You are on page 1of 9

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/271138171

Optimization and safety evaluation of concrete


gravity dam section

Article in KSCE Journal of Civil Engineering · January 2015


DOI: 10.1007/s12205-015-0139-0

CITATIONS READS

4 365

3 authors:

Arnab Banerjee D. K. Paul


University of Auckland Indian Institute of Technology Roorkee
21 PUBLICATIONS 35 CITATIONS 70 PUBLICATIONS 892 CITATIONS

SEE PROFILE SEE PROFILE

Arijit Acharyya
Jadavpur University
1 PUBLICATION 4 CITATIONS

SEE PROFILE

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

Seismic pounding analysis of curved bridges and contact mechanics View project

Seismic analysis of concrete gravity dams View project

All content following this page was uploaded by Arnab Banerjee on 22 January 2015.

The user has requested enhancement of the downloaded file.


KSCE Journal of Civil Engineering (0000) 00(0):1-8 Geotechnical Engineering
Copyright ⓒ2014 Korean Society of Civil Engineers
DOI 10.1007/s12205-015-0139-0 pISSN 1226-7988, eISSN 1976-3808
www.springer.com/12205
TECHNICAL NOTE

Optimization and Safety Evaluation of Concrete Gravity Dam Section


Arnab Banerjee*, D. K. Paul**, and Arijit Acharyya***
Received March 18, 2014/Revised July 31, 2014/Accepted August 12, 2014/Published Online January 21, 2015

··································································································································································································································

Abstract

Concrete gravity dam is a very important civil engineering structure from hydro power and irrigation point of view. It stores
a massive amount of water in its reservoir. Dam failure can cause enormous destruction in the downstream. That is why, the
safety of dam has paramount importance. The objective of this paper is divided into two categories. First of all a cost based
dam section optimization or design program, named “Optidam”, is presented. The Optidam program is able to design an
optimized concrete gravity dam section based on pseudo static analysis proposed in Indian standard. Then one dam section,
which is designed using the “Optidam”, is checked and analyzed by non-linear seismic analysis to ensure its safety. Finally, a
parametric analysis is performed to propose a dam design guidelines. Empirical relationships between the different geometric
parameters of an optimized concrete gravity dam section with cohesion and internal friction angle of soil/rock foundation are
evaluated after designing 1080 number of dams, having height ranges from 50-300 m, and situated on different foundation
characteristics.
Keywords: concrete gravity dam, optimization, Dam design guidelines, linear and non-linear seismic analysis, parametric analysis,
finite element analysis
··································································································································································································································

1. Introduction pseudo dynamic analysis of a dam structure. Indian standard (IS


6512, 2003), US Army (Engineer Manual, 1995) gives the guidelines
Hydrodynamic effect of reservoir can be considered as virtual for pseudo static analysis. In this paper, Opti-dam program has
lumped mass on the upstream face of the dam (Westergaard, 1933; been developed based on pseudo static analysis given in
Zangar, 1952) or model the upstream impounded reservoir water IS:6512-2003 (IS 6512, 2003). The Opti-dam program generates
by Eulerian approach (Calayir and Karaton, 2005; Bayraktar et al., optimum dam sections for a particular dam height when
2011; Zienkiewicz et al., 1983; Kalateh and Attarnejad, 2011) or subjected to all known of possible loading. The generated dam
Lagrangian approach (Wilson and Khalvati, 1983). section is also checked by linear and non-linear finite element
Kupfer and Gerstle (Kupfer and Gerstle, 1973) proposed the analysis.
behavior of concrete under biaxial stresses along with failure
envelops for different types of concrete based on experiments. 2. Pseudo Static Analysis as per Indian Standard
Lubliner et al. (Lubliner et al., 1989) suggested the Barcelona
model to evaluate fracture energy based damage variable to A concrete gravity dam section can be analyzed as a block,
stimulate the degradation of elastic stiffness. Lee and Fenves considering all the forces acting on it. Pseudo static analysis is
(Lee and Fenves, 1998) proposed plastic damage model for very simple and effective analysis technique for safety evaluation
cyclic loading of concrete structure. The stiffness degradation of a concrete gravity dam.
due to tension cracking and compression recovery can be There are 7 load combinations prescribed to ensure the
successfully modeled by this model. In this paper, the plastic stability analysis of a dam in Indian Standard. Table 1 gives a
damage model (Lee and Fenves, 1998) and the stress-strain summary of the load cases along with maximum permissible
curve are proposed by Sima et al. (Sima et al., 2008) have been tensile stress (IS 6512, 2003).
used to evaluate the damage. The calculated factor of safety for sliding (F) according to the
Leclerc et al. (Leclerc et al., 2003) developed a program, Eq. (1) should be always greater than 1 for a safe section (IS
known as CADAM program, to perform, pseudo static as well as 6512, 2003):

*Ph.D. Student, Dept. of Mechanical Engineering, University of Aukland, Aukland 1010, New Zealand (Corresponding Author, E-mail: abanerjee24@
gmail.com)
**Emeritus Fellow, Dept. of Earthquake Engineering, Indian Institute of Technology, Roorkee, 247667, Indian Institute of Technology, Roorkee 247667,
Indi (E-mail: dpaulfeq@gmail.com)
***Assistant Engineer, Public Health Dept., Government of West Bengal, West Bengal 711315, India (E-mail: arijit.acharyya@gmail.com)

−1−
Arnab Banerjee, D. K. Paul, and Arijit Acharyya

Table 1. Different Load Cases


Self weight Upstream Tail water Ice and wave Permissible tensile
Case No. Silt pressure Uplift pressure Earthquake load
of dam reservoir level level pressure stress
A Yes X X X X X X 0
B Yes Normal Yes Normal Yes Drained X 0
C Yes Flood Yes Flood X Drained X 0.01fc
D Yes X X X X X Yes 0
E Yes Normal Yes Normal X Drained Yes 0.02fc
F Yes Flood Yes Flood X Undrained X 0.02fc
G Yes Normal Yes Normal X Undrained Yes 0.04fc

Table 2. Factor of Safety


Load combinations Fφ Fc
A, B, C 1.5 3.6
D, E 1.2 2.4
F, G 1.0 1.2

( W – u )tanφ cA
---------------------------- + ------
Fφ Fc
F = ----------------------------------------- (1)
P
where, W is the total weight of the dam, u is the uplift force, c is
the cohesion, P is the total horizontal force, Fφ is the factor of Fig. 1. Typical Dam Section
safety against internal friction angle, Fc is the factor against
safety for cohesion. The values of and Fc are tabulated in Table 2.
The vertical stress (σ) in any section is calculated by: user, remains constant throughout the iteration process. All other
geometric parameters such as base (B), base width of upstream
W ΣM
σ = – ----- ± -------- (2) fold (Bu) and height of upstream fold (Hu) are changed in a
A Z
systematic manner to reach the optimized section.
where, M is the moment and Z is the section modulus. Calculated Prior to initiate the optimization iteration, a domain of possible
stress should be less than the permissible tensile stress for that values of each variable is determined according to the following
case as given in Table 2. formulas:
{ B } = [0.7H, 1.5H ]
3. Optimization of Dam Section and Optidam
{ Hu } = [ 0, ( H – 10 )] (4)
Programming
{ Bu } = [0.05Hu , 0.4Hu ]
Optimization of a concrete gravity dam is achieved by Optimization of these 3 variables has to satisfy a set of 14
minimizing the involved cost of construction. In this paper, the two inequalities to ensure it’s safety. In each step for every load
main costs related to the dam section geometry are considered for combination one safe dam section should pass the criteria of
example, the cost of concreting, and the cost of foundation Eqs. (1) and (2).
excavation. Due to simplicity cost of 1 m3 concerting is assumed to
be (Cc) a constant value, does not change with the elevation. 3.2 Optidam Programming
Similarly the cost of 1m excavation is also assumed constant (Cf). An excel-MATLAB based programming, named Optidam, is
So, the Total Cost (TC) of 1m thick dam monolith can be generated to optimize the dam section. To make the program
calculated by: user friendly Microsoft excel is used as an input for the Opti-dam
program. The program can take the input from that excel sheet
TC = Cc A + Cf B (3)
named ‘Optidam.xls’, ‘Sheet1’. The output is also come in the
So, the objective function of optimization is to minimize TC. same excel ‘Optidam.xls’, ‘out’. A screen shot of the input and
output is given in Fig. 2.
3.1 Geometric Parameter of a Dam Section The algorithm of Optidam is illustrated in Fig. 3.
A typical concrete gravity dam section, pointing all the geometric
parameters involved in this optimization study, is depicted in Fig. 1. 3.3 Numerical Example
In the Optidam program total height of the dam (H) is kept A 240 m high dam is to be designed for zone V. The cohesion
constant. And the crest width (C) and depth (Cd), can be given by (c) and the friction angle (φ) of the dam-foundation interface is

−2− KSCE Journal of Civil Engineering


Optimization and Safety Evaluation of Concrete Gravity Dam Section

Fig. 2. Optidam-Program Input and Output

be INR 10000 and cost of 1m excavation is taken as INR 1000.


The optimization is carried out based on the above algorithm
which gave the following optimized section.
i) Total base width is 243 m
ii) The crest width is 10 m
The optimized section is shown in Fig. 4.

4. Validation of Optidam

The dam section, designed by Optidam, is validated by


performing linear and nonlinear finite element analysis in time
domain. To perform finite element analysis the dam-foundation-
reservoir system is modeled using 2D plane-stress elements for
the dam body and 2D plane-strain element for the foundation.
Impounded reservoir water is modeled as Eularian system, using
Fig. 3. Flow Chart acoustic elements in ABAQUS, for performing the dynamic
analysis. The compressibility of reservoir water is also considered
200 kPa and 350. The normal reservoir level is 228 m and flood in time history analysis by acoustic model. To simulate the radiating
reservoir level is 234 m. Height of silt is 100 m above the boundaries, elementary boundary is replaced by absorbing
reservoir bottom level. The cost of 1 m3 concrete is assumed to boundary, using orthogonal viscous dampers, in soil domain, and
in fluid domain impedance is assigned.
The material property are tabulated in Table 3.
To perform the non-linear time history analysis non-linear
material properties are used (Sima et al., 2008). The stress-strain
curve of concrete employed for analysis is shown in Fig. 5.
To evaluate the damage, the concrete damage plasticity model
(Lee and Fenves, 1998) is used.
The damage in tension of the concrete is calculated using the

Table 3. Material Properties


Density Young Modulus Poisson Damping
Material
(kg/m3) (GPa) ratio (%)
Concrete 2400 22.36 0.16 5
Foundation 2600 5 0.3 7
Fig. 4. Optimized Dam Section

Vol. 00, No. 0 / 000 0000 −3−


Arnab Banerjee, D. K. Paul, and Arijit Acharyya

Fig. 5. Stress-Strain Curve of Concrete: (a) Compressive Stress vs. Strain, (b) Tensile Stress vs. Strain

After the failure stress, crack initiates with softening of stress-strain


curve.
The uni-axial compression stress-strain curve shows a stress
hardening after initial yield. And stress softening after maximum
stress.
ABAQUS is not able to take direct stress-strain curve for
performing Concrete Damage plasticity model. Inelastic strain
and corresponding stresses should be the input for this.
Elastic strain as shown in Figs. 8 and 9 are calculated as:
σ
εel = ----- (6)
E0
Fig. 6. Concrete Compression Damage Now inelastic strain ( ε̃in ) or cracking strain is computed using:
ε̃in = ε – εel (7)
Eq. (5):
σ
δ = 1 – -------- (5)
E0 ε
The damage in compression initiate after reaching the peak
stress. So, when ( ε < εc ′ ) no crack develops. The inelastic damage
curve for the concrete for compression and tension are given in
Figs. 6 and 7 respectively.

4.1 Concrete Damage Plasticity Model


Under uni-axial tension, stress-strain response follows a linear
elastic relationship upto the failure stress (peak stress) as in Fig. 5.
Fig. 8. Compressive Stress-Strain Curve

Fig. 7. Concrete Tension Damage Fig. 9. Tensile Stress-Strain Curve

−4− KSCE Journal of Civil Engineering


Optimization and Safety Evaluation of Concrete Gravity Dam Section

Fig. 12. De-convoluted Accelerogram

Table 4. Limiting Tensile Stress


Fig. 10. Cyclic Stress-Strain Curve (SIMULIA, 2009) Chopra (1986)(Chopra, December, 1986)
Case
Permissible tensile stress (psi) M20 (MPa)
Plastic strain is calculated by: Static ft = 1.7 fc2/3 2.38
DBE ft = 2.6 fc2/3 3.65
δ σ
εpl = ε̃in − --------------- ----- (8) MCE ft = 3.4 fc2/3 5.77
( 1 – δ ) E0
To simulate the actual behavior in cyclic load stiffness recovery
response spectrum is deconvoluted upto the depth of the
factor for tension and compression are assign to 0 and 1
considered rock foundation. Response compatible artificial
respectively. The cyclic stress strain curve is plotted in Fig. 10.
acclerogram is generated using (Seismo Soft, 2010), has shown in
Fig. 11 and the deconvoluted motion at the basebottom of the
4.2 Selection of Ground Motion
considered rock foundation (300m depth) is plotted in Fig. 12.
Response spectrum of Eurocode 8 (1994) is taken for analysis.
The response spectrum is specified for rock outcrop. The rock
4.3 Results and Discussions
foundation is also considered with the dam body. So, the
The limiting tensile stresses for concrete gravity dams for
different loading cases (Engineer Manual, 1995; Chopra, 1986)
are tabulated in Table 4.

4.3.1 Static Analysis


The stress diagram for static condition is plotted in Fig. 13.
The limiting tensile stress for the static condition is 0.1fc. The
maximum principal stress contours in Fig. 13(a) are within the
limit given in Table 4. Fig. 13(b) shows that the vertical stress
throughout the dam section for static loading is in compression,
so, the section is safe according to Table 1 case C (IS 6512,
Fig. 11. Response Compatible Accelerogram 2003).

Fig. 13. Static Stresses: (a) Maximum Principal Stress, (b) Vertical Normal Stress

Vol. 00, No. 0 / 000 0000 −5−


Arnab Banerjee, D. K. Paul, and Arijit Acharyya

Fig. 14. Factor of Safety of the Dam


Fig. 15. Tensile Damage at Heel

That is why, it can be concluded that the section is safe in static


analysis. MCE conditions.
The minimum factor of safety for sliding is 1.62 and 1.07 for
4.3.2 Linear Time History Analysis DBE and MCE condition respectively. The factor of safety for
To evaluate the stability of the dam section factor of safety is static condition is 2.17. The factor safety calculated by IS:6512-
calculated throughout the time domain as follows: 2003 is 2.05. The difference comes due to the use of factor of
safety in internal friction angle (φ) and cohesion (c).
( w – u)tanφ + cA
FOS = ---------------------------------------- (9)
P
4.3.3 Non-linear Time History Analysis
where, w is the resultant downward force, u is the uplift force, c Non linear material propertiesare assigned in the time history
is the cohesion, φ is internal friction angle, and P is the total analysis. The cumulative tensile damage at the heel is plotted in
horizontal force. Dividing the numerator and denominator by Fig. 15. Fig. 16 shows the damage when subjected to DBE and
area Eq. (9) modifies to: MCE level of earthquake.
As per the definition of damage index in Eq. (5), if the damage
Σσ22 tanφ + Σc
FOS = ----------------------------------
- (10) index is 0 then there is no yielding of concrete occurs and tensile
Σσ12
stiffness of the element completely lost. In the contrary, the
The summation is over all the elements at the base of the dam. If damage index between 0 to 1, indicates the plastic strain
size of the elements are same then dividing the numerator and softening-region of tensile stress-strain diagram, in which crack
denominator of the Eq. (10) by number of elements gives: initiates but element does not loss its full tensile stiffness. The
new stiffness, which is calculated after damage, is considered in
Σσ22 tanφ
-+c
--------------------- the subsequent stages. The damage index of the dam under DBE
n
FOS = ------------------------------ (11) and MCE condition are also shown in Fig. 15.
Σσ 12
----------- From Fig. 16 it can be observed that the tensile damage
n extends upto 7 m under DBE level of earthquake, and during
The ratio of the stabilizing and destabilizing stresses give the MCE level of shaking, damage extends upto15m in the upstream
factor of safety for sliding and is plotted in Fig. 14 for DBE and face, but the stability of the section remains unaffected.

Fig. 16. Tensile Damage: (a) DBE + Static, (b) MCE + Static

−6− KSCE Journal of Civil Engineering


Optimization and Safety Evaluation of Concrete Gravity Dam Section

4.3.4 Concluding Remarks (H, c, φ). 1080 no of dam sections of different heights, ranging
According to the previous discussion it can be concluded that from 50-300 m, internal friction angle, ranging from 350-500
the dam section generated by Optidam program is statically and with 2.50 interval, and cohesion of dam-foundation interface,
seismically safe. Stability of the dam is affected neither at MCE ranging from 100-600 kPa with 100 kPa interval, are designed
nor at DBE level of earthquake. using Optidam programing. Some assumptions made before
designing those sections are listed below.
5. Guideline to get an Optimized Section
a. Height of the upstream normal reservoir level is 0.95 times
The geometry of a concrete gravity dam depends on the height of the total dam height.
of a dam cohesion and internal friction angle of soil/rock b. Height of the upstream flood reservoir level is 0.975 times
foundation. A parametric analysis is performed on the different of the total dam height.
shape parameters of an optimized dam section such as base c. Tail water level is set to be zero for normal and 0.1 of the
width (B), base width of the upstream fold (Bu) and height of the total dam height for flood.
upstream fold (Hu). The main objective of the parametric d. Position of gallery is at one-sixth distance of the base from
analysis is to find empirical relationships between the unknown the heal
shape parameters (B, Bu, Hu) with the given constant parameters e. Height of the silt is half of the total height

Fig. 17. Parametric Analysis of Dam

Vol. 00, No. 0 / 000 0000 −7−


Arnab Banerjee, D. K. Paul, and Arijit Acharyya

f. Peak Ground Acceleration (PGA) is assumed to be 0.36 g. References


The variation of total base width (B), height (Hu) and base (Bu)
of upstream fold with the total height of the dam for those Bayraktar, A., Sevim, B., and Can Altuniçs ik, A. (2011). “Finite element
model updating effects on nonlinear seismic response of arch dam–
sections are plotted in Fig. 17.
reservoir–foundation systems.” Finite Elements Anal. Des., Vol. 47,
After a regression analysis, a predictive relation is obtained: No. 2, pp. 85-97, DOI: 10.1016/j.finel.2010.09.005.
–3
H = [3 × 10 cφ + 0.0147φ – 0.007c + 0.3597 ]B Calayir, Y. and Karaton, M. (2005). “Seismic fracture analysis of concrete
gravity dams including dam–reservoir interaction.” Comput. Struct.,
+ [– cφ + 0.106φ + 6c + 12.19 ] (12) Vol. 83, No. 19, pp. 1595-1606, DOI: 10.1016/j.compstruc.2005.02.003.
where, cohesion (c) is expressed in MPa, internal friction angle Chopra, A. K. (1986). “Earthquake analysis, design and safety evaluation
of concrete gravity dams.” Bulletin of Indian Society of Earthquake
(φ) in degree, H is the height of the dam and B is the base width
Technology, Vol. 23, No. 4, pp. 147-199.
of the dam expressed in meter.The average value of error of Eq. Engineer Manual (1995). Gravity dam design.
(12) over the 1080 number of data points is R2 = 0.13. Eurocode 8 (1994). Design provisions for earthquake resistance of
The upstream base width (Bu) of the optimized section is structures, Vol. 2, pp. 1998-1992.
obtained as: IS 6512 (2003). Criteria for design of solid gravity dams, Indian
Standard, New Delhi.
H = [–6.98cφ – 0.25φ + 260c + 19.6 ]Bu Kalateh, F. and Attarnejad, R. (2011). “Finite element simulation of
+ [7.23cφ + 9.27φ – 113.8c – 378.4 ] (13) acoustic cavitation in the reservoir and effects on dynamic response
of concrete dams.” Finite Elements Anal. Des., Vol. 47, No. 5, pp.
And the height of the upstream face (Hu) is almost 2.5 times of 543-558, DOI: 10.1016/j.finel.2010.12.004.
upstream base width (Bu) of the optimized dam sections. Kupfer, H. B. and Gerstle, K. H. (1973). “Behavior of concrete under
biaxial stresses.” Journal of the Engineering Mechanics Division,
6. Conclusions Vol. 99, No. 4, pp. 853-866.
Leclerc, M., Léger, P., and Tinawi, R. (2003). “Computer aided stability
analysis of gravity dams−CADAM.” Adv. Eng. Software, Vol. 34,
The paper presents the Optidam program which can design a
No. 7, pp. 403-420, DOI: 10.1016/S0965-9978(03)00040-1.
dam section for given constant height. Optidam program optimized Lee, J. and Fenves, G. L. (1998). “Plastic-damage model for cyclic
the section based on pseudo static analysis prescribed in Indian loading of concrete structures.” J. Eng. Mech., Vol. 124, No. 8, pp.
Standard. Factor of safety of one optimized section generated by 892-900, DOI: 10.1061/(ASCE)0733-9399(1998)124:8(892).
Optidam is evaluated by linear finite element analysis to show Lubliner, J., Oliver, J., Oller, S., and Oñate, E. (1989). “A plastic-damage
that stability is maintained. Non-linear finite element analysis is model for concrete.” Int. J. Solids Structures, Vol. 25, No. 3, pp. 299-
also performed to estimate the tensile damage. The heel of the 326, DOI: 10.1016/0020-7683(89)90050-4.
dam shows tensile damage upto 15 m and 7 m at the upstream Westergaard, H. M. (1933). “Water pressures on dams during earthquakes.”
Trans., ASCE, Vol. 98, pp. 418-432.
face for MCE and DBE level of shaking. But damthe stability of
Wilson, E. L. and Khalvati, M. (1983). “Finite elements for the dynamic
the dam section affected neither in DBE nor in MCE, which analysis of fluid-solid systems.” Int. J. Numer. Methods Eng., Vol.
proves dam sections designed by Optidam are safe. The general 19, No. 11, pp. 1657-1668, DOI: 10.1002/nme.1620191105.
characteristics of optimized dam sections is that, the down- Zangar, C. N. (1952). Hydrodynamic pressures on dams due to horizontal
stream slope of designed dams does not exceed 0.8; similarly, the earthquake effects, Technical Information Office.
upstream slope is also limited to 0.35. Zienkiewicz, O., Paul, D., and Hinton, E. (1983). “Cavitation in fluid-
Parametric study is performed on the optimized section generated structure response (with particular reference to dams under earthquake
by Optidam program for various heights, cohesion, and internal loading).” Earthquake Eng. Struct. Dyn., Vol. 11, No. 4, pp. 463-
481, DOI: 10.1002/eqe.4290110403.
friction angle. A regression analysis is performed on the 1080
Seismosoft (2010). Seismo signal, Seismosoft, Ltd.
numbers of designed dam sections. The result of regression Sima, J. F., Roca, P., and Molins, C. (2008). “Cyclic constitutive model
analysis leads to a guideline for dam designing. The parametric for concrete.” Eng. Struct., Vol. 30, No. 3, pp. 695-706, DOI: 10.1016/
analysis agrees with the fact that base width of the dam is j.engstruct.2007.05.005.
proportional to the height and inversely proportional with the SIMULIA (2009). ABAQUS theory manual, SIMULIA Worldwide.
internal friction angle and cohesion.

−8− KSCE Journal of Civil Engineering

View publication stats

You might also like