EPA 530-SW-87-014B
                                                  PB87-173837
           BACKGROUND INFORMATION  DOCUMENT  FOR THE
           DEVELOPMENT OF REGULATIONS  TO  CONTROL THE
           BURNING OF HAZARDOUS  WASTES IN BOILERS AND
           INDUSTRIAL FURNACES - VOLUME II,  INDUSTRIAL
           FURNACES
           Engineering-Science
           Fairfax, VA
           Jan 87

            •f Commerce

-------
                                   PB87-173837
  BACKGROUND INFORMATION DOCUMENT FOR THE DEVELOPMENT OF
REGULATIONS TO CONTROL THE BURNING OF HAZARDOUS WASTES  IN
             BOILERS AND INDUSTRIAL FURNACES
                        VOLUME II
                   INDUSTRIAL FURNACES
                       Submitted to

           U.S. Environmental Protection Agency
             Waste Treatment Branch  (WH-565A)
                    401 M Street, S.W.
                 Washington, D.C.  20460
                       January 1987
                      66036.00/80A-D
                   KHODvr.IV SI
                   NATIONAL  TECHNICAL
                   INFORMATION SERVICE
                      U.S. DfMsrwm of CONDUCE
                        SMMGf IELD, V*. 221(1
                       Submitted by
                   Engineering-Science
                      Two Flin'c Hill
                  10521 Rosehaven Street
                 Fairfax, Virginia  22030

-------
                      DISCLAIMER
Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.

-------
                                TABLE OF CONTENTS
SECTION 1

SECTION 2




SECTION 3
SECTION 4
SECTION 5
SUMMARY AND CONCLUSIONS                                  1-1

INTRODUCTION                                             2-1

  Objective                                              2-1
  Scope                                                  2-1

CHARACTERIZATION OF INDUSTRIAL FURNACES                  3-1.

  Introduction                                           3-1
  Portland Cement Kilns                                  3-1
  Lime Production                                        3-15
  Lightweight Aggregate Kilns                            3-16
  Blast Furnaces (Steel Industry)                        3-19
  Reverberatory Furnaces (Open Hearth Furnace -
    Steel Industry)                                      3-30
  Spent Acid Recovery Plants                             3-36
  Asphalt Concrete Plants                                3-42
  Characterization of Furnaces for Dispersion
    Modeling                                             3-47

DESTRUCTION AND REMOVAL EFFICIENCY OF HAZARDOUS
MATERIALS BY INDUSTRIAL FURNACES                         4-1

  Evaluation of Test Methods for Measuring
    Organic Emissions from Combustion Sources            4-1
  Evaluation of Test Burn Results for Cement,
    Lime, and LWA Kilns                                  4-13
  Evaluation of Test Burn Results for Asphalt
    Aggregate Kilns                                      4-34
  Evaluation of Test Burn Results for Blast Furnace      4-46
  System Modifications Necessary to Burn Hazardous
    Wastes in Ind-.strial Furnaces                        4-48

COST ELEMENTS FOR THE REGULATORY IMPACT ANALYSIS         5-1

  General                                                5-1
  Conventional Fuel Prices                               5-1
  Capital Costs                                          5-2
  Operating and Maintenance Costs                        5-17
                                        ii

-------
                                  SECTION 1

                           SUMMARY AND CONCLUSIONS
     One alternative disposal method for hazardous wastes in the United States
is to burn the wastes in industrial furnaces.  This report includes background
information for use in developing and evaluating regulatory alternatives and
in preparing a Regulatory Impact Analysis (RIA) of burning hazardous waste in
industrial furnaces.  Seven industrial furnaces were evaluated in depth; in-
cluded were cement  (wet and dry processes),  lime, lightweight aggregate kilns,
asphalt aggregate kilns, blast furnaces, spent acid recovery furnaces, and
open hearth steel furnaces.  The following characteristics make these fur-
naces likely candidates for disposal of hazardous wastes:

     o  Process operations require high temperatures (>1050°C)
     o  Combustion gases exhibit long residence times (2-10 seconds)
     o  Processes have the ability to burn low quality fuels
     o  Efficient particle removal equipment already exists on most kilns
     o  Alkaline raw meterials used in production (or downstream operations
        in the case of spent acid recovery furnaces) tends to remove acidic
        gases from combustion products before discharge to the atmosphere

     The owners and operators of furnaces used in the production of cement,
lii.e, and lightweight aggregate, and asphaltic concrete do not produce liquid
hazardous wastes.  Instead, these wastes are generated elsewhere and trans-
ported to the furnace locations for destruction by combustion.  Past waste
burning in blast furnace was similar in that the wastes were generated off-
site.  The composition of hazardous wastes which might be burned in furnaces
is highly variable depending on the generating source.   Variable parameters
include: 3tu content, ash or solids content, water content, toxic nsatals con-
tent, chlorine content, and POHCs content.

     In order to fire hazardous wastes efficiently in existing industrial
furnaces, necessary modifications will generally include:

     o  Installing waste storage and handling facilities
     o  Installing equipment to characterize the waste
     o  Installing equipment to pretreat the waste to improve its suitability
        for firing
     o  Installing or modifying the burner gun for firing the waste
     o  Modifying combustion controls
     o  Installing waste feed rate,  oxygen,  ani carbon monoxide monitors
     o  Modifying air pollution control equipment
                                     1-1

-------
     The number of furr.aces burning hazardous wastes nas not been established
definitively.  A recent survey found that 35 furnaces burn 66 million gallons
of waste per year.  The United States kiln population in these three industries
totals about 800 kilns made up of about 275 cement kilns. 4'iO lime kilns, and
about 75 lightweight aggregate kilns.  These kilns probably offer the greatest
potential for hazardous waste burning.

     Results from the trial burns of eleven industrial furnaces were evaluated
to determine their performance in burning hazardous wastes.  These furnaces
included:  three dry process cement kilns, two wet process cement kilns, one
lime kiln, one lightweight aggregate kiln, three asphalt aggregate kilns, one
blast furnace.  Testing procedures have not been standardized and quality as-
surance/quality control (QA/QC) procedures have not been established for test-
 ng of the destruction and removal efficiency (ORE) of hazardous wastes
 ombusted in furnaces.  A ORE of 99.99% for POHCs of hazardous wastes was
achieved on all of the tests evaluated in this study.  Of the furnaces evalu-
ated by EPA, particle emissions were increased significantly in one cement
kiln when the furnace was co-fired with hazardous waste.  The reason may have
been increased dust loading due to formation of salts or changes in dust
resistivity which affected the efficiency of the electrostatic precipitator
(ESP).  Metal emissions tended to be somewhat proportional to the concentration
in the waste feed.  Hydrochloric acid (HC1) emissions from the tests were
minimal even with high concentrations of chlorine in the waste feed.  Removal
is conjectured to take place in the furnace by reactions of chlorine with
alkaline materials which end up in the product or in dusts collected by the air
pollution control equipment.  The lightweight aggregate kiln was controlled
by a wet scrubber which removed HC1.  Since lightweight aggregate does not
react with chlorine like the clinker in cement or lime kilns, a scrubber
would likely be required on an aggregate kiln burning hazardous waste to
ensure that HC1 is removed.  PICs were not evident in the stack gas exhausts
from the seven tested kilns, reflecting positively on the combustion efficien-
cies in the kilns.

     Cost elements were provided for operating or retrofitting existing in-
dustrial furnaces.  Capital and operating and maintenance costs were provided
in 1982 dollars for those systems likely to be added at existing furnace
operations if hazardous wastes were to be burned in the furnaces.  This BID
includes cost elements for:

     o  Equipment to characterize the wastes
     o  Equipment to pretreat the wastes by blending, straining, or thermal
        heating
     o  New or modified burner guns to fire the waste
     o  Equipment for combustion controls.
     o  Equipment for monitoring waste feed rates, oxygen, or carbon monoxide
        levels
     o  Equipment for controlling particulate and gaseous stack emissions

     EPA developed cost elements for waste storage and handling facilities as
part of other regulatory development efforts.
                                     1-2

-------
                                  SECTION 2

                                 INTRODUCTION
     The Resource Conservation and Recovery Act (RCRA) is one of the most
important pieces of enabling legislation now being implemented by the U.S.
Environmental Protection Agency  (EPA).  It requires EPA to develop regula-
tions for the storage, handling, and disposal of hazardous wastes.  It also
requires that EPA encourage in its regulations the recycling of wastes.  On
May 19, 1980, EPA issued initial hazardous waste regulations which included
requirements for hazardous waste incinerators.  At that time, the burning
of hazardous wastes in boilers and industrial furnaces was not regulated be-
cause EPA lacked sufficient information to determine  the extent of the risk
to public health posed by such burning as well as the extent of the regula-
tory controls that would be necessary to address the  risks.  Since 1980, EPA
has researched the nature and extent of the practice  of burning hazardous
waste as fuel in boilers, and industrial furnaces.  This document presents
the results of research efforts  for industrial furnace.  A similar analysis
for industrial boilers has been  completed by EPA and  is presented in Volume
I of this document.  The risk assessment analysis is  presented in Volume III
of this document.
OBJECTIVE

     The objective of this project was to provide s li'ficient research and
investigation into the concept and practice of burning hazar-'. us wastes in
industrial furnaces such that EPA could develop and evaluate regulatory al-
ternatives.  The information and data collected by this investigation con-
stitutes a background information document (BID) for use by EPA in preparing
its regulatory impact analysis (RIA).
SCOPE

     The investigation was designed to characterize those industrial fur-
naces that have the potential to burn a hazardous waste as fuel.  An indus-
trial furnace is defined by 40 CFR 260.10 to mean any of the following en-
closed devices that are integral components of manufacturing processes and
that use controlled flame devices to accomplish recovery of materials or
energy:  cement kilns, lime kilns, aggregate kilns, phosphate kilns, coke
ovens, blast furnaces, smelting, melting and refining furnaces  (including
pyrometallurgical devices such as cupolas, reverberatory furnaces, sinter-
ing machines, roasters, foundry furnaces), titanium dioxide chloride process
                                     2-1

-------
oxidation reactors, mechane reforming furnaces, pulping liquor recovery fur-
naces, and spent acid recovery furnaces.  These furnaces aro typically high
temperature devices (above 1500°F exhaust temperatures) used in industrial
processes.

     In addition to these specific furnaces, 40 CFR 260.10 set forth criteria
for adding to the  listing of industrial furnaces.  The criteria are as follows:

     o  The design and use of the device is primarily to accomplish recovery
        of material products.

     o  The use or the device is to burn or reduce raw materials to make a
        material product.

     o  The use of the device is to burn or reduce secondary materials as
        effective substitutes for raw materials, in processes using raw mate-
        rials as principal feedstocks.

     o  The use of the device is to burn or reduce secondary materials as
        ingredients in an industrial process to make a material product.

     o  The use of the device in common industrial practice is to produce a
        material product.

     o  Other factors as appropriate.

     There are furnaces other than those listed in 40 CFK 260.10 that poten-
tially meet these criteria.  Glass melt furnaces, carbon black furnaces, and
acti/ated carbon retort kilns are examples of furnaces meeting the criterion
of reducing raw materials to produce a product.  Therefore, the universe of
industrial furnaces is larger than those listed in 40 CFK 260.10.  Additional
furnaces can oe added to the list of industrial furnaces under procedures
identified in 40 CFR 260.20.

     The quantity of hazardous waste currently being burned in industrial
furnaces is not accurately known.  In 1984, EPA conducted a mail question-
naire survey of industrial plants to estimate the quantity of waste being
burned.  Table 2.1 lists the quantity of waste burned in the industrial
furnaces covered by the survey and estimates the total quantity of waste
burned in all U.S. industrial furnaces using statistical scaling procedures.
A total of 51 million gallons per year of hazardous waste was projected to
be burned in industrial furnaces from the survey results.  This estimate is
known to be incorrect at this time since the burning of hazardous waste in
blast furnaces has recently been discontinued.  A survey conducted in April
1986 of the cement, lime, and lightweight aggregate industries found that 86
million gallons/year of hazardous waste was being burned in 35 kilns used in
these industries (Reference 1).

     Although limited, the available data indicate that appreciable quanti-
ties of hazardous waste is or has been burned in cement, lime, aggregate
kilns, blast furnaces, and spent acid recovery furnaces.  For this reason,
these furnaces have been selected for an in-depth analysis to provide infor-
mation needed to guide the evaluation of regulatory alternatives.  Because
                                     2-2

-------
                                                  TABLE 2.1

                              LISTING OF INDUSTRIAL FURNACES FROM WESTAT SURVEY
Standard
Indus*-..
Code (k)
3312
3312


3312
3341
Industry
Steel Products
Steel Products

Same Facility
Steel Products
Secondary
Smelting/
Refining
Device Type
Blast Furnace
Blast Furnace

Blast Furnace
Coke Ovens
Film Recovery
Incinerator
Waste Type
- Tar sludge
- Coke oven byproducts
- Recovered solvents
- Still bottoms
- Other
- Same
- Tar sludge
- Photo paper
- Photo & x-ray film
- Circuit boards
Quantity Reported
in Survey
(gal/yr)
208,000
5,861,729

3,907,819
8,163
35,052
(316,500
Ib)
Projected
National Quantity
(gal/yr)
395,000
9,379,000

6,253,000
15,500
3,502,000
2800
Precious Metals

Chemicals and    Fire Tube
Allied Products  Boiler
                                        - Photo fixing solutions
- Allyl chloride wastes
914,514
3,577,000
                         Recovery Unit
2819    SuIfuric Acid
        Manufacturing
                 Recovery
                 Furnace
  Same
  Heavy ends carbon
  tetrachloride
  Distillation bottoms

  Disulfide oils
  Allyl alcohols
  Acrylonitrile waste
  Other
                                                                1,470,115
607,505
  91 1,000

-------
                                             TABLE 2.1--Continned
Standard
Indust.
Code (H) industry
2819 Sulfuric Acid
Manufacturing
Same Facility
2600 Kraft Paper
K>
1
A
2631 Pa per board
Mill
2631 Paperboard
Mill
Device Type
Recovery/
Combustion
Furnace
Recove ry
Furnace
Recovery
Boiler

Recovery
Lime Kiln
Recove ry
Lime Kiln
Waste Type
- Petroleum spent acid
- Deteryent sulfona*:ion
- Sulfonation acid
- Etching acid
- Spent acids
- Waste solvents

- Waste solvents
- Waste solvents

Quantity Reported
in Survey
(gal/yr)
4,337,207
(36,200,000
Ib/yr)
(35,741,144
Ib/yr)
755

123,000
8,068,448

Projected
National Quantity
(gal/yr)
13,879,000

1 ,200

197,000
12,910,0^0

TOTAL
25,542,307
51 ,018,000

-------
of their prevalence and importance as a industrial combustion device, rever-
beratory furnaces are al»o analyzed here.  An open hearth furnace was selected
to represent reverberatory furnaces.

     The remainder of this document presents information on the selected fur-
nace deemed necessary for EPA to conduct its RIA of alternatives for regu.at-
ing the burning of hazardous waste in industrial furnaces.  Although the fur-
naces subject to regulation that are not discussed here may differ somewhat
from those selected, the information presented should be sufficiently appli-
cable to them to provide an adequate technical basis for making regulatory
decisions.  Section 3 of this volume characterizes the selected furnaces in
terms of their design, combustion methods, air pollution emissions and con-
trol tecauj-^ucc, a"H waste compatibility.  Model furnaces are described in
Section 3 to represent a "typical" unit of each type of furnace selected for
the in-depth analysis.  These models were developed for use in assessing the
health risks associated with burning hazardous waste in industrial furnaces
and for analyzing the economic impact of regulatory alternatives.  The risk
assessment results are discussed in Volume III to this BID and the eco-
nomic impact analysis is documented in the RIA.

     Section 4 describes *;he results of tests conducted to measure the per-
formance of industrial furnaces in burning hazardous wastes.  EPA funded
tests of cement kilns, lime kilns, aggregate kilns, and blast furnaces to
collect data for its development of regulations for industrial furnaces.
Included in the discussion are. the ORE, the POHCs, emissions of particulate
matter, HC1, PICs, metals and combustion gases.

     Some of the cost factors needed for the economic impact analysis are
presented in Section 5.  Some costs such as those associated with storage
facilities and permitting are not included because these are available from
other sources.  The cost items provided in  this document fall into three
categories: (1)  conventional fuel prices, (2) capital costs to modify the
furnace system to burn a hazardous waste, and (3) the major operating and
maintenance cost associated with burning a hazardous waste.
                                     2-5

-------
                                  REFERENCES

1.    Stohrer,  G.   Burning Hazardous Waste in Cement Kilns.   Report on a work-
     shop held *t Clean Site,  Inc., on April 30,  1986 in Washington,  D.C.
                                     2-6

-------
                                  SECTION 3

                   CHARACTERIZATION OF INDUSTRIAL FURNACES
INTRODUCTION

     This section presents the characterization of the industrial furnaces
chosen in Section 2 for an in-depth analysis, i.e., cement kilns, lime kilns,
lightweight aggregate kilns, blast, reverberatory, spent acid furnaces, and
asphalt aggregate kilns.  Each furnace will be described as to operation,
combustion, compatible hazardous waste fuels, and existing air pollution
control equipment.  The intent is to describe important aspect5 to their firing
hazardous waste fuels and to develop model furnaces for each of the selected
industrial furnaces.  These model furnaces will be used for the health risk
assessment documented in Volume III of the BID.
PORTLAND CEMENT KILNS

General

     The manufacture of Portland cement is one of the largest tonnage mineral
commodities in the United States.  Current (1982) United States annual produc-
tion capacity (Reference 35) was 86,607,000 tons of clinker.  This capacity
was owned by 46 companies, with the top 10 having 59.4% of the installed ca-
pacity.  These data and those which follow apply to clinker production (kiln
capacity).  Grinding capacity (finished cement production) data are very sim-
ilar but reflect the existence of plants which only grind and blend to ship
the finished product.  There are a total of 135 plants producing clinker in
274 kilns, located throughout 38 of the 50 states.  This dispersion reflects
the industry's attempts to balance sources of raw materials and energy with
the markets.  The largest plant, National Gypsum's at Alpena, Michigan, has
a clinker capacity of 2,450,000 tons per year (TPY).  The largest kiln in the
United States is Kaiser Cement's 5000 ton per day (TPD) unit at Permanente,
California.  The smallest plants, of which there are several, have clinker
capacities of less than 150 TPD (<50,000 TPY).

Process Description

     Portland cement consists of combinations of calcium, aluminum, iron ox-
ides, and gypsum formed from a variety of raw materials.  The basic chemical
compounds of Portland cement are calcium silicates;  primarily tricalcium sil-
icate (3CaO'Si02),  referred to in the industry as C3S.  Dicalcium silicate
(2CaO*Si02), tricalcium aluminate (3CaO*A120-,), and tetracalciumalamina-fer-
rite (4CaO*Al203*Fe203) are the other major components in order of abundance.


                                     3-1

-------
Calcium sulfate (CaSO^) and a variety of other compounds are added, usually
in final grinding and blending, to provide special properties but the funda-
mental compcsition remains the same.

     The 'jalcium source is most often limestone, though other materials such
as oyster shells and chalk are used.  The cement rock limestone found in the
Lehigh Valley of Pennsylvania is an example of a nearly perfectly balanced
stone *:or cement manufacture but usually clay, shale, iron ore, and other
materials must be added to provide the proper chemical formulation for the
kiln reactions and the final product.

     With a correct blend of raw materials, the mass is subjected to heat in
a kiln or other unit where a number of physical and chemical processes occur.
The temperar.ure of the solid mass is raised from ambient temperature to about
1500°C (2700°F).  The first occurrence is evaporation of free water at 100°C,
followed by continued beating to about 500°C where combined water from clay
is driven off.  At about 900°C, calcination, or CO2 evolution from carbonates,
begins leaving the oxides.  At this same temperature, other reactions begin
such as that between lime and clay dehydration products.  The material reaches
the point of incipient fusion at about 1250CC and cemer.titious product (clink-
er) begins to form with the clinker reaching a maximum temperature of about
1500°C, which is required to form the tricalcium silicate, a requirement for
quality cement.

     Although some of the intermediate chemical reactions are exothermic, the
overall chemical process requires the addition of about 1.8 x 106 Btu of heat
per ton of product (Reference 1).  This heat is supplied at the rate of 3 to
6 x 106 Btu/ton by the combustion gases from burning fossil fuel, primarily
coal.  Fuel use in cement kilns has followed the national trend of conversion
from gas and/or oil to coal.  Currently, 96% of the total installed capacity
uses coal as primary fuel compared to 30% in 1974 (Reference 2).

     Figure 3.1 (from Reference 1) portrays the basic components of a Port-
land cement plant.  The wet and dry processes differ in the use of wet and
dry grinding and in the kiln configuration.  The grinding operations are not
of much concern for this analysis, but the kiln configurations are of great
interest.  Wet process plants use "long" kilns (450-600 feet) to obtain the
necessary heat transfer for water evaporation.  Dry process plants can have
a long kiln, medium kiln, or short kiln (200 feet), depending on whether a
preheater -nd/or precalciner is used.  The problems with burning hazardous
waste in precalciner/preheater kilns are described in the dry process por-
tion of this section.

     The kilns are operated with solids flow counter-current to the combus-
tion gases, with the solids traveling down the slightly (3-8°) inclineu ro-
tating kiln.  The exhaust gases exit the kiln at temperatures of 450-700°C
depending on kiln type and energy conservation practices.  The gases typi-
cally contain from 6% to 30% of the solids feed rate as dust, 20% by volume
water vapor (wet kiln only) and up to 30% CO2/ along with 10-1000 pptn CO,
10-2000 ppm SO^, and 100-1500 ppm NOX.  The gases are transported to the
pollution control equipment with an induced draft (ID) fan.
                                     3-2

-------
ut
I
U)
         IVOIKWIUM  .,lll' .'  A   . «, A
         iioc joiil^  ','-  Al  /. ?|
            >,i.in..r,M  '.rffj
                                                                                                           S  ,J»J|1]s»«"» »«i  ••"
                                                                                                           s^ . •>'-  X! I  TV. sin. A  SA..I
                                                                                                           >fV ,    vM^JX-s1'1'1"
                                                                                                           M.£V;.< ...^M^'tx'-''
                                                   ^    ^ot^*


                                       ^..r^fl^^;''                     S
                                       -<   ^  '» ^v x.   =.?*!                    Kxi-rn   ,--  »TkS
                                        ,.   ^.   »^-s.»JC,                    loom   .,-  »?ijCi»|l  I  ';	

                                         •'1      rS'Tnl                    Bf><1"5  C1«f«"""'i»«nccMi«
                                           .1    LI I   IJ           «P»l|i SHOP      ^^.    * ^jT -'cV -  «. Si»l"P|0 IN ns.1.'
                                         . I: N   7\1 |'IOW "*      /VvS'ORtHOOU .S^.    ^^^  //   O-'   »HO in SUIK   ,
  . •I0°      i
  •IHIIK      '

' I UAOIIJG
         (i|'c-i co  t n ct  c~~\••
                                     FIGURE  3.1   -  t!U)i««Lric FlnwclidrI   for  the  Manul:anl:iir«> of  l'ortl;»n«l

                                                                hv ho Hi  Dry and Wet Pro^esHHS  (Kefernnce  1).

-------
     Air pollution control equipment in use at cement plants encompasses most
typical control systems.  Wet scrubbers, not surprisingly, are not common and
are found only on raw material handling systems, as in quarries.  Finish grind-
ing, storage, and cement transfer operations are controlled almost exclusively
with fabric filters.  Kiln emissions are controlled by fabric filters or ESPs,
often preceded by multiclones.  Clinker cooler emissions are treated by fabric
filters, ES?s, gravel bed scrubbers, and cyclone separators.  The products of
hazardous waste combustion (complete or incomplete) will appear in the clink-
er, in the kiln dust, and in the kiln exhaust gas.  That which appears in the
dust and kiln gas as a potential discharge is of much greater concern than
that which appears in the product.  One of the advantages of kiln disposal
is the potential for trace metal removal as part of the product.

     Portland cement was first made by the wet process.  Recent developments
in dry grinding and blending technology and the increased cost of energy have
made the dry process more economical.  New construction since 1971 has been
almost exclusively dry process kilns (Reference 3).  At of the end of 1982,
though, 38.8% of the installed caprcity used wet process kilns (Reference 3).

     The raw materials are the same for both processes, but the sequences and
operations for raw material blending, grinding, and crushing are different.
In the wet process, the raw material is crushed and proportioned to the ap-
proximate desired raw mix constitution, and then slurried with water to about
60-70% solids.  The slurry is ground in wet mills  (ball, tube/ball combina-
tion) to final size specifications and pumped to storage tanks.  Fine adjust-
ments of the mix may be made in the tanks before pumping to the kiln.

     The kiln length and heat requirements are the major differences between
the wet and dry processes.  The first 20-25% (100-150 feet) of a wet process
kiln is devoted to evaporating the water used for slurrying.  This results
in added water content and lower temperature of the exhaust gas compared to
dry process kilns.  Generally, wet process kilns require 5.5 to 6.0 x 106 Btu
per ton of clinker produced, while dry kilns require about 4.5 to 5.0.  Most
wet kilns are equipped with a "chain" section in the last (nearest the feed
end) 60-100 feet of the kiln as an energy conservation measure.  The chains
serve to increase the heat transfer rate from the discharge gas to the raw
feed and to aid in keeping the mass moving down the kiln.  Solids residence
time in a wet kiln is usually on the order of two hours, with gas residence
time of about 6-1C. s    'ds.  However, the important factor is not total resi-
dence time but, the time/temperature profile.  In this regard, both types of
kilns have the same basic profile for the combustion and clinkering zones.

     Referring to the solid/gas temperature profiles of Figure 3.2, the fuel
combustion zone is located in the first 50 feet of the kiln.  Burner arrange-
ments place the centerline of the flame on the cylinder axis of the kiln.  The
gases have cooled to less than 2000°F at about 40% of the kiln length for a
wet kiln.  For an average kiln length of 525 feet  (Reference 1) the effective
incineration zone is about 200 feet.  In general, this is consistent with the
dry process (preheater/precalciner) kiln design where final calcination and
clinkering occurs within this same temperature range (2000 to 2700°F), and
requires a similar length of rotary kiln.
                                     3-4

-------
CJ
O^'w/U
**O/^/*l
^dOQ
2, tr\c\
I\J\J
u.
0 9000
UJ
a:
I*T icnn
«< louu
UJ
c.
2 ioon
UJ l«£VJvJ
Q/^rt
8OU
4HP
*TVJV>






^.*~~


***^





,*•'*''










*-^





^

— • J**fc_.
C^<7/1





"'


7*




*^







/




MMf*r

,
/







.-V
/
/
y
V7




.
* 1 "*^.
' «t
p
i
1
t
1
1
V
X


                                10    20    30    40    50     60    TO     80     SO    100
                                            PERCENT CF KILN  LENGTH
                           FIGURE 3.2  - Typical  Wet Kiln  Temperature  Profile  (Reference  3).

                                         NOTIs:  Comhunl-ioii ^nnos  Flow  Count *»rcnrrr>nl to t lio fnmoul Cli,nr
-------
     The average wet process kiln has a capacity x.j produce 266,500 tons of
clinker per year, or about 850 TPD.  The kilns normally operate 24 hours per
day and seven days per week, but average 310 days per year.  The largest can
produce over 4000 TPD wnile the smallest produces no more than 220 TPD.  The
discussion later in this section describes a representative wet kiln, recog-
nizing that this representative kiln probably does not really exist.

     The hardware used to dry, calcine, and clinker the raw material is the
difference between the wet and dry processes.  The only inherent process dif-
ference is that less moisture needs to be evaporated in a dry process kiln
allowing kilns to be shorter than equivalent capacity wet kilns.  Kiln con-
figurations take three basic forms:

     o  Rotary kiln only
     o  Rotary kiln with preheater
     o  Rotary kiln with preheater and precalciner

While the physical and chemical processes are the same for all three configu-
rations, the addition of a preheater with or without precalciner improves the
process efficiency, through fuel savings and economies of larger production
units.

     A preheater system, using kiln exhaust gas, is applied to raise the
solids temperature and partially calcine the charge before feeding it to the
rotary kiln.  The preheater accomplishes the same tasks as the "back end" of
a long dry kiln but more efficiently.  The kiln is one-half to two-thirds
shorter if a preheater is used.

     Two types of preheaters are in use today; the traveling grate type and,
mot a commonly, the suspension type.  Traveling grate preheaters operate in
much the same manner as clinker coolers or traveling grate stoker boilers.
Kiln exhaust gases are pulled through a bed of pelletized raw mix supported
on a traveling grate.  The suspension type of preheater consists of from one
to as many as four cyclones arranged in a vertical tower with solids flowing
down countercurrent to kiln exhaust gases.  A graphic representation of a 4-
stage preheater kiln is shown in Figure 3.3a.  Figure 3.3b shows a 4-stage
unit with a precalciner.

     The third major configuration adds a precalciner between the back end of
the kiln and the preheater tower.  The precalciner is fueled separately from
the kiln and may use up to 60% of the total fuel burned.  Up to 95% of the
charge may be calcined in the preheater/precalciner and this means of precal-
cination results in a uniform mass being fed to the kiln for final calcination
and clinkering.  Thus, the kiln operations can be optimized for the clinkering
reactions, and the preheater/precalciner operated to optimize C02 evolution.
However, the direct contact between hot kiln gases and raw feed in the pre-
heater and the precalciner can result in condensation of sulfur and alkalies
on the feed and increase the concentration of these substances in the clinker
enough to make products of unacceptable quality.  This direct contact and re-
sultant condensation would become even more problematic if waste fuel were
burned in the precalciner.  Problems discussed below for kiln combustion of
waste fuel in preheater kilns would be magnified.  Further, the time/temper-
ature profile of the precalciner flame is much less conducive to complete
destruction of organics.  Residence time is shorter and direct gas-to-solid
                                     3-6

-------
-J
                     SECOND-STAGS
                     CYCLONES
                      THIRO-STAGg
                      CYCLONtS
                        *3UflTH-STAGS
                        CYO.ON63
COAL nilHNEH
                                                                    IIIIIOHV HMH
                                                                                   \>rr7.t-
                                   Fiqure 3. 3a  - SCHEMATIC OIAGKAM OF 4-STAGE SIISI'KNSION

-------
;fti
ww •--••*«• •«i:i
                                     CVUIUMI* ci
                                                           ^^^ _- «»• IMI
                                                         co»< m "««:«'"r •"><:l
                                                      _,Z	
                                                              	Y\
Figure
        3.11, - SCHEMATIC DIAGRAM OF 4-STAGB SUSPENSION PBEIIEATKR WITH PRBCALCIHKH

-------
charge heat transfer cools the gas strean much more rapidly  than the  gas-to-
kiln-to-solid charge transfer in a rotary kiln.

     Alkali concentration in the clinker also occurs in wet  and long  dry kilns
but to a lesser degree because more heat transfer occurs from kiln walls to
solids than gas to solids.  The alkalies may also condense in the gas stream,
but they do not deposit on the feed particles.  The condensation problem be-
comes more acute if chlorinated fuels aie burned since the alkali metal chlo-
rides condense even more readily than the oxides and can create kiln  lining
deterioration, preheater plugging, and increase the alkali recirculation in
the system.  One solution to the problem of alkali buildup commonly practiced
is bypassing some of the kil-.i exhaust to a separate control  device, thus re-
moving some of the more contaminated feed.

     Overall, dry process kilns have an average capacity of  358,000 tons of
clinker/year, or 1090 TPD.  Long rotary kilns only average 236,000 TPY, while
preheater kilns average 394,000 TPY, and preheater/precalciner kilns  average
855,600 TPY (Reference 3).  The capacity data reflect kiln age and advances
in technology as the newer kilns are larger and employ the most sophisticated
process technology.  For example, the 20 precalciner kilns of the 274 operat-
ing kilns accounted for 18.8% of the total installed capacity in 1982, where-
as there were no precalciner kilns in 1971.

     The other unit operations of the cement manufacturing process have lit-
tle relevance to the destruction of hazardous waste.  The clinker exits the
kiln to coolers where the temperature is reduced by pushing  or pulling ambi-
ent air through it.  Some preliminary size reduction is done before convey-
ing the clinker to interim storage, and then to final grinding and formula
adjustment.

Combustion
     Nearly all kilns use coal as primary fuel.  In general, lump or crushed
coal is delivered to the plant and stored in piles.  It :.s transferred from
these piles to a silo or storage bin holding a 24-hour supply,  "'rom the silo
cr storage bin, the coal is fed to a pulverizer and carried from the pulveri-
zer to the burner with ambient air preheated by mixing with exhaust air taken
from the clinker cooler.  A system of dampers is provided to achieve the de-
sire mixing of the ambient air and hot air frcm the cooler.  Secondary-com-
bustion air may be added at the burner pipe, which is a large  (approximately
24-inch diameter) tube oriented on the rotational axis of the kiln.  Auxiliary
fuel burners for startup and emergency are located along or inside the coal
pipe.  Gas and oil are the usual supplemental fuels, but wastes may be used.
Common practice, usually because of permit conditions, is to use conventional
fossil fuels rather than hazardous wastes for such occasions.  The ash con-
tribution from coal to the total loading of dust is considerably less than
that from the kiln solids.  Most operators practice readdition of collected
dust to the kiln in one form or another.  A typical coal with 8% ash burned
in a wet kiln at 5.5 x 106 Btu/ton of clinker produced will generate 40 pounds
of ash per ton of clinker produced, compared to 100-200 pounds of kiln dust
per ton, plus insufflation (pneumatic reinjection of collected dust) of 100
pounds per ton.
                                     3-9

-------
Emission and Air Pollution Control

     Trie exhaust gas leaves the kiln at temperati'-oa ranging from 400 to  525°F,
carrying the coal ash, oxides of sulfur from luel and raw material, nitrogen
oxides from the combustion process, and 
-------
                              TABLE 3.1

AIR POLLUTION CONTROL  DEVICES INSTALLED ON WET PROCESS CEMENT KILNS3
System Typv;
Electrostatic
Preci pita tor
Electrostatic
Precipitator +
Centrifugal Separator
Number
of Kilns
73
14
Percent
of Total
51 .0
9.8
Average
Reported
Efficiency
98.2%
96.4%
Fabric Filter
Fabric Filter +
Centrifugal Separator
other
Subtotal
No Data Reported
27
18
11
143
25
18.9
12.6
7.7
100.0

99.4%
99.5%
92.4%
N/A

   a  Taken from the  1980  EPA  National  Emissions Data System
      (NEDS).
                                3-11

-------
                                 TABLE  3.2

           OPERAPING CHARACTERISTICS OF DRY  PROCESS  CEMENT KILNS
Kiln Type
Wet
Long Dry
Insulated Long Dry
1 -stage Preheater3
2-stage Preheater3
4-s tag*s Preheater3
4-stage Preheater3
Bc~- End Temp.
(°F)
450-550
1300-1400
800-900
750-800
750-800
650-680
600-650
Fuel Requirement
(MMBtu/ton)
5.5
4.2
3.4
3.3
3.2
2.8
2.8
Dust Loss
(% of feed)
10-15
25-30
30-35
10-12
8-10
6-8
5-6
with Precalciner
a  Temperatures are preheater exhaust;  rotary kiln exhaust gas is 1700-
   2000°F for all types of preneaters.
                                        3-12

-------
                             TABLE 3.3

AIR POLLUTION CONTROL DEVICES INSTALLED ON DRY PROCESS CEMENT  KILN3a
        System Type
 Number
of Kilns
Percent
of Total
   Electrostatic
   Precipitator

   Electrostatic
   Precipitator +
   Centrifugal Separator
   39
   11
  29.8
   8.4
 Average
 Reported
Efficiency
  98.6%
  98.1%
Fabric Filter
Fabric Filter +
Centrifugal Separator
Other
Subtotal
No n.-,.ta Reported
42
26
13
131
36
32.1
19.9
9.9
100.0

99.4%
99.8%
94.0%
N/A

   a  Taken from the 1980 EPA National Emissions Data System
      (HEDS).
                                3-13

-------
     Several operators in the United States and abroad have burned hazardous
waste in their kilns, both as supplemental fuel and as a means of waste dispo-
sal, and have collected data on the effect of these practices.  In a previous
review for EPA, A.T. Kearney, Inc., (Reference 4), discussed the potential ef-
fects of several likely components of waste fuel on cement production.  There
are several elements which affect the quality of the product (e.g., sodium,
potassium, magnesium) but which are not usually present in hazardous wastes
in sufficient quantities relative to the feed to be significant.  There are
other elements found i.-. wastes in great enough proportions to radically af-
fect Kiln operations and piime among these is chlorine.  There are a number
of trace metals commonly found in hazardous wastes such as lead, vanadium,
beryllium, mercury, chromium, cadmium, which do not markedly affect the kiln
operations or cement quality, but which could have an impact on the environ-
ment.

     Test data collected to date support the hypothesis that the process con-
ditions required for the production of quality cement will result in the com-
plete destruction of hazardous waste.  However, even the best kiln operation
will produce poor quality cement at tiyes, so the problem of ensuring adequate
destruction of wastes becomes one of defining an appropriate set of conditions
of kiln operations.  Two aspects of waste destruction need to be considered:

     o  Destruction combustion or removal of organic constituents
     o  Collection and removal of inorganic constituents

     The ORE of a kiln involves fuel combustion variables for organics while
trace contaminant removal involves kiln chemistry and pollution control equip-
ment operations.  In practice, both are interrelated as combustion affects
kiln chemistry, and to a lesser extent, kiln chemistry affects combustion.

     Determination of those kiln operations which could relate to achievement
of some regulatory standard (e.g., 99.99% ORE or 99% HC1 removal) focused on
the types of process data routinely collected by kiln operators.  The param-
eters appearing to have the most direct connection to combustion efficiency
and therefore DRE, are kiln gas oxygen and carbon monoxide concentrations
both of which are usually monitored with continuous analyzers at existing
kilns.  The oxygen measurements are used to ensure an oxidizing atmosphere
both for clinker quality and combustion purposes.  The carbon monoxide moni-
tors are typically used as "combustibles'* monitors to prevent explosions or
fires in emission control systems, or to provide evidence of overall reducing
conditions in the kiln, unburned fuel, and unburned hydrocarbons entering the
kiln in the row materials.

     A direct measurement; of maximum clinker temperature might provide assur-
ances of sufficiently hot combustion gases as well as proper kiln operations.
However, at present there is no suitable technique for direct measurement of
clinker temperature.

     waste fuels have been burned in some types of dry process kilns witn
varying degrees of success from a process viewpoint.  It has been reported
that many cement manufacturers have conducted short-term experiments with
waste fuels, but none of their results have been reported.  Test results
have been reported for one trial of chlorinated waste and this was at St.
Lawrence Cement's Mississauga plant in Canada.  When burning chlorinated


                                     3-14

-------
waste  in  their kiln  with  a  4-stage  preheater  (no calciner),  the  alkali  bypass
duct continually  plugged  and  stack  particle emissions  increased  dramatically
because of  the control  equipment's  inability  to  handle the increased dust load.
The tests were conducted  using  a  fuel  containing 40% chlorine  by weight (tri-
chloroethylene is 89* chlorine  by weight)  and resulted in  a total chloride in-
put to the  kiln amounting to  0.3% of the clinker rate.  The fuel replacement
rate was  12%, but only  65%  of the waste fuels heating  value was  realized in
the kiln  as the rest went to  vaporizing metal chlorides (Reference 4).   Other
tests  have  been performed on  long dry  kilns in this and other  countries with
better success.   No  data  are  available on  tests  in preheatar/precalciner kilns.
Currently,  five dry  plants  burn waste  fuel and all are long Kilns.
LIME  PRODUCTION

General

      Lime  is one of  the oldest manufactured  chemicals knotra  and  the  term  ap-
plies  to a variety of calcium compounds  in the oxide and  hydroxide forms.
Over  20,000,000 tons of lime were produced in 1979; 36% of this  was  used  for
steel making (basic  oxygen  furnace),  8%  for  water purification,  6.5% for  al-
kalies, and 5.5% for pulp and paper manufacture  (Reference 5).   The  product
derives from limestone containing varying amounts of magnesium in addition
to the calcium; high magnesia limes are  referred to as dolomitic.  The  oxides
are called quick limes ana  the hydroxides are known as slaked  (slack) limes.

      The calcination of limestone to  lime is performed in rotary kilns, fluid-
ized  bed kilns, vertical shaft kilns, and rotary hearth kilns, -but 90%  of the
production comes from the horizontal  rotary kilns (Reference 6).  There are
nearly 60  companies producing lime at 155 plants in this  country, not includ-
ing such captive facilities as lime regeneration in the pulp and paper  indus-
try.   These 155 plants, located in 41 states and Puerto Rico, contain over
450 kilns, with the  top 10  manufacturers producing just under 50% of the  to-
tal output (Reference 4).

     The rotary kiln process is the only one discussed in this document.  As
previously identified, horizontal rotary kilns account for 90% of the produc-
tion;  each of the other types are suitable only for a limited scope  of  raw
materials; operating conditions of lime rotary kilns are  similar to  cement
kiln.-; and operations of lime kilns appear compatible with hazardous waste
incineration.

Process Description

     Calcination of limestone in horizontal rotary kilns  occurs  in the  same
manner as  described for Portland cement production because the raw materials
are essentially the same (limestone) and the process is equivalent (dissocia-
tion of CaO?3 to CaO and C02> except  that CaO is an intermediate in  cement
production.  Most lime kilns employ crushed or ground limestone as feed so
water evaporation requirements are minimal (comparable to dry process cement).
The charge is heated to evaporate any free moisture, which is usually less
than 1% and then to dehydrate the feed.   High-calcium limestone begins  to
dissociate at 1650°F while dolomitic  limestones dissociate over a range of
1340 to 1650°F.   The kilns are operated at maximum temperatures of 1800 to


                                     3-15

-------
2300°F to provide acceptable heat transfer rates, and at high gas flow rates
to rapidly remove the evolved CC>2 to prevent recarbonation.  Fuel require-
ments are anywhere from 4 to 10 x 106 Btu/ton of lime (Reference 6) to meet
the required heat of reaction of 2.3 x 10*> Btu/ton.  On the average, the ro-
tary kilns use about 7 x 106 Btu/ton (Reference 7) plus the energy needed
for kiln rotation, material handling, and pollution control equipment.

     Kiln production rates range from 250 to 2,500 TPD, with the average
plant producing approximately 135,000 TPY (about 450 TPD).  The largest is
Marblehead's Thorton plant which is rated at 600,000 TPY or 1,950 TPD (Refer-
ence 4).  Kiln sizes are comparable to cement kilns though somewhat shorter
and range from 6 to 14 feet in diameter and from 60 to 450 feet long (Refer-
ence 5).  Fuel is burned in typical rotary kiln manner with a burner or burn-
er pipe pointed up the kiln's rotational axis.  Kiln slope is 3-5° which is
a bit lower than that for cement kiln (6°).

Combustion
     Coal accounts for almost 70% of the fuel used in lime production while
natural gas is used for 23%.  Oil, liquified petroleum gas (LPG), electricity,
and other fuels comprise the remainder.

Emissions and Air Pollution Control

     Kiln gases are exhausted at temperatures ranging from 500 to 1400°F.
Emissions are controlled with fabric filters, E2?s, venturi scrubbers, and
occasionally, gravel bed filters.  Gas cooling required for baghouses and
ESPs is provided by stone preheaters, water evaporation, ambient air dilu-
tion, or radiation/convection heat exchangers.

     Control of emissions of particulate matter has been the major concern
with respect to lime kilns, though they also emit sulfur oxides, nitrogen
oxides, and carbon monoxide.  Typically, kiln exhaust gases contain 50% of
the fuel sulfur if high sulfur fuel is burned, resulting in an SC?2 concen-
tration of 1500-2000 ppm SO2.  The concentration of NOX is about 50-200 ppm
(average 160 ppm) and of CO is 10-150 ppm (average 110 ppm).  Wet scrubbers
result in an alkaline scrubbing medium if recirculation is practiced and will
remove HC1 to seas degree if chlorinated fuel is burned,  some acid gas ad-
sorption by the reaction of the gases with lime particles suspended in the
gas stream is also likely.
LIGHTWEIGHT AGGREGATE KILNS

General

     The term "lightweight aggregate" describes a range of special use aggre-
gates that have an apparent specific gravity considerably below normal sand
and gravel.  These lightweight aggregates will range from extremely light
materials used for insulative and non-structural concrete all the way to ex-
panded clay-like material? used for structural concrete.  Some of these, e.g.,
pumice and scoria, occur as natural aggregates while the others are produced
                                     3-16

-------
by a variety  of processes.   Of  interest  to  this  study  are  the  expanded clay-
like materials  (i.e.,  clay,  shale,  or  slate), produced by  the  rotary  kiln sin-
tering process.  The kilns used in  producing them  are  very similar  to those
used in the Portland cemsnt  and lime industries.   This class of  lightweight
aggregate is  used  in concrete in place o: sand and gravel  or stone.   Concrete
made with this  aggregate has about  th >. same strength but weighs  approximately
one-third less  than concrete made with sand and  rock.

     There are  30  plants in  operation  in 24 ftates producing lightweight ag-
gregate by the  rotary  kiln process  (Reference 7).   These plants  typically
have two or three  kilns  (Reference  8).   Thus, there should bt  >^tween 60
to 90 of these kilns in  the  United  States.  The  1979 total capacity of the
rotary kiln process was  approximately  15 x  1 O6 cubic yards of  product (12.6
x 106 tons) (Reference  2).  Six plants  have ceased operation  since 1979 and
therefore the current  capacity  is somewhat  lower.

Process Descripti-on

     A typical plant consists of:   (1) quarries  from which the raw  material
(clay, shale, or slate)  is extracted;  (2) raw material storage and  handling
facilities; (3) one or more  rotary  kilns to convert the clay-like material
into lightweight aggregate;  (4)  aggregate product  handling and storage areas;
(5) fuel storage and handling facilities, (6) a  kiln burner system  for coal
and/or liquid fuel; and  (7)  a kiln  emissions control system with a  wet scrub-
ber being the most conunon type  of control equipment.

     The raw  material  used is a highly siliceous clay,  shale,  or slate that
exhibits a bloating characteristic  when  heated.  The bloating  results because
certain minerals in the  raw  material liberate gases at the temperature of in-
cipient fusion.  The raw material is bloated to  the point  that it becomes
soft and pliable,  but  not to the point of completely melting.

     In the rotary kiln process  (Figure  3.4), the  raw  material is crushed and
introduced at the  upper end  of  a kiln similar to the type  used in the lime
industry.  (In fact, some aggregate kilns were formerly lime kilns.)   In pass-
ing through the kiln,  the material  reaches  a temperature of 1900° to  2100°F,
and begins to become plastic.   Internal  gases causes the material to  expand,
or bloat, and create a mass  of  small, unconnected  air  cells, which are re-
tained after  the material cools  and solidifies.  After leaving the kiln,  the
material is cooled and then  crushed and  graded.

     As with  cement and  lime production, maintaining the proper  temperature
in the kiln is crucial to making good lightweight  aggregate.   Heat  for the
process is provided by a burner  located  at  the clinker discharge (lower)  end
of the kiln.  The  temperature of the kiln is controlled by adjusting  the fuel
feed rate, kiln rotation speed,  and amount  of air  flow through the kiln.   Con-
trol is almost exclusively manual.  Tho  average  temperature in the hottest
zone of the kiln (near the burner) is usually held  in  the  range  of about 1900°
to 2100°F (i.e., the temperature of incipient fusion of the charge material),
which is nearly identical to the temperature range  at  the  hottest zone of a
lime kiln.  If the kiln temperature falls below  that general range, the clay
structure is not converted to the desired inorganic formation.   If the tem-
perature rises above the desired range,  the material fuses, forming a heavy
coating on the kiln interior that must be removed manually  after stopping and


                                     3-17

-------
   (I)
   (?)   sinnriiHir,
   ('j)   w:nn
   (j)   DOIAHY KIIN
   (S)   ri IIUIH CINMIR
   (i)   ionium
   (/)
   (a)
            IIOU
COAL
                                                                           (f.)
FIGURE 3.4    _ SOHFMATIC DIAGRAM OF  A TYPICAL LIGHTWEIGHT AGGRKATK  PLANT (HRFKRKNOF  2 )

-------
cooling down the kiln.  Visual observation of  the kiln contents is  the  usual
method of determining that the desired  lightweight aggregate  is being pro-
duced and not some form of the undesiraole alternatives.  The  temperature of
the clinker at the discharge  is about 1600°F.  The kiln off-gas, at a temper-
ature of about 500°F, is vented through an air pollution control system to  an
exhaust stack.

Combustion

     Kilns are typically 7 to 10 feet in diameter and 120 to  300 fee*, long
and are refractory lined.  Kiln rotation rates are 60-100 revolutions per hour
with solid material traveling down the  sloped  (3 to 4°) kiln  and hot gases  in
the opposite direction.  Production rates range from 250 to 1000 tons per day
(Reference 8).  Fuels include natural gas, ail, and coal, with a trend  toward
increasing coal use.  The energy required per  ton of product  is 3 to 6  x 10°
Btu/ton of lightweight aggregate.

     The hot clinker material discharged from  the rotary kiln  is air-cooled
to near ambient temperature.  For this cooling operation, moving grate  or ro-
tary cooler equipment is commonly used.  Some  of the off-gas  from the cooling
operation is used for combustion air for the kiln burner while the  balance  is
passed through an air pollution control system to an atmospheric vent stack.
Because of the necessity of balancing the gas  flow through both the cooler
and the kiln, this off-gas is not passed through the control device installed
on the kiln.  If necessary, the cooler section will have its own control de-
vices.

Emissions and Air Pollution Control

     Table 3.4 summarizes the air pollution systems used in the domestic pro-
duction of lightweight aggregate by the rotary kiln process (Reference  9).
These data show that for 36 producing facilities in 1979, 25 plants (or  69%)
used wet scrubbers to remove  particles from the kiln off-gas stream before
venting to the atmosphere.  Six plants  (approximately 17%) used fabric  fil-
ters for the same pollution source and one plant used an ESP.  As discussed
above, only 30 plants were operating in 1983.  No update of the data pre-
sented in Table 3.4 was found.
BLAST FURNACES (STEEL INDUSTRY)

General

     A blast furnace is essentially a large countercurrent flow chemical
reactor for reducing iron ore to pig iron for us« in the iron and steel in-
dustry.  Because of its function, it is the keystone of this industry.  Raw
materials and energy are supplied to the furnace and molten iron and slag are
withdrawn by several major auxiliary equipment items which are typically ar-
ranged as depicted in Figure 3.5.  The major raw materials include iron ore,
coke and limestone.  Energy is provided by combusting the coke supplemented
usually with one of a number of fuels.
                                     3-19

-------
                                  TABLE  3.4

      AIR POLLUTION DEVICES  INSTALLED AT  LIGHTWEIGHT AGGREGATE PLANTS
Number of Plants
Employi ng Contro1
	Systems	
       System Type
 Percent of
Total Plants
       25

        6

        1

        4

       35
Wet Scrubber

Fabric Filter

Electrostatic Preci pita tor

No Data
    69.4

    16.7

     2.8

    11 .1

   100.0
                                     3-20

-------
                              DUST CATCHER
                                  IAS WASHER
STOVE
STACK
                                                                      mX
SCALE CAR

  SKIP CAR
                                                                         BOILER STACK
   TURBO-BLOWERS
     8LAQ LADLE

     CAST HOU8E
BY-P
8TA
DAMP*
DAMPER
=*==€
ASS
CK~~^
•RS^c

BLOWER*
1
BAG HOUSE
\ 	 — ~

|\/\/\/V
|u 	 	 	
i
1
1
1
i
1
-=17 i
	 " 1
VCONVEYOR
CAST HOUSE CONTROLS
                    FIGURE 3.5 BLAST FURNACE WITH AUXILIARY EQUIPMENT
                                                                                  REFERENCE 10

-------
     The principal reactions are between coke, air, and iron ore.  Part of
the coke combines with the oxygen in the air to produce the principal iron
ore veaucing agent, carbon monoxide.  This reaction generates  the required
heat tor the process.  Another part of the coke reacts with the iron oxide
in the ore to release free iron whr'.ch melts, drips to the fcottom of the fur-
nace and collects in the hearth.  A final portion of the ca.rbon, approximate-
ly 4%, dissolves in the iron (Reference 10).

     Other important reactions are involved in the removal of  impurities from
raw materials.  Limestone is the principal fluxing material used for removing
the impurities.  It is calcined by the heat of the hot blast air in the fur-
nace.  The resulting CaO reacts with the impurities in the ore, principally
sulfur, to form a slag, which in the molten form, descends to  the hearth.
Because the slag is less dense than the molten iron collected  in the hearth,
it floats in a separate layer on it.

     When the furnace hearth becomes full of the molten iron and slag, usual-
ly evary three to four hours (Reference 10}, tap holes in the  furnace wall
are unplugged.  The molten iron is drained through an opening, known as an
iron notch, into a ladle for transport to the steelmaking facility or to a
pig casting machine.  The molten slag is removed through an opening, known
as a cinder notch, located at a higher elevation than the iron notch.  The
slag discharged from the blast furnace is either collected in  a slag ladle
for transport to disposal or into a slag pit.  Once solidified, the slag may
be used as aggregate for concrete, road bed, etc.

     The energy needed for the iron smelting process is supplied oy the com-
bustion of the coke supplemented by some other fuel as previously mentioned.
Thermal economy demands that the combustion air be preheated to about 1900°F
to 2000°F before it is blown into the furnace.  Preheating is  achieved by
passing the combustion air through a vertical regenerator chamber known as
a stove, which is heated by firing some of the gases generated by the blast
furnace reducing reactions.  Each furnace utilizes three to four stoves, one
of which is heating the blast air while the others are being heated by burn-
ing the blast furnace gas.  Because this gas is laden with dust, it is passed
through a high efficiency dust removal system before being combusted in the
stoves.

Process Description

     A blast furnace is a vertical, cylindrical structure that is composed
of four major sections:  the top, the inwall, the bosh, and the hearth  (see
Figure 3.6).  The top section receives the charged burden and  collects the
blast furnace gases before it exits the furnace and serves to  dry and preheat
the solid raw materials, commonly termed "burden".  The burden is charged in-
to the top section of the furnace through a system of two or three pressure
sealing bells.  This multiple-bell charging system allows the  solids to be
charged while preventing the escape of the blast furnace gases.  A blast fur-
nace is normally operated under high positive pressure (approximately 100 in-
ches water at the top of the furnace) to obtain increased production (Refer-
ence  10.  The walls of the top section are vertical and lined on the inside
with steel wear plates to protect the lining from abrasion from the charge.
Hot gases rising from the lower sections of the furnace dry and preheat the
raw solid materials in this top section.  These gases accumulate in the top
                                     3-22

-------
        Urv*Ml
        Hoppw

FIGURE 3.6 TYPICAL BLAST FURNACE
                                       REFERENCE  10
                3-23

-------
section and are discharged through several off-takes  (ducts) to a dust clean-
ing system.

     The main section of the furnace where the iron oxides are reduced to
free iron is known as the inwall.  It is tapered outward from the top to the
bosh section below.  It is lined with firebrick and the lower portion is wa-
ter cooled by means of steel cooling plates which are embedded in refractory.

     The coke combustion occurs in the bosh section which extends from the
widest part of the inwaJ.l down to just about the level where the combustion
air is injected through nozzles known as tuyeres.  This section is usually
lined with super-duty firebrick and is cooled by cooling plates.

     The bottom section of the furnace is known as the hearth.  It extends
from just above the tuyere level to the very bottom of the furnaca.  The
molten iron and slag formed in the upper sections of the furnace accumulate
in the hearth before they are periodically removed.  Heated combustion air
enters the furnace in the hearth section of the furnace through the tuyeres.

     Encircling the bosh section is a manifold for the hot combustion air
called a bustle pipe.  It directs the hot blast air from the furnace stoves
to the tuyeres located around the perimeter of the hearth.  Depending on the
size of the furnace, there may be from approximately 10 to 40 tuyeres.  Air
is conducted from the bustle pipe to the tuyeres through a series of connec-
tions as depicted in Figure 3.7 (Reference 11).  The connections ^e designed
for easy removal of the tuyere nozzle and blow pipe.  The blow pipe is im-
portant in the consideration of burning hazardous wastes because it the most
strategic point for injecting these wastes into the furnace as explained la-
ter.

     Blast furnaces vary considerably in size and capacity.  The 90 furnaces
in operation in the United States in 1983 ranged from approximately 90 to 150
feet high with hearth diameters of 17 to 45 feet (Retere_ce 12).  Table 3.5
shows the location of these blast furnaces.  The average height and hearth
diameter are 102 and 27 feet, respectively.  Capacities typical range from
550 to 7,000 tons of iron product per day although units of 10,000 tons per
day capacity have been built (Reference 13).  The U.S. 1983 production of
iron produced in blast furnaces was 48.7 million tons (Reference 14).

Combustion

     The smelting of iron ore in a blast furnace is highly endothermic.
Thermal energy for the process is provided primarily by the combustion of
the coke in a relatively small zone above the tuyeres.  Here the carbon in
the coke reacts with the oxygen in the hot blast air to form carbon monoxide
and carbon dioxide.  At the approximately 2,800°F temperature in the combus-
tion zone (Reference 11), the carbon dioxide is unstable and reacts with the
coke to form additional carbon monoxide.  The reaction of the coke with oxy-
gen is exothermic, but the reaction of coke and carbon dioxide is endother-
mic.  However, the overall conversion of the coke to carbon monoxide is quite
exothermic.
                                     3-24

-------
N>
01
                                                                        STACK-COOLING PLATE
                                CENTER
                                 LINE
                                 OF:
                                FURNACE
fUYERE COOLER CASTING

            TUYERE
                                                                          TUYERE STOCK

                                                                         >«-PEEP SIGHT

                                                                          TUYERE CAP


                                                                   BLOWPIPE
                              v «
                             FIGURE 3.7 SECTION THROUGH TUYERE OF BLAST FURNACE
                                                                                                 REFERENCE

-------
                                  TABLE 3.5

                     BLAST FURNACES IN THE UNITED STATES
       Company
          Location
  Number of
Blast Furnaces
Armco, Inc.
Bethlehem Steel Corp.
Inland Steel
Interlake
Lone Star Steel
LTV Steel
McLouth Steel
Products Corp.
National Steel Corp.

Rouge Steel Co.
Sharon Steel Corp.
Shenango, Inc.
United States Steel
Corp.
Weirton Steel Corp.
Wheeling-Pi ttsburg
Total
Ashland, Kentucky
Hamilton, Ohio
Middletown, Ohio
Bethlehem, Pennsylvania
Burns Harbor, Indiana
Sparrows Point, Maryland
East. Chicago, Indiana
South Chicago, Indiana
Lone Star, Texas
Aliquippa, Pennsylvania
Warren/Niles, Ohio
Youngs town, Chi o
Cleveland, Ohio
East Chicago, Indiana
Gadsden, Alabama
South Chicago, Illinois
Trenton, Michigan

Granite City, Illinois
Ecorse, Michigan
Dearborn, Michigan
Sharon, Pennsylvania
Neville Island, Pennsylvania
Braddock, Pennsylvania
Fairfiald, Alabama
Fairless Hills, Pennsylvania
Lorain, Ohio
Gary, Indiana
Geneva, Utah
Weirton, West Virginia
Steubenville, Ohio
Wheeling, West Virginia
      2
      2
      1
      4
      2
      4
      9
      2
      1
      3
      1
      2
      6
      3
      2
      1
      1

      2
      3
      3
      2
      2
      3
      2
      3
      4
      5
      3
      4
      5
                                                                     90
Source:  Reference 12.
                                     3-26

-------
     Coke is essentially  the  sole primary  raw material  for providing  the  heat
and reducing agent  for  the  modern tall blast furnace  because  of  its physical
and chemical properties.  Physical  characteristics  of greatest importance is
its hardness and compressive  strength.  These properties  give coke the  neces-
sary high abrasion  resistance and the ability to  support  the  burden and allow
the hot furnace gases to  rise uniformly through the burden.   The chemical pro-
perty of importance is  its  ability  no be effectively  utilized in the  smelting
and combu«*icjj reactions.

     Other primary  fuels, such as charcoal and coal,  were used in the devel-
oping stages of iron smelting when  much shorter furnaces  wera employed.   Some
of these materials  are  still  used as the principal  fuel in the few low-shaft
furnaces used for small-scale iron  production.

     Since the late 1950's, the injection  of an auxiliary fuel into the blast
furnace has become  a common practice to reduce coke consumption  and to  improve
control of the furnace  temperature.  The economy  of iron  production is  strong-
ly dependent upon the amount  of coke required to  produce  a ton of iron  pro-
duct.  The amount of coke required  to produce a ton of  pig iron  has dropped
from 1,760 (in 1955) to 1,260 (in 1969) pounds of coke  per ton of pig iron,
and in some furnaues the  rate is below 1,000 pounds per ton (Reference  11).
Fuel injection is one of  the  effective methods being  employed for reducing
the coke requirement.

     Several fuels  have been  used for this purpose  including  natural  gas,
coke oven gas, oil,  tar, pitch, pulverized coal,  coal-oil slurries, and re-
cently, waste organic solvents.  Although  the method  of injection differs
with the type of fuel,  the injection location is  generally at the blow  pipe 6
inches to 1-1/2 feet or more  back from the nose of  the  tuyere (Reference  11).
When a liquid fuel  or coal injection is used, a lance pipe is generally in-
serted diagonally through the side  of the  blow pipe by  means  of  an airtight
connection to conduct the fuel into the air blast.  Alternatively, it may be
injected through a  louver in  the blow pipe face plate beside  the peep sight.
Gases may be injected through specially designed  openings in  the tuyere nose
or through a lance  in the blow pipe.  The  lance for liquid fuels are  gene-
rally made by the steel company's maintenance department  and  may employ an
atomizing nozzle.

     The injection  of hydrogen bearing fuels into the blast furnace brings
about considerable  change in  the composition and  properties of the ascending
furnace gases (Reference  11).  The  fuel injection is more complex than  the
simple replacing coke carbon  by fuel carbon units.  For this  reason,  the  in-
jection rates are generally limited.  The  limiting  quantity varies with the
fuel.  For natural  gas,  the injection rate is typically limited  to between
1,000 and 2,000 standard cubic feet per ton of metal  (40  to 90 pounds per
ton) and for oil between 50 to 150 pounds per ton (Reference  11).

     Firing of a Hazardous Waste Deriv-d Fuel (HWDF) requires provisions  for
injecting depending on  the compatibility of the HWDF with the other auxiliary
fuel.  Modifications to the auxiliary fuel injection system should not  be  re-
quired if the HWDF  is of similar properties,  e.g.,  the system used for  inject-
ing fuel oil could  also be used for injecting many  liquid organic solvents.
                                     3-27

-------
     Some modifications to the auxiliary fuel injection configuration niay
be required if the HWDF only supplements another auxiliary futl.  ::he HWDF
could be blended with the fuel prior to infection or a separate fusl injec-
tion system be provided.  Blending the waste with the other auxiliar'1 fuel
requires changes in the fuel handling system but none in the burner gun.
Tnis approach requires the HWDF be compatible with the other fuel being
injected.  Also, special control provisions are needed to switch from the
blended fuel to one containing no hazardous waste in the event of a process
upset.

Emissions and Air Pollution Control

     Approximately three tons of g«ses are evolved for every ton of iron
produced (Reference 15).  The volume and composition of these gases can vary
significantly with how the furnace is operated and the amour.t of preparation
of the burden.  The amount of dust leaving the furnace can range from 0.01
to 0.15 unit weights of dust per unit or iron produced with concentrations
as high as 69 grains per standard cubic foot of gas (Reference 10).  Other
major constituents include carbon monoxide (23 to 40 volume percent), carbon
dioxide (12 to 16 volume percent), hydrogen (2 to 6 volume percent), moisture
(2 to 12 volume percent) and nitrogen (30 to 67 volume percent) (References
10, 16-18).  Gases leave the furnace at temperatures of 350 to 540°F (Refer-
ence 16).

     Because of its high carbon monoxide content, the blast furnace gas has
a heating value of about 75 to 90 Btu per cubic foot, and is used as a fuel
within the steel plant.  A survey (Reference 14) of 18 blast furnace facil-
ities found that 41.3% is burned in the stoves, L2.2% in boilers, 2.0% for
underfiring coke ovens, 3.0% is flared with the remaining gas going to soak-
ing pits and combustion engines.

     Before the blast furnace gas can be used a. a fuel AS described above,
it must be cleaned of particulate matter.  Otherwise, particulate matter
would tend to deposit in the combustion zones of these units causing pre-
mature outages and failures.  Since these units are necessary to trie opera-
tion of the iron making process and since they require a high investment of
capital, cleaning the blast furnace gas is deemed an essential part of the
iron making process, aside from any environmental consideration.

     To obtain the high level of particulate matter removal needed to use
the blast furnace gas as a fuel in the steel making plant, two or three
stages of gas cleaning are employed.  The first stage consists of either a
settling chamber or a dry cyclone to remove the large particles.  The vast
majority of steelmakers employ two stages of cleaning with a venturi scrub-
ber providing the secondary cleaning (Reference 12).  A few two stage dust
removal systems use an electrostatic precipitator as the second stage de-
vice.  Where three stage systems are used, the second stage is generally a
wet scrubber and either a venturi or an electrostatic precipitator provides
the third stage.  Any of these combinations are capable of providing a clean
fuel of less than (0.01 grains per cubic feet) for use in the steel plant
(Reference 1).
                                     3-28

-------
     Fugitive remission.,  to  the  atmosphere  occur  during the production of iron
during hot  metal  :ransfer operations  in the  cast house and when there is a
blast furnace "slip".  A "slip"  is  a  process upset  where  a stratum of the
material charged  to  a  blast furnace does not settle with  the material below
it,  thus leaving  A gas filled space between  the  two portions of the charge.
When this unsett.1 = 1  stratum of  charge collapses,  the displaced gas may cause
the  top gas pre.e ,ure to  increase above the safety limit,  thus opening a coun-
ter  weighted bleeder valve  to the atmosphere.

     There  aie  no federal New Source  Performance Standards (NSPS)  for control
of air emissions  from  blast furnaces  in the  iron and steel industry.   Since
the  blast furnace exhaust gases  are not directly vented to the atmosphere,
control of  emissions while  burning  a  HWDF  would  be  accomplished by the inter-
nal  process control  requirements previously  described.

Compatible  Wastes

     There  are  several materials that are  known  to  adversely affect the pig
iron quality and/or  the  blast furnace equipment.  Sulfur  and titanium bearing
compounds affect  the quality of  the pig iron.  The  molten iron reacts with
sulfur to form  ferric  sulfides which  alter the properties of the pig  iron.
Titanium forms  compounds with carbon  and nitrogen which are soluble at high
temperatures but  have  limited solubility at  normal  temperatures.   These com-
pounds come out of solution and  make  the metal very viscous when it flows
from the furnace  and starts to cool.   Consequently,  the iron freezes  in the
runners and causes handling problems.

     Alkali metals have  several  major adverse effects. They attack the re-
fractory in the furnace  as  well  as  the stoves used  to heat the blast  air and
are  an important  contributor to  the formation of  furnace  wall deposits and
furnace bridging  which leads to  slips (Reference  11).   Alkali metals  also re-
duce the iron production and increase the  coke rate (Reference 11).

     Kalides, in  high concentrations,  attack the  furnace  and stove refractory
••r.r:  contribute  to the formation  of  slips.  For these reasons,  the  rate of hal-
v> v-snated compound firing as auxiliary fuel must  be  limited.

     Lead and zinc are two  other very troublesome metals.   Lead is readily re-
..uced in the blast furnace  but is almost insoluble  in the hot metal.   Amounts
of lead greater than can be volatized as fume may collect beneath  the molten
iron in the hearth with  disastrous  effect  on the  lining (Reference 11).   Zinc
vaporizes and travels upward with the furnace gas.   Some  of  the vapors con-
dense in the upper part  of  the furnace lining next  to the shell.   Unless the
top  temperature is high  enough,  this  material will  accumulate in the  furnace
top and split the lining or the  shell.

     Because the  blast furnace is the keystone of a steel plant, upsets  in
the  operation caused by  the firing  of  waste  would be extremely costly.   Steel-
makers are,  therefore,  expected  to  be  very cautious to limit the content of
the materials described  above in hazardous wastes fired as auxiliary  fuel.
An upset in furnace  operation resulting from firing wastes could quickly off-
set  the fuel savings realized from  this practice.
                                     3-29

-------
     In addition, there is concern of the release of undestroyed toxic wastes
with fugitive emissions.  The casting and flushing of slag are more common
sourcss of fugitive emissions.  Because gases escape from the tap holes, in-
jection would have to be stopped during these operations.  Fugitive emissions
can also occur at the tup cf the furnace as a result of leakiag seals in the
charging system or during slips.  It is unlikely that high concentration of
toxics would escape with the off-gases.  As mentioned in Section 3, the EPA
Burner Survey showed three blast furnaces as burning a hazardous waste in
1982.  These facilities reported the following:

     o  Facility 1 - 208,000 gallons/year of tar sludge and coke oven by-
                     products .

     o  Facility 2 - Plant 1 - 5,861,729 gallons/year of recovered solvents
                     still bottoms and other wastes.
                     Plant 2 - 3,907,819 gallons/year of similar wastes.
REifERBERATORY FURNACES (OPEN HEARTH FURNACE - STE2L INDUSTRY)

General

     A reverberatory furnace delivers heat to the furnace charge by directing
a long flame above the charge.  The flame not only radiates heat directly to
the charge supported on the furnace hearth but also heats the refractory in
the furnace roof and walls which radiate and reflect thermal energy to the
charge.  The term reverberatory is derived from the fact that radiant energy
provided by the flame is reflected from the roof and walls onto the charge.

     Reverberatory furnaces are found in numerous applications.  The ma lor
use has been in the production of steel in the open hearth furnaces which is
a reverberatory furnace.  They are also widely used as rerneIt furnaces in
the aluminum industry.  Reverberatory furnaces are used for smelting copper,
lead, zinc, and tin.  The detailed analysis of reverberatcry furnaces will
be limited to open health furnaces.  There were 44 open hearth furnaces in
operation in 1984 (Reference 12) as indicated by Table 3.6.  These varied
in capacity from 262 to 420 tons per heat.  No new furnaces are being con-
structed and old ones are being phased out because of higher cost of pro-
duction per ton and lower productivity compared with basic oxygen furnaces.
The open hearth furnace accounted for only slightly ovor 7% (5,951,000 tons)
of the total 1983 United States steel production (Reference 19).

Process Description

     The open hearth furnace is used to produce steel from a mixture of scrap
and hot metal in varying proportions.  Production of steel in the furnace is
a batch operation and consists primarily of the melting the solid materials,
oxidizing the impurities, reducing the iron oxides to iron and adding the de-
sired alloying constituents.  The major impurities include carbon, silicon,
manganese, phosphorous and sulfur.  Major raw materials include the pig iron,
scrap iron and a fluxing agent (usually limestone).  Iron ore may also be
added to provide low cost iron units and oxygen for controlling the silicon
and carbon content of the steel.  Oxyoen lancing is used to accelerate the
refining period.
                                     3-30

-------
                                  TABLE 3.6

             UNITED STATES OPEN HEARTH FURNACES OPERATING IN  1984
Company
Armco, Inc.

Bethlehem
Steel Corp.

Inland Steel
Co.
Lone Star
Steel Co.
Location
Middle town,
Ohio
Sparrows Point,
Maryland

East Chicago,
Indiana
Lone Star ,
Texas
Number
of
Furnaces
6

7


7

5

Capaci ty
each Furnace
( tons /heat )
310

420


335

262

Air Pollution
Control System
venturi each
furnace
6 ESPs & 2 Ven-
turis connected
in parallel
5 ESPs connected
in parallel
3 steam-hydro
scrubbers
United States   Fairless Hills,
Steel Corp.     Pennsylvania
                Geneva, Utah
10
            400
340
connected in
parallel

9 ESPs connected
in parallel
followed by 1
secondary ESP

8 ESFs/venturi
combinations
connected in
parallel
Source:  Reference 12.
                                     3-31

-------
     A typical "heat"  (a tap-to-tap cycle) is completed  in  the  following  se-
quence:

     o  Charging the scrap into t.he furnace through hydraulically  or  elec-
        trically controlled charging doors.

     o  Adding molten  metal through the charging doors by pouring  the metal
        from a ladle into a receiving trough.

     o  Injecting oxygen into the molten metal and analyzing the metal until
        the impurities are removed.

     o  Adding the desired alloying agents.

     o  Tapping the slag and molten steel.

     Oxidizing of the  impurities takes place by the reaction with  oxygen
from the ore, from the decomposition products of the limestone  and from the
jet of pure oxygen injected into the bath.  The remcved  impurities accumu-
late in a continuous layer on the surface of the liquid  metal.

     when the molten metal bath is determined to be the  grade o2 steel re-
quired, the contents of the furnace are drained through  the tap hole  into
steel ladles.  The slag, which floats on the top of the  metal,  overflows
into a slag pot when the ladle has been filled with steel.  Alloying,  re-
carbonizing and deoxidizing agents are added to the molten metal before
the slag starts flowing.

     The tap-to-tap time of the furnace not using oxygen lancing is typi-
cally 8 to 10 hours compared to 4 to 5 hours with lancing (Reference  20).

     Figure 3.8 shows  the typical open hearth furnace design used  in  the
steel industry.  The open hearth furnace is a large reverberatory  furnace
consisting of two major sections:  (1) the melting chamber, and (2) the re-
generator (checker) chambers.  The melting chamber is where the actual steel
production takes place.  This chamber is a large shallow basin  (hearth) en-
closed by refractory bricked walls and roof.  Provisions are contained in
the walls for raw material charging and for molten metal and slag  removal.
Heat is supplied from  burners at each end of the chamber.  The  flame  tra-
vels the lenych of the chamber above the charge which rests on  the hearth.
The front vail which contains charging doors are straight but the  back wall
slopes to facilitate the flow of metal out of the furnace during tapping.
Near the bottom of the back wall is a tapping spout for  removing the  molten
slag and metal.  The furnace is reverberatory in that the hearth is heated
both directly by the burner flame and by the radiant heat from  the furnace
roof and walls.

     Raw materials are charged through the charging doors by use of a charg-
ing machine, a large vehicle which moves on tracks in front of  the furnace.
Solid materials are placed in charging boxed which are positioned  within  the
furnace by the charging machine.  When the box is inside the furnace,  it  is
rotated about its axis, dumping its contents into the furnace.  To provide
distribution over the hearth, there are at least five doors for charging.
Molten metal is charged through the charging doors by pouring it i torn a la-
dle into the receiving trough.

                                     3-32

-------
                                   TO ELF; ROSTATIC PRECIPITATOR
nrrne
SWOT
                                                                . UK

                                                                YiUE
                                                 CH£CS1 FUJE
                                        KSEKESATtYE C8MBO
                    mom
            FIGURE 3.8 OPEN HEARTH FURNACE
                                                       REFERENCE 20
                              3-33

-------
     The open hearth furnace is regenerative, i.e., it uses the heat removed
from the exhaust gases to preheat the combustion air.  A regenerator chamber
is located at each end of the furnace for this purpose.  Hot gases from the
furnace at tempera  res of 3000°F (Reference 23) are discharged through the
uptakes or ports at the ends of furnace, then through a slag pocket to re-
move heavy particles oi slag carried by the gases, and finally through the
regenerator chambers.  Flow of combustion air and exhaust gases is in a
single direction at any given time,  with combustion air being preheated in
one regenerator chamber while the furnace exhaust gases are being cooled in
the other.  The flow is reversed periodically to permit the chambers to al-
ternately be heated and cooled and to maintain a reasonably uniform combus-
tion air temperature.

     The furnace melting chamber is quite large providing gas retention times
of 2 to 3 seconds (Reference 23).  Dimensions vary widely with the furnace
capacity.  The smaller furnaces in operation today typically have inside bath
dimensions of about 20 feet in width by 50 feet in length and height from
hearth to roof of 11 feet (Reference 13).  The larger furnaces are typically
about 30 feet in width and 75 feet in length (Reference 13).  The long resi-
dence time coupled with the high temperatures (3500°F flame temperature and
exit gas temperature of 2700T) operation makes these furnaces suitable for
destroying hazardous wastes (Reference 23).

Combustion
     Heat- for the production of steel in the open hearth is provided by the
combustion of a combination of the following fuels:  natural gas, oil and
tar and pitch.  All furnaces are equipped to burn both a liquid and gaseous
fuel.  The heat input requirement is 3 to 4 million Btu per ton when molten
iron is charged compared to 4.2 to 5 million Btu per ton with all cold scrap
(Reference 20).

     The fuel is fired through two long '.approximately five feet) combina-
tion burners (i.e., capable of firing both gaseous and liquid fuels) located
at opposite ends of the melting chamber.  They are designed to throw a very
long flame across the furnace hearth to generate the high temperature needed
for steel production.  Since the liquid fuel burned may often be quite vis-
cous, some type of atomization is generally provided.  Fuel firing alternates
between burners.  Alternating of firing between burners is synchronized with
the furnace gas flow reversal such that the burners are always firing in the
direction of flow.  Switching of the burner firing is automatically controlled.

     Firing of the burners begins after the scrap is charged and is continuous
until the analysis of the metal indicates that the desired carbon-temperature
relationship is achieved during oxygen lancing.  Thereafter, the burners are
fired intermittently to maintain the proper temperature.  The firing is con-
trolled manually; monitoring of combustion gas constituents to ensure good
combustion is not practiced.

     Air for combustion is delivered from blowers through the preheating re-
generator chambers into the ports in which the burners are mounted.  A set
of dampers is provided in the ductwork to allow for the reversal of gas flow.
Air enters th<* ductwork through a forced air inlet valve; passes through the
checkers, rises through the uptakes at the end of the furnace and passes by


                                     3-34

-------
the burner where it combines with the injected fuel.  The combustion products
pass over the bath, go down the uptakes on the other side of the furnace,
pass through the checkers giving up heat to the brick, and exit through  the
waste heat boiler to control device followed by the stack.

Emissions and Air Pollution Control

     Both primary and fugitive air emissions result from the operation of an
open hearth.  Primary emissions refer to those emissions leaving the furnace
through the regeneration chambers.  Fugitive emissions from the furnace  result
from ancillary operations such as furnace charging, tapping and flame reversal.

     Particulate matter is the major pollutant found in the primary emissions.
These are generated by the refining process and by combustion of the furnace
fuels as well as the oils and grease on the metal scrap.  The greater portion
of the particulate matter is formed by the refining process and is emitted
from the furnace as metallic oxides (Reference 24).  These particulate emis-
sions consist of very small particles with nearly 50% being smaller than 5
micrometers (Reference 25).  Uncontrolled particulate emissions range from
9.4 to 22 pounds per ton of steel produced (Reference 16).  The concentration
of the particulates in the gas stream varies over a wide range during a heat
with 0.7 grains per standard cubic foot being an average value (Reference 24).

     The small particle size of the particulate emissions necessitates the
use of a high efficiency collection device.  As Table 3.5 indicates, electro-
static precipitators, venturi scrubbers, and combinations of these devices
are used to control particulate emissions from open hearth furnaces.  Host of
the air pollution control systems currently in operation have a common mani-
fold to distribute the gas flow from several furnaces evenly among the dif-
ferent control devices.  Electrostatic precipitators have demonstrated over
98% collection efficiency (Reference 24).  Outlet dust loadings of less than
0.01 grains per standard cubic foot can be achieved with a venturi scrubber
operating with a pressure drop of over 40 inches of water column (Reference
25).  There is no NSPS for open hearth furnaces brt there are standards for
other types of reverberatory furnaces.  Primary lead, brass and bronze, and
copper and secondary lead reverberatory furnaces have NSPS for particulate
emissions of 0.022 gr/dscf.

     The gases leave the furnace melting chamber at about 3000°F and are
cooled to about 1300°F as it passes through the regenerative chamber (Refer-
ence 20).  Where an electrostatic precipitator is to be used, the gases are
passed through a waste heat boiler to cool them to a temperature of 500 to
600°F (Reference 20).

     Fugitive emissions from the furnace occcur during molten metal trans-
fer, furnace charging, tapping, and flame reversal as previously mentioned.
Although the potential environmental impact of these emissions are a major
concern of burning hazardous waste, insufficient information was f^und to
permit their quantification.

     Two methods of injecting liquid HWDF are possible.  The method requiring
the least modification to the fuel handling and firing system would entail
blending the waste with the other liquid fuels being fir^a.  This required
that the waste be compatible with other liquid fuels bexng used.   Provisions
                                     3-35

-------
for switching from waste containing fuel to a conventional fuel during opera-
tions generating significant fugitive emissions (furnace charging and flame
reversal) is deemed necessary.

     Another pausible Liquid waste injection method is to provide a separate
fviei handling system and burner gun fcr the HWDF.  A waste burner guns could
be inserted into the furnace adjacent to existing guns (one at each end of the
furnace).  This approach allows independent control of the waste and other
fuel firing.  Separate fuel trains and burners are needed to fire wastes that
are incompatible with conventional fuels.

Compatible Wastes

     Burning HWDF in an open hearth furnace cot-Id adversely affect both the
furnace and the steel quality if certain materials are not avoided.  As with
blast furnace, halogenated solvents and waste containing alkali metals, lead
and tin should be burned in limit-id quantities because they could reduce the
life of the refractory lining.  Uamage to the refractory which results in
loss of operating time could easily offset the fuel savings realized from
burning hazardous wastes.  The burning of halogenated solvents would also
generate hydrochloric acid which could cause serious corrosion of the air
pollution control equipment.

     The impact on steel quality would be from metals and inorganics in the
waste as the organic compounds should essentially be destroyed at the high
temperature of over 3000°F and residence time of 2 to 3 seconds (Reference
23) common for an open hearth furnace.  These materials would be partitioned
between the flue gases, slag, and molten metal.  Because the flare is di-
rected over the surface of the furnace hearth it is expected that little of
these materials would be absorbed in the slag and metal.  Those that were
absorbed, however, should be distributed between the slag and metal as shewn
in Table 3.7 based on their ease of being oxidized relative to iron.  Since
steel properties are determined by its chemical composition, the burning of
wastes containing thos
-------
                                  TABLE 3.7

      DISTRIBUTION OF WASTE CONTAMINANT ELEMENTS BETWEEN SLAG AND METAL
Group I   - Elements almost completely taken up by slag:  Si, Al, Ti, Zn, B,
            V, Ca, and Mg

Group II  - Elements distributed between slag and metal:  Mn, P, S, and Cr

Group III - Elements almost completely in solution in metal:  Cu, Ni, Sn, Mo,
            Co, W, As, and Sb

Group IV  - Elements out of slag and metal:  Zn, Cd,  Pb, Ag, and Hg
Source:  Reference 23.
                                     3-37

-------
and has becomes contaminated with water, organics, inorganics, etc.  The acid
drops in strength and is no longer usable for its intended purpose.  In a re-
covery furnace, the spent acid is thermally decomposed at elevated tempera-
tures into SC>2, carbon monoxide, carbon dioxide and water vapor.  Heat needed
for the acid decomposition is supplied by burning fuels, typically natural gas,
oil or a liquid or gaseous waste stream.  The SO2 generated in the recovery
furnace ranges from 0-14% of the exhaust gases.

     The conversion of the S(>2 to sulfuric acid follows the contact process
as depicted in Figure 3.9.  The SC>2 is oxidized to 803 in a multi-stage con-
verter, typically having either three or four stages depending on the age of
the plant and overall conversion efficiencies desired.  The converting from
SC>2 to SO^ requires very controlled reaction temperatures for efficient con-
version.  The exhaust gases from the combustion chamber are cooled by a waste
heat boiler prior to entering the converter.  After passage through each stage
of the converter, the gases are cooled by waste heat boilers or super heaters
to maintain the desired temperature ranges.  The stages of the converter each
have a bed of solid vanadium pentoxide catalyst.  The conversion rate drops as
each stage of the converter has less SO2 than the previous stage.  Gases leav-
ing the converter are again cooled and are introduced into the 803 absorber
at about 450°F.  If oleum is to be produced, the temperature is lowered fur-
ther prior to entry into the oleum tower.

     The 803 absorber is similar to the drying tower used for the combustion
air.  The 803 is absorbed in sulfuric acid which is continuously drawn-off
as the product (Reference 26).  The sulfur dioxide, sulfur trioxide and sul-
furic acid remaining in the exhaust gases must be reduced to meet air pollu-
tion standards.

     Since the enactment of these strict air pollution regulations in 1971,
most new plants use a dual absorbtion process to lower the SO2 emissions.
The primary 303 absorber, known as the interpass absorber, is used as the
gases leaving the third stage of the converter (Reference 27).  The lower
partial pressure of the 803 in the gases returning to the fourth stage in-
creases the conversion rate and thereby lowers SO2 emissions.  The second
803 absorber is similar to the single absorber system but only absorbs 803
from the gases exiting the fourth stage of the converter.

     Plants built prior to 1960, generally had three stage converters and
overall conversion efficiencies of 95-96% (Reference 27).  In modern four
stage converter contact process plants the overall conversion efficiencies
approach 99%+.

     If oleum is produced at the acid recovery plant, the 803 leaving the
converter is cooled further and is introduced to an absorption tower with
100% sulfuric acid.

     The recovery furnace is usually a refractory lined cylindrical chamber
equipped with a burner designed to spray the spent acid into the chamber
without flame impingement on the walls.  The design of these furnaces vary
according to plant production capacity, but can be 40 feet long by 14 feet
in diameter.  The combustion air is sometimes dried before entry into the
burner in order to prevent corrosion (acid mist formation) of th«s process
equipment by the 803 combining with the water vapor to form sulfuric acid.


                                     3-38

-------
OJ


bJ
10
                                     [-SPENT A:ID
                                     -SULFUR
                                     -FUEL Oil.
                        WATER -

                         STEAM
                                               PiJS
                                       FURNACE     OUST  IMTEHEAT   GAS     GAS   ELECTROSTATIC
                                                COLLECTOR  BOILER   SCRUBBER  COOLER  PRECIPITATOnS
                                                                                                 AIR

                                                                                                11
                                                               T
                                                                S02
                                                               STRIPPER
                                                                                            [ATMOSPHERE
                        HEAT EXCHANGERS     CONVERTER
                               DRYING
TORER
ABSORPTION
                                   ACID PUMP TANK COOLER
                                       ACID COOLERS   lit ACID PUMP TANK
                             FIGURE 3.9 CONTACT PROCESS 8ULFURIC ACID PLANT BURNING SPENT ACID
                                                                                                   REFERENCE  26

-------
Since spent acid have contaminants such as water, organics and inorganics,
they have the potential to corrode and plug the converter beds and other
downstream equipment.  It is reported that spent acid usually contains about
90% sulfuric acid, 4 to 5% water, and 5 to 6% hydrocarbons (Reference 28).
The gasss are cleaned and dried after the combustion/waste heat boiler by
cyclones, electrostatic precipitators, scrubbers- and/or gas cooling towers.

Combustion
     The combustion of spent acid needs to be balanced to achieve the de-
sired range of 8-14% SO2 in the exhaust gases.  This can be accomplished by
fuel blending and combustion air adjustments.  It is important to note that
as the percentage of SO2 drops, more exhaust gases need to be processed to
create the same amount of acid (i.e., the plant exhaust gases with SO2 at 4%
in the combustion gases, will be 2.5 times that of SO2 at 10% in the combus-
tion gases) (Reference 27).  Therefore, burning multiple fuels with/and with-
out significant oxygen contents, makes the efficient combustion system very
complex (Reference 29).  The contro.1 of combustion relies on both tempera-
tures and oxygen controls.

Emission and Air Pollution Control

     The exhaust gases from a spent acid recovery furnace are processed to
sulfuric acid before being emitted to the atmosphere.  Therefore, furnace air
emissions are referenced to the tail and of the entire acid plant.  The major
pollutants emitted from a spent acid plant are SO2 and sulfuric acid mist.
The quantity of SO-, emitted is an inverse function of the sulfur conversion
efficiency (SO oxidized to 803).  This conversion i-3 always incomplete, and
is affected by the number of stages in the catalytic converter, the amount of
catalyst used, temperature and pressure, and the concentrations of the reac-
tants (SO2 and 02).  Host single absorption plants have SO2 conversion effi-
ciencies ranging from 95-98%, which corresponds to 26 to 55 Ib SO3 emitted
per ton of 100% H2SO4 produced.  The EPA performance standard for new plants
is 4 pounds SC>2 per ton of 100% H2SO4 produced which corresponds to a conver-
sion efficiency of approximately 99.7%.

     Nearly all the acid mist emitted from the spent acid plant is in the ab-
sorber exist gases.  Acid mist is created when SO3 combined with water vapor
at a temperature below the dew point of 503.  Once formed within the process
system,  this mist is so stable that only a small quantity can be removed in
the absorber.  The water vapor is produced from the water in the spent acid
and products of combustion of the fuels and hydrocarbon impurif'.as in the
spent acid.-  The strength of the acid produced also affects mist emissions.
Plant producing higher concentrations produce greater quantities of finer
more stable mist.  Another greater quantities of finer more stable mist.
Another factor affecting the mist emissions is the operating temperature of
the SO3 absorption tower.  The optimun absorber operating temperature depends
on the strength of the acid produced, throughout rates, inlet 803 concentra-
tions.  Typically, uncontrolled mist emission from a spent acid plant are 2.2
- 2.4 Ib/ton acid produced (Reference 15).  The EPA New Source Performance
Standard is 0.15 pound per ton of acid produce which requires the applica-
tion of control devices which are discussed later in this subsection.
                                     3-40

-------
     When the spent acid and fuels burned in the recovery furnace, particu-
lates, metal, and trace quantities of other pollutants may also be generated.
The mature of these pollutants will depend on the type of fuel burned and the
contaminants in the spent acid.  Those polluants that might plug or deactivate
the converter catalyst must be removed before the furnace exhaust gases enter
the converter.  Thus air pollution control equipments are used in spent acid
plants for pre-converter cleaning of the  furnace exhaust gases (for particu-
late matter, metals, acid mist, etc.) and to the process exhaust gases prior
to their being released to the atmosphere.

     The pre-converter control can be:

     o  Cyclones
     o  Scrubbers
     o  Electrostatic precipitators
     o  Dryers

     The emission controls used to meet NSPS are:

     o  Sulfur dioxide control - dual absorption systems
                               - sodium sulfate to sodium bisulfate
                                 scrubber
                               - ammonia scrubber

     o  Acid mist - electrostatic precipitator
                  - mist eliminators
                  - packed bed scrubbers
                  - molecular sieves

     The pre-converter controls are well suited to remove particulate matter,
vaporized metals and hydrochloric acid emissions.  All of these parameters
would have operational significance if allowed to contaminate or plug the
catalyst beds in the converter.  While not all spent acid plants have all
of these control devices, they must have adequate controls to prevent the
expensive replacement of the catalyst.

Compatible Wastes

     Hazardous waste burned in a spent acid recovery furnace as fuel must
not adversely affect the acid production process equipment nor significantly
alter the arid quality.  The pre-converter cleaning devices can remove some
hazardous waste contaminants that are not destroyed during the combustion,
such as metals.   However, because of the possibility that some portion of
metals reaching the converter,  operations are lively to restrict the level
of metals in the waste fuel.  They would also likely limit the amount of
chlorinated wastes burned as fuel to avoid potential conversion problems
in furnace and prevent adversely affecting the acid quality.
                                     3-41

-------
A3PHALT CONCRETE PLANTS

General

     The bituminous concrete industry composes some 3,000-4,500 plants  (Refer-
ences 30 and 31 ) r-.-ross the United States.  Their locations follow the  general
population and number of motor vehicles patterns since their product is used
in population/vehicular flated ways  (i.e., roads, parking lots, driveways).
Approximately 60% are stationry, 20% are mobile and 20% are somewhat trans-
portable (Reference 30).

     Bituminous concrete is a blended mixture of asphaltic cement (AC)  and
aggregate.  The variations of the types of mixes are numerous with different
asphaltic cement blends and different size and material combinations, of ag-
gregates (i.e., sand, rock, glass, asbestos, reprocessed asphalt pavement).
The oasic asphalt concrete manufacturing process is composed of:

     o  Heating of the aggregate to drive off moisture and to facilitate
        mixture with the AC.
     o  Screening of the aggregate.
     o  Blending the aggregate with heated AC and mixing for a homogeneous
        product.

The actual plant designs are grouped into three subsets:

     1.  Batch-mix plants - separates the heating, screening, and blending
         production steps.  Mixes one batch at a time; rotary dryer is
         started and stopped many times during the day.  Represents 91%
         (Reference 30) of reported population.

     2.  Drum-mix plants - sizes the aggregate first, then combines the
         heating and blending steps.  Usually runs for longer periods than
         a batch plant.  Represents 2.5% (Reference 30) of the reported
         population but is the majority of all new plants.

     3.  Continuous-mix plants - similar to a batch-mix design but runs con-
         tinuously for long periods based on constant demand.  Represents
         6.5% (Reference 30) of the reported population.

Process Description

     For the context of burning HWDF's, the three subsets can be looked at
as two groups;  rotary dryer heating of the aggregate only (typical batch-
mix plant) and rotary dryer heating and mixing the aggregate and AC (typical
drum-mix plant).  The continuous-mix design has the same combustion charac-
teristics as the batch-mix desig:.

     Figure 3.10 shows the schematic of a typical batch-mix plant.  Agc,jegates
are stored in outside piles of graded stone, sand, etc.  The aggregates are
fed into hoppers and then proportioned based on mix design, onto a conveyor
belt to be fed into the rotary dryer.  The aggregate is heated to about 300-
350°F depending on mix requirements and then moved by a hot elevator to the
screens for final sizing.  The screens discharge into the pug mill where the
hot aggregate is blended/mixed with the hot AC.  After mixing the batch is
                                     3-42

-------
                                                                               EXHAUST TO
                                                                               ATMOSPHERE
                                                         PRIMARY OUST
                                                          COLLECTOR
                                                           -J
                             COARSE
                            AGGREGATE
                             STORAGE
                              PILE
         FINE
      AGGREGATE
LOADER! STORAGE
         PILE
i
£.
Ul
                                           BINS
                               FEEDERS—£i   S—-
                                       CONVEYOR^-*
                                                                                              AC INJECTION
                                                                                                 TORAGE
              ^p QA8 FLOW
                                 FIGURE  3.10 TYPICAL BATCH-MIX ASPHALT CONCRETE PLANT
                                                                                                         REFERENCE 33

-------
dropped into a dump truck for transporting to the job.  Typically, several
batches are required to fill a waiting truck and the system  (i.e., conveyor,
dryer, screens, pug :nill) only operates on a per truck basis.  Recent tech-
niques have been applied to this operation that allow:

     o  If the requirement for a certain mix is large (roadway paving/resur-
        facing), a heated storage silo is used to run the system for longer
        intervals and still allow for making other batch mixes.

     o  Recycling of old asphalt pavement (RAP) is performed for economics
        of the roadway project and plant operations, as well as, reducing
        resurfacing pavement height and total pavement weight.  The previous
        layer of asphalt concrete is removed and taken to the plant.  It is
        crushed and sized as regular aggregate but stored separately.  It
        is usually heated in an auxiliary dryer by the hot exhaust gases of
        the rotary dryer and combined with the hot aggregate for mixing with
        the AC in the pug mill.

     The rotary dryers are designed to quickly heat the rock and drive off
most of its surface and pore moisture that would interfere with the proper
bonding with the AC.

     The fuel being combusted (gas, oil, waste oil, HWDF) in the burner, is
converted into heat and exhaust gases.  Outside the actual flame, the gases
contact the cold stone and quickly drop in temperature to about 350"F or less.
The desired flame pattern is short and wide but without impingement on the
sides of the dryer.

     The typical drum-mix plant is shown in Figure 3.11.  The aggregate is
taken from the storage piles and is metered to the conveyor system for speci-
fic mix requirements.  In die rotary dr^er,  the AC is sprayed onto the aggre-
gate just outside of the flame zone (ther-a are several variations to this AC
injection method).  The aggregate's moisture is driven off and the AC and ag-
gregate are mixed by the rotating action of the drum dryer.  No hot screens,
hot elevators or pug mill are needed with the drum-mix plant.  Due to its
design of mixing in the dryer, efficient operation dictates more continuous
production.  Therefore, sto: i~ge silos for the asphalt concrete are typically
found at all drum-mix plants.  Also, when RAP is used at these plants, it is
introduced into the drum after the flame zone.

     The drying/mixing zone of the drum-mix dryer, is similar to the dryer
of a batch-mix plant but modified to prevent flame impingement on the AC.
The exhaust gas temperatures of a drum-mix plant can be slightly higher than
a batch-mix plant as are the temperatures of the asphalt concrete, to compen-
sate for additional heat loss in storage.

Combustion

     The combustion zones in both the batch-mix and drum-mix dryers are simi-
lar.  The flame is short and releases sufficient heat to raise the aggregate
to approximately 300°F.  The burners use sufficient excess air to ensure
proper combustion of the fuel and the desired aggregate temperature.  Resi-
dence time at combustion conditions (>1,500°F) is short since the wet aggre-
gate cools the exhaust gases very quickly.  The burners operate efficiently
                                     3-44

-------
U<
                                         AQQnF.QAIE
                                        AND SAND FEED
                                           POMI
                      AQOHEQAIE
                     AND 8ANO FEED BINS
                                    CON VE von
                                FIGURE 3.11 TYPICAL DRUM-MIX ASPHALT CONCRETE PLANT
                                                                                                     REFERENCE   33

-------
for their normal fuels and have the potential to adequately combust some haz-
ardous waste derived fuels.

Emissions and Air Pollution Controls

     The New Source Performance Standards (NSPS) (Reference 32) for asphalt
concrete plants are as follows:

     1.  Particulate emissions - <_ 0.04 gr/dscf
     2.  Opacity - £ 20%

     The standards apply to those planes constructed or modified as of June
'1, 1973 (estimated to be 15%  (Reference 30) of all asphalt concrete plants).

     The compliance with the NSPS particulate emission standards typically
requires the use of:

     o  Baghouaes
     o  High energy scrubbers

     The typical batch-mix plant can meet the NSPS standards by use of bag-
house or scrubber, however, due to energy requirements, baghouses predominate
(Reference 30).

     The typical drum-mix plant uses a scrubber or baghouse (Reference 30)
but the use of scrubbers is prevalent due to the potential blinding of bag-
house filters with condensed hydrocarbons from the AC.  It has been reported
(Reference 33) that some drum-mix plants do not utilize any controls due to
the inherent particulate control of the AC as it forms a collection surface
for particles generated from the aggregate.   However, the AC and the heat in
the drum produces a hydrocarbon emission, that condenses and cause condensi-
ble particulate emissions and plume opacity (Reference 30).

     Plants built prior to 1973 are principally the batch-mix or continuous-
mix design and use air pollution controls such as:

     o  Cyclones
     o  Low energy scrubbers

     The typical emission regulations for these older plants are process
weight related and allow partial late emissions an order of magnitude greater
than NSPS.   The typical process weight derived allowable particulate emission
rate for a 100 ton/hour plant is 30 Ibs/hour (approximately 0.1 gr/dscf).

     Since asphalt concrete plants can burn various types of fuel and not
affect product quality, they have been users of waste oils and other blended
wast4 derived fuels.  A 1985 study (Reference 34) reports that about 80% of
all the purchased waste oil derived fuel in California were burned in 1984
by industries in the Standard Industrial Code 2951  - Paving Mixtures and
Blocks.  The majority of this code index group are asphalt concrete plants.
                                     3-46

-------
Compatible Wastes

     Most asphalt dryers could easily be adapted  to fire a liquid hazardous
waste since most are oil - or combination gas anc oi_ - rired units.  The
major compatibility problem would  likely be- the potential equipment corrosion
assoicated with firing highly halogenated wastes.  Halogen composition  speci-
fications would be required to protect the dryer  and associated air pollution
control equipment.

     Little adverse impact on product quality is  anticipated from firing haz-
ardous waste.  Organic compounds would likely be  destroyed (see test results
below) sufficiently by the combustion process that they would not significant-
ly alter the product composition and quality.  Metals entering the aggregate
from the firing of hazardous waste will be insignificant, with a few excep-
tions, compared to the levels naturally occuring  in the aggregate as dis-
cussed in the test results oresented below.
CHARACTERIZATION OF FURNACES FOR DISPERSION MODELING

     The following descriptions of furnaces, which later will be used for
dispersion modeling, were assembled from data taken from a number of sources
and are intended to be representative of the industry recognizing that the
selected plants may not exist anywhere.  They are intended to provide a basis
for analysis of the environmental impacts of burning hazardous wastes on spe-
cific segments of the industry.  The emission and stack data were used in
atmospheric dispersion models for the risk assessment and the cost analyses.
Only the furnace facilities are described; the remainder of the plant facil-
ities are not affected by burning hazardous waste except in a very limited
manner (i.e., coal mills might operate less and result in lower maintenance
and operating costs).  For selection of the lime kiln and the lightweight
aggregate kiln, this analysis utilized the same facilities used by EPA in
its NSPS analysis.  The other furnaces were selected to be representative
of the industry.

Wet Process Cement Kiln

     The typical wet kiln has a production capacity of 850 TPD for an annual
production of 266,500 tons.  Primary fuel for the kiln is coal, required in
the amount equivalent to 5.5 x 10^ Btu/ton.  Using a heating value of 12,500
Btu/pound, 7% ash, and 1.7% sulfur on an as-fired basis, the kiln will burn
7.5 tons/hour of pulverized coal.  The kiln itself is 500 feet long, 11.5
feet in diameter, and rotates at 90 revolutions per hour (rph).  Solid resi-
dence time is 2 to 2.5 hours.

     The exhaust gases from the kiln contain 25% (by volume) water vapor, 29%
CO2,  455 ppm SO2, 265 ppm NOX, and 100 ppm CO.  The gas discharges at 520°F
into a multiclone at the rate of 150,300 acfm (60,720 dscfm).  Following the
multiclone is a 4-field, 2-chamber ESP which ramoves 98.4% of inlet dust load-
ing and the stack discharges 137,700 acfm at 438°F (60,720 dscfm).  Table 3.8
summarizes the kiln stack emission parameters used for the health effects
modeling.
                                     3-47

-------
                            TABLE 3.8

           EMISSION CHARACTERISTICS OF A REPRESENTATIVE
                     WET PROCESS CEMENT KILN
      Parameters
             Values
Stack Height

Stack Diameter

Stack Gas b'low Rate


Stack Gas Temperature

Emission Rates

     Particulate Matter

     Sulfur Dioxide

     Nitrogen Oxides

     Carbon Monoxide
            170 feet

             11 feet

          137,700 acfm
           60,720 dscfm

              4388F
0.11 gr/acf

 455 ppm

 265 ppm

 100 ppm
130 Ibs/hour

365 Ibs/hour

 93 Ibs/hour

 35 Ibs/hour
                               3-48

-------
Dry Process Cement Kiln

     The configuration for  the model dry process kiln is selected both  to  re-
present a stack/control device system different from the wet process plant
and to approximate a typical dry process; a 4-stage suspension preheater and
a rotary kiln 9 feet in diameter and 30C feet long.  Capacity of the kiln  is
394,000 TPY or 1,270 TPD of clinker.  As with the we', process kiln, primary
fuel is coal which is fed to the kiln at 6.4 tons per hour based on a heat
requirement of 3 x 10^ Btu/ton clinker aud coal heating value of 12,500 Btu/
Ib, 7% ash, and 1.7* sulfur.

     The kiln exhausts 154,900 acfm (104,COO dscfm) of gas at 320°F to a
eyelone/fabric filter particle removal system.  The control system removes
99.8% of the total dust load and discharges 15 Ibs/hour of particulate mat-
ter through a 9-foot diameter stack at 120 feet above the ground.  These
data and other emission information are summarized in Table 3.9.

Lime Kiln
     Basically, the model kiln presented here is the same configuration used
by EPA in its analysis of the lime industry for the NSPS.  The industry has
not changed significantly in teras of size of technology since that review
in 1977  (Reference 17).

     The model lime kiln has a capacity to produce 500 TPD of quick lime from
1,000 TPD of limestone.  Fuel is 3% sulfur coal with a heating value of 12,500
Btu/lb, and for a requirement of 7 x 106 Btu/ton of lime, is burned at a rate
of 140 TPD.  Dust loss from the kiln is considered to be 10% of the kiln feed
rate or 8,300 Ibs/hour.

     The kiln exhausts this dust in 41,000 dscfm of flue gas at a temperature
of 700°F with 14% (volume) water vapor.  The gas contains a maximum of 1500
ppm sulfur oxides, about 150 ppm NOX, and less than 50 ppm CO.

     Particulate matter removal is provided by fabric filter which exhausts
300°F gas to the atmosphere through a short 3-foot diameter stack on top of
the baghouse.  Discharge elevation is 80 feet (Table 3.10).

Lightweight Aggregate

     The model lightweight aggregate kiln presented here is the same as that
used by EPA in its analysis of the lightweight aggregate industry for NSPS
(Reference 21).  This kiln has a production capacity of 1000 cubic yards per
day (840 tons/day).  It is fueled by pulverized coal at a rate of 157.5 tons/
day for an assumed heat requirement of 4.5 x 106 Btu/ton of aggregate.  The
coal has a heating value of 12,000 Btu/lb and a sulfur content of 4%.

     The model kiln has the following emission characteristics.  Uncontrolled
particulate emissions from this kiln are 842 Ibs/hour.  A wet scrubber is used
to reduce these emissions by 99% to 8.42 Ibs/hour.  Uncontrolled S02 emissions
are 238 Ibs/hour.   There is essentially no removal of SO2 by the wet scrubber.
                                     3-49

-------
                            TABLE 3.9

           EMISSION CHARACTERISTICS OF A REPRESENTATIVE
                     DRY PROCESS CEMENT KILN
      Parameters
              Values
Stack Height

Stack Diameter

Stack Gas Flow Rate


Stack Gas Temperature

Emission Rates

     Farticulate Matter

     Sulfur Dioxide

     Nitrogen Oxide

     Carbon Monoxide
             120 feet

               9 feet

           154,900 acfm
           104,000 dscfm

               320°P
0.011 gr/acf

  520 ppm

  310 ppm

  100 ppm
 15 Ibs/hour

540 Ibs/hour

140 Ibs/hour

 45 Ibs/hour
                               3-50

-------
                             TABLE 3.10

       EMISSION CHARACTERISTICS OF A REPRESENTATIVE LIME KILN
      Parameters
             Values
Stack Height

Stack Diameter

Stack Gas Flow Rate


Stack Gas Temperature

Emission Rates

     Particulate Hatter

     Sulfur Dioxide

     Nitrogen Oxides

     Carbon Monoxide
             30 feet

              3 feet

           67,000 acfm
           41,000 dscfm

              300°F
0.073 gr/acf

 1500 ppm

  150 ppm

   50 ppm
 42 Ibs/hour

615 Ibs/hour

 44 Ibs/hour

  9 Ibs/hour
                                3-51

-------
     The exhaust gas from the scrubber is vented to the atmosphere through a
105-foot, stack that has a 4-foot inside diameter at the exit.  This gas leaves
the stack at 160°F and a flow rate of 49,980 scfm (Table 3.11).

Blast Furnace
     The model blast furnace selected for this study is 102 feet tall with
a hearth diameter of 27 feet and normally produces 3,000 tons of molten iron
per day.  The coke rate used is 900 pounds per ton of hot metal.  This is
supplemented by 150 pounds of fuel oil per ton of hot metal.  Liquid hazard-
ous waste may be substituted for the fuel oil in any proportion for purposes
of the regulatory impact analysis.  Hot blast air is provided at the rate
of 86,000 standard cubic feet per minute and at a hot blast temperature of
2,000°F.  During the production, 100,000 standard cubic feet per minute of
off-gases are generated.

     The off-gases from the furnace contains 25.4% carbon monoxide, 12.5%
carbon dioxide, 3.5% hydrogen, on a dry volume basis.  Moisture is 1.45% on
a volume basis.  These gases leave the blast furnace and pass through a dust
removal system consisting of mechanical collectors followed by a venturi
scrubber.  The gases leaving the cleaning system has dust concentration of
0.01 grains per cubic feet.

     Forty-one percent of the blast furnace off-gases are used to fire the
blast air stove with the remainder being used to fire a boiler.  Assuming
that 15% excess air is used in combusting the 100,000 scfm of off-gases,
160,000 scfm of combustion products would be generated; 65,600 scfm in the
blast air stoves and 94,400 scfm in the boiler.  The 65,600 scfm of combus-
tion products generated in the stoves are exhausted to the atmosphere at
an average temperature of 500°F through a 200 foot stack that is 4 feet in
diameter.  The boiler combustion products (94,400 scfm) will exit through
a 100-foot stack at 350°F.  This stack is 10 feet in diameter.

Open Hea rth^ Furnace

     The model open hearth furnace selected for this study has a capacity of
320 tons of steel per heat with a tap-to-tap time of 8 hours.  Thus, it can
produce 960 tons of steel per day.  It is fueled by natural gas and oil in
approximately equal quantities on a heating value basis.  The heat require-
ment for steel production in the furnace is '3.5 x 106 Btu per ton of steel
produced.

     The model furnace has the following emission characteristics.  Uncon-
trolled emissions from the furnace is 16 pounds per ton steel produced or
640 pounds per hour.  An electrostatic precipitator is used to reduce these
emissions by 98% to 12.8 pounds per hour (0.02 gr/dscf).

     The exhaust gas from the electrostatic precipitator is vented through
a 150 foot stack that has an 8-foot inside diameter at the exit.  This gas
leaves the stack at 400°F and a flow rate of 65,000 standard cubic feet per
minute.
                                     3-52

-------
                             TABLE 3.11

            EMISSION CHARACTERISTICS OF A REPRESENTATIVE
                     LIGHTWEIGHT AGGREGATE KILX
      Parameters
                                              Values
Stack Height

Stack Diameter

Stack Gas Flow Rate


Stack Gas Temperature

Emission Rates

     Particulate Matter

     Sulfur Dioxide

     Nitrogen Oxides

     Carbon Monoxide
            105 feet

              4 feet

           67,200 acfm
           49,980 dscfm

              160°F
0.015 gr/acf

  475 ppm

  150 ppm

   50 ppm
8.4 Ibs/hour

238 Ibs/hour

 54 Ibs/hour

 11 Ibs/hour
                                3-53

-------
Spent Acid Recovery Furnu".e
     The model spent acid recovery plant has been developed with the follow-
ing characteristics:
     o  Production Rate-.

     o  Fuels:
     o  Exhaust Temperature:

     o  Exhaust Flow:

     o  Stack Height:

     o  Stack Diameter:

     o  Air Pollution Controls:


     o  Emission Rate:



Asphalt Aggregate Kiln
720 tons/day of sulfuric acid

252 tons/day sulfur total
84 tons/day elemental sulfur
126 tons/day spent acid
16,800 tons/day HWDF with 50% S and 50%
   organics

90°F

20,000 scfm

75 feet

4 feet

packed tower
dual absorption system

.08 gr/dscf particulates at 7% oxygen
4 Ibs/hr HCX
99.99% ORE
     The model asphalt concrete plant selected for the engineering cost anal-
ysis and the risk assessment is as follows:
     o  Plant Type:

     o  Production Rate:

     o  Fuel:


     c  Operational Rate:
     o  Stack Characteristics
           Temperature:
           Flow rate:
           Moisture content:
           Diameter:
           Height:

     o  Air Pollution Control:
batch-mix plant

200 tons per hour

300 gallons per hour of fuel oil or
waste derived fuel

5 hours per day
5 days per week
40 weeks oer year
150'F
20,000 >ofm
25%
4 feet
25 feet

low energy scrubber (spray bars, damper,
wet fan)
                                     3-54

-------
o  Emission Rates:              .08 gr/dscf particulates at
                                   7% oxygen
                                4 to 30 Ibs/hr HC1
                                99.99% ORE
                                3-55

-------
                                  REFERENCES
1.   Shreve, R.N.  Cheniical Process Industries.  McGraw-Hill,  1962.

2,   Addendum to the Draft Final Report:  Source Category Survey of the Clay
     and Fly Ash Sintering Industry.  Prepared by Midwest Research Institute
     for the U.S. Environmental Protection Agency.  Contract No. 68-02-3059.
     Draft Report.  April 17, 1980.

3.   Perry, R.H. and Chiiton, C.H.  Chemical Engineers Handbook.  Fifth Edi-
     tion,  McGraw-Hill, 1977.

4.   Hazelwood, D.L. and Smith, F.J.  Assessment of Waste Fuel Use in Cement
     Kilns.  Prepared by A.T. Kearney, Inc., for the U.S. Environmental Pro-
     tection Agency.  Draft Report.  March 1981.

5.   Feasibility of Using Lime Kilns to Burn Hazardous Wastes.  Prepared by
     A.T. Kearney, Inc., for the U.S. Environmental Protection Agency.  Draft
     Report.  February 1981.

6.   Standards Support and Environmental Impact Statement Volume I:  Proposed
     Standards of Performance for Lime Manufacturing Plants.  U.S. Environ-
     mental Protection Agency, Office of Air Quality Planning  and Standards,
     Research Triangle Park, North Carolina.  EPA-450/2-77-007a.  April 19V7.

7.   Personal Communication Between Harry Robinson of the Expanded Shale,
     Clay,  and Slate Institute and Robert Patrick of Engineering-Sconce.
     August 13, 1984.

8.   Source Category Survey of the Clay and Fly Ash Sintering  Industry.  Pre-
     pared by Midwest Research Institute for the U.S. Environmental Protection
     Agency.  Contract No. 68-02-3059.  Draft Final Report.  January 30, 1984.

9.   Bieser, C.O.  Identification and Classification of Combustion Source
     Equipment.  Prepared by the Process Research, Inc., for the U.S. Envi-
     ronmental Protection Agency.  EPA-R2-73-174.  February  1973.

10.  Jablin, R., et al.  Pollution Effects of Abnormal operations in Iron  and
     Steelmaking - Volume III.  Blast Furnace Ironmaking, Manual of Practice.
     Prepared by Research Triangle Institute for U.S. Environmental Protection
     Agency.  Publication EPA-600/2-78-118c.  June 1978.

11.  Blast Furnace - Theory and Practice.  Volume T..  Strassburger, J.H.,  et
     al., editors.  Gordon and Breach Science Publishers. New York.  1969.

12.  Directory of Iron and Steel Works of the United States  and Canada.  Pub-
     lished by the American Iron and Steel Institute.  Washington, D.C.  1984.

13.  Greenberg, J.  Industrial Furnaces, Ovens, Kilns, Dryers, Boilers, Incin-
     erators.  Seminar Presented by A.T. Kearney, Inc. to the  U.S. Environ-
     mental Protection Agency, Cincinnati, Ohio.  February 26, 1981.
                                     3-56

-------
14.  Campbell, R.L. Campbell & Pryor Associates Inc.  Letter to M. Benoit,
     Cadence Chemical Resources, Michigan City, Indiana, June 2, 1986.

15.  Compilation of Air Pollutant Emission Factors.  U.S. Environmental Pro-
     tection Agency.  Publication Ho. AP-42.  April 1981.

16.  Katari, V.S. and Gerstle, R.w.  Industrial Process Profiles for Envi-
     ronmental Use:  Chapter 24.  The Iron and Steel Industry.  Prepared by
     Radian Corporation for the U.S. Environmental Protection Agency.  Pub-
     lication EPA-600/2-77-023x.  February 1977.

17.  Kirk-Othmer Encyclopedia of Chemical Technology.   3rd Ed., Volume 13.
     John Wiley & Sons, Inc.  New York, New York.  1981.

18.  Trinks, W. and Mawhinney, M.H. Industrial Furnaces, Volume II.  John
     Wiley & Sons, Inc. New York.  1967.

19.  1983 Annual Statistical Report.  Published by the American Iron and
     Steel Institute.  Washington, D.C.  1983.

20.  Pollution Effects of Abnormal Operations in Iron  and Steelmaking - Vol-
     ume IV.  Open Hearth Furnace, Manual of Practice.  Prepared by Research
     Triangle Institute for the U.S. Environmental Protection Agency.  Publi-
     cation EPA-600/2-78-118d.  June 1978.

21.  Destruction & Removal of POHCs in Iron Making Blast Furnaces.  Prepared
     by Radian Corporation for the U.S. Environmental  Protection Agency.
     December 31, 1985.

22.  Atmospheric Emissions from Sulfuric Acid Manufacturing Processes.  U.S.
     Department of Health, Education and Welfare.  Division of Air Pollution,
     Cincinnati, Ohio.  1965.

23.  Hall,  F.D., et al.  Evaluation of the Feasibility of Incinerating Hazard-
     ous Waste in High Temperature Indust^ia. Processes.  Prepared by PEDCo
     Environmental, Inc.,  for the U.S. Envirorunental Protection Agency.  Pub-
     lication PB 84-159391.  February 1984.

24.  Danielson, J.  Air Pollution Engineering Manual.   U.S. Environmental
     Protection Agency.  Publication AP-40.  May 1973.

25.  Vandergrift, A.E., et al.  Particulate Pollutant  System Study, Volume
     III.  Prepared by Midwest Research Institute for  the U.S. Environmental
     Protection Agency.  May 1971.

26.  New Source Performance Standards; Inspection Manual for Enforcement of
     Sulfuric Acid Plants.  U.S. Environmental Protection Agency.   Office of
     Enforcement, Washington,  D.C.  1977.

27.  A Review of Standards of Performance  for New Stationary Source-Sulfuric
     Acid Plants.  U.S. Environmental Protection Agency.  Office of Air Qual-
     ity Planning and Standards, Research  Triangle Park, North Carolina.   1979.
                                     3-57

-------
28.  Industrial Process Profiles for Environmental Use:   Chapter 23.   Sulfur
     Oxides and Sulfuric Acid.   U.S. Environmental Protection Agency.   Office
     of Research and Development,  Cincinnati,  Ohio.   1977.

29.  Shafer, J.R.,  and Bo<;adi,  J.S.   Producing Sulfuric  Acid in Refineries.
     Chemical Engineering Practices, V.76,  October,  1980.  pp. 70-75.

30.  A Review of Standards of Performance for  New Stationary Sources,  Asphalt
     Concrete Plants.  U.S. Environmental Protection Agency.  Office  of Air
     Quality Planning and Standards.  Research Triangle  Park, North Carolina.
     1979.

31.  NEDS Source Classification Codes,  February 1985 - Population and  Through-
     put Listing.  U.S. Environmental Protection Agency. Office of Air yual-
     ity Planning and Standards.  Research  Triangle  Park, North Carolina.
     1985.

32.  Title  40,  Code of Federal  Regulations,  Part 60, Subpart I - Standards of
     Performance for Asphalt Concrete Plants.

33.  Control Technology Evaluation of the Drum-^tix Process  for Asphalt Con-
     crete  Manufacturing.  U.S. Environmental  Protection Agency.  Environmen-
     tal Research Information Center.  Cincinnati, Ohio.  1978.

34.  Assessment of  Used Solvent and  Used Oil as Fuel in  California. Califor-
     nia Air Resources Board.  January 1985.

35.  U.S. and Canadian Portland Cement Industry:   Plant  Information Summary.
     Portland Cement Association.  May 1983.
                                     3-58

-------
                                  SECTION 4

                    DESTRUCTION AND REMOVAL  EFFICIENCY OF
                  HAZARDOUS MATERIALS BY INDUSTRIAL  FURNACES
     The EPA Office of Solid Waste  (OSW)  recognized  the  need  to  collect  emis-
sion data from industrial furnaces  burning hazardous wastes as fuels  to  sup-
port its regulatory efforts.  EPA therefore developed a  program  to  tesf.  in-
dustrial furnaces burning a variety of waste streams.  This section describes
the analytical methods used to test stack gases from furnaces burning hazard-
ous wastes, identifies those testing procedures that are or should  be stan-
dardized for collecting test burn data, evaluates the QA/QC activities that
are or should be required for each  method and identifies the quality  accept-
ance criteria that were applied to  test burn data collected from seven kilns,
three asphalt aggregate kilns, and  a blast furnace.

     Results from eleven trial burns (ten EPA and one California Air  Resources
Board test) were analyzed to determine if the destruction of hazardous mate-
rials was adequate to protect the environment.  Included in the  analyses and
discussion are:  the DRE, the POHCs, emissions of particles, emissions of  HC1,
PICs, metals emissions, and combustion gases.

     The last portion of Section 4  describes the types of system modifications
that would be required of existing  production furnaces to enable them to safe-
ly burn hazardous wastes as fuels.  Major facility modifications could include:

     o  Installing waste storage and handling facilities
     o  Installing laboratory testing facilities
     o  Installing waste pretreatment facilities
     o  Modifying or adding burner  equipment
     o  Modifying or adding combustion controls
     o  Installing monitors for waste feed rate and stack gases  (CO and 02)
     o  Modifying or adding air pollution control equipment
EVALUATION OF TEST METHODS FOR MEASURING ORGANIC EMISSIONS FROM COMBUSTION
SOURCES

     The most widely employed procedures for measurement of the rates of emis-
sions of organic substances from combustion sources are the Modified Method 5
(MMS), the volatile organic sampling train (VOST) procedures, and EPA Method
23.  The first two methods are conceptually similar; both sampling trains con-
sist of a particle filter, a condenser, a bed of porous polymer sorbent, and
a condensate trap.  Their differences lie in their size, sorbent, and analyti-
cal techniques.  Some of the salient attributes of the two methods are compared


                                     4-1

-------
in Table 4.1.  Method 23 specifies that stack gas samples be drawn into inert
plastic bags at a constant rate by a lung-type sampler.  Analyses are usually
done in the field by GC with either flame ionization or electron capture de-
tection.  The strengths and weaknesses of these test methods are discussed in
this section.

Modified Method 5 (MM5)

     This method is an adaptation of EPA Method 5 (40 CFR Part 60) modified
to obtain samples for organic compound analysis as well as quantification of
particulate matter emissions.  The adaptation (Figure 4.1) is the addition of
a sample gas condenser and a sorbent resin module between the heated filter
and first impinger of Method 5 train.  The sorkxent resin moat commonly used
for hazardous waste combustion evaluation is XAO-2 which is highly effective
at trapping organic compounds with boiling points greater than 100°C (some-
times referred to as semi-volatile organic compounds or S-VOC).

     The sample is collected by isokinetically drawing stack gas through a
heated glass or quartz probe, through a heated glass fiber particle filter
and then to the condenser/resin module.  The sample gas is kept above 120°C
until it reaches the condenser where it is cooled to <20°C.  Filtration tem-
peratures up to 150°C are used to minimize organic species condensation prior
to the condenser if this does not interfere with the determination of particu-
late matter.  The sample gas and condensate pass through a resin bed located
below the condenser allowing the condensate to percolate through the bed and
collect in an impinger or condensate trap underneath the resin module.  The
sample gas is then bubbled through two more impingers in the conventional
Method 5 configuration for acid gas and additional moisture removal.

     Samples are analyzed by performing solvent extractions on the probe and
filter material as one fraction, the resin as a second fraction, and option-
ally on the condensate and impinger catches as third and fourth fractions.
The solvent extracts are concentrated and may be combined and analyzed by gas
chromatography (GC) using mass spectrometers (MS), flame ionization detectors
(FID), or electron capture detectors (ECO), as appropriate, for the organic
compounds of interest.

     MM5 Procedure Standardization

     The construction and operation of a Method 5 train is well known and well
described in the literature.  The train operation, sample recovery, choice of
sorbent resin, and analytical method to be used in a given application of MM5
are not specified.   At this time, a single description of how the MM5 train
is or may be used is not available.  The sample collection aspects including
resin choice, are discussed in various publications, but the sample recovery,
analysis, and data reduction are not well described.

     The Method 5 train and sample collection scheme has become the standard
for measurement of emissions of particulate matter and it is not surprising
that modifications to it have evolved in attempts to quantify other types of
emissions using the same equipment and techniques.  Method 5 is written as a
test procedure for determining compliance with New Source Performance Stan-
dards (NSPS).  This specificity has not been established for the MM5 proce-
dure as applied to hazardous waste ORE sampling in part because at least some


                                     4-2

-------
                                TABLE 4.1




                  COMPARISON OF MM5 AND VOST PROCEDURES
         Feature
        MM5
       VOST
Sampling Rate




Sorbent




Analysis of Condensate




Sample Recovery Technique




Analysis




Boiling Point of Analytes




Sampling Duration
1 4-40 1pm




XAD-2




Yes




Solvent Extraction




GC or GC/MS




>100°C




1 to 4 hours
0.5 to 1 1pm




Tenax*-GC




No




Thermal Desorption




GC/MS




30° to 100°C




0.3 to 1 hours
                                    4-3

-------
Teui|i«f«|ii(f Senior .    | | Stuck Wall

         I'rolw    "~~^
Type I'liut lul>«
                                                                                      lllOlinqinclor

                                                                                             Clirck V.lvt
                  Pilot Manomaler

                       necltciilallon Piiinp
                                      Thciinoineloit
                                                            lm|t(ng
-------
of the measurement objectives will be different from  test  to  test.   Many of
the objectives will be the same and these need to  be  identified  and  the method
written to unsure that the goals are clear.

     First, the relative priorities of  ,ne«suring particulate  n.atter  vind or-
ganic compound emissions need <•<•» be sec.  Most often  these will  not  conflict
and no sacrifice of one measvireuent for another is  needed.  In the caatis
where sorae relatively significant quantity of organic chemical of interest
may be found :.n the "front half" (pvobe, filter, connecting glassware)  of
che Method 5 train then the accuracy of particulate matter mecisurement  may
need to be sacrificed to obtain the S-VX.  Examples  are net  brushing the
probe to avoid contamination, renoving  some particular matter with  the
solvent used for S-VOC recovery, and/o-r operating  the heated  portion of the
train at higher than noraial tenperature (e.g., 205°C)  and  vaporizing parti-
culate matter which would otherwise deposit in the  front t'aLf of the train.
In the case of o^l-firad boilers, much  of the particulate  matter is  carbona-
ceous with little inorganic ash and the material contains  straight chain and
aromatic hydrocarbons which may or may  not be collected in the front half de-
pending on the filtration temperature.  This is different  from a coal-fired
unit where the ash is mostly (typically >95%) inorganic matter.  If  PCHCs or
PICs to be measured from hazardous waste combustion are to include naphthenic
or paraffinic hydrocarbons, then v/hether the boiler is oil- or coal-fired
will probably affect the sampling and analysis scheme.

     What compounds are selected, as POHCs and P'.Cs  also bears on the choice
of a suitable sorbent resin.  Generally, the MM 5 is used to sample for  PICs
- the higher boiling compounds, but it  is als'j used to collect samples  for
certain Appendix VIII compounds, e.g.,  toluene, monochlorobenzene.   XAD-2
is the resin most commonly us-id and recommended as  a  general  purpose sorbent
whea solvent extraction is the means of sample recovery.   XAD-2  was  selected
for its sorbert properties, ease of cleaning, and sample recovery efficiency.
Others are also used, for example, Tenax^-GC, if thermal desorption  is  the
sample recovery procedure.  However, selection of a sorbent resin usually
involves a great deal of time and effort (literature  and laboratory  research)
tnat is usually not practical on a case-by-case basis.
         last major uncertainty in regard to applying the MM5 DRE  testing  is
analyses - compound identification and quantification.  Currently, GC  is used
with one of three detection modes MS, ECO, and FID.  Each has its  own  benefit?
and disadvantages relating to sensitivity, reproducibility, compound identi-
fication, interference rejection, and analytical cost.  Each of these  factors
needs to be considered and the method written to describe when each would  be
appropriate.

     MM5 QA/QC Evaluation

     Most of the QA emphasis has been on obtaining acceptable blank values and
preventing contamination.  The large gas volume and relatively large quantity
of resin concentrates the sample, which in conjunction with the solvent ex-
traction, usually results in a greater analyzable mass than methods employing
a lower sample volume.  This greater analytical mass ten^s to decrease the
importance of trace contamination of the sample.  The resin cleaning,  blank
extraction, field trip, and laboratory handling blanks are adequate to iden-
tify problems and likely causes •
                                     4-5

-------
     There is one important aspect lacking in the method as it is being used
and this is use of field spikes to check on sample loss and recovery effi-
ciency.  As currently practiced, the procedure does not provide a means for
determining t?~get compound collection efficiency or for evaluating sample
recovery.  The nature of the sample and the field conditicns preclude the
usual option.of splitting the collected sample and spiking one split wich
a target compound to check loss and recovery.  These spikfjs and replicate
analyses can be performed in the lab using the extractate with some loss of
sensitivity.  It is also possible, and it seems highly advisable, to spike
the field samples with a tracer compound having properties similar to the
target compounds.

     Because the volume of gas sampled by MM5 is large, the total quantities
of the various S-VOCs in the samples ranged from a few to several hundred
micrograms.  Thus, the MM5 results are not greatly influenced by even a few
hundred nanograms of contaminants.

     Two QA acceptance criteria were applied to the MM5 data before they were
included in this document.

     1.  The recovery of surrogate or spike compounds added to the sample be-
         fore analysis must have been in the range of 50 to 150%.

     2.  The rate•of feed of any given compound must have been 10,000 times
         the minimum detectable limit of the MM5 procedure.

     The first of these is merely a demonstration of acceptable analytical
accuracy.  The acceptable range is wide relative to normally attainable ana-
lytical precision.  It is adequate for this analysis because the ORE calcu-
lation is insensitive to an error of a factor of three in the emission rate
measurement.

     The calculation of ORE, the primary use of the data in this document,
is based upon the concentration of the various constituents in the waste
feed stream and in the exhaust gas.  The accuracy and precision of these
concentration measurements decrease when they are near the limit of detec-
tion of the analysis methods.  Consequently, the dispersion of ORE values
that are calculated based upon these imprecise measurements becomes unac-
ceptably large.  The decision was made to include only those data that are
as accurate as is possible using the available methods.  The method chosen,
the second of the QA acceptance criteria, accomplishes this goal.

Volatile Organic Sampling Train (VOST)

     The basic details of construction and operation of the VOST are described
in the "Protocol for Collection and Analysis of Volatile POHCs Using a Vola-
tile Organic Sampling Train (VOST)1 by Envirodyne Engineers for IERL (Refer-
ence 1).  The highlights of the procedure are described below.  Stack gas is
drawn through a quartz wool particle filter in a glass or quartz probe heated
to approximately 130°C, through a three-way stopcock and through a coil con-
denser.  Following the condenser, the sample gas passes through a glass tube
containing 1.6 g of Tenax*-GC, a condensate trap, a second condenser, and a
second sorbent tube containing 1.0 g Tenax*-<3C followed by 1.0 g of activated
charcoal.  A second condensate trap is next, followed by valves, flow meters,


                                     4-6

-------
gas meters, etc.  All portions of the sample line preceding the last condenser
are glass, stainless steel, or Teflon®.  Figure 4.2 is a schematic depiction
of the train.  Previous experiments have established 20 liters as a maximum
safe sample volume for this train.  A greater sample volume incurs the risk
of stripping sorbed POHCs off the resin.  The sample rate for these tests was
1/2 liters per minute for a total of 40 minutes per pair of tubes.

     An ice water batn is used to circulate water through both condensers to
maintain sample gas temperature below 20°C through the sorbent tubes.  Tem-
perature of the probe liner, first condenser outlet, ambient air, and dry gas
meter are measured and recorded.  Leak checks of the whole train and each pair
of sorbent tubes for each run are conducted and the resulting vacuum is re-
leased by allowing ambient air in through a charcoal filter connected to the
three-way valve.

     Tubes used for the kiln tests were of the inside-inside design that are
held in the sample train with stainless steel Swagelok® fittings and ceramic-
filled Teflon* ferrules.  Other samplers have used the inside-out-inside de-
sign; a double walled sorbent cartridge/shipping container that uses O-rings
and end caps to seal the cartridges.  Stainless steel caps are used t.o seal
tube ends for shipment before and after sample collection.  After sample col-
lection, tubes are kept and shipped in chilled styrofoam containers.

     The tubes were analyzed on a GC/MS using thermal desorption with trap
and purge.  The method is described in the protocol and involves spiking
each tube or pair of tubes with an internal standard, thermally desorbing
the tubes into a water trap, and purging the water trap onto an analytical
column for component separation.  Identification and quantification are made
by elution times, characteristic ions, and ion current profile using a com-
puterized data library.

     VOST Standardization of Procedure

     The protocol provides clear and specific directions about the sample
train to be used and the method of sampling and the method of analysis.  Re-
agent preparation, sample handling, QA/QC activities, calibration and calcu-
lations are all described in detail.  The protocol states that conditioned
cartridges, as well as used ones with sorbed sample, be kept in ice water
before use and after sample collection.  This ice water storage is not re-
quired if acceptable blank levels can be maintained.

     Options regarding sample collection and recovery (analysis) efficiencies
are also provided along with evaluation criteria.  The analytical procedure
is also described very specifically.

     VOST OA/gC Evaluation

     The pre-sampling QA activities are clear and direct.  Tenax® and char-
coal cleaning, tube packing, and desorption blanking, provide sufficient as-
surance that the sample cartridges start clean.  The trip blank, field blank,
and lab blanks are intended to provide a history and background levels of
contamination and/or degradation so that the results of the sample analysis
reflect only POHCs present in the stack gas.  This history is especially dif-
ficult to create if the samples and blanks are not analyzed promptly.  Sample
                                     4-7

-------
     Glcitt Wool
     Puillculdlo
     rillor

-------
degradation, tube cross-contamination, contamination from external sources
(lab air, ambient air, etc.), and calibration and response standard degrada-
tion become more likely and less distinguishable.

     As each analysis is a one time occurrence (no way to split), a bad
analysis, a contaminated or otherwise invalid tube, means a lost data point.
However, the sample collection involves three pairs, so duplication is inher-
ent in the sampling.  One could carry this suggestion further to say that two
blanks would be desirable to yield an average value and increase the confi-
dence in the results.

     The protocol requires one exposure pair per six pairs of sample tubes.
These exposure (field or shadow) blanks are opened as if they were sample
tubes but are not installed in the train.

     One QA action which has not been done, for several reasons, is spiking
tubes in the field with one or aore target POHCs or surrogates to establish
recovery efficiencies.  The difficulty stems from two major areas.  One is
the difficulty in maintaining reagent and tube purity in a field environment
and the other is not being able to split a single sampling and spike one
portion.  The methods used to date have centered on lab simulations and the
analytical process.   A suitable field spike procedure would yield data on
sample loss (leakage out) and desorption efficiency as well as additional
data on contamination, lab QA, and overall method validity.

     The following list identifies areas where additional effort and investi-
gation could strengthen the VOST procedures.

     o  Spike blank cartridges in the field with a labeled compound to de-
        tect potential leakage during field storage and subsequent transport.

     o  Analyze the sample immediately with as little storage time in field
        and lab as possible.

     o  Conduct a detailed systematic evaluation of field handling, field
        storage,  shipping, and lab storage, to identify potential causes of
        contamination and/or leakage.  Develop guidelines to detect and pre-
        vent contamination and to leak check cartridges.

     o  Investigate the current seal design.  Do temperature changes cause
        leakage?  How can one be certain the tubes are perfectly sealed?
        How can overtightening/undertightening be prevented?  Can a pres-
        sure/vacuum tight seal be obtained repeatedly on a large number of
        tubes with no failures?

     o  Investigate cross-contamination.  Place spiked samples and clean
        blanks in the same container, store one or two weeks, and analyze.
        Do the above with loose fittings or cracked tubes and observe for
        cross contamination.  Place the sampD.es and lab blanks with their
        double seal in an atmosphere containing trace amounts of methylene
        chloride  or waste fuel vapors, store,  and analyze.
                                     4-9

-------
     QA acceptance criteria were developed for the VOC data produced during
each test.  These criteria could not be identical for all sites since  the
methods used differed.  In all cases, only those components of waste listed
in Appendix VIII  (CFR 40 Pan. 261) were Included in  t^ie ORE results; even
though there may have been other organic constituents measured in the  waste
feed and stack exhaust streams.

     It is noted that not all of the VOC sampling done during these tests
was done by the VOST procedure as has been described.  The train used  to
sample VOC at Site A consisted of a condensate trap  and one Tenax®-GC  trap
in series.  Do condenser or sorbent temperature control was used at Site A.
Impingers were inserted in the train upstream and downstream of the Tenax®-
GC tube at Site C.

     The VOC sampling trains used at Sites A and C varied from the VOST
train in that an impinger (containing water) in an ice bath was inserted in
the sample line ahead of the Tenax*-GC cartridge.  Three QA acceptance cri-
teria listed below were applied to the data collected by these trains:

     1.  If impingers were used, the contents must have been analyzed.
     2.  Both sorbent tubes must have been analyzed.
     3.  At least 70% of the total quantity of any compound found on the
         sorbent tubes must have been found on the first (Tenax*~GC) tube.

     The first two of these are completeness criteria.  While analysis of the
condensate is not normally a part of the VOST protocol, it is necessary in
these instances because of the location and the temperature of the condensate
trap.  The volume of condensate obtained from a 20-liter sample of stack gas
is approximately 1.5 ml.  Even though the compounds  of interest (mostly chlor-
inated hydrocarbons) are normally considered to be insoluble in water  they
are miscible to a small but measurable extent.  A compound soluble to  1 mg/
liter is said to be insoluble yet that translates to 1500 ng/1.5 ml of conden-
sate which is large relative to the analytical quantities of interest.  The
VOST train causes the condensate to be drawn through the resin t*d.  The re-
sin should remove the compounds from the condensate.  Since there was  no con-
tact between the condensate and the resin at these sites, it was necessary to
analyze the condensate.

     The third acceptance criteria was included to eliminate contaminated
samples from the data.  Persons who have sampled surrogate stack gases spiked
with chlorinated )• vdrocarbons under controlled conditions have reported that
at least 90% of '  -se compounds are sorbed on the first resin trap.  This is
not true of hic^ly volatile compounds, e.g., vinyl chloride, but it is for
the compounds of interest in this document.  Tenax*-GC has sufficient  affin-
ity for the? .. compounds to remove them nearly quantitatively from the  sample
gas stre-ir.  The charcoal, used as a back-up sorbent in the second cartridge,
has a or . n greater sorbent capacity and affinity for these compounds.  Thus,
if ta*. cartridges are exposed to contaminants, the second tube should  sorb
th r.. at a higher rate.  Therefore, setting the acceptance criteria at  70%
allows acceptance of some contamination but rejects  grossly contaminated
cartridges.

     At the remaining sites (B, D, and F) the VOST train (Figure 4.2)  was
used*  Acceptance criteria #1 is not applicable but  the other two are. Note
                                     4-10

-------
 that  not  all of  the  test  reports  contain  sufficient  data  to  allow  comparison
 of  the  results to  the  QA  acceptance  criteria.   These data have  been accepted
 with  the  expressed reservation  that  their quality  is unknown.   The effects
 of  application of  these QA  criteria  on  the  data are  discussed for  each test
 in  the  following pages.

 EPA Method  23  (M23)

      M23  was proposed  by  EPA  on June 11,  1980,  but has  not been promulgated
 as  a  reference method.  It  is a method  for  determination  of  the concentra-
 tions of  low boiling halogenated  organic  compounds.   It is not  applicable
 to  compounds that  might condense  at  ambient temperatures,  are sorbed on par-
 ticles  or are water  soluble.  It  is  also  not applicable to measurement of
 oxygenated  compounds (alcohols, ketones)  as these  appear  to  be  rapidly ad-
 sorbed  by the sample container.

      A  sample of gas is collected in an inert  (Tedlar*  or aluminized Mylar*,
 flexible  walled  bag.   The sample  is  conveyed directly into the  bag from the
 stack through Teflon*  tubing.   The bag  is placed inside a rigid walled,  leak
 tight container, evacuation of  this  container  causes expansion  of  the bag
 which draws sample gas into the bag.  This  system  eliminates the need for
 contact of  the sample  with  the  pump  or  other potential  sources  of  contamina-
 tion.   Figure 4.3  is a schematic  representation of the  sampling equipment the
 method  specifies that  samples be  analyzed by a  GC  with  a  FID.   Some users of
 the method  have  substituted ECDs  or  Hall  electrolytic conductivity detectors
 (HECD)  for  the FID.  Both are more sensitive to halogenated  compounds than
 is  the  FID.   The  procedure requires  that the analysis  be  done  within two
 hours of  sample  collection.   In most cases,  this requires  that  the analysis
 be  done in  the field.

     M23  Standardization  of Procedure

      The  method  as it appeared  in the Federal Register  is  clear and speci-
 fic.  The sampling and analysis equipment are described in detail.   Clear
 and specific instructions are given  about sampling and  analysis procedures,
 QA/QC activities,  calibrations, and  calculations.  The  method should be  up-
 dated to  include the HECD and the ECD in  order  to  increase its  sensitivity
 and specificity.

      QA/QC Evaluation

     The procedure describes  two  alternate  calibration  procedures:   one  em-
ploys cylinder gases with known concentrations  of chlorinated hydrocarbons,
 the other employs flash evaporation of known quantities of compounds.  Both
 require use of the bags during  the calibration  procedure which  provides  some
assurance that the sample bags  to  be used are not unusually  sorptive  of  the
compounds of interest.   An additional QA  step,  spiking  stack gas samples,
would provide assurance that water and other stack gas  constituents  do not
interfere.

     The calibration section of the method  describes  the possibility  of  cross-
contamination if a low concentration standard is prepared  in a  bag  that  has
contained a high concentration standard.  The same cross-contamination between
stack samples and between standards and stack samples is possible.  There is


                                     4-11

-------
        I

Ff.Ts? tCLASS WOOU


            ,PflOBs
                     TEr'.OM
                                  VACUUM UN£
   u
   i  !
STACK WAi.1
                8ALI  £5"        =5  MO SAU
                    T«nmfl OR

                    ALUU1NIZEC

                    UYLARBAS
                      FIGURE 4.3



              METHOD  23  SAMPLING TRAIN
                          4-12

-------
also the possibility  that  the bags may have been contaminated dui'ng manufac-
ture or during previous use.  These possibilities should be addressed  in  the
QA procedures.  This  step  could be incorporated easily by  requiring that  the
leak check be done with pure gases and that the gas be checked  for contami-
nants prior to use of  the  bag.  A similar step could be mandated between
separate uses of a bag.

     Verification of  the cleanliness of each bag prior to  its use is required
so trip, field, and laboratory blanks are not appropriate.  The requirement
for analysis of the samples within two hours of tneir collection makes  sample
splitting impractical.  The method could be made more useful by validating
the stability of more  compounds in the inert sampling bags.  The method is
applicable to more than the seven compounds that the method lists but  the
tester must perform a  test specific validation study if it is to be used
for compounds not listed in the method.

     Results of the test burns conducted, to support the regulatory program
for industrial furnaces are evaluated in the following paragraphs.  The
evaluation is presented in three separate categories by type of industrial
furnace.  Cement, lime, and lightweight aggregate kilns are discussed  as  a
single category because these furnaces share many common performance char-
acteristics.  The three tests for asphalt aggregate kilns  ere presented as
a separate category as is  the single test for blast furnaces.
EVALUATION OF TEST BURN RESULTS FOR CEMENT, LIME, AND IWA KILNS

     The initial surveys that were performed during the early stages of
development of this BIO revealed that adequate information about  the ORE  of
hazardous compounds by kilns did not exist.  EPA hae also developed exten-
sive data that reported the ORE (ORE includes both thermal destruction and
removal by control devices) of hazardous compounds by incinerators.  There
are significant differences in the two processes.  Incinerators typically
hold their combustion gases in an oxidizing atmosphere at temperatures ran-
ging from 1800° to 2500°F for times ranging from 2.0 to 3.0 seconds.  The
combustion zone temperature in kilns is typically higher (2250* to 3000T)
and the retention time is typically as long or longer (2 to 5 seconds).
Kinetic theory predicts that elementary reactions should be faster at the
higher kiln temperature, faster by a factor ranging from 4 up to  20,000,
depending upon the activation energy of the particular reaction.  This
range has been confirmed by thermal destruction analytical system (TOAS)
data for many common hazardous compounds.  These data demonstrate that
rates of destruction increase by factors that range from 17 to 12,000 when
the temperature is rained from 1900* to 3100*F (References 3, 4,  and 5).

     EPA undertook a series of testa to determine whither the destruction
of hazardous materials by co-firing in kilns was adequate to protect th«
environment.  Beginning in October 1981, EPA performed tests at six kilns.
An additional test, performed by the State of California Air Resources Board,
has been included in this data set.  The sources that have been tested are
characterized in Table 4.2.  Three dry process cement kilns, two  wet pro-
cess cement kilns, one lime kiln, and one lightweight aggregate kiln are
included.
                                     4-13

-------
                        TABLE 4.2





COMPLETED FIELD TESTS ON HAZARDOUS WASTE CO-FIRFD IN KILNS
Site
Desig-
nation
A












B
















Kiln Type/Size
Dry Process
Cement
(150,000 Ib/hr
or 1 ,800 TPD)









Wet Process
Cement
(60,000 Ib/hr
or 720 TPl»












Primary
Fuel
Pulverized
Coal ( low
sulfur)










Pulverized
Coal















Waste Description
Mixed solvents












Mixed solvents (65% a r ana-
tic, 1% chlorinated)













Source
Emission
Control
Baghouse












Electro-
static
Preci pita tor














Sampling Matrix
o Flue gas
— Modified Method 23 for vola-
tile organic compounds
— Tenax*-GC train for volatile
organic compounds
— SASS for metals and semi-
volatile organic compounds
and PCBs
— Method 5 for particles, HC1
— Continuous monitors for S02,
NOX, O2» OO, CC>2, and TUHC
o Waste dust
o Waste fuel
o Flue gas
— VOST for volatile organic
compounds
— Modified Method 5 for semi-
volatile organic compounds,
particle metals
— Impingers for HC1
— Continuous monitors for CO2,
NOX, CO. 02, and TUHC
o Waste fuel
o ESP waste dust
o Clinker
o Kiln feed
o Coal
o Slur ry wa ter

-------
                                                 TABLE 4.2—Continued
     Site
    Desig-
    nation
Kiln Type/Size
Primary
 Fuel
Waste Deccription
 Source
Emission
Control
Sampling Matrix
           Lightweight
           Aggregate
           (20,000 Ib/hr
           or 240 TPD)
                 Pulverized
                 Coal
          Solvents,  alcohols, ethers,
          still bottoms
                       Cross-flow
                       Scrubber
           o Flue gas
             — VOST for volatile organic
                compounds
             — Method 5 for particles and
                metals
             — Impingers for HCl
             — EPA Method 3 for C02 and °2
             — EPA Method 7 for NO,,
I
u>
                                                                       — EPA Method 6 for
                                                                     o Waste fuel
                                                                     o Scrubber blowdown
                                                                     o Product
                                                                     o Raw material
                                                                     o Coal
                                                                     o Scrubber water
                                                                                                       SO-.
           Lime
           (17,000 Ib/hr
           or 204 TPD)
                 Petroleum
                 Coke  (90%)
                 Gas (10%)
          Solvents,  lacquer,  thin-
          ners,  alcohols,  still
          bottoms,  paint wastes
                       Baghouse
           o Flue gas
             — VOST for volatile organic
                con pounds
             — Method 5 for particles,
                metals, HC1
             — Method 3 for O2 and CO2
             — Continuous monitors for NOX,
                CO, SO2, TUHC, and O2
           o Waste fuel
           o Baghouse dust
           o Product
           o Raw material
           o Coal

-------
                                                 TABLE 4.2—Continued
Site
Desig-
nation
E














F
















Kiln Type/Sice
Dry Process
Cement
(120,000 Ib/hr
or 1,440 TPD)











Dry Process
Cement
(120,000 Ib/hr
or 1,440 TPD)












Primary
Fuel
Coal/
Petroleum
Coke
Mixture











Coal/Coke
Mixture















Waste Description
Solvent reclamation still
bottoms (2-5% CD













Hydrocarbon solvents (2%
CD













Source
Emission
Control
Electro-
static
Preci pita tor












Electro-
static
Preci pita tor














Sampling Matrix
o Flue gas
~ Method 23 for volatile
organic compounds
— Method 5 for particles and
metals
— EPA Method 8 for SO2
— EPA Method 7 for NOX
— EPA Method 25 for TGttIO
— Impingers for HCl
— Continuous monitors for CO,
OC>2 , a nd 02
o Waste fuel
o ESP dust
o Product
o Coal
o Flue gas
— VOST for volatile organic
compounds
— Modified Method 5 for parti-
cles, metals, semi-volatile
organic compounds
— Impingers for HCl
— Continuous monitors for CC>2f
NOX, SO2, CO, O2 and TUHC
o Waste fuel
o ESP dust
o Product
o Raw material
o Coal/coke
o Quench water
0>

-------
TABLE 4.2—Continued
Site
Desig-
nation
G
















*
Kiln Type/Size
Wet Process
Cement
(70,000 Ib/hr
or 840 TPD)














Primary
Fuel
No. 6 Oil

















Waste Description
Degreaser and pharmaceuti-
cal wastes (6 to 35% Cl)
















Source
Emission
Control
Bag house

















Sampling Matrix
o Flue qaa
— Method 23 for volatile
organic compounds
— Method 5 for particles and
metals
— EPA Method 6 for SO2
— EPA Method 7 for NOX
— EPA Method 9 for opacity
— EPA Method 10 for CO
— EPA Method 3 for CO2 and O2
— Impinge rs for HC1
•— SASS for semi-volariie
organic compounds
o Process water
o Fuel oil
o Waste fuel
o ESP dust
o Product

-------
     The percent fuel replacement chlorine and heat value of the wastes burned
at the various sites are summarized in Table 4.3.  More detailed information
about the waste fuels is given in the individual test descriptions.

DREs ef POHCs
                j
     The data fi™?m all seven test sites (References 6 through 12) were re-
viewed and subjected to the QA acceptance criteria that were presented pre-
viously.  The results at each site were compared to the promulgated incine-
rator regulation individually.

     The ORE of POHCs in kilns are summarized, by site, in Table 4.4.  The
compounds listed are tho^e that were measured in the waste and also listed
(except Freon-113) in Appendix VIII (Reference 13).  Particle, HC1, and PIC
emissions are discussed in separate sections that follow the individual test
discussions.  Likewise, an overview of the implications of all the data is
presented following the individual test discussions.

     Site A

     Site A was a dry process cement plant with a production capacity of
150,000 pounds of cement clinker per hour.  The primary fuel used at the site
was low-sulfur pulverized coal.  A portion of the clinker cooler exhaust gas
is used as combustion air in the kiln.  The kiln exhaust gas temperature was
reduced from approximately 1000°F to 575°F by a water quench.  Particulate
matter emissions were controlled by a 14-compartment baghouse.  Normally,
approximately 100 tons per month of the dust collected by the baghouse are
removed from the system for disposal.  This waste rate is sufficient to main-
tain an acceptably low alkali content in the clinker product.  During these
hazardous waste combustion tests, the disposal rate was increased to 2000
tons per month.  The increased rate of disposal was necessary to remove the
excess chloride that the waste introduced into the system.  The waste, mixed
industrial solvents, was supplied by a waste fuel broxer.  The waste was ap-
proximately 30% alcohols and ketones, 40% aromatic compounds, and 20% alipha-
tic compounds.  It contained approximately 2% chlorine.  Waste fuel provided
approximately 30% of the heat input during the co-fired runs.

     A total of 31 test runs were performed.  Eighteen of these were base-
line runs with the kiln firing coal only.  Nine of these runs were done five
months before the 13 co-fired runs and nine during the same week as the co-
fired runs.  Only two compounds listed in Appendix VIII were in the waste at
concentrations sufficient to allow computation of DREs.  They were methylene
chloride (MeCl2) and methyl chloroform (1,1,1-trichloro«thane).

     Two different methods were used to measure the stack gas emission rates
of the two POHCs.  One of these was M23, the other was a single Terax* tube
that was preceded by a condensate trap.  The M23 sampling and analysis was
done in accordance with the provisional method as published in the Federal
Register.  QA/CC data showed that the results were generally credible though
a few results were discarded because of suspected contamination.  The results
of the Tenax* analyses also exhibited symptoms of contamination.  For example,
the baseline test emission rates of both POHCs were higher (Table 4.4) than
the co-fired emission rates.  On the average the Tenax* results were higher
                                     4-18

-------
                           TABLE 4.3

                 WASTE FUEL CHARACTERISTICS OF
               HAZARDOUS WASTE CO-FIRED IN KILNS
Site
A
B
C
D
Fraction of Heat
Primary Fuel by Waste (%)
Pulverized Coal
Pulverized Coal
Pulverized Coal
Petroleum Coke
30
40-60
55
8-36
Percent Chlorine Heat Value <
in Waste Waste (Btu/
2 13,800
1-4 12,400
0.9 11,800
3.0 12,900
£ Gas

Pulverized Coal
& Petroleum Coke

Pulverized Coal
& Petroleum Coke

No. 6 Oil
13
35
10
4.b
1.7
25
12,300


12,300


10,200
                              4-19

-------
                TABLE 4.4

DESTRUCTION OF  PRINCIPAL ORGANIC HAZARDOUS
          CONSTITUENTS IN KIU1S






Site Compound
A

B


C



D





E



F




G


a
b
c
methylene chloride
1,1, l-trichloroethane
1,1,1 -trichloroe thane
methyl ethyl Ice tone
toluene
methyl ethyl Icetone
methyl isobutyl Icetone
perchloroethylene
toluene
methylene chloride
methyl, ethyl ketone
1,1,1 -trichloroe thane
trichloroe thylene
perchloroethylene
toluene
methylene chloride
methyl ethyl Icetone
1,1,1 -trichloroe thane
toluene
Freon-1 1 3
methylene chloride
methyl ethyl Icetone
1,1,1 -trichloroe thane
toluene
methylene chloride
chloroform
carbon tetrachloride
Baseline
Emissions
(No Waste
Feed)
(Ib/hr)
0.0008
0.0012
0.0004
0.0070
0.0360
a
a
a
a
0.00001
0.00006
0.00004
0.00002
0.00002
0.00013
b
b
b
b
0.00308
0.03389
0.00133
<0. 00022
0.01405
0.0108
0.0886
0.5600

Co- Fired
Emissions
( Ib/hr )
0.0003
0.0006
0.0010
0.0050
0.0380
0.0008
0.0004
0.00004
0.00146
0.00002
0.00008
0.00006
0.00003
0.00003
0.00026
< 0.0882
<0.0738
<0.135
<0.0936
0.00040
0.00621
0.00109
<0. 00011
0.00745
0.754°
0.836°
2.140C
Average
Co-Fired
POHC
Feed Rate
( Ib/hr )
15
30
20
50
100
40
20
1
146
1
1-5
2
15
15
35
95
230
50
295
40
16
55
11
150
223
277
126

Test
Average
ORE (%)
99.998
99.998
99.995
99.990
99.962
99.998
99.998
99.997
99,999
99.9974
99.9995
99.9969
99.9998
99.9998
99.9997
>99.909
>99.968
>99.731
>99.968
99.999
99.96
99.998
>99.999
99.995
99.662C
98.698C
98.300C
No baseline run at this sits.
No POHC detected during baseline runs.
Baseline emission rate has
been subtracted*
                    4-20

-------
than the M23 results by a factor of  four.  «11 emission  rates were  low,  how-
ever, so that even the DREs calculated  from  the Tenax® method,  without making
allowances for the relatively high baseline  measurements, were  in excess of
99.99%.  The Tenax* results which are given  in Table  4.4, demonstrate  that
this kiln destroyed more than 99.99% of  the  POHCs  in  the waste  stream.

     Site B

     Site B was a wet process cement kiln with a production  capacity of  60,000
pounds of clinker per hour.  The primary fuel used at this site was pulverized
coal.  A portion of the clinker cooler  exhaust gas is used as combustion air
for the kiln, the remainder either passes through  cyclones prior to its  use

-------
the first tube but the baseline emission rate was 30% of the co-fired emis-
sion rate.  The ORE of 1,1,1-trichloroethane, calculated without subtracting
the baseline emission, was 99.991 to 99.999%.  No Freon-M3 was detected in
any sample.  Freon-113 was added to the waste only on test days 8 and 9.  A
significant concentration of Freon-113 was found in the sample of waste fuel
taken on test day 7.  The average ORE of Freon-113 for thase three test days
was greater than 99.3987%.

     The results of these tests, summarized in Table 4.4, indicate that the
kiln destroyed 99.99% of the hazardous material in the waste.  This does not
appear to be true of toluene.  However, toluene emission rate during the co-
fired tests was not higher than during the baseline tests.

     Site C

     Site C was a lightweight aggregate process capable of producing 20,000
pounds of expanded clay aggregate per hour. 'The kiln gases were exhausted
through a pair of cyclone dust collectors and then through a cross flow wa-
ter scrubber.  Dust from the cyclones was recycled.  The water from the spray
scrubber was exhausted to a pond.  Scrubber water was not recycled.  The dif-
ferences between this process and the cement process was that the maximum
kiln temperature was only 2100°F (opposed to 2800° to 3000°F in cement kilns)
and the product was not alkaline and thus did not scavenge HC1 from the kiln
gases.  Note that although the maximum product temperature was 700°F less
than in the cement process, the flame temperature was not less.  Less fuel
was burned per unit of product.  The temperature of the flame was equal to
the cement kiln flame temperature, but the length of time that the gases re-
main above 2800°F was shortened by the relatively high proportion of product
in the kiln.  The liquid waste was trucked directly to the site by generators.
It consisted primarily of solvents, alcohols (10%), esters (5%), and aromatic
compounds (15%).  The waste fuel provided approximately 55% of the total heat
input to the kiln during these tests.  The balance of the fuel was provided
by pulverized coal.  The POHCs of the fuel and their approximate concentra-
tions are given below:

                      POHC                  Weight Percent in Waste

          Methyl ethyl ketone (HEX)                   2.5
          Methyl isobutyl ketone (MIBK)               2.0
          Tetrachloroethylene (Perc)                  0.1
          Toluene (MePhi)                             8.3

     The VOST was used for measurement of the emission rates of the four POHCs
listed above.  The VOST was modified, however, by insertion of an impinger up-
stream of the first Tenax* trap.  This was done to alleviate the plugging of
the Tenax* traps by condensate that caused abortion of most of the first six
VOST runs.  Unfortunately, the condensate was neither measured nor saved for
analysis.  A post hoc attempt was made to account for the quantity of POHCs
that may have been lost in the condencate.  The procedure consisted of es-
timating the volume of condensate (5 ml), and measuring the concentration
of the POHCs in the scrubber effluent.  The concentrations of POHCs in the
condensate was assumed to equal the concentration in the scrubber effluent.
This calculation probably underestimated the quantity of POHCs in the con-
densate since the scrubber effluent consists of condensate that has been
diluted by scrubber water.  Not accounting for the dilution may have caused

                                     4-22

-------
the POHC concentration in the condensate  to be underestimated by  approximate
ly an order of magnitude.  Since, the quantity of  POHCs  in  5 ml of  the  scrub
ber effluent was approximately equal to the quantity on  the Tenax®  tubes  the
emission rates nay be underestinated by an order of magnitude.

     The overall test average DREs for the various POHCs are given  below.
These values are those reported by the testing organization:

                            POHC             ORE
                                            99.998
                           MIBK             99-998
                           Perc             99.997
                           Toluene          99.999

They include the estimates of the POHC concentrations  in  the  condensate  that
were based on the scrubber effluent concentrations.  As was noted,  the esti-
mated condensate concentrations may be low by a factor of 10,  therefore,  the
DREs may be high by one significant figure (99.999 may be only 99.99).   with
this adjustment, the DREs of the POHCs approach or equal  99.99%.  However,  it
is not possible to conclude that 99.99% was attained for  certain  during  these
tests.

     Site D

     Site D was a lime kiln with a capacity of 17,000  pounds  of lime  per hour.
Limestone (CaCO3) that was quarried nearby was heated  to  approximately 2000 9F
to drive off water and carbon dioxide.  The product lime  was  sold as  barn lime
and hydrated lime, Ca(OH)2.  Approximately 6.5 million Btu of  heat  per ton  of
product provided by a mixture of petroleum coke (90%)  and natural gas (10%),
was needed by the process.  Secondary kiln air was drawn,  preheated,  from the
lime cooler.  The kiln exhaust gases passed through a  series  of large radia-
tor coolers and '•.hen through a baghouse before being exhausted.

     The waste fuel consisted primarily of lacquer thinner solvents,  alcohols,
still bottoms, paint wastes, and a smaller proportion  (4%) of  chlorinated hy-
drocarbons.  The waste was spiked with perchloroethylene  (perc) and trichloro-
ethylene (TCE) to bring its chlorine concentration to  3%.  In  addition to
these compounds, four other hazardous constituents were present in  the waste
at concentrations sufficient to allow computation of their DREs.  They were:
methylene chloride (MeCl2), methyl ethyl ketone (HEX), 1 ,1 ,1-trichloroethylene
(111-TCE),  and toluene.  The waste fuel contributed from  8 to  36% of  the  heat
input to the kiln during these tests.  Waste was atomized from a  1-inch  diam-
eter pipe that was strapped to the gas burner.  The coke  was fed  through  the
annular apace surrounding the gas and waste supply pipes.

     The end si? ion rates of the two POHCs, perc and TCE, were measured by  the
VOST.  HC1 was detetiui.-.-'i by specific ion electrode measurement of  the chlo-
ride in the impingers of the Method 5 train.

     Typically,  the fraction of the various POHCs that were trapped on the
first Tenax* tube ranged from about 50 to 80%.  A few  pair of  tubes contained
less than 50% of one or more POHCs on the first tube.  The quantity of the
various POHCs found on the field and laboratory blanks ranged  from  50 to  200%


                                     4-23

-------
of the quantity found on the associated sample tubes.  These tilings appear to
be the result of contamination.  On the other hand, some 20% of the VOST re-
sults are based upon analysis of only one of the two tubes, either the Tenax®
or the Tenax'/charcoal tube having been lost (by breakage or GC/MS malfunc-
tion).  The total emission rates (Table A.4) were low and the DREs high, how-
ever.  The baseline (no waste feed) emission rates were quite near the co-
fired emission rates.  In fact, the difference between the baseline emission
rates and the co-fired emission rates were statistically significant for only
toluene and methylene chloride.  Thus, it is concluded that the kiln destroyed
at least 99.99% of the hazardous organic constituents in the waste.

     Site E

     Site E was a dry process cement plant with a capacity of 120,000 pounds
of cement per hour.  This plant had an unusually long (520 feet) kiln with an
unusually long (10 seconds) gas retention time.  The added length, and thus
retention time, was needed because the kiln did not have a raw materials pre-
heater.  The kiln gases were exhausted through cyclones and then through an
ESP.  The dust collected in the cyclones and in the first three ESP banks was
recycled.  Dust collected in the fourth ESP bank was considered waste.

     The waste burned during these tests was the still bottoms waste of a
commercial solvent reclaimer that normally was used as supplemental fuel by
another cement plant.  This waste contained approximately 11% toluene, 8%
methyl ethyl ketone (MEK), 4% methylene chloride (MeCl2), and 2% 1,1,1-tri-
chloroethane.  The balance was alcohols, esters, and alkylated aromatic com-
pounds.  The four compounds listed were taken as the POHCs for these tests.
The fuel had a chlorine content of 4.5% and a gross heating value of approx-
imately 12,300 Btu/pound.  The waste was air-atomized into the primary fuel
(coal and petroleum coke) flame.  The waste provided an average of 13% of
the heat input to the kiln during these tests.

     The stack gas concentrations of the POHCs were measured by H23.  The
lowest concentrations of the POHCs that were detectable by M23 were higher
than the concentrations of tne POHCs in the stack gas.  Thus, even though the
flow rates of the various POHCs into the kiln were appreciable it was not pos-
sible to measure them in the stack gas at 99.99% ORE.  No POHC was detected
during the co-fired runs but the maximum calculable ORE values ranged from
99.73 to 99.97%.  The data demonstrate that the OREs were at least that good
but the method used could not demonstrate 99.99% ORE.

     Site F

     Site F was a dry process cement plant with a production capacity of
120,000 pounds of clinker per  .our.  This kiln was similar to Site E in that
there was no raw material preheater so the kiln was unusually long (520 feet)
with an unusually long gas retention time.  The kil: was normally fired with
i mixture of pulverized coal and coke.  Approximately 65% of the hot air from
the clinker cooler was used as preheated combustion air, the balance was dis-
charged through a cyclone and CSP.  The kiln exhaust gases were cooled by
water sprays to 700°F and then passed through a cyclone and an ESP for par-
ticle removal.  Approximately 90% of the 225 tons per day of dust that was
collected was recycled.  The remaining 25 tons per day was wasted.
                                     4-24

-------
     A waste fuel supply pipe was attached to the underside of the coal/coke
burner.  The waste was pressure atomized.  The waste itself was composed  of
hydrocarbon solvents and resins and solids that are typical of paint wastes.
The waste was spiked with 1,',2-crichloro-1,2,2-trifluoroethane  (Freon-113).
This compound served &z a surrogate for refractory organic compounds.   The
other POHCs that were designated  (listed below) were constituents of the
was te.
               Compound                   Weight Percent in Waste Fuel

      Freon-113                                        0.8
      Methylene chloride (MeCl2)                       0.4
      Methyl ethyl ketone (MEK)                        0.9
      1,1,1-trichloroethane                            1.0
      Toluene                                         21.6

Two baseline (no waste fuel  burned) runs were done.  These were followed  by
three co-fired runs.  The percent fuel replacement during these runs was  2ii,
37, and 42% for a test average of 35%.  The  POHC emission rate data from  the
final run (42% fuel replacement were invalidated by the analytical laboratory
because the GC/MS internal standard responses were outside of the quality con-
trol range.

     The emission rates of POHCs and other organic compounds weie measured
by both the VOST and MM5 procedures.  ORE calculations were based exclusively
on the VOST analyses.  MM5 samples were analyzed by GC/MS for higher boiling
fuel constituents and for PICs.  The VOST sample results were difficult to
interpret.  There was evidence that the samples became contaminated.  There
were two prime indicators of contamination:  high blanks, especially methy-
lene chloride,  and the baseline emission rates (see Table 4.4) were higher
for all compounds than the co-fired emission rates.  Thus, contamination  may
have contributed significantly to the measured emission rates.

     The data (Table 4.4) show the ORE of all compounds, except methylene
chloride, to be well in excess of 99.99%.  The methylene chloride data were
discounted because of the potential for contamination.  There was no indica-
tion (other than the MeCl2 data) that the kiln did not achieve 99.99% DRE
of the hazardous compounds in the waste fuel.  The MM5 data, which were less
prone to be influenced by contamination because of the larger quantities  of
compounds in the samples, indicated that the DRE of other waste fuel consti-
tuents exceeded 99.99%.  It  is most likely that the MeCl2 data are in error.

     Site G

     The kiln tested at Site G was a wet process unit that was producing
70,000 pounds of clinker per hour.  The kiln exhaust gases were cooled and
then passed through a baghouse for particle removal prior to their discharge
to the stack.  The test report did not state how much of the collected dust
was recycled nor did it describe whether the clinker cooler exhaust was used
as kiln combustion air.  The primary fuel burned at the plant was No. 6 fuel
oil.

     The was':-/  was a mixture of solvents, primarily alkanes and esters.  The
chlorine content of the waste averaged 25% (range 6 to 35%).  Most of the
                                     4-25

-------
chlorine in the waste was contributed by three compounds:  methylene chlo-
ride  (MeCl->), chloroform  (MeCl3), and carbon tetrachloride (MeCl^).  These
three compounds were designated POHCs.  The waste fuel burner pipe was lo-
cated along the inside of the annular primary combustion air conduit of the
oil burner at the 2 o'clock position.  The waste was sprayed under pressure
into the oil flame but it was not air or steam atomized.  The waste fuel
provided approximately 10% (range 3 to 14%) of the heat input into the Kiln
during these tests.

     The data from this site (Table 4.4) do not appear to be similar to the
data from the other six sites.  For example, the MeCl4 emission rate measured
during the baseline run is approximately 200 times higher than the average
POHC emission rate during the co-fired runs at the ether six sites.  The base-
line results for the other two POHCs, while not so extreme, are also high.
The measured co-fired emission rates are even higher.  Note that the baseline
emission rates have been subtracted, the co-fired emission rates given in Ta-
ble 4.4 are the measured rates less the baseline rates.  The DREs were calcu-
lated using the baseline adjusted-emission rates.  Apparently, this kiln did
a relatively poor job of destroying hazardous compounds.

Particle Emissions
     The rates of emission of particles during the baseline and co-fired
tests at the various sites are given in Table 4.5.  Combustion of hazardous
waste had an effect on the emission of dust at only Site F.  The test report
states that the chlorine from the waste fuel appeared to concentrate in the
small particles.  The observation was made at several sites.  The Site F op-
erators reported that the added chloride changed the resistivity of the ce-
ment dust enough to seriously degrade the performance of the ESP.  The oper-
ators were apparently aware that this might occur and had already consulted
with the manufacturer of the precipitator to explore possible corrective
actions.  The increased emissions at Site G were the results of the breaking
of several bags in the fabric filter.  The problem first appeared during one
of the baseline runs and was unrelated to the firing of hazardous waste.  The
particle emission rates for the six co-fired runs performed after replacement
of the broken bags> was 19.5 pounds per hour (0.068 grains per dry standard
cubic foot at 7% oxygen).

     Sites F and C appear to be the only two that did not comply with the
particle emission rate standard for hazardous waste incinerators of 0.08
grains per dry standard cubic foot, corrected to 7% oxygen,  site F, ?.s was
mentioned, apparently developed a correctable dust resistivity problem as a
result of burning the chlorinated waste.  Site C was equipped with only a
cross flow water scrubber that cannot be expected to remove particles effi-
ciently.

HC1 Emissions

     The measured rates of chlorine input in the fuel (waste fuel plus pri-
mary fuel) were compared to the stack gas emission rates of gaseous chloride
in Table 4.6.  Thes« data demonstrate that the alkaline product effectively
scavenges HC1 from the combustion gases.  Several of the test reports noted
that the bulk of the chloride appeared to be associated with the fine parti-
cle fraction of the dust that was suspended in the kiln exhaust gases.  The
                                     4-26

-------
                  TABLE 4.5

      PARTICLE EMISSION RATE FROM KILNS
Particle Emissions
Baseline
Site
A
B
C
D
E
F
G
C*
Ib/hr
NMb
19.7
c
2.0
49.6
52.5
21.7

gr/dscfa
NM
0.053
c
0.013
0.069
0.164
0.069

Co-Fired
Ib/hr
NM
19.0
11.7
2.2
58.4
240.0
30.7
19.5
gr/dscfa
NM
0.052
0.182
0.013
0.080
0.828
0.117
0.068
a  Corrected to 7% Q£ as required by incinera-
   tor regulations.
k  Not measured.
c  No baseline run at this site.
d  Co-fired runs after replacement of broken
   bags.
                     4-27

-------
                          TABLE 4.6

          HYDROCHLORIC ACID EMirSION RATE FROM KILNS
Site
A
Co-Fire
B
Baseline
Co-Fire
C
Co-Fire
D
Baseline
Co-Fire
E
Baseline
Co-Fire
F
Baseline
Co-Fire
G
Baseline
Co-Fire
Chlorine in
Fuel (JL/hr)

156
13.3
129.3

22.0
1.6
28.6
36
178
22.9
149.1
NO13
351.0
Chloride in Sc^ck
Gas (Ib/hr)
Occ
1.03
<1 .8
3.52

0.051
0.20
0.44
3.7
3.2
2.9
25.3
0.34
0.96
Removal
Efficiency

99.34
86.47
97.28

97.77
87.50
98.46
89.72
98.20
87.34
83.03
99.73
a  No baseline run at this site.
0  No chlorine detected in No. 6 oil.
                             4-28

-------
rate of waste of dust collector fines had to be  increased  to prevent  exces-
sive buildup of chloride in  the product at the cement plant.  The  increased
rate of production of waste  dust was not seen to be a major hardship.   The
rate of eirassion of HC1 at Site F was significantly higher during  the  co-
fired tests than at the other sites.  It is believed that  these samples were
contaminated by the high concentration of chloride laden dust.  Sites  E and
F were similar facilities and showed similar HC1 removal on baseline  tests.
Site C was an expanded clay  aggregate kiln.  Neither the raw material  nor the
product of this plant was sufficiently alkaline  to absorb  significant  amounts
of chloride.  However, the kiln was equipped with a wet scrubber that  effi-
ciently removed the HC1 from the gas stream.  If Site F is ignored,  .he kiln
dust clinker and control equipment collectively  removed 97% or greater of the
HC1 from hazardous waste combustion on six kiln  tests.

Products of Incomplete Combustion (PICs)
     Compounds not present in the fuel or waste fuel burned appeared  in  the
exhaust gas of several sources.  There are two general mechanisms by  which
PICs may be formed.  One is formation of products of partial oxidation of
fuel components.  The formation of aldehydes during incineration of refuse
is an example of this mechanism.  Formation of low molecular weight hyaro-
carbons (C-j-Cg) during combustion of coal and heavy oil fuels  is another
example.  The second general mechanism is a reaction o* free radicals or
other molecular fragments produced in the high temperature flame to produce
different compounds.  The products of these reactions may have higher mole-
cular weights than the fuel components.  Benzene, polynuclear aromatic com-
pounds, and soot particles appear to be formed in this manner.

     There are also means by which fuel-absent compounds may appear in the ex-
haust gas that are unrelated to the combustion.  Perhaps the major source of
fuel-abse ,t compounds is the raw materials.  These materials are sedimentary,
usually rock, but sometimes oyster shells, argonlte, etc., all of which  inher-
ently contain greater or lesser degrees of biogenic material, often residual.
Levels can be as high as a few percent of the total raw materials.  This bio-
genic material contain organic compounds that often will be drivenoff at low
temperatures and av^ear in the exhaust gases without having any opportunity
to be combusted.  Other sources of fuel-absent compounds include evaporation
of lubricating oils from mechanical equipment downstream of the kiln and com-
pounds that were present in the ambient air that was used in the furnace.
The water used in wet process cement slurry or in the scrubbers at lime  and
aggregate kilns may be a source of organic compounds.  These may exist in the
fresh water supply, be introduced by other processes where water re-use  is
practiced, or they may have be
-------
have attributed the presence of several C^-Cg alcohols and ketones in exhaust
gas samples to the degradation of XAD resin.  In addition, there is evidence
that compounds sorbed by porous polymer resins are not quantitatively removed
daring their preliminary cleanup.  Successive cleanings release additional
amounts of these compounds.  Thus, it is possible that a measured compound
may be the residue of some past sample or contamination.

     Benzene emissions were reported at Sites A and B.  These emissions  (Ta-
ble 4.7) occurred during both the baseline and the co-fired test runs.  There
was not enough data to establish whether these reported emissions were due to
primary fuel coubustion, waste fuel combustion or analytical artifacts.  No
other significant PIC emissions were reported.

Metals Emissions
     The metals emissions data that were available (Table 4.8) did not indicate
that the combustion of was**? ruei increases the rate of emission of metals by
more than would be expected from their increased firing rate.  For example, the
firing rate of lead (Pb) was five times higher during the cofired run at Site B
than it was during the baseline run, and its emission rate increased by a factor
of five.  The fraction of those metals in the fuels that have been shown to con-
centrate in small particles (all except Ni in Table 4.8) emitted to the atmos-
phere was higher than the fractional emission of other metals.  The amount of
even the volatile elements emitted was less than tns amount in the waste fuel;
i.e., approximately 15% of the mercury (Hg) in tb-j waste fuel at Site B was
erai tted.

     None of the metals sought at Site A were detectable in the stack gas par-
ticles, primarily because the total mass of sample was low.  The waste burned
at Site G contained negligible concentrations of metals.  Thus, no baseline to
co-fired test comparisons were possible for these sites.  The mass emission
rates of all dust constituents was high during the co-fired tests at Site F
because the chloride increase in the dust that occurred from burning the waste
changed the dust resistivity and degraded the performance of the precipitator.
Those emissions were not representative of normal operation and were not. in-
cluded in the tabla.

     Thv metals emissions data at Site C demonstrated the effect of concentra-
tion of certain elements in small particles.  This site was equipped with a
crossflow water scrubber that was ineffective at removal of small particles.
The fraction of thfc fuel-contained volatile metals (Pb and Cr) emitted in the
stack gas was approximately four times higher at this site than it was at the
other sites.

Combustion Gases and DRE

     Table 4.9 contains average values for combustion gases plus NO2 and the
CREs of two hazardous compounds.  No single hazardous compound was common to
all sites.  The CO2 concentrations were high and, except at Site C, were not
representative of combustion conditions, since at all other sites limestone
was being calcined.  There was no indication why both the unburned hydrocar-
bons and the NOX concentrations at Site E should be so much higher than thay
were at the other sites.  Site E DRE data cannot be coansrsd to the other
sites since there was no upper bound for them.  No meaningful correlation


                                     4-30

-------
                         TABLE  4.7

        PRODUCTS OF  INCOMPLETE  COMBUSTION  FROM  KILNS
Site
A
B
C
D
E
F
G
Baseline
Compound ( Ib/hr )
benzene
PCBs
perchloroethylene
benzene
chloroform
NO ATTEMPT MADE TO MEASURE
NO ATTEMPT MADE TO MEASURE
NO ATTEMPT MADE TO MEASURE
NO COMPOUNDS NOT IN WASTE
aromatic compounds
polynuclear aro-
matic compounds
sulfur
0.4

0.007
0.29
0.001
PICs AT THIS
PICs AT THIS
PICs AT THIS
WERE FOUND IN
_—
Co-Fire
(Ib/hr)
0.6
<0.001
0.008
1.25
0.005
SITE
SITE
SITE
STACK
0.0002a
0.0002a
0.00003a
a  Compounds found during co-fired runs  that w«re  not  found
   during baseline run.
                            4-31

-------
                               TABLE 1.3

                    METALS EMISSION RATES FHOM KILNS
                                   Mass Flow Rat.es (Ib/hr)
                         Baseline Tests
                                             Co-fired Tests
Site
Metal
Fuel Total
Exhaust Gas
Fuel Total
                                                               Exhaust Gas
          None
        Detected

          Cd
          Cu
          Hg
          Ni
          Pb
          Se

          Cd
          Cr
          Pb

          Pb
          Zn
              0.007
             <0.345
             <0.002
              0.809
              0.485
              0.009
              0.098
              0.032
               <0.00001
               <0.0004
               <0.0001
                0.001
                0.012
                0.004
               <0.006
                0..001
                   13
                   676
                   009
                   113
                   396
                 0.080

                 0.034
                 0.80
                 5.0

                 0.265
                 0.221
                 0.005
                <0.001
                 0.001
                <0.0004
                 0.055
                 0.027

                 0.006
                 0.016
                 0.43

                 0.004
                 0.005
 E

 F

 G
                Mass flow rates of metals were not calculated

                   Electrostatic precipitator malfunction

                    Metals in waste fuel were negligible
                                   4-32

-------
                                               TABLE 4.9

                     AVERAGE FLUE GAS CONCENTRATIONS OF COMBUSTION GASES  FROM KILNS





1
Id
u>

Site
A
B
C
D
E
F
G
°2
Base-
line
8.3
12.3
b
7.1
5.3
6.5
12.1
(%)
Co-
Fire
8.9
12.0
13.6
5.6
5.4
7.2
11.8
co2
Base-
line
17.0
12.9
b
23.6
24.2
23.6
15.3
(%) CO (ppm)
Co-
Fire
16.2
13.1
1.5
22.0
23.3
22.4
14.7
Base-
line
64
212
b
426
3.3
37.5
166
Co-
Fire
40
190
a
646
4.0
38.7
326
TUHCe
Base-
line
a
10.2
a
8.4
1?55C
2.5
9.0
(ppm)
Co-
Fire
a
21.0
a
3.6
469°
5
12.7
N0y (ppm)
Base-
line
682
371
b
376
1600
620
136
Co-
Fire
486
477
161
422
9o9
814
6P
ORE
Methylene
Chloride
99.998
a
a
99.997
>99.909d
99.960
99.662
(%)
Toluene
a
99.990
99.998
99.999
>99 368^
99.99'j
a
a  Not measured.
b  No baseline run at this site.
c  Total gaseous normethane organic matter.
d  None detected.  ORE calculated based on minimum detection limit of Method 23.
e  Total unburned hydrocarbons.

-------
between ORE and any PIC could be found.  Nor was there any indication that
a threshold concentration of any combustion gas, beyond which ORE was atten-
uated, exists.

Regulatory Implications of Test Burn Data

     The data indicate that, where alkaline material is produced, the HC1
that results from combustion of chlorinated organic material is absorbed by
the material.  Thus, the HC1 emissions are all low.  There is no indication
that the burning of hazardous waste affected the rate of particle emissions,
except at Site F, where the excess- chloride in the dust during co-firing ap-
peared to change the resistivity of the dust and degrade the performance of
the ESP.  This was not observed at the other three sites (B, G, and E) that
were equipped with ESPs and was thought to be correctable at Site F.  Only
Site C, which was equipped with only a crossflow water scrubber, emitted
particles at a rate in excess of 0.08 grains per dry standard cubic foot,
corrected to 7% oxygen.  Thus, it is evident that hazardous materials can
be burned in lei Ins without causing excess emissions of HC1 and particles.

     The metals emissions data indicated the need for a fuel specification/
control device specification regulation.  The combination of a control de-
vice that is ineffective for small particle removal and a high concentration
of volatile, hazardous matals could result in excessive emissions.

     The ORE data demonstrated that it is possible for kilns to achieve ade-
quate (99.99%) destruction of hazardous org?nic compounds.  However, it was
not demonstrated conclusively at ill sites nor have a complete spectrum of
all equipment types and operating conditions been sampled.  Kilns are subject
to variations in design and operating conditions in response to variations
in raw material compositions and product specifications.  The population of
kilns cannot be easily categorized.  Therefore, to be environmentally cer-
tain, regulations would need to require that a trial burn be done at each
site where combustion of hazardous organic compounds was contemplated.  Care
should be exercised in the design of theje trial burns to assure that the
combination of POHC selection and stack gas test method will bound the ORE
at 99.99% or higher.
EVALUATION OF TEST BURN RESULTS FOR ASPHALT AGGREGATE KILNS

     Interest in the level of emissions from burning waste derived fuels in
asphalt concrete plants has grown as the practire has increased throughout
the industry.  The State of Texas conducted emission testing at an asphalt
concrate plant (Reference 13> due to high occurrence of chlorinated compounds
in their fuel.  The results showed that destruction efficiencies did exceed
99.99% for most chlorinated species tested.  The report concluded that down-
wind concentrations neither adversely affect the environment nor posed any
health effects frc:.i public exposure.

     Sampling performed for the Massachusetts Asphalt Pavement Association
(MAPA) on spiked waste derived fuel (Reference 14) showed that the particu-
late limits  (NSPS) were met but trichloroethylene had a destruction removal
efficiency of 99.75% and Hydrochloric acid was emitted at 4.81 pounds per
hour with an overall removal efficiency of 81.96%.
                                     4-34

-------
     EPA tested three asphalt plants to provide data to support developing
regulations for burning hazardous waste in asphalt plants (Reference 15).
The goals for the test program were to:

     o  Determine the ORE of two POHCs (perchloroethylene and chlorobenzene)
        in a batch plant and a drum-mix plant.

     o  Determine metals removal efficiency by a baghouse and a scrubber.

     o  Evaluate HC1 emissions from a baghouse-equipped unit and a scrubber-
        equipped unit.

     Initially, it was decided to sample a drum-mix plant equipped with a
baghouse (Plant A) and a batch plant equipped with a scrubber (Plant B).
When Plant B was tested, cyclonic flow in ths stack precluded sampling for
particulate (metals).  A third plant (Plant C) was added to the program to
supplement the metal iata for the "batch plant equipped with a scrubber".

     Plant A is a 1984 vintage Barber-Greene 400 ton per hour drua-raix plant
equipped with a baghouse.  The unit is described as a "six-pack* because it
is mounted on six flat bed trailers for easy relocation.  The fuel was 100%
recycled oil, primarily from industrial sources.  Due to heavy rain prior
to the test program, the aggregate had a high moisture content.  Therefore,
production was limited to 295 tons per hour during the test program.  The
average fuel consumption was 500 gallons or 100% waste oil per hour or about
1.7 gallons  per tor of concrete produced.

     Plant B is a 1950's vintage batch plant equipped with primary cyclone
and low efficiency scrubber.  The fuel oil was 100% recycled oil, primarily
crankcase oil.  The hourly production rate was 195 tons per hour burning 390
gallons per hour of 100% waste oil.  Fuel consumption was 2.0 gallons per ton
of product.

     Plant C is a 1950's vintage batch plant equipped with a primary cyclone
and a medium to high efficiency scrubber.  The fuel fired was 100% recycled
crankcase oil.  Plant C had the lowest production capacity at 76 tons per
hour, burning 165 gallons per hour of 100% waste oil.  This is equivalent
to 2.2 gallons per ton of product.

     Plant A represents a typical modern type plant (drum-mix design) with
a high efficiency particuiate collector (baghouse).  Plants B and C repre-
sents older type plants (batch process) with low to medium efficiency parti-
cuiate collectors (water scrubbers).

      Two exhaust gas test locations were sampled on the plants.  The inlet
of the control device was sampled for particulates and metals only.  The
stack was also sampled for particulates and metals to allow determination
of control equipment removal efficiencies.  Additionally, the exhaust stack
was sampled using EPA MM5 for POHCs and a MM6 for KC1.  CEMs were used for
monitoring oxygen, carbon monoxide, and total hydrocarbons.   The OEMs were
used primarily for monitoring the process to insure against sampling during
upset conditions.
                                     4-35

-------
     Table 4.10 summarizes the production data for the three units and con-
tains stack exhaust gas data including moisture, temperature, oxygen content,
and gas flow rate au standard conditions.

     Two target compounds were used to measure the ORE of the plants; per-
chloroethylene and cMorobenzene.  Ths compounds were injected irto the burn-
er fuel line between the fuel tank and the burner.  This approach provided
several advantages over spiking the fuel in the tank.

     o  The fuel tanks were not contaminated.
     o  It was not necessary to mix the fuel to avoid stratification.
     o  It was much easier (and more accurate) to monitor the actual POHC,
        injection rate using calibrated rotometers.

     The sampling method used was EPA MM5 with XAD-2 as the sorbent trap.
The analytical method followed SW846 guidelines for GC/MS.  The two target
compounds were injected into the fuel at rates to approximately 1% by weight
of the fuel.  Samples of the fuel oil were collacLew airing each test and
submitted for analysis.  The actual feed rates and feed oil concentrations
of the target POHCs are provided in Table 4.11.

     The measured DREs are presented ir Table 4.12.  Plant A is the drum-mix
plant and Plant B is the batch plant.  In ztost cases the DREs are reported
to one place beyond actual sensitivity of the method.  That estimated deci-
mal is underlined to indicate that it is an approximation.  In a few cases,
perchloroethylene was observed but it was below the reported detectable limit

-------
                                   TABLE 4.10

            AVEKAGE PRODUCTION DATA AND EXHAUST GAS CHARACTERISTICS
Plar.t	Type   Controls
Production Data
Waste   Asphalt
 Oil    Concrete
(gph)    (tph)
                                                         Exhaust Data
 Gas      Gas    Oxygen      Gas
Moist.   Temp.   Content    Flow
 (%)     (°F)      (%)     (dscfm)
        Drum-   Baghouse    501      295
        mix

        Batch   Spray       389      195
                scrubber

        Batch   Venturi     165       76
                scrubber
                     37
                     18
                     12
          280     10.1
21,900
          142     12.6     24,460
          134     16.0     16,470
                                       4-37

-------
                                  TABLE 4.11

           POHC FEED RATES AND RESULTING CONCENTRATION OF THE FUEL
                   Perchlorcethvlene
                                        Chlorobenzene
            Grams/Minute
            Percent of Fuel
              Grams/rtinutt
            Percent of Fuel
PLANT A

  Test 1
  Test 2
  Test 3
  Test 4

PLANT B
250
178
406
435
0.87
0.66
1 .40
1.50
255
269
319
377
0.88
1.00
1 .10
1.30
Test 1
Test 2
Test 3
85
86
86
0.34
0.44
0.44
185
176
176
0.74
0.30
0.90
                                     4-38

-------
                        TABLE 4.12

            MEASURED  DREs AT PLANTS A AND B
	Perchloroethylene (%)	Chlorobenzene  (%)

PLANT A

  Test 1            -  99.997                 99.99_1_
  Test 2            =  99.99J[                 99.98£
  Test 3            >  99.997^                 99.9915^
  Test 4            >  99.9997^                99.994J_

PLANT B

  Test 1            >  99.95^                  99.94
  Test 2            =.  99.99_                  99.95
  Test 3            >  99.92                   99.96
                          4-39

-------
                TABLE 4.13

METALS CONCENTRATIONS IN THE AGGREGATE  AND
       WASTE OILS OF PLANTS A AND C
Metal
Arsenic
Aluminum
Boron
Barium
Cadmi-im
Calcium
Chromium
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Silicon
Silver
Sodium
Tin
Titanium
Vanadium
Zinc
Plant A
Aggregate
(ppm)
not analyzed
1,300
0.0
4.4
4.7
150,000
2.7
5,900
11
48,000
380
5.1
8.0
2/0
<3.0
160
<30
42
7.3
15

Waste
Fuel
(ppm)
•••
295
<1.0
23.8
<0.?
42.5
287.5
570
71.8
11.5-14
5.3
8.1-9.6
3.4
69.3
1.7-3.9
0-20
0-50
1.8-2.3
10
67
Plant
Aggregate
(ppm)
«»•«
5
1,930
10
0-1
58,300
11
8,120
7
3,590
88
4
7
665
0-1
23
0-10
73
10
10
C
Waste
Oil
(ppm)
•«•
0-1
22
26
1
571
3
188
416
286
7
7
4
150
0-1
385
0-10
0-5
12
593
                   4-40

-------
     Although a few of  the metals were present  in  the oil  in  much  greater
concentrations than in  the aggregate  (Plant  A - barium,  chromium,  lead,  and
zinc; Plant B - boron,  barium,  lead,  sodium, and zinc),  the feed rate  of the
aggregate was over TOO  times  the feed rate of the  fuel  oil on a weight to
weight basis.  This precluded tracking the fate of  a given specific  metal
from the oil through  the  system.

     Another approach was taken for evaluating  the  metals  data.  Based upon
t*e relative feed rate  of the aggregate  and  fuels  oils  (hundreds of  tons per
}A ir of aggregate to  1  or 2 tons per  hour of oil)  it was hypothesized  that
the metal concentration of the  particulate catch at the control device in-
let should be similar to  the  metal composition  of  the aggregate.  'Theoreti-
cal metal loadings were calculated in pounds per hour at the  baghouse  inlet
assuming that the particulate catch there was predominantly aggregate.  The
potential contribution  from the waste oil v/as then  calculated assuming that
all of the metals in  the  oil  reachr1  the control device inlet.  Finally, the
actual metal loading was  determined from analyses  of the Method 5  particulate
catch.  The results are compared in Tables 4.14 and 4.15 for  Plants  A  and  C,
respectively.  From these results it  appears that  the waste oil contributes
i. significant portion of  aluminum, barium, chromium, iron, lead, molybdenum,
nickel, silicon, and zinc to  the particulate material at the  control device
inlet of Plant A, and boron,  barium,  chromium,  lead, sot'j.um,  and zinc  at
Plant C.  The results are not conclusive, but are  indicative  especially  that
the barium, chromium, lead, and zinc  from the waste oils are  emitted from
the asphalt kilns into  the control equipment.

     Table 4.16 shows the overall particulate control efficiencies of  tha
baghouse (99.75% efficiency)  and scrubber (99.24%  efficiency).  Table  4.17
presents the range of measured  control efficiencies for each  of the  metals
at the two plants.  The efficiencies  for the metals deduced as being from
the fuel oils are underlined  in the table.  Most metals were  controlled  as
efficiently as the total  particulate  loading in the baghouse.  However,  the
scrubber was not as efficient for several metals including lead, sodium, and
zinc.  A baghouse is generally  as efficient  for small particulate  as it  is
for large, whereas the  efficiency of  the scrubber decreases with the particle
size.  Some of the metals are possibly more concentrated in the finer  parti-
cles and therefore collected  more efficiently by a  baghouse than a scrubber.

     Plants A,  B, and C were  all sampled for HC1 emissions using a midget
impinger train.  The samples were analyzed by specific  ion electrode.  The
inlet of the venturi scrubber on Plant C was also tested.  Samples of  the
feed oils were also analyzed for total chloride  content.   The chloride con-
centrations in the waste  oils burned  at  the three plants varied significant-
ly.  The concentrations were  1.6, 0.5, and 0.1% by  weight  for Plants A,  B,
and C,  respectively.

     Plant A burned industrial waste  oil whi"Ji  contains  "extreme pressure"
(EP) lubricants which eire long  chain  chlorinated hydrocarbons.

     The stack gas samples at all locations were found  by  analysis to  be be-
low the detection level of the  method, which is approximately equivalent to
30 ppm of HC1 in the stack gas  or less than 0.5 pounds  per hour at Plants  A
and B.   The sample volume was increased  at Site C so that  the minimum  detect-
able concentration was  0.4 ppm or equivalent to  0.03 pounds per hour.  HC1
                                     4-41

-------
                                TABLE 4.14

           COMPARISON  OF  THEORETICAL METAL CONTRIBUTIONS FROM
            AGGREGATE  AND WASTE GIL TO ACTUAL ME,\3UKEb METAL
             LOADINGS  AT  THE CONTROL DEVICE INLET, PLANT A
Metal
Aluminum
Boron
Barium
Cadmium
Calcium
Chromium
Iron
Lead
Magnesium
Manganese
Molybdenum
Nicfcel
Silicon
Silver
Sodium
Tin
Titanium
Vanadium
ilinc
Aggregate*
200
<0.15
0.68
0.72
23,000
0.41
906
1.7
7,370
58.3
0.78
1.23
41.5
<0.46
24.6
<4.61
6.45
1.*2
2.30
Waste Oil
113
<0.38
9.08
<0.11
16.3
110
218
27.4
4.88
2. 01
3.38
1.28
26.5
1.0
<7.7
<19.1
0.76
<3.83
25.6
Aggregate Plus
Waste Oil
313
<.38
9.76
0.72
23,000
110
1 , 1 24
29.1
7,370
60.3
0.78
2.51
68.0
0.5
24.6
<19.1
6.45
1.12
27.9
Actual Measured Loading
(pounds per aour
x 10-2)
374
ND
7.79
0.2
25,400
101
1,165
39.8
9,330
38.2
3.75
3.68
68.6
0.6
47.9
9.63
12.8
1.68
24.2
Metal concentra'i-,  in  the  bulk aggregate multiplied by the mass flow rate
of dust in the  flu«  ,as.
                                   4-42

-------
                                TABLE 4.15

            COMPARISON  OF THEORETICAL METAL CONTRIBUTIONS FROM
             AGGREGATE  AND WASTE OIL TO ACTUAL MEASURED METAL
              LOADINGS  AT THE CONTROL DEVICE INLET, FLANT C
Metal
Arsenic
Aluminum
Boron
Barium
Cadmium
Calcium
Chromium
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Silicon
Silver
Sodium
Tin
Titanium
Vanadium
Zinc
Aggregate*
0.13
52.4
<0.05
0.27
<0.03
1,585
0.30
220
0.20
97.6
2.39
0.10
0.20
18.1
<0.0^
0.63
<0.27
1.98
0.27
0.27
Waste Oil
<0.12
2.72
3.22
3.22
0.14
70.6
0.32
23.3
51.5
35.4
0.84
0.90
0.43
18.6
<0.12
47.6
0.27
<0.62
1.48
73.4
Aggregate Plus
waste Oil
0.13
55.1
3.27
3.49
0.17
1,656
0.62
243
51.7
133
3.23
1.00
0.63
36.7
0.15
48.2
1.51
2.60
1.75
73.7
Actual Measured Loading
(pounds per hour
x 10-2)
0.23
150
0.14
0.92
0.04
5,476
0.97
445
13.48
371
7.59
0.36
2.12
10.7
0.05
14.7
0.24
3.96
0.85
21 .0
Metal concentration in  the bulk  aggregate  multiplied by the mass flow rate
of dust in the fluo> gas.
                                  4-43

-------
                            TABLE 4.16

       PARTICULATE LOADINGS *ND CONTROL DEVICE EFFICIENCIES
Plant
A
C

Sample Location
Inlet of Baghouse
Outlet of Baghouse
Inlet of Scrubber
Outlet of Scrubber
Tarticulate
Concentration
(gr/dscf )
8.9
0.020
2.32
0.015
Parti culate
Loading
llb/hr)
1536
3.96
271. S
2.07
Control
Efficiencies
(%)
99.75
99.24*

The venturi was operating at a pressure drop of 16-17.5 inches of
water.
                               4-44

-------
                TABLE 4.17

METAL CONTROL EFFICIENCIES:  PLANT A AND C
                    Air Pollution Control
                     Device Efficiencies
Metal
Arsenic
Aluminum
Boron
Barium
Cadmium
Calcium
Chromium
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Silicon
Silver
Sodium
Tin
Titanium
Vanadium
Zinc
Plant A (%)
Not analyzed
TOO
NA
0-100
99-1 00
98-100
100
100
89-100
70-100
98-1 00
100
0-100
0-100
89-100
1 0-1 00
0-100
82- JB
0-100
35-97
Plant C (%)
87-100
100
77-100
99-100
56-81
100
85-93
99
25-42
100
97-99
67-78
94
64-79
65-81
36-56
0-36
99-1 00
81-94
25-28
                   4-45

-------
was still not detected in any of the samples.  The results are presented in
Table 4.18 with the estimated equivalent HC1 fuel feed rate.

     HC1 samples were collected at th_ inlet to the scrubber at Plant C.
Because the results at the inlet of the Plant C scrubber were also below the
detectable level (0.03 pounds per hour) it appears that the chlorides may be
attracted to the aggregate which was 60% doj.omitic lime at the sites tested.

     It appears from these test results that burning chlorinated waste at
chloride inputs at the levels used in the test are not likely to result in
an air pollution problem.  It is quite possible that the chlorides react with
the calcium in the aggregate during the drying process.  This has been demon-
strated to occur during hazardous waste incineration in cement kilns.
EVALUATION OF TEST BURN RESULTS FOR BLAST FURNACE

     A blast furnace should obtain greater than 99.99% ORE of toxic organics.
However, no actual testing dar.a are currently available to substantiate this
expected performance.  The single test conducted by EPA to evaluate  :ypical
waste disposal performance of a blast furnace when burning a liquid organic
waste (Reference 16) is deemed invalid.  There were several problem/ with the
test.  One major problem was that the waste feed POHC composition viried sig-
nificantly during the test runs.  This introduced error in the calculation
of the ORE because of uncertainty in the amount of the POHC3 in th'i feed.
Another problem was that the POHC loading on some of the blanks for the VCST
analyses were higher than the sample loadings.  Also, for son® of the POHCs,
the DRE across the blast furnace was higher than that measured across the
blast furnace and stoves.  Still another shortcoming of the test.'; was that
there were no provisions made for determining whether the POHC collected
from the gas stream was PICs from the combustion of coke and fu«l oil or
was a constituent of the hazardous waste.

     Although the single test was invalid, it is deemed very likely that
POHCs would be destroyed by a blast furnace system with a ORE of over 99.99%
because the fraction escaping destruction in the blast furnace is passed
through another combustion device.  It is expected that the stoves and boil-
ers in which the off-gas is combusted should be capable of the sane level of
DRE as the industrial boilers for which testing has demonstrated over 99.99%
DRE of toxic orgarJ.cs with regard to the off-gas that is flared, results from
several studies (References 17-19) indicated over 98% DRE of the organic com-
pounds can be expected if the flares are operated such that a stable flame
is maintained, i.e., flame stays lit.  Under unstable conditions, however,
the flare destruction efficiency can be reduced to approximately 55% (Refer-
ence 17).

     To date, EPA has not conducted any tests to determine the fate of metal
constituents of hazardous waste burned in blast furnaces.  These toxic metals
would, like metals in the raw materials, be partitioned between the slag, pig
iron, and the off-gases according to their physical and chemical properties.
Those that leave the furnace in the off-gases will be pass through a high ef-
ficiency air pollution control system,  it is conceivable that baseline emis-
sions (i.e., without burning hazardous waste) of a metal could exceed even-
tually adopted limits.  la this situation, the owners and operators may be
                                     4-46

-------
                        TABLE 4.18

             COMPARISON OF HYDROGEN CHLORIDE
          EMISSION DATA AND CONTROL EFFICIENCIES
          HC1 Input     HC1 Emissions         Control
Plant	(Ib/hr)	( ib/hr)	Efficiencies  (%)

  A         62.3            <0.38              >99.4

  B         15.5            <0.56              >96.4

  C          1.3            <0.03              >97.7
                           4-47

-------
required to burn hazardous waste with metal levels below specified limits
instead 01 being required to meet mass emission rate limits.

     As with toxic metals, thera c.re no definitive data on the fate of chlo-
rides in hazardous waste burred in a blast furnace.  The chloring in chlori-
nated wastes would likely be converted to a metal halide which would probably
be removed in the slag.  Any HC1 that is formed would have to pass through a
ben of lime at some point near the top of the furnace.  The burden contains
limestone which is converted to lime as it travels downward in the furnace.
This lime would likely neutralize most of the HC1 formed from the combustion
of chlorine-containing wastes.  The air pollution control system would cap-
ture some of the HC1 that might pass out of the blast furnace.
SYSTEM MODIFICATIONS NECESSARY TO BURN HAZARDOUS WASTES IN INDUSTRIAL FURNACES

     In order to fire hazardous wastes in existing industrial furnaces certain
modifications may be necessary, including:

     o  Installing waste storage and handling facilities.

     o  Installing equipment to characterize the wastes.

     o  Installing equipment to pretreat the waste to improve its suitability
        for firing.

     o  Adding a burner gun for firing the waste or replacing the existing
        gun with a multi-fuel burner designed to burn several fuels singly
        or in combination.

     o  Upgrading combustion process controls to handle the waste fuel.

     o  Providing waste feed rate, oxygen, and carbon monoxide monitoring to
        ensure that adequate destruction of the waste is achieved.

     o  Updating the air pollution control system to meet imposed emission
        limits.

Waste Storage and Handling

     A significant number of furnaces are used in industries which do not
generate hazardous waste suitable for burning as fuel.  Thus, waste burned
in industrial furnaces is often received from off-site.  Waste may arrive at
the furnace location in tank trucks, rail tank cars, by barge, or other spe-
cial means.

     Although it is possible to feed the waste directly from these transport
units, it is desirable to provide for some short-terra on-site storage.  Un-
1jading facilities are needed to transfer the waste from the hauler to the
storage tanks.  Because the properties of tnese wastes are almo^c always very
different from the raw materials and products handled a. the furnace sites,
existing unloading facilities are often inadequate for transferring hazardous
wastes.  Possible exceptions are facilities which use oil as the primary fuel
for the furnace.  It may be possible to use the oil unloading facilities at
                                     4-48

-------
some of these plants to transfer the waste to storage.  As discussed earlier
in this document, however, most cement, lime and lightweight aggregate kilns
are fired by coal.

     The requirements for hazardous waste storage and handling are oeing
developed in a separate regulatory analysis by EPA OSW and should be con-
sulted before designing and constructing such facilities.  The subject will
therefore not be dis'^issed further in this document.

      Characterization and Pretreatment
     As previously mentioned, essentially none of the potential hazardous
wastes burned in industrial furnaces, particularly by the  cement,  lime, and
lightweight aggregate industries are produced oy these industries.  The ma-
jor exception to this generalization is in the pulp and paper industry where
wastes from pulping processes are often burnjd as fuel in  the kilns used for
regenerating lime from lime sludge.  Since the wastes are  generated off -site,
kiln operators can be selective as to which they will accept for burning.
They are not forced to chose between firing the waste in the kiln  or finding
some alternative disposal.  Specifications on the waste storage, handling,
and firing problems; to prevent potential damage to process equipment; to
protect product quality; and to meet environmental restrictions.   Laboratory
space, instruments, and personnel must be provided for characterizing the
waste received from shippers.  Samples of each shipment need to be analyzed
to determine if wastes conform to the desired specifications.  Waste not
meeting the specifications would be rejected and returned  to the shipper.

     Although it is possible to eliminate the need for pretreatment by a
judicious choice of specifications, to do so would severely limit  the quan-
tity of waste that could be fired in furnaces.  An approach that is frequent-
ly used by furnace operators is to accept only those wastes that can be up-
graded to fuel quality by one or more of the following simple, inexpensive
pretreatment processes:  wast 3 blending, solids removal by inline  straining,
and thermal trea^.rent.  More expensive pretreatment is avoided by  requiring
that wastes meet established specifications or waste properties.   The waste
properties which affect furnace operation are those on whicn decisions re-
garding pretreatment are required.  These properties include heating value,
solids content, water content, ash, halogen, and sulfur content, and aiet<-ls
and toxic substances (e.g., PCBs ) content.

     A minimum heating value of SOOO to 8000 Btu/lb or greater is  commonly
used as a guide to define whether a waste stream is being incinerated or recy-
cled for heat recover.  When wastes are below this range, they can be blended
with streams bearing higher values.  Since the blending would be done batch-
wise,  process equipment required would consist of an agitated tank and trans-
fer pumps.  Waste would be conveyed from a shipping vehicle into the agitated
tank where it would be blended with waste of higher heating value.  After the
heating value of the blended waste is adjusted, it would be transferred to an
appropriate storage tank or fed directly to the furnace.

     Water in a liquid waste impacts a furnace in two ways.  Free  or undis-
solved water in a waste stream generally causes furnce burner pulsation and
frequently leads to flame failure.  Water also to-ds to lower the  heating
value of the organic fraction since a portion of the heat generated by its
                                     4-49

-------
combustion is consumed in vaporizing and heating the moisture to the furnace
discharge temperature.  Water removal is generally not practiced at furnace
operations because the problems and costs associated with disposing of the
generated contaminated water makes limiting the war=r content of waste more
attractive.  Most orsment, lime, and aggregate plants do not have wastewater
treatment facilities and ".hose with treatment systems are limited in capacity
(used mainly for treating scrubber water).  A practical alternative to water
removal is to develop specifications that would disallow any free water and
limit dissolved water to less than 10% by volume.  If a shipment of waste is
received which contains water in excess of the specification, it may be re-
jected.  Alternatively, it may be possible to blend it with another shipment
or with waste stored on-site to meet the specification if the water concent
is not too high and the materials are compatible.  Samples of the blended
waste would be analyzed to ensure that the blended mixture met the water
content specifications before it would be transferred to a storage taak or
directly to the furnace burner.

     High solids concentration in liquid wastes will increase the apparent
viscosity; cause blockage of the burner nozzles; settle in waste solvent
linej; and may (if not combustible and not incorporated in the product) in-
crease particle loading to the air pollution control system.  To minimize
taese problems strainers should be installed in the piping system and speci-
fications should be adopted that require that the wastes be pumpable and that
particle size be smaller than the strainer screen opening.  In-line strainers
consist of one or more mesh screen baskets housed in a vessel which may be one
of a variety of geometric configurations.  When the waste is passed through
the strainer, the solid particles ar« trapped in the basket.  Waste solvents
often contain grit and debris such as o-rings, metal shavings, etc., that can
damage pumps or plug burner nozzles.  These can easily be removed by strainers.
A 1/8-inch opening screen basket is well suited for removing such grit and de-
bris.  A 100 mash screen is recommended for removing smaller particles which
might plug a burner nozzle.  Several strainer designs are commercially avail-
able differing mainly in the cleaning approach.  A good choice fur this appli-
cation is a duplex strainer which permits the cleaning of one basket while
another is on-line.  It should be recognized that the strainer is not intended
to remove large quantities of solid?.  Its purpose is to protect pumps and
prevent plugging of small restriction? in the fuel train such as burner noz-
zles.  As long as the waste is pumpable, however, solids removal should not
be required.  Puapability of a waste containing a high solids content may be
improved by blending the waste in the sane manner as described above for in-
creasing the heating value.

     Halogens such as chlorine, bromine, iodine, etc., are commonly found
in the raw materials used to produce cement and lime clinker and pig iron.
They preferentially react with the alkali metals during production giving
volatile ha 1 ides at the higher furnace temperr.tures.  As a result/ very lit-
tle halogens are found in i\ost cement and lime clinkers and iron.  The al-
kali halides are usually carried from these furnace in the gas stream.  For
this reason, additional chlorine is sometimes introduced with the cement kiln
feed in order to reduce clinker alkali content.  The potential benefits and
problems associated with burning halogenated wastes in cement and lime kilns
were discussed earlier in this document.  The raw materials used in aggregate
(lightweight and asphaltic concrete) production are not as alkaline and con-
tain much less halogens than those for pig iron, cement, and lime (Reference


                                     4-50

-------
8) and the capacity of the aggregate product  for  forming alkali halides  is
much less than that for pig iron, cement, and lime clinker.  As a  result,
the likelihood that chlorine will exit  the furnace as HC1  is much  greater for
aggregate (lightweight and asphal^-c concrete) production.  This can  have a
major impact on the air pollution control systems for these rurnaces  es  most
are susceptible to HC1 corrosion.

     The most practical means of protecting process equipment  and  product
quality against the deleterious effects of high halogen content is to estab-
lish limits on the levels that will be  accepted.  Choice of a  suitable limit
will be, to some degree, dependent on the type of process  used for making the
product.  If suitable equipment is available, waste may be blended to meet
the specification.  This would greatly  increase the quantity of wastes that
could be accepted for burning.  A simple blending arrangement  like that  des-
cribed above for increasing the heating value of waste may be  used.

     Viscosity is important to waste firing in two regards.  A primary con-
cern is that the waste be pumpable.  Reasonable pump designs and piping  pres-
sure drops require limitations of waste fuel  viscosity at  about 10,000 stan-
dard sayboIt units.  Also, if the viscosity of the waste fired is  too high,
it will not be possible to atomize the  liquid into droplets small  enough to
oxidize completely.  Good atomization can usually be achieved  if waste fuel
at the burner is less than 750 standard saybolt units (Reference 20).   This
figure is only a qeneralization as some can handle more viscous fluids white
others cannot hai.cle liquid approaching this  viscosity.  If wastes must  be
pumpable to b» adapted for burning, viscosity related problems can be satis-
factorily handled oy keeping the waste heated, by blending with other wast.-.-.
or a combination of the two.  Viscosity can be reduced by heating  the was-;e
with tarut coils or in-line heaters.  However, 400° to 500°F is the normal
limit for heating to rtd-.'.co viscosity since pumping a hot  tar  or similar
material becomes difficult above these  temperatures.  Prior to heating a
liquid waste stream, a check should be  made to ensure that undesirable re-
actions such as polymerization, nitration, oxidation, etc., will not  occur.
If preheating is not feasible based on  these  considerations, a miscible  li-
quid of lower viscosity may be added to reduce the viscosity in a  simple
blending tank like that described above for heating value adjustment.

     Where viscous materials are to be burned, provisions for  both thermal
treatment and blending to regulate the  viscosity may be necessary.  Storage
tanks will be insulated with a two- or  three-inch layer of suitable insula-
tion and heated with a side-mounted steam or  electric heat exchanger,  or
steam coils installed near the floor.  A gear pump designed for expected
temperature and viscosity is recommended to transfer the waste fuel.   The
pump and piping should be insulated and heat  traced to prevent the waste
from cooling in the lines.

     Hazardous waste also often contains various levels of ash, sulfur,  toxic
substances such as PCB, and heavy metals such as lead, cadmium, mercury, etc.
The atmospheric discharge of some of these materials is generally  limited by
local,  state, or federal regulations.  One regulatory approach is  to  restrict
the levels of these materials in the wastes being burned.  Pretreating these
wastes  by means other than blending to meet the imposed specifications is
considered economically impractical.
                                     4-51

-------
Waste Burner

     For many furnaces a burner gun must be installed to fire wastes.  As pre-
viously mentioned in Section 3, pulverized coal is the primar, fuel for the
vast majority of cement, lime, and lightweight kilns.  The burner for firing
this fuel is essentially an 8- to 24-iiich diameter tube which projects from 5
to 10 feet into the kiln.  Pulverized coal is blown into the kiln with the
primary combustion air through this tube.  Although this type of burner may
be suitable for firing some gaseous and solid wastes, it is extremely inappro-
priate for liquid wastes.  Inasmuch as liquid wastes must be vaporized before
burning can take place, they must be atomized tc achieve efficient combustion.
The pulverized coal burner has no atomizing mechanism.  Liquid wastes could
be sprayed onto the coal before it enters the pulverizer, but this practice
would cause significant equipment and safety problems.  Thus, an atomizing
gun must be installed to fire liquid wastes in pulverized coal fueled kilns.
Although gaseous waste can be simply piped into the coal burner, generally a
center-feed gas gun is installed to give better mixing of the waste wi^h the
primary combustion air and to achieve good flame pattern control.  This gun
is essentially a 1- to 2-inch diameter tube that is tapered at the firing
end.  Slots are provided in the tapered end to distribute the gas in order
to enhance mixing with the combustion air.  It is typically located in the
center of the coal tube.  It may or may not run the full length of the coal
burner tube.  It, is unlikely that gaseous wastes would be burned in a kiln,
as those wastes are typically captive waste and are burned as generated.

     A gun must also be provided to fire waste in furnace equipped only with
natural gas burners.  These burners cannot be used for firing liquid and
solid wastes.  Although it is possible to simultaneously fire a waste gas
with natural gas through this type gun, a separate waste gun is generally
used in order to provide better flame control.  Thus, a separate waste gun
is also needed for natural gas fueled kilns.

     Furnaces equipped with oil guns may not require a separate gun :o fire
waste.  A few furnaces employ a combination of burners:  oil and gas; oil and
coal; or oil, gas, and coal.  These burners may fire the fuels separately or
in combination.  Typically, only one fuel (generally oil or gas) is fired at
a time and the capability of firing the other fuel(s) is maintained for stand-
by purposes if the supply of the primary fuel is temporarily unavailable.  For
furnaces with oil burners, some liquid wastes may be fired using the standby
oil gun.  The oil supply line is disconnected from the burner and plumbed to
the waste handling system.  In order to use it for this purpose, however, the
waste viscosity, solids content, and particle size must meet the design spe-
cifications for the gun.  Since the burner turndown is generally low (about
5:1) for this type of gun, its use at low feed rates may be limited due to
poor atomization.  Another approach for using an existing oil gun is to blend
wastes and oil if they are compatible.  This approach may be used even when
the burner is not a combination burner.  However, oil is often used for start-
up purposes and furnace temperatures are low at startup.  DREs of POHCs in an
oil/waste mixture would likely decline, so oil/waste mixtures should not b-2
used for startup.  Where the waste canr.oc be fired in the oil gun, a separate
waste gun must be provided.
                                     4-52

-------
     Regardless of which gun  is used,  it  is crucial  that  it be  suitable  for
the waste being fired and  that it be operated in  such a manner  as not  to in-
terfere significantly with the overall performance of the furnace burner.  The
performance of the burner  is  critical  to  the operation of a furnace, whether
the product is cement,  lime,  lightweight  aggregate,  or any of a wide variety
of other materials.  Both  the furnace  productivity and furnace  thermal effi-
ciency in terms of heat input per unit product output are functions of the
furnace burner.  Large  variations in both productivity and tnermal efficiency
in the same furnace under  identical conditions can be caured by changes  in
the shape, length, and  location of tne flaovs in the  furnace.  Even product
quality can be affected by these changes.  Thus,  the entire plant production
capacity will be affected  by  ihe ability  of the burner to provide an optimum
flame shape, length, and location.  It is, therefore, extremely important that
the firing of hazardous waste not alter this optimum flame pattern.  When a
waste is fired through  a gun  originally installed in a furnace  to burn con-
ventional fuel, the waste  properties (e.g., heating  value, viscosity,  vola-
tility, solids content) should not differ from the conventional fuel enough
to significantly alter  the  flame pattern.  Similarly, a burner  installed for
firing a waste tc supplement  the heat  input provided by a conventional fuel
must be designed and located  such that the flame  pattern  \s not altered  sig-
nificantly.  Also, the  flame  should not impinge on the furnace  wall.   Because
of heat radiation from  the  combustion  zone of the furnace, an air-cooled
jacketed waste gun may  be  nectssary to prevent the waste  froa pre-volatiliz-
ing inside the gun.  When  the waste/fuel  ratio is large,  it may be necessary
to replace the furnace  burner with a combination  burner to achieve a good
flame pattern.

     For liquid waste,  good atomization may be achieved by:

     o  High pressure mechanical atomization
     o  Low pressure air atomization
     o  High pressure air  atomization
     o  High pressure steam atomization

In general, mechanical  atomization of  th-s waste at proper temperature  and
pressure through the use of a correctly sized nozzle tip  will provide  com-
plete combustion and a  good flame pattern.  In the simplest form, the  waste
is fed directly to the  nozzle but turndown is limited to  2.5:1  to 3:1  since
the degree of atomization drops rapidly with decrease in  pressure (Reference
20).  In a modified form involving a return flow  of  liquid, turndown up  to
10:1 can be achieved.  Major  disadvantages of mechanical  atomization are:

     o  Erosion of the  nozzle orifice
     o  Tendency to plug with solids or liquid pyrolysis  products
     o  Potential for spraying hazardous materials over a large area in  the
        event of a rupture  of plumbing to the gun due to  the high operating
        pressures

     Air or steam atomization are also suitable for  firing liquid hazardous
waste in a furnace.  These  types of atomization will likely require addition
of air or stream supplies if  these are not available.  Packaged systems  are
available to supply these commodities.   The choice of low pressure air,  high
pressure air,  or steam atomization will depend to a  large extent on the  waste
                                     4-53

-------
properties and the amount of aii that can be -.olerated.  Low pressure atomi-
zation can satisfactorily atomize low viscosity wastes but with a high air
requirement.  Atomization air requirements vary from 370 to 1000 ft^/gallon
of waste liquid fired.  This air aust be balanced with the primary combustion
air supplied with conventional fuel to maintain an optimum flame.  For high
viscosity wastes, high pressure air or steam atomization, which is more ex-
pensive, is generally required.

     Because of the importance of controlling the rlame pattern in a furnace,
not all types of burner guns are suitable for co-firing hazardous waste with
the conventional fuel.  Equally important, some types are unacceptable be-
cause of the potential environmental risk associated with their use.  The
simple tube or pipe with no atomizing nozzle is an example of an unsuitable
waste burner gun.  For liquid wastes, which constitutes the major portion of
hazardous wastes fired in furnaces,  good atomization is essential for complete
combustion of the waste and for good flame control.  The rotary cup atomizing
gun is also not suitable for firing waste in most industrial furnaces.  These
guns produce larger droplets than atomizing burners and their flam* pattern
control is poor when used over a wide range of flow.

Combustion Process Control and Safety Shutdown

     In order to safely fire hazardous waste in an industrial furnace, it
may be necessary to improve the fuel combustion controls.  Since there is a
variety of control instrumentation,  the degree of equipment change is site-
specific.  For essentially all rotary kilns used in the lightweight aggregate
industry, fuel combustion control is limited to manual adjustments of the fuel
and combustion air feed rates by an operator.  Similarily, for open hearth
furnaces and many of the older ki\ns used in cement and lime production, fuel
firing is controlled manually.

     In order to maintain satisfactory manual control of the fuel combustion
in the kiln, the operator u»ust observe several parameters.  The flame pattern
and the color of the clinker are noted as these can indicate that the proper
combustion is being obtained.  Also, the temperatures of the gases and clin-
ker are measured at various points.   Gas samples are periodically withdrawn
from the kiln exhaust for determination of oxygen ard carbon monoxide levels.
Draft pressure developed by the fans moving air through the kiln, preheaters,
and/or precalciners are observed.  Armed with these observations, the opera-
tor generally is able to manually control the fuel combustion to safely pro-
duce quality clinker.

     Although hazardous waste may be safely fired with manual control of the
fuel and combustion air feed rates,  automatic termination of waste feed is
deemed necessary to prevent the release of hazardous materials to the envi-
ronment in the event of flameout, other combustion process upsets, or air
pollution control device failure.  This requires the installation of a flame
scanner.  This device senses ultraviolet radiation from the flame.  When the
flame is lost, the scanner signals an automatic valve in the waste feed line
to close, immediately terminating the waste to the burner.  An alarm should
also be added to warn the operator of the waste fuel shut off.  It is also
deemed necessary to provide continuous monitoring of carbon monoxide levels
in the furnace exhaust gases to ensure that proper air/fuel stoichiometry is
maintained to achieve adequate combustion of the hazardous waste.  An oxygen


                                     4-54

-------
analyzer is also necessary  to correct the carbon monoxide readings  for dilu-
tion by air inleakage into  the  flue  gases.  A carbon monoxide analyzer linked
to an automatic waste field shut-off system could quickly terminate waste  feed
in the event  that poor combustion conditions cannot be eliminated by  equipment
adjustments.  Additional discussion  of rxygen and carbon dioxide monitoring
is contained  in a later subsection.

     Because  of the ever increasing  cost of energy, industrial furnaces  con-
structed in the last five years are  generally equipped with semi-automatic
process controls which incorporate logic and instrumentation for managing
fuel combustion.  These systems vary considerably in capability with  the more
sophisticated ones including metering of the fuel(s) and automatic  adjustment
of fuel:air ratio using continuous oxygen and/or carbon monoxide monitoring.
The controls  may be set up  to control single fuel or multiple fuel  firing.
They ara usually based on known heating values or air:fuel ratio requirements
for each fuel.

     The extent to which the semi-automatic control systems must be modified
to fire hazardous waste depends on the system design and desired control op-
tions.  If a  waste is to be co-fired with a conventional fuel and is  to  be
restricted to less than about 25% of the total heat input, metering of waste
fuel into the furnace by the process controls is not needed.  The furnace
could be manually baseloaded on the  waste fuel with the process controls be-
ing used to manage process  savings by adjusting the conventional fuel and
combustion air flows.  This would require no significant change in  the com-
bustion controls.  Similarly, if the system is designed for multiple  fuel
firing, the required changes may be  as minor as reprogramming the micropro-
cessor to account for the heating value different between the waste to be
fired and the fuel being replaced by the waste.

     Significant changes to a furnace's combustion controls may be  needed
when independent modulation of the waste and conventional fuel feed rate is
required.  Two such waste/fuel burning configurations are identifiable.  In-
dependent modulation is needed to fire either 100% waste fuel or 100% conven-
tional fuel when the properties of the waste are very different from  those
of the conventional fuel, e.g., liquid waste and pulverized coal.   Similarly,
co-firing at  more than 25%  of the total heat input with a conventional fuel
would require considerable  changes to the control instrumentation.  Both these
options necessitate additional flow control valves and may require  a  new mi-
croprocessor.  An automatic flame supervision system is considered  essential
whenever one  is not included as part of the combustion control instrumenta-
tion.

Waste Feed Rate Monitoring

     It may be necessary to restrict the flow of some highly toxic waste
streams to a  small fraction of the total fuel input to ensure adequate de-
struction of  a ?OHC.  If such restrictions are adopted, waste feed  rate mon-
itoring will be needed.   Additionally,  a trial burn may be required to demon-
strate the capability of a  furnace to achieve an adopted ORE.  The quantity
of POHC being fed to the furnace is needed for the ORE determination.  Where
metering pumps or feeders are employed to convey the waste to the burner, a
continuous measurement of the feed rate may be obtained by modifying  these
devices to provide a rate-dependent signal to some type of recording device.


                                     4-55

-------
The following paragraphs some of the r.ore useful flow meters that may be used
for this application.  Detaile^ information on these and other flow meters
con be found in References 21 through 24.

     There is no universally applicable flow meter.  The proper selection de-
pends on the waste properties, acceptable pressure drop, required accuracy,
and flow range to be measured.  It is critical that the operation of the flow
meter be compatible with the waste.  Properties of chief concern in th.'.s re-
gard include physical state, viscosity, solids content, electrical conducti-
vity, and corrosivity.  Pressure loss is an important consideration in that
energy must be expended to overcome it.  The flow meter must also be capable
of measuring the flow to the specific accuracy over the flow range which it
is used.

     The crifice meter, the positive-displacement meter, the flow tube meter,
the vortex shedding meter, and the turbine meter cover a wide range of ap-
plications for liquid and gaseous hazardous wastes.  These instruments are
moderately inexpensive, are capable of the needed accuracy, are of relative
simple design, and can be used over a large range of flows.  The orifice me-
ter and flow tube are differential pressure type flow measurement devices.
This type directly measures fluid velocity by measuring a differential head
(pressure) across an obstruction in the flow stream which increases the velo-
city of the fluid, thereby decreasing its pressure.  Flow equations relate
the velocity change to the pressure change.  In an orifice meter, the differ-
ential pressure between the upstream and downstream sides of an orifice plate
is measured with pressure taps on either side of the orifice plate.

     One disadvantage of the orifice meter for use in this application is that
suspended matter in the fluid may build up at the inside of the orifice plate
and affect its accuracy.  This can be avoided by keeping the solids content
low.  If it is not practical to reduce the solids content, the flow tube may
be used.  The flow tube is basically a venturi without the downstream recovery
cone.  Because it does not restrict the flow to the extent an orifice elate
does, it is applicable to streams with appreciable solids content.  It nas a
very constant discharge coefficient and is considered i-.o be highly reliable.
It is not as expensive as the venturi but considerably more expensive than
the orifice meter.

     The positive-displacement flow meters have one or more moving parts posi-
tioned in the flow stream.  The main devices are reciprocatory piston, rotary
piston, rotary-vane meter, and nutating disk.  Of these, more nutating disk me-
ters are probably used in than all the others combined.  This device consists
of a movable disk mounted on a concentric sphere.  The disk is contained in a
working chamber with spherical sidewalls and top and bottom surfaces that ex-
tend conically inward.  The disk is restricted from rotating about its own ax-
is by a radial partition that extends across the entire height of the working
chamber.  Each complete movement of the disk displaces a fixed volume of li-
quid.  The liquid enters through an inlet port and fills the spaces above and
below the disk, which fits closely and precisely in the measuring chamber.
The advancing volume of liquid moves the piston in a nutating motion until the
liquid discharges from the output port.  The major limitation of this type of
flow meter is its sensitivity to grit.
                                     4-56

-------
     With a vortex shedding meter, the fluid stream is forced past an obstruc-
tion (shedding bar) which sets up vortices  (eddies) in the fluid.  These vor-
tices cause vibrations in the shedding bar which are proportional to the flow.
A piezoelectric crystal converts these vibrations to a voltage that is ampli-
fied and transmitted to an electronic scaling iaociule.  Advantages of these
instruments include accuracy, no moving parts, and relatively low price.

     The turbine meter is a mechanical-type measurement instrument.  It oper-
ates on the turbine principle, i.e., the volume is measured by the movement
of a wheel or turbine type of impeller.  The blades of the turbine, which are
positioned within a chamber, rotate as the fluid passes through them.  The
rotor can be positioned so that it can be driven by radial or axial flow or
a combination of both.  The rotor's motion can directly drive a register.
This device can be used to measure continuous high gas flow rates with mini-
mum pressure loss.  It is not well suited for liquids containing appreciable
quantities of solids.

Oxygen and Carbon Monoxide Monitoring

     The test results presented earlier in this section demonstrated that fur-
naces used in cement, lime, and aggregate production are capable of a 99.99%
ORE.  However, good combustion conditions must be maintained to ensure high
ORES.  Like all combustion devices, industrial furnaces are subject to operat-
ing excursions.  An acceptable DR£ must be maintained during these excursions
to prevent the emissions of hazardous materials to the environment or else
the waste feed must be shut off.  Continuous monitoring of ORE is impractical
because of the complexity (and cost) of the sampling and analysis required.
Consequently, some surrogate indicator of the combustion perfornance is needed.
Combustion performance depends on operating parameters such as temperature,
feed rate of waste, and air flow rate, but monitoring these operating param-
eters does not indicate what is actually being accomplished in the furnace in
terms of the waste destruction.

     Monitoring oxygen and/or carbon monoxide levels in the flue gases does
give a continuous assessment of the effectiveness of combustion.   (For fur-
naces intentionally operated under reducing conditions, such as a blast fur-
nace, CO monitoring as a surrogate indicator of waste combustion performance
is not practical.  However, the off-gases from reducing furnaces are typically
burned in other combustion devices where CO monitoring as a surrogate indica-
tor of combustion performance is practical).  Generally accepted combustion
theory holds that low CO (carbon monoxide) flue gas levels are indicative of
a furnace operating at maximum combustion efficiency.  Operating at maximum
combustion conditions ensures minimum emissions of unburned (or incompletely
burned) organics.  In the first stage of combustion of hazardous waste fuel,
the POHCs are immediately thermally decomposed in the flame to form other,
usually smaller,  compounds termed PICs.  In this first stage of combustion
these PICs are also rapidly decomposed to form CO.

     The second stage of combustion involves the oxidation of CO to C02 (car-
bon dioxide).  The CO to CO2 step is the slowest (rate controlling) step in
the combustion process because CO is considered to be more thermally stable
(difficult to oxidize) than other intermediate products of combustion of haz-
ardous waste constituents.  Since fuel is continuously being fired, both com-
bustion stages are occurring simultaneously.
                                     4-57

-------
     Using this view of waste combustion, the "destruction" of a POHC, and
perhaps even the destruction of PI.Cs, is independent of flue gas CO levels.
Thus, CO flue gas levels cannot be correlated to ORE for POHCs and may not
correlate well with PIC destruction.  Low CO is an indicator of the status of
the CO to CO2 conversion process, the last, rate-limiting oxidation process.
Since oxidation of CO to CO^ occurs after destruction of the POHC and its
(other) intermediate (PICs), the absence of CO is a useful indication of POHC
and PIC destruction.  The presence of high levels of CO in the flue gas is a
useful indication of inefficient combustion, and at some level of elevated
CO flue gas concentrations, an indication of failure of the PIC and POHC de-
struction process.

     Instrumentation for both 02 and CO monitoring of furnace flue gas is com-
mercially available, is considered to be reliable,  and is already installed
on some industrial furnaces for combustion control.  A variety of analyzers
are used in these monitoring systems.  These are reviewed in Reference 25
which also presents a list of vendors.

     Although the analyzers could be used to manually shut-off waste feed, an
automatic shut-off system consisting of a microprocessor-controlled shut-off
valve is deemed more appropriate.  The microprocessor could time-average the
signals from the analyzers and signal the valves to stop the waste feed if a
time-averaged CO limit is exceeded.

Air Pollution Control Equipment

     All industrial furnaces are eventually are vented through some type of
dust collection systems to control particle emissions.  The efficiency of
these systems was discussed in Section 3.  Two important considerations in
the RIA of firing hazardous wastes are:

     o  How must these dust collection systems be modified so that hazardous
        waste can be fired without violating existing emission standards?

     o  What further modifications may be required when RCRA regulation of
        hazardous waste firing in industrial furnaces is promulgated?

     In order to answer the first question, an understanding is needed of the
potential impact cf firing hazardous waste on the furnace air emissions.  This
potential impact is directly related to the following waste constituents: ash,
sulfur, trace metals, toxic substances (e.g., PCB), and halogens.  These spe-
cies exit the furnace as part of product, the waste solids and liquids, and/or
in the exhaust gases.  Any fuel conversion could increase the particle loading
beyond the capacity of collection systems.  This is not likely to occur with
furnaces firing coal as waste fuels typically contain less ash than the 5 to
20% ash coal they replace.  A waste with enough noncombustible material to
create a particle emission increase for coal-fired furnace would create in-
tolerable storage, pumping, and combustion problems.

     As with conventional fuels, some of the sulfur in the hazardous waste
fuel will exit the furnace in the exhaust gases as sulfur dioxide.  The dust
collection systems on most furnaces are not very effective for FGD (spent
acid recovery furnaces are inherently equipped for SO2 control).   When burn-
ing conventional fuels, furnace operators find it more economical to use low
                                     4-58

-------
sulfur fuels than  to install  FGD.  This being the case, there  is  little  like-
lihood taat high sulfur waste fuels will be accepted for incineration at a
all but spent recovery furnaces.  Economics simply make it more practical to
limit the waste sulfur content than to upgrade the existing dust  collection
system to provide  for FGD.

     Trace metal removal efficiencies of most furnaces dust collection sys-
tems have not been established.  Because of the potential risk to the health
of the population  surrounding a furnace, a maximum allowable concentration of
some of these metals in waste fuel could be established.  This approach  would
eliminate the need to evaluate and possibly upgrade existing control systems
to ensure that the public's health is adequately protected from exposure to
atmospheric emissions of toxic metals.

     Toxic substances like PCS, homicides, and pesticides are generally lim-
iteci to such low concentrations in the waste fuel by existing  regulations
that the need for  further reduction by air pollution control equipment is
impractical and unwarranted.

     Firing halogenated wastes impact emissions in the following  ways:

     o  For cement and lime kilns and other furnaces which process alkaline
        raw material, it can change the resistivity of the dust and increase
        the exhaust dust loading if ESPs are used.

     o  It substantially increases the HC1 emissions from furnaces not pro-
        cessing alkaline raw materials which neutralize tha HC1 produced
        during combustion.  These increased HC1 emissions which can rapidly
        corrode the equipment used for controlling particle emissions from
        these furnaces.

The halogens released when halogenated wastes are broken into  their consti-
tuent elements form volatile alkali halides in the cement and  lime kilns and
other furnaces processing alkaline materials.  Some of the alkali halides,
principally those formed from potassium (and possibly sodium), exit the  fur-
nace as gaseous fumes and later condense to dusts at the lower temperatures
outside the furnace.  This increases the dust loading to the control equip-
ment.  While the increased dust loading experienced by the dust collection
system can result in increased particle emissions, it is quite feasible  that
the process and/or air pollution equipment operating conditions can be modi-
fied to adequately control emissions.  Proceis changes, which  can probably
be employed, include reducing the insufflation rates of the recovered dust,
reducing the rates at which leached dust is recycled, altering furnace drafts,
and modifying furnace kinetics.  The appropriate air pollution control device
operation change needed to counter increased loading depends on the type of
equipment being used.  For an ESP, the increased loading may be offset by
increasing the rapping frequency and modifying tne voltage ar-) current dis-
tributions.   A baghouse is generally insensitive to slight changes in load-
ing and increasing the cleaning frequency is one way of countering larger
dust loading-  For a scrubber, the increased loading can,  in some applica-
tions,  be offset by increasing the pressure drop or the amount of scrubbing
water.
                                     4-59

-------
     The raw materials used in making lightweight aggregate contain substan-
tially less alkali metals than those used in making cement and lime.  For this
reason, the formation of alkali halides in lightweight aggregate kilns does
r»ot take place to the extent that it does in cement and litae kilns.  As a re-
sult, corrosive acids, e.g., HC1, are generated when halogenated organirs are
fired in these kilns.  Furthermore, the lightweight aggregate raw materials
are not nearly as alkaline as those used in cement and lime production and
is therefore, not as capable of neutralizir  generated acids.  Thus, these
corrosive acids exit the kiln with the exhaust gases and enter the dust col-
lection system which is generally a wet scrubber.  This can create two poten-
tial problems.  If the materials of construction are not resistant to acids,
rapid corrosion will occur and the scrubber may have to be replaced.  The
second potential problem is that some states may ha.-« HC1 emission limita-
tions.  If emission standards cannot be achieved with the existing scrubber,
modification or replacement may be required.
                                     4-60

-------
                                  REFERENCES
 1.  Protocol for the Collection and Analysis of Volatile POHCs Using VOST.
     U.S. Environmental Protection Ayency.  EPA-600/S-94-007.   March 1984.

 2.  Code of Federal Regulations, Title 40, Part 60.  Appendix  A, Reference
     Method 23.

 j.  Lee, K-C., et al.  Revised Model for the Prediction of the Tii*e-Temper-
     ature Requirements for Thermal Destruction of Dilute Organic \.ipors and
     Its Usage for Predicting Compound Destructability.   Union Carbide Cor-
     poration.  South Charleston, West Virginia.  Presented at the 75th An-
     nual Meeting of the Air Pollution Control Association, New Orleans.
     June 1982.

 4.  Preliminary data.  Personal communication from B. Dellinger, University
     of Dayton Research Institute, Dayton, Ohio to C.D.  Wolbach, Acurex Cor-
     poration.  January 1983.

 5.  Duvall, D.S. and Rubey, W.A.  Laboratory Evaluation of High Temperature
     Destruction of Polychlorinated Biphenyls and Related Compounds.  SFA-
     600/2-77-228.  Municipal Environmental Research Laboratory.  Cincinnati,
     Ohio.  December 1977.

 6.  Engineering Evaluation Report C-82-080.  Supplemental Fuels Project/Gen-
     eral Portland, Inc./California Division/Los Robles  Cement Plant/Lebec,
     California.  State of California Air Resources Board, 1983.

 7.  Evaluation of Waste Combustion in Cement Kilns at General Portland, Inc.,
     Paulding, Ohio.  Research Triangle Institute and Engineering-Science,
     Inc.  March 1984.

 8.  Day, D.R. and Cox, L.A. (Monsanto Research Corporation).   Evaluation
     of Hazardous Waste Incineration in an Aggregate Kiln:  Florida Solite
     Corporation.  May 30, 1984.

 9.  Day, D.R. and Cox, L.A. (Monsanto Research Corporation).   Evaluation
     of Hazardous Waste Incineration in Lime Kilns at Rockwell Lime Company.
     October 1983.

10.  Smith,  G.E. and Rom,  J.J. (Systech Corporation).  Hazardous Waste Com-
     bustion in a Dry Process Cement Kiln.  September 1982.

11.  Evaluation of Waste Combustion in a Dry Process  Cement Kiln at Lone
     Star Industries, Oglesby, Illinois.  Research Triangle Institute and
     Engineering-Science,  Inc.  April 1984.

12.  Peters, J.A., et al.  (Monsanto Research Corporation).  Evaluation of
     Hazardous Waste Incineration in Cement Kilns at  San Juan  Cement Company.
     August 1983.

13.  Barta,  J.P. and Nabi  S. Zarr.  Emissions from the Combustion of Fuel Oil
     Containing Chlorinated Asphaltic Compounds.  Texas  Air Control Board.
                                     4-61

-------
14.  Harris,  J.C. and Schlickenrieder,  Lynn K.   Waste  Oil  Combustion at a
     Batch Asphalt Plant:   Trail Burn Sampling  and Analysis.   Presented at
     the 76th Annual Meeting of  the Air Pollution Control  Association.   June
     19-24, 1983.

15.  Baker, R.*.., et al.   Draft  Summary Test Report:   Sampling and  Analysis of
     Hazardous Waste and  Waste Oil Burned in Three Asphalt Plants.   Prepared
     for U.S. Environmental Protection Agency.   January  1986.

16.  Destruction and Removal of  POHCs in Iron Making Blast Furnaces.   Prepared
     by Radian Corporation for the U.S. Environmental  Protection  Agency.   De-
     cember 31, 1985.

17.  McDaniels, M. (Engineering-Science).  A Report on A Flare Efficiency
     Study.  Chemical Manufacturers Association,  September 1982.

18.  Pohl, J.H., et al.  (Energy & Environmental Research  Corporation).
     Evaluation of the Efficiency of Industrial Flares:  Test  Results.   U.S.
     Environmental Protection Agency.  Publication EPA-600/2-84-095.  May
     1984.

19.  Pohl, J.H., & Soellsy, N.R. (Energy & Environmental Research Corporation).
     Evaluation of the Efficiency of Industrial Flares:  Flare Head Design &
     Gas Composition.  U.S. Environmental Protection Agency.   Publication  EPA-
     600/2-85-106.  September 1985.

20.  Engineering Handbook for Hazardous Waste Incineration.  U.S. Environmen-
     tal Protection Agency.  Washington, D.C.  Publication SW  889.   September
     1981 .

21.  Fluid Meters, Their  Theory  and Application,  5th Edition.   American So-
     ciety of Mechanical  Engineers:  New York,  1959.

22.  Spring,  L.K.  Principles and Practice of Flowmeter  Engineering,  9th
     Edition.  Plimpton Press, Norwood, Massachusetts, 1967.

23.  Cheremisinoff,  N.P.   Applied Fluid Flow Measurement:   Fundamental and
     Technology.  Marcel  Dekker, Inc.,  New York,  1979.

24.  Flow:  Its Measurement and  Control in Science and Industry.  Vol.  I
     and II,  Instrument Society  of America.  Research  Triangle Park,  North
     Carolina.  1971, 1982.

25.  Continuous Air Pollution Source Monitoring Handbook.   U.S. Environmental
     Protection Agency, Cincinnati, Ohio,  EPA-825/6-79-005.   June  1979.
                                     4-62

-------
                                  SECTION 5

               COST ELEMENTS FOR THE REGULATOR* IMPACT ANALYSIS
GENERAL

     A key element of any regulatory impact analysis is an assessment of the
potential economic impact of the various regulatory options* bei'.ig considered.
This section identifies the major cost 'actors being considered by EFA in the
assessment and provides selective cost data that are not being obtained by
EFA from other sources.  Regulations for the burning of hazardous waster as
hazardous waste derived fuels (HWDF) in industrial furnaces could potentially
have an economic impact on waste producers, furnace operators, and regulatory
agencies.  This report is limited to those costs associated with burning the
wastes in existing industrial furnaces.  The actual economic impact analysis
will be presented by EFA in a separate document.

     The economic impact of burning hazardous waste in industrial furnaces
is being analyzed by EPA in terms of how the various regulatory options alter
the net fuel savings.  This net savings is the difference between the credits
associated with the fuel replacement and the increased capital and operating
costs of firing waste over firing conventional fuels.  Therefore, analysis
of the economic impact of a regulatory option requires cost data for the sig-
nificant elements of the credits and the incremental costs.  The choice of
cost elements will depend, to a large extent, on the level of detail speci-
fiad of the analysis.  These data are being obtained from a number of sources•
The items being provided in this document fall into three major categories:
(1) conventional fuel prices, (2) capital costs to modify the furnace sy tern
to burn a hazardous waste derived fuel (HWDF), and (3) the major operating
and maintenance (O&M) costs associated with burning a HWDF.

     The costs are presented so that they may easily be applied by EPA in its
analysis as they are needed for each specific industrial furnace.
CONVENTIONAL FUEL PRICES

     The prices of conventional fuels for industrial furnaces are to be used
in the analysis of fuel replacement credits.  Two sets of prices are consi-
dered.  The first set is for 1982 which was the year covered by the OSW Burn-
er Questionnaire survey of waste fuel users.  The results of this survey will
also be used in the economic impact analysis.  The 1982 prices were:
                                     5-1

-------
     o  Naturai Gas - $3.63 per million Btu
     o  Distillate Oil - $7.24 per million Btu
     o  Residual Oil - $4.62 per million Btu
     o  Cca} - $1.09 per million Btu

     The above 1982 fuel prices were ootained from documents published by
the Energy Information Administration which is the brancn of the Department
01 Energy responsible for collecting, compiling, and disseminating data on
Uniteu States energy cost and usage.  Natural gas and oil prices were devel-
oped from data provided in Reference 1;  coal prices were based on data fiom
Reference 2.  These prices were representative of the national average values
paid by industrial users in 1982.

     One element of the RIA is to estimate the impact of the regulation on
future HWDF burner operations.  Therefore, a second fuel data set was needed;
projected fuel prices for the period covered by the analysis.  Fuel prices
projected for the period 1985 through 2010, are listed in Tables 5.1 through
5.4.  These projections were taken from Reference 3 currently being used by
the EPA Office of Air Quality Planning and Standards (OA^PS) for its econom-
ical modeling as part of its effort to develop New Source Performance Stan-
dards (NSPS) for industrial boilers.  Costs are updated periodically.
CAPITAL COSTS

     When fossil iuel-fired furnaces are used to burn hazardous wastes,
capital expenditures may be required for a number of system modifications,
including: waste characterization and pretreatment, storage and handling
facilities; fuel burners; combustion control instrumentation; waste feed
metering; combustion gas oxygen and carbon monoxide monitoring; and air
pollution controls for toxic metals and HC1 emissions.

     Cost data for waste storage and handling facilities have been generated
by EPA for its economic analysis of treatment, storage and disposal facili-
ties.  Therefore, cost data for waste storage and handling are not presented
here.  Cost data for the remaining elements have been aggregated into three
groups.

     o  Waste characterization and pretreatment costs
     o  Furnace modifications costs
     '   Air pollution control device costs

The cost data presented below were developed from published data, vendor in-
formation, and engineering judgment.  The following discussion includes the
various elements comprising these three groups.

Waste Characterization and Pretreatment Costs
     A combination of restrictions on waste characteristics and pretreatment
of wastes is needed in order to avoid potential adverse impacts of firing
hazardous waste on furnace equipment, product qual:'. "v, and the environment.
The potential problems of firing these wastes could  >- avoided merely by ad-
hering to a stringent acceptance specification.  Alternatively, one could ac-
cept wastes of unrestricted quality and depend on pretreatment to make them
                                     5-2

-------
                              TABLE 5.1

        PROJECTIONS OF REGIONAL INDUSTRIAL NATURAL GAS PRICES3
                    (1982 dollars per million Btu)
     Denand Region	1985   1990   1995    2000    2005    2010

 1.  t:«w England           6.09   7.16   8.50   11.47    14.67    16.50

 2.  New York/New Jersey   4.90   5,-il   6.42    8.67    11.09    12.47

 3.  Middle Atlantic       4.29   4.. •   5.45    7.35    9,40    10.57

 4.  South Atlantic        4.82   5.70   6.76    9.13    11.67    13.13

 5.  Midwest               4.14   4.90   5.81    7.85    10.03    11.29

 6.  Southwest             4.27   4.63   5.50    7.42    9.49    10.67

 7.  Central               3.77   4.55   5.40    7.30    9.33    10.49

 8.  North Central         4.22   4.79   5.68    7.67    9.81    11.04

 9.  West                  4.69   5.44   6.45    8.71    11.14    12.53

10.  Northwest             4.90   5.08   6.03    8.15    10.41    11.72


a  Taken from Reference 3.
                                 5-3

-------
                             TABLE  5.2

   PROJECTIONS OF REGIONAL INDUSTRIAL REJ'^DUAL  FUEL  OIL  PRICESa'b
                   Cl.yB2 dollars per million  Btu)

1 .
2.
3.
4.
5.
6.
7.
8.
9.
10.
Demand Region
New England
New York /New Jersey
Middle Atlantic
South Atlantic
Midwest
Southwest
Central
North Central
West
Northwest
1985
4.
4.
4.
4.
4.
4.
4.
4.
4.
4,
09
02
01
20
34
43
40
25
49
31
1990
5.15
5.09
5.09
5.32
5.44
5.63
5.51
5.40
5.71
5.49
1995
7.
7.
7.
7.
7.
8.
7.
7.
8.
7.
32
23
22
56
73
00
83
67
10
79
2000
8.
8.
8.
9.
9.
9.
9.
9.
9.
9.
92
82
81
21
42
75
54
35
88
50
2005
11
10
10
1 1
11
12
1 1
1 1
12
11
.11
.98
.97
.48
.73
.14
.89
.65
.31
.84
2010
12.78
12.63
12.62
13.20
13.50
13.97
13.67
13.40
14.16
13.62
a  1.6% sulfur.
   0.3% sulfur * 1.6% sulfur +  $0.68/MMBtu.
   0.8% sulfur - 1.6% sulfur +  $0.35/MMBtu.
   3.0% sulfur - 1.6% sulfur -  $0.45/MMBtu.
b  Taken from Reference  3.
                                 5-4

-------
                              TABLE 5.3

    PROJECTIONS OF REGIONAL INDUSTRIAL DISTILLATE FUEL OIL PRICES3
                    (1982 dollars per million Btu)
     Demand Region	1985   1990    1 99b    2000    2005    2010

 1.  New Enyland           6.33   7.28   10.02   12.07   14.84   16.96

 2.  New York/New Jersey   6.27   7.22    9.93   11.97   14.71   '6.82

 3.  Middle Atlantic       6.24   7.18    9.88   11.90   14.63   16.72

 4.  South Atlantic        6.08   7.03    9.68   11.66   14.34   16.39

 5.  Midwest               6.20   7.13    9.82   11.83   14.54*   16.62

 6.  Southwest             6.08   7.06    9.72   11.71   14.40   16.45

 7.  Central               6.15   7.08    9.74   11.73   14.42   16.49

 8.  North Central         5.98   6.94    9.55   11.50   14.14   16.16

 9.  West                  6.04   7.01    S.65   11.63   14.29   16.34

10.  Northwest             6.04   7.01    9.65   11.63   14.29   16.34


a  Taken from Reference 3.
                                 5-5

-------
                         TABLE  5.4

PROJECTIONS OF REGIONAL DELIVERED INDUSTRIAL COAL PRICESa'b
           (January 1983 dollars oer  million Btu)

1.






2.






3.






4.






5.








Sulfur Content
Demand Region Coal Type ( Ib SO^/MKBtu)
New England Bituminous <0.80
0.80 - 1.08
1.08 - 1.67
1.67 - 2.50
2.50 - 3.33
3.33 - 5.00
>5.00
New York/New Jersey Bituminous <0.80
0.80 - 1.08
1.08 - 1.67
1.67 - 2.50
2.50 - 3.33
3.33 - 5.00
>5.00
Middle Atlantic Bituminous <0.80
0.80 - 1.08
1.08 - 1.67
1.67 - 2.50
2.50 - 3.33
3.33 - 5.00
>5.00
South Atlantic Bituminous <0.80
0.80 - 1.08
1.08 - 1.67
1.6-* - 2.50
2.50 - 3.33
3.33 - 5.00
>5.00
Midwest Bituminous <0.80
0.80 - 1.08
1.08 - 1.67
1.67 - 2.50
2.50 - 3.33
3.33 - 5.00
>5.00
Sub- <0.80
Bituminous O.SO - 1.08
1985
3.33
3.42
3.30
3.25
3.19
2.68
2.94
3.34
3.20
3.10
2.94
2.87
2.39
2.60
2. S3
2.77
2.60
2..;o
2.41
1.98
1.77
2.92
2.74
2.55
2.30
2.64
2.09
2.52
3.13
2.94
3.00
2.70
2.59
2.18
2.23
2.63
2.63
1990
3.77
3.67
3.67
3.68
3.50
3.98
3.21
3.52
3.41
3.42
3.22
3.14
2.71
2.83
3.20
3.05
2.96
2.74
2.70
2.50
2.14
3.32
3.12
2.60
2.80
2.53
2.69
2.64
3.39
3.22
3.14
2.97
2.91
2.46
2.42
2.84
2.84
1995
3.93
3.80
3.88
3.85
3.64
3.29
3.41
3.62
3.51
3.54
3.35
3.22
2.90
2.97
3.34
3.17
3.11
2.88
2.82
2.81
2.32
3.47
3.26
3.03
3.06
2.55
2.81
2.71
3.53
3.37
3.30
3.08
3.02
2.78
2.52
2.84
2.84
2000
4.01
3.92
4.03
3.96
3.68
3.48
3.56
3.74
3.63
3.66
3.47
3.31
3.14
3.13
3.45
3.30
3.25
3.00
3.04
2.96
2.54
3.66
3.42
3.31
3.17
2.70
2.96
2.87
3.66
3.48
3.45
3.21
3.14
2.91
2.67
2.92
2.92
                             5-6

-------
TABLE 5.4—Continued
Sulfur
Content
Demand Region Coal Type ( Ib SO^/MMBtu) 1985
6. Southwest Bituminous
0
1
1
2
3

7. Central Bituminous
0
1
1
2
3

Sub-Bi tumi nous
0
1
1
8. North Central Bituminous
0
1
1
Sub-Bi tumi nous
0
1
1
9. West Bituminous
0
1
1
Sub-Bituminous 1
1 0. Northwest Bituminous
0
1
1
Sub-Bi tuminous
0
1
1
<0
.80
.08
.67
.50
.33
>5
.80
- 1
- 1
- 2
- 3
- 5
.00

.08
.67
.50
.33
.00

<0.80
.80
.08
.67
.50
.33
>5
<0
.80
.08
.67
- 1
- 1
- 2
- 3
- 5
.00
.80
- 1
— 1
- 2
.08
.67
.50
.33
.00


.08
.67
.50
<0.80
.80
.08
.67
<0
.80
.08
.67
<0
.80
.08
.67
.67
<0
.80
.08
.67
<0
.80
.08
.67
- 1
- 1
- 2
.80
- 1
- 1
- 2
.80
- 1
- 1
- 2
- 2
.80
- 1
- 1
- 2
.80
- 1
- 1
- 2
.08
,67
.50

.08
.67
.50

.08
.67
.50
.50

.08
.67
.50

.08
.67
.50
* By ZCP Inc., 1850 K Street, N.W., Washington,
b Taken from Reference 3.



2
2
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
1
1
1
1
1
1
0
0
2
2
2
2
2
3
3
2
2
2
2
2
2
D.C.

.94
.89
.03
.71
.57
.95
.32
.77
.76
.95
.95
.94
.69
•34
.53
.53
.37
.43
.64
.47
.29
.91
.52
.36
.86
.80
.69
.63
.26
.53
.34
.18
.10
.17
.80
.06
.06
.05
.05
1990
3
3
2
3
2
3
2
2
2
3
3
3
2
2
2
2
2
2
1
1
1
1
1
1
0
0
2
2
2
2
2
3
3
2
2
2
2
2
2
.31
.26
.65
.09
.79
.17
.84
.97
.95
.23
.22
.15
.61
.47
.60
.60
.44
.48
.87
.66
.34
.40
.62
.48
.92
.90
.87
.76
.49
.78
.42
.37
.24
.27
.81
.14
.14
.10
.05
1995
3
3
3
3
2
3
2
3
3
3
3
3
2
2
2
2
2
2
1
1
1
1
1
1
1
1
2
2
2
2
2
3
3
3
2
2
2
2
2
.58
.51
.02
.19
.94
,39
.97
.09
.07
.34
.18
.27
.58
.52
.72
.72
.52
.63
.92
.74
.35
.50
.59
.47
.01
.04
.91
.82
.85
.60
.49
.42
.30
.29
.78
.29
.29
.29
.05
2000
3
3
3
3
3
3
3
3
3
3
3
3
2
2
2
2
2
2
2
1
1
1
1
1
1
1
3
2
2
2
2
3
3
3
2
2
2
2
2
.77
.69
.16
.21
.08
.46
.07
.14
.09
.46
.22
.40
.66
.58
.81
.81
.81
.54
.01
.76
.38
.54
.75
.59
.12
.06
.15
.98
.91
.65
.59
.58
.42
.42
.80
.36
.36
.36
.10
, 20006.







        5-7

-------
suitable for furnace firing.  Depending solely on waste specifications would
severely limit the quantity of acceptable waste.  Conversely, depending sole-
ly on pretreatment is not technically and/or economically practical for many
wastes.  Therefore, the a»ore practical approach for preventing waste-fuel
burning related problems in industrial furnaces is a combination of waste
restriction and inexpensive pretreatment processes (blending, straining of
grit and debris, and thermal treatment).

     Operational and tixicity related problems can be prevented by establish-
ing limits on waste characteristics that significantly influence furnace equip-
ment deterioration, product quality, and/or the environmental risk related to
firing wastes.  These characteristics include heating value, water content,
solids content, halogen content, viscosity, ash, toxic substances such as PCB
and metals.  Limits on these characteristics could be used as criteria for
accepting or rejecting waste from shippers or other plant operations.  Each
waste shipment would be analyzed to determine its acceptability as a fuel.
Wastes and/or HWDF which could meet the specifications as received or after
blending, in-line straining, and/or thermal treatment would be accepted for
burning while those out-of-specification wastes would be rejected.  Blending
with either a conventional fuel or another compatible waste or HWDF would be
used when possible to upgrade it to an acceptable quality, i.e., to reduce
the concentration of metals, ash, sulfur, solids, water, or halogens, and/or
to increase the heating value of wastes.  In-line strainers would be used to
remove grit and other large particles to protect pumps and to prevent plug-
ging of the burner gun.  Viscous wastes would be heated to make them pumpable
and to permit proper atomization necessary for high combustion efficiency.

     The cost to characterize and/or pretreat a waste to make it suitable for
furnace firing offsets the money saved in fuel replacement.  This section pre-
sents these offsetting costs which are required for an economic analysis of
waste firing in industrial furnaces.

     Waste Characterization

     The use of waste specifications as criteria for eliminating unsuitable
wastes requires characterization of the physical and chemical properties.
Therefore, provisions must be made for the necessary analyses.  Throe ap-
proaches for obtaining the analyses are possible.  One approach would be to
perform all analyses on-site, which would require the acquisition of analy-
tical equipment.  Laboratory space for the needed equipment could generally
be made available in an existing laboratory or by converting office space.
Another option would entail using a commercial laboratory for the required
analyses.  A third approach, is to perform some of the simple analyses in
the' plant laboratory while contracting a commercial laboratory for the more
sophisticated analyses.  The third approach was used to prepare analytical
costs.  It was assumed that analyses for metals and toxic substances would
be made by a commercial laboratory while the other analyses would be con-
ducted in a plant laboratory.  Sufficient plant laboratory space to conduct
the simple analyses was estimated to cost approximately $50,000.  Equipment
for determining compatibility, heating value, halogen content, sulfur, ash,
water content, and specific gravity could be purchased for approximately
$25,000.  Costs for having the analyses made by a commercial laboratory are
estimated later in this section along with other waste firing related O&H
costs.
                                     5-8

-------
     Blending

     Blending a waste with a  conventional  fuel  or another higher  Btu  waste
provides an inexpensive  means  of  reducing  the concentration  of  metals,  ash,
sulfur, solids, water, or halogens, while  increasing  its heating  value  for
safe disposal in an  industrial furnace.  Blending wastes is  gen-arally a
batch process.  Grab samples  taken  from each waste  lot  are used to  confirm
that they can be upgraded to  the  established specifications  by  blending.
The selected waste batch is transferred to the  blend  tank where it  is mixed
with anociier waste or fuel oil.   Samples of the materials to be blended are
tested for compatibility before mixing.  After  the  materials are  mixed, the
blended waste is either  pumped into a  Larger tank for storage or  directly
to the furnace burner(s).

     The equipment should be  designed to process wastes received  from off-
site as weil as from other plant  operations.  A simple  system would consist
of an agitation tank plus pumps and j iping.  At mini-num, it  would use a
20,000 gallon t^nk which is large enough to hold a  tank truck load  (5,000-
10,000 gallons) plus the higher quality blend stock.  A typical blending
operation might operate  one shift p« r day  for five  days per  week, blending
20,000 gallons per shift.  This schedule is especially  suited to  plan' s that
only run one shift: per day such as  asphalt concrete plants.   The  blenching
schedule includes time for sampline and analysis, transferring  it to  the
blend tank, adding high  quality material,  mixing the  tank contents, and
pumping to longer-term storage.

     The tanks should be equipped with nitrogen blanketing as it  is sometimes
required under individual State Implementation  Plans  to minimize  evaporative
losses.  Two 100 gallons/ minute  gear pumps are required per blending tank -
one for transferring materials into the tank and one  for pumping  them out.
Gear pumps were specified because they afford a reasonably tight  shut-off and
prevent leaking when not in operation.  A  100 gallons/minute capacity would
be needed to achieve rapid transfer of materials (e.g., empty the tank  in ap-
proximately 200 minutes).  Two hundred feet of  three-inch piping  was  assumed
to be neede^ for connecting tne blend tank to other storage  and handling fa-
cilities.

     The total installed cost  of  a  system with  only one 20,000  gallon blend-
ing tank was estimated at $74,000 including pumps and piping.   This system
would provide up to  20,000 gallons  of blended material  per day  or 100,000
gallons per week for the typical  five day per week  blending  operation.  It
would produce 595 gallons per  hour  of waste fuel for continuous furnace fir-
ing.  It was considered  the minimum practical size  for  blending waste for
furnace firing.  Larger  blending  capacities may be  obtained  by  using  multi-
ple units.  Costs associated with blending larger quantities may  estimated
by multiplying the price of one unit by the number  of units  required.

     Straining of Solids

     Liquid wastes with high solids content may cause several problems  includ-
ing increased apparent viscosity, burner gun nozzle blockage, obstruction of
piping,  and possibly increased particle emissions.  Because  of  these  potential
problems,  the solids content should be sufficiently restricted  to provide* a
liquid that is pumpable and free  of grit ani debris.  In-line strainers have
                                     5-9

-------
been used to protect pumps and burner guns in the even: that large particles
turn up in the waste.  Blending may be used to reduce me solids level but
other pretreatment techniques such as filtering are not generally practiced
because of the expense and the need to dispose of the removed solids which
would be classified as hazardous waste.

     Typically, two different sets of duplex strainers are needed.  A 1/8-
inch opening screen basket is requir
-------
                                     FIGURE 5.1
             EQUIPMENT DIAGRAM FOR THERMAL TREATMENT OF HAZARDOUS WASTES
RECYCLE
TO STORAGE
 STORAGE
 TANKS ARE
 INSULATED
AND HEATED
 FROM
 STORAGE
O
O
             BACK-PRESSURE
             REGULATOR
                              STEAM
 FUEL
HEATER
TO
 KILN
                  DUPLEX
                 STRAINER
                               CONDENSATE
                                                                  m
                                                                  in
                                                                  m
                                 m
                                 •s
                                 O

                                 8
                                 m
                                 8
                                 m

-------
and piping.  The installed costs of this equipment are plotted as a function
of waste feed rate in Figure 5.2 as Curve 1.

     Wastes of greater than 10,000 standard saybolt units viscosity at ambi-
ent temperature must not only be treated to insure good atomization, out must
also be kept hot enough to prevent them from setting up in the storage tank
and piping.  All waste received at the waste derived fuel storage facility
must be pumpable.  The equipment required to handle waste in this category
include the heat exchanger for heating the waste as it is pumped to the fur-
nace, provisions to recirculate the waste back to the storage tank, as well
as itarns to keep the liquid warm:

     o  Storage tank insulation
     o  Storage tank heaters
     o  Pipe tracing and insulation
     o  Addition of a spare circulating pump

Installed costs of this equipment, excluding the storage tank insulation and
heaters, are plotted as a function of waste throughput in Figure 5.2 as curve
2.  Although waste storage tank costs are not part of the authorized scope of
this study, the tank heater and insulation costs are provided.  These costs
are presented in Table 5.5.  The storage tank heater for which costs are pre-
sented is the immersion steam coil type.  Insulation costs are for three
inches of fiberglass insulation.

Furnace Modification Costa

     Burner Modification Costs
     The cost for modifying an existing burner system to fire hazardous waste
is site specific.  It depends on the existing burner type and capacity, type
of conventional fuel fired, properties of the waste, and quantity of waste to
be fired.  The least expensive approach is likely to be taken.  Some furnaces
require only that a burner gun be replumbed to fire the waste.  This would
not require significant capital expenditure.  In other instances, the hazard-
ous waste could be blended with the conventional fuel and fired with no mod-
ifications to the burner.  This approach could be used when the wasts and
conventional fuel are compatible or when burning liquid waste in the fuel
oil-fired burners.

     Costs given below are for providing the necessary burner components to
fire a liquid hazardous waste in natural gas, oil and combination fossil fuel-
fired furnaces.  It was assumed that the waste was piped to the burner, i.e.,
costs do not include a fuel handling train.  The figures were obtained from
burner vendors and, therefore, may be different from actual costs because
some furnace operators may fabricate their own waste burners.

     One approach to firing waste materials is to install a dedicated waste
burner gun in or adjacent to the conventional fuel burner.  Waste is simply
atomized into the primary fuel (natural gas or oil) flame envelope.  This
arrangement may be used for co-firing waste up to 30% of the total heat in-
put.  A good quality atomizing burner gun, properly sized for the range of
waste flows can be purchased for $10,000 to $15,000, depending on the length
of the burner.  The tc'.al installed cost in 1982 dollars, including plumbing
and flame safeguard system, is from $20,000 to $30,000.

                                     5-12

-------
                         E=S ENGINEERING-SCIENCE -
                              FIGURE  5.2


                 INSTALLED CAPITAL COSTS FOR THERMAL


                    TREATMENT OF HAZARDOUS WASTES
          100
        ui  _-
        eg  SO
        O
        G
 £  20

*t




 H  10

 O
 U

 _J



 E

 u

 o
 IU
        I-
        tf!
        Z
            2 —
             10
                                                               CURVE
                                                               CURVE 2
            2C
  50      100     200


WASTE FLOW, CPH
500
1000
CURVE-1- PUMPABLE AT AMBIENT TEMPERATURES BUT MUST BE HEATED

          FOR GOOD ATOMIZAT1ON.



CURVE-2-  MUST BE HEATED TO KEEP PUMPABLE AND TO OBTAIN GOOD

          ATOMIZATION.
                               5-13

-------
                    TABLE 5.5

STORAGE TANK HEATER AND INSULATION INSTALLED COSTS



Tank Size      Heater Costs      Insulation Costs
(gallons)     (1982 dollars)	(1982 dollars)

    500           11,000                430

  1,000           12,000                66C

  2,000           14,000                960

  5,000           17,000              1,310

 10,000           19,000              2,850

 20,000           38,000              3,640
                       5-14

-------
     It is assumed that the process controls for modulating an additional
burner are not necessary.  The furnace could be  annually based loaded on the
waste fuel with the existing process controls being used to modulate the con-
ventional fuel and combustion air flows.  Based on the detailed study of the
furnaces and their current practices, this assumption is valid and should
extend to other furnace types.

Carbon Monoxide and Oxygen Monitoring

     Continuous monitoring of carbon monoxide levels in the exhaust gases of
a furnace burning a hazardous waste is deemed necessary to ensure that good
combustion conditions are maintained to provide adequate destruction of the
POHCs and PICs.  Oxygen monitorin.., is required in conjunction with CO moni-
toring to adjust CO levels to a common excess air rate indicated by excess
oxygen content in the flue gas.  Correcting CO levels to a common flue gas
oxygen content avoids the problem of having (otherwise) high CO levels di-
luted by large quantities of excess air.  Some furnaces are already equipped
with oxygen and/or carbon monoxide monitoring to save fuel costs through in-
creased combustion efficiency, and/or for safety purposes.  Most of the CO
monitors installed on cement kilns are for detecting explosive or unburned
hydrocarbon concentrations and are nog designed to measure the low CO levels
typical of optimum combustion conditions (0-5000 ppm).  These devices typical
operate in a concentration range of 0-10%.

     The cost of CO and oxygen continuous emissions monitoring systems for a
"clean" environment is considerably less than that for a "dirty" application.
There is not a clear distinction between a "clean" and a "dirty" exhaust gas.
For purposes of this analysis, however, a "dirty" exhaust gas application is
defined as one that requires an extractive monitoring system in which the
extracted sample must be treated to reduce the gas temperature, particulate
loading, and moisture content.  A sample location in the backend of a dry
cement kiln is an example of a "dirty" application.  A "clean* application
example is a sample location downstream of the air pollution control deveice.

     Typical costs for oxygen and carbon monoxide continuous emission moni-
toring systems for clean applications are $15,000 and $20,000 respectively
(1982 dollars).  These costs include analyzers, sample transport system (if
extractive type unit), chart recorder and installation.  If an automatic data
reduction system is desired an additional cost of $20,000 would be incurred
(Reference 5).

     For a more hostile environment, the cost of the continuous emission mon-
itor systems (CB*S) will be significantly higher due to the added cost of
specialized sample conditioning equipment.  Providing an unadulterated sample
that is compatible with the analyzers is more difficult in sample locations
where there are extremely high temperature, moisture, and particulate loading.
The cost of special equipment needed to adjust these stack gas characteristics
increases with the number of characteristics that must be adjusted and the de-
gree of adjustment required.  For purposes of this analyses it will be assumed
that the conditioning system must treat a flue gas that requires a water-coded
probe and has a high moisture content and particulate loading.
                                     5-15

-------
     Waste Feed Metering

     If limitations on the hazardous waste feed rate are established based on
the trial burn results* capital outlays for flow metering will be necessary.
Two types of liquid flow measurement devices well-suited for this application
are the positive displacement meter and the orifice meter.  These units are
relatively inexpensive ana are applicable to a wide range of wastes.  Either
device can be obtained for approximately $2,000 to meter flows up to 7,000
gallons per hour, which is the maximum flow rate one would anticipate for
waste f-ad to the industrial furnaces being considered.  Some type of record-
ing device is needed to provide a permanent record of the waste faed rate.
A simple $2,000 chart recorder would serve this purpose adequately.  Instal-
lation is lively to cost an additional $1,000.  Thus, the total installed
cost of a liquid waste feed metering system will be approximately $5,000 in
1982 dollars.

Air Pollution Control Cost

     The iss.ua of particulate matter, metals, and hydrochloric acid emissions
from a furnace burning hazardous waste need to be considered on a source spe-
cific basis.  The furnaces considered in detail in this study are already
equipped with air pollution control systems for removing particulate matter
from the. furnace exhaust gases.  II is possible that most of these systems
will be able to meet adopted toxic metal emission'standards should they be
adopted.  It is assumed that metals will be controlled by either particulate
matter controls or waste fuel specifications.  Furthermore, hydrochloric acid
emissions may not be a problem if operators do not burn chlorinated waste at
chloride levels that cculd cause the adopted emission levels to be exceeded.
However, should a furnace operator chose to burn HWOF's that would cause ac-
ceptable health risks from toxic metals and hydrochloric acid emission to be
exceeded, additional air pollution controls would be required.

     The type of control system that would be installed will vary depending
on the emission limitations eventually adopted and the exhaust gas parameters.
For study purposes, costs are presented for a combination venturi scrubber
for particulate matter removal followed by a packed bed absorber for HC1 re-
moval.  The venturi scrubber will remove metals that are contained in the
exhaust gases as particulate matter while those in the vapor state will be
removed by the packed tower.  Costs are presented as a function of furnace
exhaust gas flow.  Factors are also given to estimate costs if only metals
as particulate matter or HC1 removal is required.  Other combinations of tox-
ic metals and HC1 removal are available, but the venturi/absorber is a prac-
tical technique that adequately represents the costs element needed to as-
sess the economic impact of pollution control equipment that may be required.
This approach is somewhat conservative in that the packed tower may not be
necessary in all cases to accomplish similar results.

     Installed cost for a combination venturi/absorber system may be esti-
mated from the following equation:

                             Cost » 96 x Q0.8164

where:  Q - the exhaust gas flow rate in acfm
                                     5-16

-------
This system includes a quench tower to lower the exhaust temperai_are from
550°F to saturation, a venturi scrubber for particle collection, acid gas
absorber, caustic recycle system for neutralizing the scrubber water, ID
fan, stack, and auxiliaries.  The assumed materials of construction are:

     o  High-nickel-alloy quencher and venturi throat
     o  High-grade, chemically resistant, high-temperature fiberglass shell
        for the cyclonic separator and packed tower
     o  Polypropylene tower packing
     o  Inconel or HasteXloy fan wheel with rubber-lined steel housing

     If no venturi scrubber is needed for metals control, the total system
cost is reduced by 15%.

     If acid gas absorption is not required, the total system cost is re-
duced by approximately 40%.

     Costs represented by the above equation are also based on purchased
equipment cost contained in Reference 6 with installation assumed to be 100%
of the purchased cost.  These costs are typical for a venturi pressure drop
requirement of 30-inches water column, which should be adequate for this ap-
plication.  They are indexed to 1982, the year covered by the OSW Burner
Questionnaire which is being used in the RIA.
OPERATING AND MAINTENANCE COSTS

     The operating and maintenance costs for a furnace operation are like-
ly to be greater when 100% waste fuel or a combination of waste and conven-
tional fuels are burned than when 100% conventional fuels are burned.  Addi-
tional storage and handling facilities are typically required to fire wastes
and these must be maintained.  If agitators or nitrogen blanketing are used,
operating costs will increase.  When the wastes require pretreatment before
burning, additional O&M costs are incurred.  There may also be increased
furnace O&M costs an a result of adding a waste burner gun.  However, these
costs are not expected to be significant.

     The equipment needed to meet the regulatory requirements that are even-
tually adopted will likely further increase the O&M costs associated with
firing hazardous waste over those associated with firing conventional fuels.
Air pollution control devices added to meet emission limitations will in-
crease the O&M cost burden.  Further costs will be incurred to maintain oxy-
gen and/or carbon monoxide monitoring and waste metering.  O&M costs asso-
ciated with waste metering are, however, expected to be insignificant.

     This section presents O&M costs for the following of the aforementioned
items:

     o  Waste Characterization and Pretreatment
     o  Furnace Modification Costs
     o  Air Pollution Control Devices

Increased O&M costs, associated with waste storage and handling other than
pretreatment, are provided by others.
                                     5-17

-------
Waste Characterization Pretreatment O&M Costs

     The annualized O&M costs associated with waste characterization and the
three pretreatn.3nt processes (bxending, straining, and thermal treatment) for
which capital costs were given earlier in this section are presented here.

     Waste Characterization Cost

     As previoesly discussed, furnace operators can avoid high O&M costs to
upgrade hazardous waste by limiting the materials that they will accept to
those that require only blending, straining, or thermal treatment tc make
them suitable for fuel.  This practice requires that each shipment of waste
be analyzed to determine if it meets established specifications.  Wastes
should be analyzed for heating value, halogens, sulfur, water, ash, speci-
fic gravity, toxic metals, and PCBs.  These analyses may be made by a plant
chemist or by an independent commercial laboratory.  The need for rapid re-
sults requires the analyses to be made at a laboratory near the plant.  Since
metal and PCB determinations require more sophisticated analytical procedures
and equipment, it is assumed that these are performed at a nearby commercial
laboratory while the other analyses are completed in the plant laboratory.

     O&M costs for the waste characterization approach described above will,
therefore, include simple charges for the analyses performed by the outside
firm and the labor and other costs associated with the plant's waste analyses.
Assuming 30 samples are analyzed per week for PCB by gas chromatography and
up to five metals by atomic absorption, the annual charges by the outside
laboratory is estimated at $156,000 in 1982 dollars.

     Labor for waste characterizations by the plant's chemist and his super-
visor is estimated at $26,500.  Maintenance and indirect O&M costs are taken
at 22.2% of the capital costs of the laboratory space and equipment.  These
costs are $16,600 which brings the total annual waste characterization O&M
costs to $199,100 in 1982 dollars.

     Pretreatment

     Certain elements of the pretreatment costs are generic to all three pro-
cesses.  These are listed in Table 5.6.  As shown in Table 5.6, the O&M costs
include both direct and indirect components.  The direct components include
operating labor, maintenance, and utilities.  Operating labor is taken at
$9.75/manhour.  Estimated labor requirement is two manhours for waste blend-
ing.  For straining and thermal treatment, it is ansumed that no operating
labor is needed.  Supervision is estimated at 15% of the total operating la-
bor costs (Reference 7).  Maintenance requirements are difficult to predict
accurately for these types of operations.  For such situations, maintenance
is generally taken at 2-6% of the capital costs (Reference 8).  Five percent
was used in these estimates.

     Utilities required for the pretreatment processes include electricity
to drive pumps and the blend tank agitator, nitrogen for blanketing the blend
tanks, and steam for the thermal pretreatment.  Pump power consumption is es-
timate., from the following equation:
                                     5-18

-------
                      TABLE 5.6

          COMPONENTS OF ANNUALIZED OSM COSTS
    Direct
Operating Costs
    Cost Factor*
Operating Labor
  - Operator
  - Supervisor

Maintenance

Utilities
  - Electricity
  - Steam
$9.75/manhour
15% of operator

5% of capital costs
SO.05/kilot* tt-hour
$6.00/1,000 pounds
   Indirect
Operating Costs
    Cost Factor*
Overhead



Property Tax

Insurance

Adainis tration

Capital Recovery
Costs
80% of operating
labor and main-
tenance labor

1% of capital costs

1% of capital costs

2% of capital costs

0.132 (using i •
10% and an equipment
life of IS years)
   All costs are in 1982 dollars.
                         5-19

-------
                   kilowatt-hours - 0.746 (gpm) (hd) (sg) H
                                            3960 n

where:  gpm - flow rate, United States, gallons/ninute
         hd - head of fluid, feet
         sg • specific gravity relative to water 9 60°F, 29.92 incnes mercury
          n » overall pump/motor efficiency - 40*
          H • hours of operation

Agitator horsepower for the 20,000 gallon tank is assumed to be 3 horsepower.
An electricity cost of $0.05 per kilowatt-hour (Reference 9) was used in the
power costs determination.  Unit cost and quantity of nitrogen needed for
blending the blend tank are used on data taken from Reference 6.

     Steam is needed for tne thermal pretreatment.  An average cost of $6.00/
1,000 pounds of steam at 100 pounds/square inch was used.  The indirect oper-
ating costs include the costs of overhead, taxes, insurance, administration
expense*, and capital charges.  Taxes, insurance, and administration can col-
lectively be estimated at 4% of the capital costs, wnile overhead charges
can be considered at 80% of the labor charges for both OtM.  The annualized
capital charges reflect the costs associated with capital recovery over the
depreciable life of the system and can be determined as follows:


            Capital Recovery Cost - (capital costs) x  *    * * *
                                                      (1 * i)n - 1

where:  i • annual interest rate
        n - capital recovery period

for these estimates, a useful life of 15 years and an average annual interest
rate of 10% were assumed.

     Blending O&M Costs

     The estimated annual 0«M costs for the previously described tank blend-
ing system is $36,000.  These costs are based on a five day per week blend-
ing operation using a <0,000 gallon blend tank equipped with agitation and
nitrogen blanketing.  Wastes are either transferred from other part* of the
plant or unloaded from a tank truck or rail car into the tank where they are
upgraded to the desired quality by adding othe- wastes or fuel oil.  The
waste mixture i* continuously agitated to achieve uniform blending.  Af-
ter the desired quality is obtained, the blended waste is pumped to either
another tank for long-term storage or directly to the burner.  Cost* are
based on the assumption that it take* eight hour* to prepare 20,000 gallons
of blended waste which is also the maximum daily production rate of the sys-
tem.  One operator is required two hours per day to oversee the blending
process.  Utilities include electricity to drive the agitator and transfer
pumps and nitrogen for blanketing the tank.  The nitrogen can be either
vented to the furnace or to the same system a* used for the storage tanks.

     The blending O4M costs trill vary with the amount of fuel blended.  Costs
presented here are for one 20,000 gallon tank system which will provide 595
gallon* per hour of blended waste*  OCM cost* for furnace* firing more than
the capacity of the 595 gal/hr blending system can be obtained by multiplying


                                     5-20

-------
the costs presented here for this system by the number of multiple blending
systems required.  This can be determined by dividing the production capacity
to be fueled with waste by 595 gal/hr and rounding to the nearest whole num-
ber.  For example, a 3,000 tens/day blast furnace firing 7 gallons of waste
fuel per ton pig iron produce would consume 875 gal/hr of the waste.  Two
20,000 gallon blending systems (595 gal/hr) are required to produce the nec-
essary waste fuel.

                   875 gal/hr , 1.47 (2 rounded to nearest
                   595 gal/hr         whole number)

Therefore, the O&M costs to blend the waste fuel for this furnace is two
times the costs presented for the 20,000 gallon system, i.e., $72,000.

     Straining O&M Costs

     The O&M costs associated with removing grit and debris with basket
strainers are considered insignificant for purposes of the RIA.

     Thermal Treatment O&M Costs

     Figure 5.3 presents the annualized thermal treatment O&M costs as a
function of waste feed rate for the two categories of waste for which capi-
tal coats were given earlier in this section:

     o  Curve 1 - Wastes with viscosities greater than 750 standard sayboIt
        units but less than 10,000 standard saybolt units at ambient tempera-
        tures.

     o  Curve 2 - Wastes with viscosities greater than 10,000 standard say-
        bolt units at ambient temperatures.

Provisions for heating and insulating of the storage tanks are not included
in the O&M costs presented in Figure 5.3.  Wastes with viscosities greater
than 10,000 standard saybolt units require tank heating and insulation while
wastes with viscosities between 750 and 10,000 standard saybolt units at
ambient temperatures do not.  The following are O&M costs for storage tank
heating and insulating:

                        O&M Costa - Thermal Treatment

            Tank Size (gallons)      O&M Costs (1982 dollars/year)

                     500                           470
                   1,000                           890
                   2,000                         1,710
                   5,000                         2,190
                  10,000                         3,900
                  20,000                         5,380
                                     5-21

-------
                            ENGINEERING - SCIENCE -
                             FIGURE 5.3
            ANNUALJZED O&M COSTS FOR THERMAL TREATMENT
                        OF HAZARDOUS WASTES
         100
            10    20        50     100     200

                          WASTE FLOW, CPH
500
1000
CURVE-1-  PUMPABLE AT AMBIENT TEMPERATURES BUT MUST BE HEATED
          FOR GOOD ATOM1ZATION.


CURVE-2-  MUST BE HEATED TO KEEP PUMPABLE AND TO OBTAIN GOOD
          ATOMI2AT1ON.
                               5-22

-------
Furnace Modification O&M Costs

     Burner Modification O&M Costs

     The burner gun to fire waste fuel should require  little operator atten-
tion beyond that required for the furnace firing conventional  fuels.  Main-
tenance cost will be about 5% of capital cost, capital recovery  13.2%, and
taxes, insurance, and administration, 4%.  Total annual Ot~'i cost will be
about 22.2% of the original burner gun assembly capital cost.

     Oxygen and Carbon Monoxide Monitoring O&M Costs

     Annual operating costs for several continuous monitoring  systems have
been estimated for evaluating the economic impact of NSPS for  industrial boil-
ers  (Reference 7).  These costs should also be applicable to "clean" exhaust
gas applications.  An annual O&M cost of $18,500 in 1982 dollars was estimated
for an oxygen monitoring system.  Costs for a carbon monoxide  system should be
about the same.  The major items included in this estimate are the maintenance
and performance certification.  One-half manhour per day was assumed to be re-
quired for the maintenance at a rate of $35.81/manhour, including supervisor
and overhead.  One certification test per year, costing $11.900  was assumed.

     The operating and maintenance cost of a CEMS for  dirty gas  is expected
to be higher than for a clean gas application.  The extent to  which these
cost will be higher will depend on the type of conditioning system selected
cod to some degree will be site-specific.  The labor will be approximately
30% higher for a system for a dirty gas application based on an  assumed one-
half manhour per day labor requirement.  Thus, for a dirty gas system, three
quarter-manhours per day would be required for 0/M.  At the assumed ?35.81/
manhour labor rate (1982 dollars) used above the 0/M cost would  be $21,000
(1982 dollars) which includes $11,900 for one certification test.

     Requiring 02 and CO monitors on industrial furnaces burning hazardous
waste could result in a fuel cost savings to the operators of  these devices.
This potential saving would result if the operators used the 02  monitors to
maintain low excess air (LEA) combustion of the fuels.  With LEA combustion,
less fuel is required because less heat is lost out the stack  with the com-
bustion gases.

     The magnitude of the potential fuel savings that  can be obtained by LEA
combustion must be determined individually for each furnace because it depends
on many factors.  The major factors influencing the potential  savings include:

     o  Furnace type and condition
     o  Burner type and condition
     o  Combustion control type and condition
     o  Operating load level

     The furnace type and condition have a large impact on the amount of fuel
saving that may be achieved through LEA combustion.  Some types  have design
characteristics that limit the range of LEA operation.  Also,   the flue gas
exit temperature for one type furnace can be significantly different from
those of another type.   Since the fuel savings for a given excess air reduc-
tion is temperature dependent furnaces with higher exit flue gas temperatures


                                     5-23

-------
should be capable of achieving a higher fuel savings per unit excess air re-
duction.  The condition of the furnace also impacts the fuel savings poten-
tial of LEA.  A furnace that has significant air in-leakage is .-nore difficult
to operate at low excess air levels because the air infiltration may distort
the O2 reading drastically.

     The type and condition of the burner(s) installed in the furnace also
greatly influence the fuel savings potential of LEA operation.  A burner is
designed to operate efficiently over a specific excess air range.  If oper-
ated at an excess air range lower than the design level, proper mixing of the
fuel and combustion air cannot be achieved.  Poor air and fuel mixing would
likely result in incomplete combustion of the fuel and higher fuel consump-
tion.  Gas burners generally can operate at lower excess air levels than oil
burners which generally operate at lower levels than coal burners.  There is
also a wide variation in tne excess air level operations capability of burn-
ers for a given fuel.  For example, design excess air levels for oil burners
used in asphalt dryers can vary from five tc twenty percent.  The condition
of the burner also affects the potential fuel savings because the fuel flow
through a dirty or damaged burner is difficult to control.

     Another important factor determining the potential of LEA combustion
and hence fuel savings is the type and condition of the combustion controls.
Combustion controls vary widely in complexity from the simple single-point
positioning units typically found on smaller units to the metering system of
a complex, computerized process control system.  Interfacing the 02 monitor
to these systems has limitations that are unique to each type of control sys-
tem.  The level cf LEA achievable is limited to how well the Q^ monitoring
is used by the control system.  Also, the condition of control system mech-
anical components also impacts the fuel savings potential.  Damper linkage
may flax slightly, and bearings may wear over time.  Even metering systems
are susceptible to some shortcomings, since their flow transmitters are op-
erated at temperatures and pressures that vary significantly from those at
which the transmitters were initially calibrated.

     More excess 02 is needed at low loads because of poorer mixing of the
fuel and air.  Consequently, the operating load level also impacts the fuel
savings of LEA combustion.

     Because of the influence of the factors discussed above, a detailed
breakdown of potential fuel savings by furnace type and fuel is deemed un-
justified.  For the purpose of determining the economic impact of requiring
02/CO monitoring, two estimates of this potential fuel saving is considered
adequate.  A 2% savings is estimated as a typical average value for all the
furnaces discussed in Section 3 except asphalt and lightweight aggregate
kilns.  A 4% savings is estimated for asphalt and lightweight aggregate fur-
naces because these units are typically operated at very high excess air
levels.  Both estimates are based on discussions with combustion control
equipment suppliers, information found in the literature, and limited data
on oxygen levels and furnace exit flue gas temperatures.  These estimates
are also based on the assumption that a 0.5% increase in combustion effi-
ciency is achieved per 1% reduction in the 02 level in the flue gas.
                                     5-24

-------
     Not all furnace operators would  receive the fuel saving estimated  for
LEA operation as a consequence of requiring 02/CO monitoring.  Many of  the
larger industrial boilers are already equipped with O2-trim or C>2/CO-trinu
In fact, those industrial furnaces equipped with CO-trim systems will actual-
ly be required to operate at a higher LEA  level than they are currently op-
erating at and thus wiuld consume more fuel.  Furnaces with CO-trim systems
typically operated with a CO setpoint of from 200 to 400 ppm.  A 100 ppm CO
limit would require that they operate at a higher LEA level than their  cur-
rent setpoint level.  Also, nearly all sulfur recovery furnaces are equipped
with O2-trim systems.  Thus, estimating the fuel savings potential of requir-
ing O2/CO monitors on a particular population of furnaces requires a know-
ledge of how many units in the population  already are operating at LEA  levels
(i.e., how may are using 02 or CO monitors to achieve LEA combustion.)  Once
the fraction of furnaces currently employing LEA controls is determined, the
potential fuel savings for the entire population may be estimated from  the
potential savings of a single unit.

     The savings of a single furnace  can be estimated by multiplying the to-
tal annual fuel cost by 0.02 or 0.04  depending on the type fuel burning de-
vice being considered.  The annual fuel cost is estimated by multiplying the
design heat impact by the unit fuel cost presented in the proceeding para-
graphs.  For example, the maximum potential annual fuel savings of a 150 x
106 Btu/hr, residual oil-fired furnace would be:
       .02 x 150 x
106 Btu/hr x 8760 hrs/yr x S4.62/106 Btu - $121,400
This assumes that the furnace operates 24 hours per day, 365 days a year.
This savings can be adjusted to match different assumptions regarding load
factor.

     A comprehensive survey of furnaces was not conducted to determine the
fraction of units already equipped with an LEA capacity.  Conversations with
industry representatives and control system vendors indicate the approximate
breakdown listed below:

           Furnaces                 % With Some Form of LEA Cap

     Lightweight Aggregate                       30
     Blast Furnaces
       Stoves                                    50
       Boilers                                   50
     Spent Acid Recovery                         50
     Asphalt                                      0
     Reverberatory                               50

Air Pollution Control Devices O&M Costs

     The annualized O&M ccsts for the venturi/gas absorption systems pre-
sented in Figure 5.4 are based on 8,700 hours/year operating time and the
cost factors presented in Table 5.7.  The O&M costs include direct costs
such as operating labor and materials, maintenance, replacement parts,
utilities, and collected particulate disposal.  Also included are indi-
rect costs such as overhead, insurance, taxes, and capital recovery.
                                     5-25

-------
     Estimated operating labor  requirement for  the  air  pollution control de-
vice system is four manhours per shift.   The utilities  include quench water,
scrubber water, absorber water,  caustic  soda, and electricity.
                                     5-26

-------
                                     FIGURE 5.4




                     ANNUALIZED  OPERATION AND MAINTENANCE COSTS

                       FOR THE VENTURI/GAS ABSORPTION SYSTEMS
    1C.OQC
ui
a;
=    1,000
        100,	

        10,000
                                           100,000

                                    EXHAUST GAS  RATE, a fern
1,000 000
                                         5-27

-------
                      TABLE 5.7

        COMPONENTS OF ANNUALIZED O&M COSTS FOR
            AIR POLLUTION CONTROL DEVICES
    Direct
Operating Costs
    Cost Factor*
Operating Labor
  - Operator
  - Supervisor

Maintenance

Utilities
  - Electricity
  - Steam
$9.75/manhour
15% of operator

5% of capital costs
$0.5/kiIowatt-hours
$6.00/1,000 pounds
   Indirect
Operating Costs
    Cost Factor*
Overhead



Property Tax

Insurance

Administration

Capital Recovery
Costs
80% of operating
labor and main-
tenance labor

1% of capital costs

1% of capital costs

2% of capital costs

0.132 (-using i *
10% and an equipment
life of 15 years)
*  All costs are in 1982 dollars.
                          5-28

-------
                                  REFERENCES
1.    Energy Information Administration.  Monthly Energy Review.  Washington,
     D.C., Publication DOE/EIA-0035 (83/1213]).  December 1983.

2.    Energy Information Administration.  Coal Production - 1982.  Washington,
     D.C., Publication DOE/EIA-0118 (82).  Septonber 1983.

3.    Memorandum from T- Hogan,  Energy and Environmental Analysis, Inc., to R.
     Short.  EPA:EA3.  June 22, 1983.

4.    Engineering Handbook for Hazardous Waste Incineration - Cincinnati, Ohio.
     U.S.  Environmental Protection Agency,  EPA SW-889.   1981.

5.    Dickerman, J.C. and Kelly, M.E.   Issue Paper:   Compliance Monitoring
     Costs.  Radian Corporation, Durham, North Carolina.  September 25, 1980.

6.    McCormick, R.J. and DeRosier, R.J. (Acurex Corporation).  Capi'^al and
     O&M Cost Relationships for Hazardous Waste Incineration.  Prepared for
     the U.S. Environmental Protection Agency.  EPA Contract 68-02-3176 and
     68-03-3043.  July 1983.

7.    Chilton, C.H.  Cost Engineering  in the Process Industries.  McGraw-Hill,
     New York, 1980.

8.    Peteru, M.S., and Timmerhause, K.D.  Plant Design  and Economics for
     Chemical Engineers.  McGraw-Hill,  New York,  1980.

9.    Comparative Evaluation of  Incinerators and landfills for Hazardous Waste
     Management.  Prepared by Engineering-Science for the Chemical Manufactur-
     ers Association.  May 1982.
                                     5-29

-------