Geotech Geol Eng
DOI 10.1007/s10706-014-9825-9
ORIGINAL PAPER
Assessment of the Foundation of Tunisia Ghezala Dam
H. Karoui • M. Bouassida
Received: 11 January 2014 / Accepted: 8 October 2014
Springer International Publishing Switzerland 2014
Abstract Ghezala Earth Dam was built in the city of
Mateur in the North West area of Tunisia. The filling
of the dam reservoir, with a storage capacity close to
one hundred thousand cubic meters, started in 1985
and lasted for 2 years. The dam foundation is essentially composed of consistent clayey marl layer. At the
lower level of the dam, reinforced concrete drainage,
gallery was constructed to evacuate water, when
needed, from the upstream side to downstream side.
The gallery structure is composed of eleven 15 m
length sections. During its service life (more than
26 years), an installed health monitoring system
permitted, in particular, the measurement of consolidation settlement which has occurred in the soil foundation. This follow up system made it possible
to observe the cracks which mainly developed progressively in the central portion of the drainage gallery
due to differential settlements. In this context, a 2D
numerical simulation of the behavior of dam foundation was performed by adopting the Mohr–Coulomb
model for the dam’s material and the soil foundation.
Numerical predictions of the consolidation settlement
development were compared to those measured in situ
and a good agreement was found between the two.
H. Karoui M. Bouassida (&)
Université de Tunis El Manar, Ecole Nationale
d’Ingénieurs de Tunis, LR14ES03, Ingénierie
Géotechnique, BP 37 Le Belvédère, 1002 Tunis, Tunisia
e-mail: mounir.bouassida@enit.rnu.tn
This finding highlights the effectiveness of the
adopted model for the studied case.
Keywords Settlement Gallery Disorders
Elastoplastic Consolidation Piping
1 Introduction
Detecting and analyzing the mechanisms of degradation that affect the structural health of dams is very
important especially for safety reasons. The end result
of the assessment is a list of recommended corrective
actions to maintain dams in normal functioning
conditions. Such an assessment cannot be performed
without the installation of in situ measurement devices
such as: piezometers, inclinometers, settlement
gauges, etc.
The collection of measurements over time, their
treatment and their interpretation will enable to
understand the functioning of the dam (Arfaoui 2007).
In Tunisia 26 dams were built from 1928 to date.
The assessment of these dams enabled the detection of
degradations and allowed decision makers to adopt the
best procedures for their maintenance and rehabilitation. Ghezala dam, object of this paper, is one of such
dams; its drainage gallery was affected by severe
cracks due to differential settlements.
This paper presents a model that was used to predict
the consolidation settlement of the dam foundation
123
Geotech Geol Eng
that are believed to be main causes for the development of severe cracks and significant joints’ opening
in the concrete drainage gallery observed at the
downstream side. The paper layout is as follows: first,
a presentation of Ghezala dam and the measured
developed settlements are provided. Then, a numerical
analysis, using plane strain conditions, is described.
This analysis was carried out to simulate the observed
behavior of the dam foundation. A comparison
between the in situ measurements and the predicted
results obtained by Plaxis finite element software
(version 8) is presented. This comparison gives a
helpful insight explaining the mechanism which had
occurred during the operation of Ghezala dam over its
26 years of service life. Finally, the conclusions of the
paper are drawn and presented.
2 Presentation of Ghezala Dam
Ghezala dam was built from 1981 to 1984 as a
homogeneous earth dam, 31 m in height and 560 m in
length. It is founded on a compressible marl formation
resting on calcareous stratum that is considered as an
impervious and non-compressible layer. The elevation
of High Water Level is 86.1 m (GNT Tunisian
system). Ghezala dam provides a storage capacity of
approximately one hundred thousand cubic meters
devoted to the irrigation of 900 hectares of agriculture
lands. A reinforced concrete drainage gallery,
installed in 1983, crosses the dam structure. It is
composed of eleven rectangular elements that are
15 m length, 2 m 9 2.5 m cross-section, and 0.6 m
thick. The gallery is located 25 m deep under the
elevation of full reservoir (Fig. 1). By the end of dam
construction, a settlement of 200 mm has been
measured in the central part of the gallery. In four
portions (60 m in length), several openings of separation joints (J3 to J8) and transverse cracks were
observed as sketched in Fig. 1. Among the ten main
cracks (C1 to C10) six of them have an opening greater
than 10 mm (Fig. 1). In post first fill-up of the dam
reservoir in 1985, repairing works of cracks have been
executed in the gallery structure. Later on, a progressive settlement development has resulted from the
triggered piping of estimated amount by 1 kg per week
in 2003 of soil material deposited at the downstream
side (Arfaoui 2007). The maximum settlement of
455 mm was measured in 2008 under the crest of the
123
dam (Fig. 2). Piping phenomenon is caused by water
infiltration from the gallery within the dam foundation; it induces the transport of particles at the exit
point of the downstream side.
From August to October 1985, the first repairing
activities were performed immediately before the first
filling of the reservoir. A bentonite grout injection was
used as a primary repairing action.
The progressing development of settlement during
the following years induced the opening of cracks. For
this reason second repairing activities were undertaken from 1992 to 1993. This repair mainly consisted
in the treatment of opened cracks by using water
swelling product prepared from a monomer aqueous
solution. The additional recorded movements were
approximately 120 mm. The piping phenomenon
induced transport of sediments observed at crack F3
triggered an emergency repairing action. This showed
that the second repairing activities were unsuitable to
stop the piping phenomenon; the decision makers
favored the installation, at crack locations, of PVC
drains surrounded by geotextile filter which allow the
water drainage and prevent the fines transportation.
This solution efficiently aided in stopping transport
sediment. In 1993 retrofitting of the cracks was
performed by using a technique that involves the
injection of chemical expansive product. In April 1996,
the rate of leakage increased from 3.9 to 5.1 l/s (Arfaoui
2007). In a second step, another injection technique was
tested, but it also revealed to be unsuitable to stop the
water leakage. In June 1996, the control office recommended the application of a suitable geotextile filter plate
product, which needs to be renewed once the leakage of
water is again observed in the gallery. In 2007, the
repairing of cracks was performed by the usage of a new
product that is a composite material using reinforced
polymer by ceramic steel. This product improved the
water-tightness of the drainage gallery, and the transport
of fines was prevented with much reduced leakage rate as
5 l/min. Figure 2 groups the development of observed
behavior concerning both the water storage and leakage
from which resulted the piping phenomena.
3 Analysis of Settlement Measurements
The settlement development of Ghezala dam foundation has been followed up by means of topographic
recorders made up of twenty-five points installed
Geotech Geol Eng
Fig. 1 Cross section of
Ghezala dam with
localization of cracks in the
gallery structure
Fig. 2 Debit of leakage and loss of solid quantities of Ghezala dam
underneath the sub base of the gallery at the time of
dam construction in 1982. A plot of the cumulative
settlement of the gallery, from 1982 to 2008, (Fig. 3)
shows that the maximum settlement is 45 cm in the
central portion of gallery located under the crest
elevation of the dam. The settlement linearly
decreases towards the upstream and downstream sides
quasi proportionally with the reduction of the dam
height. It is noticed that the settlement over the first
3 years of the gallery construction was approximately
26.5 cm (1985), and reached approximately 32.2 cm
in 1987, which is equivalent to approximately 71 % of
the current maximum settlement measured after
26 years of service.
123
Geotech Geol Eng
Fig. 3 Evolution of recorded settlement underneath the gallery
of Ghezala Dam
as shown in Fig. 4 were analyzed: section (A–A) is at
the upstream side with an equivalent water pressure of
25 m height; section (B–B) at the crest level of
maximum dam elevation of 30 m and, section (C–C)
at the downstream side (Al Husein 2001).
The perfect elastoplastic Mohr–Coulomb behavior
has been adopted for the compressible marl foundation
layer and the fill clayey material used to construct the
dam (Brinkgreve 2003). The geotechnical parameters
used in this study are summarized in Table 1. Three,
scenarios for the degree of saturation of the marl layer
were considered. The idea behind the scenarios is to
simulate three different periods according to water
infiltration around the gallery and the consequent
settlement development.
4 Evolution of Settlement of Gallery Structure
5.2 Numerical Modeling
Figure 3 shows that the measured settlements from
1982 to 2008 went through three different stages. The
first stage lasted approximately 1,000 days (between
1982 and 1985) during which the construction of the
dam had induced, in the central portion of gallery, the
highest rate of settlement of about 90 mm/year. The
second stage was between 1985 and 1987 when full fill
up of the dam reservoir had been maintained. In this
stage, a moderate settlement development was
recorded at a rate of 20–25 mm/year. In the third
stage (between 1987 and 2008), the rate of settlement
varied from 5 to 6 mm/year. Further, the decreased
trend of settlement development leads to deduce that
the primary consolidation is almost ended, and
therefore the behavior of the foundation of Ghezala
dam is stable.
5 Numerical Simulation of Gallery Behavior
5.1 Geotechnical Conditions of Ghezala Dam
The measured settlements, during the service life of
Ghezala dam, show that the short term undrained
conditions prevail to explain the observed behavior of
the gallery. The latter is modeled as a reinforced
concrete frame having a 2 9 2.5 m2 rectangular cross
section, 0.6 m thick and 170 m long (Fig. 5). In order
to simulate the behavior of the gallery along a given
cross section of the dam, that is subjected to various
loading conditions (Renon 2002), a plane strain
analysis is undertaken. Three different cross sections,
123
Numerical computations were conducted using the half
model presented in Fig. 5 due to the geometrical and
mechanical symmetries (Boidy 2002). Figure 5 shows
the typical plane strain model including the rectangular
cross section of the gallery overlaid by variable
thickness of dam material or height of water. The
width of the model was taken as 40 m (ten times greater
than the gallery width), so that the prediction of soil
behavior around the gallery will not be affected by the
condition of zero horizontal displacement along the
vertical border (Fig. 6). The thickness of the marl
foundation layer was set equal to 11 m, while the
thickness of the earth dam layer was variable depending on the studied section. The numerical mesh is
composed by 15 nodes triangular elements (Hicher and
Chang 2007). The gallery is modeled as a reinforced
concrete plate element having a linear elastic behavior
with normal rigidity EA and bending stiffness EI.
The initial at rest state defined by: K0 = 1 - sin u0
is considered for all materials which were treated as
normally consolidated soils. The plane strain hypothesis was used by considering the primary consolidation procedure in clayey marl layer and by assuming
anisotropic hydraulic conductivity for all materials
(vertical
and
horizontal
permeability
are:
kv = 10-8m/s and kh = 10-7m/s, respectively).
5.3 Processed Loading
The embankment is not primarily activated, and then it
is activated by single layers in accordance with the
Geotech Geol Eng
Fig. 4 Gallery cross
sections
Table 1 Geotechnical parameters and materials constitutive
of numerical model
Material
Sr
(%)
Clayey
marl
100
Dike’s
material
cd
(kN/m3)
ch
(kN/m3)
C0
(kPa)
u ()
E
(MPa)
16
20
35
25
20
90
15.4
19.4
30
20
20
50
15.35
17.5
25
20
20
16
20
30
27
15
Fig. 6 Finite element mesh of 2D model generated by Plaxis
software
Fig. 5 Cross section of the dam
staged construction of the dam. Such optimized
numerical procedure reveals efficient to simulate the
staged construction as it happens during the actual
construction of embankments. Indeed, the construction of earth dam was modeled by ten layers built in
1,090 days in conformity with the executed project.
The water level at the upstream side is modeled by
uniformly distributed pressure on the upper side of
embankment as shown in Fig. 7, which refer to the last
stage of applied loading (Bouassida et al. 2011).
Two numerical computations were carried out, by
using the mechanical model above described, in order
to predict the settlement of the gallery foundation.
Each simulation included two loading phases, the
initialization of stresses followed by, the dam weight
loading.
123
Geotech Geol Eng
Fig. 7 Loading of the numerical model in plane strain
condition
Fig. 8 Evolution of settlement of the gallery: Section (A-A)
5.4 Interpretation of Results
Measured and predicted settlements over time are
compared for the three studied sections as shown in
Figs. 8, 9 and 10. It is clear that the adopted numerical
model retraces the overall observed behavior of the
settlement development over time. The trend of the
observed behavior was quasi identical for the three
studied sections. In addition, the numerical results
predicted what was observed in situ in that the rate of
settlement development was very important in the
period of 100–1,000 days after construction. Moreover, the adopted model shows that the magnitude of
settlement depends on the thickness of dam material
and water pressure when it is applied. Finally, the
numerical model proves that the long term consolidation settlement will not affect the behavior of the
123
Fig. 9 Evolution of settlement of the gallery: Section (C-C)
Fig. 10 Evolution of settlement of the gallery: Section (B-B)
drainage gallery due to the insignificant calculated
settlement between 1,000 and 10,000 days after
construction: which is the same behavior in situ.
Therefore, it can be concluded that the adopted
mechanical model with its geotechnical parameters
simulates thoroughly the observed behavior of Ghezala dam.
6 Solutions Against Piping
The cracks opening and particle transportation can
cause the leaching of clay material which induces a
mechanism of ageing:
•
A pin-hole test should be performed for the dam
clay material to assess their sensitivity against the
risk of internal erosion.
Geotech Geol Eng
•
•
•
•
Plastic concrete diaphragm walls can be suggested
to seal the body of the dam;
The injection of low viscosity resine to the gallery
extrados, by drilling the entire length of the gallery
from its inside. This technique has been used for
the treatment of leakage in the gallery (composed
by 15 m long concrete wall) of Sayano Shushenskaya dam-Siberia, which is characterized by the
opening of cracks and transporting of solids at a
rate of 320 l/s. This technique has reduced the rate
of sediment transport to 6 liter/second in post
injection of resine.
The gallery of the dam is not equipped with
drainage devices at the contact dam-gallery.
The rectangular section of the gallery does not ease
the compaction of backfill material at the contact
embankment-gallery. Thus, water infiltration may
occur and contribute in favor of the instability of
dam.
7 Conclusions
The behavior of the foundation of Ghezala dam was
simulated by using Plaxis code to explain the origin of
disorders which had seriously affected the drainage
gallery.
Prevention solution has been formulated to stop the
piping phenomena and to reduce the leakage flow of
water from the gallery to the compressible marl layer.
A 2D plane strain model made possible the
simulation of observed settlements over 10,000 days
underneath the gallery.
The good agreement obtained between predicted
and measured settlements leads to conclude that the
adopted numerical modeling and mechanical characteristics of materials were suitable for the simulation
of the behavior of Ghezala Dam. The numerical results
as well as the measured values show that the development of consolidation settlement of the marl
foundation layer of the gallery tends to be stabilized.
References
Al Husein M (2001) Etude du comportement différé des sols et
ouvrages géotechniques. Thèse de doctorat, Université
Joseph-Grenoble
Arfaoui M (2007) Préparation des fissures du plot n4 de la
galerie de prise d’eau du barrage Ghezala. Ministère de
l’agriculture et des ressources en eaux. Tunis, 1–9
Boidy E (2002) Modélisation du comportement différé des cavités
souterraines » Thèse de doctorat, Université Joseph-Grenoble
I, pp 35–47
Bouassida M, Karoui H and Belaid M (2011) Pathology of
foundation of Ghezala dam, a Tunisian case history. Proc.
15th African Regional Conference on Soil Mechanics and
Geotechnical Engineering, 63–70
Brinkgreve R (2003) Manual of reference of Plaxis, Version 8
Hicher PY, Chang C (2007) A microstructural elastoplastic
model for unsaturated granular materials. Int J Solids Struct
44(7–8):230–234
Mc Cann, M-W Jr et al (1997) National performance of dams
program report. Hydropower Dams (4)
Renon N (2002) Simulation numérique par éléments finis des
grandes déformations des sols. Application à la Scarification. Thèse de doctorat, Ecole des Mines de Paris, pp 23–48
123