Vous êtes sur la page 1sur 10

Hydraulic conductivity of natural clays permeated with simple liquid hydrocarbons

FEDERICO FERNANDEZ AND ROBERT M. QUIGLEY


Faculty of Engineering Science, The University of Western Ontario, London, Ont., Canada N6A 5B9
Received November 2, 1984
Accepted January 3 1 , 1985
The hydraulic conductivity, k, of clayey soils is strongly influenced by the physicochemical properties of permeating liquid
hydrocarbons. Tests on natural Samia soils mixed with pure liquids at a void ratio of 0.8 yielded k values that increased from 5 x
to 1 x l op4 cm/s as the dielectric constant of the permeant decreased from 80 to 2.
Sequential permeation of compacted, water-wet samples (k = lo-' cm/s) showed no changes in hydraulic conductivity when
permeated with water-insoluble hydrocarbons of low dielectric constant (benzene, cyclohexane, xylene). These hydrophobic
liquids were forced through microchannels or macropores and displaced less than 10% of the pore water from samples at a void
ratio of unity.
Permeation with water-soluble alcohols resulted in extensive removal of the pore water and up to 10-fold increases in k.
Subsequent permeation with liquid aromatics of very low dielectric constant resulted in 1000-fold increases in k with only 30% of
the pore space occupied by the aromatics. Association liquids such as alcohol that are mutually soluble in water and the aromatics
seem to be required to initiate huge increases in k over testing periods of short duration.
Key words: hydraulic conductivity, liquid hydrocarbons, clay barriers, dielectric constant
La conductivitt hydraulique, k, des sols argileux est fortement inAuencCe par les propriCtCs physico-chimiques d'hydrocarbures
liquides permCants. Des essais sur des sols naturels de la rCgion de Sarnia, mClangCs avec des liquides purs B un indice des
vides de 0,8, ont donnC des valeurs de k qui augmentaient de 5 X lo-' B 1 X cm/s lorsque la constante diClectrique du
liquide permiant dCcroissait de 80 B 2.
La permeation ~Cquentielle d'Cchantillons humidifiks j. l'eau et compactks (k = lo-' cm/s) n'a produit aucun changement de
conductivitC hydraulique lorsque cette permtation se faisait avec des hydrocarbures insolubles dans l'eau et B faible constante
ditlectrique (benzene, cyclohexane, xylkne). Ces liquides hydrophobes sont forces B travers des macropores ou des micro-
canaux d'tcoulement et ils entrainent moins de 10% de I'eau interstitielle des Cchantilions, B un indice des vides de 1.
La permCation avec des alcools solubles dans l'eau rCsulte en une Climination importante de I'eau interstitielle et en un
accroissement de k par un facteur pouvant aller jusqu'h 10. La permCation subsCquente par des solutions aromatiques B trts
faible constante dielectrique a rCsultC en une augmentation de k par un facteur de 1000, alors que seulement 30% du volume
des pores etait occupC par les solutions aromatiques. Des liquides d'association comme les alcools, qui sont mutuellement
solubles aussi bien dans l'eau que dans les solutions aromatiques, semblent &tre nCcessaires pour amorcer d'Cnormes
augmentations de k dans des essais de courte durCe.
Mots clPs: conductivitk hydraulique, hydrocarbures liquides, barrieres argileuses, constante dielectrique.
[Traduit par la revue]
Can. Geotech. J. 22. 205-214 (1985)
Introduction
The damaging impact of liquid hydrocarbons on the hydraulic
conductivity of clayey soils has been a subject of increasing
concern in recent years. In particular, a major problem is the
integrity of clayey barriers (either natural deposits or compacted
liners) directly below industrial and toxic waste impoundments.
Very large values of hydraulic conductivity, k, have been
measured on clayey soils in which liquid hydrocarbons seem to
dominate the fluid phase (Michaels and Lin 1954; Mesri and
Olson 197 1 ; Anderson and Brown 198 1 ; Anderson et al. 1982;
Brown et al. 1983). The hydraulic conductivity k may approach
lop4 cm/ s, a value more characteristic of fine sand than clay.
At the present time, the literature on the subject appears
contradictory (Green et al. 1981) and much clarification is
required. The purpose of this paper is to present the results of
high-gradient, "constant flow rate" hydraulic conductivity
measurements on a natural soil from Sarnia, Ontario. A wide
variety of fairly simple liquids with dielectric constants, E,
varying from 80 to 2 have been used as permeants. Special
emphasis is placed on the results of sequential permeation of
compacted, water-wet soils which would be characteristic of
most liners.
Throughout this paper, k is the measured hydraulic conduc-
tivity, which is a function of both the porous media and the
liquids flowing through it. Calculated values of intrinsic
permeability showed the same trends as the hydraulic conduc-
tivity because the effects of the dielectric constant on the clay
double layers completely swamped the effects of fluid density
and viscosity (Fernandez 1984).
Materials and methods
Materials
The natural soils used for this study were obtained from the
brown, oxidized, surface crust of the thick clay deposits at
Sarnia, Ontario. These clayey soils are extensively described by
Quigley and Ogunbadejo (1976), Ogunbadejo and Quigley
(1974), and Crooks and Quigley (1984). Briefly, the brown
surface soils consist of abundant illite, chlorite, quartz, feld-
spar, carbonate, and a variable amount of smectite (0-10%)
created by oxidation weathering of unstable iron chlorite
derived from underlying Devonian shales. The index charac-
teristics and mineralogical composition of the test soil are
presented in Table 1.
The test fluids were selected t o represent a range of dielectric
constants from 8 0 (polar water) t o 2 (nonpolar benzene).
Liquids of very l ow dielectric constant (benzene, xylene,
aniline, et c. ) are only very slightly soluble in water. The
alcohols (E = 20-35) are fairly polar and are mutually soluble
with both water and nonpolar liquid hydrocarbons. The
physicochemical properties of the test fluids are presented in
Table 2. By chance, the liquids selected for study were of
similar or lesser density than water. This should be rectified in
future studies.
C
a
n
.

G
e
o
t
e
c
h
.

J
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
n
r
c
r
e
s
e
a
r
c
h
p
r
e
s
s
.
c
o
m

b
y

S
A
V
A
N
N
A
H
R
I
V
N
A
T
L
A
B
B
F

o
n

0
5
/
1
7
/
1
3
F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.

206 CAN. GEOTECH. J. VOL. 22, 1985
TABLE 1. Description of test soil
Steel Tubing
Test soil:*
Clay size (<2 pm)
Silt (2-74 pm)
Sand (74 pm -+ 2 mm)
Liquid limit
Plastic limit
Specific gravity (relative density)
Maximum dry density, pd(max)i
Void ratio at pd(max)
Water content at pd(max)
Void ratio, all water-moulded test samples
Mineralogy of test soil:
Quartz and feldspar
Carbonate
Illite
Chlorite
Smectites
Cation exchange capacity (<74 pm)
Specific surface (<74 pm)
*Sizes greater than 2mm removed from bulk soil to obtain test soil (2%
removal).
?By Harvard Miniature compaction.
Methods
The hydraulic conductivity testing equipment (Fig. 1) gener-
ates a constant flow rate, q, through the test specimens and the
induced head drop across the specimen is used to calculate k
using Darcy's law. This procedure, extensively described by
Olsen (1966), is particularly useful for the volatile, toxic liquid
hydrocarbons which are sealed in the reservoir cylinders
(syringes). The single circuit shown for one permeameter is
actually expanded to eight circuits with the reservoir syringes
mounted in a triaxial frame. The upward displacement generates
constant flow that could be varied from 6 x l op6 to 1 x
lo-' mL/s. A dial gauge mounted on the compression frame
allows continuous monitoring of the flow rates. The pressure
head at the inlet end of the specimen was measured by pressure
transducer and the outlet end was kept close to atmospheric
pressure. The total head drop was essentially equal to the
pressure head drop at high pressure heads, the elevation head
correction becoming more important at lower inlet pressures.
A permeameter cell, illustrated in Fig. 2, consists of an
aluminum cylinder (A) 5.38 cm in internal diameter. Both ends
of the cylinder are machined to contain viton O-rings (B) sealing
the contact between the cylinder and the aluminum plates (C).
The fluid outlet (D) allows collection of the effluent permeant
for constant flow assessment or chemical analysis. The two
fittings in the upper plate are the fluid inlet (E) and a valve for
escape of air during filling of the fluid chamber (F). In some later
experiments, port E was also used as the pressure transducer
mount. A brass porous disk (G) approximately 3 mm thick and a
polyethylene filter (H) 1.5 mm thick are placed on top and below
the soil sample respectively. The rigid spacer (I), placed on top
of the porous disk (G), prevents swelling of the soil sample
during permeation. The assembled cell is held together by four
threaded and sleeved rods (J) fixed to the lower aluminum plate.
Filter paper (K) is normally placed between the soil specimens
and the filter disks (G and H).
Fluid effluents from the samples were collected in poly-
ethylene bottles sealed to the outlet of each cell to prevent
evaporation losses of the volatiles. Frequent removal of these
bottles for effluent analysis kept the outlet pressures more or less
I B 1
\
Scale (crn)
,\ To Pressure Transducer
Constant Rate Hydraulic Conductivity
Upward Displacement Cell
FIG. 1. Schematic diagram of constant flow permeameter.
0 5
u
Scale (crn)
FIG. 2. Schematic of one-permeameter cell.
at atmospheric pressure; however, new collection techniques
with continuous control of the outlet pressures are being
developed.
Two different soil placement procedures had to be used
depending on whether water or hydrocarbon was present in the
soil samples. Compaction of water-wet samples was normally
done in three layers -8 mm thick using a Harvard Miniature
tamper (foot pressure 1240 kPa) and 30 tamps/layer. For this
suite of experiments, the soil was always wet-of-optimum for
the compactive effort employed, the loading spring was rarely
fully stressed, and extensive kneading of the soil occurred, thus
destroying interped macropores and channels. Soils saturated
C
a
n
.

G
e
o
t
e
c
h
.

J
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
n
r
c
r
e
s
e
a
r
c
h
p
r
e
s
s
.
c
o
m

b
y

S
A
V
A
N
N
A
H
R
I
V
N
A
T
L
A
B
B
F

o
n

0
5
/
1
7
/
1
3
F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.

FERNANDEZ AND QUIGLEY
TABLE 2. Physical and chemical properties of test permeants at -20C
Formula Water Dipole
Structural weight solubility Density Viscosity moment Dielectric
Liquid formula (8) (g/L) (g/cc) (mPa. s) (Dl constant
Acetone
Aniline
Benzene H@E 78.11 0.7 0.8780 0.647 0.0 2.28
(0.8778) (0.7243)
(C6H6) H
H
H2
Cyclohexane
(c6H12)
Ethanol
Methanol
CH3-CH-CH3
60.09 cc
0.7830 2.430
Isopropanol 1 (0.7843) (2.9765) 1.69 18.3
OH
Water
(Hz01
"Values in parentheses measured in laboratory for comparison with published values
;Refers to values <0.3 g/L.
with pure hydrocarbon liquids (Fig. 3) were so flocculated that
they could be poured into the permeameter cells and compressed
to the required void ratio. No attempt was made to destroy the
interflocc voids.
After trimming or compression of the soil samples to a thick-
ness of 2 cm, the fluid chambers were filled with the desired
permeants and the cells assembled. The syringes, filled with the
same permeants, were placed in the compression machine and
connected to each circuit. Saturation of the steel tubes was
achieved by driving the plungers until all the air was expelled
from the system. Finally, the pressure transducers were con-
nected to the circuits and the tests were ready to be commenced.
Several techiques were used to obtain the chemical composi-
tion of the effluent samples obtained during testing and the pore
fluid contained in the soil sample after completion of testing.
For samples of water with immiscible organics such as benzene,
the two phases separated and could be measured volumetrically
in a small burette. For mixtures of alcohol and water, the
percentage of water was measured using an automatic water
analyzer (Aquatest IV). For fluid samples containing three
components such as water, alcohol, and benzene, a combination
of Aquatest and nuclear magnetic resonance (NMR) was used.
Discussion
The results of a typical, constant-temperature test, run with
water on water-compacted clay, are presented in Fig. 3.
Considerable lag time is required for the pressure head (and the
calculated k) to reach equilibrium after establishment of a given
flow rate. This slow compliance is related to saturation
problems of both the equipment/soil samples and expansion of
the system as a whole. To reduce these effects all fittings were
made of steel or brass and the lengths of tubing were reduced to a
minimum. The results presented in this paper represent equili-
brium conditions for all tests run.
Another problem was created by rather small changes in
temperature in our "constant" temperature laboratory. For
certain low-volume setups with impervious (10-'cm/s) clay, a
3C increase in temperature could cause increases in pressure
head of up to 5 m for some hydrocarbon liquids that have
coefficients of thermal expansion up to five times greater than
water. Since these thermally induced variations in pressure
dissipate primarily through the clay specimens, thermal prob-
lems become less severe the more pervious the clay. It is
believed thermal problems have been eliminated from the test
results presented in this paper.
C
a
n
.

G
e
o
t
e
c
h
.

J
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
n
r
c
r
e
s
e
a
r
c
h
p
r
e
s
s
.
c
o
m

b
y

S
A
V
A
N
N
A
H
R
I
V
N
A
T
L
A
B
B
F

o
n

0
5
/
1
7
/
1
3
F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.

208 CAN. GEOTECH. J. VOL. 22. 1985
L I I
Void Ratio =0.78:
Pore Volumes
FIG. 3. Compliance characteristics of hydraulic conductivity equip-
ment.
10-9
60-
(U
0.4
10-9 10-6 10-4
Hydraulic Conductivity ( c m/ s )
FIG. 4. Hydraulic conductivity vs. void ratio, all samples moulded
and permeated with the fluid indicated.
I I
I I
q1=2.95 q2= 5.92 q3= 8.90 Q,= 19.8
*
I - I - I -
-
Flow Rate, q ( x 1 0 - ~ ml / s )
Water
0 Benzene
+ Cyclohexane
x Aniline
* Acetone
Hydraulic conductivity (pure liquid systems)
In order to establish the range of hydraulic conductivities to
be expected, a known mass of completely dry soil (dried over
P2O5 for 24 h) was mixed in each permeability cell with the
same liquid to be used as the permeant. The surface was
flattened and the cell assembled for constant flow rate permea-
tion. By using a series of soil/liquid ratios a variety of void
ratios was obtained.
The results in Fig. 4 summarize the hydraulic conductivity -
void ratio relationships for the nine test liquids. 'Three regions
are clearly visible: a low k region for polar water of dielectric
BROWN SARNI A S OI L
1 ?, , , VOI D RATI O = 1 . 0 --
U Z W
p2gw
.JX=E
I J =_I 2
- o w
l \ l j z ? ; O W Z CLua 0 ; Z
Q < I <
8 W +
I I I I I I I I I
0 20 4 0 60 8 0 1 0 0
DI EL ECTRI C CONSTANT
FIG. 5. Hydraulic conductivity vs. dielectric constant for a void
ratio of unity (data from Fig. 4).
constant E = 80; an intermediate k region for the polar alcohols
and acetone (E = 20-35); and a high k region for the nonpolar
aromatics (E = 2). The aromatic aniline produced k values
somewhat lower than the other aromatics because of its higher
viscosity and dielectric constant (see Table 2). A summary plot
of these data for a void ratio of unity is presented in Fig. 5. It
may be concluded that for the Sarnia test soil, k increases from
-loF8 to - l op3 cm/s as e decreases from 80 to 2.
Discussion
The distribution of k values as a function of the moulding
fluidlpermeant appears to be directly related to the dielectric
constant of the liquid medium and is completely compatible
with the work of Mesri and Olson (1971), who calculated their k
values from consolidation tests. Consolidation test values by
Fernandez (1984) also showed the same trends in the k values
for the Sarnia soil.
The distribution of the electric potential in the double layer
(of cations and water) around negatively charged clay particles
may be described by Gouy-Chapman theory as follows:
where
IJJx = potential at distance x from the clay particle
IJJo = potential at surface of clay particle
K = q( 8. r r e 1 ni zi2)/(e/I'T)
E = dielectric constant
1 = Boltzmann's constant
T = absolute temperature (Kj
e = elementary charge (4.803 X lo-'' esu)
ni = number of ions (i) per unit volume of bulk pore fluid
zi = valence of cations (i)
If all chemical variables are held constant except E, the
potential IJJ may be calculated and plotted versus x. As shown in
Fig. 6, the double layer of a monovalent clay in a dilute water
system (E = 80) is calculated to be about 100 nm thick, decreas-
ing dramatically to less than 15 nm for e = 2. The latter soil
system (E = 2) is highly flocculated and, on mixing, agglo-
merates into sand-sized peds or floccs with large interped pores.
In order to visually demonstrate the different fabrics of the soils
in the three regions of Fig. 4, suspensions were prepared by
C
a
n
.

G
e
o
t
e
c
h
.

J
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
n
r
c
r
e
s
e
a
r
c
h
p
r
e
s
s
.
c
o
m

b
y

S
A
V
A
N
N
A
H
R
I
V
N
A
T
L
A
B
B
F

o
n

0
5
/
1
7
/
1
3
F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.

FERNANDEZ AND QUIGLEY 209
E = 8 0 (WATER, MONOVALENT,
DI L UT E)
E = 2 0 - 3 0 (ALCOHOLS)
E = 2 ( BENZENE)
0 5 0 1 0 0 ( 0 . 1 pin)
DOUBLE-LAYER THI CKNESS, x (nm)
FIG. 6. Electric potential, $, vs. distance from clay particle for
iarious values of dielectric constant, E.
adding 0.5 g of dry soil to each of three 30 mL flasks filled with
water, ethanol, and xylene. The soil particles were dispersed
with an ultrasonic probe and then a few drops of each suspension
were placed on glass slides. Photographs taken 3 min later are
shown in Fig. 7. It is visually obvious that the clay-water
system is dispersed relative to the alcohol and xylene systems. It
is further suggested that the clay-xylene system is more
flocculated than that of ethanol.
Figure 8 illustrates the macroscopic appearance of the top of
three test specimens, moulded with water, ethanol, and ben-
zene. ~ l t h o u ~ h all are at a void ratio of 0. 8, the ethanol and
benzene samples look more porous than the water-moulded
specimen. The electron photomicrographs presented in Fig. 9
are fractured vertical sections through the soil samples. The
benzene- and alcohol-moulded samples are highly flocculated
and seem to contain large interped macropores compared with
the relatively more dispersed water-moulded sample. Section-
ing of the wet soils by a knife blade smeared the water-wet clay
(upper right) but had little effect on the hydrocarbon flocculated
specimens.
On the basis of Gouy theory and the observed fabric of the
soils, it is concluded that for a given void ratio, the high k values
in Figs. 4 and 5 are directly related to low E values, which cause
double-layer contraction, during mixing and formation of floccs
or peds with macropores between them.
Hydraulic conductivity-sequential permeation
Two-stage tests
A series of soil samples was moulded with water and
compacted by a kneading process in the permeability cells at a
void ratio of unity. Such samples would simulate a normal,
water-wet, clay barrier. The samples were then permeated with
1 pore volume (-23 mL) of reference 0.01 N CaS04 solution.
At the end of this permeation (-2.3 days with i = 500) k was
constant at 1 X l op8 cm/s and the samples are believed to have
been close to 100% saturation. This was done to prevent "early
arrival" of hydrocarbons through unsaturated samples as noted
by Brown et al. (1983). Approximately 2 pore volumes of
water-insoluble benzene, xylene, and cyclohexane (E = 2) were
then forced through separate samples at a gradient of -500. As
shown in Fig. 10 for benzene, k remained constant at - l op8
cm/s for this phase of the testing (-4.2 days) even though the
effluent fluid was 100% benzene. Exactly the same results were
obtained for xylene and cyclohexane (Fernandez 1984).
Polar methanol (E = 33) was also passed through two
water-wet samples, yielding a gradually increasing k from
FIG. 7. Flocculation characteristics, suspensions of 0.5 g clay in
30mL of liquid: ( a) water: (b) ethanol; ( c ) xylene.
- l ops for water to - l op7 for methanol (Fig. I I). At the end of
1.8 pore volumes of methanol (-4.2 days) the effluent
consisted of 80% methanol plus water. Very similar results were
obtained with ethanol.
Discussion
The complete absence of an increase in k with low dielectric
aromatics was not anticipated, so the pore fluids in the test
samples were extracted by evaporation and condensation. The
water-immiscible liquids (benzene, xylene, and cyclohexane)
spontaneously separated into two distinct phases (water and
C
a
n
.

G
e
o
t
e
c
h
.

J
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
n
r
c
r
e
s
e
a
r
c
h
p
r
e
s
s
.
c
o
m

b
y

S
A
V
A
N
N
A
H
R
I
V
N
A
T
L
A
B
B
F

o
n

0
5
/
1
7
/
1
3
F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.

210 CAN. GEOTECH.
J. VOL. 22, 1985
Scale
' 2 c m
FIG. 8. Top of actual test specimens before constant flow permea-
tion with liquid indicated.
hydrocarbon), which could be measured volumetrically. In the
case of the benzene-permeated sample, only -8% by weight of
the pore fluid consisted of benzene, the rest being water. It is
also indicated in Fig. 10 that the effluent reached 50% benzene
concentration (c/co = 0.5) at only 0.28 pore volume. These two
factors indicate that the benzene passed through interconnected
macropores in the compacted clay samples, and only water in
these large voids was replaced during the flow time of -4.2
days. Exactly the same pattern of results was obtained for
xylene and cyclohexane.
In the case of the methanol-permeated samples, -94% of the
soil pore fluid after testing consisted of methanol, the remaining
6% being water (Fig. 1 1 ) . Also, the effluent methanol concentra-
tion reached 50% at 0.75 pore volume, indicating a more
uniform flow through the micro- and macro-pores.
The sequence of photomicrographs shown in Fig. 12 shows
the influence of a drop of ethanol on a clay-water suspension.
The dispersed clay-water suspension (Fig. 12a) is immediately
flocculated (Fig. 12b), the flocc size growing gradually with
time (Fig. 12c). This flocculation is a visual indication of
double-layer collapse due to reduction in the dielectric constant
and explains the gradual increase in k illustrated in Fig. 1 1.
The two photomicrographs of Fig. 13 illustrate the relative
affinity of the test clay for water and xylene. In Fig. 13a, a drop
of xylene was added to a clay-water suspension and was
immediately repelled to the periphery of the sample, with the
Scal e
FIG. 9. Scanning electron photomicrographs of vertical surfaces of
test samples: ( a) fractured, water-moulded clay; (b) cut and smeared,
water-moulded; ( c ) alcohol-moulded; (4 benzene-nloulded.
clay 100% retained in the aqueous phase. This type of
hydrophobic reaction between water and xylene is believed to
be responsible for the behaviour illustrated in Fig. 10 (little or no
increase in k with benzene, xylene, and cyclohexane permeation
through the clay-water samples).
Figure 136 dramatically illustrates the effect of adding water
to a clay-xylene suspension. The clay floccs in the xylene
migrate to the water phase (see arrow), forming a clay-water
suspension out of a clay-xylene suspension. These two figures
illustrate the extremely hydrophobic nature of pure aromatic
C
a
n
.

G
e
o
t
e
c
h
.

J
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
n
r
c
r
e
s
e
a
r
c
h
p
r
e
s
s
.
c
o
m

b
y

S
A
V
A
N
N
A
H
R
I
V
N
A
T
L
A
B
B
F

o
n

0
5
/
1
7
/
1
3
F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.

FERNANDEZ AND QUICLEY 21 1
WATER BENZENE
- I -
- -
'-I
I
BROWN SARNIA SOI L
VOID RATIO = 0. 98
1 PV = 22. 7 mL
q = 1. 14 x mL/ s
i - 500
T PORE FLUI D
I
c 92% WATER
3
-
WATER - , BENZENE j 8% BENZENE
W
PORE VOLUMES
FIG. 10. Hydraulic conductivity and effluent concentrations vs.
pore volumes of liquid passed through soil sample.
hydrocarbons with low dielectric constant to a pure clay-water
system and the preference of clay particles for the water phase.
Three-stage tests
A series of soil samples (moulded with water and compacted
by kneading at e = 1) was sequentially permeated with water,
then alcohol, then one of benzene, xylene, and cyclohexane.
The results for benzene are presented in Fig. 14. After 1.0 pore
volume of water permeation at i = 510, the reference k was
established at - lo-' cm/s (-2.2 days). After 3.0 pore volumes
of ethanol (-6.5 days), k had gradually increased to -7 x
10-'cm/s, ethanol had reached 100% concentration in the
effluent, and c/co = 0.5 had been reached at 0.68 pore volume,
suggesting some channelling through macropores. Approxi-
mately 4 pore volumes of benzene were then permeated through
the samples (-2.5 days). After passage of -2.5 pore volumes,
there was a sudden increase in k to values close to 10-%m/s.
At this point the effluent concentration was 100% benzene and
c/co = 0.5 had been reached at only 0.39 pore volume. Exactly
the same trends were obtained with xylene and cyclohexane
after ethanol permeation.
Discussion
Permeation by an association medium, mutually soluble in
water and aromatic hydrocarbons, appears to be a critical factor
in the dramatic increase in k during the short testing times
employed in our constant flow rate permeameter. The huge
increase in k resulted in a dramatic drop in h, at the inlet side of
the test specimens, and the gradient, i, at the end of the test was
only 25 compared with a starting value with water of 510 in spite
of increases in flow rates applied during testing (Fig. 14).
The test specimens were again evaporated to dryness and the
vapour condensed for analysis of the pore fluid. After ethanol
permeation, the pore fluid consisted of -35% water and -65%
ethanol. After subsequent benzene permeation, the pore fluid
WATER METHANOL
-I
I
BROYNSARNI ASOI L I
VOI D RATI O = 1. 02
1 PV = 22. 7 mL
I
,SOIL PORE FLUI D
-
I-
3 WATER METHANOL 1 - 6% WATER
?
I +. - 94% METHANOL
I
0 1 1
W 3
2 0 . 0
ii
1 .O 2. 0 3. 0 4.0
ii
w PORE VOLUMES
FIG. 1 1. Hydraulic conductivity and effluent concentrations vs.
pore volumes passed through soil sample.
consisted of -30% benzene, -35% ethanol, and -35% water
(Fig. 14). In other words, the benzene had replaced about 50%
of the ethanol in the test sample. Since the ethanol is completely
soluble in water it was probably present in the double layers
around the clay particles causing the initial "small" increase in k.
Subsequent benzene permeation probably replaced all of the
ethanol in the macropores and must have diffused into the
double layers at the expense of ethanol. Double-layer theory
predicts a decrease in double-layer size and by inference
predicts an increase in macropore volume and hydraulic
conductivity. In the case of the test soil, k increased by a factor
of about 10 000 to a value close to [hat for a pure soil-benzene
system.
Double-layer contraction and agglomeration of the soil peds
formed by initial compaction in water might cause a small
decrease in sample volume or even cracking. In this experi-
mentation, spacers prevented soil expansion but not shrinkage,
which amounted to 1-2% vertically for the three-stage permea-
tion tests. In order to check that an annular gap did not open
between the cells and samples (and thus cause the increase in k),
a series of reversed sequence permeation tests was run on
samples moulded with one of benzene, xylene, and cyclo-
hexane, followed sequentially by ethanol and water. The results
are presented in Fig. 15 and show exactly the opposite trends to
Fig. 14. The reference k with benzene is -3 X l op4 cm/s,
which decreases to about 1 x lo-' cm/s after passage of 2. 5
pore volumes of ethanol. Subsequent permeation with water
caused a dramatic decrease in k to -2 X lo-' cm/s.
The pore fluid composition in the test specimens is also
shown in Fig. 15. Ethanol was a very efficient remover of
benzene, only - 1% of it being left after passage of 2. 5 pore
volumes of ethanol. For ethanol, c/co = 0. 5 was reached
at 0.75 pore volume, suggesting flow through a relatively
uniform system of voids. Subsequently, water removed almost
C
a
n
.

G
e
o
t
e
c
h
.

J
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
n
r
c
r
e
s
e
a
r
c
h
p
r
e
s
s
.
c
o
m

b
y

S
A
V
A
N
N
A
H
R
I
V
N
A
T
L
A
B
B
F

o
n

0
5
/
1
7
/
1
3
F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.

CAN. GEOTECH. J. VOL. 22, 1985
FIG. 12. Flocculating influence of alcohol (ethanol) on a clay-water suspension: (a) dispersed clay-water system; (b) small floccs
after addition of a drop of ethanol; ( c ) mature floccs.
FIG. 13. Clay-water-xylene systems: ( a) xylene added to clay-
water suspension; (6) water added to clay-xylene suspension.
all of the ethanol, the final pore fluid consisting of -95% water
and 5% ethanol. From these results, it is tentatively suggested
that for the clay systems studied, a polar liquid is more efficient
at removing a less polar liquid than vice versa.
Finally, it may be seen that the final k in Fig. 15 (for nearly
pure water as the pore fluid) is about 20 times greater than that
for the water-compacted sample in Fig. 14. It is believed that
this reflects two very different soil fabrics, the benzene-moulded
sample being an open flocculated system of equidimensional
peds compared with a "dispersed" system of flat peds oriented
perpendicular to the flow by compression from the foot of the
spring-loaded kneading rod.
Conclusions
The hydraulic conductivity results presented herein have
been carried out in constant flow permeameters at gradients
varying from -500 to <25. These gradients are far in excess
of probable field situations, and macropores between soil peds
are believed to play a significant role in data interpretation.
Nevertheless, the trends summarized below are believed to be
generally valid even at low gradients, which would require
much longer testing times.
1. The Sarnia test soil is a Ca-Mg silty clay containing
approximately 10-15% smectite along with illite, chlorite, car-
bonate, quartz, and feldspar. At a moulding void ratio of 1 .O, it
has a reference hydraulic conductivity, k, of -lo-' cm/s with
0.01 N CaS04 as the permeant.
2. Pure liquid hydrocarbon - clay mixtures show a huge
range in k values from lo-' for water (E = 80) to for the
simple aromatics (E = 2). The higher values of k with lower E (at
e = 1) are attributed to macropores between flocculation peds
and double-layer contraction.
3. Permeation of water-wet, compacted samples by water-
insoluble aromatics yielded no increase in k and only minor
removal of pore water, which is believed to have been displaced
from macropores between the "compaction" peds. The hydro-
C
a
n
.

G
e
o
t
e
c
h
.

J
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
n
r
c
r
e
s
e
a
r
c
h
p
r
e
s
s
.
c
o
m

b
y

S
A
V
A
N
N
A
H
R
I
V
N
A
T
L
A
B
B
F

o
n

0
5
/
1
7
/
1
3
F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.

FERNANDEZ AND QUIGLEY 213
WATER ETHANOL BENZENE
z
l o-
o CELL # 8
VOID RATIO = 1. 0 I
PORE FLUID PORE FLUID PORE FLUID
0.0 1.0 2.0 3.0 4.0 5.0 6. 0 7.0 8. 0 9. 0 10.0
PORE VOLUMES
FIG. 14. Hydraulic conductivity and effluent concentrations vs.
pore volumes for sequential permeation by water, ethanol, and
benzene.
phobic aromatics studied apparently cannot easily penetrate the
double layers of the relatively hydrophyllic water-soil system.
4. Sequential permeation of water-wet samples by water-
soluble alcohols replaced much of the water and created 10-fold
increases in k in the testing periods employed.
5. Sequential permeation of water-wet samples by alcohol,
then simple aromatics (benzene, xylene, cyclohexane) yielded
1000-fold increases in k above the alcohol values (10 000 times
larger than the values for water). Mutually soluble "association"
liquids soluble in both water and the aromatic hydrocarbons
(i.e. the alcohols) appear to have been essential in the generation
of high k values in our test soil. This sequential replacement
sequence results in pronounced contraction of the clay double
layers and extensive enlargement of both micro- and macro-
pores in the soil.
6. Extensive chemical control of effluent and pore fluid
composition is critical to understanding the clay-water-hydro-
carbon flow system.
q3 = 6.27
FLOW RATE, q x niL/s
PORE FLUID PORE FLUID $'ORE FLUID
100% BENZENE - 1 % BENZENE 0% BENZENE
- 99% ETHANOL
I r
5% ETHANOL
95% WATER
BENZENE ETHANOL WATER
---+-- WATER
0.0 1.0 2.0 3.0 4.0 5.0 6. 0 7.0 8.0 9.0 10.0
PORE VOLUMES
FIG. 15. Hydraulic conductivity and effluent concentrations vs.
pore volumes for sequential permeation by benzene, ethanol, and
water.
Acknowledgements
The research presented has been financed by research funds to
the second author by the Natural Sciences and Engineering
Research Council of Canada. Special thanks are extended to
Dr. D. Hunter, Department of Chemistry, The University of
Western Ontario, for introducing us to the quantitative analysis
of liquids by Aquatest and nuclear magnetic resonance.
ANDERSON, D. C. , and BROWN, K. W. 1981. Organic leachate effects
on the permeability of clay liners. Proceedings of the 7th Annual
Research Symposium, Ft. Mitchell, KY, EPA-60019-81-002b,
pp. 119-130.
ANDERSON, D. C., BROWN, K. W. , and GREEN, J. W. 1982. Effect of
organic fluids on the permeability of clay soil liners. I n Land
disposal of hazardous waste. Edited by D. W . Shultz. Proceedings of
the 8th Annual Research Symposium, Ft. Mitchell, KY, EPA-6001
9-82-002, pp. 179-190.
BROWN, K. W. , GREEN, J. W. , and THOMAS, J. C. 1983. The
inHuence of selected organic liquids on the permeability of clay
liners. In Land disposal, incineration and treatment of hazardous
waste. Edited by D. W. Schultz. Proceedings of the 9th Annual
Research Symposium, Ft. Mitchell, KY, EPA-60019-83-018,
pp. 114-125.
CROOKS, V. E., and QUIGLEY, R. M. 1984. Saline leachate migration
C
a
n
.

G
e
o
t
e
c
h
.

J
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
n
r
c
r
e
s
e
a
r
c
h
p
r
e
s
s
.
c
o
m

b
y

S
A
V
A
N
N
A
H
R
I
V
N
A
T
L
A
B
B
F

o
n

0
5
/
1
7
/
1
3
F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.

214 CAN. GEOTECH. J. VOL. 22, 1985
through clay: a comparative laboratory and field investigation.
Canadian Geotechnical Journal, 21, pp. 349-362.
FERNANDEZ, F. 1984. The effect of organic hydrocarbon liquids on
the hydraulic conductivity of natural soils. M.E.Sc. thesis, The
University of Western Ontario, London, Ont., 166 p.
GREEN, W. J., LEE, G. F., and JONES, R. A. 1981. Clay-soils
permeability and hazardous waste storage. Journal of the Water
Pollution Control Federation, 53, pp. 1347-1354.
MESRI, G. , and OLSON, R. E. 1971. Mechanisms controlling the
permeability of clays. Clay and Clay Minerals, 19, pp. 151-158.
MICHAELS, A. S. , and LIN, C. S. 1954. Permeability of kaolinite.
Industrial and Engineering Chemistry, 46, pp. 1239-1246.
OGUNBADEJO, T. A, , and QUIGLEY, R. M. 1974. Compaction of
weathered clays near Sarnia, Ontario. Canadian Geotechnical
Journal, 11, pp. 642-647.
OLSEN, H. W. 1966. Darcy's law in saturated kaolinite. Water
Resources Research, 2(2), pp. 287-295.
QUIGLEY, R. M., and OGUNBADEJO, T. A. 1976. Till geology,
mineralogy and geotechnical behaviour, Sarnia, Ontario. I n Glacial
till, an interdisciplinary study. Edited by R. F. Legget. Royal
Society of Canada, Special Publication No. 12, pp. 336-345.
C
a
n
.

G
e
o
t
e
c
h
.

J
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
n
r
c
r
e
s
e
a
r
c
h
p
r
e
s
s
.
c
o
m

b
y

S
A
V
A
N
N
A
H
R
I
V
N
A
T
L
A
B
B
F

o
n

0
5
/
1
7
/
1
3
F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.

Vous aimerez peut-être aussi