United States Office of Emergency and Publication 9355.4-05
Environmental Protection Remedial Response PB92-963346
Agency Washington, DC 20460 February 1992
Superfund
<&EPA Evaluation of Ground-Water
Extraction Remedies:
Phase II
Volume 1
Summary Report
-------
Publication 9355.4-05
February 1992
EVALUATION OF GROUND-WATER
EXTRACTION REMEDIES: PHASE II
Volume 1
Summary Report
Office of Emergency and Remedial Response
U. S. Environmental Protection Agency
Washington, D.C.
-------
NOTICE
The conclusions outlined in this document are intended solely for technical support to EPA personnel. They
are not intended, nor can they be relied upon, to create any rights, substantive or procedural, enforceable by
any party in litigation with the United States. The Agency reserves the right to act at variance with these
policies and procedures and to change them at any time without public notice.
For additional copies of this report please contact:
National Technical Information Service (NTIS)
U.S. Department of Commerce
5285 Port Royal Road
Springfield, VA 22161
(703) 487-4650
11
-------
INTRODUCTION TO VOLUME 1
This report is the second phase of a study to evaluate the effectivenss of ground-water extraction systems
being used to remediate ground-water contamination at hazardous waste sites. This report was prepared in the
volumes. Volume 1: Summary Report, contains an Executive Summary and chapters which discuss the
purpose, methodologies, and conclusion of the project. Volume 2: Case Studies, contain the individual
analyses of each of the 24 sites associated with this project.
ill
-------
CONTENTS
Page
Executive Summary vii
Chapter
1 Introduction 1
Project History 1
Phase II Methods and Objectives 1
Background on the Importance of NAPLs 2
2 Overview of Case Studies 3
Site Background Characteristics 3
Length of Historical Record 3
Plume Size 3
Contaminants of Concern 6
Geologic Environments 6
Issues Related to System Design 6
Remedial Objectives 6
Projected Cleanup Time 7
Extraction and Monitoring System Sizes 9
Enhancement Technologies 9
System Design Information 10
Occurrence of NAPLs 10
3 Summary of Remedial Progress 13
Plume Containment 13
Aquifer Restoration Effectiveness 15
Contaminant Mass Removal 15
Contaminant Concentrations 17
Plume Area Reduction 19
Use and Effectiveness of Enhancement Technologies .... 19
Soil Vapor Extraction 19
Ground-Water Reinjection 20
Slurry Wall Containment 20
Fracture Enhancement 21
Update on Site Data Requirements 21
Hydrogeologic Information 21
Contaminant Characteristics and Distribution 22
IV
-------
CONTENTS
(Continued)
Page
Occurrence and Implications of NAPLs 24
Waste Handling Practices Leading to NAPL
Contamination 24
Identification of NAPL Presence 24
Implications of NAPL Presence 26
-------
EXHIBITS
Figure Name Page
Figure 2-1 Geographic Distribution of Case-Study Sites 4
Table 2-1 Site Background Characteristics 5
Table 2-2 Summary of Design-Related Information 8
Table 2-3 Summary of NAPL Occurrence 12
Table 3-1 Summary of Remedial Effectiveness 14
VI
-------
EXECUTIVE SUMMARY
This summary report describes the second phase of
an evaluation of the current effectiveness of
ground-water extraction in remediating
contaminated aquifers at several hazardous waste
sites. This project involved reviewing data from
existing extraction systems and should not be
viewed as a "technology evaluation" in the typical
sense because no attempt was made to select sites
where the extraction systems had been optimized.
Due to the limited number of sites with operating
extraction systems, selection criteria were limited
to identifying those sites where extraction systems
had been in operation long enough to generate
initial performance data.
This report does not go beyond describing the
operation and conclusions associated with the 24
sites in the study. However, analysis of these
findings provides part of the basis for identifying
other guidance needs, determining modifications to
our approach to ground-water remediation, and
assessing the need for future studies and research.
A data collection guide, a screening guide for
assessing the likelihood of DNAPLs, and a
Directive clarifying EPA's approach to ground-
water remediation will be developed by EPA over
the next year.
Phase I of the study was completed in 1989 (U.S.
EPA, 1989). In this second phase, data from 17 of
the original 19 case studies were updated, and new
case studies were prepared for five additional sites.
Two of the original case studies were not updated
because more current site data either had not been
generated or could not be obtained. The second
phase of the study put special emphasis
nonaqueous phase liquids (NAPLs), which can
increase the time frame and complexity of ground-
water remediation.
The 24 case studies (U.S. EPA, 1991) that now
comprise the results of this evaluation must still be
considered a very small database from which to
draw general conclusions. The case-study sites
represent a variety of subsurface contamination
situations, geologic environments, and remedial
approaches. Records of extraction system
operation vary in length, but in most cases are
relatively short compared to the time that may be
required to complete aquifer remediation (aquifer
remediation is not always the declared remedial
objective). Because these sites represent some of
the few sites where extraction systems have
actually been in operation, they also represent
some of the earliest ground-water investigations.
Consequently, those investigations predate
application of recent advances in site
characterization methods and approaches. In most
of the cases, site data that were obtained for this
evaluation leave important questions unanswered.
These are described in more detail in Chapter 3.
Despite these shortcomings, it is possible to draw
some tentative conclusions from the results
reported here. Continued monitoring of remedial
progress at these and other ground-water extraction
sites, together with results from other ongoing
research in the field, can be expected to lead to
more effective application of ground-water
extraction technology in the future.
CONCLUSIONS
The results of the phase n study reinforced the
main conclusions of phase I and led to some
additional conclusions concerning impacts of
NAPLs on ground-water remediation. The first
four conclusions are the same conclusions drawn
in the initial study and are presented here to re-
emphasize these findings.
Conclusion 1: Data collected, both site
characterization data prior to system design and
subsequent operational data, were not sufficient to
fully assess contaminant movement or ground-
water system response to extraction.
Conclusion 2: In the majority of cases studied (15
of the 24 sites), the ground water extraction
systems were able to achieve hydraulic
containment of the dissolved-phase contaminant
plume.
Conclusion 3: Extraction systems were often able
to remove a substantial mass of contamination
from the aquifer.
Conclusion 4: When extraction systems were
started up, contaminant concentrations usually
showed a rapid initial decrease, but then tended to
level off or decrease at a greatly reduced rate.
vil
-------
This may be a result of the type of monitoring
data collected as much as a reflection of an actual
phenomenon of ground-water extraction systems.
For example, it can reflect successful remediation
as the contaminated zone shrinks and less-
contaminated ground water is puEed into the
extraction system, or poor placement of ground-
water monitoring wells.
Conclusion 5: Based on the available information,
potential NAPL presence was not addressed during
site investigations at 14 of the 24 sites. At 5 sites
they were "addressed" because they were
encountered unexpectedly during the investigation.
As a result, it is difficult to determine NAPL
presence conclusively from available site data.
Because NAPLs were not addressed in the site
investigation, they also were not addressed in the
remedial design. Consequently, a ground-water
extraction system may be performing as designed
(removing dissolved phase contaminants) even
though it will not achieve the cleanup goals within
the predicted timeframe.
j Conclusion 6: At 20 of the 24 sites, chemical data
collected during remedial operation exhibited
trends consistent with the presence of dense
nonaqueous phase liquids (DNA'PLsJp However,
t even where substantial solTand water quality data
were available, a separate immiscible phase was
rarely sampled "or observed. This is consistent
with DNAPL behavior; i.e., they can move
preferendSHy^ through very discrete pathways that
may easily be missed even in thorough sampling
schemes. DNAPL was observed at sites where
contaminant concentrations in ground water were
less than 15% of the respective solubilities.
Conclusion 7: The importance of treating ground-
water remediation as an iterative process, requiring
ongoing evaluation of system design, remediation
time frames, and data collection needs, was
recognized at all of the sites where remedial action
was continuing.
SITE BACKGROUND
INFORMATION
With the completion of phase II, moderately
detailed case studies of 24 hazardous waste sites
have been produced. Remedial performance
histories range from 1 to 12 years, involving
contaminant plumes ranging in size from less than
1 acre to more than 7,000 acres. In 19 of cases,
the contaminants of concern included volatile
organics, usually chlorinated solvents.
Ground-water remediation systems installed at
these sites reflected a wide range of intensity of
remedial effort The number of extraction wells
per site ranged from 1 to 203, and total extraction
rates of up to 9,200 gallons per minute were
reported. The number of monitoring wells at
individual sites ranged from six to 250.
Ground-water extraction systems at several sites
were supplemented by additional remedial
technologies. The most common of these were
ground-water reinjection and soil vapor extraction,
each of which were used at six sites. Soil vapor
extraction systems usually removed significant
quantities of contaminants, but the effect of this
removal on reducing aquifer cleanup time could
not be quantified. Reinjection was sometimes
used more as a means of ground-water disposal
than to increase the rate of contaminant migration.
Slurry wall containment was used at three sites;
French drains, fracture enhancement, and
intermittent pumping were used at one site each.
REMEDIAL PERFORMANCE
As of the conclusion of this second phase
evaluation, a successful aquifer cleanup has been
reported for only one of the subject sites (Emerson
Electric in Altamonte Springs, Florida). An
apparent remedial success was reported in the first
phase evaluation. However, remedial success
claims were based on limited monitoring data, and
may therefore be open to question. No new data
were available for this site during the second phase
evaluation.
Plume containment appeared effective at 15 of the
24 sites. At six other sites, the containment
effectiveness was uncertain because of insufficient
plume monitoring (chemical or water level) or
contradictory site data. Plume containment
appeared incomplete at only three sites.
Fifteen sites had data on contaminant mass
removed by the extraction systems. Amounts
removed ranged from 10 pounds to more than
203,000 pounds. Removal estimates were
provided by the parties responsible for
remediation, or were calculated as part of the
Phase II study. Information needed to make such
an estimate was unavailable for nine sites.
Vlll
-------
In light of the observation that contaminant
concentrations frequently appear to stabilize above
the cleanup goals, an effort was made to identify
these levels in the performance records for each
site. Based on data availability, this trend was
identified in either the influent to the treatment
system (11 cases), or in individual wells (9 cases).
The identification of stabilization required some
subjective judgment as to what constituted a stable
concentration and what did not. Stabilized
concentrations appear to have occurred at 17 sites.
The apparent stabilization of contaminant
concentrations may be due to a number of factors
not necessarily related to technical limitations of
ground-water extraction. These include non-
representative monitoring techniques, other
contaminant sources not previously identified,
inadequate extraction network design, and/or
inefficient operation of the extraction network.
OCCURRENCE OF NAPLs
At nine case-study sites, parties responsible for
remediation acknowledged the presence of NAPLs.
NAPLs were observed directly in eight of these
cases, and in the ninth, the determination was
based on circumstantial evidence. For at least
seven of the other 15 sites, NAPLs appeared likely
even though there was no direct confirmation or
mention of them in the site reports. The
likelihood of NAPL presence at all 24 sites was
estimated using a rating scale of 1 through 5.
Ratings were based on indirect evidence such as
high contaminant concentration in ground water,
depth of contamination in the aquifer, persistance
of contaminant plume during remediation, and
contaminant source characteristics.
Analyte concentrations in excess of the
contaminant aqueous phase solubility were
reported for ground-water samples at three sites.
This provides a strong indication of the presence
of NAPLs. One of these instances occurred at a
site where the presence of NAPLs has not been
acknowledged. At some sites where NAPL
presence was acknowledged, the maximum
contaminant concentrations reported were less than
15 percent of contaminant solubility. It is possible
that in some cases, NAPL is being removed by the
ground-water extraction systems without this being
recognized by the system operators.
IX
-------
Chapter 1
INTRODUCTION
PROJECT HISTORY
In 1989, EPA's Office of Emergency and
Remedial Response (OERR) completed a study of
19 hazardous waste sites at which ground-water
extraction systems were being used to remediate
aquifer contamination. The study utilized
available site-investigation documents, such as
remedial investigation (RI) and feasibility studies
(FS) reports, and annual or quarterly performance
monitoring reports that were generally current
through late 1988. A project report was issued as
three volumes (U.S. EPA, 1989): Volume 1,
Summary Report; Volume 2, Case Studies; and
Volume 3, General Site Data-Data Base Reports.
That study and the reports produced comprise the
first phase of the evaluation.
In late 1990, OERR initiated a second phase of the
ground-water extraction evaluation. In it the
original case studies were to be updated, and five
new case-study sites were to be evaluated with
special emphasis on the occurrence of
contaminants in the form of NAPLs.
PHASE II METHODS AND
OBJECTIVES
In general, the second phase objective was to
evaluate the remedial effectiveness of
ground-water extraction systems that had sufficient
operational data to allow an initial assessment of
system performance. These evaluations focused
on the capability of the extraction systems to
control and remove ground-water contamination.
Evaluation of subsequent treatment and disposal of
the extracted ground water is beyond the scope of
this evaluation, except for instances where
treatment or disposal issues affected the
performance of the extraction system. In this
regard, the objectives of the first and second
phases of the evaluation are the same. The second
phase updates the original 19 case studies using
current performance information and provides five
new case studies.
Site information updates covering the period from
• late 1988 through 1990 was obtained from the
same regulatory and responsible-party contacts
who provided site data in the first phase.
Remedial performance information generally was
received in the form of annual or quarterly
monitoring reports. Commonly, the annual reports
for the preceding year are compiled and released
in the first half of the following year. The timing
of these reports caused problems for some site
evaluations, because 1990 annual reports were not
always available. For most sites, performance data
were obtained through the third or fourth quarter
(September or December, respectively) of 1990.
Effort was made to locate new case-study sites
that satisfied the second phase selection criteria:
Superfund sites at which NAPLs were known to
be present and at which aquifer remediation had
been in progress long enough to produce initial
performance data. These criteria proved difficult
to satisfy. In the ends five sites were selected,
four of which were Superfund sites. The fifth,
Occidental Chemical, predates the Superfund
legislation and is therefore, strictly speaking, not
part of the program. However, it is being
administered by the State of California and the
EPA National Enforcement Investigation Center
using procedures comparable to those of
Superfund. The presence of NAPLs was
acknowledged at three of the five new case-study
sites, and NAPLs are quite likely to be present at
t one of the others.
The reporting format used in Phase II is similar to
that of the first-phase evaluation. Background
information on site history, geology, hydrogeology,
and waste characteristics is presented first. This is
followed by a description of the ground-water
extraction system, remedial objectives, and some
of the pertinent design considerations. The next
section presents a review of system performance
data. This review is based only on the site
information obtained. This information was not
always conclusive, and disagreements regarding
data interpretation has sometimes been noted
among various parties involved in the response
action. Statements presented in the performance-
evaluation sections of the case studies reflect
judgements (by others) contained in site
information packages and were not the result of
-------
this study. Conclusions drawn in the course of
this evaluation are contained in the "Summary of
Remediation" or the "Summary of NAPL-Related
Issues" sections.
Records for many sites contain no explicit mention
of NAPLs or the possibility of their presence.
Nonetheless, site data frequently contain clues
indicating that NAPLs may be present, and these
data are discussed for each case study. In the
"Summary of NAPL-Related Issues" section, the
Phase II authors of the case studies speculated on
this possibility in light of the evidence hi the site
background and performance data. This should
not be construed as an official determination by
EPA concerning the nature of the contaminants at
the site.
The format used for case study updates is similar
to that used in the original case studies, but
background issues are presented in less detail.
The updates are meant to be readable as
stand-alone documents, but to gain the fullest
understanding of the sites it is necessary to read
both the original and the update of the case study.
BACKGROUND ON THE
IMPORTANCE OF NAPLs
Phase I of this study identified several factors that
potentially increase the time frame and complexity
of ground-water remediation. These factors
include: hydrogeological factors (e.g., subsurface
heterogeneity, presence of low-permeability zones,
and presence of fractures); contaminant-related
factors (e.g., sorption to soil, presence of NAPLs);
continued leaching from source areas; and system
design parameters (e.g., pumping rate, screened
interval, and location of extraction wells). Phase
II of the study again recognizes the importance of
these factors as affecting the remediation of
ground water, and focuses on the role of NAPLs
in particular.
NAPLs have become a subject of special interest
for those involved in ground-water remediation.
In the early days of the Superfund program,
contaminants present in the subsurface in
immiscible form were not necessarily recognized
as a special threat to ground-water quality.
References can frequently be found in early site
investigation documents to "visible soil
contamination" or "soil staining," but these
conditions were not interpreted as evidence of
NAPLs.
With increased experience in the application of
pump and treat technology there has developed a
greater understanding of factors that may impede
progress in remediation of ground water. NAPLs
are now frequently identified as a key factor in the
longer-than-anticipated time frames for aquifer
restoration. NAPLs present in the subsurface act
as a residual source of ground-water contamination
that typically takes a very long time to deplete
solely by ground-water extraction. This is because
the aqueous solubility of NAPL-forming
compounds is a limiting factor; consequently,
large quantities of ground water must be pumped
to remove a significant quantity of the
contaminant. Even so, the solubility of many of
these compounds is much higher (e.g., 5 or 6
orders of magnitude) than their health-based water
quality criteria.
A more efficient way to deal with NAPL
contamination is to remove the contaminants in the
immiscible phase rather than in the dissolved
phase. To some extent this is practical for light
nonaqueous phase liquids (LNAPLs), which are
often found floating on the water table. However,
the physical behavior of dense nonaqueous phase
liquids (DNAPLs) makes them difficult to locate
and even more difficult to control, given the
current state of the science.
The emphasis on NAPLs in Phase II of the
ground-water extraction evaluation is intended to
be more empirical than theoretical. Several recent
studies present theoretical explanations and
observations on the behavior of NAPLs in
fractured and porous media (Feenstra and Cherry,
1988; Ruling and Weaver, 1991; Mercer and
Cohen, 1990). The intent in this evaluation was to
utilize field data obtained during actual
ground-water remediations to develop a sense of
the pervasiveness of the problem, and to illustrate
some of the special features associated with
NAPLs.
-------
Chapter 2
OVERVIEW OF CASE STUDIES
SITE BACKGROUND
CHARACTERISTICS
The Phase II study includes 24 case-study sites.
Table 2-1 provides a summary of their background
characteristics. The geographic distribution is
shown in Figure 2-1. The majority of the sites, 18
of 24, are east of the Mississippi River. Although
some consideration was given to geographic
distribution, the availability of performance data
for the ground-water extraction system was
considered a higher priority in site selection.
LENGTH OF HISTORICAL RECORD
Startup dates for the ground-water extraction
systems at the case-study sites range from 1974 to
December 1989.
The site with the longest record of ground-water
extraction is believed to be the Olin Corporation
facility in Brandenburg, Kentucky. Because the
process-water supply wells at Olin have gradually
evolved into a ground-water remediation system, it
is difficult to pinpoint the date when their use
became comparable to the extraction systems at
the other case-study sites. Olin's radial collector
wells have been in operation since the early 1950s,
but it was not until 1974 that their effectiveness in
controlling the spread of contaminated ground
water was recognized. In 1984 the wells were
specifically operated as part of a ground-water
remediation system. Performance data used here
to evaluate the effectiveness of the system begins
in 1984.
The IBM-Dayton site, which has the second
longest record of extraction for ground-water
remediation, began operation in 1978. After
6 years of operation, the system was shut down,
with the expectation that natural processes would
complete the restoration of the aquifer. Instead,
the contaminant plume began to expand again, and
in October 1990 ground-water extraction was
resumed.
At the Sylvester/Gilson Road Superfund site, a
ground-water extraction and recirculation system
was put into operation in December 1981. In
1982, the system was enclosed by a slurry wall.
For several years the slurry wall and pumping
system was used for containment only, and the
extracted ground water was reinjected without
treatment. In April 1986 a treatment system was
put into operation to remediate ground water
within the enclosure to alternate concentration
limits set forth in the Record of Decision (ROD).
PLUME SIZE
Table 2-1 lists the number of aquifers or aquifer
zones affected at each site and the thickness and
areal extent of the dissolved phase plume. This
information gives some indication of the relative
magnitudes of individual ground-water
contamination problems.
More than half (14 of 24, or 58%) of the sites are
listed as multi-aquifer remediation sites. In some
cases, the aquifers are composed of different
materials and have different water transmitting
properties but are not hydraulically separated.
These may more properly be considered as
separate aquifer zones. At other sites aquifer
materials are similar, but there is a significant
hydrologic distinction caused by intervening layers
of lower hydraulic conductivity. The
distinguishing feature at these sites is that separate
extraction and monitoring wells are dedicated to
individual aquifers or aquifer zones. All these
sites, therefore, require more complex extraction
and monitoring systems than might a single-aquifer
site with an otherwise similar magnitude of
contamination.
Plume thicknesses listed in Table 2-1 are estimated
maximum thicknesses of the contaminant plume.
The estimate includes the saturated thicknesses of
all aquifer zones and intervening layers between
zones. Estimates range from 20 feet for the
GenRad Corporation site to 365 feet at Tyson's
Dump.
Plume areas listed in Table 2-1 refer to the
estimated maximum lateral extent of the plume,
-------
Ponfers
IBM-San
Road
Rad
Engineering
IBM-Dayton
Blvar Slta
Harris Corp.
Sit* A
Figure 2-1
GEOGRAPHIC DISTRIBUTION OF
CASE-STUDY SITES
CASE STUDIES SUMMARY
CD
5
10
-------
TlUel-1
SITE IACKCROUND CHAKACTBJUSTK8
«*
N»
1
2
3
4
5
«
7
1
*
10
II
12
II
H
U
It
17
II
1»
20
21
22
23
24
NMM
Anabenol Corporation
Bl** * Decker
DaMotaaTCE
D» row Mot*
Eaenoi Ekdrtc
FuxUU SeafcoMtaciar
GaenlMiMi
GeafUd Corpora in
lUrri. CoroorMkxi
IBM-Dayton
IBM SM Jaw
Nichob Emi«tCTi«f
1HW { <*puralic4l
ftMOcn C'oner
•xvuuuli Rhd SMe
Sue A
Uuk rover * U|IH
VCIDM Wed FkU
VtteMenkr
Mid-SoMk Wood rraductt
Ocddeaul Cke^nl
SytKMet/OilKW Ro^
lymi'iDnap
Woura rnxxMtif
1 F«k,ID
BMUeQeet.MI
Qocticc, CMMta
Mc«i,AR
UUrop,CA
NMkw,NH
Ktaf of rnwto. r A
KOM.WA
DifcrraMm
IM3
1M5
1*74
H83
mi
mi
mi
IW4
I«K
l»77
l*7t
i«7
E«1y70>
1M1
mi
INS
IN)
INI
B«rty7t»
\9K
\m
1970
1*73
1N3
BdraellHi
WHi>4
JM.IW7
MjyINt
D«.IN7
Dec. IN5
DCC.IN4
JM.IN2
LMCIN5
LMCIN7'
A|X1I1W4
Mwc*l*7l
M>ylN2
Ju iiSS
1974
SejX. IN4
Stjx INS
Ailf. INi
Oct. INS
M*yl«M
1N>
EulylNS
JuelN2
Dec INI
Nov. INt
Jo^INt
AdHyMMrattvM
riifim
RCRA
RCRA
SupeitUd
RCRA
SMC
SMC
SUU
SIMC
SlMe * Sypafuod
SUM
Suit
SUie
Stale
SopcrfiuMl
Stale
CowMy
RCRA
SoperfiuMl
rmtaee
Supcrfuod
State
Superfiud
SaBafw^
SupciftyM
N&rf
A^tfcn
A&ctMl
2
2
1
1
1
3
3
1
2
2
5
2
1
2
4
1
3
2
2
1
1
2
1
1
PtMtArea
(Aero)
»
11
130
38
3
75
no
u
to
to
7«0
2
240
23
1,030
0.7
8-10
I2S
7,«0»
10-20
770
U
*5
14
rtmmt
IMctem
(»"<)
100
40
SO
30
SO
IM
SO
20
n
M
2SO
iOS
10
W
ISO
2S
175
120
09
170
190
110
3*S
63
Chntfcab •(&•!*»
FCE, TCB. CUaroCara
TCE,DCE,TCA.VC
TC8,DCE,VC
VOO, PaUdda
voa
TCA, DCE, Frco.
TCarCE,TCA
voa
voa
voa
voa, on
voa
DCEE.DC1FE
rC&TCE,DCE
rCB.TCE.TCA
voa
CteaKXe
voa
Or|»ict
rcr, CCA, crcoMu
reMtdda
VOO,Aneiiic
TCP. lytene, totaeae
Meuk,0r|ula
Fhniil wd (tacul liit. nod UKJ (nod
OhcW lid over (nctuced uodMone
Sill, tHd, cl»y, ud gnvd
AttwW Hud Md cby
Sud
AKinM unl ud (ravd witk till md ctoy bycra
GtacW drtn over rrMand ledimfMiiy rock
GtocM iHd, iiN, ud (ravd
SMd ud ikeH wNk d>y hyen
Sud UK) uH W ud ctoy Uyen
priCtwcH imie, ij«UiCa£, mSSumiOSE, wwi
cooftoBcnle
SM ud md wkk UUerteddcd ctey
GUcul und, >M, (ravd, ud cby
Uym* und, MH, nd cUy
Sud ud ooBUc tmatoae
AlkivhMi ud (racured bwk
GhcwJ wad ud (ravd over fractured undMoac
GtocW i»»d, (nvdi-MKl a«y ovci' (ractuml
MWlHCMie
Fracurcd mduone ud ikalc
luerbedded aHuvtel md, nil. ud ctay
OUcUl xad. (ravd, ud IM over fractured icMu
TMn i«d ud ctoy over fractured luduooe
Ulerbcddcd md, iW. ud ctay over Bcrftaw tiHy
uod
FOB - TcmcMoroewylcac
TCE - TrichlorodkylciK
TCA - 1,1,1-TricMaroclkue
VC - Vinyl chloride
DCE - Trim 1.2, DKhkxxxrChyknc
IX'KK - HKhkHMIhyl nher
IX IPI - l)Khk>niuopn>|iyl cllKi
P< f - CetiiHliloniptK-iHil
i t A - < h
-------
generally measured before the start of remediation.
Areas range from less than 1 acre at Site A to
7,600 acres at Ville Mercier.
Area estimates were made from concentration
contour maps, where these were available. At the
Mid-South Wood Products site no plume map was
available, so the estimate was based on the area of
contaminated soil. The actual plume area is
unknown and may not conform to the area of
contaminated soil. At the Emerson Electric site,
the contaminant plume was never delineated. The
extent of the plume was roughly estimated as a
circular area centered on the contaminated
monitoring well with a radius equal to the distance
to the nearest "clean" monitoring well.
CONTAMINANTS OF CONCERN
At 20 out of 24 sites, the primary contaminants
were volatile organic compounds (VOCs). Where
only a few specific compounds are significant,
they are listed by name. However, at many sites,
the number of VOCs is too great for individual
listing, and the generic abbreviation, VOCs, is
used. At a few sites, other organic compounds in
addition to VOCs are important, and the
contaminants are listed as "organics." Many of
these organic compounds are not miscible with
water, and therefore have the potential to be
present in the aquifers as a NAPL.
The case studies include two wood-treating sites,
with PAH compounds, and two pesticide sites.
Metals were significant contaminants of concern
only at the Western Processing site. The metals
involved were nickel, cadmium, zinc, chromium,
arsenic, copper, and lead. This site also had a
wide variety of organic contaminants.
GEOLOGIC ENVIRONMENTS
Various geologic environments are represented in
the collection of case studies. The sites with the
simplest hydrogeologic conditions are Emerson
Electric and Site A. In both cases, the zones with
known contamination were relatively uniform sand
of marine origin. It is interesting to note that
these two sites appear to have progressed most
rapidly toward aquifer restoration.
The remaining 22 sites had greater hydrogeologic
complexity than the two mentioned above. Many
of the aquifers are layered or interbedded
combinations of sand, silt, and clay deposited in
alluvial or glacial environments. Aquifer
restoration using extraction wells is less efficient
in heterogeneous formations than in more uniform
materials. Ground-water flow toward extraction
wells tends to take place mainly in the
higher-conductivity materials. Low-conductivity
zones, which may contain significant quantities of
contaminants, are largely bypassed. Difficulties in
remediating heterogeneous aquifers are presented
in the Phase I report (U.S. EPA, 1989), and
discussed in greater conceptual detail by Keely
(1989).
Nine sites involve contamination of fractured rock
aquifers, which are especially difficult to
remediate. The movement of ground water in
fractured rock takes place mainly in the fractures.
Usually, the fracture density is uneven, which
results in nonuniform, direction-dependent flow.
Fractured bedrock aquifers are especially difficult
to remediate as shown in several of the case study
sites, particularly at Black & Decker, Nichols
Engineering, Mid-South Wood Products, and
Tyson's Dump.
ISSUES RELATED TO SYSTEM
DESIGN
Table 2-2 summarizes information used to evaluate
extraction system design for the case study sites.
This information includes the number of
monitoring and extraction wells and the maximum
rate of extraction, which give an indication of the
level of effort expended to remediate ground water
at each site.
REMEDIAL OBJECTIVES
The remedial objectives generally have an
important influence on the design and operation of
the overall remedial system. Therefore, it is
important to recognize that the remedial objectives
for the extraction systems may differ for different
sites.
Aquifer restoration is a remedial goal for 17 out of
24 of the sites. Restoration is understood to mean
that the concentrations of contaminants in the
aquifer are to be reduced to levels that would
allow ground water to be used as it could have
been before being contaminated. The implicit
assumption is that, when these concentrations are
achieved, no further action aside from
-------
ground-water monitoring will be required.
Cleanup goals are usually maximum concentration
levels (MCLs) or other health-based criteria, such
as 10"6 excess cancer risk concentrations.
However, in some cases, alternative concentration
goals have been established on a site-specific
basis.
At the Amphenol site, the restoration goal is a
total VOC concentration of less than 5 ppb. The
General Mills site cleanup goal is 270 ppb for
trichloroethylene (TCE) in the shallow aquifer and
27 ppb in the underlying aquifers. At the
Sylvester/Gilson Road site, alternative
concentration limits for 16 key contaminants,
based on a site-specific risk assessment, are the
cleanup goals.
Remedial goals at the Savannah River Site are
similar to that for aquifer restoration but are
expressed in terms of reduction of contaminant
mass. Specifically, the goal is to remove 99
percent of the contaminant mass from affected
aquifers in 30 years. Goals stated in this manner
have proven to be a problem, because efforts to
quantify the mass of contaminants in the aquifers
using sampling data from monitoring wells have
produced highly variable results. Recent (February
1991) discovery of contaminants in NAPL form
will make accurate estimation of total contaminant
mass even more difficult.
The remedial objective at seven sites is plume
containment. This means ground-water quality
restoration is not expected within the site
containment area using the existing extraction
system.
At the Verona Well Field site, two separate, but
related, remediations are in progress. At the
Thomas Solvent Raymond Road (TSRR) source
area, aquifer restoration is being pursued. In the
well field itself, current remedial action includes a
system of blocking wells to contain the spread of
the plume and protect the remaining unaffected
wells. It may eventually be possible to
discontinue the blocking system if final remedial
actions are successful in all contaminant source
areas.
Remedial objectives at the Fairchild
Semiconductor site were changed from plume
containment to aquifer restoration in 1988 when
the remedial action at the site changed from an
interim remedial measure to a final remedial
action.
At the IBM-Dayton site, the goal was changed
from aquifer restoration to plume containment in
response to the determination that contaminants are
present in NAPL form. This change was reported
in Phase I.
Three case-study sites (Harris, IBM-Dayton, and
Ponders Corner) have well-head treatment systems.
At Ponders Corner, the extraction system consists
of two municipal production wells with treatment
for VOCs. These wells serve both to remediate
the aquifer and as a source of potable water. At
Harris and IBM-Dayton, separate ground-water
remediation systems have been installed in
contaminant source areas in addition to the
well-head treatment systems that are in operation
at down-gradient production wells.
PROJECTED CLEANUP TIME
At several aquifer restoration sites, extraction
system designers have predicted time frames
required to complete the remediation. For three of
the sites, the predicted cleanup time has passed.
Our experience with ground-water remediation and
the science involved in projecting remediation time
frames has progressed significantly since these
original estimates were made. These cleanup
timeframes were underestimated due, in part, to a
lack of knowledge of factors affecting ground-
water remediation. Projections made based on the
current understanding of fate and transport
processes and subsurface characteristics are
expected to be more representative.
At Site A, the cleanup was expected to be
complete in 60 days, but monitoring records show
that concentrations above the remedial goals were
still present at least 2 years after the onset of
pumping.
At Sylvester/Gilson Road, it was expected that the
alternate concentration levels (ACLs) would be
reached 1.7 years after the ground-water treatment
plant started operating in April 1986. However,
monitoring data show that 7 of the 16 ACL
compounds are still above the target concentrations
after more than 2 years. During this time, the
maximum concentration of one compound
-------
•Mfcl-2
SUMMARY or MSJCN-WOATED mro*MATioN
Hit
N».
1
2
3
4
5
«
7
ft
9
19
II
12
13
14
15
It
17
IS
1*
29
21
22
23
24
NMI
Amj*nwl ConnnttM
MKkADccter
OoMotooTCE
D«FaM-MoMe
Eaaeraon Ekxiric
(jCNUM MMM
GenRad Corporation
Harria Corporation
I8M Dayton
IBMSMtJoK
OBa Corporation
PoQdcn Cut MT
SmMa* Rfccr SMe
Sle A
Uta*. Tower * U»M
Vcram Wen FteU (TSRR)
(Wod*«1»Wlt)
VMc Maria
MM SDWb Wood fiDdnctt
Ooddewtii Ckealcil
^itKMC^OilKII RtMd
Tyicm'i Dui»p
Waicre PmctMinf
OtfKfln
RnlontkM
nCilonMMi
RwtortwOB
CoMriMMM
Ratantkx
ReMonMai
Ratomloa
Ratoratioi
RcMantnii1
CminlMiciil1
RcHonlioii
RMantiM
CMUJMCTK
Ratontloi1
MM
redwMm
Ra4oratkM
COMtlMMlM
RmamlM
Comii»iBciii
CoMilmioH
Rotontion
RaMMiM
Rcuomteii
Coiul«»c«l
RtUOrtlKXl
PI^KM
dm*
Ttat
SWyaci
None
Now
MM
f Mantta
1*M
Now
>5yan
NOK
N/A
rtOM
Uamuta
N/A
>l»ycm
»r«ri
«05yetn
NK«(
M«HHMl^
Wrffc
4*
54
4*
43
«
124
32
1«
I2S
-100
276
14
33
51
250
»
31
123
T
4*
«*
183
• 7S-SO
55
N*rf
Eili«cB.n
Wrfc
2
1
7
3
5
X
t
2
24
21
30
4
10
2
12
1
17
»
«
3
15
f
S
7
206
&>•.•••
IM>
(»-)
2M
10-15
ijn
180
39
»jee
310
46
310
1.000
«.ooo
65
*JOO
2.000
550
50
200
4M
ZOOO
750
42
too
300
120
200
J^MMfeff
Toll
ow
r
Y
Y
Y
N
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
N
Y
Y
7
N
Y
Y
Y
N
riM>
M*4H
(Y/N)
Y
N
Y
N
Y
Y
Y
Y
Y
Y
Y
N
Y
N
Y
Y
Y
Y
1
N
Y
Y
Y
Y
Tnml
Tta.
A»**
(YW)
Y
N
Y
N
Y
Y
N
N
N
Y
N
N
N
N
Y
Y
N
Y
7
N
Y
Y
N
n
We« potMi, rcmjectkm,
flurry mN
' RemedM*"* mctwic* well head ireMmeitl
H.M CSHH rrtrn 1" Ike DUMM Sotvetu Riynitid Koad lacitily « Ike Veroaa Wrll l-tfd We
o
10
rb
-------
(toluene) appears to be higher now than before
ground-water treatment began. Toluene is
suspected to be present in NAPL form.
Remediation at the Emerson Electric site was
judged to be complete in June 1987, after 2.5
years of ground-water extraction. However, the
original estimates were that the cleanup would take
only 9 months. It should be noted that the
determination of remedial effectiveness at Emerson
Electric was based on limited monitoring data
(samples mainly from extraction wells and little
post remediation monitoring). This site was
discussed in greater detail in Phase I (U.S. EPA,
1989).
At three other case-study sites, definite cleanup
times have been projected that have not yet
expired. Designers of the ground-water extraction
system at the Amphenol site predicted the aquifer
would be restored to the desired water quality in 5
to 10 years after system startup. This projection
was based on the estimated volume of the
dissolved contaminant plume and the assumption
that flushing several complete plume volumes of
ground water would exhaust the supply of
adsorbed contaminants. After approximately
4 years of extraction, it appears that ground-water
contaminant concentrations are being reduced at a
rate that may be consistent with projected cleanup
time.
At the Fairchild Semiconductor site, a projection
was made that aquifer restoration goals (achieving
drinking water standards in the ground water)
would be met in 1994. This estimate was based
on observed water quality improvement since
remedial extraction started in 1982.
Designers of the extraction system at the Savannah
River Site initially expected to achieve aquifer
restoration goals within 30 years of system startup.
In the six years of operation, both the system
design and the understanding of the contamination
problem have changed. No new cleanup time
projections have been made, but the system
operators now describe the 30-year timeframe as a
standard for evaluation of remedial progress rather
than a firm objective.
EXTRACTION AND MONITORING
SYSTEM SIZES
Table 2-2 lists the number of monitoring and
extraction wells at each site, as well as the
maximum ground-water extraction rate for all
wells. The number of extraction wells represents
the maximum because, at several of the sites, the
size of the system has changed over the years.
The Western Processing site has the largest
number of extraction wells with 203 well points
installed in rows coupled to a common vacuum
header. Each well point is designed to withdraw
at approximately 1 gpm.
The highest combined pumping for the case study
sites (9,200 gpm for 36 extraction wells) occurred
at the Fairchild Semiconductor site. The Fairchild
system operated at this rate for only a short time
in 1983. Since then, the extraction rate has been
steadily cut back in an effort to balance remedial
effectiveness with the need for water conservation.
Likewise, at the nearby IBM-San Jose site,
extraction rates have been reduced from 6,000
gpm (30 extraction wells) to less than 1,000 gpm.
The highest sustained extraction rate (6,200 gpm
for 10 extraction wells) is at the Olin Corporation
site, where the ground water is pumped from wells
and radial collectors next to, and extending under,
the Ohio River.
ENHANCEMENT TECHNOLOGIES
The term "enhancement technologies" as used here
and in Table 2-2 refers to remedial activities used
to augment or assist ground-water extraction in the
removal of subsurface contamination. These
technologies are not necessarily new or innovative.
Examples include:
• Soil vapor extraction can reduce cleanup
times by removing residual contaminant
sources in the vadose zone.
• Slurry wall containment can limit the
amount of water requiring treatment or
reduce the quantity of water pumped to
maintain a containment system.
• Re-injection of treated ground water can
increase hydraulic gradients and saturated
thickness in the aquifer being remediated,
and block plume movement.
-------
• Variations on the standard extraction well
design, such as French drains, enhanced
fracture zones, well-point systems, and
horizontal wells can increase extraction
effectiveness.
The most common enhancement technologies were
reinjection and soil vapor extraction, which are
used at six sites each. At the Fairchild
Semiconductor and IBM-San Jose sites, both
enhancement technologies are used. Treated
ground water was reinjected at the Harris
Corporation and Occidental Chemical sites.
However, it is not considered an enhancement
technology here because it was done solely for
disposal purposes, and the injection was to deep
aquifers that are not hydraulically connected to the
aquifers being remediated.
The next most frequently used enhancement is
slurry wall containment. The Sylvester/Gilson
Road and Western Processing sites are enveloped
by slurry walls that are not keyed into an
underlying aquitard. At Fairchild Semiconductor,
the most contaminated portion of the upper two
aquifers is enclosed by a slurry wall keyed into a
continuous low-permeability layer.
French drains are listed as an enhancement
technology at the Mid-South Wood Products site.
The extraction system at this site includes both
conventional wells and wells combined with
French drains. French drains were added to
provide an improved hydraulic connection to the
fractured rock aquifer.
A similar concern for improving hydraulic
communication in fractured rock exists at the
Black & Decker site. Here a single extraction
well was drilled in an artificially enhanced fracture
zone that was created using explosives.
Intermittent pumping is listed as an enhancement
at the GenRad site, because the extraction system
is turned off for three months every winter.
However, this is done more to prevent freezing
damage to the system than to improve the
efficiency of contaminant removal.
SYSTEM DESIGN INFORMATION
The remaining columns of Table 2-2 provide a
checklist of commonly used analytical techniques
for the design of extraction systems. Aquifer
testing is a basic method of determining the
hydraulic responsiveness of the aquifer to
pumping. It is believed to have been used at all
but four of the sites. (The Ville Mercier site was
not treated in detail in this phase of the study, and
information about the design techniques used there
was not available.)
At 18 sites, some form of ground-water flow
modeling was used to help select locations and
pumping rates of extraction wells. In most cases,
numerical or semi-analytical computer models
were used.
Some form of travel-time analysis was used for at
least ten of the sites. This analysis basically
consisted of estimating the time it would take for
the distant portion of the plume to be drawn into
an extraction well and was used as part of the
process for estimating restoration time frames.
Details of the analysis were usually not explained
in site documents obtained for this study. In a few
cases, particle tracking or streamline-generation
techniques were used to evaluate the flow of
ground water to extraction wells. In other cases,
travel-time estimates were based on comparisons
of the extraction well-pumping rate to the
estimated plume volume.
Numerical contaminant transport modeling appears
to have been used rarely.
Documents reviewed for 17 sites, explicitly
mention the importance of solute adsorption to
aquifer materials. However, it was not always
clear how this consideration was used in judging
the potential effectiveness of the extraction system.
In some cases, the estimated travel time for
contaminants to reach the extraction wells from
remote portions of the plume was increased to
account for adsorptive retardation. In other cases,
the estimate of total contaminant mass was
adjusted to account for adsorption. A third
common approach was to increase the estimated
number of pore volumes of ground water that
would have to be removed to complete the
remediation. The overall effect of all three
methods is roughly equivalent
OCCURRENCE OF NAPLs
The occurrence of contaminants as NAPLs at the
case-study sites was of special interest in Phase II.
In Phase I, NAPLs were identified as residual
10
-------
sources of contamination at the sites where they
were known or suspected to be present. However,
conditions associated with their presence and
reasons for suspecting it were not discussed in
detail.
Table 2-3 gives a summary of issues associated
with the presence of NAPLs at the case-study
sites.
At most sites studied, it was difficult to establish
NAPL presence conclusively. The exceptions
were sites where NAPLs had been directly
observed and reported in monitoring wells or soil
samples. Even though NAPLs are suspected, to
some extent, at 20 of the 24 sites, they have been
directly observed at only 8. Certain features of
NAPL behavior in the subsurface make it possible
for them to remain undetected by traditional site-
investigation procedures. This is especially true of
compounds having a density greater than that of
water (DNAPLs). Some of the more important
aspects of DNAPL behavior will be discussed
briefly in Chapter 4.
At nine sites, parties responsible for remediation
acknowledge that NAPLs are present. At three
others, it is acknowledged that NAPLs may be
present. Frequently, however, site information
contains clues indicating that NAPLs may be
present, even though this possibility was not
mentioned in site documents. This is not
surprising for older sites, because the issue of
NAPLs was not emphasized by the scientific
community until the early 1980s, and even now
the concept is relatively new.
Table 2-3 includes a column labeled "Likelihood
of NAPL Presence (1-5)". Entries in this column
give a rough quantification of NAPL likelihood on
a scale of 1 through 5. An entry of 1 indicate that
the site probably does not have NAPLs. A
likelihood of 5 was assigned only when NAPLs
have been directly observed or the parties
responsible for site remediation assert that they are
present. Entries of 2, 3, or 4 provide a range of
relative likelihoods between these extremes but do
not have precise definitions in terms of
quantitative site data.
Table 2-3 also lists several types of evidence that
were used to judge the relative likelihood of
NAPL presence. The most conclusive is direct
observation. Less conclusive, but still suggestive,
clues include the following observations:
• High concentrations in the ground water
compared to the aqueous solubilities of
the compounds,
• Unusually deep (for DNAPLs) or shallow
(for LNAPLs) concentration distributions
that do not seem to be attributable to
other hydrogeologic influences,
• Persistence of contamination in spite of
the remediation efforts,
• Source characteristics or methods of
waste disposal that would be likely to
result in the presence of NAPLs in the
aquifer.
All of these circumstantial clues are relative, being
more or less persuasive depending on the degree to
which they appear. Ground-water concentrations
greater than 100 percent of solubility, for instance,
would be considered very strong evidence of
NAPL presence, whereas concentrations in the
range of 1 to 5 percent give a questionable
indication. The relative and cumulative nature of
these clues were considered in assigning likelihood
scores in the range of 1 through 4 in Table 2-3.
To provide insight into the importance of
ground-water concentrations as an indicator of
NAPL presence, a column has been included in
Table 2-3 listing the highest reported concentration
as a percentage of aqueous solubility for each site.
In several cases, concentrations greater than
100 percent of aqueous solubility were reported in
ground-water samples. Although the co-solvent
effect is a possible explanation, this is most likely
to be an indication that the compound was present
in the sample in colloidal-size NAPL droplets. It
was considered strong evidence for NAPL
presence in the aquifer.
The final column in the table lists the chemical
species for which the relative concentration in the
preceding column was reported.
11
-------
to
T«Ur2)
SUMMARY OF NAFt OCCUMONCE
1
2
3
4
5
t
7
1
»
10
11
12
13
14
15
It
17
IS
1*
20
21
22
23
24
*PCE b diHofccd
1.1,1-TCA
IA4.TC8 -
CCI4
DBCP
DCEE
IK'ICE
PAIU
Iff
IH 1
M.» IW MM
NHH
Aafhaol Cocpocitlo.
BUrt«Dect«r
DaMotaaTCE
D« PoM-MeMe
EBOIM Ehcufc
GamlMfc
GcnR*d Carponliaii
HWTV GmpofttiM
IBM-Dipon
IBM SM Jot
NkAoh Ej^Kcrim
Otw Corporal »•
Foadcn CoTMf
ln«i>BhrrfllT
StteA
uuk POMT * UIM
VaiMWdlFkW
VMeMader
MU South Wood rrodncu
OccMcMll Omntc.1
SyKouiJGikc. Ro«l
Ty«.'.DMV
Waunrnx^tat
In MtoOK, >nd tkcnfere noM.
I.I.I TrkMnaelkMe
IA4 TrtcWontaatiM
Cifbon TcHvcMotMc
Mycydic i«o«Mllc kydnxartiani
1 i 1 TodUxofHT.f.1*
NAPL
rn-nc*
N
N
Mivdc
N
N
N
N
N
N
Y
Y
N
kbyte
N
Y
N
Y
Y
Y
Y
N
ktoyte
Y
Y
*•*> 0< Ike pwt i
NAM.
Piwne*
(I-S)
1
3
3
3
2
4
3
2
2
5
3
1
3
2
3
1
5
5
5
5
1
4
3
3
LNATUw
DNAPU
DNAPU
DMATU
DNATU
LNATLl
DNAfU
DNAfU
DNAfU
DNAfU
DNAfU
LNAfU
DNAfU
DNAfU
DNAfU
DNAfU
LNAfU
Botk
Bo*
LNAfU
DNAfU
Bock
Mrtn «f *• tS**mm fcc NATU
C^.
X
X
X
X
X
X
X
X
r^pi.nirt • COMM 1W
02
0.5
13
4.1
>*
006
37
3.2
22
0.1
>100
11.3
1
>IOO
03
2t
74
45
TCE
TCE
TCE
1A4TCB .
Toktaie
1,1.1-TCA
TCB
TCE
TCE
fCE
Fmw 113
H-14
WlfF
rre
TCE
T-U-DCE
fAlb
fCE«
1
rcr
DBCP
TakKiM
TCP
DCS
s
-------
Chapter 3
SUMMARY OF REMEDIAL PROGRESS
In the first phase of the study, four conclusions
were reached concerning the remedial performance
of the extraction systems at the sites studied.
1. The ground-water extraction systems at most
of the sites studied appeared to be achieving
hydraulic containment of aqueous
contaminant plumes,
2. Most of the extraction systems were
removing, or had removed, substantial
quantities of contaminants.
3. When extraction systems are first turned on,
contaminant concentrations usually show a
rapid initial decrease, but then tend to level
off or decrease at a greatly reduced rate.
4. The data collected prior to system design
and during operation were often not
adequate to fully assess contaminant
movement and the response of the ground-
water system to extraction.
The information obtained in the second phase of
the study seems to generally confirm these
conclusions. However, the collection of case
studies contains enough variety to provide
exceptions to each of the general conclusions.
Table 3-1 gives a concise summary of the major
indicators of remediation effectiveness at the
case-study sites.
PLUME CONTAINMENT
As shown in Table 3-1, containment of the
aqueous plume appeared to be effective at 15 of
the 24 sites. This judgment was made by
comparing the extent of known ground-water
contamination, based on ground-water monitoring,
to the capture zone of the extraction system, as
indicated by water-level measurements in
monitoring wells and piezometers.
Containment effectiveness is listed as uncertain for
six sites in Table 3-1. In each case, this is
because of a lack of enough site data on which to
base a firm determination. At the Du Font-Mobile
and General Mills sites, the delineation of capture
zones was uncertain, because there were not
enough piezometers on the downgradient side of
the extraction wells. At Emerson Electric, Site A,
and Mid-South Wood Products, both the extent of
the ground-water contamination and the hydraulic
effects of the extraction system were unclear. At
Utah Power & Light, the main problem was the
difficulty in determining the boundary of the
contaminant plume in the fractured rock aquifer.
It should be emphasized that both water-level and
water-quality measurements are required to
demonstrate that the extraction system is
effectively containing the aqueous contaminant
plume. Observation of water-quality trends alone
is insufficient.
At three of the case-study sites, the available data
indicated that hydraulic containment was not
completely effective. These sites were the
Savannah River Site, Sylvester/Gilson Road, and
Tyson's Dump.
At the Savannah River Site, the contaminant
plume extends beyond the zone of influence of the
recovery system. In the Phase I report, it was
noted that the capacity of the recovery system did
not seem to be commensurate with the magnitude
of the contamination problem. Since 1988, the
total pumping rate for the system has been
increased from an average of 436 gpm to as much
as 550 gpm. In addition, new extraction wells
have been installed in areas that were not
previously being remediated. However, it appears
that there are still portions of the plume that are
not being captured. Also, downward migration of
the plume has not been completely reversed.
At the Sylvester/Gilson Road site, the extraction
rates apparently have not been high enough to
maintain inward gradients around the entire
periphery of the slurry wall that encloses the site.
In addition, observation of vertical gradients within
the enclosed area indicate that contaminants may
be escaping by vertical migration to lower aquifer
zones. In response to these observations, new
extraction wells are to be added to the
ground-water recirculation system inside the slurry
wall. Also, consideration is being given to
13
-------
MtN*
1
2
3
4
S
«
7
1
*
10
11
U
1)
14
15
16
17
U
1*
20
21
22
23
24
N***t
Aat^iticaol Corporation)
Stock* Decker
DaMcwaTCE
DH PoM-Mobile
E*Kno* Electric
General MNh
Ge*Rad CorfccMfc*
rUUTit CvOCpOCVlMM
IBMDiyto*
IBMSmJoie
N.CM. Bnimeerai,
(Mm CorponlioB
Pomien Career
Sr»«n.ili RJvcr Srte
SHeA
UukfeKrAUfkl
Venom WeH FleU
VMc Mender
MU-9eMk Wood PnxJucU
lywM-Dw^
•Mm of Ote iwfecMor ccwpowKl only. Umttx
^Coftccntraliom «rrr reduced Mow ihc dm
cCk»j>up foal w Ike ntaeoce of other cooum
dl-rv.f>utc«l MCI,, not expbcttly specified u 1 1
C1SHK celen lu the Ifeurou Solvent Rjywoo
^j^"*
BflMlM
Efctta
Elfccifce
Unecrtti*
UlKCftti.
Eflectfw
UKcrUta
Effective
ERective
Effective
Eflccuve
BOeciivc
Eflectrw
Effective
fMW
U-cemi.
UMenalu
Effect**
Eflectlve
UwottHi
EOoctwe
tatW
PtrtM
Eff«ctt«
,^7^
Notpvea
Not five*
50-W
Natfwca
Notftvc.
Napx*
100,000
Not fives
Not (we*
Not five.
Not five*
No, five.
Not |ive«
1300
4*4flOO
NotfMII
Not five.
Not tinea
Not five.
Nc4five.
Not five*
Not five.
Not five.
Not five.
MJMMA
(*•)
Not five.
150*
ts,Me
1.4M
12*
»,000
2.0001
10*
U&O
Not inc.
>Mioir
Not five.
Not five.
l.TflO*
203,7*2
-IS
Not ffVCM
14.000 (TSRR)e
not fivcM
Not five.
-5,000"
Not five.
lOlflOO
Not ftwui
C^jTmU
ToUJVOO
TCE
TCE
TOX
MIBK
TCA
TCE
TCE
Total VOGi
TCA
Free. 113
CC14
DOTE
PCE
TCE
T-U-DCE
ToUirAffa
ToulVOOl
U-OCA
rcr
DBCP
Tokme
1J3-TCP
TCE
.s=
32)
»*,OOB
M*7
10.450
90,000
I.909.000
2300
>5,000
37,120
»4W
U.OOO
we
«2,000
4**
223.009
7^00
58,000
K.960
11,500
*5.000
4JOO
140.000
1.400,000
210,000
£1
Nolfnca
730B
iaoo
Not |jven
150
11.000
1,200
31
*\aio
200
Nolfivei
18
Not |ivcn
492
33.000
7.2W
0
19.008
11,500
15,000
4J80
17^08
340.000
13.000
CMKutnNM
30
MO
MB
None
rW
1.000
100
250
1,100
39
7
100
-4.009
50
None
None
V»M.e
2400
1,000
None
20
10.000
50,000
5.0W
•£±/
Weil ID
MtaeM
hAieiN
N/A
N/A
RW28
WdlO
G-ltl 14A
MtaeM
SB II
7
MW-1
RW-I
We»H2
lifluenl
InfhKiil
InBucnl
bribe*
Influent
N/A
Ind-en.
M*M
btOueiit
cais
rtnttn*.
1.7 yn
iyr
0.4 yn
N/A
N/A
Ino
0.5 yn
3.5 yn
3Jo yn
4 yn
7
0.8 yn
7
IP
N/A
N/A
N/A
0.3 vn
0.7 yn
N/A
.yn
0.7 yn
2yn
IP
5
5
5
N/A
Detection
200*
27
5
N/A
N/A
N/A
5
N/A
N/A
N/A
70
N/A
N/A
N/A
200"
N/A
2,900
N/A
Not five*
KKXcd eMria in Ik* colwao are «or loul COMMMMM mu*.
wp criterii ta ki» MUM 3 yem.
o«ir«lbm«c
J ROM) facility 11 UK Verona Weil BeM tile.
I
-------
increasing the rate at which treated ground water is
removed from recirculation and discharged outside
the wall.
The interim ground-water extraction system at the
Tyson's Dump site captures most, but not all, of
the solute plume. That is all that the interim
extraction system was intended to do. An
expanded extraction system designed for complete
hydraulic control of the solute plume is presently
in the final stages of construction.
Table 3-1 lists containment effectiveness as
uncertain at six of the case-study sites. This
judgment was generally reached because of lack of
information about either the extent of the
contaminant plume or the hydrodynamic conditions
generated by the extraction system.
At the Emerson Electric site, it was impossible to
judge the containment effectiveness of the
extraction system, because the extent of the
contaminant plume was never defined and no
water-level measurements were reported.
Containment effectiveness is listed as uncertain for
the General Mills site, because the site documents
that were obtained for review do not show enough
water-level measuring points to support accurate
delineation of the capture zone. The extraction
system operators assert that the portions of the
plume that exceed the cleanup levels are being
captured. In the lower of the two aquifers being
remediated, however, concentrations in excess of
the cleanup standards are consistently being
detected at monitoring wells that appear to be
outside the likely zone of influence of the
extraction well.
At Site A, no water-level measurements have been
presented to support the contention that the
recovery well has captured the contaminant plume.
The hydraulic design of the extraction well seems
to have been based on the assumption of
horizontal radial flow to the well. There is no
indication that the effects of vertical flow or
regional gradients were considered. The only
indication of the actual hydrodynamic performance
of the well is the assertion, appearing in the early
performance reports, that no drawdown could be
detected in any of the monitoring wells when the
extraction system was turned on.
Table 3-1 lists the containment effectiveness at the
Utah Power & Light site as uncertain, because the
available site data do not support an accurate
delineation of the extent of the contaminant plume
in all aquifer zones. Water-level measurement
data indicate that the extraction system's capture
zone does include all of the monitoring wells
currently reported to be contaminated.
Capture effectiveness at the Mid-South Wood
Products site is listed as uncertain, because the
available site documentation does not indicate
either the extent of the contaminant plume or the
ground-water flow patterns induced in the aquifer
by the extraction system.
At the Du Pont-Mobile site, no new information
on the hydrodynamic effects of the extraction
system has been obtained since the first phase of
the study. The potentiometric surface map
presented in the first case study appeared to show
that the contaminant plume was entirely captured.
However, some uncertainty was cast on this
conclusion by some rough, contaminant mass-
balance calculations, which indicated that only
about half of the contamination approaching the
line of extraction wells could be accounted for by
the concentrations and flow rates being extracted.
It is likely that the discrepancy is due to
inaccuracies in estimating the contaminant flux in
the plume. But, it is also possible that
contaminants are bypassing the extraction wells in
the deeper portions of the aquifer that are not
monitored.
AQUIFER RESTORATION
EFFECTIVENESS
CONTAMINANT MASS REMOVAL
Table 3-1 lists the available estimates of the initial
contaminant mass and the mass removed to date
for each of the case-study sites. In most cases, no
estimate was available for the initial mass in place.
Where such estimates were available, they have
often been proven wrong.
For instance, at the Des Moines TCE site, the
representatives of the responsible parties have, in
the past, contended that the ground-water
contamination was caused by the former practice
of pouring contaminated sludges on the ground for
dust control at the rate of 100 to 200 gallons per
15
-------
year. Considering the maximum measured TCE
concentration in the sludge and the period of years
over which this practice was followed, the total
mass of TCE disposed of can be estimated at 50 to
90 pounds. Influent concentration records for the
ground-water treatment plant, however, show that
more than 9,100 pounds of TCE were removed
from the aquifer during the first 9 months of
operation. By the end of 1989, the total mass of
TCE removed was estimated at 15,860 pounds.
This discrepancy highlights the obvious uncertainty
about the true nature of the contaminant source.
Another example is the Savannah River Site,
where the initial estimate of contaminant mass in
place was based on volume integration of the
contaminant concentrations in the plume, as
measured by the extensive ground-water
monitoring system. This calculation of
contaminant mass in place has been repeated
regularly at the Savannah River Site and used in
conjunction with estimates of the quantity of
contaminants removed as a means of monitoring
the performance of the extraction system.
Experience with this procedure has shown that the
quantity of contamination actually removed is
greater than the estimated change in contaminant
mass in the aquifer. The recent discovery of
DNAPL contamination in the aquifer provides one
explanation for this discrepancy, and highlights the
difficulty of determining what is actually present
in the subsurface.
The estimates given in Table 3-1 for the mass of
contaminants removed by the extraction system are
expected to be more reliable than the estimates of
mass in place. In some cases, the estimates of
mass removal were made by the parties
responsible for the remediations. However, most
of the estimates were made as part of this study,
and were based on treatment plant influent records
presented in the case studies.
The quantity of contaminant mass removed is
presented in Table 3-1 as a measure of extraction
system performance because it represents an
accomplishment that can be attributed
unambiguously to the system. In effect, however,
it seems to be a measure more closely associated
with the magnitude of the problem than with the
degree of remedial success. The two case-study
sites that appear to be closest to successful
remediation, Emerson Electric and Site A, are
among the three with the lowest estimates of
contaminant mass removed.
CONTAMINANT CONCENTRATIONS
The last seven columns of Table 3-1 deal with the
concentration of an indicator compound (or a
composite contamination parameter) that has been
selected for each site to serve the illustrative
purposes of the table. In many cases, the indicator
compound is also the primary contaminant at the
site, but this is not true in every case. The
indicator compounds were chosen only for use in
this table and have no official status as indicators
at the sites themselves.
The maximum reported concentration of the
indicator compound, as listed in Table 3-1, was
culled from the site records made available for this
study. In a few cases, a nearly complete record of
site-monitoring data was available, and was used
to identify the highest reported concentration.
More often, the concentration listed was the
highest mentioned either in text, tables, or figures
in a remedial investigation report for the site.
Higher concentrations may have been measured,
but were not reported in the documents made
available for the case study.
At three sites-Fairchild Semiconductor, Utah
Power & Light, and Mid-South Wood
Products—the maximum reported concentration
exceeded the aqueous solubility of the indicator
compound. These occurrences were taken as a
strong indication of the presence of NAPLs.
At three other sites-Ville Mercier, Occidental
Chemical, and Site A~the maximum reported
concentration was the initial recovery
concentration. For the first two sites, this was
probably because the information collected for the
case studies consisted mostly of summary reports.
If a complete data base of ground-water
monitoring results had been available, higher
concentrations than those reported for the
treatment plant influent probably would have been
found. At Site A, however, a fairly complete
record of site-monitoring data was made available,
and none of the reported monitoring-well
concentrations were as high as the initial
concentration reported in the treatment plant
influent.
16
-------
At several sites (particularly Utah Power & Light,
Nichols Engineering, and GenRad) the initial
recovery concentrations were very low. This is
simply the result of the extraction wells initially
not being in a high concentration portion of the
plume. In each case, however, the extraction wells
later produced at higher concentrations as
contaminants were drawn toward them.
One of the general conclusions drawn in the first
phase of the study was that ground-water
extraction frequently produces a rapid initial drop
in concentration and then levels out to relatively
constant, or slowly declining, contaminant levels.
This leveling out in concentration reduction can
result from a number of factors and can, in fact,
reflect progress in cleaning up a plume. Before
any conclusions can be drawn from looking at
concentration reduction trends, a thorough review
of extraction system design should be performed.
This was not done as part of this study;
consequently, it is not possible to determine if the
plateaus observed and described in the following
paragraphs reflect a true limitation or inefficient
design of the extraction system or sampling that
does not represent the full impact of remediation.
In most cases, the latter two occurrences are
associated with dilution of contaminant
concentrations at the monitoring point through one
of the following mechanisms:
• Selective pumping of wells in less
contaminated areas at relatively high flow
rates.
• The well is constructed such that the
water table is quickly lowered below the
contaminated zone.
• The outer edge of the plume is cleaned
up and individual monitoring wells reflect
a continuing decline in contaminant
concentrations, yet monitoring samples
are taken from a point at which ground
water from all extraction wells is
combined.
This dilution of samples with surrounding clean
ground water can mask the fact that ground-water
is being cleaned up.
Assuming the above design limitations have been
addressed, the occurrence of a stabilized
concentration can indicate that the clean up of the
affected portion of the aquifer is limited by the
kinetics of contaminant desorption or dissolution.
This could be due to the release of contaminants
from a residual source, such as adsorbed
contaminants or a NAPL.
As shown in Table 3-1, an attempt was made to
identify such a concentration plateau at each
case-study site. This identification was entirely a
matter of perception, and required the application
of subjective judgment as to what constituted a
stabilized concentration and what did not. No
precise mathematical or statistical definition of the
stabilized concentration was used. Where leveling
out of the concentration record was noted, it
frequently occurred in only part of the contaminant
plume, or in the extraction wells. The
identification of a stabilized concentration in
Table 3-1 does not constitute a prediction that it
will persist for a very long time. In several
instances in which concentrations seemed to have
stabilized in the first phase case studies, additional
data gathered in the second phase showed
concentration reductions. Examples of this are
Monitoring Well 1-S at the Amephenol site and
Extraction Well ERW-8 at Des Moines TCE (see
Case Studies 1 and 3).
No concentration plateau could be identified for
some of the sites. At the Du Font-Mobile and
Utah Power & Light sites, the concentration
records showed too much variability for plateau
identification. At the Emerson Electric site, the
initial high rate of concentration reduction was not
maintained, but the concentrations did continue to
decline steadily until the cleanup goals were
reached. The concentration records at Site A
showed a similar pattern, at least for the selected
indicator compound. At the Savannah River Site,
the record of concentration in the treatment plant
influent showed a slow, but fairly continuous,
downward trend with no obvious leveling off. At
the Mid-South Wood Products site, the
performance record was too short and available
data too limited for trends to be identified.
The initial effort at each site to identify a
concentration plateau focused on the treatment
plant influent record since this allowed the
selection of a single point for each site. (This
does not mean that treatment plant influent
concentrations are the best measurement of pump
and treat performance as indicated above.) In
10 cases, concentration plateaus were found in the
17
-------
influent records. Where plateaus could not be
found in the influent records, either because the
records were unavailable or the concentrations did
not level off, records from individual wells were
scrutinized. Several instances of leveling off were
identified in monitoring wells or individual
extraction wells, even though concentrations were
not stable for the extraction system as a whole.
Table 3-1 also lists both the approximate time after
the beginning of extraction when the concentration
plateau was reached and the remediation goal for
the indicator compound, where applicable. The
cleanup goals for the indicator compounds were
listed here primarily for comparison with the
maximum, initial recovery and plateau
concentrations. The cleanup goal is listed as
"N/A" (not applicable) when the indicator
compound is total VOCs or the remedial goal at
the site is containment rather than aquifer
restoration. For the IBM-San Jose site, the
cleanup goal for Freon 113 is different for the
different aquifers, and because no point
measurements of concentration are listed, there is
nothing to compare. At the Western Processing
site, the cleanup goal for TCE is expected to be an
ACL that is to be specified in a future record of
decision. For the Savannah River Site, the
remediation goals have not been specified in terms
of cleanup levels.
PLUME AREA REDUCTION
Reduction of the area of a contaminated aquifer is
an alternative measure of restoration effectiveness.
Remedial progress is less commonly viewed in this
way, probably because this kind of evaluation
requires mapping of contaminant concentration
values, usually in the form of concentration
isopleths. One reason such maps are not more
widely relied on to evaluate remedial progress is
that their construction makes use of interpolated
concentrations between the monitoring wells.
Consequently, the maps are partly the result of
interpretation as opposed to being a direct
measurement. The advantage of this format,
however, is that it shows the extent to which
aquifer restoration has been partially achieved.
Plume mapping is usually done as part of the
remedial investigation, but is less commonly
encountered in status reports for operating
remedial systems. At nine of the case-study sites,
contaminant plume maps were available in the
operational data reports and have been presented in
the case studies. In each of these cases, the maps
demonstrated that the area of groundwater
contamination has been reduced for some or all of
the contaminants of concern. The sites for which
plume maps were produced are:
• Fairchild Semiconductor
• GenRad Corporation
• Harris Corporation
• IBM-Dayton
• IBM-San Jose
• Nichols Engineering
• Occidental Chemical
• Sylvester/Gilson Road
- Verona Well Field
• Western Processing
Of these sites, Fairchild Semiconductor and IBM-
San Jose show the most marked reduction in
plume size.
USE AND EFFECTIVENESS OF
ENHANCEMENT TECHNOLOGIES
Supplemental remediation techniques that are
being used in addition to basic ground-water
extraction at the case-study sites, and that have the
potential to improve the ground-water remediation,
are referred to here as enhancement technologies.
These techniques are not necessarily new or
innovative. Various enhancement technologies,
including soil vapor extraction, reinjection, and
slurry wall containment are being used at several
sites. In addition, fracture enhancement was used
at the Black & Decker site. Many other
enhancement technologies are available that were
not used at the case-study sites.
SOIL VAPOR EXTRACTION
At the Fairchild Semiconductor site, soil vapor
extraction was begun as a pilot system in October
1988 and was expanded to full scale in January
1989. The system consists of 32 vapor wells
installed in the dewatered upper aquifer, in the
partially dewatered underlying aquifer, and in the
aquitard layer that separates them. Eight air inlet
wells have also been installed to facilitate vapor
sweeping in the deeper zones. By September
1990, after approximately 1 year of operation, the
system had removed 14,700 pounds of VOCs.
18
-------
contaminants is large, compared to
total of 1,500 pounds recovered in i
A pilot scale system was tested in five separate
areas of the IBM-San Jose site in 1990. Both
LNAPL petroleum hydrocarbons and VOCs were
successfully recovered. As a result of the test, a
full-scale vapor extraction system was planned for
the site.
At Ponders Corner, a vapor extraction system was
installed around the contaminant source area in
December 1987. The system consisted of
10 vertical wells and 3 horizontal vapor extraction
headers. When the system was turned on in
March 1988, it recovered tetrachloroethylene
(PCE) at a rate much higher than had been
foreseen by the designers of the vapor treatment
system. Consequently, the system operated only
intermittently, with interruptions for replacement
of the treatment system's activated carbon. During
the first month of operation, the system removed
360 pounds of PCE from the soil. Operation of
the system was permanently discontinued in April
1988, by which time it had recovered an estimated
775 pounds of PCE. Even though this mass of
the estimated
5 years by the
ground-water extraction system, it had no obvious
effect on the PCE concentrations pumped by the
extraction wells.
At the Savannah River Site, a pilot scale vapor
extraction system was tested in 1990. It recovered
a total of approximately 1,500 pounds of
contaminants in 3 weeks. As a result of the test, a
full-scale system has been proposed.
A soil vapor extraction system has been in
operation in the Thomas Solvent Raymond Road
portion of the Verona Well Field site since 1987.
It consists of 23 PVC wells of 2-inch and 4-inch
diameter. After approximately 1 year of operation,
the system had removed an estimated
45,000 pounds of VOCs from the vadose zone.
However, the rate of removal had fallen off to less
than 10 pounds per day, and the soil remediation
goals had not been met. Several reasons were put
forward by the system operators to explain this.
They included many of the same effects that
impede the restoration of aquifers by ground-water
extraction systems. For instance, it was pointed
out that VOC concentrations had been reduced to
low levels in the soil vapor so that continued
pumping resulted in low rates of mass extraction.
The rate of mass transfer from residual LNAPL
globules to the surrounding soil vapor had
apparently been reduced because the smaller
globules, with their greater ratios of surface area to
volume, had been exhausted. Also, the
concentrations of volatile constituents within the
residual LNAPL were reduced, so that they
volatilized at lower rates. Finally, it was pointed
out that the majority of the vapor flow in the
vadose zone was following preferential flow paths,
a situation that was exacerbated by the desiccation
of the soil in these areas.
GROUND-WATER REINJECTION
Reinjection has been tried at several of the
case-study sites. As reported in the first phase of
the study, reinjection wells were used briefly at the
IBM-Dayton site until their effectiveness was
destroyed by clogging. The new extraction system
at IBM-Dayton uses spray irrigation as a form of
ground-water reinjection primarily to dispose of
the treated water from the extraction wells.
However, the spray field is upgradient of the
contaminated portion of the unconfined aquifer and
may also increase the rate of groundwater flow
toward the extraction wells.
Reinjection is also being used at the
Sylvester/Gilson Road site and at Western
Processing. In both cases, the reinjection is
through trenches rather than wells. Even so, there
were problems with clogging of the trenches due
to iron precipitation at Sylvester/Gilson Road. The
ground-water treatment system, which came on
line in 1986, includes iron removal, and no further
problems with iron clogging are expected. At
Western Processing, there has been no indication
of any problems with the reinjection trenches.
At Fairchild Semiconductor, a system of
reinjection wells was put into operation in
September 1990. Their installation was preceded
by pilot testing, which apparently indicated
success. As yet, there has been no indication of
the success of the full-scale system. A system of
reinjection wells is also planned for the nearby
IBM-San Jose site.
SLURRY WALL CONTAINMENT
Slurry walls are being used at the Fairchild
Semiconductor, Sylvester/Gilson Road, and
Western Processing sites. At Fairchild, the wall
was constructed only around the most highly
contaminated portion of the plume to isolate the
19
-------
source areas. It was completed through the two
uppermost aquifers and keyed into a continuous
clay layer. Ground-water extraction within the
wall has resulted in significant aquifer dewatering,
and permitted soil vapor extraction to be
conducted at depths below the normal water table.
At Sylvester/Gilson Road, the slurry wall encloses
nearly the entire contaminant plume, but it is not
keyed into a continuous underlying aquitard.
Consequently, the containment effectiveness of the
wall is highly dependent on maintenance of inward
hydraulic gradients. It appears that the rate of net
ground-water withdrawal in the enclosed area has
not been high enough to produce inward gradients
everywhere. Contamination is also thought to be
escaping by vertical flow into the underlying
bedrock in the interior portion of the site.
The slurry wall at the Western Processing site also
depends heavily on hydraulic gradient control for
its effectiveness. A fairly elaborate gradient-
monitoring system is used to ensure that
contamination does not escape under the wall.
This system is probably effective for dissolved
constituents. However, if DNAPLs were present,
which has not been shown to be the case, the
gradients being maintained across the wall
probably would not prevent their migration to
lower aquifer zones.
FRACTURE ENHANCEMENT
The ground-water extraction system at the Black &
Decker site uses an artificially created, enhanced
fracture zone to improve the effectiveness of
extraction. It is probably due to this zone that the
system seems to provide effective plume
containment. DNAPLs have not been shown to be
present at this site. However, if they were present,
the deep fracturing produced when the enhanced
fracture zone was created might permit them to
penetrate more deeply into the bedrock than they
would otherwise have done.
UPDATE ON SITE DATA
REQUIREMENTS
In the summary of the first phase of the study,
considerable attention was paid to the types of site
information necessary for the design and operation
of effective ground-water extraction systems. The
emphasis in that discussion was on the design and
operation of systems to control and remediate
plumes of dissolved contaminants. The new case
studies and updates developed in the second phase
tend to reinforce the observations made in the
original study. In addition, with the present
emphasis on NAPLs, some of the information
requirements take on new importance.
It has been observed that a conclusive
determination of the presence or absence of
NAPLs is often difficult. At many of the sites
where circumstantial evidence suggested that
NAPLs were likely to be present, no confirmation
in the form of direct observation has been
forthcoming. Considering the important
implications that the occurrence of NAPLs can
have, it is obviously desirable to obtain the site
data that would be most helpful in reaching the
correct determination.
In this section, examples will be selected from the
new site information gathered in the second phase
of the study, to illustrate the importance of various
types of field data with respect to the selection and
design of ground-water remediation systems. The
usefulness of this information in the search for
NAPLs will also be discussed.
HYDROGEOLOGIC INFORMATION
Stratigraphy
For design of a successful ground-water
remediation system, it is important to know the
number of aquifers involved and the degree of
hydraulic interconnection between them. At
several of the case-study sites, the ground-water
extraction system was installed in more than one
aquifer. At sites like Fairchild Semiconductor and
IBM-San Jose, the sand and gravel aquifers were
clearly separated by layers of silt and clay,
although these layers were not always continuous.
At these sites, each of the contaminated aquifers
had its own set of extraction wells, which could be
operated more or less independently.
The contaminant plume at the Occidental Chemical
site occurs in an upper aquifer that is divided into
three permeable zones with partial hydraulic
interconnection. Here, some of the extraction
wells are screened in more than one zone, and
pumping from a well hi one zone creates hydraulic
gradients in other zones as well. Even so,
distinctions can be made between the behavior of
20
-------
the plume in the separate zones. For instance,
incomplete hydraulic containment has been
observed in the deep zone during certain periods
of high-volume pumping from nearby water-supply
wells. This can only be distinguished because the
monitoring system has been designed to permit
observations in the individual zones. In contrast,
the monitoring system at the Du Font-Mobile site
has not been designed to distinguish between
zones. As a result, there are persistent questions
about the effectiveness of plume capture.
Stratigraphy may also influence the movement and
detectability of DNAPLs. When downward
moving DNAPLs arrive at a layer of
lower-permeability material, they may be unable to
penetrate them. The nonaqueous liquid may then
pool on top of the low permeability layer and flow
laterally in the direction of dip. If the stratigraphic
information identifies such a situation, it may
provide an opportunity to sample the DNAPL and
perhaps even to control its migration.
Aquifer Properties
An understanding of the hydraulic properties of the
aquifer is very important for the design of the
ground-water extraction system. An interesting
example of this is the Western Processing site,
where a system consisting of many low-capacity
shallow wells was used to concentrate the capture
zone in the highly contaminated shallow soils.
This was done because it was recognized that the
underlying soils had higher hydraulic conductivity
and would yield large quantities of relatively clean
water to a system composed of a few high-
capacity extraction wells.
In dealing with NAPLs, several other hydrologic
properties of the aquifer materials besides the
saturated hydraulic conductivity are important.
These include the porosity and the complex
relationships between the degree of saturation, the
capillary pressure, and the relative permeability for
the wetting and nonwetting fluids. A consequence
of these additional porous matrix flow properties is
the phenomenon of residual saturation for NAPLs.
This is the degree of saturation below which the
NAPL is, for practical purposes, immobile. These
properties can be measured in the laboratory, and
some analytical and numerical modeling
techniques are available for using them to predict
the behavior of NAPLs. However, these
techniques are not yet developed to the stage
where they are considered reliable for widespread
practical application. More commonly, grain size
analysis may be used to obtain rough predictions
of the NAPL holding and transmitting capacity of
the soil (Mishra, et al., 1980) (Carsel and Parrish,
1988).
Potentiometric Gradients
The ability of an extraction system to capture and
remove contaminated ground water will depend
partly on the potentiometric gradients that it
creates in comparison to the external gradients. A
simple manifestation of this relationship is
illustrated in the case study for the Nichols
Engineering site. The extraction system designers
for this site have provided graphical depictions of
several alternative capture zone estimates,
depending on the magnitude of the regional
gradient.
A somewhat less obvious illustration of the
importance of the regional gradient is provided at
the Sylvester/Gilson Road site. Here, the regional
gradient has been strong enough to cause
contaminated ground water to flow out from under
the slurry wall containment system on the
downgradient side of the site. In response to this
problem, it may be necessary to increase the rate
of net ground-water withdrawal from within the
area enclosed by the wall.
The application of horizontal gradients by the
ground-water extraction system usually has little
effect on the movement of DNAPLs, which are
primarily governed by gravitational forces. In
some cases, changes in vertical gradients may
reinforce or counteract the buoyancy forces and
affect the vertical movement of the free phase.
However, this cause and effect relationship would
probably be difficult to detect, and the present
study provides no examples of it.
CONTAMINANT
CHARACTERISTICS AND
DISTRIBUTION
Identification of
Contaminants
An important step in evaluating the likelihood of
NAPLs is the identification of the compounds that
are present at the site and then- potential to persist
21
-------
in the nonaqueous phase. Creosote, toluene, and
the chlorinated ethenes and ethanes are the
contaminants that were most commonly found as
NAPLs in this study. Other compounds with
higher aqueous solubility, such as acetone,
tetrahydrofuran, and 1,4-dioxane, are less likely to
be found as NAPLs.
The solubilities of the compounds should be kept
in mind when evaluating site data. At some of the
case study sites, analytical data for ground-water
samples were reported that indicated constituent
concentrations higher than the solubility of the
compounds. There was usually no indication that
this had been noticed by the site investigators.
These occurrences should be interpreted as a
strong indication of the presence of NAPLs.
Attention should also be paid to the possibility that
several compounds that are miscible with one
another may be present as a NAPL. This is a very
common at disposal sites for used solvents. The
properties of a multicomponent NAPL may be
significantly different from the properties of the
individual constituent compounds. One effect is
that the partitioning of each individual compound
between the NAPL and the ground water will
reduce the effective aqueous solubility for each
compound.
Another effect is that compounds that are more
dense than water in pure form may be caught up
in a NAPL that floats. For instance, an LNAPL is
present at the Verona Well Field site consisting of
chlorinated ethenes and ethanes mixed with
benzene, xylene, and toluene. The proportions of
this mixture result in a NAPL that floats, even
though several of the compounds of greatest
concern would normally be expected to form
DNAPLs. A similar situation exists at the
Mid-South Wood Products site, where
pentachlorophenol (PCP), a compound with a
specific gravity of approximately 2.0, was mixed
with a light carrier oil for use in wood treatment.
The ground-water monitoring data show extensive
PCP contamination, but do not mention the
presence or nature of the carrier oil. Nonetheless,
the extraction system is designed to deal with the
resulting LNAPL.
Contaminant Distribution and
Concentration
Concentrations in both the soil and the ground
water are important clues to the likelihood of
NAPL presence. Ground-water concentrations
close to or greater than solubility indicate a high
likelihood of NAPLs. However, as has frequently
been noted, concentrations that are less than 10
percent of solubility may also indicate NAPL
presence (Feenstra and Cherry, 1988; Huling and
Weaver, 1991). At the case study sites with
acknowledged NAPLs, the range of maximum
detected ground-water concentrations was from 4.1
percent to over 100 percent of solubility.
Concentrations measured in soil samples may also
be a good indicator of NAPL presence. High soil
concentrations were noted in the update of the
Verona Well Field case study and interpreted as an
indication of NAPL presence. When soil
concentrations are interpreted in this way,
allowance must be made for the partitioning of the
contaminant between the adsorbed, dissolved, and
vapor phases that are included in the sample
(Feenstra, et al., 1991).
The vertical distribution of contaminant
concentrations may also be an indicator of NAPLs.
Sampling from a well cluster in the suspected
source area at the IBM-Dayton site showed
contaminant concentrations increasing with depth.
This was one of the clues used to support the
contention of DNAPL contamination at this site,
where there has been no direct observation of
DNAPLs.
Sorption Characteristics
The importance of contaminant sorption was
emphasized in the first phase study, both as a
retarding mechanism to aquifer restoration and as
a form of residual contaminant source. These
effects may complicate the determination of NAPL
presence on the basis of resistance to remediation,
because both adsorbed contaminants and NAPLs
can prolong the aquifer-restoration process.
For instance, at the Ponders Corner site, the
concentrations in the contaminant plume have been
relatively steady over a period of approximately
6 years. This is an indication of a residual
contaminant source, which could be due to NAPLs
or adsorbed contamination. If the soil did not
22
-------
have high sorption potential, more suspicion would
be directed to NAPLs. However, at Ponders
Comer, much of the contamination is believed to
be adsorbed to the soil in a heavily contaminated
till layer. This does not rule out NAPLs, but does
tend to cloud the evidence for them.
Identification of
Contaminant Sources
It has already been noted in this summary that a
high proportion of the known NAPL sites in the
case studies were the result of leakage from
chemical storage and handling facilities and the
direct disposal of solvents in the ground. Where
these practices are known to have taken place, the
likelihood of NAPLs resulting seems to be high.
23
-------
Chapter 4
OCCURRENCE AND IMPLICATIONS OF NAPLS
NAPLs have been directly observed at eight of the
case-study sites and their presence is suspected at
several others. This chapter will discuss the
waste-handling methods that led to release of
NAPL contamination at these sites, the signs that
revealed their presence, and their implications for
aquifer remediation.
WASTE HANDLING PRACTICES
LEADING TO NAPL
CONTAMINATION
The following table enumerates the known or
suspected sources of the NAPLs at the nine sites
where they are acknowledged to be present:
In such cases, site investigation procedures
intended to detect the presence of NAPLs should
generally be implemented. At some of the case-
study sites, compounds that, by themselves, would
be expected to sink were found to be present as
LNAPLs. At Verona Well Field, for instance,
chlorinated solvents are floating as LNAPLs on the
water table because they are part of a mixture in
which toluene is a constituent of a DNAPL
dominated by the dense compound 1,2,3-
trichloropropane.
IDENTIFICATION OF NAPL
PRESENCE
Although the presence of NAPLs is acknowledged
at several of the sites, there are several others
IBM-Dayton
IBM-San Jose
Savannah River Site
Utah Power & Light
Verona Well Field
Ville Mercier
Mid-South Wood Products
Tyson's Dump
Western Processing
Suspected leaks or spillage from storage tanks
Suspected DNAPL leaks from storage tanks; known spill of
Shell Sol hydrocarbon
Leakage from liquid waste settling basin
Leakage from underground pipeline
Leakage from buried storage tanks
Dumping in abandoned gravel pit
Leakage from waste storage lagoon
Dumping hi abandoned sand pit
Dumping of liquid wastes
In six of the nine cases listed above, the problem
was caused by faulty storage or handling of the
nonaqueous liquids. This implies that the
problems could have been avoided in two-thirds of
the cases by better design, operation, and
monitoring of the storage and handling facilities.
At the other three sites, the NAPLs were hi such
cases, site investigation procedures toluene
predominates. At Tyson's Dump, on the other
hand, toluene is a constituent of a DNAPL
dominated by the dense compound
1,2,3-trichloropropane introduced by dumping of
waste liquids into pits as a means of disposal.
where they are uncertain or are the subject of
contention. Because of their elusive nature,
especially for DNAPLs, it is often difficult to
prove beyond doubt that they are present; and, it is
even more difficult to prove their absence. Some
of the identifying signs and clues found at the
case-study sites are discussed below.
Direct Observation
NAPLs have been observed dkectly at eight of the
sites. At Verona Well Field, IBM-San Jose,
Western Processing, Tyson's Dump, and the
24
-------
Savannah River Site, NAPLS were discovered in
ground-water samples.
In the first three cases, the contaminants were
LNAPLs. LNAPLs are more likely to be
discovered in this way because they are often
present at greater than residual saturation at, and
just above, the water table. It is common to
screen monitoring wells across the water table, and
if the LNAPL saturation of the soil is greater than
the residual saturation, it can flow into the well.
Once the LNAPL has entered the well, it is likely
to be discovered during sampling.
DNAPLs are less likely to appear in ground-water
samples because of their ability to penetrate below
the water table. By penetrating deeper into the
aquifer, they travel a greater distance through the
porous material and are, therefore, less likely to be
encountered at greater than residual saturation.
Also, because of their propensity for vertical
movement, they are more likely to spread
vertically than laterally and are less likely to be
intersected by a monitoring well. Furthermore, if
they do enter a well, they tend to sink to the
bottom where they may escape detection during
sampling. In spite of these difficulties, DNAPLs
were found by ground-water sampling at the
Savannah River Site and Tyson's Dump.
The movement of DNAPLs in the subsurface is
governed primarily by gravity and, where their
downward movement is unobstructed, the depth of
penetration depends on the residual holding
capacity of the aquifer materials and the volume
and rate of contaminant release. In many cases,
the DNAPLs may not move laterally very far from
the original source area. This seems to be the case
at the IBM-Dayton site, where the residual source
area is localized near the former solvent storage
tanks. This situation is expected to facilitate the
control of plume migration.
NAPLs may also be directly observed staining or
flowing from soil samples, coating the outside of
drill rods (as at Utah Power & Light), or seeping
into surface water bodies (Mid-South Wood
Products).
High Concentrations in
Ground-Water or Soil Samples
A strong indication that NAPLs may be present is
when the ground-water samples show
concentrations that are near the aqueous solubility
of the contaminant At three of the case-study
sites, ground-water concentrations greater than the
aqueous solubiUty were reported. These sites are
Fairchild Semiconductor, Utah Power & Light, and
Mid-South Wood Products. Fairchild is not a site
where NAPLs have been acknowledged, but the
other two are. One possible explanation for
ground-water concentrations exceeding solubility
would be the co-solvent effect, but this is not
likely unless the concentration of some other
constituent is extremely high. The most likely
explanation is that NAPLs were present in
colloidal form in the sample and were not noticed
visually. This can be considered a strong
indication that NAPLs are also present in the
aquifer. It is possible that, in some cases, NAPLs
are being removed with the ground water by the
extraction systems.
The lack of any measured ground-water
concentrations close to solubility, however, is not a
good argument for the absence of NAPLs. At
several of the case-study sites where NAPLs are
acknowledged, the highest reported concentration
of the contaminant in question is considerably
below aqueous solubility. Examples are:
IBM-Dayton at 4.1 percent for PCE, Savannah
River Site at 12 percent for TCE, and Verona Well
Field at 11.3 percent for PCE. Several factors that
could account for this observation include
reduction of effective solubility due to partitioning
of the compound between water and a mixture of
nonaqueous solvents, kinetic effects limiting the
rate of dissolution of the compound from the
NAPL, dilution by the flow of ground water in the
aquifer, and dilution during the sampling process.
Because of these effects, concentrations in the
range of 1 to 10 percent of aqueous solubility may
be high enough to lead to the suspicion of NAPL
presence.
The detection of high concentrations of potential
NAPL compounds in soil samples can also
indicate that NAPLs are present. In this case,
however, the relationship between the measured
concentration and the likelihood of NAPL presence
is not as direct as it is for ground-water samples.
Analytical results for soil samples indicate the
quantity of contaminant that was present in the
sample in all forms. This includes the adsorbed
phase, the vapor phase, dissolved constituents in
the soil moisture, and the NAPL phase. By
invoking the assumption of linear equilibrium
25
-------
partitioning, Feenstra, Mackay, and Cherry (1991)
have developed a procedure for assessing the
meaning of high contaminant concentrations in soil
samples. Using a procedure similar to this, it was
determined that high concentrations of PCE
measured at the Thomas Solvent Annex portion of
the Verona Well Field site are an indication of
potential DNAPL contamination there.
Depth of Contamination
Another indicator of the possible presence of
DNAPLs is the observation of high concentrations
at greater depths in the aquifer than would
otherwise be expected in the absence of a strong
vertical gradient. This is one indicator that led to
the determination of DNAPL presence at the
IBM-Dayton site, where DNAPLs have not been
directly observed. At the Mid-South Wood
Products site, high concentrations of
pentachlorophenol also have been detected in the
deepest monitoring well, indicating that this
contaminant may have sunk deep into the bedrock
in DNAPL form.
At the Sylvester/Gilson Road site, the highest
concentrations of toluene were detected near the
water table, even though there is a downward
component of ground-water flow that has
transported other contaminants to greater depth.
This, together with the high toluene concentrations
and their resistance to remedial efforts lends
credence to the possibility that nonaqueous toluene
is present.
Resistance to Remediation
The main reason for the determination that
DNAPLs were present at the IBM-Dayton site was
the persistence of the contaminant plume and the
reappearance of high concentrations when the
extraction system was turned off. It has frequently
been noted that contaminant concentrations tend to
increase when pumping is discontinued. This
effect is considered to be an indication of a
residual source of contaminants in the aquifer.
Such a source could be of several kinds. It could
be due to continued leaching from the vadose zone
or leakage from the disposal area. It could be due
to the release of adsorbed contaminants from
highly sorptive aquifer materials or to a lessening
of the hydrodynamic dilution after the extraction
wells were turned off. At the IBM-Dayton site,
these effects were judged not to have the potential
to explain the observed magnitude of plume
resurgence, and it was concluded that DNAPLs
must, therefore, be present.
IMPLICATIONS OF NAPL
PRESENCE
The discovery of NAPLs at an aquifer restoration
site usually marks a turning point in the course of
the remediation. Aquifer restoration using
ground-water extraction alone is likely to be a very
slow process when NAPLs are present.
Ground-water extraction is an inefficient method
of removing NAPL compounds from the aquifer
because, by definition, it removes only the
dissolved constituent. Thus, to remove the NAPL,
it is necessary to wait for it to dissolve so that it
can be removed with the ground water. Because
most NAPL-forming compounds have low
solubility, large quantities of water must be
removed to extract a small amount of the
contaminant.
In a few instances, the pumping of free-phase
DNAPLs from ground-water extraction wells has
been reported. Globs of creosote have been
produced from extraction wells at the Utah Power
& Light facility. This was an unexpected
occurrence that required the retro-fitting of phase
separation equipment in the ground-water
treatment process. Phase separation has also been
provided for at the Mid-South Wood Products site
in anticipation of free-phase creosote and
pentachlorophenol recovery from the ground-water
extraction system.
There are several removal technologies that have
been used with success at LNAPL sites. These
include soil vapor extraction, free-product
skimming, and enhanced biodegradation. Both
vapor extraction and free-product skimming were
used in the remediation of LNAPLs at the Verona
Well Field site. It was estimated that 45,000
pounds of LNAPL constituents were removed by
vapor extraction, and 1,200 pounds by free-product
skimming. This should shorten the time required
for aquifer restoration, but the ground-water
concentrations are still above cleanup goals and
further restoration is expected to be slow. Both of
these removal techniques are also being applied to
the LNAPL Shell Sol spill area at the IBM-San
Jose site.
26
-------
Where DNAPLs are concerned, the current
established removal technologies are relatively
ineffective. Many new technologies are currently
being tested (e.g., use of surfactants, water
flooding, air sparging) but none of them were used
at the case-study sites.
At four of the case-study sites with acknowledged
DNAPL contamination, the remedial goal is
containment of the solute plume rather than aquifer
restoration. This goal is usually feasible when the
DNAPL has been located or there are strong
indications that it is present, and it is immobile.
Containment in the area where the DNAPL is
located can be combined with restoration of
portions of the ground-water plume that have
migrated beyond the DNAPL zone. If the residual
DNAPL source is limited to a relatively small
area, the migration of the resulting solute plume
may be fairly easy to control. At the IBM-Dayton
site, for example, it is expected that only one
extraction well located near the DNAPL source
area will eventually be sufficient to control the
migration of contaminants.
At several of the case-study sites, there seemed to
be some resistance on the part of the responsible
parties to acknowledging the existence of
DNAPLs, even though the evidence for them is
fairly strong. This resistance may be
counter-productive. Failure to recognize the
implications of DNAPL presence can result in a
much more costly and less effective remedial
action hi the long run than recognizing the
presence of the DNAPL and determining a more
appropriate remedial strategy. In some cases (as at
IBM-Dayton) the existence of a DNAPL source
may only become apparent after an effort at
aquifer remediation.
27
*U.S. G.P.O.:1992-311-893;60269
------- |