Ira. J. of Applied Mechanics and Engineering, 2003, vol.8, No.3, pp-461-482
A CRITICAL REVIEW OF DAM ANALYSIS METHODOLOGIES
P. SARKAR, S.CH, DUTTA’ and N. NANDI
Department of Applied Mechanics, Bengal Engineering College (Deemed University)
Howrah 711 103, West Bengal, INDIA
e-mail: thesarkar@netscape.net
e-mail: sedind@netscape.net
Dams in many countries including India were primarily designed on the basis of approximate hand
calculation based method. A frequent occurence of floods due to a failure of such dams and embankments
clearly points tothe need of a review ofthe procedure of dam analysis methadologis an to suggest 9 reasonably
accurate one. Te paper is limite effort forthe same, based on thee Indian dams, AS a very preliminary sep of
the whole plan of checking the safety of Indian dams, an analysis of three dams having similar features as that of
CChirpata Dam (Rajasthan), an earthen embankment dam, Koyna Dam (Maharastra), 2 concrete gravity dam and
the earthen embankment surrounding Ash Pond of Bakreswar Theemal Power Project (West Bengal, have been
carried out with the help of two-dimensional and three-dimensional finite clement discretisation under static
loading conditions. The result is then compared with the corresponding results obtained from a conventional hand
caleulation based on approximate gravity method of analysis. The comparison of the results exhibited a
‘eansierable increase in compressive tresses in two-dimensional finite element analysis with respect to what is
obtained in the conventional hand calculation based approximate gravity meted. Both of these methods
considered an in-plane stress distribution along a cross-section of the dam and hence, could not eeeognize any
possibilty of stress generation perpendicular to the cress-setion. In this context. a three-dimensional study
reveals that considerable sresses may generate even in the longitidinal direction of a dam that is generally
ignored in hand calculation based gravity method of analysis and twordimensional analysis due t the assumption
(of the two-dimensional nature of the problem. A rigorous finite element analysis also shows that even tensile
‘rest may generate under static loading conditions which ean hardly be predicted trom the conventional method.
It is also observed that hand calculation based on slip-irele method and two-dimensional analysis cannot
adequately recognize the possibility of shear fuilure at many poias in the body of the dam though their
‘vulnerability dve to itself and so itis necessary to point out by a rigorous three-dimensional modeling and
analysis, Thus, the study suggests the unavoidable necessity of three-dimensional analysis fora safe design of
dans.
Key words: dam, earthea, hand calculation based approximate method, threedimensional analysis
1. Introduction
Dams are designed to store water, which is the most important natural resource of the earth. With a
growing population and a subsequent increase in agricultural and industrial demand, there is a serious need
for many developing countries like India. However, their failure ean cause innumerable damages in the
surrounding region. So, structural safety assessment is a very important part of the dam design. Moreover, in
the context of regular flow in many locations in India, primarily duc to the failure of many earthen dams; and
also the recent earthen dam failures due to an earthquake in Gujarat, it becomes essential to check the safety
of dams of the country, particularly under crucial loading conditions. As a primary step. a critical review on
the adequacy of the frequently used dam analysis methodologies must be made. A large number of dams
failed in India. Among them seven medium-size earthen dams and fourteen small earthen dams were
damaged daring the Bhuj Earthquake in Gujarat on January 26%, 2001. The well documented damage studies
are available for Tappar Dam, Fategarh Dam, and Suvi Dam (Jain and Lets, 2001).
Earthen embankment dams are the most common type of dams used in the developing countries like
India, principally because their construction involves utilization of materials in the natural state requiring
minimum processing. Moreover, the foundation requirements for earthen embankment dams are less
stringent than that for other types of dams. Due to a very high water level, the leakage and piping are found
To whom correspondence should be addressed462 P Sarkar, §.Ch.Dutta and N.Nandi
to occur in earthen dams. Past records show that this may lead to the failure of the dam as a whole, The
failure of Kamari Dam at Kamari Irrigation project constructed near the village of Padiv at Sirohi, Rajasthan
and Chirpatia Dam constructed near the village of Chitpatia in Rajasthan (Dhariwal and Purohit, 2000) are
two such instances, A favourable situation for piping and leakage may be generated due to the occurrence of
tension in the body of the dam. Hence, itis needed to make a rigorous analysis to check whether tension may
occur in the body of such dams if they are designed by a hand calculation based approximate method, which
is frequently used in practi
For these large engineering structures, design methods are reliable only when the actual structure
stands and operates as per expectations. Since the frequent failure of dams puts a question mark on the
adequacy of the approximate method used, a detailed analysis of failed dams as well as the existing dams
with a reasonably accurate modeling of minute details is urgently needed for building safely the new dams as
well as retrofitting the existing oncs. Realistic analys:s with three-dimensional modeling of dams are rare.
For instance, even a recent study of Koyna Dam, a well known concrete gravity dam of India, was restricted
to plane stress considerations through the two-dimensional finite element method (Bhattacharjee and Leger,
1994). The results of three-dimensional modeling may particularly be different from those of two-
dimensional analysis in case of earthen dams, since Poisson's ratio of soil is high
In this context, the authors have taken up a long-term plan to make a rigorous analysis of Indian
ams under various crucial loading conditions. As a preliminary step of such studies, three dams, one of
which is @ concrete gravity dam similar to Koyna Dam and the other two are earthen dams similar to
Chirpatia Dam of Rajasthan, India and the embankment surrounding Ash Pond of Bakreswar Thermal Power
Project, West Bengal, India are analyzed with two as well as three-dimensional idealization under a few
static-loading conditions. A comparison between the results obtained from the hand calculation based
approximate gravity method of analysis, two-dimensional idealization, and three-dimensional idealization,
respectively, leads to important observations regarding methodologies of dams analysis.
2. Analysis
The details of the finite clement formulation used for two-dimensional and three-dimensional
analyses in the study are presented briefly here
2.1. Finite element formulation
A six-noded triangular clement and a twenty-noded solid element are used for the two-dimensional
and three-dimensional analyses, respectively. These well-accepted elements are used in many software (e.g.,
ANSYS 5.7) and the formulations for these elements are also available in the literature (e.g., Cook, 1995).
Both of these elements exhibit very quick convergence. The triangular element in two-dimensional modeling
is particularly used to model the geometry with sufficient accuracy near the comer of dams. To ensure a
better accuracy further, convergence study is carried out in each case comprising of each separate
combination of structural geometry and loading. The nodes at the base of the dam are restrained along all
degrees of freedom. For three-dimensional analysis, the two nodes at the ends of the dams are considered to
be five. Self-weight and fluid loading are adequately applied depending on the loading conditions
considered. The approach is briefly presented below.
2.1.1. 2-D 6-noded triangular elements
A six-noded triangular plane clement is used for the two-dimensional analysis for a typical cross-
section of the dams, This element has three nodes at three vertices and three more nodes, in the middle of
each side. The element is having two degrees of freedom at each node: translations in the nodal X direction
denoted by u and nodal ¥ direction denoted by v functions are derived. The detail of the element is shown in
Fig. L. The shape functions considered for the element are given below
11 j 20, ~MDby +10g (2g ~ kg +Us Ply —1)+ ug Ly Ly) us (4L aby + ugl4L gly), 22D)A critical review of dam analysis methodologies
463
vey, (2L) ~1)by + vp (2Ly — I) + v3 (2Ly - I+ vg (4LyLy)+ vs(4LyLy)+ vg (4Lgl;) (2.2)
where (1.¥)) (u2.¥2), (e320) (4.04) (us.¥s) and (ug.v6) are co-ordinates of the nodes J to 6 of the
element, respectively. L;,L2,L3 are the normalized co-ordinates along the sides of the element, ranging
ky
2
ON
ae i 6 3
Fig.1. Details of 2-D 6-noded triangular element
. 3-D 20-noded st
Fig.2. Details of 3-D 20-noded solid element,
elements,
This clement has twelve edges arid twenty nodes as shown in Fig.2. Each node has three translational
degrees of freedom, 1, v and w, along three mutually perpendicular axes X, ¥ and Z, respectively. The shape
functions considered for the element are presented below
w= du; (0s)(t-2)(t—n)-s—1—r=2)+ uy (14s =1)(0=r)lo—1—r-2)+
tug (I4s)(1+1)(0-r)(s+4—r—2)+ up (Is) 0 r)este—r—2)+
tty (IS) (0+ n)(es—1+r—2)+uy (74s M014 Flot r—2)+
tug (I+ s\(l+) P+ nlete+r—2)+up(l—s)(4n(te reste r—2)}+
buts? t=) 0—r) p(t +3)? linus (2 red r)+ 23)
$up(I=s)()-12 t=) uy (Is? )0-2)(t+ remy (145)(0-2 rer)
sup (I—s? t2)(04 ren, (I s)(t—t2 is rey (P-s)(-a){t-r?
sug I+s)0-o(t-- eugl+s\rta)(t-r? eu —sen(t-r?)h464 P.Sarkar, $.Ch.Dusta and N.Nandi
vadfe;(/-s) (analogous to x), @4)
w= Lfw)(I=s) (analogous to 1) 25)
where r,s, ¢are the normalized co-ordinates as shown in Fig.2, and vary within a range of - 1 to 7.1, vs w
with an adequate suffix denote the nodal displacements along X, ¥, Z axes, respectively for the node denoted
by the comesponding suffix.
Structural details
‘The concrete dam chosen in the study is considered to have a cross section very similar to the non-
overflow section of Koyna Dam located over Koyna River at Satara district in Maharastra. The height and
the width of the dam at the base are 103m and 70.4m, respectively. The width of the dam at a height of 66.5m
is 19.3m. The width of the dam atthe top is 14.8. It is considered that the height of the water level is up to
91.75m at the upstream side, whereas no water is considered in the downstream side. The upstream face is
considered as vertical. The specific weight of the material of the dam is taken as 24KN/m* with modulus of
elasticity 25.5 10* kN/m? . These dimensional data of Koyna Dam were obtained from those given for this
Dam in the literature (Chopra and Chakraborty, 1972). The total length of the dam is considered as 808m as
‘mentioned in the literature (Indian National Committee on Large Dams, 1979). As a second case study, an
earthen dam with similar features as those of Chirpatia Dam situated at Chirpatia village in Karachi Tehsil,
Rajasthan has been chosen. The height and the width of the earthen dam at the base are considered to be
102.2m and 460.9m, respectively. The width of the dam at the top is taken as 2.5m. The upstream and
downstream slope of the dam are considered as 2.5H:1V and 2H :1V , respectively. It is considered that
the maximum height of the water level is the same as the total height of the dam, The specific weight of the
material is taken as 20,5KN/m? with the modulus of elasticity 50 000kN/m? . The geometric data are
selected on the basis ofthe literature (Dhariwal and Purohit, 2000) outlining the failure and other details of
Chirpatia Dam. On the other hand, a set of reasonable values of the material constants are obtained from a
well-accepted literature (Bowles, 1996). The shear strength parameters used for the analysis are obtained
from laboratory tests of undisturbed soil samples collected from different zones of the body of an
embankment dam. The average value of the apparent cohesion (C) is obtained as 58.3KN/m? and the
average angle of shearing resistance (9) is obtained as 48.92" from the large number of tests from different
zones of an embankment dam. An analysis is also cartied out for a typically less cohesive soil having a
smaller cohesion (C) as 20.04N/m? and a smaller angle of shearing resistance (9) as 48°. Asa third case
study, an earthen embankment with a similar geometry as that surrounding the Ash Pond of Bakreswar
‘Thermal Power Project situated at Panuria village in Birbhum District, West Bengal, India has been chosen.
It is a zoned type earthen embankment recently constructed to make a reservoir for the disposal of fly ash
mixed slurry. The height and width of the embankment at the base are considered to be J8m and 9fm
respectively. The width of the embankment at the top is taken as 4m. Both the upstream and downstream
slopes of the embankment are considered 2s 2.25H : IV . There are two horizontal berms of width 3m at a
height of 6m and 12m from ground level, respectively. It is considered that the maximum height of the fluid
level is same as the total height of the embankment. For the pervious casing zone the specific weight of the
material is taken as 20.5kN/m with the modulus of elasticity 50 000kN/m? whereas for the impervious
core the specific weight of the material is taken as 19.8KN/m? with the modulus of elasticity
70.000kN/'m? , Poisson's ratios for pervious and impervious material are taken as 0.35 and 0.48,
respectively. Sets of reasonable values of these material constants are selected on the basis of a well-accepted
literature (Bowles, 1996), The shear steength parameters used for the analysis are obtained from laboratoryA critical review of dam analysis methodologies 465
tests of undisturbed soil samples collected from different zones of the body of the embankment at Bakreswar
itself. The average value of the apparent cohesion (C) is obtained as 20.0kN/im? and the average angle of
shearing resistance (9) is obtained as 48° from a large number of tests conducted on samples from different
pervious zones of the embankment. In the case of impervious material, the average value of the apparent
cohesion (C) is obtained as /6.864N/m? and the average angle of shearing resistance ($) is obtained as
27.5*. The specific weight of the water in the reservoir is taken as /0kN/m’ for all the cases. For these
dams, two-dimensional as well as three dimensional analysis are carried out to compare the results with those
obtained from the approximate hand calculation based analyses as outlined in the literature (Garg, 1996). As
mentioned earlier, twenty node solid elements are used for the three-dimensional analysis of the dams. For
the two-dimensional analysis, six noded triangular isoparametric elements are used. The study yielded
‘maximum stress at various points in the body of the dams. Special attention is focused on understanding the
possibility of occurrence of tensile stress and in the body of the dams.
4, Results and discussions
Tables 1, 2 and 3 present the maximum and minimum principle stresses (Sway and Siyiq) and
maximum shear stresses (ty,,) obtained from the two-dimensional analysis of Koyna dam, Chirpatia dam
and Ash Pond of Bakreswar Thermal Power Project, respectively. Tables 4, 5 and 6 present a comparison
between the results of the hand calculation based analysis and two-dimensional finite element analysis for
concrete gravity and earthen dams, respectively. Tables 7, 8 and 9 present the stresses at various important
points of the concrete gravity dam and the two earthen dams, respectively as obtained from the three-
dimensional finite element analysis. The normal stresses S,.,Sy, and S,. along X, ¥, and Zaxes, ie., along
the cross sectional horizontal direction, vertical direction and longitudinal horizontal direction, respectively,
and shear stresses S,,.5,- and $., along X-Y, ¥-Z, and Z-X planes, ice. along the cross sectional vertical,
longitudinal vertical and horizontal plane, respective'y, obtained from three-dimensional analysis for Koyna
dam and Chirpatia dam are listed in Tabs.7a and 8a. On the other hand, Tabs.7b, 8b and 9 exhibit the
principal and maximum shear stresses at the same points within the body of the dams
Table 1. Two-dimensional finite element analysis results for the concrete gravity dam.
Postion Six (ew/m?Y | San vf?) | toe Gv /m?
Dam bottom | US face $3953.10 =359 $2136.33,
level | Mid-seetion =51.10 =1814.70 +8818
[is face $34.10 = 440.76 237.43,
WiS face [588.43 =1018.40 $214.98
Quarter level [Mid-section | =229.32 =1164.60. $436.14
DiS face | 0.07 =1407.80. 703.86;
=249.28 =706.6 £228.69
Half level =224.84 749.65 $262.40
=142.07 =2090.00 $973.96
= 69.89 483.47 $206.79
‘Three-quarter =39.79 = 0815 418418
$0.06 =3AD28 $I7LIT
Cres iever WStface 40.02 $0.01 0.00
vest level [DIS face = 0.03 = 0.08. $0.02
“Normal stresses are considered positive if they are tensile in nature. The pair of shear stress vectors acting
‘on an element is considered positive if they form a clockwise couple.P.Sarkar, §.ChDutta and Nandi
le 2. Two-dimensional finite element analysis results for the earthen dam modeled following Chirpatia dam.
Peston Ssoc vf | San Gln?) | oan Gn?) |
Dam bottom [u/s face 546.30 359.56 4215633 |
level ‘Mid-seetion =710.69 =1814.70 +8818 |
[Dis face +110 = 0.76 4237.43,
UIS face =1018.40 4214.98
Quarter level [/Mid-section =1164.60 1456.14
DiS face 1407.80 +703.56
UIS face ~706.66 $226.69
Half level [ Mid-section 749.65 $262.40
DiS face 0.00 2090.00 $973.96
Three-quarter U/S ee = 498.43 = 483.47 $206.79
Pesce [Mid-seetion 304.84 408.15, $184.18
DiS face = 0.05 342.28 #I7LIT
Crtleve) [WS fice =104 +001 0.00 |
DIS face $3.09 = 0.08 +0.02 J
“Normal stresses are considered positive if they are tensile in nature, The pair of shear stress vectors acting
‘on an element is considered positive if they form a clockwise couple.
Table 3. Two-dimensional finite element analysis results in full reservoir condition for the embankment
used for making Ash Pond of Bekreswar Thermal Power Project.
Position, Srnax leov/m?} Sorin Gov/n? Tax (vm? } |
Curoftench | U/S side =126.03 BB 472.65 |
level__[ DIS side =132.58 =277.11 $72.26
UIS tuce ~72.68. $185.24
Ground fever -EOT Heal =155.9 $96.83
Core toe =104.86 $127.01
DIS face 40.03 $0.05
UIS face =TI9.83 $57.82
Lower berm [ Mid-section =108.18 $76.13
level [Berm inner node 14. #1722
Berm outer node 0.15 $3.56
[UIS face $102.98 $22.38
Upper berm | Mid-section = 48.91 41.65
level [Berm inner node =3.59 +28.80
Berm outer node +015 | $5.46
UIS face $3.95 +199)
Crestlevel [Mid-section $14.83 [474
DIS face: +0.16 __+0.27
“Normal stresses are considered positive if they ate tensile in nature. The pair of shear stress vectors acting
oon an element is considered positive if they form a clockwise coupleA critical review of dam analysis methodologies
467
Table 4. A comparison between the results obtained using two-dimensional analysis and conventional hand
calculation based gravity method of analysis for the concrete gravity dam.
First and second principle stress
obtained in 2-D FE analysis
Stress obtained in hand
Position calculations based analysis,
(ew/m? (ev/m2
| $3953.10
UPS side 1048.79
~359.56
Dam - -51.01
bottom | Mid-scetion 1228.93,
level ~181470 |
| $34.10
DiS side 1409.06 |
| - 440.76
~588.43
UIS side 944.78
=1018.70
~229.32
Quarter | M4id-section -903.72
level 1164.60
| ~0.07
DIS side 862.58
| -1407.80
| ~249.28
USside | - 600.44
706.66
~224.84
Half evel | Mid-section -773.88
~749.65
142.07
DIS side 947.31
-2090.00
69.89
UIS side ~476.37
~ 483.47
Three- 39.79
quarter | Mid-section = 409.10
‘evel = 408.15
+005
DIS side 341.82
-342.28
+002
Sside | 0.00
Gouee +001
level 0.08
DIS side 0.00
-0.08
“Normal stresses are considered positive if they are te
in nature.468.
PSarkar, 8.ChDuta and N.Nandi
Table 5. A comparison between the results obtained using two-dimensional analysis and conventional hand
calculation based gravity method of analysis for the earthen dam modeled following Chirpatia dam.
First and second principle stress Stress obtained in hand
iodition obtained in2-DFEunalysis | culeulatons based analysis
(avn? (ifm?
~546.30
UIS side -2917.19
| -2737.80
| Dam ~710.69
| otiom | Mid-seetion 1753.12
level =2190.10
+110
DiS side 589.05
237
= 693.54
USS side -2187.63
=2063.40
ner 538.34
Quarter | s4id-scction 1315.30
ive | 1664.89
- 0.04
DIS side =442.97
-202.11
| ~ 8747
UIS side 1458.56
1349.10
~ 406.51
Half level | Mid-seetion -874.81
=1095.66
. 0.00
DIS side 300.41
-248.01
~ 498.43
UIS side 729.00
-712.26
Three 304.84
quarter ~ 438.25,
level 2560.89
0.05
DIS side -147.20
-274.68
1.04
USS side 0.00
Crest Ald
level 43.09
DIS side 0.00
~39.26
“Normal stresses are considered positive if they are tensile in nature,A critical review of dam analysis methodologies 469
Table 6, A comparison between the results obtained using two-dimensional analysis and conventional hand
calculation based gravity method of analysis considering actual material properties of the earthen
embankment surrounding Ash Pond,
Principle stress obtained in tw0- ‘Suess obtained ia iaad
Position dimensional finite element analysis’ | calculations based analysis
Seas (ev/m?) | Sy Gdn (one
UIS face 72.68 ~ 443.16 ~363.46
Ground | Core heal =155.90 349.57 =226.49
evel | Core toe 104.86 358.89 172.48
DiS face +0.03 =007 =12.36
UIS face ~119.83 =295.47 307.74
Lower [Midsection | __-108.18 =26045 =130.82
berm
level | Berm inner node -2.14 ~36.58 =1.01
Berm outer node 40.15 -6.98 0.00
UIS face 102.98 =147.74 ~115.28
Upper | Mid-section = 48.91 =13222 =67.74
berm
level | Berm inner node 3.59 = 61.20 20.21
Berm outer node +0.15 10.77 0.00
UIS face 43.95 0.03 [ 0.00
Gest | Mid-seetion $14.83 ~ 0.03 0.00
DIS face $0.06 = 0.38 0.00
“Normal stresses are considered positive if they are tensile in nature.
Comparisons of the results of Tabs.1, 4 and 7; results of Tabs.2, 5 and 8 and the results of Tabs.3, 6
and 9 clearly indicate that the results obtained from the approximate method, those from the two-dimensional
analysis and those from the three-dimensional analysis may differ a lot in both magnitude and nature for both
the concrete gravity dam as well as earthen dams. Figures 3, 4 and 5 present a pictorial representation of,
stress distribution in the section of Koyna dam, Chispatia dam and the earthen embankment of Ash Pond,
respectively, as obtained using the above mentioned three different types of analyses. These figures clearly
indicate the difference among the stress distribution scenarios obtained by these three methods of analyses,
‘Maximum principal compressive and tensile stresses are only plotted in these figures.
Figure 3a shows that there is no possibility of tension in the body of the concrete gravity dam while
Fig.3b clearly indicates that tension may be recognized in the two-dimensional analysis. Figures 3c and 3d
clearly indicate that there may be a fairly large tensile zone at the top and the bottom of the dam. Figures 4
and 5 also show that a similar observation also holds good for the two earthen dams considered. An
approximate analysis as well as the two-