The Filtration Mechanism and Micro-observation of Soil-geotextile Systems under Cyclic Flows

Abstract

Geotextile filters have been used frequently in revetments for protection of riverbanks, levees, seashores, etc. The filters may be subjected to different flow conditions such as unidirectional or cyclic flows. Besides, the period of flow may vary from short to long due to the fluctuation of water table caused by sea waves, boats, tidal activity, or periodic drawdown of water for irrigation purposes. Due to few literatures on cyclic flow studies, this paper presents the investigation on soil-geotextile filtration mecha-nisms under long-term cyclic flows using a self-developed apparatus. The soil samples are composed mainly of sand, with 0-20% fines content of silts and clays. The factors taken into account are overburden pressure and cyclic flow periods ranging from long to short, in order to simulate a range of waves. A stereomicroscope is utilized to observe the microstructure of the geotextile filter after testing. The results show that both the overburden pressure and the fine soil content play important roles in the filtration, soil boiling, and settlement behaviors of a soil-geotextile filtration system. Furthermore, the microscopic images of the geotextile show that soil clogging under cyclic flows is not so serious as that under unidirectional flows. Besides, a bridging network can be formed under long-term cyclic flows in the areas near the filter that are not supported by marbles.

Full text

Rong-Her Chen et al.: The Filtration Mechanism and Micro-Observation of Soil-Geotextile Systems Under Cyclic Flows 101
Manuscript received July 31, 2008; revised October 9, 2008; ac-
cepted October 9, 2008.
1 Professor (corresponding author), Department of Civil Engineer-
ing, National Taiwan University, No. 1, Sec. 4, Roosevelt Rd.,
Taipei, 10617 Taiwan (e-mail: rongherchen@ntu.edu.tw).
2 Postdoctoral fellow, Department of Civil Engineering, National
Taiwan University, No. 1, Sec. 4, Roosevelt Rd., Taipei, 10617
Taiwan (e-mail: d90521023@ntu.edu.tw).
3 Former master’s student at National Taiwan University, No. 1, Sec.
4, Roosevelt Rd., Taipei, 10617 Taiwan (e-mail: bing830202@
msn.com).
THE FILTRATION MECHANISM AND MICRO-OBSERVATION OF
SOIL-GEOTEXTILE SYSTEMS UNDER CYCLIC FLOWS
Rong-Her Chen1, Chia-Chun Ho2 and Wen-Bin Chung3
ABSTRACT
Geotextile filters have been used frequently in revetments for protection of riverbanks, levees, seashores, etc. The filters may
be subjected to different flow conditions such as unidirectional or cyclic flows. Besides, the period of flow may vary from short to
long due to the fluctuation of water table caused by sea waves, boats, tidal activity, or periodic drawdown of water for irrigation
purposes. Due to few literatures on cyclic flow studies, this paper presents the investigation on soil-geotextile filtration mecha-
nisms under long-term cyclic flows using a self-developed apparatus. The soil samples are composed mainly of sand, with 0-20%
fines content of silts and clays. The factors taken into account are overburden pressure and cyclic flow periods ranging from long
to short, in order to simulate a range of waves. A stereomicroscope is utilized to observe the microstructure of the geotextile filter
after testing. The results show that both the overburden pressure and the fine soil content play important roles in the filtration, soil
boiling, and settlement behaviors of a soil-geotextile filtration system. Furthermore, the microscopic images of the geotextile
show that soil clogging under cyclic flows is not so serious as that under unidirectional flows. Besides, a bridging network can be
formed under long-term cyclic flows in the areas near the filter that are not supported by marbles.
Key words: Geotextile, filter, cyclic flow, soil erosion, bridging network.
1. INTRODUCTION
Filtration is a process that suspended or dissolved solids are
separated from a fluid as it flows through a porous media. A soil
filter is used for this purpose. In the filter design, it is based on
parameters such as channel morphology, concentration of sus-
pended solids or dissolved solids characteristics, and fluid prop-
erties such as viscosity and density. Another important factor is
the source of driving forces which may be hydrodynamic, gravity,
suction or positive seepage pressures.
The water infiltrating through the pores of a soil-geotextile
system may change the soil structure and decrease the intrinsic
coefficient of hydraulic conductivity. To prevent this phenome-
non, a suitable geotextile should be selected to impede suffi-
ciently the movement of soil particles and to build a natural filter
layer. In turn, this filter layer may restrain smaller particles mi-
gration until a stabilization state is established. In general, the
faster a natural filter establishes the smaller amount of soil parti-
cles will migrate.
The use of geotextile filters for revetments to protect river-
banks, levees, seashores has become popular. The filters may be
subjected to different flow conditions such as unidirectional or
cyclic flows. Besides, the period of cyclic flows may vary from
short to long due to the fluctuation of water table caused by sea
waves, boats, tidal activity, or periodic drawdown of water for
irrigation purposes. A stable soil-geotextile filter system must be
formed under these circumstances.
Under cyclic flow condition, soil particles may migrate or be
washed away that induces failure of a revetment. For example, a
revetment failure has been reported by Chen et al. (2008). The
revetment was composed of a layer of concrete cover lain on the
top of a soil slope. The constituent of the soil was mainly sands
with about 10% silt and clay. The filter was made of gravels
wrapped by geotextiles and placed at the drainage holes. The
revetment of 1.6 km long failed several months after it had been
completed. According to the investigation, the cause was due to
periodic drawdown of water about 2 m high per week. In addition
to this case, there are many tidal lands and harbor structures built
in western Taiwan. The tide causes variation in the groundwater
level which may lead to soil loss or ground settlement and jeop-
ardize structures. The rubble-mound groin is often used for pro-
tecting coastlines. In Taichung Port, the loss of soil at one loca-
tion of the south levee damaged a road just behind the levee. This
kind of damage also occurred in the Changbin Industrial Center
where settlements, cracks, sinkholes were found on the surface of
the road adjacent to the levee (Hsu, 2007; Liao and Chu, 2002).
The studies of the particle structure and the hydraulic char-
acteristics in the zone adjacent to a filter are essential to under-
stand the filtration mechanism under cyclic flows. However,
most of previous researches considered simpler conditions, such
as unidirectional flow, than real in-situ conditions (Giroud, 1982;
Lawson, 1982; Chen and Chen, 1986; Luettich et al., 1992; Ay-
dilek, 2006). Very few recent laboratory studies on geotextile
filters have examined conditions of dynamic, cyclic, and pulsat-
ing flows (Cazzuffi et al., 1999; Fannin and Pishé, 2001; Chew et
al., 2003; Chen et al., 2006). Due to few literatures of cyclic flow
studies, this paper presents the investigation on soil-geotextile
filtration mechanisms under long-term cyclic flows by employing
a self-developed apparatus (Chen et al., 2008). A stereomicro-
scope is also utilized to observe the inside condition of geotextile
filters as well as the adjacent soil structures.
Journal of GeoEngineering, Vol. 3, No. 3, pp. 101-112, December 2008

102 Journal of GeoEngineering, Vol. 3, No. 3, December 2008
2. FILTRATION MECHANISMS
Regarding a natural filter formation, two mechanisms based
on perpendicular flow conditions have been proposed by Rollin
and Lombard (1988), i.e., the bridging network and the vault
network formations (Fig. 1). The bridging network formation
usually occurs in non-cohesive soils. At first, large particles are
stopped at the surface of a geotextile structure. In turn, these par-
ticles retain smaller particles; this process continues until the soil
stabilizes. On the other hand, the vault network formation occurs
in non-cohesive soils with appreciable clay content or in cohesive
soils. Vaults are initiated by geotextile fibers, as shown in a pho-
tographic cross-section of a geotextile sample collected in-situ
and presented schematically in Fig. 2.
Moreover, Scheidegger (1957) divided filtration into three
classes: medium, cake, and depth filtrations. In the medium fil-
tration, particles larger than the filter entry pores are retained,
generally at surface openings or shortly inside the upstream face.
This type of filter thus behaves like a sieve. In the depth filtration,
the particles smaller than the filter pores and the dissolved mate-
rials are intercepted and retained within the filter section. In the
cake filtration, solids do not enter the filter to a great extent, but
accumulate on or in front of the surface of the filter. Soil filters
are a variation of cake filtration to some extent.
The forming process of a filter cake in soil-geotextile sys-
tems is quite complicated. Mlynarek et al. (1991) summarized
the occurrence mechanisms as the five phases presented in Fig. 3.
Nevertheless, a distinction must be made between unidirec-
tional flow and cyclic flow in filtration design using geotextiles.
Though it is believed that a natural bridging network is induced
in the soil adjacent to the geotextile during unidirectional flow,
this network may or may not develop for long-term cyclic flows.
For instance, under impacting water flow such as wave activity
(short-term cyclic flow less than 10 sec/cycle), the influence of
changing direction of flow and associated seepage forces can
destabilize such a network (Giroud, 1982; Köhler, 1993). This
study specifically focused on long-term cyclic flows due to tidal
activity or water drawdown such as for irrigation purposes de-
scribed above.
3. APPARATUS FOR CYCLIC FLOW TEST
To study the filtration mechanism of geotextiles under cyclic
flows, an apparatus was developed at National Taiwan University
(Fig. 4), with some modifications of the cyclic flow model of
ENEL, Italy (Cazzuffi et al., 1999). This apparatus is capable of
simulating cyclic flow normal to the soil-geotextile interface. It
consists of a cyclic wave generator, an acrylic sample chamber, a
water reservoir and wash-out collecting tank, and a vertical pres-
sure application system. Figure 5 shows a detailed schematic
view of the internal set-up of the chambers. The acrylic specimen
cylinders consisting of upper and lower chambers are for the
convenience to observe the state of soil erosion. The lower
chamber contains marbles which simulate the secondary armor
layer of a revetment. A porous steel plate is placed below the
lower chamber to maintain the marbles. The geotextile specimen
is laid on the marbles and clamped in the groove between upper
and lower chambers; and then the soil specimen is filled in the
upper chamber. There are four pore pressure transducers (P01,
P02, P03 and P04) placed at different positions to monitor the
fluctuation of pore pressure. The measured pore pressure can
provide information regarding various phenomena such as blind-
ing, clogging, or blocking. Ports P01 and P02, with a distance of
only 30 mm, are located just above and below the level of geo-
textile specimen, respectively. The intention is to capture the
interaction between geotextile and soil close to the interface
(Palmeira and Fannin, 2002). In addition, two settlement gages,
S01 and S02, are mounted on the top of the porous steel plate to
obtain the average settlement during testing.
Water flow
Stone filled drain
Original soil
structure
Filter zone
in soil
Bridging
network of
large
particles
Fabric
Drain
Fig. 1 Bridging network formations (Rollin and Lombard,
1988)
Water flow
Fiber
Soil
Geotextile
Fig. 2 Vault network formations (Rollin and Lombard, 1988)
Fig. 3 Typical mechanisms of flux decay of a system (Mlynarek
et al., 1991).

Rong-Her Chen et al.: The Filtration Mechanism and Micro-Observation of Soil-Geotextile Systems Under Cyclic Flows 103
Fig. 4 Photo of cyclic flow apparatus (Ho, 2007)
330
150180
220
220
S01 S02
Load
P01
P02
P03
P04
Unit : mm
Porous steel plate
Marbles
Soil
Geotextile
Porous steel plate
330
150180
220
220
S01 S02
Load
P01
P02
P03
P04
Unit : mm
Porous steel plate
Marbles
Soil
Geotextile
Porous steel plate
Fig. 5 Detailed schematic view of the internal setup of
chambers (Chen et al., 2008)
4. TEST MATERIALS
In order to understand the effect of fine soil content on the
filtration behavior of geotextile, one geotextile together with
seven soil compositions is studied. Their engineering properties
are described below.
4.1 Soils
The soil tested is composed of various weight proportions of
sand, silt and clay. The sand is Vietnam sand, classified as SP
(poorly-graded sand) based on the Unified Soil Classification
System (USCS). The properties of this sand are: specific gravity
Gs = 2.66, maximum void ratio emax = 0.76 ~ 0.77, and minimum
void ratio emin = 0.56 ~ 0.57. In order to avoid the influence of the
fines in original sand, the soil was washed and filtered out the
fines smaller than 0.074 mm before testing.
The silt soil was obtained from an alluvial soil of Xindian
Creek. This alluvial soil was air-dried at first, and coarse particles
larger than 0.074 mm were filtered out by sieves. The collected
fine particles were then mixed with water and left for sedimenta-
tion. After the slurry dried out, the non-cohesive soil remaining at
the bottom was the silt for testing. The classification of this silt is
ML.
The clay soil was sampled from the sediment of Keelung
River. The sediment soil was pretreated following the same way
as mentioned above. However, the cohesive soil was the soil
floating at top.
Seven soil specimens were prepared by weight; their com-
ponents, soil classification symbols, and hydraulic conductivities
are tabulated in Table 1. For the mixtures of G-01 to G-05, the
percentage of silt content increases from 0% to 20%. For G-06
and G-07, the amount of fines content (less than 0.074 mm) is
10%. However, G-06 contains 6.5% silt and 3.5% clay; and G-07
contains 3.5% silt and 6.5% clay. As can be seen, the hydraulic
conductivity reduces as the fines content and the clay amount in-
crease.
Table 1 The proportions of soil specimens
Tes t N o.
Vertical
Pressure
(kPa)
Sand
(%)
Silt
(%)
Clay
(%)
Classification
(USCS)
Conductivity
ks (cm/s)
a 70
G-01 b 140 100 0 0 SP 4.38 × 10−2
a 70
G-02 b 140 95 5 0 SP 3.25 × 10−3
a 70
G-03 b 140 90 10 0 SP-SM 1.43 × 10−3
a 70
G-04 b 140 85 15 0 SM 8.33 × 10−4
a 70
G-05 b 140 80 20 0 SM 4.38 × 10−4
a 70
G-06 b 140 90 6.5 3.5 SP-SM 8.41 × 10−4
a 70
G-07 b 140 90 3.5 6.5 SP-SC 7.22 × 10−5
The particle size distributions of seven specimens are shown
in Fig. 6; the characteristic particle sizes are expressed in Table 2.
Because the parts of larger particle distributions of seven speci-
mens are about the same, only the parts less than 30% passing are
shown. From G-01 to G-05, the average particle diameter, d50,
and the effective grain size, d10, decrease. Bhatia and Huang
(1995) suggested that soils with values of coefficient of curvature
above 7 should be considered as internally unstable and below
this value internally stable. As can be seen from Table 2, the co-
efficients of curvature of all specimens are less than 7; conse-
quently, they are considered as internally stable. In addition, the
geometrical stability of soil was evaluated using the criterion
proposed by Kenney and Lau (1985), which is based on a method
of describing the shape of grain-size distribution. The increment
of percent passing (H) that occurs over a designated grain size
interval of d to 4d is compared to the percent passing (F) at grain
size d. A boundary defined by a stability index, H/F = 1.0, is for
separating unstable soils from stable soils. According to the result
analyzed by this method (Fig. 7), all specimens except G-06 are
stable. However, these specimens will be further investigated
regarding the stabilities of soil-geotextile filter systems under
cyclic flow conditions.

104 Journal of GeoEngineering, Vol. 3, No. 3, December 2008
0
5
10
15
20
25
30
0.001 0.01 0.1 1
Particle size, d (mm)
Passing percentage (%)
G-01
G-02
G-03
G-04
G-05
G-06
G-07
O
90
of
geotextile
Fig. 6 Particle size distributions of test specimens
0
10
20
30
40
50
60
70
80
90
100
0 102030405060708090100
F (%)
H (% )
G-02
G-03
G-04
G-05
G-06
G-07
Unstable
Stable
Fig. 7 The result of analysis of the internal stability of
soil filter by the criterion of Kenney and Lau
(1985)
Table 2 The characteristic values of soil specimens
Tes t
No.
d10
(mm)
d15
(mm)
d30
(mm)
d40
(mm)
d50
(mm)
d60
(mm)
d85
(mm)
d90
(mm) CuCcIp
G-01 0.180 0.180 0.210 0.240 0.270 0.302 0.408 0.470 1.7 0.8 −
G-02 0.170 0.170 0.204 0.230 0.265 0.298 0.402 0.466 1.8 0.8 −
G-03 0.110 0.150 0.200 0.215 0.255 0.285 0.400 0.452 2.6 1.3 −
G-04 0.059 0.115 0.181 0.209 0.240 0.280 0.392 0.447 4.7 2.0 −
G-05 0.048 0.082 0.176 0.200 0.230 0.275 0.388 0.438 5.7 2.3 −
G-06 0.105 0.150 0.200 0.215 0.255 0.285 0.400 0.452 2.7 1.3 −
G-07 0.092 0.150 0.200 0.215 0.255 0.285 0.400 0.452 3.1 1.5 5
Note: dx = soil particle size corresponding to percent passing; Cu = coefficient
of uniformity; Cc = coefficient of curvature; IP = plasticity index.
4.2 Geotextile
With regard to the geotextile for testing, Hoare (1984) pro-
posed to adopt thin heat-bonded geotextiles for unidirectional
flow conditions and thick needle-punched geotextiles for cyclic
flow conditions. On the other hand, Giroud et al. (1998) sug-
gested that a two-layer nonwoven geotextile is suitable for pro-
tecting river or coastal revetment. In view of this, this study em-
ployed a thick two-layer needle-punched geotextile for a series of
tests; the relevant properties of the geotextile are given in Table
3.
Table 3 The properties of the geotextile
Properties and testing method Unit Symbol Va lu e
Characteristic opening size based on
hydrodynamic sieving, EN ISO 12956 mm O90 0.08
Hydraulic conductivity normal to
the plane, EN ISO 11058 cm/s kg 6.0
Elongation, EN ISO 10319 % ε max 85
Tensile strength, EN ISO 10319 kN/m Tmax 23
Mass per unit area, DS EN 965 g/m2 μA 400
Thickness, DS EN 964-1 (2 kPa) mm tGTX 3.5
4.3 Examination of Test Materials
For cyclic flow conditions, Schober and Teindl (1979) sug-
gest that the coefficient of hydraulic conductivity of geotextile
must be greater than that of soil. The Federal Waterways Engi-
neering and Research Institute in Germany (BAW, Bundesanstalt
für Wasserbau, 1993) proposes
10 for non-cohesive soil
gs
kk≥ (1)
100 for cohesive soil
gs
kk≥ (2)
where kg and ks are the hydraulic conductivities of geotextile and
soil, respectively. Since the coefficient of hydraulic conductivity
of tested geotextile, kg, is 6.0 cm/s and according to the perme-
ability of each specimen listed in Table 1, it is apparent that the
geotextile satisfies the permeability requirement for all soil
specimens. Moreover, a good soil-geotextile filter system needs
also satisfy the retention criteria listed in Table 4. For this ex-
amination, the parameters shown in Table 2 are used. The result
of each specimen also satisfies the retention criteria.
Table 4 The retention criteria for soils under cyclic flow
condition
References Base soil type Retention criterion
non-cohesive soil O90 < d50
Heerten (1982)
cohesive soil
O90 < 10 d50 and
O90 ≤ d90 and
O90 ≤ 100 μm
ASPG (1985) d40 > 60 μm O90 ≤ 1.5 d10 (Cu)1/2
and O90 ≤ d60
loose sand (Cu > 4) O90 < 0.6 d85
CFGG (1986)
loose sand (Cu ≤ 4) O90 < 0.48 d85
DGEG (1986) d40 > 60 μm O90 < d90
Cu > 5 50 μm
<
O90 < d90
PIANC (1987)
Cu < 5 50 μm < O90 < 0.7 d90
Młynarek (2000) d50 ≥ 75 μm
and Cu < 6
O90 < 0.8 d50 or
150 μm
<
O90 < d50
Particle size, d (mm)
Passing percentage (%) H (%)
F (%)

Rong-Her Chen et al.: The Filtration Mechanism and Micro-Observation of Soil-Geotextile Systems Under Cyclic Flows 105
5. TEST PROCEDURE
To investigate the effect of overburden pressure on the fil-
tration function of geotextile under cyclic flow, the specimens
were subject to loadings of 70 kPa and 140 kPa, respectively. In
addition, the wave period applied ranged from long to short pe-
riods, i.e., 600, 300, 150, and 75 seconds, respectively. The test-
ing procedure is summarized briefly as follows:
1. After the apparatus had been set up, the test soil at optimum
water content was divided into eight layers of equal weight
and placed in the upper chamber. Each layer of soil was com-
pacted until it reached the maximum dry density. With these
eight layers, the total height of the soil specimen was 45 cm.
The maximum dry density and the optimum water content
were obtained by the Standard Proctor Compaction Test
(ASTM D698). The test results are shown in Fig. 8. However,
the pure sand specimen, G-01, was prepared with the relative
density equal to 88%.
1.60
1.64
1.68
1.72
1.76
1.80
1.84
5 7 9 1113151719
Water content, ω (%)
Dry density, ρ
d
(t/m
3
)
G-02
G-03
G-04
G-05
G-06
G-07
Fig. 8 Standard Proctor compaction curves
2. The specimen was then saturated with water gradually from
the bottom until water reached the top of the specimen. This
procedure repeated three times in order to ensure full satura-
tion. Alternatively, pore pressures recorded by four transduc-
ers were compared with the elevation of piezometers, used as
a check to examine whether the specimen was fully saturated.
3. The loading device was fixed to the chamber and normal
pressure was applied in the increments of 10 ~ 20% of maxi-
mum normal load. A subsequent increment of loading was
added only after the settlement induced by previous loading
had become very small and when the pore-water pressure was
equal to the static water pressure measured by the piezome-
ters.
4. Under constant normal pressure, 70 kPa or 140 kPa, the
specimen was then subjected to cyclic flows in the order of
600, 300, 150 and 75 sec/cycle of wave period, respectively.
The test duration for each constant period was at least 48
hours until the variation in pore-water pressure became stable.
5. During testing, pore-water pressures at different positions
were recorded automatically by piezometers; and the settle-
ment at the top of specimen was monitored by settlement
gages.
6. After the test was finished, soil samples were taken at differ-
ent locations for investigation of the variation in grain size
distribution. Moreover, a stereomicroscope was utilized to
observe the inside condition of the geotextile.
6. TEST RESULTS
As stated above, pore-water pressures at various levels in the
specimen were recorded in order to examine the phenomena such
as clogging, blocking, or boiling that might occur in a filtration
system. The measured settlement can also be compared with the
pore-water pressure to find out if there exists a relationship be-
tween them.
6.1 Pore-Water Pressure
Figure 9 shows the pore-water pressure response of pure
sand, G-01, at wave periods of 600 and 150 sec/cycle, respec-
tively. The peak pore-water pressure increases as the wave period
decreases. This is because pore-water pressure has not dissipated
completely when the next cycle of flow comes up in the cases of
shorter period flows. For pure sand under various normal stresses,
the difference in pore pressure is not significant. This implies that
pure sand has an incompressible and porous structure to restrain
soil particles from migrating under the action of normal pressure.
Furthermore, the pore-water pressures of P01 and P02 are about
equal, indicating no clogging or blocking within the geotextile.
-6
-4
-2
0
2
4
6
8
10
12
14
20.0 20.1 20.2 20.3 20.4 20.5
Time, t (hour)
Pore pressure, u (kPa)
P01 (70 kPa) P02 (70 kPa) P03 (70 kPa)
P01 (140 kPa) P02 (140 kPa) P03 (140 kPa)
(a) 600 sec/cycle
-6
-4
-2
0
2
4
6
8
10
12
14
40.3 40.4 40.5 40.6
Time,
t
(hour)
Pore pressure, u (kPa)
P01 (70 kPa) P02 (70 kPa) P03 (70 kPa)
P01 (140 kPa) P02 (140 kPa) P03 (140 kPa)
(b) 150 sec/cycle
Fig. 9 Pore-water pressure of pure sand (G-01)
Water content, ω (%)
Dr
y
densit
y
,
ρ
d
(
t/m3
)
Time, t (hour)
Pore pressure, u (kPa) Pore pressure, u (kPa)
Time, t (hour)

106 Journal of GeoEngineering, Vol. 3, No. 3, December 2008
A phenomenon is noteworthy for G-02-b under the wave
period of 300 sec/cycle, i.e., the pore-water pressure response
was not uniform during testing (see Fig. 10); it decreased with
time for several hours and then remained stable. This specimen,
G-02, contains only 5% silt; the fines in voids were probably not
enough to form a dense structure even under a high normal pres-
sure. Consequently, the pore-water pressure increased gradually
in the beginning, but once it reached a certain value the fines then
started to migrate. At some locations where there was significant
loss of fines, local soil boiling might occur. In the mean time,
loss of fines also increased the hydraulic conductivity of soil and
decreased pore-water pressure. The same phenomenon can also
be found in specimen G-03-a, though it contains more silt, 10%.
In this case, boiling phenomenon was owing to the specimen
subjected to a low normal pressure. In other words, the normal
pressure was not high enough to impede fines to move.
Figures 11 and 12 present the pore-water pressure response
for silty sand specimens, G-03 and G-04, under the wave periods
of 600 sec/cycle and 150 sec/cycle, respectively. They illustrate
that fines content may cause the peak pore-pressure to increase.
As to the effect of wave period, pore-water pressure in long pe-
riod condition has enough time to transmit, hence is lower than
that in short period condition. Moreover, it is obvious that normal
pressure affects pore-water pressure as well; higher normal pres-
sure induces a higher response. This is because not only fine par-
ticles affect the response but also they are more susceptible to
form a denser structure when under load.
For specimens G-04 and G-05 that contain 15% and 20% of
silt, respectively, the soil structures tend to be more stable.
Therefore, the amplitudes of pore-water pressure are uniform, as
local soil boiling phenomenon was not observed during testing. It
seems that silt content of 10% is approximate the threshold value
for local boiling to occur.
The effect of clay content upon pore-water pressure can be
seen from Fig. 13. Figure 13(a) shows the variation in pore-water
pressure for specimen G-06 under 150 sec/cycle of wave period.
Comparing Fig. 13(a) with Fig. 11(b), specimen G-06 having
3.5% of clay content results in higher response of pore pressure
than G-03-b.
Figure 13(b) shows the variation in pore-water pressure for
G-07 under wave period of 150 sec/cycle. The peak pore-water
pressure of G-07 is also higher than that of G-06. It is under-
standable that the reason is more clay content in G-07. Moreover,
there was no local boiling in the process of testing G-07, as clay
has cohesion to limit soil migration. Thus, clayey soil will not
only reduce the potential of local boiling but also increase the
pore-water pressure response.
-10
-5
0
5
10
15
20
20 30 40 50 60
Time, t (hour)
Pore pressure, u (kPa)
Fig. 10 Envelope of peak pore pressure (P02) of G-02-b
under 300 sec/cycle of wave period
-30
-20
-10
0
10
20
30
40
20.0 20.1 20.2 20.3 20.4 20.5
Time, t (hour)
Pore pressure, u (kPa)
P01 (70 kPa) P02 (70 kPa) P03 (70 kPa)
P01 (140 kPa) P02 (140 kPa) P03 (140 kPa)
(a)
-30
-20
-10
0
10
20
30
40
30.0 30.1 30.2 30.3
Time, t (hour)
Pore pressure, u (kPa)
P01(70kPa) P02(70kPa) P03(70kPa)
P01(140kPa) P02(140kPa) P03(140kPa)
(b)
Fig. 11 Pore-water pressure of silty sand (G-03): (a) 600
sec/cycle; (b) 150 sec/cycle
-30
-20
-10
0
10
20
30
40
40.0 40.1 40.2 40.3 40.4 40.5
Time, t (hour)
Pore pressure, u (kPa)
P01(70kPa) P02(70kPa) P03(70kPa)
P01(140kPa) P02(140kPa) P03(140kPa)
(a)
-30
-20
-10
0
10
20
30
40
20.0 20.1 20.2 20.3
Time, t (hour)
Pore pressure, u (kPa)
P01(70kPa) P02(70kPa) P03(70kPa)
P01(140kPa) P02(140kPa) P03(140kPa)
(b)
Fig. 12 Pore-water pressure of silty sand (G-04): (a) 600
sec/cycle; (b) 150 sec/cycle
Time, t (hour)
Pore pressure, u (kPa)
Time, t (hour)
Time, t (hour)
Pore pressure, u (kPa)
Time, t (hour)
Time, t (hour)
Pore pressure, u (kPa) Pore pressure, u (kPa) Pore pressure, u (kPa) Pore pressure, u (kPa)

Rong-Her Chen et al.: The Filtration Mechanism and Micro-Observation of Soil-Geotextile Systems Under Cyclic Flows 107
-40
-30
-20
-10
0
10
20
30
40
50
35.0 35.1 35.2 35.3
Time, t (hour)
Pore pressure, u (kPa)
P01 (70 kPa) P02 (70 kPa) P03 (70 kPa)
P01 (140 kPa) P02 (140 kPa) P03 (140 kPa)
(a)
-40
-30
-20
-10
0
10
20
30
40
50
60
35.0 35.1 35.2 35.3
Time, t (hour)
Pore pressure, u (kPa)
P01 (70 kPa) P02 (70 kPa) P03 (70 kPa)
P01 (140 kPa) P02 (140 kPa) P03 (140 kPa)
(b)
Fig. 13 Pore-water pressure of clayey-silty sand (at 150 sec/cycle
of wave period): (a) G-06; (b) G-07
0
0.5
1
1.5
2
2.5
3
0 50 100 150 200 250 300
Settlement, S (mm)
G-02-a
G-02-b
G-03-a
G-03-b
600 sec/cycle 300 sec/cycle 150 sec/cycle 75 sec/cycle
(a)
0.0
0.5
1.0
1.5
2.0
0 50 100 150 200 250 300
Settlement, S (mm)
G-04-a
G-05-a
G-05-b
600 sec/cycle 300 sec/cycle 150 sec/cycle 75 sec/cycle
600 sec/cycle 300 sec/cycle (G-04)
150 sec/cycle (G-04) 75 sec/cycle (G-04)
(G-04)
(G-05) (G-05)(G-05) (G-05)
(b)
Fig. 14 Settlement curves of silty sand: (a) G-02 and G-03;
(b) G-04 and G-05
6.2 Settlement
The settlement curves of pure sand (G-01, not shown) are
virtually the same irrespective of normal pressures. As discussed
previously, the difference in pore-water pressure for pure sand
under different normal pressures is negligible. It can thus be con-
cluded that sand can form a structure that particles are not able to
move easily under cyclic flows; consequently, the settlement is
insignificant.
Figure 14(a) presents the settlement curves of silty sand
specimens under different normal pressures. The settlement
under 140 kPa is less than that under 70 kPa, arising from a
higher stress between particles inducing a denser structure as
well as preventing particles from moving. In particular, for
G-02-a, G-02-b and G-03-a, the settlements increase dramatically
during the action of wave period at 300 sec/cycle. Compared
with pore pressure response (e.g., Fig. 10), it is obvious that pore
pressure reduces as settlement occurs. This is because an upward
flow increases pore pressure and reduces the effective stress in
the soil small enough that coarse and fine particles can separate
and migrate away from the geotextile, i.e., a local boiling. As
water flows downwards again, soil particles migrate towards the
geotextile and some fines passing through the geotextile are col-
lected in the wash-out tank.
Furthermore, soil boiling causes fines to suspend on coarse
particles. As water flows downwards, heavier coarse particles
precipitate relatively fast than fines. In this situation, rearrange-
ment of soil particles is so significant that a sudden settlement
occurs until a stable soil structure is formed. After the soil struc-
ture becomes stable, settlement will tend to mitigate. For this
reason, soil boiling is one of the important factors that cause set-
tlement. In addition, comparing the curves of G-04-a and G-05-a
in Fig. 14(b), the settlement increases with increasing amount of
silt because silt is cohesionless thus apt to be washed away.
The variation in settlement for clayey-silty sand is shown in
Fig. 15. From the settlement curves of G-06-a and G-07-a, it is
not surprising that more clay content will combine soil particles
effectively and prevent soil boiling from happening. The other
reason is that the dry density of G-07 is higher than that of G-06
(Fig. 8); hence a denser structure has a less tendency of boiling.
In conclusion, the above results show that overburden pres-
sure as well as soil composition play important roles in the set-
tlement behavior of a soil-geotextile filtration system.
0
0.5
1
1.5
2
2.5
3
0 50 100 150 200 250 300 350
Time, t (hour)
Settlement, S (mm)
G-06-a
G-06-b
G-07-a
G-07-b
Fig. 15 Settlement curves of clayey-silty sand
Time, t (hour)
Settlement, S (mm)
Time,
t
(
hou
r
)
Settlements, S (mm)
Time, t (hour)
Pore pressure, u (kPa)
Time, t (hour)
Pore pressure, u (kPa)
Time, t (hour)
Settlement, S (mm)

108 Journal of GeoEngineering, Vol. 3, No. 3, December 2008
6.3 The Change in the Amount of Fine Soils
In order to justify the mechanism mentioned above, some
soils are taken from three positions in the chamber to evaluate
their particle size distributions. The top position is above the soil
boiling zone; the middle position is within the soil boiling zone;
and the bottom position is in the vicinity of the geotextile. Table
5 shows the amount of fines smaller than 0.074 mm before and
after testing. It can be seen that the percentages of fines in all
specimens have changed after testing. At the bottom position,
some fines are washed away or moved to other places; therefore,
the fines content decreases after testing. Generally speaking, at
the middle and bottom positions the fines content reduces more
for lower normal pressure, except for G-07-a and G-07-b at mid-
dle positions. Moreover, soil boiling makes fines move upwards;
hence the fines content increases at the top position and decreases
at the middle position. For cases where no soil boiling occurs, the
change in the fines content is not significant at both the top and
middle positions. It is clear that soil boiling is the main cause of a
significant settlement when the specimen has small amount of silt
and subjected to low normal pressure.
Table 5 The percentage of fine particle (d < 0.074 mm) at
different positions after testing
Fine particle content (%)
After testing
Test No.
Normal
pressure
(kPa)
Soil
boiling Before
testing Top Middle Bottom
a 70 Yes 5 7.29 4.03 3.94
G-02
b 140 Yes 5 6.26 5.11 4.07
a 70 Yes 10 11.74 6.63 4.01
G-03
b 140 No 10 9.63 8.93 8.12
a 70 No 15 15.89 14.51 11.61
G-04
b 140 No 15 15.47 15.49 14.35
a 70 No 20 22.14 19.26 17.77
G-05
b 140 No 20 21.08 20.44 18.47
a 70 Yes 10 10.68 5.24 5.15
G-06
b 140 No 10 9.84 9.56 8.44
a 70 No 10 9.97 9.59 8.82
G-07
b 140 No 10 10.09 9.42 9.07
Table 6 The mass of soil wash-out per unit area (unit: g/m2)
Fines content (%) Normal pressure (kPa)
Test No.
Silt Clay 70 140
G-01 0 0 79.6 63.2
G-02 5 0 421.0 352.2
G-03 10 0 5096.1 490.6
G-04 15 0 4351.6 513.6
G-05 20 0 5517.5 570.4
G-06 6.5 3.5 389.3 409.2
G-07 3.5 6.5 486.4 465.2
6.4 Soil Wash-Out
After testing, the soil wash-out is collected and the amount
per square meter is shown in Table 6. For pure sand, G-01, there
are only small amounts of soil collected under various normal
pressures. For G-03-a, G-04-a and G-05-a, much more soil are
collected under low normal pressure in comparison to high nor-
mal pressure due to small effective stress having difficulty in
confining particles effectively. Lafleur et al. (1989) defined a
threshold value, 2500 g/m2, to distinguish stable and unstable
filtration systems for base soil. As stated previously, soil boiling
occurs in G-03-a. In this situation, silt soil disperses in the water
and moves more easily; hence particle rearrangement and soil
loss are the primary causes of the settlement in this case.
The effect of clay content on soil wash-out behavior is obvi-
ous, as can be seen from comparing the results of G-03, G-06, and
G-07. Though all three specimens having the same fines content,
10%, the wash-out collected from G-03 is much more. Apparently,
this again is due to the effect of clay content which prevents soil
boiling from occurring in specimens G-06 and G-07.
6.5 Summary
According to the results discussed above, a summary is
made. For pure sand specimen, G-01, it contained no fines and
large particles formed a porous structure that water flew easily
within the soil under cyclic flow action. Consequently, the varia-
tion in pore-water pressure was regular; the settlement was small;
and the least amount of soil was collected. For specimens G-04
and G-05 that contain more than 10% of silty soil, no local soil
boiling were found due to their stable soil structures (Cu > 4 and
Cc = 1 ~ 3). In addition, the test result of specimen G-07 also
shows that the soil structure is stable (Cu > 3 and Cc = 1 ~ 3).
The characteristics of soil specimens that have boiling phe-
nomenon are presented in Table 7. The silt contents in these
specimens are 5 ~ 10%. These specimens have no plasticity and
are classified as SP and SP-SM.
As stated previously, Bhatia and Huang (1995) suggested
soils with Cu < 7 should be considered as internally stable. As
can be seen from Table 7, the Cu values range from 1.8 ~ 2.7 and
Cc values are 0.8 ~ 1.3. In addition to that, all these specimens
satisfy the criteria listed in Table 4. However, soil boiling still
occurred in some conditions, particularly under low normal pres-
sure as presented in Table 7. In view of this, the design criteria for
some soils under cyclic flow condition should be examined more
carefully and take into account the key factors such as silt content,
grain size distribution, and in-situ overburden pressure, etc.
Table 7 The characteristics of soil specimens with boiling
phenomenon
Tes t N o.
Normal
Pressure
(kPa)
Silt
(%)
Clay
(%)
Classification
(USCS) CuCcIp
a 70
G-02 b 140 5 0 SP 1.8 0.8 none
G-03 a 70 10 0 SP-SM 2.6 1.3 none
G-06 a 70 6.5 3.5 SP-SM 2.7 1.3 none

Rong-Her Chen et al.: The Filtration Mechanism and Micro-Observation of Soil-Geotextile Systems Under Cyclic Flows 109
7. MICRO-OBSERVATION OF GEOTEXTILES
After testing, a stereomicroscope is utilized to observe the
clogging condition in the geotextile. Figure 16 shows the surface
of the geotextile after testing. As can be seen, the areas where
marbles locate are darker. In order to look into details, two pieces
of geotextiles are cut for observance by a stereomicroscope.
Fig. 16 Two surface conditions of geotextile after testing (G-04-a)
Figure 17(a) shows the micro phenomenon of the geotextile
surface that contacts with marbles, and it is easy to see fine soils
adhered to fibers. Figure 17(b) shows the area between marbles,
and only a few particles are visible in this clean area.
To explain the two different behaviors, schematic drawings
are made in Fig. 18. When the water flows downwards, soil par-
ticles migrate towards the geotextile with some fines clogged
within the geotextile and some passing through the geotextile
(see Fig. 18(a)). As the water flows upwards, it will pass through
fibers and take away the fines that clogged within the geotextile,
but the particles behind the marbles are difficult to be removed
by the upward flow (Fig. 18(b)). This phenomenon will take
place as cyclic flow continues until a stable soil-geotextile filtra-
tion system is established.
In order to observe the surface condition of geotextile and to
examine the soil structure, a soil-geotextile sample is taken care-
fully from an area between marbles. In Fig. 19, a bridging net-
work is found forming adjacent to the geotextile. In this area,
most particles have been washed away and a hollow space is thus
generated. According to this observation, a network is able to
form under long-term cyclic flow. The network prevents soil
erosion and further settlement. This finding is not the same as
what Giroud (1982) and Köhler (1993) proposed that a bridging
network is unable to form under short-term cyclic flow.
Moreover, Fig. 20 shows the micro phenomenon of the cross
sections of geotextile after testing specimens G-3-a, G-6-a, and
G-7-a. These three soils contain 10% of fines, but have different
clay contents, 0%, 3.5%, and 6.5%, respectively. Their micro
phenomena are quite different. For G-03-a, the fine particles ad-
here to fibers is less than G-06-a and G-07-a. This can be ex-
plained that most fine particles clogged within the geotextile are
eroded due to soil boiling, as more settlement and wash-out are
found in comparison with the other two. On the contrary, G-07-a
containing more clay has more fine particles adhered to fibers.
This phenomenon reduced the voids of geotextile and increased
pore-water pressure. This is the reason why the pore-water pres-
sure of G-07-a is higher than those of G-06-a and G-05-a under
the same wave period condition.
(a)
(b)
Fig. 17 The surface conditions of geotextile (magnification ratio:
25) (G-04-a): (a) the area contacting with marbles; (b)
the area between marbles
(a)
(b)
Fig. 18 Schematic migration of fine particles under cyclic flows:
(a) downwards flow; (b) upwards flow

110 Journal of GeoEngineering, Vol. 3, No. 3, December 2008
Fig. 19 A bridging network formed adjacent to geotextile
(G-06-a)
(a)
(b)
(c)
Fig. 20 Micro photos of the cross sections of geotextile (magni-
fication ratio: 25): (a) G-03-a; (b) G-06-a; (c) G-07-a
8. CONCLUSIONS
This paper presents the investigation on soil-geotextile fil-
tration mechanisms under long-term cyclic flows using a self-
developed apparatus. The soil samples are composed mainly of
sand, with 0 ~ 20% fines content of silt and clay. The factors
taken into account are the effect of overburden pressure and cy-
clic flow periods ranging from long to short, in order to simulate
a range of waves. After testing, a stereomicroscope is utilized to
observe the microstructure of the geotextile filter. The results
from this study are summarized as follows:
1. For the same soil specimen, the peak pore-water pressure in-
creases as the cyclic flow period decreases. The peak pressure
also increases with the amount of cohesive soil in the speci-
men. This is due to the low hydraulic conductivity of cohesive
soil as well as the excess pore-water pressure has no enough
time to dissipate in a short period.
2. The structure of pure sand is relative stable under various
normal pressures. In consequence, the difference in pore-
water pressure, settlement, and the amount of soil loss are in-
significant.
3. For silty sands having 10-20% silt, low normal pressure could
not confine the cohesionless particles effectively. Under these
conditions, the particles in the specimens of G-03-a, G-04-a,
and G-05-a were easily washed away and caused significant
soil loss. However, some clay content in the specimens, such
as G-06-a and G-07-a, as well as higher normal pressure re-
duced soil loss significantly.
4. Boiling phenomenon occurred in specimens that have silt
content less than 10%. Soil boiling induced a sudden settle-
ment, but it did not cause so much soil loss as induced in
specimens with 10-20% silt content.
5. The micro phenomenon observed by the stereomicroscope
showed that a bridging network was able to form under long-
term cyclic flow. This network can prevent soil erosion and
further settlement.
6. Boiling phenomenon occurred in some conditions, particularly
under low normal pressure. Hence, the design criteria for cer-
tain kinds of soils under cyclic flow condition should be ex-
amined more carefully. The key factors such as silt content,
grain size distribution, and in-situ overburden should be taken
into consideration.
ACKNOWLEDGEMENTS
The financial support is provided by the National Science
Council, ROC. Special thanks also go to Dr. Yves-Henri Faure,
LIRIGM, UJF, Grenoble, France for generously supplying the
geotextile and offering valuable suggestions.
NOTATIONS
Basic SI units are given in parentheses.
Cc coefficient of curvature (dimensionless)
Cu coefficient of uniformity (dimensionless)
d
soil particle size (mm)
dx soil particle size corresponding to x percent passing (mm)

Rong-Her Chen et al.: The Filtration Mechanism and Micro-Observation of Soil-Geotextile Systems Under Cyclic Flows 111
d10 effective grain size (mm)
d50 average particle diameter (mm)
emax maximum void ratio (dimensionless)
emin minimum void ratio (dimensionless)
F passing percentage of soil at grain size d (%)
Gs specific gravity of soil (dimensionless)
H increment of passing percentage of soil for grain size
interval of d to 4d (%)
IP plasticity index (dimensionless)
kg hydraulic conductivity of geotextile normal to the plane
(m/s)
ks hydraulic conductivity of soil (m/s)
O90 geotextile characteristic opening size based on hydrody-
namic sieving. (mm)
ML low plasticity silt
S settlement (mm)
SM silty sand
SP poorly-graded sand
SP-SC poorly graded sand with clay
SP-SM poorly graded sand with silt
Tmax tensile strength of geotextile (kN/m)
t time (hour)
tGTX thickness of geotextile (mm)
u pore-water pressure (kPa)
εmax elongation of geotextile (%)
ρd dry density (t/m3)
μA mass per unit area of geotextile (g/m2)
ω water content
ABBREVIATIONS
ASPG Swiss Association of Professional on Geotextiles
CFGG French Committee on Geotextiles
DGEG Germany Committee of Soil Mechanics and Founda-
tions Engineering
LVDT linear variable differential transformer
PIANC Permanent International Association of Navigation
Congresses
USCS Unified Soil Classification System
REFERENCES
ASPG (1985). The Manual of the Geotextiles – Resistance to the
Opening Size Od. Association Suisse des Professionnels de
Géotextiles (Swiss Association of Professionals on Geotextiles),
Zurich, Switzerland (in French).
ASTM D 698-07e1. “Stand test methods for laboratory compaction
characteristics of soil using standard effort (12,400 ft-lbf/ft3
(600 kN-m/m3)).” ASTM International, West Conshohocken,
PA, USA.
Aydilek, A. H. (2006). “A semi-analytical methodology for devel-
opment of woven geotextile filter selection criteria.” Geosyn-
thetics International, 13(2), 59−72.
BAW - Bundesanstalt für Wasserbau (1993). “Code of practice for
the use of geotextile filters on waterways.” The Federal Water-
ways Engineering and Research Institute in Germany, Karlsruhe.
Bhatia, S. K. and Huang, Q. (1995). “Geotextiles filters for internally
stable/unstable soils.” Geosynthetics International, 2(3), 537−565.
Cazzuffi, D., Mazzucato, A., Moraci, N. and Tondello, M. (1999).
“A new test apparatus for the study of geotextiles behavior as
filters in unsteady flow conditions: relevance and use.” Geotex-
tiles and Geomembranes, 17(5-6), 313−329.
CFGG (1986). AFNOR G38017, Association Française de Nor-
malisation (French Committee on Geotextiles), La Plaine Saint-
Denis, France (in French).
Chen, R. H. and Chen, C. N. (1986). “Permeability characteristics of
prefabricated vertical drains.” Proceedings of the 3rd Interna-
tional Conference on Geotextiles, Vienna, Austria, 3, 785−790.
Chen, R. H., Ho, C. C. and Hsieh, A. T. (2006). “Filtration behavior
of geotextile under normal pressure and bi-directional cyclic
flow.” Proceedings of the 8th International Conference on Geo-
synthetics, Yokohama, Japan, 2, 557−560.
Chen, R. H., Ho, C. C. and Hsu, C. Y. (2008). “The effect of soil
fines content on the filtration characteristics of geotextile under
cyclic flows.” Geosynthetics International. 15(2), 95−106.
Chew, S. H., Tian, H., Tan, S. A. and Karunaratne, G. P. (2003).
“Erosion stability of punctured geotextile filters subjected to cy-
clic wave loadings─A laboratory study.” Geotextiles and Ge-
omembranes, 21(4), 221−239.
DGEG (1986). “Use and examination of civil engineering and hy-
draulic engineering.” Germany Committee of Soil Mechanics
and Foundation Engineering, Series of publications of the
DVWK 76, Hamburg, German.
DS EN 964-1 (1996). “Geotextiles and geotextile-related products─
Determination of thickness at specified pressures─Part 1: sin-
gle layers.” Dansk Standard, Charlottenlund, Denmark.
DS EN 965 (1996). “Geotextiles and geotextile-related products─
Determination of mass per unit area.” Dansk Standard, Charlot-
tenlund, Denmark.
EN ISO 10319 (1993). “Geotextiles─Wide-width tensile test.” In-
ternational Standardization Organization, Geneva, Switzerland.
EN ISO 11058 (1999). “Geotextiles and geotextile-related products
─Determination of water permeability characteristics normal to
the plane, without load.” International Standardization Organi-
zation, Geneva, Switzerland.
EN ISO 12956 (1999). “Geotextiles and geotextile-related products
─Determination of the characteristic opening size.” Interna-
tional Standardization Organization, Geneva, Switzerland.
Fannin, R. J. and Pishé, R. (2001). “Testing and specifications for
geotextile filters in cyclic flow applications.” Proceedings of
Geosynthetics 2001, Portland, Oregon, U.S.A., 423−435.
Giroud, J. P. (1982). “Filter criteria for geotextiles.” Proceedings of
the 2nd International Conference on Geotextiles, Las Vegas,
USA, 1, 103−108.
Giroud, J. P., Delmas, P. and Artières, O. (1998). “Theoretical basis
for the development of a two-layer geotextile filter.” Proceed-
ings of the 6th International Conference on Geosynthetics, At-
lanta, USA, 2, 1037−1044.
Heerten, G. (1982). “Dimensioning the filtration properties of geo-
textiles considering long-term conditions.” Proceedings of the
2nd International Conference on Geotextiles, Las Vegas, USA,
1, 115−120.
Ho, C. C. (2007). The Erosion Behavior of Revetment Using Geotex-
tile. Ph.D. Dissertation, National Taiwan University (Taiwan)
and Joseph Fourier University of Grenoble (France).
Hoare, D. J. (1984). “Geotextiles as filter.” Ground Engineering, 17,
29-44.

112 Journal of GeoEngineering, Vol. 3, No. 3, December 2008
Hsu, C. Y. (2007). “Filtration behavior of geotextile under
bi-directional cyclic flow.” Master Thesis, National Taiwan
University (in Chinese).
Kenney, T. C. and Lau, D. (1985), “Internal stability of granular
filters.” Canadian Geotechnical Journal, 22(2), 215−225.
Köhler, H. J. (1993). “The influence of hydraulic head and hydraulic
gradient on the filtration process.” Proceedings of the 1st Inter-
national Conference on Geofilters, Karlsruhe, Germany, 2,
225−240.
Lafleur, J., Mlynarek, J. and Rollin, A. L. (1989). “Filtration of
broadly graded cohesionless soils.” Journal of Geotechnical
Engineering, ASCE, 115(2), 1747−1768.
Lawson, C. R. (1982). “Filter criteria for geotextile: relevance and
use.” Journal of the Geotechnical Engineering Division, ASCE,
108(10), 1300−1317.
Liao, H. J. and Chu, C. C. (2002). “The case study of backfill soil
erosion in the tidal land of Taiwan.” http://www.ceci.com.tw/
book/53/ch53hp.htm, CECI-Technology Magazine, 53, Taipei,
Taiwan (in Chinese).
Luettich, S. M., Giroud, J. P. and Bachus, R. C. (1992). “Geotextile
filter design guide.” Geotextiles and Geomembranes, 11(4-6),
355−370.
Młynarek, J. (2000). “Geodrains and geofilters─Retrospective and
future trends.” Proceedings of the 3rd International Conference
on Geofilters, Warsaw, Poland, 1, 27-47.
Młynarek, J., J. Bogumił L., Andrél, R. and Gilles, B. (1991). “Soil
geotextile system interaction.” Geotextiles and Geomembranes,
10(2), 161−176.
Palmeira, E. M. and Fannin, R. J. (2002). “Soil-geotextile compati-
bility in filtration.” Proceedings of the 7th International Con-
ference on Geosynthetics, Nice, France, Keynote Lecture, 3,
853−869.
PIANC (1987). “Guidelines for the design and construction of flexi-
ble revetments incorporating geotextiles for inland waterways.”
Supplement to Bulletin 57, Permanent International Association
of Navigation Congresses.
Rollin, A. and Lombard, G. (1988), “Mechanisms affecting
long-term filtration behavior of geotextiles.” Geotextiles and
Geomembranes, 7(1-2), 119−145.
Scheidegger, A. E. (1957), The Physics of Flow Through Porous
Media. The MacMillan Co., NY.
Schober, W. and Teindl, H. (1979). “Filter criteria for geotextiles.”
Proceedings of the 7th European Conference on Soil Mechanics
and Foundation Engineering, Brighton, UK, 2, 121−129.