EPA/540/R-95/505
April 1995
Removal of PCBs from Contaminated Soil
Using the CF Systems® Solvent Extraction Process:
A Treatability Study
by
Science Applications International Corporation
635 West Seventh Street, Suite 403
Cincinnati, Ohio 45203
EPA Contract No. 68-C0-0048
Work Assignment No. 0-50
SAIC Project No. 01-0832-07-1124-011
Project Officer
Mark C. Meckes, Work Assignment Manager
U.S. Environmental Protection Agency
Risk Reduction Engineering Laboratory
Cincinnati, Ohio 45268
RISK REDUCTION ENGINEERING LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
TECHNICAL REPORT DATA
/Plcate read Instructions on the reierte before tompl 1
1 REPORT NO 2
FPA/54n/R-q.R/finfi
4. title anosubtitle
Removal of PCBs from Contaminate Soil Using the CF-
Systems Solvent Extraction Process: A Treatability Study
S REPORT OATE
April 1QQfi
6. PERFORMING ORGANIZATION CODE
7. AUTMORISI
Joseph Tillman, Lauren Drees, and Eric Saylor
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME ANO AOORESS
SAIC
625 West Seventh Street
Suite 403
Cincinnati, OH 45203
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-C0-0048 WA 0-50
12. SPONSORING AGENCY NAME ANO AOORESS
U.S. EPA
26 W. Martin Luther King Dr.
Cincinnati, OH 45268
13. TYPE OF REPORT AND PERIOD COVERED
Summary Final
14. SPONSORING AGENCY CODE
EPA/600/14
is. supplementary notes Project Officer = Mark Meckes (513) 569-7348
16. ABSTRACT
Soil samples were obtained from the Springfield Township Superfund Site, Davisburg
Michigan. These samples were collected from a specific area of that site which was
known to have some of the highest concentrations of polychlorinated biphenyV The
samples were shipped to Hazen Research Corporation, Golden, Colorado, where CF-System's
solvent extraction pilot plant was located. Soil samples were screened to remove
soil particles^0.25 inches, air dried, and homogenized. Six separate batches, each
weighing 100 pounds, were processed through the extractor. The operating conditions
for each of the first three batch treatments varied. Results from those treatments
were used to select operating conditions for the final three treatments. Results
showed that the process was effective in reducing the concentration of PCBs in site
soils. Removal efficiencies ranged from a low of 91.4% when two extraction cycles
were used to 99.4% when five extraction cycles were used.
17. KEY WORDS AND DOCUMENT ANALYSIS
i. DESCRIPTORS
b.IDENTIFIERS/OPEN ENOED TERMS
c. cosati Field/Group
Solvent Extraction
Wastes
PolychlorinateJBiphenyls
Treatment
Soil
Propane
Hazardous Waste Site
Cleanup
Soil Treatment
Liquified Gas
SITE
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21. NO. OF PAGES
83
20. SECURITY CLASS (This pages
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (R*V. 4-77) previous edition it oiiolctc
-------
NOTICE
This material has been funded wholly or in part by the United States Environmental Protection
Agency (EPA) under Contract No. 68-C0-0048 to Science Applications International Corporation (SAIC).
This document is intended to provide a concise presentation of the data collected and subsequent findings
regarding the subject treatability study. It has been subjected to the Agency's peer and administrative
reviews and has been approved for publication as an EPA document. Mention of trade names or
commercial products does not constitute an endorsement or recommendation for use.
ii
-------
The Superfund Technical Assistance Response Team (START) Program is designed to provide
site-specific, intensive technical support to the U.S. EPA Remedial Project Managers (RPMs) for solving
difficult site remediation problems. The Superfund Innovative Technology Evaluation (SITE) Program, on
the other hand, is designed to enhance the development of hazardous waste treatment technologies. This
is accomplished by performing demonstrations to evaluate the performance of those technologies while
also providing economic data that can be used for estimating their application at full-scale. Thus, the two
programs complement one another. SITE initially begins with an innovative technology and selects a
suitable waste site to demonstrate that technology, whereas START begins with a waste site and selects
an innovative technology (or technologies) to aid in remediation of that site.
This specific report presents the results of a START treatability study, which is part of a larger
SITE demonstration project that will evaluate the CF Systems® (CFS) process at full-scale. This
treatability study was conducted at a pilot-scale to determine the effectiveness of the CFS process in
removing organic contaminants from a significant test volume of soil (e.g., 600 lbs.) acquired from the
Springfield Township Dump (STD) Superfund site near Davisburg, Michigan. The STD site is a candidate
site for demonstration of a CFS full-scale system.
This report is published, but is available only through NTIS. However, a project summary of like
title is also published and may be obtained at no charge from the EPA's Center for Environmental
Research Information, 26 West Martin Luther King Drive, Cincinnati, Ohio 45268, using the EPA document
number .
Reference copies of the project summary will also be available in the Hazardous Waste Collection
at EPA libraries. Information regarding the availability of other reports can be obtained by calling the
Office of Research and Development Publications at (513) 569-7562. To obtain further information
regarding the SITE Program and other projects within SITE, telephone (513) 569-7696.
E. Timothy Oppelt, Director
Risk Reduction Engineering Laboratory
-------
ABSTRACT
A pilot-scale treatability study of the CF Systems® (CFS) Solvent Extraction Process was
conducted at Hazen Research, Inc. in Golden, Colorado between August 8 and August 18, 1994,
Approximately 600 pounds of dried screened feed material was batch fed into the CFS Mobile
Demonstration Unit (MDU) and treated. The feed material was a sandy soil contaminated primarily with
polychlorinated biphenyls (PCBs) and was acquired from the Springfield Township Dump (STD) Superfund
site near Davisburg, Michigan. The main portion of the study consisted of five test runs that were
conducted in two phases. Phase I of the test involved varying the number of extraction cycles for three
runs, and included three, four, and five extraction cycles all conducted for 20 minutes each. Phase II
consisted of two additional test runs using three extraction cycles at 20 minutes each. This yielded a total
of three runs at the most economical condition. A sixth additional run was conducted immediately following
the Phase II testing to test the limits of the pilot unit in treating the soil to desired levels.
The primary objectives of the treatability study were to: 1) determine the effectiveness of removing
PCBs from STD soils to the Remedial Action Standard (RAS) of <; 1 mg/kg on a dry weight basis; and
2) determine PCB concentrations in the filtrate water to ensure proper disposal. Samples of the feed soil
and product solids were collected for each test run and analyzed at a minimum for PCBs, oil and grease
(O&G), volatile solids, and moisture content. Process water used for flushing solids and separating
propane solvent from solids for each test run was analyzed for PCBs, as was the final product oil (organic
extract). The product solids, process water, and product oil were also analyzed for residual propane.
The results of the study indicated that on average 98 percent removal of PCBs was achieved for
the test runs using three extraction cycles (feed PCB concentration averaged 250 mg/kg). However, the
primary objective to attain product solids of ^1.0 mg/kg PCBs was not achieved. The four- and five-
extraction cycle runs did approach the objective at 1.8 and 2.2 mg/kg, respectively. However, the two-
extraction cycle run was well above the objective at 19 mg/kg. This indicated that the number of extraction
cycles required for attaining the lowest concentrations of PCBs in product solids was >3, but s; 5 since
there was no discernable improvement in PCB removal from 4 to 5 cycles. However, O&G analyses on
product solids gave an indication that the 5-cycle run may have outperformed all other runs on an overall
organics removal basis. PCB analyses of filtered process water resulted in concentrations of <1.0 fjg/L
for each of the five main test runs and 1.9 fjgIL for Run 6 filtrate.
This report was submitted in fulfillment of contract number 68-C0-0048
by Science Applications International Corporation (SA1C) under sponsorship of the U.S. Environmental
Protection Agency. This report covers a period from September 1993 to February 1995, and work was
completed as of February 1995.
iv
-------
TABLE OF CONTENTS
Section Page
Notice ii
Foreword iii
Abstract . . , iv
Figures vii
Tables viii
Acknowledgments x
1.0 Project Background 1-1
1.1 Description of the Springfield Township Dump 1-1
1.2 Description of the CF Systems® Solvent Extraction Process 1-4
2.0 Experimental Design 2-1
2.1 Test Objectives 2-1
2.2 Test Conditions and Optimization 2-2
2.3 Sample Locations and Frequency 2-3
2.4 Analytical Methods 2-7
3.0 Field Activities 3-1
3.1 Feed Preparation 3-1
3.2 Sample Collection 3-4
3.3 Field Measurements 3-10
3.4 Process Monitoring 3-10
4.0 Analytical Results 4-1
4.1 PCBs - Feed and Product Solids (Runs 1-5} 4-1
4.2 Moisture - Feed and Product Solids (Runs 1-5) 4-2
4.3 PCBs - Filtrate and Recycled Solvent (Runs 1-5), and Product Oil 4-2
4.4 O&G - Feed and Product Solids (Runs 1-5) 4-3
4.5 Volatile Solids - Feed and Product Solids (Runs 1-5) 4-4
4.6 Particle Size Distribution of Feed 4-5
4.7 Residual Propane - Product Solids, Filtrate, and Product Oil 4-5
4.8 Supplemental Analyses 4-7
4.9 Run 6 Results 4-10
v
-------
TABLE OF CONTENTS (Continued)
Section Page
5.0 Mass Balance 5-1
5.1 Total Materials 5-1
5.2 Solids 5-3
5.3 PCBs 5-4
6.0 Quality of the Data 6-1
6.1 Introduction 6-1
6.2 PCBs 6-4
6.3 Moisture 6-11
6.4 Noncritical Parameters 6-12
6.5 Modifications to and Deviations from the QAPP 6-17
7.0 Conclusions 7-1
7.1 Organics Removal 7-1
7.2 Mass Balance 7-5
7.3 Volume Reduction of Hazardous Waste 7-5
8.0 References 8-1
Attachment A: Physical and Chemical Properties of Propane A-1
Attachment B: Field Measurements Used for Mass Balance Calculations . B-1
vi
-------
1-2 Plan View of the Springfield Township Dump Showing Surficial Extent
of PCB Soil Contamination Above 10 mg/kg and 1.0 mg/kg 1-3
1-3 CF Systems® Process Diagram 1-5
1-4 Photograph of the Mobile Demonstration Unit Showing
Major Components 1-6
2-1 Experimental Design Flow Diagram 2-1
2-2 Two-Phase Approach for Treatability Test 2-3
2-3 Sample and Measurement Locations 2-6
3-1 Test Soil Excavation Location 3-2
3-2 Test Soil Preparation Activities 3-3
3-3 Process Operation Activities 3-5
3-4 Sample Collection and Field Measurement Activities 3-6
5-1 Total Materials Balance Diagram 5-2
6-1 Key to Sample Identification 6-2
7-1 PCB Removal Trend 7-4
7-2 Oil and Grease Removal Trend 7-4
vii
-------
LIST OF TABLES
Number Page
1-1 Major Components of the Mobile Demonstration Unit 1-7
2-1 Test Objectives 2-2
2-2 Process Conditions for All Test Runs 2-4
2-3 Summary of Critical and Noncritical Analyses 2-5
2-4 Analytical Methods Used and References 2-7
3-1 Percentage of Screened Oversize Material 3-4
3-2 Summary of Field Measurements 3-10
3-3 Process Measurement Instrumentation 3-12
3-4 Averages of Process Measurements 3-13
3-5 Ranges of Process Measurements 3-14
4-1 PCB Concentrations and Removal Efficiencies - Feed and Product
Solids (Runs 1-5) 4-1
4-2 Moisture - Feed and Product Solids (Runs 1-5) 4-2
4-3 PCB Concentrations - Filtrate and Recycled Solvent (Runs 1-5), and
Product Oil 4-3
4-4 O&G Concentrations and Removal Efficiencies - Feed and
Product Solids (Runs 1-5) 4-4
4-5 Volatile Solids Concentrations and Removal Efficiencies - Feed and
Product Solids (Runs 1-5) 4-5
4-6 Particle Size Distribution of Feed 4-6
4-7 Residual Propane - Product Solids, Filtrate, and Product Oil . 4-7
4-8 Dieldrin Analyses on Solids and Filtrate (Runs 1-5), and Product Oil 4-8
4-9 Volatile Organic Analyses on Solids, Filtrate, and Product Oil 4-8
4-10 Total Metals in Solids and Filtrate 4-9
4-11 Analytical Results for Run 6 4-10
5-1 Total Materials Balance 5-3
5-2 Solids Balance 5-4
5-3 PCB Balance 5-5
viii
-------
LIST OF TABLES (CONTINUED)
Number Page
6-1 QA Objectives for Accuracy, Precision, Method Detection Limit (MDL)
and Completeness for Critical Parameters 6-1
6-2 PCB Aroclor 1254 MS/MSD Results 6-5
6-3 Method 8080 Surrogate Recoveries, % 6-6
6-4 PCB Aroclor Field Duplicate Results 6-9
6-5 Sample Holding Times - PCB Analyses 6-10
6-6 Results of PCB Analyses of Field QC Samples 6-11
6-7 Oil and Grease MS/MSD Results 6-12
6-8 Volatile Solids Duplicates 6-13
6-9 Volatile Surrogate Recoveries, % 6-13
6-10 Volatile Organic MS/MSD Results, R4-US-04 6-14
6-11 Volatile Organic LCS Results 6-14
6-12 Pesticide MS/MSD Results, R4-US-04 6-15
6-13 Pesticide MS/MSD Results, R4-FC-04 6-16
6-14 Pesticide MS/MSD Results, R4-FL-04 6-16
6-15 Metals QC Results 6-17
7-1 Summary of Results of Analyses Conducted on Product Solids
for All Test Runs 7-1
7-2 Summary of Conclusions 7-2
ix
-------
ACKNOWLEDGMENTS
This report was prepared under the direction and coordination of Mark Meckes, the Project
Manager within EPA's Superfund Technical Assistance Response Team (START), U.S. Environmental
Protection Agency (EPA), Office of Research and Development, Risk Reduction Engineering Laboratory
(RREL), Cincinnati, Ohio. Contributors and reviewers for this report were Steve Rock and Michelle Simon
of EPA RREL, and John Markiewicz, Chris Shallice, and Dan Driscoll of CF Systems (CFS). Mr.
Markiewicz served as the CFS Project Manager and Mr. Driscoll served as the CFS Pilot Plant Field Site
Manager; both assisted in the development of the test plan and were present during the majority of the
test.
This report was prepared for EPA's START Program by Science Applications International
Corporation (SAIC) in Cincinnati, Ohio for EPA under Contract No. 68-C0-0048. Joseph Tillman served
as the Work Assignment Manager. Authors include Mr. Tillman, Eric Saylor (Project Engineer), and Lauren
Drees (Quality Assurance Coordinator). Debbie Seibel, Jo-Ann Hockemeier, and Richard Dzija provided
formatting, graphics, and editorial support, respectively. Tom Wagner and Sharon Krietemeyer of SAIC,
as well as Mr. Meckes, assisted in field activities at the Springfield Township Dump regarding the
acquisition and preparation of test feed.
A special thanks is given to Hazen Research personnel who operated the pilot plant under the
direction of CFS, provided numerous field support activities, and allowed for the use of Hazen equipment
(such as laboratory balances, scales, and filter equipment), which proved to be invaluable for conducting
this study.
The authors would specifically like to thank Rod Hodgsen (Vice President), Steve Will (Lead
Engineer), Tom Pinnow (Lead Operator), Brad Leyhock (Feed Operator), Dan Pesusich (Product
Operator), and Mark Coppen (Shipping Agent), all of Hazen Research, Inc., for their efforts in conducting
the pilot test in conjunction with CFS and the EPA START Program.
x
-------
1.0 PROJECT BACKGROUND
1.1 Description of the Springfield Township Dump
The Springfield Township Dump (STD) Superfund site is located on the 16-acre
Nickson property in Springfield Township, Oakland County, Michigan, approximately 4 miles
south of the town of Davisburg {Figure 1-1). The site comprises approximately 4 acres
(400 ft. x 500 ft.) and is enclosed by a chain-link fence (Figure 1-2).
The STD was used as an industrial waste disposal site between 1966 and 1968.
Approximately 1,500 drums are reported to have been deposited on the site, mostly in the
central or main dumping area. Subsequent to the period of waste disposal, investigations by
the Oakland County Health Department (OCHD), the Michigan Department of Natural
Resources (MDNR), and U.S. Environmental Protection Agency (EPA) found both soil and
groundwater contamination at the site. Primary soil contaminants identified included
polychlorinated biphenyls (PCBs), volatile organic compounds (VOCs), the pesticide dieldrin,
and the metals arsenic, lead, and barium.
A Record of Decision (ROD) has specified a remedial action standard (RAS) of 1.0
mg/kg for PCBs in soil, which is the State of Michigan regulatory cleanup standard for those
compounds. Figure 1-2 shows the surficial extent of PCB contamination in soil at
concentrations exceeding the RAS of 1.0 mg/kg, as defined in a more recent Remedial Design
Field Investigation (CH2M-Hill, 1992). The ROD has also specified mobile incineration for
remediating soils at the site; however, negative public opinion toward incineration has led to
consideration of other cleanup alternatives. Based upon preliminary bench-scale testing of soil
samples taken from the site, solvent extraction, which is designed to separate semivolatile
organic compounds (SVOCs) (e.g., PCBs) from soils and other media, was one of the
alternatives being considered. Therefore, a treatability test was conducted on STD soils to
determine if a particular solvent extraction process could attain the RAS of < 1.0 mg/kg for
PCBs.
1-1
-------
N
MICHIGAN
Flint
I. Fenton
Springfield
Township
Dump
Davisburg
SPRINGFIELD
TOWNSHIP *
DUMP
l#o#
I AsT ^
' £
c
T
0
miles
Pontiac
%
°>f
Figure 1-1. Location of the Springfield Township dump.
1-2
-------
STD Site Boundary (Fenced)
4
N
A
A
0 100 feet
Approximate Scale
- = Road
Site Boundary
o Rl Shallow Soil Boring Sample
• RDFI Shallow Soil Boring Sample
Rl Deep Soil Boring Sample
RDFI Deep Soil Boring Sample
•*-PCB Concentration Between 1 and 10 mg/kg
PCB Concentration Greater than 10 mg/kg
Wooded/Brush Area Boundary
NOTES:
1 "Surface" is defined here as 0.0 to 2.0 feet below
grade.
2 A10-foot zone of PCB contamination greater than
1 mg/kg and less than 10 mg/kg is shown between
areas estimated with PCBs greater than 10 mg/kg
and the limit of the area to be remediated.
Figure 1-2. Plan view of the Springfield Township dump site showing surficial extent of PCB soil
contamination above 10 mg/kg and 1.0 mg/kg. (Source: CH2M Hill, 1992)
1-3
-------
1.2 Description of the CF Systems* Solvent Extraction Process
Solvent extraction is a non-destructive treatment technology that uses a solvating
agent to extract and separate contaminants from the media being treated. The primary goal
is to isolate and concentrate the organic contaminants into a small volume of waste material,
such as oil, which can then be destroyed by way of incineration or dechlorination or, in the
case of petroleum, reused. The primary features that distinguish different solvent extraction
technologies from one another are the solvent types used and the methods by which the
solvent is recovered and continuously reused in the extraction process.
CF Systems (CFS), a Morrisen Knudson company based in Woburn, MA, uses liquefied
gases in their process as solvents to extract organics from soils, sediments, sludges, and
wastewaters. Figure 1 -3 is a simplified diagram of the CFS process. For this treatability
study, CFS chose liquefied propane as the solvent and utilized a trailer-mounted pilot plant,
which is termed the "Mobile Demonstration Unit" or "MDU." This unit has recently been
relocated to Hazen Research, Inc's. 8-acre facility in Golden, Colorado. Figure 1 -4 shows the
MDU and identifies the major components, which are listed and briefly explained in Table 1-1.
The extraction process is a series of mix-settle-decant-refill cycles, where each cycle
is defined as an extraction stage. The number of extraction stages, mixing time, and settling
time are process parameters which are determined primarily by the degree of contaminant
removal required and the physical characteristics of the soil/sediment that is being treated.
The heart of the MDU is the Extraction Vessel ("extractor"), a 50-gallon vessel
containing a central impeller and injection ports at its bottom for adding solvent or water,
depending on the process stage. In a pilot-scale extraction process a quantity of
contaminated material is either loaded manually into the extractor or slurried externally and
then pumped into the vessel. The solvent is contained in two externally jacketed pre-heated
vessels (D-5 and D-8). The solvent is pumped from D-8 to D-5 where it is heated with the
aid of a hot oil tank that heats oil to high temperatures by pumping it through the jacket for
D-5. Once the solvent is sufficiently heated (the extractor is also heated simultaneously),
liquid solvent is introduced into the extractor from D-5 through the four symmetrically spaced
1-4
-------
Solvent/Organics Mixture
Treated Solids/
Water Mixture
(Slurry)
Water
Reclaim
Water
Feed
Vent Gas
Recovery
Solvent
Feed
Polymers
Added
Solvent
Recovery
Extraction
System
Organics
Recovery
Feed
Preparation
Filtrate
Recovered
Organics
(Oil Extract)
Filtration
System
Filter Cake
(Product Solids^
NOTE
Process path used for full-
scale system only
Figure 1-3. CF Systems® process diagram.
1-5
-------
Figure 1-4. Photograph of the Mobile Demonstration Unit showing major components.
-------
Table 1-1. Major Components of the Mobile Demonstration Unit
Components
CFS Designation
Description and Function
Extraction Vessel
Vessel D-2
Solvent Pre-Heat Vessel Vessel D-5
Solvent Surge Vessel Vessel D-8
Extraction Surge Vessel Vessel D-3
Distillation Tower Vessel T-1
Product Storage Vessel Vessel D-9
Primary Compressor
Secondary Compressor
Water Decanter Vessel
Solids Filter
Hot Oil Tank
Propane Storage Tank
C-1
C-2
Vessel D-4
F-t
Hot Oil Tank
Vessel D-t 1
A 50-gallon capacity, externally-jacketed vessel, mounted with a mixer for
contacting contaminated media with solvent.
Externally-jacketed vessel in which the solvent is heated to desired temperature and
stored for eventual use in D-2.
Surge vessel for recycled propane.
A receiving vessel for contaminant-laden solvent (extract) from D-2.
A single-stage flash tower/reboiler which receives the contaminated solvent from D-
3 and separates the solvent from dissolved contaminants.
A final holding tank for concentrated organic contaminants from T-1 which were
extracted from the feed material.
Receives solvent from T-1 and recompresses it for recycle to D-8.
Scavenges residual solvent vapor from D-9 and sends it to T-1.
Decanter vessel equipped with a water-sensitive probe. Used during water
displacement to ensure that remaining solvent is entirely removed from the
extractor.
Filters solids from contaminated solvent moving from D-2 on to D-3.
Heats oil to high temperatures and pumps it through D-2 and D-5 to aid in heating
the solvent.
Blow-down tank for excess liquefied propane that will eventually be flared out into
the atmosphere.
-------
ports in the bottom of the vessel by means of pressure differential. Once the solvent is trans-
ferred, the extractor mixer is started and the first extraction stage begins. After a pre-
determined amount of time {extraction cycles can range from 5 to 120 minutes), the mixer
is turned off and the solids are allowed to settle (typically less than 5 minutes). Since
liquefied gas solvents characteristically have a low density and low viscosity, settling is readily
accomplished. The physical and chemical properties of propane are presented in
Attachment A.
When settling is complete, a major portion of the contaminated solvent is drained from
the extractor to the Extract Surge Vessel (D-3) by means of pressure differential. After
draining, the drain port is closed and clean solvent is again introduced into the bottom of the
extractor. When the desired amount of solvent has been recharged, the vessel is isolated, the
mixer is turned on, and the next extraction stage begins. This process is repeated until the
desired number of extraction stages are completed. After the final mix and settlement have
occurred, the process of removing all the solvent contained in the extractor begins. Removal
of residual propane from the solids in the MDU is accomplished by displacing the solvent with
water. It is critical that all the solvent be removed from the extractor. Any solvent remaining
in the extractor will flash, consequently re-depositing organics onto the treated solids.
Residual solvent vapor remaining after the water displacement is then recovered by a
scavenger (secondary) compressor (C-2).
After the propane is removed, the extractor is depressurized, its bottom valve is
opened, and the water/solids mixture flows into a holding tank. This mixture is then filtered
and a filter cake is produced. The filter cake is then analyzed and compared to the feed to
determine the removal efficiency of the process.
Solvent containing dissolved organics (extract) is drained from the extractor through
a solids filter (F-1) and into the Extract Surge Vessel (D-3), which serves as the supply vessel
for the Distillation Tower (T-1). Extract is fed, via pressure differential, from the bottom of
the Surge Vessel into the Distillation Tower through a pneumatic control valve.
The Distillation Tower receives the contaminated solvent from D-3 and separates the
solvent from the dissolved contaminants. The tower consists of a horizontal reservoir that
contains a tube bundle (for heat exchange) and a vertical section where vaporized solvent
1-8
-------
flows out to the main compressor (C-1). Vaporization of the solvent is accomplished by heat
transfer in the horizontal reservoir section (reboiler). Hot, compressed solvent vapor flows on
the tube side of the tube bundle while the solvent/organic contaminant mixture boils along the
outside surface. The solvent, having a very low boiling point relative to the dissolved
organics, is vaporized and flows from the top of the tower through a mist and solids filter (F-
4) and into the inlet of C-1.
The main compressor recompresses the clean solvent vapor from the Distillation Tower
to a higher pressure. This hot vapor flows on the inside of the tube bundle in the reboiler
where it gives off its heat of compression and partially condenses. Full condensation occurs
immediately downstream by means of a water-cooled condenser. The condensed, distilled
solvent then flows to the Solvent Surge Vessel (D-8) where it is stored until fresh solvent is
required in the extractor.
The organic contaminants, which represent the oily extract that is ultimately
incinerated or disposed of, are periodically drained from the bottom of the Distillation Tower
into the Product Storage Vessel (D-9). The top of D-9 is connected to the inlet of the
Secondary Compressor (C-2). Residual solvent dissolved in the organics is vaporized in D-9
and is sent through a mist/fine solids filter (F-2) and onto the Secondary Compressor. D-9 is
also heated with hot oil to aid in vaporizing residual solvent from the organics. The Secondary
Compressor recompresses this solvent vapor and recycles it into the Distillation Tower.
Periodically, the concentrated organics are drained from the bottom of D-9 into a holding
vessel.
If excess propane vapor builds up in the unit, relief valves will automatically open to
relieve the pressure. The valves will direct propane through a vent line and through the blow-
down tank (D-11) before being flared to the atmosphere.
1-9
-------
2.0 EXPERIMENTAL DESIGN
The purpose of an experimental design is to ensure that a logical sequence of events
leading up to and including the treatability test is followed, and that those laboratory analyses
and field measurements critical for evaluating the technology are identified. This will result
in a scientifically sound technical evaluation of the CFS technology's effectiveness on the STD
soils. For this pilot-scale study a specific sequence of pre-test and test activities were
conducted to achieve the technical evaluation.
Figure 2-1 shows the sequence of events leading up to the treatability test. The
preliminary events focused on acquiring the most suitable test soil possible from the STD and
included immunoassay testing of soil to identify PCBs, excavation and mixing of a significant
test volume of soil, removal of oversize material a Vi-inch diameter, and bench-scale testing
of the CFS process. These activities are discussed in more detail in Section 3.1 (Feed
Preparation).
Soil Characterization
Screening and
Sampling (at STD)
Prescreening and
Homogenization of
Test Materia) (at STD)
Testing (at CFS)
Figure 2-1. Experimental design flow cfiagram.
Pilot-Scale
Testing
(at Hazen Research)
2.1 Test Objectives
There were two primary objectives for this treatability study, the more important of
which was determining whether the CFS process, utilizing liquefied propane as an extraction
agent, was effective for removal of PCBs from the STD soils to s 1.0 mg/kg on a dry weight
basis. The 1.0 mg/kg concentration value was chosen because it is the Remedial Action
Standard (RAS) for PCBs in soils at the STD (CH2M-Hill, 1992). The second primary objective
was to determine PCB concentration in the process water used to float propane and flush out
the treated solids to determine if any PCBs solubilized in the water. Both the primary and
secondary objectives for this treatability study are summarized in Table 2-1.
2-1
-------
Table 2-1. Test Objectives
Primary Objectives
• Determine effectiveness of removing PCBs from STD soils
to the RAS of < 1 mg/kg on a dry weight basis.
• Determine PCB concentrations in the filtrate water to ensure
proper disposal.
Secondary Objectives
• Determine residual concentrations of the pesticide dieldrin in
the product solids.
• Determine the total materials mass balance for the test.
• Determine the total PCBs mass balance for the test.
• Verify the absence of PCBs in the pilot plant prior to the
start of the treatability test.
2.2 Test Conditions and Optimization
The pilot test was conducted in two phases. Phase I of the test involved the
determination of optimum process variables. CFS, in accordance with the approved Quality
Assurance Project Plan (QAPP), was allowed the opportunity to test up to three sets of
conditions, choose that condition which could best meet the primary objectives, and then
twice repeat that condition. Therefore, Phase II of the treatability test consisted of two
additional test runs at the determined optimum conditions. This would yield a total of three
optimum runs, the purpose being to demonstrate process reproducibility. Figure 2-2 illustrates
this two-phased approach.
In addition to the five test runs performed there was also an additional run conducted.
This sixth run was not designated in the QAPP, but was deemed in the field as having good
potential of supplying useful information regarding the treatment of STD soils. The process
conditions for all six runs are summarized in Table 2-2.
2-2
-------
RUN 1
RUN 2
RUN 3
PHASE I
Optimum Condition Chosen
PHASE I!
RUN 4
RUNS
| Repeat of Optimum Condition
Figure 2-2, Two-phase approach for treatability test.
2.3 Sample Locations and Frequency
A total of five process streams were sampled and analyzed during the treatability test.
These included the following:
• Untreated soil (raw feed)
• Product solids (filter cake)
• Process water (filtrate)
• Product oil (organic extract)
• Recycled solvent
The organic contaminants of concern at the STD are PCBs and the pesticide dieldrin;
the PCBs were the primary target compounds for treatment during this pilot study. The
critical process streams included the raw feed, treated product solids (filter cake), and filtrate.
On larger-scale systems, CFS would reclaim (recycle) the filtrate; however, since this is not
conducted on the pilot unit, the filtrate is a critical stream that must be checked for residual
contaminants for disposal purposes.
Table 2-3 provides a summary of critical and non-critical analyses that were conducted
on each process stream.
2-3
-------
Table 2-2. Process Conditions for All Test Runs
MAIN TEST
SET CONDITIONS
VARIED CONDITIONS
Phase
Bun No.
Feed
Loaded
(lbs}
Mixing Time
Each Cycle
(mini
Mixing
Speed
Solvent/
Feed Ratio
by Weight
Each Cycle
Extraction
Pressure
(psi)
Avg/Range
Extraction
Temp (°F)
Avg/Range
No. of
Extraction
Cycles
Propane/
Feed
Separation
Method
1
1
100
20
Full
1.5/1
315 /250-409
133 / 125-138
3
Scavenge
only
2
100
20
O/Full*
1.5/ 1
261 / 223-308
122 / 106-133
4
Scavenge
+ water
3
100
20
Full
1.5/1
238 / 182-294
117 / 93-150
5
Scavenge
+ water
II
4
100
20
Full
1.5/1
266 / 202-309
124 /98-140
3
Scavenge
+ water
5
100
20
Full
1.5/1
243 / 194-299
119 / 98-137
3
Scavenge
+ water
Added Run
6
100
20
Full
1.5/1
277 / 231-319
125 / 110-138
2
Scavenge
+ water
a During one of the four extraction cycles, the mixer was inoperable; however, a solvent flow was established by recirculating propane
from the top of the extractor and into the bottom.
-------
Table 2-3. Summary of Critical and Noncriticai Analyses
Mot Plant
Rlrwate and
Racydod
Solvant
Product Solida
(RItar
Cake)
Separation
Water
(Filtrate)
Oil Product
(Organic
Extract)
Feed
Water
Untreated
Soil {Raw Feed)
Pecticidec
mwr.
ftmss
VOCi
\\v.
Propane
Total Metals
Wmm
volatile Solids
Oil & Grease
Parcel# Slza
Total Suspended
Solids
c Denotes a critical measurement
Figure 2-3 illustrates the sample and measurement locations in respect to the CFS
process. There were four standard analyses conducted on feed and product solids samples
for all six runs. These included PCBs/pesticides, oil and grease (O&G), volatile solids, and
moisture. Since PCBs were the analyte critical to the primary objectives of the study, PCB
analyses were also conducted on a pre-test toluene rinsate of the pilot plant, filtrate for all six
runs, the recycled solvent for the five main test runs, the source water added to the extractor
to flush solids, and the oily organic extract recovered following the sixth run. Moisture, the
other critical parameter, was also conducted on the organic extract. Supplemental analyses
were conducted on specified process streams for one of the Phase II runs {Run 4). These
included volatile organics on feed, product solids, filtrate and organic extract; total metals in
the feed, treated solids, and filtrate; particle size of feed; total solids in filtrate and source
water; and residual propane solvent in the product solids, filtrate, and organic extract.
2-5
-------
Oversize
I
Water
Reclaim
i
Feed
Preparation
Feed
ry«/j
Water
Feed
EXPLANATION
» Solid Sample Location
= Liquid Sample Location
M ) = Mass Measurement
© = Volume Measurement
ir w
Solvent
Feed
Extraction
System
Solvent/Organics Mixture
Treated Solids/
Water Mixture
(Slurry)
Solvent
Recovery
Filtration
System
Filtrate
Vent Gas
Recovery
Organics
Recovery
Filter Cake
(Product Solids)
Recovered
Organics
(Oil Extract)
©0
Figure 2-3. Sample and measurement locations.
2-6
-------
2.4 Analytical Methods
Table 2-4 lists the laboratory analytical methods used for determining the presence and
concentration of each analyte in each of the matrices sampled and analyzed. More detailed
information regarding these test methods can be found in Section 6 {Quality of the Data).
Table 2-4. Analytical Methods Used and References
Analyte
Matrix
Classification
Preparation
Method
Analytical
Method
Method
Type
Ref,
PC 83
Feed/Product Solids
C
3540
8080
GC/ECD
a
Filtrate
C
3520
8080
GC/ECD
a
Toluene Rinsate/
NC
3580
8080
GC/ECD
a
Recycled Solvent
Organic Extract
NC
3580
8080
GC/ECD
•
Moisture
Feed/Product Solids
C
NA
D2216
Gray.
d
VOCs
Feed/Product Solids
NC
5030
8260
GC/MS
- a
Organic Extract
NC
5030
8260
GC/MS
a
Filtrate
NC
5030
8260
GC/MS
a
Pesticides
Feed/Product Solids
NC
3540
8080
GC/ECD
a
Organic Extract
NC
3580
8080
GC/ECD
a
Filtrate
NC
3520
8030
GC/ECD
a
Volatile Solids
Feed/Product Solids
NC
NA
2540G
Grav.
b
Total Solids
Filtrate/Feed Water
NC
NA
160.2
Grav.
c
Moisture
Organic Extract
NC
NA
D1744
Grav.
d
Propane
Product Solids
NC
NA
GC/FID
a
Filtrate
NC
NA
GC/FID
e
Organic Extract
NC
NA
GC/FID
e
Oil & Grease
Feed/Product Solids
NC
NA
9071
Grav.
a
Filtrate
NC
NA
9070
Grav.
a
Particle Size
Feed
NC
NA
D422
Grav.
d
Metals
Feed/Product Solids/Filtrate
NC
Sol./Aq
Sol./Aq
- As
3050/7060
7060/7060
AA
a
- Ba
3050/3010
6010/6010
ICP
a
- Cd
3050/3010
6010/6010
ICP
a
- Cr
3050/3010
6010/6010
ICP
a
-Cu
3050/3010
6010/6010
ICP
a
- Fe
3050/3010
6010/8010
ICP
a
- Pb
3050/3010
0010/6010
ICP
a
- Mn
3050/3010
6010/8010
ICP
a
-Hg
7471/7470
7471/7470
AA
a
- Ni
3050/3010
6010/6010
ICP
a
- Se
3050/7740
7740/7740
AA
a
- Zn
3050/3010
6010/6010
ICP
a
Key: C » Critical Aq -
NC — Nonoritical GC —
NA - Not Applicable ECD -
Sol ¦ Solid Matrix MS *
Aqueous Matrix FID
Gas Chromatograph Gray
Electron Capture Detector AA
Mass Spectrometry ICP
Flame Ionization Detector
Gravimetric
Atomic Absorption
Inductively Coupled Plasma
References:
a
b
c
d
e
Test Methods for Evaluating Solid Waste, SW-846, Third Edition, 1986.
Standard Methods for the Analysis of Water and Wastewater, 17th Edition, 19S9,
EPA Methods for Chemical Analysis of Water and Wastes, 1983.
Annual Book of ASTM Standards, American Society for Testing and Materials.
These procedures were validated by SAIC as part of the treatability study.
2-7
-------
3.0 FIELD ACTIVITIES
3.1 Feed Preparation
Preparation of the test feed material was initiated with the excavation of the material
from the STD site. Many of the survey stakes marking surveyed grid locations (longitude and
latitude) were found clearly marked and intact. From these stakes, the approximate boundary
of the > 10 mg/kg PC8 contaminated zone, as defined on maps from the Remedial Design
Field Investigation (CH2M-Hill, 1992), was located. A 10 ft. x 15 ft. grid was then
sequentially established by excavating holes with a common post-hole digger 5. ft. apart and
down to 3 ft. below surface. Immunoassay tests were conducted as the holes were
excavated to determine qualitatively the presence of PCBs. Samples were collected from the
excavated material and sent to an analytical laboratory for GC/ECD confirmation for PCBs.
Figure 3-1 shows the location within the STD where the treatability test material was
acquired. A total of approximately 1,158 lbs, of material was excavated from this location.
Approximately 623 lbs. of the material came from the test holes, and the rest came from a
trench excavated within the grid. All the material was screened to remove oversize material
> '/2-inch diameter in the field using a large piece of plastic crating material. Approximately
168 lbs. of reject (14.5% of original mass) were initially removed. The screened material was
then shipped to the CFS facility in Woburn, MA.
CFS used approximately 150 lbs. of the screened material to conduct a series of
bench-scale tests in order to establish basic operating conditions for the treatability study.
The remaining material (approximately 1,000 lbs.) and the MDU unit were shipped to Hazen
Research, Inc. in Golden, Colorado for the treatability study. At Hazen Research, the feed
material was air-dried, further screened to remove oversize > %-inch diameter, and mixed to
produce a homogenous test feed. Figure 3-2 shows the test soil preparation activities, and
Table 3-1 summarizes the results of the test soil screening for removal of oversize material.
3-1
-------
N9500
N9400
N9300
N9200
N9100
STD Site Boundary
0 100 ft
Approximate Seals
50400E 50500E 50600E 507QQE 50800E 50900E
PCB concentration between 1 and 10 mg/kg
PCB concentration greater than 10 mg/kg
Figure 3-1. Test soil excavation location.
3-2
-------
(b) Initial screening of oversize >1/2" diameter was
conducted at the STD using plastic crating for
efficiency. Oversize this large would also be
screened out during a site remediation, unless it
was reduced to a smaller size (i.e. pulverized).
(a) The test soil excavation area at the STD
consisted of a staked rectangular plot, in
which a trench was dug in the area having
the highest immunoassay readings.
(c) At Hazen Research, the material was
air-dried on a liner and mixed to
homogenize as weil as possible.
(d) The test soil was then sieved to remove
material >1/4" diameter. This oversize
material may or may not be screened
during a site remediation, depending on
the full-scale design.
Figure 3-2. Test soil preparation activities.
3-3
-------
Table 3-1. Percentage of Screened Oversize Material
Location Screened
Starting
Material (lbs.)
Material
Screened (lbs.)
% Oversize
Springfield Township Dump
1,158
168"
~ 14.5
Hazen Research
~ 626
26b
~ 4.0
Total % oversize a V* inch
~ 18.5
a Using plastic crating having approximately 54-inch openings,
b Using an ASTM sieve having 54 -Inch openings.
3.2 Sample Collection
The process streams sampled for each of the five main test runs included the raw feed
(untreated soil), filter cake (product solids), filtrate (separated process water), and the recycled
solvent (a hexane rinsate). All of these streams, except for the recycled solvent, were also
sampled for the added Run 6. In addition to these primary streams, the organic extract
(product oil) was sampled at the end of Run 6 and quality control (GO blanks on the pilot
plant and water used to flush treated solids were also collected. The following subsections
detail the sample collection procedures for all process streams sampled, in approximate
chronological order. Figure 3-3 shows some of the field process operations of the treatability
study, and Figure 3-4 shows some of the resulting sample collection and field measurement
activities.
3-2.1 Pilot-Plant Equipment Blank
The pilot-plant equipment blank was the first sample collected for the treatability test.
This sample consisted of a toluene rinsate of the MDU's interior tank and piping walls to
determine if any residual PCBs were present from past pilot tests. To rinse the unit, CFS
loaded approximately 5 gallons of toluene directly into the extractor followed by approximately
150 lbs. of liquefied propane. The propane/toluene mixture was then forced under pressure
through the process flow components. Toluene is added since, unlike propane, it is a liquid
at ambient pressures. The toluene ended up in the Product Storage Vessel (D-9) where it was
easily drained by gravity. After the initial 5 gallons of toluene rinsate were drained from the
unit and sealed in a container, another 5 gallons of fresh toluene were added to the extractor.
The entire process was repeated, and a sample of the rinsate was collected directly into 40-
rnL vials as it drained from D-9. This sample was analyzed for PCBs.
3-4
-------
(a) Feed loading
(b) Propane extraction
•Vessel D-2
(c) Flushing product solids (d) Vacuum filtration
-as a slurry
Figure 3-3. Process operation activities.
3-5
-------
Product
Sotvwtf
Recovery
Column
Column gn \
Rabollf £| j
(a) Sampling product solids
•after filtration
(c) Sampling oily organic extract {the
froth is volatilizing propane)
(b) Sampling filtrate
-from holding tank
(d) Weighing process outputs to
determine mass balances
Figure 3-4. Sample collection and field measurement activities.
3-6
-------
3.2.2. Tap Water Feed Blank
CFS added normal tap water to aid the extractor in floating out any residual propane
solvent and flushing treated solids from the extractor at the end of a run. Since the process
water potentially contacts PCBs and would be analyzed for that parameter at the completion
of each run, it was essential to analyze the source of the water for PCBs to ensure that the
target analyte was not in the water from the beginning. The tap water feed sample was
collected by filling 1,000-mL bottles directly from a hose.
3.2.3. Raw Feed
Samples of the raw feed material were collected just prior to loading the feed for each
of the six test runs. The amount of feed loaded for each of the runs was always measured
out as 100 lbs. of air-dried material and contained in two 5-gallon buckets. A grab sample
of the feed was collected directly into 250-mL jars from the approximate middle of the batch
loaded (at the 50-lb. interval). The only exception to this procedure was during the collection
of the field duplicate sample for Run 4. To account for field variability, the field duplicate was
collected from the top of the second bucket and thus, spaced about 50 lbs. of feed away
from the primary sample.
3.2.4 Filter Cake
After the product solids are piped from the extractor to a lined drum as a slurry, the
slurry must undergo a process to separate the majority of the water from the solids. For this
treatability test, an existing vacuum filtration system at Hazen Research was utilized. This
system consists of a 4 ft. x 4 ft. x 0.8 ft. pan filter lined with a 5- to 10-micron filter mesh,
which is attached to a vacuum pump. The pump applies a vacuum along the entire surface
area of the pan and transfers filtrate to a holding tank (refer to Figure 3-3).
While the vacuum pump was in operation, the slurry was transferred from the lined
drum and poured onto the mesh one bucket at a time. It took several hours for the water to
be filtered out. The efficiency of the process was at its greatest when the solids were spread
over the entire pan surface area; this formed a seal which increased the effect of the applied
vacuum.
3-7
-------
The filter cake material, which visually appeared similar to the raw feed before it was
air-dried, was sampled using two different procedures. The original plan was to collect grab
samples from the pile of solids placed in the center of the pan filter, after the majority of
water drained off; this was prior to spreading the solids out evenly across the filter mesh. The
procedure was followed for the first three runs and resulted in wet samples. Because the
solids are vigorously mixed in the extractor, and then discharged as a slurry, there is every
reason to believe that the mass of solids discharged following each of the six runs was
homogenous. However, during vacuum filtration of the slurry, there is a possibility of
segregating fines, especially when spreading the solids over the entire surface area of the pan
filter to maximize vacuum filtration efficiency.
For the last three runs, a composite sample of filter cake was taken following
spreading of the solids and after the dewatering process was completed. After dewatering,
the solids were accumulated into the center of the filter mesh; fines were then scraped from
the mesh and mixed into the pile with a shovel. Small amounts of solids were then collected
from different locations of the dewatered pile of solids using a clean trowel and accumulated
and mixed in a clean bucket. The samples were collected from the bucket in 250-mL sample
jars. For the field duplicate sample collected for Run 4 treated solids, the entire compositing
process was repeated to account for field variability.
3.2.5 Filtrate
During the vacuum filtration process, the filtrate is collected from the bottom of the
pan filter and pumped into an adjacent holding tank. At the bottom of this tank there was a
discharge hose which allowed for easy sample collection of filtrate directly into sample
bottles. To obtain representative samples of the filtrate, approximately 5 gallons of filtrate
were purged into a bucket before collecting the sample. It was observed that some fines do
penetrate the filter cloth and enter the filtrate tank, since the initial filtrate purge had a light
brownish color that diminished somewhat as the tank was emptied. For collection of the Run
4 duplicate sample, an additional 5 gallons of filtrate were purged prior to collecting the
duplicate, again to account for field variability.
3-8
-------
3.2.6 Recycled Solvent
Sampling to determine whether PCBs remained in the recycled propane involved a
unique procedure. To sample liquefied propane, CFS uses a large steel propane canister
("bomb" sampler) that weighs roughly 2214 lbs. empty and can hold approximately 1 gallon
of liquid propane. To acquire a propane sample, the bomb is fitted to a valve connected to
a line extending from the Solvent Surge Tank (D-8). After each of the five test runs, the
bomb was filled with the liquefied propane that had been recovered from the preceding run.
To trap any residual organics (i.e., PCBs) and separate them from the solvent, the
preweighed bomb was propped at an angle while the top valve was opened and the propane
was slowly vented. Once the propane had been fully vented, any nonvolatile species should
have remained within the interior of the bomb. Hexane was used to rinse the bomb to collect
residual. Approximately 80 mL of hexane was poured into one end of the bomb, the valve
was then closed, and the bomb was shaken and rotated to thoroughly rinse the interior of the
bomb. The 80 mL of hexane was then collected from the bomb directly into two 40-mL vials.
For the Run 4 duplicate sample, the bomb was refilled with Run 4 recovered solvent and the
same sampling procedure was followed.
3.2.7 Qrggnic Extract
The organic extract was accumulated in the organic extract product tank each run, and
was sampled only at the end of the entire treatability study. This would ensure ample sample
volume of product oil and would only necessitate a one-time handling of the extract, which
contained the concentrated organics and was by far the most toxic of the process streams.
The extract is gravity drained from the Product Storage Tank (D-9) into a plastic
container. There is a liner and a bermed area beneath this portion of the pilot plant. Prior to
sampling the extract, a small volume of oil was purged. To safely sample this material, a
wide-mouth, glass, pyrex cup was used to intercept the oil as it drained from the tank (Figure
3-4). Once in the pyrex cup, the oily extract was poured into 40-mL vials. Following
collection of this primary sample, another small volume of oil was purged and a field duplicate
sample was collected using the same procedure.
3-9
-------
3.3 Field Measurements
Field measurements were taken during ail six process runs to determine if ail materials
loaded into the MDU could be accounted for. This was necessary to determine mass balance
closure for each run. Table 3-2 summarizes the field measurements that were recorded during
the treatability test, the instrument used for the measurement, and the measurement
frequency. The actual data are presented in Section 5 (Mass Balance) and in Attachment 6.
Table 3-2. Summary of Held Measurements
Measurement
instrument Used
Instrument
Readability
Frequency
INo. of
Measurements)
Mass of soil feed Platform scale
Mass of process water added Mass flow meter
Mass flow meter
Mass of solvent added to
extractor
Mass of sand filler added
Mass of F-1 filter
Mass of slurry
Mass of filter cake
Mass of filtrate
Mass of organic extract
Propane sample canister
Platform scale
Platform Scale/
Lab balance
Platform scale
Platform scale/Lab
balance
Platform scale/Lab
balance
Lab balance
Lab balance
± % lb
± 0.1 lb
± 0.1 lb
± % lb
± % lb/
± .01 gram
± % Ib/y« lb'
± % lb/
± .01 gram
± 54 lb I
± .01 gram
± 0.1 g
± .01 gram
Each run (6)
Each run (6)
Each run (6)
Each run following
Run 1 (5)
Before and after
changeout (2)
Each run (6)
Each run (6)
Each run (6)
After Run 6 (1)
Before and after
filling (5)
a Two platform tcalM, having diffarent readability, ware uaad at diffarant timaa.
3.4 Process Monitoring
Process monitoring was conducted by the Superfund Technical Assistance Response
Team (START), and concurrently by Hazen Research personnel who were operating and
monitoring the MDU under the direction of CF Systems personnel. START and Hazen
Research took readings every half hour, while staggering the readings so that process data
was recorded every 15 minutes. Identical process log forms, tailored for this treatability test,
were used by both parties.
3-10
-------
The MDU functions are largely enacted by a computer within the control room. A
computer graphic display is used to make adjustments to the process conditions and monitors
process measurements. As a backup, standard analog measurement devices such as pressure
and temperature gauges, rotometer, etc. are installed throughout the various MDU
components. A portion of the process measurement readings could be taken both in the
control room and the analog devices, in these cases, the reading taken from the control room
was considered to be the actual reading.
The primary process measurements taken are as follows:
• Coolant water inlet and outlet temperature
• Extractor vessel pressure and temperature
• Solvent preheat vessel pressure
• Extract surge vessel pressure
• Solvent surge vessel pressure
• Still pressure and temperature
• Main compressor pressure and temperature
• Scavenger compressor pressure and temperature
• Combustible gas sensor readings
• Hot oil tank pressure, temperature, and level
• Nitrogen supply seal pot, hot oil, and vent header pressure
• Water displacement pressure (D-4)
• Solids filter pressure (F-1 filter)
• Propane pump pressures
Table 3-3 summarizes the instruments used and their readability for each of the primary
process measurements. Tables 3-4 and 3-5 provide the average and ranges, respectively, for
the primary process measurements recorded during the treatability test.
3-11
-------
Table 3-3. Process Measurement Instrumentation
Measurement
Instrument
Reading
Increment
Instrument
Ranges
Coolant Water
Temperature
Analog Gauge
2"F
0-250'F
Extractor Vessel Pressure
Computer Graphic Display
1 psi
NA
Extractor Vessel
Temperature
Computer Graphic Display
1*F
NA
Solvent Preheat Vessel
Pressure
Computer Graphic Display
1 psi
NA
Solvent Surge Pressure
Computer Graphic Display
1 psi
NA
Still Pressure
Computer Graphic Display
1 psi
NA
Still Temperatures
Computer Graphic Display
1°F
NA
Main Compressor Pressure
Computer Graphic Display
1 psi
NA
Main Compressor
Temperature
Computer Graphic Display
1*F
NA
Scavenger Compressor
Pressure
Computer Graphic Display
1 psi
NA
Scavenger Compressor
Temperature
Computer Graphic Display
1°F
NA
Combustibles
Computer Graphic Display
0.0001 %
NA
Hot Oil Tank Pressure
Analog Gauge
5 psi
0-60 psi
Hot Oil Tank Temperature
Analog Gauge
5°F
50-550°F
Hot Oil Tank Level
Magnehelic Gauge
O.S inches
0-25" H20
Nitrogen Supply Seal
Pot Pressure
Analog Gauge
50 psi/100 psi
0-2,000 psi/
0-4,000 psi
Nitrogen Supply Hot Oil
Pressure
Analog Gauge
5 psi/100 psi
0-200 psi/
0-4,000 psi
Nitrogen Supply Vent
Header Pressure
Analog Gauge
5 psi/100 psi
0-200 psi/
0-4,000 psi
Water Displacement
Pressure
Analog Gauge
10 psi
0-400 psi
F-1 Filter Pressure
Analog Gauge
5 psi
0-50 psi
Propane Pump Pressures
Analog Gauge
10 psi
0-600 psi
NA = Not Applicable
3-12
-------
Table 3-4. Averages of Process Measurements
Run
Coolant Water
Temparatu#e
m
Extractor Vaaaal
(D 21
Solvant Prahaat Praaaura (p«i)
(D-6J
Extract Surge Praaaura (pal)
ID-31
Servant Sur^a
Praaaura
toei
Still iT-K
Main
Compressor
(c-n
Mat
Outlat
Press.
(psil
fe
Press
ipait
Temp.*
m
Praaa.
ipail
Tamp
m
1
63
67
316
133
463
148
206
01
67,64,87
200
187
2
63
64
261
122
426
144
163
70
61,60,76
168
140
3
64
64
238
117
427
140
147
76
68,63,70
147
138
4
63
63
266
124
468
142
168
70
68,67,76
160
147
6
66
64
243
119
466
141
146
74
64,62,72
146
143
3
64
64
2??
126
462
130
160
02
64,64,78
160
144
Bun
Scavangar Compressor
iC-21
CombuaUtdaa
Hot OH Tank
Nitrogan Supply Pressure
ipalJ
W»tw
Displacement
Prosaura
lpsj|
Filtar
{F* 11 Praaaura
(pal}
Propane Pump Praaaura*
ipai>
Press.
fpait
Tamp.
m
Area 1
Araa 2
Praaa.
Ipatl
Tamp.
m
Laval
lin.J
Seel
Pot
Hot Oil
Vant
Header
P*2
P 3
1
27
162
0.1221
0.2011
16
284
11
660
200
20
123
0
166
178
2
11
88
0.1221
0.407
13
283
11
660
88
21
107
0
216
106
3
6
83
0.13
0.2062
16
222
11
646
67
22
130
0
163
163
4
23
103
0.68
0.4803
16
276
13
647
113
42
206
1
176
187
6
16
04
0.1322
0.3687
16
261
12
660
71
41
210
0
132
146
6
22
111
0.1300
0.3040
14
266
12
660
76
47
76
0
t«1
181
a Bottom, middle, and top readings, respectively.
-------
Table 3-5. Ranges of Process Measurements
Run
C a dent Water Tempo turn
m
Extractor V
©-2J
fiaeaei
Setoent Preheat
Praeeure (pail
©¦61
Extract Surge
Praeeure (pail
©-3I
t Surpe Preeetw
m-m
-------
Table 3-5. Ranges of Process Measurements (Continued)
Run
Nitrogen Supply Pram**
M
Wfltw OkplAMnMnt
Pri—urt
im)
©-4)
RH«r IF-1}
Ptmmjt*
(P«)
Preparta Pump Prmmurm
itml
4
626-460 82-136
40-44
206-206
0-10
148-210
168-220
5
d2S-«?6 SO-SO
40-46
180-226
0
120-160
136-160
a
860 70-96
40-60
74-60
0
166-170
166-166
3-15
-------
4.0 ANALYTICAL RESULTS
This section summarizes the laboratory analytical results used in evaluating the
effectiveness of the CF Systems* pilot-scale solvent extraction process for treating soils
acquired from the Springfield Township Dump. The data are presented in the selected order
of importance, beginning with PCS and moisture results for raw feed, filter cake (product
solids), filtrate (process water), recycled solvent, and organic extract (product oil). For all
parameters, raw feed and filter cake results are reported on a dry weight basis. The quality
of the data are discussed fully in Section 6.
4.1 PCBs - Feed and Product Solids (Runs 1-5)
The raw feed material and product solids were analyzed for total PCBs for each of the
five main test runs. Table 4-1 presents the PCS concentrations for feed versus product solids
on a dry weight basis and provides the percent removal results.
Table 4-1. PCS Concentrations and Removal Efficiencies - Feed and
Product Solids (Runs 1-5)
Run
Number
Feed (mg/kg)*
Product Solids (mg/kg)*
Percent
Removal
1
210
4.9
97.7
2
240
1.8
99.3
3
340
2.2
99.4
4
310b
4.0b
98.7
5
220
5.8
97.4
• Aroclor 12S4 was the only PCS identified; concentration! are rounded to two significant digits,
b Average concentration of analyses of field duplicate samples rounded to two significant digits.
c Two values are given, the first pertains to all five runs and the second pertains to Runs 1, 4, and 5, all of which involved
three extraction cycles.
4-1
-------
The results indicate that the CFS process was able to achieve high removal efficiencies
ranging from 97.4 to > 99 percent. However, the primary objective of attaining product solid
concentrations of < 1.0 mg/kg by the specified test method was not attained. Two of the
runs, however, did attain results just above and below the Toxic Substances Control Act
(TSCA) regulatory limit of 2.0 mg/kg. There is not enough PCS data to indicate that using five
extraction cycles at 20 minutes each (conducted during Run 3) produced better PCB
extraction results than the 4-cycle run used for Run 2. However, the data shows that the 4-
and 5-cycle runs (Runs 2 and 3, respectively) did outperform the other three runs consisting
of three cycles. The average of the product solids concentrations from the three 3-cycle runs
is 4.9 mg/kg. The average of the product solids concentrations from Runs 2 and 3 is 2.0
mg/kg.
4.2 Moisture - Feed and Product Solids (Runs 1-5)
Moisture was a critical analysis conducted on the raw feed and product solids primarily
to ensure accurate PCB values for those samples on a dry weight basis. Table 4-2 presents
the moisture results for the feed material as loaded into the extractor for each of the five main
test runs (after air drying), and the moisture contents of each of the corresponding product
solids following filtration.
Table 4-2. Moisture - Feed and Product Solids (Runs 1-5)
Run Number
Feed Moisture (%)
Product Solids
Moisture (%)
1
2.44
11.4
2
1.38
15.4
3
1.45
15.9
4
2.33
15.3
5
1.55
15.4
Avitmaa
1.83
4.3 PCBs - Filtrate and Recycled Solvent (Runs 1 -5), and Product Oil
The second primary objective of the treatability study was to determine if any PCBs
were solubilized in the process water used for floating residual propane from solids and for
4-2
-------
flushing the solids from the extractor. The recovered solvent was sampled at the end of each
of the five main test runs to evaluate the solvent recycling process, and the organic extract
was sampled at the end of the test to conduct a mass balance for determining if all extracted
organics were recovered from the pilot unit. These latter two evaluations were conducted as
secondary objectives.
Table 4-3 presents the PCB results for filtrate (process water), recycled solvent, and
product oil (organic extract).
Table 4-3. PCB Concentrations - Filtrate and Recycled Solvent (Runs 1 -5), and Product Oil
Run
Number
Filtrate
U/Q/U
Recycled Solvent
U/g/L)
Product Oil (Extract)
(mg/ko)b
< 1
< 1
< 1
< 1
< 1
< 38
< 16
< 16
< 25"
< 27
a Average concentration of analyses of field duplicate sample* rounded to one, two or three significant digits,
b The oily organic extract was accumulated during the entire teat in the extract product tank and was sampled
at the conclusion of Run 8.
4.4 O&G - Feed and Product Solids (Runs 1-5)
Oil and grease (O&G) analysis (SW-846 9071) is an inexpensive gravimetric test
method that gives gross determinations of low-volatile organic content in a variety of
matrices. Because of its broad application, it can be used as an indicator analysis for virtually
any study involving semivolatile and nonvolatile contaminants. Thus, O&G analysis can
provide comparability of data among those studies, even though specific target analytes of
those studies are different. O&G analysis was conducted on feed and product solids for each
test run to make a semi-quantitative evaluation of the effectiveness of the CFS solvent
extraction process in removing other organic compounds in the feed, in addition to the target
PCB analyte. Table 4-4 presents the results of O&G analyses conducted on feed and product
solids for the five main test runs.
4-3
-------
Table 4-4. O&G Concentrations and Removal Efficiencies - Feed and
Product Solids (Runs 1 -5)
Run
Percent
Number
Feed (mg/kg)
Product Solids (mg/kg)
Removal
1
4,480
112
97.5
2
4,560
73
98.4
3
5,870
< 20
> 99.6
4
5,460
133
97.6
5
5,140
93
98.2
Ittiipoiiiiiiiiiiii
iiiiiiJiiiiiiiiBS
& 98.3
The results of the O&G analyses on feed indicate that there is a significant amount of
semivoiatile and nonvolatile organic material present in the soil, other than PCBs. Also, based
on the limited amount of data, it appears that the increased number of extraction cycles (5)
conducted during Run 3 did result in the highest percentage of O&G removal and generated
the "cleanest" product solids based upon O&G. It should be noted, however, that this
interpretation of results is based only on one O&G analysis per run, and thus should be
supplemented with additional data.
4.5 Volatile Solids - Feed and Product Solids (Runs 1-5)
Volatile solids analysis (EPA 2540G5 is another inexpensive gravimetric test method
that provides a more gross estimation of solids content than O&G. Unlike O&G, volatile solids
includes volatile materials, inorganic forms of carbon, and certain organics not expected to
be extracted by propane (i.e., humic material from decayed leaves, etc.) that may be present
in the feed solids. Table 4-5 presents the results of the volatile solids analyses conducted on
feed and product solids for each of the five main test runs.
The results of the volatile solids analyses should be viewed only as a rough
approximation of the amount of organic matter in the solid fractions before and after
treatment. These results are largely inconclusive, based on the fact that the product solids
for Runs 4 and 5 were reported as higher than the feed material. It is unlikely that volatile
solids content went up. The reported QC results (see Section 6) would suggest that analytical
4-4
-------
Table 4-5. Volatile Solids Concentrations and Removal Efficiencies -
Feed and Product Solids (Runs 1-5}
Run Percent
Number Feed <%) Product Solids {%) Change
1 3.0 3.0 NC
2 3.7 1.4 62
3 3.3 1.0 70
4 3.9 4.4 NC
5 3.6 4.3 NC
NC - Not Calculated
error could be responsible for the anomalous results and that there is no difference between
the feed and product solids. However, based on the reported values in Table 4-5, it appears
that more volatile solids were extracted during Runs 2 and 3, which utilized four and five
extraction cycles, respectively.
4.6 Particle Size Distribution of Feed
Particle size distribution of the feed material was determined primarily to identify the
amount of fine-grained sand and fines (silt and clay) present in the feed material. The particle
size of the material being treated can be an important aspect when evaluating the
effectiveness of a solvent extraction process, especially when chlorinated organic compounds
(e.g., PCBs) are the contaminant of concern. Table 4-6 presents the results of the particle
size distribution analysis of feed, which shows that the feed was approximately 85 percent
sand. The porosity of a sandy soil may diminish the importance of a mixing stage.
4.7 Residual Propane - Product Solids, Filtrate, and Product Oil
Propane was the extracting agent used by CFS during the treatability test. As part of
the solvent extraction process, the propane is recycled and reused in extraction cycles
conducted after the initial extraction takes place (refer to Subsection 1.2 for process descrip-
4-5
-------
Table 4-6. Particle Size Distribution of Feed*
Particle Size Analysis (ASTM D422)
% Accumulative % Accumulative
Sieve No.
Mesh Size (mm)
Passing
Retained
Classification
4
100
100
0
8
50
97.6
2.4
Coarse Sand
16
25
96.3
3.7
30
12.5
92.3
7.7
Medium Sand
50
4.75
52.7
47.3
100
1.18
19.6
80.4
200
0.600
14.9
85.1
Fine Sand
Hydrometer Analysis
Particle Size (mm)
0.036
5.9
94.1
0.018
4.4
95.6
0.009
2.9
97.1
0.005
1.5
98.5
a Distribution does not include oversize material a %-in diameter, which was removed prior to extraction treatment.
tion). Because both the solvent arid process water contact the contaminated solids, those
process streams, as well as the product oil, were analyzed for residual propane. Although
propane itself is a rather innocuous gas at low concentrations, it is still common to test for
residual solvent in any solvent extraction process to ensure that treated solids are not
hazardous by characteristic, and to better evaluate the solvent recovery efficiency of the
process.
Analyzing for propane at low levels is uncommon, and an approved EPA test method
does not exist. Therefore, methods were developed to conduct the analyses of the product
solids, filtrate, and product oil. The concentration of propane in the product solids was
determined by leaching the solid sample with reagent water and separating the propane in a
packed GC/FID column. The concentration of propane in the filtrate was measured by direct
injection, and the concentration of propane in the product oil was also measured by direct
4-6
-------
injection following a 1,000-fold dilution with hexadecane. These methods are fully described
in a separate validation study report {SAIC Methods Laboratory, October 17, 1994). Table
4-7 presents the analytical results for determining residual propane in product solids, filtrate,
and product oil. The data indicate, as expected, that virtually none of the solvent is retained
in the solids and filtrate, and a relatively significant amount is present in the product oil. In
commercial applications, the oil would go under additional solvent distillation.
Table 4-7. Residual Propane - Product Solids, Filtrate, and Product Oil
Process Stream Concentrations
Analyte Product Solids
(mg/kgj Filtrate (mg/L) Product Oil (mg/L)
Propane 9.83 11.3 12,500*
a Average concentration of analyses of field duplicate samples.
4.8 Supplemental Analyses
Supplemental analyses were performed on feed, product solids, and filtrate samples
collected during Run 4, and on product oil samples collected following all test runs. In
addition, the pesticide dieldrin was analyzed in solids and filtrate for each test run, and in the
product oil. The primary purpose for conducting these noncritical analyses was to better
characterize those process streams. Pesticides and volatile organics were analyzed in all four
of those matrices because the pesticide dieldrin and VOCs had been reported to occur in the
STD soils {CH2M-Hill, 1992). The solids and filtrate were also analyzed for total metals
because metals (specifically arsenic, barium, and lead) were detected in the STD soils
-------
Table 4-8. Dieldrin Analyses on Solids and Filtrate (Runs 1-5), and Product Oil
Process Stream Concentrations
Run Number
Feed"
Product Solids"
Filtrate
Product Oil
(zug/kg)
i//g/kg)
(pg/U
Smg/kgi
1
< 3.3
< 3.3
< 0.1
...
2
< 3.3
< 3.3
< 0.1
...
3
< 3.3
< 3.3
< 0.1
...
4
< 3.3"
< 3.3"
< 0.1"
...
5
< 3.3
< 3.3
< 0.1
...
Average
< 3.3
< 0.1
TOTAt < 33",b
a Average concentration of analyses of field duplicate samples,
b The product oil was sampled at the end of the entire test, following Run 6,
Table 4-9. Volatile Organic Analyses on Solids, Filtrate, and Product Oil
Process Stream Concentrations
Analyte
Feed
Product Solids
Filtrate
Product Oil
(/vg/kg)
(//g/kg)
U/g/L)
(mg/kg)
Acetone
< 5
35
210
27
Carbon Disulfide
< 5
< 5
< 5
1.3
Chloroform
< 5
< 5
< 5
8
Benzene
< 5
3.7
< 5
55
Toluene
1.7
29,000
13
225,000
Ethylbenzene
< 5
6.8
< 5
92
m,p-Xylenes
< 5
21
< 5
310
o-Xylene
< 5
5.8
< 5
94
4-8
-------
Table 4-10. Total Metals in Solids and Filtrate
Process Stream Concentrations
Analyte
Feed
(mg/kgj
Product Solids
{mg/kg)
Filtrate
(mg/L)
Aluminum
4600
4600
0.73
Antimony
14
46
0.080
Arsenic
9.0
6.2
0.059
Barium
1200
1700
0.29
Beryllium
< 1
< 1
< 0.005
Cadmium
7.7
9.2
0.0054
Calcium
15,000
16,000
37
Chromium
91
150
0.032
Cobalt
< 10
< 10
< 0.050
Copper
40
130
0.058
Iron
11,000
30,000
1.5
Lead
670
1,100
0.18
Magnesium
2,800
2,700
8.3
Manganese
170
270
0.27
Mercury
< 0.1
0.2
0.0030
Molybdenum
< 13
15
0.12
Nickel
21
59
< 0.040
Potassium
< 1,000
< 1,000
32
Selenium
1.4
< 1.0
< 0.005
Silver
< 2
< 2
< 0.010
Sodium
< 1,000
< 1,000
20
Thallium
< 2
< 2
< 0.010
Vanadium
< 10
< 10
< 0.050
Zinc
1900
2700
0.80
4-9
-------
4.9 Run 6 Results
A sixth run was added to the treatability study and conducted immediately following
the five runs comprising the main treatability test. This run was conducted to determine the
effectiveness of the CFS process if only two 20-minute extraction cycles were conducted.
Table 4-11 summarizes the results of all analyses conducted on Run 6 feed, product solids,
and filtrate. As indicated by the data, Run 6 resulted in the poorest performance of all runs
conducted.
Table 4-11. Analytical Results for Run 6
Analyte
Process Stream PCBs Moisture O&G Volatile Solids
Feed 220 mg/kg 1.90% 7,060 mg/kg 5.1%
Product Solids 19 mg/kg 22.5% 279 mg/kg 4.1%
Filtrate 1.9 pqII NA NA NA
Percent Removal in Solids; 914% — 96i0% 20%
NA = Not Analyzed
4-10
-------
5.0 MASS BALANCE
Mass balance determinations are a necessary function in any measurement study that
involves the generation of one or more products from an input feed, especially when a treated
output material is produced from a contaminated input material; as was the case in the CF
Systems treatability study. As a secondary objective of the study, cumulative and individual
run mass balances have been determined by comparing the masses of the feed with the
masses of the various product fractions. The mass balances indicate how efficient the MDU
was in separating and enabling recovery of both the concentrated oily extract and product
solids.
This section presents mass balances on a total materials basis, a solids basis, and on
a contaminant concentration basis for PCBs. All field measurements used in the calculations
are presented in Attachment B.
5.1 Total Materials
The purpose of the total materials balance is to account for all material loaded into the
unit for each run and ensure that a significant amount of the material did not simply remain
in one or more of the process components. All materials loaded into the unit or exiting the
unit were examined in the total materials mass balance, except for the liquified propane output
since the propane was flared off at the end of the test and there was not an available method
to measure the amount flared. Feed soil and water were the input parameters that were
examined. Oil extract, slurry, and F-1 filter solids were the output parameters examined. The
mass of each of the input and output streams is shown diagrammaticaliy in Figure 5-1 and
summarized in Table 5-1.
As indicated in Table 5-1, the total input materials mass for Runs 1-6 was
approximately 792,600 grams and the total output materials mass was approximately
777,000 grams. Therefore, the unit achieved a total materials balance recovery of
approximately 98 percent for the entire treatability study.
5-1
-------
INPUT MATERIALS
274,400 grams
Soil Feed
518,200 grams
Water
792,600 grams
Total Input
Materials
I
OUTPUT MATERIALS
3,701 grams
Oil Extract
770,000 grams
Slurry
F-1 Filter Solids
3,378 grams
Materials
Recovery
98%
777,000 grams
Total Output
Materials
Figure 5-1. Total materials balance diagram.
5-2
-------
Table 5-1. Total Materials Balance
Input (grams!
Output (grams)
Run #
Feed soil1
Water
TOTAL
Oil Extract Slurry
F-1 Filter
TOTAL
Material
Recovered {%)
1
45,400
52,600
98,000
- 71,600
485
72,100
73.6
2
45,800
80,800
126,600
— 147,400
485
147,900
117.0
3
45,800
93.8002
139,600
- 116,200
485
116,700
83.6
4
45,800
88,500
134,300
— 134,800
640
135,400
101.0
5
45,800
99.5003
145,300
- 141,200
640
141,800
97.6
6
45,800
103.0004
148,800
3,700 158,800
640
163,100
110.0
TotaJ
792.600
777.000 1
ft'-,$8.0 '
1 Run# 2-6 includa the addition of 453.6 g of sand to fill void spaca in axtractor.
2 Includes 16,260 g of rinse water used to flush remaining soil from extractor.
3 Includes 10,396 g of rinse water used to flush remaining soil from extractor.
4 Includes the addition of rinsate water. Determined by taking the average of two previous rinsate additions,
5.2 Solids
The purpose of the solids mass balance is to ensure that all solids loaded into the
system exited during flushing or were filtered out and do not remain in the extractor or any
other CFS process component in significant volumes. The input parameter examined was the
feed soil minus the moisture and O&G contents. The output parameters examined were the
product solids (minus the moisture and O&G contents) and the F-1 filter solids.
Table 5-2 summarizes the total solids mass balance and percent recovery for each of
the runs in the treatability study. The 88.7 percent value is considered to be a reasonable
solids recovery because loss of all solids (e.g., through the pan filter or solids removed with
solvent regeneration) is very difficult to account for. The discrepancy in the data from Runs
1 and 2 was the result of caking of solids in the bottom of the extractor. The solids could not
be properly flushed out due to a broken water pump. These solids were flushed out later with
Run 2 solids after the pump was repaired.
5-3
-------
Table 5-2. Solids Balance
Input' (grams)
Output1 (grams)
Run #
Feed soil
Product Solids
F-1 Filter
Total
Percent Recovery (%)
1
44,100
18,100
485
18,800
42.4
2
45,000
72,400
485
72,900
162.0
3
44,900
36,600
485
37,100
82.4
4
44,500
21,700
641
22,400
50.3
5
44,900
44,100
641
44,700
99.8
6
44,600
41,400
641
42,000
94.2
TOTAL
- 268,000
liaiiiil
3,380
237.700
88.7
1 Dry weight basis (moisturo and O&G content subtracted out).
5.3 PCBs
The purpose of performing a PCS mass balance is to show that the unit effectively
separated the PCBs from the feed soil by accounting for them in the oil extract, fiiter cake,
and filtrate. The mass balance would determine whether the PCBs exited the unit in the
process output streams or remained in the system by coating the interior of various process
components. Table 5-3 summarizes the PCB mass balance as a total for the entire study.
The low recovery is believed to be primarily due to the inability to drain all of the oil from the
extract product tank and associated piping. If this is the case, then it is imperative that an
equipment decontamination step be implemented following solids treatment to acquire a better
PCB balance.
5-4
-------
Table 5-3. PCB Balance
Run #
Input (mg|
Feed soil
Oil Extract
Output img)
Product
Solids
Filtrate
Total
Percent
Recovery (%)
1
9,300
—
84.3
< 0.05
—
—
2
10,700
—
126
< 0.09
—
—
3
15,200
—
79
< 0.06
—
_
4
13,700
—
82
< 0.10
—
_
5
9,800
—
248
< 0.09
—
_
6
9,800
—
754
0.19
—
—
TOTAL.,.
? \>V, i<-
¦ ' 63k0
a Oily axtract was sampled at the and of all six runs. Tha vaiua is tha average of analyses of field duplicate samples.
5-5
-------
6.0 QUALITY OF THE DATA
6.1 Introduction
The data quality objectives established for this pilot-scale treatability study of the CFS
liquefied gas solvent extraction technology were based on project requirements and, thus,
designed to ensure that the data generated during the study would be of known and
acceptable quality to achieve the project's technical objectives. This section reports the
QA/QC results for each of the critical measurements in terms of the four data quality
indicators: accuracy, precision, method detection limits (MDLs), and completeness; other QC
information (e.g., QC blanks and holding times) is also discussed.
The objectives that were established in terms of the four data quality indicators for
critical parameters are summarized below in Table 6-1.
Table 6-1. OA Objectives for Accuracy, Precision, Method Detection Limit (MDL),
and Completeness for Critical Parameters
Parameters
Matrix
Method*
Accuracy6,
% Recovery
Precision,
RPDe
MDLd/Units
Completeness,
%
PCBs
Soil Feed
3540/8080
50-150
40
3 fJQlQ dry
80
Product Solids
3540/8080
50-150
40
0.1 UQtq dry
80
Filtrate
3520/8080
50-150
40
0.01 tjqiL
80
Moisture
Soil Feed
02216
NA-
25
0.1%
80
Product Solids
D2216
NA*
25
0.1%
80
a Method references are provided in Table 2-4,
b Determined by matrix spikes for PCB Aroclor 1254.
c Relative Percent Difference,
d Per specific compound, if applicable,
e Balance calibration verification checks must meet laboratory-specified acceptance criteria.
Samples in this section are identified by the number assigned at the time of sampling.
The coded sample identification system shown in Figure 6-1 was employed to relate each
sample to the run number, sample medium, chronological order of collection, and quality
assurance status. This self-explanatory code provided START field and laboratory personnel
with the most important aspects of each sample without having to access a cross reference.
6-1
-------
For this project, ail OA samples were collected during Run 4, except for the oily organic
extract which was collected at the end of the study, following Run 6.
Run No. (Run 4)
Sample Number
R4 - US - 04 - MS
Process Stream
OA Designation
US = Untreated Soil (Feed)
FC = Filter Cake (Product Solids)
D = Field Duplicate
MS = Laboratory Matrix Spike
FL = Filtrate (Process Water)
QE s Organic Extract (Product Oil)
TF = Tap Water Feed
TB = Trip Blank
PPEB = Pilot Plant Equipment Blank
(Toluene Rinsate)
6.1.1 Accuracy
Accuracy is the degree of agreement of a measured value with the true or expected
value. Matrix spikes are aliquots of sample spiked with a known concentration of target
analyte(s) used to document the accuracy of a method in a given sample matrix. For matrix
spikes, recovery is calculated as follows;
Figure 6-1. Key to sample identification.
%R = Cl ~C° x 100
C,
where: C,
Ca
Ct
measured concentration in spiked sample aliquot
measured concentration in unspiked sample aliquot
actual concentration of spike added
-------
A laboratory control sample (LCS) is a blank matrix spiked with representative target
analytes used to document laboratory performance. LCS results are useful in differentiating
laboratory problems from matrix problems. For LCSs, recovery is calculated as follows:
%R « x 100
C,
where: Cm = measured concentration of LCS
Ct = true concentration of LCS
For the PCB/pesticide analyses, surrogates were added to all samples and blanks to
monitor extraction efficiencies. Surrogates are compounds which are similar to target
analytes in chemical composition and behavior. Surrogate recoveries will be calculated as
shown above for LCSs.
6-1.2 Precision
Precision is the agreement among a set of replicate measurements without assumption
of knowledge of the true value. When the number of replicates is two (duplicates), as was
the case for this treatability study, precision is determined using the relative percent difference
(RPD):
RPD =
-------
6.1.3 Method Detection Limits
Method detection limits (MDLs) are based on project requirements and the sample
matrix to be analyzed. Obtaining the desired MDLs is a function of sample size, size of
extract, and the ability to clean up matrix interferences. The MDLs for the two critical
parameters (PCBs and moisture) were presented in Table 6-1. These objectives were met for
the solid samples. For the filtrate samples, however, a detection limit of 1 //g/L was reported.
Review of the original 0.01 //g/L objective indicates that this goal may have been
unnecessarily restrictive. The 1 //g/L limit should be sufficient in documenting the PCB
concentration in the filtrate samples.
6.1.4 Completeness
Completeness is defined as the ratio of the number of valid measurements to the total
number of measurements planned. For this treatability study, a completeness goal of 80
percent was established with respect to critical parameters and critical objectives. All PCB
and moisture results have been deemed useable. The only sample not collected as planned
was the propane solvent feed blank. Due to an inappropriate sampling technique, this sample
was invalidated. No PCBs were expected in the propane solvent.
6.2 PCBs
6.2.1 Extraction Methods
Solids samples (feed, treated filter cake) were Soxhlet-extracted with acetone:hexane
(1:1) for 18 hours as described in Method 3540 (SW-846). The extracts were subjected to
florisil cleanup (Method 3620).
Aqueous samples (filtrate, tap water feed) were continuously extracted with methylene
chloride for 18 hours as described in Method 3520 (SW-846). The extracts were subjected
to florisil cleanup.
Recovered solvent samples were concentrated prior to analysis. The pilot-plant
equipment blank (toluene rinsate) could not be concentrated and was analyzed as received.
Organic extract samples were diluted in hexane and analyzed.
6-4
-------
6.2.2 Analytical Method
Method 8080 calibration was performed using Aroclor 1254 at five concentration
levels; 0.1, 0.5, 1, 2, and 4 //g/mL. A quadratic curve was prepared. The calibration curve
was verified after every 10 samples using the midpoint standard. If the results for this
calibration check standard differed from the actual value by more than 15 percent, then
reanalyses of the affected samples were required. Continuing calibration requirements were
met for all reported results.
6.2.3 Quality Control
Matrix Soikes/Matrix Spike Duplicates
Table 6-2 presents the results for the PCB matrix spike and matrix spike duplicates
conducted on Run 4 product solids (FC) and filtrate (FL). These matrix spike analyses were
not performed with the original sample analyses due to laboratory oversight. The samples
were spiked and analyzed outside sample holding times only to provide recovery and precision
information. The results for the original analyses performed within holding times were
reported. Due to the high concentration of PCBs present in the feed (US) and product oil (OE),
spiking these matrices was not feasible.
The sample concentrations listed in Table 6-2 are those obtained from the reanalyses.
It should be noted that these values correlate well with the original results.
Table 6-2. PCB Aroclor 1254 MS/MSD Results
Spike
Sample Cone.
MS
MS
MSD
MSD
Sample
(mg/kg)
(mg/kg)
(mg/kg)
% R
(mg/kg)
% R
RPD
R4-FC-04MS
4.9 mg/kg
4,1 mg/kg
8.2 mg/kg
84
9.1 mg/kg
102
10
R4-FI-04MS
10.0 //g/l
< 1.0*/g/L
9.0 ijqli.
90
7.8 mn.
78
14
6-5
-------
Surrgq?tes
Table 6-3 presents the Method 8080 surrogate recoveries. The PCB surrogate used
was decachlorobiphenyl and the pesticide surrogate used was tetrachioro-m-xylene (TCMX).
These two surrogates were added to all sample and blanks prior to extraction to
monitor extraction efficiency during Method 8080 analyses. Acceptance limits were 60 to
150 percent.
Table 6-3. Method 8080 Surrogate Recoveries, %
Sample
Decachlorobiphenyl
T etrachloro-m-xylane
Feed
R1-US-01
R2-US-02
R3-US-03
R4-US-04
R4-US-G4D
R5-US-05
R6-US-06
Product Solids
R1-FC-01
R2-FC-02
R3-FC-03
R4-FC-04
R4-FC-04D
R5-FC-05
R5-FC-05 (Re-extract)
R6-FC-06
R6-FC-06 (Re-extract)
Filtrates, Field Blank
R1-FL-01
R2-FL-02
R3-FI-03
R4-FL-04
79
81
81
82
86
82
78
92
94
91
84
91
26'
114
21*
112
21*
24*
26*
29*
74
77
75
75
76
75
74
74
79
72
74
80
20*
102
3*
103
40*
40*
45*
43*
6-6
-------
Table 6-3. Method 8080 Surrogate Recoveries, % (Continued)
Sample
Decachlorobiphenyl
Tetrachloro-m-xylene
R4-FL-04 (Re-extract)
30*
46*
R4-FL-04D
30*
43*
R5-FL-G5
27*
43*
R5-FL-05 (Re-extract)
37*
63
R6-FL-06
26*
17*
R6-FL-06 (Re-extract)
33*
36*
RO-TF-01
83
65
Solvents/Oils
R0-PPEB-01
D
D
R1-RS-01
178*
_
R2-RS-02
178*
__
R3-RS-03
176*
_
R4-RS-04
167*
__
R4-RS-04D
170*
R5-RS-05
177*
__
R6-OE-G1
NS
NS
R6-OE-01D
NS
NS
NS = No surrogate added
0 = Surrogate diluted
• Outside control limits
- No recovery possible due to non-target interfarents
Re-extractions of filter cakes R5-FC-05 and R6-FC-06 were performed due to low
surrogate recoveries. These reanalyses were performed outside holding times; acceptable
surrogate recoveries were obtained. These re-extractions yielded results which were similar
to the original analyses (slightly lower). Since the reanalyses confirmed that poor recovery
was not a problem in the original analyses, the original results were reported. If a recovery
problem existed in the original analyses, the re-extracation results should have been
significantly higher. As discussed above, this was not the case.
All surrogate recoveries for the filtrate samples were low. Corresponding method
blanks also had low surrogate recoveries. Three of the filtrate samples were re-extracted and
6-7
-------
reanalyzed (R4-FL-04, R5-FL-05, R6-FL-06); surrogate recoveries were again low in the
samples but acceptable in the blank. PCB results identical to the original results were
obtained. It appears from the re-extractions that a matrix interference exists in the water
samples which affects the surrogate recoveries. However, based on the MS/MSD results
presented in Table 6-2, PCB recoveries are not affected.
For the recycled solvent samples (RS), high levels of non-PCB interferents prevented
proper quantitation of surrogate recoveries and increased MDLs. The samples were subjected
to analysis by GC/MS to confirm the absence of PCBs. No PCBs were identified.
Laboratory Control Sample (LCS)
An LCS and LCS duplicate were extracted and analyzed with the filtrate samples as
a measure of laboratory performance. The LCS consisted of the same PCB Aroclor 1254 used
for spiking. The acceptance criteria for recovery were the same as those used for the matrix
spikes. Recoveries of 96 and 91 percent were obtained, with an RPD of 5. Surrogate
recoveries for these LCSs were within acceptance limits.
No LCS was extracted with the original solid sample analyses due to laboratory
oversight. As a check on the extraction procedure for the solid samples, it was requested that
an LCS be performed with the filter cake reanalysis previously discussed. A recovery of 105
percent was obtained for this LCS.
Laboratory Duplicates
Precision objectives for critical PCS measurements were established (Table 6-1) as the
relative percent difference (RPD) between matrix spike and matrix spike duplicate (MS/MSD)
analyses. However, spiking was not feasible for the highly concentrated feed samples and
oil extracts. Laboratory duplicates (duplicates initiated in the laboratory) were requested on
these samples to evaluate precision for the critical PCB analysis.
Unfortunately, these duplicate analyses were not performed. However, since neither
of these matrices was involved in the assessment of primary project objectives, the lack of
precision information is not critical. Field duplicate samples of the oil extracts were collected
and analyzed. These results are discussed below.
6-8
-------
Field Duplicates
Table 6-4 presents the results of the PCB analyses of field duplicate samples for the
Run 4 solids and filtrate, and Run 6 product oil.
Table 6-4. PCB Aroclor Field Duplicate Results
Sample Sample Result Result 2 RPD
R4-US-04 350 mg/kg 260 mg/kg 30
R4-FC-04 4.0 mg/kg 3.9 mg/kg 2.5
R6-OE-01 11,200 mg/kg 11,300 mg/kg 0.9
R4-FL-04 < 1 yqlL < 1 //g/L NC
NC = Not calculated
Laboratory Method Blanks
All parameters analyzed in the laboratory require the analysis of a method blank, also
known as a reagent blank, with each batch of samples analyzed, or every 20 samples,
whichever is more frequent. A method blank consists of an aliquot of reagent water carried
through all preparation and analysis steps, and is designed to document that the analytical
equipment and reagents are free of contamination and interferences. If method blanks are
observed to be above MDl for a given parameter, the effect on the sample data should be
evaluated. If the amount found in the blank is less than 5 percent of the amount found in the
samples, the data can be flagged without reanalysis.
No PCBs were detected in any of the method blanks analyzed.
6.2.4 Sample Holding Times
Holding time requirements for all sample matrices and analyses were specified in the
QAPP. Table 6-5 presents a sample holding time chronology for the critical PCB
measurements showing the collection and analysis dates for each primary sample. The
analyses of all primary samples for PCBs were performed within specified holding times.
6-9
-------
Table 6-5. Sample Holding Times - PCB Analyses
Data Date Date Holding Time Met
Sample No.
Collected
Extracted
Analyzed
(Extraction/Anal
Feed"
R1-US-01
8/9/94
8/23/94
9/3/94
Yes/Yes
R2-US-02
8/10/94
8/23/94
9/3/94
Yes/Yes
R3-US-03
8/11/94
8/23/94
9/3/94
Yes/Yes
R4-US-04
8/15/94
8/23/94
9/3/94
Yes/Yes
R4-US-04D
8/1S/94
8/23/94
9/3/94
Yes/Yes
R5-US-05
8/16/94
8/23/94
9/4/94
Yes/Yes
R6-US-06
8/16/94
8/23/94
9/4/94
Yes/Yes
Product Solids*
R1-FC-01
8/10/94
8/23/94
9/3/94
Yes/Yes
R2-FC-02
8/11/94
8/23/94
9/3/94
Yes/Yes
R3-FC-03
8/11/94
8/23/94
9/3/94
Yes/Yes
R4-FC-04
8/16/94
8/23/94
9/3/94
Yes/Yes
R4-FC-04D
8/16/94
8/23/94
9/3/94
Yes/Yes
R5-FC-05
8/16/94
8/23/94
9/7/94
Yes/Yes
R6-FC-06
8/18/94
8/23/94
9/7/94
Yes/Yes
Filtrate and Field Blank"
R1-FL-01
8/10/94
8/12/94
9/7/94
Yes/Yes
R2-FL-02
8/11/94
8/17/94
8/30/94
Yes/Yes
R3-FL-03
8/12/94
8/17/94
8/30/94
Yes/Yes
R4-FL-04
8/16/94
8/22/94
9/2/94
Yes/Yes
R4-FL-04D
8/16/94
8/22/94
9/2/94
Yes/Yes
R4-FL-05
8/16/94
8/22/94
9/2/94
Yes/Yes
R6-FL-06
8/18/94
8/22/94
9/7/94
Yes/Yes
RO-TF-01
8/10/94
8/17/94
8/30/94
Yes/Yes
Non-Aqueous Media*
RO-PPEB-01
8/9/94
8/23/94
8/23/94
Yes/Yes
R1-RS-01
8/11/94
8/23/94
8/23/94
Yes/Yes
R2-RS-02
8/11/94
8/23/94
8/23/94
Yes/Yes
R3-RS-03
8/15/94
8/23/94
8/23/94
Yes/Yes
R4-RS-04
8/15/94
8/23/94
8/23/94
Yes/Yes
6-10
-------
Table 6-5. Sample Holding Times - PCB Analyses (Continued)
Sample No.
Date
Collected
Date
Extracted
Date
Analyzed
Holding Time Met
(Extraction/Analyses)
Non-Aqueous Media (Con't)
R4-RS-Q4D
8/15/94
8/23/94
8/23/94
Yes/Yes
R5-RS-05
8/16/94
8/23/94
8/23/94
Yes/Yes
R6-OE-01 MS
8/18/94
8/25/94
8/26/94
Yes/Yes
R6-OE-01D
8/18/94
8/25/94
8/26/94
Yes/Yes
R6-0E-01 MS (Reanalysis)
8/18/94
10/18/94
10/19/94
No/No0
R6-OE-01D (Reanalysis)
8/18/94
10/18/94
10/19/94
No/Noc
a 14 days to extraction; 40 days to analysis,
b 7 days to extraction; 40 days to analysis,
c Those samples were analyzed by a different laboratory at the indicated later date, due to poor precision results acquired
initially. The results are presented in Table 8-4.
6.2.5 Field QC Sgmplg?
Field QC samples for this treatability study included a pilot plant equipment blank, a
sample of the tap water that was used as process water and ultimately became filtrate, and
field duplicates conducted on the feed and process products. Field duplicate data have
already been discussed. Table 6-6 presents the results of the analyses conducted on the pilot-
plant equipment blank and tap water. A description of how these samples were collected is
described in Section 3 (Field Activities).
Table 6-6. Results of PCB Analyses of Field QC Samples
Sample
Type
PCBs
RO-PPEB-01
Toluene rinsate of pilot plant
< 1.9 mg/L
RO-TF-01
Tap water
< 1 //g/L
6.3 Moisture
6.3.1 Mgth
-------
6.3.2 Quality Control
A laboratory duplicate was performed for one solid sample to assess precision. Results
obtained were 2.44 percent and 2.34 percent, with an RPD of 4.2, indicating good precision
during this analysis.
6.4 Noncritica! Parameters
6.4.1 Oil and Grease
Solid samples were Soxhlet-extracted with freon as described in Method 9071. The
extract was dried and weighed.
An LCS was extracted and analyzed with the solid samples; a recovery of 93 percent
was obtained.
Two MS/MSD analyses were performed; these results are summarized in Table 6-7.
Table 6-7. OH and Grease MS/MSD Results
Sample
Sample Cone
(mg/kg)
Spike
Added
MS
(mg/kg!
%
Recovery
MSO
(mg/kg)
%
Recovery
RPD
R4-US-04
5460
10,300
17,000
112
15,100
94
12
R4-FC-04
133
11,800
12,000
101
11,500
96
4
6.4.2 Volatile Solids
Solid samples were dried and then ignited at 550°C as described in Standard Methods
2540G. Originally, the ignition time was 30 minutes rather than 60 minutes as specified in
Method 2540G. Two laboratory duplicate analyses were performed; these results are
presented in Table 6-8. Reanalyses were requested to verify that a sufficient ignition time
was used. Samples were ignited for 60 minutes, cooled, and weighed. These samples were
then re-ignited for 30 minutes; no significant difference was observed.
6-12
-------
Table 6-8. Volatile Solids Duplicates
Sample
Sample, %
Duplicate, %
RPD
R4-US-04
3.60
4.19
15
R4-FC-04
4.23
4.66
9.7
6.4.3 Volatile Oraanics (Toluene)
Method 8260 was used for volatile organic analysis. For the feed and filtrates, sample
aliquots were directly purged. For the filter cake and organic extract samples, methanolic
dilutions were performed.
Table 6-9 summarizes the surrogate recoveries obtained.
Table 6-9. Volatile Surrogate Recoveries. %
Sample
1,2-Dichloro-
ethane-d4
Toluene-dg
4-Bromo-
fluorobenzene
R4-US-04
104
111
90
R4-FC-04
115
111
83
R4-FL-04
101
108
103
R6-OE-01
NS
NS
NS
TB-01
103
106
105
R4-FC-04 (dilution)
115
125*
120
* Outside QC limits
NS = no surrogates added
One MS/MSD pair was analyzed for volatile organics. These results are presented in
Table 6-10. The LCS results associated with this MS/MSD pair are summarized in Table 6-
11.
6-13
-------
Table 6-10. Volatile Organic MS/MSD Results, R4-US-04
Compound
Spike Added
(*/Q/kgj
Sample Cone
&g/kg|
MS
%
Recovery
MSD
%
Recovery
RPD
1,1-
Dichloroethene
50
NO
48.7
97
43.1
86
12
Benzene
50
ND
45.6
91
44.2
88
3
Trichloroethene
50
ND
37,7
75
36.5
73
3
Toluene
50
1.7
45.8
88
43.3
83
6
Chlorobenzene
50
NO
48.9
98
46.7
93
5
Table 6-11. Volatile Organic LCS Results
Compound % Recovery
1,1 -Dichloroethane
85
Benzene
91
Trichloroethene
77
Toluene
91
Chlorobenzene
93
A trip blank was shipped with the filtrate sample (R4-FL-04). Minimal concentrations
of acetone (29 //g/L) and chloroform (27 //g/U were observed. No chloroform was detected
in the filtrate sample. Acetone is a common laboratory contaminant; its presence in the trip
blank cannot be attributed to sample shipping procedures. At the level detected, there is
minimal impact to project results.
6.4.4 Propgng
Aqueous filtrate samples were directly injected into a gas chromatograph equipped with
a flame ionization detector (GC/F1D). Solid filter cake samples were leached with water; the
leachate was then injected into the GC. The oil extract was diluted with hexadecane and then
injected into the GC. All analyses were performed in triplicate.
A five-point calibration curve was prepared. A linear curve was obtained. All
continuing calibration checks were < 20 percent different than the initial results.
6-14
-------
R4-FL-04 was spiked in triplicate. Recoveries of 81, 86, and 81 percent were
obtained.
6.4.5 Pesticides
Pesticide analyses were performed using Method 8080. The extracts analyzed were
the same as those generated for PCB measurements. Five-point calibrations using specific
pesticide compounds were prepared; quadratic curves were generated.
An LCS containing six representative pesticides was extracted and analyzed with both
solid samples and aqueous samples. All recoveries were within laboratory control limits.
MS/MSD pairs were analyzed for one feed sample, one filter cake sample, and one
filtrate. These results are presented in Tables 6-12 through 6-14.
As can be seen in Tables 6-12 and 6-13, the presence of Arocior 1254 at significant
concentrations interfered with recoveries of 4,4'-DDT, dieldrin, and endrin in the feed sample
and 4,4'-DDT and dieldrin in the filter cake sample.
Table 6-12. Pesticide MS/MSD Results, R4-US-04
Compound
Spike Added
(Ml/kg)
Sample Cone
U/g/kg)
MS
%
Recovery
MSD
%
Recovery
RPD
G-BHC
116
NO
103.0
102
113
112
9
Heptaclor
116
ND
70.6
70
77.2
76
8
Aldrin
232
ND
55.1
54
63.9
63
15
4,4'-DDT
232
NO
2,780*
1,370*
2,740*
1,360*
*
Dieldrin
232
ND
•
•
2,130*
1,050*
•
Endrin
232
ND
1,052*
516*
1,370*
680*
~
NO = Not detected
* High PCB concentrations interfered with pesticide results.
6-15
-------
Table 6-13. Pesticide MS/MSD Results, R4-FC-04
Compound
Spike Added
iiiQlkg)
Sample Cone
(^g/kg)
MS
%
Recovery
MSD
%
Recovery
RPD
G-BHC
116
ND
109
102
98.3
85
18
Heptaclor
116
ND
126
108
114
98
10
Aldrin
116
ND
81.4
70
74.1
64
9
4,4'-DDT
232
ND
468*
202*
492
212*
~
Dieldrin
232
ND
517*
223*
526
227*
•
Endrin
232
ND
145
62
114
49
23
ND - Not detected
* High PCS concentrations interfered with pesticide results.
Table 6-14. Pesticide MS/MSD Results.
R4-FL-04
Compound
Spike Added
U/g/kg|
Sample Cone
U/Q/kg)
MS
%
Recovery
MSD
%
Recovery
RPD
G-BHC
1.0
ND
0.94
94
0.96
96
2
Heptaclor
1.0
ND
0.49
49
0.49
49
0
Aldrin
1.0
ND
0.35
35
0.38
38
8
4,4*-DDT
2.0
ND
0.71
36
0.62
31
15
Dieldrin
2.0
ND
1.47
74
1.45
72
3
Endrin
2.0
ND
1.61
80
1.58
79
1
6.4.6 Metals
Metals were quantitated by Method 6010 using inductive coupled plasma (ICP)
spectrophotometry, except for arsenic, selenium, and mercury. These were analyzed using
graphite furnace atomic absorption (Methods 7060 and 7740) and cold vapor atomic
absorption (Method 7470}.
A matrix spike and duplicate was performed for one solid sample (R4-FC-04) and one
aqueous sample (R4-FL-04). Results are presented in Table 6-15.
6-16
-------
Table 6-15. Metals QC Results
R4-FC-04 R4-FL-04
Metal
MS, % R
Duplicate, RPD
MS, % R
Duplicate, RPD
As
250
8
106
0.2
Ba
36
44
105
4
Cd
70
1
104
b
Cr
69
6
104
2
Cu
a
58
104
6
Fe
a
34
108
5
Pb
a
2
a
10
Mri
34
17
99
2
Hg
73
31
70
4
Ni
88
0.2
110
b
Se
93
b
105
b
Zn
a
15
96
2
a Inappropriate spiking laval used,
b RPD not calculated; sample or duplicate concentration not detected.
6.5 Modifications to and Deviations from the QAPP
Modifications to and deviations from the QAPP occurred during the course of the
project involving both field and laboratory activities. Laboratory/method deviations have been
discussed in this section. Other deviations and their effect on data quality are discussed in
this subsection.
Field Mo difica tions/De via tions
* An additional run was performed during the treatability study. Run 6 was added
at the suggestion of the developer and with concurrence of the EPA Project
Manager to provide additional useful information regarding the CFS process. This
run consisted of using just two extraction cycles at 20 minutes each for treating
the feed material. The feed, product solids, and filtrate were analyzed for PCBs,
and moisture content was determined for the solids. The results of Run 6 are
discussed separately in Section 4 because they are not a part of the original
experimental design.
6-17
-------
• There was a change in the recycled propane sampling procedure for determining
residual PCS content in the solvent. It was decided in the field that bubbling vented
propane from a pressurized sample canister into a solvent exchange media (hexane)
was futile since any semivolatile compound could not be vaporized from the
canister once a pressure drop existed. It was decided in the field to vent the
propane canister, rinse its interior walls thoroughly with hexane, and collect the
rinsate in 40-mL vials. Because this modification was not proposed until after the
collection of the propane solvent feed sample, no valid sample was obtained.
These modifications were documented and submitted to EPA project and OA
management.
6-18
-------
7.0 CONCLUSIONS
This section summarizes conclusions reached regarding the performance of the CF
Systems® pilot plant for treating STD soils during a treatability study performed between
August 8 and August 18, 1994 in Golden, Colorado. Because the primary objective of the
treatability test was, in essence, to determine which test conditions would produce the
"cleanest" solids product, it is appropriate to compare directly the results of the analyses that
were conducted on the product solids generated for all six test runs. Table 7-1 presents an
overall summary of the analytical results for PCBs, moisture, O&G, and volatile solids for all
six test runs. Table 7-2 summarizes the conclusions discussed in the following subsections.
Table 7-1. Summary of Results of Analyses Conducted on Product Solids for All Test Runs
Run No. and Extraction Cycle Condition
Analyte (units!
Run 1
(3 cycles)
Run 2
{4 cycles!
Run 3
(5 cyclesl
Run 4
{3 cycles)
Run 5
(3 cycles!
Run 6
{2 cycles)
PCBs (mg/kg)
4.9
1.8
2.2
3.9'
5.8
19
Moisture (%)
11,4
15.4
15.9
15.3
15.4
22.5
Oil & Grease (mg/kg)
112
73
< 20
133
93
279
Volatile Solids (%)
3.0
1.4
1.0
4.4
4.3
4.1
a Average concentration of analyses of field duplicate samples rounded to two significant digits.
7.1 Organics Removal
The analytical test data indicates that the primary goal of producing solids having
< 1.0 mg/kg PCB concentration was not attained. Of the five main test runs, the closest PCB
concentration to the RAS of 1.0 mg/kg was 1.8 mg/kg in Run 2 product solids, which was
approximately a 99.3 percent removal efficiency as calculated from the Run 2 feed
concentration of 240 mg/kg. The average PCB concentration in product solids samples
collected from the five main test runs was 3.7 mg/kg, which translates into an average
removal efficiency of approximately 98.6 percent from an average feed concentration of 260
mg/kg.
7-1
-------
Table 7-2. Summary of Conclusions
Contaminant Removal
Contaminant
Results
Conclusion
PCBs
Average removal of approximately
98.6% for the five main runs, which
included three runs using three cycles,
and one run each using four and five
cycles. Lowest value for product
solids = 1.8 mg/kg using 4 extraction
cycles
To attain product solids having PCB
concentrations s1.0 mg/kg, per the
test method used, more aggressive
operating conditions may be
necessary {i.e., use of a solvent that
is more polar than propane, additional
extraction cycle, etc.). It is also
possible that the target goal would
have been met, with the conditions
used, if a lower feed concentration
had existed.
O&G
Average removal > 98%. Lowest
value for product solids < 20 mg/kg
Mass Balance
No goal; CFS process effective in
removing on average > 98% of O&G
in the STD soil
Category
Results {% recovery)
Conclusion
Total Mass Balance
Solids Balance
PCB Balance
98.0
88.7
63.0
98% of all materials accounted for
{which is considered a very good
mass balance closure). Variations in
solids balance is considered minor.
The PCB balance is low and is
believed to be due to an inability to
completely drain all of the extract
from the pilot plant, which is not
uncommon for pilot-scale operations.
Process Operation
Operation
Results
Conclusion
Extraction Efficiency
Based on the results of ail six runs.
Runs 2 and 3 (four- and five-20-
minute extraction cycles respectively)
produced solids having the lowest
PCB concentrations. These
concentrations (1.8 and 2.2 mg/kg)
are essentially equal based upon an
analytical perspective.
Based on the results of all six runs,
the number of extraction cycles
required for attaining the lowest
concentrations of PCBs in product
solids was > 3 but <. 5. However,
five extraction cycles appears best for
overall organics recovery based on
limited O&G data.
One of the four Run 2 extractions
involved a different type of mixing,
yet PCBs apparently were extracted.
Therefore, diffusion of PCBs from the
STD sandy soil into solvent likely
occurs within the 20-minute cycle
regardless of the mixing method.
7-2
-------
Table 7-2. Summary of Conclusions (Continued)
Contaminant Removal
Contaminant
Results
Conclusion
Water Separation
All filtrate from the five main test runs
below detection limits for PCBs. PCBs
detected at very low levels in Run 6
filtrate.
PCBs were not solubilized in the
filtrate at or above 1 //g/L when >_ 3
extraction cycles were used.
Oil Recovery
12,500 mg/L of residual propane in
product oil
CFS solvent recovery process
effective in recovery of most all
propane for recycle.
Waste Reduction
Mass of oily extract was 1.35% of
total feed soil mass.
Volume of oily extract was 2.1% of
total feed soil volume.
CFS process effective in consolidating
organic contaminants present in the
STD soil into a much reduced mass
and volume.
There is not enough data to indicate whether the additional fifth extraction cycle
conducted during Run 3 would benefit PCB removal. The two concentration values for Run
2 (1.8 mg/kg) and Run 3 (2.2 mg/kg) are essentially equal since they are within the range of
field sampling and analytical error. However, the other organic type analyses conducted (O&G
and volatile solids) can be used to supplement the interpretation of results, with respect to
organics removal in general.
As Table 7-1 indicates, when PCB, O&G, and volatile solids are evaluated together, the
five cycles used for Run 3 appear to have performed best for overall organics removal; the
two cycles used for Run 6 performed the worst. The performance of the runs relative to one
another may be best presented in an illustrated fashion. Figures 7-1 and 7-2 show the
removal of PCBs and O&G respectively, for each test run as the descent in contaminant
concentration from starting feed to product solids as sloped lines. Both figures show the
disparity in performance between test runs for the respective parameters, which may not be
as apparent when simply looking at percent removal values. Figure 7-1 clearly shovys that
Runs 2, 3, and 4 came closer to the test objective assuming a feed concentration equal to the
average of ail runs (260 mg/kg). For O&G removal. Figure 7-2 indicates that Run 3 produced
the "cleanest" solids, while Runs 1, 2, 4, and 5 had approximately equal success.
7-3
-------
400
300 "
200
100-
Run No.
m. -mm TeSt 0bjGCtiV8
3 extraction-cycle run
- 4 extraction-cycle run
- 5 extraction-cycle run
•• 2 extraction-cycle run
1 2 3 4 5
10 15
Product Solids Concentration (PCBs in mg/kg)
20
Figure 7-1. PC8 removal trend.
\
V V %
I ~ V",
s\. \
\M2 \
\ X
\©\\
(J) Run No.
3 extraction-cycle run
—4 extraction-cycle run
— 5 extraction-cycle run
..... 2 extraction-cycle run
30 50
100 150 200 250
Product Solids Concentration (O&G in mg/kg)
300
Figure 7-2. Oil and grease removal trend.
7-4
-------
Both figures indicate that implementing two extraction cycles did not produce relatively good
results.
7.2 Mass Balance
The CFS process recovered 98 percent of all material loaded into the unit for each of
the six runs. This is considered to be a very good total materials mass balance closure. The
CFS process recovered 88.7 percent of total solids loaded into the unit. Although lower than
the total materials recovery, this value still indicates the unit's ability to recover a significant
amount of the solids loaded into the system. The observed presence of some solids in the
filtrate water may account for a small amount of unaccountable solids. The PCB balance was
low (63 percent). The low PCB recovery may be due to oil that was retained in the organic
recovery tank and associated piping, which consequently could not be drained out.
7.3 Volume Reduction of Hazardous Waste
The CFS solvent extraction process is not capable of destroying PCBs and other
contaminants present in the STD soil but is a means of separating those contaminants from
the soil, thereby reducing the volume of hazardous waste that must be treated and the
cleanup costs involved. The cumulative mass of the wet contaminated feed soil for all six
runs of the treatability study was approximately 274,000 grams. The mass of the oily extract
sampled at the completion of Run 6 was approximately 3,700 grams. Therefore, the process
reduced the overall mass of the contaminated material to 1.35 percent of its original waste
mass. The volume of the feed soil (SG = 1.34 g/mL) and oil extract (SG = 0.87 g/mL) were
approximately 204 and 4.3 liters, respectively. Therefore, the process reduced the overall
volume of the contaminated material to 2.1 percent of its original waste volume. The highly
concentrated oily extract from the CFS process is destroyed by either incineration or chemical
dechlorination.
7-5
-------
8.0 REFERENCES
1. American Society for Testing and Materials. Annual Book of ASTM Standards.
2. CF Systems. Personal Communication from Joseph Tillman of SAIC to John
Markiewicz of CF Systems. September 1994.
3. CF Systems. Pilot-Plant Process Description {No Date).
4. CH2M-Hill. Remedial Design Field Investigation. Springfield Township Dump, Oakland
County, Michigan, January 1992.
5. CH2M-Hill Predesign Report. Springfield Township Dump, Oakland County, Michigan
(No Date).
6. Commercial Testing and Engineering Co. Results of Particle Size Analysis: ASTM
D422. Correspondence of August 23, 1994.
7. Lockheed Analytical Services. SAIC PCB Project - Analytical Reports, December 1994.
8. SAIC. Quality Assurance Project Plan (Category III) for Pilot-Scale Testing of the CF
Systems* Liquefied Propane Extraction Process for Removal of PCBs from Soil, from
the Springfield Township Dump Superfund Site Near Davisburg, Michigan, February
1994.
9. SAIC Methods Laboratory. Analysis of Propane in Process Water, Filter Cake, and
Organic Extracts Produced During the Pilot-Scale Testing of the CF Systems Process
for Waste Extraction, October 17, 1994.
10. SAIC Methods Laboratory. Determination of PCB in Oil Extracts from the Pilot-Scale
Testing of the CF Systems Process, October 20, 1994.
11. U.S. Environmental Protection Agency. Standard Methods for the Examination of Water
and Wastewater, 17th Edition, 1989.
12. U.S. Environmental Protection Agency. Methods for Chemical Analysis of Water and
Waste. EPA/4-79-020, Revised March 1986,
13. U.S. Environmental Protection Agency. Test Methods for Evaluating Solid Waste (SW-
846), 3rd Edition. November 1986.
8-1
-------
attachment a
PHYSICAL AND CHEMICAL PROPERTIES OF PROPANE
Chemical Formula:
Synonyms:
CAS Registry No.:
Chemical FamBy:
TWA Exposure Limits:
IDLH:
Physical Description:
CaHB
Liquified Petroleum Gas (LPG)
74-98-6
Hydrocarbon
1000 ppm
20,000 ppm
Colorless, odorless gas. Product contains an odorant (unpleasant odor).
Chemical and Physical Properties
Molecular Weight: 44.1
SoiubiRy: 0.01%
ionization Potential: 11.07 eV
Vapor Density: 1.52
Percent, Volatile by Volume (%): 100
Upper Explosive Limit (UEL): 9.5%
Boiling Point: -44°F
Rash Point: -156°F
Specific Gravity: 0.51
Vapor Pressure at 100°F: 205 pslg
Freezing Point* 306°F
Lower Explosive Limit (LEL): 2.1%
Inhalation: dizziness, disorientation
Congestion: excitation, frostbite
First Aid
Eyes:
Skin:
Breath:
Immediately flush eyes with large amounts of water
immediately flush with water
Obtain respiratory support
Recommendations for Respirator Selection
10,000 ppm:
20,000 ppm:
Escape:
References
Supplied air respirator, self-contained breathing apparatus
Supplied air respirator (continuous flow mode), self-contained breathing apparatus with full
facepiece, supplied air respirator with full facspiece
Any appropriate escape-type, self-contained breathing apparatus
• Liquified Petroleum Gas or Propane Material Safety Data Sheet Health and Safety Plan, Pilot-Scale
Testing of CF Systems* Uquhied Propane Extraction Process, for Removal of PCBs from Soil, from the
Springfield Township Dump Superfund Site near Davisburg, Michigan, February 1994.
• National Institute for Occupational Safety and Health Pocket Guide to Chemical Hazards, June 1990,
pp. 186,187.
A-1
-------
ATTACHMENT B
FIELD MEASUREMENTS USED FOR MASS BALANCE CALCULATIONS
RUN 1
Feed soB:
Propane added:
Slurry weight:
Filter Cake:
Filtrate Water
Propane bomb full:
Propane bomb empty:
Filter mesh tare:
FBter mesh with solids:
Water added:
100 lbs (45,400g)
150.2 + 150 + 150 = 450.2 lbs (204,200g)
158 lbs (71,600g)
621 .S3g* + 17,463.6g + 1,276.5gb « 19,400g
48,700.1 g + 4009.56g = 52,800g
12,140g
11,990g
1,241g
2,31 Og
116 lbs (52,600g)
RUN 2
Feed sol:
Propane added:
Slurry weight:
Flter Cake:
Fltrate Water:
Propane bomb full:
Propane bomb empty:
Flter mesh tare:
Filter mesh with solids:
Water added:
RUN 3
101 lbs (45,800g)
150 lbs X 4 - 600 lbs (272,150g)
325 lbs (147,400g)
80,468.64g + 700.220* + 1,276.50"
81.199.4g + 4,136.45g* - 85,400g
12,075g
10,251 g (after rinsate: 10,252g)
l,241g
4,400g
178 lbs (80,750g)
82,400g
Feed sol:
Propane added:
Slurry weight
Flter cake:
Filtrate Water
Propane bomb Ml:
Propane bomb empty:
Filter mesh tare:
Flter mesh with solids:
Water added:
101 lbs (45,800g)
156 + 134.4 + 148.3 + 153 + 151 - 742.7 lbs (336,900g)
256 lbs (116,200g)
41,025.82 + 902° + 901.5* - 42,800g
4,104* + 56,405.26 = 60,600g
12,097g
10,253g
1,282q
2,019g
170.8 IbS (77.474.88g) + 16,266gd = 93,750g
RUN 4
Feed soil:
Propane added:
Slurry:
Filter Cake:
Filtrate Water
Propane bomb full:
Propane bomb empty:
Filter mesh tare:
Filter mesh with solids:
Water added:
101 lbs (45,800g)
157 + 160 + 150 = 467 lbs (211,850g)
297 lbs (134,800g)
20,250.13g + 3,951.7g* - 24,200g
82,173.04g + 14.053.25g* = 96,200g
12,114g (Duplicate: I1,760g)
10,288g (Duplicate: 10,290g)
1,250g
2,710g
195 lbs (88,450g)
B-1
-------
RUNS
Feed sol:
Propane added:
Slurry.
Filter Cake:
Filtrate Water
Propane bomb full:
Propane bomb empty.
Filter mesh tare:
Filter mesh with solids:
Water added:
101 lbs (45,800g)
430.59 lbs (195,350g)
287.5 + 24d - 311.5 lbs (141,200g)
48,922.96g + 703.9g* + 902gc = 50,600g
86,403.81 g + 4,039g - 90,400g
11,903g (after venting with hexane: I0,345g)
10,291 g
1,280g
2,834g
195 lbs (88,452g) + 10,896gd = 99,450g
RUNS
Feed soil:
Propane added:
Slurry weight:
Filter cake:
Filtrate Water:
Propane bomb full:
Propane bomb empty:
Filter mesh tare:
Filter mesh with solids:
Water added:
Oi Extract:
101 lbs (45,800g)
300 lbs (136,100g)
350 lbs (158,800g)
50,599.6g + 573.2g = 51,200g
91,800.1 g + 4,151.8g « 96,000g
NA
NA
1,205g
3,477g
197 lbs + 300 lbs* - 227 lbs (102,950g)
3,311.7 + 388.9" - 3,701 g
*Note: The Hazen filter cake moisture sample was taken before SAIC weighed the sol in Runs 1-3 and 5
and therefore was added In above.
The CF Systems filter cake sample was taken before SAIC weighed the soil in Runs 1 and 2 and
after In Runs 3 through 6.
a SAIC sample
b CF Systems moisture and analytical sample
c CF Systems moisture sample
d Rlnsate water
e Estimation of rlnsate water based upon average of previous additions
8-2
------- |