INTERNATIONAL SOCIETY FOR

SOIL MECHANICS AND

GEOTECHNICAL ENGINEERING

This paper was downloaded from the Online Library of

the International Society for Soil Mechanics and
Geotechnical Engineering (ISSMGE). The library is
available here:

https://www.issmge.org/publications/online-library

This is an open-access database that archives thousands

of papers published under the Auspices of the ISSMGE and
maintained by the Innovation and Development
Committee of ISSMGE.
 Paper No. NASTU

NUMERICAL ANALYSIS OF CONCRETE FACED ROCKFILL DAMS

DURING EARTHQUAKES

1 2 4
Sinan TURKMEN , Kutay OZAYDIN , Mehmet BERILGEN3, Havvanur KILIC

ABSTRACT

In this study, the seismic behavior of a rock fill dam with concrete covered upstream slope is investigated
through numerical analysis. The lateral displacements and accelerations generated under the influence of a
set of base rock motions are computed for different reservoir water levels and ground motion intensities. It
is observed that computed displacements and accelerations are influenced by the reservoir water level,
shaking intensity and by the characteristics of the ground motion used. The behavior of same dam is also
analyzed if it was constructed without a concrete cover but as a conventional clay cored dam. The results
of the analyses have shown that larger displacements can be expected for a dam with clay core and
somewhat smaller accelerations compared with the concrete covered dam.

Key words: dam dynamic analysis, concrete faced rock fill dam, Quake/W

INTRODUCTION

Rock fill dams with concrete covered upstream slopes are being widely used all over the world. Also in
Turkey this type of dams are started to be preferred due to difficulties in finding proper core material.
Kurtun Dam is the first rock fill dam with concrete covered upstream slope built in Turkey (Kurtun Dam,
1997-2003).

Kurtun Dam is built as part of the Harsit Project being developed in Eastern Black Sea region and located
at 27 km northwest of Gumushane. The power generation capacity is 85 MW and expected to produce
198GWh electricity annually, constituting 8.2% of the total energy generation capacity of Harsit Project.
Rock fill embankment construction has started in 1997 and completed on 28.04.1999. The construction
was halted for about 18 months before the construction of concrete membrane cover and after the
completion of concrete cover, the parapet wall on the dam crest and the fill behind that were completed.
The water filling of the reservoir started on 08.02.2002 and the water level reached +630.0m elevation on
28.05.2002. The height of Kurtun Dam is 133m above the stream bed, the upstream embankment slope is
1.4(H):1(V) and downstream slope is 1.5(H):1(V), with a crest length of 300m. The dam is constructed
over a narrow valley, the stream bed is 40m wide, the average side slope angles of the left and right sides
of the valley are 610 and 520, respectively.

1
Civil Engineer, Msc., Department of Civil Engineering, Yildiz Technical University, Fax:
+90.212.383 51 33
2
Professor, Department of Civil Engineering, Yildiz Technical University, e-mail:
[email protected]
3
Assoc. Professor, Department of Civil Engineering, Yildiz Technical University, e-mail:
[email protected]
4
Assist. Professor, Department of Civil Engineering, Yildiz Technical University, e-mail:
[email protected]
 5th International Conference on Earthquake Geotechnical Engineering
January 2011, 10-13
Santiago, Chile
A view of Kurtun Dam from upstream is shown in Figure 1, (DSI, 2003). The main geological formations
at the dam site are granodiorite, diabase, andezite and limestone. The dam is seated on granodiorite
formation. The cross section of the dam and kinds of fill material used are shown in Figure 2, and the
properties of fill material are given in Table 1.

Figure 1. View of Kurtun dam from the upstream (DSI, 2003)

Figure 2. Cross section and fill material of Kurtun Dam (Ozkuzukiran, 2005)
Table 1. Properties of Kurtun Dam Fill Materials (Ozkuzukiran, 2005)
 5th International Conference on Earthquake Geotechnical Engineering
January 2011, 10-13
Santiago, Chile

Two Dimensional Dynamic Analyses And Earthquake Records Used

Two dimensional dynamic analysis of the dam is performed using Quake/W (2007) software utilizing
equivalent linear analysis technique. The earthquake motion is applied at the base of the dam where the
nodes are fixed in both directions. In order to investigate the effect of stored water on the dynamic behavior
of the dam dynamic analyses are performed for both empty and full reservoir conditions. The computed
values of accelerations, displacements and stresses at various points on the dam body are compared. For
the full reservoir condition 111m water height is considered and the hydrodynamic effect of the water is
taken into account using the relationship proposed by Westergaard (1931) for impervious surfaces.

(1)

In the above, p(z) is the hydrodynamic pressure at depth z from the water surface, a is the instantaneous
horizontal acceleration, γ is unit weight of water, h is the water height. Because the value of a will change
with time and z changes with depth, during the earthquake the value of the hydrodynamic water pressure
acting on the dam will change continuously. Therefore the values of water pressure are computed for each
of the nodes in contact with water.

In the dynamic analysis of Kurtun Dam, real earthquake motions recorded at base rock level are used. The
nearest strong motion record, 1992 Erzincan earthquake horizontal motion, and two other strong motion
records during 1994 Northridge and 1989 Loma Prieta earthquakes are used in the dynamic analyses. The
 5th International Conference on Earthquake Geotechnical Engineering
January 2011, 10-13
Santiago, Chile
characteristics of the earthquake motions used are summarized in Table 2, and the records are shown in
Figure 3. The east-west component of the Erzincan acceleration record with 0.01s time intervals had
duration of 15s and maximum value of 0.49791g, whereas E-W Northridge record had 15s duration, 0.55g
maximum, and Loma Prieta record had 10s duration and 0.3484 maximum value.

Table 2. Properties of Earthquake Records Used in the Analyses

Record Max. acceleration (g) Mw (Magnitude)
1992 Erzincan Earthquake 0.4979 6.9
1994 Nortridge Earthquake 0.5500 6.7
1989 Loma Prieta Earthquake 0.3484 6.9

Figure 3. Base rock motions used in dynamic analyses a)

1992 Erzincan strong motion record ( EW)
b) 1994 Northridge strong motion record (EW)
c) 1989 Loma Prieta strong motion record (NS)

The materials except those in zones 3B and 3C are shown to have very little effect on the dynamic
behavior of the dam (Saboya et al, 1993) therefore to simplify the analysis it is assumed that the dam is
composed of only the materials 3B and 3C. The material properties used are given in Table 3 (Turkmen,
2009).

Table 3. Properties of embankment materials

γ E G Poisson’s
(kN/m )
3 (kPa) (kPa) Ratio
3B Material 21.0 2.444.200 977.680 0.25
3C Material 21.0 2.506.600 1.002.600 0.25
Concrete Cover 23.5 28.500.000 12.076.000 0.18

In order to evaluate the dynamic behavior of the dam and understand the effect of various
parameters, the values of horizontal acceleration and horizontal displacements computed through
dynamic analysis for points A, B, C, D, E, F and G on the dam body as shown in Figure 4 are
considered. Point A, C and G are on the concrete covered upstream face, points B and D are on
the downstream face, point E is on the crest and point F at the base of the dam.
 5th International Conference on Earthquake Geotechnical Engineering
January 2011, 10-13
Santiago, Chile

Figure 4. Cross section of the dam and calculation output points (Türkmen, 2009)

Results of Dynamic Analysis Using 1992 Erzincan Earthquake Record

The results of dynamic analyses for the cases of empty and full reservoir conditions using 1992
Erzincan Earthquake (EW) record are shown as change of acceleration and horizontal displacement
with time, at various points on the dam, in Figures 5 and 6, respectively.

Figure 5. For empty reservoir condition- at points A, B, C, D and G a) horizontal acceleration-time

history b) horizontal displacement-time history

Figure 6. For full reservoir condition- at points A, B, C, D and G a) horizontal acceleration-

time history b) horizontal displacement-time history
 5th International Conference on Earthquake Geotechnical Engineering
January 2011, 10-13
Santiago, Chile
As it is observed from Figure 5a, when the reservoir is empty smaller accelerations develop at points A, C
and D on the concrete covered face than at the points B and D on the downstream side. When the reservoir
is full, it is observed from Figure 6a that except a very high initial value at point G just above the base
(which is probably due to fixed boundary condition imposed at the base) somewhat higher accelerations
develop on the concrete covered upstream face. In Figures 5b and 6b variation of horizontal displacements
computed to develop at various points on the dam for the cases of empty and full reservoir conditions,
respectively, are illustrated. As it would be expected, displacements are increased when the reservoir is
full due to the hydrodynamic pressures acting on the dam, and for the full reservoir condition
displacements on the concrete covered upstream face are increasing with depth and larger than the
downstream side. For points A and B which are on two sides of the crest and above the water level in the
reservoir, computed accelerations and displacements for both empty and full reservoir conditions are
about the same, and for the both points displacements are increased when the reservoir is full but the
accelerations are smaller. In Table 4 the maximum values of computed accelerations and horizontal
displacements at various points on the dam for both empty and full reservoir conditions are summarized.

Table 4. Maximum horizontal acceleration and displacement values at various points on the dam
(1992 Erzincan Earthquake)
Point Empty reservoir Full reservoir
dmax (m) amax(g) dmax (m) amax(g)
A 0.146 0.829 0.204 0.700
B 0.151 0.849 0.225 0.700
C 0.100 0.634 0.375 0.844
D 0.147 0.790 0.281 0.638
(1)
G 0.017 0.276 0.663 3.443
(2)
0.209
(1) high initial value (2) max. value after 1 s

In the investigation of the dynamic behavior of the dam, in addition to the acceleration and displacement
values computed at various points on the surface of the dam, the effects along a vertical section through
the centerline (E-F) are also examined. In Figure 7 the variation computed maximum horizontal
accelerations and displacements along the centerline of the dam for both empty and full reservoir
conditions are shown.

Figure 7. Variation with height along mid vertical section (E-F) a) maximum horizontal
acceleration b) maximum horizontal displacement (1992 Erzincan Earthquake)

As it can be observed from Figure 7a and 7b smaller acceleration but larger displacement values are
obtained along the mid vertical section of the dam for the full reservoir condition compared to the empty
reservoir case. Hydrodynamic pressures cause an increase in the displacements, and for both empty and
 5th International Conference on Earthquake Geotechnical Engineering
January 2011, 10-13
Santiago, Chile
full reservoir conditions the displacements are observed to increase with depth from the crest level but
then start to decrease towards the base because the nodes at the base are fixed in the numerical model.

Results Of Dynamic Analysis With 1994 Northridge Earthquake Record

The maximum horizontal acceleration and displacement values computed as a result of dynamic analysis
using 1994 Northridge Earthquake (EW) record are summarized in Table 5.

Table 5. Maximum horizontal acceleration and displacement values at various points on the dam
(1994 Northridge Earthquake)
Point Empty reservoir Full reservoir
dmax (m) amax(g) dmax (m) amax(g)
A 0.070 0.702 0.114 0.608
B 0.072 0.695 0.120 0.613
C 0.048 0.646 0.202 0.691
D 0.060 0.681 0.132 0.590
(1)
G 0.007 0.213 0.614 3.063
(2)
0.336
(1) high initial value (2) max. value after 1 s

It is observed that displacements are increased when the reservoir is full due to hydrodynamic pressures.
The effect of full reservoir is not very clear on the accelerations developed at different points of the dam.
The very high value of the acceleration at the start of the excitation at point G near the base is attributed to
fixed boundary condition assigned to the base nodes. The variation of horizontal accelerations and
displacements along the height of the dam at mid section for both empty and full reservoir conditions are
shown in Figure 8.

Figure 8. Variation with height along mid vertical section (E-F) a) maximum horizontal
acceleration b) maximum horizontal displacement (1994 Northridge Earthquake)

From Figure 8a it is noticed that presence of full reservoir has a slight effect on accelerations but a very
strong effect on displacements. Increasing displacements with depth from crest of the dam start to decrease
towards the base where the nodes are assumed to be fixed.

Results Of Dynamic Analysis With 1989 Loma Prieta Earthquake Record

The maximum horizontal acceleration and displacement values computed as a result of dynamic analysis
using 1989 Loma Prieta Earthquake (NS) record as base excitation are summarized in Table 6. The effect
of full reservoir is again much more pronounced on displacements than on accelerations at various points
on the dam.
 5th International Conference on Earthquake Geotechnical Engineering
January 2011, 10-13
Santiago, Chile
Table 6. Maximum horizontal acceleration and displacement values at various points on the dam
(1989 Loma Prieta Earthquake)
Point Empty reservoir Full reservoir
dmax (m) amax(g) dmax (m) amax(g)
A 0.079 0.507 0.110 0.480
B 0.080 0.526 0.116 0.487
C 0.056 0.402 0.206 0.693
D 0.073 0.471 0.127 0.425
(1)
G 0.009 0.252 0.678 3.122
(2)
0.260
(1) high initial value (2) max. value after 1 s
The variation of horizontal accelerations and displacements with height along mid section of the dam
shown in Figure 9 also illustrates the strong effect of hydrodynamic pressures on the displacements
expected to occur by the dam during earthquake shaking.

Figure 9. Variation with height along mid vertical section (E-F ) a) maximum horizontal
accelerations b) maximum horizontal displacements (1989 Loma Prieta Earthquake)

Effect of Shaking Intensity On Dynamic Behavior Of The Dam

In this investigation strong motion records from 3 different earthquakes are used to analyze the dynamic
behavior of a concrete faced rock fill dam. Two of these records (1992 Erzincan and 1994 Northridge)
had close maximum horizontal acceleration values (0.50g and 0.55g, respectively) and same duration
(15s). The third one (1989 Loma Prieta) had lower shaking intensity (amax=0.35g) and shorter duration
(10s). Results of dynamic analyses have shown that dynamic behavior is affected by the base excitation
used. In order to study the effect of shaking intensity alone, same earthquake record (1989 Loma Prieta) is
used with maximum acceleration being scaled to three different values (0.20g, 0.35g and 0.50g) and the
results are compared. In Figure 10 the variation of computed horizontal displacements with height along
the mid section of the dam subjected to different shaking intensities are shown for empty and full
reservoir conditions.
 5th International Conference on Earthquake Geotechnical Engineering
January 2011, 10-13
Santiago, Chile

Figure 10. Variation with height of maximum horizontal displacement values using Loma
Prieta earthquake record scaled to 0.20g, 0.35g and 0.50g a) empty reservoir condition
b) full reservoir condition

It is observed that the maximum horizontal displacements along the mid vertical section of the dam are
affected by the shaking intensity. The computed values of maximum horizontal displacements for empty
and full reservoir conditions are summarized in Table 7.

Table 7. Maximum horizontal displacement values computed by using same earthquake motion
scaled to different maximum acceleration values
Maximum horizontal displacement (m)
amax (g)
Empty reservoir Full reservoir
0.20 0.051 0.100
0.35 0.081 0.163
0.50 0.108 0.273

Dynamic Analysis of The Dam Without A Concrete Surface Cover

The probable dynamic behavior of the same dam if it was built with a clay core rather then with a concrete
surface cover on the upstream side is also investigated for the sake of comparing the behavior of these two
different dam types. The section of the dam analyzed is shown in Figure 11 and the material properties
used are summarized in Table 8.

Figure 11. Clay cored dam section

 5th International Conference on Earthquake Geotechnical Engineering
January 2011, 10-13
Santiago, Chile
Table 8. Properties of fill material for clay cored dam (Louadj et al, 2007)
E G
Fill type Poisson’s Ratio
(MPa) (MPa)
1 1383.7 519.02 0.33
2 2330.0 932.00 0.25
3 2444.2 977.68 0.25
4 2506.6 1002.60 0.25

In the dynamic analysis 1992 Erzincan Earthquake (EW) record is used as the base excitation. In Table 9
the computed maximum horizontal acceleration and displacement values at different points on the dam for
the cases of both empty and full reservoir conditions are summarized.

Table 9. Maximum horizontal acceleration and displacement values at various points on the dam
(1992 Erzincan Earthquake)
Empty reservoir Full reservoir
Point
dmax (m) amax(g) dmax (m) amax(g)
A 0.231 0.622 0.427 0.516
B 0.232 0.622 0.441 0.506
C 0.151 0.614 0.729 0.723
D 0.169 0.625 0.300 0.594
(1)
G 0.013 0.308 1.225 3.908
(2)
0.574
(1) high initial value (2) max. value after 1 s

It is observed that the horizontal displacements are very much influenced by the hydrodynamic pressures
caused by the water in the reservoir, especially for the upstream side. Also for the points A and B which
are above the water level, the computed displacements are almost doubled due to hydrodynamic pressures
acting on the dam body. On the other hand the influence of water on horizontal accelerations expected to
develop is much less. The variation of horizontal accelerations and displacements with height along the
mid vertical section of the dam shown in Figure 12 illustrates the same trends.

Figure 12. Variation with height along mid vertical section (E-F) a) maximum horizontal
acceleration b) maximum horizontal displacement (1992 Erzincan Earthquake)

Comparison of the dynamic analyses results summarized in Table 4 and Table 9 reveals that larger
horizontal displacements can be expected to occur during an earthquake on a clay cored dam compared to
a concrete faced rock fill dam, whereas somewhat smaller accelerations are expected to develop on a clay
cored dam.
 5th International Conference on Earthquake Geotechnical Engineering
January 2011, 10-13
Santiago, Chile
RESULTS AND CONCLUSIONS

Based on the results of numerical analyses carried out for this investigation the following conclusions can
be reached.
1) When the results of dynamic analyses on concrete faced rock fill dams using two different
earthquake records of similar maximum acceleration and duration (1992 Erzincan and 1994
Northridge) as base excitation are compared:
(a) computed horizontal accelerations and displacements for both empty and full reservoir
conditions are noted to be somewhat larger under the influence of 1992 Erzincan
Earthquake, which can be an indication that properties of the earthquake motion other then
the maximum acceleration might be also be influential on dynamic behavior;
(b) for both earthquake motions the computed displacements are found to be much larger when
the reservoir is full compared to those for empty cases, which considered to be a strong
indication for the influence of hydrodynamic pressures on the dam behavior during an
earthquake.
2) When the results of dynamic analyses on concrete faced rock fill dams using the same
earthquake record (1989 Loma Prieta) scaled to three different maximum acceleration values
( 0.20g, 0.35g and 0.50g) as base excitation are compared:
(a) for both empty and full reservoir conditions computed maximum horizontal displacements
are observed to get larger with increasing shaking intensity which is clearly demonstrated
by the figures showing their variation with height along the mid vertical section of the dam;
(b) the computed horizontal displacements are observed to be much larger for the case of full
reservoir conditions and increase with depth from the water surface as the hydrodynamic
pressures increase, but approaches zero after a certain depth which is due to fixed boundary
conditions imposed on the nodes at the base;
3) When the results of dynamic analyses using same base excitation on concrete faced and clay
cored dams of same geometry are compared:
(a) for both empty and full reservoir conditions computed horizontal displacements are noted to
be considerably larger for the clay cored dam than the concrete faced dam with similar
variation with height at mid vertical section of the dam;
(b) for both empty and full reservoir conditions computed horizontal accelerations are noted to
be smaller for the clay cored dam compared to the concrete faced dam.

REFERENCES

DSI (1997- 2003) Kurtun Dam, Ankara (in Turkish)

GEO-SLOPE (2007) Dynamic Modelling with QUAKE/W an Engineering Methology, Third Edition,
Canada.
Louadj, S., Bahar, R., Laouami, N., Slimani, A., (2007), “Response of a Rockfill Dam Spatially
Varying Earthquake Ground Motion”, 4th International Conference on Earthquake Geotechnical
Engineering, 1181.
Özkuzukuran, R.S., (2005), “Settlement Behaviour of Concrete Faced Rockfill Dams: A Case Study”,
Orta Dogu Teknik Üniversitesi.
Saboya, F. Jr., and Byrne, P. M., (1993) “Parameters for Stress and Deformation Analysis of Rockfill
Dams”, Can. Geotech. J., 30, 690-701.
Strasser, M., Stegmann, S., Bussmann, F., Anselmetti, F.S., Rick, B., Kopf, A., (2007), “Lake Lucerne
as Model for Ocean Margins”, Marine Geology.
Westergaard, H.M., (1931), Water Pressures on Dams During Earthquakes, ASCE Transactions,
No:1835, 418-433 pp.
Türkmen (2009) “Analysis of slope stability during earthquake” MSc thesis, Yildiz Technical Univ.,
Science and Technology Institute, Istanbul, Turkey (in Turkish).