Documente Academic
Documente Profesional
Documente Cultură
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
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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.
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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.
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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.
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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.
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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)
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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.
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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.
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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.
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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
2003 by CRC Press LLC
SOFTbank E-Book Center Tehran, Phone: 66403879,66493070 For Educational Use.
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 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