving several groups of indivi-
duals. We would like to thank Dr. Richard
Flagan, Dr. David Golden, Dr. Randall
Seeker and Dr. Larry Waterland, members of
the GA Circulating Bed Combustor Test
Advisory Panel, for their suggestions and
insights in establishing the goals of the
test program. The actual sampling and
coordination of the test was carried out
under the direction of Mr. Gary Murchison
of the CARS. Without the support of Mr.
Harmon Wong-Woo, Deputy Executive Officer,
CARS, Mr. Peter Venturini, Chief of the
Stationary Source Division, Mr. Dean
Simeroth, Chief of the Engineering
Evaluation Branch and Mr. Bob Adrian,
Manager of the Evaluation Section, this
program would not have come about.
Additionally, Mr. William Rickman and Ms.
Wendy Lessig of GA Technologies provided
the data from the continuous analyzers.
Dr. Lynn Teuscher of Tracer Technologies
kindly supplied the sulfur hexafluoride
data.
Financial support for the project was
initially provided from CARS contract
A2-157-32 and was supplemented by funds
from the Environmental Protection Agency to
the Department of Health Services to the
CARS and finally to the University of
California, Davis, under contract numbers
A3-118-45 and A4-059-45. Dr. Harry Freeman
was the EPA Project Officer.
DISCLAIMER
The statements and conclusions in
this article are those of the authors and
not necessarily those of the California Air
Resoures Board or the Environmental
Protection Agency. The mention of commer-
cial products, their source or their use in
connection with material reported herein is
not to be construed as either an actual or
implied endorsement of such products.
Chang, Daniel P.Y. and Nelson W. Sorbo
in collaboration with the Stationary
Source Division, California Air
Resources Board, 1985. GA
Technologies Circulating Bed Combustor
Test Report. Appendix IV, Final Report
(Draft) contract «A2-157-32.
GA Technologies, 1983. The Circulating
Bed Combustor for Waste Disposal,
Brochure G-392. GA Technologies, San
Diego, California.
Rickman, W.S., 1984. Circulating Bed
Waste Incineration, 6-4999(6). GA"
Technologies, Inc., San Diego, CA.
Waterland, Larry, 1984. Pilot-Scale
Investigation of Surrogate Means of_
Determining POHC Destruction.
Presented at the 77th Annual Meeting
of the Air Pollution Control
Association, San Francisco,
California, (June 24-29, 1984).
198
-------
SUMMARY OF TESTING AT CEMENT KILNS COFIRING HAZARDOUS WASTE
Marvin Branscome, Wayne Westbrook
Research Triangle Institute
Research Triangle Park, NC 27709
Robert Mournighan
U.S. Environmental Protection Agency
Cincinnati, OH 45268
Jon Bolstad, John Chehaske
Engineering - Science
Fairfax, VA 22030
ABSTRACT
The incineration of chlorinated and other liquid organic wastes was investigated in
6- to 9-day test programs at dry- and wet-process cement kilns. Testing was conducted
initially under baseline conditions with no burning of waste fuels. During the waste
fuel burn, the waste fuel replaced a portion of the coal and coal/coke used as the
primary fuel. Different replacement rates were used for each test day. The test program
included monitoring for principal organic hazardous constituents (POHCs), products of
incomplete combustion (PICs), particulate matter, S02, NO , HC1, CO, C02 02, and_total
hydrocarbons. The fate and distribution of chlorine and metals were also determined.
' Test results for POHCs showed that these cement kilns generally could achieve a
destruction and removal efficiency (ORE) of 99..99 percent or greater. Most of the addi-
tional chlorine introduced with the waste fuel is removed with the kiln's waste dust.
An increase in HC1 emissions was observed as the quantity of chlorine entering the kiln
with the waste fuel increased. The data indicate that waste combustion changes the
distribution of lead so a higher percentage of the lead is removed with the waste
dust.
At the wet-process kiln, no significant difference in emission rates of parti-
culate matter and PICs was found between the baseline and waste burn test conditions.
Toluene and benzene were found in trace quantities and are attributable to coal combus-
tion. Particulate matter results at the dry-process kiln are inconclusive because of
a malfunctioning electrostatic precipitator (ESP).
The burning of waste fuel at the wet-process kiln decreased S02 emissions from an
average of .636 ppm at the baseline to 265 ppm. S02 concentrations at the dry-process
kiln remained relatively low with a baseline range of 2 to 12 ppm compared to 5 to 38
ppm during the waste burn.
INTRODUCTION
Preliminary feasibility studies and
tests have indicated that the high-
temperature combustion process of cement
kilns may offer an effective alternative
to other disposal methods for hazardous
waste. The promising characteristics of
cement kilns include:
• Existing high-temperature combustion
process at 1,350° to 1,650 °C (2,500° to
199
-------
3,000 °F) with a gas residence time on
the order of seconds.
• Numerous plants scattered through out
the country, which could handle large
quantities of combustible hazardous
waste liquids.
• Large-scale equipment in place, includ-
ing process control and pollution
control; relatively small capital in-
vestment required.
• Instantaneous temperature excursions
unlikely because of the huge thermal
inertia.
• The alkaline environment in a cement
kiln absorbs HC1 from chlorinated waste
combustion.
• Kilns are operated under draft (slight
vacuum); therefore, there would be
little outward leakage of fumes, mostly
inward leakage of air.
• Ash may be incorporated into the pro-
duct.
• Energy savings from substitution of
waste fuel for oil, coal, or gas.
PURPOSE
The purpose of this study was to
develop additional data on the destruction
of hazardous waste in cement kilns. These
data will be used, along with data from
other tests, to evaluate the environmental
impacts of thi.s method of waste disposal.
To this end, the test program was designed
to:
• Calculate DREs of principal organic
hazardous constituents (POHCs).
• Compare baseline operation (no waste
fuel burned) to the operation with
waste fuel.
• Determine if products of incomplete
combustion (PICs) are formed and, if
so, identify them.
• Determine the fate and distribution of
metals.
• Measure the effect of fuel burning on
HC1 emissions, and determine the fate
and distribution of chlorine through a
material balance.
Examine the effects of waste fuel
burning and process conditions on other
pollutants (e.g., particulate matter,
NO , S02, and total hydrocarbons).
P\
APPROACH
The primary focus of the sampling
efforts was the kiln's electrostatic
precipitator exhaust gas. Volatile
organics were collected in Tenax® and
Tenax®/charcoal sorbent tubes with the
volatile organic sampling train (VOST).
The sorbent cartridges were thermally
desorbed and analyzed by gas chromato-
graphy/mass spectroscopy (GC/MS). Less
volatile organic compounds were collected
by XAD resin in a Modified Method 5 (MM5)
sampling train. Both sorbents were ana-
lyzed to determine POHCs identified in the
waste fuel and were also examined for
products of incomplete combustion. Parti-
culate matter emissions were determined
from the MM5 sampling train. Metals
emissions were estimated from analysis of
the MM5 particle catch.
HC1 emissions were sampled through a
separate sampling train. The impinger
solution was analyzed for HC1 by ion
chromatography. The exhaust gas was
analyzed by continuous emission monitors
for 02, CO, C02, NO , S02, and total
hydrocarbons.
Process samples taken included the raw
material feed, cement product (clinker),
ESP dust, primary fuel, waste fuel, and
process water. The major process streams
were analyzed for metals and chlorine
content to attempt a material balance. In
addition, the waste fuel and water were
analyzed for organic compounds by GC/MS.
Process operating parameters were recorded
at 15-minute intervals throughout each
test period.
The testing sequence first established
baseline operating conditions burning only
the primary fuel and no hazardous waste.
The hazardous waste was then burned at
different fuel replacement rates. This
sequence permitted comparisons between the
baseline and waste fuel test conditions
and also provided information on the
impacts of waste combustion in the cement
kiln.
PROBLEMS ENCOUNTERED
The major problem encountered was
200
-------
methylene chloride contamination of blank
samples at levels roughly equal to those
found from stack gas, samples. The result
is a bias toward high methylene chloride
emission rates and a bias toward ToW DREs.
Blank values were relatively low or nil
for the other POHCs and yielded meaningful
results. For the dry-process kiln test,
one set of samples was invalidated because
of lengthy storage time, cracks in sorbent
tubes, .arid contamination. The ESP malfunc-
tioned during the test on the dry-process
kiln, and no meaningful data on controlled
particulate matter emission rates were
obtained for this test.
RESULTS
DREs
The designated POHCs for the two tests
were methylene chloride, methyl ethyl
ketone (MEK), toluene, 1,1,1-trichloro-
ethane (TCE), and l,l,2-trichloro-l,2,2-
trifluoroethane (Freon 113). The concen-
tration of the POHCs in the waste fuel
ranged from 0.3 to 4 percent. The waste
fuel was spiked with Freon 113 to deter-
mine the ORE of a compound that is diff-
icult to destroy by combustion and that
was unique to the waste fuel. The ORE
results are summarized in Tables 1 and 2.
.Results for methylene chloride are environ-
mentally conservative and biased low
because sample levels and blank levels
were about the same. According to the
VOST protocol, blank corrections were not
applied because blank levels could not be
distinguished from sample levels. No
significant blank problems were experi-
enced with the other POHCs. ORE calcula-
tions for Freon 113, MEK, and 1,1,1-TCE
show 4 to over 5 nines. Toluene was shown
to originate from coal combustion. Toluene
emission rates were unchanged by the
cofiring of waste fuel. Therefore, the
ORE for toluene entering only with the
waste fuel is likely to be much higher
than was measured. Even without subtract-
ing the baseline level of toluene, DREs of
99.99 percent or greater were achieved at
the dry-process kiln and one day at the
wet-process kiln. The highest DREs for
both tests were observed for compounds
that had the lowest baseline and blank
levels (Freon 113 and 1,1,1-TCE) with DREs
on the order of 99.999 percent or greater.
The MM5 results from the dry-process
kiln test were also examined to determine
the DREs of compounds that were not pre-
designated as POHCs. Styrene, ethyl-
benzene, o-xylene, and benzaldehyde were
present in the waste fuel at concentra-
tions of 10 to 20 g/L.• Benzaldehyde was
detected in the stack gas for both the
baseline and waste burn tests and yielded
a ORE of 99.998 percent without correction
(subtraction) of the baseline level. None
of the other three compounds were found in
the stack gas, which resulted in DREs
greater than 99.999 percent for each based
on detection limit values.
A review of kiln tests indicate that
the highest DREs and lowest emission rates
were observed when baseline or blank levels
of the POHC were clearly not a problem.
The two tests with apparently the least
interference in measurements were at
Rockwell Lime (5) and Stora Vika (1,2),
where wastes similar to those previously
discussed were burned. At Stora Vika,
DREs consistently exceeded 4 nines and
included 6-7 nines for several compounds.
Similarly, at Rockwell Lime the DREs con-
sistently exceeded 4 nines and were as
high as 6 nines for several chlorinated
compounds. Background and blank inter-
ferences were noted at St. Lawrence Cement
(9); however, DREs exceeding 4 nines'were
obtained on a worst-case basis by not sub-
tracting background levels, the results
for the test at Los Robles were limited by
method detection limits rather than by
interferences, but DREs exceeding 4 nines
were generally observed (8). A test with
PCBs at Peerless Cement (12) found more
PCBs entering with slurry feed (river water)
in the cooler end of the kiln than was
emitted during the PCB burn. However,
DREs of 4 to & nines were reported 'without
correction for the contribution from the
water.
PICs
Organic compounds that were potential
PICs from waste combustion were emitted at
similar rates during the waste fuel and
baseline tests. At the wet-process kiln,
compounds such as toluene, benzene,s
xylene, biphenyl, naphthalene, and methyl-
naphthalenes were observed in the stack
emissions with coal as the only fuel.
.During both the baseline and waste burn
tests, benzene emission rates ranged from
15 to 50 mg/s. Emission rates of the
other compounds were on the order of 1 to
10 mg/s. The highest emission rates were
observed on a test day with a kiln upset
(ring formation) with coal as the only
201
-------
TABLE 1. DREsa AT THE WET-PROCESS KILN
Compound
tj
Methylene chloride"
Freon 113
Methyl ethyl ketone
1,1, 1-Trichloroethane
Toluene
59%
99.998
99.991
99.991
99.952
Percent DRE at Given Replacement Rate
43%
99.995
99.978
99.991
99.940
61% ,
99.956
99.990
99.996
99.974
39%
99.975
>99.999
99.983
99.996
99.951
58%
99.993
>99.999
99.997
99.999
99.988
Average
99.983
>99.999
99.988
99.995
99.961
TABLE 2. DREs AT THE DRY-PROCESS KILN
Compound
h
Methylene chloride"
Freon 113
Methyl ethyl ketone
1,1, 1-Trichloroethane
Toluene
Percent DRE at Given Replacement Rate
25%
99.94
99.999
99.997
>99.999
99.992
37%
99.99
99.999
99.999
>99.999
99.998
Average
99.96
99.999
99.998
>99.999
99.995
. Uncorrected for blanks.
c Blank values were comparable to sample values for methylene chloride.
Trace quantities of toluene are produced from coal combustion,.
TABLE 3. CHLORINE RESULTS FOR THE WET-PROCESS KILN
Condition
Baseline
Day 5
Day 6
Day 7
Day 8
Day 9
Cl in fuel
(kg/hr)
6
32
21
41
72
128
HC1 emissions
(kg/hr)
0.57
0.27
1.0
1.5
2.3
5.4
Cl in dust
a)
0.2
0.7
0.7
0.7
1.0
1.7
Cl in clinker
(ppm)
<200
705
<200
<200
<200
<200
TABLE 4. CHLORINE RESULTS FOR THE DRY-PROCESS KILN
Condition
Baseline
Day 3
Day 4
Day 5
Clinker
Cl in fuel
(kg/hr)
10
60
71
72
results were <125
HC1 emissions
(kg/hr)
1.3
2.2
5.5
27.
ppm chlorine.
Cl in recycle
dust (%)
0.73
1.3
1.4
1.9
Cl in waste
dust (%)
3.5
5.7
6.1
8.1
202
-------
fuel. No statistically significant in- ,
crease in emission rates was observed when
the waste fuel was burned. No polychlorj-v
nated dibenzodioxins or dibenzofurans were "
found in any samples at a detection limit-,-
of less than 1 ppb by weight in the stack'
gas. Similar results were obtained at the
dry-process kiln except that the quantity
of these compounds was lower (roughly 10
times the detection limit). The emission
rates at this kiln were on the order of
0.3 mg/s.
Particulate Matter Emissions
No statistically significant differ-
ence was noted in particulate matter
emissions between the baseline and waste
fuel burn at the wet-process kiln. The
control device was an ESP, and emissions
averaged about 0.65 Ib/ton. The maximum
chlorine loading reached 4.7 kg/Mg of
clinker.
A review of particulate matter results
from kiln tests conducted to date shows
similar results for pairs of tests:
• The testing of two kilns equipped
with baghouses (San Juan and
Rockwell Lime) showed no increase
in particulate matter emissions
when chlorinated wastes were
burned (5,11). Emissions were
0.65 and 0.25 Ib/t for the cement
and lime kilns, with chlorine
inputs of approximately 5.5 and
2.7 kg Cl/Mg product.
• The testing of two kilns'equipped
with ESPs (St. Lawrence Cement's
dry-process kiln and Alpha Cement)
showed a decrease in emissions
when wastes were burned that were
low in chlorine content (3,10).
Baseline emissions of 1.1 Ib/t
decreased to 0.7 and 0.8 Ib/t.
• The.testing of two kilns equipped
with ESPs (the wet-process kiln we
tested and Marquette Cement)
showed no change in particulate
matter emissions when chlorinated
wastes were burned at rates of
1.1-4.7 kg Cl/Mg clinker (7).
Emissions were 0.65 and <1 Ib/t
for these two kilns.
The testing of two kilns equipped
with EPSs (St. Lawrence's wet-
process kiln and Stora Vika)
yielded results for different
wastes and different chlorine
-'- .loadings. Emissions were posi-
""'tively correlated with chlorine
loading; however, the extent of
the increase in emissions varied
for different compounds and different
kilns. Chloride accumulation, as evi-
denced by ring formation and subsequent
release or pluggage, begins to occur
in the range of 6-9 kg Cl/Mg clinker
(1,2,9).
Although increased chlorine loading
at the St. Lawrence and Stora Vika kilns
increased particulate matter emissions,
there was no known attempt to compensate
for changes in the dust's resistivity.
However, emissions were still comparable
to those observed at other tests. Base-
line test results at these two kilns
ranged from 0.2 to 0.5 Ib/t compared
to 0.5 to 1.1 Ib/t during the cofiring
of waste fuels. The results do not
include a kiln upset from a chloride
ring formation at St. Lawrence Cement
when emissions averaged about 3 Ib/t.
The tendency for chloride rings to form
during high chlorine loading provides an
incentive to the kiln's operator to limit
the chlorine entering the kiln. Limiting
the chlorine input may avoid plugging and
process disruptions as well as limit
chlorine concentrations in the dust going
to the ESP. The test data suggest that
particulate matter emissions from chlori-
nated waste combustion are controllable
by a properly operating control device
and a reasonable limit on chlorine loading
to avoid ring formation and excessive dust
loading. Adjustments may be required on a
site-by-site basis to optimize ESP perfor-
mance and thus account for changes in dust
resistivity.
Fate of Chlorine
Results for chlorine are summarized
in Tables 3 and 4 and show that 90 to 99
percent of the chlorine is retained in
the process solids. At the dry-process
kiln, the waste fuel contained an average
of 2 percent chlorine and was fired at an
average rate of about 1.2 kg Cl/Mg clinker.
The waste fuel at the wet-process kiln
contained 1 to 4 percent chlorine and
was fired at an average rate of 2.2 kg Cl/
Mg clinker (maximum of 4.7). HC1 emis-
sions, percent chlorine in the dust, and
percent chlorine in the recycled dust
increase as the total chlorine input
203
-------
increases. On the first waste fuel test
day (second day of waste fuel burning) at
the wet-process kiln, HC1 emissions were
lower than during the baseline period.
Chlorine was detected in the clinker.
It is possible that an equilibrium chlorine
cycle had not been obtained at this point.
On the last day of testing at the dry-
process kiln, the chlorine cycle evidently
shifted to the cooler end of the kiln.
More chlorine exited with the stack gas
and waste dust than in previous tests,
and the chlorine concentration of the re-
cycled dust appeared to increase although
the total chlorine input remained un-
changed.
A review of other tests showed that
during the combustion of highly chlorinated
wastes at St. Lawrence Cement, over 99
percent of the chlorine was retained in
the process solids and HC1 emissions were
<1 Ib/hr during both the baseline and waste
fuel burns (9). The total chlorine input
was up to 6.8 kg Cl/Mg clinker. An in-
crease in HC1 emissions from 0.6 to 1.0
Ib/hr was observed at Los Robles (8), and
an increase from 2.4 to 5.8 Ib/hr was
observed at Alpha Cement (10). An. .
increase from <0.2 to 0.8 Ib/hr was
observed at San Juan Cement (11), and at
Rockwell Lime the increase was from-0.2
to 0.4 Ib/hr (5). During the tests at San
Juan Cement, the waste fuel was highly
chlorinated (up to 35 percent Cl) and was
fired at an average rate of 5.5 kg Cl/Mg
clinker. Over 99 percent of the chlorine
was retained in the process solids, primari-
ly the clinker. For other tests, most of
the chlorine was removed with the waste
dust. These tests indicate that HC1 emis-
sions can increase from chlorinated waste
combustion; however, 90 to over 99 percent
of the chlorine is retained in the process
solids.
Fate of Lead
The lead content of the waste fuel
ranged from roughly 100 to 1,000 ppm for
both kiln tests. Lead emissions at the
wet-process kiln increased from about 1.5
mg/s at the baseline to about 6.9 mg/s
during the waste burn. The malfunctioning
ESP prevented useful results at the dry-
process kiln test, but a previous test at
the same kiln with a similar waste fuel
showed an increase in lead emissions from
5 to 9 mg/s. A material balance showed
that over 99 percent of the lead was
retained in the process solids. The lead
204
concentration in the waste dust increased
at both kilns. At the wet-process kiln,
the increase was from 395 to 1,530 ppm
compared to,an increase from 116 to
2,650 ppm for the dry-process kiln.
A review of lead emissions during
other kiln tests shows varying results.
At St. Lawrence Cement, burning lead-
contaminated waste oil with a low chlorine
content did not affect lead emissions (3).
Similarly, burning chlorinated aliphatics
did not increase lead emissions, but burn-
ing PCBs resulted in an increase from ~1.5
mg/s at the baseline to ~4.6 mg/s (9).
Lead emissions appeared to increase also
at Alpha Cement (from ~4 to ~17 mg/s) (10)
and Rockwell Lime (from <0.4 to 0.47 mg/s)
(5). The lead content in the waste dust
also increased in most cases.
For perspective, consider that total
lead emissions arfe relatively small and on
the order of emissions from several auto-
mobiles burning leaded gas. Also consider
that over 99 percent of the lead is retain-
ed in the process solids. By comparison,
a boiler burning used oil emits 50 to 60
percent of the lead, and may emit a
higher percentage because of losses
during soot blowing (3,6).
S02, NO , Total Hydrocarbons, and CO
Emissions
At the wet-process kiln, S02 concen-
trations decreased from an average of 636
ppm to 265 ppm when the waste fuel re-
placed the sulfur-containing coal (4.3
percent sulfur). Approximately 60 percent
of the total sulfur was retained in the
process solids for both test conditions.
At the dry-process kiln, S02 concentra-
tions were low, with baseline concentra-
tions of 1.5 to 12 ppm. During the waste
burn, S02 concentrations ranged from 5 to
38 ppm and were shown to be strongly
affected by 02 input. The kiln exit
oxygen increased to 6.7, 7.3, and 7.5
percent on successive test days with
corresponding S02 concentrations of 38,
13, and 5 ppm. Approximately 99 percent
of the sulfur entering with the fuel was
retained by the process solids at the dry
kiln.
Similar results have been observed at
other tests, i.e., a reduction in S02 from
waste combustion. At Alpha Cement (10),
S02 concentrations decreased from 78 to
33 ppm, from 93 to 18 ppm at Marquette
-------
Cement (7), and were constant at 27 ppm
at Los Robles (8). . ,, ,.
i . . • '' *-"/"'
NO concentrations ranged from 37p_,to.
480 p$m at the wet-process kiln and "from-
600 to 800 ppm at the dry-process kiln'.'
The NO concentrations at the dry-process
kiln we>e strongly affected by kiln oxygen
input: an increase in kiln exit 02 from
an increased air rate yielded correspond-
ing increases in NO . NO emissions are
not obviously affected by burning waste
fuels. Other investigators have found
that NO concentrations are primarily
affecteS by oxygen input, primary/secondary
air ratio, and temperatures (4).
Concentrations of total hydrocarbon
ranged from 6 to 7 ppm during three base-
line tests at the wet-process kiln and
increased to 21 ppm during a kiln blockage
for one baseline test. The overall average
was 10 ppm compared to an average of 21
ppm (16 to 28 ppm) during the waste fuel
burn. The baseline test at the dry-process
kiln revealed concentrations of 2 to 4 ppm
compared to daily averages of 1, 5, and 9
ppm for the three waste burn tests. Tests
at San Juan Cement, Rockwell Lime, and
Stora Vika showed that total hydrocarbon
concentrations on the order of 10 ppm pr
less were attained during both the base-
.line and waste burn test conditions.
Carbon monoxide concentrations at
the wet-process kiln averaged 212 ppm
(100 to 400 ppm) during the baseline com-
pared to 190 ppm (130 to 340 ppm) during
the waste burn. The baseline tests at the
dry-process kiln revealed CO concentrations
of 35 to 40 ppm compared to an average of
39 ppm during the waste fuel burn. The
difference between CO concentrations
during the baseline and waste fuel test
conditions is not significant.
REFERENCES
1. Ah!ing, Bengt. 1978. Combustion Test
with Chlorinated Hydrocarbons in a
Cement Kiln at Stora Vika Test Center.
Swedish Water and Air Pollution
Research Institute, Stockholm, Sweden.
2. Ahling, Bengt. 1979. Destruction of
Chlorinated Hydrocarbons in a Cement
Kiln. Environ. Sci. Tech.
13(11):1377-1379.
3. Berry, E. E.,. L. P. MacDonald, and D.
{.- J. Skinner, 1975. Experimental
ir,.,,Burning of Waste Oil as a Fuel in
Cement Manufacture. Environment
' Canada. Report No. EPS 4-WP-75-1.
4. Carter, W.A., an R.C. Benson. 1982.
Application of Combustion Modification
Technology For NO Control to Cement
Kilns. Joint Symposium on Stationary
Combustion NOX Control, USEPA/EPRI,
Dallas, Texas, November 1-4.
5. Day, D. R. , and LA. Cox. 1983.
Evaluation of Hazardous Waste Incine-
ration in Lime Kilns at Rockwell Lime
• Company. (Draft Report). EPA Con-
tract-No. 68-03-3025. U.S. Environ-
.-.. mental Protection Agency, Cincinnati,
Ohio.
6. Fennel ly'« P., et al. 1984. Environ-
mental Characterization of Waste Oil
Combustion in Small Boilers. Hazard-
ous Waste, 1(4):489-505.
7. Higgins, G.M., and A. J. Helmstetter.
1982. Evaluation of Hazardous Waste
Incineration in a Dry Process Cement
Kiln.
8. Jenkins, A.C., et al. 1982. Supple-
mental Fuels Project, General Portland,
Inc., Los Robles Cement Plant. State
of California Air Resources Board.
Report C-82-080..
9. MacDonald, L.P., D.J. Skinner, F.J.
Hopton, and G.H. Thomas. 1977.
Burning Waste Chlorinated Hydrocarbons
in a Cement Kiln. Fisheries and Env-
ironment Canada. Report No. EPS
4-WP-77-2.
10. Nesselbeck, E.R.. 1981. Baseline and
Solvent Fuels Stack Emissions Tests.
Prepared for Energy and Resource Re-
covery Corporation, Albany, New York.
January.
205
-------
EVALUATION OF HAZARDOUS WASTE DESTRUCTION
IN A BLAST FURNACE
Radford C. Adams, Thomas A. Buedel,
Carol A. McCarthy, and Michael A. Palazzolo
Radian Corporation
Research Triangle Park, North Carolina 27709
ABSTRACT
At least one steel company utilizes organic waste liquids as a heat
and carbon content source to partially replace the coke that is used to
charge the blast furnace. The waste liquids fed to the blast furnace are
likely to contain hazardous constituents. Temperature and residence time
in the blast furnace favor total destruction of the principal organic
hazardous constituents (POHCs) of the waste fuel but verification of
destruction efficiencies has not been attempted up to now. Also,
reduction reactions that occur in a blast furnace may promote the
formation of products of incomplete combustion (PICs).
Tests were conducted while feeding waste fuel to a blast furnace
located at a major steel mill. The primary objective of the test program
was to determine the fate of the POHCs of the waste fuel and to look for
formation of PICs, notably dioxins and dibenzofurans.
INTRODUCTION
At least one steel company is
burning liquid hazardous wastes in
blast furnaces to supplement fuel
and coke requirements. Tests were
conducted during the burning of a
waste fuel containing hazardous
constituents (POHCs) in a blast
furnace at a major steel plant.
The blast furnace tested under this
program is typical of blast fur-
naces used for iron production
throughout the steelmaking
Industry. The blast furnace had
been retrofitted with a liquid
feed system for injecting waste
fuel, replacing natural gas,
into the combustion zone.
Blast furnaces are prime
candidates for thermally destroy-
ing POHCs due to the extremely
high temperatures that are gen-
erated 1n the combustion zone.
On the other hand, unburned POHCs
are subjected to reducing
conditions when leaving the com-
bustion zone and entering the
iron oxide reduction area. As a
consequence, hazardous waste
destruction may be limited or
products of incomplete combustion
(PICs) may form.
A blast furnace produces mol-
ten iron from iron ore and other
iron bearing feed materials. A
moving bed of iron ore, carbon
(as coke), and limestone descends
through the blast furnace tower.
In the combustion zone, located
between the moving bed and the
hearth of the furnace, oxygen of
hot blast air and steam react with
the carbon monoxide and hydrogen.
Temperatures in the combustion zone
exceed 3000°F. CO and H7 travel
upward through the moving bed and
react with the iron oxide (as iron
ore) to produce elemental iron.
The molten iron and slag are collec-
206
-------
ted in discrete layers on the
hearth of the furnace and are
removed through tap holes at reg-
ular intervals.
Figure 1 depicts the overall
blast furnace system. Excess CO
and H7 are produced to drive the
iron oxide reduction reaction to
completion. Therefore* blast
furnace off gases have a heating
value of about 90 Btu/scf. This
energy is used to produce steam for
air compression and other process
"steam requirements. Only about 1/4
of the blast furnace gas is needed
to heat the blast air. A set of
three stoves are used to heat t|ie
blast air to approximately 1800 F
(only one stove is shown in Fig-
ure 1). The stoves, which are
filled with a refractory brick
checkerwork, cycle between com-
busting blast furnace gas, and
heating blast air so that one
stove provides hot blast air
while the other two stoves are
being heated up. Products of
combuston from the stoves are
emitted through a stack. Dust is
removed from the blast furnace
before entering either of the com-
bustion devices.
PURPOSE
The objective of this test
program was to evaluate typical
waste disposal performance of a
blast furnace when burning a liq-
uid organic waste. A major goal
was to determine the fate of the
major components of the waste and
their products of incomplete com-
bustion (PICs), if any, that are
emitted from the blast furnace and
a downstream combustion process.
For this program, the blast fur-
nace stove set was selected as the
downstream combustion process.
The alternative would have been
the plant power house where blast
furnace gas is burned in boilers
that produce process steam. Speci-
fically, the following objectives
were to be met.
1. Determine the fate and destruc-
tion removal efficiency (DRE)
of waste fuel POHCs (principal
IRON ORE
COKE
LIMESTONE
VENTURI
CYCLONE SCRUBBER
COOLER
WASTE OIL
\
SLA(
r
3 TAP
METAL TAP
.
FROM AIR BLOWER
TO POWER HOUSE
SAMPLING POSITIONS
Figure 1. Blast Furnace Process- Flow
207
-------
organic hazardous constituents)
by monitoring blast furnace gas,
scrubber water* and combustor
flue gas for Identified POHCs.,;
2. Determine PIC formation and,
if detected, their fate.
3. Determine relative emissions
of volatile organic components
of waste fuel from the waste
fuel storage tank compared
with emissions of these com-
ponents from the combustor.
It was also an objective to deter-
mine if any dioxin/dibenzofurans
were present at levels greater
than one part per trillion in the
emission/effluent streams.
APPROACH
Blast furnace process streams
were sampled simultaneously to meet
the data requirements. These are:
• waste fuel to identify and
quantitate POHCs and chlorine
content,
• blast furnace gas to determine
fate and ORE of POHCs and to
Identify and quantitate PICs
(sampling position is down-
stream of dust collector and
wet scrubber),
• stove stack gas to determine
fate and ORE of POHCs and fate
of.identified PICs,
• dust collector sample to
archive for expansion of per-
formance evaluation if needed,
and
• waste fuel storage tank
emissions to compare magnitude
of tank emissions with stove
emissions.
Identify and Quarvhjtate PQHCg
Jn_Feed
The waste fuel was fed to the
blast furnace continuously at about
60 gpm. The pumping rate was 100
gpmj thus, excess liquid was re-
circulated back to the feed tank.
The_,liquid was sampled just before
the blast furnace burner at inter-!
vals of one hour and composited for
and quantisation was accomplished
by gas chromatography/mass
spectroscopy (GC/MS).
The feed tank also served as
storage tank. Additions to the
feed tank during sampling and re-
sultant variations in composition
could not be avoided. Tank trucks
were off loaded to the feed tank
during each sampling run. The
waste liquid was blended with No.6
fuel oil to maintain a preset
level in the feed tank. The com-
bination of these two events
created cyclic waste to fuel
ratios of the waste fuel fed to
the blast furnace. During sam-
pling, this ratio averaged 0.8 -to
1.0, 0.6 to 1.0 and 0.6 to 1.0,
respectively for each run.
Fate and DRE of POHCs
Destruction and removal
efficiencies (DRE) were, deter-
mined for two cases; across the
blast furnace and across the blast
furnace and combustion process.
For both cases, any removal by
the particulate control devices
(a cyclone dust collector and a
wet scrubber in series) was
accounted for. Since the POHCs
were unknown at time of sampling,
both volatile (VOST) and semi-
volatile (modified method 5)
sampling methods were employed.
Stove products of combustion were
sampled at near atmospheric
pressure from the stack. Blast
furnace gas after cleaning was
sampled at inlet to the stoves
at a positive pressure of 50
Inches w.c. Table 1 shows the
fixed gas compositions for each
location. Hydrogen should account
for a substantial portion of the
four percent not identified 1n the
stove inlet gas. Analysis of POHC
data of the scrubber waters was
not completed in time to Include
in this paper.
Identify and Determine
Fate of PICs
208
-------
The following outlet streams
were sampled for compounds not
found in the waste fuel:
• •. • - • • • ;, ::nr.
• scrubber makeup water - - '••-
• scrubber discharge water
• blast furnace gas inlet to
stoves
• blast furnace gas outlet
of stoves
Classification of these compounds
as PICs is avoided because they may
have originated in the blast air or
the iron ore/coke mixture.
Waste Fuel Storage Tank Emissions
The storage tank gas phase was
sampled by integrated bag sampling
and analyzed by GC/flame ioniza- ;
tion detection. A portion of the
sample was passed through a Tenax
resin trap for later analysis by
GC/MS. Liquid temperature in the
tank averaged 117 F.
PROBLEMS ENCOUNTERED
Two major types of problems
were encountered:
1, The potential for CO leaks in-
troduced the need for specific
safety precautions during sam-
pling.
The duct that was sampled for
blast furnace gas was at a posi-
tive pressure of 50 inches w.c.
Also, high concentrations of CO
and H? if released created both a
CO poisoning and fire hazard to
sampling personnel. Special sam-
pling equipment was devised to
protect personnel and equipment
and to maximize sample collection
efficiency. The steel company
safety rules required that their
safety personnel be present at
all times. They continually moni-
tored for dangerous CO levels and
they had the authority to evacuate
the area when CO levels were ex-
ceeded. Air masks required to be
worn during periods of maximum
risk, e.g., when probe changes were
made. The special probe for modi-
fied method 5 sampling was inserted
through a packing gland that allow-
ed movement of the probe to the
next traverse point with a minimum
of gas release. Pressure connec-
tions were used throughout the
sampling train and both positive
pressure and negative pressure leak
tests were made. The heated filter
box was fitted with a fan to purge
any possible release into the box
of explosive gases.
The site specific limitations
were as follows:
Site specific sampling in-
stallations and operating
inflexibility required some
sacrifice of best test
protocoli • • •
TABLE 1. FIXED GAS ANALYSIS RESULTS
Lack of consistent feed com-
position was described
earl ier.
Location
Parameter
Stove outlet
CO
Concentration
(Volume percent)
Stove inlet 09
CO2
CO,
N2
, 4.3
22.0
15.6
54.1
5.9
21.8
69.5
209
-------
• The waste liquid feed could
not be shut off without up-
setting the blast furnace
carbon and heat balances. The
baseline data could not be
col1ected.
• Access problems prevented
traversing for sampling and
for velocity measurements
two diameters of the round
ducts at the stove inlet and
stove outlet sampling posi-
tions. Federal Register
methods stipulate traversing
on two diameters at right
angles.
The latter limitation introduced
significant error in the velocity
and flow measurements at the inlet
location. Therefore, process flow
measurements were used. The outlet
velocity measurements appeared to
be valid when compared with a mass
balance calculated from the inlet
process flow measurements and
gas composition.
RESULTS
Analysis of all data had not
been completed 1n time to include
1n this report. This Includes
primarily the fate of PICs and the
tank emissions results. Complete
results will be reported to the
EPA's Hazardous Waste Engineering
Research Laboratory in a final
report. Program accomplishments
that can be reported now are are
as follows:
- Characterization of liquid
waste fed to the blast fur-
nace,
- DRE's across the blast
furnace with and without
downstream combustion,
- Determination of dioxin/
dibenzofuran emissions,
and
- Blast furnace operating data.
Three sampling runs were conducted
on successive days.
Liquid Waste Characterization
A waste broker supplies the
organic liquid waste to the steel
company. The waste is blended
with No. 6 fuel oil to create a
waste fuel for blast furnace use.
Table 2 shows POHC average concen-
tration and feed rates for the
three sampling runs. Toluene is
the predominant POHC in the fuel,
although chlorinated POHCs account
for over 6400 ppm of the liquid.
The potential for consider-
able variation in waste fuel
composition was discussed earlier.
Table 3 summarizes the variabil-
Table 2. Concentration of POHCs in the Waste Fuel
Compound
l,l-D1chloroethene
Chloroform
1, Ij.-Tr1chl oroethane
Trlchloroethene
Benzene
Tet rachl oroethene
To! uene
m/p-Xyl ene
o-Xylene
Naphthalene
Total Chlorine
Waste Fuel
(ppm by wt.)
1,210
254
999
1,185
553
2,800
56,200
940
5,280
880
835
Feed Rate
(Kg/hr)
15.7
3.3
13.0
15.4
7.2
36.3
m.O
12.2
68.8
11.4
10,8
210
-------
Table 3. Variation of POHCs in Feed
Chloroform
1>1,1-Trichloroethane
Benzene
Toluene
o-Xylene
Range, ppm by Weight
33-469
817-1215
543-569
49901-65417
3050-9500
CV. %
70
16
2
12
56
ity of POHCs in the feed as ex-
pressed by the coefficient of
variation (CV) of the standard
deviation of 3 samples. Table 3
shows standard deviations varying
from 2% from the mean to 70% from
the mean. All other POHCs fall
between the values shown. Surpris-
ingly, benzene exhibits a CV of
only 2%. Precision this good is
highly unusual. Close examination
of the analytical records has so
far not revealed any anomalies or
errors.
ORE
Destruction/removal efficiency
(DRE) is defined as the disappear-
ance of a feed constituent due to
thermal destruction followed by
removal by an air pollution control
device, expressed as a percentage.
In this paper* (1-DRE) will be re-
ported expressed as one part per
100,000 parts by weight. If the
breakpoint between an acceptable
and unacceptable DRE is 99.99%, a
(1-DRE) less than 10 exceeds 99.99%
and 1s acceptable. Table 4
summarizes (1-DRE) results for the
blast furnace before combustion
inlet to stoves). Best DRE's were
accomplished with the combustion
device. This should be typical
of all blast furnaces.
DJoxin/Dibenzofuran EmJssions
Blast furnace gases before
and after combustion and scrubber
waters were analyzed by high reso-
lution GC/MS. Neither class of
compound was detected at the 1 ppt
detection limit for the gas sample
and the 1 ppb detection limit for
the water samples.
Table 4. (1-DRE) - Blast Furnace
Before Combustion After Combustion
(Parts per 100,000 parts)
1,1 Dichloroethene
Chloroform
1,1,-Trichloroethane
Trichloroethene
Benzene
Tetrachl oroethene
To! uene
m/p-Xylene
o-Xyl ene
Naphthalene
ND-24
ND-15
ND-32
ND-3
842-1437
<1
9-16
8
12
105
ND-<1
2-6
10-16
ND-1
2-470
<1
<1
<1-40
2-7
7-38
ND: not detected
211
-------
Blast Furnace Operation
Table 5 summarizes the oper-
ating data, for the most part
collected from plant instru-
ments. At steady state, the
blast furnace produces 4000
ton/day of metal. Total oxygen
in the blast exceeds that of
air because the air is oxygen
enriched. Maintaining the blast
air temperature above about
1725 F and constant is critical
for maintaining combustion
zone reactions. Therefore,
the blast cycle for each stove
is kept,fairly short to avoid
excessive temperature cycling.
Table 5. Summary of Blast Furnace Operating Data
Averaae ,Value
Parameter
Run 01
Run 02
Run 03
Blast Air Temperature, F .,
Blast Air Flowrate, dscfm x 10
Blast Air Pressure, psig
Total Oxygen in Blast, %
Flame Temperature, °F (calculated)
Furnace Top Temperature, F
Top Gas HV, Btu/scf (calculated) 3
Total Blast Gas Volume, scfm x 10
Waste Fuel Injection, gpm
Waste Fuel HV, Btu/lb
1743
116.5
31.5
23.4
3389
337
90.7
128.3
60.6
14,776
1770
116.6
30.4
23.4
3487
379
89.3
127.1
59.5
14,708
1756
116.2
30.2
23.4
3402
325
91.5
127.9
60.4
14,851
ACKNOWLEDGEMENTS
The authors gratefully
acknowledge the cooperation of
the steel company management
Including especially the environ-
mental control coordinator and
the blast furnace operating
superintendent. We are particu-
larly grateful to the safety
engineers who made a safe sampling
operation possible by continually
monitoring the work areas for dan-
gerous levels of CO. This project
is sponsored by the U. S. EPA
Hazardous Waste Engineering
Research Laboratory. The tech-
nical project monitor for EPA is
Mr. Robert Mournighan.
212
-------
FIELD EVALUATION OF SULFURIC ACID REGENERATION UNIT
BURNING HAZARDOUS WASTE AS FUEL
R. C. Adrian, P. K. Ouchida
California;Air Resources Board
Sacramento, California 95814
ABSTRACT
Rarely does an existing chemical processing facility provide the qualities needed for a
hazardous waste destruction process without significant modifications to the facility.
The system which is the subject of this study is a waste sulfuric acid regeneration unit.
It is designed for combustion of many organic wastes from petroleum processing that
contain high sulfur levels and have a high heating value. The furnace operates above
1600°F, has a large flame volume and-long residence time. The combustion products are
passed through a scrubber and a wet electrostatic precipitator for production of sulfuric
acid. Thus, the organic feed is subjected to thermal, chemical and catalytic oxidation
and combustion products are subject to absorptive and electrostatic separation.
The testing program for the system includes varying the operating conditions to evaluate
emission effects of combustion air, furnace temperature, firing rate and fuel heat
content. Baseline conditions are evaluated, with no added hazardous materials, burning
spent acid and auxiliary natural gas fuels. Waste burn effects incorporate the addition
of surrogates including carbon tetrachloride, Freon 113, 1,2,4 trichlorobenzene and penta
chlorophenol. Analysis.of emissions provide for DRE determination of the surrogates,
POHCs dioxins and furans, and normal gaseous emissions.
INTRODUCTION
In a continuing investigation of
alternative methods of hazardous waste
disposal, the California Air Resources
Board (CARB) has evaluated several
existing commercial scale incineration
systems. The selection of processes to
be tested was guided not only by the
severe Environmental Protection Agency
(EPA) requirements of high temperature
and long residence time but also by
adjunct oxidative chemical reactions and
possibilities of removal or
neutralization of products of incomplete
combustion (PICs).
Emission testing and chemical
analysis of effluents for hazardous waste
processing tend to be very labor
intensive as well as requiring very
specialized apparatus. It is thus
neccessary to thoroughly prepare a test
plan well in advance of testing.
Analytical limitations may be a major
factor in the sequence and duration of
the tests.
Stauffer Chemical Company (SCC) has
for years operated a spent sulfuric acid
recovery plant in an industrial section
of Los Angeles. The spent acid is
primarily refinery alkylation acid,
containing a variable percentage of
organic compounds, some of which would be
considered hazardous materials.
PURPOSE
The goals of this project were
multifaceted. The primary objective was
to measure emissions to determine if the
California Department of Health Services
(DHS) and the South Coast Air Quality
213
-------
Management District (SCAQMD) would be
justified in issuing operating permits
for burning hazardous wastes in
conjunction with the normal plant
feedstocks. Additional objectives
included evaluation of the destruction
and removal efficiency (ORE) of the
principal organic hazardous constituents
(POHCs) across the burner,
characteristics of PICs formed in
combustion, and evaluation of operating
conditions (failure modes) to define
permit limitations in terms of
temperature, feed composition, and firing
conditions.
PROCESS DESCRIPTION
The SCC unit evaluated in this
project was essentially a double
contact-double absorption sulfuric acid
plant. The facility incorporates a
proprietary combustion chamber into which
molten sulfur, alkylation acid, and waste
materials can be injected either singly
or in combination. The combustion
chamber is ell-shaped, permitting either
in-line firing (face firing) or impinging
firing by injecting one or more of the
fuels into the branch of the chamber
(dutch-oven firing). Thus turbulence and
residence time can be modified to
optimize combustion efficiency.
Additional heat to provide an
approximately 1800°F temperature at the
combustion chamber exhaust was supplied
by augmenting natural gas.
Contiguous with the combustion
chamber is an ash settling chamber which
can also act as a secondary combustion
chamber when a large flame volume is
present. Normal feed to the system may
contain as much as 10% ash; however,
little of the ash drops out at this point
or in the immediately following waste
heat boiler. The waste heat boiler
reduces the combustion gas temperature to
about 700°F, while also producing process
steam and electrical power.
Following the waste heat boiler, the
cooled gas enters a direct spray,
brick-lined scrubber (quench tower) which
further cools the gas by flashing water
from recirculating weak acid. The
scrubber also removes most of the ash
that may have carried through the boiler
along with other minor impurities.
After this first scrubber, the gas is
further cooled, condensed water is
removed and the gas is compressed before
passing through the first of two wet
electrostatic precipitators (E$Ps).
In various stages the gas is passed
through a strong acid drying column and
then through the first stage of a
catlytic converter. $03 is then
removed in a first stage absorber.
Dilution air is added to promote $03
conversion and the gas passes through the
converter second stage. The balance of
S03 is removed in a second absorber.
The gas then passes through a second ESP
before being vented through the stack.
APPROACH
Emission standards for combustion of
hazardous waste materials have not yet
been developed except for the EPA
acceptance of 99.99% or in some cases
99.9999% combustion efficiency for POHCs.
Prior to drafting the test plan, a
panel of experts in the fields of
combustion, chemistry, kinetics, and
testing met to discuss the best test
program for the Stauffer system. The
test plan incorporated the panel
recommendations. To evaluate the
destruction efficiency of this process,
it was concluded that the fuel normally
encountered, spent alkylation acid,
represented conditions equivalent to the
most extreme chemical decomposition
likely to be encountered if POHCs could
be identified and remain constant.
However, the evaluation of decomposition
of certain added organics would provide a
basis for comparison with other processes
and hazardous material conversion
efficiency.
It was concluded that the addition of
a synthetic waste mixture consisting of
carbon tetrachloride, 1,1,1
trichloroethane, 1,2,4 trichlorobenzene
and hexachlorobenzene would provide a
surrogate equivalent of the most
difficult to destroy organic compounds
and produce the least desirable PICS.
Subsequently, it was found that
hexachlorobenzene was not available and
pentachlorophenol was substituted.
Heating value, viscosity and laboratory
detection limits were considered when
determining the composition'of this
waste.
214
-------
5TAUFFER CHEMICAL COMPANY TEST OBJECTIVES
The test objectives for the Stauffer
Chemical Company test were to allow : .,- ,
determination of the baseline emissions
while the plant is burning spent acid and
natural gas. Specifically the following
were to be determined:
1) Emissions while the plant was
burning spent acid and typical
waste fuels.
2) Destruction and removal
. efficiency of the process under
at least three different
operating conditions by using a
special mixture of organic
compounds as surrogates for the
waste fuel .
3) HC1 emissions from the stack.
4) Flow rate through the process
using
5)
6)
7)
02, C02, CO, CH4, TUHC,
NOX and S02 concentrations at
the exit of the boiler and the
stack.
Dioxin and dibenzofuran
concentrations in the exhaust
gases under different operating
conditions at the exit of the
boiler and at the stack.
Effects on emissions of modified
modes of operation.
Also, as part of the test program,
the capability to perform on site
chromatographic (GC) analysis was to be
continued to be developed.
To accomplish the objectives set
forth above, the following operating
conditions were to be evaluated.
a) Baseline conditions: Spent
acid was to be fed into the
furnace at a rate of 20 to 30
gallons per minute. Natural
gas was to be used to provide
the balance of the heat
requirements. Furnace
conditions were to be varied
in order to determine the
operating range (envelope)
for this process.
b) Normal waste fuel
conditions: TTie furnace was
to be operated under normal
steady-state conditions
except when the effects of
operating parameters were
being evaluated. The
synthetic waste was to
include carbon tetrachloride,
hexachlorobenzene,
trichlorobenzene,
chlorophenols and sulfur
hexafluoride.
c) Modified Furnace Operating
Mode: The waste used in
Section b was also be to be
used for these tests, except
perlite was to be added to
represent typical solids
content for expected waste
fuels.
i) Waste injected through
burner in front face
(when furnace at 1600°F
and again when 1% excess
02 in furnace).
ii) Waste injected through
burner in dutch oven at
furnace conditions of
1600°F and 1% excess 02
in furnace.
The total test time was to be
approximately 7 days.
Parameters such as temperature, CO,
TUHC, 02 and NOX were to be used in
determining the acceptable ranges.
Operating curves were to be obtained for
excess 02 ranging from 1% to 4% and
temperature from 1600°F to 2000°F.
SCC was to monitor and record all
operating conditions necessary to
determine if the desired conditions have
been reached and was be responsible for
determining the length of time that these
conditions can be expected to be
maintained.
TEST PROGRAM
Testing actually took place over a
three-week span. The first week was used
in getting equipment set up, calibration,
getting the process stabilized and
running the baseline tests. The second
215
-------
week was taken up with runs using the
spent acid/synthetic waste mixture under
normal firing conditions but with some "
variation in operating conditions.
During the third week the failure modes
of low temperature and burner
reconfiguration were examined. Extremely
low or high temperatures could not be met.
Samples were taken at two locations;
in a horizontal duct between the waste
heat boiler and the quench cooler, and in
the stack about 90 feet above ground
level. At each point, four Tedlar bag
samples for volatile organics were taken
per day. At the stack, two modified M-5
trains were run per day using XAD-2, one
HC1 train per day except the four days on
which dioxin trains were run (also with
XAD-2), and one M-5 train per day. At
the boiler, one modified M-5 train per
day for semi-volatiles, and intermittent
bag samples for S02 and CO were taken.
Additionally, SCC personnel obtained
process samples during the test periods
for spent acid fuel, synthetic waste,
stack precipitator rundown, quench cooler
rundown and product acid.
The synthetic waste nominally
contained 22.39% carbon tetrachloride,
21.22% 1,1,1 trichloroethane, 3.00%
pentachlorophenol, 10.61% 1,2,4
trichlorobenzene, 30.62% diesel oil,
1.16% methanol and 11.00% perlite. This
augmenting mixture was fired at about ten
pounds per minute. This mixture was used
throughout the program.
PROBLEMS ENCOUNTERED
Testing was originally planned to
begin in September 1984; however, delays
in receiving necessary experimental
permits and variances resulted in an
actual start time of February 26, 1985.
Once testing had begun, there were
surprisingly few delays caused by
mechanical problems. Perlite in the
synthetic waste mixture created a high
erosive rate in the pumps and some
clogging of the fuel system, but these
were usually repaired at night or on
weekends with little effect on the test
schedule.
The fittings of the Teflon-lined
stainless steel flexible sampling line at
the waste heat boiler were found to
deteriorate in the 13% SOg, high
temperature and high moisture atmosphere,
arid the line was replaced with Teflon.
'Also, at this sampling point a 50:1
dilution system was used to avoid damage
to-'the continuous monitoring instruments.
The on-site chromatograph could not
be used for quantisation during the test
because of a shortage of standards in the
proper concentration range. These data
were reduced after the test.
It was not found possible to measure
the combustion chamber exhaust flow rate
by SFs injection. These rates were
instead calculated by mass and heat
balance around the waste heat boiler.
The low temperature failure mode test
(1600°F) was found to be impractical.
Unburned carbon and ash were not carried
through the combustion chamber and caused
an early shutdown. The test was later
carried out at 1700°F successfully.
RESULTS
Field tests for the project were not
completed until March 15, 1985, and
laboratory analyses are not yet
complete. Therefore, a discussion of the
test data cannot be made at this time.
216
-------
NONSTEADY INDUSTRIAL BOILER WASTE COFIRING TESTS
Robert 0. DeRosier, Howard B. Mason, Ursula Spannagel, C. Dean Wolbach
Acurex Corporation
Mountain View, CA 94039
ABSTRACT
Hazardous waste thermal destruction field tests were conducted on a 13.9 kg/s
(110,000 Ib/hr) single burner package watertube boiler to evaluate the effect of nonsteady
and off-design operation on destruction and removal efficiency (ORE) of principal organic
hazardous constituents (POHC's). These tests were done to extend the earlier results for
11 boilers tested at nominally steady conditions which showed a mass weighted average ORE
of 99.998 percent. The test series comprised triplicate baseline runs at steady
conditions; 24 nonsteady runs with gas/waste cofiring; and 21 runs with oil/waste
cofiring. For the baseline and selected nonsteady runs, POHC's were sampled with the
volatile organic sampling train (VOST). for the majority of the nonsteady runs, POHC's
were quantified with a "minf-VOST" protocol adapted for shorter sampling times and onsite
analysis with the GC-Hall detector. The raw waste, containing methyl methacrylate, was
spiked with carbon tetrachloride and monochlorobenzene. ORE results showed minimal effect
of nonsteady operation on thermal destruction. The average ORE for 112 combinations of
POHC and test runs was 99.998 percent. Only six values of DRE were less than
99.99 percent. The DRE showed no discernable correlation with carbon monoxide emissions.
Two additional tests are planned to assess the general applicability of these results.
INTRODUCTION
Cofiring of combustible hazardous
wastes with conventional fuels in
industrial boilers is widely practiced in
industry for energy recovery and as an
economical means of disposal. Regulation
of boiler cofiring was temporarily exempted
from the 1981 Resource Conservation and
Recovery Act (RCRA) which limited
incineration DRE to 99.99 percent or
higher. In the interim, EPA's office of
Solid Waste and Emergency Response has
conducted a regulatory impact assessment of
cofiring thermal destruction pursuant to a
boiler regulation. To support this
assessment, the Hazardous Waste Engineering
Research Laboratory has performed a boiler
cofiring field test program (1).
An initial series of 11 boiler field
tests were conducted to obtain a
representative sampling of waste thermal
destruction over the spectrum of
design/waste combinations in use in
industry, (2) through (5). The 11 test
sites were selected to cover the range of
industry practice with emphasis on gas- or
oil-fired watertubes which is the
predominant cofiring application. The test
boilers were operated at nominally steady
load, excess air and waste-to-fuel ratio to
obtain a reference data base under steady
operation. Within this context, the boiler
settings corresponded to normal plant
practice. A boiler test at each site
generally consisted of triplicate POHC
measurements at the nominal steady set
point.
Composite results for the 11 test
boilers showed generally high levels of
thermal destruction. The composite DRE for
a specific POHC averaged over all sites was
99.99 percent or higher for each of the
volatile POHC's quantified. For
semi volatile POHC's, average DRE's of
99.96 to 99.98 were experienced for four
POHC's in a wood-waste-fired stoker
operated at very high levels of exces.s air
which partially quenched the combustion.
The site specific DRE (averaged over all
volatile POHC's) exceeded 99.99 percent for
all sites except one boiler where unstable
217
-------
burner operation was experienced. No
definitive trend of ORE with carbon
monoxide emissions was observed. However,
the range of CO variation was small.
The test discussed in this paper was
conducted to extend the earlier data base
in two ways. First, nonsteady and
off-design operating conditions were tested
to evaluate how ORE is affected by upsets
or transients. Second, a wide range of
steady levels of load, excess air and
waste-to-fuel ratios were tested to
quantify ORE variations over the boiler
operating envelope. As a secondary
objective, a simplified version of the
volatile organic sampling and analysis
protocol was implemented to facilitate this
type of transient and parametric testing.
APPROACH
The test boiler was a forced draft
Combustion Engineering type 30-A-12 package
watertube with a maximum capacity of
13.9 kg/s (110,000 Ib/hr) of superheated
steam. Natural gas or No. 6 oil is fired
through a dual air register Coen dual-fuel
burner. The burner has been retrofit with
two steam atomized waste guns to fire a
distillation byproduct containing methyl
methacrylate. The waste was spiked with
varying concentrations (0.5 to 4.5 percent)
of carbon tetrachloride (CC14) and
monochlorobenzene (CsHsCl) to broaden the
range of POHC's quantified and to introduce
chlorinated compounds so that onsite
GC-Hall analyses could be performed. Waste
firing typically supplied 15 to 40 percent
of total heat input to the boiler. The
boiler normally operates in the range of
45,000 to 75,000 Ib/hr with daily load
variations of 20,000 Ib/hr and occasional
spot steam demands of 15,000 Ib/hr. The
boiler is normally operated automatically
with a microprocessor controlled oxygen
trim system. Waste flow is regulated
manually.
Six test series, summarized in Table 1,
were performed from April 24 to May 11,
1984. The initial boiler performance tests
were performed to determine operational
characteristics needed to plan the
subsequent transient and off-design tests.
Boiler characteristics monitored included:
CO/02 relationships for gas and gas/waste
cofiring; temperatures and emissions over
the operating range; thermal response to
transients; and 02/fuel/waste response to
transients. Baseline tests were run for
comparison to earlier steady-state tests
and as a reference for subsequent nonsteady
tests. The baseline tests used the full
sampling and analysis protocol as in the
previous 11 boiler tests:
• Continuous flue gas monitoring for
02, CO, C02, NOX and unburned
hydrocarbons
• Particulate and semivolatile
organic sampling by the modified
Method 5 train; post-test
semivolatile analyses by GC/MS
0 Volatile organic sampling by VOST;
post-test analyses by direct
desorbtion on GC/MS
• HC1 sampling by a modified Method 6
train; analyses by titration
• Grab sample of waste and fuel oil;
GC/MS POHC analyses
For the nonsteady tests, alternate volatile
sampling protocols were needed with shorter
sampling times and onsite analysis
capability for refinement of the test
matrix. Three approaches were tested: a
semi-continuous VOST; a mini VOST; and a
total organic chloride (TOCL) monitor.
The semi-continuous VOST used a single
tenax trap directly plumbed to the stack
via a heated sample line and gas
conditioning system. A sample was drawn
through the trap for approximately five
minutes after which the trap was directly
desorbed to a GC-Hall and the cycle
repeated. Tests of this protocol during
April 30 to May 4 showed good system
response and sample recovery. Results were
suspect, however, due to possible POHC
deposition on the sample delivery system.
The TOCL protocol used a Hall detector
to quantify total organic chlorides. Flue
gas samples scrubbed, for HC1 by a water or
NaOH impinger, were drawn through a heated
line directly to the Hall detector. The
detector response was read on a recorder.
Trial runs with this protocol showed
qualitative correspondence with POHC
throughput. Although the protocol needs
further development, particularly for
calibration procedures, it does show
promise for indicating relative levels of
POHC emissions.
218
-------
The bulk of volatile organic sampling
for the nonsteady tests was done with a
simplified mini-VOST protocol. A single
tenax trap was used with the VOST train for
sample extraction at the stack. A 10 liter
sample was drawn at 0.5 1/min. Following
sampling, the traps were taken to an onsite
GC-Hall and thermally desorbed. Trap
preparation prior to testing was also done
onsite. 'The nonsteady test protocol from
May 7 to May 11 generally used the
mini-VOST together with continuous monitors
for flue gas criteria species, and waste
grab samples. Occasional full VOST runs
with post-test GC/MS were made for
reference. No modified Method 5 sampling
was done for nonsteady runs.
The variables tested in the nonsteady
and off-design test series included:
• Upper and lower ranges of load,
excess air and waste flowrate
• Increasing and decreasing excess
air, load and waste flowrate
• Waste startup
• Poor waste atomization
• Waste atomizer spatial orientation
relative to oil guns
• Sootblowing
Table 2 summarizes the baseline and
nonsteady test runs. Runs with
semi-continuous VOST are excluded because
of suspected hysteresis in the POHC
concentrations entering the tenax trap.
The values for load, excess 03, and waste
flow are nominal values from plant board
data. During transient runs, frequent or
continuous recording of these parameters
was made. The boiler operational settings
were selected to deliberately induce
smoking or CO excursions in the range of
200 to 1,000 ppm.
RESULTS
For the cofiring of the spiked waste,
volatile POHC quantisation was made for
three steady baseline tests, using full
VOST, 24 gas/waste nonsteady tests (3 with
full VOST), and 21 oil/waste nonsteady
tests (3 with full VOST). On an overall
basis, these test series produced 112 ORE
data points for the composite of baseline
and nonsteady quantitation of carbon
tetrachloride, monochlorobenzene and methyl
methacrylate thermal destruction.
ORE results show notably high levels of
thermal destruction even during substantial
upset conditions (Table 3). Of the 112 ORE
values, only 6 were less than
99.99 percent. One of these low values, a
ORE of 99.7 for CC14, is suspect due to
apparent coelutriation of other compounds
in the gas chromatograph during tenax
desorption. For conservatism, the entire
GC response was assumed CC14. Excluding
this ORE value, the overall average ORE for
the entire test program was 99.998 percent,
which is the same average as for the
earlier 11 steady-state tests.
Although there were significant
variations in ORE from one run to another,
the average ORE values were in a fairly
narrow range. Table 4 compares the average
ORE values for the three POHC's and for gas
and oil firing. The ORE for
methyl-methacrylate, averaged for both the
VOST and modified method 5 measurements,
was higher than the organic chlorides.
Excluding the one outlier point, there was
not a significant difference between the
average ORE values for CC14 and
monochlorobenzene, or between gas and oil
firing. There was also a lack of marked
difference in ORE due to boiler operating
condition (Table 5). Although several of
these conditions, particularly atomizer
upsets and low 02 firing, produced
significantly degraded combustion
efficiency, the average DRE's were mostly
in the range 99.992 to 99.999 with no
consistent trend in ORE variation. In this
regard, there was a general lack of
correlation with combustion efficiency
indicators such as CO emissions, as shown
in Figure 1. There was, however, a general
agreement in magnitude between a pseudo-DRE
for natural gas computed for the gas-fired
runs (99.997) and the average POHC ORE
(99.998).
The lack of ORE correlation or of
consistent variation with operating
conditions may mean that the near burner
flame region is very efficient in
destroying or at least pyrolyzing POHC's
thereby suppressing effects of any physical
or chemical variation. The resultant low
levels of POHC in the stack appear to be in
the range of background emissions. It is,
also suspected that gummy organic deposits
on the boiler surfaces upstream, of the
sampling station were modulating the POHC
219
-------
concentration somewhat by serving as a
reservoir for adsorbtion and desorbtion.
Significant quantities of products of
incomplete combustion (PIC) were quantified
for nine compounds not present in the waste
in detectable concentrations. The highest
PIC concentrations were evident for
tetrachloroethene and dichloromethane. The
program average PIC/POHC mass ratio was 15
but individual run ratios varied from 0.3
to 190.
Two additional field tests are planned
in 1985 to assess the generality of these
results. Additional development and
application of the mini-VOST protocol will
also be performed.
ACKNOWLEDGEMENT
This effort was sponsored by the
Environmental Protection Agency under
contract 68-02-3176. Robert Olexsey of the
Hazardous Waste Engineering Research
Laboratory, and Marc Turgeon of the Office
of Solid Waste and Emergency Response were
the Project Officers. Their support and
assistance both on- and offsite is greatly
appreciated. The host site is gratefully
acknowledged for contributing personnel and
logistical support necessary to make the
test possible.
REFERENCES
1. Olexsey, R. A., "Incineration of
Hazardous Wastes in Power Boilers:
Emissions Performance Study Rationale
and Test Site Matrix," in Proceedings
of the Tenth Annual Research Symposium,
Incineration and Treatment of Hazardous
Waste, EPA-600-9-84-022,
September 1984.
2. Castaldini, C. et. al. "Field Tests of
Industrial Boilers Cofiring Hazardous
Wastes", Proceedings of the Tenth
Annual Research Symposium, Incineration
and Treatment of Hazardous Waste, P.57,
EPA-600-9-84-022, September 1984.
3. Adams, R. et. al. "Field Tests of
Industrial Boilers and Industrial
Processes Disposing of Hazardous
Wastes" Proceedings of the Tenth Annual
Research Symposium, Incineration and
Treatment of Hazardous Waste, P.62,
EPA-600-9-84-022, September 1984.
4. Chehaske, J. "Summary of Field Tests
for an Industrial Boiler Disposing of
Hazardous Wastes" Proceedings of the
Tenth Annual Research Symposium,
Incineration and Treatment of Hazardous
Waste, P.70, EPA-600-9-84-022,
September 1984.
5. Castaldini, C. et. al. "Engineering
Assessment Report, Hazardous Waste
Cofiring in Industrial Boilers," Acurex
Technical Report TR-84-159/EE, June,
1984.
220
-------
£= . 3
CHf—
C O
i- O
i— O
CL-*->
E O
d.
C C r— •—
O O 3 3
O E Li- Li.
• r- 4-1 . O
+-> T- \— •— t-
E C LO l_ LO
O O O OJ O
O £ >• D->-
CM- • •• C
O VI J W1 O
0 *J 0 •"-> -r- J
•O Cr- C 4-> O
C OJ •*- CU (O «—
ra -^ -t- M J=
tn OJ V ••-
-D C <-> C E •"->
ic ro vi « O O
O I- o
C
ro « en jc o o
O 1- « t- -^ O
_J I- 1 <-> «t "
CM U
O l-
CLI
O
o
§
CO
UJ
0
c
o
c c.
CJ C
u ~~
t,
UJ
CO
-------
TABLE 2. SUMMARY OF VOST RUMS
PRIMARY
DATE
25-ApE
25-Apr
25-Apr
27-Apr
27-Apr
27-Apr
08-Hay
08-May
OS-May
OS-May
OS-May
OS-May
OS-May
OS-May
OS-May
OS-May
OS-May
09-May
09-May
09-May
09-May
09-May
09-May
09-May
09-May
09-May
09-May
09-May
09-May
09-May
09-May
09-May
10-May
10-May
10-May
10-May
10-May
10-May
10-May
10-May
10-May
10-May
10-May
10-May
10-May
10-May
10-May
11-May
11-May
11-May
11-May
11-May
11-May
11-May
11-May
11-May
12-May
TIME
1445
1606
1732
1446
1626
1755
1154
1430
1447
1600
1642
1710
1740
2108
2224
2332
2357
20
643
921
1038
1142
1311
1410
1508
1702
1801
1946
2018
2153
2301
2346
716
945
1030
1132
1408
1452
1647
1750
1831
1917
2012
2055
2134
2222
2313
7
1330
1430
1528
1723
2033
2118
2225
2319
9
FUEL
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
OIL
STEAM
LOAD
(lb/hr)
64000
58000
57500
62000
57000
60000
60000
58000
60000
57000
59000
55000
54000
60000
61000
62000
76000
73000
62000
63000
57000
53000
51000
52000
25000
25000
28000
27000
25000
55000
35000
32000
32000
28000
28000
27000
27000
40000
44000
46000
50000
47000
47000
50000
47000
46000
50000
50000
50000
62000
60000
60000
60000
60000
60000
61000
60000
02
(%)
6.40
6.70
7.00
5.38
5.48
4.70
5.15
4.92
5.25
4.70
2.58
2.20
2.62
2.37
2.90
4.30
3.38
3.25
2.98
3.90
3.98
2.87
3.07
3.70
4.00
4.08
4.00
3.50
4.17
3.40
4.30
4.92
4.80
4.72
6.33
5.22
5.22
5.10
5.00
4.36
4.64
4.69
4.62
4.63
4.87
4.87
3.64
3.85
5.32
3.88
5.29
5.40
4.48
4.30
4.30
4.86
5.00
WASTE
FLOW
(gpm)
3.7
3.7
3.8
4.0
3.2
3.7
3.9
3.9
1.4
2.4
0.8
1.9
3.0
3.0
4.2
2.8
2.9
2.9
0.0
1.9
3.5
1.0
1.4
3.5
2.9
3.3
3.1
0.9
2.6
2.1
1.5
1.5
0.0
0.0
1.1
0.5
0.4
0.9
2.7
3.0
2.6
1.5
2.7
3.4
0.9
2.9
3.1
2.9
0.0
2.9
2.9
2.5
3.0
2.9
3.0
3.5
3.3
TEST
CONDITIONS
TSB BASELINE UNSPIKED WASTE,
TSB BASELINE UNSPIKED WASTE,
TSB BASELINE UNSPIKED WASTE,
TSB BASELINE, VOST
TSB BASELINE, VOST
TSB BASELINE, VOST
HIGH GPM BASELINE
HIGH GPM BASELINE
LOW GPM
NO WASTE ATOMIZATION
LOW WASTE GPM/02
LOW 02 WASTE INCREASE
LOW 02 BASELINE
WASTE OIL STARTUP, VOST
LOW WASTE ATOM STEAM (VARY) ,
SOOT BLOW
EA TRANSIENTS
EA TRANSIENTS
GAS BASELINE
START UP DUAL VOST
LOW 02
LOW 02 & WASTE GPM
LOW 02 & WASTE GPM
GAS BASELINE
LOAD REDUCTION
LOW LOAD, MODERATE 02, VOST
LOW LOAD, LOW 02
LOW LOAD/02/ WASTE GPM
WASTE & GPM TRANSIENTS
LOAD INCREASE
LOAD DECREASE
SOOTBLOW LOW LOAD
BASELINE OIL
BASELINE OIL
WASTE STARTUP
LOW LOAD/02/GPM
LOW LOAD BASELINE
LOAD INCREASE
BASELINE OIL, MODERATE LOAD
HIGH WASTE GPM/MODERATE LOAD
MODERATE 02/HIGH GPM
WASTE GPM TRANSIENTS
START INVERTED SPRAY
INVERTED SPRAY
INVERTED SPRAY
SOOTBLOW
MIN. USABLE WASTE ATOMIZATION
MIN. USABLE WASTE ATOMIZATION
LOAD INCREASE, NO WASTE
WASTE STARTUP
HIGH BASELINE
HIGH BASELINE
REDUCED WASTE ATOMIZATION
SOOTBLOW
POOR WASTE ATOMIZATION, VOST
HIGH 02, VOST
HIGH 02, VOST
VOST
VOST
VOST
VOST
222
-------
TABLE 3. ORE GROUPED BY TEST CONDITIONS
DATE TIME GC # FUEL LOAD
25-Apr
27-Apr
10-May
10-May
10-May
10-May
10-May
10-May
09-May
08-May
27-Apr
25-Apr
08-May
25-Apr
27-Apr
08-May
12-May
11-May
11-May
27-Apr
25-Apr
11-May
10-May
08-May
11-May
11-May
08-May
09-May
09-May
10-May
10-May
08-May
11-May
10-May
10-May
10-May
09-May
09-May
10-May
09-May
09-May
09-May
09-May
10-May
09-May
09-May
09-May
08-May
09-May
08-May
08-May
11-May
10-May
08-May
09-May
1408
945
716
1647
1750
1831
1410
1740
1626
1732
1430
1606
1755
1154
9
1528
2319
1446
1445
7
2313
1600
2225
2033
. 2224
20
2108
1030
1917
2108
1430
2134
2012
2055
1508
2301
1452
2153
921
1702
1946
1132
1801
1311
1142
1710
1038
1642
1447
2118
2222
2332
2346
MM5
MM5
96
90
87
101
103
104
67
48
FULL VOST
FULL VOST
36
FULL VOST
FULL VOST
35
FULL VOST
144
FULL VOST
FULL VOST
FULL VOST
122
118
40
FULL VOST
1 TRAP VO
FULL VOST
53
77
92
105
FULL VOST
143
109
107
108
69
83
97
81
57
FULL VOST
76
94
74
66
62
46
60
42
38
1 TRAP VO
111
50
86
GAS
GAS
OIL
OIL
OIL
OIL
OIL
OIL
GAS
GAS
GAS
GAS
GAS
GAS
GAS
GAS
OIL
OIL
OIL
GAS
GAS
OIL
OIL
GAS
OIL
OIL
GAS
GAS
GAS
OIL
OIL
GAS
OIL
OIL
OIL
OIL
GAS-
GAS
OIL
GAS
GAS
GAS
GAS
OIL
GAS
GAS
GAS
GAS
GAS
GAS
GAS
OIL
OIL
GAS
GAS
59000
59000
27000
28000
32000
44000
46000
50000
52000
54000
57000
57500
58000
58000
60000
60000
60000
60000
61000
62000
64000
50000
50000
57000
60000
60000
61000
73000
25000
28000
47000
60000
62000
47000
47000
50000
25000
35000
40000
55000
63000
25000
27000
27000
28000
51000
53000
55000
57000
59000
60000
60000
46000
62000
32000
CONDITIONS
TSB BASELINE
TSB BASELINE
LOW LOAD BASELINE
LOW LOAD BASELINE
LOW LOAD BASELINE
BASELINE OIL, MODERATE
HIGH GPM/MODERATE LOAD
MODERATE 02/HIGH GFM
GAS BASELINE
LOW O2 BASELINE
BASELINE
TSB BASELINE
TSB HIGH GPM BASE
TSB BASELINE
BASELINE
TSB HIGH GPM BASE
HIGH 02
HIGH BASELINE
HIGH O2
BASELINE
TSB BASELINE
MIN. DSABLE ATOM
MIN. USABLE ATOM
NO WASTE ATOMIZER
POOR ATOM
REDUCED ATOM.
LOW ATOM STEAM (VARY)
EA TRANSIENTS
GPM TRANSIENTS
WASTE STARTUP
GPM TRANSIENTS
WASTE OIL STARTUP
WASTE STARTUP
INVERTED SPRAY
START INVERTED SPRAY
INVERTED SPRAY
LOAD REDUCTION
LOAD DECREASE
LOAD INCREASE
LOAD INCREASE
START UP DUAL VOST
LOW LOAD, MODERATE 02
LOW LOAD/02/GPM
LOW LOAD/O2/GPM
LOW LOAD, LOW O2
LOW O2 & GPM
LOW O2 & GPM
LOW 02 WASTE INCREASE
LOW O2
LOW GPM/02
LOW GPM
SOOTBLOW
SOOTBLOW
SOOT BLOW
SOOTBLOW LOW LOAD
CCL4
NA
NA
99.996
NA
NA
3 99.9988
5 99.9996
99.99997
99.998
99.9997
99.997
NA
99.9995
NA
100
99.99990
99.99994
99.98
99.9997
99.996
NA
99.9997
99.9998
99.998
99.9998
99.99989
100
100
99.9990
99.989
99.9997
99.9997
99.998
99.9990
99.9989
99.9994
99.997
99.9995
99.998
99.9997
99.99994
99.9997
99.998
99.994
99.998
99.987
99.999904
100
99.9998
99.998
99.7
100
99.9993
99.998
99.9996
CHLORO-
BENZENE
NA
NA
99.9996
NA
NA
99.9997
99.9998
99.99993
99.9986
100
99.99991
NA
99.997
NA
99.99988
99.99991
99.99988
99.998
99.9998
99.99989
NA
99.99989
99.998
99.98
99.9998
99.9997
100
99.994
99.9994
99.993
99.99989
99.997
100
99.99992
99.9998
99.99994
99.9990
99.9997
99.9996
99.99993
99.9994
99.9997
99.998
99.995
99.9989
99.995
100
99.9998
99.9989
99.998
100
99.9998
99.9998
99.97
•99.9998
MMA
99.9997
99.999987
99.99992
99.99989
99.99987
99.999897
99.9997
99.9996
99.9998
99.9997
99.9997
99.9988
99.996
99.9996
99.9997
99.9995
223
-------
TABLE 4. ORE SUMMARIES
POHC
MMA
CC14 + C1 +
CC14
Cl(f>
CC14
CC14
CH
en*
No. tests
16
b 96
48
48
21
27
21
27
Series
All runs
All runs
All runs
All runs
Oil firing
Gas firing
Oil firing
Gas firing
Average ORE
99.9995
99.9949
(99.9980)3
99.9919
(99.9981)3
99.9980
99.9975
99.9874
(99.9985)3
99.9991
99.9971
^Excluding one low (99.7 percent) value for CC14.
"Monochlorobenzene
224
-------
TABLE 5. ORE SUMMARIES
ORE percent
Test series
Baseline
02 transients
GPK transients
Load transients
Low 02, load, waste
Atomization upsets
Inverted atomizer
Sootblowing
Total
Points
14
1
5
5
10
6
3
4
48
CC14
99.9974
100
99.9971
99.9988
99.9674
(99.9972)3
99.9995
99.9991
99.9992
99.9919
(99.9981)3
C1<|>
99.9994
99.994
99.9979
99.9995
99.9983
99.9962
99.9999
99.9924
99.998
aExcluding one low value for CC14.
225
-------
99.99999
99.9999 -
7? 99.999 -|
c
O)
<| 99.99
UJ
99.9
99
90
CCL4
OIL AND GAS
D
200 400
CO average (ppm)
600
Figure 1. CC14 ORE versus CO average.
226
-------
MOUSE - A COMPUTERIZED UNCERTAINTY SYSTEM
FOR ENVIRONMENTAL ENGINEERING ANALYSES
Albert J. Klee
Hazardous Waste Engineering Research Laboratory
U. S. Environmental Protection Agency
Cincinnati, Ohio 45268
ABSTRACT
Environmental engineering calculations involving uncertainties expressed as probabil-
ity distributions are far beyond the capabilities of hand analysis for any but the simpl-
est of models. There exist a number of computer simulation languages involving Monte
Carlo methods that certainly can do the job, but learning such languages and implementing
them on all computers is not the quickest nor the easiest of tasks.
MOUSE (an acronym for Modular Oriented Uncertainty System Emulator) deals with the
problem of uncertainties in models that consist of one or more algebraic equations. It
was designed to be used by those with little or no knowledge of computer languages or
programming. It is compact (and thus can run on almost any digital computer), easy and
fast to learn, and has most of the features needed for substantive uncertainty analysis
(built-in probability distributions, plotting and graphing capabilities, sensitivity
analysis, interest functions for cost analyses, etc.).
MOUSE has been used within USEPA for studying the migration of pollution plumes in
streams, for analyzing the uncertainties in establishing regulations for hazardous wastes
in landfills, and for investigating the variabilities inherent in pollutioq_control cost
estimation.
INTRODUCTION
If we define a model as a physical or
symbolic representation of reality, we find
among the set of all models one called the
mathematical model of which one particular
type consists of a series of one or more
algebraic equations. Mathematical models
of this type are extremely important for
they are found almost everywhere, includ-
ing economics", engineering, and science.
The use of an equation is understood by
almost everyone; in a somewhat "inelegant"
sense, numbers "go into" the equation and
an answer is obtained. For example, con-
sider the following very simple equation,
Y = AB
[1]
where Y might stand for the cross-sectional
area of a heating duct, given that A is its
height and B is its width.
equation 1 one has to know
and B. If A is equal to 2
and B is equal to 15, then
Often, however, we are hot
values of A or of B. A mi
might be 20; in such case,
equal to 60, not 30. The
certainty about the input
B, clearly the greater our
about the output variable,
To "solve"
the values of A
for example,
Y = 2(15) = 30.
sure of the
ght be 3 and B
Y would be
greater our un-
variables A and
uncertainty
Y.
TRADITIONAL APPROACHES TO UNCERTAINTY
The most often encountered approaches
to uncertainty in mathematical models are
(1) the best value approach, (2) the con-
servative approach, and (3) sensitivity
analysis. The first two are single-value
approaches. The "best" in "best value"'is
not precisely defined; generally it refers
227
-------
to some measure of central tendency such
as an average or a mode. In our duct
problem, we might suppose that the value of
A of 2 and of B of 15 are average values.
Hopefully, the answer of 30 is also some
sort of average value. As we shall see,
however, this is not always the case. Ad-
mittedly a simple technique, in reality the
best value approach is more a matter of
ignoring uncertainty than it is of any
conscious effort to come to grips with it.
The conservative approach does make an
attempt to consider the consequences of
uncertainty. As the name implies, the in-
put values selected are not the average or
most likely ones but rather those that pro-
duce conservative results with regard to
the consequences of over- or underestimat-
ing. For example, in the conservative
approach for the duct example, the values
of A and B selected to go into equation I
would be greater than their average values
of 2 and 15 since overestimation is probab-
ly better than underestimation in this
case. If "best" values were used, there is
a good chance that the duct area would be
underestimated.
The popular, traditional approach to
the problem of uncertainty is sensitivity
analysis. Sensitivity analysis is a combi-
nation of both the best value and conserva-
tive approaches since it usually starts
with a best value estimate, followed by a
change or perturbation in one of the input
variables (holding all other input vari-
ables at their previous values). The
perturbation can be either an increase or a
decrease in the value of the variable and
hence can be either of a "conservative"
nature or a "liberal" one. For example, to
examine the effects of a modest change in A
in equation 1, we might increase the value
of A by 10% over its "best" estimate value
of 2. A 10% increase in A (to 2.2) results
in an estimated Y-value of 33. If a value
of A of 2.2 is "reasonably likely" to oc-
cur, then the sensitivity analysis suggests
that a value of Y of 33 is also "reasonably
likely" to occur.
PROBLEMS WITH TRADITIONAL APPROACHES
As has been mentioned, the best value
approach really does not address the prob-
lem of uncertainty at all. For one thing,
best input values do not necessarily have
high probabilities of occurring. There is
another difficulty with the best value
approach which is not generally recog-
nized. The difficulty arises when the
algebraic model contains non-linear ele-
ments such as multiplications or divisions
and the variables are correlated. If
variables A.and B of equation 1 were
correlated, for example, the average of Y
would not be equal to the average of A
times the average of B. In point of fact,
if A and B were positively correlated,
then the average of Y would be greater
than the product of the averages; con-
versely, if A and B were negatively cor-
related, the average of Y would be less
than the product of the averages.
The conservative approach also has its
deficiencies. For one thing, in a complex
calculation involving many equations and
many input variables (some of which may be
correlated) it may not be obvious what
values of the input variables constitute
"conservative" ones with respect to the
output. Secondly, because conservative
input values generally are those with a
low probability of occurring, the esti-
mates obtained by using such values per-
force will not have a high probability of
occurring. It should be recognized that
the point estimates involved in using
either the best value or the conservative
approach do not utilize all of the infor-
mation that is usually available. One
usually has at least some idea of the un-
certainties in the input figures.
Sensitivity analysis, being largely an
amalgamation of elements of both the best
value and conservative approaches, suffers
the defects of both methods. An arbitrary
change in the value of an input variable,
even though the change falls within the
expected range of the variable, tells us
little about the likelihood of occurrence
Of the new estimate obtained. In other
words, if we know little about the likeli-
hood of such a change occurring, it fol-
lows that we know little about the likeli-
hood of the calculated output occurring.
Furthermore, in sensitivity analysis all
other variables are held at their previous
values, the so-called "all other things
being equal" view of the world. The prob-
lem is that "all other things" are seldom
equal. In actuality, the change we intro-
duce in a variable under sensitivity
analysis may well be either mitigated or
intensified by what is happening to the
other variables. In short, sensitivity
analysis does not show the combined net
228
-------
effect of changes in all variables or the
likelihood of various changes occurring
together. Viewed in this manner, the
traditional sensitivity analysis can be
misleading.
ALTERNATIVE SOLUTIONS TO THE UNCERTAINTY
PROBLEM
Let us examine three alternative solu-
tions to the problem of uncertainty. The
first method, illustrated in Figure 1, is
Direct or Complete Enumeration. The model
of equation 1 is employed, and we assume
the uncertainties of A and B as given in
the two probability distributions for these
variables shown in the upper left-hand
corner of the figure. In other words, we
suppose that there is a 25% chance that A
is equal to 1, 50% that it is equal to 2,
and 25% that it is equal to 3. For B,
there is a 50-50 chance that it is equal
to either 10 or 20. (For this simple
example, we assume no correlation between
the two input variables.) In complete
enumeration we list all of the possible
combinations of the input variables and
then calculate the probabilities of these
combinations occurring. In this example
there are 3 choices for A and 2 for B, re-
sulting in 6 possible outcomes for Y. The
probabilities of these combinations are
•shown in the middle top of the figure.
Since some of the combinations are duplica-
tions, the table of combinations of A and B
may be simplified to the 5 entries shown at
the upper right of the figure for method 1.
The average value of Y is shown to be 30.
At the bottom of the figure for method 1 is
a graph of the frequency or probability
distribution of Y. Note that the most
likely value is not the average but rather
values to either side of it. Furthermore,
one of the extreme values (Y = 60) has a
higher probability of occurrence than has
the average value. As can be seen, the
complete enumeration method tells us every-
thing about the distribution of Y, including
its mean, standard deviation, minimum,
maximum, and the probability of occurrence
of any value of Y.
The second method is the Probability
Calculus method. The method, as the name
implies, requires some knowledge of the
'calculus of probabilities (sometimes known
in engineering as the "propagation of
error"). Using the model of equation 1 as
before, the method is also illustrated in
Figure 1. The error formula is given in
the figure, and involves three terms and
knowledge of the variances of A and B.
The latter are calculated, as is shown in
the figure, from the probability distribu-
tions of A and B given previously in the
Direct Enumeration method. The error
formula shows the variance of Y to be 225,
i.e., its standard deviation = 15. The
probability calculus method produces no
more than the mean and the standard devi-
ation of the output (i.e., Y) distribu-
tion. The standard deviation alone, how-
ever, is not sufficient to determine the
nature of the uncertainty in a mathemati-
cal model.
The third method is a form of Monte
Carlo simulation known as Model Sampling.
The idea of Model Sampling is relatively
simple:
a.
b.
A value for each of the input
variables is drawn at random from
its respective probability dis-
tribution, and the model is com-
puted using this particular set
of values.
The above process is repeated
many times. Since the results
vary with each iteration, the
outputs themselves (i.e., the
Y's) are gathered in the form of
a probability distribution. Thus
the uncertainties of the model's
in- puts are transferred to the
output which can then be studied
and subsequently utilized ;ui
decision processes .
The procedure is shown schematically in
Figure 1. The output of the Monte Carlo
simulation method becomes almost identical
to that of complete enumeration as the
number of iterations becomes large. Un-
like Direct Enumeration, however, large
and/or complex problems are tractable and
continuous uncertainty distributions are
easily handled. The Monte Carlo simula-
tion method forms the basis for MOUSE,
the computerized uncertainty analysis
system which is the subject of this paper.
MOUSE, A COMPUTERIZED UNCERTAINTY ANALYSIS
SYSTEM
Monte Carlo simulation requires the
use of a digital computer for any substan-
tive problem. It is important, therefore,
to consider the desirable characteristics
229
-------
METHOD 1: DIRECT (COMPLETE ENUMERATION)
MODEL: Y =AB
A
1
2
3
p(A) B p(B) A
B AB p(AB)
.25 10 .50 1 10 10 .125
.50 20 .50 2 10 20 .250
.25 3 10 30 .125
1 20 20 .125
2 20 40 .250
AB p(AB)
10
20
30
40
•0
.125
.375
.125
.250
.125
* 3 20 60 .125
.40-
.30-
.10-
.00
|
I
Note: Average = Y =
Ai = 2(15)
= 30
0 10 20 30 40 50 60 Y
PROBABILITY DISTRIBUTION OF Y (= AB)
METHOD 2: PROBABILITY CALCULUS
MODEL: Y = AB
ERROR FORMULA IS var (AB) = vir (Y) = A 2 var (B) + B 2 var (A) + var(A) var (B)
A = 2 and § = 15
£f*I A (* - *)2 P(A) (A . A)2
.25 1 1 .25
.50 2 0 .00
.25 3 1 .25
var(A) = .50
therefore,
and
(B) £ (B - B)2 p(B) (B - B)2
.50 10 25 12.5
.50 20 25 12.5
var(B)= 25.0
(0>50M25-0) _ 225
•Id (Y) = 15
METHOD 3: MONTE CARLO SIMULATION
START: I - 1
i
random -f random
•arnpl* I I sample
record Y|
repeat n llmtt
FINISH
FROM COLLECTION OF Y'» OBTAIN:
1. MEAN
2. STANDARD DEVIATION
3. COEFFICIENT OF VARIATION
4. MINIMUM
5. MAXIMUM
6. GRAPH OF FREQUENCY
DISTRIBUTION
7. GRAPH OF CUMULATIVE
FREQUENCY DISTRIBUTION
Figure 1. Alternative Solutions to the Uncertainty Problem
230
-------
of any computerized uncertainty analysis
system.
Table 1 presents a comparison among
general purpose computer languages, general
purpose simulation languages, and MOUSE
(the system introduced in this paper) for
uncertainty problems that deal in sets of
algebraic equations. Table 1 is not de-
signed to devalue the capabilities of
either general purpose computer languages
or general purpose simulation languages.
MOUSE is a restricted special purpose
simulation language and the table makes its
comparison assuming that the problem at
hand is of this restricted form, i.e., that
the model consists of one or more algebraic
equations. For such restricted models,
MOUSE is clearly superior to the other two
groups of languages. It is concise, power-
ful, and convenient to use. MOUSE can
solve uncertainty problems faster and
easier than can other languages. With
MOUSE, the'user's attention is on problem-
solving, rather than on the details of
coding a program to compute a solution.
Further, MOUSE programs are easier to
understand, easier to explain to others,
and easier to modify than are general
purpose languages.
APPLICATIONS OF MOUSE
It is not possible to go into the de-
tails of MOUSE in this paper (a detailed
example of a MOUSE application within
USEPA will be presented in the following
paper titled "Uncertainties and Incinera-
tion Costs: Estimating the Margin of
Error," by Gordon Evans) but a typical
output is shown in Figure 2. Other appl i -
cations of MOUSE in EPA have included: a
facilities design tool cost model to con-
struct and close a surface impoundment or
landfill facility, a waste pile costing
model required to construct and close a
waste pile facility, the use of engineer-
ing fault tree analysis in failure analy-
sis of RCRA land disposal facilities, cost
models for systems for the incineration
of hazardous wastes, and an investigation
of a model to determine an appropriate
level for regulating organic toxicants in
hazardous wastes.
231
-------
TABLE 1. A COMPARISON OF COMPUTER LANGUAGES FOR UNCERTAINTY ANALYSIS
| Language
Element |
Simplicity
Built-in Proba-
bility functions
Built-in Inter-
est functions
Automatic Output
Decimal Numbers
Automatic Sens-
itivity Analysis
Correlation
Among Variables
De-bugging
Simple
Transportability
Aid To Deriving
Probability
Distributions
Ability to
Expand by
Modules
General Purpose
Languages
(e.g., FORTRAN,
PASCAL, BASIC)
compl ex
no
no
no
yes
no
must be
programmed
no
any
computer
no
yes
General Purpose
Simulation Language
GPSS | SIMSCRIPT
simple
a few
no .
yes
no
no
no
no
mini and
mainframes
no
no
complex
yes
no
yes
yes
no
must be
programmed
no
mini and
mainframes
no
no
SLAM
complex
yes
no
yes
yes
no
must be
programmed
no
mini and
mainframes
no
yes
MOUSE
simple
yes
yes
yes
yes
yes
yes
yes
any
computer
yes
yes
232
-------
DISTRIBUTION FOR QUANTITY FACTOR
NUMBER OF ITERATIONS = 1000
MEAN =
MINIMUM =
MAXIMUM =
19814.26758
5150.11523
73638.«
STANDARD DEVIATION = 12020.29688
COEFFICIENT OF VARIATION, 7. = 60.66485
LOHER NUMBER OF
LIMIT ENTRIES
6600.
11300.
20700.0000
39500.
44200.
48900.
53600.
67700.
72400.
PERCENT CUMULATIVE CUMULATIVE
ENTRIES X ENTRIES COMPLEMENT
DISTRIBUTIONS
* = FREQUENCY DISTRIBUTION
0 = CUMULATIVE DISTRIBUTION
21.
221.
253.
185.
95.
55.
37.
46.
32.
22.
10.
5.
12.'
3.
2.
1,
2.10
22.10
25.30
18.50
9.50
' 5.50
3.70
4.60
3.20
2.20
1.00
0.50
1.20
0.30
0.20
0.10
2.10
24.20
49.50
68.00
77.50
83.00
86.70
91.30
94.50
96.70
97.70
98.20
99.40
99.70
99.90
100.00
97.90
75.80
50.50
32.00
22.50
17.00
13.30
8.70
5.50
3.30
2.30
1.80
0.60
0.30
0.10
0.00
*0«s
»*«l
**«
«HH
««
*#«
***J
•iHHH
««i
*»*<
«*
*S
iHHfr'
»
*
*
««*#********»**»»*«
Figure 2. Statistics, Histogram and Graphs Produced
by MOUSE for a Typical Problem
233
-------
UNCERTAINTIES AND INCINERATION COSTS:
ESTIMATING THE MARGIN OF ERROR
Gordon M. Evans
United States Environmental Protection Agency
Hazardous Waste Engineering Research Laboratory
Cincinnati, Ohio 45268
ABSTRACT
It is standard practice in cost estimation to place percentage confidence bands around
final estimates in order to indicate the expected margin of error. This envelope of varia-
bility offers no useful information of the expected occurance of any point estimate within
that band. An estimation procedure that utilizes an in-house computer program overcomes
this limitation by producing estimates in the form of frequency distributions which repre-
sent the the probabilities associated with points along the entire range of expected out-
comes. This paper presents the results of an application of this technique to the problem
of estimating the costs involved in the incineration of hazardous wastes. The end result
is the production of information that allows a decisionmaker to explicitly relate the
consequences of a wrong decision (in terms of dollars lost) to the margin of error he
chooses to accept.
INTRODUCTION
The goal of this paper is to provide
the reader with a description of an economic
uncertainty analysis conducted on hazardous
waste incineration costs at the U.S. Envi-
ronmental Protection Agency's (USEPA)
Hazardous Waste Engineering Research Labora-
tory (HWERL). This project utilized MOUSE
(Modular Oriented Uncertainty Systems
Emulator) (I), a computer utility program
developed by Dr. Albert J. Klee of that
Laboratory. This computer program allows an
analyst to define a model's crucial vari-
ables in probabilistic terms. These prob-
abilities provide information on the uncer-
tain nature of the crucial variables and are
specified through sampling the opinions of
revelant experts. The primary objective in
applying this stochastic process to a cost
engineering model (in this case, one con-
structed to provide estimates on the costs
associated with the incineration of haz-
ardous wastes) is simply to provide added
information to the decision process.
With uncertainty analysis, the decision-
maker is offered cost estimates which allow
him to explicitly incorporate his own risk
preferences into the decision process. These
estimates are generated with the help of the
MOUSE program and come in the form of fre-
quency distributions. In essence, these
frequency distributions reflect expert opin-
ion on the various sources of uncertainty
inherent in the system under question. In
the case at hand, the final MOUSE estimates
show the range of values that selected dir-
ect and indirect cost items may take, the
frequency with which each particular value
within that range is predicted to occur, and
the associated statistical information re-
garding each distribution. Thus, with the
MOUSE output, the decision maker is not
limited to a single point estimate for each
system component; instead, he recieves
both graphical and statistical information
234
-------
that will allow him to ascertain with great
confidence the margin of error associated
with any estimate falling within that range.
The utility of this added information
is highlighted through reference to the
typical cost estimating situation. Nor-
mally, such a procedure produces an esti-
mate in the form of a single point. The
problem lies in the fact that this tradi-
tional estimating method offers no useful
information on the possible range of out-
comes or the probability of their occurance.
Therefore, the traditional point estimate is
subject to some overall margin of error.
It is important to understand that this
term "margin of error" is often used without
fully understanding it's meaning. As used
within this paper, the phrase "margin of
error" encompasses two separate and distinct
notions; measurement error and the error that
results from the effect of the system's
underlying uncertainty (2). It should be
clear that the decisionmaker, when offered a
point estimate, is seldom given complete
information regarding the meaning of the
margin of error. It has been standard prac-
tice to attempt its representation through
the construction of an envelope of varia-
bility around the point in question (e.g.,
"the probable accuracy is +_ 40 percent") (3).
While this technique provides an indication
as to the possible range of outcomes (taking
into account the potential for measurement
error), it says nothing significant about the
probability associated with any particular
point within that range. For all intents and
purposes, the occurence of each outcome with-
in that confidence band is equally likely.
Thus, while the technique of constructing an
envelope of variability tries to deal explic-
itly with the margin of error concept by
offering a solution to the problem of meas-
urement error, it fails to provide the deci-
sionmaker with any useful information on the
underlying uncertainty. Nonetheless, it is
possible to obtain a realistic appraisal of
the margin of error. By utilizing a cost
model which incorporates the uncertainty
approach afforded via the MOUSE technique, a
decisionmaker is given all the information he
needs to fully comprehend the real margin of
error associated with a given estimate.
Complete information on the margin of
error has added significance when one con-
siders the constraints placed on decisions
that are often made within institutional
settings. It is not at all unrealistic to
assume that the average decisionmaker (or
the institution he represents) will attempt
to avoid risky situations; he will tend to
be risk adverse. Economists recognize that
people are primarly motivated to pursue
their own self-interest. Thus it is assumed
that the negative consequences to a deci-
sionmaker are likely to be greater for
underestimating the cost of a given project
(here the cost to incinerate a hazardous
waste) than for overestimating that same
cost (4).
To help emphasize this point consider
for a moment the plight of a risk adverse
corpor'ate executive (or government bureau-
crat) who is asked to choose between two
competing investment projects, "A" or "B".
Both of these projects will accomplish the
same goals and both will involve consider-
able capital expenditures. Suppose our
executive is offered cost estimates for both
projects; each with 40 percent envelopes of
variability attached to them. Suppose as
well that Project A has the lower point es-
timate value. With all other factors being
equal, and recognizing the risk adverse
nature of our decisionmaker, we may assume
that he will likely choose Project A. After
all, he possesses no real information on
either project's underlying uncertainty. As
far as he is concerned, all values falling
within the two 40 percent confidence bands
have an equal likelihood of occurring. As
such, the expected value of Project A re-
mains less than that of Project B.
Despite the addition of the envelopes
of variability, this decisionmaker was of-
fered no real understanding of the true
margin of error associated with either Proj-
ect's estimate. Now, if in retrospect, he
finds that the expenditures on Project A
were in fact lower than estimated, he can
be certain that his credibility within
that institutional setting will be enhanced.
More importantly from his perspective, he
will have successfully avoided placing
himself (and the institution he repre-
sents) into the risky situation he had
235
-------
hoped to avoid. Conversely, if the actual
expenditures end up being significantly
greater than estimated, our risk adverse
executive will find himself in a most
uncomfortable position.
Regardless of the final outcome, the
point that needs to be emphasized is that
this decisionmaker did not have access to
any information that would have allowed him
to ascertain the true margin of error asso-
ciated with either estimate, thereby giving
him final control over the level of risk he
chooses to accept. By offering a decision-
maker a cost estimate in the form of a MOUSE
generated frequency distribution, he will be
able to explicitly relate the negative
consequences of a wrong decision (in terms
of dollars lost) to an acceptable margin of
error. In other words, he will now be able
to quantify his attitudes toward risk and
act accordingly.
In terms of this example, suppose that
a HOUSE output had indicated that the mean
estimate for Project A was less than that
for Project B. This is similar to the pre-
vious situation where the value of the point
estimate offered for Project A was smaller.
But suppose that after viewing the frequency
distributions for both estimates it was
obvious to this decision maker that the
range of estimates generated for Project A
exhibited a much greater variance. This
risk adverse decisionmaker, now able to
explicitly consider the system's uncer-
tainity, might well find that selecting
Project B actually minimizes the margin of
error thereby reducing the level of risk he
accepts. With the simple assumptions of-
fered here, it is quite plausible that at a
95% confidence level the estimated cost of
Project B would be lower. Once our deci-
sionmaker determines the level of risk that
is acceptable to him (5 percent) he is able
to select from among the project estimates
that which h.as the lowest cost and which, at
the same time, satisfies his preferences
toward risk.
By using an example of estimating the
cost of hazardous waste incineration, this
paper will show how the application of un-
certainty analysis (via MOUSE) can improve
the nature of the information provided in
the decisionmaking process. Specifically,
this paper will use the incineration issue
to provide an answer to one major question:
"How much more information regarding the
margin of error does uncertainty analysis
provide a decision maker over the tradi-
tional practices?"
APPROACH
Before an uncertainty analysis could be
conducted, it was necessary to develop a
mathematical model which would represent the
engineering and cost relationships for each
of the 3 major incinerator configurations
(i.e., rotary kiln, liquid injection, and
multiple hearth). The specification of these
relationships had to be flexible enough to
allow for a representation of incinerators
operating under a variety of design para-
meters and utilizing various hazardous waste
streams. Once specified, these engineering
relationships could then be used to determine
the type and size of the incinerators'
structural components and to specify the
resources (e.g. utilities, labor) necessary
for its' operation. Experts could then be
used to specify the cost of these items in
probabilistic terms, and once formatted,
these judgments could be incorporated into
the MOUSE program.
Upon reviewing the incinerator cost
literature, it became clear that none of the
existing cost models readily lent themselves
to the MOUSE programing requirements (5,6,7).
For example, the cost model developed by
Industrial Economics (6) for the USEPA's
Office of Solid Waste was constructed using
a linear programming approach (i.e., finding
the least cost configuration on a production
frontier). Unfortunately however, a linear
programming approach was incompatible with
the MOUSE technique. One model, however,
was nearing completion and was being con-
structed in a format ideally suited to the
MOUSE programming requirements.
Under contract to the EPA, Acurex, Co.
was in the draft stage of a report entitled
"Capital and O&M Cost Relationships for
Hazardous Waste Incineration" (8). Their
model asked its user to specify the incin-
erator's parameters from a menu of design
and operating configurations. These design
236
-------
parameters included items such as the type
of air pollution control device (APCD),
energy recovery system, and geographical
location of the facility (which would effect
input cost variables such as fuel and labor).
Users were also asked to specify the chem-
ical composition of the waste stream they
wished to simulate in order to establish the
systems stoichiometric combustion require-
ments.
By using the parametric information
from the selected design options the user
worked through a series of engineering cal-
culations, which in turn completed the spec-
ification of the cost equations. Ultimatly
these equations produced point estimates for
numerous items including total capital cost
and unit disposal cost, among others.
With the Acurex report serving as the
basis for the underlying mathematical model,
there now existed an expert on the incin-
erator's component costs; the author of the
Acurex report. His judgment of the uncer-
tainties that surround each of the input
cost variables would provide the basis for
the specification of appropriate probability
distributions.
The construction of the computer pro-
gram consisted of two activities. First the
mathematical relationships expressed within
the Acurex report had to be encoded into the
MOUSE language and second, the expert's un-
certainty judgments had to be fitted to
probability distributions for their eventual
encoding into MOUSE.
Given the clear presentation and organ-
ization of the Acurex report, the processing
of encoding of the cost equations provided
few technical problems. In the end, the
cost relationships were reduced to a com-
puter program of about 600 lines. With that
portion of the program written, the next
step was to solicit and encode our expert's
uncertainty judgments.
Under ideal conditions an analyst
should seek to spend a considerable amount
of time with the revel ant expert to insure
that an accurate assignment of probabilities
takes place. A variety of interview tech-
niques exist to help the analyst elicit
these judgments (9). Research in this area
has shown that an expert's judgment on a
system's uncertainty is often clouded by a
variety of biases. These biases are catego-
rized as being either motivational or cog-
nitive in nature. Motivational biases are
defined as the conscious or unconscious
influences on one's judgment that are caused
by personal interests or previous commit-
tments. Cognitive biases, on the other
hand, operate on a persons judgment in a
more subtle fashion. They can be viewed as
problems created by ones intuitive per-
ception of probabilities. Experience has
shown that a structured elicitation process,
such as an interview, helps the expert being
surveyed to detect and reduce the influence
that these biases have on the determination
of uncertai nties.
However, for the project at hand,
access to our cost expert was limited, and
as such it was decided to expedite the proc-
ess by providing him with a list of the cost
variables in question and allowing him to
make the uncertainty determinations on his
own. Recognizing the negative influence
that bias may have on judgement, our expert
was also given a short paper detailing the
various types of bias and their effect.
Lastly, to insure that he had a complete
understanding of the project's methodology,
he was provided with supporting materials
that included an explanation of the aims of
our uncertainty analysis and a listing of
the probability distributions processed by
MOUSE.
The final product of his probability
estimates was limited to the specification
of envelopes of variability for each of the
individual input variables question. For
example, the actual cost of a compressor, in
his estimation, could be off the given point
estimate by as much as + 25 %, with the
original point estimate~~bei ng offered as the
most likely. Because of the limited scope
of this information on the cost uncertain-
ties, it was decided to utilize only the
three most basic probability distributions
provided by the MOUSE program: continuous
uniform, triangular, and trapezoidal (Figure
1). In the final version of the model, the
continuous uniform distribution accounted
for 48 percent of the 151 distributions
237
-------
0.25 0.75
Continous Uniform Distribution
0.25 0.60
Triangular Distribution
0.75
0.25 0.35 0.65
Trapezoidal Distribution
0.75
(Figure 1)
238
-------
specified, while the trapezoidal and tri-
angular accounted for 46 and 6 percent,
respectively.
Although the procedures followed in
assigning these probabilities were less than
ideal, they nonetheless produced acceptable
estimates of the uncertainties involved and
allowed the model to be excercised with some
degree of confidence. The failure to employ
a more structured interview process with our
cost expert has been recognized as a major
shortcoming of this analysis and will be
given a higher priority in future applicat-
ions of uncertainty analysis.
RESULTS
As stated earlier, this paper tries to
answer one major question: "How much more
information does the uncertainty analysis
provide?"
In order to answer this question, it
was first necessary to construct a scenario
which would reflect a typical hazardous
waste incineration facility (10). The one
selected for this purpose was a commercial-
scale rotary kiln hazardous waste inciner-
ator operating at 80 million Btu/hr. It was
assumed that this facility was burning a
combination of a low Btu liquid waste, a
medium Btu sludge, and a high Btu container-
ized waste. In addition, the facility was
assumed to be efficiently operating at 80
percent of its peak capacity (where peak
capacity is defined as a 24-hour, 7-day,
52-week operating schedule). Lastly, it was
determined that this facility would employ a
waste heat boiler, allowing for an energy
recovery credit to be deducted from the
annual operating costs.
After calculating the point estimates
generated by the Acurex model, the MOUSE
version of the model was exercised. While
this paper will focus on the estimates of a
few of the key individual cost components,,
the MOUSE model is capable of producing
frequency distributions for any cost items
found within the model. It should be noted
that the costs estimates generated by the
uncertainty model compare quite favorably to
those costs reported in the hazardous waste
literature for similar operations (11,12).
Figure 2 shows the MOUSE output for the
unit disposal cost for incinerating the
hazardous waste mix specified in the base
facility on a per pound basis. The fre-
quency distribution shows a range of esti-
mates falling between $0.02658/1b and
$0.050378/lb ($68.92/ton to $100.76/ton)
with a mean value of $0.03646/lb
($72.92/ton). The point estimate, generated
by the Acurex Model without the benefit of
the uncertainty analysis, is $0.0353/lb
($70.60/ton). Keep in mind that the unit
disposal cost reported here simply refers to
costs faced by the waste disposer (averaged
over the yearly throughput) and is not
meant to reflect fees charged to waste gen-
erators.
From the information supplied by the
MOUSE model in Figure 1* one can see that
there is a 47 percent probability that the
actual value of the unit disposal costs will
be less than or equal to the value of the
point estimate. More importantly to the
risk adverse decision maker, this means that
there is a 53 percent chance that the actual
unit disposal cost will exceed the value of
the point estimate. This 53 percent margin
of error is likely to be too great. Suppose
that this decisionmaker feels comfortable
accepting no more than a 5 percent proba-
bility that he will underestimate the unit
disposal cost. Given this risk preference,
he would choose a value that corresponds to
the 5 percent level in the cumulative com-
plement column, or conversely the 95 percent
level of the cumulative entries column.
This value is 0.0418/lb ($83.60/ton).
At first glance the difference between
the value of the point estimate and that at
the 95 percent probability level appears
small ($0.0065/lb or $13.00/ton). However,
when one considers that the total yearly
throughput of waste at the facility under
question is 42,048 tons, the difference be-
tween the two estimates (on an annual basis)
is $546,623. Making an error of this magni-
tude could be of significant consequence to
any decision maker. Thus, the information
gained on the margin of error by using the
uncertainty approach, as measured by this
dollar differnce, is appreciable.
Assume that this risk adverse decision-
239
-------
CO CM
CO LO
t-n oo
in o
to o
• •
en o
u ii
CO
I—I
o:
00
I—I
a
LU
cs
co
0£. SB
-l
(— LU
SB O
LU
I , Q
O DC.
I—I e^
U. O
o l-
o to
u
=«=
UJ
CO
oo
i-i LU
o: i—
irocrit—iococnt—i
C3COLf)oorHcn^o«*cMCTir--LnroocotD"^-i—iv|-'^-'^-«^-d->vl-<^-Di
O CD O CD CD CD CD CD CD CD CD O CD OOOOOOOOOOOOOOOLU
OOOOOOOOOOOOOOCDCDOOCDOCDOOOOOOOO
240
-------
maker had simply decided to construct a 40
percent envelope of variability around the
original Acurex estimate. Now suppose that
in order to obtain the most conservative
estimate he had taken the upper limit of
that range as his estimate. (The Acurex
model predicts an accuracy of _+ 40 percent.)
The maximum value of an estimate within that
confidence band is $0.0494/lb. or
$98.80/Ton. By accepting this conservative
estimate over the MQUSE value (taken at the
5 percent margin of error level) this
decisionmaker would have needlessly over-
estimated the yearly unit disposal cost by
$639,130. Assuming that a 5 percent margin
of error is an acceptable level of risk,
this upper limit confidence band estimate
represents a needless degree of caution.
Another example of the gain in usable
information obtained from uncertainty ana-
lysis occurs when the point estimate for the
total capital investment of this same base
facility is compared to one generated by the
MOUSE model. Figure 3 shows the MOUSE out
put for that cost component. Compare that
to the point estimate generated by the
Acurex model, $8,614,669. By examining the
MOUSE output, one finds that this point
estimate corresponds to an approximate level
of cumulative entries of 12 percent, indica-
ting that there is an 88 percent probability
the actual value will exceed this point es-
timate.
If the decisionmaker wishes to reduce
his margin of error (and thus his level of
risk) from the 88 percent level to a more
acceptable 5 percent level (as in the pre-
ceding example), he will choose $11,164,258
as the value of his estimate for the total
capital cost. The difference between these
two estimates is $2,549,589. This is an in-
crease of close to 30 percent over the point
estimate generated by the original model.
Once again this is a significant difference
which reflects the dollar value associated
with the margin of error. One should note
that the MOUSE estimate taken at this 5
percent risk level falls well within a 40
percent envelope of variability constructed
around the the point estimate, and as before,
it provides the added information on proba-
bility which prevents an over-estimation of
the total capital cost by $896,278 (given
the risk preference of the decisionmaker).
CONCLUSION
After viewing the significant dif-
ference between the values of the final cos-
estimates generated with and without the
MOUSE program, it is reasonable to conclude
that this particular decision process was
enhanced through the provision of additiona'
information on uncertainty. The advantage
gained by offering cost estimates in the
form of frequency distributions is that the
decisionmaker has the ability to incorporate
his risk preferences at the onset of the
decision process.
In coming to this conclusion, one im-
portant issue has been purposely overlooked.
The success of this analysis ultimately
rests with the decision maker. He needs to
bring to the decision process an apprecia-
tion of his personal preferences toward
risk. There is no reason to believe that
the typical decision maker will have taken
the time and effort to objectivly review his
attitudes toward risk, especially when con-
sidering that the opportunities for such
reflection are rare. The chances are such
that he will never have been placed in a
situation which has required him to do so.
It is hoped that through the adoption
and regular use of an uncertainty approach
to cost estimation, the decision maker will
be forced to deal with the concept of risk.
Then, and only then, will he be motivated to
recognize and act upon his own risk
preferences.
This paper has concerned itself with
the problems faced by the risk adverse
decision maker, yet it should be made clear
that the information obtained from these
frequency distributions can benefit anyone,
regardless of their preferences toward risk.
All that is required is for one to judge
what margin of error is acceptable and
select that corresponding value from the
distribution. As this paper has shown, when
properly conducted, uncertainty analysis can
vastly improve both the quality and quanity
of information regarding an estimate's
margin of error. The end result will be a
more reliable decision process.
241
-------
o
«^- LO
r~~ CM
Lf> r-t
CO CO
• •
CO CO
o
1—4
CM
II II
n
o
t—t
ss
CO
LU
o:
C3
>—4
LL.
i—i O
o o:
t-t «a:
Li_ a
LL. Z
LU 3
o (—
«-> oo
S
CO
I
a.
s
o
e
z
o
*—4
Z3
CQ
00
Q
S§8
• • •
ID in r-.
.-< uo *d-
co oo en
to cy> LO
o CM «*
co o oo
A *» •»
l»- CO CT>
II II II
CO
I—I
a:
oo
t—t
a
n
=**=
CO,
i — i
cc
to
I— I
a
o-
UJ
ce
i — I LU
\- s:
=> O
CJ> O
> oo
H-1 I t I
I— 1-1
•a: a:
01 I—
LU HH
i—lOLOi-Hi—ICMLfJOi—I
'—1OO
ooooT-Hcoir>CTi<*ocoir><3-ooocoi-n'— i
OO CM AO IO CO •* O
orot— )roi^.oocr)O
i — cocncricricncno
OOOC3OOOOOOOOC3OOOCDOCDOCDOC5OOOOOC33
—1OC3CDCDOC3OOOCOOC3OOC3OOOC3OC3C5OOOOOOO
U-OOOOOCDOC3OCDOOOC3OC3OCDCDOOOOOOOOO —I
CJiOOOOOOOOOOOOOOOOOOOOOOOOOOOOLL.
-
242
-------
REFERENCES
1. Klee, Albert 0., 1985. MOUSE Manual . 10.
Hazardous Waste Engineering Research Lab-
oratory, USEPA, Cincinnati, Ohio. An
unpublished users quide.
2. Goddard, Haynes C. Using Uncertainty An-
alysis to Facilitate Environmental Dec-
ision Making. To appear i n The Environ-
menta Professional. 11.
3. Peters, Max S. and Klaus D. Timmerhaus,
1974. Plant Design and Economics for
Chemical Engineers. McGraw-Hill. Chap-
ter 4.
4. Friedman, Milton and Rose, 1981. Free
To Choose. Avon. pp. 197-199.
5. Kapner, Mark, Efim Livshits, Amitava
Podder and David Woodbridige, 1981.
The Economics of Hazardous Uaste Incin-
eration. Hittman Assoc., Columbia,
Maryland. Prepared for USEPA under
Contract No. 68-03-2566 T3006.
6. Cost Model for New Hazardous Waste
Inci nerators, 1983. Industrial
Economics, Cambridge, Massachusetts.
Prepared for Office of Solid Waste,
USEPA.
12.
7.
9.
Vogel, Greg, Irwin Frankel, and Neil
Sanders, 1983. Hazardous Waste Incin-
eration Costs. Proceedings of the Eighth
Annual Research Symposium. Industrial
Environmental Research Laboratory,
USEPA, Cincinnati, Ohio. EPA-600/
9-83-003.
McCormick, R.J. and R.J. DeRosier, 1983.
Capital and O&M Cost Relationships for
Hazardous Waste Incinerators. Acurex
Co., Mountain View, California. Prepared
for USEPA under Contract No. 62-02-3176
and 68-03-3043.
Matheson, James E. and Carl-Axel S. von
Hoi stein, 1979. A Manual for Encoding
Probability Distributions. SRI Inter-
national, Menlo Park, California. Pre-
pared for the Defense Advanced Research
Projects Agency: SRI Project 7078.
Frankel, Irwin, Neil Sanders and Greg
Vogel, 1983. Profile of the Hazardous
Incineration Manufacturing Industry.
Proceedings of the Eighth Annual
Research Symposium. Industrial Environ-
mental Research Laboratory, USEPA,
Cincinnati, Ohio. EPA-600/9-83-003.
National Rural Electrical Co-op Assoc-
iation, 1983. Economics of PCB Disposal:
A summary of report findings. The
Hazardous Waste Consultant, 11(2):
4.11, 1984. Est. PCB disposal costs
are in the range of $40-150/Ton.
California Air Resources Board, 1983.
Air Pollution Impacts of Hazardous
Waste Incineration: A summary of report
fi ndi ngs. The Hazardous Waste Con-
sultant, 11(3): 2.13, 1984.Est.
incineration costs are in the range
of $37-395/Ton.
243
-------
UPDATE ON CALIFORNIA PROGRAM TO RESTRICT
HAZARDOUS WASTE LAND DISPOSAL
Jan Radimsky, P.E.
Division of Toxic Substances Control
California Department of Health Services
Sacramento, California 95814
ABSTRACT
In 1983, California adopted regulations restricting land disposal of certain hazardous
wastes in order to reverse its hazardous waste management's reliance on land disposal
and to stimulate development of technological alternatives to land disposal.
Realizing the real high "long-term" costs of land disposal, demonstrated by the multi-
million cleanup costs estimates for the State's Superfund sites, California government
sought a way to stimulate development of the needed alternative technology facilities
by developing regulations with a definite schedule of phasing out of land disposal of
specific hazardous wastes of concern. Such a schedule guarantees those who develop
treatment facilities for the restricted wastes that they will not be competing with
low-cost land disposal because the land disposal alternative will not be available to
generators of such wastes.
Implementation of these regulations during the last two years has had positive effects
on hazardous waste management in California. Several issues have arisen affecting moni-
toring of compliance with current and implementation of future land disposal restric-
tions in California.
INTRODUCTION
In 1983, California adopted regulations
restricting land disposal of certain
hazardous wastes in order to reduce
its reliance on land disposal and to
stimulate development of technological
alternatives to land disposal. During
explorations of status of alternative
technologies, it became apparent that
a great number of effective treatment,
destruction, or recycling technologies
were available. However, they were
not utilized in California because of
availability of "cheap" land disposal.
Commercial disposal facilities' opera-
tors could not afford to make the
investment necessary to develop and
operate alternative technology facili-
ties and be competitive with low costs
of land disposal.
PURPOSE
Realizing the high "long-term" costs
of land disposal, demonstrated by the
multimillion dollar cleanup cost esti-
mates for the State's Superfund sites,
California government sought a way to
stimulate development of needed alter-
native technology facilities by develop-
ing regulations with a definite schedule
of phasing out of land disposal of spe-
cific hazardous wastes. .Such a schedule
guarantees those who develop treatment
facilities for the restricted wastes
that they will not be. competing with
low-cost land disposal, because the
land disposal alternative will not be
available to generators of such wastes.
This "phased" schedule reflects need
to allow time 'for designing, siting,
and permitting of alternative tech-
nology facilities.
244
-------
Major features of the California land
disposal restrictions are:
1. Land disposal restrictions apply
to all forms of land disposal,
including, but not limited to,
landfill, surface impoundment,
waste piles, deep well injection,
land spreading, and coburial with
municipal garbage.
2. Restrictions apply to five cate-
gories of wastes (so-called
"restricted wastes"), four of
which include only the liquid
phase, and for each component,
there is a threshold concentra-
tion (see Table I).
3. Categorically exempted are the
following wastes:
a. Injected drilling fluids and
produced wastes from production
of natural gas or crude oil.
b. Mining overburden.
c. Contaminated soil from site
cleanup pursuant to the Depart-
ment's app rova1.
4. Variances from land disposal restric-
tions can only be made for a specific
waste stream or site-limited land dis-
posal method. Emergency variances can
be granted if treatment or recycling
facilities have unplanned shutdown.
5. Restrictions are being implemented
according to a schedule of tentative
dates included in the regulations
(see Table II). Dates for restric-
tions are subject to confirmation by
the Department and become effective
only when this agency determines that
available treatment and recycling
capacity is available in the State
of California.
California's approach relies on the ini-
tiative and cooperation of the hazardous
waste processing industry to develop the
treatment and recycling facilities needed
for implementation of this land disposal
restriction program.
TABLE I: CALIFORNIA RESTRICTED HAZARDOUS WASTES
a. Liquid hazardous wastes containing free cyanides at concentrations greater
than or equal to 1,000 mg/1.
b. Liquid hazardous wastes containing the following dissolved metals (or ele-
ments) or compounds of these metals (or elements) at concentrations greater
than or equal to those specified below:
Arsenic and/or compounds (as As)
Cadmium and/or compounds (as Cd)
Chromium (VI) and/or compounds (as Cr VI)
Lead and/or compounds (as Pb)
Mercury and/or compounds (as Hg)
Nickel and/or compounds (as Ni)
Selenium and/or compounds (as Se)
Thallium and/or compounds (as Th)
500 mg/1
100 mg/1
500 mg/1
500 mg/1
20 mg/1
134 mg/1
100 mg/1
130 mg/1
c. Liquid hazardous wastes having a pH less than or equal to two (2.0).
d. Liquid hazardous wastes containing polychlorinated biphenyls at concentra-
tions greater than or equal to 50 mg/1.
e. Hazardous wastes containing halogenated organic compounds in total concen-
trations greater than or equal to 1,000 mg/kg.
245
-------
TABLE II: CALIFORNIA LAND DISPOSAL RESTRICTIONS SCHEDULE
June 1, 1983
January 1, 1984
January 1, 1985
July 1, 1985
Cyanide Wastes
Toxic Metals
Acid Wastes
PCB Wastes
Halogenated Organic
Waste Liquids
Halogenated Organic
Waste Sludges and Solids
EXPERIENCE WITH IMPLEMENTATION OF
INDIVIDUAL RESTRICTIONS
Cyanide Waste Restriction
The assessment of "restricted" cyanide
waste generation and availability of
treatment and recycling capacity con-
ducted in 1983 produced the following
information:
• Less than 7,000 tons of restricted
cyanide wastes were generated in
California.
• About 2,000 tons of these wastes
were already treated, mos.tly at
on-site facilities.
• There was no commercial recycling
capacity for the "restricted? cya-
nide wastes.
• Less than 5,000 tons of "restricted"
cyanide wastes were disposed of at
off-site land disposal facilities.
• Restriction of cyanide wastes led
to construction of three new facil-
ities capable of treating their
wastes and to increased utiliza-
tion of the only off-site cyanide
waste treatment facility in exis-
tence prior to the restriction
implementation.
Restriction for cyanide wastes was imple-
mented as originally scheduled in the
regulations.
A year later, in June 1984, a review of
hazardous waste manifests and on-site
disposal reports was conducted to deter-
mine the effectiveness of the land dis-
posal restrictions. The following were
the findings:
• Almost 100 percent reduction of
land disposal of restricted cyanide
wastes was achieved.
• A lesser amount of restricted cya-
nide waste was being generated
(approximately 3,000 tons per year
prior to the restrictions and 2,600
tons per year after the restriction)
in spite of improvement in the
State's economy. Also generation
of nonrestricted cyanide wastes
increased by 42.8 percent.
• There was a 79 percent increase in
wastes going to treatment facilities.
• Only 5 companies were responsible
for over 50 percent of the prere-
striction volumes of restricted
wastes and 60 percent of the post-
restriction volumes.
Because of the large number of variables,
it is difficult to make any precise con-
clusion as to what the impact and effec-
' tiveness of the land disposal restrictions
were without detailed evaluation of the
•generators and treatment at facilities
involved. Among the variables which
affect such evaluations are: unprecise
reporting on hazardous waste manifests,
246
-------
changes in generated volume dtie to' the
status of the economy, individual company
production changes, on-site treatment,and
waste reduction efforts of generators'."
Heavy Metal and Acid Waste Restriction
The land disposal restriction for heavy
metal wastes was also implemented as
scheduled in the regulations. This was
made possible primarily due to Califor-
nia's definition of land disposal, which
excludes surface impoundments used for
treatment, as long as the residue is
rendered nonhazardous within a year or,
if it remains "hazardous", it is removed
from the surface impoundment within a
year.
This practice accounts for more than 50
percent of the restricted acid and heavy
metal waste treatment, with the remain-
der, being neutralized and/or precipi-
tated in the tanks.
Solid residue is not subject to the land
disposal restrictions, and it is being
landfilled. Any variation in the gene-
ration of these wastes can easily be
accommodated by the large capacity 'of
existing in surface impoundments.
Liquid PCS-Containing Waste Restriction
Implementation of this portion of land
disposal restriction program did not
have any impact on management of liquid
PCB-containing wastes. This is because
of the Toxic Substances Control Act,
which already bans land disposal of liq-
uid PCB-containing wastes with concen-
trations over 500 ppm, and allows liquids
with 50 to 500 ppm of PCB concentrations
only with specially EPA-approved "secure
landfills" which do not exist in Califor-
nia. Implementation of this restriction
in California precludes the possibility
of establishment of such a landfill in
the future.
No variance requests were submitted in
connection with the first two land dis-
posal restrictions because of the excess
capacity available and simplicity of
treatment.
Liquid Halogenated Organic Waste
Restrictions (See Tables III and IV)
In implementation of this restriction,
the Department has run into the inher-
ent weakness of the California restric-
tions program, its dependence on indus-
try's cooperation in development of
needed treatment and recycling facil-
ities . As the treatment most generally
applicable to this category — incin-
eration is very expensive and has high
up-front capital costs -- industry has
been reluctant to build new facilities.
There is a risk that generators may not
send their wastes to be incinerated in
California, but will opt to send these
wastes to land disposal facilities in
other states which do not have to abide
by California land disposal restrictions.
This fact, combined with difficulties
in siting and permiting of incinerators
in California, resulted in insufficient
increase of thermal treatment (wet air
oxidation and incineration) capacity
due to the land disposal restrictions.
It is anticipated that several projects
currently in planning will move ahead
with the prospect of a national land
disposal restriction program as well
as the demand for disposal without
future liability potential increases.
A preland disposal restriction evalua-
tion determined that although there
is capacity for thermal treatment and
recycling of 36,200 tons, only 22,200
tons are thermally treated or recycled.
This is because the existing thermal
treatment facilities have stringent
limits on wastes to be treated such as
minimum heat value, maximum chlorine
concentration, maximum or minimum
organic content, etc. Without building
new thermal treatment facilities which
could accept a broad range of wastes,
about 23,000 tons cannot be recycled or
thermally treated. Over 18,500 tons of
these wastes are being generated by a
single company, which requested and
received a variance from land disposal
restrictions for a period of 5 years as
long as the low-concentration, nonvola-
tile halogenated organic-containing
wastes are disposed of to a double-lined,
leachate collection-equipped pond and
the company pursues process and waste
247
-------
disposal modification to eliminate gene-
ration of restricted liquid waste alto-
gether. Remaining nonrecyclable or
incinerable wastes will be allowed to
be solidified.
Solid Halogenated Waste and Lab Packs
Land Disposal Restriction (See Tables
III and IV)
The restriction of solid halogenated
organic wastes was tentatively sched-
uled for July 1, 1985. In the absence of
Table III. Management and Quantities of Hazardous Wastes
Containing Halogenated Organic Compounds in California (1983)
Type of Waste
Solvents
Still Bottoms,
Organic Liquids,
and Sludges
Dry-Cleaning Wastes
Waste Waters
Solids and Lab Packs
Total
Recycled or
Incinerated
(TPY)
21,800
500
„
300
22,600
Land
Disposal
(TPY)
2,400
2,800
3,800
28,400 (IS.SOO)-/
4,100
41,500
Total
Quantity
(TPY)
24,200
3,300
3,800
28,400
4,400
64,100
Table IV. Availability of Treatment for Recycling
Capacity for Hazardous Wastes Containing
Halogenated Organic Compounds in California
(1985)
Waste Description
Solvents
Still Bottoms,
Organic Liquids,
and Sludges
Dry-Cleaning Wastes
Waste Waters
Total Liquids
Solids, Lab Packs
Total Volume
Total
Quantity
(TPY)
24,200
3,300
3,800
28,400
59,700
4,400
64,100
Not Recyclable
or Incinerable
(TPY)
800
1,800
-
20,900 (18,50(
23,500
4,100
27,600
Recycling
Incineration
Capacity
Available
(TPY)
23,400
1,500
3,800
))-/ 7,500
36,200
0(300)-/
36,200
I/ Incinerated out of state.
2/ 18,500 TPY were generated by one company.
248
-------
art o££-si_te. commercial rotary kiln in
California, this restriction will not
be implemented this year. the' Depart-
ment is considering postponing this
deadline to coincide with ,the date .indi-
cated in the Resource Conservation and
Recovery Act Amendments of 1984. Even
that deadline appears difficult to meet
as the siting and permitting of needed
incinerators may require more time.
PROBLEMS ENCOUNTERED AND ISSUES TO BE
RESOLVED
Major problems which were encountered
during the implementation of the above
land disposal restrictions include:
1. Obtaining of good detailed data on
volume and composition of wastes is
difficult as reporting on haz-
ardous waste manifests is not
always detailed enough and often
plagued by errors.
Restricted wastes can be listed
under several waste categories
(California's waste categories
are not identical to EPA's) and,
therefore, manual review of all
manifests was necessary to obtain
needed information. Much more
detailed reporting is necessary
in the future to enable us to eval-
uate treatability . and recyclabil-
ity of specific wastes.
Continuous technical review of data
reported on manifests and in annual
(biennial) reports by generators'
disposal, treatment, and recycling
facilities' operators is needed.
As we strive to maximize waste
reduction, recycling and treatment
of hazardous wastes annual reports
will become an increasingly impor-
tant source of information. The
format of annual reports needs to
be expanded to sufficiently provide
detailed description of waste gener-
ation, composition, and information
on the management practices which
will enable us to determine recycla-
bility, treatability, and waste
reduction potential of each, waste
stream. At this time, the avail-
able information on existing waste
streams obtained from manifests is .
not adequate for that purpose.
Manifests have their place in that
they help to track wastes from
"cradle to grave", but should not
and cannot be the sole source of
information on hazardous waste
management.
California's ban depends on the
waste disposal industry's cooper-
ation and initiative to develop
new needed treatment and recycling
facilities. Because of the indus-
try' s reluctance to build a rotary
kiln incinerator, the last ban is
not implementable for either solid
halogenated wastes or lab packs.
Lack of a nationwide land disposal
restrictions program was probably
the primary reason for industry's
hesitance to develop incineration
facilities in California because
of the anticipated exodus of wastes
to neighboring states if the incin-
eration facilities were built and
restriction implemented in Califor-
nia. Recent RCRA changes and their
implementation should assist Cali-
fornia in implementation of future
land' disposal restrictions as long
as the national restrictions will
apply to underground injection of
hazardous wastes. If underground
injection is considered acceptable
for hazardous wastes, there may be
little incentive for building treat-
ment and recycling facilities for
hazardous wastes.
Future land disposal restrictions
in California as well as in the
United States will probably be
developed in the legislative arena.
Examples of such legislation are '
the recent RCRA amendments, as well
as the recent California legisla-
tive bill, AB 3566 (Katz), which
bans use of surface impoundment
for storage and treatment of
"restricted" wastes after July 1,
1985. This date may be postponed
due to the "shortness" of notice
to the dischargers (on October 1,
1984) who will not be able to com-
ply so quickly and will need addi-
tional time to build tank storage
and treatment facilities.
249
-------
FUTURE OF THE PROGRAM
At the present time, the California
hazardous waste management program
is being expanded to include any strat-
egy, which would ultimately result in
reduction of volume of hazardous waste
going to land disposal. These strate-
gies include economic and regulatory
incentives for waste reduction, waste
reduction technical assistance, waste
recycling and reuse support, as well
as restrictions on land disposal of spe-
cific wastes.
The emphasis of the California hazard-
ous waste management program will be on
reduction of hazardous waste generation
which is the ultimate waste management
solution, as it decreases the need for
transportation, treatment, recycling,
and disposal facilities, all of which
have some negative environmental and
public health impact potential, as well
as high costs and potential for long-
term liability for industry.
250
-------
THE THERMAL DECOMPOSITION CHARACTERISTICS OF
A SIMPLE ORGANIC MIXTURE
John L. Graham, Douglas .L. Hall* and Barry Dellinger
University of Dayton Research Institute
Dayton, Ohio 45469
ABSTRACT
Through previous efforts, the University of Dayton's Environmental Sciences
Laboratory has gathered extensive data on the gas phase thermal decomposition of pure
organic compounds. In this report, the thermal decomposition of a simple mixture is
examined. Specifically, the effect of oxygen concentration on the thermal stability of
the components and the formation of thermal reaction products are examined. Also, the
thermal stability of the components in the mi.xture are compared with their stability as
pure compounds.
The hazardous waste mixture consisted of carbontetrachloride, monochlorobenzene,
l,l,2-trichloro-l,2,2-triflouroethane (Freon 113), trichloroethylene, and toluene.
Thermal decomposition studies were conducted in atmospheres in which oxygen was in excess,
stoichiometric, and absent (absolute pyrolysis) with respect to complete combustion. For
comparison, the components were run as pure compounds in an atmosphere with stoichiometric
oxygen available.
Results indicate that the order of stability of the five components was strongly
effected by oxygen concentration. Oxygen concentration had a pronounced influence on the
thermal stability of monochlorobenzene, toluene, and trichloroethylene in the mixture,
but no influence on the stability of the Freon 113 or carbontetrachloride. Furthermore,
with the exception of Freon 113, the thermal stability of each component in the mixture
was less than its stability as a pure compound. The stability of Freon 113 was identical
as a pure compound and in the mixture.
It was also found that oxygen concentration had a significant effect on .the produc-
tion of thermal reaction products. In general, the numbe'r and complexity of thermal
reaction products.increased with decreasing oxygen concentration. In all cases, products
ranged from simple chlorinated aliphatics to complex polynuclear ardmatics, with the
majority being chlorinated aromatics.
251
-------
VOST APPLICATIONS AT THE USEPA
COMBUSTION RESEARCH FACILITY
by
Robert W. Ross, II, F. C. Whitmore, R. H. Vocque,
T. H. Backhouse and B. M. Cottlngham
Versar, Inc.
P. 0. Box 1838
Pine Bluff, Arkansas 71613
and
Richard A. Carnes
Environmental Scientist
USEPA Combustion Research Facility
Jefferson, Arkansas 72079
ABSTRACT
The volatile organic sampling train (VOST) has been used to collect stack samples at
the EPA's Combustion Research Facility (CRF) in Jefferson, Arkansas for the past year.
During this time strengths and weaknesses of this sampling and analysis technique have
become apparent. Among the advantages are ease of operation of commercially available
equipment, short time required to take a sample, and rapid analytical turn around time -
results are typically available from the on-site laboratories 1 1/2 hours after the
sample is collected. The disadvantages of this technique include the time consuming
nature of sample tube preparation and blanking, blockages in the flow path of the
analytical thermal desorption unit, and high breakage rate of glass sample tubes of the
I/I design. Solutions to these problems are presented along with extensive spike and
recovery data collected during validation of innovative modifications to the basic VOST
methodology.
INTRODUCTION
The research program at the
Combustion Research Facility (CRF) has,
to date, been largely concerned with the
study of the incineration of several
Environmental Protection Agency (EPA)
soups in a pilot scale, rotary kiln,
fired afterburner incinerator. Among
other non-conventional attributes, this
system is provided with sampling ports
in the kiln transfer duct which conducts
the kiln exhaust gases to the after-
burner, and in the afterburner transfer
duct carrying the combustion gases to
the pollution control system. The Soups
contain compounds of a variety of
boiling points requiring sampling with a
variety of procedures including the
volatile organic sampling train (VOST).
In many ways the operation at the
CRF is very different from the normal
procedures in that immediately after the
VOST sample has been taken, it is
removed from the train, capped and
delivered to the laboratory for
analysis. This unusual manner of
operation has resulted in a number of
variations from the usual manner of
using the VOST system and it is these'
variations that will be the subject of
252
-------
this paper. In addition, the use of the
VOST in a water saturated gas stream
(the exit gases from the scrubber are
saturated at 81°C) introduces peculiar
analytical problems which are also
treated below. The nature of these
problems and the solutions introduced at
the CRF are seen in Table 1.
The dry gas meters have been
calibrated on site and showed a low
and varible average bias of 7% to
9%. Recalibration using a wet test
meter shows a consistently low
average bias of 5%.
Thermal Desorption System - NuTech
TABLE 1. MODIFICATIONS TO VOST PROTOCOL*2)
PROBLEM
SOLUTION
PROBLEM
SOLUTION
PROBLEM
SOLUTION
The time consuming nature of sample tube preparation and
blanking especially when many samples will be taken in a
given experiment.
Tenax® is aspirated into the tube, weighed, and thermal
conditioning takes place on an easily constructed 24-port
manifold located in a forced draft oven. Cleaning is
accomplished by steam stripping requiring approximately
four hours.
The need to transfer large volumes of water/steam
(derived from the saturated combustion gases) during
analytical thermal desorption of sample tubes leads to
low and variable Principle Organic Hazardous Constituent
(POHC) recoveries.
Bulk of water entrained in the tubes is transferred
during a preheating step without purge gas flow.
Breakage of glass sample tubes during analysis and field
sampling may easily exceed 30%.
Stainless steel (S.S.) tubes of several designs have
shown good POHC recoveries in extensive trials.
It is recognized that the solutions
listed in Table 1 represent deviations
from and/or expansions of the most
recently enunciated*2^ VOST sampling
and analysis protocol. To support and
validate these changes extensive QC data
will be given when these and other
elements of the CRF investigations with
VOST techniques are discussed below.
APPARATUS
The VOST sampling and analysis
equipment is listed in functional order
below, along with comments identifying
specific problems encountered in its use.
Sampling System-NuTech Model 280B -
Model 320 - This is not the Model
320 probably familiar to most
current NuTech users: (1) it has no
provision for cryo-trapping; (2) it
can desorb only two sample tubes
(I/I design) at a time; (3) tubes
are no longer sealed into the system
with duck-bill valves, but connected
with SWAGELOK® reducers,
ceramic-filled TEFLON® ferrules
and stainless steel capillary
tubing; (4) this tubing looses its
flexibility after a period of
service and becomes a source
of glass sample tube breakage; (5)
the thermal desorption chamber
itself is of improved design, more
accessible and more convenient to
use and (6) the Valco® 6-port
valve and associated tubing are of
253
-------
narrow bore, keeping instrument dead
volume to a minimum but causing
problems in wet sample processing.
Purge and Trap System - Tekmar Model
LSC-1 - This equipment, familiar to
many in the analytical community,
has given the dependable performance
that users have come to expect.
Gas Chromatograph - Hewlett Packard
Model 5880A - This system,
configured for packed column* 5) use
with FID, has given predictably
dependable performance.
VOST Sorbent Cartridge (sample tube)
Conditioning Equipment - This
apparatus was constructed on site
from materials, largely, on hand. A
twenty four-port manifold was
assembled from Swagelok®, 1/4 inch
straight-run tees and elbows of #316
stainless steel. Fittings as well
as connecting lengths of 1/4 inch
#302 seamless stainless steel tubing
wore sonicated through three changes
of 1:1, acetone: methanol (Burdick
and Jackson) and oven dried before
assembly. The manifold
configuration is such that it
supports VOST tubes at a 45° angle
above the horizontal and was
installed in a forced draft oven
capable of maintaining 250°C. The
oven itself is placed in a forced
draft laboratory hood for safety
reasons. The manifold is supplied
with UPC grade nitrogen which
receives final polish through a bed
of activated carbon and silica gel.
OVERVIEW OF VOST QA/QC
As staff skills in VOST techniques
have grown, the QA/QC program has matur-
ed in the direction of more realistic
mimicry of incinerator operation and
sampling conditions. While the work of
many investigators has been valuable,
only two <3,4) are cite
-------
TABLE 2. COMPARISON OF P&T AND TDS ANALYSES OF IDENTICAL STANDARD SOLUTIONS
(values in yg)
P&T
TDS
P&T
TDS
P&T
*Value exaggerated by high and variable blank
TDS
Freon® 113
Carbon tetrachloride
Trichloroethylene
Chlorobenzene
3.00
3.00
3.00
3.00
3.01
3.05
3.04
3.03
0.300
0.300
0.300
0.300
0.282
0.327
0.279
0.274
0.0300
0.0300
0.0300
0.0300
0.0370
0.0906*
0.0353
0.0354
TABLE 3. COMPARISON OF P&T AND TDS SPIKES PLACED AT
SEVERAL POINTS IN A PAIR OF VOST TUBES
(values in tig)
Freon® 113
Carbon tetrachloride
Trichloroethylene
Chlorobenzene
P&T
3.00
3.00
3.00
3.00
TDS
tube #2
exit
3.04
3.07
3.13
3.12
TDS
tube #2
entry
3.01
3.05
3.04
3.03
TDS
tube #1
exit
3.00
3.00
3.01
2.97
TDS
tube #1
entry
-0-
0.023
-0-
-0-
standard solution. It is of interest to
note that TDS and P&T data are virtually
identical for all locations except tube
#l-entry. POHC recovery at this
location (the charcoal portion of the
Tenax®/charcoal tube) is virtually
zero. This experiment has been
repeated several times with the same
results, and no explanation can be given.
Phase 3 - Flash Evaporation Spikes
The current VOST protocol^)
requires the preparation and anlysis of
standards generated by the flash
evaporation technique. Suffice it to
say that these exercises have limited
relevance as a diagnostic procedure for
VOST sampling and analysis validity.
The results of CRF performance are
displayed in Table 4.
Phase 4 - VOST Spike and Recovery by
Incinerator Mimicry
In an effort to lend more relevance
to VOST QA/QC, a practice has been
initiated at the CRF. Periodically
several pairs of VOST tubes are
inoculated with various amounts of POHC
and products of incomplete combustion
(PIC) of current interest. In pairs,
these tubes are installed in a full
dress sampling train and 20 dry standard
liters (dsl) of steam-laden ambient air
are drawn through them from a flask of
boiling OFW. Routine analysis of these
tubes allows calculation of spike and
recovery data (see Table 5). This QC
device may represent a practical
compromise between the somewhat sterile
approach of the protocol*2^ and the
quite complex apparatus used*3) by
TABLE 4. RESULTS OF ANALYSIS OF FLASH EVAPORATION STANDARDS
% Recovery @
Spike = 3.0ug
% Recovery @
spike = 0.30ug
% Recovery @
spike = O.OSOvg
Methylene chloride
Chlororform
Freon® 113
Carbon tetrachloride
91
91
101
86
62
87
86
93
95
106
102
68 (continued)
255
-------
TABLE 4. (continued)
% Recovery @
Spike = 3.0pg
% Recovery @
spike = O.SOug
% Recovery @
spike = O.OSOyg
Trl chloroethylene
Benzene
Toluene
Chlorobenzene
92
93
93
93
89
90
109
91
95
97
114
104
TABLE 5. ANALYTICAL RESULTS FOR INCINERATOR MIMICRY SPIKED SAMPLES
% Recovery @
spike = 30.OUR
% Recovery @
spike = 3.0uK
Chloroform
Freon® 113
Carbon tefcrachloride
Trichloroethylene
Benzene
Toluene
Chlorobenzene
104
72
97
105
105
106
101
115
65
97
100
101
102
94
some other workers.
Phase 5 - Rapid Packing and Blanking of
Sorbent Cartridges
Sampling tubes of the I/I design may
be rapidly and uniformly packed as
follows: (1) a glass wool plug is placed
in one end of the tube, (2) this end is
connected to a water aspirator while the
other end is placed in a container of
Tenax GC and tapped gently; (3) when the
tube is filled, the resin is held in
place by a second plug of glass wool.
(5) tubes are cooled with nitrogen flow
and put into immediate service or sealed
for storage. It should be pointed out
that the compounds listed above are the
major contaminants routinely found, and
that scant other material is found using
the analytical method of reference (5).
Also, glass wool, glass sampling tubes,
Tenax® GC, charcoal, Swagelok®
fittings and Teflon® ferrules are
given no pre-treatment but are used as
received from their various suppliers.
Phase 6 - Analysis of Wet VOST Samples
Contaminant levels for caroon
tetrachloride, benzene and toluene may
be reduced to values <. 0.005 yg per
pair of tubes as follows: (1) packed
tubes are washed by percolation with
25-30 ml of OFW and allowed to soak 15
min. or more, (2) tubes are then
connected to the manifold in the cold
oven with nitrogen set at ~
30 ml/min/tube or 40 psig when all 24
tubes are in place; (3) oven is set for
180*C and turned on; (4) temperature
increases rapidly to set point in ~
one hr. where it is held for > 3 hrs.;
A pair of packed sampling tubes may
easily contain 6 ml. of entrained
condensate after use in stack sampling
at the CRF. The NuTech Thermal
Desorption System in use at the CRF
cannot accomodate this volume of liquid
in a prompt or reproducible fashion due
to the very narrow bore valve and tubing
of which it is constructed. If thermal
desorption with purge gas flow is
attempted, the results are low, erratic,
sometimes zero recovery of POHC and
PIC. A simple remedy for this problem
is as follows: (1) the bulk, of .the
256
-------
en.tcai.tied condensate is removed by , :
several sharp bursts of vacuum
charcoal filtered air is used)"-over a ;
period one min.; (2) the remaining
liquid is vaporized during a 15 min.
pre-heating period during which the
tubes are connected in the thermal
desorption chamber but no purge gas is
flowing, (3) during pre-heating much of
the liquid transfers, as steam, to the
sparger where it condenses again and
thus carries out no sparing; and (4)
after pre-heating, the analysis is
conducted normally, starting with the
desorb purge flow cycle. This seemingly
radical departure from protocol has been
validated by saturating tubes with
water, spiking them with known
quantities of authentic POHC and PIC,
removing the water pneumatically/
thermally and conducting their analysis
to measure recovery of the spikes.
Table 6 displays the average recoveries
Swagelok® design - #302 seamless
S.S., 1/4" o.d. x 5.5", usable
internal volume ~ 1.8 cm3
containing ~ 0.30 gm Tenax® GG.
Welded design - #302 seamless S.S.
5/8" o.d.x 3.0", fabricated by
welding 5/8" - 1/4" reducers on
either end of S.S. tube, usable
internal volume ~ 11.5 cm3
containing ~ 1.8 gm Tenax® GC.
The results listed in Table 7 were
encouraging to the point that a third
design, a S.S. replica of the I/I glass
design, was fabricated locally and field
tested at the CRF.
This third design was constructed as
follows: a 2 3/4" length of 5/8" o.d.
#302 seamless S.S. tubing was provided
TABLE 6. VALIDATION OF PROPOSED PROTOCOL MODIFICATION FOR ANALYSIS
OF WET VOST SAMPLES
% Recovery
spike = SO.Opg
% Recovery
@ spike = 3.0pg
% Recovery
@ spike = O.SOug
Methylene chloride
Chloroform
Freon ® 113
Carbon tetrachloride
Trichloroethylene
Benzene
Toluene
Chlorobenzene
98.0
94.3
99.8
107.2
102.1
101.9
102.5
99.8
99.5
100.2
103.4
104.1
103.2
102.8
99.9
92.4
138.5
102.7
108.5
92.1
104.3
107.2
102.1
97.7
of each of eight compounds, determined
in triplicate, at each of three spiking
levels. The analytical system was
calibrated using purge & trap
methodology as the performance standard.
Phase 7- Excessive Breakage of VOST Tubes
In field sampling applications,
workers generally concede that a
breakage rate of 30% is not unusual for
glass VOST tubes of the I/I design. CRF
experience parallels that of other
workers and the cost here in lost
analytical data is considerable. As a
result of these considerations, tubes of
two different designs were
assembled/fabricated locally and
evaluated in spike and recovery studies
(see Table 7).
with female threads at each end; end
fittings were machined from rod stock of
the same material with matching male
threads, wrench flats and a taper to 1/4"
o.d. tube size; end fittings were
installed on the 5/8" tube with Teflon®
tape seals; the assembled unit was
demonstrated leak free @ 80 psig helium;
it was weighed, charged with TENAX GC®,
weighed again and inscribed with an
identifying number.
It is important to note that each of
the prior experimental designs suffered
from serious defects which have been
eliminated in the most recent one. The
Swagelok® design could only contain ~
20% the amount of resin required by
protocol^2*, and exhibited high
257
-------
TABLE 7. RECOVERY DATA FOR METAL VOST TUBES OF TWO DESIGNS
% Recovery @ S.Oyg
Swagelok® Welded
% Recovery @ O.SOjjg
Swagelok® Welded
Methylene chloride
Chloroform
Freon®
Carbon tetrachloride
Trichloroethylene
Benzene
Toluene
Chlorobenzene
105
105
106
105
105
105
102
98
100
111
108
102
110
114
107
100
122
122
122
126
120
121
122
120
96
88
90
46
78
91
76
79
pressure drops when tested in the
sampling train. The welded design
rusted badly at the butt seals and, in
fact, spit rusty water during thermal
conditioning.
After steam stripping, thermal
conditioning and demonstration of
analytical blank, these S.S. replica
tubes were evaluated in recovery studies
involving 19 compounds at 3 spike levels
(Table 8). These levels were S.Oyg
compound per pair of VOST tubes. In
addition, 1 ml of water was placed in
each tube of every pair and the
analytical system was calibrated using
P&T methodology. It is worth noting
that among the several chemical compound
classes included in these data, the
results are good. The abrupt threshold
effect shown .by acetonitrile is not
surprising in view of its miscibility
with water, while the high recoveries
shown by readily polymerizeable styrene
at least suggest that catalytic effects
TABLE 8. RECOVERY DATA FOR STAINLESS STEEL REPLICA VOST TUBES
Acetonitrile
Methylene chloride
Chloroform
1,2-Dichloroe thane
1,1, 1-Tr ichloroethane
Carbon Tetrachloride
Tr i chloroe thylene
Benzene
Tetrachloroethylene
Zso-octane
Toluene
Chlorobenzene
Ethylbenzene
Octane
Styrene
Ortho-xylene
1 , 3-Dichlorobenzene
1 , 4-Di Chlorobenzene
Nonane
% Recovery @
spike = S.Oyg
116
102
105
101
103
97
103
103
104
104
103
99
100
106
104
105
105
104
104
%Recovery @
spike = O.SOpg
110
106
107
104
104
92
104
105
104
105
104
94
88
105
88
99
82
73
104
% Recovery @
spike = 0.030pg
-0-
105
108
95
105
90
107
102
91
100
106
114
110
110
86
99
95
94
113
and O.SOjjg, and 0.03ug for each
exerted by the steel walls of these
258
-------
tubes may not be marked.
Additional evidence of a low order
of catalysis exerted by the walls, ,o.f the
S.S. replica VOST tubes may be drawn
from limited studies with additional
compounds at the spiking levels
indicated:
1,1-dlchloroethylene (0.20vig); trans-
1,2-dichloroethylene (0.60vig);
bromodichloromethane (0,60pg);
1,2-dichloropropane (0.20ug); trans-
1,3-dichloropropene (0.60pg); and
bromoform (0.60pg). Each of these
compounds was recovered at 88% or
better except bromoform whose recoveries
ranged from 42-113% with 71% as the
average.
Table 9 displays the results of a
tubes were prepared. Each set was made
up of one blank pair of tubes plus two
spiked pairs. Each spiked pair
contained 18 compounds at O.SOjjg per
compound per tube pair. On day #1, 20
dsl of steam-laden air were drawn
through every pair of tubes which were
then sealed, and the tube pairs in set 1
were analyzed. On day #2, the tubes of
set 2 were analyzed, while set 3 tubes
were analyzed on day #5.
SUMMARY AND CONCLUSIONS
VOST techniques are invaluable in
the characterization of volatile
emissions from hazardous waste
incinerators. Practitioners of these
techniques, the CRF among them, have
identified several nagging operational
TABLE 9. RECOVERY AND STABILITY DATA FOR S.S. REPLICA VOST TUBES
EVALUATED BY INCINERATOR MIMICRY
Acetonitrile
Methylene chloride
Chloroform
1,2-Dichloroethane
1,1, 1-Tr ichloroethane
Carbon Tetrachloride
Trichloroethylene
Benzene
Tetrachloroethylene
Iso-octane
Toluene
Chlorobenzene
Ethylbenzene
Octane
Styrene
Ortho-xylene
1 , 3-Dichlorobenzene
1 , 4-Dichlorobenzene
% Recovery @
*TET=0 hrs.
-0-
7
82
93
88
88
98
99
96
88
103
100
100
106
99
103
87
71
%Recovery @
TET=48 hrs.
4
55
90
96
93
90
97
98
93
104
106
99
98
112
101
101
93
79
% Recovery @
TET = 120 hrs
-0-
37
87
92
94
85
96
98
92
103
98
98
99
106
96
102
86
69
*TET = Total Elapsed Time Since Spiking
study undertaken to demonstrate recovery
by incinerator mimicry (see phase 4
section above) as well as compound
stability in S.S. replica VOST tubes.
This investigation was conducted as
follows. Initially, three sets of VOST
problems that are frequently
encountered. This report presents
validated solutions to some of these
problems in the hope that fewer VOST
pitfalls may lead to wider VOST usage,
and the harvesting of more and better
259
-------
data on volatile emissions.
It must be emphasized here that
none of the remedies suggested is
infallible. For example, S.S. replica
VOSX tubes possess certain unmistakable
advantages over their glass
counterparts. Nonethless, investigators
contemplating the use of such steel
tubes should demonstrate that compounds
of interest to them are well behaved in
these devices.
Appreciation is extended to Donald A.
Oberacker;(USEPA-Ci) for his thoughtful
review of this work. Every effort has
been made to incorporate his uniformly
helpful suggestions.
This research was conducted under
USEPA Contract No. 68-03-3128 by Versar,
Inc., which gratefully acknowledgers
this support.
REFERENCES
Chang, R., R. Carnes, and F.
Whitmore. Helium Tracer Measurements
at the USEPA Combustion Research
Facility. Hazardous Waste. June,
1985 (in press).
Hansen, E. M.. Protocol For The
Collection and Analysis of Volatile
POHC's Using VOST. a contractor's
final report prepared by Envirodyne
Engineers, Inc., St. Louis, Missouri
under EPA contract No. 68-02-3697,
Technical Directive 003, report
dated February, 1984.
Jungclaus, G. and Gorman, P..
Evaluation of a Volatile Organic
sampling Train (VOST). a
contractors draft final report
prepared by Midwest Research
Institute, Kansas City, Missouri
under EPA contract No. 68-01-5915,
draft report dated July 2, 1982.
Pellizari, E., et al.. Sampling of
Organic Compounds in the Presence of
Reactive Inorganic Gases with Tenax
GC, Anal. Ghent.. 1984, 56, 793-798.
Method #8010 found in EPA/SW-846,
Test Methods for Evaluating Solid
Waste - Physical/Chemical Methods.
second edition, 1983.
NOTICE AND ACKNOWLEDGEMENT
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 recommendation
for use.
- 559-111/20686
260
------- |