£EPA
         United States
         Environmental Protection
         Agency
             Office of Research and
             Development
             Washington, DC 20460
EPA/60Q/R-01/092
September 2001
Innovative Technology
Verification Report
         Field Measurement
         Technologies for Total
         Petroleum Hydrocarbons in Soil

         Dexsil® Corporation
         PetroFLAG™ System

-------
                                  EPA/600/R-01/092
                                  September 2001
Innovative Technology
   Verification Report
      Dexsil® Corporation
     PetroFLAG™ System
               Prepared by

             Tetra Tech EM Inc.
         200 East Randolph Drive, Suite 4700
            Chicago, Illinois 60601

            Contract No. 68-C5-0037
             Dr. Stephen Billets
        Characterization and Monitoring Branch
          Environmental Sciences Division
          Las Vegas, Nevada 89193-3478
        National Exposure Research Laboratory
         Office of Research and Development
         U.S. Environmental Protection Agency
              EH/

-------
                                       Notice
This document was prepared for the U.S. E nvironmental Protection Agency (EPA) Superfund
Innovative Technology Evaluation Program under C ontract No. 68-C5-0037. The document has
been subjected to the EPA's peer and administrative reviews and has been approved for publication.
Mention of corporation names, trade names, or commercial products does not constitute endorsement
or recommendation for use.

-------
                      UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                                     Office of Research and Development
              ENVIRONMENTAL TECHNOLOGY VERIFICATION PROGRAM
                                VERIFICATION STATEMENT
 TECHNOLOGY TYPE:    FIELD MEASUREMENT DEVICE

 APPLICATION:           MEASUREMENT OF TOTAL PETROLEUM HYDROCARBONS

 TECHNOLOGY NAME:   PetroFLAG™ SYSTEM

 COMPANY:               DEXSIL® CORPORATION
 ADDRESS:                ONE HAMDEN PARK DRIVE
                            HAMDEN, CT 06517

 WEB SITE:                http://www.dexsil.com

 TELEPHONE:             (203) 288-3509
VERIFICATION PROGRAM DESCRIPTION

The U.S. Environmental Protection Agency (EPA) created the Superfund Innovative Technology Evaluation (SITE) and
Environmental Technology Verification (ETV) Programs to  facilitate deployment of innovative technologies through
performance verification and information dissemination. The goa  1 of these programs is to further environmental protection
by substantially accelerating the acceptance and use of improved and cost-effective technologies. These programs assist and
inform those involved in design, distribution, permitting,  and purchase of environmental technologies.  This document
summarizes results of a demonstration of the PetroFLAG™ System developed by Dexsil * Corporation (Dexsil).

PROGRAM OPERATION

Under the SITE and  ETV Programs, with the full participa tion of the technology developers, the EPA evaluates and
documents the performance of innovative technologies by developi ng demonstration plans, conducting field tests, collecting
arid analyzing demonstration data, and preparing reports. The technologies are evaluated under rigorous quality assurance
(Q A) protocols to produce well-documented data ofknown quality . The EPA National Exposure Research Laboratory, which
demonstrates field sampling, monitoring, and measurement tec hnolbgies, selected Tetra Tech EM Inc. as the verification
organization to assist in field testing seven field measuremen t devices for total petroleum hydrocarbons (TPH) in soil. This
demonstration was funded by the SITE Program.

DEMONSTRATION DESCRIPTION

In June 2000, the EPA conducted a field demonstration of the PetroFLAG™ System and six other field measurement devices
for TPH in soil. This verification statement focuses on the Pe troFLAG™ System; a similar statement has been prepared for
each of the other six devices. The performance and cost of th e PetroFLAG™ System were compared to those of an off-site
laboratory reference method, "Test Methods for Evaluating Solid Waste" (SW-846) Method 8015B (modified). To verify
a wide range of performance attributes, the demonstration ha d both primary and secondary objectives. The primary objectives
included (1) determining the method detection limit, (2)  evaluating the accuracy and precision of TPH measurement,
(3) evaluating the effect of interferents, and (4) evaluating the effect of moisture content on TPH measurement for each
device. Additional primary objectives were to measure samp le throughput and estimate TPH measurement costs. Secondary
objectives included (1) documenting the skills and training re quired to properly operate the device, (2) documenting the
portability of the device, (3) evaluating the device's durab ility, and (4) documenting the availability of the device and
associated spare parts.

The PetroFLAG™ System was demonstrated by using it to analyze 66 soil environmental samples, 79 soil performance
evaluation (PE) samples, and 36 liquid PE samples. In add ition to these 181 samples, 10 extract duplicates prepared using
the environmental samples were analyzed. The environmenta 1 samples were collected in four areas contaminated with
gasoline, diesel, or other petroleum products, and the PE sa mples were obtained from a comme rcial provider. Dexsil chose
not to analyze soil samples collected in a fifth area because De xsil believed that the natural organic material in the area wou Id
adversely impact the PetroFLAG™ System's ability to accurately measure TPH. In addition, Dexsil chose not to analyze
                          The accompanying notice is an integral part of this verification statement.                 September 2001
                                                  iii

-------
low- and medium-concentration-range weathered gasoline so il PE samples because according to Dexsil, the PetroFLAG™
System was not sensitive to weathered gasoline concentrations of less than 1,000 milligrams per kilogram.

Collectively, the environmental and PE samples provided the di fferent matrix types and the different levels and types of
petroleum hydrocarbon contamination needed to perform a comprehensive evaluation of the PetroFLAG™ System. A
complete description of the demonstration and a summary of its results are  available in the  "Innovative Technology
Verification  Report:  Field Measurement  Devices fo r Total  Petroleum  Hydrocarbons  in Soil—Dexsil * Corporation
PetroFLAG™ System" (EPA/600/R-01/092).

TECHNOLOGY DESCRIPTION

The PetroFLAG™ System manufactured by Dexsil is based on  emulsion turbidimetry, which involves measurement of the
light scattered by an emulsion. With the PetroFLAG™ Sy stem, a proprietary, nonpolar, orga nic solvent mixture composed
of alcohols, primarily methanol, is used to extract petr oleum hydrocarbons from soil samples.  A proprietary developer
solution that is polar in nature and that acts as an emulsifier  is added to a sample extract in order to precipitate the aromat ic
and aliphatic hydrocarbons and form uniformly sized micelles.  Light at a wavelength of 585 nanometers is passed through
the emulsion, and the amount of light scattered by the emulsion  at a 90-degree angle is measured using a turbidimeter. The
TPH concentration in the emulsion is then determined by comparing the turbidity reading for the emulsion to that for a
reference standard or to a standard calibration curve. Accordi ng to Dexsil, the TPH concentration thus measured is a function
of the mean molecular weight of th e hydrocarbons present in the sample.

During the demonstration, extraction of petroleum hydrocarbons in a given soil sample was typically completed by adding
10 milliliters (mL) of proprietary methanol mixture extraction solvent to 10 grams of the sample. To form an emulsion, 2 mL
of sample  extract was then decanted into a vial containing 4  mL of developer solution.  The emulsion was analyzed using
the PetroFLAG™ Analyzer (turbidimeter)  to obtain a direct measurement of the TPH concentration in the soil sample.

VERIFICATION OF PERFORMANCE

To ensure data usability, data quality indicators for accuracy,  precision, representativeness, completeness, and comparability
were assessed for the reference method based on project-s pecific QA objectives. Although the reference method results
generally exhibited a negative bias, based on the results for the data quality indicators, the reference method results were
considered to be  of  adequate quality.  The  bias was considered  to be  significant primarily for  low- and medium-
concentration-range soil samples containing diesel, which made up only 13 percent of the total number of samples analyzed
during the demonstration. The reference method recoveries obs erved during the demonstration were typical of the recoveries
obtained by most organic analytical methods for environmental  samples. In general, the user should exercise caution when
evaluating the accuracy of a field measurement device by comp aring it to reference methods because the reference methods
themselves may have limitations. Key demonstration fi ndings are summarized below for the primary objectives.

Method Detection Limit: Based on the TPH results for seven low-concentr ation-range diesel soil PE samples, the method
detection limits  were determined to be 20 and 6.32 milligrams per kilogram for the PetroFLAG™ System and reference
method, respectively.

Accuracy and Precision: Seventy-one of 97 PetroFLAG™ System results (  73 percent) used to draw conclusions regarding
whether the TPH concentration in a given sampling area or samp  le type exceeded a specified action level agreed with those
of the reference  method; 26 PetroFLAG™ System conclusi  ons were false positives. There were no false negatives.

Of 91 PetroFLAG™ System results used to  assess measurem ent bias, 11 were within 30 percent, 9 were within 30 to
50 percent, and 71 were not within 50 percent of the refere nee method results; 82 PetroFLAG™ System results were biased
high, and 9 were biased low.

For soil environmental samples, the PetroFLAG™ System results  were statistically (1) the same as the reference method
results for one of the four sampling areas and (2) different fro m the reference method results for three of the sampling areas.
For soil PE samples, the PetroFLAG™ System results were sta tistically (1) the same as the reference method results for high-
concentration-range diesel samples and (2) different from the  reference method results for blank samples, high-concentration-
range weathered gasoline samples, and low- and medium-concentr ation-range diesel samples. For liquid PE samples, the
PetroFLAG™ System results were statis tically different from the reference method results for both weathered gasoline and
diesel samples.

The PetroFLAG™ System results correlated highly with the refe rence method results for one of the four sampling areas and
diesel soil PE samples (the square of the correlation coefficient [R 2] values were greater than 0.90, and F-test probability
values were less than 5 percent). The PetroFLAG™ System  results correlated moderately w ith the reference method results
for two of the four sampling areas (R2 values were 0.84 and 0.86, and F-test pr obability values were less than 5 percent).
The PetroFLAG™ System results correlated weakly with the re ference method results for one of the four sampling areas and
                            The accompanying notice is an integral part of this verification statement.                   September 2001

                                                      iv

-------
weathered gasoline soil PE samples (R 2 values were 0.42 and 0.10, respectively, and F-test probability values were greater
than 5 percent).

Comparison of the PetroFLAG™ System and reference method me dian relative standard deviations (RSD) showed that the
PetroFLAG™ System and the reference method exhibited sim  ilar overall precision.  Specifically, the median RSD ranges
were 6 to 19 percent and 5.5 to 16 percent for the PetroFLAG™  System and reference method, respectively.  The analytical
precision was about the same for the PetroFLAG™ System (a me dian relative percent difference of 5) and reference method
(a median relative percent difference of 4).

Effect oflnterferents: The PetroFLAG™ System showed a mean response of less than 5 percent for neat methyl-tert-butyl
ether (MTBE) and  tetrachloroethene (PCE) and for soil spiked with humic acid. The device's mean responses for neat
Stoddard solvent; turpentine; and l,2,4-trichlorobenzenewere42. 5,103,and 16 percent, respectively. The reference method
showed varying mean responses for MTBE (39 percent); PC E (17.5 percent); Stoddard solvent (85 percent); turpentine
(52 percent); 1,2,4-trichlorobenzene (50 percent); and humic acid (0 percent). For the demonstration, MTBE and Stoddard
solvent were included in the definition of TPH.

Effect of Moisture Content: The PetroFLAG™ System showed a statistically significant decrease (17 percent) in TPH results
when the soil moisture content was increased from 9 to 16 percent for weathered gasoline soil PE samples; the reference
method TPH results were unaffected. Both PetroFLAG™ Sy stem and reference method TPH results were unaffected when
the soil moisture content was increased from less than 1 to 9 percent  for diesel soil PE samples.

Measurement Time: From the time of sample receipt, Dexsil required 50  hours, 40 minutes, to prepare a draft data package
containing TPH results for 181 samples and 10 extract duplicates compared to 30 days for the reference method, which was
used to analyze 199 samples and 13 extract duplicates.

Measurement Costs: The TPH measurement cost for 181 samples and  10 extract duplicates  was estimated to be $6,390,
including the capital equipment purchase cost of $695, for the PetroFLAG™ System compared to $38,560 for the reference
method.

Key demonstration findings are summarized below for the secondary objectives.

Skill and Training Requirements: The PetroFLAG™ System can be operated by one person with basic wet chemistry skills.
The sample analysis procedure for the device can be learned in the field with a few practice attempts.

Portability: The PetroFLAG™ System is battery-operated and requi res no alternating current power source. The device can
be easily moved between sampling areas in the field, if necessary.

Durability and Availability of the Device : All items in the PetroFLAG™ System are available from Dexsil.  During a
6-month warranty period, Dexsil will supply replacement parts for the device by overnight courier service at no cost. During
the demonstration, none of the device's re usable items malfunctioned or was damaged.

In summary, during the demonstration, the PetroFLAG™ System exhibited the following desirable characteristics of a field
TPH measurement device: (1) good precision, (2) lack of sensitiv ity to interferents that are not petroleum hydrocarbons (PCE
and humic acid), (3) low measurement costs, and (4) ease of us e. In addition, the PetroFLAG™ System exhibited moderate
sample throughput. Based on action level conclusions and st atistical correlations, the PetroFLAG™ System TPH results
compared well with those of the reference method; however , the device exhibited a high bias, and its TPH results were
determined to be statistically different from those of the refe rence method. In addition, turpentine and 1,2,4-trichlorobenzen e
biased the device's  TPH results high. Moreover, an increase  in soil moisture content biased  the device's TPH results low
for weathered gasoline soil PE samples. Collectively, the de monstration findings indicated that the user should exercise
caution when considering the device for a specific field TPH measurement application.

Original
signed by

Gary J. Foley, Ph.D.
Director
National Exposure Research Laboratory
Office of Research  and Development
 NOTICE: EPA verifications are based on an evaluation of technology performance under specific, predetermined criteria and
 appropriate quality assurance procedures. The EPA makes no expressed or implied warranties as to the performance of the technology
 and does not certify that a technology will always operate as verified.  The end user is solely responsible for complying with any and
 all applicable federal, state, and local requirements.
                            The accompanying notice is an integral part of this verification statement.                   September 2001
                                                       V

-------
                                      Foreword
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the
nation's natural resources. Under the mandate of national environmental laws, the agency strives to
formulate and implement actions leading to a compatible balance between human activities and the
ability of natural systems to support and nurture life. To meet this mandate, the EPA's Office of
Research and  Development provides  data and  scientific support that can  be used to solve
environmental problems, build the scientific knowle dge base needed to manage ecological resources
wisely, understand how pollutants affect public health, and prevent or reduce environmental risks.

The National Exposure Research Laboratory (NERL) is  the agency's center for investigation of
technical and management approaches for identifyi ng and quantifying risks to human health and the
environment. Goals of the laboratory's research program are to (1) develop and evaluate methods
and technologies for characterizing and monitoring air, soil, and water; (2) support regulatory and
policy decisions; and (3) provide the scientific s upport needed to ensure effective implementation
of environmental regulations and strategies.

The EPA's  Superfund Innovative Technology Evaluation (SITE) Program evaluates technologies
designed for characterization and remediation of contaminated Superfund and Resource Conservation
and Recovery Act sites. The SITE Program was created to provide reliable cost and performance
data  in  order  to  speed  acceptance  and use  of innovative remediation,  characterization,
and monitoring technologies by the regulatory and user community.

Effective measurement and monitoring  technologies  are needed to assess the  degree of
contamination at a site, provide data that can be used to determine the risk to public health or the
environment, supply the necessary cost and performance data to select the most appropriate
technology, and monitor the success or failure of a remediation process. One component of the EPA
SITE Program, the Monitoring and Measurement Technology (MMT) Program, demonstrates and
evaluates innovative technologies to meet these needs.

Candidate technologies can originate within the fe deral government or the private sector. Through
the  SITE Program, developers are given the opportun ity to conduct a rigorous demonstration of their
technologies under actual field conditions.  By comp leting the demonstration and distributing the
results, the agency establishes a baseline for acceptance and use of these technologies.  The MMT
Program is administered by the Environmental Sc iences Division of NERL in Las Vegas, Nevada.
                                           Gary J. Foley, Ph.D.
                                           Director
                                           National Exposure Research Laboratory
                                           Office of Research and Development
                                           VI

-------
                                       Abstract
The PetroFLAG™ System developed by Dexsil® Corporation (Dexsil) was demonstrated under the
U.S. Environmental Protection Agency Supe rfund Innovative Technology Evaluation Program in
June 2000 at the Navy Base Ventura County site in Port Hueneme, California. The purpose of the
demonstration was to collect reliable performance and cost data for the PetroFLAG™ System and
six other field measurement devices for total petroleum hydrocarbons (TPH) in soil. In addition to
assessing ease of device operation, the key objectives of the demonstration included determining the
(1) method detection limit, (2) accuracy and precision, (3) effects of interferents and soil moisture
content on TPH measurement, (4) sample throughput, and (5) TPH measurement costs for each
device.  The demonstration involved analysis of both performance evaluation (PE) samples and
environmental samples collected hi four areas contam inated with gasoline, diesel, or other petroleum
products. The performance and cost results for a  given field measurement device were compared to
those for an off-site laboratory reference  method, "Test Methods for Evaluating Solid Waste"
(SW-846) Method 8015B (modified).  During the demonstration, Dexsil required 50 hours,
40 minutes, for TPH measurement of 181 samples and 10 extract duplicates. The TPH measurement
costs for these samples were estimated to be $6,390 for the PetroFLAG™ System compared to
$38,560 for  the reference method. The method detection limits were determined to be 20 and
6.32 milligrams per kilogram  for the PetroFLAG™  System and reference method, respectively.
During the demonstration, the PetroFLAG™ System exhibited good precision and ease of use. The
device's mean responses for interferents that are considered to be petroleum hydrocarbons were
mixed (0 and 42.5 percent for neat methyl-tert-butyl ether and Stoddard solvent, respectively). The
device's mean responses for interferents that are not considered to be petroleum hydrocarbons were
also mixed  (1.5,  103, and  16 percent for  neat  tetrachloroethene;  turpentine;  and  1,2,4-
trichlorobenzene,  respectively, and 2.5 percent for soil spiked with humic acid). In addition, an
increase in soil moisture content biased the device's TPH results low for weathered gasoline soil PE
samples. Based on action level conclusions and statistical correlations, the PetroFLAG™ System
TPH results compared well with those of the refe rence method; however, the device exhibited a high
bias, and its TPH  results were determined to be  statistically different from those of the reference
method. Collectively, the demonstration findings indicated that the user should exercise caution
when considering the device for a specific field TPH measurement application.
                                           vn

-------
                                      Contents


Chapter                                                                         Page

Notice	ii

Verification Statement	,	  iii

Foreword  	  vi

Abstract 	t	vii

Figures	  xi

Tables	xii

Abbreviations, Acronyms, and Symbols	  xiv

Acknowledgments  	 xvi

1      Introduction	1
        1.1    Description of SITE Program  	1
        1.2    Scope of Demonstration	4
        1.3    Components and Definition of TPH  	4
              1.3.1   Composition of Petroleum and Its Products 	4
                     1.3.1.1  Gasoline	5
                     1.3.1.2  Naphthas 	5
                     1.3.1.3  Kerosene	  . 6
                     1.3.1.4  JetFuels	6
                     1.3.lr.5  Fuel Oils	7
                     1.3.1.6  Diesel	7
                     1.3.1.7  Lubricating Oils	7
              1.3.2   Measurement of TPH  	7
                     1.3.2.1  Historical Perspective  	7
                     1.3.2.2  Current Options for TPH Measurement in Soil 	8
                     1.3.2.3  Definition of TPH  	9

2      Description of Emulsion Turbidimetry and the PetroFLAG™ System  	11
       2.1    Description of Emulsion Turbidimetry  	11
       2.2    Description of the PetroFLAG™ System  	12
              2.2.1   Device Description  	12
              2.2.2   Operating Procedure	14
                                         vin

-------
                              Contents (Continued)



Chapter                                                                         Page

       2.3    Developer Contact Information	15

3      Demonstration Site Descriptions  	16
       3.1    Navy Base Ventura County Site  	17
              3.1.1   Fuel Farm Area	17
              3.1.2   Naval Exchange Service Station Area 	18
              3.1.3   Phytoremediation Area	18
       3.2    Kelly Air Force Base Site  	19
       3.3    Petroleum Company Site	19

4      Demonstration Approach  	21
       4.1    Demonstration Objectives 	21
       4.2    Demonstration Design	21
              4.2.1   Approach for Addressing Primary Objectives  	23
              4.2.2   Approach for Addressing Secondary Objectives   	26
       4.3    Sample Preparation and Management	31
              4.3.1   Sample Preparation	31
              4.3.2   Sample Management	33

5      Confirmatory Process  	37
       5.1    Reference Method Selection 	37
       5.2    Reference Laboratory Selection	39
       5.3    Summary of Reference Method	39

6      Assessment of Reference Method Data Quality 	48
       6.1    Quality Control CheckResults	48
              6.1.1   GRO Analysis	48
              6.1.2   EDRO Analysis	51
       6.2    Selected Performance Evaluation Sample Results	57
       6.3    Data Quality	60

7      Performance of the PetroFLAG™ System	61
       7.1    Primary Objectives	61
              7.1.1   Primary Objective PI: Method Detection Limit	63
              7.1.2   Primary Objective P2: Accuracy and Precision 	64
                     7.1.2.1 Accuracy  	64
                     7.1.2.2 Precision	74
              7.1.3   Primary Objective P3: Effect of Interferents  	75
                     7.1.3.1 Interferent Sample Results	77
                     7.1.3.2 Effects of Interferents on TPH Results for Soil Samples	80
              7.1.4   Primary Objective P4: Effect of Soil Moisture Content	87
              7.1.5   Primary Objective P5: Time Required for TPH Measurement  	87
       7.2    Secondary Objectives	90
              7.2.1   Skill and Training Requirements for Proper Device Operation  	90
              7.2.2   Health and Safety Concerns Associated with Device Operation 	91
                                          IX

-------
                             Contents (Continued)
Chapter

              7.2.3   Portability of the Device	91
              7.2.4   Durability of the Device 	92
              7.2.5   Availability of the Device and Spare Parts  	92

8      Economic Analysis	93
       8.1     Issues and Assumptions	93
              8.1.1   Capital Equipment Cost 	93
              8.1.2   Cost of Supplies 	93
              8.1.3   Support Equipment Cost	94
              8.1.4   Labor Cost	94
              8.1.5   Investigation-Derived Waste Disposal Cost	94
              8.1.6   Costs Not Included	94
       8.2     PetroFLAG™ System Costs 	95
              8.2.1   Capital Equipment Cost 	95
              8.2.2   Cost of Supplies 	95
              8.2.3   Support Equipment Cost	96
              8.2.4   Labor Cost	96
              8.2.5   Investigation-Derived Waste Disposal Cost	96
              8.2.6   Summary of PetroFLAG™ System Costs  	96
       8.3     Reference Method Costs 	96
       8.4     Comparison of Economic Analysis Results	96

9      Summary of Demonstration Results	98

10     References	103

Appendix      Supplemental Information Provided by the Developer  	105

-------
                                       Figures

Figure                                                                            Page
1 -1.    Distribution of various petroleum hydrocarbon types throughout boiling point
       range of crude oil 	6
5-1.    Reference method selection process 	38
7-1.    Summary of statistical analysis of TPH results	62
7-2.    Measurement bias for environmental samples  	67
7-3.    Measurement bias for soil performance evaluation samples	70
7-4.    Linear regression plots for environmental samples	73
7-5.    Linear regression plots for soil performance evaluation samples  	74
                                           XI

-------
                                       Tables
Table                                                                           Page

1-1.    Summary of Calibration Information for Infrared Analytical Method	8
1-2.    Current Technologies for TPH Measurement	9
2-1.    PetroFLAG™ System Method Detection Limits and Response Factors for
       Petroleum Products  	12
2-2.    PetroFLAG™ System Components  	13
3-1.    Summary of Site Characteristics	17
4-1.    Action Levels Used to Evaluate Analytical Accuracy  	23
4-2.    Demonstration Approach 	27
4-3.    Environmental Samples 	32
4-4.    Performance Evaluation Samples	34
4-5.    Sample Container, Preservation, and Holding Time Requirements 	36
5-1.    Laboratory Sample Preparation and Analytical Methods	39
5-2.    Summary of Project-Specific Procedures for GRO Analysis	40
5-3.    Summary of Project-Specific Procedures for EDRO Analysis	44
6-1.    Summary of Quality Control Check Results for GRO Analysis	52
6-2.    Summary of Quality Control Check Results for EDRO Analysis  	56
6-3.    Comparison of Soil and Liquid Performance Evaluation Sample Results	58
6-4.    Comparison of Environmental Resource Associates Historical Results to
       Reference Method Results  	59
7-1.    TPH Results for Low-Concentration-Range Diesel Soil Performance Evaluation
       Samples	63
7-2.    PetroFLAG™ System Response Factors 	65
7-3.    Action Level Conclusions	66
7-4.    Statistical Comparison of Petro FLAG™ System and Reference Method TPH
       Results for Environmental Samples  	68
7-5.    Statistical Comparison of PetroFL AG™ System and Reference Method TPH
       Results for Performance Evaluation Samples	72
7-6.    Summary of Linear Regression Analysis Results	75
                                         xn

-------
                              Tables (Continued)
Table                                                                         Page
7-7.    Summary of PetroFLAG™ System and Reference Method Precision for Field
       Triplicates of Environmental Samples 	76
7-8.    Summary of PetroFLAG™ System and Reference Method Precision for Extract
       Duplicates	77
7-9.    Comparison of PetroFLAG™ System and Reference Method Precision for
       Replicate Performance Evaluation Samples	78
7-10.   Comparison of PetroFLAG™ System and Reference Method Results for
       Interferent Samples	79
7-11.   Comparison of PetroFLAG™ System and Reference Method Results for Soil
       Performance Evaluation Samples Containing Interferents	81
7-12.   Comparison of Results for Soil Performance Evaluation Samples at Different
       Moisture Levels 	88
7-13.   Time Required to Complete TPH Measurement Activities Using the PetroFLAG™
       System  	89
8-1.    PetroFLAG™  System Cost Summary 	95
8-2.    Reference Method Cost Summary  	97
9-1.    Summary of PetroFLAG™ System Results for the Primary Objectives 	99
9-2.    Summary of PetroFLAG™ System Results for the Secondary Objectives	  102
                                        Xlll

-------
                   Abbreviations, Acronyms, and Symbols
AC
AEHS
AFB
API
ASTM
bgs
BTEX
BVC
CCV
CFC
CFR
DC
DER
Dexsil
DRO
EDRO
EPA
EPH
ERA
FFA
FID
GC
GRO
ICV
IDW
ITVR
kg
L
LCS
LCSD
MCAWW
MDL
Means
mg
min
Greater than
Less than
Less than or equal to
Plus or minus
Microgram
Alternating current
Association for Environmental Health and Sciences
Air Force Base
American Petroleum Institute
American Society for Testing and Materials
Below ground surface
Benzene, toluene, ethylbenzene, and xylene
Base Ventura County
Continuing calibration verification
Chlorofluorocarbon
Code of Federal Regulations
Direct current
Data evaluation report
Dexsil® Corporation
Diesel range organics
Extended diesel range organics
U.S. Environmental Protection Agency
Extractable petroleum hydrocarbon
Environmental Resource Associates
Fuel Farm Area
Flame ionization detector
Gas chromatograph
Gasoline range organics
Initial calibration verification
Investigation-derived waste
Innovative technology verification report
Kilogram
Liter
Laboratory control  sample
Laboratory control  sample duplicate
"Methods for Chemical Analysis of Water and Wastes"
Method detection limit
R.S. Means Company
Milligram
Minute
                                        xiv

-------
            Abbreviations, Acronyms, and Symbols (Continued)
mL
MMT
MS
MSD
MTBE
n-Cx
NERL
NEX
ng
nm
ORD
ORO
OSWER
PC
PCB
PCE
PE
PHC
PPE
PRA
PRO
QA
QC
R2
RPD
RSD
SFT
SITE
STL Tampa East
SW-846
TPH
UST
VPH
Milliliter
Monitoring and Measurement Technology
Matrix spike
Matrix spike duplicate
Methyl-tert-butyl ether
Alkane with "x" carbon atoms
National Exposure Research Laboratory
Naval Exchange
Nanogram
Nanometer
Office of Research and Development
Oil range organics
Office of Solid Waste and Emergency Response
Petroleum company
Polychlorinated biphenyl
Tetrachloroethene
Performance evaluation
Petroleum hydrocarbon
Personal protective equipment
Phytoremediation Area
Petroleum range organics
Quality assurance
Quality control-
Square of the correlation coefficient
Relative  percent difference
Relative  standard deviation
Slop Fill Tank
Superfund Innovative Technology Evaluation
Severn Trent Laboratories in Tampa, Florida
"Test Methods for Evaluating Solid Waste"
Total petroleum hydrocarbons
Underground storage tank
Volatile petroleum hydrocarbon
                                        xv

-------
                                Acknowledgments
This report was prepared for the U.S. Environm ental Protection Agency (EPA) Superfund Innovative
Technology Evaluation Program under the direction and coordination of Dr. Stephen Billets of the
EPA National Exposure Research Laboratory (NERL)—Environmental Sciences Division in Las
Vegas, Nevada. The EPA NERL thanks Mr. Ernest Lory of Navy Base Ventura County, Ms. Amy
Whitley of Kelly Air Force Base, and Mr. Jay Simonds of Handex of Indiana for their support in
conducting field activities for the project. Mr. Eric Koglin of the EPA NERL served as the technical
reviewer of this report. Mr. Roger Claff of the American Petroleum Institute, Mr. Dominick De
Angelis of ExxonMobil Corporation, Dr. Ileana Rhod es of Equilon Enterprises, and Dr. Al Verstuyft
of Chevron Research and Technology Company served as the peer reviewers of this report.

This report was prepared for the EPA by Dr. Kirankumar Topudurti and Mr. Eric Monschein of
Tetra Tech EM Inc.  Special acknowledgment is gi ven to Mr. Jerry  Parr of Catalyst Information
Resources,  L.L.C.,  for serving  as the  technical  consultant for the project.   Additional
acknowledgment and thanks are given to Ms. Jeanne Kowalski, Mr. Jon Mann, Mr.  Stanley
Labunski, and Mr. Joe Abboreno of Tetra Tech EM Inc. for their assistance during the preparation
of this report.
                                         xvi

-------
                                              Chapter 1
                                             Introduction
The U.S. Environmental Protection Agency (EPA) Office
of Research and Development (ORD) National Exposure
Research Laboratory (NERL) conducted a demonstration
of seven innovative field measurement devices for total
petroleum hydrocarbons (TPH) in soil. The demonstration
was conducted as part of the EPA Superfund Innovative
Technology  Evaluation  (SITE)  Monitoring   and
Measurement Technology (MMT) Program using TPH-
cdntaminated soil from five areas located in three regions
of the United States. The demonstration was conducted at
Port Hueneme, California,  during the week of June 12,
2000.  The purpose of the  demonstration was to obtain
reliable performance and cost data on field measurement
devices in order to provide (1) potential users with a better
understanding of the devices' performance and operating
costs under  well-defined field conditions and  (2) the
developers with documented results that will assist them in
promoting acceptance and use of their devices. The TPH
results obtained using the seven field measurement devices
were compared  to the TPH results  obtained  from  a
reference laboratory chosen for the demonstration, which
used a reference method modified for the demonstration.

This  innovative  technology verification  report  (ITVR)
presents demonstration performance results and associated
costs for the PetroFLAG™ System developed by Dexsil ®
Corporation (Dexsil).  Specifically, this report describes
the SITE Program, the scope of the demonstration, and
the  components  and definition of TPH (Chapter 1);
the  innovative  field measurement  device and  the
technology upon which it is based (Chapter 2); the three
demonstration sites  (Chapter  3);  the  demonstration
approach (Chapter 4);  the selection of the reference
method and  laboratory  (Chapter 5); the  assessment of
reference  method  data  quality  (Chapter  6);  the
performance of the field measurement device (Chapter 7);
the economic analysis for the field measurement device
and reference method (Chapter 8);  the  demonstration
results in summary form (Chapter 9); and the references
used to prepare the ITVR (Chapter 10).  Supplemental
information  provided by Dexsil  is presented in the
appendix.

1.1    Description of SITE Program

Performance verification of innovative  environmental
technologies is an integral  part of the regulatory and
research mission of the EPA. The SITE Program was
established by the EPA Office of Solid Waste and
Emergency Response (OSWER) and  ORD under the
Superfund Amendments and Reauthorization Act of 1986.
The  overall  goal  of the SITE Program is to conduct
performance verification studies and  to promote the
acceptance of innovative technologies that may be used
to achieve long-term protection  of human health and
the environment.   The program is designed to  meet
three  primary  objectives:    (1)  identify and remove
obstacles to the development and commercial use of
innovative  technologies,  (2)  demonstrate  promising
innovative technologies and gather reliable performance
and cost information to support site characterization and
cleanup activities, and (3) develop procedures and policies
that  encourage the use of  innovative technologies at
Superfund sites  as well as  at  other  waste  sites  or
commercial facilities.

The  intent  of a  SITE  demonstration is  to obtain
representative,  high-quality performance and cost data on
one or more innovative technologies so that potential users
can assess the suitability of a given technology  for a
specific application.   The SITE  Program includes the
following elements:

   MMT Program—Evaluates innovative technologies
   that sample, detect, monitor,  or measure hazardous
   and toxic substances. These technologies are expected

-------
to  provide  better,   faster,  or  more  cost-effective
methods  for  producing  real-time  data during  site
characterization  and  remediation   studies   than  do
conventional technologies.

    Remediation  Technology   Program—Conducts
    demonstrations of innovative treatment technologies to
    provide reliable performance, cost, and applicability
    data for site cleanups.

•   Technology  Transfer  Program—Provides  and
    disseminates technical information in the form of
    updates,  brochures,  and  other  publications  that
    promote  the  SITE  Program   and  participating
    technologies. The Technology Transfer Program also
    offers technical assistance, training, and workshops to
    support the technologies.  A significant number of
    these activities are performed by EPA's Technology
    Innovation Office.

The TPH  field measurement device demonstration was
conducted as part of the MMT Program, which provides
developers of innovative hazardous  waste  sampling,
detection, monitoring, and measurement devices with an
opportunity to  demonstrate  the  performance  of their
devices under actual field conditions. These devices may
be used to sample, detect, monitor, or measure hazardous
and toxic substances in water, soil gas, soil, and sediment.
The technologies include chemical sensors for in situ (in
place) measurements, soil and sediment samplers, soil gas
samplers, groundwater samplers, field-portable analytical
equipment, and other systems that support field sampling
or data acquisition and analysis.

The MMT Program promotes acceptance of technologies
that can be used to (1) accurately assess the degree of
contamination at  a  site, (2) provide  data to evaluate
potential effects on human health and the environment,
(3) apply data to assist in selecting the most appropriate
cleanup action,  and (4) monitor  the  effectiveness of a
remediation process.  The program places a high priority
on  innovative technologies that provide more  cost-
effective, faster, and safer methods for producing real-time
or near-real-time data than do conventional, laboratory-
based technologies.   These innovative technologies are
demonstrated under field conditions, and the results are
compiled, evaluated, published, and disseminated by the
ORD. The primary objectives of the MMT Program are as
follows:
•   Test and verify the performance of innovative field
    sampling and  analytical technologies that enhance
    sampling,  monitoring,  and  site  characterization
    capabilities

•   Identify  performance   attributes  of   innovative
    technologies to address field sampling,  monitoring,
    and characterization problems in a more cost-effective
    and efficient manner

•   Prepare protocols, guidelines, methods, and  other
    technical publications that enhance acceptance of
    these technologies for routine use

The MMT Program is administered by the Environmental
Sciences Division  of the NEPvL in Las Vegas, Nevada.
The NERL is the EPA center for investigation of technical
and  management  approaches  for   identifying   and
quantifying risks to human health and the environment.
The NEPvL mission components include (1) developing
and evaluating  methods and technologies for sampling,
monitoring, and characterizing  water,  air, soil,  and
sediment; (2) supporting regulatory and policy decisions;
and (3) providing the technical support needed to ensure
effective implementation of environmental  regulations
and strategies.    By demonstrating  innovative  field
measurement devices for TPH in soil, the MMT Program
is supporting the development and evaluation of methods
and  technologies   for  field  measurement  of  TPH
concentrations in a variety of soil types.  Information
regarding the selection of field measurement devices for
TPH is available in American Petroleum Institute (API)
publications (API 1996, 1998).

The MMT Program's technology verification process is
designed to conduct demonstrations  that will  generate
high-quality data so that potential users have reliable
information regarding device performance and cost.  Four
steps  are inherent in the process: (1) needs identification
and technology selection, (2) demonstration planning
and  implementation,   (3)  report   preparation,   and
(4)  information distribution.

The first step of the  verification process begins  with
identifying technology needs of the EPA and the regulated
community.   The EPA  regional  offices, the  U.S.
Department of Energy, the U.S. Department of Defense,
industry, and state environmental regulatory agencies are
asked  to  identify technology  needs  for  sampling,
monitoring, and measurement of environmental media.
Once  a need is identified, a search is conducted to identify

-------
suitable technologies that will address the need.  The
technology search and identification process consists of
examining industry  and trade publications, attending
related  conferences, exploring leads  from technology
developers and industry experts, and reviewing responses
to Commerce Business Daily announcements. Selection of
technologies for field testing includes evaluation of the
candidate technologies based on  several criteria.   A
suitable technology for field testing

•   Is designed for use in the field

•   Is  applicable to  a  variety  of  environmentally
    contaminated sites

    Has  potential for solving problems that  current
    methods cannot satisfactorily address

•   Has estimated costs that  are lower  than those  of
    conventional methods

•   Is likely to achieve better results than current methods
    in areas such as data quality and turnaround time

•   Uses techniques  that are easier or safer than current
    methods

•   Is commercially available

Once  candidate   technologies  are  identified,  their
developers are asked to participate  in a developer
conference.  This conference gives the  developers  an
opportunity to describe their technologies' performance
and to learn about the MMT Program.

The second step of the verification process is to plan and
implement a demonstration that will generate high-quality
data to assist potential users  in selecting a technology.
Demonstration   planning   activities  include   a
predemonstration sampling and analysis investigation that
assesses existing conditions at the proposed demonstration
site or sites.  The objectives of the predemonstration
investigation are  to  (1) confirm available information
on   applicable  physical,  chemical,   and  biological
characteristics of contaminated media at the sites to justify
selection of site areas for the demonstration; (2) provide
the technology developers with an opportunity to evaluate
the  areas, analyze representative samples, and identify
logistical requirements; (3)  assess the overall logistical
requirements for  conducting  the  demonstration;  and
(4) provide the reference laboratory with an opportunity
to  identify  any matrix-specific  analytical  problems
associated with the  contaminated media and to propose
appropriate solutions.  Information generated through the
predemonstration investigation is  used to  develop the
final demonstration design and sampling and analysis
procedures.

Demonstration planning activities also include preparing
a detailed demonstration plan that describes the procedures
to be used to verify the performance and cost of each
innovative   technology.    The   demonstration   plan
incorporates  information  generated  during   the
predemonstration investigation as  well as input from
technology developers, demonstration site representatives,
and technical peer reviewers. The demonstration plan also
incorporates the quality assurance (QA) and quality control
(QC) elements needed to produce  data  of sufficient
quality to document the performance and cost of each
technology.

During the demonstration, each innovative technology is
evaluated  independently  and, when  possible  and
appropriate, is compared to a reference technology.  The
performance and cost of one innovative technology are not
compared to those of another technology evaluated in the
demonstration.  Rather, demonstration  data are used to
evaluate  the individual performance, cost, advantages,
limitations, and field applicability of each technology.

As part of the third step of the verification process, the
EPA publishes  a verification statement and a detailed
evaluation of each technology in an ITVR.  To ensure its
quality, the ITVR is published only  after comments from
the technology developer and external peer reviewers are
satisfactorily addressed.  In addition, all demonstration
data used to evaluate each innovative technology are
summarized in  a  data evaluation  report  (DER)  that
constitutes a complete record of the demonstration.  The
DER is  not published  as an EPA  document,  but an
unpublished copy may be obtained form the EPA project
manager.

The fourth step of the verification process is to distribute
information  regarding demonstration results.  To benefit
technology developers and potential technology users, the
EPA distributes demonstration bulletins and  ITVRs
through  direct  mailings,  at  conferences, and  on  the
Internet.  The ITVRs  and additional information on the
SITE Program are available on the EPA ORD web site
(http://www.epa.gov/ORD/SITE).

-------
1.2     Scope of Demonstration

The purpose of the demonstration was to evaluate field
measurement devices for TPH in soil in order to provide
(1) potential users with a better  understanding of the
devices' performance and costs under well-defined field
conditions and (2) the developers with documented results
that will assist them in promoting  acceptance and use of
their devices.

Chapter 2 of this ITVR describes both the technology upon
which the PetroFLAG™ System  is based and the  field
measurement device itself. Because TPH is a "method-
defined parameter," the performance results for the device
are compared to the results  obtained using an off-site
laboratory measurement method—that  is, a reference
method. Details on the selection of the reference method
and laboratory are provided in Chapter 5.

The demonstration  had both primary  and secondary
objectives.   Primary  objectives  were  critical  to the
technology verification and required the use of quantitative
results  to  draw  conclusions  regarding  each  field
measurement device's performance as well as to estimate
the cost of operating the device.   Secondary objectives
pertained to information that was useful but  did not
necessarily require the use of quantitative results to draw
conclusions regarding the performance of each device.
Both the primary and secondary objectives are discussed
in Chapter 4.

To meet the demonstration  objectives,  samples  were
collected from five individual areas  at three sites. The first
site is referred to as the Navy Base Ventura County (BVC)
site; is located in PortHueneme, California; and contained
three  sampling areas. The Navy BVC site lies in  EPA
Region 9.  The second site is referred to as the Kelly Air
Force Base (AFB) site; is located in San Antonio, Texas;
and contained one sampling area. The Kelly AFB site lies
in EPA Region 6. The third site is referred to as the
petroleum company (PC) site, is located in north-central
Indiana, and contained one sampling area. The PC site lies
in EPA Region 5.

In preparation for the demonstration, a predemonstration
sampling and analysis investigation was completed at the
three  sites  in January 2000.  The purpose  of this
investigation was to assess whether the sites and sampling
areas  were appropriate  for evaluating the seven  field
measurement  devices  based  on the  demonstration
objectives. Demonstration field activities were conducted
between June 5 and 18, 2000.  The procedures used to
verify the performance and costs of the field measurement
devices are documented in a demonstration plan completed
in June 2000 (EPA 2000). The plan also incorporates the
QA/QC elements that were needed to generate data of
sufficient quality to document field measurement device
and reference laboratory performance and costs.  The plan
is  available  through   the  EPA   ORD  web   site
(http://www.epa.gov/ORD/SrTE) or from the EPA project
manager.

1.3     Components and Definition of TPH

To  understand the term "TPH," it  is  necessary  to
understand the composition of petroleum and its products.
This  section  briefly  describes  the  composition  of
petroleum  and  its products and  defines  TPH from  a
measurement standpoint.   The  organic compounds
containing only hydrogen and carbon that are present in
petroleum and its derivatives are collectively referred to as
petroleum hydrocarbons (PHC). Therefore, in this ITVR,
the term "PHC"  is used to identify sample constituents,
and the term "TPH" is used to identify analyses performed
and  the   associated  results  (for  example,  TPH
concentrations).

1.3.1   Composition of Petroleum and Its Products

Petroleum is essentially a mixture of gaseous, liquid, and
solid  hydrocarbons that occur  in sedimentary  rock
deposits. On the molecular level, petroleum is a complex
mixture of hydrocarbons; organic compounds of sulfur,
nitrogen, and oxygen; and compounds containing metallic
constituents, particularly vanadium,  nickel,  iron,  and
copper. Based on the limited data available, the elemental
composition of petroleum appears to vary over a relatively
narrow range: 83 to 87 percent carbon, 10 to 14 percent
hydrogen, 0.05 to 6  percent sulfur,  0.1  to 2 percent
nitrogen, and 0.05 to  1.5 percent oxygen.  Metals are
present in petroleum at concentrations of up to 0.1 percent
(Speight 1991).

Petroleum in the crude state (crude  oil)  is a mineral
resource,  but when  refined it  provides  liquid fuels,
solvents, lubricants, and many other marketable products.
The  hydrocarbon  components  of crude  oil  include
paraffinic, naphthenic, and  aromatic groups.   Paraffins
(alkanes)  are  saturated,  aliphatic hydrocarbons  with
straight or branched chains but without any ring structure.

-------
Naphthenes  are  saturated,  aliphatic  hydrocarbons
containing one or more rings, each of which may have one
or more paraffinic side chains (alicyclic hydrocarbons).
Aromatic  hydrocarbons contain  one or more aromatic
nuclei, such as benzene, naphthalene, and phenanthrene
ring systems, that may be linked  with (substituted)
naphthenic rings or paraffinic side chains. In crude oil, the
relationship  among  the   three  primary  groups  of
hydrocarbon components is a result of hydrogen gain or
loss between any two groups.    Another class  of
compounds that is present in petroleum products such as
automobile gasoline but rarely in crude oil is known as
olefins.   Olefins  (alkenes) are  unsarurated, aliphatic
hydrocarbons.

The distribution of paraffins, naphthenes, and aromatic
hydrocarbons depends on the source of crude oil.  For
example, Pennsylvania crude oil contains high levels of
paraffins (about 50 percent), whereas Borneo crude oil
contains  less than 1 percent paraffins.   As  shown in
Figure 1 -1, the proportion of straight or branched paraffins
decreases  with increasing  molecular weight or  boiling
point fraction for a given crude oil; however, this is not
true for naphthenes or aromatic hydrocarbons.   The
proportion of  monocyclonaphthenes  decreases  with
increasing molecular weight or boiling point fraction,
whereas the opposite is true  for polycyclonaphthenes
(for example, tetralin and decalin) and polynuclear
aromatic hydrocarbons;  the  proportion  of mononuclear
aromatic hydrocarbons  appears  to  be  independent of
molecular weight or boiling point fraction.

Various petroleum products  consisting of carbon and
hydrogen  are  formed when crude oil  is subjected to
distillation and other processes in a refinery. Processing of
crude oil results in petroleum products with trace quantities
of metals and organic compounds that contain nitrogen,
sulfur,  and oxygen.  These products include liquefied
petroleum gas, gasoline, naphthas, kerosene, fuel oils,
lubricating oils,  coke,  waxes, and asphalt.   Of these
products,  gasoline,  naphthas,  kerosene, fuel oils,  and
lubricating oils  are  liquids  and may  be  present at
petroleum-contaminated sites.  Except for gasoline and
some naphthas, these  products are made primarily by
collecting particular boiling point fractions of crude oil
from a distillation column.  Because this classification of
petroleum products is based on boiling point and not on
chemical composition, the composition of these products,
including the ratio of aliphatic to aromatic hydrocarbons,
varies depending on the source of crude oil.  In addition,
specific information (such  as boiling  points and carbon
ranges) for different petroleum products, varies slightly
depending on the source of the information. Commonly
encountered forms and blends of petroleum products are
briefly described below.  The descriptions are primarily
based on information in books written by Speight (1991)
and Gary and Handwerk (1993).  Additional information
is provided by Dryoff (1993).
1J.1.1
Gasoline
Gasoline  is  a major  exception  to  the  boiling  point
classification  described  above  because  "straight-run
gasoline" (gasoline directly recovered from a distillation
column) is only a small fraction of the blended gasoline
that is commercially available as fuel.   Commercially
available gasolines are complex mixtures of hydrocarbons
that boil below 180 °C or at most 225 °C and that contain
hydrocarbons with 4 to  12 carbon atoms per molecule. Of
the commercially available gasolines, aviation gasoline has
a narrower boiling range (38 to 170 °C) than automobile
gasoline (-1 to 200 °C).  In addition, aviation gasoline may
contain high  levels of paraffins (50  to 60  percent),
moderate  levels of naphthenes (20 to 30 percent), a low
level of aromatic hydrocarbons  (10 percent), and no
olefins, whereas automobile gasoline may contain up to
30  percent olefins  and  up to  40 percent  aromatic
hydrocarbons.

Gasoline composition can vary widely depending on the
source of crude  oil.  In addition, gasoline composition
varies from region to region because of consumer needs for
gasoline with a high  octane rating  to prevent engine
"knocking."  Moreover, EPA regulations regarding the
vapor pressure of gasoline, the chemicals used to produce
a high  octane rating,  and cleaner-burning fuels have
affected gasoline composition. For example, when use of
tetraethyl  lead to produce gasoline  with  a high octane
rating was banned by the EPA, oxygenated fuels came into
existence. Production of these fuels included addition of
methyl-tert-butyl  ether (MTBE),  ethanol,  and  other
oxygenates.  Use of oxygenated fuels also  results in
reduction  of air pollutant emissions (for example, carbon
monoxide and nitrogen oxides).

13.1.2     Naphthas

"Naphtha" is a generic term applied to petroleum solvents.
Under  standardized distillation  conditions,  at  least
10 percent of naphthas should distill below 175 °C, and at
least 95 percent of naphthas should distill  below 240 °C.
Naphthas  can be both aliphatic and aromatic and contain

-------
     0
           Lighter oils
                             *• Increasing nitrogen, oxygen, sulfur, and metal content
                                                                          *•  Heavier oils and residues
                                                                    Polynuclear aromatic hydrocarbons
                          Mononuclear aromatic hydrocarbons


         V'? Monocyclonapnthenes *"
                                                                             Polycyclonaphthenes
               Straight and branched paraffins
                          100
                                             200                300
                                                   Boiling point, °C
Source: Speight 1991

Figure 1-1. Distribution of various petroleum hydrocarbon types throughout boiling point range of crude oil.
                            400
500
hydrocarbons with 6 to  14 carbon atoms per molecule.
Depending on the intended use of a naphtha, it may be free
of aromatic hydrocarbons (to make it odor-free) and sulfur
(to make it less toxic and less corrosive).  Many forms of
naphthas are commercially available, including Varnish
Makers' and Painters' naphthas (Types I and IT), mineral
spirits  (Types I through IV),  and  aromatic  naphthas
(Types I and II).  Stoddard solvent is an example of an
aliphatic naphtha.
13.13
           Kerosene
Kerosene is a straight-run petroleum fraction that has a
boiling point range of 205 to 260 °C. Kerosene typically
contains hydrocarbons with 12 or more carbon atoms per
molecule.  Because of its use as an indoor fuel, kerosene
must be free of aromatic and unsaturated hydrocarbons as
well as sulfur compounds.
13.1.4
           Jet Fuels
Jet fuels, which are also known as aircraft turbine fuels,
are  manufactured by  blending gasoline, naphtha,  and
kerosene in varying proportions. Therefore, jet fuels may
contain a carbon  range  that  covers  gasoline through
kerosene.  Jet fuels are used  in  both military  and
commercial aircraft.  Some examples of jet fuels include
Type A, Type A-l, Type B, JP-4, JP-5, and JP-8. The
aromatic hydrocarbon content of these fuels ranges from
20 to 25 percent.  The military jet fuel JP-4 has a wide
boiling point range (65 to 290 °C), whereas commercial jet
fuels, including JP-5 and Types A and A-l, have  a
narrower boiling point range (175 to 290 °C) because of
safety considerations. Increasing concerns over combat
hazards associated with JP-4 jet fuel led to development of
JP-8 jet  fuel, which has  a flash point of 38 °C  and a
boiling  point range  of 165  to 275 °C.  JP-8 jet fuel
contains hydrocarbons  with  9 to  15 carbon  atoms per
molecule. Type B jet fuel has a boiling point  range of
55 to 230 °C and a carbon range of 5 to 13 atoms per
molecule.   A  new specification  is currently  being
developed by the American  Society for Testing and
Materials (ASTM) for Type B jet fuel.

-------
13.1.5
Fuel Oils
Fuel oils  are  divided  into two  classes:  distillates and
residuals.  No. 1 and 2 fuel oils are distillates and include
kerosene, diesel, and home heating oil. No. 4,5, and 6 fuel
oils are residuals or black oils, and they all contain crude
distillation tower bottoms (tar)  to which cutter stocks
(semirefined or refined distillates) have been added. No. 4
fuel oil contains the most cutter stock, and No. 6 fuel oil
contains the least.

Commonly available fuel oils include No. 1,2,4,5, and 6.
The boiling points, viscosities, and densities of these fuel
oils increase with  increasing number designation.  The
boiling point ranges for No. 1,2, and 4 fuel oils are about
180 to 320, 175 to 340, and 150 to 480 °C, respectively.
No. 1 and 2 fuel oils contain hydrocarbons with 10 to 22
carbon atoms per molecule; the carbon range for No. 4 fuel
oil is 22 to 40  atoms per molecule.  No. 5 and 6 fuel oils
have a boiling point range of 150 to 540 °C but differ in
the amounts of residue they contain: No. 5 fuel oil contains
a small amount of residue, whereas No. 6 fuel oil contains
a large amount. No. 5 and 6 fuel oils contain hydrocarbons
with 28 to 90 carbon atoms per molecule.   Fuel oils
typically contain about 60 percent aliphatic hydrocarbons
and 40 percent aromatic hydrocarbons.
1.3.1.6
Diesel
Diesel is primarily used to operate motor vehicle and
railroad diesel engines. Automobile diesel is available in
two grades: No. 1 and 2. No. 1 diesel, which is sold in
regions with cold climates, has a boiling point range of 180
to 320 °C and a cetane number above 50. The cetane
number is similar  to the octane number of gasoline; a
higher number corresponds to less knocking. No. 2 diesel
is very similar to No. 2 fuel oil. No. 2 diesel has a boiling
point range of  175 to 340 °C  and a minimum  cetane
number of 52. No.  1 diesel is used in high-speed engines
such as truck and bus engines, whereas No. 2 diesel is used
in other diesel engines.  Railroad diesel is similar to No. 2
diesel but has a higher boiling point (up to 370 °C) and
lower cetane number (40 to 45).  The ratio of aliphatic to
aromatic hydrocarbons in diesel is about 5. The carbon
range for hydrocarbons present in diesel is 10 to 28 atoms
per molecule.

1.3.1.7     Lubricating Oils

Lubricating oils can be distinguished from other crude oil
fractions by their high boiling points (greater than 400 °C)
and  viscosities.   Materials  suitable for  production  of
lubricating oils are composed principally of hydrocarbons
containing 25 to 35 or even 40 carbon atoms per molecule,
whereas residual stocks may contain hydrocarbons with
50 to 60  or more (up to 80 or so) carbon atoms per
molecule. Because it is difficult to isolate hydrocarbons
from the  lubricant  fraction  of petroleum,  aliphatic  to
aromatic hydrocarbon ratios are not well documented for
lubricating oils. However, these ratios are expected to be
comparable to those of the source crude oil.

1.3.2  Measurement of TPH

As   described in Section  1.3.1,  the  composition  of
petroleum and its products is complex and variable, which
complicates TPH measurement. The measurement of TPH
in soil is further complicated by weathering effects. When
a petroleum product is released to soil, the product's
composition  immediately  begins  to  change.    The
components with lower boiling points are volatilized, the
more water-soluble components migrate to groundwater,
and  biodegradation  can affect many other components.
Within a short period, the contamination remaining in soil
may have only some characteristics in common with the
parent product.

This section  provides  a historical  perspective on  TPH
measurement,  reviews  current  options  for  TPH
measurement  in soil,  and  discusses the definition  of
TPH that was used for the demonstration.

13.2.1     Historical Perspective

Most environmental  measurements are  focused on
identifying and quantifying a particular trace element (such
as lead)  or  organic  compound  (such  as benzene).
However,  for some "method-defined"  parameters, the
particular substance being measured may  yield different
results depending on  the measurement  method used.
Examples of such parameters include oil and grease and
surfactants. Perhaps the most problematic of the method-
defined parameters is TPH. TPH arose as a parameter for
wastewater analyses in the 1960s because of petroleum
industry  concerns that the  original "oil and  grease"
analytical  method, which is gravimetric in nature, might
inaccurately characterize petroleum industry wastewaters
that  contained naturally occurring vegetable  oils and
greases  along with  PHCs.   These  naturally occurring
materials are typically long-chain fatty acids (for example,
oleic acid, the major component of olive oil).

-------
Originally, TPH was defined as any material extracted
with a particular solvent that is not adsorbed by the silica
gel used to remove fatty acids and that is not lost when
the solvent is  evaporated.   Although  this definition
covers most of the components of petroleum products, it
includes many  other organic  compounds  as well,
including  chlorinated  solvents,  pesticides,  and other
synthetic organic chemicals. Furthermore, because of the
evaporation step in the gravimetric analytical method, the
definition   excludes  most  of  the petroleum-derived
compounds in gasoline that are volatile  in nature.  For
these reasons, an infrared analytical method was developed
to measure TPH.  In this method,  a calibration standard
consisting of three components is analyzed at a wavelength
of 3.41  micrometers, which corresponds to an aliphatic
CH2 hydrocarbon stretch.  As shown in Table 1-1, the
calibration  standard is designed  to mimic a petroleum
product having a relative distribution of aliphatic  and
aromatic compounds as well as a certain percentage of
aliphatic CH2 hydrocarbons.  The infrared analytical
method  indicates that any compound that is extracted by
the solvent, is not adsorbed by silica gel, and contains a
CH2 bond is a PHC.  Both the gravimetric and infrared
analytical   methods  include  an  optional,  silica  gel
fractionation step to remove polar, biogenic compounds
such as fatty acids, but this cleanup step can also remove
some petroleum degradation products that are polar in
nature.

In the   1980s, because of the  change  in focus from
wastewater analyses to characterization of hazardous
waste sites that contained contaminated soil, many parties
began to adapt the existing wastewater analytical methods
for application to soil. Unfortunately, the term "TPH" was
in common use, as many states had adopted this term
(and the wastewater  analytical  methods) for cleanup
activities  at  underground storage tank (UST) sites.
Despite  efforts by the API and others to establish new
analyte  names  (for example, gasoline range organics
[GRO] and diesel range organics [DRO]), "TPH" is still
present in many state regulations as a somewhat ill-defined
term, and most state programs still have cleanup criteria for
TPH.

1.3.2.2     Current Options for TPH Measurement
           in SoU

Three widely used technologies measure some form of
TPH in soil to some degree. These technologies were used
as starting points in deciding how to define TPH for the
demonstration.  The three technologies and the analytes
measured are summarized in Table 1-2.

Of the three  technologies, gravimetry and infrared are
discussed in Section 1.3.2.1. The third technology, the gas
chromatograph/flame ionization detector (GC/FID), came
into  use because of the documented shortcomings of the
other two technologies. The GC/FDD had long been used
hi the petroleum refining industry as a product QC tool to
determine the boiling point distribution of pure petroleum
products. In the 1980s, environmental laboratories began
to apply this technology along with sample preparation
methods developed for soil samples to measure PHCs at
environmental levels (Zilis, McDevitt, and  Parr 1988).
GC/FID methods measure all organic compounds that are
extracted by the solvent and that can be chromatographed.
However, because of method limitations, the very volatile
portion  of gasoline compounds containing four  or five
carbon atoms per molecule is not addressed by GC/FID
methods; therefore, 100  percent  recovery cannot be
achieved for pure  gasoline.   This  omission  is not
considered  significant because  these  low-boiling-point
aliphatic compounds (1) are not expected to be present in
environmental samples (because of volatilization) and
(2) pose less  environmental  risk  than  the aromatic
hydrocarbons hi gasoline.
Table 1-1. Summary of Calibration Information for Infrared Analytical Method
Standard
Constituent
Hexadecane
Isooctane
Chlorobenzene
Constituent Type
Straight-chain aliphatic
Branched-chain aliphatic
Aromatic
Portion of Constituent
in Standard
(percent by volume)
37.5
37.5
25
Number of Carbon Atoms
Aliphatic
CH3
2
5
0
CH2
14
1
0
CH
0
1
0
Aromatic
CH
0
0
5
Average
Portion of Aliphatic CH2 in
Standard Constituent
(percent by weight)
91
14
0
35

-------
Table 1-2. Current Technologies forTPH Measurement
Technology
Gravimetry
Infrared
Gas chromatograph/flame ionization detector
What Is Measured
All analytes removed from the sample by the
extraction solvent that are not volatilized
All analytes removed from the sample by the
extraction solvent that contain an aliphatic CH2
stretch
All analytes removed from the sample by the
extraction solvent that can be chromatographed
and that respond to the detector
What Is Not Measured
Volatiles; very polar organics
Benzene, naphthalene, and other aromatic
hydrocarbons with no aliphatic group attached;
very polar organics
Very polar organics; compounds with high
molecular weights or high boiling points
The primary limitation of GC/FID methods relates to the
extraction solvent used. The solvent should not interfere
with the analysis, but to achieve environmental levels of
detection  (in the low milligram per kilogram [mg/kg]
range) for soil, some concentration of the extract is needed
because the sensitivity of the FID is in the nanogram (ng)
range.   This  limitation  has resulted  in  three  basic
approaches for GC/FED  analyses for GRO,  DRO, and
PHCs.

For GRO analysis,  a GC/FID method was developed as
part of research sponsored by API and was the subject of
an interlaboratory validation study (API 1994); the method
was first published in 1990.  hi this method, GRO  is
defined as the sum of the organic compounds in the boiling
point range  of 60  to  170 °C, and the  method uses a
synthetic calibration standard as both a window-defining
mix and a quantitation standard. The GRO method was
specifically incorporated  into EPA "Test Methods for
Evaluating Solid Waste" (SW-846) Method 8015B in 1996
(EPA 1996).  The GRO method uses the purge-and-trap
technique for sample preparation, effectively limiting the
TP,H components to the volatile compounds only.

For DRO analysis, a GC/FID method was developed under
the sponsorship of API as a companion to the GRO method
and was interlaboratory-validated in 1994. In the  DRO
method, DRO  is defined as the  sum  of the organic
compounds in the boiling point range of 170 to 430 °C. As
in the GRO  method, a synthetic calibration  standard  is
used  for  quantitation.   The DRO  method  was  also
incorporated into SW-846 Method 8015B in  1996.  The
technology used in  the DRO method can  measure
hydrocarbons with boiling points up to 540 °C. However,
the hydrocarbons with boiling points in the range of
430 to 540 °C  are specifically excluded from SW-846
Method 8015B so as not to include the higher-boiling-
point petroleum products.  The  DRO  method uses  a
solvent  extraction  and concentration step,  effectively
limiting the method to nonvolatile hydrocarbons.

For PHC analysis, a GC/FED method was developed by
Shell  Oil Company (now Equilon Enterprises).   This
method was interlaboratory-validated along with the GRO
and DRO methods  in an API study in 1994.  The PHC
method  originally  defined PHC  as  the sum of the
compounds  in  the  boiling point  range  of  about
70 to 400 °C, but it now defines PHC as the sum of the
compounds in the boiling point range of 70 to 490 °C.  The
method  provides options for instrument calibration,
including use of synthetic standards, but it recommends
use of products similar to the contaminants present at the
site of concern. The PHC method has not been specifically
incorporated into SW-846; however, the method has been
used as  the basis for the TPH methods in several  states,
including Massachusetts, Washington, and Texas.  The
PHC method uses solvent microextraction and thus has a
higher detection limit than the GRO and DRO methods.
The PHC method also begins peak integration after elution
of the solvent peak for n-pentane.  Thus, this method
probably cannot measure some volatile compounds  (for
example, 2-methyl pentane and MTBE) that are measured
using the GRO method.
1.3.23
Definition of TPH
It is  not possible to establish a definition of TPH that
would  include crude oil and its refined products  and
exclude other  organic compounds.  Ideally, the TPH
definition selected for the demonstration would have

•  Included compounds that are PHCs, such as paraffins,
   naphthenes, and aromatic hydrocarbons

•  Included,  to  the  extent possible, the  major liquid
   petroleum products (gasoline, naphthas, kerosene, jet
   fuels, fuel oils, diesel, and lubricating oils)

-------
 •   Had little inherent bias based on the composition of an
     individual manufacturer's product

 •   Had  little  inherent  bias  based on the  relative
     concentrations of aliphatic and aromatic hydrocarbons
     present

 •   Included much of the volatile portion of gasoline,
     including all weathered gasoline

 •   Included MTBE

 •   Excluded crude oil residuals beyond the extended
     diesel range organic (EDRO) range

 •   Excluded nonpetroleum organic compounds (for
     example,   chlorinated   solvents,  pesticides,
     polychlorinated  biphenyls  [PCB],   and naturally
     occurring oils and greases)

 •   Allowed TPH measurement using a widely accepted
     method

o«   Reflected accepted TPH measurement practice in
     many states

 Several states, including Massachusetts, Alaska, Louisiana,
 and North Carolina, have implemented or are planning to
 implement a TPH contamination cleanup approach based
 on the aliphatic and aromatic hydrocarbon fractions of
 TPH.  The action levels for the aromatic hydrocarbon
 fraction  are more stringent than those for the aliphatic
 hydrocarbon fraction. The approach used in the above-
 mentioned  states   involves   performing   a  sample
 fractionation procedure and two analyses to determine the
 aliphatic and aromatic hydrocarbon concentrations in a
 sample.  However, in most applications of this approach,
 only a few samples are subjected to the dual aliphatic and
 aromatic  hydrocarbon analyses because of the  costs
 associated with performing sample fractionation and two
 analyses.

 For the demonstration, TPH was not defined based on the
 aliphatic and aromatic hydrocarbon  fractions because

     Such a definition is used in only a few states.
•   Variations exist among the sample fractionation and
    analysis procedures used in different states.

•   The   repeatability  and   versatility  of   sample
    fractionation and  analysis procedures are not well
    documented.

•   In some states, TPH-based action levels are still used.

•   The associated analytical costs are high.

As stated in Section 1.3.2.2, analytical methods currently
available for measurement of TPH each exclude some
portion of TPH and are unable to measure  TPH alone
while excluding all other organic compounds, thus making
TPH a method-defined paramete r. After consideration of
all the information presented above, the GRO and DRO
analytical methods were selected for TPH measurement for
the demonstration.  However, because of the general
interest in higher-boiling-point petroleum products, the
integration range  of the DRO method was extended to
include compounds with boiling points up  to 540 °C.
Thus, for the demonstration, the TPH concentration was
the sum of all organic compounds that have boiling points
between 60 and 540 °C and that can be chromatographed,
or the sum of the results obtained using the GRO and DRO
methods.   This approach  accounts for most gasoline,
including  MTBE, and  virtually  all other  petroleum
products and excludes a portion (25 to 50 percent) of the
heavy lubricating  oils.  Thus,  TPH measurement  for the
demonstration included PHCs as well as some organic
compounds that are not PHCs. More specifically, TPH
measurement  did not exclude nonpetroleum  organic
compounds such as chlorinated solvents, other synthetic
organic  chemicals such as pesticides and  PCBs, and
naturally  occurring oils and greases.   A  silica  gel
fractionation step  used  to  remove  polar,  biogenic
compounds such as fatty acids in some GC/FID methods
was not included in the sample preparation step because,
according to the State of California, this step can also
remove some petroleum degradation products that are also
polar in  nature  (California  Environmental  Protection
Agency 1999). The step-by-step approach used to select
the reference  method for the  demonstration and  the
project-specific procedures implemented for soil  sample
preparation and analysis using the reference method are
detailed in Chapter 5.
                                                    10

-------
                                              Chapter 2
           Description of Emulsion Turbidimetry and the PetroFLAG™ System
Measurement of TPH in soil by field measurement devices
generally involves extraction of PHCs from soil using an
appropriate solvent followed by measurement of the TPH
concentration in the extract using an optical method. An
extraction solvent is selected that will not interfere with the
optical measurement of TPH in the extract.   Some field
measurement devices use light in the visible wavelength
range, and others use light outside the visible wavelength
range (for example, infrared and ultraviolet light).

The  optical measurements made by field measurement
devices  may   involve  absorbance,  reflectance,  or
fluorescence. In general, the optical measurement for a
soil extract is compared to a  calibration curve in order to
determine the TPH concentration. Calibration curves may
be developed by (1) using a series of calibration standards
selected based on the type of PHCs being measured at a
site  or (2) establishing a correlation between off-site
laboratory  measurements  and field measurements for
selected, site-specific soil samples.

Field measurement  devices may  be  categorized as
quantitative, semiquantitative, and  qualitative.  These
categories are explained below.

•   A quantitative measurement device measures TPH
   concentrations ranging from its reporting limit through
   its linear range. The measurement result is reported as
   a single, numerical value that has an  established
   precision and accuracy.

•   A semiquantitative measurement device measures
   TPH concentrations above its reporting limit.  The
   measurement result may be reported as a concentration
   range with lower and upper limits.

•   A  qualitative measurement device indicates the
   presence or  absence of PHCs  above or  below a
    specified value (for example, the reporting limit or an
    action level).

The PetroFLAG™ System is a field measurement device
capable of  providing quantitative  TPH  measurement
results.  Measurements made  using the  PetroFLAG™
System are  based on emulsion turbidimetry, which is
described in Section 2.1.  Two-point calibration curves
for the  PetroFLAG™   System may  be  developed
using calibration standards.    Section 2.2.2 discusses
development of calibration curves for the device.

Section 2.1  describes the technology  upon which  the
PetroFLAG™ System is based, Section 2.2 describes  the
PetroFLAG™  System itself, and Section 2.3 provides
Dexsil contact information.  The technology and device
descriptions  presented below are not intended to provide
complete  operating  procedures for  measuring  TPH
concentrations in soil using the PetroFLAG™ System.
Detailed operating procedures for the device, including soil
extraction, TPH  measurement, and TPH concentration
calculation  procedures,   are  available  from  Dexsil.
Supplemental information provided by Dexsil is presented
in the appendix.

2.1    Description of Emulsion Turbidimetry

Measurement of TPH in  soil  using the  PetroFLAG™
System is based on emulsion turbidimetry. Turbidimetry
may be described as measurement of the attenuation, or
loss in  intensity, of a light beam as  the beam passes
through a solution with particles large enough to scatter the
light.  Emulsion turbidimetry involves measurement of the
light scattered by an emulsion (in an emulsion, one liquid
is stably dispersed hi a second, immiscible liquid). The
relationship  between the amount  of light scattered
(turbidity) and the concentration of the emulsion may be
expressed as shown in Equation 2-1.
                                                   11

-------
                      T = kbc                 (2-1)
where
       T =  Turbidity
       k =  Proportionality constant
       b =  Light path length (centimeter)
       c =  Concentration of emulsion (milligram per
             liter [mg/L])
When PHCs in a soil sample are measured using emulsion
turbidimetry, the sample  is  extracted using a  water-
miscible solvent, and the sample extract is added to a vial
containing  an  aqueous,  polar  developer solution.
According to Dexsil, the wate r-miscible solvent extracts all
nonpolar hydrocarbons, and the developer solution acts as
an  emulsifier,  causing  the  aromatic  and aliphatic
hydrocarbons  in solution to precipitate out and form
uniformly  sized micelles.   Micelles are  electrically
charged, colloidal particles composed of aggregates of
large molecules and are  stable for some time. The vial
containing  the  resulting  emulsion  is  placed  in  a
turbidimeter to measure the amount of light scattered by
the emulsion. In the turbidimeter, light at a wavelength of
585 nanometers (nm) is passed through the emulsion, and
the amount of  light  scattered by  the  emulsion at a
90-degree angle is measured. A wavelength of 585 nm is
typically used because the maximum amount of light is
scattered by the  emulsion at this wavelength. The total
concentration of PHCs in the emulsion is then  determined
by comparing the turbidity reading for the emulsion to that
for a reference standard or to a standard calibration curve
prepared using more than one concentration of mineral oil.
According to Dexsil, the TPH concentration thus measured
is  a function of the mean  molecular  weight  of the
hydrocarbons present in the sample.

2.2     Description of the PetroFLAG™ System

The    PetroFLAG™  System,  a  quantitative  device
manufactured by Dexsil, has been commercially available
since January 1995. Dexsil currently holds three patents
on the  device: patent No.  5,756,357; 5,928,950; and
6,117,682.   This  section describes the device  and
summarizes its operating procedure.

2.2.1    Device Description

The PetroFLAG™ System extracts PHCs in soil using a
proprietary organic solvent mixture composed of alcohols,
primarily methanol. The device also uses a proprietary
developer solution that is polar in nature and that acts as
the emulsifying agent.   The  developer solution  also
contains water and surfactants that stabilize the emulsion.

According to Dexsil, the PetroFLAG™ System can be
used  to  measure the  petroleum products  listed in
Table 2-1.   Dexsil reports that  the  device  does not
distinguish between aromatic and aliphatic hydrocarbons
and that it responds to compounds in the C 8 through C36
carbon range. Method detection limits (MDL) claimed by
Dexsil for the device are also listed in Table 2-1 and range
from  10 mg/kg for hydraulic fluid to 1,000 mg/kg for
weathered gasoline.   In addition, Table 2-1  lists the
device's response factors, which are based on mineral oil.
The response factors range from 2 for weathered gasoline
to 10 for transformer oil, indicating that the device is more
sensitive to transformer oil th an weathered gasoline. If no
information   is   available  regarding  the  type  of
contamination in a sample, Dexsil recommends use of an
average response factor of 5.
Table 2-1.  PetroFLAG™ System Method  Detection Limits  and
Response Factors for Petroleum Products
Method Detection Limit
Petroleum Product (milligram per kilogram)
Mineral oil
Transformer oil
Grease
Hydraulic fluid
Transmission fluid
Motor oil
No. 2 fuel oil
No. 6 fuel oil
Diesel
Gear oil
Low-aromatic diesel
Pennsylvania crude oil
Kerosene
Jet A fuel
Weathered gasoline
15
15
15
10
19
19
25
18
13
22
27
20
28
27
1,000
Response Factor
10
10
9
8
8
7
7
6
5
5
4
4
4
4
2
Source:  Dexsil 1997
The PetroFLAG™ System consists of three components:
the (1) PetroFLAG™ System starter kit, (2) PetroFLAG™
Reagents  kit,  and  (3)   PetroFLAG™  High-Range
Extraction  Vials  kit.   The  items  contained in each
                                                    12

-------
component are listed in Table 2-2.  The PetroFLAG™
System starter kit contains both reusable items, including
the PetroFLAG™ Analyzer (turbidimeter), digital balance,
and digital timer, and enough disposable items, including
polypropylene tubes, filter-syringe assemblies, calibration
standards, extraction solvent, and developer solution, to
perform 10 soil  sample analyses.   The PetroFLAG™
Reagents kit can be ordered in multiples of the supply
requirements  for  10  sample analyses.    For  high-
concentration-range samples,  the  PetroFLAG™  High-
Range Extraction Vials kit contains enough proprietary
high-range extraction solvent (proprietary glycol ether) and
pipettor tips  to  perform  10  sample  extractions.  An
Eppendorf™ 1-milliliter (mL) pipettor can be purchased
separately from Dexsil.

The PetroFLAG™ Analyzer emits light at a factory-preset
wavelength of 585 nm.  The analyzer is 5.75 inches long,
3.5 inches wide, and 2 inches high and weighs 0.6 pound.
The power supply for the analyzer is one 9-volt battery,
which  according   to   Dexsil  can  last   for  about
18,000 readings.

The PetroFLAG™ Analyzer allows a user hi the field to
select the response factor that is appropriate for the PHCs
suspected to be present at a site. The response factor can
be changed at any time without affecting the calibration
data stored in the analyzer. The microprocessor  hi the
analyzer uses the calibration data and the response factor
selected by the user to  convert the optical reading into a
TPH concentration in mg/kg.  The user is not required to
perform any calculations unless a soil to solvent mass ratio
other than one is used to measure the TPH concentration or
the sample extract is diluted.  In  such cases, the TPH
concentration reported by the analyzer must be multiplied
by the dilution factor used.

To   assist  the   user   in  obtaining  accurate   TPH
measurements, the PetroFLAG™ Analyzer automatically
performs  several internal  checks and informs the user
when an  error has  occurred.   For example, if a TPH
concentration is outside the linear range of the analyzer
but is still  within  the  quantifiable range, the analyzer
displays a blinking reading.  If the TPH concentration is
outside the quantifiable  range of the analyzer, the analyzer
displays a  blinking "EEEE."  According to Dexsil,
acceptable precision for the linear and quantifiable ranges
is  plus or  minus  (±)  10  and ±20 percent precision,
respectively.
Table 2-2. PetroFLAG™ System Components

PetroFLAG™ System starter kit
•  1 battery-powered, handheld PetroFLAG™ Analyzer (turbidimeter)
   (one 9-volt battery included)
•  1 battery-powered ACCULAB" digital balance (one 100-gram
   calibration weight and two CR2032 cell batteries included)
•  1 battery-powered digital timer (one G-13 button cell battery
   included)
•  12 plastic screw-capped, polypropylene tubes
•  12 filter-syringe assemblies
•  1 breaktop vial containing zero calibration standard (proprietary
   methanol mixture)
*  1 breaktop vial containing calibration standard (1,000 milligrams
   per liter of mineral oil)
•  10 breaktop vials containing extraction solvent (proprietary
   methanol mixture)
•  10 glass vials containing proprietary developer solution
•  User guide
•  Training video and interactive compact disk
•  Material safety data sheet for proprietary methanol mixture
•  Carrying case
PetroFLAG™ Reagents kit
•  12 plastic screw-capped, polypropylene tubes
•  12 filter-syringe assemblies
•  1 breaktop vial containing zero calibration standard (proprietary
   methanol mixture)
•  1 breaktop vial containing calibration standard (1,000 milligrams
   per liter of mineral oil)
•  10 breaktop vials containing extraction solvent (proprietary
   methanol mixture)
•  10 glass vials containing proprietary developer solution
•  Material safety data sheet for proprietary methanol mixture
PetroFLAG™ High-Range Extraction Vials kit
•  10 glass vials containing high-range extraction solvent (proprietary
   glycol ether)
•  10 tips for Eppendorf™ 1-milliliter pipettor
For cases in which a TPH concentration is outside the
linear or quantifiable range of the PetroFLAG™ Analyzer,
the user must select the most appropriate corrective action
from among several options. According to the Dexsil user
guide, the user may (1) decrease the soil to  solvent mass
ratio when the proprietary methanol mixture is used as the
solvent or  (2)  use  the  high-range extraction  solvent
(proprietary glycol ether) to generate 1 mL of extract that
is subsequently diluted in 10 mL of proprietary methanol
mixture. Although it is not mentioned in the user guide,
when the TPH concentration is outside the linear range but
within the quantifiable range, the user may also dilute the
sample extract produced using the proprietary methanol
mixture and reanalyze the  extract, thus eliminating the
re-extraction step and reducing the amount of additional
supplies required to perform the TPH measurement.

The  PetroFLAG™  Analyzer is  also equipped  with a
built-in temperature  sensor that  measures  the ambient
temperature during TPH measurement. The analyzer uses
                                                        13

-------
the  sensor's  temperature  readings  to  correct  for
measurement  fluctuations  caused   by  temperature
variations.   However, the temperature corrections are
valid only for ambient temperatures within 10 °C of the
calibration  temperature.   Therefore, if the  ambient
temperature deviates from the calibration temperature by
more than  10 °C, an error  condition results,  and the
analyzer has to be recalibrated.  According to Dexsil, the
operating temperature range for the analyzer is 4 to 45 °C,
and  the analyzer  does not have an operating humidity
restriction.

The PetroFLAG™ Analyzer also automatically performs
an  optical  system check.    The  analyzer  has  two
independent optical channels. During TPH measurement,
if the analyzer responses on the two channels are different,
the analyzer displays an error message.

Dexsil recommends using its HYDROSCOUT * meter to
measure the moisture content of soil samples in the field
so that the TPH concentrations can be reported on a dry
weight basis.   However, for the demonstration, the
developers were not required to report TPH concentrations
on a dry weight basis; therefore, Dexsil did not use the
HYDROSCOUT® meter.

According to Dexsil, 16 analyses can be performed by one
field technician using the PetroFLAG™ System in about
1 hour. The device is easy to operate and is designed to be
used by those with basic wet chemistry skills.  The
PetroFLAG™ System starter kit includes a user guide that
must be  followed to properly operate  the  device  as
well  as a free training video and interactive compact disk.
In addition,  during regular  business  hours,  Dexsil
provides  technical support over the  telephone  at no
additional cost. Depending on the size of a project, Dexsil
also provides on-site training upon request.

According  to  Dexsil,  the  PetroFLAG™  System  is
innovative  because it responds  to  a broad  range  of
petroleum products regardless of their composition and
extent  of weathering.   Also, the proprietary methanol
solvent mixture used in the device was selected to provide
consistent extraction efficiencies for a range of soil types
and  conditions, including varying levels of moisture
content and ionic strength.
2.2.2   Operating Procedure

The PetroFLAG™ System can be calibrated using known
standards. During the demonstration, Dexsil generated a
two-point calibration curve using an extraction solvent
blank (proprietary methanol mixture) for zero calibration
and a calibration standard (1,000 mg/L of mineral oil).

During the demonstration, extraction of PHCs in a given
soil sample was typically completed by adding 10 mL of
proprietary  methanol  mixture  extraction  solvent  to
10 grams of the sample. If the TPH measurement for a
sample  was outside  the  calibration range  of  the
PetroFLAG™ Analyzer, Dexsil either (1) decreased the
soil to solvent mass  ratio or (2)  used  the PetroFLAG™
high-range extraction solvent (proprietary glycol ether) to
generate 1 mL of extract that was subsequently diluted in
10  mL  of proprietary methanol mixture.   To form an
emulsion, 2  mL of filtered sample extract was  then
transferred to a vial containing 4 mL of developer solution.
The emulsion was  analyzed using the PetroFLAG™
Analyzer at a wavelength of 585 nm.

During  the  demonstration,  Dexsil  generally  used  a
response factor of 2 for samples containing primarily
weathered gasoline and a response factor of 6 for samples
containing primarily diesel.   When no dilutions were
performed, the PetroFLAG™ Analyzer provided a direct
measurement of the TPH concentration in a sample; no
calculations were required. When a soil to solvent mass
ratio  other than one was used  to  measure the TPH
concentration in a sample, the concentration reported by
the analyzer was multiplied by the dilution factor used.

In addition to the internal checks automatically performed
by the PetroFLAG™ Analyzer, Dexsil also analyzed field
duplicates and performed  a temperature check before
analyzing each batch of samples. The temperature check
was performed to verify that the PetroFLAG™ Analyzer
was operating in an ambient temperature within 5 °C of the
calibration temperature. According to Dexsil, although the
analyzer automatically  performs a  temperature check
during each  measurement to verify that the ambient
temperature is within 10 °C of the calibration temperature,
TPH measurements  made using the analyzer are more
accurate when the ambient temperature is within 5 °C of
the calibration temperature.
                                                    14

-------
2.3    Developer Contact Information

Additional information about the PetroFLAG™ System
can be obtained from the following source:

   Dexsil Corporation
   Dr. Ted B. Lynn
   One Hamden Park Drive
   Hamden,CT06517
   Telephone: (203) 288-3509
   Fax: (203) 248-6523
   E-mail: tblynn@dexsil.com
   Internet: www.dexsil.com
                                               15

-------
                                                Chapter 3
                                  Demonstration Site Descriptions
This  chapter  describes  the  three  sites  selected  for
conducting the demonstration.  The first site is referred to
as the Navy BVC site; it is located in Port Hueneme,
California, and contains three sampling areas. The second
site is referred to as the Kelly  AFB site; it is located in
San Antonio, Texas, and contains one sampling area. The
third site is referred to as the PC site; it is located in north-
central Indiana and contains one  sampling area.  After
review of the information available on these and other
candidate sites, the Navy BVC, Kelly AFB, and PC sites
were selected based on the following criteria:

    Site Diversity—Collectively, the three sites contained
    sampling areas with the different soil types and the
    different levels and  types of PHC contamination
    needed to  evaluate the seven field  measurement
    devices selected for the demonstration.

•   Access and Cooperation—The site representatives
    were interested in supporting the demonstration by
    providing site access for collection of soil samples
    required for the demonstration. In addition, the field
    measurement devices were to be demonstrated at the
    Navy BVC site using soil samples from all three sites,
    and the Navy BVC site representatives were willing to
    provide the site support facilities required for  the
    demonstration and to support a visitors' day during the
    demonstration.    As  a testing location  for  the
    Department  of Defense  National  Environmental
    Technology Test Site program, the Navy BVC site is
    used to demonstrate technologies and systems  for
    characterizing or  remediating soil, sediment, and
    groundwater contaminated with fuel hydrocarbons or
    waste oil.

To ensure that  the sampling areas were selected  based
on  current  site characteristics,   a predemonstration
investigation was conducted. During this investigation,
samples were collected from the five candidate  areas
and  were  analyzed  for  GRO  and  EDRO  using
SW-846  Method 8015B (modified) by  the  reference
laboratory, Severn Trent Laboratories in Tampa, Florida
(STL Tampa East). The site descriptions in Sections 3.1
through  3.3  are based   on   data  collected  during
predemonstration investigation sampling activities, data
collected during demonstration  sampling  activities, and
information provided by the site representatives. Physical
characterization of samples was performed in the field by
a geologist during both predemonstration investigation and
demonstration activities.

Some of the predemonstration investigation samples were
also  analyzed by the PetroFLAG™ System developer,
Dexsil, at its facility. Dexsil used reference laboratory
and  PetroFLAG™ System results to prepare for the
demonstration.

Table 3-1  summarizes key site characteristics, including
the contamination type, sampling depth intervals,  TPH
concentration ranges, and soil type hi each sampling area.
The TPH concentration ranges and soil types presented in
Table 3-1  and  throughout this report are  based on
reference  laboratory  TPH results  for  demonstration
samples and soil characterization completed during the
demonstration, respectively. TPH concentration range and
soil type information obtained during the  demonstration
was  generally consistent with the information obtained
during the predemonstration investigation except for the
B-38 Area at Kelly  AFB.  Additional information on
differences between demonstration and predemonstration
investigation activities and  results  is  presented hi
Section 3.2.
                                                     16

-------
Table 3-1. Summary of Site Characteristics
Site
Navy Base
Ventura
County
Kelly Air
Force Base
Petroleum
company
Sampling Area
Fuel Farm Area
Naval Exchange
Service Station
Area
Phytoremediation
Area
B-38 Area
Slop Fill Tank
Area
Contamination Type"
EDRO (weathered diesel
with carbon range from
n-C10 through n-C40)
GRO and EDRO (fairly
weathered gasoline with
carbon range from n-C6
through n-C14)
EDRO (heavy lubricating oil
with carbon range from
n-CH through n-C^.)
GRO and EDRO (fresh
gasoline and diesel or
weathered gasoline and
trace amounts of lubricating
oil with carbon range from
n-C6 through n-Cw)
GRO and EDRO
(combination of slightly
weathered gasoline,
kerosene, JP-5, and diesel
with carbon range from
n-Cs through n-C^)
Approximate
Sampling Depth
Interval
(foot bgs)
Upper layer11
Lower layer6
7 to 8
8 to 9
9 to 10
10 to 11
1.5 to 2.5
23 to 25
25 to 27
2 to 4
4 to 6
6 to 8
8 to 10
TPH Concentration
Range (mg/kg)
44.1 to 93.7
8,090 to 15,000
28.1 to 280
144 to 2,570
61 7 to 3,030
9.56 to 293
1,130 to 2,140
43.8 to 193
41 .5 to 69.4
6. 16 to 3,300
37.1 to 3,960
43.9 to 1,210
52.4 to 554
Type of Soil
Medium-grained sand
Medium-grained sand
Silty sand
Sandy clay or silty sand and gravel
in upper depth interval and clayey
sand and gravel in deeper depth
interval
Silty clay with traces of sand and
gravel in deeper depth intervals
Notes:

bgs   = Below ground surface
mg/kg = Milligram per kilogram

3   The beginning or end point of the carbon range identified as "n-Cx" represents an alkane marker consisting of V carbon atoms on a gas
    chromatogram.
b   Because of soil conditions encountered in the Fuel Farm Area, the sampling depth intervals could not be accurately determined. Sample collection
    was initiated approximately 10 feet bgs, and attempts were made to colled 4-foot-long soil cores. This approach resulted in varying degrees of
    core tube penetration up to 17 feet bgs. At each location in the area, the sample cores were divided into two samples based on visual observations.
    The upper layer of the soil core, which consisted of yellowish-brown, medium-grained sand, made up one sample, and the lower layer of the soil
    core, which consisted of grayish-black, medium-grained sand and smelled of hydrocarbons, made up the second sample.
3.1     Navy Base Ventura County Site

The Navy BVC site in Port Hueneme, California, covers
about 1,600 acres along the south California coast.  Three
areas at the Navy BVC site were selected as sampling areas
for the demonstration: (1) the Fuel  Farm Area (FFA),
(2) the Naval Exchange (NEX) Service Station Area, and
(3) the Phytoremediation Area (PRA).  These areas are
briefly described below.
3.1.1   Fuel Farm Area

The FFA is  a tank farm in the southwest corner of the
Navy BVC site.  The  area contains five tanks and was
constructed to refuel ships and to supply heating fuel for
the Navy BVC site. Tank No. 5114 along the south edge
of the FFA was used to store marine diesel.  After Tank
No. 5114  was deactivated in  1991, corroded pipelines
leading into and out of the tank leaked and contaminated
the surrounding soil with diesel.
                                                         17

-------
The horizontal area of contamination in the FFA was
estimated  to be about 20 feet wide and 90 feet long.
Demonstration samples were collected within several
inches  of the three  predemonstration  investigation
sampling  locations  in  the  FFA using a  Geoprobe®.
Samples were collected at the three locations from east to
west and about 5 feet apart. During the demonstration, soil
in the area was found to generally  consist of medium-
grained sand, and the soil cores contained two distinct
layers.  The upper layer consisted of yellowish-brown,
medium-grained sand with no hydrocarbon odor and TPH
concentrations ranging  from 44.1 to 93.7 mg/kg; the
upper  layer's TPH  concentration range  during the
predemonstration investigation was 38 to 470 mg/kg. The
lower layer consisted of grayish-black, medium-grained
sand  with  a strong  hydrocarbon  odor  and  TPH
concentrations ranging  from 8,090 to  15,000 mg/kg;
the lower  layer's  TPH  concentration range  during the
predemonstration   investigation  was   7,700  to
11,000 mg/kg.

Gas  chromatograms  from  the   predemonstration
investigation and the demonstration showed that FFA soil
samples contained (1) weathered diesel, (2) hydrocarbons
in  the  n-C10  through  n-C28  carbon  range with the
hydrocarbon  hump   maximizing  at  n-C17,  and
(3) hydrocarbons in the n-C 12 through n-C40 carbon range
with the hydrocarbon hump maximizing at n-C20-

3.1.2  Naval Exchange Service Station Area

The NEX Service Station Area lies in the northeast portion
of the Navy BVC  site. About 11,000 gallons of regular
and unleaded gasoline was released from UST  lines in this
area between September 1984 and March 1985. Although
the primary  soil contaminant hi  this area is gasoline,
EDRO is also of concern because (1) another spill north of
the area may have resulted  in a commingled plume of
gasoline and diesel  and  (2) a significant  portion of
weathered  gasoline is associated with EDRO.

The horizontal area of contamination in the NEX Service
Station Area was estimated to be about 450 feet wide and
750 feet long.  During the demonstration, samples were
collected at the  three  predemonstration  investigation
sampling locations in the NEX Service Station Area'from
south to north and about 60 feet apart using a Geoprobe *.
Soil in  the  area  was  found to  generally  consist of
(1)  brownish-black,  medium-grained  sand   in  the
uppermost depth interval and (2) grayish-black, medium-
grained sand in the three deeper depth intervals. Traces of
coarse sand were also present in the deepest depth interval.
Soil samples  collected  from the  area  had a  strong
hydrocarbon odor.  The water  table in the area was
encountered at about 9 feet below ground surface (bgs).
During the demonstration, TPH concentrations ranged
from 28.1 to 280 mg/kg in the  7- to 8-foot bgs depth
interval;  144  to  3,030  mg/kg  in the  8-  to  9- and
9- to 10-foot bgs depth intervals; and 9.56 to 293 mg/kg
in the 10- to  11-foot bgs depth interval.  During the
predemonstration investigation, the TPH concentrations in
the (1) top two depth intervals (7  to 8 and 8 to 9 feet bgs)
ranged from 25 to 65 mg/kg and (2) bottom depth interval
(10 to 11 feet bgs) ranged from 24 to 300 mg/kg.

Gas  chromatograms   from   the  predemonstration
investigation and the demonstration showed that NEX
Service  Station Area soil samples contained (1)  fairly
weathered gasoline  with a high aromatic hydrocarbon
content and (2) hydrocarbons in the n-C6 through n-C14
carbon range. Benzene, toluene, ethylbenzene, and xylene
(BTEX)   analytical  results   for  predemonstration
investigation samples from the 9- to 10-foot bgs depth
interval  at the middle  sampling location revealed a
concentration of 347 mg/kg; BTEX made up 39 percent of
the total GRO and 27 percent of the TPH at this location.
During the predemonstration investigation, BTEX analyses
were conducted at the request of a few developers to
estimate the aromatic hydrocarbon content of the GRO;
such analyses  were not conducted for  demonstration
samples.

3.1.3  Phytoremediation Area

The PRA lies  north of the FFA and west of the  NEX
Service Station Area at the Navy BVC site.  The PRA
consists of soil from a fuel tank removal project conducted
at the Naval Weapons Station in Seal Beach, California.
The area is contained within concrete railings  and  is
60 feet wide,  100 feet long, and about 3 feet deep. It
consists of 12 cells of equal size (20 by 25 feet) that have
three different types of cover: (1) unvegetated cover, (2) a
grass and legume mix, and (3) a native grass mix.  There
are four replicate cells of each cover type.

In the PRA, demonstration samples were collected from
the 1.5- to 2.5-foot bgs depth interval within several inches
of  the  six  predemonstration  investigation sampling
locations using a  split-core sampler.    During the
demonstration, soil at four adj acent sampling locations was
found to generally  consist of  dark  yellowish-brown,
silty sand with  some clay and no  hydrocarbon odor. Soil
                                                   18

-------
at the two remaining adjacent sampling locations primarily
consisted of dark yellowish-brown, clayey sand with no
hydrocarbon odor, indicating the absence of volatile PHCs.
The TPH concentrations in the demonstration samples
ranged from 1,130 to 2,140 mg/kg; the TPH concentrations
in the predemonstration investigation samples ranged from
1,500 to 2,700 mg/kg.

Gas  chromatograms   from  the  predemonstration
investigation and the demonstration showed that PRA soil
samples  contained  (1) heavy  lubricating  oil  and
(2) hydrocarbons in the n-C 14 through n-C4(H carbon range
with the hydrocarbon hump maximizing at n-C 32.

Dexsil chose not to demonstrate the PetroFLAG™ System
using the soil samples collected in the PRA because Dexsil
believed that the natural organic material in the area would
adversely impact the device's performance.

3.2    Kelly Air Force Base Site

The Kelly AFB site covers approximately 4,660 acres and
is about 7 miles from the center of San Antonio, Texas.
One area at Kelly AFB, the B-38 Area,  was selected as a
sampling area for the demonstration. The B-38 Area lies
along the east boundary  of Kelly AFB  and is part of an
active UST farm that serves the government vehicle
refueling station  at  the base.   In December  1992,
subsurface soil contamination resulting from leaking diesel
and gasoline USTs and associated piping was discovered
in this area during UST removal and upgrading activities.

The B-38 Area was estimated to be about 150 square feet
in size.  Based on discussions with site representatives,
predemonstration investigation samples were collected in
the 13- to 17- and 29- to 30-foot bgs depth intervals at four
locations in the area using a Geoprobe®.   Based on
historical  information,  the  water table in  the  area
fluctuates  between  16 and 24 feet bgs.  During  the
predemonstration investigation, soil in the area was found
to generally consist of (1) clayey silt in the upper depth
interval above the water  table with a TPH concentration
of 9 mg/kg and (2) sandy clay with significant gravel in the
deeper depth   interval  below  the water  table with
TPH concentrations ranging from 9 to 18 mg/kg.  Gas
chromatograms from  the predemonstration investigation
showed that B-38 Area soil samples contained (1) heavy
lubricating oil and (2) hydrocarbons hi the n-C 24 through
n-C30 carbon range.
Based on the low TPH  concentrations and the type of
contamination  detected  during  the predemonstration
investigation   as   well  as   discussions  with   site
representatives   who  indicated  that   most  of  the
contamination in the B-3 8 Area can be found at or near the
water table, demonstration samples were collected near the
water table. During the demonstration, the water table was
24 feet bgs.  Therefore, the demonstration samples were
collected in the 23- to 25- and 25- to 27-foot bgs depth
intervals at three locations in the  B-38  Area using a
Geoprobe®.  Air Force activities in the area during the
demonstration prevented the sampling team from accessing
the fourth location sampled during the predemonstration
investigation.

During the demonstration, soil in the area was found to
generally consist of (1) sandy clay or silty sand and gravel
hi the upper  depth interval with a  TPH concentration
between 43.8 and 193 mg/kg and (2) clayey sand and
gravel  in the  deeper  depth  interval with   TPH
concentrations between 41.5 and 69.4 mg/kg. Soil samples
collected in the area had little  or no hydrocarbon odor.
Gas chromatograms from the demonstration showed that
B-3 8 Area soil samples contained either (1) fresh gasoline,
diesel, and hydrocarbons in the n-C 6 through n-C25 carbon
range with the hydrocarbon hump maximizing at n-C 17;
(2) weathered gasoline with trace amounts of lubricating
oil and hydrocarbons in the n-C6 through n-C30 carbon
range  with   a   hydrocarbon   hump  representing  the
lubricating oil between n-C2o and n-C30; or (3) weathered
gasoline  with   trace   amounts  of lubricating   oil
and hydrocarbons in the n-C6 through n-C40 carbon range
with a hydrocarbon hump representing the lubricating oil
maximizing at n-C31.

3.3     Petroleum Company Site

One area at the PC site in north-central Indiana, the  Slop
Fill Tank (SFT)  Area, was selected as a sampling area for
the demonstration.  The SFT Area lies in the west-central
portion of the PC site  and is part of an active fuel tank
farm.  Although the primary soil contaminant hi this area
is gasoline, EDRO is also of concern because of a heating
oil release that occurred north of the area.

The SFT Area was estimated to be 20 feet long and 20 feet
wide.  In this area, demonstration samples were collected
from 2 to 10 feet bgs at 2-foot depth intervals  within
several inches of the five predemonstration investigation
                                                    19

-------
sampling locations using a  Geoprobe®.  Four  of the
sampling locations were spaced 15 feet apart to form the
corners of a square, and the fifth sampling location was at
the center of the square. During the demonstration, soil in
the area  was found to generally  consist of brown to
brownish-gray, silty clay with traces of sand and gravel in
the deeper depth intervals. Demonstration soil samples
collected in the area had little or no hydrocarbon odor.
During the demonstration, soil in the three upper depth
intervals  had TPH concentrations ranging from 6.16 to
3,960 mg/kg, and soil  in the  deepest depth interval had
TPH concentrations ranging  from 52.4 to 554 mg/kg.
During the  predemonstration investigation, soils in the
three upper depth intervals and the deepest depth interval
had TPH concentrations ranging from 27 to 1,300 mg/kg
and from 49 to 260 mg/kg, respectively.
Gas   chromatograms  from  the  predemonstration
investigation and the demonstration showed that SFT Area
soil samples contained (1) slightly weathered gasoline,
kerosene, JP-5, and diesel and (2) hydrocarbons in the
n-C5 through n-C20 carbon range. There was also evidence
of  an  unidentified  petroleum  product  containing
hydrocarbons in the n-C24 through n-C32 carbon range.
BTEX  analytical  results   for   predemonstration
investigation  samples from the deepest depth  interval
revealed concentrations of 26, 197, and 67 mg/kg at the
northwest,  center,  and southwest  sampling locations,
respectively.  At the northwest location, BTEX made up
13 percent of the total GRO and 5 percent of the TPH. At
the center location, BTEX made up 16 percent of the total
GRO and 7 percent of the TPH. At the southwest location,
BTEX made  up  23  percent of the  total GRO and
18 percent of  the TPH.   BTEX analyses were  not
conducted for demonstration samples.
                                                   20

-------
                                              Chapter 4
                                     Demonstration Approach
This chapter presents the objectives (Section 4.1), design
(Section 4.2), and sample preparation and management
procedures (Section 4.3) for the demonstration.

4.1    Demonstration Objectives

The primary goal of the SITE MMT Program is to develop
reliable performance and cost data on innovative, field-
ready technologies. A SITE demonstration must provide
detailed and reliable performance and cost data  so that
potential technology users have adequate information to
make  sound  judgments   regarding   an  innovative
technology's applicability to a specific site and to compare
the technology to conventional technologies.

The demonstration  had both primary  and secondary
objectives.   Primary  objectives were  critical  to the
technology evaluation and required the use of quantitative
results to  draw conclusions regarding  a technology's
performance.     Secondary  objectives  pertained   to
information that was useful but did not necessarily require
the use of  quantitative results  to  draw  conclusions
regarding a technology's  performance.

The primary  objectives  for the  demonstration  of the
individual field measurement devices were as follows:

P1. Determine the MDL

P2. Evaluate  the accuracy  and precision  of TPH
    measurement for  a  variety  of  contaminated  soil
    samples

P3. Evaluate  the  effect  of  interferents  on   TPH
    measurement

P4. Evaluate the effect of soil moisture content on TPH
    measurement
P5. Measure the time required for TPH measurement

P6. Estimate costs associated with TPH measurement

The secondary  objectives for the demonstration of the
individual field measurement devices were as follows:

SI. Document the skills and training required to properly
    operate the device

S2. Document health and safety concerns associated with
    operating the device

S3. Document the portability of the device

S4. Evaluate  the durability  of the device based on its
    materials of construction and engineering design

S5. Document the availability of the device and associated
    spare parts

The objectives forthe demonstration were developed based
on input from MMT Program stakeholders, general user
expectations of field measurement devices, characteristics
of the demonstration areas, the time available to complete
the  demonstration, and  device capabilities  that  the
developers intended to highlight.

4.2    Demonstration Design

A predemonstration sampling and analysis investigation
was conducted to assess existing conditions and confirm
available  information  on  physical   and  chemical
characteristics of soil in each demonstration area.  Based
on information from the predemonstration investigation as
well as available historical data, a demonstration design
was developed to address the demonstration objectives.
Input regarding  the demonstration design was  obtained
                                                   21

-------
from  the   developers   and  demonstration   site
representatives. The demonstration design is summarized
below.

The demonstration involved analysis of soil environmental
samples, soil performance evaluation  (PE) samples, and
liquid PE samples.  The environmental samples  were
collected  from three  contaminated  sites, and the PE
samples  were obtained  from  a  commercial  provider,
Environmental Resource  Associates (ERA)  in Arvada,
Colorado. Collectively, the environmental and PE samples
provided the different matrix types and the different levels
and types of PHC contamination needed to perform a
comprehensive demonstration.

The environmental samples  were soil core  samples
collected from the demonstration areas at the Navy BVC,
Kelly AFB, and PC sites described in Chapter 3. The soil
core samples collected at the Kelly AFB and PC sites were
shipped to the Navy BVC site 5 days prior to the start of
the field  analysis activities.   Each  soil core sample
collected from a specific depth interval at a particular
sampling location in a given area was homogenized and
placed in individual sample containers. Soil samples were
then provided to the developers and reference laboratory.
In addition, the PE samples were obtained from ERA for
distribution to the developers and reference laboratory.
Field analysis of all environmental and PE samples was
conducted near the PRA at the Navy BVC site.

The field measurement  devices  were evaluated based
primarily  on how they  compared with the reference
method selected for the demonstration. PE samples  were
used to  verify that reference method performance was
acceptable. However, for the comparison with the device
results, the reference method results were not adjusted
based on the recoveries observed during analysis of the PE
samples.

The sample collection and homogenization procedures
may have resulted in GRO losses of up to one order of
magnitude in environmental samples.  Despite  any such
losses, the homogenized samples were expected to contain
sufficient levels of GRO to allow demonstration objectives
to be achieved.  Moreover, the  environmental sample
collection and homogenization procedures implemented
during the demonstration ensured that the developers and
reference laboratory received the same sample material for
analysis,  which  was required  to  allow meaningful
comparisons of field measurement device and reference
method results.

To facilitate effective use of available information on both
the  environmental  and   PE  samples  during   the
demonstration, the developers and reference laboratory
were  informed  of (1)  whether each  sample was  an
environmental or PE sample, (2)  the area where each
environmental  sample   was  collected,  and  (3)  the
contamination type and  concentration range of each
sample. This information was included in each sample
identification number.  Each sample  was identified as
having a low (less than 100 mg/kg), medium (100 to
1,000 mg/kg), or high (greater than 1,000 mg/kg) TPH
concentration range. The concentration ranges were based
primarily on predemonstration investigation results or the
amount of weathered gasoline or diesel added during PE
sample preparation. The concentration ranges were meant
to be used only as a guide by the developers and reference
laboratory.  The gasoline used for PE sample preparation
was 50 percent weathered; the weathering was achieved by
bubbling nitrogen gas into a known volume of gasoline
until the volume was reduced by 50 percent. Some PE
samples also contained interferents specifically added to
evaluate the effect of interferents on TPH measurement.
The   type   of   contamination  and   expected  TPH
concentration ranges were identified; however, the specific
compounds used as interferents were not identified.  All
PE samples were prepared in triplicate as separate, blind
samples.

During  the demonstration,  Dexsil field  technician's
operated   the   PetroFLAG™  System,   and   EPA
representatives made observations to evaluate the device.
All the developers were given the opportunity to choose
not to analyze samples collected in a particular area or a
particular class of samples, depending on the intended uses
of their devices.  Dexsil chose not to analyze soil samples
collected in the PRA because Dexsil believed that  the
natural organic material in the area would adversely impact
the PetroFLAG™ System's ability to accurately measure
TPH.  In addition, Dexsil chose not to analyze low- and
medium-concentration-range weathered gasoline soil PE
samples because according to Dexsil, the PetroFLAG™
System  was not sensitive  to  weathered  gasoline
concentrations of less than 1,000 mg/kg.

Details of the approach used to address the primary and
secondary objectives for the demonstration are presented
in Sections 4.2.1 and 4.2.2, respectively.
                                                    22

-------
4.2.1   Approach for Addressing Primary Objectives

This section presents the approach used to address each
primary objective.

Primary Objective PI: Method Detection Limit

To determine the MDL for each field measurement device,
low-concentration-range soil PE  samples  containing
weathered gasoline or diesel were to be analyzed.  The
low-range PE samples  were prepared using methanol,
which facilitated preparation ofhomogenous samples. The
target concentrations of the PE samples were set to meet
the following criteria: (1) at the  minimum acceptable
recoveries set by ERA, the samples contained measurable
TPH concentrations, and (2) when feasible, the sample
TPH  concentrations  were  generally  between  1  and
10 times the MDLs claimed by the developers and the
reference  laboratory, as recommended by 40  Code  of
Federal Regulations (CFR)  Part   136,  Appendix B,
Revision  1.1.1.   Dexsil and the reference laboratory
analyzed seven diesel PE samples to statistically determine
the MDLs for EDRO soil samples.  Dexsil chose not to
analyze the low-range weathered  gasoline PE samples
because according to Dexsil, the PetroFLAG™ System
was not sensitive to weathered gasoline concentrations of
less than 1,000 mg/kg. The reference laboratory analyzed
seven low-range weathered gasoline PE samples; however,
during  the  preparation  of these samples,  significant
volatilization ofPHCs occurred because of the matrix used
for preparing the  samples.  Because  of the problems
associated  with preparation  of low-range  weathered
gasoline PE samples, the results for these samples could
not be  used to determine the MDL for the reference
method.

Primary Objective P2: Accuracy and Precision

To estimate  the accuracy and precision of each field
measurement device, both environmental and PE samples
were analyzed. The evaluation of analytical accuracy was
based on the assumption that a field measurement device
may  be used  to  (1)  determine  whether the  TPH
concentration in a given area exceeds an action level or
(2) perform a preliminary characterization of soil in a
given area.  To evaluate whether the TPH concentration in
a soil sample exceeded  an action level, the developers and
reference laboratory were asked to determine whether TPH
concentrations in a given area or PE sample type exceeded
the action levels listed in Table 4-1.  The action levels
chosen  for environmental samples were based on  the
predemonstration investigation analytical results and state
action levels.   The action levels  chosen  for the  PE
samples were based in part on the ERA acceptance limits
Table 4-1. Action Levels Used to Evaluate Analytical Accuracy
Site
Navy Base Ventura
County
Kelly Air Force Base
Petroleum company
Area
Fuel Farm Area
Naval Exchange Service Station Area
Phytoremediation Area"
B-38 Area
Slop Fill Tank Area
Performance evaluation samples (GRO analysis)
Performance evaluation samples (EDRO analysis)
Typical TPH Concentration Range3
Low and high
Low to high
High
Low
Medium
Medium
High
Low
Medium
High
Action Level (mg/kg)
100
50
1,500
100
500
200
2,000
15
200
2,000
Notes:

mg/kg =  Milligram per kilogram

'    The typical TPH concentration ranges shown cover all the depth intervals in each area. Table 4-2 shows the depth intervals that were sampled
    in each area and the typical TPH concentration range for each depth interval. The action level for each area was used as the basis for evaluating
    sample analytical results regardless of the typical TPH concentration ranges for the various depth intervals.
b    Dexsil chose not to analyze soil samples collected in the Phytoremediation Area because Dexsil believed that the natural organic material in the
    area would adversely impact the PetroFLAG™ System's ability to accurately measure TPH.
                                                     23

-------
for PE samples; therefore, each PE sample was expected
to have  at least the TPH concentration  indicated in
Table 4-1.  As stated above, Dexsil did not analyze low-
and medium-concentration-range weathered gasoline soil
PE samples, so  primary  objective P2  could not  be
addressed for the PetroFLAG™ System based on these
samples. In addition, because of the problems associated
with preparation of the low-range weathered gasoline PE
samples, the results for these samples could not be used to
address primary objective P2 for the reference method.

Neat (liquid) samples of weathered gasoline and  diesel
were  also  analyzed by  the developers  and reference
laboratory to evaluate accuracy and precision. Because
extraction of the neat samples was not  necessary, the
results for these samples provided accuracy and precision
information strictly associated with the analyses and were
not affected by extraction procedures.

Sample  TPH   results   obtained  using  each   field
measurement device  and  the  reference method  were
compared to  the action levels presented in Table 4-1 in
order to determine whether sample TPH  concentrations
were above the action levels. The results obtained using
the device  and reference method were compared to
determine how many times the device's results agreed with
those of the  reference method for a particular area or
sample type.  In addition,  the ratio of the TPH results of a
given device to the TPH results of the reference method
was calculated. The ratio was used to develop a frequency
distribution in order to determine how many of the device
and reference method results were within 30 percent,
within 50 percent, and outside the 50 percent window.

To complete a preliminary characterization of soil in a
given area using a field measurement device, the user may
have to demonstrate to a regulatory agency that (1) no
statistically  significant  difference  exists between the
results of the laboratory method selected for the project
(the reference method) and the device results, indicating
that  the  device  may be  used as  a substitute for the
laboratory method, or (2) a consistent correlation exists
between  the device  and laboratory  method results,
indicating that the device results can be adjusted using the
established correlation.

To evaluate whether a statistically  significant difference
existed between a given field measurement device and the
reference method results, a two-tailed, paired Student's
t-test was performed. To  determine whether a consistent
correlation  existed between the TPH results of a  given
field measurement device and the reference method, a
linear regression was performed to estimate the square of
the correlation coefficient (R2), the slope, and the intercept
of each regression equation. Separate regression equations
were developed for each demonstration area and for the PE
samples that did not contain interferents. The reliability of
the regression equations was tested using the F-test; the
regression equation probability derived from the F-test was
used to evaluate whether the correlation between the TPH
results of the device and the reference method occurred
merely by chance.

To evaluate analytical precision, one set of blind field
triplicate environmental samples was collected from each
depth interval at one location in each demonstration area
except the B-38 Area, where site conditions allowed
collection of triplicates in the top depth interval only.
Blind triplicate low-, medium-, and high-concentration-
range PE samples were also used to  evaluate analytical
precision because TPH concentrations in environmental
samples  collected during  the demonstration sometimes
differed from the analytical results  for predemonstration
investigation samples.  The low- and medium-range PE
samples were prepared using methanol as a carrier, which
facilitated preparation of homogenous samples.

Additional information regarding  analytical precision was
collected by  having  the developers  and  reference
laboratory analyze extract  duplicates. Extract duplicates
were  prepared  by extracting  a  soil sample  once  and
collecting two aliquots of the extract.  For environmental
samples, one sample  from  each  depth  interval was
designated as an extract duplicate.  Each sample designated
as an extract duplicate was co llected from a location where
field triplicates were collected. To evaluate a given field
measurement device's ability to precisely measure TPH,
the relative standard deviation (RSD) of the device and
reference method TPH results for triplicate samples was
calculated. In addition, to evaluate the  analytical precision
of the device and reference method, the relative percent
difference (RPD) was calculated using the TPH results for
extract duplicates.

Primary Objective P3: Effect of Interferents

To  evaluate  the effect of interferents on each  field
measurement device's ability to accurately measure TPH,
high-concentration-range  soil PE  samples containing
weathered gasoline or diesel with or without an interferent
were analyzed.  As explained in Chapter 1, the definition
of TPH is quite variable. For the purposes of addressing
                                                     24

-------
primary objective P3, the term "interferent" is used in a
broad sense and is applied to both PHC and non-PHC
compounds. The six different interferents evaluated during
the demonstration were MTBE; tetrachloroethene (PCE);
Stoddard solvent; turpentine (an alpha and beta pinene
mixture); 1,2,4-trichIorobenzene; and humic  acid.  The
boiling points and vapor pressures of (1) MTBE and PCE
are similar to those of GRO; (2) Stoddard solvent and
turpentine are similar to  those of GRO and EDRO; and
(3) 1,2,4-trichlorobenzene and humic acid are similar to
those of EDRO.  The solubility, availability, and cost of
the interferents  were also considered during  interferent
selection.   Specific reasons for the selection of the six
interferents are presented below.

•   MTBE is an oxygenated gasoline additive that  is
    detected  in  the  GRO   analysis   during  TPH
    measurement using a GC.

•   PCE is not a petroleum product but is detected in the
    GRO analysis during TPH measurement using a GC.
    PCE may also be viewed as a typical halogenated
    solvent that may be  present in some environmental
    samples.

•   Stoddard solvent is an aliphatic naphtha  compound
    with a carbon range of n-C 8 through n-C 14 and is partly
    detected in both the GRO and EDRO analyses during
    TPH measurement using a GC.

•   Turpentine is not a petroleum product but has a carbon
    range of n-C9 through n-C15 and is partly detected in
    both the GRO  and  EDRO analyses during  TPH
    measurement using a GC.  Turpentine may also be
    viewed as a substance  that behaves similarly to  a
    typical naturally occurring oil or grease during TPH
    measurement using a GC.

•   The  compound  1,2,4-trichlorobenzene  is not  a
    petroleum  product but is  detected  in the EDRO
    analysis.  This compound may also be viewed as a
    typical  halogenated semivolatile organic  compound
    that behaves similarly to a  chlorinated pesticide or
    PCB during TPH measurement using a GC.

•   Humic  acid  is  a  hydrocarbon  mixture  that  is
    representative of naturally occurring organic carbon in
    soil and was suspected to be detected during EDRO
    analysis.
Based  on  the principles  of operation  of  the  field
measurement devices, several of the interferents  were
suspected to be detected by the devices.

The PE samples containing MTBE and PCE were not
prepared with diesel  and the PE samples  containing
1,2,4-trichlorobenzene and humic acid were not prepared
with weathered gasoline because these interferents were
not expected to impact the analyses and because practical
difficulties such as solubility constraints were associated
with preparation of such samples.

Appropriate control  samples  were  also  prepared and
analyzed to address primary objective P3. These samples
included processed garden soil, processed garden soil and
weathered gasoline, processed garden soil and diesel, and
processed garden soil and humic acid samples. Because of
solubility constraints, control samples containing MTBE;
PCE;  Stoddard solvent;  turpentine; or 1,2,4-trichloro-
benzene  could not be prepared.   Instead, neat (liquid)
samples of these interferents were prepared and used as
quasi-control samples to evaluate the effect of each
interferent on the field measurement device and reference
method results. Each PE sample was prepared in triplicate
and submitted to the developers and reference laboratory
as blind triplicate samples.

To evaluate the effects of interferents on a given field
measurement device's ability to accurately measure TPH
under primary objective P3,  the means  and standard
deviations of the TPH results for triplicate PE samples
were calculated. The mean for each group of samples was
qualitatively evaluated to determine whether the data
showed any trend—that  is, whether an increase in the
interferent concentration resulted in an increase or decrease
in the measured TPH concentration. A one-way analysis
of variance was performed to determine whether the group
means were the same or different.

Primary Objective P4: Effect of Soil Moisture
Content

To evaluate the effect of soil moisture content, high-
concentration-range soilPE samples containing weathered
gasoline or diesel were analyzed.  PE samples containing
weathered gasoline were prepared at two moisture levels:
9 percent moisture and 16 percent moisture. PE samples
containing diesel  were also prepared at  two moisture
levels:  negligible moisture (less than  1  percent) and
9 percent moisture. All the moisture levels were selected
                                                    25

-------
based  on  the  constraints  associated  with  sample
preparation. For example, 9 percent moisture represents
the minimum moisture level for containerizing samples in
EnCores and 16 percent moisture represents the saturation
level of the soil used to prepare PE samples.  Diesel
samples  with negligible moisture could be prepared
because they did not require EnCores for containerization;
based on vapor pressure data for diesel and weathered
gasoline, 4-ounce jars were  considered to be appropriate
for containerizing diesel samples but not for containerizing
weathered  gasoline  samples.   Each PE sample was
prepared in triplicate.

To measure the effect of soil moisture content on a given
field measurement device's ability to accurately measure
TPH under primary objective P4, the means and standard
deviations of the TPH results for triplicate PE samples
containing weathered gasoline and diesel at two moisture
levels were calculated. A two-tailed, two-sample Student's
t-test was performed to determine whether the device and
reference method results were impacted by moisture—that
is, to determine whether an increase in moisture resulted in
an increase or  decrease  in  the TPH concentrations
measured.

Primary Objective P5: Time Required for TPH
Measurement

The sample throughput (the number of TPH measurements
per  unit  of time)  was  determined  for  each field
measurement device by measuring the time required for
each activity associated with TPH measurement, including
device setup, sample extraction, sample analysis, and data
package preparation.  The EPA provided each developer
with investigative  samples stored in  coolers.   The
developer unpacked the coolers and checked the chain-of-
custody forms to verify  that it had received the correct
samples.  Time measurement began when the developer
began to set up its device.  The total time required to
complete  analysis  of all  investigative  samples was
recorded.  Analysis was  considered to be complete and
time measurement stopped when the developer provided
the EPA with a summary table of results, a run log, and
any supplementary information that the developer chose.
The summary table listed all samples analyzed and their
respective TPH concentrations.

For the reference laboratory, the total analytical time began
to be measured  when the  laboratory received all  the
investigative samples, and tune measurement continued
until the EPA representatives received a complete data
package from the laboratory.

Primary Objective P6: Costs Associated with TPH
Measurement

To estimate the costs associated with TPH measurement
for each field measurement device, the following five cost
categories were identified: capital equipment, supplies,
support equipment, labor, and investigation-derived waste
(IDW) disposal.  Chapter 8 of this ITVR discusses the
costs estimated for the PetroFLAG™ System based on
these cost categories.

Table 4-2 summarizes the demonstration approach used to
address the primary objectives and includes demonstration
area characteristics, approximate sampling depth intervals,
and the rationale  for the analyses performed by  the
reference laboratory.

4.2.2  Approach for Addressing Secondary
       Objectives

Secondary  objectives were  addressed  based on field
observations made during the demonstration. Specifically,
EPA representatives observe d TPH measurement activities
and documented them in a field logbook. Each developer
was given the opportunity to review the field logbook at
the end of each day of the demonstration. The approach
used to address each  secondary objective for each field
measurement device is discussed below.

•   The skills and training required for proper device
    operation (secondary objective SI) were evaluated by
    observing and noting the skills required to operate the
    device  and prepare the data package during  the
    demonstration  and by discussing  necessary  user
    training with developer personnel.

•   Health  and safety concerns associated with device
    operation (secondary objective S2) were evaluated by
    observing  and noting possible health and safety
    concerns during the demonstration, such as the types
    of  hazardous  substances   handled  by  developer
    personnel during  analysis, the number of times that
    hazardous  substances  were  transferred  from one
    container to another during the analytical procedure,
    and direct exposure  of  developer  personnel  to
    hazardous substances.
                                                    26

-------
        Table 4-2. Demonstration Approach
Site
Navy
BVC
Kelly
AFB
PC
Area
FFA
NEX
Service
Station
Area
PRA"
B-38
Area
SFT
Area
Approximate
Sampling Depth
Interval (foot bgs)
Upper layer0
Lower layer"
7 to 8
8 to 9
9 to 10
10 to 11
1.5 to 2.5
23 to 25
25 to 27
2 to 4
4 to 6
6 to 8
8 to 10
Sample Matrix
Ottawa sand
(PE sample)
Processed garden soil (PE sample)
Objective
Addressed*
P2
Objective
Addressed"
P1, P2
P2
Soil Characteristics
Medium-grained sand
Medium-grained sand
Silty sand
Sandy clay or silty sand and
gravel in upper depth interval and
clayey sand and gravel in deeper
depth interval
Silty day with traces of sand in
deeper depth intervals
Soil Characteristics
Fine-grained sand
Silty sand
Contamination Type
Weathered diesel with carbon range from
n-C,0 through n-C41)
Fairly weathered gasoline with carbon range
from n-C6 through n-C,4
Heavy lubricating oil with carbon range from
n-C,4 through n-C40
Fresh gasoline and diesel or weathered
gasoline and trace amounts of lubricating oil
with carbon range from n-C6 through n-C40
Combination of slightly weathered gasoline,
kerosene, JP-5, and diesel with carbon range
from n-C5 through n-CK
Contamination Type
Weathered gasoline"1'
Diesel
Weathered gasoline"
Diesel
Typical TPH
Concentration
Range"
Low
High
Low to
medium
Medium to
high
High
Low
High
Low
Medium
Typical TPH
Concentration
range"
Low
Medium and
high
Rationale for Analyses
by Reference Laboratory
Only EDRO because samples did not
contain PHCs in gasoline range
GRO and EDRO because samples
contained PHCs in both gasoline and
diesel ranges
Only EDRO because samples did not
contain PHCs in gasoline range
GRO and EDRO because samples
contained PHCs in both gasoline and
diesel ranges
Rationale for Analyses
by Reference Laboratory
GRO and EDRO because weathered
gasoline contains significant amounts of
PHCs in both gasoline and diesel
ranges
Only EDRO because diesel does not
contain PHCs in gasoline range
GRO and EDRO because weathered
gasoline contains significant amounts of
PHCs in both gasoline and diesel
ranges
Only EDRO because diesel does not
contain PHCs in gasoline range
to
-0

-------
        Table 4-2. Demonstration Approach (Continued)
Sample Matrix
Not applicable (neat liquid PE
sample)
Processed garden soil (PE sample)
Objective
Addressed*
P2
(Continued)
P3
Soil Characteristics
Not applicable
Silly sand
Contamination Type
Weathered gasoline
Diesel
Blank soil (control sample)
Weathered gasoline
Weathered gasoline and MTBE (1,100 mg/kg),
PCE (2,810 mg/kg), Stoddard solvent
(2,900 mg/kg), or turpentine (2,730 mg/kg)
Weathered gasoline and MTBE (1 ,700 mg/kg),
PCE (13,100 mg/kg), Stoddard solvent
(15,400 mg/kg), or turpentine (12,900 mg/kg)
Diesel
Diesel and Stoddard solvent (3,650 mg/kg) or
turpentine (3,850 mg/kg)
Diesel and Stoddard solvent (18,200 mg/kg)
or turpentine (19,600 mg/kg)
Diesel and 1 ,2,4-trichlorobenzene
(3,350 mg/kg) or humic acid (3,940 mg/kg)
Diesel and 1 ,2,4-trichlorobenzene
(16,600 mg/kg) or humic acid (19,500 mg/kg)
Humic acid (3,940 mg/kg)
Humic acid (19,500 mg/kg)
Typical TPH
Concentration
range"
High
High
Trace
High
Trace
Rationale for Analyses
by Reference Laboratory
GRO and EDRO because weathered
gasoline contains significant amounts of
PHCs in both gasoline and diesel
ranges
Only EDRO because diesel does not
contain PHCs in gasoline range
GRO and EDRO because processed
garden soil may contain trace
concentrations of PHCs in both gasoline
and diesel ranges
GRO and EDRO because weathered
gasoline contains significant amounts of
PHCs in both gasoline and diesel
ranges
Only EDRO because diesel does not
contain PHCs in gasoline range
GRO and EDRO because (1) Stoddard
solvent contains PHCs in both gasoline
and diesel ranges and (2) turpentine
interferes with both analyses
Only EDRO because 1,2,4-trichloro-
benzene and humic acid do not interfere
with GRO analysis
Only EDRO because humic acid does
not interfere with GRO analysis
The contribution of trace concentrations
(less than 15 mg/kg) GRO found in
processed garden soil during the
predemonstration investigation was
considered to be insignificant evaluation
of the effect of humic acid interference,
which occurs in the diesel range.
K)
CO

-------
Table 4-2. Demonstration Approach (Continued)
Sample Matrix
Not applicable (neat liquid PE
sample)
Processed garden soil (PE sample)
Objective
Addressed'
P3
(Continued)
P4
Soil Characteristics
Not applicable
Silty sand
Contamination Type
Weathered gasoline
Diesel
MTBE
PCE
Stoddard solvent
Turpentine
1 ,2,4-Trichlorobenzene
Weathered gasoline (samples prepared at
9 and 16 percent moisture levels)
Diesel (samples prepared at negligible [less
than 1 percent] and 9 percent moisture levels)
Typical TPH
Concentration
range"
High
Not
applicable
High
Not
applicable
High
Rationale for Analyses
by Reference Laboratory
GRO and EDRO because weathered
gasoline contains significant amounts of
PHCs in both gasoline and diesel
ranges
Only EDRO because diesel does not
contain PHCs in gasoline range
Only GRO because MTBE and PCE do
not interfere with EDRO analysis
GRO and EDRO because Stoddard
solvent contains PHCs in both gasoline
and diesel ranges
GRO and EDRO because turpentine
interferes with both analyses
Only EDRO because 1 ,2,4-trichloro-
benzene does not interfere with GRO
analysis
GRO and EDRO because weathered
gasoline contains significant amounts of
PHCs in both gasoline and diesel
ranges
Only EDRO because diesel does not
contain PHCs in gasoline range
Notes:

AFB   =  Air Force Base
bgs    =  Below ground surface
BVC   =  Base Ventura County
FFA   =  Fuel Farm Area
mg/kg  =  Milligram per kilogram
MTBE  =  Methyl-tert-butyl ether
NEX  =  Naval Exchange
PC    =  Petroleum company
PCE  =  Tetrachloroethene
PE    =  Performance evaluation
PHC  =  Petroleum hydrocarbon
PRA  =  Phytoremediation Area
SFT  =  Slop Fill Tank
     Field observations of all sample analyses conducted during the demonstration were used to address primary objectives PS and P6 and the secondary objectives.
     The typical TPH concentration range was based on reference laboratory results for the demonstration. The typical low, medium, and  high ranges indicate TPH concentrations of less than
     100 mg/kg; 100 to 1,000 mg/kg; and greater than 1,000 mg/kg, respectively.

     Because of soil conditions encountered in the FFA during the demonstration, the sampling depth intervals could not be accurately determined. Sample collection was initiated approximately
     10 feet bgs, and attempts were made to collect 4-foot-long soil cores.  For each sampling location in the area, the sample cores were divided into two samples based on visual observations.
     The upper layer of the soil core made up one sample, and the lower layer of the soil core made up the second sample.
     Dexsil chose not to analyze soil samples collected in the PRA because Dexsil believed that the natural organic material in the area would adversely impact the PetroFLAG™ System's ability
     to accurately measure TPH.

-------
OJ
o
        Table 4-2. Demonstration Approach (Continued)

        "    Dexsil chose not to analyze low- and medium-concentration-range weathered gasoline soil PE samples because according to Dexsil, the PetroFLAG™ System was not sensitive to weathered
             gasoline concentrations of less than 1,000 mg/kg.

        '    Because of problems that arose during preparation of PE samples with low concentrations of weathered gasoline, the results for these samples were not used to evaluate the field measurement
             devices.

-------
•   The portability of the device (secondary objective S3)
    was evaluated by observing and noting the weight and
    size of the device and additional equipment required
    for TPH measurement as well as how easily the device
    was set up for use during the demonstration.

•   The durability of the device (secondary objective S4)
    was evaluated by noting the materials of construction
    of the device and additional equipment required for
    TPH measurement.  In addition, EPA representatives
    noted likely device failures or repairs that may be
    necessary  during  extended  use  of the device.
    Downtime required to make device repairs during the
    demonstration was also noted.

•   The availability of the device and associated spare
    parts  (secondary objective  S5) was evaluated by
    discussing the availability of replacement devices with
    developer personnel and determining whether spare
    parts were available in retail stores or only from the
    developer. In addition, the availability of spare parts
    required during the demonstration was noted.

Field observations of the analyses of all the samples
described in Table 4-2 were used to address the secondary
objectives for the demonstration.

4.3    Sample Preparation and Management

This section presents sample preparation and management
procedures used during the demonstration. Specifically,
this  section describes  how  samples  were collected,
containerized,  labeled, stored, and shipped during the
demonstration.   Additional details about  the sample
preparation and management procedures are presented in
the demonstration plan (EPA 2000).

4.3.1  Sample Preparation

The sample preparation procedures for both environmental
and PE samples are described below.

Environmental Samples

For  the  demonstration,  environmental  samples were
collected  in  the   areas  that  were   used  for  the
predemonstration investigation:  (1) the FFA and NEX
Service Station Area at the Navy BVC site; (2) the B-38
Area at the Kelly AFB site; and (3) the SFT  Area at the
PC site.   Samples were collected in all areas using a
Geoprobe®.

The liners  containing environmental  samples  were
transported to the sample management trailer at the Navy
BVC site, where the liners were cut open longitudinally.
A geologist then profiled the  samples based on soil
characteristics to determine where the soil cores had to be
sectioned. The soil characterization performed for each
demonstration area is summarized in Chapter 3.

Each core sample  section was then transferred  to  a
stainless-steel bowl. The presence of any unrepresentative
material such as sticks, roots, and stones was noted in a
field logbook, and such material was removed to the extent
possible using gloved hands. Any lump of clay in the
sample that was greater than about 1/8 inch in diameter
was  crushed  between   gloved   fingers   before
homogenization. Each soil sample was homogenized by
stirring it for at least 2 minutes using a stainless-steel
spoon or gloved hands until  the  sample was  visibly
homogeneous.    During  or  immediately  following
homogenization, any free water was  poured from the
stainless-steel  bowl containing the soil sample into a
container designated for IDW. During the demonstration,
the field sampling team used only nitrile gloves to avoid
the possibility of phthalate contamination from handling
samples with plastic gloves.   Such contamination had
occurred during the predemonstration investigation.

After sample homogenization, the samples were placed in
(1) EnCores of approximately 5-gram capacity for GRO
analysis; (2) 4-ounce, glass jars provided by the reference
laboratory for EDRO and percent moisture analyses; and
(3) EnCores of approximately 25-gram capacity for TPH
analysis.   Using a  quartering  technique, each  sample
container was filled by alternately spooning soil from one
quadrant of the mixing bowl and then from the opposite
quadrant until the container was full. The 4-ounce,  glass
jars were filled after all the EnCores for a given sample
had been filled.  After a sample container was filled, it was
immediately  closed  to  minimize  volatilization  of
contaminants.  To minimize the time required for sample
homogenization and filling of  sample containers,  these
activities  were simultaneously  conducted  by   four
personnel.

Because of the large number of containers being filled,
some time elapsed between the filling of the first EnCore
and the filling of the  last.  An attempt was made to
                                                    31

-------
eliminate any bias by alternating between filling EnCores
for the developers and filling EnCores for the reference
laboratory.   Table 4-3  summarizes the demonstration
sampling depth intervals, numbers of environmental and
QA/QC samples collected, and numbers of environmental
sample analyses associated with the demonstration of the
PetroFLAG™ System.

Performance Evaluation Samples

All PE samples for the demonstration were prepared by
ERA and shipped to the sample  management  trailer at
the Navy BVC site.  PE samples consisted of both soil
samples  and liquid samples.   ERA prepared soil PE
samples  using  two  soil matrixes:  Ottawa sand  and
processed garden soil (silty sand).
                            To prepare the soil PE samples, ERA spiked the required
                            volume of soil based on the number of PE samples and the
                            quantity  of soil per PE sample requested.  ERA then
                            homogenized the soil by manually mixing it. ERA used
                            weathered gasoline or diesel as the spiking material, and
                            spiking was done at three levels to depict the three TPH
                            concentration ranges:  low,  medium,  and high.   A
                            low-range sample was  spiked to correspond to a TPH
                            concentration of less than 100 mg/kg;  a medium-range
                            sample was spiked to correspond to a TPH concentration
                            range of 100 to 1,000 mg/kg; and a high-range sample was
                            spiked to correspond to a TPH concentration of more than
                            1,000 mg/kg.

                            To spike each low- and medium-range soil sample, ERA
                            used methanol as a "carrier" to distribute the contaminant
Table 4-3. Environmental Samples
Site
Navy
BVC
Kelly
AFB
PC
Area
FFA
NEX
Service
Station
Area
B-38
Area
SFT
Area
Depth
Interval
(foot bgs)
Upper layer
Lower layer
7 to 8
8 to 9
9 to 10
10 to 11
23 to 25
25 to 27
2 to 4
4 to 6
6 to 8
8 to 10
Number of
Sampling
Locations
3
3
3
3
3
3
3
3
5
5
5
5
Total
Total Number of
Samples, Including
Field Triplicates, to
Dexsil and Reference
Laboratory3
5
5
5
5
5
5
5
3
7
7
7
7
66
Number of
MS/MSD"
Pairs
1
1
1
1
1
1
1
1
1
1
1
1
12
Number of
Extract
Duplicates0
1
1
1
1
1
1
1
1
1
1
1
1
12
Number of
TPH Analyses
by Dexsil
6
6
6
6
6
6
6
4
8
8
8
8
78
Number of Analyses
by Reference
Laboratory
GRO
0
0
8
8
8
8
8
6
10
10
10
10
86
EDRO
8
8
8
8
8
8
8
6
10
10
10
10
102
Notes:

AFB   = Air Force Base
bgs   = Below ground surface
BVC   = Base Ventura County
FFA     = Fuel Farm Area
MS/MSD = Matrix spike and matrix spike duplicate
NEX     = Naval Exchange
PC   =  Petroleum company
SFT  =  Slop Fill Tank
    Field triplicates were collected at a frequency of one per depth interval in each sampling area except the B-38 Area. Because of conditions in the
    B-38 Area, triplicates were collected in the top depth interval only. Three separate, blind samples were prepared for each field triplicate.
    MS/MSD samples were collected at a frequency of one per depth interval in each sampling area for analysis by the reference laboratory. MS/MSD
    samples were not analyzed by Dexsil.
    Because of site conditions, Dexsil did not analyze extract duplicates for the upper clay layer in the FFA and the 10- to 11-foot bgs depth interval
    in the NEX Service Station Area.
    All environmental samples were also analyzed for moisture content by the reference laboratory.
                                                      32

-------
evenly throughout the sample.  Soil PE samples were
spiked with interferents  at two different levels ranging
from 50 to 500 percent of the TPH concentration expected
to be present.  Whenever possible, the interferents were
added at levels that best represented real-world conditions.
ERA analyzed the samples containing weathered gasoline
before shipping them  to the Navy BVC site.   The
analytical  results  were  used   to  confirm  sample
concentrations.

Liquid PE samples consisted of neat materials. Each liquid
PE sample consisted of approximately 2 mL of liquid in a
flame-sealed, glass ampule. During the demonstration, the
developers and reference laboratory were given a table
informing them of the amount of liquid sample to be used
for analysis.

ERA grouped like PE samples together in a resealable bag
and placed all the PE samples in a cooler containing ice for
overnight shipment to the Navy BVC site. When the PE
samples arrived at the site, the  samples were labeled with
the appropriate sample identification numbers and placed
in appropriate coolers for transfer to the developers on site
or for shipment to the reference laboratory as summarized
in Section 4.3.2. Table 4-4 summarizes the contaminant
types and concentration ranges as well as the numbers of
PE samples and analyses associated with the demonstration
of the PetroFLAG™ System.

4.3.2  Sample Management

Following sample containerization, each environmental
sample was assigned a unique sample designation defining
the sampling  area,  expected  type of  contamination,
expected concentration range, sampling location, sample
number,  and QC identification, as appropriate.  Each
sample container was labeled  with the  unique  sample
designation, date, time, preservative, initials of personnel
who had filled the container, and analysis to be performed.
Each  PE  sample was also assigned a  unique  sample
designation that  identified it as a PE sample.  Each PE
sample   designation  also  identified  the  expected
contaminant type and range, whether the sample was soil
or liquid, and the sample number.

Sample custody began when samples were placed in iced
coolers in the possession  of the designated field sample
custodian. Demonstration samples were divided into two
groups to allow  adequate time for the developers and
reference laboratory to extract and analyze samples within
the method-specified holding times presented in Table 4-5.
The two  groups of  samples  for reference laboratory
analysis were placed in coolers  containing ice and chain-
of-custody forms and were sh ipped by overnight courier to
the reference laboratory on the first and third days of the
demonstration. The two groups of samples for developer
analysis were placed in coolers  containing ice and chain-
of-custody  forms  and  were  hand-delivered  to  the
developers at the Navy BVC site on the same days that the
reference laboratory received its two groups of samples.
During the demonstration, each developer was provided
with a tent to provide shelter from direct sunlight during
analysis of demonstration samples. In addition, at the end
of each day, the developer placed any samples or sample
extracts  in its custody in  coolers, and  the coolers were
stored in a refrigerated truck.
                                                    33

-------
Table 4-4. Performance Evaluation Samples
Sample Type
Typical TPH
Concentration
Range3
Total Number
of Samples to
Dexsil and
Reference
Laboratory
Number of
MS/MSD"
Pairs
Number of
Analyses by
Dexsil
Number of
Analyses by Reference
Laboratory*1
GRO
EDRO
Soil Samples (Ottawa Sand) *~ '-".',1 '' - ,"- ,. '_!-" {^-. .'-VT "*--C/w.~4> -~ ,"-* "~ , „" / "--._*" ""I * -~zT.~
Weathered gasoline
Diesel
Low
7
7
0
0
0"
7
7
0
7
7
SoifjSamples (Processed Garden Soil) ^,,jr 'V-',^1' -JL ,"*'>ij '.Ltr-:;'""'"*'' — - ,-;,.-« -.7£ir-Z"-'" • "-"S-l^i;'-'* -«-"' -f!.'/^"' *- --
Weathered gasoline
Diesel
Blank soil (control sample)
MTBE (1 ,100 mg/kg) and weathered gasoline
MTBE (1 ,700 mg/kg) and weathered gasoline
PCE (2,810 mg/kg) and weathered gasoline
PCE (13,100 mg/kg) and weathered gasoline
Stoddard solvent (2,900 mg/kg) and weathered
gasoline
Stoddard solvent (15,400 mg/kg) and weathered
gasoline
Turpentine (2,730 mg/kg) and weathered
gasoline
Turpentine (12,900 mg/kg) and weathered
gasoline
Stoddard solvent (3,650 mg/kg) and diesel
Stoddard solvent (1 8,200 mg/kg) and diesel
Turpentine (3,850 mg/kg) and diesel
Turpentine (19,600 mg/kg) and diesel
1 ,2,4-Trichlorobenzene (3,350 mg/kg) and
diesel
1 ,2,4-Trichlorobenzene (16,600 mg/kg) and
diesel
Humic acid (3,940 mg/kg) and diesel
Humic acid (19,500 mg/kg) and diesel
Humic acid (3,940 mg/kg)
Humic acid (19,500 mg/kg)
Weathered gasoline at 16 percent moisture
Diesel at negligible moisture (less than
1 percent)
Medium
High
Medium
High
Trace
High
Trace
High
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
^^^^m0ef^ie^'i^ael^^^i~2^ry"fi'y^'^-'f,'L I^r^f*111""" ^-5. — '
Weathered gasoline
Diesel
MTBE
High
3
3
6
0
1
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
0"
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
tiff TSl: " ;4 ^1-4. v-lfS
1
0
0
3
3
6
3^
5
0
0
5
3
3
3
3
3
3
3
3
3
3
3
3
0
0
0
0
0
0
5
0
3
5
3
5
5
3
3
3
3
3
3
3
3 .
3
3
3
3
3
3
3
3
3
3
5
5
L,*. -.•?- £ ^'Z&ff-K *
5
0
6
5
3
0
                                                          34

-------
Table 4-4. Performance Evaluation Samples (Continued)
Sample Type
Typical TPH
Concentration
Range'
Total Number
of Samples to
Dexsil and
Reference
Laboratory
Liquid Samples (N«at Material) (Continued) . ' * ". 7
PCE
Stoddard solvent
Turpentine
1 ,2,4-Trichlorobenzene
Not applicable
High
Not applicable
Total
6
6
6
6
125
Number of
MS/MSD'
Pairs
Number of
Analyses by
Dexsil
„ -' :~ ,.'•'* ~r
0
0
0
0
6
6
6
6
6
115
Number of
Analyses by Reference
Laboratory0
GRO
EDRO
'• -- • •
6
6
6
0
90
0
6
6
6
125
Notes:

mg/kg    =  Milligram per kilogram
MS/MSD =  Matrix spike and matrix spike duplicate
MTBE =  Methyl-tert-butyl ether
PCE  =  Tetrachloroethene
    The typical TPH concentration range was based on reference laboratory results for the demonstration. The typical low, medium, and high ranges
    indicate TPH concentrations  of less than 100 mg/kg; 100 to 1,000 mg/kg; and greater than 1,000 mg/kg, respectively.  The typical TPH
    concentration range for the liquid sample concentrations was based on the definition of TPH used for the demonstration and knowledge of the
    sample (neat material).
    MS/MSD samples were analyzed only by the reference laboratory.
    All soil performance evaluation samples were also analyzed for moisture content by the reference laboratory.
    Dexsil chose not to analyze low- and medium-concentration-range weathered gasoline soil performance evaluation samples because according
    to Dexsil, the PetroFLAG™ System was not sensitive to weathered gasoline concentrations of less than 1,000 mg/kg.
                                                               35

-------
Table 4-5. Sample Container, Preservation, and Holding Time Requirements

Parameter*
GRO
EDRO
Percent moisture
TPH
GRO and EDRO
Notes:
± = Plus or minus

Medium
Soil
Soil
Soil
Soil
Liquid



Container
Two 5-gram EnCores
Two 4-ounce, glass jars with Teflon™-lined lids
Two 4-ounce, glass jars with Teflon™-lined lids
One 25-gram EnCore
One 2-milliliter ampule for each analysis



Preservation
4±2°C
4±2°C
4±2°C
4±2°C
Not applicable


Holding Time (days)
Extraction Analysis
2" 14
14" 40
Not applicable 7
Performed on site0
See note d


    The reference laboratory measured percent moisture using part of the soil sample from the container designated for EDRO analysis.

    The extraction holding time started on the day that samples were shipped.

    If GRO analysis of a sample was to be completed by the reference laboratory, the developers completed on-site extraction of the corresponding
    sample within 2 days. Otherwise, all on-site extractions and analyses were completed within 7 days.

    The reference laboratory cracked open each ampule and immediately added the specified aliquot of the sample to methanol for GRO analysis and
    to methylene chloride for EDRO analysis. This procedure was performed in such a way that the final volumes of the extracts for GRO and EDRO
    analyses were 5.0 milliliters and 1.0 milliliter, respectively. Once the extracts were prepared, the GRO and EDRO analyses were performed within
    14 and 40 days, respectively.
                                                               36

-------
                                              Chapter 5
                                       Confirmatory Process
The  performance results for  each field measurement
device were compared to those for an off-site laboratory
measurement method—that is, a reference method. This
chapter  describes the rationale for the selection of the
reference method (Section 5.1) and reference laboratory
(Section  5.2) and summarizes project-specific  sample
preparation and analysis procedures associated with the
reference method (Section 5.3).

5.1    Reference Method Selection

During the demonstration, environmental and PE samples
were analyzed for TPH by the reference laboratory using
SW-846 Method 8015B (modified). This section describes
the analytical methods considered for the demonstration
and provides a rationale for the reference method selected.

The reference method used was selected based on the
following criteria:

   It is  not a field screening method.

   It is  widely used and accepted.

•  It measures light (gasoline) to heavy (lubricating oil)
   fuel  types.

•  It can provide separate measurements of GRO and
   EDRO fractions of TPH.

•  It meets project-specific reporting limit requirements.

The analytical methods considered for the demonstration
and the  reference method selected based on  the above-
listed criteria  are  illustrated  in  a flow diagram in
Figure 5-1. The reference method selection process is
discussed below.
Analytical methods considered for the demonstration were
identified based on a review of SW-846, "Methods for
Chemical Analysis of Water and Wastes" (MCAWW),
ASTM, API, and state-specific methods. The analytical
methods considered collectively represent six different
measurement technologies.  Of the methods reviewed,
those identified  as  field screening  methods,  such as
SW-846 Method  4030,  were eliminated  from  further
consideration in the reference method selection process.

A literature review was conducted to determine whether
the remaining methods are widely used and accepted in the
United States (Association for the Environmental Health of
Soil  [AEHS]  1999).  As a result of this review, state-
specific methods such as the Massachusetts Extractable
Petroleum Hydrocarbon  (EPH) and Volatile Petroleum
Hydrocarbon (VPH) Methods (Massachusetts Department
of Environmental Protection 2000), the Florida Petroleum
Range Organic (PRO) Method (Florida Department of
Environmental Protection 1996), and Texas Method 1005
(Texas Natural Resource Conservation Commission 2000)
were  eliminated  from  the  selection  process.   Also
eliminated were  the  gravimetric and infrared methods
except forMCAWWMethod418.1 (EPA 1983). The use
and acceptability of MCAWW Method 418.1 will likely
decline because the extraction solvent used in this method
is Freon 113, a chlorofluorocarbon (CFC), and use of
CFCs will eventually be  phased out under the Montreal
Protocol.  However, because several states still accept the
use of MCAWW Method 418.1 for measuring TPH, the
method was retained for further consideration in the
selection process (AEHS  1999).

Of the remaining methods, MCAWW Method 418.1, the
API PHC Method, and SW-846 Method 8015B can all
measure light (gasoline)  to heavy (lubricating oil) fuel
types.  However,  GRO and  EDRO fractions cannot be
                                                   37

-------
                    Measures light
                  (gasoline) to heavy
                    (lubricating oil)
                      fuel types?
                                                                           Reference method selected
nethods
3d 413.1
3d 413.2
id 8440
)d 9071
•nethods
od413.1
od413.2
od418.1
athod
d 801 SB
)d 8440
Jd 9071
                                                                         SW-846 Method 8015B (modified)
   Yes->
MCAWW Method 41 8.1
API PHC Method
SW-846 Method 801 5B
^

          Yes->
                 MCAWW Method 418.1
                    API PHC Method
                 SW-846  Method 801 SB
                                                                                     Meets
                                                                            project-specific reporting limit
                                                                                  requirements?
        Considered a field
       screening method?
                             Yes—»
                Provides
          separate measurements
            of GRO and EDRO
            fractions of TPH?
                                              MCAWW Method 418.1
                                            AP PHC Method
ASTM Method D 5831 -96
 SW-846 Method 4030
 SW-846 Method 9074
                                                                   reference method
sr Testing and Materials, DRO = diesel range organics, EPH = extractable petroleum hydrocarbon, GC/FID = gas chromatograph/flame
s of Water and Wastes," PHC = petroleum hydrocarbon, PRO = petroleum range organics, SW-846 = "Test Methods for Evaluating
 measurements and, when modified, can also provide EDRO measurements.

-------
measured separately using MCAWW Method 418.1. As
a result, this method was eliminated from the  selection
process.

Both the API PHC Method and SW-846 Method 8015B
can be  used to separately measure the GRO and DRO
fractions of TPH. These methods can also be modified to
extend the DRO range to EDRO by  using a calibration
standard that includes  even-numbered alkanes in the
EDRO range.

Based on a review of state-specific action levels for TPH,
a TPH  reporting limit of 10 mg/kg was  used for the
demonstration. Because the TPH reporting limit for the
API PHC  Method  (50  to 100 mg/kg) is greater  than
10 mg/kg, this method was eliminated from the selection
process (API 1994). SW-846 Method 8015B (modified)
met the reporting limit requirements for the demonstration.
For   GRO,   SW-846   Method  8015B   (modified)
has a reporting limit of 5 mg/kg,  and for  EDRO, this
method has a reporting limit of 10 mg/kg.  Therefore,
SW-846 Method 8015B (modified) satisfied all the criteria
established for selecting the reference method.  As an
added benefit, because  this is  a  GC method,  it also
provides   a  fingerprint  (chromatogram)  of  TPH
components.

5.2     Reference Laboratory Selection

This section provides the rationale for the selection of the
reference laboratory. STL Tampa East was selected as the
                                        reference laboratory because it (1) has been performing
                                        TPH  analyses  for  many  years, (2) has passed many
                                        external audits by successfully implementing a variety of
                                        TPH  analytical methods,  and (3) agreed to implement
                                        project-specific analytical requirements. In January 2000,
                                        a project-specific audit of the laboratory was conducted
                                        and determined  that  STL  Tampa  East  satisfactorily
                                        implemented   the   reference  method   during   the
                                        predemonstration investigation. In addition, STL Tampa
                                        East successfully analyzed double-blind PE samples and
                                        blind field triplicates for GRO and EDRO during the
                                        predemonstration investigation. Furthermore, in 1998 STL
                                        Tampa East was one of four recipients and in 1999 was
                                        one of six recipients  of the Seal of Excellence Award
                                        issued  by  the  American   Council  of   Independent
                                        Laboratories.  In each instance, this award was issued
                                        based on the results of PE sample analyses and client
                                        satisfaction surveys. Thus, the selection of the reference
                                        laboratory  was based  primarily on performance and not
                                        cost.

                                        5.3    Summary of Reference Method

                                        The laboratory sample preparation and analytical methods
                                        used for the demonstration are summarized in Table 5-1.
                                        The SW-846 methods listed in Table 5-1 for GRO and
                                        EDRO analyses were tailored to meet the definition of
                                        TPH  for the project (see Chapter 1).  Project-specific
                                        procedures for soil  sample preparation and analysis for
                                        GRO and EDRO are summarized in Tables 5-2 and 5-3,
                                        respectively.  Project-specific procedures  were applied
Table 5-1. Laboratory Sample Preparation and Analytical Methods
Parameter
          Method Reference (Step)
                                                                             Method Title
GRO
EDRO
Based on SW-846 Method 5035 (extraction)

Based on SW-846 Method 5030B (purge-and-trap)
Based on SW-846 Method 8015B (analysis)

Based on SW-846 Method 3540C (extraction)
Based on SW-846 Method 8015B (analysis)
Percent moisture   Based on MCAWW Method 160.3"
Closed-System Purge-and-Trap and Extraction for Volatile Organics
in Soil and Waste Samples
Purge-and-Trap for Aqueous Samples
Nonhalogenated Volatile Organics by Gas Chromatography

Soxhlet Extraction
Nonhalogenated Volatile Organics by Gas Chromatography

Residue, Total (Gravimetric, Dried at 103-105 °C)
Notes:

MCAWW = "Methods for Chemical Analysis of Water and Wastes"
SW-846  = Test Methods for Evaluating Solid Waste"

'   MCAWW Method 160.3 was modified to include calculation and reporting of percent moisture in soil samples.
                                                     39

-------
Table 5-2. Summary of Project-Specific Procedures for GRO Analysis
SW-846 Method Reference (Step)
5035 (Extraction) v. . _ ,. >. .
Low-level (0.5 to 200 micrograms per kilogram) or high-level (greater
than 200 micrograms per kilogram) samples may be prepared.
Samples may be collected with or without use of a preservative
solution.
A variety of sample containers, including EnCores, may be used when
high-level samples are collected without use of a preservative.
Samples collected in EnCores should be transferred to vials containing
the extraction solvent as soon as possible or analyzed within 48 hours.
For samples not preserved in the field, a solubility test should be
performed using methanol, polyethylene glycol, and hexadecane to
determine an appropriate extraction solvent.
Removal of unrepresentative material from the sample is not discussed.
Procedures for adding surrogates to the sample are inconsistently
presented. Section 2.2.1 indicates that surrogates should be added to
an aliquot of the extract solution. Section 7.3.3 indicates that soil
should be added to a vial containing both the extraction solvent
(methanol) and surrogate spiking solution.
Nine ml of methanol should be added to a 5-gram (wet weight) soil
sample.
When practical, the sample should be dispersed to allow contact with
the methanol by shaking or using other mechanical means for 2 min
without opening the sample container. When shaking is not practical,
the sample should be dispersed with a narrow, metal spatula, and the
sample container should be immediately resealed.
5030B . •»• '•*_"' ,' _• -i _, Ei* •**"•«
Because the project-specific reporting limit for GRO was 5 milligrams
per kilogram, all samples analyzed for GRO were prepared using
procedures for high-level samples.
Samples were collected without use of a preservative.
Samples were containerized in EnCores.
Samples were weighed and extracted within 2 calendar days of their
shipment. The holding time for analysis was 14 days after extraction.
A full set of quality control samples (method blanks, MS/MSDs, and
LCS/LCSDs) was prepared within this time.
Because the reference laboratory obtained acceptable results for
performance evaluation samples extracted with methanol during the
predemonstration investigation, samples were extracted with methanol.
During sample homogenization, field sampling technicians attempted to
remove unrepresentative material such as sticks, roots, and stones if
present in the sample; the reference laboratory did not remove any
remaining unrepresentative material.
The soil sample was ejected into a volatile organic analysis vial, an
appropriate amount of surrogate solution was added to the sample, and
then methanol was quickly added.
Five ml of methanol was added to the entire soil sample contained in a
5-gram EnCore.
The sample was dispersed using a stainless-steel spatula to allow
contact with the methanol. The volatile organic analysis vial was then
capped and shaken vigorously until the soil was dispersed in methanol,
and the soil was allowed to settle.
„-- V ••'." -"- - 5 _: ' %J,i. "'---*--- ~ H
Samples were screened with an automated headspace sampler (see
SW-846 Method 5021) connected to a GC equipped with a flame
ionization detector.
According to the reference laboratory, a sample extract concentration
equivalent to 10,000 ng on-column is the minimum concentration of
GRO that could result in carryover. Therefore, if a sample extract had
a concentration that exceeded the minimum concentration for
carryover, the next sample in the sequence was evaluated as follows:
(1) if the sample was clean (had no chromatographic peaks), no
carryover had occurred; (2) if the sample had detectable analyte
concentrations (chromatographic peaks), it was reanalyzed under
conditions in which carryover did not occur.
                                                          40

-------
Table 5-2. Summary of Project-Specific Procedures for GRO Analysis (Continued)
SW-846 Method Reference (Step)
Project-Specific Procedures
5030B (Purge-and-Trap) (Continued) _ 7.; r - "r .,...* ', - ''-?';- .\~'"**- ' ".' i'-'':.f " " -_ " - "' •
The sample purge device used must demonstrate adequate
performance.
Purge-and-trap conditions for high-level samples are not clearly
specified. According to SW-846, manufacturer recommendations for
the purge-and-trap devices should be considered when the method is
implemented. The following general purge-and-trap conditions are
recommended for samples that are water-miscible (methanol extract):
Purge gas: nitrogen or helium
Purge gas flow rate: 20 mL/min
Purge time: 15 ± 0.1 min
Purge temperature: 85 ± 2 °C
Oesorb time: 1 .5 min
Desorb temperature: 1 80 °C
Backflush inert gas flow rate: 20 to 60 mUmin
Bake time: not specified
Bake temperature: not specified
Multiport valve and transfer line temperatures: not specified
A Tekmar 2016 autosampler and a Tekmar LSC 2000 concentrator
were used. Based on quality control sample results, the reference
laboratory had demonstrated adequate performance using these
devices.
The purge-and-trap conditions that were used are listed below. These
conditions were based on manufacturer recommendations for the purge
device specified above and the VOCARB 3000 trap.
Purge gas: helium
Purge gas flow rate: 35 mL/min
Purge time: 8 min with 2-min dry purge
Purge temperature: ambient temperature
Desorb time: 1 min
Desorb temperature: 250 °C
Backflush inert gas flow rate: 35 mL/min
Bake time: 7 min
Bake temperature: 270 °C
Multiport valve and transfer line temperatures: 115 and 120 °C
»^*««i^:: ??2^^
GC Conditions
The following GC conditions are recommended:
Column: 30-meter x 0.53-millimeter-inside diameter, fused-silica
capillary column chemically bonded with 5 percent methyl
silicone, 1.5-micrometer field thickness
Carrier gas: helium
Carrier gas flow rate: 5 to 7 mL/min
Makeup gas: helium
Makeup gas flow rate: 30 mL/min
Injector temperature: 200 °C
Detector temperature: 340 °C
Temperature program:
Initial temperature: 45 °C
Hold time: 1 min
Program rate: 45 to 100 °C at 5 °C/min
Program rate: 100 to 275 "C at 8 °C/min
Hold time: 5 min
Overall time: 38.9 min
The HP 5890 Series II was used as the GC. The following GC
conditions were used based on manufacturer recommendations:
Column: 30-meter x 0.53-millimeter-inside diameter, fused-silica
capillary column chemically bonded with 5 percent methyl
silicone, 1 .5-micrometer field thickness
Carrier gas: helium
Carrier gas flow rate: 15 mL/min
Makeup gas: helium
Makeup gas flow rate: 15 mL/min
Injector temperature: 200 °C
Detector temperature: 200 °C
Temperature program:
Initial temperature: 25 °C
Hold time: 3 min
Program rate: 25 to 120 °C at 25 °C/min
Hold time: 4 min
Program rate: 120 to 245 °C at 25 °C/min
Hold time: 5 min
Overall time: 20.4 min
Calibration
The chromatographic system may be calibrated using either internal or
external standards.
Calibration should be performed using samples of the specific fuel type
contaminating the site. When such samples are not available, recently
purchased, commercially available fuel should be used.
The chromatographic system was calibrated using external standards
with a concentration range equivalent to 100 to 10,000 ng on-column.
The reference laboratory acceptance criterion for initial calibration was a
relative standard deviation less than or equal to 20 percent of the
average response factor or a correlation coefficient for the least-
squares linear regression greater than or equal to 0.990.
Calibration was performed using a commercially available,
10-component GRO standard that contained 35 percent aliphatic
hydrocarbons and 65 percent aromatic hydrocarbons.
                                                           41

-------
Table 5-2.  Summary of Project-Specific Procedures for GRO Analysis (Continued)
SW-846 Method Reference (Step)
Project-Specific Procedures
80l^{Anafi^te);{Qo«artu«i)i: ' ~ . ,:" * % : " , . -/.% : *: - ':'_ ',""'•• 1:
Calibration (Continued)
Initial calibration verification is not required.
CCV should be performed at the beginning of every 12-hour work shift
and at the end of an analytical sequence. CCV throughout the 12-hour
shift is also recommended; however, the frequency is not specified.
CCV should be performed using a fuel standard.
According to SW-846 Method 8000, CCV should be performed at the
same concentration as the midpoint concentration of the initial
calibration curve; however, the concentration of each calibration point is
not specified.
A method sensitivity check is not required.
Initial calibration verification was performed using a second-source
standard that contained a 10-component GRO standard made up of
35 percent aliphatic hydrocarbons and 65 percent aromatic
hydrocarbons at a concentration equivalent to 2,000 ng on-column. The
reference laboratory acceptance criterion for initial calibration
verification was an instrument response within 25 percent of the
response obtained during initial calibration.
CCV was performed at the beginning of each analytical batch, after
every tenth analysis, and at the end of the analytical batch. The
reference laboratory acceptance criteria for CCV were instrument
responses within 25 percent (for the closing CCV) and 15 percent (for
all other CCVs) of the response obtained during initial calibration.
CCV was performed using a commercially available, 10-component
GRO standard that contained 35 percent aliphatic hydrocarbons and
65 percent aromatic hydrocarbons.
CCV was performed at a concentration equivalent to 2,000 ng
on-column.
A method sensitivity check was performed daily using a calibration
standard with a concentration equivalent to 100 ng on-column. The
reference laboratory acceptance criterion for the method sensitivity
check was detection of the standard.
Retention Time Windows
The retention time range (window) should be established using
2-methylpentane and 1 ,2,4-trimetnylbenzene during initial calibration.
Three measurements should be made over a 72-hour period; the results
should be used to determine the average retention time. As a minimum
requirement, the retention time should be verified using a midlevel
calibration standard at the beginning of each 12-hour shift. Additional
analysis of the standard throughout the 12-hour shift is strongly
recommended.
The retention time range was established using the opening CCV
specific to each analytical batch. The first eluter, 2-methylpentane, and
the last eluter, 1 ,2,4-trimethylbenzene, of the GRO standard were used
to establish each day's retention time range.
Quantitation
Quantitation is performed by summing the areas of all chromatographic
peaks eluting within the retention time range established using
2-methylpentane and 1,2,4-trimethylbenzene. Subtraction of the
baseline rise for the method blank resulting from column bleed is
generally not required.
Quantitation was performed by summing the areas of all
chromatographic peaks from 2-methylpentane through
1,2,4-trimethylbenzene. This range includes n-C10. Baseline rise
subtraction was not performed.
Quality Control
Spiking compounds for MS/MSDs and LCSs are not specified.
According to SW-846 Method 8000, spiking levels for MS/MSOs are
determined differently for compliance and noncompliance monitoring
applications. For noncompliance applications, the laboratory may spike
the sample (1) at the same concentration as the reference sample
(LCS), (2) at 20 times the estimated quantitation limit for the matrix of
interest, or (3) at a concentration near the middle of the calibration
range.
The spiking compound mixture for MS/MSDs and LCSs was the
10-component GRO calibration standard.
MS/MSD spiking levels were targeted to be between 50 and
1 50 percent of the unspiked sample concentration. The reference
laboratory used historical information to adjust spike amounts or to
adjust sample amounts to a preset spike amount. The spiked samples
and unspiked samples were prepared such that the sample mass and
extract volume used for analysis were the same.
                                                           42

-------
Table 5-2. Summary of Project-Specific Procedures for GRO Analysis (Continued)
SW-846 Method Reference (Step)
Project-Specific Procedures
8Q15B (Analysis) (Continued) - . ,,"-•-..,- .-
Quality Control (Continued)
According to SW-846 Method 8000, in-house laboratory acceptance
criteria for MS/MSDs and LCSs should be established. As a general
rule, the recoveries of most compounds spiked into a sample should fall
within the range of 70 to 1 30 percent, and this range should be used as
a guide in evaluating in-house performance.
The LCS should consist of an aliquot of a dean (control) matrix that is
similar to the sample matrix.
No LCSD is required.
The surrogate compound and spiking concentration are not specified.
According to SW-846 Method 8000, in-house laboratory acceptance
criteria for surrogate recoveries should be established.
The method blank matrix is not specified.
The extract duplicate is not specified.
The reference laboratory acceptance criteria for MS/MSDs and LCSs
were a relative percent difference less than or equal to 25 with 33 to
115 percent recovery. The acceptance criteria were based on
laboratory historical information. These acceptance criteria are similar
to those of the methods cited in Figure 5-1 .
The LCS/LCSD matrix was Ottawa sand.
The spiking compound mixture for LCSDs was the 10-component GRO
calibration standard.
The surrogate compound was 4-bromofluorobenzene. The reference
laboratory acceptance criterion for surrogates was 39 to 163 percent
recovery.
The method blank matrix was Ottawa sand. The reference laboratory
acceptance criterion for the method blank was less than or equal to the
project-specific reporting limit.
The extract duplicate was analyzed. The reference laboratory
acceptance criterion for the extract duplicate was a relative percent
difference less than or equal to 25.
Notes:

±
ccv
GC
LCS
LCSD
Plus or minus
Continuing calibration verification
Gas chromatograph
Laboratory control sample
Laboratory control sample duplicate
min      = Minute
mL      = Milliliter
MS      = Matrix spike
MSD     = Matrix spike duplicate
ng       = Nanogram
SW-846  = Test Methods for Evaluating Solid Waste"
                                                             43

-------
Table 5-3. Summary of Project-Specific Procedures for EDRO Analysis
SW-846 Method Reference (Step)
Project-Specific Procedures
3540C{Brtracflooi r .. ",.*•.._-'-- - I "L : - : ~-
Any free water present in the sample should be decanted and
discarded. The sample should then be thoroughly mixed, and any
unrepresentative material such as sticks, roots, and stones should be
discarded.
Ten grams of soil sample should be blended with 10 grams of
anhydrous sodium sulfate.
Extraction should be performed using 300 mL of extraction solvent.
Acetone and hexane (1:1 volume per volume) or methylene chloride
and acetone (1 :1 volume per volume) may be used as the extraction
solvent.
Note: Methylene chloride and acetone are not constant-boiling
solvents and thus are not suitable for the method. Methylene
chloride was used as an extraction solvent for method
validation.
The micro Snyder column technique or nitrogen blowdown technique
may be used to adjust (concentrate) the soil extract to the required final
volume.
Procedures for addressing contamination carryover are not specified.
During sample homogenization, field sampling technicians attempted to
remove unrepresentative material such as sticks, roots, and stones. In
addition, the field sampling technicians decanted any free water present
in the sample. The reference laboratory did not decant water or remove
any unrepresentative material from the sample. The reference
laboratory mixed the sample with a stainless-steel tongue depressor.
Thirty grams of sample was blended with at least 30 grams of
anhydrous sodium sulfate. For medium- and high-level samples, 6 and
2 grams of soil were used for extraction, respectively, and proportionate
amounts of anhydrous sodium sulfate were added. The amount of
anhydrous sodium sulfate used was not measured gravimetrically but
was sufficient to ensure that free moisture was effectively removed from
the sample.
Extraction was performed using 200 mL of extraction solvent.
Methylene chloride was used as the extraction solvent.
Kudema Danish and nitrogen evaporation were used as the
concentration techniques.
According to the reference laboratory, a sample extract concentration of
100,000 micrograms per mL is the minimum concentration of EDRO
that could result in carryover. Therefore, if a sample extract had a
concentration that exceeded the minimum concentration for carryover,
the next sample in the sequence was evaluated as follows: (1 ) if the
sample was clean (had no chromatographic peaks), no carryover
occurred; (2) if the sample had detectable analyte concentrations
(chromatographic peaks), it was reanalyzed under conditions in which
carryover did not occur.
^OtSB-(AnilxsH\-^- " ','< ' ,' ~ >_<...'(£'.: rf'" \ ^ '\ *»£,;. . *"V **? "' • §•-" f'. ** - :''J3r -^ •
GC Conditions
The following GC conditions are recommended:
Column: 30-meter x 0.53-millimeter-inside diameter; fused-silica
capillary column chemically bonded with 5 percent methyl
silicone, 1 .5-micrometer field thickness
Carrier gas: helium
Carrier gas flow rate: 5 to 7 mL/min
Makeup gas: helium
Makeup gas flow rate: 30 mL/min
Injector temperature: 200 °C
Detector temperature: 340 °C
Temperature program:
Initial temperature: 45 °C
Hold time: 3 min
Program rate: 45 to 275 °C at 12 °C/min
Hold time: 12 min
Overall time: 34.2 min
An HP 6890 GC was used with the following conditions:
Column: 30-meter x 0.53-millimeter-inside diameter, fused-silica
capillary column chemically bonded with 5 percent methyl
silicone, 1. 5-micrometer field thickness
Carrier gas: hydrogen
Carrier gas flow rate: 1 .9 mL/min
Makeup gas: hydrogen
Makeup gas flow rate: 23 mL/min
Injector temperature: 250 °C
Detector temperature: 345 °C
Temperature program:
Initial temperature: 40 °C
Hold time: 2 min
Program rate: 40 to 345 °C at 30 °C/min
Hold time: 5 min
Overall time: 1 7.2 min
                                                          44

-------
Table 5-3. Summary of Project-Specific Procedures for EDRO Analysis (Continued)
SW-846 Method Reference (Step)
Project-Specific Procedures
8015B (Analysis) (Continued) ->...- ;: " . '1 -
Calibration
The chromatographic system may be calibrated using either internal or
external standards.
Calibration should be performed using samples of the specific fuel type
contaminating the site. When such samples are not available, recently
purchased, commercially available fuel should be used.
ICV is not required.
CCV should be performed at the beginning of every 12-hour work shift
and at the end of an analytical sequence. CCV throughout the 12-hour
shift is also recommended; however, the frequency is not specified.
CCV should be performed using a fuel standard.
According to SW-846 Method 8000, CCV should be performed at the
same concentration as the midpoint concentration of the initial
calibration curve; however, the concentration of each calibration point is
not specified.
A method sensitivity check is not required.
The chromatographic system was calibrated using external standards
with a concentration range equivalent to 75 to 7,500 ng on-column. The
reference laboratory acceptance criterion for initial calibration was a
relative standard deviation less than or equal to 20 percent of the
average response factor or a correlation coefficient for the least-
squares linear regression greater than or equal to 0.990.
Calibration was performed using a commercially available standard that
contained even-numbered alkanes from C1Q through C^.
ICV was performed using a second-source standard that contained
even-numbered alkanes from C10 through Cta at a concentration
equivalent to 3,750 ng on-column. The reference laboratory
acceptance criterion for ICV was an instrument response within
25 percent of the response obtained during initial calibration.
CCV was performed at the beginning of each analytical batch, after
every tenth analysis, and at the end of the analytical batch. The
reference laboratory acceptance criteria for CCV were instrument
responses within 25 percent (for the closing CCV) and 1 5 percent (for
all other CCVs) of the response obtained during initial calibration.
CCV was performed using a standard that contained only even-
numbered alkanes from C10 through C40
CCV was performed at a concentration equivalent to 3,750 ng
on-column.
A method sensitivity check was performed daily using a calibration
standard with a concentration equivalent to 75 ng on-column. The
reference laboratory acceptance criterion for the method sensitivity
check was detection of the standard.
Retention Time Windows
The retention time range (window) should be established using
C10 and C28 alkanes during initial calibration. Three measurements
should be made over a 72-hour period; the results should be used to
determine the average retention time. As a minimum requirement, the
retention time should be verified using a midlevel calibration standard at
the beginning of each 12-hour shift. Additional analysis of the standard
throughout the 12-hour shift is strongly recommended.
Two retention time ranges were established using the opening CCV for
each analytical batch. The first range, which was labeled diesel range
organics, was marked by the end of the 1 ,2,4-trimethylbenzene or n-C10
peak, whichever occurred later, through the n-octacosane peak. The
second range, which was labeled oil range organics, was marked by the
end of the n-octacosane peak through the tetracontane peak.
Quantitation
Quantitation is performed by summing the areas of all chromatographic
peaks eluting between n-C]0 and n-octacosane.
Quantitation was performed by summing the areas of all
chromatographic peaks from the end of the 1 ,2,4-trimethylbenzene or
n-C10 peak, whichever occurred later, through the n-octacosane peak.
A separate quantitation was also performed to sum the areas of all
chromatographic peaks from the end of the n-octacosane peak through
the tetracontane peak. Separate average response factors for the
carbon ranges were used for quantitation. The quantitation results were
then summed to determine the total EDRO concentration.
All calibrations, ICVs, CCVs, and associated batch quality control
measures were controlled for the entire EDRO range using a single
quantitation performed over the entire EDRO range.
                                                          45

-------
Table 5-3. Summary of Project-Specific Procedures for EDRO Analysis (Continued)
SW-846 Method Reference (Step)
Project-Specific Procedures
8015B (Analysis):(Continued) , . , .. ... ,,-.-.,,» ..- --"-"'-. -:
Quantitation (Continued)
Subtraction of the baseline rise for the method blank resulting from
column bleed is appropriate.
Because phthalate esters contaminate many types of products
commonly found in the laboratory, consistent quality control should be
practiced.
The reference laboratory identified occurrences of baseline rise in the
data package. The baseline rise was evaluated during data validation
and subtracted when appropriate based on analyst discretion.
Phthalate peaks were not noted during analysis.
Quality Control
Spiking compounds for MS/MSDs and LCSs are not specified.
According to SW-846 Method 8000, spiking levels for MS/MSDs are
determined differently for compliance and noncompliance monitoring
applications. For noncompliance applications, the laboratory may spike
the sample (1) at the same concentration as the reference sample
(LCS), (2) at 20 times the estimated quantitation limit for the matrix of
interest, or (3) at a concentration near the middle of the calibration
range.
According to SW-846 Method 8000, in-house laboratory acceptance
criteria for MS/MSDs and LCSs should be established. As a general
rule, the recoveries of most compounds spiked into a sample should fall
within the range of 70 to 130 percent, and this range should be used as
a guide in evaluating in-house performance.
The LCS should consist of an aliquot of a clean (control) matrix that is
similar to the sample matrix.
No LCSD is required.
The surrogate compound and spiking concentration are not specified.
According to SW-846 Method 8000, in-house laboratory acceptance
criteria for surrogate recoveries should be established.
The method blank matrix is not specified.
The extract duplicate is not specified.
The spiking compound for MS/MSDs and LCSs was an EDRO standard
that contained even-numbered alkanes from CIO through Cw.
MS/MSD spiking levels were targeted to be between 50 and
150 percent of the unspiked sample concentration. The reference
laboratory used historical information to adjust spike amounts or to
adjust sample amounts to a preset spike amount. The spiked samples
and unspiked samples were prepared such that the sample mass and
extract volume used for analysis were the same.
The reference laboratory acceptance criteria for MS/MSDs and LCSs
were a relative percent difference less than or equal to 45 with 46 to
124 percent recovery. The acceptance criteria were based on
laboratory historical information. These acceptance criteria are similar
to those of the methods cited in Figure 5-1 .
The LCS/LCSD matrix was Ottawa sand.
The spiking compound for LCSDs was the EDRO standard that
contained even-numbered alkanes from C10 through C40.
The surrogate compound was o-terphenyl. The reference laboratory
acceptance criterion for surrogates was 45 to 143 percent recovery.
The method blank matrix was Ottawa sand. The reference laboratory
acceptance criterion for the method blank was less than or equal to the
project-specific reporting limit.
The extract duplicate was analyzed. The reference laboratory
acceptance criterion for the extract duplicate was a relative percent
difference less than or equal to 45.
Notes:

CCV  =
GC
ICV   =
LCS   =
LCSD =
min   =  Minute
Continuing calibration verification
Gas chromatograph
Initial calibration verification
Laboratory control sample
Laboratory control sample duplicate
mL      =  Milliliter
MS      =  Matrix spike
MSD     =  Matrix spike duplicate
n-Cx     =  Alkane with "x" carbon atoms
ng       =  Nanogram
SW-846  =  Test Methods for Evaluating Solid Waste"
                                                              46

-------
(1)  if a  method  used offered  choices (for example,
SW-846 Method  5035 for GRO extraction states that
samples may be  collected  with or without use of a
preservative solution), (2) if a method used did not provide
specific details (for example, SW-846 Method 5035 for
GRO extraction does not specify how unrepresentative
material should be handled during sample preparation),
or (3) if a modification to a method used was required
in order to meet demonstration objectives (for example,
SW-846 Method 8015B for EDRO analysis states that
quantitation is performed by summing the areas of all
chromatographic peaks eluting between the end of the
1,2,4-trimethylbenzene or n-C10 peak, whichever occurs
later, and the n-octacosane peak; however,  an additional
quantitation  was performed to sum the  areas  of all
chromatographic peaks  eluting from  the  end of the
n-octacosane peak through the tetracontane peak in order
to meet demonstration objectives).

Before analyzing a liquid PE sample, STL Tampa East
added an aliquot of the liquid PE sample to the extraction
solvent used for soil samples. A specified  aliquot of the
liquid PE sample was diluted in methanol for  GRO
analysis and in methylene chloride for EDRO analysis
such that the final volume of the solution for GRO and
EDRO analyses was 5.0 and 1.0 mL, respectively.  The
solution was then analyzed for GRO and EDRO using the
same procedures as are used for soil sample extracts.
                                                   47

-------
                                              Chapter 6
                         Assessment of Reference Method Data Quality
This chapter assesses reference method data quality based
on QC check results and PE sample results. A summary of
reference method data quality is included at the end of this
chapter.

To ensure that the reference method results were of known
and adequate quality, EPA representatives performed a
predemonstration audit and an in-process audit of the
reference laboratory. The predemonstration audit findings
were used in developing the predemonstration design. The
in-process  audit was  performed when  the laboratory
had analyzed a sufficient numb er of demonstration samples
for both GRO and EDRO  and had prepared its first data
package.   During  the  audit,  EPA  representatives
(1) verified that the laboratory had properly implemented
the EPA-approved demonstration plan and (2) performed
a critical review of the first data package.  All issues
identified during the  audit were fully addressed by the
laboratory  before  it  submitted  the subsequent  data
packages to the EPA. The laboratory also addressed issues
identified during  the EPA final review of the  data
packages. Audit findings are summarized in the DER for
the demonstration.

6.1    Quality Control Check Results

This section summarizes QC check results for GRO and
EDRO analyses performed using the reference method.
The QC checks associated with soil sample analyses for
GRO and EDRO included  method blanks, surrogates,
matrix spikes and matrix spike duplicates (MS/MSD), and
laboratory control samples and laboratory control sample
duplicates (LCS/LCSD). In addition, extract duplicates
were analyzed  for soil environmental samples.  The QC
checks associated with liquid PE sample analysis for GRO
included  method  blanks,  surrogates, MS/MSDs, and
LCS/LCSDs.   Because liquid PE sample analyses for
EDRO did not include a preparation step, surrogates,
MS/MSDs, and LCS/LCSDs were not analyzed; however,
an instrument  blank was  analyzed  as a method blank
equivalent. The results for the QC checks were compared
to project-specific acceptance criteria. These criteria were
based on the reference laboratory's historical QC limits
and its  experience in analyzing the predemonstration
investigation samples using the reference method.  The
reference  laboratory's QC limits were  established as
described in   SW-846  and  were within the general
acceptance criteria recommended by SW-846 for organic
analytical methods.

Laboratory duplicates were also analyzed to evaluate the
precision associated with percent moisture analysis of soil
samples.  The acceptance criterion  for the  laboratory
duplicate results was an RPD less than or equal to 20. All
laboratory duplicate results met this criterion. The results
for the laboratory duplicates are not separately  discussed
in this  ITVR  because soil sample  TPH  results  were
compared on a wet weight basis except for those used to
address primary object P4 (effect of soil moisture content).

6.1.1    GRO Analysis

This section summarizes the results for QC checks used by
the reference laboratory during GRO analysis,  including
method blanks, surrogates, MS/MSDs, extract duplicates,
and LCS/LCSDs. A summary of the QC check results is
presented at the end of the section.

Method Blanks

Method blanks were analyzed to verify that steps in the
analytical procedure did not introduce contaminants that
affected analytical results. Ottawa  sand and  deionized
water  were used as method blanks for soil and  liquid
                                                   48

-------
samples, respectively.  These blanks underwent all the
procedures required for sample preparation. The results
for all method blanks met the acceptance criterion of being
less than or equal to the required project-specific reporting
limit (5 mg/kg). Based on method blank results, the GRO
analysis results were considered to be valid.

Surrogates

Each soil investigative and QC sample for GRO analysis
was  spiked with  a surrogate, 4-bromofluorobenzene,
before extraction to determine whether significant matrix
effects existed within the sample and to estimate the
efficiency of analyte recovery during sample preparation
and analysis. A diluted, liquid PE sample was also spiked
with the surrogate during sample preparation.  The initial
surrogate spiking levels for soil and liquid PE samples
were  2 mg/kg  and 40  microgram per  liter (ng/L),
respectively.  The  acceptance   criterion was 39  to
163 percent surrogate recovery. For samples analyzed at
a  dilution  factor  greater than  four,  the surrogate
concentration was diluted to a level below the reference
laboratory's reporting limit for the reference method;
therefore, surrogate recoveries for these samples were not
used to assess impacts on data quality.

A total of 101 surrogate measurements were made during
analysis of environmental and associated QC samples.
Fifty-six of these samples were analyzed  at a dilution
factor less than or equal to four. The surrogate recoveries
for these 56 samples ranged from 43 to 345 percent with a
mean recovery of 150 percent and a median recovery of
136 percent. Because the mean and median recoveries
were greater than 100 percent, an overall positive bias was
indicated.

The surrogate recoveries for 16 of the 56 samples did not
meet the acceptance criterion. In each case, the surrogate
was recovered at a concentration above the upper limit of
the acceptance  criterion.   Examination  of the  gas
chromatograms for  the 16 samples revealed that some
PHCs or naturally occurring interferents present in these
environmental samples  coeluted with  the  surrogate,
resulting in higher surrogate recoveries. Such coelution is
typical for hydrocarbon-containing samples analyzed using
a GC/FHD technique, which was the technique used in the
reference method.   The  surrogate  recoveries  for QC
samples such as method blanks and LCS/LCSDs met the
acceptance criterion, indicating that the laboratory sample
preparation  and  analysis procedures were in  control.
Because  the  coelution  was   observed  only  for
environmental  samples  and  because  the  surrogate
recoveries for QC samples met the acceptance criterion,
the  reference   laboratory  did  not  reanalyze  the
environmental samples with high surrogate recoveries.
Calculations performed to evaluate whether the coelution
resulted  in  underreporting  of  GRO  concentrations
indicated an insignificant impact of less than 3 percent.
Based on the surrogate results for environmental  and
associated QC  samples,  the  GRO analysis results for
environmental samples were considered to be valid.

A total of 42 surrogate measurements were made during
the analysis  of soil PE and associated QC samples.
Thirty-four of these samples were analyzed at a dilution
factor less than or equal to four. The surrogate recoveries
for these 34 samples ranged from 87 to 108 percent with a
mean recovery of 96 percent and a median recovery of
95 percent.  The surrogate recoveries for all 34 samples
met the acceptance criterion.  Based  on the  surrogate
results for soil PE and associated QC samples, the GRO
analysis results for soil PE samples were considered to be
valid.

A total of 37 surrogate measurements were made during
the analysis of liquid PE and associated QC samples.
Six of these samples were analyzed at a dilution factor
less than or equal  to four.  All six samples were  QC
samples (method blanks and LCS/LCSDs). The surrogate
recoveries for these six samples ranged from  81  to
84 percent, indicating a small negative bias. However, the
surrogate recoveries for all six samples met the acceptance
criterion. Based on the surrogate results for liquid PE and
associated QC samples, the GRO analysis results for liquid
PE samples were considered to be valid.

Matrix Spikes and Matrix Spike Duplicates

MS/MSD results were evaluated to determine the accuracy
and precision of the analytical results with respect to the
effects of the sample matrix. For GRO analysis, each soil
sample designated as an MS or MSD was spiked with the
GRO  calibration standard at an initial spiking level of
20 mg/kg.  MS/MSDs were also prepared for liquid PE
samples. Each diluted, liquid PE sample designated as an
MS or MSD was spiked with the GRO calibration standard
at an initial  spiking level of 40 ug/L.   The acceptance
criteria for MS/MSDs were 33 to 115 percent recovery and
an RPD less  than or equal to 25. When the MS/MSD
percent recovery acceptance criterion was not met, instead
                                                    49

-------
of attributing  the  failure  to  meet the criterion  to an
inappropriate  spiking  level,  the  reference  laboratory
respiked the sample at  a more appropriate and practical
spiking level. Information on the selection of the spiking
level and calculation of percent recoveries for MS/MSD
samples is provided below.

According  to  Provost  and Elder (1983), for percent
recovery data to be reliable, spiking levels should be at
least five times the unspiked sample concentration. For the
demonstration, however, a  large number of the unspiked
sample concentrations were expected to range between
1,000 and 10,000 mg/kg, so use of such high spiking levels
was  not practical. Therefore, a target spiking level of 50
to 150  percent of the unspiked sample concentration was
used for the demonstration.  Provost and Elder (1983) also
present an alternate approach for calculating percent
recoveries for MS/MSD samples (100 times the ratio of the
measured concentration  in  a spiked sample to  the
calculated concentration in the sample). However, for the
demonstration, percent  recoveries were calculated using
the traditional approach (100 times the ratio of the amount
recovered to the amount spiked) primarily because the
alternate approach is not commonly used.

For environmental samples, a total of 10 MS/MSD pairs
were analyzed.  Four sample pairs collected in the NEX
Service Station Area were designated as MS/MSDs. The
sample matrix in this area primarily consisted of medium-
grained sand. The percent recoveries for all but one of the
MS/MSD samples  ranged  from 67 to 115 with RPDs
ranging from 2  to 14.  Only one MS sample with a
162  percent recovery did not meet the percent recovery
acceptance  criterion;  however, the  RPD  acceptance
criterion  for the MS/MSD and the percent recovery and
RPD acceptance criteria for the LCS/LCSD  associated
with the analytical batch for this sample were met. Based
on the MS/MSD results, the GRO analysis results for the
NEX Service Station Area samples were considered to be
valid.

Two sample  pairs collected  in  the B-38 Area  were
designated as MS/MSDs. The sample matrix in this area
primarily consisted of sand  and clay.   The  percent
recoveries for the MS/MSD samples ranged from 60 to 94
with RPDs of 1 and 13. Therefore, the percent recoveries
and RPDs for these samples met the acceptance criteria.
Based on the MS/MSD  results, the GRO analysis results
for the  B-38 Area samples were considered to be valid.
Four  sample pairs  collected in the SFT Area were
designated as MS/MSDs.  The sample matrix in this area
primarily consisted of silty clay. The percent recoveries
for the MS/MSD samples ranged from 0 to 127 with RPDs
ranging from 4 to 21.  Of the four sample pairs, two
sample pairs met the percent recovery acceptance criterion,
one sample pair exhibited percent recoveries less than the
lower acceptance limit, and one  sample pair exhibited
percent recoveries greater than the upper acceptance limit.
For the two sample pairs that did not meet the percent
recovery  acceptance  criterion,  the  RPD acceptance
criterion for the MS/MSDs and the percent recovery and
RPD acceptance criteria for the LCS/LCSDs associated
with the analytical batches for these samples were met.
Because of the varied percent recoveries for the MS/MSD
sample pairs, it was not possible to conclude whether the
GRO analysis results for the SFT Area samples had a
negative or positive bias.   Although one-half of the
MS/MSD results did not meet  the percent  recovery
acceptance criterion, the out-of-control situations alone did
not constitute adequate grounds for rejection of any of the
GRO analysis results for the SFT Area samples.  The out-
of-control situations  may have  been associated  with
inadequate spiking levels  (0.7 to 2.8 times the unspiked
sample  concentrations  compared   to  the  minimum
recommended value of 5 times the concentrations).

Three soil PE sample pairs were designated as MS/MSDs.
The sample matrix for these samples consisted of silty
sand.  The percent recoveries for these samples ranged
from 88 to 103 with RPDs ranging  from 4 to 6.  The
percent recoveries and RPDs for these samples met the
acceptance criteria.  Based on the MS/MSD results, the
GRO  analysis  results for the soil  PE  samples were
considered to be valid.

Two liquid PE sample pairs were designated as MS/MSDs.
The percent recoveries for these samples ranged from 77
to 87 with RPDs of 1 and  5.  The percent recoveries and
RPDs for these samples met the acceptance criteria. Based
on the MS/MSD results, the GRO  analysis results for the
liquid PE samples were considered to be valid.

Extract Duplicates

For GRO  analysis, after soil sample extraction, extract
duplicates were  analyzed  to  evaluate  the  precision
associated with the  reference laboratory's  analytical
procedure.  The reference laboratory sampled duplicate
                                                   50

-------
aliquots of the GRO extracts for analysis. The acceptance
criterion for extract duplicate precision was an RPD less
than or equal to 25. Two or more environmental samples
collected in each demonstration area whose samples were
analyzed for GRO (the NEX Service Station, B-38, and
SFT Areas) were designated as extract duplicates. A total
of  10  samples designated as extract duplicates  were
analyzed for GRO. The RPDs for these samples ranged
from 0.5 to 11. Therefore, the RPDs for all the extract
duplicates met the acceptance criterion.   Based on the
extract duplicate results, the GRO analysis results were
considered to be valid.

Laboratory Control Samples and Laboratory
Control Sample Duplicates

For GRO analysis, LCS/LCSD results were evaluated to
determine the  accuracy  and precision associated with
control samples prepared by the reference laboratory. To
generate a soil LCS or LCSD, Ottawa sand was spiked
with the GRO  calibration standard at a spiking level of
20 mg/kg.  To generate an LCS or LCSD for liquid PE
sample analysis, deionized water was spiked with the GRO
calibration standard at a spiking level of 40 ug/L.  The
acceptance criteria for LCS/LCSDs were 33 to 1 ISpercent
recovery and an RPD less than or equal to  25.  The
LCS/LCSD acceptance criteria were based on the reference
laboratory's historical data.

Ten pairs of soil LCS/LCSD samples were prepared and
analyzed. The percent recoveries for these samples ranged
from 87 to 110 with RPDs ranging from 2 to 14. In
addition, two pairs of liquid LCS/LCSD  samples were
prepared and analyzed.  The percent recoveries for these
samples ranged from 91 to 92 with RPDs equal to 0 and 1.
Therefore, the percent recoveries and RPDs for the soil and
liquid LCS/LCSD samples met the acceptance criteria,
indicating that the GRO analysis procedure was in control.
Based on the LCS/LCSD results, the GRO analysis results
were considered to be valid.

Summary of Quality Control Check Results

Table 6-1 summarizes the QC check results  for GRO
analysis. Based on the QC check results, the conclusions
presented below were drawn regarding the accuracy and
precision of GRO analysis results for the demonstration.

The project-specific percent recovery acceptance criteria
were met for most environmental  samples and all PE
samples. As expected, the percent recovery ranges were
broader for the environmental samples than for the PE
samples. As indicated by the mean and median percent
recoveries, the QC check results generally indicated a
slight negative bias  (up  to  20 percent)  in the GRO
concentration measurements; the exceptions  were the
surrogate recoveries for environmental samples and the
LCS/LCSD recoveries for soil PE samples. The observed
bias did not  exceed  the  generally  acceptable bias
(±30 percent) stated in SW-846 for organic analyses and is
typical  for  most  organic  analytical  methods  for
environmental  samples.  Because the percent recovery
ranges were sometimes above and sometimes below 100,
the observed bias did not appear to be systematic.

The project-specific RPD acceptance criterion was met for
all samples. As expected, the RPD range and the mean and
median  RPDs  for MS/MSDs associated with  the soil
environmental  samples were greater than those for other
QC checks and matrixes listed in Table 6-1.  The low
RPDs observed indicated good precision in the GRO
concentration  measurements  made   during  the
demonstration.

6.1.2  EDRO Analysis

This section summarizes the resu Its for QC checks used by
the reference laboratory during EDRO analysis, including
method  and  instrument blanks,  surrogates, MS/MSDs,
extract duplicates, and LCS/LCSDs.  A summary of the
QC check results is presented at the end of the section.

Method and Instrument Blanks

Method and instrument blanks were analyzed to verify that
steps in  the analytical procedures  did  not introduce
contaminants that affected analytical results. Ottawa sand
was used as a method blank for soil samples. The method
blanks underwent all the procedures required for sample
preparation. For liquid PE samples, the extraction solvent
(methylene chloride) was  used as an instrument blank.
The results for all method and instrument blanks met the
acceptance criterion of being less  than or equal to the
required project-specific reporting limit (10 mg/kg). Based
on  the  method  and  instrument blank  results,  the
EDRO analysis results were considered to be valid.
                                                   51

-------
        Table 6-1. Summary of Quality Control Check Results for GRO Analysis
QC Check'
Surrogate
MS/MSD
Extract
duplicate
LCS/LCSD
Matrix
Associated
with QC Check
Soil
environmental
samples
Soil PE
samples
Liquid PE
samples
Soil
environmental
samples
Soil PE
samples
Liquid PE
samples
Soil
environmental
samples
Soil
environmental
andPE
samples
Liquid PE
samples
No. of
Measurements
Used to
Evaluate Data
Quality
56
34
6
20 (10 pairs)
6 (3 pairs)
4 (2 pairs)
10 pairs
10 pairs
2 pairs
Accuracy (Percent Recovery)
Acceptance
Criterion
39 to 163
33 to 115
Actual
Range
43 to 345
87 to 108
81 to 84
0 to 162
88 to 103
77 to 87
No. of
Measurements
Meeting
Acceptance
Criterion
40
34
6
15
6
4
Mean
150
96
83
81
94
83
Median
136
95
84
80
92
85
Not applicable
33 to 115
87(0110
91 to 92
20
4
100
92
100
92
Precision (Relative Percent Difference)
Acceptance
Criterion
Actual
Range
No. of
Measurements
Meeting
Acceptance
Criterion
Mean
Median
Not applicable
s25
1 to 21
4 to 6
1to5
0.5 to 1 1
2 to 14
Oto1
10 pairs
3 pairs
2 pairs
10 pairs
10 pairs
2 pairs
11
5
3
5
6
0.5
12
5
3
4
6
0.5
NJ
        Notes:

        s          =  Less than or equal to
        LCS/LCSD =  Laboratory control sample and laboratory control sample duplicate
        MS/MSD   =  Matrix spike and matrix spike duplicate
        PE        =  Performance evaluation
        QC        =  Quality control
             During the demonstration, 12 method blanks (10 for soil samples and 2 for liquid samples) were analyzed. The method blank results met the project-specific acceptance criteria.

-------
Surrogates

Each soil investigative and QC sample for EDRO analysis
was spiked with a surrogate, o-terphenyl, before extraction
to determine  whether significant matrix effects existed
within the sample and to estimate the efficiency of analyte
recovery during sample preparation and analysis.  For a
30-gram sample, the spike concentration was 3.3 mg/kg.
For samples with higher EDRO concentrations, for which
smaller sample amounts were used during extraction, the
spiking   levels   were  proportionately  higher.    The
acceptance criterion  was 45  to  143 percent surrogate
recovery.  Liquid PE samples for EDRO analysis were not
spiked with a surrogate  because the analysis  did not
include a sample preparation step.

A total of 185 surrogate measurements were made during
analysis of environmental and associated QC samples. Six
of  these samples did not  meet the percent recovery
acceptance criterion.  Four of the six samples  were
environmental samples.  When the reference laboratory
reanalyzed the four samples, the surrogate recoveries for
the samples met the acceptance criterion;  therefore, the
reference laboratory reported  the EDRO concentrations
measured  during  the reanalyses.   The remaining two
samples for which the surrogate recoveries did not meet
the acceptance criterion were LCS/LCSD samples; these
samples had low surrogate recoveries. According  to the
reference laboratory, these low recoveries were due  to the
extracts  going  dry  during the extract  concentration
procedure. Because  two samples were laboratory QC
samples, the reference laboratory reanalyzed them as well
as  all the other  samples in the QC  lot; during  the
reanalyses, all surrogate recoveries met the acceptance
criterion. The surrogate recoveries for all results reported
ranged from 45 to 143 percent with mean and median
recoveries of 77 percent, indicating an overall negative
bias.  The surrogate recoveries for all reported sample
results met the  acceptance criterion.  Based on  the
surrogate results for  environmental and associated QC
samples, the EDRO analysis results were considered to be
valid.

A total of 190 surrogate measurements were made during
analysis of soil PE and associated QC samples.  Five of
these samples  did  not  meet the  percent recovery
acceptance criterion.  In  each case, the surrogate was
recovered  at a concentration below the lower limit of the
acceptance criterion. Three of the five samples were soil
PE  samples,  and the  remaining  two samples  were
LCS/LCSDs.  The reference laboratory reanalyzed the
three soil PE samples and the LCS/LCSD pair as well as
all the other samples in the QC lot associated with the
LCS/LCSDs;  during  the  reanalyses,  all  surrogate
recoveries met the acceptance criterion.  The surrogate
recoveries for all results reported ranged from 46 to
143  percent  with  mean  and  median  recoveries of
76 percent,  indicating  an  overall negative  bias.   The
surrogate recoveries for all reported sample results met the
acceptance criterion.  Based on the surrogate results for
soil PE and  associated QC samples, the EDRO analysis
results were  considered to be valid.

Matrix Spikes and Matrix Spike Duplicates

MS/MSD results were evaluated to determine the accuracy
and precision of the analytical results with respect to the
effects of the sample matrix. For EDRO analysis, each soil
sample designated as an MS or MSD was spiked with the
EDRO calibration standard at an initial spiking level of
50 mg/kg when a 30-gram sample was used during
extraction. The initial spiking levels were proportionately
higher when smaller sample amounts were used during
extraction.  The acceptance criteria for MS/MSDs were
46 to 124 percent recovery and an RPD less than or equal
to 45.  When the MS/MSD percent recovery  acceptance
criterion was not met, instead of attributing the failure to
meet the criterion to an inappropriate spiking level, the
reference laboratory  respiked the samples  at a target
spiking level between 50 and 150 percent of the unspiked
sample concentration. Additional information on spiking
level selection for MS/MSDs is presented in Section 6.1.1.
No MS/MSDs were prepared for liquid PE samples for
EDRO analysis because the analysis did not include a
sample preparation step.

For environmental samples, a total of 13 MS/MSD pairs
were analyzed. Two sample pairs collected  in the FFA
were designated as MS/MSDs. The sample matrix hi this
area primarily consisted of medium-grained sand.  The
percent recoveries for the MS/MSD samples ranged from
0 to 183 with RPDs of 0 and 19. One of the two sample
pairs exhibited percent recoveries less than the lower
acceptance limit. In the second sample pair, one sample
exhibited  a  percent  recovery  less   than  the  lower
acceptance limit, and one sample exhibited a  percent
recovery greater than the upper acceptance limit. For both
sample pairs, the  RPD acceptance criterion for the
MS/MSDs and the percent recovery and RPD acceptance
criteria for the LCS/LCSDs asso ciated with the analytical
                                                   53

-------
batches for these samples were met. Because of the varied
percent recoveries for the MS/MSD sample pairs, it was
not possible to conclude whether  the EDRO analysis
results for the FFA samples had  a negative  or positive
bias.  Although the MS/MSD results did not meet the
percent recovery acceptance criterion, the out-of-control
situations alone did not constitute adequate grounds for
rejection of any of the EDRO analysis results for the FFA
samples.  The out-of-control situations may have been
associated with inadequate spiking levels (0.1 to 0.5 times
the unspiked  sample  concentrations  compared  to the
minimum  recommended   value  of  5  times  the
concentrations).

Four sample pairs collected in the NEX Service Station
Area were designated as MS/MSDs. The sample matrix in
this area primarily consisted of medium-grained sand. The
percent recoveries for the MS/MSD samples ranged from
81 to 109 with RPDs ranging from 4 to 20. The percent
recoveries and RPDs for these samples met the acceptance
criteria.  Based on the MS/MSD  results, the EDRO
analysis results for the NEX Service Station Area samples
were considered to be valid.

One sample pair collected in the PRA was designated as an
MS/MSD.  The sample matrix in this area primarily
consisted of silty sand. The percent recoveries for the
MS/MSD samples were 20 and 80 with an RPD equal to
19. One sample exhibited a percent recovery less than the
lower acceptance limit, whereas the percent recovery for
the other sample met the acceptance criterion.  The RPD
acceptance criterion for the MS/MSD and the percent
recovery  and RPD acceptance criteria for the LCS/LCSD
associated with the analytical batch for this sample pair
were  met.   Although the  percent recoveries for the
MS/MSD sample pair may indicate a negative bias,
because the MS/MSD results for only one sample pan-
were available, it was  not possible to conclude that the
EDRO analysis results for the PRA samples had a negative
bias.  Although one of the percent recoveries for the
MS/MSD did not meet the acceptance criterion, the out-of-
control situation alone did not constitute adequate grounds
for rejection of any of the EDRO analysis results for the
PRA samples. The out-of-control situation may have been
associated with inadequate  spiking levels (0.4 times the
unspiked sample concentration compared to the minimum
recommended value of 5 times the concentration).

Two  sample  pairs  collected  in  the  B-38 Area were
designated as MS/MSDs. The sample matrix in this area
primarily  consisted of sand  and clay.   The percent
recoveries for the MS/MSD samples ranged from 25 to 77
with  RPDs of 6 and 11. Of the two sample pairs,  one
sample pair met the percent recovery acceptance criterion,
and one sample pair exhibited percent recoveries less than
the lower acceptance limit. For the sample pair that did
not meet the percent recovery acceptance criterion, the
RPD  acceptance  criterion  for the MS/MSDs and the
percent recovery  and RPD  acceptance  criteria for the
LCS/LCSDs associated with the analytical batch for the
sample pair were met. Although the percent recoveries for
one MS/MSD sample pair indicated a  negative bias,
because the percent recoveries for the other sample  pan-
were acceptable,  it was not possible to conclude that the
EDRO analysis results for the  B-38 Area samples had a
negative bias. Although one-half of the MS/MSD results
did not meet the percent recovery acceptance criterion, the
out-of-control situations alone did not constitute adequate
grounds for rejection of any of the EDRO analysis results
for the B-38 Area samples. The out-of-control situations
may have been associated with inadequate spiking levels
(1.4 times the unspiked sample concentrations compared
to the minimum  recommended  value  of 5  times the
concentrations).

Four  sample  pairs  collected  in  the SFT  Area were
designated as MS/MSDs. The sample matrix in this area
primarily consisted of silty clay.  The percent recoveries
for the MS/MSD samples ranged from 0 to 223 with RPDs
ranging from 8 to 50.  Of the four sample pairs, three
sample pairs had one sample each that exhibited a percent
recovery  less than the lower acceptance limit and one
sample pair had one sample  that exhibited  a percent
recovery  greater  than the upper acceptance limit.  The
RPD acceptance criterion was met for all but one  of the
MS/MSDs. The percent recovery and RPD acceptance
criteria for the LCS/LCSDs asso ciated with the analytical
batches for these samples were met. Because of the varied
percent recoveries for the MS/MSD sample pairs,  it was
not possible to conclude whether the  EDRO analysis
results for the SFT Area samples had a negative or positive
bias.  Although one-half of the MS/MSD results did not
meet the percent recovery acceptance criterion and one of
the four sample pairs did not meet the RPD acceptance
criterion,   the  out-of-control  situations  alone did  not
constitute  adequate grounds for rejection of any  of the
EDRO analysis results for the SFT Area samples. The
out-of-control situations may have been associated with
inadequate spiking levels (0.4 to 0.7 times the unspiked
                                                   54

-------
sample  concentrations  compared  to  the  minimum
recommended value of 5 times the concentrations).

Five soil PE sample pairs were designated as MS/MSDs.
The sample matrix for these samples primarily consisted of
silty sand.  The percent recoveries for these samples
ranged from 0 to 146 with RPDs ranging from 3 to 17. Of
the five sample pairs, three sample pairs met the percent
recovery acceptance criterion, one sample pair exhibited
percent recoveries less than the lower acceptance limit, and
one sample pair exhibited percent recoveries greater than
the upper acceptance limit. For the two sample pairs that
did not meet the percent recovery acceptance criterion, the
RPD  acceptance criterion for the MS/MSDs and  the
percent  recovery and  RPD acceptance criteria for the
LCS/LCSDs associated with the analytical batches for
these samples were met. Because of the varied percent
recoveries for the  MS/MSD sample pairs, it  was not
possible to conclude whether the EDRO analysis  results
for the soil PE samples had a negative or positive bias.
Although the percent recoveries for two of the five sample
MS/MSD pairs did not meet the acceptance criterion, the
out-of-control situations alone did not constitute adequate
grounds for rejection of any of the EDRO analysis results
for the soil PE samples.

Extract Duplicates

For EDRO analysis, after soil sample extraction, extract
duplicates were analyzed  to  evaluate  the  precision
associated with the  reference  laboratory's analytical
procedure.  The reference laboratory sampled duplicate
aliquots of  the EDRO  extracts  for analysis.    The
acceptance criterion for extract duplicate precision was an
RPD less than or equal to 45. One or more environmental
samples collected  in  each  demonstration area  were
designated as extract duplicates.   A total of 13 samples
designated as extract duplicates were analyzed for EDRO.
The RPDs for these samples ranged from 0 to 11 except for
one extract duplicate pair collected in the SFT Area that
had an RPD equal to 34.  The RPDs for all the extract
duplicates met the  acceptance criterion.  Based on the
extract duplicate results, all EDRO results were considered
to be valid.

Laboratory  Control Samples and Laboratory
Control Sample Duplicates

For EDRO analysis, LCS/LCSD results were evaluated to
determine  the accuracy and precision associated with
control samples prepared by the reference laboratory. To
generate a soil LCS or LCSD, Ottawa sand was spiked
with the EDRO calibration standard at a spiking level of
50 mg/kg. The acceptance criteria for LCS/LCSDs were
46 to 124 percent recovery and an RPD less than or equal
to 45. The LCS/LCSD acceptance criteria were based on
the reference laboratory's historical data. No LCS/LCSDs
were prepared for liquid PE samples for EDRO analysis
because the analysis did not include a sample preparation
step.

Twenty-two pairs of LCS/LCSD samples  were prepared
and analyzed.  The percent recoveries for these samples
ranged from 47 to 88 with RPDs ranging from 0 to 29.
Therefore, the percent recoveries and RPDs for these
samples met the  acceptance criteria, indicating that the
EDRO analysis procedure was in control.  Based on the
LCS/LCSD  results, the  EDRO analysis results  were
considered to be valid.

Summary of Quality Control Check Results

Table 6-2 summarizes the QC check results  for EDRO
analysis.  Based on the QC check results, the conclusions
presented below were drawn regarding the accuracy and
precision of EDRO analysis results for the demonstration.

The project-specific percent recovery acceptance criteria
were met for all  surrogates  and LCS/LCSDs.  About
one-half of the MS/MSDs did not meet the percent
recovery acceptance criterion. As expected, the MS/MSD
percent recovery range was broader for  environmental
samples than for  PE samples.  The mean and median
percent recoveries for all the QC check samples indicated
a negative  bias  (up  to   33  percent) in the  EDRO
concentration measurements. Although the observed bias
was  slightly greater than  the generally acceptable bias
(±30 percent) stated in SW-846 for organic analyses, the
observed recoveries were  not atypical for most organic
analytical methods for environmental samples. Because
the percent  recovery ranges were sometimes  above and
sometimes below 100, the observed bias did not appear to
be systematic.

The project-specific RPD acceptance criterion was met for
all samples  except one environmental MS/MSD sample
pair.   As expected,  the RPD range and  the mean and
median RPDs  for MS/MSDs associated  with the soil
environmental samples were greater than those for other
QC checks  and matrixes listed in Table  6-2.  The low
                                                   55

-------
Table 6-2. Summary of Quality Control Check Results for EDRO Analysis
QC Check"
Surrogate
MS/MSD
Extract
duplicate
LCS/LCSD
Matrix
Associated
with QC Check
Soil
environmental
samples
Soil PE
samples
Soil
environmental
samples
Soil PE
samples
Soil
environmental
samples
Soil
environmental
andPE
samples
No. of
Measurements
Used to
Evaluate Data
Quality
179
185
26 (13 pairs)
10 (5 pairs)
1 3 pairs
44 (22 pairs)
Accuracy (Percent Recovery)
Acceptance
Criterion
45 to 143
46 to 124
Actual
Range
45 to 143
46 to 143
0 to 223
0 to 146
No. of
Measurements
Meeting
Acceptance
Criterion
179
185
14
6
Mean
77
76
67
75
Median
77
76
79
78
Not applicable
46 to 124
47 to 88
44
77
80
Precision (Relative Percent Difference)
Acceptance
Criterion
Actual
Range
No. of
Measurements
Meeting
Acceptance
Criterion
Mean
Median
Not applicable
s45
OtoSO
3 to 17
Oto34
Oto29
12 pairs
*!
5 pairs
13 pairs
22 pairs
17
7
6
6
'16
,- 4
2
5
Notes:

s         =  Less than or equal to
LCS/LCSD =  Laboratory control sample and laboratory control sample duplicate
MS/MSD  =  Matrix spike and matrix spike duplicate
PE       =  Performance evaluation
QC       =  Quality control
     During the demonstration, 22 method blanks for soil samples and 2 instrument blanks for liquid samples were analyzed. The blank results met the project-specific acceptance criteria.

-------
RPDs observed indicated good precision in the EDRO
concentration   measurements   made  during   the
demonstration.

6.2     Selected Performance Evaluation Sample
        Results

Soil and liquid PE samples were analyzed during the
demonstration to  document the reference method's
performance  in analyzing  samples  prepared under
controlled conditions. The PE sample results coupled with
the QC check results were used to establish the reference
method's performance  in such  a  way  that the overall
assessment of  the reference method  would  support
interpretation of the PetroFLAG™ System's performance,
which is discussed in Chapter 7.  Soil PE samples were
prepared by adding weathered gasoline or diesel to Ottawa
sand or processed garden soil.   For each  sample, an
amount of weathered gasoline or diesel was added to the
sample matrix in order to prepare a PE sample with a low
(less than 100 mg/kg), medium (100 to 1,000 mg/kg), or
high (greater than  1,000  mg/kg) TPH concentration.
Liquid PE samples consisted of neat materials. Triplicate
samples of each type of PE sample were  analyzed by the
reference laboratory except for  the  low-concentration-
range PE samples, for which seven replicate samples were
analyzed.

As described in Section 4.2, some  PE  samples  also
contained interferents.  Section 6.2 does not discuss the
reference method  results  for PE samples containing
interferents because the  results  address  a  specific
demonstration objective. To facilitate comparisons, the
reference   method  results  that  directly   address
demonstration objectives are  discussed along with the
PetroFLAG™ System results in Chapter 7. Section 6.2
presents a comparison of the reference  method's mean
TPH results for selected PE samples to the certified values
and performance acceptance limits provided by ERA, a
commercial PE sample provider that prepared the PE
samples for the demonstration. Although the reference
laboratory reported sample results  for GRO  and EDRO
analyses separately, because ERA provided certified values
and performance acceptance limits, the reference method's
mean TPH results (GRO plus EDRO analysis results) were
used for comparison.

For soil samples  containing weathered gasoline,  the
certified values used for comparison to the reference
method  results were based on mean TPH results for
triplicate samples analyzed by  ERA using a GCMD
method. ERA extracted the PE samples on the day that PE
samples  were  shipped  to the  Navy  BVC  site for
distribution to the reference laboratory and developers.
The reference laboratory completed methanol extraction of
the demonstration samples within 2 days of receiving
them. Between 5 and 7 days elapsed between the time that
ERA and the time that the reference laboratory completed
methanol extractions of the demonstration samples. The
difference in extraction times is not believed to have had
a significant effect on the reference method's T^H results
because the samples  for GRO analysis were containerized
hi EPA-approved EnCores and were stored at 4 ± 2 °C to
minimize volatilization. After methanol extraction  of the
PE samples,  both ERA and the reference laboratory
analyzed  the  sample  extracts  within the appropriate
holding times for the extracts.

For soil samples containing diesel, the certified values
were established by calculating  the TPH concentrations
based  on the amounts  of diesel spiked into known
quantities of soil; these  samples were not analyzed by
ERA.  Similarly, the densities of the neat materials were
used as the certified values for the liquid PE samples.

The performance acceptance  limits for soil PE samples
were based on ERA's historical data on percent recoveries
and RSDs from multiple laboratories that had  analyzed
similarly prepared ERA PE samples using a GC method.
The performance acceptance limits were determined at the
95 percent confidence level using Equation 6-1.

Performance Acceptance Limits = Certified Value x
(Average Percent Recovery ± 2(Average RSD))     (6-1)

According to SW-846, the 95 percent confidence  limits
should be treated as warning limits, whereas the 99 percent
confidence limits should be treated as control limits. The
99 percent confidence limits are calculated by using three
tunes the average RSD in  Equation 6-1 instead of two
times the average RSD.

When  establishing the performance  acceptance limits,
ERA did not account  for variables among the multiple
laboratories, such  as different extraction and analytical
methods,  calibration  procedures, and chromatogram
integration ranges (beginning and end points).  For this
reason, the performance acceptance limits should be used
with caution.
                                                   57

-------
Performance acceptance limits for liquid PE samples were
not  available because  ERA  did not have historical
information on percent recoveries  and RSDs for the neat
materials used in the demonstration.

Table 6-3 presents the PE sample types, TPH concentration
ranges, performance acceptance limits, certified values,
reference method mean TPH concentrations, and ratios of
reference method mean TPH concentrations to certified
values.

In addition to the samples listed in Table 6-3, three blank
soil PE samples (processed garden soil) were analyzed to
determine whether the soil PE sample matrix contained a
significant TPH concentration. Reference method GRO
results for all triplicate samples were below the reporting
limit of 0.54 mg/kg.  Reference method  EDRO results
were calculated by  adding  the results for DRO and oil
range organics (ORO) analyses. For one of the triplicate
samples, both the DRO and ORO results were below the
reporting limits of 4.61 and 5.10 mg/kg, respectively. For
the remaining two triplicates, the DRO and ORO results
were 1.5 times greater than the reporting limits. Based on
the TPH  concentrations  in  the medium-  and  high-
concentration-range soil PE samples listed in Table 6-3,
the contribution of the processed garden soil to the TPH
concentrations was  insignificant and ranged between
0.5 and 5 percent.

The reference method's mean  TPH results for the soil PE
samples listed in Table 6-3 were within the performance
acceptance limits except for the low-concentration-range
diesel samples. For the low-range diesel samples, (1) the
individual TPH concentrations for all seven replicates were
less than the lower  performance acceptance limit and
(2) the upper 95 percent confidence limit for TPH results
was also less than the lower performance acceptance limit.
However, the reference method mean and individual TPH
results for the low-range diesel samples were within the
99  percent confidence interval of 10.8 to 54.6 mg/kg,
indicating that the reference method results met the control
limits but not the warning limits.  Collectively,  these
observations   indicated  a   negative  bias  in  TPH
measurements for low-range diesel samples.
Table 6-3. Comparison of Soil and Liquid Performance Evaluation Sample Results
TPH Performance
Concentration Acceptance Limits
Sample Type" Range (mg/kg) Certified Value
Soil Sample (Ottawa Sand) ~,
Diesel
-
Low
" " , '"i' *• V.?,
18.1 to 47.4
Soil Samples (Processed Garden'Soit)' *„__".; - • ~ < , *- •" f'"^
Weathered gasoline

Weathered gasoline at
16 percent moisture
Diesel

Diesel at less than 1 percent
moisture
Medium
High
High
Medium
High
High
389 to 1,548
1,1 10 to 4,430
992 to 3,950
220 to 577
1,900 to 4,980
2, 100 to 5,490
'*',- *!L-'';ii*1'
37.3 mg/kg
i'V-'.-V;*"*.
1,090 mg/kg
3,120 mg/kg
2,780 mg/kg
454 mg/kg
3,920 mg/kg
4,320 mg/kg
Reference Method Reference Method Mean
Mean TPH TPH Concentration/
Concentration Certified Value (percent)
"V- ' H" "~ \ '" „ '
15.4 mg/kg
*-, **''' "* ^J
41
' !/:- "" ^T *&*
65
65
69
56
69
67
Liquid Samples " ,/X, *f '' ~ **^, $ . ' - J • ~, »,' '"^V"^,, f' $j? f^'^^j, ^ *> r fg"^> "fj"^ff ^^C'" °A
Weathered gasoline
Diesel
High
High
Not available
Not available
814,100 mg/L
851 ,900 mg/L
648,000 mg/L
1,090,000 mg/L
80
128
Notes:

mg/kg =  Milligram per kilogram
mg/L  =  Milligram per liter

a    Soil samples were prepared at 9 percent moisture unless stated otherwise.
                                                    58

-------
As noted above, Table 6-3 presents ratios of the reference
method mean TPH concentrations to the certified values
for PE samples.   The ratios for weathered gasoline-
containing soil samples ranged from 65 to 69 percent and
did not appear  to depend on whether the samples were
medium-  or high-range  samples.   The  ratio for neat,
weathered gasoline (liquid sample) was 80 percent, which
was 11 to 15 percentage points greater than the ratios for
the soil samples.  The difference in the ratios  may be
attributed to  (1) potential loss of volatiles during  soil
sample transport and  storage and  during soil  sample
handling when extractions were performed and (2) lower
analyte recovery during soil sample  extraction. The  less
than 100 percent ratios  observed indicated a negative bias
in  TPH  measurement  for  soil and liquid  samples
containing weathered gasoline. The observed bias for the
liquid samples did not exceed the generally acceptable bias
(±30 percent) stated in SW-846 for most organic analyses.
However, the bias for soil  samples exceeded the acceptable
bias by up to 5 percentage points.

The ratios for diesel-containing soil samples ranged from
41 to 69 percent and  increased with increases in the TPH
concentration range.   The  ratio for neat diesel (liquid
sample) was 128 percent, which was substantially greater
than the ratios for soil samples. Collectively, the negative
bias observed for soil samples and the positive bias
observed  for liquid samples indicated  a low  analyte
recovery during soil sample extraction because the soil and
liquid samples were analyzed using the same calibration
procedures but only the soil samples required  extraction
before analysis. The extraction procedure used during the
demonstration is an EPA-approved method that is widely
used by  commercial laboratories in the  United States.
Details  on the extraction  procedure are presented in
Table 5-3 of this ITVR.

The positive  bias  observed for liquid samples did not
exceed the generally acceptable bias stated in SW-846.
The negative  bias observed for high-concentration-range
soil samples exceeded the acceptable bias by an average of
2 percentage points. However, the negative bias observed
for low- and  medium-range  samples  exceeded  the
acceptable  bias  by  29  and  14  percentage  points,
respectively, indicating a negative bias.

Because the reference method results exhibited a negative
bias for soil PE samples when compared to ERA-certified
values, ERA's historical data on percent recoveries and
RSDs  from  multiple  laboratories   were  examined.
Table 6-4 compares ERA's historical percent recoveries
and RSDs to the reference method percent recoveries and
RSDs obtained during the demonstration. Table 6-4 shows
that ERA's historical recoveries also exhibited a negative
bias for all sample types except weathered  gasoline in
water and that the reference method recoveries were less
than ERA's historical recoveries for all  sample types
except diesel in water.  The ratios of reference method
mean recoveries to ERA historical mean recoveries for
weathered gasoline-containing samples indicated that the
reference method TPH results were 26 percent less than
ERA's  historical  recoveries.   The  reference  method
recoveries for diesel-containing (1)  soil  samples were
34 percent less than the ERA historical recoveries and
(2) water samples were 63 percent greater than the ERA
historical recoveries.   In all cases,  the RSDs for the
reference method were  significantly  lower than ERA's
historical RSDs, indicating that the  reference method
Table 6-4. Comparison of Environmental Resource Associates Historical Results to Reference Method Results
                             ERA Historical Results
                                                                         Reference Method Results
Sample Type
Weathered gasoline in soil
Diesel in soil
Weathered gasoline in water
Diesel in water
Mean
Recovery
(percent)
88.7
87.7
109
78.5
Mean Relative
Standard Deviation
(percent)
26.5
19.6
22.0
22.8
Mean Reference Method Mean
Recovery3 Recovery/ERA Historical
(percent) Mean Recovery (percent)
66
58
80
128
75
66
73
163
Mean Relative
Standard Deviation3
(percent)
7
9
5
6
Notes:

ERA  = Environmental Resource Associates

'    The reference method mean recovery and mean relative standard deviation were based on recoveries and relative standard deviations observed
    for all concentration ranges for a given type of performance evaluation sample.
                                                     59

-------
achieved significantly  greater precision.  The greater
precision observed for  the reference method during the
demonstration may be  associated with the fact that the
reference method was implemented by a single laboratory,
whereas ERA's  historical RSDs were based on results
obtained from multiple laboratories that  may have used
different analytical protocols.

In summary, compared to ERA-certified values, the TPH
results for all PE sample types except neat  diesel exhibited
a negative bias to a varying degree;  the TPH results for
neat diesel exhibited a  positive bias of 28 percent.  For
weathered gasoline-containing soil samples, the bias was
relatively independent of the TPH concentration range and
exceeded the generally acceptable bias stated in SW-846
by up to 5 percentage points. For neat gasoline samples,
the bias did not exceed the  acceptable bias.  For diesel-
containing soil samples, the bias increased with decreases
in the TPH concentration range, and the bias for low-,
medium-,   and  high-range   samples  exceeded  the
acceptable  bias  by 29,  14, and  2  percentage points,
respectively.   For neat  diesel samples, the observed
positive bias did not exceed the acceptable bias. The low
RSDs (5,to 9 percent)  associated  with the reference
method indicated good precision in analyzing both soil and
liquid samples.  Collectively, these observations suggest
that caution should be exercised during  comparisons of
PetroFLAG™ System and reference method results for
low- and medium-range soil samples containing diesel.

6.3    Data Quality

Based on the reference method's performance in analyzing
the QC check samples and selected PE  samples, the
reference  method results were considered to  be  of
adequate quality for the  following  reasons: (1)  the
reference  method was  implemented with  acceptable
accuracy (±30 percent) for all samples except low- and
medium-concentration-range  soil  samples  containing
diesel, which made up only 13 percent of the total number
of samples analyzed during the demonstration, and (2) the
reference method was implemented with good precision
for all samples (the overall RPD range was 0 to 17). The
reference method results generally exhibited a negative
bias.  However, the bias was considered to be significant
primarily  for low-  and  medium-range soil samples
containing diesel because the bias exceeded the generally
acceptable  bias  of ±30 percent stated  in SW-846  by
29 percentage points for low-range and 14  percentage
points for medium-range samples. The reference method
recoveries  observed were typical  of  the  recoveries
obtained by   most  organic analytical methods  for
environmental samples.
                                                    60

-------
                                               Chapter 7
                           Performance of the PetroFLAG™ System
To verify a wide range of performance attributes, the
demonstration had both primary and secondary objectives.
Primary  objectives were  critical  to  the technology
evaluation  and were intended to produce quantitative
results regarding a technology's performance. Secondary
objectives provided information that was useful but did not
necessarily  produce quantitative  results regarding  a
technology's performance.  This chapter discusses the
performance of the PetroFLAG™ System based on the
primary objectives (excluding costs associated with TPH
measurement) and secondary objectives. Costs associated
with  TPH  measurement (primary  objective  P6) are
presented in Chapter 8. The demonstration results for both
the primary and secondary objectives are  summarized in
Chapter 9.

7.1    Primary Objectives

This section  discusses the performance  results for the
PetroFLAG™ System based on primary objectives PI
through P5, which are listed below.

PI. Determine the MDL

P2. Evaluate  the  accuracy  and  precision  of  TPH
    measurement for  a  variety of  contaminated  soil
    samples

P3. Evaluate  the   effect  of   interferents   on  TPH
    measurement

P4. Evaluate the effect of soil moisture content on TPH
    measurement

P5. Measure the time required for TPH measurement

To address  primary objectives PI through P5, samples
were collected from four different sampling areas.  In
addition, soil and liquid PE samples were prepared and
distributed to Dexsil and the reference laboratory; of these
samples, Dexsil chose not to analyze low- and medium-
concentration-range weathered gasoline soil PE samples
because according  to Dexsil, the PetroFLAG™ System
was not sensitive to weathered gasoline concentrations less
than 1,000 mg/kg. In addition, Dexsil chose not to analyze
soil samples collected in the PRA because Dexsil believed
that the  natural organic material  in the  area would
adversely impact the  PetroFLAG™ System's ability to
accurately measure TPH.  The numbers and types of
environmental samples collected in each sampling area and
the numbers  and  types of PE samples prepared are
discussed in Chapter 4.

Primary objectives  PI through P4 were addressed using
statistical and nonstatistical approaches, as  appropriate.
The statistical tests  performed to address these objectives
are illustrated in the flow diagram in Figure 7-1. Before a
parametric test was  performed, the Wilk-Shapiro test was
used to determine  whether the PetroFLAG™ System
results and reference method results or, when  appropriate,
their  differences   were  normally  distributed  at  a
significance level  of 5  percent. If the results or their
differences   were   not  normally  distributed,  the
Wilk-Shapiro test was performed on transformed results
(for example,  logarithm and square root transformations)
to verify the  normality assumption.  If the normality
assumption was not met, a  nonparametric test was
performed.  Nonparametric tests are not as powerful as
parametric tests because the nonparametric  tests do not
account for the magnitude of the  difference between
sample results. Despite this limitation, when the normality
assumption was not met, performing a nonparametric test
was considered to be a better alternative than performing
no statistical comparison.

For  the  PetroFLAG™   System,   when   the  TPH
concentration in a given sample was reported as below the
reporting limit, one-half the reporting limit was used as the
                                                   61

-------
0)
£   
-------
TPH concentration for that sample, as is commonly done,
so that necessary calculations could be performed without
rejecting the data. The same approach was used for the
reference method except that the appropriate reporting
limits were used  in calculating the TPH concentration
depending  on  which TPH  measurement components
(GRO, DRO, and  ORO) were reported at concentrations
below the  reporting  limits.    In  addition,  for  the
PetroFLAG™ System, when the TPH concentration in a
given sample was reported as a "greater than" value, the
specified value was used as the TPH concentration for that
sample so that necessary calculations could be performed
without rejecting  the data.  Caution was exercised to
ensure that these necessary data manipulations did not alter
the conclusions.

The  reference method  GRO results were adjusted for
solvent dilution associated with the soil  sample moisture
content because the method required use of methanol, a
water-miscible  solvent, for extraction  of soil samples.
Although the PetroFLAG™ System also required use of a
water-miscible solvent for extraction of soil samples, the
device's TPH results were not adjusted at the request of
Dexsil.  In addition, based on discussions with Dexsil, a
given TPH  result for  the PetroFLAG™ System was
rounded to the  nearest integer when it was less than or
equal to 99 mg/kg or 99 mg/L and to the  nearest 10 when
it was greater than 99 mg/kg or 99 mg/L.  Similarly, based
on discussions  with the reference laboratory,  all TPH
results for the reference method were rounded to three
significant figures.

7.1.1  Primary Objective PI: Method Detection
       Limit

To determine the MDLs for the PetroFLAG™ System and
reference method, both Dexsil and the reference laboratory
analyzed seven low-concentration-range soil PE samples
containing diesel.  Because the PetroFLAG™ System and
reference method results were both normally distributed,
the MDLs for the soil PE samples containing diesel were
calculated using  Equation  7-1  (40  CFR  Part  136,
Appendix B, Revision 1.1.1). An MDL thus calculated is
influenced by TPH concentrations because the standard
deviation will likely  decrease with a decrease in TPH
concentrations.  As a result, the MDL will be lower when
low-range samples are used  for  MDL determination.
Despite this limitation, Equation 7-1 is commonly used and
provides a reasonable estimate of the MDL.
                              1-0=0.99)
                                               (7-1)
where
    S  =  Standard deviation of replicate TPH results
    t,
     n-l,l-o=0.99)  _
                  Student's t-value appropriate for a
                  99  percent confidence level  and a
                  standard deviation estimate with n-1
                  degrees of freedom  (3.143 for n = 7
                  replicates)
Because GRO compounds were not expected to be present
in the soil PE samples containing diesel, the  reference
laboratory performed only  EDRO  analysis  of these
samples and  reported the sums of the DRO and ORO
concentrations as the TPH results.  The PetroFLAG™
System and reference method results for these samples are
presented in Table 7-1.
Table 7-1.  TPH Results for Low-Concentration-Range Diesel Soil
         Performance Evaluation Samples
PetroFLAG™ System Result
(mg/kg)
Reference Method Result (mg/kg)







MDL
72
61
75
68
76
80
77
20
12.0
16.5
13.7
16.4
17.4
17.2
14.8
6.32
Notes:
MDL   =  Method detection limit
mg/kg  =  Milligram per kilogram
Based on the TPH results for the low-concentration-range
diesel soil PE samples, the MDLs were determined to be
20 and  6.32 mg/kg for  the PetroFLAG™ System and
reference  method,  respectively.   Because  the  ORO
concentrations in  all these samples  were below  the
reference  laboratory's  estimated   reporting  limit
(5.1 mg/kg), the MDL for the reference method was also
calculated using only DRO results.  The MDL for the
reference  method   based on  the DRO  results was
6.29 mg/kg, whereas the MDL for the reference method
based on the EDRO results was 6.32 mg/kg, indicating that
the ORO concentrations below the reporting limit did not
impact the MDL for the reference method.
                                                    63

-------
The MDL of 20 mg/kg for the PetroFLAG™ System was
greater than the MDL of 13 mg/kg claimed by Dexsil for
soil samples containing diesel. The MDL of 6.32 mg/kg
for the reference method compared well with the MDL of
4.72 mg/kg published hi SW-846 Method 8015C for diesel
samples extracted  using a pressurized fluid  extraction
method and analyzed for DRO.

7.7.2  Primary Objective P2: Accuracy and
       Precision

This section discusses  the ability of the PetroFLAG™
System to  accurately and precisely measure  TPH
concentrations in a variety of contaminated soils.  The
PetroFLAG™ System TPH results were compared to the
reference method TPH results. Accuracy and precision are
discussed in Sections 7.1.2.1 and 7.1.2.2, respectively.

7.1.2.1     Accuracy

The accuracy of PetroFLAG™ System measurement of
TPH was assessed by determining

•   Whether   the   conclusion   reached  using   the
    PetroFLAG™ System agreed  with that reached using
    the reference method regarding whether the TPH
    concentration in a given sampling  area or soil type
    exceeded a specified action level

    Whether the PetroFLAG™ System results were biased
    high or low compared to the reference method results

•   Whether the PetroFLAG™  System  results  were
    different from  the reference method results at a
    statistical significance level  of  5  percent when a
    pairwise comparison was made

•   Whether a significant correlation existed between the
    PetroFLAG™ System and reference method results

During examination of these four factors, the data quality
of the reference method and PetroFLAG™ System TPH
results was considered.  For example, as discussed in
Chapter 6, the reference method generally exhibited a low
bias. However, the bias observed for all samples except
low- and medium-concentration-range diesel soil samples
did  not   exceed   the  generally  acceptable  bias  of
±30 percent stated in SW-846  for organic  analyses.
Therefore, caution was exercised during comparison of the
PetroFLAG™  System  and reference  method results,
particularly those for low- and medium-range diesel soil
samples. Caution was also exercised during interpretation
of statistical test conclusions drawn based  on a small
number of samples. For example, only three samples were
used for each type of PE sample except the low-range
diesel  samples;  the small number of samples used
increased the probability that the results being compared
would be found to be statistically the same.

As discussed in Chapter 2, during the demonstration, the
PetroFLAG™ System TPH results were determined using
response factors that Dexsil considered  to be appropriate
for the PHCs suspected to be present in the demonstration
samples. The two response factors used by Dexsil during
the demonstration were 2 (for weathered gasoline) and 6
(for No. 6  fuel oil).  Table 7-2  presents the  response
factors used for  both soil  and liquid demonstration
samples.  Although no demonstration samples contained
No. 6 fuel oil, Dexsil used the response factor for No. 6
fuel oil instead of the  diesel response factor so that the
PetroFLAG™ System TPH results would agree with those
of the  reference method, which was known to have a
negative bias.   Based on the information provided to
Dexsil  regarding  the  types of PHCs  suspected  to be
present hi the samples provided, the response factors used
seemed to be appropriate for some samples (for example,
FFA samples) but not for  others (samples containing
MTBE and PCE). All PetroFLAG™ System TPH results
presented in this ITVR are based on the response factors
used by  Dexsil during the demonstration.   Dexsil's
perspective  on the selection  of response factors  is
presented in the appendix.

The following sections discuss how the PetroFLAG™
System results compared with the reference method results
by addressing each of the four factors identified above.

Action Level Conclusions

Table 7-3 compares action level conclusions reached using
the PetroFLAG™ System and reference method results for
environmental and soil PE samples.  Section 4.2 of this
ITVR explains how the action levels were selected for the
demonstration.   Of the environmental  samples,  the
percentage of samples  for which the conclusions agreed
ranged from 60 to 88. Of the PE samples, the percentage
of samples for which the conclusions agreed ranged from
33 to 100.  Overall, the  conclusions were the  same for
73 percent of the samples.
                                                    64

-------
Table 7-2. PetroFLAG™ System Response Factors
Sampling Area or Sample Type
Fuel Farm Area
Naval Exchange Service Station Area
B-38 Area
Slop Fill Tank Area
Soil performance evaluation samples
Liquid performance evaluation samples
Contamination Type
Weathered diesel
Weathered gasoline
Fresh gasoline and diesel or weathered gasoline and trace
amounts of lubricating oil
Slightly weathered gasoline, kerosene, JP-5, and diesel
Weathered gasoline .
Weathered gasoline with interferents
Diesel
Diesel with interferents
Blank
Blank with humic acid
Weathered gasoline
Diesel
Methyl-tert-butyl ether
Tetrachloroethene
Stoddard solvent
Turpentine
1 ,2,4-Trichlorobenzene
Response Factor Used'
6
6
6
2
2
2
6
6
2
6
6
6
6
6
6
6
6
Note:
    The response factors of 2 and 6 correspond to weathered gasoline and No. 6 fuel oil, respectively, for the PetroFLAG™ System; the response factor
    for diesel is 5.
The  least agreement observed for the environmental
samples was for those from the FFA.  This observation
appeared to be associated with the significant negative bias
observed for the reference method TPH results for low-
concentration-range diesel soil PE samples discussed in
Chapter 6.

The  lack of agreement observed  for  two types  of  PE
samples, blank soil and blank soil containing humic acid,
appeared to be associated with the background reading, or
noise, that the PetroFLAG™ System measured in blank
soil,  which was at least one  order of magnitude greater
than  the concentrations reported by ERA, the PE sample
provider. According to ERA, the blank soil samples used
during the demonstration contain (1) less than 16 mg/kg of
PHCs based on  an infrared  analysis and (2)  less than
2 mg/kg of GRO and less than 10 mg/kg of DRO using a
GC analysis (Tetra Tech 2001). Of the soil PE samples
containing high-concentration-range weathered gasoline,
the reference method results for the samples whose action
level conclusions did not agree were within 5 percent of
the action level, making it difficult to accurately assess
whether the sample concentrations were above or below
the action level.  The lack of agreement observed for soil
PE samples containing low-range diesel appeared to be
associated  with the significant negative bias of the
reference method TPH results for such samples discussed
in Chapter 6.

When the action level conclusions did not agree, the TPH
results were further  interpreted  to assess whether the
PetroFLAG™ System conclusion was conservative. The
PetroFLAG™ System conclusion was considered to be
conservative when the PetroFLAG™ System result was
above the action level and the reference method result was
below the action level. A regulatory agency would likely
favor a field measurement device whose results are
conservative; however, the party responsible for a site
cleanup  might  not  favor  a  device  that  is  overly
conservative  because  of  the  cost  associated with
unnecessary cleanup.  Of the 26 PetroFLAG™ System
action  level conclusions that did  not agree with the
reference method conclusions, all 26 conclusions were
conservative.
                                                    65

-------
Table 7-3. Action Level Conclusions
Sampling Area or Sample Type
Fuel Farm Area
Naval Exchange Service Station Area
B-38 Area
Slop Fill Tank Area
PE sample
PE sample
Soil PE sample
containing
weathered
gasoline in
Soil PE sample
containing
diesel in
Blank soil
(9 percent moisture content)
Blank soil and humic acid
(9 percent moisture content)
High-concentration range
(9 percent moisture content)
High-concentration range
(16 percent moisture content)
Low-concentration range
(9 percent moisture content)
Medium-concentration range
(9 percent moisture content)
High-concentration range (less
than 1 percent moisture
content)
High-concentration range
(9 percent moisture content)
Action
Level
(mg/kg)
100
50
100
500
10
200
2,000
2,000
15
200
2,000
2,000
Total
Total Number
of Samples
Analyzed
10
20
8
28
3
6
3
3
7
3
3
3
97
Percentage of Samples for
Which PetroFLAG™ System
and Reference Method
Conclusions Agreed
60
75
88
78
67
50
67
33
57
100
100
100
73
When Conclusions Did Not Agree,
Were PetroFLAG™ System
Conclusions Conservative or Not
Conservative?"
Conservative
•L^^^^^^^^aaMg^^aHH^^^^^s
HRB^B^^HI
HB^B^^^^^^^^^^^^B^KJ^^^BHjB!BiBBBB9gj
R^^^^^^^^^^^^^^^^^^^^^^^^^^H^^H^Ugl
BIB
^••IlliiniH8111'1''1
•S^^K^^BS^^^^^BHiB^^^^^^HKSil
Notes:

mg/kg =  Milligram per kilogram
PE   =  Performance evaluation

"   A conclusion was considered to be conservative when the PetroFLAG ™ System result was above the action level and the reference method result
    was below the action level. A conservative conclusion may also be viewed as a false positive.
Measurement Bias

To determine the measurement bias, the ratios of the
PetroFLAG™ System TPH results to the reference method
TPH results were calculated. The observed bias values
were grouped to identify the number of PetroFLAG™
System results within the following ranges of the reference
method results: (1) greater than 0 to 30 percent, (2)greater
than 30 to 50 percent, and (3) greater than 50 percent.

Figure 7-2 shows the distribution of measurement bias for
the environmental samples.  Of the four sampling areas,
the best agreement between the PetroFLAG™ System and
reference  method results  was  observed for  samples
collected  from  the B-38  Area;  for  these  samples,
50 percent of the PetroFLAG™ System results were within
50  percent  of  the  reference  method  results.   The
percentages of samples whose device results  were within
50  percent  of the  reference  method  results   were
significantly lower for the FF A (10 percent), NEX Service
Station  Area (20 percent), and  SFT Area (7 percent).
Measurement bias for the NEX Service Station Area was
assessed using 3,330 mg/kg as the TPH result for the two
PetroFLAG™ System results reported  as  greater than
3,330 mg/kg (see Table 7-4). As a result, the percent bias
for the two samples may have actually been higher than
calculated (3 0 percent). Excluding B-3 8 Area samples, the
PetroFLAG™ System results for environmental samples
were biased high except for 2 of 58 samples (3 percent).
The high bias of these results could not be explained based
solely on the negative bias associated  with reference
method TPH results.  For example, for the SFT  Area
samples, some of the high bias observed was attributable
to  the  response   factor   used  in  calculating  the
PetroFLAG™ System TPH results. The low bias observed
for the B-38  Area samples was  attributable to both the
response factor used and the PetroFLAG™ System's
lack of sensitivity to low-concentration-range weathered
                                                     66

-------
                      Fuel Farm Area
                  Total number of samples: 10
             X)to30
                           >30to50
                        Bias, percent
                                            >50
                         8-38 Area
                  Total number of samples: 8
                                                                      >0to30
                           >30to50
                         Bias, percent
>50
             Naval Exchange Service Station Area
                 Total number of samples: 20
             >0to30
                           >30to50
                         Bias, percent
                                           >50
                                                              24

                                                            |20
                                                            8 16
                                                          S. £12
                                                          «•- ^
                                                          o c
                                                          Si  8
                                                          I >
                                                          I W  4
                     Slop Fill Tank Area
                 Total number of samples: 28
                                                                     >0to30
                          >30 to 50
                         Bias, percent
 Notes:   > -  Greater than; •  PetroFLAG™ system result biased low compared to reference method result;
         biased high compared to reference method result

Figure 7-2. Measurement bias for environmental samples.
                                PetroFLAG™ system result
gasoline.  However, the high bias observed for the NEX
Service Station Area samples could not be explained.

Figure 7-3 shows the distribution of measurement bias for
selected soil PE samples.  Of the four sets of samples
containing PHCs and the one set of blank samples, good
agreement  between  the  PetroFLAG™  System  and
reference  method  results was observed for  the  high-
concentration-range  weathered  gasoline  and  diesel
samples.  All PetroFLAG™ System results for the high-
range  diesel samples  were within  30  percent  of the
reference  method  results.   The PetroFLAG™  System
results for three of the six high-range weathered gasoline
samples were within 50 percent of the reference method
results. The remaining PetroFLAG™ System results for
soil PE samples, including blank  samples, were biased
high by greater than 50 percent.  The high bias observed
for the blank PE samples may have been associated with
the background reading, or noise, that the PetroFLAG™
System measured in blank soil as discussed above under
"Action Level Conclusions."  The high bias observed for
low-range diesel samples appeared to be associated with
the significant negative bias of the reference method TPH
results for such samples.  The high bias of greater than
50 percent observed for the remaining samples cannot be
explained based solely on the negative bias associated with
the reference method.

Pairwise Comparison of TPH Results

To evaluate whether a statistically significant difference
existed between the PetroFLAG™ System and reference
                                                      67

-------
Table 7-4. Statistical Comparison of PetroFLAG™ System and Reference Method TPH Results for Environmental Samples
Sampling Area
Fuel Farm Area
Naval Exchange
Service Station
Area
B-38 Area
TPH Result (mg/kg)
PetroFLAG™
System
320
25,000
280
24,500
300
19,500
3,240
13,570
230
23,200
1,170
630
1,550
460
830
>3,330
>3,330
850
750
2,040
1,860
740
1,080
1,610
1,860
470
530
>3,330
>3,330
910
23
44
39
26
150
130
36
92
Reference
Method
68.2
15,000
90.2
12,000
44.1
13,900
1,330
8,090
93.7
12,300
28.8
144
617
293
280
1,870
1,560
9.56
270
881
1,120
14.2
219
1,180
1,390
15.2
54.5
2,570
3,030
15.9
79.0
41.5
61.4 .
67.3
193
69.4
43.8
51.6
Statistical Analysis Summary
Statistical Test
and Null Hypothesis
Statistical Test
Wilcoxon signed rank test
(nonparametric)
Null Hypothesis
The median of the differences
between the paired observations
(PetroFLAG™ System and reference
method results) is equal to zero.
Statistical Test
Two-tailed, paired Student's t-test
(parametric)
Null Hypothesis
The mean of the differences between
the paired observations
(PetroFLAG™ System and reference
method results) is equal to zero.
Were PetroFLAG™ System and
Reference Method Results
Statistically the Same or Different?
Different
Different"
Same
Probability of Null
Hypothesis Being
True (percent)
0.2
0.00
58.52
                                                         68

-------
Table 7-4.  Statistical Comparison of PetroFLAG™ System and Reference Method TPH Results for Environmental Samples (Continued)
Sampling Area
Slop Fill Tank
Area
TPH Result (mg/kg)
PetroFLAG™
System
610
470
1,030
580
110
83
100
250
6,980
2,100
1,570
1,070
1,410
870
960
460
2,360
760
860
360
1,510
830
1,170
340
2,140
15,260
580
56
Reference
Method
105
269
397
339
6.16
37.1
43.9
52.4
3,300
1.270
588
554
834
501
280
185
1,090
544
503
146
938
517
369
253
151
3,960
1,210
121
Statistical Analysis Summary
Statistical Test
and Null Hypothesis
Statistical Test
Two-tailed, paired Student's t-test
(parametric)
Null Hypothesis
The mean of the differences between
the paired observations
(PetroFLAG™ System and reference
method results) is equal to zero.
Were PetroFLAG™ System and
Reference Method Results
Statistically the Same or Different?
Different
Probability of Null
Hypothesis Being
True (percent)
0.00
Notes:

>     =  Greater than
mg/kg =  Milligram per kilogram

a    When the PetroFLAG™  System results of >3,330 mg/kg were not included in the statistical analysis, a two-tailed, paired  Student's t-test
    (parametric) showed that the PetroFLAG™ System and reference method results were still statistically different (the probability was 0.00 percent).
                                                              69

-------
 (9
 ^  82
 k.  S 1
 rt»  J5 '
                        Blank soil
                 Total number of samples: 3
            X) to 30
                          >30 to 50
                        Bias, percent
                                            >50
                                                          *1
                                                           •o" ?2
                                                           1
                                                           Si
                                                           *i1
                                                           E I*
                                                                      Diesel in medium-concentration range
                                                                           Total number of samples: 3
                                                                       >0to30
                                                                                     >30 to 50
                                                                                   Bias, percent
>50
 If1
                   Weathered gasoline In
                  high-concentration range
                 Total number of samples: 6
            >0 to 30
                           >30 to 50
                        Bias, percent
                                            >50
                                                                        Diesel in high-concentration range
                                                                            Total number of samples: 6
                                                                      >0 to 30
                                                                                     >30 to 50
                                                                                   Bias, percent
                                                                                                      >50
     7

%*•
 ll5
 Is4
 *E3
 k. fi,
 » ^2
 I*'
     0
              Diesel in low-concentration range
                 Total number of samples: 7
            >0to30
                           >30 to 50
                         Bias, percent
                                            >50
                                                             Notes:

                                                             > - Greater than

                                                             HI   PetroFLAG™ system result  biased  low  compared to
                                                                  reference method result
                                                             •   PetroFLAG™ system result biased high compared to
                                                                  reference method result
Figure 7-3.  Measurement bias for soil performance evaluation samples.
                                                       70

-------
method TPH results, a parametric test (a two-tailed, paired
Student's t-test) or a nonparametric  test (a Wilcoxon
signed rank test) was  selected based on the approach
presented in  Figure 7-1.  Tables  7-4 and 7-5 present
statistical comparisons of the PetroFLAG™ System and
reference method results for  environmental  and  PE
samples,  respectively.     The   tables   present   the
PetroFLAG™ System and reference method results  for
each sampling area or PE sample type, the statistical test
performed and the  associated null hypothesis  used to
compare the results, whether the results were statistically
the same or different, and the probability that the results
were the same.

Table  7-4 shows that the  PetroFLAG™ System and
reference method results for all sampling areas except the
B-38  Area were statistically different at  a significance
level  of 5 percent.  Specifically, the probability of the
results being the same was less than 5 percent for the FFA,
NEX Service Station Area, and SFT Area.  The statistical
test conclusions  appeared  to  be  reasonable  because
compared  to  the  reference   method   results,   the
PetroFLAG™ System results were (1) biased high for the
FFA samples by up to a factor of seven, (2) biased high for
the NEX Service Station Area samples by up to two orders
of magnitude, and (3) biased high for 26 SFT Area samples
by up to a factor of 18 and biased low for 2 SFT Area
samples by a factor of two.  The  pairwise comparison
for the NEX Service Station Area was performed using
3,330 mg/kg as the TPH result for the four PetroFLAG™
System results reported  as  greater than  3,330 mg/kg.
When the PetroFLAG™  System results of greater than
3,330 mg/kg were not included in the statistical analysis,
the probability  that  the PetroFLAG™  System  and
reference method  results  were   the same  remained
unchanged, indicating that the results were still statistically
different.

Table 7-5 shows that the PetroFLAG™ System  and
reference method results were statistically the same at a
significance  level of 5 percent for high-concentration-
range diesel soil PE  samples.  The PetroFLAG™ System
and reference method results for all other PE sample types
were statistically different. Of the PetroFLAG™ System
results for PE samples that were statistically different from
the reference  method results, the PetroFLAG™ System
results for (1) blank soil samples were biased high by at
least an order of magnitude,  (2) weathered gasoline soil
samples and medium-range soil diesel soil samples were
biased high by up to a factor  of two, (3) low-range diesel
soil samples were biased high by up to a factor of six, and
(4) liquid samples were biased low by up to 71 percent for
neat weathered gasoline and by up to 30 percent for neat
diesel. Li addition, the PetroFLAG™ System results for
the liquid PE samples were biased low when compared to
the  sample densities; specifically, the  PetroFLAG™
System results were biased low  by 76 percent for neat
weathered gasoline and by 8 percent for neat diesel. The
low bias observed for neat weathered gasoline was largely
associated  with the response factor selected by Dexsil
for neat weathered gasoline samples.  Specifically, had
Dexsil selected the response factor for weathered gasoline
(2) instead of the response factor used for diesel samples
during the  demonstration (6), the device results for neat
weathered gasoline samples would have been three times
higher; moreover, the bias would have been reduced to
34 percent when the device results were compared to the
sample density and to 8 percent when the device results
were compared to the average TPH results of the reference
method.

Correlation of TPH Results

To determine  whether a  significant correlation existed
between the PetroFLAG™ System and reference method
TPH results, linear regression analysis was performed. A
strong correlation between the PetroFLAG™ System and
reference method results would indicate that the device
results could be adjusted using the established correlation
and that field decisions could be made using the adjusted
results in situations where the device results may not be the
same as off-site laboratory results.  Figures 7-4 and 7-5
show the linear regression plots for environmental and soil
PE  samples,  respectively.   Table  7-6  presents  the
regression model, square of the correlation coefficient (R 2),
and probability that the slope of the regression line is equal
to zero (F-test probability) for each sampling area and soil
PE sample type. Because Dexsil chose to analyze only the
high-concentration-range   samples   during   the
demonstration, the range  of corresponding reference
method TPH results was  small (1,740 to 2,180 mg/kg).
Therefore, the  regression analysis for weathered gasoline
samples was less robust than that for other samples.

Table  7-6  shows  that R2  values for (1) environmental
samples except B-38 Area samples ranged from 0.84 to
0.96 and (2) soil PE samples except weathered gasoline
samples were at  least 0.89.   The  R2  values  were
considerably lower for B-38  Area samples  (0.42) and
weathered  gasoline soil PE samples (0.10).  The linear
regression analysis for the NEX Service Station Area was
conducted using 3,330 mg/kg for the four PetroFLAG™
                                                    71

-------
Table 7-5. Statistical Comparison of PetroFLAG™ System and Reference Method TPH Results for Performance Evaluation Samples
Sample Type
TPH Result
PetroFLAG™
System
Reference
Method
Statistical Analysis Summary
Statistical Test .
and
Null Hypothesis
Were PetroFLAG™
System and Reference
Method Results
Statistically the Same or
Different?
Probability of
Null Hypothesis
Being True
(percent)
Soil Samples (Processed Garden Soil) (TPH Results in Milligram per Kilogram) ° • : '
Blank (9 percent moisture content)
Weathered
gasoline
Diesel
High-concentration range
(9 percent moisture content)
High-concentration range
(16 percent moisture content)
Low-concentration range
(9 percent moisture content)
Medium-concentration range
(9 percent moisture content)
High-concentration range
(9 percent moisture content)
High-concentration range
(less than 1 percent moisture
content)
170
160
170
3,760
3,650
3,360
2,550
2,920
2,740
72
61
75
68
76
80
77
510
530
520
2,970
2,860
2,980
3,310
3,490
3,020
5.12
13.1
13.5
1,880
2,020
2,180
1,740
1,980
2|050
12.0
16.5
13.7
16.4
17.4
17.2
14.8
226
265
267
2,480
2,890
2,800
2,700
2,950
3,070
Statistical Test
Two-tailed, paired
Student's t-test
(parametric)
Null Hypothesis
The mean of the
differences between the
paired observations
(PetroFLAG™ System and
reference method results)
is equal to zero.
Different
Different
Different
Different
Different
Same
Same
0.11
1.67
0.78
0.00
0.11
29.33
22.19
Liquid Samples (Neat Materials) (TPH Results In Milligram per Liter) R ' * > ' ' • ,-X ,1 ".•' ^\-
Weathered gasoline
Diesel
213,860
178,890
200,200
783,900
763,000
815,260
656,000
611,000
677,000
1,090,000
1,020,000
1,160,000
Statistical Test
Two-tailed, paired
Student's t-test
(parametric)
Null Hypothesis
The mean of the
differences between the
paired observations
(PetroFLAG™ System and
reference method results)
is equal to zero.
Different
Different
0.09
0.70
                                                        72

-------
    25.000

% ^20,000
«!* o>
           Comparison of Fuel Farm Area results
                      5,000        10,000       15,000
                 Reference method TPH result (mg/kg)
                                                                        Comparison of B-38 Area results
                                                                150

                                                                100
                                                                50
                                                                                               R2 = 0.421
                                                                                * *
                                                                           50        100       150       200
                                                                       Reference method TPH result (mg/kg)
Comparison of Naval Exchange
Service Station Area results
i inn

CO e o 4nn

2 = 1 400
< to 1,^00 •
-J o>
U. C I
a T1 7HO 3
* o.
» 1—
Q- n
4 ^^^ *
^^^


*^x*
\xr
t£*
r*

0 800 1,600 2,400 3,200
Reference method TPH result (mg/kg)
                                                                    Comparison of Slop Fill Tank Area results
                                                                16,000 -•	
                                                                             1,000     2,000     3,000     '
                                                                          Reference method TPH result (mg/kg)
Notes:

mg/kg  -  Milligram per kilogram
                                   R2 = Square of the correlation coefficient
Figure 7-4. Linear regression plots for environmental samples.
System results reported as greater than 3,330 mg/kg.
When the PetroFLAG™ System results of greater than
3,330 mg/kg were not included in the linear regression
analysis, the R2 value for this area decreased to 0.73. The
R2 value for the regression model for diesel  soil PE
samples  (0.98) was significantly higher than  that for
weathered gasoline soil PE samples (0.10); therefore, it
was  not  surprising  that the R2  value for the combined
regression model for these PE samples was 0.89.

The probability of the slopes of the regression lines being
equal to zero was 0.00 percent for all environmental and
soil PE samples except B-38 Area and weathered gasoline
soil  PE samples, indicating that there  was less than a
                                                       5 percent probability that the PetroFLAG™ System and
                                                       reference method results correlated only by chance.  The
                                                       probability that the PetroFLAG™ System and reference
                                                       method results correlated by chance was higher for B-38
                                                       Area samples (8.14 percent) and weathered gasoline soil
                                                       PE samples  (55.03  percent).   Based on  the  R2 and
                                                       probability   values,  the  PetroFLAG™  System  and
                                                       reference method results were considered to be (1) highly
                                                       correlated for FFA samples, diesel soil PE samples, and
                                                       weathered  gasoline  and  diesel  soil  PE  samples;
                                                       (2) moderately correlated for NEX Service Station Area
                                                       and SFT Area samples; and (3) weakly correlated for B-3 8
                                                       Area samples and weathered gasoline soil  PE samples.
                                                      73

-------
PetroFLAG™ System
TPH result (mg/kg)
Comparison of weathered gasoline
performance evaluation sample results
Q onn
9 Ann .
4 enn

ft .
*«
^^*
^^^*+
^ —
^f^^^
R2- 0 10!



0 800 1,600 2.400
Reference method TPH result (mg/kg)
                                                       7.1.2.2
           Precision


e
I«
w E
l!
u_ »-
8*
« K
tt. ^



Comparison of dlesel
performance evaluation sample results
o f\nt\
2.500
o nnn
1.500-
1.000
cnn

0<
*X^
^^X%
^^^^
^^r^
^r
^^^^
_^^^ R2 — n nfll

A ^^^

r








0 800 1.600 2,400 3,200
Reference method TPH result (mg/kg)
        Comparison of weathered gasoline and diesel
           performance evaluation sample results

          1                      *•.    ^<
                  800      1.600      2,400     3,200
               Reference method TPH result (mg/kg)
Notes:
mg/kg = Milligram per kilogram

R2   = Square of the correlation coefficient
Figure 7-5.  Linear regression plots for soil performance evaluation
          samples.
Both environmental and PE samples were analyzed to
evaluate the precision associated with TPH measurements
using the PetroFLAG™ System and reference method.
The results of this evaluation are summarized below.

Environmental Samples

Blind field triplicates were analyzed to evaluate the overall
precision of the sampling, extraction, and analysis steps
associated  with TPH measurement.  Each set of field
triplicates was collected from a well-homogenized sample.
Also, extract  duplicates  were  analyzed  to  evaluate
analytical precision only.  Each set of extract duplicates
was  collected by  extracting  a given soil sample and
collecting  two aliquots  of the  extract.    Additional
information on  field triplicate  and extract duplicate
preparation is included in Chapter 4.

Tables 7-7 and 7-8 present the PetroFLAG™ System and
reference method results for field triplicates and extract
duplicates,  respectively.  Precision  was estimated using
RSDs for field triplicates and RPDs for extract duplicates.

Table 7-7 presents the TPH results and RSDs for 11 sets of
field triplicates analyzed using the PetroFLAG™ System
and reference method. For the PetroFLAG™ System, the
RSDs ranged from 7  to 38 percent with a  median of
19 percent. The RSDs for the reference method ranged
from 4 to 39 percent with a median of 16 percent. The
median RSDs for the PetroFLAG™ System and reference
method indicated about the same level of precision. The
PetroFLAG™ System and reference method RSDs did not
exhibit  consistent trends  based  on  soil type,   PHC
contamination type, or TPH concentration.

Table 7-8 presents the TPH results and RPDs  for 10 and
12  sets  of  extract  duplicates  analyzed  using the
PetroFLAG™ System and reference method, respectively.
For the PetroFLAG™ System, the RPDs ranged from 1 to
26 with a median off. The RPDs for the reference method
ranged from 0 to 11 with a median of 4. The median RPDs
for  the   PetroFLAG™  System  and   reference
method indicated about the same level of precision. The
PetroFLAG™ System and reference method RPDs did not
exhibit consistent trends based on PHC contamination type
or TPH concentration. As expected, the median RPDs for
extract duplicates were less than the median RSDs for
field triplicates for both the PetroFLAG™ System and
reference method.  These findings indicated that greater
                                                    74

-------
Table 7-6. Summary of Linear Regression Analysis Results
Sampling Area or Sample Type
Environmental Samples .-
Fuel Farm Area
Naval Exchange Service Station Area'
B-38 Area
Slop Fill Tank Area
Soil Performance Evaluation Samples
Weathered gasoline6
Diesel
Weathered gasoline and diesel
Regression Model
(y = PetroFLAG™ System TPH result,
x = reference method TPH result)
"-.'•'. '"• "'' :'";. .'""'- ', ' -•' '' ° '•''<" .-
y'=1.68x -1-444.10
y=1.06x + 711.08
y = 0.66x+ 17.48
y = 2.99x - 382.22
•c *
V. . '
y=1.02x +1,1 40.67
y=1.06x+118.68
y=1.17x + 225.73
Square of
Correlation
Coefficient
. J ... "-..;'/ .(;..'
0.96
0.86
0.42
0.84
' - "' '„ v,
0.10
0.98
0.89
Probability That Slope of
Regression Line Was Equal
to Zero (percent)
.-!. i"'"'*°--: i-'-":-V;','^-|
0.00
0.00
8.14
0.00
*"'-"' '*.' i, i
0 " . , b ' *» . ?
55.03
0.00
0.00
Notes:

mg/kg =  Milligram per kilogram

"  The linear regression analysis for the Naval Exchange Service Station Area was conducted using 3,330 mg/kg for the four PetroFLAG™ System
   results reported as greater than 3,330 mg/kg. When the PetroFLAG™ System results of greater than 3,330 mg/kg were not included in the linear
   regression analysis, the square of the correlation coefficient value for this area decreased to 0.73, but the probability that the slope of the regression
   line was equal to zero remained the same.
" ' Because Oexsil chose to analyze only the high-concentration-range samples during the demonstration, the range of corresponding reference
   method TPH results was small (1,740 to 2,180 mg/kg). Therefore, the regression analysis for weathered gasoline samples was less robust than
   that for other samples.
precision was achieved when only the analysis step could
have contributed to TPH measurement error than when all
three steps (sampling, extraction, and analysis) could have
contributed to such error.

Performance Evaluation Samples

Table 7-9 presents the PetroFLAG™ System and reference
method TPH results and RSDs for seven sets of replicates
for soil PE samples and two sets of triplicates for liquid PE
samples.   The  PetroFLAG™ System  RSDs for  the
replicate sets of soil samples ranged from 2 to 9 percent
with a median of 6 percent. The PetroFLAG™ System
RSDs for the two triplicate sets of liquid samples were
3 and 9 percent with a median of 6 percent.

For the reference method, the RSD calculated for the blank
soil  samples  was  not  considered in  evaluating  the
method's  precision  because one of the three blank soil
sample results (5.12 mg/kg) was estimated by adding one-
half the reporting limits  for the GRO, DRO, and ORO
components of TPH measurement.  The RSDs for  the
remaining six replicate sets ranged from 6 to  13 percent
with a median of 8 percent.   The RSDs for the two
triplicate sets of liquid samples were 5 and 6 percent with
a median  of 5.5 percent.   The median  RSDs for  the
PetroFLAG™ System and reference method indicated
about  the  same  level of precision.  Finally,  for  the
reference method, the median RSD for the soil PE samples
(8 percent) was  less than that for the environmental
samples (16 percent), indicating that greater precision was
achieved for the samples prepared under more controlled
conditions  (the  PE  samples).     Similarly  for  the
PetroFLAG™ System,  the median RSD for soil  PE
samples  (6  percent)  was  less   than   that  for  the
environmental samples (19 percent).

7.1.3  Primary Objective P3: Effect of
       Interferents

The effect of interferents on TPH measurement using the
PetroFLAG™ System and reference method was evaluated
through  analysis of  high-concentration-range soil  PE
samples that contained weathered gasoline or diesel with
or without an interferent.  The six interferents used were
MTBE;   PCE;   Stoddard   solvent;   turpentine;
1,2,4-trichlorobenzene; and humic acid. In addition, neat
(liquid) samples of each interferent except humic acid were
used as quasi-control samples to evaluate the effect of each
interferent  on  the  TPH  results  obtained  using  the
PetroFLAG™ System and the reference method. Liquid
interferent samples were submitted for analysis as blind
                                                      75

-------
Table 7-7. Summary of PetroFLAG™ System and Reference Method Precision for Field Triplicates of Environmental Samples
Sampling Area
Fuel Farm Area
Naval Exchange Service
Station Area
B-38 Area
Slop Fill Tank Area
Field Triplicate
Set
1
2
3
4
5
6
7
8
9
10
11
PetroFUAG™ System
TPH Result
(milligram per kilogram)
320
280
300
25,000
24,500
19,500
830
750
1,080
>3,330
2,040
1,610
>3,330
1,860
1,860
850
740
470
23
39
26
1,410
2,360
1,510
870
760
830
960
860
1,170
460
360
340
Relative Standard
Deviation (percent)
7
13
19
38"
36"
28
29
30
7
16
16
Reference Method
TPH Result
(milligram per kilogram)
68.2
90.2
44.1
15,000
12,000
13,900
280
. 270
219
1,870
881
1,180
1,560
1,120
1,390
9.56
14.2
15.2
79
61.4
67.3
834
1,090
938
501
544
517
280
503
369
185
146
253
Relative Standard
Deviation (percent)
34
11
13
39
16
23
13
14
4
29
28
Notes:

>  = Greater than

'  The relative standard deviations were calculated for triplicate sets 4 and 5 using 3,330 milligrams per kilogram for the PetroFLAG™ System results
   reported as >3,330 milligrams per kilogram.
                                                               76

-------
Table 7-8. Summary of PetroFLAG™ System and Reference Method Precision for Extract Duplicates
Sampling Area
Fuel Farm Area
Naval Exchange Service
Station Area
B-38 Area
Slop Fill Tank Area
Extract
Duplicate
Set
1
2
3
4
5
6
7
8
9
10
11
12
PetroFLAG™ System
TPH Result
(milligram per kilogram)
300
Not analyzed3
20,500
18,500
1,030
1,130
1,550
1,660
1,820
1,900
470
Not analyzed3
20
26
43
45
1,400
1,420
850
890
970
950
450
460
Relative Percent
Difference
Not calculated3
10
9
7
4
Not calculated3
26
5
1
5
j~-
2
2
Reference Method
TPH Result
(milligram per kilogram)
44.1
44.1
13,700 .
14,000
226
213
1,190
1,170
1,420
1,360
15.5
14.9
79.6
78.4
41.4
41.5
829
838
528
473
271
289
189
181
Relative Percent
Difference
0
2
6
2
4
4
2
0
1
11
6
4
Note:
   Not enough extract was generated to analyze the extract duplicate; therefore, a relative percent difference could not be calculated for this extract
   duplicate set.
triplicate samples.  Dexsil and the reference  laboratory
were  provided  with  flame-sealed  ampules  of  each
interferent and were given specific instructions to prepare
dilutions of the liquid interferents  for analysis.   Two
dilutions of each interferent we re prepared; therefore, there
were six PetroFLAG™ System and reference method TPH
results for each interferent.   Blank soil was mixed with
humic acid at two levels to prepare quasi-control samples
for this interferent.    Additional details regarding the
interferents are provided in Chapter 4. The results for the
quasi-control interferent samples are discussed first below,
followed  by  the  effects  of the  interferents on the
TPH results for soil samples.
7.13.1     Interferent Sample Results

Table  7-10  presents  the  PetroFLAG™ System  and
reference method  TPH results,  mean TPH results,  and
mean responses for triplicate sets of liquid PE samples and
soil  PE  samples  containing humic acid. Each mean
response was calculated by dividing the mean TPH result
for a triplicate set by  the interferent concentration  and
multiplying by 100. For liquid PE samples, the interferent
concentration was  estimated using its density and purity.

The mean responses for the PetroFLAG™ System ranged
from 0 to 103 percent  for the liquid interferent samples;
the mean response for humic acid was 2.5 percent. The
                                                      77

-------
Table 7-9. Comparison of PetroFLAG™ System and Reference Method Precision for Replicate Performance Evaluation Samples
Sample Type
Replicate Set
PetroFLAG™ System
TPH Result
i^^i^i^i^ii^^^niil^i^Pn^alii^S^
Blank (9 percent moisture content)
Weathered
gasoline
Diesel
High-range TPH
concentration
(9 percent moisture
content)
High-range TPH
concentration
(16 percent moisture
content)
Low-range TPH
concentration
(9 percent moisture
content)
Medium-range TPH
concentration
(9 percent moisture
content)
High-range TPH
concentration
(9 percent moisture
content)
High-range TPH
concentration (less
than 1 percent
moisture content)
1
2
3
4
5
6
7
170
160
170
3,760
3,650
3,360
2,550
2,920
2,740
72
61
75
68
76
80
77
510
530
520
2,970
2,860
2,980
3,310
3,490
3,020
Relative Standard
Deviation (percent)

3
6
7
9
2
2
7
^flff^^B^S^^rtSa^iib^Crt'hTS^wft^Tn^l^^ip'l^^ter^ * "**"j!~V _ *? ^f^? "^>Z
Weathered gasoline
Diesel
8
9
213,860
178,890
200,200
783,900
763,000
815,260
9
3
Reference Method
TPH Result
Relative Standard
Deviation (percent)

5.12
13.1
13.5
1,880
2,020
2,180
1,740
1,980
2,050
12.0
16.5
13.7
16.4
17.4
17.2
14.8
226
265
267
2,480
2,890
2,800
2,700
2,950
3,070
45
7
8
13
9
8
6
ijf iT^^t" l^^lf^ Ji
656,000
611,000
677,000
1,090,000
1,020,000
1,160,000
5
6
                                                          78

-------
Table 7-10.  Comparison of PetroFLAG™ System and Reference Method Results for Interferent Samples
Interferent and Concentration"
PetroFLAG™ System
TPH Result
Mean TPH
Result
Mean Response"
(percent)
l^Ai^er^mi^p^^HRe^^^m^amp^mKf' •"- . ^ -„ _1
Methyl-tert-butyl ether
(740,000 milligrams per liter)
Tetrachloroethene
(1,621,000 milligrams per liter)
Stoddard solvent
(771,500 milligrams per liter)
Turpentine
(845,600 milligrams per liter)
1 ,2,4-Trichlorobenzene
(1,439,000 milligrams per liter)

Humic acid at 3,940 milligrams per
kilogram
Humic acid at 19,500 milligrams
per kilogram
<4,020
<1 0,020
<4,020
<2,020
<1 0,020
<2,020
42,110
12,530
34,890
>669,930
18,540
18,180
329,660
332,660
340,680
337,670
332,140
322,140
874,250
844,690
864,730
880,260
897,790
853,700
129,920
128,320
91,430
462,510
127,170
438,210
artJanlSoilVCi'Ptt
120
110
140
410
350
390
3,010
2,340
29,840
235,550
334,330
327,300
861,220
877,250
116,560
342,630
Results in MM!
120
380
0
0
2
14°
43
42
102
104
8
24
Reference Method
TPH Result
Mean TPH
Result
Mean Response"
(percent)
" ;*• "»; vw T , - if-^f1" -- ~- .."v^s- s>
, .-v, *J - .J« • , »ti ;. -^. ; - JH .- - .
309,000
272,000
270,000
303,000
313,000
282,000
269,000
270,000
277,000
290,000
288,000
307,000
561,000
628,000
606,000
703,000
Not reported
713,000
504,000
459,000
442,000
523,000
353,000
349,000
711,000
620,000
732,000
754,000
756,000
752,000
284,000
299,000
272,000
295,000
598,000
708,000
468,000
408,000
688,000
754,000
38
40
17
18
78
92
55
48
48
52
fiS»i^liiiin»F«« -, £^~^>-£rrti "I'.iS:. s*
3
2
8.99
8.96
8.12
69.3
79.1
78.5
9.00
76.0
0
0
Notes:
>  =  Greater than
<  =  Less than
a   A given liquid interfered concentration was estimated using its density and purity.
b   The mean response was calculated by dividing the mean TPH result for a triplicate set by the interferent concentration and multiplying by 100.
°   When the PetroFLAG™ System result of >669,930 milligrams per liter (an analytical outlier) was not considered, the mean response was
    calculated to be 1 percent.
                                                               79

-------
TPH results for a given triplicate set and between the
triplicate sets showed good agreement for Stoddard solvent
and turpentine. The mean responses for MTBE (0 percent)
and  Stoddard  solvent  (42.5 percent)  indicated  that
Stoddard  solvent can be measured as TPH using the
device. The mean responses for turpentine (103 percent)
and 1,2,4-trichlorobenzene (16 percent) indicated that both
of these interferents will likely result in false positives
during TPH measurement.  The mean response for PCE
(8 percent) was affected by one of the  six high-level
samples that had a TPH result (greater than 669,930 mg/L)
up to an order of magnitude greater than the other five
results, making the mean response questionable. When the
PetroFLAG™ System result of greater than 669,930 mg/L
(an analytical outlier) was not considered,  the mean
response  for PCE (1.5 percent)  indicated that  this
interferent would not result in either false positives or false
negatives  during  TPH measurement.   Also, the mean
response of 2.5 percent for humic acid indicated that humic
acid would not result in either false positives or false
negatives during TPH measurement.

The mean responses for the reference method ranged from
17 to 92 percent  for the liquid interferent samples; the
mean response for humic acid was 0 percent.  The TPH
results for a given triplicate set and between the triplicate
sets showed good agreement.  The mean responses for
MTBE (39 percent) and Stoddard solvent (85 percent)
indicated that these compounds can be measured as TPH
using the reference method.  The mean responses for PCE
(17.5   percent);   turpentine   (52   percent);   and
1,2,4-trichlorobenzene (50 percent) indicated that these
interferents will likely result in false positives during TPH
measurement. The mean response of 0 percent for humic
acid indicated that humic acid would not result in either
false positives or false negatives during TPH measurement.

7.1J.2     Effects of Interferents on TPH Results for
           Soil Samples

The effects of interferents on TPH measurement for soil
samples containing weathered gasoline  or diesel were
examined  through analysis  of PE  samples  containing
(1)   weathered  gasoline   or   diesel   (control)  and
(2) weathered gasoline or diesel plus a given interferent at
two levels. Information on the selection of interferents is
provided in Chapter 4.
Triplicate sets of control samples and samples containing
interferents  were  prepared  for   analysis  using  the
PetroFLAG™  System  and  reference  method.    A
parametric   or  nonparametric  test was  selected for
statistical evaluation of the results using  the approach
presented in Figure 7-1.

TPH results for samples with and without interferents,
statistical tests performed, and statistical test conclusions
for both the PetroFLAG™ System and reference method
are presented in Table 7-11.  The null hypothesis for the
statistical tests  was that mean TPH results for samples
with and without interferents were equal. The statistical
results for each interferent are  discussed below.

Effect of Methyl-Tert-Butyl Ether

The effect of MTBE was evaluated for soil PE samples
containing weathered gasoline.  Based on the liquid PE
sample (neat material)  analytical  results, MTBE  was
expected to  have no effect on the TPH results for the
PetroFLAG™ System; however, it  was expected to  bias
the reference method results high.

For the PetroFLAG™ System, MTBE biased the TPH
results low; the bias was statistically significant only at the
high MTBE level. This observation appeared to contradict
the conclusions drawn from the analytical results for liquid
PE samples  containing MTBE (quasi-control samples);
however, the apparent contradic tion was attributable to the
fact that  the quasi-control sample analyses  could predict
only a positive bias  (a negative bias is equivalent  to a
negative  concentration).

For the reference method, at  the interferent levels used,
MTBE was  expected to bias the  TPH results high by
21 percent (low level) and 33 percent (high level).  The
expected bias  would be lower (17  and 27 percent,
respectively) if MTBE in soil samples was assumed to be
extracted as  efficiently  as  weathered gasoline in soil
samples. However, no effect  on TPH measurement was
observed for  soil PE  samples analyzed during the
demonstration.  A significant amount of MTBE, a highly
volatile compound, may have been lost during PE sample
preparation, transport, storag e, and handling, thus lowering
the MTBE concentrations to levels that would not have
increased the TPH results beyond the reference method's
precision (7 percent).
                                                    80

-------
        Table 7-11.  Comparison of PetroFLAG™ System and Reference Method Results for Soil Performance Evaluation Samples Containing Interferents
Sample Matrix and
Interferent*
IliS'^ib'SitfylilS'Sfi'iP
Weathered gasoline
Diesel

Weathered
gasoline

MTBE
(1,100mg/kg)
MTBE
(1,700mg/kg)
PCE
(2,810mg/kg)
PCE
(13,100 mg/kg)
PetroFLAG™ System
TPH Result
(mg/kg)
Mean TPH
Result
(mg/kg)
(LUtSt* ^A^^||fe|^.li^3^!*i?fei?iBi£^^S
s¥fl^™l!!^^^^^^
3,760 ~
3,650
3,360
2,970
2,860
2,980

3,140
3,280
3,670
2,390
2,040
1,840
4,570
5,960
7,780
>20.000
2,890
2,690

3,590
2,940

3,360
2,090
6,100
8,530
Statistical Tests
Were Mean
TPH Results
for Samples
With and
Without
Interferents
the Same or
Different?
Probability of
Mean TPH
Results for
Samples With
and Without
Interferents Being
the Same
(percent)
Reference Method
TPH
Result
(mg/kg)

Not applicable
Not applicable
1,880
2,020
2,180
2,480
2,890
2,800

One-way
analysis of
variance
(parametric) and
Tukey (honest,
significant
difference)
pairwise
comparison of
means
(parametric)
Kruskal-Wallis
one-way
analysis of
variance
(nonparametric)
and Kruskal-
Wallis pairwise
comparison of
means
(nonparametric)
Mean with
interferent at
high level
was different
from means
without
interferent
and with
interferent at
low level
Same
0.07
30.12"
1,900
1,750
2,210
2,150
2,320
2,560
2,540
2,160
2,450
4,740
4,570
4,040
Mean TPH
Result
(mg/kg)

2,030
2,720

1,950
2,340
2,380
4,450
Statistical Tests
Were Mean
TPH Results
for Samples
With and
Without
Interferents
the Same or
Different?
mmmmmwm
Probability of
Mean TPH
Results for
Samples With
and Without
Interferents
Being the Same
(percent)
$%?$$&&
Not applicable
Not applicable

One-way analysis of
vanance
(parametric) and
Tukey (honest,
significant
difference) pairwise
comparison of
means (parametric)
W®WW^i&*?'>
Same
Mean with
interferent at
high level
was different
from means
without
interferent
and with
interferent at
low level
11.21
0.00
00

-------
        Table 7-11. Comparison of PetroFLAG™ System and Reference Method Results for Soil Performance Evaluation Samples Containing Interferents (Continued)
Sample Matrix and
Interfered'

Weathered
gasoline
(Continued)
Diesel
Weathered
gasoline

Stoddard
solvent
(2,900 mg/kg)
Stoddard
solvent
(15,400 mg/kg)
Stoddard
solvent
(3,650 mg/kg)
Stoddard
solvent
(18,200 mg/kg)
Turpentine
(2,730 mg/kg)
Turpentine
(12,900 mg/kg)
PetroFLAG™ System
TPH Result
(mg/kg)

8,410
6,120
7,080
15,930
32,780
26,720
20,850
18,860
17,040
39,300
28,670
27,080
10,440
6,630
7,560
58,200
51,870
> 100,000
Mean TPH
Result
(mg/kg)
H*$fi {IBB
UpRssMsssBis
7,200
25,140
18,920
31,680
8,210
70,020
Statistical Tests
Were Mean
TPH Results
for Samples
With and
Without
Interferents
the Same or
Different?
w^^w^S^^lww^wSM^ssiJ^s^^^^^-' -ij5EF|K
One-way
analysis of
variance
(parametric) and
Tukey (honest,
significant
difference)
pairwise
comparison of
means
(parametric)
All three
means (with
and without
interferents)
were
significantly
different from
one another
All three
means (with
and without
interferents)
were
significantly
different from
one another
All three
means (with
and without
interferents)
were
significantly
different from
one another
Probability of
Mean TPH
Results for
Samples With
and Without
Interferents Being
the Same
(percent)
Reference Method
TPH
Result
(mg/kg)

0.02
0.00
0.00C
4,350
4,760
4,110
10,300
14,300
11,000
4,390
4,640
4,520
8,770
6,580
8,280
4,410
3,870
4,440
12,800
11,200
14,600
Mean TPH
Result
(mg/kg)
#C«l!lNlill§!isiHiB$
4,410
11,900
4,520
7,880
4,240
12,900
Statistical Tests
Were Mean
TPH Results
for Samples
With and
Without
Interferents
the Same or
Different?

One-way analysis of
variance
(parametric) and
Tukey (honest,
significant
difference) pairwise
comparison of
means (parametric)
All three
means (with
and without
interferents)
were
significantly
different from
one another
All three
means (with
and without
interferents)
were
significantly
different from
one another
All three
means (with
and without
interferents)
were
significantly
different from
one another
Probability of
Mean TPH
Results for
Samples With
and Without
Interferents
Being the Same
(percent)
ifiSftSlfff
0.00
0.00
0.00
oo
K)

-------
        Table 7-11. Comparison of PetroFLAG™ System and Reference Method Results for Soil Performance Evaluation Samples Containing Interferents (Continued)
oo
Sample Matrix and
Interferenf
PetroFLAG™ System
TPH Result
(mg/kg)
Mean TPH
Result
(mg/kg)
Statistical Tests
Were Mean
TPH Results
for Samples
With and
Without
Interferents
the Same or
Different?

Diesel
Turpentine
(3,850 mg/kg)
Turpentine
(19,600 mg/kg)
1,2,4-Trichloro-
benzene
(3,350 mg/kg)
1 ,2,4-Trichloro-
benzene
(16,600 mg/kg)
17,430
17,040
20,470
62,700
51,980
45,300
3,770
3,990
3,900
10,100
8,550
9,920
18,310
53,330
3,890
9,520
One-way
analysis of
variance
(parametric) and
Tukey (honest,
significant
difference)
pain/vise
comparison of
means
(parametric)
All three
means (with
and without
interferents)
were
significantly
different from
one another
All three
means (with
and without
interferents)
were
significantly
different from
one another
Probability of
Mean TPH
Results for
Samples With
and Without
Interferents Being
the Same
(percent)

0.00
0.00
Reference Method
TPH
Result
(mg/kg)

5,860
5,810
5,610
15,000
13,300
13,300
3,220
3,750
3,550
7,940
6,560
6,690
Mean TPH
Result
(mg/kg)

5,760
13,900
3,510
7,060
Statistical Tests
Were Mean
TPH Results
for Samples
With and
Without
Interferents
the Same or
Different?

Kruskal-Wallis one-
way analysis of
variance
(nonparametric) and
Kruskal-Wallis
pairwise
comparison of
means
(nonparametric)
One-way analysis of
variance
(parametric) and
Tukey (honest,
significant
difference) pairwise
comparison of
means (parametric)
Mean without
interferent
was same as
mean with
interferent at
low level;
mean with
interferent at
low level was
same as
mean with
interferent at
high level
Mean with
interferent at
high level
was different
from means
without
interferent
and with
interferent at
low level
Probability of
Mean TPH
Results for
Samples With
and Without
Interferents
Being the Same
(percent)
vttSSSMa^Mf
WwilSPTWTf*
2.65
0.01

-------
Table 7-11. Comparison of PetroFLAG™ System and Reference Method Results for Soil Performance Evaluation Samples Containing Interferents (Continued)
Sample Matrix and
Interferent"

Diesel
(Continued)
Humic acid
(3,940 mg/kg)
Humic acid
(19,500 mg/kg)
PetroFLAG™ System
TPH Result
(mg/kg)

2,870
2,880
2,660
3,140
2,680
3,150
Mean TPH
Result
(mg/kg)

2,800
2,990
Statistical Tests
$&$>* id£i»&u&u&i£iii iJC£it&*
«i^^Bs«MSlw^iSj^H|t
Kruskal-Wallis
one-way
analysis of
variance
(nonparametric)
and Kruskal-
Wallis pairwise
comparison of
means
(nonparametric)
Were Mean
TPH Results
for Samples
With and
Without
Interferents
the Same or
Different?
Probability of
Mean TPH
Results for
Samples With
and Without
Interferents Being
the Same
(percent)

Same
39.32
Reference Method
TPH
Result
(mg/kg)

2,150
2,080
2,360
2,660
2,420
2,270
Mean TPH
Result
(mg/kg)

2,200
2,450
Statistical Tests

One-way analysis of
variance
(parametric) and
Tukey (honest,
significant
difference) pairwise
comparison of
means (parametric)
Were Mean
TPH Results
for Samples
With and
Without
Interferents
the Same or
Different?
Probability of
Mean TPH
Results for
Samples With
and Without
Interferents
Being the Same
(percent)

Mean without
interfered
was same as
mean with
interferent at
high level;
mean with
interferent at
low level was
same as
mean with
interferent at
high level
3.87
Notes:
>      =  Greater than
mg/kg  =  Milligram per kilogram
MTBE  =  Methyl-tert-butyl ether
PCE   =  Tetrachloroethene

*   All samples were prepared at a 9 percent moisture level.

b   When the PetroFLAG™ System result of >20,000 mg/kg (an analytical outlier) was not considered, a Kruskal-Wallis one-way analysis of variance (nonparametric) and a Kruskal-Wallis pairwise
    comparison of means (nonparametric) showed that the mean without the interferent was the same as the means with the interferent at the low and high levels; however, the mean with the
    interferent at the low level was not the same as the mean with the interferent at the high level.  The probability of mean TPH results for samples with and without interferents being the same
    was 4.39 percent.
0   When the PetroFLAG™ System result of >100,000 mg/kg (an analytical outlier) was not considered, a Kruskal-Wallis one-way analysis of variance (nonparametric) and a Kruskal-Wallis pairwise

    low level was the same as the mean with the interferent at the high level. The probability of mean TPH results for samples with and without interferents being the same was 4.39 percent.

-------
Effect of Tetrachloroethene
Effect of Stoddard Solvent
The effect of PCE was evaluated for soil PE samples
containing weathered gasoline.  Based on the liquid PE
sample  (neat material)  analytical results,  PCE  was
expected to have no effect on the TPH results for the
PetroFLAG™ System; however, it was expected to bias
the reference method results high.

Table  7-11  shows that  PCE  did  not   affect  the
PetroFLAG™ System TPH results for weathered gasoline
soil PE samples, which confirmed the conclusions drawn
from the results of the neat PCE analysis. However, when
the  PetroFLAG™  System  result  of  greater  than
20,000 mg/kg (an analytical outlier) was not considered,
a   Kruskal-Wallis   one-way  analysis  of  variance
(nonparametric)   and   a  Kruskal-Wallis  pairwise
comparison  of means (nonparametric) showed that the
mean TPH result without the interferent was the same as
the mean TPH results with the interferent at the low and
high levels  and that the mean  TPH  result with  the
interferent at the low level was not the same as the mean
TPH result with the interferent at the high level. When the
analytical outlier was not considered, a simple comparison
of mean TPH results showed that the mean TPH result was
biased high at the low PCE level and that it was biased low
at the high  PCE level.  Therefore, no  conclusion was
drawn regarding the effect of PCE on the TPH results for
the PetroFLAG™ System.

For the reference method, at the interferent levels  used,
PCE was expected to bias the TPH results  high by
24 percent (low level) and 113 percent (high level). The
expected bias would be  lower  (20 and 92 percent,
respectively) if PCE in soil samples was assumed  to be
extracted as efficiently  as  weathered gasoline in soil
samples.  The statistical tests showed that the probability
of the three  means  being equal was less than 5 percent.
However, the tests also showed that at the high level, PCE
biased the TPH results high, which appeared  to be
reasonable based on the conclusions drawn from  the
analytical results for neat PCE.  As to the reason for PCE
at the low level having no effect on the TPH results,
volatilization during PE sample preparation, transport,
storage,  and handling  may  have lowered  the  PCE
concentrations to levels that would not have increased the
TPH results beyond the reference method's precision
(7 percent).
The effect  of  Stoddard  solvent was evaluated for
weathered gasoline and diesel soil PE samples. Based on
the liquid PE sample (neat material) analytical results,
Stoddard  solvent was  expected to  bias  both the
PetroFLAG™ System and reference method results high.
For the PetroFLAG™ System, at the interferent levels
used, Stoddard solvent was expected to bias the TPH
results high by 34 percent (low level) and 180 percent
(high level) for weathered gasoline soil PE samples and by
52  percent (low level) and  260 percent (high level) for
diesel soil PE samples. The expected bias would be higher
(160 and 860 percent, respectively, for weathered gasoline
soil PE samples) and lower  (43 and  210  percent,
respectively, for diesel soil PE samples)  if Stoddard
solvent in soil samples was  assumed to be extracted as
efficiently  as weathered gasoline and diesel  in soil
samples. The statistical tests showed that the mean TPH
results with and without the interferent were different for
both weathered gasoline and diesel soil PE samples, which
confirmed the conclusions drawn from the analytical
results for neat Stoddard solvent.

For the reference method, at the interferent levels used,
Stoddard solvent was expected to bias the TPH results high
by 121 percent (low level) a nd 645 percent (high level) for
weathered gasoline soil PE samples and by 114 percent
(low level) and 569 percent (high level) for diesel soil PE
samples.  The expected bias would  be lower (99 and
524 percent, respectively, for weathered gasoline soil PE
samples and 61 and 289 percent, respectively, for diesel
soil PE samples) if Stoddard solvent in soil samples was
assumed to be  extracted as efficiently  as weathered
gasoline and diesel in soil samples. The statistical tests
showed that the  mean TPH results with and without the
interferent were different for both weathered gasoline and
diesel soil PE samples, which confirmed the conclusions
drawn from the analytical results for neat Stoddard solvent.

Effect of Turpentine

The effect  of turpentine was evaluated for weathered
gasoline and diesel soil PE samples. Based on the liquid
PE sample (neat material) analy tical results, turpentine was
expected to bias both the PetroFLAG™  System and
reference method results high.
                                                    85

-------
For the PetroFLAG™ System, at the interferent levels
used, turpentine was expected to bias the TPH results high
by 78 percent (low level) and 370 percent (high level) for
weathered gasoline soil PE samples and by 135 percent
(low level) and 687 percent (high level) for diesel soil PE
samples.  The expected bias would be higher (380 and
1,770 percent, respectively, for weathered gasoline soil PE
samples) and lower (110 and 560 percent, respectively, for
diesel soil PE samples) if Stoddard solvent in soil samples
was assumed to be extracted as efficiently as  weathered
gasoline and diesel in soil samples.  The statistical tests
showed that the mean TPH results with  and without the
interferent were different for both weathered gasoline and
diesel soil PE samples, which confirmed the conclusions
drawn from the analytical results for neat turpentine.

For  weathered gasoline  soil  PE samples,  when  the
PetroFLAG™ System result of greater than 100,000 mg/kg
(an analytical  outlier)  was  not considered, a Kruskal-
Wallis one-way analysis of variance (nonparametric) and
a  Kruskal-Wallis  pairwise  comparison  of means
(nonparametric)   were  performed  because   the  two
remaining results could not be tested to determine whether
they were normally distributed.  The nonparametric tests
showed that (1)  the  mean TPH  result  without  the
interferent and the mean TPH result with the interferent at
the low level were equal and (2) the mean TPH results with
the interferent at the low and high levels  were equal,
indicating that turpentine at the low level did not affect the
TPH results for the weathered gasoline soil PE samples but
that turpentine at the high level did affect the TPH results.
The conclusion reached for the interferent at the low level
was unexpected and did not  seem reasonable based on a
simple comparison of means that differed by a factor of
three. The anomaly might have been associated with the
nonparametric test used to evaluate the effect of turpentine
on TPH results for weathered gasoline soil PE samples, as
nonparametric  tests do not account for the magnitude of
the difference between TPH results.

For the  reference  method, at the interferent levels  used,
turpentine was expected to bias the TPH results high by
69 percent (low level) and 327 percent (high level) for
weathered gasoline soil PE samples and by 72 percent (low
level) and 371 percent (high level) for diesel soil PE
samples.  The expected bias would be lower (56 and
266 percent, respectively, for weathered gasoline soil PE
samples and 39 and 200 percent, respectively, for diesel
soil PE samples) if turpentine in soil samples was assumed
to be extracted as efficiently as weathered gasoline and
diesel in soil samples. The statistical tests showed that the
mean TPH results with and without the interferent were
different for weathered gasoline soil PE samples, which
confirmed  the conclusions drawn from  the  analytical
results for neat turpentine.  However, 'for diesel soil PE
samples, (1) the mean TPH result without the interferent
and the mean TPH result with the interferent at the low
level were equal and (2) the mean TPH results with the
interferent at the low and high levels were equal, indicating
that turpentine at the low level did not affect the TPH
results for the diesel soil PE samples but that turpentine at
the high level did affect the TPH results. The conclusion
reached for the interferent at the low level was unexpected
and did not seem reasonable based on a simple comparison
of means that differed by a factor of three. The anomaly
might have been associated with the nonparametric test
used to evaluate the effect of turpentine on TPH results for
diesel  soil  PE samples, as nonparametric tests do  not
account for the magnitude of the difference between TPH
results.

Effect of 1,2,4-Trichlorobenzene

The effect  of 1,2,4-trichlorobenzene was evaluated for
diesel soil PE samples.  Based on the liquid PE sample
(neat material) analytical results, 1,2,4-trichlorobenzene
was expected to bias  both the  PetroFLAG™ System and
reference method results high.

For the PetroFLAG™ System, at the interferent levels
used, 1,2,4-trichlorobenzene was expected to bias the TPH
results high by  18 percent (low level) and  90 percent
(high level). The expected bias would be lower (15 and
74 percent, respectively) if 1,2,4-trichlorobenzene in soil
samples was assumed to be  extracted  as efficiently as
diesel in soil samples.  The statistical tests showed that the
mean  TPH results with and without the interferent were
different, which confirmed the conclusions drawn from the
analytical results for neat 1,2,4-trichlorobenzene.

For the reference method,  at  the interferent levels used,
1,2,4-trichlorobenzene was expected to bias the TPH
results high by 62 percent (low level)  and 305 percent
(high level). The expected bias would be lower (33 and
164 percent, respectively) if 1,2,4-trichlorobenzene in soil
samples was assumed to be  extracted  as efficiently as
diesel in soil samples.  The statistical tests showed that the
probability of three  means being  equal  was less than
5 percent. However,  the tests  also showed that when the
interferent was present at the high level, TPH results were
biased high.   The effect observed  at the  high level
confirmed the conclusions drawn from  the  analytical
                                                     86

-------
results for neat 1,2,4-trichlorobenzene. The statistical tests
indicated that the mean TPH result with the interferent at
the low level was not different from the mean TPH result
without the interferent, indicating that the low level  of
1,2,4-trichlorobenzene did not affect TPH measurement.
However, a simple comparison of the mean TPH  results
revealed that  the  low level  of 1,2,4-trichlorobenzene
increased the TPH results to nearly the result based on the
expected bias of 33 percent. Specifically, the mean TPH
result with the interferent at the low level was 3,510 mg/kg
rather than the  expected value of 3,620 mg/kg.  The
conclusions drawn from the statistical tests were justified
when the variabilities  associated with the mean TPH
results were taken into account.

Effect of Humic Acid

The effect of humic acid was evaluated for diesel soil PE
samples.  Based on the  analytical results for soil PE
samples containing  humic acid, this  interferent  was
expected to have no effect on the TPH results for the
PetroFLAG™  System and reference method.

Table 7-11  shows that humic acid  did not affect the
PetroFLAG™  System  TPH results for  diesel soil PE
samples, which confirmed the conclusions drawn from the
analytical results for soil PE samples containing  humic
acid.

For the reference method, humic acid appeared to have
biased the TPH results low.  However, the bias decreased
with an increase in the humic acid level. Specifically, the
negative  bias  was  19 percent at the  low level  and
10 percent at the high level. For this reason, no conclusion
was drawn  regarding the effect of humic acid on TPH
measurement using the reference method.

7.1.4  Primary Objective P4: Effect of Soil
       Moisture Content

To  measure the  effect  of soil moisture  content on the
ability of the PetroFLAG™ System and reference method
to accurately measure TPH, high-concentration-range soil
PE samples containing weathered gasoline or diesel at two
moisture levels were analyzed. The PetroFLAG™ System
and reference method results were converted from a wet
weight basis to a dry weight basis in order to evaluate the
effect of moisture content on the sample TPH results. The
PetroFLAG™  System and reference method dry weight
TPH results were normally distributed; therefore, a two-
tailed, two-sample Student's  t-test was performed to
determine whether the device and reference method results
were  impacted by  soil moisture content—that is, to
determine whether an increase in soil moisture  content
resulted  in  an  increase  or decrease  in  the  TPH
concentrations measured. The null hypothesis for the t-test
was that the two means were equal or that the difference
between the means was equal to zero. Table 7-12 shows
the sample moisture levels, TPH results, mean TPH results
for sets of triplicate samples, whether the mean  TPH
results at different moisture levels were the same,  and the
probability of the null hypothesis being true.

Table 7-12 shows that PetroFLAG™ System results for
diesel  soil samples  at different moisture levels  were
statistically the same at a significance level of 5  percent
but that this was not the case for weathered gasoline soil
samples.  Therefore, the PetroFLAG™ System results for
weathered gasoline soil samples were impacted  by soil
moisture  content.  Based on a simple comparison of the
results, this conclusion appeared to be reasonable.

Table 7-12 also shows that reference method results for
weathered gasoline soil samples and diesel soil samples at
different moisture levels were statistically the same at a
significance level of 5  percent; therefore, the reference
method results were not impacted by soil moisture content.
Based  on  a simple comparison of the results,  this
conclusion appeared to be reasonable.

7.1.5  Primary Objective PS: Time Required for
       TPH Measurement

During the demonstration,  the time  required for  TPH
measurement activities, including PetroFLAG™  System
setup,  sample extraction and analysis and data package
preparation, and PetroFLAG™ System disassembly, was
measured.  During the  demonstration, two Dexsil  field
technicians performed the TPH measurement activities
using the PetroFLAG™  System.  Time measurement
began  at  the  start of each demonstration day when the
technicians began to set up the PetroFLAG™ System and
ended  when  they disassembled the device.  Tune not
measured included (1) the time spent by the technicians
verifying that they had received all  the demonstration
samples indicated on chain-of-custody forms and (2) the
times when both technicians took breaks.  In addition to
the total time required  for TPH measurement, the time
required to perform sample extraction, perform  sample
analysis, and prepare the data package for the first and last
analytical batches of soil samples was  measured.   The
                                                    87

-------
        Table 7-12.  Comparison of Results for Soil Performance Evaluation Samples at Different Moisture Levels
Sample Type and Moisture Level
Weathered gasoline at 9 percent
moisture level
Weathered gasoline at 16 percent
moisture level
Diesel at less than 1 percent
moisture level
Diesel at 9 percent moisture level
PetroFLAG™ System
TPH Result on Dry
Weight Basis
(milligram per
kilogram)
4,140
4.020
3,700
3,040
3,520
3.260
3,340
3,520
3,040
3,280
3,150
3,270
Mean TPH
Result
(milligram per
kilogram)
3,950
3,270
3,300
3,230
Were Mean TPH
Results at Different
Moisture Levels the
Same or Different?"
Different
Same
Probability of
Null Hypothesis
Being Trueb
(percent)
2.36
67.18
Reference Method
TPH Result on
Dry Weight Basis
(milligram per
kilogram)
2,070
2,220
2,400
2,070
2,390
2,440
2,740
3,180
3,070
2,720
2,970
3,100
Mean TPH
Result
(milligram per
kilogram)
2,230
2,300
3,000
2,930
Were Mean TPH
Results at Different
Moisture Levels the
Same or Different?*
Same
Same
Probability of
Null Hypothesis
Being Trueb
(percent)
66.52
71.95
00
oo
        Notes:
             A two-tailed, two-sample Student's t-test (parametric) was used to evaluate the effect of soil moisture content on TPH results.

             The null hypothesis for the t-test was that the two means were equal or that the difference between the two means was equal to zero.

-------
number and type of samples in a batch were selected by
Dexsil.

The time required to complete TPH measurement activities
using the PetroFLAG™ System is shown in Table 7-13.
When a given activity was performed by the two field
technicians simultaneously, the time measurement for the
activity was the total time spent by both technicians. The
time required for each activity was rounded to the nearest
5 minutes.

Overall, Dexsil required 50 hours, 40 minutes, for TPH
measurement of 66 soil environmental samples, 79 soil PE
samples, 36 liquid PE samples, and 10 extract duplicates.
Information  regarding  the  time  required  for  each
measurement   activity   during   the   entire   3-day
demonstration  and  for extraction,  analysis, and  data
package preparation for the first and last batches  of soil
samples is provided below.

PetroFLAG™  System setup required 25 to 40 minutes
each  day,  totaling 1 hour,  35  minutes, for the entire
demonstration.   This  activity  included PetroFLAG™
System setup; performing zero and standard calibrations of
the PetroFLAG™ Analyzer; performing calibration checks
of the digital balance; and organizing extraction, dilution,
analysis, and waste disposal supplies. The setup time was
                   measured on days 2 (25 minutes) and 3 (40 minutes) of the
                   demonstration; therefore, the average setup time rounded
                   to the nearest 5 minutes (30 minutes) was used to estimate
                   the setup time for day 1 of the demonstration and the total
                   setup time for the 3-day demonstration.

                   The sample extraction, sample analysis, and data package
                   preparation times were not separately measured during the
                   demonstration because  these  activities were performed
                   concurrently by two field technicians. To report 191 TPH
                   results,   Dexsil   performed  221   sample   extractions,
                   39 dilutions, and 213 TPH analyses for soil samples and
                   45 dilutions and 50 TPH analyses for liquid PE samples.
                   A total of 47 hours, 10 minutes, was required to perform
                   these activities; therefore, an average of 10 minutes was
                   required to perform these activities for one sample.  The
                   263 TPH analyses included sample reanalyses as well as
                   analyses of  10  extract duplicates.   Throughout the
                   demonstration, both technicians performed measurement
                   activities.

                   The time required to perform extraction, analysis, and data
                   package preparation for the first and last batches of soil
                   samples was also recorded.  A total of 2 hours, 20 minutes,
                   was  measured for the first batch of samples, which
                   required 15  extractions,  11  dilutions,  and  16  TPH
                   analyses; therefore, an average of 9 minutes was required
Table 7-13. Time Required to Complete TPH Measurement Activities Using the PetroFLAG™ System
Time Required3
Measurement Activity
PetroFLAG™ System setup
Sample extraction and analysis and data
First Sample Batch11
30 minutes0
2 hours, 20 minutes
Last Sample Batch"
40 minutes
1 hour, 50 minutes
3-Day Demonstration Period
1 hour, 35 minutes'
47 hours, 1 0 minutes
package preparation
PetroFLAG™ System disassembly
Total
40 minutes'
3 hours, 30 minutes
40 minutes
3 hours, 10 minutes
1 hour, 55 minutes'
50 hours, 40 minutes
Notes:
    The time required for each activity was rounded to the nearest 5 minutes.
    The first sample batch required 15 soil sample extractions, 11 dilutions, and 16 TPH analyses. The last sample batch required 13 soil sample
    extractions and 13 sample extract TPH analyses.
    The setup time was measured on days 2 (25 minutes) and 3 (40 minutes) of the demonstration; therefore, the average setup time rounded to the
    nearest 5 minutes (30 minutes) was used to estimate the setup time for day 1  of the demonstration and the total setup time for the 3-day
    demonstration.
    The sample extraction, sample analysis, and data package preparation times were not separately measured during the demonstration because
    the field technicians performed these activities concurrently.
    The disassembly time was measured on days 2 (35 minutes) and 3 (40 minutes) of the demonstration; therefore, the average disassembly time
    rounded to the nearest 5 minutes (40 minutes) was used to estimate  the disassembly time for day 1 of the demonstration and the total
    disassembly time for the 3-day demonstration.
                                                      89

-------
to perform these activities  for one sample.  A total of
1 hour, 50 minutes, was measured for the last batch of
samples, which required  13 extractions and 13  TPH
analyses; therefore, an average of 8 minutes was required
to perform these activities for one sample.

The decrease in the average analysis time between the first
and last batches of soil samples appeared to be associated
with the additional dilutions performed for the first batch
of  samples.   The average analysis  time  for all the
demonstration samples (10 minutes) exceeded the average
analysis times for the first and last batches of samples and
also appeared to be associated with the additional dilutions
required throughout the demonstration.

PetroFLAG™  System  disassembly  required  35  to
40 minutes each day, totaling 1 hour, 55 minutes, for the
entire demonstration.  Disassembly included packing up
the PetroFLAG™ System and associated supplies required
for TPH  measurement.   The  disassembly time  was
measured on days 2 (35 minutes) and 3 (40 minutes) of the
demonstration;  therefore,  the average disassembly time
rounded to the nearest 5 minutes (40 minutes) was used to
estimate the  disassembly  time   for  day  1 of the
demonstration and the total disassembly time for the 3-day
demonstration.

During  the weeks following the demonstration, Dexsil
spent additional time revising the data package to address
EPA  comments.  The revisions  included correcting
calculation and typographical errors for a few samples.
The amount of additional time that Dexsil spent finalizing
the data package could not be quantified and was not
included  as  part  of the  time  required  for   TPH
measurement.

For the reference method, time measurement began when
the  reference laboratory received  all  the  investigative
samples and continued until the EPA received the first
draft data package from the laboratory.  The reference
laboratory took 30 days to deliver the  first draft data
package to the EPA.  Additional time  taken by the
reference laboratory to address EPA comments on all the
draft laboratory data packages was not included as part of
the time required for TPH measurement.

7.2     Secondary Objectives

This section discusses  the  performance results for the
PetroFLAG™ System in terms of the secondary objectives
stated in Section 4.1. The  secondary objectives  were
addressed based on (1) observations of the PetroFLAG™
System's performance during  the  demonstration  and
(2) information provided by Dexsil.

7.2.1  Skill and Training Requirements for
       Proper Device Operation

Based on observations made during the demonstration, the
PetroFLAG™ System is easy to operate, requiring one
field technician with basic wet chemistry skills acquired on
the job or in a university. For the demonstration, Dexsil
chose to conduct sample analyses using two technicians in
order to increase the sample throughput.  One technician
performed sample extractions while the other performed
sample analyses.

The  sample analysis  procedure for the PetroFLAG™
System can be learned in the field with a few practice
attempts. The device contains a user guide that must be
followed to properly operate the device and a free training
videotape and  interactive compact  disk.  In addition,
during regular business hours, Dexsil provides technical
support  over   the telephone  at no  additional  cost.
Depending on the size of a project, Dexsil also provides
on-site training upon request at no additional cost.

The carrying case of the PetroFLAG™ System contains a
foam workstation that allows the user to keep enough items
to perform  10  analyses organized and easily accessible.
Also, items in the  PetroFLAG™ Reagents  kit  and
PetroFLAG™   High-Range  Extraction  Vials  kit are
organized in a configuration that resembles a test tube
rack.  Reagents in the PetroFLAG™ System are provided
in premeasured, sealed vials. As a result, the likelihood of
user error during sample analysis is minimized. The filter-
syringe assembly provided as part of the device is small,
lightweight, and designed such that little effort is required
to filter sample extracts.

The PetroFLAG™ Analyzer allows a user in the field to
select the response factor that is appropriate  for the PHCs
suspected to be present at a site.   Table  2-1 lists the
device's  response factors,  which range  from  2 for
weathered gasoline to 10 for transformer oil. Selection of
an appropriate  response factor  is critical for obtaining
accurate TPH results using the PetroFLAG™ System and
may require significant user experience, especially when
the sample contains a mixture of PHCs.

The response factor can be changed at any  time without
affecting the calibration data stored in the analyzer. The
                                                    90

-------
microprocessor in the analyzer uses the calibration data
and the response factor selected by the user to convert the
optical reading into a TPH concentration in mg/kg. The
user is not required to perform any calculations unless a
soil to solvent mass ratio other than one is used to measure
the TPH concentration or the sample extract is diluted. In
such cases, the TPH concentration reported by the analyzer
must be multiplied by the dilution factor used. After the
demonstration, Dexsil made minimal revisions to the TPH
results reported in the field.  Specifically, of the 191 TPH
results reported in the field at the end of the demonstration,
fewer than 5 percent were corrected based on EPA review
of the data package.  The corrections primarily involved
transcription errors.

To  assist  the  user in   obtaining  accurate  TPH
measurements, the PetroFLAG™ Analyzer automatically
performs several internal check s and informs the user when
an  error  has  occurred.    For example,  if  a  TPH
concentration is outside the linear range of the analyzer but
is still within the quantifiable range, the analyzer displays
a blinking reading. If the TPH concentration is outside the
quantifiable  range of the analyzer, the analyzer displays a
blinking   "EEEE."  According  to  Dexsil,  acceptable
precision  for  the linear  and  quantifiable ranges is
±10 percent  and ±20 percent precision, respectively.

For cases in which a TPH  concentration is outside the
linear or quantifiable range of the PetroFLAG™ Analyzer,
the user must select the most appropriate corrective action
from among  several options. According to the Dexsil user
guide, the user may (1) decrease the soil to solvent mass
ratio when the proprietary methanol mixture is used as the
solvent or (2)  use the high-range extraction  solvent
(proprietary glycol ether) to  generate 1 mL of extract that
is subsequently diluted hi  10 mL of proprietary methanol
mixture.  Although it is not  mentioned in the user guide,
when the TPH concentration is outside the linear range but
within the quantifiable range, the user may also dilute the
sample extract produced using the proprietary methanol
mixture and reanalyze the extract, thus eliminating the
re-extraction step and reducing the amount of additional
supplies required to perform the TPH measurement.

The  PetroFLAG™ Analyzer is also equipped  with a
built-in temperature sensor that measures the ambient
temperature during TPH measurement. The analyzer uses
the  sensor's  temperature   readings   to  correct  for
measurement  fluctuations   caused  by  temperature
variations. However, the temperature corrections are valid
only for ambient  temperatures  within  10  °C  of the
calibration temperature.    Therefore,  if  the  ambient
temperature varies from the calibration temperature by
more than 10 °C, an  error condition  results, and the
analyzer has to be recalibrated.

The PetroFLAG™ Analyzer also automatically performs
an  optical  system  check.   The  analyzer has  two
independent optical channels.  During TPH measurement,
if the analyzer responses on the two channels are different,
the analyzer displays an error message.

7.2.2  Health and Safety Concerns Associated
       with Device Operation

Sample analysis using the PetroFLAG™ System requires
handling  small  quantities  of a potentially hazardous
proprietary methanol mixture (extraction solvent and zero
calibration standard),  proprietary  glycol  ether  (high-
concentration-range extraction solvent), and proprietary
developer solution in sealed containers. Therefore, the
user should  employ good laboratory  practices  during
sample analysis.  Example guidelines for good laboratory
practices are  described in ASTM's "Standard Guide for
Good Laboratory Practices in Laboratories Engaged in
Sampling and Analysis of Water" (ASTM 1998).

During  the  demonstration,  Dexsil field technicians
operated the PetroFLAG™ System in modified Level D
personal protective equipment (PPE) to prevent eye and
skin contact  with reagents.  The PPE included safety
glasses, disposable gloves, work boots, and work clothes
with  long pants.   Sample analyses  were performed
outdoors in a well-ventilated area; therefore, exposure to
the extraction solvents, zero calibration  standard, and
developer solution through inhalation was not a concern.
Health and  safety  information  for  these reagents is
included in material safety data sheets available from
Dexsil. In addition, the user should exercise caution when
handling the extraction and developer solution vials, which
are made of glass.

7.2.3  Portability of the Device

The PetroFLAG™ System is easily transported between
sampling areas in the field. Each PetroFLAG™ System
starter kit weighs 11 pounds  and is housed in a  plastic
carrying case that is 14 inches long, 19 inches wide, and
5.5 inches high.  The PetroFLAG™  Analyzer alone
weighs 0.6 pound and is 5.75 inches long, 3.5 inches wide,
and 2 inches high.   Each  PetroFLAG™ Reagents kit,
which contains enough supplies for 10 analyses, weighs
                                                    91

-------
5  pounds and  is housed in  a cardboard box  that is
13 inches long,  16 inches wide, and 9 inches high. Each
PetroFLAG™ High-Range Extraction  Vials kit, which
contains  enough   supplies  to  perform   10   high-
concentration-range extractions, weighs less than 3 pounds
and is housed in a cardboard box that is 12 inches long,
8 inches wide, and 8 inches high. The Eppendorf™ 1-mL
pipettor, which can be purchased separately from Dexsil,
weighs less than 1 pound and is housed in a cardboard box
that is 12 inches long, 8 inches wide, and 8 inches high.
The PetroFLAG™ Analyzer, ACCULAB ® digital balance,
and digital timer are battery-operated.  Because no AC
power source is required, the system can  be easily
transported between sampling areas.

To operate the PetroFLAG™ System, a sample preparation
and analysis  area  is required  that is  large enough  to
accommodate the items  in one PetroFLAG™ System
starter kit. A staging area may also be  required to store
PetroFLAG™ Reagents kits and  PetroFLAG™ High-
Range Extraction Vials kits; the size of the staging area
depends on the number of samples to be analyzed and is
thus project-specific.   For the demonstration,  Dexsil
performed sample preparation and analysis under one 8- by
8-foot tent that  housed two 8-foot-long, folding  tables;
three  folding  chairs;  one 20-gallon laboratory pack for
flammable waste; and one 55-gallon drum for general
refuse.

7.2.4  Durability of the Device

The PetroFLAG™ System starter kit  contains several
reusable items,  including the PetroFLAG™  Analyzer,
ACCULAB®  digital  balance,  and digital timer.   An
Eppendorf™  1-mL pipettor, a reusable item, is used to
perform high-concentration-range extractions.  Based on
observations  made  during   the   demonstration,  the
PetroFLAG™ System and the pipettor are durable; none of
the reusable items malfunctioned or was damaged.  The
reusable items are manufactured or distributed by Dexsil
or by scientific equipment suppliers and, except for the
pipettor, are housed by Dexsil hi a hard-plastic carrying
case to prevent damage to the items during transport. The
items  were unaffected by the varying  temperature and
humidity conditions encountered between 8:00 a.m. and
5:00 p.m. on any given day of the demonstration. During
the daytime, the temperature ranged from about  17 to
24 °C, while the relative humidity ranged from 53 to
88 percent. During sample analysis, the light, disposable
items in the device were  kept in the racks of the
PetroFLAG™ Reagents kits, so wind speeds up to 20 miles
per hour did not affect device operation.

7.2.5  Availability of the Device and Spare Parts

During the demonstration, neither the Eppendorf™ 1-mL
pipettor nor any of the reusable items in the PetroFLAG™
System starter kit required replacement. Had any of these
items required replacement, only a digital timer would
have been available in local stores.  A replacement  item
can be obtained from Dexsil by  overnight courier service
if the order is placed by 2:00 p.m. eastern time. If the need
for a replacement item is identified after 2:00 p.m. eastern
time and the item is needed the next day, each item may be
obtained from various distributors of Dexsil products; the
ACCULAB® digital balance and  Eppendorf™ 1-mL
pipettor may also be obtained from scientific equipment
suppliers.   Spare parts for reusable items  such as the
PetroFLAG™  Analyzer are not provided by Dexsil.
Dexsil recommends that malfunctioning reusable items be
returned to Dexsil for service; according to Dexsil, repairs
should not be attempted in the field by the user. Because
Dexsil provides a 6-month warranty for reusable items,
Dexsil will replace such items and supply them to the user
by overnight courier service at no additional cost during
the warranty period.

The  power  supplies for the PetroFLAG™  Analyzer
(one 9-volt battery), ACCULAB® digital balance  (two
CR2032 cell batteries), and digital timer (one G-13 button
cell battery) can be purchased from local stores  and
replaced in the field if necessary.

Disposable items in the PetroFLAG™ System should be
obtained from Dexsil or its authorized distributors. All the
disposable items, including the breaktop  vials containing
proprietary methanol  extraction solvent  mixture, are
manufactured only  by  Dexsil.  The disposable items
provided to a given user on a given occasion all come from
the same lot. Because Dexsil conducts QC checks for each
lot individually, if the user performs analyses with items
from more than one lot or uses reagents obtained from a
scientific supply store, Dexsil assumes no responsibility
for the quality of the sample analysis results.
                                                   92

-------
                                              Chapter 8
                                         Economic Analysis
As discussed throughout this ITVR, the PetroFLAG™
System was demonstrated by using it  to analyze soil
environmental samples, soil PE samples, and liquid PE
samples. The environmental samples were collected from
three contaminated sites,  and the PE  samples  were
obtained from a commercial provider, ERA. Collectively,
the environmental and PE samples provided the different
matrix types and the different levels and types of PHC
contamination  needed  to perform  a  comprehensive
economic analysis for the PetroFLAG™ System.

During the demonstration, the PetroFLAG™ System and
the off-site laboratory reference method were each used to
perform nearly 200 TPH analyses.  The purpose of the
economic analysis was to estimate the total cost of TPH
measurement for  the PetroFLAG™ System and then
compare this cost to that for the reference method.  The
cost per analysis was not estimated for the PetroFLAG™
System because the cost per analysis would increase as the
number of samples analyzed decreased.  This increase
would be primarily the result of the distribution of the
initial capital equipment cost  across a smaller number of
samples. Thus, this increase in the cost per analysis cannot
be fairly compared to the reference laboratory's fixed cost
per analysis.

This chapter provides  information  on  the  issues and
assumptions   involved  in  the  economic  analysis
(Section 8.1), discusses the costs associated with using the
PetroFLAG™  System (Section 8.2), discusses the  costs
associated with using the reference method (Section 8.3),
and presents a comparison of the economic analysis results
for the PetroFLAG™ System and the reference method
(Section 8.4).
8.1    Issues and Assumptions

Several factors affect TPH measurement costs. Wherever
possible in this chapter, these factors are identified in such
a way that decision-makers can independently complete a
project-specific economic analysis.  The following five
cost categories were included in the economic analysis for
the demonstration: capital equipment, supplies, support
equipment, labor, and IDW disposal.  The issues and
assumptions associated with these categories and the costs
not included in the analysis are briefly discussed below.
Because the reference method costs were based on a fixed
cost per analysis, the issues and assumptions discussed
below apply only  to the PetroFLAG™ System unless
otherwise stated.

8.1.1  Capital Equipment Cost

The capital equipment cost was the cost associated with
purchase of the  PetroFLAG™ System starter  kit used
during the demonstration.   The  purchase price  was
obtained from a standard price list provided by Dexsil.

8.1.2  Cost of Supplies

The cost of supplies was estimated based on the supplies
required to analyze all demonstration samples using the
PetroFLAG™ System that were not included in the capital
equipment cost category. Supplies used by Dexsil during
the demonstration included (1) PetroFLAG™ Reagents
kits, (2) PetroFLAG™ High-Range Extraction Vials kits,
and (3)  an Eppendorf™ 1-mL pipettor to accurately
measure and transfer  high-concentration-range sample
extract into proprietary methanol solvent mixture. During
                                                   93

-------
the demonstration, the quantity of supplies used by Dexsil
were noted each day.  The purchase price of these supplies
was obtained from a standard price list provided by Dexsil.
Because a user  cannot  return the unused supplies, no
salvage value for supplies that were not used during the
demonstration was included in the cost of supplies.

8.1.3   Support Equipment Cost

Because of the large number of samples analyzed during
the demonstration, the EPA  provided support equipment,
including a  tent, tables,  and  chairs, for  the  field
technicians'  comfort during sample  extraction  and
analysis.   For  the  economic  analysis,  the  support
equipment costs were estimated based on price quotes
from independent sources.

8.1.4   Labor Cost

The labor cost was estimated based on the time required
for PetroFLAG™ System  setup,  sample  preparation,
sample analysis,  and  summary data package preparation.
The  data package included,  at a minimum,  a result
summary  table,   a run   log,  and any supplementary
information submitted by Dexsil. The measurement of the
time required for Dexsil to complete all analyses and
submit the data package to the EPA was rounded to the
nearest half-hour. However, for the economic analysis, it
was assumed that a field technician who had worked for a
fraction of a day would be paid for an entire 8-hour day.
Based on this assumption, a daily rate for a field technician
was used in the analysis.

During the demonstration, EPA representatives evaluated
the skill level required for the field technicians to complete
analyses and calculate TPH concentrations.  Based on the
field  observations, a field technician  with basic wet
chemistry skills acquired on the job or in a university and
a few hours of device-specific  training was considered to
be qualified to operate the PetroFLAG™ System.  For the
economic analysis, an hourly rate of $16.63 was used for
a field technician (R.S. Means Company [Means] 2000),
and a multiplication factor of 2.5 was applied to labor costs
in order to account for  overhead costs.  Based on this
hourly rate  and  multiplication factor,  a daily  rate of
$332.60 was used for the economic analysis.

8.1.5  Investigation-Derived Waste Disposal Cost

During the demonstration, Dexsil was provided with two
20-gallon laboratory packs for collecting hazardous wastes
generated (one for flammable wastes and one for corrosive
wastes) and was charged for each laboratory pack used.
Unused samples and  sample  extracts, residual solvent
generated from sample  extractions and dilutions,  used
EnCores, and unused chemicals that could not be returned
to Dexsil were disposed of in a laboratory pack.  Dexsil
was required  to provide any  containers necessary to
containerize individual wastes prior to their placement in
a  laboratory  pack; however,  Dexsil did not require
additional containers.   Items such as used PPE  were
disposed of with municipal garbage hi accordance with
demonstration  site  waste  disposal  guidelines;  the
associated waste disposal cost was not included in the IDW
disposal cost estimate.

8.1.6   Costs Not Included

Items whose costs were not included in  the economic
analysis are identified below along with a rationale for the
exclusion of each.

Oversight of Sample Analysis Activities.  A typical user
of the PetroFLAG™ System would not be required to pay
for customer  oversight of sample   analysis.    EPA
representatives audited all activities associated with sample
analysis during the demonstration, but costs  for  EPA
oversight were not included in the  economic analysis
because these  activities  were project-specific.  For the
same  reason,  costs for EPA oversight of the reference
laboratory were also not included in the analysis.

Travel  and Per Diem  for Field Technicians.    Field
technicians may  be  available  locally.   Because the
availability of field technicians is primarily a function of
the location of the project site, travel and per diem costs
for field technicians were not included in the economic
analysis.

Sample Collection and Management. Costs for sample
collection and management activities,  including sample
homogenization and labeling, were not included in the
economic analysis because these activities were project-
specific and  were  not  device-  or reference  method-
dependent.

Shipping.  Costs  for shipping  (1)  the PetroFLAG™
System and necessary  supplies to the demonstration site
and (2) sample coolers to the reference laboratory were not
included in the economic analysis because such costs vary
depending on the shipping distance and the service used
                                                    94

-------
(for example, a courier  or  overnight shipping versus
economy shipping).

Items Costing Less Than $10.  The cost of inexpensive
items such as ice used for sample preservation in the field
was not included in  the economic analysis because the
estimated cost was less than $ 10.

8.2    PetroFLAG™ System Costs

This section presents information on the individual costs of
capital equipment, supplies, support equipment, labor, and
IDW disposal for the PetroFLAG™ System  as well
as a summary of these costs.  Additionally, Table 8-1
summarizes the PetroFLAG™ System costs.

8.2.1  Capital Equipment Cost

The capital equipment cost was the cost associated with the
purchase  of the  PetroFLAG™  System  starter  kit
(Item No. PETROFLAG™-MTR-01).  Dexsil does not
rent the PetroFLAG™ System starter kit.  Table 2-2  lists
the components  of the kit, which contains the  reusable
equipment and disposable supplies required to perform
10  TPH measurements.   Supplies required to perform
additional  measurements  are sold  separately.    The
PetroFLAG™ System starter kit can be purchased from
Dexsil for $695.
Reusable items of the PetroFLAG™ System starter kit that
can be purchased  separately  from Dexsil  if additional
quantities are needed include the PetroFLAG™ analyzer
($530), ACCULAB * digital balance ($ 132.50), and digital
timer ($10.60).  During the demonstration, no additional
quantities of these items were required.

8.2.2  Cost of Supplies

Supplies  used during  the demonstration  included the
following: (1) PetroFLAG™ Reagents kits  (Item No.
PF-SRP); (2)  an Eppendorf™ 1-mL pipettor (Item No.
PC-1ML-PP),  and  (3)  PetroFLAG™   High-Range
Extraction Vials kits (Item No. PF-HRD). All supplies are
available from Dexsil.  The  cost for the  supplies are
described below.

During the demonstration, Dexsil used 28 PetroFLAG™
Reagents kits.  Table 2-2 lists the items included in a
PetroFLAG™ Reagents kit.  Each kit includes enough
disposable supplies to perform 10 soil analyses. Because
the price per kit decreases as the number of kits purchased
increases, the purchase price per kit was $ 106. To perform
high-concentration-range extractions,  Dexsil used  an
Eppendorf™ 1-mL pipettor ($110) and four PetroFLAG™
High-Range Extraction Vials kits ($20 each).   Each
PetroFLAG™ High-Range Extraction Vials kit contains
enough disposable extraction vials and replacement tips for
Table 8-1. PetroFLAG™ System Cost Summary
Item
   Quantity
Unit Cost ($)
Itemized Cost3 ($)
Capital equipment
    Purchase of PetroFLAG™ System starter kit                1 unit
Supplies
    PetroFLAG™ Reagents kit (supplies for 10 soil analyses)      28 units
    PetroFLAG™ High-Range Extraction Vials kit (supplies for 10    4 units
                      695


                      106
                      20
                      695


                     2,968
                       80
high-concentration-range extractions)
Eppendorf™ 1-mL pipettor
Support equipment
Tent
Tables and chairs (two each)
Labor
Field technicians
Investigation-derived waste disposal
Total Cost"

1 unit

1 unit
1 set for 1 week

6 person-days
1 20-gallon container


110

159
39

332.60
345


110

159
39

1,996
345
$6,390
Notes:
    Itemized costs were rounded to the nearest $1.
    The total dollar amount was rounded to the nearest $10.
                                                    95

-------
the pipettor to perform 10 high-range extractions. Thus,
the total cost of supplies used  by Dexsil during the
demonstration was $3,158.
8.2.3  Support Equipment Cost

Dexsil was provided with one 8- by  8-foot tent for
protection  from   inclement  weather  during   the
demonstration as well as two tables and two chairs for use
during sample preparation and analysis activities.  The
purchase cost for the tent ($159) and the rental cost for two
tables and two chairs for 1 week ($39) totaled $198.

8.2.4  Labor Cost

Two field technicians were required for 3 days each during
the demonstration to complete all sample analyses  and
prepare the summary data package. Based on a daily labor
rate of $332.60 per person, the total labor cost for the
PetroFLAG™ System was $ 1,996 (rounded to the nearest
$1).

8.2.5  Investigation-Derived Waste Disposal Cost

Dexsil used one laboratory pack to collect flammable
hazardous waste generated during the demonstration.  The
IDW  disposal cost  included the purchase cost  of the
laboratory  pack ($38)  and  the cost  associated  with
hazardous waste disposal  in  a  landfill ($307) (Means
2000). The total IDW disposal cost was $345.

8.2.6  Summary ofPetroFLA G™ System Costs

The total cost for performing nearly  200 TPH analyses
using the  PetroFLAG™ System and  for preparing a
summary data package was $6,390 (rounded to the nearest
$10).   The TPH analyses were performed for 66  soil
environmental samples, 79 soil PE samples, and 36 liquid
PE samples. In addition to these 181 samples,  10 extract
duplicates were analyzed for specified soil environmental
samples.   When Dexsil performed multiple extractions,
dilutions, reanalyses, or field duplicates for a sample, these
were not included in the number of samples  analyzed.
During  the  demonstration,  the multiple  extractions,
dilutions,  reanalyses, and  field duplicates collectively
required 9 additional PetroFLAG™ Reagents kits (about
47 percent more).

The total cost of $6,390 for analyzing the demonstration
samples using the PetroFLAG™ System included $695 for
capital equipment; $3,158 for supplies; $198 for support
equipment; $1,996 for labor; and $345 for IDW disposal.
Of these five costs, the two  largest were the  cost of
supplies (49 percent of the total cost) and the labor cost
(31 percent of the total cost).

8.3    Reference Method Costs

This  section presents  the costs  associated with the
reference  method used  to analyze  the  demonstration
samples for TPH.  Depending on the nature of a given
sample, the reference laboratory analyzed the sample
for GRO,  EDRO,  or  both  and  calculated the  TPH
concentration by  adding the GRO and EDRO  concen-
trations, as appropriate. The reference method costs were
calculated using unit cost information from the reference
laboratory invoices. To allow an accurate comparison of
the PetroFLAG™ System and reference method costs, the
reference  method costs  were  estimated  for the  same
number of samples as was analyzed by Dexsil.   For
example,  although  the  reference  laboratory analyzed
MS/MSD samples for TPH and all soil samples for percent
moisture, the associated sample analytical costs were not
included in the reference method costs because Dexsil did
not analyze MS/MSD samples for TPH or soil samples for
percent moisture during the demonstration.

Table 8-2 summarizes the reference method costs, which
totaled $38,560.   This  cost  covered  preparation of
demonstration samples and their analysis for TPH.  In
addition, at no additional cost, the reference laboratory
provided (1) analytical  results  for internal QC check
samples such as method blanks and LCS/LCSDs and (2) an
electronic  data deliverable  and two paper copies of full,
EPA Contract Laboratory Program-style  data packages
within 30  calendar days  of  the receipt  of the last
demonstration sample by the reference laboratory.

8.4    Comparison of Economic Analysis
       Results

The total costs for the PetroFLAG™ System ($6,390) and
the reference method ($3 8,560) are listed in Tables 8-1 and
8-2, respectively. The total TPH measurement cost for the
PetroFLAG™ System was 83 percent less than that for the
reference method.  Although the PetroFLAG™  System
did not provide the same level of detail (for example,
carbon ranges) as  the reference method analytical results
or comparable QA/QC data, the PetroFLAG™  System
provided TPH analytical results on site at significant cost
savings. In addition, use of the  PetroFLAG™ System in
                                                   96

-------
Table 8-2. Reference Method Cost Summary
Item
Soil environmental samples
GRO
Extract duplicates
EDRO
Extract duplicates
Soil performance evaluation samples
GRO
EDRO
Liquid performance evaluation samples
GRO
EDRO
Total Cost"
Number of Samples Analyzed

56
9
66
10

45
79

27
24

Cost per Analysis ($)

111
55.50
142
71

111
142

111
106.50

Itemized Cosf ($)

6,216
500
9,372
710

4,995
11,218

2,997
2,556
$38,560
Notes:

"    Itemized costs were rounded to the nearest $1.
"  '  The total dollar amount was rounded to the nearest $10.
the field will likely  produce  additional  cost  savings
because the results will be available within a few hours of
sample collection; therefore, critical decisions regarding
sampling and analysis can be made in the field, resulting
in a more complete data set.   However, these savings
cannot be accurately estimated and thus were not included
hi the economic analysis.
                                                       97

-------
                                               Chapter 9
                               Summary of Demonstration Results
As discussed throughout this ITVR, the PetroFLAG™
System was demonstrated by using it to analyze 66 soil
environmental samples, 79 soil PE samples, and 36 liquid
PE samples. In addition to these 181 samples, 10 extract
duplicates prepared using the environmental samples were
analyzed. The environmental samples were collected from
four individual areas at three contaminated sites, and the
PE samples were obtained from a commercial provider,
ERA.  Collectively, the environmental and PE samples
provided the different matrix types and the different levels
and types of PHC contamination needed to perform a
comprehensive evaluation of the PetroFLAG™ System.

The PetroFLAG™ System performance and cost data were
compared to  those for an off-site laboratory reference
method, SW-846  8015B  (modified).   As discussed in
Chapter 6, the reference method results were considered to
be of adequate quality  for the following reasons: (1) the
reference  method  was  implemented  with acceptable
accuracy (±30 percent) for all the samples except low- and
medium-concentration-range soil  samples  containing
diesel, which made up only 13 percent of the total number
of samples analyzed during the demonstration, and (2) the
reference method was implemented with good precision
for all samples.  The reference method  results generally
exhibited  a negative  bias.   However,  the  bias was
considered  to be significant primarily  for low- and
medium-range soil  samples containing diesel.   The
reference  method  recoveries  observed  during the
demonstration were typical of the recoveries obtained by
most  organic analytical  methods for  environmental
samples.

This chapter compares the perfo rmance and cost results for
the PetroFLAG™ System  with those for the reference
method, as appropriate. The performance and cost results
are discussed in detail in Chapters 7 and 8, respectively.
Tables 9-1 and 9-2 summarize the results for the primary
and secondary objectives, respectively. As shown in these
tables, during the demonstration, the PetroFLAG™ System
exhibited the following desirable characteristics of a field
TPH measurement device: (1) good precision, (2) lack of
sensitivity  to  interferents  that  are  not  petroleum
hydrocarbons (PCE and hurnic acid), (3) low measurement
costs, and (4) ease of use.  hi addition, the PetroFLAG™
System exhibited moderate sample throughput.

Based on   action  level  conclusions  and  statistical
correlations,  the  PetroFLAG™  System  TPH  results
compared well with  those of  the  reference  method;
however, the device exhibited a high bias, and its TPH
results were determined to be statistically different from
those of the reference method. In addition, turpentine and
1,2,4-trichlorobenzene biased the PetroFLAG™ System
TPH results high; the reference method TPH results were
also biased high by turpentine and 1,2,4-trichlorobenzene,
but the extent of the  impact on the reference method was
significantly  less.   These findings  indicated that  the
accuracy of TPH measurement using the device may be
significantly  impacted by naturally occurring oil and
grease and chlorinated semivolatile organic contaminants
such as chlorinated pesticides and PCBs present in soil
samples.  Also, the  device exhibited sensitivity to soil
moisture content during TPH measurement of weathered
gasoline  soil  samples but  not diesel  soil  samples.
Specifically, the TPH results for weathered gasoline soil
samples were biased  low (17  percent) when the soil
moisture  content was  increased from 9 to 16 percent.
Collectively, the demonstration findings indicated that the
user should exercise caution when considering the device
for a specific field TPH measurement application.
                                                   98

-------
        Table 9-1.  Summary of PetroFLAG™ System Results for the Primary Objectives
         Primary Objective
                                       Evaluation Basis"
                                                                                                                          Performance Results
                                                            PetroFLAG™ System
                                                                                                                                .  Reference Method
         P1
Determine the method
detection limit
Method detection limit based on TPH analysis of
seven low-concentration-range diesel soil PE samples
                                                                                     20 mg/kg
6.32 mg/kg
         P2  Evaluate the accuracy
              and precision of TPH
              measurement
vo
                      Comparison of project-specific action level
                      conclusions of the PetroFLAG™ System with those of
                      the reference method for 66 soil environmental and
                      31 soil PE samples
                                                 71 of 97 PetroFLAG™ System conclusions (73 percent) agreed with those of the reference method;
                                                 26 PetroFLAG™ System conclusions were false positives.
                                    Comparison of PetroFLAG™ System TPH results with
                                    those of the reference method for 66 soil
                                    environmental and 25 soil PE samples
                                                                        11 of 91 PetroFLAG™ System results (12 percent) were within 30 percent of the reference method
                                                                        results; 7 PetroFLAG™ System results were biased high, and 4 were biased low.

                                                                        9 of 91 PetroFLAG™ System results (10 percent) were within 30 to 50 percent of the reference method
                                                                        results; 8 PetroFLAG™ System results were biased high, and 1 was biased low.

                                                                        71 of 91 PetroFLAG™ System results (78 percent) were not within 50 percent of the reference method
                                                                        results; 67 PetroFLAG™ System results were biased high, and 4 were biased low.
                      Pain/vise comparison of PetroFLAG™ System and
                      reference method'TPH results for (1) soil
                      environmental samples collected from four areas;
                      (2) soil PE samples, including blank, weathered
                      gasoline, and diesel soil samples; and (3) liquid PE
                      samples consisting of neat weathered gasoline and
                      diesel
                                                 For soil environmental samples, the PetroFLAG™ System results were statistically (1) the same as the
                                                 reference method results for one of the four sampling areas and (2) different from the reference method
                                                 results for three of the sampling areas.

                                                 For soil PE samples, the PetroFLAG™ System results were statistically (1) the same as the reference
                                                 method results for high-concentration-range diesel samples and (2) different from the reference method
                                                 results for blank samples, high-concentration-range weathered gasoline samples, and low- and medium-
                                                 concentration-range diesel samples.

                                                 For liquid PE samples, the PetroFLAG™ System results were statistically different from the reference
                                                 method results for both weathered gasoline and diesel samples.
                                    Correlation (as determined by linear regression
                                    analysis) between PetroFLAG™ System and
                                    reference method TPH results for (1) soil
                                    environmental samples collected from four areas and
                                    (2) soil PE samples, including weathered gasoline and
                                    diesel soil samples
                                                                        The PetroFLAG™ System results correlated highly with the reference method results for one of the four
                                                                        sampling areas and diesel soil PE samples (R2 values were greater than 0.90, and F-test probability
                                                                        values were less than 5 percent).

                                                                        The PetroFLAG™ System results correlated moderately with the reference method results for two of the
                                                                        four sampling areas (R2 values were 0.84 and 0.86, and F-test probability values were less than
                                                                        5 percent).

                                                                        The PetroFLAG™ System results correlated weakly with the reference method results for one of the four
                                                                        sampling areas and weathered gasoline soil PE samples (R2 values were 0.42 and 0.10, respectively,
                                                                        and F-test probability values were greater than 5 percent).

-------
        Table 9-1. Summary of PetroFLAG™ System Results for the Primary Objectives (Continued)
Primary Objective
P2 Evaluate the accuracy
and precision of TPH
measurement
(continued)
P3 Evaluate the effect of
interferents on TPH
measurement
P4 Evaluate the effect of
soil moisture content
on TPH measurement
Evaluation Basis"
Overall precision (RSD) for soil environmental, soil
PE, and liquid PE sample replicates
Analytical precision (RPD) for extract duplicates for
soil environmental samples (10 for the PetroFLAG™
System and 12 for the reference method)
Mean responses for neat materials, including MTBE;
PCE; Stoddard solvent; turpentine; and
1 ,2,4-trichlorobenzene, and for soil spiked with humic
acid (two triplicate sets each)
Comparison of TPH results (one-way analysis of
variance) for weathered gasoline and diesel soil PE
samples without and with interferents at two levels
Interferents for weathered gasoline soil PE samples:
MTBE, PCE, Stoddard solvent, and turpentine
Interferents for diesel soil PE samples: Stoddard
solvent; turpentine; 1,2,4-trichlorobenzene; and humic
acid
Comparison of TPH results (two-sample Student's
t-test) for weathered gasoline and diesel soil PE
samples at two moisture levels: 9 and 16 percent for
weathered gasoline samples and less than 1 and
9 percent for diesel samples
Performance Results
PetroFLAG™ System
Soil environmental samples (1 1 triplicates)
RSD range: 7 to 38 percent
Median RSD: 19 percent
Soil PE samples (7 replicates)
RSD range: 2 to 9 percent
Median RSD: 6 percent
Liquid PE samples (2 triplicates)
RSDs: 3 and 9 percent
Median RSD: 6 percent
RPD range: 1 to 26
Median RPD: 5
MTBE: 0 percent
PCE: 1.5 percent
Stoddard solvent: 42.5 percent
Turpentine: 103 percent
1,2,4-Trichlorobenzene: 16 percent
Humic acid: 2.5 percent
MTBE, a petroleum hydrocarbon, caused
statistically significant interference only at the
high level.
PCE results were inconclusive.
Stoddard solvent, a petroleum hydrocarbon,
caused statistically significant interference at
both levels for weathered gasoline and diesel
samples.
Turpentine caused statistically significant
interference at both levels for weathered
gasoline and diesel samples.
1,2,4-Trichlorobenzene caused statistically
significant interference at both levels.
Humic acid did not cause statistically
significant interference at either of the two
levels.
Soil moisture content had a statistically
significant impact on weathered gasoline
sample results.
Reference Method
Soil environmental samples (1 1 triplicates)
RSD range: 4 to 39 percent
Median RSD: 16 percent
Soil PE samples (6 replicates)
RSD range: 6 to 13 percent
Median RSD: 8 percent
Liquid PE samples (2 triplicates)
RSDs: 5 and 6 percent
Median RSD: 5.5 percent
RPD range: 0 to 1 1
Median RPD: 4
MTBE: 39 percent
PCE: 17. 5 percent
Stoddard solvent: 85 percent
Turpentine: 52 percent
1 ,2,4-Trichlorobenzene: 50 percent
Humic acid: 0 percent
MTBE, a petroleum hydrocarbon, did not cause
statistically significant interference at either of the two
levels.
PCE caused statistically significant interference only at
the high level.
Stoddard solvent, a petroleum hydrocarbon, caused
statistically significant interference at both levels for
weathered gasoline and diesel samples.
Turpentine caused statistically significant interference
(1) at both levels for weathered gasoline samples and
(2) only at the high level for diesel samples.
1 ,2,4-Trichlorobenzene caused statistically significant
interference only at the high level.
Humic acid results were inconclusive.
Soil moisture content did not have a statistically
significant impact.
o
o

-------
Table 9-1. Summary of PetroFLAG™ System Results for the Primary Objectives (Continued)
Primary Objective
P5 Measure the time
required for TPH
measurement (sample
throughput)
P6 Estimate TPH
measurement costs
Evaluation Basis"
Total time from sample receipt through preparation of
the draft data package
Total cost (costs of capital equipment, supplies,
support equipment, labor, and IDW disposal) for TPH
measurement of 66 soil environmental samples,
79 soil PE samples, 36 liquid PE samples, and
1 0 extract duplicates
Performance Results
PetroFLAG™ System
50 hours, 40 minutes, for TPH measurement
of 66 soil environmental samples, 79 soil PE
samples, 36 liquid PE samples, and 10 extract
duplicates
$6,390 (including the capital equipment
purchase cost of $695 for the PetroFLAG™
System starter kit)
Reference Method
30 days for TPH measurement of 74 soil environmental
samples, 89 soil PE samples, 36 liquid PE samples, and
13 extract duplicates
$38,560
Notes:

IDW   =
mg/kg =
MTBE =
PCE   =
Investigation-derived waste      PE
Milligram per kilogram           R2
Methyl-tert-butyl ether           RPD
Tetrachloroethene              RSD
Performance evaluation
Square of the correlation coefficient
Relative percent difference
Relative standard deviation
     All statistical comparisons were made at a significance level of 5 percent.

-------
Table 9-2. Summary of PetroFLAG™ System Results for the Secondary Objectives
Secondary Objective
S1 Skill and training
requirements for proper
device operation
S2 Health and safety concerns
associated with device
operation
S3 Portability of the device
S4 Durability of the device
S5 Availability of device and
spare parts
Performance Results
Trie device can be operated by one person with basic wet chemistry skills.
The device's user guide, training videotape, and interactive compact disk are considered to be adequate
training materials for proper device operation. The sample analysis procedure for the device can be learned
in the field with a few practice attempts.
The device's digital display provides a sample's TPH concentration in milligrams per kilogram on a wet
weight basis; minimal effort is required to calculate a TPH concentration when a soil to solvent mass ratio
other than one is used or when sample extract dilution is involved. At the end of the demonstration, Dexsil
reported 191 TPH results. Of these, fewer than 5 percent required corrections, which primarily involved
transcription errors.
No significant health and safety concerns were noted; when the device is used in a well-ventilated area,
basic eye and skin protection (safety glasses, disposable gloves, work boots, and work clothes with long
pants) should be adequate for safe device operation.
The device is battery-operated and requires no alternating current power source. The device can be easily
moved between sampling areas in the field, if necessary.
The device is provided in a hard-plastic carrying case to prevent damage to the device. During the
demonstration, none of the device's reusable items malfunctioned or was damaged. The moderate
temperatures (1 7 to 24 °C) and high relative humidities (53 to 88 percent) encountered during the
demonstration did not affect device operation.
All items in the device are available from Dexsil. During a 6-month warranty period, Dexsil will supply
replacement parts for the device by overnight courier service at no cost unless the reason for a part failure
involves misuse of the device.
                                                         102

-------
                                             Chapter 10
                                             References
AEHS.  1999.  "State Soil Standards Survey."  Soil &
    Groundwater. December 1999/January 2000.

API. 1994. "Intel-laboratory Study of Three Methods for
    Analyzing  Petroleum  Hydrocarbons  in  Soils."
    Publication Number 4599. March.

API. 1996. "Compilation of Field Analytical Methods for
    Assessing Petroleum Product Releases." Publication
    Number 4635. December.

API.  1998.  "Selecting Field Analytical Methods: A
    Decision-Tree Approach." Publication Number 4670.
    August.

ASTM.  1998. "Standard Guide  for Good Laboratory
    Practices in Laboratories Engaged in Sampling and
    Analysis of Water." Designation: D 3856-95. Annual
    Book of ASTM Standards.  Volume 11.01.

California  Environmental Protection Agency.   1999.
    Memorandum Regarding  Guidance for Petroleum
    Hydrocarbon  Analysis.  From Bart Simmons, Chief,
    Hazardous  Materials  Laboratory.  To Interested
    Parties. October 21.

Dexsil.  1997.  "PetroFLAG™  Hydrocarbon Analyzer
    User's Manual." May 6.

Dryoff, George V. Editor. 1993. "Manual of Significance
    of Tests for Petroleum Products." ASTM Manual
    Series: MNL  1. 6th Edition.

EPA.  1983.  "Methods for Chemical Analysis of Water
    and Waste."  Revision. Environmental Monitoring
    and Support  Laboratory.  Cincinnati,  Ohio.  EPA
    600-4-79-020. March.
EPA. 1996. "Test Methods for Evaluating Solid Waste."
    Volumes 1A through 1C. SW-846. Third Edition.
    Update HI. OSWER. Washington, DC.  December.

EPA. 2000. "Field Measurement Technologies for Total
    Petroleum  Hydrocarbons in  Soil—Demonstration
    Plan." ORD. Washington, DC. EPA/600/R-01/060.
    June.

ESC. 2001.  E-mail Regarding Cost of Supplies Used
    During the Demonstration. From Matt Teglas, ESC.
    To Sandy  Anagnostopoulos, Tetra Tech EM  Inc.
    April 17.

Florida Department of Environmental Protection.  1996.
    "FL-PRO Laboratory  Memorandum."   Bureau of
    Waste Cleanup.   Accessed  on April 21. On-Line
    Address: www.dep.state.fl.us/labs/docs/flpro.htm

Fritz, James  S., and  George  H.  Schenk.    1987.
    Quantitative Analytical Chemistry. Allyn and Bacon,
    Inc. Boston, Massachusetts.  Fifth Edition.

Gary, J.H., and G.E. Handwerk. 1993.   Petroleum
    Refining: Technology and Economics. Marcel Dekker,
    Inc. New York, New York.

Massachusetts Department of Environmental Protection.
    2000. "VPH/EPH Documents." Bureau of Waste Site
    Cleanup.  Accessed on April 13. On-Line Address:
    www.state.ma.us/dep/bwsc/vp_eph.htm

Means.    2000.    Environmental  Remediation  Cost
    Data-Unit Price. Kingston, Massachusetts.

Provost,  Lloyd  P.,  and Robert  S. Elder.    1983.
    "Interpretation of Percent Recovery Data." American
    Laboratory. December. Pages 57 through 63.
                                                 103

-------
Speight, J.G.  1991.  The Chemistry and Technology of    Texas Natural Resource Conservation Commission. 2000.
   Petroleum. Marcel Dekker, Inc.  New York, New        "Waste Updates."  Accessed on April 13.  On-Line
   York.                                                Address:   www.tnrcc.state.tx.us/permitting/
                                                        wastenews.htm#additional
Tetra  Tech EM  Inc.  2001.   Record  of Telephone
   Conversation Regarding Blank Soil PE Samples.     Zilis, Kimberly, Maureen McDevitt, and Jerry Parr. 1988.
   Between  Kirankumar  Topudurti,  Environmental        "A Reliable Technique  for Measuring Petroleum
   Engineer,  and Jeff Lowry, Environmental Resource        Hydrocarbons in the Environment." Paper Presented
   Associates. June 10.                                   at the Conference on Petroleum Hydrocarbons and
                                                        Organic Chemicals in Groundwater. National Water
                                                        Well Association (Now Known as National Ground
                                                        Water Association). Houston, Texas.
                                                 104

-------
                                                Appendix
                     Supplemental Information Provided by the Developer
This appendix contains supplemental information provided
by Dexsil. After the introduction, this appendix discusses
the PetroFLAG™ System, the effect of naturally occurring
organic compounds  on  TPH measurement using  the
device, and the importance of the choice of the response
factor.

Introduction

Dexsil would like to  thank the EPA for undertaking the
demonstration project.   As a developer of field test kit
technologies, Dexsil believes that programs like the SITE
Program are a  valuable  contribution to environmental
efforts everywhere and that the EPA, as a government
agency, is the only organization with the credentials to
conduct these types of unbiased evaluations. Dexsil also
believes that the subject of TPH measurement of soil is
especially important because of the unique complexities
associated with both definition of TPH as an analyte and
comparison of  dissimilar laboratory or  field analytical
methods. Dexsil would also like to thank Terra Tech EM
Inc. for doing such a great job in organizing and running
the field demonstration and coordinating all the logistical
aspects of the project.

As a result of the predemonstration phase of the study and
subsequent discussions with developers, contractors, and
the EPA, there were a number of issues associated with
measurement of TPH in soil  that had to  be resolved.
Because TPH  itself is defined by the method used to
measure  it,  the choice  of a reference  method  and
verification of the  method's performance were the two
major concerns of the developers. Although the study was
not intended to be  an evaluation  and verification  of
laboratory methods, Dexsil would like to thank the EPA
for taking  the  first step in developing and  validating
laboratory analytical methods for TPH in soil. The results
obtained in this study should provide a good starting point
for further work on method validation. The discussions hi
the previous sections of this ITVR illustrate the difficulties
associated  with independent validation of a reference
method.

The  following  discussion has been contributed in the
interest of presenting an evaluation of some of the aspects
of the demonstration that are of particular importance to
PerroFLAG™ System users. The preliminary statistical
analyses presented were performed as  references for the
discussion  and  are not intended to  replace the analysis
presented in the previous sections of this ITVR. Dexsil
made every effort to follow standard statistical methods.
There might be small differences in the specific results
obtained by  the EPA  and Dexsil,   but the overall
conclusions drawn below should still be valid.  It is hoped
that this information will help field technicians develop a
more  complete  understanding  of  the  use  of  the
PerroFLAG™ System as well as other  field technologies
in  light  of  the  difficulties  associated  with  TPH
measurement of soil.

The PetroFLAG™ System

The  PetroFLAG™  System used  during  the  SITE
demonstration was the standard production analyzer sold
by Dexsil. This analyzer has been in use in the  field since
1995. All the reagents used were standard reagents, and
 This appendix was written solely by Dexsil. The statements presented in this appendix represent the developer's point of view and summarize
 the claims made by the developer regarding the PetroFLAG™ System.  Publication of this material does not represent the EPA's approval or
 endorsement of the statements made in this appendix; performance assessment and economic analysis results for the PetroFLAG™ System are
 discussed in the body of this ITVR.
                                                    105

-------
the  analytical procedures  used  were  the  standard
procedures described in the device's user manual and in
SW-846 Method 9074.

Designed in the early 1990s to overcome environmental
and technical problems associated with the  then current
methods  for  measurement of TPH  in  soil,   the
PetroFLAG™  System uses  no  chlorofluorocarbons, is
generally not  affected by soil moisture content below
15 percent, and is fully field-portable. The key to the
versatility of the  PetroFLAG™ System is the patented
solvent/developer solution system.  The water-miscible
solvent is designed to extract all nonpolar hydrocarbons
across a broad range of molecular weights  and classes.
The extracted  hydrocarbons  are then precipitated by the
developer solution to form a turbid aqueous suspension.

The PetroFLAG™ System provides a predictable and
reproducible response for a broad range of hydrocarbon
mixtures found   in  the  environment.   Because  the
PetroFLAG™  System responds to hydrocarbons based on
their nonpolar nature and not on a particular functional
group or fluorescence spectrum, the accuracy of a final
TPH result is a predicable function of the "mean molecular
weight" of the  contaminant hydrocarbon. Depending on
the makeup of the hydrocarbon, the  response of the
PetroFLAG™  System can be adjusted to  the specific
contaminant present using a response factor determined at
the factory or in the field. Because the response factors are
simply  multiplicative  constants  applied  to  the  raw
PetroFLAG™ System readings, the results obtained in the
field can be corrected as confirmatory laboratory results
become available.

The Effect of Naturally Occurring Organic
Compounds

With the PetroFLAG™ System, any naturally occurring,
nonpolar, hydrocarbon-like compounds may be quantified
as TPH. For this reason, naturally occurring hydrocarbons
tend to elevate  the device's TPH results. This can occur if
surface soils are included hi  a sample or if a project site
has had a history of industrial activity during which soil
layers may have been mixed during excavation activities.
If naturally occurring hydrocarbons are present at a site, an
estimate of the  background concentration can be obtained
by using both device and laboratory results for samples
collected either on site or off site.  Provided that  the
laboratory is reliable, either option is acceptable.  The
former option requires more samples  to  arrive at a
statistically valid background concentration estimate; the
latter option can be implemented using fewer samples but
requires  that  the  "blank"  samples  be  truly free of
hydrocarbon contamination.

For the NEX Service Station Area samples, both during the
predemonstration investigation and the demonstration, the
regression analyses indicate that background organics were
present at concentrations between 400 and 600 nag/kg.
Despite the presence of nonhydrocarbon organics and the
associated increased variability in the NEX Service Station
Area samples, the correlation of the PetroFLAG™ System
results with the laboratory results was still reasonable.
Because  naturally  occurring  organic  compounds  are
generally not uniformly distributed throughout soil strata
and because they are not included hi the definition of TPH,
if they are present, they tend to  add variability  to  the
sample TPH results. Although the regression analysis of
the data for the other sampling areas resulted in a positive
intercept, the intercepts were not statistically significant,
indicating that significant background concentrations were
not present. It should be noted that ERA made the PE
samples  from processed  garden  soil,  which   has
approximately 100 to 200 mg/kg of organic background
material as measured by the PetroFLAG™ System. Both
the predemonstration investigation sample results and the
demonstration soil PE  sample  results showed this
background concentration.

The Importance of the Choice of the Response
Factor

As is the case with all the field devices evaluated in the
study, the PetroFLAG™ System does not measure all the
individual compounds that are quantified as TPH by a
laboratory method using GC analysis.  The PetroFLAG™
System measures a specific property of hydrocarbons and,
using a conversion algorithm, converts this measurement
into  an equivalent TPH  value.   In the case  of  the
PetroFLAG™ System, the conversion algorithm is simply
an empirically determined  multiplier applied  to  the
turbidity  reading  for the  final solution.  Because  the
 This appendix was written solely by Dexsil. The statements presented in this appendix represent the developer's point of view and summarize
 the claims made by the developer regarding the PetroFLAG™ System.  Publication of this material does not represent the EPA's approval or
 endorsement of the statements made in this appendix; performance assessment and economic analysis results for the PetroFLAG™ System are
 discussed in the body of this ITVR.
                                                    106

-------
conversion multiplier, or  response  factor,  is  roughly
proportional  to the mean molecular weight or carbon
number of the PHCs, simply knowing which petroleum
product is present allows predetermined response factors
to be used for all soil samples from a site. Determining the
correct   response   factor   for  unknown  or  mixed
contaminants requires some preliminary analysis, but once
the contamination has been characterized, the appropriate
response factor can be selected, and any results already
obtained can be corrected using this response factor.

Accurate laboratory characterization of the hydrocarbon
present is crucial to the conversion of field readings into
meaningful TPH results. Not only is it important to choose
the correct analytical method, but the laboratory must also
be proficient in using the chosen method. Because the
PetroFLAG™ System is designed to measure TPH, it is
important that a TPH method be specified as the reference
method.  Using a specific petroleum method, such as a
GRO,  DRO,  EPH,  or VPH method,  may result  in
underestimation of the TPH concentration.   Similarly,
using a laboratory that has not demonstrated proficiency in
using TPH methods will lead to inaccurate results. Dexsil
recommends that any prospective laboratory be thoroughly
evaluated before it is used and that its performance be
monitored.   This evaluation should  involve laboratory
analysis of soils spiked with the appropriate hydrocarbons,
duplicate field samples, and samples that are both wet and
dry.

The sample  set used in the study provided  a good
illustration of the different factors that must be considered
in  choosing  the  correct  response  factor.    Dexsil
recommends  (and in some cases, it is  required) that
confirmatory split samples be sent to a qualified laboratory
as part of a  remediation project.  The response factor
chosen must always be a balance between  the correct
response factor  for  the  contaminant present  and  the
response factor that will match the laboratory results.

The choice of the correct response factors for the study
was somewhat complicated by  the  predemonstration
investigation results. On one hand, the reference method
indicated  that   the   PetroFLAG™   System  was
overestimating TPH concentrations. On the  other hand,
Dexsil's  laboratory  analysis of the  predemonstration
investigation samples indicated that the device obtained
correct results using the predetermined response factors for
the hydrocarbon mixtures present. To resolve this conflict,
Dexsil looked at the results of the PE sample analyses.
The EDRO PE sample results indicated that the reference
method results were  biased low: the recoveries were
54 and 64 percent for the two samples analyzed. For these
two samples, the Dexsil laboratory recoveries were 97 and
112 percent, respectively; and the PetroFLAG™ System
recoveries  (without a  dilution effect correction) were
97 and 103 percent, respectively.  The ratio of the average
recoveries predicted the slope of the regression line for the
comparison of the PetroFLAG™ System and reference
laboratory  results.   This  information, combined with
5 years of historical data, indicated that Dexsil's choice of
response factors based on the  predetermined values for
each of the sampling areas would provide accurate results.

The drawback in choosing conservative response factors as
described above was that the PetroFLAG™ System results
might be high relative to the  reference  method  results
if the reference method results were biased low.  The
choice of response factors was conservative because, from
a remediation point of view, a false positive is preferable
to a false negative.   Choosing a response factor that
underreported the results to match the reference method
recoveries would have resulted in underestimation of the
true  TPH concentration, and  if the reference method
recoveries improved for some soil types, the PetroFLAG™
System results would then be biased low relative to the
reference method results. As discussed in Chapter 7, this
is indeed what happened. An analysis of the data for both
the FFA and SFT Area samples (primarily siJty clay)
indicated  that the PetroFLAG™ System results were
biased high relative to the reference method results,  hi
comparison,  the device results from the NEX Service
Station  Area samples  (primarily  sand)  showed  no
significant bias relative to the  reference  method results.
Insufficient data were available to perform a regression
analysis of the B-38 Area sample data. Most of the data
for the B-38 Area samples were  at or below the method
quantitation limit for the PetroFLAG™ System.

The results for the PE samples confirmed the low bias  of
the reference method results.  The reference method's
diesel sample recoveries were 55,70, and 67 percent, and
 This appendix was written solely by Dexsil. The statements presented in this appendix represent the developer's point of view and summarize
 the claims made by the developer regarding the PetroFLAG™ System. Publication of this material does not represent the EPA's approval or
 endorsement of the statements made in this appendix; performance assessment and economic analysis results for the PetroFLAG™ System are
 discussed in the body of this ITVR.
                                                    107

-------
its gasoline sample recoveries were 61  and 62 percent.
The PetroFLAG™ System recoveries were 114, 75, and
75 percent for the diesel samples and 115 and 98 percent
for the gasoline samples, respectively.  Recoveries for the
hydrocarbon-like compounds such as Stoddard solvent and
turpentine were also lower for the reference method than
for the PetroFLAG™ System (see Chapters 6 and 7).
Although analysis of PE samples does not provide a direct
indication of the performance of a  particular method for
field  samples,  PE  sample analysis  provides the only
independent test of the "accuracy" of either a laboratory or
field method. The recoveries obtained for PE samples
could be considerably worse or better than those obtained
for field samples. For the demonstration, the PE samples
were made from relatively uniform, easily extractable soil.
Field samples may not be so easy to extract.

In a  situation  where the  reference  method used by
regulators is biased high or low, the PetroFLAG™ System
response  factor  can be  adjusted  to provide data  that
matches the reference method data. For example, using a
response factor of 10 instead of 6 for the FFA samples
brings the PetroFLAG™ System data into agreement with
the reference method data. For the SFT Area samples,
using a response factor of 4 instead of 2 brings the device
results into better agreement  with the reference method
results. Changing the response factor does not, however,
change the correlation, which was very good for these two
sample sets. The device data for the NEX Service Station
Area  samples (primarily sandy  soil)  agreed with the
reference method data and would not be recalculated.

The above discussions are in no way exhaustive, but they
should provide some guidance for interpretation of the data
generated during the study.   Overall, the PetroFLAG™
System performed as expected, exhibiting good correlation
with the reference method, reasonable recoveries for the
PE samples,  and predictable results for the liquid PE
samples.
This appendix was written solely by Dexsil.  The statements presented in this appendix represent the developer's point of view and summarize
the claims made by the developer regarding the PetroFLAG™ System.  Publication of this material does not represent the EPA's approval or
endorsement of the statements made in this appendix; performance assessment and economic analysis results for the PetroFLAG™ System are
discussed in the body of this ITVR.
                                                     108

-------