TECHNOLOGY EVALUATION REPORT
RETECH, INC., PLASMA CENTRIFUGAL FURNACE
Volume I
Science Applications International Corporation
EPA Contract No. 68-CO-0048
Work Assignment WA-019
SAIC Project No. 01-0831-07-0222-XXX
Project Officer:
Laurel J. Staley
Risk Reduction Engineering Laboratory
Cincinnati, Ohio 45268
RISK REDUCTION ENGINEERING LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
-------
NOTICE
The information in this document has been funded by the U.S. Environmental
Protection Agency under Contract No. 68-CO-0048 and the Superfund Innovative
Technology Evaluation (SITE) Program. It has been subjected to the Agency'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 for use. .
-------
-------
FOREWORD
The Superfund Innovative Technology Evaluation (SITE) Program was authorized by
the 1986 Superfund Amendments and Reauthorization Act (SARA). The Program is
a joint effort by EPA's Office of Solid Waste and Emergency Response (OSWER) and
Office of Research and Development (ORD) to enhance the development of hazardous
waste treatment technologies necessary for implementing new cleanup standards that
require greater reliance on permanent remedies. This is accomplished by performing
technical demonstrations that provide engineering and economic data on selected
technologies .-
This project consists of an analysis of the Retech, Inc.-Plasma Centrifugal Furnace
The Demonstration Tests took place at the Department of Energy's Component
Development and Integration Facility located in Butte, Montana. The demonstration
effort was directed at obtaining information on the performance and cost of the
process in order to assess the technology's potential applications at other hazardous
waste sites. This Technology Evaluation Report describes the field activities and
laboratory results from the Demonstration Tests. An interpretation of the available
data, an economic analysis, and a discussion of the potential applicability of the
technology is provided in the Applications Analysis Report.
Additional copies of this report may be obtained at no charge from the EPA's Center
for Environmental Research Information, 26 West Martin Luther King Drive, Cincinnati
Oh,o, 45268, using the EPA document number found on the report's front cover'
Once this supply is exhausted, copies can be purchased from the National Technical
HI
-------
Information Service, Ravensworth Building, Springfield, Virginia, 22161, (703) 487-
4600. Reference copies will be available in the Hazardous Waste Collection at the
EPA libraries. Information regarding the availability of other reports can be obtained
by calling the SITE Clearinghouse Hotline at (800) 424-9346 or (202) 382-3000 in
Washington, D.C.
E. Timothy Oppelt
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
IV
-------
ABSTRACT
A demonstration of the Retech, Inc. Plasma Centrifugal Furnace (PCF) was conducted
under the Superfund Innovative Technology Evaluation (SITE) Program at the
Department of Energy's (DOE's) Component Development and Integration Facility in
Bune, Montana. The furnace uses heat generated from a plasma arc to melt and
vitrify solid feed material. The feed soil was a mixture of Silver Bow Creek soil and
10% by weight No. 2 diesei oil, spiked to provide 28,000 ppm zinc oxide and 1,000
ppm hexachlorobenzene in the soil/oil mixture.
Pre-treatrnent soil and scrubber liquor/makeup sampling was performed to characterize
the material inputs to the process. Following treatment, the vitrified soil, scrubber
liquor, and stack gas were sampled to determine the technology's suitability for use
in destroying and immobilizing contaminants in the test soil. The results from this
testing were used to draw conclusions on the technology.
The following conclusions were derived from the test results: (1) the treated soil did
not leach any metals at levels above the regulatory limits; (2) the process achieved a
Destruction and Removal Efficiency.(ORE) of greater than 99.99% for the Principal
Organic Hazardous Constituent (POHC); (3) the air pollution control system did not
reduce the level of paniculate emissions to below the ,RCRA limit; (4) a high
percentage of the metals fed to the furnace are encapsulated in the treated soil; (§)
the PCF is advantageous over other incinerator technologies in that it can success
fully immobilize heavy metals in the slag, however, this treatment option can be more
expensive than conventional incineration systems.
-------
CONTENTS
SECTION PAS!
NOTICE H
FOREWORD iii
ABSTRACT v
FIGURES . . . ... ix
TABLES x
ABBREVIATIONS AND SYMBOLS .... . . xii
1. EXECUTIVE SUMMARY •'...- ..1
INTRODUCTION . , 1
CONCLUSIONS AND RESULTS 3
2. INTRODUCTION . 7
SITE PROGRAM OBJECTIVES '. 7
PROJECT BACKGROUND 8
3. PROCESS DESCRIPTION . . ., .10
TEST SOIL 10
TEST LOCATION '. .'...'.....11
DETAILED PROCESS DESCRIPTION , ''. 12
v«
-------
CONTENTS (CONTINUED)
4. FIELD OPERATIONS DOCUMENTATION 23
. TEST SUMMARY 23
DESCRIPTION OF OPERATIONS j _. 25
OPERATIONAL LOG FOR THE DEMONSTRATION TESTS 30
5. SAMPLING AND ANALYSIS PROGRAM .35
SAMPLING PROTOCOLS . . 35
ANALYTICAL PROTOCOLS . 50
6. PERFORMANCE AND DATA EVALUATION 59
INTRODUCTION 59
TOXICITY CHARACTERISTIC LEACHING PROCEDURE 61
DESTRUCTION AND REMOVAL EFFICIENCY (S3
ACID GAS REMOVAL AND PARTICULATE EMISSIONS .;.......... 65
AIR EMISSIONS '. .' 67
TEST SOIL AND TREATED SLAG . . . . 72
SCRUBBER LIQUOR . 75
CONTINUOUS EMISSION MONITORS . . , 76
FURNACE OPERATION , 80
7. CONCLUSIONS . . I . . , 33
LEACHABILITY OF TREATED SOIL FOR INORGANIC COMPOUNDS ... 83
LEACH ABILITY OF TREATED SOIL FOR ORGANIC COMPOUNDS ..... 84
DESTRUCTION AND REMOVAL EFFICIENCY OF TARGET ANALYTES . . 84
STACK GAS EMISSIONS . 35
AIR POLLUTION CONTROL SYSTEM 37
CONTINUOUS EMISSION MONITORS -...'• 89
SYSTEM PERFORMANCE AND RELIABILITY 90
vii
-------
CONTENTS (CONTINUED)
COST OF COMMERCIAL OPERATION 90
8. QUALITY ASSURANCE .-•••' 92
INTRODUCTION , 92
PROCEDURES DEFINING DATA QUALITY CONTROL AND USABILITY . 94
ANALYTICAL QUALITY CONTROL . 97
AUDIT FINDINGS 160
MODIFICATIONS AND DEVIATIONS FROM THE QAPP ........... 163
SPECIAL STUDIES i 166
SAMPLE HOLDING TIMES 170
CONCLUSIONS AND LIMITATIONS OF DATA 172
VII!
-------
FIGURES
MSJIMEiER
1. SCHEMATIC OF PLASMA CENTRIFUGAL FURNACE SYSTEM ..13
2. RETECH'S CENTRIFUGAL FURNACE WITH TRANSFERRED
PLASMA ARC 17
3. SCHEMATIC OF SAMPLING LOCATIONS FOR THE
DEMONSTRATION TESTS 36
4. CO PLOT FOR DEMONSTRATION JEST 3 77
5. C02 PLOT FOR DEMONSTRATION TEST 3 77
6. O2 PLOT FOR DEMONSTRATION TEST 3 ,..;....... 78
7. NOX PLOT FOR DEMONSTRATION TEST 3 78
8. THC PLOT FOR DEMONSTRATION TEST 3 79
, ix
-------
TABLES
NUMBER EMI
1. DEMONSTRATION TEST OPERATING CONDITIONS 26
2. OPERATING PARAMETERS MONITORED BY THE DAS 28
3. OPERATING PARAMETERS CALCULATED BY THE DAS 29
4. SUMMARY OF SELECTED PROpESS PARAMETERS 32
•5. SAMPLES COLLECTED DURING DEMONSTRATION TESTS 38
6. VOST SAMPLE IDENTIFICATION .,.,.... 47
7. TCLP RESULTS FOR DEMONSTRATION TESTS ............ 62
8. ORGANIC COMPOUNDS IN THE DEMONSTRATION TEST SOIL .. 64
9. - DRE RESULTS FOR DEMONSTRATION TESTS 64
10. PARTICULATE RESULTS FOR DEMONSTRATION TESTS 86
11. STACK GAS COMPOSITION DURING THE
DEMONSTRATION TESTS ..,..' 68
12. METALS IN THE DEMONSTRATION TEST FEED SOIL
AND TREATED SOIL . 74
13. VOLATILE SURROGATE SUMMARY DATA . . 100
14. FEED SOIL VOLATILE DUPLICATE SAMPLE RESULTS
(TEST 1) 100
15. SCRUBBER WATER VOLATILE DUPLICATE SAMPLE RESULTS . 103
16. SURROGATE RECOVERIES FOR VOST ANALYSES 105
17. VOST METHOD SPIKE RESULTS 105
-------
TABLES (CONTINUED)
18. VOST DUPLICATE SAMPLE RESULTS 107
19. . SEMIVOLATILE FEED SAMPLE SURROGATE RECOVERIES .... 111
20. FEED SOIL MATRIX SPIKE RESULTS (TEST 1) 111
21. DUPLICATE SEMIVOLATILE FEED SAMPLE RESULTS ....... 113
22, SEMIVOLATILE TREATED SOIL SURROGATE RECOVERIES ... 114
23. SEMIVOLATILE TREATED SOIL MATRIX SPIKES 115
24. SCRUBBER WATER SEMIVOLATILE SURROGATE RECOVERIES 117
25. SCRUBBER WATER SEMIVOLATILE DUPLICATE RESULTS 119
26. SEMIVOLATILE EMISSION SAMPLE SURROGATE RECOVERIES 121
27. DUPLICATE SEMIVOLATILE EMISSION SAMPLE RESULTS .... 122
28. FEED SOIL METAL RESULTS 125
29. FEED SOIL METAL MATRIX SPIKES RESULTS 128
30. TREATED SOIL METAL DUPLICATE RESULTS ........ 130
31. TREATED SOIL METAL MATRIX SPIKE RESULTS {TEST 1) .... 132
32. SCRUBBER WATER SAMPLE METAL DUPLICATE RESULTS ... 134
33. METALS EMISSIONS DUPLICATE SAMPLE RESULTS (TEST 2) . 140
34. RECOVERY CHECK RESULTS FOR MULTIPLE METALS TRAIN . 142
35. TCLP METALS DUPLICATE SAMPLE RESULTS 144
36. TCLP METALS SPIKED SAMPLE RESULTS 146
37. TCLP SEMIVOLATILE MATRIX SPIKE RESULTS (TEST 3) 149
38. TCLP SEMIVOLATILE SURROGATE SPIKE RECOVERY DATA ..150
39. PCDD/PCDF INTERNAL AND SURROGATE TREATED SOIL
RECOVERIES . . 153
40. PCDD/PCDF INTERNAL AND SURROGATE GAS RECOVERIES . 154
41. PCDD/PCDF DUPLICATE SOIL SAMPLE RESULTS 155
42. PCDD/PCDF DUPLICATE GAS SAMPLE RESULTS 157
XI
-------
ABBREVIATIONS AND SYMBOLS
A amps
ACLs Alternate Concentration Limits
BTEX Benzene, Toluene, Ethyl! benzene, and Xylene
CDIF Component Development and Integration Facility
CEM •- Continuous Emission Monitor
cf • cubic feet
CFR ' Code of Federal Regulations
DAS Data Acquisition System
DOE Department of Energy
DOT Department of Transportation
DRE Destruction and Removal Efficiency
dscf dry standard cubic feet
$ U.S. Dollar
EPA Environmental Protection Agency
°F degree Fahrenheit
ft feet
gal gallons
gpm gallons per minute
gr . . grains
hr hour
INEL Idaho National Engineering Laboratory
kg kilograms
XII
-------
ABBREVIATIONS AND SYMBOLS (CONTINUED)
kW
Ib
L
m
ORD
OSWIER
PCDD
PCDF
PCF
PIC
POHC
ppbv
ppm
ppt
psia
psig
%
RCRA
SAIC
SARA
SBC
scfm
SITE
TCDD
TCDF
TCLP
THC
kilowatts
pounds
liters
meters
milligrams
Office of Research and Development
Office of Solid Waste and Emergency Response
polychlorodibenzodioxin
polychiorodibenzofuran
Plasma Centrifugal Furnace
Product of Incomplete Combustion
Principal Organic Hazardous Constituent
parts per billion, by volume
parts per million
parts per trillion
pounds per square inch, absolute
pounds per square inch, gauge
percent
Resource Conservation and Recovery Act
Science Applications International Corporation
Superfund Amendment and Reauthorizatfon Act
Silver Bow Creek
•
standard cubic feet per minute
Superfund Innovative Technology Evaluation
2,3,7,8-tetrachlorodibenzodioxin
2,3,7,8-tetrachlorodibenzofuran
Toxicity Characteristic Leaching Procedure
Total Hydrocarbons
xiii
-------
ABBREVIATIONS AND SYMBOLS (CONTINUED)
Tentatively Identified Compounds
V volts :
VOST Volatile Organic Sampling Train
wk week
XIV
-------
SECTION 1
EXECUTIVE SUMMARY
INTRODUCTION
This report summarizes the activities and results of Demonstration Testing of the
Plasma Centrifugal Furnace (PCF) technology developed by Retech-, Incorporated
(Retech). The study was conducted at the U.S. Department of Energy's (DOE's)
Component Development and Integration Facility (CDIF) in Butte, Montana under the-
Superfund Innovative Technology Evaluation (SITE) Program developed by EPA.
The Retech technology, a Plasma Centrifugal Furnace, is a thermal process which uses
the heat generated from a plasma torch to decontaminate metal and organic
contaminated waste. This is accomplished by melting metal bearing solids and, in the
process, thermally destroying organic contaminants. The molten soil forms a hart.
glass-like nonleachable mass on cooling. The waste feed used in the Demonstration
Tests was comprised of heavy metal-bearing soil from the Silver Bow Creek Superfund
Site mixed with 10% by weight No. 2 diesel oil. The mixture was spiked to provide
28,000 ppm of zinc oxide and 1,000 ppm of hexachlorobenzene. In Addition to
complete monitoring of the system, sampling of all input and output streams was
performed during each of three Demonstration Tests. !
-------
The goal of the Demonstration Tests .was to obtain specific operating, design,
analytical, and cost information to evaluate the performance of the pilot-scale Plasma
Centrifugal Furnace (PCF-6) under actual operating conditions. Under the SITE
Program, the feasibility of utilizing the PCF technology as a viable hazardous waste
treatment system at other sites throughout the country was also studied. To this csrtd,
the specific critical test objectives were:
• to characterize the residues produced at optimum operation including
Destruction and Removal Efficiency (ORE), fate and transport of metals, and
residue quality;
t
• to identify pre- and post-feed waste treatment requirements.
• to evaluate the ability of the Plasma Centrifugal Furnace to effectively vitrify
inorganic and metal constituents within a soil into a monolithic nonleachable
mass; and .
•• to determine if the furnace can meet 99.99% ORE for target analytes in a soil
contaminated with up to 10% organics.
The SITE Demonstration Tests at the CDIF Were conducted between July 22 and 26,
1991. For a one-week period during the Demonstration Tests, EPA SITE Program
staff along with their evaluation contractor, Science Applications International
Corporation (SAIC), were present to observe and record data on the operation of the
technology and to perform sampling and analytical work. Quality assurance/quality
control (QA/QG) audit teams from S-Cubed, an EPA Risk Reduction Engineering
Laboratory (RREL) contractor, validated the test protocols in both the on-site tasks and
in the main laboratory. .
-------
CONCLUSIONS AND RESULTS
Presented below is a summary of the conclusions and results relating to the defined
objectives of the test program.
• The Toxicity Characteristic Leaching Procedure (TCLP) was performed on both
the feed soil and the treated slag. With regard to the metals, the feed soil only
exhibited significant leachability characteristics for calcium (with an estimated
mean leachate concentration of 175 mg/L and a 95% confidence limit of 162
to 188 mg/L ) and the spiked zinc (with an estimated mean of 982 mg/L and
a 95% confidence limit of 948 to 1,017 mg/L). Sodium was also present in
the leachate at 1,475 mg/L {95% confidence interval of 1,100 to 1,850 mg/L),
but was not selected as a tracer compound since it is a weakly dissociable
metal and, therefore, behaves differently from typical metals regulated by TCLP
. tests. The treated soil does not show strong leachability for any metals except
sodium which leached at approximately the same level as in the feed soil. Both
tracer metals, calcium and zinc, showed significant .reductions in leaching
properties as a result of treatment*
• The only organic compounds that were found to be leachable from the feed soil
were 2-methy(naphthalene and naphthalene: Although the feed was spiked
with high levels of hexachlorobenzene (1,000 ppm), it did not leach 'from the
feed soil. No organic compounds were found to leach from the treated slag.
• The ORE is based on the concentration of the target analyte in the feed soil and
the amount captured in the stack gas. For the Demonstration Tests, the 95%
confidence interval for the estimated mean of the hexachlorobenzene spiked i
into .the feed soil was 864 to 1,080 ppm. Hexachlorobenzene was riot ;
detected in the stack gas during any of the three tests. Therefore, all DREs
determined were based on the detection limit from each of the tests. For the
-------
Principal Organic Hazardous Constituent (POHC), hexachloroberizene, the
average DRE values ranged from ^99.9968 to > 99.9999% for all the
Demonstration Tests.
2-Methylnaphthalene was found in the feed soil with an estimated mean
concentration between 390 ppm to 526 ppm. Again, none of this compound
i
was detected in the stack gas, so the DRE range for all three Demonstration
Tests, based on detection limit from each of the tests was > 99.9939 to
> 99.9996.
A group of volatile compounds, total xylenes, was found in the feed soil at an
estimated mean concentration between 128 and 139 ppm. Over the course of
all the tests, this led to an average DRE range of > 99.9929 to > 99.9934%.
Because the collection volume of gas analyzed for volatile compounds remained
i i
•constant for each test, DREs calpulated for xylenes are reported as an average
for the entire demonstration. : .
The components of the PCF can be broken down'into two main categories: the
thermal treatment section and the exhaust gas treatment system. The furnace
unit demonstrated that it was entirely capable of processing the waste feed,
however, the gas treatment system did not perform up to expectations. The
air. pollution control device allowed an average of 0.374' grains/dscf of
particulates to be emitted to" the atmosphere throughout the three tests. This
exceeded the RCRA regulatory limit of 0.08 grains/dscf. Only very small
amounts of particulate matter, organic compounds, or inorganic compounds
were found in the scrubber sump at the conclusion of the tests, indicating poor
gas treatment efficiency.
-------
HCI emissions were very low fdr all three tests, ranging from 0.007 to 0.0017
Ibs/hr. Because of the low chlorine input, the regulatory requirement of less
than 4 Ibs/hr was met.
NOX emissions were high, averaging approximately 5,000 ppm (uncorrectecl to
• 7% oxygen) during the Demonstration Tests. Because of the low flowrates of
the stack gas, the emission rates averaged approximately 2.5 Ibs/hr.
Small quantities of volatile and semivolatile organic compounds were formed
as products of incomplete combustion (PICs) in the plasma furnace. The
volatile compound found most abundantly in the stack gas was benzene at
approximately 19 ppbv. Benzene and substituted benzenes are prevalent in
many forms throughout the feed diesel oil and hence benzene, is a readily
formed PIC. The most dominant semivolatile organic compound released in the
stack gas'was benzoic acid.
The entire system is a high maintenance item. During the course of the
Demonstration Tests, the exhaust gas blower failed twice (because of the high
paniculate loading in the flue gas),.the torch developed a deionized water leak,
and numerous preventive maintenance activities took place. The on-line factor
for the process during the Demonstration Tests was 70%, but a more realistic
on-line factor could be considered to be approximately 60%. . -
The configuration of the furnace for the Demonstration Tests allowed the
treatment of soils contaminated with heavy metals and hazardous organic
compounds with no free liquid in the soil. Based on observations during the
Demonstration Tests, it is anticipated that, with a different feeder, the Retech
process could treat a wide variety of organic and inorganic wastes in either a
solid or liquid matrix. However, judicious selection of an effective air pollution
-------
control system downstream of the furnace is necessary before remediation can
proceed.
• Successful operation of the PCF is limited by several logistical considerations.
The furnace used during the Demonstration Tests must be erected in a climate-
controlled environment with access to a 3-phase power supply of at least 480
volts and 1,600 amps! Plant cooling water with adequate heat rejection
{cooling tower) is necessary to supply all cooling circuits on the furnace.
• Several cost scenarios can vary the unit cost of operation for the furnace. The
cost of operation is strongly dependent on two factors: the on-line factor and
the feed rate. The present configuration of the feeder, furnace, and slag
collector allows an average feed rate of 120 Ibs/hr. However, feed rates of
500 and 1,000 Ibs/hr could be achieved with a few minor furnace
' - modifications. For a feed rate of 500 Ibs/hr with an on-line factor of 70% is
considered then it is estimated that the cost per ton for this technology is
" $1,816/ton of contaminated waste. For a feed rate of 2,200 Ibs/hr with the
same on-line factor the cost would be $757/ton.
-------
SECTION 2
INTRODUCTION
SITE PROGRAM OBJECTIVES
In response to the Superfund Amendments and Reauthorization Act (SARA) of 1986,
the U.S. Environmental Protection Agency (USEPA) established a formal program
called the Superfund Innovative Technology Evaluation (SITE) Program. The SITE
Program was established to accelerate the development, demonstration, and
implementation of innovative technologies at hazardous waste sites across the
country. There are four parts to the SITE Program: '
1. To identify and, where possible, remove impediments to the development and
commercial use of alternative technologies.
2. To conduct a demonstration program of the more promising innovative
technologies to establish reliable performance and cost information for site
characterization and cleanup decision-making.
3. To develop procedures and policies that encourage selection of available
alternative treatment remedies at Superfund sites. .
4. To structure a development program that nurtures emerging technologies.
-------
The objective of the first part of the program is to identify and evaluate these
impediments and remove them or design, methods to promote expanded use of
alternative technologies. The demonstration portion of the SITE Program is a
significant ongoing effort involving the Office of Research and Development (ORD),
the Office of Solid Waste and Emergency Response (OSWER), USEPA Regions, and
the private sector. The demonstrations willj provide Superfund decision-makers with
the information necessary to evaluate the use of these technologies in future cleanup
actions. The third part of the SITE Program focuses on establishing methods for
selecting treatment technologies for Superfund sites from the expanding range of
available remedies, including these innovative technologies. Finally, the SITE Program
provides a means of assisting in the development of emerging technologies towards
a mutual goal.
PROJECT BACKGROUND
A demonstration of the Retech Plasma Centrifugal Furnace (PCF) technology has been
performed under the SITE Program. This system used an .innovative thermal
technology to treat soils and debris contaminated with hazardous organic chemicals,
inorganic chemicals, and heavy metals. The process claims to vitrify metal-bearing
solids and inorganic material into a monolithic nonieachable phase, and thermally
destroy organic chemicals. .
The study was conducted at the DOE's CDlF located in Butte, Montana, operated for
the DOE by MSE, Inc. The test material 'used was a blend of soil' [mine tailings
obtained from the nearby Silver Bow Creek (SBC) Superfund Site] and 10% by weight
No. 2 diesel oil. The mixture was spiked with high concentrations of zinc oxide and
hexachlorobenzene. The combination of the; soil and oil provided a test material which-
was high in concentrations of both heavy metals and organic material. The zinc oxide
8
-------
and the* hexachlorobenzene were spiked to ensure traceable compounds throughout
the testing period.
This technology is significantly different from conventional incineration technologies
in that the temperatures in the treatment zone are,much higher and the gas flows are
«.
much lower than those typically encountered. Additionally, the treated soil is vitrified
into a glass-like mass. The tests obtained process data on the system performance
and the fate of metals and organics in the system. Removal or stabilization of
inorganic and metal contaminants was assessed by pre- and post-test sampling and
analysis. The ORE of this thermal technology was also determined by pre- and post-
test sampling and analysis for organic materials. The results of this testing are
presented in this report.
-------
SECTION 3
PROCESS DESCRIPTION
TEST SOIL
Soil collected from the Silver Bow Creek (SBC) Superfund Site in Butte, Montana was
placed in 11 32-gallon plastic containers with lids. The soil from the SBC site is
categorized as a heavy metal-bearing soil and was previously characterized as part of
the Streambank Tailings and Revegetation Study (STARS) conducted by the EPA in
1988 [1]. Zinc oxide and hexachlorobenzene were spiked into the dry SBC soil.
Spiked SBC soil and 10% by weight No. 2;diesel oil were combined together in a
cement mixer to provide a mixture containing 28,000 ppm zinc oxide and 1,000 ppm
hexachlorobenzene. This mixture was homogenized by the rotating action of the
mixer. To permit ease of handling and loading into the Retech Plasma Centrifugal
Furnace, the blended mixture was poured from the mixer into 5-gallon metal
containers and sealed for transport and storage.
Based on previous analyses, the teachability of the metals in the SBC soil was very
low overall. Therefore, it was determined to spike the feed soil with a nonhazardous
metal, zinc, to ensure an initial metal concentration in the feed high enough to allow
an adequate evaluation of the effectiveness of treatment by the PCF in .vitrifying
inorganic constituents into a nonleachable mass.
10
-------
Diesel oil was mixed with the feed soil to show that the process could treat wastes
contaminated with high levels of organics. Ten percent (10%) was the maximum
level of liquid combustibles that could be fed to the furnace equipped with the type
of feeder present at the test site; levels of dieel oil greater than 10% ignite in the
feeder because of the heat from the process. It is possible to treat contaminated
wastes with a higher percentage of organics using a different feeder configuration
assuming that the gas treatment system is sized correctly. Although the composition
of the diesel oil was estimated prior to the Demonstration Tests, the soil/diesel oil feed
mixture was left in the containers for an extended period of time after mixing activities
and prior to the analysis conducted as part of the Demonstration Tests. Therefore,
at the time of the tests, it was not certain that the concentrations of organic
components in the diesel oil were suitable to evaluate the ability of the furnace to
treat wastes contaminated with up to 10% organics. Hexachlorobenzene was spiked
into the feed to ensure a traceable organic compound for evaluating DRE.
TEST LOCATION
The Demonstration Tests were conducted using actual hazardous waste at the DOE
CDIF in Butte, Montana. The use of this facility for the Demonstration Tests was the
result of an interagency agreement between the EPA and the DOE. The CDIF is an
engineering-scale development test facility, operated for the DOE by MSE, Inc. The
CDIF is a major DOE test facility and part of the Idaho National Engineering Laboratory
(INEL). '
Retech's equipment was erected near the southwest corner of a 100-foot by 95-foot
Component Test Building, which has a ten-ton bridge crane and a roll-up access door
12 feet wide by 20 feet high. The building is supplied with 480-volt power and is
served by a recirculating cooling tower water system with heat rejection by a cooling
tower. Macadam roads provide good access to the building.
11
-------
DETAILED PROCESS DESCRIPTION
The Retech Plasma Centrifugal Furnace is a remedial action process for -soils
contaminated with hazardous chemical wastes and/or heavy metals. The PCF is a
thermal treatment process designed to convert contaminated soil into a chemically
inert and stable glass and crystalline product. Figure 1 is a schematic of the pilot-
scale unit utilized in the SITE Demonstration Tests. This unit is designated as a PCF-
6, according to the six-foot diameter of its primary chamber.
The entire system is comprised of a thermal treatment system and an exhaust gas
treatment system. The thermal treatment system consists of the feeder, the primary
chamber, the torch, the afterburner, the secondary chamber, and the collection
chamber. The exhaust gas treatment system consists of a quench tank, a jet
scrubber, a packed-bed scrubber, a demister, and a stack blower.
The Feeder
Hazardous waste is initially loaded manually into a screw feeder from sealed 5-gallon
containers. The feeder unit is 3.5 feet high ;x 2.5 feet wide x 10 feet long and can
hold approximately 120 pounds of feed. Tne outer stationary cylinder of the feeder
is sealed at both ends. The fill end has a hinged door that opens to allow the inner
cylinder to be loaded. The inner cylinder has welded internal screw flights that push
the dirt charge toward the discharge chute as it is rotated by a hydraulic drive. After
the feeder is loaded, it is sealed, and the waste is fed semi-continuously into the
furnace through the discharge chute. An elbow in the discharge chute limits the size
of waste particles fed into the furnace to be less than four inches in diameter. The
operator controls the rotation of the feeder.
12
-------
SUK FEEDER
OXVOEN LANCE
> PtASMA TORCH
VENT TO
ATMOSPHERE
EXHAUST OA8 TREATMENT SYSTEM
Figure 1. Schematic of Plasma Centrifugal Furnace System
-------
When the feeder is empty, it is purged of potentially toxic gases to the primary
chamber with an air purge. The feeder isolation valve is then closed, and the feeder
is refilled. During the Demonstration Tests, the system operated in the semi-batch
fashion, described here, at a rate of approximately 120 pounds per hour. Other
feeders (not evaluated during these Demonstration Tests) that allow an increase in the
feed rate and offer no limitation on the organic content of the waste feed can be used
by the furnace.
In order to access to the inside of the primary chamber for maintenance and repair
services, the feeder must be moved away from the primary chamber before opening
the Hd; it must be repositioned before the torch may be lit. During the Demonstration
Tests, the ten-ton bridge crane in the CDIF was used to move the feeder.
The Primary Chamber .
The waste material drops from the feeder into the primary chamber, which is a
1 f i
rotating tub with a central orifice at the bottom. Solid material is retained in the
primary chamber by centrifugal force. The primary chamber walls have an inner shell
and water jacket welded between. Fifteen gallons per minute of cooling water at 30
psig circulate between the shell and the jacket. At the copper throat, an area of high
heat flux, the cooling water flow area is reduced to increase the velocity of the
cooling water to 40 ft/sec, increasing the cooling in this area.
Ports are located in the head of the lid for: the plasma torch, an off-axis feeder port,
an oxygen lanqe, and four view ports. The entire lid subassembly is attached to a
hinged structure that allows the lid to be hydrauiically tilted through a 45° angle. A
mechanical safety link provides a backup to the lid lift hydraulic cylinder locks.
14
-------
The cavity below the primary chamber is maintained slightly positive with respect to
furnace pressure so that any leakage at the furnace lip seal at the top of the spinning
furnace wall will be in toward the hot furnace. This leakage will cool the furnace lip
seal and help prevent contamination of the drive cavity below.
The primary chamber can hold approximately 1,800 pounds of contaminated waste;
however, the size of the collection chamber limits the amount of waste that can be
in the primary chamber to 1,000 pounds. The primary chamber has a diameter of 6
feet and a height of approximately 3.5 feet.
The Plasma Torch
i
The Retech PCF makes use of a transferred plasma torch. The plasma torch uses
electrical discharges to add energy to plasma torch gases in order to increase the gas
temperature beyond .that normally attainable by chemical reaction. The plasma torch
produces a transferred arc that directly contacts a conducting portion of the primary
chamber {either the copper throat or steel "doughnuts") located at the bottom of the
spinning primary chamber. The heat generated by the plasma torch brings the waste
material to temperatures sufficient to melt soil. The melting point for typical soil is
on the order of 3,000°F. The waste is melted by this extreme heat, incorporating any
inorganic and metal components into a stable material. Organic components are
volatilized by the heat of the plasma and oxidized by the air used as the plasma gas.
Oxygen may also be added from an oxygen lance in the primary chamber to enhance
combustion of organics.
The plasma torch used in the PCF-6 is Retech's Model RP-250T plasma torch,
developed for melting superalloys in the metallurgical industry. For this application,
the torch, rated at 500 kW, used approximately 20 scfm of air as plasma gas. The
torch runs on direct current (DC) provided by a power supply that uses 3-phase input.
15
-------
Figure 2 shows a simplified presentation of the transferred plasma torch. Air is
injected tangentially at an intermediate axial position inside the torch tube. One of the
DC electric arc termination points is higher up inside the tube on an electrode. The
arc travels out the end of the tube and terminates on the rotating copper throat (or
i
steel doughnuts) below. The electrode and nozzle are cooled by a high velocity flow
of distilled water. The torch is mounted on a spherical ball swivel joint that has x-,
y-, and z-axis hydraulic positioning capability. The torch subassemblies are mounted
on the primary chamber lid.
The Afterburner
The gases evolved from the melting of the feed are drawn through the copper throat
of the primary chamber and pass through a natural gas afterburner located in the
secondary chamber, just downstream of the copper throat. The afterburner is utilized
to combust any products of incomplete combustion (PICs) by providing additional heat
input {beyond that supplied by the plasma torch) through another ignition source. It
is sized to provide 200,000 Btu/hr and operates on a natural gas flame. The organics
that are volatilized and oxidized are then drawn off to the exhaust gas treatment
system.
Secondary Chamber
A camera port in the secondary chamber allows observation of the gases and slag
exiting the throat. If needed, oxygen may be added from oxygen jets located in the
secondary chamber to enhance combustion of organics. A sleeve extending from the
•copper throat of the primary chamber, past the afterburner, and down to the end of
the baffles in the secondary chamber, reduces the volume of gas required to be
16
-------
ARC
TERMINATION
NOZZLE
WATER-COOLED
COPPER ELECTRODE
TANGENTIAL QAS
INJECTION (USUALLY AIR)
ARC TERMINATION
6LA® BATH
SPINNING
REACTOR
WELL
6LAQ AND OAS
REMOVAL
Figure 2. Retech's Centrifugal Furnace with Transferred Plasma Arc
-------
heated in the secondary chamber by the afterburner, thus ensuring enough heat is
available to complete the combustion of volatilized organics.
i i
The secondary chamber walls have three inches of refractory lining to abate heat loss
and protect the steel walls. These walls also form a jacketed vessel ,with 25 gpm
cooling water circulating between them to maintain a safe operating temperature.
Collection Chamber
The molten mass falls from the secondary chamber into a heavy pig mold located in
the collection chamber. The collection chamber is a water-cooled, jacketed chamber
that is bolted to the bottom of the secondary chamber. It houses a pig mold that can
hold approximately 1,000 pounds of melted solids.
The collection chamber is cylindrically shaped on a horizontal axis. One end is closed
off with a bolt-on, water-cooled blind flange. The other end has a water-cooled,
hinged door equipped with a viewport. Pig molds may be loaded and unloaded
through this door. Ten gpm of plant water at 30 psig cool this chamber subassembly.
The pig mold is made of 3/4-inch thick steel plate so it can withstand the sustained
he'at load of the cooling slag and the shock of the cascading charge. The mold has
a rectangular plan view and both elevation views reveal a draft angle to aid in removal
of the casting from the pig mold. The pig mold is on a removable tray, somewhat
simplifying chamber decontamination.
18
-------
Exhaust Gas Treatment System
Figure 2 shows the exhaust gas treatment system designed for the Demonstration
Tests. The gas stream is initially cooled in a quench tank before going to a jet
scrubber that is designed to remove particulates. From the jet scrubber, the gas
passes to a packed-bed scrubber to remove additional acid gases that may be present.
A demister then removes moisture droplets entrained in the flow.
A mildly caustic solution is used in the quench tank, jet scrubber, and packed-bed to
remove acid gases as well as particulates. The water passing through the exhaust gas
treatment components exits from the bottom of each unit, back into the scrubber
sump. The caustic reservoir that supplies the solution to the gas treatment units is
maintained at a pH of 8.5. This is achieved by means of a 0.1 gph positive
displacement pump that feeds the reservoir with sodium hydroxide, based on a signal
from a pH sensor located at the discharge j>ort of the scrubber pumps.
The scrubber-sump is equipped with a 50-ton chiller to cool the scrubber water
circulating through the exhaust gas treatment equipment, s'o that all the moisture can
be removed from the-exhaust gases. The chiller utilizes an external heat exchange
system. The chiller coolant is previously cooled by means of an additional internal
heat exchanger located within the chiller. The scrubber water is continuously cooled
and recirculated through the exhaust gas treatment system.
i
Flowmeters are installed to ensure that correct flow rates of caustic solution are being
applied to each of the gas treatment units. Flowrates are adjusted i by means of
metering valves at the inlet to each of the treatment units. \
During normal operation, there is a negative pressure on the furnace due to the effect
of the stack blower. This assures that, if there is a leak in the system, any leak would
19
-------
be drawn into the furnace and not out into-the operator area. All connections are
water-cooled and O-ring sealed. The system is leak-checked before each test.
Water Cooling Systems
All parts of the system exposed to high temperature are water cooled. There were
three cooling water systems involved In the furnace components for the
Demonstration Tests: a closed-circuit deionized water (D.I.) system; a closed-circuit
water with rust inhibitor system; and a plant cooling water system.
The closed D.I. system cools the plasma torch, the ram, the electrode and the nozzle.
The D.I. system pump delivers a total of 60 gpm at 80 psig. A heat exchanger
transfers the heat gained by the above circuits to the plant cooling water system.
* ^
The closed water with rust inhibitor system has just one circuit-the primary chamber
cooling circuit, flowing at'100 gpm arid 50 psig (at the pump discharge). A heat
exchanger is sized to pass a maximum of 300,000 BTU/hr from the furnace cooling
circuit to the plant water system. -
The plant cooling water system must not only cool the two cooling systems described
above (60 gpm and 100.gpm, respectively), but must also cool the furnace lid,
primary chamber, drive lid, secondary chamber, collection chamber, chiller, and
hydraulic unit.
20
-------
The Control System
The Retech control system is designed to be fail-safe. Numerous interlocks and alarm
circuits have been incorporated to reduce the risk of release of toxic material or
equipment damage due to operator error or process failure.
I
The torch cannot be lit unless all of the console interlocks are satisfied. These
interlocks include the torch gas pressure, the cooling water systems, the hydraulic
system, the exhaust gas treatment system, the stack blower, the afterburner, and the
chiller. The furnace is interlocked to shutdown if any of these systems or equipment
fail. If the emergency stop is activated, only necessary support systems are left
operating. ;
The plasma torch gas controls include a regulator, solenoid actuated on/off valves, gas
rotameYters, needle valves, and check valves. A pressure gauge and pressure switch
are also mounted in the line on the way to the torch. Before torch start-up, the
* t
pressure regulator is set and the gas flow rate established. The operator controls the
on/off valve from the control console. Power to the torch is automatically cut if the
gas pressure falls below a predetermined level.
Power to the torch is controlled from the control console. Current can be varied by
the operator with a twist of a dial. Cooling water to the torch is turned on manually
at a supply manifold. Switches in the supply lines provide an interlock to cut power
to the torch either at start-up or during operation if the water flows fall below preset
minimums. A starter unit, used to initiate an arc inside the torch at start-up, is
energized by the operator at the control console and de-energized by a timer in the
controller. Torch position is controlled by the operator at the control console with a
joy-stick. Position sensor outputs are used to limit deflection to safe locations.
21
-------
I * i
The centrifugal speed is monitored and controlled at the console. A magnetic pickup
uses the drive gear to generate the signal. Flow switches in the furnace cooling water
return manifolds prevent the torch from operating if the furnace cooling system fails.-
The lid cannot be lowered hydraulically without a safety link being disconnected, thus
protecting the equipment.
The output from the continuous emissions monitors are displayed at the console. An
interlock is included to assure that the stack blower is turned on before the furnace
is started. A pressure differential switch exists across the blower which shuts the
*
system down in the event of blower failure.
Level controllers are equipped on the caustic makeup tank and reservoir for the water
pump. If the level switches, in either case, are tripped on the low condition, an alarm
lights an indicator on the control console.
To ensure complete combustion, the afterburner is interlocked to the feeder. If the
•afterburner is lost, the feeder is automatically stopped. The chiller is interlocked to
• the torch, the afterburner, and the feeder.
22
-------
SECTION 4
•
FIELD OPERATIONS DOCUMENTATION
This section provides an overview of the field operations including a summary of the
tests, a detailed description of typical operations, and an operational log of the
Demonstration Test activities.
TEST SUMMARY ° - .
Demonstration testing of the Retech Plasma Centrifugal Furnace took place the we?ek
of July 22, 1991. The goal of this demonstration was to determine if the unit could
effectively thermally treat soil contaminated with metals and- up to 10% organic
material and create a nonleachable matrix. The testing objectives are defined in
Section 1 of this report and more specifically in the "Demonstration Plan for the
Plasma Centrifugal Furnace Technology" [2].
For the demonstration, three test runs were performed using Silver Bow Creek mine
tailings mixed with 10% by weight No. 2 diesel oil, and spiked to provide 28,000 ppm
zinc oxide and 1,000 ppm hexachlorqbenzene in the soil/oil mixture. The
Demonstration Tests took place as scheduled on July 22, 24, and 26, 1991. It was
anticipated that 960 pounds of the spiked soil would be treated for each test (2,800
23
-------
pounds total). Instead, 1,440 pounds of the spiked soil were fed into the furnace and
1,137 pounds of treated material were .generated during the three Demonstrate
Tests Additional details regarding this discrepancy are provided later in this section
and in Section 6, "PERFORMANCE AND DATA EVALUATION." For each
Demonstration Test, samples of the feed soil, the treated soil, the scrubber liquor, and
the stack emissions were collected to evaluate the performance of the technology.
Samples were collected and analyzed in accordance with the Demonstration Test Plan
with only minor changes in some of the sampling and analytical methods. These
changes are detailed in Section 8.
The unit was installed into three levels of the MSE Component Test Building. The
uppermost level of the system was located at ground level. This level housed the
primary chamber and the feed screw, which extended to an elevation of approximately
17 feet. The feed screw was located on top of ithe unit and a platform with stairs
allowed easy access to facilitate feeder loading. Twelve feet directly below, on the
second level, was the secondary chamber,-which received.the soil and combustion
gases for supplementary treatment with the assistance of an afterburner. The Sower
level (13, feet below the second level) contained the air pollution control system,
• designed to ensure that the treatment off-gas emitted to the atmosphere remained
within permissible levels. The treated gas was directed back up to ground level,
outside the building. Here, the exhaust gas ductwork was constructed to facilitate
gas sampling. A vacuum blower drew the gas through the air pollution control system
and maintained negative pressure in the furnace. After passing through the sampling
configuration, exhaust gas exited at the top of the building through the stack. The
entire process was controlled on the first level at a central control panel. Here the
torch operator bontrolled the feed screw and the position of-the torch. To assist in
torch control, cameras, which gave a visual indication of the torch position, were
. installed in the primary chamber. All process parameters were monitored by an
automated data acquisition system (DAS) which collected information every 30
24
-------
-seconds. Auxiliary equipment was monitored by MSE personnel performing visual
inspections and recording data every 30 minutes.
r
Prior to the Demonstration Tests, MSE performed treatment tests using
uncontaminated soil to familiarize themselves with equipment operation. During these
shakedown tests, modifications were made to the original system design. These
modifications included the installation of the afterburner in the throat of the furnace
between the primary and secondary chambers, the installation of a chiller on the gas
treatment system to compensate for this additional heat input, and the elimination of
the surge tank. The detailed discussion of the process description is included in
Section 3 of this report.
DESCRIPTION OF OPERATIONS
Typically, operational procedures are conducted under standard process conditions.
These operating conditions, coinciding with those used during the Demonstration
Tests, are summarized in Table .1. To facilitate the treatment of different types of
waste, the system has the potential for alternate conditions or configurations for
parameters such as the feed rate, the feed material, and the air pollution control
system; generally, the conditions remain the same for all operations. The operational
procedures for the present configuration are described below. >
As part of preparation for operation, the non-current-carrying end of the electrode is
coated with Dow Corning vacuum grease and the current-carrying electrode thread is
treated with an application of Copper Cote®. These materials are utilized as an aid to
light-off procedures. Before initiating operations, the lid is opened to inspect the torch
and the gas lines leading into the furnace. A generalinspection of the entire system
also takes place prior to light-off. Approximately 100 pounds of mild steel, in the
form of 1 -inch thick "doughnuts," are placed inside the chamber, encircling upper end
25 :
-------
Table 1. Demonstration Test Operating Conditions
Operating'Range
Variable. Limited by 10% organics or no
free liquid (for this feeder.)
Feed Composition
Mild Steel 'Doughnut* Mass
Torch Gas (Air) Flowrate
rorch Ga* (Air) Temperature
Oxygen Lance Rowrate
25 (during feeding)
Auxiliary Torch Gas (e.g.. Argon. Helium) Rowrate
Afterburner Gas (Natural Gas) Rowrate
Furnace Well Rotation Rate During Treatment
Furnace Well Rotation Rate During Pour
Primary Chamber Temperature
Afterburner Temperature
Scrubber Liquor Generated
Off-Gas Flowrate
of the throat, This provides a conductive surface on which the torch arc can be
initiated and a lip that prevents untreated material from, spilling into the secondary
chamber. Typically, a 4- to 8-inch layer of previously treated material coats the inside
of the primary chamber.
After a final check of process -equipment, the torch arc is struck! The furnace
chamber temperature initially rises rather rapidly, and off-gas temperatures climb
steadily. Before any material can be fed into the furnace/operating procedures require
that the primary chamber temperature must be at least 2,100°F, and -the afterburner
temperature 1,800°F. Three to five hours of operating time are necessary to achieve
these temperatures, depending on the amount of ;residual heat in the primary chamber.
I !
26
-------
Feed material is generally placed into 5-gallon pails to facilitate ease and convenience
during feeding. The pails are placed on the feeder platform prior to startup.
When the furnace is at or near operating conditions, the feed screw is manually
loaded. Feeding is performed by opening the access door to the feed screw and
placing soil into the screw chamber. A specially-designed bucket holder facilitates
dumping the contents of the bucket and allows the material to pass down a shoot into
the feeder. As material enters the feeder, the screw is turned by hand to evenly
distribute the feed load along the screw. After an appropriate amount of feed
(approximately 120 pounds) has been loaded into the feeder, the access door to the
feeder is secured and purged with air of any residual contamination. The purged air
is bled to the primary chamber of the furnace. This eliminates potential for worker
exposure to toxic fumes when loading of the feeder is resumed. This feeding process
has potential for automation, but is presently manually operated. !
The feed screw is placed in motion,, allowing material to enter the furnace. It takes
approximately one hour to completely feed the 120 pounds, of material in the screw
to the furnace. When empty, the feeder is reloaded and additional material is fed to
the furnace as described above until a maximum of 1,000 pounds has been fed. This
corresponds to the capacity of the pig molds into which the molten soil is poured.
Process operations are continuously observed by operators and their supervisors.
Process measurements shown in Table 2 are recorded manually or electronically by
the DAS. Important process information is also calculated by this system. These
calculated values are presented in Table 3.
Once fed into the furnace, treatment of the waste material is initiated with the plasma
torch. The area closest to the copper throat and the mild steel doughnuts is the first
location to be preheated. After the melted material in this vicinity is heated to
conducting temperature, the torch is moved slowly to heat more of the bottom of the
27
-------
Table 2. Operating Parameters Monitored by the DAS
•lYTemperatures ..:
Argon Supply
Oxygen Supply
Service Air Supply
Furnace Off-Gas
Scrubber Outlet Gas
Stack Ga*
Primary Chamber Gas
Afterburner Gas
Secondary Chamber Gas
Melt
Stack Gas Composition '•' '
High CO
. Middle CO
Low CO
cy
SO,
NO
NO.
THC
•-Pressures
; Argon Supply
; Oxygen Supply
Service Air Supply
Stack Gas Flowmetar Differential
Stack Gas
Drive Chamber
• .'. • '' •'•. ' ...••.:••":••::;•••::• gj^SOWrataS
Argon Supply
Oxygen Supply
' Service Air Supply
; Afterburner Air
Afterburner Gas
i Miscellaneous
. Torch Current
Torch Voltages
Feeder Position
: ' Primary Chamber Rotation Spaed
Switched Input Sensing
primary chamber and eventually the sidewall. This is continued until the entire
contents of the primary chamber have been melted by the torch. Following this
preheat period, the screw feeder is rotated:to charge material at a uniform rate into
the furnace. The feeder can be recharged and feeding repeated as previously
described. After all the desired charge is melted, the natural gas afterburner (located
downstream of the primary chamber, in the^ secondary chamber) is extinguished, and
the furnace spin .rate is slowed to approximately 25 rpm allow the pool to move
inward and the melted soil to pour out ;of the bottom of the reactor, into the
secondary chamber. The molted mass falls from the secondary chamber, through the
28
-------
Table 3. Operating Parameters Calculated by the DAS
•'••.,•'.-. '••:- •.,.. '":.;-,-.. . Heat Losses . ': . „
Scrubber Cooling
Centrifuge Cooling
Torch Ball Cooling
Drive Chamber Cooling
X-Y Plate Cooling
Torch Cooling '
Demister Cooling
Primary Lid Cooling
Secondary Combustion Chamber Cooling
Side Ports Cooling '
Total
Argon Supply Mass
Service Air Supply Mass
Stack Gas Mass
rates ' - ' ',.;•'. • •'-•..'.. ';
Oxygen Supply Mass
Stack Gas Volume
Total Mass In - Total Mass Out
'..' ' .:•. . . ''•' '!'"""•;;:'• '. •: '' ''"'.-' • • 'Power • •• ' :,: ' •_ . • ..'. . ;.;:
Torch
collection chamber, and into a heavy pig mold. The pouring process takes five to ten
minutes for a 600-pound pig. The molten mass solidifies into a hard monolith in
approximately 12 hours and can be disposed of in an appropriate landfill or otherwise
utilized. The organics that are volatilized and oxidized are drawn off to the exhaust
gas treatment system for subsequent conditioning prior to discharge to the
atmosphere. ;
Routinely, the scrubber sump is drained, cleaned and charged with fresh water.
Recharging activities for the scrubber sump are also performed whenever the waste
feed is altered. New filters are installed into the scrubber recirculation lines as
required, depending on particulate accumulation.
29
-------
OPERATIONAL LOG FOR THE DEMONSTRATION TESTS
Presented below are the field notes for the Retech Demonstration Tests. The notes
briefly and chronologically summarize the operational events that took place during the
tests. ; ! . .
Test 1 ;
The Retech Demonstration Tests began on Monday, July 22, 1991. Activities for
Test 1 began at approximately 7:00 a.m. At 7:10 a.m., the pig mold, which would
hold the treated material, was vacuumed out to remove any residual soiL A new
torch electrode had been installed the previous evening and was coated with Dow
Corning vacuum grease and Copper Cote®. The primary chamber was prepared for
closure by coating the sealing surface with a new gasket. By 7:35 a.m., the furnace
lid was fastened in place. At 7:40 a.m., the feed screw was placed on the furnace
and secured into place. A process problem with a thermocouple and a plugged stack
flowrate annubar were noted at 8:35 a.m. .
After the process checkout had been completed at 9:05 a.m., the torch arc was
struck, and the furnace warm-up was initiated. More than an hour and a half later,
stack gas became visible from the stack. The gas appeared brown in color with a
yellowish tint. Continuous emission monitoring equipment indicated that the stack
gas exhibited high NOX values in the range of 7,000 to 10,000 ppm.
At 12:20 p.m., the open feed buckets were monitored with the organic vapor analyzer
(OVA) field instrument to estimate the amount of volatile organics being emitted to
the air around the feeder. The OVA did not detect any volatile organics in the air in
the vicinity of the feed. Loading of the feed screw began at 12:25 p.m. By 12:35
p.m., the feeder loading was complete with 120 pounds of contaminated soil.
30
-------
Because of Demonstration Test sampling activities, the time required to load the
feeder was slightly longer than that required for typical operations.
By 1:10 p.m., the action of the screw feeder to the furnace was initiated for the first
feed load. At this time, the primary reaction chamber temperature was 2,156°F,
while the afterburner baffle temperature was 1,855 °F. At 1:50 p.m., the sump pump
tripped and the system was checked to determine the cause-. The 1:50 p.m. trip of
the scrubber occurred because one of the scrubber sump pumps was being starved
of scrubber water. This caused the pump motor to overheat and trip the pump.
Three minutes later, the scrubber sump pump feed line was cleared and the furnace
was again activated. Feeding could not be resumed immediately because the primary
chamber temperature was too low. By 2:20 p.m., the feeding resumed, and 20
minutes later, re-loading of the feed screw was started. As soon as the loading of the
feeder was complete (at 2:49 p.m.)., the action of the. screw feeder was again-
initiated. The.second feed load was fed by 2:50 p.m. At 3:25 p.m., scrubber pump
#2 was taken off-line. At 3:30 p.m., the feed screw loading was started and by .3:39
p.m. feeding of third load began. -By 4:15 p.m., the feed screw loading was started,.
and the feeding of the fourth and final began at 4:25 p.m.
At 4:37 p.m. and 4:48 p.m., the afterburner tripped and was restarted.; At 5:03 p.m.
the afterburner tripped again. A low pressure differential (20" H2O) across the blower
was rioted at this time. Normal operation requires a pressure differential of 24" H20,
thus, the operators experienced difficulties maintaining negative pressure in the
primary chamber. The cause for the drop in pressure differential across the blower
was particulate buildup within the blower. The resulting low negative pressure in the
furnace was the cause of the afterburner failures. Operations were restarted, but
feeding .was not recommenced immediately due to inadequate temperatures in the
afterburner. By 5:51 p.m., sampling and feeding of the final load were resumed. It
was decided not to reload the feeder with any more soil as the problem with the
particulates would only be aggravated. Therefore, the test was terminated.
31 [
-------
At 6:14 p.m., the feeder was turned off thP afto K
:;
Test 2
.....
*•*«
"
*•
Table 4. Summary of Selected Process Parameters
ParameTor
T««t Soil Fad flb«)
roned Son Pournd flh.l
™^^^™™^™"«
Scrubbar Water Ganoratad (onll
Mild Steel Addod
32
-------
blower tripped the torch, and adequate furnace pressure could not be maintained. The
torch arc was reinitiated, and the test proceeded until a pour could be completed.
At 2:27 p.m., the afterburner was shut down, and the melt was poured. By 2:37
p.m., the system was shut down, secured, and Test 2 was complete.
IgsjJi
Demonstration Test 3 was conducted on July 26, 1991. The troublesome blower
was replaced by a blower with a larger motor. Lighting of the torch took place at
approximately 8:20 a.m. The furnace began to heat up at 8:23 a.m.
At 9:53 a.m., the furnace tripped. A leak, which caused distilled water to drip into
the primary chamber, had developed in the torch cooling system. The problem was
caused by a side arcing of the torch which burned a hole in the torch casing. The
furnace was opened and repairs were made by grinding clean the !torch tip and
welding the hole shut. "By 12:40 p.m., the repairs were complete. The torch was
again struck, and the furnace began to heat back up.
At 3:30 p.m., the furnace was near processing temperatures, and feeder loading
started.. By 3:35 p.m., feeder loading was completed. The feeding of the first load
to the furnace was started at 3:50 p.m. By 4:47 p.m., the loading of the feeder with
a second load had begun. At 4:54 p.m., the feeder was fully loaded. This was
followed immediately by the feeding of the second load to the furnace. Approximately
one hour later, feeder loading of the third load was started and promptly completed.
By 6:00 p.m., the feeding of the third load had begun.
At 6:07 p.m. the afterburner/tripped, temperature in the top baffle dropped, and the
feeder was stopped. The feeder was restarted at 6:11 p.m. At 6:24 p.m., the
33
-------
afterburner tripped again, the furnace temperature dropped, and the feeder was
stopped. The afterburner was relit at 6:27 p.m., and the feeder was restarted at 6:30
p.m. .It was later determined that the afterburner trips had been caused by the
operator of the scrubber unit. During the course of the test, the two 10 micron filters
at the discharge of the sump pumps had become clogged. The operator changed out
each of the filters while the furnace was still operational. This caused a pressure drop
in the scrubber supply water which, in turn, tripped the afterburner.
The time was 6:59 p.m. when the feeder loading activities began for the fourth load
of feed. At 7:06 p.m., when the feeder was loaded, the feeding of the fourth load
started. Feeder loading began at 7:59 p.m. and was finished eight minutes later. The
feeding of a fifth load immediately followed. By 9:06 p.m., the feeding of the fifth
load was completed and the feeder was stopped. At this point, the torch operator
prepared the melt for pouring. At 9:23 p.m., the pouring of the melt was initiated.
By 9:26 p.m., the melt was poured, the system was powered down, and the test was
complete. i .
34
-------
SECTION 5
SAMPLING AND ANALYSIS PROGRAM
SAMPLING PROTOCOL
As pail of the SITE Program, a sampling strategy was designed and employed to
evaluate the performance of the Plasma Centrifugal Furnace technology developed by
.Retech, Inc. The sampling .protocol was developed based on the critical test
objectives stated in Section 1. -
Extensive monitoring of process equipment was also performed to collect data
required for analytical and economic calculations necessary to properly evaluate this
technology. These monitoring parameters are presented in Section 4 of this report.
Samples were collected in accordance with the "Demonstration Plan for Plasma
Centrifugal Furnace Technology." Minor changes to the original sampling plan were
made. These changes are discussed in detail in Section 8 of this report. Figure 3
presents a schematic of the sampling locations for the Demonstration Tests.
To evaluate the technology, three replicate tests were performed using similar process
conditions. Accordingly, the sampling plan for each test was similar. The
demonstration consisted of three tests to ensure that adequate information could be
35
-------
to
O)
Figure 3. Schematic of Sampling Locations for the Demonstration Tests
-------
gathered to properly evaluate the technology and limit analytical bias from one set of
data. The types of samples collected for each of the tests are presented in Table 5.
Originally, it was also planned to collect samples of the scrubber solids. Due to the
limited amount of mass trapped in the scrubber filters (less than 0.5% wt.), it was not
possible to obtain enough sample for the required analyses. Hence, the scrubber solid
and liquid samples were collected and analyzed together.
All liquid, solid, and gas samples were collected for analysis of the parameters listed
in Table 5. Sampling containers were obtained from l-Chern Research and were
cleaned to EPA protocol specifications. Miscellaneous sample containers, such as
petri dishes, XAD-2 resins, Tenax®, etc. were provided by the sampling subcontractor
and precleaned to specifications outlined in the Demonstration Plan. '•
Process measurements were made by MSE. The process was equipped with sensing
equipment fitted-with transmitters (i.e., thermocouples, flowmeters, etc.) to monitor
all important parameters. The data acquisition system (DAS) checked each monitoring^
point and logged information every 30 seconds. Table 2 presents the process
measurements made by MSE using the automated DAS. From these measurements,
the DAS calculated several operational parameters as shown in Table 3. Visual
monitors were provided in the control area to display these parameters to the
operators. Additional measurements of the auxiliary equipment were collected by
MSE at 30-minute intervals. These, measurements included:
* furnace temperatures and pressures;
• torch parameters (i.e., pressure and flow of torch gas, and torch power supply);
• deionized and centrifuge water temperatures, pressures, and flow rates; and
• system pressures for the lid, feeder,.exhaust, drive, and scrubber.
37
-------
Table 5. Samples Collected During Demonstration Tests
Sample Location*
—=3======
==3ssss==3ss=======
SI
S2
S4
S4
S4
Sample Name
Feed Soil
. Treated Soil
Scrubber Liquor
Before Test
Scrubber Liquor
After Test
Scrubber Discharge
Solids*
Sample Type
TEST 1 '-
=====
Discrete
Discrete
Discrete
Composite
Composite
Composite
Composite
Composite
Discrete
Discrete
Discrete
Discrete
Discrete
Grab
Grab
Grab
Grab '
Grab
Grab
Grab
Grab
Grab'
Grab
Grab
Grab
Volatile® .
Somivolatiles
Metals Scan
Dioxins/Furans
TCLP
Higher Heating Value
Chloride
Moisture (as received)
Semivolatiles
Metals Scan
Dioxin/Furans
TCLP
• Bulk Density
Volatiles
Semivolatiles
Metals Seen
Dioxins/Furans
Volatiles
: Semivolatiles
Metals Scan
Dioxins/Furans
Volatiles
Semivolatiles
Metals Scan
Dioxins/Furans
Number of ;
Samples."'-
P ;:
3
3
3
V
1
1
1
1
1
1"
1
1
1
1
1
1
1«
1
1
1
1C
1
1
1
1-
'""SO ;'i
1
3*
3»
r
1
1
1
1
1
H
1 .
1
r
1
i
1
r
i
i
t
r
(Contimued)
38
-------
Table 5; (Continued)
TEST 2
lie Location' ••
- Sample Name
Sample Type
Parameter ;:
TEST 1 (CONTINUED)
S5
Stack Gas
"
VOST
(SW-846 M0030)
Gas Canister
(EPA
Compendium
MTO-14)
EPA MM5
Multiple Metals
EPA MM5
EPA M5
CEMs
Volatiles
Samivolatiles
Metals Scan
Dioxins/Furans
HCI
Participates
O,, CO,, CO. NO». SO,. THC
Number of
"Samples •;
•;p ••'•
6
3*
1
1
1
1
-:-.-.:D '•'•
1
r 1
1
ST
S2
S4
_______
•
-
Treated Soil
Scrubber Liquor
Before Test
Discrete
Discrete
Discrete
Composite
Composite
Composite
Discrete
Discrete
Discrete
Discrete
Discrete
Grab
Grab
Grab
Grab
Volatiles
Semivolatiles
Metals Scan
Dioxins/Furans :
TCLP
Higher Heating Value
Semivolatiles
Metals Scan
Dioxin/Furans
TCLP
Bulk Density
Volatiles
Semivolatiles
Metals Scan
Dioxins/Furans
3
3
3
1°
1
1
1
1
1
1
1
1
1e
1
3" .
3b
"1 '
1
1
1
1
1
1
1e
(Continued)
39
-------
Table 5. (Continued)
. • Sample Location*
. Sample Name '
Sample Type'- ;;: ':
•..-.,• '. • : 'Parameter • ' •'[.
Number of
Samples ;
P
• -:D • ;:
TEST 2 (CONTINUED)
S3
S4
S4
S5
Scrubber Makeup
Scrubber Liquor
After Test
Scrubber Discharge
Solids*
Grab
Grab
Grab
Grab
Grab !
Grab
Grab
Grab
Grab :
Grab
Grab
Grab i •
VOST ;
(SW-486 MO030)
Gas Canister
(EPA Compendium
MTO-14)
EPA MEWS
Multiple Metals :
EPA MM5
EPA MS
OEMs i
Volatiles
Semivolatiles
Metals Sean
Dioxins/Furans
Volatiles
Semivolatiles
Metals Scan
Dioxins/Furans
Volatiles
Serqivolatilas
Metals Sean
Dioxins/Furans
Volatiles
Semtvoiatiles
Metals Scan
Dioxins/Furans '
HCI
Particulates
02. COZ, CO. NO.. S02. THC
1
1
1
r
1
1
1
1e
1
1
1
1*
S
3°
1
1
1
1
9
1
1
1
1*
1
i
1
r
1
1*
1
1
Continuous
(Continued)
40
-------
Table 5.' (Continued)
Sample Location*
Sample Name
S1
S2
S4
S4
S4
Feed Soil
Treated Soil
Scrubber Liquor
Before Test
Scrubber Liquor
After Test
Scrubber Discharge
Solids*
Sample Type ;;
Parameter ,
Number «f
Samples:
P
•O "f
TESTS
Discrete
Discrete
Discrete
Composite
Composite
Composite
Discrete
Discrete
Discrete
Discrete
Discrete
Grab
Grab
Grab
Grab
Grab
Grab
Grab
Grab
Grab
Grab
Volatiles
Semivolatiles
Metals Scan
Dioxins/Furans
TCLP
Higher Heating Value
Semivolatiles
Metals Scan
, Dioxin/Furans ;
TCLP
Bulk Density
: Volatiles
Semivolatiles
Metals. Scan
Dioxins/Furans
Volatiles
Semivolatiles
Volatiles
Semivolatiles
Metals Scan ,
Dioxins/Furans
3
3
3
1«
1
1
1
1
1
1
1
1
1
1
1e
1
1
1
1
1
r
3-
3"
1«
1
1
1
1
,
1
1
1
1';
1
t
1
1
1
r
, (Continued)
41
-------
Table 5. (Continued)
Gas Canister
(EPA
Compendium
MTO-14)
. SeopU locations keyed to Figure-4.
„„
r 35
In
Primary Sample
Duplicate Sample . !
r calculated parameters included: I
"" I '
plant water temperatures and flowrates; and !
off-gas heat removal data and chiller operating data.!
* starting the Demonstration Tests, extensive planning and evaluations were
cted to ensure that representative samples would be cof lected to achieve project
42
-------
objectives. Samples were collected at the locations noted in Figure 3. Below is a
discussion of the procedures used to collect samples at these locations.
Feed Soil
Samples of the feed soil were collected during each loading of the feed screw. These
samples were collected by obtaining a scoop of material from the top and middle of
each 5-gallon pail that was fed to the furnace. These samples were collected using.
a small (1-cup) stainless steel scoop and depositing the sample in a 4-L glass jar.
After each collection, the jar was sealed and agitated by hand to thoroughly mix its
contents.
At the end of the test, the material that had been collected in the jar was portioned
* . •
intp aliquots and stored in separate containers for each.selected analysis. 'Samples
for all analytical parameters were collected using this procedure with the exception
of the samples obtained for volatile organic analysis (VGA). Samples for VOA were.
collected at intervals specified.in the QAPP during feeder loading. These discrete
samples were immediately placed into 40-mL VOA vials and stored at 4°C.
Treated Soil
After each test, the treated soil was poured into a rectangular steel mold ("pig") and
allowed to cool. After the material had solidified and cooled to approximately 100"F,
sampling was conducted. These events took place the morning after treatment for
Tests 1 and 2 and two days later for Test 3. The treated soil from Test 3 was too hot
to sample the morning following treatment.
43
-------
Sampling of the treated soil was conducted using a drill equipped with a diamond-
tipped 2-inch coring bit. The drill was mounted on a specially designed stand which
spanned the pig and could be firmly attached in place. This provided a sturdy, stable
platform to operate the drill. Once mounted, the stand allowed the drill to operate
similar to a drill press. The drill bit was cooled by using a Milli-Q water system.
Cooling water was placed in a container, which could be pressurized by hand, and
piped to the drill bit. When drilling was initiated, the cooling water was allowed to
flow to the bit. The drill bit was lowered into place at selected locations within the
pig to collect sample cores. Each 4- to 5-inch core sample took approximately 5
minutes to obtain. When the bit was raised from the slag, the core sample remained
in the bit. There were, on occasion, some broken pieces- that remained in the drilled
1 i
hole. The broken fragments were removed from the hole using a stainless steel.
screwdriver. The core within the bit was dislodged by removing the bit from the drill,
sliding a length of pipe down the internal shaft of the bit, and ramming the core until
it fell from the bit. The entire core length (unbroken) depended on the depth of the
f - !
treated soil.in the pig. In most cases-, the core was recovered in-broken fragments.
" w • ; - '-, *
Pieces obtained from each core ranged in size from 1 to 2 inches."
Several cores were obtained from different locations within the treated soil pig for
each test. After collecting a sufficient amount of material from each test, the sample
cores were composited into a single sample for each test. Compositing was
performed by placing the core fragments between two cotton cloths. A hammer was
used to break the fragments into smaller pieces. After reducing the fragment size to
less than one-half inch, the pieces were thoroughly mixed and placed into sample
containers for analysis of each designated analytical parameter.
Samples of the Milli-Q water were also collected and shipped to the laboratory for
analysis to check for contamination and account for any potential loss of material from
the vitrified soil during the drilling process.
44
-------
Scrubber Liquor
Samples of the scrubber liquor were collected before and after each test. The
samples collected at the beginning of the test were obtained from a tap in the
scrubber recirculation lines. The scrubber was flushed clean at the beginning of each
test and recharged with fresh tap water. Before samples were collected, the
recirculation pumps were turned on to allow the water to thoroughly rnix. A bucket
was placed under the tap, and scrubber water was allowed to flow until 1 to 2 gallons
were collected and the sample tap flushed. This was performed to ensure that the
samples collected were representative of the actual contents of the scrubber and not
line contamination. After purging the line, the 1-L sample bottles were placed under
the tap and filled. Samples for volatile organic analysis were also collected using this
technique and .substituting the 1-L sample bottles with 40-mL VOA vials.
Samples of the scrubber liquor were also collected immediately following each test
run. When the system was secured, samples were obtained from the same tap as the
pre-test. These samples were collected using the same procedures described above.
Scrubber Caustic
As with the scrubber liquor, a representative sample of the scrubber caustic in the
sump was collected after Test 2. This sample was obtained from the discharge end
of the scrubber makeup pump. The tubing from the discharge end was removed from
its connection into the scrubber tank and placed into the sampling containers. The
pump was allowed to operate until the sample containers were filled.
45
-------
Stack Gas
The stack gas was sampled for a number of parameters. All gas samples were
collected downstream of the blower. The stack gas was collected from the exhaust
duct located on the outside of the MSE test building. Several sampling ports were
installed in the duct to accommodate the gas sampling trains. Each port was installed
so that the spacing between them was at least 8 pipe diameters to facilitate isokinetic
sampling when required. All piping in the exhaust duct was 3" sch 10 304L SS with
an ID of 3.26". The sampling ports themselves were 3" x 3" tees with threaded
adapters for the sampling probes. All stack gas probes (with the exception of the
standard pitot tube and thermocouple probe) were designed to allow only the nozzle
to protrude into the gas stream and thus prevent significant flow disturbances.
A port similar to the sample train ports, only with a 1" adaptor, was also installed to
facilitate continuous emission monitoring equipment. This port was located upstream
of all other ports to rnaintain sample integrity that may have been lost from in-leakage
due to the installation.and removal of other sample trains.
Volatiles .
The stack gas was sampled for volatile organic compounds (VOCs) using SW-846
Method 0030, volatile organic sampling train (VOST). This method is designed to
provide analytical information on volatile organics with boiling points less than 100°C.
Gas samples were collected on pairs of Tenax®-Tenax®/charcoal cartridges as
described in the method. VOST samples were collected for each Demonstration Test.
Tests 1 and 2 were conducted utilizing a single sample train to collect VOST samples
throughout each test. Test 3 employed both a primary train and a duplicate train
which operated during the same time frame as the primary train. To prevent overload
on the resin cartridges, varying volumes of samples were collected during each test.
Table 6 presents information on the samples collected for each test run.
46
-------
Table 6. VOST Sample Identification
Time
Sample ID#
Sample Description
Test 1 (07/22/91)
1400
1540
2530
1715
1735
1800
1825
1841
1841
1841
SAIC-0129
SAIC-0130
SAIC-0131
SAIC-0132
SAIC-1033
SAIC-0134
SAIC-0134
SA1C-0161
SAIC-0165
SAIC-0199
20-L Pair
20-L Pair
Field Blank
10-L Pair
10-L Pair
5-L Pair
54. Pair
34. Pair
Condaneate
Trip Blank
Test 2 (07/24/91)
1100
1116-1156
1210-1250
1302-1322
1336-1356
1429-1442
1443
SAIC-0255
SAIC-0256
SAIC-0257
SAIC-0258
SAIC-0259
SAIC-0260
'SAIC-0272
Field Blank
20-L Pair
204. Pair
104. Pair •
10-L Pair
6.754. Pair
Condencate
Test 3 (07/26/91)
2145
2145
1558-1638
1558-1638
1649-1729
1648-1728
1737-1757
1738-1758
1808-1828
1809-1829
1841-1851
1841-1851
2127-2130
2127-2130
1900
SAIC-0381
SAIC-0382
SAIC-0386
SAIC-0387
SAIC-0389
SAIC-0399
SAIC-0390
SAIC-0391
SA1C-0392
SAIC-0393
SAIC-0394
SAIC-0395
SAIC-0396
SAIC-0397
SAIC-0398
Primary Condencate
Duplicate Condeneate
Duplicate 20-L Pair
Primary 20-L Pair
Duplicate 20-L Pair
Primary 20-L Pair
Duplicate 10-L Pair
Primary 10-L Pair
Duplicate 10-L Pair
Primary 10-L Pair
Duplicate 5-L Pair
Duplicate 5-L Pair
Duplicate 1.694. Pair
Primary 1.69-L Pair
Field Blank
47
-------
In the event of breakthrough of the VOST cartridges or saturation of the GC detector,
gas canister samples were also collected as a backup for the VOST samples. The gas
canister samples were collected per EPA Compendium Method TO-14. As with the
VOST, Tests 1 and 2 were sampled with a single canister sampling train and Test 3
included a duplicate canister sampling train. Six-liter samples were collected in
evacuated stainless steel canisters.
Semivolatiles and Dioxins/Furans
Samples for semivolatiles and dioxins/furans were each collected from the stack using
the Modified Method 5 (MM5) sampling trains in accordance with the method
specified in SW-846 (Method 0010). The MM5 train is designed to sample gaseous
and paniculate pollutants with boiling points greater than 100°C. Samples are pulled
from the stack isokinetically and then passed through a filter and a porous polymeric
resin to trap the components of interest. .I
Samples were collected nominally at approximately 0.5 scfm until a total of 106 ft3
i
of sample were obtained. During sample retrieval in .the field, each component of the
sample train was removed and rinsed with methylene chloride. This rinse was
collected and sent to the laboratory for analysis. For the dioxin/furan MM5 sample
trains, toluene was used instead of methylene chloride. All other procedures took
place as described in the method.
r ' j '
One MM5 sample train was used to collect semivolatile stack samples for Tests 2 and
3. Test 1 semivolatile stack samples were collected in duplicate. One sample train
was used to collect dioxin/furan-stack samples in Tests 1 and 3, while Test 2
sampling was performed in duplicate.
48
-------
Metals
Samples of the stack gas for metals analysis were collected using the multiple metals
train. This method is similar the MM5 method sampling methodology and is described
in SW-846 draft Method 0012.
For the multiple metals method, nitric acid, peroxide, potassium permangenate and
sulfuric acid are utilized in the impinger system to remove contaminates from the gas
stream. Samples are collected isokinetically and particulates are captured on a filter
(similar to Method 5) which is later digested for analysis. For these tests, a minimum
of 30 ft3 was required to meet Demonstration Test objectives. Samples were
collected at a rate of approximately 0.5 scfm.
One multiple metals train was used to collect gas samples for Tests 1 and 3. A
duplicate sample train was installed for Test 2. ' '. '
Particulates/HCI
The stack gas was sampled for particulates and hydrogen chloride (HCI) using the EPA
Method 5 (M5) sampling train. This method involves isokinetic sampling of the stack
gas similar to the MM5 sample trains. Here the impinger system employs sodium
hydroxide and silica gel to determine contamination levels in the gas stream.
The gas was sampled at a rate of 0.5 scfm until a total of 30 cubic feet were
obtained for each sample. After the samples were collected, the train was removed
to a clean area for sample recovery utilizing acetone and deionized water. .
49
-------
Continuous Emission Monitoring -
Continuous Emission Monitors (CEMs) were tjsed for each of the test runs to monitor
the stack gas concentrations of CO2, CO, O2, SO2, NOX, and THC. Samples were
collected by inserting a stainless steel probe into one of the 1" sampling ports. Gas
was withdrawn and transported through a heated Teflon® sample line to the
instrument trailer. Once inside the trailer, the gas was conditioned before entering the
analytical equipment. Conditioning was performed by passing the gas through several
short-stemmed impingers immersed in an ice bath to remove any water that was
entrained in the gas. Following the impingers, the gas was directed through a glass
fiber filter which removed partieulates. The gas was pulled through the system by a
Teflon®-bladder diaphragm pump. Gas exiting the pump was sent to a manifold which
supplied sample to each of the CEM instruments.
ANALYTICAL PROTOCOLS
Analytical protocols were selected to provide reliable data. In most cases, the
Demonstration Test samples were analyzed using standard EPA-approved methods.
There were some variations and modifications made to these methods as noted in
Section 8 of this report. The following describes the analytical methods used for this
demonstration. ...
Volatile Organics
Analyses for VOCs of all solid and liquid samples obtained during the Demonstration
Tests were done in accordance with SW-84& Method 8240. This method is a GC/MS
method in which the sample is introduced to a purging tube and an inert gas is
bubbled through the sample. The volatiles are removed from the sample and swept
50
-------
into the gas phase. The gas is then collected and concentrated in a trap that contains
a sorbent material. After the sample has been sufficiently purged, the trap is rapidly
heated to desorb the volatiles into a gas chromatograph. The sample is separated in
the gas chromatograph and sent to a mass spectrometer for detection. The mass
spectrometer is calibrated by spiking reagent water with pollutants of interest and
analyzing under the same conditions as the samples.
Before purging the standards and samples, internal standards and surrogates are
added to the purge tube. Quantitation of the sample is performed by comparing the
response of the samples to that of the standard. Corrections are made for the
recovery of the internal standards.
The VOST samples were analyzed using SW-846 draft Method 5041 and SW-846
Method 8240. Method 5041 was required to desorb the volatiles from the VOST
tubes. For the analysis of the VOST tubes, the GC/MS was first calibrated using a.
flash evaporation technique which involved the loading of the standards on a pair of
VOST cartridges. The VOST cartridge was then placed in a clamshell heater to rapidly
desorb the volatiles. An inert carrier gas backflushed the volatiles off the VOST
cartridges. The volatiles passed through a sparge tube with 5 mL of reagent water
containing surrogate and internal standards. The volatiles were collected on the
concentrator tube in the purge and trap device. Following the desorption, the analysis
proceeded as described in SW-846 Method 824O. Samples were analyzed under the
same conditions as the standards.
If the results of the VOST tubes had exceeded the calibration range of the GC/MS, it
would have been necessary to analyze the gas canister sample for volatiles. Since the
results of the VOST samples were satisfactory, the gas canisters were not analyzed.
51
-------
ftemivolatiles
Semivolatile samples collected during the Demonstration Tests were analyzed using
SW-846 Method 8270. Method 8270 is a GC/MS method in which a sample extract
is injected into a gas chromatograph. The chromatograph splits the components of
interest which are then detected by the mass spectrometer. Surrogate standards
were added to the sample at the time of extraction to measure the extraction
efficiencies. Internal standards were added to the sample extracts before analysis.
Calibration of the mass spectrometer was performed in accordance with the
procedures prescribed in the method. Method 8270 was used to analyze the feed
soil, treated soil, TCLP extract for the treated soH> scrubber liquor before and after
each test, the scrubber caustic, and the stack gas samples.
To prepare the samples for analysis by SWJ-846 Method 8270, the compounds of
interest must first be extracted from the media of interest. Liquid samples were
extracted using Method 3520, continuous liquid-liquid extraction. Solid samples,
including the XAD-2 resin from the MM5 sample train, were extracted using Method.
3540, soxhlet extraction. The treated soil from the test had to be pulverized into a
fine powder before it was extracted. This was accomplished by running the material
through a crusher and placing the smaller pieces in a mechanical mortar and pestle
until a fine powder was obtained.
Dioxins and Furans
Analysis for polychlorinated dioxins and furans was performed using SW-846 draft
Method 8290, which is a high resolution gas chromatography/high resolution mass
spectrometry (HRGC/HRMS) method. It was not anticipated that dioxin and furans
would be present in any of the matrices. However, hexachlorobenzene, which has the
potential of forming these compounds, was spiked into the feed soil. Method 8290
52
-------
was selected to detect dioxins and furans at very low levels. The sampling plan called
for the analysis of the treated soil and the stack gas to determine if these compounds
were present. If these compounds were detected at significant levels in these
matrices, the remaining samples would be analyzed. Since dioxin and furans were not
detected at significant levels in the stack gas or the treated soil, these additional
samples were not analyzed.
Method 8290 is a rigorous method for the detection of polychlorodibenzodioxins and
polychlorodibenzofurans (PCDDs/PCDFs) which involves matrix-specific sample
preparation and cleanup. Before extraction, the samples are spiked with specific
amounts of nine isotopically (13C12) labeled PCDDs/PCDFs. The samples then undergo
the extraction procedure and extract cleanup. The final extract is prepared by adding
two recovery standards to determine the percent recoveries of the PCDD/PCDF
congeners and the hexa-, hepta-, and octachlorinated PCDD/PCDF congeners. One
to two //L of the extract are injected into an HRGC/HRMS capable of performing
selected ion monitoring and resolving powers of at least a 10 percent valley definition.
Compounds are detected and identified based on their elution at their exact retention
time, per the calibration, and the simultaneous detection of the two most abundant
ions in the molecular ion region. Confirmation is performed by a comparison of the
ratios of the integrated ion abundance of the molecular ion species to their theoretical
abundance ratios. Quantitation is achieved in conjunction with a multipoint calibration
curve for each targeted compound.
Metals
All of the sample matrices collected during the Demonstration Tests were analyzed
for metals. All liquid samples were digested per Method 3010 of SW-846. The filters
that collected paniculate matter from the multiple metals trains were digested using
SW-846 Method 3050. A modification of Method 3051 was used for the digestion
53
-------
of all soil samples for metals analyses. This is a microwave digestion procedure. The
modification to this method requires the use of hydrofluoric acid in combination with
hydrochloric acid and nitric acid to completely dissolve the sample rather than to leach
the sample. A total digestion was particularly necessary for the metals analyses of
the treated soil because of the limited leaehability of metals in this matrix. The
method was also used for the feed soil samples so that reported values are
comparable when performing a material balance. The glassified soil was crushed prior
to digestion as described for semivolatile analysis.
Analyses of the digestates were performed using primarily SW-846 Method 6010
which is inductively coupled plasma atomic emission spectroscopy method (ICP). This
method measures element-emitted light by optical spectrometry. The quantity of light
emitted is directly proportional to the amount of material present in the sample.
Arsenic concentrations were measured using SW-846 Method 7060. This method is
a graphic furnace procedure in which a sampJe of the digestate is atomized In a
graphic tube furnace. The absorption of. a specific wavelength of light is proportional
to the arsenic concentration. Selenium was measured by SW-846 Method 7740
which is also a graphic furnace technique. The procedure is the same as the arsenic
method but uses a different wavelength of light. Mercury was analyzed by Method
7471 which is a cold-vapor atomic absorption method. The mercury is reduced to the
elemental state and aerated from solution in a closed system. The vapor passes
through a .cell position in the light path of an atomic absorption spectrophotometer.
Absorbance is measured on a strip chart recorder where the height of the peak is a
function of the mercury concentration.
Toxicitv Characteristic Leaching Procedure
Feed soil and treated soil were tested for leachable metals and semivolatiles organic
compounds using the Toxicity Characteristic Leaching Procedure (TCLP). TCLP
54
-------
analyses were conducted in accordance with the procedures outlined in SW-846
Method 131.1. The TCLP method is designed to determine the mobility of organic and
inorganic analytes present in liquid, solid, and multiphase wastes. The method
involves selecting an appropriate extraction fluid, .then extracting the waste by
agitation for 18 ± 2 hours. The extract is then collected by filtration and subjected
to the required analytical procedures. Semivolatiles were extracted and analyzed
using the procedures defined in SW-846 Methods 3520/8270. Metals were analyzed
per SW-846 methods outlined above.
TCLP requires that solid waste pass through a 9.5 mm standard sieve. Since the
treated soil was a solid monolith, particle size reduction was necessary. This was
accomplished by crushing and grinding the material as described previously.
Chloride Analysis
Samples collected in the M5 sample train were analyzed for HCI using a Dionex Model.
2010 ion chromatograph following Method 27 frorn the "FGD Chemistry and
Analytical Methods. Handbook", Radian Corporation, Volume 2, July 1984. This
method involves ion separation on an ion exchange column and detection of these
ions conductimetrically.
Chloride analysis of feed soil samples were performed in accordance with ASTM
Method D-2361.
Higher Heating Value
The higher heating value of the test soil was determined using a bomb calorimeter in
accordance with ASTM Method D2015-85. The method involves charging a known
55
-------
quantity of material inside a bomb calorimeter, igniting it, and determining the amount
of heat released. . .
Parrlnnlate Matter
The particulate determination from the M5; sample train was determined using a
gravimetric procedure. A pre-weighed filter was used in the sample train, then
reweighed after the test. The acetone rinsates generated during sample recovery
were allowed to evaporate and the residue was also weighed. The amount of residue
from the rinsate was added to the residue collected on the filters to determine the
total particulate catch.
. Density
The bulk density of the treated soil was determined! using EPA Method 25A. The
samples were placed in a pre-weighed graduated cylinder to a prescribed volume. The
cylinder was re-weighed and the density was determined.
Continuous Emission Monitoring j
Gas emissions were continuously monitored using on-line instrumentation. The data
obtained from these instruments were recorded every 2 seconds using a computer
based data acquisition system. Although measured, Sb2 data was later determined
to be unusable because of interferences caused by high NOg levels in the gas stream.
All CEM instruments were calibrated using certified gas standards. Below is a
discussion of the analysis method for each parameter.
56
-------
Carbon Monoxide
A Bendix Model 85-105CA analyzer was used to measure CO in the stack gas
according to EPA Method 10. The instrument operates by using a dispersive infrared
analyzer which measures the concentration of CO by infrared absorption at a
characteristic wavelength.
Carbon Dioxide
A MSJA Model 303 non-dispersive infrared analyzer was used to detect the
concentration of C02 in the stack gas according to EPA Method 3A. This monitor
operates by measuring the absorption of infrared radiation at a characteristic
wavelength.
Oxygen . • '..'-'
A Taylor Model 540A oxygen analyzer was used to determine the concentrations of
O2 in the stack gas in accordance with EPA Method 3A. The analyzer measures 02
concentrations based on the magnetic properties of O2.
Sulfur Dioxide
A TECO Model 40 pulsed fluorescent analyzer was used to measure SO2 in the stack
gas continuously (EPA Method 6C). The instrument detects S02 based on the
absorption of ultraviolet radiation. As the molecule returns to the ground energy
state, fluorescence occurs. The amount of fluorescence is related to the
concentration of SO2.
57
-------
Nitrogen Oxides
A TECO Model 10 analyzer was used for NOX measurements in the stack gas
according to EPA Method 7E. The instrument works by converting all of the nitrogen
oxides present in the sample to nitric oxide. The nitric oxide is then reacted with
ozone. This reaction produces a chemiluminescence proportional to the NOX
concentration in the sample. The chemiluminescence is detected by a high-sensitivity
photomultiplier.
Total Hydrocarbons :
! i
A Beckman Model 400A was used .to continuously measure the concentration of
hydrocarbons present in the stack gas in accordance with EPA Method 25A. This
instrument uses a hydrogen flame ionization detector which ionizes the sample as it
passes through the flame. The carbon atoms in the sample are ionized to produce
. positive cations and electrons. These charged particles are collected and produce a
•current which is measured. The current generated is directly proportional to the
concentration of hydrocarbons present in the sample.
58
-------
SECTION 6
PERFORMANCE AND DATA EVALUATION
fRODUCTION
ree Demonstration Tests were performed to evaluate the effectiveness of the
ech, inc. Plasma Centrifugal Furnace (PCF-6, in treating the waste feed matrix
- •
e un,ts at hazardous waste treatment faciiities throughout the country To
«ate this evaluation, the foiiowing "crftioar and "noncriticar objectives were
[caj:
tc .characterize the residues produced at optimum operation induding
Removal Efficiency ™i fate °n
to identify pre- and post-feed waste treatment
requirements.
to evaluate the abiiity of the furnace to effectively vitrify inorganic and meta,
constituents within a soil into a mo I'th'
iQUlt} i ripSS I xin^* ^lyi^io VA/OO
spiked into the soil at a level of 28,000 ppm.)
59
-------
• to determine of the furnace can meet 39.99% ORE for the target analytes in
a soil contaminated With up to 10% organics. (Hexachlorobenzene was spiked
into the feed soil at 1,000 ppm so that a ORE of 99.99% could be positively
determined. DREs for other target analytes, i.e., SW-846 Method 8240 and
8270 compounds as listed in the Quality Assurance Project Plan (Section 5 of
the Demonstration Plan), were determined if these compounds were'present at
high enough levels in the feed soil.)
fMoncritical:
• to achieve heat and mass balances;
• to characterize the performance of the process;
• to identify the heed for secondary treatment;
• to isolate operational problems in the field;
• to identify solutions to potential problems;
• to identify government policy and regulatory requirements;
• to evaluate potential uses of the process;-
• to provide a comparison against competitive technologies;
• to develop operating costs; and
• to determine the useful life of the equipment.
In addition to allowing an evaluation of the technology for potential Superfund
applications, the activities and results of this testing will also provide assistance to
DOE in their evaluation of the technology for the remediation of hazardous waste sites
under their jurisdiction.
All detailed analytical results are presented in Volume II of this Report. Generally, only
results that have undergone data reduction appear in this section.
60
-------
TOXIC1TY CHARACTERISTIC LEACHING PROCEDURE
The PCF is designed to encapsulate inorganic compounds in the vitrified slag and
render the treated soil-nonleachable. Testing activities have demonstrated that the
process can effectively bind inorganic compounds into the treated soil. The Toxicity
Characteristic Leaching Procedure (TCLP) was performed on both the feed soil and the
produced slag. The feed soil was tested to establish initial values for the teachability
of organic and inorganic compounds. The vitrified slag underwent TCLP to meet the
testing objectives.
TCLP analysis of the feed soil for metals showed that the only elements which
exhibited significant leachability characteristics were: calcium, sodium, and the spiked
zinc. Table 7 summarizes the results of the TCLP metals analysis of the feed soil.
The presence of sodium in the leachate is not unexpected because of its high
concentration in the soil and the fact that it is a weakly dissociable metal. This means
that sodium, unlike other metals, is readily soluble. If the solution is even slightly
acidic (as in the TCLP) -this phenomenon is enhanced. None of the eight RCRA
characteristic metals found in the feed soil leachate were above the regulatory limit.
The evaluation of the leachability of the vitrified slag was based on calcium and zinc.
Calcium was chosen, in addition to zinc, because of its tendency to leach from the
feed soil. Sodium was not monitored because of its unusual solubility characteristics
as explained above. .
The treated soil TCLP metals analysis is also shown in Table 7. None of the metals,
with the exception of sodium, showed any strong characteristic for leaching. Sodium
is probably-present in the leachate for the reasons stated above and was not
* i
considered in this evaluation. Both tracer metals, calcium and zinc, showed
significant reductions in leaching properties in the treated soil as compared to the
feed. In fact, all of the metals, with the exception of aluminum and iron showed
reduced leachability characteristics. The leachability for the aluminum in both feed
-------
Table 7. TCLP Results for Demonstration Tests
.Compound
Nickel
•Average' Feed Soil
Leachate
Concentration
(mg/L)
Treated Soil Leachate Concentration
• iTest^l :" .-'•••^
(mg/U ; t
, -:: .'••«T.eat:2
-------
The only organic constituents that were found to be teachable from the feed soil were
2-methylnaphthalene and naphthalene, as shown in Table 7. Although the feed soil
was spiked with a high level of hexachlorobenzene (1,000 ppm), it did riot leach from
the soil. No organic compounds were found to leach from the treated slag.
The Toxicity Characteristic Leaching Procedure requires samples to be ground into
small particles, in this manner, a large amount of surface area is available for
leaching. Since the PCF produces a monolithic slag after treatment, the surface area
per pound of treated soil is much smaller than that of the feed soil. The TCLP
results,, therefore, present a conservative assessment of the actual teachability of the
monolithic slag.
-DESTRUCTION AND REMOVAL EFFICIENCY
The Destruction and Removal Efficiency (ORE), used to determine organic destruction,
• is determined by analyzing for the Principal Organic Hazardous Compound'{POHC) in
the feed soil and the stack gas. The DRE may be calculated as follows:
mass in - mass out
DRE {%) « , x 100%
mass out
For these tests, the POHC was hexachlorobenzene. The estimated mean level of
v hexachlorobenzene, based on all the feed soil samples for the three tests was 972
ppm (see Table 8). The 95% confidence interval for the estimated mean was 864 to
1,080 ppm. No hexachlorobenzene was detected in the stack gas, therefore, all
hexachlorobenzene DREs determined are based on the detection limit from the
appropriate tests. Table 9 gives these DREs based on the 95% confidence interval
of the feed soil and the detection limit for each test.
63
-------
Table 8. Organic Compounds in the Demonstration Test Soil
Compound
Benzene
lb/100 Ibfeed
ppm
Volatile*1
9.91E-05
. 0.991
Ethyl Benzene
2.84E-03
28.4
Toluene
1.8 IE-OS
18.1
Xyiene
1.34E-02
134
Semivolatil«c°
Hexachlorobanzone
9.72E-OS
972
•Methyl naphtha) arw
4.S8E-OS
458
Naphthalene
1.521-05
150
Phenanthrene
6.62E-06
66.2
1,2-biohloroethane and methyl ethyl ketone were both detected a few samples, but only at low levels.
Other compounds were detected in a few samples, but only, at low levels.
Table 9. ORE Results for Demonstration Tests
_ ' Compound ,,
Test!
..... Testl
':-*•• .Duplicate
Tast 2
Test3
Hexachlorobenzene
Lower 95% Confidence Interval Limit
Mean
Upper 95% Confidence Interval Limit
Lower 95% Confidence Interval Limit
Mean
Upper 95% Confidence Interval Limit
>99.9964
>99.9968
>99.9971
>99.9982
> 99.9984
> 99.9986
>99.9990
> 99. 9991
>S9.9tS2
2-Methylnaphthatene
>99.9853
>99.9872
>99.9891
> 99.9930
> 99.9939
> 99.9948
> 99.9958
> 99.9964
> 99 .9969
>99.999S9
> 99. 9999©
>SS.§SS91
> 99.99960
> 99 .99965
> 99.99970
As can be seen from Table 9, the estimated average ORE values for this test ranged
from > 99.9968% to > 99.9999% for a highly chlorinated compound,
hexachlorobenze.ne. It can be reasonably assumed that this level of DRE (if
measurable) can be achieved for most chlorinated or halogenated compounds.
64
-------
The Silver Bow Creek soil was mixed wjth 10% by weight No. 2 diesel oil, in addition
to being spiked with hexachlorobenzehe. Analysis of the mixed feed soil indicated
that sufficient 2-methylnaphthalene, another semivolatile compound, was present at
sufficiently high levels in the feed to determine a significant ORE for each test. The
95% confidence interval for 2-methylnaphthalene was 390 to 526 ppm with an
estimated mean of 458 ppm as shown in Table 8. This level of contamination in the
feed soil leads to the range of DREs given in Table 9, again based on detection limits
because none of this compound was detected in the stack gas.
Total xylenes, a group of volatile compounds, were also found in sufficient quantities
in the feed soil to determine a significant ORE. The 95% confidence interval for the
estimated mean of the total xyienes was 128 to 139 ppm. The DREs associated with
this confidence interval for the three tests were >99.9929% to >99.9934%.
Throughout each of the Demonstration Tests, multiple Volatile Organic Sampling Train
(VOST) samples were taken. VOST samples were obtained using SW-846 Method
0030 which designates specific volumes collected over a short duration. The DREs
presented are an average over all three tests based on the 95% confidence interval
of xylenes in the feed soik Averages of all- the DREs can be taken because the
detection limits obtained are the same for all the VOST samples.
Overall, the PCF appeared to be very efficient in destroying both volatile and
semivdlatile compounds when both the primary reaction chamber and the afterburner
were operating.
ACID GAS REMOVAL AND PARTICIPATE EMISSIONS
Measured HCI emission rates ranged from 0.0007 to 0.0017 Ibs/hr. The chlorine
concentration in the feed soil for Test 1 was 0.066%. This leads to a HCI removal
efficiency of 98.-5%. Because of the low chlorine input, the regulatory requirement
65
-------
of the larger of either 4 Ibs/hr or 99%: removal [40 CFR (07/01/91 Edition)
§264.343(b}] was met. The removal efficiency may not be meaningful because of the
low chlorine input. However, it appears that even if the chlorine in the feed had been
higher, an HCI removal efficiency of 99% could be achieved.
As shown in Table 10, the particulate emissions during each of the three tests
exceeded the regulatory limit of 0.08 grains/dscf [40 CFR (07/01/91 Edition)
§264.343(c)]. These emission rates have not been corrected for 7% oxygen. The
7% oxygen correction factor is 14%/(21 % - O2%). The oxygen correction is required
by RCRA for all hazardous waste incinerators except those operating under the
condition of oxygen enrichment. The purpose for correcting for 7% O2 in
conventional incineration systems is to account for the dilution factor in the stack gas
caused by using excess air for combustion. .For the Retech process, pure oxygen is
fed to the primary chamber through an oxygen lance as soon as feeding of the first
' ' •
batch of soil begins. The O2 is continually introduced to the furnace throughout the
* * •
remainder of the treatment process. The O2'content of. the stack gas when no soil is
being fed to the furnace (Ke, between feeding cycles, during recharging of the feeder),
is in excess of 21%. As stated above,-RCRA does not require the 7% O2 correction
factor for hazardous waste incinerators operating under oxygen enrichment [40 CFR
(07/01/91 Edition) §264.343(c)3. Therefore, the PCF is exempt from this correction
factor. During feeding of the soil to the furnace, the stack gas 02 level drops to
approximately 11%. Therefore, if the correction is-to be applied only during the
feeding cycle (when presumably the particulates are being generated), then the values
given in Table 10 should be increased by a factor of 1.4.
Table 10. Particulate Results for Demonstration Tests
, "„-.,,- "Parameter '...;;•
Paniculate Concentration (grains/dscf)
* Particulato Emissions (Ib/hr)
Test 1 !
0.341
0.342 j
Test 1 ;
Duplicate
0.240
0.238
Test 2
0.422
0.418
Test 3
0.410
0.423
66
-------
The amount of particulates captured by the air pollution control system was extremely
small. This is demonstrated, in part, by the low level of scrubber solids present in the
sump. There was less than 0.5% total solids in the scrubber sump tank. The
consequence of this high particulate loading during the Demonstration Test was a
substantial buildup of particulate matter in the exhaust blower after the air pollution
control system. The pressure differential across the exhaust blower was reduced
because of the particulate build-up, and the first two Demonstration Tests were
shortened due to this problem. A larger blower was installed for the third
Demonstration Test, which was completed as scheduled, but particulate build-up still
persisted.
The Silver Bow Creek Superfund Site soil was extremely dry and dusty before it was
mixed with the diesel oil. Even after mixing the soil with the diesel oil, it was very
free-flowing with no standing liquid. It is possible that the fines from this feed may
not have been retained in the melted soil in the primary reaction, chamber and simply
passed through the treatment process and the scrubbing unit and into the exhaust
blower and stack gas. If the dust did pass .through the treatment process, it would
be expected that a well-designed wet scrubbing system, would be capable of capturing
the particulates. If the feed soil does not contain any halogenated compounds, then
the process does not require a wet scrubbing system and a baghouse could be used
to control the particulate emissions. Judicious selection of the most suitable air
pollution control device is necessary before effective implementation of this
technology can be achieved.
AIR EMISSIONS
The air emissions consisted primarily of products of incomplete combustion (PICs) and
particulates. Table 11 presents a summary of the semivolatile organic compounds
emitted in the stack gas. For.the case of the emitted semivolatile organics, the most
67
-------
Table 11. Stack Gas Composition During the Demonstration Tests
Compound % ' ' :
Te«t 1
lbs/100 Ibi feed
•>', ;sppm ,. •• r;
Teat 2 i , •;
Ibt/iOO lb« feed
,:; ;v:ppm; ;;
;;ib«/1tJOtt>»
•-;•• ".fend ••• "?•
SemlvolatUec
bti(2-Ethylhexvl>phthalat»
4-Ntaophenol
<2.15E-O6
3.77E-04
<4.95E-07
<3.28E-O7
O.93E-07 B
<5.41E-O5 B
2.13E-05 J
7.36E-05 B
1.68E-O5 B
0.13E-O7 J
that thfe compound was dotecsad in a blank.
J Eitimatad result. Indicates that retult is lasc than the quantitation limit. The quantitation limit it defined at five timae the detection limit.
< Not detected at or above the detection limit.
68
-------
dominant compound released in the stack gas is benzoic acid at an average
*i* *- t
concentration of approximately 4 ppm. The occurrence of benzoic acid in the stack
gas is to be expected as both toluene (from the diesel oil) and chlorine (from the
hexachlorobenzene) were present in the feed soil. These two compounds, with the
addition of heat, readily form benzoic acid as shown in the basic reaction presented
below: .
CI2 H20, OH'
C6H5CH3 — > CeH5CCI3 > C6H5COOH
heat
(Toluene) (Benzotrichloride) (Benzoic Acid)
The water and hydroxide required to complete the reaction were provided by the
scrubber.
A group of nitrated compounds was found in-the stack gas at low levels (< 0.3 ppim).
Thes.e compounds were formed because of the higlvlevels of NOX and trace quantities
of organic compounds in the stack gas interacting with the scrubber liquor spray.
Other compounds such as the phthalate groups and naphthalene were also present
but were found in the field blanks as well.
In addition to the target compounds identified by the SW-846 Method 8270 list, the,
next 20 highest peaks from the chromatograms were investigated for compound
identification and semi-quantification. A review of the chromatograms and the
spectral data showed that some Tentatively Identified Compounds (TICs) were present
in the gas stream. The first Demonstration Test yielded a relatively clean
chromatogram that was somewhat comparable to the field blank. No TICs could be
positively identified, but it appeared that some of the unknowns contained oxygen
somewhere in the molecular structure "and some of the unknowns were nitrogen-
containing compounds.
69
-------
Test 2 data appears to be notably different frbm the Test 1 data. The chromatogram
indicated that several higher molecular weight compounds were detected. The TIC
data shows that these compounds were unidentifiable carbonic acids. It should be
noted that there were no positive identifications made on the specific type of carbonic
acid, but the pattern was present in several of the TICs. Carbonic acids are formed
by reacting carbon dioxide with water at high temperature. This may suggest that the
torch cooling water leak detected during Test 3 could have developed during Test 2
(see Section 4). Because of this leak, Test 3 was aborted early and restarted after the
leak had been repaired. Compounds that eluted in the early stages of the
chromatogram contained nitrogen and oxygen within their molecular structure.
Data from Test 3, collected after the leak was repaired, was very similar to the TIC
information gathered from Test 1. In general, only low levels of semivolatile organic
compounds were identified in the gas stream.
i
Very low.levels of volatile organic compounds were detected in the exhaust gas
stream. The most common of these compounds was benzene at an average
concentration of approximately 19 ppbv. Benzene and substituted benzenes are
prevalent in many forms throughout diesel oil and hence benzene is a readily formed
PIC. Other identified compounds in the stack gas tended to be chlorinated orgariics
which were not identified in the feed soil at the detection limits achieved for the
testing and therefore can possibly be PICs.
Sampling and analysis for polychlorodibenzodioxins (PCDDs) and
polychlorodibenzofurans (PCDFs) in the exhaust gas stream was accomplished during
the Demonstration Tests. The results of these analyses indicated that no PCDDs or
PCDFs were formed in the stack gas. Although some PCDDs and PCDFs were
detected in some of the samples analyzed, the levels detected were lower than the
corresponding blank sample detection limit.; For example, a particular isomer of a
PCDD was detected with 10 picograms of catch; however, the field blank reported
70
-------
a'nondetect with a detection limit of 20 picograms of catch. The variation of the
detection limits is a function of the matrix being analyzed and the resolution of the
analytical instruments used to quantify the samples.
Metal emissions (summarized in Table 11) were almost exclusively in the solid phase.
Very little of the metals was found in the impingers of the sampling trains. The only
significant vapor phase metals detected were calcium and mercury. A very volatile
metal such as mercury is expected to be found in the vapor phase. Arsenic, copper,
iron, lead, potassium, and zinc were in abundance in the stack gas in the solid phase.
The copper in the stack gas is suspected to originate from the throat of the furnace
and the torch electrode as copper was not present in high quantities in the feed soil.
The presence of iron in the stack gas, at 66 ppm, was a consequence of the high
levels of this element in the feed soil and the need for using a mild steel doughnut for
start-up purposes. Arsenic, at approximatefy 6 ppm in the stack gas, and lead, at 4
pprn, were not retained in the treated soil as they are volatile metals (arse'nic
sublimates) and most probably evaporated from the soil while it was being treated in
the furnace. Potassium was found in the stack gas because of its high initial levels
in the feed soil (see TEST SOIL AND TREATED SLAG, below). The high level of zinc
in the stack gas, at approximately 125 ppm, was a consequence of the high spiking
level of this element and the high volatility of this metal in the temperature range
encountered within the furnace. Additionally, the presence of chlorine with the zinc
at this temperature range rapidly increases the volatility of zinc. In all cases (except
mercury) the air pollution control system should have captured the metals. As shown
later, the scrubber liquor did not contain high levels of metals after each of the tests.
Based on these results, it appears that not all of the volatile metals are captured in the
molten soil at the completion of treatment. If this is the case, then these volatile
metals should be captured by the gas treatment system (assuming it is correctly
designed). A percentage of the volatile metals originally found in the soil would
71
-------
appear in the scrubber liquor, therefore, possibly requiring treatment of the liquor prior
to disposal.
TEST SOIL AND TREATED SLAG
As stated previously, the feed soil consisted of a mixture of Silver Bow Creek
Superfund Site soil, which is classified as a high metal-bearing soil, and 10% by
weight No. 2 diesel oil. Into this mixture, zinc oxide and hexachlorobenzene were
.spiked to provide 28,000 ppm and 1,000 ppm, respectively. (The corresponding
amount of zinc is 22,500 ppm zinc). Section 3, TEST SOIL, provides additional
information regarding the preparation of the waste feed.
: ! '' •
Analysis of the feed soil showed that it contained volatile compounds consistent with
'those associated with diesel fuel: benzene, toluene, ethyl benzene, and xyiene
(BTEX). These compounds were detected in the quantities presented in Table 8. The
se"mivolatile compounds found most .predominantly in the feed soil were the spiked
hexachlorobenzene and 2-methylnaphthalehe (see Table 8). Other diesel-based
compounds were also found in the soil, but at levels that could not be accurately
quantified by SW-846 Method 8270. Gas chromatography/rnass spectrometry
(GC/MS) analysis also indicated the presence of large numbers of TICs. These TICs
are typical- of those found in diesel oil and consisted mainly of compounds of
naphthalene and benzene. The metals found most abundantly in the feed soil were
aluminum, calcium, iron, potassium, sodium, and zinc.
Volatile organic analysis was not performed on the treated soil as no volatile
compounds were considered to exist in the slag after it had reached its melting point
temperature. The only semivolatile organic compounds found in the treated soil were
low levels of two phthalate compounds which were probably sampling or analytical
72
-------
contaminants. This agrees with the TCLP analysis of the slag discussed -earlier in
which no semivolatile compounds leached from the slag.
For the first Demonstration Test, 480 pounds of feed soil was fed to the furnace and
a steel "doughnut" weighing 74.5 pounds was added. As the feed soil contained
10% organics, which appeared to have been destroyed, a calculated 506.5 pounds
of slag should have been poured into the collection chamber. The mass of treated soil
collected in the slag chamber was 277 pounds, the remainder being retained within
the reaction chamber to provide a "skull" for the next test. (A skull is a layer of
melted material around the interior of the primary chamber which reduces the
chamber's volume and acts as an insulator to protect the refractory.) Therefore, a
mass balance that yields meaningful results cannot be performed on this technology
since a portion of material from each test can potentially remain in the reactor at the
end of treatment. It is possible, though to compare the concentration of the inorganic
elements in the feed soil with that of the collected slag, taking into account the
destruction of the 10% organics (by weight) and assuming that none of the elements
are concentrated in the poured slag. Table 12 gives the concentrations of the metals
in feed and in the slag for all threexrf the tests. The:feed is an average of alJ feed
samples from the three tests. This table shows that a large percentage of the metals
from the feed soil are retained within the vitrified slag. Exceptions to this trend are
generally the volatile metals: arsenic, lead, mercury, and zinc. These volatile metals
have been found, as stated earlier, to be exiting, the system through the exhaust
stack. In addition, some of these metals can be found in the scrubber liquor.
To evaluate the fate of the feed soil metals, a comparison can be made of the
behavior of a non-volatile metal such as aluminum to that of a volatile metal such as
zinc. Test 3 provides a good basis for examination since 600 pounds of waste were
fed into the furnace and 595 pounds were poured. The feed soil utilized during Test
3 contained 29.6 pounds of aluminum. A total of 26.0 pounds were detected in the
treated soil. ,This represents 88% of the aluminum originally present in the feed soil.
73
-------
Traces of aluminum {0.00340 pounds total) were also detected in the stack gas. A
small increase in the aluminum concentration from the pre-test scrubber liquor to the
post-test scrubber liquor was also noted. The reminder of the aluminum which has
not been accounted for may be due to sampling and analytical variation.
I :
For zinc,.13.9 pounds were fed to the furnace in the feed soil during Demonstration
Test 3. A total of 5.24 pounds (38%) was retained in the treated soil. The stack gas
contained 0.332 pounds of zinc. A large portion of the zinc plated out as paniculate
matter in the blower and the long exhaust gas duct. Again, sampling and analytical
Table 12. Metals in the Demonstration Test Feed Soil and Treated Soil
Element
Lead
Manganece
Mercury
Nickel
Sodium
Vanadium
Zinc
'Average Feed Soil
(ppm)
49,400 B
201
• 508
1~2.500
23.6 J
591
36,900 B
426
4.650
814
1.00
NA
19.360
10.2OO
77.3 J
23.200
', : Treated Soil
Test 1
(ppm)
51.200 ;
12.0. J
523 ;
28,800
500
780
160,000
98.3 r J
5,720
1,850
<0.133
270
15,500
8.180 B
59.8 J-
6,480
Test2
(ppm)
46,000
16.0 J
480
- 20,500
510
1 .BOO
150,000
11i " J
4,600
1,900
<0.133
265
16,000
8,650
53.5 J
9,050
TestS
(ppm)
43,700
11.3 J
4S3
18,000
617
827
213,000
120 J
4,670
2.S70
<0.133
287
14,700
7,830
50.0 J
8,800
B Indicates that this compound was detected in a blank.
J Estimated result. Indicates that the result is less than the quantitation limit. The quantitation limit is defined and 5 times.
the detection limit
NA Not Analyzed
74
-------
variation may have contributed, in. part, to the apparent discrepancy between the zinc
in the feed soil and the zinc in the treated soil and the stack gas.
The increase in iron concentration can be explained by the initial presence of the
carbon steel doughnut. The concentrations of chromium and nickel both increase in
the treated slag because, prior to the Demonstration Tests, stainless steel had been
incorporated into the furnace to form part of the skull. The increase in the copper
concentration is probably because of the melting of portions of the copper throat
during treatment. In addition, soil treated prior to the Demonstration Tests, and hence
part of the skull, had been high in calcium, manganese, and potassium.
Only the treated soil (not the feed soil) was analyzed for PCDDs and PCDFs". The
levels of PCDDs and PCDFs in the treated soil were very low. However, as described
earlier for the stack gas emissions, the detection limits for the blank samples were
higher than the levels of PCDDs and PCDFs detected in the samples. It is therefore
reasonable to conclude that no PCDDs or PCDFs were formed by the treatment
process, and if any dioxins/furans were in the feed soil, they were destroyed by the
intense heat of the process.
SCRUBBER LIQUOR
The pre-test scrubber liquor for each of the three Demonstration Tests contained very
little in the way of organic compounds. A metals scan on the pre-test scrubber liquor
showed that, generally, only low levels of inorganic elements were present. This was
expected since, prior to each test, the scrubber sump was. flushed and filled with
deionized water.
The post-test scrubber liquor did not contain any significant quantities of organic
compounds. Nitrated compounds and phthalates were the only compounds present.
75
-------
The nitrated compounds were most likely produced from the high levels of NOX in the
exhaust gas reacting with the water from the scrubber and any organic compounds
present. The phthalates and volatile organic compounds probably entered the
scrubber sump from the scrubber make-up liquor. The lack of organic compounds in
the scrubber liquor and, as stated earlier, the absence of volatile or semivolatiles
organic compounds in the exhaust gas, indicates that combustion of the organic
compounds was complete. .
The scrubbing unit was very inefficient in the capture of the inorganic compounds.
The scrubber did capture some of the volatile metal elements but not at the levels that
would typically be expected from a well-designed system. As stated previously, the
exhaust gas contained a variety of metals that should have been captured by the
scrubbing unit. The types of metals found in;the scrubber liquor were similar to those
found in the stack gas; that is, arsenic, iron, and zinc were the elements in
abundance. Higb sodium levels found in the liquor were a consequence of the
scrubber make-up (sodium hydroxide). j
CONTINUOUS EMISSION MONITORS
Throughout each of the three tests, CO; C02, O2, NOX, and Total Hydrocarbons (THC)
were monitored continuously to present a real time image of the combustion process
and to determine if regulatory standards were being exceeded. Figures 4 through 8
present CEM plots for Demonstration Test 3, typical of those generated throughout
the demonstration. SO2 was also measured, but the data collected was not
considered suitable. High levels of NOX in stack gas are known to interfere with SO2
meters, and the interference indicates the presence of SO2 when none is actually
present.
76
-------
u
Q
X
O
z
O
s
O
ffl
sc.
<
o
3
a.
a,
X
o
o
at
ce
u
u
Ul
a.
-
I-
2.0 r-
>9 'r-
18 j-
17 h
16
15
14
13
12
11
10
9
8
7
6 h
5 i-~
3
2
1
15:21
16:19 17:16 18:14 19:12
TIME OF DAY
Figure 4. CO Plot for Demonstration Test 3
20:09
.21:07
15:21 16:19 17:16 18:14 19:12
TIME OF DAY
20:09
21:07
22:04
Figure 5. CO, Plot for Demonstration Test 3
77
-------
UI
o
X
o
o
-------
o
03
u
o
a:
o
a:
a.
a,
31-
I
1 t-
i
i
-1— L_ 1 L.
15:21
16:19 17:16 18:14 19:12
20:09
21:07 22:04
TIME OF DAY
Figure 8.* THC Plot for -Demonstration Test 3
Since the installation of the afterburner in the secondary combustion chamber the
level of Total Hydrocarbons exiting th§ system has been low (<4 ppm) even with at
least 10%. organics in the feed soil. This gives a good indication that effective
thermal destruction of the organic compounds is occurring. Another indication of the
ability of the process to treat organic contaminated media is the low levels of CO in
the exhaust (approximately 1.4 ppm) and the level of CO2 (approximately 8%). O2
monitors showed significant variation throughout the treatment process as pure O2
was fed to the primary chamber at approximately 18 scfm while waste was being fed
to the furnace.
High levels of NOX are a consequence of this process if air is used as the torch gas.
The torch gas passes through the extremely hot arc of the plasma, thus oxides of
nitrogen are readily formed. Testing to date, has shown that the average
concentration of NOX in the stack gas is approximately 5,000 ppm (uncorrected to 7%
O2 as explained earlier). The oxygen lance operated intermittently rather than
continuously during Tests 1 and 3. During Test 2, however, the oxygen was fed at
a steady rate over the entire treatment time, so the NOX values corrected to 7% O2
79,
-------
may easily be calculated, if desired. This correction is not required by RCRA
regulations because the system operates under oxygen enrichment. The average
uncorrected NOX value during Test 2 was 5,467 ppm; the average NOX value corrected
to 7% O2 was 8,514 ppm.
The uncorrected average NOX value for testing (5,000 ppm) corresponds to an
emission rate of 2.5 Ibs/hr (based on operation 365 days/yr, 24 hrs/day). Federal
requirements state that the NOX emission rate must be less that 9.2 Ibs/hr (40 tons/yr
based on operation 365 days/yr, 24 hrs/day). The PCF does not exceed this 9.2-
Ibs/hr standard, but the concentration in the stack gas is high. The emission rate is
low because of the low flowrate of the exhaust gas (approximately 110 scfm). This
flowrate is dependent on the size of the torch used. The "feed rate of the soil does not
influence the level of NOX in the exhaust assuming the same torch is used for the
different feed rates. This is because the torch uses the same amount of torch gas
regardless of the soil feed rate. If a torch larger than that used in the Demonstration
Tests is "to be employed, then the use of a NOX-reduction technology should be
investigated. ; .
FURNACE OPERATION
Since all three Demonstration Tests were designed to be identical in nature, operating
conditions during the tests were relatively constant. These operating conditions are
described in detail in Section 4, FIELD OPERATIONS DOCUMENTATION. The feed
material was identical in each case, a mixture of Silver Bow Creek soil and 10% by
weight No. 2 diesel oil, spiked to provide 28,000 ppm zinc oxide (22,500 ppm zinc)
and" 1,000 ppm hexachlorobenzene. The feed rate for each test was 120 Ibs/hr.
Although the mass of material to be fed during each test was anticipated to be 960
pounds', the actual weight of the feed was 480, 360, and 600 pounds for Tests 1,
2, and 3, respectively. The corresponding weight of the treated soil generated during
80
-------
the tests was 277, 265, and 595 pounds. The difference between the mass fed and
the mass poured during each test can be accounted for by the retention of material
inside the chamber as part of the skull as described earlier in this section. As
demonstrated by these values, the skull was progressively built up throughout the
course of the Demonstration Tests.
The torch power ranged from an average of approximately 410 kW during Test 3 to
nearly 460 kW during Tests 1 and 2. The total power consumption of the torch
ranged from 3,308 kWh (Test 1) to 4,720 kWh {Test 3). As anticipated, the total
power consumption for Test 3 was greater than the other two tests because of the
extent of its duration. The torch gas in each case was air with a flowrate of 23 to 24
scfm.
The furnace is operated so that a minimum temperature of 2,100°F is achieved in the
primary reaction chamber and a minimum temperature of 1,800°F is reached in the
afterburner before feeding of the waste is initiated. The reactor chamber temperature,
once it stabilized, achieved an average value of approximately 2,250°F. The
afterburner temperature averaged around 1 ,-800°F (slightly higher during Test 3) once
the system reached operating range. The off-gas flowrate was maintained at
approximately 110 scfm during all three Demonstration Tests.
The scrubber liquor generated during each of Tests 1 and 2 was close to 150 gallons.
During Test 3, this value was greatly exceeded due to frequent blowdpwns of the
scrubber in an attempt to reduce particulate loading on the blower downstream.
Nearly 800 gallons of scrubber liquor were generated during Test 3.
The PCF-6 is a high maintenance process that is subject to frequent stoppage because
of equipment failure. During the first Demonstration Test, a stoppage occurred when
a scrubber sump pump overheated and tripped the system. The test was also
shortened because of particulate build-up in the exhaust blower. This same problem
81
-------
caused the second Demonstration Test to be completed prior to treatment of all the
feed. While warming up the primary chamber for the third Demonstration Test, the
torch developed a cooling water leak. This led to the furnace being out of operation
for approximately three hours while the leak: was repaired. Experience of operating
the PCF-6 has shown that secondary arcing within the primary chamber is the most
common form of equipment breakdown. Torch cooling water leaks result when this
occurs, and the torch ram needs to be welded, plugging these pinhole leaks. With
respect to preventive maintenance, torch electrode replacement is the most regular
of the procedures that need to be carried out to ensure uninterrupted operation. The
electrodes must be replaced approximately every 60 to 100 hours of furnace
operation..
The plasma torch provides a substantial amount of thermal energy to the.feed,
f ! o •
however, to protect the equipment from being damaged by this heat, cooling circuits
are utilized. For the PCF-6 the cooling circuits that remove the highest percentage of
the heat from the process are the torch cooling circuit (31 %), the primary chamber
cooling circuit (39%), arid the scrubber unit (10%). For optimum operation of the
furnace, it would be anticipated that the majority of the heat removed from the
system would be from the collection chamber and the scrubbing unit. For the
Demonstration Tests, the specific energy consumption was approximately 8 kWh/ib.
Physical data indicates that, ideally, a specific energy requirement for melting soils is
approximately 0.3 kWh/lb. Therefore, there is considerable room for design and
i
operational improvement of the PCF 'systeml
82
-------
SECTION 7
CONCLUSIONS
A number of conclusions have been drawn based on this SITE Demonstration Test.
These conclusions have been briefly addressed in the EXECUTIVE SUMMARY {Section
1) and are discussed in detail in this section.
LEACHABILITY OF THE TREATED SOIL FOR INORGANIC COMPOUNDS
The PCF process is designed to encapsulate inorganic compounds in a vitrified slag
and render the treated soil non-leachable. The Toxicity Characteristic Leaching
Procedure (T.CLP) was performed on both the feed soil and the treated slag to
determine if the process can effectively bind inorganic compounds into the treated
soil. The feed soil was tested to establish initial values for the leachability of inorganic
and organic compounds. The vitrified slag underwent TCLP to meet testing
objectives. For inorganics, the feed soil only exhibited significant leachability
characteristics for: calcium (175 mg/L), sodium {1,475 mg/L)> and the spiked zinc
(982 mg/L). Sodium, was not selected as a tracer compound since it is a weakly
dissociable metal and thus behaves differently from typical metals. The treated soil
did not show strong leachability for any metals except sodium which leached at
approximately the same level as in the feed soil. Both tracer metals, calcium and zinc,
showed significant reductions in leaching properties as a result of treatment.
83
-------
Overall, testing activities demonstrated that the process can effectively bind inorganic
compounds into the treated soil.
LEACHABILITY OF THE TREATED SOIL FOR ORGANIC COMPOUNDS
The only organic compounds that were found to be leachable from the feed soil for
the Demonstration Tests were naphthalene and 2-methy(naphthalene. Although the
feed soil was spiked with high levels of hexachlorobenzene (1,000 ppm), it did not
leach from the soil. No organic compounds were found to leach from the treated slag.
DESTRUCTION AND REMOVAL EFFICIENCY OF TARGET ANALYTES
The Destruction and Removal Efficiency (DRE) is based on the concentration of the
target analyte in the feed soil and the stack gas. For the Demonstration Tests the
mean level of hexachlorobenzene, based on all the feed soil samples for the three
Demonstration Tests, was 972 ppm. The 95% confidence interval for the estimated
mean of the hexachlorobenzene spiked into the feed soil was .864 to 1,080 ppm. No
hexachlorobenzene was detected in the stack gas in any of the three tests, therefore,
all hexachlorobenzene DREs were determined based on the detection limit from the .
appropriate test. The estimated average DRE for each of the three Demonstration
Tests,.based on the confidence interval of the feed soil and on. the analytical
instruments' detection limits for the stack gas analysis, were > 99.9976%,
>99.9991%, and > 99.9999%, respectively. . "
The concentration of 2-methylnaphthalene, a semivolatile compound, was found to
be between 390 and 526 ppm in the feed soil. Again, none of this compound was
v
detected in the stack gas for any of the Demonstration Tests, leading to estimated
84
-------
average DREs, based on instrument detection limits, of > 99.9906%, > 99.9964%,
and > 99.9996%, for Demonstration Tests 1, 2, and 3, respectively.
Total xylenes, volatile compounds, were found in the feed soil with a 95% confidence
interval of 128 and 139 ppm. This led to an estimated average DRE of total xylenes,
for all tests, based on instrument detection limits, of > 99.993%. The reasoning for
presenting only one DRE for a volatile compound is given in Section 6.
Overall, the PCF process appears to be very efficient in destroying both volatile and
semivolatile compounds when both the primary reaction chamber and the afterburner
are opesrating.
STACK GAS EMISSIONS
HCL arid Particulate Matter Emissions
HCI emissions were very low for all three tests. Measured HCI emission rates for the
Demonstration Tests ranged from 0.0007 to 0.0017 Ibs/hr. The chlorine
concentration in the feed soil for Test 1 was 0.066%. This leads to a HCI removal
efficiency of 98.5%. The removal efficiency may not be meaningful because of the
low chlorine input. However, it appears that if the feed contained a higher
concentration of chlorine, then a HCI removal efficiency of 99% could be achieved.
The average paniculate emissions to the atmosphere 'for each of the three
Demonstration Tests was 0.374 grains/dscf. This exceeded the RCRA regulatory limit
of 0.08 grains/dscf [40 CFR (07/01/91 Edition) §264.343(c}].
85
-------
Volatile and Semivolatile Oroanic'Compound Emissions
Small quantities of volatile and semivolatile compounds were formed as products of
incomplete combustion (PICs) in the plasma furnace. The most dominant semivolatile.
compound released in the stack gas was benzoic acid, at an average concentration
of approximately 4 ppm. The occurrence of benzoic acid in the stack gas is to be
i :
expected as both toluene (from the diesel oil) and chlorine (from the
hexachlorobenzene) were present in the feed soil. These two compounds, with the
addition of heat, readily form benzoic acid as discussed in Section 6. Low levels of
nitrated compounds were found in the stack gas at low levels « 0.3 ppm). These
compounds were formed because of the high levels of NOS and trace quantities of
organic compounds in the stack gas interacting with the scrubber liquor spray,, The
volatile compound found in the greatest abundance in the stack gas was benzene, at
approximately 19 ppbv. Benzene and substituted benzenes are prevalent in many
forms throughout diesel oil in the feed, and hence benzene is a readily formed PIC.
Ploxins/Furans Emissions .
Sampling and analysis for polychlorodibenzodioxins (PCDDs) and
polychlorodibenzofurans (PCDFs) in the exhaust gas stream was accomplished during
the Demonstration Tests. The results of these analyses indicated that no PCDDs or
PCDFs were formed in the stack gas. Although some PCDDs and PCDFs were
detected in some of the samples analyzed, the levels detected were lower than the
corresponding blank sample detection limit.
Metal Emissions '
The only significant vapor phase metals detected were calcium and mercury. A very
volatile metal such as mercury is expected to be found in the vapor phase. Arsenic,
86
-------
cpppeir, iron, lead, potassium, and zinc were in abundance in the stack gas in the solid
phase,, Lead, at approximately 4 ppm in the stack gas, and arsenic, at 6 ppm, were
not retained in the treated soil as they are volatile metals (arsenic sublimates) and
most probably evaporated from the soil while it was treated in the furnace. Potassium
was found in the stack gas because of its high initial levels in the feed soil. The high
level of zinc in the stack gas, at approximately 125 ppm, was a consequence of the
high spiking level of this element and the high volatility of zinc in the temperature
range encountered within the furnace. With the exception of mercury, the air
pollution control system should have captured the metals.
The Demonstration Tests results show that not all of the volatile metals were captured
in the molten soil at the completion of treatment. If this is the" case, these volatile
metals should be captured by a correctly designed gas treatment system.
AIR POLLUTION CONTROL SYSTEM -
Particuilates Captured bv the Air Pollution Control System
The amount of particulates captured by the air pollution control system was extremely
small. This is demonstrated, in part, by the low level of scrubber solids present in the
sump. There was less than 0.5% total solids in the scrubber sump tank. The high
particulate loading during the Demonstration Tests caused substantial build-up of
particulate matter in the exhaust blower downstream of the air pollution control
system.
The Silver Bow Creek Superfund Site soil was extremely dry and dusty* both before
and after it was mixed with diesel oil. The mixed feed soil was very free-flowing with
no standing liquid. It is possible that the fine particles of this dusty feed may not
have been retained in the primary reaction chamber in the melted soil and simply
87
-------
passed through both the treatment process and the scrubbing unit to the exhaust
blower and stack. A well-designed scrubbing system should be capable of capturing
the particulates.
Qroanics Captured bv the Air Pollution Control System
The pre-test scrubber liquor for each of the three Demonstration Tests contained very
little organics. This was expected since, prior to each test, the scrubber sump was
flushed and filled with deionized water. The post-test scrubber liquor did. not contain
any significant quantities of organic compounds. Only nitrated compounds and
phthalates were present in the scrubber liquor. The nitrated compounds were most
« \ , •
likely from the high levels of NOX in the exhaust gas reacting with the water from the
scrubber and any organic compounds present. The phthalates and volatile organic
compounds probably entered the scrubber sump from the scrubber make-up liquor or
in the case of phthalates were possibly a laboratory contaminant. The lack of organic
compounds in the scrubber liquor and the absence of volatile or semivolatile organic
compounds in the exhaust gas, indicates that combustion of the organic compounds
was complete. "
Inorganics Captured bv the Air Pollution Control System
The scrubbing unit was very inefficient in capturing inorganic compounds. Only small
quantities of inorganic compounds, mainly volatile metals, were captured by the
scrubber, and not at the levels that would be expected from a well-designed system.
The metals found in the scrubber liquor were similar to those found in abundance in
the stack gas; that is, arsenic, iron, and zinc. High sodium levels found in the liquor
were a consequence of the scrubber make-up (sodium hydroxide).
88
-------
CONTINUOUS EMISSION MONITORS .
Throughout each of the Demonstration Tests, CO, C02, O2, NOX, and Total
Hydrocarbons (THC) were monitored continuously to present a real time! image of the
combustion process and to determine if regulatory standards were being exceeded.
SO2 was also measured but the data collected was not considered suitable. High
levels of NOX in a stack gas are known to interfere with S02 meters, and the
interference indicates the presence of S02 when none actually exists. During the
Demonstration Tests, the THC exiting the system was low (< 4 ppm), even with 10%
organics in the feed. The exhaust gas contained low levels of CO (approximately 1.4
ppm) and a level of approximately 8% CO2. These levels of THC, CO, and CO2
give a good indication that effective thermal destruction of the organic compounds is
occurring.
Oxyggin monitors showed significant variation throughout the treatment process as
pure O2 is fed to the primary chamber at approximately 18 scfm while waste is being
fed to the furnace. High levels of NOX are a consequence of this process if air is used
as the torch gas, as it was during the Demonstration Tests. The torch gas passes
through the extremely hot arc of the plasma, thus oxides of nitrogen are readily
formed. NOX emission rates during the Demonstration Tests averaged approximately
5,000 ppm (2.5 Ibs/hr). However, because of the low stack gas flowrates (1 lOscfrn)
the total emissions of NOX were below the regulatory requirements of 9.2 Ibs/hr (40
tons/yr based on operation 365 days/yr, 24 hrs//day). If a torch larger than that used
in the Demonstration Tests is to be employed, then the use of a NOX reduction
technology should be investigated.
89
-------
SYSTEM PERFORMANCE AND RELIABILITY
i
The components of the PCF can be broken down into two main categories; the
thermal treatment section and the gas clean-up system. The furnace unit
demonstrated that it was entirely capable of processing the waste feed, however, the
gas clean-up system did not perform up to expectations. In fact very little particulate
matter or organic or inorganic compounds were found in the scrubber sump at the
conclusion of the tests.
The entire system is a high maintenance unit. During the course of the Demonstration
Tests, the exhaust gas blower failed twice (because of the high particulate loading in
the flue gas), the torch developed a deionized water leak, and numerous preventive
maintenance activities took place. The on-line factor for the process during the
Demonstration Tests was 70%.
• COST OF COMMERCIAL OPERATION
Several cost scenarios can vary the unit cost of operation for the furnace. The cost
of operation is strongly dependent on two factors; the on-line factor and the feed
rate. The present configuration of the feeder, furnace, and slag collector allows an
average feed rate of .120 Ibs/hr. However, feed rates of 500 and 1,000 Ibs/hr could
be achieved with a few minor furnace modifications. A larger Plasma Centrifugal
Furnace is operational in Switzerland that can process contaminated soil at a feed rate
of 2,200 Ibs/hr. If the PCF used during the Demonstration Tests is operated at a feed
rate of 500 Ibs/hr, with an on-line factor of 70% and a total treatment volume of
2,000 tons, then it is estimated that the cost per ton of this technology is $1,816/ton
of contaminated waste treated. For the larger-scale PCF, operating at a feed rate of
2,200 Ibs/hr, and the same on-line factor and total treatment volume, the cost is
90
-------
estimated at $757/ton. Although the on-line factor for the Demonstration Tests was
70%, a more realistic on-line factor would be approximately 60%.
91
-------
SECTION 8
QUALITY ASSURANCE
Quality Assurance (QA) may be defined as a system of activities whose purpose is to
provide to the producer or user of a product or a service with assurance that it meets
defined standards of quality with a stated level of confidence. A QA program is a
means of integrating the quality planning, quality assessment, quality control (QC),
and quality improvement efforts o'f various groups in an organization and to enable
* I • i ' .
operations to meet user requirements at an economic level, included are all actions
taken by personnel, and the documentation of laboratory performance as specified in
the Quality Assurance Project Plan (QAPP). The QA program is an essential part of
a sound analytical protocol used by individuals and laboratories to detect and correct
problems in a measurement process or to demonstrate statistical control. The
objective of this quality assurance program is ,to reduce measurement errors to agreed-
upon limits and to produce results of acceptable and known quality.
INTRODUCTION
To achieve testing objectives (see Section 1|) and to obtain data of known quality, a
detailed QAPP was prepared for this Demonstration Test. The QAPP specified the
necessary guidelines to ensure that the measurement system was in control. The
QAPP also detailed information on the process measurements as well as the analytical
92 .
-------
approach to ensure data of high quality could be obtained to verify vendor claims and
achieve project objectives. Certain key measurements were used to determine the
leachability of the treated soil and the Destruction and Removal Efficiencies (DREs) for
the organic material. Measurements designated as critical for evaluating leachability
and DREs (previously defined as numerical objectives for the demonstration) consisted
of sernivolatile organic compounds in the feed soil, semivolatiles in the stack gas,
along with sernivolatile and metal TCLP measurements for the treated soil.
Specifically, hexaehlorobenzene and zinc were spiked into the feed soil to evaluate
these parameters. In addition to the above critical measurements, other parameters
which were designated as secondary critical measurements included metals,
semivolatiles, dioxins/furans, particulates, HCI/chlorine, carbon monoxide, for all
matrices and semivolatiles and metals TCLP for the feed soil. Other non-critical
parameters included carbon dioxide, oxygen, nitrogen oxides, sulfur dioxide, total
hydrocarbons, higher heating value, bulk density, moisture, and volumetric flow. Each
of these measurement's, along with their'respective QC data is discussed in the
following sections. . •
The QAPP outlines several modifications made to standard analytical methods. These
modifications were made to allow collected data to be evaluated in terms of project
objectives. Modifications were made to surrogate spike compounds, matrix spike
compounds, and target analyte lists, along with sample preparation methods. These
deviations are presented in the following sections and discussed in detail later in
Section 8 under "MODIFICATIONS AND DEVIATIONS FROM THE QAPP." During the
Demonstration Test, there were no modifications or deviations made pertaining to the
collection of field data.
As part of the QAPP, audits of both the field and laboratory operations were
performed. The intent of these audits was to ensure that measurement techniques
were performed in accordance with the guidance set forth in the QAPP. Results of
these audits are discussed in this, section under "AUDIT FINDINGS."
93
-------
The following sections describe procedure's used to determine the quality and usability
of the measured data obtained for this Demonstration Test.
PROCEDURES DEFINING DATA QUALITY CONTROL AND USABILITY
To assess the quality, of the data generated for this Demonstration Test, two
important data quality indicators are of primary concern: precision and accuracy.
Precision can be defined as the degree of mutual agreement characteristics of
independent measurements as the result of repeated applications of the process under
specified conditions. Accuracy is the degree of agreement of a measured value with
the true or expected value.
i
Precision is measured by matrix spike and sample duplicates. In most cases precision
was evaluated by expressing, as a percentage, the difference between results of
matrix spike or sample duplicates for a single parameter. The relative percent
difference (RPD) was calculated as:
Maximum Value - Minimum Value x 100%
RPD * [{Maximum Value + Minimum Value}/2]
For data sets with greater than two points, the coefficient of variance (CV) or the
relative standard deviation (RSD) were used to assess precision. The CV (or RSD) can
be calculated as:
Standard Deviation x 100 .
CV(orRSD) = Mean
94
-------
To determine the agreement between a measured value and the true value, accuracy
was measured. Accuracy can be expressed as the relative error (%RE) calculated as:
Measured Value - Theoretical Value x 100% ;
%RE = Theoretical Value
To determine the accuracy of several organic and inorganic parameters, known
!
quantities of analytical and/or surrogate standards were spiked into selected samples.
Equipment used to provide data for this project were tested for accuracy through the
analysis of calibration check standards and laboratory control samples. To determine
the recovery of these spikes, the following equation was used:
r - c
^«« ^us
% Recovery of Spike = x 100%
Csa
where: C.s = analyte concentration in spiked sample
Cue s= analyte concentration in unspiked sample
- C.B = analyte concentration added to sample
During the course of the sample analyses, several samples were spiked to determine
the matrix effect on analyte recoveries. These samples were selected at random by
the analytical laboratory performing the work. i • .
Another important aspect of assessing data quality is completeness. Gompletenejss
is a measure of the amount of valid data produced from the total effort compared to
the total amount of data originally planned for the project. Specific completeness
objectives were delineated in the QAPP for all measurements.
95
-------
For this demonstration, the completeness objectives were met for all analytical
parameters. Some samples had poor surrogate recoveries or other QC indicators
outside QAPP control limits; however, most of these data points are still useable and
are discussed in greater detail in the following sections. There were only a few minor
losses of data due to problems associated with poor quality control indicators. In
addition, minor losses of data were also due to laboratory preparation of samples and
breakage of sample containers during shipment. Some of the semivolatile feed soil
samples were analyzed two days beyond their specified extract holding times which
presented another possible loss of data. Reviewing the QC information and the data
associated with these samples, however, indicates that these data are still useable for
this Demonstration Test. (See discussion later in this section under "SAMPLE
HOLDING TIMES.")
*
To determine if a measurement is valid, it must be reproducible and comparable.
Comparability expresses the extent with which one data set can be compared to
another. To generate comparable' results, standard metho'ds, which are widely
accepted, along with strict analytical protocols were used. These methods were
clearly specified in the QAPP and reviewed and approved before data or samples were
collected. Methods that were non-standard, such as the Multiple Metals Train (MMT),
were also cited in the QAPP and the appropriate analytical protocol were appended.
This allowed the results generated by this test to be reproducible by ©ther
i :
investigators using these methods. ;
Although several precautions were taken to generate data of known quality through
the control of the measurement system, the data must also be representative of true
conditions. Representativeness refers to the degree with which analytical results
accurately and "precisely represent actual conditions present at locations chosen for
sample collection. During the development: of the QAPP and the analysis of the
samples, great care was taken to ensure that the samples were representative of
actual process conditions. Specific sampling methods are discussed in Section 5.
96
-------
Of critical concern, was the sampling of the treated soil." Since the treated soil was
poured from the furnace as a molten mass, it was not possible to collect a well-mixed
grab sample. To obtain a representative sample, the molten mass was allowed to cool
and harden into a monolith. A diamond-tipped drill bit was used to cut 2" cores from
random locations in the monolith. When the cores were removed, they were collected
for homogenization. In the field, the collected cores were smashed with a sledge
hammer between cotton cloth to reduce their size. The shattered fragments were
thoroughly mixed and placed into appropriate sample containers for shipment to the
laboratory. In the laboratory, the sample was further reduced in size by grinding the
fragments in a mechanical mortar and pestle. Once the material was in a powder-like
form, aliquots were collected for specific analyses.
ANALYTICAL QUALITY CONTROL j .
: - 4 - ^ -
Quality control was measured in each of the analytical parameters through the use of
surrogate spikes, matrix spikes, blanks, and field and laboratory duplicates. Surrogate
spikes are added to the matrices before extraction and provide an indication of the
extraction efficiency. Matrix spikes provide an indication of how well target analytes
can be recovered in sample matrices. Matrix spike and laboratory duplicates provide
information on laboratory precision. Blanks provide an indication of contamination
that may have been introduced to the samples in the field, during shipment, and
during laboratory preparation. Field duplicates also provide information concerning
sample representativeness. Good agreement among field duplicates indicates that the
'matrix sampled is homogeneous and that representative samples have been collected.
In. addition to these QG parameters, criteria was placed on the analytical laboratory
to ensure that the results were performed using acceptable methodologies as outlined
in the QAPP. For each measured parameter, internal QC procedures were designed
to control and assess the performance of the analytical system. The following
sections discuss the results of these QC indicators for each analytical parameter.
97
-------
Each of the following sections addresses, the accuracy and precision associated with
all measurement parameters. As previously noted, completeness objectives were
satisfied for every parameter.
Volatile Organlcs
Volatile organic samples were collected from the feed soil, scrubber liquor before and
after each run, and the scrubber makeup. In addition, volatile emissions were
analyzed by the collection of volatile organic sample train (VOST) samples as
discussed in Section 5. Samples of the feed soil were collected from the 5-gallon feed
pails during the loading of the feeder. Liquid samples were collected from the
scrubber sump by allowing water to purge the samples lines. Once the line was
purged, the flow was diverted into standard 40-mL VOA vials. These samples were
analyzed using SW-846 Method 8240. This method uses a purge-and-trap analyte
concentration procedure prior to analysis by gas chromatograph/mass spectrometry
(GC/MS). - "
As outlined in the QAPP, each day the GC/MS was operated, a mass axis calibration
was performed and the spectral fragmentation pattern was checked through the use
of bromofluorobenzene (BFB). The BFB was required t© pass the criteria specified in
the method. The instrument was calibrated for the target volatiles through the
analysis of standards across five concentration ranges: Each day, this calibration was
checked to ensure that instrument conditions did not change and that the initial
calibration data was valid. Calibrations were required to pass criteria for system
performance check compounds and calibration check compounds as specified by SW-
846 Method 8240. Calibrations were verified every 12 hours. All samples were
spiked with internal standard compounds to provide accurate quantitations relative to
the calibration standards. Internal standard irecoveries were acceptable for each
98
-------
analysis. Each sample was also spiked with surrogate standards to monitor the
purging efficiency of the sample. Spike data is summarized in the following sections.
Feed Soil Volatiies
Volatile organics in the feed soil was designated as a secondary critical parameter for
this demonstration. Specifically, xylene concentrations were used in the same manner
as semcvolatiles in the feed soil when determining DREs for evaluating project '
objectives. Due to the high levels of target analytes in the feed soil, the samples were
analyzed using the SW-846 Method 8240 medium level protocol. This entailed
extracting a known quantity of the sample with a known volume of methanol. A
portion of the methanol extract was then analyzed using conventional purge-and-trap
procedures. Surrogate recoveries for the feed soil volatile organic samples are
summarized in Table-13. As noted in this table, all surrogate recoveries met the
specified recovery control limits.
Matrix spikes for the feed soil were not performed because of a mistake made on the
chain of custody records sent to the analytical laboratory. In viewing the results of
all of the feed soil samples, it can be seen that there is very good agreement among
these results (see Appendix D). The standard deviation for analyte concentrations
were well within QAPP precision specifications. These data were expected to show
good agreement because the feed soil was homogenized prior to sampling. In
addition, surrogate spike recoveries, as noted above, were within specified control
limits. Therefore, it is believed that the absence of matrix spikes has a limited impact
on data quality and does not adversely affect project objectives.
Two sets of duplicate samples of the feed soil were collected and analyzed during the
first test run. Results of these duplicates are presented in Table 14. This table
summarizes the results of the detected compounds. Compounds that do not appear
on the table were not detected. The feed contained high levels (ppm quantities) of
99
-------
Table 13. Volatile Surrogate Summary Data
Dichloroethano-d
4-Bromo-d
4-Bromofluorobenzene
———• ——.^1^—
Toluona-ds
Number of
«d -up*—. .Does not ta*-.
. -
u
Include* one aoiubber make-up water sample.
Table 14. Feed Soil Volatile Duplicate Sample Results (Test 1)
Benzene
•••••••••
1,2-Dichloroethane
Ethyl Benzene
Methyl Ethyl Ketone
^•z™—»™
Methyiene Chloride
••• " • ""
Toluene
•B^—•—•
Xyienes
i ' ~" —
Bonzene
•B«w»aaBB»-»»»-
Ethyl Benzene
«••••••••^^^•^—•^—^
Methylene Chloride
Toluene
«^«B«—«^-«
Xyienes
==
Indicates that this compound was not detected;
as 5 times the detection limit.
NC Not calculated.
^"detected be.ow the quantitation Hmits. The <,uantitation limit to dOf,ned
100 . .
-------
organic compounds that are typical' of. diesel fuels. The feed soil samples show
excellent agreement between samples. The compounds methylene chloride, 1,2-
dichloiroethane, and MEK were detected in one sample but not the others. This was
most likely caused by the introduction of laboratory contamination or values detected
near the detection limit. As noted in Table 14, RPD values were not calculated for
these parameters. ! ' '
i
Field blanks collected for the feed soil for the first two tests were clean with the
exception of low levels of methylene chloride. The field blank for Test 3 contained
benzene at 12 //g/kg, chlorobenzene at 19 //g/kg, and some methylene chloride.
•• \
These low levels have no significant impacts on data quality. A complete summary
of the volatile feed soil data is presented in Appendix D.
The only volatile compound used for computing DRE were total xylenes because of
their high concentration in the" feed soil. This is seen in a review of the project
conclusions (see Section 7). Xylenes were well within precision objectives as shown
in Table 14 and because all surrogates were within specified objectives for all
samples, only data of known and sufficient quality were used for evaluating this
conclusion in relation to stated project objectives. ;
Volatile Scrubber Water Samples . . .
Scrubber water samples (a secondary critical measurement) were collected and
analyzed per the procedures outlined in the QAPP with no modifications. Results of
these samples indicated little or no significant volatile compounds were detected.
Scrubber samples were spiked with surrogate standards before analysis. These
results are also presented in Table 13. For the scrubber liquor samples, all surrogates
were recovered within the control limits.
101
-------
As with the feed soil, matrix spikes were not performed in this matrix because of an
error on the chain of custody records. Matrix spikes for these samples would have
been of limited use since there were no significant levels of volatile target compounds
found in these samples. Matrix spikes would have indicated potential recovery
problems associated with specific target analytes, however, surrogate spikes showed
reasonable recoveries indicating no recovery problems for general chemical compound
classification. The absence of .matrix spikes is therefore not believed to adversely
affect the quality of data for these samples. In addition, no critical project conclusions
were based upon data from scrubber samples. Specifically, scrubber water data were
only used in supporting emission data for evaluation of the emission control system.
Duplicate samples of the scrubber water were collected and analyzed before and after
each Demonstration Test. These data are presented in Table 15. The scrubber water
samples are also in good agreement for the compounds that were detected. In
general these samples were very clean with little or no target' analytes detected.
There are some variations that yield high RPD values for methylene chloride.
Methylene chloride has'also been identified as a laboratory contaminant which
probably attributed to the poor precision of this compound in these samples.
• The field blank collected for the scrubber liquor contained low levels of toluene, MEK
and methylene chloride. A sample of the scrubber make-up was collected and
analyzed in the unlikely event that unexplained volatile results were obtained in the
scrubber. No target compounds were detected except MEK at 44 //g/L, methylene
chloride at 1.6 ,/g/L, and toluene at 1.6 ^g/L. These results have no impact on
scrubber volatile data quality.
Volatile Organic Sampling Train Gas Samples
Samples of the gas stream were collected using the volatile organic sampling train
(VOST) to determine the types and quantity of volatile organic emissions generated
102
-------
Table 15. Scrubber Water Volatile Duplicate Sample Results
"""• %I "" / ' -'-'"' , T- '••,'"- -:
PRE-TEST
Chlorobenzene
Methyl Ethyl Ketone
Methyleine Chloride
POST-TEST
Methyleine Chloride
PRE-TEST
Methyl Ethyl Ketone
Methyleine Chloride
Xylanes;
POST-TEST
Methyl Ethyl Ketone
Methyleine Chloride
PRE-TEST
Methyleine Chloride
POST-TEST
Benzems
Chloroform
Dibi-omomethane
Methyl Ethyl Ketone
• Methyleine Chloride ' '
Toluenei
" 'Primary,
Tect 1
•
31 J
1.8 JB
44 B
Test 2
1600
1.1 JB
1.6 J
,
200
15 B
TeetS
22 B
0.85 J
2.4 J
a
1.3 J
26 J
0.51 J
:i% *? , ;•
3.4 J
33 J
20 B
16 B
"—
1800 J
14 JB
•
260
5.4 • B
6.2 B
0.88 J
2.3 J
0.78 J
•
22 J
0.63 J
, V. .. V."
'
NC
6
167
93
12
171 ':
NC
26
94
1
112
3
4 •
NC i
NC ;
17
21
ObjacJiveB
Preciaion -
30
30
. 30
30
30
30
30
30
30
30
- 30
30
30
30
30
30
* Indicates that this compound was not detected.
B Indicates that this compound was detected in a blank. -
J Estimated Result. Indicates that this compound was detected below the quantitation limit. The quantitation limit is defined
and !5 times the detection limit.
NC Not calculated.
103
-------
during the Demonstration Tests. These data are considered as secondary critical
measurements used primarily for evaluating emission levels and to determine if PICs
were being formed. Total xylenes, as previously noted, were used in determining
DREs as stated in the conclusions and these data were considered critical. VOST
samples were collected under the guidance of SW-846 Method 0010. For each of the
*
Demonstration Tests, several VOST samples were collected. The normal sampling
procedure calls for the collection of 20 L of sample to be collected for each Tenax®
pair. To ensure that the samples did not become overwhelmed and saturate the resin,
several pairs were collected at various sample volumes (see Section 5) during each
Demonstration Test. In addition, before the melt was poured into the pig, a fresh set
of Tenax® traps was installed to measure any sudden release of volatile contaminants
during the pouring process. When these samples were returned to the laboratory,
they were analyzed in accordance with Methods 5041 and 8240. Similar procedures,
were used to calibrate and tune the mass spectrometer for VOST analysis. Calibration
of the instrument was performed by using a flash volatilization technique. Here,
standards were loaded onto a clean set of Tenax® traps per Method 5041
specifications. These standards were then desorbed -and analyzed using the same
procedures employed while analyzing samples.
Surrogate recoveries for VOST analyses are summarized in Table 16. As noted in this
table, there were some poor surrogate recoveries that were not within specified
control limits. 'It is not possible to re-run.VOST samples once they have been
desorbed from their Tenax® cartridges. In reviewing the raw data, it can be noted that
several of the poor recoveries occurred for Test 3. This could have been caused by
chromatographic interference. Sample SAIp-390, 10-L duplicate, had very poor
recovery for all of the surrogates. This would indicate a potential problem with this
sample. Some samples' contained two surrogates that were not within the control
limits. Data from these samples may be in question due to poor desorption or
analytical interference. Samples that contain one surrogate out of the control
window have not been significantly affected. Although some surrogates were not
104
-------
Table 16. Surrogate Recoveries for VOST Analyses
" ''Compound
'« „ , '•-
1 ,2-Bromofluorobenzene
1 ,2-Dicihloroethana-d4
TolueniB-dBb
-• % \
•*toxfcif ;
8e*tilts !
•»
24
24
23
=• S««BH ^
•Recovery
W*
71
75
90
^Stfi
Oav
<%l
47
24
56
MIR
«6J
4
6
3
Max:
<%) '
139
107
300
No, :0f Bssetts ..
fititstda limits
tow.. ,.
8
2
4
High
0
0
1
1.*. ?"• *^ >. *. * >>>'~f ~~
•Cottfrol UnnrJts
' *%) --- %
50 - 150
50 - 150
50 • 150
• Numboir of field samples and field duplicates. (Does not include matrix spikes, blanks, and recovery checks.)
b Recovery of toluene-de was not calculable in one of the VOST samples due to interferences.
recovered well in some of the VOST tubes, it does not appear that this seriously
|
affects VOST data quality because the quantities detected in the VOST samples do
not contribute significantly to process emissions.
One set of method spikes (MS and MSD) was performed for the VOST analyses. This
was done by spiking a blank VOST tube and then performing the analysis as specified
by Method 5041. These results are summarized in Table 17. Accuracy and precision
objectives were not specified in the QAPP for these samples, however, the results
indicate the precision for this matrix was acceptable for each- compound except
toluene which had a RPD of 38 percent. Accuracy was acceptable for these spikes
with the exception of high- recoveries for benzene {about 135 percent) and low
chlorobenzene recoveries (60 and 51 percent).
Table 17. VOST Method Spike Results
' ,„ - 'I, ,Corop«ipd, % \~?'
"" """ - ' :' ,-. %* ' % %
Benzene
Chlorobeinzene
1,1-Dichloroethene
Toluene
Trichloroethene
* >"'" $0ilce-1
% Recovery- ff.^
138
60
103
117
109
$Pik* a
' '%--R*C€lYB*V - f ••"
' 133
51
98
80
111
-~ * RPJ>
y> ,^* •''«'« 4"" •£»"•:*>•
4
16
5
38
2
105
-------
Duplicate VOST samples were collected during Test 3. This was accomplished by
installing a second VOST into the exhaust duct. Samples were collected throughout
the duration of the test. The analytical results of these duplicates are presented in
Table 18. A total of six duplicate pairs were collected for this run. No specific
control limits for precision were placed in the QAPP for the VOST duplicates. The
precision for most compounds exceeded the generally accepted guidance of 30
percent RPD. Benzene appeared to have higher RPD values than most other
compounds. Specific reasons for this discrepancy are not clear. Values for methylene
chloride also exhibit high RPDs in some samples which were most likely due to
laboratory contamination. Although there was high variability associated with these
duplicate results, it does not appear that this seriously affected VOST data quality
because the quantities detected in the VOST samples did not contribute significantly
to process emissions as noted by the project conclusions. If, in the worst case, the
VOST values are off by a factor of 2, volatile emissions would still be considered
irrelevant and would not impact project objectives. In addition, no xylenes were
detected in the stack gas and therefore DRE calculations are not affected.
; '
Field blanks were collected on each run day. Field blanks (FB) were clean with the
exception of low levels of methylene chloride detected in the FB for Test 3. Eight mg
of 1,1',1-triehloroethane were detected in the Test 2 FB. Trip blanks were analyzed
for Test 1 and Test 2. No detectable levels of volatile contaminants were found in
these samples. These blank results did hot have any significant impact on data
quality. A complete presentation of all of the VOST data is presented in Appendix C.
I i
For two 10-L primary and duplicate samples (SAIC-0391 and SAIC-0392) and a 20-L
primary sample (SAIC-0130), only the Tenax®/charcoal tubes were analyzed because
the front-half cartridges broke during shipment. The back half cartridge is used if high
levels are encountered and breakthrough occurs in the first Tenax® tube. Since very
few compounds were detected at significant levels in any of the other samples, and
a total of 30 samples splitamong the 3 tests were analyzed, loss of data is expected
106
-------
Table 18. VOST Duplicate Sample Results (Test 3)
;tese»ijfltjan
1" Pair
—
21* Pair
-
4* Pair
K^y,-;-*-- ,;-::
Benzene
Chloroform
Methylene Chloride
Toluene
1,1,1 -Trichloroethane
Benzene
Chloroform
Methylene 'Chloride
Toluene
1,1,1 -Trichloroethane
Benzene
Chloroform
Methylene Chloride
Toluene
1 , 1 . 1 -Trichloroethane .
Vinyl Chloride
Benzene
Chloroform
Ethyl Benzene
Methylene Chloride
Tetrachloroethene
Toluene
1,1,1 -Trichloroethane
Trichloroethene
Xylenes
••Primary v>
3.36
0.23
.
0.52
0.20
4.91
3.11
0.24
•
0.12
9.94
1.04
•
.
0.23
3.13
4.05
2.78
•
1.09
•
0.50
0.39
*
9
' '
:.. Duplicate
2.48
0.26
0.35
•
0.17
9.86
4.O1
0.33
0.70
0.11
7.59
, 0.70
0.38
1.32
0.22
*
36.47 ;
6.26
1.14
0.52
. 0.28
6.84
0.22
0.36
1.28
"',-KrtU
30
12
NC
NC.
16
67
25
32
NC
9
27
39
NC
NC
4
160
77
NC
71
NC
173
56
NC
NC jj
(Continued)
107
-------
Table 18. (Continued)
> r^JS
'»~**?^$
5* Pair
6*1 Pair
*$$^&8^r V-x , '<* , yf '"..•. "
-•**Ss XKft-&\cA •*" S% •* v £.-.•• s\ x ^ v •" f •• J
^^?^*"<3mjw«xf ^' '"' ^ *
C^Xk-.^ i' t£>'" 1'<- * ' ; , '- ," ' "-
Benzene
Carbon Tetrachloride
Chloroform ',
Mathyione Chloride
Toluene ;
1,1.1 -Trichloroethane
Benzene
Chloroform
Methylene Chloride
Toluene
, Primary -^
' -V<#aks-;
, - pi&V - -, ,
25.44
e
8.50
1.02
3.09
0.51
116.31
6.43
3.57
s:97
Duplicate! ^,
, " v^iw r ^
, 0j?bV % >
37.39
0.32
9.48
2.S7
2.99
0.44
23.59
3.93
7.49
4.91
vir-Vv.,^ y
^iW \
/•"i ^ '« ^
38
NC
11
98
3
15
133
48
71
35
* • Indicates that this compound was not detected.
NC - Not calculated
to be minimal even though these 3 samples
samples does not impact project completeness
Volatile Organic Gas Canister Samples
were not analyzed. Loss of these three
objectives.
As back-up stack gas sample for volatile organies, gas canister samples were collected
in the event that the VOST tubes became saturated. These canisters were collected
in accordance with EPA Compendium Method TO-14. There were very few
compounds that exceeded the instrument calibration during the VOST analyses
(benzene and chloromethane in two samples). None of the detected compounds
saturated the GC/MS detector. A preliminary review of the VOST data indicated that
the analysis of canisters was not required and therefore, as stated in the QAPP these
analyses were not performed.
108
-------
Semivolatile Orqanics
Semivolatile organic samples were collected from several locations for each of the
Demonstration Tests. The semivolatile analyses for the feed soil and the stack gas
were designated as critical parameters for this demonstration and were used to
determine DREs. All other matrices were secondary critical measurements..
Semivolatile samples were collected for the feed soil, treated soil, scrubber liquor
before and after each test, the scrubber make-up, and the stack gas. Below is a
discussion of each of these analyses.
Feed Soil Semivolatiles
• Samples of the feed soil were collected from each 5-gallon pail that was fed to the
unit during each of the D.emonstration Tests, A small metal scoop was used to collect
approximately 100 g of soil from several locations within each of the 5-gallon palls.
This sample was placed into a glass jar for compositing. After the addition of each
scoop, the jar was agitated by hand to facilitate mixing. At the end of the
Demonstration Tests, appropriate portions of the composite sample were sent to the
laboratory for semivolatile analyses. To ensure that a representative sample was
studied, 60 grams of the composite sample was extracted using SW-846 Method
3550. Since the samples contained high levels of diesel and hexachlorobenzene,
special care was required to bring the target analytes to within the detection limits of
the instrument without exceeding the calibration range.. Sample extracts were
"cleaned" to remove possible interferences- of the diesel fuel by gel permeation
chromatography (GPC) and neutral alumina prior to analysis. Because of the high
levels of target compounds in the samples, surrogate spiking was performed at similar
levels with three surrogate compounds: 2-fluorobiphenyl, methylnaphthalene-d10, and
nitrobenzene-dg. These surrogates were selected because of their chemical similarities
to the compounds of interest in the feed. In addition, consideration as to the amount
of spike needed and the high concentration required, was also a factor in selecting
109
-------
. Tho usual SW-846 Method 8270 surrogates added
these project specific surrogates. The usual SW ** „ ot tnese
at typical concentrations would not sat,s1y pro,ect QC oblect,ves
three surrogates was approved prior to the demonstration.
Rowing
8270.
extraction, the extracts were analysed in
used to provide all semivolatile results for this study.;
for
established laboratory control limits.
110
-------
Table 19. Semivoiatile Feed Sample Surrogate Recoveriec
,*» ;
•Ow,
'**
o Wd out^too ml M4 *4*c«>«a. IM induing M«*s
appears that the quantitations are about 1.5 times higher than the average of the
other feed samples. These data, therefore, were not used for the calculation of DREs.
It appears that the incorrect amount of internal standard was added to this extract
causing mjs-quantitation of detected compounds. No other limitations for these data
are indicated by these surrogate recoveries. : .
Because of the high levels of hexachlorobenzene and diesel fuel oil in the feed
samples, it was not possible to perform a matrix spike using normal procedures
outlined In SW-846 Method 8270. It was not feasible to spike all of the matrix spike
compounds at levels high enough to be detected after diluting the extract. Since
hexachlorobenzene was one of the semivolatile compounds designated as critical in
the QAPP for these Demonstration Tests, it was spiked (at 1,000 ppm) into Silver
Bow Creek tailing soil that was previously mixed with No. 2 diesel oil and spiked-with
zinc oxide but no hexachlorobenzene. The matrix spike results for the feed soil are
presented in Table 20. As noted by the data presented in this table, the results of
Table 20. Feed Soil Matrix Spike Results (Test 1)
MS ^
fcPO
SAIC- 159/1 60 MS
Hexachlorobenzene
123
116
120
5.9
Del. - 152
50
111
-------
these spikes are satisfactory. In addition to matrix and surrogate spikes, a smgle
sample was to be extracted twice to obtain extraction efficiency information on this
matrix. For the target ana.ytes. this double extraction was proposed because of the
££ complex matrix and because on,y hexachlorobenzene cou,d feasib.y be sp,ke
,„ o L feed soil.' While this procedure was performed by the Moratory, both
extracts were inadvertent* combined and run as a single sample rather than analyzed
separately. However, because most of the hexachlorobenzene spike was recovered
in the feed samples, and because al, samples had sim«ar ,eve,s of hexachlorobenzene
w,th very ,itt.e difference in concentration, the lack of these additiona, extracts
results has not greatly impacted data quality.
Replicate samples were collected during each test run. For each test, the QAPP
specified that three primary and two duplicate samples be collected. Since the same
feed stock of material was used for each test, and each sample was aa aliquot of the
•feed comfiosite, if was not possible to distinguish between a specific primary and
duplicate sample. Selecting a primary and a random duplicate sample from a spec,,c
run could bias precision and accuracy information. Results for the.feed soil samples
are presented in Table 21. Data presented in this table.is evaluated two ways. The
first part presents test-specific information pertaining to each of the five aliquots from
the composite feed sample for each test. The summary information provides an
evaluation of the statistics of all three Demonstration Tests. These results ind.cate
good agreement among the detected compounds. This indicates that the soil was
well-mixed and homogeneous samples were collected.
Field blanks collected on each run day were Clean. The results from the laboratory
extraction blanks were also clean. Data quality was therefore not impacted by
contamination introduced in the field or the laboratory for these samples. A Complete
presentation of these sample results is presented in Appendix D.
112
-------
Table 21'. Duplicate Semivolatile Feed Sample Results
- ^ Compound '"'"
Sample 1
Cus&Br
$fljnp}e 2 ''
fc/gft*r) --
Sample 3
0>S&0
Sajnpte 4
<^tfkfl>
Sample £
"fygfcgr^
Mean
•. -,
Sttf,
~0«v<%"
BSD
-•- "^ +-•%"«?
Test 1
Acenaphthene
Oibenzofuran
Fluorene
Haxachlorobenzena
2-Metrayinaphthalene
Naphthalene .
Phananthrene
Dibenzofuran
Fluorene
Hexachlorobenzene
2-Methylnaphthalene
Naphthalene
Phenanthrene
•
*
•
850
470
140
60 J
•
*
•
750
480
120
60 J
16 J
•
36 J
990
440
130
65 J
•
•
•
900
380
110
55 J
Test 2
*
3
1100
420
140
60 J
Acenaphthene
Anthracene
Dibenzofuran
FJuorene
Hexachlorobenzene
2-Methyinaphthale.ne
Naphthalene
Phenanthrene
Suifcrftary™" ' J <•*••*•
Hexachlorobenzene
2-Mothyinaphthalene
Naphthalene
Phananthrene
17 ' J
9.7 J
16 J
43 J
980
490
170
72 J
•
• '
1000
400
140
60 J
14 J
36 J
950
420
1-60
65 J
Test 3
» •
* *
e «
as
a *
V ft
»0
a »
"\-'X Mteft ""'
927
429
141
63
•
•
15 J
45 J
1100
310
180
74 J
20 J
36 J
950
400
130
70 J
•
13 J
37 J
900
440
130
60 J
NC
NC
NC
878
442
126
60
NC
NC
NC
88 •
; 39
; 11
4
NC
NC
NC
1.14
8.82
9.05
5.89
*
«
•
800
370
120
60 J
«
*
16 J
46 J
930
550
170
68 J
Slsfldatd Deviation. „
107
59
21
6
•
*
12 J
30 J
780
430
130
54 J
NC
NC
960
402
138
63 •
NC
NC .
108
'• 20
15
6
NC
NC
11.29
5.10
10.75
3.78
NC
" NC
15
41
948
445
163
67
NC
; NC
; 2
7
133
102
22
.' 9
NC
NC
12.83
18.14
13.98
23.03
13.65
13.46
,"""% * ' ,"," »•- '• ,, ''
-"'-* B5DV" , >^ ,%'v,'
11.57
13.69
15.13
9.66
rial inis compouno was not oetecteu.
* * Sample results not used because of high surrogate recoveries as noted previously.
B Indicates that this compound was detected in a blank.
J Estimated Result. Indicates that this compound was detected below the quantitation limit. The quantitation limit is defined
as 5 times the detection limit.
NC Not calculated.
113
-------
Semivolatile Treated Soil
Once the molten mass had cooled, samples of the treated soil were collected for
semivolatile analyses. Samples were obtained by using a drill core as described in
Section 5 and earlier in this section. These samples were sent to the laboratory for
further size reduction and extraction using SW-846 Method 3540. In addition to the
core samples, water used to cool the drill bit was sampled and analyzed with the
treated soil samples. These cooling water samples were collected to measure any
losses of semivolatiie organic material due to the drilling. These liquid samples were
f
extracted using SW-846 Method 3520. No modifications were required to either
extraction procedure because it was anticipated that these samples would contain
little or no semivolatiie organic material. Each sample was spiked with the normal
surrogates required for SW-846 Method 8270. All of the samples were analyzed
using SW-846 Method 8270. The quality control procedures; tuning, calibration, etc,
utilized for these extracts are described later in this section.
» " *
Table 22 presents the surrogate results for these samples. Included in this table are
the surrogate results of the cooling water samples. The results of the surrogates are
excellent. All of the recoveries are within the specified limits with the exception of
Table 22. Semivolatiie Treated Soil Surrogate Recoveries
-stft-
Max. ' .
No; of tloaulJs "
V*,
2-Fluorobiphenyl
9O
80
98
30
115
2-Fluorophenol
89
73
94
25
121
2,4,6-Tribromophonol
94
85
104
19
122
N!trobanzena-dt
89
77
98
23
120
Phenol-d(
91
75
97
24
113
Terphonyl-d1*
103
15
77 121
18
137
Number of routine field samples and field duplicates, not including blanks, matrix spikes, and recovery checks.
114
-------
one high recovery of terphenyl-d14 (188%) in the cooling water sample that was
collected before the water was subjected to drilling.
Results of the matrix spikes performed on the treated soil are presented in Table 23.
As indicated by these results, there does not appear to be a recovery problem
i
associated with this matrix.
Table 23. Semivolatile Treated Soil Matrix Spikes
-. ,:
* ;#*w;^
SAIC-0220 MS
SAIC-0220 MS
SAIC-0220 MS
SAIC-0220 MS
SAIC-0220 MS
SAIC-0220 MS
SAIC-0220 MS
SAIC-0220 MS
SAIC-0220 MS
SAIC-0220 MS
SAIC-0220 MS
SAIC-0221 MS
SAIC-0221 MS
SAIC-0221 MS
SAIC-0221 MS
SAIC-0221 MS
SAIC-0221 MS
SAIC-0221 MS
SAIC-0221 MS
SAIC-0221 MS
SAIC-0221 MS
SAIC-0221 MS
<"••''• ,
1 .2,4-Trichlorobenzene
1 ,4-Dichlorobenzene
2-Chlorophenol
2,4-Dinitrotoluene
4-Chloro-3-methylphenol
4-Nitrophenol
Acenaphthene
N-Nitrosodipropylamine
Pentachlorophenol
Phenol
Pyrene
1 , 2,4-Trichlorobenzene
1 ,4-Dichlorobenzene
2-Chlorophenol
2,4-Dinitrotoluene
4-Chloro-3-methylpheno!
4-Nitrophenol
Acenaphthene
N-Nitrosodipropylamine
Pentachlorophenol
Phenol
Pyrene
'' % ^ v ^
. X»>v>^V^
97
82
99
98
101
78
88
89
84
87
101
94
82
96
107
98
84
90
86
85
88
100
,3*&w&
«;,
103
83
94
93
. 86
92
93
76
98
- 84
112
85
70
84
105
99
90
89
88
87
79
104
S-> f
Mean.
100
83
97
96
94
85
91
83.
91
86
107
90
76
90
106
99
87
90
87-
86
84
102
^
;;^;
6.0
1.2
5.2
5.2
16.0
16.5
5.5
. 15.8
15.4
3.5
10.3
10.1
15.8
13.3
1.9
1.0
6.9
1.1
2.3
2.3
10.8-
3.9
V '" fife"
"fXSf
38-107
28-104
25-102
• 28-89
26-103
11-114
31-137
41-126
17-109
26-90
35-142 .
38-107
28-104
25-102
28-89
26-103
11-114
31-137
41-126
17-109
26-90
35-142
%
;>££«<:
23
27
50
47
33
50
19
.38
47
35
36
: 23
27
50
47
33
50
19 .
38
47
35
36
115
-------
Duplicate samples were collected and analyzed for each test run. Duplicate results
for the semivolatile treated soil samples are not presented here because there were
no detectable levels of semivolatile target compounds present in the primary or
duplicate samples. There were some low level phthalate compounds detected, but
they were qualified as laboratory blank contaminants. The core water.samples also
had some low level phthalate contamination which were suspected to be from
laboratory contamination with no other target compounds detected. The core water
samples indicate that there was no loss of semivolatile target compounds during the
drilling process.
Semivolatile Scrubber Water Samples
Samples of the scrubber water were collected before and after each Demonstration
Test. These samples were collected to evaluate any accumulation of semivolatiles in
the scrubber liquor and were considered secondary critical measurements. These
samples were collected from a tap in the scrubber sump using prescribed procedures.
These samples were extracted using SW-846 Method 3520. The extracts were
analyzed under instrument conditions outlined later in this section.
Pre-demonstration testing indicated that the generally accepted surrogates for SW-846
Method 8270 were not recovered properly in the scrubber liquor samples. It was
postulated that high levels of NO2 in the exhaust gas was nitrating the acid
surrogates, resulting in poor recoveries of the acid spike compounds. To compensate
for these poor recoveries, three additional acid surrogates were selected for spiking
in this matrix and in the MM5 resin samples: 13C-pentaehlorophenol,.* 2,4-
dinitrophenol-d3, and 4,6-dinitro-2-methylphenol-d2. These compounds were selected
and pre-approved because they exhibited satisfactory recoveries in pre-demonstration
test activities and because they were highly substituted phenolic compounds which
were less subject to nitration. Recovery criteria that was placed on these compounds
are based on the non-labeled analogs recovery criteria presented in SW-846 Method
116
-------
8270. The surrogate recoveries are presented in Table 24 for the scrubber water
semivolatile samples. It should be- noted that the soil matrix originally proposed for
this demonstration contained significant quantities of pentachlorophenol, anticipating
the use of this compound for determining DREs. Because a different soil matrix was
ultimately chosen, and the primary compound of interest was hexachlorobenzene, acid
surrogate recoveries were consequently less important.
Reviewing the data presented in this table indicates that the base/neutral surrogates
were recovered very well from this matrix. These were critical for project conclusions
when evaluating DREs for hexachlorobenzene and 2-Methylnaphthalene. Recoveries
for the base/neutral surrogates that were outside control limits exceeded the recovery
range. There is, however, a wide scatter of results for the acid surrogates. As
anticipated, the routine 8270 surrogates were not recovered well in this matrix, with
the exception of 2,4,6-tribromophenol. The most probable reason for this is that the
large bromine atoms prevent the NOX from attacking the phenolic \ ring of this
molecule. Other acid surrogates showed marginal to good recoveries. Recovery
Table 24. Scrubber Water Semivolatile Surrogate Recoveries
""'<; ^CBnvoSrtr' V> ^
"\ ,'f $&>- -"-x, ' ^~^--#^\\ * '"
••'•. ' *ff' ''£„ ",?; '-'. , ',
•v^A w, f~' -""S"? j^-- "ffS* T^
2-Fluorobipheny)
2-Fluorophenol
2,4-Dinitrophenol-d3
2,4,6-Tribromophenol
4,6-Dinitro-2-m8thy)ph8nol-d,
"C-Pentiichlorophenol
Nitrob6n;:ene-d5
Phenol-d,;
Terphenyl-dM
•• XV ^ '"
:*io.*t
^fresuf*?*-"
% -c .• ^
'#^»x^;
% I % % % % <•
12
12
12
12
12
12
12
12
12
Mean
%ft«(»vaty
' «?S.
74
16
120
38
85
56
91
42
110
'St&,%
I^BV,.
{%}m
14
29
70
37
41
35
15
28
20
Mn,
'*W'
% •. %
**V >
52
0
0
0
0
0
75
0
92
Max.
mr
f<,ffff \
•.
95
83
223
103
117
95
130
84
155
No., oi Results
^'0&St8
-------
check samples and method blanks indicated satisfactory recoveries of the special acid
surrogates and the normal acid surrogates. These results indicate that acid
compounds may not have been accurately quantitated or detected in the scrubber
water matrix. The scrubber sump is maintained at a high pH with sodium hydroxide
to neutralize acid gases. The results indicate that the scrubber solution, along with
the NO2 gases produced by the process, have an impact on acid compounds. As
indicated by the results there were very few detectable SW-846 Method 8270
compounds found in the scrubber water samples and these data were not used when
evaluating critical project conclusions. As with volatile data from the scrubber liquor
these data were used in conjunction with emission data to evaluate emission system
efficiency.
Matrix spikes were not performed on this matrix because an insufficient amount of
sample was sent to the laboratory. This has limited impact on data quality since most
of these samples contained no significant levels of semivolatile compounds and were
not used in calculating DREs to satisfy the project objectives.
Duplicate samples were collected for each Demonstration Test at the beginning and
/ •
end of each run. Table 25 summarizes the results of the duplicates. For the pre-test
samples, there were very few compounds detected. Most of the values were flagged
with a "J" qualifier indicating that they are estimated values below the quantitation
limit. Because values were detected below the standard laboratory quantitation limits,
agreement among" duplicates should not be expected to meet QAPP specifications.
However^ there is good agreement between some of the detected quantities as
indicated by the RPD values.
The duplicate sample for Test 1 (SAIC #106) was lost during the laboratory extraction
procedure. Since there were no significant quantities of target compounds detected
in the post-test samples, the loss of this sample has no impact on data quality. As
with the pre-test samples, the post-test samples contained very few target
118
-------
compounds. Most of these values were also flagged with the "J" qualifier indicating
that these values are estimated. Although some nitrophenolic compounds wore
detected in one of the two duplicate samples, this does not appear to be a matter of
concern because they do not impact project objectives. Since these quantities are
close to the detection limits, it is not surprising that they appear in one sample and
not the other. ;
Table 25. Scrubber Water Semivolatile Duplicate Results
f- ' : ' ' , -' ••
..X" ' 'V''1"V> 'V'' ''••'" •-,,. V ' X*
-' ,-,> / ^ Comp«uttdn»f¥ ','?", "
r '• ^^tf«'^ >J;
T«8t1
*'••&>,>•-, -ffff '< .sr •• ' •.
* ihipitea**
,^>^ Nii»0fiJ»,v" :,
•X v-:- ^ .. ' ^
i.V^^
- ' ^6$i«e6v»» Xs "
«*&> WDJ •*?-'•'"
POST-TEST
bi£;{2-Ethylhaxyl)phthalate
Nitrobenzene
2-Nitrophenol
2.6 JB
9.3 J
«
2.3 J
6.0 J
5.9 J
12
43
NC
30
30
30
Test 2
PRE-TEST . ' • :
bis(2-EthylhexyI)phthalate
2-Methylnaphthalene
Naphthalene
1.8 J
2.0 J
3.0 J
3.0 J
2.0 J
3.2 J
50
' O
6
- 30
30
3O
POST-TEST
4-Aminobiphenyl
2,4-Oinitrophenol
.bista-EthylhexyDphthelate
Nitrobenzene
2-Nitrophenol
e
25 J
2.9 J
5.0 J
9.7 J
3.0 J
ft
*
4.8 J
«
NC
NC
NC
4
NC
30
30
: 30
30
30
Test 3
PRE-TEKT
Benzole Acid
10J
POST-TEST
Naphthalene
1.2J
8.3 J
1.2J
19
0
30
30
* Indicates that this compound was not detected.
B Indicates that this compound was detected in a blank.
J Estimated Result. Indicates that this compound was detected below the quantitation limit. The quantitation limit is defined
as 5 times the detection limit.
NC Not calculated.
119
-------
The field blank (SAIC #280) was lost during the extraction procedure. Based upon
sample results, which showed no significant concentrations of target analytes, the
loss of this sample has no impact on overall data quality.
Semivolatile Emissions
Samples of the stack gas were collected for all three Demonstration Tests to
determine the types and quantities of semivolatile emissions and to provide
information for ORE calculations and, therefore, these measurements were considered
critical. Samples were collected using an EPA Modified Method 5 (MM5) sampling
train with XAD-2 resin. Specifics of this procedure are presented in Section 5. After
each Demonstration Test, the sample trains were broken down and thoroughly rinsed
to remove any semivolatile analytes from the filter housing, probe, and nozzle (PNR).
The solutions were recovered from the impingers and sent to the laboratory for
analysis. When the samples arrived in the laboratory they were composited and
extracted using SW-846 Method 3540 for the XAD and 3510 for the impinger
solutions and PNRs. Samples were spiked with the normal SW-846 Method 8270
surrogates and the special acid surrogates as noted above. These surrogates were
added to the MM5 samples because of the hjgh levels of NOX anticipated in the stack.
A summary of the surrogate recoveries for the semivolatile emission samples is
presented in Table 26. As noted in the table, there were several surrogates that did
not meet the specified recovery criteria. Some of the surrogate recovery information
x i .
was lost because of dilutions that were required to bring target compounds into the
calibration range, specifically benzoic acid. However, some samples analyzed without
dilutions had surrogates that were not detected (recoveries of 0 percent). As
previously noted with the scrubber liquor samples, normal 8270 surrogates were not
expected to be detected due to nitration effects caused by high NOX concentrations.
This is why other acid surrogates were added to the matrix. Specifically, substituted
phenols with large steric hinderences were chosen to prevent nitration. One method
blank had poor recovery of 2,4-dinitrophenbl-d3, (3 percent) which may indicate that
120
-------
Table 26. Semivolatile Emission Sample Surrogate Recoveries
Compound
f *" '" ' -, "*
2-Huorobiphenyl
2-Fluorophenol
2,4-Dinitrophenol-d3
2,4,6-1'ribromophenol
4,6-Dinitro-2-methylphenol-dj
"C-PemtaehlorophenoI
Nitrobenzene-dfi
Phenol- d6
Terphenyl-du
ResultB* .
3"
1"
4
1"
4
4
4
1"
4
ft/lean
137
0
106
0
115
23
91
. 2
105
*,./
(%1
45
NC
98
NC
.20
29
23
NC
28
Mm
•(%)
103
0
0
0
92
0
63
2
72
Max *
-
188
0
234
0
140
61
120
2
140
No. of Results
Outside tMts
tow
0
1
1
1
0
2
0
1
0
«<8V
2
0
1
0
0
0
0
0
1
^ tiroits
30 - 115
25 - 121
D - 191
19 - 122
D - 181
14 - 176
,23 -. 120
24 - 113
18 - 137
• Number in parentheses is number of routine field samples, including duplicates analyzed. Blanks, matrix spikes, and
recovery checks are not included.
b Number of results less than number of samples because surrogates were diluted out in some samples.
NC - Not calculated.
P - Deitected. •
there was a problem with this surrogate standard. Low recoveries, 27 and 18
percent, were also noted in the recovery check samples. The carbon-labeled
pentachlorophenol surrogate was not recovered in the Test 1 or Test 1 duplicate
samples, due to a dilution effect. However, this surrogate was recovered within
acceptable limits for every other sample. Recovery of acidic compounds in this matrix
were, therefore, considered to be acceptable, but were not considered critical for
evalueiting project objectives. The base/neutral surrogates were recovered reasonably
well with the exception of some high and low recoveries of 2-fluorobiphenyl. The
specific reason for these QC outliers has not been determined but does not impact
project objectives because all other base/neutral surrogates were within QC limits.
A second sample train was installed into the stack for Test 1 to collect a duplicate
semivolatile stack sample. The results of this duplicate pair is presented in Table 27.
Most of the targeted compounds detected in these two runs are flagged with the ™J"
qualifier. Results that are above the method quantitation limits show good agreement,
with the exception of the phthalate compounds. Phthalates are commonly reported
121
-------
r '
low levit:presence^ oabsence,
" '• ' '
•appear^'
to.tH|ve impactedidata .quality s|
"11
The field blank collected for these samples was clean with the exception of some Sow
level phthalates and 120 tig of naphthalene. It is important to note that this is
approximately the same level of naphthalene that was detected in all of the process
samples. ".Therefore, the levels of naphthalene that were detected in the emission
t samples vfere a tresulrbf field blank contamination and weretvdisregarded
\ evaluating the serrjivolatile emission dafca from these^ests.
Metals
Several parameters were tested for metal content at various sampling locations 4
•throughout the system. Metals were denoted as a secondary critical parameter^ r
Samples for metals analysis were collected from the feed soil, the treated soil, the
Table 27. Duplicate Semivolatile Emission Sample Results
•> % v-^f '' *"^ ' *"
^^r^'S^^v:
Acfltophsnon*^ £
'3.
Baruoic Acid •*. ' , '-.•
J"r
Dibutylphthalate -"
m- '.f
Ci*thyiphthalate ":
2.4-O!nitrophanoI
bt»(2-Ethv)hoxYl)phthnl«to
Naphthalene
2-Nitrophonol ^_
4-Nitrophenol "••-
"^\ ^ '"frtewry J
""k-^pbWjM
•*« '• •
•**• I
23SJ.-10
* _, ;
o ."•
17.B2 J
28.22 B
10.69 :.
t. •• -" -
12.S6 J -.
'^ SuptSeate-^ "' ,
"*, -%pf»V)': ,- ,N
'4 2.57 J
k 242^6
0.51 J
49.51
•••
0.82 JB
7.90
9.24 J
* ° -^
' RPB-' '!' ' '^
.. -- -^"r"R*i ^-j
t . *IC '~~
!•- '5 . -»•
•':• NC ft
,;NC
NC
189
30
*JC
NC
0 Indicates that this compound was not datected.
B Indicates that this compound was detected in a blank. f
J Estimated Result. Indicates that this compound was detected below the quantitation limit. The quantitation limit is dofined
as 5 times the detection limit.
NC Not calculated.
122
-------
^ .scrubber Jicfyof before and after the test, the scrubber makeup and the stack gas. For v-
-£- ' ,. ' "~S. ' • '- ~ .' . ' . •":
ithese,matrices, most of metals data was generated using SW-846 Method?601Q, ion.
coupled plasma (ICP). A few metals could not be analyzed using this method.
Therefore, mercury analysis was performed using SW-846 Method 7470/7471, cold
vapor; arsenic by SW-846 Method 7060, atomic absorption; and selenium by SW-846
Method 7740, atomic absorption. To insure that data of known quality was obtained,
the instruments were calibrated in accordance with procedures outlined in the
. appropriate methods and/or the QAPP. ~
For the ICP, a mixed standard calibration was performed each day using a multipoint
calibration curve. For this calibration to be acceptable, the measured value of the high
calibration standard had to be within ±10% of the true value. Once the calibration
met. this acceptance criteria, a calibration check was performed at a frequency of
10%. Reagent blanks were also analyzed at a 10% frequency and all target metals
had to be less,than five times the method detection limit. Matrix spike and matrix
spike duplicates were analyzed for each.sample matrix. Duplicate samples were also
collected in the field and analyzed for metals. ICP interference checks were performed
at "the beginning, middle, and end of each analysis. These-checks required a 80 to
120% recovery of the true value for EPA check sample elements.
For the atomic absorption and the cold vapor analyses, similar analytical constraints
were placed on instrument performance. Before analyzing samples, a multipoint
calibration was performed. Calibrations checks were performed after each multipoint
calibration and after every 10 samples. Reagent blanks were also analyzed at a 10%
frequency. The results of these blanks required that the target analyte was less than .
five times the method detection limit. Matrix spikes and matrix spike duplicates were
performed for each matrix. As with the other metal parameters, field duplicates were
collected and analyzed.
123
-------
Feed Soil Metals Ti*, ~ ' ;
« *1 -- • i ,
Samples were collected of the feed soil under similar conditions as described for the
feed soil semivolatile organics. At the end of the Demonstration Tests, samples were
collected from the composite jar and sent to the laboratory for metals determination
using SW-846 procedures specified in the QAPP. Several samples were collected
from the composite jar which were randomly collected aliquots of the same
composite. Five samples were collected for each test run. As with the semivolatiles,
it was not possible to distinguish between these aliquots by specifying random
primary and duplicate samples since they were all taken from the same jar. Therefore,
the results are presented together for each test. Table 28 presents the results for the
feed soil metals. As with the semivolatile samples, the data is presented two ways.
The first part presents test-specific information with the appropriate means and
standard deviations. The summary portion provides information on all of the test runs
since the soil used for each test was from the same lot. As demonstrated by the
results presented in Table 28, good agreement was achieved for each test along with.
good statistical results for all of the tests.
In order to allow a comparison of the feed soil and the treated soil results, the
laboratory used a modified digestion procedure which uses a combination of
hydrofluoric, hydrochloric, and nitric acids. This was done because of the difficulty
encountered with breaking down the treated soil matrix during pre-demonstration
activities. Since a combination of both acids was required for the digestion of the
treated soil, this procedure was also used for the feed soil so that the results could
be evaluated equivocally.
Matrix spikes were performed on the feed soil. Because of the high levels' of zinc
oxide in the feed soil, special considerations were made for the spiking of zinc in this
matrix. Two samples which did not contain any zinc oxide spike (but which had been
previously spiked with both diesel and hexachlorobenzene) were sent to the laboratory
124
-------
Table 28. Feed Son Metal Results
Compound •-
Arsenic
Barium
Cadmium
Calcium
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Mercury
Potassium
Sodium
Vanadium
Zinc
Aluminum
Arsenic:
Barium
Cadmium
Calcium
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Mercury
Potassium
Sample I s
%< \
3.96
12.18
8.13
NC
6.54
14.20
8.96
4.92
5.05
4.02
13.03
6.34
6.79
5.24
8.72
1.93
2.34
8.15
2.61
16.24
• 4.42
25.28
1 1 .74
4.39
3.74
6.20
13.65
6.58
2.33
125
-------
Table 28. (Continued)
0^*,
: O •»•%• •> v %
v Sample 3'
Sample 4 :
Santpte 5' :
'"?*::
w
Test 2 (Continued)
Vanadium
Zinc
Aluminum
Araam'c
Barium
Cadmium
Calcium
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Mercury
Potassium
Sodium
Vanadium
Zinc
Summary ,\
Aluminum
Arsenic
Barium
Cadmium
Calcium
Chromium •
Copper
Iron
Lead
10000
83
22000
9800.
84
21000
10000
69
2700O
10000
73
23000
10000
88
27000
Test 3
49000
200
500
.
12000
29
580
36000
410
4300
700
1.1
. 19000
11000
81
21000
50000
200
490
e
12000
-27
560
36000 "
430
4600
780
1.3
19000
10000
. 80
23000
;B^%l*$£fe^fe,
49600
203
507
6
12467
24
587
36867
427
49000
200
480
5.5
12000
28
620
35000
430
4300
740
1.2
19000
' 10000
79
23000
V*st«xi8td
48000
260
510
*
12000
24
7SO
38000
420
4400
, 800
• 1.2
20000
10000
83
22000
ftwfofeirj:
2798
25
43
1
640
:4.
68
1727
27
52000
220
520
•
13000
26
600
38000
410
4800
660
1.2
19000
11000
87
21000
9960
79
2400O
89
8
2828
% Rsr> ,"
-6* / , .. v
0.90
10.10
11.79
"
49600
216
500
SMC
12200
27
624
36600
420
4480
736
1.2
19200
10400
82
22000
1S17
26
16
NC
447
2
79
1342
10
217
57
0.07
447
548
3
1000
3.06
12.07
3.16
NC
3.67
7.18
12.70
3.67
2.38
4.84
7.78
5.89
2.33
5.27
3.86
4.55
", t ^V-Vfcs^"^/-*' *''"-
5.64
12.13
- 8.49
14.27
5.13
17.24
11.65
4.68
6.28
126
-------
Table 28. (Continued)
Summary ,
Magnesium
Manganese
Mercury
Potassium
Sodium
Vanadium
Zinc
JMeW
4667
. 755
1.3
19200
10153
79
23067
f ' * «r
' Standard Deviation ,
424
87
0.1
775
452
6
1831
•.-.-. v -. ..•. ^ v -.%
-v% % ,> --RSD *- ' - ,' "
3. OS
1 1 .49
7.98
4.03
4.45
7.97
7.94
specifically for the zinc spiking. The laboratory was also sent a portion of the zinc
oxide lot which was used for the spiking of the Demonstration Test soil. The
unspiked soil was spiked in the lab for MS and MSD analyses at the same level as the
zinc oxide feed soil spike. Results of these spikes are presented in Table 29. The
remaining metals were spiked into two actual feed soil samples. These results are
also summarized and presented in Table 29. As noted by the data presented in this
table, precision and accuracy objectives, as specified in the QAPP, were met for all
parameters except silver. Silver was recovered poorly in the MS and MSD. However
the precision was excellent for these recoveries. The •poor recovery results have no
impact on data quality since silver was not detected in any of the feed soil samples
and was not a critical analyte.
Field blanks were collected for the feed soil at the time of sampling. The results of
the field blanks indicated a small quantity of aluminum contamination for each of the
tests. The amount of this contamination was 670, 360, and 340 ppm for Tests 1,
2, and 3, respectively. For the worst case, highest contamination and lowest
aluminum feed soil concentration, this represents 1.5 percent of the measured value
detected in routine samples. Therefore, field blank contamination for aluminum has
little or no impact on data quality. Additionally, a small quantity (76 ppm) of iron was
detected in the Test 3 field blank. This has no impact on data quality since reported
iron concentrations were several orders of magnitude higher than the field blank.
127
-------
Table 29. Feed Sofl Metal (Matrix Spikes Results (Test 1)
Compou«j--;v
^ ^ v "° *" §* ^
* f '•'. ' * '
Nickel
Potassium
Selenium
Silver
Sodium
Thallium
Vanadium
Zinc'
'^r^***H*M*> "%•
70
98
109
94
92
94
94
96
98
70 Q
90
94
97
102
98
90
83
22 Q
90
98
' 94
• 98
1>s?x
71. Q
91
108
96
92
94
88
97
102
68 Q
94
94
100
84
97
110
81
22 Q
85
97
89
92
',. ' \/'
w$i*n.
71
95
109
95
92
94
91
97
100
69
92
94
99
93
98
100
82
22
88
98
92
95
* *i»^i
1
1.4
7.4
0.9
2.1
0.0
0.0
6.6
1.0
4.0
2.9
4.3
0.0
3.0
19.4 -
1.0
20.0
2.4
0.0
B.7
1.0
5.5
6.3
^•<""""^--&bl*
sassass' '" '. ISB±
65-140
65-140
65-140
65-140
65-t40
65-140
65-140
65- 1140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
75-120
^M"*'*
- / •#, ''•.'••.
-, Hce,cts1wv< A
v - ' y&Bl-.'- '^''
••"
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
Q Outside control limits.
• Zinc was not actually spiked into the Test 1 feed soil, but rather into a sample of feed soil previously spiked with diesel oil
and hexachlorobenzene but no zinc oxide.
Method blanks for metals contained consistent 50 and 220 ppm levels of copper and
iron, respectively. The copper contamination is of significance since this level
represents approximately 5 to 10 percent £>f the detected copper in the feed soil
samples. Since copper was not present in large enough quantities to effectively track
process performance, the blank contamination has little effect on project objectives.
Copper was not used in the evaluation of the technology. The iron present in the field
128
-------
blank and method blank totals approximately 1 to 2 percent of the feed soil contesnt
and does not have a significant impact on data quality. All feed soil metals data
appears in Appendix D.
Treated Soil Metals
a
Samples of the treated soil were collected after the melt had cooled and hardened.
Samples were collected in accordance with procedures outlined in Section 5 and
previously discussed in this section. To effectively evaluate the metal content of the
treated soil, the samples were crushed into a powder form and subjected to
microwave digestion. A mixture of hydrofluoric, hydrochloric, and nitric acids was
used to break down the glass monolithic matrix. The QAPP specified that a primary
and a duplicate sample be collected for each test run. Table 30 presents the results
of the treated soil metal duplicates for Tests 1, 2, and 3. As noted in the table,
precision objectives met the specified QAPP criteria for all metals with the exception
of arsenic which had a RPD of 38 percent in Test~2. Arsenic concentrations were
very low (less than 20 ppm) for all of the Demonstration Test results; hence, this QC
outlier does not have an impact on data quality.
• i
A matrix spike analysis and a matrix spike duplicate analysis were performed on a
treated soil sample.from Test 1. The results of these analyses are presented in Table
31. As noted in the table, all p.f the spiked compounds met precision and accuracy
objectives for all of the metals with the exception of silver, and one high recovery for
aluminum. The matrix spike duplicate for aluminum was recovered at 143% while
both the MS and MSD for silver were recovered at 16 and 17%, respectively. These
QC outliers do not have a significant impact on overall data quality for the treated soil
metals because neither silver nor aluminum were used in the evaluatioti of project
objectives.
129
-------
Table 30. Treated Soil Metal Duplicate Results
. •**$$:.&
Barium
Calcium
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Nickel
Potassium
Sodium
Vanadium
23 no
Aluminum
Arsenic
Barium
Calcium
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Nickel
Potassium
Sodium
Vanadium
?r^|^:v% ":
i
48000
13
530
27000
520
800
160000
120
6000
1900
290
16000
8200
61
6600
1
. 45000
13
460
21000
520
1700.
150000
130
4200
1900
260
16000
8700
60
?'•£&' -i
last 1
49000
14
510
27000
500
820
160000
95
5500
1900
280
15000
8600
61
6900
rest 2
47000
19
500
20000
500
1300
150000
100
5000
. 1900
270
16000
8600
59
.-,••* ,
2.06
. 7.41
3.85
0.00
3.92
2.47
0.00
23.26
8.70
0.00
3.51
6.45
4.76
0.00
4.44
4.35
37.50
8.33
S.OO
3.85
26.67
0.00
26.09
17.39
0.00
6.06
0.00
1.16
1.68
••.?ssr,.
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30 .
30
30
30
30
30
'30
30
30
• 30
*30
30
30
30
130
-------
Table 30. (Continued)
""%•.- ^:C«m$>ottBtf'> '" ' " "\
^Ptimafy
^.tongfca* ',,
'Duplicate
img/kfl) , ' / -
•ftPS - ",,
"• V.% •. * ' '
, objwfivsi /
^, ,. CRPO}"', ^
Test 2 (Continued)
Zinc
9300
8800
5.52
30
Test3
Aluminum
Arsenic
Barium
Calcium
Chromium
Copper
Iron
Lead
Magnesium
Manganese :
Nickel
Potassium
Sodium
Vanadium
Zinc ..
45000
11
480
18000
610
820
220000
120
5000
2600
290
15000
7800
49
• 8600
41000
13
420
18000
610
830
210000
120
4100
2500
280
14000
7800
52
9100
9.30
16.67
13.64
0.00
0.00
1.21
4.65
0.00
19.78
3.92.
3.51
• 6.90
0.00
5.94
5.6.5
30
30
30
30
30
30
30
30
30
30
30
30-
30 -
30
30
The laboratory method blanks performed along with the analyses of these samples
were all clean. Analysis of the drill core cooling water sample, which was collected
to measure any losses during sampling, contained very low levels of metals. All
metals detected in this sample were less than 10 ppm with the .exception of the
following: calcium, 18 ppm; potassium, 62 ppm; and sodium, 40 ppm. These results
indicate that there were no significant losses of metal compounds from the drilling
• • *
process because the amounts found in these samples were several orders of
magnitude lower than routine analyses of the treated soil samples. All of the treated
soil data may be found in Appendix E.
131
-------
Table 31. Treated SoO Metal Matrix Spikes Results (Test 1)
xi,, ^' v,"
Compotmd, ,v "f,
•> s v -^
, i ,*$^
Aluminum
Antimony
Arsenic
Barium
Beryllium
Csdmiurn
Calcium
Chromium
Copper
Iron
Lead
Magnasium
Manganese
Nickel
Potassium
Selenium
Silver
Sodium
Thallium
Vanadium
Zinc
M^lC^^oov^ss """- - , ,' , 7 '-
" -t$"?'**^$*.
-f^IlK
132
94
107
• 101
90
88
86
' 94
94
93
95
98
99
93
120
86
16Q
95
92
94
88
'TO V..
r ^MSD ;^,j
_ V •, :
143 Q
94
111
103
90
90
96
95
96
89
. 98
96
100
93
126
84
17 Q
96
90
95
89
* Stean V "•
138
94
109
102
90
89
91
95
95
91
97
97
100
93
123
85
17
96
91
95
89
*'\\
' "J5P&
,- v"/ * '
•/ f f J-"-
8.0
• o.o
3.7
2.0
0,0
2.2
11.0
1.1
2.1
4.4
3.1
2.1
1.0
0.0
4.9
2.4
6.1
1.0
2.2
1.1
1-1
-> , "^^-^lObJBfitfvea1; l,,""^\
AccuracV ' '••
t%^««s.r- :
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
.65-140
65-140
65-140
65-140
65-140
75-120
Pwsi^Qfl;.. "
, , -iRPOJ'--^^
30
. 30 '
30
30
30
30
30
30
30
30
30
30
. ' 30
30
30
30 •
30
30
30
30
30
Q - Outside control limits.
Scrubber Water Metal Samples
Samples of the scrubber water were collected before and after each Demonstration
Test. The samples were tested for metal content by using procedures outlined in SW-
846. Most metal results were obtained by digesting the samples using SW-846
Method 3010 and analyzing the digestate by SW-846 Method 6010, ICP. Other
132
-------
metals (arsenic, mercury, and selenium) were evaluated using procedures outlined in
SW-846 Methods 7060, 7470, and 7740, respectively.
One primary and one duplicate sample were collected at the beginning and the end of
each demonstration test from the scrubber sump. Sampling procedures are described
in Section 5. Table 32 presents the results of the scrubber water duplicate samples.
Values were reported in this table only if they appeared in both samples. In general,
most of the QAPP objectives for precision were met. With this much data, it is not
surprising that some of the metals do not meet the objectives. The post-test
duplicates for Test 2 show the greatest amount of variability and several metals do
not meet the objectives. The reason for this is unclear. This may indicate that the
contents of the scrubber after this test were not well-mixed when the samples were
collected or that the sample line was not purged sufficiently. In general, these QC
results are good and do not imply that data quality has been impacted for the scrubber
water metal samples. In addition, these data have a minimum impact on the project
because scrubber water results are only indirectly used for evaluation, of the
technology and are not used for ORE or teachability claims.
Matrix spikes were not performed on this matrix for metals with the exception of
arsenic. Arsenic was recovered high at values of 152 and 174 percent recovery. The
matrix spikes for the additional metals was not performed. Since there were several
duplicates collected for this matrix, and since only low levels were detected, the
impact of not having spiked sample results is limited and does not seriously affect
data quality. These data were not used in determining previously defined numerical
objectives and, overall; are not critical in evaluating process performance.
Laboratory method blanks that were analyzed along with the scrubber liquor samples
were free of contamination and therefore do not impact data quality.
133
-------
Table 32. Scrubber Water Sample Metal Duplicate Results
,. Ccwpo^;t:J^
if! ij£l^ns»y- ;H-t-;% -
^'f^'^ctrngSJ -.-• <-^ ?
>*-"<* ftoftlltfisHf \* *"
* -,,, Jmgfljf , " .,-
^^*tt>- :,j
•. * ' ^ r
"<' Objftfcjvw'%0-
' -fRPD* ^ '
Test 1
PRE-TEST
Coppor
Iron
Lead
Magnesium
Manganese
Nickel
Potassium
Sslenlum
Sodium
Vanadium
Zinc
0.45
0.021
0.011
13
0.45
0.91
13
O.053
4.1
0.12
0.13
14
0.0099
350
0.050
0.61
0.50
0.024
0.01 1
12
0.57
1.00
16
0.054
4.1
0.14
0.18
14
0.0089 •
I 36©
0.053
0.77
10.53
13.33
0.00
8.00
23.53
9.42
20.69
1.87
0.00
15.38
32.26
O.OO
10.64
2.S2
S.S3
23.19
20
20
20
20
20
20
20
20
20
20
20
20'
20
• 20
20
20
POST-TEST
Aluminum
Arsenic
Barium
Cadmium
Calcium
. Chromium
Copper
Iron
Lead
Magnesium
Manganese
Mercury
3.0
4.9
0.17
0.085
9.8
0.99
4.1
34
2.6
3.2
0.61
O.O88
2.9
s.s
0.17
0.093
10.0
1.00
4.2
34 .
2.7
3.3
0.63
0.032
3.39
11. §4
0.00
8.99
2.02
1.01
2.41
0.00
3.77
3. OS
3.23
93.33
20
20
20
20
20
20
20
20
20
20
20
20
~™~ "IcontJ
134!
-------
Table 32. (Continued)
Conupotfnd ' ':
•f-. •• •,; ,
POST-TEST (CONTINUED)
Nickel
Potassium
, Selenium
Sodium
Vanadium
Zinc
PRE-TEST
Aluminum
Arsenic
Barium
Cadmium
Calcium
Chromium
Copper
Iron
Lead
Manganese
Mercury
Molybdenum
Nickei
Potassium
Selenium
Silicon
Silver
Sodium
Vanadium
Zinc
" prfrnsty %
' -ftne/JJ *" :
Te
0.41
36
0.12
4200
0.12
110
1.1
1.5
0.045
- 0.026
4.6
0.90
2.7
28
0.78
0.30
0.0040
0.14
0.31
4.8
.014
24
0.020
180
0.048
27
Duplicate
- 4ms/LJ
st 1 (Continued)
0.42
37
0.12
4400
0.12
120
Test 2
1.1
1.5
0.045
0.024
4.6-
0.90
2.7
28
0.80
0.30
0.0028
0.14
0.31
4.9.
.012
24
0.020
190
0.047
28
J-"|IPD^ "„;
2.41
2.74
0.00
4.65
0.00
8.70
0.00
0.00
0.00
8.00
0.00
0.00
0.00
0.00
2.53
0.00
35.29
0.00
. 0.00
2.06
15.38
0.00
0.00
5.41
2.11
3.64
/"ajjjaCjSVOS/^
', ' IRPDJ ^
20
20
20
, ' 20
20
20
; 20
I 20
\ 20
20
20
: 20 '
20
20
20
' 20
20 '
20
20
20
20
20
20
20
20
20 .
(coiit.)
135
-------
Table 32. (Continued)
st 2 (Continued)
1.7
3.9
0.15
4.8
0.085
11
0.81
0.013
4.6
22
', ' 2.3
1.0
0.45
! 0.13
1.3
0.35 '
23
0.091
68
0.073
2000
0.058
0.14
100
Test 3
i
0.43
: 0.32
-V'^app >;;:;
41.86
66.67
23.53
0.00
48.89
0.00
65.56
42.42
64.71
62.50
66.67
26.09
37.84.
0.00.
14.29
47.83
8.33
18.91
38.10
4.03
18.18
6.67
44.44
82.35
'
13.04
8.96
-I 6$«<$y«*' ""•>
..?•,-"- -IRPD*"'"' '^ -
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20 ~~
20
20
20
20
20
20
20
20
20
•
20
. 20
(cent.)
136
-------
Table 32. (Continued)
- ,> -&X»p i
OujJlitate"
"•• ' ImgAJ
' **&, "^'
••"\"*' "*•? " '"
% A,s, ,...,... ^ % ^ - '•'
Va&feetive* ; ^
-,-«<• \- " JRPOh--- '^ "
Test 3 (Continued)
PRE-TEST (CONTINUED)
Barium
Boron
Cadmium
Calcium
Chromium
Copper
Iron
Lead
Manganese
Mercury
Molybdenum
Nickel
Potassium
Silicon
Silver
Sodium
Strontium
Vanadium
Zinc
0.030
1.5
0.025
4.7
1.10
1.5
41
0.68
0.42
0.0028
0.31
0.28
8.4
29
0.013
650
0.020
' 0.028
26
0.027
1.4
0.028
4.8
0.96
1.4
37
0.63
0.38
0.0027
0.28
0.26
7.7
24
. 0.013
600
0.020
0.026
24
10.53
6.90
11.32
2.11
13.59
6.90
10.26
7.63
10.00
3.64
10.17
7.41
8.70
18.87
0.00
8.00
0.00
7.41
. 8.00
POST-TEST
Aluminum
Arsenic
Barium
Boron <
Cadmium
Calcium
Chromium
Cobalt
1.1
3.3
0.080
1.1
0.090
4.4
1.9
0.022
1.1
7.7
0.079
1.1
• 6.078
4.0
1.9
0.024
0.00
80.00
1.26
0.00
14.29
9.52
0.00
8.70
20
1 20
20
20
20
20
20
20
20
20
20
20
20
: 20
20
20
20
20
20
20
20
20
20
20
20
20
; 20
(cont.)
137
-------
Table 32. (Continued)
Tost 3 (Continued)
PQST.TEST (CONTINUED)
Copper
3.4
3.3
2.99
20
46
44
4.44
20
2.1
2.1
0.00
20
Manganese
0.53
0.52
1.90
20
0.018
0.020
10.53
20
0.40
0.40
0.00
20
Nickel
0.96
0.97
1.04
20
13
13
0.00
20
0.020
0.017
16.22
20
Silicon
43
42
2.35
20
0.057
0.048
17.14
20
1200
1200
0.00
20
0.019
0.018,
5.41
20
0.071
0.069
2.86
20
Z!nc
65
64
1.55
20
Metal Emissions Data ,
For each of the Demonstration Tests, samples of the stack gas were collected to
evaluate metal emissions. Samples were collected using the Multiple Metals Train
(MMT) which is discussed in Section 5. At the completion of each Demonstration
'Test, the impinger solutions were collected, the filter was recovered, and the sample
train components were rinsed. In the' laboratory, the probe and nozzle rinses (PNR)
were combined with the filters and digested as single sample. All of the data
«
generated for the metals emissions were obtained from the filter and PNR.
138
-------
The analyses of the impinger solutions yielded non detectable quantities for most
metals, or values at or near the detection limits. Since these results were
insignificant, they were not included in the calculation of the total sample train catch.
Mercury was the only metal that was found in the impingers at a significant level.
This is not surprising given the high volatility of mercury and the operating
temperatures of the process. Raw data for these measurements can be found in
Appendix C.
For Test 2, a primary' and a duplicate MMT were installed into the stack to collect
duplicate metal emission samples. The results of these duplicates are presented in
Table 33. As seen by the calculated RPD values, the results of the duplicates are
poor. Most of the reported values in the duplicate train are a factor of 2 higher than
the primary train. A check of raw data and calculations indicated no source of error
in the reporting of the results. The results of the duplicates indicate that there is
some inconsistency associated with the stack gas flow or sampling procedures. It is
possible that one of the trains was not located correctly inside the stack and an
isokinetic sample was not collected. There were also several problems noted with the
blower during testing activities which may have caused flow irregularities. -Several
metals were also detected in the field blank which is discussed below. Limitations on
these data are discussed later in this section under "CONCLUSIONS AND
LIMITATIONS OF DATA."
Field blanks for the metal trains showed detectable levels for most of the target
analytes. Aluminum, calcium, and sodium were detected at levels greater that 1,000
fjg. The amount of aluminum, calcium, and sodium (combined) in the field blanks
represents approximately 25, 10, and 15 percent of the catch for Tests 1, 2, and 3,
respectively. Magnesium was detected at 480 fjg in the field blank. This represents
about 50 percent of the catch for the Demonstration Test samples.' All other metals
detected in the field blank were insignificant in comparison with the Demonstration
Test results. Zinc, which was of critical importance to this demonstration, was
139
-------
Table 33. Metals Emissions Duplicate Sample Results (Test 2)
=1 "^ \ \X "» -.^X-
ComjMJurn* rX^-/ *$
Aluminum
Antimony
Arsenic
Barium
Beryllium
Cadmium
Calcium
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Mercury
Nickel
Potassium
Selenium
Silver
Sodium
Thallium
Vanadium
Zinc
•. «" v,\W £>s^ f* <•• \ ^
^^S^^P^x', ,/ -
^Jm^^rtoprti^ " of,"-
1.396B
0.046
3.323 B
0.113 B
0.0005
O.057
2.525 B
0.930 B
6.646 B
43.197 B
3.256 B
0.498 B
0.326 B
0.0043 B
0.236 B
9.969
0.015
0.009
5.981 B
0.017
0.093
116.299 B
" ' ''" ** i'
Duplicate - ;
: ^Ipprni, *?*?
2.800 B
0.136
9.753 B
0.242 B
0.0010
0.119
5.089 B
1.951 B
1 3.994 B
89.053 B
: S.785 B
0.933 B
0.763 B
0.0080 B
0.509" B.
21.203
0.034
0.019
12.7228
0.042
0.195
250.197 B
rv?^^S:
66.92
98.90
• 98.35
72.68
36.67
70.45
67.35
70.88
71.20
68.83
70.29
60.80
80.26
60.16
73.29
72.78
77.55
71.43
72.08
84.75
70.83
45.78
B - Indicates that this compound was detected in the field blank.
detected at 160 fjg. This is several orders of magnitude lower.than values found in
test samples; therefore, field blank contamination has little or no effect on the
evaluation of the ability of the process to treat zinc. Field blank data may be found
in Appendix C. Other metals data was not used in evaluating project objectives other
than total emissions. The blank data does not impact total emissions output as zinc
is the dominant metal in the gas stream as stated by project conclusions. Method
140
-------
blanks analyzed along with these samples were clean with the exception of small
quantities of iron (5.5//g) being detected in one blank. These levels are insignificant
i
in comparison the levels of iron detected in the Demonstration Test samples.
Matrix spike recovery checks were prepared and analyzed for the multiple metals train
by spiking a blank filter and clean impinger solutions with the targeted analytes.
Recovery information for these spikes is presented in Table 34. This table shows that
the recovery for most of the analytes fell within the accuracy objectives specified by
the laboratory. Silver was the only outlier with a 20 percent recovery from the filter
media. Since silver was not a metal of great interest to this project, this poor
recovery has no impact on data quality.
TCLP Analyses .
teachability of the treated soil was considered an extremely critical parameter for
evaluation in these Demonstration Tests. Samples were subjected to TCLP and then
analyzed for semivolatiles and metals. The TCLP extractions were performed in
accordance with SW-846 Method 1311. The TCLP leachate was analyzed using SW-
846 Method 8270 for semivoiatiles and SW-846 Method 6010 for most metals.
Other metals were analyzed as previously specified in this sectioh. Detailed
discussions of the extraction, digestion, and quality control procedures utilized to
ensure proper sample preparation, instrument tuning, and calibration have been
outlined previously in this section for both semivolatiles and metals.
TCLP Metals
Samples of the feed and treated soil were collected to determine the teachability of
metals. In general, low levels of metals were detected in the leachate samples except
for-calcium, zinc, and sodium in the feed soil. -Zinc, which was spiked at high levels,
leached freely from the feed soil and was effectively contained in the treated material.
141
-------
Table 34. Recovery Check Results for Multiple Metals Trains
," Comptama^^B-^1^
Silver
Copper
Manganese
Vanadium
Nickel
Barium
Aluminum
Chromium
Lead
Thallium
Calcium
Sodium
Iron
Potassium
Selenium
Arsenic
' Mercury
Aluminum*
Antimony
Cadmium
Copper
Manganese
Thallium
Vanadium
V ', *\>^mwJi*iiM - 4\ V -1 ,
Filter
Filter
Filter
Filter
Filter
Filter
Filter
Filter
Filter
Filter
Filter
Filter
Filter
Filter
Rlter
Filter
Filter
Rlter
Rlter
Rlter
Filter
Rlter
Liquid
Liquid
Liquid
Liquid
Liquid
Liquid
Liquid
Liquid
Liquid
-'" ; CV % fcwwwwy' ' ,- ' ,"",' " "
91
101
20 Q
83
86
86
87
88
90
90
90
92
93
94
• 94 •
94
9§
96
96 /
98
101
85
107
99
92
93
94
95
95
' 95 '
95
142
-------
Table 34. (Continued)
' > „.•''' %% ' •••'•. ',••
- - vr*,- , „ ' Compound*" , "™£r - ":
Barium
Beryllium
Chromium
Magnesium
Nickel
Silver
Sodium
Zinc
Iron
Calcium
Lead
Potassium
Selenium
"" '' RftBiflJquiif ' "
Liquid
Liquid
Liquid
Liquid
Liquid
Liquid
Liquid
Liquid
Liquid
Liquid
Liquid
Liquid
Liquid
,_ %%>-- <-f
-------
Table 35. TCLP Metals Duplicate Sample Results
* ^ I --t}'e£
Compound xt^*V
s' V "i1"'1:
FEED SOIL
Barium
Cadmium
Calcium
Copper
Iron
Magnesium
Manganese
Mercury
Potassium
Sodium
Vanadium
Zinc
TREATED SOIL
Aluminum
Barium
Calcium
Copper
Iron
Manganese
Sodium
Zinc
TREATED SOIL
Aluminum
Barium
Calcium
Copper
Iron
^- ^-- "'f $.§•*'•'•• ^ •* •.*••"
4t*¥?/^?4lnrmiy * "V.f "' /
s. V^V.-. g*.-?.: . jss&Af 1 "* s s
r$1*^ &ns/M „ --
•
0.16
0.066
190
6.2
0.110
7.8
5.0
0.0003
4.9
1600
0.094
1000
•
0.23
0.087
1.8
0.12
2.1
0.050
1SOO
0.39
' 0.28
0.090
1.3
0.17
2.4
- \v Ouplteate V" , ':-
"v^'^liHBJO K '!
Tostl
0.12
0.064
170
3.7
0.055
7.S
; 4.6
0.0008
4.2
1400
0.097
1000
.
0.34
0.074
2.5
0.23
3.1
0.061
' 1500
0.51
Tost 2
'
0.55
0.080 '
3.8
0.54
3.5
'./'"tffib ---^-v
s •. /• ^ ^ v
28.57
3.08
11.11
50.51
66.67
1.27
3.33
90.91
15.38
13.33
3.14
0.00
15.87
16.15
32.56
62.86
38.46
13.82
0.00
26.67
65:06
•11.76
38.04
104.23
37.23
'; \X»fitovr ^
", --^iRPOJ ^',
30
30
30
30
30 '
30
30
30
30
30
30
30
-
30
~~30
30
30 '
30
30 •
30
30
30
30
30
30
30
(cont.)
144
-------
Table 35. (Continued)
•. % % Compound '„
^•. .. v •. r •••• -v^t
^ , ;,>f«mBiy' - ,
, kngfo
Duplicate
ta*g/U '
; 'RPD * '
- " Objectives
' lBPt» ^ " -
Test 2 (Continued)
TREATED SOIL (CONTINUED) '
Manganose
Sodium
Zinc
0.051
1400
0.32
0.071
1400
0.40
32.79
0.00
22.22
30
30
30
Test 3
TREATED SOIL
Aluminum
Barium
Calcium
Copper
Iron
Manganese
Sodium
Zinc
0.39
0.0.70
2.9
0.11
4.3
0.13
1400
0.37
0.25
0.080
1.2
0.50
58
0.36
1400
0.23
43.75
13.33
82.93
127.87
172.39
93.88
0.00.
46.67
3O
30
30
30
30
30
30
30
however, does not significantly impact project'conclusions because of how this data
is evaluated, as explained in Section 6. Data evaluations account for variability
associated with this matrix. •
Spiked samples were analyzed for the treated and feed soil. These results are
summarized in Table 36. The feed soil spikes exhibit good precision and accuracy
values for all metals with the exception of calcium, sodium, and zinc. This is most
likely due to the high levels of these compounds found in the unspiked samples, lit
is possible that these metals were not spiked at high enough levels to exceed the
native concentration found in the matrix for these metals. Hence, this behavior can
be expected and does not imply that routine samples results have been impacted. The
treated soil spikes showed excellent precision and accuracy values for allimetals with
145
-------
Table 36. TCLP Metals Spiked Sample Results
: ?•* v4
$$S& ^•v':v% Recoveries" ' f'> ™.
|£gjp
S^-
-,%*••. **W*9w ^f '
f ff
- *PP^
••wV ^ "\
s%<- V^Qbjifl
," Aecawv, ,'i
*£»VT:*%
v'gSS'fv
TEST SOIL
SAIC-0309 MS
SAIC-0309 MS
SAIC-0309 MS
SAIC-0309 MS
SAIC-0309 MS
SAIC-0309 MS
SAIC-O309 MS
SAIC-0309 MS
SAIC-0309 MS
SAIC-0309 MS
SAIC-0309 MS
SAIC-0309 MS
SAIC-030S MS
SAIC-0309 MS
SAIC-0309 MS
SAIC-0309 MS
SAIC-0309 MS .
SAIC-0309 MS •
SAIC-0309 MS
SAIC-0309 MS
SAIC-0309 MS
SAIC-0309 MS
Aluminum
Antimony
Arsenic
Barium
Beryllium
Cadmium
Calcium
Chromium
Copper
Iron
Lead
Magnesium
Manganese '
Mercury
Nickel
Potassium
Selenium
Silvar
Sodium
Thallium
Vanadium
Zinc
SAIC-0314 MS
SAIC-0314 MS
SAIC-O314.MS
SAIC-0314 MS
' SAIC-0314 MS
SAIC-0314 MS
Aluminum
Antimony
Arsenic
Barium
Beryllium
Cadmium
103
91
88
101
101
96
48 QX
100
86
102
97
96
82
117
99
94
75
• 97
OQX
107
100
OQX
96
89
93
. 101
101
96
138 QX
99
108
102
97
100
107
115
98
106
74 Q
96
304 QX
102
99
1110QX
100
• 90
91
101
101
96
NC
ISO
97
102
97
98
95
116
99
100
75
97
NC
105
100
NC
7.0
2.5
5.5
0.0
0.0
„ 0.0
NC
1.0
22.7
0.0
0.0
4.1
26.5
1.7
1.0
12.0
. 1.3
. 1-0
NC
4.8
1.0
-NC
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-1 4O
65-140
65-140
65-140
65-140
65-140
65-140
75-120
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
TREATED SOIL
100
110
96
'. 102
102
97
102
108
97
103
103
98
i
101
109
97
103
103
98
2.0
1.8
1.0
1.0
1.0
1.0
65-140
65-140
65-140
65-140
65-140
65-140
30
30
30
30
30
30
leont.r
146
-------
Table 36. (Continued)
~ FwfoHD "<
f %
Element
-. •-
' %' Recoveries
MS ;
"MSD"
Mean
RPD
-'
Objectives
Accuracy
i°& flee.*
Precision
mpoi - ••
TREATED SOIL (CONTINUED)
SAIC-0314MS
SAIC-0314 MS
SAIC-0314 MS
SAIC-0314 MS
SAIC-0314 MS
SAIC-0314 MS
SAIC-0314 MS
SAIC-0314 MS
SAIC-0314 MS
SAIC-0314 MS
SAIC-0314 MS
SAIC-0314 MS
SAIC-0314 MS
SAIC-0314 MS
SAIC-0314 MS
.SAIC-0314 MS
Calcium
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Mercury
Nickel
Potassium
Selenium
Silver
Sodium
Thallium .
Vanadium
Zinc
105
101
99
103
104
103
100
99
102
121
64 Q
97
430 OX
102
101
101
107
102
101
104
106
104
101
98
103
118
54 Q
98
100
108
102
101
106
100
100
104
105
104
101
99
103
120
59
98
1OO
105
102
101
1.9
1.0
2.0
1.0
1.9
1.0
1.0
1.0
1.0
2.5
16.9
1.0
NC
5.7
1.0
0.0
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140
65-140,
65-140
65-140
65-140
65-140
65-140,
75-1 20 '
30
30 •
30
30
30
30
30
30
30
30
30
30
30
30
30
30
Q Outside control limits.
• X Amount of spike added was significantly less than the natural analyte concentration in the sample; recovery not meaningful
due to dilution or background interference.
NC Not calculated.
the exception of sodium. Problems associated with the sodium recovery are
discussed below and discussed in Section 6. .
Method blanks for the TCLP metals samples were free of significant contamination
that may have effected results with the exception of sodium. Sodium was detected
in a method blank at a level similar to those found in project samples. This could
explain the poor precision and accuracy values for sodium in the treated soil spikes.
Although sodium contamination was detected in the method blanks, this does not
present a significant impact on data quality. Since sodium hydroxide was used in the
147
-------
scrubber as a neutralizing agent, sodium was not a heavily weighted factor used to
evaluate critical p'roject objectives.
TCLP Semivolatiies
Samples of the feed and treated soil were collected for evaluation of leachabie
semivolatile compounds. There were no target compounds detected in either of these
matrices. Duplicates were collected and analyzed for each test as described in the
TCLP metals section. Since none of the targeted compounds were detected, it was
not possible to evaluate the results of the sample duplicates from a precision
standpoint. The investigators of this project were surprised to find that none of the
hexachlorobenzene leached from the feed soil. To confirm these results, a special
study, which is later discussed under "SPECIAL STUDIES," was conducted.
i
Matrix spikes were performed on both the treated soil and feed soil leachates. These
results are summarized in Table 37. The feed soil show one poor recovery for phenol
and some high-RPD values for 2-chlorophenoi, 4-chloro-3-methylphenol and phenol.
Recoveries for the treated soil were all acceptable and the RPD values were within the
control limits. Since no target compounds were detected in the feed soil, the poor QC
associated with a few of the spike compounds has no impact on TCLP data quality.
Surrogates spiked into the TClP semivolatile leachate samples are summarized in
Table 38. As noted in the table, there are some recoveries that are outside the
control limits. Based upon the sample results, these surrogate recoveries have little
impact on the data because there were no semivolatile compounds detected in the
leachates and these data were not used to evaluate numerical project objectives. The
project conclusions simply state that no sernivolatiles were found to leach from the
treated soil.
148
-------
Table 37. TCLP Semivolatile Matrix Spike Results (Test 3)
J 'JFtotofi '
•.'•., , •&<(.•(. ' '•^
, ', ,Ctt»jtt*»U*;;'
^% ' <
, : ,* •. - « \
' •.•••. :<
, * ••
~ %^eeovftnr«e - v
W$^
-• %
'«SD
- -Mean
"w&"
•f f,
•J- %
f % f %
'^Ot>je
-------
Table 38. TCLP Semivolatile Surrogate Spike Recovery Data
• ^^ i. .. v
-4V-M4Wi - „'
^Recover? ,
--.^' *.
JLtftt
jMfci >l
pvtmiA -;
' -Uhrtite x -
-/,. m - -
•> <• -'
FEED SOIL
2-Ruorophonol
2,4,6-Tribromophonol
Nitrobenzene-d(
Phonol-ds
Torphonyi-d,,
4
4
4
4
4
4
58
53
71
83
70
110
32
29
27
5
38
10
33
10
31
78
14
101
TREATED SOIL
2-Fluorobiphenyr
2-Ruorophenol
2.4.6-Tribromophanol
Nitrobehzene-ds
Phenol-dt
Terphenyl-d,4
7
7
7
7
7
7
76
70
77
93
84
105
27
19
12
14
2S
19
39
37
58
76
43
77
105
70
89
88
92
124
2
1
0
0
0
0
0
0
0
0
0
0
43-116
21-100
10-123
35-114
10-94
33-141
115
94
95
109
104
128
1
0
0-.
0
0 '.
'0
0
0
0
0
4
0
43-116
21-100
10-123
3S-1 V4
10-94
33-141
* Number of routine field samples and field duplicates, not including blanks, matrix spikes, and recovery checks.
/
Dioxin and Furans
Samples were collected during each of the Demonstration Tests to determine the
presence of polychlorinated dibenzodioxins and poiychlorinated dibenzofurans
(PCDDs/PCDFs). PCDD/PCDF samples were collected because hexachlorobenzene,
which is a potential PCDD/PCDF-forming compound under the thermal processing
conditions encountered for these tests, was spiked into the feed soil . The sampling
strategy was to analyze the stack and treated soil samples first to determine if these
compounds were present. If PCDDs/PCDFs were detected in these matrices, then all
of remaining samples would be analyzed. Since the samples that were initially
analyzed only contained'trace quantities of PCDDs/PCDFs, the remaining samples
150
-------
were not analyzed
- ..... - - --. -i: -r
L and E' respectively
A" S0//dand stack samples wo
«««««..»».„.„„ ,;™ •™"~ —.. ~ „.,„„,.,
„„„.„,„.
-
^2*7 -~*.— -~ ndt:;r0 at ij and °-29 •* ~7
eCDF Wh'Ch w« Seated at 7 p "'0'1' b/a"k eXCePt for 2-3-4.7 -
at
sh e at 7 ps M^,"' a" eXCePt for 2-3-4.7 8-
™
151
-------
TMQL. Blank results were, nonetheless, deemed acceptable by the analytical
laboratory and indicate no significant impact on sample results.
Each sample was spiked with surrogate standards, alternate standards, and internal
standards to monitor the quality of the results generated. For each sample extracted,
nine carbon-labeled PCDD/PCDF congeners representing the tetra- through octa-
chlorinated PCDDs/PCDFs were added to the sample. These standards measured the
overall method efficiency and provided a correction for the unlabeled analog (isotope
dilution mass spectrometry). Five surrogates and two alternate compounds were
added during the sample preparation to measure efficiencies of extract preparation
steps.
The summarized results of the surrogate and internal standards are presented in Table
39 for the treated soil samples and Table 40 for the stack gas samples. As noted in
these tables, most of the recoveries were within the acceptance ranges. Recoveries
of an internal standard that are not within the control does not necessarily indicate a
problem since the corresponding target analyte is corrected for the recovery. Only
erroneous recoveries would indicate a problem with the quantitations. Internal
standard recoveries were satisfactory. Poor surrogate recoveries were noted for only
a few analytes and were usually only slightly below or above acceptance limits.
Based upon the sample results, the poor recoveries have little or no effect on overall
data quality. . ; - .
Duplicate samples for the stack emission were collected during Test 2. Duplicate
samples for the treated soil were also collected and analyzed for each of the
Demonstration Tests. Table 41 summarizes the results of the duplicates for the
treated soil samples and Table 42 presents the results of the stack gas duplicates.
These tables indicate that the majority of the duplicate results do not meet the QAPP
objectives. Compounds that were beyond the control limits were most likely due to
sample interferences since most of the hits were very .low and close to the TMQL.
152
-------
table 39. PCDO/PCDF Internal and Surrogate Treated Soil Recoveries3
{ •- f *.f •• •• >v.
"" OjropOBWl '
•.,-"
'Sfe-.f
'M68»i
. Recovery '
-t%* ~
' f " < 1
Sttf.
O«w.
fflMK,
%WiO
**m "
**«!
1%)
No, o* Rosults ,
Outside ..ttmlls
uw; '] s «$& "
s ' V
- ,\- ,JM
'" s '"
5onW v^^,
Urrifes '"- " -
,Wl ^ V
Surrogate Standards
13C-HpCDF-789
"C-HxCDD-478
13C-HxCDF-478
13C-PeCDF-234
37CI-TC:DD
8
8
8
8
9C
13C-HxCDF-234
"C-HxCDF-789
8
8
80
94
79
92
56
84
81
20
,16
17
18
23
46
64
50
71
28
107
119
106
118
97
Alternate Standards
18
18
49
49
111
106
0
0
0
0
3
0
0
0
0
0
40
40
40
40
40
-
-
-
-
130
130
130
130
130
0
0
0
0
25
25
.
-
130
130
Internal Standards ' '
"C-2378-TCDD
13C-2378-TCDF
"C-PeCDF-123
"C-PeCDD-123
13C-HxCDF-678
13C-HxCDD-678
13C-HpCDF-67-8
t3C-HpCDD-678
"C-OCDD
9C
9e
8
8
'8
8
8
.8
8
59
53
80
98
79
84
74
87
94
24
. 21
29
32
24
22
16~
20
24
29
24
45
67
• 48
49
44
51
54
102
85
125
149
127
124
89
112
118
2
3
0
0
0
0
0
o
0
0
0
0
2
0
0-
0
0
0
40
40
40
40
40
40
25
• 25
25
-
-
-
.
.
. "
.
-
130
130
130
130
130
130
130
130
130
* Includes one treated soil core water sample.
b Number of routine field samples and field duplicates analyzed. Does not include blanks, matrix spikes, or recovery checks.
c Includes confirmation as well as full screen analysis results.
153
-------
Table 40. PCDD/PCDF Internal and Surrogate Gas Recoveries
. - \ , ' ^
> ^ •* ts;,*,'.11'
S^.h
b 1?
Compound: """";•
^t||*^
s»##
^--"-&»:^\..sv
"' ^"%v?H v X-iS..
*$»&*&
'Racovafy "
$»<$Wto s:
•4^ >•.-"•. .. i.,-.,, ^
*>&f»
*•>>:, . .
s«C/
oW,
^i%&
«"•/ :
, **" •••
tiim '•
f*J ;
* .>•.•.
f •, ••
Max
1%l'
,N««^fe ft««uU*'':
:.;Q(3t6iti6:L?nnitst '••
' \jytt..'..
' High sx
^^«t'-*-'y-- '-- £
v dunirc} ~ „«
- ^v:»-N*»" \""
--;" ™$ "* - -
13C-HpCDF-789
13C-HxCDD-478
13C-HxCDF-478
13C-PeCDF-234
27CI-TCDD
Alternate Standards
13C-HxCDF-234
13C-HxCDF-789
4
4
4
4
8k
4
4
97
167
118
1O8
104
102
72
14
- 78
7
8
8
19
16
82
108
?11
98
95
111
279
126
118
116
79
52
120
88
0
0
0
0
o
0
0
0
2
0
0.
0
0
0
70
70
70
70
70
40
40
Internal Standards
13C-2378-TCDD
13C-2378-TCDF
13C-PeCDF-123
13C-PeCDD-123
13C-HxCDF-678
13C-HXCDD-678
13C-HpCDF-678
13C-HpCDD-678
13C-OCDO
8k
8h
4
4
4
4
4
4
4
77
74
65
73
82
84
76
74
59
39
33
16
32
8
41
7
17
30
34
35
42
53
7S
36 '
71
56
28
151
131
77
120
94-
135
86
94
86
2
2
0
. 0
0
0
0
0
0
1
1
0
0
0
1
0
0
0
40
40
46
40
40
40
25
2S
25
-
-
-
-
-
-
-
-
-
•
•
-
-
-
-
130
130
130
730
130
130
130
130
130
130
130
130
130
130
130
130
• Number of routine field samples and duplicates analyzed. Does not include blanks, matrix spikes, or recovery checks.
k Includes comfirmation as welt -as full screen analysis results.
154
-------
Table 41. PCDD/PCDF Duplicate Soil Sample Results
W-T ''' *
' ' < :'
Compound
Treated Soil -Test 1
1234678-HpCDD
•OCDD
2378-TCDF
23467«-HxCDF
1234678-HpCDF
OCDF
Total PnCOD
Total HxCOD
Total HpCDD
Total TCDF
Total HxCDF
Total HpCDF
Treated Soil - Test 2
1234678-HpCDD
OCDD
2378-TCDF
234678-HxCDF
OCDF
Total PoCDD
Total HxCDD
Total HpCDD
Total TCDF '
Total P«iCDF
Total HxCDF.
Treated Soil - Test 3
123678-HxCDD
123789-HxCDD
1 234678-HpCDD
•. f % % f *•
' •• 'jPnntHEY
' ,- " fpp$ - " - - *
Duplicate
- '
-------
Table 41. {Continued)
v - "A Si'/?
-' Compound ,^''*l
Treated Soil - Test 3
OCDD
2378-TCDF
23478-PeCDF
123478-HxCDF
123678-HxCDF
234678-HxCDF
1234678-HpCDF
1234789-HpCDF
OCOF
Total TCDD
Total PaCDD
Tote) HxCDD
Total HpCDD
Total TCDF
Total PeCDF
Total HxCDF
Total HpCDF
4^\wrf^wy^% *""v '
•tf^^tifyti ---'l*
„ "- 3Kg)iie
-------
Table 42. PCDD/PCDF Duplicate Soil Sample Results
. ,y ,, •..,,,," - S"fy. ;
** *" ** % ^ !
, CojnpotinS, «1 \ s :
" 'V •• <% %-> *" *
*""-•• ' - Pom**?,
'V 6»ptV) /-
0 ^OupKC8t« '
te*v» ' ^ "
%«'--'-% % , ' :
"• % •>.
,flPDv'^''
Stack Gas - Test 2
2378-TCDD
12378-PeCDD
123478-HxCDD
123678-HxCDD
123789-HxCDD
1234678-HpCDD
OCDD
2378-TCDF
12378-PeCDF
23478-PeCDF
123478-HxCDF
123678-HxCDF
234678-HxCDF
123789-HxCDF
1234678-HpCDF
1234789-HpCDF
OCDF
Total TCDD
Total F'eCDD
Total HxCDD
Total HpCDD
Total TCDF
Total PeCDF
Total HxCDF
Total HpCDF
2.89E-04
3.2SE-04
1 .49E-03 Q
5.95E-04
2.68E-03 Q
4.65E-03 B
2.10E-02 B
6.46E-03
2.39E-03
3.76E-03
9.31E-03
3.72E-03
4.97E-03 B
6.21 E-04
1.17E-02B
1.71E-03
6.29E-03 B
3. 61 E-04 Q
1 .96E-03 Q
1 .49E-02 Q
9.57E-03
2.89E-02 Q
3.08E-02
3.41 E-02
1.96E-02
2.98E-04
2.69E-04
2.45E-04
4.91E-O4
7.36E-04
3.61E-03B
1.61 E-02 B
5.64E-03
1.69E-03
2.82E-03
7.16E-03
2.81E-03
4.35E-03 B
2.05E-04
9.37E-03 B
1.17E-03
6.48E-03 B
2.98E-O4
1 .08E-03 Q
5.15E-03
7.44E-03
2.07E-02
2.48E-02 Q
2.38E-02
"1 .55E-02
3.07
19.16 •
152.29
19.48
113.82
25.18
26.42
13.55 •
34.31
38.57
26.11
27.87
13.30
100.73
22.12 .
37.50
2.98
19.12
57.89
97.26
25.04
33.06 .
21.58
35.58
23.36
- "^ 3&}ft«JV9*
-\J ^{RPOJ
25
25
25
! 25
25
25
25
25
25
; 25
25
' . 25
25
25
25
25
25
2s'
25
25
25
25
25
25
25
B - Indicates that this compound was detected in a blank.
Q - Indicates that this value is an estimated value.
157
-------
Several of these compounds were also found in method blanks where indicated.
Based upon the low sample results, no significant impacts on data quality arise from
the duplicate results.
Stack Gas Particulates
Paniculate data was collected for each of the Demonstration Tests. Particulate
loading was determined using a gravimetric procedure. For Test 1, a primary and a
duplicate sample were collected. Precision, as indicated by the RPD of the duplicates,
was measured at 35 percent. This exceeded QAPP control limits. This -is an
indication that the particulates within the stack were not well-mixed or that there was
a problem in the collection of an isokinetic sample. This is in agreement with the high
RPDs associated with the MMT duplicate samples which indicated the particulate
distribution was not uniform within the stack. As previously mentioned, this may
have been a result of problems associated with the blower which may have caused
irregular flow patterns in the stack. Data users should be cautioned that the values
associated with the particulate measurements may have an error range of ±35
percent of the measured value. As noted by the paniculate summary results, the
potential error associated with these data has no effect on conclusions pertaining to
particulate emission since the reported values were well above the regulatory limits.
This is discussed in further detail in Section 6.
One field blank (filter) for particulates showed 2.3 mg of loading;, the second blank
showed no particulate loa'ding. These values are very small in comparison to the
catch on actual sample filters and, therefore, have no significant impact on sample
results.
158
-------
Chloride .
•Chloride content in the stack gas was measured for each test. Chloride samples were
analyzed using ion chromatography. Results of multipoint calibrations and calibration
checks met laboratory specifications. Three samples were spiked and analyzed.
Spike recoveries were at 95, 89, and 83 percent with a control range of 75 to 125
percent. Preparation blanks prepared for and analyzed with these samples were free
of contamination. A laboratory duplicate sample was analyzed, these results were
compared against the original analysis and yielded an 8 percent RPD. Quality control
parameters for these measurements indicate no limitations on measured results. -
Continuous Emission Monitoring
Flue gas was monitored continuously for O2 by EPA Method 3A, NOX by EPA Method
7E, SO2 by EPA Method 66, CO by EPA Method 10, CO2 by EPA Method 3A, and
THC by EPA Method 25A. Samples were collected from the stack throjjgh stainless
steel tubing and passed through a series of impingers for gas conditioning. Sample
gas them passed to a manifold for distribution to each of the monitors.
Each analyzer was calibrated at the beginning and end of each Demonstration Test.
The calibration consisted of three points which bracketed the average concentration
of interest for most compounds. The NOX calibration only went up to 888 ppm, which
was lower than the average concentration detected for each Demonstration Test. For
Tests 2 and 3, a cylinder that contained 5,070 ppm was used to check the upper limit
of the calibration (it should be noted that this is close to reported NOX concentrations).
Results of these checks indicated that the analyzer measurements were within an
average of 5% of the known cylinder value. Calibration checks, zero, and drift checks
were acceptable for all of the gases for each demonstration test.
159
-------
After calibrating the analyzers for Test 1. audit cylinders were analyzed to check the
calibrations for NO, and CO. Audi, cylinders were EPA Protocol No. 1 certified
Cylinder concentrations were certified at 2,351 ppm for NOX and 40.4 ppm for CO
Results of these checks showed a RPD of 2 ,o 3 percent for the NO, and a RPD of 5
percent for the CO. The results of the QC checks associated with the CEMs indicate
that the data obtained are of acceptable quality.
Physical Property Measurements
Demonstration Test soil samples were analyzed for higher heating value (HHV) bulk
density, and moisture (all non-critical), as well as chloride content. Quality control
efforts for the HHV consisted of the analysis of known standards three times
Recoveries of these checks were 99. 99, and 100 percent of the theoretical values
Three duplicate sample to measure the-precision of ithe measurements all yielded RPD
of 0 percent. Four duplicates were analyzed to provide QC for the density
measurements. These results yielded identical results for two samples and a RPD of
0.4 and 2.26 percent for the other two. Moisture measurements were performed
tw,ce on a sample and yielded identical results. Quality control for" the chloride
measurements consisted of a spiked sampie and the anaiyses of a reference materiai.
The sp,ke recovery was 85 percent and the reference material was 106 percent of the
theoretical value. These results imply that the there are no limitations associated with
the physical properties measurements.
AUDIT FINDINGS
At the start of the Demonstration Test, a field Technical Systems Review was
conducted by an EPA RREL QA contractor. The results of this audit were satisfactory
wrth only minor concerns which were corrected in the field. Project organization and
QA management were reviewed and deemed satisfactory for this proiect The
160
-------
auditors also reviewed the sampie custody and sample shipping. All procedures were
be,ng performed ,n accordance with ,he QAPP and pertinent DOT regulations.
A review of the solid and liouid sampiing reveaied some concerns with the
col 7 7 S0"d SamPleS' UneqUa' POrt'°nS °f
-------
The auditors Md a deoriefiho -^ to discuss their
management and competed a corrective action recommendat,ons «CAR, form.
* i c 1 QQI The results of this audit
L chloride sampies as per the QAPP to correct the probtem ,dent,f,ed ,n the aud«.
Chloride data quality was not affected.
addressed and appropriate corrective action was ta.en as
this response, data puaiity was not affected and the data was deem d
for accomplish,nfl proiect ogives. Copies of this and aU other aud,t
reports can be found in Appendix L.
162
-------
MODIFICATIONS AND DEVIATIONS FROM THE QAPP
Because a substantial amount of time passed between the original approval of the
QAPP and the Demonstration Test, it was necessary to make some modifications to
reflect current laboratory practices. Changes were also made based upon experience
*•
gained through pre-demonstration activities. Below is a summary discussion of each
of these changes and how they have impacted the overall data quality.
I
• The target lists for SW-846 Methods 8240 and 8270 were modified to reflect
the compounds that were currently being analyzed by the laboratory on a
routine basis. Since the compounds of interest were still included, and the
detection limits were not affected, semivolatile data quality was not impacted.
• Microwave digestion was used for metal samples rather that standard SW-846
procedures. During pre-demonstration activities, it was noted that the SW-846
procedures were not capable of completely digesting the treated soil samples.
A microwave digestion that incorporated the use of hydrofluoric acid to better
digest the treated soil was used on both the feed and treated soil to provide
consistency. The basic procedure was a modification of Method 3051 which
has not yet been EPA-approved. This procedure adequately fulfilled the
objectives of the Demonstration Test.
• Lead and thallium were analyzed using ICP rather than the graphite furnace
method. The quality of these data were non impacted by this change.
• For SW-846 Methods 8240 and 8270, the top 20 Tentatively Identified
Compounds (TlCs) were tentatively identified and semi-quantitated for each
sample.
163
-------
• Sixty grams of the feed soil were analyzed using SW-846 Method 3550. This
was done to obtain a more representative sample of the feed. The
hexachlorobenzene was spiked in the mixture as a solid crystal and may have
been subjected to high variability if a small extract aliquot was taken. Because
a Jarge sample size was used, these samples required large dilutions and
• -
surrogate spiking at levels similar to that of the hexachlorobenzene spike.
Three surrogates, d5-nitrobenzene, 2-fluorobiphenyl, and d3-methylnaphtha!ene
where chosen for surrogate spiking because of their ability to represent the
compounds of interest in the feed and their availability. These modifications
were made to improve the quality of data obtained from the feed samples.
• Matrix spikes for metals were modified for zinc. Soil that did not contain the
zinc oxide spike was spiked at the feed concentration (28,000 ppm) to provide
matrix spike information. The unspiked feed (no zinc) was also spiked with
arsenic, cadmium, chromium, and lead at levels approximately five times the
concentrations of the feed samples. . .
i
• Aluminum, calcium, iron, magnesium, and potassium were added to the target
list for all metal parameters.
« The draft Method 5041 was used in place of SW-846 Method 5040 for VOST
analyses. Method 5041 incorporates the use of a capillary column. VOST data
was not impacted by this change. •
• The scrubber and stack samples were spiked with three additional surrogates
because of the anticipated matrix recovery problems associated with the
routine SW-846 Method 8270 acid surrogates. The three surrogates (C13-
pentachlorophenol, 2,4-dinitrophenol-d3, and 4,6-dinitro-2-fnethylphenoI-d2)
were selected based upon studies prompted by pre-demonstration results.
Adding thes'e surrogates enhanced recovery information for these matrices.
164
-------
Matrix spikes for the feed soil were modified. It was not feasible to spike the
routine spiking compounds at levels which would be detectable after diluting
the extracts to bring the hexachlorobenzene into the calibration range. Soirthat
did not contain the hexachlorobenzene spike was spiked at the feed
concentration, 1,000 ppm of hexachlorobenzene, to provide matrix spike
information. This provided an assessment of the recoverability of the
hexachlorobenzene in the feed soil. These samples were to be re-extracted
with fresh solvent to assist in the evaluation of the extraction efficiencies of
the feed sampiiss.
{
Composite samples of the feed soil for semivolatile and metal analyses were
collected instead of discrete samples as specified in the QAPP. This was done
to ensure representativeness of the feed soil samples. ,
Scrubber solid samples were not collected and analyzed as specified in the
QAPP. Insufficient quantities of solids were generated during the
demonstration to provi.de enough samples mass for the specified analyses,,
Gas canister samples of the stack gas that were collected for volatile organics
compound analysis were not analyzed. This was an option provided in the
QAPP in the event that the VOST cartridges were overloaded. VOST data was
determined to be of sufficient quality such that the analysis of these samples
was not warranted.
The only samples analyzed for PCDDs/PCDFs were the stack; gas arid the
treated soil samples. This was an option provided by the QAPP in the event
that no significant PCDD/PCDF compounds were detected in these samples.
Since no significant contamination was detected, the remaining PCDD/PCDF
samples were not analyzed.
165
-------
Duplicate sample analysis for each tpct CQ •
not be required for ' ' "
'" "» QAPP "«*
SPECIAL STUDIES
and ensure that
were prompted by
- of
r
studies
°f
It was postuiated that these surroo, „
bV the process. Addmonl he "
Which may have also Tsed ^h
Pour experiments Wera ^ 1^ '^ '°
•n these Samp,es aTd
four .periments
that
2°pheno1- and 2-4-6-
UqU°r extracts'
5°diUm "Vdroxide
"** "" 3CW
*"
d""""~"
166
-------
and nitric acid solution were analyzed separately along with an XAD resin blank
to assure proper QA/QC.
RESULTS; Nitric acid appeared to have very little effect on the spiked XAD.
Very small quantities of any of the compounds of interest were transferred to
the eluent and, while recoveries of all BNA spikes from the XAD resin were
low, all were within acceptable ranges. Surrogate spike recoveries were also
acceptable. Matrix compounds spiked separately on the resin and in'the nitric
acid solution indicated that extraction of these compounds was not a problem.
A separate analysis of blank XAD resin showed that the resin was free of any
contamination.
. j
2) Scrubber liquor, obtained from the developer, was spiked with the BNA matrix
spike. This was performed in duplicate to determine the precision of the data.
Extraction and analysis were conducted by SW-846 Method .3520 and SW-846
Method 8270, respectively.
RESULTS: Recovery results of the matrix spike for phenol, 2-chlorophenol, and
4-chlorophenol and the acid surrogate spikes were 0 percent.
Pentachlorophenol {PCP) recovery averaged around 25 percent. Although this
is; a generally low recovery, it is still considered acceptable. Surprisingly, 2-
nitrophenol, and 2,4-dinitrophenol were detected by SW-846 Method 8270.
These compounds were not spiked in the study. It was concluded that the
disappearance of the spike compounds and the appearance of the nitrated
compounds was caused by the presence of NOX in the scrubber solution which
apparently reacted with the spikes.
• •
3) Additional scrubber liquor was spiked with the BNA matrix solution and two
types of analyses were performed on both unspiked and spiked scrubber liquor.
The two analytical methods were SW-846 Method 8270 and HPLC for PCP
167
-------
analysis only. In addition, a GC/MS search was performed on both spiked and
unspiked scrubber liquor to tentatively identify and quantify the next 20 highest
peaks which were not part of the 8270 target list.
RESULTS; Results of Study (2) suggested that the nitrification phenomena
required further investigation. Results of this study were similar to that of the
second experiment. Several nitro-phenolic compounds were detected in the
.spiked solution while no significant amounts of these compounds were
detected in the unspiked samples. HPLC analysis for PCP showed that this
compound remained stable, as would be expected, since it has no open sites
for nitrification. The purpose of HPLC analysis was to eliminate the question
of extraction efficiency; verification that all of the PCP remained was required.
The results of the TICs found other nitrated phenolic compounds which were
not present in the unspiked liquor. The presence of the nitrated TICs suggested
that the matrix spikes were being nitrated.
4} The final experiment was similar to-the first experiment,-except instead of using
nitric acid, scrubber liquor was used in its place.
RESULTS; After determining that nitrification had occurred in solution it was
necessary to determine if this sample effect could take place on XAD resin. To
test this, scrubber liquor was poured through spiked resin, then the resin and
eluent were collected and analyzed. While this is not exactly the same
conditions as collecting a sample with the MM5 train on site, it was expected
to give an indication of possible reactions. Carbon-labeled PCP was also spiked
into these sampjes to track PCP reactions. . (Note that PCP was originally a
target analyte when these experiments were'performed.) Results of these
samples indicated that nitrification could occur on the XAD. The experiments
suggested that PCP probably remained stable and therefore would be detected
in both the scrubber solution and gas.
168
-------
These studies concluded that the routine acid surrogates for SW-846 Method 8270
would not be detected in the scrubber liquors of the XAD resin samples. Therefore
is was necessary to select surrogates that would be stable in these matrices.
Surrogates were selected for spiking and modifications to the QAPP were made
accordingly before the demonstration test began. Additional surrogates used for the
demonstration have been discussed in earlier portions of this section. As previously
noted,, pentachlorophenol was originally a target analyte but was later not considered
critical to accomplishing project objectives when a different soil was used for the
demonstration. Hence, the recovery of acid compounds became less critical
The second study related to the digestibility of the treated soil. It was noted for
analyses conducted during pre-demonstration that the normal SW-846 digestion
procedures were only able to leach, and not completely digest the treated soil. This
indicated that these procedures were inadequate for the digestion of this matrix. This
was obvious from the sample results and from the appearance of the remaining
sample in the digestate. '
Experiments were conducted using a lithium-borate digestion which was better able
to breakdown the glassified soil matrix. Since these earlier experiments, microwave
digestion techniques were developed and appeared to be- a superior digestion
procedure. At the time of the Demonstration Tests, the microwave digestion
procedure was not an EPA-approved method. The microwave digestion was a
modification of SW-846 method 3051. The modification to the method required the
use of hydrofluoric acid in combination with hydrochloric acid and nitric acid to
completely dissolve the sample rather than to leach the sample. A total digestion was
necessary for the analyses of the treated soil due to the limited teachability of metals
in this matrix. The method was used for the feed soil samples so va|ues could be
comparable when performing a material balance.
169
-------
One final study was conducted to determine the effectiveness of the feed soil spiking.
After spiking was completed, samples were collected and analyzed for semivolatiles,
TCLP semivolatiles, and TCLP metals. The results of the semivolatile analyses
assisted in identifying special sample requirements for surrogate spiking, sample
dilution, and extract clean-up to remove hydrocarbon interference from the diesel fuel.
Results of the semivolatile analysis (with no extract cleanup) found hexachlorobenzene
at 294 ppm, naphthalene at 71 ppm, and 2-methylnaphthalene at 134 ppm. TCLP
semivolatile results identified four leachable constituents: hexachlorobenzene (34.
ppm), 2-methylnaphthalene (21 ppm), naphthalene (10 ppm), and phenanthrene (5
ppm). The TCLP metals analysis identified only three metals: barium at 54 ppb,
cadmium at 80 ppb, and zinc at 960 ppm. Results of these studies indicated that the
spiking levels were sufficient to evaluate DRE for hexachlorobenzene and teachability
of zinc.
Upon learning of the preliminary results of the semivolatile leachate samples for the
Demonstration Tests, the investigators were surprised to learn that hexachlorobenzene
was not detected in the feed soil leachate-samples. To verify these results fresh
* » * ' ' ' •
samples were collected of the remaining feed material, before it was shipped for
disposal. These samples were subjected to TCLP and analyzed for semivolatiles by
an independent laboratory. The-results of this second analyses confirmed that the
hexachlorobenzene was no longer leachable, hence no further investigations were
conducted.
SAMPLE HOLDING TIMES
Holding times were specified in the QAPP for each analytical parameter. Most
samples were analyzed within the required holding time with the following exceptions:
170
-------
Description: Semivolatile feed soils SW-846 Method 8270, extract -holding time
violated by 2 days.
Samples: #113 - Test 1 Feed
#116 -Test 1 Feed
; «-
#117 - Test 1 Feed
#127-Test 1 Feed Blank
#166-Test 1 Feed
#167-Test 1 Feed
#159 - Feed Matrix Spike
#160 - Feed Matrix Spike Dup
#243 - Test 2 Feed
#206 - Test 2 Feed
#205 - Test 2 Feed
#207 -Test 2 Feed
#208 - Test 2 Feed
#204 - Test 2 Feed
Cause: Due to the high levels of hexachlorobenzene in the feed samples and the
potential interference of the diesel, the feed sample extracts went
through extract clean-up. Upon analyzing these samples, it was
discovered that the hexachlorobenzene was not detectable.
Investigations within the laboratory determined that the neutral alumina
removed the hexachlorobenzene from the extract. The fraction of the
extract that remained, which went through GPC and not the neutral
alumina, was analyzed and the hexachlorobenzene was detected.
Several shots on the mass spectrometer were required to bring the
results within the calibration range and obtain surrogate recovery
information. These efforts took some time within the laboratory and
valid results were not obtained until two days past the extract holding
time.
Impact: In viewing the results of the Test 3^ feed soil samples, which were
analyzed within the required extract holding time, these results agree
very well. It appears that the integrity of the sample extract was not
171
-------
impacted and the results for these samples are useable for calculating
DREs.
CONCLUSIONS AND LIMITATIONS OF DATA
Upon review of all data related to this project, along with the QC associated with
these results, the data appears to be of sufficient quality to provide proper evaluation
of the Demonstration Test objectives. Overall, most of the QAPP requirements were
met with regard to precision, accuracy and completeness. There were very few
samples that were lost due to holding time difficulties, laboratory preparation, and
poor quality control results.
Among parameters of secondary critical concern were the semivolatile analyses of the
scrubber samples. As noted during the discussion of these data, poor and highly
variable recoveries were noted for the 8270 and special acid surrogates for some of
these'samples. These results imply that-acid compounds (phenols) may have been
chemically attacked causing new compounds to be formed which are not part of the
8270 target list. In addition, there may have been analytical problems in the
laboratory. Certainly phenol-d5 and 2-Fluorophenol were most likely nitrated and
therefore not detected as suggested by pre-demonstration studies. Whatever the
reason for some samples showing low recoveries the associated semivolatile scrubber
samples contained a very limited amount of low level target hits, hence these poor
recoveries have limited impact on this project.
One of the limitations that may be associated with .these results relates to the
apparent variability between the duplicate samples collected of the stack gases. It
' was noted in the paniculate and metal, duplicate samples that there was some
variability associated with these measurements. As mentioned in the discussion of
these results, this variability may be due to low stack flowrates which were caused
172
-------
by a faulty blower during Tests 1 and.2. The blower was noted as having flow
problems and trouble maintaining the required negative pressure in the reactor
chamber. This could have caused interruptions in stack flow patterns that may have
influenced these samples. Other sources of error in the stack duplicate samples may
include sample train installation, sample train maintenance or filter changes, and the
differences in sampling intervals. If these variables were not exactly the same for the
primary and duplicate trains, some differences can be expected in the results.
Although there were problems identified with these duplicate samples, the overall
quality of the data is adequate for the objectives of this project based upon data used
for evaluating furnace performance.
The stack data for all parameters appear to be satisfactory. Stack flowrate
measurement for all of the sample trains during each test were in close agreement
with each other. No significant problems were noted during the review of the data.
VOST samples that were collected were provided adequate data such that the gas
canisters did riot require analysis. There were only a few target compounds in the
VOST tubes that exceeded the calibration range of the GC/MS. These compounds
were benzene and chloromethane. It is important to note that VOST data that
exceeded the calibration range was very rare. These few data points that exceeded
the calibration range did not saturate the GC/MS, hence the quantitations associated
with these values appear to be reasonable. Due to the limited volatile emissions for
this process, these few data points do not limit the use of these data.
All data presented in this report have NOT been corrected for field or laboratory blank
contamination. Where appropriate, the raw data have been flagged when laboratory
method blank contamination is suspected as. interfering with sample results. Field
blank contamination is not denoted on the raw data. Field blank contamination of
naphthalene was, however, significant for the stack gas samples. Similar values of
naphthalene were detected in the field -blank as in routine samples. Naphthalene data
in the semivolatile emission samples is highly suspected as field contamination and,
173
-------
therefore, was not used in evaluation of project objectives. Small quantities of several
metals were found in the field blajik for the MMT. The source of this contamination
has not been identified. Most of the contaminant concentrations were small in
respect to routine samples and therefore has a limited impact on data quality. Other
blank contamination was limited in scope and has no direct impact on data associated
with this project.
* ,
As noted in the Appendices and the Report, all data has been converted to a lbs/100
Ibs feed basis where appropriate. This was done to facilitate the engineering review
of these data. Feed rates for each test were targeted at 120 Ibs/hr. Because of
process interruptions and unforseen events, this feed rate was not precisely achieved
for each test. Feed rates were determined by summing the total feed material fed to
the unit divided by the total time required to feed this material minus any process
interruptions. These parameters were obtained by the evaluation of field notes,
process logs, and field observation. There may be a limited amount of error
•
associated with this evaluation which does not significantly affect the calculated
results.
No information was obtained for the second extraction of a feed sample to determine
if hexachlorobenzene was completely extracted from the soil during normal extraction
procedures. This procedure was placed in the QAPP as a precautionary measure to
evaluate extraction efficiency if the hexachlorobenzene was recovered poorly from
feed samples. The laboratory did extract a sample twice, but accidently combined the
two extracts and analyzed them as a single" sample. Fortunately, hexachlorobenzene
was recovered in the routine feed samples at anticipated pre-spiked levels and the
results of this study were not necessary. Therefore data quality has not suffered from
the loss of this study.
In closing, data generated for this project is rated as satisfactory and is of sufficient
quality to provide for the proper evaluation of test objectives.
174
------- |