EPA/540/5-91/005
July 1992
Technology Evaluation Report:
SITE Program Demonstration Test
Horsehead Resource Development Company, Inc.
Flame Reactor Technology
Monaca, Pennsylvania
Risk Reduction Engineering Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45268
-------
Program Demonstration Test, Horsehead Resource Develop-
ment, Inc., Flame Reactor Technology, M'onaca, Pennsylvani
9. PERFORMING ORGANIZATION NAME AND ADDRESS
TECHNICAL REPORT DATA
{H<:ctc read /nssruciivns on the rrirrse before enmpirii
1. REPORT NO.
EPA/540/5-91/005
4. TITLE AND SUOTITLE
Technology
2,
valuation Report: SITE
REPORT DATE
July 1992
7. AU'T HtJftlS)
Ken Partymiller, et al
8. PERFORMING ORGANIZATION REPORT NO.
PRC-
Houstoni Texas 77380
12. SPONSORING AGENCY NAME AND ADDRESS
Risk Reduction Engineering laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
Cin., OH
i. PERFORMING ORGANIZATION CODE
1U. PROGRAM bLEMfcN I NO.
1 1. CONTRACT/GRANT NO.
Contract No. 68-CO-0047
13. TYPE OF REPORT AND PERIOD COVERED
.-ProJerJ: Report ______ _
14. SPONSORING AGENCY CODE |)j-q j g~£ £
EPA/GOO/14
10. SUPPLEMENTARY NOTES
Marta K. Richards, Project Manager
Donald A. Oberacker, Project Manager
FTS and Commercial: 513/569-7783
FTS and Commercial: 513/569-7510
16. ABSTRACT
A SITE demonstration of the Horsehead Resource Development (HRD) Company, Inc.
Flame Reactor Technology was conducted in March 1991 at the HRD facility in
Monaca, Pennsylvania. For this demonstration, secondary lead smelter soda slag
was treated to produce a potentially recyclable lead- and zinc-enriched oxide
product and a nonhazardous (based on the regulatory requirements of the Toxicity
Characteristic test) effluent slag. The lead and^zinc in the oxide product were
concentrated about threefold from the feed concentrations. The effluent slag was
determined to be nonhazardous based on extraction by the Toxicity Characteristic
Leaching Procedure.
Potential wastes that might be treated include industrial residues, Resource
Conservation and Recovery Act wastes, Superfund wastes, and other wastes
contaminated with metals and organic wastes.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPT ORS
High-temperature test
Lead
Zinc
Thermal recovery method
Waste treatment
18. DISTRIBUTION STA1 tMtN i
RELEASE TO PUBLIC.
ti.IDENTIFIERS/OPEN ENDED TERMS
19. SHCURIT Y CLASS p'/iis Report)
JiNOASSUElEU
20. SECUHITY CLASS phis paste)
UNCI ASS I FTFf)
c. COS AT I Field/Croup
21. NO. OF PAGES
127
77. PRICE
EPA Foim 2220-1 (Rov. 4-77) previous F.uniON 11 obsolete '
i
-------
NOTICE
The information in this document has been prepared for the U.S. Environmental
Protection Agency (EPA) Superfund Innovative Technology Evaluation (SITE) Program under
Contract No. 68-C0-0047. This document has been subjected to the EPA's peer review and
administrative review, and it has been approved for publication as a U.S. EPA document.
Mention of trade names or commercial products does not constitute an endorsement or
recommendation of use.
ii
-------
FOREWORD
The Superfund Innovative Technology Evaluation (SITE) Program was authorized in the
Superfund Amendments and Reauthorization Act of 1986 (SARA). The SITE Program is
administered by the U.S. Environmental Protection Agency (EPA) Office of Research and
Development (ORD). The purpose of the SITE Program is to accelerate the development and use
of innovative cleanup technologies applicable to Superfund and other hazardous waste sites. This
is accomplished through demonstrations designed to provide performance and cost data on
selected technologies.
A field demonstration was conducted under the SITE Program to evaluate the Horsehead
Resource Development Company, Inc., (HRD) Flame Reactor technology. The technology
demonstration took place at HRD's facility in Monaca, Pennsylvania. The demonstration effort
was directed to obtain information on the performance and cost of the technology regarding its
utility for treating hazardous wastes. Documentation consists of two reports: (1) an Applications
Analysis Report, which interprets the data and discusses the potential applicability of the
technology, and (2) this Technology Evaluation Report, which describes the field activities and
laboratory results.
Copies of this report can be purchased from the National Technical Information Service,
Ravensworth Building, Springfield, Virginia 22161, 703/487-4600. Requests should include the
document number found on the report's cover. Reference copies are available at EPA libraries in
the Hazardous Waste Collection. Furthermore, the SITE Clearinghouse hotline at 800/424-9346 or
202/382-3000 in Washington, D.C., can supply information about the availability of all SITE
reports.
E. Timothy Oppelt, Director
Risk Reduction Engineering Laboratory
iii
-------
ABSTRACT
This report evaluates the Horsehead Resource Development Company, Inc., (HRD) Flame
Reactor technology's ability to remove and recover volatile metals, such as lead and zinc, from
waste while producing a vitrified slag that meets applicable disposal requirements.
The HRD Flame Reactor technology is a patented high-temperature thermal process
designed to treat industrial residues and wastes containing metals. During processing, the waste
material is introduced into the HRD Flame Reactor. After introduction to the Flame Reactor, the
waste material is subjected to temperatures in excess of 2,000°C by exposure to reducing gases
produced by the combustion of solid or gaseous hydrocarbon fuels in oxygen-enriched air. At
these temperatures, volatile metals in the waste are fumed, and organic compounds should be
destroyed. The waste materials react rapidly, producing a potentially nonhazardous vitrified slag
and gases, including steam and metal vapors. Metal vapors further react in the combustion
chamber and cooling system to produce metal-enriched oxide product that is collected in the
baghouse portion of the oxide product recovery system. The resulting metal-enriched oxide
product may be recycled to recover the metals. The amount of waste reduced to oxide and slag
depends on the chemical and physical properties of the waste material.
The HRD Flame Reactor Demonstration was conducted under the SITE Program at HRD's
facility in Monaca, Pennsylvania, in the spring of 1991. During the demonstration, secondary lead
smelting slag was treated to produce a lead- and zinc-enriched metal oxide product and a slag that
was determined to be nonhazardous based on current regulatory requirements (waste extraction by
the Toxicity Characteristic Leaching Procedure [TCLP] and subsequent extract analysis). Greater
than 75 percent of the 0.0411 weight percent cadmium, 5.41 weight percent lead, and 0.416 weight
percent zinc in the waste was recovered in the potentially recyclable metal oxide product. In the
metal oxide product, concentrations of cadmium, lead, and zinc were 0.128, 17.4 and 1.38 weight
percent, respectively. The weight of the raw waste was reduced by 36.6 weight percent.
Based upon limited data, atmospheric emissions of metals from the Flame Reactor could be
of concern. The planned addition of a modern emission control should reduce this concern.
The HRD thermodynamic model was found to be appropriate to establish preliminary
operating conditions for the Flame Reactor, including waste feed and consumable flow rates.
During the SITE Demonstration, the HRD Flame Reactor experienced no major
operational problems. Several auxiliary systems, such as the oxide product recovery system,
cooling water system, and feed system experienced problems that did not affect operations of the
Flame Reactor. As these auxiliary systems are to be upgraded, their problems were not considered
to be significant.
The HRD Flame Reactor system processed waste from the National Smelting and Refining
(NSR) site, under very rigorous testing conditions, at a cost of $932 per ton. Data supplied by
HRD for other studies show that the HRD Flame Reactor can process similar waste for as little as
$208 ton.
iv
-------
TABLE OF CONTENTS
Section Page
NOTICE ii
FOREWORD iii
ABSTRACT iv
LIST OF TABLES viii
LIST OF FIGURES x
LIST OF ABBREVIATIONS, ACRONYMS, AND SYMBOLS xi
CONVERSION FACTORS xv
ACKNOWLEDGEMENTS xvi
1.0 EXECUTIVE SUMMARY 1
1.1 OVERVIEW OF THE HRD SITE DEMONSTRATION 2
1.2 TECHNOLOGY DESCRIPTION 3
1.3 ANALYTICAL RESULTS 4
1.4 QUALITY ASSURANCE PROCEDURES 6
1.5 CONCLUSIONS 6
1.6 COMMENTS 7
2.0 INTRODUCTION 9
2.1 REPORT ORGANIZATION 9
2.2 BACKGROUND 10
2.2.1 SITE Program 10
2.2.2 Technology Demonstration Program Objectives 11
2.2.3 HRD Flame Reactor Technology 11
2.2.4 Technology Evaluation Criteria 12
2.2.5 Demonstration Preparation 13
2.2.6 Project Organization and Responsibilities 13
3.0 SITE BACKGROUND 14
3.1 NSR SITE 14
3.2 HRD FACILITY 14
4.0 HRD FLAME REACTOR PROCESS DESCRIPTION 16
4.1 WASTE FEED SYSTEM 18
4.2 FLAME REACTOR 19
4.3 SLAG SEPARATOR 21
4.4 COMBUSTION CHAMBER 23
4.5 OXIDE PRODUCT RECOVERY SYSTEM 23
v
-------
TABLE OF CONTENTS
(Continued)
Section Page
5.0 DEMONSTRATION PROCEDURES 25
5.1 DEMONSTRATION PREPARATION 25
5.2 DEMONSTRATION OPERATIONS 25
5.3 SAMPLING PROGRAM 26
5.3.1 Sampling Locations 27
5.3.2 Sample Size, Sampling Frequency,
and Analytical Parameters 29
5.3.3 Sampling Methods 32
5.4 DEVIATIONS FROM THE DEMONSTRATION PLAN 37
6.0 PERFORMANCE DATA AND EVALUATION 39
6.1 METHOD SELECTION FOR SAMPLE DIGESTION 39
6.2 CHARACTERIZATION OF THE WASTE FEED 42
6.2.1 Metal Content 42
6.2.2 TCLP Results 44
6.2.3 General Chemical Analyses for
Waste Feed Characterization 44
6.2.4 Summary of the Waste Feed 48
6.3 CHARACTERIZATION OF THE METAL OXIDE PRODUCT 50
6.4 CHARACTERIZATION OF THE EFFLUENT SLAG 50
6.5 MASS BALANCE 53
6.5.1 Materials Inventory Data 53
6.5.2 Weight Reduction 56
6.5.3 Percent Recovery 56
6.6 HRD FLAME REACTOR PROCESS RELIABILITY 57
6.7 STACK MONITORING AND EMISSIONS SAMPLING RESULTS 60
6.7.1 Introduction to the BIF Rule 60
6.7.2 Metal Emissions 62
6.7.3 HC1 Emissions 62
6.7.4 Particulate Emissions 62
6.7.5 Continuous Emission Monitoring 64
6.8 HRD FLAME REACTOR OPERATIONS PARAMETERS 67
6.9 THE HRD THERMODYNAMIC MODEL 67
vi
-------
TABLE OF CONTENTS
(Continued)
Section Paee
7.0 QUALITY ASSURANCE/QUALITY CONTROL 72
7.1 ANALYTICAL METHODS 72
7.2 SAMPLE HOLDING TIMES 78
7.3 MATRIX SPIKE/MATRIX SPIKE DUPLICATE RECOVERIES 78
7.4 LABORATORY DUPLICATES 86
7.5 FIELD DUPLICATES 94
7.6 REFERENCE STANDARDS 94
7.7 FIELD BLANKS, EQUIPMENT BLANKS, AND SYSTEM BLANKS 100
7.8 BLANK AIR RUN 101
7.9 METHOD BLANKS AND REAGENT BLANKS 101
7.10 INSTRUMENT CALIBRATION 102
7.11 DETECTION LIMITS 102
7.12 REPLICATE ANALYSES 103
7.12 QUALITY ASSURANCE CONCLUSIONS 104
8.0 COST OF DEMONSTRATION 106
8.1 EPA SITE CONTRACTOR COSTS 106
8.1.1 Phase 1: Planning 106
8.1.2 Phase II: Demonstration 107
8.2 DEVELOPER COSTS 108
REFERENCES 109
Appendices
A VERSAR ANALYTICAL RESULTS
B ENGINEERING-SCIENCE EMISSIONS TEST REPORT
C HRD ANALYTICAL RESULTS
D PITTSBURGH MINERAL AND ENVIRONMENTAL TECHNOLOGY, INC.,
MINERALOG1CAL CHARACTERIZATION REPORT AND RESULTS
E TECHNICAL SYSTEMS REVIEWS
F VERSAR STANDARD OPERATING PROCEDURE FOR TO TAL CARBON
AND TOTAL ORGANIC CARBON
G STANDARD OPERATING PROCEDURE USED FOR HYDROFLUORIC
ACID DIGESTION METHOD COMPARISON TESTING
Note: Appendices A through G, as referenced here and in the body of the report,
are available in limited number. Please mail your request to:
Donald A. Oberacker or Marta K. Richards
U.S. Environmental Protection Agency
26 West Martin Luther King Drive
Cincinnati, OH 45268
vii
-------
LIST OF TABLES
Table Page
1 BOILING POINTS OF VARIOUS METALS 22
2 HRD DEMONSTRATION SAMPLING FREQUENCY 30
3 COMPARISON OF METAL DIGESTION PROCEDURES USING EPA METHOD
3050, THE HYDROFLUORIC ACID METHOD, THE LITHIUM
TETRABORATE FUSION METHOD, AND THE HRD METHOD 41
4 ANALYSIS OF THE WASTE FEED AFTER
DIGESTION BY EPA METHOD 3050 43
5 TCLP RESULTS FOR THE WASTE FEED 45
6 PARTICLE SIZE DISTRIBUTION OF THE WASTE FEED 46
7 SUMMARY OF OTHER CHEMICAL ANALYSES 47
8 ANALYTICAL DATA FOR OXIDE PRODUCT DIGESTION
BY EPA METHOD 3050 52
9 TCLP RESULTS FOR THE EFFLUENT SLAG 54
10 ANALYTICAL DATA FOR EFFLUENT SLAG
DIGESTION BY EPA METHOD 3050 55
11 MATERIALS INVENTORY DATA FROM THE HRD
SITE DEMONSTRATION 57
12 MASS BALANCE CLOSURE AND RECOVERY OF
METALS PRESENT IN THE WASTE FEED 59
13 METAL EMISSION RATES FOR THE HRD FLAME REACTOR 64
14 PARTICULATE RESULTS 66
15 RESULTS OF CONTINUOUS EMISSION MONITORING 68
16 FLAME REACTOR OPERATING PARAMETERS
FOR THE SITE DEMONSTRATION 69
17 OPERATING AND RECOVERY DATA FROM THE HRD THERMODYNAMIC
MODEL AND THE SITE DEMONSTRATION 71
18 QA OBJECTIVES FOR PRECISION, ACCURACY, COMPLETENESS, AND
DETECTION LIMITS 73
19 ANALYTICAL METHODS FOR SOLID SAMPLES 75
viii
-------
LIST OF TABLES
(Continued)
Table Page
20 ANALYTICAL METHODS FOR LIQUID SAMPLES 76
21 ANALYTICAL METHODS FOR GAS SAMPLES 77
22 DATES OF SAMPLE COLLECTION, PREPARATION, AND ANALYSIS
FOR PROCESS SAMPLES 79
23 DATES OF SAMPLE COLLECTION, PREPARATION, AND ANALYSIS
FOR STACK SAMPLES 83
24 TCLP MATRIX SPIKE RECOVERIES 87
25 WASTE FEED MATRIX SPIKE RECOVERIES 88
26 SPIKE RECOVERIES FOR STACK SAMPLES 89
27 LABORATORY DUPLICATE RESULTS FOR
GENERAL CHEMISTRY ANALYTES 90
28 LABORATORY DUPLICATE RESULTS FOR
WASTE FEED SAMPLES 91
29 LABORATORY DUPLICATE RESULTS FOR
OXIDE PRODUCT SAMPLES 92
30 LABORATORY DUPLICATE RESULTS FOR
EFFLUENT SLAG SAMPLES 93
31 COMPARISON OF CERTIFIED ASSAY VALUES AND
LABORATORY RESULTS FOR LEAD BLAST FURNACE
SLAG STANDARD REFERENCE MATERIAL 96
32 COMPARISON OF CERTIFIED ASSAY VALUES AND
LABORATORY RESULTS FOR FLAME REACTOR
OXIDE STANDARD REFERENCE MATERIAL 97
33 COMPARISON OF CERTIFIED ASSAY VALUES AND
LABORATORY RESULTS FOR EURO-STANDARD 877-1
FURNACE DUST STANDARD REFERENCE MATERIAL 98
34 COMPARISON OF CERTIFIED ASSAY VALUES AND
LABORATORY RESULTS FOR EURO-ENCHANTILLON
876-1 STANDARD REFERENCE MATERIAL 99
ix
-------
LIST OF FIGURES
Figure Page
1 HRD FLAME REACTOR PROCESS FLOW SCHEMATIC 17
2 FLAME REACTOR TWO-STAGE SYSTEM 20
3 SAMPLING AND MONITORING LOCATIONS 28
4 SCHEMATIC DIAGRAM OF THE METALS SAMPLING TRAIN 34
5 SCHEMATIC DIAGRAM OF THE HC1 SAMPLING TRAIN 36
x
-------
LIST OF ABBREVIATIONS, ACRONYMS, AND SYMBOLS
ARAR applicable or relevant and appropriate requirements
ASTM American Society for Testing and Materials
BIF Boiler and Industrial Furnace
Btu British thermal unit
Btu/lb Btu per pound
eC degrees Celsius
CAR corrective action recommendation
Cd cadmium
CEM continuous emission monitor
CERCLA Comprehensive Environmental Response, Compensation,
and Liability Act
CFR Code of Federal Regulations
CH4 methane or natural gas
cr chlorine ion
Cl2 chlorine
CO carbon monoxide
C02 carbon dioxide
CV cold vapor atomic absorption spectroscopy
dscf dry standard cubic feet
EAF electric arc furnace
EPA U. S. Environmental Protection Agency
EP extraction procedure
°F degrees Fahrenheit
Fe iron
FeO ferrous oxide
Fe203 ferric oxide
FR Federal Register
g gram
GFAA graphite furnace atomic absorption spectroscopy
gr grain
H2 hydrogen
H20 water
H202 hydrogen peroxide
xi
-------
LIST OF ABBREVIATIONS, ACRONYMS, AND SYMBOLS
(Continued)
HC1 hydrogen chloride gas or hydrochloric acid
HF hydrofluoric acid
HN03 nitric acid
HPLC high pressure liquid chromatography
hr hour
HRD Horsehead Resource Development Company, Inc.
ICP inductively coupled plasma atomic emission spectroscopy
kg kilogram
KMn04 potassium permanganate
L liter
lb pound
lb/hr lb per hour
M5 EPA Method 5
MCAWW Methods for the Chemical Analysis of Water and Wastes
(U.S. EPA, 1979)
mg milligram
mg/kg milligram per kilogram
mg/L mg per liter
mL milliliter
min minute
MS matrix spike
MSD matrix spike duplicate
N2 nitrogen
NA not applicable
Na sodium
NaOH sodium hydroxide
NC not calculable
ND not detected
NESHAPS National Emissions Standards for Hazardous Air Pollutants
ng nanogram
ng/kg ng per kilogram
NIST National Institute of Science and Technology
xii
-------
LIST OF ABBREVIATIONS, ACRONYMS, AND SYMBOLS
(Continued)
NL formerly National Lead Company
N02 nitrogen dioxide
NOx nitrogen oxides
NPL National Priorities List
NSPS new source performance standards
NSR National Smelting and Refining Company, Inc.
02 oxygen
ORD EPA Office of Research and Development
OSHA U.S. Occupational Safety and Health Administration
PaDER Pennsylvania Department of Environmental Resources
Pb lead
ppm parts per million
PRC PRC Environmental Management, Inc.
PSD particle size distribution
psi pounds per square inch
QA quality assurance
QAPP quality assurance project plan
QC quality control
RCRA Resource Conservation and Recovery Act
RD&D Research, Development, and Demonstration
RREL U.S. EPA Risk Reduction Engineering Laboratory
RFP request for proposal
RPD relative percent difference
RSD relative standard deviation
S sulfur
SARA Superfund Amendments and Reauthorization Act
scfm standard cubic feet per minute
sec second
SiOj silicon dioxide (sand)
SITE Superfund Innovative Technology Evaluation
SLS secondary lead smelter
S02 sulfur dioxide
xiii
-------
LIST OF ABBREVIATIONS, ACRONYMS, AND SYMBOLS
(Continued)
so3
sulfur trioxide
SOP
standard operating procedure
SW-846
U.S. EPA Test Methods for Evaluating Solid Waste
TC
Toxicity Characteristic
TCLP
Toxicity Characteristic Leaching Procedure
THC
total hydrocarbons
TOC
total organic carbon
TSR
Technical Systems Review
tpy
tons per year
microgram
Zn
zinc
xiv
-------
CONVERSION FACTORS
English (US)
Factor
Length:
1 inch (in)
1 foot (ft)
1 mile (mi)
x
x
x
2.54
0.305
1.61
Area:
1 square foot (ft2)
0.0929
Volume:
1 gallon (gal)
1 cubic foot (ft3)
x
x
3.78
0.0283
Mass:
1 grain (gr)
1 pound (lb)
1 ton (t)
x
x
X
64.8
0.454
907
Pressure:
1 pound per square inch (psi)
0.0703
Energy: 1 British Thermal Unit (Btu) x 1.05
1 kilowatt hour (kWh) x 3.60
Temperature: ("Fahrenheit - 32) x 0.556
Metric
centimeter (cm)
meter (m)
kilometer (km)
square meter (m2)
liter (L)
cubic meter (m3)
milligram (mg)
kilogram (kg)
kilogram (kg)
kilogram per
square centimeter
(kg/cm2)
kilojoule (kJ)
megajoule (MJ)
"Celsius
XV
-------
ACKNOWLEDGEMENTS
This report was prepared under the direction and coordination of Donald A. Oberacker and
Marta K. Richards, U.S. Environmental Protection Agency (EPA) Superfund Innovative Technology
Evaluation (SITE) Project Managers, in the Risk Reduction Engineering Laboratory (RREL),
Cincinnati, Ohio. Contributors and reviewers of this report included Gregory Carroll, Gordon Evans,
Ivars Licis, Guy Simes, and Robert Stenburg of U.S. EPA's Risk Reduction Engineering Laboratory
and John Pusateri and Regis Zagrocki of Horsehead Resource Development Company, Inc. (HRD).
This report was prepared for EPA's SITE Program by Jack Brunner, Jack Fishstrom, Tony
Gardner, Bill Jones, and Ken Partymiller of PRC Environmental Management, Inc., (PRC) and
Damian Cercone, Jeremy Flint, John Newton, Jr., and Susan Perry of Versar, Inc. PRC and Versar,
Inc., performed the process sampling; Engineering-Science, Inc., performed the stack sampling; and
Versar Laboratories, Inc., performed the chemical analyses for this SITE Demonstration.
The authors would like to acknowledge the help and support provided by the HRD Flame
Reactor Division and its operations staff in the planning and preparing for the SITE Demonstration,
the SITE Demonstration itself, and in preparing this report.
xvi
-------
1.0 EXECUTIVE SUMMARY
In response to the Superfund Amendments and Reauthorization Act of 1986 (SARA), the
U.S. Environmental Protection Agency (EPA) has established the Superfund Innovative
Technology Evaluation (SITE) Program to accelerate the development, demonstration, and use of
new or innovative technologies that offer permanent, long-term cleanup solutions at Superfund
sites. The SITE Program is administered by the EPA Office of Research and Development (ORD)
and has four primary goals:
• Identify and remove obstacles to the development and commercial use of
alternate technologies.
• Structure a development program that nurtures emerging technologies.
9 Demonstrate promising innovative technologies to establish reliable
performance and cost information for site characterization and cleanup
decision-making.
• Develop procedures and policies that encourage the selection of available
alternative treatment remedies at Superfund sites, as well as other waste
sites and commercial facilities.
As a part of the SITE Program, EPA solicits proposals from innovative waste treatment
technology developers who have expressed an interest in participating in the SITE Program. Based
on these proposals, EPA selects technologies for inclusion in the demonstration portion of the
SITE Program. One of the selected technologies is the Horsehead Resource Development
Company, Inc., (HRD) Flame Reactor.
The demonstration of the HRD Flame Reactor technology occurred at HRD's facility in
Monaca, Pennsylvania, using secondary lead smelter (SLS) rotary kiln soda slag from the National
Smelting and Refining Company, Inc., (NSR) Superfund site in Atlanta, Georgia. The goal of the
HRD Flame Reactor technology is to produce a marketable metal oxide product and nonhazardous
effluent slag.
The primary objectives of the HRD Flame Reactor SITE Demonstration include the
following:
• Evaluate the technology's ability to treat waste materials to form a
recyclable metal oxide product and a nonhazardous fused slag
1
-------
• Evaluate the system's reliability
• Develop overall economic data on the technology
Secondary objectives were also defined. These are objectives that would be of interest to
potential technology users but concern testing auxiliary systems rather than the actual Flame
Reactor. Secondary objectives include the following:
• Assess the airborne emissions from the process
• Verify the predictions of the HRD thermodynamic process model so that
the model can be used to predict costs for other projects
Due to their environmental concern (cadmium and lead), economic value (zinc), and
concentration in the waste (lead and zinc); cadmium, lead, and zinc were the volatile
metals of primary concern during the SITE Demonstration.
1.1 OVERVIEW OF THE HRD SITE DEMONSTRATION
The HRD Flame Reactor was demonstrated in February and March 1991. Seventy-two
tons of waste material from the NSR site in Atlanta, Georgia, were treated during all phases of
testing. This waste material was granular, SLS slag containing carbon, iron, sodium, sulfur, lead,
silicon, chlorine, zinc, arsenic, cadmium, and many other metals and inorganic chemical
compounds, including water. This waste material was considered a Resource Conservation and
Recovery Act (RCRA) characteristic waste because of high cadmium and lead concentrations in
the Toxicity Characteristic Leaching Procedure (TCLP) extracts of the waste. This waste was
chosen as it was readily available, it contained high concentrations of several recoverable metals
(lead and zinc), it contained no organic compounds (waste containing organic compounds could
not be handled under HRD's state permits), and it was representative of a waste type available in
large quantities throughout the country.
Once at the HRD facility, the waste material was dried and passed through a hammermill
prior to treatment in the HRD Flame Reactor. The demonstration test runs included a series of
shakedown runs to establish optimal operating conditions, a blank run with no waste treatment,
four test runs (including one that was not used for interpretation of results due to sampling
2
-------
problems), and a series of additional runs to produce effluent slag with improved durability and to
process remaining waste.
Extensive process operating data and numerous analytical samples were collected. The
operating data include raw waste feed rate, processed oxide product and effluent slag production
rates, natural gas and oxygen consumption rates, electrical consumption, and temperatures
throughout the system. Laboratory analyses of the waste feed include metals, moisture, sulfur,
chloride, fluoride, carbon, total organic carbon, and energy and ash content. Effluent samples
(oxide product from the oxide product recovery system and processed effluent slag) were analyzed
for metals. The waste feed and effluent slag were also extracted by the TCLP test and the extract
was analyzed for metals. Concentrations of carbon monoxide (CO), carbon dioxide (C02), oxygen
(Oz), nitrogen oxides (NOx), sulfur dioxide (SOz), and total hydrocarbons (THC), as well as
hydrogen chloride gas (HC1), metals, and particulate in the stack gases were also measured.
1.2 TECHNOLOGY DESCRIPTION
According to the developer, the HRD Flame Reactor is designed to thermally treat
granular solids, soil, flue dust, slag, and sludge containing metals. The goal of the treatment
process is to yields two products: (1) a heavy metal oxide product that can potentially be recycled
by metal producers, and (2) a vitrified slag that, if tested and shown to be nonhazardous, can be
used as aggregate. The high-temperature reactor processes wastes with a hot reducing gas
produced by the combustion of solid or gaseous hydrocarbon fuels in oxygen-enriched air. In the
reactor, feed materials react in less than 0.5 second, allowing high waste throughput (HRD,
1989a).
Volatile metals in the waste material, including cadmium, lead, and zinc, are vaporized
and oxidized, then captured downstream in a oxide product recovery system. Nonvolatile metals
are predominantly encapsulated in the slag product. According to HRD, for optimum reaction
conditions, the waste feed should contain less than 5 percent total moisture, and at least 80 percent
of the feed should be sized less than 200 mesh. Waste material might require pretreatment by
drying and by physical size reduction. In order to produce a fluid slag, the fusion temperature of
the nonvolatile feed materials should not exceed 1,400°C. Deviations from these specifications,
such as total moisture content up to 15 percent, higher fusion temperatures, or particle sizes up to
20 mesh, are acceptable but tend to decrease throughput and reduce the recovery of metals in the
oxide product (HRD, 1989a).
3
-------
After drying and size reduction, if required to meet the above specifications, pretreated
waste material is transferred to temporary storage bins. From the temporary storage bins, the
waste feed is transferred to the day bins, where it is metered by a screw feeder to a surge hopper
and then pneumatically injected into the HRD Flame Reactor. In the Flame Reactor, the waste
feed is heated to a high temperature (greater than 2,000°C) by the fuel-rich combustion of
natural gas or coal and oxygen-enriched air. The high temperature forces water, volatile metals,
and volatile inorganic compounds into the gas phase. Organic compounds and carbon in the waste
are believed to be totally destroyed or removed. Nonvolatile and noncombustible materials are
fused by the high temperatures and fall through the reactor into the horizontal slag separator. The
effluent slag exits through the slag tap and then cools. Gaseous matter is drawn by reduced
pressure into a combustion chamber, where air is introduced and oxidation occurs. The oxidized
gases are cooled in a heat exchanger, and the enriched-metal oxide product is collected and
recovered in a baghouse. The oxide product from this collection system is discharged through a
screw conveyor into bulk storage bags for potential recycling (HRD, 1989a).
1.3 ANALYTICAL RESULTS
During the HRD Flame Reactor SITE Demonstration, a comprehensive sampling and
analysis program was undertaken to characterize the SLS slag waste feed, the oxide product, the
effluent slag, and stack gas emissions.
The analytical results of the SLS slag waste feed show that iron, lead, silicon, and sodium
account for 28.9 percent of the slag. In addition, seven other metals (aluminum, calcium, copper,
magnesium, potassium, tin, and zinc) are present at concentrations greater than 0.1 percent. The
remainder of the slag is composed mainly of carbon, various inorganic compounds (such as
compounds containing chloride and sulfur), water, and chemically bound oxygen.
The results of the TCLP extraction and analysis show that the SLS slag waste feed is a
hazardous waste, as defined by RCRA, because both lead and cadmium leach at levels above the
allowable RCRA limits.
The data show that metals are concentrated in the oxide product. The major metal
constituents of the oxide product are lead (17.4 percent), sodium (15.7 percent), iron
(3.22 percent), and zinc (1.38 percent). Seven other metals (antimony, arsenic, cadmium, calcium,
copper, potassium, and tin) plus silicon are present at concentrations of greater than 0.1 percent.
4
-------
Notably, over 75 percent of the three target metals (cadmium, lead, and zinc) in the waste feed
was captured in the oxide product.
The effluent slag TCLP data show that, although the leachability of some metals (such as
arsenic and barium) increased as compared to the SLS slag waste feed, the effluent slag is not a
RCRA hazardous waste. In addition, data show that the predominant metals in the effluent slag
are aluminum, calcium, iron, silicon, and sodium accounting for 38.7 percent of the effluent slag
mass.
The total effluent slag and oxide product from the demonstration weighed 36.6 percent
less than the total weight of the waste feed. This weight reduction of 36.6 percent is due to the
essentially complete conversion of carbon to C02, of moisture to steam, of chloride to HC1 gas,
and of sulfur to S02.
The stack gas was sampled for N0X,S02, HC1, metals, and particulate. The results show
that emission rates for arsenic, chromium, and lead exceed the proposed Tier II screening limits
contained in the EPA document Guidance on Metal and Hydrogen Chloride Controls for
Hazardous Waste Incinerators (EPA, 1989b). In addition, HCl emissions exceed the current limits
for incinerators [40 Code of Federal Regulations (CFR) 264.343(6)] (CFR, 1988).
The only permit limitation currently placed on the HRD Flame Reactor is for 500 parts
per million (ppm) for SOz. Although the unit operated below this limit most of the time
(averaging less than 300 ppm), the limit was briefly exceeded once during a test resumption.
Information was also collected on the reliability of the HRD Flame Reactor. The HRD
Flame Reactor itself had no major operational problems affecting its performance during the
demonstration. However, according to HRD staff, the reactor's oxide product collection system
was undersized for the feed rate used during the demonstration. This caused the first sampling
run to be scrapped, because isokinetic sampling conditions could not be maintained. Additionally,
the heat exchanger and associated cooling water system developed several leaks.
5
-------
1.4
QUALITY ASSURANCE PROCEDURES
The main quality assurance (QA) objective of all SITE Demonstrations is to produce
well-documented sampling and analytical data of known quality. To accomplish this goal, a
detailed and comprehensive Quality Assurance Project Plan (QAPP) was developed before the
demonstration. The QAPP contains specific QA targets for precision, accuracy,
representativeness, completeness, and comparability. It also specifies (1) the analytical methods to
be used, (2) data for holding times, (3) number and types of blanks, (4) matrix spikes and matrix
spike duplicates, (5) laboratory duplicates, (6) reference standards, and (7) method detection
limits.
The waste feed and effluent slag from the reactor are nonhomogeneous and, because of
their physical and mineralogical properties, difficult to digest and analyze for metals. Therefore,
a method selection study was undertaken to select the best analytical method for digesting the
samples to determine metal concentration. Based on this study, a slightly modified version
(reduced sample size) of EPA SW-846 Method 3050 was chosen. However, the results of using
EPA Method 3050 digestion of chromium in a high-silicon content matrix and silicon are known
to be poor; consequently, results obtained for chromium and silicon from a digestion method
developed by HRD are also presented in this report.
1.5 CONCLUSIONS
Based on the results of the HRD SITE Demonstration, the following conclusions were
drawn concerning the performance of the HRD Flame Reactor technology.
• The HRD Flame Reactor technology processed SLS slag and produced both
a potentially recyclable metal oxide product and an effluent slag meeting
RCRA TCLP standards.
• The HRD Flame Reactor achieved a net weight reduction of 36.6 percent
when the waste feed was processed into oxide product and effluent slag.
• During the demonstration, the HRD Flame Reactor had no major
operational problems; however, auxiliary systems such as the oxide product
collection system, cooling water system, and feed system experienced
problems that did not affect the operation of the Flame Reactor.
6
-------
• The HRD thermodynamic model can be used to set preliminary operating
conditions and to determine order of magnitude estimates for parameters
used in a cost estimate, such as fuel and oxygen flow rates.
• The HRD Flame Reactor system processed SLS slag from the NSR site at a
cost of $932 per ton. Data supplied by HRD for other studies show that the
HRD Flame Reactor can process similar waste for as little as $208 ton.
• A site-specific risk analysis is required to assess the impact of the HRD
Flame Reactor stack emissions. Based on limited data, the atmospheric
emissions of metals could be a concern, however, due to data limitations,
no conclusions could be reached on metal emissions.
1.6 COMMENTS
Based on the SITE Demonstration, the following comments were noted. These comments
should be taken into account when the HRD Flame Reactor technology is considered for use to
process waste material.
• HRD recommends that 80 percent of the feed be smaller than 200 mesh
(0.0029 inch or 75 microns) for optimal recovery of volatile metals.
Coarser feed material would decrease smelting efficiency and impede slag
fusion, both of which decrease the recovery of volatile metals.
Pretreatment can be performed by HRD to meet these specifications.
• The recommended total moisture content is less than 5 percent (free and
chemically bound) as moisture consumes energy in the Flame Reactor.
HRD operates a drier for feed pretreatment and desiccation.
• The HRD Flame Reactor oxide product collection system was designed to
handle the gases generated while treating electric arc furnace (EAF) dust.
During this demonstration, SLS slag was processed under conditions which
generated larger volumes of off-gases. In addition, the heat exchanger
used at the facility is old and could operate only at about 80 percent of
capacity. Therefore, a larger volume of ambient cooling air was needed to
maintain the temperature of the inlet gas to the oxide product collection
system baghouse. As this system heated up during operation, the
introduction of even larger volumes of air were necessary to control the
baghouse temperature. Eventually, a larger volume of cooling air was
required than could be drawn through the system by the induced draft
baghouse blower, resulting in a positive pressure in the reactor. Under
positive pressure, the reactor emits CO and metal vapors through the slag
tap hole and must be shut down. Positive pressure typically shut down the
reactor after approximately 4 hours of operation. For a full-scale
Superfund remediation, the gas handling capacity of the oxide product
collection system should be increased to allow continuous operation.
7
-------
There may be a cost involved in recycling the lead oxide dust. This cost
depends on the current lead market (supply and demand for lead), the
concentration of lead as well as impurities or contaminants in the material
to be recycled, the amount of oxide product, and the cost of handling the
oxide product. It should be noted that the amount of oxide produced during
the demonstration was approximately 25 weight percent of the total dried
waste feed.
The cost and time estimates to perform Tier III metal emission assessments
are considerable. These calculations are both waste and site specific and
must done on a case by case base. Future HRD Flame Reactor work might
not be done at Monaca, Pennsylvania - and the calculations would have to
be redone. While this calculation is definitely desirable, its cost is
prohibitive and beyond the scope of this SITE project. If the reader of this
report is very interested in the HRD technology, he can contact HRD and
discuss concerns before selecting this technology.
Potential HRD technology users should be aware of, and make sure that
they satisfy the requirements of, all applicable local, state, and federal
regulations, such as RCRA revisions, the revised Clean Air Act, and state
hazardous waste regulations.
-------
2.0 INTRODUCTION
The purpose of this Technology Evaluation Report is to provide a comprehensive
description of the HRD Flame Reactor SITE Demonstration and its results. It is intended for
engineers and others making a detailed evaluation of the technology for a specific site and waste
situation. These technical evaluators can use this report to understand, in detail, the performance
of the technology during the demonstration and the advantages, risks, and costs of the technology
for a specific application. This information can be used to produce conceptual designs in
sufficient detail to make preliminary cost estimates for the demonstrated technology. For a
discussion of advantages, disadvantages, and limitations of the technology, refer to the
Applications Analysis Report for the HRD Flame Reactor.
2.1 REPORT ORGANIZATION
This report is organized in eight sections. Section 1.0 is an executive summary. Section
2.0 presents introduction and background information on the SITE Program in general, the HRD
Flame Reactor technology, and the HRD SITE Demonstration. Section 3.0 describes the hazardous
waste site and the HRD facility. Section 4.0 provides a detailed description of the HRD Flame
Reactor technology. Section 5.0 describes the demonstration operations, including demonstration
preparation and the sampling program, and provides a summary of the demonstration. Section
6.0 discusses the analytical results and performance data in relation to the objectives of the HRD
SITE Demonstration. Section 7.0 presents the quality assurance and quality control objectives and
results for the HRD SITE Demonstration. Section 8.0 presents the EPA and developer costs for
the technology demonstration. A list of references is provided at the end of this report.
This report also includes seven appendices. Appendix A is a tabulation of analytical
results from Versar, Inc. Appendix B is the Engineering-Science, Inc. emissions test results.
Appendix C is HRD's analytical results. Appendix D is the mineralogical characterization report
prepared by Pittsburgh Mineral and Environmental Technology, Inc. for HRD. Appendix E
contains the field and laboratory Technical Systems Reviews performed as a part of the HRD
Demonstration. Appendices F and G are the analytical methods employed for total carbon and
total organic carbon analyses as well as the hydrofluoric acid digestion procedure.
9
-------
2.2 BACKGROUND
Past hazardous waste disposal practices and the environmental and human health impacts
of those practices caused Congress to enact the Comprehensive Environmental Response,
Compensation, and Liability Act (CERCLA) of 1980 [reference: Public Law (PL) 96-510]. The
original act established a Hazardous Substance Response Trust Fund to handle emergencies at
uncontrolled hazardous waste sites and to clean up the sites; this fund has become known as
Superfund. EPA has investigated hazardous waste sites and established national priorities for site
cleanups. The ultimate objective of these investigations is to develop plans for permanent site
cleanups, although EPA does initiate short-term removal actions when necessary. The National
Priorities List (NPL) is EPA's list of the Nation's top-priority hazardous waste sites that are
eligible to receive federal cleanup assistance under the Superfund Program.
Congress recently expressed concern over the use of land-based disposal and containment
technologies to address problems caused by releases of hazardous substances at hazardous waste
sites. Because of this concern, the 1986 reauthorization of CERCLA, SARA, mandates that EPA
select, to the maximum extent practicable, remedial actions at Superfund sites that create
permanent solutions to the sites' effects on human health or the environment. In doing so, EPA is
directed to consider use of alternative or resource recovery technologies.
2.2.1 SITE Program
EPA has established the SITE Program to accelerate the development, demonstration, and
use of new or innovative technologies that offer permanent site cleanup. The program is
administered by ORD.
Each year EPA solicits proposals to demonstrate innovative technologies. The most
promising technologies are chosen for participation in the SITE Demonstration Program. ORD
and EPA regional personnel match these technologies with a list of potentially appropriate sites.
The Demonstration Program is designed to develop detailed and reliable performance and
cost data on the innovative alternative technologies so that potential users have sufficient
information to make sound judgments about the applicability of the technology to a specific site
and to compare it to other currently available technology alternatives. The program also identifies
10
-------
the governmental policy and regulatory requirements applicable to the technology and the
hazardous substances being treated or destroyed.
2.2.2 Technology Demonstration Program Objectives
The SITE Program mandate is to seek cost-effective alternatives to the traditional practice
of using land disposal and containment for the remediation of hazardous waste sites. To address
this mandate, the following general objectives were developed for the SITE Demonstration
Program:
• Determine the effectiveness of the process from an assessment of analytical
results
• Identify the potential need for pre- and posttreatment processing of raw
and treated materials
• Identify the types of wastes and media to which the process can be applied
• Identify the hazardous substances being treated or destroyed by the process
• Identify any potential process system operating problems and their possible
resolutions
• Determine the approximate capital, operating, and maintenance costs
• Determine the projected long-term operating and maintenance costs
• Identify governmental policy and regulatory requirements applicable to the
process
2.2.3 HRD Flame Reactor Technology
The HRD Flame Reactor technology is a patented, hydrocarbon-fueled, flash-smelting
system that treats residues and wastes containing metals. The reactor processes wastes with a hot
(greater than 2,000°C in the gas injection chamber) reducing gas produced by the fuel-rich
combustion of solid or gaseous hydrocarbon fuels in oxygen-enriched air. Within the reactor, the
feed materials react rapidly, allowing a high waste throughput. The end products are a potentially
recyclable metal-enriched oxide product and a potentially nonhazardous, fused slag (a glass-like
solid when cooled). The weight and volume reduction achieved (of waste feed to both the oxide
product and effluent slag) depends on the chemical and physical properties of the waste but
typically ranges from 10 to 40 percent (HRD, 1989a).
11
-------
According to HRD, the Flame Reactor technology can be applied to granular solids, soil,
flue dust, slag, and sludge containing heavy metals. The volatile metals in the waste material,
such as cadmium, lead, and zinc, are vaporized, then oxidized and captured downstream in a
oxide product recovery system. Nonvolatile metals are encapsulated in the slag product. At the
elevated temperature of the process, organic compounds, if present, should be destroyed (HRD,
1989a).
Under most conditions, waste pretreatment is required. For high metals recovery, the
process requires that waste feeds be dry enough (less than 5 percent total moisture) to be
pneumatically conveyed and fine enough (80 percent less than 200 mesh) to react rapidly. Waste
containing larger-sized particles (up to 20 mesh) can be processed; however, the efficiency of
metals recovery is usually reduced (HRD, 1989a).
A more detailed process description is provided in Section 4.
2.2.4 Technology Evaluation Criteria
A Demonstration Plan was prepared before the HRD SITE Demonstration (EPA, 1990).
The plan includes a Sampling and Analysis Plan designed to address the general demonstration
program objectives. The Demonstration Plan outlined the following primary objectives:
• Evaluate the technology's ability to treat waste materials to form a
potentially recyclable metal oxide product and a nonhazardous fused slag
• Evaluate the system's reliability
• Develop overall economic data on the technology
The Demonstration Plan also outlined secondary objectives. These are objectives that
would be of interest to potential technology users but concerned testing auxiliary systems rather
than the actual Flame Reactor. Secondary objectives include the following:
• Assess the airborne emissions from the process
• Verify the predictions of the HRD thermodynamic process model so that
the model can be used to predict costs for other projects
12
-------
Evaluation of the HRD Flame Reactor technology wai based on its ability to treat waste
material containing cadmium, iron, lead, zinc, and other metals at the HRD facility and produce a
potentially recyclable oxide product containing the more volatile metals and an effluent slag with
minimal leaching potential containing the less volatile metals.
2.2.5 Demonstration Preparation
HRD conducted startup testing to check the HRD Flame Reactor system for problems that
would prevent smooth operation, such as the viscosity of the slag and clumping of the waste feed.
It was determined that on-site pretreatment (drying and crushing) would be adequate.
The next phase of the demonstration was the shakedown period. The shakedown period
was used to set the feed rates, reactor temperatures, oxygen content of the combustion air, and
other operating parameters. The developer's thermodynamic process model was used to
precalculate operating set points. These set points were then adjusted during shakedown by
evaluating the condition of the slag generated.
About 4 weeks after the startup procedures and shakedown were completed, one
background run was conducted to establish a process baseline. During this run, only natural gas
was fired into the reactor without waste feed. Following the completion of the background run,
four replicate test runs were conducted.
2.2.6 Project Organization and Responsibilities
For the SITE Demonstration, EPA and HRD cooperated to collect data of known quality to
evaluate the performance of the HRD Flame Reactor technology. EPA had overall responsibility
for the project: overseeing, reviewing, auditing, and approving technical and quality assurance
aspects. HRD was responsible for providing and operating all of the demonstration equipment.
13
-------
3.0 SITE BACKGROUND
This section describes background information for the NSR site, which is the source of the
waste treated during the demonstration, and the HRD facility where the demonstration was
performed.
3.1 NSR SITE
The waste treated during the HRD SITE Demonstration was transported to the
demonstration site from the NSR Superfund site. The NSR site is located at 430 Bishop Street in
the northwest portion of Atlanta, Georgia, in an industrialized area that is intermixed with
residential communities. Approximately 1 acre of the 4-acre site is owned by the Southern
Railroad Company (which is owned by Norfolk Southern Corporation), and the remaining 3 acres
are owned by Atlanta Forge and Foundry Company. The waste treated is located on the property
belonging to Atlanta Forge and Foundry Company (NL Industries, 1989; 1990).
The facility has been operated by various owners for approximately 80 years. During a
portion of this time, lead smelting and refining activities were performed at the site. The most
recent operations at the facility involved the recovery of lead from storage batteries and other
lead-bearing scrap and secondary lead smelting activities. NSR purchased the facility from NL
Industries on June 30, 1981, and operated the facility until March 1984, at which time NSR filed
for bankruptcy. Since 1984, the facility has been inactive (EPA, 1989a; NL Industries, 1990).
During the 3 years that NSR operated the facility, approximately 350 tons of processed
rotary-kiln SLS slag from the NL Industries' Superfund site in Pedricktown, New Jersey, were
shipped to the NSR facility in Atlanta for possible recycling. This waste material was stored in
two bunkers at the NSR site. Seventy-two tons of this material were collected, loaded in bulk
storage bags in closed trailers, and manifested for shipment to the HRD facility for treatment
during the SITE Demonstration.
3.2 HRD FACILITY
The HRD Flame Reactor pilot plant is located in Monaca, Pennsylvania, and is operated
by HRD, a division of Horsehead Industries, Inc. The HRD Flame Reactor plant and associated
facilities occupy about 3 acres on a 5-acre site. The plant and facilities include the main building
14
-------
that houses the reactor; an auxiliary storage building; liquid oxygen (02)and nitrogen (N2) storage
facilities; an oxide product or off-gas handling and collection system employing a baghouse; a
cooling tower for the closed-loop, noncontact cooling water system; and a pretreatment facility
containing a waste feed dryer and a hammermill. The facility is presently operating under
authority of an EPA RD&D permit (U.S. EPA I.D. No. PAD 98 111 0570) and a Pennsylvania
Department of Environmental Resources (PaDER) hazardous waste storage and treatment permit
for research testing of EAF dust (RCRA code K061 hazardous waste), and certain characteristic
wastes. These operating permits have allowed extensive testing of the HRD Flame Reactor.
The main building, measuring 40 by 80 feet and 60 feet high, presently contains the feed
handling and storage equipment, the reactor and slag separator, the effluent slag cooling and
conveying table, the control room connected to a computer in the main office building, and the
motor control center. It also includes maintenance and spare parts storage. The auxiliary storage
building, liquid 02 storage facilities, baghouse, cooling tower, and pretreatment facility are
located in the area outside the main reactor building. Adjacent to the Flame Reactor building is
an office building housing administrative and engineering offices and the computer center.
15
-------
4.0 HRD FLAME REACTOR PROCESS DESCRIPTION
The HRD Flame Reactor technology is designed to thermally treat granular solids, soil,
flue dust, slag, and sludge containing metals. The treatment process yields two products: a heavy
metal, metal-enriched oxide product that can potentially be sold to metal producers and a
potentially nonhazardous effluent slag that can be used as aggregate. The high-temperature
reactor processes wastes with a very hot reducing gas produced from the fuel-rich combustion of
solid or gaseous hydrocarbon fuels in oxygen-enriched air. After entering the reactor, the waste
feed reacts in less than 0.5 second, allowing high waste throughput (HRD, 1989a).
Metals in the waste feed, such as cadmium, lead, and zinc, are vaporized in the Flame
Reactor and oxidized in the combustion chamber. These volatile metal oxides are subsequently
captured downstream in a oxide product recovery system. Nonvolatile metals are generally
encapsulated in the effluent slag product. For optimum reaction conditions, the waste feed should
contain less than 5 percent total moisture, and at least 80 percent of the feed should be sized finer
than 200 mesh. Waste material might require pretreatment by drying and physical size reduction.
In order to produce a fluid slag, the fusion temperature of the nonvolatile feed materials should
not exceed 1400°C. Fluxing agents (such as sand) can be added to improve effluent slag fluidity.
Variations from these specifications are acceptable but tend to decrease throughput and reduce the
recovery of metals in the oxide product.
Figure 1 presents a schematic of the HRD Flame Reactor process. The process consists of
five sections (HRD, 1989a):
• Waste Feed System
• Flame Reactor
• Slag Separator
• Combustion Chamber
• Oxide Product Recovery System
16
-------
FIGURE 1
HRD FLAME REACTOR PROCESS FLOW SCHEMATIC
WASTE FEED
FUEL
COMBUSTION AIR
COMPRESSOR
STACK
OAS
EXHAUST
OXYQEN
COMBUSTION
CHAMBER
HEAT
EXCHANGER
EFFLUENT
SLAQ
POST-COMBUSTION AIR
OXIDE PRODUCT
DAY
BIN
COOLING
WATER
SYSTEM
SLAQ
SEPARATOR
DUST
COLLECTOR
Source: EPA, 1990
17
-------
4.1
WASTE FEED SYSTEM
Waste feed system operations include (1) waste feed and solid fuel storage and handling,
(2) metering and injection of waste feed and fuel into the reactor, and (3) metering and injection
of Oz and compressed air.
The waste feed storage and handling system consists of storage facilities, portable bins,
day bins, and several conveying systems. The waste material to be fed into the reactor can be
delivered to the site by rail or by truck. The waste material is stored in a storage building next to
the main building prior to processing. If pretreatment of the waste (drying and crushing) is
necessary, the waste is transferred to another building that contains the pretreatment equipment.
After pretreatment, a loader empties the feed material into portable bins, which are moved to the
Flame Reactor building. From the portable bins, waste is transferred to the day bins. The
portable bins are placed on a discharge stand and the bottom discharge slide-gate is opened.
Dusting is controlled by a seal located between the gasketed opening of the stand and the flange of
the portable bin slide-gate. The waste is fed into a screw conveyor that empties into the tubular,
day bin filling system.
Of the three day bins, two are used for waste, and one is used for solid fuel. Normally,
solid fuel such as coal fines can be used to reduce costs; however, natural gas was chosen for the
HRD SITE Demonstration because (1) it is more likely to be used in a site remediation and (2) it
has a uniform composition. Each day bin has a capacity of 150 cubic feet and is mounted on a set
of three, shear-beam, load cells that measure the day bin weight for inventory and process
control. Material from each day bin is metered and pneumatically injected into surge hoppers
prior to entering the reactor.
To calculate waste feed rate, the process control system records the loss of weight over
time. The system uses a 10-minute average waste feed rate to control the feed system. Material is
discharged from a day bin, through a live-bottom feeder, into a surge hopper set above a variable
speed screw feeder with a rated capacity of 60 pounds per minute. The feeder accurately controls
the flow of material into the reactor via a 2-inch, pneumatic injection line.
The gases used in the demonstration were 02, ambient air, and natural gas. Oz is stored
on-site as a cryogenic liquid in a 9,000-gallon storage tank. It is used to enrich the ambient air
for combustion. Compressed air produced by a compressor, operating at 1,000 standard cubic feet
18
-------
per minute (scfm) at 40 pounds per square inch (psi), is used to combust the fuel. A second
compressor, operating at 100 psi provides air for process controllers and for injecting solids into
the reactor. Natural gas, supplied by pipeline from the local utility company, was used as the fuel
during this demonstration. A 6,000-gallon liquid N2 tank is also on-site, but it was not used in the
demonstration. The N2is used to blanket coal fines as an added safety feature when coal is used
as the fuel source.
4.2 FLAME REACTOR
The HRD Flame Reactor, shown on Figure 2, is a two-stage system consisting of a fuel
burner system (first stage) and the metallurgical reactor (second stage). Carbon-based combustion
and gasification reactions occur in the burner system, followed by metal smelting reactions in the
metallurgical reactor. The reactor is 15 feet tall, positioned vertically, with an internal diameter
of 23 inches (HRD, 1989b).
The fuel burner system consists of a mixing head, upper pilot, lower pilot, and gas
injection chamber. In the mixing head, fuel and 02-enrichedair (typically 50 percent to
80 percent 02by volume) are mixed. This fuel-rich mixture then ignites in the upper pilot and is
stabilized by expansion into the lower pilot. Injecting 02-enriched air in the gas injection
chamber helps control the reducing conditions, adjust the stoichiometry (CO:COz ratio), and
further stabilize the flame in the Flame Reactor. Because highly 02-enriched air is used, flame
temperatures greater than 2,000°C are realized in the Flame Reactor. A different burner design
is employed when solid fuel is used as the energy source.
Fine, dry, waste feeds containing metals are metered with a screw feeder and
pneumatically injected into the reactor (second stage) at a location just below the exit of the
burner (see Figure 2). The waste feed reacts in the high-temperature, reducing gas stream. CO
from the incomplete combustion of the fuel reduces the metal compounds in the waste feed by the
following reactions:
Combustion of natural eas fCH^)
CH4+ 3/20z - CO + 2H20
CH4 + 202 - C02 + 2H20
CH4 + C02 + 02 - 2CO + 2H20
CH4 + 1 /202 CO + 2H2
19
-------
FIGURE 2
FLAME REACTOR TWO-STAGE SYSTEM
OXYGEN-ENRICHED AIR
NATURAL OAS
OXYGEN-ENRICHED AIR
WASTE FEED
MIXING
HEAD
UPPER PILOT
LOWER PILOT
GAS INJECTION
CHAMBER
BURNER
SYSTEM
(STAGE I)
FEED INJECTION
CHAMBER
REACTOR
)» METALLURGICAL
REACTOR
(STAGE II)
SAMPLING
CHAMBER
TRANSITION SECTION
(TO SLAG SEPARATOR)
J
Source: EPA, 1990
20
-------
Reduction/Smelting of Volatile Metals
Iron:
Zinc:
ZnO + CO - Zn (vapor) + C02
CdO + CO - Cd (vapor) + COz
Fe304+ CO - 3FeO + C02
Cadmium:
Zinc-Iron:
ZnFe204+ 2CO - Zn (vapor) + 2FeO + 2C02
PbS04 + 2CO - Pb (vapor) + SOz + 2COz
PbS04 + CO - PbO + S02 + C02
Lead:
Lead:
The nonvolatile components of the waste feed fuse, forming the effluent slag.
The energy required for fusion and reduction lowers the temperature to between 1,500°C
and 1,700"C. In this temperature range, several elemental metals are above their boiling point,
shown in Table 1, and volatilize into the gas stream. Recovery of cadmium, lead, and zinc is of
particular interest because of their economic value.
The reactor vessel is water-cooled to assure that a layer of the molten slag solidifies on the
inner reactor walls. The slag layer protects the reactor walls from intense heat and reduces the
reactor heat loss. Molten material is conveyed down the reactor walls by gravity and by the
combustion gases. At the end of the reactor, the molten metal is accelerated through a tapered
transition section into the horizontal slag separator (HRD, 1989a).
4.3 SLAG SEPARATOR
The reactor continuously discharges material into a refractory-lined, water-cooled
cyclonic separator that separates molten slag from reactor off-gases. Off-gases contain mainly
CO, hydrogen (H2),and any metal vapors recovered from the waste feed. The effluent slag
contains 35 to 75 percent of the mass of metals from the waste feed.
The slag separator is positioned horizontally between the flame reactor and the combustion
chamber (see Figure 1). The gases, particulate, and metal vapors flow toward the combustion
chamber, countercurrent to the slag. The molten slag runs out through a tap hole on the discharge
end of the unit. Occasionally a small amount of effluent slag is carried over to the combustion
chamber.
21
-------
TABLE 1
BOILING POINTS OF VARIOUS METALS
Metal
Boiling Point (4C)
Aluminum
1501
Antimony
1,380
Arsenic
6132
Barium
1,640
Cadmium
765
Calcium
1,490
Chromium
2,670
Copper
2,600
Iron
2,750
Lead
1,740
Magnesium
1,110
Mercury
357
Nickel
2,730
Potassium
774
Selenium
685
Silver
2,210
Sodium
883
Thallium
1,460
Tin
2,260
Zinc
907
Note:
1 Decomposes
2 Sublimes
#C = degrees Celsius
Source: CRC Handbook of Chemistry and
Physics. 71st Edition, 1990-1991.
22
-------
4.4
COMBUSTION CHAMBER
The slag-free reactor off-gases are combusted again with air in a refractory-lined
combustion chamber. The metal vapors oxidize and condense as solids, while combustible gases
such as CO and H2arc burned. The gas stream from the combustion chamber includes metallic
oxides, C02, water (II20), sulfur trioxide (S03),and NOx. For the SITE Demonstration, the
temperature of the off-gases after the combustion chamber was typically between 700 and
1,000°C (HRD, 1989b). Reactions in the combustion chamber include:
CO + 1/20, - C02
H2+ l/202- H20
Metal (vapor) + l/202 - MetalO
S02+ 1 /202 - S03
MetalO + S03 -> MetalS04
N2 + x02 - 2NOx
4.5 OXIDE PRODUCT RECOVERY SYSTEM
The metal oxide product (MetalO) recovery system is designed to cool the gas stream and
capture the metal oxides formed in the combustion chamber. The system consists of a heat
exchanger, a tempering air inlet damper, and a baghouse for dust collection. A fan, located
between the baghouse and the stack, provides the induced draft to power the system.
The gas is cooled by a shell-and-tube heat exchanger and by the addition of ambient air.
The heat exchanger has water on the shell side and hot gases on the tube side. The addition of
ambient air is controlled by a damper. The damper is located just before the baghouse and is used
to maintain the baghouse temperature below 200°C. Because of the typically high particulate
level in the gas stream, the heat exchanger tubes require frequent cleaning (HRD, 1989a).
The main component of the oxide product recovery system is a jet-pulsed baghouse
designed to recover metal oxide product from the gas stream. The recovery system emits
off-gases through the plant stack and discharges metal oxide product into enclosed bulk storage
bags for recovery. A rotary air lock, screw conveyors, and a sealed boot connection reduce the
possibility of fugitive emissions. The baghouse collects the oxide product dust on 8,900 square
feet of cloth.
23
-------
The bag cleaning procedure consists of short, high-pressure pulses of air through the bags
to dislodge the particles trapped on the surface. The pulses are initiated on a timed cycle based on
typical gas flow rates and dust loadings. The particles fall by gravity into a screw conveyor below
the baghouse bags. The screw conveyor transports the oxide product in one of two enclosed bulk
storage bags. While one storage bag is filling, the other can be removed and replaced with an
empty storage bag.
According to HRD the oxide product from the baghouse contains approximately 25 to
65 percent of the mass of the waste feed. This percentage is highly dependent upon the amount of
volatile metals in the waste feed. Specific recoveries for volatile metals are generally very high.
Based on past testing, the baghouse oxide product accounts for greater than 90 percent of the
volatile metals in the waste feed. The remainder is encapsulated in the effluent slag, with a
minimal fraction lost to the atmosphere as stack emissions (HRD, 1989a).
24
-------
5.0 DEMONSTRATION PROCEDURES
This section contains a discussion of the demonstration preparation, operations, and
sampling program.
5.1 DEMONSTRATION PREPARATION
Prior to the demonstration, a Demonstration Plan for the HRD Flame Reactor was
prepared (EPA, 1990). The plan consists of four sections. The first section provides background
information for the SITE Program and the HRD Flame Reactor technology. The second section
contains the operating plan for the demonstration. The third section is a QAPP that presents
testing and quality assurance objectives, a detailed sampling plan, and quality control measures.
The fourth section contains a Health and Safety Plan. The Demonstration Plan was approved by
RREL's quality assurance officer.
5.2 DEMONSTRATION OPERATIONS
Upon arrival at the demonstration site, the SLS slag waste, obtained from the NSR site,
was transferred to the feed preparation equipment. The feed preparation equipment is designed
to dry and crush the material, and a dryer and a hammermill are used. Once the SLS slag was
adequately dried and crushed for use as waste feed, it was loaded into portable bins and
transported to a storage facility adjacent to the Flame Reactor building.
Startup testing of the demonstration equipment began after the feed preparation
equipment had processed a sufficient quantity of SLS slag to produce 5 to 10 tons of dry, crushed
waste feed material. Prior to initial system startup, EPA and the SITE Team reviewed the
Demonstration Plan with HRD personnel. During startup, the HRD Flame Reactor system was
checked for any problems that would prevent smooth operation of the equipment, such as the
viscosity of the slag and clumping of the waste feed. No problems were found.
The next phase was the shakedown period. The feed rates, reactor temperatures, oxygen
content of the combustion air, and other operating parameters were set. A thermal chemical
process model was used to precalculate operating set points. The set points were then adjusted
during these runs by evaluating the condition of the effluent slag generated. The production of a
free-flowing, low-lead content effluent slag indicates the attainment of the proper values for the
25
-------
set points. Ihe actual demonstration was performed with the values set during the shakedown
runs.
Approximately 4 weeks after the startup procedures and shakedown were completed, the
demonstration began. The first run was a background run to establish a process baseline. During
the background test, only natural gas was fired in the Flame Reactor; no waste feed was admitted
to the system. Only stack gas samples were collected during the background run.
Waste feed processing commenced following the completion of the background testing.
Four test runs were conducted. The results first run were invalid due to fluctuations in the stack
gas temperature, pressure, and flow rate. During all test runs, solid samples were collected for
6 hours while the stack samples were collected for approximately 2-1/2 hours during the middle
of each sampling day. During all three runs, the Flame Reactor had to be shut down to allow the
oxide product collection system to cool.
During the four test runs, samples were collected as they entered or exited the system at
various points. These samples included waste feed, effluent slag, oxide product, and stack gas
emissions. The number of samples collected at each location, the frequency, and the rationale for
sampling and analysis parameters are discussed in Section 3.4 of the Demonstration Plan and in
Section 5.3 of this report.
After the demonstration was complete, the remaining waste feed was processed through
the HRD Flame Reactor system. Although detailed chemical analyses were not performed on the
samples from these runs, the effluent slag was analyzed and found to be a nonhazardous waste
and, therefore, could be disposed of in a nonhazardous waste landfill. The oxide product was
analyzed and found to be enriched in the volatile metals. Several potential recycling opportunities
are being explored for this lead-rich product.
5.3 SAMPLING PROGRAM
The primary objective of the demonstration was to collect sufficient data so that the
effectiveness of the HRD Flame Reactor could be evaluated for (1) removing volatile metals from
high metal-bearing wastes to produce a potentially recyclable, metal-enriched oxide and
(2) producing a nonhazardous fused slag. The following sections describe the specific sampling
objectives and sample point locations.
26
-------
5.3.1
Sampling Locations
To evaluate the effectiveness of the HRD Flame Reactor for treating waste from the NSR
site, samples were collected from several points. Figure 3 is a block diagram of the HRD Flame
Reactor, showing the locations where the following process operating parameters were monitored
and recorded:
Monitoring
Location Description
1 Natural gas feed rate (scfm)
1 Oxygen feed rate (scfm)
1 Air feed rate (scfm)
1 Waste feed rate (Ib/hr, lb/test)
1 Reactor heat loss (Btu/hr)
2 Effluent slag flowrate (lb/hr, lb/test)
3 Combustion chamber temperature (°C)
4 Temperature of cooling chamber exit gas (°C)
5 Oxide product flowrate (lb/hr, lb/test)
6 Stack temperature ("C)
Also shown on Figure 3 are the locations of each sample point:
Sampling
Location Sample
1 Waste feed
2 Effluent slag from slag separator
5 Oxide product
6 Stack gases to determine particulate and metal
emissions
6 Stack gases to determine HC1 emissions
6 Continuous emissions monitoring
(CO,, CO, 02, NOx, S02, THC)
Samples of the waste feed (Point 1, Figure 3) were collected from the surge hopper located
above the screw feeder and below each feed bin. Samples of the effluent slag from the slag
separator (Point 2, Figure 3) were collected from the shaker conveyer before the slag was
discharged to a collection bin. Samples of the oxide product (Point 5, Figure 3) were collected at
the discharge end of the screw conveyor located along the bottom of the baghouse.
27
-------
FIGURE 3
SAMPLING AND MONITORING LOCATIONS
O
NabralGas
Oxygen
Air
Waste Fe«o
Combust'©*
A*
Non-Comae!
Cooling
Water
Cooling
Air
Rame
Slag
Corrbustten
CooKflg
CKambttf/
Bsghouse
Abactor
Sep* a tot
Chamber
H*at
S*dwingBr
©
©
Effluent Slag
©
Non-Con lacl
Ceotifig
WlWf
T
usa
Pr<
©
Bagfiousa DusC
(Oxkte Product)
to
s
T
A
C
K
/'////
Source: EPA, 1990
28
-------
Samples of the fuel feed (natural gas) were not collected. Instead, copies of ..he vendor
specifications for the natural gas supplied for this demonstration were obtained.
Samples of the gases (Point 6, Figure 3) exiting the baghouse were sampled in a 48-inch
circular duct. The gas samples were obtained by traversing the duct using the two existing
sampling ports (90° apart) on a 10-foot 2-inch straight piece of the duct. Twenty-four sampling
points were used (12 per traverse).
Measurement of waste feed rates (Point 1, Figure 3) for inventory and process control was
accomplished using load cells. The output of the load cell was recorded during testing. Effluent
slag and oxide product weights were recorded using a floor scale in the Flame Reactor building.
The floor scales and load cells were calibrated several times by adding known weights to the units.
Measurement of temperatures throughout the HRD Flame Reactor facility was performed
by the HRD computerized process control system.
Equipment and impinger solution blank samples for quality assurance purposes were also
collected. These included a system blank of an EPA modified Method 5 metals train that was
assembled and leak-tested; blanks of the impinger solutions and filter and probe washes were
collected also. Other quality assurance samples included reagent blanks of the test reagents, such
as sodium hydroxide (NaOH), acetone, water, nitric acid (HN03), hydrogen peroxide (H202), and
potassium permanganate (KMn04).
5.3.2 Sample Size, Sampling Frequency, and Analytical Parameters
A summary of the sampling program is presented in Table 2. Prior to collecting the
samples shown in this table, the HRD Flame Reactor operated with no waste feed (only natural
gas and air/oxygen were introduced into the reactor). During this period of time, a single
background stack gas sample was collected. The background sample required approximately
2-1/2 hours to collect and consisted of 120 minutes of continuous emission monitoring for SOjand
NOxand stack gas sampling for metals, particulate, and HC1 using two EPA Method 5 stack
sampling trains. After the background sample was collected, four test runs, over 4 days, were
conducted processing waste feed. The solid samples were collected over a period of 6 hours.
About 2 hours after each run began, stack gas sampling was begun. Continuous emission monitors
29
-------
TABLE 2
IIRD DEMONSTRATION SAMPLING FREQUENCY
(sheet 1 of 2)
Number
Total
of
number
samples
Number
of
Analysis
Sample description
Sample type
Sampling frequency
per run
of runs
samples
1. Waste feed
Composite1
One grab every 15 minutes
6
3
18
Total chloride, Btu content, metals2, percent
(hourly)
ash, water content (moisture), particle size
distribution, teachable metals3, TOC4, total
carbon, sulfur, chloride, fluoride
2. Effluent slag
Composite1
One grab every 15 minutes
6
3
18
Metals2, leachablc metals3
(hourly)
3. Oxide product
Composite
One grab every 15 minutes
1
3
3
Metals2
(daily)
4. Stack gas (M5-metals)
EPA Method 5
120 minutes5
a. Acetone probe wash
Liquid
1
4
4
Particulate, metals6
b. HNOj probe wash
Liquid
1
4
4
Metals6
c. Particulate filter
Solid
1
4
4
Particulate, metals6
d. Impingers 1-3
Liquid
1
4
4
Metals7
e. 50 mL from impingers 1-3
Liquid
1
4
4
Mercury
f. Impinger 4
Liquid
1
4
4
Mercury
5. Stack gas (HC1)
EPA Method 5
120 minutes5
a. Impingers
Liquid
1
4
4
Total chloride, moisture
6. Blanks/Background
a. Alkaline solution (NaOH)
Grab
Once before test
1
1
1
Metals3
b. Acetone wash
Grab
Once before test
1
1
1
Metals3
c. Water
Grab
Once before test
1
1
1
Metals3
d. HPLC water8
Grab
Once before test
1
1
1
Metals3
e. HNOs
Grab
Once before test
1
1
1
Metals3
f. h2o2
Grab
Once before test
1
1
1
Metals3
g. KMn04
Grab
Once before test
1
1
1
Metals3
-------
TABLE 2
HRD DEMONSTRATION SAMPLING FREQUENCY
(sheet 2 of 2)
Number
Total
of
number
samples
Number
of
Analysis
Sample description
Sample type
Sampling frequency
per run
of runs
samples
7. System blank (M5-metals)
EPA Method 5
a. Acetone probe wash
Liquid
Once after all runs
1
1
1
Particulate, metals6
b. HNOj probe wash
Liquid
Once after all runs
1
1
1
Metals6
c. Particulate filter
Solid
Once after all runs
1
1
1
Particulate, metals6
d. Impingers 1-3
Liquid
Once after all runs
1
1
1
Metals6
e. 50 mL from impingers 1-3
Liquid
Once after all runs
1
1
1
Metals6
f. Impinger 4
Liquid
Once after all runs
1
1
1
Metals6
8. System blank (M5-HCI)
EPA Method 5
a. Impingers
Solid
Once after all runs
1
1
1
Total chloride, moisture
9. Duplicate samples
a. Waste feed
Solid
One during Run 2
1
1
1
Archived
b. Effluent slag
Solid
One during Run 2
1
1
1
Archived
c. Oxide product
Solid
One during Run 2
1
1
1
Archived
10. Decontamination water
Liquid
Once after demonstration
1
1
1
Metals2
11. Equipment rinsate blanks
Liquid
Once during each run
1
3
3
Metals2
Notes:
1 Composite samples consisted of a grab sample taken every 15 minutes. Every four samples were homogenized and split into sample aliquots for each hour of sampling. Field
duplicate samples were collected from individual grab samples.
2 Metals included aluminum, antimony, arsenic, barium, beryllium, cadmium, calcium, chromium, copper, iron, lead, magnesium, manganese, mercury, potassium, selenium, silicon,
silver, sodium, thallium, tin, and zinc.
3 Leachable metals as extracted by the TCLP included arsenic, barium, cadmium, chromium, lead, mercury, selenium, and silver.
4 Total organic carbon
5 All stack sampling was performed concurrently during each 120-minute sampling episode.
6 Metals analysis included antimony, arsenic, barium, beryllium, cadmium, chromium, lead, mercury, silver, and thallium.
7 Metals analysis included all metals listed in note (3) except mercury.
8 High pressure liquid chromatography grade (high purity) water
Source: EPA, 1990
-------
(CEM) operated for all 6 hours of each run. Run 1 results were discarded because of fluctuations
in the stack gas temperature, flowrate, and pressure.
As shown in Table 2, the waste feed and effluent slag from the slag separator were
collected every 15 minutes. Sampling of the waste feed began at the start of each test run.
Sampling of the effluent slag from the slag separator was initiated 15 minutes after each run
began to account for the residence time of the molten slag in the Flame Reactor, in the slag
separator, and on the water-cooled conveyor.
Hourly composite samples of the waste feed and effluent slag from the slag separator were
made by compositing the four samples taken during each hour. Therefore, a total of six composite
samples were collected for each test run.
A subsample of the oxide product was collected every 15 minutes. At the end of each run,
a composite sample was made from the 24 subsampies.
Gas samples for metals, particulate, and HC1 were collected at the outlet of the baghouse.
CEM samples were collected at a location on the baghouse outlet duct near the point where the
stack gas samples were collected.
5.3.3 Sampling Methods
The following sections describe the sampling collection procedures that were followed to
collect the solid and gas samples.
5.3.3.1 Solids Sampling
(1) Waste Feed. Samples of the waste feed were collected from each of the two feed
bins through one of several ports on the surge hoppers above the screw feeders. Every 15 minutes
the port was opened, and a small sample scoop was used to collect a sample from the surge hopper.
The sample was placed in a clean plastic container and temporarily sealed to prevent
contamination. After four samples were collected, they were mixed thoroughly, and an hourly
composite sample was prepared. Each run produced six composite samples.
(2) Effluent Slae from the Slag Separator. Samples of the effluent slag were collected
32
-------
with a stainless steel sample scoop. Care was taken to collect the effluent slag from various
locations and at regular intervals on the shaker conveyer so that representative samples were
collected. A grab sample was taken every 15 minutes. Each sample was allowed to cool in a
stainless steel bucket that was covered with clean aluminum foil to prevent contamination. After
cooling, the sample was placed in a clean sample jar to achieve constant sample volume prior to
compositing. Some size reduction was required to place the effluent slag in the sample jars. A
small rubber mallet was employed to reduce pieces to no larger than 1/4-inch in size. Every
effort was made to include all particles, including very fine particles, with the collected samples.
Further size reduction was accomplished in the laboratory by a mortar and pestle prior to analysis.
Hourly composite samples were taken from four consecutive grab samples, for a total of six
composite samples per run.
(3) Oxide Product. Samples of the oxide product were obtained with a scoop from the
discharge end of the screw conveyor located at the bottom of the baghouse. A grab sample was
taken every 15 minutes. After all of the grab samples were collected, they were homogenized by
manually mixing with a plastic trowel. A single composite sample was prepared from the grab
samples. Therefore, one sample was collected and analyzed for each run.
5.3.3.2 Gas Sampling
(1) Stack Gases to Determine Metals Emissions. During all runs, two EPA
Method 5 (M5) stack gas sampling trains were used. The first M5 train was modified and used to
sample for metals. The second M5 train was used to sample for the stack HC1 concentration.
Total particulate loading was determined in the metals train. This section provides a description
of the metals/particulate stack sampling train. Subsection (2) below provides a description of the
HC1 stack sampling train.
Sampling for metals was conducted using a modified version of the Method 5 (M5)
sampling train. This method has been proposed for acceptance by the EPA and is currently being
validated (EPA, 1991a). A schematic of the sampling train is shown on Figure 4. A detailed
description of this method is contained in the Demonstration Plan (EPA, 1990).
The front half of the sample train was similar to a standard M5 train, except that all of the
components that come into contact with the sampled gas were constructed of glass and Teflon®
(with the exception of a stainless steel nozzle). The probe and filter holder were borosilicate glass.
33
-------
FIGURE 4
SCHEMATIC DIAGRAM OF THE METALS SAMPLING TRAIN
• thermometer
HEATED
AREA \
THERMOCOUPLE
"S*TYPE PILOT
THERMOCOUPLE
/
HECK VALVE
QUARTZ FIBER
PARTICULATE FILTER
PROBE
STACK
WALL
PILOT
MANOMETER
SILICA GEL
ICE BATH
JOOfrt
103 ml 4*,KMnQ,/10%HjSG
5*- HNCyiO%H^L
EMPTY
THERMOCOUPLES
BYPASS
VALVE
VACUUM
ORIFICE
VALVE
DRY
GAS
METER
AIRTIGHT
PUMP
Source: EPA, 1991b
34
-------
A Teflon*-coated screen was used to support the filter. Both probe and filter holder were heated
to 120°C ± 14°C to prevent moisture condensation. A high-purity, high-efficiency
(99.95 percent determined on 0.3-micron dioctyl phthalate particles) quartz fiber filter without
organic binder was used.
This sampling train is designed to capture metals in both the solid and vapor phases. Solid
samples were collected as particulate in the filter and probe rinse. The back half of the impinger
train was modified to capture the metals in the vapor phase. The impingers were arranged as
follows:
1. Empty knockout impinger (Modified Greenburg-Smith)
2. Greenburg-Smith impinger containing a 100 mL solution of 5 percent HN03and
10 percent HjOj
3. Same as Number 2 above
4. Greenburg-Smith impinger containing a 100 mL solution of 4 percent KMn04and
10 percent H2S04
5. Modified Greenburg-Smith impinger containing about 200 grams of indicating
silica gel
An ice bath was used to maintain an impinger exit temperature of less than or equal to
20°C. All connections were made with leak-tight, ground glass joints.
The flue gas was withdrawn from the stack isokinetically for 120 minutes collecting a gas
sample of at least 50 dry standard cubic feet (dscf). Nominally, the expected stack flow rate was
13,000 dscf. At the end of the test run, the train was checked for leaks and then disassembled
into three parts: (1) probe, (2) filter and connecting glassware, and (3) impingers. Each part of
the sample train was sealed to prevent sample loss or contamination. These parts of the sample
train were then transferred to the field laboratory for recovery.
(2) Stack Gases to Determine HC1 Emissions. The HC1 sampling method involved
pulling an integrated gas sample from the stack through distilled water. Sampling of HCl was
performed concurrently with stack sampling for metals. A schematic diagram of the EPA
Method 5 sampling train that was used for HCI sampling is shown on Figure 5.
35
-------
FIGURE 5
SCHEMATIC DIAGRAM OF THE HCI SAMPLING TRAIN
THERMOMETER
THERMOCOUPLE
*8* TYPE PILOT
PILOT ,
MANOMETER
THERMOCOUPLE
CHECK VALVE
HEATED
AREA
OUART7 FIBER
PARTICULATE FILTER
STACK
SILICA GEL
ICE BATH
ICOrr/
0 I N NaOH
EMPTY
IX ai
distilled water
THERMOCOUPLES
VALVE
BY PASS
ORIFICE
VALVE
METFH
AIRTIGHT
PUMP
.VACUUM
LINE
Sourc*: 40 CFR 60, Appsndlx A
36
-------
The sampling train was placed in a port located at the same height on the stack as the
metals M5 train. The sample was introduced to the train through a heated borosilicate glass probe
and a quartz fiber particulate filter before entering the impingers. Although a particulate filter
was required in this sampling train, it was not used to measure total particulate. The sample was
collected at a single point.
Four impingers were used to collect the HC1 sample. The first impinger was empty. The
second impinger contained 100 mL of distilled water, and the third impinger contained 100 mL of
0.1 normal (N) sodium hydroxide. The final impinger contained silica gel as a final water catch.
The gas then proceeded through a vacuum pump, flow rate meter, and dry gas meter.
The HC1 sample was collected for 120 minutes at a nominal flow rate of 1 cubic foot per
minute. After each test was completed, the volume of liquid in each impinger was determined by
pouring the contents into a graduated cylinder. The final impinger containing the silica gel was
weighed. The total stack gas moisture was determined as the sum of the impinger contents (less
the 100 mL of NaOH and 100 mL of distilled water) and the moisture in the silica gel. The first
three impingers were then recovered into a leak-free sample bottle, followed by a water rinse of
each of the impingers and the connecting glassware.
In the distilled water, the HCl gas was solubilized and formed chloride ions (CI-). The
concentration of the chloride ions in the sample was then determined in the laboratory. Reagent
blanks of the distilled water were collected for analysis with the flue gas samples.
(3) Continuous Emission Monitoring. CEMs were used to continuously monitor the
concentration of SOz, NOx, CO, C02,02, and THC in the flue gas. A detailed description of the
CEMs is in the Engineering-Science report, Emissions Testing at Horsehead Resource
Development Company (Appendix B of this report). The NOx analyzer was calibrated by EPA
Method 7E [40 CFR 60, Appendix A, July 1989]. The 02and C02 analyzers were calibrated using
EPA Method 3A [40 CFR 60, Appendix A, July 1989]; the CO analyzer by Method 10 [40 CFR
60, Appendix A, July 1989]; the THC analyzer by Method 25A [40 CFR 60, Appendix A, July
1989]; and the SOz monitor by Method 6C [40 CFR 60, Appendix A, July 1989],
5.4 DEVIATIONS FROM THE DEMONSTRATION PLAN
During the demonstration, one major change to the HRD Demonstration Plan and two
37
-------
minor changes to the air sampling procedures were made.
The demonstration plan specified that three background stack samples would be collected.
However, before the demonstration began, the EPA Project Managers, with the approval of the
EPA quality control manager and auditors, decided that only one blank run would be conducted.
Although the auditors commented that one run would not be statistically significant (a minimum
of three is required for statistical significance), the EPA Project Managers and the technology
vendor agreed that, because the background run was not representative of background
contamination, only one background run would be conducted and that the three treatment runs
would not be corrected for background contamination.
The EPA Method 5 stack sampling procedures had to be modified in the field because the
Flame Reactor off-gas temperature gradually increased during the first run. Because the
increasing temperature led to increasing gas velocities, isokinetic sampling (required by EPA
Method 5) could not be maintained. After Run I was discarded because of this problem, the
required sample volume was decreased from 100 ft8 to 50 ft8. This decision traded reduced
detection limits for valid cu sampling. At this time, glass sampling probe nozzles were replaced
with stainless steel nozzles.
The HC1 stack gas sampling procedure was also modified. The QAPP specified that the
HC1 train would be operated at a nominal flow rate of 1 scfm for 120 minutes. After the second
run, the pH in the final impinger was slightly acidic (pH 5.5), another impinger was added
(100 mg of NaOH) for Run 3. After Run 3, the pH was checked again and found to be slightly
acidic (pH 5). For Run 4, the additional impinger was used and the sampling time reduced to
60 minutes. The Run 4 final pll was 7. This should not be interpreted to mean that there may
have been losses of HC1 during sampling. HC1 is a strong acid and is stable in solution up to
37 percent. (The new EPA-approved Method 26 protocol for sampling HC1 and Cl2 (chlorine gas)
uses an acidic solution to trap HC1.) Therefore, even though acid conditions existed in the
impingers, all the IIC1 was captured and results were not impacted.
38
-------
6.0 PERFORMANCE DATA AND EVALUATION
To evaluate the technology's performance with respect to the demonstration objectives (see
Section 2.2), detailed analytical and performance data were gathered and evaluated. This section
discusses these analytical and performance data.
Subsections are as follows:
• Method Selection for Sample Digestion -- discussion of how the digestion
procedure for the metal analysis was chosen
• Characterization of the Waste Feed — presentation of the analyte
concentrations in the waste feed, the TCLP results for the waste feed, and
the results of particle size distribution, moisture, and chemical constituent
analysis
• Characterization of the Metal Oxide Product -- presentation of the analyte
concentrations for the oxide product and the percent recovery of the
volatile metals
• Characterization of the Effluent Slag — presentation of the TCLP and
analyte concentrations for the effluent slag
• Mass Balance -- discussion of the mass balance performed on the system
and the weight reduction of the oxide product and effluent slag, as
compared to the waste feed
• HRD Flame Reactor Process Reliability — discussion of the HRD Flame
Reactor reliability
• Stack Monitoring and Emissions Sampling Results -- presentation of the
airborne emissions from the HRD Flame Reactor, including metals, HC1,
S02, NOx, and particulate emissions
• HRD Flame Reactor Operating Parameters — presentation of the ranges of
operating parameters of the HRD Flame Reactor during the SITE
Demonstration
• The HRD Thermodynamic Model -- discussion of the accuracy of the
HRD model in predicting the operation of the HRD Flame Reactor while
processing waste feed
6.1 METHOD SELECTION FOR SAMPLE DIGESTION
The Demonstration Plan (EPA, 1990) recommends that EPA Method 3050, from
39
-------
SW-846 (EPA, 1986), be used to digest all of the solid samples from the HRD SITE
Demonstration. EPA Method 3050 uses nitric and hydrochloric acids to digest the sample. The
Demonstration Plan also presented two alternative digestion methods (American Society for
Testing and Materials [ASTM] Method E886-A, Lithium Tetraborate Fusion and ASTM Method
E886-B, Aqua Regia Dissolution) that could be used if EPA Method 3050 could not completely
digest the samples.
Because of the potential analytical problems associated with digesting the waste feed,
oxide product, and effluent slag, analyses were performed using various digestion methods. These
analyses compared digestion methods using lithium tetraborate (ASTM Method E-886A),
hydrofluoric acid (HF) (see to Appendix G), EPA Method 3050, and the HRD digestion method.
Although analytical digestion by the HRD method was also considered, this method is not
yet validated by EPA. The HRD digestion method uses a two-stage microwave bomb digestion
procedure. The first stage uses HF and perchloric acid, and the second stage uses nitric and
hydrochloric acids.
Table 3 presents the analytical results obtained from analysis of split samples of waste
feed, effluent slag, and oxide product after digestion by EPA Method 3050, HF, lithium
tetraborate, and the HRD method. The choice of method is not obvious. Each digestion method
has advantages and disadvantages when all of the metals are considered. For the HRD SITE
Demonstration, the metals of primary concern were lead and zinc; therefore, selection of the
digestion method that was used was based mainly on these two metals.
For the effluent slag, the lead and zinc concentrations were about the same for all four
digestion methods. However, for the waste feed, the lead concentration obtained from both the
HF and lithium tetraborate method was one-third that obtained from EPA Method 3050 and the
HRD method. In addition, reported lead concentrations in the oxide product from using the HF
Method were one order of magnitude below the other three digestion methods.
Based on the available data, a slightly modified EPA Method 3050 (using a sample size of
0.2 rather than 1.0 gram) was selected as the best method to digest all of the samples collected
during the HRD SITE Demonstration. Appendix E contains a copy of the August 1, 1991,
"Response to: Technical Systems Review (TSR) of Versar Laboratories, Inc., Springfield,
Virginia," which contains the supporting data used to make the selection. However, the results of
40
-------
TABLE 3
COMPARISON OF METAL DIGESTION PROCEDURES USING
EPA METHOD 3050, THE HYDROFLUORIC ACID METHOD,
THE LITHIUM TETRABORATE FUSION METHOD, AND THE HRD METHOD
(mg/kg)
WMte:F««d"
: hf ......
Lithium
: Tetraborate
HRD
¦ Metal
3050
Arsenic
425
546
<23
620
Cadmium
389
452
<5
340
Chromium
40
251
536
240
Lead
59,000
16,600
17,600
71,100
Selenium
108
55
<69
<10
Thallium
<9.8
3
<95
60
Zinc
3,380
5,010
3,220
5,600
. Effluent Slag
Metal
Method
3050
HF
Lithium
Tetraborate
HRD
Arsenic
<12
757
251
290
Cadmium
8
7.5
<5
1,600
Chromium
99
686
1,760
1,880
Lead
5,740
7,740
6,460
6,590
Selenium
<7.1
43
<70
<10
Thallium
104
2
<96
130
Zinc
1,250
1,590
1,630
2,870
Oxide Product
Metal
Method
3050
HF
Lithium
Tetraborate
' HRD
Arsenic
1,110
1,300
134
1,460
Cadmium
1,380
1,710
<5
1,640
Chromium
288
372
320
350
Lead
128,000
13,200
143,000
197,000
Selenium
74
47
347
<10
Thallium
68
9
113
90
Zinc
20,400
19,700
17,600
22,100
fe
< = less than; mg/kg = milligram per kilogram
41
-------
the method selection project also indicate that the waste feed and effluent slag are
nonhomogeneous and difficult to digest. Additionally, the analytical results of using EPA Method
3050 are known to be poor for chromium and silicon. Consequently, results from the HRD
method, performed on samples by HRD, as well as EPA Method 3050, performed by the SITE
Team, are reported for these two metals.
To determine whether complete digestion of the metals occurred, a sequential digestion
using EPA Method 3050 was conducted on the three matrices. First, the samples were digested
using EPA Method 3050. Then the digestate was filtered, and the liquid portion was then
analyzed to produce the sample concentration. The solid portion was redigested using EPA
Method 3050, filtered, and the new liquid portion was analyzed to determine the concentration, if
any, of metals not digested by the previous extraction. For lead and zinc, the filtrate
concentration was always less than 2.3 percent of the sample concentration. For chromium, the
filtrate concentration was 12.3 percent of the sample concentration. Since chromium was not of
primary concern, these results were considered acceptable. To improve the ability of EPA
Method 3050 to digest all subsequent samples, the sample size was reduced to one-fifth that
specified in the analytical method. Because almost 98 percent of the lead and zinc are extracted in
the first digestion, it was determined that sequential digestion was not required.
6.2 CHARACTERIZATION OF THE WASTE FEED
The waste feed was digested by a slightly modified EPA Method 3050, which used only
20 percent of the normal sample size, and extracted by the Toxicity Characteristic Leaching
Procedure (TCLP). The digestates and extracts were analyzed by appropriate SW-846 methods to
determine metal content. Waste feed samples were also analyzed by various chemical methods to
determine characteristics such as particle size distribution and moisture content.
6.2.1 Metal Content
The metals data for the waste feed are presented in Table 4. Sodium, iron, and lead
account for the majority of the metals as well as the total mass of the waste feed. Seven other
metals, aluminum, calcium, copper, magnesium, potassium, tin, and zinc, as well as silicon, are
present in the waste feed at concentrations of greater than 0.1 percent. A significant percentage
(2.89 to 23.8 percent according to HRD analyses) of the waste feed is silicon.
42
-------
TABLE 4
ANALYSIS OF THE WASTE FEED AFTER DIGESTION BY EPA METHOD 3050
(weight percent)
Analyte
Mean Weight1
Standard
Deviation
Range
Aluminum
0.596
0.0800
0.490-0.787
Antimony
0.0373
0.00503
0.0278-0.0455
Arsenic
0.0515
0.0132
0.0428-0.104
Barium
0.0861
0.00312
0.0804-0.0940
Beryllium
<0.00011
NA
<0.00011
Cadmium
0.0411
0.00345
0.0356-0.0512
Calcium
0.653
0.0702
0.552-0.835
Chromium2
0.00877
0.00148
0.00631-0.0113
Copper
0.185
0.0239
0.146-0.259
Iron
10.3
0.753
9.56-13.0
Lead
5.41
0.414
4.82-6.17
Magnesium
0.228
0.0559
0.163-0.346
Manganese
0.0753
0.00706
0.0672-0.0903
Mercury
0.000068
0.000010
0.000054-
0.000087
Potassium
0.244
0.0255
0.204-0.284
Selenium
0.00727
0.00290
0.00400-0.0175
Silicon2
0.276
0.0716
0.176-0.444
Silver
0.000339
0.000096
0.000160-
0.000540
Sodium
12.2
0.478
11.5-13.2
Thallium
0.0253
0.00424
0.0181-0.0317
Tin
0.282
0.0129
0.261-0.314
Zinc
0.416
0.0744
0.321-0.681
Notes:
1 Average of 18 hourly composites, six each from Runs 2, 3, and 4
2 Due to matrix interferences, analytical results are known to be lower than actual
concentrations for chromium and silicon. The analytical data from the HRD method
indicate that the values for chromium are as follows: mean - 0.024; standard deviation -
0.006; and range - 0.016 to 0.034. The values for silicon are as follows: mean - 8.10;
standard deviation - 5.32; and range - 2.89 to 23.8.
NA = Not applicable; < = less than
43
-------
6.2.2 TCLP Results
The results of the TCLP extraction and analysis, presented in Table 5, show that the waste
feed is a hazardous waste, as defined in the RCRA Toxicity Characteristic (TC) Rule, because
both lead and cadmium are leached at levels above the toxicity characteristic criteria [40 CFR
261.24], The data also show that cadmium was present in a more leachable form than lead in the
waste feed because, although the total cadmium concentration is one hundredth that of lead, the
cadmium leached at 12.8 mg/L, compared to 5.75 mg/L for lead.
6.2.3 General Chemical Analyses for Waste Feed Characterization
The waste feed was also characterized by using a series of general chemical analyses. The
analyses used to characterize the waste feed material included PSD, moisture content, total carbon,
Btu content, ash content, chloride concentration, fluoride concentration, and sulfur concentration.
Table 6 summarizes the PSD results, and Table 7 summarizes the remaining analyses.
HRD recommends that 80 percent of the feed be less than 200 mesh for optimal recovery
of volatile metals, because a small particle is required for efficient heat and mass transfer in the
reaction zone of the Flame Reactor.
As shown in Table 6, 66.6 percent of the waste feed used for this demonstration was less
than 200 mesh, which approached HRD's recommended sizing. The complete distribution is
shown in Table 6. The standard deviation shows that the waste feed PSD for all of the runs was
relatively consistent.
As received from the NSR site, the waste feed had a moisture content of up to 30 percent.
Before being treated during the demonstration, the waste feed was dried to an average total
moisture content of 3.3 percent. The recommended moisture content is less than 5 percent total
moisture (free and chemically bound), because elevated moisture content increases fuel
consumption and adversely affects pneumatic conveying of the waste feed into the Flame Reactor.
Both total carbon and total organic carbon (TOC) analyses were performed. The average
total carbon content of the waste feed was 15.0 percent. The average TOC content of the waste
feed was 14.9 percent. The carbon in the waste feed was believed to contain unreacted coal
particles from the secondary lead smelting process. The waste feed carbon content, which is
44
-------
TABLE 5
TCLP RESULTS FOR THE WASTE FEED
(mg/L)
Extract Concentration
Analyte
Mean1
Standard
Deviation
Range
RCRA
TC Criteria2
Arsenic
0.213
0.0124
<0.210-0.264
5.0
Barium
0.0347
0.0143
0.0177-0.0675
100.0
Cadmium
12.8
2.63
7.61-15.8
1.0
Chromium
0.184
0.0334
0.140-0.283
5.0
Lead
5.75
0.705
4.35-6.80
5.0
Mercury
<0.010
NA
<0.010
0.2
Selenium
0.0716
0.0480
<0.030-0.160
1.0
Silver
<0.050
NA
<0.050
5.0
Notes:
Average of 18 values; when the analyte was not detected, the detection limit is
reported and was used in the calculations.
2 From 40 CFR 261.24
NA = Not applicable
< = less than
mg/L = milligrams per liter
45
-------
TABLE 6
PARTICLE SIZE DISTRIBUTION1 OF THE WASTE FEED
(percent by weight)
Mesh Size
Sieve
Opening
(microns)
Mean2
Standard
Deviation
Range
30 < x
600
7.92
1.01
5.74-9.69
60 < x < 30
250
8.74
0.693
7.34-10.1
100 < x < 60
150
7.53
0.694
6.40-9.25
200 < x < 100
75
9.16
0.897
7.25-11.0
325 < x < 200
45
2.29
0.308
1.87-3.04
x < 325
<45
64.3
1.54
61.6-67.8
Notes:
1 Particle size distribution was determined by ASTM Method D422.
2 Average of 18 weight percent values, six each from Runs 2, 3, and 4
< = less than
46
-------
TABLE 7
SUMMAHY OF OTHER CHEMICAL ANALYSES
Analyte
Method1
Mean
Units
Standard
Deviation
Range
Moisture
ASTM D2216
3.35
percent
0.538
2.26-4.07
Total Carbon
Versar SOP
15.0
percent
2.26
9.56-19.4
Organic Carbon
Versar SOP
14.9
percent
2.01
11.3-19.6
Btu Content
ASTM D2015
1,670
Btu/pound
179
1,240-1,880
Ash Content
ASTM D3174
81.6
percent
0.549
80.6-82.4
Chloride
SW-846 9252
24,600
mg/kg
2,650
21,200-
28,900
Chlorine
as Chloride
ASTM D3177
and MCAWW
300.0
16,600
mg/kg
2,080
13,200-
20,800
Fluorine
as Fluoride
ASTM D3761
130
mg/kg
15.1
106-166
Sulfur
as Sulfate
ASTM D4239
and MCAWW
300.0
15.7
percent
1.10
14.3-19.3
Sulfur
Calculated from
Sulfate
5.25
percent
0.367
4.77-6.44
Notes:
1 ASTM = American Society for Testing and Materials, ASTM, 1991
MCAWW = Methods for the Chemical Analysis of Water and Wastes, EPA, 1979
SW-846 = Test Methods for Evaluating Solid Waste, EPA, 1986
The Versar SOP for total carbon and total organic carbon (TOC) is contained in
Appendix F.
mg/kg = milligram per kilogram
Btu = British thermal unit
47
-------
consumed in the Flame Reactor, is not a feed characteristic that affects the recovery of volatile
metals.
The average Btu content of the waste feed material was 1,660 Btu per pound (Btu/lb).
Residues from high temperature processes such as secondary lead smelting are typically expected
to have low heating values. However, almost all of this can be attributed to the 15.0 percent
carbon, as the carbon content of the waste feed was believed to contain unreacted coal particles.
If all of the Btu content of the waste feed was from the carbon, the heating value of the carbon
would be 11,100 Btu/lb. (If the waste feed has a Btu content of 1,660 Btu/lb, which is
attributable to 15 percent of its composition then that 15 percent composition [carbon] must have a
heating value of 1,660/0.15 or 11,100 Btu/lb.) A poor grade of bituminous coal has a heating
value of between 10,500 and 11,500 Btu/lb (Perry, 1985). Therefore, it is reasonable to assign all
of the heating value of the waste feed to the 15.0 percent carbon content.
The waste feed averaged 81.6 percent inorganic residue or ash. This was not surprising
considering the fact that the waste feed is residue from a high temperature process (secondary lead
smelting). The remaining 18.4 percent is volatile material composed almost entirely of carbon
(15.0 percent) and free moisture (3.3 percent).
The waste feed was analyzed for three chemical constituents: chloride, fluorine, and
sulfur. The waste feed averaged 2.46 percent chloride by weight (24,600 mg/kg). Fluorine was
present at an average concentration of 0.0130 percent by weight (130 mg/kg). The sulfur content
of the waste feed averaged 5.25 percent by weight.
6.2.4 Summary of the Waste Feed Composition
The composition of the waste feed discussed in the previous sections is summarized as
48
-------
follows:
Constituent
Percent
Carbon
Sodium
Iron
Silicon
Lead
Sulfur
Moisture (H20)
Chloride
Miscellaneous metals
TOTAL
15.0
12.2
10.3
8.10 (by HRD analysis)
5.40
5.25
3.35
2.46
3.00 (includes HRD analysis for chromium)
65.0 percent
Of the above constituents, only moisture and carbon are present in the form presented
above. The other constituents are present as various chemical compounds (such as oxides and
sulfates). HRD provided laboratory analyses by X-ray diffraction analysis and scanning electron
microscopy. These techniques can produce semiquantitative mineralogical data. The complete
summary of analytical results are presented in Appendix C. Compounds with significant
concentrations are presented below;
Compound
Thenardite
Hydrous Iron Oxides
Halite
Caracolite
Concentration
>10 percent
>10 percent
3-10 percent
3-10 percent
Chemical formula
Na2S04
FexO„
NaCl
Na3Pb2(S04)3Cl
The analytical results presented above indicate that of the remaining 35 percent of the
waste feed, a significant portion was oxygen bound as oxides and sulfates. Therefore, a
methodology was developed to estimate the oxygen content of the waste feed. The estimated
weight percent of oxygen in the waste feed was calculated using the following assumptions. An
example calculation is shown for the first assumption.
It was assumed all the lead in the waste feed (5.41 percent) was present as
caracolite (Na3Pb2(S04)3Cl), which contains 2.50 percent 02.
49
-------
The calculation for oxygen content is as follows:
The waste feed is 5.40 percent lead. Assume a 100 gram basis.
Therefore, there are 5.40 grams of lead in the waste feed. The
molecular weight of lead is 207.19 grams/mole. The 5.40 grams of
lead equals 0.0261 moles (5.40 grams/207.19 grams/mole =
0.0261 moles).
All lead is assumed to be present as caracolite Na3Pb2(S04)sCl. For
every 2 moles of lead, there is 1 mole of caracolite. Therefore,
there are 0.0130 moles of caracolite (0.0261/2).
Each mole of caracolite contains 12 moles of oxygen (six moles of
02). Therefore, there are 0.156 moles of oxygen in the waste feed
from caracolite.
The 0.156 moles of oxygen (molecular weight 15.99 g/mole) equals
2.50 grams. Because we are working on a 100 gram basis, the waste
feed is 2.50 percent oxygen from caracolite.
Using the same procedure, the remaining number of moles of
sodium (Na), sulfur (S), and chlorine (CI) bound in other
compounds, such as NaCI, Na2S04,can be calculated.
• It was then assumed that all the remaining sulfur (the sulfur not bound in
the caracolite) was present as thenardite (Na2S04), which contains
8.0 percent 02.
• All the remaining chloride (the chloride not bound in the caracolite) was
assumed to be present as halite (NaCI), which contains no 02.
• It was assumed that all excess sodium (not bound in the caracolite,
thenardite, or halite; about 4.26 weight percent) was present as sodium
oxide (Na20), which contains 1.5 percent 02.
• Further it was assumed that all the iron was present as ferric oxide (Fe2Os),
which contains 4.4 percent 02.
• Finally, all the silicon was assumed to be present as silicon dioxide (Si02),
which contains 9.2 percent 02.
(1)
(2)
(3)
(4)
(5)
Using this methodology, the oxygen content of the waste feed is estimated at 25.6 percent.
It should be noted that this is not exact. It is known that small quantities of lead are present as
metallic lead and as sulfates, and that iron is present in several oxidation states (FeO, Fe2Os,and
others). However, using these assumptions, more than 90 percent of the waste feed has been
characterized; 65 percent from constituent analysis and 25.6 percent from estimated oxygen
content.
50
-------
6.3
CHARACTERIZATION OF THE METAL OXIDE PRODUCT
Table 8 presents the metal analysis data for the oxide product. A comparison of the oxide
product data with the waste feed and effluent slag data clearly show that (1) the volatile metals
(such as antimony, arsenic, cadmium, lead, potassium, sodium, and zinc) were concentrated in the
oxide product, and (2) the nonvolatile metals (such as aluminum, calcium, and iron) were
concentrated in the effluent slag. The major metal constituents of the oxide product were lead
(17.4 percent), sodium (15.7 percent), iron (3.22 percent), and zinc (1.38 percent). Seven other
metals (antimony, arsenic, cadmium, calcium, copper, potassium, and tin) plus silicon were
present at concentrations of greater than 0.1 percent. According to HRD data, silicon was present
in the oxide at 10.5 percent. The oxide product contained a small concentration of nonvolatile
species because the horizontal slag separator is not 100 percent efficient. Some slag, including
iron particles, was entrained with the off-gas stream.
Recovery of the volatile metals is a measure of the efficiency of the Flame Reactor
process. Typically, the recovery of the volatile metals is the total mass of volatile metals in the
oxide product compared with the total mass in the waste feed. EPA and HRD have no fully
satisfactory explanation for the relatively low recoveries of arsenic, barium, or thallium in the
oxide product. These metals do not behave as one would expect based solely on their boiling
points. Arsenic is very volatile. Barium and thallium are less volatile, having boiling points
similar to that of lead. The presence of metal species such as oxides or chlorides may explain the
anomalous recoveries. The recoveries of these volatile metals warrant further study, which is
outside the scope of this report.
One objective of this SITE Demonstration was to form a recyclable metal oxide product
enriched in lead and zinc. The Flame Reactor was successfully optimized by HRD to do this.
Arsenic and barium - and possibly thallium - have negative economic impact on oxide product
recycling and were not desired in this product.
The HRD Flame Reactor oxide product was enriched in the volatile metals lead, zinc, and
cadmium, demonstrating greater than 75 percent recovery for all three. Section 6.5.3 discusses the
recovery of all metals in the oxide product. Recycling opportunities are currently being discussed
with various consumers.
51
-------
TABLE 8
ANALYTICAL DATA FOR OXIDE PRODUCT DIGESTION BY EPA METHOD 3050
(weight percent)
Range
Analyte
Mean Weight1
Standard
Deviation
Aluminum
0.0562
0.00734
0.0459-0.0623
Antimony
0.125
0.00403
0.122-0.131
Arsenic
0.110
0.00680
0.101-0.117
Barium
0.0282
0.00310
0.0248-0.0323
Beryllium
<0.0001
NA
<0.0001
Cadmium
0.128
0.0139
0.108-0.138
Calcium
0.202
0.0343
0.155-0.234
Chromium2
0.0300
0.00154
0.0278-0.0313
Copper
0.161
0.0168
0.138-0.178
Iron
3.22
0.267
2.91-3.56
Lead
17.4
1.10
15.9-18.4
Magnesium
0.0327
0.00441
0.0266-0.0368
Manganese
0.0265
0.00370
0.0214-0.0300
Mercury
0.000013
0.000002
<0.000010-
0.00014
Potassium
0.707
0.0548
0.630-0.751
Selenium
0.00520
0.00102
0.00415-0.00659
Silicon2
0.127
0.0102
0.113-0.137
Silver
0.00269
0.000622
0.00190-0.00342
Sodium
15.7
1.40
13.7-16.8
Thallium
0.00746
0.000243
0.00714-0.00773
Tin
0.660
0.0342
0.612-0.687
Zinc
1.38
0.272
1.00-1.62
Notes:
1 Average of three composite samples from Runs 2, 3, and 4; when an analyte was present
below the detection limit, the detection limit was used to calculate the average.
2 Due to matrix interferences, analytical results are known to be lower than actual
concentration for chromium and silicon. The analytical data from the HRD method
indicate that the values for chromium are as follows: mean - 0.034; standard deviation -
0.002; and range - 0.032 to 0.037. The values for silicon are as follows: mean - 10.5;
standard deviation - 5.8; and range - 3.75 to 18.0.
NA = Not applicable; < = less than
52
-------
6.4
CHARACTERIZATION OF THE EFFLUENT SLAG
Table 9 presents the mean, range, and standard deviation of the TCLP extraction and
analysis results on the effluent slag, as well as the RCRA TC criteria. These results show that the
effluent slag was nonhazardous for all eight characteristic metals. In fact, when compared to the
waste feed, the leachable levels of cadmium, chromium, and lead were all reduced to
nondetectable values, and the level of leachable selenium was reduced by half. The levels of
leachable arsenic and barium both increased, although they were still below the TC criteria.
Barium, a nonvolatile metal, was concentrated in the effluent slag by the technology, so increased
leachability may be reasonable. However, the concentration of arsenic in the effluent slag was
less than that in the waste feed, so a decrease in the TCLP value of arsenic is expected. Because
the TCLP value of arsenic was higher in the effluent slag than in the feed, the technology may
convert arsenic into a more leachable state.
The effluent slag was also digested by EPA Method 3050 and analyzed for metals.
Table 10 presents these results, which show that the effluent slag was composed of more than
50 percent metals. The major components were iron, sodium, silicon, aluminum, and calcium,
which accounted for more than 48 percent of the effluent slag mass and more than 95 percent of
the metals. Seven metals (barium, copper, lead, magnesium, manganese, potassium, and zinc)
were present in concentrations of greater than 0.1 percent. Based upon HRD results, the effluent
slag contained 10.2 percent silicon.
Based upon the TCLP extraction procedure, the effluent slag from the HRD Flame
Reactor did not leach any metals above the TC criteria; therefore, the effluent slag was considered
to be nonhazardous.
6.5 MASS BALANCE
A mass balance was performed on the HRD Flame Reactor process using materials
inventory data (total waste feed, oxide product, and effluent slag) and the metals concentration
data for all three streams. Mass balance is an accounting of where chemicals in the waste feed are
partitioned in the products after processing. Mass balance closure is a determination of the
amount of each chemical present in the waste feed that can be accounted for in the products.
Stack emissions are not included because they are small in relation to the other streams. For
example, lead, the largest stack emission, totaled 0.2 pounds compared to approximately
-------
TABLE 9
TCLP RESULTS FOR THE EFFLUENT SLAG
(mg/L)
Extract Concentration
Mean1
Standard
Deviation
RCR A
TC criteria2
Analyte
Range
Arsenic
0.474
0.188
<0.210-0.930
5.0
Barium
0.175
0.0420
0.109-0.281
100.0
Cadmium
<0.050
NA
<0.050
1.0
Chromium
<0.060
NA
<0.060
5.0
Lead
<0.330
NA
<0.330
5.0
Mercury
<0.010
NA
<0.010
0.2
Selenium
0.0326
0.010
<0.030-0.730
1.0
Silver
<0.050
NA
<0.050
5.0
Notes:
1 Average of 18 values; for analytes that were not detected, the detection limit is
reported and used in calculations.
2 From 40 CFR 261.24
NA = Not applicable
mg/L = milligram per liter
54
-------
TABLE 10
ANALYTICAL DATA FOR EFFLUENT SLAG DIGESTION BY EPA METHOD 3050
(weight percent)
Mean Weight1
Standard
Range
Aluminum
1.53
0.138
1.33-1.85
Antimony
0.0357
0.0412
0.0100-0.190
Arsenic
0.0262
0.0291
0.00921-0.134
Barium
0.165
0.0136
0.139-0.183
Beryllium
0.000101
0.000008
<0.00087-
0.000110
Cadmium
0.000373
0.000254
<0.00023-0.00135
Calcium
1.30
0.0973
1.06-1.45
Chromium2
0.00890
0.00786
0.00339-0.0385
Copper
0.344
0.0324
0.273-0.389
Iron
20.4
1.60
16.7-22.8
Lead
0.552
0.252
0.156-1.14
Magnesium
0.543
0.0847
0.441-0.761
Manganese
0.175
0.0268
0.132-0.231
Mercury
<0.000010
NA
<0.000010
Potassium
0.238
0.0194
0.199-0.269
Selenium
0.00344
0.00345
<0.00226-0.0176
Silicon2
0.327
0.0979
0.183-0.525 1
Silver
0.000394
0.000082
0.000250-
0.000510
Sodium
15.5
1.06
12.8-16.8
Thallium
0.0689
0.00862
0.0535-0.0852
Tin
0.0796
0.0150
0.0544-0.111
Zinc
0.113
0.0287
0.07110-0.168
Notes:
1 Average of 18 hourly composites, six each from Runs 2, 3, and 4; when an analyte was
present below the detection limit, the detection limit was used to calculate the average.
2 Due to matrix interferences, analytical results are known to be lower than actual
concentrations for chromium and silicon. The analytical data from the HRD method
indicate that the values for chromium are as follows: mean - 0.040; standard deviation -
0.023; and range - 0.013 to 0.11. The values for silicon are as follows: mean - 10.2;
standard deviation - 4.7; and range - 7.77 to 28.4.
NA = Not applicable; < = less than
55
-------
2,000 pounds of lead in the oxide product for the entire demonstration.
6.5.1 Materials Inventory Data
During the demonstration, the masses of the waste feed, oxide product, and effluent slag
were determined by HRD. The materials inventory data were obtained from calibrated scales.
The amount of waste feed used was calculated from the differences in the load cell readings of the
day bin feed hoppers. The masses of oxide product and effluent slag were determined by
weighing the full bulk storage bags of oxide product and full metal bins of effluent slag. Because
the bags and bins were not always full at the end of a run, the materials inventory data are
meaningful when data from all four demonstration runs are summed. Table 11 presents the
materials inventory data.
Because the metals concentration from Run 1 samples are unknown (samples were not
analyzed because of sampling problems), there is an element of uncertainty in applying the metals
concentration data to the combined materials inventory data. Because the metal concentrations in
samples from Runs 2, 3, and 4 were consistent, we assume the partitioning in Run 1 was similar,
therefore the risk of error in the overall materials inventory data is believed to be small.
In addition, the materials inventory includes a line item referred to as cleanout. This
material results from cleaning the oxide product recovery system and material deposited in the
shell and tube heat exchanger and combustion chamber. The cleanout material contained oxide
product mixed with slag that was entrained with the off-gas. Therefore, this material was a
mixture of oxide product and effluent slag.
6.5.2 Weight Reduction
For all four test runs, a total of 47,300 pounds of waste feed were processed, generating
11,200 pounds of oxide product and 15,300 pounds of effluent slag. The total mass of oxide
product and effluent slag only account for 56.1 percent of the waste feed mass. Therefore, the
process has a net weight reduction of 43.9 percent. After the demonstration was complete, a total
of 3,460 pounds of treated waste was cleaned out of the combustion chamber and heat exchanger.
This is 7.32 percent of the total waste feed processed. Therefore, if the cleanout material is
included, the weight reduction is 36.6 percent. This information is summarized in Table 11.
56
-------
TABLE 11
MATERIALS INVENTORY DATA FROM THE HRD SITE DEMONSTRATION
(pounds)
Run I1
Run 2
Run 3
Run 4
Total2
Waste feed
14,700
11,100
10,300
11,100
47,300
Oxide Product
3,360
2,400
2,250
3,210
11,200
Effluent Slag
4,340
5,170
1,640
4,160
15,300
Cleanout
3
1,640
3
1,810
3,460
Notes:
Although concentration data were not collected for this run because of fluctuations in
stack gas temperature, the materials inventory data are reported.
Due to rounding errors, the total weight does not equal the sum of the weights of the
four runs.
No combustion chamber and heat exchanger cleanout material was weighed on these days.
57
-------
The weight reduction was primarily the result of essentially complete conversion of carbon
to carbon dioxide (CO2)(15.0 percent), moisture to steam (3.35 percent), and chloride to hydrogen
chloride (HC1) gas (2.46 percent). In addition, sulfur was partially converted to sulfur dioxide
(S02)(2.26 percent). The remaining sulfur (2.99 percent) was trapped in either the oxide product
or the effluent slag. These values are the average of data from Runs 2, 3, and 4.
The remaining 13.5 percent weight reduction is attributed to the strong reducing
conditions of the Flame Reactor and analytical variations.
6.5.3 Percent Recovery
Because no metal concentration data are available for Run 1, only data from Runs
2, 3, and 4 were used to calculate percent recoveries. The materials inventory data for these runs
include: 32,500 pounds of waste feed, 7,860 pounds of oxide product, and 11,000 pounds of
effluent slag. Table 12 presents metal recovery data in two forms. The first form is the raw
percent recovery when the oxide product is compared to the feed. Because mass balance closure is
less than 100 percent, due mainly to build up of material in the combustion chamber and heat
exchanger, the percent recoveries are low. For lead, zinc, and cadmium, the percent recoveries
are 77.7, 80.0, and 75.0, respectively. The second method presents a normalized percent recovery.
This method scales the percent recovery data based on the mass balance closure, so that the sum of
the oxide product percent recovery and effluent slag percent recovery is 100 percent. Using this
method, the percent recoveries of lead, zinc, and cadmium are 95.8, 89.7, and 99.6, respectively.
6.6 HRD FLAME REACTOR PROCESS RELIABILITY
The HRD Flame Reactor process consists of five sections: (1) the feed system, (2) the
Flame Reactor, (3) the slag separator, (4) the combustion chamber, and (5) the oxide product
recovery system. The Flame Reactor, slag separator, and combustion chamber had no operational
problems that affected their performance during the demonstration. The feed system had one
feed system jam that did not affect the SITE Demonstration results. However, HRD reports that
the oxide product recovery system was undersized for the feed rate used during the
demonstration.
During Run 1, a feed screw conveyor was jammed for approximately 30 minutes. The
other feed conveyor rate was increased to keep the feed rate to the reactor constant. Because
58
-------
TABLE 12
MASS BALANCE CLOSURE AND RECOVERY
OF METALS PRESENT IN THE WASTE FEED
(percent)
Analyte
Mass Balance
Closure
Oxide Product
Recovery1
Normalized
Oxide Product
Recovery2
Aluminum
88.9
2.27
2.56
Antimony8
113
81.1
71.5
Arsenic3
68.7
51.6
75.1
Barium3
72.3
7.90
10.9
Beryllium8
58.6
23.9
40.7
Cadmium3
75.3
75.0
99.6
Calcium
74.5
7.48
10.0
Chromium3,4
90.6
34.6
38.2
Copper
83.5
20.9
25.1
Iron
74.6
7.53
10.1
Lead
81.2
77.7
95.8
Magnesium
83.5
3.46
4.14
Manganese3
87.0
8.50
9.77
Mercury8
9.44
4.49
47.6
Potassium
103
69.8
68.0
Selenium3
33.2
17.3
52.0
Silicon4
73.6
31.2
42.3
Silver3
230
191
83.0
Sodium
73.5
30.9
42.0
Thallium3
99.0
7.11
7.19
Tin
66.1
56.6
85.6
Zinc
89.2
80.0
89.7
Notes:
In this calculation the weight percent of the metal in the effluent slag is the mass
balance closure minus the oxide product recovery.
In this calculation the weight percent of the metal in the effluent slag is 100 minus
the normalized oxide product percent recovery.
Analyte is present in the waste feed at less than 0.1 weight percent.
Silicon and chromium analytical results from HRD's analytical procedure were
used.
59
-------
Run 1 was discarded (because of temperature and pressure fluctuations in the stack gas), the
demonstration results were not affected.
The oxide product recovery system consists of a shell and tube heat exchanger, a baghouse,
an induced draft fan, and a stack. This equipment was sized to handle the gases generated by the
HRD Flame Reactor processing 20,000 tons per year of EAF dust. However, for reasons
explained below, when processing waste feed during the SITE Demonstration at a rate of only
7,880 tons per year (0.9 tons/hour or 30 pounds per minute), the capacity of the oxide product
recovery system was exceeded.
During each demonstration test run, the Flame Reactor was forced to shut down because
of back pressure at the slag tap hole. Under normal conditions, the reactor operates under a
slightly negative pressure to prevent escape of gases, including CO and metal vapors. The
negative pressure is maintained by the induced draft baghouse blower. If this blower is unable to
draw a sufficient draft, the process reactions quickly cause a pressure rise which results in the
escape of gases.
During the demonstration test runs, the heat exchanger restricted the flow of off-gases
causing back pressure at the slag tap hole and subsequent shutdown of the reactor. The
water-cooled shell and tube heat exchanger was obtained as salvage from a nearby zinc smelter.
Approximately 20 percent of its tubes have been intentionally sealed because they are broken due
to corrosion and leak water. There were no problems at the beginning of each test run because the
system was cool and the refractory material lining the system before the heat exchanger behaved
as a heat sink, lowering the temperature and, therefore, the gas volume through the heat
exchanger. As the refractory material heated, the temperature of the gases entering the heat
exchanger increased until the gas volume was so high that a large pressure drop occurred across
the heat exchanger and the induced draft baghouse blower was unable to compensate.
The tap hole back pressure usually occurred during the demonstration after about 4 hours
of steady operation. At this time, the control room operator manually shut down the Flame
Reactor. The system was cooled by drawing ambient air through the oxide product collection
system. At the same time, oxide which accumulated in the heat exchanger tubing was removed to
reduce restrictions. The total time of each shut down was 2 to 3 hours.
For a full-scale Superfund site remediation project, the gas handling capacity of the oxide
60
-------
product recovery system could be increased to allow the process to run 24 hours/day with regular
maintenance.
Another problem developed when, on the second day of the demonstration, the cooling
tower began leaking noncontact cooling water. Although the addition of makeup water was
required to maintain the necessary level in the tower, the problem did not affect the operation of
the cooling system.
6.7 STACK MONITORING AND EMISSIONS SAMPLING RESULTS
During the HRD SITE Demonstration, stack gases were sampled for metals, HC1, and
particulate emissions, and were continuously monitored for SOz, nitrogen oxides (NOx), 02, C02,
CO, and THC. The metals and particulate emissions were determined using an EPA Modified
Method 5, isokinetic, multiple metals sampling train, HC1 emissions were determined by a single
point EPA Method 26 sampling train. The continuous emission monitors used the following: EPA
Method 6C for S02, EPA Method 7E for NOx, EPA Method 3A for 02and C02, EPA Method
10 for CO, and EPA Method 25A for THC. All the standard EPA methods can be found in
40 CFR 60, Appendix A, and the multiple metals train is discussed in the Methods Manual for
Compliance with the Boilers and Industrial Furnaces (BIF) Regulations [40 CFR 266, Appendix
IX].
6.7.1 Introduction to the BIF Rule
On February 21, 1991, the final BIF rule was published [40 CFR 266, Subpart H; 56 FR
7134-7240]. This rule is the applicable or relevant and appropriate requirement (ARAR) that the
air emissions from the HRD Flame Reactor may be required to meet. Therefore, a brief
introduction to the rule is presented before stack monitoring and emissions sampling are discussed.
The BIF rule became effective on August 21, 1991.
The HRD Flame Reactor is classified as a smelting, melting, and refining furnace by the
BIF rule [56 FR 7143]. However, smelting, melting, and refining furnaces that process hazardous
waste solely for metal recovery are conditionally exempt from the majority of 40 CFR 266,
Subpart H regulations, and therefore, need only comply with 40 CFR 266.101 (Management Prior
to Burning) and 40 CFR 266.112 (Regulation of Residues). These furnaces are conditionally
exempt because (1) EPA does not believe it prudent to regulate a whole potential class of devices
61
-------
and wastes that it has not fully evaluated and (2) EPA wishes to study further whether regulating
these furnaces under the Clean Air Act may be more appropriate, particularly if technology-based
controls on toxic air emissions are likely to apply.
Even though EPA might classify the HRD Flame Reactor as conditionally exempt because
the B1F rule promulgated risk-based emission levels for metals and HC1, the state or federal
permitting authority could possibly apply these standards by imposing the omnibus authority of
RCRA [40 CFR 270.32(b)(2) and RCRA Section 3005(c)(3)] to protect human health and the
environment. Therefore, the B1F regulatory limits for metals, particulate, and HCl emissions are
presented below.
The BIF rule established a three-tiered permitting structure to control emissions of HCl,
chlorine (CI2), and 10 toxic metals listed in Appendix VIII of 40 CFR 261. The list of 10 toxic
metals is further broken down into four carcinogenic metals (arsenic, beryllium, cadmium, and
chromium) and six noncarcinogenic metals (antimony, barium, lead, mercury, silver, and
thallium). Tier I of the three-tiered permitting structure limits feed rates, Tier II sets emission
rate screening limits, and Tier III requires a site-specific risk assessment. EPA expects the
majority of facilities to elect to comply with Tier III standards to obtain more flexible permit
limits.
Tier III standards require 1) emissions testing to determine the emission rate for each metal
and 2) air dispersion modeling to predict the maximum, annual, off-site, ground-level
concentration for each metal. These concentrations are then compared to the acceptable ambient
levels specified in Appendix IV and V of the BIF rule [56 FR 7223], Because dispersion modeling
is beyond the scope of the SITE Demonstration, actual Tier III limits for the HRD Flame Reactor
were not calculated; therefore. Tier II screening limits will be presented. These limits are not the
regulatory limits for the Flame Reactor and are presented for comparison purposes only. In
addition, because the HRD stack is shorter than the Flame Reactor building, and because the
surrounding terrain within 5 kilometers (3.1 miles) of the stack equals or exceeds the elevation of
the physical stack height, the Tier II screening limits are the conservative (restrictive) limits.
The HRD Flame Reactor may be subject to additional air emissions regulations when EPA
promulgates regulations under the authority of the Clean Air Act Amendments of 1990. The
Clean Air Act Amendments concerning hazardous air pollutants may potentially address many of
the same sources regulated under 40 CFR 266. For example, the New Source Performance
62
-------
Standards (NSPS) of Section 111 of the Clean Air Act may be ARARs for a HRD Flame Reactor
unit installed at a Superfund site, especially if the pollutants emitted and the technology employed
are sufficiently similar to a pollutant and source category regulated by the NSPS. Also, EPA has
established National Emissions Standards for Hazardous Air Pollutants (NESHAPS) for arsenic,
beryllium, and mercury for certain categories of sources [40 CFR 61].
6.7.2 Metal Emissions
The metals emissions in the stack gas were calculated from the EPA Modified Method
5 sampling train data. The Modified Method 5 sampling train collects two types of samples. The
first type, or "front-half," is solid matter collected on a filter. The second type, or "back-half,"
collects gaseous emissions by absorbing them in an acid solution. The analytical results, shown in
Table 13, are the sum of both halves. Table 13 presents the metals emission rates for the blank
run and the three valid test runs (Runs 2, 3, and 4), as well as the BIF Rule Tier II screening
limits. The Tier II screening limits are presented for comparison only. EPA believes that most
facilities will choose to comply with the less restrictive Tier III limits. Tier III limits are not
presented because they involve site-specific dispersion modeling and risk analysis.
The HRD Flame Reactor emitted lead, chromium, and arsenic at rates above the Tier II
screening limits during the SITE Demonstration; however, the detection limit for arsenic is too
high for comparison to Tier II limits. The Tier II levels presented are restrictive because of the
short stack and the complex terrain. (See Section 6.7.1 of this report.)
6.7.3 HC1 Emissions
HC1 emissions during the HRD Flame Reactor Demonstration were between 38.5 and
46.4 pounds per hour (lb/hr). This high emission rate could be expected because the Flame
Reactor had no acid gas control system, and the waste feed contained, on average, 2.46 percent
chloride by weight. The BIF rule has promulgated risk-based emission limits on HC1. The Tier II
screening limit is 0.091 g/sec (0.72 lb/hr) [56 FR 7232], The addition of a wet scrubber should
control HC1 emissions to below the applicable standards.
6.7.4 Particulate Emissions
Because the HRD Flame Reactor process uses a baghouse to capture the metal oxide
63
-------
TABLE 13
METAL EMISSION RATES OF THE HRD FLAME REACTOR
(grams per hour)
Test
Tier II
Above
Blank
Test
Test
Screening
Limits1
t
Metal
Run
Run 2
Run 3
Run 4
Below
Antimony
0.502
0.39s
0.352-3
0.372,3
14
Below
Arsenic
0.392
0.292
0.272,8
0.292,3
0.11
Above
Barium
0.041
0.55
0.72
0.62
2400
Below
Beryllium
0.00572'3
0.0192'3
0.0182,3
0.0192,3
0.20
Below
Cadmium
0.122
0.242
0.172
0.172
0.26
Below
Chromium
0.0532
1.02
0.702
1.02
0.040
Above
Lead
12
122
122
122
4.3
Above
Mercury
0.0413
1.1
1.3
0.96
14
Below
Silver
0.015
0.19
0.0132'3
0.0142-3
140
Below
Thallium
0.722,3
2.32,8
2.22,3
2.32,8
14
Below
Notes:
1 Emission limits are based on emission of a single metal. Tier III standards based on
site-specific dispersion modeling are less restrictive. (See Section 6.7.1 of this report).
2 The concentration of the metal was below the analytical detection limit in the combined
EPA Modified Method 5 sampling train back-half samples for this run.
3 The concentration of the metal was below the analytical detection limit in the combined
EPA Modified Method 5 sampling train front-half samples for this run.
64
-------
product, particulate emissions from the Flame Reactor should be low when the baghouse is
maintained and operated properly.
During analysis of the demonstration samples, problems occurred with the gravimetric
analysis, preventing accurate determination of the particulate emissions for all but the blank run.
Therefore, a worst case analysis (when the concentration of a metal was below the analytical
detection limit, the detection limit was used in the emissions calculation) of the test run
particulate emissions was performed.
Particulate emissions are calculated from the sum of two gravimetric analyses. The first
analysis is conducted on the filter. The EPA Method 5 sample train uses a filter that exhibits
99.95 percent efficiency on 0.3-micron dioctyl phthalate smoke. The filter is weighed before and
after sampling. The resulting difference is equal to the particulate weight on the filter. The
blank run and Run 3 were the only runs with positive differences. Negative numbers were
replaced with 1 mg for a worst case analysis. The second gravimetric analysis is the acetone probe
wash. The probe (that part of the EPA Method 5 train that carries the gas sample from the source
to the filter) is rinsed with acetone to remove any particulate matter deposited on the probe. In
the laboratory, the acetone is completely evaporated, and the residual is weighed. A laboratory
error occurred, and thus residuals were weighed only to ± 10 mg. The blank run and Run 2 were
the only runs without negative numbers. For a worst case analysis, each run was assumed to show
10 mg of dried acetone residual.
Even under a worst case scenario, the test run particulate emissions were lower than the
particulate emissions for the blank run. The blank run particulate emissions were slightly higher
than runs containing waste, because no oxide product was formed to act as seed nuclei for particle
formation and growth. Therefore, the particles formed were smaller and were not captured by the
baghouse. The sources of the blank run particulate emissions may include residue material in the
system and lime used to condition the baghouse bags.
The permit limit specified in the EPA RD&D permit for particulate emission is 0.02 grains
per dscf. As shown in Table 14, all particulate emissions were below this standard. Table 14 also
presents the particulate emissions in lb/hr and in grains/dscf corrected to 12 percent C02and
7 percent Ozfor comparison purposes.
65
-------
TABLE 14
PARTICULATE RESULTS
Test
Run
Actual
Grains/dscf Corrected to
Pounds
per
hour
mg/dscf
Grains/dscf
12 percent
co2
7 percent
o2
Blank
0.517
0.00797
0.0282
0.0786
0.522
21
0.191
0.00295
0.00887
0.0164
0.364
31
0.294 | 0.00454
0.0114
0.0233
0.548
41
0.213 j 0.00328
0.00894 | 0.0168
0.356
Notes:
1 Particulate emissions were calculated using a worst case scenario,
dscf = dry standard cubic feet
mg = milligrams
66
-------
6.7.5
Continuous Emission Monitoring
Emissions of S02, NOx, 02, C02, and THC were continuously monitored for the blank run
and for each 6-hour test run. Table 15 presents the average emission for each run.
The HRD Flame Reactor currently has an air quality permit issued by PaDER that limits
S02 emissions to less than 500 ppm for commercial operations. For the demonstration, for which
the limit is not effective, the S02 emissions were below 500 ppm except for a 2-minute period
during Run 2. The maximum SOz emission was 514 ppm, which occurred immediately following
startup, after a system shut down was required to cool the oxide product recovery system.
6.8 HRD FLAME REACTOR OPERATING PARAMETERS
Table 16 shows the operating parameters for the Flame Reactor during the blank run and
the three valid test runs. The three test runs were intended to be replicate runs. The operating
parameters show that, within operating limits, the operating conditions were the same for all three
runs.
6.9 THE HRD THERMODYNAMIC MODEL
HRD has developed a thermodynamic model for the HRD Flame Reactor process. The
model was used to set Flame Reactor operating parameters for the shakedown runs. Because the
SLS slag waste feed is different in chemical composition from the EAF dust used to develop the
model, the results of the model were not extremely accurate. Without improvements, it is believed
that the model can be used only to set preliminary operating conditions and to determine order of
magnitude estimates of the required amount of fuel and oxygen.
In the HRD Thermodynamic Model, oxygen and natural gas flow rates are calculated, and
oxide product recovery and effluent slag formation are estimated by entering waste composition
data and expected operating conditions into the model. The waste composition data used were
based on chemical analyses and the estimated mineralogy of the waste feed. The operating
parameters used were the waste feed rate, fuel composition (natural gas assumed to be 100 percent
methane), oxidant composition (the combined total gas input, about 80 percent oxygen and
20 percent nitrogen), estimated heat loss, target temperature, and carbon monoxide/carbon
dioxide ratio. Using these data, the model calculated the required waste feed, oxygen, air, and
67
-------
TABLE 15
RESULTS OF CONTINUOUS EMISSION MONITORING
(average concentrations)
Gas
Units
Blank Run
Test Run 2
Test Run 3
Test Run 4
S02
ppm dry
ND
268
272
290
NOx
ppm dry
173
16.0
15.8
18.5
o2
percent
19.6
18.5
18.3
18.3
co2
percent
3.39
3.99
4.77
4.40
CO
ppm dry
4.17
ND
14.21
1.28
THC
ppm dry, as
propane
1.64
1.61
1.7
0.91
Notes:
The CO analyzer performed erratically during Test Run 3. Therefore, the CO data for
Test Run 3 are suspect.
ND = Not detected
ppm ¦= parts per million
68
-------
TABLE 16
FLAME REACTOR OPERATING PARAMETERS
FOR THE SITE DEMONSTRATION
Operating Parameter
Range of Values
: o1
31 ¦
¦ 41
Natural gas feed rate (scfm)
242-251
334-356
331-353
333-357
Oxygen feed rate (scfm)
404-413
618-651
586-659
629-676
Air feed rate (scfm)
159-162
158-170
157-173
155-161
Waste feed rate (ib/min [lb/test])
0
29.72
[10,700]
29.82
[10,800]
30.02
[10,800]
Reactor slag flow rate (Ib/min [lb/test])
0
14.4
[5,170]
4.55
[1,640]
11.6
[4,160]
Combustion chamber temperature (°C)
308-329
315-733
215-832
423-726
Cooling chamber exit gas temperature (°C)
92-99
100-306
78-401
154-372
Oxide product flowrate (lb/min [lb/test])
0
6.67
[2,400]
6.25
[3,250]
8.92
[3,210]
Stack temperature (°C)
73-70
50-150
50-147
70-155
Notes:
Test Run number. Test Run 0 was the background run. Test Run 1 was not a valid run.
2 Average feed rate for entire test calculated from the change in weight of the feed bins
over time,
scfm = standard cubic feet per minute
lb/min = pounds per minute
lb/test = pounds per test
°C = degrees Celsius
69
-------
natural gas flow rates, and estimated the oxide product recovery and effluent slag formation,
based on the system achieving an equilibrium.
The operating and recovery information estimated from the HRD Thermodynamic Model,
and the actual data from the demonstration test are presented in Table 17. As shown in Table
17, the HRD Thermodynamic Model was unable to predict with the same accuracy as with
previous waste feeds (1) the necessary oxygen and natural gas flows, (2) the amount of oxide
product recovered, (3) the amount of effluent slag produced, and (4) the percent recoveries of
zinc and lead.
The flow rates necessary for oxygen and natural gas were significantly higher than
predicted by the HRD thermodynamic model. The formation of effluent slag in the actual
demonstration was much less than predicted by the model, and the recovery of oxide product was
much greater. Although total oxide product recovery was greater than predicted, the recoveries
obtained during the demonstration for zinc (89.7 percent) and lead (95.8 percent) were less than
predicted by the model (99 percent and 99.6 percent, respectively).
HRD offered several explanations of why the model was not able to accurately predict the
conditions of the demonstration. First, because the model was set up for EAF dust and not for
SLS slag, the waste composition data for the SLS slag did not fit directly into the model.
Therefore, HRD had to make several assumptions concerning the mineralogy of the SLS slag and
adjust the waste composition data to facilitate using this data in the model.
Also, certain constituents (such as sodium) that are assumed by the model to partition
entirely into the effluent slag were found to be present in the oxide product. This would partially
explain the lower-than-predicted effluent slag formation and the higher-than-predicted oxide
product recovery.
70
-------
TABLE 17
OPERATING AND RECOVERY DATA FROM THE HRD
THERMODYNAMIC MODEL AND THE SITE DEMONSTRATION
. ,
Effluent
Slag
Formation
{percent
of feed)
Oxide
Product
Recovery
(percent
, of feed)
Zinc
Recovery
(percent :
of total
line)
Lead
Recovery
(percent
of total
lead)
Oxygen
Flow
(scfm)
Natural
i Gas
Flow
(scfm)
Feed
Rate
(Ib/min)
HRD Model
488
208
30
62.6
8.3
99.0
99.6
SITE
Demonstration
614
310
30
30.1
22.0
89.7*
95.81
Notes:
* Because the HRD model assumes a 100 percent closure on the mass balance, the normalieed percent recovery is
presented. The raw percent recovery for line and lead are 80.0 and 77.7 percent, respectively,
scfm = standard cubic feet per minute
Ib/min = pounds per minute
71
-------
7.0 QUALITY ASSURANCE/QUALITY CONTROL
The overall quality assurance objective was to provide well-documented sampling and
analytical data of known quality in support of the demonstration objectives. A QAPP was
prepared to incorporate all of the elements of a Category II QAPP (EPA, 1987; EPA, 1991b).
Specific QA targets for precision, accuracy, completeness, and detection limit are listed in
Table 18 and are summarized in the sections to follow. These sections present the analytical
methods used and data for holding times, blanks, matrix spikes and spike duplicates, laboratory
duplicates, reference standards, and detection limits.
7.1 ANALYTICAL METHODS
In selecting appropriate methods to prepare and analyze the solid, liquid, and vapor phase
samples from the HRD SITE Demonstration, the specific analytes of interest, the sample matrices,
and the minimum detectable concentrations needed for the project were taken into account.
Tables 19, 20, and 21 summarize the methods chosen. Most of the analytes were analyzed by
using the EPA-approved methods or American Society for Testing and Materials (ASTM) methods
as specified in the QAPP (EPA, 1990). Sulfur was analyzed by ASTM Method D3177 followed by
EPA Methods for the Chemical Analysis of Water and Wastes (MCAWW) Method 300.0, instead of
ASTM Method D4239 as originally specified, because it was determined during the laboratory
audit that the method used would be more appropriate for samples with high metal content
(ASTM, 1991; EPA, 1979).
Metals, TCLP metals, chloride, total organic carbon, Btu content, ash content, moisture
content, and particle size distribution analyses were performed on the waste feed. Metals, TCLP
metals, and moisture content analyses were performed on the effluent slag samples. Metals and
moisture content analyses were performed on the oxide product samples. For the stack gas sample
trains, particulate (metals sampling train), chloride (HCl sampling train), metals, and mercury
(metals/mercury train) were quantified.
The standard EPA-approved digestion technique for metals preparation is
SW-846 Method 3050 (EPA, 1986). However, because of potential analytical problems involving
incomplete digestion, several alternative digestion methods were considered. Three methods,
72
-------
TABLE 18
QA OBJECTIVES FOR PRECISION, ACCURACY,
COMPLETENESS, AND DETECTION LIMITS
(sheet 1 of 2)
Measurement
Matrix Type
Method
Reference
Measurement::
Units
Detection
! Limits
Precision
(RPD)
Accuracy
(Percent
Recovery)
Completeness
(percent)
1. Metals
Metals
(TCLP)
Solid'
Gas2
Solid (Waste
feed and
effluent slag
samples)
SW-8463
Methods
SW-8463
Methods
SW-846
Methods
mg/kg
Mg/dscf
mg/L
Various*
Various4
Various4
25
25
25
NA
73-130
70-130
90
100
90
2. Chloride
Stack gas
samples
SW-846
' 9252
ppm
1
25
70-130
100
3. Particulate
Stack gas
samples
EPA
Method 5
gram/dscf
0.0015
gram/
filter*
NA
NA
90
4. Total
Chloride
Solid
SW-846
9252
percent
0.1
26
NA
90
5. Total
Fluoride
Solid
ASTM D3761
percent
0.1
25
NA
90
6. Total
Sulfur
Solid
ASTM D3177
and MCAWW
300.0
percent
0.1
25
NA
90
7. Total
Carbon
Solid
Versar SOP
percent
0.1
25
NA
90
8. Total
Organic
Carbon
Solid
Versar SOP
percent
0.1
30
NA
90
9. Particle
Distribution
Solid
(Waste feed)
D422
ASTM
NA
NA
NA
NA
90
10. Moisture
Content
Solid
(Waste feed)
D2216
ASTM
percent
NA
30
NA
90
Notes:
1 Metals analysed in the solid samples (effluent slag, oxide product, and waste feed) include aluminum, antimony, arsenic, barium,
beryllium, cadmium, calcium, chromium, copper, iron, lead, magnesium, manganese, mercury, potassium, silicon, silver, sodium,
thallium, tin, and line.
2 Metals analyzed in the stack samples include antimony, arsenic, barium, beryllium, cadmium, chromium, lead, mercury, silver,
and thaJlium.
3 Test Methods for Evaluating Solid Waste, Volume One: Laboratory Manual, Physical/Chemical Methods; and Volume Two
Field Manual, Physical/Chemical Methods, SW-846, Third Edition, Office of Solid Waste, U.S. Environmental Protection
Agency, Document Control No. 955-011-00000-1, 1986.
73
-------
TABLE 18
QA OBJECTIVES FOR PRECISION, ACCURACY,
COMPLETENESS, AND DETECTION LIMITS
(sheet 2 of 2)
4 Typical instrument detection limits for metals analysed by inductively coupled plasma atomic emission spectroscopy
(ICP) using EPA Method 6010 were as follows:
Typical Detection Limits for Metals
Metal
Liquid Matrix
(uk/L)
Solid Matrix
fma/kg)
Impinger Contents
fliit/dscf)
Filter
(tfg/dscf)
Aluminum
25
6.5
—
Antimony
30
8.5
0.26
0.030
Arsenic
21 (4 for GFAA6)
9.0
0.18
0.097
Barium
1
0.5
0.0085
0.0037
Beryllium
1
1
0.0086
0.011
Cadmium
5
2.5
0.043
0.027
Calcium
20
4.5
--
--
Chromium
6
2
0.052
0.054
Copper
E
2.5
—
--
Iron
10
3
—
—
Lead
33 (1 for GFAA6)
17.5
0.28
0.19
Magnesium
500
5
—
—
Manganese
4
1
—
—
Mercury
(0.2 for CV7)
(0.1 for CV7)
0.0086
0.0011
Potassium
744
242
—
—
Selenium
(3 for GFAA6)
26
--
—
Silicon
—
38
—
—
Silver
5 (0.5 for GFAA6)
1
0.0041
0.011
Sodium
29
10
—
--
Thallium
(1 for GFAA6)
75
1.4
0.81
Tin
13
6.5
—
—
Zinc
2
1
--
--
Note that the actual detection limits varied, depending on the concentration of the analytes present and the presence of any
interference. The stack gas sampling trains generate liquid matrices for which the detection limits are listed. The detection
limits for the impinger contents and filter were based on the average sample volume of Runs 2, 3, and 4 (55.5 dscf) and a
final impinger volume of the system blank (475 ml).
5 Estimate of the detection limit was baaed on performing gravimetric analysis of the filter by using a scale accurate to
0.0001 grams.
6 Graphite furnace atomic absorption spectroscopy
7 Cold vapor atomic absorption spectroscopy
NA = Not applicable
RPD = Relative percent difference
fig = microgram
dscf = dry standard cubic foot
mg = milligram
L = liter
74
-------
TABLE 19
ANALYTICAL METHODS FOR SOLID SAMPLES
(Waste Feed, Oxide Product, and Effluent Slag)
Analyte
Preparation
Method1
Analysis
Method1
Metals
Mercury
Other metals: aluminum, antimony, arsenic,
barium, beryllium, cadmium, calcium, chromium,
copper, iron, lead, magnesium, manganese,
potassium, selenium, silicon, silver, sodium,
thallium, tin, and sine
Leachable Metals (use TCLP, Method 1311 to
prepare extract)
Mercury
Selenium
Other metalB: arsenic, barium,
cadmium, chromium, lead, and
silver
Other Analvtes3
Particle size distribution
Moisture content
Chloride
Btu content
Ash content
Total organic carbon
Total carbon
Sulfur as sulfate
Fluorine as fluoride
Chlorine as chloride
7471
3050
7470
7740
3005
7471
6010
7470
7740
6010
ASTM D422
ASTM D2216
9252
ASTM D2015
ASTM D3174
Verear SOP4
Verear SOP4
ASTM D3177, followed by
MCA WW 300.0
ASTM D3761
ASTM D3177, followed by
MCAWW 300.0
Notes:
1
2
3
4
All methods are from SW-846 unless specified otherwise.
Method 3050 was modified to use 0.2 gram of sample instead of 1 gram.
Analysis was completed only for samples from waste feed.
Versar SOP is included in Appendix F of this document.
75
-------
TABLE 20
ANALYTICAL METHODS FOR LIQUID SAMPLES
(Decontamination Water and Equipment Rinsate Blanks)
Analyte
Preparation
Method1
Analysis
Method1
Metals
Arsenic
7060/7740
7060
Lead
3020
7421
Mercury
7470
7470
Selenium
7060/7740
7740
Thallium
3020
7841
Other metals2
3010
6010
Notes:
1 All methods are from SW-846 unless specified otherwise.
2 Other metals include aluminum, antimony, barium, beryllium, cadmium, calcium, chromium,
copper, iron, magnesium, manganese, potassium, silicon, silver, sodium, tin, and zinc.
76
-------
TABLE 21
ANALYTICAL METHODS FOR GAS SAMPLES
(Stack Gas)
Analyte
Sampling
Method
Front Half
Back Half
Preparation
Method1
Analysis
Method'
Preparation
Method1
Analysis
Method1
Metals2
Metals
Method 5s
Antimonv
3050
6010
3010
6010
Arsenic
3050
6010
3010
6010
Barium
3050
6010
3010
6010
Beryllium
3050
6010
3010
6010
Cadmium
3050
6010
3010
6010
Chromium
3050
6010
3010
6010
Lead
3050
6010
3010
6010
Mercury
3050
7470
3010
6010
Silver
3050
6010
3010/7761
7761
Thallium
3050
7470
7470
7470
Particulate
Method 53
Desiccation
Gravimetric
Chloride
Method 26
9252
Moisture
Method 4
Gravimetric
Notes:
1 All methods are from SW-846 unless specified otherwise.
2 The 10 toxic metals listed in Section 3005(c)(3) of RCRA were the target analytes for the
stack samples.
3 EPA, 1991a. The Method 5 sampling train has a front half (consisting of a probe and filter
assembly) used to collect particulate, and a back half (consisting of impingers) to transfer
metal vapors to a aqueous phase for analysis. Each half is analyzed independently and the
results combined.
77
-------
lithium tetraborate fusion, hydrofluoric acid dissolution, and a microwave bomb technique
developed by HRD, were tested and evaluated before the final method was selected. Discussion
of preliminary studies and rationale for comparing and selecting the various alternative digestion
methods is presented in Section 6.1. A modified EPA Method 3050, with a reduced sample size
(from 1.0 gram to 0.2 grams), was selected as the digestion method.
7.2 SAMPLE HOLDING TIMES
The dates of sample collection, preparation, and analysis are listed in Tables 22 and 23.
These tables indicate whether sample holding times were met. The sample holding times specified
in the analytical methods were met for all samples except for fluoride and sulfur in the waste feed
samples and the mercury stack gas samples. Maximum holding times for fluoride and sulfur are
28 days. For this test the actual holding times for fluoride and sulfur were 62 and 63 days,
respectively. Maximum holding time for mercury gas samples is 28 days. For this test the actual
holding time was 34 days for all of the mercury samples.
Although the fluoride and sulfur waste feed sample holding times were exceeded by more
than a month, it is believed that the quality of data were not compromised. The waste feed
samples were collected from a waste pile that had been in place for many months before sample
collection occurred, therefore, exceeding the holding time is believed to make little difference.
Holding times for the mercury stack gas samples were exceeded by 6 days, because the
samples were held until consensus was reached by HRD and EPA regarding which analytical
method should be used.
7.3 MATRIX SPIKE/MATRIX SPIKE DUPLICATE RECOVERIES
Matrix spike and matrix spike duplicates were performed on the TCLP analyses for waste
feed, effluent slag, and stack gas samples from Run 2 to ensure the accuracy and precision of
these analyses. Matrix spikes were not analyzed for metals because of the high metal content
(greater that 10 percent) of the samples. When matrix concentrations are very high, normal
spiking concentrations cannot be used as they would be overwhelmed by the high concentration of
the analyte in the sample. In addition, it would be very hard do dissolve such high concentrations
of analytes to accurately prepare such solutions. In addition, special spiking solutions are very
expensive. Therefore, duplicates were determined not to be appropriate by the analytical
78
-------
TABLE 22
DATES OF SAMPLE COLLECTION, PREPARATION,
AND ANALYSIS FOR PROCESS SAMPLES
(sheet 1 of 4)
; Sample/Analyte
Holding
Time
Collection
Date
Preparation/
Extraction
Date
Analysis
Date
Holding
Time
Met?
Waste Feed
Metals
6 months1
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
5/20, 3/21
3/21
3/22
3/23
3/21
--
6/27
6/27
6/27
Archived
Yes
Yes
YeB
Archived
TCLP Metals
6 months
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
3/21
3/22
3/23
3/21
4/3, 7/112
4/3, 7/112
4/3, 7/112
4/10-4/12, 7/112
4/10-4/12, 7/112
4/10-4/12, 7/112
Archived
Yes
Yes
Yes
Archived
Total Chloride
28 days
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20,3/21
3/21
3/22
3/23
3/21
4/16
4/16
4/16
4/18
4/18
4/18
Archived
Yes
Yes
Yes
Archived
Fluoride
28 days
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
3/21
3/22
3/23
3/21
5/17
5/17
5/22
5/24
6/24
5/25
Archived
No
No
No
Archived
Sulfur
28 days
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
3/21
3/22
3/23
3/21
4/25, 5/21
5/21-22
5/22-23
5/23
5/24
5/25
Archived
No
No
No
Archived
Total Carbon
6 months
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run S, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
3/21
3/22
3/23
3/21
—
4/18, S/24-252
4/18, 5/24-2S2
4/18, 5/24-252
Archived
Yes
Yes
Yes
Archived
79
-------
TABLE 22
DATES OF SAMPLE COLLECTION, PREPARATION,
AND ANALYSIS FOR PROCESS SAMPLES
(sheet 2 of 4)
Sample/Analyte
Holding
Time
Collection
Dat«
Preparation/
Extraction
Date
Analysis
Date
Holding
Time
Met?
Waste Faed (continued)
TOC
6 months
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
3/21
3/22
3/23
3/21
—
4/16, 5/24-252
4/16,5/24-25J
4/16, 5/24-25*
Archived
Yes
Yes
Yes
Archived
Ash Content
6 months
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
3/21
3/22
3/23
3/21
—
4/12
4/12
4/12
Archived
Yes
Yes
Yes
Archived
Moisture Content
6 months
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
3/21
3/22
3/23
3/21
—
4/9
4/9
4/9
Archived
Yes
Yes
Yes
Archived
Particle Size Distribution
6 months
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
3/21
3/22
3/23
3/21
—
5/20-6/6
5/20-6/6
5/29-6/6
Archived
Yes
Yes
Yes
Archived
Btu Content
6 months
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
3/21
3/22
3/23
3/21
—
6/27
6/27
6/27
Archived
Yes
Yes
Yes
Archived
Chlorine
6 months
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
3/21
3/22
3/23
3/21
5/21-22
5/21-22
5/22-23
5/23-24
5/24
5/25
Archived
Yes
Yes
Yes
Archived
80
-------
TABLE 22
DATES OF SAMPLE COLLECTION, PREPARATION,
AND ANALYSIS FOR PROCESS SAMPLES
(sheet 3 of 4)
Holding
:Time
Met?
Sample/Parameter
Holding
Time
Collection
Date
Preparation/
Extraction
Date
Analysis
Date
Effluent Slag
Metals
6 months1
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
3/21
3/22
3/23
3/21
--
6/26
6/26
6/26
Archived
Yes
Yes
Yes
Archived
TGLP Metals
6 months
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
8/21
3/22
3/23
3/21
4/8
4/8
4/8
4/15-4/17
4/15-4/17
4/15-4/17
Archived
Yes
Yes
Yes
Archived
Moisture Content
6 months
Run 1, Hr. 1-7
Run 2, Hr. 1-6
Run 3, Hr. 1-6
Run 4, Hr. 1-6
Run 2, Hr. 2 (Field Dup.)
3/20, 3/21
3/21
3/22
3/23
3/21
—
5/13
5/13
5/13
Archived
Yes
Yes
Yes
Archived
Oxide Product
Metals
6 months1
Run 1, Composite of Hr. 1-6
Run 1, Hr. 7
Run 2, Composite of Hr. 1-3
Run 2, Composite of Hr. 4-6
Run 3
Run 4
Run 2, Composite of Hr. 1-3
(Field Duplicate)
3/20
3/21
3/21
3/21
3/22
3/23
3/21
—
6/27
6/27
6/27
6/27
Archived
Archived
Yes
Yes
Yes
Yes
Archived
Moisture Content
6 months
Run 2 Composite
Run 3
Run 4
3/21
3/22
3/23
4/11
4/11
4/11
Yes
Yes
Yes
81
-------
TABLE 22
DATES OF SAMPLE COLLECTION, PREPARATION,
AND ANALYSIS FOR PROCESS SAMPLES
(sheet 4 of 4)
Sample/Analyte
Holding
Time
Collection
Date
Preparation/
Extraction
Date
Analysis
Date
Holding
Time
Met?
Blanks
Metals
6 months1
Sand Blanks
Equipment Blank - Oxide
Product (Run 1)
Equipment Blank - Effluent
Slag (Run 2)
Air Blank (Run 2)
Equipment Blank - Feed
(Run 3)
Decontamination Solution -
Waste
3/19
3/20
3/21
3/21
3/22
3/23
—
4/24
4/24
4/24
4/24
4/24
Archived
Yes
Yes
Yes
Yes
Yes
TCLP Metals
6 months
Sand Blanks
3/19
--
--
Archived
Notes:
The holding time for mercury is 28 days.
This sample was reanalyzed.
82
-------
TABLE 23
DATES OF SAMPLE COLLECTION, PREPARATION, AND ANALYSIS
FOR STACK SAMPLES
(sheet 1 of 3)
Holding
Collection
Preparation/
Holding
Sample/Anolyte
Time
Date
Extraction Date
Analysis Date
Time Met?
Particulate
6 months
Run 1 - Front Half Acetone
Rinse
3/20
7/2, 7/8
Yes
Run 1 - Filter
3/20
6/13, 6/14, 8/71
Yes
Run 2 - Front Half Acetone
Rinse
3/21
7/2, 7/8
Yes
Run 2 - Filter
3/21
6/13, 6/14, 8/71
Yes
Run 3 - Front Half Acetone
Rinse
3/21
7/2, 7/8
Yes
Run 3 - Filter
3/21
6/13, 6/14, 8/71
Yes
Run 4 - Front Half Acetone
3/22
7/2, 7/8
Yes
Rinse
3/22
6/13, 6/14, 8/71
Yes
Run 4 - Filter
8/23
7/2, 7/8
Yes
Run 5 - Front Half Acetone
3/23
6/13, 6/14
Yes
Rinse
3/23
7/2, 7/8, 8/71
Yes
Run 5 - Filter
3/23
6/13, 6/14
Yes
System Blank - Acetone Rinse
3/23
7/2, 7/8
Yea
System Blank - Filter
Reagent Blank - Acetone
28 days
HCl
3/20
3/20
4/U
Yes
Run 1 - Impinger 1-3 Contents
3/20
3/20
4/11
Yes
and Rinse
Run 1 - Filter
3/21
3/21
4/11
Yes
Run 2 - Impinger 1-3 Contents
3/21
3/21
4/U
Yes
and Rinse
Run 2 - Filter
3/21
3/21
4/11
Yes
Run 3 - Impinger 1-3 Contents
3/21
3/21
4/11
Yes
and Rinse
Run 3 - Filter
3/22
3/22
4/11
Yes
Run 4 - Impinger 1-3 Contents
3/22
3/22
4/11
Yes
and Rinse
3/21
3/21
4/11
Yes
Run 4 - Filter
3/21
3/21
4/11
Yes
Reagent Blank - D.I.2 Water
3/23
3/23
4/11
Yes
Reagent Blank - 0 1N NaOH
System Blank - Impingers 1-3
6 monthB
Metal*
3/20
7/9
7/11
Yes
Run 1 - Front Half HNO,
Rinse
3/20
4/16
4/23
Yes
Run 1 - Impinger 1-3 Contents
and Rinse
83
-------
TABi,E 23
DATES OF SAMPLE COLLECTION, PREPARATION, AND ANALYSIS
FOR STACK SAMPLES
(sheet 2 of 3)
Holding
Collection
Preparation/
Holding
Sample/Parameter
Time
Date
Extraction Date
Analysis Date
Time Met?
Met all (continued)
Run 2 - Front Half HN03 Rinse
3/21
7/9
7/11
Yes
Run 2 - Impinger 1-3 Contents
and Rinse
3/21
4/16
4/23
Yes
Run 3 - Front Half HNOj
Rinse
3/21
7/9
7/U
Yes
Run 3 - Impinger 1-3 Contents
and Rinse
3/21
4/16
4/23
Yes
Run 3 - O.lN HN03 Nonle
Rinse
3/21
7/9
7/11
Yes
Run 4 - Front Half HNOj
Rinse
3/22
7/9
7/11
Yes
Run 4 - Impinger 1-3 Contents
and Rinse
5/22
4/16
4/23
Yes
Run 4 - O.lN HNOj Noizle
Rinse
3/22
7/9
7/11
Yes
Run 6 - Front Half HNOj
Rinse
3/23
7/9
7/11
Yes
Run 5 - Impinger 1-3 Contents
and Rinse
3/23
4/16
4/23
Yes
Run 5 - O.lN HNOj Nozzle
Rinse
3/23
7/9
7/11
Yes
SyBtem Blank - O.lN HNO,
Nozzle Rinse
3/23
7/9
7/11
Yes
Reagent Blank - E% HNOj/
10% H202
5/23
7/9
7/11
Yes
Reagent Blank - D.I.2 Water
3/25
7/9
7/11
Yes
Reagent Blank - O.lN HNOj
3/23
7/9
7/11
Yes
Reagent Blank - 8N HC1
5/23
7/9
7/11
Yes
Reagent Blank - Filter
3/23
7/9
7/11
Yes
System Blank - Impinger 1-3
4/23
Contents and Rinse
3/23
4/16
Yes
System Blank - Front Half
HNOj Rinse
3/23
7/9
7/11
Yes
Mercury /Metals
28 days
Run 1 - Impinger Rinse 1
50 mL Aliquot
Run 1 - KMn04/8N HC1 Rinse
3/20
4/16
4/24
No5
Run 2 - Impinger Rinse 1
3/20
4/16
4/24
No3
50 mL Aliquot
3/21
4/16
4/24
No3
84
-------
TABLE 23
DATES OF SAMPLE COLLECTION, PREPARATION, AND ANALYSIS
FOR STACK SAMPLES
(sheet 3 of 3)
Holding
Time
Collection
Date
Preparation/
Holding
Time Met?
Sample/Parameter
Extraction Date
Analysis Date
Mercmy/Metmls
(continued)
Run 2 - KMn04/8N HC1 Rinse
8/21
4/16
4/24
No3
Run S - Impinger Rinse 1
50 mL Aliquot
8/21
4/16
4/24
No3
Run S - KMn04 HC1 Rinse
8/21
4/16
4/2S
No3
Run 4 - Impinger Rinse 1
Ho3
SO mL Aliquot
3/22
4/16
4/24
Run 4 - KMn04/8N HC1 Rinse
8/22
4/16
4/24
No3
Run S - Impinger Rinse 1
SO mL Aliquot
3/23
4/16
4/24
No3
Run 8 - KMn04/8N HC1 Rinse
8/23
4/16
4/24
No3
Reagent Blank - KMn04
8/23
4/16
4/24
No3
System Blank - KMn04/8N
HC1 rinse
3/23
4/16
4/24
No3
System Blank - Impinger
Rinse 1 SO mL Aliquot
3/23
4/16
4/24
No3
Notes:
These filters were reweighed.
D.l. Water = distilled, deionized water
These samples were held for analysis until extraction method was selected.
85
-------
laboratory. For a spike to be recovered, it would have to be almost 25 percent of the sample
concentration. Therefore, laboratory duplicates, as discussed in Section 7.5, were used. The
chloride, metals, and mercury gas train samples, were analyzed for matrix spike (MS) and matrix
spike duplicate (MSD) recoveries. MS analyses were also performed to determine chloride,
fluoride, sulfur, and carbon recoveries from the Run 2 waste feed samples.
The target recovery window for all of the MS and MSD recoveries is 70 to 130 percent,
and the specified range of relative percent differences (RPDs) for matrix spike duplicates is 0 to
25 percent. Tables 24, 25, and 26 show the sample results, spiked sample results, spikes added,
percent recoveries, and relative percent differences for MS and MSD samples.
All recoveries were within the target recovery windows except the following TCLP metals:
barium, cadmium, lead, and silver for waste feed; and silver for stack gases. Most of these
samples were out of specification because they were spiked at inappropriate levels. Reanalyses
were performed on the waste feed sample with the correct spike concentrations (that is equivalent
to the regulatory levels for the RCRA Toxicity Characteristic [TC] rule, or 0.5 to 5 times the
result if the result is less than half of the regulatory level). All of the reanalysis recoveries were
within the target window of 70 to 130 percent.
For the stack gas sample, only the MS and MSD metal recovery results for silver analyzed
by inductively coupled plasma atomic emission spectroscopy (ICP) were low (32.4 percent and
10.3 percent). This sample did not require reanalysis because silver was also analyzed by graphite
furnace atomic absorption spectroscopy (GFAA) and the MS and MSD results were acceptable.
After reanalysis, all RPDs for spike duplicates were within the specified limits of 0 to
25 percent, except the stack gas sample that was not reanalyzed. The RPD for leachable silver in
the stack gas sample analyzed by ICP was 104 percent, but the RPD for the GFAA silver analysis
was 3.70 percent, which is well within the control limits.
7.4 LABORATORY DUPLICATES
Laboratory duplicate samples were divided in the laboratory and analyzed as an indicator
of precision. One sample of each sample matrix (waste feed, oxide product, and effluent slag)
from Run 2 was analyzed in duplicate for all analytes. Tables 27, 28, 29, and 30 contain
laboratory duplicate results. The precision objective was an RPD of less than or equal to
86
-------
TABLE 24
TCLP MATRIX SPIKE RECOVERIES
Matrix Spike
Matrix Spike Duplicate
Analytes
Sample
Result
(mg/L)
Spiked
Sample
Result
(mg/L)
Spike
Added
(mg/L)
Percent
Recovery1
Spiked
Sample
Result
(mg/L)
Spike
Added
(mg/L)
Percent
Recovery1
Relative
Percent Difference2
TCLP - WASTE FEED
Arsenic
<2103
4.51
5.00
90.2
4.81
5.00
96.2
6.44
Barium4
0.0165
197
200
98.6
195
200
915
1.12
Cadmium4
16.1
18.1
2.00
97.7
18.0
2.00
95.0
2.80
Chromium
0.193
4.45
5.00
85.1
4.57
5.00
87.5
2.81
Lead4
4.73
15.3
10.0
105
14.1
10.0
93.2
12.3
Mercury
<0.010
0.186
0.200
93.2
0.186
0.200
93.2
0
Selenium
0.151
0.853
1.00
70.1
0.844
1.00
69.3
1.15
Silver4
<10.0
0.0176
0.0200
88.0
0.0163
0.0200
81.6
155
TCLP - EFFLUENT SLAG
Arsenic
0.616
4.70
5.00
81.7
4.51
5.00
77.8
4.89
Barium
0.200
88.9
100
88.7
86.0
100
85.8
3.32
Cadmium
<0.0500
0.884
1.00
88.4
0.830
1.00
83.0
6.30
Chromium
<0.0600
4.66
5.00
93.1
4.58
5.00
91.6
1.62
Lead
<0.330
4.25
5.00
85.0
4.08
5.00
81.5
4.20
Mercury
<0.0100
0.187
0.200
93-5
0.184
0.200
92.0
1.62
Selenium
0.0730
0.685
1.00
79.5
0.780
1.00
78.4
24.6
Silver
<0.0530
0.795
1.00
79.5
0.780
1.00
78.0
1.90
Percent recovery = (matrix spike concentration - sample concentration) x 100/matrix spike concentration
RPD = (|matrix spike recovery - matrix spike duplicate recovery| x 100)/(matrix spike recovery + matrix spike duplicate recovery)/2
< = Below detection limit
Reanalysis results are given for this metal.
-------
TABLE 25
WASTE FEED MATRIX SPIKE RECOVERIES
Matrix Spike
Spiked
Sample :
Sample
Spike
Result
Result
Added
Percent
Analytes
(mg/L)
(mg/L)
(mg/L)
Recovery'
Chloride
20.8
42.6
23.0
94.8
Fluoride
0.131
0.339
0.217
96.8
Sulfur
166
346
192
93.8
Total Carbon
120,000
365,000
236,000
104
Total Organic Carbon
150,000
325,000
156,000
112
Notes:
1 Percent recovery = (matrix spike concentration - sample concentration) x 100/matrix spike concentration
mg/L = milligram per liter
88
-------
TABLE 26
MATRIX SPIKE RECOVERIES FOR STACK SAMPLES
00
vo
Matrix Spike
Matrix Spike Duplicate
Sample/Analytes
Sample
Result
Qtg/sample)
Spiked
Sample
Result
(Mg/sample)
Spike
Added
(Mg/sample)
Percent
Recovery1
Spiked
Sample
Result
Oig/sample)
Spike
Added
(Mg/sample)
Percent
Recovery1
Relative
Percent
Difference2
METALS - STACK
Antimony
20.6
2,350
2,460
91.6
2,360
2,460
91.9
0.327
Arsenic
<13.4
2,350
2,460
92.1
2,310
2,460
90.8
1.42
Barium
29.5
1,280
1,270
98.5
1,280
1,270
98.0
0.509
Beryllium
<0.64
1,220
1,270
95.5
1,210
1,270
94.9
1.05
Cadmium
<3.18
1,190
1,270
93.8
1,170
1,270
92.2
1.72
Chromium
<3.82
1,190
1,270
93.3
1,190
1,270
93.3
0
Lead
<21.0
2,460
2,460
97.9
2,510
2,460
98.5
0.611
Silver
<3.18
412
1,270
32.42
132
1,270
10.32
1042
Thallium
<106
2,360
2,460
92.5
2,420
2,460
95.0
2.67
Silver by GFAA3
0.68
2.08
1.27
110
2.03
1.27
106
3.70
METALS/Mercuiy - STACK4
162
227
64.1
102
220
64.1
90.7
11.8
METALS/Mercury - STACK5
9.62
13.9
3.42
124
13.7
3.42
119
4.12
Chloride - STACK
2,580
6,760
4,340
96
6,760
4,340
96
0
Notes:
1 Percent recovery = (matrix spike concentration - sample concentration) x 100/matrix spike concentration
2 This value is outside control limits (0 to 25 relative percent difference or 70 to 130 percent recovery).
3 GFAA = Graphite furnace atomic absorption spectroscopy
4 For impingers 1 through 3 — see Section 5.3
5 For impinger 4 (special KMn04 impinger) — see Section 5.3
< = less than
tig = microgram
-------
TABLE 27
LABORATORY DUPLICATE RESULTS
FOR GENERAL CHEMISTRY ANALYTES
Analyte
Sample
Result
Duplicate
Result
Mean
Concentration
Relative
Percent
Difference'
WASTE FEED (mg/kg)
Chloride
29,100 mg/kg
29,300 mg/kg
29,200 mg/kg
0.685
Chlorine
21,200 mg/kg
20,400 mg/kg
20,800 mg/kg
3.85
Fluoride
138 mg/kg
124 mg/kg
131 mg/kg
10.7
Sulfur
166,000 mg/kg
166,000 mg/kg
166,000 mg/kg
0
A*h Content
82.2 percent
81.9 percent
82.0 percent
0.366
Total Carbon
11.8 percent
13.2 percent
12.5 percent
11.2
Total Organic Carbon
14.8 percent
15.3 percent
15.0 percent
3.33
Moisture Content
3.23 percent
3.28 percent
3.26 percent
1.53
Btu Content
2,020 Btu/lb
2,020 Btu/lb
2,020 Btu/lb
0
Btu Content
1,760 Btu/lb
1,710 Btu/lb
1,740 Btu/lb
2.87
Particle Size Distribution
30 < X
9.45 percent
9.69 percent
9.57 percent
2.51
60 < X < SO
8.28 percent
8.28 percent
8.28 percent
0
100 < X < 60
7.58 percent
7.16 percent
7.37 percent
5.70
200 < X < 100
8.69 percent
8.85 percent
8.77 percent
1.82
325 < X < 200
1.94 percent
2.39 percent
2.16 percent
20.8
325 < X
64.1 percent
63.6 percent
6S.8 percent
0.784
Notes:
1 Relative percent difference = (|sample result - duplicate result) * mean concentration) X 100
< = less than
mg/kg = milligram per kilogram
Btu/lb = British thermal unit per pound
90
-------
TABLE 28
LABORATORY DUPLICATE RESULTS
FOR WASTE FEED SAMPLES
Sample/Analyte
Sample
Result :
(mg/kg)
Duplicate
Result
(mg/kg)
Mean
Concentration
(mg/kg)
Relative
Percent
Difference1
WASTE FEED
Aluminum
5,470
6,940
5,710
8.23
Antimony
449
359
404
22.1
Arsenic
1,040
414
729
86.53
Barium
846
865
856
2.22
Beryllium
<1.00
<0.99
NC2
NC2
Cadmium
512
870
441
32.13
Calcium
6,820
6,860
6,840
0.585
Chromium
98.3
66.7
82.5
38.33
Copper
2,590
2,080
2,340
22.0
Iron
103,000
96,800
99,800
6.21
Lead
51,600
44,000
47,800
16.0
Magnesium
2,180
2,310
2,240
5.80
Manganese
840
764
802
9.48
Mercury
0.64
0.62
0.63
3.18
Potassium
2,700
2,320
2,510
15.5
Selenium
62.2
112
87.1
57.13
Silicon
2,000
3,610
2,800
52.13
Silver
5.38
1.58
3.48
1093
Sodium
127,000
128,000
128,000
0.781
Thallium
269
240
255
11.6
Tin
2,950
2,800
2,880
6.21
Zinc
6,810
4,630
5,720
38.13
Notes:
1 Relative percent difference = (|sample result - duplicate result| -r mean concentration) x 100
2 NC = Not calculable
3 This value is outside of the control limits of 0 to 25 relative percent difference.
< = less than
rag/kg = milligram per kilogram
91
-------
TABLE 29
LABORATORY DUPLICATE RESULTS
FOR OXIDE PRODUCT SAMPLES
Sample/ Analyte
Sample
Result
(mg/kg)
Duplicate
Result
(mg/kg)
Mean
Concentration
(mg/kg)
Relative
Percent
Difference1
OXIDE PRODUCT
Aluminum
623
462
542
29.72
Antimony
1,230
1,260
1,240
2.42
Arsenic
1,130
1,130
1,130
0
Barium
275
267
271
2.00
B«ryllium
<0.95
<0.95
<0.95
NC
Cadmium
1,370
1,370
1,370
0
Calcium
2,340
1,860
2,100
22.8
Chromium
308
308
3,08
0
Copper
1,670
1,620
1,650
2.00
Iron
31,800
31,300
31,600
1.58
Lead
180,000
173,000
177,000
3.95
Magnesium
348
334
341
4.10
Manganese
282
279
281
1.07
Mercury
0.136
0.165
0.150
19.0
Potassium
7,410
7,450
7,430
0.638
Selenium
65.9
65.0
65.5
1.53
Silicon
1,130
1,230
1,180
8.47
Silver
27.4
27.8
27.6
1.45
Sodium
168,000
163,000
165,000
3.03
Thallium
<71.4
<71.4
<71.4
NC
Tin
6,870
6,800
6,840
1.02
Zinc
16,200
16,200
16,200
0
Notes:
1 Relative percent difference = (|sample result - duplicate result| -r mean concentration) X 100
2 This value is outBide of the control limits of 0 to 25 relative percent difference.
NC = Not calculable
< = leas than
mg/kg = milligram per kilogram
92
-------
TABLE 30
LABORATORY DUPLICATE RESULTS
FOR EFFLUENT SLAG SAMPLES
Sample/Analyte
Sample
Result
(mg/kg)
Duplicate
Result
(mg/kg)
Mean
Concentration
(mg/kg)
Relative
I! Percent
Difference1
EF
FLUENT SLAG
Aluminum
14,300
14,800
14,500
3.45
Antimony
103
93.2
98.4
10.4
Arsenic
104
89.6
97.0
15.2
Barium
1,640
1,680
1,660
2.41
Beryllium
<1.00
<1.00
<1.00
NC
Cadmium
<2.50
<2.50
<2.6
NC
Calcium
13,800
13,700
13,500
2.96
Chromium
64.4
63.1
63.8
2.04
Copper
3,570
3,660
3,610
2.49
Iron
192,000
197,000
195,000
2.56
Lead
1,560
1,640
1,600
5.00
Magnesium
5,750
5,820
5,790
1.21
Manganese
2,130
2,170
2,150
1.86
Mercury
<0.10
<0.10
<0.10
NC
Potassium
2,080
2,200
2,140
5.61
Selenium
<26.0
40.8
NC
NC
Silicon
2,170
2,600
2,390
18.0
Silver
S.06
4.53
3.80
38.6*
Sodium
150,000
154,000
152,000
2.63
Thallium
535
599
567
11.4
Tin
766
786
776
2.58
Zinc
1,270
1,310
1,290
3.10
Notes:
1 Relative percent difference = ()sample result - duplicate result| -s- mean concentration) X 100
2 This value is outside of the control limits of 0 to 25 relative percent difference.
NC — Not calculable
< = less than
mg/kg = milligram per kilogram
93
-------
30 percent for total organic carbon and moisture content, and less than or equal to 25 percent for
all other analytes. When sample results were lesa than 5 times the detection limit, the control limit
was less than or equal to 100 percent.
In the waste feed duplicate sample analyzed for metals and silicon, all RPDs were within
the control limits of less than or equal to 25 percent (or 0 to 100 percent for low concentration
samples) except arsenic, cadmium, chromium, silicon, silver, and zinc (86.5, 32.1, 38.3, 52.1, 109,
and 38.1, respectively). Silver results were less than five times the detection limit, so the RPD of
109 percent is just outside the control limits of 0 to 100 percent. For metals analysis of the oxide
product, only the RPD of 29.7 percent for aluminum was greater than 25 percent. The moisture
content duplicate of 49.2 percent RPD for the oxide product was within the limit of 100 percent
RPD for sample results near the limit of detection. For metal analysis of effluent slag, all RPDs
were less than or equal to 25 percent except for silver (38.6 percent). Data that fall outside of the
quality control (QC) limits have been flagged as such, and the EPA Project Managers have been
notified.
7.5 FIELD DUPLICATES
Field duplicates were collected during Run 2, and archived, as specified in a November
2, 1990 memo concerning changes to the HRD SITE QAPP (EPA, 1990). Because a large number
of waste feed (18 samples), oxide product (3 samples), and effluent slag (18 samples) samples were
collected and analyzed, the collection of duplicate samples was redundant and the field duplicates
were not analyzed.
7.6 REFERENCE STANDARDS
The following reference materials were analyzed with each batch (Batch 1 was the waste
feed samples, Batch 2 was the oxide product samples, and Batch 3 was the effluent slag samples)
of metal samples: (1) lead blast furnace slag reference material, (2) Flame Reactor oxide reference
material, (3) EURO-Standard 877-1 Furnace Dust reference material, (4) EURO-Enchantillon
876-1 certified reference material, and (5) National Institute of Science and Technology (NIST)
1633a Trace Elements in Coal Fly Ash certified standard reference material. Analytical results for
the first four standards were compared to the certified values for an indication of accuracy.
Because of the high level of metals in the demonstration samples, the NIST 1633a trace element
standard was considered inappropriate. Relative standard deviations (RSD) and percent recoveries
were calculated for each of three replicate samples - one analyzed with the waste feed samples
94
-------
(Batch 1), one analyzed with the effluent slag samples (Batch 2), and one analyzed with the oxide
product samples (Batch 3). These results are shown in Tables 31, 32, 33, and 34. Reference
material recoveries for each batch give an indication of accuracy for that particular batch or
matrix; and although no acceptance criteria were specified, mean recoveries and RSDs give an
indication of precision and accuracy for the entire metal analytical results.
For the lead blast furnace slag standard, mean recoveries for cadmium (62.0 percent), lead
(83.8 percent), and zinc (126 percent) were all greater than 60 percent. Chromium had a mean
recovery of only 17.7 percent. Although this recovery is very low, it is not an unexpected result
for chromium digestion by Method 3050. The rationale for selection of this method is given in
Section 6.1.
For the Flame Reactor oxide product, precision and accuracy were acceptable, with all
recoveries greater than 75 percent and all RSDs less than 3 percent.
For EURO-Standard 877-1 furnace dust standard, all mean recoveries were greater than
75 percent except silicon (27.1 percent). Arsenic recovery was 214 percent. All RSDs were below
5 percent, except for these same constituents. The RSD for arsenic was 15.0 percent and the RSD
for silicon was 9.76 percent. The analytical results for potassium were very close to the limits of
detection.
For EURO-Enchantillon 876-1, all mean recoveries were greater than 75 percent, except
for silicon (22.7 percent). All RSDs were less than 5 percent except for silicon, with an RSD of
6.34 percent.
One of the five reference materials, the NIST 1633a coal fly ash standard reference
material, had low recoveries for every analyte except arsenic. This indicates that the Method
3050 digestion method is not appropriate for the coal fly ash matrix, probably because the matrix
is resistant to the acid leaching procedure. However, the low recoveries of metals from the coal
fly ash standard reference material do not necessarily transfer to the oxide product matrix
samples. The coal fly ash standard contains only trace elements of metals, which is not
comparable to the high concentration wastes associated with the Flame Reactor process. Thus, the
NIST 1633a coal fly ash standard reference material does not appear to be an appropriate indicator
of accuracy for the metal oxide product. Two other reference materials tested, the EURO 877-1
furnace dust standard and the HRD Flame Reactor oxide product standard, are much more similar
95
-------
TABLE 31
COMPARISON OF CERTIFIED ASSAY VALUES AND LABORATORY RESULTS FOR LEAD
BLAST FURNACE SLAG STANDARD REFERENCE MATERIAL
(weight percent)
Analyte
Certified
Value
Batch 1
Percent
Recovery1
Batch 2
Percent
Recovery1
Batch 3
Percent
Recovery1
Mean
Result
Relative
Standard
Deviation
Mean
Percent
Recovery
Cadmium
0.002
0.00150
75.0
0.00140
70.0
0.00082
41.0
0.00124
29.6
62.0
Chromium2
0.035
0.00659
18.8
0.00615
17.6
0.00580
16.6
0.00618
6.4
17.7
Lead
1.360
1.21
89.0
1.13
83.1
1.08
79.4
1.14
5.75
83.8
Zinc
10.110
10.0
98.9
18.9
187
9.25
91.5
12.7
42.2
126
Notes:
Percent recovery = (certified concentration value - sample concentration) x 100/certified concentration value
It was recognized during method selection that chromium results would be biased low by using EPA Method 3050 for sample digestion;
however, this method was selected for its overall applicability to digestion and analysis of a variety of metals, including lead
and zinc, which were critical to the project.
-------
TABLE 32
COMPARISON OF CERTIFIED ASSAY VALUES AND LABORATORY RESULTS
FOR FLAME REACTOR OXIDE STANDARD REFERENCE MATERIAL
(weight percent)
Analyte
Certified
Value
Batch 1
Percent
Recovery1
Batch 2
Percent
Recovery1
Batch 3
Percent
Recovery1
Mean
Result
Relative
Standard
Deviation
Mean
Percent
Recovery
Cadmium
0.562
0.494
87.9
0.487
86.6
0.490
87.2
0.490
0.716
87.2
Chromium
0.033
0.0258
78.2
0.0270
81.8
0.0268
81.2
0.0265
2.42
80.3
Lead
7.23
6.54
90.4
6.86
94.9
6.73
93.1
6.71
2.40
92.8
Zinc
35.8
35.2
98.3
36.5
102
36.1
101
35.9
1.86
100
Notes:
Percent recovery = (certified concentration value - sample concentration) x 100/certified concentration value
-------
TABLE 33
COMPARISON OF CERTIFIED ASSAY VALUES AND LABORATORY RESULTS FOR
EURO-STANDARD 877-1 FURNACE DUST STANDARD REFERENCE MATERIAL
(weight percent)
Analyte
Certified
Value
Batch 1
Percent
Recovery1
Batch 2
Percent
Recovery1
Batch 3
Percent
Recovery1
Mean
Result
Relative
Standard
Deviation
Mean
Percent
Recovery
Aluminum
0.044
0.0378
85.9
0.0395
89.9
0.0370
84.1
0.0381
3.36
86.4
Arsenic
0.014
0.0345
246
0.0312
223
0.0256
183
0.0304
15.0
214
Calcium
3.23
2.85
88.2
2.86
88.5
2.66
82.4
2.79
4.04
86.4
Chromium
0.017
0.0147
86.5
0.0143
84.1
0.0136
80.0
0.0142
3.94
83.5
Copper
0.025
0.0221
88.4
0.0219
87.6
0.0211
84.4
0.0217
2.44
86.8
Iron
62.07
60.5
97.5
60.8
98.0
58.3
93.9
59.9
2.28
96.5
Lead
1.00
0.785
78.5
0.835
83.5
0.767
76.7
0.796
4.42
79.6
Magnesium
0.28
0.235
83.9
0.244
87.1
0.232
82.9
0.237
2.63
84.6
Manganese
1.37
1.19
86.9
1.19
86.9
1.12
81.8
1.17
3.46
85.2
Potassium
0.058
<0.023
<39.7
0.0688
119
<0.023
<39.7
NC
NC
NC
Silicon
1.08
0.263
24.4
0.320
29.6
0.295
27.3
0.293
9.76
27.1
Sodium
0.23
0.219
95.2
0.227
98.7
0.228
99.1
0.225
2.18
97.8
Zinc
1.16
0.929
80.1
0.934
80.5
0.861
74.2
0.908
4.49
78.3
Notes:
1 Percent recovery = (certified concentration value - sample concentration) x 100/certified concentration value
< = less than
NC = not calculated
-------
TABLE 34
COMPARISON OF CERTIFIED ASSAY VALUES AND LABORATORY RESULTS FOR
EURO-ENCHANTILLON 876-1 STANDARD REFERENCE MATERIAL
(weight percent)
Analyte
Certified
Value
Batch 1
Percent
Recovery1
Batch 2
Percent
Recovery1
Batch 3
Percent
Recovery1
Mean
Result
Relative
Standard
Deviation
Mean
Percent
Recovery
Aluminum
0.34
0.283
83.2
0.279
82.1
0.268
78.8
0.277
2.81
81.5
Arsenic
0.023
0.0252
110
0.0242
105
0.0229
99.6
0.0241
4.77
105
Cadmium
0.13
0.108
83.1
0.106
81.5
0.101
77.7
0.105
3.45
80.8
Calcium
3.43
3.11
90.7
3.03
88.3
2.85
83.1
3.00
4.43
87.5
Chromium
0.17
0.138
81.2
0.135
79.4
0.127
74.7
0.133
4.28
78.2
Copper
0.42
0.377
89.8
0.365
86.9
0.360
85.7
0.367
2.38
87.4
Iron
24.85
21.9
88.1
21.5
86.5
20.3
81.7
21.2
3.92
85.4
Lead
7.82
6.37
81.4
6.66
85.2
6.18
79.0
6.40
3.78
81.8
Magnesium
1.31
1.10
84.0
0.0504
3.85
1.11
84.7
0.753
80.9
57.5
Manganese
2.84
2.48
87.3
2.40
84.5
2.29
80.6
2.39
4.00
84.2
Potassium
1.63
1.40
85.9
0.0514
3.15
1.39
85.3
0.947
81.9
58.1
Silicon
1.72
0.363
21.1
0.398
23.1
0.411
23.9
0.391
6.34
22.7
Sodium
1.98
1.85
93.4
1.83
92.4
1.83
92.4
1.84
0.625
93.4
Tin
0.094
0.0832
88.5
0.0824
87.7
0.0816
86.8
0.0824
0.971
87.7
Zinc
23.29
21.8
93.6
21.0
90.2
21.0
90.2
21.3
2.17
91.3
Notes:
Percent recovery = (certified concentration value - sample concentration) x 100/certified concentration value
-------
to the oxide product matrix and should be more true indicators of accuracy than the coal dust
reference material.
7.7 FIELD BLANKS, EQUIPMENT BLANKS, AND SYSTEM BLANKS
The field blanks consisted of sand. These blanks were collected before the test but, at the
direction of the EPA Project Managers, were not analyzed. Equipment blanks were collected and
analyzed for the oxide product sample scoop rinsate, effluent slag sample scoop rinsate, and feed
equipment rinsate. An air blank was also collected and analyzed, as was a sample of the
decontamination solution.
For each type of stack gas sample (metals, particulate, HC1), system blanks and reagent
blanks were collected and analyzed. System blanks were below detection limit for all metals
except barium, chromium, and silver. All reagent blanks were below detection limit except for
barium and mercury in the hydrochloric acid and chromium in the nitric acid/peroxide mixture.
The barium and chromium in the system blanks was probably carried over from the
reagent blanks. For both metals, the system blank contamination was lower than the reagent blank
contamination. Contaminants were detected at up to five times the detection limit, with sample
concentrations approximately five times the blank concentration. Even when blank contamination
was included with the sample results, barium was well below the Tier II Screening Limits, and
chromium was well above the screening limits. Adjusting the data for blank contamination would
not make a difference in the outcome of the emissions analysis.
Silver was high in the system blank and mercury was high in the HC1 reagent blank.
There is no known explanation for this contamination. However, emissions for both metals are
below the Tier II Screening Limits (based on the highest sample concentration) and there would be
no change in the outcome of the emissions analysis even if the data was blank-corrected.
Most blank results were below or near the detection limit. None was greater than 10 times
the detection limit. The levels of contamination found in the blanks are insignificant when
compared to the actual samples.
100
-------
7.8
BLANK AIR RUN
The HRD SITE Demonstration Plan specified that three background stack samples would
be collected (EPA, 1990). However, before the demonstration began, the EPA Project Managers,
with the concurrence of the EPA QA manager and EPA QA auditors, decided that only one blank
run would be conducted. Although the QA auditors commented that one run would not be
statistically significant (a minimum of three is required for statistical significance), the EPA
Project Managers and the technology vendor both agreed that, because the background run was
not truly representative of background contamination, only one background run would be
conducted and that the three treatment runs would not be corrected for background
contamination.
The blank run exhibited elevated metal and particulate emissions. For arsenic, antimony,
and particulate, the blank run emissions were higher than the treatment run emissions. The three
treatment runs were not corrected to account for the contaminant levels in the blanks, but the
highest concentration sample results were used to determine whether or not permissible limits
were exceeded.
7.9 METHOD BLANKS AND REAGENT BLANKS
One method blank, laboratory grade pure water that underwent all the sample preparation
(EPA Method 3050 digestion) steps, was prepared for each set of metal samples of similar matrix
(such as waste feed, oxide product, and effluent slag). Laboratory reagent blanks were analyzed
for each batch of reagents.
A contamination level of greater than 10 times the detection limit was considered outside
control limits. Most blanks had concentrations below the detection limit. All laboratory blanks
with metals concentrations of greater than 10 times the detection limit are indicated below.
• The method blank for metals in the effluent slag had concentrations for calcium
and magnesium of greater than 10 times the detection limit. The calcium
concentration was 48.2 mg/kg, and the detection limit was 4.5 mg/kg. The
magnesium concentration was 22.6 mg/kg, and the detection limit was 0.5 mg/kg.
• The sulfur reagent blank for waste feed had a concentration of 3,990 mg/kg and a
detection limit of 10 mg/kg.
101
-------
Because the average sample concentration of calcium and magnesium in the effluent slag
was 13,000 mg/kg and 5,430 mg/kg, respectively; the relatively low levels of these metals detected
in the method blank are not of concern. The level of contamination found in the sulfur reagent
blank is also insignificant when compared with the actual samples, as the average concentration of
sulfur in the samples was 52,500 mg/kg.
7.10 INSTRUMENT CALIBRATION
Calibration standards were prepared and analyzed, as stated in the QAPP, in accordance
with specifications provided in the methods. All initial and continuing calibration verification
recoveries were within control limits. All initial and continuing calibration blanks were below or
near the detection limit. For metals, the levels of elements detected above the instrument
detection limits in the various blanks may be considered typical of the method. All blank
concentrations were less than 10 times the detection limit.
7.11 DETECTION LIMITS
Method detection limits were lower than analyte concentrations, allowing evaluation of all
sample matrices. The typical instrument detection limits for metal sample analysis using ICP and
GFAA are shown in Table 18. Detection limits varied, depending on sample matrix and sample
size.
The particulate results had a higher detection limit than specified in the QAPP due to
analytical problems with the particulate filter weights and the acetone blowdown weights. Seven
filters were weighed after the SITE Demonstration (the blank run, 4 test runs, the system blank,
and the reagent blank). Only three of these filters were found to have positive weight: the blank
run (44.8 mg), Run 3 (7.0 mg), and the reagent blank (1.0 mg). The other filters had negative
weights. The reason for this problem is unknown but is probably related to a weighing error. In
addition, a weighing error occurred with the acetone blowdown residuals, which were only
weighed to an accuracy of 10 mg, rather than to the 0.1 mg accuracy specified in the QAPP. A
worst case value of 10 mg was used in all calculations for the acetone blowdown weight.
102
-------
7.12 REPLICaTE ANALYSIS
The HRD SITE Demonstration collected samples from three replicate runs (three runs
operated under the same conditions). The analytical results from all three runs are combined for
each matrix and presented in Tables 4 (waste feed), 8 (oxide product), and 10 (effluent slag). To
determine variability associated with the HRD Flame Reactor process and the relationship to the
sample matrices, standard deviations were calculated for metals based on 18 samples each for the
waste feed and effluent slag and three samples for the oxide product.
For waste feed, all standard deviations were below 40 percent. Arsenic, selenium, silicon,
and silver had RSDs between 25 percent and 40 percent.
Standard deviations for effluent slag showed much greater variability than standard
deviations for waste feed. All RSDs were below 40 percent, except cadmium (68.0 percent),
chromium (88.3 percent), and lead (45.6 percent). Antimony, arsenic, and selenium all had RSDs
greater than 100 percent (J 15 percent, 111 percent, and 100 percent, respectively).
For the oxide product, all standard deviations were below 25 percent.
Possible explanations for the high variation in sample replicates, most of which occur in
the effluent slag matrix, include the following:
• Nonhomogeneity of the sample matrix
• Analytical problems
Because effluent slag samples show more variability than the waste feed and oxide product
samples, the problem is probably related to lack of sample homogeneity, especially because of the
difficulty in preparing a composite sample from this matrix.
Most of the metals showing variability among samples collected over the entire testing
period are classified as volatile metals (antimony, arsenic, cadmium, lead, selenium, and zinc). It
is possible that variation in the results was affected by (1) temperature changes during processing,
and (2) analysis that affected the amount of volatilization and, thus, the metals concentrations.
103
-------
7.13 QUALITY ASSURANCE CONCLUSIONS
In general, most of the analytical results are credible based on the supporting QA/QC data
and were of acceptable quality for evaluation of the demonstration objectives. The primary
objectives that utilized the analytical results were as follows:
• Evaluate the technology's ability to treat waste materials to form a potentially
recyclable metal oxide product and a nonhazardous fused slag
• Evaluate the system's reliability
These objectives involved the three primary constituents of concern; lead, zinc, and
cadmium, which are of interest because of their hazardous nature and high toxicity, and also
because the recovery of these metals in the oxide product is a major objective of the Flame
Reactor technology. Lead is of greatest concern because it is present at high levels in the lead slag
waste feed. Zinc is also of concern because much of HRD's previous data was generated from
treatment of EAF dust containing high levels of zinc.
All accuracy and precision data for lead were within the specified control limits.
For cadmium and zinc in the waste feed, the relative percent difference between the
sample and sample duplicate of the laboratory duplicate analysis was slightly high. Considering
the good overall precision of the 18 replicate samples and the consistency of the concentrations of
waste feed samples collected throughout the test, the precision of the waste feed measurements is
believed to be better than the RPDs from the sample/sample duplicate analysis indicate. Thus, the
duplicate RPDs were used as a conservative measure of the actual precision of the data.
The recovery of cadmium from the lead blast furnace slag reference material was only
62.0 percent. Although this one measure of accuracy indicates a possible low bias for cadmium in
the effluent slag, the test objectives were not impacted for several reasons. First, the ability of
the process to form a nonhazardous slag was determined in accordance with regulatory protocols
on the basis of metals analysis using the TCLP test procedure. QC data for the TCLP analysis
indicated a high level of accuracy. Second, the concentrations of recyclable metals in the metal
oxide product were determined to be high enough for successful recovery based on the total
metals data obtained. Even if cadmium is actually present at slightly higher levels than reported,
104
-------
the oxide product would still be suitable for metals recovery, so the test objective would not be
affected.
The system's reliability in meeting the first objective was demonstrated by the replicate
sample results.
The chromium and silicon results are known to be biased low (but supplemental results for
split samples digested by mineral acid digestion are available for comparison). A high recovery
value for arsenic from the EURO-877-1 furnace dust standard indicates that arsenic results may
be biased high. However, these analytes are not as critical as lead, cadmium, and zinc, and no
major test conclusions were based on the affected data.
There was also a problem with the quality of air emissions data for particulate (due to a
measurement error), and the concentrations determined in the air blanks for metals and particulate
showed high background levels (higher than the test run concentrations for particulate, arsenic,
and antimony). These problems could have affected the secondary objective of air emissions
assessment. However, low quality results for particulate provided numbers sufficient for
worst-case evaluation of particulate emissions, and the highest concentration results were used to
determine whether or not permissible particulate emissions limits were exceeded. Thus, the data
available were sufficient for a worst-case assessment of particulate emissions, but data of higher
quality would have allowed for a more comprehensive assessment.
Barium, chromium, and mercury were found in reagent blanks; and barium, chromium,
and silver were found in system blanks, with all except silver at levels lower than the sample
concentrations. The blank contamination had no effect on the outcome of the emissions analysis
(i.e., whether Tier II Screening Limits were met).
105
-------
8.0 COST OF DEMONSTRATION
The cost (rounded to the nearest $100) of conducting the HRD SITE Demonstration was
approximately $532,900. This cost includes site characterization and preparation, collection and
loading of the SLS slag at the NSR site in Atlanta, transportation of the SLS slag to the HRD
Flame Reactor facility in Monaca, demonstration planning, demonstration field work, chemical
analyses, and report preparation. The developer's portion of this cost was $122,500, and the
balance of $491,100 was allocated to the SITE Program. EPA costs for labor and travel are not
included in this cost breakdown.
8.1 EPA SITE CONTRACTOR COSTS
Each SITE Project is divided into two phases. Phase I is for planning and Phase II is for
demonstration of the technology. Costs for each phase are presented below, along with a list of
the activities performed during each phase. Phase I costs are actual costs incurred before the
HRD SITE Demonstration; Phase II costs include actual costs plus estimates for labor to complete
the Technology Evaluation Report, Applications Analysis Report, and the technology
demonstration videotape.
8.1.1 Phase I: Planning
Phase I activities included the following:
• Site sampling and treatability testing
• Sampling and Analysis Plan development
• Demonstration Plan development
• Site subcontractor procurement
• Waste handling and transportation
106
-------
Costs for Phase T are summarized below by cost category:
Labor $120,200
Equipment and supplies 9,900
Travel 5,600
Waste handling 16.000
TOTAL $151,700
8.1.2 Phase II: Demonstration
Phase II activities included the following:
• Site preparation, mobilization, and demobilization
• Sample collection and field oversight
• Chemical analysis (field and off-site)
• Report and videotape preparation
• Demonstration waste disposal
Costs for Phase II are summarized below by cost category:
Labor $294,200
Equipment and supplies 28,000
Travel 16,000
Waste disposal 2.200
TOTAL $340,400
107
-------
8.2
DEVELOPER COSTS
This section presents costs incurred by HRD in preparing and conducting the
demonstration. These costs including:
• Field work
• Consumables (primarily oxygen and natural gas)
• Utilities (mainly electricity)
• Equipment and supplies
• Analytical costs.
HRD costs for processing 72 tons of SLS slag during the demonstration are presented
below:
Labor $46,500
Consumable 11,000
Utilities 9,000
Travel 1,100
Analytical 50,700
Equipment and supplies 4 200
TOTAL $122,500
108
-------
REFERENCES
American Society for Testing and Materials (ASTM), 1991. Methods Published Annually by the
American Society for Testing and Materials.
Code of Federal Regulations (CFR), 1988. Protection of Environment, Title 40, Parts 190 to 299,
260.10 Subpart B: Definitions, 264 Subpart O Incinerators.
Chemical Rubber Company (CRC), 1990. CRC Handbook of Chemistry and Physics, 1990-1991.
Federal Register (FR), 1990. 55 FR 17862. Standards for Owners and Operators of Hazardous
Waste Incinerators and Burning of Hazardous Wastes in Boilers and Industrial Furnaces
Proposed and Supplemental Proposed Rule, Technical Corrections and Request for
Comments. Federal Register Volume 55, No. 82, April 27, 1990.
Horsehead Resource Development Company, Inc. (HRD), 1989a. HRD Flame Reactor Process -
A High Temperature, Gas-Fired Flash Smelting Process. A proposal in response to U.S.
EPA RFP-004, March 3, 1989.
Horsehead Resource Development Company, Inc. (HRD), 1989b. HRD Flame Reactor Process -
Product Literature, April 1989.
NL Industries, 1989. A Project Work Plan (Phase I, Tasks A, B, and D) for the National Smelting
and Refining Site, 451 Bishop Street, Atlanta, Georgia, September 4, 1989.
NL Industries, 1990. Results of Task B - Subsurface Investigation.
Perry, 1985, Perry's Chemical Engineer's Handbook, 6th Edition, McGraw-Hill Book Company,
New York.
United States Environmental Protection Agency (EPA), 1979. Methods for the Chemical Analysis
of Water and Wastes, EPA-600/4-79-020, Revised March 1983, Environmental Monitoring
and Support Laboratory, Cincinnati, Ohio, U.S. Environmental Protection Agency (EPA),
and subsequent EPA-600/4 Technical Additions.
United States Environmental Protection Agency (EPA), 1986. Test Methods for Evaluating Solid
Waste, Volumes IA-1C and subsequent revisions: Laboratory Manual, Physical/Chemical
Methods; and Volume II: Field Manual, Physical/Chemical Methods, SW-846, Third
Edition, Office of Solid Waste, Document Control No. 955-001-00000-1.
United States Environmental Protection Agency (EPA), 1987. Preparation Aid for HWERL's
Category II Quality Assurance Project Plan, HWERL Document Control No. PA QAPjP,
0007-GFS.
United States Environmental Protection Agency (EPA), 1989a. Administrative Order on Consent,
in the Matter of National Smelting and Refining Site, Atlanta, Georgia, and NL Industries,
Inc., Houston, Texas, EPA Docket No. 89-26-C.
109
-------
United States Environmental Protection Agency (EPA), 1989b. Guidance on Metal and Hydrogen
Chloride Controls for Hazardous Waste Incinerators. Volume IV of Hazardous Waste
Incineration Guidance Series. Office of Solid Waste, Waste Treatment Branch,
Washington, D.C., U.S. Environmental Protection Agency.
United States Environmental Protection Agency (EPA), 1990. Demonstration Plan for the
Horsehead Resource Development Company Flame Reactor. Prepared by PRC
Environmental Management, Inc. for the Superfund Innovative Technology Evaluation
Program.
United States Environmental Protection Agency (EPA), 1991a. Methods Manual for Compliance
with the BIF Regulations. EPA/530-SW-91-001.
United States Environmental Protection Agency (EPA), 1991b. Preparation Aids for the
Development of Category II Quality Assurance Project Plans. EPA/600/8-91/004,
February 1991.
United States Environmental Protection Agency (EPA), 1992. Applications Analysis Report.
Horsehead Resource Development Company, Inc. Flame Reactor Technology. U.S. EPA
SITE Program (to be published).
110
------- |