a.
It...

4.1.1.7..a5 ‘—

if.
I a
22. a; .

1.

what $1.: . €qu

1 .1Vuuiﬂwﬁam. .f‘
I

Z

in. a»...

It 24!.1.

raft 93
bl u.......3...
A {V I... r
. :7!!!

 

 

 

 

 

n . s,
in 4.4. ...., .u .

 

 

THtﬁi-“i

MICHIGAN STATEU

IIIIIIIII IIZIII IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

0405 6372

 

 

 

 

 

 

 

 

 

 

 

This is to certify that the

thesis entitled

ALTERATION OF HYDRAULIC CUNDUCTIVITY DUE TO
BATCH ORGAND-MDDIFICATIUN
OF AN AQUIFER MATERIAL

presented by

James Michael Grant

has been accepted towards fulﬁllment
of the requirements for

M.S. (mgﬁeh‘CiVil Engineering

ﬁwﬂﬂ/

Major professor

 

Date {/Zf/Q5

0-7639 MS U is an Afﬁrmative Action/Equal Opportunity Institution

 

 

 

LIBRARY

Michigan State
University

 

 

 

PLACE IN RETURN BOX to roman this checkout from your record.

TO AVOID FINES return on or Moro dd. duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU In An Afﬁnndivo Action/Equal Opportunhy Inﬂation
_ _ Wm:

ALTERATION OF HYDRAULIC CONDUCTIVITY DUE TO
BATCH ORGANO-MODIFICATION
OF AN AQUIFER MATERIAL
By

James Michael Grant

A THESIS

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

MASTER OF SCIENCE

Department of Civil and Environmental Engineering

1996

ABSTRACT
ALTERATION OF HYDRAULIC CONDUCTIVITY DUE To
BATCH ORGANO-MODIFICATION
or AN AQUIFER MATERIAL
By

James Michael Grant

F alling-head permeability tests in a ﬁxed-ring consolidometer were performed on
Oshtemo B soil, an aquifer type material, before and after batch treatment with
hexadecyltrirnethyl ammonium chloride (HDTMA-Cl). Additionally, the hydraulic
conductivity of the treated soil was explored at two different packing states, water
saturated and air dry. It was found that batch treatment of the soil with HDTMA-Cl
caused an initial reduction to the hydraulic conductivity at low effective stress states.
Increasing the effective stress on an organo-modiﬁed soil sample caused the hydraulic
conductivity to decrease at a lesser rate than that of an untreated soil sample. It was
hypothesized that batch organo-modiﬁcation imposed a higher degree of orientation, i.e.
more parallel arrangement, on the clay fabric and that the hydrophobic clay assemblages
existed predominately within the pore space created by the non-clay fraction of the soil.
The implications of these results were discussed in terms of their relevance to engineered

systems.

This thesis and all of the work that it represents, both previous to and during my graduate
studies, is dedicated to my parents Harry and Judith Grant. I am grateful and indebted to
them for their ﬁnancial support and spiritual motivation and encouragement, without
which I would not be where I am currently.

This work is also dedicated in part to Mr. James Wackerlin. If it had not been for his

input of the current market place and opinions on the importance of, and lack thereof,

higher degrees in the sciences, I may have never ventured into a graduate program. A
decision that I know I would have regretted.

iii

ACKNOWLEDGMENTS

I am extremely grateful to Dr. Roger Wallace and Dr. Thomas Voice for all of the
knowledge they contributed and patience they exercised as this research evolved. On
many occasions when everyone involved had thought processes leading down opposite
paths, they managed the team well enough to get things back on track and keep the goals
of the project in focus.

There is no way that I can express my inﬁnite appreciation for Dr. Thomas
Wolff. I feel very fortunate to have wandered into his soil mechanics class as a
misguided, undergraduate student with an undeclared engineering major. Dr. Wolff‘s
wisdom of the geotechnicai science and his command of its practical application sparked
an interest in me that eventually led to a graduate degree and a young, conﬁdent civil
engineer. He has helped me set professional and personal standards, both in and out of

the classroom, that I will always be grateful for.

iv

TABLE OF CONTENTS

LIST OF TABLES ...................................................... vi

LIST OF FIGURES .................................................... vii

I. INTRODUCTION .................................................... 1

II. METHODS AND MATERIALS ......................................... 8

2.1. Soil Batch Preparations ......................................... 8

2.2. Sample Preparation ........................................... 10

2.3. Vacuum Saturation ............................................ 14

2.4. Sample Consolidation and Hydraulic Conductivity Measurement ....... 15

2.5. Sample Disassembly .......................................... 17

III. RESULTS AND DISCUSSION ........................................ 19

IV. ENGINEERING SIGNIFICANCE ...................................... 33

V. CONCLUSIONS .................................................... 42

APPENDIX A : PROCEDURAL DETAILS ................................. 44

APPENDIX B : LOAD FRAME CALIBRATION DATA ...................... 47
APPENDIX C : SAMPLE SPREADSHEET OUTPUT FOR DATA

REDUCTION AND ERROR ANALYSIS ..................... 49

BIBLIOGRAPHY ...................................................... 50

LIST OF TABLES

Table 1: Variation of the Measured Hydraulic Conductivity ..................... 20
Table 2: Variation of the Measured Dry Density .............................. 22
Table B1: Load Frame #1 Calibration Data .................................. 47
Table BZ: Load Frame #2 Calibration Data .................................. 48

vi

LIST OF FIGURES

Figure 1: Consolidation Load Frame and a Cross-Section of the F ixed-Ring
Consolidometer ................................................. 6

Figure 2: Oshtemo B Soil Adsorption of HDTMA-Cl ........................... 9

Figure 3: Vacuum-Dessicator System used to Saturate Oshtemo B Samples
with Water .................................................... 12

Figure 4: Hydraulic Conductivity vs. Effective Stress for Treated and Untreated
Oshtemo B Soil ................................................ 24

Figure 5: Porosity vs. Effective Stress for Treated and Untreated Oshtemo B Soil . . . . 25

Figure 6: Effective-Pore-Size Index vs. Effective Stress for Treated and

Untreated Oshtemo B Soil ....................................... 26
Figure 7: Schematic Representation of an In-Situ Sorption Zone Constructed by

the Injection of a Cationic Surfactant Solution ........................ 34
Figure 8: Schematic Illustration of a Typical Jet Grouting Operation .............. 39

Figure 9: Typical Triple Liner System Showing the Redundancy in the Design of
Horizontal Barriers ............................................. 4O

vii

I. INTRODUCTION

Volatile organic compounds (V OCs) present in groundwater and soil reportedly
have a potential to produce adverse health effects. Natural deposits of low organic matter
content soil, clay soil, and aquifer materials have little sorptive capacity to serve as
temporary or permanent sinks for VOCs, especially for nonionic organic compounds
(N OCs) that are poorly soluble in water. The native inorganic cations are strongly
hydrated in the presence of water, creating hydrophilic clay surfaces and interlayers,
hence a poorly water soluble organic molecule cannot compete for the mineral adsorption
sites (Lee et al. , 1989). This results in the mobility and transport of organic contaminants
in the subsurface, especially compounds that possess high water solubilities. The limited
sorptive capacity of some such soils can be greatly improved by replacing the natural,
metal cations with large organic cations through ion-exchange reactions (Lee et al. , 1989;
Jaynes and Boyd, 1991). These organic cations ﬁx on the clay giving it the surface
properties of an organic phase which results from the alkyl hydrocarbon tails that project
out ﬁ'om the clay surface (Jaynes and Boyd, 1991). The organic cations that have been
studied most extensively are quaternary ammonium compounds (QACs) of the general
form [(CH3)3NR]+ where R is an alkyl or an aromatic hydrocarbon (Boyd et al. , 1991).

When R is a relatively large alkyl hydrocarbon, as is the case for hexadecyltrimethyl

2

ammonium (HDTMA) where R=(CH2),,CH3, then the modiﬁed clays are referred to as
organophilic clays due to the character that is imparted to the clay surfaces and the
interlayers (Boyd et al., 1991).

Organo-modiﬁcation of a soil results in a greater sorption capacity for dissolved
organic contaminants and hence improved retardation capabilities for low organic matter
content soils and aquifer media. Mechanistically, the organic phase that results from the
ion-exchange process behaves analogous to the partitioning observed in natural soil
organic matter (J aynes and Boyd, 1991). It has been shown that the organic phase
derived from HDTMA treatment formed an effective partition medium that was 10 to 30
times more effective on a unit weight basis than natural organic matter (Lee et al. , 1989;
J aynes and Boyd, 1991). The enhanced sorption characteristics could be managed to
intercept a migrating plume and retard the advective and the diffusive advance of a
contaminant mass through a region. This implies that modiﬁed soils could become useful
remedial tools for containment technologies.

Organo-modiﬁed soils have been suggested to have applications as containment
barriers (Alther et al. , 1990; Boyd et al., 1991), as in-situ sorptive zones (Boyd et al. ,
1991; Burris and Antworth, 1992), for stabilization of hazardous wastes (Alther et al. ,
1990; Boyd et al. , 1991; Brown et al. , 1992), for water and wastewater treatment
(Srinivasan and Fogler, 1989; Alther et al. , 1990), and as an immediate response to
terrestrial chemical spills to limit the extent of contamination (Boyd et al. , 1991). Each
application requires the blockage of or the bulk transport of ﬂuid through a treated soil

region. An evaluation of the modiﬁed soil's hydraulic conductivity in different chemical

3

environments is necessary to evaluate the advective transport characteristics and hence
feasibility of potential organo-modiﬁed soil technologies.

Hydraulic conductivity is probably the most important parameter for assessing
contaminant migration (Evans, 1991). Saturated hydraulic conductivity is a macroscopic
property of porous media that depends on the medium properties as measured by the
intrinsic permeability and the ﬂuid properties of viscosity and speciﬁc weight. Intrinsic
permeability captures the integrated inﬂuence of the pore geometry and depends on
porosity and pore size distribution of the matrix. These properties are all macroscopic
characterizations of the media which are dependent on other features of the soil matn'x.
Of importance are the grain size distribution and grain shape, clay type and content,
structure, packing, mixing, cementation, stress state and chemical characteristics of the
pore liquids. Allred and Brown (1994) used the Hagen-Poiseuelle equation to show the
dependency of the hydraulic conductivity on the porosity, the effective pore radius, and

the pore ﬂuid properties:

2
K_n_*r__v (1,
8*}:

where K is the hydraulic conductivity, n is the porosity, r is the effective pore radius, 7 is
the speciﬁc weight of the ﬂuid, u is the ﬂuid viscosity, and 8 is a constant arising out of
the geometry of the system. Gross volumetric parameters such as porosity cannot entirely

account for changes in the hydraulic conductivity of ﬁne grained soils as they are better

4

related to pore-size distributions (Garcia and Lovell, 1981). When no microscopic data is
available for soils, the effective-pore-size index (r2) from equation (1) becomes a useful
tool for making conjectures at this level. Its relative magnitude when comparing soils is a
direct reﬂection of the integrated pore shapes and is an indication of whether a large
number of small pores or a small number of large pores are contributing to the gross
porosity.

Hydraulic conductivity is signiﬁcantly inﬂuenced by the soil's structure which
consists of both the geometric particle arrangement and the interparticle forces of a soil
(Lambe, 1958a; Acar and Ghosh, 1986). The geometrical relationships established by the
soil particles is known as the soil fabric. The clay ﬁaction of the soil fabric can be
characterized at two levels, a ﬁrst-order and a second-order recognition pattern (Yong and
Warkentin, 1975). The second-order recognition level describes the geometrical
arrangement of single clay particles within individual, identiﬁable groups or group units
called fabric units. The ﬁrst-order recognition level describes the geometrical
arrangement of the fabric units. The response behavior of clay soils to mechanical
manipulation is through the fabric unit interactions and the various bonding mechanisms
between these units; therefore, physical soil testing can only provide an evaluation of the
soil's fabric at the ﬁrst-order recognition level (Yong and Warkentin, 1975).

The response of a soil mass to changes in applied stress depends almost
exclusively on the effective stresses in that mass. Effective stress is deﬁned as the
difference between an engineering total stress and a measurable pore water pressure. It is

not a directly measurable, physical quantity and should not be confused as a grain-to-

5

grain contact stress. The effective stress can be increased by increases in the total stress
of a soil mass (i.e., increased bearing loads on a soil or greater depths of overburden
material) and by decreases in the pore water pressure (i.e., ﬂuctuations in the groundwater
table). The porosity and effective pore radius, and hence the hydraulic conductivity, are
dependent on the effective stress state; therefore, the ﬂuid transport properties of a soil
mass can be expected to change with varying effective stresses.

A ﬁxed-ring consolidometer (Figure 1) is an apparatus that allows the
measurement of a soil's macroscopic volumetric properties at different effective stress
states. A controlled total stress can be applied to a soil sample and vertical deformations
can be measured during and at the conclusion of primary consolidation (i.e., complete
dissipation of the induced pore water pressure by drainage such that the total stress equals
the effective stress). The attached vertical standpipe (Figure 1) can be employed to
conduct falling-head permeability tests. This allows for the measurement of hydraulic
conductivity at varying effective stresses. The consolidometer has the advantages of
simulating ranges of overburden pressures that could be encountered in the ﬁeld and of
reducing the likelihood of any signiﬁcant side-wall leakage, and hence channeling, during
permeability testing.

The scope of this investigation is to provide data on the hydraulic conductivity
alterations induced by batch organo-modiﬁcation and the resultant response of the soil to
differing stress states. A preliminary model of the soil fabric's alteration by batch
treatment with the cationic surfactant (HDTMA) is developed. A ﬁxed-ring

consolidometer will be employed to impose varying effective stress states and measure

cmﬁsoEomcoo mcEivoxE 05 .o cowoomlmmoco a 25 use; 33 com—.2328 3 2:2...

 

 

 

 

a
t
it“
If
£235
ii;
‘5 I
I
a
I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

“\FN Eda.”— O 3
gﬂgg '3 s !E
Egg 0
is c! g
I RVA L
E I HE \ L

 

 

 

 

 

ex.

 

 

, if...
E
S
i

 

i

i

E
33

 

 

 

 

 

 

Egg\ 4.... .n trim
x. #24; «my. .34... ......... .3 I

 

 

 

 

h}.

EIS
NEIL? a. E

ale-Id.- 425.110

 

 

 

 

 

 

 

7

gross volumetric parameters on an untreated and on a treated aquifer type material. The
treated soil will be packed in a wet state and in a dry state to assess any changes in
mechanical response induced by drying a hydrophobic soil. Falling-head permeability
tests will be run at each stress state to measure the hydraulic conductivity's degree of
dependence on the effective stress before and after modiﬁcation. The macroscopic data
that is collected will be used to draw conjectures regarding the alteration of the soil fabric

at the ﬁrst-order level as induced by the batch organo-modiﬁcation process.

II. METHODS AND MATERIALS

MW Approximately 50 kg (110 lbs.) of Oshtemo soil

was obtained from a pit located at the Kellogg Biological Station (KBS) between
Kalamazoo and Battle Creek, MI. It was taken from the B-horizon at a depth of
approximately 60 cm (2 ft.) from the surface. The collected soil was air dried, sieved
through a US standard #20 (0.84 mm), and then homogenized using the quartering
method (ASTM C702-80). The homogenized Oshtemo B soil is a well graded , sandy
loam with: 0.4% organic matter content; 78% sand, 19% clay, and 3% silt; 71.6% Ca2+,
22.5% Mg“, 3.7% K+, and 2.2% Na+ as exchangeable cations. The major clay minerals
included one type of swelling clay (vermiculite) and three types of non-swelling clays
(illite, kaolinite, and hydroxy-aluminum interlayered vermiculite). Three separate soil
batches were prepared from this homogenized stock and were given the designations
"untreated-dry", "treated-dry", and "treated-wet" where the term "treated" refers to the
organo-modiﬁcation of the soil and the "wet" or "dry" designation distinguishes the soil
state at the time of packing.

The organo-modiﬁed soils were prepared by adding amounts of
hexadecyltrimethyl ammonium (HDTMA) chloride equal to the maximum amount of
inorganic exchangeable cations that can be replaced by HDTMA-C1. This corresponds to
the ﬁrst plateau of the HDTMA-Cl adsorption isotherm (Figure 2) which is 46.1
mmole/kg (4.61 meq/ 100 g). 160 ml of HDTMA-C1 was added to batch suspensions of
840 ml high performance liquid chromatography (HPLC)-grade water and 25 g of

8

 

 

 

 

120 . i . 1 1 i t
m . .
U)
U
Q)
E 100 - , _
a: O HDTMA Adsorption O
.5 A Cation Release 0
*5 A 80 — -
0 T o
B e
o _ O
.5 E 60 - O _
a E
8
'2 40 - I
<
2
'5
I 20 - HDTMA loading in _
modified soil
0 1 l 1 l 1 l 1 l
-7 ‘ —6 -5. —4 —3
log CHDTMA (CHDTMA in M)

Figure 2: Oshtemo B Soil Adsorption of HDTMA-Cl

Oshtemo B soil and then stirred. A rapid mixing stage of 120 rpm for 1 minute and a
slow mixing stage of 20 rpm for 30 minutes were imposed on the suspensions.
Preliminary jar tests indicated that rapid mixing at 120 rpm for 1 minute produced the
least turbidity when the soil was settled out of suspension, indicating a higher degree of
ﬂocculation. The organo-modiﬁed soil was settled out of suspension and then washed

once with HPLC-grade water by decanting the HDTMA-C1 solution, ﬁlling and agitating

 

10

the beakers with HPLC-grade water, and then decanting again. Insuring that no ﬁnes
were lost, the treated Oshtemo B soil was transferred to a HPLC-grade water bath and
mixed occasionally to prevent layering and to maintain complete water saturation of the
soil. The treated soil was covered by approximately 10 cm (3.9 in.) of bath water at all
times. The treated-dry batch was obtained by air drying soil from this bath.
12W A ﬁxed-ring consolidometer (Figure 1) was used for
testing all of the Oshtemo B soil samples. This device allowed the total volume of a soil
sample to be measured at different stress states. A dial gage and a linear vertical
displacement transducer (LVDT) mounted on the load frame measured the change in
thickness of the soil samples at each effective stress. The measured sample thicknesses at
the conclusion of consolidation (Lload) were actually volumes since the soil was conﬁned
to a constant cross-sectional area (A). The speciﬁc gravity of solids (G5) was measured
using the pyncometer test (ASTM D854) and it was assumed that the volume of soil
solids (V s) remained constant; therefore, any measured changes of the sample volume (V)
associated with varying the effective stress were attributed to void volume (V v) changes.
This allows for calculation of the soil's volumetric parameters at different stress states.

The porosity (n) at each effective stress state was calculated using the following equation:

 

n:_1:____': ' (2)

 

11

Here M, was the oven dry mass of the soil solids and the rest of the terms were as deﬁned
above.

The term "index" will be used throughout these procedures as reference to the
process of taking dial gage readings at four points, each 90 degrees to the previous point,
around the top of the upper porous stone assembly and then averaging these values to
obtain an equivalent "height" for that speciﬁc assemblage of the ﬁxed-ring
consolidometer (Figure 1). Initially, a 2.54 cm (1 in.) "height" datum (me) was
established for a ﬁxed position of the dial gage and load frame pedestal by inserting a
machined, metal puck of 2.54 cm (1 in.) thickness and 6.35 cm (2.5 in.) diameter with
ﬁlter paper on each surface within the completely assembled ﬁxed-ring consolidometer
and indexing. As long as the dial gage and load frame pedestal position was kept
constant, the thickness of any soil within the sample ring of the consolidometer could be
calculated, even if the assembly was removed from the load frame, by indexing against
the 2.54 cm (1 in.) reference datum. The establishment of this datum was the ﬁrst step
for all of the sample preparations, regardless of the soil batch that was being tested.

The treated-wet samples were prepared by adding treated Oshtemo B soil slurry
from the HPLC-grade water bath to a ﬁxed-ring consolidometer within a desiccator
system (Figure 3) that allowed for drainage and vacuum saturation. The empty
consolidometer was assembled without the base plug, stand-pipe, or upper porous stone
assembly and set within the desiccator. A vacuum was applied to the system to deair the
HPLC-grade water that would be used for saturation and to vacuum saturate the base

components of the consolidometer. The vacuum was collapsed on the desiccator only

L325 5? maisom m 2.850 2.838 8 v8: Eoﬁam Loﬁoﬁmooieaaog "n 952...

 

 

12

agggggﬂggﬂggon.
gggngnggt

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

cleanse-3.28.12 3.3a
lip—1.38%.";
Engaging 3:983!
23$
53/ 5.5 385
5!" IX)
IX .
3.62
au¢3a8I\
82:81:
.5852 age
L, L, I.
v girl
«I G G
".1888.“ «1.8886
b \h/
g

 

 

 

 

 

 

13

and the water within was drained by a siphon to a level equal to the top surface of the
lower porous stone. Appendix A contains more detail on this procedure.

Approximately 160 g (0.35 lb.) of the treated soil slurry was poured into the sample ring
of the consolidometer. The slurry was scooped from random regions of the HPLC-grade
water bath and transferred to a tared container before it was poured into the sample ring.
Light tamping with a copper rod was implemented during pouring to avoid layering
effects associated with differential settling velocities of the soil grains and to create a ﬂat
upper surface on the soil sample. Drainage was allowed until there was no standing water
on top of the soil sample. An attempt was made to minimize the amount of ﬂuid
associated with the slurry. This seemed to result in greater reproducibility in the initial
dry density of the samples and sped up the drainage process. The upper ﬁlter paper and
porous stone assembly were set on the soil sample slowly to minimize tilt and the
desiccator was rescaled in preparation for the sample vacuum saturation procedure.

The treated-dry and untreated-dry samples were packed with the intention of
replicating the initial conditions (dry density) of the wet samples as closely as possible.
The compaction device employed had the following speciﬁcations: 1.47 kg (3.24 lb)
hammer, 664 g (1.46 lb) hammer head, 7.5 cm (2.9 in.) hammer drop. The hammer head
was machined to a 6.35 cm (2.5 in.) diameter and a 2.54 cm (1 in.) thickness so that it
would just ﬁt in the efﬂuent reservoir and was the same diameter as the sample ring of
the ﬁxed-ring consolidometer, thus preventing the loss of soil during packing. After the
initial indexing to establish Lref was completed, the empty consolidometer was assembled

without the base plug, stand-pipe, or upper porous stone assembly and either 130 g (0.29

14

lb.) of untreated-dry Oshtemo B soil or 140 g (0.31 lb.) of treated-dry Oshtemo B soil
was placed in the ring. Best results were obtained when vacuum grease was used around
the consolidometer's O-ring to prevent leakage. The untreated-dry samples were tamped
4 times and the treated-dry samples were tamped 2 times. The upper ﬁlter paper and
porous stone assembly were set on the soil and the sample was indexed to obtain the
packed height of the specimen (me). The soil sample and the consolidometer were set
inside the desiccator and it was rescaled in preparation for the sample vacuum saturation
procedure.

WED. A vacuum is applied to the entire system to reestablish
the negative headspace pressure in the desiccator for the treated-wet samples, and to deair
the HPLC-grade water for the dry packed samples. While the vacuum was maintained on
the desiccator only, it was ﬂooded with deaired HPLC-grade water by opening stopcock
no. 1 (Figure 3) to the ﬂask. When the level was just below the lip of the
consolidometer's effluent reservoir, ﬂow was discontinued. Slowly removing the rubber
stopper on the reservoir ﬂask and exposing the headspace to atmospheric pressure
increases the total head, and hence the differential head driving the ﬂow and drastically
reduces the time necessary to ﬂood the desiccator. The vacuum was maintained on the
desiccator over night (an average of 15 hours). Whenever there was any leakage at the
consolidometer's O-ring seal there were small air bubbles drawn out at the steel—to-steel
interface and the sample was repacked. After 15 hours, the water level within the efﬂuent
reservoir should have risen above the upper porous stone unless delivery of water was

blocked through the base components of the consolidometer or the soil sample. Water

15

was always observable within the efﬂuent reservoir for this set of experiments; however,
if this were not true it could be an indication of insufficient saturation of the soil sample.
The vacuum within the desiccator was collapsed. The desiccator top was removed and
the base plug and stand-pipe were attached to the consolidometer while it was still
immersed. It was best to ﬁll the bottom half of the stand-pipe with water before attaching
to avoid introducing air bubbles into the base components of the consolidometer.
Wrapping the base plug with Teﬂon tape was also helpful to prevent leakage. The
consolidometer assembly was removed from the system, dried off, and then indexed to
calculate the saturated height of the soil sample (Lm). The difference between the
saturated thickness and the packed thickness of the dry soil samples represents the change
in volume due to swelling. Due to the uncontrolled nature of the swelling interface of the
soil, it was difficult to maintain constant bulk densities during saturation. The untreated
Oshtemo B soil displayed a greater tendency to swell due to its hydrophilic nature. The
stainless-steel ball bearing was set into place in preparation for the consolidation
sequence. Appendix A contains more detail on the vacuum saturation procedures.

‘ _-.-uo ‘ 0001...: or an ,- ua ._ on. ' Ufa. -- ‘u‘r Theload
frame was tared for the consolidometer assembly then the saturated soil samples were set
in the load frame (see Appendix B for calibration data) and subjected to a loading
sequence from 24 kPa (0.25 ton/ﬁz) to 766 kPa (8.0 ton/ﬁz). Appendix A contains
detailed procedures for taring the load frame. A load increment ratio (LIR) of unity,
which means that the load was doubled each increment. The consolidation procedure was

a modiﬁed form of the conventional consolidation test (ASTM D243 5-90) that employed

16

a rapid-loading sequence. Each load increment was imposed on the soil for one hour
before a falling-head permeability test was performed and the next load was applied.
Evidence suggests that for a LIR of unity, variable loading times greater than the time for
90% consolidation (t90) have an insigniﬁcant affect on the void ratio vs. effective stress
relationship as would be obtained using the classical approach of a 24 hour load duration
(Olson, 1986). Taylor's square root of time ﬁtting method was used to obtain tgo on the
order of 1-2 minutes for the untreated Oshtemo B soil. It was assumed that the treated
soil would possess a t90 on the same order of magnitude and that a load duration of one
hour would still be adequate.

Once the soil sample had consolidated under each effective stress, a conventional
falling-head permeability test with constant tail water level (ASTM D5084-90) was run .
The ﬁxed-ring consolidometers (Figure 1) were outﬁtted with a standpipe to allow
permeability tests to be run. The inner diameters (a) were 0.258 cm2 (0.040 inz) for the
glass and 0.335 cm2 (0.052 in2) for the plastic standpipes. This allowed adequate time
scales to measure hydraulic heads at intermediate points and validate Darcy's Law. This
was done for each falling-head test by plotting the logarithm of the hydraulic head vs. the
cumulative time to insure that a linear relationship existed. This checked for the
continuity of ﬂow and that there was no system leakage during experimentation. The
sample thickness was noted (how), the temperature of the permeant was measured, and
then 2 or 3 tests, corresponding to 3 ml-8 ml of ﬂushed deaired HPLC-grade water, were
conducted at each state of stress. The total volume of permeant ﬂushed through the soil

by the end of the entire loading sequence varied from 34 ml to 48 ml which corresponds

17

to approximately 1.0 to 1.5 pore volumes. The hydraulic conductivity for each test was
corrected for temperature effects on the permeant's ﬂuid properties using a correction
factor (RT). All values reported correspond to an equivalent hydraulic conductivity at 20
C (68 F). The hydraulic conductivity for each trial was calculated using:

2.303*a*L *R (bl-Ah)

K: Load T*lO 3
A*t gag—Ah) ()

 

where Ah was the height of the tailwater datum, hl was the initial hydraulic head, h2 was
the ﬁnal hydraulic head, and t was the elapsed time for falling head. All other terms are
deﬁned previously. The average hydraulic conductivity value was reported for each
effective stress (Table 1). The measured hydraulic conductivity values from the falling-
head tests in the consolidometer are actually equivalent values that account for headloss
through the entire system. Experiments were conducted that determined the contribution
of the porous stones to ﬂuid resistance was negligible, hence the calculated hydraulic
conductivity is representative of the soil sample only. A spreadsheet was constructed for
the data reduction and error analysis of each experiment (see Appendix C for a sample
output).

WW At the conclusion of the loading sequence and the last
falling-head permeability test, the ﬁnal dial gage reading was taken and the ﬁxed-ring
consolidometer was removed from the loading ﬂame. The air pressure to the load frame
assembly was vented by turning the load dial to the "unload" position. The treated-wet

samples were removed from the consolidometer and placed in an oven to dry at 110 C.

18

This would allow for calculation of the dry weight of soil solids (W,). The oven-dried,
treated-wet samples and the dry packed samples were visually inspected for channeling
along the sample ring wall and for migration of ﬁnes. Ultrasonic baths were used to
clean the porous stones at the conclusion of each experiment. It was assumed that no
microbial growth was incurred during stone storage to contribute to ﬂow resistance in

subsequent experiments.

III. RESULTS AND DISCUSSION

Hydraulic conductivity is dependent on many factors which can be broadly
deﬁned as ﬂuid property effects or as soil matrix effects. The intrinsic permeability is a
soil property that represents the integrated inﬂuence of the pore geometry on the ﬂuid
conductance of the material. When the effective stress is varied on a soil mass, the
geometric alteration is captured in this macroscopic parameter which, as deﬁned earlier,
depends on porosity and pore size distribution. The dominance of particular clay
aggregate arrangements can be associated with certain types of engineering behavior
(Collins and McGown, 1974), thus the mechanical response of the treated and untreated
soil to increasing effective stress will be used to develop a hypothesis of the effect that
batch HDTMA-Cl treatment has on the Oshtemo B soil's hydraulic conductivity and
geometric particle arrangement. The experimental data are presented in Tables 1 and 2,
and in Figures 4 through 6.

The measured hydraulic conductivity of a sample in the ﬁxed-ring consolidometer
has some uncertainty associated with it. Uncertainty is a quantity that indicates the
variability of the best estimated value due to calibration errors, data acquisition errors,
sample variability, data reduction errors, and other uncontrolled errors. As shown in
Table 1, hydraulic conductivity was measured with a maximum relative uncertainty of
approximately 4% for one standard deviation. In many experiments the uncertainty was

lower. Uncertainty was estimated for each experiment (see sample spreadsheet output in

19

20

Table 1: Variation of the Measured Hydraulic Conductivity

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Soil Type Load Hydraulic Conductivity (K) Average K Relative

(tsf) (cm/s) (cm/s) Uncertainty

(%)

Trial #1 Trial #2 Trial #3

Untreated 0.25 1.84e-3 2.26e-3 2.25e-3 2.12e-3 3.33

Dry 0.50 1.31e-3 1.32e-3 1.28e-3 1.30e-3 2.94

1.0 6.68e-4 7.04e-4 6.40e-4 6.71e-4 2.79

2.0 2.62e-4 2.29e-4 2.72e-4 2.54e-4 2.73

4.0 1.09e-4 8.07e-5 8.80e-5 9.26e-5 2.67
_ __ 8'°___ 4'29; “9265 283's __le__3'55 ELM

I Treated 0.25 6.89e-4 6.91e-4 8.02e-4 7.27e-4 2.93

Dry 0.50 5 81e-4 5.91e-4 6.66e—4 6.13e-4 2.87

1.0 4.54e-4 4.51e-4 5.14e-4 4.73e-4 2.84

2.0 3.30e-4 3.24e-4 3.63e-4 3.3964 2.88

4.0 2.29e-4 2.22e—4 2.48e—4 2.33e-4 2.87
I __ ﬁn“- __450 5i 163 #3264 _ 28'7 n
I Treated 0.25 2.26e-4 2.05e-4 2.49e-4 2.27e-4 2.70 I

I wet 0.50 183e-4 1.65e-4 1.99e-4 1.82e-4 2.73

1.0 1.31e-4 1.21e-4 1.50e-4 1.34e-4 2.70

2.0 8.59e-5 8.26e-5 1.04e-4 9.08e-5 2.72

4.0 5.02e-5 5.11e-5 6.61e-5 5.58e-5 2.72

8.0 2.55e-5 2.93e-5 3.79e-5 3.09e-5 2.72

 

 

 

 

   

 

 

 

 

21

Appendix C) based on the assumptions that the measurement error was a random variable
with a normal probability distribution function and an expected value of zero, and that the
absolute uncertainty associated with an individual measurement was small enough to
permit the analytical function to be approximated by the ﬁrst linear term of a Taylor
series expansion. The hydraulic conductivity was most sensitive to uncertainty in the
hydraulic head measurements made during a falling-head permeability test. This could
be improved by providing a standpipe with a greater cross-sectional area for slower drops
and having more precise increments for the head measurements.

The measured volumetric parameters of a soil sample at a certain stress state also
had uncertainty associated with it. The relative uncertainty was estimated for each
experiment (see Appendix C) based on the same assumptions as previously mentioned,
and are presented in Table 2 for the dry density calculations. The volumetric parameters
were most sensitive to variances in the measured total volumes of the soil samples at each
effective stress. This could be improved by a procedural reﬁnement. The total sample
volume was dependent on the sample thickness at a particular effective stress (Load). The
absolute uncertainty associated with Lload was dominated by the uncertainty in the
measured initial sample thickness (Linn). It was assumed that the soil sample's initial
thickness after the load frame head was set in place was unchanged from the calculated
Lin,t by indexing. The indicator used to prevent sample compression was the
displacement of a marked line on the load frame's pedestal stem (Figure 1). This line

could only realistically be judged for displacement with an accuracy of 0.1 mm.

 

 

II

F Soil Type

 

t_._

I Untreated
Dry

 

Treated
Dry

Treated
Wet

 

22

Table 2: Variation of the Measured Dry Density

 

 

 

 

 

 

 

 

 

 

 

Load Dry Density (yd) Average yd Relative
(tsf) (g/cm3) (g/cm3) Uncertaint
y (%)
Trial #1 Trial #2 Trial #3
0 1.35 1.32 1.31 1.33 0.33
0.25 1.50 1.46 1.50 1.49 0.37
0.50 1.56 1.53 1.58 1.56 0.39
1.0 1.62 1.59 1.66 1.62 0.40
2.0 1.68 1.68 1.73 1.70 0.42
4.0 1.74 1.74 1.80 1.76 0.44
1.79 1.80 1.87 1.87

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

23

Initially, the batch treatment of Oshtemo B soil with HDTMA-Cl caused a
reduction in the soil's hydraulic conductivity to water at low effective stress states.
Figure 4 shows that the hydraulic conductivity decreased by approximately 50% when
comparing the untreated-dry and the treated-dry samples, and by as much as 90% when
comparing the untreated-dry and the treated-wet samples at an effective stress of 24 kPa
(0.25 tsf). This implied that the initial handling of the treated sample, which involved
chemical treatment and physical compaction, had already caused a bulk reduction in the
pore space available for ﬂuid conductance. Figures 5 and 6 supported this by illustrating
the initial reduction in porosity and effective-pore-size index at low effective stresses.

Increasing the effective stress had a greater effect on the untreated soil such that
the treated-dry Oshtemo B soil had a larger hydraulic conductivity at higher effective
stress states. Figure 4 shows that the hydraulic conductivity of the untreated soil
decreased almost two orders of magnitude over the same effective stress increase for
which the conductivity of the treated soil decreased only one half to one order of
magnitude. The amount of rearrangement and compression that took place in the soil
matrix as the effective stress increased was dependent on its rigidity, which is ultimately
a function of the soil structure. The data implied that the batch modiﬁcation somehow
altered the soil matrix to reduce its compressibility. The reduction in the medium's
compressibility suggested that the treated soil did not have as much opportunity to
rearrange due to physical compression because its fabric had already been altered by

chemical treatment and compaction. This was reﬂected in the lesser degree that

24

 

 

Hydraulic Conductivity (cm/s)

0.01

 

IIII'

untreux-d-dry

 

0-001 treated-dry

l IIIIII
c»

treated-wet

0.0001

 

 

 

000001 1 rLlrrrrl l 1 11....
0.1 1 10

Effective Stress (tsf)

 

 

Figure 4: Hydraulic Conductivity vs. Effective Stress for Treated
and Untreated Oshtemo B Soil

 

25

 

 

 

 

 

 

 

0.5
0.45 -_ untreﬁed-dry
I o
g I treated-dry *
3 0.4 '_ o
o : wax
a : t '1 "‘
: reate -wet .
0.35 1
3 o
0.3- J rrrrrrrl r rlrrr*u
0.1 1 1

Effective Stress (tsf)

 

Figure 5: Porosity vs. Effective Stress for Treated and Untreated Oshtemo B Soil

 

26

 

 

Effective-Pore-Size Index (cm"2)

0.01

 

0.001

IIIII

l

I

 

0.0001

untreated-dry

treated-dry

treated-wet

 

 

0.1

Effective Stress (tsf)

 

Figure 6: Effective-Pore-Size Index vs. Effective Stress for

Treated and Untreated Oshtemo B Soil

 

in

C0

 

27

decreases in the batch treated soil's porosity (Figure 5) and other effects (Figure 6)
occurred as the effective stress increased.

As indicated by equation (1), the hydraulic conductivity is proportional to the
porosity (n) and to the effective-pore-size index (r2); therefore, the observed decrease in
conductivity as effective stress is increased can be discussed in terms of changes in these
parameters. Porosity is simply a ratio of the void volume to the total soil volume
(equation (2)) and contains no information describing the pore space geometry.
Effective-pore-size index describes the distribution of space within the void volume,
which the hydraulic conductivity is more dependent on. It should be noted that the data
reported in Figure 6 is actually K/n and has incorporated the ﬂuid speciﬁc weight and
viscosity and the geometric constant from equation (1) into the effective-pore-size index
term. This is based on the assumption that the ﬂuid properties remained constant during
experimentation, which is not unreasonable to make since correction factors are used to
adjust all of the reported hydraulic conductivities to an equivalent value at 20 C.

The results presented from this investigation are collected using physical soil
testing procedures in a ﬁxed-ring consolidometer. Mechanical soil testing provides a
macroscopic evaluation of the material; therefore, the data gathered by such means can
provide only an indirect assessment of soil fabric as it relates to the ﬁrst-order
characterization of the soil matrix geometry (Yang and Warkentin, 1975). Conjectures
beyond the general orientation of the clay fabric units, i.e. oriented or random
arrangements, is not possible without an analysis involving microscopy procedures to

verify the clay particle interactions.

28

The proposed interpretation was that the batch organo-modiﬁcation procedure
caused ﬂocculation of the clay fabric at the ﬁrst-order characterization into a more
parallel arrangement. The preparation of the dry packed samples provided the initial
evidence that the clay fabric units in the batch treated soil were more oriented. Table 2
shows that initially the untreated-dry sample had a lower dry density than the treated-dry
sample even though a greater compactive effort per unit mass (4 tamps/130 g for
untreated vs. 2 tamps/140 g for treated) was employed in the packing of the untreated
Oshtemo B soil. The compactive effort measured the mechanical stabilization or
densiﬁcation energy imposed on a sample. The soil's resistance to this energy input
provided an indication to the degree of randomness of the soil's clay matrix. It has been
shown that the more random the clay fabric array is, the greater the resistance to
compaction is (Lambe, 1958b; Haug and Wong, 1992). The lower dry density of the
untreated Oshtemo B soil sample reﬂected the greater ability of the soil's clay fraction to
resist compaction relative to that of the treated sample. The untreated Oshtemo B soil
was expected to have a predominately ﬂocculated clay fabric due to more than 90% of the
exchangeable sites being occupied by divalent cations (71.6% Ca2+ and 22.5% Mg”).
The treated-dry Oshtemo B soil was noticeably "slippery" to the touch and conﬁgured
itself into a denser packing when compacted, indicating that the chemically induced
arrangements of the clay aggregates facilitate grain slippage. This was consistent with
observations that the more parallel the clay units are, the less stable and more readily

remolded a soil was upon compaction (Lambe, 1958b; Haug and Wong, 1992). This

 

 

 

 

sug

29

suggested that the ﬂocculation that resulted from batch HDTMA-Cl addition to the
Oshtemo B soil was of a more parallel nature than the untreated soil's clay fabric.
Comparing the untreated-dry and the treated-dry results (Figure 5) showed that
about 10% of the initial loss of conductivity may be attributed to the lost porosity. The
remaining reduction in conductivity at an effective stress of 24 kPa (0.25 tsf) should be
attributed to the change in the effective-pore—size index as shown in Figure 6. The
initially larger porosity and effective-pore-size index of the untreated soil were
indications that a more random arrangement, and hence a greater void volume, were
present. This also suggested a greater potential for rearrangement of the granular matrix
due to the unstable state of a random array. The greater response of the untreated soil's
hydraulic conductivity as effective stress was increased was attributed to this
rearrangement which was facilitated by the alignment of the initially random clay fabric.
It has been shown that one dimensional compression of hydrophilic soils tends to align
the clay fabric normal to the direction of pressure application and to decrease the spacing
between adjacent fabric units (Lambe, 195 8a,b; Leonards, 1962; Yong and Warkentin,
1975). The batch treated Oshtemo B soil exhibited a less dramatic reduction in hydraulic
conductivity because the chemical alteration of the soil fabric seemed to be at the expense
of the larger pores as indicated by the reductions in porosity (Figure 5) and effective-
pore-size index (Figure 6). The treated soil's response to increasing effective stress
supported the conjecture that the organo-modiﬁcation process imposed a higher degree of
orientation on the clay fabric. It appeared from Figure 5 and Table 2 that the compressive

behaviors of the treated and the untreated soils converged at approximately 4 tsf as their

30

macroscopic volumetric parameters became equal. This suggested that once all of the
soil's clay fabric was predominately oriented and parallel to one another that the
mechanical behavior was similar as effective stress increased.

The chemical rearrangement of the clay fabric resulted in macroscopically
measured differences between the treated and the untreated soil response to initial
compaction, one-dimensional consolidation, and permeation. At the microscopic level,
modiﬁcation of Oshtemo B soil with HDTMA-Cl has been shown to cause both
aggregation and dispersion of the layered silicates, depending on the HDTMA loading
levels (Xu and Boyd, 1994a,b). At loading levels below the cation exchange capacity
(CEC), adsorbed HDTMA-Cl replaced the inorganic exchangeable cations (Ca2+, Mgz”,
K“, and Na“) and compensated the negative charges of the layered silicates. A face-to-
face, parallel type association developed due to the mutual attraction between the alkyl
tails of the adsorbed HDTMA-Cl on the faces of adjacent clay sheets, and due to the
tendency of the hydrophobic moieties of the adsorbed HDTMA-Cl to be removed from
the water phase (Xu and Boyd, 1994a,b). At loading levels above the CEC, HDTMA
adsorption by hydrophobic bonding caused the development of positive charges on the
clay surface and ultimately dispersion (Xu and Boyd, 1994a,b). In the experiments
reported herein the soil was treated at an HDTMA-Cl loading of about 100% the CEC,
corresponding to the ﬁrst plateau of Figure 2. Therefore, the face-to-face association
should have been extensive because most of the accessible exchange sites were saturated
with HDTMA-Cl, with little or no signiﬁcant HDTMA-Cl adsorption by hydrophobic

bonding. This face-to-face association imposed a higher degree of orientation on the clay

31

fabric at a microscopic level. The measured macroscopic properties (volumetric
parameters, hydraulic conductivities, and effective-pore-size indices) for a HDTMA-Cl
loading of 100% the CEC were consistent with these microscopic observations

Drying the treated soil after the batch modiﬁcation procedures altered the clay
fabric such that the hydraulic conductivity of the treated-dry samples was always greater
than the treated-wet samples (Figure 4). Initially the treated-dry Oshtemo B soil had a
hydraulic conductivity one half order of magnitude greater than the treated-wet sample.
Figures 5 and 6 show that the mechanical response and loss of pore volume of the treated-
dry soil were indistinguishable from that of the treated-wet soil; however, the effective-
pore-size index was larger. This indicated that the differences in hydraulic conductivity
were due minimally to the total void volume that was available and essentially all due to
differences in the distribution of pore space. Figures 4 and 6 suggested that drying the
batch treated soil caused the clay aggregates to arrange themselves such that larger spaces
were able to conduct ﬂuid transport than when the treated soil remained water saturated.
There was a shift from a large number of smaller pores in the treated-wet samples to a
small number of larger pores in the treated—dry samples. This data also supported the
conjecture that the treated clay fabric existed predominately within the pore space created
by the granular skeleton as opposed to adhering to the soil minerals like a hydrophilic,
natural clay would. The data showed that only the void space distribution was affected
by the alteration of the treated clay fabric induced by drying (Figure 6), not the bulk
mechanical response to effective stress increases (Figure 5). The non-clay fraction of the

soil, i.e. sand and silt particles, were assumed to be unaffected by the ion exchange of

 

by
Di

CI

f2

32

HDTMA-Cl and therefore could be expected to be surrounded by ﬁlms of water. It is
hypothesized that the hydrophobic nature of the organo-modiﬁed clay limits its adherence
or interaction with the sand and silt; therefore, limiting the treated clay to preferentially
exist within the granular pore space until compression was signiﬁcant enough to mask
this effect.

Finally, a question can be raised regarding the effect of the permeant on the clay
fabric. The HPLC-grade water that was used for these experiments had a low electrical
conductance and hence a low ionic strength. The degree of clay ﬂocculation in an
untreated soil is dependent on the ionic strength of the solution ﬂowing through it
(Shackelford, 1994). Initially, after saturation, the inorganic salts that were precipitated
on the dry soil have diffused into solution to satisfy some equilibrium between the pore
water solution and the soil . There is some ionic strength associated with this
equilibrium. As the falling-head permeability tests were carried out, the pore space was
ﬂushed with a solution that had a lower ionic strength. The lower the ionic strength of
the solution, the more dispersion that occured during continued ﬂushing as the clay
structure attempted to reach a new chemical equilibrium. Back-to-back falling-head
permeability tests indicated a decreasing hydraulic conductivity with increasing time at a
constant stress state, which is consistent with a dispersion effect due to ﬂushing. An
analysis of the conductivity lost due to dispersion relative to that lost due to consolidation
showed that the permeant effect contributed to less than 10% of the overall measured

decrease in hydraulic conductivity for all of the experiments conducted.

IV. ENGINEERING SIGNIFICANCE

Superfund sites are predominately contaminated by organic chemicals. Of the ten
treatability groups of contaminants found in soils, nine are groups of organic chemicals.
The organic chemicals that possess high water solubilities, namely one- and two-carbon
chlorinated solvents (i.e., trichloroethylene, carbon tetrachloride) and the aromatic
hydrocarbon constituents of gasoline (i.e., benzene, toluene, xylene), become of special
interest due to their low retardation factors and hence high mobility in-situ. Organo—
modiﬁed soils become viable alternatives as remedial tools because they possess an
increased potential to contain such contaminants within an engineered treatment zone and
to act as barriers to further migration. The attractiveness of treated soil technologies is
further enhanced by the commercial availability of large quantities of cationic surfactant
(QACs) at reasonable costs (i.e., $0.50 to $1.00 per 1b.), the wide variety of QACs
available allowing for versatility in the production of organo-clays and adaptation to
differing site characteristics, and the wide usage of QACs in other applications (i.e.,
detergents, fabric softeners, antistatic sprays, swimming pool additives) suggesting
environmental acceptance (Boyd et al. , 1991).

Underground injection of cationic surfactants to increase the sorptive properties of
subsurface materials has been suggested as a potential technology for greatly reducing the
mobility of an organic contaminant plume (Burris and Antworth, 1992; Boyd et al. ,
1991). A constructed sorption zone (Figure 7) in itself does not provide a ﬁnal solution
for containment of the contaminant mass; however, the retardation of such constituents

33

530m Eat—cram £555 a S :o_to_c_ 2: .3 83528 ocoN Steam Bars 5 “o cognacognoz owe—=28 K 2:2...

 

34

 

 

 

 

 

 

 

 

 

35

within a deﬁned treatment zone suggests that a manageable remediation technique is
feasible when coupled with contaminant degradation mechanisms (Burris and Antworth,
1992). In-situ remediation, whether biotic or abiotic, can often be a slow process thus the
enhanced retardation of a migrating plume provides time for slow reactions to occur. The
advective transport of the dissolved contaminant to the sorption sites becomes important,
hence the hydraulic conductivity of an aquifer material before and after modiﬁcation is an
important parameter for assessing the feasibility of such a technology. System failures
due to hydraulic bypassing may result if the conductivity to the contaminated
groundwater were reduced signiﬁcantly. This could cause ﬂow across the sorptive zone's
face or around the zone entirely, allowing large fractions of the contaminant mass to
never contact the interior sorption sites of the zone. Insigniﬁcant retardation would result
and the integrity of the sorption zone would be lost.

Preliminary results from this investigation (Figure 4) indicated that batch
treatment of an aquifer type material with HDTMA reduced the hydraulic conductivity by
about 90% at low effective stresses. As the effective stress increased, the differences
between the untreated and the treated soil's hydraulic conductivity decreased. It was
hypothesized that the observed decrease in conductivity due to organo-modiﬁcation was
due, in part, to a reduction in the pore size caused by rearrangement of the clay
aggregates in the pore space. These results allowed some conjectures to be made with
regard to the feasibility of sorptive zones. First, the reduced hydraulic conductivity may
cause hydraulic bypassing at shallow depths (i.e., less than 4 tsf of overburden) for

aquifer material of a similar nature to Oshtemo B soil. Some kind of a modeling effort

36

would be necessary to evaluate if the magnitude of the reduction was signiﬁcant enough
to direct the streamlines completely around the sorptive zone in-situ. It may be possible
to compensate for the decrease in hydraulic conductivity by constructing a larger zone to
intercept the contaminant plume. Second, the convergence of the treated-wet and the
untreated-dry soil's hydraulic conductivity values at higher effective stresses suggested
that at sufficient depths or high external loads, in-situ treatment can be implemented
without producing decreases in the hydraulic conductivity. Third, the associated loss of
porosity (Figure 5) due to organo-modiﬁcation indicated that preferential ﬂow paths may
develop in-situ as a result of local settlement effects. This could promote faster routes
through the sorptive zone and again reduce the mass of contaminant delivered to the
interior sorptive sites.

It is important to note the limitations of the conjectures being made based on the
measured data of this investigation. These are preliminary observations of the batch
HDTMA-Cl treatment effects on a sandy loam material. The observed reductions in the
hydraulic conductivity and in the effective-pore-size index may, in part, be induced by
abrasive effects that occur during batch stirring procedures. In-situ applications such as
the introduction of cationic surfactant solutions by underground injection wells would not
involve such mixing and would involve more complex chemical dynamics. The
efﬁciency of the ion-exchange process and the soil fabric alteration by abrasive motions
of mixing come into question.

The abrasive motions induced by the stirring process used in this investigation

were more directly related to currently practiced in-situ and ex-situ soil mixing

37

techniques. With some adaptation, these systems could become useful as organo-
modiﬁcation tools. Soil mixing applications that could beneﬁt most directly fiom the
modiﬁed soil's increased sorption capacity and reduced hydraulic conductivity are
vertical and horizontal barrier technologies. The more commonly designed vertical
barrier systems are implemented using excavation-and-replacement methods that involve
ex-situ soil mixing or using jet grouting methods that accomplish the soil mixing in-situ.
The most common horizontal barrier systems are compacted clay liners such as those
used for hazardous and solid waste landﬁlls and for underground storage tanks (U STs).
Slurry trench cutoff walls are the most common vertical barrier technology
(Evans, 1991). Typically a well-graded soil is mixed ex-situ with a bentonite clay and
then placed in a trench to form a barrier to horizontal groundwater ﬂow. The bentonite
swells in the presence of water, plugging up the pore space and reducing the hydraulic
conductivity. This serves to redirect the groundwater in a region of contaminated soil,
preventing clean groundwater ﬁom entering the zone or any pumping, extraction, and
treatment systems. In the presence of contaminants the swelling of bentonite is a
reversible process (Evans, 1991). At sites where contaminant migration may destroy the
barrier, modiﬁed soil becomes an attractive candidate as an admixture material. Evidence
from this investigation indicated that a batch treated soil slurry (treated-wet samples) had
a reduced hydraulic conductivity to water ﬂow (Figure 4) and could thus contribute to the
plugging of the pore space. Additionally, the enhanced sorptive capabilities of a
modiﬁed clay would induce swelling in the presence of organic contaminants thus

maintaining the integrity of the barrier wall. It is suggested that organo-modiﬁed clays or

38

soils should be investigated as feasible components of slurry trench cutoff walls to
enhance their resistance to ﬂow in the presence of organic contaminants.

Jet grouting is an in-situ shallow and deep soil mixing technology. It involves the
injection of pumpable materials into a soil and is commonly used for chemical
stabilization processes in clays and for structural and water control processes in sands.
The schematic diagram presented in Figure 8 suggests the feasibility of in-situ injection
and mixing of soil with an abrasive action that is similar to that of the batch stirring used
in this investigation. The "cement" and "additives" in Figure 8 could be replaced by an
aqueous cationic surfactant solution and a natural soil to create a treated soil slurry, or the
aqueous surfactant solution alone could be used as the pumpable suspension. The
modiﬁed soil's hydraulic properties could be such that the region could act as a barrier or
as a sorptive zone depending on the classiﬁcation of the native soil and on the chemical
composition of the injected suspension. It is implied that jet grouting, with some
adaptation, could provide a means for in-situ construction of vertical barriers or sorption
zones.

The use of compacted clays to minimize advective transport is well established.
The most common uses of compacted natural clays are in the construction of hazardous
and municipal landﬁlls as shown in Figure 9. There is evidence suggesting that waste-
soil interactions can have a negative effect on barrier technologies, causing the compacted
clay fabric to ﬂocculate and allow routes of exposure (Shackelford, 1994). This is
reﬂected in the redundancy in the design of the barrier layers and the leachate collection

systems, along with the EPA's concern that both geomembranes and natural clay liners

53280 95.55 E. .833 a B cogs... ozoEogum a 9:6:

 

 

39

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

EoEom .3556: Co cameo 05 5 352:8”. 05 8305 839W 3:: 03.5. Bib a 2:9...

 

 

40

 

 

 

 

 

41

may be subject to degradation in the presence of organic contaminants (Alther et al. ,
1990). To reduce the uncertainty in long-term effectiveness it becomes necessary to
minimize the rates of contaminant migration by minimizing the hydraulic conductivity
and rate of diffusion, thus maximizing the retardation. Organo-modiﬁed clays possess
the properties to minimize the rate of organic contaminant transport. The additional
adsorptive capacity of the treated clays becomes a useful tool for the collection of organic
contaminants that migrate past the geomembrane in response to concentration induced
diffusive ﬂuxes or to advective ﬂuxes through liner defects. Conceptually a composite
barrier layer could be constructed that would include a modiﬁed clay component, or an
actual modiﬁed soil layer could be incorporated to reduce the redundancy in design.
Swelling is induced as the organic chemicals partition into the organic phase of the
modiﬁed clays which serves to further reduce hydraulic conductivity and increase the
time for chemical breakthrough. Furthermore, evidence from this investigation indicates
that reductions in the hydraulic conductivity and the bulk density of treated soil could be
accomplished with less compactive effort than that required for the untreated soil. The
model that is hypothesized suggests that the batch organo—modiﬁcation process oriented
the clay fabric in a more parallel array, thus a more compact assemblage is attained with
less compactive effort when compared to the unmodiﬁed soil. This suggests that a
compacted, modiﬁed clay liner could be constructed in the ﬁeld with less mechanical
manipulation, and hence in less time, than a compacted, natural clay liner. It is proposed

that organo-modiﬁed clays would be an effective attenuating component of a composite

liner system.

V. CONCLUSIONS

The batch organo-modiﬁcation of an aquifer type material with HDTMA-Cl
caused an initial reduction in the soil's hydraulic conductivity to water at low effective
stress states. The experimental results showed evidence that the loss of hydraulic
conductivity resulted from a loss of porosity and an alteration of the pore size distribution
at low effective stresses.

The batch treatment process also altered the compressibility properties such that
treated soils exhibited less loss of hydraulic conductivity as the effective stress was
increased. At higher effective stress states, the treated soil possessed a greater hydraulic
conductivity if it was packed dry and an equal hydraulic conductivity if it was packed wet
relative to the untreated soil that was packed dry.

The treated soil could not rearrange due to physical compression or conduct ﬂuid
ﬂow as signiﬁcantly as the untreated soil because its fabric had been altered. It was
speculated through the engineering response of the natural and the modiﬁed soils that
batch treatment processes changed the initially random clay fabric of the untreated soil to
a more oriented, parallel type arrangement.

Additionally, the collected data showed that the void space distribution of the
organo-modiﬁed soil was dependent on the state at which it was packed. The change in
effective-pore-size index with an unchanging porosity between the wet and the dry

packed treated samples indicated that the geometry within the pore space varied. The

42

43

measured behavior suggested that the treated clay aggregates existed predominately

within the pore space created by the non-clay fraction, i.e. the sand and silt grains.

APPENDIX A

 

APPENDIX A

PROCEDURAL DETAILS

The treated-wet samples were prepared using a ﬁxed-ring consolidometer that was
marked on the outside at the height corresponding to the top surface of the lower porous
stone (the base of the soil sample). This assembly was placed within the desiccater
system (Figure 3) without the consolidometer's base plug, stand-pipe, or upper porous
stone. Best results were obtained when vacuum grease was used around the
consolidometer O-ring to prevent leakage. HPLC-grade water was added to the 4000 m1
vacuum ﬂask until it was just below the vacuum spigot and to the desiccater until the
level was just above the elbow joint for the stand-pipe. This insured that the static water
level was above the top surface of the lower porous stone. The top of the desiccater was
set in place and the 3-way stopcocks were sealed from the atmosphere by closing all the
channels of stopcock no. 1 and closing the "to atmosphere" channel of stopcock no. 2. A
vacuum was applied to both the desiccater and the ﬂask for 1 hour. This allowed for
deairing of the water within the system and to vacuum saturate the base components and
lower porous stone of the ﬁxed-ring consolidometer. Magnetic stir bars were used for
agitation to promote faster deairing. The vacuum was collapsed on the desiccater only by
opening stopcock no. 2 such that the ﬂask was still isolated from atmospheric pressure.
The top of the desiccater was removed and the water within the sample ring was drained
by a siphon to a level equal to the outer marking on the consolidometer. The treated soil
slurry was added to the consolidometer as described previously.

44

45

Once the treated soil slurry was in place, stopcock no. 3 was adjusted so that the
channel to the ﬂask was closed, then stopcock no. 2 was scaled from the atmosphere by
closing the "to atmosphere" channel. This isolated the ﬂask from atmospheric pressure
and exposed only the desiccater to the vacuum. A vacuum was applied to the desiccater
for 30 mins. to reestablish the negative headspace pressure. Vacuum saturation
procedures then proceeded as described previously with the ﬂooding of the desiccator
assembly.

For the dry packed samples, HPLC-grade water was used to ﬁll the 4000 ml
reservoir ﬂask to a level just below the vacuum spigot and to ﬁll the desiccator to a level
ﬂush with the base plate. The soil sample and consolidometer were set inside the
desiccator. The top of the desiccator was put into place and all the stopcock spigots were
sealed from the atmosphere. A vacuum was applied to the system for 1 hour to deair all
of the HPLC-gradc water that would be used for saturation. Stir plates were used for
agitation to promote faster deairing. Stopcock no. 3 was adjusted to isolate the reservoir
ﬂask from the vacuum, but to continue application to the desiccator. Stopcock no. 1 was
opened to ﬂow from the ﬂask, allowing the deaired HPLC-grade water to ﬂood the
desiccator to a level just below the lip of the consolidometer's efﬂuent reservoir, at that
time stopcock no. 1 is rescaled. The vacuum was then maintained on the desiccator
overnight and the procedures continued as previously described.

The load frame had to be tared for the mass of the consolidometer and the soil
sample before beginning the loading sequence. The consolidometer assembly was set on

the load time pedestal and the low pressure regulator was increased until the pedestal

46

just began to rise. The pressure was then backed off a little to stabilize the pedestal in a
stationary position. The load dial was set in the "off' position and the pressure transducer
was zeroed. The load frame head was then lowered to just sit snug on the stainless-steel
ball bearing, being very careful not to apply excessive mass to the consolidometer and
induce some settling, and tightened into place. The displacement dial gage was tightened
into position such that the tip contacted the consolidometer and there was enough play
leﬂ in the stem to allow movement as the sample consolidated. The initial dial gage
reading was noted from the linear vertical displacement transducer (LVDT). The ﬁrst
load increment was dialed using the low pressure regulator (see Appendix B for
calibration data on each pressure transducer-load frame assembly that was used). At time

zero the load dial is set on "load" to apply each pressure increment.

APPENDIX B

APPENDIX B

LOAD FRAME CALIBRATION DATA

NOTE: The calibration data was obtained from the load frame manufacturer.

W Serial N0. =270
Model N0. = S-450

Table B1: Load Frame #1 Calibration Data

 

 

 

 

 

 

 

 

 

 

 

Pressure Transducer Load Cell

(psi) (lbs.)

0.80 10.00

1.20 20.00
2.00 40.00
4.80 100.00
9.50 200.00
18.80 400.00
37.30 800.00
56.00 = 1200.00 II

 

regression analysis: Y(lbs.) = 21 .5339284*X(psi) - 4.7530335
r2 = 0.999988

based on a 2.5 in. diameter soil sample:
Z(tsf) = 0.31585362*X(psi) - 0.069720584

This equation was used to calculate the required transducer pressure that would
correspond to the desired sample pressure.

47

48

W Serial N0. = 269
Model N0. = S-450

Table B2: Load Frame #2 Calibration Data

 

 

 

 

 

 

 

 

 

 

 

 

Pressure Transducer 1 Load Cell
(psi) (lbs.)
0.80 10.00
1.30 20.00
2.20 40.00
4.90 100.00
9.70 200.00
18.70 400.00
37.50 800.00
56.30 1200.00

 

 

 

 

regression analysis: Y(lbs.) = 21 .4663518*X(psi) - 6.334829
r2 = 0.99997

based on a 2.5 in. diameter soil sample:
Z(tsf) = 0.314862428*X(psi) - 0.0929175

This equation was used to calculate the required transducer pressure that would
correspond to the desired sample pressure.

APPENDIX C

APPENDIX C

SAMPLE SPREADSHEET OUTPUT FOR DATA REDUCTION
AND ERROR ANALYSIS

 

49

BIBLIOGRAPHY

BIBLIOGRAPHY

Acar, Y. B. and Ghosh, A, 1986, “Role of Activity In Hydraulic Conductivity of

Compacted Soils Permeated With Acetone, ” InternationaLSxmmsiumm

Enmanmentalﬁeatechnalamalalumez. Hsai- -Yang Fang, Eds Pennsylvania,
pp. 403-412.

Allred, Barry and Brown, Glenn 0., 1994, “Surfactant Induced Reductions in Soil
Hydraulic Conductivity,” 9W. Spring, pp-174-184-

Althcr, George K, Evans, Jeffrey C. and Pancoski, Stephen E.,1990, “No Feet of Clay,”
Mamas August

Boyd, S. A. J,a,ynes W. F. and Ross, B. S., 1991, “Immobilization of Organic
Contaminants by Organo-Clays: Application to Soil Restoration and Hazardous

Waste Containment ”QuartiLSuhstamesandﬁedimeneinﬂater. V 1, 1313131-
200.

Brown, Robert E., Jindal, Bipender S. and Bulzan, Joseph D., 1992, “A Critical Review
of the Effectiveness of Stabilization and Solidiﬁcation of Hazardous Organic

Wastes,” Hazardausﬂastes, pp- 43-60-

Burris, David R. and Antworth, Christopher P., 1992, “In-situ Modiﬁcation of an Aquifer
Material by a Cationic Surfactant to Enhance Retardation of Organic

ContmninantS”.Imunal_Qf_CamamlnantHltdralagx,V 10 PP 325- 337

Collins, K. and McGown, A., 1974, “The Form and Function of Microfabric Features in a
Variety of Natural Soils,” Qegtcchnigue, V. 24, No. 2, pp. 223-254.

Evans, Jeffery C.,1991,“Geotechnics of Hazardous Waste Control Systems,’ ’Emmdamn
Engineeﬁngﬂandhaok, Hsai-Yang Fang, Eds., Van Nostrand Reinhold, New
York, pp. 750-777.

Garcia-Bengochea, Ignacio and Lovell, C. W., 1981, “Correlativc Measurements of Pore

Size Distribution and Permeability in Soils,” Benneahﬂiujndﬂrmmdmten

50

 

51

W T. F. Zimmie and C- 0- Riggs, Eds,
American Society for Testing and Materials, Philadelphia, pp. 137-150.

Haug, Moir D. and Wong, Lionel C., 1992, “Impact of Molding Water Content on
Hydraulic Conductivity of Compacted Sand-Bentonite,” WW

m1, V. 29, No. 2, April.

Jaynes, W. F. and Boyd, S. A, 1991, “Clay Mineral Type and Organic Compound
Sorption by Hexadecyltrimethlyammonium-Exchanged Clays,’ 3211ng

SadeturtAmerigalaumal V 55 pp 43-48

Lambe, T. W. 1,958a, “The Structure of Compacted Clay,” MW
andEonndatiansDixisiarLASQE V 84 N0 8M2 pp 1654- “O 1654- 34

Lambe, T. W., 1958b, “The Engineering Behavior of Compacted Clays, 1931mm

WV 84, N0 3M2 PP 1655- “0
1655-35.

Lee, J. F ., Crum, J. R. and Boyd, S. A., 1989, “ Enhanced Retention of Organic
Contaminants by Soils Exchanged with Organic Cations,” W

andleghngmgy, V. 23, No. 11, pp. 1365-1372.

Leonards, G. A., 1962, “Engineering Properties of Soils,” Wm G. A.
Leonards, Eds., McGraw Hill Book Co., New York, pp. 66-240.

Olson, R. E., 1986, “State of the Art. Consolidation Testing,” ggnsgh’daﬁgn 91: $9115

W R N Yong and F C Townsend Eds»
American Society for Testing and Materials, Philadelphia, pp. 7- 70.

Shackelford, Charles D., 1994, “Waste-Soil Interactions that Alter Hydraulic
Conductivity,” .. , ‘ -..
W142, David E. Daniel and Stephen J. Trautwein, Eds,
American Society for Testing and Materials, Philadelphia, pp. 111168.

 

Srinivasan, Keeran R. and Fogler, Scott H., 1989, “ Use of Modiﬁed Clays for the
Removal and Disposal of Chlorinated Dioxins and Other Priority Pollutants From
Industrial Wastewaters,” Chemgsphere, V. 18, No. 1-6, pp. 333-342.

Xu, Shihe and Boyd, Stephen A., 1994a, “Cationic Surfactant Sorption to a Vermiculitic
Subsoil via Hydrophobic Bonding”, Manuscript.

Xu, Shihe and Boyd, Stephen A., 1994b, “Cation Exchange Chemistry of
Hexadecyltrimethyl-ammonium in Vermiculitic Subsoils,” Manuscript.

 

52

Yong, Raymond N. and Warkentin, Benno P., 1975, W
Elsevier Scientiﬁc Publishing Co., Amsterdam.

MICHIGAN STATE UNIV. LIBRARIES
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
31293014056372