United States Society on Dams

Analysis of Seismic Deformations

of Embankment Dams

August 2022

Prepared by the USSD Committee on Earthquakes

Sub-Committee on Seismic Deformation Analysis of Embankment Dams

iii
 U.S. Society on Dams
USSD, as the United States member of the International Commission on Large Dams, is
dedicated to:

COMMUNITY AND CONNECTIONS

• Strengthen and promote connection to ICOLD.
• Create Opportunities and mechanisms for higher education institutions and researchers to share
advances in the field.
• Create networking and mentoring opportunities for professionals to share, grow, and access
information for collaboration and encourage the free flow of information.
ADVOCACY AND AWARENESS
• Build awareness of the challenges, values, and benefits of dams and levees to the public
• As a profession, let people know who we are, what we do, and why.
• Generate, join and support coalitions which are aligned with USSD’s mission and values.
RECRUITMENT AND GROWTH
• Foster career paths that will grow the talent pool for our industry.
• Strengthen diversity, equity and inclusion within the community of practice.
• Grow USSD through outreach beyond our membership.
EDUCATION AND PROFESSIONAL DEVELOPMENT
• Identify advancements in dams and levees and needs for training.
• Develop and deliver industry-relevant education and training.
• Create a structured curriculum for dam and levee professionals.

The information contained in this report regarding commercial products or firms may not be used for
advertising or promotional purposes and may not be construed as an endorsement of any product or firm by
the United States Society on Dams. USSD accepts no responsibility for the statements made or the opinions
expressed in this publication.

Copyright © 2022 U. S. Society on Dams

Printed in the United States of America

ISBN 979-8-9870559-1-5

U.S. Society on Dams

13918 E. Mississippi Ave., #61160
Aurora, CO 80012
info@ussdams.org
Phone: 303-792-8753
Fax: 303-792-8782
www.ussdams.org

2
 FOREWORD

Observations of dam performance during earthquakes over the past 50 years have focused
the attention of the dam engineering profession on the importance of reliable estimates of
seismically induced deformations of embankment dams. It is now widely recognized that
earthquake damage of an embankment dam is closely related to the dam’s earthquake-
induced deformations. It is also known that reliable assessment of future seismic
performance requires appropriate evaluation of expected dam seismic deformations.
Thus, over the past few decades, there has been a marked increase in the use of numerical
analysis in engineering practice for estimating seismic deformations of embankment
dams.

This report is intended to provide guidance on the evaluation of seismic deformations of

embankment dams and on the use of numerical analysis procedures for such evaluations.
It is intended to serve as a reference on available analysis methods for consultants,
regulators, and owners involved in performing, reviewing, and procuring analyses of
seismic stability and deformations of embankment dams. Because engineering practice in
this field will likely continue to evolve rapidly, this guidance is expected to require
periodic review and updating in the future.

The document was prepared by the Committee on Earthquakes of the United States
Society on Dams (USSD) and provides a consensus view of the Committee on current
United States practice for analysis of embankment dam seismic deformations. It
culminates the efforts of the USSD Earthquakes Committee over several years. A
subcommittee consisting of Drs. Lelio Mejia (Chair), Sam Abbaszadeh, Richard
Armstrong, Michael Beaty, and Jack Montgomery, reviewed previous drafts of the
document, reworked the document outline to reflect current practice direction and needs,
and prepared the document in its current form. Committee member Dr. Kate Darby
worked with the subcommittee to compile review comments, revise figures, combine
edits, and review the finished document. The subcommittee is very thankful to Dr. Darby
for her input and help with completing the document.

The subcommittee is grateful to the committee members who worked on early versions of
the document including Mr. Donald Babbitt, Dr. Michael Beaty, Dr. Mohsen Beikae,
Professor Ross Boulanger, Mr. Joseph Ehasz, and Dr. Faiz Makdisi. The contribution of
the following reviewers who provided helpful comments on the draft of this document is
also gratefully acknowledged: Mr. Donald Babbitt, Professor Ross Boulanger, Professor
I. M. Idriss, Mr. Richard Kramer, Dr. Faiz Makdisi, Ms. Erica Medley, Dr. Nicolas
Oettle, Dr. Nicholas Paull, and Mr. Paul Ridlen.

Finally, the Committee and Subcommittee wish to thank the members of the USSD
Publication Review Committee (PRC), Dr. Dean Durkee, Professor Jonathan Bray, and
Dr. Alan Rauch, who reviewed the final draft of the document and provided thoughtful
and constructive comments. Dr. Dean Durkee served as Chair of the PRC and
coordinated the review on behalf of the USSD Board of Directors.

iii
The Committee welcomes feedback from the dam engineering community as the
document is used in practice and welcomes comments for improvement and updating of
the guidance in future editions as practice evolves.

Subcommittee on Seismic Deformation Analysis of Embankment Dams

Dr. Lelio Mejia, Chair; Dr. Sam Abbaszadeh; Dr. Richard Armstrong; Dr. Michael
Beaty; Professor Jack Montgomery

USSD Committee on Earthquakes

Dr. Lelio Mejia, Chair; Professor Jack Montgomery, Vice Chair; Dr. Chris Krage, Y.P.
Vice Chair

iv
 TABLE OF CONTENTS

1.0 Introduction .............................................................................................................. 1

1.1 Background .......................................................................................................... 1
1.2 Purpose and Scope ............................................................................................... 1
1.3 Report Organization ............................................................................................. 2
2.0 Seismic Performance of Dams ................................................................................. 2
2.1 Potential Effects of Earthquakes on Embankment Dams..................................... 2
2.2 Case History Examples ........................................................................................ 3
2.3 Potential Failure Modes Initiated by Shaking-Induced Dam Deformation........ 10
3.0 Overview of Deformation Analysis Approaches ................................................... 11
3.1 Historical Development...................................................................................... 11
3.2 Types of Deformation Analysis Approaches ..................................................... 14
3.3 General Analysis Steps....................................................................................... 18
4.0 Simplified Analyses ............................................................................................... 24
4.1 Overview ............................................................................................................ 24
4.2 Empirically Based Correlations ......................................................................... 25
4.3 Analytically Based Correlations......................................................................... 27
4.4 Flow Slide Analyses ........................................................................................... 28
5.0 Sliding Block Analyses .......................................................................................... 29
5.1 Overview ............................................................................................................ 29
5.2 Basis and Limitations ......................................................................................... 30
5.3 Common Steps to Perform a Sliding-Block Analysis ........................................ 32
5.4 Step 1 – Perform Limit-Equilibrium Slope Stability Analysis .......................... 33
5.5 Step 2 – Develop Average Acceleration Histories ............................................. 34
5.6 Step 3 – Double-integration of acceleration history........................................... 36
6.0 Nonlinear Deformation Analyses .......................................................................... 37
6.1 Overview ............................................................................................................ 37
6.2 Framework for Nonlinear Analysis .................................................................... 37
6.3 Approaches to NDA ........................................................................................... 40
6.4 Analysis Process ................................................................................................. 47

v
 6.5 Special Considerations ....................................................................................... 50
6.6 Future Directions ................................................................................................ 52
7.0 Review and Documentation of Analysis Results ................................................... 53
7.1 Overview ............................................................................................................ 53
7.2 General Approach to Review and Documentation ............................................. 54
8.0 Considerations on Use of Deformation Analyses .................................................. 55
8.1 Objectives ........................................................................................................... 55
8.2 General Considerations ...................................................................................... 55
8.3 Additional Considerations for NDA .................................................................. 56
9.0 Concluding Remarks .............................................................................................. 56
10.0 References .............................................................................................................. 59

Appendices

Appendix A: Selected Analytically Based Correlations for Seismic Deformation Analysis

Appendix B: Key Features of Selected Constitutive Models from US Practice

Appendix C: Review and Documentation Considerations Specific to Analysis Approach

vi
 1.0 Introduction
1.1 Background

Reliable analysis of earthquake-induced deformations has been recognized since the mid
1960’s as a key element of the seismic stability evaluation of embankment dams.
Analyses of earthquake-induced deformations are now typically used to assess the
expected seismic performance of embankment dams. Furthermore, expected seismic
performance, expressed in terms of acceptable deformations, is commonly used as a
criterion to guide the seismic design of new embankment dams and the seismic retrofit of
existing dams.

Assessing the expected seismic performance of embankment dams often requires

evaluating potential earthquake damage to a significant degree of refinement (e.g.,
evaluating the spatial distribution and magnitude of dam deformation). Because
earthquake damage of an embankment dam is often closely related to seismically induced
deformations of the dam, reliable evaluation of expected seismic performance of a dam
generally requires appropriate evaluation of the magnitude and pattern of such
deformations.

The state of engineering practice for the seismic stability evaluation of embankment dams
has advanced considerably over the past several decades. Major advances have been
made during this time in the understanding of soil behavior under cyclic loading and in
the analysis of embankment dam seismic deformations. Since the introduction of the
sliding block method of analysis by Newmark (1965), progress in the evaluation of
embankment seismic deformations has been fostered by developments in the finite
element and finite difference methods of analysis and in the formulation of constitutive
models of soil mechanical behavior.

It is now widely recognized by the profession that reliable estimates of seismic

deformations are essential to the evaluation of the seismic stability of dams. Numerical
analyses of embankment seismic deformations are now routinely applied in the design of
new dams and in the evaluation of existing dams, and nonlinear effective stress analyses
are gaining wide appreciation within the engineering profession as a useful tool for the
seismic stability analysis of embankment dams.

1.2 Purpose and Scope

The USSD Earthquakes Committee has identified a need for guidance within the
practicing engineering community regarding the evaluation of seismic deformations of
embankment dams. This guidance document was prepared in response to that need. The
Committee envisions this document serving as a reference to consultants, regulators, and
owners involved in performing, reviewing, and procuring analyses of seismic stability
and deformations of embankment dams.

1
These guidelines are intended to serve as a primer regarding the selection of analysis
methods, the implementation of analysis procedures, and the documentation and
interpretation of results. They provide an overview of current approaches for the
evaluation of seismic deformations of embankment dams and best practices for the
numerical analysis of seismic deformations and the interpretation of analysis results. A
range of tools, which may be used in a phased analysis program, are discussed. Such
tools include screening evaluations, simplified analyses, and nonlinear deformation
analyses.

The focus of the guidelines is U.S.-centric as they attempt to provide a consensus view of
current U.S. practice as opposed to general international practice. The Committee
recognizes that international practice has contributed to U.S. practice over the years, and
vice versa, and hopes that this document will foster continued collaboration efforts.
Although a review of international practice is beyond the scope of these guidelines, it is
hoped that they will spur preparation of an analogous summary for international practice.

Readers should note that the document is not intended to be: a) an instruction manual on
procedures for seismic deformation analysis, b) a comprehensive text on the evaluation of
the seismic stability or seismic safety of embankment dams, c) a guide for the
engineering characterization of embankments and their foundations for analysis of
stability or seismic deformations, or for the selection of design earthquake ground
motions, d) a reference on the assessment of consequences of seismic deformations or of
potential failure modes, or e) a reference on risk reduction measures or on dam seismic
design.

1.3 Report Organization

In Chapter 2, selected case histories of dam performance during past earthquakes are
described to provide a perspective on how seismic loading may affect the stability of
embankment dams. This is followed in Chapter 3 by an overview of current analysis
approaches, and in Chapters 4, 5, and 6 by a description of selected methods of analysis,
including simplified methods, sliding block methods of analysis, and methods for
nonlinear analysis of seismic deformations.

Chapter 7 provides recommended best practices on the evaluation, review and

documentation of analysis results, depending on the choice of method of analysis. In
Chapter 8, the document discusses special considerations regarding the use of
deformation analysis for evaluating the seismic stability and safety of embankment dams.
A summary of key takeaway points from the document is provided in Chapter 9.

2.0 Seismic Performance of Dams

2.1 Potential Effects of Earthquakes on Embankment Dams

Post-earthquake observations have shown that dams may be affected by earthquakes in

multiple ways. Potential consequences of earthquakes on embankment dams were

2
discussed originally by Sherard et al. (1963), Sherard (1967), and summarized by Seed
(1973) amongst others. Some of these effects are listed below.

1. Slope failure of the embankment

2. Sliding of the embankment through its foundation
3. Embankment deformation and settlement
4. Embankment cracking
5. Shearing of the embankment and its foundation by surface fault rupture
6. Differential tectonic displacement of the reservoir and the dam foundation
7. Reservoir rim land-sliding leading to overtopping
8. Failure of the spillway or outlet works

Ways in which earthquakes can affect dams have since been investigated and refined
considerably, and the interdependency between components of a dam is well recognized.
For example, earthquake effects on a dam leading to distress (or failure) of the outlet
works (and vice versa) are now typically considered in evaluating the seismic safety of
embankment dams.

This report, however, is focused on the seismic deformation analysis of embankment

dams and does not address potential consequences of the embankment deformations (e.g.,
dam overtopping or cracking) or mitigation measures (e.g., reservoir level restrictions). A
brief discussion of the consequences of embankment deformations and of potential
failure modes that can be initiated by earthquakes is provided in Section 2.3 below, and
more extensive discussions are presented by USBR (2011), and USBR and USACE
(2019).

Whether the aforementioned effects lead to dam failure and reservoir uncontrolled release
depends partly on how intensely they manifest themselves in a dam during and after an
earthquake. The severity of some of these effects will be closely related to the
embankment deformations induced by the earthquake. Such is the case, for example, with
the severity of embankment crest settlement or of embankment cracking. The following
case histories illustrate the close relation between embankment dam damage and
earthquake-induced deformations.

2.2 Case History Examples

Five case histories are described below as examples of the extent of embankment dam
damage and deformation that may be induced by strong earthquake shaking. These case
histories are:

1. The near failure of Lower San Fernando Dam during the moment magnitude
(Mw) 6.6 1971 San Fernando Earthquake,

3
 2. The satisfactory performance of Los Angeles Dam during the Mw 6.7 1994
Northridge Earthquake,
3. The heavy damage of Austrian Dam during the Mw 6.9 1989 Loma Prieta
Earthquake,
4. The acceptable performance of Aratozawa Dam during the Mw 6.9 2008
Iwate-Miyagi-Nairiku Earthquake, and
5. The failure of Fujinuma Dam during the great Mw 9 2011 Tohoku Earthquake.
The primary source of information for these case histories is the three-volume series of
reports by the USSD Earthquakes Committee on observed performance of dams during
earthquakes (USCOLD, 1992, 2000; USSD, 2014).

2.2.1 Lower San Fernando Dam, California (1971)

Lower San Fernando Dam developed a major slide in its upstream slope and crest from
shaking by the February 9, 1971, Magnitude Mw 6.6 San Fernando earthquake. Aerial
views of the dam before and after the earthquake are shown in Figure 2-1. The
liquefaction-induced slide collapsed one of two outlet towers and nearly caused
uncontrolled release of the reservoir. The earthquake epicenter was about 7 miles from
the dam. Based on the postulated extent of the fault rupture plane (NOAA, 1973), the
closest distance from the dam to the fault rupture is estimated to have been less than one
mile.

(a) (b)
Figure 2-1. Aerial Views of Lower San Fernando Dam (a) Before and (b) Days After the
1971 San Fernando Earthquake (one of two outlet towers collapsed and lies below water).

Initially constructed between 1912 and 1916, the embankment was founded on stiff clay
alluvium with lenses of sand and gravel. Most of the original embankment was placed by
hydraulic filling, a process that resulted in upstream and downstream shells of loose to
medium dense sands and silts and a central core of clayey soil. The embankment was
raised a few times between 1917 and 1930, by placement of rolled fills, to a maximum
height of about 135 feet. A thin blanket of compacted shale and gravelly material was
placed on the lower downstream slope in 1930 and a rolled fill berm was added for
stability on the downstream slope in 1940.

4
The earthquake caused liquefaction of the lower upstream hydraulic fill shell such that
large blocks of the upstream shell moved into the reservoir, riding over liquefied soil. The
latter was found to have extruded upwards, between nearly intact blocks of the shell, and
to have flowed upstream up to 250 feet from the dam upstream toe. The downstream shell
exhibited settlements and horizontal displacements of up to about one foot but remained
stable after the earthquake. This case history was pivotal to engineering practice in the
U.S. because it highlighted the vulnerability of hydraulic fill dams to gross instability
during strong earthquake shaking. Additional information on the dam’s earthquake
performance is provided by USCOLD (1992), and a detailed study of the dam failure was
presented by Seed et al. (1973).

The case history was also extensively studied by Seed et al. (1988) and Castro et al.
(1992) to back-calculate the residual strength of the liquefied hydraulic fill. It has also
been used in several studies as an important data point for empirical correlations between
post-liquefaction residual strength and soil penetration resistance (e.g., Seed and Harder,
1990; Idriss and Boulanger, 2007).

2.2.2 Los Angeles Dam, California (1994)

Los Angeles Dam was completed in 1979 to replace the Upper and Lower San Fernando
dams which were damaged by the 1971 San Fernando earthquake. The dam impounds a
10,000-acre-foot reservoir. It is a modern embankment with a maximum height of 155
feet, founded on sedimentary rock. The embankment has upstream and downstream
shells of silty sand, a silty clay core, a central vertical chimney drain, and a downstream
near-horizontal drainage blanket. As shown in Figure 2-2, the dam crest is 30-feet wide
and the slopes have inclinations of about 3:1 downstream and 3.5:1 upstream. The
upstream face of the dam is lined with 3 inches of asphalt concrete.

Figure 2-2. Maximum Cross Section of Los Angeles Dam (Davis and Bardet, 1996).

The epicenter of the January 17, 1994, Mw 6.7 Northridge Earthquake was 6 miles from
the dam. The earthquake’s focal depth was 12 miles and the fault rupture extended
beneath the dam (Wald et al. 1996) such that the closest distance from the fault plane to
the dam was about 5 miles. Measured peak ground accelerations at the dam during the
earthquake are listed Table 2-1.

5
 Table 2-1. Recorded Peak Accelerations at Los Angeles Dam During 1994 Northridge
Earthquake.
Peak Acceleration (g)
Location
Transverse Longitudinal Vertical
Crest 0.60 0.42 0.38
Foundation 0.27 0.32 0.12
Right Abutment 0.42 0.33 0.32

The embankment underwent a maximum crest settlement of 3.5 inches and about one
inch of horizontal movement. Seepage quantities and piezometer levels in the dam
increased noticeably but returned to normal within a short time after the earthquake. The
earthquake caused minor cracking of the upstream asphalt concrete liner and of the dam
crest pavement. However, these cracks did not extend into the embankment materials.
The dam exhibited relatively minor damage under strong ground motion and its
performance was deemed satisfactory. Additional details on the case history are provided
by USCOLD (2000).

2.2.3 Austrian Dam, California (1989)

Austrian Dam was strongly shaken by the October 17, 1989, Mw 6.9 Loma Prieta
earthquake. The dam is located about 0.4 miles northeast of the San Andreas fault zone.
The epicenter of the earthquake was 7 miles south of the dam, but the fault ruptured
northward past the site and small fault surface displacements were reported on strands of
the Sargent Fault several hundred feet northeast of the dam right abutment (Aydin et al,
1992). The closest distance from the dam to the fault rupture was less than 1 mile. The
dam impounds a 6,200-acre-foot, water-supply reservoir. The reservoir level was about
100 feet below the dam crest at the time of the earthquake.

Constructed in 1949-1950, the dam is a 200-foot-high embankment with a crest length of

700 feet. It is a nearly homogeneous, gravelly, clayey sand embankment, compacted to
approximately 90 percent of ASTM D-1557 maximum density. Essentially all soils and
highly weathered rock were removed from the dam foundation prior to the embankment
construction. Excavation for repairs after the earthquake disclosed that strip drains
beneath the downstream zone had been blocked by waste materials.

Ground movements and surface cracks were observed on the downstream right abutment
above the spillway chute, on the upstream left abutment between the dam crest and the
inclined outlet, and along the right side of the reservoir upstream of the dam (Figure 2-3).
Aydin et al. (1992) suspected that the cracks upstream of the dam may have been
associated with an unmapped strand of the Sargent fault. However, based on trenching,
the dam owner's consultants concluded that the cracks were induced by ground shaking
(Wahler Associates, 1990).

Embankment crest movements caused by the earthquake resembled the dam’s long-term
settlement pattern. The maximum seismic settlement was 2.8 feet. Horizontal movement
was 1.1 feet downstream, near the spillway wall on the right abutment. The maximum
upstream movement was 0.5 feet and occurred at the left quarter point of the

6
embankment. The embankment movements were concluded to be due to settlement and
spreading of the fill during the strong earthquake shaking, followed by downslope creep
(Harder et al., 1998).

Figure 2-3 shows a plan view of the dam and the zones of cracking observed at the dam
after the earthquake. Longitudinal cracks up to 14 feet deep occurred within the upper 25
percent of the upstream and downstream faces. Shallower longitudinal cracks were found
on much of the downstream face. Crest cracking was confined to the abutment contact
areas. A transverse crack was traced to a depth of 30 feet down the left abutment, where
the dam had been constructed on weathered, highly fractured rock. On the right abutment,
transverse cracking and embankment separation from the spillway structure occurred to a
depth of 23 feet. Additional information on the case history is presented in USCOLD
(1992) and Harder et al. (1998). The case history was recently used by Boulanger (2019)
to examine the dam’s seismic deformations with nonlinear analysis methods.

Figure 2-3. Plan View of Austrian Dam with Areas of Cracking Observed After 1989
Loma Prieta Earthquake (after Wahler Associates, 1990).

2.2.4 Aratozawa Dam, Japan (2008)

Aratozawa Dam is a 74-m-high, zoned earth and rockfill embankment founded on rock,
with an upstream slope of 2.7:1 and a downstream slope of 2:1. Figure 2-4 shows a
transverse cross section of the dam. The embankment has a 10-m-wide and 414-m-long
crest and a volume of 3.05 million m3. It was completed in 1991, placed in service in
1998, and impounds a reservoir of 14,139 million m3.

7
 Figure 2-4. Transverse Cross Section of Aratozawa Dam (from PWRI, 2008).

The reservoir level was 6 m below the spillway crest when the Mw 6.9 Iwate-Miyagi-
Nairiku Earthquake of June 14, 2008 occurred. The earthquake’s epicenter was centered
15 km from the dam and the fault rupture passed near the dam right abutment (Aydan,
2016). During the main shock, strong motion instrumentation at the dam recorded the
peak accelerations listed in Table 2-2.

Table 2-2. Recorded Peak Accelerations at Aratozawa Dam During 2008 Iwate-Miyagi-
Nairiku Earthquake.
Peak Acceleration (g)
Location Elevation (m)
Transverse Longitudinal Vertical
Crest 278.5 0.54 0.46 0.63
Mid core 250.0 0.55 0.49 0.48
Bottom gallery 203.7 1.03 0.92 0.70

The crest above the clay core settled up to 400 mm and the rockfill shells settled a
maximum of 200 mm. A 200-mm-high step formed at the embankment/spillway contact.
The upstream shell deformed laterally towards the reservoir between 24 and 43 mm.
Despite the deformations, there was no apparent damage to the hand-placed rock facing
on the downstream or upstream sides of the reservoir.

Small openings were caused by the earthquake at the joints between blocks of the
inspection gallery beneath the impervious core zone. Approximately 1 liter per minute of
leakage was observed through fine cracks in the gallery ceiling, and at the boundary
between the upstream side wall and the gallery invert. The gallery was dry before the
earthquake. The measured leakage declined steadily after the earthquake.

Strong motion data from the main shock and 185 aftershocks demonstrated that large
strains induced by the strong motion reduced the shear wave velocity of the core material,
but that such velocity recovered with time. These strains led to significantly increased
pore pressures in the core. At a piezometer in the lower part of the core, about 42% of the
excess pore pressure dissipated in one day and the remainder dissipated in 3 months.

8
This case history is particularly useful for analytical model validations applicable to
rockfill embankments with thin compacted clay cores. The dam is one of a few large
earth-core rockfill dams that is well instrumented and has been subjected to strong
shaking from a large magnitude earthquake. Additional information on the case history is
presented in USSD (2014).

2.2.5 Fujinuma Dam, Japan (2011)

Fujinuma Dam failed due to the March 11, 2011, Mw 9.0 Tohoku Earthquake. The dam is
located 80 kilometers from the 600-kilometer-long earthquake subduction rupture. The
nearest strong motion seismograph, located 2.8 kilometers away from the dam, registered
a peak ground acceleration of 0.315g. The duration of shaking with accelerations greater
than 0.05 g exceeded 100 seconds. The reservoir was nearly full when the earthquake
occurred. Reportedly, an upstream slide lowered the crest over a significant length and
the damaged dam was overtopped for approximately 20-25 minutes immediately after the
earthquake, before it breached completely. Figure 2-5 shows a view of water overtopping
the dam embankment after the earthquake. A more detailed description and examination
of the failure is presented by Harder et al. (2011), JCOLD (2012) and Pradel et al. (2013).

Figure 2-5. Water Overtopping Fujinuma Dam After the 2011 Tohoku Earthquake (photo
by M. Yoshizawa).

The dam had a maximum height of about 18.5 meters and, together with a 6-m-high
auxiliary dam, retained a reservoir with a volume of approximately 1.5 million cubic
meters (~1,200 acre-feet). Construction of Fujinuma Dam began in 1937 but was halted
during World War II and the dam was completed in 1949. The main dam had a crest
width of 6 meters and an upstream slope ranging from 2.5H:1V to 2.8H:1V, to a
relatively steep 1.5H:1V upper slope. The downstream face had a slope of 2.5H:1V and

9
steepened at the downstream toe. The auxiliary dam is believed to have had a similar
geometry.

Limited information on the design of the dam is available. Some reports indicate the dam
was a homogeneous embankment, although a post-failure inspection disclosed that there
were several types of fill within the embankment which may reflect the intermittent
construction. At least some of the fill was placed in relatively thick lifts, so the degree of
compaction may have been relatively low. The inspection also indicated that the
foundation did not appear to have been completely cleared and grubbed and that an
organic residual soil was left in place.

A large slide also occurred in the upstream slope of the auxiliary dam, perhaps during the
earthquake shaking or, alternatively, because of rapid reservoir drawdown after the main
dam breach.

2.3 Potential Failure Modes Initiated by Shaking-Induced Dam Deformation

The above case histories illustrate some of the potential behaviors and failure modes of
embankment dams due to seismic loading. Potential failure modes of embankment dams
that can be initiated by earthquake-induced deformations of a dam include, amongst
others:

1. Overtopping due to loss of freeboard by crest settlement

2. Erosion and piping through embankment cracks
3. Erosion and piping into or along damaged outlet works or spillway structures
4. Structural collapse due to sliding within embankment and/or foundation
It is important to note that dam failure may also be initiated by earthquake effects other
than embankment deformations, such as overtopping due to loss of operation of the
spillway or outlet works or due to seiches in the reservoir. In addition, dam deformations
(or other effects) that may trigger emergency response, loss of function, or disruption of
operations are typically of concern even if they do not lead to uncontrolled release of
reservoir water.

Although the dam did not release its reservoir, the slide in the upstream shell of the
Lower San Fernando Dam during the 1971 San Fernando Earthquake is a clear example
of how structural collapse due to sliding within an embankment could lead to dam
overtopping and immediate reservoir release. Furthermore, this case history illustrates
how embankment deformations may cause loss of outlet works, which in turn could lead
to a dam breach by overtopping.

Failure of Fujinuma Dam after the 2011 Tohoku Earthquake appears to have been
initiated by large embankment deformations associated with either embankment sliding
or crest slumping. Deformation of Austrian Dam during the 1989 Loma Prieta
Earthquake led to major cracking of the embankment, which could have led to failure of
the dam by internal erosion had the reservoir been full at the time of the earthquake.

10
The preceding case histories (and numerous others available in the technical literature)
illustrate how earthquake-induced deformations can initiate some of the above failure
modes, and possibly others, and lead to breach of an embankment dam. To evaluate the
likelihood that these and other failure modes might be initiated in an embankment dam
and assess their likely progression requires reliable estimates of the deformations that
may be induced in the dam by an earthquake. Reliable deformation estimates are also
required to assure that designs of new dams and of dam retrofits meet deformation
criteria for desired seismic performance. This report is intended to provide guidance on
the use of available analysis procedures to develop estimates of earthquake-induced
deformations of embankment dams in accordance with current practice.

3.0 Overview of Deformation Analysis Approaches

3.1 Historical Development

The initial development of seismic deformation analysis procedures for embankment

dams may be traced back to the 1960s. A timeline of selected events in the development
of embankment dam seismic analysis procedures and case-history implementation of
such procedures is shown in Figure 3-1. Analysis procedures generally evolved in
parallel with the development of dynamic soil laboratory tests, constitutive models of soil
behavior, and analysis software in geotechnical engineering. For simplicity, however, the
development of soil laboratory tests, constitutive models, and geotechnical analysis
software is not traced in Figure 3-1.

An early major contribution was made by Ambraseys (1960a, b, c) who utilized 1-D and
2-D shear-beam methods of analysis with visco-elastic material behavior to investigate
the dynamic response of embankment dams. Newmark (1965) developed a method to
estimate earthquake-induced embankment displacements using a sliding block analysis
technique.

The first application of the finite element method (FEM) for dynamic analysis of
embankment dams was made by Clough and Chopra (1966). This application involved
utilizing the FEM to calculate the vibration mode shapes and frequencies of a triangular
earth dam cross section and the dynamic response of the section for input time histories
representing the 1940 El Centro earthquake. The FEM was incorporated into the Seed-
Lee-Idriss analysis procedure, which was first introduced conceptually by Seed (1966)
and first applied to analyze the failure of the Sheffield Dam after the 1925 Santa Barbara
earthquake (Seed et al., 1969). The analysis of the Sheffield Dam failure noted the need
to allow for the progressive development of liquefaction to explain the failure. The
procedure was later refined with various improvements, including the use of the
equivalent linear procedure (Idriss and Seed, 1968), and used to analyze the performance
of the upper and lower San Fernando Dams during the 1971 San Fernando earthquake
(e.g., Seed et al., 1975a, 1975b) and was eventually formalized by Seed (1979a). In the
Seed-Lee-Idriss procedure, the FEM is used to calculate the pre-earthquake and
earthquake-induced stresses, which are subsequently used to estimate excess pore water

11
pressures, strains and deformations, and stability factor of safety (Idriss and Duncan,
1988).

Concurrently with the aforementioned developments, studies were being performed to

better understand the behavior of soils under cyclic loading. Seed and Lee (1966) studied
the liquefaction of saturated sands using cyclic triaxial tests while Drnevich (1967)
studied the dynamic shear modulus and damping characteristics of soils at low strains
using resonant column tests. Cyclic simple shear tests were subsequently used by Seed
and Peacock (1971) and Finn et al. (1971) to study the cyclic strength of sands.

In the 1970s, analysis techniques were improved and applied to case histories of dam
performance during earthquakes. The simplified procedure for evaluating soil
liquefaction potential was proposed by Seed and Idriss (1971). The equivalent-linear
method to approximate nonlinear soil behavior for dynamic response analysis, which had
been developed by Idriss and Seed (1968), and the variations of modulus and damping
values with shear strain proposed by Seed and Idriss (1970) were subsequently
incorporated into the one-dimensional (1D) dynamic analysis program SHAKE
(Schnabel et al., 1972), and the two-dimensional (2D) FEM analysis programs QUAD4
(Idriss et al., 1973) and FLUSH (Lysmer et al., 1975).

The San Fernando earthquake occurred in 1971 and resulted in the near-failure of the
Lower San Fernando Dam and significant damage to the Upper San Fernando Dam.
Since then, the performances of these two dams have been important case history studies
for benchmarking deformation analysis approaches. The first study of the 1971
earthquake performance of the dams was carried out by Seed et al. (1973). In his Rankine
Lecture, Seed (1979a) provided a summary of key considerations in the earthquake-
resistant design of earth and rockfill dams and an overview of methods of stability and
deformation analysis available at that time and formalized the Seed-Lee-Idriss analysis
procedure.

New computer programs capable of performing nonlinear deformation analyses were

developed in the 1980s including TARA-3 (Siddarthan and Finn, 1982); DYNAFLOW
(Prevost, 1981); DIANA (Kawai, 1985); FLAC (Cundall and Board, 1988); and
DYNARD (Moriwaki et al., 1988). Some of these programs have been enhanced over
time and are used today. Another notable contribution in the 1980s was the Earthquake
Engineering Research Institute monograph “Ground motion and soil liquefaction during
earthquakes” by Seed and Idriss (1982). This monograph summarized techniques for
liquefaction analysis and earthquake ground motion characterization.

Several earthquakes occurred in the 1980s that produced useful case histories of
embankment dam performance including the 1980 Mammoth Lakes earthquake, the 1987
Whittier Narrows earthquake, and the 1989 Loma Prieta earthquake. Instrumentation at
several dams shaken by those earthquakes provided strong ground motion recordings
which were utilized to evaluate the capabilities of analysis techniques of that time.
Example studies included the analysis of Long Valley Dam for the 1980 Mammoth Lake

12
earthquake (Griffiths and Prevost, 1988) and the analysis of Puddingstone and Cogswell
Dams for the 1987 Whittier Narrows earthquake (Seed et al., 1989).

In the 1990s, multiple numerical analyses were conducted of well-recorded earthquake

dam response. Notable dynamic analyses of dams included those of: La Villita Dam
(Elgamal, 1992), Lenihan Dam (Makdisi et al., 1991; Mejia et al., 1992), Cogswell Dam
(Boulanger et al., 1995), and Long Valley Dam (Zeghal and Ghafar, 1992). Empirical
studies of dam performance were also published, most notably by Fong and Bennett
(1995) and Swaisgood (1998), which correlated measures of crest settlement and
cracking to simple metrics of shaking and dam configuration. Improvements to
constitutive models for soil liquefaction based on effective stresses were also developed
and used with common numerical analysis programs. Examples of such models include
the cycle-counting Roth model (Roth et al., 1991), the coupled effective stress Wang
model (Wang et al., 1990), and the UBCSAND model (Beaty and Byrne, 1998).

Since 2000, additional improvements have been made in simplified and advanced
analysis techniques. Notable empirical studies were published by Pells and Fell (2002)
and Swaisgood (2003, 2014). Correlations based on large sets of sliding block analyses,
such as the regression model by Bray and Travasarou (2007), were developed to allow
earthquake-induced embankment deformations to be estimated from simple ground
motion intensity measures.

New constitutive models have been introduced recently for modeling the dynamic
behavior of various soil types. Examples include: (1) UBCHYST (Naesgaard et al.,
2015), for non-liquefiable soils; (2) PM4Silt (Boulanger and Ziotopoulou, 2018a), for
low-plasticity soils subject to cyclic softening; (3) UBCTOT (Beaty and Byrne, 2008), a
decoupled total stress model for liquefiable soils; and (4) PDMY (Elgamal et al., 2003),
SANISAND (Taiebet and Dafalias, 2008), PM4Sand (Boulanger, 2010; Boulanger and
Ziotopoulou, 2015), and WANG2D (Wang and Ma, 2018; Wang and Ma, 2019), coupled
effective stress models for liquefiable soils.

Analyses of dam seismic performance case histories since 2000 have included those of
the Lower San Fernando Dam (Ming and Li, 2003; Naesgaard and Byrne, 2007; Dawson
and Mejia, 2012; Chowdhury, 2019), Upper San Fernando Dam (Beaty and Byrne, 2008;
Chowdhury, 2019); and Lenihan Dam (Hadidi et al., 2014; Armstrong et al., 2018).
Recently, Boulanger (2019) used nonlinear procedures and the PM4Silt model to analyze
the seismic response of Austrian Dam during the Loma Prieta earthquake, and Dawson
and Mejia (2020) used a parallel Iwan model (Chiang and Beck, 1994) to perform a
three-dimensional (3D) nonlinear seismic response analysis of Lenihan Dam for the same
earthquake. Mejia and Dawson (2019) also used the Lenihan Dam case history to
illustrate the use of 3D nonlinear analysis procedures in evaluating the potential for
earthquake-induced cracking of embankment dams.

Such types of case history analyses as well as simulations of centrifuge tests (e.g., Kutter
et al., 2017; Manzari et al., 2017) have provided useful evaluations of the capabilities of
numerical platforms and constitutive models for analysis of embankment seismic

13
deformations. Several summaries of the state-of-practice for liquefaction evaluation have
also been published since 2000 including those by Youd et al. (2001), Seed et al. (2003),
Idriss and Boulanger (2008), and the National Academies of Sciences, Engineering, and
Medicine (NASEM, 2016).

Several of the analysis methods and procedures mentioned above are used in current
practice and research in the U.S. and are the subject of subsequent sections of these
guidelines.

Figure 3-1. Timeline of Selected Events in the Development and Case-History

Implementation of Methods for Analysis of Seismic Deformations of Embankment
Dams.

3.2 Types of Deformation Analysis Approaches

Methods of analysis of earthquake-induced deformations used in practice may be

grouped into the following types of analysis approaches: (1) simplified analyses, (2)
sliding block analyses, and (3) nonlinear deformation analyses. An overview of these
analysis approaches is provided in this section together with a discussion of post-
earthquake stability analyses and screening procedures for embankment dams. More

14
detailed descriptions of the analysis methods that are grouped under these approaches are
given in Chapters 4, 5, and 6.

Table 3-1 summarizes the general types of inputs and outputs associated with the various
analysis methods classed within the above approaches. It is important to emphasize that
the entries in Table 3-1 refer to the types of inputs/outputs applicable to an analysis
approach, as opposed to the specific input/output parameters pertinent to a method of
analysis grouped under any one approach. Applicability of types of inputs/outputs to
multiple approaches in the table does not imply that the inputs/outputs or their associated
uncertainties are considered equally in the analysis methods grouped under those
approaches. Various analysis methods may produce the same types of outputs (e.g., crest
settlement) but the outputs may involve differing levels of applicability and reliability
depending on the situation. For many situations where both types of methods might be
deemed applicable, the outputs of simplified analysis methods are generally subject to
greater uncertainty than the outputs of advanced analysis methods. The user should be
aware of specific uncertainties and limitations associated with each analysis method and,
whenever feasible, should consider using more than one method for the analysis of dam
seismic deformations.

3.2.1 Simplified Analyses

Simplified analyses include the use of empirically-based regression models and

regression models based on the results of multiple systematic sliding block analyses (i.e.,
analytically-based regression models). Analyses of post-earthquake flow slide
deformations using simplified methods, such as limit-equilibrium or runout analysis
procedures, are also included in the category of simplified analyses.

In empirically-based regression models, deformation (usually characterized by crest

settlement) is estimated based on regression analysis of observed dam deformations
during earthquakes. Simple measures of embankment characteristics (typically
embankment height) and earthquake characteristics (e.g., peak ground acceleration, 𝑃𝑃𝑃𝑃𝑃𝑃,
and earthquake magnitude, 𝑀𝑀) are regressed against observed deformations (e.g.,
settlements). An example of an empirical model is the correlation by Swaisgood (2014).
Analytically-based regression models are similar except that the deformation data are
obtained from deformation analyses models with parameters describing selected
embankment characteristics (e.g., embankment period of vibration and yield coefficient,
𝑘𝑘𝑦𝑦 ) and earthquake shaking characteristics (e.g., 𝑃𝑃𝑃𝑃𝑃𝑃 and peak ground velocity, 𝑃𝑃𝑃𝑃𝑃𝑃).
An example of an analytically-based regression model is that developed by Bray and
Travasarou (2007).

3.2.2 Sliding Block Analyses

Sliding block analyses assume that deformations of a dam result from sliding of a mass or
block 1 of the dam by yielding along the base of the block. The original concept, as

1
The terms slide block and slide mass are used interchangeably throughout this document.

15
published by Newmark (1965), assumes that the soil block is rigid and neglects the
dynamic response of the dam. The earthquake motion is applied to the base beneath the
block, and relative displacements occur when the applied base accelerations exceed the
yield acceleration of the sliding block. This approach to a sliding block analysis is often
referred to as a Newmark analysis.

Sliding block analyses are occasionally performed using acceleration histories

corresponding to the base of the dam as described by Newmark (1965). Preferably, the
analyses include the dynamic response of the dam (e.g., as described by Makdisi and
Seed, 1978). In the latter case, the analyses are based on estimates of the average
horizontal accelerations of the sliding block developed from 1D or 2D dynamic response
analyses. If the average accelerations of the sliding block are calculated assuming that
block sliding does not occur (i.e., they are calculated independently of the sliding
displacement of the block), the analysis is referred to as decoupled (Lin and Whitman,
1983). If the accelerations and sliding displacements of the block are analytically
dependent on each other and are calculated jointly, the analysis is referred to as coupled
(Rathje and Bray, 2000).

3.2.3 Nonlinear Deformation Analyses

Nonlinear deformation analyses (NDAs) include 2D and 3D continuum-based numerical

analyses in which soil yielding and unrecoverable plastic deformations are simulated.
These types of analyses provide the most insight into the seismic deformation response of
a dam, although they are more complex than simplified and sliding block analyses.
Nonlinear deformation analyses are routinely conducted using a 2D formulation.
However, in some situations significant 3D effects may be expected (e.g., embankments
in narrow canyons with highly irregular/steep foundations) and 3D analyses may be
appropriate. In addition, in cases where a refined assessment of potential embankment
cracking is desired, nonlinear 3D analyses offer the capability to directly estimate the
location and extent of seismically induced transverse and longitudinal cracking
throughout an embankment (Mejia and Dawson, 2019).

Table 3-1. General Types of Inputs and Outputs Applicable to

Deformation Analysis Approaches Used in U.S. Practice.
SIMPLIFIED(2) SLIDING BLOCK
Inputs - Outputs(1) NDA
Empirical Analytical Decoupled Coupled
Inputs
Embankment geometric
X X X X X
characteristics
Yield acceleration X X X
Elastic stiffness X(3) X X
Shear strength X X X X
Permeability X(4) X(4)
Embankment period X(5)
Modulus reduction and
X(3) X X(6)
damping curves

16
 Constitutive model specific
X(6)
parameters
Earthquake magnitude(7) X X(5)
Ground motion intensity
X X
measures
Ground motion time series X X X
Outputs
Slip surfaces X X X
Sliding block displacements X X X
Crest settlement X X
Shear strains X
Excess pore water pressures X
Deformation patterns X
Notes:
(1) The table entries refer to the types of inputs/outputs applicable to an analysis approach, as opposed to
specific input/output parameters pertinent to a method of analysis grouped under any one approach.
Applicability of types of inputs/outputs to multiple approaches does not imply that the inputs/outputs or
their associated uncertainties are considered equally in the analysis methods grouped under those
approaches. Multiple analysis methods may produce the same types of outputs (e.g., crest settlement) but
the outputs may involve differing levels of validity and uncertainty depending on the situation.
(2) Simplified analyses in this document include empirically-based correlations, analytically-based
correlations, and flow slide analyses – see Chapter 4 for more information.
(3) Decoupled analyses include Newmark Analysis, which does not consider the dam’s response nor
require this type of input.
(4) Only required if steady-state or transient seepage is modeled.
(5) Not required for all analytically-based models.
(6) Additional model-specific parameters are required depending on the constitutive models used.
(7) Moment magnitude is the earthquake magnitude scale most commonly used in practice.

3.2.4 Post-Earthquake Stability Analyses

Post-earthquake stability analyses are commonly performed in practice to check the

potential for gross instability of a dam once the earthquake shaking has ceased. A post-
earthquake analysis provides an indication of the potential for gravity-driven
deformations that may result from seismically induced strength loss (e.g., from soil
liquefaction). Post-earthquake stability analyses are commonly performed using limit-
equilibrium techniques but can also be done using finite element or finite difference
analysis methods with suitable soil constitutive models. The analysis considers primarily
changes in strength due to the seismic loading but may also consider post-shaking
changes in pore pressure, effective stress, and strength. If a limit-equilibrium analysis is
conducted, the resulting factor of safety may be taken as an indication of the likelihood of
large deformations. When performed with nonlinear finite element or difference methods,
deformations can be estimated directly as the numerical mesh will deform during the
post-earthquake analysis until static equilibrium is reached. An important consideration
in these analyses is the potential for pore pressure redistribution, void ratio changes, and
delayed failures after the earthquake. Analyses of pore pressure redistribution and void
ratio changes are possible with nonlinear deformation analyses. Examples of such
analyses are presented by Naesgaard et al. (2006) and Kamai and Boulanger (2013).

17
3.2.5 Screening Procedures

Information collected from case histories of earthquake embankment dam performance

have been used to identify embankment and earthquake shaking characteristics that may
be indicative of adequate or inadequate performance in an earthquake. If previous case
histories indicate that a dam should perform adequately in a future earthquake,
deformation analyses may not be necessary to assess the expected seismic performance of
the dam, provided other key dam safety issues are duly considered (e.g., the potential
consequences of dam failure).

Criteria for screening-based procedures were first suggested by Seed et al. (1978) and
have undergone subsequent refinements (e.g., Seed 1979a). As an example of screening
criteria for low to medium height dams (i.e., embankment heights less than 60 m or 200
feet), the USACE summarized in 2010 eight conditions that, if satisfied, suggest that
deformation analyses are not required to confirm adequate performance in an earthquake
(Perlea and Beaty, 2010). These eight conditions are as follows.

1. Dense dam and foundation materials that are not liquefiable and do not
contain sensitive clays.
2. Well-built, densely compacted dam with relative compaction 𝑅𝑅𝑅𝑅 > 95% 2 or
relative density 𝐷𝐷𝑟𝑟 > 80%.
3. Dam slopes of 3:1 (H:V) or flatter, and/or the phreatic line well below the
downstream face.
4. 𝑃𝑃𝑃𝑃𝑃𝑃 ≤ 0.2g at the base of the embankment. On a case-by-case basis, 𝑃𝑃𝑃𝑃𝑃𝑃 ≤
0.35g for a compacted clay embankment on rock or stiff clay foundation.
5. Static factor of safety before the earthquake greater than 1.5 for all non-
surficial failure surfaces.
6. At the time of the earthquake, freeboard greater than 3 to 5 percent of the dam
height (including alluvial foundation) but not less than 0.9 m (3 feet). Special
attention should be given to the presence and suitability of filters for dams
with modest freeboard.
7. No appurtenant features related to the safety of the dam would be harmed by
small movements of the embankment.
8. No historic incidents at the dam that may indicate a limitation in its ability to
survive an earthquake.
3.3 General Analysis Steps

3.3.1 Geological and Geotechnical Model Development

The development of suitable geological and geotechnical models of the embankment dam
and foundation represents an important first step, if not the most important, in conducting

2
The ASTM D1557 standard is the assumed reference for maximum density.

18
a deformation analysis. Considerable guidance has been developed and published
regarding various aspects of this process, so only a brief discussion is provided herein.
For further detail, the reader is referred to the technical literature including general
references, such as: USBR (1998), USBR (2015) and Fell et al. (2015); and more
specialized references, such as: EPRI (1990); Mayne et al. (2001); Idriss and Boulanger
(2008); and Lees (2016).

As a first task, the information from as-built drawings, geological maps, geologic studies,
construction records, and site investigation exploration programs should be synthesized
to create a model in which the spatial distribution of materials and embankment zones are
defined. For each material zone, available basic index properties (e.g., plasticity index,
gravel content, sand content, fines content, unit weight), stiffness properties (e.g., shear
wave velocity), penetration resistance (e.g., CPT, SPT), and shear strength properties
(e.g., cohesion and friction angle) should be summarized. Additional material properties
may be required for analysis of phenomena such as liquefaction. In typical project
applications the scope of field and laboratory investigations is commonly increased in a
phased manner with corresponding increases in the level of geotechnical modeling,
material characterization, and analysis complexity.

Statistical evaluation of the data may be warranted if the data sample size is significant,
although any statistical evaluation should be guided by an understanding of the applicable
geomorphology and construction processes. Although representative properties are
typically defined for each material, it is often important to conduct analyses to evaluate
the sensitivity of the results to the selection of material properties. Spatial visualization of
the data can be achieved by plotting data on plan and profile views in which the material
units are delineated. Profile views should be selected to help visualize the embankment
and foundation characteristics that may affect earthquake-induced phenomena. For
example, common profile views include the following sections of an embankment dam:
(1) along the longitudinal axis, (2) at the maximum embankment cross section and other
transverse sections, and (3) near significant discontinuities in the foundation surface or
adjacent to embedded structures.

The profile views may be used directly in subsequent deformation analyses or idealized
into one or more design cross sections for analysis purposes. Development of 3D models
of embankment zoning and foundation stratigraphy may also be useful to improve the
material characterization of the embankment and foundation.

3.3.2 Seismic Hazard Analysis and Ground Motion Development

The evaluation of the seismic hazard, selection of seismic load criteria, and the
characterization of earthquake ground motions are critical steps for the analysis of dam
seismic deformations. The details of the seismic hazard assessment and seismic load
characterization are likely to depend on the regulatory environment and scope for seismic
safety evaluation of a dam and on the types of anticipated deformation analyses, but all
analysis methods require a clear understanding of the seismic hazard. For a screening or
simplified analysis, the seismic loading is commonly represented using ground motion

19
intensity measures (e.g., peak ground acceleration and earthquake magnitude). For a
dynamic response analysis, ground motion time series must be selected. Regardless of the
analysis method, the selected ground motion parameters or time series must appropriately
consider the seismic hazard and represent the design loading. It is also important to note
that, depending on the situation, simplicity in ground motion characterization (e.g., with
intensity measures alone) may not be sufficient to properly characterize the seismic
loading for dam evaluation and may lead to underestimation of uncertainty in the loading
and in the resulting dam response.

The selection of the seismic loading characteristics for analysis, including the number of
earthquake ground motion time series, involves multiple considerations, the specifics of
which are beyond the scope of this document. Because of the subject’s importance in the
analysis of dam response and deformations, readers and analysts in particular, even if not
involved in the development of the ground motions for analysis of a dam, are encouraged
to be familiar with the subject. A few references on seismic hazard assessment and the
estimation and selection of earthquake ground motions for seismic analysis include
Kramer (1996), Bozorgnia and Bertero (2004), and NIST (2011). General guidance on
the selection of earthquake ground motions for dam projects has been published by,
amongst other organizations, USCOLD (1999), ICOLD (2016) and FERC (2018).
Multiple examples of project-specific applications of such types of guidance are found in
the literature (e.g., Mejia et al., 2002, 2005; Phalen et al., 2014; and Hunt et al., 2016).

3.3.3 Phased Approach to Deformation Analyses

A seismic deformation analysis may be performed once the design cross sections,
material properties, and seismic loading for a dam are characterized. As the effort, time,
cost, and necessary expertise generally increase with the complexity of a deformation
analysis, it is common to implement a phased analysis approach that begins with simple
evaluations and adds sophistication as needed (e.g., see Perlea and Beaty 2010). The final
level of analysis sophistication required will depend on factors such as the importance of
the project, the consequence of unsatisfactory performance of the dam, the purpose of the
analysis, the intensity of the earthquake ground motion, and the complexity of possible
material behavior in the embankment and foundation under earthquake shaking (all of
which dictate the necessary degree of confidence and refinement in the analysis results).

A screening level assessment may be used initially. If the screening level assessment
indicates little potential for significant deformation, additional analyses may not be
warranted. If there is no potential for earthquake-induced stiffness or strength loss in the
embankment or foundation, simplified analyses may be performed to assess the order-of-
magnitude of likely seismic deformations. This level of analysis may be sufficient if only
a rough estimate of deformations is needed, and the estimated deformations are small.
The next level of analysis sophistication corresponds to sliding block analyses, which
allow factoring-in the characteristics of the earthquake ground motions and the seismic
response of the dam. These analyses are often sufficient to assign a dam evaluation into
one of the following two categories: (1) cases where small deformations are expected
with high confidence and that do not require refined definition of the spatial distribution

20
of deformation, and (2) cases of moderate or large deformations, of poorly defined
deformations, of potential for significant stiffness or strength degradation, or cases that
require a refined definition of the spatial distribution of deformation. It is this second
category where advanced analyses are usually required and provide the most value.

There are several reasons that nonlinear deformation analyses are performed as part of a
phased dam evaluation, such as: (1) critical phenomena are not adequately captured in or
simulated with the simpler analysis approaches (e.g., excess pore water pressure
generation, complex geometries or material behaviors), (2) deformations computed using
simpler methods do not allow for confident decision making, (3) the dam has unusual
components or features that may not lend themselves to simplified analysis procedures,
(4) the analysis objectives include design optimization or evaluation of robustness, (5) an
improved understanding of potential seismic response is beneficial to the safety and
economic assessment of a project, and (6) performance-based or risk-informed design
calls for advanced analysis of potential deformations and associated uncertainties.

It is important to note that for some project applications simplified analysis methods may
not be applicable or appropriate to begin with, or the need for advanced analyses is
otherwise identified early in the project. In such cases, a phased approach is not likely to
be applicable to the analysis of dam seismic deformations, although a phased program
may nonetheless be useful for the process of dam investigations and geotechnical dam
characterization.

3.3.4 Special Considerations Regarding Liquefaction and Cyclic Softening

Soil liquefaction in a dam or its foundation is a key phenomenon that may significantly
affect the magnitude and pattern of seismically-induced displacements. Thus, this
potential soil behavior should be addressed regardless of analysis approach. Simplified
deformation analysis methods are often not appropriate when liquefaction is anticipated,
and the potential for and extent of liquefaction should be assessed in establishing the
validity of these methods. Procedures and considerations for performing liquefaction
triggering analyses and estimating post-liquefaction residual shear strengths are not
presented in this document and are addressed elsewhere (e.g., see Idriss and Boulanger,
2008; NASEM, 2016). The potential for cyclic softening in low-OCR clay-like soils, and
its relevance to the selected analysis methods, should also be assessed as part of each dam
analysis program.

The potential for liquefaction or cyclic softening of the embankment or foundation

materials often requires special modeling in a deformation analysis. Both of these
phenomena can lead to increased inertia-driven permanent deformations as well as
increased gravity-driven deformations due to the decrease in material stiffness and
strength. The impacts of material strength-loss on overall embankment stability are often
assessed through limit-equilibrium analyses in which the yield coefficient or factor of
safety are determined using a reduced strength. Nonlinear deformation analyses with soil
constitutive models that directly simulate the anticipated changes in soil behavior are also
commonly used to evaluate the impacts of strength loss on embankment stability and

21
deformation. Such analyses are able to capture directly changes in material response and
strength and the impacts of those changes on embankment deformations.

3.3.5 Volumetric Strains and Deformations

Embankment dam seismic deformations are associated with the plastic shear and
volumetric strains induced in the dam and foundation materials during earthquake
shaking. Additional shear and volumetric strains may occur after the earthquake ceases
due to redistribution and dissipation of excess pore water pressures generated by the
shaking.

During shaking, volumetric strains in saturated materials under undrained conditions are
very small and shear strains are the main component of seismic deformations. Volumetric
strains are also generally small in dense or cohesive soils, even if unsaturated, because
these materials are not susceptible to compression (i.e., densification) under vibratory
loading, and in fact may dilate under shear. However, volumetric strains during shaking
can be significant in unsaturated soils susceptible to densification under vibratory loading
such as dry loose-to-medium-dense sands, gravels, and rockfill. Volumetric strains during
shaking may also be significant in saturated rockfill as these materials can exhibit drained
behavior during the shaking because of their large permeability.

Because volumetric strains may be a significant component of seismic deformations of

embankment dams, they must be considered in the evaluation of dam seismic
performance. In fact, in materials with significant tendency for compression under cyclic
loading, such as loose to medium dense rockfill, seismic compression associated with
volumetric strains may control overall seismic deformations of a dam. For example, this
appears to have been the case at Fena Dam during the Mw 7.8 1993 Guam Earthquake
(Mejia, 2017).

A key consideration in the evaluation of potential seismic volumetric deformations is that

the plastic volumetric strain characteristics of coarse cohesionless soils can be
significantly different under saturated conditions than under dry conditions. For example,
it is well known that saturated rockfill is significantly more compressible under
earthquake shaking (and monotonic loading) than dry rockfill, all other conditions being
equal.

Various approaches are commonly used in practice to evaluate deformations associated

with volumetric compression, depending on the approach used to evaluate overall dam
seismic deformations. Empirically-based correlations (e.g., Swaigsood, 2014) implicitly
include deformations associated with volumetric compression. On the other hand, most
analytically-based correlations (e.g., Bray and Travasarou 2007) and sliding block
analyses account for shear deformations only, and evaluation of dam seismic
deformations using these approaches must be supplemented with a separate assessment of
potential volumetric compression. Depending on the soil constitutive models used for
analysis, nonlinear deformation analyses can directly include shear as well as volumetric
strains and deformations. The capability and limitations of any constitutive model should

22
be well understood before it is used in estimating volumetric deformations as many
models do not address key aspects of seismically-induced volumetric strains.

Procedures commonly used to estimate volumetric strains of unsaturated sands (seismic

compression) are generally based on studies of settlements under cyclic loading in the
laboratory. Keystone studies include those by Silver and Seed (1971), Martin et al.
(1975) and Tokimatsu and Seed (1987). More recent studies of the laboratory seismic
compression of sands include those by Duku et al. (2008) and Lasely et al. (2016). Post-
liquefaction reconsolidation settlements of sands are often estimated using these types of
data as well (e.g., Ishihara and Yoshimine, 1992; Wu et al., 2003; Idriss and Boulanger,
2008).

It is occasionally assumed that the above studies and procedures can be used to
approximate volumetric strains in cohesionless soils other than sands, such as gravelly
sands and sandy gravels. However, studies of gravelly soils (e.g., Kokusho, 2007)
indicate that the deformation behavior of such soils can be significantly different than
that of sands. Additional research and data are needed to facilitate the evaluation of
volumetric strains and deformations in well graded and coarse-grained soils.

These procedures are occasionally used to estimate dam volumetric deformations in a

simplified manner by applying them to vertical profiles in a dam section assuming 1D
behavior. The procedures have also been used with 2D and 3D nonlinear finite element
methods to simulate volumetric strains in seismic analyses of dams. Finn (1993) used the
formulation proposed by Byrne (1991), and the measured deformations of Matahina Dam
(New Zealand) during the Mw 6.5 Edgecumbe Earthquake, to simulate volumetric
deformations in 2D nonlinear analyses of the dam for the design earthquake motions. Zou
et al. (2013) used 3D nonlinear analyses to simulate the seismic response of Zipingpu
Dam (China), including shear and volumetric deformations of the dam, during the Mw 7.9
2008 Wenchuan Earthquake.

A unique advantage of nonlinear deformation analyses is that they offer the ability to
calculate directly seismically induced volumetric deformations of dams provided that the
constitutive models used in the analysis are properly formulated to calculate volumetric
strains (Chapter 6). Note that the above limitations associated with data availability for
coarse granular soils also apply for these types of analyses.

3.3.6 Instrumentation Data Utilization

Measured data from piezometers, seepage weirs, survey monuments, inclinometers,

strong motion accelerometers, and other instruments help in understanding the behavior
of a dam. These types of data can also be employed to calibrate the static and seismic
analysis of a dam and its material engineering properties. Basic considerations in utilizing
instrumentation data for steady-state seepage analyses and dynamic analyses are
discussed below. A recent example of using instrumentation data in a dam seismic
evaluation was presented by Hadidi et al. (2014).

23
For a steady-state seepage analysis, the reasonableness of the calculated total water heads
can be evaluated by comparison with those measured from piezometers. When
applicable, horizontal and vertical permeabilities may be adjusted within reasonable
ranges so that the calculated total water heads provide a reasonable representation of the
measured values. When using piezometer recordings, it is important to consider
differences between the available data and the assumptions made in the seepage and
deformation analyses about reservoir levels and possible transient seepage conditions.

Strong ground motion data recorded at a dam from previous earthquakes may be used to
evaluate the reasonableness of a seismic deformation analysis model. In this evaluation,
the deformation analysis model is typically subjected to an input motion equal to or
deconvolved from a recorded motion on a nearby outcrop of competent material. When
using recorded motions, their suitability as input motions for analysis should be carefully
considered. Comparison of the analysis results with the instrumentation data should focus
on several factors, including (1) earthquake-induced permanent deformations, (2)
acceleration histories and other ground motion characteristics (such as acceleration
response spectra and Fourier amplitude spectra) at the crest and at other locations on the
embankment, and (3) changes in measured pore pressures.

Calculated results that agree reasonably well with the measurements would suggest that
the geometry and other analysis inputs, including estimated material behaviors (e.g., soil
liquefaction or cyclic softening), provide a reasonable representation of the embankment
dam response at the loading level of the recorded earthquake. However, the possibility
that compensating errors may have fortuitously contributed to the finding of reasonable
agreement should be considered. If, on the other hand, the calculated results differ
significantly from the measured data, there may be one or more aspects of the analysis
model or overall modeling approach that are not representative of field behavior. For
example, a case in which deformations are significantly overpredicted may indicate that
the adopted strength and stiffness parameters for one or more of the embankment zones
are lower than those of the embankment materials in the field, and that additional effort to
better define the shear strength of the materials is required.

4.0 Simplified Analyses

4.1 Overview

Simplified analyses are often useful first-step tools in evaluating the seismic performance
of embankment dams. These analyses may provide an appreciation of likely dam
response and establish a frame of reference for assessing the results of an advanced
analysis. For example, the process of estimating the yield acceleration, 𝑘𝑘𝑦𝑦 , of a number of
potential sliding surfaces may suggest potential deformation patterns. Simplified analysis
may also provide an initial basis for estimating earthquake displacements and the
potential for loss of stability and a flow slide. It is important to note that, although
simplified analyses are useful tools, advanced analyses are often required to make final
decisions regarding the potential performance of many embankment dams during future
earthquakes. This is particularly the case for high-hazard and critical dams.

24
A key feature of a simplified analysis is that it can be performed relatively quickly with
minimal resources and effort. In fact, simplified analyses may often be completed with no
or minor aid from advanced computer programs. Due to their simplicity and relative cost
effectiveness, simplified analyses are a common first step in phased deformation analysis
programs. There are however multiple limitations associated with simplified analyses (as
discussed in the following sections) and the corresponding uncertainty in the results
should be considered. In this regard, it is generally considered good practice to use
multiple methods of analysis, whenever feasible.

For purposes of this guideline, simplified analyses are categorized into three broad
categories:

• Empirically-Based Correlations
• Analytically-Based Correlations
• Flow Slide Analyses
The application, limitations, advantages and disadvantages of each of these methods are
discussed subsequently.

4.2 Empirically-Based Correlations

Empirically-based correlations use regression models that relate observations of dam

performance following an earthquake to selected characteristics of the dam and the
earthquake loading (e.g., Jansen, 1990; Swaisgood, 2014). Empirically-based regression
models are often used to obtain order-of-magnitude estimates of deformations for
comparison with the results of more in-depth analyses. These methods are generally
useful for assessing dam response to a deterministically defined earthquake scenario.

One empirical correlation that is commonly used in practice to estimate seismically

induced crest settlement is the correlation developed by Swaisgood (2003, 2014).
Swaisgood compiled a database summarizing the seismic response of 82 dams and
focused primarily on cases where liquefaction was not a significant factor. The types of
dams considered in the study included: 35 earth core rockfill dams, 29 earthfill dams, 11
concrete-faced rockfill dams, and 7 hydraulic fill dams.

Data from these case histories, in terms of normalized crest settlement (NCS), are shown
in Figure 4-1. NCS is the measured crest settlement as a percentage of dam height plus
alluvium thickness (see Figure 4-1). For discussion of the level of damage shown in
Figure 4-1, the reader should refer to Swaisgood (2014).

25
 Figure 4-1. Settlement of Embankment Dams During Earthquake (Excluding Cases of
Settlement Due to Liquefaction) (from Swaisgood, 2014).

Swaisgood (2014) performed a regression analysis of data from the 82 case studies and
derived a best-fit relationship between NCS and peak ground acceleration (PGA) and
earthquake magnitude, M. Note that the relationship does not quantify the dispersion in
the data or consider other basic factors that are known to affect embankment dam seismic
deformations such as dam configuration and material properties. Furthermore, only 4
case histories of large subduction earthquakes (i.e., M>8.5) are included in the database.

Therefore, this and other similar correlations should generally be used with caution.
Special care should be exercised for cases that are not well represented in the databases
used to derive the correlations. The limitations of any one correlation should be
considered in using the relationship, together with the range of supporting observations.
Before using a correlation, users are advised to become familiar with its database. A side
benefit is that useful insights into the possible seismic behavior of a dam may be gained

26
by identifying dams within a correlation database that are relatively similar to the dam in
question.

4.3 Analytically-Based Correlations

Analytically-based correlations are based on the displacement results of analytical studies

rather than case history observations. All of the methods in this category are based on
data sets of displacements obtained from analyses such as those used in the rigid sliding
block method (see Chapter 5). These correlations provide approximations of
displacements calculated from sliding-block-type analysis methods, based on descriptors
of the slide mass (e.g., yield acceleration, fundamental period of vibration) and input
ground motion (e.g., peak ground acceleration (PGA), peak velocity (PGV), spectral
acceleration, and earthquake moment magnitude (Mw)).

Some of the analytically-based methods aim to simplify the rigid sliding block analysis
procedure, while others attempt to address limitations of the original Newmark approach
(such as the assumption of rigid soil behavior above the slide plane). Regardless, they are
all constrained by key aspects of the sliding-block analysis approach, which include: (a)
displacements are assumed to occur on a sliding surface and deformation implies rigid-
plastic soil behavior along the surface; and (b) the methods are restricted in application to
compacted clayey embankments and dry or dense cohesionless soils which experience
limited reduction in shear strength due to cyclic loading.

Table 4-2 presents a listing of simplified methods based on analytical correlations that are
commonly referenced in the literature. A few of these methods are briefly discussed in
Appendix A.

Table 4-2. List of Simplified Methods Based on Analytical Studies.

Assumed
Predicted No. of
Key Input Sliding Block
Method Displacement Acceleration
Parameters Response
Value Records
Model
Newmark (1965) PGA, PGV, ky Rigid Upper Bound 4

Sarma (1975) PGA, Tp, ky Rigid Mean 9

Franklin & Chang (1977) PGA, ky Rigid Upper Bound 179

Makdisi and Seed (1978) PGA, Mw, ky, TD Decoupled Range 4

Hynes-Griffin & Franklin

PGA, ky Decoupled Upper Bound 354
(1984)
Coupled Mean and
Mw, ky, Ts,
Bray & Travasarou (2007) (Rigid/Non- Standard 688
Sa (1.5Ts)
rigid) Deviation
Mean and
Jibson (2007) PGA, Mw, Ia, ky Rigid Standard 2270
Deviation

27
 Mean and
PGA, PGV, Ia,
Saygili & Rathje (2008) Rigid Standard 2383
ky
Deviation
PGA, Mw, ky, Ts,
Coupled
Bray, Macedo, and Sa (1.5Ts) [for Selected
(Rigid/Non- 810
Travasarou (2018) subduction percentile
rigid)
events]
Mw, PGV, ky, Ts,
Coupled
Sa (1.3Ts) [for Selected
Bray & Macedo (2019) (Rigid/Non- 6,711
near-fault percentile
rigid)
events]
Notes:
Coupled = Sliding block assumed to be non-rigid and block response is coupled with displacement
Decoupled = Sliding block assumed to be non-rigid and block response is decoupled from displacement
Rigid = Sliding block assumed to be rigid
Ia = Arias intensity (m/s)
ky = Critical or yield acceleration (g)
Mw = Earthquake moment magnitude
PGA = peak ground acceleration (g)
PGV = peak ground velocity (cm/s)
Tp = Predominant period of earthquake acceleration record
Sa (𝑥𝑥Ts) = Spectral acceleration of the input ground motion at a period of 𝑥𝑥Ts (g)
TD = Fundamental period of the dam
Ts = Initial fundamental period of sliding mass

Although they are useful screening tools for estimating seismic deformations of dams,
use of these simplified methods requires caution and careful consideration. The analyst
should be aware that, depending on the specific seismic setting and site conditions,
simplified correlations may not represent adequately key aspects of the performance of a
dam. For example, simplified correlations should not be used where liquefaction or cyclic
softening may be significant to the seismic response of a dam.

Except for the Bray et al. (2018) procedure, which provides seismic slope displacement
estimates using ground motion records from subduction zone earthquakes, other
simplified seismic displacement analysis procedures presented in Table 4-2 are based on
ground motion records from shallow crustal earthquakes. Therefore, the Bray et al.
(2018) procedure should be given preference over similar procedures, for analysis of
dams subject to subduction earthquakes.

4.4 Flow Slide Analyses

This category of deformation analysis applies to cases with materials susceptible to

liquefaction or otherwise significant strength loss due to seismic loading. The primary
use of this type of analysis is to evaluate the post-earthquake deformations caused by
gravity acting on embankments that become unstable and undergo flow sliding due to soil
liquefaction.

A flow slide analysis often incorporates an analysis of stability, such as with the limit-
equilibrium method, to assess the potential for a flow slide. Post-liquefaction shear
strengths are used in the stability analysis. Estimates of the extent of strength loss across

28
the embankment and foundation are often obtained with simplified analyses of
liquefaction triggering or with dynamic response analyses such as those based on the
equivalent linear method. Post-liquefaction shear strengths are often estimated from
available empirical correlations (e.g., Olson and Stark, 2002; Idriss and Boulanger, 2007;
Robertson, 2010; Weber, 2015; Kramer and Wang, 2015). It should be noted that
available correlations can provide widely ranging estimates for post-liquefaction residual
strengths, and this is often a key source of uncertainty in a flow slide analysis. Thus,
when characterizing the residual strength, it is prudent to consider more than one
correlation to account for this uncertainty, and to consider the variability associated with
each correlation.

Results from the stability analysis can be used to judge whether large seismic
displacements are expected. For example, a factor of safety (FOS) significantly less than
1.0 is a likely indicator of large displacements. If limit-equilibrium analyses indicate that
the embankment may not be stable after the anticipated strength loss, a NDA is often
performed to estimate the potential extent of deformation. A runout analysis for
estimating large displacements may be required in some cases. Approaches for simplified
runout analyses of flow slide deformations have been proposed by Lucia et al. (1981),
Hungr (1995), Tan et al. (2000) and Olson et al. (2000). An analytically-based correlation
for flow failures, based on parametric FLAC simulations, has been published by Rauch et
al. (2007) and Pace et al. (2008).

In cases when significant strength loss due to liquefaction or cyclic softening and flow
failure are expected, post-earthquake gravity-driven deformations may be much larger
than deformations driven by earthquake-induced inertia forces. In such cases, it may not
be necessary to calculate inertia-driven deformations. This might be the case, for
example, if a runout analysis indicates that post-earthquake gravity-driven deformations
are sufficient to initiate breach of a dam. If refined estimates of inertia-driven and post-
shaking displacements are needed, a NDA should be utilized, as discussed in Chapter 6
(see Sections 6.3.1 and 6.4.4).

5.0 Sliding Block Analyses

5.1 Overview

Sliding-block (or Newmark-type) methods are commonly used methods for estimating
seismically induced permanent displacements of slopes and embankments. Newmark
(1965) first proposed this concept to evaluate the seismic stability and perform seismic
deformation analysis of dams and embankments. Laboratory experiments (Goodman and
Seed, 1966; Wartman et al., 2003, 2005) and back-analyses of natural slopes after
earthquake-induced landslides (Wilson and Keefer, 1983) have shown that Newmark’s
method can provide reasonable estimates of slope displacements if the slope geometry,
soil properties, and earthquake ground motions are known, and key assumptions of the
method are satisfied.

Sliding-block analyses are most appropriate for competent embankments and foundations
where the deformation mechanism may be approximated by a well-defined slide mass,

29
and deformation behavior along the failure surface may be assumed to be rigid-plastic.
Sliding-block methods are generally not appropriate for situations involving liquefaction
or softening of soils, except in cases where the above conditions regarding deformation
mechanism and sliding behavior may be assumed to be generally applicable and any
inherent uncertainties are considered acceptable. In those types of cases, sliding-block
methods can be very useful even if soil liquefaction or cyclic softening is involved.
Examples of the use of sliding-block methods for analysis of seismic deformations in
situations involving liquefaction or cyclic softening of soils have been presented by
Moriwaki et al. (1989), Seed (1987), and Olson and Johnson (2008), amongst others.

5.2 Basis and Limitations

In the original Newmark (1965) method, the slide mass is represented as a single rigid
block with resistance to sliding mobilized along the sliding surface as shown in Figure 5-
1. The slip surface, irrespective of its shape, is represented by an inclined plane. As
discussed in Chapter 4, subsequent modifications to the original Newmark method have
considered the effect of deformability and dynamic response of the dam or slope during
earthquake loading, and the coupling of the slide mass acceleration with the slide mass
displacement.

30
 Figure 5-1. Newmark’s Sliding Block Method Concept (after Wilson and Keefer, 1985
from Rauch, 1997). Note that in this figure the yield acceleration is labeled ay and is
oriented parallel to the slope. Elsewhere in this document, the yield acceleration is
labeled ky and is assumed to be horizontal as is common in dam engineering practice.

The sliding-block approach is based on the concept that accelerations induced during the
earthquake may temporarily exceed the available shear resistance along the base of a
slide mass. The available resistance can be expressed as a yield acceleration (ky), which is
the horizontal acceleration of the slide mass that causes yielding on the slide plane. The
yield acceleration depends on the geometry of the slope and sliding surface as well as the
mass of the slide block and the shear strength along the sliding surface. The applied
loading is expressed as the average horizontal acceleration of the slide mass. In a

31
decoupled analysis, the slide mass acceleration is assumed to be independent of the
sliding displacement and is typically calculated assuming that there is no yielding on the
slide plane.

Although the Newmark method is a useful tool in engineering practice, it has the
following limitations:

1. The method is only applicable to cases where the deformation mechanism

may be approximated by a well-defined slide mass and deformation behavior
along the failure surface may be assumed to be rigid-plastic.
2. The analysis does not describe the variation in displacement across the slide
mass, the direction of displacement at any particular point, or account for
complex or multiple displacement geometries.
3. The effect of excess pore pressure generation, liquefaction, or cyclic softening
on the stiffness and strength of susceptible soils is not considered. Although
these effects can be addressed through use of degraded strengths in the
estimation of ky, or by specifying a ky that is a function of computed
displacement, this is an approximate approach that requires additional effort
and involves uncertainty.
4. The assumption of a rigid block and foundation neglects the flexibility of the
slide block and foundation and its effects on the seismic response of the block
and the accumulation of displacement.
5. Neglecting the effect of soil yielding on the block response can affect the
estimated displacement.
The simplest approach for performing Newmark analyses is to assume that the dam
accelerations equal the base accelerations (as described by Newmark, 1965). Dynamic
response analysis of the dam is not performed. Although this approach ignores the
dynamic response of the dam, it has often been used for relatively small dams.

When the accelerations of the slide mass are expected to vary significantly from the base
ground motion, dynamic response analyses using equivalent-linear techniques may be
used to estimate the average slide mass accelerations. This is the case, for example, of
medium-sized to large dams, or when the dam is founded on deep alluvial deposits or
complex foundation conditions. A common tool to perform equivalent-linear two-
dimensional dynamic response analyses of dams is the computer program QUAD4M
(Hudson et al., 1994). One-dimensional response analyses with computer programs such
as SHAKE (Schnabel et al., 1972) are satisfactory to calculate the dynamic response of
horizontally stratified soil deposits but are generally not satisfactory to estimate the
average accelerations of a two-dimensional slide mass.

5.3 Common Steps to Perform a Sliding-Block Analysis

The following steps are generally required to perform a sliding-block analysis (Makdisi
and Seed, 1978).

32
 1. Perform a limit-equilibrium slope stability analysis to calculate the yield
acceleration, ky, for various critical slip surfaces.
2. Perform a dynamic response analysis to determine the average horizontal
acceleration time history for each critical slide mass.
3. Perform the calculation of displacement for each slide mass and ky and the
corresponding average acceleration time history using the Newmark double
integration procedure.
Each of these steps is discussed briefly below.

5.4 Step 1 – Perform Limit-Equilibrium Slope Stability Analysis

The objective of this step is to estimate the yield acceleration, ky, for each slip surface
that is being considered for the sliding-block analysis. A pseudo-static, limit-equilibrium
stability analysis is commonly used to estimate ky (Duncan et al., 2014). In this analysis, a
horizontal acceleration is applied to the slope stability model and the factor of safety for
the critical slip surface is calculated. A series of analyses are performed using different
values of applied acceleration until a factor of safety of unity (1.0) is obtained by trial and
error or by interpolation.

The value of shear strength appropriate for modeling competent or compacted clay soils
in the limit-equilibrium analysis (i.e., the yield strength) may be less than the peak
undrained strength. A reduced shear strength is often used to represent the stress at which
the material behavior may be assumed to transition from near-elastic to near-plastic.
Cyclic strength tests can be used to define this yield strength for a clayey soil. However,
when these tests are unavailable or impractical, it has been common practice to use about
80% of the peak undrained strength for compacted clay soils as suggested by Makdisi and
Seed (1978). This 20% reduction in strength is based primarily on data from cyclic tests
performed on undisturbed and compacted clayey samples and was considered to apply
provided the earthquake-induced shear strain amplitudes are no greater than about one-
half the static failure strain. The latter was found to be the case from response analyses of
selected dam models for strong ground motion.

It should be noted that a 20% strength reduction is generally considered conservative for
dense or compacted embankment soils, except perhaps in the case of long-duration,
strong shaking from large magnitude earthquakes. The dynamic strength of soils is
affected by many factors including the maximum induced strain, the rate of loading,
cyclic degradation, and yielding at stress levels less than the peak static strength.
Therefore, selection of an appropriate shear strength for analysis of embankment seismic
deformations requires thoughtful consideration of those factors.

In addition to the critical slip surface under pseudo-static loading, additional slip surfaces
should generally be considered and analyzed to check if other surfaces are associated
with larger calculated seismic displacements (and are thereby more critical) due to the
structural response of the dam. For example, although a slide mass extending one-half the

33
height of an embankment may have a higher yield acceleration than a full-height slide
mass, the calculated seismic displacement for the partial-height slide mass may be larger
than that for the full-height mass because the amplitude of the average acceleration is
likely to be higher for the former than the latter.

An important factor that should be considered when selecting the slip surfaces for
analysis is how each surface might affect the overall performance of the embankment
dam. For example, slip surfaces that include a significant portion of the crest or that
engage critical elements such as internal filters are generally of greater concern than
surfaces associated with shallow sliding of a dam slope.

Slip surfaces to consider for embankment dams may include the following:

1. The slip surface that results in the minimum FOS without a seismic
coefficient;
2. A slip surface that includes at least 50% of the crest width;
3. A slip surface that includes the entire crest width; and
4. Partial and full dam height slip surfaces.
Figure 5-2 illustrates typical slip surfaces that might be analyzed for displacements in the
downstream direction. It should be noted that only circular slip surfaces are shown in
Figure 5-2, but the possibility of non-circular slip surfaces should also be considered and
evaluated. This is particularly important in cases where a non-circular deformation
mechanism is potentially defined by weak zones or layers in the dam and foundation.

1
2
3

Figure 5-2. Schematic Depiction of Possible Slip Surfaces to Consider for a Sliding
Block Deformation Analysis of Dam (for clarity, only shown downstream).

5.5 Step 2 – Develop Average Acceleration Histories

The average horizontal acceleration time history of the slide block is often obtained from
a dynamic response analysis of the dam. At each time step, the average acceleration of
the slide block may be obtained using the method described by Chopra (1966), which
consists of calculating the resultant of the forces along the boundary of the slide block
and dividing that resultant by the mass of the block. In special cases, such as slopes or
embankments that are small (e.g., < 10 ft in height) and stiff, the average acceleration
history of a full-height slide mass may be approximated using the ground motion at the
base of the slope or embankment. The latter is often assumed equal to the corresponding
free-field motion.

34
A key aspect of a dynamic response analysis is how the stress-strain and damping
behavior of the materials is simulated. A simple method that is commonly used in
geotechnical analysis to approximate such behavior during earthquake shaking is the
equivalent-linear procedure. This method was first introduced and used for evaluating the
seismic response of horizontally-layered soil deposits by Idriss and Seed (1968) and Seed
and Idriss (1970). The procedure involves selecting elastic modulus and damping values
that approximate the anticipated nonlinear soil behavior in linear visco-elastic
calculations. The elastic modulus and damping parameters are adjusted iteratively until
the adjusted values are compatible with the estimated cyclic shear strains. The method
can provide an estimate of the overall dynamic response of the embankment, including
cyclic stresses and estimates of the average acceleration history of a pre-determined slide
mass. However, because material behavior is assumed to be visco-elastic, no residual or
permanent deformations are calculated.

The equivalent-linear method has been shown to be a practical and effective approach to
simulate the recorded acceleration response of dams in numerous studies (e.g., Lai and
Seed, 1985; Seed et al., 1989; Mejia et al., 1991; Makdisi et al., 1991; Mejia and
Boulanger, 1993; Boulanger et al., 1995; Mejia and Dawson, 2010). Two-dimensional
and three-dimensional dynamic response analyses using the equivalent-linear method
have been used successfully to simulate the recorded accelerations at the crest and at
other points of embankment dams using recorded free-field outcrop motions as input to
the analyses. Thus, it is reasonable to infer that the method is an effective approach to
calculate the average accelerations of potential slide masses in dams for use in sliding-
block analyses of seismic deformations.

Care and experience are necessary in using the method because, as noted by Kramer
(1996), the inherent linearity of equivalent linear analysis may lead to erroneous
resonances. Also, the use of an effective shear strain in an equivalent-linear analysis can
lead to an over-softened or under-softened response over portions of an earthquake time
history, depending on the shear-strain response history and the factor used to adjust the
peak cyclic shear strain to the equivalent strain. Accordingly, estimates from equivalent-
linear analyses should be carefully reviewed, particularly for analyses that produce
estimates of cyclic shear strains greater than approximately 1%.

Some examples of computer programs for 2D dynamic response analysis of dams include
QUAD4M (Hudson et al. 1994), QUAKE/W (GeoStudio, 2014), FLAC (Itasca, 2016),
and PLAXIS (Plaxis, 2016). These programs are typically based on finite element or
finite difference method formulations. Figure 5-3 presents an example of a typical finite
element mesh that may be used with QUAD4M. One-dimensional analyses are not able
to adequately represent the geometry of most dams and generally do not produce
reasonable approximations of dam acceleration response. However, 1D analyses of
appropriate soil columns may provide reasonable estimates for initial evaluations of
deep-sliding seismic displacement in dams with gentle slopes.

35
Figure 5-3: Example of a QUAD4M model discretization.

5.6 Step 3 – Double-integration of acceleration history

The permanent displacement of the sliding block is calculated by integrating the portions
of the average acceleration record that are above the yield acceleration (ky) to obtain the
relative velocity of the block. When the average acceleration drops below the yield
acceleration the block will decelerate due to the resistance acting on the base until the
relative velocity reaches zero or another acceleration pulse exceeds the yield acceleration.
The relative velocity record of the block is then integrated to obtain the accumulated
displacement. This procedure is illustrated in Figure 5-1. A number of tools can be used
to perform these calculations including general purpose computing programs such as
MATLAB and Python or specialized software such as SLAMMER (Jibson et al., 2013).

The integration is usually performed twice for each earthquake record. One integration is
performed for one polarity of the motion (i.e., positive or negative orientation), and the
second for the opposite polarity. A Newmark-type analysis generally reflects only the
positive or negative polarity of the acceleration trace, and the two orientations for the
same earthquake record can produce significantly different values of displacement. It is a
common practice to select the maximum calculated displacements of the two polarities.
This is especially the case when a limited number of time histories is used for the
analysis.

It is important to note that integration of the difference between the average acceleration
of the sliding block and the yield acceleration, both of which are defined as horizontal
vectors, produces a horizontal displacement of the sliding block. For sliding blocks that
intersect the crest of a dam, the crest settlement may be estimated vectorially from the
block sliding-surface displacement that is kinematically consistent with the calculated
horizontal displacement of the block.

36
 6.0 Nonlinear Deformation Analyses
6.1 Overview

Nonlinear deformation analyses (NDAs) are a useful approach to estimate permanent

deformations of embankment dams due to earthquakes. NDAs are commonly performed
using a dynamic finite element or finite difference program with one or more nonlinear
constitutive models that capture the response of the various materials in the dam and
foundation. The earthquake loading is commonly applied to the base of the model and
permanent deformations are calculated as plastic shear strains accumulate due to yielding
of the dam and foundation materials. The accuracy of the deformation estimates depends
on the numerical formulation, which must realistically capture the propagation of seismic
waves and inertial effects, and the constitutive model, which must be able to reproduce
the aspects of soil behavior important to the problem being studied. These include not
only the strength and stiffness of the soil but may also include more detailed aspects of
soil behavior such as nonlinear stress-strain response, strain-softening, generation of
excess pore pressures, and groundwater flow.

6.2 Framework for Nonlinear Analysis

The basic framework for NDAs includes two main components: the numerical platform
used to solve the governing equations and the constitutive models used to represent the
soil behavior. Each of these components will be discussed in some detail below.

6.2.1 Numerical Platform

NDAs are commonly performed using a finite element or finite difference program.
Examples of programs which have been used for the seismic response of dams in the U.S.
include FLAC (Itasca, 2016), PLAXIS (PLAXIS, 2016) and LS-DYNA (LSTC, 2015).
This list is not exhaustive and represents only some of the more common software
packages used in current practice. The selected platform must be able to appropriately
model dynamic wave propagation of the earthquake motion and plastic deformations
associated with soil yielding. In many cases, the ability to address large displacements
and the associated changes in model geometry is needed. Each of these requirements
poses numerical challenges and the selected program should have evidence of extensive
verification by the program developers to ensure that these important analysis aspects can
be appropriately modeled within the program. This is separate from the validation
exercises which are discussed in the next section.

The application of a numerical platform for an NDA involves modeling choices which
may have a large impact on the results. This includes the selection of a numerical grid to
represent the internal and overall geometry of the structure, boundary conditions for each
stage of the analysis, loading characteristics and application protocols (e.g., loading type,
steps, and locations), and numerical analysis parameters (e.g., numerical damping,
integration algorithms, iteration tolerances, dynamic time increments). The discussion of
these is beyond the scope of this report, but the effects of these choices should be

37
investigated and documented by the user for the problem being studied. This may be
accomplished through a well-defined sensitivity study which examines the effects of
changing these parameters on the simulation results.

6.2.2 Constitutive Model

Constitutive models provide the numerical framework that relates increments of strain to
increments of stress. The response of the constitutive model will determine the amount of
plastic strain which occurs in a material and therefore how much permanent deformation
of the dam is predicted. The response of the constitutive model depends on the initial
conditions (i.e., state of stress from static initialization), the input parameters of the
model, and the imposed loading. The behavior of nonlinear constitutive models may
evolve during the analysis based on the loading and the response of the global analysis
model.

The choice of constitutive models for advanced analysis should be based on how well
each model represents the soil behavior that is considered important to the problem under
consideration. The simplest models typically used to represent soil in NDAs are those
based on the Mohr-Coulomb failure criteria (Figure 6-1a). In these models, the stress
states and increments below the yield surface are linear elastic, while those that attempt
to exceed the yield surface lead to plastic strains, which may include plastic dilation or
contraction. While these models cannot simulate multiple details of complex soil
behavior, they have been useful for modeling competent soils such as compacted clays,
dense rockfill, or weathered bedrock. A simple extension of these models may include
nonlinear elasticity or hardening/softening behavior, which can more accurately capture
stiffness degradation and damping at smaller strain levels. The latter type of models is
referred to herein as extended Mohr-Coulomb models.

More sophisticated models include nonlinear effective stress models that may be coupled
with pore water response and flow. These models often require a significant number of
input parameters which must be calibrated for each soil but can produce more realistic
responses compared with simpler models (Figure 6-1b). These types of models are
referred to herein as coupled effective-stress plasticity models. All models should be
validated against laboratory data, case histories, and design correlations that represent the
types of loading and soil behavior of significant concern to an analysis. The sensitivity of
the model predictions to key input parameters should be understood by the user to
properly consider uncertainty in the results.

38
 0.15
(a)

0.1

Shear stress ratio (τ/σ′vc )

0.05

-0.05

-0.1

-0.15
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5
Shear strain (%)
0.15
(b)

0.1
Shear stress ratio (τ/σ′vc )

0.05

-0.05

-0.1

-0.15
-6 -4 -2 0 2 4 6
Shear strain (%)

Figure 6-1. (a) Comparison of Strain Controlled Cyclic Direct Simple Shear Test Data on
Kaolinite Clay (after Mortezaie and Vucetic, 2013) and a Mohr-Coulomb Model with a
Similar Undrained Strength and User-Selected Secant Modulus (Red). (b) Comparison of
Stress Controlled Cyclic Direct Simple Shear Test Data on Ottawa F-65 Sand (after El
Ghoraiby and Manzari, 2019) and the PM4Sand Constitutive Model (Red).

The use of more complex models does not guarantee the results will provide a better
basis for evaluating a dam. It is necessary to check that each model being used can
adequately simulate the key material behaviors in each part of the structure. The use of
both complex and simple constitutive models in a single analysis may be appropriate. The
strength and limitations of each model as used in an analysis should be well understood
and the models should be selected, applied and the results evaluated with judgment.
Some general attributes of an appropriate constitutive model for use in NDAs are:

39
 • A sound basis in mechanics.
• Ability to simulate key aspects of soil behavior important to the anticipated
dam response and the objectives of the analysis.
• Satisfactory reproduction of the soil response from applicable laboratory tests,
such as stress-strain behavior, pore pressure generation or dissipation and
strength.
• Input parameters that can be defined using common laboratory or in-situ tests
in a reasonably transparent manner.
• Behavior consistent with laboratory test data under stress paths analogous to
those that may occur during earthquake loading.
• Proper validation through the simulation of well documented case histories or
physical models.
Key features of selected constitutive models from U.S. practice, including references for
further reading, are described in more detail in Appendix B. In addition to the Mohr-
Coulomb model, the types of models discussed therein include extended Mohr-Coulomb
models and coupled effective-stress plasticity models, as listed in Table 6-1.

Table 6-1. Constitutive Models from U.S. Practice Described in Appendix B.

Type of Model Name of Model
Linear elastic-plastic Mohr-Coulomb
Extended Mohr-Coulomb UBCHYST
Roth Cyclic-Counting
UBCTOT
Coupled effective-stress plasticity PDMY
PM4Sand
PM4Silt
SANISAND
UBCSAND
Wang

6.3 Approaches to NDA

This section briefly describes some of the available approaches to performing an NDA,
including simplified gravity turn-on methods, decoupled analyses, and fully or partially
coupled effective stress analyses. This section is not meant to be an exhaustive discussion
of all possible approaches, but rather to summarize some of the more common procedures
encountered in practice. The complexity of the analysis method for a given project should
be consistent with the quality and completeness of the site characterization, the
embankment and foundation zoning and material characteristics, the size and importance
of the dam, the sensitivity of dam performance to the amplitude and pattern of
deformations, and the resources available for the analytical work. Important parameters
that must be resolved prior to selecting an appropriate analysis approach include an
understanding of the geomorphology of the site, the 3D definition of soil strata, material

40
properties for each soil unit, and the definition of seismic loading. For evaluation of
existing dams, it is also important to understand the construction history of the dam and
the data on as-built properties of the materials. Appropriately defining these parameters is
a critical step in any deformation analysis, not just for NDAs.

Advanced analyses for existing dams are often valuable when (1) the uncertainty in the
predicted performance from simplified techniques, such as those described in the
previous sections, significantly impacts the outcome of a dam safety or performance
evaluation, (2) the additional insight gained from understanding possible deformation
patterns could improve the evaluation of dam performance and/or the selection and
design of remediation options, (3) some aspect of soil-structure interaction is important to
the evaluation, or (4) interim safety modifications are being made, such as a reservoir
restriction, that could benefit from an improved understanding of potential deformations.
It is not uncommon for simplified techniques to be adequate for evaluating the safety of
an existing structure while complex methods are required for designing and optimizing a
selected remediation. The analysis of some remediated structures can be significantly
more complex than those of typical non-remediated dams, and simplified tools may not
be applicable. Advanced analyses may also be needed in cases where the loading or soil
behavior is not adequately represented in the simplified approaches (e.g., tailings dams or
loading from large subduction zone earthquakes).

There is often a benefit in using multiple numerical platforms and constitutive models
when analyzing potential deformations of an embankment dam to understand the
sensitivity of the solution to the modeling approach. Understanding where differences
arise between various constitutive models can provide insights into which aspects of the
material behavior may be important for a particular problem. Perlea and Beaty (2010)
summarized experiences with deformation analyses using multiple programs and
constitutive models and highlighted the importance of parametric studies. Montgomery
and Abbaszadeh (2017) compared results for a hypothetical embankment dam founded
on a liquefiable layer using two constitutive models. The reader is referred to these two
studies for additional information on these types of comparative analyses.

This report is primarily focused on 2D (plane strain) analyses as these are far more
common than 3D analyses for seismic cases. This is due to the significantly greater effort
required for 3D analyses compared with 2D analyses (Mejia and Dawson, 2010). Cases
where 3D analyses may be needed include analysis of remediations with a 3D footprint
(e.g., zoned soil improvement), dams in narrow canyons (crest length less than 3 to 5
times the dam height; Mejia and Seed, 1983), and dams with irregular rock surface
topography that may lead to variations in shaking or abrupt changes in embankment
deformations across the footprint of the dam. If a 3D analysis is to be performed, it is
important to verify that the selected constitutive model has been formulated for use in 3D
as plane strain models cannot capture 3D stress states. Even in cases where a 3D model is
used, 2D analyses are commonly analyzed in parallel to ensure that reasonable results are
being obtained from the 3D model.

41
6.3.1 Simplified Gravity Turn-on Analyses

NDAs are often used to estimate potential post-earthquake deformations by solving for
equilibrium of a structure with reduced, post-cyclic shear strengths. These analyses are
typically performed by evaluating the initial static stress conditions within the
embankment using drained strengths. Zones which are estimated or judged to undergo
strength loss during the earthquake cyclic loading are then assigned a reduced strength
and the model is used to analyze equilibrium under gravity loading. The selection of
zones on which to impose strength reduction may be based on judgment or results from
another type of analysis (e.g., the Seed-Lee-Idriss method with an equivalent-linear
dynamic response analysis and a liquefaction triggering analysis, Chapter 3). The
deformation that occurs before the model reaches equilibrium will depend on the number,
location, and size of zones which are assigned a reduced strength and the magnitude of
the strength reduction. Depending on the numerical platform, the model may not reach
equilibrium before the numerical solution must be stopped due to excessive distortion of
the numerical grid.

These types of analyses have the advantage of being conceptually simple to apply and
understand. They also can make use of relatively simple constitutive models (e.g., Mohr-
Coulomb) which require few parameters and can be more computationally efficient than
the advanced models discussed in the next two sections. The main limitations to this type
of analysis are that the effects of inertial loading on deformations are ignored and that the
extent and magnitude of strength reduction are estimated using some other procedure.
Uncertainty introduced by these limitations should be considered when interpreting
deformation results from these types of analyses.

The gravity turn-on approach can be modified to apply the seismic loading to the model
and impose the reduced strengths at either the start of the earthquake or at some other
pre-determined time. These modified analyses attempt to account for the effect of inertial
response on the estimated deformations in a simple way. However, this modified
approach may produce unrealistic estimates of inertially-driven displacements resulting
from base-isolation effects associated with low strengths imposed uniformly and
instantaneously over significant zones or layers of the model.

Simplified NDAs are most appropriate for cases where significant strength loss is
expected over a large zone of a model and the deformations are likely to be dominated by
post-earthquake stability rather than inertial effects during the earthquake loading. These
types of analyses were used to evaluate the stability of Duncan Dam by Pillai and
Salgado (1994) and Casitas Dam by France et al. (2000). Oaks and Scott (2014)
presented results from an analysis of Scoggins Dam which evaluated different values of
undrained shear strength in the soft clay foundation and considered gravity-only and
gravity-with-seismic loadings.

42
6.3.2 Decoupled Analyses

Decoupled analyses consider the dynamic response of the soil during earthquake loading
but do not couple the responses of the soil and the pore water. Without this coupling,
these simulations cannot simulate complex soil behavior and pore-water pressure
response. These types of analyses often use simplified constitutive models, which
commonly utilize a Mohr-Coulomb failure criterion with either a drained or an undrained
strength depending on the expected response of the material. Some practice-oriented
analysis methods use the Mohr-Coulomb model with simplified techniques to generate
excess pore pressures in response to cyclic loading, which can be useful for
approximating some aspects of liquefaction response. It is important to note that while
Mohr-Coulomb-based models are often easier to use, they offer a crude representation of
true soil behavior (i.e., Figure 6-1a) and uncertainty introduced by this approximation
should be considered when interpreting deformation results from such models.

Decoupled models represent an additional level of complexity compared to the simplified

gravity turn-on NDAs described above as the onset of strength loss and inertial loading
are directly considered. A decoupled analysis may also need to be followed with a post-
earthquake gravity only analysis with post-cyclic strengths to estimate additional
deformations due to softening.

This type of analysis is often useful for modeling competent soils such as compacted
clays, dense rockfill, or weathered bedrock. The Mohr-Coulomb models commonly used
in a decoupled analysis may be extended to include hysteretic behavior and/or strain-
softening which may be important to some problems. Examples of some of these types of
models are discussed in Appendix B.

6.3.3 Coupled Effective Stress Analyses

Coupled effective stress analyses explicitly consider the coupled response of the soil
skeleton and the pore water. These types of analyses can reproduce many important
aspects of dynamic soil behavior including excess pore pressure generation and
dissipation and are often most useful for analyses that examine the possibility of
liquefaction (Figure 6-1b). Coupled effective stress analyses are the most complex of the
approaches discussed in this report and are also the most computationally intensive. The
accuracy of any analysis depends on the ability of the constitutive model to reproduce the
aspects of soil behavior that are important to the problem at hand, on the quality of the
calibration of model parameters, and, for coupled analyses, on the ability of the numerical
platform to perform coupled fluid and mechanical calculations. Details on some
constitutive models that can be used in coupled analyses for liquefaction are discussed in
Appendix B.

Some dams may be analyzed using coupled effective-stress models in some zones of the
analysis mesh and simpler models in other zones to lessen the computational effort
needed for the analysis. It may also be advantageous to perform coupled effective stress
analyses with two conditions for groundwater flow: a) undrained conditions and

43
b) allowing transient flow. This allows the modeler to examine the effects of fluid flow
on the results.

6.3.4 Need for Validation

Any analysis method applied to dams should have some track record of validation. In this
context, validation is understood as demonstration that the method is capable of
reasonably approximating key aspects of material and system behavior that are important
to the problem under consideration. Validation is a critical exercise for all analysis
methods, not just NDAs, and should consider uncertainties in both the results of the
experiments or case histories being used for validation and the response of the numerical
model. Validation studies should include a demonstration of the sensitivity of the model
response to reasonable ranges in model parameters and should discuss the overall
protocols for selecting properties.

Two types of validation are needed for NDAs: comparison of constitutive model behavior
to laboratory element tests and comparison of overall modeling procedure results to well-
documented case histories or physical model test results. The use of well-documented
case histories for validation is important to ensure an analysis model can perform well in
simulating field conditions that may not be realized in the laboratory or in physical model
tests, but it is important to also consider how uncertainties in a case history affect the
comparison with the model. The validation exercises discussed in this section are
separate from the verification of the numerical framework discussed in the previous
section. The burden of validation for constitutive models often falls on the developer of
the model whereas the validation of the overall NDA framework may fall on the analyst.
These two types of validation are discussed in more detail below.

Validation of a constitutive model should demonstrate that the model can reasonably
reproduce the results of relevant element tests that explore the aspects of soil behavior
important to the problem being analyzed. This validation may be performed using results
from laboratory tests and applicable empirical design relationships. An example
validation exercise is illustrated in Figure 6-2 in terms of the effects of sustained shear
stress ratio on cyclic resistance to liquefaction obtained from a constitutive model and
compared to published information.

Common relationships important for the seismic analysis of earth dams include modulus
reduction and damping curves, liquefaction triggering relationships, and stress-dilatancy
relationships. The specific relationships or laboratory test results selected for validation
will vary with the problem being analyzed. For some materials, common empirical design
relationships may not be appropriate (e.g., mine tailings or coal combustion residuals).
For these materials, it may be necessary to perform cyclic testing to compare material
response with the selected model.

The validation of a constitutive model is often performed by the model developer and
should evaluate the capabilities of the model over its range of intended uses. The user
should be familiar with how the validation was performed and what behavior the model is

44
and is not able to capture, and how well (or not) it approximates key behaviors pertinent
to the analysis at hand. The user may need to perform their own validation if working
outside the range of properties or loading conditions considered by the model developer.
The validation of the constitutive model should be well documented as discussed in
Chapter 7.

Figure 6-2. Example Validation of UBCSAND Against the Static Shear Stress Correction
Factor Developed by Idriss and Boulanger (2003) (Figure from Beaty and Byrne, 2011).

Validation of the overall NDA framework should be performed through evaluation of

well documented case histories or physical models (e.g., centrifuge tests). When
available, case histories or physical models that are similar to the dam (in key respects of
its configuration, materials, and loading) should be evaluated. It is often desirable to
analyze at least one case where significant deformations were observed and one where
limited deformations were observed to demonstrate that the modeling approach can
distinguish between these cases. The general procedures and methods used for
developing the analysis model, selecting input parameters, and setting boundary
conditions and other modeling factors should be consistent between the validation
analyses and the dam analyses. If these procedures are significantly altered for the dam
analyses, additional validation should be performed. To ensure consistency, it is desirable
for the analyst that is responsible for the dam analyses to be the same analyst that
performs the validation. For projects where this is not practical, the analyst should have a
thorough understanding of the techniques used in developing and performing the
validation analyses. The validation of the NDA framework should be well documented as
discussed in Chapter 7.

For some problems, there may be a lack of well-documented case histories or

experiments with similar geometries or loading conditions to use for validation. This may
happen for tailings dams, where the combination of unique geometries and materials may

45
make it difficult to identify a similar validation case. In these situations, the validation
may need to be performed on the individual components of the NDA framework that are
important to the problem under consideration (e.g., large strain behavior, fluid-
mechanical coupling, cyclic softening) rather than on the entire framework at once.

Many constitutive models have gone through extensive validation which is often
documented or referenced in the manual of the model. As an example, the user manual
for UBCSAND version 904aR (Beaty and Byrne, 2011) shows results for single element
simulations versus laboratory test results and several common empirical design
relationships for liquefaction analyses, which serves as validation of the model itself. The
manual also presents results from a FLAC simulation of the Upper San Fernando dam
which serves as a validation example for the overall NDA procedure (Figure 6-3).
References are provided to other validation examples including additional case histories
and centrifuge tests.

Figure 6-3. Comparison of Observed Displacements at the Upper San Fernando Dam
with Simulations Using UBCSAND (Figure from Beaty and Byrne, 2011).

It should be noted that validation of a model against one or more case histories or
physical models does not mean that the model will perform well in all situations or under
all loading paths. The analyst or reviewer must critically examine the results of each
simulation to ensure that the model is performing as expected and that the results are
reasonable. Sometimes, the details of a constitutive model’s response can be relatively
unimportant for one structure and very important for another structure. An example of

46
this is the response of a model developed for liquefaction analyses and applied to
conditions with large static shear stresses such as those encountered within the shells of
embankment dams (Figure 6-4). Ziotopoulou et al. (2014) examined the response of
several constitutive models to direct simple shear type loading under varying static shear
stresses. Each of the models had been developed for liquefaction analyses and validated
by the respective authors. The results were compared with common liquefaction
triggering relationships and this comparison demonstrated that each model had
limitations in predicting the behavior of sands under static shear stresses. Recognizing
these limitations is an important outcome of validation exercises.
2

DR =
1.5
75%

55%
Kα

35%

0.5
Ziotopoulou & Boulanger (2016):
PM4Sand - ver 3
σ' = 100 kPa
vc (b)
0
0 0.1 0.2 0.3 0.4
Static shear stress ratio, α
Figure 6-4. Results Demonstrating the Range of Responses from Single-Element
Simulations of Undrained Direct Simple Shear Tests Using: (a) PM4Sand Model and (b)
PDMY03 Model. The Trends in These Results Can Be Compared with the Relationships
Shown in Figure 6-2. (Courtesy of Prof. R. Boulanger)

6.4 Analysis Process

The typical analysis consists of defining the problem geometry and numerical mesh or
grid and other input parameters, establishing the initial conditions of the dam, and
estimating the response of the dam to the applied ground motions, including any post-
shaking deformations.

6.4.1 Defining Geometry and Numerical Mesh

The first stage in a numerical analysis is defining the problem geometry and the
numerical mesh or grid. Details on meshing algorithms and best practices are beyond the
scope of this report, but some guidance specific to NDAs for dams is provided. Particular
care should be taken when defining the mesh to achieve a sufficiently refined grid in
critical areas of the dam and foundation (Figure 6-5). These areas may include locations
with complex strain patterns or soil units where significant pore pressure generation is
anticipated, as using too few elements can result in a smearing of values in areas of high
gradients. The mesh density should also be sufficiently refined to ensure that all

47
frequencies of interest in the ground motion can be numerically propagated (e.g.,
Kuhlemeyer and Lysmer, 1973). Parametric studies using different mesh densities are
useful in establishing an appropriate mesh size for a specific problem.

The bottom and lateral boundaries of the mesh should be located beyond the areas where
significant permanent shear strains may occur that affect the dam response. Consideration
should also be given to keeping the boundaries far enough away from the dam that
inaccuracies in predictions at these boundaries do not impact the dam response.

Compliant or free-field boundary conditions assist with reducing reflection of waves

from the boundaries of the model and simulating “free-field” conditions at and beyond
the model boundaries. The bottom boundary is ideally located at a depth of significant
impedance contrast which typically corresponds to an abrupt increase in shear wave
velocity. A compliant boundary may be used within a deep, uniform soil deposit to limit
the overall extent of the model. This type of boundary should be located far enough
below the dam where deformations are expected to be small and such that unintended
boundary effects will not have a significant influence on the dam response. The behavior
of compliant boundaries located within soft soil deposits should be evaluated for
unrealistic boundary deformations that may impact the estimated response of the dam. An
example of where these boundaries might be located for a dam on rock is shown in
Figure 6-5. The compliant boundary was located within the rock unit to reduce the
occurrence of model base bending and rotation during the analysis. This can occur when
using compliant boundaries along the sides and the base of the model due to a loss of
equilibrium at the boundaries as deformations and stress redistribution occur within the
model (Itasca, 2016).

Figure 6-5. Example Dam Mesh Showing Locations of Dynamic Boundary Conditions
(after Montgomery and Abbaszadeh, 2017).

6.4.2 Static Analysis of Initial Conditions

A static analysis is used to establish the initial conditions and may include simulating the
construction of the dam, the reservoir filling, and an analysis of long-term steady state
seepage. The complexity in the static analysis sequence should consider the level of detail
required for the dynamic analysis. In many cases, the static analysis can be simplified to
reduce the computational effort. Examples of simplifications may include using elastic or
Mohr-Coulomb constitutive models rather than more complex models, combining aspects
of the construction sequence (e.g., building the entire dam before raising the reservoir
instead of simulating multiple construction stages of the dam and seasonal fluctuations in

48
the reservoir during construction), or by using decoupled steady state seepage analyses to
establish initial pore pressures rather than simulating the coupled hydro-mechanical
process of filling a reservoir. These simplifications are often justifiable as the primary
purpose of the static analysis is usually to establish a reasonable initial stress state for the
dynamic stage and not to model potential deformations during construction. The potential
for variations in initial conditions due to these simplifications can be addressed through
parametric study.

The reasonableness of stress distributions should be confirmed at the end of the static
analysis and before start of the dynamic evaluation. Factors that should be evaluated
include the magnitude and orientation of the principal stresses, the overall distribution of
the lateral stress coefficient (Ko) where Ko = σʹh / σʹv, the static shear stress coefficient (α)
where α = τxy / σʹv, and flow nets from seepage analysis. For existing dams, piezometer
data can be used to ensure the seepage analysis is appropriately capturing the pore
pressures within the dam. Records from inclinometers or survey markers can also be
helpful for assessing the reasonableness of the static portion of the analysis.

6.4.3 Dynamic Analysis of Earthquake Loading

The seismic phase of the analysis is started with the model in equilibrium under the static
stresses. During this stage, dynamic boundary conditions are applied to allow for the
seismic waves to propagate through the model. The specific choice of dynamic boundary
conditions will depend on the numerical platform, site characterization, model geometry,
and the goals of the analysis. Each input ground motion should be analyzed for the full
duration of motions capable of inducing permanent deformations and until the predicted
dam response has stabilized.

Input ground motions should be applied in a manner consistent with the selected
boundary conditions and the goals of the analysis. Ground motions may be applied as
acceleration or shear stress time histories depending on whether rigid or compliant
boundaries are being utilized (e.g., Itasca, 2011). Recorded or assumed outcrop motions
may need to be deconvoluted to ensure they are suitably represented by the input motions
at the base of the model (e.g., Mejia and Dawson, 2006).

Using vertical motions in addition to horizontal motions is generally more appropriate

than applying horizontal motions only, as the combined motion represents a closer
approximation of the seismic loading conditions. The inclusion of vertical motions often
has a minor impact on the deformation estimates when compared with deformations
calculated using the horizontal motions only (e.g., USBR, 2015; Perlea and Beaty, 2010).
However, that is not always the case. For example, Bureau et al. (2008) found that
including vertical motions in a NDA increased settlements and deformations for the dam
and motions they analyzed by factors ranging from a small percentage up to a factor of
two. Given the potential effects of vertical motions on calculated dam deformations, it is
generally advisable to consider vertical motions in NDAs of embankment dams.
Parametric analyses may be useful to examine the effects of vertical motions on the
calculated response of a dam.

49
6.4.4 Post-earthquake Stability Analysis

The potential for strength loss can be an important factor in the overall response of dams
where large portions of the foundation or embankment contain soils that are susceptible
to liquefaction or cyclic softening. Current constitutive models and techniques for NDA
may be unable to adequately capture the potential for strength loss due to cyclic softening
or void redistribution in liquefied soils, induced by the earthquake. This is often the case
for coupled effective stress models because the stress-strain behavior of these models is
based on trends observed in the laboratory, which do not reflect the effects of void
redistribution in the field. Thus, the constitutive models do not simulate the post-
liquefaction strength loss that is back-analyzed from flow slides and this strength loss
must be considered separately.

A common method in current practice is to perform a post-earthquake stability analysis

using Mohr-Coulomb-type constitutive models with post-earthquake undrained shear
strengths. This method is similar to the simplified gravity turn-on NDA approach
discussed above, except that the post-earthquake stability analysis occurs after the
earthquake stage in the analysis and therefore, any deformations obtained in this stage
further deform the embankment model at the end of shaking. The results of the
earthquake stage may also be used to judge which zones should be assigned a reduced
shear strength in the post-earthquake analysis. Details on this approach and
considerations for estimating which zones have “liquefied” are discussed by Boulanger et
al. (2015) and Chowdhury (2019).

For zones where liquefaction triggers, the reduced strength is often estimated from post-
liquefaction residual strength correlations (e.g., Idriss and Boulanger, 2007; Weber, 2015;
Kramer and Wang, 2015). For soils susceptible to cyclic softening, the post-earthquake
strengths may be estimated from laboratory and in situ strength tests. This decoupled
approach can be used to check stability, but it is unable to capture important aspects of
soil response like the potential for strength loss during shaking, or for pore pressure
redistribution to lead to delayed deformation or failure (e.g., as was the case for the
Mochikoshi tailings dam, Ishihara, 1984). Future developments in constitutive models
may be better able to capture these behaviors explicitly in a NDA and reduce the need for
a separate post-earthquake analysis.

6.5 Special Considerations

Advanced analysis techniques are often sensitive to one or more input parameters, such
as the magnitude of loading or the parameters of the constitutive model. The analyst
should perform sufficient parametric studies to identify the key parameters for a given
model and their influence on the predicted response. For material properties, field
explorations or laboratory tests can be used to refine estimates of important parameters.
Sensitivity to loading can be more challenging as dams with highly nonlinear materials
such as liquefiable soils or sensitive clays may perform in a brittle fashion. In these cases,
it is helpful to develop fragility relationships that relate one or more response measures

50
for the dam (e.g., crest settlement) to an intensity measure for the earthquake loading
(e.g., PGA, spectral acceleration, or Arias Intensity). Fragility relationships are key
components in a risk-based evaluation but may also be valuable for deterministic
evaluations. Examples where these relationships have been developed from the results of
NDAs are given by Ruthford et al. (2011) and Oaks and Scott (2014).

Many constitutive models and numerical procedures developed for simulating the shear
response of soils under cyclic loading do not appropriately capture the volumetric
response associated with consolidation (e.g., Ramirez et al., 2018) or seismic
compression (Lasley et al., 2016). For problems where estimating volumetric strains may
be important (e.g., reconsolidation of liquefied deposits, seismic compression of
unsaturated soils or rockfills), volumetric strains may need to be estimated using
simplified approaches, as discussed in Chapter 3.

Simplified approaches can also be coupled to numerical simulations to estimate

volumetric strains for a specific profile. For example, Ziotopoulou and Boulanger (2013)
found that reasonable estimates of reconsolidation strains for liquefied deposits could be
estimated by reducing the elastic modulus of the soil after shaking had ended. They based
their modulus reduction factors on the empirical approach developed by Ishihara and
Yoshimine (1992). Lasley et al. (2016) combined equivalent linear site response analysis
with an empirical damage model to estimate seismic compression of an unsaturated fill.
Few studies have examined the ability of numerical models to estimate the volumetric
response of embankment dams under seismic loading and this remains an important area
for future research.

NDA results may also be sensitive to the reservoir level assumed in a simulation. For
many dams, it is appropriate to perform the seismic deformation analysis assuming the
reservoir is at full pool. However, depending on the dam and typical reservoir operations,
the critical reservoir level for seismic loading may not be at full reservoir pool. This is
particularly true for flood control dams evaluated using risk assessment procedures
(Ruthford et al., 2011) as the full pool may only occur for a short duration each year and
is also the case for pumped storage projects. The reservoir operation may need to be
considered and simulations performed for several reservoir conditions.

NDAs of dams on weak foundations or with large potential for strength loss may result in
large deformations. Simulation of these deformations is a challenge for many finite
element and finite difference programs due to the numerical formulation and distortion of
the grid or mesh. Excessive distortion may cause the solution process to stop unless re-
meshing techniques are used to allow the simulation to continue. Displacement estimates
at this large level of deformation have considerable uncertainty, which should be
recognized and clearly documented. Recently developed numerical analysis approaches
that may be better suited to this type of problem are discussed in Section 6.6.

Current analysis methods are not able to track certain physical phenomena that may be
important to the seismic response of dams. It is important to recognize these limitations
as certain failure modes may be dependent on these processes. Examples of these

51
processes include crack propagation, response of unsaturated zones, strains due to
reconsolidation of liquefied layers or seismic compression, and potential mixing of soil
layers or reservoir water with soil during large deformations. Nonetheless, reasonable
solutions for some of these problems may be obtained using currently available models,
provided the model parameters and analysis platforms are adequately selected to
approximate the controlling aspects of the phenomena. For example, Mejia and Dawson
(2019) showed that 3D nonlinear dynamic analysis procedures together with a strength-
of-materials approach are capable of reasonably simulating the approximate location and
extent of earthquake-induced transverse and longitudinal cracking in embankment dams.
Some of these issues may be addressed by moving away from continuum-based modeling
approaches as is discussed in the next section.

6.6 Future Directions

NDAs remain an active area for research and new methods are continually being
developed, which may periodically replace the methods used in current practice. Soga et
al. (2015) provide an overview of current numerical methods available for large
deformation analyses of geomaterials. Some of the methods that have shown potential for
use in NDAs of earth dams include the material point method (MPM), discrete element
method (DEM), and Arbitrary Lagrangian Eularian (ALE) methods. Each of these
methods is briefly described below.

The MPM is a particle-based numerical technique where the continuum is represented by

a series of particles that move through a fixed grid (e.g., Jassim et al., 2013). The
properties of the continuum are stored in the particles while the fixed mesh is simply used
to compute gradients and solve equations of motion. Recent developments in the MPM,
including extensions to solve coupled hydro-mechanical problems, are discussed by Soga
et al. (2015). Advantages of the MPM include the ability to model large deformations
without mesh distortion, the ability to simulate coupled problems, and the use of
advanced constitutive models. Disadvantages are high computational cost and the need to
use higher order shape functions to avoid numerical errors. Yerro et al. (2019)
demonstrated the use of MPM to perform a runout analysis of the Oso landslide (Keaton
et al., 2014), an important issue for tailings dams and some embankment dams.

The DEM is a numerical technique which simulates the interaction of particles, such as
soil grains, by directly modeling the contact mechanics between the individual particles.
The method was initially developed by Cundall and Strack (1979) and has been extended
by subsequent researchers. Recent developments include coupling the DEM with fluid
simulation techniques to solve coupled problems including liquefaction (e.g., El Shamy
and Abdelhamid, 2014). Advantages of the DEM include the ability to directly simulate
the micro-mechanical aspects of granular soil behavior without the use of advanced
constitutive models. Disadvantages include the need to simulate a very large number of
particles and the high computational cost. The ability of the DEM to simulate full-scale
problems is still beyond the capabilities of current practice, although hybrid approaches
may be promising for future applications (e.g., Guo and Zhao, 2014).

52
The ALE method is a continuum approach which combines aspects of Lagrangian and
Eulerian numerical approaches (e.g., Van den Berg et al., 1996; Hu and Randolph, 1998).
This is accomplished by taking a traditional Lagrangian step in which the material and
mesh deform together followed by an Eulerian step in which the mesh is moved and
smoothed before remapping the field variables to the new mesh. Then the process is
repeated (Wang et al., 2015). In this manner, relatively large deformations may be
simulated without excessive mesh distortion. Advantages of this method include the
similarity of the formulation to the numerical approaches commonly used in NDAs,
including the ability to use similar constitutive models. Disadvantages include the need to
develop remeshing and remapping techniques to transfer the properties and state
variables between the old and new meshes.

7.0 Review and Documentation of Analysis Results

7.1 Overview

A seismic deformation analysis is often a complex and multi-phased endeavor. The

application of the analysis procedure and the resulting estimates should be reviewed and
evaluated to ensure the analyses are properly formulated and are adequate to support the
conclusions regarding dam response. This technical review should include an assessment
of how well the analysis has simulated the anticipated material behaviors and global
responses. The analysis and review should be documented to provide current and future
stakeholders the opportunity to understand the corresponding processes, results, and
limitations. The level of effort and detail required for the review and documentation can
range from modest to substantial depending on the complexity of the analysis program.

The technical review of an analysis should occur throughout the analysis process. The
review may address many aspects of the work, including the following:

• Suitability of the analysis method (or methods) for the structure and seismic
loading being evaluated.
• Characterization of the dam and foundation to support the selected analysis
approach.
• Implementation of the analysis methods.
• The processes used to detect and correct errors in model coding or
specification of parameters.
• The simulation of key behaviors anticipated for the dam and constituent soils.
• Parametric evaluations performed to identify key factors in the analysis and to
assess robustness of the dam to seismic loading.
• The adequacy of documentation to support the conclusions regarding
embankment behavior.
A seismic evaluation of a dam may involve multiple participants, and documentation is
the means of effectively communicating the analysis approach, evaluation, and results to

53
the participants. Additional technical reviews may be required by owners and regulators
which will rely upon this documentation. Future safety assessments also benefit from
thorough descriptions of past analyses and evaluations. Transparent documentation is a
necessity to meet these objectives.

7.2 General Approach to Review and Documentation

The activities of review and documentation are closely linked. Good documentation can
succinctly demonstrate that the analysis and conclusions have been adequately reviewed.
Technical items to be considered during the review and documentation process may
include the following:

1. Identification of the potential seismic failure modes addressed by the analysis.

2. Description of the site characterization and seismic loading in sufficient detail
to support the analysis method, results, and conclusions. Details of the
characterization may be included in the documentation by reference to
supporting documents.
3. The number, uncertainty, and basis for the required analysis parameters.
4. Identification of the parameters of most importance to the estimated response.
5. A description of the estimated embankment and foundation response
including both general trends and significant local behaviors.
6. Limitations of the approach that are significant to the assessment of dam
response.
7. Uncertainties in the estimates.
8. Assessment of the adequacy of the analyses to the overall dam evaluation
objectives.
9. Quality control efforts including verification of coded subroutines, accurate
specification of input parameters, etc.
Documentation should be tailored to the types of analyses performed and the eventual use
of the documentation. Primary objectives include conveying the approaches,
assumptions, interpretations, parameters, response estimates, and evaluations of the
analysis in sufficient detail to allow sound assessments of the analysis reliability by
knowledgeable participants (Boulanger and Beaty, 2016). Documentation should also
convey the conclusions of the study, explain the basis for those conclusions in a clear and
effective manner, and identify their limitations and appropriateness.

7.2.1 Considerations for Analysts

Documentation is a means of communication but can also serve a role in quality control
efforts by ensuring key aspects of the analysis are reviewed by the analyst and
communicated to qualified independent reviewers. This provides value to owners through
quality assurance and analysis validation.

54
7.2.2 Considerations for Reviewers and Regulators

Reviewers and regulators are often tasked with making recommendations or decisions on
safety and risk that rely, in part, on analysis results. These external reviews should ensure
there was adequate quality control and documentation of the analysis process.

7.2.3 Considerations for Owners

Owners must manage the risks of their facilities and satisfy safety and regulatory
requirements, while also making economic or planning decisions based on these potential
risks. Seismic deformation analyses can be an important aspect of defining these risks.
The required effort to prepare the documentation and provide quality assurance should be
explicitly considered when laying out an analysis program or when requesting proposals
from potential consultants. Independent peer review is a valuable addition to an analysis
program, particularly for high risk or complex projects. Performing these reviews at
intermediate analysis stages (e.g., site characterization and model development) as well
as near the conclusion of the analysis can be an efficient way of executing a robust
analysis program.

7.2.4 Analysis-specific Considerations

The appropriate scope for the review and documentation of a seismic deformation
analysis will depend on the analysis approach. Review and documentation considerations
specific to the analysis approaches described in the preceding sections of this document
are discussed in Appendix C.

8.0 Considerations on Use of Deformation Analyses

8.1 Objectives

The planning, implementation and use of seismic deformation analyses requires

significant engineering judgment. This chapter provides general perspectives on the
development and application of a seismic deformation analysis.

8.2 General Considerations

Developing reliable estimates of embankment deformations is the primary objective of a

seismic deformation analysis. However, there are significant uncertainties in each
analysis method which affect the accuracy of the deformation estimates (e.g.,
uncertainties in the numerical or empirical procedures, limitations in the site
characterization, ability of the constitutive models to fully simulate key soil behaviors,
material variability, characterization of input motions, uncertainty in critical parameters
such as residual or post-liquefaction shear strength). In spite of uncertainty, a key purpose
of a deformation analysis is to improve the understanding of potential embankment
behavior in a rational and physically reasonable manner. Such understanding may be
combined with knowledge of the site, materials, seismic environment, failure modes and

55
processes, and case history precedent to develop assessments of potential dam behavior
and risk.

As discussed in previous sections, deformation analyses are often performed in a phased

manner beginning with simplified approaches and continuing to advanced methods as
needed. The analyses evolve as each step provides an improved understanding of
potential dam behaviors, key factors, and primary uncertainties. This evolution may
induce additional refinements in the analyses, site exploration, or testing. The simplicity
or complexity of an analysis program should be justified based on the effectiveness with
which key behaviors can be modeled and the degree to which a more refined analysis
may or may not affect final decisions regarding dam safety, risk mitigation efforts, or
remedial works. A key aspect of both simplified and advanced analyses is the need for
adequate site characterization that directly considers geologic processes, site investigation
limitations, and uncertainties.

Seismic dam evaluations are typically team efforts that may involve multiple specialists
including geologists, seismologists, and structural and geotechnical engineers. Analyses
benefit from coordination and communication between these disciplines at key stages of
the analysis process. The contributions of technical reviewers can also add significant
value to an analysis program.

8.3 Additional Considerations for NDA

NDAs often provide insights that simpler analysis methods cannot, such as estimated
patterns of deformation, potential for progressive failure mechanisms or processes,
interactions between embankments and embedded structures or remediation elements,
dependence of response on ground motion characteristics, or the relative effectiveness of
remediation options. NDAs can incorporate improved material and structural behaviors
as compared to simplified approaches. Additionally, NDAs often allow for a better
definition of sources of uncertainty and bias than is possible with simplified approaches.
The insights afforded by NDAs should be assessed considering the limitations of the
analysis approach and site characterization. In principle, NDAs can lead to improved
decisions with reduced uncertainty relative to assessments based on simpler and more
approximate analyses. These benefits are associated with an increase in complexity and
required technical skill.

9.0 Concluding Remarks

Reliable evaluation of seismically induced deformations is required to assess the
expected seismic performance and safety of embankment dams and in particular, to
assure the safety of existing dams and the adequacy of new dam designs and dam retrofits
in seismic settings. These guidelines are intended to provide guidance on the evaluation
of seismic deformations of embankment dams and on the use of numerical analysis
procedures for such evaluations. The guidelines are meant to serve as a reference to
engineers involved in performing, reviewing, and procuring analyses of seismic stability
and deformations of dams.

56
The guidelines may be viewed as a primer on the selection of analysis methods, the
implementation of analysis procedures, and the documentation and interpretation of
analysis results based on U.S. practice. They provide an overview of current approaches
for the evaluation of seismic deformations of embankment dams and best practices for
the numerical analysis of seismic deformations and the interpretation of analysis results.
A range of tools, which may be used in a phased manner, are discussed. Such tools
include screening evaluations, simplified analyses, and nonlinear deformation analyses.

The methods for seismic deformation analysis are classed into the following types of
analysis approaches: a) simplified analyses, b) sliding block analyses, and c) nonlinear
deformation analyses. Simplified analyses include the use of empirically based
correlations, analytically-based correlations, and simplified methods for flow slide
analysis. These types of analyses may be used for preliminary assessments of low-hazard
and small dams or as points of comparison to check the results of more advanced analysis
methods. However, even for small dams, more advanced methods are often required to
provide the level of refinement necessary to adequately represent in an analysis key
aspects of dam seismic response (such as ground motion characteristics, embankment
zoning, and material behavior).

Sliding block analyses are based on the simplifying assumption that seismic shear
deformations of an embankment dam result from displacement of a slide mass along a
discrete failure surface when the inertia forces on the slide mass exceed the resistance
along the failure surface. Sliding block analyses are useful in cases where the
mechanisms of embankment deformation and of material behavior may be approximated
by those assumed in the sliding block method. However, these types of analyses should
not be used for cases where loss of strength or stiffness may be widespread and lead to
spatially distributed deformations throughout extended zones of the embankment or
foundation. In such cases, nonlinear deformation analyses are likely to be required to
adequately simulate the overall response and mechanism of dam seismic deformation.

The guidelines include brief discussions of methods for analyses of post-earthquake

stability and of run-out deformations, for cases where gross instability might be expected
due to potential strength loss in the dam and/or foundation materials. However, detailed
descriptions of methods for analysis of liquefaction triggering or of cyclic softening,
which could lead to severe strength loss, are outside the scope of the document.

Nonlinear deformation analyses, NDAs, are commonly performed using finite element or
finite difference methods with constitutive models that simulate the nonlinear stress-
strain behavior of the dam and foundation materials. These types of analyses typically
require more information and effort than other methods, but they can provide
significantly greater insight into the anticipated response of a dam. NDAs are valuable
when the uncertainty in estimated performance resulting from the use of simpler analyses
may sway the outcome of a dam safety evaluation or when a better understanding of
seismic deformation patterns could improve the evaluation of dam performance.

57
NDAs are classed into gravity turn-on analyses, decoupled analyses, and fully or partially
coupled effective stress analyses. Nonlinear constitutive models are grouped into linear
elastic-plastic models (specifically the Mohr-Coulomb model), extended Mohr-Coulomb
models, and coupled effective-stress plasticity models, depending on their formulation of
soil stress-strain and pore-pressure response. Selected constitutive models are described
in the guidelines to provide the reader with a basic understanding of key features of the
models and their ability to simulate various aspects of soil stress-strain behavior. Proper
validation of analysis methods and of constitutive models is a critical step for the reliable
evaluation of dam seismic deformations, particularly when using advanced nonlinear
analysis models.

A phased analysis approach that begins with simple evaluations and adds sophistication
as needed is usually considered an effective approach to seismic deformation analysis of
embankment dams. The final level of analysis sophistication required for a dam depends
on the project importance, the analysis purpose, the results of previous analysis phases,
and the complexity of expected seismic response and material behavior in the
embankment and foundation under earthquake shaking. These factors dictate the degree
of refinement that may be necessary in simulating the response and mechanism of dam
deformation, and the degree of uncertainty that may be accepted, for evaluating the
seismic performance of a dam. A phased approach often allows efficient staging of
seismic deformation analyses of increasing complexity and of the corresponding
characterization of the embankment and foundation materials, and of progress of a dam
safety evaluation commensurate with available funding and resources.

However, for some project applications simplified analysis methods may not be
applicable or appropriate to begin with, or the need for advanced analyses may be
otherwise identified early in the project. In such cases, a phased approach will likely not
be applicable to the analysis of dam seismic deformations, although a phased program
may nonetheless be desirable for the process of dam investigations, geotechnical
characterization, and modeling.

Besides uncertainty in the earthquake ground motions, the largest source of uncertainty in
estimating the seismic performance of an embankment dam lies in the characterization of
the embankment and foundation materials. Thus, generally the most important aspect of a
seismic deformation analysis is proper geotechnical characterization of the materials and
of their likely seismic behavior. Such characterization must be substantiated by adequate
information regarding the zoning of the embankment and foundation and the engineering
properties of the materials. The degree of refinement used for characterization of the
materials should be consistent with the level of analysis sophistication.

The importance of adequate documentation of all aspects of a seismic deformation

analysis, ranging from the characterization of the dam and foundation materials to the
analysis approach and to the analysis results, cannot be overemphasized. In general, the
level of documentation should be consistent with the level of analysis sophistication and
the corresponding refinement in material characterization.

58
Proper documentation of a dam seismic deformation analysis is necessary to allow
adequate interpretation of the analysis results and sound assessment of dam performance
and safety. In addition to thoughtful interpretation of the analysis results, the assessment
of dam performance requires careful consideration of uncertainty associated with all
aspects of the analysis, including uncertainty in geotechnical characterization and
analytical model uncertainty. In consideration of the latter and whenever feasible, it is
generally considered good practice to perform parametric analyses and to use more than
one analysis method or model at the various stages of a phased analysis approach of dam
seismic deformations.

Case histories of back analysis of dam performance during past earthquakes (e.g. Mejia et
al., 1992; Boulanger et al., 1995; Zeghal and Abdel-Ghafar, 1992; Naesgard and Byrne,
2007; Beaty and Byrne, 2008; Mejia and Dawson, 2012; Hadidi et al., 2014; Chowdhury,
2019; and Boulanger, 2019) provide a rough measure of overall uncertainty in estimated
embankment seismic response and deformations under circumstances of dam and
foundation characterization typical, or perhaps at the higher end, of the range encountered
in U.S. practice. Such case histories suggest that overall uncertainty in the simulated
seismic response and deformations of embankment dams in practice is considerable. For
example, calculated permanent seismic displacements at the dam crest, from available
case histories, are commonly a factor of two or three off from the observed
displacements. Accordingly, it is critically important to consider such uncertainty in the
interpretation of the results of seismic deformation analyses of embankment dams and in
the evaluation of their seismic safety.

Because methods of analysis of dam seismic deformations will continue to evolve with
time, possibly involving major changes and improvements, it is anticipated that this
guidance document will be reviewed and updated periodically in the future. A preview of
possible upcoming changes in analysis methods may be drawn from the brief discussion
of new analysis techniques provided in Chapter 6 of the guidelines. Those techniques are
still under research but stand to represent major future advances in engineering practice
for the analysis of embankment dam seismic deformations.

10.0 References
Al Atik, L. and Sitar, N. (2010). “Seismic earth pressures on cantilever retaining
structures.” Journal of Geotechnical and Geoenvironmental Engineering, 136(10), pp.
1324-1333.

Ambraseys, N.N. (1960a). “The seismic stability of earth dams.” Proceedings of 2nd
world conference on earthquake engineering, 2. pp. 1345-1363.

Ambraseys, N.N. (1960b). “On the seismic behavior of earth dams.” Proceedings of 2nd
world conference on earthquake engineering, 1. pp. 331-356

Ambraseys, N.N. (1960c). “On the shear response of a two-dimensional truncated wedge
subjected to an arbitrary disturbance.” Bull Seismol Soc Am 50, pp. 45-46.

59
Ambraseys, N.N. and Menu, J.M. (1988). “Earthquake-induced ground displacements.”
J. Earthquake Engineering and Structural Dynamics, 16(7), pp. 985-1006.
https://doi.org/10.1002/eqe.4290160704.

Ambraseys, N.N. and Srbulov, M. (1994). “Attenuation of earthquake-induced ground

displacements.” J. Earthquake Engineering and Structural Dynamics, 23(5), pp. 467-487.
https://doi.org/10.1002/eqe.4290230502.

Amirzehni, E. (2016). Seismic assessment of basement walls in British Columbia. PhD

Dissertation. University of British Columbia.

Armstrong, R.J., Boulanger, R.W., and Beaty, M.H. (2013). “Liquefaction effects on
piled bridge abutments: centrifuge tests and numerical analyses.” Journal of
Geotechnical and Geoenvironmental Engineering, 139(3), pp. 433-443.

Armstrong, R.J., Park, D.S., and Gullen, A. (2018). “Cyclic soil behavior of common
constitutive models used in non-linear deformation analysis of embankment dams.”
Proceeding 2018 USSD Conference.

Arulmoli, K., Muraleetharan, K.K., Hossain, M.M., and Fruth, L.S. (1992). VELACS:
Verification of Liquefaction Analyses by Centrifuge Studies, Laboratory Testing
Program. Report No. 90–0562, The Earth Technology Corporation, Irvine, CA.

Ashford, S.A. and Sitar, N. (2002). “Simplified method for evaluating seismic stability of
steep slopes.” J. Geotech. Geoenviron. Eng., 128(2), pp. 119-128.

Aydin, A., Johnson, A. M., and Fleming, R. W. (1992). “Right-lateral-reverse surface

rupture along the San Andreas and Sargent faults associated with the October 17, 1989,
Loma Prieta, California, Earthquake.” Geology, v. 20, p. 1063-1067.

Aydan, Ö. (2016). “Some Considerations on a Large Landslide at the Left Bank of the
Aratozawa Dam Caused by the 2008 Iwate–Miyagi Intraplate Earthquake.” Rock Mech
Rock Eng 49, 2525–2539.

Babbitt, D.H. (2014). “Transverse cracking of embankment dams caused by earthquakes

revisited.” Proceedings of 34th USSD Conference, San Francisco, California, April.

Barrero, A.R., Taiebat, M., and Lizcano, A. (2015). “Application of an advanced

constitutive model in nonlinear dynamic analysis of tailings dam.” Proc., Sixty Eighth
Canadian Geotechnical Conference, Quebec, Canada.

Beaty, M. (2009). Analysis for Post-Earthquake Deformation, Presentation, Davis,

California.

Beaty, M.H. and Byrne, P.M. (1998). “An effective stress model for predicting
liquefaction behaviour of sand.” In P. Dakoulas, M. Yegian, and R. D. Holtz (Eds.),
Geotechnical Earthquake Engineering and Soil Dynamics III, ASCE Geotechnical

60
Special Publication No. 75, Vol. 1, Proceedings of a Specialty Conference, pp. 766-777.
Seattle: ASCE.

Beaty, M.H., and Byrne, P.M. (2000). “A Synthesized Approach for Predicting
Liquefaction and Resulting Displacements.” In Proc. of the 12th World Conference on
Earthquake Engineering, Auckland, New Zealand, January 30-February 4. (Paper No.
1589).

Beaty, M.H., and Byrne, P.M. (2001). “Observations on the San Fernando Dams.” In
Proc., 4th International Conference on Recent Advances in Geotechnical Earthquake
Engineering and Soil Dynamics, San Diego, California, March 26-31. University of
Missouri-Rolla.

Beaty, M. and Byrne P. (2008). “Liquefaction and Deformation Analyses Using a Total
Stress Approach.” J. Geotechnical and Geoenvironmental Eng., ASCE 134(8),
1059-1072.

Beaty, M.H., Ruthford, M., and Yule, D.E. (2010). “Seismic Evaluation of Semi-
Embedded Outlet Tower.” In Proc. 30th USSD Annual Meeting and Conference, United
States Society on Dams, Sacramento, April 12-16, pp. 265-280.

Beaty, M.H. and Byrne, P.M. (2011). “UBCSAND constitutive model: Version 904aR.”
Documentation report on Itasca UDM Web Site, February.

Beaty, M., Schlechter, S., Greenfield, M., Bock, J., Dickenson, S., Kempner Jr., L., and
Cook, K. (2014). “Seismic Evaluation of Transmission Tower Foundations at River
Crossings in the Portland-Columbia River Region.” In Proc. 10th National Conference in
Earthquake Engineering, EERI, Anchorage, AK, DOI: 10.4231/D3NP1WK0P.

Beaty, M. and Dickenson, S. (2015). “Numerical Analysis for Seismically-Induced

Deformations in Strain-Softening Plastic Soils.” 6th International Conference on
Earthquake Geotechnical Engineering, Christchurch, New Zealand.

Beikae, M., Luebbers, M., Barneich, J., and Osmond, D. (1996). “Seismic Deformation
Analyses of Eastside Reservoir Dams.” Sixteenth Annual USCOLD Lecture Series, Los
Angeles, California, July 22-26.

Boulanger, R.W., Bray, J.D., Merry, S.M., and Mejia, L.H. (1995). “Three-dimensional
dynamic response analyses of Cogswell dam.” Canadian Geotechnical Journal, 32(3),
pp. 452-464.

Boulanger, R.W. (2010). A sand plasticity model for earthquake engineering

applications. Report No. UCD/CGM-10-01, Center for Geotechnical Modeling,
Department of Civil and Environmental Engineering, University of California, Davis,
CA, 77 pp.

Boulanger, R.W. and Ziotopoulou, K. (2015). PM4Sand (Version 3): A sand plasticity
model for earthquake engineering applications. Report No. UCD/CGM-15/01, Center for

61
Geotech. Modeling, Department of Civil and Environ. Eng., University of California,
Davis, CA, 112 pp.

Boulanger, R. W., Montgomery, J., and Ziotopoulou, K. (2015). "Ch. 10: Nonlinear
deformation analyses of liquefaction effects on embankment dams." In: Ansal, A., Sakr,
M. (Eds.), Perspectives on Earthquake Geotechnical Engineering. Springer Nature
Publishing, Switzerland, doi: 10.1007/978-3-319-10786-8_10

Boulanger, R. and Beaty, M. (2016). “Seismic Deformation Analyses of Embankment

Dams: A Reviewer’s Checklist.” In Proc., USSD 2016 Annual Conference, United States
Society on Dams, Denver, CO, pp. 535-546.

Boulanger, R.W. and Ziotopoulou, K. (2017). PM4Sand (Version 3.1): A sand plasticity
model for earthquake engineering applications. Report No. UCD/CGM-17/01, Center for
Geotech. Modeling, Department of Civil and Environ. Eng., University of California,
Davis, CA, 112 pp.

Boulanger, R.W., and Ziotopoulou, K. (2018a). PM4Silt (Version 1): A silt plasticity
model for earthquake engineering applications. Report No. UCD/CGM-18/01, Center for
Geotechnical Modeling, Department of Civil and Environmental Engineering, University
of California, Davis, CA.

Boulanger, R. and Ziotopoulou, K. (2018b). “On NDA practices for evaluating

liquefaction effects.” In Proc., GEESD-V Conference, Austin, TX.

Boulanger, R.W. (2019). “Nonlinear Dynamic Analyses of Austrian Dam in the 1989
Loma Prieta Earthquake.” Journal of Geotechnical and Geoenvironmental Engineering,
ASCE, 145(11): 05019011, 10.1061/(ASCE)GT.1943-5606.0002156.

Boulanger, R.W., Munter, S.K., Krage, C.P., and DeJong, J.T. (2019). “Liquefaction
evaluation of interbedded soil deposit: Çark Canal in 1999 M7.5 Kocaeli Earthquake.”
Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 145(9): 05019007,
/10.1061/(ASCE)GT.1943-5606.0002089.

Bozorgnia, Y. and Bertero, V.V. (2004). Earthquake engineering: from engineering

seismology to performance-based engineering. CRC Press, 1150 pages.

Bray, J.D. and Travasarou, T. (2007). “Simplified procedure for estimating earthquake-
induced deviatoric slope displacements.” J. Geotechnical and Geoenvironmental Eng.,
ASCE 133(4), pp. 381-392.

Bray, J.D. and Luque, R. (2017). “Seismic performance of a building affected by

moderate liquefaction during the Christchurch earthquake.” Soil Dynamics and
Earthquake Engineering, 102, pp. 99-111.

Bray, J.D., Macedo, J., and Travasarou, T. (2018). “Simplified procedure for estimating
seismic slope displacements for subduction zone earthquakes.” J. Geotechnical and

62
Geoenvironmental Eng., ASCE 144(3):04017124.
https://doi.org/10.1061/(ASCE)GT.1943-5606.0001833.

Bray, J.D. and Macedo, J. (2019). “Procedure for estimating shear-induced seismic slope
displacement for shallow crustal earthquakes.” J. Geotechnical and Geoenvironmental
Eng., ASCE 145(12):04019106. https://doi.org/10.1061/(ASCE)GT.1943-5606.0002143.

Bureau, G., Rettberg, W. A. and Eymann, J. (2008). “Influence of vertical shaking on

embankment dam seismic response.” Proc., United States Society on Dams Annual
Conference, Portland, OR.

Byrne, P. M. (1991). “A Cyclic Shear-Volume Coupling and Pore-Pressure Model for

Sand.” Proceedings of Second International Conference on Recent Advances in
Geotechnical Earthquake Engineering and Soil Dynamics, Paper No. 1.24, 47-55.

Byrne, P.M., Park, S.S., Beaty, M., Sharp, M.K., Gonzalez, L., and Abdoun, T. (2004).
“Numerical modeling of liquefaction and comparison with centrifuge tests.” Canadian
Geotechnical Journal, 41(2), pp. 193-211.

Castro, G., Seed, R.B., Keller, T.O. and Seed, H.B. (1992). “Steady-state strength
analysis of Lower San Fernando Dam Slide.” Journal of Geotechnical Engineering,
ASCE, 118(3), pp. 406-427.

Cheng, Z. (2018). Manual for Dafalias-Manzari 04 – V3. Available from Itasca UDM
website: https://www.itascacg.com/udms/sanisand-dafalias-manzari-2004

Cheng, Z., Dafalias, Y.F., and Manzari, M.T. (2013). “Application of SANISAND
Dafalias-Manzari model in FLAC3D. In Continuum and Distinct Element Numerical
Modeling in Geomechanics.” Proc., 3rd International FLAC/DEM Symposium,
Hangzhou, China, October 22-24.

Chiang, D.Y. and Beck, J.L. (1994). “A new class of DEM for cyclic plasticity – I.
Theory and application.” Int. J. Solids Structures 31(4), pp. 485-496.

Chopra, A.K. (1966). Earthquake Effects on Dams. Dissertation, University of California,

Berkeley.

Chou, J-C, Kutter, B.L., and Travasarou, T. (2010). “Numerical Analyses of Centrifuge
Models of the BART Transbay Tube.” Proc., Recent Advances in Geot. Earthq. Eng. and
Soil Dynamics, Paper No. 8.08.

Chowdhury, K. H. (2019). Evaluation of the State of Practice Regarding Nonlinear

Seismic Deformation Analyses of Embankment Dams Subject to Soil Liquefaction Based
on Case Histories. Dissertation, University of California, Berkeley.

Clough, R.W. and Chopra, A.K. (1966) “Earthquake stress analysis in earth dams.”
Journal of the Engineering Mechanics Division, ASCE, 92(2), 197-212.

63
Cundall, P.A. and Strack, O.D.L. (1979). “A discrete numerical model for granular
assemblies.” Géotechnique, 29(1), pp. 47-65. doi:10.1680/geot.1979.29.1.47

Cundall, P.A. and Board, M. (1988). “A Micro-Computer Program for Modeling Large-
Strain Plasticity Problems.” Proceedings, 6th International Conference on Numerical
Methods in Geomechanics, Innsbruck Austria, April, pp. 2101- 2108.

Dafalias, Y.F. and Manzari, M. (2004). “Simple plasticity sand model accounting for
fabric change effects.” J. of Engineering Mechanics, ASCE, 130(6), pp. 622-634.

Dafalias, Y.F., Papadimitriou, A.G., and Li, X.S. 2004. “Sand plasticity model
accounting for inherent fabric anisotropy.” Journal of Engineering Mechanics, ASCE,
130(11), pp. 1319-1333.

Davis, C.A. and Bardet, J.P. (1996). “Performance of Two Reservoirs During 1994
Northridge Earthquake.” J. Geotechnical Engineering, ASCE, 122(8), August, pp. 613-
622.

Dawson, E.M., Roth, W.H., Nesarajah, S., Bureau, G., and Davis, C.A. (2001). “A
practice-oriented pore pressure generation model.” In Proceedings, 2nd FLAC
Symposium on Numerical Modeling in Geomechanics, October 29-31, Lyon, France.

Dawson, E.M. and Mejia, L.H. (2010). “3D Analysis of the Response of Seven Oaks
Dam.” Fifth Internal Conference on Recent Advances in Geotechnical Earthquake
Engineering and Soil Dynamics. San Diego, CA, May 24-29.

Dawson, E.M. and Mejia, L.H. (2012). “Updates to a practice-oriented pore pressure
generation model.” GeoCongress 2012, Geotechnical Special Publication No. 225. R.D.
Hryciw et al. eds. Oakland, CA, March 25-29, Geo Institute, ASCE.

Dawson, E.M. and Mejia, L.H. (2020). “Three-dimensional analysis of Lenihan Dam for
the Loma Prieta earthquake.” Proceedings of 40th Annual USSD Conference and
Exhibitionhttp://ussd.conferencespot.org/2020/activity

Deng, Y.H., Dashti, S., Hushmand, A., Davis, C., and Hushmand, B. (2016). “Seismic
response of underground reservoir structures in sand: evaluation of class-c and c1
numerical simulations using centrifuge experiments.” Soil Dynamics and Earthquake
Engineering, 85, pp. 202-216.

Dobry, R. and Taboada, V.M. (1994). “Possible lessons from VELACS model No. 2
results.” Proc. Intl. Conf. on Verification of Numerical Procedures for the Analysis of
Soil Liquefaction Problems, K. Arulanandan and R. F. Scott, eds., Balkema, Rotterdam,
The Netherlands, 2, pp. 1341-1352.

Dobry, R., Taboada, V., and Liu, L. (1995). “Centrifuge modeling of liquefaction effects
during earthquakes.” Keynote Lecture, Proc., 1st Int. Conf. on Earthquake Geotechnical
Engineering (IS-Tokyo), K. Ishihara, ed., Balkema, Rotterdam, The Netherlands, 3, pp.
1291-1324.

64
Drnevich, V.P. (1967). Effect of strain on the dynamic properties of sand. Ph.D. thesis,
University of Michigan, Ann Arbor, Michigan.

Duku, P.M., Stewart, J.P., Whang, D.H., and Yee, E. (2008). “Volumetric strains of clean
sands subject to cyclic loads.” J. Geotechnical Engineering, ASCE, 134(8), pp. 1073-
1085.

Duncan, J.M., Wright, S.G., and Brandon. T.L. (2014). Soil Strength and Slope Stability.
Second Edition. John Wiley & Sons Inc. ISBN-13: 9781118651650. 336 pp.

El Ghoraiby, M.A., and Manzari, M.T. (2019). "LEAP-2020 - Stress-strain response of

Ottawa F65 sand in Cyclic Simple Shear." Report, George Washington University,
Washington, D.C.

El Shamy, U. and Abdelhamid, Y. (2014). “Modeling granular soils liquefaction using

coupled lattice Boltzmann method and discrete element method.” Soil Dynamics and
Earthquake Engineering, 67, pp. 119-132.

Electric Power Research Institute (EPRI). (1990). Manual on estimating soil properties
for foundation design, Palo Alto, CA, Electric Power Research Institute, Vol. 1, EPRI
TR-102293.

EPRI. (1993). Guidelines for determining design basis ground motions, Palo Alto, CA,
Electric Power Research Institute, Vol. 1, EPRI TR-102293.

Elgamal, A.W. (1992). “Three-dimension seismic analysis of La Villita Dam.” J.

Geotechnical Engineering, ASCE, 118(12), pp. 1937-1958.

Elgamal, A., Yang, Z., Parra, E., and Ragheb, A. (2003). “Modeling of cyclic mobility in
saturated cohesionless soils.” International Journal of Plasticity, 19(6), pp. 883-905.

Elgamal, A., Lu, J., and Forcellini, D. (2009). “Mitigation of liquefaction-induced lateral
deformation in a sloping stratum: Three-dimensional numerical simulation.” Journal of
Geotechnical and Geoenvironmental Engineering, ASCE, 135(11), pp. 1672-1682.

Federal Energy Regulatory Commission (FERC). (2018). Engineering guidelines for the
evaluation of hydropower projects. Chapter 13 – Evaluation of earthquake ground
motions. Washington, D.C.

Federal Highway Administration (FHWA). 2011. LRFD Seismic Analysis and Design of
Transportation Geotechnical Features and Structural Foundations Reference Manual.
Publication No. FHWA-NHI-11-032, GEC No. 3, August 2011 (Rev. 1). NHI Course No.
130094. U.S. Department of Transportation, Federal Highway Administration.

Fell, R., Macgregor, J.P., Stapledon, D.H., Bell, G., and Foster, M. (2015). Geotechnical
Engineering of Dams. Second Edition. Taylor Francis, London.

65
Finn, W.D.L. (1993). “Practical studies of the seismic response of a rockfill dam and a
tailings impoundment.” Proceedings, Third International Conference on Case Histories in
Geotechnical Engineering, June 1-4, pp. 1347-1360.

Finn, W.D.L., Pickering, D.J., and Bransby, P.L. (1971). “Sand liquefaction in triaxial
and simple shear tests.” Journal of the Soil Mechanics and Foundation Division, 97(4),
pp. 639-659.

Fong, F.C. and Bennett, W.J. (1995). “Traverse cracking on embankment dams due to
earthquake.” Proceedings of 1995 ASDSO Western Regional Conference.

France, J.W., Adams, T., Wilson, J., and Gillette, D. (2000). “Dynamic deformation
analysis and three-dimensional post-earthquake stability analysis for Casitas Dam,
California.” Soil Dynamics and Liquefaction 2000, pp. 123-147.

Franklin, A.G. and Chang, F.K. (1977). Earthquake resistance of earth and rock-fill
dams, Report 5: Permanent displacements of earth embankments by Newmark sliding
block analysis. Miscellaneous Paper S-71-17, Soils and Pavement Laboratory, U.S. Army
Engineer Waterways Experiment Station, Vicksburg, Miss.

Friesen, S., Balakrishnan, A., Driller, M., Beaty, M., Arulnathan, R., Newman, E., and
Murugaiah, S. (2014). “Lessons learned from FLAC analyses of seismic remediation of
Perris Dam.” Proc., United States Society on Dams Annual Conference, San Francisco,
CA, 39-55.

GeoStudio. (2014). Dynamic Modeling with QUAKE/W. GEOSLOPE, Calgary, AB,

Canada.

Goodman, R.E. and Seed, H.B. (1965). Displacements of slopes in cohesionless materials
during earthquakes. Rep. No. H21, Inst. of Trans. and Traffic Engineering. Univ. of
Calif., Berkeley, California.

Goodman, R.E., Seed, H.B, (1966). Earthquake-induced displacements in sand

embankments. Journal of the Soil Mechanics and Foundations Division 92, 125-146.

Griffiths, D.V. and Prevost, D.V. (1988) “Two- and three-dimension dynamic finite
element analyses of Long Valley Dam.” Geotechnique, 38(3), 367-388.

Guo, N. and Zhao, J. (2014). “A coupled FEM/DEM approach for hierarchical multiscale
modelling of granular media.” International Journal for Numerical Methods in
Engineering, 99(11), pp. 789-818.

Hadidi, R., Moriwaki, Y., Barneich, J., Kirby, R., and Mooers, M. (2014). “Seismic
Deformation Evaluation of Lenihan Dam Under 1989 Loma Prieta Earthquake.” In Proc.
10th National Conference in Earthquake Engineering, EERI, Anchorage, AK.

66
Harder, L. F. Jr. and Boulanger, R. W. (1997). "Application of Kσ and Kα Correction
Factors." Proc., NCEER Workshop on Evaluation of Liquefaction Resistance of Soils,
Nat. Ctr. For Earthquake Engrg. Res., State Univ. of New York at Buffalo, 195-218.

Harder, L.F., Bray, J.D., Volpe, R.L. and Rodda, K.V. (1998). “Performance of Earth
Dams During the Loma Prieta Earthquake.” Performance of the built environment (T.L.
Holzer, ed.), U.S. Geological Survey Professional Paper 1552D.

Harder, L.F., Kelson, K.I., Kishida, T., and Kayen, R. (2011). Preliminary Observations
of the Fujinuma Dam Failure Following the March 11, 2011 Tohoku Offshore
Earthquake, Japan. GEER Association Report No. GEER-25e – Preliminary, June 6.

Hu, Y., and Randolph, M. F. (1998). “A practical numerical approach for large
deformation problems in soil.” International Journal for Numerical and Analytical
Methods in Geomechanics, 22(5), 327-350.

Hudson, M., Idriss, I., Beikae, M. (1994). User's Manual for QUAD4M: A Computer
Program to Evaluate the Seismic Response of Soil Structures Using Finite Element
Procedures and Incorporating a Compliant Base. University of California, Davis, U.S.A.

Hungr, O. (1995). “A Model for the Runout Analysis of Rapid Flow Slides, Debris
Flows, and Avalanches.” Can. Geotech. J. 32, pp. 610-623.

Hunt, D.B., Beutel, J., Weber, C. (2016). “Developing time histories with acceptable
record parameters for Dillon Dam.” Proceedings, United States Society on Dams 2016
Annual Meeting and Conference, Denver, Colorado, April.

Hutabarat, D., and Bray, J.D. (2019). “Effective stress analysis of liquefiable site in
Christchurch to discern the characteristics of sediment ejecta.” Earthquake Geotechnical
Engineering for Protection and Development of Environment and Constructions, Silvestri
and Moraci (eds), Associazione Geotecnica Italiana, Rome, Italy, 2923-2931.

Hynes-Griffin, M.E. and Franklin, A.G. (1984). Rationalizing the seismic coefficient
method. Miscellaneous Paper No. GL-84-13, Geotechnical Laboratory, U.S. Army
Engineer Waterways Experiment Station, Vicksburg, Miss.

Idriss, I.M. and Seed, H.B. (1968). “Seismic response of horizontal soil layers.” J. Soil
Mechanics and Foundations Div. ASCE, 94(4), pp. 1030-1031.

Idriss, I.M., Lysmer, J., and Seed, H.B. (1973). QUAD-4: a computer program for
evaluating the seismic response of soil structure by variable damping finite element
procedures. Report No. EERC 73-15. University of California at Berkeley.

Idriss, I.M. and Duncan, J.M. (1988). “Earthquake response analysis of embankment
dams.” Advanced Dam Engineering for Design, Construction, and Rehabilitation,
Springer, pp 239-255.

67
Idriss, I.M. and Boulanger, R.W. (2003). “Estimating Kα resistance of sloping ground.”
Proc., Eighth U.S.‐Japan Workshop on Earthquake Resistant Design of Lifeline Facilities
and Countermeasures Against Liquefaction, Technical Report MCEER‐03‐0003,
Multidisciplinary Center for Earthquake Engineering Research.

Idriss, I.M. and Boulanger, R.W. (2007). “Residual shear strength of liquefied soils.” Pp.
621–634 in Proceedings of the Modernization and Optimization of Existing Dams and
Reservoirs: 27th Annual USSD Conference, Philadelphia, Pennsylvania, March 5-9.
Denver, CO: United States Society on Dams.

Idriss, I.M. and Boulanger, R.W. (2008). Soil Liquefaction During Earthquakes,
Earthquake Engineering Research Institute, MNO-12.

International Commission on Large Dams (ICOLD). (2016). Selecting Seismic

Parameters for Large Dams – Guidelines. ICOLD Bulletin 148.

Ishihara, K. (1984). “Post-earthquake failure of a tailings dam due to liquefaction of the

pond deposit.” In Proc., Inter. Conf. on Case Histories in Geotechnical Engineering,
Rolla, Missouri, May 6-11, Vol. 3, pp. 1129-1143.

Ishihara, K., and Yoshimine, M. (1992), “Evaluation of Settlements in Sand Deposits

Following Liquefaction during Earthquakes,” Soils and Foundations, Vol. 32, No. 1, pp.
173-188.

Itasca Consulting Group, Inc. (2008). FLAC – Fast Lagrangian Analysis of Continua,
Ver. 6.0 User’s Manual. Minneapolis: Itasca.

Itasca Consulting Group, Inc. (2009). FLAC3D – Fast Lagrangian Analysis of Continua
in 3 Dimensions, Ver. 4.0 User’s Manual. Minneapolis: Itasca.

Itasca Consulting Group, Inc. (2011). FLAC, Fast Lagrangian Analysis of Continua,
User’s Guide, Version 7.0, Itasca Consulting Group, Inc., Minneapolis, MN.

Itasca Consulting Group, Inc. (2016). FLAC, Fast Lagrangian Analysis of Continua,
User’s Guide, Version 8.0. Itasca Consulting Group, Inc., Minneapolis, MN.

Jansen, R.B. (1990). “Estimation of embankment dam settlement caused by earthquake.”

International Water Power & Dam Construction, December, pp. 35-40.

Jassim, I., Coetzee, C., and Vermeer, P.A. (2013). “A dynamic material point method for
geomechanics.” In Proceedings of the International Conference on Installation Effects in
Geotechnical Engineering (ICIEGE), pp. 15-23.

Japan Committee on Large Dams (JCOLD) (2012), “Review of the cause of Fujinuma-
Ike failure - Summary report - January 25, 2012”,
http://www.jcold.or.jp/e/activity/FujinumaSummary-rev.120228.pdf

68
Jibson, R.W. (2007). “Regression models for estimating coseismic landslide
displacement.” Engineering Geology, 91(2-4), pp. 209-218.
https://doi.org/10.1016/j.enggeo.2007.01.013.

Jibson, R.W., Rathje, E.M., Jibson, M.W., and Lee, Y.W. (2013). SLAMMER—Seismic
Landslide Movement Modeled using Earthquake Records (ver.1.1, November 2014): U.S.
Geological Survey Techniques and Methods, book 12, chap. B1, unpaged.

Kamai, R., and Boulanger, R.W. (2013). “Simulations of a centrifuge test with lateral
spreading and void redistribution effects.” Journal of Geotechnical and
Geoenvironmental Engineering, ASCE, 139(8), pp. 1250-1261.

Kawai, T. (1985). Summary Report on the Development of the Computer Program,

DIANA - Dynamic Interaction Approach and Non-Linear Analysis, Science University of
Tokyo.

Keaton, J.R., Wartman, J., Anderson, S., Benoit, J., deLaChapelle, J., Gilbert, R., and
Montgomery, D.R. (2014). The 22 March 2014 Oso Landslide, Snohomish County,
Washington. GEER Association Report No. GEER-036, July 22.

Khosravifar, A., Elgamal, A., Lu, J., and Li, J. (2018). "A 3D model for earthquake-
induced liquefaction triggering and post-liquefaction response." Soil Dynamics and
Earthquake Engineering, 110, pp. 43-52.

Kiernan, M. and Montgomery, J. (2020). “Numerical Simulations of the Fourth Avenue

Landslide Considering Strain Softening.” Journal of Geotechnical and Geoenvironmental
Engineering. 146(10) doi: 10.1061/(ASCE)GT.1943-5606.0002349

Kokusho, T. 2007. Liquefaction strengths of poorly-graded and well-graded granular

soils investigated by lab tests. Pp. 159–184 in 4th International Conference on
Earthquake Geotechnical Engineering, Thessaloniki, Greece.

Kramer, S.L. (1996). Geotechnical Earthquake Engineering, Prentice Hall, Inc., Upper
Saddle River, New Jersey, 653 pp.

Kramer, S.L. and Smith, M.W. (1997). “Modified Newmark model for seismic
displacement of compliant slopes.” Journal of Geotechnical and Geoenvironmental
Engineering, 123(7), 635-644.

Kramer, S.L. and Wang, C.H. (2015). “Empirical model for estimation of the residual
strength of liquefied soil.” Journal of Geotechnical and Geoenvironmental Engineering,
141(9), 04015038.

Kuhlemeyer, R.L. and Lysmer, J. (1973). “Finite Element Method Accuracy for Wave
Propagation Problems.” J. Soil Mech. & Foundations, Div. ASCE, 99(SM5), pp. 421-427,
May.

69
Kutter, B.L. et al. (2017). “LEAP-GWU-2015 Experiment Specifications, Results, and
Comparisons.” Journal of Soil Dynamics and Earthquake Engineering, June.

Kwok, A., Stewart, J., Hashash, Y., Matasovic, N., Pyke, R., Wang, Z., and Yang, Z.
(2007). “Use of Exact Solutions of Wave Propagation Problems to Guide Implementation
of Nonlinear Seismic Ground Response Analysis Procedures.” J. Geotechnical and
Geoenvironmental Eng., ASCE 133(11), pp. 1385-1398.

Lai, S.S. and Seed, H.B. (1985). Dynamic Response of Long Valley Dam in the Mono
Lake Earthquake Series of May 25-27, 1980, EERC 75-30, University of California,
Berkeley.

Lasley, S. J., Green, R. A., Chen, Q., & Rodriguez-Marek, A. (2016). Approach for
estimating seismic compression using site response analyses. Journal of Geotechnical
and Geoenvironmental Engineering, 142(6), 04016015.

Lees, A. (2016). Geotechnical finite element analysis, a practical guide. ICE Publishing.
One Great George Street, Westminster, London WXP1 3AA.

Li, X.S., Shen, C.K., and Wang, Z.L. (1998). “Fully coupled inelastic site response
analysis for 1986 Lotung earthquake.” J. Geotech. Geoenvironmental Engineering,
124(7), pp. 560-573.

Li, X.S. and Dafalias, Y.F. (2012). “Anisotropic critical state theory: role of fabric.”
Journal of Engineering Mechanics, 138(3), pp. 263-275.

Lin, J.-S., and Whitman, R.V. (1983). ‘‘Decoupling approximation to the evaluation of
earthquake-induced plastic slip in earth dams.’’ Earthquake Engrg. and Struct. Dyn., 11,
pp. 667-678.

Livermore Software Technology Corporation (LSTC). (2015). LS-DYNA Release 8.0.

Livermore Software Technology Corporation (LSTC), Livermore, CA.

Lode, W. (1926). “Versuche iiber den Einfluss der mittleren Hauptspannung auf das
Fliessen der Metalle Eisen, Kupfer, und Nickel.” Z. Physik, 36, pp. 913-939.

Lucia, P., Duncan, J.M., and Seed, H.B. (1981). “Summary of Research on Case
Histories of Flow Failures of Mine Tailings Impoundments.” Mine Waste Disposal
Technology, Proceedings Bureau of Mines Technology Transfer Workshop, Denver,
July.

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
75-30, Earthquake Engineering Research Center, University of California, Berkeley.

Ma, F.G., Wang, Z.-L., and LaVassar, J. (2006). “Liquefaction analyses of a hydraulic fill
earthfill dam.” Varona & R. Hart (eds), FLAC and Numerical Modeling in

70
Geomechanics – 2006, Proceedings of the Fourth International FLAC Symposium,
Madrid, Spain, 29-31 May 2006, Paper #08-03. Minneapolis: Itasca.

Ma, F.G. and Wang, Z.L. (2008). “Implementation and three dimensional example
applications of a bounding surface hypo-plasticity model for sand as a C++ UDM for
FLAC3D.” 1st FLAC/DEM Symposium on Numerical Modeling. 25-27 August 2008,
Minneapolis, MN, USA.

Ma, F.G., French, J., Change, C.Y., Wang, Z.L., and Aw, E.S. (2016). “Racking study of
the Sierra and Lundy tunnel with surrounding soil layer liquefying under seismic loading
using FLAC3D.” Proc., Applied Numerical Modeling in Geomechanics –– Gomez,
Detournay, Hart & Nelson (eds), Paper 03-03.

Macedo, J., Zuta, J., and Aguilar, Z. (2015). “Seismic Analysis of an Instrumented Earth
Dam in Terms of the Variability of the Response Design Spectra.” In Proc., 6th
International Conference on Earthquake Geotechnical Engineering.

Makdisi, F.I. and Seed, H.B. (1977). A simplified procedure for estimating earthquake-
induced deformations in dams and embankments. Report No. UCB/EERC-77/19,
Earthquake Engineering Research Center, University of California, Berkeley.

Makdisi, F.I. and Seed, H.B. (1978). “Simplified procedure for estimating dam and
embankment earthquake-induced deformations.” J. Geotechnical Engineering, 104(7),
pp. 849-867.

Makdisi, F.I., Seed, H.B., and Idriss, I.M. (1978). “Analysis of Chabot Dam during the
1906 earthquake.” Earthquake Engineering and Soil Dynamics, Proceedings of ASCE
Geotechnical Division Specialty Conference, Vol. II, pp 569-587, Pasadena, CA.

Makdisi, F.I., Change, C.Y., Wang, Z.L., and Mok, C.M. (1991). Analysis of the
recorded response of Lexington Dam during various levels of ground shaking.
Proceedings of SMIP91 Seminar, California Geological Survey.
https://www.conservation.ca.gov/cgs/Pages/Program-SMIP/Seminar/smip91_toc.aspx

Makdisi, F.I., Wang, Z.L., and Edwards, W.D. (2000). “Seismic Stability of New
Exchequer Dam and Gated Spillway Structure.” Proc., Twentieth Annual USCOLD
Lecture Series, Seattle, Washington, July 10-14.

Manzari, M.T. and Dafalias, Y.F. (1997). “A two-surface critical plasticity model for
sand.” Geotechnique, 47(2), pp. 255-272.

Manzari, M.T. et al. (2017). “Liquefaction Experiment and Analysis Projects (LEAP):
Summary of Observations from the Planning Phase.” Journal of Soil Dynamics and
Earthquake Engineering, June.

Martin, G. R., Finn W. D. L. and Seed H. B. (1975) “Fundamentals of Liquefaction under

Cyclic Loading,” J. Geotech., Div. ASCE, 101(GT5), 423-438.

71
Martin, K. and Malvick, E.J. (2016). “Nonlinear modeling of sediment retention
structures for dam removal project.” Proc., United States Society on Dams Annual
Conference. Denver, CO.

Mayne, P.W., Christopher, B.R., and DeJong, J.T. (2001). Manual on Subsurface
Investigations. FHWA Publication No. FHWA-NHI-01-031, 394 pgs.

McKenna, F., Fenves, G.L., and Scott, M.H. (2000). “Open system for earthquake
engineering simulation.” University of California, Berkeley, CA.

Mejia, L.H. (2017). “Performance of Fena Dam during the 1993 Guam Earthquake.”
Proceedings, United States Society on Dams 2017 Annual Meeting and Conference,
Anaheim, California.

Mejia, L.H. and Seed, H.B. (1983). “Comparison of 2-D and 3-D dynamic analyses of
earth dams,” Journal of the Soil Mechanics and Foundation Division, ASCE, 109, GT11,
1383-1398.

Mejia, L.H., Sykora, D., Hynes M., Fung K., and Koester, J. (1991). “Measured and
calculated dynamic response of rock-fill dam.” Proceedings Second International
Conference on Geotechnical Earthquake Engineering and Soil Dynamics, University of
Missouri-Rolla, St. Louis, Missouri.

Mejia, L.H., Sun, J.I., Salah-mars, S., Moriwaki, Y., and Beikae, M. (1992). “Non-linear
dynamic response analysis of Lexington Dam.” Proceedings of SMIP92 Seminar,
California Geological Survey. https://www.conservation.ca.gov/cgs/Pages/Program-
SMIP/Seminar/smip92_toc.aspx

Mejia, L.H. and Boulanger, R.W. (1993). “Calibrated dynamic response analysis of
Stafford Dam.” Proceedings, Third International Conference on Case Histories in
Geotechnical Engineering, June 1-4, pp. 321-328.

Mejia, L., Gillon, M., Walker, J., and Newson, T. (2002). “Seismic Load Evaluation
Criteria for Dams of Two New Zealand Owners.” International Journal on Hydropower
and Dams, 9(4).

Mejia, L., Macfarlane, D., Read, S., and Walker, J. (2005). Seismic Criteria for Safety
Evaluation of Aviemore Dam. Proceedings, United States Society on Dams 2005 Annual
Meeting and Conference, Salt Lake City, Utah, June.

Mejia, L.H., and E. M. Dawson (2006). “Earthquake Deconvolution for FLAC” in FLAC
and Numerical Modeling in Geomechanics. Proceedings of the 4th International FLAC
Symposium, Madrid, Spain, May 2006, pp. 211-219. P. Varona & R. Hart, eds.
Minneapolis, Minnesota: Itasca Consulting Group Inc.

Mejia, L.H. and Dawson, E. (2010), “3D Analysis of the Seismic Response of Seven
Oaks Dam.” Proceedings of the 5th International Conference on Recent Advances in
Geotechnical Earthquake Engineering and Soil Dynamics, San Diego, CA, May.

72
Mejia, L.H. and Dawson, E. (2019). “Evaluation of earthquake-induced cracking of
embankment dams.” Proceedings, United States Society on Dams 2019 Annual Meeting
and Conference, Chicago, Illinois, April.

Ming, H.Y. and Li, X.S. (2003). “Fully coupled analysis of failure and remediation of
Lower San Fernando Dam.” J. Geotechnical and Geoenvironmental Eng., ASCE 129(4),
pp. 336-349.

Montgomery, J., Boulanger, R.W., Armstrong, R.J., and Malvick, E.J. (2014).
"Anisotropic undrained shear strength parameters for non-linear deformation analyses of
embankment dams." 2014 ASCE GeoCongress, Atlanta, GA.

Montgomery, J. and Abbaszadeh, S. (2017). “Comparison of two constitutive models for

simulating the effects of liquefaction on embankment dams.” Proc., United States Society
on Dams Annual Conference, Anaheim, CA.

Montgomery, J., Boulanger, R.W., and Ziotopoulou, K. (2017). “Effects of spatial

variability on the seismic response of the Wildlife Liquefaction Array.” Proc., 3rd
International Conference on Performance-based Design in Earthquake Geotechnical Eng.
(PBD-III), Vancouver, BC, Canada.

Moriwaki, Y., Beikae, M., and Idriss, I.M. (1988). “Nonlinear Seismic Analysis of the
Upper San Fernando Dam Under the 1971 San Fernando Earthquake.” Proceedings, 9th
World Conference on Earthquake Engineering, Tokyo and Kyoto, Japan, Vol. VIII, pp.
237- 241.

Moriwaki, Y., Idriss, I.M., Moses, T.L., and Ladd, R.S. (1989). “A re-evaluation of the L
Street slide in Anchorage during the 1964 Alaska earthquake.” Proc. 12th Int. Conf. Soil
Mech. And Found. Engrg., Vol 3, Balkema, The Netherlands, 1583-1586.

Mortezaie, A. R., & Vucetic, M. (2013). “Effect of frequency and vertical stress on cyclic
degradation and pore water pressure in clay in the NGI simple shear device.” Journal of
Geotechnical and Geoenvironmental Engineering, 139(10), pp. 1727-1737.

Naesgaard, E. (2011). A Hybrid Effective Stress – Total Stress Procedure for Analyzing
Soil Embankments Subjected to Potential Liquefaction and Flow. PhD Dissertation,
University of British Columbia.

Naesgaard, E., Byrne, P.M., Seid-Karbasi, S.M. (2006). “Modeling flow liquefaction and
pore water redistribution mechanisms.” Proceedings of the 8th U.S. National Conference
on Earthquake Engineering.

Naesgaard, E. and Byrne, P.M. (2007). “Flow liquefaction simulation using a combined
effective stress – total stress model.” 60th Canadian Geotechnical Conference, Canadian
Geotechnical Society, Ottawa, Ontario.

73
Naesgaard, E., Byrne, P.M., and Amini, A. (2015). “Hysteretic Model for Non-
liquefiable Soils (UBCHYST5d).” Documentation report on Itasca UDM Web Site,
November 2015.

National Academies of Sciences, Engineering, and Medicine (NASEM). (2016). State of

the Art and Practice in the Assessment of Earthquake-Induced Soil Liquefaction and Its
Consequences. Washington, DC: The National Academies Press.
https://doi.org/10.17226/23474.

National Oceanic and Atmospheric Administration (NOAA). (1973). San Fernando,

California, Earthquake of February 9, 1971, V. III Geological and Geophysical Studies.
Washington, D.C., U.S. Government Printing Office.

Newmark, N.M. (1965). “Effects of earthquakes on dams and embankments.”

Géotechnique, 15(2), pp. 139-160.

National Institute of Standards and Technology (NIST). (2011). Selecting and scaling
earthquake ground motions for performing response-history analyses. NIST GCR 11-
917-15.

Oaks, R. and Scott, B. (2014). “Seismic issues at Scoggins Dam.” Proc., United States
Society on Dams Annual Conference, San Francisco, CA, 57-77.

Olson, S.M., Stark, T.D., Walton, W.H., and Castro, G., (2000). "1907 Static
Liquefaction Flow Failure of the North Dike on Wachusett Dam", ASCE J. Geotechnical
and Geoenvironmental Eng, Vol. 126, No. 12, ASCE, December.

Olson, S.M., and Stark, T.D. (2002). “Liquefied strength ratio from liquefaction flow
failure case histories.” Canadian Geotechnical Journal 39, pp. 629-647.

Olson, S.M. and Johnson, C.I. (2008). “Analyzing liquefaction-induced lateral spreads
using strength ratios.” ASCE Journal of Geotechnical and Geoenvironmental
Engineering, 134(8), August.

Pace, T. G., Schaefer, J. A., O’Leary, T. M., and Rauch, A. F. (2008). “Simplified
estimation of seismic deformation for risk analysis.” Proc., 28th Annual USSD Conf,
United States Society on Dams, Portland, Oregon, April 28 - May 2, pp. 521-532.

Parra, E. (1996). Numerical modeling of liquefaction and lateral ground deformation

including cyclic mobility and dilation response in soil systems. Ph.D. thesis, Rensselaer
Polytechnic Institute, Troy, NY.

Pells, S. and Fell, R. (2002). “Damage and cracking of embankment dams by earthquake
and the implication for internal erosion and piping.” Proceedings 21st International
Congress on Large Dams, Montreal. ICOLD, Paris Q83-R17, International Commission
on Large Dams, Paris.

74
Perlea, V.G. and Beaty, M.H. (2010). “Corps of Engineers practice in the evaluation of
seismic deformation of embankment dams.” Proc., Fifth International Conference on
Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics. San
Diego, CA, paper SPL 6.

Phalen, J., Makdisi, F., Mejia, L., Mooers, M. (2014). “Risk of seismic deformation of a
1930’s embankment dam in a highly active seismic environment.” Proceedings, United
States Society on Dams 2014 Annual Meeting and Conference, San Francisco, CA, April.

Pillai, V.S. and Salgado, F.M. (1994). “Post-liquefaction stability and deformation
analysis of Duncan Dam.” Canadian Geotechnical Journal, 31(6), pp. 967-978.

Plaxis. (2016). Plaxis 2D and 3D. Plaxis bv., Delft, Netherlands.

Pradel, D., Wartman, J., and Tiwari, B. (2013). “Failure of the Fujinuma Dams during the
2011 Tohoku Earthquake.” Proceedings, 2013 ASCE GeoCongress, San Diego, CA.

Prevost, J.H. (1981). DYNAFLOW: A Nonlinear Transient Finite Element Analysis

Program, Princeton University, Department of Civil Engineering, Princeton, NJ.

Prevost, J.H. (1985). “A simple plasticity theory for frictional cohesionless soils.” Soil
Dynamics and Earthquake Engineering, 4(1), pp. 9-17.

Public Works Research Institute (PWRI). (2008). Report of Damage to Infrastructures

and Buildings by the 2008 Iwate-Miyagi Nairiku Earthquake, Technical Note of Public
Works Research Institute, No.4120, (In Japanese),
http://www.nilim.go.jp/lab/bcg/siryou/tnn/tnn0486.htm

Ramirez, J., Barrero, A. R., Chen, L., Dashti, S., Ghofrani, A., Taiebat, M., & Arduino, P.
(2018). Site response in a layered liquefiable deposit: evaluation of different numerical
tools and methodologies with centrifuge experimental results. Journal of Geotechnical
and Geoenvironmental Engineering, 144(10), 04018073.

Rathje, E. M., and Antonakos, G. (2011). “A unified model for predicting earthquake-
induced sliding displacements of rigid and flexible slopes.” Eng. Geol. 122 (1): 51–60.

Rathje, E.M. and Bray, J.D. (2000). “Nonlinear coupled seismic sliding analysis of earth
structures.” J. Geotech. Geoenviron. Eng., 126(11), pp. 1002–1014.

Rathje, E.M. and Bray, J.D. (2001). “One- and two-dimensional seismic analysis of solid-
waste landfills.” Can. Geotech. J., 38(4), pp. 850–862.

Rauch, A.F. (1997). EPOLLS: An Empirical Method for Predicting Surface

Displacements Due to Liquefaction-Induced Lateral Spreading in Earthquakes. PhD
Dissertation. Virginia Polytechnic Institute and State University. May 5, 1997.
Blacksburg, Virginia.

75
Rauch, A. F., Pace, T. G., Yankey, G., Dingrando, J. S., and Schaefer, J. A. (2007).
“Liquefaction under dams and levees: Back-of-the-envelope predictions of deformation.”
Proc., Dam Safety 2007, ASDSO, Austin, Texas, September, 19 pages.

Rix, G.J., de Melo, L., Saucier, C.L., and Morrison, M.A. (2017). “Seismic deformation
analysis of Blue Ridge Dam.” Proc., United States Society on Dams Annual Conference,
Anaheim, CA.

Rocscience. (2002). Slide 2D limit equilibrium slope stability for soil and rock slopes.
User’s Guide.

Roth, W.H., Bureau, G., and Brodt, G. (1991). “Pleasant Valley Dam: An Approach to
Quantifying the Effect of Foundation Liquefaction.” 17th International Congress on Large
Dams, Vienna, Austria.

Roth, W.H., Inel, S., Davis, C., and Brodt, G. (1993). “Upper San Fernando Dam 1971
Revisited.” 10th Annual Conference of the Association of State Dam Safety Officials,
Kansas City, Missouri, USA.

Ruthford, M., Perlea, V., Serafini, D., Beaty, M., Anderson, L., and Bowles, D. (2011).
“Risk assessment of Success Dam, California: earthquake induced potential failure
modes.” Geo-Risk 2011: Risk Assessment and Management, pp. 923-930.

Ryan, M.J., Makdisi, F.I., Wang, Z.L., and Yiadom, A. (2006). “Seismic Stability of San
Pablo Dam.” Proc., 100th Anniversary Earthquake Conference Commemorating the 1906
San Francisco Earthquake, San Francisco, CA. April.

Sarma, S.K. (1975). “Seismic stability of earth dams and embankments.” Geotechnique.
25(4), 743-761. Doi:10.1680/geot.1975.25.4.743.

Saygili, G. and Rathje, E.M. (2008). “Empirical predictive models for earthquake-
induced sliding displacements of slopes.” J. Geotechnical and Geoenvironmental
Engineering, ASCE, 134(6), pp. 790-803. https://doi.org/10.1061/(ASCE)1090-
0241(2008)134:6(790).

Schnabel, P B., Lysmer, J., and Seed, H B. (1972). SHAKE: A computer program for
earthquake response analysis of horizontally layered sites. Report No. EERC 72-12,
University of California, Berkeley, Calif., Earthquake Engineering Research Centre.

Seed, H.B. (1966). “A method for earthquake resistant design of earth dams.” Journal of
the Soil Mechanics and Foundation Division, ASCE, 92. No. SM1, Proceedings Paper
4615, 13-41.

Seed, H.B. (1973). “Stability of Earth and Rockfill Dams During Earthquakes.”
Embankment-Dam Engineering, Casagrande Volume, Hirschfeld and Poulos Eds.,
Krieger Publishing, John Wiley & Sons.

76
Seed, H.B. (1979a). “Considerations in the earthquake-resistant design of earth and
rockfill dams.” Geotechnique, 29(3), pp. 215-263.
https://doi.org/10.1680/geot.1979.29.3.215

Seed, H.B. (1979b). “Soil liquefaction and cyclic mobility evaluation for level ground
during earthquakes.” Journal of the Geotechnical Engineering Division, ASCE, Vol. 105
No. GT2, pp. 201-255.

Seed, H.B. (1987). “Design problems in soil liquefaction.” ASCE Journal of

Geotechnical Engineering, Vol. 113, No. 8, August.

Seed, H.B. and Lee, K.L. (1966). “Liquefaction of saturated sands during cyclic loading.”
Journal of the Soil Mechanics and Foundation Division, ASCE, 92. No. SM6,
Proceedings Paper 4972, 105-134.

Seed, H.B., Lee, K.L., and Idriss, I.M. (1969). “Analysis of Sheffield Dam failure.”
Journal of the Soil Mechanics and Foundation Division, ASCE, 95. No. SM6,
Proceedings Paper 6906, 1453-1490.

Seed, H.B., and Idriss, I.M. (1970). Soil moduli and damping factors for dynamic
response analysis. Report No. EERC 70-10, Earthquake Engineering Research Center,
University of California, Berkeley.

Seed, H.B. and Idriss, I.M. (1971). “Simplified procedure for evaluating soil liquefaction
potential.” Journal of the Soil Mechanics and Foundations Division, 97(7), pp. 1249-
1273.

Seed, H.B. and Peacock, W.H. (1971). “Test procedures for measuring soil liquefaction
characteristics.” Journal of the Soil Mechanics and Foundation Division, ASCE, 97. No.
SM8, Proceedings Paper 8330, pp. 1099-1119.

Seed, H.B., Lee, K.L., Idriss, I.M., and Makdisi, F. (1973). Analysis of the slides in the
San Fernando Dams during the earthquake of Feb. 9, 1971. Earthquake Engineering
Research Center 73-2, University of California, Berkeley, California.

Seed, H.B., Idriss, I.M., Lee, K.L., and Makdisi, F. (1975a). “Dynamic analysis of the
slide in the Lower San Fernando Dam during the earthquake of February 9, 1971.”
Journal of the Geotechnical Engineering Division, ASCE, Vol. 101.

Seed, H.B., Duncan, J.M., and Idriss, I.M. (1975b). “Criteria and methods for static and
dynamic analysis of earth dams.” Criteria and assumptions for numerical analysis of
dams. Swansea, University College, 564-588.

Seed, H.B., Martin, P.P., and Lysmer, J. (1976). “Pore-Water Pressure Changes During
Soil Liquefaction.” Journal of the Geotechnical Engineering Division, ASCE, Vol. 102,
No. GT4, pp 323-346.

77
Seed, H.B., Makdisi, F.I., and De Aba, P. (1978). “Performance of Earth Dams During
Earthquakes.” Journal of Geotechnical Engineering, ASCE, Vol. 104, No. GT7, pp. 967-
994.

Seed, H.B. and Idriss, I.M. (1982). Ground Motions and Soil Liquefaction During
Earthquakes. Earthquake Engineering Research Institute, Oakland, CA, 134 pp.

Seed, H.B., Seed, R.B., Harder, L.F. and Jong, H. (1988). “Re-evaluation of the slide in
the Lower San Fernando Dam in the earthquake of Feb. 9, 1971.” Report No.
UCB/EERC-88/04, University of California, Berkeley, California.

Seed, R.B., Bray, J.D., Boulanger, R.W., and Seed, R.W. (1989). “Seismic response of
the Puddingstone and Cogswell Dams in the 1987 Whittier Narrows Earthquake.” SMIP
89 Seminar Proceedings.

Seed, R.B. and Harder, L.F. (1990). “SPT-based analysis of cyclic pore pressure
generation and undrained residual strength.” Proceedings, H. Bolton Seed Memorial
Symposium, BiTech Publishers Ltd., Vancouver, Canada.

Seed, R.B., Cetin, K.O., Moss, R.E.S., Kammerer, A., Wu, J., Pestana, J., Riemer, M.,
Sancio, R.B., Bray, J.D., Kayen, R.E., and Faris, A. (2003). “Recent Advances in Soil
Liquefaction Engineering: A Unified and Consistent Framework.” Keynote presentation,
26th Annual ASCE Los Angeles Geotechnical Spring Seminar, Long Beach, CA.

Sherard, J.L. (1967). “Earthquake considerations in earth dam design.” Journal of the
Soil Mechanics and Foundations Division, ASCE, Vol. 93, No. SM4.

Sherard, J.L., Woodward, R.J., Gizienski, S.F., and Clevenger, W.A. (1963). Earth and
Earth-Rock Dams, John Wiley and Sons, New York.

Siddharthan, R. and Finn, W.D.L. (1982). TARA-2: two dimensional nonlinear static and
dynamic response analysis. Soil Dynamics Group, University of British Columbia,
Vancouver.

Silver, M. L., and Seed, H. B. (1971). “Volume changes in sands during cyclic loading.”
J. Soil Mech. and Found. Div., 97(9), pp. 1171-1182.

Soga, K., Alonso, E., Yerro, A., Kumar, K., and Bandara, S. (2015). “Trends in large-
deformation analysis of landslide mass movements with particular emphasis on the
material point method.” Géotechnique, 66(3), pp. 248-273.

Stark, T.D., Beaty, M.H., Byrne, P.M., Castro, G.V., Walberg, F.C., Perlea, V.G., Axtell,
P.J., Dillon, J.C., Empson, W.B., and Mathews, D.L. (2012). “Seismic Deformation
Analysis of Tuttle Creek Dam.” Canadian Geotechnical Journal, 49(3), pp. 323-343.

Swaisgood, J.R. (1998). “Seismically-induced deformation of embankment dams.”

Proceedings of sixth national conference on earthquake engineering. Seattle, Washington.
May 31-June 4.

78
Swaisgood, J.R. (2003). “Embankment Dam Deformations Caused by Earthquakes.”
Proceedings of 2003 Pacific Conference on Earthquake Engineering.

Swaisgood, J.R. (2014). “Behavior of embankment dams during earthquake.” Journal of

Dam Safety, ASDSO, 12(2), pp. 35-44.

Taiebat, M. and Dafalias, Y.F. (2008). “SANISAND: Simple anisotropic sand plasticity
model.” International Journal for Numerical and Analytical Methods in Geomechanics,
32(8), pp. 915-948.

Tan, P., Moriwaki, Y., Seed, R.B., and Davidson, R.R., (2000). "Dynamic run-out
analysis methodology (DRUM)", Tailings and Mine Waste '00. Balkema, Rotterdam.

Tasiopoulou, P., Taiebat, M., Tafazzoli, N., and Jeremic, B. (2015). “Solution
verification procedures for modeling and simulation of fully coupled porous media: static
and dynamic behavior.” Coupled Systems Mechanics, 4(1), pp. 67-98.

Taylor, D.W. (1953). Letters to South Pacific Division. United States Army Corps of
Engineers, San Francisco, 14 April and May 20.

Tokimatsu, K., and Seed, H. B. (1987). “Evaluation of settlements in sands due to

earthquake shaking.” J. Geotech. Engrg., 113(8), 861-878.

United States Bureau of Reclamation (USBR). (1998). Earth Manual. U.S. Department of
the Interior, U.S. Government Printing Office, Second Edition, Denver, Colorado.

United States Bureau of Reclamation. (2011). Interim Dam Safety Public Protection
Guidelines – Examples of Use. U.S. Department of the Interior, Denver, Colorado.
https://www.usbr.gov/ssle/damsafety/documents/PPGExamples201108.pdf

United States Bureau of Reclamation. (2015). Design Standards No. 13 – Embankment

Dams, Chapter 13 – Seismic Analysis and Design. U.S. Department of the Interior, U.S.
Government Printing Office, Denver, Colorado.

United States Bureau of Reclamation (USBR) and U.S. Army Corps of Engineers
(USACE). (2019). Best Practices in Dam Safety and Levee Safety Risk Analysis. Version
4.1. https://www.usbr.gov/ssle/damsafety/risk/methodology.html

United States Committee on Large Dams (USCOLD). (1992). “Observed Performance of

Dams During Earthquakes, Volume I.” Report of the Earthquakes Committee of
USCOLD, July.

USCOLD. (1999). “Updated Guidelines for Selecting Seismic Parameters for Dam
Projects.” Report of the Earthquakes Committee of USCOLD, April.

USCOLD. (2000). “Observed Performance of Dams During Earthquakes, Volume II.”

Report of the Earthquakes Committee of USCOLD, October.

79
United States Society on Dams (USSD). (2014). “Observed Performance of Dams During
Earthquakes - Volume III.” Report of the Earthquakes Committee of USSD, February.

Van den Berg, P., De Borst, R., and Huétink, H. (1996). “An Eulerean finite element
model for penetration in layered soil.” International Journal for Numerical and
Analytical Methods in Geomechanics, 20(12), pp. 865-886.

Wagner, N. and Sitar, N. (2016). “Seismic Earth Pressures on Deep Stiff Walls.” In Proc.,
Geotechnical and Structural Engineering Congress 2016, ASCE, pp. 499-508.

Wahler Associates, 1990, Austrian Dam-Investigation and remedial construction

following the October 17, 1989, Loma Prieta earthquake: Report prepared for the San
Jose Water Company.

Wald, D. J., Heaton, T. H., and Hudnut, K. W. (1996). “The slip history of the 1994
Northridge, California, earthquake determined from strong-motion, teleseismic, GPS, and
leveling data.” Bull. Seism. Soc. Am. 86, 49-70.

Wang, D., Bienen, B., Nazem, M., Tian, Y., Zheng, J., Pucker, T., and Randolph, M. F.
(2015). “Large deformation finite element analyses in geotechnical engineering.”
Computers and Geotechnics, 65, pp. 104-114.

Wang, Z.L. (1990). Bounding Surface Hypo-plasticity Model for Granular Soils and Its
Application. Ph.D. Dissertation. University of California at Davis.

Wang, Z.L. and Ma, F.G. (2018). WANG2D: A Plasticity Sand Model Updated Based on
Simple Shear Test Data. Documentation report on Itasca UDM Web Site, May.

Wang, Z.L, Dafalias, C.F., and Shen, C.K. (1990). “Bounding Surface Hypoplasticity
Model for Sand.” Journal of Engineering Mechanics, American Society of Civil
Engineering, 116(5), pp. 983-1001.

Wang, Z.L. and Makdisi, F. (1999). “Implementing a bounding surface hypoplasticity

model for sand into the FLAC program.” C. Detournay & R. Hart (eds), FLAC and
Numerical Modeling in Geomechanics, Proceedings of the International FLAC
Symposium, Minneapolis, MN, USA, 1-3 September 1999: 483-490. Rotterdam:
Balkema.

Wang, Z.L. and Dafalias, Y.F. (2002). “Simulation of post-liquefaction deformation of

sand.” Proc., 15th ASCE Engineering Mechanics Conference, June 2-5, 2002, Columbia
University, New York, NY.

Wang, Z.L., Dafalias, Y.F., Li, X.S., and Makdisi, F.I. (2002). “State pressure index for
modeling sand behavior.” J. Geotech. Geoenviron. Eng. 128(6), pp. 511-519.

Wang, Z.L., Makdisi, F.I., and Egan, J. (2006). “Practical Applications of a Non-linear
Approach to Analysis of Earthquake-Induced Liquefaction and Deformation of Earth
Structures.” Soil Dynamics and Earthquake Engineering, 26(2-4), pp. 231-252.

80
Wang, Z.L., Makdisi, F.I., and Ma, F.G. (2012). “Effective stress soil model calibration
based on in-situ measured soil properties.” J. Geotech. Geoenviron. Eng. 138(7), pp. 869-
875.

Wang, Z.L. and Ma, F. G. (2019). Bounding surface plasticity model for liquefaction of
sand with various densities and initial stress conditions. Soil Dynamics and Earthquake
Engineering. Volume 127, 105843.

Wartman, J., Bray, J.D., Seed, R.B., 2003. Inclined plane studies of the Newmark sliding
block procedure. Journal of Geotechnical and Geoenvironmental Engineering 129, 673-
684.

Wartman, J., Seed, R.B., Bray, J.D., 2005. Shaking table modeling of seismically induced
deformations in slopes. Journal of Geotechnical and Geoenvironmental Engineering 131,
610-622.

Weber, J.P. (2015). Engineering evaluation of post-liquefaction strength. Ph.D. diss.,

University of California, Berkeley. 158 pp. Available at
http://digitalassets.lib.berkeley.edu/etd/ucb/text/Weber_berkeley_0028E_15600.pdf;
accessed 26 October 2016

Wilson, R.C., Keefer, D.K., 1983. Dynamic analysis of a slope failure from the 6 August
1979 Coyote Lake, California, earthquake. Bulletin of the Seismological Society of
America 73, 863-877.

Wilson, R.C. and Keefer, D.K. (1985). “Predicting areal limits of earthquake induced
landsliding.” Evaluating Earthquake Hazards in the Los Angeles Region: An Earth-
Science Perspective (J.I. Ziony, ed.), U.S. Geological Survey Professional Paper 1360,
316-345.

Wu, J., Seed, R.B., and Pestana, J.M. (2003). Liquefaction Triggering and Post
Liquefaction Deformations of Monterey 0/30 Sand under Uni-Directional Cyclic Simple
Shear Loading. Geotechnical Engineering Report No. UCB/GE-2003/01, April 2003,
Univ. of California, Berkeley.

Yang, Z. (2000). Numerical modeling of earthquake site response including dilation and
liquefaction. Ph.D. thesis, Columbia University, New York, NY.

Yang, Z. and Elgamal, A. (2002). “Influence of permeability on liquefaction-induced

shear deformation.” Journal of Engineering Mechanics, 128(7), pp. 720-729.

Yang, Z., Elgamal, A., and Parra, E. (2003). “A computational model for cyclic mobility
and associated shear deformation.” Journal of Geotechnical and Geoenvironmental
Engineering, 129(12), pp. 1119-1127.

Yang, D., Naesgaard, E., Byrne, P.M., Adalier, K. and Abdoun, T. (2004). “Numerical
model verification and calibration of George Massey Tunnel using centrifuge models.”
Canadian Geotechnical Journal 41(5), pp. 921-942.

81
Yang, Z. and Elgamal, A. (2008). Multi-surface cyclic plasticity sand model with Lode
angle effect. Geotechnical and Geological Engineering, 26(3), 335-348.

Yegian, M.K., Marciano, E.A., and Ghahraman, V.G. (1991). “Earthquake-induced

permanent deformations: probabilistic approach.” J. Geotechnical Engineering., ASCE.
117(1). https://doi.org/10.1061/(ASCE)0733-9410(1991)117:1(35).

Yerro, A., Soga, K., and Bray, J.D. (2019) “Runout evaluation of Oso landslide with the
material point method.” Canadian Geotechnical Journal. 56(9).
https://doi.org/10.1139/cgj-2017-0630

Youd, T.L., Idriss, I.M., Andrus, R.D., Arango, I., Castro, G., Christian, J.T., Dobry, R.,
Finn, W.D.L., Harder, L.F., Hynes, M.E., Ishihara, K., Koester, J.P., Liao, S.S.C.,
Marcuson, W.F., Martin, G.R., Mitchell, J.K., Moriwaki, Y., Power, M.S., Robertson,
P.K., Seed, R.B., and Stokoe, K.H. (2001). “Liquefaction resistance of soils: summary
report from the 1996 NCEER and 1998 NCEER/NSF workshops on evaluation of
liquefaction resistance of soils.” J. Geotechnical and Geoenvironmental Eng., ASCE
127(10), 817–33.

Zeghal, M. and Abdel-Ghaffar, A.M. (1992). “Analysis of behavior of earth dam using
strong-motion earthquake records.” J. Geotech. Geoenvironmental Engineering, 118(2),
pp. 266–277.

Ziotopoulou, K. (2017). “Seismic response of liquefiable sloping ground: Class A and C

numerical predictions of centrifuge model responses.” Special issue of Soil Dynamics and
Earthquake Engineering, dx.doi.org/10.1016/j.soildyn.2017.01.038

Ziotopoulou, K., Boulanger, R.W., and Kramer, S.L. (2012). “Site response analysis of
liquefying sites.” In GeoCongress 2012: State of the Art and Practice in Geotechnical
Engineering (pp. 1799-1808).

Ziotopoulou, K., Maharjan, M., Boulanger, R.W., Beaty, M.H., Armstrong, R.J., and
Takahashi, A. (2014). “Constitutive modeling of liquefaction effects in sloping ground.”
Tenth U.S. National Conf. on Earthquake Eng., Frontiers of Earthquake Eng., July 21-25,
Anchorage Alaska.

Ziotopoulou, K., and Boulanger, R. W. (2013). “Numerical modeling issues in predicting

post-liquefaction reconsolidation strains and settlements.” Proc., 10th Int. Conf. on Urban
Earthquake Engineering, Tokyo Institute of Technology, Tokyo, 469–475

Ziotopoulou, K. and Boulanger, R.W. (2016). “Plasticity modeling of liquefaction effects

under sloping ground and irregular cyclic loading conditions.” Soil Dynamics and
Earthquake Eng., 84, pp. 269-283, 10.1016/j.soildyn.2016.02.013.

Zou, D., Zu, B., Kong, X., Liu, H. and Zhou, Y. (2013). “Numerical simulation of the
seismic response of the Zipingpu concrete face rockfill dam during the Wenchuan
earthquake based on a generalized plasticity model.” Computers and Geotechnics, 49, pp.
111-122.

82
 Appendix A: Selected Analytically-Based Correlations for Seismic
Deformation Analysis

This appendix summarizes key aspects of analytically-based correlations for estimating

seismic deformations of embankment dams. The reader should refer to the source
publications to obtain a full understanding of each method and their potential uses and
limitations.

Makdisi and Seed (1978)

This method employs a correlation developed using displacement results from sliding
block analyses that improved on the original Newmark method (see Chapter 5) by
considering the 2D dynamic response of an embankment. Two-dimensional finite
element equivalent-linear analyses were performed on four embankment sections using
four input motions representing earthquakes with magnitudes ranging from 6.5 to 8.2.
Dam section heights of 75, 135, and 150 feet (23, 41, and 46 m) were considered in the
development of the method. The analyses were used to determine average acceleration
histories for pre-defined slide masses. These histories were then used in sliding-block
analyses to calculate the resulting displacement of the slide masses. A series of charts
were developed from these analysis results for use in estimating seismically induced
displacements of embankments. These charts are shown in Figure A-1. One-dimensional
shear slice response calculations were the primary basis to determine the range of average
acceleration ratios (kmax/ümax) in the left plot of Figure A-1.

83
 ümax

kmax y
ky
h

Rigid Base

Figure A-1. Makdisi-Seed Simplified Method Charts (after Makdisi and Seed, 1977).

Although this method has the advantage of considering the dynamic response of the
embankment, it has limitations requiring consideration of the following: a) the evaluation
of sliding displacement is decoupled from the estimate of dynamic response (see Sections
3.2.2 and 5.2), which can produce displacement values differing from those calculated
using coupled analyses (Lin and Whitman, 1983; Rathje and Bray, 2000), b) the method
is based on only 4 ground motions (3 recorded motions, only one of which is from a
M>6.5 event, and one synthesized motion), which likely underestimates the potential
variability in calculated seismic displacement and the range of the displacement bounds
shown in Figure A-1, and c) the range of embankment heights considered in the method
development was only 75 ft to 150 ft.

Additional comments to be considered in applying this method are: 1) the dam was
assumed to be founded on a rigid base, which is not a satisfactory approximation for
many dam foundations, and 2) the method requires estimating the peak acceleration at the
crest of the embankment. A procedure to estimate this acceleration was proposed by
Makdisi and Seed (1978). However, the procedure involves multiple assumptions and
simplifications, which introduce uncertainty in the results, and is not commonly used.

In lieu of performing a 2D dynamic response analysis of a dam, empirical correlations

between the peak acceleration recorded at the crest and the peak acceleration recorded or

84
estimated at the base of multiple dams during past earthquakes (e.g., see Harder et al.,
1998) have been used in practice to estimate the peak crest acceleration for use with the
Makdisi and Seed method. It should be noted that for several of the case histories in these
charts, the motion at the base of the dam has been estimated based on motions recorded at
a short distance in the free field away from the toe of the dam. If the estimated motions at
the base of the dam can be considered as the free-field outcrop motion, then such charts
can be used to make a rough estimate of the crest acceleration in forward applications,
with due consideration for the significant uncertainty associated with these types of
correlations.

Bray and Travasarou (2007)

A semi-empirical predictive relationship was proposed by Bray and Travasarou (2007)

for estimating earthquake-induced permanent displacements of slopes using a nonlinear
sliding block model. This relationship is based on the results from analyses using 688
pairs of horizontal ground motion components from 41 earthquakes. The flexibility of the
block, and the interaction between yielding and seismic loading, were considered by
using a nonlinear coupled stick-slip deformable sliding model (Rathje and Bray, 2000).
The vibration characteristics of the sliding block are captured through an estimate of its
initial fundamental period, Ts.

The Bray and Travasarou (2007) correlation requires four input parameters: (1)
earthquake magnitude, (2) the yield coefficient of the selected potential sliding block, (3)
the Ts of this block, and (4) the spectral acceleration of the ground motion at a period of
1.5 times Ts. The correlation can be used to estimate the probability of exceedance of a
given displacement, the probability of negligible displacement (less than one centimeter),
and the displacement associated with a given probability of exceedance.

Advantages of the Bray and Travasarou (2007) method include simplicity of use,
consideration of block flexibility, coupling of dynamic response and displacement,
consideration of a large number of earthquake records, and adaptability to a fully
probabilistic framework. A limitation of the method is that it is based on 1D analysis
which is found to underestimate the seismic demand of shallow sliding at the top of 2D
systems (such as embankment dams) where topographic amplification can be significant.
Bray and Travasarou suggested that for these cases, the input PGA can be amplified by
25% for moderately steep slopes and by 50% for steep slopes.

Additional comments to be considered in applying this procedure are: (1) the period of
the potential sliding block is calculated using either the equation for a semi-infinite
horizontal soil layer over a rigid foundation or the equation for a triangular soil mass over
a rigid foundation; (2) the base of the sliding block is assumed to coincide with the base
of the earthfill model used to develop the method, as shown in Figure A-2(a), (3) the
height of the sliding block assumed in developing the method is equal to the full height of
the method’s earthfill model [H in Figure A-2(a)], and (4) the free-field outcrop motion is
applied as the input motion to the base in forward applications.

85
In the interest of simplicity, the Bray and Travasarou (2007) method assumed a flexible
sliding mass that can slide atop a rigid base, consistent with the approach recommended
by Kramer and Smith (1997). However, in practical applications, the potential sliding
block is often sitting on the embankment material (or foundation soil material), which is
not rigid, and the dynamic response of the sliding block is affected by the response of the
entire embankment-foundation system to the free-field outcrop motion. Because the
height of the sliding block is assumed in the method to equal the height of the earthfill
model in Figure A-2(a), adjustments need to be considered when applying the method to
potential sliding blocks that extend only part way the height of a dam, such as the sliding
block of height h in Figure A-2(b). For shallow sliding blocks at the top of 2D systems,
Bray and Travasarou (2007) indicate that the input PGA can be amplified in
consideration of ground motion amplification effects and the stiffness of shallow blocks.
The above notwithstanding, the method has been shown to yield reasonable
approximations to the displacements observed at several dams during past earthquakes
(see Bray and Travasarou, 2007).

Figure A-2. (a) Earthfill and sliding block models used to develop the Bray and
Travasarou (2007) method (from Bray and Travasarou, 2007). (b) Dam with sliding
blocks that extend part way the height of the dam (block of height h) and the full height
of the dam (block of height H).

Saygili and Rathje (2008)

The Saygili and Rathje model is a simplification of that shown in Figure A-2(a) in that
the sliding block is assumed to be a rigid block, as shown in Figure A-3, instead of a
deformable block.

Earth Fill

(a)
Potential Slide Plane

Figure A-3. (a) Earthfill slope system. (b) Rigid sliding block model used to develop the
Saygili and Rathje (2008) method (from Rathje and Bray 2000).

Saygili and Rathje computed Newmark displacements using motions selected from the
PEER NGA database. The initial selection included motions from earthquakes ranging
from Mw=5 to 7.9 and distances from 0.1 to 100 km. The analysis excluded motions
recorded at soft soil sites, below the ground surface, above the ground floor of a building,

86
in buildings taller than four stories, and on the crest or abutments of dams. Additionally,
motions with high-pass filter corner frequencies larger than 0.25 Hz or low-pass filter
corner frequencies less than 10 Hz were removed. Saygili and Rathje used 2,383 motions
to compute Newmark displacements for four yield accelerations (ky) values of 0.05, 0.1,
0.2 and 0.3. A number of correlations were developed utilizing single and multiple
ground motion parameters including PGA in g, PGV in cm/s, mean period of the
recorded motion (Tm in seconds), Arias Intensity (IA in m/s), and two measures of
duration based on Arias Intensity (D5-95 and D5-75 in seconds). The goal for developing
multiple correlations was to identify which ground motion parameters would minimize
the standard deviation of the displacement prediction.

The conclusion of Saygili and Rathje was that the combination of PGA, PGV, and IA
resulted in the smallest standard deviation in the displacement prediction. The Saygili and
Rathje (2008) displacement predictive models have the advantages of simplicity, a
curated database of input motions, choice of regression equations, and the capacity to be
incorporated into probabilistic hazard assessments.

The limitations of the Saygili and Rathje method include all of those associated with the
rigid sliding-block approach (see Section 5.2). Additional considerations in applying this
procedure are: 1) the method does not consider the effect of slide depth on the estimated
displacements, 2) the method does not account for the dynamic response of the dam, and
3) the free-field outcrop motion is applied as the input motion to the assumed rigid base
below the sliding block in forward applications.

It is noted that Rathje and Antonakos (2011) developed a predictive model to estimate
seismic displacements of rigid and flexible sliding blocks by extending the Saygili and
Rathje (2008) work for uncoupled conditions. However, the Rathje and Antonakos model
has not been widely used in practice.

Bray and Macedo (2019)

The semi-empirical predictive relationship proposed by Bray and Travasarou (2007) was
updated by Bray and Macedo (2019) using 6,711 pairs of horizontal ground motion
records from the NGA-West2 database and incorporating a modification from Bray et al.
(2018) to the equation of motion solution scheme. As compared to the Bray and
Travasarou (2007) correlation, the Bray and Macedo (2019) correlation lowers the
negligible displacement threshold from 1 cm to 0.5 cm, decreases the degraded period
from 1.5Ts to 1.3Ts, increases the recommended input PGA modification factor for
moderately steep slopes from 25 to 30%, and includes a method to account for near-fault
pulse motions with the addition of a PGV parameter for cases where the PGV is greater
than 115 cm/s.

The Bray and Macedo (2019) correlation generally estimates smaller seismic slope
displacement values than Bray and Travasarou (2007) at lower magnitudes and
fundamental periods for all values of ky evaluated. The same advantages and limitations
identified for the Bray and Travasarou (2007) correlation also apply to the Bray and

87
Macedo (2019) correlation. The additional comments provided above for the Bray and
Travasarou (2007) procedure apply to this procedure as well.

88
 Appendix B: Key Features of Selected Constitutive Models from US
Practice
General

This appendix presents a brief discussion of ten constitutive models which have been
used recently in US dam engineering practice. Table B-1 presents a summary of key
aspects of these models. These models are only meant to provide an example of the range
of models that have been used in practice. This is not an exhaustive list of the models
available or an endorsement of any model or its accuracy. Many other types of
constitutive models are available in various numerical platforms and the reader is
encouraged to contact their software developer to discuss options.

The constitutive models are grouped according to the following types: Mohr-Coulomb
model, extended Mohr-Coulomb models, and coupled effective-stress plasticity models.
The application cases referenced at the end of each constitutive model subsection are
offered as examples and references for the reader and are not meant to be an endorsement
of those projects or their authors.

The variety of model formulations and responses discussed in this appendix highlight the
importance of examining the response of a selected model to ensure that the model is
capturing the aspects of material behavior important to the problem being analyzed. A
user should not assume that a model will perform adequately for all problems, all
densities, all stress levels, or all loading conditions. Users should produce their own
example responses using single element simulations to verify that the model is
performing as expected and explore the limitations of the model.

Table B-1. Summary of Key Aspects of Nonlinear Deformation Analysis Models.

Stress- Post-
Pore- Analysis
strain 2D/3D Liquefaction
Model Pressure Platform References
Constitutive Capability Strength
Response Availability
Framework Simulation
FLAC, Itasca (2016),
Mohr- Linear- May Not in model
FLAC3D, Plaxis (2016)
Coulomb elastic, include framework.
PLAXIS, 2D, 3D Mckenna et al.
(Drucker- perfectly- volumetric Post-shaking
OpenSees, (2000),
Prager) plastic dilation analysis
LS-DYNA LTSC (2015)
Linear- May Not in model Naesgaard et al.
elastic, include framework. (2015),
UBCHYST FLAC 2D
Softening volumetric Post-shaking Naesgaard
after yield dilation analysis (2011)
Strength
Linear- Roth et al.,
Roth Effective reduces to
elastic, FLAC, (1991)
Cyclic- Stress, 2D, 3D residual
perfectly- FLAC3D Dawson and
Counting Decoupled strength after
plastic Mejia, (2012)
triggering

89
 Strength
Linear-
Total reduces to
elastic, tri- Beaty and
UBCTOT Stress, FLAC 2D residual
linear Byrne (2008)
Decoupled strength after
yielding
triggering
Nested Not in model Elgamal et al.
Effective
Yield CYCLIC, framework. (2003)
PDMY Stress, 2D,3D
Surface OpenSees Post-shaking Yang and
Coupled
Plasticity analysis Elgamal (2008)
Not in model
Bounding Effective Boulanger and
FLAC, framework.
PM4Sand Surface Stress, 2D Ziotopoulou,
OpenSees Post-shaking
plasticity Coupled (2015)
analysis
Not in model
Bounding Effective Boulanger and
framework.
PM4Silt Surface Stress, FLAC 2D Ziotopoulou
Post-shaking
plasticity Coupled (2018a)
analysis
Not in model
Bounding Effective
framework. Taiebat and
SANISAND Surface Stress, FLAC3D 2D, 3D
Post-shaking Dafalias (2008)
plasticity Coupled
analysis
Not in model
Hyperbolic Effective Beaty and
FLAC, framework.
UBCSAND Hardening Stress, 2D, 3D Byrne (1998,
PLAXIS Post-shaking
Plasticity Coupled 2011)
analysis
Not in model Wang et al.
Wang2D Bounding Effective
FLAC, framework. (1990)
and Surface Stress, 2D, 3D
FLAC3D Post-shaking Wang et al.
Wang3D plasticity Coupled
analysis (2006)

Mohr-Coulomb (Drucker-Prager)

One of the most common soil constitutive models used in NDAs is the Mohr-Coulomb
plasticity model. This model is commonly available in numerical programs used in
geotechnical engineering and requires only a few input parameters. These parameters
define the initial elastic stiffness and the yield surface (commonly defined by a friction
angle and cohesion) beyond which the soil will yield. Additional parameters may be
specified to allow for dilation or contraction during plastic yielding. The elastic-perfectly
plastic behavior of Mohr-Coulomb models is a crude approximation of real soil behavior,
especially under cyclic loading (e.g., Figure 6-1). Despite this, Mohr-Coulomb models
have been commonly used to model competent soils which are not expected to undergo
large strains or softening during loading.

Selecting Mohr-Coulomb properties for a dynamic analysis requires additional

considerations beyond those used in static analyses. The analyst may choose the elastic
modulus to account for cyclic degradation and to add numerical damping (e.g., Rayleigh
Damping) to increase the amount of energy dissipation during small strain cycles which
might otherwise be linear elastic. These values can be selected based on experience,
complementary equivalent-linear analyses or through an iterative analysis. The strength
of material zones in a dynamic analysis must also be carefully selected. The use of

90
frictional strengths may result in unintended coupling where the strength of the zone
changes in response to dynamic pore pressures. A method to avoid this is to specify an
undrained strength (cohesion) with a friction angle of zero which does not change during
the simulation. This undrained strength can be calculated based on the initial state of
stress within the dam. Adjustments may be made based on anisotropic consolidation, rate
effects, cyclic degradation or strength anisotropy. Further discussion of how undrained
strengths can be specified for NDAs of dams can be found in Montgomery et al. (2014).

Extensions of the Mohr-Coulomb framework can be used to add behavior like strain-
softening, hysteretic damping, modulus degradation and pore pressure generation. These
extensions retain the same limitation of an elastic-perfectly plastic model but may be able
to approximate other aspects of soil behavior which can be important for dam analyses.
Three of these extended Mohr-Coulomb models are discussed below.

Extended Mohr-Coulomb Models

UBCHYST

UBCHYST is a total stress model developed for dynamic analyses of soil systems
subjected to earthquake loading. Its primary use is to simulate the hysteretic stress-strain
and damping behavior of competent, non-degrading soils during seismic loading.
Descriptions of the model formulation can be found in Naesgaard (2011) and Naesgaard
et al. (2015). Its characteristics include a reduction in secant modulus with strain during
loading and unloading, the emulation of strain ratcheting that may occur in the presence
of a static shear-stress bias, and the option of imposing a permanent modulus reduction as
a function of maximum past shear stress ratio to reflect degradation of the soil fabric. The
model is intended for use with undrained shear strength parameters in low permeability
clayey and silty soils. Constant drained strengths are typically used for highly permeable
or non-saturated granular soils.

The formulation of UBCHYST is similar to the Mohr-Coulomb model except that the
elastic modulus degrades with increasing shear strain. The amount of degradation is
calculated by assuming that the tangent modulus (Gt) is a function of the peak shear
modulus (Gmax) multiplied by a reduction factor that is a function of the developed stress
ratio (η = τxy/σv′) and the margin in stress ratio to reach failure (defined as point on Mohr-
Coulomb failure envelope at same effective vertical stress, σv′). Some of the important
aspects of model behavior in UBCHYST are shown in Figure B-1 in terms of the secant
modulus and change in shear stress (Δτ). This includes increasing hysteretic behavior and
reduction in secant modulus (Gs) with greater strain (Figure B-1a) and/or greater number
of cycles (Figure B-1b). The model can also capture shear strain accumulation when
loaded under a static shear bias (Figure B-1c) and can be modified to include a reduction
in shear strength with increasing strain (Figure B-1d). A comparison of the model
response with results from a direct simple shear (DSS) test is shown in Figure B-2.

91
Figure B-1. Example Aspects of UBCHYST Behavior (Figure from Naesgaard et al.,
2015).

92
 Figure B-2. Comparison of the Test Data from Cyclic DSS With a Static Shear Bias
(Left) With a Schematic of UBCHYST Behavior (Right) (Figure from Naesgaard et al.,
2015).

The input parameters required for the UBCHYST model are the same as those of the
Mohr-Coulomb model with the addition of several parameters for calibrating the model
to selected modulus reduction and damping curves. Some features of UBCHYST include
its ability to incorporate hysteretic and ratcheting behaviors in a simple manner.
Limitations include its reliance on the stresses on vertical and horizontal planes instead of
principal stresses to estimate stress ratios, potential difficulties in calibrating the model to
damping curves at strains larger than about 0.1%, and its comparatively limited vetting
within industry. The model has been implemented in the explicit finite-difference
program FLAC (Itasca, 2011).

UBCHYST has been used in a number of applications, often as a secondary model in

soils not susceptible to liquefaction or major strength loss. The application of UBCHYST
to embankment dams includes a project that allowed model calibration to measured dam
response (Macedo et al., 2015), and the analysis of Lenihan (Lexington) Dam and its
response to the 1989 Loma Prieta earthquake (Hadidi et al., 2014). It has been used in
research at UC Berkeley to study the effect of seismic earth pressures on deep stiff walls
(Wagner and Sitar, 2016), and at University of British Columbia for studying basement
walls (Amirzehni, 2016).

Roth Cyclic-Counting Model

The Roth Cyclic-Counting Model is a decoupled effective stress model that uses the
Mohr-Coulomb linear-elastic-perfectly-plastic constitutive model to simulate soil stress-
strain behavior and an empirically based procedure to simulate pore pressure generation.
Pore pressure is generated in response to shear stress cycles, using the cyclic-stress
approach of Seed (Seed et al., 1976; Seed, 1979b) incrementally during shaking. The

93
model has been implemented in the explicit finite-difference program FLAC (Itasca,
2008), as well as FLAC3D (Itasca, 2009), through a routine that integrates pore pressure
generation with the dynamic effective-stress analysis.

Pore pressures are continuously updated for each element of the analysis mesh in
response to shear stress cycles. Effective stresses are decreased with increasing pore
pressure, and as the corresponding shear strength is exceeded during cyclic loading,
increments of permanent deformation are accumulated. The pore-pressure generation
routine is illustrated schematically in Figure B-3. The shear stress time history of each
element is monitored, and shear stress cycles are counted in half cycles, which are
detected through shear stress reversals. As soon as a stress half-cycle is detected, the
excess pore pressure is incremented by an amount dependent on the cyclic stress ratio of
that half-cycle. Excess pore pressure is accumulated at each stress reversal as a function
of the amplitude of the reversal, 2τcy, which is defined as shown in Figure B-3.

Figure B-3. Pore Pressure Generation Routine in the Roth Cyclic-Counting Model.

The increment of excess pore pressure due to each stress cycle detected is computed
based on the cyclic strength curve. The increment in excess pore-pressure ratio for the
reversal is taken as:

Δu/ σv′ = 0.5/Ni (1)

where Ni is the number of cycles to liquefaction under the cyclic stress ratio:

CSRi = τcy/ σv′ (2)

94
Ni is obtained from the relationship between liquefaction resistance and number of
cycles, as shown in Figure B-3. Pore pressure increases are accompanied with decreases
in effective stress and in shear strength. Once the strength decreases to the post-
liquefaction residual strength it is kept at that value in the analysis. The elastic shear
modulus is also reduced in proportion to the square root of the effective stress (or any
other user specified function). An example of the type of stress-strain response obtained
with the model in a hypothetical element test under uniform stress-controlled cyclic
simple shear is shown in Figure B-4. The shear stress ratio in the element equals the
applied stress ratio (0.22) at the start of the test and progressively decreases with
increasing excess pore pressure until it reaches the value of the residual shear strength
ratio (0.1).

0.3
shear stress ratio, τ/σ vc

0.2

0.1

0.0

-0.1

-0.2

-0.3
-1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.0
shear strain (%)
Figure B-4. Example Stress-Strain Response Obtained with Roth Cyclic-Counting Model
Under Uniform Cyclic Simple Shear (Courtesy Dr. E. Dawson)

The main input parameters for use of the model are the Mohr-Coulomb shear strength
parameters, the value of the elastic shear modulus, the value of the normalized clean-sand
standard penetration test blow count, (N1)60cs, which is used to define the cyclic resistance
of the soil, and the value of the post-liquefaction residual strength. The Seed Simplified
Procedure contains correction factors for adjusting the cyclic strength for initial vertical
effective stress greater than 1 atmosphere (Kσ) and for initial static shear stress (Kα). Both
factors are currently included in the model following the recommendations of Idriss and
Boulanger (2008).

The Roth cycle counting model has been employed on various projects involving
dynamic deformation analysis of dams (Roth et al., 1991; Roth et al., 1993; Dawson et al,
2001, Dawson and Mejia, 2012). The model is simple to understand, only requires a few
basic input parameters, does not require a separate step to simulate flow failures, and can
be used within a 3D framework. Its main limitation is in the representation of soil stress-
strain behavior through the elasto-plastic Mohr-Coulomb model, as discussed in the
previous section on Mohr-Coulomb models and shown in Figure 6-1a and Figure B-4.
The limitations of this formulation include reliance on viscous damping for stress states

95
below yielding, linear stress-strain behavior for stress conditions below yield, and the
inability to evaluate coupled pore water flow.

UBCTOT

UBCTOT is a simple total stress model capable of estimating the onset of liquefaction
and post-liquefaction stress-strain behavior in sand-like soils. The initial formulation of
the model is described in Beaty and Byrne (2008). UBCTOT uses a cumulative damage
(or cycle counter) approach to estimate the onset of liquefaction. Trends from laboratory
data and theoretical formulations are used to estimate the evolution of liquefaction with
time on an element by element basis. The model has been implemented in the explicit
finite-difference program FLAC (Itasca, 2011).

The original formulation of UBCTOT separated soil behavior into two regions: pre- and
post-liquefaction. A linear elastic, fully plastic Mohr Coulomb model is used to simulate
the pre-liquefaction behavior. Stiffness and strength properties are adjusted in each
element at the triggering of liquefaction. The stress-strain behavior of liquefied soil is
represented using a tri-linear model to simulate the development of degraded stress-strain
loops, and the strength of the liquefied soil is set equal to the selected residual (liquefied)
strength (Figure B-5). An example response of the model in cyclic DSS is shown in
Figure B-6.

Figure B-5. Constitutive Response of UBCTOT Showing the Transition from Elastic
Behavior to Liquefaction at Point A and Then to a Residual Strength at Point B (after
Beaty and Byrne, 2008).

96
 Figure B-6. Comparison of the UBCTOT Model and Experimental Data in Cyclic DSS
(after Beaty and Byrne, 2008).

UBCTOT is a practice-based model with reliance upon commonly used approaches for
evaluation of liquefaction resistance and residual strength (e.g., semi-empirical triggering
curves, Kσ and Kσ relationships, and common recommendations for residual strength).
Limitations of the original formulation include its reliance upon viscous damping for
stress states below yielding, linear elastic-plastic stress-strain behavior for pre-liquefied
conditions, and the inability to evaluate coupled flow. Modifications to the original
formulation have been made to address some of these shortcomings and allow its use for
low plasticity silts and clays (Beaty and Dickenson, 2015). These modifications include
nonlinear, hysteretic behavior prior to liquefaction that softens as a function of estimated
pore pressure generation.

Applications of the UBCTOT model include dams and slopes. Examples for the original
model formulation include the back-analysis of the Upper and Lower San Fernando
Dams for the 1971 San Fernando earthquake (Beaty and Byrne, 2000, 2001), and the
evaluation of the Massey Tunnel (Yang et al., 2004). The modified model has been used
to simulate the response of a low plasticity foundation layer at Tuttle Creek Dam (Stark
et al., 2012), low plasticity silts within the foundation of tall electrical transmission
towers (Beaty et al., 2014), and a back-analysis of the 4th Avenue slide in Anchorage
(Beaty and Dickenson, 2015).

Coupled Effective-Stress Plasticity Models

PDMY

The PDMY material (Yang and Elgamal, 2002; Elgamal et al., 2003; Khosravifar et al.,
2018) is a three-dimensional elasto-plastic material model for simulating the essential

97
response characteristics of pressure sensitive soil materials under general loading
conditions. Such characteristics include dilatancy (shear-induced volume contraction or
dilation) and non-flow liquefaction (cyclic mobility), typically exhibited in sands or silts
during monotonic or cyclic loading. The model was developed based on multi-surface-
plasticity theory (Prevost, 1985), with a non-associative flow rule to reproduce the
dilatancy effect. The yield surfaces are conical Drucker-Prager type surfaces as shown in
Figure B-7. The model was recently updated (Khosravifar et al., 2018) to comply with
the known dependence of liquefaction triggering to the number of loading cycles,
effective overburden stress (Kσ), and static shear stress (Kα) expected on sloping grounds.

Figure B-7. Conical Multi-Surface Yield Criteria in Principal Stress Space for the PDMY
Model (Figure from Khosravifar et al., 2018).

The PDMY model has been modified over the past two decades with extensive
calibration through numerous sources of data including downhole-array records,
laboratory tests, shake-table and centrifuge experiments. A key component of this
modeling effort lies in earlier calibration of the soil model (based on recorded response of
Nevada Sand at a relative density (Dr) of about 40%) under situations of liquefaction and
lateral spreading response (Parra, 1996; Yang, 2000; Elgamal et al., 2003, 2009; Yang et
al., 2003). The calibration process included data from centrifuge model tests to simulate
the 1D dynamic response of level and mildly sloping saturated cohesionless soil sites.
These tests (Dobry and Taboada, 1994; Dobry et al., 1995) were performed in a 1D
laminated container that emulated a prototype soil layer of 10 m depth and infinite lateral
extent. Data from the anisotropically consolidated, undrained cyclic triaxial tests
VELACS (verification of liquefaction analysis by centrifuge studies) No. 40-58
(Arulmoli et al., 1992) was also employed. In light of the relative complexity of the
model and input parameters, the calibration is developed such that the user can extract the
input parameters based solely on relative density (DR) or SPT (N1)60 values for clean
sand. The calibrated input parameters are provided in Khosravifar et al. (2018). Figure B-
8 shows an example model response in undrained cyclic simple shear loading for
(N1)60=5.

98
 0.2 0.2

Undrained cyclic DSS simulation

τ
(N1)60 = 5, σ′vc = 100 kPa, α = s/σ′vc = 0.0

Shear stress ratio (τ/σ vc)

Shear stress ratio (τ/σ vc)
0.1

′
0.1

0 0

-0.1 -0.1

-0.2 -0.2
-0.1 -0.05 0 0.05 0.1 0 0.2 0.4 0.6 0.8 1

Shear strain γ12 Vertical effective stress ratio (σ′v/σ′vc)

Figure B-8. Example Model Response in Undrained Cyclic Simple Shear Loading for
(N1)60=5 (Figure from Khosravifar et al., 2018).

The soil model has been available in OpenSees (Mckenna et al., 2000) since the early
2000s and was also implemented in FLAC and FLAC3D finite-difference frameworks. It
was verified that similar results are obtained using FLAC, FLAC3D and OpenSees for
the implemented model. As such, the soil constitutive model has been compiled as a
dynamic link library (DLL) with corresponding versions for FLAC (Versions 7 and 8)
and FLAC3D (Versions 5 and 6). Some features of the model include availability of 2D
and 3D versions implemented in multiple programs, the relatively straightforward
formulation of the multi-yield-surface framework, and extensive validation against
centrifuge test results. Some limitations include the large number of required input
parameters and a relatively complicated calibration procedure required for densities
outside of the example calibrations presented by Khosravifar et al. (2018).

The PDMY model has been successfully used to model the dynamic behavior of sands in
various situations. These include the behavior of underground reservoirs (Deng et al.,
2016) and retaining walls (Al Atik and Sitar, 2010). PDMY has also been calibrated and
validated for liquefaction problems using empirical correlations (Khosravifar et al., 2018)
and laboratory tests (Yang et al., 2003).

PM4Sand

The sand plasticity model PM4Sand (Version 3.1) was developed by Boulanger and
Ziotopoulou (2015) and Ziotopoulou and Boulanger (2016) for geotechnical earthquake
engineering applications. The initial version of the model was developed by Boulanger
(2010). The model follows the basic framework of the stress-ratio controlled, critical state
compatible, bounding surface plasticity model for sand presented by Dafalias and
Manzari (2004). Modifications to the Dafalias and Manzari model were developed and
implemented to improve its ability to approximate the stress-strain responses important to
geotechnical earthquake engineering applications; in essence, the model was calibrated at
the equation level to provide for better approximation of the trends observed across a set
of experimentally- and case history-based design correlations. The formulation and

99
implementation details can be found in Boulanger and Ziotopoulou (2015, 2017) and
Ziotopoulou and Boulanger (2016).

PM4Sand is a stress-ratio controlled model which means that the response in the model is
controlled by the ratio of mean to deviatoric stress rather than the absolute value of either.
The response of PM4Sand is controlled by three surfaces (Figure B-9): the yield surface
(limit of elastic behavior), the dilatancy surface, and the bounding surface. As the
shearing continues, the dilatancy and bounding surface collapse to the critical state
surface. This behavior is cast in terms of relative state ξR (i.e., the difference between the
relative density DR and the relative density at critical state DRcs for the current confining
pressure), such that the soil properties can change during the simulation as a function of
the change in state (i.e., changes in mean effective stress and/or in void ratio). Figure B-
10 shows an example response of the model in cyclic DSS.

Figure B-9. Schematic of Yield, Critical, Dilatancy, and Bounding Lines in q-p Space
(After Dafalias and Manzari, 2004). Relative Location of Dilatancy and Bounding Lines
Corresponds to Dense-of-Critical States of Stress (Figure from Boulanger and
Ziotopoulou, 2015).

Figure B-10. Undrained Cyclic DSS Loading Response from PM4Sand for DR = 55%
With Vertical Effective Consolidation Stress (σ´vc) of 1 atm and With Initial Static Shear
Stress Ratios of 0.0 (Figure from Boulanger and Ziotopoulou, 2015).

100
PM4Sand Version 3 has 22 input parameters, from which only three are considered
primary and are required as model inputs (initial relative density, shear modulus
coefficient, and contraction rate parameter). The other 19 (2 flags and 17 secondary
parameters) can be left with their preset default values if no other information is available
or calibrated to the desired response based on available laboratory data. Some features of
the model are that it directly incorporates critical state behavior, it can approximate the
response of sands across a range of stress levels and densities, and it has been extensively
validated. Its complexity provides for a more realistic modeling of the complex behaviors
exhibited by liquefying sands. Some limitations of the model are its plane strain
formulation and its inability to model compression/crushing under loading paths with
large increases in mean stress (e.g., at the tips of piles or cone penetrometers). The model
is coded as a user defined material in a dynamic link library (DLL) for use with the
commercial program FLAC 7.0 (Itasca 2011) and FLAC 8.0 (Itasca 2016) and is
accompanied by a report wherein the behavior of the model is illustrated by simulations
of element loading tests covering a broad range of conditions.

PM4Sand has been used successfully in a variety of projects including analyses for dams,
buildings, tunnels and levees. These include buildings affected by liquefaction in
Christchurch (Bray and Luque, 2017), seismic design of a sediment retention structure
(Martin and Malvick, 2016) and the analysis of Blue Ridge Dam (Rix et al., 2017).
PM4Sand has also been extensively validated against centrifuge test results and case
histories (e.g., Ziotopoulou, 2017; Montgomery et al., 2017; Ziotopoulou et al., 2012).

PM4Silt

The plasticity model PM4Silt (Version 1) was developed by Boulanger and Ziotopoulou
(2018a) for representing the undrained monotonic and cyclic loading responses of
saturated low-plasticity silts and clays that exhibit stress-history normalized behaviors, as
opposed to the responses of purely non-plastic silts and sands. The model builds on the
framework of PM4Sand and is a stress-ratio controlled, critical state compatible,
bounding surface plasticity model. The formulation and implementation details can be
found in Boulanger and Ziotopoulou (2018b).

PM4Silt has as similar bounding surface formulation to PM4Sand (Figure B-9) with
modifications to the parameters and response to better approximate the undrained
monotonic and cyclic loading of low-plasticity silts and clays. The goal of the
modifications was to obtain reasonable approximations of undrained monotonic shear
strengths, undrained cyclic shear strengths, and shear modulus reduction and hysteretic
damping responses across a range of initial static shear stress and overburden stress
conditions. The model does not include a cap, and therefore is not suited for simulating
consolidation settlements or strength evolution with consolidation stress history. The
model is cast in terms of the state parameter (i.e., the difference between the void ratio e
and the void ratio at critical state ecs for the current confining pressure) relative to a linear
critical state line in void ratio versus logarithm of mean effective stress. Figure B-11
shows an example response of the model in cyclic DSS.

101
 Figure B-11. Undrained Cyclic DSS Loading Response from PM4Silt for an Undrained
Shear Strength Ratio (su/σ´vc) With Vertical Effective Consolidation Stress (σ´vc) of 1 atm
and With Initial Static Shear Stress Ratio of 0.0 (Figure from Boulanger and Ziotopoulou,
2018).

PM4Silt Version 1 has 24 input parameters, from which only three are required as model
inputs (undrained shear strength, the shear modulus coefficient and the contraction rate
parameter). The other 21 (2 flags and 19 optional parameters) can be left with their preset
default values if no other information is available or calibrated to the desired response
based on available laboratory data. Some features of the model are that it directly
incorporates critical state behavior, it can approximate important shear modulus and
damping correlations for low plasticity silts and clays and can directly incorporate
undrained shear strength. Some limitations of the model are that it is not formulated to
approximate anisotropic strengths and is currently limited to plane strain applications.
The yield surface in the model also does not include a cap and therefore is not suited for
simulating consolidation processes, predicting consolidation settlements, or predicting the
evolution of undrained shear strength with consolidation stress history. The model is
coded as a user defined material in a dynamic link library (DLL) for use with the
commercial program FLAC 8.0 (Itasca 2016) and is accompanied by a report wherein the
behavior of the model is illustrated by simulations of element loading tests covering a
broad range of conditions.

PM4Silt is a relatively new model but has been used in several recent dam studies
including to evaluate the seismic stability of Blue Ridge Dam and to analyze the
performance of Austrian Dam in the 1989 Loma Prieta earthquake (Boulanger 2020). It
has also been used to evaluate cyclic softening of fine-grained soils in non-dam related
studies (e.g., Hutabarat and Bray, 2019; Boulanger et al., 2019) and to analyze the Fourth
Avenue landslide in Anchorage during the 1964 earthquake (Kiernan and Montgomery,
2020).

SANISAND

The SANISAND (Simple Anisotropic Sand) family of models includes several models
which fall within the general framework of critical state soil mechanics and bounding
surface plasticity. This family of models has been developed to capture the complex

102
stress-strain behavior of sands under a variety of loading conditions and stress levels. The
development of this family of models began with the models proposed by Manzari and
Dafalias (1997) and Dafalias and Manzari (2004) and has undergone several
modifications and additions since then. These modifications include the addition of
inherent anisotropy (Dafalias et al., 2004), plastic strains under constant stress ratio
(Taiebat and Dafalias, 2008) and the anisotropic critical state (Li and Dafalias, 2012).
The details of the formulation for these different versions can be found in the respective
references.

All of the SANISAND models are formulated within the framework of critical state soil
mechanics. The critical state line is defined as a power law in void ratio-stress space
(Figure B-12a). The response of the model is controlled by the position of the yield
surface (the limit of elastic behavior) relative to the dilatancy surface and the bounding
surface. These surfaces are shown in Figure B-12b. As shearing continues, the dilatancy
and bounding surfaces collapse to the critical state surface at which point shearing
continues with no further change in volume. This behavior is cast in terms of the state
parameter (i.e., the difference between the void ratio and the void ratio at critical state for
the current confining pressure), such that the soil properties can change during the
simulation as a function of the change in state (i.e., changes in mean effective stress
and/or in void ratio). The yield surface in SANISAND is either defined as a narrow cone
and the model response is controlled by the stress-ratio (Dafalias and Manzari, 2004) or
the yield surface is defined using a cap which allows for plastic strain at a constant stress-
ratio (Taiebat and Dafalias, 2008). Figure B-13 shows the response of a single element
subject to cyclic simple shear loading.

Figure B-12. (a) Illustration of the Critical State Line for SANISAND in Void Ratio-
Stress Space and (b) Schematic Illustration of the Locations of the Yield, Dilatancy,
Critical State and Bounding Surfaces for SANISAND in Deviatoric (q) and Mean
Effective Stress (p) Space (Figures from Taiebat and Dafalias, 2008).

103
 Figure B-13. Undrained Cyclic Simple Shear Loading of the SANISAND model for a
Void Ratio of 0.833 and a Mean Effective Stress of 100 kPa for Two Values of Ko
(Figure from Cheng et al., 2013).

Various versions of SANISAND have been implemented into several numerical

programs including OpenSEES, PLAXIS, FLAC and FLAC3D. These implementations
do not all use the same version of SANISAND and so the user must ensure that the
implementation they are using has the desired features. The version implemented in
FLAC3D uses the Dafalias and Manzari (2004) model and has 16 input parameters which
can be selected based on results of triaxial tests or using relationships from the literature
(Cheng, 2018). Some features of the SANISAND family of models include their ability
to capture details of sand behavior not commonly incorporated into constitutive models
including Lode angle (relates the magnitude of the intermediate principal stress to major
and minor stresses; Lode, 1926) dependence, particle crushing and anisotropy. Some
limitations include a relatively complex formulation and calibration procedure. This
model also does not include a mechanism for damage/fabric accumulation under uniform
symmetric loading, which results in a repeating stress-strain pattern after a certain strain
level (shown in Figure B-13) rather than strain accumulation as is observed in lab tests.

SANISAND models have been successfully used to model the liquefaction behavior of
sands in a variety of situations. These include the behavior of tailings dams (Barerro et
al., 2015) and centrifuge tests (Tasiopoulou et al., 2015). SANISAND has also been
extensively verified and calibrated against data from a variety of laboratory tests (e.g.,
Taiebat and Dafalias, 2008).

UBCSAND

UBCSAND is a 2D effective stress plasticity model developed primarily for sand-like

soils having the potential for liquefaction under seismic loading (e.g., sands and silty
sands with a relative density less than about 80%). Descriptions of the model can be
found in Beaty and Byrne (1998), Byrne et al. (2004), and most recently in Beaty and
Byrne (2011). The model simulates the hysteretic shear stress-strain behavior of the soil
using an assumed hyperbolic relationship. The associated volumetric response of the soil

104
skeleton is based on a flow rule that is a function of the current stress ratio (or φmob), the
stress ratio at constant volume conditions (or φcv), and the stress ratio history. The model
reasonably simulates the cyclic response for both pre-liquefaction and post-liquefaction
conditions.

UBCSAND is a stress-ratio controlled model which means that the response in the model
is controlled by the ratio of mean to deviatoric stress rather than the absolute value of
either. The model adopts a Mohr-Coulomb type yield surface (Figure B-14a) which is
able to expand as the stress ratio changes until reaching the peak friction angle. The
amount of plastic strain that occurs is calculated through an assumed hyperbolic
relationship for the plastic modulus (Gp) (Figure B-14b). The ratio of plastic deviatoric
strain (dγp) to plastic volumetric strain (dεvp) is calculated based on the difference
between the current stress ratio and the critical state stress ratio. For stress ratios higher
than the critical ratio, the model predicts dilation while for stress ratios less than the
critical state friction angle the model predicts contraction, which allows for the
generation of excess pore pressure in coupled simulations. Example model responses
from DSS simulations are shown in Figure B-15.

Figure B-14. The Yield Surface (upper plot) and Relationship Between the Stress Ratio
and Plastic Strain Increment (lower plot) Used in the Formulation of the UBCSAND
Model (from Beaty and Byrne, 2011).

105
 Figure B-15. Laboratory DSS and UBCSAND (Version 904aR) Using Generic Input
Parameters (from Beaty and Byrne, 2011).

UBCSAND has 14 input parameters of which two are considered primary. The remaining
12 parameters (2 flags and 10 other parameters) can be set using a function provided by
the developers which correlates the remaining parameters to the (N1)60cs value. Some
features of UBCSAND include the ability to simulate liquefaction behavior within a
relatively simple model framework, an effective stress formulation that allows
mechanical and transient seepage response to be fully coupled, a limited number of
model parameters that typically have well understood definitions, the availability of
generic model parameters for a hypothetical sand that are a function of relative density
(or corrected penetration resistance), and its history of use for seismic analysis projects
since the early 2000s. Some limitations include its tendency to produce a limited strain

106
development under symmetric loading cycles and a lack of a critical state framework to
capture the effects of changes in density on the behavior of the model. In addition,
UBCSAND has been modified considerably since its initial publication in the mid-1990s.
There are many versions in current use, not all of which have the same capabilities,
behaviors, or input parameters. Various versions of the model have been implemented in
both FLAC (Itasca 2011) and Plaxis (2016). The user of UBCSAND must be cautious to
confirm the selected version is consistent with their expectations. Version 904aR is
considered to be the most validated and distributed version and the one that has seen the
widest use within the dam engineering community.

UBCSAND has been applied to the seismic evaluation of many embankment dams.
These include the evaluation of Tuttle Creek Dam and the effect of liquefiable foundation
sand on the deformations of the structure (Stark et al., 2012). The model was used to
analyze the non-remediated and remediated structure and was useful for optimizing the
dam modifications. UBCSAND was also used in analyses of Success Dam, which
included one cross section incorporating an embedded concrete control tower (Beaty et
al. 2010). Other projects have included foundation stability evaluations of electrical
transmission towers (Beaty et al., 2014), the BART tube (Chou et al., 2010), the Massey
Tunnel (Yang et al., 2004), and research studies for piled embankments (Armstrong et al.,
2013).

Wang2D and Wang3D

The Wang2D and Wang 3D models are bounding surface plasticity models developed to
simulate sand behavior under a variety of loading conditions. The original formulation of
the model was developed within the general framework of bounding surface plasticity
(Figure B-16) and is described by Wang (1990) and Wang et al. (1990). After some use,
several significant sand behaviors were incorporated including: a) the critical state and/or
residual strength of sand as an ultimate state through the use of the 'state pressure index'
(Wang and Makdisi, 1999; Wang et al., 2002; Wang et al., 2012), b) simulation of post-
liquefaction deformation of sand (Wang and Dafalias, 2002), and c) the capability to
simulate asymmetric cyclic loading, that is, the effect of initial shear stress on cyclic
resistance. The model can successfully simulate laboratory results for sand from
conventional triaxial (i.e., axisymmetric loading), true triaxial, hollow cylindrical
(torsional shear) and simple shear tests, for both monotonic and cyclic loading, under
drained and undrained conditions. It was specifically formulated to capture the
contractive and dilative behavior induced by shear stress, under “rotational shear” (or
“neutral loading”). This allows prediction of buildup of pore pressure and liquefaction.

The Wang2D and Wang3D models are stress ratio-controlled models which means that
the response in the models is controlled by the ratio of mean to deviatoric stress rather
than the absolute value of either. This leads to cone-shaped yield and failure surfaces in
3D stress space (Figure B-16). The mean stress, deviatoric stress, and stress ratio are
defined as follows,

(𝜎𝜎11 +𝜎𝜎22 +𝜎𝜎33 )

𝑝𝑝 = , 𝑠𝑠𝑖𝑖𝑖𝑖 = 𝜎𝜎𝑖𝑖𝑖𝑖 −𝑝𝑝𝑝𝑝𝑖𝑖𝑖𝑖 , 𝑟𝑟𝑖𝑖𝑖𝑖 = 𝑠𝑠𝑖𝑖𝑖𝑖 /𝑝𝑝 (3)
3

107
The surfaces can also be visualized in terms of the deviatoric stress invariant (J) and
stress ratio invariant (R), which are defined as follows,

𝐽𝐽2 = 0.5 ∗ 𝑠𝑠𝑖𝑖𝑖𝑖 𝑠𝑠𝑖𝑖𝑖𝑖 and 𝑅𝑅 = 𝐽𝐽/𝑝𝑝 (4)

The failure surface (R-Rf = 0) and the loading surface (R-Rm= 0), defined in the stress
space J-p, are as shown in Figure B-16, where nij is a unit vector in stress space along the
direction of the deviatoric plastic strain increment. The ratio of the current pressure in a
soil element, p, to the critical pressure, pc, corresponding to the current void ratio e, was
introduced as the state pressure index Ip= p / pc. This index is incorporated in the
dilatancy line Rd to provide a curved dilatancy line that is unique to each pc:

𝑅𝑅𝑑𝑑 = 𝑅𝑅𝑝𝑝 + �𝑅𝑅𝑓𝑓 − 𝑅𝑅𝑝𝑝 �𝐼𝐼𝑝𝑝 , 𝑅𝑅𝑑𝑑 ≤ 𝑅𝑅𝑓𝑓 (5)

Where Rp is a constant that defines the slope of the phase transformation line. Reversals
in loading are tracked using the variables ρ (the distance from the current stress ratio to
the reversal stress ratio), 𝜌𝜌̅ (the distance from the reversal stress ratio to the maximum
pre-stress surface, Rm) and 𝜌𝜌𝜌𝜌 (the distance from reversal point to the dilatancy surface,
Rd). All the above surfaces and variables are explained in Figure B-16. Significant
modification to the Wang model was incorporated recently (Wang and Ma, 2019). The
new Wang model was calibrated and validated using a set of simple shear laboratory test
data (Wu et. al., 2003) as shown in Figure B-17. Figure B-18 compares the model
response to empirical relationships for Kα and Kσ.

Figure B-16. Definitions of Surfaces and Variables of Wang's Plasticity Model for Sand
(Wang, 1990 and Wang et al., 2002).

108
 0.3
0.3
0.2
0.2
0.1
0.1
0.0
CSR

0.0

CSR
-0.1
-0.1
-0.2
-0.2
-0.3
-0.3
0.0 0.2 0.4 0.6 0.8 1.0
0.0 0.2 0.4 0.6 0.8 1.0
σ'v/σ'v0 σ'v/σ'v0
0.3 0.3
0.2 0.2
0.1 0.1
0.0 0.0
CSR

CSR
-0.1 -0.1
-0.2 -0.2
-0.3 -0.3
-6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
Shear strain , % Shear strain , %

Figure B-17. Normalized shear stress ratio vs vertical effective stress ratio: CSR=0.200,
(MS23j), initial σ’v = 81kpa, Dr=43%. Test data (in blue, after Wu et. al., 2003) shows 8
cycles of loading to reach a ru of 95%, while the model prediction shows 11 cycles (in
red).

Figure B-18. Computed Kα for Monterey Sand in comparison with relationship from
Harder and Boulanger (1997) and Computed Kσ for Monterey sand compared with
relationship from Idriss and Boulanger (2008) and Youd et al. (2001). Figures after Wang
and Ma (2019).

109
The Wang models require eight input parameters which may be calibrated based on in-
situ tests or laboratory test results. Model parameters have previously been calibrated
based on sand test data from conventional triaxial, true triaxial, hollow cylindrical, and
simple shear tests. Some features of the model include a critical-state-based formulation,
the ability to model cyclic and monotonic loading under drained and undrained
conditions, the inclusion of Kα and Kσ effects based on results of laboratory tests, and the
ability to simulate post-liquefaction deformation and plastic modulus softening using a
residual strength. Some limitations of the model are the cone-type yield surface (Figure
B-16). In addition, the ability to model consolidation and rotational shear stresses are
disabled in current versions. The original version of the model was written in general
stress and strain tensors for 3D conditions and implemented into SUMDES. A 2D
formulation was later derived for plane strain conditions and implemented into FLAC
(Wang and Makdisi, 1999). The model is currently available as a user-defined model for
FLAC and FLAC3D (Ma and Wang, 2008).

Since 1999, the Wang2D model, as a User Defined Model (UDS) in FLAC, was used in a
number of projects for evaluating the earthquake-induced deformation and seismic
stability of dams. Examples include FLAC dam analyses performed for Success dam
(Wang et al. 2006), New Exchequer dam (Makdisi et al., 2000) and San Pablo dam (Ryan
et al., 2006). A 3D version of the Wang model was also used in analyzing soil structure
interaction for a tunnel of the Bay Area Rapid Transit (BART) system (Ma et al. 2016).
The model has also been compared with case histories of site response and pore water
pressure generation during the Lotung earthquake (Li et al., 1998) and of deformation at
the Lower San Fernando Dam (Ma et al., 2006).

110
 Appendix C: Review and Documentation Considerations Specific to
Analysis Approach
Items that should be addressed during review and documentation depend upon the
complexity of both the analysis method and implementation as well as any special
aspects of the structure being analyzed. Boxes C.1 and C.2 present detailed lists of items
that should typically be considered, although these lists are not exhaustive. Brief
comments are provided below.

Simplified and Sliding Block Analyses

The review and documentation of simplified analyses and sliding block analyses should
include the description of the analysis approach and assumptions, the analysis approach
limitations, and how those limitations may impact the estimated dam response and
deformations. Despite the relative simplicity of these analyses, the careful development
and documentation of a sufficient site characterization is a priority, including
identification of potential limitations. Items that should be considered in the review and
documentation are listed in Box C.1. An assessment of the adequacy of the analysis
approach and results to support the conclusions and recommendations should be
provided.

111
 Box C.1 Considerations for review and documentation of simplified analyses
and/or sliding block analyses

 Seismic failure modes

o Potential seismic failure modes.
o Limitations of analysis procedures in simulating key dam and material
behaviors.

 Site characterization
o Basis for dam and foundation zoning and material parameters.
o Extent of soils susceptible to liquefaction or marked cyclic softening

 Empirical analyses
o basis for input parameters.
o assessment of uncertainty in results.

 Limit-equilibrium, post-earthquake, and yield acceleration analyses

o analysis program; solution method(s); cross sections including strength
parameters, unit weights, pore pressures.
o significant shear surfaces with associated yield acceleration values; basis for
selection of critical surfaces.

 Dynamic response analyses

o basis for selection of input ground motions; approach used to estimate average
acceleration response of sliding mass.
o selection of stiffness models (e.g., VS profile) and relationships for modulus
reduction and damping.
o appropriateness of 1D or 2D analysis models.

 Newmark-type analyses
o use of rigid or deformable sliding mass.
o displacement estimates for both orientations of each earthquake record;
estimates of likely displacements and uncertainties.

 Conclusions
o Primary uncertainties and limitations of the analyses.
o Conclusions and recommendations with due consideration of the geology,
material characteristics, seismic hazard, potential consequences, and key
sources of uncertainty.

Nonlinear Deformation Analyses

Because of their greater complexity, the review and documentation for NDAs is generally
more extensive than for the simpler analysis methods. Adequate demonstration of the
suitability and limitations of an NDA is dependent on transparent and clear description of

112
the modeling input, process, and response estimates. Insufficient documentation can
undermine the review of an advanced analysis and reduce its value.

The documentation of an NDA should typically describe, or include references to, the site
characterization and the basis for selection of the input parameters, the overall analysis
approach including all aspects of significance to the dam response estimates, the key
features and basis for selection of the constitutive models, the model calibration process
and its suitability, the estimated element behavior from calibration and final model
analyses, the representative dam response estimates, presentation of parametric analysis
results, and an assessment of the appropriateness of the analysis method and results.

Boulanger and Beaty (2016) provide a detailed list of items that should be considered
when documenting and reviewing an NDA of an embankment dam. Boulanger and
Ziotopoulou (2018b) provide additional discussion on the subject. Items that should be
considered in the review and documentation of NDAs are listed in Box C.2.

113
 Box C.2 Considerations for review and documentation of NDAs
(after Boulanger and Beaty, 2016)

 Seismic failure modes

o Potential seismic failure modes.
o Dam and material behaviors important to the identified failure modes.
o Limitations in the ability of the analysis procedure to simulate key behaviors.

 Numerical modeling procedure

o Description of the analysis process.
o Significant modeling assumptions regarding embankment construction,
reservoir filling and seepage, seismic response, post-earthquake evaluation, etc.
o Description of the analysis model including boundary conditions, model extent,
material zoning, and numerical grid refinement.

 Validation of numerical modeling procedure

o Reference or description of how the constitutive models and numerical
modeling procedures have been validated for similar systems.
o Demonstration of how the constitutive models simulate material behaviors
consistent with empirical trends over the conditions of interest (e.g., CRR versus
N, effects of initial confining stress and static shear stress ratio α).

 Site characterization
o Basis for material parameters.

 Calibration of constitutive model

o Description of calibration processes.
o Constitutive model parameters.
o Demonstration of key material behaviors of the calibrated model under
representative initial stress and loading conditions.

 Input ground motions

o Appropriateness of the input ground motions in view of the seismic hazard and
numerical modeling approach.
o Significant characteristics of the input ground motions (e.g., source and site,
response spectra, duration, Arias Intensity, CAV, PGV, PGA).
o Protocols for application of motions to the model (e.g., fixed or compliant base).

 Initial static stress conditions

o Reasonableness of estimated static stresses, including distributions of effective
stress, K0, and α.
o Static pore pressure estimates in view of available piezometer data.

114
 Box C.2 (Continued)

 Dynamic response
o Description of dynamic analysis and summary of results.
o Acceleration and displacement time series and response spectra for key points;
stress paths and stress-strain responses for representative elements; and
deformed shapes with contours of strain, excess pore pressure ratio,
displacement, etc.

 Post-earthquake deformations
o Description of post-earthquake analysis and summary of results.
o Detailed results for one (or more) ground motions.
o Key factors addressed (e.g., estimates of residual strength, cyclic degradation,
remolding, rate effects, dilation, cracking, void redistribution).

 Parametric analyses
o Description of parametric analysis program including basis for selection of
investigated parameters.
o Identification of behaviors and parameters of most importance.
o Tabulated and graphical summaries of analysis results for both dynamic and
post-earthquake responses.

 Conclusions
o Primary uncertainties and limitations in the NDA model.
o Conclusions and recommendations drawn from the analyses, with due
consideration given to the geology, material characteristics, system
configuration, seismic hazard, estimated response, potential consequences, and
key sources of uncertainty.

115
 U.S. Society on Dams
13918 E. Mississippi Ave., #61160
Aurora, CO 80012
Phone: 303-792-8753
Website: USSDams.org
E-mail: info@ussdams.org

______________________________________________________________________________
73