ARTICLE IN PRESS

Soil Dynamics and Earthquake Engineering 27 (2007) 191–199

www.elsevier.com/locate/soildyn

Visualization technique for liquefaction process in chamber experiments

by using electrical resistivity monitoring
Motoharu Jinguujia,, Selcuk Toprakb, Sunao Kunimatsua
a
National Institute of Advanced Industrial Science and Technology (AIST), Institute for Geo-Resources and Environment, PO305-8567,
AIST Tsukuba Central#7, Higashi 1-1-1, Tsukuba, Ibaraki, Japan
b
Civil Engineering Department, Pamukkale University, Kinikli Campus, Denizli, Turkey
Received 23 August 2006; accepted 23 August 2006

Abstract

A new technique for real-time visualization of soil liquefaction by using electrical resistivity monitoring is presented. The technique
utilizes the relationship between porosity and electric resistivity of soil, which was defined first by Archie in 1942. With the technique, the
change of density (including relative density) of saturated soil with respect to depth can be monitored in real-time before, during, and
after the dynamic excitation applied. A series of small chamber experiments on single-layer, two-layer, and five-layer soil models were
performed and soil liquefaction was monitored in each case. The experiments show that the technique can be used successfully to
visualize the change in density of saturated soil throughout the process of pore pressure build-up, liquefaction, and post-liquefaction
during shaking experiments. The effects of sand properties, stratification, duration and magnitude of dynamic excitation, and application
of successive dynamic excitations on the liquefaction and post-liquefaction behavior of soils are discussed. The technique is particularly
promising for large-scale chamber experiments where it can be used to control and monitor relatively large soil mass in addition to
determine its state of liquefaction during and after the shaking.
r 2006 Elsevier Ltd. All rights reserved.

Keywords: Liquefaction; Resistivity; Chamber experiment; Shaking table; Visualization

1. Introduction Liquefaction susceptibility of sands depends on their

compositional characteristics, such as particle size, shape,
The study of liquefaction and its effect on structures by gradation, and relative density. Experience from field
using chambers on shaking tables is increasingly being observations, and field and laboratory experiments by
utilized in earthquake engineering research in many many researchers showed this dependency clearly as can be
countries. The size of chambers generally changes from a found in numerous publications (e.g., [1,2]). A parameter
small size of about 0.01 m2 to a large size of 50 m2. With the that correlates well with these characteristics of soils can be
construction of the largest shaking table in the world with useful for liquefaction monitoring of soils. Such a
an area of 300 m2 (20
15 m) in Japan, almost real-size parameter is the formation factor, which is calculated by
structures like small buildings and foundations in soil can measuring electrical resistivity of soil and pore water. This
be tested under earthquake loadings. Increasing size parameter has been shown to correlate well with the above-
requires that the soil mass in the chambers should be mentioned characteristics of soils (e.g., [3,4]).
controlled and monitored more vigorously in order to Herein, we introduce a new technique for visualization
characterize its liquefaction behavior and its effect on the of the liquefaction phenomenon in the soil during the
structures. chamber experiments. This technique measures and ob-
serves the change of density or porosity of sand using
monitoring of electrical resistivity and the relationship
Corresponding author. Tel.: +81 298 61 8293; fax: +81 298 61 3612. between porosity and electric resistivity of sand in real
E-mail address: [email protected] (M. Jinguuji). time. The technique is described and evaluated here with

0267-7261/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.soildyn.2006.08.004
 ARTICLE IN PRESS
192 M. Jinguuji et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 191–199

small chamber experiments on single layer, two-layer, and

five-layer sand samples. This technique is also extended to Current Electrode
large chamber and field experiments, the result of which is
under evaluation and will be published in subsequent
papers.
V
2. Principle and method of resistivity monitoring

For saturated soils, the following Archie’s law is I V

applicable:
F ¼ ft =fw ¼ anm , (1)
V
where F is the formation factor, ft is the electrical
resistivity of soil, fw is the electrical resistivity of pore
water, and n is the porosity of the soil whose resistivity is
measured. The constants a and m depend on the soil type
and are determined from the results of laboratory
Current Electrode
experiment. The relative density of a particular soil mass
can be determined from the porosity of the soil. Eq. (1) is
Fig. 1. Schematic view of the method using current electrodes at fixed
generally used for clean sand and under the condition of position.
resistivity of pore water is relatively low (especially for fine
sand).
Herein, we propose two methods that can be applied for this way is the average resistivity of soil between each of
the electrical resistivity measurement of sands in liquefac- two potential electrodes. Therefore, it is assigned to the
tion experiments using a chamber. The first method mid-point height between respective potential electrodes.
(Method I) uses electric current electrodes at fixed positions The advantage of this method is that the high-speed
and measures the difference of electric potential between sampling can be applied because this method utilizes fixed
any two other electrodes by using the commutated direct position of current electrodes. The new equipment devel-
current. Two different types of current electrodes can be oped in this study has the capability to pass an electric
utilized to transmit current. One type is plane- or mesh- current of 20 mA and scan every one of 16 potential
type electrodes which are placed at the top and bottom of electrodes into separate data channels in 0.3 s. This method
the sand sample. The advantage of using this type of is suitable for the laboratory experiments that use a small
electrodes is that a set of needle electrodes that are partially vessel or chamber.
inserted into a cylindrical or rectangular vessel at certain The second method utilizes an electrode cable on which
depths from the outside can be used to measure electric electrodes are placed at equal intervals or line electrodes
potential of the soil between them. The disadvantage is that are placed along the inner wall like the application of
the top face of sand sample is covered by a plane- or mesh- Method I. In this method, the electrodes can be used as a
type electrode and it interferes with the use of measurement current or potential electrode. The electric potential can be
cables such as for a pore-pressure meter or an acceleration measured by using different methods, such as the 4-pole
meter. The other type is a set of continuous line electrodes method. Fig. 2 shows a schematic view of an electrode
which are installed at certain depths along the inner walls cable, electrodes, and the application of the 4-pole method
of a cylindrical or rectangular vessel. The line electrodes do of Method II. The black, solid rectangles on the cable show
not obstruct the use of other measurement cables because the location of electrodes. Because electrodes can be used
the top face is open. Also the electrodes can be used both to as current or potential electrodes, the distribution of
transmit current and to measure potential. resistivity in the vertical direction around the electrode
Fig. 1 shows a schematic view of a vessel, electrodes, and cable is continuously measured by changing a set of 4
the use of Method I. Two metal mesh electric current electrodes. During each time a 4-electrode set is selected,
electrodes are installed at the top and bottom part of the the top and bottom electrodes serve as current electrodes
vessel to transmit parallel current to the soil inside the and intermediate electrodes serve as potential electrodes
vessel. The line electrodes shown as circles in the figure are (Fig. 2). The current, I at current electrodes and the
installed at certain depths along the inner wall to serve as potential, V, at intermediate electrodes are measured and
potential electrodes. Note that the line electrodes in the the resistivity of sand is calculated. The resistivity
figure serve as the needle electrodes described before. The determined in this way is the average resistivity of the
electric current, I, transmitted by the current electrodes and spherical volume between two potential electrodes. Then,
the electric potential, V, at potential electrodes are the selection moves up or down one electrode, and a new 4-
measured and the distribution of vertical electrical resis- electrode set is selected. This process continues progres-
tivity in the sand is calculated. The resistivity determined in sively in the up or down direction until any single scanning
 ARTICLE IN PRESS
M. Jinguuji et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 191–199 193

32CH Electric Scanner

Main Control Unit

Induces and measures I, V

Scanning of electrodes using 4-Pole method

Control

A/D

V I

Computer
Electrodes or scanner
1) Records the measurements
2) Calculates resistivity change
3) Calculates relative density
change
Fig. 2. Schematic view of the method using electrode cable.
4) Plots various graphs

is stopped. The new equipment developed for this method Fig. 3. Liquefaction monitoring system.
can scan an entire 32 set of 4-pole electrodes into separate
data channels in maximum 0.1 s. However, this scanning
speed is very closely related to the frequency of alternative Electrode
DC transmitted wave. Because of the effect of induced
polarization, the sampling time is practically much higher
Electrode
than 0.1 s. Generally, this method is suitable for large-scale
experiments. Porous plate
Fig. 3 shows the integrated system used in this study for Copper mesh
the real-time monitoring of soil conditions before, during,
A3
and after liquefaction occurs. In other words, this system
provides real-time visualization of liquefaction phenomena
in the soil. The installed electrodes in the chamber or the
electrode cable are connected to a 32-channel electric
scanner. This scanner is linked to a main control unit which
P1
induces and measures alternative DC current and potential
100mm

600mm

as described in Methods I and II. The main control unit is Specimen

connected to a portable computer. The computer via the P2 A2
software developed in this project controls this unit,
records the measurements, calculates resistivity change
and change rate of resistivity and plots various resulting
graphs. The computer also receives the data from pore
water pressure meters and accelerometers placed in the A1
Plate of
sand sample. acrylate Copper mesh

3. Liquefaction monitoring of single-layer sand samples

Laboratory tests were performed on single-layer sand Water

samples to confirm the effectiveness of the resistivity
monitoring during liquefaction. In order to facilitate
comparisons, pore water pressure and acceleration mea- 200mm
surements were also taken during the experiments. Fig. 4
shows the cylindrical chamber that was used for the Fig. 4. Schematic view of the small cylindrical chamber used in the
laboratory test. The bottom and top mesh electrodes are resistivity experiments.
 ARTICLE IN PRESS
194 M. Jinguuji et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 191–199

current electrodes that transmit electric current to the sand 3, the excess pore water pressure dissipation and densifica-
sample. Because of this arrangement of current electrodes, tion continues initially and reaches a stable state after-
samples about the height of the chamber can be used for wards. The results show that the relative density of the soil
resistivity measurements. The potential electrodes are sample increased from about 30 to 50% after the
needle electrodes that are partially inserted into the excitation.
chamber from the outside. The locations of the three
accelerometers (A1, A2 and A3) as well as two pore 4. Factors that affect the resistivity behavior of liquefied
pressure transducers (P1 and P2) on the chamber are sands
shown in Fig. 4.
The medium grain size sand produced in Toyoura beach Some factors are known to affect the liquefaction of
in Japan was used in the experiments. The soil sample was sands such as sand type, duration and amplitude of
prepared using the boiling method. The height of the sand excitation, and application of successive excitations. The
sample in the chamber was about 50 cm. The remaining influence of these factors on the resistivity behavior of
10 cm at the top of chamber was filled with water. The box liquefied sands was tested in the following experiments.
was placed on a small shaking table that was controlled by Fig. 6 shows the chamber used in the experiments. The line
an electric linear motor and 4-s excitation was applied in electrodes along the walls of the chamber were placed at
the experiment. The waveform of the excitation was sine- 1.5 cm intervals. The line electrodes can be continuously
wave with a frequency of 5 Hz. The amplitude of the measured by changing a set of 4 electrodes, like Method II.
acceleration was 600 cm/s2. The tests with different shaking Using this chamber with Method II eliminates any need for
accelerations and durations are reported in Jinguuji et al. water on the top of the sample because the method can
[5]. adapt to any change in the height of the sample by the
Fig. 5a shows the change of resistivity starting from 5 s densification during the liquefaction. The 4-pole method
before the beginning of the shaking. The resistivity values can be applied even when the height of the sample changes,
are converted to relative density of soil by using Archie’s because the current electrode is not fixed on the top. The
law. Fig. 5b shows the change of relative density of the Toyoura and Keisa No. 6 sands were used in the
sand during the test as determined from the change of experiments. The porous stone is placed at the bottom of
resistivity in Fig. 5a. P1 and P2 in the figure show the the chamber and the soil sample was prepared using the
locations of pore-pressure meters. Fig. 5c shows the pore- boiling method. The initial relative density of the sand
pressure change at P1 and P2 and P2-P1 shows the pressure sample was about 30%. The box was placed on a small
difference between P1 and P2. Fig. 5d shows the accelera- shaking table and 4-s excitation was applied. The wave-
tion measured at the bottom of the chamber. Because the form of the excitation was sine-wave with a frequency of
accelerations recorded by three instruments were almost 5 Hz. The amplitude of the acceleration was 150 cm/s2.
the same, only the recording at the bottom is provided in Toyoura and Keisa No. 6 sands were used to test the
the figure. effect of sand type. Both sands are commonly used for
Three zones are inferred from the resistivity pattern in liquefaction experiments in Japan. Particle size of Toyoura
Fig. 5a as shown by solid lines. Zones 1 and 2 show the sand is finer than Keisa No. 6 sand. The D50 values for
resistivities before and after the excitation, respectively. Toyoura and Keisa No. 6 sands are about 0.125 and
Although the soil is in liquefied state in Zone 2 as indicated 0.275 mm, respectively. The sample heights in this experi-
by the pore water pressure increase up to the overburden ment were about 13 cm. The relative density of both
pressure, no significant resistivity change was observed. samples was about 30%. Figs. 7a and b show the resistivity
This is because the soil was already in very loose state results from the experiments for Toyoura and Keisa No. 6
before the excitation and the density of soil did not change sand samples, respectively. The origin of abscissa starts at
significantly during the liquefaction. The border of Zones 2 5 s which reflects that the shaking initiated 5 s after the
and 3 marks the start of the dissipation of excess pore recording started. The legends next to the figures show the
water pressure and the beginning of densification of the color scale for the resistivity change relative to each
liquefied soil. This observation can be seen by comparing sample’s initial resistivity. Note that the relative resistivity
the pore pressures at point A for P2 and at point B for P1 change rather than the measured resistivity values were
in Fig. 5b. The relative density increases quickly during this used in Fig. 7, because it showed changes of the values
transition. Because the border of Zones 2 and 3 clearly more clearly. Comparison of the figures shows that the
identifies the start of the excess pore pressure dissipation densification front has higher slope in Keisa No. 6 sand
and the beginning of densification, it can be used to than it did in finer grain Toyoura sand. This is expected as
determine the status of liquefaction at a particular time and the dissipation of excess pore water pressure is faster in
depth. We call this border as ‘‘densification front’’. coarser grain sand. Also, more volume of Keisa No. 6 sand
Densification starts at the lower parts of the sample at sample densified to a higher degree.
first and moves up as excess pore water pressure dissipates In order to test the effects of the shaking duration,
upward. The slope of the border is controlled by the extent Toyoura sand samples were shaken for 4 and 30-s periods.
of the excess pore pressure dissipation in the soil. In Zone Fig. 8 shows the result of resistivity monitoring starting
 ARTICLE IN PRESS
M. Jinguuji et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 191–199 195

Fig. 5. Change of relative density and pore pressure during liquefaction (modified from [5]).

from 5 s before the beginning of the shaking. The results process reached constant state in the entire sample as soon
indicate that the densification started and completed at the as the shaking stopped. The comparison of Figs. 8a and b
lower parts of the sample much before the shaking stopped. shows that more volume of soil has higher resistivities
The densification process progressed upward and the during the long duration shaking. For comparison
 ARTICLE IN PRESS
196 M. Jinguuji et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 191–199

Line Electrodes
18 lines
300 mm

Mesh Electrode

150mm

600 mm
Injection

Fig. 6. Schematic view of the small rectangular chamber used in the electrical resistivity experiments: (a) Toyoura sand; (b) Keisa No. 6 sand.

Fig. 7. Change of electrical resistivity measured in two different kinds of sand: (a) 4-s excitation; (b) 30-s excitation.

purposes, the resistivity scale in Fig. 8b was kept the same 5. Liquefaction behavior in multi-layer sand samples
as the one in Fig. 8a. The actual resistivity in the long
duration test, however, reached about 190 Om levels. Natural geologic systems in general have heterogeneous
To test the effects of successive excitations, 4-s shaking structure and it is easy to think that field settings would
was applied consecutively six times to the Toyoura sand introduce complicacy into predictions of liquefaction
sample. Fig. 9 shows the results from the experiment. There occurrence and behaviour [6]. For example, liquefaction
were about 100s intervals between each excitation which in multi-layer soils, which is the more common case in
were not shown in the figure. The arrows indicate the nature rather than the exception, is more complex than the
beginning of each shaking. Each successive excitation single-layer soil. Different permeability levels of the layers
resulted in some increase in the resistivity; however, the affect the dissipation of excess pore water pressure. Low
level of change of resistivity decreases after each excitation. permeability layers may act like barriers and may entrap
This is consistent with the expectation that the soil is the water pressure. To monitor and visualize the behavior
densified more after each shaking and the change of of multi-layer soil using electrical resistivity measurements,
densification after each cycle reduces. experiments were performed with 2-layer and 5-layer soil
 ARTICLE IN PRESS
M. Jinguuji et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 191–199 197

Fig. 8. Electrical resistivity measurements at different shaking durations.

Fig. 9. Electrical resistivity measurements during successive six 4-s shakings.

samples. A chamber similar to the one shown in Fig. 6, confined to the bottom layer. No resistivity change, and
except for the width of 30 cm, was used in the tests. The soil hence no density increase, occurred in the top layer and just
samples were prepared by dropping sand from the top into below the boundary of the two layers.
the water. When the layers are more complex and finer sand is
Two-layer models were prepared using two types of sand included in the intermediate layers, the liquefaction
with different particle sizes: Keisa No. 5 for the lower part behavior of the sample becomes more complex. Fig. 11
and Keisa No. 6 for the upper part. The heights of the shows the layer order and the resistivity change of the 5-
layers are 16.8 and 11.4 cm, respectively, for the bottom layer sample that consists of three types of sand: Keisa
and top layers. The D50 values for Keisa No. 5 and 6 sands Nos. 5, 6, and 8. The D50 value of Keisa No. 8 is 0.121 mm.
are about 0.527 and 0.275 mm, respectively. The soil In Fig. 11, the resistivity change is expressed as resistivity
sample was liquefied by hitting a hammer five times at change rate because the differences of initial resistivity of
the bottom of the chamber. Fig. 10 provides the result of each layer are big, hence it is difficult to show all changes of
resistivity monitoring during the test. Comparison of this resistivity. The resistivity change rate, RC, is expressed as
figure with previous resistivity monitoring of single layer ðR  Ri Þ=Ri , where R and Ri are resistivity at any measured
shows that existence of a finer sand layer at the top affected time and initial resistivity, respectively. Again, the pattern
the soil liquefaction significantly. Density increase was is different from the one for single-layer samples. The
 ARTICLE IN PRESS
198 M. Jinguuji et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 191–199

Fig. 10. Electrical resistivity change of two sand layers that consist of different particle size.

Fig. 11. Electrical resistivity change rate during liquefaction of the 5-layer sand sample.

figure shows that electrical resistivity increases in relatively 6. Summary and conclusions
coarse sands such as Keisa No. 5 or 6, indicating a density
increase. The Keisa No. 8 layers have relatively low This paper presents a new technique to monitor and
permeability and therefore act like barriers. The discharged visualize the behavior of saturated soils during the
water from the coarser sands is trapped at these layers and liquefaction and post-liquefaction stage. The technique
the electrical resistivity or density will decrease at the utilizes the unique relation between soil electrical resistivity
layers. In fact, the observation from outside the chamber and porosity and determines the change of relative density
that is made of acryl transparent material confirms the with respect to depth and time. The results from the
water films at the fine layers. This experiment was made laboratory experiments on single-layer samples and com-
using the small chamber, but further study will clear this parisons with pore water and accelerometer measurements
phenomenon in future. during the tests show clearly that the electrical resistivity
 ARTICLE IN PRESS
M. Jinguuji et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 191–199 199

measurements capture the behavior of liquefied sands. Masanori Hamada and his students for the support and
Resistivity monitoring was able to show the effects of assistance.
different well-known factors such as grain size, duration
and magnitude of excitation, and application of successive
dynamic excitations on liquefaction of soils. The technique
provides a tool to investigate and compare the liquefaction References
in single- and multi-layer soils. The difference in liquefac-
tion pattern in single-layer, two-layer, and five-layer soils [1] National Research Council. Liquefaction of soils during earthquakes.
during the small chamber experiments indicates the Washington, D.C.: National Academy Press; 1985.
[2] Kramer SL. Geotechnical earthquake engineering. New Jersey:
complexity of the liquefaction phenomenon in the field. Prentice-Hall; 1996.
In addition to studies of liquefaction in small chambers, the [3] Archie GE. The electrical resistivity log as an aid in determining some
technique is expected to be useful for large chamber reservoir characteristics. A.I.M.E. Transactions 1942;146:54–62.
liquefaction experiments on shaking tables or field experi- [4] Arulmoli K, Arulanandan K, Seed HB. New method for evaluating
ments with explosives. liquefaction potential. J. Geotech. Eng. 1985;111(1):95–114.
[5] Jinguuji M, Kunimatsu S, Izumi H, Mochizuki T. Development of
visualization technique of relative density of sand during liquefaction
Acknowledgments using resistivity and consideration of the results. Japan Soc. Civ. Eng.
2001;680(III-55):201–9 (in Japanese).
Some of the tests were performed at Waseda University [6] Toprak S, Holzer TL. Liquefaction potential index: field assessment.
J. Geotech. Geoenviron. Eng. 2003;129(4):315–22.
Geotechnical Engineering Laboratory. We thank Prof.