See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/312580584

Evaluation of the Behavior of Earth and Rockfill Dams during Construction

and First Impounding Using Instrumentation Data and Numerical Modeling

Article  in  Journal of Rock Mechanics and Geotechnical Engineering · February 2017

DOI: 10.1016/j.jrmge.2016.12.003

CITATIONS READS

30 1,074

2 authors:

Mohammad Rashidi S. Mohsen Haeri

University of Texas at El Paso Sharif University of Technology
15 PUBLICATIONS   105 CITATIONS    102 PUBLICATIONS   964 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Effect of liquefaction induced lateral spreading on piles View project

Rockfill and earthfill dams View project

All content following this page was uploaded by Mohammad Rashidi on 10 August 2017.

The user has requested enhancement of the downloaded file.

 Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725

Contents lists available at ScienceDirect

Journal of Rock Mechanics and

Geotechnical Engineering
journal homepage: www.rockgeotech.org

Full Length Article

Evaluation of behaviors of earth and rockfill dams during construction

and initial impounding using instrumentation data and numerical
modeling
Mohammad Rashidi a, *, S. Mohsen Haeri b
a
University of Texas at El Paso, El Paso, USA
b
Sharif University of Technology, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history: In this study, the behavior of Gavoshan dam was evaluated during construction and the first impounding.
Received 13 August 2016 A two-dimensional (2D) numerical analysis was conducted based on a finite difference method on the
Received in revised form largest cross-section of the dam using the results of instrument measurements and back analysis. These
11 December 2016
evaluations will be completed in the case that back analysis is carried out in order to control the degree
Accepted 18 December 2016
Available online 1 February 2017
of the accuracy and the level of confidence of the measured behavior since each of the measurements
could be controlled by comparing it to the result obtained from the numerical model. Following that, by
comparing the results of the numerical analysis with the measured values, it is indicated that there is a
Keywords:
Earth and rockfill dams
proper consistency between these two values. Moreover, it was observed that the dam performance was
Initial impounding suitable regarding the induced pore water pressure, the pore water pressure ratio ru, settlement, induced
Numerical modeling stresses, arching degree, and hydraulic fracturing probability during the construction and initial
Instrumentation impounding periods. The results demonstrated that the maximum settlement of the core was 238 cm at
Settlement the end of construction. In the following 6 years after construction (initial impounding and exploitation
Pore pressure period), the accumulative settlement of the dam was 270 cm. It is clear that 88% of the total settlement of
the dam took place during dam construction. The reason is that the clay core was smashed in the wet
side, i.e. the optimum moisture content. Whereas the average curving ratio was 0.64 during dam con-
struction; at the end of the initial impounding, the maximum amount of curving ratio in the upstream
was 0.81, and the minimum (critical) amount in the downstream was 0.52. It was also concluded that this
dam is safe in comparison with the behaviors of other similar dams in the world.
Ó 2017 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by
Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).

1. Introduction (ICOLD, 1982), such as characterizing the dam’s overall behavior

Engineering and economical investigations in dam construction
projects throughout the world indicate that, in many cases, rockfill
dams with impervious clay cores are the best selection for the final
design (Rezaei and Salehi, 2011). This approach makes the inves-
tigation of different issues affecting the stability of rockfill dams
worthwhile. In the field of earth and rockfill dams, monitoring of
typical physico-mechanical behaviors is a fundamental issue.
Measurements of displacements, total stresses, pore water pres-
sures, and arching ratio can be used to carry out a number of tasks
* Corresponding author.
E-mail address: mrashidi@miners.utep.edu (M. Rashidi).
Peer review under responsibility of Institute of Rock and Soil Mechanics,
Chinese Academy of Sciences.
(Pagano et al., 1998), checking the behavior of specific zones,
obtaining information about the in situ mechanical properties of
embankment soils (Marsal and Resendiz, 1975), and finally, sup-
porting the difficult task of evaluating dam safety and efficiency
(Gould and Lacy, 1993).
Duncan (1996) and Kovacevic et al. (1994) are the main con-
tributors to the state of the art in finite element analyses of
embankment deformation behaviors, mainly for zoned earth and
rockfill dams. They discussed the analysis methods and their limi-
tations, available constitutive models of stressestrain relationship,
and areas of uncertainty. As Duncan (1996) pointed out, most an-
alyses from the literature are of Class C1 according to Lambe (1973),
which may account for the generally good agreement between
predicted and observed deformation behaviors.

http://dx.doi.org/10.1016/j.jrmge.2016.12.003
1674-7755 Ó 2017 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-
NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
710 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725
An important component of embankment dam modeling is the
consideration of collapse settlement of susceptible rockfills and
earthfills on wetting. The collapse settlement most affects the
initial impounding for the upstream dam abutment, but the
collapse settlement has also been observed in the downstream dam
abutment following wetting due to rainfall, leakage or tail-water
impoundment. Incorporating collapse settlement into constitutive
models adds further complexity and greater uncertainty in the
estimation of material parameters in laboratory and field condi-
tions due to the dependency of collapse settlement on compacted
water content, compacted density, applied stress and material
properties. Justo (1991) and Naylor (1997) proposed methods for
incorporation of collapse settlement of rockfill into constitutive
models. Naylor (1991) performed the finite element analysis for
Beliche dam, a central core earth and rockfill dam, by considering
the collapse settlement of the upstream rockfill in the modeling.
The effect of pore water pressure dissipation in earthfill during
construction was also considered by a number of authors including
Eisenstein and Law (1977) and Cavounidis and Hoeg (1977),
amongst others. For these cases, the incremental embankment
construction was modeled as a two-stage process, the first stage
modeling the new layer construction using undrained properties
for the core and the second stage modeling pore water pressure
dissipation. In most cases, pore water pressure development in the
core is ignored. For wet placed earthfill, where high pore water
pressures are developed during construction, the core is often
modeled using undrained strength and compressibility parameters,
and the permeability is assumed as sufficiently low such that pore
water pressures will not dissipate during the period of
construction.
Zhou et al. (2011) assessed the settlement behavior of Shuibuya
dam during construction, initial impounding, and two years after
operation. They carried out two-dimensional (2D) numerical ana-
lyses using finite element method (FEM), and compared the results
with the data measured by the instruments in terms of settlements.
In addition, back analysis was performed by utilizing hybrid generic
algorithms (HGAs). The results represented this technique as a
successful one for controlling the dam deformation. Furthermore,
the results demonstrated that the settlement increased apparently
after the initial impounding, but the rate of settlement decreased
and tended to stabilize over time.
Mahin Roosta and Alizadeh (2012) assessed the nonlinear
behavior of rockfill material using numerical analyses and labora-
tory tests. To estimate the collapse settlement phenomena in the
rockfill dam during inundation, strain-hardening and strain-soft-
ening models in the fast Lagrangian analysis of continua (FLAC)
software (Itasca, 2002) were modified based on the data obtained
from laboratory tests. The results helped dam engineers to better
predict the nonlinear behavior and collapse settlements in the
upstream shell. In addition, the effect of impounding process on the
overall stability of a high dam was considered by Luo et al. (2015).
The dam behavior predicted using the three-dimensional (3D)
numerical modeling is in good agreement with field measurement
in terms of stress and strain distribution characteristics in the dam
and foundation, especially when the feedback parameters were
adopted for numerical analysis.
As can be seen, previous researches did not consider all pa-
rameters in rockfill dams such as settlement, pore water pressure,
vertical stress, and arching ratio during the periods of construction
and initial impounding in a comprehensive study. This paper pre-
sents the evaluation of these parameters in an earth and rockfill
dam in stages of dam construction and operation using instru-
mentation data and numerical modeling. It first introduces the
Gavoshan rockfill dam, then reviews the results of settlement, pore
water pressure, vertical stress, and arching ratio during
construction and the first impounding. Finally, these results are
compared with the monitored ones.
2. Main feature of Gavoshan dam
Gavoshan dam is 123 m high, the length of the crest is 650 m
and the volume of the container is approximately 550
106 m3
(Mahab Ghodss Engineering Consulting Co., 2009). The dam is a
rockfill dam with a clay core and is located in Kurdistan Province,
Iran. The purpose of building the Gavoshan dam was to harness
Gaveh Rood River and to supply water to irrigate approximately
31,000 ha of downstream lands and also to provide 63
106 m3 of
drinking water and 11 MW of electrical energy to Kermanshah
Town. To make clear the dam’s behavior, accurate instruments were
installed in five sections, including piezometers, measurement cells
for total soil mass pressure, settlement meters and inclinometers,
as shown in Fig. 1. Taking into account the foundation type, the
height of the earthfill and the depth of the valley, the profile located
at 0 þ 365 km was considered as the most critical cross-section of
the Gavoshan dam regarding the accumulation of pore water
pressure, the maximum settlement and tolerance, as presented in
Fig. 2. In rockfill dams, the most critical section regarding stability is
always the highest section assuming that other conditions are
similar in all points. Also most of the stresses, displacements and
pore water pressures usually take place in this section. Since there
are a number of sections in Gavoshan dam where instruments are
set up, it is impossible to analyze all these sections within the
limited duration of the present research, only the stressestrain
behavior on critical sections is analyzed. Clearly, if the behavior of
the dam in critical sections is understood accurately, this behavior
could be generalized to all sections of the dam.
The construction of Gavoshan rockfill dam started in October
2000 (Fig. 3). Till 19 June 2004, 115 m of the dam was constructed
and the initial impounding had begun. Four months after the initial
impounding, on 22 October 2004, the construction was completed.
Considering raining seasons in Kurdistan area and also the fact that
the central core of the dam body could not be constructed in all
seasons, only the 6 months between May and October in each year
are suitable for earthfilling with clay materials. Also, considering
the high volume of shell materials in the upstream and down-
stream central cores and the fact that it is impossible to simulta-
neously execute the construction of the central core and the shell, a
part of the upstream and downstream shell is not constructed at
the same time with the core, and in the seasons in which the
earthfilling of central core is stopped, the construction of the
remaining shell is executed.
3. Material properties and numerical modeling
Gavoshan dam was modeled using FLAC software, based on the
finite difference method. Mohr-Coulomb elastoplastic model was
used for the earthfill, filters and core. As the well-known failure
model in soil mechanics, the model states that failure occurs when
the shear stress s and the effective normal stress s0 acting on any
element in the material satisfy the following linear equation:
js j ¼ c þ s0 tan f (1)
where c and f are the cohesion and the internal friction angle,
respectively. In terms of the principal effective stresses, the Cou-
lomb’s yield criterion can be expressed by the following equations:
9
f13 ¼ ðs0 1  s0 3 Þ=2  ðs0 1 þ s0 3 Þsin f=2  c cos f ¼ 0 =
0 0 0 0
f12 ¼ ðs 1  s 2 Þ=2  ðs 1 þ s 2 Þsin f=2  c cos f ¼ 0 (2)
;
s0 1  s0 2 ¼ s0 3 ðtriaxial compressionÞ
M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725 711

Fig. 1. The plan view and sections of Gavoshan dam.

Fig. 2. Critical cross-section of Gavoshan dam.

Fig. 3. Construction and impounding periods of Gavoshan dam.

712 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725
This failure criterion can be rewritten in terms of deviatoric
stress and mean effective stress which is a more common expres-
sion for defining failure behavior in advanced soil models:
f1 ¼ q  sin fðp0 þ c cot fÞ
In the Mohr-Coulomb model, it is assumed that the material
behaves elastically until the failure surface is approached. However,
in reality, plastic deformations begin well before failure occurs. But
still this model has been used because of its simplicity and ability to
obtain reasonable solutions to important practical problems in
geotechnical engineering.
3.1. Material properties
The material properties used in the numerical analysis were
extracted from certain soil mechanics laboratory tests on fine and
coarse materials of the dam. The initial values of the parameters in
the Mohr-Coulomb model for various zones of the dam are pre-
sented in Table 1. These values are used for preliminary calcula-
tions. The calculations will be completed in the case that back
analysis is carried out to control the degree of the accuracy and the
level of confidence of the measured behavior since each of the
measurements could be controlled by comparing it to that obtained
from the numerical model. This process and the solution algorithm
are described in Fig. 4.
Table 2 lists the back-analyzed material properties. The differ-
ences between the initial and the back-analyzed input parameters
are induced by the quality control of compaction and the actual
natural characteristics of the incorporated materials. Back analysis
is an effective method to check and modify the parameters of soils.
Moreover, the results of back analysis are beneficial to safety
assessment of project itself, and can improve the design of subse-
quent structures of the same type (Yu et al., 2007).
3.2. Numerical modeling and analysis
Gavoshan dam was modeled in 26 layers. The height of the
earthfill layers was defined to be between 3 m and 6 m. First, the
cofferdam was modeled at the time when it was constructed, and
then the layers were modeled. The Gavoshan dam impounding was
done in 11 layers similar to its construction, and the height of the
layers was 7e13 m. The impounding of each layer was carried out
considering the elevation of the water table behind the dam which
was recorded by instruments. The first impounding of the dam was
completed in 36 months. It should be noted that in the numerical
analysis, the rock foundation was not modeled due to its minor
settlement during construction (5 cm), low permeability (approx-
imately 1
106 m/s), and also the concrete slab at the bottom of
the clay core and its thickness of 200 cm. Fig. 5 shows the finite
element model of the critical cross-section. Four-node quadrilateral
elements with a maximum length-to-width ratio of 2 were used for
mesh generation.
Figs. 6e11 present the results of numerical analysis in the two
phases of dam impounding, i.e. from the filling level of 108 m
Table 1
Initial parameters of dam body materials.
Elements Dry unit Unit weight at
weight (kN/m3) saturation (kN/m3)
Core 17.5 20.7
Upstream and downstream 22.6 23.8
shell
Filters and transition 20 21.2
before the beginning of the first impounding to the water level of
110 m after impounding.
3.3. Effects of initial impounding on rockfill dam
(3)
One of the crucial issues in earth and rockfill dams is the
tolerance of dam during initial impounding. Displacements of the
dam in the initial impounding could have a complex nature. To
analyze these complex displacements, Nobari and Duncan (1972)
investigated the effects of impounding on a dam with mixed
cross-section. They stated the effect of water pressure as follows:
(1) The water pressure in the upstream core has increased the
total lateral tension and pushed the dam downwards.
(2) The effect of water pressure on the upstream foundation,
which has a high compressibility, could be highly effective on
changes in the shape of dam.
(3) The decrease in the effective stress on the upstream shell
caused flotation phenomenon which led to buoyant uplift to
some extent in this area.
(4) The decrease in shear strength parameters and elastic
modulus of the upstream shell material caused wetting and
rock softening, which eventually induced settlement.
The effects of impounding mentioned above are shown in
Fig. 12. It should be pointed out that the settlement induced in
impounding is the most important factor which causes the de-
creases in frictional resistance and hardness of rocks of impounding
layer underneath the heavy weight of that layer (Naylor, 1991).
In this study, the method provided in Justo (1991) was mainly
used for unloading analysis of the lower layers. Because of the
impounding effects, the decrease coefficient of internal friction angle
of material was assumed to be 20% (Alonso and Oldecop, 2000;
Soroush and Aghaei, 2005). The decrease coefficient of Young’s
modulus was assumed to be 50% with regard to the studies that have
been carried out (Saboya and Byrne, 1993; Varadarajan et al., 2003).
4. Results and discussion
In this section, the data recorded by the instruments in Gav-
oshan dam including settlements, total vertical stresses and pore
water pressures at the end of construction and initial impounding
are evaluated and then compared with those obtained by numer-
ical analysis.
4.1. Settlement
The internal settlements of the dam are categorized into three
groups: vertical, horizontal, and rotational movements. Vertical
movements show the settlements in terms of material weight,
compaction, and consolidation of the dam body. Horizontal
movements mainly refer to the upstream movements that occur
during impounding in the dam storage, which is induced by the
faster reduction in the effective stress in the upstream materials
than that in other parts of the dam. Downstream movement is
Internal friction Cohesion Elasticity Poisson’s Dilation Hydraulic
angle ( ) (kPa) modulus (MPa) ratio angle ( ) conductivity (m/s)
27 50 26 0.3 0 1
108
40 e 124 0.24 7e10 Freely drained
35 e 64 0.28 3e4 1
104
 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725 713

Fig. 4. Solution algorithm for back analysis.

Table 2
Back-analyzed material parameters.

Elements Dry unit Unit weight at Internal friction Cohesion Elasticity Poisson’s Dilation Hydraulic
weight (kN/m3) saturation (kN/m3) angle ( ) (kPa) modulus (MPa) ratio angle ( ) conductivity (m/s)

Core 17.5 20.7 30 50 16 0.33 0 1
108

Upstream and downstream shell 22.6 23.8 43 e 72 0.2 8 Freely drained
Filters and transition 20 21.2 36 e 35 0.26 5 1
104

Fig. 5. Mesh generation of dam body at the critical cross-section.

714 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725

Fig. 6. Vertical displacement of dam at the filling height of 108 m before the starting of initial impounding (unit: m).

Fig. 7. Vertical displacement of dam at the impounding height of 110 m (unit: m).

Fig. 8. Horizontal displacement of dam at the filling height of 108 m before the starting of initial impounding (unit: m).

Fig. 9. Horizontal displacement of dam at the impounding height of 110 m (unit: m).

Fig. 10. Pore water pressure of dam at the filling height of 108 m before the starting of initial impounding (unit: Pa).

produced by the horizontal water pressure of the dam storage. surveying points, inclinometers (settlement tubes), and/or settle-
Rotational movements that appear in the upstream and down- ment gages to measure these deformations in earth dams. In this
stream slopes are induced by the lower shear strength of materials study, only the vertical settlements in the Gavoshan dam obtained
in the foundation or dam body. It is a conventional way to use from inclinometers are analyzed.
 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725
Fig. 11. Vertical stress of dam at the filling height of 108 m before the starting of initial impounding (unit: Pa).
Fig. 12. Impounding effects in a rockfill dam (Naylor, 1991).
To measure the displacement of the dam critical cross-section
shown in Fig. 13, three tubes of inclinometers were installed at the
downstream and upstream and near the axis of the dam. Fig. 14
shows the settlement variation of the core with the dam height in
8 different time intervals from the dam construction. According to
Fig. 3, the dam construction was completed in October 2004. At this
time, the maximum settlement was 238 cm. The initial impounding
of Gavoshan dam begun from June 2004 when the final stages of
earthfilling of the dam body had begun, and continued up to April
2006. During this period, the displacement of the dam body at this
level was 256 cm according to Fig. 14. The settlements in April 2007
and 2009 were 258 cm and 270 cm, respectively. It was observed that
from the initial impounding to the dam exploitation (approximately
5 years), the settlement was only 32 cm, which was 12% of the total
settlement of the dam. While 88% of the settlements (238 cm) took
place in the construction period, as seen in the figure. The readings of
some settlement meter plates were not available due to the blockage
in inclinometer tubes, which impeded the settlement meter from
going inside the tubes and arriving at the lower levels.
Fig. 15 presents the settlements in the inclinometer tubes versus
the earthfilling levels at 9 time intervals. Due to inclinometers
malfunction, the readings of the inclinometers embedded 75 m
below the dam in the upstream were not available. Considering the
fact that the maximum dam settlement occurred in the middle
sections, it was not possible to obtain accurate results.
Fig. 16 shows the settlement obtained from the settlement
meter tubes in the downstream at 9 time intervals. When earth-
filling was completed in October 2004, the maximum settlement
was 207 cm, which occurred at the height of 66 m of the dam body.
In April 2006, January 2007 and April 2009, the maximum settle-
ments of the downstream shell were 218 cm, 219 cm and 220 cm,
respectively. It could be concluded from Fig. 16 that at the time of
715
initial impounding, the settlement of the downstream shell was
10 cm and it kept stable afterwards. It was also observed that in the
following three years after the beginning of dam exploitation, i.e.
from 2006 to 2009, the settlement of the dam body was only 2 cm.
Fig. 17a compares the settlements obtained from 3 inclinometer
tubes at the end of construction (November 2004). As can be
observed, the difference in the upstream and downstream core
settlements is not significant due to the differences in behavior
characteristics and compressibility of the core, filter, drainage and
shell. Considering that the materials of the upstream and down-
stream dam body and the central core were similar, the difference
in the settlements in the upper part of the dam was very little.
Fig. 17b shows the settlements in April 2008, about 5 years after
the beginning of the initial impounding. It can be seen that the
settlements increased to some extent in comparison with the ones
at the beginning of the initial impounding. It is noted that the in-
crease in the settlement of the upstream shell is greater than that of
the downstream shell due to the hydrostatic pressure of water on
the upstream shell.
4.2. Comparison of settlements obtained from instrumentation
measurements and numerical modeling
Figs. 18 and 19 compare the settlements of the central core in
June 2004 (towards the end of the construction) and April 2006
(after the initial impounding) obtained from instrument measure-
ments and numerical modeling, respectively. It can be seen that the
numerical results are very close to the actual ones in both periods of
time.
Figs. 20 and 21 compare the downstream shell settlements
obtained from instruments and numerical modeling in the two
periods mentioned above. At the end of construction, these results
716 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725
Fig. 13. Inclinometers on the critical cross-sections.
are close to each other. This trend keeps constant in the lower part
of the dam after the initial impounding. But in the upper part of the
dam, the results from instruments are obviously different from the
numerical ones. The reason is the low density of the materials in
the upper part of the downstream shell and heavy snow and rain in
cold seasons of the year in Kurdistan Province, where the annual
amount of rain in common climate conditions is 600 mm.
Fig. 14. Core settlement obtained from I3002 along dam height at different times.
The pressure meters in the upstream section were damaged
before the construction was completed. Readings of the remaining
pressure meters in September 2003 are compared with the nu-
merical results, as shown in Fig. 22. The results were not so
different up to that date. The difference occurred at the height of
45e65 m, located in the transferring area of the dam. The speci-
fications of the materials in the transferring area are different
from those of the filter. However, because the specifications of
these materials were not available, we had to consider the spec-
ifications of the materials of the filter as the one of the materials of
the transferring area. The readings of inclinometers are also
compared with the numerical results at different levels of the dam
in relation to time, as illustrated in Fig. 23. As it is observed, the
numerical results are very close to the actual ones. It is also shown
that the settlement of the dam increases significantly after 48
months (at the end of construction and the start of initial
impounding).
4.3. Pore water pressure
Vibrating wire piezometers were installed at different levels, i.e.
1430.5 m, 1455 m, 1483 m, 1503 m, and 1526 m, as shown in Fig. 24.
Fig. 25 shows the pore water pressure at 1430.5 m level. As seen in
the figure, the electrical piezometers installed on the first level are
all readable and the results indicate the accurate function of the
piezometers and the natural behavior of dam. With the addition of
earthfilling layers, the pore water pressure increases. It is clear that
from July 2003 (the 34th month), the growth rate of pore water
pressure is reduced. This indicates the drop process of pore water
pressure.
Fig. 26 indicates the pore water pressure at 1430.5 m level
before and after the initial impounding. Before the initial
impounding starts, the graph is symmetrical, and the pore water
pressure decreases as the distance from the dam axis increases. But
after the impounding begins, the upstream clay core has a higher
pore water pressure than the downstream clay core. Meanwhile,
the pore water pressure of the downstream clay core decreases
after the initial impounding due to the pore water pressure drop
occurring after construction and initial impounding.
Fig. 27 shows the trends of pore water pressure at electrical
piezometers EP 306, EP 304 and EP 309 with regard to the time
passing by the piezometer EP 304 which is located in the upstream
clay core. The figure shows that after 50 months (when the
 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725 717

Fig. 15. Settlement of upstream central core obtained from I3003 along dam height at different times.

Fig. 16. Settlement of downstream central core obtained from I3001 along dam height at different times.

construction was completed), the pore water pressure at EP 304
increases. While the piezometer EP 309, which is located in the
downstream clay core, shows a decrease in the pore water pressure
in this period, unlike the piezometer EP 304. Piezometer EP 306
located in the middle of the clay core indicates the behavior be-
tween those of EP 304 and EP 309 after the beginning of initial
impounding. The main difference is the higher pore water pressure
at EP 306 in comparison with those at the other two piezometers.
The only noticeable point in this graph is the dramatic decrease in
pore water pressure growth rate in the 32nd month (July 2003)
which is caused by the drop in pore water pressure in this month.
It seems strange for the decrease in pore water pressure in July
2003 at 1430.5 m level regarding the progress in the construction
process of Gavoshan dam in the summer and also the prediction
that the pore water pressure would increase. To make sure the
accuracy of the readings of pore water pressure at the first layer
level, one piezometer at the second level, i.e. EP 340, was investi-
gated, as illustrated in Fig. 28. It is clear that the decrease rate of the
pore water pressure at EP 340 is less than that at EP 306 in the first
layer.
The pore water pressures measured in June 2003 at different
levels of the dam are shown in Fig. 29. It is observed from the figure
that the pore water pressure at 1430.5 m level is larger than the
water pressures produced in the upper layers due to greater loads.
Additionally, it is noticed that in this figure, the pore water pressure
of the second layer is close to that of the first layer. The reason is the
high optimum moisture and low density of the second level. From
Fig. 30, it is seen that the water content of clay is more than the
optimum water content in the second level due to its low density
percentage according to technical reports, which is 98%. Because of
the weight of the overlying layers, high moisture and long distance
of fill, the pressure of the water trapped inside the clay increases.
The ratio of pore water pressure is defined as the ratio of the
water pressure produced at each point to the vertical load at each
point, which can be expressed as follows:
u
ru ¼ (4)
gh
where ru is the ratio of pore water pressure, u is the pore water
pressure produced, g is the specific weight of the material, and h is
the height of load at each point.
Fig. 31 shows the calculated ratios of pore water pressure at
three levels of the dam body where electrical piezometers were
718 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725
Fig. 17. Settlements at different heights in clay (I3002), upstream (I3003) and down-
stream central cores (I3001).
Fig. 18. Comparison of settlements obtained from instruments and numerical
modeling in relation to the height of the central core at the end of construction.
Fig. 19. Comparison of settlements obtained from instruments and numerical
modeling in relation to the height of the central core after the initial impounding.
installed. It is clear from the figure that the maximum values of ru
calculated according to Eq. (4) are 0.47, 0.63 and 0.42 in the first,
second and third layers, respectively. These values were based on
the data measured in June 2003 when the dam fill level was 73 m.
Since that, most of the piezometers in the second and third layers
did not function any longer.
In order to obtain the measured values of ru, the total vertical
stress readings from all the tension cells were needed, but actually,
only the total vertical stresses at the first level were available. It is
worth mentioning that at the second level, only electrical pie-
zometers were installed, without total tension cells. Also the
installed total tension cells in the third level were mostly un-
readable. Based on these conditions, the ratio of pore water
Fig. 20. Comparison of settlements obtained from instruments and numerical
modeling in relation to the height of the downstream central core at the end of
construction.
 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725
Fig. 21. Comparison of settlements obtained from instruments and numerical
modeling in relation to the height of the downstream central core after the initial
impounding.
Fig. 22. Comparison of settlements obtained from instruments and numerical
modeling in relation to the height of the downstream central core in September 2003.
pressure was measured for the first level in June 2003 and
compared with the calculated one, as shown in Fig. 32. It could be
seen that the maximum ru calculated for the first level was 0.46,
much less than the measured value of 0.73 due to curving
phenomenon.
4.4. Comparison of pore water pressures from instrumentation
measurements and numerical analyses
Fig. 33 shows the comparison of pore water pressures in the first
layer obtained from instrumentation measurements and numerical
modeling at the end of construction. It can be observed that in the
boarder of the fills, the pore water pressures are close to each other;
but in the core, the maximum pore water pressure measured by the
piezometers is approximately 980 kPa, less than the value of
1280 kPa from the numerical modeling. Fig. 34 shows the
719
comparison of pore water pressures in April 2006 (after initial
impounding). It can be seen that the difference in the pore water
pressures obtained from numerical modeling and measurements
decreases. To explain this difference, we first inspect the history of
pore water pressure at piezometer EP 306 located in the first layer.
As illustrated in Fig. 35, during construction, the pore water pres-
sure obtained from numerical modeling is greatly different from
the measured one, while this difference becomes less in the 50th
month (starting of initial impounding).
As mentioned before, the readings of piezometers located in
the second layer were available in September 2003. Fig. 36
compares the results of pore water pressure from numerical
modeling and measurements of the electrical piezometers
installed in the second layer in June and September 2003. It can
be observed that the results are very close to each other in the
second layer, and the largest difference was 20e40 kPa. Fig. 37
shows the comparison of pore water pressures measured by the
piezometers EP 339 and EP 336 located in the second layer with
the numerical results.
As mentioned before, the average water content in the second
layer (1455 m) is always higher than the desired one, as well as the
average water content in the first layer (1430.5 m). It means that in
the second layer, the clay is smashed in the completely wet side and
has caused the increase in the saturation degree of clay, supporting
the first premise of the numerical analysis that liquid and gas are
compressible. Therefore, the numerical results well match the in-
strument readings.
The above-mentioned reason is more applicable for matching
the results in the second layer. The difference of pore water
pressure between numerical modeling and measurements of in-
struments located in the first layer is significantly great, so we
cannot consider the layer moisture as modeling error. Fig. 38
shows the sketch map of Gavoshan dam situated on rocky foun-
dation. In the numerical analysis, the foundation was not modeled
due to its minor settlement during construction (5 cm), low
absorbance (the hydraulic conductivity K is approximately
1
106 m/s), and concrete slab at the bottom of the clay core
with the thickness of 2 m. It can be seen in Fig. 38 that the
permeability coefficient of the rock of dam shell (K ¼ 5
105 m/
s) is close to that of the dam foundation (K ¼ 5
106 m/s). During
construction, the pore water pressure gradually drops in the first
layer from the sides of the concrete slab. Even when the concrete
slab breaks at some points in the analyzed section of the dam, the
pore water pressure decreases in the first layer. But in the second
layer, since the piezometers in this layer are far from the foun-
dation of the dam, the pore water pressure in this level is not
affected.
Since the dam foundation has been neglected and the bottom of
the dam has been assumed to be impermeable in the modeling, the
pore water pressure in the bottom layer measured by instruments
is significantly different from the numerical result. Therefore, it
could be concluded that when the hydraulic conductivities of the
rockfill shell and the foundation are close, under any circumstances,
the dam foundation should be modeled.
It was observed that after the initial impounding, the pore water
pressure obtained from the instruments was close to the result
obtained from numerical modeling in the lower sections. After the
initial impounding, the upstream and the foundation of the dam
are completely saturated and the water can no longer permeate
into the dam foundation through the dam core. As can be seen in
Fig. 35, the pore water pressures obtained from numerical analysis
and instrument measurement get closer to each other after 2 years.
Therefore, the dam foundations in any case should be modeled
along with the dam body in order that the analysis results would be
very close to reality.
720 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725
Fig. 23. Comparison of settlements at different levels of dam obtained from inclinometers and numerical modeling in relation to time.
4.5. Vertical stress
To obtain the vertical stress, total pressure cells were installed at
two levels, i.e. 1430.5 m and 1483 m, as seen in Fig. 39. Fig. 40 shows
the changes in the total vertical stress measured by the pressure
cells installed in 1430.5 m level in accordance with earthfilling and
impounding at different times. The change process of the vertical
stress is steady and agrees with the progress of dam construction
and impounding except for the cell PC 330 which shows a very low
reading in comparison with the adjacent cells. This is due to the
accurate instrument being damaged. As can be seen, the pressure
meter cells inside the filters show a different behavior in compar-
ison with those in the clay core. Pressure cell PC 312 is located in the
upstream filter and the vertical stress measured by this cell in-
creases after impounding because the materials of the shell and the
upstream filter become saturated. Pressure cell PC 333 located in
the downstream gives higher pressure readings due to the higher
specific weight of the materials of dam shell above the cell,
compared with those of the materials of clay and filter. The pressure
cells installed in the clay core show that with the progress in
earthfilling in Summer 2003, the increase rate of the total vertical
stress decreases due to curving.
Fig. 41 shows the changes in the vertical stress in the core and
the upstream and downstream filters at the first level at different
points and times. It can be seen that with the increase in the levels
of earthfilling, the stress level in the core and the filter also in-
creases. The maximum stress is mostly related to the pressure cell
PC 324 which is located 5 m downstream of the dam axis inside
clay. At the end of construction, the cell’s reading was 1550 kPa. It
can be observed that when impounding begins, the vertical stress
in the upstream area has an ascending trend. The highest vertical
stress of 2002 kPa was obtained by the cell PC 312 in April 2006.
The minor difference in the stress levels in the core and filter in-
dicates that there was no severe curving. In order to accurately
address this issue, it is essential to display the changes in the ratio
of the vertical stress to the load in the width of the clay and filters.
4.6. Comparison of vertical stresses obtained from measurements
and numerical modeling
Fig. 42 displays the development of the vertical stress from PC
321 at different times. It can be seen that the difference in the
numerical and actual vertical stresses is 150e200 kPa from the very
start of construction till the end of the initial impounding. The
vertical stress obtained by the numerical analysis is greater than
the actual one, both of which develop in the same trend. The reason
is that the soil around the cells was compacted to a lower density in
comparison with other parts of the core to prevent damage to the
pressure cells. Consequently, the stiffness of soil in the dam body
was mostly larger than that around the cells.
4.7. Arching ratio and hydraulic fracturing probability
Shell materials including rockfill and gravel were stiffer than the
materials used in the core. Difference of elasticity modulus between
these two kinds of materials made various responses to settlement.
In addition, the friction between the core and shell materials caused
stress transfer from the core to the shell, which could create a low-
stress area in the core. This phenomenon is called arching in earth
dams. The following equation presents the arching ratio:
sv
Ar ¼ (5)
gh
 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725 721

Fig. 26. Pore water pressure at 1430.5 m level before and after initial impounding.

Fig. 27. Pore water pressure in relation to time at the electrical piezometers EP 306, EP
304 and EP 309.

Fig. 24. Locations of electrical piezometers at 5 different levels of critical cross-section

(unit: m).
where sv is the vertical total stress, and gh is the soil overburden
pressure at the point (Terzaghi, 1936, 1943).
The vertical stress is a major factor to prevent cracking in the
core. During construction and initial impounding, decreased
vertical stress due to arching phenomenon can produce horizontal
cracks because the water pressure is higher than the vertical stress.
This is called hydraulic fracturing that makes cracks develop from
the upstream to the downstream. Also, it makes a serious damage
to the dam body, possibly leading to dam failure. To prevent arching
in the core, the arching ratio should be less than 1 at the locations of
pressure cells in the core. The higher the value of Ar is, the less the

Fig. 25. Pore water pressure at 1430.5 m level of dam body.

722 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725

Fig. 28. Pore water pressure in relation to time at the electrical piezometers EP 306
and EP 340. Fig. 31. Calculated ratios of pore water pressure for the electrical piezometers installed
at three different levels of the dam body.

Fig. 32. Comparison of calculated and measured ratios of pore water pressure at
1430.5 m level.

Fig. 29. Pore water pressure at different layers of the dam body in June 2003.

Fig. 33. Comparison of pore water pressures obtained from measurements and nu-
merical modeling for the first layer in June 2004.

Fig. 30. The average water content in clay rockfill layers.

arching phenomenon in the core is, and the lower the probability of
hydraulic fracturing would be.
Fig. 43 presents the changes in the arching ratio in the width
of core and filters at the first level in August 2002, June 2004
(towards the end of construction and beginning of the
impounding), and April 2006. It can be seen that the average
arching ratio in June 2004 was 0.64. This indicates that only 64%
of the load weight on the core in this level was transferred to the
materials underneath it, and the remaining 36% was transferred
to the shell and side filters of the core. It is observed that 2 years Fig. 34. Comparison of pore water pressures obtained from measurements and nu-
merical modeling for the piezometers located in the first level in April 2006.
 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725 723

5. Conclusions

The purpose of this study was to present a general evaluation of

dam function during construction and the initial impounding using
the data obtained from instruments at Gavoshan earth and rockfill
dam and also numerical modeling regarding the stressestrain
behavior of the dam. The following results are drawn:

Fig. 35. Comparison of pore water pressures obtained from measurements and nu-
merical modeling at the location of electrical piezometer EP 306 in the first level.
after impounding of Gavoshan dam, the average arching ratio
remains 64%, and the arching ratio in the upstream area increases
and even reaches 0.81. But in the downstream areas, this value
decreases. The most critical arching occurred in April 2006, as
much as 0.52, at PC 330 located in the downstream clay core. This
means that by the time that impounding begins, the amount of
arching in the downstream core increases. Therefore, the arching
ratio in Gavoshan rockfill dam was between 0.81 and 0.52,
showing that the dam was located on a safe side in terms of
hydraulic fracturing at the end of construction and during the
first impounding.
(1) The maximum settlement of the core was 238 cm at the end
of construction. In the following 6 years after construction
(initial impounding and exploitation period), the accumula-
tive settlement of the dam was 270 cm. It is clear that 88% of
the total settlement of the dam took place in dam con-
struction due to the clay core being compacted in the wet
side of optimum water content.
(2) The average curving ratio was 0.64 in the first layer of the
dam. At the end of the initial impounding, the curving ratio
was kept constant at 0.64, but the maximum amount in the
upstream was 0.81 and the minimum (critical) amount in the
downstream was 0.52. This means that by the time when
impounding begins, the amount of curving in the down-
stream of the core increases. Therefore, the arching ratio in
Gavoshan rockfill dam was between 0.52 and 0.81, showing
that the dam was located on a safe side in terms of hydraulic
fracturing at the end of construction and during the initial
impounding.
(3) The assumptions that the dam is constructed in 26 layers and
the impounding occurs in 11 stages are appropriate. Obvi-
ously, the assumption of fewer layers in the construction

Fig. 36. Comparison of pore water pressures obtained from measurements and numerical modeling in the second layer.

Fig. 37. Comparison of pore water pressures obtained from measurements and numerical modeling for EP 336 and EP 339 in the second layer.
724 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725

Fig. 38. Drop of pore water pressure in the first layer from the sides of the concrete slab (unit: m/s).

Fig. 40. Changes in the total vertical stress measured by the pressure cells installed at
the first level at different times.

Fig. 39. Locations of total pressure cells at different levels on the critical cross-section.

Fig. 41. Changes in the total vertical stress in the dam body at the first level at different
times.
time would change the results of pore water pressure due to
the thick layers being placed and their sudden weight addi-
tion. Placing layers with the thickness more than 8 m causes (4) Before the initial impounding starts, the pore water pressure
extra pore water pressure in the initial stage of construction, is symmetrical between the upstream and the downstream.
and it is better to assume the number of layers in a way that With the increase in the distance from the clay core to the
their thickness would be between 4 m and 6 m for analyzing upstream and downstream, the pore water pressure de-
earth and rockill dams. creases. However, after impounding, the upstream clay core
 M. Rashidi, S.M. Haeri / Journal of Rock Mechanics and Geotechnical Engineering 9 (2017) 709e725
Fig. 42. Comparison of total vertical pressures from measurements and numerical
modeling for the cell PC 321 located at the first level.
Fig. 43. Changes in arching ratio in the width of the dam body at the first level at
different times.
has a higher pore water pressure in comparison with the
downstream. Also the pore water pressure of the down-
stream clay core decreases after the initial impounding It is
also noted that the clay’s water content is more than the
optimum one in some layers. This has led to the increase in
the pore water pressure in this part.
(5) The total vertical stresses, obtained from the numerical
analysis, were proved to have the same trend with the data
recorded by the pressure cells, but a small difference in
quantity was found between them. This was caused by the
local arching phenomena at the installation place of the
pressure cells, due to inadequate compaction around these
instruments which has caused a low-stress zone.
Finally, it is worth mentioning that the behavior of this dam on
its largest cross-section was reasonable in terms of settlement,
stress and pore water pressure at the end of construction and
during the initial impounding.
Conflict of interest
The authors wish to confirm that there are no known conflicts of
interest associated with this publication and there has been no
significant financial support for this work that could have influ-
enced its outcome.
View publication stats
725
References
Alonso EE, Oldecop LA. Fundamentals of rockfill collapse. In: Unsaturated soils for
Asia, Proceedings of the Asian Conference on Unsaturated Soils. Rotterdam: A.A.
Balkema; 2000. p. 3e13.
Cavounidis S, Hoeg K. Consolidation during construction of earth dams. Journal of
the Geotechnical Engineering Division, ASCE 1977;103(GT10):1055e67.
Duncan JM. State of the art: limit equilibrium and finite-element analysis of slopes.
Journal of Geotechnical Engineering 1996;122(7):577e95.
Eisenstein Z, Law STC. Analysis of consolidation behavior of Mica dam. Journal of
the Geotechnical Engineering Division, ASCE 1977;103(GT8):879e95.
Gould JP, Lacy HS. Seepage control in dam rehabilitation. In: Proceedings of the
Geotechnical Practice in Dam Rehabilitation. ASCE; 1993. p. 240e55.
International Commission on Large Dams (ICOLD). Automated observation for the
safety control of dams. 1982. p. 1e20. Paris, France.
Itasca Consulting Group, Inc. FLAC version 4 user manual. Minneapolis, USA: Itasca
Consulting Group, Inc.; 2002.
Justo JL. Collapse: its importance, fundamentals and modeling. In: Advances in
rockfill structures. Dordrecht, Netherlands: Springer; 1991. p. 97e152.
Kovacevic N, Potts DM, Vaughan PR. Finite element analysis of a rockfill dam. In:
Proceedings of the 8th International Conference on Computer Methods and
Advances in Geomechanics. Morgantown, USA; 1994. p. 2459e64.
Lambe TW. Predictions in soil engineering. Geotechnique 1973;23(2):151e202.
Luo D, Lin P, Li Q, Zheng D, Liu H. Effect of the impounding process on the overall
stability of a high arch dam: a case study of the Xiluodu dam, China. Arabian
Journal of Geosciences 2015;8(11):9023e41.
Mahab Ghodss Consulting Engineering Co. Gavoshan dam construction and oper-
ation phase of the national project. Report of the Gavoshan dam instrumen-
tation results in 2000 to 2009. Tehran, Iran: Mahab Ghods Consulting
Engineering Co; 2009.
Mahin Roosta R, Alizadeh A. Simulation of collapse settlement in rockfill material
due to saturation. International Journal of Civil Engineering 2012;10(2):93e9.
Marsal RJ, Resendiz D. Earth dams and rockfill. Mexico City, Mexico: Limusa; 1975
(in Spanish).
Naylor DJ. Collapse settlement e some developments. In: Applications of compu-
tational mechanics in geotechnical engineering. Rotterdam: A.A. Balkema; 1997.
p. 37e54.
Naylor DJ. Finite element methods for fills and embankment dams. In: Advances in
rockfill structures. Springer; 1991. p. 291e340.
Nobari ES, Duncan JM. Effect of reservoir filling on stress and movement in earth
and rockfill dam. Report TE-72-1. Department of Civil Engineering, University of
California; 1972.
Pagano L, Desideri A, Vinale F. Interpreting the settlement profiles of earth dams.
Journal of Geotechnical and Geoenvironmental Engineering 1998;124(10):923e
32.
Rezaei M, Salehi B. The effect of changing the geometry and compaction degree on
arching of earth dams. In: Geo-Frontiers Congress 2011. American Society of
Civil Engineers (ASCE); 2011. p. 3207e16.
Saboya JRF, Byrne PM. Parameters for stress and deformation analysis of rockfill
dams. Canadian Geotechnical Journal 1993;30(4):690e701.
Soroush A, Aghaei A. Uncertainties in mechanical behavior of rockfills during first
impounding of rockfill dams. In: Proceedings of the 73rd Annual Meeting of
ICOLD. Tehran: Iran; 2005. Paper No. 186eS5.
Terzaghi K. Stress distribution in dry and in saturated sand above a yielding trap-
door. In: Proceeding of the 1st International Conference on Soil Mechanics and
Foundation Engineering. Cambridge, USA; 1936. p. 307e11.
Terzaghi K. Theoretical soil mechanics. New York: John Wiley and Sons Inc; 1943.
Varadarajan A, Sharma KG, Venkatachalam K, Gupta AK. Testing and modeling two
rockfill materials. Journal of Geotechnical and Geoenvironmental Engineering
2003;129(3):206e17.
Yu Y, Zhang B, Yuan H. An intelligent displacement back-analysis method for earth-
rockfill dams. Computers and Geotechnics 2007;34(6):423e34.
Zhou W, Hua J, Chang X, Zhou C. Settlement analysis of the Shuibuya concrete-face
rockfill dam. Computers and Geotechnics 2011;38(2):269e80.
Mohammad Rashidi received his BS degree in Civil
Engineering from University of Tabriz, and his MS degree
in Civil Engineering from Sharif University of Technology.
He is currently a PhD student at the University of Texas at
El Paso. His research interests include numerical analysis,
monitoring and instrumentation, computational me-
chanics and geomechanics, dam and geotechnical engi-
neering, pavement, chemical stabilization of aggregate
soil, laboratory testing and characterization of engineering
materials. He has presented and published several scien-
tific papers in various international conferences and
journals.