United States
Environmental Protection
Agency
National Risk Management
Research Laboratory
Cincinnati, Ohio 45268
Research and Development
EPA/600/S-96/001
May 1996
&EPA Project Summary
Hydrogeologic Characterization
of Fractured Rock Formations:
A Guide for Groundwater
Remediators
Andrew J. B. Cohen, Kenzi Karasaki, Sally Benson, Gudmundur Bodvarsson,
Barry Freifeld, Pascual Benito, Paul Cook, John Clyde, Kenneth Grossenbacher,
John Peterson, Ray Solbau, BhaskarThapa, Don Vasco, and Peter Zawislanski
Abstract
A field site was developed in the
foothills of the Sierra Nevada, California,
to develop and test a multidisciplinary
approach to the characterization of
ground-water flow and transport in
fractured rocks. Nine boreholes were
drilled into the granitic bedrock, and a
wide variety of instruments and
methodologies were tested. Fracture
properties were measured on outcrops
and in boreholes using acoustic
televiewer, digital borehole color
scanner, and by down-hole camera logs.
Conventional geophysical logs were
collected. In addition, thermal-pulse and
impeller flowmeter logging, fluid
replacement and conductivity logging,
packer-injection profiling tests, and
ordinary open-hole pumping tests were
conducted. Transmissive fractures were
identified by integrating results from
hydrologic and geophysical
measurements, and the hydrogeologic
structure of the formation was
hypothesized. Cross-hole seismic
surveys yielded tomograms of inter-
borehole rock properties. Visualization
software was used in combination with
geophysical logs to interpolate inter-
borehole properties, and a detailed 3-D
model of the subsurface was
constructed. Other referenced work at
the site includes cross-hole hydrologic
tomography, tracer tests, fracture-
specific morphology studies, and
development of an automated data
acquisition system used to collect data
and monitor and control test parameters
during borehole testing. A novel aspect
of the project report is its guidebook
format. A description of each tool and
methodology, the strengths and
shortcoming of each, how they compare
with one another, and suggestions of
how best to analyze and integrate data
are presented.
The Project Summary was developed
by the National Risk Management
Research Laboratory's Subsurface
Protection and Remediation Division,
Ada, OK, to announce key findings of the
research project that is fully documented
in a separate report (see Project Report
ordering information at the back). The
reader is encouraged to visit the Internet
Web page for further information:
http://www.epa.gov/ada/kerrlab.html
1 Introduction
Fundamental to every ground-water
remediation effort is a description of the
subsurface distribution of contaminants and
fluid flow properties. The hydrogeologic
complexity of fractured formations makes
their characterization very difficult. This
-------
difficulty is the primary reason that the
majority of fractured portions of
contaminated formations have not yet been
investigated. Knowledgeofnewandexisting
site characterization tools and
understanding of analysis methods are
necessary first steps toward improved
remediation of fractured formations.
1.1 Format of Project Report
The project report describes the work
and findings of a hydrogeologic
characterization study of a saturated,
fractured, granitic rock aquifer in the foothills
of the Sierra Nevada, California. A wide
variety of new and traditional hydrologic
and geophysical characterization tools and
methodologies were tested. The numerous
field experiments and analyses have
provided many insights that should greatly
benefit the remediation community. The
project report was designed in a guidebook
format. First, an overview of the problems
associated with remediating fractured
aquifers is presented. Case histories are
referenced as examples. Second, a brief
description of the methods and results of
the characterization effort at the
experimental field site are presented. The
following chapters comprise the bulk of the
report. Each chapter describes a particular
characterization phase, and a general
strategy for hydrogeologic characterization
is presented. Each tool and method is
described in detail. Descriptions of the
tools, how they are used, what are the
strengths and shortcomings of each, how
they compare with one another, as well as
how best to analyze and integrate the data
collected, are discussed. Findings obtained
at other remedial sites where fractured
materials are present are also referenced.
Determination of subsurface flow properties
is the emphasis of this project rather than
contaminant sampling and plume
delineation. However, issues regarding the
effect of incorrect characterization of flow
properties on prediction of contaminant
behavior are addressed. The findings and
recommendations are not necessarily
limited to fractured crystalline rock
formations; many apply to fractured
subsurface formations in general.
1.2 Experimental Field Site
Studies were conducted at the Raymond
Field Site, which is operated by the E. O.
Lawrence Berkeley National Laboratory and
the U.S. Geological Survey. Sponsorship
has been provided by the Office of Civilian
Radioactive Waste Management of the
Department of Energy, Atomic Energy of
Canada, Ltd., and bythe U.S. Environmental
Protection Agency. The site was established
to develop and test a multidisciplinary
approach to the characterization of ground-
water flow and transport in fractured rocks.
No ground-water contamination was
present. Research began in 1992 and is
ongoing.
The site is situated in the western foothills
of the central Sierra Nevada, California,
approximately 60 km (37 mi) south of
Yosemite Valley and about 5 km (3 mi)
southeast of the town of Raymond (Figure
1). Nine vertical boreholes penetrate
fractured granitic bedrockto depths ranging
between 75 and 90 m. They are arranged
in a triangular pattern and are spaced no
more than 61 meters apart (Figure 2).
CALIFORNIA
,
Figure 1. Location of Raymond Field Site
The conceptual model of the site
subsurface is based on the integration of all
work conducted at the site. Borehole
geophysical logs indicate that hundreds of
fractures intersect each well. However,
borehole flow logs reveal there are only
several distinct hydraulically conductive
fractures in each. Many of the transmissive
fractures are subhorizontal and westwardly
dipping, and are associated with pegmatite
dikes. A few are subvertical or of different
orientation. Most occur within or near one
of two zones of relatively low electrical
resistivity and increased borehole diameter,
both indications of altered rock. These
zones dip gently to the west and are
separated by about 25 meters. Fractures
with minimal transmissivities occur in other
portions of the well. The conceptual model
consists of two subhorizontal and
hydraulically conductive fracture zones that
behave as confined units imbedded within a
relatively impermeable rock matrix
(Figure 3).
1.3 Data Acquisition System
An innovative, automated data acquisition
system was developed for the field site.
The new system was built around a 486 PC
which was used for controlling a sampling
table, opening and closing borehole valves,
and for logging pressure and chemical
concentration data from all the
measurement locations throughout the site.
Pressures are monitored in upto 29 different
packed-off intervals while flow rates and
water chemistry are monitored and analyzed
simultaneously. The output from the
pressure transducers and chemical
concentration measuring devices are
available in real-time on the computer. In
addition, controlled parameters such as
flow and sampling rates can be adjusted
during the test. A schematic of the new data
acquisition system is shown in Figure 4.
Pressure transducers, tracer injection or
tracer recovery instruments, and pumps
were placed within isolated zones during
various hydrologic tests. Flowmeters and a
spectrofluorometer were located at the
surface. The packers have feed-throughs
so that both fluids and electric signals can
be passed to the surface. Data can be
recorded at a rate of 1 Hz. Electrical signals
from the well heads are transferred in
underground conduits to terminal blocks in
the computer room. Analysis of fluorescent
tracers is accomplished using a flow-through
cell in a fluorometer, and ionic tracer
concentration is determined using ion
specific electrodes. A computer-controlled,
144 bottle sampling table was built so that
samples could be taken back to the lab for
further analysis. In several months of field
use, the fully automated data collection
system has proven to be highly reliable,
even for extended multi-week tests.
2 Methods and Results
2.1 Surface Fracture
Characterization
Physical properties of fractures exposed
on outcrops in the area within a several
hundred meter radius from the well field
were measured. The report describes
findings from several hundred
measurements of fracture orientation, trace
length, spacing, weathering, secondary
mineralization, and relative displacement.
Other measurements included fracture
aperture, roughness, and planarity, detailed
measurements of fracture spacing made at
nine outcrops, and mapping of a large
fracture pavement.
Regional and site-specific fracture sets
were identified. Regional sets consist of
-------
SW-4
SE-4
z
[m]
0 -,
-10 -
-20 -
-30 -
-40 -
-50 -
-60 -
-70 -
-80 -
-90 -
-100 -
I I i 1 I I I I « I
potentiomebric
surface
FRACTURED
GRANITIC
BEDROCK
(SATURATED)
I casing
unscreened and uncased borehole
Figure 2. Plan and cross-sectional view of boreholes at the Raymond Field Site.
two orthogonal and steeply dipping tectonic
fracture sets, and unloading fractures that
are subparallel to the topographic surface.
The spatial distribution of fracture density,
aperture, and infilling characteristics of these
sets is, in general, highly heterogeneous on
the scale of the well field, and there is no
apparent systematic structure. The
characteristics of these sets were heavily
altered due to their exposure at the surface.
Other fracture sets exposed in the vicinity of
the well field are associated with aplite and
pegmatite dikes. Pegmatites with
thicknesses on the order of centimeters
were often observed to contain a
discontinuous open fracture nearthe center
and chemical alteration associated with fluid
flow. The average continuity of pegmatitic
dikes is on the order of 10 m, with some
extending 30 meters or more.
Surface fracture characterization provided
information useful to other phases of the
site characterization. Integration of
conventional geophysical logs, hydrologic
logs, and borehole fracture detection
methods revealed that the upper
transmissive zone in many wells is
associated with a pegmatite dike. The
observation that pegmatite dikes at the
surface are laterally continuous for tens of
meters and are often fractured and exhibit
signs of fluid flow provided support for
inference that these intervals are portions
of a continuous band of fracturing associated
with a westwardly dipping pegmatite.
Similarly, tectonic fractures were observed
to be planar and laterally extensive, which
justified the use of extrapolating these
fractures linearly to other boreholes as a
means of hypothesis testing and model
development.
2.2 Well Drilling
Boreholes were drilled using an air-
percussion rotary drill. Boreholes were
cased at shallow depths where the rock was
weathered and weak, and left uncased and
unscreened below these zones. Detailed
logs of the wells were made by a geologist
as drilling proceeded. The location of some
fractures was identified by observing drill
behavior and cuttings. Drilling rates
increased in fracture zones. Fluid-bearing
intervals were encountered, and changes
in fluid flow with depth were measured.
These observations, in combination with
changes in drilling rate and cutting
characteristics, indicated the possible
locations of transmissive fracture zones.
Total fluid flow observed at the final drilling
depth was measured and used as an
indication of the relative discharge capacity
of each well. Comparison of the geologist's
logs with conventional geophysical logs and
fluid flow logging results show that several
of the dominant transmissive fracture zones
were identified during drilling.
2.3 Conventional Geophysical
Logging
Three-arm caliper, fluid conductivity, 16-
and 64-inch normal resistivity, natural
gamma, temperature, single-point
resistance, spontaneous potential, and
lateral logs were collected in each well.
Results and uses of these logs are described
in Section 3.6.
2.4 Pumping Tests
Constant rate pumping tests were
performed for several well combinations.
Pressure transducers that can detect water
level changes on the order of 0.1 mm were
used to measure drawdown. Measurements
in the pumping well and observation wells
were recorded as often as every 10 seconds.
Wellsthat exhibited low yields during drilling
could not be sustained since water levels
would fall beneath the upper fracture zone,
and well bore storage effects almost
completely dominated the pumping well
pressure response. Observation well
pressure transients typically deviated from
an ideal confined aquifer response at early
pumping times. Some fit the ideal curve
almost exactly. Transmissivities calculated
from standard semi-log pressure transient
analysis were very similar from well to well,
-------
SW-4
, I
0-0
•25-
z[m]
-50-
-75
•100-
SE-4
I
WEST
.. - - IP
--"*
"'
|0
o
o
o
lower
I iilpiiaiftr:! resistivity *: i"O'O Ol'in'• m
„ westward"],' dipping
Figures. Conceptual model of hydrogeologic structure at the Raymond Field Site.
and did not reflect the large degree of
heterogeneity found from borehole flow
logging tests. The report describes how
highly heterogeneous fractured formations
can respond like homogeneous porous
formations. Potential misinterpretations of
pressure transients, in general, are also
discussed. Vertical flow profiling in
observation wells shows that the boreholes
act as short-circuiting fluid pathways.
2.5 Detection and Measurement
of Subsurface Fractures
Several technologies were explored as a
means to observe and measure subsurface
fractures. Acoustic televiewer (ATV) logs
provide images of the acoustic reflectivity of
the borehole wall. Fracture signatures are
observable and enable measurement of
fracture depth, orientation, dip, and apparent
aperture. ATV logs were collected for all
nine wells. Color television camera logs
were obtained for all nine wells and were
used to observe fracture alteration and
mineralization not observable on the ATV
logs. Seven wells were probed with a new
digital borehole scanner(DBS). The scanner
digitally records the reflected intensity of
the red, blue, and green light wavelengths
from the borehole wall as a white light
source rotates and simultaneously
illuminates the borehole wall. High resolution
color images of the borehole wall were
constructedfromthescannerdata. Fracture
orientation, surface roughness, aperture,
identification of fracture mineralization, and
other microscale properties are measurable.
A stereonet of measured fractures from
the ATV logs shows that at least three
fracture sets are present within the well
field. Comparison of the ATV logs with the
color television and DBS logs revealed that
fractures can be divided into more than
three sets, as differences occur in
mineralization that are not observable on
the ATV logs. In general, these comparisons
revealed that in the ATV logs: 1) closely
spaced fractures or zones of altered rock
that were gouged out during drilling appear
as a single, large fracture zone with no
distinguishable orientation, dip, or aperture;
2) mineral infilled fracturesthatwere slightly
gouged-out during drilling appear as open
fractures; 3) sealed and mineral infilled
fractures not affected by drilling were not
detected; 4) partially sealed or very small
fractures were not detected or were difficult
to identify; and 5) the top and bottom of
steeply dipping fractures were gouged out
and appearto havegreaterdip. Subsequent
borehole hydrologic tests showed that only
several of the hundreds of observed
fractures in each borehole are transmissive.
Accordingly, spacing distribution of borehole
fractures did not correlate with the hydrologic
properties around each borehole.
Quantitative measurements from the
television camera logs were not possible,
but qualitative determinations of whether or
not fractures were infilled were possible.
Rock discoloration around individual
fractures could be seen, and therefore
helped locate fractures that might be
hydrologically altered. The location of many
pegmatite dikes was identified because their
large, reflective crystals were easily
recognizable on the image. This was very
valuable since it was later found that these
dikes were one of the significant conductive
fracture sets. Orientation and dip of
pegmatites were determined by identifying
their trace on ATV logs.
The DBS provided borehole wall images
of the greatest detail and quality. Many
more fractures were observable compared
to the ATV logs. The photographic quality
of the logs enabled more precise
measurement of fracture aperture,
orientation, and dip. Mineral infilling in
fractures was clearly visible, and enabled
distinction between open and sealed
fractures. Fluid alteration such as oxidation
stains around some fractures was clearly
observable. Individual fractures within zones
that were gouged out during drilling were
measurable. Several disturbed zones were
highly transmissive, and the DBS enabled
detection and measurement ofthe individual
fractures within these zones.
2.6 Borehole Flow Logging
I mpellerflowmeter logging, thermal-pulse
flowmeter logging, fluid replacement and
conductivity logging, and straddle-packer
injection profiling were performed to
determine the location of fluid-bearing
fractures and their respective trans-
missivities. Impellerflowmeter profiling was
performed in three wells. All ninewellswere
profiled with a thermal-pulse flowmeter.
Fluid conductivity logging was performed in
seven wells, and straddle-packer injection
tests were conducted in all nine wells.
Results were integrated with the borehole
scanning logs described above, and with
the conventional geophysical logs to
determine the specifictransmissive fractures
within an observed borehole flow interval.
The impellerflowmeterwas used to profile
the vertical flow in a well while it was pumped.
The instrument is composed of an impeller-
type flowmeter mounted above a hollow
shaft that passes through the center of an
inflatable packer. The well was pumped at
a constant rate by a downhole submersible
-------
486 PC
I
Multiplexer &
Digital Multimeter
Automated Valves
Fluorometer
Pressure Trandueers
Flow meters
Ion Specific Electrodes
Automated Sampling Table
Figure 4. Schematic of data acquisition and control system at the Raymond Field Site.
pump situated near the upper portion of the
well. The assembly was successively
inflated at different depths and vertical flow
measurements were taken (Figure 5a). The
drawdown transient throughout the entire
test duration was measured by a pressure
transducer situated near the upper portion
of the well. Flow rate into the borehole from
a particular interval was calculated from the
difference in vertical flow measured over
the interval. Apparent formation
transmissivity was determined from the
transient drawdown, which responded as a
confined aquifer. Transmissivities of
particular flow intervals were calculated as
the product of fractional inflow and formation
transmissivity. The minimum measurable
flow rate of the impeller was 4 L/min. This
relatively high stall velocity prohibited the
profiling of low-yielding wells. Hence, the
procedure was only conducted in three
relatively high yielding wells. In addition,
flow intervals near the lower portions of the
wells could not be defined because vertical
flow rates were below the stall velocity of
the instrument (Figure 5a).
The thermal-pulse flowmeter (also
referred to as heat-pulse flowmeter) was
used to profile multiple wells during the
pumping of a single, high yielding well (Figure
5b). The instrument consists of a heating
grid and temperature sensors emplaced
within a central tube that passes through an
inflatable packer. Vertical flow velocity is
determined by heating a small parcel of
water at the heating grid and observing the
time of arrival of the parcel at temperature
sensors located above and below the grid.
Response of one of the sensors indicates
flow direction. Differences in vertical flow
with depth indicated fluid flow into or out of
a borehole, but no quantitative estimate of
the conductivity of such intervals could be
determined. The thermal-pulse flowmeter
method detected more flowing intervals
than the impeller method. This is consistent
with the fact that it has a lower minimum
detection limit. Vertical flows on the order
of 0.05 L/min were measurable. In theory,
the thermal-pulse flowmeter could be used
in the same manner that the impeller was
used in this study, thereby allowing
calculation of interval conductivities. For
some wells, measurements at the same
depth at different times were different due
to the evolving transient flow field. This
made the interpretation of some portions of
some logs difficult. In one well, the thermal-
pulse flowmeter did not detect a flowing
interval identified by the impeller method.
This is most probably an artifact of the test
methodology, not the instrument. A
particular fracture may contribute flow
differently when a well other than the one
the fracture intersects is being pumped. In
addition, the vertical flow transient in an
observation well associated with flow into a
conductive fracture could diminish to
unobservable levels by the time the well is
profiled.
Fluid replacement and conductivity
logging was done in seven of the wells. The
method consists of replacing the fluid in the
well bore with deionized water, which has a
much lower electrical conductivity than the
formation water. Replacement was
achieved by simultaneously pumping
formation water near the upper portion of
the well and replacing with deionized water
at the bottom using downhole tubing. After
replacement was complete, the well was
pumped at a low and constant rate and a
time sequence of upward and downward
logs of fluid electric conductivity were
collected. Figure 5c shows the downward
logs forwell SE-1. The conductivity logs for
the seven wells exhibit noticeable peaks at
conductive fracture locations. Therefore,
determination of the particular fracture
contributing to flow was straightforward,
requiring only a brief inspection of the
acoustic televiewer and/or television logs.
Many more conductive fractures were
identified bythis method than by the impeller
or thermal-pulse flowmeter. Equally
significant is that the measurements were
more precise and enabled quick and
confident assessment of the locations and
relative magnitudes of the conductivity of
particular conductive fractures. The results
of the tests are amenable to analysis
whereby the transmissivity of particular
fractures is determined, although this was
not done in this study. The fluid conductivity
logs have higher resolution in general, which
results in part because measurements were
made at many more depths compared to
the other methods.
Straddle-packer injection tests were done
in all nine wells. Water was injected at a
constant rate into a 6 meter packed-off
interval in a borehole. Fluid pressure within
the interval was measured with a pressure
transducer. Two transducers were also
mounted above and below the test interval
to monitor for possible flow leakage around
the packers and/or flow short-circuiting due
to interconnecting transmissive fractures.
For each test interval injection was continued
until a pseudo-steady flow was attained.
This was on the order of minutes to tens of
minutes. This procedure was repeated at
different depths, and the transmissivity of
each interval was calculated using the
observed flow rates and injection pressures,
and assuming radial flow. Short-circuiting
occurred in high conductivity intervals and
transient injection pressure and flow rates
persisted because of flow-line resistance
effects. Therefore, the conductivities of the
highest conductivity intervals were
underestimated. The results are consistent
with the othertechniques but in some cases
indicate very low transmissivity intervals
while the others do not.
2.7 Integration of Geophysical
and Hydrologic Logs
Conventional geophysical logs were used
in conjunction with flowmeter logs in order
to identify the particular hydraulically con-
ductive fractures and/or fractured zones
-------
a) Impeller
b) Heat-Pulse
c) Fluid Conductivity
d) Packer Injection
flow less than
stall velocity
Q [L/min]
-3 -2-101
1 1
[juS/cm] Q/AH [L/min/m]
60 120 180 0 1 2 3
H 1 1 I 1
Figure 5. Example results from the various flow logging techniques performed in well SE-1. The impeller measurements were taken during constant rate
pumping of approximately 15 L/min just above the most shallow measurement. Measurements below 31 m were not possible because of the high
stall velocity of the instrument. Heat-pulse measurements were taken during pumping in well 0-0. Negative flow indicates downward flow.
Upward and downward flow is converging and exiting borehole near 30 m. Fluid conductivity profiles exhibit peaks where formation water is
flowing into well through discrete fractures while the well is pumped above 10 m. Measurements are in micro siemans per centimeter. Straddle-
packer injection tests show the relative magnitudes of transmissivity for 3 m intervals. Values are given as flow rate achieved per given injection
pressure expressed as equivalent hydraulic head.
intersecting the boreholes. ATV, DBS, tele-
vision, 16-inch normal resistivity, caliper,
and natural gamma logs were the most
useful forthis purpose. After specific trans-
missive fractures were identified in each
well, fractures and fracture zones that inter-
sect multiple wells were identified by inter-
polating fracture geophysical properties
using 3-D visualization software.
Figure 6 shows geophysical and flowmeter
measurements, and the identification of the
particular transmissive fractures in well
SE-1, for example. Comparison of the
geophysical logs with one another indicated
that caliper logs show the locations of
increased borehole diameter. Zones of
intense fracturing such as in areas where
there are many closely spaced, sub-
horizontal fractures were easily detected.
Some caliper anomalies are associated with
individual fractures. Peaks on the natural
gamma logs indicated the presence of
pegmatitic dikes. Regions of resistivity less
than 700 ohm-m are associated with
hydraulically conductive fractures. Clays
and ferric oxides associated with
hydraulically altered fractures exhibit high
electrical conductivities compared to the
parent rock.
In general, several different types of
fractures are transmissive: fractured
pegmatitic dikes, closely spaced
subhorizontal fractures, subvertical tectonic
fractures, and to a minor degree various
partially infilled aplite veins which are very
weakly transmissive. The fractured
pegmatite dikes are often found within or
nearthe zones of subhorizontal and closely
spaced fractures. Two subhorizontal
fracture zones comprised of the westwardly
dipping, subhorizontal altered fractures and
pegmatites were delineated based on
similarities in fracture geophysical properties
in adjacent boreholes. These two zones
define general hydrologic structures within
the well field (Figure 3). Other important
features include subvertical tectonic
fractures which may connect these zones.
The report describes how specific
transmissive fractures were identified by
integration of different combinations of
geophysical and flowmeter logs.
2.8 Computer Visualization
Visualization software enabled viewing of
the well field and how particular fractures
are positioned relative to boreholes.
Borehole lateral logs were used in
conjunction with surface mapping to
determine the x,y,z coordinates of the well
field. The true 3-D perspective of all wells
was constructed from these coordinates.
This formed the initial 3-D model upon
which to superimpose other features. One
very useful feature of the model was
6
-------
Natural Resistivity
Gamma (16 in)
SE-1
taliper ATV
Fluid Identified
Electrical Conductive
Conductivity Fractures
u
10 -
20 -
30 -
40 -
50 -
60 -
70 -
I • • i | I ' ' | I ' i ' 1 'I1 1 " 1 ' ' ' I IT!
0 -3 ISO 0 3000 16 18 20 HEP*™ ° 60 120 180 **»•»
,•["•• [QhttrriKtleftS] (ctntjitrtjleisl [azimuth] [(.iS/cm] I&lftf\t,i1h\ _
-
- f f
: t $
- \ \
: f. f
f ™' <-_
- •«* ^f™*
: \ r
: f [
: 1
: F*
^
t' ' ' '
'13
i '
Tf
~S'
™
•Ir-"1*-
1*
!••••£.
: i 7 ^
JT r- \Tf
If
: |
i
a ii
= ! r |
: 1 ^
f>w
t™_
1
~~ J^ J"
: ^ B
_
'
L.
-
-
—
-
_
— j
';, • .: ... - , , - H _:
•'
• *
i'-
'
i, , -
_-
—
-
• • ^,.
_
-
-
-
—
-,
\
0.
»
D
Figure 6. Conventional geophysical logs and borehole fluid replacement and conductivity log of well SE-1. Fractures identified astransmissive shown at
right.
-------
deducing subsurface fracture connectivity
and structure. For example, a transmissive
fracture identified by integrating borehole
flowmeter and geophysical data was
measured using the acoustic televiewer
log, and a plane representing the azimuth
and dip of that fracture at its proper location
in a well could be displayed. This plane was
represented as a disk, and was used to see
where that fracture would theoretically
intersect other wells. The logs of other
wells were inspected to see if such a feature
was present at and around the extrapolated
location, and in this way quickly determined
which features extended between wells.
Extrapolation also considered non-planar
fractures and fracture measurement error.
A complex 3-D model of the site was
developed via this hypothesis testing and
data synthesis approach.
2.9 Cross-Hole Seismic
Imaging
Cross-borehole seismic surveys were
performed to deduce the location of fractures
and fracture zones between wells. The
method consists of equipping a borehole
with a piezoelectric source transducer, and
an adjacent borehole with a string of
hydrophones. The transducer creates a
pressure pulse which is measured by the
hydrophones. Afterthetransducerispulsed
at a particular depth and measurements
are made, it is lowered and pulsed again.
This procedure is repeated along the depth
of the borehole. The seismic data from
each receiver during each depth
transmission is analyzed to determine the
travel time and amplitude of each received
seismic pulse. A numerical matrix that
incorporates the positions of each ray path
and its respective travel time and/or
amplitude is inverted and the velocity and/
or seismic attenuation of different pixel
elements of a 2-D grid representing the
plane between boreholes is determined.
Figure 7 shows the concatenated seismic
velocitytomograms ofwell pairs SW3-SW2,
SW2-00,00-SE2, and SE2-SE3. The results
support the conclusions regarding the
general dual-layer, westwardly dipping
fracture zones.
2.10 Interwell Tracer Tests
Radially convergent and two-well partial
recirculation tracer tests using both
conservative and reactive tracers were
conducted. The objectives of this test were:
(1) to obtain an estimate of transport
parameters; (2) to compare performance of
several different tracers; and (3) to compare
test methodologies and identify advantages
and shortcomings of these methods as
tools to describe fracture flow properties.
The upper fracture zone in each well was
isolated with pneumatic packers during both
tracer tests. A minimum of two packers
were used in each well with a total of 22
packers in 9 wells. Twenty-nine (29)
transducers were located within and around
packed-off zones throughout the well field.
For the radially convergent test, well 0-0
was first pumped at a constant rate for a few
days prior to the tracer injection to establish
a quasi-steady flow field. A three tracer
mixture of deuterium, fluoresceine, and
microspheres was subsequently injected at
the upper fracture zone in well SW-3. A
mixture of bromide and fluoride was injected
in the same zone two and a half hours later.
The pumping continued forabout one week,
during which time the pressure in all 29
zones and the flow rate atthe discharge line
were continuously monitored and recorded.
Also monitored in real-time were the
fluoresceine and bromide concentration in
the discharge line using an in-line
fluorometer and ion specific electrode as
described in section 1.3. Sample water was
taken from the discharge line every ten
minutes using the automated sampling table
SW-3
0-0
SE-3
-3C 00
-0.3
-10,00 1000
DISTANCE [mj
VELOCITY DEVIATiON [km/s]
3000
0.3
Figure 7. Concatenated seismic velocity tomograms from adjacent well pairs. Image represents
cross-section between well SW-3 and SE-3. Negative velocity deviation represents seismic
velocity less than mean value. An upper and lower zone is apparent.
-------
C/Co
0.001
0.0008
0.0006
0.0004
0.0002
n
•
I
flouresceine
bromide
deuterium
20 40 60
time [hrs from injection]
80
Figure 8. Normalized breakthrough curves of three conservative tracers during radially convergent
test.
for later analysis in the laboratory.
Figure 8 shows the breakthrough curves
of fluoresceine, bromide, and deuterium
normalized to the injection concentration.
Microspheres were not detected in the
pumped water. It is possible that the
microspheres became negatively charged,
coalesced and, hence, became stuck within
the fracture zone. The background
concentration of fluoride was too high for
the fluoride concentration result to be
meaningful.
The first tracer arrival occurred at about
10 hours after injection. The tracer
breakthrough curves did not lie on top of
each other, even through the three tracers
are presumably conservative (Figure 8).
The evidence suggests that the most
significant factor responsible for this
behavior was that effective injection tracer
concentrations were different than predicted
because of borehole mixing and storage
effects. After the mixture of tracers was
injected, the tracers may have separated
within the injection zone, which was roughly
ten times the volume of the injected fluid.
Due to its higher density, bromide may have
caused the tracer mixture plume to sink to
the bottom of the injection zone. The mass
of fluoresceine that arrived at the pumping
well was calculated to be about 15% of the
total mass injected. Almost all of the
remaining mass was recovered by pumping
outfromthe injection well. Othercontributing
factors may include differences in the degree
of reaction (or non-reaction) between the
tracers. Fluoresceine has been reported
elsewhere to react (fluoresce more) with
certain minerals intherockgiving increases
in apparent mass.
A partial-recirculation injection and
recovery tracer test was conducted using
both reactive and conservative tracers. The
extraction and injection wells were the same
as in the previous test. The extraction and
reinjection rates were 7.6 and 0.7 L/min,
respectively. Bromide and fluoresceine
were used as conservative tracers and
lithium was used as a sorbing tracer. A
specially designed low volume tracer
injection system was used to minimize the
well bore effects observed in the previous
test (Figure 9). The test was halted after 22
days, when the fluoresceine concentration
dropped to approximately one tenth of the
peak arrival concentration.
There was a significant difference
between the recovery of bromide and lithium
during the 22 days in which the test was
conducted. The percentage of bromide
recovered was 80% of the injected mass,
and less than 30% of the lithium was
recovered. In addition, the breakthrough
curve for bromide was much steeper with a
peak arrival two days before the peak for
lithium. A simple convective-dispersive
transport equation was used to determine
transport parameters from field data. A
linear relationship was assumed to exist
between fluid concentration and adsorbed
phase, with instantaneous chemical
equilibrium. Transport is modeled along a
1-D stream tube between the injection and
withdrawal well. This is a reasonable
approach for this test since the ratio of fluid
injection to discharge is small. A dispersion
length of 32 meters was calculated from the
best-fit model. As described subsequently,
the relevance of this parameter is very
questionable.
3 Conclusions and
Recommendations
3.1 Surface Fracture
Characterization
Observation and measurement of surface
geologic features is a standard and usually
initial investigative phase in any subsurface
characterization effort. A non-intrusive,
relatively low cost surface study combined
with previously documented geologic studies
can provide a means to determine the
dominant types and orientations of fracture
sets present, as well as their physical
properties, spatial distribution, and probable
modes of genesis. The relative ease and
low cost of performing surface
characterization and the pertinent
information that it reveals makes it a practical
necessity, but it is not a substitute for
borehole studies. For example, most of the
fractures observable at the surface may not
be hydrologically significant. Dominant
conductive fractures in the subsurface are
not typically observable on the surface, or
only limited fractures may be observable at
the surface, simply because few bedrock
outcrops are present.
3.2 Well Drilling
Air drilling is the most appropriate method
for the installation of wells in crystalline
bedrock in terms of its relative cost,
penetration rate, and potential to yield
relevant hydrogeologic information.
Observation of drill cuttings, drilling rate,
and flow out of the borehole during drilling
may be used to infer lithology and the
location of transmissive fractures, for
example. The rate of penetration offers an
obvious cost advantage over other methods,
but a greater savings is probably due to the
increased information gained concerning
fluid-bearing zones. Careful observation
during drilling and a descriptive log can
provide valuable hydrogeologic information.
Of great significance is the fact that the
depth of transmissive fractures may be
-------
Injection Tank
Discharge to Sarnpkr and
Fluyi firi'p tHi
I
I
Packer
Figure 9. Equipment layout for two well tracer tests with partial recirculation. Tracer is injected into
small annulus between borehole wall and injection casing between packers. Tracer exits
through several feed-through valves evenly distributed over injection interval. Not to scale.
observable, because water will be brought
to the surface with the cuttings when such
fractures are encountered. Flow can be
measured with a bucket at the surface, and
changes in flow with depth may indicate
that the drill bit encountered another
transmissive fracture or fracture zone. In
addition, the flow measured at the total
depth may be used as a relative measure of
well yield. This information can be used to
assess which wells are relatively high
yielding and, therefore, are the best
candidates for pumping tests or borehole
flow profiling tests, for example.
3.3 Pumping Tests
Drawdown transients in different types of
fractured formations can behave in a manner
describable by the Theis or Cooper-Jacob
solutions, but this behavior does not
necessarily indicate the aquifer can be
considered an equivalent porous medium.
Irrelevant parameter values can be derived
from analysis of drawdown transients, and
analysis of late-time drawdown may be
essentially unrelated to the region of the
aquiferthat is of interest. Well bore storage
can mask the effects of important
heterogeneities, and semilog analysis of
drawdown in the pump well will lead to
underestimates of transmissivity. Local
transmissivity heterogeneities can produce
apparent anisotropic behavior in the
drawdown responses in observation wells.
The installation of wells can connect
previously unconnected fractures and create
short-circuit pathways. Pumping of wells
can further initiate transport of contaminants
to previously uncontaminated areas. In
orderto maximize the information necessary
for characterization and minimize cost and
the potentially negative effects of pumping,
pumping tests should not be conducted as
tests in and of themselves. Other hydrologic
tests which are necessary in characterizing
a fractured formation yield more valuable
information, and can simultaneously provide
pressure transients amenable to standard
pumping test analysis. For example,
impeller-flow meter profiling allows one to
determine which fractures are conductive,
and pressure transducers can be installed
in wells during the test to measure pressure
transients. A drilling method that enables
observation of fluid emanating from the
borehole can provide information regarding
which well(s) are best suited for pumping.
Pumping of wells during any type of test
should be minimized in duration, and efforts
should be made to analyze the entire
pressure transient rather than resort only to
late-time data analysis.
3.4 Detection and Measurement
of Subsurface Fractures
The identification and measurement of
fractures intersecting boreholes is an
absolute necessity. Besides its use in
determining the general fracture trends or
sets, a database of fracture locations and
orientations in boreholes enables one to
determine fracture-specific geophysical
properties through comparison with
standard geophysical and flow logs. The
combination of a flowmeter log and an
acoustic televiewer log may enable the
determination of particular hydraulically
conductive fractures. The measured
orientation and dip may be used to associate
these fractures with fracture sets, and project
to intersections with other borehole wells.
Although the digital borehole scanner
provides the highest resolution data, it is
relatively expensive. However, given the
capabilities of the tool and imminent reduc-
tions in its cost, it will be the best tool for use
in the near future. At present, ATV and
television logs used together provide very
useful fracture data, especially when
coupled with standard geophysical logs.
Fractures can be easily mapped from the
ATV logs, and television logs can be used to
check if particular fractures are altered and/
or infilled. The ATV was most useful for
identifying the location of fractures and
determining their orientation. Measure-
ments of individual fracture properties, such
as aperture or roughness are not reliable.
The determination of orientation and dip of
individual fractures was very useful for later
use, when the data was integrated with
visualization software and measured frac-
tures were extrapolated to see where they
might intersect other boreholes. Most of
the information gained from coring is ob-
tainable through the combined data from
the ATV, television, and standard geophysi-
cal logs, which are more economical and
timely.
The large data set resulting from the
fracture measurements makes a statistical
analysis of spatial properties very tempting.
However, because of the effects of drilling
on the appearance of individual fractures,
and on the fact relatively few of the many
fractures detected conduct fluid, statistical
descriptions of fracture aperture and spacing
(based on ATV data alone) should not be
10
-------
used to estimate general hydrologic
properties.
3.5 Borehole Flow Logging
Flow logging is a critical necessity in the
characterization study. It provides a means
to identify and quantify the transmissivities
of onlythe relatively few fractures orfracture
zones which are in fact conductive. This
information provides knowledge of the
general structure of the aquifer, from which
all future remedial planning emanates.
Results from flow logging are best when
integrated with other geophysical logs.
A necessity of all flowmeters is that
pumping or injection of the well-bore is
required. Therefore, a flow field may be
established in the contaminated field, and
contaminants can migrate to previously
uncontaminated areas. In orderto minimize
these effects, profiling should first be
conducted in wells under natural conditions.
In theory, either the impeller or thermal-
pulse flowmeter could be used at this stage,
although the lower sensitivity of the latter is
probably the better choice since flows are
likely to be very low. Vertical borehole flow
will probably be greatest soon after drilling
when fractures are being connected by the
installation of the well and the system is in
a highly transient state. After this initial
profiling, the method of profiling multiple
wells during the pumping of a single well
should be implemented. The highestyielding
well should be used as the pumping well,
and driller's logs can providethis information.
Profiling in the pump well with the impeller
flowmeter allows determination of
transmissivities of intervals adjacent to the
well, and by installing pressure transducers
in neighboring wells, a multi-well pumping
test is effectively conducted. This eliminates
the need to perform a separate pumping
test. Alternatively, one well can be used for
pumping and other wells profiled with a
heat-pulse flowmeter, for example. The
information gained is somewhat different,
but perhaps more informative since one
can determine which fractures are
transmissive in other wells, although not
quantitatively. If the impeller or thermal-
pulse is used, it is recommended that a
downhole inflatable packer be used around
the flow casing to increase the sensitivity
and avoid borehole variation and turbulence
effects and to reduce the number of
calibrations needed.
The use of packer injection tests is not
recommended under most situations. The
method is expensive and time consuming
and creates the greatest non-equilibrium
condition in the aquifer. A method like the
fluid conductivity logging yields precise
locations of transmissive fractures and can
be analyzed to determine the
transmissivities of these fractures. The
packer-injection tests are probably not
appropriate as a means to detect flowing
fractures but perhaps as a later investigative
phase once flowing fractures are found and
some quantitative assessment is sought.
3.6 Integration of Geophysical
and Hydrologic Logs
Conventional geophysical logs were used
in conjunction with flowmeter logs in order
to identify the particular hydraulically
conductive fractures and/or fractured zones
intersecting the boreholes. It was found
that by integrating the flowmeter results
with the acoustic televiewer, television, 16-
inch normal resistivity, caliper, and gamma
logs, the particular conductive fracture or
fractured zone could be determined. After
the conductive fractures were identified,
interpolation of properties between wells
was made based on similarities in various
fracture geophysical properties, and the
general hydrologic structure of the aquifer
was deduced. The integration of various
geophysical logs is an essential component
in any characterization effort. All of the
conventional geophysical logs collected may
be potentially significant for use at other
sites. Reference is made in the full reportto
other works where different tools have been
used.
3.7 Computer Visualization
Visualization software is nowan affordable
reality, and allows one to deduce
complexities not possible from traditional 2-
D plots of borehole data. One very useful
feature of some visualization software
routines is their ability to represent spatially
distributed data in a true three-dimensional
perspective, and which allow real-time
manipulation of the viewing perspective.
Hypothesis testing via interpolation of
fracture properties between wells is
extremely beneficial. Visualization should
be a commonplace tool for subsurface
characterization.
3.8 Interwell Tracer Tests
In general, tracer tests were the most
difficult field test to construct and operate,
even with the highly sophisticated and
expensive equipment used at the Raymond
Field Site. Well bore storage and mixing
effects in the injection zone severely altered
the intendedtestconfiguration and rendered
the tracer breakthrough curves amenable
to arrival time analysis only. Even with the
implementation of the special injection
system that minimized borehole effects,
the usefulness of the test results is very
questionable. The theoretical mixing lengths
that yielded the best model fit to the data
were larger than the well spacing between
the injection and withdrawal wells. This
result generally points out the shortcomings
of using the convection-dispersion equation
to model transport in highly heterogeneous
fracture rock. Based on the field experience
at this site, tracer tests for characterizing
fracture formations as part of a remediation
program are problematic. Highly controlled
and sophisticated and expensive equipment
is needed, and even in the best of
circumstances there will most likely be
considerable ambiguity associated with the
derived parameters.
Notice:
The information in this project summary
and in the original report is the result of a
research project funded jointly by the U.S.
Environmental Protection Agency and the
U.S. Department of Energy. Neither the
original report nor this project summary
should be interpreted as official EPA
guidance for site characterization at sites
located in fractured rock formations.
11
-------
Andrew J. B. Cohen, Kenzi Karasaki, Sally Benson, Gudmundur Bodvarsson, Barry Freifeld, Pascual Benito, Paul Cook, John Clyde, Kenneth
Grossenbacher, John Peterson, Ray Solbau, Bhaskar Thapa, Don Vasco, and Peter Zawislanski are with the Earth Sciences Division, Ernest
Orlando Lawrence Berkeley National Laboratory, University of California, Berkeley, CA 94720
Stephen R. Kraemerwasthe EPA project officer (see below).
The complete report is entitled "Hydrogeologic Characterization of Fractured Rock Formations: A Guide for Groundwater Remediators" and is
referenced as LBL-38142/UC-800 (144 pages, color figures). A limited number of copies of the full report are available from Subsurface Remediation
Information Center, P.O. Box 1198, Ada, OK 74821. A digital copy is available on the home page http://www.epa.gov/ada/kerrlab.html
The full report is available to DOE and DOE Contractors from the:
Office of Scientific and Technical Information
P.O. Box 62
Oak Ridge, TN 37831
(615)576-8401
The full report is available from the:
National Technical Information Service
U. S. Dept. of Commerce
5285 Port Royal Rd.
Springfield, VA 22161
The EPA Project Officer can be contacted at:
U.S. Environmental Protection Agency
National Risk Management Research Laboratory
Subsurface Protection and Remediation Division
P.O. Box1198
Ada,OK74821-1198
------- |