1

LARGE-SCALE PHYSICAL MODEL TESTS ON SAND-FILLED

GEOTEXTILE TUBES AND CONTAINERS UNDER WAVE ATTACK
P. van Steeg
1
, E. Vastenburg
1
, A. Bezuijen
1
, E. Zengerink
2
and J. de Gijt
3

Geotextile encapsulated sand elements, such as geotextile tubes or geotextile containers, are considered
more and more as an alternative material for coastal protection. Uncertainty with respect to the stability
of geotextile elements under wave attack is one of the reasons why these systems are not applied widely
yet. Therefore, large-scale physical model tests focusing on the stability of geotextile containers and
tubes under wave attack have been performed in the Delta Flume of Deltares. The first part focuses at
the stability of geotextile containers under wave attack; the second part describes the stability of
geotextile tubes under wave attack. For the geotextile containers, observed failure mechanisms are
described and analysed qualitatively. For the geotextile tubes, a physical sound formula based on a
theoretical approach and verified with large-scale physical model tests is suggested for both stability and
deformation (settlement) under wave attack. It was concluded that Froude scaling is not suitable for
geotextile containers (due to sand migration within the containers) but can be applied for tubes with a
relatively high filling degree.
INTRODUCTION
Geotextile encapsulated sand elements, such as geotextile tubes, geotextile con-
tainers, geotextile bags or geotextile mattresses can be used in several hydraulic
applications. Examples and suggestions are given in CUR (2004, 2006), Pilarc-
zyk (2000), and Oh and Shin (2006). However, the elements are not regularly
used for coastal defence works. One of the main reasons given in Bezuijen and
Vastenburg (2008) is the uncertainty on the behaviour of these elements under
wave loads. Therefore, this behaviour is studied by conducting large-scale phys-
ical model tests on geotextile tubes and geotextile containers under wave attack.
Formulas to predict the stability are scarce. In Pilarczyk (2000) some formulas
are given but these are based on small-scale physical model tests. The research
described in this paper leads to a formula, based on large-scale physical model
tests, providing a dimensionless stability number for geotextile tubes under
wave attack and a qualitative description of geotextile containers under wave
attack.
REVIEW OF STABILITY CONCEPTS OF GEOTEXTILE TUBES AND
CONTAINERS
Nine failure mechanisms of geotextile elements under wave attack are
discussed in Lawson (2008). Sliding instability, overturning instability, bearing


1
Deltares, P.O. Box 177, 2600 MH Delft, The Netherlands, [email protected],
[email protected], [email protected]
2
TenCate Geosynthetics, P.O. Box 236 Almelo, The Netherlands [email protected]
3
Delft University of Technology / Gemeentewerken Rotterdam, P.O. Box 6633, 3000 AP Rotterdam,
[email protected]


2
instability, global instability, scour of foundation, and foundation settlement are
related to external loads. Geotextile rupture, erosion of fill through the
geotextile, and deformation of contained fill are internal failure mechanisms.
An important, but almost forgotten, failure mechanism is given in Venis (1968).
He performed several tests using various sizes of sandbags under current attack.
He concluded that, from a certain size, the size at where the sandbag started to
shift was independent of the model scale. At a certain critical velocity the sand
in the bags started to migrate which led to instability of the sandbags. Venis
(1968) concluded that the Froude law scaling is not applicable related to
situations where sand migration within the bags occurred. An illustration is
given in Fig. 1.

Start of sand transport
caterpillar mechanism
sliding
L
ucrit,cp
Froude law
u
crit


Figure 1. Stability as function of element size (represented by L) according to Venis
(1968).
Several formulas based on physical small-scale model tests are summarized in
Pilarczyk (2000) and CUR (2006). For stability, usually the failure mechanisms
sliding instability or overturning instability are considered. The used
dimensionless stability numbers are usually based on Froude scaling laws such
as H
s
/(D) =C or H
s
/(B) =C where H
s
=significant wave height, B =width
of element, D =height of element, =relative density (
sand
-
water
)/
water
, =
density and C =constant. In Recio and Oumeraci (2009), stability formulas
including deformation of the elements are reported.
TEST SET-UP AND TEST RESULTS
General
Large-scale physical experiments on the stability of geotextile tubes as well as
on geotextile containers under wave attack have been performed in the Delta
Flume of Deltares. This flume has a width of 5 m, a length of 235 m and a
depth of 7 m. Irregular waves, with a JONSWAP spectrum and a wave
steepness based on the peak wave period of s
0,p
=0.03 were generated at the
wave board during each test. The number of waves during each test was
approximately N =1000. In case damage occurred, the test was aborted. If no
damage occurred, a new test with a higher wave height was started. During all
stable
unstable


3
test series a Geolon

PE180L geotextile was used, thus only woven geotextiles

were tested. The opening size of the geotexile pores was O
90
=0.170 mm.
Geotexti le contai ners
The width of the containers B was 2.75 m, the length L (perpendicular to the
direction of the flume) was 5 m (the same as the flume width) and the average
thickness D was 0.55 m. The filling percentage p
A
, defined as actual fill divided
by maximum fill possible with the given circumference of the container, was
approximately 44 %. A stack of containers was placed on an existing 1:3
concrete slope. The seaward slope of the containers was 1:2. Above the
containers, that formed a berm with a width of approximately 6 m, a smooth
1:3 concrete slope was present. The test program consisted of two subsets, both
with a different water level. During the first test series, the water level was 0.75
H
s
above the top of the upper container. During the second test series, the water
level was equal to the top of the upper container. Measurements that were
performed are (i) wave measurements (wave gauges), (ii) profile measurements
of the structure (mechanical profiler), (iii) registration of water movement in
the container (electromagnetic velocity meter) and (iv) resistance measurements
(as measure for the density of the fill material) by pressing a cone through the
sand before and after each test (Penetrologger).
Table 1. Test program geotextile containers.
Test Rc/Hs (-) Hs (m) Tp (s) N (-) Hs(BD) (-) HsD (-) damage
t 1-1a 0.75 1.05 4.57 1015 0.86 1.91 minor reshaping
t 1-2 0.75 1.19 5.09 1015 0.97 2.16 damage
t 2-1 0 0.76 3.97 1014 0.62 1.38 no reshaping
t 2-2 0 0.92 4.35 973 0.75 1.67 minor reshaping
t 2-2a 0 0.90 4.32 2084 0.73 1.64 severe reshaping
t 2-2b 0 0.90 4.32 2124 0.73 1.64 damage
t 2-3 0 1.08 4.67 1011 0.88 1.96 damage
t 2-4 0 1.21 5.05 1036 0.98 2.20 damage
t 2-5 0 1.34 5.36 1038 1.09 2.68 damage


4.25 m
1:3
1:2
4.25 m
3.15 m
1:2
1
0.75Hs

Figure 2. Schematized test set-up of geotextile containers; test set-up prior to test-run;
test-run.
The test program is given in Table 1. In the first test series (t1: R
c
/H
s
=0.75),
two tests were performed. During Test t1-2 damage occurred. In the second test
14
12 & 13
11
8
6 & 7
4
2 & 3
1


4
series (t2: R
c
/H
s
= 0) seven tests were performed. During Test t2-2, little
horizontal displacement of some containers was observed. To determine
whether this was a time-based process, the test was repeated two times as Test
t2-2a and Test t2-2b. It was observed that the processes of reshaping and
deformation was indeed time-based.
Geotexti le tubes
The tested geotextile tubes were placed on a plateau. This plateau was made of
concrete, had a horizontal floor located 3.60 m above flume bottom and a sea-
ward slope angle of 1:2.5. Eight different configurations were tested. In five
configurations (F1, F3, F4a, F4b and F5), single placed tubes with varying
filling percentages and sizes were tested. Configuration T1 consisted of a single
tube with a bar placed at the landward side of the tube to simulate a trench or
settlement due to the weight of the installed tube. Configuration P2 consisted of
two tubes placed behind each other. Configuration P3 consisted of a so-called
2-1 stack; two tubes placed behind each other with a third tube on top. At the
landward side of configuration P3, a bar was placed. J ust like in practice, the
prefabricated tubes were filled hydraulically with a sand-water mixture in-situ.
To reduce the friction between installed tube and concrete flume walls, smooth
wooden plates were attached to the flume walls. An impression of the filling
process is given in Fig. 4 (a demonstration beside the wave flume). The dimen-
sions of the prefabricated (empty) tubes were determined based on the
theoretical shape using the Timoshenko method as given in CUR (2006). It
turned out that the calculated shape of the tube matched remarkably well with
the measured shape of the in-situ filled tube. Measurements that were carried
out included wave measurements (wave gauges), profile measurements
(mechanical profiler), displacement measurement (using overlay photographs)
and sand migration (by colouring the sand in the tube with injections and
visually inspect these after testing).
Table 2. Overview tested configurations geotextile tubes.
D(m) B (m) a
1
(m) R100% pA (%)
F1 single tube, low filling rate 0.57 2.19 1.12 0.75 66
F3 single tube 0.79 2.04 1.11 0.75 80
F4a single tube, high filling rate, close to seaside 0.82 1.52 0.85 0.57 109
F4b single tube, high filling rate 0.82 1.52 1.22 0.57 109
F5 single tube 0.74 2.02 0.91 0.76 72
T1 single tube, bar at landward side 0.88 2.03 1.03 0.76 85
P2 two tubes 0.84 1.99 0.86 0.77 77
P3 2-1 stack, bar at landward side 0.70 1.41 0.65 0.57 91
1)
ais distance between most seaward point of tube and seaward side of horizontal part of supporting plateau


5
F4a
F5
T1
P3
P2
F1
F3
F4b

Figure 3. Tested configurations for geotextile tubes.

Figure 4. Filling of tube (demonstration besides flume); Wave impact on tube during
test; analysis of sand migration (visual inspection after test by opening tube).
All single placed tubes (F1, F3, F4a, F4b, F5 and T1) failed due to sliding
instability (in landward direction). For the tubes with a lower filling percentage
(e.g. F1), some sand migration, based on the color injections, were observed
leading to minor deformation of the tube. For all configurations, some
settlement of the tubes was observed. For Configuration P2, failure occurred
due to sliding instability of the landward tube in landward direction. For
Configuration P3, global instability occurred (a slip circle in seaward direction).
ANALYSIS GEOTEXTILE CONTAINERS
During the tests a combination of four failure mechanisms was observed. These
are:
1. Erosion of fill through geotextile.
2. Global instability.
3. Sliding instability.
4. Caterpillar Mechanism.
The four failure mechanisms are described below:


6
- Erosion of fill through geotextile: Based on the sieving curve of the sand
and opening sizes of the geotextile, 27 % percent of the sand could
theoretically move through the geotextile (assuming no natural filter).
Actual erosion of fill, based on comparison of the profiler measurements
before and after the test series, is estimated between 0 % and 8 %.
- Global instability: a slip circle containing containers 2, 3, 4, 6, 7 and 8
(see Fig. 2) was visually observed.
- Sliding instability: sliding of several containers was observed.
- Caterpillar Mechanism: The measured deformation was large (based on
profiler measurements and visual observation) which is largely due to
sand movement within the containers. Penetrologger measurements
showed a lower packing density after the tests and velocities of more than
1 m/s were measured at several locations within the containers indicating
internal sand movement. It was observed that the deformation led to a
new mechanism indicated as Caterpillar Mechanism. This Caterpillar
Mechanism is not fully understood but it was concluded that moving sand
led to a mechanism that can be compared with the rolling of a caterpillar.
It was not possible to quantify the contributions of the four observed
mechanisms; the combination of observed physical processes was too complex
and the processes influenced each other too much. Based on observations and
described processes of Venis (1968), it is very likely that the sand movement
inside the containers is not dependent on the size of the containers (assuming
not very small experimental set-ups). This implies that the scaling with respect
to Caterpillar Mechanism is 1:1. In other words, in a prototype situation, start
of displacement would occur at the same wave height as in the large-scale
experimental set-up in the Delta Flume. The stability of containers in this test
set-up is significantly lower than the stability of containers in earlier small-
scale model tests. The difference in stability is explained due to the following
two aspects: (i) The migration of sand caused a caterpillar mechanism, which
contributed significantly to the instability of the containers; (ii) The presence of
a 1:3 smooth slope above the containers causes a severe wave run-down, which
might have affected the stability of the containers negatively.
ANALYSIS: STABILITY SINGLE PLACED GEOTEXTILE TUBES
Theoretical approach
For all configurations, except Configuration P3, horizontal sliding instability
was the normative failure mechanism. Usually this mechanism is described
using dimensionless stability numbers H
s
/(D) or H
s
/(B). In Deltares (2010), a
theoretical derivation is described which led to the following equation:

2
( cos sin )
( )
rep
D L
H
f
C fC BD

o o

<
+ A
(1)


7
Eq. (1), with representative wave height H
rep
, shape factor =B/D, wave
velocity coefficient , drag coefficient C
D
, lift coefficient C
L
, friction coefficient
f and slope angle , forms the physical basis for the chosen dimensionless
stability number. Assuming no severe wave run-down the can be replaced
by a +. Since the shape factor is a function of the width B and height D of
the tube ( =f(B, D)), the stability number H
rep
/((BD)) is, from a theoretical
point of view, not a solid stability number. However, for engineering purposes,
this stability number turns out to be very useful as will be demonstrated below.
All parameters of Eq. (1) are described in the following sections.

Representative wave height H
rep
. Breaking waves results into transmitted
wave energy E
t
, and so-called blocked energy E
b
, consisting of dissipated and
reflected energy. The blocked energy can be splitted into energy components on
the tube E
b,tube
, and on the structure underneath E
b,struc
. Since the interest lies in
the wave energy on the tube, the following equation with reduction parameter ,
representtative wave height H
rep
, significant incident wave height H
s
, blocked
wave energy on tube E
b,tube
, and incident wave energy E
i
is suggested:

, rep b tube
s i
H E
H E
_ = = (2)
The energy components, with transmission coefficient for a situation with only
underneath structure C
tr,struc
, and transmission coefficient for a situation with
underneath structure and tube C
tr,tube&struc
are given by:
With tube:
, , , & i b tube b struc tr tube struc
E E E E = + + with
2
, & , & tr tube struc tr tube struc i
E C E = (3)

Without tube:
, , i b struc tr struc
E E E = + with
2
, , tr struc tr struc i
E C E = (4)

Combining Eq. (2), Eq. (3) and Eq. (4) gives:

2 2
, , tr struc tr struc tube
C C _
+
= (5)
The transmission coefficient can be determined in various ways. In this analysis,
the method described in Van der Meer et. al. (2003) is applied. A graphical
representation, which is restricted to a situation with perpendicular wave attack
(wave angle = 0) and a water level equal to the level of the highest point of
the tube (crest height R
c
=0), of this approach is derived and given in Fig. 5.
The representative wave height H
rep
of Eq. (1) is described by:

rep s
H H _ = (6)



8

0.0
0.2
0.4
0.6
0.8
1.0
0 0.5 1 1.5 2 2.5 3 3.5 4
H
s
/D
X
p =1
p =2
p =3

Figure 5. Design graph for reduction coefficient (applicable for Rc = 0, = 0).
Friction coefficient f, slope angle . The friction coefficient f between the tube
and its foundation layer can be determined by performing friction tests as
described in Deltares (2010). The representative slope angle is known for
each situation and can be determined as described in Deltares (2010).

Drag coefficient C
D
, lift coefficient C
L
, wave velocity coefficient . In case of
no lift forces, lift coefficient C
L
is equal to zero. Eq. (1) is rewritten by:

2
1
( cos sin )
D
H
f
B C
o o

<
A
(7)
In case of no drag forces, drag coefficient C
D
is equal to zero. Eq. (1) is
rewritten by:

2
1
( cos sin )
L
H
f
D f C
o o

<
A
(8)
Literature gives some values for drag coefficient C
D
, lift coefficient C
L
(e.g.
Recio, 2008) and wave-velocity coefficient (according to Pilarczyk (2000): =
1-1.5). Since it is unlikely that drag coefficient C
D
, lift coefficient C
L
and wave-
velocity coefficient will be determined for each specific situation and the fact
that the friction coefficient f has not much influence since the lift forces are
relatively small (fC
L
), these parameters are collected within dustbin
coefficient C
x
which can be determined based on experiments. Eq. (1) is
rewritten by:
Onlydrag
,1 1
( cos sin )
s
s
H
N C
B f
_
o o
= <
A +
with
1 2
1
D
C
C
= (9)
Onlylift
,2 2
( cos sin )
s
s
H
N C
D f
_
o o
= <
A +
with
2 2
1
L
C
f C
= (10)
Drag and lift
,3 3
( cos sin )
s
s
H
N C
BD f
_
o o
= <
A +
with
3 2
( )
D L
C
C fC

=
+
(11)


9
Test results as function of deri ved dimensionl ess stabi li ty numbers (Ns,x)
Test results as function of three dimensionless stability numbers are given in
Fig. 6. On the horizontal axis, dimensionless stability number N
s,1
(upper
graphs), N
s,2
(middle graphs) or N
s,3
(lower graphs) is given. On the vertical
axis, displacement per test, made dimensionless with width of tube (x/B) (left
graphs), and cumulative displacement, made dimensionless with width of tube
(x
cum
/B), (right graphs) are given.

Figure 6. Test results are function of derived dimensionless stability numbers.
In the upper graphs of Fig. 6, it can be seen that there is a data collapse
indicating that the chosen dimensionless stability number (based on drag forces
F
D
and no lift forces F
L
, tube height D has no influence) is a representative
stability number. However, Configuration F1 (tube with low filling percentage)
seems to be an outlier. Probably this is caused by lift forces, which played a
significant role for this specific configuration, where the tube was relatively
thin and wide. In the middle graphs, it can be seen that there is almost no data
collapse indicating that the used dimensionless stability number (based on lift
forces F
L
and no drag forces F
D
, tube width B has no influence) cannot be


10
applied. In the lower graphs, data collapse can be seen clearly, indicating that
the chosen dimensionless stability number (based on drag forces F
D
and lift
forces F
L
, width B and height D both included) is a proper dimensionless
stability number. The only outlier is Configuration F4a. This can be explained
due to the position of the tube at its supporting foundation layer. The tube of
Configuration F4a was in a (relatively) more seaward position than the tubes
during other configurations. Assuming an accepted cumulative sliding distance
of 5% of the tube width (B), the following equation is suggested:
0.65
( cos sin )
s
H
BD f
_
o o
<
A +
(12)
Eq. (12) is verified by the physical model tests under the following conditions:

Crest height relative to water level R
c
= 0 m
Breaker parameter
p,toe
= 2
Friction coefficient f = 0.55
Angle of foundation layer 0
o
< < 5
o
Filling percentage 66 % < p
A
< 109 %
Radiusof tubewhen100%filled 0.57 < R
100%
< 0.75
Characteristic distance 1.2 < a/D < 2.0
Accepted relative sliding distance x
cum
/B < 0.05
ANALYSIS STABILITY MULTIPLE PLACED TUBES
Installing two tubes behind each other (Configuration P2) does not significantly
increase the stability. The tube at landward side started to shift due to hydro-
static pressure caused by water entrapped between the two tubes and hydrostatic
pressure caused by the wave action at seaward side of the tubes. Therefore, the
stability relation as given in Eq. (12) should also be used for a structure consis-
ting of two tubes placed behind each other. Width B should then be chosen as
the width of a single tube.
Placing two tubes behind each other with a third tube on top (Configuration P3),
resulted into a so-called slip circle. In the performed tests, the slip circle in
landward direction was blocked by applying a fixed bar at the landward side of
the tube. At the seaside, such a bar was not placed resulting in a slip circle in
seaward direction. A stability analysis based on acting forces on the stack of
tubes is carried out and described in Deltares (2010). The calculated results
approximated the results obtained from the physical model tests.
ANALYSIS SETTLEMENT OF GEOTEXTILE TUBES
Based on the tests with a single placed tube, the relative deformation is deter-
mined and given in Fig. 7. On the horizontal axis, filling percentage p
h

(defined as actual height divided by the height of a tube which is maximum
filled) and filling percentage p
A
(defined as the actual fill area divided by the


11
maximum fill area that is possible given the circumference of the geotextile) are
given. On the vertical axis, relative vertical deformation of the tube S
tube
, based
on initial height D
ini
and height after test program D
aft
, is given.

aft ini
tube
ini
D D
S
D

= . (13)
0
4
8
12
16
0 20 40 60 80 100 120
filling percentage p(%)
s
e
t
le
m
e
n
t

o
f

t
h
e

t
u
b
e
,

S
tu
b
e

(
%
)
filling percentage ph
filling percentage pA
Equation 14
Equation 15

Figure 7. Relative vertical deformation of the geotextile tubes as function of pA and ph.
Based on regression analysis the following formulas are given:
0.41 32.3
tube h
S p = + (14)
0.28 32.6
tube A
S p = + (15)
Eq. (14) and Eq. (15) are applicable for same conditions as given for Eq. (12).
CONCLUSIONS
Due to the Caterpillar Mechanism, geotextile containers become unstable when
located in the breaker zone. Therefore, it is recommended to avoid the use of
containers in the zone where waves break (between SWL and 2H
s
below sea
level) without applying proper measures (e.g. protection cover). In case internal
sand migration is possible, Froude scaling is not applicable and dimensionless
stability numbers such as H
s
/(D) and H
s
/((BD)) cannot be used.
For tubes, which usually have a relatively high filling degree, Froude scaling is
applicable. A physical sound dimensionless stability number has been derived
(Eq. (11)) to determine the stability against sliding. This dimensionless stability
number includes the significant wave height H
s
, tube height D, tube width B,
relative density of the tube , reduction parameter to include wave energy
transmission , slope angle between tube and supporting foundation layer ,
and friction between tube and foundation layer f. Eq. (12) (stability) and Eq.
(14) and Eq. (15) (settlement) can be used as a first approximation for the
design of structures consisting of geotextile tubes under wave attack. However,
for specific projects where geotextile tubes are applied in a wave-breaking zone
(and where it is very likely that physical conditions will be different from those


12
in the described experiments), it is recommended to perform additional
physical model tests to study the stability for specific characteristics of that
structure. The methodology described in this paper can serve as a basis for
design of those physical model tests and interpretation of test results.
ACKNOWLEDGEMENTS
The study on the stability of geotextile containers was sponsored by Delft
Cluster. The contribution to the physical model on the stability of geotextile
tubes by Ten Cate and De Vries & van de Wiel is very much acknowledged.
The contribution of Mark Klein Breteler was essential to produce Deltares
(2008) and Deltares (2010), which forms the foundation of this paper.
REFERENCES
Bezuijen, A., E. Vastenburg, 2008. Geosystems, possibilities and limitations for
applications. EuroGeo 4, paper nr. 282.
CUR, 2004. In Dutch: CUR 214: Geotextile zandelementen (CUR 214:
Geotextile encapsulated sand elements), Stichting CUR, Gouda.
CUR, 2006. In Dutch: CUR 217: Ontwerpen met geotextiele zandelementen
(CUR217: Design with geotextile encapsulated sand elements), Stichting
CUR, Gouda.
Deltares, 2008. Large scale physical model tests on the stability of
geocontainers, Deltares report H4595, May 2008, Delft.
Deltares, 2010. Large scale physical model tests on the stability of geotextile
tubes, Deltares report 1200162-000, February 2010, Delft.
Lawson, C.R., 2008. Geotextile containment for hydraulic and environmental
engineering, Geosynthetics International, 15, No 6, 384-427.
Oh, Y.I., E.C. Shin, 2006. Using submerged geotextile tubes in the protection
of the E. Korean Shore, Coastal Engineering 53, 879-895, Elsevier.
Pilarczyk, K., 2000. Geosynthetics and geosystems in hydraulic and coastal
engineering, 2000, Balkema Rotterdam.
Recio, J., H. Oumeraci, 2009. Process based stability formulea for coastal
structures made of geotextile sand containers Coastal Engineering, 56 (5),
p. 632-658, May 2009.
Van der Meer, J.W., B. Wang, A. Wolters, B. Zanuttigh, M. Kramer 2003
Oblique wave transmission over low crested structures Proc 4
th
Int. Coastal
Structures Conf, Portland, OR, 26-30 Aug 2003. ASCE, reston, VA, pp567-
579.
Venis, W.A., 1968. Closure of estuarine channels in tidal region, Behaviour of
dumping material when exposed to currents and wave action, De
Ingenieur, 50, 1968.


13
KEYWORDS CSt2011
p0021
LARGE-SCALE PHYSICAL MODEL TESTS ON SAND-FILLED
GEOTEXTILE TUBES AND CONTAINERS UNDER WAVE ATTACK
Van Steeg, Paul
Vastenburg, Erik
Bezuijen, Adam
Zengerink, Edwin
de Gijt, J arit

Caterpillar Mechanism
Container
Delta Flume
Geotextile
Physical model
Stability
Tube
Wave attack