0068 c26

2003 by CRC Press LLC
26
Dams and Appurtenant
Facilities
26.1 Introduction
26.2 Dams and Earthquakes
Overview Performance of Embankment Dams Performance
of Concrete Dams Performance of Spillways and Outlet Works
Dams and Faulting Reservoir-Triggered Seismicity
26.3 Seismic Vulnerability of Existing Dams
The Need for Vulnerability Assessment National Inventory of
Dams Seismic Vulnerability Ranking for Multiple Dams
26.4 Seismic Evaluation of Dams
Seismic Parameters for Dams Dam Analysis Parameters
Analysis of Embankment Dams Analysis of Concrete Dams
Analysis of Intake/Outlet Towers Limitations of Current
Analysis Methodologies Physical Testing, Modeling, and
Centrifuge Studies Post-Earthquake Inspection
26.5 Seismic Upgrade of Existing Dams
General Seismic Upgrade of Embankment Dams Seismic
Upgrade of Concrete Dams Seismic Upgrade of Appurtenant
Structures
26.6 Seismic Design of New Dams
General New Embankment Dams New Concrete Dams
New Appurtenant Structures
26.7 Seismic Instrumentation of Dams
Acknowledgments.
Dening Terms
References
26.1 Introduction
Dams and reservoirs located near urbanized areas represent a potential risk to the downstream population
and property in the event of uncontrolled release of the reservoir water due to earthquake damage. This
chapter reviews the seismic performance of existing dams, describes procedures for analysis and safety
evaluation, and briey summarizes design features that improve performance under earthquake loading.
In addition to the dam structure, the main structural and mechanical components of reservoir outlet
works must be investigated to assure safe release of reservoir water in case of emergency. It should be
noted that earthquake-triggered landslides may also affect the reservoir shoreline at some distance away
from the dam.
Gilles J. Bureau
Consulting Engineer
Piedmont, CA
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
26.2 Dams and Earthquakes
26.2.1 Overview
The rst failure of a dam due to earthquake reported in the literature is Augusta Dam, GA, during the
1886 Charleston, SC earthquake. Worldwide, fewer than 30 dams have failed completely during earth-
quakes [USCOLD, 2000]. These were primarily tailings or hydraulic ll dams, or relatively small embank-
ments of questionable design. Few large embankment or concrete dams have been severely damaged.
One gravity dam failed as a result of fault rupture across its foundation [USCOLD, 2000]. No arch dam
has ever suffered seismic damage that threatened the safe impoundment of its reservoir.
There are more than 75,000 dams of all sizes listed in the U.S. National Inventory of Dams [U.S. Army
Corps of Engineers, 2000] and thousands of large dams have been built worldwide. Hence, the record
may appear outstanding. However, except for several well-known cases, few dams have been tested by
ground motion equivalent to their Design Basis Earthquake [USCOLD, 1999]. Conversely, a few dams
have experienced signicant damage under moderate shaking. Performance data and detailed references
regarding the approximately 400 dams that have been subjected to signicant earthquake shaking are
provided by USCOLD [1984, 1992b, 2000].
26.2.2 Performance of Embankment Dams
Embankment dams comprise rockll, earthll, hydraulic ll, or tailings dams. Their seismic performance
has been closely related to the nature and state of compaction of the ll material. Well-compacted modern
dams can withstand substantial earthquake shaking with no detrimental effects. In particular, earth dams
built of compacted clayey materials on competent foundations and rockll dams have demonstrated
excellent stability under extreme earthquake loading. In contrast, old embankments built of poorly
compacted sands and silts or founded on loose alluvium, hydraulic ll dams, and tailings dams represent
nearly all the known cases of failures. The following paragraphs summarize the experience and lessons
learned from the most notable case histories.
26.2.2.1 1906: San Francisco earthquake (M 8.3, estimated)
This event affected about 30 medium-sized earthll dams within 50 km of the fault rupture trace (15 of
these were less than 5 km away). Most survived the shaking with only minor damage. This satisfactory
performance demonstrated the ability of clayey dams to withstand extreme seismic loading, despite the
questionable methods of compaction used for these historic facilities.
26.2.2.2 1925: Santa Barbara earthquake (M 6.3)
This earthquake caused catastrophic slope sliding failure of the 25-ft-high Shefeld Dam in Santa Barbara,
CA. This was the rst recognition that shaking of embankments with low relative density materials may
cause liquefaction failures.
26.2.2.3 1971: San Fernando earthquake (M 6.5)
Engineers concerns regarding the vulnerability of dams constructed of poorly compacted, saturated ne
sands and silts were conrmed in 1971. The Lower Van Norman Dam, a 140-ft-high hydraulic ll dam,
experienced widespread liquefaction and major slope failures, as shown in Figure 26.1. Flooding of the
downstream area with its 70,000 residents was barely avoided, due to an unusually low reservoir level.
The 80-ft-high Upper Van Norman Dam was also severely damaged. This experience triggered numerous
reassessments of other dams and led to the development of modern numerical methods of dynamic
analysis of dams. Following that earthquake, questionable or unsafe embankments in California were
upgraded or decommissioned, or owners were mandated to operate the reservoirs at restricted levels.
26.2.2.4 1985: Mexico earthquake (M 8.1)
Two large dams, La Villita (197 ft high) and El Inernillo (485 ft high) were affected. Although neither
experienced signicant damage, these dams were shaken from 1975 to 1985 by a string of closely spaced
seismic events, ve of magnitudes greater than 7.1. Cumulative earthquake-induced settlements of La
Villita Dam, an earth-rockll embankment with a wide, central clayey core, approach 2 ft and have
increased in the latest events, perhaps due to progressive weakening of some of the materials. The
deformations of El Inernillo Dam, an earth-core rockll dam, have remained small and consistent from
one event to the next.
26.2.2.5 1989: Loma Prieta earthquake (M 7.1)
A wide region around the San Francisco Bay was affected. About 100 embankment dams of various sizes
were within 100 km of the epicenter, including most of those previously shaken by the 1906 earthquake.
The bilateral fault rupture propagation reduced the duration of the strong phase of shaking of the 1989
event to about 10 sec. Given the season, most of the reservoirs were only lled to between 10 and 50%
of maximum capacity, and all but one dam performed well. Austrian Dam, a 200-ft-high earth dam
about 12 km from the epicenter, suffered substantial transverse abutment cracking and settled nearly
3 ft. The reservoir was half full at the time of the earthquake. Overall damage to the dam was extensive,
considering the short duration of shaking. Austrian and other dams affected by the Loma Prieta earth-
quake must be capable of withstanding earthquakes considerably more demanding in intensity and
duration of shaking than experienced in 1989.
26.2.2.6 1990: Philippines earthquake (M 7.7)
Five large earth and rockll dams were located between 1.5 and 12.5 miles from the fault rupture trace.
Ground motion was estimated at these sites at between 0.35 and 0.70 g. None of the dams failed but they
all experienced settlement, deformations, and cracking. One of the dams, Diayo Dam, 197 ft high,
experienced a major slump along the total length (660 ft) of its upstream slope. The scarp of that slump
was about 1 ft high on the downslope side.
26.2.2.7 1994: Northridge earthquake (M 6.7)
The hypocenter was centered about 32 km west-northwest of the San Fernando Valley. This earthquake
was signicant for two reasons: (1) it reemphasized the seismic hazard associated with blind thrust faults
FIGURE 26.1 Damaged Lower San Fernando Dam, 1971 San Fernando earthquake, looking west. Shown as Color
Figure 26.1.
in California and (2) it was the second signicant event to affect the San Fernando Valley in less than
25 years. More than 100 dams were located within 75 km of its epicenter, including most of those shaken
in 1971. Eleven earth and rockll dams experienced cracking and slope movements but none threatened
life and property. The 125-ft-high Lower Van Norman Dam (repaired since its damage in the 1971 San
Fernando event, discussed previously) again suffered noticeable damage. Since 1971, the dam had been
operated for ood control with an empty reservoir. It experienced longitudinal cracks several hundred
feet long, up to 3.5 in. wide and 5 ft deep, and sand boils and a sinkhole along the upstream face. At the
crest, maximum settlement was 8 in. and maximum horizontal movement about 4 in. upstream. The
82-ft-high Upper Van Norman Dam (also operated since 1971 with an empty reservoir) experienced
transverse cracks near its abutments and along the downstream slope. These cracks were up to 60 ft long
and 3 in. wide. Maximum crest settlement was about 2.4 ft, with over 6 in. of horizontal upstream
movement. Between the Van Norman dams, and replacing them as a water supply facility, was the new
130-ft-high Los Angeles Dam. Ground shaking was very strong at the site (0.42 g peak ground acceleration
[PGA] at the dam left abutment and 0.85 g at an instrument 4400 ft away). However, the dam showed
only minor deformations and supercial cracking of the concrete lining of the reservoir. The crest moved
2.2 in. horizontally and settled 3.5 in. at the maximum section. Los Angeles Dam, constructed in 1977
to demanding seismic requirements, withstood the Northridge earthquake. In contrast, the Van Norman
dams, designed and built of hydraulic ll in 1915, suffered major damage both in 1971 and in 1994.
26.2.2.8 1999: Kocaeli earthquake, Turkey (M
W
7.4)
Gokce Dam, a 200-ft-high earth core rockll dam and Kirazdere Dam, a 356-ft-high earthll embankment
with clay core, sand and gravel lters, and rockll shells are located within the area of strong damage. The
only observed effect at Gokce Dam was a longitudinal crack along the upstream side of the crest, about
0.33 in. wide. Kirazdere Dam is located within 2 to 3 km of the epicenter and in close proximity to the
causative fault, the North Anatolian Fault. There was about 7 ft of right-lateral movement within about
1 mi of the dam. A few longitudinal cracks, each about 0.1 in. wide, occurred on the crest gravel road.
Overall, both dams performed satisfactorily and demonstrated the high seismic resistance of rockll dams.
26.2.2.9 2001: Bhuj (Gujarat), India earthquake (M
W
7.7)
This event resulted in widespread soil effects and liquefaction in low-lying estuaries and young alluvial
deposits. Strong ground motion lasted more than 85 sec, and lower-level shaking several min. Numerous
embankment dams were damaged in the epicentral area, including seven medium-sized (40 to 120 ft
high) earth dams (Rudramata, Niruna, Sasoi, Fategadh, Suvi, Kaswati, and Tapar). Fourteen smaller dams
were also damaged, some extensively. The reservoirs were very low at the time of the earthquake but
liquefaction of the foundation caused moderate to severe failure of the upstream and, locally, the down-
stream slopes of the dams.
26.2.3 Performance of Concrete Dams
Concrete dam designs include cylindrical arch, thin arch with double curvature, multiple arch, gravity,
buttress, hollow gravity, and combinations of these types, or composite embankment and concrete
structures. Overall, the earthquake performance of concrete dams has been satisfactory, and thus can be
inferred to indicate that they are more earthquake-resistant than embankment dams. Most concrete dams
have been built to design standards higher than some embankment dams, and are less susceptible to
aging, deterioration, seepage, and poor maintenance. However, only about 100 concrete dams have been
signicantly shaken by earthquakes, with only about 15 of these experiencing PGA greater than 0.20 g.
No signicant damage has ever been suffered by an arch dam, although several have experienced sub-
stantial ground motion. However, the true test of a nearby major earthquake, shaking a large thin arch
with the reservoir at full capacity, has yet to occur. Concrete dams are prone to major damage in cases
of fault movement across their foundation. The only reported case of complete failure is Shih-Kang Dam,
a concrete buttress gravity dam intersected by the surface rupture of the 1999 Chi-Chi, Taiwan earthquake
(M
W
7.6). Earthquake experience with concrete dams include those described below.
26.2.3.1 1906: San Francisco earthquake (M 8.3, estimated)
Lower Crystal Springs Dam, a 127-ft-high, curved-concrete gravity dam built of interlocking blocks,
withstood that earthquake without a single crack. The primary fault rupture, with a local right-lateral
slip of about 10 ft, passed less than 600 ft from the dam.
26.2.3.2 1957: Great Britain earthquake (Intensity VIII, British Scale)
This rare event was centered about 6.4 km from Blackbrook Dam, a 100-ft-high gravity dam with
upstream brick and downstream stone facings. This is the only dam in Great Britain damaged by an
earthquake. The mortar of the stone facing was cracked. All of the large coping stones topping the parapet
walls on both sides of the crest were lifted from their mortar bed and dropped back, crushing it in the
process.
26.2.3.3 1962: Kwangdong, China earthquake (M 6.1)
Hsinfengkiang Dam, a 344-ft-high buttress dam with abutment gravity sections, was completed in 1959.
Earthquake swarms occurred after the rst lling of the reservoir, and the dam was strengthened to
increase its seismic capacity. The 1962 earthquake induced shaking stronger than considered for the
modied design. Cracks formed near the crest, at a change in section of the nonoverow blocks on each
side of the spillway. The earthquakes were considered to be possible cases of reservoir-triggered seismicity.
The dam was strengthened again, and placed back in service.
26.2.3.4 1967: Koyna, India earthquake (M 6.5)
Koyna Dam, a 338-ft-high straight gravity dam built of rubble concrete in 1963, was shaken by a nearby
earthquake of magnitude 6.5. This earthquake was also suspected to be due to reservoir-triggered seismicity.
Koyna Dam developed substantial longitudinal cracking near the top, at the location of a sharp change of
slope of the downstream face, used to increase height and reservoir capacity. Another design weakness
was the use of varying concrete strengths based on static analysis and decreasing from bottom to top,
which could not accommodate large dynamic stresses that occurred in the upper portion of the dam.
Similar design features are avoided in modern structures. Koyna Dam was repaired and is still in service.
26.2.3.5 1971: San Fernando earthquake (M 6.5)
During that earthquake, the 372-ft-high Pacoima Dam, a thick arch, was subjected to an estimated peak
base acceleration (PBA) of about 0.70 g. An unprecedented horizontal acceleration of 1.25 g was recorded
on rock at the left abutment, slightly above the dam crest. The dam was used for ood control and the
reservoir level was at about mid-height. Neither cracks nor relative block movements were reported. The
left abutment was strengthened with post-tensioned anchors to stabilize two large rock wedges that moved
several inches during the earthquake. The large accelerations and rock wedge movements in the left
abutment illustrate ground motion amplications (ridge) effects caused by a peculiar topographic con-
guration.
26.2.3.6 1990: Manjil, Iran earthquake (M 7.6)
Sed Rud Dam, a 348-ft-high gravity buttress dam built in 1962, is less than 10 mi from the epicenter.
PGA was estimated at 0.72 g. The seven gravity monoliths and 23 head buttress units of the dam structure
were designed with pseudostatic horizontal coefcients up to 0.25 g. The dam suffered severe cracking
at lift joints in the upper part of the buttresses, accompanied by a 20-mm shear displacement toward
downstream. Leakage occurred through the cracks and lowered the reservoir. Sed Rud Dam represents
an example of a concrete dam subjected to shaking substantially more severe than its design loads. The
dam suffered signicant damage but had overall satisfactory performance. It was repaired using post-
tensioned anchors.
26.2.3.7 1994: Northridge earthquake (M 6.7)
Pacoima Dam was strongly shaken by this earthquake 23 years after the San Fernando earthquake. Peak
accelerations at the top of the left abutment were again amplied by the local narrow ridge topography
and shattered condition of the rock, and reached 1.76 g horizontal and 1.60 g vertical. The dam performed
satisfactorily with a half-full reservoir. For the rst time, this earthquake provided evidence of the opening
and closing of vertical contraction joints. The joint between the left abutment and the left end of the
arch opened a maximum of 2 in., due to movement of the rock wedges on the upper abutment. Minor
horizontal cracking of concrete at the left end of the dam, and several minor horizontal and vertical
block offsets occurred at the joints. The rock mass at the left abutment that had been stabilized after the
San Fernando earthquake did not fail but several of the anchors were overstressed.
26.2.3.8 1999: Chi-Chi, Taiwan earthquake (M
W
7.6)
Shih-Kang Dam, a large gravity, radial-gated, water supply dam, is 50 km from the epicenter. It failed
due to differential thrust fault movement at its north abutment. Over two thirds of the dam body were
uplifted about 29 ft vertically, and displaced 6.5 ft horizontally (Figure 26.2). Damage was conned to
the two bays overlying the fault rupture. The sixteen other bays were essentially intact. The dam expe-
rienced horizontal accelerations up to 0.50 g, and its performance would have been excellent had it been
located outside of the rupture trace. The reservoir slowly drained through the failed bays, without causing
major ooding. Another concrete gravity dam with radial gates, the Chi-Chi Diversion Dam, is less than
10 km from the epicenter and experienced no damage.
26.2.4 Performance of Spillways and Outlet Works
Earthquake reconnaissance reports have often focused more on the dams than on their appurtenant
facilities. However, spillways and outlet works are also exposed to potential earthquake damage. Outlet
works may lose functionality at a critical time, due to possible loss of power supply. Spillway damage is
rarely critical, as the probability of experiencing a signicant earthquake during or immediately before
major ooding is extremely low. Tall intake or outlet towers and their equipment, however, are particularly
vulnerable. The 56-ft-high, 33-ft-wide hoist tower at Koyna Dam suffered considerable damage during
the 1967 earthquake, and the precast concrete blocks and the main reinforced concrete framework cracked
in many places [Gupta and Rastogi, 1976]. At the Van Norman dams, two of three tall outlet towers were
lost during the 1971 San Fernando earthquake, including one complete collapse (Figure 26.3). Emergency
lowering of the reservoir required pumps to be brought to the site. During the 1995 Hyogo-Ken Nanbu,
Japan earthquake (M
JMA
7.2), the foundation of the intake tower controlling the lower pool level at Koyoen
Reservoir moved substantially, tilting the tower. The intake remained functional but the access bridge of
FIGURE 26.2 Damage to Shih-Kang Dam due to fault offset, 1999 Chi-Chi, Taiwan earthquake.
that structure was shoved through the access door, illustrating how intake/outlet towers and their access
bridges may respond differently to the shaking, and pound against each other or fail at supports and
connections.
26.2.5 Dams and Faulting
Many dams are built across streams or rivers that follow existing fault traces. While such faults are not
necessarily active, the potential for differential movement across the dam foundation must be taken
seriously and investigated. Sherard and coworkers [1974] discussed at length the problem of active faults
across dam foundations. Two of these authors recently updated their 1974 paper on the same subject
[Allen and Cluff, 2000]. The 88-ft-high San Andreas Dam, composed of a main and saddle dam, was
nearly intersected by the 1906 rupture of the San Andreas Fault, which passed between the two embank-
ments but did not cause damage. The fault rupture actually offset the submerged embankment of the
Old San Andreas Dam, which, at the time of the earthquake, was abandoned in the north arm of the
reservoir. Another example was Hebgen Dam, MT, during the West Yellowstone earthquake of 1959
(M 7.1). The fault rupture trace passed about 600 ft from the dam right abutment, and displayed about
16 ft of vertical upward movement. Simultaneously, the bedrock underlying the entire dam, spillway, and
FIGURE 26.3 Failed outlet tower, Lower Van Norman Dam, 1971 San Fernando earthquake. Shown as Color
Figure 26.3.
outlet conduit subsided uniformly by about 10 ft. The faulting did not affect Hebgen Dam, which
experienced only moderate slumping and cracking. However, contrary to the two near-misses cited here,
the example of Shih-Kang Dam (Section 26.2.3.8) reminds us what fault movement can do to a dam.
Building a dam across an active fault should be avoided, if possible. In 1975, the Oroville, CA
earthquake (M 5.7) ruptured the Cleveland Hill Fault, part of the Sierra Nevada Foothills Fault System
then considered inactive. Subsequently, the U.S. Bureau of Reclamation (USBR) prudently abandoned
its plans to construct Auburn Dam, a proposed 685-ft-high, double curvature thin arch, as faults of the
same system were encountered during foundation excavation. This decision was made although detailed
dynamic analyses showed that the dam would be undamaged in an M 6.5 local event.
Recognizing the existence of an active or capable fault across the proposed alignment of a dam requires
drastic measures and, preferably, abandoning the site, as USBR did in the case of Auburn Dam. However,
an alternative site may not be available and a conservatively designed dam may need to be considered.
The decision to proceed with construction of a dam across an active fault represents one of the most
severe design challenges but has been made in a few cases. First, the rupture mechanism and amplitude
of relative displacements must be rigorously assessed. Then, special features must be included in the
design to accommodate such displacements. An example is Coyote Dam in California, which was built
in 1936 across the main trace of the Calaveras Fault. The fault was then recognized as a main active
seismogenic feature, capable of about 3 ft of vertical movement and nearly 20 ft of lateral displacement.
A decision was made to construct a dam at the site, with the use of special design features, such as an
oversize, compacted clayey core with transition layers and extra freeboard (22 ft, for a dam height of
125 ft). Another case is Cedar Springs Dam, 215 ft high, completed in 1971 after exploratory trenches
in the foundation area showed the presence of several faults with evidence of Holocene activity (less than
11,000 years old). The original design of the dam was modied after the discovery of the faults. The dam
height was reduced nearly one third compared with the initial plans, and the dam section was completely
redesigned to include thick rockll exterior shells and wide transition zones of well-graded coarse sand-
gravel mixture protecting the enlarged clay core. All of the core materials were imported from a distant
borrow area to obtain better quality materials than available onsite. In another example, several poten-
tially active branches of the Dunstan Fault were discovered when excavating the foundation of the
335-ft-high Clyde Dam, New Zealand, a concrete gravity dam [Hatton et al., 1991]. The design of the
dam was modied to include a slip joint directly above the fault, with a rubber-sealed, steel-sheathed
wedge plug, 328 ft long. The joint was designed to accommodate 6.5 ft of strike-slip and 3.3 ft of dip-slip
movements. A few other examples of dams built across active faults are described in Sherard and
coworkers [1974] and Allen and Cluff [2000].
Another type of challenge is encountered when it is discovered that an existing dam was unknowingly
built on an active fault. This was the case with the 240-ft-high Matahina Dam, New Zealand, constructed
in 1960 on recently recognized active splays of the Waiohau Fault, a major tectonic feature of the Alpine
fault system and one of the longest and most active regional strike-slip faults. Various alternatives were
considered to improve the safety of the dam, and the selected design consisted of placing a new thick
lter, transition zones, and rockll shell on the downstream face of the existing dam [Mejia et al., 1997].
This design will allow the lter zone to heal any potential cracking induced by the fault rupture, thereby
maintaining the integrity of the reservoir. Additional information on features that improve the seismic
performance of dams is presented in Section 26.6. In addition to the risk of sudden fault rupture, fault
creep may also cause distress within a dam. Fault creep is a gradual, continuous, relative displacement
that occurs generally at a low rate of slip. Fault creep has been observed across Bajina Basta and Lipovica
dams in Yugoslavia [Bozovic and Markovic, 1999] but is considered to be a sufciently rare phenomenon
that can usually be dismissed in seismic studies for dam projects [Allen and Cluff, 2000].
26.2.6 Reservoir-Triggered Seismicity
The creation of articial lakes formed by large dams received considerable attention in the 1960s and
1970s as a potential source of seismicity, caused by lling of the reservoirs. The earthquakes that affected
Hsinfengkiang Dam, China (1962) and Koyna Dam, India (1967) were believed to be clear cases of
reservoir-induced seismicity. More than 30 other examples are known where the impounding of large
reservoirs may have initiated or enhanced seismic activity [Gupta and Rastogi, 1976; USCOLD, 1986,
1997]. The most recognized of these other dams include Aswan, Egypt; Hoover (Lake Mead), NV and
AZ; Kariba, Zambia; Kremasta, Greece; Monteynard, France; Nurek, Russia; Oroville, CA; and Vajont,
Italy. Reservoir-induced seismicity has been potentially associated with incremental effective stressing
and warping of the Earths crust due to the reservoir loading. However, although all of the reservoirs
were quite deep (typically 300 ft or more), stress and pore pressure increases induced by reservoir loading
are considerably less than stress changes associated with earthquakes.
Models relating pore pressure increases to the mechanics of rock fractures were also developed over
the years to explain the relationship between earthquakes and dams but obtaining reliable information
regarding the initial and modied state of stress at the depth of several kilometers where earthquakes
occur has not been practical. While experts still agree that lling a reservoir modies the stress regime
within crustal layers and reduces the effective shear strength of the rock mass, it is now generally agreed
that such changes cannot cause fractures in intact rock. However, it is now believed that preexisting
faulted rock with a high in situ state of stress can be brought to slip by the reservoir impoundment.
Furthermore, most reservoir-induced seismic events have occurred in areas affected by Quaternary
faulting. Hence, earthquakes were most likely triggered rather than induced by the reservoir, and such
terminology is now considered to be more appropriate [USCOLD, 1997]. While the possibility of
reservoir-triggered earthquakes should be considered for any reservoir deeper than 250 to 300 ft, expe-
rience suggests that the maximum reservoir-triggered earthquake should not exceed the design earth-
quake that must otherwise be specied for any site located within an area of recognized potential seismic
activity. Furthermore, the likelihood of a reservoir-triggered earthquake being associated with signicant
surface fault displacement is exceedingly remote [Allen and Cluff, 2000].
26.3 Seismic Vulnerability of Existing Dams
26.3.1 The Need for Vulnerability Assessment
Dam owners and regulators must ensure that dams are safely operated and present no risk to the public
in case of an earthquake. While most existing or new dams in recognized seismic regions have been
evaluated and analyzed for seismic loads (see Section 26.4), dams located in areas of moderate or
infrequent seismicity have been given less systematic attention. In such cases, owners of many dams or
ofcials in charge of dam safety programs may consider comparative assessment of the seismic risk
associated with their dams and establish priorities, as needed. A methodology to perform such tasks was
developed as part of a general earthquake risk and loss estimation program for the State of South Carolina
[URS Corporation et al., 2002] and can be applied to any state using the National Inventory of Dams
(NID) and the concepts described in Section 26.3.3.
26.3.2 National Inventory of Dams
In the United States, various sources of information can be consulted to obtain information on dams,
including (1) the National Inventory of Dams (NID) and (2) federal, state, or local agencies that have
jurisdiction over dams in a given area. The NID is maintained and periodically updated by the U.S. Army
Corps of Engineers (COE), under legislation enacted by Congress in 1986 as the Water Resources
Development Act (P.L. 99-662). The NID was implemented in 1989 and has been updated several times.
The COE has prime responsibility for maintenance and update of the NID and has been working closely
with other involved federal and state agencies. The NID is accessed from an Internet site (http://
www.tec.army.mil/nid/index.html) or from a CD-ROM. Its current version includes 57 elds, which for
each dam describe name(s), type, purpose, year completed or modied, owner, location, dimensions,
reservoir storage capacity, hydraulics, downstream hazard, etc. The NID also provides information on
the availability of an emergency action plan (EAP), inspection requirements, and state or federal agencies
having jurisdiction on the facility. The NID is an evolving database and information may change over
time, especially if the operation or conguration of any dam is modied.
26.3.3 Seismic Vulnerability Ranking for Multiple Dams
Various risk factors and weighting points can be used to approximately quantify the total risk factor
(TRF) of any dam [Bureau and Ballentine, 2002]. The TRF depends on the dam type, age, size, the
downstream risk potential, and the dam vulnerability, which depends on the seismic hazard of the site.
The TRF is expressed as:
TRF = [(CRF + HRF + ARF) + DHF] PDF (26.1)
The dam structure inuence is represented by the sum (CRF + HRF + ARF) of capacity, height, and
age risk factors. The downstream hazard factor (DHF) is based on population and property at risk. The
vulnerability rating is a function of the site-dependent seismic hazard and observed performance of
similar dams, as dened by a predicted damage factor (PDF). Additional information regarding these
various risk factors is provided in the following sections.
Structure Inuence
Three factors quantify the risk of a dam and its reservoir. The capacity risk factor (CRF) and the height
risk factor (HRF) indicate that high dams or large reservoirs can cause signicant ooding. The age
rating factor (ARF) expresses that old dams are often more vulnerable than modern dams because of
possible deterioration, lack of maintenance, use of obsolete modes of construction (concrete masonry
or hydraulic ll), insufcient compaction, reservoir siltation, or insufcient foundation treatment. The
dam risk factors are dened in Tables 26.1 and 26.2.
Downstream Risk
The overall downstream hazard factor (DHF) is dened as:
DHF = ERF + DRI (26.2)
The downstream evacuation requirements factor (ERF) depends on the human population at risk.
The downstream damage risk index (DRI) is based on the value of private, commercial, industrial, or
government property in the potential ood path. These factors should preferably be obtained from a
combination of detailed dam breach, inundation mapping, and economic studies. The DHF should be
updated whenever new information becomes available or when the dam is repaired, modied, or raised
(Table 26.3).
When it is not cost-effective to obtain the ERF and DRI from detailed studies, the downstream hazard
potential rating of the NID can be used to obtain a substitute value of the DHF, as shown in Table 26.4.
TABLE 26.1 Denition of Dam Size Risk Factors
Contribution to Total Risk Factor (weighting points)
Risk Factor Extreme High Moderate Low
Capacity (acre-feet) (CRF) >50,000 (6) 50,0001,000 (4) 1,000100 (2) <100 (0)
Height (feet) (HRF) >80 (6) 8040 (4) 4020 (2) <20 (0)
TABLE 26.2 Denition of Dam Age Rating Factors
Dam Age <1900 19001925 19251950 19501975 19752000 >2000
ARF 6 5 4 3 2 1
Seismic Vulnerability Rating
Dam vulnerability curves developed by Bureau and Ballentine [2002] from observed seismic performance
of dams during earthquakes [USCOLD, 1992b, 2000] can be used to compute a predicted damage index
(PDI). The PDI depends on the dam type and on the site seismic hazard and tectonic environment. The
expected ground motion at the dam site for the scenario earthquake considered is expressed by the
earthquake severity index (ESI), a robust estimate of the severity of shaking for dam evaluation purposes
[Bureau et al., 1985]. The ESI is expressed as:
ESI = PGA (M 4.5)
3
(26.3)
where the PGA is measured in g, and M is the Richter or moment magnitude (M
W
, if available) of the
causative event. The PDI depends on the ESI at each dam site, for each postulated earthquake scenario,
and is obtained from graphical relationships shown in Figure 26.4. As is well known, hydraulic ll and
tailings dams are clearly the most severely affected, based on historic experience. Arch dams have
performed best but the corresponding data are limited.
Several comments are appropriate regarding Figure 26.4. First and foremost, the relationships plotted
do not predict failure or nonfailure of any specic dam for a given ESI. This requires detailed site-specic
geologic, geotechnical, and seismic response studies. The vulnerability curves are intended only to be
applied to a large family of dams and help identify the potentially most critical facilities in regional
studies. The PDI rates only the relative vulnerability of each dam type, and includes a signicant
uncertainty, especially when extrapolated to large ESI values, which can be quantied from the standard
deviations associated with the mean estimates.
It should be recognized that many other important factors may affect the present or future risk
associated with existing dams: (1) availability or lack of construction and maintenance records,
(2) availability or lack of instrumentation data and surveillance records, (3) level of effort expended in
safety evaluations, (4) planned upgrade of the dam, (5) planned enlargement of the reservoir, and
(6) planned downstream developments.
From the computed PDI, a PDF is assigned to each dam, as dened by the following equation:
PDF = 2.5 PDI (26.4)
Bureau and Ballentine [2002] empirically selected the coefcient 2.5 to provide consistency between
vulnerability estimates obtained from site-specic ground motion estimates dened by the ESI or, when
TABLE 26.3 Denition of Downstream Risk Factors
Contribution to Total Risk Factor (weighting points)
Risk Factor Extreme High Moderate Low
Evacuation requirements (ERF) (persons) >1,000 (12) 1,000100 (8) 1001 (4) None (1)
Downstream damage risk index (DRI) High (12) Moderate (8) Low (4) None (1)
TABLE 26.4 Downstream Hazard Factor Based on NID
NIDs Downstream Hazard
Potential Rating Loss of Human Lives
Economic, Environmental, or
Lifeline Losses
Downstream Hazard
Factor (DHF)
Low None expected Low, generally limited to owners
property
2
Signicant None expected Yes 12
High Likely, one or more
expected
Yes or probable but not strictly
required
24
such estimates were not available, from a combination of a dam type factor and a seismic zone factor
derived from the Uniform Building Code Zone Factor.
The last step of the assessment is to rank the dams by decreasing TRF and assign to each a Risk Class
ranging from I (low risk) to IV (extreme risk), as shown in Table 26.5. The vulnerability ranking of over
4500 dams, including all dams in South Carolina plus some in neighboring states, was compiled in that
fashion [URS Corporation et al., 2002].
The risk class can be used to establish the need for more detailed seismic safety evaluations and to
establish priorities for such evaluations. The procedure presented in this section can be used to quickly
assess the potentially most vulnerable facilities in a large dam inventory. The risk classication based on
the TRF provides guidance to dam safety ofcials to select appropriate evaluation procedures and to
assign priorities for seismic safety evaluations of the most critical dams.
26.4 Seismic Evaluation of Dams
Since the late 1960s, considerable progress has been made in the understanding of earthquake effects on
concrete or embankment dams. Better appraisals of the seismic response of dams have followed the rapid
development of computer-based analytical procedures and the installation of instruments for accurate
FIGURE 26.4 Predicted damage index (PDI).
TABLE 26.5 Denition of Dam Risk Class
Total Risk Factor (TRF) Dam Risk Class
2 to 25 I (low)
25 to 125 II (moderate)
125 to 250 III (high)
> 250 IV (extreme)
recording of earthquake motion at various locations on a dam and in its immediate vicinity. Considerable
progress has been made in the denition of the seismic input. In addition to the development of site-
specic seismic criteria, dam analysis requires denition of the geometry and conguration of the
structure and its foundation, characterization of the construction and foundation materials and their
static and dynamic properties, and inclusion of the effects of internal water pressures and the inuence
of the reservoir. In some cases, local site topographic features may need to be taken into consideration
in dam response calculations. For new dams, achieving an optimal conguration requires an iterative
process of adjusting the use of available construction materials and locating the dam within a site of a
given topography. The construction engineer and the analyst may have conicting objectives to resolve
issues when selecting a dam type and estimating the projects behavior when subjected to an earthquake.
The problem of seismic response and behavior of dams and their foundations, when formulated in
general terms, is extremely complex. One must always keep in mind that even the best analytical methods
do not provide precise numerical answers. They only provide insight into the design problems under
consideration.
Depending on the type and size of the dam, its seismic exposure, and other factors such as regulatory
requirements, numerical analysis may be conducted using simplied or complex procedures. Ideally, one
would incorporate three-dimensional interactive response of the dam body, the reservoir water, and the
dam foundation. For demanding seismic input or unusual conditions regarding dam or foundation
material properties or geometry, e.g., the possible presence of joints, it may be best to perform a nonlinear
response analysis of the damfoundationreservoir system, which is always a major undertaking. The
following sections describe input parameters and analysis procedures for seismic analysis of embankment
and concrete dams.
26.4.1 Seismic Parameters for Dams
The selection of site-dependent seismic input is essential to the safety evaluation of dams. The effort
required in such an endeavor increases with the complexity of the analyses to be performed. Detailed
information on this topic is provided in current guidelines for selecting seismic parameters for dam
projects [USCOLD, 1999]. The present discussion only briey describes the general requirements of each
type of analysis. It does not address how to select distance and magnitude parameters and frequency
content of the input motion, which is discussed elsewhere in this volume. Various earthquake levels
considered for the analysis of dams include the operating basis earthquake (OBE), the maximum credible
earthquake (MCE), the maximum design earthquake (MDE) and, occasionally, the reservoir-triggered
earthquake (RTE) [see USCOLD, 1999, for a complete denition of these terms]. Because the primary
requirement of earthquake-resistant design is to protect public safety and property, seismic criteria and
analysis parameters for dams, as for the evaluation of critical nuclear facilities, are often selected more
conservatively than for conventional structures.
26.4.1.1 Seismic Input for Embankment Dams
Simplied procedures for embankment dam analysis only require the peak ground acceleration (PGA)
and velocity (PGV) as input parameters [e.g., Newmark, 1965], or the PGA and the magnitude of the
causative event [Bureau et al., 1985]. Other simplied methods need a response spectrum and the
specied magnitude [Makdisi and Seed, 1977]. In all other cases, such as the classic (Newmarks) double-
integration method of dam evaluation and in equivalent-linear (EQL) or nonlinear (NL) detailed analyses,
one or several horizontal acceleration time histories must be specied, depending on whether two- or
three-dimensional analysis is considered. It is customary to use the largest component of horizontal
motion in the upstreamdownstream direction.
Various natural strong motion records are normally considered to select acceleration time histories
as seismic input. Selection criteria consist of magnitude and duration of the causative earthquake, mode
of fault rupture, distance, subsurface conditions at the recording station, and possible presence of near-
eld effects. A digitization interval of 0.02 sec is sufcient for embankment dam analysis.
In most cases, the inuence of the vertical motion has been considered negligible. However, this
conclusion was based on EQL response analyses of at-sloped earth dams, which are essentially unaffected
by that component of motion. This is not true in the case of steep dam faces, for example, as encountered
in rockll dams [Von Thun and Harris, 1981], or in the case of NL analysis. As NL response is stress-
path-dependent and because failure criteria depend on both normal and shear stresses, calculated per-
manent dynamic deformations will differ whether vertical excitation is included or not. Such differences
can be appreciable, up to about 20% of the maximum deformation response [Bureau, 1997]. Therefore,
the vertical component of motion must preferably be included in NL dam response analyses. This
component, especially, may dominate the seismic input in the case of near-eld thrust faulting.
For earthll dams, the duration of shaking is a most signicant parameter, as it governs the extent of
damage and buildup of excess pore pressures in the foundation or in the saturated portions of the
embankment. Hence, considerable care must be given to selecting time histories of appropriate duration
of the strong phase of shaking.
26.4.1.2 Seismic Input for Concrete Dams
Except for the use of a horizontal seismic coefcient in the case of the simplied analysis of gravity dams,
concrete dam analysis generally requires the use of one or several sets of horizontal and vertical acceler-
ation time histories. For sites of moderate seismic activity, horizontal and vertical response spectra may
be sufcient, as long as the allowable stresses are not exceeded. If that is not the case, acceleration time
histories are required to obtain stress time histories and assess how many cycles exceed the allowable
strength. As concrete dams respond at higher frequencies than embankment dams, it is customary to
use time histories digitized at 0.01-sec intervals. Concrete arch dam analysis requires the use of three
statistically independent components of motion. As for embankment dams, the largest horizontal com-
ponent of motion is typically applied in the upstreamdownstream direction.
26.4.1.3 Seismic Criteria for Appurtenant Structures
Spillway walls can be analyzed with horizontal load coefcients, such as used for retaining walls. However,
intake/outlet towers, which typically range from a few feet to more than 400 ft high, often respond within
a range of spectral amplications that dictates dynamic analysis. As failure of most towers would rarely
cause uncontrolled ooding, signicant damage might be acceptable for the MCE, and OBE-level seismic
criteria are sufcient if acceptable performance is demonstrated. Towers required for emergency draw-
down of the reservoir must be analyzed to the most demanding MCE or DBE requirements.
A building code methodology (such as the Uniform Building Code) can be used for towers located
in areas of moderate seismic hazard. However, this approach represents minimum seismic requirements.
The code is primarily intended for buildings, and the applicability of the R
w
factor to towers is question-
able, as it implies more ductile behavior than available in many of these often under-reinforced structures.
Detailed dynamic analysis is necessary for towers susceptible to amplifying ground motion signicantly.
In that case, seismic input is best described by horizontal and vertical response spectra.
26.4.1.4 Natural vs. Articial Acceleration Records
It has been common practice to scale and modify natural earthquake records to match a specied peak
acceleration and spectral content (see Figure 26.5). For earth or concrete dam analysis purposes, it is
desirable to minimize the modications of natural records, as the number of induced stress cycles is
more important to assess performance than the peak stress values. Current procedures for matching a
specied spectral shape consist of scaling a natural record to the specied PGA and selectively adjusting
the Fourier amplitude spectrum in the frequency domain, keeping the phase angles unchanged. Down-
ward rather than upward scaling is recommended. The process is iterative. Modied Fourier amplitudes
and original phase angles are then recombined in the time domain. The new acceleration history resem-
bles the original record but matches the targeted response spectral shape. Spectrum-compatible records
should preferably be baseline corrected to achieve zero velocity and displacement at the end of the
excitation. Without such correction, solutions based on absolute rather than relative displacements will
superimpose a global translation of the analysis model to the earthquake-induced deformations. This
does not affect the computed stresses, which depend on strains but the translation of the base of the
model must be subtracted from the calculated displacements.
26.4.1.5 Other Factors of Signicance
Fault mechanism and local source and path variability strongly inuence ground motion [Somerville
et al., 1995]. Frequently, horizontal shear wave velocity pulses near a rupturing fault have larger amplitudes
FIGURE 26.5 Spectrum-compatible acceleration time history.
normal to the fault than parallel to it. Similarly, ground motions are more severe on the up-thrown wall
of a reverse fault than on the stationary wall. Hence, the direction of arrival of ground motion pulses may
be signicant for dams located close to an active or capable fault. Except when records from an instru-
mented dam are used to back-calculate its observed response, input for dam analysis uses acceleration
histories recorded at other sites. Hence, there is no particular relationship between the dam alignment
and the fault rupture mechanism and direction of propagation associated with such records. The base
motion may be applied either as if it represents seismic waves traveling from upstream to downstream,
or the reverse. Hence, it is prudent to rst use a specied time history, then its opposite (all acceleration
values multiplied by 1). In NL analyses, calculated permanent displacements will differ when a horizontal
base motion is replaced by its opposite. Parametric analyses to dene the safe reservoir level of an existing
hydraulic ll dam have shown that maximum computed nonrecoverable deformations may more than
double locally when switching the input acceleration history to its opposite [URS Corporation, 2002].
Similarly, several records intended to represent a same earthquake scenario may lead to different defor-
mations. Therefore, using a single set of acceleration histories to represent a specied evaluation earthquake
is considered insufcient. Successive use of three to four sets of acceleration histories, plus their opposites,
is recommended.
For near-fault sites, the possible presence of large shear wave velocity pulses (ing) in the ground
motion, rst recognized by Bolt in 1971 [see Bolt, 1996], is of potential signicance [Somerville and
Graves, 1996]. Directivity and focusing effects can increase such pulses by a signicant factor [Bolt,
1996]. Dams built perpendicular to a strike-slip fault may experience less severe rupture directivity effects
than if parallel to such fault. Similarly, dams along a fault-controlled mountain front may experience
the largest components of motion transverse to their axis. Hence, an unfavorable direction of arrival of
the largest near-eld pulses may increase response and contribute to larger cumulative displacements or
a sharp increase in pore pressures. Such effects must be considered in the specication of the seismic
input for dams near active faults and, especially, large or loose embankments that may respond at periods
of vibration close to the period of those pulses. The alignment of a new dam can be optimized with
respect to critical paths of traveling seismic waves through a parametric analysis.
For analysis purposes, seismic input is normally taken as uniform along the dam footprint or at some
depth within the foundation, although phase characteristics and amplitudes vary both transversely and
longitudinally along the dam bottom. Such variations were observed at Los Angeles Dam during the 1994
Northridge earthquake [Davis and Sakado, 1994]. Abutment motions (free-eld) were more severe than
below the dam, and transverse components included more prominent near-eld pulses than the longitu-
dinal components. At this time, insufcient strong motion records at the base of dams and the lack of a
rigorous theoretical basis prevent analyzing embankment dams using nonuniform base excitation, a
potential shortcoming for large embankments. However, base shaking with input derived from free-eld
records, as typically used, should provide a response more conservative than motions recorded in depth.
26.4.1.6 Deterministic vs. Probabilistic Analysis
Depending on the location of a dam and which agencies have jurisdiction, deterministic or probabilistic
criteria have been used to develop seismic loads for dam safety evaluation. The Division of Safety of
Dams (DSOD) of the State of California requires jurisdictional dams to be evaluated for the MCE, which
represents the largest conceivable earthquake along a recognized capable or active fault or within a
recognized tectonic province. The largest magnitude and shortest distance to the fault are considered for
the MCE, and no consideration is given to its probability of occurrence. This deterministic approach
results in signicant variations of the risk associated with various dams, depending on whether the
controlling faults have a very high or very low rate of slip. Conversely, some agencies now use probabilistic
risk assessment procedures for dam safety management [U.S. Bureau of Reclamation (USBR), 1997].
When the deterministic approach is required, it may be desirable to perform an adjunct probabilistic
seismic hazard analysis of the site and compute the return period of the specied earthquake motion.
Then proper judgment may be used to select analysis parameters and avoid compounding multiple
excessively conservative assumptions.
26.4.2 Dam Analysis Parameters
26.4.2.1 Analysis Parameters for Embankment Dams
This section discusses soil and other properties required for the seismic analysis of embankment dams.
Field exploration and laboratory testing programs are essential to dene the physical and strength
properties of these dams and their foundations.
Static Parameters
Most analysis procedures require static properties (unit weight, moisture content, and total and effective
stress strength parameters). The effective static shear strength is essential to some NL dynamic analyses
[Dawson et al., 2001; Roth et al., 1991]. Finite element analyses used to dene the initial state of static
stresses often rely on hyperbolic soil models [Duncan et al., 1984] and variations of the initial tangent
static modulus E
i
with the conning pressure, as originally suggested by Janbu [1963]:
E
i
= K P
a
(/P
a
)
n
(26.5)
in which K is a constant, the minor principal stress, P
a
the atmospheric pressure, and n an exponent
dening the rate of variation of E
i
with . Depending on the type of analysis contemplated, some or all
of the following dynamic properties may be required: shear and bulk modulus, damping ratio, cyclic
shear strength, rate of pore pressure buildup, and residual shear strength.
Modulus and Damping Ratio
Strain-dependent equivalent dynamic shear moduli and damping ratios as rst introduced by Seed and
Idriss [1970a] (see Figure 26.6), are essential to EQL analyses. The dynamic shear modulus decreases
with the average induced shear strain, while damping increases. Strain-controlled cyclic triaxial tests and
resonant column tests may be used to obtain these parameters. Generic modulus degradation (G/G
max
)
and damping curves may be used when it is not practical to perform dynamic testing. Geophysical
measurements are recommended to dene the low-strain dynamic modulus G
max
. Recent generic modulus
relationships for sandy or clayey soils depend on the mean effective conning pressure (sands) or the
plasticity index (clays). For NL analysis based on elasto-plastic constitutive models, modulus degradation
and damping curves are not required. This is because stiffness degradation and damping occur through
stress-strain looping and plastic ow. The low-strain modulus and initial damping are the only parameters
needed.
Cyclic Shear Strength
When materials susceptible to liquefying are present, stress-controlled cyclic triaxial and cyclic simple-
shear tests were originally used to develop cyclic strength curves. These tests cannot duplicate eld
conditions and failure modes observed in dams and, for existing dams and foundations, eld penetration
tests are now preferred to assess the potential for liquefaction. In situ standard penetration tests (SPT)
or cone penetration tests (CPT) are used in ne-grained materials, and Becker hammer tests (BHT) in
coarse materials. In the absence of comparison holes, the SPT may be more reliable than the CPT [Babbitt
and Verigin, 1996] but CPT testing provides more continuous information. The penetration resistance
depends on the depth where the test is performed, hammer and sampler types, hole size, possible use of
drilling mud and the percentages of nes (materials passing the standard U.S. Sieve #200) of the soil
tested. SPT tests are normally corrected in terms of the modied penetration resistance N
1
(60)
cs
of an
equivalent clean sand [Seed, 1987]. The number 60 indicates that blow counts for the standard test with
a rope and drum system correspond to drilling hammer impact energy of 60% of the energy delivered
by free fall of the hammer. Correction factors for the nes are found in the literature [Idriss, 1999; Martin
et al., 1994; Tokimatsu and Yoshimi, 1983].
Corrected SPT tests can be used to obtain normalized cyclic strength curves [Dawson et al., 2001],
based on empirical liquefaction charts [Idriss, 1999; Seed and Idriss, 1982], and magnitude-scaling factors
and the number of equivalent uniform stress cycles corresponding to a given earthquake magnitude
[Idriss, 1999]. Cyclic strength curves relate the applied uniform cyclic stress ratio (CSR) and the number
of cycles to liquefaction. SPT-based cyclic strength curves are conservative, as SPT-based liquefaction
curves represent a lower-bound threshold of the CSR causing liquefaction.
It is essential to recognize that the empirical liquefaction charts and, therefore, the SPT-based cyclic
strength curves apply to cyclic simple-shear condition and a normalized connement of 2000 psf. Hence,
normalized cyclic strength curves must be corrected for the initial static shear and connement conditions
prevailing within the dam section. As the overburden pressure increases, the CSR causing excess pore
pressure of 100% decreases. The CSR also depends on the initial static shear stress ratio (shear stress
divided by normal stress) acting on the soil element. Determining the inuence of the initial state of
stress on the CSR is done through dynamic laboratory testing of isotropically and anisotropically con-
solidated specimens, or through the use of empirical functions relating static state of stress and cyclic
loading resistance [Seed, 1983; subsequently modied by Seed and Harder, 1990 and others (see Idriss,
1999)]. These functions correct the CSR through two factors, K
related to the initial state of static shear,

and K
related to the effective vertical stress. The CSR causing an excess pore pressure of 100% at the
level and of initial stresses is provided as:
(CSR)
=
= (CSR)
= 0
K
(26.6)
Rate of Pore Pressure Buildup
Also needed for dynamic NL effective-stress analysis is the rate at which excess pore pressures build up
during earthquake loading. Fully coupled analyses relate volume changes and excess pore pressure buildup
through constitutive models calibrated with the results of laboratory tests. If volume changes are ignored,
the rate of pore pressure buildup can be expressed numerically by relationships such as that proposed
by Martin and Seed [1978]:
U
g
/
0
= 0.64 arc sin (N

eq
/N
ref
)
0.5/
(26.7)
where U
g
is the generated pore water pressure;
0
is the effective overburden pressure; N

eq
is the current
number of cycles; N
ref
(reference) is the number of equivalent uniform cycles causing an excess pore water
FIGURE 26.6 Typical strain-dependent average shear modulus and damping factors.
pressure of 100%; and is a material-dependent coefcient. A coefcient between 0.7 and 0.8 is typical
of sandy soils.
Post-Liquefaction Residual Shear Strength
Another parameter signicant to the seismic analysis of dams is the post-liquefaction, undrained residual
strength, S
u,r
. This strength can be measured in the laboratory [Castro et al., 1987, 1989; Vaid and Thomas,
1994]. The applicable testing procedures are difcult to implement and sensitive to subjective interpre-
tation. In lieu of laboratory testing, the residual strength can be developed from back-analysis of historic
liquefaction failures in materials of known corrected penetration resistance [Seed and Harder, 1990; R.B.
Seed, 1999]. The use of a post-liquefaction residual strength is justied for numerical analysis purposes
but remains controversial if very low eld penetration resistance is encountered, as such condition rarely
yields undisturbed samples and is difcult to reproduce in the laboratory. Anderson and coworkers [1996]
have suggested in studies of Tellico Dam, TN, that laboratory residual shear strengths can be substantially
higher than those derived from SPT data or developed from the back-calculation of case histories of ow
slides.
In the relationship between S
u,r
and N
1
(60)
cs
, the denition of N
1
(60)
cs
differs from that used to evaluate
the potential for liquefaction, as the clean sand correction factors are lower (R.B. Seed, 1999). The N
1
(60)
cs
obtained with such correction factors can be used with the Seed and Harder relationships to estimate
S
u,r
. The post-liquefaction strength is used as a lower-bound strength limit in NL response analysis or
in post-earthquake static slope stability analyses.
26.4.2.2 Analysis Parameters for Concrete Dams and Appurtenant Structures
Seismic analysis of concrete dams and appurtenant structures requires the denition of the foundation
materials, typically bedrock, and of the dam concrete or masonry. Of concern is the presence of contrac-
tion (vertical) and construction (horizontal, or lift joints), which may represent a weakness and cause
nonlinear response.
Static Properties
For most concrete or masonry dams, cores are recovered and tested in the laboratory for unconned
compression and tension (direct or splitting tests). Cores of sufcient diameter, 6 in. or larger, and
appropriate length-to-diameter ratios must be recovered [Electric Power Research Institute (EPRI), 1992].
As needed, direct shear tests shall be performed on in-place or carefully sampled lift joints. In the absence
of testing, the tensile strength of mass concrete is taken as 10% of the compressive strength. The default
shear strength is 12% of the compressive strength. The static modulus of elasticity (E), if not measured
in the laboratory, can be derived from empirical formulas based on the compressive strength [Raphael,
1984]. A small reduction in modulus is often used to account for creep and other long-term effects.
Poissons ratio is also required, and can be either measured or assigned an average value.
Dynamic Properties
The strength and stiffness of concrete show an increase under rapid (earthquake) loading condition,
compared with slow testing. Current practice uses dynamic strength increase factors (20% for compres-
sion, 30% for shear, and 40% for tension). Tensile strength increase factors up to 50% have sometimes
been justied, based on rapid loading test data [American Concrete Institute (ACI), 1987] or on the
concept of apparent tensile strength [Raphael, 1984]. The modulus of elasticity is also typically increased
by 25% for dynamic load condition.
Strength of Joints
Of interest to dam safety is the condition of joints. Lift joints must withstand earthquake-induced
cantilever tensile stresses, especially near the top of arch dams. Lift surfaces often have tensile and shear
strengths lower than mass concrete. Under seismic loads, vertical and horizontal joints can open or break
shear keys, if any are present. Careful preparation of lift joints increases their strength. Prior to pouring
new concrete, modern dam construction requires surface preparation of the in-place concrete by remov-
ing loose particles and laitance (by brushing or air/water jetting), thoroughly moistening (prewetting)
previously placed concrete surfaces, and carefully vibrating the new concrete upon placement. Wet sand
blasting or high-pressure water jetting can be used if previously placed concrete has signicantly set.
Mortar coats or bonding agents (e.g., epoxies) are sometimes added between old and new concrete.
Depending on placement conditions, the tensile strength of lift joints varies from less than a third to
more than 90% of the strength of intact concrete. The results of research studies on the strength of lift
joints can be used to select strength reduction factors for various methods of surface treatment and
concrete placement. Applicable references include Tarbox et al. [1979], Wall et al. [1986], EPRI [1992],
Pacelli et al. [1993], Lger et al. [1997], and Harza Engineering Company [1997]. Average strength
reduction factors were obtained from these references as follows:
Lift joints with no prior surface treatment. These achieve between 25 and 80% of the tensile strength
of intact concrete. The mean strength reduction factor is 0.60, based on a compilation of published
values.
Joints formed by placing new concrete on a dry concrete surface. Average strength reduction factors
range from 0.55 to 0.90, depending on surface preparation treatments, with a mean value of 0.75.
Joints formed by placing new concrete on a wet concrete surface. Average strength reduction factors
range from 0.53 to 0.96, depending on surface preparation and moisture condition, with a mean
value of 0.80.
Reservoir and Silt Loading
Reservoir water pressure and any silt pressure loads must be applied to the upstream face of the dam.
Based on experience and in the absence of specic silt sampling and geophysical measurements, one can
assume an average density of 85 pcf and a compressive wave velocity of 1000 fps for reservoir bottom
sediments.
26.4.3 Analysis of Embankment Dams
The seismic safety of embankment dams is governed by whether loss of strength might occur within the
dam or its foundation, and whether nonrecoverable deformations remain within acceptable limits. Large
deformations reduce freeboard and often cause longitudinal or transverse cracking. The past 25 years
have resulted in signicant progress in methods and tools to evaluate the seismic performance of embank-
ment dams. The simplest of these methods relies on empirical correlations and simplied procedures
derived from observed or calculated seismic response data, and requires few input parameters. Field
penetration data can be interpreted to assess the potential for liquefaction. Detailed analysis techniques
include EQL (decoupled) solutions, and NL nite element and nite difference coupled or decoupled
formulations. Information on applicable computer programs for dam engineering has been compiled in
a USCOLD publication [1992a] and Bureau [1997] presented a review of various applicable procedures
and some examples of their application.
26.4.3.1 Simplied Analysis Procedures
Simplied procedures are used for small dams analysis or to assess the need for detailed studies of large
dams. Two common procedures are described here.
Newmarks Method
Newmark [1965] computed earthquake-induced displacements in embankments by assuming that move-
ments occur when inertia forces on a rigid block of soils above a xed potential failure surface exceed
its sliding resistance. For planar sliding surfaces, he related the maximum displacement to the peak
acceleration (A) and velocity (V) of the input motion and the yield acceleration (N, or K
y
). The yield
acceleration is the horizontal load coefcient (in g) that results in a factor of safety of exactly 1.0 for the
sliding block. Newmark assumed a number of effective pulses for his standardized earthquake not greater
than 6 and suggested that, for large magnitudes, the number of effective pulses (and computed displace-
ments) be taken proportional to the square root of the estimated duration of shaking. The method can
be applied to any soil mass and planar, circular, or noncircular failure surfaces by double-integrating the
increments of an applied acceleration time history above the yield acceleration, for a downslope direction
of movement. The sliding soil mass is dened by the slip surface with the lowest K
y
in conventional static
slope stability analyses. Computer programs such as STABL [Siegel, 1975] or UTEXAS3 [Wright, 1992]
can be used to obtain K
y
. The input acceleration time history is the specied ground motion, in the case
of small dams, or is obtained by dynamic analysis at a suitable central location along the assumed slip
surface, in the case of large dams. The main limitations of the method is that it assumes a well-dened
sliding block that must be predened and does not account for any progressive loss of soil strength during
earthquake shaking. However, K
y
may be adjusted as a function of time to consider strength degradation.
A derivative of Newmarks method consists of estimating crest settlement by vectorially combining the
displacements of the upstream and downstream slopes [Vrymoed, 1996].
MakdisiSeeds Procedure
A dam responds as a exible body, and accelerations vary as a function of depth within the embankment.
To take this into account, Makdisi and Seed [1977] estimated the peak crest acceleration (
max
) from a
specied response spectrum and a square-root-of-the-sum-of-the-squares (SRSS) combination of the
spectral accelerations of the rst three modes of dam vibration. By interpreting the results of EQL nite
element analyses of several dams, they related the average peak acceleration ratio of the sliding mass
(K
max
) and
max
to the depth of the assumed failure surface. Then, for several magnitudes, they expressed
the normalized peak displacement of the soil mass,
max
/K
max
gT
0
, as a function of K
y
/K
max
(Figure 26.7).
In their investigation, Makdisi and Seed used the examples of clayey dams of medium height (75 to
150 ft). Hence, their procedure applies best to similar dams. For dams higher than 200 ft, it may be
prudent to increase calculated displacements proportionately to the dam height. Due to the assumptions
of no loss of strength during shaking and EQL properties, the procedure is questionable for severe ground
shaking (0.50 g or greater) and only applies to dams built of materials experiencing little or no loss of
strength during shaking (such as densely compacted sands or cohesive clays).
FIGURE 26.7 Simplied estimation of normalized displacements by MakdisiSeeds procedure.
26.4.3.2 Empirical Methods
Empirical methods are based on the observed or computed performance of existing dams and correlate
crest settlement with peak ground motion parameters.
Bureau et al.s Method [1985, 1987]
Observed performance of concrete face and earth core rockll dams was used to develop an empirical
relationship between the earthquake severity index (ESI, see Section 26.3.3) and the relative crest settle-
ment for this type of dam. The dam is assumed founded on bedrock or hard soils, although several of
the dams used in developing the correlation were on alluvial foundations. The original correlation was
developed for compacted rockll, a material that does not develop signicant loss of strength during
shaking. In 1987, the authors tested the correlation with friction angles lower than encountered in rockll,
using the results of physical model tests on dry sand embankments [Roth et al., 1986]. The extended
correlations can be used for dams built of densely compacted granular materials, using the applicable
friction angle (see Figure 26.8).
Swaisgoods Method [1995, 1998]
Swaisgood estimated seismic crest settlements by statistical treatment of data collected from the seismic
performance review of about 60 existing dams. In 1995, he related the crest settlement (CS), expressed
in percent of combined dam and alluvium thickness, to a seismic energy factor (SEF) and three constants
based on type of dam construction (K
typ
), dam height (K
dh
), and alluvial thickness (K
at
). Similar to the
ESI, the SEF depends on the magnitude and peak ground acceleration of the causative earthquake. In
1998, Swaisgood streamlined his approach and expressed the crest settlement as the product of the SEF
and a resonance factor (RF) differentiating between rockll, earthll, or hydraulic ll dams. As is the
case with other simplied procedures, Swaisgoods method is questionable when applied to loose
embankments.
26.4.3.3 Other Simplied Approaches
Other simplied approaches for estimating dam deformations can be found in the literature [e.g., Jansen,
1987; Romo and Resendiz, 1981]. If liquefaction is of concern to the dam or its foundation, the simplied
FIGURE 26.8 Crest settlement estimates for rockll dams and sand embankments based on Earthquake Severity
Index (ESI).
procedure of Seed and Idriss [1970b] can be implemented for dams with at slopes. A better approach
is to assess the liquefaction potential from corrected eld penetration data [Seed et al., 1983; Seed, 1983].
26.4.3.4 Equivalent-Linear Response Analyses
Equivalent-linear (EQL) analyses typically use two-dimensional numerical models of the maximum dam
section. Static analysis is rst required to establish the initial state of stress, such as using the computer
program FEADAM84 [Duncan et al., 1984]. Gradual embankment construction and progressive reservoir
lling can be simulated. Then dynamic response is computed with a different computer program. EQL
response is sometimes obtained for representative soil columns within the dam section using SHAKE91
[Idriss and Sun, 1992]. Most frequently, two-dimensional nite element programs are used, such as
FLUSH [Lysmer et al., 1975], SuperFLUSH [Civil Systems, Inc., 1980], DYNDSP [Von Thun and Harris,
1981] or QUAD4M [Hudson, Idriss, and Beikae, 1994]. These calculate total stress and strain response
using iterated shear moduli and damping coefcients compatible with the average strain induced within
each element of the model. SuperFLUSH and QUAD4M include simulation of a compliant base, which
improves the solution. QUAD4M allows indirect calculation of dam deformations based on the concept
of sliding wedges and seismic coefcients. If three-dimensional effects are expected, two-dimensional
models can be stiffened so that the fundamental period of the modeled section matches that of the three-
dimensional dam, or three-dimensional analysis is performed with TLUSH [Mejia and Seed, 1983).
Following the response analysis, induced stresses can be compared with stresses causing liquefaction
[Seed, 1983], or computed acceleration histories used in Newmarks Method to obtain displacement
estimates. Attempts to convert EQL strains into nonrecoverable deformations have been made using the
concepts of stiffness softening or strain potential [Serff et al., 1976]. Such concepts required signicant
judgment in their application, and are now rarely implemented.
The reliability of EQL analyses decreases when the specied ground motion becomes very demanding.
After such analyses, it is desirable to perform conventional stability analyses of the upstream and down-
stream slopes of the dam, using computer programs such as STABL or UTEXAS3 and assigning post-
liquefaction residual strength properties (see Section 26.4.2.1) to the affected zones of the embankment.
26.4.3.5 Nonlinear Analysis
Recently developed methods of dam earthquake analysis include nonlinear (NL) nite element or nite
difference analysis. These methods apply when loss of strength, large deformations, or liquefaction are
a concern for the embankment or its foundation. A signicant advantage of NL analysis is that the same
numerical model can be used for both static and dynamic conditions. Post-earthquake stability can also
be evaluated by pursuing the analysis through a period of quiet time after the end of the excitation and
verifying whether the dam maintains a stable conguration.
NL analyses include elasto-plastic (EPNL) and direct nonlinear (DNL) solutions. Dynamic pore
pressures are semicoupled or fully coupled with deformations and volume changes. EPNL (two-dimen-
sional) computer programs include DYNAFLOW [Prevost, 1981; Elgamal et al., 1984], DYNARD [Mori-
waki et al., 1988] and FLAC [Itasca Consulting Group, 1992]. A three-dimensional version of FLAC
(FLAC
3D
) was released in 1995. DNL (two-dimensional) programs include TARA-3 and TARA-3L [Finn
and Yogendrakumar, 1989] and GEFDYN [Coyne and Bellier/ECP/EDF-REAL, 1991). The Bureau of
Reclamation (USBR) has also used ADINA/BM [Bathe, 1978] with hyperbolic and cap models and an
endochronic pore pressure generator based on computed strains [Harris, 1986].
Constitutive Models
The previously mentioned programs use various constitutive models. TARA-3 and TARA-3L use total or
effective stresses, hysteretic cyclic shear behavior, and undrained strength parameters. Response depends
on the mean effective normal stress and hyperbolic stress-strain curves, with the tangent shear and bulk
moduli being continuously updated during the calculations. Excess pore pressures are coupled with the
strain response through the MartinFinnSeed model [1975]. Permanent deformations accumulate due
to gravity action and consolidation of the softened soils. If liquefaction is triggered, the specied residual
strength replaces the undrained shear strength.
GEFDYN relies on the HujeuxAubry constitutive model [Aubry et al., 1982]. This model attempts
to reproduce fully coupled uid-soil behavior based on elasto-plastic strain softening/hardening and the
concept of critical state, where soils continue to deform at constant stress and void ratio. GEFDYN
requires common soil parameters (effective friction angle and cohesion; stress-dependent moduli and
Poissons ratio) and critical state, dilatancy, deviatoric, and isotropic parameters [Martin and Niznik,
1993].
FLAC and FLAC
3D
are explicit, nite difference programs. Constitutive equations are solved incre-
mentally [Cundall, 1976], thus allowing large strains, material anisotropies, sliding interfaces, and other
nonlinearities. For dam analysis, the MohrCoulomb constitutive model has been shown to be partic-
ularly applicable [Roth et al., 1986]. Other models are built in the program or can be coded through a
macro programming language. At every calculation step, incremental strains are computed in each
elementary zone and resulting stress increments derived from the applicable constitutive relationship.
Zone stresses and gridpoint displacements are updated, and new incremental strains are computed. Mass-
and stiffness-dependent Rayleigh damping is used at low strain. At higher strains, damping occurs
primarily through hysteretic looping. A semicoupled empirical procedure [Roth et al., 1991; Dawson
et al., 2001], based on the concept of cumulative damage, has been used in FLAC to generate excess pore
pressures at each calculation step. As an illustration of such procedure, Figure 26.9 shows the two-
dimensional nite element model of a large embankment dam, the time history of the computed
settlement at the crest center, and shear stress and excess pore pressures histories obtained for a typical
model grid zone.
In another NL approach using Cundalls equations, Beikae [1996] extended Newmarks method to
calculate three-dimensional seismic displacements in an embankment. The procedure uses a Lagrangian
formulation, coded in the computer program BLOCK
3D
. It simulates gravity, hydrostatic, and seismic
forces on elementary soil blocks with xed masses representing the geometry of the embankment. Soil
blocks can move, expand, compress, and distort in space relative to each other. The equations of motions
are solved explicitly at the gravity center of each elementary block. A signicant limitation is that
BLOCK
3D
applies to materials not susceptible to developing excess pore pressures.
NL Computer Program Validation
An important step in the use of NL analysis is to perform calibration tests and verications. TARA-3 was
validated with centrifuge testing [Finn, 1991] and using the example of Matahina Dam, New Zealand
[Finn et al., 1992]. GEFDYN has been tested using the observed performance of El Inernillo Dam (see
the proceedings of the International Benchmarks Workshops on Numerical Analysis of Dams, Bergamo,
Italy, 1992, Paris, 1994, and Denver, CO, 1999 [ICOLD, 1992, 1994, 1999]). Verications of FLAC for
dam analysis purposes include prediction of centrifuge liquefaction testing results [Inel et al., 1993] and
comparison between observed and back-calculated performances of South Haiwee Dam [Dames and
Moore, 1991], Los Leones Dam [Bureau et al., 1994), Los Angeles Dam [Bureau et al., 1996], and Upper
San Fernando Dam [Dawson et al., 2001].
26.4.4 Analysis of Concrete Dams
The seismic safety evaluation of concrete dams is primarily governed by whether unacceptable stresses
might occur within the dam or sliding of one or several blocks could be triggered along joints, the
foundation, or the abutments. Evaluation methods have evolved considerably over the past 30 years.
Earlier dams were designed through simple force and moment equilibrium considerations, while modern
safety evaluations involve sophisticated linear-elastic or NL dynamic numerical response analyses. The
effects of foundation exibility on strains and stresses within the dam body must be accounted for. Of
possible concern to the stability of concrete dams are uplift forces at lift joints and at the interface between
the dam and its foundation, and joint opening. The 1992 USCOLD publication (see Section 26.4.3) also
describes computer programs applicable to concrete dams. In addition to concrete, water, and foundation
properties, viscous damping ratios of 3 to 5% for the OBE and between 5 and 10% for the MCE are
considered appropriate input parameters.
26.4.4.1 Simplied Analysis Procedures
Some old arch dams have been designed using force equilibrium methods, which do not consider cantilever
action. Simplied analysis procedures based on such methods are now used only for small gravity dams,
and assume that the dam is a two-dimensional rigid body that can slide on a rigid foundation. The updated
Engineering Guidelines for the Evaluation of Hydropower Projects [Federal Energy Regulatory Com-
mission (FERC), 2000] includes a methodology for such analysis. The guidelines require iterative cracked-
base analysis with zero tensile strength and full reservoir head in the cracked zone.
FIGURE 26.9 Nonlinear two-dimensional nite difference model of an embankment dam and typical analysis
results.
26.4.4.2 Detailed Analysis of Gravity Dams
Many computer programs are available to analyze gravity dams. The most frequently used programs
include multipurpose, three-dimensional codes such as ANSYS [Swanson Analysis Systems, Inc., 1970],
ADINA [Bathe, 1978], SAP90 and SAP2000 [Wilson and Habibullah, 1988], and specialized two- or
three-dimensional programs, such as EAGD-84 [Fenves and Chopra, 1984] and EACD-3D [Fok et al.,
1986]. Most of these programs have been continuously updated since their rst release. All perform time-
history response analysis but the general purpose programs also include response spectrum analysis.
Leclerc and coworkers [2002] recently introduced CADAM, an interactive program for gravity dams,
based on response spectrum analysis.
Analysis of gravity dams is generally done using two-dimensional nite element models of the max-
imum section of the dam. Depending on the dam geometry, one may need to consider other sections
or three-dimensional analysis. Plane-strain or plane-stress two-dimensional formulations can be used,
depending on the spacing of construction joints. Added masses (Westergaards theory) represent the
reservoir water, if more rigorous solutions are not provided. The original added mass concept is based
on simplifying assumptions of vertical upstream face, rigid dam section, and incompressible water but
was modied by Kuo [1982] for other orientations of the upstream face. Some computer programs
simulate hydrodynamic pressures using a nite-element formulation and compressible or incompressible
uid elements (e.g., EACD-3D). A massless foundation is generally assumed, except for the half-space
of EAGD-84. In that program, seismic input is entered at the bottom of the dam (top of the half-space)
but, in others, the base of the foundation model is excited. A two-dimensional analysis ignores the forces
that resist the thrust on the abutments of a curved concrete gravity dam. Plane-strain or plane-stress
analysis, depending on which is applicable, is likely to be conservative for stress evaluation purposes,
and three-dimensional analysis would generally reduce the maximum calculated tensions and compres-
sions. In two-dimensional analysis, neglecting effects and crack opening patterns related to the valley
topography may lead to overestimating the factor of safety against sliding [Brand, 1993]. However, the
two-dimensional approach is often sufcient to demonstrate the safety of straight or some curved
concrete gravity dams, and is a cost-effective solution if safety is demonstrated.
Chopra [1988] used EAGD-84 to identify several factors of signicance to gravity dam response: the
inuence of water compressibility, dam exibility, and vertical component of earthquake motion on
hydrodynamic pressures; the inuence of wave propagation through the foundation rock; and the effect
of seismic waves absorption by the reservoir bottom sediments. In EAGD-84, the dam monolith is
idealized as an assemblage of planar, four-node, nonconforming, linear-elastic elements and the foun-
dation as an isotropic, innite half-space. The reservoir water is represented as a compressible uid of
constant depth and innite length toward upstream. Hydrodynamic pressures are simulated through the
solution of the two-dimensional wave equation. The absorption of compressive uid waves by the bottom
sediments and foundation rock is accounted for through a wave absorption coefcient. Tailwater is not
represented in the model. Constant modal damping is used for the dam, and hysteretic damping for the
foundation. A low foundation damping (e.g., 0.5%) and a large wave reection coefcient (e.g., 0.8) for
the elastic half-space imply near-full reection of water pressure waves by the reservoir bottom, a
conservative approach. Stresses are calculated at element centers.
Despite major progress in numerical analysis procedures, consideration of the inuence of uplift on
the sliding stability relies upon simple principles. Dynamic uplift pressure uctuations are not considered,
and static uplift pressures are combined with the earthquake response to obtain the effective normal
stresses at the damfoundation interface. In the cracked-base approach [Federal Energy Regulatory
Commission, 2000], full reservoir head is applied to any zone experiencing normal tension along the
base of the dam by iteratively adjusting the length of cracking and relaxing or deleting the corresponding
dam elements [Brand, 1993]. This procedure is best implemented using computer programs with inter-
face elements (e.g., ANSYS, SAP2000, FLAC). Improved formulations regarding the simulation of planar
discontinuities and concrete cracking have led to the consideration of nonlinear concrete behavior
[Tarbox et al., 1979] and MohrCoulomb damfoundation interfaces [Chavez and Fenves, 1993; Bureau,
1996a; Bu, 2001]. Most importantly, it should be recognized that various well-accepted programs may
yield different results because of their different formulations. Using three successive two-dimensional
analyses (SAP90, EAGD-84, or FLAC) for the same dam, peak tension was reduced by 30% and peak
compression increased by 60% when a sliding interface with no bond strength was placed between the
dam and its foundation; sliding stability results, however, were consistent [Bureau, 1996a]. Overall, a
thorough understanding of available evaluation methods is required, especially in the case of marginally
stable gravity dams.
26.4.4.3 Detailed Analysis of Arch Dams
Three-dimensional response analysis is typically required for arch dams. Early analyses of arch dams
were based on the trial load method but that method has now been replaced by nite element analysis.
The geometry of such dams is complex and must be represented with sufcient renement. Dynamic
effects must be superimposed with static effects. The latter, in the case of arch dams, must incorporate
temperature effects, which are less signicant and often ignored for gravity dams. Stresses are typically
resolved as horizontal arch and vertical cantilever stresses along the upstream and downstream faces of
the arch (see Figure 26.10). A correction factor may be used to obtain principal stresses. Seismic response
analyses of arch dams can be conducted using specialized computer programs such as EADAP [Ghanaat
and Clough, 1989], ADAP-88 [Fenves et al., 1989], GDAP and ADAP-NS [Quest Structures, 1993], and
SCADA [Hall, 1997]. General purpose three-dimensional codes, such as SAP90, SAP2000, or ANSYS (see
Section 26.4.4.2) can also be considered but a deciency of these codes is that reservoir interaction effects
can be represented only through the use of added masses. As the upper portion of an arch is more exible
than the top of a gravity dam, reservoir water compressibility can lead to signicant pressure wave effects
that modify response appreciably. Specialized programs include uid elements and compressible water
formulation. As for gravity dams, a large portion of the foundation bedrock must be included in the
analysis of arch dams.
GDAP is a graphics-based nite-element analysis program for linear-elastic static and dynamic analyses
of concrete arch dams. It represents damreservoir interaction effects with an added-mass matrix com-
puted using a nite-element formulation for an incompressible uid volume. To achieve an efcient
substructure solution, the full added-mass matrix is converted into an equivalent diagonal matrix that
preserves hydrodynamic forces due to rigid body accelerations of the dam in the stream, cross-canyon,
or vertical directions. ADAP-NS (an enhanced version of ADAP-88) has capabilities to represent potential
zones of weakness in the dam, such as lift joints, damfoundation interface, and signicant cracks.
ADAP-NS uses three-dimensional solid elements similar to those employed in GDAP to model the arch
and the foundation bedrock, and three-dimensional uid elements to represent the reservoir water. It
uses a substructure solution procedure to carry out static and dynamic analyses. Any portion of the dam
body bounded by contraction joints, lift joints, or the damfoundation interface is considered as a linear
substructure, for which a constant stiffness is used throughout the solution process. Nonlinear effects
are restricted to joint elements whose possible opening is inuenced by preexisting normal stresses.
Normal and tangential stiffnesses and the assumed tensile strength control joint behavior. A large normal
stiffness simulates joint closing. The tensile resistance of joints is cancelled when normal tensile stresses
reach the specied tensile strength. Joints with no initial tensile strength open freely, if subjected to tensile
normal forces. An example three-dimensional nite element model of an arch dam with discrete repre-
sentation of selected lift and contraction joints is shown in Figure 26.11.
Another form of NL analysis, used in SCADA, relies on the concept of smeared cracks. Such a solution
not only attempts to model the opening and closing of contraction joints but also the formation of cracks
in the concrete as a result of excessive tensile stresses. The smeared crack approach simulates nonlinearities
by placing conditions on the stresses of the elements (shells) that represent the arch. The smeared crack
approach is computationally more efcient and exhibits better convergence than discrete joint models.
It allows formation of cracks based on the most critical orientation of induced stresses.
As in the case of gravity dams, the sliding stability of each contact block must be evaluated. Peak
static-plus-seismic driving and resisting forces can be compared, or a step-by-step, time-dependent factor
of safety can be computed for each block as the ratio of transient resisting-to-driving forces. The results
of a typical block sliding stability analysis are presented in Figure 26.12. For such transient analyses, it
should be noted that instantaneous dynamic factors of safety less than 1.0 do not necessarily indicate
unacceptable behavior, if the duration of such occurrences is very short and, therefore, insufcient to
induce signicant sliding of the dam.
FIGURE 26.10 Arch dam stress analysis results, plotted as arch and cantilever stress contours.
26.4.4.4 Detailed Analysis of Buttress Dams
The evaluation of such dams normally requires three-dimensional considerations and is, therefore, similar
to the evaluation of arch dams (see Figure 26.13). In the case of long-crested buttress gravity or multiple
arch dams, it may be sufcient to evaluate only two adjacent buttresses and the portion of the dam
between. Buttresses have often been built slender and are very sensitive to cross-valley seismic loads.
Hence, one must successively take the largest component of horizontal motion as parallel or perpendicular
to the dam crest. Some buttresses near the abutments may be affected by the possible presence of joints,
shear zones, or foliation parallel to the valley slopes, and earthquake effects on the stability of rock blocks
bounded by such features must be investigated.
26.4.5 Analysis of Intake/Outlet Towers
Intake/outlet towers can reach signicant heights (more than 400 ft tall) and may signicantly amplify
ground motion. For towers of simple conguration, Chopra and Fok [1984] concluded that the rst two
modes of tower response provide sufcient accuracy to compute shear forces and moments. For large,
complex, or very exible towers, dynamic response analysis is the most appropriate evaluation method-
ology. In addition to hydrodynamic interaction effects, the response of exible towers can be inuenced
by the presence of internal equipment and wall openings, and by soilstructure or structure-to-structure
interaction (access bridges). Finite element models with exible three-dimensional beam elements and
distributed or lumped masses (stick model) represent most towers adequately [Bureau, 1993b; USCOLD,
1995]. As few modes of vibration contribute signicantly to dynamic response, 20 joints (nodal points)
or less are generally sufcient to represent the tower shaft. Added mass coefcients are typically used to
approximate hydrodynamic effects for various tower sectional shapes [Goyal and Chopra, 1989]. More
complex formulations using incompressible or compressible uid elements are also available. Most intake/
outlet towers are either founded on an enlarged base on the reservoir bottom or cantilevered into bedrock
or hard soil. In situ monitoring of ambient and low-level forced vibrations of intake/outlet towers
embedded in rock have shown apparent points of xity several tens of feet below ground surface [Bureau
and Scawthorn, 1986; Bureau, 1985]. Therefore, a signicant depth of the foundation rock may need to
be included in the analytical model. Plane strain axisymmetric or three-dimensional thick plate or solid
elements can also represent complex tower wall congurations and the surrounding foundation [Dungar
and Jackson, 1975]. Approximate three-dimensional analyses of outlet towers embedded in embankment
FIGURE 26.11 Three-dimensional nite element model of arch dam with representation of contraction and lift
joints. (Numerical model developed by Quest Structures, Orinda, CA.)
dams have been done using two-dimensional, equivalent-linear, plane-strain nite element models with
unbalanced free-eld boundary conditions [Bureau and Udaka, 1982] (see Figure 26.14). Light access
footbridges have little effect on tower response. Some towers, however, are connected to massive rein-
forced concrete bridges with possible intermediate piers which must be included with the tower shaft in
a detailed model (see Figure 26.15).
Once a representative model has been developed, response is rst calculated using elastic response
spectra and gross section properties. As for concrete dams, viscous damping ratios of 3 to 5% for the
OBE, and between 5 and 10% for the MCE are considered appropriate. Induced shear and moment loads
FIGURE 26.12 Nonlinear block sliding stability analysis for a thick arch dam.
4
5
6
7 8 9 10
11 12
13
14
15
16
17
18
(a) Dam-Foundation Blocks for Evaluation of Factor of Safety
(b) Computed Factors of Safety at Critical Time Steps
Computed @ 3.155 sec
Computed @ 4.515 sec
Required FS
14
12
10
8
6
4
2
0
2 1 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Block No.
F
a
c
t
o
r

o
f

S
a
f
e
t
y
(c) Computed Instantaneous Factor of Safety (Typ.)
F
a
c
t
o
r

o
f

S
a
f
e
t
y
10
9
8
7
6
5
4
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10
Block No. 11
Time (sec)
are compared to the gross (uncracked) capacity of the concrete or masonry. The SRSS method is
appropriate to combine multidirectional loading components. If the uncracked capacity is not exceeded,
such analysis is sufcient. If cracking is shown to occur, induced forces and moments are resisted by the
steel reinforcement. Approximate or detailed nonlinear response analysis may be required to compute
the post-cracking response and verify that the ultimate structural capacity (bending or shear) will not
be exceeded. Cracking, especially, reduces the stiffness of a tower and lengthens the period of its principal
modes of vibration, which may increase or decrease earthquake loads, depending on the peak of the
specied response spectrum.
Cracking of reinforced concrete is a highly nonlinear phenomenon. If sufcient ductility is available,
idealized elasto-plastic moment vs. rotation curves can be developed for the tower shaft. For most towers,
however, it is prudent to limit cumulative ductility factors to 2.0 in both bending and compression
[Chopra and Liaw, 1975]. A possible approach to simulate cracking of a tower is an iterative equivalent-
linear procedure [Bureau, 1993b]. The analysis properties of overloaded sections are modied to simulate
crack formation and propagation. Where the cracking moment M
cr
is exceeded, a procedure similar to
that recommended for beams in ACI-318, Building Code Requirements for Reinforced Concrete, is
used to obtain the ultimate shear and moment capacity. An equivalent (reduced) moment of inertia I
e
is also derived from the gross (I
g
) and cracked (I
cr
) section properties. The cracked moment of inertia
I
cr
is the sum of the moments of inertia of the compression concrete and vertical steel reinforcement
bars, and reduced moments of inertia are substituted for gross moments of inertia in cracked sections.
Vertical steel provides the principal resistance to bending moments. In the ACI-318 procedure for
estimating the ultimate capacity of reinforced concrete sections, an ultimate strain of 0.003 in./in. in the
extreme ber of the compression face of the tower section considered and a rectangular compressive
stress block are employed. Moment-axial load interaction curves must be considered in bending capacity
FIGURE 26.13 Finite element model of triple arch dam and its foundation.
calculations. A typical load-moment interaction diagram is shown in Figure 26.16. The ultimate shear
capacity of the shaft section can also be computed, using the provisions of ACI-318, as the sum of the
shear capacity of the concrete (V
c
) and the capacity of the shear reinforcement steel (V
s
).
The most overstressed shaft element is assumed to crack rst, and the tower response is reevaluated.
If overstressing extends beyond the initial cracked zone, cracking propagates upward or downward (or
both) along the shaft, and the process is iterated. Stability against further cracking may be achieved or
not, and this approximate procedure indicates with a reasonable degree of condence if the ultimate
seismic capacity of the tower will be exceeded. For towers embedded in embankment dams, one must
also verify that cracking would not cause leakage or piping, even if the tower has sufcient capacity.
26.4.6 Limitations of Current Analysis Methodologies
The ability of engineers to sample and test embankment dam materials meaningfully lags behind the
ease of using, developing, or obtaining access to complex theoretical constitutive soil models and powerful
FIGURE 26.14 Approximate three-dimensional embankment tower seismic interaction model.
numerical tools. First and foremost, one must keep in mind that modern computerized analysis tools
operate at a level of apparent accuracy that far exceeds the engineers ability to obtain reliable measure-
ments of eld material properties and assess the variability of such properties within the dam or its
foundation. Hence, regardless of their sophistication, the quality of numerical analyses is always limited
by the quality of the material properties to be used as a basis to develop numerical models.
FIGURE 26.15 Three-dimensional beam model of outlet tower, access bridge, and piers.
FIGURE 26.16 Typical moment-axial load interaction and cracked moment of inertia relationships for outlet tower
shaft section.
Similarly, and despite careful verication and acceptance by review authorities, analysis tools for
concrete dams include both useful features and potential limitations in their ability to solve the complex
problem of dam response. As Professor Ray Clough stated in 1993, in judging the seismic response
of a structure, the displacements that occur are much more important than the stresses [Clough and
Ghanaat, 1993]. Case histories have led to difculties in questions regarding very large, calculated elastic
stresses and the absence of conspicuous earthquake-induced cracking (e.g., Lower Crystal Springs Dam
in 1906; Pacoima Dam in 1971 and 1994). Nonlinear analysis suggests that cantilever and arch stress
calculations may merit less importance than that given to date, but it raises the question of what are
acceptable computed amplitudes for joint opening and dam movements.
For both embankment or concrete dam analysis, excessive conservatism is less forgiving in a nonlinear
environment, and sophisticated solutions require more realistic assumptions, and perhaps new consid-
erations. Until increased knowledge is gained, we must recognize the limits of our ability to duplicate
the numerous and complex phenomena that are part of the seismic behavior of dams. Common sense,
engineering judgment, and eld and construction experience remain the most important components
of any numerical analysis.
26.4.7 Physical Testing, Modeling, and Centrifuge Studies
Physical testing and modeling of dams can be used to conrm analysis assumptions, develop insight into
performance, identify modes of vibration, response or failure, and validate numerical models. Detailed
discussion of these aspects is beyond the scope of this chapter, and only a few comments are appropriate.
Observations of the behavior of dams in the linear-elastic range, using full-size shaking such as induced
by mechanical vibrators, yields results that generally agree with computed values at low levels of excita-
tion. A successful way to test dams and their appurtenant facilities has been the use of underwater pressure
wave generators [Ostrom and Kelly, 1977] or explosive charges, placed at some depth in the reservoir,
which apply low-level forced vibrations to the structure tested. While these low-level forced vibrations
are useful for calibration of analytical models, it is difcult to extrapolate from those results to compute
the anticipated response to strong seismic motion.
Physical model (reduced-size) testing of earthquake effects on dams has been conducted in various
research projects. The literature is rich with descriptions and results of model testing for stresses and
deections in concrete dams, soilstructure interaction and deformation, slope stability, and excess pore
pressure generation in embankment dams [Clough and Niwa, 1982; Hall, 1997]. Scaling is a major issue
in model testing and, especially, dynamic model testing. With the availability of powerful computer tools,
computational ability has progressively replaced the ability to test, and the use of physical modeling is
now limited to large universities or government laboratories for special cases. However, ability to suc-
cessfully test embankment dams of relatively simple geometry has been enhanced by the development
of testing devices that allow improved feedback controls and dynamic excitation at the base of the model
(which is loaded statically through the rotation of the centrifuge). Detailed information and a compre-
hensive list of publications on centrifuge testing are provided in the proceedings of specialty conferences
on the subject; see Centrifuge 88, Centrifuge 91, and Centrifuge 94, published by A.A. Balkema.
26.4.8 Post-Earthquake Inspection
Competent operations personnel and engineers must inspect dams that have been subjected to signicant
earthquake ground motion. Post-earthquake inspections must include initial and follow-up site visits.
Immediate inspection is critical to decisions regarding continued safe operation. Dam inspections are
essential whether damage has occurred or not, as relevant information collected provides insight regard-
ing the structural performance of the affected facility and may be used to subsequently calibrate or verify
numerical analysis models. A post-earthquake inspection plan should be prepared, and should include
the crest and faces of the dam, abutments, gates and spillways, outlet works, and penstocks and power-
plant, if any. Inspections should focus on identifying any foundation or slope movements, soil cracks,
ssures and settlement, increased or reduced seepage, evidence of concrete overstressing or joint move-
ment, and reservoir margin defects, and actual or imminent landslides. Collection of recorded data from
strong motion accelerometers, settlement monitoring devices, inclinometers, tiltmeters, etc. must proceed
immediately. Crest and slope surveys may need to be performed if the dam is not instrumented. To make
inspections more meaningful, inspectors should follow the applicable guidelines for inspections of dams,
such as available from USBR [1983], ICOLD [1988], and FERC [2000].
26.5 Seismic Upgrade of Existing Dams
26.5.1 General
Dams can be modied to increase their seismic capacity and repaired, strengthened, or replaced if
damaged by earthquakes. Such modications to dams started in the 1930s but systematic improvement
programs for state and federal dams really began after the 1971 San Fernando earthquake. In the United
States, more than 140 dams have been modied to better resist ground motion. Other dams have been
abandoned, replaced, or converted to ood control instead of water storage.
New dams can be designed and constructed to resist the most severe earthquake loads, as demonstrated
by the outstanding performance of several modern dams. Many procedures used to upgrade existing
dams apply to new dams. The philosophy for new design or for improvement of existing facilities to
efciently resist seismic loads requires a thorough analysis of the observed behavior of different types of
dams during past earthquakes and an understanding of why performance has been poor or satisfactory.
26.5.2 Seismic Upgrade of Embankment Dams
Embankment dams and foundations can be modied to prevent earthquake-induced overtopping, inter-
nal erosion, liquefaction, or a combination of these failure modes. A compendium of seismic upgrade
methods used for such dams is provided in the following paragraphs. Several of these methods are often
simultaneously implemented. The name of a representative completed dam upgrade is provided for each
main type of improvements. The reader should consult the literature for more details.
Dam improvements: Embankments can be strengthened by enlarging, attening, or adding berms
to upstream or downstream slopes (Bradbury Dam, CA); increasing freeboard by raising the crest
(Calaveras Dam, CA); installing crack stopper zones (Austrian Dam, CA); driving piles in upstream
slope (Sardis Dam, MS); implementing compaction grouting (Pinopolis West Dike, SC); and
constructing lters and drained buttresses (Steinaker Dam, UT).
Foundation improvements: Foundation improvements include performing vibrootation-vibro-
compaction (Scoeld, Dam, UT); compaction grouting; removing or replacing subsurface mate-
rials with compacted soils (Austrian Dam, CA) or soil-cement mixture (Jackson Dam, WY);
performing deep dynamic compaction (Folsom Dam, CA); installing drains (Hinckley Dam, NY)
or stone columns (Lopez Dam, CA); and constructing impervious cutoffs (Hebgen Dam, MT).
26.5.3 Seismic Upgrade of Concrete Dams
With regard to concrete dams, improvements may involve either the structure or its foundation, and are
intended to limit overstressing, prevent sliding, control joint opening, and stabilize buttresses. Some of
the improvements of embankment dam foundations can be used for concrete dam foundations.
Dam improvements: Methods that have been implemented include thickening the cross section
with concrete or shotcrete, or adding buttresses of roller-compacted concrete (RCC) or mass
concrete (Gibraltar Dam, CA); thickening or restraining buttresses laterally (Weber Dam, CA);
epoxy-grouting the concrete (Sed-Rud Dam, Iran); installing post-tensioned anchors through
the dam (Railroad Canyon Dam, CA); installing cross-braces or strut multiple arches and
Ambursen dams (Sutherland, CA); and installing relief drains.
Foundation or abutment improvements: Drainage, grouting, rock bolting, and anchorage with rock
bolts or post-tensioned anchors (Pacoima Dam, CA).
26.5.4 Seismic Upgrade of Appurtenant Structures
Means to improve facilities appurtenant to dams are the following:
Increasing freeboard by spillway lowering or gate removal (Henshaw Dam, CA)
Plugging outlet tunnels or conduits exposed to differential fault movements (Calaveras Dam, CA)
Constructing bypass outlets (Lake Madigan Dam, CA)
Anchoring intake/outlet tower with post-tensioned multistrand cables (Tolt Dam, WA)
Adding steel brace and shotcrete to outside of tower wall (San Gabriel Dam, CA)
Building a moment-resisting structural frame around the tower (Shin Tsuroko Dam, Japan)
Enlarging the diameter of the tower base mat and adding mass above
Filling the lower portion of the inside shaft with concrete (proposed)
Strengthening access bridge piers, replacing concrete deck with lighter steel deck
Tying the bridge deck to the tower
Building bulkhead where outlet tunnel joins tower (Lake Herman Dam, CA)
Upgrading spillway gates (Matahina Dam, New Zealand).
26.6 Seismic Design of New Dams
26.6.1 General
Seismic considerations affect many decisions in the design of new dams, ranging from site selection to
the type of instrumentation to be installed. The selected dam type and design features should be tailored
to the site conguration, geology, and the local and regional seismic environments. Topographic effects,
joint orientations, and valley shape can affect the seismic response, and should be accounted for.
26.6.2 New Embankment Dams
The foundation conditions should be thoroughly investigated and unsuitable materials removed,
replaced, or compacted, and drainage features installed. The embankment must be designed to the highest
standards. Particular considerations should be given to rockll dams with concrete (CFRD) or asphalt
concrete faces. Details of the connections of the face and toe slabs, and anchorage of the toe slab to its
foundation should be carefully considered. Waterstops should be carefully designed and redundant
waterstops utilized as needed. Drainage materials below the face slab should be placed on well-compacted
rockll. Listed below are various design or construction measures for the seismic design of new embank-
ment dams.
Increase freeboard to lengthen lateral seepage paths at normal operating reservoir level and accom-
modate earthquake-induced settlement or seiches.
Increase the crest width to produce longer seepage paths through transverse cracks that might
develop during a seismic event.
Excavate foundation to very dense materials or bedrock; or densify or replace loose layers with
well-compacted materials.
Provide gentle shape, free of sharp and reentrant edges, to the foundation/core contact. Keep
transverse upstream foundation slopes across the core to less than 1:4 (v:h), and slopes along the
dam alignment to less than 1:2 (v:h). This will reduce or prevent transverse cracking. At the
abutments of large dams, in the upper 100 ft, build transverse foundation slopes across the core
zone horizontal or to a gentle slope toward upstream. This preserves watertight contact after
shaking and reduces settlement. Flare the core contact, along the upper portions of the abutments,
to provide longer seepage paths in case of abutment ssures.
Thoroughly compact all zones of the embankment. This reduces earthquake-induced loss of
strength, deformations, or settlement.
Upstream from the water barrier or below the water table, avoid using ll materials that tend to
develop excess pore water pressures.
Provide high capacity internal drainage zones to intercept seepage from any earthquake-induced
transverse cracking and prevent saturation in embankment zones designed to remain dry.
Place lter zones along fractured foundation bedrock to prevent piping of the embankment
through enlargement of these fractures. Widen lter and drain zones to maintain material conti-
nuity in case of offset and to heal transverse cracks.
Select gradations for the transition zones that will be self-healing and facilitate healing of any
cracks within the core zone.
Avoid brittle materials for use as water barriers. Use more plastic materials in areas where tensions
are prone to develop during earthquakes.
26.6.3 New Concrete Dams
Buried narrow gorges in wide valley foundations may affect the dynamic response of concrete buttresses
or monoliths. Such a condition must be mitigated, for example by plugging the largest gorges with mass
concrete, or by selecting a different type of dam, such as rockll embankment. A concern under seismic
condition is the performance of combination structures, such as earth or rock structures wrapped around
the end of concrete gravity dams and spillway structures, and composite gravity arch dams. Potentially
critical are dynamic interaction and differential displacements of adjacent structures with incompatible
periods of vibration. Design details that improve performance of concrete dams are listed as follows:
Design a regular and smooth geometry for the dam structure (symmetry is desirable but not
essential). Increase the crest width to reduce incidence of through-cracking. Limit the ratio of
crest length to dam height and reduce mass near top of the dam to control crest deections and
minimize distortions.
Avoid slope changes along the downstream face of gravity dams to eliminate local stress concen-
trations.
Minimize discontinuities in dam body, such as those caused by heavy parapet walls and thrust blocks.
Maintain continuous compressive loading along the foundation by shaping the damfoundation
interface or adding a plinth to support and transfer dam loads.
Improve quality of foundation bedrock by excavating, replacing with concrete or grouting poor
quality rock, shears, cavities, etc.
Brace, widen, or enlarge the bottom portion of buttress walls in buttress dams to improve resistance
earthquake loads parallel to the dam axis.
Provide contraction joints with strong interlocking through use of oversized shear keys. Prepare,
clean, and wet lift surfaces to increase bond strength.
Maintain low concrete placement temperatures to minimize locked-in tensile stresses and reduce
the occurrence of shrinkage cracks.
Slope and batter concrete interface in composite earth-concrete dams to maintain compressive
embankment loading against the concrete; use plastic soils and wider lter zones in the embank-
ments in the interface area.
26.6.4 New Appurtenant Structures
Seismic consideration must also be given to the design of appurtenant facilities and, especially, intake/
outlet towers. To mobilize a higher ductility in such structures, a key requirement is to provide adequate
reinforcement and caging (i.e., connement) for the concrete. Vertical and horizontal steel should be
provided in several curtains, and bar sizes and spacing selected to adequately conne the concrete upon
shear load reversals and prevent excessive tensile stresses in bending and compression buckling of the
vertical steel. For highly seismic sites, inclined reinforced concrete towers, supported on the foundation
over their entire length, may be considered. This solution is a compromise between structural safety and
water quality requirements. If a vertical tower is preferable, then holding it down with post-tensioned
anchors may provide a solution. The tower should be sited far from the dam, if concerns are raised
regarding potential damage to the dam caused by tower failure.
Internal or external elements, such as operating platforms, decks, access ladders, polar cranes, electrical
control panels, control valves, slide or radial gates, and their guides and supports must be adequately
anchored, especially since amplications of the ground motion to the top of the tower subject internal
equipment to large lateral forces. Care must be taken to prevent access bridge decks from overloading
the tower. Light steel trusses or wooden decks are recommended. If a heavy bridge cannot be avoided,
it should be structurally disconnected from the tower, supported on heavy piers or shear walls, and
equipped with a collapsible or compressible joint at the bridge-tower connection.
26.7 Seismic Instrumentation of Dams
High-quality performance data and strong-motion records on and near dams are essential to better
understand the behavior of dams during earthquakes and calibrate and improve methods of numerical
analysis. Conventional dam instrumentation should be provided, plus accelerometers or, preferably,
strong-motion instruments producing digital records [USCOLD, 1989]. A network of instruments should
preferably be deployed at selected locations in the valley, including sufciently far from the dam to obtain
free-eld condition, at the base of the dam and top of the abutments, and at or near the center of the
dam crest and along the downstream face. Three instruments are the bare minimum to consider. If a
planned reservoir is large enough to be considered as a potential source of reservoir-triggered seismicity,
a network of sensitive seismographic instruments may be installed around and beyond the reservoir
perimeter to record local microseismic activity and establish a baseline prior to reservoir lling.
Acknowledgments
This chapter was inspired by 15 years of involvement of the author as vice chairman of the Committee
on Earthquakes of the U.S. Society on Dams (USSD, formerly USCOLD). Permission to borrow materials
and information from relevant committee publications was received from Mr. Larry Stephens, Executive
Director, USSD, and is gratefully acknowledged.
Dening Terms
Added masses Added masses may be used to represent the effect of hydrodynamic pressures in the
reservoir water and lengthening of the natural periods of vibration of a dam (compared with
the empty reservoir) as a result of earthquake shaking. The concept was originally developed
by Westergaard [1933] for a vertical dam face and was subsequently extended to any dam face
orientation by Kuo [1982]. Added masses can also be used for intake/outlet towers surrounded
by the reservoir water [Chopra and Fok, 1984].
Contraction joint An articial, generally vertical, joint constructed (cut) in concrete dams to reduce
tensions and prevent cracking of the concrete, after placement, as a result of cooling during
curing.
Cracked-base analysis For gravity dams, this method of analysis is used to evaluate the inuence of
tensile stresses along the damfoundation interface. Where normal tensions are computed across
the interface, uplift is assumed to occur, and the bond strength of the concrete and bedrock is
set to zero. Full reservoir pressure is also applied to the uplifted zone. The analysis is repeated
with these new assumptions, and the process iterated and the length of the cracked zone
increased, as needed, until no further tensions develop.
Design basis earthquake (DBE) The maximum earthquake employed for the design or analysis of
a dam, usually specied in terms of acceleration time histories or response spectrum at the site
or both. It may be dened in terms of magnitude and epicentral distance from the dam site.
Equivalent-linear analysis In an equivalent-linear analysis, embankment materials properties
(damping coefcient and shear modulus) are iteratively adjusted in a series of successive linear-
elastic analyses to become compatible with the average computed shear strains within the
elements of the numerical model. The average strain is taken as a xed percentage of the peak
strain in each element. Iterations are stopped when convergence is achieved for the estimated
properties, based on the computed average strains.
Excess pore pressure Corresponds to an increase of internal pore water pressure within a saturated
soil medium as a result of a reduction in skeleton volume caused by earthquake shaking.
Fling Very large velocities and displacements observed in strong ground motion records recovered at
near-fault sites, resulting from tectonic deformation.
Foliation Parallel alignment of planar fabric elements in rock.
Freeboard Represents the elevation differential between the crest of a dam and the lowest portion of
the spillway crest. A large freeboard provides extra safety, as it reduces earthquake loads and
the possibility of overtopping.
Free-eld The regions of the ground surface that are not inuenced by man-made structures. Also
designates a medium containing no structure (free-eld prole) and where boundary effects
do not signicantly inuence the response of the medium to earthquake motion.
Hydraulic ll dam Designates a dam built by deposition of loose granular materials, typically silts
and sands, conveyed as a slurry to the site by hydraulic pumps and pipes.
Lift joint The surface between two consecutive concrete pours during the construction of concrete
gravity or arch dams.
Operating basis earthquake (OBE) In the case of dams, represents the level of ground motion at
the dam site with a 50% probability of not being exceeded in 100 years [USCOLD, 1999].
Maximum credible earthquake (MCE) (Also Maximum considered earthquake) The largest earth-
quake considered in an analysis, that appears possible along a recognized fault or within a
geographically dened tectonic province. Little regard is given to its probability of occurrence.
Usually the largest reasonably conceivable event. MCE sometimes also used to denote maximum
credible earthquake, although this term is not favored now.
Moment magnitude (M
W
) The moment magnitude M
W
of an earthquake is proportional to the
logarithm of the seismic moment M
0
. M
0
is a measure of the energy released and is related to
the physical characteristics of the causative fault rupture.
Rayleigh damping In nite element analysis, the Rayleigh damping matrix [D] is proportional to
the mass matrix [M] and stiffness matrix [K] of a exible system, and is typically expressed by
the formulation [D] = a[M] + b[K], where a and b are two constants.
Reservoir-triggered earthquake (RTE) The maximum level of ground motion capable of being
induced at a dam site by the lling, drawdown or simple presence of the reservoir. Consideration
of the RTE is generally limited to dams higher than 300 ft.
Reservoir-triggered seismicity Seismic activity potentially caused by the lling of a large reservoir.
Response spectrum A plot of the maximum values of acceleration, velocity, and displacement
response of an innite series of single-degree-of-freedom systems subject to a time-dependent
dynamic excitation. The maximum response values are expressed as a function of the undamped
natural period of each system, for a specied viscous damping coefcient.
Spillway A channel, pipe, tunnel, or overow or gated concrete structure intended to evacuate excess
ood ows and prevent overtopping of the dam.
Tailings dam A dam primarily intended to store solid waste (tailings) from mining operations. The
dam may either be a conventional structure, behind which tailings are impounded, or be built
from the tailings themselves, using hydraulic ll construction methods.
Thrust fault A fault along which relative movement of the overhanging side of the fault occurs upward
(reverse displacement), and primarily due to a compressive tectonic stress regime.
Unbalanced free-eld boundary condition In a two-dimensional nite element model, this condi-
tion corresponds to non-symmetric absorbing lateral boundary conditions for the free-eld
condition, e.g., as implemented in the computer program SuperFLUSH. This assumption differs
from those used in the computer programs LUSH and FLUSH, where the left and right boundary
conditions are identical, due to the simplifying assumption of horizontal soil layers for the free-
eld condition. The unbalanced free-eld condition was developed to analyze structures embed-
ded in embankment dams, the geometry of which signicantly differs from a horizontally
layered system.
Waterstop Designates a at sheet of metal or PVC placed across a joint in a concrete dam to prevent
leakage through the joint.
References
Allen, C.R. and Cluff, L.S. 2000. Active Faults in Dam Foundations: An Update, Proc. 12th World
Conference on Earthquake Engineering, January 29February 5, Auckland, New Zealand, Vol. II,
paper no. 2490, Earthquake Engineering Research Institute, Oakland, CA.
American Concrete Institute. 1987. Mass concrete. ACI 207.1, American Concrete Institute, Detroit, MI.
American Concrete Institute. 2002. Building Code Requirements for Reinforced Concrete, ACI Standard
318-02, American Concrete Institute, Detroit, MI.
Anderson, R.J., Martin, P.P., and Wagner, C.D. 1996. A Comparison of the Predictions of Tellico Dams
Seismic Response by Simplied and Advanced Methods of Analysis, Proc. 16th Annual USCOLD
Lecture Series, Los Angeles, July 2226, pp. 3147.
Aubry, D., Hujeux, J.C., Lassoudire, F., and Meimon, Y., Eds. 1982. A Double Memory Model with
Multiple Mechanisms for Cyclic Soil Behavior, in International Symposium on Numerical Methods
in Geomechanics, Zurich, A.A. Balkema, Rotterdam, pp. 313.
Babbitt, D.H. and Verigin, S.W. 1996. General Approach to Seismic Stability of Embankment Dams, in
Earthquake Engineering for Dams, Western Regional Technical Seminar, Association of State Dam
Safety Ofcials (ASDSO), April 1112, Sacramento, pp. 197211.
Bathe, K.J. 1978. ADINA/BM: A General Computer Program for Nonlinear Analysis of Mines Structures,
Final report to Ofce of Assistant-Director of Mining, U.S. Department of the Interior, Contract
no. 30255008.
Beikae, M. 1996. A Seismic Displacement Analysis Technique for Embankment Dams, Proc. 16th Annual
USCOLD Lecture Series, Los Angeles, July 2226, pp. 91109.
Bolt, B.A. 1996. The Joint Synthesis of Seismic Acceleration, Velocity and Displacement, in Earthquake
Engineering for Dams, Western Regional Technical Seminar, Association of State Dam Safety Of-
cials, April 1112, Sacramento, pp. 6886.
Bozovic, A. and Markovic, M. 1999. Neotectonics and Dams: Guidelines, Committee on Seismic Aspects
of Dam Design, International Committee on Large Dams (ICOLD), Paris, France.
Brand, B. 1993. FERCs Evolving Policy on Three-Dimensional Stability Analysis of Concrete Gravity
Dams, in Proc. WaterPower 93, pp. 821830.
Bu, S. 2001. Seismic Evaluation of Concrete Gravity Dams Using FLAC, Proc. Second International
FLAC Symposium, Lyon-Ecully, France.
Bureau, G. 1985. Seismic Safety and Rehabilitation of Dam Inlet/Outlet Structures, 15th International
Congress on Large Dams, Lausanne, Switzerland, Q. 59, R. 17.
Bureau, G. 1993a. Seismic Safety Evaluation of Priest Dam, California, Proc. International Workshop
on Dam Safety Evaluation, Grindelwald, Switzerland, April 2628, Vol. 2, pp. 149160.
Bureau, G. 1993b. Seismic Safety of Intake/Outlet Towers, Proc. International Workshop on Dam Safety
Evaluation, Grindelwald, Switzerland, April 2628, Vol. 2, pp. 12291240.
Bureau, G. 1996a. Assessment of Conventional and Advanced Procedures for Seismic Safety Evaluation
of Concrete Gravity Dams, USCOLD Annual Lecture, July 2226, Los Angeles, CA.
Bureau, G. 1996b. Numerical Analysis and Seismic Safety Evaluation of Embankment Dams, Invited
lecture, Boston Society of Civil Engineers Section of American Society of Civil Engineers, Dam
Inspection, Analysis, and Rehabilitation Seminar, November 2, Waltham, MA, 54 pp.
Bureau, G. 1997. Evaluation Methods and Acceptability of Seismic Deformations in Embankment
Dams, 29th International Congress on Large Dams (ICOLD), Florence, Italy, May, Q. 73, R. 11,
pp. 175200.
Bureau, G. and Ballentine, G.D. 2002. A Comprehensive Seismic Vulnerability and Loss Assessment of
the State of South Carolina Using HAZUS. Part VI. Dam Inventory and Vulnerability Assessment
Methodology, 7th National Conference on Earthquake Engineering, July 2125, Boston, Earth-
quake Engineering Research Institute, Oakland, CA.
Bureau, G.J. and Scawthorn, C. 1986. Seismic Reevaluation of Lower Crystal Springs Outlet System, in
Seismic Evaluation of Lifeline Systems: Case Studies, Proceedings of a session sponsored by the
Technical Council on Lifeline Earthquake Engineering of the American Society of Civil Engineers,
Boston, October 27. Wang, L.R.L and Whitman, R.V., Eds., American Society of Civil Engineers,
New York.
Bureau, G. and Ghanaat, Y. 2000. Seismic Evaluation of a Historic Curved Gravity Dam, Proc. Dam
Safety 2000 Conference, Providence, RI, September 2629, Association of State Dam Safety Ofcials
(ASDSO).
Bureau, G. and Udaka, T. 1982. Seismic Interaction of Control Towers Embedded in Embankment
Dams, 8th European Conference on Earthquake Engineering, September, Athens, Greece, Vol. 6,
pp. 8390.
Bureau, G., Edwards, A., and Blmel, A.S. 1994. Seismic Design of Stage IV Raising, Los Leones Dam,
Chile, Proc. 11th Association of State Dam Safety Ofcials (ASDSO) Conference, Boston, Sep-
tember, Suppl., pp. 7786.
Bureau, G., Inel, S., Davis, C.A., and Roth, W.H. 1996. Seismic Response of Los Angeles Dam, During
the 1994 Northridge Earthquake, Proc. USCOLD Annual Meeting, Los Angeles, July 2226, pp.
281295.
Bureau, G., Volpe, R.L., Roth, W.R., and Udaka, T. 1985. Seismic Analysis of Concrete Face Rockll
Dams, in Concrete Face Rockll Dams: Design, Construction and Performance, ASCE International
Symposium on CFRDs, Detroit, Oct. 21, pp. 479508, American Society of Civil Engineers, New
York.
Bureau, G., Volpe, R.L., Roth, W.R., and Udaka, T. 1987. Seismic Analysis of Concrete Face Rockll
Dams, Closure, ASCE J. Geotechnical Eng. Div., 113(October), 10, pp. 12551264.
Castro, G. et al. 1987. On the Behavior of Soils during Earthquakes: Liquefaction, in Soil Dynamics and
Liquefaction, Cakmak, A.S., Ed., Elsevier, Amsterdam, pp. 169204.
Castro, G. et al. 1989. Re-Evaluation of the Lower San Fernando Dam. Report 1: An Investigation of the
February 9, 1971 Slide, Report no. GL-892, U.S. Army Corps of Engineers, Waterways Experiment
Station, Vicksburg, MS.
Chavez, J.W. and Fenves, G.L. 1993. Earthquake Analysis and Response of Gravity Dams Including Base
Sliding, EERC Report no. UCB/EERC-9307, December, Earthquake Engineering Research Center,
University of California, Berkeley.
Chopra, A.K. 1988. Earthquake Response Analysis of Concrete Dams, in Advanced Dam Engineering
for Design, Construction and Rehabilitation, Jensen, R.B., Ed., Van Nostrand Reinhold, New York,
pp. 416465.
Chopra, A.K. and Fok, K.L. 1984. Evaluation of Simplied Earthquake Analysis Procedures for Intake-
Outlet Towers, 8th World Conference on Earthquake Engineering, July, San Francisco, Vol. 7,
Earthquake Engineering Research Institute, Oakland, CA, pp. 467474.
Chopra, A.K. and Liaw, C.Y. 1975. Earthquake-Resistant Design of Inlet-Outlet Towers, ASCE J. Struc-
tural Division, 101 (ST7), 13491366.
Civil Systems, Inc. 1980. SuperFLUSH, Users manual, Vols. I to III, report to Kozo Keikaku Engineering,
Inc., Tokyo, October.
Clough, R.W. and Ghanaat, Y. 1993. Concrete Dams: Evaluation for Seismic Loading, Proc. International
Workshop on Dam Safety Evaluation, April 2628, Grindelwald, Switzerland, Vol. 4, pp. 137167.
Clough, R.W. and Niwa, A. 1982. Earthquake Simulator Research on Arch Dam Models, in Dynamic
Modeling of Concrete Structures, SP 735, American Concrete Institute, Detroit, pages 83105.
Coyne and Bellier/ECP/EDF-REAL 1991. GEFDYN, a Computer Program for Geomechanics Finite
Element Analysis: Two/Three-Dimensional Quasi-Static/Dynamic Coupled Mechanical-Hydraulics
Software for Nonlinear Geomaterials Analysis, Paris, February.
Cundall, P.E. 1976. Explicit Finite-Difference Methods in Geomechanics, 2nd International Conference
on Numerical Methods in Geomechanics, Blacksburg, VA, June.
Dames and Moore. 1991. Stability Evaluation: South Haiwee Dam, Inyo County, Report to Los Angeles
Department of Water and Power, July.
Davis, C.A. and Sakado, M.M. 1994. Response of the Van Norman Complex to the Northridge Earth-
quake, Proc. 11th Association of State Dam Safety Ofcials Conference, Boston, September, pp.
241255.
Dawson, E.M., Roth, W.H., Nesarajah, S., Bureau, G., and Davis, C.A. 2001. A Practice-Oriented Pore-
Pressure Generation Model, Proc. 2nd International FLAC Symposium, October, Lyon-Ecully,
France, Itasca Consulting Group, Minneapolis, MN.
Duncan, J.M., Seed, R.B., Wong, K.S., and Ozawa, Y. 1984. FEADAM84: A Computer Program for Finite
Element Analysis of Dams, Geotechnical Engineering Research Report no. SU/GT/8403, Depart-
ment of Civil Engineering, Stanford University, Stanford, CA, November.
Dungar, R. and Jackson, E.A. 1975. The Seismic Analysis of the Bellmouth Spillway and Valve Tower for
an Earth Dam, in Numerical Analysis of Dams, Naylor, Stagg and Zienkiewicz, Eds., pp. 604624.
Electric Power Research Institute (EPRI). 1992. Uplift Pressures, Shear Strengths, and Tensile Strengths
for Stability Analysis of Concrete Gravity Dams, Report no. EPRI TR100345, Electric Power
Research Institute, Palo Alto, CA.
Elgamal, A.M., Abdel-Ghaffar, A.M., and Prevost, J.H. 1984. Nonlinear Earthquake-Response Analysis
of Earth Dams, Report no. 84-SM-14, December, Princeton University, Princeton, NJ.
Federal Energy Regulatory Commission (FERC). 2000. Engineering Guidelines for the Evaluation of
Hydropower Projects, Ofce of Hydropower Licensing, Washington, D.C., June 2000 Peer Review
Copy.
Fenves, G. and Chopra, A.K. 1984. EAGD-84: A Computer Program for Earthquake Analysis of Concrete
Gravity Dams, Report no. UCB/EERC8411, August, Earthquake Engineering Research Center,
University of California, Berkeley.
Fenves, G.L., Mojtahedi, S., and Reimer, R.B. 1989. ADAP-88: A Computer Program for Static and
Dynamic Analysis of Arch Dams, Report no. UCB/EERC89/12, November, Earthquake Engineer-
ing Research Center, University of California, Berkeley.
Finn, W.D.L. 1991. Estimating How Embankment Dams Behave during Earthquakes, Water Power and
Dam Construction, London, April, pp. 1722.
Finn, W.D.L. and Yogendrakumar, M. 1989. TARA 3-FL: Program for Analysis of Liquefaction-Induced
Flow Liquefaction, University of British Columbia, Vancouver, Canada.
Finn, W.D.L., Gillon, M.D., Yogendrakumar, M., and Newton, C.J. 1992. Simulating the Seismic Response
of a Rockll Dam, Proc. NUMOG-4, A.A. Balkema, Rotterdam, pp. 379391.
Fok, K.L., Hall, J.F., and Chopra, A.K. 1986. EACD-3D Users Manual, Department of Civil Engineering,
University of California, Berkeley; Hydrodynamic and Foundation Flexibility Effects in Earth-
quake Response of Arch Dams, ASCE J. Struct. Eng., 112 (no. 8), 18101828.
Forrest, M., Bureau, G., Lazarte, C., and Hutchings, S. 2000. Seismic Rehabilitation of Weber Dam,
Proc. Association of State Dam Safety Ofcials (ASDSO) 2000 West Region Annual Conference
and Technical Seminar, Portland, OR, May 1519, pp. 289300.
Ghanaat, Y. and Clough, R.W. 1989. EADAP: Enhanced Arch Dam Analysis Program, Report no. UCB/
EERC8907, November, Earthquake Engineering Research Center, University of California, Ber-
keley.
Goyal, A. and Chopra, A.K. 1989. Earthquake Analysis and Response of Intake-Outlet Towers, Report
no. UCB/EERC89/04, July, Earthquake Engineering Research Center, University of California,
Berkeley.
Gupta, H.K. and Rastogi, B.K. 1976. Dams and Earthquakes, Elsevier, Amsterdam.
Hall, J.F. 1997. Efcient Numerical Analysis of Arch Dams: Users Manual for SCADA Smeared Crack
Arch Dam Analysis, Report no. EERL-96-01, Earthquake Engineering Research Laboratory, Cal-
ifornia Institute of Technology, Pasadena, CA.
Hall, R.L., Chowdhury, M.R., and Matheu, E.E. 2001. Seismic Testing of a 1/20-Scale 2D Model of Koyna
Dam, in Wind and Seismic Effects, Proc. 32nd Joint Meeting U.S.Japan Cooperative Program in
Natural Resources Panel on Wind and Seismic Effects, April, NIST SP 963, National Institute of
Standards and Technology, Gaithersburg, MD, pp. 131140.
Harris, D.W. 1986. Dynamic Effective Stress Finite Element Analysis of Dams Subjected to Liquefaction,
Report RECERC864, Embankment Dams Branch, Division of Dam and Waterway Design,
Engineering and Research Center, U.S. Department of the Interior, Bureau of Reclamation, Denver,
CO, December.
Harza Engineering Company. 1997. Criteria Evaluation Memorandum, Big Tujunga Dam Reanalysis
Study, Unpublished memorandum, prepared for Los Angeles County Department of Public
Works, January.
Hatton, J.W., Foster, P.F., and Thomson, R. 1991. The Inuence of Foundation Conditions on the Design
of Clyde Dam, Proc. 17th International Congress on Large Dams, Vienna, Austria, Q. 66, R. 10,
vol. III, pp. 157178.
Hudson, M., Idriss, I.M., and Beikae, M. 1994. QUAD4M: A Computer Program to Evaluate the Seismic
Response of Soil Structures Using Finite Element Procedures and Incorporating a Compliant Base,
May, Department of Civil and Environmental Engineering, University of California, Davis.
ICOLD (International Committee on Large Dams). 1988. Inspection of Dams Following Earthquakes:
Guidelines, Bulletin 62.
ICOLD (International Committee on Large Dams). 1992. Third Benchmark Workshop on Numerical
Analysis of Dams, Bergamo, Italy, International Committee on Large Dams and Italian Committee
on Large Dams.
ICOLD (International Committee on Large Dams). 1994. Fourth Benchmark Workshop on Numerical
Analysis of Dams, Paris, France, International Committee on Large Dams and French Committee
on Large Dams.
ICOLD (International Committee on Large Dams). 1999. Fifth Benchmark Workshop on Numerical
Analysis of Dams, June 25, Denver, CO, International Committee on Large Dams, Bureau of
Reclamation and U.S. Committee on Large Dams.
Idriss, I.M. 1999. An Update of the Seed-Idriss Procedure for Evaluating Liquefaction Potential, Proc.
TRB Workshop on New Approaches in Liquefaction Analysis, Washington, D.C., January 10.
Idriss, I.M. and Sun, J.I. 1992. Users Manual for SHAKE91, November, Department of Civil and
Environmental Engineering, University of California, Davis.
Inel, S., Roth, W.H., and De Rubertis, C. 1993. Nonlinear Dynamic Effective-Stress Analysis of Two Case
Histories, Session on Geotechnical Aspects of Recent Earthquakes, Proc. 3rd International Con-
ference on Case Histories in Geotechnical Engineering, St. Louis, MO, June 16.
Itasca Consulting Group. 1992. FLAC: Fast Lagrangian Analysis of Continua. vol. I. User's Manual; vol. II.
Verication Problems and Example Applications, Itasca Consulting Group, Minneapolis, MN.
Janbu, N. 1963. Soil Compressibility as Determined by Oedometer and Triaxial Tests, Proc. European
Conference on Soil Mechanics and Foundation Engineering, Wiesbaden, Germany, vol. 1,
pp. 1926.
Jansen, R.B. 1987. The Concrete Face Rockll Dam. Performance of Cogoti Dam under Seismic Loading,
Discussion of a paper presented at ASCE Symposium on Concrete Face Rockll Dams, ASCE J.
Geotech. Eng. Div., 113 (no. 10), 11351136.
Kuo, J.S.H. 1982. Fluid-Structure Interactions: Added Mass Computations for Incompressible Fluid,
Report no. UBC/EERC82/09, Report to National Science Foundation, August, University of Cal-
ifornia, Berkeley.
Leclerc, M., Lger, P., and Tinawi, R. 2002. CADAM, Users Manual Version 1.4.3, Ecole Polytechnique
de Montreal, Canada.
Lger, P. et al. 1997. Failure Mechanisms of Gravity Dams Subjected to Hydrostatic Overload: Inuence
of Weak Lift Joints, Proc. 19th International Congress on Large Dams, Florence, Italy, Q.75, R.2,
vol. IV, pp. 1137.
Lysmer, J., Udaka, T., Tsai, C.-F., and Seed, H.B. 1975. FLUSH: A Computer Program for Approximate
3-D Analysis of Soil-Structure Interaction Problems, Report no. EERC 7530, November, Earth-
quake Engineering Research Center, University of California, Berkeley.
Makdisi, F. and Seed, H.B. 1977. A Simplied Procedure for Estimating Earthquake-Induced Deforma-
tions in Dams and Embankments, EERC Report no. UCB/EERC77/19, Earthquake Engineering
Research Center, University of California, Berkeley.
Martin, G.R., Finn, W.D.L., and Seed, H.B. 1975. Fundamentals of Liquefaction under Cyclic Loading,
ASCE J. Geotech. Eng. Div., 101 (no. GT5), 423438.
Martin, P.P. and Niznik, J.A. 1993. Prediction of Static and Dynamic Deformation Response of Blue
Ridge Dam, Georgia, U.S.A., Proc. International Workshop on Dam Safety Evaluation, April 2628,
Grindelwald, Switzerland, vol. 2, pp. 133147.
Martin, P.P. and Seed, H.B. 1978. Apollo: A Computer Program for the Analysis of Pressure Generation
and Dissipation in Horizontal Sand Layers During Cyclic or Earthquake Loading, Report no.
UCB/EERC 7821, Earthquake Engineering Research Center, University of California, Berkeley.
Martin, P.P., Niznik, J.A., Kleiner, D.E., and Wagner, C.D. 1994. TVAs Seismic Safety Assessment Program
of Its Embankment Dams, 18th International Congress on Large Dams, Durban, South Africa,
Question 68.
Mejia, L.H. and Seed, H.B. 1983. Comparison of 2-D and 3-D Dynamic Analysis of Earth Dams, ASCE
J. Geotech. Eng., 109 (no. GT11), 13831398.
Mejia, L.H., Gillon, M., Freeman, S., and Berryman, K. 1997. Design Criteria for Fault Rupture at the
Matahina Dam, New Zealand, Int. J. Hydropower Dams, 4 (2), 120123.
Moriwaki, Y., Beikae, M., and Idriss, I.M. 1988. Nonlinear Analysis of the Upper San Fernando Dam
under the 1981 San Fernando Earthquake, Proc. 9th World Conference on Earthquake Engineer-
ing, Tokyo and Kyoto, Japan, vol. III, Earthquake Engineering Research Institute, Oakland, CA,
and Japanese Meteorological Agency, Tokyo, Japan, pp. 237241.
Newmark, N.M. 1965. Effects of Earthquakes on Dams and Embankments, Rankine Lecture,
Gotechnique 15 (2), 139160.
Ostrom, D.K. and Kelly, T.A. 1977. Method for Dynamic Testing of Dams, ASCE J. Power Div., 103 (no.
P01), 2736.
Pacelli, W.A., Andriolo, F.R., and Sarkaria, G.S. 1993. Treatment and Performance of Construction Joints
in Concrete Dams, Int. Water Power Dam Construction, 45 (11), 2631.
Prevost, J.H. 1981. DYNAFLOW: A Nonlinear Transient Finite Element Analysis Program, Princeton
University, Princeton, NJ.
QUEST Structures. 1993. GDAP: Graphics-Based Dam Analysis Program Users Manual, March,
prepared for U.S. Army Corps of Engineers, Vicksburg, MS.
Raphael, J.M. 1984. Tensile Strength of Concrete, ACI J. Technical Paper No. 8117, (March-April),
158165.
Romo, M.P. and Resendiz, D. 1981. Computed and Observed Deformations of Two Embankment Dams
under Earthquake Loading, in Dams and Earthquakes, Proceedings, Paper 30, Institute of Civil
Engineers, London, Thomas Telford, London, pp. 267274.
Roth, W.H., Bureau, G., and Brodt, G. 1991. Pleasant Valley Dam: An Approach to Quantifying the
Effect of Foundation Liquefaction, Proc. 17th International Congress on Large Dams, Vienna,
June, pp. 11991223.
Roth, W.H., Scott, R.F., and Cundall, P.A. 1986. Nonlinear Dynamic Analysis of a Centrifuge Model
Embankment, 3rd U.S. National Conference on Earthquake Engineering, August 2428, Charles-
ton, SC, vol. I, Earthquake Engineering Research Institute, Oakland, CA, pp. 506516.
Seed, H.B. 1983. Earthquake-Resistant Design of Earth Dams, Proc. ASCE Symposium on Seismic
Design of Embankments and Caverns, ASCE National Convention, Philadelphia, May 1620, pp.
4164, American Society of Civil Engineers, New York.
Seed, H.B. 1987. Design Problems in Soil Liquefaction, ASCE J. Geotech. Eng., 113 (8), 827845.
Seed, H.B. and Idriss, I.M. 1970a. Soil Moduli and Damping Factors for Dynamic Response Analysis,
Report no. EERC/7010, December, Earthquake Engineering Research Center, University of Cali-
fornia, Berkeley.
Seed, H.B. and Idriss, I.M. 1970b. A Simplied Procedure for Evaluating Soil Liquefaction Potential,
Report no. EERC 709, Earthquake Engineering Research Center, University of California, Berkeley.
Seed, H.B. and Idriss, I.M. 1982. Ground Motions and Soil Liquefaction During Earthquakes, Mono-
graph Series, Earthquake Engineering Research Institute, Berkeley, CA.
Seed, H.B., Idriss, I.M., and Arango, I. 1983. Evaluation of Liquefaction Potential Using Field Perfor-
mance Data, ASCE J. Geotech. Eng., 109 (3), 458482.
Seed, R.B. 1999. Engineering Evaluation of Post-Liquefaction Residual Strength, in New Approaches in
Liquefaction Analyses, Proc. TRB Workshop, January 10, Washington, D.C.
Seed, R.B. and Harder, L.F. Jr. 1990. SPT-Based Analysis of Cyclic Pore Pressure Generation and
Undrained Residual Strength, Proc. H. Bolton Seed Memorial Symposium, vol. 2, BiTech Pub-
lishers, Canada, May.
Serff, N., Seed, H.B., Makdisi, F.I., and Chang, C.K. 1976. Earthquake-Induced Deformations of Earth
Dams, EERC Report no. EERC/764, Earthquake Engineering Research Center, University of
California, Berkeley.
Sherard, J.L., Cluff, L.S., and Allen, C.R. 1974. Potentially Active Faults in Dam Foundations,
Gotechnique, 24 (3), 367428.
Siegel, R.A. 1975. Computer Analysis of General Slope Stability Problems; STABL Users Manual, Joint
Highway Research Project, Report No. JHRP-75-8, Project No. C-36-36K, File No. 6-14-11, Purdue
University, Lafayette, IN, June.
Somerville, P.G. and Graves, R.W. 1996. Strong Ground Motions of the Kobe, Japan Earthquake of
January 17, 1995, and Development of a Model of Forward Rupture Directivity Effects Applicable
in California, Proc. Association of State Dam Safety Ofcials Western Regional Technical Seminar,
April 1112, Sacramento, pp. 89108.
Somerville, P.G., Smith, N.F., Graves, R.W., and Abrahamson, N.A. 1995. Accounting for Near-Fault
Rupture Directivity Effects in the Development of Design Ground Motions, Proc. ASME/SSME
Conference, Hawaii, July, American Society of Mechanical Engineers, New York.
Swaisgood, J.R. 1995. Estimating Deformation of Embankment Dams Caused by Earthquakes, Associ-
ation of State Dam Safety Ofcials Western Regional Conference, Red Lodge, MT, May 2225.
Swaisgood, J.R. 1998. Seismically-Induced Deformations of Embankment Dams, 6th U.S. National
Conference on Earthquake Engineering, June, Seattle, WA, Earthquake Engineering Research Insti-
tute, Oakland, CA.
Swanson Analysis Systems, Inc. 1970. ANSYS, Houston, PA.
Tarbox, G.S., Dreher, K.J., and Carpenter, L.R. 1979. Seismic Analysis of Concrete Dams, Thirteenth
International Congress on Large Dams, New Delhi, Q. 51, R. 11, pp. 963994.
Tokimatsu, K. and Yoshimi, Y. 1983. Empirical Correlation of Soil Liquefaction Based on SPT N-Value
and Fines Content, Soils and Foundations, 23 (4), 5674.
Uniform Building Code. 1997. With State of California Amendments, International Conference of Building
Ofcials, Whittier, CA.
URS Corporation. 2002. Magalia Dam and Reservoir, Feasibility Study to Modify Restricted Reservoir
Level, Report to Paradise Irrigation District, January, Paradise, CA.
URS Corporation et al. 2002. Comprehensive Seismic Risk and Vulnerability Study for the State of South
Carolina, Report to the South Carolina Emergency Preparedness Division and HAZUS-Compat-
ible CD.
U.S. Army Corps of Engineers. 1989. National Inventory of Dams (NID), April 2000 update, U.S. Army
Corps of Engineers, Washington, D.C.
U.S. Bureau of Reclamation. 1983. Safety Evaluation of Existing Dams (SEED Manual), U.S. Depart-
ment of the Interior, Denver, CO.
U.S. Bureau of Reclamation. 1997. Guidelines for Achieving Public Protection in Dam Safety Decision-
Making, April, Denver, CO.
USCOLD (U.S. Committee on Large Dams). 1984. Bibliography on Performance of Dams during
Earthquakes, compiled by Philip Gregory, University of California, Berkeley.
USCOLD (U.S. Committee on Large Dams). 1986. Bibliography on Reservoir-Induced Seismicity.
Committee on Earthquakes, December, Denver, CO.
USCOLD (U.S. Committee on Large Dams). 1989. Strong Motion Instruments at Dams: Guidelines for
their Selection, Installation, Operation and Maintenance, U.S. Committee on Large Dams, Denver,
CO.
USCOLD (U.S. Committee on Large Dams). 1992a. Directory of Computer Programs in Use for Dam
Engineering in the United States, Committee on Methods of Numerical Analysis of Dams, March,
Denver, CO.
USCOLD (U.S. Committee on Large Dams). 1992b. Observed Performance of Dams during Earth-
quakes, Committee on Earthquakes, July, Denver, CO.
USCOLD (U.S. Committee on Large Dams). 1995. Guidelines for Earthquake Design and Evaluation
of Structures Appurtenant to Dams, Committee on Earthquakes, May, Denver, CO.
USCOLD (U.S. Committee on Large Dams). 1997. Reservoir Triggered Seismicity, Committee on
Earthquakes, April, Denver, CO.
USCOLD (U.S. Committee on Large Dams). 1999. Updated Guidelines for Selecting Seismic Parameters
for Dam Projects, Committee on Earthquakes, April, Denver, CO.
USCOLD (U.S. Committee on Large Dams). 2000. Observed Performance of Dams during Earthquakes,
vol. II, Committee on Earthquakes, October, Denver, CO.
Vaid, Y.P. and Thomas, J. 1994. Post-Earthquake Liquefaction Behavior of Sand, Proc. 13th International
Conference on Soil Mechanics and Foundation Engineering, New Delhi, India.
Von Thun, L. and Harris, C.W. 1981. Estimation of Displacements of Rockll Dams Due to Seismic
Shaking, Proc. International Conference on Recent Advances in Geotechnical Engineering and
Soil Dynamics, St. Louis, MO, April 26May 3, vol. I, pp. 417423.
Vrymoed, J.L. 1996. Seismic Safety Evaluation of Two Earth Dams, in Earthquake Engineering for Dams,
Proc. Western Regional Technical Seminar, Association of State Dam Safety Ofcials, April 1112,
Sacramento, Association of State Dam Safety Ofcials, Lexington, KY, pp. 215234.
Wall, J.S. et al. 1986. Testing of Bond between Fresh and Hardened Concrete, Proc. RILEM and LCPC
International Symposium on Adhesion between Polymers and Concrete, Aix-en-Provence, France,
September 1619, Chapman & Hall, London, pp. 335344.
Westergaard, H.M. 1933. Water Pressures on Dams during Earthquakes, Trans. ASCE, 98(paper
no. 1835), 418472.
Wilson, E.L. and Abibullah, A. 1988. SAP90: A Series of Computer Programs for the Static and Dynamic
Finite Element Analysis of Structures, Computers and Structures, Inc., Berkeley, CA.
Wright, S.G. 1992. UTEXAS3 Version 1.2, a Computer Program for Slope Stability Calculations, Shinoak
Software, Austin, TX.

0068 c26

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

0068 c26

Încărcat de

Drepturi de autor:

Formate disponibile

2003 by CRC Press LLC

related to the initial state of static shear,

= 0.64 arc sin (N

is the effective overburden pressure; N

S-ar putea să vă placă și