United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
Las Vegas, NV 89193-3478
Research and Development
EPA/600/S4-90/028 July 1991
& EPA Project Summary
Field Comparison of Ground-
Water Sampling Devices for
Hazardous Waste Sites: An
Evaluation Using Volatile
Organic Compounds
Karl F. Pohlmann, Ronald P. Blegen, and John W. Hess
To determine whether ground-water
contamination has occurred or
remediation efforts have been effective,
tt Is necessary to collect ground-water
samples In such a way that the samples
are representative of ground-water con-
ditions. Unfortunately, formation of stag-
nant water within conventional monitor-
Ing wells requires that these wells be
purged prior to sampling, a procedure
that may Introduce significant bias Into
the determination of concentrations of
sensitive constituents such as volatile
organic compounds (VOCs). The use of
In situ ground-water sampling devices,
which minimize or eliminate the need for
well purging, may help alleviate some of
the difficulties associated with sampling
ground water at hazardous waste sites.
In this study, several ground-water sam-
pling devices, Including two In situ sys-
tems, were fie Id-tested to determine their
capability for yielding representative
VOC data.
Sampling devices Included In the field
study were a bladder pump, a bladder
pump below an Inflatable packer, a bailer,
a bailer with a bottom-emptying device,
an In situ Westbay MP System™1, two In
situ BAT™' Ground-Water Monitoring
Systems, and a prototype BAT well probe.
The devices were field-tested at a site
contaminated by a VOC plume, and the
'Mention of trade names or commercial products does
not constitute endorsement or recommendation for
comparison was based on the ability of
the devices to recover representative
concentrations of the VOCs. The data
show that, under these field conditions,
the sample variability associated with
the bladder pump and In situ samplers Is
of similar magnitude and that there Is
little difference In apparent accuracy and
precision of these devices. In contrast,
both bailers Introduced significant varl-
abl IKy Into VOC determinations and were
found to be less accurate and precise
than the other devices tested. The study
Indicates that the tested In situ devices
may eliminate the need for well purging
prior to sample collection and that the
resulting samples are at least as repre-
sentative as those collected with a blad-
der pump In a conventional monitoring
well.
This Project Summary was developed
by EPA's Environmental Monitoring Sys-
tems Laboratory, Las Vegas, NV, to an-
nounce key findings of the research
project that Is fully documented In a
separate report of the same title (see
Project Report ordering Information at
back).
Introduction
The ability to collect ground-water
samples representative of aquifer
hydrochemical conditions is a major con-
cern in any ground-water investigative ef-
fort. Unfortunately, there are many factors
in the sampling and analysis process that
can introduce variability into determinations
Printed on Recycled Paper
-------
of chemical constituent concentrations,
greatly influencing the ability to obtain accu-
rate results. Examples include well drilling
method, well design, the materials used in
well construction, well development and
purging, sampling device, sample handling
and preservation, and analytical technique.
Of these, maintenance of sample integrity
from the subsurface sampling point to the
sample container has received consider-
able interest because of the great potential
impact on sample representativeness pos-
sible during this phase of the sampling
process.
Concern about obtaining representative
samples of ground water containing volatile
.Organic compounds (VOCs) has arisen be-
cause VOCs are common contaminants at
hazardous waste sites and they are sensi-
tive to sampling method. To address this
concern, sampling devices have been de-
veloped to minimize impact on sample qual-
ity. These devices are generally used in
conventional monitoring wells and there-
fore require removal of stagnant water from
the well prior to sampling. While well purg-
ing is generally considered necessary to
collect representative ground-water
samples, the purging process itself may
introduce considerable bias into the sam-
pling results and may produce waste-water
disposal problems and undesirable expo-
sure of sampling personnel to potentially
hazardous materials.
Among the newer commercially available
sampling devices are two devices that need
to have little or no water purged from the
system prior to sampling. These devices
are not used in conventional monitoring
wells but are stand-alone systems installed
directly into the subsurface, similar to con-
ventional monitoring well installations. Be-
cause of the nature of their design and
installation, these devices collect samples
almost directly from the formation and so
are often referred to as in situ systems. The
existence of these devices raises three im-
portant questions:
•Do sampling methods exist that eliminate
the need for well purging prior to sample
collection?
•How valid or representative are the result-
ing samples?
•Are the proposed in s/Yusampling methods
inherently invalid because of the necessity
of well purging?
This study was initiated to address the
concerns expressed by these questions.
The primary objective of the study was to
compare several conventional and in situ
devices for their ability to collect represen-
tative VOC samples.
Descriptions of Sampling
Devices
Seven ground-water sampling devices,
all of which are commercially available,
were utilized in this study. The devices may
be categorized as grab samplers (bailers),
positive displacement mechanisms (blad-
der pump, bladder pump below an inflatable
packer), and in situ devices (Westbay MP
System™, two BAT™ Ground-Water Moni-
toring Systems, and a prototype BAT well
probe). To minimize the potential for bias
resulting from contact with the materials of
which the samplers were constructed, the
devices were selected from the most chemi-
cally inert materials availablef rom the manu-
facturer.
Two bailers were utilized in the study.
The first was a Teflon bailer with a Teflon-
coated stainless steel haul line. The design
of this bailer, with side ports and a closed
top, was chosen to minimize the possibility
of stagnant water entering the top of the
device as it was retrieved from the well. The
second bailer was similar in design but
utilized a bottom emptying device (BED) to
transferthe sample directly from the bottom
of the bailer to the sample container.
Two Teflon bladder pumps were used in
the study, one in conjunction with an inflat-
able packer mounted above the pump and
the other without the packer. The packer is
designed to reduce purge volumes by iso-
lating stagnant well water above the pump
and preventing that water from migrating
downward to the pump intake. Each pump
was dedicated to a single well, although the
bladder pump without the packer was re-
moved from its well after purging to allowfor
sampling with the bailers.
The BAT Ground-Water Monitoring Sys-
tem consists of sealed components with
hydraulic interconnections between com-
ponents accomplished through the use of
hypodermic needles, flexible seals, and in-
duced pressure gradients. The primary com-
ponent of the system is the BAT filter tip
which is threaded onto the bottom of stan-
dard 5.1-cm-diameter monitoring well cas-
ing and permanently installed at the desired
depth. Ground-water samples are collected
in sealed, evacuated glass vials contained
in a sample container housing which is
lowered down the casing to the filter tip.
Contact between the container housing and
the top of the filter tip causes a double-
ended hypodermic needle to puncture septa.
in both the sample container and the filter tip
cap, causing ground water to flow from the
formation into the sample vial. When full,
the sample is retrieved, and both septa are
resealed as the syringe is withdrawn from
the bottle and filter tip cap. The vials used fo
sample collection may then be sent direct!
to the laboratory for analysis. Because on!
the small volume of water contained wrthi
and immediately outside the filter tip is ii
contact with the sampling device, the vol
ume of stagnant water which needs purgini
is significantly smaller than in convention.
monitoring wells. Twof ittertips were utilize
in this study, one with a filter consisting c
polytetrafluoroethylene (PTFE), and on
consisting of high-density polyethylen
(HDPE). These devices are referred to i
the report as the "PTFE filter tip" and th
•HDPE filter tip,* respectively.
A third BAT device used in this study wa
a modified version of the PTFE filter ti|
which was installed in the screened interv.
of a conventional 5.1 -cm-diameter monitoi
ing well. This device, referred to in the repoi
as the "well probe," utilizes two Viton™' G
rings to seal off the well bore above am
below the screened interval. With the we
screen isolated, the well probe functions a
a sampling port from which water contain*
within the screened interval alone is ex
tracted, therefore, reducing the purge vo!
ume required. Sampling procedures ar
identical to those for the in situ filter tips.
The Westbay MP System (referred to i
the report as the "multi-port") allows dis
crete samples or measurements to betake
at multiple levels within a single borehole
The system consists of various lengths c
casing joined by regular or yatved port cot
plings and a variety of specialized tools an
probes to access the ports and retrieve th
samples or measurement data from th
environment outside the sealed casing
Sampling probes accessing a measure
ment port draw the sample from direct!
outside the casing, so there is no exposur
of the sample to fluids contained inside th
casing. For this reason, there is no need t
purge standing water inside the casing, an<
purge volume requirements aresignificantl
reduced over the requirements of conver
tional monitoring wells. The multi-port com
ponents chosen for this study were con
structed of stainless steel.
Procedure
The study was approached by conduct
ing comparisons of seven ground-watt
sampling devices, including two types of /
situ devices, at a field site over a VOC
'Mention of trade names or commercial products doe
not constitute endorsement or recommendation ft
use.
-------
contaminated shallow aquifer. The aquifer
consisted of alluvial sands and gravels, with
some silt and day interbeds, overlying, at a
depth of 9.5 m below ground surface, thick
clay/silt beds containing thin, interbedded
sands and some gravel. The water table
depth was approximately 3 m below ground
surface, resulting in a total unconf ined aqui-
fer thickness of 6.5 m.
Ground water at the site is contaminated
by a variety of organic and inorganic com-
pounds related to a major industrial site
located approximately 3 km hydraulically
upgradient. Much of the organic contamina-
tion is the result of an underground storage
tank leak which released approximately
113,500 Lof benzene into the ground water.
Movement of the benzene plume
downgradient has brought it into contact
with a variety of other organic compounds,
which may have been mobilized and trans-
ported away from their original disposal
areas. The VOCs of highest concentration
at the study site, benzene, chlorobenzene,
1,2-dichlorobenzene, and 1,4-dichloroben-
zene, were each present at the initiation of
the study in concentrations up to 1600 u.g/L
To minimize geologic and hydrochemical
variability between the sampling devices it
was necessary to install the wells close to
each other. However, the wells also had to
be spaced far enough apart so that the
effects of pumping at one well would not
impact the distribution of VOCs near adja-
cent wells. Based on site geology and cap-
ture zone calculations, a 6-m spacing was
chosen. Six boreholes were drilled on a
rectangular grid pattern by using a dual-
tube percussion hammer drill rig. Two of the
holes on opposite corners of the grid were
logged by using magnetic induction and
natural-gamma logging methods. These logs
were used to study geologic variability across
the site and to aid in locating the sampler
intake depths. Results of the logging indi-
cated that geologic conditions were similar
at the two corners of the site and that well
intake depths of 6 to 7 m would result in
lower clay content and potentially better
flow characteristics.
Five of the six wells were cased with 5.1 -
cm-diameter Teflon casing. Three of these
wells were screened over a 0.3 m-interval
(for installation of the bladder pump, blad-
der pump/packer, and well probe), while the
other two had HOPE and PTFE filter tips
installed. The sixth well consisted of the
stainless steel multi-port. The well screens,
filter tips, and the deepest multi-port mea-
surement port were all installed at depths of
about 6m.
The sampling devices were evaluated to
compare their relative accuracy and preci-
sion and to determine if the non-pumping in
situ methods yielded representative data.
The comparison was based on the ability of
each device to deliver representative
samples from the subsurface environment
to the ground surface and into an appropri-
ate sample container. Because VOCs are
common contaminants at hazardous waste
sites, the effects the tested devices had on
VOC recovery were of primary importance.
As virtually all field sampling techniques
introduce bias into VOC determinations, a
true assessment of accuracy, and therefore
representativeness of ground-water condi-
tions, is not possible in a field study. How-
ever, because of the physical and chemical
properties of most VOCs, losses of VOCs
from a sample are much more likely than
increases. Therefore, a relative approxima-
tion of accuracy was based on the concen-
trations of VOCs recovered during the sam-
pling process (i.e., those devices which
recovered the highest VOC concentrations
were considered the most accurate).
Because the sampling devices were the
source of variability of most interest to this
study, all other sources of variability present
in the study were minimized. In addition, it
was necessary to make the assumption that
geologic and hydrochemical conditions at
each of the six well sites were identical.
However, despite the assumption to the
contrary, the natural variability in lithology
and stratigraphy of alluvial materials sug-
gests that geologic conditions were cer-
tainly not identical at each of the six wells.
The variability in geologic conditions may
be an important element of the total variabil-
ity observed in this study but could not be
adequately quantified in a way that would
relate to effects on VOC concentration vari-
ability. Likewise, the assumption that
hydrochemical conditions were the same at
each of the six well sites was made in order
to form a basis for the sampler comparison.
To test this assumption a "survey" sampling
round was conducted in which an initial set
of samples was collected from three of the
wells. These wells were utilized for the
survey sampling because they were the
only wells that could be sampled with the
same device, a Teflon bladder pump, and
because their locations represented a fairly
complete coverage of the study site. Be-
cause of their design, the other three wells
could not be sampled with the bladderpump
and, therefore, were not included in the
survey sampling. The results indicated that
mean concentrations at each of the three
sampled wells were within one standard
deviation of each other and within labora-
tory error. However, some variability in VOC
concentrations was evident in the survey
sampling, and the possible contribution of
this small variability to the overall variability
observed in the sampling results was kept in
mind during the analysis.
The VOC concentration data resulting
from the comparisons were statistically ana-
lyzed by means of a two-way analysis of
variance to determine if the sampling de-
vices and/or sampling time introduced a
significant source of variability to the data.
Univariate analyses of variance were then
used to determine if significant differences
existed between device sample means,
based on recovery of each individual com-
pound. Finally, Tukey multiple comparison
tests were conducted to identify which indi-
vidual pairs of sample means during each
sampling event were significantly different
at the 5 percent level. The devices were
then grouped accordingly by bracketing
those devices whose mean concentrations
showed no significant differences.
The comparisons in this study were di-
vided into three experiments: A, B, and C.
Experiment A consisted of eight sample
rounds over a 19-week period and involved
all of the devices except the BED bailer.
Experiment B, which utilized only the blad-
der pump, bailer, PTFE filter tip, and multi-
port, consisted of multiple replicate samples
collected with each device during a single
sampling event. Experiment C included the
four devices used in Experiment B, with the
addition of the HOPE filter tip and BED
bailer. Four sample rounds were conducted
at 12-week intervals during this experiment.
Results and Discussion
The results of Experiment A suggest that
both of the filter tip devices and the well
probe recovered benzene and chloroben-
zene with an accuracy greater than that of
the bailer and at levels rivaling those ob-
tained with the bladder pump. Before deter-
mining the source of the anomalously low
concentrations, the multi-port produced VOC
samples which were much less accurate
than those collected with the bladder pumps,
filter tip, and well probe devices. Additional
experiments with the multi-port have shown
that replacement of the perforated VOA
bottle septum with a new septum can pre-
vent sample degradation and allow this
system to extract accurate samples which
can be preserved until the time of analysis.
Experiment B, the design of which was
based on the results and problems noted
during the first experiment, confirmed many
of the original findings. The statistical analy-
sis indicated that for eight of the nine de-
tected VOCs, concentrations recovered by
the PTFE filter tip were significantly higher
than, or not significantly different from, con-
centrations recovered by the bladderpump.
Although the multi-port was slightly less
accurate than the bladder pump and filter
-------
tip, precision was comparable for all three.
Samples collected with the bailer appeared
to be less accurate than those collected with
all other devices. However, the bailer sam-
pling procedures used did not allow for the
collection of true replicate samples, and,
therefore, complete assessments of the
accuracy and precision of the bailer could
not be made. Experiment B also confirmed
the ability of the multi-port to collect more
accurate samples when the perforated
sample vial septum is replaced.
The results of Experiment C followed
several of the trends established during the
previous two experiments, but provided
longer-term information. This experiment
suggested that the bladder pump and the
filter tips were the most accurate of the
devices tested and that the bailers were the
least accurate. The multi-port was found to
be somewhat less accurate than the blad-
der pumps and filter tips. The multi-port,
bladder pump, and HOPE filter tip provided
the most precise samples, while the bailers,
the BED bailer in particular, provided the
least precise samples.
Conclusions and
Recommendations
The variability observed in the survey
sampling, although less than the stated
analytical error, indicates that some spatial
variability may exist at the site and that the
assumption of identical conditions between
devices may not have been strictly valid.
Because of the nature of this study, a com-
plete understanding of the hydrochemical
variation between the well installations could
not be determined independently of the
tested devices. As a result, the possibility of
spatial variability contributing to the overall
observed variability cannot be discounted.
If spatial variability typical of many field sites
is included in the final analyses, these ex-
periments suggest that the variability asso-
ciated with the bladder pump and in situ
samplers is of similar magnitude and that
there may be little difference in the accuracy
and precision of these devices. The bailers,
on the other hand, which sampled the same
well as the bladder pump, were not sub-
jected to the uncertainty of varying spatial
conditions, so the samplers were the pri-
mary source of variability.
This study shows that, under field condi-
tions, in situ devices may provide samples
with essentially the same precision and
accuracy as bladder pumps and greater
precision and accuracy than bailers. It ap-
pears that the designs of the two types of in
situ devices tested virtually eliminate the
need for well purging prior to sample collec-
tion and can provide representative VOC
samples in sand and gravel aquifers.
In addition to collecting representative
samples and minimizing purging volumes
the in situ devices allow samples to tx
collected quickly, while reducing both expo
sure of sampling personnel to potentially
hazardous materials and the volume of purge
water to be disposed of. Furthermore, these
sampling systems are relatively easy tc
operate and maintain, the standardize
sampling methodology reduces variability
potentially introduced by sample handling
and fewer sampling personnel are requirec
to obtain samples. The few operational dif
ficulties experienced with the two in sih
devices are described fully in the report.
To obtain a more accurate evaluation o
these sampling devices, it is suggested tha
additional studies be developed and imple
mented, such as application in various
hydrogeologic conditions, including low
permeability environments, or areas witt
different contaminants. Other installatioi
methods and devices at varied depths alsc
should be investigated. Further studies
performed at a variety of sites and involving
these and other commercially available //
situ sampling devices, are needed to im
prove understanding of the applicability o
these devices to a variety of monitoring
situations.
.S. GOVERNMENT PRINTING OFFICE: IHI - M8-028/40020
-------
Karl F. Pohlmann, Ronald P. Blegen, and John W. Hess are with the Water Resources
Center, Desert Research Institute, Las Vegas, NV 89120.
Jane E. Denne is the EPA Project Officer (see below).
The complete report, entitled field Comparison of Ground-Water Sampling Devices for
Hazardous Waste Sites: An Evaluation Using Volatile Organic Compounds," (Order
No. PB91-181 776/AS; Cost: $23.00, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Las Vegas, NV 89193-3478
United States
Environmental Protection
Agency
Center for Environmental
Research Information
Cincinnati, OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use $300
EPA/600/S4-90/028
------- |