Numerical Study of Near-Field Earthquake Effects on Concrete Gravity Dams

NUMERICAL STUDY OF NEAR-FIELD EARTHQUAKE EFFECTS ON
CONCRETE GRAVITY DAM

MS by Research Thesis
submitted by
K. Jagan Mohan
(2008 14002)
in partial fulfilment of the requirements for the degree
of
Master of Science (by Research)
in
Civil Engineering (Structural Engineering)
Earthquake Engineering Research Centre,
International Institute of Information Technology, Hyderabad 500 032
December 2012
CERTIFICATE
It is certified that the work contained in this thesis, titled Numerical study of near-
field earthquake effects on concrete gravity dams by Mr. K. Jagan Mohan, has been
carried out under my supervision and is not submitted elsewhere for a degree, to my
knowledge.
__________________________________________________
Advisor: Dr. Ramancharla Pradeep Kumar
Professor
Earthquake Engineering Research Centre
International Institute of Information Technology
December 2012 Hyderabad 500 032
INDIA
Dedicated to my mother Rama Devi, father Chinna Subbaiah and dearest friend
Rupak Ashle, whom I miss in this world.
Acknowledgements
My sincere gratitude goes primarily to my thesis supervisor, Professor Dr. Ramancharla
Pradeep Kumar, firstly because of his vast knowledge shared with us in several fields, be it
technical or non-technical cannot be overstated, and secondly for his personal care,
guidance and supervision over me, for my better career.
I would also like to thank Prof. M Venkateshwarlu for teaching us Mathematical
Foundations of Solid Mechanics and Theory of Elasticity and Assistant Prof. Dr. D. Neelima
Satyamfor their technical and non-technical support.
I would like to give my sincere thanks to all my colleagues (CASE), EERC members (Admin
Staff, Project Staff, Research Staff and Office boy), M.Tech students and friends who
contributed directly or indirectly in various ways for my research and well being.
I would like to give my true respect to my guardians, brothers, sisters and family members
for their support and special care and love towards me.
I owe to IIIT-H and every single person who attended and took care of me when I was
hospitalised. I would like to thank IIIT-H for releasing medical funds for my recovery.
Finally, I would like to dedicate this thesis to my dearest friend Rupak Ashle, who is no
longer alive. He was been a biggest support to me, which no one else gave in my entire life.
He is the person who always wanted to see me get MS degree deep from his heart and he
always wants to take pride in speaking about my degree. Here this is the thesis for him.
Miss him deep from my heart. Wish his soul would rest in peace.
K. Jagan Mohan
Earthquake Engineering Research Centre
International Institute of Information Technology
Hyderabad, INDIA
Abstract
Dam is one of the biggest structures built on the Earth. It is known as a life line
structure, as it serves the purpose of irrigation, hydro-electric power generation, flood control,
domestic and industrial water supply etc., which are important for human existence. This makes
dam as a reliable structure. For this reason, dam should always be designed for highest safety,
resisting worst forces of nature. India is a country with over 5,100 large dams. India is also a
seismically active country with over 1,000 active faults. 1988 Bihar earthquake, 1991 Uttarkashi
earthquake, 1993 Killari earthquake, 1997 Jabalpur earthquake, 1999 Chamoli earthquake, 2001
Bhuj earthquake, 2002 Andaman earthquake, 2004 Sumatra earthquake, 2005 Kashmir earthquake,
2011 Sikkim earthquake are some of the earthquakes that has hit India in the recent past. Also
events like 1992 Landers, 1994 Northridge, 1995 Hyogoken-Nambu and few other events that took
place around the world proved how devastating an earthquake could be, particularly if it is near-
field. Near-field ground motions could cause more damaging effects on structures, as they were
observed to differ dramatically from the characteristics of their far-field counterparts. The
propagation of fault rupture towards a site at very high velocity causes most of the seismic energy
from the rupture to arrive in a single or multiple large long period pulse of motion, which occurs at
the beginning of the record. This characteristic of near-field ground motions could cause damage to
a wide range of structures including dams. Several dams that were built in India, which are in highly
seismic zones are prone to near-field ground motions. In this regard, behavior of a concrete gravity
dam subjected to near-field ground motion should be studied.
In the proposed study, a concrete gravity dam is selected from the National Importance
Dams of India and is numerically modelled along with its foundation and soil strata, using Applied
Element Method (AEM). The dam models are analysed for the components of near-field and far-
field ground motions. The comparisons between these two were drawn to understand the effects of
near-field ground motions on dams. In another study, vertical displacement is given at the bed rock
level to create reverse fault effect, while changing the location of dam alongside fault. Variation of
displacements and stresses at the base, crest, toe, heel, neck, and on upstream and downstream
sides are evaluated. For understanding the initiation and propagation of cracks, eect at various
parts of the dam body is also studied. Failure of dam is observed, when subjected to near-field
ground motions caused by reverse fault and strike-slip fault. Also, when subjected to fault motion,
failure in the dam body is observed to be more, when the dam is modelled on hanging wall than
when it is modelled on foot wall. This study proved life-line structure like dam has failed to near-
field ground motion. Thus, dams should be designed to resist severe earthquakes.
Contents
1. Introduction - - - - - - - - - 1
1.1 Prevalent seismic hazard in India - - - - - - 2
1.2 Performance of concrete gravity dams subjected to earthquakes - - 5
1.3 Literature review - - - - - - - - 8
1.3.1 Analytical studies - - - - - - - 8
1.3.2 Experimental studies - - - - - - 9
1.3.3 Numerical studies - - - - - - - 9
1.4 Scope of present study - - - - - - - 13
1.5 Organization of the thesis - - - - - - - 13
2. Numerical Modelling of concrete gravity dam
2.1 Introduction - - - - - - - - 15
2.2 Selection of concrete gravity dam - - - - - - 15
2.2.1 Dam Geometry - - - - - - - 16
2.2.2 Material properties - - - - - - - 19
2.3 Ground motions & their characteristics - - - - - 19
2.3.1 Selection of ground motions - - - - - - 20
2.3.2 Characteristics of ground motions - - - - - 20
2.4 Near-field ground motions - - - - - - - 25
2.5 Far-field ground motions - - - - - - - 35
2.6 Numerical method
2.6.1 Introduction - - - - - - - 45
2.6.2 Mathematical formulation - - - - - - 46
2.6.3 Element size - - - - - - - - 47
2.6.4 Material modelling - - - - - - - 47
2.6.5 Boundary conditions- - - - - - - 48
2.6.6 Effect of number of connecting springs - - - - 48
2.6.7 Modelling limitations - - - - - - 49
2.7 Summary - - - - - - - - - 49
3 Linear Earthquake response of concrete gravity dam
3.1 Introduction - - - - - - - - 50
3.2 Eigen value analysis - - - - - - - 50
3.3 Near-field earthquakes
3.3.1 Response of S1 subjected to fault normal component - - 55
3.3.2 Response of S1 subjected to fault parallel component - - 56
3.4 Far-field earthquakes
3.5 Comparison of near-field and far-field response of S1
3.5.1 Fault normal and transverse response of S1 - - - - 76
3.5.2 Fault parallel and longitudinal response of S1 - - - 76
3.6.1 Fault normal and transverse response of S2 - - - - 77
3.6.2 Fault parallel and longitudinal response of S2 - - - 78
3.7 Comparison of Response Spectrum of near-field and far-field ground motions 79
3.8 Summary - - - - - - - - - 82
4 Non-linear Earthquake response of concrete gravity dam
4.1 Introduction - - - - - - - - 83
4.2 Comparison of near-field and far-field ground motions
4.2.1 S1 and S2 subjected to Tabas ground motion - - - 83
4.2.2 S1 and S2 subjected to Loma Prieta ground motion - - - 85
4.2.3 S1 and S2 subjected to Landers ground motion - - - 85
4.2.4 S1 and S2 subjected to Northridge ground motion - - - 86
4.2.5 S1 and S2 subjected to Kobe ground motion - - - - 89
4.3 Understanding the behaviour of models S1 and S2 with amplified accelerations
4.3.1 S1 and S2 subjected to Tabas ground motion - - - 101
4.3.2 S1 and S2 subjected to Loma Prieta ground motion - - - 103
4.3.3 S1 and S2 subjected to Landers ground motion - - - 105
4.4 Summary - - - - - - - - - 107
5 Non-linear response of concrete gravity dam subjected to fault motion
5.1 Introduction - - - - - - - - 116
5.2 Numerical model - - - - - - - - 117
5.3 Dynamic soil-structure interaction- - - - - - 121
5.3.1 Direct method - - - - - - - 121
5.3.2 Sub-structure method - - - - - - 122
5.4 Non-linear response of concrete gravity dam subjected to fault motion - 122
5.4.1 Dam modelled on foot wall - - - - - - 123
5.4.2 Dam modelled on hanging wall - - - - - 123
5.5 Summary - - - - - - - - - 124
6 Conclusions and Scope for future Study
6.1 General conclusions - - - - - - - 130
6.2 Future work - - - - - - - - 131
Appendix
A.1 History and necessity of dams - - - - - - 132
A.2 Causes of dam failures - - - - - - - 134
A.3 Special focus on earthquake effects - - - - - 135
A.4 Large dams and importance of seismic study on large dams in India - 140
References - - - - - - - - - - 149
List of Figures
1.1 Seismic zonation map of India - - - - - - - 2
2.1 National importance dams of India & seismic zonation map & top 100 active faults 16
2.2 Cross-section and material properties of Structure 2 - - - - 17
2.3 Cross-section and material properties of Structure 1 - - - - 18
2.4 Tabas near-field ground motion - - - - - - - 25
2.5 Loma Prieta - Los Gatos near-field ground motion - - - - - 26
2.6 Loma Prieta - Lexington Dam near-field ground motion - - - - 27
2.7 Erzincan near-field ground motion - - - - - - 28
2.8 Cape Mendocino near-field ground motion - - - - - 29
2.9 Lander's near-field ground motion - - - - - - 30
2.10 Northridge - Rinaldi near-field ground motion - - - - - 31
2.11 Northridge - Olive View near-field ground motion - - - - 32
2.12 Kobe near-field ground motion - - - - - - - 33
2.13 Kobe - Takatori near-field ground motion - - - - - 34
2.14 Tabas-Sedeh far-field ground motion - - - - - - 35
2.15 Loma Prieta - Hollister City Hall far-field ground motion- - - - 36
2.16 : Loma Prieta - Sunnyvale far-field ground motion - - - - 37
2.17 Erzincan far-field ground motion- - - - - - - 38
2.18 Cape Mendocino far-field ground motion - - - - - 39
2.19 Landers far-field ground motion - - - - - - - 40
2.20 Northridge-Anacapa Island far-field ground motion - - - - 41
2.21 Northridge-Anaheim far-field ground motion - - - - - 42
2.22 Kobe-99999HIK far-field ground motion- - - - - - 43
2.23 Kobe-99999FUK far-field ground motion - - - - - 44
2.24 Element components for formulating stiffness matrix (SOURCE: Kimuro Meguro and Hatem
2001) - - - - - - - - - - 46
2.25 Quarter portion of stiffness matrix - - - - - - 46
2.26 Material models for concrete and steel (a) Tension and compression concrete Maekawa model
(b) Bi-linear stress strain relation model for steel reinforcement (Kimuro Meguro and Hatem 2001)
- - - - - - - - - - 48
3.1 Eigen values of first ten mode shapes of S1 - - - - - 51
3.2 Eigen values of first ten mode shapes of S2 - - - - - 52-53
3.3 Linear displacement response of S1 to 5 near-field fault normal ground motions - 59
3.4 Linear displacement response of S1 to 5 near-field fault parallel ground motions - 60
3.5 Linear displacement response of S2 to 5 near-field fault normal ground motions - 64
3.6 Linear displacement response of S2 to 5 near-field fault parallel ground motions - 65
3.7 Linear displacement response of S1 to 5 far-field transverse ground motions - 69
3.8 Linear displacement response of S1 to 5 far-field longitudinal ground motions - 70
3.9 Linear displacement response of S2 to 5 far-field transverse ground motions - 74
3.10 Linear displacement response of S2 to 5 far-field longitudinal ground motions - 75
3.11 Acceleration Response Spectrum of 5 near-field fault normal ground motions - 80
3.12 Acceleration Response Spectrum of 5 near-field fault parallel ground motions - 81
3.13 Acceleration Response Spectrum of 5 far-field transverse ground motions - 81
3.14 Acceleration Response Spectrum of 5 far-field longitudinal ground motions - 82
4.1 Crack propagation observed in S2 to Tabas near-field fault normal ground motion 85
4.2 Crack propagation observed in S2 to Landers near-field fault normal ground motion 86
4.3 Crack propagation observed in S2 to Landers near-field fault parallel ground motion 86
4.4 Crack propagation observed in s1 to Northridge near-field fault normal ground motion 87
4.5 Crack propagation observed in s1 to Northridge near-field fault parallel ground motion 88
4.6 Crack propagation observed in S2 to Northridge near-field fault normal ground motion 88
4.7 Crack propagation observed in S2 to Northridge near-field fault parallel ground motion 89
4.8 Crack propagation observed in s1 to Kobe near-field fault normal ground motion 91
4.9 Crack propagation observed in s1 to Kobe near-field fault parallel ground motion 91
4.10 Crack propagation observed in S2 to Kobe near-field fault normal ground motion 92
4.11 Crack propagation observed in S2 to Kobe near-field fault parallel ground motion 92
4.12 Non-linear displacement response of S1 to 5 near-field fault normal ground motions 93
4.13 Non-linear displacement response of S1 to 5 near-field fault parallel ground motions 94
4.14 Non-linear displacement response of S2 to 5 near-field fault normal ground motions 95
4.15 Non-linear displacement response of S2 to 5 near-field fault parallel ground motions 96
4.16 Non-linear displacement response of S1 to 5 far-field transverse ground motions 97
4.17 Non-linear displacement response of S1 to 5 far-field longitudinal ground motions 98
4.18 Non-linear displacement response of S2 to 5 far-field transverse ground motions 99
4.19 Non-linear displacement response of S2 to 5 far-field longitudinal ground motions 100
4.20 Crack propagation observed in S1 to Tabas near-field fault normal ground motion 101
4.21 Crack propagation observed in S1 to Tabas near-field fault parallel ground motion 102
4.22 Crack propagation observed in S2 to Tabas near-field fault parallel ground motion 103
4.23 Crack propagation observed in S1 to Loma Prieta near-field fault normal motion 104
4.24 Crack propagation observed in S2 to Loma Prieta near-field fault normal motion 104
4.25 Crack propagation observed in S2 to Loma Prieta near-field fault parallel motion 104
4.26 Crack propagation observed in S1 to Landers near-field fault normal ground motion 105
4.27 Crack propagation observed in S1 to Landers near-field fault parallel ground motion 106
4.28 Amplified non-linear displacement response of S1, subjected to 5 near-field fault normal
ground motions - - - - - - - - - 108
4.29 Amplified non-linear displacement response of S1, subjected to 5 near-field fault parallel
ground motions - - - - - - - - - 109
4.30 Amplified non-linear displacement response of S2, subjected to 5 near-field fault normal
ground motions - - - - - - - - - 110
4.31 Amplified non-linear displacement response of S2, subjected to 5 near-field fault parallel
ground motions - - - - - - - - - 111
4.32 Amplified non-linear displacement response of S1, subjected to 5 far-field transverse ground
motions - - - - - - - - - 112
4.33 Amplified non-linear displacement response of S1, subjected to 5 far-field longitudinal ground
motions - - - - - - - - - 113
4.34 Amplified non-linear displacement response of S2, subjected to 5 far-field transverse ground
motions - - - - - - - - - 114
4.35 Amplified non-linear displacement response of S2, subjected to 5 far-field longitudinal ground
motions - - - - - - - - - 115
5.1 Displacement time history of 2.5m with rise time of 5sec as input- - - 118
5.2 Failure of springs observed in the base when 2.5m displacement is given in 5sec - 118
5.3 (i) model of S2a (ii) displacement of 2.5m as input to S2a - - - - 119
5.4 (i) model of S2b (ii) displacement of 2.5m as input to S2b - - - - 119
5.5 (i) model of S2c (ii) displacement of 2.5m as input to S2c - - - - 120
5.6 (i) model of S2d (ii) displacement of 2.5m as input to S2d - - - - 120
5.7 (i) model of S3 (ii) displacement of 2.5m as input to S3 - - - - 120
5.8 Accelerations recorded at 4 different locations on the surface, when base is subjected to 2.5m
upward displacement with a rise time of 5sec - - - - - 121
5.9 Failure of dam (at 100m-200m) subjected to 2.5m displacement in the base - 125
5.13 Failure of dam subjected to free-field acceleration recorded at 250m - - 126
5.16 Displacement response at the crest of dam (dam placed at 4 different locations), subjected to
2.5m displacement with a rise time of 5 sec - - - - - - 128
5.17 Displacement response at the crest of dam; acceleration recorded at 4 different locations
when base is subjected to 2.5m displacement with a rise time of 5 sec - - 129
A1.1 Kallanai Dam in India dated 2
nd
century AD - - - - - 132
A3.1 Fault Normal, Fault Parallel & Vertical Accelerogram of 1994 Northridge earthquake 136
A3.2 1992 Landers earthquake, showing the Forward and Backward Directivity region 137
A3.3 Surface faulting caused major damage to Shii-kang Dam- - - - 138
A4.1 State wise distribution of large dams (existing and ongoing) in India - - 140
A4.2 Distribution of large dams in India - decade wise - - - - 141
A4.3 State wise distribution of large dams (completed) in India - - - 141
A4.4 State wise distribution of large dams (under construction) in India - - 142
A4.5 National Importance Dams of India on Seismic Zonation Map - - - 143
A4.6 National Importance Dams of India placed on Fault & Seismic Zonation map of India 143
A4.7 National Importance Dams of India on Seismic Zonation Map plotted with top 100 active fault
- - - - - - - - - - - 144
List of Tables
1.1 Important earthquakes in Himalayan Frontal Arc (Kamalesh Kumar (2008),
http://gbpihed.nic.in) - - - - - - - - 3
1.2 Important earthquakes in Peninsular India (Kamalesh Kumar (2008),
1.3 Important earthquakes in Northeastern region of India (Kamalesh Kumar (2008),
1.4 Concrete dams subjected to significant shaking (PHGA > 0.3g) (Courtesy: USSD Proceedings
2012) - - - - - - - - - - 5
2.1 Details of near-field - fault normal ground motions - - - - 23
2.2 Details of near-field - fault parallel ground motions - - - - 23
2.3 Details of far-field - transverse ground motions - - - - - 24
2.4 Details of far-field - longitudinal ground motions - - - - - 24
3.3 Response of S1 subjected to 5 near-field fault normal ground motions - - 55
3.4 Response of S1 subjected to 5 near-field fault parallel ground motions - - 56
3.5 Comparison of response of S1 subjected to 5 near-field fault normal and fault parallel ground
motions - - - - - - - - - 57
3.6 Response of S2 subjected to 5 near-field fault normal ground motions - - 61
3.7 Response of S2 subjected to 5 near-field fault parallel ground motions - - 62
3.8 Comparison of response of S2 subjected to 5 near-field fault normal and fault parallel ground
motions - - - - - - - - - 63
3.9 Response of S1 subjected to 5 far-field transverse ground motions - - 66
3.10 Response of S1 subjected to 5 far-field longitudinal ground motions - - 67
3.11 Response of S1 subjected to 5 far-field transverse and longitudinal ground motions 68
3.12 Response of S2 subjected to 5 far-field transverse ground motions - - 71
3.13 Response of S2 subjected to 5 far-field longitudinal ground motions - - 72
3.14 Response of S2 subjected to 5 far-field transverse and longitudinal ground motions 73
3.15 Fault normal and transverse response of S1 subjected to 5 near-field and far-field ground
motions - - - - - - - - - - 76
3.16 Fault parallel and longitudinal response of S1 subjected to 5 near-field and far-field ground
motions - - - - - - - - - - 76
3.17 Fault normal-transverse and fault parallel-longitudinal response of S1 subjected to 5 near-field
and far-field - - - - - - - - - 77
3.18 Fault normal and transverse response of S1 subjected to 5 near-field and far-field ground
motions - - - - - - - - - - 78
3.19 Fault parallel and longitudinal response of S1 subjected to 5 near-field and far-field ground
motions - - - - - - - - - - 78
3.20 Fault normal-transverse and fault parallel-longitudinal response of S2 subjected to 5 near-
field and far-field - - - - - - - - - 79
3.21 Linear displacement response at the crest of S1 and S2 subjected to near-field and far-field
ground motions - - - - - - - - - 79
4.1 Comparison of non-linear response of S1 & S2 subjected to near-field & far-field ground
motions - - - - - - - - - - 92
4.2 Comparison of non-linear response of S1 & S2 subjected to amplified near-field & far-field
ground motions - - - - - - - - - 106
5.1 Non-linear displacement response at the crest of dam, between direct method and sub-
structure method - - - - - - - - - 122
A2.1 Causes of failures of dams around the world with percentage - - - 134
A4.1 Dams with height 100 m and above or with storage capacity of 1 km
3
and above
completed - - - - - - - - - - 145
A4.2 Dams with height 100 m and above or with storage capacity of 1 km
3
and above under
construction - - - - - - - - - - 148
1
Chapter 1
Introduction and Literature Review
1.0 Introduction
Dams are impressive constructions in our world and it is a fascinating chapter of our
history to investigate their origin. The history shows, that these constructions are not
innovations of nowadays, because the first predecessors have existed even 6000 years before
our modern times. A dam is a barrier or structure across a stream, river, or a waterway for the
purpose of confining and controlling the flow of water. Depending upon requirements,
construction of a dam can vary in size and material from small earthen embankments to
massive concrete structures. Primary purpose of dams being irrigation, hydro-electric power
generation, flood control, domestic and industrial water supply etc., makes these structures as
one of the life line structures. As such, dams are cornerstones in the water resources
development of river basins.
Dams are now built to serve several purposes and are therefore known as multipurpose.
With rapid growth of population in India and the consequent demand over water for various
purposes, it has now become necessary not only to construct new dams with revised design
procedures which can sustain worst forces of nature but also to rehabilitate and maintain
existing ones. Natural disasters like earthquake, landslide, cyclone, flood, drought, etc., are quite
common in different parts of India. These can create catastrophe leading to the loss of life,
property damage and socio-economic disturbances. Such losses have grown over the years due
to increase in population and misuse of natural resources.
Among all these natural disasters earthquakes are one of the worst and it is also known
that it is impossible to prevent earthquakes from occurring. However, the disastrous effects of
these can be greatly minimized. This can be achieved through scientific understanding of their
nature, causes, and areas of influence. By identifying the areas, population and structures
vulnerable to hazards, earthquake disaster mitigation and preparedness strategies to those
would reduce miseries to mankind. The study of life line structure like dam is thus required to
design resisting worst forces of nature. One such force of nature is earthquake. Before analyzing
the structure for earthquakes, it is first important to know the prevalent seismic hazard in India.
2
1.1 Prevalent seismic hazard in India
USGS estimates that around 5 lakh earthquakes hit the Earth every year, 1 lakh
of those can be felt, and very few cause damage [http://earthquake.usgs.gov/learn/facts.php].
Moreover, in Indian-Subcontinent, particularly the north-eastern and north-western regions are
the most earthquakes-prone regions of the world. 1988 Bihar earthquake, 1991 Uttarkashi
earthquake, 1993 Killari earthquake, 1997 Jabalpur earthquake, 1999 Chamoli earthquake,
2001 Bhuj earthquake, 2002 Andaman earthquake, 2004 Sumatra earthquake, 2005 Kashmir
earthquake, 2011 Sikkim earthquake are some of the worst hit earthquakes, which cumulatively
have caused over 1 lakh death toll.
Seismic zonation map clearly shows that India is highly vulnerable to earthquake
hazard. During last 100 years, India has
witnessed more than 650 earthquakes of
magnitude 5.0 [Kamalesh Kumar, 2008].
In addition to very active northern and
north-eastern range, the recent events of
1993 Killari (Maharashtra) and Jabalpur
(Madhya Pradesh) in the Peninsular India
have started raising doubts as the disasters
caused by these earthquakes are
alarmingly increasing. Earthquake events
reporting from the Himalayan mountain
range, Andaman and Nicobar Islands, Indo-
Gangetic plain as well as from peninsular
region of India belongs to subduction
category and a few events had also been
under intra-plate category. Figure 1.1: Seismic zonation map of India
Himalayan Frontal Arc (HFA) ranging about 2,500 km long extending from Kashmir in
the west to Assam in the east undergoes subduction process, making it one of the most
seismically active regions in the world. The Indian plate came into existence after initial rifting
of the southern Gondwanaland in late Triassic period. Later the force from spreading of the
Arabian Sea on either side of the Carlsberg ridge caused to continue drifting since mid-Jurassic
to late Cretaceous time to finally collide with the Eurasian plate [Kamalesh Kumar, 2008]. This
led to the formation of Himalayan mountain range and the present day seismicity in this region
is due to the continuous collision between Indian and Eurasian plates. Some of the most
3
important earthquakes that have occurred during the past century in Himalayan Frontal Arc are
tabulated below in Table 1.1.
Table 1.1: Important earthquakes in Himalayan Frontal Arc (Kamalesh Kumar (2008), http://gbpihed.nic.in)
Place Year Magnitude Casualty
Kangra Valley April 4, 1905 8.6 >20,000
Bihar-Nepal border January 1, 1934 8.4 >10,653
Quetta May 30, 1935 7.6 about 30,000
North Bihar 1988 6.5 1000 approximately
Uttarkashi October 20, 1991 6.6 >2,000
Chamoli March 29, 1999 6.8 >150
Hindukush November 11, 1999 6.2 no death reported
Sikkim September 18, 2011 6.9 about 111
The Peninsular India which was once considered as a stable region has started to
experience the earthquakes in increased number because of intra-plate mechanism. Even
though the magnitudes of these are less and recurrence intervals are larger than those of the
HFA, it started to create panic among the inhabitants in this region. Some of the most important
earthquakes that have occurred in Peninsular India in the past are tabulated below in Table 1.2.
Table 1.2: Important earthquakes in Peninsular India (Kamalesh Kumar (2008), http://gbpihed.nic.in)
Place Year Magnitude Casualty
Kachchh June 16, 1819 8.5 No record
Jabalpur June 2, 1927 6.5
Indore March 14, 1938 6.3
Bhadrachalam April 14, 1969 6.0
Koyna December 10, 1967 6.7 >200
Killari (Latur) September 30, 1993 6.3 >10,000
Jabalpur May 22, 1997 6.0 >55
Bhuj January 26, 2001 7.6 >20,000
North-eastern region of India which is one of the six most seismically active regions of
the world lies at the junction of the Himalayan arc to the north and the Burmese arc to the east.
Eighteen large earthquakes with magnitude 7.0 occurred in this region during the last
hundred years (Kayal, 1998). High seismicity in the north-eastern region may be attributed to
the collision tectonics in the north (Himalayan arc) and subduction tectonics in the east
4
(Burmese arc). Some of the most important earthquakes that have occurred in this region of
India in the past are tabulated below in Table 1.3.
Table 1.3: Important earthquakes in Northeastern region of India (Kamalesh Kumar (2008), http://gbpihed.nic.in)
Place Year Magnitude Remark
Cachar March 21, 1869 7.8
Numerous earth fissures
and sand craters
Shillong Plateau June 12, 1897 8.7 About 1542 people died
Sibsagar August 31, 1906 7.0 Property damage
Myanmar December 12, 1908 7.5 Property damage
Srimangal July 8, 1918 7.6
4500 sq km area
suffered damage
S-W Assam September 9, 1923 7.1 Property damage
Dhubri July 2, 1930 7.1
Railway lines, culverts
and bridges cracked
Assam January 27, 1931 7.6 Destruction of Property
N-E Assam October 23, 1943 7.2 Destruction of Property
Upper Assam July 29, 1949 7.6 Severe damage
Upper Assam August 15, 1950 8.7
About 1520 people died.
One of the largest
known earthquake in
the history
Indo-Myanmar
border
August 6, 1988 7.5 No casualty reported.
Seismologists seem not to believe that the frequency in the occurrence of earthquakes
has increased. Unfortunately, earthquakes of higher magnitudes which use to occur in
uninhabited areas or virtually uninhabited areas have hit some thickly populated areas.
Consequently, they have killed thousands of people. Increase in the loss of life and property
damage is due to increasing vulnerability of human civilization to these hazards. This can be
understood by the fact that Kangra event of 1905 (MW=8.6) and Bihar-Nepal of 1934 (MW=8.4)
killed about 20,000 and 10,653 people respectively. On the other hand 1897 and 1950 events of
the northeast (MW=8.7 each) caused death to about 1542 and 1520 people. This is because
Kangra and Bihar-Nepal events struck in densely populated areas of Indo-Gangetic plain. On the
other hand, the north-eastern region was thinly populated in 1897 and 1950 [Kamalesh Kumar,
2008]. The concentration of population has became denser since the time when such major
5
earthquake occurred in these regions, creating more alarming situation and the devastation it
would become if such event occurs now. There are several examples, where high number of
casualties and deaths occurred when the event occurred during early morning hours and quite
opposite when they occurred during the day time even when the epicenter is too near to the
inhabited areas. These examples clearly tell that the time of occurrence of the event and the
epicenter also matters, to quantify loss of life and damage to property.
1.2 Performance of concrete gravity dams subjected to earthquakes
The rst failure of a dam due to earthquake reported in the literature was Augusta Dam,
Georgia, during the 1886 Charleston, South Carolina earthquake. However, the milestone in the
seismic analysis of dams turned after the 1967 Koyna earthquake in India where damage was
caused to the upstream and downstream side of the concrete gravity dam and 1971 San
Fernando earthquake in California where damage was caused to embankment dams (San
Fernando dams) and also to an arch-gravity dam (Pacoima dam). Although such ground motions
caused problems to dams, no serious damages were observed. However, during some
earthquake events, concrete gravity dams were uprooted when blind faults which were lying
below the dam body turned active. These very few events have shown that the earthquake
hazard continues to be a serious threat to dams, as the failure of a full reservoir concrete gravity
dam could cause catastrophe on the downstream.
In the epicentral area of the earthquake, a number of concrete gravity dams have
experienced ground shaking. However, only about 20 dams have been subjected to 0.3g PHGA
or higher without apparent damage. Some of these concrete dams performance to earthquakes
are tabulated below.
Table 1.4: Concrete dams subjected to significant shaking (PHGA > 0.3g) [Courtesy: USSD Proceedings 2012]
Dam
(completed)
Country
Ht.
(m)
Earthquake
name and date
Dist. to
fault
(km)
Mag. PHGA (g) Remarks
Concrete Gravity Dams
Lower Crystal
Springs (1890)
USA 47
San Francisco
Apr 18, 1906
0.4 8.3
0.52 to
0.68 (est.)
Not the slightest
crack
Koyna (1963) India 103
Koyna
Dec11, 1967
3.0 6.5 0.63 (cc) Cracks on both faces
Williams
(1895)
USA 21
Loma Prieta
Oct 17, 1989
9.7 7.1 0.6 (est.) No damage
6
Bear Valley
(1912, 1988)
USA 28
Landers
Jun 28, 1992
45.0 7.4 0.18
Multiple arch
modified to gravity
dam in 1988.
Big Bear
Jun 29, 1992
14.5 6.6 0.57
No damage, except
slight displacement
of crest bridge
girders.
Gohonmatsu
(1900)
Japan 33
Kobe
Jan 17, 1995
1.0 7.2 0.83
No damage on this
masonry dam
Shih-Kang
(1977)
Taiwan 21.4
Chi Chi
Sep 21, 1999
0.0 7.6
0.51 h
0.53 v
Vertical disp. of 9 m,
Rupture of concrete.
Mingtan
(1990)
Taiwan 82
Chi Chi
Sep 21, 1999
12.0 7.6
0.4 to 0.5
(est.)
No damage
Kasho (1989) Japan 46.4
Western Tottori
Oct 6, 2000
3.0-8.0 7.3 0.54
Cracks in control
building at crest
Uh (___) Japan 14
Western Tottori
Oct 6, 2000
1.0-3.0 7.3 1.16
Small crack at
spillway base
Takou (2007) Japan 77
Tohoku
Mar 11, 2011
109.0 9.0 0.38
Cracking of gate-
house walls at crest.
Miyatoko
(1993)
Japan 48
Tohoku
Mar 11, 2011
135.0 9.0 0.32 No damage
Concrete Arch Dams
Gibraltar
(1920, 1990)
USA 52
Santa Barbara
Jun 29, 1925
? 6.3
> 0.3
(est.)
No damage. Modified
in 1990 with RCC.
Pacoima
(1929)
USA 113
San Fernando
Feb 9, 1971
5.0 6.6
0.6 to
0.8
No cracks in arch.
Open joint between
arch and thrust block.
Northridge
Jan 17, 1994
18.0 6.8 0.53
Open joint (2)
between arch and
thrust block
Ambiesta
(1956)
Italy 59
Gemona-Friuli
May 6, 1976
20.0 6.5 0.36 No damage
Rapel (1968) Chile 111
Santiago
Mar 3, 1985
45.0 7.8 0.31
Damage to spillway
and intake tower.
Maule
Feb 27, 2010
232.0 7.8 0.302
Dam performed well.
Cracked pavement.
Techi (1974) Taiwan 185
Chi Chi
Sept 21, 1999
85.0 7.6 0.5
Local cracking of curb
at dam crest.
7
Shapai (2003) China 132
Wenchuan
May 12, 2008
20.0 8.0
0.25 to
0.50
(est.)
No Damage
Concrete Buttress Dams
Hsinfengkiang
(1959)
China 105
Reservoir
Mar 19, 1962
1.1 6.1 0.54
Horizontal cracks in
top part of dam
Sefid Rud
(1962)
Iran 106
Manjil
Jun 21, 1990
Near
dam
site
7.7 0.71 (est.)
Horizontal cracks
near crest, minor
disp. of blocks
Notes: Legend: Ht.=height, est.=estimated, Dist.=distance, Mag.=magnitude (ML
or MB for less than 6.5 and MS above 6.5), cc=cross canyon, h=horizontal,
v=vertical. PHGA=Peak Horizontal Ground Acceleration, disp.=displacement
Table 1.4 illustrates about the worldwide performance of concrete (Arch, Buttress &
Gravity) dams subjected to ground motions > 0.3g. From the table it can be concluded that
concrete dams have performed well when subjected to high intensity accelerations. There might
be several reasons why concrete dams have performed well and consistently well than that
predicted by design or analysis when shaken by an earthquake. However, the dams present in
highly seismic zones are always under threat as some dams have performed less than what was
expected. Several factors like magnitude, epicentral distance, PHGA, range of frequency can
solely vary the performance of dams subjected to earthquakes. A thorough understanding on
the ground motions should be studied for that respective area before the construction of dam.
For this huge number of strong motion recorders should be placed at or near the dam
sites, which would increase our knowledge over the performance of severely shaken concrete
dams and that knowledge could be applied in designing future dams. The most significant
factors other than magnitude to be considered in determining the response of concrete dams
are the epicentral distance to the dam, PHGA and also the spectral acceleration at the natural
frequency of the dam. PHGAs get amplified from the base of the dam to the crest and peak
accelerations at the crest would be greater when the reservoirs are full. The epicentral distance
also has got its effect when the high velocity energy pulse to hit the dam, causing near-field
earthquake effect on dam. Also if the natural frequency of the dam matches with the frequency
of the ground motion there would be questions raised over the performance of concrete dams.
Even though there are several potential failure modes like foundation problems,
settlement, base sliding etc., general accepted failure mode for concrete dams during
earthquake is cracks in the concrete of the dam body. Most of the concrete dams listed in Table
1.4 when subjected to severe shaking were observed with cracks in the concrete and that too at
8
the change in location of geometry. While concrete dams are designed to withstand severe
shaking and have performed well in the past, it should not be considered as a positive sign of
their performance in the future. Utmost care in design and construction practices should be
taken and special attention towards possible faults located near the dam should be given.
1.3 Literature review
Failure of a concrete gravity dam subjected to an earthquake is a very rare case. Even
though there are a very few number of concrete gravity dams which suffered minor damages
when subjected to earthquakes, the effect that an earthquake can cause even on such a huge
construction is already known. One such example is failure of Shih-Kang dam. Since past few
decades, several researchers have conducted research even by considering site conditions, dam-
foundation, and dam-foundation reservoir interactions.
For the current study, a thorough literature review has been conducted to understand
the past research work of earthquake effects on concrete gravity dams. From the conclusions
through conducted literature review, the inspiration to take our study in a new angle has been
marked. The studies presented for literature review are categorized as:
Analytical Studies
Experimental Studies
Numerical Studies
1.3.1 Analytical Studies
Tatsuo Ohmachi (1999) had done fundamental study on near-field effects on earthquake
response of arch dams. Effects of directivity of near-field and vertical ground motions on linear
response of an arch dam were also studied. The strong motion record observed at Pacoima Dam
station during the 1994 Northridge earthquake, California was taken as an input motion and
configuration of Morrow Point dam as a model of an arch dam had been used. Also, while
investigating directivity, the variation of stresses at all elements on u/s and d/s face of the dam
due to different angle of incident were studied. In accordance with remarkable directivity or
polarization in the horizontal ground motion, the arch dam shows a remarkable variation in both
arch and cantilever stresses. The range of variation was different between the arch and cantilever
stresses. The large variation amounting to the factor of 3 and even 4 was seen in the arch stress
on the upstream face below the crest around arch crown or abutment, while similar amount of
variation in the cantilever stress was seen on the downstream face near the deepest bottom. The
contribution of vertical component to the total response is about 30%.
9
1.3.2 Experimental Studies
William P. Donlon (1989) had investigated nonlinear seismic response of concrete gravity
dams experimentally using small-scale models. Of primary interest are crack formation, crack
opening and closing, and sliding along crack planes. Stability of structure after cracking was also
studied. Three small-scale models of a single monolith of Pine Flat Dam were tested to determine
the extent of such behaviour and its effect on structural stability. The models were constructed of
one polymer-based and two-plaster-based materials developed for these experiments. The
plaster-based materials fulfil the strength, stiffness, and density requirements established by the
laws of similitude, while the polymer-based material fulfils only the stiffness and density
requirements and are used only in the lower part of the dam where cracking is not expected.
Tests were performed with and without water in the reservoir. The results of the experiments
indicate that the neck region of a concrete gravity dam is most susceptible to cracking, although
crack propagation files can differ as a result of variations in excitation, material properties and
construction techniques. These results also indicate alternate design techniques which could
improve the seismic stability of a cracked gravity dam.
Jean Proulx (1997) had done experimental and numerical investigation of dam-reservoir-
foundation interaction for a large gravity dam. Forced-vibration tests were completed on Outardes
3 gravity dam, located in north-eastern Quebec, Canada. The experimental results were
subsequently used as a basis for a numerical correlation study to evaluate the performance of
state-of-the-art finite element programs for earthquake analysis of concrete dams. The
experimental procedure was presented and involved in the recording of acceleration responses
on the 84-m-high dam under harmonic loading. Hydrodynamic pressures were also recorded at
several locations in the reservoir, up to a distance of 90m from the dam face. Extensive studies
were carried out with two- and three-dimensional models. Numerical results were compared
with complete frequency responses for accelerations and pressures obtained on site. It was
demonstrated that the two-dimensional approach could only predict the fundamental resonance
of the system, and that a three-dimensional for the dam-reservoir-foundation system including
water compressibility could reproduce the experimental behaviour with accuracy.
1.3.3 Numerical Studies
Anil K Chopra (1981) had proposed a general procedure to analyze the response of
concrete gravity dams, including the dynamic effects of impounded water and flexible foundation
rock, to the transverse and vertical components of earthquake ground motion. The analysis
procedure developed was specifically for the two-dimensional analyses of gravity dam monoliths
supported on the surface of a visco-elastic half-plane and impounding water in the reservoir
10
with horizontal bottom. The problem was analysed under the assumption of linear behaviour for
the concrete, foundation rock and water. Based on the analytical procedures developed, a
computer program has been written to evaluate numerically the responses of concrete gravity
dams, including various effects of both the impounded water and the foundation rock.
S.S. Bhattacharjee (1993) had studied the seismic fracture analysis of concrete gravity
dam using finite element method. He has proposed the smeared crack analysis model based on
the non-linear fracture behaviour of concrete. The features that he had considered in the
development of the model are (i) the strain softening of concrete due to micro cracking; (ii) the
rotation of the fracture band with the progressive evolution of micro crack damage in finite
elements; (iii) the conservation of fracture energy; (iv) the strain-rate sensitivity of concrete
fracture parameters; (v) the softening initiation criterion under biaxial loading conditions; and
(vi) the closing-reopening of cracks under cyclic loading conditions. A two-dimensional seismic
response analysis of Koyna dam was then performed to demonstrate the application of the
proposed non-linear fracture mechanics model. His study had shown that the continuum
mechanics approach could efficiently predict the localized cracking response of concrete gravity
dams if applied with appropriate constitutive models and the seismic fracture response of Koyna
dam had been satisfactorily reproduced in the analyses using the non-linear smeared fracture
model.
Abdolrahim Jalali (2000) had studied the aspects of concrete dams response to near-field
ground motions. In this investigation the effects of near-field ground motions on concrete dams
are assessed using a collection of some records from actual earthquakes. All of these records
exhibit main characteristics of near-field ground motions. Another record, the 1940 El-Centro
motion in which near-field effects are absent, is used as a reference. The effects of near-field
ground motions on displacements, and stresses of upstream and downstream faces of Morrow-
Point Arch dam, and on stresses, displacements, and base sliding of Pine Flat Concrete gravity
dam have been evaluated.
D.K. Paul (2002) had studied the seismic safety of 260.5 m high rock-fill Tehri dam
located in seismically active zone. Sequential non-linear static analysis, free vibration analysis
and the earthquake response analysis of the rock-fill dam had been carried out taking into
account the effects of material property variation with confining pressure, non-linearity, the
schedule of raising the dam and the schedule of reservoir filling. The non-linear sequential and
earthquake response analyses had given insight of the dam response with respect to stresses,
plastic deformations, peak accelerations and displacements. Quantitatively, the high rock-fill dam
was found to undergo plastic deformations mostly near the u/s and d/d slopes, which were taken
11
care of by providing a sumptuous riprap composed of blasted rock from a quarry of competent
quartzite, both in u/s and d/s. The top portion of the fill dam was build with the same material to
withstand severe accelerations. The designers considered that the dam based on the detailed
dynamic analysis carried out for several worst case scenarios and different hazard levels, suffer
deformation to an extent within acceptable limits under postulated MCE. Finally their study
concluded that the dam design was structurally safe to withstand the MCE.
Tatsuo Ohmachi (2003) had studied near-field effects of hidden seismic faulting on a
concrete dam. The 2000 Western Tottori earthquake (Ms 7.3), Japan, was caused by a hidden
seismic fault underlying the Kasho Dam, a 46 m high concrete gravity dam. Strong-motion
accelerometers registered peak accelerations of 2000 gal at the top and 500 gal in the lower
inspection gallery. Integration of the acceleration records in the gallery showed a permanent
displacement of 28 cm to the north, 7 cm to the west, and an uplift of 5 cm. This dam survived
the earthquake without serious damage. However, the reservoir level dropped suddenly by 6 cm
followed by damped free vibration that continued several hours. Based on numerical simulation
and field observations, the water level change is attributed to ground displacement in the near-
field and subsequent seiching of the reservoir. The vibration period of the dam in the u/s-d/s
direction changed noticeably during the main shock, probably due to hydrodynamic pressure
variation. The earthquake caused cracking of concrete floor beams in a sub-gate control room,
which was repaired by post-tensioning with steel bars, resulting in increased beam rigidity and
micro-tremor instruments were used to evaluate the effectiveness of the repair work. Based on
the experiences at this site, author had strongly recommended that the earthquake response not
only of the dam body but of appurtenant structures such as sub-gates control rooms should be
considered in the structural design of dams.
Vahid Lotfi (2004) had studied dynamic analysis of concrete arch dams based on finite
element-boundary element (FE-BE) procedure including non-uniform ground motion. In this
technique, dam body was discretised by finite elements, while foundation rock domain was
handled by three dimensional boundary element formulations with no restrictions imposed on
geometry of canyon shape. Based on this methodology, the program 'MAP-75' was enhanced and
the response of Morrow Point arch dam was studied for uniform and non-uniform ground
excitations. The efficiency of the technique relies heavily on an interpolation scheme, which was
used for evaluation of foundation impedance matrix and force vector. Finally, the responses were
compared for uniform and non-uniform ground motions for different types of excitation
directions. It was observed that response for low frequency was practically the same for both
uniform and non-uniform ground motions. It was also observed that for moderate frequencies,
the response was mainly higher for uniform ground motion in comparison with non-uniform
12
ground motion. While at high frequencies, the opposite behaviour was noticed. This was true for
all three types of excitation direction being considered.
Worakanchana Kawin (2005) had studied the failure mechanism of Shih-Kang dam by
Applied Element Method (Pradeep Kumar Ramancharla, 2001). Static non-linear analysis and
parametric study including dip angle of the fault, fault location were studied. Also, the redesign of
Shih-Kang dam was proposed. It was found that concrete dam like the case of Shih-Kang dam can
resist very low amount of fault induced ground rupture and deformation. From numerical result,
Shih-Kang dam damage mechanism starts from the separation of the dam from its foundation,
crack from the top of the dam, shear crack and compression failure. According to the parametric
study, it was found that the normal fault, if occur under the dam will damage the dam at the lower
displacement than the reverse fault. Also, different fault location affects in the different shear
span length. The longer the shear span's length, the more displacement the dam can resist. The
proposal for the rehabilitation of Shih-Kang dam was proposed by placing the slip joint and
reinforcement, FRP or expansive concrete to control the crack width of the dam. To place the slip
joint, the accurate knowledge of fault characteristic in the area must be known.
Rajib Sarkar (2007) had studied the response of a dam subjected to dynamic loading in a
combined effect of the interaction among dam, reservoir and foundation systems. They adopted
the profile of the Koyna dam for the study. Also, nonlinear concrete properties had been taken
into account through concrete damaged plasticity model to simulate the damage induced in the
dam body under a real-time earthquake motion. Tensile damage to the dam structure occurred
during the earthquake motion had been studied and the same had been studied by making few
changes in parameters like varying the height of the reservoir and the foundation modulus
values, to show the influence of reservoir and foundation materials on the dynamic response of
concrete gravity dams. They concluded that with the decrease in the foundation modulus, the
displacements increased and if reservoir depth considered was less than 0.7 times the full-
reservoir depth, the reservoir had no significant impact on the dynamic behaviour of the dam
structure. Also the tensile damage patterns of the systems considering (i) only the dam structure,
and (ii) the dam with the foundation were almost same. However, when the reservoir was
considered along with the dam and the foundation, the damage pattern obtained was different
and it showed that the crack initiates at the point of slope change in the downstream side of the
dam structure and then grows at nearly the same level.
Javad Moradloo (2008) had studied nonlinear dynamic analysis of concrete arch dam. A
three dimensional smeared crack model was used to consider nonlinear behaviour of mass
concrete. The proposed model considers major characteristics of mass concrete under three
13
dimensional loading conditions. Those characteristics were pre-softening behaviour, softening
initiation criteria, fracture energy conservation and strain rate effect. After verification of the
proposed model by some available numerical tests, dynamic analysis of Morrow Point concrete
arch dam under the three components of the Taft earthquake scaled to 1.0g was carried out. In
the analysis, complete fluid-structure interaction was considered accounting for fluid
compressibility and absorptive reservoir boundary condition approximately. The coupled
equation resulting from dam-reservoir interaction was solved using staggered method. The
deduced results showed that crack pattern had good agreement with the contour of maximum
principal stresses and the proposed algorithm also gives reliable solution even in large time steps.
1.4 Scope of present study
Earthquakes and dams both are not new to this world. However, when the effect of strong
motion over different range of structures is known and importance of life line structure like dam
is also known it should be seen that the dams are constructed to resist earthquakes. India is a
country with over 1,000 active faults and 5,100 large dams constructed and few more under
construction. Most of these dams are in highly seismic zones under the threat to experience
severe earthquakes. In an earthquake event, factors like magnitude, peak ground acceleration,
velocity pulses, permanent displacement, epicentral distance, directivity, orientation of fault, local
site conditions, soil-structure interaction etc., individually have varying effects on dams. However,
in our study we have restricted ourselves with ground motions recorded with epicentral distance
less than 10 km, having forward directivity effects, for different types of faults. For complete
analysis other parameters should have also been considered. However, with limitation of data and
complexity in solving problem, we had certain limitations.
Current study would give a scope for proper understanding of how the effects would vary
between near-field and far-field earthquakes. Even though very few concrete dams have
undergone minor failures when subjected to earthquakes, near-field earthquakes have capacity
to destabilize the dams. This current study would give an understanding of how near-field
earthquakes affect concrete gravity dams. Codal provisions in the country haven't considered or
revised near-field effects for seismic design of dams. In this regard, future scope of our work
would lead towards including near-field earthquake effects.
1.5 Organization of thesis
Thesis work on "Numerical Modelling of Near-Field Earthquake Effects on Concrete Gravity Dam"
is totally divided into 6 chapters.
14
Chapter (2) describes the selection of dam and its geometrical details. It also explains about the
ground motions considered and their characteristics. Finally Applied Element Method (AEM)
(Hatem, 1998) the numerical modelling used is described with its formulation and limitations.
Chapter (3) describes the comparison of linear earthquake response between components of
near-field and far-field on non-over flow section of dam with foundation system and on non-over
flow section of dam with foundation and base.
Chapter (4) describes the comparison of non-linear earthquake response between components
of near-field and far-field on non-over flow section of dam with foundation system and on non-
over flow section of dam with foundation and base. Initiation of cracks and its propagation is also
described.
Chapter (5) describes the non-linear response of concrete gravity dam subjected to fault motion.
Parametric study is done, by modeling the dam on hanging wall and foot wall. The behaviour of
dam, initiation of cracks and its propagation is also described.
Conclusions and future scope of work are given in Chapter (6).
Appendix (A1) gives the information on history and necessity of dams.
Appendix (A2) gives different causes of dam failures.
Appendix (A3) gives a special focus on near-field earthquake effects.
Appendix (A4) gives the information on large dams in India.
15
Chapter 2
Numerical Modelling of Concrete Gravity Dam
2.1 Introduction
This chapter mainly focuses in describing the selection process of a concrete gravity
dam in India, for the analysis based on its importance. Geometrical details of dam, its material
properties, and ground motions considered for the analysis and their characteristics are
described. Later a numerical method called Applied Element Method [Hatem et.al. 1998, 2000],
its mathematical formulation, modeling limitations and its application in analysis of concrete
gravity dam subjected to near-field and far-field ground motions are described.
2.2 Selection of concrete gravity dam
By the time India got independence in 1947, there were less than 300 large dams in
India. At present this number has grown to about 5,100 [National Register of Large Dams
2009], with 181 dams of national importance. India is a seismically active country, with history
of major earthquakes having occurred in the past. North-eastern and north-western parts of
India are seismically very active as the Indo-Australian plate is sub-ducting under Eurasian
plate at this region. Along with intra-plate earthquakes, India has witnessed several severe
earthquakes.
These earthquakes resulted in the failures of different range of structures from small
buildings to major dams. The 1967 Koyna earthquake is one example, which caused minor
damage to the upstream and downstream side of Koyna dam without causing flooding. As it is a
concrete gravity dam, it has withstood the event. However, there are several other examples in
India, where hundreds of earth dams have failed due to earthquakes. As the reservoir storage
capacity of major concrete dams is comparatively very high from earth dams, failure of it might
lead to catastrophe. 1999 Chi-Chi, Taiwan earthquake is one example during which, the Shih-
Kang dam has failed due to differential thrust fault movement. Over two-thirds of the dam body
were uplifted about 9m vertically and displaced 2m horizontally. This clearly indicates how
devastating an earthquake could be even towards dams, which are known as the world's largest
constructions.
16
In this concern, a large dam has been selected from India, by taking few parameters into
consideration. Depending upon services utilized from dams for multi-purposes, NRLD has
classified a list of 181 dams among 5,100 dams as National Importance Dams. Due to
unavailability of complete data of 181 dams from 2009 report, data from previous report
available has been considered. By the time report of National Importance Dams was made in
2002, there were 4,525 dams and 70 National Importance Dams [Appendix-A2]. These dams are
placed on India map with the information of their latitude-longitude. Later the map is
overlapped with seismic zonation map of India along with 1040 active faults and lineaments
[Geological Survey of India, 2000]. Top 100 active faults are identified based on the cumulative
energy generated from those faults in the previous earthquakes. Finally a map is prepared by
over lapping National Importance Dams of India, Seismic Zonation map of India and top 100
active faults of India. For the study, among 70 national importance dams, dams >100m (23
dams) are listed and dams in seismic
zones IV and V (11 dams) are further
short listed. In those final 11 short listed
dams, the seismic safety of 103m high
concrete gravity Koyna dam located in
seismic zone IV had been widely debated
and because of that probably this dam has
been studied very extensively. Failure of
such a large dam would evidently create
catastrophe. So, for our study we thus
have considered Koyna dam for the
analysis as it has undergone MW 6.3
magnitude earthquake on 11
th
December
1967. It has also experienced 17
earthquakes of MW5.0 and also over 150
earthquakes of MW 4.0 (Harsh K Gupta
et.al. 2002). Figure 2.1: National Importance Dams of India on Seismic
Zonation map & top 100 active faults
2.2.1 Dam geometry
Koyna hydro electric project is located in Maharashtra, India. The dam is primarily
constructed for the generation of power, with some irrigation on the banks of river Krishna. The
principle feature of the project is Koyna dam, which is a concrete gravity structure. With 892 sq.
km of catchment area, the dam and the reservoir are located on the Indian Peninsular shield,
17
one of the oldest continental blocks on the earths surface. Prior to the year 1962, this shield
was referred to as a stable rock. However, the December 11th, 1967 earthquake of MW 6.3 in the
Koyna region contradicted all these beliefs. However, the dam in the region withstood this
significant seismic activity without much damage. For the study we have considered two models
first one is a non-overflow section of the dam along with foundation and the second model is a
non-overflow section of dam with foundation and base. These two models are considered and
studied individually.
2.2.1.1 Non-overflow section of dam with foundation (Structure S1)
The structure is 103m high and 70m wide at the base and 15m wide at the crest. The
dam on the downstream is planned and designed out of conventional style by providing large
neck portion in an inclined position. The change of cross section of dam at the neck happens at
67m height from the base. Width at this height of 67m, where the cross section gets changed is
20 m. Also the dam is provided with a concrete foundation of 100m x 5m (fig. 2.3). The whole
dam is divided into many blocks; with separation joints being provided between two adjacent
blocks, each being 15m wide. Even though the numerical modeling is in 2D it considers the
effect of thickness therefore, thickness of 15m is considered for modeling. The bottom of the
dam is considered as fixed for this model and analysis is done individually on it.
2.2.1.2 Non-overflow section of dam with foundation and base (Structure S2)
Foundation conditions depend upon the geological character and thickness of the strata
which should carry the weight of dam, their inclination, permeability, and relation of underlying
strata, existing faults and fissures. The foundation will limit the choice of type to a certain
extent, although such limitations can frequently be modified, considering the height of the
proposed dam. For the foundation base, homogeneous strata of granite rock of length 800m and
height 50m is modeled. The dam with concrete foundation is placed at the centre of the base for
the study. The dimensions and material properties of the dam remains the same as S1.
Figure 2.2: Cross-section and material properties of Koyna dam with foundation and base (Structure 2)
50 m
800 m
18
Figure 2.3: Cross-section and material properties of Koyna dam with foundation (Structure 1)
19
2.2.2 Material Properties
For the dam, properties like density, Poisson's ratio, compression resistance and tension
resistance values are taken from a reference paper [Rajib Sarkar et al., 2005]. In which, they
have studied the non-linear behavior of Koyna dam by considering that the properties of Koyna
dam taken are actual from its design.
Properties of Dam
Youngs Modulus (Ec): 3.1027x10
7
kN/m
2
Poissons Ratio (): 0.2
Density of Concrete (): 2.643 t/m
3
Tension Resistance (t): 2.58x10
3
kN/m
2
Compression Resistance (c): 2. 41x10
4
kN/m
2
Even though the site is one of the oldest continental blocks of earth's surface, the dam is
assumed to be built on hard strata and analyzed. Therefore, homogeneous granite rock is
considered as base.
Properties of base
Youngs Modulus (E): 6.0x10
7
kN/m
2
Poissons Ratio (): 0.2
Density of Granite rock (): 2.7 t/m
3
Tension Resistance (t): 1.6x10
4
kN/m
2
Compression Resistance (c): 1.6x10
5
kN/m
2
2.3 Ground motions and their characteristics
Seismic inputs are the earthquake data that are necessary to perform different types of
seismic analysis. In the context of seismic analysis and seismic design of structures, various
earthquake data can be considered depending upon the nature of analysis to carry out. Seismic
inputs for structural analysis are provided either in the time domain or in the frequency
domain, or in both time and frequency domains. In addition, a number of earthquake
parameters are also used as seismic inputs for completeness of the information, which are
required to perform different types of analysis. These include magnitude, intensity, peak ground
acceleration, peak ground velocity, peak ground displacement, duration, predominant ground
20
frequency etc. Selection of ground motions used in the study and their characteristics are
described in coming sections.
2.3.1 Selection of ground motions
For the study, we have considered 10 near-field ground motions (fault normal and fault
parallel components) and for the same near field events 10 far-field ground motions (transverse
and longitudinal components) are considered and among these 10 only 5 ground motions
depending on their characteristics have been selected for the study.
The PGA's of 10 near-field fault normal ground motions ranges between 0.432 - 1.088,
PGA's of 10 near-field fault parallel ground motions ranges between 0.37 - 0.978 are considered.
For the same events, PGA's of 10 far-field transverse ground motions ranges between 0.027 -
0.496 and PGA's of 10 far-field longitudinal ground motions ranges between 0.0262 - 0.515 are
selected. The PGA's and duration of ground motions ranges from low to high and the frequency
content ranges from resonating to non-resonating frequencies. The details of the ground
motions are listed from Table 2.1 to Table 2.4 and the ground motion records and their Fourier
amplitude spectrums are shown from fig. 2.4 to fig. 2.23
2.3.2 Characteristics of ground motions
It is necessary to describe the characteristics of the ground motion that are of
engineering significance and to identify a number of ground motion parameters that reflect
those characteristics. For engineering purposes, three characteristics of earthquake motions (1)
amplitude, (2) frequency content, and (3) duration of the motion are important to be studied.
Plenty of different ground motion parameters have been proposed, each of which provides
information about one or more of these characteristics. In practice, it is usually necessary to use
more than one of these parameters to characterize a particular ground motion adequately
(Kramer, 1996). These (amplitude, frequency, duration) characteristics differ dramatically
between near-field and far-field ground motions.
Current study is on comparison of the behavior of a concrete gravity dam subjected to
effects caused by near-fault ground motions with the effects caused by far-field ground motions
of the same event. Near-fault ground motions are different from ordinary motions in that they
often contain strong coherent dynamic long period pulses and permanent ground
displacements. The dynamic motions are dominated by a large long period pulse of motion that
occurs on the horizontal component perpendicular to the strike of the fault, caused by rupture
directivity effects. Forward rupture directivity causes the horizontal strike-normal component
of ground motion to be systematically larger than the strike-parallel component at periods
21
longer than about 0.5 seconds [Paul Somerville, 2005]. However, near fault recordings from
recent earthquakes indicate that the pulse is a narrow band pulse whose period increases with
magnitude, causing the response spectrum to have a peak whose period increases with
magnitude, such that the near-fault ground motions from moderate magnitude earthquakes
may exceed those of larger earthquakes at intermediate period. Parameters like rupture
directivity, recordings close to the epicenters, faulting mechanism and duration can cause
changes in the characteristics of near-field ground motions.
In near-field ground motions, the directivity effects play their role with varied
characteristics. Forward rupture directivity effects occur when two conditions are met: the
rupture front propagates towards the site, and the direction of slip on the fault is aligned with
the site [Paul Somerville, 2005]. The conditions for generating forward rupture directivity
effects are readily met in strike slip faulting, where the rupture propagates horizontally along
strike either unilaterally or bilaterally, and the fault slip direction is oriented horizontally in the
direction along the strike of the fault. However, not all near-fault locations experience forward
rupture directivity effects in a given event. Backward directivity effects, which occur when the
rupture propagates away from the site, give rise to the opposite effect: long duration motions
having low amplitudes at long periods. The conditions required for forward directivity are also
met in dip-slip faulting. The alignment of both the rupture direction and the slip direction updip
on the fault plane produces rupture directivity effects at sites located around the surface
exposure of the fault (or its updip projection if it does not break the surface) [Paul Somerville,
2005].
Amplitude: Horizontal accelerations have commonly been used to describe the ground
motions. The peak horizontal acceleration for a given component of motion is simply the largest
(absolute) value of horizontal acceleration obtained from the accelerogram of that component.
The largest dynamic forces induced in certain types of structures (very sti) are closely related
to the peak horizontal accelerations [Kramer, 1996].
Frequency: Earthquakes generate intricate loading with components of motion spanning a
wide range of frequencies. This frequency content will describe how the amplitude of ground
motion is distributed among dierent frequencies. The frequency content of any ground motion
would strongly inuence the motion of structure. The broad band width of the Fourier
amplitude spectrum is the range of frequencies over which certain level of Fourier amplitude is
exceeded.
Generally band width is measured at a level of 1/2 times of maximum Fourier amplitude.
22
The Fourier transform of an accelerogram ( ) t x is given by,
}
= dt e t x X
t i
) (
2
1
) ( (2.1)
Where, ( ) t x is the acceleration record and is frequency.
Duration: The duration of strong ground motion can have a strong inuence on earthquake
damage. It is related to the time required for accumulation of strain energy by rupture along the
fault. There are dierent procedures for calculating the duration of ground motion, out of which
we have considered Trifunac and Brady (1975) method for calculating the duration of ground
motion.
Trifunac and Brady Duration (1975) is based on the time interval between the points at which
5% and 95% of the total energy has been recorded.
Details of the ground motions with magnitude, epicentral distance, PGA, duration and
frequency range are given below from table 2.1 to table 2.4. Further the Ground motions and
their Fourier Amplitude Spectrums are given from fig. 2.4 to fig. 2.23.
23
Table 2.1: Details of near-field - fault normal ground motions
Sl.
No
Near-Field - Fault Normal ground motions
Earthquake
Date of
occurrence
Station MW
Epicentral
Distance
(Km)
PGA
(g)
Duration
(sec)
Frequency
(Hz)
1 Tabas
16-09-1978
Tabas 7.4 1.2 0.9 18.52 0.012-5.444
2 Loma Prieta
18-10-1989
Los Gatos 7.0 3.5 0.718 6.24
0.024-1.099
3 Loma Prieta
18-10-1989
Lex. Dam 7.0 6.3 0.686 3.27
0.024-0.83
4
Erzincan
13-03-1992
Meteorogical
Station
6.7 2.0 0.432 7.14
0.024-1.489
5 Cape Mendocino
25-04-1992
Petrolia 7.1 8.5 0.638 15.84
0.073-2.637
6
Landers
28-06-1992
Lucerne
valley
7.3 1.1 0.713 13.38
0.015-17.71
7 Northridge
17-01-1994
Rinaldi 6.7 7.5 0.89 7.01
0.049-1.709
8 Northridge
17-01-1994
Olive View 6.7 6.4 0.732 5.82
0.012-3.284
9 Kobe
17-01-1995
Kobe 6.9 3.4 1.088 7.5
0.269-2.625
10 Kobe
17-01-1995
Takatori 6.9 4.3 0.786 10.69
0.024-0.952
Table 2.2: Details of near-field - fault parallel ground motions
Sl.
No
Near-Field - Fault Parallel ground motions
Earthquake
Date of
occurrence
Station MW
Epicentral
Distance
(Km)
PGA
(g)
Duration
(sec)
Frequency
(Hz)
1 Tabas
16-09-1978
Tabas 7.4 1.2 0.978 17.64
0.012-6.519
2 Loma Prieta
18-10-1989
Los Gatos 7.0 3.5
0.458
10.67
0.024-1.172
3 Loma Prieta
18-10-1989
Lex. Dam 7.0 6.3
0.37
5.08
0.024-2.075
4
Erzincan
13-03-1992
Meteorogical
Station
6.7 2.0
0.457
10.085
0.024-1.904
5 Cape Mendocino
25-04-1992
Petrolia 7.1 8.5
0.655
17.1
0.378-2.979
6
Landers
28-06-1992
Lucerne
valley
7.3 1.1
0.799
14.224
1.373-17.50
7 Northridge
17-01-1994
Rinaldi 6.7 7.5
0.389
9.545
0.049-4.932
8 Northridge
17-01-1994
Olive View 6.7 6.4
0.595
6.06
0.732-3.271
9 Kobe
17-01-1995
Kobe 6.9 3.4
0.575
7.94
0.635-3.186
10 Kobe
17-01-1995
Takatori 6.9 4.3
0.424
13.13
0.024-1.953
24
Table 2.3: Details of far-field - transverse ground motions
Sl.
No
Far-Field - Transverse ground motions
Earthquake
Date of
occurrence
Station MW
Epicentral
Distance
(Km)
PGA
(g)
Duration
(sec)
Frequency
(Hz)
1 Tabas
16-09-1978
Sedeh 7.35 177.9
0.027
29.48 0.024-1.147
2 Loma Prieta
18-10-1989
HCL 6.93 47.9
0.215
13.62
0.024-1.464
3 Loma Prieta
18-10-1989
Sunnyvale 6.93 42.13
0.209
25.29
0.024-1.806
4 Erzincan
13-03-1992
Erzincan 6.69 8.97
0.496
7.35
0.024-1.465
5 Cape Mendocino
25-04-1992
Petrolia 7.01 53.34
0.178
19.84
0.292-3.625
6 Landers
28-06-1992
Anaheim 7.28 146.11
0.0353
25.62
0.012-2.00
7 Northridge
17-01-1994
Anacapa
Island
6.69 77.39
0.0367
13.90
2.294-6.933
8 Northridge
17-01-1994
Anaheim 6.69 70.45
0.0659
18.40
0.073-4.956
9 Kobe
17-01-1995
99999HIK 6.9 135.63
0.148
11.08
0.50-2.795
10 Kobe
17-01-1995
99999FUK 6.9 196.18
0.0422
32.62
0.280-2.783
Table 2.4: Details of far-field - longitudinal ground motions
Sl.
No
Far-Field - Longitudinal ground motions
Earthquake
Date of
occurrence
Station MW
Epicentral
Distance
(Km)
PGA
(g)
Duration
(sec)
Frequency
(Hz)
1 Tabas
16-09-1978
Sedeh 7.35 177.9
0.0262
29.66 0.024-4.785
2 Loma Prieta
18-10-1989
HCL 6.93 47.9
0.247
17.40
0.024-1.36
3 Loma Prieta
18-10-1989
Sunnyvale 6.93 42.13
0.207
21.175
0.024-1.831
4 Erzincan
13-03-1992
Erzincan 6.69 8.97
0.515
7.46
0.024-1.806
5 Cape Mendocino
25-04-1992
Petrolia 7.01 53.34
0.154
20.78
0.012-2.722
6 Landers
28-06-1992
Anaheim 7.28 146.11
0.0471
27.58
0.012-2.197
7 Northridge
17-01-1994
Anacapa
Island
6.69 77.39
0.0673
12.30
1.635-6.665
8 Northridge
17-01-1994
Anaheim 6.69 70.45
0.0724
19.72
0.024-4.468
9 Kobe
17-01-1995
99999HIK 6.9 135.63
0.147
17.02
0.732-3.027
10 Kobe
17-01-1995
99999FUK 6.9 196.18
0.0337
41.90
0.402-2.868
25
2.4 Near-Field ground motions and their Fourier amplitude spectrums
Figure 2.4: Tabas near-field (a) Fault normal ground motion (b) Fault normal Fourier amplitude spectrum (c) Fault
parallel ground motion (d) Fault parallel Fourier amplitude spectrum
Tabas is a city in and capital of Tabas County, Yazd Province, IRAN. The city has
experienced a severe earthquake on September 16
th
, 1978. Tectonics of east-central Iran as part
of the Alpine-Himalayan orogenic belt has been the subject of numerous discussions as the
region has a complex tectonic history of repeated folding, magmatism and metamorphism. The
nearest prominent tectonic boundary of global significance is the main Zagros thrust line to the
southwest, which marks the plate boundary between two continental masses of varied tectonic
environment. Fault normal and fault parallel ground motion records of near-field considered for
the study are produced in fig. 2.4. Tabas station which is 1.2 km from epicenter recorded the
magnitude of the near-field ground motion as MW 7.4. PGA's of fault normal and fault parallel
are 0.9g and 0.978g respectively. Predominant frequency ranges between 0.012-5.444 for fault
normal and 0.012-6.519 for fault parallel.
Tabas - Tabas Station
26
Figure 2.5: Loma Prieta - Los Gatos near-field (a) Fault Normal ground motion (b) Fault normal Fourier amplitude
spectrum (c) Fault Parallel ground motion (d) Fault parallel Fourier amplitude spectrum
Loma Prieta is a Northern California mountain in the Santa Cruz Mountains, CA, USA. On
October 18
th
, 1989 a right-lateral strike-slip along the San Andreas fault which lasted for about
15 seconds resulted in a severe earthquake. This major earthquake struck the San Francisco Bay
Area. Fault normal and fault parallel ground motion records of near-field considered for the
study are produced in fig. 2.5. Los Gatos station which is 3.5 km from epicenter recorded the
magnitude of the near-field ground motion as MW 7.0. PGA's of fault normal and fault parallel
are 0.718g and 0.458g respectively. Predominant frequency ranges between 0.024-1.099 for
fault normal and 0.024-1.172 for fault parallel.
Loma Prieta Los Gatos Station
27
Figure 2.6: Loma Prieta - Lexington Dam near-field (a) Fault Normal ground motion (b) Fault normal Fourier
amplitude spectrum (c) Fault Parallel ground motion (d) Fault parallel Fourier amplitude spectrum
Another near-field record at Lexington Dam station which is at 6.3 km from epicenter
recorded 7.0 MW. Fault normal and fault parallel ground motion records of near-field considered
for the study are produced in fig. 2.6. PGA's of fault normal and fault parallel are 0.686g and
0.37g respectively. Predominant frequency ranges between 0.024-0.83 for fault normal and
0.024-2.075 for fault parallel.
Loma Prieta Lexington DamStation
28
Figure 2.7: Erzincan near-field (a) Fault Normal ground motion (b) Fault normal Fourier amplitude spectrum (c)
Fault Parallel ground motion (d) Fault parallel Fourier amplitude spectrum
Erzincan is the capital of Erzincan Province in the eastern Anatolian region of Turkey.
On March 13
th
, 1992 a severe earthquake has struck Erzincan. Slip along the North Anatolian
Fault, which is a major active right-lateral moving strike-slip fault in northern Anatolia which
runs along the transform boundary between the Eurasian Plate and the Anatolian Plate is the
reason. The Erzincan basin lies on the intersection of this fault on its northern side. Fault normal
and fault parallel records of near-field considered for the study are produced in fig. 2.7. Erzincan
station which is 2.0 km from epicenter recorded the magnitude of the near-field ground motion
as MW 6.7. PGA's of fault normal and fault parallel are 0.432g and 0.457g respectively.
Predominant frequency ranges between 0.024-1.489 for fault normal and 0.024-1.904 for fault
parallel.
Erzincan - Erzincan Station
29
Figure 2.8: Cape Mendocino near-field (a) Fault Normal ground motion (b) Fault normal Fourier amplitude spectrum
(c) Fault Parallel ground motion (d) Fault parallel Fourier amplitude spectrum
Cape Mendocino located on the Lost Coast entirely within Humboldt County, California,
USA, is the westernmost point on the coast of California. Three earthquakes with epicenters
nearby at Petrolia and offshore west of Cape Mendocino, on 25, 26 April 1992 showed that the
Cascadia subduction zone is both capable of producing large earthquakes and generate
tsunamis. Fault normal and fault parallel records of near-field considered for the study are
produced in fig. 2.8. Petrolia station which is 8.5 km from epicenter recorded the magnitude of
the near-field ground motion as MW 7.1. PGA's of fault normal and fault parallel are 0.638g and
Landers - Petrolia Station
30
Figure 2.9: Lander's near-field (a) Fault Normal ground motion (b) Fault normal Fourier amplitude spectrum (c)
Fault Parallel ground motion (d) Fault parallel Fourier amplitude spectrum
Landers is a small town in California, USA. On 28 June 1992 five separate fault segments
in sequence: Johnson Valley, Landers, Homestead Valley, Emerson, and Camp Rock experienced
main shock rupture generating magnitude MW 7.3 earthquake. The Landers quake is also
sometimes referred to as "the Landers sequence." The mechanism by which Landers ruptured is
called "cascading rupture" whereby the rupture of one fault triggers the rupture of a
neighboring fault. Fault-normal and fault parallel records of near-field considered for the study
are produced in fig. 2.9. The 7.3 MW ground motion recorded at Petrolia station is 1.1 km from
epicenter. PGA's of fault normal and fault parallel are 0.713g and 0.799g respectively.
Predominant frequency ranges between 0.015-17.71 for fault normal and 1.373-17.50 for fault
parallel.
Landers - Petrolia Station
31
Figure 2.10: Northridge - Rinaldi near-field (a) Fault Normal ground motion (b) Fault normal Fourier amplitude
Northridge is a community located in the San Fernando Valley region of the city of Los
Angeles, California, USA. On 17 January 1994 an earthquake of magnitude MW 6.7 has struck this
place. Despite the area's proximity to the San Andreas Fault, the Northridge earthquake did not
occur along this fault, however, on the previously undiscovered Northridge blind thrust fault
(also known as Pico thrust fault). The 6.7 MW ground motion recorded at Rinaldi station is 7.5
km from epicenter. Fault normal and fault parallel records of near-field considered for the study
are produced in fig. 2.10. PGA's of fault normal and fault parallel are 0.89g and 0.389g
respectively. Predominant frequency ranges between 0.049-1.709 for fault normal and 0.049-
4.932 for fault parallel.
Northridge Rinaldi Station
32
Figure 2.11: Northridge - Olive View near-field (a) Fault Normal ground motion (b) Fault normal Fourier amplitude
Another near-field record of the 6.7 MW Northridge ground motion recorded at Olive
View station is 6.4 km from epicenter. Fault normal and fault parallel records of near-field
considered for the study are produced in fig. 2.11. PGA's of fault normal and fault parallel are
0.732g and 0.595g respectively. Predominant frequency ranges between 0.012-3.284 for fault
normal and 0.732-3.271 for fault parallel.
Northridge Olive ViewStation
33
Figure 2.12: Kobe near-field (a) Fault Normal ground motion (b) Fault normal Fourier amplitude spectrum (c) Fault
Parallel ground motion (d) Fault parallel Fourier amplitude spectrum
Kobe is the 5th largest city in Japan and is the capital city of Hyogo Prefecture on the
southern side of the mainland of Honshu. On 17 January 1995 an earthquake of magnitude MW
6.9 has struck this city. The great Hanshin earthquake (also Kobe earthquake) began north of
the island of Awaji, which lies just south of Kobe. It spread toward the southwest along the
Nojima fault on Awaji and toward the northeast along the Suma and Suwayama faults, which
run through the center of Kobe. Observations of deformations in these faults suggest that the
area was subjected to east-west compression, which is consistent with previously known
crustal movements. Like other earthquakes recorded in western Japan between 1891 and 1948,
the 1995 earthquake had a strike-slip mechanism that accommodated east-west shortening of
the Eurasian plate due to its collision with the North American plate in central Honshu. The 6.9
MW ground motion recorded at Kobe station is 3.4 km from epicenter. Fault normal and fault
parallel records of near-field considered for the study are produced in fig. 2.12. PGA's of fault
normal and fault parallel are 1.088g and 0.575g respectively. Predominant frequency ranges
between 0.269-2.625 for fault normal and 0.635-3.186 for fault parallel.
Kobe Kobe Station
34
Figure 2.13: Kobe - Takatori near-field (a) Fault Normal ground motion (b) Fault normal Fourier amplitude
Another near-field record of the 6.9 MW Kobe ground motion recorded at Takatori
station is 4.3 km from epicenter. Fault normal and fault parallel records of near-field considered
for the study are produced in fig. 2.13. PGA's of fault normal and fault parallel are 0.786g and
Kobe Takatori Station
35
2.5 Far-Field ground motions and their Fourier amplitude spectrums
Figure 2.14: Tabas-Sedeh far-field (a) Transverse ground motion (b) Transverse Fourier amplitude spectrum (c)
Longitudinal ground motion (d) Longitudinal Fourier amplitude spectrum
Tabas event recorded far-field at Sedeh station at an epicentral distance of 177.9 km had
a magnitude of MW 7.35, with transverse and longitudinal PGA's as 0.027g and 0.0262g
respectively. Predominant frequency ranges between 0.024-1.147 for transverse and 0.024-
4.785 for longitudinal. The amplitude of acceleration varied with high rate between fault normal
of near-field & transverse of far-field and between fault parallel of near-field and longitudinal of
far-field. The Fourier amplitude also varied with high rate. Maximum Fourier amplitude of near-
field fault normal is 0.715 where as maximum Fourier amplitude of far-field transverse is 0.028.
A conclusion could be drawn from it that even though the predominant frequency range is more
and matches with the frequency of dam considered, it cannot show the effect as the amplitude of
far-field is very less. Transverse and longitudinal records of far-field considered for the study
are produced in fig. 2.14.
Tabas Sedeh Station
36
Figure 2.15: Loma Prieta - Hollister City Hall far-field (a) Transverse ground motion (b) Transverse Fourier
amplitude spectrum (c) Longitudinal ground motion (d) Longitudinal Fourier amplitude spectrum
Loma Prieta event recorded far-field at Hollister City Hall station at an epicentral
distance of 47.9 km had a magnitude of MW 6.93, with transverse and longitudinal PGA's as
0.215g and 0.247g respectively. Predominant frequency ranges between 0.024-1.464 for
transverse and 0.024-1.36 for longitudinal. Transverse and longitudinal records of far-field
considered for the study are produced in fig.2.15.
Loma Prieta Hollister City Hall Station
37
Figure 2.16: Loma Prieta - Sunnyvale far-field (a) Transverse ground motion (b) Transverse Fourier amplitude
spectrum (c) Longitudinal ground motion (d) Longitudinal Fourier amplitude spectrum
Another Loma Prieta event considered for the study recorded far-field at Sunnyvale
station at an epicentral distance of 42.13 km had a magnitude of MW 6.93, with transverse and
longitudinal PGA's as 0.209g and 0.207g respectively. Predominant frequency ranges between
0.024-1.806 for transverse and 0.024-1.831 for longitudinal. Transverse and longitudinal
records of far-field considered for the study are produced in fig. 2.16.
Loma Prieta Sunnyvale Station
38
Figure 2.17: Erzincan far-field (a) Transverse ground motion (b) Transverse Fourier amplitude spectrum (c)
Erzincan event considered for the study recorded far-field at Erzincan station at an
epicentral distance of 8.97 km had a magnitude of MW 6.69, with transverse and longitudinal
PGA's as 0.178g and 0.154g respectively. Predominant frequency ranges between 0.024-1.465
for transverse and 0.024-1.806 for longitudinal. Transverse and longitudinal records of far-field
considered for the study are produced in fig.2.17.
Erzincan - Erzincan Station
39
Figure 2.18: Cape Mendocino far-field (a) Transverse ground motion (b) Transverse Fourier amplitude spectrum (c)
Cape Mendocino event considered for the study recorded far-field at Petrolia station at
an epicentral distance of 53.34 km had a magnitude of MW7.01, with transverse and longitudinal
considered for the study are produced in fig. 2.18.
Cape Mendocino - Petrolia Station
40
Figure 2.19: Landers far-field (a) Transverse ground motion (b) Transverse Fourier amplitude spectrum (c)
Landers event considered for the study recorded far-field at Anaheim station at an
PGA's as 0.035g and 0.047g respectively. Predominant frequency ranges between 0.012-2.00 for
Landers AnaheimStation
41
Figure 2.20: Northridge-Anacapa Island far-field (a) Transverse ground motion (b) Transverse Fourier amplitude
Northridge event considered for the study recorded far-field at Anacapa Island station at
an epicentral distance of 77.39 kmhad a magnitude of MW6.69, with transverse and longitudinal
Northridge Anacapa Station
42
Figure 2.21: Northridge-Anaheimfar-field (a) Transverse ground motion (b) Transverse Fourier amplitude
Northridge event considered for the study recorded far-field at Anaheim station at an
Northridge - AnaheimStation
43
Figure 2.22: Kobe-99999HIK far-field (a) Transverse ground motion (b) Transverse Fourier amplitude spectrum (c)
Kobe event considered for the study recorded far-field at 99999HIK station at an
PGA's as 0.148g and 0.147g respectively. Predominant frequency ranges between 0.50-2.795 for
Kobe 99999HIK Station
44
Figure 2.23: Kobe-99999FUK far-field (a) Transverse ground motion (b) Transverse Fourier amplitude spectrum (c)
Kobe event considered for the study recorded far-field at 99999FUK station at an
Kobe 99999FUK Station
45
2.6 Numerical Method
2.6.1 Introduction
Numerical methods for the analysis of structures can be broadly classified in to two. The
first one is based on continuum mechanism. Finite Element Method (FEM) [Jr. William Weaver,
James M. Gere, 1966] is one such example. However, it cannot perform the analysis up to
collapse because of limitations that exist in representation of cracks and separation distance
between elements. FEM can answer only one question will the structure fail or not? it cant tell
how the structure collapse
On the other hand, second category of numerical methods is based on discrete element
methods, like Extended Discrete Element Method [Williams J.R, Hocking G, and Mustoe G.G.W,
1985; A.A. Balkema, Rotterdam, 1985] for nonlinear analysis of structures. This method can
track the behavior from zero loading to total collapse of structure. However, this method is less
accurate than FEM in small deformation range. So this can answer only the second question
how does the structure collapse?
To follow total structural behavior from small deformation range to complete collapse, a
unique, efficient and accurate technique is required. Tagel Din Hatem (1988) gave a new
method of analyzing the structural behavior from zero loading, crack initiation & propagation,
separation of structural members till the total collapse with reliable accuracy, and with
relatively simple material models. The method is now known as Applied Element Method
(AEM) and is widely in usage.
Applied element method is a discrete method in which the elements are connected by
pair of normal and shear springs which are distributed around the element edges is shown in
fig. 2.24. Stresses and deformations of each and every element are represented by these shear
and normal springs. The motion of element is rigid body motion and the internal deformations
of the element are taken by springs only. Elements can be connected by any number of springs.
However, for improving the accuracy, increasing number of elements give better results. The
general stiffness matrix components corresponding to each degree of freedom are determined
by assuming unit displacement and the forces are at the centroid of each element. The element
stiffness matrix size is 6x6. The stiffness matrix components diagram is shown in fig. 2.25 and
the first quarter portion of stiffness matrix is shown in fig. 2.26. Finally the global stiffness
matrix is generated by summing up all the local stiffness matrices for each element.
46
2.6.2 Mathematical formulation
The governing dynamic equation for a structure is given in equation 2.2
g
u M t f Ku u C u M A = + + ) ( (2.2)
Where [M] is mass matrix; [C] is damping matrix; [K] is nonlinear stiffness matrix; f(t) is
incremental applied load vector U and its derivatives are the incremental displacement,
velocity and acceleration vectors respectively. The above equation is used in AEM and is solved
numerically using Newmark's beta method.
Figure 2.24: Element components for formulating stiffness matrix [Kimuro Meguro and Hatem 2001]
Figure 2.25: Quarter portion of stiffness matrix
For mass matrix the elemental mass and mass moment of inertia are assumed lumped at the
element centroid so that it will act as continuous system. The elemental mass matrix in case of
square shaped elements is given in equation 2.3
(
(
(
=
(
(
(
0 . 6 /
4
2
2
3
2
1
t D
t D
t D
M
M
M
(2.3)
Where D is the element size; t is element thickness and is the density of material. From the
above equation it is noticed that [M1] and [M2] are the element masses and [M3] is the mass
47
moment of inertia about centroid of the element. The mass matrix is a diagonal matrix. If
element size is small response of the structure is very near to the continuous/distributed mass
system. If damping is present, response of the structure will get reduced. Damping matrix is
calculated from the rst mode as follows:
n
M C 2 = (2.4)
Where is damping ratio and n is rst natural frequency of the structure. The general equation
for free vibration without damping is:
0 = + Ku u M (2.5)
For a non-trivial solution, determinant of the above matrix must be equal to zero and solution of
determinant of the matrix gives natural frequencies of the structure.
2.6.3 Element size
Element size in a structure is one important parameter to be considered in any
numerical modeling. Large element size decreases structures displacement and leads to
increasing stiffness and failure load of the structure. For any numerical analysis three important
requirements (convergence, stability and accuracy) are necessary.
1. Convergence - As element size decreases, numerical solution should close to the
exact/theoretical solution.
2. Stability - The numerical solution should be stable in the presence of numerical roundo
errors.
3. Accuracy - The numerical procedure should provide results that are close to the exact
solution.
For the study, the element size of the models is fixed as 1m x 1m after few parametric
studies. With the limitations in the AEM, this element size has been fixed. Limitations in AEM are
described in section 2.6.6.
2.6.4 Material modeling
The material model used in this analysis is Maekawa compression model [Tagel-
Din Hatem, 1998]. In the model, initial Young's modulus, fracture parameter, representing the
extent of the internal damage of concrete, and the compressive plastic strain are introduced to
define the envelope for compressive stresses and compressive strains, so that unloading and
reloading can be conveniently described. Also in the model, the tangent modulus is calculated
according to the strain at the spring location. After peak stresses, spring stiffness is assumed as
a minimum value to avoid negative stiffness. This leads to the difference between calculated
stresses and stress corresponding to the spring strain. The residual stresses thus obtained are
48
redistributed by applying the redistributed force values in the reverse direction. For concrete
springs subjected to tension, spring stiffness is assumed as the initial stiffness till it reaches the
crack point. After cracking, stiffness of springs subjected to tension are assumed zero. For
reinforcement, bi-linear stress strain relation is assumed. After yield of reinforcement, steel
spring stiffness is assumed as 0.01 of the initial stiffness. For cracking criteria [Tagel-din-Hatem,
1998], principal stress based on failure criteria is adopted. The models for concrete, both in
compression and tension and the reinforcement bi-linear model are shown in fig. 2.26 (a) & (b).
Figure 2.26: Material models for concrete and steel (a) Tension and compression concrete Maekawa model (b) Bi-
linear stress strain relation model for steel reinforcement (Kimuro Meguro and Hatem 2001)
2.6.5 Boundary conditions
In actual field condition, soil does not have boundary. As, modeling of whole earth is not
possible, we have considered a portion of it for our study. For this small portion, appropriate
boundary conditions should be applied. Wave generated from source in field can travel infinite
distance, until its strength becomes negligible. To restrict the generated waves from source, a
boundary condition should be used where, the waves will be absorbed and radiation of waves is
minimized. Bottom of the models (S1 and S2) are modeled as fixed, and sides of the base in
model S2 is modeled as absorbing boundaries. . The non-reflecting boundaries approximate the
behaviour of the waves to propagate past the edges of domain. Outgoing waves are absorbed by
placing dampers at the boundaries. This boundary effect is modeled in AEM [Ahmed, 2011] for
the current study.
2.6.6 Effect of number of connecting springs
The number of connecting springs between elements is one of the important factors that
should be taken into account. Obviously, in nonlinear analysis, increasing the number of
connecting springs between elements leads to better results of crack propagation. It is assumed
49
that 2n springs are connecting two elements together. Each spring represent a distance of
(D/2n). In horizontal and vertical degrees of freedom case, the number of connecting springs
has no effect on the element stiffness as decreasing the number of connecting springs leads to
increasing of area represented by each spring. Finally, the total area becomes the same as that
represented by one whole element. Rotation of an element if mainly resisted by shear springs
together with normal springs.
In case of two connecting springs, the numerically obtained rotational stiffness is
smaller than the theoretical value by 25% which is quite large. However, this error reduces to
less than 1% if the number of connecting springs is 10 or more [Tagel-din-Hatem, 1998]. This
effect is dominant if the element size is small because the relative rotation between adjacent
elements becomes small.
2.6.7 Modeling limitations
AEM is a discrete element method and it has got some limitations. Only two models
(steel and concrete) and two materials (soil and concrete) are considered. Any other material
can be used by giving its properties. However, material models are only concrete and steel. For
the current study, reservoir condition is not considered as water has not been modeled in AEM.
Therefore hydrostatic and hydrodynamic pressures are not considered. Also two loads cannot
be given simultaneously on a structure. Therefore, when ground motions are given as input, the
structure is subjected to only ground motion, which will not happen in real time. Several forces
like hydrostatic, hydrodynamic, silt load, wave pressure, uplift etc. cannot be given
simultaneously and have to be studied individually. Work is being done to incorporate these
limitations.
2.8 Summary
In this chapter, a concrete gravity dam based on its importance is selected. Later two
models, concrete gravity dam with foundation and concrete gravity dam with foundation and
base are modeled using applied element method. Material properties considered for modeling
S1 and S2 are also illustrated. Later ground motions considered for the study, their origin and
type of fault are explained along with their Fourier amplitude spectra. Duration and
predominant frequency range of the ground motion records are also tabulated. Further
introduction, mathematical formulation, material modeling and boundary conditions of AEM are
discussed.
In the next chapter, linear earthquake response of structures S1 and S2 subjected to
components of selected near-field and far-field ground motion records is studied.
50
Chapter 3
Linear Earthquake Response of Concrete Gravity Dam
3.1 Introduction
The main focus of this chapter is to study the linear earthquake response of (i) a
concrete gravity dam with foundation (Structure 1 - S1) as one model and (ii) a concrete gravity
dam with foundation and base (Structure 2 - S2) as another model. The above two models are
subjected to fault normal and fault parallel components of near-field ground motions,
transverse and longitudinal components of far-field ground motions. Comparison among these
components are drawn for understanding the linear behavior of concrete gravity dam.
3.2 Eigen value analysis
Dynamic characteristics of any structure could be evaluated by its natural frequencies
and vibration mode shapes. This could be achieved through eigen value analysis. The analysis
deals with undamped free vibration of the structure and does not represent response due to any
loading, but yields the natural frequencies (eigen values) and corresponding vibration mode
shapes (eigen vectors) of the structure when there is no dissipation of energy due to damping.
The amplitude of the free vibration will depend on the initial conditions and in the absence of
damping, the vibration will continue without any decay. The solution of equation (3.1) for any
structure give the eigenvalues and their corresponding eigen vectors.
0 = + Ku u M
(3.1)
in which [M] and [K] are the global mass matrix and stiffness matrix of the dam and are obtained
by assembling the mass matrices and stiffness matrices of relevant elements of dam model, and
{u} is nodal displacement vector. In this study, a linear Applied Element Method code is used for
carrying out 2D eigen value analysis for the concrete gravity dam.
S1 has significant structural modes of peak frequencies 2.98, 7.78 and 15.26 Hz. Another
model (S2) has significant structural modes of peak frequencies 2.82, 6.97 and 12.97 Hz. Above
mentioned frequencies corresponds to the first, second and third modes of respective models.
Fourier analysis of accelerograms used to compute the displacements of the dam models
give the predominant frequency range of that particular ground motion. If the natural frequency
51
of the structure matches with the freqeuncy of the ground motion, the phenomenon of
resonance will occur causing failure to the structure. In the next section this study is discussed.
Figure 3.1: Eigen values of first ten mode shapes of S1
52
Figure 3.2(a): Eigen values of 1 - 5 mode shapes of S2
Mode 1
Frequency: 2.82 Hz
Time Period: 0.355 Sec
Mode 2
Frequency: 6.97 Hz
Mode 3
Frequency: 12.97 Hz
Mode 4
Frequency: 19.34 Hz
Mode 5
Frequency: 26.05 Hz
53
Figure 3.2(b): Eigen values of 6 - 10 mode shapes of S2
Mode 6
Frequency: 32.62 Hz
Mode 7
Frequency: 38.60 Hz
Mode 8
Frequency: 44.13 Hz
Mode 9
Frequency: 49.76 Hz
Mode 10
Frequency: 55.41 Hz
54
Table 3.1: Eigen values of first ten mode shapes of S1
Modes Frequency (Hz) Time Period (sec)
Mode 1 2.98 0.335
Mode 2 7.78 0.128
Mode 3 15.26 0.065
Mode 4 23.54 0.042
Mode 5 32.03 0.031
Mode 6 39.0 0.025
Mode 7 47.11 0.021
Mode 8 54.70 0.018
Mode 9 62.69 0.016
Mode 10 69.70 0.014
Table 3.2: Eigen values of first ten mode shapes of S2
Modes Frequency (Hz) Time Period (sec)
Mode 1 2.82 0.355
Mode 2 6.97 0.143
Mode 3 12.97 0.077
Mode 4 19.34 0.051
Mode 5 26.05 0.038
Mode 6 32.62 0.030
Mode 7 38.60 0.026
Mode 8 44.13 0.022
Mode 9 49.76 0.020
Mode 10 55.41 0.018
55
3.3 Near field earthquakes
3.3.1 Response of S1 subjected to fault normal component
The first ten natural frequencies and mode shapes of S1 are given in table 3.1 and
represented in fig. 3.1. PGA's, predominant frequency ranges and displacements of S1 at
different cross sections for 5 near-field fault normal ground motions are given in table 3.3.
Arbitrary element numbers (535, 2285, 3496, and 4130) are considered at different cross
sections to study the variation of response of the structure from base to the crest.
Table 3.3: Response of S1 subjected to 5 near-field fault normal ground motions
Earthquake
PGA
(g)
Predominent
Frequency
Range
(Hz)
Response of
element no
535 @ 6m ht
from base
(mm)
Response of
element no
2287 @
36m ht
from base
(mm)
Response of
element no
3496 @
72m ht
from base
(mm)
Response of
element no
4130 @
108m ht
from base
(mm)
Tabas
0.9
0.012-5.444
0.056 0.713 3.34 9.79
Loma Prieta LG 0.718 0.024-1.099
0.053 0.798 3.33 8.59
Landers 0.713 0.015-17.71
0.039 0.556 2.29 6.21
Northridge R 0.89 0.049-1.709
0.059 1.124 5.44 15.04
Kobe 1.088 0.269-2.625
0.060 1.054 4.84 13.05
Fourier amplitude spectrum of each ground motion considered is smoothened to 10
th
degree polynomial to obtain predominant frequency range and is given from fig. 2.4 to fig. 2.23.
Tabas near-field fault normal (NF-FN) ground motion has a predominant frequency range of
0.012-5.444 Hz. This matches with the frequency of the S1 in its first mode (2.98Hz).
Displacements at different sections from base to the crest clearly show the displacement at the
crest remains maximum for all ground motions. The motion at the base does not have
frequencies which are present in motion at the crest and the peak displacements at the base are
much smaller as compared to those at the crest. Displacement at the crest of S1 subjected to
Tabas NF-FN ground motion is 9.79 mm. Loma Prieta near-field fault normal ground motion
recorded at Los Gatos station has a predominant frequency range of 0.024-1.099 Hz. The
structure remained stable as the frequency of S1 does not match with the frequency of the
ground motion. Displacement at the crest of S1 is 8.59 mm. Landers near-field fault normal
ground motion recorded at Lucerne Valley Station has a predominant frequency range of 0.015-
17.71 Hz. Even though S1 has first 3 modal frequencies (2.98, 7.78, and 15.26) in predominant
frequency range, peak amplitude of Fourier spectrum is very less. Therefore, ground motion had
56
no serious effect on the structure. The displacement at the crest of S1 is 6.21 mm. Northridge
near-field fault normal ground motion recorded at Rinaldi station has a predominant frequency
range of 0.049-1.709 Hz. Even though S1 does not lie in the predominant frequency range,
structure resulted to higher displacements compared to other ground motions considered in the
study. This is because Northridge earthquake has occurred due to previously undiscovered
inactive blind thrust fault, which became active, releasing high energies in short duration.
Whereas, Loma Prieta earthquake has caused due to strike-slip faulting with low amplitudes.
The energy released in the thrust fault will be more than strike-slip fault. Accumulation of
energy observed on the hanging wall side has characteristics that disturb the structure with
higher displacements. Displacement at the crest of the dam subjected to this ground motion is
15.04 mm. Kobe near-field fault normal ground motion has a predominant frequency range of
0.269-2.625, which does not match with the natural frequency of the dam. Even though the fault
that caused Kobe earthquake is a strike-slip fault, PGA and peak amplitude of Fourier spectrum
are high. The energy released in the ground motion is high, thus causing higher displacements at
the crest compared to other ground motions considered in the study. The displacement at the
crest of the dam is 13.05 mm.
3.3.2 Response of S1 subjected to fault parallel component
Similar to the displacements observed at different cross sections of the dam subjected to
near-field fault normal ground motions, results are drawn for near-field fault parallel ground
motions also and are produced in table 3.4.
Table 3.4: Response of S1 subjected to 5 near-field fault parallel ground motions
Earthquake
PGA
(g)
Predominent
Frequency
Range
(Hz)
Response of
element no
535 @ 6m ht
from base
(mm)
Response of
element no
2287 @
36m ht
from base
(mm)
Response of
element no
3496 @
72m ht
from base
(mm)
Response of
element no
4130 @
108m ht
from base
(mm)
Tabas
0.978
0.012-6.519
0.049 0.729 3.19 9.41
Loma Prieta LG 0.458 0.024-1.172
0.033 0.544 2.36 6.49
Landers 0.799 1.373-17.50
0.041 0.572 2.51 7.09
Northridge R 0.389 0.049-4.932
0.044 0.830 3.94 10.92
Kobe 0.575 0.635-3.186
0.055 1.018 4.75 13.22
57
Natural frequency of S1 matches with the predominant frequency range of Tabas near-
field fault parallel ground motion. The PGA of fault parallel (0.978g) is higher than fault normal
(0.9g). However, the displacement is less since it is a thrust fault, where the permanent
displacement is not observed in fault parallel direction. Displacement at the crest of the dam
subjected to fault normal ground motion is 9.41 mm. Loma Prieta - Los Gatos near-field fault
parallel PGA observed is lower (0.458g) compared to fault normal PGA (0.718). The structure
also does not lie in the predominant frequency range. The amplitude of the Fourier spectrum is
also low and the displacement observed is thus less than fault normal. The displacement at the
crest of the dam is 6.49 mm. Landers near-field fault parallel ground motion has a predominant
frequency range which matches with the frequency of the structure in its first three modes. Also
compared to the fault normal, PGA of fault parallel is higher and the amplitude of the Fourier
spectrum is also higher. The displacement observed is thus higher. The displacement at the
crest of the dam is 7.09 mm. Similarly the Northridge - Rinaldi near-field fault parallel ground
motion also has less PGA and the peak amplitude of the Fourier spectrum is also low. Even
though the structure lies in the frequency of the ground motion, lesser amplitude could not
produce higher displacement. The displacement at the crest of the dam is 10.92 mm. Kobe near-
field fault parallel ground motion has a predominant frequency range which matches with the
frequency of the structure in its first mode. Therefore, even though the amplitude is less, the
displacement observed was little higher. Displacement at the crest of the dam is 13.22 mm.
Table 3.5: Comparison of response of S1 subjected to 5 near-field fault normal and fault parallel ground motions
Maximum linear displacement response at the crest of S1 subjected to near-field ground
motions
Fault Normal
(Displacement at the crest in
mm)
Fault Parallel
(Displacement at the crest in mm)
Tabas
9.79 9.41
Loma Prieta LG
8.59 6.49
Landers
6.21 7.09
Northridge R
15.04 10.92
Kobe
13.05 13.22
Depending upon the parameters like location and orientation of the fault rupture plane,
and the location of the hypocenter on the rupture plane, forward rupture directivity effects are
observed mostly in fault normal directions. This directivity effect is observed stronger in thrust
fault where both the directivity and fling effects are in fault normal direction. This is not seen in
strike-slip faults where the directivity effect is observed in fault normal direction and
permanent displacement is observed in fault parallel direction. Even though directivity is
observed in fault normal, because of permanent displacement oriented in fault parallel
58
direction, effect of directivity will be less compared to thrust fault, when compared on same
parameters. When rupture directivity pulses of smaller earthquakes (Loma Prieta, Northridge,
Kobe) with magnitude range of 6.7 to 7.0 are compared with large earthquakes (Tabas,
Landers) with magnitude range of 7.2 to 7.6, the narrow band nature of these pulses caused
their elastic response spectra to have peaks. The fault normal spectra of smaller earthquakes
are observed to be different from larger earthquakes. Their spectra have larger amplitudes in
the intermediate period range of 0.5 to 2.5 sec. Whereas, for large earthquakes response spectra
have bump at around 4 sec. Strong motion recordings of the recent large earthquakes confirm
that the near fault pulse is a narrow band pulse whose period increases with magnitude
(Somerville et al. - 1997). Loma Prieta, Northridge and Kobe which are smaller in magnitude
which have peak narrow band pulses with short periods at the beginning of the record carried
most of the energy in them. Whereas Tabas and Landers, which are large earthquakes have peak
narrow band pulses with long periods at intermediate range with less seismic energy. Adding to
these earthquakes their orientation and kind of fault has produced displacements as given in
Table 3.3. Compared with fault normal, fault parallel components behave as expected with much
smaller at long periods.
In a thrust fault, even if PGA in fault parallel motion is higher or lower when compared
with fault normal motion, displacement response of the structure observed are less in fault
parallel. This is observed from Table 3.5. Tabas and Northridge ground motions are cause
because of thrust faulting. Tabas ground motion has PGA higher in fault parallel direction.
However, the displacement observed is low compared to fault normal. Similarly Northridge
ground motion which is caused by thrust fault has low PGA in fault parallel motion and
displacement observed is also low. In strike slip fault when the frequency of structure matches
with the predominant frequency range of the ground motion, the displacement response
observed is different with fault parallel motion as input. Frequency of Landers and Kobe
earthquakes matches with the frequency of the structure. Fault parallel motion of Landers
earthquake has PGA higher than fault normal motion and the displacement observed is higher.
However, fault parallel motion of Kobe has less PGA than fault normal motion; even then the
displacement observed is higher. Thus from the observed results, strike slip faults will have
higher displacement response in fault parallel direction irrespective of PGA compared with fault
normal, if the frequencies matches between structure and ground motion.
59
Figure 3.3: Linear displacement response of S1, subjected to 5 near-field fault normal ground motions
60
Figure 3.4: Linear displacement response of S1, subjected to 5 near-field fault parallel ground motions
61
3.3.3 Response of S2 subjected to fault normal component
The displacements observed at different cross sections of S2 subjected to near-field fault
normal ground motions are produced in Table 3.6.
Table 3.6: Response of S2 subjected to 5 near-field fault normal ground motions
Earthquake
PGA
(g)
Predominent
Frequency
Range
(Hz)
Response of
element no
40035 @
51m ht from
base (mm)
Response of
element no
42013 @
86m ht
from base
(mm)
Response of
element no
43095 @
122m ht
from base
(mm)
Response of
element no
43630 @
153m ht
from base
(mm)
Tabas
0.9
0.012-5.444
0.443 1.888 5.52 11.65
Loma Prieta LG 0.718 0.024-1.099
0.312 1.350 4.64 10.14
Landers 0.713 0.015-17.71
0.389 1.212 3.32 6.96
Northridge R 0.89 0.049-1.709
0.444 2.849 9.79 20.96
Kobe 1.088 0.269-2.625
0.440 2.531 8.72 18.33
Tabas near-field fault normal ground motion has a predominant frequency range which
matches with the frequency of S2 in its first mode (2.82 Hz). Displacement at the crest of the
structure is 11.65 mm. Loma Prieta near-field fault normal ground motion recorded at Los Gatos
station has a predominant frequency range that does not match with the frequency of the structure.
Displacement at the crest of the structure is 10.14 mm. Landers near-field fault normal ground
motion recorded at Lucerne Valley Station has a predominant frequency range which matches with
the first 3 modes of structure. However, the amplitude of Fourier spectrum is very less, having no
serious effect on the structure. The displacement at the crest of the dam is 6.96 mm. Northridge
near-field fault normal ground motion recorded at Rinaldi station has a predominant frequency
range that does not match with the frequency of the structure. However, the structure was
subjected to higher displacements compared to other ground motions considered for the study.
This is because Landers and Loma Prieta earthquakes are caused due to the strike-slip faulting
mechanism. Whereas, Northridge earthquake has occurred due to previously undiscovered inactive
blind thrust fault. Displacement at the crest of the dam subjected to this ground motion is 20.96
mm. Predominant frequency range of Kobe near-field fault normal ground motion does not match
with the natural frequency of the dam. Even though the fault that caused Kobe earthquake is a
62
strike-slip fault, PGA and peak amplitude of Fourier spectrum are high. The energy released in the
ground motion is high, thus causing higher displacements at the crest compared to other ground
motions considered in the study. Displacement at the crest of the dam is 18.33 mm.
3.3.4 Response of S2 subjected to fault parallel component
The displacements observed at different cross sections of S2 subjected to near-field fault
parallel ground motions are produced in Table 3.7.
Table 3.7: Response of S2 subjected to 5 near-field fault parallel ground motions
Earthquake
PGA
(g)
Predominent
Frequency
Range
(Hz)
Response of
element no
40035 @
51m ht from
base (mm)
Response of
element no
42013 @
86m ht
frombase
(mm)
Response of
element no
43095 @
122m ht
from base
(mm)
Response of
element no
43630 @
153m ht
from base
(mm)
Tabas
0.978
0.012-6.519
0.361 1.291 4.33 9.79
Loma Prieta LG 0.458 0.024-1.172
0.201 1.077 3.58 7.84
Landers 0.799 1.373-17.50
0.429 1.607 4.70 10.07
Northridge R 0.389 0.049-4.932
0.354 1.978 6.67 15.13
Kobe 0.575 0.635-3.186
0.285 1.642 5.64 11.89
Natural frequency of S2 matches with the predominant frequency range of Tabas near-field
fault parallel ground motion. The PGA of fault parallel (0.978g) is higher than fault normal (0.9g).
However, the displacement is less since it is a thrust fault, where the permanent displacement is not
observed in fault parallel direction. Displacement at the crest of the dam subjected to fault normal
ground motion is 9.79 mm. Loma Prieta - Los Gatos near-field fault parallel PGA observed is lower
(0.458g) compared to fault normal PGA (0.718). The structure also does not lie in the predominant
frequency range. The amplitude of the Fourier spectrum is also low and the displacement observed
is thus less than fault normal. The displacement at the crest of the dam is 7.84 mm. Landers near-
field fault parallel ground motion has a predominant frequency range which matches with the
frequency of the structure in its first three modes. Also compared to the fault normal, PGA of fault
parallel is higher and the amplitude of the Fourier spectrum is also higher. The displacement
observed is thus higher. The displacement at the crest of the dam is 10.07 mm. Similarly the
63
Northridge - Rinaldi near-field fault parallel ground motion also has less PGA and the peak
amplitude of the Fourier spectrum is also low. Even though the structure lies in the frequency of the
ground motion, lesser amplitude could not produce higher displacement. The displacement at the
crest of the dam is 15.13 mm. Kobe near-field fault parallel ground motion has a predominant
frequency range which matches with the frequency of the structure in its first mode. The
displacement at the crest of S2 is 11.89 mm.
Table 3.8: Comparison of response of S2 subjected to 5 near-field fault normal and fault parallel ground motions
Maximum linear displacement response at the crest of S2 subjected to near-field ground motions
Fault Normal
Fault Parallel
Tabas
11.65 9.79
Loma Prieta LG
10.14 7.84
Landers
6.96 10.07
Northridge R
20.96 15.13
Kobe
18.33 11.89
Structure 1 (S1) and Structure 2 (S2) are different in geometry. Addition of hard rock as
base for S2 gave displacements results which are different from S1. Irrespective of higher or lower
PGA's between fault normal and fault parallel, in both thrust and strike-slip fault observed results
gave fault parallel displacements less than fault normal displacements. Since Landers earthquake
mechanism involved rupture in 5 faults (cascading rupture) observed displacements are different
in fault normal and fault parallel. Also the amplitude in the Fourier spectrum have peaks in the
frequency range where the structure is matched in its 2nd mode, the displacements are higher.
64
Figure 3.5: Linear displacement response of S2, subjected to 5 near-field fault normal ground motions
65
Figure 3.6: Linear displacement response of S2, subjected to 5 near-field fault parallel ground motions
66
3.4 Far-field earthquakes
3.4.1 Response of S1 subjected to transverse component
The displacements observed at different cross sections of S1 subjected to far-field
transverse ground motions are produced in Table 3.9.
Table 3.9: Response of S1 subjected to 5 far-field transverse ground motions
Earthquake
PGA
(g)
Predominent
Frequency
Range
(Hz)
Response of
element no
535 @ 6m
ht from
base (mm)
Response of
element no
2287 @
36m ht
from base
(mm)
Response of
element no
3496 @
72m ht
from base
(mm)
Response of
element no
4130 @
108m ht
from base
(mm)
Tabas
0.027
0.024-1.147
0.001 0.021 0.09 0.25
Loma PrietaSVL 0.209 0.024-1.806
0.013 0.211 0.94 2.97
Landers 0.0353 0.012-2.00
0.002 0.040 0.18 0.46
Northridge A 0.0659 0.073-4.956
0.006 0.147 0.77 2.17
Kobe 0.148 0.50-2.795
0.012 0.208 0.92 2.43
Tabas far-field transverse ground motion has a predominant frequency range which does
matches with the frequency of S1. The epicentral distance was 177.9 km where far-field ground
motion was recorded. This is the reason why PGA is very less (0.027g). Longer duration motion is
observed in far-field from which, it can be assumed that the recorded data at Sedah station might be
in backward rupture directivity. Displacement at the crest of the structure is 0.25 mm. Loma Prieta
far-field transverse ground motion recorded at Sunnyvale station has a predominant frequency
range that does not match with the frequency of the structure. Epicentral distance is 47.9 km and
the motion has a PGA of 0.209. The duration of the ground motion is also higher from near-field.
Displacement at the crest of the structure is 2.97 mm. Landers far-field transverse ground motion
recorded at Lucerne Valley Station has a predominant frequency range which does not match with
the frequency of structure. Similar to Tabas earthquake, Landers motion is recorded far-field has
higher epicentral distance with 146.11 km. PGA is very low with 0.0353g. The displacement at the
crest of the dam is 0.46 mm. Northridge far-field transverse ground motion recorded at Anaheim
station has a predominant frequency range that match with the frequency of the structure. The
67
epicentral distance and PGA are less when compared to other ground motions. However, as the
structure is in the range with the ground motion displacement at the crest of the dam was high with
2.17 mm. Predominant frequency range of Kobe far-field transverse ground motion does not match
with the natural frequency of the dam. However, the PGA is high when compared with other ground
motions. This gave displacement at the crest of the dam as 2.43 mm.
3.4.2 Response of S1 subjected to longitudinal component
longitudinal ground motions are produced in Table 3.10.
Table 3.10: Response of S1 subjected to 5 far-field longitudinal ground motions
Earthquake
PGA
(g)
Predominent
Frequency
Range
(Hz)
Response of
element no
535 @ 6m
ht from
base (mm)
Response of
element no
2287 @
36m ht
from base
(mm)
Response of
element no
3496 @
72m ht
from base
(mm)
Response of
element no
4130 @
108m ht
from base
(mm)
Tabas
0.0262
0.024-4.785
0.002 0.019 0.07 0.19
0.014 0.200 0.76 2.13
Landers 0.0471 0.012-2.197
0.004 0.067 0.29 0.79
Northridge A 0.0724
0.024-4.468 0.005 0.105 0.54 1.58
Kobe 0.147 0.732-3.027
0.009 0.137 0.56 1.47
Natural frequency of S1 matches with the predominant frequency range of Tabas far-field
longitudinal ground motion. The PGA of longitudinal is lower than transverse. As the epicentral
distance is more, the PGA is less and thus the displacement is also low. The displacement is less for
longitudinal motion since it is a thrust fault, where the permanent displacement is not observed in
this direction. Displacement at the crest of the dam subjected to longitudinal ground motion is 0.19
mm. Loma Prieta - Sunnyvale far-field longitudinal PGA observed is lower compared to transverse
PGA. However, when compared to other longitudinal ground motions the record has high PGA as it
is recorded at lower epicentral distance. Even though the structure does not lie in the predominant
frequency range the displacement observed at the crest of S2 is 2.13 mm which is higher than other
displacements caused to structure by other ground motions. Landers far-field longitudinal ground
68
motion has a predominant frequency range which does not match with the frequency of the
structure. PGA of longitudinal motion is higher than transverse motion and the amplitude of the
Fourier spectrum is also higher. The displacement observed is thus higher. The displacement at the
crest of the dam is 0.79 mm. Northridge - Anaheim far-field longitudinal ground motion has higher
PGA. And even though the structure lies in the frequency of the ground motion, it could not produce
higher displacement. The displacement at the crest of the dam is 1.58 mm. Kobe far-field
longitudinal ground motion has a predominant frequency range which matches with the frequency
of the structure in its first mode. However, as the characteristics of near-field fault parallel and far-
field longitudinal ground motions are different, the effect of higher displacements observed in near-
field fault parallel when compared to fault normal are not seen in far-field longitudinal. The
displacement observed is thus lower and is 1.47 mm.
Table 3.11: Response of S1 subjected to 5 far-field transverse and longitudinal ground motions
Maximum linear displacement response at the crest of S1 subjected to far-field ground motions
Transverse
Longitudinal
Tabas
0.25 0.19
Loma Prieta LG
2.97 2.13
Landers
0.46 0.79
Northridge R
2.17 1.58
Kobe
2.43 1.47
Except in the case of Landers earthquake, irrespective of higher or lower PGA's between
transverse and longitudinal ground motions, displacements caused in the structure when subjected
to longitudinal ground motions are less. As Landers event is an event caused by 5 faults at a time,
the site conditions and characteristics of ground motions were different which produced higher
displacements in longitudinal motion.
69
Figure 3.7: Linear displacement response of S1, subjected to 5 far-field transverse ground motions
70
Figure 3.8: Linear displacement response of S1, subjected to 5 far-field longitudinal ground motions
71
3.4.3 Response of S2 subjected to transverse component
The displacements observed at different cross sections of S1 subjected to far-field longitudinal
ground motions are produced in Table 3.12.
Table 3.12: Response of S2 subjected to 5 far-field transverse ground motions
Earthquake
PGA
(g)
Predominent
Frequency
Range
(Hz)
Response of
element no
40035 @
51m ht
from base
(mm)
Response of
element no
42013 @
86m ht
from base
(mm)
Response of
element no
43095 @
122m ht
from base
(mm)
Response of
element no
43630 @
153m ht
from base
(mm)
Tabas
0.027
0.024-1.147
0.010 0.049 0.14 0.32
0.129 0.725 2.17 4.86
Landers 0.0353 0.012-2.00
0.017 0.076 0.23 0.49
Northridge A 0.0659 0.073-4.956
0.047 0.293 1.13 2.50
Kobe 0.148 0.50-2.795
0.086 0.439 1.43 2.96
Tabas far-field transverse ground motion has a predominant frequency range which does
matches with the frequency of S2. The epicentral distance where the far-field ground motion
recorded is very high also the PGA is very less. Displacement at the crest of the structure is 0.32
mm. Loma Prieta far-field transverse ground motion recorded at Sunnyvale station has a
predominant frequency range that does not match with the frequency of the structure. Epicentral
distance is 47.9 km and the motion has a PGA of 0.209. Displacement at the crest of the structure is
4.86 mm. Landers far-field transverse ground motion recorded at Lucerne Valley Station has a
predominant frequency range which does not match with the frequency of structure. Similar to
Tabas earthquake, Landers motion is recorded far-field has higher epicentral distance with 146.11
km. PGA is very low with 0.0353g. The displacement at the crest of the dam is 0.49 mm. Northridge
far-field transverse ground motion recorded at Anaheim station has a predominant frequency
range that match with the frequency of the structure. The epicentral distance and PGA are less
when compared to other ground motions. As the structure is in the range with the ground motion,
displacement at the crest of the dam was 2.50 mm. Predominant frequency range of Kobe far-field
transverse ground motion does not match with the natural frequency of the dam. However, the PGA
72
is high when compared with other ground motions. This gave displacement at the crest of the dam
as 2.96 mm.
3.4.4 Response of S2 subjected to longitudinal component
longitudinal ground motions are produced in Table 3.13.
Table 3.13: Response of S2 subjected to 5 far-field longitudinal ground motions
Earthquake
PGA
(g)
Predominent
Frequency
Range
(Hz)
Response of
element no
40035 @
51m ht
from base
(mm)
Response of
element no
42013 @
86m ht
from base
(mm)
Response of
element no
43095 @
122m ht
from base
(mm)
Response of
element no
43630 @
153m ht
from base
(mm)
Tabas
0.0262
0.024-4.785
0.012 0.041 0.09 0.20
0.082 0.454 1.49 3.21
Landers 0.0471 0.012-2.197
0.023 0.127 0.45 0.96
Northridge A 0.0724
0.024-4.468 0.030 0.172 0.64 1.39
Kobe 0.147 0.732-3.027
0.063 0.293 1.02 2.19
Natural frequency of S2 matches with the predominant frequency range of Tabas far-field
longitudinal ground motion. The PGA of longitudinal is lower than transverse. As the epicentral
distance is more, the PGA is less and thus the displacement is also low. Displacement at the crest of
the dam subjected to longitudinal ground motion is 0.20 mm. Loma Prieta - Sunnyvale far-field
longitudinal PGA observed is lower compared to transverse PGA. However, when compared to
other longitudinal ground motions the record has high PGA as it is recorded at near epicentral
distance. Even though the structure does not lie in the predominant frequency range the
displacement observed at the crest of S2 is 3.21 mm which is higher than other displacements
caused to structure by other ground motions. Landers far-field longitudinal ground motion has a
predominant frequency range which does not match with the frequency of the structure. PGA of
longitudinal motion is higher than transverse motion and the amplitude of the Fourier spectrum is
also higher. The displacement observed is thus higher. The displacement at the crest of the dam is
0.96 mm. Northridge - Anaheim far-field longitudinal ground motion has higher PGA. Even though
73
the structure lies in the frequency of the ground motion, it could not produce higher displacement.
The displacement at the crest of the dam is 1.58 mm. Kobe far-field longitudinal ground motion has
a predominant frequency range which matches with the frequency of the structure in its first mode.
As explained in section 3.4.2 the displacement observed is lower and is 2.19 mm.
Table 3.14: Response of S2 subjected to 5 far-field transverse and longitudinal ground motions
Maximum linear displacement response at the crest of S2 subjected to far-field ground motions
Transverse
Longitudinal
Tabas
0.32 0.20
Loma Prieta LG
4.86 3.21
Landers
0.49 0.96
Northridge R
2.50 1.39
Kobe
2.96 2.19
As the geometric details of S1 and S2 are different, the results observed are different. Since
S2 has a base of hard rock with a depth of 50 m, amplification of acceleration, velocity and
displacement at the crest are higher and thus produced higher displacements in S2 than S1. The
results are higher in S2; however, the behaviour remained same. Except with Landers ground
motion, displacements at the crest in both S1 and S2 are similar with higher displacements
observed in transverse motion than in longitudinal motion.
74
Figure 3.9: Linear displacement response of S2, subjected to 5 far-field transverse ground motions
75
Figure 3.10: Linear displacement response of S2, subjected to 5 far-field longitudinal ground motions
76
3.5.1 Fault normal and transverse response of S1
With parameters like type of fault, magnitude, PGA, epicentral distance, frequency range
and two models (S1 and S2) taken into picture, the displacements are tabulated. Comparing
near-field fault normal and far-field transverse ground motions acting on S1, the displacements
on the crest of S1 are produced in Table 3.15. As PGA's are higher in near-field compared with
far-field, the results obtained can clearly show how a structure would behave if it undergoes
near-field earthquake effects.
Table 3.15: Fault normal and transverse response of S1 subjected to 5 near-field and far-field ground motions
Maximumlinear displacement response at the crest of S1 subjected to near-field fault normal
and far-field transverse ground motion
Near-Field - Fault Normal
Far-Field - transverse
Tabas
9.79 0.25
Loma Prieta LG
8.59 2.97
Landers
6.21 0.46
Northridge R
15.04 2.17
Kobe
13.05 2.43
3.5.2 Fault parallel and longitudinal response of S1
Even though most of the time directivity effects are higher in fault normal, sometimes
PGA is also observed to be higher in fault parallel. However, the effects are mostly conventional
giving lesser displacements in fault parallel direction. In Table 3.16 comparison between
displacements in the structure caused by near-field fault parallel and far-field longitudinal
motions are tabulated. As PGA's in near-field fault parallel are higher the effect on the structure
is more compared with far-field longitudinal. As the ground motions in far-field have different
range of epicentral distance and PGA's proper comparison could not be drawn. However, as
PGA's are much higher in near-field the displacements observed are higher.
Table 3.16: Fault parallel and longitudinal response of S1 subjected to 5 near-field and far-field ground motions
Maximum linear displacement response at the crest of S1 subjected to near-field fault parallel
and far-field longitudinal ground motion
Near-Field - Fault Parallel
(Displacement at the crest in
mm)
Far-Field - Longitudinal
Tabas
9.41 0.19
Loma Prieta LG
6.49 2.13
77
Landers
7.09 0.79
Northridge R
10.92 1.58
Kobe
13.22 1.47
Table 3.16 gives the displacements at the crest of S1 subjected to fault normal and fault
parallel ground motions of near-field and S1 subjected to transverse and longitudinal ground
motions of far-field. The results obtained can be clearly explained that displacements observed
in thrust faults are higher in fault normal and lower in fault parallel when compared for the
same near-field earthquake, when the structure is in frequency range. The same case is also
observed in strike-slip fault. However, when S1 has undergone through Kobe near-field ground
motion, frequency of fault normal motion has not matched with the structures frequency,
leaving the structure to undergo less displacements compared to fault parallel which is in the
frequency range of the structure. Landers ground motion remained with different results
leaving fault parallel in near-field to be higher than fault normal and longitudinal displacements
to be higher than transverse in far-field.
Table 3.17: Fault normal-transverse and fault parallel-longitudinal response of S1 subjected to 5 ground motions
Max. linear displacement response at the crest of S1 - Near-field & far-field ground motions
Near-Field Far-Field
Fault Normal Fault Parallel Transverse Longitudinal
Tabas
9.79 9.41 0.25 0.19
Loma Prieta LG
8.59 6.49 2.97 2.13
Landers
6.21 7.09 0.46 0.79
Northridge R
15.04 10.92 2.17 1.58
Kobe
13.05 13.22 2.43 1.47
3.6 Comparison near-field and far-field response of S2
3.6.1 Fault normal and transverse response of S2
As mentioned in section 3.4.1 parameters like type of fault, magnitude, PGA, epicentral
distance, frequency range and two models (S1 and S2) influence the displacements in the
structures and obtained displacements are tabulated. Comparing near-field fault normal and
far-field transverse ground motions acting on S2, the displacements on the crest of S2 are
produced in Table 3.17. As PGA's are higher in near-field compared with far-field, the results
obtained can clearly show how a structure would behave if it undergoes near-field earthquake
effects. Also as S2 has a base of hard rock with a depth of 50 m, displacements at the crest are
observed to be higher compared to S1.
78
Table 3.18: Fault normal and transverse response of S1 subjected to 5 near-field and far-field ground motions
Maximum linear displacement response at the crest of S2 subjected to near-field fault normal
and far-field transverse ground motion
Near-Field - Fault Normal
Far-Field - Transverse
Tabas
11.65 0.32
Loma Prieta LG
10.14 4.86
Landers
6.96 0.49
Northridge R
20.96 2.50
Kobe
18.33 2.96
3.6.2 Fault parallel and longitudinal response of S2
Table 3.18 gives comparison between displacements in the structure caused by near-
field fault parallel and far-field longitudinal motions. As PGA's in near-field fault parallel are
higher the effect on the structure are higher compared with far-field longitudinal. Thus, as PGA's
are much higher in near-field the displacements observed are also higher. Also, as S2 has a base
of 50 m strong rock, the displacements obtained are further higher than S1.
Table 3.19: Fault parallel and longitudinal response of S1 subjected to 5 near-field and far-field ground motions
Maximum linear displacement response at the crest of S2 subjected to near-field fault parallel
and far-field longitudinal ground motion
Near-Field - Fault parallel
Far-Field - Longitudinal
Tabas
9.79 0.20
Loma Prieta LG
7.84 3.21
Landers
10.07 0.96
Northridge R
15.13 1.39
Kobe
11.89 2.19
Table 3.19 gives the displacements at the crest of S2 subjected to fault normal and fault
parallel ground motions of near-field and S2 subjected to transverse and longitudinal ground
motions of far-field. As explained in section 3.4.2 the results obtained can vary depending upon
type of fault and the structure lying in frequency range. Similar to the displacements observed
in S1, displacement in S2 is also same with increased scale. However, the failure pattern
remained same. The frequency of S2 lies in the range of near-field fault normal motion of Kobe
earthquake thus it gave higher displacement and fault parallel motion does not lie in the
frequency range of the structure thus gave lower displacement. Remaining all other ground
79
motions irrespective of PGA, fault normal had higher displacements than fault parallel and far-
field transverse had higher displacements than longitudinal. Displacements at the crest
amplified in S2 as it had strong rock of 50 m as base.
Table 3.20: Fault normal-transverse and fault parallel-longitudinal response of S2 subjected to 5 near-field and far-
field
Max. linear displacement response at the crest of S2 - Near-field & far-field ground motions
Near-Field Far-Field
Fault Normal Fault Parallel Transverse Longitudinal
Tabas
11.65 9.79 0.32 0.20
Loma Prieta LG
10.14 7.84 4.86 3.21
Landers
6.96 10.07 0.49 0.96
Northridge R
20.96 15.13 2.50 1.39
Kobe
18.33 11.89 2.96 2.19
Table 3.20 is the complete picture of chapter 3, in which the crest displacements are
observed for two structures (S1, S2).
Table 3.21: Linear displacement response at the crest of S1 and S2 subjected to near-field and far-field ground
motions
Maximum linear displacement response at the crest of S1and S2 subjected to both near-field
and far-field ground motions
S1 S2
Near Field Far-Field Near-Field Far-Field
FN FP Trans. Long. FN FP Trans. Long.
Tabas
9.79 9.41 0.25 0.19 11.65 9.79 0.32 0.20
LomaPrieta
8.59 6.49 2.97 2.13 10.14 7.84 4.86 3.21
Landers
6.21 7.09 0.46 0.79 6.96 10.07 0.49 0.96
Northridge
15.04 10.92 2.17 1.58 20.96 15.13 2.50 1.39
Kobe
13.05 13.22 2.43 1.47 18.33 11.89 2.96 2.19
3.7 Comparison of response spectrum of near-field and far-field ground motions
Response spectra are very useful tools of earthquake engineering for analyzing the
performance of structures and equipment in earthquakes, since many behave principally as
simple oscillators (single degree of freedom systems). A response spectrum is simply a plot of
the peak of steady-state response (displacement, velocity or acceleration) of a series of
oscillators of varying natural frequency, which are forced into motion by the same base
vibration or shock. The resulting plot can then be used to pick off the response of any linear
system, given its natural frequency of oscillation. One such use is in assessing the peak response
80
of structures to earthquakes. In the current study, a comparison is drawn between response
spectrums of components of near-field ground motions and components of far-field ground
motions, to understand the effect of ground motion for the natural time period of the dam
(0.335 sec). Fig. 3.11 to fig. 3.14 are drawn to describe the response spectrum of specific ground
motions for time period varying from 0-5 seconds.
Fig. 3.11 explains that the Tabas near-field fault normal ground motion can cause
damage to structures with low period, implying that the ground motion contains high frequency
short periods. Loma Prieta near-field fault normal component can cause damage to structures
with periods >0.6 sec. Response spectrum for Landers near-field fault normal ground motion
has less amplitude and could not cause much damage to structures. However, it has a peak at 0.1
sec, which states, it could be dangerous to the short structures with natural period of 0.1 sec.
Northridge ground motion also has very less effect, however it could cause damage to the
structures within period of 0.1 0.3 sec. Response spectrum of Kobe near-field fault normal
ground motion explains it could be dangerous to structures with natural time period between
0.3-0.4 seconds.
Figure 3.11: Acceleration Response Spectrum of 5 near-field fault normal ground motions
Fig. 3.12 explains the similarity of the fault normal response spectrum to the response
spectrum of near-field fault parallel components of Tabas, Loma Prieta, Landers, and Northridge
ground motions. Response spectrum of Kobe fault parallel ground motion is found to have effect
on structures with period 0.3 to 0.6 seconds.
81
Figure 3.12: Acceleration Response Spectrum of 5 near-field fault parallel ground motions
Fig. 3.13 and fig. 3.14 explains the response spectra of far-field transverse and
longitudinal components. As these ground motions are recorded far-field, they show less
energy. These could not cause any damage to structures.
Figure 3.13: Acceleration Response Spectrum of 5 far-field transverse ground motions
82
Figure 3.14: Acceleration Response Spectrum of 5 far-field longitudinal ground motions
3.8 Summary
The results are discussed individually in this chapter. Final conclusions could be drawn
from table 3.20, in which displacement responses of both structures (S1 and S2) subjected to
near-field and far-field ground motions were given. S1 is a concrete gravity dam with
foundation, where the whole structure is carrying same material property throughout. When
the natural frequency of the structure lies in the frequency of the ground motion, it showed
higher displacements. Even though the range of magnitude, PGA, epicentral distance and
frequency are near, displacements in thrust faults are observed to be higher than strike-slip
faults. S2 is a concrete gravity dam with foundation and base, where the structure had different
material properties for dam and base, where the base is modeled with hard rock properties.
With a base depth of 50 m, the displacements at the top of the dam are high in S2 compared
with S1 in all the cases.
With limitation in the modeling, reservoir effect is not added, making dam-foundation-
reservoir interaction not possible. However, in reality dam-foundation-reservoir interaction
makes the response of the structure to differ from numerical studies. The results thus obtained
in this linear analysis without considering reservoir effect are much lower than what could be in
real time. However, this study is also important when the dam has to be studied with empty
reservoir condition. Overall, Chapter 3 explains the linear displacement behaviour of S1 and S2.
Non-linear analysis of the two structures is explained in Chapter 4.
83
Chapter 4
Non-linear Earthquake Response of Concrete Gravity Dam
4.1 Introduction
In the previous chapter, linear analysis is conducted on the models considered for the
study. Life line structures like dams should always be considered to be designed for highest
safety. Near-field effects on a structure like dam can have adverse effects which can start with
minor cracks to complete collapse. Events that took place across the world proved how
devastating an earthquake could be. In this concern, a proper comparison between near-field
and far-field should be drawn. The difference between these two would alert the technocrats in
considering the near-field effects in designing the structure.
Now, in order to reach a safe economical seismic structural design during seismic
events, nonlinear analysis is indispe nsable as linear analysis would not allow the structure to
yield. In nonliner analysis, forces at which structure yields could be obtained. This makes the
designers to consider the forces which could cause damage to the structure. The material and
geometic details of Structure-1 (S1) and Structure-2 (S2), and material model considered for the
study are already disussed in Chapter-2. The main focus of this chapter is to study the non-
linear earthquake response of S1 and S2. These two models are subjected to fault normal and
fault parallel components of near-field ground motions, transverse and longitudinal
components of far-field ground motions. Comparision between these componets are drawn for
understanding the non-linear behavior of concrete gravity dam. Behavior of dam subjected to
these ground motions has been studied by understanding the crack propagation of non-
overflow section of dam with out the effect of reservoir. Later input ground motions which
could not cause damage to the structure are amplified to 50% are given as input and the failure
is studied. Increasing the amplitude still contains the same duartion and frequency
characteristics.
4.2 Comparison of near-field and far-field ground motions
4.2.1 Comparison of near-field and far-field response of S1 and S2 subjected to
Tabas ground motion
Tabas earthquake event had happened by a thrust fault. Its near-field fault normal
ground motion has a PGA of 0.9g. The fundamental frequency of S1 lies in the frequency range of
84
ground motion. Yet, no failure was observed in S1. Displacement at the crest is 9.78 mm. As
structure has not yielded, its non-linear response was almost equal to that of linear response. S1
thus remained in limits of linearity. PGA of fault parallel ground motion is 0.978g. First mode
frequency of S1 is in predominant frequency range of this record. Yet no higher displacements
were observed. Displacement at the crest is 9.38 mm. Minimum displacements in S1 can be
because of several reasons like (1) in a near-field ground motion energy is released in single or
multiple pulses at the beginning of event, which sometimes even become PGA of that specific
ground motion. Energy released in the record of Tabas near-field ground motion was less with
less amplitude at the beginning. (2) Duration of ground motion also can change the response of
the structure. According to Trifunac and Brady, Tabas ground motion's 5%-95% of energy is
distributed in stretched duration. (3) Also, the effects of an event caused by a thrust fault will be
lesser on foot wall side. The ground motion used in the analysis might be recorded on foot wall
side as observed displacements are very less for its magnitude and frequency range. Like the
effect of near-field ground motions, transverse and longitudinal components of far-field record
also could not have any effect on the structure and S1 behaved linearly. The intensity of the
ground motion at longer epicentral distances might also be a reason for lesser displacements in
case of far-field ground motion.
Failure of dam in S2 has been observed in case of near-field fault normal component. As
model S2 has a base of 50m, ground motion record given as input at the fixed base gets
amplified as it reaches the surface and to the crest of the structure. Displacement at the crest is
11.71mm. Total of 77 springs have failed. Initially bending tensile were observed at the neck of
the dam between 116m - 199m, where there is change in cross section. Bending compressive
cracks were later observed at the heel of dam at 50m (fig. 4.1). No cracks caused by shear were
observed. Fault parallel component of near-field ground motion could not cause damage to the
structure and the structure remained safe with linear response. Transverse and longitudinal
components of far-field ground motion also could not show effect on the structure.
Displacements at the crest of S1 and S2 for the components of near-field and far-field are
given in table 4.1. Near-field fault normal and fault parallel have distinct features compared to
far-field components. Most of the time, high energies are released in fault normal. In very few
cases it is so in fault parallel. PGA observed in this case is higher in fault parallel yet no higher
displacements or failure is observed in S1 or S2. Far-field sites located with higher epicentral
distances lack higher intensity to cause damage to the structure. Thus no serious effect on
structure could be seen. In this case, Tabas near-field and far-field ground motions are recorded
at epicentral distances of 1.2 km and 177.9 km respectively. Intensity in the ground motion
enough to cause failure to the structures at higher epicentral distances could be seen in very few
85
Figure 4.1 Crack propagation observed in S2 subjected to Tabas near-field fault normal ground motion
cases. S1 stood without failure in near-field fault parallel and in both components of far field
ground motion records.
Loma Prieta ground motion
Frequency range of components of near-field and far-field ground motion records of
Loma Prieta considered for the study does not match with the fundamental natural periods of
S1 and S2. Thus, when S1 and S2 are subjected to each component of near-field and far-field
ground motions, failure was not observed in the structure. Response of S1 and S2 remained
linear without any failure. As the epicentral distance of far-field ground motion is 42.13 km,
displacements observed in case of both near-field and far-field for S1 and S2 are little higher
compared to other ground motions. However, no failure has been observed. Displacements at
the crest of S1 and S2 for the components of near-field and far-field are given in Table 4.1.
Landers ground motion
Landers near-field fault normal and fault parallel ground motions frequency range
matches with first 3 modal frequencies of S1. Yet no failure was observed when analyzed for S1
as amplitudes in Fourier amplitude spectrum are less. PGA of fault parallel is higher and thus
displacements observed in case are little higher. However, higher PGA could not cause damage
to the structure. Also, transverse and longitudinal components of far-field ground motions also
could not have effect on the structure
Failure in of dam in S2 is observed when it is analyzed for both fault normal and fault
parallel components of near-field ground motion. Ground motion given at the base of S2 got
amplified on reaching the surface and to the top of the structure. Displacement at the crest in
fault normal case is 90.51mm. Total of 176 springs got failed. Bending tensile cracks were
observed at the height of 119m from base, at the change in cross section of dam (fig. 4.2). Also,
86
Figure 4.2: Crack propagation observed in S2 subjected to Landers near-field fault normal ground motion
Figure 4.3: Crack propagation observed in S2 subjected to Landers near-field fault parallel ground motion
failure of dam in S2 is observed, when it is analyzed for fault parallel case. Displacement at the
crest of S2 is 136.29 mm. Failure of 191 springs took place and all of them were observed at the
change in cross section of dam at the neck, at the height of 119 m. Bending tensile cracks were
initially observed on the downstream side and followed throughout the section (fig. 4.3).
Transverse and longitudinal components of far-field ground motion could not have effect on the
structure, because of 146.11km epicentral distance.
Displacement of the dam at the crest in models S1 and S2 for the components of near-
field and far-field are given in Table 4.1. As the PGA of fault parallel is higher than fault normal
and with matching frequency range, the structure went under higher displacement resulting in
the formation of cracks. No damage was observed at the base of the dam. With high epicentral
distance and slight intensity, components of far-field ground motion could not cause failure in
the structure.
Northridge ground motion
Northridge-Rinaldi near-field ground motion has pushed the structure beyond the limits
of linearity. The displacement at the crest of the dam for fault normal component is 178.31mm.
In this case, failure in S1 has took in even when the structures natural frequency does not match
with the frequency range of fault normal ground motion. Failure of 393 springs has been
observed in S1. Initially failure of cracks has been noticed at the heel of the dam at 5m height
87
from base, the section where foundation starts. Later bending tensile cracks were observed
between 71m - 74m from base, at the change in cross section on the downstream side. Ground
motion has been so intense that crack took place all through the section (at neck portion).
Predominant frequency range of fault parallel component matched with the first mode
frequency of S1. Displacement at the crest of S1 is 99.71mm when subjected to fault parallel
ground motion. Total of 315 springs failed and initial failure has been noticed at 71 m height
from base, at neck where there is change in cross section (fig. 4.5). Bending tensile cracks are
observed in this area. Transverse and longitudinal components of far-field ground motions
could not cause damage to the structure. Even though the PGA's of the components of far-field
are less, 70.45km of epicentral distance gave a chance for intensity to be carried. This is the
reason Northridge far-field ground motions have little higher displacements compared to other
far-field ground motions. However, they could not fail the structure.
Figure 4.4: Crack propagation observed in s1 subjected to Northridge near-field fault normal ground motion
88
Figure 4.5: Crack propagation observed in s1 subjected to Northridge near-field fault parallel ground motion
Northridge near-field fault normal ground motion had severe effect on S2. Total of 1155
springs failed. Displacement at the crest is 154.74 mm. Initially cracks were observed at the heel
of S2. Severe cracks were observed throughout the section at the base, from toe to heel.
Compression cracks and shear cracks were observed at the base. Bending compressive cracks
were observed at the toe. Whereas, at the neck bending tensile cracks were observed. The
frequency of ground motion does not match with the initial frequency of S2. However, once a
structure enters into non-linearity, frequency of structure keeps on changing, which might meet
the frequency of ground motion at certain point, where the structure undergoes severe failure.
With ground motion given as input at the base, it amplified on reaching the top of the structure
(fig. 4.6). Even though PGA of Northridge fault parallel motion is less, the structure has failed
and resulted in higher displacements. Displacement at the crest of the dam is 145.84 mm.
Figure 4.6: Crack propagation observed in S2 subjected to Northridge near-field fault normal ground motion
89
Figure 4.7: Crack propagation observed in S2 subjected to Northridge near-field fault parallel ground motion
Failure of 346 springs has been observed. Initially bending tensile cracks were observed at the
change in cross section of dam. These cracks propagated throughout the section from upstream
to downstream at 119 m height from base (fig. 4.7). Frequency range of transverse and
longitudinal components of far-field ground motion matches with the frequency of S2. However,
with lesser amplitude in ground motion and Fourier amplitude spectra far-field components
could not cause failure in the structure. However, with shorter epicentral distance, the intensity
of the ground motion was carried, which caused higher displacement in the structure compared
to other ground motions, which could not cause damage to the structure.
given in Table 4.1. In a near-field ground motion, sometimes PGA can be at the very beginning of
the ground motion releasing high energy in the form of high amplitude velocity pulse.
Northridge-Rinaldi ground motion record is one such example, and that effect is clearly
observed in S1 and S2 with high displacements when compared with other ground motions.
Far-field ground motions could not cause damage to the structure.
Kobe ground motion
Even though magnitude and PGA of Kobe ground motion are higher in near-field, it
could not cause severe damage to the structure as it is a strike-slip fault. Directivity effects
caused in fault normal or fault parallel components of near-field ground motions in case of
strike-slip faults generally used to less compare to the ones in reverse faults. In strike-slip faults,
permanent displacement of the fault would cause certain effects on structure. Displacement at
the crest caused by fault normal component is 13.14mm. Failure of 27 springs has been
observed in S1. All of these are caused by bending tension at the neck. Failure of springs has
been noticed between 71m - 74m on the downstream side (fig. 4.8). In case of fault parallel
component, displacement at the crest is 13.23mm. Total 343 springs have failed. Even though
the PGA of fault normal is higher than fault parallel, little higher displacement took place in fault
parallel. The reason is, the frequency of S1 matches with the frequency of fault parallel and does
90
not match with fault normal. Cracks that were noticed are caused because the structure
experienced bending tension in that region. Failures of springs were seen first on the
downstream side and the crack propagated throughout the section, at the change in cross
section of damat 71m height from base (fig. 4.9). The epicentral distance of 135.63km in case of
far-field ground motion record could not cause any damage to the structure for both transverse
and longitudinal components. The frequency range of the transverse and the longitudinal
ground motion records matches with the fundamental frequency of the structure. However, as
the amplitude of ground motions and Fourier amplitude spectrum are less, they could not cause
damage to the structure.
796 springs have failed in S2 when subjected to fault normal ground motion record and
initial failure has been observed at the base of the dam. Later huge number of springs failed at
the neck. Compressive cracks were observed at the base and bending tensile cracks were
observed at the neck (fig. 4.10). Displacement at the crest is 85.02 mm. Fault parallel ground
motion resulted in failure of 242 springs. Initially cracks were observed at the base of the dam.
Bending tensile cracks were later observed at the change in cross section of neck., at the height
of 119 m height from base. Failure of springs took place throughout the section from upstream
to downstream face. Displacement at the crest of dam is 11.84 mm (fig. 4.11). Even though the
frequency range of ground motion matches with the structural frequency, with an epicentral
distance of 135.6 km, transverse and longitudinal components of far-field ground motion could
not cause damage to S2.
given in Table 4.1. Fault normal frequency range does not match with the natural frequency of
structure. However, the structure resulted in higher displacement and thus more number of
springs have failed in S2. This was not the case in S1 as in S2, with a base of 50 m, amplification
of ground motion takes place as it reaches the crest. Even though the fundamental frequency of
the S1 and S2 matches with the frequency range of fault parallel, due to less amplitude, it could
not produce higher displacements or failure. Dam in the models S1 and S2 remained safe when
subjected to far-field ground motions.
Displacement response of near-field and far-field ground motions of S1 and S2 are given
from fig. 4.12 to fig. 4.19. These response histories are scaled to a single value, though which it
could be easily understood how displacements varied between ground motions. With the
analysis conducted on S1 and S2 for both the components of near-field and far-field, models S1
and S2 have failed only in near-field, making this as an important note that these near-field
events can occur anywhere along the fault and would severely cause damage to the structure
91
lying near the fault line.
Figure 4.8: Crack propagation observed in s1 subjected to Kobe near-field fault normal ground motion
Figure 4.9: Crack propagation observed in s1 subjected to Kobe near-field fault parallel ground motion
92
Figure 4.10: Crack propagation observed in S2 subjected to Kobe near-field fault normal ground motion
Figure 4.11: Crack propagation observed in S2 subjected to Kobe near-field fault parallel ground motion
Table 4.1: Comparison of non-linear response of S1 & S2 subjected to near-field & far-field ground motions
Non-linear displacement response (in mm) at the crest of S1and S2 subjected to both near-field
S1 S2
Tabas
9.78 9.38 0.26 0.19 11.71 9.78 0.32 0.21
LomaPrieta
8.60 6.49 2.96 2.12 10.14 7.84 4.87 3.21
Landers
6.11 7.12 0.47 0.79 90.51 136.29 0.49 0.95
Northridge
178.31 99.71 2.15 1.58 154.74 145.49 2.49 1.40
Kobe
13.14 13.23 2.43 1.47 85.02 11.84 2.96 2.19
In few cases failure was not observed even for near-field events. For understanding the
behaviour of dam and propagation of cracks, same ground motion for which failure has not been
observed are amplified to 50% of their original acceleration and response of models are studied.
This would keep the characteristics of ground motion similar; however with varied amplitude.
And failures in S1 and S2 caused by amplified acceleration are described in next section 4.3 with
Table 4.2.
93
Figure 4.12: Non-linear displacement response of S1, subjected to 5 near-field fault normal ground motions
94
Figure 4.13: Non-linear displacement response of S1, subjected to 5 near-field fault parallel ground motions
95
Figure 4.14: Non-linear displacement response of S2, subjected to 5 near-field fault normal ground motions
96
Figure 4.15: Non-linear displacement response of S2, subjected to 5 near-field fault parallel ground motions
97
Figure 4.16: Non-linear displacement response of S1, subjected to 5 far-field transverse ground motions
98
Figure 4.17: Non-linear displacement response of S1, subjected to 5 far-field longitudinal ground motions
99
Figure 4.18: Non-linear displacement response of S2, subjected to 5 far-field transverse ground motions
100
Figure 4.19: Non-linear displacement response of S2, subjected to 5 far-field longitudinal ground motions
101
4.3 Understanding behaviour of models S1 and S2 with amplified accelerations
Table 4.2 gives the displacement at the crest of S1 and S2 for fault normal and fault parallel
components of near-field, transverse and longitudinal components of far-field ground motions,
which are amplified by 50%. To understand the characteristics of ground motions for which failure
was not observed with original acceleration, amplification is done. Response and behavior of those
models failed only after amplification are described in this section.
4.3.1 Comparison of near-field and far-field response of S1 and S2 subjected to Tabas
ground motion
Tabas near-field ground motion is recorded at a very close epicentral distance of 1.2 km.
PGA of fault normal is 0.9g. The fundamental frequency of S1 lies in the predominant frequency
range of ground motion. Yet, no failure was observed in S1. With amplification of this component of
ground motion, the PGA got amplified from 0.9g to 1.44g. With this ground motion as input, failure
of dam has been seen in S1. Displacement at the crest is 103.16 mm. Total 338 springs have failed.
Initial crack started at the change in cross section of dam at 71 m height from base on the
downstream side. These are observed to as bending tensile cracks. Failure of more number of
springs took place through the section from downstream to upstream. Failure of springs can be
Figure 4.20: Crack propagation observed in S1 subjected to Tabas near-field fault normal ground motion
102
Figure 4.21: Crack propagation observed in S1 subjected to Tabas near-field fault parallel ground motion
linked as formation and propagation of cracks and are observed between 69m-74m (fig. 4.20).
Failure of 236 springs was observed in fault parallel amplified ground motion, which has an
amplified PGA of 1.56g. Displacement at the crest of S1 is 129.55 mm. As the PGA is higher and the
structures frequency matching with the ground motion frequency more displacement is observed.
Bending tensile cracks were observed at the change in cross section of dam at 71 m from base. More
springs failed throughout the section between 71 m- 74 m (fig. 4.21).
Near-field fault parallel ground motion with original acceleration could not cause damage to
the structure. However with amplified acceleration, S2 has undergone failure of 594 springs.
Displacement at the crest of S2 is 263.07 mm. In this case also, fault parallel component caused
more crest displacement than fault normal. Initially bending tensile cracks were observed at the
change in cross section of neck at 116 m height from base. Later compressive cracks were observed
at the heel of the dam at 50 m height from base of S2 (fig. 4.22). At neck portion, crack propagation
took throughout the section from upstream to downstream. Displacement at the crest is given in
table 4.2.
106
Figure 4.27: Crack propagation observed in S1 subjected to Landers near-field fault parallel ground motion
Table 4.2 gives the non-linear amplified displacement response at the crest of structures S1
and S2, subjected to the components of near-field and far-field ground motions. S1 and S2 have not
failed for some original accelerations. Those which have not failed for original accelerations and
failed for amplified accelerations are shown in bold format.
Table 4.2: Comparison of non-linear response of S1 & S2 subjected to amplified near-field & far-field ground motions
Non-linear amplified displacement response at the crest of S1and S2 subjected to both near-field
S1 S2
Tabas
0.40 0.32 223.64 0.51 0.33
LomaPrieta
10.41 4.72 3.39 7.80 5.14
Landers
0.75 1.27 95.03 292.15 0.78 1.53
Northridge
100.45 506.31 3.43 2.52 182.82 203.22 3.99 2.24
Kobe
221.98 31.92 3.89 2.35 216.17 42.91 4.74 3.51
107
4.4 Summary
Tabas and Northridge ground motions are caused by rupture along a thrust fault. As
directivity and fling effects in case of thrust fault are both along fault normal side, high energies are
released in this event. However, the effect of forward directivity could not cause severe effect on
foot wall side. Tabas ground motion record used in the analysis got higher PGA, magnitude and
matching frequency range with structure. However, displacements observed are significantly not
high compared to other ground motions. In case of Northridge, even though the structure frequency
does not lie in the frequency range of fault normal record failure was noticed to a high degree and
it is because the ground motion had long periods and high energy is released in initial pulse of
acceleration.
Kobe and Loma Prieta ground motions are cause due to rupture along strike-slip fault. In
this case, directivity effects are observed in fault normal direction and fling effects in fault parallel
direction. If structures natural frequency matches with the frequency range of ground motion, then
the response would be higher mostly for fault normal component. Fault parallel component could
not cause severe effect in case of strike-slip fault and is clearly observed in the behavior of S1 and
S2 subjected to these ground motion records as input. Displacement response of near-field and far-
field ground motions of S1 and S2 are given from fig. 4.28 to fig. 4.35. These response histories are
scaled to a single value, though which it could be easily understood how displacements varied
between ground motions.
In this chapter, comparison of non-linear response of S1 and S2 subjected to both near-field
and far-field are discussed. In the next chapter, effects of fault motion on S1 and S2 will be analyzed.
Also the effect of hanging wall and foot wall is also studied.
108
Figure 4.28: Amplified non-linear displacement response of S1, subjected to 5 near-field fault normal ground motions
109
Figure 4.29: Amplified non-linear displacement response of S1, subjected to 5 near-field fault parallel ground motions
110
Figure 4.30: Amplified non-linear displacement response of S2, subjected to 5 near-field fault normal ground motions
111
Figure 4.31: Amplified non-linear displacement response of S2, subjected to 5 near-field fault parallel ground motions
112
Figure 4.32: Amplified non-linear displacement response of S1, subjected to 5 far-field transverse ground motions
113
Figure 4.33: Amplified non-linear displacement response of S1, subjected to 5 far-field longitudinal ground motions
114
Figure 4.34: Amplified non-linear displacement response of S2, subjected to 5 far-field transverse ground motions
115
Figure 4.35: Amplified non-linear displacement response of S2, subjected to 5 far-field longitudinal ground motions
116
Chapter 5
Fault Motion Analysis
5.1 Introduction
In the previous chapters, structures S1 and S2 are modelled and analysed for the
components of near-field and far-field ground motions. The recorded ground motion data used
for the analysis was a free-field data, where the instruments might be placed either on the
ground or at the bed rock level. This input acceleration data was used to study the behaviour of
structures S1 and S2. Later, comparison on the response and the behaviour of structures S1 and
S2 are made. In this chapter, the study on behaviour of dam in models S1 and S2 subjected to
uplift displacement generating the effect of reverse fault motion as direct method and sub-
structure method is carried. In direct method, the effect of fault motion on dam in S2 is studied
by placing dam at different locations on foot wall and hanging wall. In sub-structure method,
free-field accelerations are recorded at different locations on foot wall and hanging wall while
only the sub structure was subjected to fault motion are used to analyse super structure S1.
Comparison is later drawn between S1 and S2.
In real time, local soil conditions will have strong inuence on the characteristics of
ground motions like amplitude, frequency, and duration. The level to which this inuence is
monitored depends upon the geological, geographical, and geotechnical conditions. The most
inuential factors are the properties of soil layers over the bedrock, topography of the region,
type of the bed rock, depth of the overlying soil etc.,. The effect of these factors on response of
the structures can be observed when the structure is analysed numerically for free eld ground
motion and when the structure experiences real earthquake. For the seismic analysis of
overground or underground structures, consideration of soil-structure interaction is essential.
In order to understand the soilstructure interaction problem, knowledge on the earthquake
wave propagation through the soil medium is necessary. The effect of dynamic soil-structure
interaction depends on the stiffness and mass properties of the structure, the stiffness of soil,
and the damping characteristics of both soil and the structure.
During earthquake excitation, structures interact with the surrounding soils which
undergo deformations. These deformations could cause the movement at the supports or at the
interface region of the soil and the structure. These effects could not be seen when the super
structures are analysed numerically with free field recorded ground motions. At the end,
response of the structure varies with a large extent, when in reality structure interacts with soil
120
(i)
(ii)
Figure 5.5: (i) model of S2c (ii) displacement of 2.5m as input to S2c
(ii)
(ii)
Figure 5.6: (i) model of S2d (ii) displacement of 2.5m as input to S2d
(i)
(ii)
Figure 5.7: (i) model of S3 (ii) displacement of 2.5m as input to S3
120
(i)
(ii)
Figure 5.5: (i) model of S2c (ii) displacement of 2.5m as input to S2c
(ii)
(ii)
Figure 5.6: (i) model of S2d (ii) displacement of 2.5m as input to S2d
(i)
(ii)
Figure 5.7: (i) model of S3 (ii) displacement of 2.5m as input to S3
121
Figure 5.8: Accelerations recorded at 4 different locations on the surface, when base is subjected to 2.5m upward
displacement with a rise time of 5sec
5.3 Dynamic soil structure interaction
Dynamic soil-structure interaction problems can be treated in different ways. Rigorous
or exact analysis could be done and appropriate results could be obtained under particular
conditions. Idealization of realistic problem and bounded problem both does have their
advantages and disadvantages. In this chapter, idealization of realistic problem is studied as
direct method and bounded problem as sub-structure method.
5.3.1 Direct method
In direct method dam is placed at different locations on foot wall and hanging wall and
displacement is given as an input at the bedrock level (fig. 5.3 fig. 5.6). Giving dynamic
displacement produces free field ground motion on the surface. This causes the energy in free-
field ground motion to sum up with kinematic interaction and inertial interaction produced by
dynamic displacement to act at the structure and soil interface [T.K Datta, 2010]. Advantage
with this method is that the above expressed effect could be useful to solve the problem
realistic. However, this method does also have a disadvantage. As some portion of the soil
adjacent to the foundation undergoes vibration, some energy emitted from the ground motion is
absorbed by the soil mass, as the energy produced within the soil will gradually die in the form
122
of radiating waves. These interaction effects could change the response of structure. Another
disadvantage with the method is that, the structure is surrounded with infinite soil on all
directions (except vertical), which is impossible to model with the heterogeneities in the soil
medium. Later effective dissipation of radiation energy produced in the event is hard to model.
This approach is a direct method of solving soil-structure interaction.
5.3.2 Sub-structure method
In the sub-structure method of analysis, both the sub-structure and super structure are
modelled separately and are treated as two independent models. The connection between two
models could be established by taking the forces on the surface, where these forces are in turn
used as input on the super structure to achieve results [T.K. Datta, 2010]. The foundation
medium is treated as half-space to absorb the radiation waves. For some particular models,
portion of soil can also be included in the super structure and can be analysed. Base is modelled
individually as a substructure. Displacement given at the base generates acceleration on the
surface and is recorded (fig. 5.8). This is in turn used as input on model S1 to study the
behaviour of dam.
Many software packages are now available to solve dynamic soil-structure interaction
problems. However, with the complexities imbibed in modelling through direct method,
restricts us to use sub-structure method (analyse the super-structure with already recorded
ground motions), where the inputs in the form of time histories are given to the structures. With
limitations even in AEM, to understand the effect of fault motion on the behaviour of dam efforts
are made to understand the effect on structure using direct method and sub-structure method.
This study is described in section 5.4.
5.4 Non-linear response of concrete gravity dam subjected to fault motion
Using direct method and sub-structure method, non-linear displacement response at the
crest of dam, when dam placed on foot wall and hanging wall of the fault is given in table 5.1.
Table 5.1: Non-linear displacement response of damfor direct and sub-structure methods
Direct Method Sub Structure Method
Dam placed
between
Displacement at the
crest of the dam
(m)
Free field
Acceleration
recorded at
Displacement at the
crest of the dam
(m)
Foot wall side
100m-200m 0.005 150m 0.002
200m-300m 9.791 250m 0.264
Hanging wall
side
500m-600m 1.521 550m 1.095
600m-700m 2.198 650m 1.732
123
5.4.1 Dam modelled on foot wall
Fig. 5.9 and fig. 5.10 show the behaviour of dam on foot wall side using direct method.
When the dam is modelled between 100m-200m on the foot wall, the rupture could not reach
the base of the dam and the energy levels at this distance could not affect the structure.
However, the interface between the structure and the soil at the toe of the dam got separated by
cracks being observed at this location. Displacement observed at the crest of dam is 0.005m (fig.
5.9). No failure is observed in the behaviour of dam. When the dam is positioned between
200m-300m, the dam experienced fault rupture right under it. This caused failure in the dam to
a major extent. Cracks were observed throughout the foundation. Cracks were also seen at the
toe, heel, and change in cross section at neck, downstream, upstream and even through the
middle of dam. Failure of the whole structure is observed in tension and compression cracks. As
the rupture hit the base, the dam has undergone major lateral movement. Rocking effect has
been observed as the dam is pushed backwards lifted from the toe. Displacement observed at
the crest is 9.791 (fig. 5.10).
Sub-structure method is used to analyse S1 with free field ground motion recorded on
foot wall at 150m of S3. S1 has not undergone any failure. PGA of 0.19 g was observed.
Displacement at the crest of dam is 0.002m. When S1 is analysed for free field ground motion
recorded at 250m, it has under gone failure. PGA recorded at this location is 2.3g, which is very
high. The acceleration recorded at this station is high as fault rupture is observed under it.
Initial failure was observed as cracks were generated at the heel of the dam and later the cracks
were observed on the upstream side, between 55m-75m. Cracks were later seen throughout the
base and at the neck, on downstream side. Lot of tension and compression cracks were
observed throughout the structure. Displacement at the crest is 0.264m (fig. 5.13).
Displacement responses at the crest of dam are given in fig. 5.16 and fig. 5.17.
5.4.2 Dam modelled on hanging wall
Fig. 5.11 and fig. 5.12 shows the behaviour of dam on hanging wall side using direct
method. When the dam is modelled between 500m-600m on the hanging wall side, the effect of
radiation waves caused failure of dam. Radiation is observed more on the hanging wall as the
waves keep on reflecting and stays within the beak of the hanging wall. As the dam lies much
closer to this beak, failure was observed. Tension, compression and shear cracks were observed
throughout the structure. Cracks were seen both on downstream and upstream side.
Displacement at the crest of dam is 1.521m (fig. 5.11). When the dam is modelled between
600m-700m on the hanging wall side, the dam experienced cracks at the change in cross section
of neck, caused much or less by radiation of energy. Cracks were seen both on downstream and
124
upstream side. Initial cracks started on the upstream side. Cracks are observed between 60m-
70m, mostly at the change in cross section, near neck of the dam. Displacement at the crest of
dam is 2.198m (fig. 5.12). This is due to the effect of fault movement, as the structure moved
away from the fault on hanging wall side; the structure is lifted to front side, lifted at its heel.
Therefore huge displacement is observed at the crest.
Sub-structure method is used to analyse S1 with free field ground motion recorded on
hanging wall at 550m of S3. PGA observed at this location is 1.154g. Initially cracks are observed
at the change in cross section of dam. Cracks are observed both on downstream and upstream
side. Later cracks were also observed at the base of the dam. Downstream side at the base has
also experienced cracks. Displacement at the crest is 1.095m (fig. 5.14). When S1 is analysed for
free field ground motion recorded at 650m, it has under gone failure. PGA recorded at this
location is 1.484g. The acceleration recorded at this station is higher than record measured at
550m, which is near to the fault. Initial failure was observed as cracks were generated on the
upstream side, between 55m-75m. Cracks were later seen throughout the base and at the neck,
on downstream side. Lot of tension and compression cracks were observed throughout the
structure. Displacement at the crest is 1.732m (fig. 5.15). Displacement responses at the crest of
dam are given in fig. 5.16 and fig. 5.17.
5.5 Summary
Both in direct and sub-structure methods, failure of the structure was not seen when the
structure is analysed being away from the fault on foot wall side. As the structure nears the fault
on foot wall side, huge displacements are observed at the crest of the dam. This is the reason
because; the fault rupture took place right under the dam in direct method and also at the
element where the free-field ground motion is recorded in sub-structure method. When the
structure is placed on the hanging wall side, the effect of reflection of waves produced in the
beak of hanging wall caused emission of huge radiation energy and thus failure in the structure.
Cracks are observed to be more on downstream side, as the downstream side faces towards the
fault. The same is observed in sub-structure method also. The displacements are also observed
lesser. When the dam is placed little away from the fault on the hanging wall side, the actual
effect of hanging wall is observed. PGA is observed higher and so the displacement at the crest
also. The interaction of structure with disturbances in the soil along with displacement has
affected the structure. Both in direct method and sub-structure method, failure is observed both
at the base of the damand at the change in cross section of dam.
The study of fault motion analysis showed how devastating an earthquake could be,
when the fault lies just below the dam. 1999 Chi-Chi, Taiwan earthquake, which was witnessed
125
by the whole world, demonstrates the same effect, where the dam is subjected to the
displacement of fault lying below it, causing 9m vertical and 2m horizontal displacement in the
dam. As the failure of dam could release the reservoir, which could inundate thousands of
square kilometres on the downstream, it is necessary to built dams which can sustain extreme
loading conditions. With this study on near-field earthquake effects and fault motion analysis on
dam, it could be concluded that the design of these structures should be done with great care
and by considering all possible severe effects. Finally, as near-field ground motions proved to
have specific characteristics which are dangerous compared to their far-field, it is necessary to
consider the near-field earthquake effects while designing the dam
Figure 5.9: Failure of dam (at 100m-200m) subjected to 2.5m displacement in the base
126
Figure 5.13: Failure of dam subjected to free-field acceleration recorded at 250m
127
128
Figure 5.16: Displacement response at the crest of dam (dam placed at 4 different locations), subjected to 2.5m displacement with a rise time of 5 sec
129
Figure 5.17: Displacement response at the crest of dam; acceleration recorded at 4 different locations when base is subjected to 2.5m displacement with a rise time of 5 sec
130
Chapter 6
Conclusions and Future Scope
6.1 General conclusions
The analysis considered linear and non-linear earthquake response of structures S1 and S2
subjected to near-field and far-field ground motions of the same event. Displacement of the
structures is obtained and damage of the structure is described with cracks and their
propagation. The analysis was further extended to fault motion analysis. The main conclusions
can be summarized as below
Linear and non-linear analysis:
Near-field ground motions have severe effect on the structure. Several parameters like
magnitude, PGA, epicentral distance, frequency range and type of fault played their role in
determining the displacement of the structure. Shift in the stiffness of the structure took to the
section above the neck. This gave higher displacements at the crest and tensile failure at the
change in cross section at neck. Until the structure entered non-linearity, the displacements
were similar like linear. Structures which have not failed in non-linear analysis gave similar
displacement results like linear.
Ground motions recorded in the events caused by thrust fault had severe effects than strike-slip
fault. Displacements observed at the crest of the dam in S2 are higher than S1. This was
observed to be so because, S2 has a base and ground motion at the base amplified
displacements at the crest. In specific, fault normal components had more effect than fault
parallel and totally near-field ground motions caused more damage to the structure where as
far-field ground motions could not cause any damage to both S1 and S2.
Fault motion analysis:
Displacement at the base developed thrust fault effect at the surface. Dam placed on hanging
wall had more damages than dam placed on foot wall. In one case, the fault was observed just
below the dam. In this case damage to the dam observed was very high. Far from the fault on
foot wall side, no effect on the dam was observed. On hanging wall side, higher displacements
and failures were observed at locations where the frequency of ground motion caused by fault
motion matched with the frequency of the structure. In sub-structure method, where the
131
accelerations recorded on the foot wall side could not cause effect on the structure. Whereas,
the accelerations recorded on the hanging wall created damage in the structure.
Study on near-field earthquake effects has started in the last two decades. This is
because of lack of enough number of strong ground motion recorders very near to the fault. The
adverse characteristics of these near-field ground motions are still not being considered in
designing of structures except in very few important projects. Life line structures like dams
should be designed by considering these effects as the current study proved that these ground
motions could even cause damage to the dams. Codes should speak about these effects and
should consider them seriously. Effects of these characteristics should be included in codes.
Microzonation of the whole land mass of India would eventually benefit, which makes near-field
effects to be considered in those areas, which are in highly seismic zone and very near to active
faults.
6.2 Future work
This work can further be continued in various directions
1. Obtaining design spectrum which considers near-field earthquake effects.
2. Study of fault motion analysis varying rise time of fault and displacement parameters.
3. Comparison on study of behaviour of dam when dam is on foot wall, by modeling
downstream side of dam towards fault and by modeling upstream side of dam towards
fault.
4. Comparison on study of behaviour of dam when dam is on hanging wall, by modeling
downstream side of dam towards fault and by modeling upstream side of dam towards
fault.
5. Study of near-field effects on dam, by 3D modeling.
6. Similarly study of fault motion analysis on dam in 3D.
7. Reservoir effect can be considered, which makes complete dam-foundation-reservoir
interaction problem.
132
Appendix A1
History and necessity of dams
Dams are not new to this world. And humans are not the only ones to construct them. In
fact, humans learnt to build dams from Beavers. For almost 6000 years, dams have enabled
civilization to collect and store water. Throughout the world, histories of dams have been
successful in upholding and enhancing the quality of life. At present, the oldest dams believed to
be known are very few:
The earliest known dam is situated in Jawa, Jordan, 100 km northeast of the capital
Amman. This gravity dam featured a 9 m high and 1 m wide stone wall, supported by a
50 m wide earth rampart. This structure is dated to 3000 BC.
The Ancient Egyptian Sadd Al-Kafara (in Arabic means Dam of the Pagans) at Wadi Al-
Garawi, located about 25 kilometers south of Cairo, was 11 m high, 102 m long at its
base and 87 m wide. The structure was built around 2800 or 2600 B.C as a diversion
dam for flood control. However, it got collapsed by heavy floods and over topping during
construction or shortly afterwards.
The oldest surviving and standing dam in the world is believed to be the Quatinah
barrage in modern day Syria. The dam is assumed to date back to the reign of the
Egyptian Pharaoh Sethi (13191304 BC), and was enlarged in the Roman period and
again in between 1934-1938. It still supplies the city of Homs with water.
The Kallanai is a massive dam of unhewn
stone, over 300 m long, 4.5 m high and 20
m wide, across the main stream of the river
Kaveri in India. The basic structure dates to
the 2nd century AD. The purpose of the
dam was to divert the waters of the Kaveri
across the fertile Delta region for irrigation
via canals. Figure A1.1: Kallanai Dam in India dated
2
nd
century AD
The Romans were also great dam builders, with many examples such as the three dams
at Subiaco on the river Anio in Italy. Many large dams also survive at Mrida in Spain.
In China, Iran, Netherlands, and in few other places, there are still many dams existing,
which were built 2000 years ago.
133
The history of dams described above have support to tell about the importance of
storing water, and the hard work done by ancestors to preserve water. Even in the present
world, with rapid growth of population and the consequent demand for water for various
purposes, it has become necessary to construct more number of dams. However, due to lack of
technology, people in the past have failed in retaining and rehabilitating the dams. However,
there is no unique way to store huge water other than dams. This is why in the present world
with new growing technologies; we see different shapes in dams. Some dams are tall and thin,
while others are short and thick. And even dams are made from a variety of materials such as
rock, earth, concrete etc. varying from small earthen embankments to massive concrete
structures. Considering all these parameters, to reach the needs of humans and their activities,
construction of dams has became the most important and necessary item which can't be ignored
from very beginning of planning for a dam to selection of site, until its construction and
maintenance.
134
Appendix A2
Causes of dam failures
By the end of 20
th
century, there are over 45,000 large dams built in 150 countries
(International Commission on Large Dams ICOLD). No doubt, the dams provide the mankind
with sufficient benefits. However, if any dam breaks or breaches, the large volume of water
stored in the reservoir gets suddenly released and flows in the downstream valley resulting in a
catastrophe. Thus the analysis of "Dam Failure" has attained significance in concerns of Dam
Safety. Every structure which is built will have a life time and so for dams. However, the failures
can also occur before the structures life time with more than a few reasons.
Occurrences of failures reveal that depending on the type of dam, the causes of failures are of
several types. However, the maximum number of failures can be seen in earthen dams as
concrete masonry dams are stronger because of material properties. A study of dam failures in
the world has revealed the percentage distribution of dam breaks and its featured causes of
failures.
Table A2.1: Causes of failures of dams around the world with percentage
Cause of failure % Cause of failure
Foundation problems 40
Inadequate spillways 23
Poor Construction 12
Uneven Settlement 10
High pore pressure 5
Acts of war 3
Embankment slips 2
Defensive materials 2
Incorrect operations 2
Earthquakes <1
135
Appendix A3
Special Focus on Earthquake Effects
An earthquake is a result of sudden release of strain energy from the rocks in the earth
crust, which in turn manifests themselves by shaking and sometimes displacing the ground on
the earths surface. The amount of energy released by an earthquake is so huge that it can
collapse any structure in its vicinity if its magnitude is very high. If the structure lies very close
to the source of an earthquake, energy released at the beginning of the event might cause more
damage, than to the structure which lies far. These additional characteristics can be observed
very close to the epicenter and their intensity reduces to the farther distances depending upon
the rupture magnitude and also the properties of the soil. These are thus known as near-field
earthquakes. Even though the percentage cause of failure of dams due to earthquakes are very
less, there is a possibility that the dam might be very close to an active fault or inactive fault
which might become active, causing a rupture which produces the effects of near-field
earthquake on dam and might lead to the failure of the dam. The catastrophe that a dam can
make over other causes of failures is very high in terms of loss of property and life. Earthquakes
alone have got several effects on structures; however, they are not limited to shaking and
ground rupture. Landslides, avalanches, fires, soil liquefaction, tsunami, floods, and tidal forces
are few of secondary effects. Therefore the study of a structure subjected to an earthquake is
necessary and for one of the life line structures like dam subjected to near-field effects, it is
compulsory to design it as an earthquake resistant irrespective of seismic sector it is present in.
Near Field Earthquakes
Near-Field earthquake is caused by shear dislocation that begins at a point on the fault
and spreads at a velocity that is almost as large as the shear wave velocity. The propagation of
fault rupture toward a site at very high velocity causes most of the seismic energy from the
rupture to arrive in a Single Large Long Period Pulse of motion which occurs at the beginning of
the record [Somerville and Graves 1993]. This pulse of motion represents the cumulative effect
of almost all of the seismic radiation from the fault. The radiation pattern of the shear
dislocation on the fault causes this large pulse of motion to be oriented in the direction
perpendicular to the fault, causing the strike-normal peak velocity to be larger than strike-
parallel peak velocity.
136
Strike normal refers to the horizontal component of motion normal to the strike of the
fault
Strike parallel refers to the horizontal
component of motion parallel to the strike of the
fault
Figure A3.1: FN, FP & V Accelerogram of
1994 Northridge earthquake
Important features of near-fault ground motions
Near fault ground motion comprise of velocity pulse. And the two main causes being directivity
and fling
Large velocity pulse
Two causes of large velocity pulses
Directivity
Fling
Directivity: It is related to the direction of rupture front. It is a two-sided velocity pulse due to
constructive interference of shear waves generated from parts of the rupture located between the
site and epicenter. It occurs at sites located close to the fault however, away from the epicenter or
near the epicenter depending on the wave propagation. The two kinds of directivity are
a) Forward Rupture Directivity
b) Backward Rupture Directivity
a) Forward Directivity: This occurs when these conditions are met. When shear wave
velocity coincides with the rupture velocity, the rupture propagates toward the site (site
away from the epicenter), and when the direction of slip on the fault is aligned with the
site Forward Rupture Directivity effect occurs. It is readily met in strike-slip faulting. And
not all near fault locations will experience forward rupture directivity in an event.
137
b) Backward Directivity: A backward directivity effect occurs when the rupture propagates
away from the site (site near the epicenter).
Figure A3.2: 1992 Landers earthquake, showing the Forward and Backward Directivity region
Fling: It is a one-sided velocity pulse due to tectonic deformation. It is related to the permanent
tectonic deformation at the site. Fling occurs at sites located near the fault rupture, independent
of the epicenter location.
Near-Field effects on dams & importance on study of near-field earthquake effects
The Northridge1994 (Mw 6.7) and Hyogoken-Nanbu-1994 (Mw 6.8) earthquakes have
revealed that near-field ground motions have very damaging effects on structures, if they were
not adequately taken into account with seismic design guidelines. Beginning with Landers
earthquake (Mw 7.3) of 1992, strong motion data began to be recorded from near-field stations
located within a few kilometers of the plane of fault rupture. These ground motions were
observed to differ dramatically from their far-field counterparts. They were characterized by
distinct large amplitude single or multiple pulses, large velocity pulses, forward rupture
directivity and larger ratio of vertical-to-horizontal components ratio (V/H), which was viewed
as damaging criteria. Other records from U.S (for example Pacoima Dam site) and Japan show
similar pattern. The Near-Field pulse-like velocity and displacement time histories associated
with a strong earthquake can greatly affect a wide range of different types of structures.
138
Concerns about the seismic safety of concrete dams have been growing during recent
years, partly, because the population at risk in locations downstream of major dams continues
to expand and also because it is increasingly evident that the seismic design concepts in use at
the time most existing dams were built were inadequate. Since the Northridge and Hyogoken-
Nanbu (Kobe) earthquakes, there has been much discussion about the adequacy of design
practice of concrete dams. The hazard posed by large dams has been demonstrated since 1928
by the failure of many dams of all types and in many parts of the world. However, no failure of a
concrete dam has resulted from earthquake excitation; in fact the only complete collapses of
concrete dams have been due to failures in the foundation rock supporting the dams.
The 1999 Chi-Chi, Taiwan earthquake (MW 7.6) has witnessed collapse of Shih-Kang
dam, which is 50 km from the epicenter. This
dam has failed due to differential thrust
fault movement. Over two thirds of the
dam body were uplifted about 9 m
vertically and displaced 2 m horizontally.
The dam experienced horizontal
accelerations up to 0.5g. However, the
damage was confined to only two bays
overlying the fault rupture. The reservoir
slowly drained through the failed bays,
without causing major ooding. This
example clearly shows how disastrous an
earthquake could be if an active fault is
lying near the dam site. Figure A3.3: Surface faulting caused major damage to Shii-kang Dam
In fact, lots of major dam sites are located alongside major active faults and so could be
subjected to near-field ground motions from large earthquakes. Knowledge of ground motion in
the near-field region of large earthquakes is limited by the scarcity of recorded data. The near-
field of an earthquake (also called near-source or near-fault region) is the region within which
distinct pulse-like particle motions are observed due to a coherent release and propagation of
energy from the fault rupture process. For damaging earthquakes, the near-field region may
extend several kilometers outward from the projection on the ground surface of the fault
rupture zone and its extension to the surface, particularly in the direction of rupture
propagation. The near-field ground motions are characterized by high peak acceleration (PGA),
high peak velocity (PGV), high peak displacement (PGD), pulse-like time history, and unique
spectral content. The nature of near-field ground motions differs significantly from that of far-
139
field ground motions. Therefore it is crucial that these near-field effects be identified and
thoroughly understood, and that appropriate mitigation measures are found to deal with these
special ground motions.
144
with length and seismic earthquake events. Geological survey of India has compiled all the
available geological, geophysical and seismological data for the entire India and has published a
seismotectonic map in 2000. Using a software package OpenJUMP GIS, seismic zonation of India,
fault data given by GSI and National Importance Dams given by NRLD are integrated. Fig. A4.6
shows the seismic zonation map of India, with faults and national importance dams. Fig. A4.7
shows the seismic zonation map of India with national importance dams and top 100 active
faults. The top 100 active faults are selected based on the energy released around these faults in
the events of earthquakes from 1064 AD 2009 AD.
Figure A4.7: National Importance Dams of India on Seismic Zonation Map plotted with top 100 active faults
NATIONAL REGISTER OF LARGE DAMS Information compiled up to April 2002
DAMS OF NATIONAL IMPORTANCE
Table A4.1a: Dams with height 100 mand above or with storage capacity of 1 km
3
and above completed
Sl
No
Name of Dam
Year of
comple-
tion
River Nearest city Type
Height
above
lowest
foundation
Length of
dam
Volume
content of
dam
Gross
Storage
capacity
Reservoir
Area
Effective
Storage
capacity
Pur-
pose
Designed
Spillway
capacity
m m
10
3
m
3
km
3
km
2
km
3
m
3
/sec
Andhra Pradesh
1 Nagarjunasagar Dam 1960 Krishana Hyderabad TE/PG 125.00 4865.00 7725.00 11.56 284.90 5.44 I/H 42509.00
2 Sriamasagar 1977 Godavari Nizamabad TE/PG 43.00 15600.00 13095.00 3.17 450.82 2.32 I/H 45300.00
3 SrisailamH.E.Project 1984 Krishna Hyderabad TE/PG 145.00 512.00 1953.00 8.72 616.42 4.25 H 38369.00
4 Somasila 1989 Pennar Nellore TE/PG 39.00 760.75 1875.00 2.21 212.28 1.99 I/S 22375.00
Chhattisgarh
5 Manimata (Hasdeo) Bango 1990 Hasdeo Bilaspur TE 87.00 2509.00 3.42 188.47 3.05 I/H 42842.00
Gujarat
6 Ukai 1972 Tapi Fort Songadh TE/PG 81.00 4927.00 25390.00 8.51 7.09 I/H/C 49490.00
7 Kadana 1979 Mahi Samtrampur TE/PG 66.00 1551.00 2364.00 1.54 1.20 I/H 49497.00
Himachal Pradesh
8 Bhakra Dam 1963 Sutlej Nangal PG 226.00 518.00 4130.00 9.62 168.35 7.19 I/H 8212.00
9 Pong Dam 1974 Beas Mukerian TE/ER 133.00 1951.00 35500.00 8.57 260.00 7.29 I/H 12375.00
10 Chamera 1994 Ravi Banikhet PG 140.00 295.00 1342.00 0.39 9.50 0.11 H 21760.00
Jammu &Kashmir
11 Salal 1986 Chenab Jammu ER/PG 113.00 486.80 8000.00 0.29 9.37 H 224000.00
Jharkhand
12 Maithon Dam 1957 Barakar Asansol TE/PG 56.00 4789.00 4170.00 1.35 106.19 0.61
I/H/S/C
14150.00
13 Panchet Hill 1959 Damodar Asansol TE/PG 49.00 6777.00 5560.00 1.58 153.00 0.25
I/H/S/C
17842.00
14 Tenughat 1978 Damodar Hazaribagh TE/PG 64.87 6494.00 1.02 64.80 1.00
I/H/S/
1600.00
Karnataka
15 Krishnarajasagar 1931 Cauvery Mysore PG 42.67 2621.00 849.00 1.37 129.00 1.16 I/S 9912.00
16 Tungabhadra 1953 Tunga-bhadra Hospet TE/PG 49.38 2440.22 1441.56 3.77 378.13 3.71 I/H 18403.00
17 Bhadra Dam 1965 Bhadra Shimoga TE/PG 76.81 1708.00 2625.00 2.03 117.25 1.79 I/H/S 3020.95
18 Linganamakki Dam 1965 Sharavathy Shimoga TE/PG 61.28 2750.00 4080.00 4.44 316.65 4.29 H 8070.00
19 Malaprabha 1972 Mala-prabha Belgaum PG 43.13 154.52 - 1.07 129.50 0.97 I 5238.60
ABBREVIATION:: Earth:TE # Rockfill :ER # Gravity/Masonry : PG # Irri : I # Hydel : H # Flood contrl : C #Water supply : S # Navigation : N # Pisc.cult : F
Dams with height 100 mand above or with storage capacity of 1 km
3
and above completed
Sl
No
Name of Dam
Year of
comple-
tion
Height
above
lowest
foundation
Length of
dam
Volume
content of
dam
Gross
Storage
capacity
Reservoir
Area
Effective
Storage
capacity
Pur-
pose
Designed
Spillway
capacity
m m
10
3
m
3
km
3
km
2
km
3
m
3
/sec
20 Hidkal 1977 Ghata-prabha Belgaum
ER/TE/PG
62.48 10183.00 14072.00 1.44 78.04 1.39 I/H/S 4615.00
21 Narayanpur 1982 Krishna Lingasagar TE/PG 29.72 10637.50 18270.00 1.07 132.00 0.86 I 37945.00
22 Hemavathy Project 1979 Hemavathy Hassan TE/PG 58.50 4692.00 5782.50 1.05 91.62 0.93 I/S 3624.52
23 Supa 1987 Kalinadi Dandeli PG 101.00 322.00 1020.00 4.18 123.00 3.76 H 2987.00
24 Mani 1989 Varahi Shimoga TE/PG 59.00 565.00 1000.00 1.01 56.50 0.93 H 1954.00
25 Lakhya 1994 Lakya Kudremukh TE 108.00 1048.00 11000.00 0.25 5.72 0.17
T'ling/S
33.00
26 Almatti 1999 Krishna Bijapur TE/PG 52.24 1564.85 1307.55 6.42 754.25 6.00 I/H 31007.00
Kerala
27 Kakki Dam(KSEB) 1966 Kakki Vandi-periyar PG 114.00 336.19 725.00 0.46 17.51 0.45 H 1783.96
28 Idukki Dam (KSEB) 1974 Periyar Idukki Arch 169.00 366.00 467.00 1.99 59.83 1.46 H/C 5012.00
29 Cheruthoni (KSEB) 1976 Cheruthoni Idukki PG 138.00 650.00 1700.00 1.99 59.83 1.46 H/C
30 Kulamavu Dam (KSEB) 1977 Muavothupuzha Idukki PG 100.00 385.00 453.00 1.99 59.83 1.46 H/C
31 Idamalayar (KSEB) 1985 Idamalayar
Kotha-mangalam
PG 100.00 373.00 860.00 1.15 28.30 1.02 H/C 3013.00
Madhya Pradesh
32 Gandhisagar 1960 Chambal Mandasur PG 62.17 514.00 7.32 660.00 6.80 I/H 21238.00
33 Tawa 1978 Tawa Itarsi TE/PG 57.91 1905.00 2.31 200.55 2.05 I 20500.00
34 Bargi (R.A.B.S.Project) 1988 Narmada Jabalpur PG 69.80 5385.50 8771.00 3.92 267.97 3.18 I/H 43617.00
35 Indira Sagar 2006 Narmada Punasa PG 92.00 208.00 12.20 913.48 9.75 I/H/C
Maharashtra
36 Koyna 1964 Koyna Patan PG 103.00 805.00 1555.00 2.80 115.00 2.64 H 3883.00
37 Paithan (Jayakwadi) 1976 Godavari Paithan TE/PG 41.30 10415.00 13410.00 2.91 398.00 2.17 I/H 18153.00
38 Ujjani 1980 Bhima Indapur TE/PG 56.40 2534.00 3320.00 3.14 336.50 1.44 I/H 15010.00
39 Isapur 1982 Penganga Nanded TE/PG 47.00 4120.00 11216.00 1.25 0.95 I 10480.00
40 Totladoh (Pench Hydel) 1989 Pench ramtek PG 74.50 680.00 972.60 1.24 77.71 1.09 H 12072.00
ABBREVIATION :: Earth:TE # Rockfill :ER # Gravity/Masonry : PG # Irri : I # Hydel : H # Flood contrl : C #Water supply : S # Navigation : N # Pisc.cult : F
Dams with height 100 mand above or with storage capacity of 1 km
3
and above completed
Sl
No
Name of Dam
Year of
comple-
tion
Height
above
lowest
foundation
Length of
dam
Volume
content of
dam
Gross
Storage
capacity
Reservoir
Area
Effective
Storage
capacity
Pur-
pose
Designed
Spillway
capacity
m m
10
3
m
3
km
3
km
2
km
3
m
3
/sec
Orissa
41 Hirakud 1957 Mahanadi Sambalpur TE/PG 59.00 4800.00
19330000.00
8.14 743.00 5.82 I/H/C/S 42450.00
42 Balimela 1977 Sileru Chitrakonda TE 70.00 1821.00
13856000.00
3.70 169.08 2.68 I/H/N 10930.00
43 Rengali 1985 Branmani Angol PG 70.50 1040.00 789.00 5.15 378.00 4.40 I/H 46960.00
44 Upper Kolab 1985 Kolab Jeypore PG 55.00 646.00 478510.00 1.22 122.00 0.94 I/H/S 10020.00
Punjab
45 Thein (Ranjit Sagar) 2000 Ravi Pathankot TE/ER 47.00 878.00 21920.00 3.28 2.34 I/H 19329.00
Rajasthan
46 Ranapratapsagar 1969 Chambal Kota PG 53.59 1143.00 567.00 2.90 198.29 1.57 I/H/S 18400.00
47 Mahi Bajaj Sagar 1985 Mahi Banswara TE/PG 74.52 3109.00 2.18 134.00 1.83 I/H 23786.00
Tamil Nadu
48 Mettur Dam 1934 Cauvery Mettur PG 70.41 1615.00 1546.00 2.71 153.46 2.65 I/H 11179.00
49 Sholayar Dam 1971 Sholayar Pollachi TE/PG 105.16 1244.00 2533.00 0.15 5.26 0.14 I/H 1475.00
Uttaranchal
50 Ramganga 1974 Ramganga P.Garhwal TE/PG 128.00 715.00 10000.00 0.25 19.72 0.22 I/H 8467.00
Uttar Pradesh
51 Rihand 1962 Rihand Sonbhadra PG 91.46 934.45 16827.00 10.60 468.00 8.90 H 13342.78
52 Matatila 1964 Betwa Jhansi TE 45.72 6315.15 5012.05 1.13 138..85 1.02 I/H 15850.00
53 Rajghat Betwa Lalitpur TE/ PG 43.80 10790.00 2.172 2453.00 1.945 I/H 33893.00
54 Jamrani 1990 Gola Kathgodam PG 140.00 465.00 0.21 0.21 I/H 3630.00
West Bengal
55 Kangsabati 1965 K.Bati/ Kumari Bankura TE 41.00 10400.00 11300.00 1.04 124.32 0.90 I/H 5715.00
Table A4.2: Dams with height 100 mand above or with storage capacity of 1 km
3
and above under construction
Sl
No
Name of Dam
Year of
comple-
tion
Height
above
lowest
foundation
Length of
dam
Volume
content of dam
Gross
Storage
capacity
Reservoir
Area
Effective
Storage
capacity
Pur-pose
Designed
Spillway
capacity
m m
10
3
m
3
km
3
km
2
km
3
m
3
/sec
Andhra Pradesh
1 Kandleru U/C Kandaleru Nellore TE 49.00 10752.00 16720.00 1.92 142.32 1.69 I
Gujarat
2 Karjan U/C Karjan Rajpipla PG 100.00 903.00 1440.00 0.63 0.58 I 17286.00
3 Sardar Sarovar U/C Narmada Bharuch PG 162.00 1210.00 17700.00 9.50 508.00 I/H/C 84949.00
Jharkand
4 Icha at Kuju Kharkai Chaibasa TE/PG 39.50 6600.00 3456.00 1.04 127.00 0.93 I 26000.00
5 Chandil U/C Subarnarekha Jamshed-pur TE 56.50 720.00 1480.00 1.96 174.10 1.61 I 32340.00
6 North Koel U/C North Koel Dalton ganj TE/PG 67.86 342.75 473.00 1.17 71.20 0.96 I/
H
16000.00
Karnataka
7 Kachige U/C Kachige Kudremukh TE 115.00 650.00 6000.00 0.09 20.90 0.06 T/S
Orissa
8 Indravati (Upper Indavati project) U/C Indravati Navrangpur PG 45.00 535.00 332750.00 2.30 110.00 1.49 I/
H
1121.50
9 Kapur (Upper Indavati project) U/C Kapyr Navrangpur PG 64.00 537.00
383600000.00
2.30 110.00 1.49 I/
H
10 Muran (Upper Indavait project) U/C Muran Navrangpur PG 65.00 604.00 700000.00 2.30 110.00 1.49 I/
H
7728.50
11 Podgada(Upper Indavati project) U/C Podgada Navrangpur PG 71.00 462.00 3100000.00 2.30 110.00 1.49 I/
H
Rajasthan
12 Bisalpur U/C Banas Deoli TE/PG 39.50 574.00 1096000.00 1.10 212.30 0.89 I/S 29047.00
Uttaranchal
13 Tehri U/C Bhagirathi Tehri TE/ER 260.50 575.00 27032.00 3.54 42.00 2.615 I/
H
11800.00
14 Kishau U/C Tons Dehradun PG 236.00 680.00 9500.00 1.81 21.70 1.33 I/
H
23019.00
15 Lakhwar U/C Yamuna Dehradun PG 204.00 451.78 2800.00 0.58 29.50 0.33 I/
H
8000.00
16 Vyasi U/C Yamuna Dehradun PG 88.00 203.00 2250.00 1.16 9.65 I/
H
8000.00
References
[1] A.A. Balkema, Rotterdam, a text book on Numerical Methods of Engineering, Theory and
Applications, January 1985.
[2] A K Chopra, H Tan, Modeling of Dam-Foundation Interaction in Analysis of Arch Dams,
Earthquake Engineering, The 10
th
World Conference on Earthquake Engineering, 1992.
[3] Abdolrahim Jalali, Tatsuo Ohmachi, Aspects of Concrete Dams Response to Near-Field
Ground Motions, The 12
th
World Conference on Earthquake Engineering, 2000.
[4] Anil K Chopra, P Chakrabarthi, Earthquake Analysis of Concrete Gravity Dams Including
Dam-Water-Foundation Interaction, Earthquake Engineering and Structural Dynamics, Volume
9, pp. 363-383, 1981.
[5] Chopra A K, Zhang L, Base Sliding Response of Concrete Gravity Dams Due to Earthquakes,
Report No. UCB/EERC-91/05, Earthquake Engineering Research Center, University of California,
Berkeley, California, 1991.
[6] D K Paul, Seismic Cracking of Tehri Dam, Proceedings of 12
th
Symposium on Earthquake
Engineering, Volume 1, pp. 979-990, December 2002.
[7] Dam Safety Organization, Central Water Commission National Register of Large dams,
2009
[8] Douglas Dreger, Gabriel Hurtado, Anil K. Chopra, Shawn Larsen, Near-Fault Seismic Ground
Motions, Earthquake Engineering Research Center, College of Engineering, University of
California, Berkeley, Report No. EERC 2007-08, October 2007.
[9] Earthquake facts, http://earthquake.usgs.gov/learn/facts.php, United States Geological
Survey (USGS).
[10] Erol Kalkan, Sashi K. Kunnath, Effects of Fling Step and Forward Directivity on Seismic
Response of Buildings, Earthquake Spectra, Volume 22, No.2, pp. 367-390, 2006
[11] Javad Moradloo, Mohammad Taghi Ahmadi, Shahram Vahdani, Non-Linear Dynamic
Analysis of Concrete Arch Dam, The 14
th
World Conference on Earthquake Engineering,
October 2008.
[12] Jean Proulx, Patrick Paultre., Experimental and Numerical Investigation of Dam-Reservoir-
Foundation Interaction for a Large Gravity Dam, Canadian Journal of Civil Engineering, 24, pp.
90-105, 1997.
[13] Jr. William Weaver, James M. Gere, a text book on Matrix Analysis of Framed Structures,
1966
[14] Kamalesh Kumar, a text book on Basic Geo-technical Earthquake Engineering, 2008.
[15] Larry K Nuss, Norihisa Matsumoto, Kenneth D Hansen, Shaken but not Stirred
Earthquake Performance of Concrete Dams, USSD Proceedings, 2012.
[16] M Emin Emiroglu, Influences on Selection of Type of Dam, International Journal of Science
and Technology, Volume 3, Number 2, pp. 173-189, 2008.
[17] Mohammed Ahmed Hussain, Study of The Effects on The Ground Surface Subjected to
Seismic Base Fault Motion, PhD Thesis (2012), International Institute of information
Technology, Hyderabad, India.
[18] Paul G Somerville, Engineering Characterization of Near-Fault Ground Motions, NZSEE
conference, 2005.
[19] Petter E Skrikerud, Hugo Bachmann, Discrete Crack modeling for Dynamically Loaded
Unreinforced Concrete Structures, Earthquake Engineering and Structural Dynamics, Volume
14, pp. 297-315, 1996.
[20] Rajib Sarkar, D K Paul, L Stempniewski, Influence of Reervoir and Foundation on the Non-
linear Dynamic Response of Concrete Gravity Dam, ISET Journal of Earthquake technology,
Volume 44, Number 2, pp. 377-389, June 2007.
[21] R Pradeep Kumar, K Meguro, Applied Element Simulation of non-Linear Behaviour of Dip-
Slip Faults for Studying Ground Surface Deformations, Seismic Fault-induced Failures, pp. 109-
114, January 2001.
[22] S S Bhattacharjee, P Leger, Seismic Cracking and Energy Dissipation in Concrete Gravity
Dams, Earthquake Engineering and Structural Dynamics, Voulume 22, pp. 991-1007, 1993.
[23] Semih Kucukarslan, Dam-Reservoir Interaction Including The Reservoir Bottom Effects in
Time Domain, 16
th
ASCE Engineering Mechanics Conference, July 2003.
[24] Stevel L Kramer, a text book on Geo-technical Earthquake Engineering, 2008.
[25] Tagel-Din-Hatem, A New Efficient Method for Nonlinear, Large Deformation and Collapse
analysis of Structures, PhD Thesis (1998), The University of Tokyo, Japan.
[26] T K Dutta, a text book on Seismic Analysis of Structures, 2010.
[27] Tatsuo Ohmachi, Abdolrahim Jalali., Fundamental Study on Near-Field Effects on
Earthquake Response of Arch Dams, Earthquake Engineering and Engineering Seismology,
Volume 1, Number 1, pp. 1 11, September 1999.
[28] Tatsuo Ohmachi, Naoyuki Kojima, Atsushi Murakami, Nobuhiko Komaba, Near-Field
Effects of Hidden Seismic Faulting on a Concrete Dam, Journal of Natural Disaster Science,
Volume 25, Number 1, pp. 7-15, 2003.
[29] Vahid Lotif, Dynamic Analysis of Concrete Arch Dams based on FE_BE Procedure Including
Non-uniform Ground Motion, Dam Engineering, Volume 25, issue 1, pp. 23-50, September 2004
[30] Worakanchana Kawin, Kimiro Meguro, Failure Mechanism of Shih-Kang Dam by Applied
Element Method, New Technologies for Urban Safety of Mega Cities in Asia, pp.119-128,
October 2005.
[31] William P Donlon, Experimental investigation of the nonlinear seismic response of
concrete gravity dams, PhD Thesis (1989), California Institute of Technology, California, USA.
[32] Williams J.R, Hocking G, and Mustoe G.G.W, a text book on The Theoretical Basis of the
Discrete Element Method, NUMETA 1985.

Numerical Study of Near-Field Earthquake Effects on Concrete Gravity Dams

Transféré par

Informations du document

Description originale:

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Numerical Study of Near-Field Earthquake Effects on Concrete Gravity Dams

Transféré par

Droits d'auteur :

Formats disponibles

NUMERICAL STUDY OF NEAR-FIELD EARTHQUAKE EFFECTS ON

CONCRETE GRAVITY DAM

Vous aimerez peut-être aussi