Content uploaded by Ece Eseller-Bayat
Author content
All content in this area was uploaded by Ece Eseller-Bayat on Jun 29, 2015
Content may be subject to copyright.
Induced-Partial Saturation for Liquefaction Mitigation:
Experimental Investigation
M. K. Yegian, F.ASCE
1
; E. Eseller-Bayat, M.ASCE
2
; A. Alshawabkeh, M.ASCE
3
; and S. Ali
4
Abstract: The technical feasibility of a new liquefaction mitigation technique is investigated by introducing small amounts of gas/air into
liquefaction-susceptible soils. To explore this potential beneficial effect, partially saturated sand specimens were prepared and tested under
cyclic shear strain controlled tests. A special flexible liquefaction box was designed and manufactured that allowed preparation and testing
of large loose sand specimens under applied simple shear. Partial saturation was induced in various specimens by electrolysis and
alternatively by drainage-recharge of the pore water. Using a shaking table, cyclic shear strain controlled tests were performed on fully and
partially saturated loose sand specimens to determine the effect of partial saturation on the generation of excess pore water pressure. In
addition, the use of cross-well radar in detecting partial saturation was explored. Finally, a setup of a deep sand column was prepared and
the long-term sustainability of air entrapped in the voids of the sand was investigated. The results show that partial saturation can be
achieved by gas generation using electrolysis or by drainage-recharge of the pore water without influencing the void ratio of the specimen.
The results from cyclic tests demonstrate that a small reduction in the degree of saturation can prevent the occurrence of initial
liquefaction. In all of the partially saturated specimens tested, the maximum excess pore pressure ratios ranged between 0.43 and 0.72.
Also, the cross-well radar technique was able to detect changes in the degree of saturation when gases were generated in the specimen.
Finally, monitoring the degree of partial saturation in a 151 cm long sand column led to the observation that after 442 days, the original
degree of saturation of 82.9% increased only to 83.9%, indicating little tendency of diffusion of the entrapped air out of the specimen. The
research reported in this paper demonstrated that induced-partial saturation in sands can prevent liquefaction, and the technique holds
promise for use as a liquefaction mitigation measure.
DOI: 10.1061/共ASCE兲1090-0241共2007兲133:4共372兲
CE Database subject headings: Liquefaction; Saturation; Drainage; Shake table tests; Radar
.
Introduction
Liquefaction of loose saturated sands has been observed in almost
every past moderate to large earthquake, most recently in 1995
Hyogoken Nanbu 共Kobe兲 and in 1999 Adapazari 共Turkey兲. Dur-
ing an earthquake, saturated loose sands can lose shearing resis-
tance, associated with a sudden increase in pore water pressure,
often resulting in large lateral spreading, settlement, and founda-
tion and building damage.
Over the past three decades, intensive efforts have been made
by the geotechnical research community to understand the mecha-
nism of liquefaction, and to develop methods for evaluating liq-
uefaction potential at a site during a given seismic event. Recent
research 共Mitchell et al. 1995兲 has focused on the use of various
soil remediation measures to reduce or eliminate liquefaction po-
tential. Such measures include: Improved site condition through
densification, enhanced drainage, increased effective stress 共hence
strength兲, and change of soil fabric using grout. These remedial
measures for protecting structures from liquefaction-induced
damages are expensive and often are only applied in projects
involving large and important structures. The potential vulnerabil-
ity of existing structures founded on liquefiable soils continues to
be of major concern worldwide. Cost-effective liquefaction miti-
gation techniques, which can be easily and widely used for new
and, more important, for existing structures, are urgently needed.
Research results of many investigators have shown that a
small reduction in the degree of saturation of a fully saturated
sand can result in a significant increase in shear strength against
liquefaction. Martin et al. 共1975兲 explained that a 1% reduction in
the degree of saturation of a saturated sand specimen with 40%
porosity can lead to 28% reduction in the pore water pressure
increase per cycle. According to Yang et al. 共2003兲, a reduction in
saturation by 1% led to a reduction in the excess pore pressure
ratio from 0.6 to 0.15 under pure horizontal excitation. Chaney
共1978兲 and Yoshimi et al. 共1989兲 have shown that the resistance to
liquefaction was about two times that of fully saturated samples
when the degree of saturation reduced to 90%. Also, Xia and Hu
共1991兲 demonstrated that minute quantities of entrapped air can
significantly increase the liquefaction strength of a sand speci-
men. Their laboratory data demonstrated that a reduction in the
degree of saturation from 100 to 97.8% led to greater than 30%
increase in liquefaction strength. Fig. 1 shows their experimental
results plotted in terms of normalized cyclic shear stress
1
COE Distinguished Professor of Civil and Environmental
Engineering, Northeastern Univ., Boston, MA 02115.
2
Graduate Student, Dept. of Civil and Environmental Engineering,
Northeastern Univ., Boston, MA 02115.
3
Associate Professor, Dept. of Civil and Environmental Engineering,
Northeastern Univ., Boston, MA 02115.
4
Engineer, Parsons, Boston, MA 02115.
Note. Discussion open until September 1, 2007. Separate discussions
must be submitted for individual papers. To extend the closing date by
one month, a written request must be filed with the ASCE Managing
Editor. The manuscript for this paper was submitted for review and pos-
sible publication on June 3, 2005; approved on January 30, 2006. This
paper is part of the Journal of Geotechnical and Geoenvironmental
Engineering, Vol. 133, No. 4, April 1, 2007. ©ASCE, ISSN 1090-0241/
2007/4-372–380/$25.00.
372 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2007
共
d
/2
3c
⬘ 兲 as a function of the applied number of cycles, N. Ishi-
hara et al. 共2002兲 studied the liquefaction resistance of partially
saturated sands. They concluded that the resistance to liquefaction
increased with decreasing value of pore pressure parameter B.
When the B value dropped to zero with a degree of saturation of
S=90%, the cyclic strength became twice as much as that of the
fully saturated condition.
It is noted that the primary thrust of the previous researchers,
with the exception of Ishihara et al. 共2002兲, was to demonstrate
the importance of achieving 100% saturation in laboratory sand
specimen to avoid overestimating the strength of the specimen
against liquefaction. It is the intent of this paper to demonstrate
that the higher strength of partially saturated sands can be poten-
tially exploited to develop a new cost effective liquefaction miti-
gation measure. The research reported in this paper evaluated the
reduction in liquefaction potential as a result of introducing small
amounts of gas through electrolysis and air through drainage-
recharge of the pore water. The first phase of the research in-
volved the design and fabrication of a liquefaction box made of
Plexiglass that permitted the application of cyclic simple shear
strains with the use of a shaking table. Then, hydrogen and oxy-
gen gases were generated through electrolysis in the saturated
sand specimen and both saturated and partially saturated speci-
mens were tested using the shaking table facility. In the second
phase of the research, an alternative method was employed to
induce partial saturation by draining the initially saturated speci-
mens and then recharging them slowly from the top. The cyclic
tests were then performed on these specimens to again evaluate
the potential reduction in the liquefaction-induced excess pore
pressure. Finally, the detection of partial saturation using cross-
well radar was investigated, and the long-term sustainability of
partial saturation was monitored.
Liquefaction Box and Test Setup
A special flexible liquefaction box 共Fig. 2兲 was designed and con-
structed. A sand specimen in the box was subjected to a simple
shear deformation generated by a shaking table excitation at the
bottom of the box. The liquefaction box was designed by consid-
ering strength, flexibility, and workability, allowing 共1兲 the prepa-
ration of a sizable loose saturated sand specimen; 共2兲 the
introduction of gas/air into the specimen; 共3兲 the measurement
and detection of gas/air using cross-well radar; and 共4兲 the per-
formance of cyclic strain-controlled tests with pore water pressure
measurements to evaluate the effect of partial saturation on lique-
faction potential.
The liquefaction box provided the ability to subject a soil
specimen to controlled shear strains at different frequencies, in-
duced by the shaking table. Fig. 2 shows the dimensions and
details of the box. The box walls and base are made of Plexiglas
material. Two sides of the box are designed to rotate about their
bottom edges and the other two sides are fixed to the base of the
box, which in turn is fixed on the shaking table. The tops of the
two movable sides are joined together at one end of an aluminum
cross bar. The other end of the cross bar is fixed to a steel column
in front of the shaking table. A special construction joint sealant
共Sikaflex-15LM兲 is used as a flexible watertight joint between the
Plexiglas sides. The sealant is a strong adhesive and acts as a
flexible membrane allowing large deformations without rupture.
The box is sufficiently large to minimize boundary effects, and
to allow the preparation of representative samples as well as the
placement of two pore pressure transducers and two electrode
plates for performing electrolysis and generating gases in the sand
pores.
The sample preparation involved first bolting the entire box on
top of the shaking table and then attaching one end of the cross
bar to the tops of the two movable sides and the other end bolting
to the top of the fixed steel column located off the shaking table.
During specimen preparation, two miniature pore pressure trans-
ducers 共PDCR 81兲 were inserted, one close to the top and the
other close to the bottom of the specimen, as shown in Fig. 2共a兲.
The transducers were small enough 共0.6 cm
3
兲 not to affect the
behavior of the sand during the cyclic tests. A linear variable
Fig. 1. Effect of degree of saturation on liquefaction strength from
laboratory tests 共from Xia and Hu 1991, ASCE兲
Fig. 2. 共a兲 Liquefaction box and setup for testing of specimens
partially saturated through electrolysis; 共b兲 plan view of the
liquefaction box
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2007 / 373
displacement transformer 共LVDT兲 was mounted at the top of one
of the movable sides to monitor the controlled shear strains in-
duced in the specimens. A specimen was subjected to cyclic shear
strains by inducing cyclic displacements at the base of the box
through the shaking table. The motion of the fixed-base induced
rotation of the two movable walls with their tops fixed by the
crossbar. The pore pressure and LVDT responses were measured
using a PC-based data acquisition system. Fig. 3 shows a photo-
graph of the shaking table, the flexible liquefaction box, and a
typical test setup.
Cyclic Tests on Sand with Gas Generated through
Electrolysis
In this phase of the research, electrolysis, which is the ionization
of hydrogen and oxygen gases when a current is sent to electrodes
in water, is used to entrap gas molecules in saturated specimen.
Electrolysis has been used for dewatering fine-grained soils and
studied by geotechnical engineers 共Casagrande 1949; Esrig 1968兲.
Recently, electrolysis and electroosmosis have been further devel-
oped as effective and inexpensive methods for the enhancement
of in situ remediation of contaminated soils 共Acar and Alsha-
wabkeh 1993兲. Also, Thevanayagam and Jia 共2003兲 have explored
the use of electrokinetics to grout soils for liquefaction mitigation
applications.
Electrolysis was selected as an efficient application to induce
partial saturation since it introduces gas into the soil pores with-
out application of any pressure. Water electrolysis produces oxy-
gen and hydrogen gases at the anode and cathode, respectively, as
follows:
At the cathode: 4H
2
O+4e
−
→ 4OH
−
+2H
2
共1兲
At the anode: 2H
2
O−4e
−
→ 4H
+
+O
2
共2兲
While the gas quantities produced by electrolysis are not high
enough to produce any safety hazard 共especially H
2
gas兲, they are
significant enough to change the degree of saturation. Further,
electrolysis changes the pore fluid pH 共Acar and Alshawabkeh
1993兲. The experiments showed that this pH change would not
cause any significant change in the physics of saturation, pore
pressure development, and liquefaction.
Two rectangular meshes 共20 cm⫻ 33 cm兲 made of titanium
coated mixed metal oxides 共MMO for high electrolysis efficiency
and to prevent electrode corrosion兲 were used as electrodes. The
cathode was placed at the bottom of the box, where twice the
number of gas molecules 共hydrogen兲 was produced when com-
pared to gas molecules 共oxygen兲 produced at the anode, which
was hung at the top, as shown in Fig. 2共a兲. After saturated speci-
mens were prepared, various tests were performed to determine
the level of current and length of time needed to generate an
appreciable amount of gases in the sand specimens, as well as to
note the effect of gas generation on the density and average de-
gree of saturation 共Ali 2003兲. The process of electrolysis was
maintained until hydrogen bubbles were generated at the cathode
and migrated through the soil specimen toward the anode. The
large sized titanium meshes used as electrodes ensured the gen-
eration and vertical migration of gases to be well distributed
within the specimen. Visual inspection and probing of specimens
partially saturated through electrolysis confirmed that the process
indeed reduced the average degree of saturation of the specimens.
For comparison purposes, cyclic tests were performed on fully
saturated and partially saturated specimens developed through
electrolysis.
Saturated Specimen Preparation
Pluviation is a sample preparation method used to prepare soil
samples with different relative densities and gradation. Pluviation
can be performed in water 共wet pluviation兲 or in the dry 共dry
pluviation兲. According to the study done by Vaid and Negussey
共1998兲, pluviation in water is favored over dry pluviation in that it
results in initially fully saturated sand samples and replicate sand
samples can be produced more conveniently. Also, research per-
formed by Frost and Yang 共2002兲 in imaging of void distributions
in sand during shearing and variation in the initial void distribu-
tions due to sample preparation demonstrated that wet pluviated
specimen showed a higher number of larger voids than air pluvia-
tion and moist tamping. Therefore, fully saturated specimens were
prepared by raining dry sand very slowly from a specific height
共20 cm distance above the water level兲 into a predetermined
amount of water placed in the liquefaction box. The sand used
was Ottawa sand, which is uniform sand with rounded particle
shape.
Since saturation could not be controlled by measuring the B
value as is typically done in a triaxial cyclic test, the degree of
saturation was calculated using phase relationships. The volume
of the specimen was obtained by carefully measuring the average
height and using a height versus volume chart that was prepared
for the box. Knowing the weight of water placed in the box at the
start of the sample preparation, the mass of dry sand, and the total
volume of the saturated specimen, the volume of solids, volume
of voids, the average void ratio, and the average degree of satu-
ration were calculated using phase relations. The void ratio for the
loose Ottawa sand specimen was about 0.74. The maximum and
the minimum void ratios of the Ottawa sand were measured to be
0.5 and 0.8, respectively. Thus, the relative density of the loose
saturated sand specimen was estimated to be 20%. The saturated
density of the soil specimen was about 1.96 g / cm
3
.
Cyclic Tests on Fully Saturated and Gas-Generated
Specimens
For comparison purposes, two cyclic tests on fully saturated
specimens and two cyclic tests on partially saturated specimens
induced by electrolysis were performed.
Fig. 3. Typical test setup showing the flexible liquefaction box fixed
on the top of the shaking table
374 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2007
Cyclic Tests on Fully Saturated Specimens
Following the specimen preparation and testing procedures de-
scribed earlier, two fully saturated sand specimens were prepared
for cyclic testing. The degrees of saturation for the two specimens
that were prepared using wet pluviation were 99.4 and 99.5% and
the void ratio was 0.74 for both specimens.
Constant amplitude sinusoidal shear strains at 4 Hz were ap-
plied for about 15 s and pore pressure measurements from the
two transducers and displacement measurements from the LVDT
were recorded throughout the shaking. A 16 bit data acquisition
card was used to obtain the data at a sampling rate of 100 Hz. The
first 5 s of the cyclic shear strains induced in the specimen is
shown in Fig. 4 This strain history was used in all tests.
Fig. 5 shows the pore pressure build-up in the top and the
bottom transducers in one of the fully saturated specimens tested.
The initial pore water pressures of the two transducers matched
well with the hydrostatic pressures determined by the locations of
the bottom and top transducers 41 and 16 cm, respectively. Re-
sults from the two specimens were consistent and initial liquefac-
tion occurred within one or two cycles of excitation. The param-
eter, r
u
, defined as the ratio of maximum excess pore water
pressure and the initial effective stress, was calculated for both
bottom and top transducers. As shown in Table 1, r
u
for the two
fully saturated specimens was 1.0 and 1.05 from the bottom trans-
ducers, indicating initial liquefaction. The maximum excess pore
pressure ratios for the top transducer from the two tests were 0.81
and 0.93. The values of r
u
slightly less than one for the top trans-
ducer, even though both specimens liquefied, are probably be-
cause of the free draining surface of the specimen. In addition,
possible leakage may have occurred along the top transducer
cable that in these two tests were extended vertically out of the
specimens. In later tests, the transducer cables were all extended
laterally out of the specimens to avoid vertical diffusion of gas/air
and water pressure along the cables. However, this reduction of
pore pressure in the top transducer did not affect the results of the
study because the main purpose of the investigation was to com-
pare the pore pressure responses of saturated and partially satu-
rated specimens.
Cyclic Tests on Partially Saturated Specimens
In this phase of the research, partially saturated sand specimens
developed through electrolysis were tested under controlled cyclic
shear strains. After setting the box on the shaking table, the cath-
ode and anode meshes were placed within the box. The same
instrumentation setup 共pore pressure transducers and LVDT loca-
tions兲 used in the fully saturated tests was maintained with slight
differences in the locations of the transducers. Fig. 2共a兲 shows the
schematic of the setup for testing the specimens partially satu-
rated through electrolysis. After preparing a fully saturated speci-
men, hydrogen and oxygen gases were generated in the saturated
specimen for 11
Ⲑ
2 h for the first test and for 3 h for the second
test, at a current of 525 mA.
As the electrolysis process proceeded, a water layer slowly
formed on top of the originally fully saturated specimen and no
change in the original volume of the saturated sand was detected.
This was clear evidence that while gases were generated in the
soil specimen, the gases were indeed trapped in the specimen and
forced the water out of the specimen without changing the rela-
tive density of the sand. Hence, it was demonstrated that the
electrolysis process can generate, at least under laboratory condi-
tions, a controlled amount of gases without changing the void
ratio of the specimen. Since the volume of voids did not change,
the volume of gas entrapped within the soil was almost equal to
the volume of the water layer formed on top of the saturated
specimen surface. Based on this, the degree of saturation, the void
ratio, and the density of the partially saturated specimen were
calculated using phase relations. In both specimens, a degree of
saturation of about 96.3% was achieved. The amount of gas pro-
duced within the specimen was also calculated by Faraday’s law
of equivalence of mass and charge, and the equation of state for
gases
Gas collected, 共n兲 = moles of O
2
+ moles of H
2
=
冉
1
4
I
F
+
1
2
I
F
冊
⌬t
共3兲
Fig. 4. Cyclic shear strain history induced in the fully and partially
saturated sand specimens
Fig. 5. Comparison of excess pore water pressures generated in fully
saturated Ottawa sand specimen, and partially saturated Ottawa sand
specimen prepared through electrolysis 共S =96.3% 兲 共sample size:
21 cm⫻ 33 cm⫻ 42 cm兲
Table 1. Effect of Degree of Saturation on Maximum Excess Pore
Pressure Ratio, r
u
. Partial Saturation Induced through Electrolysis
Max excess pore pressure
ratio, r
u
Sand specimen
Degree of
saturation
S 共%兲
Bottom
transducer
Top
transducer
Fully saturated 共1兲 99.4 1.00 0.81
Partially saturated 共1兲 96.3 0.70 0.53
Fully saturated 共2兲 99.5 1.05 0.93
Partially saturated 共2兲 96.3 0.70 0.43
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2007 / 375
V =
nRT
P
共4兲
where I⫽current applied 共525 mA兲; ⌬t⫽time of current genera-
tion 共3 h for the second test兲; and F⫽Faraday’s constant
共96,485 C/ mol兲. P⫽pressure in Pa 共1 atm=101,325 Pa at 25°C兲;
V⫽volume of gas produced; n⫽number of moles, which is cal-
culated from Eq. 共3兲; R⫽ideal gas law constant
共8.314 Pa m
3
/mol兲; and T⫽absolute temperature 共K兲. Total
amount of gas produced in 3 h was 0.044 mol. The volume of gas
produced was calculated as 1,077 cm
3
using Eq. 共4兲, for 100%
efficiency and accordingly the degree of saturation would be
92.1%. Since the electrolysis efficiency is usually less than 100%
and some of the gas bubbles were able to find a way through the
wires and escape to the surface, the actual degree of saturation of
the specimen after electrolysis 共96.3%兲 was higher than that cal-
culated by Faraday’s law and the equation of state for gases.
Controlled cyclic shear strains were induced in the partially
saturated specimens again using the excitation history shown in
Fig. 4. The bottom and top pore pressure transducers were moni-
tored during shaking and typical test results for the first 5 s of the
excitation are plotted in Fig. 5. The maximum excess pore pres-
sure ratios for the two specimen tests were less than 0.7 for the
bottom transducer and less than 0.53 for the top transducer, as
shown in Table 1. The slightly smaller pore pressure ratio of the
top transducer is again because of the likely drainage occurring
during cyclic loading near the free surface of the specimen.
Fig. 5 presents a comparison of the pore pressure generation in
the fully and partially saturated specimens. The test results show
that a small reduction in the degree of saturation 共from 99.5 to
96.3%兲 led to significant reduction in the excess pore pressure.
The maximum excess pore pressure ratios in the fully and par-
tially saturated specimens using electrolysis are compared in
Table 1. The results indicate that a reduction in the degree of
saturation by about 3% prevented the onset of initial liquefaction.
Cyclic Tests on Sand with Air Entrapped by
Drainage-Recharge
In the preceding section, it was demonstrated that electrolysis can
induce partial saturation in sands leading to the prevention of
liquefaction and reduction in the potential generation of excess
pore water pressures. In this section, the results of investigations
conducted to assess the beneficial effect of induced partial satu-
ration using an alternative technique herein referred to as
drainage-recharge method is presented. In this method, after pre-
paring a fully saturated sand specimen, the pore water was slowly
drained from the bottom of the liquefaction box and then the
drained water was reintroduced from the top of the specimen at a
slow rate. Fig. 6 shows a schematic of the test setup. After rein-
troducing all the drained water, a significant amount of water
remained above the surface of the sand specimen indicating en-
trapment of air during recharge. The degree of saturation of the
specimen was calculated using the volume of the surface water as
a measure of the volume of the entrapped air.
During the drainage-recharge tests, the void ratio of the two
specimens tested remained at 0.74. The degrees of saturation
achieved were 86.2 and 86.5%. Visual evidence through the
Plexiglas sides also confirmed the presence of almost uniformly
distributed small bubbles of air trapped in the sand specimen.
Similar to the gas-generated specimen tests, cyclic strain-
controlled tests were performed on both the saturated and air-
entrapped specimens to investigate the effectiveness of the
drainage-recharge method in reducing the pore pressure build-up
during shaking.
Cyclic Tests on Fully Saturated and Air-Entrapped
Specimens
One cyclic test on a fully saturated specimen and two cyclic tests
on air-entrapped specimens were carried out using the shaking
table facility.
Cyclic Test on Fully Saturated Specimen
A fully saturated Ottawa sand specimen was prepared in the liq-
uefaction box following the wet pluviation procedure described
earlier. The degree of saturation of the specimen was 99.7%.
The specimen was then subjected to the sinusoidal shear strain
history shown in Fig. 4. Within one or two cycles, the specimens
liquefied, and the maximum excess pore pressure ratios were
computed for the bottom and top transducers as 1.0 and 1.04,
respectively, as shown in Table 2. The pore pressure response
plots for the first 5 s of the excitation are shown in Fig. 7. The
results are consistent with the fully saturated specimens described
earlier, indicating reproducibility of test results.
Cyclic Tests on Air-Entrapped Specimens
Two partially saturated specimens were prepared using the
drainage-recharge procedure resulting with degrees of saturation
of 86.2 and 86.5%. Applying the cyclic shear strains shown in
Fig. 4, pore pressures in both bottom and top transducers were
monitored during the tests.
Fig. 7 shows a typical plot of the measured pore pressures
during the cyclic strain application. The maximum excess pore
pressure ratios computed from the two specimens tested ranged
between 0.63 and 0.72 and are summarized in Table 2. Thus,
air-entrapped specimens never liquefied, although significant ex-
cess pore pressures nevertheless developed because of the very
low relative density of the sand 共20%兲 and the high amplitude of
the applied shear strains 共0.2%兲.
Fig. 6. Test setup for inducing partial saturation using drainage-
recharge method. The drawing shows a cross section perpendicular to
the direction of shaking.
376 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2007
In summary, the cyclic test results on the fully and partially
saturated specimens, prepared using the drainage-recharge
method, are compared in Fig. 7. As shown in Fig. 7, significant
reductions in the excess pore pressures were observed in the air-
entrapped 共partially saturated兲 specimen compared with the fully
saturated test results. The decrease in the degree of saturation
from 99.7 to about 86% using the drainage-recharge method re-
duced the maximum excess pore pressure ratio from 1 to approxi-
mately 0.65. Settlements of the fully and partially saturated
specimens were recorded after 15 s 共60 cycles兲 of strain applica-
tion. Table 2 shows a summary of the settlements and axial
strains. The results show that the partially saturated specimen
compared to fully saturated specimen experienced less than half
the axial strain 共5.1% for the fully saturated and typically 2.1%
for the partially saturated specimens兲.
The test results demonstrated that the drainage-recharge
method can induce partial saturation without change in the void
ratio of the sand. Further, the results confirm the earlier conclu-
sion made from the tests on partially saturated sands prepared by
electrolysis, that a small reduction in the degree of saturation of a
loose sand can prevent liquefaction and reduce the excess pore
pressures and settlement of the sand.
Effect of Sand Type
The test results presented earlier were on specimens of Ottawa
sand that have round shaped particles and uniform gradation. To
confirm the applicability of the results to a natural sand that has
more angular shape and a well graded distribution of particles,
selected tests were performed on saturated and partially saturated
specimens. Fig. 8 shows a comparison of the gradation curves of
the Ottawa and the natural sands tested.
The partially saturated specimen of the natural sand was pre-
pared again using the drainage-recharge method. The void ratios
of the fully and partially saturated specimens were the same, 0.75.
The induced cyclic shear strains were again as in the other tests.
Fig. 9 shows a comparison of the excess pore pressures generated
in the fully and partially saturated specimens. Again, significant
reductions in the excess pore pressures were observed due to
induced partial saturation. Whereas the fully saturated specimen
under the applied shear strains liquefied, the excess pore pressures
in the partially saturated specimen never reached the initial effec-
tive stress. The maximum excess pore pressure ratios 共r
u
兲 in the
partially saturated specimens were less than 0.66, as shown in
Table 3. Induced partial saturation in the natural sand also re-
duced the axial strain from 4.7% for the fully saturated to 1.7%
for the partially saturated.
In summary, the tests demonstrated that partial saturation in
natural sands can prevent initial liquefaction as was observed in
the Ottawa sand specimens. Based on the tests, it appears that the
benefits of partial saturation with respect to liquefaction strength
and settlement are slightly better when the sand is angular than
well graded.
Table 2. Effect of Partial Saturation Induced through Drainage-Recharge on Liquefaction-Induced Pore Pressure and Settlement
Max excess pore pressure ratio, r
u
Sand specimen
Degree of
saturation
S 共%兲
Bottom
transducer
Top
transducer
Settlement
共cm兲
Axial strain
共%兲
Fully saturated 99.7 1.00 1.04 1.71 5.1
Partially saturated 共1兲 86.2 0.72 0.63 0.82 2.4
Partially saturated 共2兲 86.5 0.68 0.66 0.65 1.9
Fig. 7. Effect of entrapped air on the excess pore water pressure
generation in the partially saturated Ottawa sand specimen
共S = 86.2% 兲, prepared by drainage-recharge during shear strain
controlled cyclic test 共sample size: 21 cm⫻ 33 cm⫻ 34 cm兲
Fig. 8. Gradation curves of the Ottawa and natural sands tested
Fig. 9. Effect of entrapped air on the excess pore water pressure
generation in the partially saturated natural sand specimen
共S = 84.2% 兲, prepared by drainage-recharge, during shear
strain-controlled cyclic test 共sample size: 21 cm⫻ 33 cm⫻ 34 cm兲
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2007 / 377
Gas/Air Detection and Air Diffusion Tests
When partial saturation was induced by both procedures—gas
generation and drainage-recharge, free water layer was formed on
the soil surface. The free water volume produced on top of the
soil surface is estimated to be equal to the gas/air volume gener-
ated in the voids. This volume of generated water at the specimen
surface was evidence of the gas/air presence within the soil speci-
men. However a more scientific approach for detection of pres-
ence of gas/air bubbles in a sand specimen was explored as a
possible field technique. Therefore, the next phase of the research
involved the potential use of cross-well radar 共CWR兲 in detecting
the presence of entrapped gas/air and its rate of long-term
diffusion.
Gas Detection Test by Cross-Well Radar
Radar has been used 共Trop et al. 1980; Lange 1983; Knoll and
Clement 1999兲 in determining the water content of saturated soils,
and in identifying large subsurface voids. In this research, a pre-
liminary effort was made to determine if the application of a
nondestructive test, such as radar, holds promise for the charac-
terization of a three-phase system consisting of sand, water, and
gas. Cross-well radar is an electromagnetic 共EM兲 geophysical
method for high resolution detection, imaging, and mapping of
soils. A typical CWR system has two main components: transmit-
ter and receiver. The transmitter radiates a short EM pulse into the
soil, which is refracted, diffracted, and reflected primarily as it
meets any contrast in dielectric permittivity and electric conduc-
tivity. These contrasts are because of the existence of different
materials in the media.
Exploratory tests were performed in the SoilBED Laboratory
of the Center for Subsurface Sensing and Imaging Systems
共CenSSIS兲, at Northeastern University using the experimental
setup shown in Fig. 10. A partially saturated specimen was pre-
pared using the electrolysis method described earlier. The trans-
mitter and receiver antennas were placed within the middepth of
the specimen as shown in Fig. 10.
The degree of saturation in the specimen prior to electrolysis
was 99.6%. Initially, transmission measurements were taken for a
fully saturated specimen, and the response data were reduced in a
vector network analyzer. Transmission measurements are the re-
sponses of the media in dB, which is defined as dB
=20 log共A
r
/A
t
兲, where A
r
and A
t
⫽amplitudes of the received and
transmitted signals. Measurements were made over a frequency
range of 0.4– 2.2 GHz.
Fig. 11 shows the transmission measurement data for a fully
saturated specimen, which is the response of the background field,
i.e., soil and water without gases. Next, gas generation was initi-
ated in the same specimen, and the response data was recorded
every half an hour. After 40 h of gas generation through electroly-
sis, free water of an average depth of 1.47 cm was observed on
the soil surface and the degree of saturation of the specimen was
calculated as 91.5%. Fig. 11 also shows the transmission mea-
surements obtained from the partially saturated specimen tested
after 40 h of electrolysis. Included in Fig. 11 are the measure-
ments on the same specimen 24 h after the gas generation was
terminated. The pattern of the variation of each response is gov-
Fig. 10. Experimental test setup for gas/air detection test by
cross-well radar
Fig. 11. Comparison of transmission response dB for 共1兲 a fully
saturated specimen; 共2兲 a partially saturated specimen prepared
through 40 h of electrolysis; and 共3兲 the partially saturated specimen
24 h after termination of gas generation
Table 3. Results of Cyclic Tests on Fully and Partially Saturated Natural Sand Specimens
Max excess pore pressure ratio,
r
u
Sand specimen
Degree of
saturation
S 共%兲
Bottom
transducer
Top
transducer
Settlement
共cm兲
Axial strain
共%兲
Fully saturated 99.1 0.97 0.98 1.57 4.7
Partially saturated
a
84.2 0.62 0.66 0.58 1.7
a
Partial saturation by drainage-recharge.
378 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2007
erned by the transmission and reflection characteristics of the an-
tennas, Farid et al. 共2006兲. The resonant frequency of both anten-
nas was 1.1 GHz.
A comparison of the transmission responses shown in Fig. 11
indicates that the received signal intensity changed appreciably as
gases were generated in the specimen. Further, 24 h after halting
the gas generation, there was little change in the transmission
measurements, thus indicating that there was little if any loss of
gas shortly after its generation. Even after 1 week, entrapped
gases could be seen in the sand through the Plexiglas walls of the
box.
The fact that there are differences between the responses of the
partially and fully saturated specimens suggests that cross-well
radar holds potential promise as a field technique for detection of
presence of gas in soils and quantification of degree of satura-
tion. Further research is needed to explore the technical feasibility
of using cross-well radar in large sand deposits under field
conditions.
Long-Term Air Diffusion Test
The cross-well radar test confirmed the short-term sustainability
of induced partial saturation in sands. A further test was con-
ducted to investigate whether or not air bubbles would remain
entrapped for a long time or have the tendency to quickly diffuse
out of liquefaction susceptible sands. All the earlier tests were
performed on samples 34–42 cm deep. In reality, in a deep soil
layer the water pressure will be higher and may force the air out
of the voids. On the other hand, one can argue that it would be
more difficult for the air molecules to find a path and escape
through a deep soil layer. To evaluate the potential long-term
tendency of diffusion of air from a thick soil layer, the degree of
saturation of a 151 cm column of partially saturated sand is being
monitored. Fig. 12共a兲 shows the test setup used. A 184 cm plastic
tube with an outer diameter of 10.12 cm 共4in.兲 was rigidly fixed
to a concrete column in the basement of the engineering building
to minimize the effect of ambient vibrations. A 151 cm column of
loose fully saturated Ottawa sand specimen was prepared in the
tube again by the wet pluviation method. The void ratio and the
degree of saturation of the specimen were calculated as 0.80 and
96.7%, respectively. Partial saturation was induced in the speci-
men using the drainage-recharge method.
From daily measurements of the volume of water above the
sand, and the sand height, the degree of saturation of the speci-
men was computed using phase relations. Fig. 12共b兲 shows the
measured data to date. The results indicate that the initial degree
of saturation of 82.1% only slightly increased to 83.9%, after 442
days of monitoring. It is noted that this small increase in the
degree of saturation was recorded within the first few days after
partial saturation was induced. Visual observations showed rear-
rangement of air bubbles in isolated regions within the specimen
until equilibrium was achieved. The long-term monitoring of sus-
tainability of induced partial saturation in a deep 共151 cm兲 sand
specimen led to the conclusion that under hydrostatic conditions,
small well-distributed air bubbles can remain trapped for a long
time.
Summary and Conclusions
Using induced-partial saturation 共IPS兲 in loose sands as a measure
for liquefaction mitigation was investigated experimentally. A
flexible liquefaction box that permitted the application of cyclic
simple shear strains in large loose sand specimens using a shaking
table was designed and manufactured. This new box eliminates
the limitations associated with the fixed walls that are typical for
conventional boxes used for liquefaction or other types of soil-
structure tests using a shaking table.
Fully saturated loose sand specimens were prepared in the
flexible box using the wet pluviation method. The relative density
of the sand was about 20%. Using the process of electrolysis,
oxygen and hydrogen gases were uniformly generated in satu-
rated sand specimens, resulting in a degree of saturation of
96.3%. Cyclic shear strains induced in the fully saturated sands
yielded maximum excess pore pressure ratios close to unity, indi-
cating initial liquefaction under about 2 cycles of shear strain with
a frequency of 4 Hz, and an amplitude of 0.2%. Cyclic tests on
sand specimens partially saturated through electrolysis yielded
maximum excess pore pressure ratios significantly smaller than
unity, thus confirming visual observations that initial liquefaction
was never achieved in these specimens. These tests indicate the
potential beneficial effect of a small reduction in the degree of
saturation in sands on liquefaction potential.
An alternative technique to induce partial saturation, referred
to as drainage-recharge method, was also employed, and its effect
on liquefaction potential was investigated. In this method, partial
saturation was achieved by slowly draining water from a speci-
Fig. 12. 共a兲 The test setup used to investigate the long-term sustainability of air bubbles in a partially saturated sand column and 共b兲 long-term
monitoring of the degree of saturation
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2007 / 379
men, and then slowly reintroducing the drained water to the speci-
men from its top. The degree of saturation of the specimens pre-
pared in this manner was about 86%. Cyclic simple shear tests on
these specimens resulted in maximum excess pore pressure ratios
smaller than unity, again confirming visual observations that the
partially saturated specimens did not liquefy, although consider-
able excess pore pressures were still generated under the large
amplitude of shear strains applied 共0.2%兲 to the loose sand with a
relative density of about 20%. The axial strains at the end of 60
cycles of excitation in the partially saturated specimens were sig-
nificantly smaller than in the fully saturated specimens. The re-
sults obtained from these tests confirm earlier conclusions arrived
at from testing specimens subjected to electrolysis that induced-
partial saturation can prevent liquefaction and lead to a reduction
in the excess pore water pressures.
The cross-well radar technique was employed to explore its
potential applicability in detecting partial saturation in sands.
Tests on both fully and partially saturated sand specimens showed
that received signal intensity obtained from a partially saturated
specimen deviated significantly from the signals obtained from a
fully saturated sand specimen. Thus, the cross-well radar tech-
nique holds promise as a potential field method for detecting and
maybe quantifying degree of saturation of sands.
A 151 cm column of loose sand was prepared to study the
potential long-term diffusion of entrapped air. Partial saturation
was induced in the sand column using the drainage-recharge
method. After 442 days, the degree of saturation of the sand col-
umn only slightly increased from about 82.9 to 83.9%.
The experimental results reported in this paper demonstrated
that IPS holds promise as a liquefaction mitigation measure. To
advance this concept for eventual use in field applications, further
research is required on: The behavior of a sand-water-air mixture
under varying field conditions and subjected to seismic excitation;
long-term diffusion of air under large overburden and small hy-
draulic gradient; and development of cost-effective field methods
for inducing and verifying partial saturation in liquefaction sus-
ceptible sands.
Acknowledgments
This research was funded by the National Science Foundation
through the Geoenvironmental Engineering and Geohazard Miti-
gation Program, under Grant Nos. CMS-0234365 and CMS-
0509894. The support of NSF and Program Director Dr. Richard
J. Fragaszy is greatly appreciated. The writers thank Professor
Carey Rappaport and Dr. Arvin M. Farid for their help in the use
of the cross-well radar. Special appreciation is expressed to Mr.
Richard Holt from NDT Corporation for his valuable insights into
the field behavior of partially saturated sands.
References
Acar, Y. B., and Alshawabkeh, A. N. 共1993兲. “Principles of electrokinetic
remediation.” Environ. Sci. Technol.,27共13兲, 2639–2647.
Ali, S. 共2003兲. “Experimental setup and procedures for the preparation
and cyclic testing of air entrapped sand.” Master’s Rep. Northeastern
Univ., Boston.
Casagrande, L. 共1949兲. “Electro-osmosis in soils.” Geotechnique,1共3兲,
159–177.
Chaney, R. 共1978兲. “Saturation effects on the cyclic strength of sands.”
Proc., ASCE Special Conf. on Earthquake Engineering and Soil Dy-
namics, New York, 342–358.
Esrig, M. I. 共1968兲. “Pore pressures, consolidation, and electrokinetics.”
J. Soil Mech. and Found. Div.,94共4兲, 899–921.
Farid, M., Alshawabkeh, A. N., and Rappaport, C. M. 共2006兲. “Calibra-
tion and validation of a laboratory experimental setup for CWR in
sand.” Geotech. Test. J.,29共2兲, 158–167.
Frost, D., and Yang, C.-T. 共2002兲. “Image analysis of void distributions in
sand during shearing.” U.S.-Japan Seminar on Seismic Disaster Miti-
gation in Urban Area by Geotechnical Engineering, Anchorage,
Alaska.
Ishihara, K., Tsukamoto, Y., Nakazawa, H., Kamada, K., and Huang, Y.
共2002兲. “Resistance of partly saturated sand to liquefaction with ref-
erence to longitudinal and shear wave velocities.” Soils Found.,
42共6兲, 93–105.
Knoll, M. D., and Clement, W. P. 共1999兲. “Vertical radar profiling to
determine dielectric constant, water content, and porosity values at
well locations.” Proc., Symp. on the Application of Geophysics to
Engineering and Environmental Problems, Oakland, Calif., 821.
Lange, J. N. 共1983兲
. “Microwave properties of saturated reservoirs.” Geo-
physics,48共3兲, 367–375.
Martin, G. R., Seed, H. B., and Finn, D. L. 共1975兲. “Fundamentals of
liquefaction under cyclic loading.” J. Geotech. Engrg. Div., 101共5兲,
423–438.
Mitchell, J. K., Baxter, C. D. P., and Munson, T. C. 共1995兲. “Performance
of improved ground during earthquakes.” Soil improvement for earth-
quake hazard mitigation, ASCE Geotechnical Special Publication No.
49, 423–438.
Thevanayagam, S., and Jia, W. 共2003兲. “Electro-osmotic grouting for liq-
uefaction mitigation in silty soils.” ASCE Conf. Proc. Paper, Grouting
and Ground Treatment, New Orleans, 1507–1517.
Topp, G. C., Davis, J. L., and Annan, A. P. 共1980兲. “Electromagnetic
determination of soil water content measurement in coaxial transmis-
sion lines.” Water Resour. Res.,16共3兲, 574–582.
Vaid, Y. P., and Negussey, D. 共1998兲. “Preparation of reconstituted sand
specimens.” Advanced triaxial testing of soil and rock, ASTM STP
977, Philadelphia, 405–417.
Xia, H., and Hu, T. 共1991兲. “Effects of saturation and back pressure on
sand liquefaction.” J. Geotech. Engrg.,117共9兲, 1347–1362.
Yang, J., Savidis, S., Sato, T., and Li, X. S. 共2003兲. “Influence of vertical
acceleration on soil liquefaction: New findings and implications.”
Proc., Soil and Rock America 2003, Cambridge, Mass., Vol. 1.
Yoshimi, Y., Yanaka, K., and Tokimatsu, K. 共1989兲. “Liquefaction resis-
tance of a partially saturated sand.” Soils Found.,29共2兲, 157–162.
380 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2007