. (rt... 231.: 1. 7.!
A V .(«anu 9.3 .91”
. 584 ‘1: . 1.7 ..
u 3.5x, ...r...MN.....'.A.v.a.h....

.thal

. .. LﬁmeNﬁ

H .‘ n n .
w. , . . . . ‘ HA.» .
. .. . . r .
,. "$3.? .. in.» ﬂow. ..
. WI. . IIIDooOII..Un , . .. . A... . s . ._
$.WuHM»...WHWt.IR5M.WW-ﬂ?£. .nﬂﬁWv‘ . ‘ , . V . . .

taut“

o .

twig..." .
we..- . n::¢.m...aﬂ.n.z ..
.YrﬂNUl‘wiuﬁmmyNiu. mu . 5%}... A

 

. mi .‘mwM.

.. . ‘. . . . .7 . . ... .. ‘.. . .. .Ifoovmvhrr.ﬂ.ﬁ‘m
.. . ... “.muzﬂsb; a . n . s“ , . . . olts
.%|1¥.. ..-i,...!.f.h . .. .. . . . .

 

 

O
‘ . . .cho “ION“ cunt 1.th

¢ - 0'
7.1. . .: ifﬁ. I

Out;

1 »VI .
lit. > a

‘ 19m

‘

. 13f ...u 1 I...
...H.H.l.l. LU f5. n- .thtohl.‘o
. 0.

vi‘

0.5. .‘ nl.

.Pgﬂu‘“ I 9|.Lo.v tit.
PI

5.;

:4!
v0 I.
n .

.“1

1%...

l
3

.310“ .. Id

1 .ll‘.

- 2-.

 

 

 

:1. ANN:

.vv. I.$OW ..

u . .
114:0

 

.5:

f' 3h;
4, 1w j’

«'1 a: "i" "
‘.. L‘"
'
h- .
‘ k :‘."
H
Y

1“

v
I
IN
..‘

if t.
....m.4.~... 1m.

5

.'-l
V I

M

‘35)!99f
I: 'A ‘ '1.
‘ . .

WT

.e

YDE»!.'.IV;.IAV .
sub-\I ‘ .

..II. un .77.
.{nt’trlﬂ'p'
. .1 ‘ . .1

 

 

THESIS

l
(DH)

 

Michigan State
University

 

 

 

This is to certify that the

thesis entitled

GROUND-WATER FLOW IN A CLAY-RICH TILL
IN SOUTHEASTERN MICHIGAN: THE ROLE OF
FRACTURES

presented by

Carol Lee Luukkonen

has been accepted towards fulfillment
of the requirements for

M - S - degree in JeologL

94/ A4

professor

 

Date 9%?" 174 4, /??¥

0.7639 MS U is an Afﬁrmative Action/Equal Opportunity Institution

micucANS"

LIBRARY ‘llllll \l l\\\\\\\\l llll

\\\\\\\\l

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINE return on or before date due.
MAY BE RECAU£D with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1M mu

GROUND-WATER FLOW IN A CLAY-RICH TILL
IN SOUTHEASTERN MICHIGAN: THE ROLE OF FRACTURES

By

Carol Lee Luukkonen

A THESIS
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Geological Sciences

1997

ABSTRACT

GROUND-WATER FLOW IN A CLAY-RICH TILL
IN SOUTHEASTERN MICHIGAN: THE ROLE OF FRACTURES

By

Carol Lee Luukkonen

A four layer steady-state ground-water flow model was developed to simulate ground-
water ﬂow at the study site near Milan in southeastern Michigan using data collected during this
and previous investigations. The fractured clay-rich till units overlying bedrock were treated as
equivalent porous media. Particle tracking was used to investigate the observed distribution of
tritium at depth and the proposed upward hydraulic gradient beneath the Arkona Road Landﬁll
site. Leakances between layers were systematically varied to simulate the effect of fractures and to
observe effects on simulated heads.

Particle-tracking results indicate that the model does not accurately represent the fractures
which create localized flow systems at a smaller scale than the grid spacing of the model. The
minimum travel time for particles from the top of layer one to layer two was 15.2 years, however,
the minimum travel time to layer three where tritium was detected in the water was 6,000 years.
Particle tracking does indicate that all ground-water ﬂow is downward. Finally simply increasing

the leakances between layers does not alone improve the simulated heads.

ACKNOWLEDGMENTS

I wish to extend special thanks to my thesis advisor, Dr. Grahame Larson, for the many
hours he spent at the study site measuring fractures, for his assistance during model construction
and calibration, for his comments that improved this thesis, and for not giving up on me. I also
extend special thanks to Dr. Michael Velbel and Dr. William Cambray for taking the time to serve
on my committee and for their comments that improved this thesis. I am in debt to all of my
committee members for the knowledge I have gained during this study and during coursework.
Thanks also go to the US. Geological Survey for employing me during this study and for
increasing my knowledge of modeling, data analysis, writing, and ﬁeld work. And last but
deﬁnitely not least, this study would not have been possible without the support and encouragement
of my son, Ben, who kept me company while I worked and often clicked the mouse button for me,
and my husband, Dave, who kept me focused on the study and always came up with good

suggestions when I needed them the most.

iii

TABLE OF CONTENTS

List of Tables ................................................................................................................... vi

List of Figures .................................................................................................................. vii
Introduction ...................................................................................................................... 1
Previous Studies ............................................................................................................... 2
Study Area Description ..................................................................................................... 5
Geology ............................................................................................................... 7

Hydrology ............................................................................................................ 10

Fractures ............................................................................................................. 14

Simulation of Ground-Water Flow .................................................................................... 21

Conceptual Model ................................................................................................ 21

Numerical Model ................................................................................................. 23

Model Grid and Layers ............................................................................ 23

Boundary Conditions ............................................................................... 24

Sources and Sinks .................................................................................... 25

Parameter Values .................................................................................... 25

Model Layers .............................................................................. 25

Ground-Water Recharge Rates .................................................... 26

Hydraulic Properties ................................................................... 26

Model Calibration .................................................................................... 29

Simulation Results ................................................................................... 31

Particle Tracking .................................................................................................. 35

Tritium Data .................................................... ' ....................................... 37

Proposed Upward Flow ........................................................................... 38

Analysis of Fractures ........................................................................................... 39

Model Limitations ............................................................................................................ 41
Summary .......................................................................................................................... 42
Conclusions ...................................................................................................................... 44
Appendix .......................................................................................................................... 46
List of References ............................................................................................................. 68

LIST OF TABLES

Table 1. Hydraulic Conductivity Values ........................................................................... 12
Table 2. Hydraulic Conductivities Assigned to Strata Codes ............................................. 27
Table 3. Optimal Parameter Estimates ............................................................................. 31
Table 4. Effects of Changing Leakances on Simulated Heads ........................................... 40

vi

LIST OF FIGURES

Figure 1. Location of the Envotech Resource Center Site .................................................. 6

Figure 2. Generalized Cross-Section ................................................................................ 8

Figure 3. Fracture Orientations Measured by RMT (1992) ............................................... 15
Figure 4. Fracture Orientations on the Sloping Surface of the Unit 2 Till .......................... 18
Figure 5. Fracture Orientations in the Gully of the Unit 2 Till ........................................... 19
Figure 6. Relation Between Measured and Simulated Heads for Layer One ....................... 32
Figure 7. Relation Between Measured and Simulated Heads for Layer Two ...................... 33
Figure 8. Relation Between Measured and Simulated Heads for Layer Three .................... 34
Figure 9. Relation Between Measured and Simulated Heads for Layer Four ...................... 36
Figure 10. Model Grid for Layer One ............................................................................... 46
Figure 11. Model Grid for Layer Two .............................................................................. 47
Figure 12. Model Grid for Layer Three ............................................................................ 48
Figure 13. Model Grid for Layer Four .............................................................................. 49
Figure 14. Land Surface Elevation ................................................................................... 50
Figure 15. Top of Layer Two ........................................................................................... 51
Figure 16. Top of Layer Three ......................................................................................... 52
Figure 17. Top of Layer Four .......................................................................................... 53
Figure 18. Measured Ground-Water Levels in Layer One ................................................ 54
Figure 19. Measured Ground-Water Levels in Layer Two ................................................ 55

vii

Figure 20.
Figure 21.
Figure 22.
Figure 23.
Figure 24.
Figure 25.
Figure 26.
Figure 27.
Figure 28.
Figure 29.

Figure 30.

Figure 31.

Measured Ground-Water Levels in Layer Three .............................................. 56
Measured Ground-Water Levels in Layer Four ................................................ 57
Estimated Horizontal Hydraulic Conductivity of Layer One ............................. 58
Estimated Leakance Between Layers One and Two .......................................... 59
Estimated Transmissivity of Layer Two .......................................................... 60
Estimated Leakance Between Layers Two and Three ....................................... 61
Estimated Transmissivity of Layer Three ......................................................... 62
Estimated Leakance Between Layers Three and Four ....................................... 63
Simulated Ground-Water Levels in Layer One ................................................. 64
Simulated Ground-Water Levels in Layer Two ................................................ 65
Simulated Ground-Water Levels in Layer Three .............................................. 66
Simulated Ground-Water Levels in Layer Four ................................................ 67

viii

INTRODUCTION

The study of fractures in clay and methods of modeling ground-water ﬂow through a
fractured media are becoming important areas of research. Studies have shown that the presence
of fractures can increase the bulk hydraulic conductivity of till and clay deposits and can provide
avenues of ground water and contaminant transport at faster rates than possible through the matrix
(Gerber and Howard, 1992; Herzog and others, 1989; Grisak and Cherry, 1975; Grisak and others,
1976; and McKay, 1991). The presence of fractures in many deposits has been conﬁrmed by the
detection of tritium or other isotopes at depth beneath older water (Foote, 1989; Hendry, 1982 and
1983; Rudolph and others, 1991; and Ruland and others, 1991), order of magnitude differences in
ﬁeld and laboratory measured values of hydraulic conductivity (Fredericia, 1990; Grisak and
Cherry, 1975; Keller and others, 1986 and 1988; McKay and others, 1993; Ruland and others,
1991; Thomson, 1990; and Williams and Farvolden, 1967), or visual observations of fractures at a
site (Herzog and others, 1989).

The presence of fractures in two clay-rich till units in southeastem Michigan has been
conﬁrmed both by the detection of tritium at depth (Foote, 1989) and by visual observation (RMT,
1992). This site is located in southeastern Michigan approximately 2.5 miles east of the city of
Milan in Augusta Township, Washtenaw County. The fractured till units are separated by a
lacustrine clay and overlain by a lacustrine sand. Subcropping beneath this area are shale and
limestone units. Based on an extensive hydrogeological investigation, RMT predicted that an

upward hydraulic gradient exists beneath the Arkona Road Landﬁll which overlies the fractured

2
clay-rich till units in this area. RMT developed a ground-water-flow model to evaluate ground-

water ﬂow in the upper lacustrine sands at the site, however, their model did not incorporate the
underlying till and bedrock units.

This study was undertaken to develop a model to explain the presence of tritium at depth
and to investigate the proposed upward hydraulic gradient. This model will include the fractured
till units underlying the upper lacustrine sands, as well as the bedrock units underlying the site.
Particle tracking, in which hypothetical water particles are tracked as they move through the
simulated ground-water-ﬂow system, will be used to determine whether water can move downward
at a fast enough rate to allow tritiated water to reach the observed depths and whether there is an

upward hydraulic gradient beneath the Arkona Road landﬁll.

PREVIOUS STUDIES

The Arkona Road Landﬁll (ARL), located within the study area, was operated as a waste
disposal site from April 1973 until its closure in March 1979 (RMT, 1992). The landﬁll is located
within the upper fractured clay-rich till unit. Subsequent investigations in 1987 led to the
conclusion that there was potential for contaminants to migrate beyond the landﬁll boundaries and
that additional work was needed to address the potential for releases (RMT, 1988). Visual
observations near the ARL revealed seepage from the landﬁll cover and some surface water
samples near the ARL suggest contributions from the landﬁll (RMT, 1992).

Although ground-water samples collected by RMT do not clearly demonstrate water
chemistry changes due to the ARL, the presence of fractures implies that an effective barrier to
contaminant migration from the landﬁll may no longer exist. A similar conclusion has been

suggested at other sites underlain by tills and argillaceous sediments. For example, Herzog and

3
others ( 1989) discovered contamination in a monitoring well in Illinois and noted organic chemicals

had migrated much faster than expected from a disposal facility. Subsequent studies by Herzog
and others revealed that previously undiscovered fractures in the glacial till were responsible for
the rapid migration of chemicals. Also, Grisak and Cherry (1975) studied a clay-loarn till and
lacustrine clay in southeastern Manitoba and found that hydraulic conductivity and storativity of
these deposits were primarily controlled by the fractures.

Study of a glacial till in Alberta likewise revealed that pre-bomb ground water overlies
tritiated ground water (Hendry, 1982). Hendry suggested that movement of ground water through
fractures could explain the presence of tritiated ground water at depth. Rudolph and others (1991)
found variations with depth of major ions, oxygen 18, and deuterium at their study site in a
fractured lacustrine clay near Mexico City. Ruland and others (1991) observed tritium deeper than
expected in tills in southwestern Ontario.

Grisak and others (1976) observed that ground-water velocities in fracture networks were
much larger than those in unfractured till masses. The presence of ground-water ﬂow through
fractures alone, however, does not completely account for the observed concentrations of chlorides
and other soluble salts in much of the till waters in the Plains Region. Diffusion must be considered
also (Grisak and others, 1976). Day (1977) found that diffusional exchange with pore water could
explain the absence of tritium and oxygen 18 at depth observed in a ﬁactured till in Manitoba.
Grisak and others (1980) studied solute transport through a cylindrical column of fractured clayey
glacial till. The breakthrough curves indicated that calcium and chloride were affected by diffusion
from the fractures into the porous matrix of the till. Thus, "analyses of contaminant migration
through fractured till and of ground water origin and age based on the distribution of isotopes such
as '80, D, 3H, and 1‘c in fractured un should account for the effect of matrix diffusion" (Grisak

and others, 1980).

4
Methods of modeling fracture systems have been developed by many researchers

(Berkowitz and Braester, 1991; Grisak and Pickens, 1980; Hull and others, 1987; Moreno and
others, 1990; Rowe and Booker, 1990 and 1991; Snow, 1969; Sudicky and McLaren, 1992;
Sudicky and Frind, 1982; Thoma and others, 1992; and Zimmerman and others, 1992). Fracture
systems can be modeled using (1) equivalent porous media, (2) discrete fracture network, or (3)
dual porosity models (Anderson and Woessner, 1992). Harrison and others (1992) observed that
the discrete network approach ignores diffusion, and that equivalent porous media and dual
porosity model approaches have questionable results when extended to greater depths where the
fracture frequency is lower. Long and others (1982), Long and Witherspoon (1985), and Long and
Billaux (1987) have investigated treating a fractured network as an equivalent porous media and
Long and others (1982) found that a fractured system behaves more like a homogeneous,
anisotropic material with an increase in fracture density, nonuniform orientations, uniform
apertures, and constant gradients for each representative elementary volume. Subsequent studies
by Long and Witherspoon (1985) observed that "systems with longer but less dense fractures
behave more like porous media than systems with shorter, but more dense fractures". Systems
with longer fractures tended to have a higher degree of interconnection which Long and
Witherspoon found to affect fracture permeability. Long and Billaux (1987) concluded that an
equivalent porous media approach was generally not valid for their two-dimensional fracture
network due to the low degree of fracture interconnection. However, Long and Billaux's analyses
indicated that a three dimensional system may be well connected. Other researchers doubt that
equivalent porous media models adequately represent fractured networks (Cacas and others, 1990a
and 1990b). Gale (1982) found that systems with a low matrix hydraulic conductivity and a low

fracture density could not be appropriately modeled as an equivalent porous media. Berkowitz and

5
others (1988), however, observed that if field estimates of hydraulic parameters were available, an

equivalent porous media model could be constructed.

A problem in representing a fractured system as an equivalent porous media lies in
determination of the effective hydraulic parameters. Keller and others (1986) showed that the bulk
permeability of a clayey till near Saskatoon, Saskatchewan exceeded the matrix permeability by
two orders of magnitude. Fredericia (1990) observed that laboratory measured values probably
only represent the hydraulic conductivity of the matrix material between the fractures. Keller and
others (1988) concluded that bulk permeability was greater in fractured till compared to
unfractured till. Grisak and others (1976) observed that ground-water velocities were many orders
of magnitude larger in fracture networks than in unfractured till masses.

Additional problems result since ﬁeld-measured values of hydraulic conductivity can be
inﬂuenced by smearing during installation of measuring equipment (McKay and others, 1993).
McKay and others (1993) found that use of specially designed augers that minimize smearing
produced higher estimates of hydraulic conductivity. Measurements of hydraulic conductivity are
further affected by the physical scale (McKay and others, 1993; and Brunet, 1993). Bruner (1993)

found that the bulk hydraulic conductivity increases as larger volumes of till are tested.

STUDY AREA DESCRIPTION

Approximately 2.5 miles east of the city of Milan, Envotech Resource Center (ERC) has
proposed locating a hazardous waste management facility in Augusta Township, Washtenaw
County, Michigan (Figure 1). The ARL site is located approximately three-quarters of a mile west
of the proposed waste management facility and an abandoned limestone quarry is located

approximately a quarter mile south of the ARL site. Data on the geology and hydrogeology of the

.85 “2:00 8.508». £8.95 2: .3 5:83 ._ 2:9..—

 

 

 

 

.Ewu 80.3 Good 0
gm
mZOkwm 3
(my? E0 omeaOmn.
: WOKZOI
I J w<~$ t
520 :(ucu a
:2
«.02 lo
(m ZOFaw mm mhm<>> an. e. .12 =96
\ .375 O <20xm<
03029.
T c \

 

 

 

 

7

proposed facility site and surrounding area is available from numerous investigations (RMT, 1992;
Neyer, Tiseo, and Hindo, 1987; Huntec, Ltd., 1967; and McNamee, Porter, and Seeley, 1965).

In 1965, McNamee, Porter, and Seeley advanced 29 soil and rock borings and 11 test
holes for Martin-Marietta Company who was considering developing the site as a limestone
quarry. In 1967, Huntec, Ltd. investigated the elevation and stratigraphy of the bedrock using
resistivity and seismic surveys with borings into the bedrock and in 1987 Neyer, Tiseo, and I-Iindo
(NTI-I) advanced 50 borings to depths ranging from 55 to 125 feet with 5 borings to the bedrock
for NTI-I who was assessing the property's potential for development of hazardous waste facilities
for Envotech. Between 1987 and 1990 RMT advanced 278 monitoring wells and soil borings
throughout the proposed study area for the Augusta Development Corporation, a corporation
related to Envotech. These borings have provided detailed information on the subsurface
stratigraphy of the area.

The proposed landfill area includes 518 acres and an adjacent waste reception area
includes an additional 40 acres. The land surface slopes gently towards Lake Erie. Sugar Creek
crosses the northeast comer of the site. The York and Augusta Drains cross in the south and the

Murray Drain crosses the southwest comer of the site (RMT, 1992).

Geology

At the study site, glacial deposits were deposited during the Wisconsin glacial (Dorr and
Eschman, 1970). The sequence of deposits at the site, which includes two till units separated by a
lacustrine clay, indicates the presence of multiple advances of ice. At the Milan study site the
glacial deposits range from 84 to 119 feet thick (RMT, 1992) and overlie shale and limestone
bedrock units of Devonian age (Figure 2). Rocks of Mississippian, Pennsylvanian, Permian,

Triassic, Jurassic, Cretaceous, and Tertiary age are missing in this area.

Approximate
Depth

 

O
10

 

 

 

 

 

 

 

 

 

 

 

 

 

50 ..
. - ‘ ° . . - . . Li
a - LOWER GLACIALTILL. - . . o
' ' ° ' b. . . , ‘ o
- I 0' ’ ' O ' " , o . O .. -
- ' ' o TRAVERSE SHALE .. °
100 WW 0‘ , , o
no- LI1%1;L:lvlefl._lletlt:
L] T I J r 1 DUNDEE LIIMESJTONIE r T 1 LI ‘ 14
T I t t I I t l I l l I

Figure 2. Generalized Cross-Section.

9
The glacial deposits underlying the proposed facility site have been divided into four units

(RMT, 1992). These units include: 1) surﬁcial lacustrine deposits, 2) upper glacial till unit, 3)
lacustrine silty clay deposits, and 4) lower glacial till unit. Lacustrine/outwash lenses are found
within the glacial till units. Surface exposures reveal that the glacial till units are weathered and
fractured.

The upper lacustrine deposits, unit 1, range from 5 to 15 feet thick and consist primarily of
sand and silt. Unit 1 was deposited at the bottom and on the shore of a glacial lake (RMT, 1992).
According to RMT (1992) a melting glacier occupying the Lake Erie Basin blocked drainage to the
east and caused the formation of a glacial lake in front of the ice margin. Lake level changes,
shoreline location changes and different sediment sources, as well as more recent reworking by
wind and streams, have introduced some variability into this deposit (RMT, 1992).

The upper glacial till, unit 2, ranges from 20 to 50 feet thick and consists primarily of a
silty clay or clay. The till is relatively homogeneous but does contain some sand and sand and
gravel lenses. This upper till unit was deposited by the ice of the Erie-Huron Lobe (Kunkle, 1961,
as referenced in RMT, 1992) and extends to the cast into the present Lake Erie Basin (RMT,
1992). The upper till unit is visibly weathered and fractured in surface exposures. These fracture
surfaces show geochemical changes, including iron oxide and possible manganese oxide staining.
Calcite and gypsum precipitation is also visible. The number of fractures decreases with depth. At
about 35 feet below the surface of this till unit, secondary mineral deposition is rare (RMT, 1992).

Underlying the upper till unit is a lacustrine clay, unit 3, which averages 10 feet thick over
the site and consists primarily of a silty clay. This lacustrine clay was deposited at the bottom of a
pro-glacial lake and was later over-run by the glacier that deposited the overlying till unit (RMT,

1992).

10
The lower glacial till, unit 4, is approximately 55 feet thick over the area and consists of

multiple till units. Unit 4 is slightly coarser-grained than the upper till, unit 2. Multiple earlier
advances of the ice in the Erie-Huron Lobe deposited this unit (RMT, 1992).

The two Devonian bedrock units subcrop beneath the Milan study site and are, on the
average, 110 feet below the ground surface. The contact between the Dundee Formation and the
overlying Traverse Shale crosses the study area from southwest to northeast. These units dip
gently towards the northwest. The Traverse Shale is very ﬁne grained with numerous fossils and
some pyrite. This shale ranges from 0 feet thick in the southeast portion of the site to 40 feet thick
in the northwest portion of the site (RMT, 1992). This bedrock unit is permeable in the northwest
section of the study area, possibly due to bedding plane fractures (Snow, 1967). The Traverse
Shale unconformably overlies the Dundee Limestone. The Dundee Limestone is a ﬁne to coarse
grained dolornitic limestone and dolomite with fossils (RMT, 1992). Snow (1967) reported that

the Dundee is about 100 feet thick.

Hydrology

MS (1992) investigations at the Envotech proposed facility site reveal that ground-
water-ﬂow directions are towards the southeast and northeast. Wetlands lie on top of the unit 2
clay and within the unit 1 sand. Their hydrology is probably inﬂuenced most by recharge from
precipitation, ground-water flow in the sand, and ground water discharge to adjacent surface water
bodies (RMT, 1992). Water from the unit 1 sand is occasionally used for residential water supply.
Other limited sources of water to wells include the discontinuous sand or sand and gravel lenses

within the unit 2 and unit 4 tills, however, these are not regionally continuous (RMT, 1992).

1 1
Kunkle (1961, as referenced in RMT, 1992) found that regionally the water table dips

slightly towards the southeast to Lake Erie. The creeks and drains within the Milan study site
control the shape of the water table surface and serve as sites for discharge of water from the
shallow portions of the ﬂow system (RMT, 1992). During both high and low water conditions
measured by RMT (1992), the creeks and drains formed an east-west trending ridge on the water
table surface with the highest elevations beneath the waste reception area and the northwest comer
of the landfill area. Generally, ground water ﬂows northeast to Sugar Creek and Branch No. 2 of
Sugar Creek and southeast to the York and Augusta Drain. RMT (1992) observed that the York
and Augusta Drains and Sugar Creek form hydrogeologic divides to horizontal flow in the unit 1
sand.

Ground water ﬂow is primarily horizontal through unit 1, the surﬁcial lacustrine deposits,
and is generally towards the surface drains, subsurface drains, and streams (RMT, 1992). Flow is
primarily vertical in units 2, the upper glacial till, 3, lacustrine silty clay, and 4, the lower glacial
till. RMT (1992) observed an exception to the downward ﬂow conditions that occurred near the
former quarry operation where ground water flow discharged to the surface or to the quarry lake,
and beneath the ARL site and part of the proposed facility site where ground water ﬂow was
upward and controlled by the hydraulic head in the limestone. RMT (1992) also found that the
lacustrine and outwash lenses within the tills "have little effect on the overall pattern of downward
ﬂow across the site. The exceptions to this are the thick (10 to 20 feet) sandy deposits at the
surface of the bedrock where strong horizontal ﬂow components are found".

Within the Traverse Shale in the western part of the area, ground water ﬂow is vertical.
Ground water flow is towards the south in the Dundee Limestone (RMT, 1992). Water is
available to wells in the Dundee Limestone when a signiﬁcant number of joints and fractures are
present (Twenter and others, 1976, as referenced in RMT, 1992). Solution-enlarged fractures

produce the permeable zones in part of the bedrock below the site (Snow, 1967, as referenced in

l2
RMT, 1992). The piezometric surface in the Dundee Limestone shows an increase in head away

from the center of the ERC towards the north, west, and east. McNamee, Porter, and Seeley (1965)
and Twenter and others (1976), as referenced in RMT (1992) reported some ﬂowing wells, which
suggest the presence of localized upward gradients, within three miles of the ERC. Review of 682
well logs collected by RMT reveal that 7 wells, of which 4 are completed in bedrock, had static
water levels at or above the ground surface. These ﬂowing wells were located north and west of the
site. Of the bedrock wells, the closest is approximately 2/3 mile west of the ARL site and the
farthest is 1 mile north of the ARL site.

RMT (1992) measured hydraulic conductivities in each of the units at the site (Table 1).
Laboratory values were determined using a permeameter on undisturbed samples. Field values
were determined using bail-down or slug tests in which the recovery rate of the well was measured
after a predetermined amount of water was removed from the well. RMT interpreted the recovery
rate data using analytical solutions developed by Hvorslev (1951) or Bouwer and Rice (1976).

Table 1. Hydraulic Conductivity Values.
( (t) denotes field value and n denotes number of measurements )

 

 

MEASURED VALUES FOR ESTIMATED VALUES FOR
UNIT HYDRAULIC CONDUCTIVITY, HYDRAULIC CONDUCTIVITY,
feet/day (RMT, 1992) feet/day (this study)
1 SE 3.4 9-03 to 1.53 [(f), n=13] 6.07 to 22.6
NW 0.964 to 7.65 [(0, n=5]
2 8.5 9-05 to 4.82 9-03 [(0, n=24] 4.0 9-04 to 1.5 9-04
4 2.55 9—03 [(0, n=1] 4.3 9-07 to 9.8 9-04

bedrock 3.12 9-05 to 1.93 [(1). 0:6] 0.134 to 13.39

 

13
For unit 1, hydraulic conductivity values determined from ﬁeld tests ranged from 3.4 X 10'

3 to 1.53 ft/d in the southeast (n=13) and from 0.964 to 7.65 ft/d in the northwest (n=5). The higher
values were in the area of the proposed waste reception area and reﬂect the coarser grained
deposits in this area. The geometric means of these measurements were 7.37 X 10'2 in the southeast
and 3.4 in the northwest.

For unit 2, hydraulic conductivity values determined from ﬁeld tests ranged from 8.5 X 10'
5 to 4.82 x 10'3 ft/d. The geometric mean using these 24 measurements was 2.58 x 10“. For both
ﬁeld and laboratory determined values RMT calculated 95 % conﬁdence intervals. For unit 2, the
upper limit of the 95 % conﬁdence interval calculated from ﬁeld values, which RMT interpreted to
be an effective hydraulic conductivity for the unit, was 2.83 X 10“ ftld. The lower limit of the 95
% conﬁdence interval calculated from laboratory values, which RMT interpreted to be the matrix
hydraulic conductivity, was 4.25 X 10” ftld. For unit 3, the geometric mean using 194 laboratory
measurements was 4.34 X 10"5 ftld.

For unit 4, a hydraulic conductivity value of 2.55 X 10‘3 RM was determined from one
ﬁeld test. RMT interpreted the matrix hydraulic conductivity to be 2.83 X 10'5 ft/d which is within
the 95 % conﬁdence interval calculated from 149 laboratory measurements. Finally, for the
limestone RMT determined hydraulic conductivity values from ﬁeld tests that ranged from 3.12 X
10" to 1.93 ftld. The geometric mean using these 6 measurements was 8.22 X 10'3 ftld. Snow
(1967) also determined a mean hydraulic conductivity value using airlift data of 5.67 ftld. RMT
estimated the hydraulic conductivity of the Traverse Shale to range from 2.83 x 10" to 2.83 x 10‘8

ftld.

l4

RMT (1992) observed fractures in the unit 2 till (Figure 3). After excavating an area for
cover for the Arkona Road Landﬁll, an approximately 2.5 acre sloping surface of unit 2 was
exposed to a depth of 35 feet below the unit 1 sand. Six test pits were dug with two each at 10, 20,
and 35 foot depths. At the two deeper depths the 2 to 4 foot deep test pits were oriented
approximately north-south and east-west, while the shallow pair were both oriented north-south.
Smeared clay surfaces were removed and the orientations of fractures measured.

RMT observed a frequency of 6 to 7 fractures per foot in the top 10 feet of unit 2 with a
spacing of about 0.15 foot in horizontal exposures. The fractures in vertical exposures appeared to
be spaced at about 0.67 foot but these were less visible and may have been underestimated. These
fractures showed a predominant strike of 80° and a possible second set striking 150° with both sets
generally dipping within 10° of vertical (n=66). Observed products of geochemical changes on
fracture surfaces included a dark mineral (possibly manganese oxide), iron oxide, gypsum, and
calcite precipitates. The unfractured matrix was slightly oxidized near fracture surfaces (RMT,
1992).

At a depth of 20 feet below the unit 1 sand, RMT (1992) observed a frequency of 3 to 4
fractures per foot with a spacing of about 0.30 feet in horizontal exposures. The frequency of
fractures in vertical exposures was 2 to 3 per foot with a spacing of about 0.40 foot. Three
fracture sets were observed with strikes of 90°, 130°, and 20° with the majority of fractures near
vertical (n=48). Observed products of geochemical changes on fracture surfaces included iron
oxide, gypsum, and calcite precipitates. Some oxidation of the unfractured clay matrix was

observed near fracture surfaces (RMT, 1992).

15

.32: .926 a-.. as»... 92. 8989 25E

8%... .8393 no sous—auto n A

98 85 83898 .89?
5329305532"— 2323 u >...

38." 69 83896 Beanie:
no 3288 moans—ooh 2:89.: n ma.

239283 .«e 38:: u a

 

‘ ”928m:

as .t. 5
S u an a

Elm—DgnmrNF—ZD

9N u >..~
2-3 2.-..
”air a

£8: :5. 3 cases: 289..on 238.... .m came.

.5528 24240: t v .EZD >450 N :75
»
3* n 5 n 30 u 5

n4 n>m
Gena...—

84 u 5
2. u an

 

   

EQEcNtNP—ZD EDEc—rag

16
At the 35 foot depth the frequency and orientations of fractures was nearly identical to

those at the 20 foot depth (n=122). Fracture surfaces rarely showed secondary mineral deposition,
although the unfractured clay matrix was oxidized near fracture surfaces (RMT, 1992).

RMT made additional observations of the fracturing in the unit 2 clay in a gully eroded
into the west wall of the quarry where approximately 20 to 25 feet of unit 2 was exposed. Fracture
set orientations were observed striking 60°, 110°, and 15° (n=78). Thus these appear to be rotated
somewhat from those in the test pits. Due to weathering, any precipitation on fracture surfaces or
oxidation of the matrix could not be determined. Between 10 to 20 feet deep, fractures were
spaced about 0.5 to 1.0 foot. Below 10 feet fractures were observed ranging from several feet to
greater than 5 feet in length.

The unit 4 till is not visible in surface exposures at the ERC site. RMT (1992) observed
exposures of this unit at a quarry approximately 10 miles to the south operated by Holman, Inc.
near Dundee, Michigan. Near-vertical fractures were observed to have a frequency of 0.3 to 1.0
per foot with a spacing of about 1 to 3 feet. These fractures were observed to be up to 5 to 6 feet
long. Fracture set orientations were observed striking 145°, 60°, and 5° (n=40). This unit was also
weathered so any geochemical changes on fracture surfaces could not be determined.

Errors observed during measurement of fracture frequency and orientation data from other
nearby quarries was reported by Chaivre (1972). Chaivre investigated measurement of fractures in
Upper Silurian and Devonian rocks by different operators and methods. His results indicated that
the size of the sample or number of fractures measured was important. In Chaivre's observations
large fractures trended the same as smaller ones thus the more obvious ones could be measured
exclusively as long as a sufﬁcient number was available to measure. Also, he observed that some
error was introduced due to Operator variability in measurements and in selecting what to measure.

Chaivre accounted for these sources of error and presented results of fracture set orientations for

l7
jointing in the Dundee Limestone from three different quarries. Fractures generally strike 121 -

180° and 46 - 75° with near vertical dips.

Since Chaivre’s results indicated that operator variability introduced error into the fracture
measurements and the number and experience level of RMT personnel who measured the fractures
is unknown, additional measurements were made during this study. Along the sloping exposed
surface of unit 2, fracture measurements were made at 10, 20 and 35 foot depths (Figure 4).
Along the vertical walls of eroded channels in the surface at each depth, a shovel was used to
excavate the upper weathered material exposing unweathered portions of unit 2. In these
unweathered sections visible fractures were measured. The criteria for selecting which fractures to
measure was based on the partings along planar surfaces in the clay and on visible staining and
oxidation along fracture surfaces.

Fractures were observed at varying orientations at the 10 foot depth with most having
strikes ranging from 20 - 140° with a major group at 85° (n=23). These fractures were near
vertical and most showed visible oxidation of the clay matrix along fracture surfaces. At the 20
foot depth two preferred orientations were evident (n=32). One fracture set ranged from 60 - 120°
and the other near 10° with all fractures having near vertical dips. Most of these fractures showed
visible oxidation of the clay matrix near fracture surfaces. Finally at the 35 foot depth most
fractures were observed to strike around 35° , 90° or 125° (n=30). Measurements at this depth
were made from two separate excavations, one oriented north-south and the other oriented east-
west. Fractures at this depth ranged in dip from 57° to 90°.

In an eroded gully located to the west of the slope additional fractures were measured
(Figure 5). Strikes of the near vertical fractures were measured at four depth intervals which
included 0 - 10, 15 - 20, 20 - 25, and 25 - 30 feet. At the 0 - 10 foot depth fracture strikes are

oriented almost randomly with small groups at 18 to 32° and 92 to 112° (n=15). At the 15 - 20

10 foot depth (n=23)
[N-S]

20 foot depth (n=32)
[N-S]

 

   

35 foot depth (n=30)
[N‘S, E'W]
LEGEND:
n = number of measurements
[ ] = excavation orientation
Rose diagrams plotted using the
equal area calculation method

with a 10° class interval. Inner
circles indicate percentages.

Figure 4. Fracture Orientations on the Sloping Surface of the Unit 2 Till.

19

15-20 foot depth (n=6)

       
   

 
   
  
   
    
    
  
   
      
   
  

 
 
   

 
 
 
 
 
   
     

II
l\~. 1' e‘ﬂﬁ'lh’e
it”? I/ §‘\‘ - 'I~o

II
\

  

agar"

 
  

 

with a 10° class interval. Inner
circles indicate percentages.

Figure 5. Fracture Orientations in the Gully of the Unit 2 Till.

20
foot depth most fractures strike at approximately 5° or 100° (n=6). At the 20 - 25 foot depth most

fractures strike at approximately 32° with a grouping ranging from 2 to 70° (n=27). At the 25 - 30
foot depth most fractures strike at approximately 10° with a grouping ranging from 350 to 32°
(n=17). Within the gully the spacing and extent of the fractures could be determined. A larger set
of fractures was spaced 1.5 to 2 feet apart and extended vertically up to 15 feet in length. A second
more closely spaced set was observed 2 to 6 inches apart and extended vertically from 1 to 5 feet in
length.

Comparison of fracture measurements made during this study with those made by RMT
(1992) reveal that the orientations generally match. Along the sloping surface of unit 2 at the 10
foot depth both measurements indicate a major group near 85° and a grouping ranging from 20 to
140°. The wide range of these measurements is probably due to desiccation and weathering of the
upper portions of the unit. At the 20 foot depth, both measurements indicate a major group near
90° and a minor group near 10°. However, measurements made during this study do not show a
second minor group at 130°. At the 35 foot depth, both measurements show groups near 90°, 30°,
and 130°. Possibly more directions are evident in measurements made during this study since both
north-south and east-west orientations were measured.

In the gully, RMT apparently combined measurements made from different depths, thus it
is difﬁcult to directly compare those measurements with ones made during this study. However, the
major group RMT observed at 60° falls within the range observed at to 20-25 foot depth. The
group RMT observed at 110° falls close to the measurements obtained at the 0—10 and 15-20 foot
depths. Finally, the group observed at 15° closely matches the average of the measurements
obtained at the 25-30 foot depth. The fracture spacings observed during this study generally agree

with those made in the gully by RMT.

21
Overall the measurements made along the sloping surface of unit 2 indicate a general east-

west trend in fracture orientations. The measurements made in the adjacent gully do not show this
same trend. However, the orientation of the exposed surfaces in the gully may have contributed to
the absence of an east-west group of fracture orientations. For example, at the 25-30 foot depth
the gully wall trended 300°, thus, an east-west trending set of fractures would have likely been

missed.

SIMULATION OF GROUND-WATER FLOW

Conceptual Model

The conceptual model is the ﬁrst step in model design and includes (1) deﬁnition of units,
(2) water budget, and (3) deﬁnition of ﬂow system. The purpose of a conceptual model is to
organize the available data in a way that represents the ﬁeld situation as closely and as simply as
possible, since the actual ﬁeld situation can not be duplicated exactly. The model boundaries
include the ARL site and most of the proposed hazardous waste management facility area. Since no
natural hydrologic boundaries exist around this area the boundaries must be extended enough away
from the main area of interest so as not to affect simulation results in the central model area.

The layers of the model consist of the units described by RMT (1992). The upper layer
will consist of the unit 1 sand. Water can enter or exit this layer from the boundaries, creeks,
drains, and the underlying layer. Water can also enter in the form of precipitation. The second layer
consists of the unit 2 till and the unit 3 lacustrine clay. These two units were combined due to their
similar hydraulic conductivities. Water can enter or exit this layer from the overlying and

underlying layers. The third layer consists of the unit 4 till. Water can enter or exit this layer from

22
the overlying and underlying layers. The fourth layer consists of the bedrock limestone unit. Water

can enter or exit this layer from the boundaries and the overlying layer. Water can also enter this
layer in the form of precipitation at the quarry location.

Ground-water ﬂow in the unit 1 sand is primarily towards Sugar Creek and the York,
Augusta, and Murray Drains. Ground-water ﬂow is primarily vertical in the clay units 2, 3, and 4.
Ground-water ﬂow is primarily towards the south in the Dundee Limestone.

Ground-water ﬂow is presumably faster within the fractures than within the intervening
matrix material as indicated by the presence of tritium at depth. To use a standard ﬁnite difference
code such as MODFLOW (McDonald and Harbaugh, 1988), a fractured material must be
represented as an equivalent porous media in which primary and secondary hydraulic
conductivities are replaced with an effective hydraulic conductivity. With an appropriate
representative elementary volume of material, the values for effective parameters can be
determined using aquifer tests, water balances or inverse models, or ﬁeld descriptions of fracture
apertures and interconnections (Anderson and Woessner, 1992). However, utilization of these
techniques at the Milan study site was not possible during this study.

At this site, fracture frequency and orientations have been measured, however, Long and
Witherspoon (1985) found in their studies that this information alone was inadequate to determine
permeability or whether the system behaved as a porous media. Field and laboratory measurements
of hydraulic conductivity were conducted at this site in each unit by RMT (1992) and can be used
to estimate effective hydraulic conductivities. Knowledge of these values and the selection of a cell
size much larger than the fracture spacing will be assumed to be adequate to represent this system
as an equivalent porous media. The accuracy of this assumption will be indicated by how well

simulated heads match measured heads.

23
Numerical Model

Model Grid and Layers

The grid spacing used in the model must be large enough to prevent excessive computer
storage requirements and computation time, yet small enough to represent the system accurately
(Anderson and Woessner, 1992). Within most of the study area, the large number of soil borings
and monitoring wells located approximately 200-300 feet apart provides good information on
lithologic characteristics and water levels. However, these soil borings and monitoring wells are
concentrated within the ARL and proposed waste reception and landﬁll areas. Outside of these
areas lithologic and hydrologic information is limited with available borings 800 or more feet
apart. Based on the high and low data densities, a grid spacing of 150 feet was chosen to represent
ﬂow conditions within this area.

The axes in the model need to be aligned with the principal directions of hydraulic
conductivity (K, , K,, and K1). If the study site shows some trend other than east-west and north-
south, the axes of the model need to be rotated. Since fracture measurements made during this
study and by RMT indicate a general east-west trend, the model grid does not need to be rotated.
Therefore the model grid consists of 33 rows and 50 columns and covers 1.33 square miles
approximately 2 miles east of the city of Milan. Vertically the model consists of four layers with
layer 1 representing the upper unit 1 sand, layer 2 representing the upper unit 2 till and the unit 3
lacustrine clay, layer 3 representing the lower unit 4 till, and layer 4 representing the bedrock
limestone unit. The bedrock shale unit is implicitly included in the leakance term between layers 3

and 4.

Boundary Conditions

Boundaries of ground-water-ﬂow systems can be physical or hydraulic. Physical
boundaries are formed by an impermeable body of rock or a body of surface water. Hydraulic
boundaries include ground-water ﬂow divides and streamlines. These boundaries are represented in
a mathematical model by the following types of conditions: 1) speciﬁed head boundaries where the
head in the cell is given, 2) speciﬁed ﬂow boundaries where the ﬂux across the cell boundary is
given, and 3) head-dependent ﬂow boundaries where ﬂux across a cell boundary is determined
using a speciﬁed head in the cell (Anderson and Woessner, 1992). External boundaries for the
model were chosen based on the conceptual model outlined above. External boundaries for the
model area were chosen to be distant from the main area of interest to remove effects of the
speciﬁed boundary conditions. The boundaries of the model extend approximately 4,800 feet east
of ARL, 600 feet west of ARL, 1,050 feet north of ARL, and 1,950 feet south of ARL.

Layer 1 boundaries were speciﬁed head on all sides except where Sugar Creek intercepted
the north and east boundaries and along the south boundary at the quarry location. The river cells
were assigned head-dependent ﬂow boundaries. The cells at the quarry were assigned speciﬁed
ﬂow boundaries with ﬂow equal to zero, therefore these were no-ﬂow boundaries. The quarry
excavation extends through units 1, 2, and 3, thus no ﬂow is possible across this excavation. The
upper boundary of layer 1 was assigned as speciﬁed ﬂow equal to the average recharge rate for the
area. Layer 2 and 3 boundaries were speciﬁed ﬂow on all sides with ﬂow equal to zero. Since ﬂow
in these units is primarily vertical it is assumed that no ﬂow is possible in the horizontal direction
across the external boundary. Layer 4 boundaries were speciﬁed head boundaries on all sides. The

lower boundary of layer 4 was assumed to be a no-ﬂow boundary.

Sourcesand Sinks

Sources and sinks are another avenue for water movement into and out of the model area.
Sources and sinks are represented by the same types of boundary conditions described above and
can be considered as internal boundary conditions. Speciﬁed head, speciﬁed ﬂow, with ﬂow equal
to zero, and head-dependent ﬂow boundaries were used to deﬁne interior sources and sinks in the
model area. For layer 1 no-ﬂow cells were assigned to the quarry location. Head-dependent ﬂow
cells were assigned to the locations representing Sugar Creek, and the York, Augusta, and Murray
Drains. For layer 2 and 3 no-ﬂow cells were assigned to the quarry location. For layer 4 speciﬁed

head cells were assigned to the quarry location.

Parameter Value

Input parameters required for the model include 1) the geometry of model layers, 2)
ground-water recharge rates, and 3) the aquifer hydraulic properties. Each of these will be

described in the following paragraphs.

M21213
Information on unit thicknesses was taken from the well and boring logs prepared during
earlier site investigations (RMT, 1992; Neyer, Tiseo and Hindo, 1987; and McNamee, Porter and
Seeley, 1965). The top of layer 1 was taken to be the water table. This surface was deﬁned using
data from water level measurements in 41 wells, as well as, 23 river and 59 drain elevations. Well
data was averaged from seven measurements taken between 12/10/87 and 6/1/88. This range was
chosen since it bracketed March 16, 1988 when a large number of monitoring wells at the site were

measured. An average of head measurements over a range of time more accurately represents

26
steady-state conditions than a single point in time. The measurements for any well installed

immediately prior to or during this range were inspected for the point at which equilibrium was
reached. Only those measurements for the well at equilibrium were included in the initial estimate
of the water table.

The bottom of layer 1 was deﬁned using 454 measurements from well and boring logs.
The bottom of layer 2 was deﬁned using 296 measurements from well and boring logs. The bottom
of layer 3 was deﬁned using 140 measurements, of which 24 were from well and boring logs and
116 were taken from Figure 4-13 (RMT, 1992). The bottom of layer 4 was taken to be 100 feet

beneath the bottom of layer 3 based on Snow's (1967) observations.

Ground-Wye; Rﬂggge Rag
RMT (1992) estimated a recharge value of 4.5 inches per year based on mass balance

calculations for the site and on literature values for similar soils in similar climatic regimes. This
constant value was used for layer 1. This value is very close to the range of 5.0 to 5.2 inches per
year predicted for this area using the analysis relating base ﬂow characteristics of streams to land

use and basin characteristics in the Lower Peninsula of Michigan determined by Holtschlag (1994).

Hydraulic Promggg
The hydraulic properties, including hydraulic conductivity, transmissivity, leakance, and
streambed and drain conductances, control simulated ground-water ﬂow through and between
model layers. These properties were calculated as the product of a matrix containing initial
estimates and a constant determined during model calibration. The methods and information used
to determine the initial estimates are described in the following paragraphs.
The hydraulic conductivity of layer 1 and the transmissivities of layers 2 and 3 were

determined by assigning hydraulic conductivities to each lithology described in the well and boring

27
logs for each layer (Table 2). Using the procedures described by McDonald and Harbaugh (1988,

p. 5-2), effective hydraulic conductivities were computed for each location. These values were then
contoured over the model area. For layers 2 and 3, these values were multiplied by the layer
thickness to determine transmissivity. Initial estimates of horizontal hydraulic conductivity for
layer 1 range from 9.5 to 35.3 ftld. The initial estimate for transmissivity of layer 2 ranges from
2.8 X 10 ‘4 to 2.03 X 10 '2 ftzld. The initial estimate for transmissivity of layer 3 ranges from 1.905

x 10" to 3.276 x 10 '2 ft2/d.

Table 2. Hydraulic Conductivities Assigned to Strata Codes.

 

 

03:333va STRATA LITHOLOGIC
CODE DESCRIPTION
(feet per day)
0.00028 13 sandy clay / sandy, silty clay
16 clay and sand / silty, clay
19 till
40 top soil
0.0028 15 silt / sandy silt
0.28 28 sand with clay
28 20 sand / sandy till
21 fine sand / silty sand
280 25 sand and gravel

 

The initial estimate for transmissivity of layer 4 was taken from Figure 4-14 (RMT,
1992). This ﬁgure shows bedrock hydraulic conductivity derived from Snow (1967). These values
were multiplied by the assumed 100 foot thickness of the layer. The initial estimate for

transmissivity of layer 4 consisted of zones equaling 1.42, 14.2 and 142 ft2/d.

28
Leakances were determined using a similar method. Hydraulic conductivities were

assigned to each lithology described in the well and boring logs for each layer. An effective vertical
hydraulic conductivity was computed for each location using the equation described by Freeze and
Cherry (1979, p. 34).

K: = d / ((di/ Ki) + (din/Km»
where Kz = effective vertical hydraulic conductivity,

d = thickness of layered system,

(1, = thickness of layers above and below contact, and

K; = hydraulic conductivity of layers above and below contact.

The leakance equals the vertical hydraulic conductivity divided by the thickness of the layered
system (d) and has units of UT. Thus, for example for layer 1, the leakance equaled:

1/ ((dl/Kl) 4' (dz/1(2))
The leakance for layer 3 included vertical hydraulic conductivity and thickness values for the
Traverse Shale. There was none for layer 4 since this was the bottom layer of the model. The initial
estimate of leakance for layer 1 ranged from 4.14 x 10 *5 to 2.8 x 10 4 Id. The initial estimate of
leakance for layer 2 ranged from 5.17 X 10 ‘9 to 5.27 X 10 ° Id. The initial estimate of leakance for
layer 3 ranged from 5.16 x 10 '9 to 2.91 x 10 '5 Id.

In the model, streambed conductance is the product of the hydraulic conductivity of the
streambed materials, stream length, and stream width, divided by the streambed thickness
(McDonald and Harbaugh, 1988, p. 6-5). These values were assumed to be constant within the
model area. Stream width, as measured by RMT (1992) in December 1987, was approximately
10.5 ft. Stream length is equal to the length of the cell (150 ft) and streambed thickness is
arbitrarily assumed to be 1 ft. Hydraulic conductivity of the streambed materials was assumed to
be 0.5 ftld. Drain conductances are calculated in the same way as for streams. The drain width,

as measured by RMT (1992) in December 1987, was approximately 7.8 ft. Drain length is equal to

29
the length of the cell (150 ft) and drain thickness is arbitrarily assumed to be 1 ft. Hydraulic

conductivity of the drain was assumed to be 0.2 ftld. Streambed conductances equaled 800 ftzlday

and drain conductances equaled 200 ft2/day.

Model Calibration

Model calibration requires adjustments to input parameters to reduce the difference
between measured and simulated hydraulic heads. The nonlinear regression method,
MODFLOWP, developed by Cooley and Naff (1990) and modiﬁed by Hill (1992) was used with
MODFLOW to calibrate the model. In MODFLOWP, optimal parameter values are calculated
that minimize the sum of squared errors. Scaled sensitivities are also computed by MODFLOWP
and give an indication of how much a change in the parameter value will affect the simulated
heads. A low sensitivity indicates little effect on the simulated heads and that the parameter can not
be accurately estimated by regression.

Parameter estimation was used to determine the optimal estimates for the multiplicative
constants for matrices containing hydraulic conductivity of layer 1, transmissivity of layer 4, and
leakances for layers 1, 2, and 3. The transmissivities of layers 2 and 3, as well as the streambed
and drain conductances had very low sensitivities and could not be estimated by MODFLOWP.
Initially the multiplicative constants were assigned a value of one. '

Weights were assigned to the head measurements for each layer. In general more weight
was assigned to measurements whose measurement variances were low. A standard deviation,
which when squared equals the variance, was calculated for each well from the measurements
collected from 12l10/87 to 6/1/88. A total of 41 wells were included for layer 1, 26 for layer 2, 6
for layer 3, and 14 for layer 4. An additional 8 measurements for cells at the quarry location for

layer 4 were included and assigned a standard deviation of 0.22 which was the lowest standard

30

deviation calculated for the wells in each unit. As an example, water levels for well 152 B
measured on 10-Dec-87, 5-Jan-88, 22—Feb-88, 16-Mar-88, 4-Apr-88, 11-May-88, and 1-Jun-88
were 672.72, 673.33, 673.62, 673.52, 673.64, 672.31, and 671.67 feet respectively (RMT, 1992).
The calculated standard deviation for these measurements is 0.70. The weight, which is the inverse
of this value squared, was 2.04 for well 152 B in layer 2.

Optimal parameter estimates are given in Table 3. MODFLOWP indicated that all of the
leakances were correlated. Correlation between parameters complicates the estimation and
interpretation of parameter values. Coordinated changes in parameter values would produce
identical results in terms of the calibration criteria (Hill, 1992). Thus the leakances can not be
uniquely estimated.

The estimated values for the hydraulic conductivity of layer 1, the transmissivity of layer
4, and the leakances for layers 1, 2, and 3 were computed as the matrix of initial estimates times
the optimal parameter value determined by nonlinear regression. The resulting range in hydraulic
conductivity for layer 1 is 6.07 to 22.6 ft/d, in transmissivity for layer 4 is 13.4 to 1339 ftzld, in
leakance for layer 1 is 1.1 X 10 '5 to 8.02 X 10 4 Id, in leakance for layer 2 is 1.08 X 10 "° to 1.1

x 10 '7 Id, and in leakance for layer 3 is 1.53 x 10 "0 to 8.6 x 10 '7 /d (Table 1).

31
Table 3. Optimal Parameter Estimates.

 

 

 

PARAMETER INITIAL OPTIMAL
ESTIMATE ESTIMATE

Hydraulic Conductivity of Layer One 1.0 0.639
Transmissivity of Lay9r Four 1.0 9.43

Leakance for Lay9r One 1.0 0.0395
Leakance for Lay9r Two 1.0 0.0209
Leakance tor Lay9r Three 1.0 0.0296

Simulation Results

Generally, simulated heads were consistent with measured heads. Near the model

boundaries where no water level data was available, initial heads were estimated. In layer 1 the

greatest differences between simulated and estimated heads (-27 to 10 ft) are near the model

boundaries and mostly to the west. Within most of the central model area simulated heads are

within 4 feet of measured values (Figure 6). In layer 2 the greatest differences between simulated

and estimated heads (~20 to 29 ft) are near the model boundaries. Within most of the central model

area simulated heads are within 5 feet of measured values (Figure 7). In layer 3 the greatest

differences between simulated and measured heads (-28 to 10 ft) are near wells 204, 501, and 502.

Within most of the central model area simulated heads are within 8 feet of measured values (Figure

8). In layer 4 the greatest differences between simulated and estimated heads (-66 to 51 ft) are near

32

 

 

 

mo
° 9
09
e
e
e e .
e
99 .
o
0
975+»
e
a . 0
3 .
e
3 e
1 9° 9
:1 9
.§ 0 °
a:
970i» °
9
. o
o
00
e
e
o
as. f
as 970 975
MeasuredI-Ieads

Figure 6. Relation Between Measured and Simulated Heads for Layer One.

 

33

 

Simulated Heads

 

 

 

 

Measured Heads

Figtne7.Re1ationBetweenMeasuredandSimulatedl-IeadsforLayerTwo.

34

 

Simulated Heads
9

 

 

 

0
600
O O
O
can «t
920 : ¢ : : 4. 2 : I :
625 can 65 60 66 60 $5 990 5 670
Measured Heads

Figure 8. Relation Between Measuredand Simulated Heads for Layer Three.

35
the model boundaries and mostly to the east. Within most of the central model area simulated heads

are within 10 feet of measured values (Figure 9).

Generally, estimated ranges for the hydraulic parameters are consistent with measured
ﬁeld values (Table 1). Estimated values for the layer 1 sand were higher than the ﬁeld measured
values, however, when compared to literature values (Freeze and Cherry, 1979) the estimated
values seem reasonable for a unit that is primarily sand. Estimated values for layer 2 are within
the range of ﬁeld measured values. Estimated values for layer 3 are lower than the ﬁeld measured
value. However, only one ﬁeld measurement was collected and it is not known whether this value is
representative of the unit or not. Comparison of these estimated values with literature values
(Freeze and Cherry, 1979) show that they seem reasonable for a unit that is primarily glacial till.
Finally, estimated values for layer 4 are higher than the ﬁeld measured values. However, Snow
(1967) estimated an average hydraulic conductivity value of 5.67 ft/d for the limestone which is in

closer agreement with the model estimated values.

Particle Tracking

Particle tracking provides a way to investigate the ground-water ﬂow directions. Cell-by-
cell ﬂows are output from MODFLOW and used as input for the particle tracking program,
MODPATH, documented by Pollock (1989). MODPATH was used to track the particles as they
moved from cell to cell throughout the steady-state three-dimensional ﬂow system. In a particle
tracking simulation, hypothetical particles are placed on the faces of a cell or within a cell and
tracked for a speciﬁed amount of time or until they leave the ground-water ﬂow system. The path
of the particle as well as the stopping location can be determined as it moves through the system.

Particles can leave the ground-water ﬂow system at the boundaries or at sinks such as rivers.

36

 

 

 

 

3 case
3 .
E e
O
E
a 820V
0 O
6150 O
610 t : :
610 615 62)
MeeeuredHeads

Figure 9. Relation Between Measured and Simulated Heads for Layer Four.

37
Particle tracking was used to investigate the presence of tritium at depth and the proposed upward

ﬂow directions at the ARL site.

Tritium Data

Large amounts of tritium, a radioactive isotope of hydrogen, were ﬁrst released into the
atmosphere around 1952 during a period of atomic testing. Before that date, Payne (1972)
estimated that natural tritium concentrations in precipitation ranged from 5 to 20 tritium units.
Using the half-er of tritium of 12.3 years, ground water that is older than 45 years old would now
have a tritium concentration less than 2.0 tritium units (TU). Ground water (tritiated water) that
has been recharged since that date would be expected to have tritium concentrations greater than
2.0 TU.

At the study site in southeastern Michigan, Foote (1989) collected samples from 17
different elevations in a soil boring and from 15 nested wells. In the extracted water samples from
the soil boring tritiated water was found at depths exceeding 22 m, while non-detectable values
were found as shallow as 8 m depth. In the ground-water samples from the nested wells, tritiated
water was found at depths of at least 25 m in the glacial sediments, while non-detectable values
were found as shallow as 14 m depth (Foote, 1989).

Particle tracking was used to determine whether the model could account for fractures in
the clay units and explain the presence of tritium at depth. For this simulation four particles were
placed on the upper face of each active cell not including constant head cells in layer 1 and tracked
forward until they left the ground-water ﬂow system. Thus, pathlines and ending locations were
determined for 5,944 particles. Of these particles, 19 % were removed from the ground-water ﬂow

system in layer 4. The rest were removed at the boundaries, drains, and river of layer 1. That most

38
particles leave the ground-water ﬂow system in layer 1 is expected due to the evidence that ground-

water ﬂow in the upper sand unit is primarily towards the surface and subsurface drains and river.
For those particles that traveled through layers 2 and 3 before being removed from the
ground-water ﬂow system in layer 4, the travel times were determined. The minimum travel time to
layer 2 was 15.2 years, while the minimum travel time to layer 3 was 6000 years. Thus the model
does allow for recent water to move to the upper till unit and lacustrine clay unit. However, it does
not allow for recent water to move down to the lower till unit as was observed to be the case with
Foote's (1989) measured tritium concentrations. These particle-tracking results show that the

model does not adequately account for the local ﬂow systems created by fractures in the till units.
Proposed Upward Flow

To investigate the proposed upward ground-water ﬂow direction at the ARL site, three
particle tracking simulations were run. Twenty-ﬁve particles were placed on the upper face of each
cell not including constant head cells in layers 2, 3, and 4. In each simulation particles were tracked
forward to determine whether they moved upward or downward. In the ﬁrst simulation with
particles on the upper face of layer 2 which is also the lower face of layer 1, 2 % stop in layer 1 at
the boundary, drain, and river cells. The rest stop in layer 4. In the second simulation with particles
on the upper face of layer 3 all particles travel downward and stop in layer 4. In the third
simulation with particles on the upper face of layer 4, all particles stop in layer 4. Thus particle

tracking indicates that the ground water is ﬂowing downward within all of the model area.

39
Analysis of Fractures

The presence of fractures complicates the interpretation of ground-water ﬂow in this area.
Fractures allow ground water movement at faster rates than possible through the clay matrix. The
above ground-water ﬂow model treats the fractured clay units as equivalent porous media. At a
large enough scale the effect of ﬁactures is minimized and an effective hydraulic conductivity can
be determined for the fractured matrix. However, the above particle tracking results indicate that
the model does not adequately represent the fractured units since water can not get to the depth that
tritium has been detected in a short enough time. An additional series of runs were done to examine
the effect of increasing the hydraulic conductivity of the fractured units by allowing greater leakage
between them. After increasing the leakage, these new simulated heads were compared to the
simulated heads calculated by parameter estimation.

In seven groups of runs, the following leakages were modiﬁed to determine the effect: 1)
leakage between layer 1 and layer 2 (kvl), leakage between layer 2 and layer 3 (kv2), and leakage
between layer 3 and layer 4 (kv3); 2) kvl; 3) kv2 and kv3; 4) kv3; 5) kvl and kv3; 6) kv2; 7) kvl
and kv2. In each group the leakages were multiplied by 0.01, 0.1, 5.0, 10.0, and 50.0. The runs
using 0.01 and 0.1 were done in order to determine the effect of decreasing the leakage. The results
are shown in Table 4. The values were calculated by taking the simulated head value minus the
measured value for the well and squaring the result.

The last column shows the value using simulated heads determined by parameter
estimation. In each case except for group two, layer 1 heads were improved by increasing the
leakage. Layer 2 simulated heads were improved in group four. Layer 3 heads were improved in

groups one, two, and part of group six. Layer 4 heads were only improved when the leakage was

Table 4. Effects of Changing Leakances on Simulated Heads.

 

LEAKANCE

LEAKANCE

LEAKANCE

LEAKANCE

LEAKANCE

 

23mg NLJJBEER MULTIPUER MULTIPLIER MULTIPUER MULTIPLIER MULTI PLIER JIONDAELL
or 0.01 or 0.10 or 5.0 or= 10.0 or= 50.0
kv1.kv2,kv3 1 5.94 5.93 5.75 5.97 5.19 5.92
2 70199.93 39.79 41.52 42.55 45.54 39.39
3 2927.72 322.15 29725 295.94 243.47 307.49
4 22.91 22.93 25.59 29.91 95.95 23.21
kv1 1 5.94 5.93 5.92 5.92 5.92 5.92
2 190139 419.49 94.97 99.09 10925 39.39
3 933.94 524.93 279.95 27521 27225 307.49
4 22.92 22.99 23.29 23.29 23.27 23.21
kv2,kv3 1 5.94 5.93 5.79 5.79 5.72 5.92
2 104.10 93.39 195.19 399.34 1123.92 39.39
3 33921 290.14 419.79 505.54 994.34 307.49
4 22.79 22.93 24.55 25.97 29.99 23.21
kv3 1 5.93 5.93 5.91 5.91 5.91 5.92
2 100.31 74.04 34.41 34.20 34.11 39.39
3 979.03 499.93 997.34 794.53 979.39 307.49
4 22.91 22.92 23.32 23.34 23.35 23.21
kv1,kv3 1 5.94 5.93 5.91 5.91 5.91 5.92
2 93.99 90.91 79.99 92.49 105.30 39.39
3 312.53 239.54 990.30 79420 997.99 307.49
4 22.90 22.99 23.40 23.43 23.49 23.21
kv2 1 5.94 5.93 5.90 5.90 5.79 5.92
2 103.59 99.99 50.99 97.91 91.52 39.39
3 90125 790.59 230.39 279.79 354.97 307.49
4 22.90 22.94 23.72 23.99 24.05 23.21
kv1,kVZ 1 5.94 5.93 5.79 5.79 5.77 5.92
2 37.90 31.31 92.37 79.75 99.94 39.39
3 903.37 799.77 329.43 523.95 991.49 307.49
4 22.79 22.94 23.99 24.33 24.79 23.21

 

41
decreased. These results indicate that simply increasing the leakage to better represent the

increased ﬂow possible through the fractures does not improve the model.

MODEL LIMITATIONS

The ground-water ﬂow model simulates steady-state conditions within the model area. All
inputs are assumed to be constant throughout the simulation. Changes in recharge or other
parameters over time can not be represented by this model. The model combines primary and
secondary hydraulic conductivities by treating the fractured units as equivalent porous media. To
more accurately represent this area and the small-scale localized ﬂow systems a model that
includes fractures should be considered.

The model boundaries were simulated as constant head cells for layers 1 and 4 and
determined on the basis of water level measurements in nearby wells. The solution is constrained
by and dependent upon the values chosen for these boundaries. More detailed water level
measurements should be obtained to more accurately represent the boundaries. To investigate the
effect of these boundaries on the interior of the model these values should be changed and moved.
An even better technique would be to ﬁrst model a larger area using natural ground-water ﬂow
divides and physical boundaries. Then successively smaller areas could be modeled using output
from the larger model to determine the boundary conditions where natural and physical boundaries
do not exist. Additionally modeling a larger area would enable inclusion of pumping wells as
stresses on the ground-water system in the area and improve the calibration.

The model cells are assumed to represent homogeneous hydraulic properties. Local ﬂow
systems smaller than the cell dimensions can not be represented. The hydraulic properties of the

layers were determined using existing well and soil boring data. The estimates for the layers could

42
be further improved by collection of more lithologic data. Determination of effective hydraulic

conductivities for the fractured units could be improved by more aquifer tests and more detailed
fracture measurements.

The accuracy of the particle tracking results depend on the assumptions and accuracy of
the ground-water ﬂow model and the estimate of effective porosity of the ﬂow system. The particle
tracking assumes no dispersion or diffusion. Similarly the results of the analysis of decreasing and
increasing leakage to simulate the presence of fractures depend on the accuracy of the ground-

water ﬂow model.

SUMMARY

Fractures provide a means for ground water movement and contaminant migration at faster
rates than possible through the matrix alone. At the study site, Augusta Development Corporation
has proposed locating a hazardous waste management facility overlying a fractured clay till unit.
Investigation of the role of fractures on ground-water ﬂow is necessary to ensure the protection of
water quality in this area. A four layer steady-state ground-water ﬂow model was developed to
simulate the ground-water flow system in this area.

The glacial deposits underlying the pr0posed facility site can be divided into four units
(RMT, 1992). These units include: 1) surﬁcial lacustrine deposits, 2) upper glacial till unit, 3)
lacustrine silty clay deposits, and 4) lower glacial till unit. Lacustrine/outwash lenses are found
within the glacial till units. Surface exposures reveal that the glacial till units are weathered and
fractured. These glacial units overlie the Traverse Shale and the Dundee Formation. The contact
between the Dundee Formation and the overlying Traverse Shale crosses the study area from

southwest to northeast.

43
RMT's (1992) investigations reveal that ground-water ﬂow directions are towards the

southeast and northeast. Wetlands lie on top of the unit 2 clay and within the unit 1 sand. Their
hydrology is probably inﬂuenced most by recharge from precipitation, ground-water ﬂow in the
sand, and ground-water discharge to adjacent surface water bodies. Ground-water ﬂow is primarily
horizontal through unit 1 and is generally towards the surface drains, subsurface drains, and
streams. Flow is primarily vertical in units 2, 3, 4, and the Traverse Shale. Ground-water ﬂow is
towards the south in the Dundee Limestone.

The ground-water ﬂow model developed for this area consists of four layers with layer 1
representing the upper unit 1 sand, layer 2 representing the upper unit 2 till and the unit 3
lacustrine clay, layer 3 representing the lower unit 4 till, and layer 4 representing the bedrock
limestone unit. The bedrock shale unit is implicitly included in the leakance term between layers 3
and 4. The model area is a rectangular grid covering 1.33 square miles approximately 2 miles east
of the city of Milan in southeastern Michigan. This area is divided into a grid of equally spaced
cells in 33 rows and 50 columns with each cell 150 feet on a side.

The model was calibrated using nonlinear regression. The estimated values for the
hydraulic conductivity of layer 1, the transmissivity of layer 4, and the leakances for layers 1, 2,
and 3 were computed as the matrix of initial estimates times the optimal parameter. The resulting
range in hydraulic conductivity for layer 1 is 6.07 to 22.6 ft/d, in transmissivity for layer 4 is 13.4
to 1339 ftzld, in leakance for layer 1 is 1.1 x 10'5 to 8.02 x 10 4 Id, in leakance for layer 2 is 1.08
x 10 "° to 1.1 x 10 '7 Id, and in leakance for layer 3 is 1.53 x 10 "° to 8.6 x 10 '7 Id.

Generally, simulated heads were consistent with measured heads. Within most of the
central model area simulated heads are within 4 feet of measured values for layer 1, within 5 feet of
measured values for layer 2, within 8 feet of measured values for layer 3, and within 10 feet of

measured values for layer 4.

44
Particle tracking was used to investigate the observed distribution of tritium at depth.

Foote (1989) collected samples from within a soil boring and from nested wells and found tritiated
water at depth beneath non-tritiated water. The particle tracking results indicate that water can not
reach layer 3 where tritium was detected within a reasonable amount of time. Particle tracking was
also used to investigate the upward hydraulic gradient beneath the Arkona Road Landﬁll proposed
by RMT (1992). The results show that all ground water movement is downward within the model
area. Finally, leakances between layers were systematically varied to simulate the effect of
fractures and to observe effects on simulated heads. However the results indicate that simply

increasing the leakances between layers is not sufﬁcient to improve the simulated heads.

CONCLUSIONS

At the Milan study site, approximately 2.5 miles east of the city of Milan in southeastern
Michigan, 3 four layer steady-state ground-water-ﬂow model was developed to simulate ground
water ﬂow in an attempt to explain the presence of tritium at depth observed by Foote (1989) and
the upward hydraulic gradient beneath the Arkona Road Landﬁll proposed by RMT (1992).
Simulation results using particle tracking indicate that water recharging this area can not travel
from the land surface to the depth where tritium was detected within a reasonable amount of time.
Simulation results also indicate that all ground water movement is downward within the model
area.

The fact that this model can not reproduce the results observed by Foote (1989) is not
surprising. The model described in this study does not explicitly account for the presence of
fractures, rather, fractures are implicitly included in the effective hydraulic conductivities

determined for each layer. Furthermore movement of water through fractures would occur at a

45

much smaller scale than represented by the cell dimensions. Fractures would allow local movement
of water at faster rates than indicated by the effective hydraulic conductivity for the layer.
Therefore, the presence of fractures near the sampled wells would allow water containing tritium to
reach the observed depths.

The fact that this model does not show an upward hydraulic gradient beneath the Arkona
Road Landﬁll is also explained by the fact that the model can not represent small-scale ﬂow
systems. The wells beneath the Arkona Road Landﬁll showing the upward gradient are screened in
localized outwash units within the glacial till and likely do not represent gradients present within
the surrounding till. Additionally, water levels taken from well logs within a three mile radius of the
area show that in only 4 logs out of 682 are heads in the Dundee Limestone at or above the land
surface. These wells likely intersect a joint or fracture in the limestone. These isolated locations
where the gradient is upward would occur at a much smaller scale than represented by the cell

dimensions.

APPENDIX

.95 93 é one .892 .2 can:

duo 33.32 D
4.8235 \\

due 55.... m

.33 93.55528 I
2052559

 

47

65. 923 .9 25 .952 .: 2&2

38332.02 D
202.:

 

48

.30 332.02 B
202,258

8.5.93.9 one 982 .2 25mm

 

49

due 9a: 2.25.0200 I
2222553

59.292392 25 .892 .2 ecu...—

 

50

.8965 8255 9.3 .3 25mm

02Fw<m
000.NOF.N 000.FOF.N 000:00FN 000.00F.N 000.00F.N 9.00.501” <000.00F.N

 

 

 

\de.

_ x.) Heel _ . _

 

 

l

 

 

M0006 FN

1000KFN

000.0 FN

000.0 FN

000.0NN

SNII-LLHON

51

.030. 5.3..— .«o mob .2 250E

02.55
000.NOF.N 000.FOF.N 000.00F.N 000.00F.N 000.00F.N 000KOF.N 000.00F.N

 

 

 

 

 

000.0 FN

000K FN

000.0 FN

000.0 FN

000.0NN

IE)NIH.L!:ION

52

 

 

 

 

 

219000
218,000)-
217000
216000

220,000 —

SNIFLLHON

2,186,000 2,187,000 2,188,000 2,189,000 2,190,000 2,191,000 2,192,000

EASTING

Figure 16. Top of Layer Three.

 

 

 

 

220,000 -

219,000 -

 

SNIHLHON

2,186,000 2,187,000 2,188,000 2,189,000 2,190,000 2,191,000 2,192,000

EAST] NG

Four.

Figure 17. Top of Layer

54

6:0 5.34 5 £054 “8.036.580 003802 .00 850E

 

02.0.05
000.090 000.090 000.FOF.0 000.020 000.030 000.00F.0 000K000 000.00F.0
_ _ .va 00 r _ 4 0
I 3.... 8... as .. 8% a

000.0 F0

000K F0

SNIHLHON

000.0 F0

000.0 F0

000.000

 

 

 

 

55

.23. 6.03 5 £05..— 08036555 005232 .3 am

02_._.w<m
000.00F.0 000.00F.0 000.FOF.0 000.00F.0 000.00F.0 000.00F.0 000.\.0F.0 000.00F.0

 

flu

I

 

_

0.100

@M «a... . w my) .. M .
0 09.0 b?

 

 

n

1F

50

 

.

l

 

L

 

 

000.0 F0

000.0F0

000.50

000.0F0

000.0F0

000.000

SNIHLHON

56

695,—. 3.3.0 5 £26..— uBakFégEO 00580—2

62.0.05
000.00F.0 000.FOF.0 000.00F.0 000.00F.0 000.00F.0 000KOF.0 000.00F.0

 

 

 

l

l

 

 

.8 2:5

000.0 F0

000K F0

000.0 F0

9Nll-LLHON

000.0 F0

000.000

57

.500 8.34 5 £33 333.0580 00.5802 .00 PEME

022.05
000.00F.0 000.00F.0 000.FOF.0 000.00F.0 000.00F.0 000.00F.0 000KOF.0 000.00F.0

  

 

   

000.0 F0

qqqoqqqq
§OOOGO

0
2.
0

000K .0

000.0 F0

SNIHLHON

000.0 F0

 

1 83%

 

 

 

 

58

25 £3 Mo 8:38:28 33?: 3:88: 3.850 am 2:?

02:05
000.00F.0 000. FOF.0 000.00F.0 000.00F.0 000.00F.0 000.00F.0 000.00F.0

 

 

 

_ _
é
ﬁ Sosa

Soda
/\\lll Soda

1 000.000

SNII-LLHON

 

 

59

000.00F.0 000.FOF.0 000.00F.0 000.00F.0 000.00F.0 000.00F.0 000.00F.0

.030. 05 0:0 00.34 002300 8.5—«o..— Bmﬁumm .mN 250E

022.05

 

1

 

1

{A C _ _

d .

DCQ //.;..'.'.;-— _

_

M

1

l

l

 

 

000.0 F0

000.0 F0

000.0 F0

000.0 F0

000.000

SNIHLHON

60

.93. 5.93 0o £23555 395mm .E 2..me

022.060
000.00F.0 000..FOF0 000.00F.0 000.00F.0 000.00F.0 000K020 000.00F.0

 

 

 

 

 

000.0 F0

000K F0

000.0 F0

000.0 F0

000.000

9Nll-LLHON

.025. 05 03 00.8 03330 305.3 085850 .00 «50$
2....
000.

0.0. 8.... -. . .. .
. //.k r

,F A
(m.
26.3..“ 8o 2m 3 8; com;
< . . .
. J {K . R
1 833
v a... g m
o 0 _%.0m
I 0 a
V 3.“ z o W
. .o/.\/.\Av [Soda I
N
9
.o
. $8.2m
0
_

 

       

 

 

 

62

 

 

 

 

SNII-LLHON

 

EASTING

2,186,000 2,187,000 2,188,000 2,189,000 2,190,000 2,191,000 2,192,000

Figure 26. Estimated Transmissivity of Layer Three.

63

as". 2.... 8.5. 5.3 803.00 8.5.8.. 3550 .R paw...

0Z....0<m
000.00F.0 000.FOF.0 000.00F.0 000.00F.0 000.00F.0 000KOF.0 000.00F.0
_ _ _

 

0 _/

O
can.

 

 

  

 

 

1 000.0 F0

1 000K F0

1 000.0 F0

1 000.0 F0

1 000.000

I.’-)N|H.LEION

0:0 00.03..— 5 0.0.6..— 00330080 003.250 .00 0.50.0

02....0<w
000.00 F.0 000. F0 F.0 000.00 F.0 000.00 F.0 000.00 F.0 000.50 F.0 000.00 F.0

 

000.0 F0

1 000K F0

 

1 000.0 F0

0
0 O
-N N Am 80.23
p

A A”... 0.

MN“ “N0 m
1 08.80
i\i|l.|’.__, (\I‘W/ﬂ/ M NV

9Nll-LLHON

 

 

 

 

 

65

.030. 00.03 a. 0.0.0..— ..0336550 0033.30 .00 050.".

02....0<m
000.00F.0 000.FOF.0 000.00F.0 000.00F.0 000.00F.0 000.00F.0 000.00F.0

 

 

\ 3. p
00 a

000

 

)

.00/1V.

\Z/ _/

\{\\I 83.8
.5

      

000.0 F0

9Nll-LLUON

I 000.000

 

 

.025. 33 s 223 .sgégoﬁ Banana .8 2am

02_._.m<m
oooﬁm _. .N 80. 5 _. .N Soda _. .N 89mm _. .N ooodm _. .N Soﬁa—um ooodw _. .N
_ _ _ _ _ _ _

 

 

 

 

 

1 086 Pm

COOK N

I 086 _.N

085 E

1 ooodNN

9NIl-LLUON

67

.Som 534 5 £264 hank/6:905 nous—scam Am 25wE

02Fw<m
ooonﬁm coo._.m—..N Goodmwﬁ ooodmﬁm ooodmﬁm ooonPN Soda—um
a

 

 

 

 

a 4 .

m m #835
f

/ K838

086 _.N

 

|EJNIHLHON

1 Good 8

 

\J 1 Soda
g _

 

LIST OF REFERENCES

LIST OF REFERENCES

Anderson, M. P. and Woessner, W. W., 1992. Applied Groundwater Modeling: San Diego:
Academic Press, Inc., 381 p.

Berkowitz, B., Bear, J., and Braester, C., 1988. Continuum models for contaminant transport in
fractured porous formations. Water Resources Research, v. 24, no. 8, p. 1225-1236.

Berkowitz, B. and Braester, C., 1991. Solute transport in fracture channel and parallel plate
models. Geophysical Research Letters, v. 18, no. 2, p. 227-230.

Bouwer, H. and Rice, RC, 1976. A slug test for determining hydraulic conductivity of unconﬁned
aquifers with completely or partially penetrating wells. Water Resources Research, v. 12, no.
3, p. 423-428.

Bruner, DR, 1993. Scale effect of hydraulic conductivity measurements in pre-Illinoian and
Wisconsinan tills in Iowa. Geological Society of America Abstracts with Programs, v. 25, no.
3, p. 9.

Cacas, M.C., Ledoux, B., de Marsily, G., Tillie, B., Barbreau, A., Durand, B., Feuga, B., and
Peaudecerf, P., 1990a, Modeling fracture flow with a stochastic discrete fracture network:
calibration and validation. l. The flow model. Water Resources Research, v. 26, no. 3, p. 479-
489.

Cacas, M.C., Ledoux, B., de Marsily, G., Tillie, B., Barbreau, A., Durand, B., Feuga, B., and
Peaudecerf, P., 1990b, Modeling fracture flow with a stochastic discrete fracture network:
calibration and validation. 2. The transport model. Water Resources Research, v. 26, no. 3, p.
491-500.

Chaivre, K.R., 1972. The feasibility of determining joint orientations by sampling all sizes of
joints. M.S. thesis, Wayne State University, 97 p.

Cooley, R.L. and Naff, R.L., 1990. Regression modeling of ground-water ﬂow: US. Geological
Survey Techniques of Water-Resources Investigations, book 3, chap. B4, 232 p.

Day, M.J., 1977. Analysis of movement and hydrochemistry of groundwater in the fractured clay
and till deposits of the Winnipeg area, Manitoba. M.S. thesis, University of Windsor.

Dorr, John A., Jr., and Eschman, Donald F., 1970. Geology of Michigan: Ann Arbor, Mich.,
University of Michigan Press, 476 p.

68

69

Foote, GS. 1989. Application of environmental isotopes as a test for fracture ﬂow in argillaceous
glacial sediments. M.S. thesis, Michigan State University, 67 p

Fredericia, J ., 1990. Saturated hydraulic conductivity of clayey tills and the role of fractures.
Nordic Hydrology, v. 21, p. 119-132.

Freeze, RA. and Cherry, J .A., 1979. Groundwater: Englewood Cliffs, New Jersey: Prentice Hall,
Inc., 604 p.

Gale, J .E., 1982. Assessing the permeability characteristics of fractured rock. GSA Special Paper
189, p. 163-181.

Gerber, RE. and Howard, Ken W.F., 1992. Migration of contaminants through till: Implications
for the selection of landﬁll sites. Geological Society of America Abstracts with Programs, v.
24, no. 7, p. 300.

 

Grisak, GE. and Cherry, J .A. 1975. Hydrologic characteristics and response of fractures till and
clay conﬁning a shallow aquifer. Canadian Geotechnical Journal, v. 12, no. 1, p. 23-43.

Grisak, G.E., Cherry, J .A., Vonhof, J.A., and Blumele, JP, 1976. Hydrogeologic and
hydrochemical properties of fractured till in the Interior Plains Region. Glacial Till, Special
Publication No. 12 of the Royal Society of Canada, ed by RF. Legget, p. 304—335.

Grisak, GE. and Pickens, J .F., 1980. Solute transport through fractured media, 1. The effect of
matrix diffusion. Water Resources Research, v. 16, no. 4, p. 719-730.

Grisak, G.E., Pickens, J.F., and Cherry, J.A., 1980. Solute transport through fractured media. 2.
Column study of fractured till. Water Resources Research, v. 16, no. 4, p. 731-739.

Harrison, B., Sudicky, E.A., and Cherry, J .A., 1992. Numerical analysis of solute migration
through fractured clayey deposits into underlying aquifers. Water Resources Research, v. 28,
no. 2, p. 515-526.

Hendry, MJ. 1982. Hydraulic conductivity of a glacial till in Alberta. Ground Water, v. 20, no. 2,
p. 162-169.

Hendry. M.J., 1983. Groundwater recharge through a heavy-textured soil. Journal of Hydrology, v.
63, p. 201-209.

Herzog, B.L., Grifﬁn, R.A., Stohr, C.J., Folimer, LR, and Morse, W.J., 1989. Investigation of
failure mechanisms and migration of chemicals at Wilsonville, Illinois. Ground Water
Monitoring Review, v. 9, no. 2, p. 82-89.

Hill, M.C., 1992. A computer program (MODFLOWP) for estimating parameters of a transient,
three-dimensional, ground-water ﬂow model using nonlinear regression: US. Geological
Survey Open-File Report 91-484, 358 p.

Holtschlag, DJ ., 1994. A generalized estimate of ground-water recharge rates in the lower
peninsula of Michigan: US. Geological Survey Water-Supply Paper 2437, 81 p.

70

Hull, L.C., Miller, J .D., and Clemo, T.M., 1987. Laboratory and simulation studies of solute
transport in fracture networks. Water Resources Research, v. 23, no. 8, p. 1505-1513.

Hvorslev, M.J., 1951. Time lag and soil permeability in ground water observations: US. Army
Corps Engrs. Waterways Exp. Sta. Bull. 36, Vicksburg, Mississippi.

Huntec, Ltd., 1967. Report on a combined geophysical survey. Milan, Michigan, for the Martin-
Marietta Corporation, Toronto, Ontario. 25 p., 13 plates.

Keller, C.K., Van der Karnp, G., and Cherry, J .A. 1986. Fracture permeability and groundwater
flow in clayey till near Saskatoon, Saskatchewan. Canadian Geotechnical Journal, v. 23, p.
229-240.

Keller, C.K., Van der Karnp, G., and Cherry, J .A. 1988. Hydrogeology of two Saskatchewan tills,
I. fractures, bulk permeability, and spatial variability of downward ﬂow. Journal of
Hydrology, v. 101, p. 97-121.

Kunkle, George Robert, 1961. The ground water geology and hydrogeology of Washtenaw County
and the Upper Huron River Basin. Ph.D. Dissertation, University of Michigan, 242 p.

Long, J .C.S., Remer, J .S., Wilson, CR, and Witherspoon, P.A., 1982. Porous media equivalents
for networks of discontinuous fractures. Water Resources Research, v. 18, no. 3, p. 645-658.

Long, J CS. and Witherspoon, P.A., 1985. The relationship of the degree of interconnection to
permeability in fracture networks. J ournal of Geophysical Research, v. 90, no. B4, p. 3087-
3098.

Long, J CS. and Billaux, D.M., 1987. From ﬁeld data to fracture network modeling: an example
incorporating spatial structure. Water Resources Research, v. 23, no. 7, p. 1201-1216.

McDonald and Harbaugh, 1988. A modular three-dimensional ﬁnite difference ground-water ﬂow
model. Techniques of Water-Resources Investigations, Book G, Chapter A1. US. Geological
Survey, Washington.

McKay, L.D., 1991. Groundwater ﬂow and contaminant transport in a fractured clay till. Ph D
thesis, University of Waterloo.

McKay, L.D., Cherry, J.A., and Gillham, R.W. 1993. Field experiments in a fractured clay till 1.
hydraulic conductivity and fracture aperture. Water Resources Research, v. 29, no. 4, p. 1149-
1162.

McNamee, Porter and Seeley, 1965. Report on groundwater, section 29, 30, 31, and 32. Augusta
Township, Washtenaw County, Michigan, for the Cement and Lime Division of Martin-
Marietta Corporation. 17 p., 9 plates.

Moreno, L., Tsang, C., Tsang, Y., and Neretnieks, 1., 1990. Some anomalous features of flow and
solute transport arising from fracture aperture variability. Water Resources Research, v. 26,
no. 10, p. 2377-2391.

 

71

Neyer, Tiseo and Hindo, 1987. Report on subsurface investigation. Prepared for Augusta
Development Corporation. Project No. 871590W. October 16, 1987.

Payne, BR, 1972. Isotope Hydrology: Adv. Hydrosci, v. 8, p. 95-138.

Pollock, D.W., 1989. Documentation of computer programs to compute and display pathlines
using three-dirnensional fmite-difference ground-water flow model: US. Geological Survey
Open-File Report 89-381, 188 p.

Rowe, R.K. and Booker, J .R., 1990. Contaminant migration through fractured till into an
underlying aquifer. Canadian Geotechnical Journal, v. 27, p. 484-495.

Rowe, R.K. and Booker, J .R., 1991. Modeling of two-dimensional contaminant migration in a
layered and fractured zone beneath landﬁlls. Canadian Geotechnical Journal, v. 28, p. 338-
352.

RMT, 1988. Preliminary hydrogeologic investigations of Arkona Road Landﬁll. May 1988.

RMT, 1992. Envotech Resource Center, Hydrogeologic report for hazardous waste management
facilities. Madison, Wisconsin.

Rudolph, D.L., Cherry, J .A., and Far'volden, RN, 1991. Groundwater ﬂow and solute transport in
fractured lacustrine clay near Mexico City. Water Resources Research, v. 29, no. 9, p. 2187-
2202.

Ruland W.W., Cherry, J .A., and Feenstra, S., 1991. The depth of fractures and active ground-
water ﬂow in a clayey till plain in southwestern Ontario, v. 29, no. 3, p. 405-417.

Snow, David T., 1967. Hydraulic properties of the Dundee Limestone near Milan, Michigan, and
an evaluation of quarry seepage for the Martin-Marietta Corp., 42 p.

Snow, David T., 1969. Anisotropic permeability of fractured media. Water Resources Research, v.
5, no. 6, p. 1273-1289.

Sudicky, EA. and Frind, E.O., 1982. Contaminant transport in fractured porous media: Analytical
solutions for a system of parallel fractures. Water Resources Research, v. 18, no. 6, p. 1634-
1642.

Sudicky, EA. and McLaren, R.G., 1992. The Laplace Transform Galerkin Technique for large-
scale simulation of mass transport in discretely fractured porous formations. Water Resources
Research, v. 28, no. 2, p. 499-514.

Thoma, S.G., Gallegos, DP, and Smith, D.M., 1992. Impact of fracture coatings on
fracture/matrix ﬂow interactions in unsaturated, porous media. Water Resources Research, v.
28, no. 5, p. 1357-1367.

Thomson, David Gray, 1990. Hydraulic evidence of Wisconsinan-aged fractures in a deep clayey
till. M.S. thesis, University of Waterloo, 179 p.

72
Twenter, F.R., Knutilla, R.L., and Nowlin, J .O., 1976. Water resources of Washtenaw County,
Michigan. U.S.G.S. and Washtenaw County Metropolitan Planning Commission. 74 p.

Williams, RE. and Farvolden, RN, 1967. The inﬂuence of joints on the movement of ground
water through glacial till. Journal of Hydrology, v. 5, p. 163-170.

Zimmerman, R.W., Chen, D., and Cook, N.G.W., 1992. The effect of contact area on the
permeability of fractures. Journal of Hydrology, v. 139, p. 79-96.

"‘1111111111111“