vvEPA
            United States
            Environmental Protection
            Agency
           Environmental Monitoring
           Systems Laboratory
           P.O. Box 93478
           Las Vegas NV 89193-3478
EPA/600/R-93/109
June 1993
            Research and Development
Automated On-Site
Measurement of
Volatile Organic
Compounds in Water

A Demonstration of
the A+RT, Inc.
Volatile Organic
Analysis System
                                        5280GR93QAD.COV

-------
                                         NOTICE


The information in this document has been funded wholly or in part by the U.S. Environmental Protection
Agency  under Contract No. 68-CO-0049 to Lockheed Engineering &  Sciences Company and under
Cooperative Agreement No. CR-817552-01-0 to the University of Houston.  It has been subjected to the
Agency's peer and administrative review, and it has been approved for publication as an EPA document.
Mention of corporation names, trade names, or commercial products does not constitute endorsement or
recommendation for use.

-------
                                          Abstract
A demonstration of the performance of the Analytic and Remedial Technology, Inc. Volatile Organic Analy-
sis System (AVOAS) is described.  The AVOAS is designed to provide automated collection and analysis
of water samples containing volatile organic compounds (VOCs).  In this demonstration, the AVOAS was
used to monitor VOCs  at a pilot pump-and-treat ground-water remediation facility. The demonstration was
conducted by the U.S.  Environmental Protection Agency (EPA) through the Superfund Innovative Technol-
ogy Evaluation (SITE) Program at a Superfund site in Woburn, Massachusetts in May, 1991.
The AVOAS consists of a manifold which permits  sampling from multiple locations; an  injector  which
extracts the VOCs from aqueous samples by a proprietary process similar to purge and trap and injects
them into a gas chromatograph; and software that provides system control and storage of data. The pro-
posed advantages of the AVOAS are that it eliminates the steps typically associated with collection and
analysis of water samples and provides real-time results.
The demonstration was performed  at a site where the ground water is contaminated with volatile chlori-
nated hydrocarbons. The performance of the AVOAS was compared with that of an established method
(EPA Method 502.2).  In addition to on-line monitoring of the ground-water treatment system, samples
spiked with measured  concentrations of analytes were processed for purposes of evaluation and quality
control.  The demonstration was designed to detect sources of variability between the field and conven-
tional  laboratory techniques.  On the  basis of results from this short-term study, we conclude that the
AVOAS is capable of providing the benefits of automated sampling and analysis, as proposed. AVOAS
recoveries for samples spiked with known concentrations of VOCs  were  higher than those  of Method
502.2, and were close to 100% for most analytes. Precision was within limits acceptable for VOC analyses
(within 30 percent relative standard deviation). Further development of the AVOAS is encouraged, and a
longer term evaluation  at other sites is recommended.
This evaluation report is submitted by Lockheed Engineering & Sciences Company in partial fulfillment of
Contract Number  68-CO-0049 with the U.S.  EPA Environmental Monitoring Systems Laboratory, Las
Vegas, Nevada.

-------
                                   Table of Contents
Notice	 ii
Abstract	iii
Abbreviations and Acronyms	xi
Acknowledgments	xiii
Executive Summary	xv

1 Introduction 	1
     SITE Program Background	1
     AVOAS Technology Description 	1
     AVOAS Demonstration Overview  	2 .
         Purpose of the AVOAS Demonstration	2
         Demonstration Site Description 	3
     Project Organization 	3
2 AVOAS Demonstration Design and Activities	9
     Demonstration Design 	9
         Rationale Behind the AVOAS Demonstration Design  	9
         AVOAS Demonstration Evaluations	10
     Demonstration Components	11
         Sample Types Analyzed	11
         Performance Evaluation Study	13
     Quality Assurance for the AVOAS Demonstration	15
         Data Quality Objectives	•	15
         Quality Control Samples	17
     Demonstration Methods	17
     EPA Method 502.2 Requirements and Modifications	17
         Sample Analysis	19
         Data Management	19
         Data Analysis	19
     AVOAS Demonstration Activities	21

3 AVOAS Demonstration Results and Discussion	23
     Operational Results	23
     Quantitative Results for Comparison Studies	23
         Comparison of the AVOAS and Method 502.2 Results for Treatment Train Samples	23
         Comparison of the AVOAS and Method 502.2 Results for Spiked Purged Effluent
          Water Samples 	29
         Transport Study Results	34
         Linearity Study Results	36

-------
     Quality Assurance and Quality Control Results	38
         Results for QC Samples	38
         Results for Data Quality Objectives	42
         Performance and Systems Audits	42
     Qualitative Observations	43
         Ruggedness and Adaptability	43
         Maintenance Requirements	44
         Facility and Supplies Requirements	44
         System Costs	45
     Summary of Results and Discussion	45

4 Conclusions and Recommendations	47
     Conclusions	47
         Advantages of the AVOAS	47
         Limitations of the AVOAS	...48
         Limitations of the AVOAS Demonstration	48
     Recommendations	.....48
         Recommendations Specific to the AVOAS	48
         Recommendations for Future Studies	49

References	50
                                            vi

-------
                                     List of Figures
                                                                                     Page
1-1.  Schematic of the AVOAS	2
1-2.  Area map of the Wells G and H Superfund Site, Wobum, Massachusetts	4
1-3.  Schematic of the Unifirst Ground-Water Treatment System	5
1-4.  Organizational Structure of the AVOAS Demonstration	6
2-1.  AVOAS Demonstration Evaluations 	11
3-1.  Treatment Train Sample Results. (Overall ports and times.)	28
3-2.  Spiked Purged  Effluent Water Results	29
3-3a. Transport Study Results for Medium- and High-concentration Spiked Samples for
        Trichloroethene	34
3-3b. Transport Study Results for Medium- and High-concentration Spiked Samples for
       Tetrachloroethene	35
3-4.  Quality Control  Check Standard Results	40
                                             VII

-------
                                     List of Tables
                                                                                   Page
2-1       Sample Types Analyzed for the AVOAS Demonstration	12
2-2       Summary of Analyte Concentrations (ng/L), by Sample Type	13
2-3       Summary of Method Detection Limits (ng/L) for the AVOAS and EMSL-LV
          Instruments, by Sample Loop Size, Dilution, and Sample Type	14
2-4       Data Quality Objectives for the AVOAS Demonstration	16
2-5       Method 502.2 Quality Control Requirements	18
2-6       Instrument Components and Operating Conditions	20
2-7       Chronology of AVOAS Demonstration Activities	21
2-8       Schedule of Sample Analyses for the AVOAS and EMSL-LV	22
3-1 (a,b)  Treatment Train Sample Results for EMSL-LV and the AVOAS On-line Analyses  	24
3-1 (c)    Treatment Train Sample Results for the AVOAS Off-Line Analyses	24
3-2(a)    Summary of Mean, Standard Deviation, and %RSD for Port Measurements for
          EMSL-LV, AVOAS On-LJne, and AVOAS Off-Line Analyses of Treatment Train Samples
          Across the Six Days of the Demonstration	25
3-2(b)    Summary of Mean, Standard Deviation, and %RSD for Sample Time Measurements for
          EMSL-LV, AVOAS On-Line, and AVOAS Off-Line (a.m. and p.m.) Analyses of Treatment
          Train Samples Across the Six Days of the Demonstration	26
3-3        Difference in Mean Percent Recoveries for Treatment Train Samples Analyzed by the
          AVOAS and EMSL-LV	27
3-4       Summary of Analysis of Variance Results for Treatment Train Samples	27
3-5(a)    Low-Concentration Spiked Purged Effluent Water Sample Results for EMSL-LV and
          AVOAS Analyses	30
3-5(b)    Medium-Concentration Spiked Purged Effluent Water Sample Results for EMSL-LV and
          AVOAS Analyses	31
3-5(c)    High-Concentration Spiked Purged Effluent Water Sample Results for EMSL-LV and
          AVOAS Analyses	32
3-5(d)    Superhigh-Concentration Spiked Purged Effluent Water Sample Results for EMSL-LV
          and AVOAS Analyses	33
3-6       Difference in Mean Percent Recoveries for Spiked Purged Effluent Water Samples
           Analyzed by the AVOAS (Off-Line) and EMSL-LV	34
3-7(a)    Transport Study Results for Medium Spiked Samples	35
3-7(b)    Summary of Transport Study Results for Medium Spiked Samples	36
3-8(a)    Transport Study Results for High Spiked Samples	37
3-8(b)    Summary of Transport Study Results for High Spiked Samples	37
3-9(a,b)  Quality Control Check Standard Results for EMSL-LV and the AVOAS 	38
3-10(a,b) Percent Recoveries for Quality Control Check Standards for  EMSL-LV and the AVOAS	39
3-11 (a,b) Percent Differences for Continuing Calibration Check Standards for EMSL-LV and the
          AVOAS	41
3-12(a)   Percent Recoveries for Spiked Purged Effluent Water Samples for Each Demonstration
           Day for EMSL-LV	43
3-12(b)   Percent Recoveries for Spiked Purged Effluent Water Samples for Each Demonstration
          Day for the AVOAS	**

                                            ix

-------
                          List of Abbreviations and Acronyms
1,1 -DCA      1,1-dichloroethane
1,1 -DCE      1,1 -dichloroethene
1,1,1 -TCA     1,1,1 ,-trichloroethane
1,2-DCA      1,2-dichloroethane
A+RT         Analytic and Remedial Technology, Inc.
AFMMP       Advanced Field Monitoring Methods Program
ANOVA       Analysis of variance
AVOAS       A+RT Volatile Organic Analysis System
CCCS        continuing calibration check standard
cis-1,2-DCE   cis-1,2-dichloroethene
CLP          Contract Laboratory Program
DQO         data quality objective
ELCO         electrolytic conductivity detector
EMSL-LV      Environmental Monitoring Systems Laboratory-Las Vegas
EPA          Environmental Protection Agency
EPC          Environmental Project Control, Inc.
GC           gas chromatograph
HCI           hydrochloric acid
LESC         Lockheed Engineering & Sciences Company
MDL          method detection limit
Method 502.2  EPA Method 502.2, "Volatile Organic Compounds in Water by Purge and Trap Capillary
               Column Gas Chromatography with Photoionization and Electrolytic Conductivity
               Detectors in Series"
%D           percent difference
%R           percent recovery
PCE          tetrachloroethene (perchloroethene)
PEW         purged effluent water
PE           performance evaluation
PID           photoionization detector
ppb           parts per billion
ppm          parts per million
PQL          practical quantitation limit
QA           quality assurance
QAPjP        Quality Assurance Project Plan
QC           quality control
QCCS        quality control check standard
RSD          relative standard deviation
RT           retention time
SARA         Superfund Amendments and Reauthorization Act
                                            xi

-------
                   List of Abbreviations and Acronyms (continued)
SITE          Superfund Innovative Technology Evaluation
SPEW        spiked purged effluent water
TCE          trichloroethene
trans-1,2-DCE  trans-1,2-dichloroethene
TT            treatment train
UV            ultraviolet
VC            vinyl chloride
VOA          volatile organic analysis
VOC          volatile organic compound
                                           xii

-------
                                    Acknowledgments
We wish to acknowledge the support of all of those who helped plan and conduct this demonstration,
interpret the data, and prepare this evaluation  report.  In particular, for demonstration site access and
relevant background information, Barbara Newman, Wells  Q and H Remedial Project Manager (EPA
Region I); for on-site coordination, Jeffrey Lawson (Environmental Project Control, Inc.); for QA oversight
and support for the performance evaluation study, Elizabeth Schultz (Trillium, Inc.); for the preliminary
evaluation and QA plan, James Smith  (Trillium, Inc.); for preparation of the Demonstration Plan, Deb
Chaloud (Lockheed); for field sampling and logistics, Kevin Cabbie (Lockheed); for project management,
Scott Hulse (Lockheed) and Deirdre O'Leary (Lockheed); for EPA project management, Eric Koglin and
Stephen Billets (EPA); for project management of the EPA/University of Houston Cooperative Agreement,
John Nocerino (EPA); for data review, Grade Martucci (Lockheed); for data reduction, John Zimmerman
and Bryant Hess (Lockheed); and for word processing support, Julia Jackson, Lillian Steele, Pat Craig,
and Jan Aoyama (Lockheed).  In addition, we gratefully acknowledge the comments  and suggestions of
those who reviewed this report Reviewers included Alan Crockett (EG&G Idaho, Inc.), Christian Daughton
(EPA), Mitchell Erickson (Argonne National Laboratory), Jeff RosenfekJ and Mark Silverstein (Lockheed),
and Kaveh Zarrabi (University of Nevada, Las Vegas).
                                             xiii

-------
                                   Executive Summary
An automated system designed to monitor volatile organic compounds (VOCs) in a water-treatment pro-
cess stream was evaluated under the U.S. Environmental Protection Agency (EPA) Superfund Innovative
Technology Evaluation (SITE) Program. The monitoring system evaluated was the Analytic and Remedial
Technology,  Inc., Volatile Organic Analysis System (AVOAS). The AVOAS permits unattended sampling
from multiple locations, with extraction of VOCs from the samples by a proprietary process similar to purge
and trap, and analysis by gas chromatography (GC).
The demonstration was conducted under the guidance of the EPA Environmental Monitoring Systems Lab-
oratory in Las Vegas, Nevada (EMSL-LV), as part of the Monitoring and Measurement Technologies por-
tion of the SITE Program. A mandate of the SITE Program is to identify cost-effective alternatives to
existing  methods used in characterizing and remediating contamination at Superfund sites. SITE demon-
strations provide a mechanism for evaluating selected technologies that are at least comparable in quality
and cost to established methods.
Currently, most methods for processing samples used to monitor remediation at Superfund sites require a
number  of steps. These include sample collection, preservation, transport, storage, and analysis at off-site
laboratories.  The principal drawbacks to these procedures are the  need for trained sampling personnel,
the turnaround time for results, the cost per analysis, and, particularly for VOCs, the potential for changes
in sample integrity between the time of sample collection and analysis. The  proposed advantages of the
AVOAS  are that it replaces conventional sampling and analysis procedures and provides results within
less than an hour of sampling (i.e., nearly real-time results). Further,  the AVOAS continuously monitors the
function of the various components of a treatment system, permitting process control capability. Malfunc-
tions in a treatment system can be quickly detected as increased concentrations of VOCs, signalling the
need for maintenance or other measures to correct the problem.
For the application evaluated in this demonstration, the AVOAS was incorporated as a dedicated compo-
nent of a treatment system being tested at an EPA Region I Superfund site for remediating ground water
contaminated with chlorinated VOCs.  The treatment system was a series of discrete units designed to
progressively reduce contaminant concentrations. The AVOAS measured VOC concentrations after each
treatment unit.
The demonstration compared the performance of the AVOAS with that of EPA Method 502.2.  The field
portion of the demonstration was conducted May 18 through 24, 1991, at the Wells G and H Superfund
Site in Wobum, MA.  For the demonstration, the ground-water treatment system was operational, and
samples were collected and analyzed in the monitoring mode for which the AVOAS was designed. Sam-
ples were also collected  using standard methods for VOCs in water, and were analyzed by the AVOAS
operated in the "off-line" mode and by Method 502.2 at the EMSL laboratory in Las Vegas, NV.  In addition
to samples from the treatment system, samples spiked with  measured concentrations of analytes and a
variety of quality assurance and quality control (QA/QC) samples were analyzed by each method.
A total of 96 samples analyzed by the AVOAS and the EMSL-LV instrument during the demonstration were
used to assess the performance of the AVOAS. The AVOAS  ran an  average of 22.3 hours per day, which
included approximately 8 hours per day of unattended operation. Only minor breakdowns occurred, and
recalibration  and maintenance requirements were minimal. The qualitative and quantitative findings gen-
erated in this demonstration provided the basis for the following observations about the performance of the
AVOAS:
                                             xv

-------
 1) Concentration values generated by the AVOAS for virtually all samples were higher than those mea-
   sured by Method 502.2; however, AVOAS analyte recovery values were closer to 100%.  The cause of
   this finding cannot be resolved from  the data available from  this short-term  study.  The precision
   obtained by the AVOAS was comparable to that obtained by Method 502.2, and was within the expected
   range.
2) The AVOAS performed well in analyzing  samples with VOC concentrations in the low-ppb to low-ppm
   range.
3) The proprietary system used by the AVOAS for stripping VOCs from water is simple in design and func-
   tion and more adaptable to field automation than the traditional purge-and-trap procedure.
4) The demonstration showed that the AVOAS is capable of providing the benefits of automated sampling
   and analysis, as proposed. That is, it provides real-time  results and eliminates manual sample collec-
   tion, handling, shipping, and analysis steps as well as the potential for losses in sample integrity.  In
   addition, the potential cost savings for use of the AVOAS in comparison to use of established methods
   could be substantial, particularly for processing large numbers of samples (i.e., for long-term, continu-
   ous monitoring).
The following items were identified as potential limitations to the use of the AVOAS. Many of these are not
so much limitations, as factors that  must be kept in mind or planned for  by prospective users of the
AVOAS.  The AVOAS developer recognized some of these  limitations during the demonstration, and has
since taken steps to correct equipment or procedures.
1) The AVOAS requires a temperature-controlled environment and an atmosphere free of contamination.
   Although maintaining these conditions was not a problem at the location where this demonstration was
   conducted, extra measures to ensure these conditions may be required in other field situations or indus-
   trial facilities.
2) Water samples high in paniculate matter tend to clog a filter in the injector inlet. The impact of this prob-
   lem can be  minimized if water is pre-filtered before being conveyed to the AVOAS. Also, cleaning and
   replacement of the filter can be incorporated as part of the routine maintenance procedures.
3) The possibility of carryover in the smallest sample loop of the AVOAS was suggested by the finding of
   an increase in trace concentrations of several analytes  in  ground-water treatment system (i.e., treat-
   ment train) samples between morning and afternoon analyses. The AVOAS developer has modified the
   design of the manifold to reduce dead volume significantly, and expects this change to greatly reduce
   the carryover problem.
4) Many of the (relatively minor) problems  encountered with  the AVOAS during the field demonstration
   were due to problems in the system software, most of which were corrected during the demonstration.
Based on the findings of this study, the following recommendations are made:
1) Further development of the AVOAS is encouraged, and additional studies should be performed.  Perfor-
   mance data should be generated for this system when used at other sites, by different operators, and in
   the fully automated mode.
2) Instrument calibration should be checked regularly on a schedule based on the performance of the
   detector. A daily calibration check is recommended. If this is not feasible, at a minimum, detector per-
   formance should be monitored daily over a period of several weeks after instrument installation. An
   appropriate calibration schedule should be developed and followed for the long-term  use of the system.
   especially for unattended, remote operation.
3) A mechanism for introducing surrogate standards into the AVOAS with each analysis would provide an
   ongoing indication of analytical instrument performance.
4) Depending on the data quality objectives  for instrument performance at a particular site, blank samples
   should be periodically analyzed on all sampling loops, to ensure that carryover from high-concentration
   samples into low-concentration samples is not occurring.
5) During the initial stages of long-term monitoring, concentrations for each sampling port, as well as data
   on instrument precision should be closely monitored to generate expected baseline conditions. Periodic
   confirmatory analyses by established methods should be  perfoomed to verify instrument results.


                                             xvi

-------
                                        Section 1
                                       Introduction
     The Analytic and Remedial Technology, Inc.,
Volatile Organic Analysis System (AVOAS)  was
evaluated under the U.S. Environmental Protection
Agency (EPA) Superfund Innovative Technology
Evaluation (SITE) Program.   The AVOAS is
designed to monitor volatile organic compounds
(VOCs) in water-treatment process streams.  The
system provides automated sampling from multi-
ple locations, with extraction of VOCs from the
samples by a proprietary process similar to purge
and  trap,  and analysis by gas chromatography
(GQ. The purpose of the demonstration described
in this report was to compare the performance of
the AVOAS with that of an established method.
     This  section provides an overview of the
SITE Program and descriptions of the AVOAS, the
site of the demonstration, and the project organiza-
tion. The  rest of the report is organized into  four
sections.  Section 2 describes the AVOAS demon-
stration; Section 3 presents and discusses the quan-
titative and qualitative results of the evaluation;
and Section 4 presents the conclusions and recom-
mendations based on the findings of the demon-
stration.

SITE Program Background

     The Superfund Program was initiated in 1980
to identify, prioritize, and remediate uncontrolled
hazardous  waste sites. Subsequently, the problems
associated with hazardous waste sites proved to be
far more complex man originally anticipated. In
1986, the  U.S.  Congress  enacted the Superfund
Amendments and Reauthorization  Act (SARA),
requiring the EPA to promote more effective haz-
ardous waste site identification and cleanup.  The
SITE Program addresses Section 311(b) of SARA,
which requires the EPA to establish "a program of
research,  evaluation,  testing,  development,  and
demonstration of alternative or innovative treat-
ment technologies...which  may be utilized in
response actions to achieve  more permanent  pro-
tection of human health and welfare and the envi-
ronment"
     Two categories of technologies are addressed
under the SITE Program: (1) treatment technolo-
gies that may serve as alternatives to land disposal,
and (2) technologies for measuring and monitoring
contaminants at hazardous waste sites. This latter.
portion of the SITE Program is administered by the
Advanced  Field Monitoring Methods  Program
(AFMMP).  The Environmental Monitoring  Sys-
tems Laboratory at Las Vegas, Nevada (EMSL-LV)
is the lead laboratory for the AFMMP, and pro-
vided oversight of the demonstration described in
this report

AVOAS Technology  Description

     The AVOAS was designed by Gary Hopkins
and Doug Mackay of Analytic and Remedial Tech-
nology, Inc. (ATO, Inc., Menlo Park, CA).  The
AVOAS consists of a  sampling  manifold and a
purge-and-trap  concentrator  ("injector")   con-
nected to a gas chromatograph (GQ. An integrator
processes the  GC signal data, and a computer is
used to control the  system and to store the data.
The innovative components are the sampling mani-
fold, the injector, and the computer software.  The
manifold permits on-line collection of samples
from  multiple locations. The injector collects a
measured volume of water, strips the VOCs from
the aqueous to the gas phase, traps them on a sor-
bent, and thermally desorbs them into the GC. The
computer software handles data storage and  con-
trols  the analytical system. With use of a modem,
data  can be transmitted off-site, allowing remote
operation and control of the  AVOAS, eliminating
the need for the physical presence of the operator
at the site.   Figure 1-1 is  a schematic  of the
AVOAS.
     Among the features that can be controlled are
the sampling sequence (i.e., the order in which the
manifold valves are actuated);  the flush time for

-------
                          VolaHUudSampIt
      Injector
   Liquid Proem
    MttUfoU
     Simpfc
    MMifoM
                                                              Integrator
                                                                                 Computer
                             Wast*
            Samplfi
           A-SuapbSofaMidi
           B • OfMte Sunpta CoOtcttoa PBrt
           C-SifflpltUiMPamp
          UqAtPracaMKtfbld:
InjKtor:
 G.SWpptejCdi
 H-ProcaiVili«
 J-IaJcctiaiValTC
 I-TrapudlicaMr
           E-MuHMoopVih*
           F'UqvMSuvhPnmi)
Hgura 1-1. Schematic of the AVOA3.
the sampling lines; the sample size; and the inte-
grator operating parameters.  Sample size is con-
trolled via  a  switching  valve,  which  permits
selection among sample loops of different internal
volumes. To optimize detection of target analytes,
the sample volume can be selected based on either
prior knowledge of potential concentration ranges
at a given sampling point or by computer logic. In
the computer logic  mode, the  microcomputer is
programmed to repeat a sample analysis using a
different loop volume if the first measurement is
out of the range of instrument detection or calibra-
tion. Sample sizes fromlOO jiL (the dead volume
of the valve) to 10 mL or more are possible (the
upper limiting volume is defined by the maximum
sample processing time required for any applica-
tion). The sampling loops are flushed with VOC-
free water between samples.  To minimize sample
carryover (i.e.,  adsorption of compounds in sam-
ples to analytical surfaces they contact, followed
by desorption into subsequent samples), the mani-
fold and all  sample loops and sample conveyance
lines are constructed of stainless  steel.  External
samples, including calibration standards and qual-
ity control (QQ samples, are introduced into the
system through a line between the manifold and
injector.  Water samples from each sample port can
be collected at the manifold for off-site, confirma-
tory laboratory analysis. The AVOAS can be used
with virtually any GC and integrator, and can be
                adapted to different purgeable analytes by selection
                of the appropriate GC column, temperature pro-
                gram, and detector. Earlier versions of the system
                have been used in site-specific research applica-
                tions. The version evaluated in this demonstration
                had modifications intended to make the AVOAS a
                commercially marketable product
                     For the application discussed in this report,
                the AVOAS functioned as a dedicated component
                of a treatment system  being demonstrated at a
                Superfund site to remediate ground water contami-
                nated with chlorinated VOCs. The treatment sys-
                tem consisted of a series of discrete units designed
                to provide a stepwise reduction in  contaminant
                concentrations. The AVOAS measured VOC con-
                centrations at selected  locations in  the ground-
                water treatment system.

                AVOAS Demonstration Overview

                Purpose of the AVOAS Demonstration
                     The purpose of this demonstration was to
                compare the performance of the  AVOAS  while
                operating at the site of a remediation with that of
                an established method conducted in  a fixed-base
                laboratory.  EPA Method 502.2, "Volatile Organic
                Compounds in Water by Purge and Trap Capillary
                Column Gas Chromatography with  Photoioniza-
                tion and Electrolytic Conductivity Detectors in

-------
Series - Revision 2," was selected for comparison,
because operation of the AVOAS most closely fol-
lows the specifications of that method.  This SITE
demonstration was  not intended to evaluate the
ground-water treatment system.
     The primary goal of the demonstration was to
generate data of  sufficient  quality, quantity,  and
type,  by both the AVOAS  and the  comparison
method, to permit a sound interpretation of results.
A demonstration plan prepared prior to the AVOAS
evaluation identified the specific study objectives
and described in detail the sampling, analysis, and
QA/QC activities required  to meet those objec-
tives.
     Qualitative measures of system performance
(ruggedness, reliability, maintenance requirements,
facility and supply  requirements, and cost) were
not specifically defined in the AVOAS  Demonstra-
tion  Plan; however, these measures were infor-
mally evaluated and are discussed in this report.
Performance characteristics such as specificity and
interferences, which are determined by the detec-
tion system (i.e., the chromatographic column, gas
chromatograph, and detector), were not considered
in this demonstration because these characteristics
are not a function of the innovative components of
the AVOAS.

Demonstration Site Description
     The AVOAS demonstration was performed at
the Wells G and H Superfund Site in Woburn, Mas-
sachusetts. The site covers 330 acres and includes
the aquifer and land in the vicinity of two former
municipal drinking water wells (Wells G and H).
Figure 1-2 is an area map of the site.
     Wells G and H had been used as a source of
municipal water until,  in 1979, the Massachusetts
Department of Environmental Protection detected
contamination by chlorinated VOCs in water sam-
ples  from the wells.  The primary contaminants
include  trichloroethene (TCE) and tetrachloroet-
hene (PCE), with lesser amounts of 1,1-dichloroet-
hene  (1,1-DCE),  cis-l,2-dichloroethene (cis-1,2-
DCE), trans-l^-dichloroethene  (trans- 1,2-DCE),
1,1-dichloroethane (1,1-DCA), and 1,1,1-trichloro-
ethane (1,1,1-TCA). In December 1982, the EPA
put the site on the National Priority List, and in
September 1989, a Record of Decision was issued
requiring that the ground water beneath the site be
treated to remove the contamination.
     A pilot demonstration (conducted under a dif-
ferent program) of a ground-water treatment sys-
tem installed in the Unifirst Facility presented the
opportunity to simultaneously evaluate the perfor-
mance of the AVOAS.  The treatment system pilot
demonstration was May 18 through May 30,1991,
and the field portion of the AVOAS SITE demon-
stration was conducted over the 6-day period of
May 18 through 24. During the pilot demonstra-
tion, the treatment process included units for ultra-
violet/chemical oxidation, carbon adsorption, and
reductive dehalogenation.  Figure 1-3 is a sche-
matic of the ground-water treatment system. Sam-
pling ports  at  selected  locations   permitted
monitoring the concentration of VOCs before and
after each treatment step. The sampling ports used
to evaluate the AVOAS included port S-3 (after the
carbon drum, before the oxidation unit) and  port S-
4 (after the reductive dehalogenation unit,  before
the oxidation unit).  Sample collection from two
ports was specified in the Demonstration Plan, on
the  expectation that differences in concentration
would provide information about the performance
of the AVOAS in measuring more than one concen-
tration.  Samples from each port were  analyzed
separately; i.e., samples from S-3 and S-4 were not
physically combined.  Water from port S-6 (treated
effluent) was used to prepare spiked samples and
analytical and transport blanks.

Project Organization
     The  EMSL-LV and  its  prime  contractor,
Lockheed  Engineering  &  Sciences  Company
(LESQ, worked together to plan, conduct, evalu-
ate, and report the results of the AVOAS demon-
stration.    The  EMSL-LV   Technical   Leader
arranged scheduling and the approach of the dem-
onstration.  The Technical Leader was also respon-
sible for communicating with EPA Region I  staff to
obtain access to the  site, and for background and
health and safety information.  Oversight of all
remedial activities at the Wells G and H Superfund
Site  is the responsibility of  the EPA Region  I
Remedial Project Manager.  The  site remediation
contractor for the potentially responsible parties is
Environmental Project Control, Inc. (EPQ.  For
the AVOAS demonstration, ETC  was responsible
for overseeing or conducting all of the operational
aspects of the  ground-water pump-and-treat pro-
cess.  The Johnson Company designed the treat-
ment system. EPC arranged the subcontract with
A*RT, Inc. to provide the on-line analytical capa-
bility.

-------
                                                           MASSACHUSETTS
                                                                Ground-Water
                                                                Extraction Well
                          Location of the
                          Ground-Water
                        Treatment System
                          andtheAVOAS
                                             Unifirst
                                             Facility
                                                      Golden
                                                      School
                                                                        OAD1047GR92
Figure 1-2. Area map of the Wells G and H Superfund Site, Woburn, Massachusetts.

-------
           voc
       CONTAMINATED
       INFLUENT FROM
      GROUND-WATER
      EXTRACTION WELL
            VOC-FREE
            EFFLUENT
            (TO STORM
             SEWER)
     SAMPLE LINES FROM
      TREATMENT TRAIN
       SAMPLE PORTS  "
                                                                SAFETY/
                                                        DECONTAMINATION AREA
                             MULTI-MEDIA/
                             PRESSURE '
                               FILTER
                                       REDUCTIVE  /
                                    DEHALOGENATION
                                          UNIT
                                               A RT, INC.
                                            VOLATILE ORGANIC
                                            ANALYSIS SYSTEM
    0123
    I  I   I  I
      SCALE IN FEET
                               LEGEND
 SAMPLING
   PORT

GATEVALVE
Sample Pa
S-1
S-2
S-3
S-4
S-5
S-6
a

Influent from ground- water extraction well
After reductive denaJpaenatfon unit
After carbon drum, bei
Betoreoxidationunlt
After oxidation wit
Treated effluent
yecoddatian unit



Rgure 1-3. Schematic of the UnMrat Ground-Water Treatment System.

-------
     Figure 1-4 shows the organizational structure
for the AVOAS demonstration.  Key participants
included:
• Statistician (LESC):  Responsible for designing
  the evaluation and overseeing data analysis and
  interpretation of results, including verifying and
                             validating data, generating data tables and plots,
                             and programming and executing statistical rou-
                             tines.

                             Demonstration Design Consultant (EPA Cooper-
                             ative  Agreement,  University   of   Houston):
                             Assisted in the development of the AVOAS dem-
                  1
             EPA/Region I
      Wells G and H Superfund Site
        Remedial Project Manager
                Region I
             Site Contractor
                 (EPC)
    Treatment
   System Design
   (Johnson Co.)
Treatment System
   Monitoring
  (A+RT,Inc.)
           EPA/EMSL-LV
     Monitoring and Measurement
        Technologies Program
AVOAS Demonstration Project Oversight
             Ground-water
           Treatment System
               (Unifirst)
                  Demonstration
                  and QA Plans
                                        Field Logistics/
                                       Safety and Health
                                         EPA Method
                                        502.2 Analyses
                                                                                   Database
                                                                                 Management
                                                              Data
                                                          Interpretation
   EMSL-LV • Environmental Monitoring Systems Laboratory-Las Vegas
       LESC - Lockheed Engineering & Sciences Company
         QA - Quality Assurance
       SITE - Superfund Innovative Technology Evaluation
   A+ RT, Inc. • Analytic and Remedial Technology, Inc.
        EPC - Environmental Project Control, Inc.
Figure 1-4. Organizational Structure of the AVOAS Demonstration.

-------
onstration design.  Provided data interpretation
and report review.

Supervising Chemist (LESC):  Responsible for
ensuring the technical and scientific quality of
the project  Responsible for overseeing or per-
forming technical activities related to planning
and conducting the evaluation and interpreting
results, including overseeing field and reference
laboratory operations, preparing sample spikes,
and approving any modifications to the Demon-
stration Plan.

AVOAS Developer (A+RT, Inc.):  Responsible
for installing and operating the AVOAS at the
field site, including calibrating the instrument,
collecting treatment train  samples,  analyzing
samples, and providing copies of the data gener-
ated by the AVOAS.

Project QA  Manager (LESC):  Responsible for
developing the QAPjP and for overseeing the
review and interpretation of the QA data.
Logistics  and Sampling  Coordinator (LESC):
Responsible for all aspects of field operations,
  including site reconnaissance, sample collection
  and shipment, sample chain-of-custody, decon-
  tamination of equipment and disposal of hazard-
  ous wastes in the field,  and documentation of
  field activities.
• Laboratory Analyst  (LESC):  Responsible for
  analyzing  samples  in  accordance with  EPA
  Method S02.2 and with the Demonstration Plan
  and the QAPjP, and for documenting and report-
  ing data in the required format.

• Data Base Manager (LESC): Responsible for the
  acquisition and formatting of all data generated.

• Health and Safety Coordinator (LESQ:  Respon-
  sible for  ensuring that all field  and laboratory
  health and safety issues were addressed, and for
  preparation and approval of the Site Safety and
  Health Plan.
• Report Coordinator and Writer (LESQ:  Respon-
  sible for coordinating input from project partici-
  pants and preparing this report

-------
                                         Section 2
                      AVOAS Demonstration Design and Activities
Demonstration Design

Rationale Behind the AVOAS
Demonstration Design

     An important objective of the AVOAS dem-
onstration design was to account for factors that
could affect the comparability 'of results.  These
included: 1) sample-related factors, including sam-
ple matrix effects and changes in sample integrity
during collection, preservation, transport, and stor-
age and 2) variability due to analysis, including
differences in instrument hardware and function,
day-to-day variation in operating conditions, drift
in instrument calibration, and variability in the lab-
oratory technique of different analysts.

Sources of Variability in Samples
     Sample Matrix Effects —  Chromatographic
retention times and measured concentrations of tar-
get analytes can be .affected by  the chemical and
physical characteristics of the sample, i.e., the sam-
ple matrix.    Matrices in  this  demonstration
included 1)  reagent-grade  water for malting  the
standards for the initial calibration and for the daily
instrument calibration checks and 2) ground-water
treatment system (i.e., treatment train) water.  To
reduce the potential variability due to matrix, treat-
ment train effluent water from port S-6 was purged
to remove VOCs and was used  for travel  blanks,
AVOAS instrument  blanks,  and  for preparing
spiked samples.
     Changes  in Sample  Integrity —  Sample
integrity refers to how well the  sample composi-
tion  and concentration  at the  time of. analysis
reflect those characteristics at the time the  sample
was collected.   For demonstrations in which two
methods are being compared, maintaining  unifor-
mity in  the sample pairs analyzed by the test and
the comparison  method  is  critical  for comparing
data from both methods. Sample integrity can be
compromised by collection procedures,  sample
container characteristics, conditions during ship-
ping (e.g., temperature  or barometric pressure
changes),  component  degradation  or alteration
(e.g.,  due to biological or photochemical  pro-
cesses), and handling steps during analysis. Main-
taining the integrity of samples containing low-
molecular-weight  chlorinated  hydrocarbons,  the
target analytes for the AVOAS  demonstration, is a
particular challenge because those compounds can
easily volatilize into the atmosphere or sorb to con-
tainer materials.  In addition,  residues in sample
containers or in the atmosphere in which the sam-
ples are stored can contaminate samples, leading to
biased or false-positive results (e.g., increased con-
centrations or the presence of compounds not actu-
ally present in the medium sampled).
     The potential for changes in sample integrity
was a concern for samples  shipped to EMSL-LV,
because they required additional handling steps
and time between sampling  and analysis. To mini-
mize variability, the guidelines of  Method 502.2
for collecting, preserving, shipping, and handling
samples containing VOCs were followed. In addi-
tion, holding time — the time between sample col-
lection and analysis — was minimized as much as
possible.   A  transport study  was performed in
which results  for samples spiked with known ana-
lyte concentrations and shipped, were compared
with  results  for  corresponding  samples  not
shipped.

Sources of Analytical Variability

     Differences  in Instrument Hardware  and
Operation — The manner  in  which the AVOAS
collects and handles samples when operated in the
automated ("on-line" mode) did  not permit direct
comparison of sample results with those of Method
502.2.  In the on-line mode, the  AVOAS conveys
samples from the sampling ports to the analytical
system. Samples are introduced directly into the
injector without exposure to the atmosphere, and
are analyzed immediately.  In  contrast, the estab-

-------
lished methods, including  Method 502.2 involve
manual collection of  samples into  containers,
transport to a laboratory, and refrigerated storage
prior to analysis.  The sample is exposed to the
atmosphere during collection, and usually  again
just before analysis. To help identify any effects on
the data due to these inherent differences in proce-
dure, the AVOAS was used in both the on-line,
automated mode and in off-line mode for analysis
of treatment train samples.  For the off-line proce-
dure, the line between the manifold and injector
was momentarily opened and placed in the sample
vial.  The sample water was then pumped through
the sampling loop of the AVOAS injector.
     Another difference between the AVOAS and
typical Method 502.2 analyses is the  means by
which sample concentrations are  adjusted to fall
within the  linear range of  the  detector.   The
AVOAS injector  uses sampling loops of different
volumes that can be selected to adjust for analyte
concentrations.  For example, for high-concentra-
tion samples, a small sample volume is used so that
a small quantity of analyte is introduced into the
analytical system.  By contrast, the standard purge-
and-trap apparatus used with the EMSL-LV instru-
ment makes all injections using the same sample
volume.  Adjustments to keep the sample concen-
tration range within the linear range of the detector
are made by diluting the sample. For this demon-
stration, sample dilution was required to bring the
highest concentration analyte, PCE, to  within the
linear range of the EMSL-LV detector. Potential
sources of variability in the EMSL-LV data arise
from the dilution procedure, and for the AVOAS
data, from the performance of the sampling loops.
     Finally, the method by which the AVOAS
strips VOCs from samples (A*RT, Inc. proprietary)
is  different from standard purge-and-trap tech-
niques, so that differences  in VOC stripping effi-
ciency are  possible.   Although  this  difference
should be compensated for through analysis of
standards, stripping efficiency for the different
sample loop sizes may vary; this source of variabil-
ity was not evaluated in this demonstration.
     Day-to-Day Variations in Instrument Operat-
ing Conditions — Day-to-day variability in instru-
ment operating conditions could arise from several
sources.  The GC used in this demonstration with
the AVOAS is very basic in design. It was installed
on location at the industrial  facility  where the
ground-water  treatment system is housed.   The
demonstration  began  within  3  days  after the
AVOAS was installed.  During the demonstration.
in addition to processing SITE demonstration sam-
ples, the AVOAS was used in the on-line monitor-
ing mode and for off-line analyses of samples from
a second treatment system. The AVOAS was oper-
ating for over 22 hours per day during the demon-
stration. Samples of widely varying concentrations
were analyzed sequentially, providing the potential
for contaminant carryover from high-concentration
samples.   In contrast,  the EMSL-LV  instrument
had been allowed to stabilize in the laboratory for
nearly 2 weeks prior to the demonstration.  Only
AVOAS demonstration samples were analyzed by
that instrument during  this study, and  the instru-
ment was used  less than 8 hours per day.
     To help detect day-to-day sources of variabil-
ity, quality control  check standards, continuing cal-
ibration check standards,  and instrument blanks
were analyzed daily. To identify effects due to the
time of day, treatment train samples were collected
and analyzed by the AVOAS in both the morning
and the afternoon.
     Drift in Instrument Calibration — Changes
in instrument response factors or calibration curves
can be a significant source of measurement vari-
ability  in VOC analysis.  Both  the AVOAS and
EMSL-LV instruments were calibrated prior to the
demonstration, and their calibration was monitored
daily using the  CCCS.
     Other Sources of Variability — Data for
spiked  samples can have inherent variability.  For
example, analyst spiking technique can vary, and
actual concentrations of spiking solutions can be
different from  intended concentrations. To mini-
mize these sources of  variability, the stock solu-
tions for spiked samples were prepared in  batches
and sealed in ampoules by one chemist before the
study.  Spiked  samples  and  sample aliquots were
prepared by the same chemist at die field site.

AVOAS Demonstration Evaluations

     The assessment of the AVOAS performance
relative to that of Method S02.2 was based on three
evaluations, which are depicted schematically in
Figure 2-1 and  are  described below.
1. The true concentrations of VOCs in the ground-
  water  treatment  train water were  unknown.
  Therefore, to assess the comparability of results
  obtained by  the AVOAS and by Method 502.2,
  samples spiked with known concentrations of
  analytes (i.e., spiked purged effluent  water sam-
  ples) were prepared at the field site and were
  analyzed by  both instruments.  Data were gener-
                                             10

-------
    1)  SPIKED PURGED
       EFFLUENT WATER SAMPLES

       (Prepared at the field site in
       WobunuMA)
                                                    -•»•   AVOrtS Off-line analysis
    2)  TREATMENT TRAIN SAMPLES

       (From ground-water treatment system
       at the field site in Wobura. MA)
   3)   TRANSPORT STUDY SAMPLES

       (Piepated and analyzed in Las Vegas, NV)
      L- low-concentration spiked sample
      M. medium
-------
 TABLE 2-1. SAMPLE TYPES ANALYZED FOR THE AVOAS DEMONSTRATION
  Simple Name
Sample Description
Use
                                Purpose
  Purged Effluent Water
  (PEW)
  Spiked Purged
  Effluent Water
  (SPEW)
  Treatment Train (TT)
  Quality Control Check
  Standard (QCCS)
  Continuing
  Calibration Check
  Standard (CCCS)
Water from treatment train sample port
S-6, purged to remove VOCs
Purged effluent water spiked with nine
compounds in four concentration
ranges; analyzed by the AVOAS (off-
line) and EMSL-LV.  SPEW samples
were made in the field from ampoules
of stock solutions prepared prior to the
demonstration.

Samples from treatment train ports S-3
and S-4, analyzed by the AVOAS (on-
line and off-line) and EMSL-LV.
QC standard required by EPA Method
502.2 containing 5 VOCs at low and
very low u,g/L levels, analyzed by both
the AVOAS and EMSL-LV.
A calibration check standard analyzed
daily by both the AVOAS and EMSL-
LV.
For making up matrix spike samples for
analysis by the AVOAS and by EMSL-
LV.

As a blank sample for the AVOAS and as
a travel blank for samples shipped to
EMSL-LV.

For making up matrix spike samples
analyzed by EMSL-LV for the transport
study.

To compare the accuracy and precision
obtained by the AVOAS (off-line) and by
EMSL-LV.

In the transport study, to compare results
between transported and non-
transported samples.

To determine whether AVOAS off-line
results are representative of AVOAS
on-line results

To compare AVOAS and EMSL-LV
measurement performance

To indicate any loss of instrument
sensitivity.
To verify instrument performance in
terms of response factors.
To minimize variability due
to matrix effects of the
treatment train water.
To provide samples of
known concentration in a
variety of concentration
ranges.
To provide actual
environmental samples.
To provide samples of
known concentration so
that instrument
performance can be
monitored.

To provide samples of
known concentration so
that ongoing instrument
performance can be
monitored.
Treatment Train Samples

     The protocol for collection of treatment train
(TT) samples was as follows:  Each morning dur-
ing the routine AVOAS on-line collection and anal-
ysis of a sample from port S-3 or S-4, water from
the port sampled AVOAS was also collected auto-
matically into four 40-mL volatile organic analysis
(VOA) vials.  The first vial collected was analyzed
by the AVOAS immediately after the on-line analy-
sis  of the same sample, using the off-line proce-
dure.    The  TT samples  for  EMSL-LV  were
preserved by acidifying to pH<2 with HC1. In the
afternoon, the alternate port (S-3 or S-4) was sam-
pled in the same fashion.  Both morning and after-
                                  noon TT samples were shipped that day to EMSL-
                                  LV by overnight express carrier.

                                  Purged Effluent Water

                                        Purged effluent water (PEW) was prepared as
                                  follows:  Four gallons of water from sampling port
                                  S-6 (final effluent)  was collected  into a  5-gallon
                                  Pyrex solution bottle on May 16 and was purged
                                  continuously using  high-purity  helium  gas.  The
                                  purged water was poured into four 1-L bottles and
                                  was analyzed in the field using the AVOAS instru-
                                  ment to verify the absence  of VOCs.  Three 1-L
                                  bottles of the purged water were sent to EMSL-LV
                                  for use in the transport study. The remaining liter
                                                    12

-------
TABLE 2-2. SUMMARY OF ANALYTE CONCENTRATIONS (ng/L), BY SAMPLE TYPE
Treatment
Train1
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene
Port
S-3
NO
6.7
0.68C
3.6C
19
37
ND
75
2583
Port
S-4
NO
5.9
0.82C
3.0C
13
31
ND
60
2550
Spiked Purged Effluent Water"
Low
22.4
4.6
13.8
9.1
28.1
9.1
9.0
213
23.7
Medium
37.4
91.1
97.5
45.6
23.7
91.1
39.2
53.8
88.7
High
224
45.6
138
91.1
281
91.1
90.2
213
237
Super
High
747
0
1946
911
467
1822
788
1067
1774
QCCS"
EMSL-
LVand
AVOAS
20.0
2.0
0
0
0
10.0
20
2.0
15.0
cccs"
EMSL-
LV
20
20
20
20
20
20
400
20
20
AVOAS
400
400
400
400
400
400
400
400
400
Transport Study6
Medium
37.4
91.1
97.5
45.6
23.7
91.1
39.2
53.8
88.7
High
227
45.6
138
91.1
281
91.1
90.2
213
237
NO:  Not detected by either the AVOAS or EUSL-LV.
CCCS: Continuing calibration check standard.
QCCS: Quality control check standard.
' Mean concentration. ng/L. Values are approximate and are mean values for all samples collected (a.m. and p.m.) by the AVOAS over the 6-day
   demonstration.
11 Target concentration, fig/L
e Detected by AVOAS, but not by EUSL-LV.
was used to prepare the spiked samples and was
also used as an instrument blank for the AVOAS.

Spiked Purged Effluent Water
     Spiked purged effluent water (SPEW) sam-
ples were freshly prepared each day of the demon-
stration at  the field site.  The solutions used to
spike the PEW had been prepared for each of four
spiking concentration  ranges and  sealed in glass
ampoules at EMSL-LV prior to the  field demon-
stration.  New ampoules were opened daily to pre-
pare SPEW samples,  and the  spiking procedure
was performed by the same chemist to minimize
variability  due to spiking technique. To prepare
each SPEW sample, a 40-mL VGA vial was filled
most of the way with  PEW.  The spiking solution
was  added directly into the VOA  vial using a
micropipette.  Purged  effluent water  was added to
fill the vial, and the vial was capped immediately
and checked for bubbles. If bubbles  were present,
the vial was discarded, and a fresh spiked sample
was prepared.
     Duplicate spike-sample vials were made from
the same ampoule. One sample vial of each con-
centration range was analyzed by the AVOAS in
the off-line mode,  generally within 1 hour of prep-
aration. The other aliquots were preserved (acidi-
fied  with HC1 to pH<2), chilled at 4 °C.  and
shipped to EMSL-LV in coolers. The aliquot ana-
lyzed in the field was not preserved.
     To help maintain analyst objectivity,  neither
the AVOAS  nor  the EMSL-LV  analysts were
informed beforehand of the exact SPEW spiked
concentrations.

Performance Evaluation Study
     A performance evaluation (PE) study  was
conducted prior to the demonstration on behalf of
EPC, Inc. by Trillium, Inc. The PE study had two
objectives: 1) the  determination of method detec-
tion limits (MDLs) and practical quantitation limits
(PQLs) for  both  the EMSL-LV and A+RT,  Inc.
instruments for each of the nine analytes targeted
for the AVOAS demonstration and 2) the analysis
of "blind" (i.e., the composition and concentration
                                              13

-------
 are unknown to the analyst) samples, to provide an
 impartial determination of instrument performance
 in measuring low (ppb level)  concentrations of
 analytes.   EMSL-LV  analyses  were performed
 using modified EPA  Method 502.2 procedures.
 For the AVOAS analyses,  the PE samples were
 introduced  in the  sample manifold in the off-line
 mode.  At the time of the PE study, the AVOAS
 was not yet installed  at the field site, so the PE
 samples were analyzed at the A^RT, Inc. research
 and development facility.

 Determination  of Method Detection and
 Practical Quantitation Limits

      The MDL of a method is defined as its perfor-
 mance in measuring an analyte in a sample matrix,
 regardless of the  sample's origin (Closer et al.,
 1981).  For the PE study, the MDL was set as three
 times the  standard deviation of  the  analytical
                      results for seven replicate analyses, in |ig/L. The
                      PQL was set as 10 times the MDL.
                          A representative ground-water sample was
                      used by each laboratory to make spiked samples
                      for the MDL and PQL determinations. The spiking
                      concentration of each analyte was  1 fig/L.  Seven
                      replicates each  of  spiked and unspiked aliquots
                      were analyzed  alternately,  and the MDLs  and
                      PQLs were calculated for each compound. Table
                      2-3 gives the EMSL-LV and A+RT, Inc. MDLs for
                      each sample type analyzed for the demonstration.
                      The MDLs for  the AVOAS depended on sample
                      volume, as determined by sample loop size. In the
                      Trillium, Inc. study, the 10.0-mL sample loop was
                      used on the AVOAS, and MDLs were calculated
                      for the other loop  volumes.   The MDLs for the
                      EMSL-LV  instrument depended  on the dilution
                      factor used.
TABLE 2-3. SUMMARY OF METHOD DETECTION LIMITS ftig/L) FOR THE AVOAS AND EMSL-LV
INSTRUMENTS, BY SAMPLE LOOP SIZE, DILUTION, AND SAMPLE TYPE
 Compound
                                           AVOAS1
                                                  EMSL-LV
                              10-rnL
                              Loop
           5-fflL
           Loop
1.0-mL
 Loop
0.2-mL
 Loop
  No
Dilution
 1:10
Dilution
  B, L     II.QCCS      H        S,TT
Samples   Simple*    Samplai   Samplai
 1:100
Dilution
                     B,L,M,
                      QCCS       H       S,TT
                     Samples   Samples    Samples
vinyl chloride
1,1-dichloroethene
trans-1 ,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroettiene
tetrachloroethene
0.08
0.05
0.01
0.02
0.02
0.07
0.05
0.03
0.01
0.16
0.1
0.02
0.04
0.04
0.14
0.10
0.06
0.02
0.8
0.5
0.1
0.2
0.2
0.7
0.5
0.3
0.1
4.0
2.5
0.5
1.0
1.0
3.5
2.5
1.5
0.5
0.08
0.16
0.17
0.08
0.24
0.12
0.19
0.22
0.3
0.8
1.6
1.7
0.8
2.4
1.2
1.9
2.2
3.0
8.0
6.0
17
8.0
6.0
23
19
22
30
1 Method Detection Limits (MDLs) were measured using the 10-mL loop; MDLs for the other loop sizes wen calculated from those values.
   For the demonstration, each sample type was analyzed using the loop or dilution indicated in the table headings:
 B:    Blank
 L:    Low-concentration spiked purged effluent water sample.
 M:    Medium-concentration spiked purged effluent water sample.
 H:    High-concentration spiked purged effluent water sample.
 S:    Superhigh-concentration spiked purged effluent water sample.
 QCCS: Quality control check standard.
 77:    Treatment tram sample.
                                                14

-------
Analysis of Blind Performance Evaluation
Samples

     Two PE samples were sent to each of the par-
ticipating laboratories by Trillium, Inc.  Each of the
two  samples  was provided in three 40-mL vials
and  a 1-L bottle. The compounds in one of the
samples were present at concentrations of approxi-
mately 1.7 to 2.5 (Ag/L, and in the other at approxi-
mately 0.1 to 0.6 |ag/L. Trillium, Inc. determined
that  the EMSL-LV and AVOAS results were com-
parable.

Transport Study
     The effect  of shipping  on sample integrity
was  supposed to be studied concurrently with the
AVOAS demonstration, using the AVOAS to ana-
lyze   transported and  nontransported  samples.
However, scheduling commitments for the AVOAS
required the transport study to be delayed to a later
date. The transport study was performed from
May 30 to June 13, using the EMSL-LV instrument
for all comparisons, as described below.
     The spiked samples used for  the transport
study were made in the same manner as the SPEW
samples, using the PEW that had been shipped to
EMSL-LV during the demonstration  and stock-
solution ampoules originally  intended for the
Wobum transport study.    Two concentration
ranges, medium and high, were tested. Four 40-mL
VOA vials of each of the two concentration ranges
were prepared  daily  for  5  days, using fresh
ampoules each day. Two vials from each concen-
tration range were shipped and a corresponding set
was not shipped.  The  transported vials were pre-
served by acidifying to pH<2; the nontransported
vials were not The transported vials were shipped
in coolers at approximately 4 °C in the same man-
ner as the demonstration samples had been, using
the  same  overnight   carrier  service  (Federal
Express).  Samples were sent from Las Vegas to
the central Federal Express shipping hub, and were
returned the  following morning to Las Vegas.
Nontransported samples were analyzed within 6
hours of preparation;  transported samples were
analyzed within 30 hours of preparation.  For the
first 3 days of the study the samples were prepared
and analyzed by one chemist, and for the last 2
days by a different chemist

Quality Assurance for the AVOAS
Demonstration
     The purpose of  the  QA program  for the
AVOAS demonstration was to ensure that the qual-
ity of data produced would be sufficient to allow
formulation of sound conclusions about the perfor-
mance of the AVOAS relative to the comparison
method.
     According to the description of project cate-
gories specified in the EMSL-LV Quality Assur-
ance Program Plan (U.S. EPA, 1987), the AVOAS
demonstration is considered a Category n project.
Category n projects apply to measurement activi-
ties whose results are used by a client organization,
but not directly in regulatory, enforcement, legal,
or policy matters. The QAPjP addressed the key
elements required for Category n projects, and was
prepared in accordance with the guidance in Prepa-
ration Aids for the Development  of Category n
Project  Plans  (U.S. EPA, 1991).   The  QAPjP
clearly identified and delegated  responsibilities
related to all aspects of QA for this demonstration,
defined  data  quality  objectives  (DQOs), and
described the measures required  to  achieve  the
DQOs.

Data Quality Objectives

     The EPA recommends that DQOs include, at
a minimum, five indicators of data quality (Stanley
and Vemer  1983):  representativeness, complete-
ness, comparability, accuracy, and precision. Table
2-4 summarizes the DQOs for the AVOAS demon-
stration.
     DQOs were set to ensure that the study was
well enough designed and carried out for the com-
parisons to be as tree from confounding and extra-
neous sources of  variability as  possible.   In
addition, because a fair assessment of the AVOAS
required that the performance of Method 502.2 be
known, it was important that DQOs be specified
for that method.   DQOs were not set  for  the
AVOAS because  the performance of that  system
was unknown and its assessment was an objective
of the demonstration.
     Representativeness — Representativeness is
defined as "the degree to which the data accurately
and precisely represent a characteristic or a param-
eter, variation of a property, a process characteris-
tic, or an operation condition" (Stanley and Vemer,
1985). Representativeness was a particular  issue in
this demonstration because the AVOAS is designed
to collect samples directly from a water treatment
stream with no manual  intervention. In  normal
operations, the samples are never  exposed to air.
For this demonstration, analysis of spiked samples
(i.e., the SPEW, QCCS, and CCCS samples) by the
AVOAS required that the operator disconnect the
                                             15

-------
sample line and introduce the samples in the off-
line mode. An estimate of the degree of represen-
tativeness of spiked samples was made by compar-
ing the results of TT samples analyzed on-line and
off-line by the AVOAS and by EMSL-LV.  Repre-
sentativeness was  optimized by  using consistent
sample collection and preparation techniques and
by taking measures to ensure that sample integrity
was maintained to the highest degree possible.
     Completeness — The completeness of a data
set can be reduced by contamination, loss of sam-
ples due to vial breakage, and problems with cali-
bration, among other things. Completeness can be
monitored through use of appropriate QC samples,
i.e.,  instrument  and travel  blanks, CCCSs,  and
QCCSs.  The AVOAS demonstration plan  was
designed to ensure maximum use of the resulting
data.  Because each analysis provided more than
one type of data, for both QC and statistical analy-
sis purposes, the completeness objective for this
demonstration was 100%.
                                       Comparability — Comparability is defined as
                                  "the confidence with which one data set can be
                                  compared to another" (Stanley and Verner, 1985).
                                  Analytical results from SPEW, TT,  and QC sam-
                                  ples were used to evaluate the comparability of the
                                  two methods.

                                       Accuracy — Accuracy refers to the degree of
                                  agreement of  a  measured value with the  true, or
                                  reference value  for  each analyte (Taylor, 1987).
                                  Accuracy was assessed using data from daily anal-
                                  yses of SPEW and QCCS samples and, for EMSL-
                                  LV analyses,  surrogate compound  recoveries.
                                  Accuracy was computed as the percent recovery
                                  (%R) of measured concentrations relative to spiked
                                  (target) concentrations for each analyte:
                                  where:
                                      €„, = measured concentration of analyte (p.g/L)
                                      Ct = target concentration of analyte (jig/L)
 TABLE 2-4. DATA QUALITY OBJECTIVES FOR THE AVOAS DEMONSTRATION
 DQO Element
Demonstration Purpose
Computation
DQO
 Representativeness
 Completeness

 Comparability
Assesses how well off-line AVOAS measurements        Difference in mean %R for on-  None set.
represented on-line AVOAS measurements to determine    line and off-line treatment train
whether on-line AVOAS and EMSL-LV measurements could  samples.
be compared.                                 Analysis of variance.
Assesses the amount of useable data collected.          %C

Determines how well AVOAS and EMSL-LV data compare.
                       100%

                       None set.
                  SPEW samples:
                   off-line AVOAS/EMSL-LV.

                  TT samples:
                  On-line AVOAS/ofMine AVOAS.
                  On-line AVOAS/EMSL-LV.
                  Off-line AVOAS/EMSL-LV.
                                           • %R
                                           • Difference in mean %R.

                                           Difference in mean %R.
Accuracy
Precision
Compares agreement of AVOAS measurements with target %R
concentrations and with EMSL-LV measurements.
Compares variability among replicate measurements for the %RSD
AVOAS and EMSL-LV.
•%R within 100 ±30%
for Method 502.2*
%RSD 530% for
Method 502.2*
%fl:     Percent recovery, computed as Xfl = (Cm/Ct) • 100, when Cm = measured analyte concentration (fig/L),
          C, = target analyte concentration (ng/L).
Difference in mean %R: Difference in mean percent recovery, computed as mean %fl tor data set 1 minus mean %R for data set 2.
XRSD:   Percent relative standard deviation,  computed as %RSD = (standard deviation/mean) • 100.
%C:     Percent completeness, computed as %C = 100 • V/n, where
          V = number of valid measurements, n = total number of measurements.
' OOOs were not set for the AVOAS because the performance of that system was unknown and its assessment was an objective of the
   demonstration.
                                                 16

-------
     The accuracy DQO for the EMSL-LV instru-
ment, based on Method 502.2 performance data
and on performance data presented in a document
evaluating various water analysis methods (EPA,
1988), was 100 ±30%R for measured concentra-
tions  relative to target concentrations  of spiked
compounds.
     Precision —  Precision  is  defined as  the
degree of mutual  agreement  among  individual
measurements  (Taylor, 1987).   Precision values
were computed for each compound and each sam-
ple type over the 6 days of the demonstration for
the AVOAS on-line and off-line and the EMSL-LV
measurements.  Precision  for this demonstration
was expressed as percent relative standard devia-
tion (%RSD), calculated as:
%RSD =
                   x 100
where:
      a = standard deviation
      \i = mean value of results over the 6 days of
          the demonstration
     The precision DQO for Method 502.2 analy-
ses was <30 %RSD.

Instrument Calibration
     The AVOAS and EMSL-LV instruments were
calibrated within 2 days before the start of the
demonstration.   A 5-point calibration curve was
used for the AVOAS,  and a 4-point calibration
curve was used for the  EMSL-LV instrument To
monitor instrument response during the demonstra-
tion, single-point calibration checks were per-
formed daily using  the CCCSs.  The calibration
standards for both instruments came from the same
supplier and lot so as to eliminate potential vari-
ability due to differences in concentration that can
occur in stock solutions from different sources.
     The performance, expressed as percent differ-
ence (%D) between the initial and continuing cali-
bration responses for each analyte, was calculated
as follows:
     %D = (C1-C2)xlOO/C1
where:
     GI = response in the initial calibration
     C2 = response in each continuing
          calibration
     A 20% difference between initial  and con-
tinuing calibration was  set as the acceptance crite-
rion for the EMSL-LV instrument.
Quality Control Samples
     Quality Control Check Standard — To moni-
tor  instrument  sensitivity  and performance, the
QAPjP required daily analysis of a quality control
check  standard (OjCCS)  by  each instrument.
QCCSs were made from commercially prepared
stock standards (Restek) having certified concen-
trations of five target analytes.  The QCCS samples
for the AVOAS  were made daily at the field site by
the LESC supervising chemist using PEW; a new
ampoule  of stock  standards  was  used each day.
The QCCS for the EMSL-LV instrument was made
by  the EMSL-LV analyst  using reagent-grade
water, the same stock solution ampoule was used
for the duration of the demonstration.
     Continuing Calibration  Check Standards —
The QAPjP required daily analysis of continuing
calibration check standards (CCCSs), to provide an
ongoing  indication of instrument  performance.
CCCSs were one of the standards that had been
used to calibrate the instruments prior to the dem-
onstration.  For  the EMSL-LV  instrument, the
CCCS concentration was 20 jig/L; for the AVOAS,
it was 400 ng/L.  The CCCS for the AVOAS was .
made daily by the AVOAS operator from a fresh
stock standard ampoule; the EMSL-LV daily check
was performed using a standard  prepared at the
beginning of the demonstration.
     Instrument Blank Samples  —  Instrument
blanks  were samples of VOC-free  water (PEW,
reagent-grade water, or VOC-free water from sam-
ple port S-6) analyzed to verify that the analytical
systems were not contaminated. For the systems to
be considered free of contamination, target analyte
concentrations were required to be below the low-
concentration SPEW sample or the MDL, which-
ever was less.
     Travel Blank Samples — Travel blank sam-
ples were used  to determine whether routine sam-
ples may  have been contaminated during shipping
or handling.  Travel blanks were unspiked  PEW
sent with each batch of TT and SPEW samples to
EMSL-LV. Travel blank samples were also used in
the transport study.

Demonstration Methods

EPA Method 502.2 Requirements and
Modifications
     The procedures for instrument setup, initial
calibration,  continuing  calibration checks,  addi-
tional QA/QC measures, and associated corrective
                                            17

-------
actions are specified in EPA Method 502.2 and are
summarized in Table 2-5.  Some modifications to
the method were used for this demonstration:
1. Method 502.2 requires the use of a photoioniza-
   tion detector (FED),  and electrolytic conductiv-
   ity detector (ELCD). in  series.  The AVOAS
   instrument  was  used with  only  the  ELCD
   because the target analytes in this study were
   halogenated hydrocarbons, which are best mea-
   sured  using this detector.  Both PID and ELCD
   were used with the EMSL-LV instrument; how-
   ever, the PID served only  as an operator perfor-
   mance  guide,  and   the   PID   data  were  not
   included in  the demonstration database.
                                   2. Compounds  with  chromatographic  retention
                                     times greater than that of PCE were not detected
                                     in any of the samples. Therefore, the analytical
                                     run times  were shortened to increase sample
                                     throughput (i.e., the number of samples that can
                                     be analyzed in a given amount of time).

                                   3. Method 502.2 requires that reagent water forti-
                                     fied with target analytes at low concentrations
                                     (just above the method PQLs). be analyzed as a
                                     laboratory-fortified blank once each day or once
                                     every 20 samples.  This QC  procedure was not
                                     included in the demonstration, since five  of the
                                     eight  daily analyses were performed  on spiked
                                     water samples.
TABLE 2-5. METHOD 502.2 QUALITY CONTROL REQUIREMENTS
 QC Element
Speciflcationi
Acceptance Criterli
Corrective Action
 1.  Initial calibration
 2.  Continuing calibration
    verification
 3. Initial demonstration of
   precision and
   accuracy
   Note:  These analyses
   were performed as
   part of the PE study.

 4. Quality Control Check
   Standards
   (commercially
   prepared)
 5. Instrument/reagent
   blank
   Purged  effluent  water
   blank
 7. Surrogate
All target compounds and
surrogate(s) at three-five
concentration levels

One or more standards analyzed
on each working day
Four to seven replicates of a low
level (0.5 • 5 ng/L) water sample
• Analyze once each day at the
 end of sample analysis
• Compounds at concentrations
 known to analysts
• Daily, prior to standard or
 sample analysis
• Following samples containing
 analytes > the high standard
< 10% RSD over concentration range     Plot calibration curves for
   required to use mean RF for quantitation analytes with RSOs >10%
RF or response (height or area) for each
analyte 120% of RF or response for that
analyte in the initial calibration standard

• The mean accuracy (% recovery) should
 be 80-120 %, and the
 %RSD should be < 20 %
• The MOL must be sufficient to detect
 analytes at the regulatory levels
• Recommended recoveries are
 70-130%
• Monitor any loss of instrument sensitivity
 over the day, based on low-level analyte
 recoveries

No analytes present >MDL or low level
standard, which ever is less
• One per sample shipment batch  No analytes present >MDL or low level
• Trip blank for laboratory       standard, which ever is less
 samples
• In each blank, TT sample,
 spiked sample, and standard
•2-bromo-1-chloropropane
Recommended recovery is 80 -120 %
•Repeat the test using a fresh
 standard
• If test fails, recalibrate or prepare
 a new curve

Perform new initial calibration or
perform maintenance, and rerun
the analyses
None specified, limits are advisory
»Repeat until blank meets the
 acceptance criteria
• Decontaminate instrument
                                  None specified, flag sample
                                  results with 'B'
• Check calculation, standard
 solution
• No other specified, flag
 associated sample data as
 estimated,' J1
%RSD:   % Relative standard deviation
MDL:     Method Detection Limit
RF:      Response factor
TT:      Treatment Train
                                                      18

-------
4. A surrogate compound was not included in the
   QC  requirements  for the AVOAS, because a
   sample  containing a surrogate could not be
   introduced into the AVOAS in the on-line mode.

Sample Analysis

     Table 2-6 summarizes the components and
operating conditions used for the AVOAS and
EMSL-LV analytical systems during the demon-
stration.

AVOAS
     A  Scientific Research Instruments Model
8610 GC equipped with a Tremetrics Model 1000
ELCD  was used with  the AVOAS for  the field
demonstration.   The GC  signal integrator was a
Spectra-Physics, Model SP4400. An IBM-compat-
ible 286 microcomputer was used to control the
operation of the analytical system and to collect
and store the analytical data.  The AVOAS was
operated at the  field site by its inventor.   Off-line
TT samples were generally analyzed immediately
after collection.  SPEW samples were  analyzed
within several hours of preparation.

EMSL-LV Method 502.2 Analyses

     EMSL-LV analyses were performed using a
Hewlett Packard  (HP)  Model 5890 Series 2 GC
with an OI Model 4420 ELCD. The purge-and-
trap unit was an 01 Model 4460 with an MPM-16
autosampler.  The integrator was an HP 3396. All
analyses for the demonstration were performed by
the same analyst, who was experienced in opera-
tion of the instrument Samples were received the
morning following  shipment from Woburn and
were usually analyzed mat day.  Exceptions were
samples from the first  day of the demonstration
were analyzed by EMSL-LV 2 days after they were
shipped from Wobum, and those from the last day
were analyzed 4 days after they were shipped.

Data Management
     Data  for both the AVOAS and EMSL-LV
were handwritten onto  forms.  The AVOAS data
were checked  for compound  identification and
integration errors, completeness, and consistency
at the field site by the AVOAS operator. EMSL-LV
results were checked initially by the LESC analyst
and were reviewed by the  project QA manager.
Hard copies of the chromatograms and correspond-
ing data (including chromatographic peak retention
times, peak areas, calculated concentrations, injec-
tion volumes, injection  times, and sample identifi-
ers) for all analyses, and instrument logbook pages
were submitted for review.  The LESC data base
manager reviewed the field and laboratory results,
identified any data that appeared unusual, and rec-
onciled the  results prior to data base  entry.  All
database entries were checked by a second per-
son.  Final hard copy and  electronic results were
accompanied by case narratives that described any
problems or data interpretation difficulties, and
associated corrective actions.

Data Analysis

     In addition to the computations described in
the QA/QC  section, comparisons of differences in
mean %Rs for sample pairs were computed. The
advantage of using paired samples (with the pairs
being prepared and sampled under as nearly identi-
cal conditions as possible) is that the effect of
uncertainty in the true concentrations of samples is
reduced;  also,  accuracy  comparisons thereby
become insensitive to any  possible differences in
precision between  the two methods being com-
pared. The analyses were done mostly  in terms of
%Rs rather  than actual measurements  simply for
ease of interpretation and comparison across differ-
ent concentration levels and  compounds.  This
approach was taken for  the SPEW and transport
study data, because of uncertainties in actual spik-
ing concentrations. For TT samples, this approach
was necessary because the true  concentration of
those samples was unknown and possibly varied
with time.
     For the purpose of obtaining approximate tar-
get concentrations to compute %Rs for TT sam-
ples, a weighted least-squares regression line was
fit through the  origin for SPEW samples for the
AVOAS off-line and, separately, for  EMSL-LV
analyses.   Mean measured concentrations  were
adjusted by the respective line to provide estimates
of target concentrations for each port.  These esti-
mates were averaged when  both AVOAS  and
EMSL-LV estimates were  available.   Measure-
ments were converted to approximate %Rs so that
results could be expressed on the same scale across
different  true  concentration  ranges  and com-
pounds.  This conversion does not affect the preci-
sion results  or the hypothesis tests of whether the
mean recoveries are equal.  The target concentra-
tion value is rough, and is used only for converting
differences in mean measurements between mea-
surement modes so as to approximate  differences
in mean %Rs.
                                            19

-------
TABLE 2-6. INSTRUMENT COMPONENTS AND OPERATING
Instrument
Gas Chromatograph
Column
Carrier gas, flow rate
Oven temperature program
Purge and Trap Sampler
Purge gas, flow rate
Purge time
Desorb time
Bake time
Trapping temperature
Oesorb preheat
Oesorb temperature
Bake temperature
Valve/transfer line temp.
Hall Detector
Reactor temperature
Reactor base temperature
Makeup gas, flow rate
Reaction gas, flow rate
n-Propanol flow
Photoionization Detector
Intensity
mV Signal
. Base temperature
Integrator
Computer
Software
AVOAS System
SRI 8610
DB-624, 30-m (J&W)
He, 6 cc/min
Initial 35°C
0.5 min 408C
2.5 min 45°C
10.0 min 50°C
12.5 min 65°C
A'RT
He, 12 cc/min
10 min
2 min
8 min
30°C
180°C
215°C
215°C
130°C
Tremetrics
885°C
250'C
He, 25 cc/min
H2, 25 cc/min
60 jiL/min
Not used
-
—
—
Spectra-Physics 4400
IBM-Compatible 286
A*RT
CONDITIONS
EMSL-LV
HP 5890 Series 2
RTX502.2, 105-m (Restek)
He, 10 cc/min
45°C, 60C/minto143°C
(no hold time)
01 4460 with MPM-16 auto-sampler
He, 40 cc/min
11 min
2 min
12 min
28"C
165°C
180"C
180°C
80°C
01 Model 4420
8008C
240°C
He, 30 cc/min
H2, 85 cc/min
50 u,l/min
01 Model 4330
6
11.9
2358C
HP 3396
IBM 286 PC
HP File server
20

-------
TABLE 2-7. CHRONOLOGY OF AVOAS DEMONSTRATION ACTIVITIES
 Predemonstration Activities

   Project planning, experimental design

   Preparation of Demonstration and QA Plans

   Performance Evaluation Study
                    Date

                    March, 1991

                    April, 1991

                    April 12-15,1991
   Preparation of ampoules of stock solutions for spiked purged effluent water samples at EMSL-LV  May 8,1991

 Demonstration Activities

   Installation of AVOAS                                                   May 14*17,1991

   Preparation of purged effluent water at field site                                  May 16,1991

   Demonstration:AVOAS analyses                                            May 18-24,1991

   EMSL-LV Method 502.2 analyses                                           May 20-May 28,1991

 Transport Study                                                         May 30-June 13,' 1991
     The %Rs for each of the method pairs were
computed for each day, and the mean values for
each method over days were computed. The differ-
ences were assessed  using  paired-sample t-tests
with a hypothesis of no difference in mean %Rs.
Confidence  intervals  were also computed.   For
similar performance between two systems or treat-
ments, one would expect a difference in mean %R
of zero, with some allowance for random variabil-
ity. The major advantage  of paired comparison t-
tests is their simplicity and ease of interpretation.
In  statistical terms,  the  paired-sample t-test
amounts to blocking all other effects, and focusing
only on the one effect of  interest  Robustness  is
attained at the cost of degrees of freedom.
     Paired-sample t-tests were used to  assess all
pairs of sample data.  An Analysis of Variance
(ANOVA) was used to evaluate TT sample data
with respect to time of sample collection, day-to-
day variability, mode of analysis (i.e., AVOAS on-
line, AVOAS off-line, and  Method 502.2), and
sample  port  An ANOVA  models  other effects
which are either expected to be present or are pos-
sibly present  and cannot be  ruled  out   In  an
ANOVA analysis, not all other effects which might
be present are accommodated.  On  the other hand,
if the model fitted for the other effects reasonably
accommodates those that are in fact present one
obtains  more  degrees of  freedom  for error esti-
mates,  and  hence better sensitivity  (power).  In
addition, the ANOVA allows one to assess those
other effects.
AVOAS Demonstration Activities

     A chronology of the activities and schedules
for the AVOAS  demonstration  are  included  in
Tables 2-7 and 2-8.
     Activities prior to the demonstration included
site reconnaissance, preparation of the Demonstra-
tion and QA Plans, the PE study, and preparation
of spiking ampoules.  Demonstration participants
arrived at the Woburn field site on May 13. While
the AVOAS was being installed, tested, and cali-
brated by A+RT, Inc., the EMSL-LV demonstration
participants  selected sample ports to be used for
the demonstration, set up an area for preparing and
shipping samples, and prepared  and  checked the
PEW.
     Daily activities at  the  field site during  the
demonstration included preparing SPEW, CCCS,
and QCCS samples; collecting TT samples; prepar-
ing and shipping  samples to EMSL-LV; entering
and reviewing data;  and analyzing  the required
AVOAS samples.  Daily activities at the EMSL-LV
laboratory included preparing QCCS and CCCS
samples, receiving and logging  in samples, per-
forming the surrogate spikes, analyzing samples,
and  entering  and reviewing  data  on the  data
spreadsheets.
     Activities performed after the field portion of
the demonstration included the transport study,
data validation and verification, and data  analysis
and interpretation.
                                               21

-------
TABLE 2-8. SCHEDULE OF SAMPLE ANALYSES FOR THE AVOAS AND EMSL-LV
AVOAS
5/18/91
B(S6)
TTF (S3)
TTF (S3)
CCCS
L
H
S
M
QCCS
TIN (S4)
TTF (S4)
5/20/91
B(S6)
TTF (S4)
TTF(S4)
CCCS
S
H
M
L
QCCS
TTN (S3)
TTF (S3)
15/21/91
TTN (S4)
TTF (S4)
CCCS
L
M
M
S
QCCS
TTN (S3)
TTF(S3)
B (PEW)
5/22/91
TTN (S4)
TTF (S4)
CCCS
S
L
M
H
QCCS
TTN (S3)
TTF (S3)
B (PEW)
5/23/91
TTN (S3)
TTF (S3)
CCCS
H
M
S
L
QCCS
TTN (S4)
TTF (S4)
B (PEW)
5/24/91
TTN (S3)
TTF (S3)
CCCS

TTN (S4)
TTF(S4)
L
M
H
S
QQCS
EMSL-LV
5/20/91
(Samples
5/18/91)
B (Water)
CCCS
B (PEW)
L
M
H
S
TTA (3)
TTP (3)
QCCS

fl (S6):
B (PEW):
B (Water):
CCC:
QCCS:
Sur:
TTA:
TTF:
TTN:
TTP:
5/21/91
from (Samples from
5/20/91)
B (Water)
Recal'
B (PEW)
L
M
H
S
TTA (2)
TTP (2)
B (PEW)
QCCS
5/22/91
(Samples from
5/21/91)
B (Water)
CCCS
B (PEW)
H
L
M
S
TTA (2)
TTP (2)
QCCS

Recalibrated with 2, 20, 50, WO ppb standards.
VOC-lree sample from port 56 (blank)
VOC-lree purged effluent water (blank)
Reagent-grade water (blank)
Continuing calibration check standard.
Quality control check standard.
Reagent-grade water spiked with 2-bromo-l-
chloropropane surrogate
Treatment train sample, a.m.
Treatment train oil-line sample (AVOAS).
Treatment train on-line sample (AVOAS).
Treatment train sample, p.m.
5/23/91
(Samples
5/22/91)
B (Water)
CCCS
Sur
B (PEW)
L
M
H
S
TTA (2)
TTP (2)
QCCS
(S3):
(S4):
(2) or (3):
H:
L:
M:
S:
5/24/91
from (Samples from
5/23/91)
Sur
CCCS
B (PEW)
L
M(2)
H
S
TTA (2)
TTP (2)
QCCS

5/28/91
(Samples from
5/24/91)
Sur
CCCS
B (PEW)
L.
M
M
S
TTA (2)
TTP (2)
QCCS
QCCS
Port S-3 sample.
Port S-4 sample.
Number of sample replicates, analyzed.
High spiked purged effluent water samples.
Low spiked purged effluent water samples.
Medium spiked purged effluent water samples.
Superhigh spiked purged effluent water samples.
22

-------
                                        Section 3
                    AVOAS Demonstration Results and Discussion
Operational Results

     The AVOAS demonstration went smoothly.
Both the AVOAS and the EMSL-LV instruments
were operational for the entire demonstration, and
all of the samples specified in the Demonstration
Plan were collected.

     Initial installation and preliminary testing of
the AVOAS took 2-1/2 days. The problems that
occurred during installation were typical for an ini-
tial field site setup, and included delays and dam-
ages in shipping and the need to modify the site
facilities (e.g., installing an air conditioner and
vent).

     During the ground-water treatment system
pilot study of May 18 through May 30, 1991, the
AVOAS processed 65 samples for the SITE dem-
onstration (not including QC samples), 192 auto-
mated  treatment system  samples,  52 samples
analyzed for another treatment project, and 41 cali-
bration and blank samples.  The AVOAS ran an
average of  22.3 hours per day and processed an
average of 24.4 samples per day, at an average run
time of 55 min per sample. Further details of the
AVOAS field performance,  as  reported by the
developer, are included as Appendix A.
     Deviations  from  the Demonstration Plan
were minor, and their impact on the results is
judged to be negligible. Modifications were made
because of the large number of samples scheduled
to be processed by the AVOAS during the same
period, beyond those required for this demonstra-
tion.  Deviations included:

1. Rescheduling the transport study until after the
  demonstration, with analysis of the  transport
  study samples by EMSL-LV rather than by the
  field team.

2. Adjusting the planned sequence for sample anal-
  ysis.  Originally, the plan was to analyze the
  CCCS as the first sample of the day, and the
  QCCS as the last  The heavy  analysis schedule
  for  the  SITE demonstration  and  for  other
  projects  required  modifying  the  sampling
  sequence so that the TT samples from port S-3
  or S-4, which were analyzed last for routine
  monitoring purposes, would be the first samples
  for the AVOAS demonstration, analyzed as on-
  line TT samples.

3. Using  the routine TT sample  from port S-6
  instead of a PEW sample as a system blank.
  This sample was used only if the S-6 sample
  was found to have target analyte concentrations
  <1.0 u.g/L.

4. The AVOAS requires flushing of sample loops
  between analyses, and has a built-in mechanism
  for producing VOC-free water from TWP water
  for that purpose. However, at the demonstration
  site, the tap water contained a substance (possi-
  bly residual flux from new copper tubing) that
  caused a  foaming  problem. Bottled,  purified
  water was used instead for flushing.

Quantitative Results for Comparison
Studies

Comparison of the AVOAS and Method
502.2 Results for Treatment Train Samples
     Tables 3-l(a-c) are compilations of results for
TT  samples  analyzed by EMSL-LV and by the
AVOAS in the on-line and off-line modes. Tables
3-2(a,b) summarize mean, standard deviation, and
%RSD for EMSL-LV and AVOAS data by port and
time.
     Table 3-3 summarizes differences in mean
%Rs for analytical mode comparisons for each of
the  three paired-analysis combinations (AVOAS
on-line/AVOAS off-line; AVOAS off-line/EMSL-
LV; AVOAS on-line/EMSL-LV) for samples from
ports S-3 and S-4 and for bom ports combined,
over the  demonstration  period. Paired-sample t-
tests of the hypothesis of no significant difference
in mean  %Rs were performed, and significance
results are shown in Table 3-3.
                                            23

-------
TABLE 3-1 (a). TREATMENT TRAIN SAMPLE RESULTS
Compound
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
trichloroethene
tetrachloroethene
MOL
(ng/L)
6.0
17.0
8.0
6.0
23.0
22.0
30.0

1
a.m.
S-3
7.1
U
U
12.8
26.6
53.1
2042


p.m.
S-4
6.5
U
U
U
37.9
49.3
2177
Detected
2
a.m.
S-4
U
U
U
11.6
22.8'
50.3
1882
FOR EMSL-LV
ANALYSES
Concentration, |ig/L, for Demonstration

p.m.
S-3
U
U
U
10.1
22.6'
49.3
2006
3
a.m. p.m.
S-4 S-3
U U
U U
U U
8.3 11.3
19.8' 25.3
41.4 50.2
1915 1824
4
a.m. p.m.
S-4 S-3
U 6.2
U U
U U
7.6 13.5
19.4* 24.3
42.9 54.9
1837 2188




Days 1 through 6
5
a.m.
S-3
6.4
U
U
13.5
27.8
56.1
2204
p.m.
S-4
U
U
U
8.2
17.3'
39.4
2157

a.m.
S-3
7.9
U
U
17.4
38.5
73.1
2133
6
p.m.
S-4
5.9'
U
U
10.1
27.7
45.3
2207

TABLE 3-1 (b). TREATMENT TRAIN
Compound
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
trichloroethene
tetrachloroethene
UDL
(H9/L)
2.5
0.5
1.0
1.0
3.5
1.5
0.5

1
a.m.
S-3
9.2
U
4.1
16.5
30.3
56.3
1924
SAMPLE RESULTS


p.m.
S-4
11.7
2.4
4.8
21.0
51.7
82.6
2713
Detected
2
a.m.
S-4
9.8
U
2.6
12.8
20.1
48.1
1986
FOR THE AVOAS ON-LINE ANALYSES
Concentration, ng/L, for

p.m.
S-3
6.8
1.4
4.1
17.6
33.8
64.3
2124
3
a.m. p.m.
S-4 S-3
4.0 7.4
U 2.2
2.7 4.4
12.5 19.7
23.5 36.2
48.6 74.4
2058 2368
Demonstration
4
a.m. p.m.
S-4 S-3
7.1 7.1
U 1.6
3.3 3.7
13.4 20.0
25.0 33.2
50.1 72.6
2249 2370
Days 1 through 6

a.m.
S-3
5.8
U
3.4
19.3
36.8
77.3
2366
5
p.m.
S-4
4.9
1.7
3.5
15.5
30.5
63.0
2385

a.m.
S-3
3.7
U
3.7
23.8
45.8
95.1
3030
6
p.m.
S-4
5.2
1.5
3.4
17.7
36.6
72.3
2758

TABLE 3-1 (C). TREATMENT TRAIN
SAMPLE RESULTS
FOR THE AVOAS OFF-LINE ANALYSES
Detected Concentration, ug/L, for
Compound
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
trichloroethene
tetrachloroethene
UDL
(H9/L)
2.5
0.5
1.0
1.0
3.5
1.5
0.5
1
a.m.
S-3
7.3
U
4.2
16.7
35.0
58.0
1944

p.m.
S-4
9.1
1.7
3.8
15.7
39.2
63.3
2406
2
a.m.
S-4
5.0
U
2.6
12.2
26.4
49.3
2123

p.m.
S-3
5.3
1.3
3.8
17.8
35.7
64.7
2195
3
a.m. p.m.
S-4 S-3
3.2 6.9
U 1.5
2.6 4.2
11.8 20.5
25.3 37.4
46.6 72.3
2039 2365
Demonstration
4
a.m. p.m.
S-4 S-3
5.3 6.4
U 0.9
3.0 3.5
12.9 18.5
25.8 36.4
50.9 71.0
2289 2391
Days 1 through 6

a.m.
S-3
3.1
U
3.2
17.6
39.8
73.6
2219
5
p.m.
S-4
5.3
1.7
3.2
14.7
32.3
64.2
2278

a.m.
S-3
3.3
U
3.7
23.1
45.5
93.2
2921
6
I-J'
6.1
1.1
3.8
16.3
36.3
70.1
2806
Vinyl chloride and 1,2-dichloroethane were not detected by either EMSL-LV or the AVOAS, so those compounds are not included in this table.
MDL:   Method Detection Limit
U:     Below instrument detection limits.
':     Estimated value; measured concentration below calculated method detection limit.

                                                              24

-------
     Table 3-4 summarizes the results of analysis
 of variance (ANOVA) computations, performed to
 determine the significance of effects  on the data
 attributable to port (S-3 or S-4); time of day (mom-
 ing- or afternoon-collected samples); day-to-day
 differences over  the  demonstration period;  and
 mode or analytical system differences  (AVOAS
 on-line, AVOAS off-line, and EMSL-LV).  Mea-
 sured concentrations were used for the ANOVA
 computations.  Results for each compound were
 computed separately for the EMSL-LV dat«. the
 AVOAS data, and for the AVOAS and EMSL-LV
 data combined.

 Mode Comparisons

     AVOAS On-line and Off-line Mode Compari-
 sons — For all compounds and ports, the measured
concentrations were greater for on-line than for
off-line  measurements  (except  for  1,1,1-TCA).
The difference in mean %R results indicated, how-
ever, that this difference was not significant or only
slightly significant for most pairs. The difference
in mean %R was greater for the two most volatile
compounds (1,1-DCE and trans-1,2-DCE) than for
the other compounds. It must be pointed out that
the concentrations for the higher volatility com-
pounds in the TT samples were very low (close to
instrument  detection  limits) for 1,1-DCE, trans-
1,2-DCE, and 1,1-DCA compared to the lower vol-
atility compounds, particularly  TCE and  PCE,
which had concentrations of around 65 (ag/L and
over 2300 jig/L, respectively. For TCE and PCE,
the differences in mean %R between on-line and
off-line sample analyses were very low (less than
TABLE 3-2(a). SUMMARY OF MEAN, STANDARD DEVIATION, AND %RSD FOR PORT MEASUREMENTS FOR
EMSL-LV, AVOAS ON-LINE, AND AVOAS OFF-LINE ANALYSES OF TREATMENT TRAIN SAMPLES ACROSS
THE SIX DAYS OF THE DEMONSTRATION

Compound
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
trichloroethene
tetrachloroethene

Compound
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
trichloroethene
tetrachloroethene

Hean
EHSL
LV
*
U
U
13.1
27.5
56.1
2066

Hcan
EHSL
LV
6.2
U
U
9.2C
24.2
44.8
2029


Concentration, iig/L
• AVOAS
On-Llne
6.7
b
3.9
19.5
36.0
73.3
2364

AVOAS
Off-line
5.4
b
3.8
19.0
38.3
72.1
2339

Concentration, iig/L
• AVOAS
On-Llne
7.1
b
3.4
15.5
31.2
60.5
2358
AVOAS
Off-lint
5.7
b
3.2
13.3
30.9
57.4
2324

Port 3


Standard Deviation
EHSL-
LV
—
—
—
2.5
5.7
8.7
142

AVOAS
On-Llne
1.8
b
0.4
2.5
5.3
13.1
373
Port 4
AVOAS
Off-lint
1.8
b
0.4
2.4
3.9
11.9
323

EHSL-
LV
—
—
-
19.1
20.6
15.5
6.9 .

Standard Deviation
EHSL-
LV
0.4
—
—
1.7
7.6
4.4
168
AVOAS
On-Llne
3.0
—
0.8
3.3
11.6
14.8
325
AVOAS
Off-line
1.9
—
0.5
1.9
6.0
9.7
270
EHSL-
LV
6.8
—
—
18.0
31.6
9.7
8.3

%RSD
AVOAS
On-Llne
27.4
b
9.3
12.9
14.8
17.9
15.8

%RSD
AVOAS
On-Llne
42.9
—
23.3
21.6
37.1
24.4
13.8


AVOAS
Off-line
33.8
b
10.4
12.4
10.2
16.4
14.0


AVOAS
Off-line
34.2
—
17.1
13.6
19.3
16.8
11.6
U: Below instrument detection limits.
1 Six of 12 values were 'U'; no computations.
* Trans-1,2-dichlomothene was detected in p.m. samples only.
" Based on n=5, because one value was 'U.'
                                             25

-------
 TABLE 3-2(b). SUMMARY OF MEAN, STANDARD DEVIATION, AND %RSD FOR SAMPLE TIME MEASURE-
 MENTS FOR EMSL-LV, AVOAS ON-LINE, AND AVOAS OFF-LINE (A.M. AND P.M.) ANALYSES OF TREATMENT
 TRAIN SAMPLES ACROSS THE SIX DAYS OF THE DEMONSTRATION
Compound
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroelhane
cis-1,2-dichloroethene
1,1,1-trichloroelhane
trichloroethene
tetrachloroethene
Compound
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
trichloroethene
tetrachloroethene

Mean
EMSL
LV
a
U
U
11.9
25.8
52.8
2002

Mean
EMSL
LV
6.2
U
U
10.6e
25.9
48.1
2093


Concentration, ug/L
On
6.6
U"
3.3
16.4
30.3
62.3
2269

OH
4.3
U"
3.2
15.7
33.0
61.9
2256

Concentration, ng/L
On
7.2
1.8
4.0
18.6
37.0
71.5
2453
Off
6.5
1.4
3.7
17.3
36.2
67.6
2407
A.M.-
Collected
Samples

Standard Deviation
EMSL-
LV
0.8
-
-
3.6
7.1
11.5
148
P.M.-
On
2.6
b
0.6
4.5
9.6
19.7
408
Collected
Off
1.7
b
0.6
4.4
8.5
18.1
348
Samples
EMSL-
LV
10.5
-
—
30.4
27.5
21.7
7.4

Standard Deviation
EMSL-
LV
0.3
—
—
1.9C
6.9
5.2
150
On
2.4
0.40
0.5
2.0
7.5
7.2
240
Off
1.4
0.33
0.3
2.1
2.3
4.0
211
EMSL-
LV
4.8
—
-
18.3C
26.5
10.9
7.2

%RSD
On
39.0
b
17.4
27.3
31.7
31.6
18.0

%RSD
On
34.0
22.5
13.8
10.9
20.4
10.0
9.8


Off
36.7
b
19.7
27.7
25.8
29.3
15.4


Off
21.7
23.3
9.1
12.2
6.3
5.9
8.8
U: Below instrument detection limits.
1 Six ol 12 values were 'U'; no computations.
b Trans-12-dichloroethene was detected in p.m. samples only.
c Based on n=5, because one value was 'U.'

5% overall). The ANOVA results support the find-
ing of no significant difference due to on-line or off-
line modes, except for the high volatility, low con-
centration analytes 1,1-DCE and trans- 1,2-DCE.
     AVOAS and EMSL-LV Comparisons —  The
very high concentrations of PCE required sample
dilution to bring PCE concentrations to within the
linear range of the  EMSL-LV detector.  Sample
dilution reduced the quantity of the lower concen-
tration  analytes (1,1-DCE,  trans- 1,2-DCE,  1,1-
DCA,  cis-1,2-DCE, and 1,2-DCA) to below the
MDLs of the EMSL-LV instrument.  Furthermore,
the MDLs for the EMSL-LV instrument were some-
what higher than those  for the AVOAS; i.e., the
AVOAS  instrument had greater sensitivity than the
EMSL-LV instrument.  For  these reasons, the
AVOAS results indicate detectable concentrations
of analytes for a number of samples for which con-
centrations were below the  MDLs of the EMSL-
LV detector.
     For analytes detected  by both instruments,
virtually all of the AVOAS concentrations for both
the on-line and off-line modes were higher than the
EMSL-LV values, and all differences in mean %R,
with the exception of 1,1-DCE and PCE for port S-
3, were statistically significant at the 10% level.
For the on-line AVOAS and EMSL-LV compari-
sons, the difference  for TCE was approximately
21% for samples from both ports,  and  for PCE,
approximately 12% for samples from both ports.
The ANOVA results reinforce the finding of a sig-
nificant effect due to analytical system. Figure 3-1
                                            26

-------
 shows EMSL-LV and AVOAS on-line and off-line    of all compounds measured for port S-3 samples,
 TT sample results, averaged over port and time of    except for 1,1-DCE, were greater than those for
 day-                                                  port S-4 samples. The ANOVA results indicate that
      Sample Port Differences — For all AVOAS    ** differences were statistically significant for 1,1-
 on-line and off-line measurements, concentrations    DG\, cis-l,2-DCE,  1,1,1-TCA, and TCE.  It is
                                                       possible that these differences reflect actual differ-


 TABLE 3-3. DIFFERENCE IN MEAN PERCENT RECOVERIES FOR TREATMENT TRAIN SAMPLES ANALYZED
 BY THE AVOAS AND EMSL-LV.
On-line - Off-line AVOAS
Compound
1,1-dichloroethene*
trans- 1 ,2-dichloroethene*
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
trichloroettiene
tetrachloroethene
Ports
19S
37 NS
4M
2NS
•6S
2NS
1 NS
Port 4
25 NS
26 NS
6NS
12 M
1NS
5NS
INS
Off-lint -
Port3
•3NS
c
c
31 S
29 S
23 S
11 M
EVSL-LV
Port 4
27 S
c
c
45 S
22 S
21 S
12 S
On-lint -
Ports
16 NS
c
c
34 S
23 S
21 S
12 M
EMSL-LV
Port 4
52 S
c .
c
57 S
23 S
26 S
13S
Vinyl chloride and 1,2-dichloroethane were not detected by either EMSL-LV or the AVOAS, so those compounds are not included in this table.
1 For EMSL-LV, two of six samples for Port S-3 and four of six samples for Port S-4 had concentrations below Instrument detection limits.
6 All morning AVOAS measurements had concentrations below instrument detection limits.
e Compound not detected by EMSL-LV Instrument; no computation possible.
NS:   Not significant (p> 0.10).
M:    Moderately significant (0.05 0.10).
M:    Moderately significant (0.05 


-------
ences in water composition for samples from each
port.  However, for the EMSL-LV measurements,
the significance of port among analytes was vari-
able.  The data are confounded by higher concen-
trations detected by the AVOAS in some afternoon
collected/analyzed samples compared to morning
samples.
                                  Time Differences — No  significant  differ-
                             ences for time of sample collection were detected
                             for any compound measured  by the  EMSL-LV
                             instrument; however, for the AVOAS, time  of col-
                             lection/analysis effects were  significant or  highly
                             significant for every compound.  The results for the
                             AVOAS afternoon analyses  were always  greater
      Treatment Train Sample Results for 1,1-Dichloroethene

        20-1
        -5
        10-
         5-
                               Treatment Train Sample Results (or 1,1-Dlchloroethane
                             I
                                                         10-1
                                                          8-
                                                          4-
                                                          2-
                        3  DAY *
                . AVOAS ON4JNC
             5      6

        AVOAS OFf-UNE
                                               3  DAY *
                                                                 • AVOAS ON4JNE
                                                                                • AVOAS OFF-LINE
     Treatment Train Sample Results for cls-1,2-Dlchloroethene    Treatment Train Sample Results for 1,1,1 -Trlchloroethane
        30

        25

        20 •

        15

        10

        5

        0
                                 SO-i


                                 40-


                                 30-


                                 20-


                                 10-
           1       2

          •- AVOAS ON4JN8
3  DAY*
            5      8

AVOAS OFF-LW8   -M- EMSL-LV
        Treatment Train Sample Results for THchloroethene
       100.


        80.


        60.


        40


        20
                                                    DAY
                                  -m- AVOAS OfHJM  -e- AVOAS OfF-UHE   -M-EMSL-LV



                                Treatment Train Sample Results for Tetrachloroethene

                                3500.
                                3000
                             3


                             1 z*00
                                                      8
                                2000
                                                        1500
                  2      3  DAY 4      5
          »- AVOAS ON4JNI  -«- AVOA3 OFF-UNI   -M- EMSL-LV
                                    1       2

                                   I- AVOAS ON4JNI
                                               3  DAY *
           5      •

AVOAS OFHJM   -M- EMSL-LV
Rgure 3-1. Treatment Train Sample Results. (Over all ports and times.)

                                                  28

-------
than those for the morning analyses. For example,
none of the morning AVOAS measurements con-
tained detectable concentrations of trans-1,2-DCE,
whereas all afternoon samples did.  This finding is
consistent with carryover from higher concentra-
tion samples analyzed between the morning and
afternoon analyses.  The superhigh-concentration
SPEW samples were analyzed between the mom-
ing and afternoon TT analyses using the same loop
as that used for the TT samples. The TT samples
contained much lower concentrations of all ana-
lytes (except TCE and PCE) than the superhigh-
concentration SPEW samples.
     Day-to-Day Differences — For the EMSL-
LV data, no significant differences in results from
day to  day  were observed.  For the AVOAS, day-
to-day  differences were statistically significant for
all compounds but one (trans-1,2-DCE). Although
the differences are significant, the magnitude is not
great  Most values for TCE and PCE were within
± 30%  of the 6-day mean concentrations.  Possible
explanations for this finding could include:   1)
actual changes in treatment train water concentra-
tions over the demonstration period;  2) factors
intrinsic  to  the AVOAS, such  as  variations  in
instrument function over the demonstration period;
or 3) random variability.  The fact that the EMSL-
LV measurements show no significant day-to-day
difference suggests that variation in treatment train
water over the 6-day period is not an explanation.

Precision for Treatment Train Samples
     Percent RSD values for both the AVOAS and
EMSL-LV were less than 25%, except for 1,1-DCE
for  the AVOAS, which was 35% for the on-line
mode and 33% for off-line.  Precision results for
the  AVOAS in the  off-Ike mode tended to be
slightly better  than  those for  the  on-line  mode,
except for 1,1,1-TCA.
     For EMSL-LV, the %RSDs for TCE and PCE
were 18% and 7% for all data over the 6 days. For
the  AVOAS the corresponding %RSDs for TCE
and PCE were 20% and 12% for the off-line mode
and 22% and 14% for the on-line mode.
     For all compounds  except PCE, the %RSD
values for port S-4 were the same or greater than
those of port S-3.

Comparison of the AVOAS and Method
502.2 Results for Spiked Purged Effluent
Water Samples
     Tables 3-5(a-d)  are  compilations  of  the
SPEW  sample results of the AVOAS and EMSL-
LV  analyses for each of the nine compounds at
each of the four concentration ranges.  Table 3-6
summarizes data for differences in mean %Rs. The
results of paired-sample t-tests of the hypothesis of
no significant difference in mean %Rs between the
AVOAS and EMSL-LV are included in that table.
Plots of mean %Rs for both methods are shown in
Figure 3-2.
       Mean %Recoveries Averaged Ovw Compound*
        AVOAS SPIKED PURGED EFFLUENT WATER SAMPLES
     140
     130
     120
     110
     100
      90
      80
      70
      60
      SO
          LOW
                       DAY

                       -e-MOM
                                -8UPHWMH
       Mean %necovsns9j Averaged Over Compounds
        EMSU.V SPIKED PURGED EFFLUENT WATER SAMPLES
     140
     130

     120
     110
     100
      90
      80
      70

      60
      SO
         1     2

        ••LOW  -«.
3  DAY*
 S      8

-M-SUPMMH
       MMH %Racov*riM Averaged Over Compounds
                   and Levels
          SPWED PURGED EFFLUENT WATER SAMPLES
     140
     130
     120
     110

     100
      90

      80
      70
      80
      SO
                    3  DAY4
                 .AVOAS
Figure 3-2. Spiked Purged Effluent Water Results.
                                             29

-------
     In every case, the mean %R was greater for    the spiked target values; i.e., the %Rs were less
the AVOAS than for the corresponding EMSL-LV    than 100%, whereas the AVOAS results were usu-
analysis. All but three of the differences were sig-    ally close to or above the spiked values. The over-
nificant (p<0.05).  For most analytes, the EMSL-    all %R averaged over all compounds, days, and
LV measured concentrations were near or below    concentration levels for EMSL-LV was 74%; that
TABLE 3-5
-------
for the AVOAS was 104%. Possible explanations
for these findings include:  1)  differences in per-
formance of the instruments, 2)  variability in ana-
lyst technique, or 3)   transport or holding time
effects.   The data indicate that a combination of
these sources may have contributed to the observed
results.
     The day-to-day patterns for the AVOAS and
EMSL-LV are similar, though hot identical.  Some
exceptions can be noted. For example, the AVOAS
results for the medium-concentration SPEW sam-
ple for day 4 had markedly lower concentrations of
all analytes compared to the other days  and to
those for the EMSL-LV instrument; for day 3, the
TABLE 3-5(b). MEDIUM-CONCENTRATION SPIKED PURGED EFFLUENT WATER SAMPLE RESULTS FOR
EMSL-LV AND AVOAS ANALYSES
EMSL-LV
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene
AVOAS
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene
Detected Concentration,
Target
Cone.
(Mfl)
37.4
91.1
97.5
45.6
23.7
91.1
39.2
53.8
88.7
1
38.9
67.0
76.0
39.4
20.5
74.0
33.1
40.5
75.0
2
29.5
55.0
66.0
38.2
19.2
67.0
33.6
40.6
66.0
3
26.2
53.0
62.0
36.0
17.9
62.0
32.2
38.2
63.0
4
28.0
48.9
57.0
32.9
16.2
58.0
28.9
34.6
55.0
Detected Concentration,
Target
Cone.
(ug/i)
37.4
91.1
97.5
45.6
23.7
91.1
39.2
53.8
88.7
1
60.2
75.0
98.1
55.8
25.5
110
45.5
56.5
88.6
2
56.6
72.0
99.6
53.6
25.1
108
42.4
57.2
89.8
3
51.8
67.8
90.9
52.2
23.2
104
40.0
51.1
82.3
4
34.7
43.2
60.5
34.4
15.1
62.9
28.0
34.1
52.2
ng/L, by Day
5
33.9
58.0
61.0
34.5
16.7
62.0
28.1
35.3
56.0
6
27.0
44.3
52.0
29.4
14.2
60.0
26.9
30.0
46.8
Mean
Cone.
(Mrl)
30.6
54.4
62.3
35.1
17.5
63.8
30.5
36.5
60.3
ng/L, by Day
5
54.5
71.1
71.1
53.3
23.5
97.4
43.6
51.9
80.2
6
58.0
70.1
97.2
55.0
24.0
112
42.6
55.6
91.3
Mean
Cone.
(ng/L)
52.6
66.5
86.2
50.7
22.7
98.9
40.4
51.1
80.7
Summary
Standard
Deviation
4.9
7.8
8.2
3.6
2.2
5.8
2.8
4.1
9.8
%RSD
16.0
14.4
13.2
10.4
12.9
9.1
9.3
11.2
16.3
Mean
%R
82
60
64
77
74
70
78
68
68
Summary
Standard
Deviation
9.2
11.7
16.4
8.1
3.8
18.4
6.3
8.7
14.6
%RSD
17.6
17.6
19.1
16.0
16.8
18.6
15.6
17.0
18.1
Mean
%R
141
73
88
111
96
109
103
95
91
Target Cone.: Concentration of spiking solution.
Mean Cone.: Mean concentration over 6 days.
The standard deviation listed is for the measured concentrations (not for the %Rs).
%flSO:     Percent relative standard deviation.
Mean %R:   Mean percent recovery.
                                                31

-------
EMSL-LV  analyses  of  the  low-concentration
SPEW samples had markedly lower concentrations
for all components than on the other days.
     Precision for the AVOAS and for the EMSL-
LV instruments for the combined values across all
concentrations and compounds was similar (12.8
and  13.0%).   Results for all four concentration
ranges were <30 %RSD for both the AVOAS and
the EMSL-LV data,  with the exception of 1,1,1-
TCA in the low-concentration samples. The higher
(31 to 32%) RSD for the 1,1,1-TCA replicates was
TABLE 3-5
-------
a consequence of an exceptionally high concentra-
tion of this compound in the sample analyzed at
both locations on the last day of the demonstration.
It  is possible that 1,1,1-TCA or a coeluting con-
taminant (i.e., a compound with a chromatographic
retention time  close  to  that  of 1,1,1-TCA)  was
introduced into the  samples  at  the site  during
SPEW sample preparation.  The average precision
for the AVOAS across all concentrations and com-
pounds was 12.6 %RSD; that for EMSL-LV was
13.3%. This difference is so small as to be of no
practical significance.
TABLE 3-5(6). SUPERHIGH-CONCENTRATION SPIKED PURGED EFFLUENT WATER SAMPLE RESULTS FOR
EMSL-LV AND AVOAS ANALYSES
EMSL-LV
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene
AVOAS
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene
Detected Concentration,
Target
Cone.
(1*9/1.)
747
—
1946
911
467
1822
788
1067
1774
1
580
—
1330
720
345
1280
650
810
1460
2
510
—
1240
680
305
1270
580
730
1240
3
580
—
1170
620
286
1310
530
690
1390
4
450
-
970
540
261
1080
493
590
1020
Detected Concentration,
Target
Cone.
(M9/L)
747
—
1946
911
467
1822
788
1067
1774
1
1211
—
2024
1109
547
2368
899
1277
2227
2
1038
—
1954
969
489
1861
784
1220
2129
3
1131
—
1823
949
456
1946
788
1127
1929
4
930
—
1754
881
430
1667
687
1071
1958
jig/L, by Day
5
730
—
1340
700
306
1380
570
750
.1310
6
630
—
1210
720
329
1280
550
720
1200
Mean
Cone.
(H9/L)
580
-
1210
663
305
1267
562
715
1270
ug/L, by Day
5
1218
—
1897
945
463
1973
747
1121
1935
6
1415
—
2114
1111
535
2494
955
1324
2450
Hem
Cone.
(M/L)
1157
—
1928
994
487
2053
810
1190
2150
Summary
Standard
Deviation
96.7
-
135
70.9
29.9
99.9
53
73.1
155
%RSD
16.7
-
11.2
10.7
9.8
7.9
9.4
10.2
12.2
Mean
%R
78
-
62
73
65
70
71
67
72
Summary
Standard
Deviation
167
—
132
94.6
46.3
313
99.1
99.2
208
%RSD
14.4
—
6.8
9.5
9.5
15.2
12.2
8.3
9.9
Mean
%fl
1S5
-
99
109
' 104
112
103
112
119
Target Cone.: Concentration of spiking solution.
Mean Cone.: Mean concentration over 6 days.
The standard deviation listed is for the measured concentrations (not for the %Rs).
%RSD:     Percent relative standard deviation.
Mean %R:   Mean percent recovery.
                                                 33

-------
TABLE 3-6. DIFFERENCE IN MEAN PERCENT RECOVERIES FOR SPIKED PURGED EFFLUENT WATER
SAMPLES ANALYZED BY THE AVOAS (OFF-LINE) AND EMSL-LV

vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroettiene
tetrachloroethene
Low Spike
58
6NS
20
31
25
41
23
23
16
Medium Spike
59
13
25
34
22
39
25
27
23
High Spike
40
13 M
18 M
25
20
22
17
23
21
Superhigh
Spike
77
b
37
36
39
43
31
45
47
Mem, All
59
11
25
32
26
36
24
29
27
" All values significantly different, except as noted (p < 0.05).
* /, 1-dichlonethene not in superhigh-concentration sample.
NS:   Not significantly different (p > 0.10).
U:    Moderately significant (O.OS 
-------
              Transport Study - Medium Spited Samples
                         TETRACHLOROETHENE
                                  Transport Study - High Spited Samples
                                          TETRACHLOROETHENE
           1*0

           140

           120

           100

            SO

            CO

            40
                            160

                            140

                            120

                            100

                             80

                             80

                             40
                                 1
                                DAY
                    TRANSPORTED
     4        5


. NOT TRANSPORTED
 3
OAV
                                                                         > TRANSPORTED
     4        5


• MOT TRANSPORTED
Figure 3-3b. Transport Study Results for Medium- and High-concentration Spiked Samples for
Tetrachloroethene.
TABLE 3-7(a). TRANSPORT STUDY RESULTS FOR MEDIUM SPIKED SAMPLES*
Detected Concentration, |ig/L,
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene
Target
Cone.
(M9/L)
37.4
91.1
97.5
45.6
23.7
91.1
39.2
53.8
88.7

Trim
36.5
36.6
61.0
61.0
68.0
68.0
39.2
38.4
19.4
18.8
66.0
65.0
33.5
33.3
40.9
39.7
65.0
57.0
1
Not
Trim'
35.5
53.0
60.0
33.8
16.2
60.0
31.7
34.1
54.0
2 3
Trim
47.9
38.3
74.0
61.0
81.0
68.0
46.4
38.5
22.2
18.8
80.0
64.0
36.8
32.8
46.9
38.5
76.0
63.0
Not
Trim
36.3
51.0
70.0
75.0
78.0
80.0
41.6
43.1
21.9
20.8
48.4
70.0
35.8
36.0
42.3
46.8
67.0
79.0
Trim
35.8
54.0
60.0
99.0
39.4
50.0
26.1
30.5
17.3
20.8
34.3
38.7
25.2
27.8
27.8
33.3
37.8
45.9
Not
Trim
55.0
44.2
82.0
66.0
97.0
70.0
47.3
38.6
24.9
18.9
56.0
65.0
37.6
31.6
50.0
39.7
86.0
63.0
by Simple
4
Trim
103
86.1
97.6
98.6
94.1
95.4
53.3
52.2
25.5
29.5
89.7
101
41.9
47.7
56.0
54.7
81.9
87.2
Not
Trim
84.1
92.5
93.5
102
91.9
105
51.6
56.2
26.1
26.3
95.8
89.6
44.9
42.5
59.4
56.1
87.4
84.7
5
Trim
80.6
81.7
92.3
94.1
98.2
91.6
59.4
51.8
28.6
24.8
103
87.7
46.1
23.8
62.5
51.9
80.3
80.9
Not
Tram
93
127
102
130
99.5
128
56.8
78.4
27.1
36.3
99.6
127
44.6
61.5
57.7
77.4
86.9
110
" Results for duplicate samples for each treatment.
6 Duplicate sample not analyzed.
Trans:    sample transported.
Not trans: sample not transported.
                                                       35

-------
 TABLE 3-7(b).  SUMMARY OF TRANSPORT STUDY RESULTS FOR MEDIUM SPIKED SAMPLES
Mean Concentration
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene
Tram
59.8
79.9
75.4
43.6
22.6
72.9
34.9
45.2
67.5
Not Tram
68.7
85.9
89.9
49.7
24.3
79.0
40.7
51.5
79.8
Standard
Deviation
Trans
25.3
17.9
20.2
10.8
4.3
23.8
8.3
11.1
16.7
Not Tram
31.7
23.4
20.5
13.3
5.8
25.6
9.3
13.0
16.6
%RSD
Tram
42.1
22.4
26.8
24.7
19.1
32.6
23.8
24.4
24.7
Not Tram
46.1
27.2
22.8
26.7
24.0
32.3
22.8
25.2
20.8
Mean %R
Tram
160.6
87.7
77.3
95.6
96.4
80.1
89.0
84.0
76.1
Not Tram
183.8
94.3
92.2
109.0
103.8
86.8
103.8
96.6
89.9
Difference
in
Mean %R
23.2
6.7
14.9
13.4
7.2
6.7
14.8
11.7
13.9
Mean Concentration: Mean concentration tor duplicate samples lor live sample pairs.
%RSD:            Percent relative standard deviation.
Mean %R:          Mean percent recovery for duplicate samples (or live sample pairs.
Difference in mean %R: Difference in mean percent recovery between samples not transported and those transported.
     Several instances for which transport effects
may have influenced the results can be observed.
The  estimated loss in TCE concentration for the
medium-range transported sample was approxi-
mately 12%. However, the 90% confidence inter-
val expands the potential loss  to  approximately
28%, which is consistent with the  results for the
AVOAS and EMSL-LV differences in mean %R
for SPEW samples.
     Most of the %Rs for both transported and
nontransported  samples,  at  both concentration
ranges, were less than 100%. It is impossible to
attribute with certainty a cause for these results;
however,  explanations could include:  1) analyte
concentrations in the standards were less than the
certified amounts, and 2) chemist spiking or ana-
lytical technique. For the SPEW samples, EMSL-
LV %Rs also tended to be <100%. However, the
%Rs of the SPEW samples  analyzed  by the
AVOAS were close to 100%. The transport study
samples were made using the same spiking solu-
tions as those for the SPEW  samples. In contrast,
the QCCS samples,  which were prepared at the
respective locations,  had %Rs close to 100% for
both the AVOAS and EMSL-LV. The stock solu-
tion used to make the QCCSs was not the same as
that used for the SPEW samples.
     The results suggest that inter-analyst variabil-
ity was significant; that is, the difference in recov-
eries obtained by the two chemists was larger in
magnitude than any transport effect  This inter-
analyst variability occurred despite the fact that
each analyst had recalibrated the instrument before
their respective days on the study.  The analyst
variability seems to differ for the medium- and
high-concentration spikes.  One explanation may
be the difference  in dilution techniques used by the
analysts for the high-concentration samples.  How-
ever, wide confidence  intervals for each of the
spiking concentrations preclude definitive conclu-
sions. The analyst-to-analyst difference is similar
in magnitude to  the AVOAS - EMSL-LV differ-
ences found in the TT and SPEW sample compari-
sons, leading one to speculate whether the real
differences between the AVOAS  and  EMSL-LV
mean %Rs are due to transport phenomena, differ-
ences between instruments,  differences between
analysts, or a combination of these or other factors.

Linearity Study Results
     Data for off-line analyses  of the SPEW sam-
ples were used to check the linearity of response of
the AVOAS. The results suggest little evidence for
nonlinearity for most of the target analytes at most
                                              36

-------
TABLE 3-8(a). TRANSPORT STUDY RESULTS FOR
HIGH SPIKED SAMPLES
Detected Concentration, ng/L, by Sample
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene

1
Target Not
Cone. Trant Tram
227
45.6
138
91.1
281
91.1
90.2
213
237
282 255
252 295
46.9 43.2
42.7 48.0
116 121
107 122
76.0 81.0
71.0 79.0
235 237
220 274
74.0 90.0
74.0 95.0
75.0 77.0
72.0 88.0
181 187
162 213
197 203
195 233

Trant
265
234
47.8
51.0
121
118
76.0
70.0
235
240
70.0
53.0
75.0
78.0
182
184
213
199
2
Not
Trant
301
249
54.0
53.0
128
124
78.0
78.0
250
260
69.0
60.0
82.0
86.0
195
200
207
200

Tram
331
358
56.0
55.0
121
124
76.0
78.0
214
215
58.0
65.0
75.0
72.0
193
197
183
188
3
Not
Trant
355
311
59.0
50.0
137
120
85.0
77.0
258
235
77.0
71.0
84.0
75.0
201
178
213
195
4



5
Not Not
Tram Tram Trant Trant
341
303
44.7
41.6
96.3
89.9
61.4
58.1
136
168
57.2
50.7
60.2
53.8
157
131
177
144
301
328
41.6
49.9
82.7
99.6
53.7
63.0
157
182
44.0
59.7
51.4
58.3
121
143
120
159









350
323
45.2
45.1
103
97.5
66.9
61.9
192
180
63.3
56.9
59.6
58.4
178
143
189
156
322
288
42.6
37.8
93.2
78.7
60.9
50.6
176
150
57.1
47.7
58.9
59.8
138
113
145
123
a Results for duplicate samples lor each treatment.
b Duplicate sample not analyzed.
Trans: sample transported.
Not trans: sample not transported.
TABLE 3-8(b). SUMMARY OF TRANSPORT STUDY RESULTS FOR HIGH SPIKED SAMPLES
Mean
Concentration
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1 ,2-dichloroethane
trichloroethene
tetrachloroethene
Trant
303.9
47.6
109.4
69.5
203.5
62.2
67.9
170.8
184.1
Not Trtnt
300.5
47.9
110.6
70.6
217.9
67.0
72.0
168.7
179.8
Standard Deviation
Tram Not Tram
43.6
4.9
12.2
7.2
33.9
8.4
8.8
21.8
20.6
32.0
6.6
20.3
12.4
46.6
16.8
13.6
36.6
39.9
Tram
14.3
10.3
11.2
10.3
16.7
13.5
13.0
12.7
11.2
%RSD
Not Tram
10.6
13.7
18.4
17.5
21.4
25.1
18.9
21.7
22.2
Mean
Trant
135.6
104.4
79.3
76.3
72.4
68.3
75.3
80.2
77.7
*R
Not Tram

II
132.4
105.1







80.2
77.5
77.5
73.5
79.9
79.2
75.9







Dltt. In
lean %R
•1.5
0.7
0.9
1.2
5.1
5.2
4.6
•1.0
•1.8
Mean Concentration:    Mean concentration for duplicate samples for five sample pairs.
%RSD:                 Prcent relative standard deviation.
Mean %ft:              Mean percent recovery for duplicate samples for five sample pain.
Difference in mean %R: Difference in mean percent recovery between samples not transported and those transported.

                                                              37

-------
concentrations; however, responses for TCE and
PCE show some nonlinearity in the superhigh-con-
centration range. The elevated concentrations of
those compounds (in the TT, as  well as in the
superhigh-concentration SPEW samples) required
use of the smallest sampling loop.  One explana-
tion for this finding  could be carryover of those
compounds in that loop.

Quality Assurance and Quality Control
Results

Results for QC Samples

QCCS Results: Percent Recoveries
     Measured concentrations for the QCCS sam-
ples are shown in Tables 3-9(a,b).  Percent recov-
ery results  are presented in Tables  3-10(a,b).
Figure 3-4 shows QCCS %R results for each com-
pound for both the AVOAS and EMSL-LV.
    Method 5022 — With the exception of 1,1-
DCE, all but two (of 30) data points for the EMSL-
LV QCCS samples were within the DQO of 100
±30 %R.  The theoretical spike value for 1,1-DCE
was 2.0 ug/L; five of the  six %Rs for 1,1-DCE
were between 175 and 195%, or 3 to 4 ug/L. The
impact of the 1,1-DCE results in this demonstra-
tion was minimal. Two other results exceeded the
DQO: one for VC and one for 1,1,1-TCA.
    AVOAS — AVOAS results for VC were more
variable than those of EMSL-LV, with %Rs rang-
ing from 64-170%. Four of the six recoveries for
VC exceeded ±30 %R of the theoretical spiked
concentration. In contrast to the EMSL-LV results,
only one 1,1-DCE result exceeded ±30% of the tar-
get value. TCE, PCE, and 1,1,1-TCA each had one
value that exceeded ±30% of the target value.
    For both EMSL-LV and the AVOAS, all com-
pounds (except  1,1,1-TCA for  EMSL-LV) had
their highest %R on the first day of analysis. For
TABLE 3-9(8). QUALITY CONTROL CHECK STANDARD RESULTS FOR EMSL-LV
Detected Concentration, jig/L, by Day
Compound
vinyl chloride
1,1-dichloroethene
1,1,1-trichloroethane
trichloroethene
tetrachloroethene
Target Cone.
(ng/L)
20.0
2.0
10.0
2.0
15.0
1
27.2
3.9
10.0
1.9
16.7
2
22.6
3.6
7:9
1.6
13.7
3
23.9
3.9
9.0
1.8
15.1
4
25.0
3.6
10.1
1.8
14.8
5
25.3
3.5
8.4
1.6
13.7
6
21.4
2.4
6.5
1.8
14.0
Vean Cone.
(H9/L)
24.2
3.5
8.7
1.8
14.7
Summary
Standard
Deviation
2.1
0.6
1.4
0.1
1.2

%RSD
8.5
16.0
15.8
7.0
7.9

TABLE 3-9(b). QUALITY CONTROL CHECK STANDARD RESULTS FOR THE AVOAS
Datictid Concantratlon, (ig/L, by Day
Compound
vinyl chloride
1,1-dichloroethene
1,1,1-trichloroethane
trichloroethene
tetrachloroethene
Targat Cone.
(M9/M
20.0
2.0
10.0
2.0
15.0
1
33.9
2.3
15.2
5.4
18.3
2
29.8
1.7
11.4
2.1
13.9
3
12.8
1.1
8.0
1.4
9.6
4
15.9
1.4
7.2
1.4
10.9
5
25.6
1.7
9.2
1.9
11.9
6
27.7
1.9
10.8
2.3
14.1
Mean Cone.
(H9/M
24.3
1.7
10.3
2.4
13.1
Summary
Standard
Deviation
8.2
0.4
2.9
1.5
3.1

%RSD
33.8
24.4
28.1
61.9
23.3
%RSD:     Percent relative standard deviation
Mean % R:  Mean percent recovery.
                                            38

-------
the following days, results for 1,1,1-TCA, TCE,
and PCE tended to be near or less than 100% R.
The cause for either of these observations is uncer-
tain.

QCCS Results:  Precision

     Method 502.2 — Percent RSDs for  EMSL-
LV analyses ranged from 7 to  16% among com-
pounds.

     AVOAS — For the AVOAS QCCS samples,
although the mean %Rs for each compound across
days were good (84 to 122%), the %RSDs tended
to be high, ranging from 23 to 62%.  The 62 %RSD
for TCE is a result of an inexplicably high %R of
approximately 270% on the first day of analysis.
     The variability for the EMSL-LV QCCS sam-
ples  was  notably less than  that for the AVOAS,
probably  because the  QCCS used  by EMSL-LV
was prepared from a single ampoule spiked into
reagent water, and the  same solution was used for
the entire demonstration; for the QCCS used for
the AVOAS. A new ampoule was opened daily and
a fresh standard  was prepared.  As one would
expect, this procedure led to less variability in the
EMSL-LV results than those of the AVOAS for the
QCCS analyses. Variabilities in spiking technique
and in analyte concentrations in the commercially
prepared spiking  solutions  would be expected to
contribute  to the lower  precision values for the
AVOAS.

Surrogate Recoveries

     A surrogate  compound was not used for the
AVOAS analyses. Surrogate (2-bromo-l-chloro-
propane) recoveries ranged from 85 to 116%, all
within  the  80 to 120% acceptance limits for
Method 502.2.

Instrument Blank Results

     Method 5022 — Instrument blanks showed
no analytes present at concentrations greater than
the MDLs.

     AVOAS — The blank samples run on the
AVOAS showed small amounts of one or more of
TABLE 3-10(a).
PERCENT RECOVERIES FOR QUALITY CONTROL CHECK STANDARDS
Percent Recovery, by
Compound
vinyl chloride
1,1-dichloroetherte
1,1,1-trichloroethane
trichloroethene
tetrachloroethene
1
136
195
100
95
111
2
113
180
79.0
80.0
91.3
3
120
195
90.0
90.0
101
4
125
180
101
90.0
98.7
Oiy
5
127
175
84.0
80.0
91.3

6
107
120
65.0
90.0
93.3

Hem
%R
121
174
86.5
87.5
97.8
FOR EMSL-
Summtry
Standard
Deviation
10.4
27.8
13.6
6.1
7.6
LV

%RSD
8.6
16.0
15.8
7.0
7.8

TABLE 3-1 0(b).
PERCENT RECOVERIES FOR QUALITY CONTROL CHECK STANDARDS
Percent Recovery, by
Compound
vinyl chloride
1,1-dichloroethene
1,1,1-trichloroethane
trichloroethene
tetrachloroethene
1
170
115
152
270
122
2
149
85.0
114
105
92.7
3
64.0
55.0
80.0
70.0
64.0
4
79.5
70.0
72.0
70.0
72.7
Diy
5
128
85.0
92.0
95.0
79.3

6
139
95.0
108
115
94.0

Mem
%R
122
84.2
103
121
87.5
FOR THE AVOAS
Summary
Standard
Deviation
41.3
20.6
28.8
75.3
20.5

%RSD
33.9
24.5
28.0
62.3
23.4
                                           39

-------
the target analytes; however, with the exception of
PCE, all were at concentrations less than the MDL
or PQL of the instrument. The PQL for the  10-mL
sampling loop used for blanks was 0.10 (ig/L for
PCE; PCE  was present  in all blanks at approxi-
mately 0.3 to 1.2 \ig/L. However, because the low-
est sample concentration of PCE in this study was
15 |lg/L (in the QCCS), no data were invalidated
by  the  presence of  the  small amounts  of PCE
detected in the blanks.

Travel Blank Results
     Travel blank  samples were not run for the
AVOAS, because  samples  were   not  shipped.
             Quality Control Check Standard Results
                       VINYL CHLORIDE
          Quality Control Check Standard Results
                   1,1-OICHLOROETHENE
        200-]
        150-
        100-
         50
                                                          200<
                                                          140-
                                                          100-
                                                           50
                  2      J  DAY  *
                       AVOAS
             Quality Control Check Standard Results
                    1,1,1-TOCHLOROETHANE
                                                                             DAY
                                                                         AVOAS    -«. EMSL-LV
          Quality Control Check Standard Results
                   TRtCHLOROETHENB
        200-|
        150
        100-
        50
                                                          200-1
                                                          150-
                                                          100-
                                                          50 -*T
                         3  DAY *

                    -•- AVOAS    -*- EMSL-U
                2      3  DAY  *
                 -•- AVOAS   -e- EKSL-CV
                                       Quality Control Check Standard Results
                                               TETRACHLOROETHENE
                                  200-1
                                  150-
                                  100-
                                   50 JT
                                            2      3  DAY  *

                                              HH AVOAS   -O- BWH.V
Figure 3-4. Quality Control Check Standard Results.
                                                  40

-------
Travel blanks analyzed by EMSL-LV during the    Continuing Calibration Check Standard
demonstration  contained  concentrations  of PCE    Results
below the MDL;  1,1,1-TCA was detected in one         „	
                                                                         aeck Standard results
no demonstration data were impacted by contami-
nation during transport.
TABLE 3-11(8). PERCENT DIFFERENCES
EMSL-LV
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene
for EMSL-LV and the AVOAS are shown in Tables
3-ll(a,b).
FOR CONTINUING CALIBRATION CHECK STANDARDS

1
6.4
•5.6
•7.9
-8.2
•9.6
•14.8
•8.3
•10.6
•14.4

2
0.7
2.3
2.6
1.0
2.7
•7.1
2.0
1.1
3.2
Percent Difference,
3
5.8
6.4
4.4
4.8
4.9
•5.6
4.3
1.1
3.2
byOiy
4
6.8
5.4
2.7
2.4
7.2
•5.3
6.8
4.0
4.3

5
3.9
4.5
1.4
1.3
1.3
•10.0
0.6
•0.7
•0.2
FOR

6
0.85
0.30
•7.1
•6.2
•2.6
•12.7
-2.2
•3.8
•6.3

TABLE 3-11(b). PERCENT DIFFERENCES
AVOAS
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene
FOR CONTINUING CALIBRATION CHECK STANDARDS

1
•8.3
2.3
•6.4
•8.6
-7.0
-20.1
•3.6
•10.2
•7.5

2
28.4
16.3
-0.18
-5.6
•3.3
•13.6
1.7
•5.9
•4.0
Percent Difference,
3
•10.7
9.9
•9.3
•9.1
•6.2
•22.9
1.9
•10.8
•8.3
byDiy
4
35.5
21.6
3.1
•4.0
•2.0
•7.4
•1.5
•6.2
•5.8

5
30.5
24.9
•1.2
•7.8
•5.5
•18.8
•2.9
•9.9
•6.8
FOR THE

6
24.9
21.2
•6.8
•13.2
•11.8
•30.1
•9.5
•16.4
•18.6
Percent Difference * C,-C2 • 100/C,; C, = response for initial calibration; C2 a response for continuing calibration.
 EMSL-LV data are the mean of two measurements.
Target concentrations for all analytes lor EMSL-LV were 20 fig/L, and for the AVOAS, 400 (ig/L.
                                                   41

-------
     Method 502.2 — All %Ds for the EMSL-LV
CCCSs were within the acceptance criteria of <20
%D.  The EMSL-LV instrument was recalibrated
once, on the second day of the demonstration
because %Ds of nearly  15% were noted for two
compounds on the first day.  All EMSL-LV sam-
ples were analyzed within acceptable instrument
calibration.
     AVOAS — Over the 6 days of analysis, 10 of
54 of the %Ds for individual analytes were >20%.
These high %Ds occurred with three compounds:
VC, 1,1-DCE, and  1,1,1-TCA.  Two of the VC
responses exceeded 30 %D. If the data associated
with those two CCCSs are considered suspect, 10
of the 30 spiked sample results would be invalid.
Most of these are for VC, for which the majority of
QCCS and SPEW samples had %Rs in excess of
100 ±30%.

Results for Data Quality Objectives

Representativeness

     To determine whether or not the data from
spiked samples introduced into the AVOAS in the
off-line mode were representative of data obtained
when the instrument was operated in the automated
mode, the results for AVOAS on-line and off-line
TT samples were compared.  Vinyl chloride and
1,2-DCA were not  detected by the AVOAS  or
EMSL-LV in the TT samples.  Of the remaining
seven analytes, the difference between the on-line
and off-line AVOAS results was not significant or
only slightly significant for most compounds. The
TT sample results  indicate that  data  generated
using off-line spiked samples should be representa-
tive of data generated by the AVOAS in an on-line,
automated mode.

Completeness

     The QC data for the EMSL-LV instrument
indicated acceptable conditions for all measure-
ments. For the AVOAS, although data in excess of
the QC limits occurred, no statistical outliers were
identified.  All of the AVOAS data values were uti-
lized  in performing the statistical tests and  in
developing  the  conclusions for this demonstra-
tion.  Completeness is therefore considered to be
100%.

Accuracy

     Comparisons  of accuracy results  for the
AVOAS and EMSL-LV were made using %R data
for QCCS and SPEW samples.  The QCCS data
were discussed previously and are summarized in
Table 3-10(a,b).
     SPEW  sample  differences  in  mean  %R
results were discussed previously.  Percent recov-
eries for individual days are presented in Tables 3-
12(a,b).  A total of 210 %Rs for SPEW  samples
were collected on each of the instruments.
     Method 5022 — Percent recoveries ranged
from 48 to 122%, with no results above the limit
and 39% of the results below 70 %R.

     AVOAS — Fourteen percent of the  AVOAS
SPEW sample results were in excess of ±30% of
the target concentrations; recoveries for those sam-
ples were all greater than 100%.  Only three of the
24 VC recoveries were within 100 ±30 %R. Ana-
lytes other than VC with %Rs that exceeded 100
±30% included  1,2-DCA, 1,1,1-TCA, TCE, and
PCE. With the exception of VC, virtually all of the
analyte results were within a 100 ±50 %R window.

Precision

     Precision for TT and SPEW sample results
was discussed earlier. Precision for AVOAS and.
EMSL-LV QCCS sample results are not compara-
ble because of the different QCCS preparation pro-
cedures used.

Comparability

     Results concerning the comparability of the
AVOAS and Method  502.2 were discussed previ-
ously in the sections on TT, SPEW, and QCCS
sample results.

Performance and Systems Audits

     Performance audits use samples spiked with
known concentrations of analytes to quantitatively
evaluate the measurement capability of a system.
No samples were specifically designated as perfor-
mance audits in this demonstration, because many
of the samples were  inherently similar to perfor-
mance audit samples. That is, they were  samples
spiked with known concentrations  of target  ana-
lytes.
     Systems audits are qualitative evaluations of
operational details, performed to ensure  that the
protocols of  the  QAPjP  are  properly  imple-
mented.  Systems audits were conducted at both
the EMSL-LV laboratory and the Wobum site early
in the demonstration by experienced QA auditors
from LESC.  Facilities, equipment, and operations
(such as sample collection and  handling, record-
                                           42

-------
keeping, chain-of-custody and  sample tracking,
data reporting, and QA procedures) were audited.
The EMSL-LV audit indicated satisfactory condi-
tions and compliance with the QAPjP. For the field
site, no serious problems were found. The absence
of written operating procedures for the AVOAS
was noted.  Also,  high  variability  in  the early
QCCS results for the AVOAS was observed; how-
ever, it was too early in the study to determine the
significance of or to identify a cause for that find-
ing.

Qualitative Observations

     The short duration of this demonstration did
not permit a comprehensive qualitative evaluation
of the AVOAS.   However, several observations
regarding system features were made incidental to
the field demonstration and these are summarized
below.

Ruggedness and Adaptability

     The AVOAS hardware is well-designed and
ruggedly constructed.  The system  is sturdy and
without extraneous frills or packaging.  The basic
components are visible and accessible, facilitating
maintenance and  troubleshooting.   The AVOAS
components are designed to  accommodate field-
automated analyses:  the stripping cell  is self-
cleaning  between sampling events', sample port
and sample volume can be automatically or manu-
TABLE 3-1 2(a). PERCENT RECOVERIES FOR SPIKED PURGED EFFLUENT WATER SAMPLES FOR EACH
DEMONSTRATION DAY FOR EMSLV
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dich!oroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,2-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene

1
92.9
87.7
71.9
71.4
76.6
61.5
78.7
76.5
80.2

1
107.0
92.1
76.3
74.6
79.5
67.0
80.9
80.7
83.6

2
89.7
87.7
71.9
73.5
76.6
71.4
76.5
78.3
81.6

2
79.2
66.2
57.4
59.3
60.2
68.1
66.5
66.6
66.7
Low
3
47.8
61.4
53.8
56.0
61.3
56.0
67.6
61.9
64.6
High
3
76.2
71.7
59.6
60.4
62.0
58.2
63.2
63.8
64.8
Spikl
4
92.0
85.5
71.2
71.4
74.5
67.0
75.4
75.0
76.0
Split*
4
104.0
88.2
74.1
75.7
76.3
71.4
79.8
74.6
76.0
Medium Spike
5
96.4
92.1
74.9
76.8
81.6
77.9
86.5
86.3
87.8
6
90.2
109.6
71.2
73.5
74.5
122.9
74.3
75.5
83.2
1 2
104.0 78.9
73.5 60.4
77.9 67.7
86.4 83.8
86.5 81.0
81.2 73.5
84.4 85.7
75.3 75.5
84.6 74.4
3
70.1
58.2
63.6
78.9
75.5
68.1
82.1
71.0
71.0
4
74.9
53.7
58.5
72.1
68.4
63.7
73.7
64.3
62.0
5
90.6
63.7
62.6
75.7
70.5
68.1
71.7
65.6
63.1
6
72.2
48.6
53.3
64.5
59.9
65.9
68.6
55.8
52.7
Super High Spike
5
120.7
103.5
90.1
88.9
87.3
94.4
87.6
87.7
86.1
6
102.2
84.9
69.8
68.1
68.4
75.7
73.2
69.9
68.4
1 2
77.8 68.3
i a
68.3 63.7
79.0 74.6
73.9 65.3
70.3 69.7
82.5 73.6
75.9 68.4
82.3 69.9
3
77.6
a
60.1
68.1
61.2
71.9
67.3
64.7
78.4
4
60.2
a
49.8
59.3
55.9
59.3
62.6
55.3
57.5
5
97.7
a
68.9
76.8
65.5
75.7
72.3
70.3
73.8
(
343
i
622
79.0
70.4
70.3
69.8
67.5
67.6
 1,1-dicMonethene was not spiked into superhigh-concentration spiked purged effluent water samples.
                                             43

-------
ally selected; and the system can be adapted to dif-
ferent purgeable analytes by using the appropriate
GC column, temperature program, and detector.

Maintenance Requirements

     The maintenance requirements of this instru-
ment are similar to those for other analytical instru-
mentation.  Routine GC  maintenance procedures
are required, e.g., changing gas cylinders, cleaning
detectors, and  replenishing integrator paper.  Sys-
tem-specific maintenance  includes  cleaning  or
changing the  inlet water filters to  the  injector
(about once a day); changing the peristaltic pump
tubes on the injector (about  once a week); and
changing the drying trap on the injector (about
once a week).  These routine maintenance proce-
dures do not appear to be time-consuming or diffi-
cult
     According to the AVOAS developer, they can
be completed in less than 15 min and can be done
at the time of the continuing calibration check.

Facility and Supplies Requirements

     The AVOAS requires an environment consis-
tent with that found in an analytical laboratory:  a
sturdy bench or tabletop (approximately 80 x 30
in), a clean source of water, constant temperature
(ideally  20°C, ± 3°C), and a relatively clean, VOC-
free  environment   If telecommunications are
desired, a dedicated  telephone line  is necessary.
The instrument is not suitable for quick setup in a
harsh environment (e.g., as  required for emer-
TABLE 3-12(b). PERCENT RECOVERIES FOR SPIKED PURGED EFFLUENT WATER SAMPLES FOR EACH
DEMONSTRATION DAY FOR THE AVOAS
Low Spiki
Compound
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene
1
145.7
97.8
96.3
101.3
96.5
95.9
97.6
96.0
92,0
2
164.3
98.0
99.1
104.5
101.3
104.1
97.1
104.0
98.4
3
125.1
82.5
86.9
95.5
93.9
93.9
93.4
92.1
88.9
4
135.4
94.5
74.1
95.1
97.9
104.5
98.8
93.4
90.7
5
141.1
91.9
78.4
103.7
106.7
114.4
106.2
100.8
93.2
6
142.9
98.2
99.0
106.7
97.3
189.5
102.5
105.6
104.7
1 2
160.9 151.3
82.3 79.0
100.6 102.1
122.4 117.6
107.4 105.7
120.6 118.0
116.0 108.2
104.9 106.3
99.9 101.2
High Splk*
vinyl chloride
1,1-dichloroethene
trans-1,2-dichloroethene .
1,1-dichloroethane
cis-1,2-dichloroethene
1,1,1-trichloroethane
1,2-dichloroethane
trichloroethene
tetrachloroethene
1
146.1
97.7
95.7
98.7
95.0
94.4
95.4
97.3
96.3
2
157.0
101.1
97.8
98.5
96.3
96.0
91.6
102.0
98.2
3
118.3
87.3
85.8
91.8
88.3
89.5
88.2
93.0
91.2
4
130.7
89.7
82.8
88.5
77.3
75.8
76.4
80.2
79.7
5
132.5
103.0
74.6
102.6
101.6
110.3
102.9
103.2
98.9
6
136.7
103.7
98.1
98.9
94.4
97.8
99.0
104.2
105.7 .
1 2
162.1 138.9
a a
104.0 100.4
121.7 106.4
117.1 104.7
130.0 102.1
114.0 99.5
119.7 114.3
125.5 120.0
Medium Spike
3
138.4
74.5
93.2
114.5
97.7
114.0
102.0
94.9
92.8
4
92.8
47.4
62.1
75.4
63.8
69.1
71.5
63.4
58.9
5
145.7
78.0
72.9
117.0
99.2
106.9
111.3
96.5
90.4
6
155.0
76.8
99.6
120.4
100.8
123.1
108.6
103.4
102.9
Superhlgh Spike
3
151.4
a
93.7
104.1
97.5
106.8
100.0
105.6
108.7
4
124.5
a
90.1
96.7
92.0
92.0
87.1
100.4
110.4
5
163.1
a
97.5
103.7
99.1
108.3
94.8
105.1
109.1
6
189.4
a
108.6
121.9
114.5
136.9
121.2
124.1
138.1
 /, 1-dichloroethene not spiked into superhigh-concentration spiked purged effluent water samples.

                                              44

-------
gency-response  screening   analyses),   but   is
designed for long-term monitoring in an undis-
turbed location.  A constant 20-A, 120-V power
source is required.  Any extended interruptions in
power require a warmup  time to ensure  that  the
detection system has reached a stable equilibrium.
After a power interruption, a calibration check is
required to ensure that the system instrument has
not changed.
     A water supply is required for flushing  the
lines  and  sample loops between analyses. The
AVOAS includes a sparging system that uses high-
purity inert gas to purge a water source of VOCs.
At  most sites, local tapwater could be used as a
source for rinse-water. Otherwise, bottled purified
water can be used.
     The instrument should be operated in a con-
stant-temperature environment because a fluctua-
tion of more than a few degrees would probably
alter the sparging efficiencies of the target volatile
analytes.   This situation was not encountered or
studied during the demonstration.
     Provision must  be made for treatment and
disposal of waste water generated by the AVOAS
during its operation.  Several liters of treatment
train water (500 mL/min for several minutes)  are
generated during sampling.  The amount depends
on the length of the sampling lines.  At the Wobum
site, the contaminated waste water was collected
and recycled into the treatment system, so waste
water disposal was not a problem.
     In addition to compressed gases used to oper-
ate  the GC, the AVOAS requires zero-grade helium
for  the injector purge  gas, and  standard-grade
helium for  the rinse-water  purification  system.
Other supplies are the same as those required  for
standard GC analyses of VOCs in water.

System Costs

     The AVOAS is  currently not for sale. The
developer intends to initially lease AVOAS units at
rates  that  will depend upon the  specific system
configuration; a one-time fee for system setup and
operator training will be charged.  An additional
expense in using the AVOAS would be that for an
operator.   Although an  operator was present  for
about 14 hr/day during the demonstration, during
routine monitoring, an operator's presence would
be required for at most 1 hr/day.
     The AVOAS was leased at the Wells G and H
Superfund Site for $8000 per month. Assuming
that 20 samples/day could be processed under rou-
tine monitoring operations (not including QC sam-
ples), the cost per sample over a month would be
less than $15, not including operator labor cost.
The cost per sample for comparable analysis by
conventional methods is approximately $225-275,
depending on the method used and the  number of
target analytes. However, because of sample turn-
around time (usually on the order of several days),
conventional methods are not typically suited to
high-frequency, long-term monitoring, so the costs
may not be directly comparable.

Summary of Results and Discussion

1. For  most of the TT  samples, concentrations
  measured by the AVOAS in  the on-line mode
  were slightly higher than those measured  off-
  line; however, the difference was not significant
  for most compounds.  This indicated that the
  findings  for samples  analyzed off-line (e.g.,
  SPEW and QCCS samples) could be considered
  representative  of  on-line performance,  and
  could be used for comparisons of the AVOAS
  and EMSL-LV results.

2. Nearly all recoveries for both  TT and SPEW
  samples were higher for the AVOAS than for the
  EMSL-LV analyses. However, the overall mean
  %R for EMSL-LV (for all compounds, concen-
  tration ranges, and days) for SPEW samples  was
  74%, in comparison to 104% for the AVOAS.
  These findings seem to indicate that the AVOAS
  gave results more nearly equivalent to the target
  concentrations than  EMSL-LV.  QCCS  %R
  results for EMSL-LV close to 100% and CCCS
  results  consistently within acceptable ranges
  would seem to argue against a  problem in instru-
  ment calibration.  Transport or holding time
  effects are a possible explanation for the finding
  of lower EMSL-LV analyte concentrations.

3. The effect of transport on sample results is open
  to question.   The variability  due  to  analyst
  appears to exceed that due to transport.

4. Treatment  train  samples analyzed  by  the
  AVOAS in the afternoon had  slightly higher
  analyte concentrations than those analyzed in
  the morning, especially for  the  less  volatile
  compounds.  This finding is consistent with  car-
  ryover  from  superhigh-concentration  SPEW
  samples analyzed between the morning and the
  afternoon. Both TT and superhigh-SPEW sam-
  ples were collected through the smallest sam-
  pling loop. The  increases in  concentration are,
  however, low enough to be of no practical  sig-
  nificance.
                                             45

-------
5. AVOAS %R results for vinyl chloride were van-.
   able and substantially exceeded those for other
   compounds analyzed by that system, and they
   were proportionately higher than those of the
   EMSL-LV analyses. The source of the problem
   cannot  be  determined from the demonstration
   data. If the AVOAS were considered for use at a
   site having vinyl chloride as a contaminant, the
   system's performance in measuring that com-
   pound would need to be more rigorously evalu-
   ated.

     The  question of false-positive results  arises
with respect to compounds detected by  the AVOAS
but not by EMSL-LV in the TT samples. In all of
the sample pairs for which the AVOAS had detect-
able concentrations of compounds but EMSL-LV
did  not,  the  concentrations measured by  the
AVOAS were less than the detection limits for the
EMSL-LV instrument Thus, it is unlikely that the
AVOAS results reflect false positives.
     Although  false-negative  results (i.e.,  not
detecting a contaminant when in fact it is present)
were not studied in this evaluation, such erroneous
results  are possible,  and steps should  be taken to
minimize them. The consequence of a false-nega-
tive result could be  to not recognize  a failure or
decline in  function of some component in the treat-
ment system. False-negative results could poten-
tially occur if there  were a problem in either the
sampling or the analytical system; for example, if a
sampling line were clogged or if the manifold were
not functioning properly. Ongoing instrument cali-
bration checks are important, as are confirmatory
analyses,  particularly of samples from effluent
ports, where the risk of actual VOC concentrations
exceeding  measured values is of most concern; i.e.,
where the  potential exists for releasing water con-
taining VOC concentrations above the allowable
effluent limits.
     The data from this study were also evaluated
by Stanley Deming and John Palasota, of the Uni-
versity of Houston Department of Chemistry under
a cooperative agreement with the EMSL-LV. Their
findings are summarized in a technical report sub-
mitted to the EPA through a cooperative agree-
ment.  Although their approach to data analysis
was somewhat different (i.e., ratios of paired data
rather than differences in mean %Rs  were evalu-
ated), the findings were  similar.  That is, the
AVOAS gave results that were comparable to those
obtained using Method 502.2.   Bias exists; the
AVOAS results are high compared to the EMSL-
LV results. The source for this bias is not possible
to determine from the data of this study.  Possible
explanations could include calibration differences
between the two locations or degradation of sam-
ples due to sample shipment and holding time. The
QC data do not indicate a calibration problem, and
the potential for a holding time effect is not sup-
ported by the TT sample data or the transport study
results.  Whether bias would occur in  general (i.e.,
for routine use of the AVOAS) is yet to be learned.
     An interpretation of the data from this dem-
onstration, and any extension of the results to other
similar evaluations, must be tempered by several
considerations. Throughout the demonstration the
AVOAS was being operated by its inventor, who
was there daily monitoring and adjusting its perfor-
mance.  Also, the demonstration was conducted
shortly after the system was installed, so that any
long-term effects, such as loss in performance due
to wear and tear on the system or contamination of
its various components (e.g., sampling Ones, mani-
fold, sampling loops, sparging unit) could not be
evaluated.
                                             46

-------
                                         Section 4
                          Conclusions and Recommendations
Conclusions

     The AVOAS  demonstration went smoothly,
and was considered by both the AVOAS developer
and EMSL-LV to be a fair evaluation of its perfor-
mance under the study conditions.  All  samples
specified by the demonstration design were ana-
lyzed, and all of the data were judged to be of suffi-
cient quality to be  included in the data base upon
which the following conclusions for this demon-
stration are made.  The qualitative and quantitative
findings generated  in this demonstration provided
the basis for the following conclusions regarding
the performance of the AVOAS:
1. Mean  concentration values  generated  by the
   AVOAS  for all  analytes in all SPEW and TT
   treatment train samples were higher than those
   measured by Method 502.2.  The cause of this
   bias is difficult to determine from the data avail-
   able.  The bias  is more likely due to  sample
   holding time and/or transportation effects, ana-
   lyst techniques for calibration, preparing spiked
   samples, or analysis, or some combination of
   these, than to the instrument itself.
2. The precision obtained by the AVOAS was com-
   parable to that obtained by the laboratory instru-
   ment, within the range of expected variability
   for samples containing VOCs.
3. The AVOAS performed well in analyzing sam-
   ples with VOC concentrations in the low-ppb to
   ppm range.

4. The  system used by the AVOAS for stripping
   VOCs from water (one of the proprietary fea-
   tures  of  the AVOAS) is simple in design and
   function, self-cleaning, and more adaptable to
   field automation  than the traditional purge-and-
   trap procedure.
5. The  fact that  the AVOAS  was  operational
   shortly after installation, and was running nearly
   24 hours per day for the duration of this demon-
  stration indicates the potential reliability of the
  system is high.

6. Based on the findings of this demonstration, the
  AVOAS appears to have the potential to provide
  data  comparable to  that  which  would  be
  obtained using established sampling and analyti-
  cal methods. The discrepancies in accuracy and
  precision are not substantial enough to offset the
  advantages of  the  system and are probably
  attributable factors other than instrument perfor-
  mance.

Advantages of the AVOAS

1. The AVOAS is simple in design, and has easily
  accessible  components.  The  LESC  scientist
  present  at the field site observed that the minor
  instrumental problems encountered during the 6
  days of analysis were relatively easy to diagnose
  and repair.

2. The AVOAS appears to be capable of providing
  the benefits of automated sampling and analysis
  discussed in Section 1 of this report; i.e., rapid
  results and the  elimination of separate sample
  collection, handling,  shipping,  and  analysis
  steps.
3. The potential cost savings per sample for use of
  the AVOAS in comparison to use of established
  methods could be substantial. Savings would be
  realized not only in the activities related to sam-
  pling and analysis,  but  in the increased effi-
  ciency brought about by real-time monitoring of
  the performance of ground-water treatment sys-
  tems.  We must emphasize that, with a dedi-
  cated,   on-line  system, sampling  frequency
  would probably be greater than if conventional
  methods were used. For that reason, the cost per
  sample would be lower.  In addition, the likeli-
  hood of detecting a change in system perfor-
  mance would be increased.
                                             47

-------
 Limitations of the AVOAS

     Although the following items were identified
 as potential limitations to the AVOAS, many of
 them are merely factors to be kept in mind by pro-
 spective users of the AVOAS. The AVOAS devel-
 oper has since taken steps to correct equipment or
 procedural problems.
 1. The AVOAS requires a temperature-controlled
   environment in an atmosphere free of contami-
   nation.  Extra measures to ensure these condi-
   tions may be required in a typical field situation
   or industrial facility.

 2. Water samples high in paniculate matter tend to
   clog a filter in the injector inlet This problem
   can be minimized if water is pre-filtered before
   being conveyed to the AVOAS or if the inlet fil-
   ters are cleaned or changed as part of the routine
   daily maintenance.

 3. Internal standards cannot be  analyzed by the
   AVOAS  in the on-line mode.   Such standards
   provide an ongoing indication of instrument per-
   formance.  It is  possible to indirectly monitor
   instrument  performance  through  calibration
   checks and observation  of analyte concentra-
   tions for each port  However, this approach is
   not recommended, and  is not diagnostic for
   problems  with the analytical  versus the sam-
   pling and sparging portions of the system.

4. Carryover  in the smallest  sample loop of the
   AVOAS  was suggested  by the  finding of an
   increase in trace  concentrations of several ana-
   lytes in TT samples between the morning and
   afternoon analyses. The corresponding  EMSL-
   LV data do not reflect these findings. Because
   of the low concentrations involved, the possibil-
   ity of carryover is not a serious impediment to
   use of the AVOAS for higher concentration ana-
   lytes. The AVOAS developer  has modified the
   design of the manifold to reduce dead  volume
   significantly, and expects this change to greatly
   reduce the carryover problem.

5. The volume of sample injected by the smallest
   sampling loop was difficult to measure  accu-
   rately because the dead volume of the  sample
   valve (i.e., the internal volume of valve  parts,
   aside from sample volume measurement hard-
   ware) was disproportionately large. The devel-
   oper  is  examining  better  approaches  to
   estimating  volumes  for each  of the sampling
   loops.
6. Most of the problems, albeit minor, encountered
  with the AVOAS during the field demonstration
  were due to problems in the software.  Some
  were solved during the demonstration, and plans
  were made to correct the others shortly after the
  demonstration was completed.

Limitations of the AVOAS Demonstration

     This demonstration was designed to evaluate
specific aspects of the AVOAS performance rela-
tive to an established method; the demonstration
was not a comprehensive evaluation of all of the
proposed advantages or potential limitations of the
system. The conditions of this demonstration were
atypical of those that would probably exist under
routine monitoring operations.  A number of sam-
ple types were analyzed in the off-line mode and
the demonstration was performed with the AVOAS
inventor closely checking and adjusting the sys-
tem's performance.  The capability of the AVOAS
to be operated unattended from a remote location
under software control  was not evaluated;   nor
were the repeatability and reliability of calibration
and maintenance procedures, or the long-term per-
formance of the system.

Recommendations

     Recommendations are made here specifically
for the AVOAS and for future SITE Program dem-
onstrations. A number of the issues encountered for
the AVOAS demonstration design  apply  to  the
demonstration of virtually any technology. Reso-
lution of these issues would facilitate collection of
high quality data for future SITE demonstrations.

Recommendations Specific to the AVOAS

1. Further development of the AVOAS is encour-
  aged.  A fair evaluation at a future stage of
  development would require monitoring the sys-
  tem's performance when operated totally in the
  automated mode, without interruption or inter-
  vention. Performance data should be generated
  for this system when used at other sites  and by
  different operators.

2. Instrument calibration should be checked regu-
  larly, on a schedule based on the performance of
  the detector. A daily calibration check is recom-
  mended.  If this is not feasible, at a minimum,
  detector performance should be monitored daily
  over a period of several weeks after instrument
  installation. An appropriate calibration schedule
  should be developed and followed for the long-
                                             48

-------
   term use of the system, especially for unat-
   tended, remote operation.

3. A mechanism for introducing surrogate or inter-
   nal standards into the AVOAS with each analy-
   sis would provide an  ongoing indication of
   analytical instrument performance.

4. Depending on the  data  quality  objectives for
   instrument performance at a  particular  site,
   blank samples should be periodically analyzed
   on all sampling loops used for analysis to ensure
   that carryover from high-concentration into low
   concentration samples is not occurring.

5. In this demonstration, response factors for three
   of the sampling loops were based on estimated
   loop sizes. Errors in estimating the sample loop
   sizes are thought to have influenced the calcu-
   lated response factors for some of the loops, par-
   ticularly the smallest size. A procedure for more
   accurately determining sample loop  size should
   be developed.  Alternatively, each sample loop
   should be individually calibrated.

6. During the initial stages of long-term monitor-
   ing, concentrations  and precision for each sam-
  pling port  should  be  closely  monitored  to
  generate baseline conditions.   Also,  periodic
  confirmatory analyses by  established  methods
  should be performed to verify AVOAS results.
7. An instruction  and procedural manual for the
  system  should be written.   QA/QC procedures
  should be included as part  of the standard oper-
  ating procedures.

Recommendations for Future Studies
1. A more rigorous evaluation of holding time and
  transport effects on measured concentrations of
  VOCs in water samples is recommended.
2. An investigation of the  stripping efficiency for
  various VOCs from water,  in purge and trap and
  similar  processes (e.g., the method used by the
  AVOAS), for samples of different  concentra-
  tions and volumes is recommended.
3. A study to distinguish the variability due to cali-
  bration-standard spiking procedures of various
  analysts from the variability due to  instrument
  performance is recommended.
                                              49

-------
                                        References
Glaser, J.A., D.L. Foerst, G.D. McKee, S.A. Quave, and W.L. Budde.  1981.  'Trace Analyses for Waste
Waters," Environ. Sci. Technol. 15:1426 - 35.
Stanley, T.W. and  S.S. Vemer.   1983.  Interim Guidelines and Specifications for Preparing Quality
Assurance Project Plans. EPA/600/4-83/004. U.S. EPA, Washington, D.C.
Stanley, T.W. and S.S. Verner.  1985.  The U.S. Environmental Protection Agency's Quality Assurance
Program. In: J.K. Taylor and T.W. Stanley (eds.). Quality Assurance for Environmental Measurements.
ASTM STP 867, pp. 12-19. American Society for Testing and Materials. Philadelphia, Pennsylvania.
Taylor, J.K.   1987.  Quality Assurance of Chemical Measurements.  Lewis Publishers, Inc., Chelsea,
Michigan. 328pp.
U.S. Environmental Protection Agency. 1987. "Quality Assurance Program Plan."  EPA/600/X-87/241.
Environmental Monitoring Systems Laboratory, USEPA, Las Vegas, Nevada.
U.S. Environmental Protection Agency.  1988.  "Availability, Adequacy, and Comparability of Testing
Procedures for the Analysis of Pollutants Established under Section 304(h) of the Federal Water Pollution
Control Act" EPA/600/9-87/030. USEPA, Cincinnati, Ohio.
U.S. Environmental Protection Agency.  1991.  "Preparation Aids for the Development of Category  H
Quality Assurance Project Plans."  EPA/600/8-91/004. USEPA, ORD, Washington D.C.
                                             50

-------