PACIFIC EARTHQUAKE ENGINEERING

RESEARCH CENTER

PEER Workshop on Liquefaction

Susceptibility

Armin W. Stuedlein1
Besrat Alemu1
T. Matthew Evans1
Steven L. Kramer2
Jonathan P. Stewart3
Kristin Ulmer4
Katerina Ziotopoulou5

1
School of Civil and Construction Engineering,
Oregon State University, Corvallis, Oregon
2
Department of Civil and Environmental Engineering,
University of Washington, Seattle, Washington
3
Department of Civil and Environmental Engineering,
University of California, Los Angeles, California
Geoscience and Engineering Department,
4

Southwest Research Institute, San Antonio, Texas

5
Department of Civil and Environmental Engineering,
University of California, Davis, California

PEER Report No. 2023/02

Pacific Earthquake Engineering Research Center

Headquarters at the University of California, Berkeley
PEER 2023/02 May 2023
May 2023
 PEER Workshop on Liquefaction
Susceptibility

Armin W. Stuedlein1
Besrat Alemu1
T. Matthew Evans1
Steven L. Kramer2
Jonathan P. Stewart3
Kristin Ulmer4
Katerina Ziotopoulou5

1 School of Civil and Construction Engineering, Oregon State University, Corvallis, Oregon

2
Department of Civil and Environmental Engineering, University of Washington, Seattle, Washington
3
Department of Civil and Environmental Engineering, University of California, Los Angeles, California
4
Geoscience and Engineering Department, Southwest Research Institute, San Antonio, Texas
5
Department of Civil and Environmental Engineering, University of California, Davis, California

PEER Report 2023/02

Pacific Earthquake Engineering Research Center
Headquarters at the University of California, Berkeley
May 2023
ii
 ABSTRACT
Seismic ground failure potential from liquefaction is generally undertaken in three steps. First, a
susceptibility evaluation determines if the soil in a particular layer is in a condition where
liquefaction triggering could potentially occur. This is followed by a triggering evaluation to
estimate the likelihood of triggering given anticipated seismic demands, environmental conditions
pertaining to the soil layer (e.g., its depth relative to the ground water table), and the soil state. For
soils where triggering can be anticipated, the final step involves assessments of the potential for
ground failure and its impact on infrastructure systems. This workshop was dedicated to the first
of these steps, which often plays a critical role in delineating risk for soil deposits with high fines
contents and clay-silt-sand mixtures of negligible to moderate plasticity. The workshop was hosted
at Oregon State University on September 8-9, 2022 and was attended by 49 participants from the
research, practice, and regulatory communities.

Through pre-workshop polls, extended abstracts, workshop presentations, and workshop breakout
discussions, it was demonstrated that leaders in the liquefaction community do not share a common
understanding of the term “susceptibility” as applied to liquefaction problems. The primary
distinction between alternate views concerns whether environmental conditions and soil state
provide relevant information for a susceptibility evaluation, or if susceptibility is a material
characteristic. For example, a clean, dry, dense sand in a region of low seismicity is very unlikely
to experience triggering of liquefaction and would be considered not susceptible by adherents of a
definition that considers environmental conditions and state. The alternative, and recommended,
definition focusing on material susceptibility would consider the material as susceptible and would
defer consideration of saturation, state, and loading effects to a separate triggering analysis. This
material susceptibility definition has the advantage of maintaining a high degree of independence
between the parameters considered in the susceptibility and triggering phases of the ground failure
analysis.

There exist differences between current methods for assessing material susceptibility – the
databases include varying amount of test data, the materials considered are distinct (from different
regions) and have been tested using different procedures, and the models can be interpreted as
providing different outcomes in some cases. The workshop reached a clear consensus that new
procedures are needed that are developed using a new research approach. The recommended
approach involves assembling a database of information from sites for which in situ test data are
available (borings with samples, CPTs), cyclic test data are available from high-quality specimens,
and a range of index tests are available for important layers. It is not necessary that the sites have
experienced earthquake shaking for which field performance is known, although such information
is of interest where available. A considerable amount of data of this type are available from prior
research studies and detailed geotechnical investigations for project sites by leading geotechnical
consultants. Once assembled and made available, this data would allow for the development of
models to predict the probability of material susceptibility given various independent variables
(e.g., in-situ tests indices, laboratory index parameters) and the epistemic uncertainty of the

iii
predictions. Such studies should be conducted in an open, transparent manner utilizing a shared
database, which is a hallmark of the Next Generation Liquefaction (NGL) project.

Keywords: liquefaction, susceptibility, ground failure, testing, NGL

iv
v
ACKNOWLEDGMENTS AND DISCLAIMER
Execution of this Workshop was made possible by the sponsors (acknowledged below), the
participants and, critically, the Oregon State University graduate student and post-doctoral
assistants including Ms. Mahrooz Abed, Dr. Sandeep Chitta, Dr. Ali Dadashiserej, Dr. Amalesh
Jana, Mr. Logan McGinn, Mr. Nicholas Peterson, Ms. Talenta Pitso, Ms. Jiayao Wang, and Mr.
Eshan Yazdani. The writers also thank the Kearney Dean of Engineering, Scott A. Ashford,
College of Engineering, Oregon State University for support of Workshop expenses.

This Workshop and report were funded by the Pacific Earthquake Engineering Research (PEER)
Center, under Contract No. 1179-NCTRIN and the College of Engineering, Oregon State
University. The opinions, findings, conclusions, and recommendations expressed in this
publication are those of the author(s) and do not necessarily reflect the views of the Pacific
Earthquake Engineering Research (PEER) Center, the Regents of the University of California,
Oregon State University, and the State of Oregon.

vi
vii
CONTENTS
ABSTRACT ................................................................................................................................. III

ACKNOWLEDGMENTS AND DISCLAIMER ..................................................................... VI

CONTENTS.............................................................................................................................. VIII

LIST OF TABLES ...................................................................................................................... XI

LIST OF FIGURES ................................................................................................................. XIII

1 INTRODUCTION, PURPOSE, AND SCOPE ................................................................1

1.1 INTRODUCTION..................................................................................................1
1.2 PURPOSE AND SCOPE OF THIS WORKSHOP .............................................1
1.3 WORKSHOP ORGANIZING COMMITTEE ...................................................2
1.4 ORGANIZATION OF THIS REPORT...............................................................3

2 ORGANIZATION OF THE WORKSHOP ....................................................................5

2.1 APPROACH ...........................................................................................................5
2.2 LIMITATIONS ......................................................................................................8
2.3 LESSONS LEARNED ...........................................................................................8

3 PRE-WORKSHOP POLL ..............................................................................................11

3.1 OVERVIEW .........................................................................................................11
3.2 SUMMARY OF RESPONSES ...........................................................................11
3.3 DISCUSSION .......................................................................................................20

4 SCOPE AND SUMMARY OF DISCUSSIONS ............................................................21

4.1 SESSION 1: State-of-Practice and Limitations ................................................21
4.1.1 Overview ....................................................................................................21
4.1.2 Breakout Questions ....................................................................................22
4.1.3 Summary of Discussions............................................................................22
4.2 SESSION 2: Where Do We Want to Be in 5 TO 10 Years? Model
Development, Resource Needs AND Gaps.........................................................25
4.2.1 Overview ....................................................................................................25

viii
 4.2.2 Breakout Questions ....................................................................................26
4.2.3 Summary of Discussions............................................................................29
4.3 SESSION 3: Opportunities for Synthesizing Laboratory- and Field-
based Observations ..............................................................................................31
4.3.1 Overview ....................................................................................................31
4.3.2 Breakout Questions ....................................................................................33
4.3.3 Summary of Discussions............................................................................33

5 INTERPRETATION OF KEY ISSUES ........................................................................39

5.1 NEED FOR A DEFINITION OF LIQUEFACTION SUSCEPTIBILITY ....39
5.2 HOW SHOULD SUSCEPTIBLITY BE DEFINED .........................................39
5.3 APPLICATION OF CURRENT SUSCEPTIBLITY MODELS .....................42

6 RESEARCH NEEDS .......................................................................................................44

7 SUMMARY AND CONCLUSIONS ..............................................................................48

REFERENCES.............................................................................................................................52

APPENDIX A: LIST OF PARTICIPANTS ..............................................................................56

APPENDIX B: EXTENDED ABSTRACTS..............................................................................58

APPENDIX C: PRE-WORKSHOP POLL RESULTS ..........................................................156

APPENDIX D: WORKSHOP AGENDA ................................................................................182

APPENDIX E: SESSION 1 PRESENTATIONS ....................................................................186

APPENDIX F: SESSION 2 PRESENTATIONS ....................................................................188

APPENDIX G: SESSION 3 PRESENTATIONS ....................................................................190

ix
x
LIST OF TABLES
TABLE 3.1 TABULATED DISTRIBUTION OF 100 POINTS FOR INFLUENCE OF
VARIOUS PARAMETERS ON LIQUEFACTION SUSCEPTIBILITY (POLL
QUESTION #3). ...............................................................................................................13

TABLE 3.2 RESPONSES TO QUESTIONS ABOUT SUSCEPTIBILITY ASSESSMENT

AND CHARACTERIZATION. ......................................................................................16

TABLE 6.1 SUMMARY OF SOME AVAILABLE LABORATORY DATA DERIVED

FROM FIELD OBSERVATIONS. ................................................................................46

TABLE 6.2 SUMMARY OF SOME AVAILABLE DATA DERIVED FROM

LABORATORY INVESTIGATIONS. ..........................................................................47

xi
xii
LIST OF FIGURES
FIGURE 3.1 EMPLOYMENT ROLES OF WORKSHOP ATTENDEES (POLL
QUESTION #1). ...............................................................................................................11

FIGURE 3.2 EXPERIENCE LEVELS OF WORKSHOP ATTENDEES (IN

YEARS; POLL QUESTION #2). ...................................................................................12

FIGURE 3.3 RELIABILITY OF CPT-BASED ASSESSMENT OF

LIQUEFACTION SUSCEPTIBILITY (POLL QUESTION #4). ...............................14

FIGURE 3.4 FINES CONTENT (IN PERCENT) BEYOND WHICH THE FINE-
GRAINED FRACTION CONTROLS SOIL BEHAVIOR (POLL QUESTION #5). ...
....................................................................................................................15

FIGURE 3.5 FINES CONTENT ABOVE WHICH GOOD QUALITY

UNDISTURBED SAMPLES CAN BE OBTAINED (POLL QUESTION #6). .........15

FIGURE 3.6 FACTORS THAT LIQUEFACTION SUSCEPTIBILITY SHOULD

BE RELATED TO (POLL QUESTION #10). ..............................................................17

FIGURE 4.1 PARTICIPANTS’ RESPONSES TO QUESTION #1 DURING THE

SESSION 2 BREAKOUT. ...............................................................................................27

FIGURE 4.2 PARTICIPANTS’ RESPONSES TO QUESTION #2 DURING THE

SESSION 2 BREAKOUT. ...............................................................................................27

FIGURE 4.3 PARTICIPANTS’ RESPONSES TO QUESTION #3 DURING THE

SESSION 2 BREAKOUT. ...............................................................................................28

FIGURE 4.4 PARTICIPANTS’ RESPONSES TO QUESTION #4 DURING THE

SESSION 2 BREAKOUT. ...............................................................................................28

FIGURE 4.5 PARTICIPANTS’ RESPONSES TO QUESTION #5 DURING THE

SESSION 2 BREAKOUT. ...............................................................................................29

xiii
xiv
1 INTRODUCTION, PURPOSE, AND SCOPE

1.1 INTRODUCTION
The typical progression of engineering analysis of soil liquefaction involves three steps:
determination of liquefaction susceptibility, evaluation of liquefaction triggering for one or more
earthquake scenarios, and assessment of the consequences of liquefaction triggering. Although
each of these steps is associated with considerable epistemic uncertainties, the basic framework
for engineering analyses of liquefaction triggering and the consequent deformations or instability
has been established. However, these analyses hinge upon whether a particular stratum is deemed
susceptible to liquefaction, with considerable risk or cost associated with incorrectly assessing
susceptibility. The Next Generation Liquefaction (NGL) Project aims to advance the state of the
art in liquefaction research in part through the provision of consensus-based, probabilistic
methodology for assessment of liquefaction potential and risk, and includes components ranging
from the development of case history and laboratory databases, supporting studies, and model
development. This PEER Workshop, held on September8 - 9, 2022 on the Oregon State University
campus in Corvallis, OR, was conducted as a supporting study under the umbrella of the NGL
Project to seek consensus on liquefaction susceptibility.

Geotechnical engineers have historically divided soil behavior into “sand-like” and “clay-like” due
to their significantly different responses during static loading. This precedent serves the profession
well for sands and clays, but falls short for transitional (or, equivalently, intermediate) soils (clayey
sands, nonplastic sandy silts and silts, and low-plasticity clayey silts) as well as for interlayered
deposits, for which the assessment of liquefaction susceptibility is difficult. Many of the currently
available susceptibility and triggering models are largely based on the interpretation of field
performance data from sites that have or have not exhibited surficial evidence of liquefaction as
typically characterized through observations of sand boils, ground cracks, or large permanent
deformations. While the interpretation of case histories is useful, the NGL Project seeks to evaluate
susceptibility through a separate process in which laboratory data is being carefully parsed to
distinguish “sand-like” from “clay-like” behavior. Nonetheless, significant questions regarding the
linkage between physical (e.g., consistency limits) and correlated (e.g., CPT-based soil behavior
type index, I ) quantities and threshold soil behavior (from “sand-like” to “intermediate”, and from
c

“intermediate” to “clay-like”) exist.

1.2 PURPOSE AND SCOPE OF THIS WORKSHOP

The goals of the PEER Workshop on Liquefaction Susceptibility were to: (1) organize and conduct
a one-and-a-half day long workshop fully-aligned with the ongoing efforts of the Next Generation
Liquefaction (NGL) team that is focused on developing improved data resources and models
related to liquefaction susceptibility and triggering; and (2) prepare a summary report describing

1
the outcomes of the workshop and the specific consensus-based recommendations on the needed
elements of next-generation liquefaction models and the steps needed to produce such models.
Accordingly, the Workshop organizers sought to identify challenges and research opportunities
for improved assessments of liquefaction susceptibility, centered on three broad themes:

1. The current state-of-the practice and its limitations;

2. The linkage between laboratory observations, and field characterization and
response; and,
3. Options for future susceptibility models that could be used, for example, in
conjunction with liquefaction triggering models or hazard mapping.

Vehicles for exploring these three themes included the solicitation of extended abstracts on the
topic of liquefaction susceptibility in response to several prompts, a pre-Workshop poll of
participants, and the Workshop itself, which included a mix of brief presentations, break-out
sessions, and moderated discussion sessions. Key to advancing the objectives of the Workshop,
the organizers sought to draw participants from a broadly diverse set of expert practitioners,
governmental agency representatives, and academicians.

1.3 WORKSHOP ORGANIZING COMMITTEE

The Workshop organizing committee (OC) included faculty members and researchers drawn from
the NGL Project and PEER member universities, and include (in alphabetical order):

Besrat Alemu, Workshop Secretary and graduate research assistant, School of Civil and
Construction Engineering, Oregon State University, Corvallis, OR 97331;

T. Matthew Evans, Professor, School of Civil and Construction Engineering, and

Associate Dean for Faculty and Staff Advancement, College of Engineering, Oregon State
University, Corvallis, OR 97331;

Steven L. Kramer, Professor Emeritus, Department of Civil and Construction

Engineering, University of Washington, Seattle, WA 98195;

Jonathan P. Stewart, Professor, Department of Civil and Construction Engineering,

University of California, Los Angeles, CA 90095;

Armin W. Stuedlein, Workshop Chair and Professor, School of Civil and Construction
Engineering, Oregon State University, Corvallis, OR 97331;

Kristin J. Ulmer, Research Engineer, Geoscience and Engineering Department,

Southwest Research Institute, San Antonio, TX 78238; and,

Katerina Ziotopoulou, Associate Professor, Department of Civil and Environmental

Engineering, University of California, Davis, CA 95616.

2
The OC was initially formed by PEER Workshop grant PIs Armin W. Stuedlein, Jonathan P.
Stewart, and T. Matthew Evans. Discussions by this subset identified the need to more strongly
link the OC to current NGL efforts and diversify membership in the OC. Subsequently Steven
Kramer (NGL), Kristin Ulmer (NGL), and Katerina Ziotopoulou (PEER member faculty) were
invited to help organize the Workshop. Pre-workshop organizational activities were recorded and
filed by Workshop Secretary Besrat Alemu.

1.4 ORGANIZATION OF THIS REPORT

This report is divided into several chapters which provide an overview and summary of the
Workshop. The organization of the Workshop is presented in Chapter 2 and includes the
considerations and planning activities, limitations encountered, and a brief summary of lessons
learned which may be useful to those planning future similar workshops. Chapter 3 summarizes
the Pre-Workshop poll of participant’s views on issues associated with the assessment of
liquefaction susceptibility and which helped guide the conduct of the Workshop. Chapter 4
presents comprehensive descriptions of each of the three Workshop sessions, including
presentations to participants, breakout session questions and participant responses, and
corresponding discussions which concluded each of the sessions.

Chapter 5 presents the organizing committee’s interpretations of key issues raised during the
Workshop discussions. Research needs that were identified by Workshop participants and the
organizing committee are summarized in Chapter 6. Chapter 7 provides concluding remarks and
is followed by a bibliography of references cited. Four appendices accompany this report,
including the list of Workshop participants (Appendix A), extended abstracts addressing various
issues related to liquefaction susceptibility contributed by Workshop participants (Appendix B),
the full pre-Workshop poll and results (Appendix C), and concluding with the detailed Workshop
Agenda (Appendix D). Appendices E, F, and G of this report points the reader to the invited
Workshop presentation slides in the form of electronic supplements which are posted on the PEER
Report webpage.

3
4
2 ORGANIZATION OF THE WORKSHOP

2.1 APPROACH
The workshop was organized to facilitate information gathering and data-informed collaborative
discussions among participants, with the goal of answering three fundamental questions:

• What is the current state-of-practice and what are its limitations?

• Where should the professional practice be in 5 to 10 years, specifically considering
model development and resource needs?
• What opportunities exist for synthesizing laboratory- and field-based observations?

These fundamental questions guided the work of the OC, including the invitation of participants,
attendee pre-Workshop activities, development of the Workshop agenda, and design of breakout
group activities. Each of these aspects is discussed briefly below.

Invitation of Participants. From the outset, the OC worked to ensure balanced participation by
identifying individuals from academia and the public and private sectors who are actively working
in the area of liquefaction susceptibility and potential, specifically related to transitional soils.
Balance in individual expertise was also considered when developing the invitation list, including
laboratory and in-situ testing, numerical simulations, statistical modeling, and regional-scale
assessments. In addition to invitees identified by the OC, an announcement of the Workshop and
solicitation of participants was issued to US academics through the USUCGER email list service.
Applications submitted by interested participants were to include their name, position, and
affiliation, a statement describing their primary interest in participating, a title and summary of the
extended abstract which would be submitted if selected, and a list of three research products that
represent the theme of the Workshop. Four participants were selected by the OC following
submission of applications. Due to travel restrictions associated with the COVID-19 pandemic or
other factors, several participants joined virtually or were unable to participate. The final list of
participants included 33 academics, 10 practitioners, and six state or federal government
employees from 10 U.S. states and five countries outside of the U.S. The final list of participants
is included in Appendix A.

Pre-Workshop Activities. To inform, and better guide discussions during the Workshop,
attendees were asked to complete two main pre-workshop activities: (i) develop an extended
abstract for review by the OC (see Appendix B); and (ii) complete a short anonymous pre-
Workshop poll (see Chapter 3 and Appendix C). After reviewing the extended abstracts, the OC
identified 19 of 31 submissions whose authors would be invited to present their work at the
Workshop. The results from the pre-Workshop poll were synthesized and used in the development
of the Workshop Agenda and the breakout session activities.

5
Extended Abstracts. Solicitation of extended abstracts by the OC served several purposes,
including: (1) a means for participants to communicate their current perceptions and/or research
inquiries on liquefaction susceptibility to other participants prior to the Workshop (through pre-
Workshop distribution of the abstracts), and (2) identification of potential gaps in the Workshop
Agenda during the planning stage.

The instructions for submitting extended abstracts included the request to address three broad
Workshop themes (Section 1.2) by responding to one or more prompts regarding liquefaction
susceptibility and modeling identified by the OC, summarized by:

1. What is the state-of-the-art?

2. What are the consequences of incorrectly assessing liquefaction susceptibility,
and under what conditions are these consequences most acute?
3. What are the relevant geological, material characteristics, in-situ tests, or
modeling processes which drive the challenges associated with the methods
currently in use?
4. What data resources would help improve understanding?
5. How should the next generation of liquefaction susceptibility models be
formulated to advance the state-of-the-art?
6. Have you experienced a case where the determination of susceptibility proved
to be pivotal, and what were the considerations associated with the application
of typical (i.e., state-of-the-practice) susceptibility procedures?
7. Can you describe a case where liquefaction susceptibility was assessed using
methods beyond those typically applied, given the importance of the project
and consequences of liquefaction?

The received extended abstracts addressed these seven prompts from a broad range of perspectives
and experiences. The breadth and the depth of current views of, and research thrusts by,
participants in the area of liquefaction susceptibility are evident in the rich collection of abstracts
found in Appendix B. These abstracts were reviewed by the OC and selected contributors were
invited to present their abstracts during the Workshop.

Workshop Agenda. Each day of the Workshop was organized into a combination of formal
presentations, breakout group activities, and larger group discussions, with a half-day session
dedicated to each of the three fundamental questions listed above. Each session began with a group
of presentations followed by breakout groups and concluding with moderated discussions amongst
all participants. Pacing was deliberate and activities varied to ensure that participants remained
engaged and ample time was provided for “sidebar conversations”, which were often observed to
carry over into the broader group discussions. Appendix D provides the Workshop Agenda, which
is briefly summarized as:

6
 Day 1, Welcome and Opening Remarks
Welcome, Agenda, and Workshop Objectives
Overview and Summary of Pre-Workshop Poll

Day 1, Session 1: State-of-Practice and Limitations

Parts A & B: Presentations
Breakout Session
Report
Discussion/Debriefing

Day 1, Session 2: Where Do We Want to Be in 5 – 10 Years? Model Development,

Resource Needs/Gaps
Parts A & B: Presentations
Breakout Session
Report
Discussion/Debriefing

Day 2, Session 3: Opportunities for Synthesizing Laboratory- and Field-based

Observations, Consensus Recommendations
Parts A & B: Presentations
Breakout Session
Report
Discussion/Debriefing

Day 2, Closing Remarks

Observations on Workshop Discussions
Closing Remarks

Breakout Session Activities. The OC used the extended abstracts selected for presentation and
the results from the anonymous pre-Workshop poll to design the breakout session activities.
Workshop attendees were provided with selected poll results for each session and asked to reflect
on those responses in various ways. In this way, breakout session activities were focused and had
prescribed “deliverables” to facilitate discussions amongst the attendees at the end of each session.
Importantly, all breakout activities were conducted outdoors. This served to change the tone of the
conversations from those conducted in the main venue and provide variety in the flow of the day,
effectively disrupting the relative comfort zones of the attendees and forcing engagement in the
breakout activities. The 37 in-person attendees were randomly divided into six groups for the first
and third breakout sessions. This was a deliberate decision intended to make each group unique,
rather than prescriptive. For example, some groups were academic-heavy while others contained
a higher proportion of practitioners, and some groups were geographically diverse and others more
homogenous. This resulted in diversity of thought and opinion when group discussions were
reported back to the entire audience, further encouraging discussion amongst all attendees. The
second breakout session occurred late in the afternoon, and the OC anticipated relative lower
participant energy levels. Accordingly, this session was programmed to be interactive in nature.
Specific questions were developed using results from the pre-Workshop poll in an effort to probe

7
attendees’ opinions about liquefaction susceptibility. The questions were printed on poster boards
and attendees were asked to throw rocks into cups or complete evolutionary histograms to indicate
their responses to the questions.

Virtual Participants. Given that several countries as well as individuals are still experiencing the
impacts of the pandemic, in addition to individual circumstances, the workshop aimed to provide
hybrid participation. Out of the 49 total participants, 12 were virtual. Virtual participants were able
to view slides projected in the Workshop room, as well as engage in the in-person moderated
discussions, facilitated through a suite of four fixed cameras and microphones. The virtual
participants were formed into an additional seventh group for the breakout activities and were
placed into a virtual breakout room to hold their discussions during the first and third breakout
sessions. For the second breakout session, virtual participants received the questions via an online
poll and provided their replies there.

2.2 LIMITATIONS
The workshop managed to provide a venue for participants to share their opinions and experiences
and to elicit the participants’ thoughts on the questions posed. Participant turnout was good,
although not everyone invited was able to participate in-person or in limited instances, at all. An
increasing amount of anecdotal and scientific evidence has emphasized the limitations of virtual
meetings compared to face-to-face interactions, meetings, and learning experiences. The workshop
was not an exception to this, although the virtual group managed to have vibrant conversations
and efficiently report back to the audience during breakout sessions as well as pose questions after
presentations. While the virtual participation was as carefully as possible considered and
facilitated, the virtual attendees missed out on the in-person technical and networking interactions
and most likely were not as vocal as they would likely have been in person. The presence of some
of the virtual attendees was also understandably not continuous, given either time zone differences
or the fact that virtual events have been found to be challenging to keep up with.

2.3 LESSONS LEARNED

On Attendance and Participation. The workshop succeeded in bringing key researchers and
practitioners together who shared their experiences and voiced their opinions on the three main
questions that the Workshop aimed to resolve. Virtual participants were reasonably vocal and
encountered no accessibility issues (i.e., screen sharing, view of the room, acoustics). One OC
member participated virtually, and a relevant lesson learned in this regard was that this likely
helped with keeping virtual participants more accountable and engaged.

On In-Person Activities. Graduate students from the host institution joined the workshop and
actively assisted with the in-person activities, and execution of the breakout sessions in particular.
The graduate student assistance represented a critical component in ensuring that all participants
and breakout groups knew what was going on at any given moment.

8
On Debriefing Sessions. Breakout Sessions 1 and 3 asked attendees to hold conversations and
provide answers to specific questions (see Sections 4.1 and 4.3). The answers of each group were
shared with the rest of the audience by a selected group representative. In retrospect, one
opportunity for improvement would have been to request the notes of each group.

On the Pre-Workshop Poll. The pre-Workshop poll (Appendix C) was helpful in two respects:
(1) it helped the OC prepare the breakout sessions and guide fruitful conversations on controversial
topics, and (2) it helped attendees to think deeper about certain issues and to be prepared beyond
collecting their thoughts for their extended abstracts (Appendix B). The poll’s anonymity also
helped the attendees provide their honest opinions and remain open to revisiting them later during
the workshop. Anecdotal evidence indicates that at least one participant claimed that their opinion
on liquefaction susceptibility changed following participation in the Workshop. A lesson learnt
was that the poll could have been redistributed after the end of the workshop to probe whether
attendees changed their minds regarding any of the questions.

9
10
3 PRE-WORKSHOP POLL

3.1 OVERVIEW
Prior to the Workshop, a poll (Appendix C) was developed to gain information about the attendees'
professional backgrounds and experience, and their interpretations of a number of issues relating
to the assessment of liquefaction susceptibility. Responses to the poll were useful both in a general
sense and for guiding design of the Workshop Agenda (Appendix D). A brief presentation
synthesizing the responses was provided at the beginning of the Workshop.

The poll consisted of 20 questions relating to current practice, problematic soil conditions,
limitations of current procedures, use of additional (e.g., geologic) information to guide
liquefaction susceptibility assessments, and thoughts on how liquefaction susceptibility
assessment procedures could be improved. Some of the questions lent themselves to statistical or
numerical interpretation and others involved open-ended written responses. This chapter presents
a brief summary of the questions and interpreted general categories, when possible, of the
responses. Not all attendees responded to the poll and not all respondents responded to each
individual question. A complete listing of the responses can be found in Appendix D.

3.2 SUMMARY OF RESPONSES

The first two questions were oriented toward the backgrounds of the attendees. The OC sought
input from the research and practitioner communities and invited participants from both (Figure
3.1). The average level of experience of all respondents was approximately 19 years (Figure 3.2).
25

20
Frequency

15

10

Figure 3.1 Employment roles of Workshop attendees (Poll Question #1).

11
 20

18
Mean = 19.1 years
16

14
Frequency

12

10

0
0 to 5 5 to 10 10 to 15 15 to 20 20 to 25 25 to 30 30 to 35 35 to 40 40+
Years of Experience

Figure 3.2 Experience levels of Workshop attendees (in years; Poll Question #2).

The organizers sought to evaluate the extent to which participants considered liquefaction
susceptibility to be a separate issue from triggering versus a reflection of the likelihood of
triggering. The third poll question probed the relative extents to which material characteristics
(which would not reflect triggering issues) and state/environment characteristics (which would
influence triggering) were considered to be significant with respect to susceptibility. Attendees
were given 100 points to distribute among 17 parameters that could be taken to influence
susceptibility, recognizing that the list was not exhaustive and that a number of the parameters
were correlated with each other. The ranges, means, and coefficients of variations (COV) of the
responses are summarized in Table 3.1. The top 10 vote-getters, in terms of their means, were
plasticity index, PI (by a large margin), degree of saturation, soil behavior type index, relative
density, mineralogy, fines content, depositional environment, age, and (in a three-way tie that
brings the total number of parameters to 11) CPT tip resistance, clay content, and the water content-
to-liquid limit, wc/LL, ratio.

12
Table 3.1 Tabulated distribution of 100 points for influence of various parameters on
liquefaction susceptibility (Poll Question #3).

Parameter Minimum Maximum Mean COV

Plasticity index 0 80 18.7 1.03

Degree of saturation 0 30 9.5 0.84
Soil behavior type index, Ic 0 40 9.4 1.11
Relative density 0 60 9.4 1.27
Mineralogy 0 100 7.6 2.29
Fines content 0 20 6.4 1.01
Depositional environment 0 20 5.7 0.96
Age 0 20 5 1.04
CPT tip resistance 0 20 4.8 1.35
Clay content 0 20 4.8 1.23
wc/LL 0 40 4.8 1.63
Shear wave velocity 0 20 4.6 1.27
SPT resistance 0 20 3.3 1.56
Undrained strength ratio 0 35 2.6 2.49
CPT friction ratio 0 10 2 1.59
Undrained strength 0 8 0.8 2.24
Compression index 0 10 0.6 3.11

Of the parameters in the Top 10 list, five (plasticity index, soil behavior type index, mineralogy,
fines content, and clay content) can be interpreted as inherent material characteristics, five (degree
of saturation, relative density, age, CPT tip resistance, and wc/LL ratio) can be interpreted as
state/environmental characteristics, and one (depositional environment) can be interpreted either
way. The sum of the means of the “inherent material characteristics” category was 43.7 and the
corresponding sum of the “state/environmental” category was 29.8, revealing a general sense that
liquefaction susceptibility was more strongly influenced by material characteristics than
state/environmental characteristics. The increasing use of cone penetration testing for
characterization of liquefaction resistance motivated a question regarding the reliability of CPT-
based assessment of liquefaction susceptibility. The responses, shown graphically in Figure 3.3,
indicate that respondents considered CPT-based assessments of liquefaction susceptibility to be
generally reliable but improved with complementary measurements of fines content and with
cyclic laboratory test data. No respondents rated CPT-based assessments as highly reliable, two as
not generally reliable, and one as poor.

13
 16

14

12

10
Frequency

0
Highly reliable Reliable Reliable in certain Reliable in certain Not generally Poor; do not trust
soils where soils where fines reliable
previous cyclic contents have
test data have also been
been obtained measured

Figure 3.3 Reliability of CPT-based assessment of liquefaction susceptibility (Poll

Question #4).

Susceptibility issues are often centered on the effects of fine-grained soils, either alone or, more
commonly, mixed with coarse-grained soils. In mixed soils with small fractions of fine-grained
particles, the fines are contained within the voids of a skeleton of coarse-grained particles that
resist applied stresses through interparticle forces at their contacts. With large fractions of fine-
grained particles, however, the coarse-grained particles are not in contact with each other and
essentially “float” in a matrix of fine-grained particles that provide the resistance to applied
stresses. Workshop attendees were asked to state their understanding of the fraction of fine-grained
particles at which the transition occurs from coarse-grained control of behavior to fine-grained
control. As shown in Figure 3.4, the majority of the responses indicated that the fine-grained
fraction would control behavior at fines contents above 25 to 35%. Interestingly, none of the
respondents selected 50%, which is the fines content used in the Unified Soil Classification System
to distinguish coarse-grained from fine-grained soils. The respondents’ thresholds are generally
consistent with recent literature on this topic (Cubrinovski and Ishihara 2002; Thevanayagam et
al. 2002; Simpson and Evans 2016; Park and Santamarina 2017).

14
 12

10

8
Frequency

0
10 to 15 15 to 20 20 to 25 25 to 30 30 to 35 35 to 40 40 to 45
Fines Content (%)

Figure 3.4 Fines content (in percent) beyond which the fine-grained fraction controls
soil behavior (Poll Question #5).

With laboratory testing providing a potential means for characterization of liquefaction

susceptibility of transitional soils, the conditions under which good quality “undisturbed” samples
can be obtained is of interest. The attendees were asked to provide their view on the fines content
above which good quality samples could be obtained. Figure 3.5 shows a very wide range of
responses with fines contents ranging from nearly zero to 60%. The average fines content was
approximately 30%.

12

10

8
Frequency

0
0 to 10 10 to 20 20 to 30 30 to 40 40 to 50 50 to 60 60 to 70 70 to 100
Fines Content (%)

Figure 3.5 Fines content above which good quality undisturbed samples can be obtained
(Poll Question #6).

15
Attendees were then asked a series of questions related to their views on how liquefaction
susceptibility should be assessed and characterized. The responses to three of these questions are
presented in Table 3.2. The responses show very clear preferences for probabilistic
characterization of liquefaction susceptibility and for the use of geologic information and advanced
laboratory testing in the assessment of liquefaction susceptibility.

Table 3.2 Responses to questions about susceptibility assessment and characterization.

Poll Question Should susceptibility be characterized in a binary Binary Probabilistic

#11 (yes/no) or probabilistic manner? 2 30

Poll Question Should geologic information be quantified and used Yes No

#12 in liquefaction susceptibility assessment? 26 7

Poll Question Should advanced laboratory testing be used in Yes No

#14 liquefaction susceptibility assessment? 33 0

Poll Question Should NGL develop a database specifically for the Yes No
#18 study of susceptibility? 27 5

Building on the previous questions, attendees were asked to state their opinions on the factors that
susceptibility should be related to. A set of factors ranging from material characteristics, state,
and cyclic behavior was provided and the respondents asked to indicate which of these factors
susceptibility should be related to, with the option of indicating any and/or all that apply. The
intent of this question was to gain insight into whether respondents believed that susceptibility
should be treated as a material characteristic or as being influenced by state and environment. The
responses, illustrated in Figure 3.6, did not provide much clarity with respect to the question’s
objective – many respondents checked multiple boxes with the apparent (and not unreasonable)
thought that all information is helpful.

16
 Degree of saturation

Insitu soil state (density, effective stress) conditions.

Depositional environment (e.g., marine, lacustrine,

fluvial, etc)

Soil behavior characteristics (e.g., parameters related

to shape of hysteresis loops, pore pressure ratio ru,
parameters related to whether undrained strength
normalizes by effective stress).

Soil material characteristics (e.g., fines content,

plasticity, gradation, angularity).

0 10 20 30 40
Frequency

Figure 3.6 Factors that liquefaction susceptibility should be related to (Poll Question
#10).

The preceding questions lent themselves well to a tabular summary of responses. Other questions
were more open-ended and solicited written responses. The remainder of this chapter will
summarize responses to those questions (indicated in bold font) in terms of broad categories of
common responses supplemented in some cases by representative and/or interesting responses.
The responses to all of the poll questions are provided in Appendix C.

How can geologic information be used in liquefaction susceptibility assessment (Poll

Question #13)?

Most common responses included:

• By characterizing depositional environment;

• For characterizing soil variability/continuity;
• For constraint/guidance in probabilistic models; and,
• Only qualitatively.

Representative and/or interesting responses included:

• Evaluation of stratigraphic continuity and variability to assess the extent of liquefiable

soils and liquefaction hazard, and guiding site exploration programs;
• Not sure, other than by crude correlation to nearby tested material. Factors like age are
triggering factors, to me; and,

17
 • Essentially as a “prior”, i.e. before we even poke holes in the ground. This could be
achieved using geospatial models for probability of susceptibility (similar to Zhu et
al.’s global model for liquefaction probability).

How should advanced laboratory testing (e.g., test type: monotonic, cyclic; specimen types:
reconstituted, intact) be used in assessment of liquefaction susceptibility (Question #15)?

Most common responses included:

• Cyclic favored over monotonic; both when possible;

• Intact samples favored over reconstituted; difficulty/expense of sampling noted; and,
• Cyclic simple shear favored over cyclic triaxial.

Representative and/or interesting responses included:

• A combination of monotonic and cyclic testing can be used to demonstrate whether the
soil behavior will be sand like vs clay like. I think this type of testing is applicable for
critical infrastructure projects and research efforts to develop more simplified
relationships for use in general geotechnical engineering practice;
• Get the best samples possible with the soil type of interest. If good intact samples are
not possible, reconstitute in manner that approximates actual depositional processes.
Examine shapes of hysteresis loops, highest ru achieved, tendency for dilation upon
phase transformation, rate of stiffening upon dilation.
• The key is not whether testing would be useful, but how to incentivize/require it on
routine (or not so routine) projects. If very few people are willing to pay for it, this
discussion isn’t very purposeful. The incentive should be worked out before the specific
test details.

How should such tests be interpreted in terms of the potential for susceptibility (Question
#16)?

Most common responses included:

• In terms of reduction of stiffness observed;

• In terms of excess pore pressure level reached in tests; and,
• In terms of shapes of hysteresis loops.
Representative and/or interesting responses included:
• The hysteresis loop at large strains is the single most important piece of information
which can definitively establish the susceptibility to liquefaction;
• Significant stiffness reduction (near zero) under the anticipated seismic loads; and,

18
 • Pore pressure generation, stiffness degradation characteristics with ru, hysteretic
behavior, and post-seismic tests should be evaluated and compared with established
literature.

What information should Next Generation Liquefaction susceptibility models provide to the
user that current models aren’t providing now (Question #17)?

Most common responses included:

• Most comments mentioned need for the probability of susceptibility; and,

• Many triggering-related comments.

Representative and/or interesting responses included:

• Better understanding of transitional soil response and system response of a deposit that
may have partial saturation or interlayering; probabilistic estimates and integration with
triggering models and consequences;
• Three classes: susceptible, transitional, not-susceptible; and,
• A probabilistic liquefaction susceptibility model, material characteristics based, that is
pegged to cyclic hysteresis behavior observed in the laboratory.

How can new technologies be applied to susceptibility assessment (Question #19)?

Most common responses included:

• Use machine learning;

• Improved sampling, in-situ testing; and,
• Use drones in reconnaissance to identify sites with and without manifestation.

Representative and/or interesting responses included:

• New technologies allow for better collection of large datasets and should be used for
transparent dissemination of data. Using new technologies for more efficient collection
and publication of geotechnical in-situ data should also be a priority;
• The development of a cheap, downhole (borehole?) based in-situ cyclic testing
apparatus could improve the ability to establish liquefaction susceptibility;
• Perhaps machine learning could be used to parse out trends/material characteristics
which give rise to certain hysteretic features; and,
• Develop a database with: (1) lab index tests; (2) Ic values from CPT; and (3) advanced
testing. Perform advanced regression, perhaps including machine learning, to relate
susceptibility to different indicators.

19
What do you view as the most significant challenge to advancing liquefaction susceptibility
assessment (Poll Question #20)?

Most common responses included:

• Difficulty and cost;

• Collecting data needed for new models; and,
• Inertia – professional resistance to using new procedures.

Representative and/or interesting responses included:

• Complexity/uniqueness of soil composition;

• Being able to assess susceptibility in a way that is economically feasible for typical
engineering projects;
• Utilizing multiple CPT-based criteria in a way that allows us to capture both aleatory
variabilities and epistemic uncertainties; and,
• Settling on a clear definition of liquefaction susceptibility. Although we might consider
a probabilistic treatment as convenient, this definition ought to be binary for maximum
clarity given that liquefaction can either occur given sufficient loading intensity and
duration, or not.

3.3 DISCUSSION
The pre-Workshop poll was successful in providing an indication of the thoughts and opinions of
workshop participants prior to the Workshop. It effectively pointed out that the term
“susceptibility” meant different things to different respondents. Most respondents appeared to
consider susceptibility to be a function of material characteristics, but many use the term in relation
to triggering considerations. However, the respondents also indicated a strong desire to have a
clear and unambiguous definition of susceptibility.

The poll also confirmed the types of sites where susceptibility considerations are important but
difficult to deal with: sites with transitional soils, sites with interbedded soil layers, and sites with
gravelly soils. Each of these types are encountered frequently for large and small projects and the
judgement of susceptibility can have significant consequences on project cost and schedule.

Responses to the poll indicated a strong belief that susceptibility assessment is fraught with
uncertainty and there is a strong desire to see it characterized in a probabilistic manner. Geologic
information and advanced laboratory testing were also viewed as having significant potential
benefit for the assessment of liquefaction susceptibility. Finally, strong support was expressed for
establishment of a susceptibility database that could be used to develop improved susceptibility
models.

20
4 SCOPE AND SUMMARY OF
DISCUSSIONS

4.1 SESSION 1: STATE-OF-PRACTICE AND LIMITATIONS

4.1.1 Overview

The seven presentations comprising the first session of this workshop aimed to describe the state-
of-practice as well as its limitations. The first presentation by Professor Çetin (Middle East
Technical University) provided an overview of available methods and approaches for predicting
liquefaction susceptibility and was followed by six presentations from practitioners (private
companies and federal agencies) who shared their experiences and solutions from projects that
have featured challenging soils from a liquefaction susceptibility perspective.

Thomas Weaver (US Nuclear Regulatory Commission, NRC) described NRC regulations for
liquefaction evaluation as provided in regulatory guide US NRC 2003 which includes liquefaction
susceptibility criteria. Erik Malvick (Division of Safety of Dams, California Department of Water
Resources) emphasized the complexities that gravels can pose in liquefaction evaluations and
discussed the importance and challenges associated with completing adequate site
characterizations including cyclic lab testing. Dr. Malvick cautioned against the use of statistical
models without accounting for the quality of the data underpinning the models. Pedro Espinosa
(ENGEO, Inc.) presented their experience with the assessment of liquefaction susceptibility of a
natural shoal sand unit beneath the fills at Treasure Island. They found that currently available
simplified procedures were unable to capture the behavior of these shoal materials, which instead
were characterized using material-specific cyclic testing on high-quality samples. This general
approach of performing high-quality material-specific cyclic testing to guide assessments of
susceptibility was also described for various example sites in the Pacific Northwest by Sam Sideras
(Shannon & Wilson, Inc.), Matt Gibson (Clarity Engineering, LLC), and Brice Exley (Haley &
Aldrich, Inc.). In many cases, these projects have produced in situ data (CPT) and laboratory data
that could be shared as part of a broader research exercise. Refer to the corresponding extended
abstracts in Appendix B for additional information on these presentations and other relevant
industry experience-based contributions.

The moderated discussion that followed the group breakout addressed participants’ responses to
four questions. The questions aimed to solicit the participant’s opinions on the need to resolve
issues of terminology and provide updated definitions, and the mechanics (hysteretic behavior)
and/or methodologies (for assessment of susceptibility) that specific terms associated with
susceptibility may or may not imply. The questions are presented below and are followed by a
summary of the corresponding discussion.

21
4.1.2 Breakout Questions

1. Should the profession establish a clear and unambiguous definition of

liquefaction susceptibility?
2. Should susceptibility be clearly distinct from triggering, i.e., should it be a
function of soil characteristics and independent of factors such as soil state
(relative density and effective stress), saturation, loading, and other
environmental variables that influence triggering?
3. Should susceptibility be judged on the ability for triggering to occur at some
state under some intensity of loading, or on the applicability/appropriateness
of available/future liquefaction triggering models?
4. Based on your discussion in this breakout session, propose a clear definition
of liquefaction susceptibility that has the potential to advance the state of
practice.

4.1.3 Summary of Discussions

Workshop participants discussed the questions in a group setting for 60 minutes and reconvened
to present their answers and then discuss.

Question #1: There was a broad consensus to establish a clear and unambiguous definition of
liquefaction susceptibility. Most of the groups provided additional commentary on the need to also
have a clear and unambiguous definition of the term “liquefaction”, which may be used in
connection with strength loss due to pore pressure increase in sands, cyclic softening of clays, flow
liquefaction, and cyclic mobility. Two groups pointed out the fact that while a delineation would
be desirable, it also needs to be viewed within the context and scope of the evaluation or analysis
performed. While simpler approaches might require clear terminology, advanced performance-
based procedures that apply more advanced modeling of soil responses (e.g., time series of excess
pore pressure or shear strain) might not. One group pointed out that ultimately, susceptibility and
triggering criteria guide estimates of strength and subsequently inform the selection of the
appropriate tools to assess the performance.

Question #2: There was also broad consensus that susceptibility and triggering should be distinct
steps in a liquefaction evaluation, although most groups acknowledged that there had been debate
regarding the role of seismic loading in drawing the distinction. Several groups proposed a two-
step approach wherein: (1) compositional (i.e., material) factors are first accounted for and, if these
factors dictate, then (2) environmental factors (soil state, saturation, age, etc.) are taken into
consideration. The role of geologic history (through the overconsolidation ratio, aging and/or
cementation, and fabric) was identified as a gray zone and could be viewed as either compositional
factors or environmental/state factors. It was also noted that a given soil could exhibit different
behaviors under different loading intensities. The need for defining triggering was also raised (e.g.,
is it 100% excess pore pressure ratio or 3% single amplitude shear strain?). The groups that viewed
the need for clearly distinguishing susceptibility from triggering or even sequentially tracking

22
composition, environment, and loading, pointed out that the distinction would be helpful towards
establishing a probabilistic framework, and would accommodate future changes in design loads or
in the perception of hazards overall. One group mentioned that in clearly established Simplified
Methods like the CPT-based liquefaction triggering evaluation models, susceptibility and
triggering are inherently linked, so there may not be a need to parse them out if one is working
with soils that fit within the said methodology, unless the analysis is conducted within a
performance-based design framework. However, data presented by several speakers in Session 3
(i.e., Professors Maurer, Moug, and Stuedlein) suggested that soils exhibiting hysteretic behavior
or field performance that is associated with liquefaction exhibit a wide range in soil behavior type
indices which commonly exceed typical thresholds (e.g., 2.6) selected to inform liquefaction
susceptibility assessments. For performance-based design, separating the probability of
susceptibility, triggering, and consequences is desirable, so as to explicitly account for the
uncertainty in each of the models.

Question #3: The groups expressed diverse views on whether susceptibility of a soil should be
judged based on: (1) the ability of the material to trigger irrespective of its current state and the
anticipated shaking intensity (i.e., a material behavioral criteria), (2) the applicability or
appropriateness of currently available or future liquefaction triggering models, or (3) whether the
soil was likely to trigger given environmental factors (e.g., an unsaturated soil would be considered
non-susceptible) or whether the consequences of liquefaction were likely to be significant (e.g., a
dilatant soil would be unlikely to have large deformations). A point made in connection with option
(3) was that susceptibility should not be judged as a function of loading because in some materials
cyclic stress-strain loops can have very different shapes under strong vs. moderate imposed stress
demands. Moreover, it was argued that if susceptibility is evaluated independent of loading, a
cleaner parsing of uncertainties in liquefaction evaluations is possible (e.g., demand uncertainties
would be independent of susceptibility uncertainties).

Question #4: Not all groups had the time to develop a definition of susceptibility. Three groups
provided the following preliminary definitions, noting the challenge of the task:

• “[Susceptibility is defined as] material composition that leads to behavior that looks
like liquefaction.” This definition was accompanied by the stated need to study what
“looks like liquefaction” through research.
• “[Susceptibility describes] screening based on material characteristics for soil that has
the potential for liquefaction and develop rapid decrease of stiffness and large strain
accumulation.” This group mentioned that they attempted to also mention
consequences of liquefaction in their definition, and that a definition of liquefaction
would be necessary to implement this definition. The group also supported the
development of a holistic framework set within a performance-based design paradigm.
• “[Susceptibility describes the ability to develop] 100% ru with instantaneous zero
stiffness.” This group admitted not reaching a consensus on the definition although they
agreed on being specific about “susceptibility to liquefaction” or “liquefaction
susceptibility” recognizing the breadth of soils exhibiting various [hysteretic]

23
 behaviors and advocating to honor the potential soil behavior. The group also
mentioned they were challenged in decoupling both loading and the types of analyses
from the definition of susceptibility, implying that any definition of susceptibility
should include a proposed type of analysis (or suite of analyses).

The presentation of the groups’ answers was followed by an open discussion, largely focused on
the considerations that should be included in a definition of liquefaction susceptibility (e.g.,
loading, potential consequences, etc.). By the end of this session, it became apparent that a source
of confusion is that for some, “susceptibility” means whether or not a soil can liquefy under certain
conditions, while for others it means whether a soil is likely to liquefy. A summary of the
discussion follows, grouped anachronistically by general theme.

Initially, Professor Idriss (University of California, Davis; retired) emphasized the need to define
liquefaction before defining susceptibility, and later on proposed that one should not worry about
the consequences of a certain behavior but rather about the likelihood for that certain behavior to
occur. Professor Scott Brandenberg (University of California, Los Angeles) made two related
points (1) susceptibility is a fundamental soil response that should be evaluated probabilistically
and not in a binary manner and (2) susceptibility should be distinguished from screening, which is
mainly related to anticipated performance (e.g., a structure on dense sand could pass a screening
criterion even though the soil is susceptible). Professor Jonathan Bray (University of California,
Berkeley) argued that the mechanical behavior of a soil and its consequences should be of most
importance in judging susceptibility. Professor Scott Olson (University of Illinois) also argued that
the soil properties that one seeks to define depend on the consequences one expects and that
consequences need to be tied into susceptibility, particularly because there are different
consequences for soils with different composition and states. Professor Brady Cox (Utah State
University) promoted adopting a more holistic approach wherein one views the whole profile
instead of any one layer, and defining the limit of appropriateness for simplified approaches and
corresponding analyses instead of defining a soil as susceptible. Professor Laurie Baise (Tufts
University) suggested that liquefaction susceptibility be defined within a geologic perspective
which contributes a different scale that becomes important when considering the risk to
infrastructure.

A group of participants made a motion suggested to move past the term “susceptibility”. Professor
Pedro Arduino (University of Washington) proposed defining something different that would be
applicable to a broader range of soils, such as quick clays, removing the word “liquefaction” so
that the “susceptibility” could represent universal application. Professor Ross Boulanger
(University of California, Davis) was in favor of moving past “liquefaction susceptibility criteria”
and focusing on the question of how one will obtain material properties, specifically strength,
cyclic strength, and post-earthquake strength. Towards this end, he proposed the terms “cyclic
strength evaluation criteria” or “cyclic mobility criteria”, with multiple criteria which can
complement one another in the various stages of an analysis.

Dr. Matt Gibson (Clarity Engineering, LLC) proposed to include considerations related to loading
and consequences in the derivation of susceptibility, in order to provide usable advice to clients.

24
Dr. Andrew Makdisi (US Geological Survey) communicated a concern regarding practitioners
with less experience or who deal infrequently with advanced analyses, and their potential reception
of more complex definitions and approaches. Professor Jon Stewart (University of California, Los
Angeles) argued that current practice suffers from ambiguity in the definition of susceptibility and
from the lack of clear and consistent guidelines for its assessment. He suggested that this effort
provide clarity by putting forward a material behavior-based definition, indicating that
consequences and susceptibility need not be coupled because consequences should be addressed
in subsequent stages of a liquefaction risk assessment. Professor Steve Kramer (University of
Washington) discussed that from a practical standpoint, having a base definition of susceptibility
should come with a set of screening criteria against triggering (e.g., saturation, high densities) so
that unnecessary sophisticated analyses can be avoided. In this context, a soil can for example be
called “susceptible, but unlikely to liquefy”.
The Session 1 discussion substantiated the clear need for conducting the Workshop in view of
the lack of consensus on some fundamental aspects of how susceptibility is defined and applied
in projects.

4.2 SESSION 2: WHERE DO WE WANT TO BE IN 5 TO 10 YEARS?

MODEL DEVELOPMENT, RESOURCE NEEDS AND GAPS

4.2.1 Overview

The focus of Session 2 was to identify improvements in assessing liquefaction susceptibility and
its implementation that the profession would ideally achieve in the next 5 to 10 years. Six
presenters provided their perspectives on this prompt.

Professor Shideh Dashti (University of Colorado, Boulder) described her vision to incorporate a
spectrum of soil behaviors into systems level triggering and consequence models, highlighting a
need to separate performance within the profile from surface manifestation and the need for
additional case histories, centrifuge experiments, and numerical simulations. Professor Jonathan
Bray (University of California, Berkeley) shared lessons learned from liquefaction of silty soil
observed in Adapazari, Turkey following the 1999 Kocaeli earthquake and in Christchurch, New
Zealand following the 2010-2011 Canterbury Earthquake Sequence, including his
recommendations to test soil that can be sampled effectively and to consider depositional
environment and soil system response. Professor Dharma Wijewickreme (University of British
Columbia) presented a particle fabric imaging technique intended to understand the shear response
of silts, indicating that X-ray tomography can provide knowledge about the particulate fabric that
could improve our understanding of complex silt behavior.

Professor Laurie Baise (Tufts University) presented global geospatial liquefaction models that
provide regional susceptibility evaluations that could be incorporated into local liquefaction
assessment as a prior and then updated in a Bayesian framework with local geotechnical
information. Dr. Christine Beyzaei (National Institute of Standards and Technology) shared her

25
recommendations for developing susceptibility models to include quantitative site-specific
methods and qualitative regional methods, including the development of easily accessible
susceptibility maps for each state and more widely available interactive databases of borings and
CPTs. Dr. Andrew Makdisi (US Geological Survey) outlined his workflow to incorporate
uncertainty in susceptibility criteria into probabilistic liquefaction hazard analysis and highlighted
research needs including better estimates of uncertainties at all stages (i.e., susceptibility,
triggering, and effects). Refer to the corresponding extended abstracts in Appendix B for additional
information on these presentations and other relevant contributions.

Following the presentations, attendees participated in a breakout activity and a moderated

discussion. During the breakout activity, workshop participants were invited to respond to
questions that were developed as a follow-up to the responses of the pre-workshop poll (Chapter
3; Appendix C). These questions encouraged the participants to consider what liquefaction
susceptibility models should be like in the next 5 to 10 years. The questions are presented below,
and are followed by a summary of the corresponding moderated discussion.

4.2.2 Breakout Questions

1. Do you agree that liquefaction susceptibility should be incorporated into liquefaction

hazard analysis in a probabilistic manner?
2. Do you agree that geologic information should be quantified and incorporated into
liquefaction susceptibility assessment?”
3. How can geological information be used?
4. Separating the idea of liquefaction susceptibility from liquefaction triggering and its
consequences, what factors should future liquefaction susceptibility models consider
(check all that apply)?
5. Which existing and new technologies for the improvement of liquefaction
susceptibility assessment hold promise (check all that apply)?

Question #1: This question is similar to Question #11 from the pre-Workshop poll (Appendix C),
which asked “Do you agree that liquefaction susceptibility should be incorporated into liquefaction
hazard analysis in a probabilistic manner?” One of the purposes of asking this question again was
to see if the presentations and discussions in Sessions 1 and 2 had changed participants
perspectives on this topic. The responses of the participants after the Session 2 presentations are
shown in Figure 4.1. A strong majority of attendees agreed that liquefaction susceptibility should
be incorporated in a probabilistic manner (32 compared to six), similar to views communicated
within the pre-Workshop poll (30 compared to two).

26
 Question 1: Do you agree that liquefaction susceptibility should be
incorporated into liquefaction hazard analaysis in a probabilistic manner?
35
30
25

Frequency
20
15
10
5
0
Yes No

Figure 4.1 Participants’ responses to Question #1 during the Session 2 breakout.

Question #2: Some of the presentations during Session 2 addressed the need to incorporate types
of information other than site-specific geotechnical data, such as geologic and geospatial
information. To assess the participants’ perception of the usefulness of geological data, Question
#2 asked “Do you agree that geologic information should be quantified and incorporated into
liquefaction susceptibility assessment?” This was essentially the same question as Question #12
in the pre-Workshop poll (Appendix C). The participants’ responses during the breakout activity
are documented in Figure 4.2. In both the pre-Workshop poll and during the breakout activity, it
was nearly unanimous that geologic information should be quantified and incorporated.
Question 2: Do you agree that geologic information should be quantified and
incorporated into liquefaction susceptibility assessment?
40
35
30
Frequency

25
20
15
10
5
0
Yes No

Figure 4.2 Participants’ responses to Question #2 during the Session 2 breakout.

Question #3: This question followed up on the second question by asking “How can geological
information be used?” Four separate options were provided with the responses shown in Figure
4.3. This question was asked in an open-ended manner in Question #13 of the pre-Workshop poll
(Appendix C), and the responses to that pre-Workshop question guided the options provided in
Question #3 of the breakout activity. The majority of participants stated that they strongly or very
strongly felt that geologic age and geologic information or depositional environment could be
used. Many also agreed that geologic information is useful for interpolation or extrapolation. The
responses were somewhat more diverse, however, when asked if geologic information was useful
when correlated to field or laboratory test results. Many expressed “very strong” agreement with
this statement, but there was a nearly uniform distribution across the “low,” “moderate,” and
“strong” responses.

27
 Question 3: How can geological information be used?
16
Very Low
14
Low
12
Moderate

Frequency
10 Strong
8 Very Strong
6
4
2
0
Geologic Age? Geologic information / Geologic Info useful for Geologic info useful
depositional interpolation / when correlated to field
environemnt? extrapolation / lab results

Figure 4.3 Participants’ responses to Question #3 during the Session 2 breakout.

Question #4: This question asked the participants: “Separating the idea of liquefaction
susceptibility from liquefaction triggering and its consequences, what factors should future
liquefaction susceptibility models consider (Check all that apply)?” The list of factors given as
optional responses were selected based on responses to Question #3 from the pre-Workshop poll
(Appendix C). The responses generated during the second breakout session are presented in Figure
4.4. Plasticity (e.g., plasticity index, PI) received the strongest support from the participants. Other
factors received moderate support, such as (in order of most to least votes) probability of
susceptibility, CPT data, fines content, cementation, probabilistic range of a given predictor
variable, mineralogy, and degree of saturation. Clay content and clay activity were selected by a
few participants, but generally received little support from most participants, given that these
properties are correlated with PI.

Question 4: What factors should future liquefaction susceptibility models

consider?
40
35
30
Frequency

25
20
15
10
5
0

Figure 4.4 Participants’ responses to Question #4 during the Session 2 breakout.

28
Question #5: This question asked the participants to consider a list of existing and new
technologies for the improvement of liquefaction susceptibility assessment and to identify “Which
do you think hold promise (check all that apply)?” Some of the existing and new technologies that
were proposed in this question were selected from the responses to Question #19 from the pre-
Workshop poll. The participants’ responses are shown in Figure 4.5. The top four technologies
that received the most votes, in order of most to least votes, are: (1) improved sampling methods
to obtain higher (highest) quality samples, (2) laboratory cyclic testing for the quantification of
hysteretic behavior, (3) machine learning/artificial intelligence and/or logistical regression, and (4)
X-ray tomography to quantify soil fabric.

Question 5: Which existing and new technologies for the improvement of

liquefaction susceptibility assessment hold promise (check all that apply)?
30
25
Frequency

20
15
10
5
0

Figure 4.5 Participants’ responses to Question #5 during the Session 2 breakout.

4.2.3 Summary of Discussions

During the moderated discussion, the results of the poll questions during the breakout activity were
displayed on the screen for participants to review and discuss. Those in the minority who
responded “No” to Question #1 regarding the need for incorporating susceptibility in a
probabilistic manner were invited to share their perspective. One participant stated that from their
perspective, the “susceptibility” question is related to choosing an appropriate analysis to estimate
strengths. According to this participant's view, tracking the uncertainties associated with the
estimated strengths is appropriate whereas tracking the uncertainty related to the decision of a
suitable type of analysis is not appropriate. Another participant said that they had originally
answered “Yes” in the pre-Workshop poll, but then answered “No” during the breakout activity,
citing a similar concept: if the “susceptibility” question is a decision about an appropriate analysis
to use (e.g., a semi-empirical simplified approach for soils with sand-like behavior) then perhaps
a probabilistic approach would not be helpful.

29
This led to a discussion about the perceived differences and/or similarities between the Boulanger
and Idriss (2006; B&I06) and Bray and Sancio (2006; B&S06) models. One author from each of
these studies, Professor Ross Boulanger (University of California, Davis) and Professor Jon Bray
(University of California, Berkeley), shared their perspectives. A summary of their statements is
provided below, and an interpretive commentary on this discussion by the OC is provided in
Section 5.3.

• Professor Boulanger: The B&I06 model was intended to define: (1) which soils
could be analyzed using case history-based relationships using similar soils
(“sand-like” soils that can undergo “liquefaction”) vs. (2) which soils could be
confidently sampled and tested in the lab and should be analyzed for undrained
strengths using the suite of methods currently available (“clay-like”). The
transition zone between these two categories should be sampled and carefully
considered to decide which analysis methods to use. He proposed separate
names for the B&I06 and B&S06 criteria, rather than apply the term
“susceptibility” to the models.
• Professor Bray: The B&S06 criteria were intended to be used to identify which
soils have similar stress-strain curves to soils that “liquefy.” The criteria were
intended to assess the engineering response of such soils within a strength-based
approach near buildings interacting with soil.
• Both agree that sand-like soils cannot be confidently sampled (without
expensive and rare procedures like in-situ soil freezing and coring) and tested
in the laboratory, which means that these criteria are necessary to identify which
soils could be analyzed using simplified procedures for sand-like soils. They
both agree that B&I06 and B&S06 models are not fundamentally the same.

One comment in response to this discussion was that these differences appear at face value to be
more qualitative than quantitative, which can be more difficult for practicing engineers to
accommodate in quantitative ways (e.g., weighting each method). Others commented that the lines
between “susceptibility,” “liquefaction,” and “manifestation” were appearing blurry given some
example scenarios (e.g., sites in Adapazari, Turkey) where ground deformation was not observed
in the free field but was observed under buildings which imposed additional static shear, reflecting
a need to carefully distinguish what components of liquefaction evaluations should be considered
in each step.

In response to Questions #2 and #3 of the breakout activity (Figures 4.2 and 4.3), several attendees
generally agreed that additional information such as geological and geospatial data is helpful in
assessing susceptibility. However, the discussion did not yield specific suggestions of what
information to use or what is most impactful in assessing liquefaction susceptibility. There was
relatively little discussion related to the responses to Question #4 regarding the factors that future
susceptibility models should consider. One participant wondered why participants chose fines
content, and how they would describe the fundamental aspect of fines content in liquefaction
susceptibility. In response, another participant cited that the fines content determines whether

30
coarser or finer fractions dictate the soil behavior. The applicability of the factors proposed in
Question #4 were discussed in presentations during Session 3 on the following day (e.g., Professors
Maurer, Moug, and Stuedlein).

A participant asked whether efforts should focus primarily on high-end engineering practice that
is performed somewhat rarely (e.g., using sophisticated cyclic laboratory tests) vs. relatively
routine practice that uses simplified procedures without cyclic laboratory tests. This aspect was
also explored by participants in the moderated discussions of Sessions 1 and 3. Another participant
suggested that it might be helpful to distribute the pre-Workshop poll to a broader audience to
gather feedback from a larger sample size.

There was relatively little discussion related to Question #5 about promising technologies for
improving susceptibility assessments. However, it is interesting to note that Professor
Wijewickreme’s (University of British Columbia) presentation on X-ray tomography may have
generated some new interest, as this technology was listed in the top four factors that future
liquefaction susceptibility models should consider. There was also some detailed discussion during
the moderated breakout for Session 3 about sampling techniques to obtain high quality samples,
which was one of the options provided in response to Question #5.

4.3 SESSION 3: OPPORTUNITIES FOR SYNTHESIZING

LABORATORY- AND FIELD-BASED OBSERVATIONS

4.3.1 Overview

The six presentations comprising the third session of this workshop ranged from CPT-based
interpretations of susceptibility viewed through the lens of critical state soil mechanics, field
observations following earthquakes, linkage of laboratory-based observations of cyclic responses
to CPT- or dynamic, in-situ testing, and linkage of liquefaction susceptibility criteria (or
alternatively, criteria for cyclic strength evaluation) to the estimation of cyclic strength for
assessment of deformations (i.e., consequences).

Professor Scott Olson (University of Illinois) kicked off the third session by presenting a
framework for performing consequence-based susceptibility assessments derived using CPT data,
using the Q approach (Saye et al. 2017). This framework treats soils as susceptible or
insusceptible to specific ground failure mechanisms through consideration of both material
characteristics and soil state. Professor Olsen identified preliminary limiting boundaries for flow
liquefaction, lateral spreading, and post-liquefaction settlement through CPT-based
compressibility-adjusted normalized cone tip resistance. Professor Diane Moug (Portland State
University) leveraged a recently-developed database of cyclic laboratory test data on transitional
soils from sites for which CPTs are also available. She identified the central tendency and
dispersion of cyclic resistance ratio of the silt specimens in the context of normalized cone tip
resistance-based liquefaction triggering and cyclic softening curves. This work clearly identified
that the soils exhibiting sand-like and “transitional” hysteretic behavior can exhibit a very wide

31
range of soil behavior type indices, Ic, which exceeded 2.95 in some cases. Professor Brett Maurer
(University of Washington) used insights gained from the assessments of case history data from
the 2010-11 Canterbury Earthquake Sequence in Christchurch, NZ to address questions related to:
(1) the link between Ic and liquefaction susceptibility, and (2) the role of increased information
(fines content, PI, etc.) on improved susceptibility assessments. Professor Maurer concluded that
material characteristic-based assessments of susceptibility (e.g., using Atterberg limits) may
provide different predictions of susceptibility than the soil behavior type index, and that an Ic = 2.6
provides a reasonable median (i.e., probability of occurrence of 50%) threshold for liquefaction
susceptibility of Christchurch soils. However, he cautioned that the relationship between Ic and
susceptibility is uncertain, and its uncertainty appears to be greater than what is commonly
appreciated.

Professor Ross Boulanger (University of California, Davis) led the next set of three presentations.
He suggested that available liquefaction susceptibility criteria serve different purposes, with the
B&I06 criteria intended to be used to map the outcome of the liquefaction susceptibility
assessment to the appropriate means for determining or estimating cyclic strength. See Chapter
4.2.3 for an in-depth summary of the discussion related to this viewpoint. Professor Scott
Brandenberg (University of California, Los Angeles) presented a series of cyclic direct simple
shear (CDSS) test results of three fine-grained soils with similar PI (8 and 9) but differing amounts
of certain clay minerals and salinity. He showed that: (1) the shapes of stress-strain hysteresis loops
and the rate of strain accumulation with the number of cycles provides the most insight regarding
the potential for sand-like and clay-like behavior, and (2) straight and parallel critical state lines
(CSL) and normal consolidation lines (NCL) provide another means for determining hysteretic
behavior (parallel indicating clay-like, non-parallel indicating sand-like). Professor Armin
Stuedlein (Oregon State University) closed the third session of presentations by providing
examples of the ultimate hysteretic behavior obtained from CDSS tests on a large number silt soils.
This data was used to identify general ranges in both hysteretic metrics and PI for which soil might
be categorized as sand-like and clay-like. Those ranges are in general agreement with the B&S06
criteria, as summarized in Stuedlein et al. (2023), and use of ultimate hysteretic metrics was
justified in view of the large number of cycles of loading associated with the Cascadia Subduction
Zone. Refer to the corresponding extended abstracts in Appendix B for additional information on
these presentations and other relevant contributions.

The group breakout session immediately following these presentations was designed prior to the
Workshop and intended to follow up on themes addressed in the presentations with a focus on data
resources that could be leveraged by the NGL Project. The moderated discussion that followed the
group breakout addressed participants’ responses to six questions. The questions are presented
below and are followed by a summary of the corresponding discussion.

32
4.3.2 Breakout Questions

1. Given your experiences with the pre-Workshop poll and the discussion at this
Workshop, has your opinion of how susceptibility should be defined changed
or evolved, and if so, how?
2. Discuss the definitions of liquefaction susceptibility proposed yesterday. Rank
the definitions in terms of most helpful (i.e., Rank 1) for providing clarity to
the state of practice, with or without modifications as your group sees fit.
3. What currently determinable soil characteristics and behaviors are most suited
to support judgments of liquefaction susceptibility? What aspects of
transitional soil behavior in cyclic tests should be used to judge their
susceptibility? How should susceptibility criteria based on laboratory tests be
formulated?
4. Should one type of parameter (e.g., CPT Ic) always be backed up with
measurements of another (e.g., PI, wc/LL)?
5. If a database were to be developed to support the development of new,
potentially probabilistic, susceptibility models, what information should be
included in that database?
6. What could a probabilistic liquefaction susceptibility model look like (e.g.,
continuous distribution, or discretized into ranges of susceptibility from low,
moderate to high) and how would it be informed? Consider for example your
level of confidence in CPT-based parameters (e.g., Ic, etc.) relative to index
test-based parameters (e.g., PI).

4.3.3 Summary of Discussions

Question #1: This question “…has your opinion of how susceptibility should be defined changed
or evolved, and if so, how?” sought to identify how the discussions in the Workshop may have
resulted in a shift in the perception of liquefaction susceptibility determinations. One of the groups
concluded that whereas prior to the Workshop, they generally felt that susceptibility
determinations should be linked to cyclic strength estimation, they now generally felt that
susceptibility should be decoupled from assessments of cyclic strength. Another group reflected
upon the need for clarity in communication of how a project-specific determination of liquefaction
susceptibility will or has been made. This suggestion stemmed from the nuances in liquefaction
susceptibility assessment identified over the course of the Workshop. Another group concurred
with the need for clarity in communications of susceptibility determination given the critical role
of susceptibility determinations in practice.

Question #2: The second question anticipated participants desire for a clear definition of
liquefaction susceptibility and attempted to identify a potential consensus on a definition. The first
group to report their findings indicated that it was challenging to rank the liquefaction

33
susceptibility definitions, let alone to define liquefaction susceptibility, within the time available
at the Workshop. However, this group felt that developing a clear definition is a critical need. At
least one other group concurred with this sentiment.

Question #3: Participants identified the plasticity index as the most readily determinable soil
characteristic to begin an assessment of liquefaction susceptibility. The water content to liquid
limit ratio was also identified as contributing information to liquefaction susceptibility, although
it was recognized as largely being associated with soil state (e.g., loose or dense of the critical
state) rather than a material composition-type variable. Participants identified the quantification of
metrics of the hysteresis of cyclic test data (e.g., maximum excess pore pressure ratio, minimum
tangent shear modulus, and other variables described by Professors Bray and Stuedlein) as adding
potential value to objective assessments of ultimate cyclic behavior. It was recognized however
that cyclic testing was expensive and largely justifiable for a small percentage of typical projects.
Participants from industry commented on the need for government and regulatory agencies to drive
what is important in liquefaction susceptibility and cyclic strength evaluation criteria.

An example provided during this discussion is that what may be important for a department of
transportation may not be as relevant to an agency responsible for regulating dam safety. The range
in possible design guidelines or codes which address liquefaction susceptibility (and perhaps more
generally, liquefaction assessment), some with respect to performance-based design, were
identified by industry consultants to include:

• ASCE 7 - Minimum Design Loads for Buildings and Other Structures;

• ASCE/SEI 41 - Seismic Evaluation and Retrofit of Existing Buildings;
• ASCE 61 - Seismic Design Standard for Piers and Wharves; and,
• Department of Transportation Geotechnical Design Manuals (e.g., WSDOT).

Question #4: The fourth question intended to identify the need for using interrelated variables
obtained through disparate means (i.e., “should one type of measurement always be accompanied
by another?”). The discussion groups appeared to uniformly agree that assessments of liquefaction
susceptibility should always include disparate measurements of composition (e.g., CPT-based soil
behavior type index and the plasticity index). One of the driving reasons for this sentiment
appeared to be associated with the need to assess the spatial variability of the various strata at a
given site, typically achieved using relatively inexpensive CPTs, the information of which can be
interpreted through the lens of soil indices (e.g., Atterberg limits) obtained from generally more
sparsely distributed soil samples which require relatively expensive drilled boreholes. Participants
pointed to the findings presented earlier in the session, which in aggregate identified strong
regional discrepancies between the CPT-based soil behavior type index and either inferences of
liquefaction from observations of case histories (Professor Maurer) or the results of cyclic
laboratory tests (Professor Moug and Professors Stuedlein and Evans). Specifically, relatively
strong correlation of the soil behavior type index and plasticity index to observed/lack of observed
manifestation of liquefaction in Christchurch, New Zealand, was noted by Professor Maurer
(University of Washington), whereas trends between CPT qc and cyclic resistance by Professor

34
Moug (Portland State University) and between CPT Ic and hysteretic metrics quantifying degrees
of strength and stiffness loss in cyclic direct simple shear tests by Professors Stuedlein and Evans
(Oregon State University) for the transitional soils of the Pacific Northwest could not be
established. In particular, Ic approaching and exceeding 2.95 were identified as exhibiting transient
loss of stiffness and excess pore pressure ratios of 100% in laboratory test results (Professors
Stuedlein and Evans). Finally, participants noted the potential for using multivariate analyses to
reduce the uncertainty in the probabilistic assessments of liquefaction susceptibility which are
beginning to appear in various publications.

Question #5: The fifth question related to the desired data types for a database that could be used
to develop new liquefaction susceptibility models including, for example, the probability of a
material being susceptible to liquefaction. The consensus of the participants is that such a database
ought to include as many types of information as is possible for the exploration and laboratory test
programs associated with a typical project, including but not limited to:

• Water content;
• Atterberg limits;
• Grain size distributions;
• Estimates of, or quantified, mineralogical composition;
• Assessments of age and cementation;
• Oedometric and constant rate-of-strain compression data and indices;
• Monotonic undrained shear strength;
• Cyclic stress-strain curves and post-cyclic undrained shear strength;
• Post-cyclic strength and deformation;
• Penetration resistance (SPT, CPT);
• Shear wave velocity;
• Measurements representing sample quality (e.g., reconsoldiation volumetric
strains, compression ratio indices, shear wave velocity) and quality assessment
designations;
• Latitude and longitude of explorations; and,
• Groundwater depth and/or elevation.

One discussion group emphasized the need for the provision of an entire record of the testing
program, including the protocols used for preparing specimens, conducting the tests, and checks
on saturation where truly undrained tests are conducted (e.g., Skempton’s pore pressure parameter
B). Another discussion group emphasized the need to report the entire dataset including time series
of a particular test, which likely include several stages (consolidation, cyclic testing phase, post-
cyclic testing phase) for completeness and to understand the history of a given specimen. The need
for ensuring the longevity of any such database was identified by participants; that is, steps

35
necessary to ensure that the data doesn’t simply vanish from an online repository or become
obsolete through updates/versioning of the supporting platform would be critical. Likewise, the
need to minimize the complexity of any given database was identified as critical for improving the
longevity of a database. Another participant circled back to the need for providing “flags” to
database entries to clearly distinguish the particular region from which any given data originated
due to the obvious differences in material responses to transient/cyclic loading presented by
speakers earlier in the session.

The idea of limiting access to a database to those who made contributions to such a database was
raised by a participant following reflection on the structure of the New Zealand Geotechnical
Database (http://www.nzgd.org.nz/). This geospatial database of geotechnical exploration data
requires that any user accessing the database contribute new subsurface information developed
following the access to the database. Members of the NGL database thought that such restrictions
may be difficult to impose and manage.

An industry participant commented on the need for the development of guidance on drilling,
sampling, and laboratory testing of transitional soils. This comment elicited a fairly comprehensive
discussion of the drilling, sampling, and handling protocols used alternatively in the post-
Canterbury Earthquake Sequence investigations (Professor Bray; University of California,
Berkeley) and those implemented for the regional study of transitional soils in the Pacific
Northwest (Professor Stuedlein; Oregon State University). Professor Bray suggested that the
protocols used by various research teams be posted to the PEER website as a starting point for
practitioners, and that the protocols used could be compared and compiled in a guidance document
following acquisition of PEER funding. Participants roundly endorsed this potential future
activity.

Question #6: The final question discussed by the workshop participants related to the possible
composition of a future probabilistic liquefaction susceptibility model. One discussion group noted
that it was necessary to account for the uncertainty of a given variable in such a model (e.g., spatial
variability and measurement error in PI) as well as epistemic uncertainty in the model. One
participant suggested a probabilistic liquefaction hazard analysis (PLHA) framework which
treated susceptibility criteria (or models) as a branch in a logic tree, which could be weighted
according to a variety of factors, including the epistemic uncertainty of each susceptibility model,
and incorporated into the larger PLHA outcome. Discussers noted that the intent of a given
probabilistic liquefaction susceptibility model would need to be clearly stated so that an end user
would have an appropriate context for the potential use and interpretation of such a model.

Closing Remarks: Professor Idriss closed out the session and the Workshop with observations on
the presentations and discussions, noting both the critical importance of the Workshop as well as
the struggles of many participants to define liquefaction susceptibility. Reflecting on the need for
professional responsibility for design of a given structure, Professor Idriss identified the
assessment of susceptibility as a significant concern, noting that the charge of the engineer is to
assess the risk of a given facility exhibiting unacceptable performance, if the facility is subjected
to a shaking intensity sufficiently large to result in unacceptable performance. Regarding the

36
question on how to quantify the risk of unacceptable performance, Professor Idriss suggested that
susceptibility should assessed in terms of “What is the susceptibility of the material to a particular
performance?” and “What are the properties we need to describe that material?” At the same time,
Professor Idriss noted the usefulness of the B&I06 criteria as it points to the specific purpose of
the criteria, namely, determining engineering procedures for estimating cyclic resistance, as well
as the B&S06 criteria, which is focused on the displacement potential. Professor Idriss concluded
by noting that the discussions in this session emphasized the need to have this Workshop, but also
the need to continue to meet in such a manner to continue working through the complexities of
liquefaction susceptibility assessment.

37
38
5 INTERPRETATION OF KEY ISSUES

5.1 NEED FOR A DEFINITION OF LIQUEFACTION SUSCEPTIBILITY

Both the pre-workshop poll and the discussions during the workshop indicated a wide range of
interpretations of the term “susceptibility.” While both the plenary and breakout session
discussions established a clear need for such a definition, these discussions did not converge to a
consensus on what it should be.

The interpretations of susceptibility appeared to fall into two primary camps: one based on inherent
material characteristics (independent of soil state and other environment-related characteristics),
and one that included triggering- and/or consequence-related considerations, such as the likelihood
of observing surface manifestation in different forms. One presentation, for example, suggested
the development of different consequence-based susceptibility indices for flow failure, lateral
spreading, and post-earthquake settlement. Some participants also suggested that the term
“susceptibility” not be used and that hazard assessments involving potentially liquefiable soils
move directly to triggering analyses and the models that should be used to evaluate them.

Some of the differences in susceptibility interpretations resulted from the colloquial use of the term
in common language; examples of explaining susceptibility to non-technical people such as
managers or the general public were given. The material-based interpretation of susceptibility is,
in broad terms, based on the possibility of the material liquefying, whereas the consequence-based
interpretation suggests its likelihood. The Oxford dictionary defines susceptibility as “the state or
fact of being likely or liable to be influenced or harmed by a particular thing.” The concept of
likelihood implies a degree of conditionality upon loading, saturation, soil state, etc., in order for
that influence to actually occur.

A few issues generated considerable discussion during the workshop but did not lead to a
consensus among the workshop participants. These issues were discussed by the Workshop
Organizers and authors of this report, with input from select Workshop participants, following the
Workshop. This chapter represents the authors’ interpretation of the discussion of these issues,
which are offered in the spirit of clarifying the issues and identifying paths forward to the
development of improved models for characterization of liquefaction susceptibility. The authors’
interpretations and recommendations provided herein are examples of possible paths forward and
do not represent a consensus decision reached during the workshop.

5.2 HOW SHOULD SUSCEPTIBLITY BE DEFINED

Recognizing the differences in procedures and the limited data upon which current susceptibility
models are based, the majority of workshop participants indicated a desire for future liquefaction
susceptibility models to be expressed in a probabilistic form (30 of 32 in the pre-Workshop Poll,

39
Section 3.2; 32 of 38 in Breakout Session #2, Section 4.2.2). Such a form would also be compatible
with the further development of a performance-based framework for liquefaction hazard
assessment. Since its inception, PEER has developed, refined, and implemented procedures for
performance-based earthquake engineering (PBEE; Deierlein et al 2003). The PEER framework,
formalized through its well-known “triple integral,” is fundamentally probabilistic in nature and
assumes a Markovian independence between ground motion intensity, engineering demand,
physical damage, and loss. In this framework, demands are related to ground motion intensity
measures, damage is related to engineering demand parameters, and loss is related to damage
measures. The framework is modular, and its components can be updated and improved
independently. PEER’s PBEE framework is being used in the development of new performance-
based design and assessment procedures and there is general agreement that future developments
will also be probabilistically-oriented. Workshop participants reported broad support for this
probabilistic approach in the development of future susceptibility models (e.g., responses to
Question #1 from the Session 2 breakout activity, Section 4.2.2).

The advancement of liquefaction hazard assessment procedures will also require the probabilistic
characterization of predictive models and their parameters. Within a PEER-like probabilistic
framework, the introduction of susceptibility into a liquefaction hazard assessment can be made
clearer and more efficient by also assuming Markovian independence between susceptibility,
triggering, and consequences. In order to establish this independence, the quantities used to
establish susceptibility should not be related to those that influence triggering. This format is
inconsistent with the consequence-based interpretation of susceptibility since it mixes elements of
material, loading, and response characteristics. The material-based interpretation, on the other
hand, characterizes susceptibility in terms of inherent material characteristics that are independent
of density, saturation, effective stress, and other environment-related characteristics that influence
triggering and consequences. It can, however, lead to some counter-intuitive circumstances; for
example, a very dense, well-graded, dry sandy gravel would be considered susceptible to
liquefaction under this definition even though it would be virtually impossible for it to liquefy in
that condition. However, it would not liquefy because it would not trigger – not because the
material itself was inherently non-susceptible (i.e., the same particles could be rearranged in a
different environment to a condition in which it could liquefy).

Environmental variables that may be commonly thought to be linked to or control liquefaction

susceptibility, but which are not material-type variables, are better served to be evaluated during
the liquefaction triggering assessment. Such variables include the degree of saturation, age,
cementation, effective stress magnitude, and relative density (i.e., state). These environmental
variables are well-suited to screen various layers for liquefaction as part of the liquefaction
triggering assessment. Clear examples of the effects of such variables on cyclic resistance come in
the form of available adjustments to cyclic resistance for partial saturation and geologic age.
During the first breakout session, representatives from two discussion groups discussed their view
of susceptibility assessment as a two-step approach wherein one would first evaluate the
compositional factors, and, if those dictate it, then assess environmental factors such as state and
saturation. Factors like loading history and overconsolidation ratio were also found to challenge

40
the definitions as they could be interpreted by some as both material and state variables. However,
we find such characteristics better suited for screening certain layers during the triggering
evaluation stage of analysis.

The pre-Workshop poll identified inherent material-specific characteristics (e.g., plasticity index,
fines content, CPT-based soil behavior type, mineralogy, clay content) as being more influential
with respect to susceptibility than environment-specific characteristics that include triggering- and
consequence-related factors. Throughout the workshop discussions, this trend appeared to persist.
However, as noted above, several participants expressed their view that susceptibility should also
be related to state and/or loading. For example, according to this view, a tightly-packed, dilative
material either would not liquefy, or even if it did experience high excess pore pressures under
extreme loading, large deformations would be unlikely. However, this perspective on
susceptibility can cause confusion in practice, because state is fundamentally associated with
liquefaction triggering assessments.

We support the material-based interpretation of liquefaction susceptibility as being more

fundamentally appropriate for the advancement of practice in liquefaction hazard assessment.
Some clarity in the application of this material-based definition of susceptibility may be gained by
referring to it as “material susceptibility” or possibly “inherent susceptibility” in order to
distinguish its use in technical analyses from the more colloquial likelihood-oriented use in
common communication. It is clear from the workshop discussions that great care should be taken
in the use of technical terminology and that definitions be made explicit in both technical
communication and the description of potential hazards to non-technical audiences. A finding of
material susceptibility does not indicate whether or not significant liquefaction hazards are likely
to occur at a particular site (i.e., that liquefaction is likely to be triggered), it merely indicates that
the potential for their occurrence should be evaluated using appropriate laboratory tests, models,
and analyses. Similarly, a finding of non-susceptibility does not mean that other hazards, such as
cyclic softening, do not exist and should not be carefully evaluated.

The workshop participants broadly agreed that in its current form, the CPT-based Soil Behavior
Type Index, Ic, represents a useful parameter for characterization of liquefaction susceptibility, but
one for which there can be appreciable uncertainty when applied to a particular soil. Evidence was
presented for regions such as Christchurch, New Zealand, and the Pacific Northwest, USA, that Ic
is a parameter that varies with the mineralogy, depositional environment, and post-depositional
geological processes which give rise to unique soil fabric. This parameter is also influenced by
anthropogenic processes such as ground improvement, as a result of changes in the lateral stress
state and thus Ic (e.g., Nguyen et al. 2014). Further strong statistical evidence appears to point to
Ic = 2.6 as an approximate median of a distribution of Ic that can separate potential sand-like and
clay-like behavior of different soils based on observations of liquefaction manifestation. This
implies that a non-negligible percentage of soil materials characterized with an Ic > 2.6 could be
falsely flagged as non-susceptible to liquefaction (and vice versa), as was documented in
presentations during the workshop. It is emphasized here that Ic represents both material and state
characteristics, and that the deterministic use of a Ic alone could result in inaccurate determinations
of susceptibility. Participants endorsed the discontinued use of Ic = 2.6 as a strict cut-off between

41
sand-like and clay-like behavior. This need to avoid the use of strict cutoffs is not unique to Ic but
would likely apply to any soil parameter used in liquefaction susceptibility assessments.

5.3 APPLICATION OF CURRENT SUSCEPTIBLITY MODELS

Two widely used susceptibility models, i.e., those of Boulanger and Idriss (2006) and Bray and
Sancio (2006), referred to hereafter as B&I06 and B&S06, respectively, were the subject of
extended discussion in the Workshop. These models were developed by different researchers based
on a varying number of tests on different soil materials and employed somewhat different methods
to interpret test results. As a result, the two models have broad areas of agreement in their
characterization of soil behavior, but also areas where their results can be interpreted as
inconsistent. The discussion centered on the B&I06 and B&S06 models and whether their findings
are in fact different in terms of how they should be applied in practice.

Professor Ross Boulanger (University of California, Davis) explained that the B&I06 model had
the objective of guiding the choice of analysis method and stated his preference for it to be
interpreted as a “cyclic strength evaluation procedure criterion,” with the following logic:

• A determination of sand-like soil behavior would indicate that semi-empirical

procedures for liquefaction triggering (which are dominated by data from granular
soil sites and for which sampling in an intact state is relatively impractical) should
be applied, and as appropriate, undrained monotonic, cyclic, and residual shear
strengths would be obtained via correlations.
• A determination of clay-like soil behavior would indicate that procedures for
strength characterization for clay, which often include sampling, laboratory testing,
and consideration of stress history effects, should be applied with the goal of
estimating appropriate undrained monotonic, cyclic, and remolded shear strengths
of the material. Ground failure assessment in this domain is referred to as cyclic
softening and is principally related to strength and stiffness degradation from
undrained cyclic loading.

Professor Jonathan Bray (University of California, Berkeley) stated the B&S06 model was
developed under a different framework related to the engineering response and consequences of
the material being cyclically loaded. The B&S06 model examined the response of slightly plastic
silt soils and concluded that, since their response (i.e., excess pore water pressure ratio, ru > 90%
and similar ‘banana-shaped’ cyclic shear stress vs. cyclic shear strain loops) were similar to those
of medium dense to dense sands whose response is termed liquefaction, their response should also
be termed liquefaction. The B&S06 model did not specify the method to evaluate the engineering
response and consequences of these materials.

Both Boulanger and Idriss (2006; 2008) and Bray and Sancio (2006; 2008) have recommended
that soils that can be reliably sampled should be sampled and tested to refine the assessment of
susceptibility as well as triggering, and consequences of liquefaction.

42
Following the Workshop, the report writers discussed whether the B&I06 and B&S06 criteria in
fact represent different objectives given epistemic uncertainties associated with the models, how
the typical practitioner may use the differing criteria, and how they relate to the development of
future susceptibility models. These discussions led to the following conclusions:

• After considerable discussion, the interpretation and applicability of available

liquefaction susceptibility criteria appears to hinge entirely on the definition of the
terms “liquefaction” and “sand-like” and “cyclic softening” and “clay-like”. The
report writers therefore identify the critical need for the profession to accompany
any liquefaction susceptibility model with a clear statement of how liquefaction
susceptibility is defined (i.e., as a material behavior vs. as part of the strength or
consequence determination). We also emphasize points clearly stated by Boulanger
and Idriss (2006; 2008) and Bray and Sancio (2006; 2008) that the determination
of “non-liquefiable” or “clay-like” should precipitate an assessment of cyclic
softening potential and its consequences.

• Both models make use of the results of laboratory tests to judge the behavior of
cyclically loaded soils, and significantly more test data has become available since
the models were published. Epistemic uncertainties (e.g., differences in the PI
ranges specified by the B&I06 and B&S06 models) may be reduced by continued
experimental laboratory research, case history interpretation, and development of
refined deterministic and probabilistic models, the latter of which was clearly
identified as a need in the Workshop.

• Regarding application in practice, the original papers and closures, as well as

discussions in the Workshop, describe how the developers of the B&I06 and
B&S06 models view their intended use. Professors Boulanger and Bray present
their views on their models in their extended abstracts (see Appendix B). However,
practitioners often interpret the end result of both models as indicating whether or
not a material is susceptible to liquefaction. It is incumbent upon geotechnical
engineers to recognize the differing intent, applicability, and limitations of these
models in their use and interpretation in practice.

The existing susceptibility models have been widely and beneficially used since their publication
some 17 years ago, and differences in their intents, terminologies, and use in practice have helped
illustrate the complexity of cyclically loaded soil behavior over important ranges of soil
characteristics. The models are supported by experimental data and the expert interpretation of
that data by their developers. Differences in the models can largely be attributed to differences in
the data they are based upon, and differences in the developers’ interpretation of that data. Future
liquefaction susceptibility models should supplement that data with available data generated since
their publication, new data that fills gaps in the current state of knowledge, and should define their
terms and intended model application in practice carefully and explicitly.

43
6 RESEARCH NEEDS

There was general consensus among Workshop attendees that improved procedures to evaluate
liquefaction susceptibility are needed. Moreover, these procedures should be probabilistic, in that
they should capture variability in the underlying data, and should be formulated in a manner that
facilitates evaluation of epistemic uncertainties. Here we briefly describe the vision for the
research that would lead to the development of such procedures, the research tasks (or scope) that
would support the development of such procedures, and the ways in which epistemic uncertainty
could be inferred from the results.

Vision: Our vision for next-generation susceptibility models is that they should predict whether a
material could exhibit fundamental soil behavior that is characteristic of granular media. This
includes a rapid and substantial reduction of stiffness, and potentially strength, associated with the
development of high pore pressure, and the potential development of surface manifestation and/or
permanent vertical and/or horizontal deformations. A concise phrasing for such characteristics is
material susceptibility, which emphasizes the soil behavior and thus may be a preferred term to
“susceptibility”. As such, the models would not reflect environmental conditions (mainly
saturation) or information about soil state (such as water content or relative density). The
susceptibility models should consider alternate predictor variables, reflecting different levels of
information available for different applications. The most basic of these parameters would be
routinely available from most professional geotechnical reports, such as Ic or PI. More advanced
metrics can and should also be considered. The models should be probabilistic (i.e., the outcome
of the model is a probability of material susceptibility) that would reflect aleatory variability.
Different levels of aleatory variability could be anticipated when different predictor variables are
used.

Scope: The scope of the research that would realize this vision mainly involves the development
of a database targeted at susceptibility studies. The current NGL database (Brandenberg et al.
2020) was originally developed for field case histories of manifestations (or lack thereof) of
liquefaction from past earthquakes. That database structure is not directly applicable to this
problem; however, the structure of the database has recently been adapted for laboratory data
(Hudson et al. 2022), without corresponding earthquake information or field performance. Such
laboratory test data can be used to study many different aspects of liquefaction problems, including
the present focus on material susceptibility. The contents of the database were discussed
extensively in the Workshop, and are envisioned to include the following:

1. Each entry would be from a given site. The sites would ideally have wide
geographic distribution to capture different geological characteristics;
2. Each site should have in-situ CPTs that are essentially co-located with borings
with samples. The horizontal separation distance between the CPT sounding
and boring should be greater than 1.5 m to limit the effects of drilling on the

44
 sounding or of the sounding on the sampling, respectively, and smaller than 3
meters to minimize differences in material characteristics due to inherent soil
variability;
3. Cyclic tests for layers of interest in the profile should have been performed,
with the results provided digitally in the database. The results should allow for
plots of stress-strain hysteresis and stress paths;
4. Metadata related to the tests should be provided, including method of
sampling, method of sample preparation, pre-test consolidation of specimens
and B-values (CTX) or normal strain time histories (CDSS), as applicable,
and the results of tests to assess sample disturbance (Chapter 4.3);
5. Index tests performed on material from the samples used in (3) should be
provided, including Atterberg limits, water content, and gradation curves.
These soil indices should be obtained from the cyclic test specimen itself
when feasible, and from cuttings or other test specimens within close
proximity to the cyclic test specimen; and,
6. Where available, results of monotonic undrained tests should be provided.
Ideally, a suite of such tests should be provided to allow strength
normalization to be evaluated using established procedures (e.g., Ladd 1991).

A particularly encouraging outcome of the Workshop is that a large amount of data of this type is
available derived from field observation studies, research-based laboratory investigations, and
design-based laboratory investigations from investigations by professional engineers (Tables 6.1
and 6.2). While further investigations are always welcome, it is likely not mandatory for further
data development to occur for the envisioned study to be successful. Rather, what is needed is to
assemble and archive the data into a perpetually available and usable form.

Once the database is developed, a research team would interpret the cyclic test results for each
specimen to assign a numerical or qualitative indicator of soil behavior (e.g., sand-like,
intermediate, clay-like). Alternate variables would then be examined through regression (including
AI methods) to investigate their predictive power. Ineffective parameters would have large
variability (i.e., a statistical distribution of a model characterized by a large standard deviation)
whereas effective parameters would have reduced variability.

In the development of predictive models from the database, several types of epistemic uncertainties
could be considered:

1. Models in which a region flag is included (model provides predictions specific

to that region, such as Christchurch) versus models in which all global data
are grouped. The latter (grouped) data would have larger and quantifiable
uncertainties;

45
 2. Different assignments of susceptibility from cyclic tests could be made by
different investigators (e.g., see Chapters 4.3.3 and 5.3). Differences between
the resulting models is a quantifiable form of epistemic uncertainty; and,
3. Different methods of regression could be used and different forms of
probabilistic models could be provided.

Table 6.1 Summary of some available laboratory data derived from field observations.

Basis Event and Location Laboratory Testing Reference

Field Observations 1989 Loma Prieta, Moss Index, consolidation, cyclic Boulanger et al. (1998)
Landing, USA triaxial tests
Field Observations 1994 Northridge, Los Index and laboratory vane Holzer et al. (1999)
Angeles, USA shear tests
Field Observations 1999 Kocaeli, Adapazari, Index, cyclic triaxial, and Bray and Sancio (2006)
Turkey direct simple shear tests
Field Observations 1999 Chi-Chi, Wufeng, Index, consolidation, Chu et al. (2008)
Taiwan triaxial, cyclic triaxial tests
Field Observations 2010-11 Canterbury Index, cyclic triaxial tests Beyzaei et al. (2018)
Earthquake Sequence,
Christchurch, New Zealand
Field Observations Controlled blasting Index, consolidation, Jana and Stuedlein (2022)
monotonic direct simple
shear, cyclic direct simple
shear
Field Observations Vibroseis shaking, Port of Index, consolidation, Dadashiserej et al. (2022)
Longview, USA monotonic direct simple
shear, resonant column-
torsional shear, cyclic direct
simple shear
Field Observations Vibroseis shaking and Index, consolidation, Jana et al. (2023)
controlled blasting, Port of monotonic direct simple
Longview, USA shear, resonant column-
torsional shear, cyclic direct
simple shear

46
Table 6.2 Summary of some available data derived from laboratory investigations.

Basis Laboratory Testing Reference

Research Index, consolidation, cyclic direct Sanin and Wijewickreme (2006)
simple shear
Research Index, consolidation, monotonic Dahl et al. (2014)
direct simple shear cyclic direct
simple shear

Research Index, consolidation, monotonic Wijewickreme et al. (2019)

direct simple shear cyclic direct
simple shear

Research Index, consolidation, monotonic Jana and Stuedlein (2022)

direct simple shear cyclic direct
simple shear

Research Index, consolidation, monotonic Stuedlein et al. (2023)

direct simple shear cyclic direct
simple shear

Research / Design Investigations Index, cyclic triaxial, cyclic direct Dickenson et al. (2022)
simple shear
Design Investigations Index, cyclic direct simple shear Presented by P. Espinosa at
Workshop (Appendix B)
Design Investigations Index, cyclic direct simple shear Presented by S. Sideras at Workshop
(Appendix B)
Design Investigations Index, cyclic direct simple shear Presented by D. Moug at Workshop
(Appendix B)
Design Investigations Index, consolidation, triaxial, Presented by M. Gibson at Workshop
monotonic and cyclic direct simple (Appendix B)
shear
Design Investigations Index, cyclic direct simple shear Presented by B. Exley at Workshop
(Appendix B)

47
7 SUMMARY AND CONCLUSIONS

The typical progression of engineering analysis of soil liquefaction involves three steps:
determination of liquefaction susceptibility, evaluation of liquefaction triggering for one or more
earthquake scenarios, and the assessment of the consequences of liquefaction triggering. Although
each of these steps is associated with considerable epistemic uncertainties, the basic framework
for engineering analyses of liquefaction triggering and the consequent deformations or instability
have been established. However, these analyses hinge upon whether a particular stratum is deemed
susceptible to liquefaction, with considerable risk or cost associated with incorrectly assessing
susceptibility. The uncertainty associated with the determination of susceptibility represents a
significant contribution to the overall uncertainty associated with the assessment of ground failure
risk.

The objectives of this PEER Workshop on liquefaction susceptibility were to identify the means
to improve data resources and models related to liquefaction susceptibility to reduce uncertainties
in the assessment of liquefaction susceptibility (aligned with the goals of the Next Generation
Liquefaction project) and to summarize the outcomes of the workshop discussions on the needed
elements of, and steps needed to develop, Next-Generation Liquefaction models. Workshop
organizers sought to identify challenges and research opportunities for improved assessments of
liquefaction susceptibility, centered on three broad themes:

1. The current state-of-the practice and its limitations;

2. The linkage between laboratory observations, and field characterization and
response; and,
3. Options for future susceptibility models that could be used, for example, in
conjunction with liquefaction triggering models or hazard mapping.

Workshop participants were invited to submit extended abstracts on the topic of liquefaction
susceptibility in response to the following prompts:

1. What is the state-of-the-art?

2. What are the consequences of incorrectly assessing liquefaction susceptibility,
and under what conditions are these consequences most acute?
3. What are the relevant geological, material characteristics, in-situ tests, or
modeling processes which drive the challenges associated with the methods
currently in use?
4. What data resources would help improve understanding?
5. How should the next generation of liquefaction susceptibility models be
formulated to advance the state-of-the-art?

48
 6. Have you experienced a case where the determination of susceptibility proved
to be pivotal, and what were the considerations associated with the application
of typical (i.e., state-of-the-practice) susceptibility procedures?
7. Can you describe a case where liquefaction susceptibility was assessed using
methods beyond those typically applied, given the importance of the project
and consequences of liquefaction?

The breadth and depth of current perspectives on liquefaction susceptibility and related concerns
by Workshop participants are clearly demonstrated in the submitted abstracts included within this
report (Appendix B).

A pre-Workshop poll (Appendix C) with questions drawn in part from the information provided
by participants in their extended abstracts served to help refine the Workshop Agenda (Appendix
D) and focused discussion points. The Workshop discussions (Session 1; Section 4.1) clearly
identified that the term “susceptibility” could mean a variety of different things to different
participants. Whereas most participants considered susceptibility to be a function of material
characteristics alone, many linked the term and act of assessing susceptibility to triggering
evaluations. The participants overwhelmingly concurred on the need to have a clear and
unambiguous definition of liquefaction susceptibility. Several groups of participants identified
preliminary forms of such a definition, however, the Workshop participants were unable to
converge on a shared definition in the time available.

Participants clearly indicated a powerful belief that susceptibility assessment is fraught with
uncertainty and there is a clear desire to see it characterized in a probabilistic manner (Session 3;
Section 4.3). Information derived from geologic investigations and cyclic laboratory tests were
also viewed as having significant potential benefit in the assessment of liquefaction susceptibility
(Session 2; Section 4.2). Strong support was also expressed for establishment of a susceptibility
database that could be used to develop improved susceptibility models (Session 3; Section 4.3).

The Workshop Organizing Committee (OC) synthesized three key issues identified over the course
of the discussions and provided their opinions thereof in Chapter 5, including the: (1) need for a
definition of liquefaction susceptibility, (2) means by which such a definition should be developed,
and (3) the differences between current susceptibility models. The OC suggests that the
interpretation and applicability of available liquefaction susceptibility criteria is driven by the
definition of the terms “liquefaction” and “sand-like” and “cyclic softening” and “clay-like”.
Epistemic uncertainties related to differing datasets underpinning the criteria can be treated
through additional laboratory testing and field observations. The OC emphasizes that the
geotechnical engineer is responsible for recognizing the differing intent, applicability, and
limitations of these models in their use and interpretation in practice.

A meaningful outcome of the Workshop is recognizing that a large amount of data that could
populate a susceptibility database is available from both researchers and practitioners. The sources
identified range from field observation including post-earthquake reconnaissance studies, research
investigations deploying in-situ dynamic test methods, as well as research- and design-based

49
laboratory investigations (Chapter 6). Workshop participants appeared to agree that the assembly
and archival of the available data in a perpetual and usable form is both viable and necessary.

The Workshop facilitated deep, meaningful, and vigorous discussions on a critical component
comprising the overall set of steps to assess the risk of seismically-induced ground failure: the
assessment of liquefaction susceptibility. Although a broad consensus on how liquefaction
susceptibility should be defined was not achieved, Workshop participants expressed that the event
added significant value to their understanding of liquefaction susceptibility, that the discussions
served to crystalize various perspectives, and identified clear data resource needs and model
development goals.

50
51
REFERENCES
Boulanger, R. W., and Idriss, I. M. (2006). “Liquefaction susceptibility criteria for silts and
clays.” Journal of Geotechnical and Geoenvironmental Engineering, 132(11), 1413-1426.
http://dx.doi.org/10.1061/(ASCE)1090-0241(2006)132:11(1413)

Boulanger, R.W. & Idriss, I. M. (2008). Closure to “Liquefaction Susceptibility Criteria for Silts
and Clays.” Journal of Geotechnical and Geoenvironmental Engineering, 134(7), 1027-1028.
https://doi.org/10.1061/(ASCE)1090-0241(2008)134:7(1027)

Brandenberg, S.J., Zimmaro, P., Stewart, J.P., Kwak, D.Y., Franke, K.W., Moss, R.E., Çetin,
K.Ö., Can, G., Ilgac, M., Stamatakos, J. and Weaver, T. (2020). “Next-generation liquefaction
database.” Earthquake Spectra, 36(2), 939-959. https://doi.org/10.1177/8755293020902

Bray, J. D., and Sancio, R.B. (2006). “Assessment of the liquefaction susceptibility of fine-
grained soils.” Journal of Geotechnical and Geoenvironmental Engineering, 132(9), 1165-1177.
http://dx.doi.org/10.1061/(ASCE)1090-0241(2006)132:9(1165)

Bray, J. D., & Sancio, R. B. (2008). Closure to “Assessment of the Liquefaction Susceptibility of
Fine-Grained Soils” Journal of Geotechnical and Geoenvironmental Engineering, 134(7), 1031-
1034. https://doi.org/10.1061/(ASCE)1090-0241(2008)134:7(1031)

Cornell, C.A. (1968). “Engineering seismic risk analysis,” Bull. Seism. Soc. Am., 58, 1583-1606.
https://doi.org/10.1785/BSSA0580051583

Cubrinovski, M., and Ishihara, K. (2002). “Maximum and minimum void ratio characteristics of
sands.” Soils and Foundations, 42(6), 65-78. https://doi.org/10.3208/sandf.42.6_65

Dadashiserej, A., Jana, A., Stuedlein, A.W., Evans, T.M., Zhang, B., Xu, Z., Stokoe, I.I. and
Cox, B.R. (2022). “In-situ and Laboratory Cyclic Response of an Alluvial Plastic Silt Deposit.”
Proc., 20th International Conference on Soil Mechanics and Geotechnical Engineering, Sydney,
Australia, 2021. 7 pp.

Dahl, K. R., DeJong, J. T., Boulanger, R. W., Pyke, R., and Wahl, D. (2014). “Characterization
of an alluvial silt and clay deposit for monotonic, cyclic, and post-cyclic behavior.” Canadian
Geotechnical Journal, 51(4), 432-440. https://doi.org/10.1139/cgj-2013-0057

Deierlein, G.G., Krawinkler H., and Cornell, C.A. (2003). “A framework for performance-based
earthquake engineering.” Proceedings, 2003 Pacific Conference on Earthquake Engineering,
Christchurch, New Zealand.

Dickenson, S. E., Khosravifar, A., Beaty, M. H., Bock, J., Moug, D., Schlechter, S. M., and Six,
J. (2022). “Cyclic and Post-Cyclic Behavior of Silt-Rich, Transitional Soils of the Pacific
Northwest; A Database for Geoprofessionals in Practice and Research,” Data report prepared for

52
the Oregon Department of Transportation, Bridge Engineering Section, Salem, Oregon, by New
Albion Geotechnical, Inc., Reno, NV

Hudson, K.S., Zimmaro, P., Ulmer, K., Carlton, B., Stuedlein, A., Jana, A., Dadashiserej, A.,
Brandenberg, S.J., Stamatakos, J., Kramer, S.L. and Stewart, J.P. (2022). “Laboratory
component of next-generation liquefaction project database.” Proceedings, PBDIV, 4th
Conference on Performance-based Design in Earthquake Geotechnical Engineering, Beijing,
Springer, 1865-1874.

Jana, A., Dadashiserej, A., Zhang, B., Stuedlein, A. W., Evans, T. M., Stokoe II, K.H., and Cox,
B.R. (2023). “Multi-directional VibroSeis Mobile Shaking and Controlled Blasting to Determine
the In-Situ Dynamic Nonlinear Inelastic Response of a Low Plasticity Silt Deposit.” Journal of
Geotechnical and Geoenvironmental Engineering, 149(3), 04023006.
https://ascelibrary.org/doi/abs/10.1061/(ASCE)GT.1943-5606.0002924

Jana, A. and Stuedlein, A. W. (2021). “Monotonic, Cyclic and Post-Cyclic Response of an

Alluvial Plastic Silt Deposit.” Journal of Geotechnical and Geoenvironmental Engineering,
147(3): 04020174. https://doi.org/10.1061/(ASCE)GT.1943-5606.0002462

Jana, A. and Stuedlein, A.W. (2022). “Dynamic, In-situ, Nonlinear-Inelastic Response and Post-
Cyclic Strength of a Plastic Silt Deposit.” Canadian Geotechnical Journal, 59(1), pp. 111-128.
https://doi.org/10.1139/cgj-2020-0652

Ladd, C. C. (1991). “Stability evaluation during staged construction.” Journal of Geotechnical

Engineering, 117(4), 540-615. http://dx.doi.org/10.1061/(ASCE)0733-9410(1991)117:4(540)

Park, J., and Santamarina, J. C. (2017). Revised soil classification system for coarse-fine mixtures.
Journal of Geotechnical and Geoenvironmental Engineering, 143(8), 04017039.
https://doi.org/10.1061/(ASCE)GT.1943-5606.0001705

Sanin, M. V., & Wijewickreme, D. (2006). “Cyclic shear response of channel-fill Fraser River
Delta silt.” Soil Dynamics and Earthquake Engineering, 26(9), 854-869.
https://doi.org/10.1016/j.soildyn.2005.12.006

Saye, S. R., Santos, J., Olson, S. M., & Leigh, R. D. (2017). “Linear trendlines to assess soil
classification from cone penetration test data.” Journal of Geotechnical and Geoenvironmental
Engineering, 143(9), 04017060. https://ascelibrary.org/doi/full/10.1061/(ASCE)GT.1943-
5606.0001729

Simpson, D. C., and Evans, T. M. (2016). “Behavioral thresholds in mixtures of sand and kaolinite
clay.” Journal of Geotechnical and Geoenvironmental Engineering, 142(2), 04015073.
https://doi.org/10.1061/(ASCE)GT.1943-5606.0001391

Stewart, J. P., Chiou, S. J., Bray, J. D., Graves, R. W., Somerville, P. G., & Abrahamson, N. A.
(2002). “Ground motion evaluation procedures for performance-based design.” Soil Dynamics

53
and Earthquake Engineering, 22(9-12), 765-772.
https://doi.org/10.1016/S0267-7261(02)00097-0

Stuedlein, A. W., Dadashiserej, A., Jana, A., and Evans, T. M. (2023). “Liquefaction
susceptibility and cyclic response of intact nonplastic and plastic silts.” Journal of Geotechnical
and Geoenvironmental Engineering, 149(1), 04022125.
http://dx.doi.org/10.1061/(ASCE)GT.1943-5606.0002935

Thevanayagam, S., Shenthan, T., Mohan, S., & Liang, J. (2002). “Undrained fragility of clean
sands, silty sands, and sandy silts.” Journal of Geotechnical and Geoenvironmental Engineering,
128(10), 849-859. https://doi.org/10.1061/(ASCE)1090-0241(2002)128:10(849)

US NRC. (2003). Regulatory Guide 1.198 Procedures and Criteria for Assessing Seismic Soil
Liquefaction at Nuclear Power Plant Sites. US NRC.
https://www.nrc.gov/docs/ML0332/ML033280143.pdf

Wijewickreme, D., Soysa, A., and Verma, P. (2019). “Response of natural fine-grained soils for
seismic design practice: A collection of research findings from British Columbia, Canada.” Soil
Dynamics and Earthquake Engineering, 124, 280-296.
https://doi.org/10.1016/j.soildyn.2018.04.053

54
55
APPENDIX A: LIST OF PARTICIPANTS

56
First Name Last Name Affiliation Country
Pedro Arduino University of Washington USA
Laurie Baise Tufts University USA
Christine Beyzaei National Institute of Standards and Technology USA
Jason Bock Geotechnical Resources, Inc USA
Ross Boulanger University of California, Davis USA
Scott Brandenberg University of California, Los Angeles USA
Jonathan Bray University of California, Berkeley USA
Ashly Cabas North Carolina State University USA
Trevor Carey University of British Columbia Canada
Kemal Onder Cetin Middle East Technical University Turkey
King Chin GeoEngineers, Inc. USA
Sandeep Chitta Oregon State University USA
Brady Cox Utah State University USA
Misko Cubrinovski University of Canterbury New Zealand
Ali Dadashiserej Jacobs Solutions, Inc. USA
Shideh Dashti University of Colorado, Boulder USA
Stephen Dickenson New Albion Geotechnical, Inc. USA
Pedro Espinosa ENGEO, Inc. USA
T. Matthew Evans Oregon State University USA
Brice Exeley Haley & Aldrich, Inc. USA
Kevin Franke Brigham Young University USA
Matthew Gibson Clarity Engineering, LLC USA
Russell Green Virginia Tech USA
I.M. Idriss University of California, Davis USA
Amalesh Jana Oregon State University USA
Robert Kayen University of California, Berkeley / USGS USA
Steven Kramer University of Washington USA
Andrew Makdisi United States Geological Survey USA
Erik Malvick California Department of Water Resources USA
Brett Maurer University of Washington USA
Nason McCullough Jacobs Solutions, Inc. USA
Robb Moss San Luis Obispo USA
Ramin Motamed University of Nevada, Reno USA
Diane Moug Portland State University USA
Mitsu Okamura Ehime University Japan
Scott Olson University of Illinois, Urbana-Champaign USA
Susan Ortiz Oregon Department of Transportation USA
Scott Schlechter Geotechnical Resources, Inc USA
Samuel Sideras Shannon & Wilson, Inc. USA
John Stamatakos Southwest Research Institute USA
Jonathan Stewart University of California, Los Angeles USA
Christopher Stouffer ENGEO, Inc. USA
Armin Stuedlein Oregon State University USA
Kristin Ulmer Southwest Research Institute USA
Tom Weaver US Nuclear Regulatory Commission USA
Dharma Wijewickreme University of British Columbia Canada
Derek Wittwer US Bureau of Reclamation USA
Paolo Zimmaro University of Calabria Italy
Katerina Ziotopoulou University of California, Davis USA

57
APPENDIX B: EXTENDED ABSTRACTS

58
 LIQUEFACTION OF SILTY SOIL

Jonathan D. Bray
Univ. of California, Berkeley, CA, USA
[email protected]

INTRODUCTION

This contribution assesses the current state-of-the-art and linkage between the engineering
response of silty soil in the field and laboratory. Insights from the comprehensive investigations
of the silty soil sites in Adapazari and Christchurch are shared. Emphasis is placed on the
engineering response of silty soils that exhibit cyclic mobility similar to that of clean sands.

RESPONSE TO PROMPTS

State-of-the-art

Ground failure in Adapazari, Turkey during the 1999 Kocaeli earthquake (Bray et al. 2004) led
to research that produced the Bray & Sancio (2006) susceptibility criteria (referred to as B&S06).
The criteria were developed based on an extensive database of laboratory tests performed on high-
quality samples retrieved from natural fine-grained alluvial soil deposits after documenting
liquefaction effects at several field case histories sites in Adapazari. Bray & Sancio (2006)
performed over 100 CTX tests and a dozen CSS tests. Donahue et al. (2008) then performed about
50 CSS tests on laboratory-prepared specimens of Adapazari silt of various plasticity indices (PI).
Markham et al. (2018), Beyzaei et al. (2018) and Mijic et al. (2021) each performed dozens of
CTX and CSS tests. Observations in other earthquakes (e.g., 1994 Northridge, 1999 Chi-Chi, and
2010-11 Canterbury sequence) and additional research confirmed the applicability of the B&S06
criteria. The total number of tests and case histories from these studies exceeds the number of data
points used to develop the current empirically based liquefaction triggering procedures.

The B&S06 criteria (i.e., PI ≤ 12 & w/LL ≥ 0.85) are based on the engineering response and
consequences of the material that is cyclically loaded. The B&S06 model examined the response
of a range of slightly plastic silty soils and found their cyclic shear stress vs. cyclic shear strain
curves looked like those of a medium dense to dense sand whose response is termed liquefaction.
If the phenomenon is referred to as liquefaction for a medium dense sand composed of angular
fine sand particles, then a slightly plastic silt under cyclic loading that generates high excess pore
water pressures (i.e., excess pore water pressure ratio, ru > 90%) that produces similar ‘banana-
shaped’ cyclic shear stress vs. cyclic shear strain loops should also be referred to as liquefaction.
Building settlement occurred at both sand and slightly plastic silt sites in New Zealand, Taiwan,
and Turkey. Sediment ejecta were produced along the edges of buildings at some sand sites and at
some slightly plastic silt sites in the field case histories. The engineering response of the sand and
slightly plastic silty soils and their consequences in terms of Performance-Based Earthquake
Engineering (PBEE) are similar so they should both be classified as liquefiable. This is especially
relevant in evaluating the consequences of liquefaction in terms of displacements.

It is noteworthy that Ishihara (1996) found the cyclic stress ratio causing 5% double-amplitude
strain in 20 cycles in laboratory tests of high fine-content soils “did not change much for the low

59
plasticity range” (i.e., PI ≤ 10), “but increases thereafter with increasing plasticity index.” Thus,
an independent study categorized cyclic resistance of slightly plastic fine-grained soils based on a
PI-based criterion, which in this case was PI ≤ 10.

Boulanger and Idriss (2006) (referred to as B&I06) employed a different liquefaction

susceptibility framework focused on the engineering tools they recommended to be used to
evaluate the cyclic strength of sand-like or clay-like materials. Given the different intended
outcomes of the B&I06 and B&S06 criteria, the different PI ranges associated with the B&I06 and
B&S06 criteria do not necessarily constitute different recommendations regarding the assessment
of susceptibility. Instead, the B&I06 susceptibility criterion is intended to direct engineers to the
use of their CPT and SPT based liquefaction triggering for soils they believe are captured well by
their liquefaction triggering CPT and SPT databases, and to direct them not to use their CPT and
SPT based liquefaction triggering methods for soils they term to behave clay-like. Conversely, the
B&S06 criteria categorizes slightly plastic silty soils with field and laboratory responses like those
of some clean sands, in which the term liquefaction is commonly used, as liquefiable without
specifying the method to evaluate the engineering response and consequences of these materials.
However, they state laboratory testing, if performed on high-quality specimens, would provide the
most relevant insight. In fact, Boulanger & Idriss (2008) and Bray & Sancio (2008) both
recommend that soils that can be reliably sampled should be sampled and tested to refine the
assessment of susceptibility as well as triggering, response, and consequences of liquefaction.

The empirical databases used to develop liquefaction triggering procedures consist primarily
of liquefaction triggering data from sand sites. Often clean sand equivalent adjustments are made
to penetration resistances to account for the difference between nonplastic silty soil and sand and
the differences between slightly plastic silt and sand. The basis for these adjustments is not clear.
Recent work by Bray & Olaya (2023) provides data that examine the trends of clean sand
equivalent cone penetration resistance (CPT) adjustments to silty soils using soil behavior type
index (Ic). Their data are from natural soil deposits in Christchurch. Further research is warranted.

Relevant geological, material characteristics, and in-situ tests

Consideration of geologic processes is crucial in proper site characterization, ranging from its
relevance to the composition, fabric and microstructure of soil at a given location in the profile to
the spatial variability of the sediments at a site (Bray & Olaya 2023). Often site characterization is
performed without consideration of the depositional environment of a site. Most current standard-
of-practice in situ test methods are not sensitive enough to detect the fabric and microstructure of
sediments. Furthermore, disturbed sampling may cause mixing of fine- and coarse-grained
fractions, resulting in an incorrect characterization of the engineering properties of the soil.
Better integration of qualitative geologic information about the soils at a site and the
quantitative information from in situ and laboratory engineering tests is essential for quantifying
and minimizing the uncertainties associated with site characterization (Beyzaei et al. 2020). At the
site scale, one potential way to do this is to use proxies for depositional environments. At the fabric
and microstructure scale, use of multiple existing in situ tests that induce different levels of strain
(e.g., Vs and CPT) should be used with continuous high-quality soil sampling to characterize soil
deposits. New in situ test methods that are sensitive to the fabric and microstructure of soil should
be developed.

60
CONCLUDING REMARKS AND RECOMMENDATIONS

The cyclic response of silty soil is less understood than that of clean sand. Additional field,
laboratory, and numerical studies are required to advance the profession’s understanding of the
liquefaction susceptibility and cyclic response of silty soils, especially those with low plasticity.

ACKNOWLEDGEMENTS

Recent work was supported financially by the U.S. National Science Foundation through Grant
CMMI-1956248, CMMI-1561932, CMMI-1407364, CMMI-1332501, and CMMI-1266418.
Additional support by the Faculty Chair in Earthquake Engineering Excellence at UC Berkeley.

REFERENCES

Beyzaei, C.Z., Bray JD, Cubrinovski M, Riemer M, Stringer M. (2018). Laboratory-based

characterization of shallow silty soils in southwest Christchurch. Soil Dyn. Earthquake Eng,
110, 93–109.
Beyzaei, C.Z., Bray, J.D., Cubrinovski, S., Bastin, S., Riemer, M., Stringer, M., Jacka, M., van
Ballegooy, S., and Wentz, R. (2020) "Characterization of Silty Soil Thin-Layering and
Groundwater Conditions for Liquefaction Assessment," Canadian Geotechnical J.,
10.1139/cgj-2018-0287.
Boulanger, R.W., Idriss, I.M. (2008). Closure to “Evaluation of cyclic softening in silts and clays”
by Ross W. Boulanger and I. M. Idriss. J. Geotech. Geoenviron. Eng., ASCE, 135(2), 308.
Bray, J. D., R. B. Sancio, H.T. Durgunoglu, A. Onalp, T. L. Youd, J. P. Stewart, R. B. Seed, O.K.
Cetin, E. Bol, M. B. Baturay, C. Christensen, and T. Karadayilar (2004) “Subsurface
Characterization at Ground Failure Sites in Adapazari, Turkey,” J. Geotech. Geoenviron. Eng.,
ASCE, V. 130(7), 673-685.
Bray, J. D. and Sancio, R. B. (2006). Assessment of the liquefaction susceptibility of fine-grained
soils. J. Geotech. Geoenviron. Eng., ASCE, 132(9), 1165–1177.
Bray, J. D. and Sancio, R. B. (2008). Closure to “Assessment of the liquefaction susceptibility of
fine-grained soils” by Jonathan D. Bray and Rodolfo B. Sancio. J. Geotech. Geoenviron. Eng.,
ASCE, 134(7), 1031–1034.
Bray, J.D., and Olaya, F.R. (2023) “Evaluating Liquefaction Effects,” J Geotech Geoenviron Eng.,
ASCE, in press.
Donahue, J.L., Bray, J.D., and Riemer, M.F. (2008) “The liquefaction susceptibility, resistance,
and response of silty and clayey soils,” GeoEngineering Report, Depart. of Civil and Environ.
Engineering, Univ. of California, Berkeley, Jan 31.
Markham, C.S., Bray, J.D., Cubrinovski, M. and Riemer, M. (2018). Liquefaction resistance and
steady-state characterization of shallow soils within the Christchurch central business district.
J Geotech Geoenviron Eng, ASCE 144(6), 04018032.
Mijic, Z., Bray, J.D., Riemer, M.F., Rees, S.D., and Cubrinovski M. (2021). Cyclic and monotonic
simple shear testing of native Christchurch silty soil, Soil Dyn. Earthquake Eng, 148.

61
 SUSCEPTIBILITY CRITERIA FOR SELECTING ENGINEERING PROCEDURES

Ross W. Boulanger
University of California, Davis, CA, USA
[email protected]

PURPOSE OF SUSCEPTIBILITY CRITERIA

Liquefaction susceptibility criteria are used in conjunction with a hierarchy of simplified to

complex engineering analysis procedures for estimating seismic deformations in a diverse array of
soil and soil-structure infrastructure systems. Liquefaction susceptibility criteria should be
developed to be universally applicable across this hierarchy of analysis procedures and systems.

Estimating deformations for a geotechnical structure can require estimating the strains (from
small to large) that might develop in a wide range of soil types (from cohesionless to cohesive)
across a range of states (from loose to dense of critical state) subjected to a wide range of loading
intensities, with the responses described by various terms including liquefaction and cyclic
softening. Methods for predicting strains generally require knowledge of the earthquake-induced
shear stresses (i.e., demand) and soil shear strength (i.e., capacity). The engineering procedures
that are appropriate for estimating a soil's strength (monotonic or cyclic; drained or undrained), or
more generally its stress-strain response to earthquake loading, depends on the nature of the soil.

For saturated cohesionless soils, such as clean sands or nonplastic silts, the loss of strength
and/or development of strains under earthquake loading is generally referred to as liquefaction.
Liquefaction has become a colloquial term that encompasses phenomena such as flow liquefaction,
excess pore pressure ratio of 100%, shear strains in excess of a specified failure criterion, and
surface manifestations in the field. The potential for liquefaction triggering is commonly evaluated
using correlations based on in-situ tests (SPT, CPT, Vs etc.) rather than laboratory testing of field
samples because conventional tube sampling techniques cause excessive disturbance to
cohesionless soils and frozen sampling techniques are usually uneconomical.

For saturated cohesive soils, such as clays and plastic silts, the loss of strength and development
of strains under earthquake loading is generally referred to as cyclic softening. Cyclic softening
phenomena share similarities with liquefaction phenomena and can be described using similar soil
mechanics theories. Procedures for evaluating cyclic softening focus on estimating monotonic and
cyclic undrained strengths using information from laboratory testing, in situ testing, and empirical
correlations. In contrast to cohesionless soils, conventional tube sampling techniques can usually
be used to obtain reasonably high-quality samples for laboratory strength testing.

Comparisons of different liquefaction susceptibility criteria should begin with clarification of

their intended purpose. Is a liquefaction susceptibility criterion intended to differentiate between
"liquefaction" and "cyclic softening" on the basis of differences in (1) their phenomenological
responses to earthquake/cyclic loading, (2) some fundamental soil mechanics characteristic
influencing their responses, or (3) the types of engineering procedures that are appropriate for
evaluating potential responses? Different purposes can lead to significantly different criteria,
which can lead to problems when the application of a criterion is inconsistent with its purpose.

62
EXAMPLE OF CRITERIA WITH DIFFERENT PURPOSES

For example, consider the liquefaction susceptibility criteria by Bray and Sancio (2006) and
Boulanger and Idriss (2006). Bray and Sancio (2006) used cyclic test results for a wide range of
soils from Adapazari and the observed field performances of those soils during the 1999 Kocaeli
earthquake to conclude that silts and clays with PI ≤ 12 and water contents (w c) greater than 85%
of the Liquid Limit (LL) are liquefiable, while soils with 12 < PI < 18 and wc > 0.8LL are more
resistant to liquefaction but still susceptible to cyclic mobility (Figure 1a). Boulanger and Idriss
(2006) suggested that the emphasis should be put on determining which engineering procedures
are most appropriate for evaluating cyclic strengths, and recommended that clays and silts with PI
≥ 7 be evaluated using cyclic softening procedures, whereas silts and clays with lower PI should
be considered as likely exhibiting sand-like behavior (and evaluated using liquefaction
correlations) unless shown otherwise through detailed laboratory and in situ testing (Figure 1b).

Figure 1. Susceptibility criteria by: (a) Bray & Sancio (2006) and (b) Boulanger & Idriss (2006).

The differences between the guidance provided by Bray and Sancio (2006) and Boulanger and
Idriss (2006) have sometimes been perceived in practice as being greater than they really are, in
large part because of semantics. The commonality between these two sets of guidance is well
illustrated by the following passage from Bray and Sancio (2006),
"Based on the results of the cyclic testing performed in this study, a soil may be susceptible to
liquefaction if the ratio of the water content to liquid limit is greater than 0.85 (wc/LL>0.85)
and the soil plasticity index in less than 12 (PI<12). Soils that do not meet these conditions but
have plasticity index less than 18 (PI<18) and water content to liquid limit ratio greater than
0.8 (wc/LL>0.8) may be moderately susceptible to liquefaction. These soils, especially those
satisfying the first set of requirements, should be tested in the laboratory to assess their
liquefaction susceptibility and strain potential under the loading conditions existing in the field.
Soils with PI>18 did not liquefy at low effective stresses. However, structures founded on these
soils, and for that matter, any soil, may undergo significant deformations if the cyclic loads
approach or exceed the dynamic strength of the soil."
Bray and Sancio further clarify their recommendations in their 2008 closure to discussions,
"The authors contend that field sampling and laboratory testing currently offer the most reliable
way to evaluate the liquefaction susceptibility, resistance, and response of fine-grained soils."

63
The above recommendations are in good agreement with those of Boulanger and Idriss (2006), as
illustrated by the following passage from their closure (Idriss & Boulanger 2008),
"Effective communication regarding issues of liquefaction requires a clear understanding of
the technical definitions used by different individuals. For the paper, definitions for terms were
chosen such that the names for soil type, soil behavior, and analysis methodology were
reasonably consistent: (1) "liquefaction" was reserved for describing the behavior of sand-like
or cohesionless soils that would be appropriately evaluated using semi-empirical SPT- or CPT-
based "liquefaction" correlations; (2) "cyclic softening" was used to describe the behavior of
clay-like or cohesive soils that would be appropriately evaluated using procedures developed
for, or modified from those for clays; and (3) the recommended criteria were called
"liquefaction susceptibility criteria," because they distinguished between these two cases.
Thus, the two sets of guidance differ in the terminology used to describe cyclic loading behavior
(e.g., cyclic softening versus liquefaction), but they both agree in recommending laboratory testing
as the preferred basis for evaluating the cyclic strengths and potential strains for low-plasticity,
fine-grained soils, regardless of what the behavior may be called. In this regard, there is no
significant consequential difference between the practical intent of the two sets of guidance.
The differences in terminology between these two sets of guidance are a consequence of
seeking to avoid two common misuses of criteria in practice: (1) equating "nonliquefiable" with
the absence of a possible problem, such that no further analysis of potential deformations is
performed, and (2) equating "liquefiable" with the requirement that cyclic strengths be evaluated
using SPT or CPT based liquefaction triggering correlations, which can be overly conservative for
low—plasticity fine-grained soils. The criteria by Bray and Sancio (2006) reduces the potential for
the first misuse by including a broader range of soils within the "liquefiable" criteria, after which
the fine print says to determine cyclic strengths by performing lab tests. The criteria by Boulanger
and Idriss (2006) reduces the potential for second misuse by mapping the criteria to the choice of
engineering procedures, after which the fine print says to evaluate potential deformations in
"nonliquefiable" soils using appropriate procedures (e.g., cyclic softening procedures).
CONCLUDING REMARKS AND RECOMMENDATIONS

Alternative names for criteria with different intended purposes could provide clarity for
practice and avoid legacy issues associated with the term "liquefaction susceptibility criteria." For
example, the Bray and Sancio (2006) guidance might be called "Cyclic deformation susceptibility
criteria" whereas the Boulanger and Idriss (2006) guidance might be called "Cyclic strength
evaluation criteria." A clear distinction between criteria that provide different types of guidance
would also help facilitate recognition that they can be complementary tools in application.

REFERENCES

Boulanger, R. W., & Idriss, I. M., (2006). Liquefaction susceptibility criteria for silts and clays. J.
Geotechnical and Geoenvironmental Eng., ASCE 132(11), 1413–426.
Bray, J. D., & Sancio, R. B. (2006). Assessment of the liquefaction susceptibility of fine-grained
soils. J. Geotechnical and Geoenvironmental Eng., ASCE 132(9), 1165–177.
Idriss, I. M., & Boulanger, R. W. (2008). Soil liquefaction during earthquakes. Earthquake
Engineering Research Institute.

64
 PARTICLE FABRIC IMAGING FOR UNDERSTANDING SHEAR RESPONSE OF
SILTS

Dharma Wijewickreme and Ana Valverde

University of British Columbia, Vancouver, BC, Canada
[email protected]

INTRODUCTION

The knowledge from experimental research has shown the significant effect of particle
structure (fabric) on the monotonic and cyclic shear behavior of silts, in addition to the well
understood influence of void ratio (e) and effective confining stress (vc). It has been shown that
3-D imaging can be used to examine the soil fabric of sands (coarse-grained soils). Due to
technology advancements, it is now possible to examine the fabric of finer-grained silt size
material. With this background, a research program using X-ray μ-CT imaging technology, has
been undertaken at the University of British Columbia (UBC) to support characterizing the
mechanical response of natural silts. This extended-abstract presents the initial outcomes of this
work and demonstrates the suitability of X-ray μ-CT imaging methodologies for understanding
the fabric of silt-size particle matrices.

CURRENT UNDERSTANDING

Liquefaction susceptibility of soils under seismic shaking has been studied globally with much
of the focus on the performance of saturated loose sands. Mainly as a result of the liquefaction-
induced damaged observed in the 1991 Chi-Chi, 1999 Kocaeli, and 2011 Christchurch
earthquakes, seismic performance of silty soils has also been receiving increased attention.

Soil fabric refers to the spatial arrangement of individual particles, particle groups, and pore
spaces in soils. Initial recognition of this factor was made by Casagrande and Carillo (1944) via
the ideas of inherent and induced anisotropy of soils (Arthur et al. 1977). Significant effect of
particle fabric and microstructure on the mechanical behavior of soils has been noted by Oda
(1972).

Potential effect of fabric on the behavior of silts

Constant volume monotonic direct simple shear (DSS) testing at UBC has shown that
reconstituted Fraser River Delta silt specimens prepared using slurry deposition method exhibit
lower shear strength (at all levels of confinement) compared to those from counterpart undisturbed
specimens; moreover, the undisturbed specimens display a strain hardening response in contrast
to the behavior observed for reconstituted specimens. These trends are displayed in spite of the
reconstituted specimens having a denser matrix compared to that for relatively undisturbed
specimens. The void ratio (e) and vertical effective stress (v) states after consolidation as well as
after reaching relatively large shear strain levels ( ~15%), as shown in Figure 1 (left side), shows
that the lines for the reconstituted silt are at significantly different locations from that noted for the
undisturbed silt. Cyclic DSS testing has shown that reconstituted natural undisturbed silt generally
exhibits a weaker response compared to that observed from the undisturbed specimens of the same
material. (Figure 1, right side). These DSS results are only explainable by the potential differences

65
in the particle fabric between the two specimen types, highlighting the need to account for the
effect of particle fabric, in addition to the traditionally well studied effects of e and vc.
1.20

Undisturbed specimens
1.10
Initial

1.00
At 15% shear strain

0.90 Reconstituted specimens

e

Initial
0.80
At 15% shear strain
0.70

0.60
1 10 100 1000
' v (kPa)

Figure 1. Behavior of undisturbed and reconstituted specimens of Fraser River silt: Left Side - e-
log v curves for initial consolidation versus those at 15% shear strain; Right side - Cyclic
Resistance Ratio versus Number of cycles for γ=3.75% (Wijewickreme et al. 2019).

Initial findings towards understanding of fabric

The research herein was undertaken using two silt-sized materials with particle sizes ranging
between 2 μm to 74 μm: (i) standard-size silica particles with spherical and irregular shapes; (ii)
Fraser and River silt. Imaging was undertaken using ZEISS Xradia 520 Versa equipment (Zeiss
International, Germany). A dry specimen of soil containing spherical-shaped silica zone (45 and
63 μm size range) overlying irregular-shaped silica layer (40 and 63 μm size range) was imaged,
and the results are shown in Figure 2. The ability of μ-CT imaging to identify/distinguish particle
shapes as well as layering in matrices of silt size particles are notable.

Figure 2. Images from a silica matrix with spherical particles overlying and irregular-shaped
grains.

The outcomes from μ-CT imaging of Fraser River silt are presented in Figure 3. The 3D images
in Figure 3, do not visually indicate any layering or bedding, confirming the non-segregation
(uniformity) expected by the reconstituted specimens formed using slurry deposition. The digital
particle size distributions (PSDs) of the silt from the 3 sub-samples taken from the same parent
specimen are shown in Figure 3 (Right Side); excellent agreement amongst the PSDs is evidence
of very good uniformity within the parent sample. Particle orientation data derived from the same
subsamples shown through rose diagrams in Figure 4, illustrates that the principal axes of the
particles in the subsamples mainly align in directions close to the horizontal - in accord with the
previous observations related to particle orientations for gravity deposited specimens.

66
 Figure 3. Findings for three subsamples (X, Y, and Z) obtained from the same parent
reconstituted specimen of Fraser River silt. Left - Representative raw and processed images;
Right - PSDs.

Figure 4. Particle principal axis orientation for three subsamples of reconstituted Fraser River
silt.

CONCLUDING REMARKS

The research outcomes highlight the potential of X-ray μ-CT imaging to understand the particle
fabric of silt, and in turn, support understanding the mechanical behavior of silts. The findings are
in accord with those known from the mechanical laboratory element testing.

ACKNOWLEDGEMENTS

Funding support from the Natural Sciences and Engineering Research Council of Canada
(NSERC) and X-ray μ-CT imaging services provided by the UBC Pulp and Paper Centre are
gratefully acknowledged.

REFERENCES

Arthur, J. R. F., Chua, K. S., & Dunstan, T. (1977). Induced anisotropy in a sand, Geotechnique,
7(1), 13-30.
Casagrande, A. & Carillo, N. (1944). Shear failure of anisotropic materials, Journal of Boston
Society of Civil Engineers, 31(4), 122-135.
Oda, M. (1972). Initial Fabrics and their Relations to Mechanical Properties of Granular Material.
Soils and Foundations, 12(1), 17–36.
Wijewickreme, D., Soysa, A., & Verma, P. (2019). Response of natural fine-grained soils for
seismic design practice: A collection of research findings from British Columbia, Canada. Soil
Dynamics and Earthquake Engineering, 124, 280–296.

67
 LINKING HYSTERETIC BEHAVIOR TO LIQUEFACTION SUSCEPTIBILITY

Armin W. Stuedlein and T. Matthew Evans

Oregon State University, Corvallis, OR; USA
[email protected] / [email protected]
Adapted from Stuedlein et al. (2023)

INTRODUCTION

The term “liquefaction susceptibility” has been interpreted differently by various researchers
and practitioners, and the potential for confusion is not surprising. Ambiguity stems in part from
the need to link the judgement of a soil’s behavioral response to seismic loading to its cyclic
resistance, which is: (1) commonly estimated using in-situ penetration resistance and/or shear
wave velocity, in the case of sand-like soils, and (2) then if judged sand-like, the cyclic resistance
is estimated based on case histories where surface evidence of liquefaction (e.g., ejecta) was or
was not observed. Historically, soil liquefaction has been related to large-deformation flow failure,
transient development of zero effective stress, or a particular (though often arbitrary – for
transitional soils) cyclic strain failure criterion (Dadashiserej et al. 2022), which injects further
potential for ambiguity in the assessment of whether a particular soil specimen or deposit will
experience liquefaction. Furthermore, the profession must separate lack of observed ejecta from
the possibility that liquefaction has or has not occurred. For example, sites where manifestation of
liquefaction has not been documented may include soils at depth which have liquefied. This
occurrence could impose the transfer of drag loads to deep foundations, or significant settlement
to piled-structures when liquefaction occurs below the depth of piling.

Liquefaction susceptibility should not address whether a soil will liquefy under a given cyclic
stress demand; rather, it should identify whether or not liquefaction can occur given any seismic
demand. Such an interpretation is fully consistent with the recognition that the transient loss of
strength (associated with zero effective stress) and stiffness are considered the hallmarks of
liquefaction phenomena. Considering transitional soils (e.g., silty sands, clayey sands, sandy silts,
silts, low plasticity clayey silts and clays), efforts to quantify the reasonableness in performance
of certain CPT-based soil behavior type thresholds (i.e., Ic) coupled with a given liquefaction
triggering model have revealed significant uncertainty (Maurer et al. 2019). Thus, it appears that
observed hysteretic behavior of high-quality laboratory specimens can shed the clearest light on
what characteristics of soils are those which can be correlated to liquefaction susceptibility. This
abstract describes some of the findings regarding liquefaction susceptibility based on a series of
direct simple shear tests on the transitional soils of deposits found in Southwest Washington and
Western Oregon, conducted as part of a long-range study of these materials.

HYSTERETIC SOIL BEHAVIOR: SAND-LIKE, CLAY-LIKE, OR INTERMEDIATE

BEHAVIOR?

Identifying the hysteretic behavior of fine-grained soils in term of sand-like, clay-like, and
intermediate behavior is important for establishing the potential for transient loss of shear stiffness
and strength during seismic loading. However, these assessments have often been made somewhat
subjectively. Quantitative hysteretic metrics are evaluated for suitability in the consistent
identification of soil behavior. Figure 1a presents selected hysteretic metrics, including the

68
difference in the cyclic shear stress at  = 0, cyc, the minimum tangent shear modulus, Gtan,min,
the angle of the hysteresis curves just prior to and following shear stress reversal,  (computed in
the cyc− plane) and ru,max, each calculated for N = 3% and the last cycle of each test, Nmax. Large
cyclic shear strain amplitudes (i.e., greater than 5%) are considered in addition to  = 3% owing to
the anticipated intensity and duration of loading associated with the subduction zone events
anticipated in the Pacific Northwest.

To minimize the effect of scaling on the interpreted hysteretic behavior, cyc and Gtan,min were
normalized by the corresponding maximum cyclic shear stress, cyc,max, (i.e., cyc/cyc,max and
Gtan,min/cyc,max, respectively). Selected cyc,max-normalized cyclic hysteretic loops presented in Figs.
1b, 1c, 1d, and 1e are accompanied by the hysteretic metrics for Nmax and exhibit a range in
behaviors which evolve with  and N. Qualitatively, the hysteretic behavior of Specimen F-2-6
(Fig. 1b) could be described as intermediate for N = 3% and sand-like for the last loading cycle (i.e.,
Nmax) with its inverted S-shaped cyclic stress-strain hysteresis (indicative of low dissipated strain
energy). Quantitively,  = 7 and 16, and cyc/cyc,max = 0.60 and 0.47 for N = 3% and Nmax,
respectively. Importantly, this specimen exhibits non-zero and zero shear stiffness, with
Gtan,min/cyc,max = 10.1 and 0, and ru,max = 93 and 99%, for N = 3% and Nmax, respectively. The
evolution in the minimum transient shear stiffness and corresponding maximum excess pore
pressure ratio throughout loading is objectively quantified using the hysteretic metrics, which
indicate that the ultimate hysteretic behavior is sand-like.

In contrast, clay-like behavior is qualitatively characterized by wide stress-strain loops with

non-zero shear stiffness and relatively low generated excess pore pressure. Specimen E-3-2 (Fig.
1c) presents an example with clearly clay-like behavior which did not evolve throughout cyclic
loading, quantified with cyc/cyc,max = 0.76 and 1.00,  = 8 and , Gtan,min/cyc,max = 20.4 and
1.26, and limited ru,max = 8 and 79% for N = 3% and Nmax, respectively. Specimens A-BL-3 and A-
BL-5 exhibit frequently-observed evolutionary hysteretic behavior, whereby shear strains of 3%
or greater suggested a clay-like response, but upon continued loading the specimen transitioned to
sand-like behavior at Nmax with Gtan,min/cyc,max less than 2 and ru,max greater than 95%, indicative of
the substantial transient loss of stiffness and strength associated with transient liquefaction.

That the hysteretic behavior of these transitional soil specimens can evolve throughout loading
highlights the role of earthquake duration (i.e., N) on the potential for exhibiting sand-like
behavior. Thus, short duration crustal earthquakes may not produce sufficient loading cycles to
trigger sand-like behavior, whereas longer duration (e.g., subduction zone) earthquakes, which can
produce greater than 100 cycles of loading, depending on the power law exponent b describing the
CRR-N relationship (Boulanger and Idriss 2015; Stuedlein et al. 2021), can lead to the transient
loss of strength. Furthermore, it was observed that specimens that exceeded  = 3% in the first
cycle (i.e., subjected to large CSR) often required a number of additional cycles to satisfactorily
establish the ultimate hysteretic behavior, indicating that significantly larger shear strains than
those associated with common cyclic failure criteria are necessary to make determinations of sand-
like or clay-like behavior.

69
 40

Cyclic Shear Stress, cyc (kPa)

F-2-5 Angle of the hysteresis prior to &
30 ru,max = 98% following shear stress reversal 

20
Minimum tangent
10
shear modulus
0 Gtan,min
cyc
-10 Cyclic shear stress
-20 difference at  = 0

-30
(a)
-40
-15 -10 -5 0 5 10 15
Shear Strain,  (%)
1.5 1.5
Normalized Cyclic Shear Stress,

Normalized Cyclic Shear Stress, Normalized Cyclic Shear Stress,

F-2-6, PI = 0, OCR = 2.4, cyc,max = 20 kPa E-3-2, PI = 27, OCR = 2.1, cyc,max = 47 kPa
Hysteretic metrics, Nmax: Hysteretic metrics, Nmax:
1.0 1.0
ru,max = 99% ru,max = 79%
Gtan,min /cyc,max = 0.00 Gtan,min /cyc,max = 1.26
0.5 0.5 cyc /cyc,max = 1.00
cyc /cyc,max = 0.47
cyc/cyc,max

cyc/cyc,max
 = 16o  = 23o
0.0 0.0
N = 3%
-0.5 -0.5
Nmax
-1.0 -1.0
N = 3%: Intermediate Behavior N = 3%: Clay-Like Behavior
Nmax: Sand-Like Behavior (b) Nmax: Clay-Like Behavior (c)
-1.5 -1.5
1.5 1.5
Normalized Cyclic Shear Stress,

A-BL-3, PI = 11, OCR = 4.2, cyc,max = 12 kPa A-BL-5, PI = 19, OCR = 4.2, cyc,max = 15 kPa
Hysteretic metrics, Nmax: Hysteretic metrics, Nmax:
1.0 1.0
ru,max = 100% ru,max = 96%
Gtan,min/cyc,max = 0.04 Gtan,min/cyc,max = 1.93
0.5 cyc /cyc,max = 0.71 0.5
cyc /cyc,max = 0.74
cyc/cyc,max
cyc/cyc,max

 = 25o  = 27o
0.0 0.0

-0.5 -0.5

-1.0 -1.0
N = 3%: Clay-Like Behavior N = 3%: Clay-Like Behavior
Nmax: Sand-Like Behavior (d) Nmax: Sand-Like Behavior (e)
-1.5 -1.5
-15 -10 -5 0 5 10 15 -15 -10 -5 0 5 10 15
Shear Strain,  (%) Shear Strain,  (%)

Figure 1. Quantification of hysteretic behavior, including: (a) selected metrics considered for the
identification of hysteretic soil behavior at N = 3% and Nmax, and quantified examples of
hysteretic behavior from cyclic direct simple shear tests on intact specimens (this study) for Nmax:
(b) initially intermediate behavior transitioning to sand-like, (c) clay-like, and (d, e) clay-like
behavior transitioning to sand-like behavior.

The variation of ru,max with cyc/cyc,max and Gtan,min/cyc,max for N = 3% and Nmax are presented
in Fig. 2 for nearly 50 representative soil specimens prepared from thin-walled tube samples
obtained from mud-rotary boreholes from selected study sites. These data suggest that for the
typical strain-based cyclic failure criterion of  = 3%, none of the specimens tested exhibited sand-
like behavior, with each exhibiting cyc/cyc,max ⪆ 0.55, Gtan,min/cyc,max > 5, and ru,max < 95% (Figs.
2a and 2b). The hysteretic metrics for N = 3% further suggest that an approximate boundary of 90%
⪅ ru,max < 95% is consistent with precedent-based qualitative judgments of intermediate behavior,
whereas those specimens with ru,max ⪅ 90% also tend to exhibit cyc/cyc,max ⪆ 0.55, Gtan,min/cyc,max
> 2 for both N = 3% and Nmax, providing a quantitative basis that is consistent with precedent-based
judgments of clay-like behavior. When shear strain amplitudes exceed 3%, the hysteretic behavior
of many specimens that previously exhibited intermediate and clay-like behavior transition to
sand-like behavior, with Gtan,min/cyc,max < 2, and ru,max generally greater than or equal to 95%, which

70
quantifies their significant transient loss of strength and stiffness (Figs. 2d and 2e; compare to Fig.
1).

100
Generally Sand-Like Generally Sand-Like
Pressure Ratio, ru,max (%)
Maximum Excess Pore

Generally Intermediate Generally Intermediate

90
Generally Clay-Like

80
Generally Clay-Like
70
Clay-Like Behavior

60 ru,max < 50% Intermediate Behavior

(typ.; see Table S5)
Sand-Like Behavior Nmax
N = 3%
50
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2
(a) Normalized Cyclic Shear Stress Difference at  = 0, (d) Normalized Cyclic Shear Stress Difference at  = 0,
cyc /cyc,max cyc /cyc,max
100
Generally Sand-Like Generally Sand-Like
Maximum Excess Pore
Pressure Ratio, ru,max

Generally Intermediate Generally Intermediate

90
Generally Clay-Like

80
Generally Clay-Like

70

60
ru,max < 50%
(typ.; see Table S5) N = 3% Nmax
50
0 5 10 15 20 25 30 0 2 4 6 8 10
(b) Normalized Minimum Tangent Shear Modulus, (e) Normalized Minimum Tangent Shear Modulus,
Gtan,min/cyc,max Gtan,min/cyc,max
30

25
Plasticity Index, PI

20
2 sp. 2 sp.
15 2 sp. 2 sp. Generally not
Susceptible
2 sp. 2 sp.
10

2 sp. 2 sp. 2 sp. 2 sp. Generally

5
Susceptible
6 NP sp. N = 3% 6 NP sp. Nmax
0
0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
(c) Water Content-to-Liquid Limit Ratio, (f) Water Content-to-Liquid Limit Ratio,
wn /LL wn /LL

Figure 2. Variation of selected hysteretic metrics with excess pore pressure ratio and
liquefaction susceptibility assessment for: (a - c) N = 3%, and (d – f) Nmax, indicating variation of
maximum excess pore pressure ratio with: (a, d) normalized cyclic shear stress difference at  =
0, (b, e) normalized minimum tangent shear modulus, and (c, f) the plasticity index-water
content-to-liquid limit ratio of selected specimens. Note: (1) the number of specimens where
markers coincide is indicated, and (2) non-plastic (NP) specimens assigned wn/LL = 1.0 for
plotting purposes.

71
 Based on the large-strain observations associated with Nmax, approximate quantitative
guidelines for identifying cyclic behavior may be summarized as:
• Clay-like behavior: transitional soils with ru,max ⪅ 90%, cyc/cyc,max ⪆ 0.55, and
Gtan,min/cyc,max ⪆ 2;
• Sand-like behavior: transitional soils with ru,max ⪆ 95% and Gtan,min/cyc,max ⪅ 2; and,
• Intermediate behavior: transitional soils with 90% ⪅ ru,max < 95%, Gtan,min/cyc,max ⪆ 2
and cyc/cyc,max > 0.55.

LINKING HYSTERETIC METRICS TO SOIL INDEX-BASED LIQUEFACTION

SUSCEPTIBILITY CRITERIA

The liquefaction susceptibility and framework for evaluating transient cyclic characteristics
discussed above recognize that judgments of anticipated hysteretic behavior in the absence of site-
specific cyclic data are necessary. Accordingly, the specimen behavior deduced using the
quantitative criteria for N = 3% and Nmax are assessed in terms of correlation to mineralogy and state
through the PI and 𝑤𝑛 /𝐿𝐿 ratio, similar to the Bray and Sancio (2006; B&S06) criteria. Comparison
of Figures 2c and 2f serves to reinforce the need to assess hysteretic behavior of transitional soils
at large strain amplitudes (i.e., greater than 5%) given the lack of sand-like behaviors for N = 3%.
Figure 2f shows that no specimen determined to exhibit clay-like behavior using hysteretic metrics
is characterized with 𝑤𝑛 /𝐿𝐿 > 0.85 and 𝑃𝐼 < 12. In contrast, only one sand-like specimen with 𝑃𝐼
= 19 notably deviates from the 𝑃𝐼 = 12 boundary separating the susceptible and moderately
susceptible soils from not susceptible using the B&S06 criteria, whereas eight sand-like specimens
are characterized with 𝑃𝐼 > 7 associated with the clay-like threshold proposed in the Boulanger
and Idriss (2006; B&I06) criteria. Based on the large-strain cyclic responses of specimens
exhibiting sand- and clay-like hysteretic behavior (i.e., associated with Nmax), it appears that in the
absence of site-specific cyclic test data, transitional soils with 𝑃𝐼 > 12 and/or 𝑤𝑛 /𝐿𝐿 < 0.85 may
be reliably judged as clay-like, whereas soils with 𝑃𝐼 ⪅ 12 and 𝑤𝑛 /𝐿𝐿 > 0.85 may be reliably
judged as sand-like provided that the associated soil deposit experiences sufficient loading cycles
to trigger large-strain behavior.

CONCLUDING REMARKS AND RECOMMENDATIONS

Evaluation of seismic risk associated with soil liquefaction follows three general steps:
evaluation of liquefaction susceptibility, assessment of cyclic resistance and liquefaction
triggering, and estimation of the consequences of liquefaction. Given that the second and third
steps are inextricably linked to the first step, a clear definition of soil liquefaction, constrained to
the transient loss of stiffness and strength, is necessary to complete the liquefaction hazard
analysis. The occurrence of liquefaction, and therefore liquefaction susceptibility, of transitional
soil may be quantified objectively using hysteretic behavior, as demonstrated by the results of the
study discussed herein, independent of the engineering procedure which may be recommended for
estimation of cyclic resistance or the consequences of liquefaction triggering.

ACKNOWLEDGEMENTS

Mr. Ali Dadashiserej, lead student researcher, and Dr. Amalesh Jana performed the cyclic
direct simple shear tests described in this study; their effort and dedication is gratefully

72
acknowledged. This material is based upon work supported by the National Science Foundation
under grant CMMI-1663654, the Oregon Department of Transportation (ODOT) under Grant
SPR-304-911 and various ODOT projects, the Port of Portland, the Cascadia Lifelines Program
(CLiP), and the Pacific Earthquake Engineering Research (PEER) Center under Grant 1175-
NCTRSA. Any opinions, findings, and conclusions or recommendations expressed are those of
the authors and do not necessarily reflect the views of the above-mentioned sponsors.

REFERENCES

Boulanger, R. W., & Idriss, I. M. (2006). Liquefaction susceptibility criteria for silts and clays. J.
Geotch. Geoenv. Eng., (11), 1413-1426.
Boulanger, R. W., & Idriss, I. M. (2015). Magnitude scaling factors in liquefaction triggering
procedures. Soil Dynamics and Earthquake Engineering, 79, 296-303.
Bray, J. D., and Sancio, R.B. (2006). Assessment of the liquefaction susceptibility of fine-grained
soils. Journal of Geotechnical and Geoenvironmental Engineering, 132(9), 1165-1177.
Dadashiserej, A., Jana, A., Stuedlein, A. W., & Evans, T. M. (2022). Effect of strain history on the
monotonic and cyclic response of natural and reconstituted silts. Soil Dynamics and
Earthquake Engineering, 160, 107329.
Maurer, B. W., Green, R. A., van Ballegooy, S., & Wotherspoon, L. (2019). Development of
region-specific soil behavior type index correlations for evaluating liquefaction hazard in
Christchurch, New Zealand. Soil Dynamics and Earthquake Engineering, 117, 96-105.
Stuedlein, A., T. Evans, A. Dadashiserej & A. Jana (2021). Cyclic Response and Softening of
Western Oregon Silts and Assessment within the Simplified method Framework with Updates
Relevant for Application to the Cascadia Subduction Zone. Final Report, Version 2, Cascadia
Lifelines Program, Oregon State University, Corvallis, OR.
Stuedlein, A.W., Dadashiserej, A., Jana, A., Evans, T.M. (2023). On the Liquefaction
Susceptibility and Cyclic Response of Intact Nonplastic and Plastic Silts. Journal of
Geotechnical and Geoenvironmental Engineering, Vol. 149, No. 1, 04022125.

73
 CYCLIC BEHAVIOR OF LOW PLASTICITY FINE-GRAINED SOILS OF VARYING
SALINITY, AND CYCLIC FAILURE DUE TO DYNAMIC SOIL-STRUCTURE
INTERACTION

Scott J. Brandenberg
Department of Civil and Environmental Engineering, University of California, Los Angeles, CA,
United States
[email protected]

Jonathan P. Stewart
Department of Civil and Environmental Engineering, University of California, Los Angeles, CA,
United States
[email protected]

MOTIVATION

Liquefaction susceptibility of fine-grained soils is often based on plasticity index, with lower
plasticity soils considered susceptible to liquefaction and higher plasticity soils considered
susceptible to cyclic softening. Pore fluid chemistry exerts a significant impact on the behavior of
plastic clay minerals because cations from dissolved salts interact with the negatively charged
surface of the clay minerals, thereby suppressing the diffuse double layer and causing a flocculated
structure. The presence of dissolved cations also reduces the plasticity for a soil with a given
mineral composition. However, very little research has been performed to understand the influence
of dissolved cations on the cyclic response of fine-grained soils, and whether plasticity index is a
sufficient indicator of susceptibility.

This abstract addresses themes 1. the current state-of-the-practice and its limitations, and 3.
options for future susceptibility models that could be used, for example, in conjunction with
liquefaction triggering models or hazard mapping. We address theme 1. by presenting direct simple
shear test data for fine-grained soils with PI=9, but with varying mineralogy and pore fluid
chemistry. These soils behave differently during cyclic loading despite having the same PI, which
is not well captured by state-of-the-practice methods that utilize PI to assess liquefaction
susceptibility. We then address theme 3. by briefly laying out a vision for a data-driven
susceptibility model consisting of field performance observations, geotechnical site investigations,
and laboratory tests.

RESPONSE TO PROMPT(S)

Prompt 1. What is the state-of-the-art?

The state of the art in evaluating susceptibility of fine-grained soils is to utilize the soil
plasticity index, PI. For example, Boulanger and Idriss (2007) indicate that a soil is “clay-like”
and therefore not susceptible to liquefaction when PI>7, though it is susceptible to cyclic softening
which should be evaluated using a separate set of procedures. Bray and Sancio (2006) indicate that
loose soils with PI < 12 are susceptible, 12 < PI < 18 and wc/LL>0.8 are marginally susceptible,
and PI >18 are not susceptible. Both of these methods utilize PI to assess susceptibility, while
Bray and Sancio (2006) also use wc/LL which is related to soil state rather than mineral

74
composition. This raises an important question about whether susceptibility should be a sole
function of soil composition, or whether soil state should be included.

Figure 1 illustrates three blends of minerals with the same PI (Table 1) that exhibited
significantly different behavior under stress-controlled cyclic loading. The blends consisted of a
mix of non-plastic silt with either bentonite or kaolinite, and the pore fluid was either fresh water
or a 35 g/L NaCl solution. Vertical consolidation stress was 50 kPa, and OCR was 1 for all three
blends. The SBSW blend exhibits a gradual accumulation of shear strain and relatively “fat”
hysteresis loops that are generally consistent with clay-like behavior, whereas the SKFW blend
exhibits a more abrupt increase in strain amplitude and narrower hysteresis loops with a significant
flat portion in the middle of the loops, which is more consistent with sand-like behavior. The
SBFW blend is intermediate between the other two blends.

Table 1. Properties of mixtures used in experimental program

a
ID % silt % Bentoniteb % Kaolinitec Pore fluid Gs LL PL PI
SBFW 95 5 0 Fresh water 2.64 31.2 22.6 8.6
SBSW 90 10 0 Saline water 2.67 31.9 23.1 8.8
SKFW 78 0 22 Fresh water 2.63 30.0 21.4 8.6
a Sil-co-sil #45 ground silica, Non-plastic
b LL = 455, PL = 40, PI = 416
c
LL = 66, PL = 36, PI = 30

Figure 1. Cyclic behavior of three mineral blends with PI near 9 (Eslami 2017).

75
Prompt 3. Options for future susceptibility models

Future susceptibility models require data capable of tying together field performance
observations from earthquakes, geotechnical site investigation data, and laboratory testing data.
Susceptibility cannot be derived solely from field case history data because it is generally
impossible to discern whether a particular site did not exhibit surface evidence of liquefaction
because it is not susceptible, or because it was not shaken by strong enough ground motion.
Laboratory tests provide insights into fundamental behaviors that cannot be gleaned from field
performance alone, but laboratory testing alone cannot address system responses and field
performance. We suggest that a robust publicly available database that synthesizes field
performance, site investigation, and laboratory test data is the best path toward developing new
susceptibility models. Furthermore, these models should account for uncertainties in susceptibility
assessment. For example, if only soil behavior type index, Ic, is available from a CPT test, there is
significant uncertainty with respect to susceptibility, which should be quantified by a large
standard deviation. If Atterberg limits are available, uncertainty should be reduced. Finally, if site-
specific cyclic testing is performed, uncertainty should be relatively small. Such a framework
provides an incentive for engineers to conduct thorough site investigations to reduce uncertainty,
thereby generally reducing hazard.

CONCLUDING REMARKS AND RECOMMENDATIONS

Liquefaction susceptibility of fine-grained soils has historically been based upon plasticity
index, perhaps in combination with other metrics. We show that soils with the same PI may exhibit
significantly different behavior during stress-controlled cyclic loading, indicating that PI is an
insufficient indicator of susceptibility. What remains unclear at this time is whether additional
parameters might provide predictive power in addition to PI, or whether these deviations in
behavior should be handled as aleatory variability and incorporated into a stochastic analysis
framework. This finding points to the need for a robust, publicly accessible database of cyclic
testing data on fine grained soils to further develop models.

An important issue for our community to clarify is whether the word “susceptible” should refer
solely to compositional characteristics (i.e., mineralogy, plasticity), or whether it should also
include soil state in some manner (water content, OCR). Our opinion is that susceptibility should
be based on compositional characteristics, such that a soil that is not susceptible to liquefaction
will not liquefy regardless of how strongly it is shaken. Soil state and shaking intensity should be
included to assess whether susceptible soils will liquefy.

Regarding terminology surrounding susceptibility, there is some risk in using the phrase “non-
susceptible” without further clarification because engineers may misinterpret “non-susceptible” as
meaning “not problematic” or not worthy of additional ground failure evaluation. In reality, a soil
that is not susceptible to liquefaction likely is susceptible to another mechanism of strength loss
and/or deformation. For future susceptibility models, we suggest clearly stating mechanisms that
should be evaluated based on susceptibility of a specific soil to those mechanisms. For example, a
soil may be susceptible to liquefaction, susceptible to cyclic softening, or susceptible to other
potential mechanisms (e.g., seismic compression).

76
ACKNOWLEDGEMENTS

This material is based on work supported by the National Science Foundation under award
1563638. Any opinions, findings, and conclusions or recommendations expressed in this material
are those of the authors and do not necessarily reflect the views of the National Science
Foundation. We thank Mandro Eslami and Jason Buenker for their contributions to this work via
their PhD studies.

REFERENCES

Boulanger, R. W., & Idriss, I. M. (2007). Evaluation of cyclic softening in silts and clays. Journal
of geotechnical and geoenvironmental engineering, 133(6), 641-652.
Brandenberg, S.J., Buenker, J.M., & Stewart, J.P. (2022). Evaluation of ground failure potential
due to soil-structure interaction and vertically propagating shear waves. Journal of
Geotechnical and Geoenvironmental Engineering, accepted for publication.
Bray, J. D., & Sancio, R. B. (2006). Assessment of the liquefaction susceptibility of fine-grained
soils. Journal of geotechnical and geoenvironmental engineering, 132(9), 1165-1177.
Eslami, M. (2017). “Experimental mapping of elastoplastic surfaces for sand and cyclic failure of
low-plasticity fine-grained soils.” PhD Dissertation, University of California, Los Angeles.

77
 A DATA ARCHIVE OF CYCLIC AND POST-CYCLIC BEHAVIOR OF SILT-RICH,
INTERMEDIATE SOILS OF THE PACIFIC NORTHWEST

Arash Khosravifar Stephen Dickenson Diane Moug

Portland State University, New Albion Geotechnical, Inc., Portland State University,
Portland, OR, USA Reno, NV, USA Portland, OR, USA
[email protected] [email protected] [email protected]

Silt-rich soil deposits are prevalent in the Pacific Northwest (PACNW) region of the USA as
well as other parts of the world. While the majority of past research has been focused on the cyclic
behavior of sands and clays, few studies have investigated the cyclic response of intermediate fine-
grained soils that fall in between classical sand and clay types (e.g., Vaid 1994; Polito & Martin
2001; Bray & Sancio 2006; Idriss & Boulanger 2008; Dahl et al. 2014; Wijewickreme et al. 2019;
Jana & Stuedlein 2021.) The cyclic behavior of silt has been documented as intermediate between
the generalized and short-hand characterization of soil behavior as either “sand-like” or “clay-
like”, thereby adding a level of complexity to seismic vulnerability studies involving silt.

This abstract addresses the workshop theme of “The linkage between laboratory observations,
and field characterization and response” by introducing a data archive compiled of over 200
cyclic shear tests performed on intact soil samples from 37 sites/projects in Oregon, Washington,
Alaska and British Columbia. The data archive is comprised of predominantly unpublished test
results from Direct Simple Shear (DSS) tests and Triaxial compression (TX) tests. The tests
provide researchers and practitioners with a basis for laboratory evaluation of (i) cyclic resistance
for a range of reference shear strains, (ii) post-cyclic stress-strain-strength behavior, and (iii) post-
cyclic one-dimensional volumetric strain. This data archive will help advance the field by
improving our understanding of the effects of stress history and overconsolidation ratio on cyclic
resistance, and correlations between cyclic and post-cyclic responses and various soil properties.
Figure 1 shows the location of sites/projects included in this data archive. The Data Report will be
available under Dickenson et al. (2022) as listed in the References.

Figure 1. Project site locations.

78
Data in this study

The data in this archive includes over 200 cyclic shear tests on silt-rich soil deposits from 37
sites in the PACNW of the USA and regions of British Columbia and Alaska. These tests were
performed by several soil lab testing facilities in support of various, primarily transportation,
projects in these regions. The soil specimens are characterized as low-plasticity silt (ML), low
plasticity clay (CL), high plasticity silt (MH), high plasticity clay (CH), and silty sand (SM) based
on their USCS classification. Figure 2 shows that the soils presented in this study are characterized
as being susceptible to liquefaction or cyclic softening based on screening methods by Idriss and
Boulanger (2008) using the illustration method developed by Armstrong and Malvick (2016). The
fines contents (FC) for these soils range from 18% to 100% and the plasticity index (PI) values
range from nonplastic (NP) to 47. The intact soil samples were extracted from shallow depths
down to a depth of 76 m. The depositional environments of the soils in this archive include fluvial
(e.g., overbank, floodplain, glacial outwash), estuarine (e.g., mudflat, slough, inter-tidal zone),
coastal near-shore (in shallow to intermediate water depths), general alluvial, and mine tailings
(e.g., gravel processing and wash tailings). The data archive will be updated as additional projects
and test data are provided.

Figure 2. Atterberg limits and fines contents of the soils used in this study and the screening
liquefaction and cyclic softening criteria by Idriss and Boulanger (2008) using the illustration by
Armstrong and Malvick (2015).

79
CONCLUDING REMARKS

The data archive presented in this abstract is intended to be used by researchers and
practitioners in evaluating the cyclic and post-cyclic response of silt-rich intermediate soils.
Considering the scarcity of cyclic shear data on intermediate soils, this data archive provides a
benchmark in evaluating the cyclic behavior of silt soils whose cyclic behavior transitions between
sand-like and clay-like behaviors. This data archive includes silts from a variety of depositional
environments in the PACNW, British Columbia and Alaska, thus the samples support assessment
of the influence of factors such as mineralogy, fabric, composition, consistency, density, stress-
history, and aging on the cyclic and post-cyclic behavior of the soil.

ACKNOWLEDGEMENTS

The compilation of data was initiated with funding from the Oregon Department of
Transportation. The efforts have continued with direct and in-kind support from many
organizations and geotechnical consultancies in the Pacific Northwest who shared test data and
geotechnical data reports including Jan Six (formerly with the Oregon Department of
Transportation), Tony Allen (Washington State Department of Transportation), Scott Schlechter,
Jason Bock, and Jack Gordon (GRI Inc.), Don Anderson and Nason McCullough (Jacobs), and
Park Piao, Sam Sideras, Bill Perkins, and Bob Mitchell (Shannon & Wilson).

REFERENCES

Armstrong, R. J., & Malvick, E. J. (2016). Practical considerations in the use of liquefaction
susceptibility criteria. Earthq Spectra:32:1941–50. https://doi.org/10.1193/071114EQS100R.
Bray, J. D., & Sancio, R. B. (2006). Assessment of the liquefaction susceptibility of fine-grained
soils. J. Geotech. Geoenviron. Eng., 132:1165–77.
Dahl, K. R., DeJong, J. T., Boulanger, R. W., Pyke, R., & Wahl, D. (2014). Characterization of an
alluvial silt and clay deposit for monotonic, cyclic and post-cyclic behavior. Can. Geotech.
J;51:432 –40.
Dickenson, S. E., Khosravifar, A., Beaty, M. Bock, J., Moug, D., Schlechter, S. & Six, J. (2022).
Cyclic and Post-Cyclic Behavior of Silt-Rich, Transitional Soils of the Pacific Northwest: A
Data Archive for Geo-professionals in Practice and Research. Geotechnical Engineering Data
Report. New Albion Geotechnical, Inc.
Idriss, I. M., & Boulanger, R. W. (2008). Soil liquefaction during earthquakes. 2nd Ed. Oakland,
Calif.: Earthquake Engineering Research Institute.
Jana, A., & Stuedlein, A. W. (2021). Monotonic, Cyclic, and Postcyclic Responses of an Alluvial
Plastic Silt Deposit. J. Geotech. Geoenviron. Eng;147: 04020174.
Polito, C. P., & Martin II, J. R. (2001). Effects of nonplastic fines on the liquefaction resistance of
sands. J. Geotech. Geoenviron. Eng., 127:408–15. https://doi.org/10.1061/(ASCE)1090-
0241(2001)127:5(408).
Vaid, Y. P. (1994). Liquefaction of silty soils. In: Prakas S, Dakoulas, P (Eds.) Ground failures
under seismic conditions. Reston, Va.: ASCE; October, pp. 1–16.
Wijewickreme, D., Soysa, A., & Verma, P. (2019). Response of natural fine-grained soils for
seismic design practice: A collection of research findings from British Columbia, Canada. Soil
Dyn Earthq Eng 124, 280–96. https://doi.org/10.1016/j.soildyn.2018.04.053.

80
 SUSCEPTIBILITY OF CYCLICALLY-INDUCED DEFORMATION OF LOW
PLASTICITY SILTS IN THE PACIFIC NORTHWEST

Stephen Dickenson Sam Sideras Arash Khosravifar

New Albion Geotechnical, Inc., Shannon & Wilson, Lake Portland State University,
Reno, NV, USA Oswego, OR, USA Portland, OR, USA
[email protected] [email protected] [email protected]

CYCLIC STRAIN ACCUMULATION IN LOW PLASTICITY, TRANSITIONAL SOILS

This abstract provides a brief summary of on-going research on the cyclic and post-cyclic
behavior of non-plastic to moderate plasticity silt-rich, transitional soils vulnerable to the
generation of excess pore pressure. The collection and synthesis of over 200 cyclic Direct Simple
Shear (cycDSS) tests on silt specimens from the Pacific Northwest (PACNW), Alaska, and British
Columbia (Dickenson et.al. 2022) have facilitated the development of trends for the influence of
excess pore pressure generation on the rate of shear strain accumulation, post-cyclic shear strength
and shear stiffness, and volumetric strain due to reconsolidation. This effort supports the broad
theme of the linkage between laboratory observations and field response. The field response of
note in this abstract concerns cyclic shear strain accumulation and modeling of permanent ground
deformations due to excess pore pressure generation, with Ru-values ranging from 0 to 0.95.

INFLUENCE OF EXCESS PORE PRESSURES ON CYCLIC STRAIN

Laboratory testing of intact soil and reconstituted specimens has provided the basis for
modeling of the rate of shear strain accumulation as a function of excess pore pressure. The work
of Tasiopoulou et al. (2020) provides a notable example of shear strain development per cycle of
loading for clean sands (Figure 1) loaded to a state of initial liquefaction (i.e., single amplitude
shear strain of ∼3% during cyclic loading), then subjected to additional loading. The semiempirical
relationship illustrated highlights the importance of cyclic demand (τcyc) and cyclic resistance (DR)
on the post-liquefaction rate of strain. This relationship provides valuable trends for calibration of
constitutive models used in two-dimensional nonlinear deformation analysis (NDA).

Figure 1. Post-liquefaction shear strain curves for clean sands in cyclic undrained laboratory
testing (Tasiopoulou et al. 2020).

81
 In order to develop practice-oriented trends in the rate of shear strain accumulation in low-
plasticity silts for calibration of constitutive models such as PM4Silt, the co-authors have evaluated
data from the PACNW silt database, as well as from current projects. With the goal of establishing
trends that are broadly applicable for a range of cyclic load amplitude and number of cycles, and
cyclic resistance, a relationship that included excess pore pressure (R u) was preferred. This
approach has facilitated the development of trends in the shear strain mobilized per cycle for Ru
values ranging from < 0.10 to 0.95, thus applicable for situations leading up to, and including,
“liquefaction” (or more appropriately “cyclic degradation” for silt that does not exhibit post-cyclic
softening behavior similar to that of loose to medium dense sand). Characterization of the strain
increment per cycle for silt is important in the PACNW where the seismic hazard is dominated by
large magnitude (M 8.5 to 9.2), long-duration Cascadia Subduction Zone earthquakes.

Pore-pressure based models for evaluating behavior such as the post-cyclic shear strength of
fine-grained soils have been presented in which the reduction in undrained (constant-volume)
shearing resistance is directly related to the maximum Ru during cyclic loading (Ajmera et al. 2019;
Dickenson et al. 2022; Egan et al. 1984). A similar approach has been adopted in the current
investigation. A curve-fit approximation that includes a function of Ru in cycDSS tests has been
applied to model the shear strain per cycle for a silt from the PACNW.

PROJECT-SPECIFIC APPLICATION IN PORTLAND, OREGON

A subset of the cycDSS data presented in the proceedings of this PEER Workshop by the
second author is applied in this project summary. Pertinent test results for the 6 specimens include
the following; natural water content 34 to 42%, Plasticity Index 0 to 5%, fines content 45 to 60%.
The overconsolidation ratio was 1.5 for all specimens. The results of 6 tests are summarized to
highlight the general trend of cyclic strain accumulation as a function of; cyclic demand (τcyclic),
cyclic resistance (as correlated with the static undrained shear strength, Su_st), and the excess pore
pressure at each cycle (f[Ru]). The influence of progressive softening of the silt on the shear strain
per cycle is captured by way of the Ru function, which was obtained by curve fit to the cycDSS
tests as previously noted. The trends provided in Figure 2 illustrate that the strain per cycle is well-
correlated with both CSR and Ru, and that strain accumulation is small for Ru less than about 0.4,
then increases rapidly up to shear strains in excess of 1.0 as the Ru values exceed 0.8.

Figure 2. Shear strain per cycle as a function of strength normalized cyclic load amplitude and
excess pore pressure function for a low-plasticity silty soil.

82
CONCLUDING REMARKS AND RECOMMENDATIONS

With respect to the workshop prompts addressing aspects of liquefaction susceptibility and its
modeling our observations based on the cycDSS data include the following;
1. Incorrect assessment of the excess pore pressure generation and associated shear strain
mobilization per cycle of loading for silt-rich soil commonly results in overprediction of
permanent ground deformations due to long-duration earthquake motions.
2. Relevant aspects of hazard assessment that present challenges, or highlight current limitations,
in current practice-oriented methods for transitional soils include; (i) characterization of
“liquefaction” and “cyclic degradation” type behavior for non- to low-plasticity silt-rich
transitional soils, (ii) assessing the timing of cyclic degradation and the associated coupling of
kinematic and inertial effects for structures founded on soils subject to permanent seismic
deformations, and (iii) for projects involving NDA and transitional soil units, cyclic lab testing
is necessary for calibration of constitutive models and has been demonstrated in several cases
to reduce inherent uncertainty and conservatism associated with the application of (overly)
simplified methods of characterization.
3. The co-authors have experienced numerous cases where the cyclic and post-cyclic behavior of
silt-rich soil was pivotal for dynamic soil-foundation-structure interaction of major bridges,
port waterfront structures, and buried pipelines. In many cases, the mis-use of liquefaction
modeling procedures developed for clean sand with associated fines correction factors has been
problematic. There is a pressing need in practice for additional guidance on the post-cyclic
behavior of silt-rich, transitional soils.
4. Characterization of rate effects on both static undrained strength and cyclic pore pressure
generation in cycDSS testing of low plasticity silt warrants continued research.

ACKNOWLEDGEMENTS

This investigation was initiated with funding from the Oregon Department of Transportation, and
has been continued with direct and in-kind support from many organizations and geotechnical
consultancies in the Pacific Northwest.

REFERENCES

Ajmera, B., Brandon, T., & Tiwari, B. (2019). Characterization of the Reduction in Undrained Shear
Strength in Fine-Grained Soils due to Cyclic Loading. Journal of Geotechnical and Geoenvironmental
Engineering, ASCE, 145(5): 04019017.
Dickenson, S. E., Khosravifar, A., Beaty, M. Bock, J., Moug, D., Schlechter, S. & Six, J. (2022). Cyclic
and Post-Cyclic Behavior of Silt-Rich, Transitional Soils of the Pacific Northwest: A Data Archive for
Geo-professionals in Practice and Research. New Albion Geotechnical, Inc.
Egan, J. A., Moriwaki, Y., & Moses, T. L. (1984). Site Characterization for Seismic Hazards Evaluations,
Proc. of the 1984 Seminar on Earthquake Engineering in Alaska, Alaska Academy of Engineering and
Sciences, 1600 – 1619.
Tasiopoulou, P., Ziotopoulou, K., Humire, F., Giannakou, A., Chacko, J., & Travasarou, T. (2020).
Development and Implementation of Semiempirical Framework for Modeling Postliquefaction Shear
Deformation Accumulation in Sands. Journal of Geotechnical and Geoenvironmental Engineering,
ASCE, 146(1): 04019120.

83
 CPT-BASED PROBABILISTIC PREDICTION OF LIQUEFACTION SUSCEPTIBILITY

Brett W. Maurer
University of Washington, Seattle, Washington, United States
[email protected]

THE CURRENT STATE-OF-PRACTICE AND ITS LIMITATIONS

Cone-penetration-test (CPT) models for predicting liquefaction occurrence and consequences

offer advantages over other in-situ tests. However, because soil samples are not recovered, soils
are often not characterized directly or tested further in the laboratory. Thus, the standard-of-
practice is to infer liquefaction susceptibility from CPT data. Most often, the CPT soil behavior
type index, or Ic, is used for this purpose. First proposed by Jeffries and Davies (1993), Ic was
modified by Robertson and Wride (1998) to better fit the Robertson (1990) Q – F classification
scheme, where Q and F are the normalized CPT penetration resistance and normalized CPT friction
ratio, respectively. This modification has become widely used in practice. In the domain of Q and
F, circular arcs of constant Ic approximate boundaries between soil behavior types. Ic = 2.60, for
example, is the commonly assumed boundary between silt mixtures and sand mixtures, and thus,
is often used to binomially predict susceptibility, such that soils with Ic < 2.6 are inferred to be
susceptible. While the Ic = 2.6 threshold is common, the Ic boundaries are approximate and warrant
continuous retraining, having been established ca. 1990 using an unpublished database that is
described neither quantitatively nor qualitatively by Robertson (1990). Moreover, Ic is unlikely to
be a perfectly efficient or sufficient predictor of susceptibility. For these reasons, studies have
shown the Ic < 2.6 criterion to be suboptimal (e.g., Li et al. 2007; Pease 2010).

REGION-SPECIFIC MODELS LINKING LAB AND FIELD CHARACTERIZATIONS

Using 574 split-spoon soil samples obtained parallel to CPTs, Maurer et al. (2019) developed
probabilistic models for predicting liquefaction susceptibility in Canterbury, New Zealand. This
effort provides important insights that are applicable beyond the region of study, with broader
implications for liquefaction hazard modeling. Using the measured liquid limit, plastic limit, and
natural moisture content of each sample, four different criteria based on Atterberg limits were used
to classify susceptibility: Boulanger and Idriss (2006) [B&I06]; Bray and Sancio (2006) [B&S06];
Polito (2001) [P01]; and Seed et al. (2003) [Sea03]. The classifications made by these criteria were
then related to the measured Ic. In the following, the Maurer et al. (2019) models are first
susceptibly summarized, after which several remarks and conclusions are presented. The model
developed using the B&I06 criterion is first shown in greater detail. This criterion is often favored
in practice because it was explicitly developed to determine the most appropriate model for
subsequent prediction of cyclic behavior, based on whether the soil’s expected cyclic response is
“sand-like” or “clay-like.” It is important to note that the “susceptibility,” as predicted by Maurer
et al. (2019): (i) is that defined by the developers of the respective criteria; and (ii) that these
definitions are not consistent. Shown in Figure 1a are frequency distributions of samples classified
by the B&I06 criterion as a function of the measured Ic. The optimal deterministic Ic threshold for
binomial prediction of susceptibility, which for this dataset and criterion was Ic = 2.5, is also
plotted. Shown in Figure 1b is the probability that soil is susceptible, modeled using a cumulative
log-normal distribution, as defined in Maurer et al. (2019). Analogous results are shown for all

84
four susceptibility criteria in Figure 2a. These results can also be reconceptualized to represent the
probability density of any Ic threshold value, as shown in Figure 2b.

45 1
BI06
40 (a) Susceptible 0.9 (b)
Bin Data

Ic = 2.5

Susceptible to Liquefaction
Not Susceptible 0.8
35

Probability that soil is

Intermediate (Test) 0.7
30
Frequency

0.6
25 ± 1σ
0.5
20
0.4
15 0.3
10
Range of
0.2 commonly-used
5 0.1 deterministic Ic
thresholds
0 0
1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 1.5 2 2.5 3 3.5
Soil Behavior Type Index, Ic Soil Behavior Type Index, Ic
Figure 1. Results using the Boulanger and Idriss (2006) criterion: (a) frequency distributions of
samples classified as a function of Ic; (b) the probability of susceptibility as a function of Ic.
1 2
BI06 BI06
Probability Density of Ic Threshold

0.9 (a) 1.8 (b)

BS06 BS06
Susceptible to Liquefaction

0.8 1.6
Probability that soil is

Sea03 Sea03
0.7 1.4
P01 P01
0.6 1.2
0.5 1
0.4 0.8
0.3 0.6
Range of
0.2 common 0.4
0.1 deterministic 0.2
Ic thresholds
0 0
1.5 2 2.5 3 3.5 4 1.5 2 2.5 3 3.5 4
Soil Behavior Type Index, Ic Soil Behavior Type Index, Ic
Figure 2. Results using the B&I06, B&S06, Sea03, and P01 criteria: (a) the probability of
susceptibility as a function of Ic; (b) probability density of Ic threshold values.

CONCLUDING REMARKS AND RECOMMENDATIONS

The global applicability of the susceptibility models developed from data in Canterbury cannot
be known, and thus, recommendations for or against the use of these models elsewhere cannot be
made. Nonetheless, several important conclusions can be derived from these models:

1. While the most common deterministic Ic thresholds for discriminating susceptibility (i.e., Ic =
2.4-2.6) appear to be reasonable medians, the relationship between Ic and susceptibility, as
predicted by any of these criteria, is quite uncertain. Using the B&I06 criterion in Canterbury,
for example, there is a 15% probability that soil with Ic ≈ 2.3 is not susceptible, and similarly, a
15% probability that soil with Ic ≈ 2.75 is susceptible. It should also be recognized that the

85
 existing criteria based on Atterberg limits do not provide uncertainty quantification, and as such,
that component of uncertainty is not accounted for here.
2. Various criteria based on Atterberg-limit data often provide very different predictions of
susceptibility. This is of course unsurprising, especially when considering that the definition of
“susceptibility” is inconsistent among developers of susceptibility criteria.
3. Given that the uncertainty between Ic and susceptibility is nontrivial, it should arguably be
accounted for in any rigorous probabilistic treatment of liquefaction hazard. Accounting for this
uncertainty will be most consequential when Ic is asymmetrically distributed within a given soil
profile. In these cases, accounting for the uncertainty of the Ic threshold will result in the
predicted liquefaction hazard differing from that computed using a median threshold. This
assumes, of course, that the median is not systematically unreasonable for the profile (i.e.,
biased), which would be an altogether different and more consequential problem.
4. The models developed in Canterbury provide a methodology that can be repeated for other
regions or at global scale. The uncertain relationship between Ic and susceptibility predicted
from Atterberg limits also suggests that predictor variables yet to be determined (i.e., other than
Ic) could provide more efficient and/or sufficient predictions of susceptibility. Efforts to better
predict susceptibility via CPT data, and to define the uncertainty therein, are needed.

REFERENCES

Boulanger, R.W., & Idriss, I. M. (2006). Liquefaction susceptibility criteria for silts and
clays. Journal of geotechnical and geoenvironmental engineering, 132(11), 1413-1426.
Bray, J. D., & Sancio, R. B. (2006). Assessment of the liquefaction susceptibility of fine-grained
soils. Journal of geotechnical and geoenvironmental engineering, 132(9), 1165-1177.
Jeffries, M.G. & Davies, M.P. (1993). Use of CPTu to estimate equivalent SPT N60. Geotechnical
Testing Journal, 16(4), 458-468.
Li, D. K., Juang, C. H., Andrus, R. D., & Camp, W. M. (2007). Index properties-based criteria for
liquefaction susceptibility of clayey soils: a critical assessment. Journal of Geotechnical and
Geoenvironmental Engineering, 133(1), 110-115.
Maurer, B. W., Green, R. A., van Ballegooy, S., & Wotherspoon, L. (2019). Development of
region-specific soil behavior type index correlations for evaluating liquefaction hazard in
Christchurch, New Zealand. Soil Dynamics and Earthquake Engineering, 117, 96-105.
Pease, J. W. (2010, May). Misclassification in CPT liquefaction evaluation. In Proc. 2nd
International Symposium on Cone Penetration Testing (pp. 9-11).
Polito, C. (2001, March). Plasticity based liquefaction criteria. In Proc., 4th Int. Conf. Recent
Advances in Geotechnical Earthquake Eng and Soil Dynamics. San Diego, CA, March 26-31.
Robertson, P. K. (1990). Soil classification using the cone penetration test. Canadian geotechnical
journal, 27(1), 151-158.
Robertson, P. K., & Wride, C. E. (1998). Evaluating cyclic liquefaction potential using the cone
penetration test. Canadian geotechnical journal, 35(3), 442-459.
Seed, R. B., Cetin, K. O., Moss, R. E., Kammerer, A. M., Wu, J., Pestana, J. M., ... & Faris, A.
(2003). Recent advances in soil liquefaction engineering: a unified and consistent framework.
In Proceedings of the 26th Annual ASCE Los Angeles Geotechnical Spring Seminar.

86
 RELATING CYCLIC BEHAVIOR TO CPT DATA FOR INTERMEDIATE FINE-
GRAINED SOILS

Diane Moug, Arash Khosravifar

Portland State University, Portland, Oregon, USA
[email protected], [email protected]

The Cone Penetration Test (CPT) is an established method for evaluating the liquefaction
susceptibility of sand and granular soils. However, there remains large uncertainty when
evaluating the cyclic strength of soils intermediate to sands and clays (e.g., non-plastic silts, clayey
silts). This abstract addresses the workshop theme of “The linkage between laboratory
observations, and field characterization and response”, specifically by discussing relationships
between intermediate soil cyclic behavior, cyclic strength, soil state and CPT data. Developing
relationships between the cyclic response of intermediate soils and CPT data currently has several
limitations, including: (1) uncertainties in use of CPT-based screening criteria for cyclic behavior
and (2) limited understanding of how CPT data responds to changes in intermediate soil properties
and state parameters that affect cyclic behavior. This abstract discusses recent work to address
these limitations through a database of geotechnical project data from fine-grained soil sites and
numerical cone penetration modeling in non-plastic and low-plasticity silt.

Figure 1. CRR – qt1N data from fine-grained soil projects in Oregon and Washington: (a) Ic ≤
2.6, (b) 2.6 < Ic ≤ 2.95, and (c) Ic ≥ 2.95 (from Moug et al. 2022).

CPT-based intermediate soil liquefaction susceptibility and cyclic strength

The cyclic strength of soil (i.e., cyclic resistance ratio, CRR) is often estimated from the
normalized cone tip resistance (qt1N). Estimates of CRR from qt1N vary depending on whether the
soil is considered to exhibit sand-like (e.g., Boulanger & Idriss 2014 (2004?)) or clay-like (e.g.,
Boulanger & Idriss 2006) behavior. Therefore, a reliable CPT-based screening method for
intermediate behavior will be a notable step forward for liquefaction evaluation. Soil behavior type
index (Ic), as defined by Robertson (1990), is often used as screening criteria: soils with Ic > 2.6
are generally not susceptible to liquefaction but should be sampled and lab tested (Robertson &
Wride 1998).

A preliminary analysis of 11 fine-grained soil sites in western Oregon and Washington

supports use of Ic for liquefaction screening (Moug et al. 2022). Fine-grained alluvial soils are
ubiquitous in western Oregon and Washington, therefore, region-specific CPT-based relationships

87
will strongly benefit the region. The data for this study are shown in Figure 1 with CRR-qt1N pairs
binned by Ic≤ 2.6, 2.6 < Ic ≤ 2.95, and Ic > 2.95. The CRR for these soils were evaluated using
cyclic direct simple shear tests performed on intact soil samples. The qt1N values were evaluated
using CPT data measured in the same soil unit from which the intact soil samples were obtained.
The project data indicate that soils with Ic ≤ 2.6 are consistent with the Boulanger & Idriss (2014)
relationship for sand with 70% fines content; Ic > 2.95 are consistent with typical clay-like behavior
but CRR values are generally lower than expected from the relationship; and, soils with 2.6 < Ic ≤
2.95 appear to transition between the sand and clay relationships. This analysis provides a basis
for in-depth examination of CRR-qt1N relationships for regional soils, and for intermediate soils
more broadly, to constrain liquefaction screening criteria and CRR-qt1N relationships.

Figure 2. Centrifuge model (Price et al. 2019) and Figure 3. Compression behavior of
simulated penetration data (Moug & Price 2023) for PI PI = 0 and PI = 6 silt and clay
= 0 and PI = 6 silt and clay mixtures. mixtures (data from Price 2018).

Influence of intermediate soil properties and state on CPT data

Linking cyclic behavior and strength of intermediate soils to CPT data will benefit from
investigations into the relationship of CPT data to soil properties (e.g., compressibility, critical
state line (CSL) position) and state parameters (e.g., ξo). A primary influence of the transition of
CRR-qt1N relationships between sands and clays is soil compressibility. Soil compressibility has
recently been incorporated by Saye et al. (2021) into a CPT-based liquefaction susceptibility and
triggering evaluation procedure. The influence of compressibility is shown in Figure 2 and 3 with
geotechnical centrifuge model data, numerical modeling, and laboratory compression data (Price
et al. 2019; Moug et al. 2019; Moug & Price 2023). Figure 2 shows centrifuge model-measured qt
at varying penetration velocities for a PI = 0 non-plastic silt (SIL-CO-SIL) and a PI = 6 silt and
kaolin clay mixture. Additionally, simulated qt from a direct axisymmetric cone penetration model
with the MIT-S1 constitutive model (Pestana & Whittle 1999) calibrated for the PI = 0 silt and a
PI = 6 mixture are shown in Figure 2. It should be noted that the PI = 0 soil is a highly angular and
dilative soil, with behavior that likely deviates from naturally-deposited silt soils. The cone
penetration data show that qt values are over an order of magnitude larger for PI = 0 than PI = 6,
which is largely attributed to higher compressibility of the PI = 6 soil mixture. The laboratory-
measured compression behavior and MIT-S1 limiting compression curves (LCC), are shown in
Figure 3. The LCC for PI = 6 is located at much lower mean effective stress (p’) conditions than

88
PI = 0, corresponding to the higher compressibility of the PI = 6 soil. These cone penetration data
show the strong influence of compressibility on qt for intermediate soils. Additionally,
compressibility affects the soil’s CSL, which is further discussed below.

ξo, defined as the difference between the initial void ratio (e) and the equivalent void ratio at
critical state conditions for the same p’, is considered an indicator for whether liquefiable soil will
have contractive (ξo ≥ -0.05) or dilative (ξo < -0.05) shear behavior. Several researchers have
proposed CPT-based methods for interpreting ξo (Plewes et al. 1992; Robertson 2009; Been et al.
1986, 1987), however, given challenges to routinely characterizing ξo (e.g., obtaining intact high
quality samples and characterizing critical state lines), these are generally based on limited data
and soil types. Cone penetration simulations, as shown in Figure 4, provide insight into how PI =
0 and PI = 6 soils respond to changes in ξo for undrained and drained penetration conditions. It
should be noted that for many intermediate soils, the response will be partially drained at the
standard penetration rate. Figure 4 shows that the e – p’ response for soil around the penetrating
cone is related to the CSL position: by the cone tip, soil has loaded to the CSL for drained and
undrained penetration conditions. Therefore, CSL position (which is also influenced by
compressibility) and ξo have a strong influence on qt. Similarly, stress and porewater pressure
conditions at the cone shoulder and friction sleeve, are also influenced by the position of the CSL
and ξo. This work demonstrates that direct cone penetration simulations across intermediate soil
types, ξo, and drainage conditions can provide a fundamental basis (i.e., CSL position and shape)
to link ξo, soil behavior, and soil properties to CPT data.

Figure 4. Simulated e – p’ paths for soil adjacent to a penetrating cone at initial p’ = 66.7 kPa
and various ξo for (a) PI = 0 drained penetration, (b) PI = 0 undrained penetration, (c) PI = 6
drained penetration, and (d) PI = 6 undrained penetration. Dashed lines are the CSL for triaxial
compression loading based on MIT-S1 calibration parameters.

CONCLUDING REMARKS AND RECOMMENDATIONS

The work described in this abstract looks at CRR-qt1N relationships for fine-grained and
intermediate soils, including analysis of pairs of laboratory-characterized CRR values and nearby
CPT profiles, and numerical simulations of cone penetration in intermediate soils. This abstract
recommends additional research into CPT-based liquefaction screening criteria and cyclic strength
evaluation for intermediate soils, including how CPT-based criteria and relationships are affected
by soil properties, ξo and penetration drainage conditions. Compiling and synthesizing data from
existing fine-grained soil sites, specifically pairing CPT profiles next to boreholes with high
quality intact soil sampling for geotechnical projects, will provide useful data to link intermediate
soil behavior to CPT data. Numerical simulations for natural intermediate soils over a range of ξo,
initial stress conditions, and penetration drainage conditions will provide a critical fundamental
understanding for interpreting CPT data.

89
ACKNOWLEDGEMENTS

Many individuals and organizations that provided data and support for this study: Oregon
Department of Transportation, GRI Inc., Northwest Geotech Inc., Geocon Northwest Inc., CH2M
Hill, Shannon & Wilson, GeoEngineers, Washington State Department of Transportation, and
Portland State University. Support for the numerical cone penetration model and calibration came
from Dr. Ross Boulanger and Dr. Adam Price.

REFERENCES

Been, K., Crooks, J., Becker, D., & Jefferies, M. (1986). The cone penetration test in sands: part
I, state parameter interpretation. Géotechnique, 36(2), 239-249.
Been, K., Crooks, J., Becker, D., & Jefferies, M. (1987, September). The cone penetration test in
sands: part II, general inference of state. Géotechnique, 37(3), 285-299.
Boulanger, R. W., & Idriss, I. M. (2004). Evaluating the potential for liquefaction or cyclic failure
of silts and clays. Davis, California: Center for Geotechnical Modeling.
Boulanger, R. W., & Idriss, I. M. (2006). Liquefaction susceptibility criteria for silts and clays. J.
Geotch. Geoenv. Eng., (11), 1413-1426.
Boulanger, R. W., & Idriss, I. M. (2014). CPT and SPT based liquefaction triggering procedures.
Report No. UCD/CGM.-14, 1.
Moug, D. M. & Price A. B. (2023). Examination of cone penetration in non-plastic silt with a
direct cone penetration model. Submitted for proceedings of 2023 ASCE Geo-Congress.
Moug, D. M., Boulanger, R. W., DeJong, J. T., & Jaeger, R. A. (2019). Axisymmetric simulations
of cone penetration in saturated clay. Journal of Geotechnical and Geoenvironmental
Engineering, 145(4).
Pestana, J. M., & Whittle, A. J. (1999). Formulation of a unified constitutive model for clays and
sands. International Journal for Numerical and Analytical Methods in Geomechanics, 23(12),
1215-1243.
Plewes, H.D., Davies, M.P., & Jefferies, M.G. (1992). CPT based screening procedure for
evaluating liquefaction susceptibility. In Proceedings of the 45th Canadian Geotechnical
Conference, pp. 41-49.
Price, A. B. (2018). Cyclic strength and cone penetration resistance for mixtures of silica silt and
kaolin. PhD dissertation, Department of Civil & Environmental Engineering, University of
California, Davis.
Price, A. B., Boulanger, R. W., & DeJong, J. T. (2019). Centrifuge modeling of variable-rate cone
penetration in low-plasticity silts. Journal of Geotechnical and Geoenvironmental
Engineering, 145(11), 04019098.
Robertson, P. K. (1990). Soil classification using the cone penetration test. Canadian geotechnical
journal, 27(1), 151-158.
Robertson, P. K., & Wride, C. E. (1998). Evaluating cyclic liquefaction potential using the cone
penetration test. Canadian geotechnical journal, 35(3), 442-459.
Robertson, P. K. (2009). Interpretation of cone penetration tests—a unified approach. Canadian
geotechnical journal, 46(11), 1337-1355.
Saye, S. R., Olson, S. M., & Franke, K.W. (2021). Common-origin approach to assess level-ground
liquefaction susceptibility and triggering in CPT-compatible soils using ΔQ. Journal of
Geotechnical and Geoenvironmental Engineering. 147(7):04021046.

90
 USING CONE PENETRATION TESTS IN WILLAMETTE SILTS

Susan C. Ortiz
Oregon Department of Transportation, Salem, Oregon, USA
[email protected]

LABORATORY TEST RESULTS CORRELATED TO IN-SITU CPT

This research focused on relating engineering parameters and cyclic resistance of large strain
liquefaction triggering of lightly overconsolidated, low to non-plastic silts at seven different study
sites. Specific attention was given to preconsolidation (’p), undrained shear strength (su) and
liquefaction triggering using sand-like, clay-like, and the Common-origin Q methods. Laboratory
tests from undisturbed samples were paired with measured data and relationships from adjacent
cone penetration test sounding to develop relationships.

With the use of 24 good quality one-dimensional constant rate of strain consolidation tests, a
regional silt-specific ’p relationship was developed. A regional silt-specific correlation for
Willamette Silt, using the power law, improved the predictability of ’p from both Agaiby and
Mayne (2019) and Mayne (2007) models which underestimated and overestimated, respectively.

Figure 1. Preconsolidation stress models (Mayne 2007; Agaiby & Mayne 2019; Ortiz 2022).

Using monotonic direct simple shear tests results from Willamette Silts, two approaches
were found to be reliable predictors for su. A regional silt-specific cone factor was regressed and

91
as well as a regional silt-specific SHANSEP parameters where both methods were found to provide
a good estimate of su.

Figure 2. Estimation of undrained shear strength using region silt-specific Nkt and SHANSEP
parameters (Ortiz 2022).

Assessment of Cyclic Resistance Against Large Strain Triggering

Cyclic resistance of Willamette Silts using both sand-like and clay-like models indicate
underestimating and overestimating, respectively (Boulanger & Idriss 2016; Boulanger & Idriss
2007). Investigation of the proposed Common Origin-Q model (Saye et al. 2021) indicates close
approximation to sand-like models and generally underestimates the cyclic resistance.

Figure 3. Cyclic Resistance of Willamette Silt using Sand-like and Clay-like models (Boulanger
& Idriss 2007, 2016; Ortiz 2022).

92
 Figure 4. Cyclic Resistance of Willamette Silt using Common Origin-Q method (Ortiz 2022).

CHALLENGES WITH WILLAMETTE SILT

Soil throughout Southwest Washington and Western Oregon are placed by catastrophic
floods known as the Missoula Floods. Repeated flooding of water up to 400 feet, ponding for
decades, resulted in deep soil deposits in the Willamette Valley. These flood deposits, Willamette
Silts, are low to non-plastic silts which are lightly overconsolidated and generally do not behave
like normally consolidated silty sands and sandy silts. Designing projects which are founded in
Willamette Silts continue to prove to be difficult to model. Willamette silts show that neither sand-
like nor clay-like models for cyclic resistance are particularly good. This is troublesome given that
the use of sand-like model results in an over-conservative design increasing the cost of the project
and the clay-like models overestimate CRR which compromise safety and long-term performance.

CONCLUDING REMARKS AND RECOMMENDATIONS

Willamette Silts are difficult soils to collect, test and model. Given the stress history and
characteristics of the soil, models generated from global databases are not well-suited for use in
design. Examples of this include; ’p where standard-of practice methods both over- and under-
estimate laboratory test results, su where regional silt-specific Nkt and SHANSEP parameters
improve predictability, and assessment of large strain liquefaction triggering where sand-like or
clay-like models under- and over-predict cyclic resistance. Lesson learned from this research
include: 1) regional specific CPT correlations show promise for improved CPT correlations, 2) an
increased dataset would improve the robustness of the proposed models, 3) laboratory testing is
best.

ACKNOWLEDGEMENTS

The Oregon Department of Transportation sponsored for this research with Dr. Armin W.
Stuedlein as the principal investigator.

93
REFERENCES

Agaiby, S. S., & Mayne, P. W. (2019). CPT evaluation of yield stress profiles in soils. Journal of
Geotechnical and Geoenvironmental Engineering, 145(12), 04019104.
Boulanger, R. W., & Idriss, I. M. (2007). Evaluation of cyclic softening in silts and clays. Journal
of geotechnical and geoenvironmental engineering, 133(6), 641-652.
Boulanger, R. W., & Idriss, I. M. (2016). CPT-based liquefaction triggering procedure. Journal of
Geotechnical and Geoenvironmental Engineering, 142(2), 04015065-04015065.
Mayne, P. W. (2007). NCHRP Synthesis 368: Cone penetration testing. Transportation Research
Board, Washington, DC, 118.
Ortiz, S.C. (2022). Region-specific investigation of the Cone Penetration Test for preliminary
classification and estimation of preconsolidation stress, undrained shear strength, and cyclic
resistance of transitional silts. Oregon State University.
Saye, S. R., Olson, S. M., & Franke, K. W. (2021). Common-Origin Approach to Assess Level-
Ground Liquefaction Susceptibility and Triggering in CPT-Compatible Soils Using Δ Q.
Journal of Geotechnical and Geoenvironmental Engineering, 147(7), 04021046.

94
HIGH-RESOLUTION VP & VS MEASUREMENTS FROM DIRECT-PUSH CROSSHOLE
(DPCH) TESTING T0 AID IN ASSESSING LIQUEFACTION SUSCEPTIBILITY

Brady R. Cox
Utah State University, Logan, Utah, USA
[email protected]

CURRENT STATE-OF-THE PRACTICE AND ITS LIMITATIONS

Currently, in-situ liquefaction susceptibility is primarily evaluated based on results obtained

from at least one of the following site characterization tests/methods: (1) the standard penetration
test (SPT), (2) the cone penetration test (CPT), and (3) shear wave velocity (Vs) measurements. In
simplified terms, the SPT- and CPT-based methods attempt to account for the impact of soil
density on liquefaction resistance, while the Vs -based methods attempt to account for the impact
of small-strain soil stiffness/modulus on liquefaction resistance. Each method has strengths and
weaknesses in regards to evaluating liquefaction susceptibility, and it is beyond the scope of this
abstract to elaborate on these points. Suffice it to say that both density and small-strain stiffness
are important, and integrating these approaches with one another would be ideal. While sufficient
space is not available in this abstract to elaborate on various approaches for integrating penetration-
based and Vs-based liquefaction susceptibility methods, some thoughts are provided on how high-
resolution measurements of compression wave velocity (Vp) and Vs can be used together to
supplement traditional liquefaction susceptibility analyses. Whereas much attention has already
been given to Vs in regards to evaluating liquefaction susceptibility, relatively little effort has been
put forth regarding the importance of Vp in evaluating liquefaction susceptibility. Vp measurements
can be useful in at least two ways: (1) in determining in-situ degree of saturation (or lack thereof),
and (2) in determining in-situ soil void ratio. Currently, we neglect the positive benefits of partial
saturation below the hydrostatic ground water level in both forward (i.e., design) and backward
(i.e., case histories) liquefaction susceptibility evaluations, predominantly because we don’t have
an easy way to determine degree of saturation in-situ and/or understand if partial saturation will
persist over long time periods. Furthermore, given the importance of relative density/void ratio in
all of our laboratory testing for liquefaction susceptibility, we have no good way of measuring it
in the field, and simply hope that the penetration resistance from either SPT or CPT captures its
effects. High-resolution, in-situ measurements of seismic wave velocities can help shed light on
both of these issues. The newly-developed direct-push crosshole (DPCH) method is presented as
an in-situ testing technique that can provide high-resolution measurements of both Vp and Vs.
Several liquefaction case histories from the 2010-2011 Canterbury Earthquake Sequence (CES)
are provided to illustrate how these measurements can be used to determine in-situ degree of
saturation and void ratio.

SATURATION AND VOID RATIO FROM IN-SITU VP & VS MEASUREMENTS

The DPCH test is a new, invasive, near-surface seismic testing method that combines the
desirable characteristics of borehole-based crosshole seismic testing with the relative inexpense
and speed of direct-push testing methods like CPT (Cox et al. 2019). DPCH allows for higher
resolution Vs and Vp measurements than possible with more common methods like seismic CPT
(SCPT). In particular, it is very difficult to obtain accurate Vp measurements with SCPT due to
compression waves traveling directly down the rods. An example of high-resolution Vp and Vs

95
measurements made via DPCH is shown in Figure 1, along with a borehole stratigraphy log and
CPT results collected within 2m of the DPCH measurements. This particular data is from the Cobra
Reserve site in the Halswell suburb of Christchurch, NZ. No observations of surficial liquefaction
manifestation were observed at this site after either the September 2010 Darfield or February 2011
Christchurch earthquakes. Standard CPT-based methods predicted moderate liquefaction in both
earthquakes, making this site a false-positive liquefaction case history. While the soil is very soft
in terms of both qc and Vp, the Vp measurements indicate the soil is not fully saturated (Vp <
1500m/s) for many meters below the hydrostatic ground water level (GWL). Research by Ishihara
and Tsukamoto (2004) indicates that soils with Vp ~700m/s have cyclic resistance ratios (CRRs)
that are ~ 24% greater than an equivalent fully saturated soil. Additional information on attempts
to refine liquefaction susceptibility predictions at various false positive case history sites from the
CES using high-resoluiton Vp and Vs measurements, as well as other refinements like site-specific
fines content correction factors, may be found in Cox et al. (2018), Boulanger et al. (2018), and
McLaughlin et al. (2019).

Figure 1. Site investigation data at the Cobra Reserve site: (a) soil classification from sonic
borehole samples, (b) friction ratio (Rf) and cone tip resistance (qc) from CPT testing, (c)
normalized soil behavior type index (IC) from CPT testing, (d) Vp from DPCH testing, and (e) Vs
from DPCH testing. The ground water table based on piezometer readings is indicated by a
horizontal dashed line and an inverted triangular symbol in each panel (Cox et al. 2019).

96
 High-resolution Vp and Vs measurements from DPCH can also be used to directly measure void
ratio (e) in-situ based on the theory of poroelasticity (Foti et al. 2002). Stolte and Cox (2019)
applied this method at 10 sandy case history sites from the CES and report on its strengths and
weaknesses. Results from Rawhiti Domain site are provided in Figure 2.

Figure 2. Site investigation data at the Rawhiti Domain site: (a) soil classification from
continuous sonic borehole samples, (b) CPT cone tip resistance qc, (c) Vs, (d) Vp, and (e) in-situ
estimates of void ratio from three CPT-based Dr empirical relationships and representative
ranges of emin and emax from laboratory testing and seismic-based estimates of in-situ void ratio
indicated by circular markers at two assumed values for Poisson’s ratio of the soil skeleton, with
νSK = 0.15 always yielding lower void ratio estimates than νSK = 0.35. (Stolte & Cox 2019).

CONCLUDING REMARKS

High resolution Vp and Vs measurements from DPCH testing can shed light on important
factors like in-situ degree of saturation and void ratio that can strongly influence liquefaction
susceptibility. These factors are either ignored or only approximately accounted for in current in-
situ liquefaction susceptibility methods based on SPT, CPT, and/or Vs.

REFERENCES
Boulanger, R.W., Khosravi, M., Cox, B.R., & DeJong, J.T. (2018). Liquefaction Evaluation for an
Interbedded Soil Deposit: St. Teresa’s School, Christchurch, New Zealand. IACGE 2018
Geotechnical and Seismic Research and Practices for Sustainability. Chongquing, China, 21-
22 October 2018.
Cox, B.R., McLaughlin, K.A., van Ballegooy, S., Cubrinovski, M., Boulanger, R., Wotherspoon,
L. (2017). “In-Situ Investigation of False-Positive Liquefaction Sites in Christchurch, New

97
 Zealand: St. Teresa’s School Case History," 3rd International Conference on Performance-
based Design in Earthquake Geotechnical Engineering, Vancouver, Canada, 16-19 July 2017.
Cox, B.R., Stolte, A.C., Stokoe, K.H. II, & Wotherspoon, L.M. (2019). A Direct-Push Crosshole
Test Method for the In-Situ Evaluation of High-Resolution P- and S-wave Velocity. ASTM
Geotechnical Testing Journal, 42(5), 1101-1132. (https://doi.org/10.1520/GTJ2017038).
Foti, S., Lai, C., & Lancellotta, R. (2002). Porosity of fluid-saturated porous media from measured
seismic wave velocities. Géotechnique. 52(5): 359-373.
Ishihara, K. & Tsukamoto, Y. (2004). Cyclic strength of imperfectly saturated sands and analysis
of liquefaction. Proceedings of the Japan Academy, 80B(8): 372-391.
McLaughlin, K.A., Cox, B.R., Wotherspoon, L., Boulanger, R., van Ballegooy, S., & Cubrinovski,
M. (2019). In-Situ Investigation of False-Positive Liquefaction Sites in Christchurch, New
Zealand: Palinurus Road Case History. ASCE Geo-Congress 2019: 8th International
Conference on Case Histories in Geotechnical Engineering. Philadelphia, PA, 24-27 March
2019.
Stolte, A.C., & Cox, B.R. (2019). Feasibility of In-situ Evaluation of Soil Void Ratio in Clean
Sands Using High Resolution Measurements of Vp and Vs from DPCH Testing. AIMS
Geosciences, 5(4), 723-749. (doi: 10.3934/geosci.2019.4.723).

98
 GEOSPATIAL MODELS FOR LIQUEFACTION SUSCEPTIBILITY

Laurie G. Baise
Tufts University, Medford, MA, USA
[email protected]

GEOSPATIAL LIQUEFACTION MODELS

As demonstrated in Zhu et al. (2015, 2017), we have developed a geospatial approach to

liquefaction modeling that can be applied anywhere in the world rapidly after an earthquake using
using globally available datasets. The method relies on geospatial parameters as proxies for
important soil properties including soil density and soil saturation and includes shaking intensity
through PGA and PGV from the USGS ShakeMap. The models are developed statistically and
current models use a logistic regression formulation which results in a probability of liquefaction
which is in turn interpreted as an estimate of spatial extent. The global geospatial liquefaction
model is described by the following equations.

1
PGA, PGV, and Vs30 thresholds (depend on model
1+𝑒−𝑋
P(x) = formulation) (1)
{ 0 Otherwise

P(x) is the probability of liquefaction which lies between zero and 1; and X includes
explanatory variables that describe density, saturation and loading conditions. In Zhu et al. (2017),
several candidate models (M) are developed each using a different set of inputs (explanatory
variables). As an example, Model 2 has the following model form given by:
𝑋 = 𝐴 + 𝑏1 . 𝑙𝑛(𝑃𝐺𝑉) + 𝑏2 . 𝑙𝑛(𝑉𝑠30) + 𝑏3 . 𝑝𝑟𝑒𝑐𝑖𝑝 + 𝑎4 . 𝑑𝑤 + 𝑎5 . 𝑤𝑡𝑑 (2)

This model also includes thresholds by heuristically assigning zero probability when PGV<3
cm/s, Vs30>620 m/s, and precip > 1700 mm (the precip threshold was recommended in Rashidian
& Baise 2020). Ongoing efforts are updating the geospatial liquefaction database and evaluating
new explanatory variables and moving toward a Bayesian modeling approach to quantify
parameter and model uncertainty (Baise et al. 2021; Zhan et al. 2022).

OPTIONS FOR FUTURE SUSCEPTIBILITY MODELS

Geospatial Liquefaction Models are designed to leverage geospatial information about the
geology and hydrology of any point on the globe. Geospatial proxies exist for soil density (e.g.
Vs30) and soil saturation (e.g. distance to closest water body). They use global models and datasets
as well as geospatial tools to calculate such proxies as distance from or elevation above surface
water bodies. Geospatial data captures information about hydrologic conditions and geologic
depositional environments are useful for identifying low-lying areas related to alluvial and coastal
sediments.. In Zhu et al. (2017), we demonstrated that removing the earthquake loading from the
geospatial models results in a map of liquefaction susceptibility as shown in Figure 1 as shown in
comparison to a surficial geology-based liquefaction susceptibility map for the San Francisco Bay

99
area. Geospatial liquefaction models are regional in nature and can provide an estimate of
liquefaction susceptibility.

Figure 1. a) Liquefaction susceptibility from a geospatial model (Zhu et al. 2017); b) Surficial
geology based liquefaction susceptibility (Witter et al. 2006).

CONCLUDING REMARKS AND RECOMMENDATIONS

The geologic depositional environment and hydrologic conditions of a site have long been
understood to be important for determining liquefaction susceptibility as described by Youd and
Perkins (1978). The geospatial liquefaction modeling approach provides a mechanism to bring in
uniform proxy information for depositional environment and saturation at any location on the
globe and should be used in areas lacking more detailed local information as a regional estimate
of liquefaction susceptibility.

ACKNOWLEDGEMENTS

The work has been supported by USGS Award G20AP00029 and G22AP00048. Weiwei Zhan,
a postdoctoral scholar is currently working on updating the database and enhancing the geospatial
models. He has developed the algorithms for adding liquefaction and nonliquefaction samples with
geospatial explanatory variables to the dataset. We are also currently working with Prof. Babak
Moaveni at Tufts to apply Bayesian Updating as a way of quantifying the model uncertainty and
updating the geospatial liquefaction models when new data is available. We have collaborated with
Eric Thompson at the USGS on implementation of the model as part of USGS ground failure
products.

100
REFERENCES

Baise, L.G., Akhlaghi, M., Chansky, A., Meyer, M. & B. Moaveni. (2021). Updates to a Global
Geospatial Liquefaction Model. Final Technical Report for the USGS Award G20AP00029.
Rashidian, V. & Baise, L.G. (2020). Regional efficacy of a global geospatial liquefaction model.
Engineering Geology. April 2020.
Youd, T. L., & D. M. Perkins (1978). Mapping liquefaction-induced ground failure potential. J.
Geotech. Eng. Div. 104, no. 4, Paper Number 13659, 433–446.
Zhan, W., Baise, L.G., Chansky, A., Asadi, A., Sanon, C., Akhlaghi, M.M., and B. Moaveni.
(2022). OpenLIQ: A geospatial database of global seismic liquefaction hazards. Poster for the
annual SSA Meeting.
Zhu, J., Baise, L.G., & Thompson, E.M. (2017). An Updated Geospatial Liquefaction Model for
Global Application, Bull. Seism. Soc. Am. 107 (3), doi: 10.1785/0120160198
Zhu, J., Daley, D., Baise, L.G., Thompson, E.M., Wald, D.J., & Knudsen, K.L. A (2015). A
Geospatial Liquefaction Model for Rapid Response and Loss Estimation. Earthquake Spectra,
31 (3), 1813-1837.

101
ACCOUNTING FOR SOIL AGING IN ASSESSING LIQUEFACTION SUSCEPTIBILITY

Russell A. Green
Virginia Tech, Blacksburg, VA, USA
[email protected]

CURRENT STATE-OF-THE PRACTICE AND ITS LIMITATIONS

Geologic age and origin of the soil has been long recognized as having a significant influence
on its susceptibility to liquefaction triggering (e.g., Seed 1979), where the soil fabric changes with
the age of the deposit, generally resulting in an increased resistance to liquefaction triggering.
However, this influence has been largely expressed qualitatively, e.g., relating depositional setting
and age to relative liquefaction susceptibility (Youd & Hoose 1977). These qualitative
relationships for relative liquefaction susceptibility largely define the current state-of-practice. The
lack of quantitative metrics needed to explicitly incorporate aging effects into liquefaction hazard
analyses often has resulted in ignoring these effects on liquefaction susceptibility. Alternatively,
“engineering judgement” is sometimes used to conclude that a deposit is not susceptible to
liquefaction based on its age. This latter approach is often taken when the computed liquefaction
hazard with aging effects ignored implies that costly mitigation measures are needed.
Unfortunately, “engineering judgement” is not consistently applied across all projects and is
sometimes biased towards a desired outcome, rather than being technically well founded.

WHAT IS THE STATE OF THE ART?

Seed (1979) proposed an early method for accounting for aging on liquefaction resistance by
computing the ratio of the CRR of an aged soil to that of a young deposit of the same soil:
𝐶𝑅𝑅
𝐾𝐷𝑅 = 𝐶𝑅𝑅 𝑎𝑔𝑒𝑑 (1)
𝑦𝑜𝑢𝑛𝑔

where KDR is referred to as the “liquefaction strength gain factor” due to aging effects. More
recently, several approaches have been developed to compute a numerical index for soil aging.
Bwambale and Andrus (2019) provide an excellent overview of the various proposed aging
indices. Correlations relating these indices to KDR defines the state-of-art in accounting for aging
effects on liquefaction susceptibility. One promising index is the ratio of measured to estimated
small-strain shear wave velocities (VS): MEVR (Hayati & Andrus 2009; Andrus et al. 2009).

Hayati and Andrus (2009) argue that the time since last disturbance is more relevant to
liquefaction triggering susceptibility than geologic age. To estimate time since last disturbance,
Andrus et al. (2009) proposed using MEVR as an index:
𝑉
𝑀𝐸𝑉𝑅 = 𝑉𝑆−𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 (2)
𝑆−𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑

where VS-measured is directly measured, and the estimated VS (i.e., VS-estimated) is determined using
correlations relating VS and penetration resistance. The underlying premise of the MEVR index is
that the measurement of penetration resistance mobilizes intermediate to large strains that
inherently disturb the soil fabric and, thus, is not that sensitive to aging effects (i.e., penetration

102
resistance correlates to the small-strain shear wave velocity, VS, of the soil, if the soil were young,
regardless of the age of the soil). In contrast, the measurement of VS directly in the soil is a small-
strain measurement and is sensitive to aging effects (i.e., it is the VS of the aged soil). Thus, the
ratio of directly measured VS to that estimated from penetration resistance should be able to serve
as an index for the time since last disturbance.

HOW SHOULD THE NEXT GENERATION OF LIQUEFACTION SUSCEPTIBILITY

MODELS BE FORMULATED TO ADVANCE THE STATE-OF-THE ART?

Green et al. (2022) proposed the use of the K-factor to account for intrinsic soil properties and
state variables on liquefaction triggering in simplified models. K is used in place of K in
simplified stress-based triggering frameworks, but conceptually K and K are very different.
Numerically, K and K are similar for young, normally consolidated sandy soils when the factor
of safety (FS) against liquefaction triggering is close to one, but may differ significantly for other
scenarios and/or conditions (e.g., aged deposits and when FS ≠ 1).

K is based on equating the shear strain induced in a given soil at given initial stress state and
subjected to a given shear stress to that induced when the soil is confined at a reference initial
stress state, all else being equal. The same concept can be applied to assess the strength gain due
to aging effects by equating induced shear strains in young and aged soils, where young soils
represent the reference condition. As can be surmised from Figure 1, the ratio of the cyclic
resistance ratio (CRR) in the aged and young sands corresponding to a given induced shear strain
() is equal to the ratio of small-strain shear modulus (Gmax) of the aged and young sands, with this
ratio being independent of . Thus:
2
𝐶𝑅𝑅𝑎𝑔𝑒𝑑 𝐺𝑚𝑎𝑥−𝑎𝑔𝑒𝑑 𝑉𝑠𝑎𝑔𝑒𝑑 2 ∙𝜌𝑡 (𝑀𝐸𝑉𝑅∙𝑉𝑠𝑦𝑜𝑢𝑛𝑔 ) ∙𝜌𝑡
𝐾𝐷𝑅 = =𝐺 = 𝑉𝑠 2 ∙𝜌 = = 𝑀𝐸𝑉𝑅 2 (3)
𝐶𝑅𝑅𝑦𝑜𝑢𝑛𝑔 𝑚𝑎𝑥−𝑦𝑜𝑢𝑛𝑔 𝑦𝑜𝑢𝑛𝑔 𝑡 𝑉𝑠𝑦𝑜𝑢𝑛𝑔 2 ∙𝜌𝑡

Figure 1. Shear stress–shear strain ( - ) response of soil for the same soils, one aged and
young. The ratio of the CRR corresponding to the same  for the two soils is K.

103
 Figure 2 shows a comparison of KDR computed using a relationship derived from regressing
compiled soil aging case histories (Saftner et al. 2015) and one derived based on the K-factor, per
Eq. (3). As may be observed from this figure, the two are in amazingly close agreement, especially
given the large uncertainty in aging data. This comparison gives credence for accounting for aging
effects on liquefaction susceptibility via the K-factor, without the need of using the KDR-factor or
qualitative judgement regarding liquefaction susceptibility of aged deposits.

Figure 2. Comparison of the liquefaction strength gain factor, KDR, regressed by Saftner et al.
(2015) [Sea15] from compiled case histories and an expression derived using the K-factor.

CONCLUDING REMARKS AND RECOMMENDATIONS

The use of the K-factor in place of K to analyze liquefaction case histories to develop new
cyclic resistance ratio curves would advance the state-of-the art in assessing the liquefaction
susceptibility of aged deposits, among other advances in the state-of-the art of assessing
liquefaction triggering potential.

REFERENCES
Andrus, R.D., Hayati, H., & Mohanan, N.P. (2009). Correcting liquefaction resistance for aged
soils using measured to estimated velocity ratio. Journal of Geotechnical and
Geoenvironmental Engineering, 135(6), 735-744.
Bwambale, B. & Andrus. R.D. (2019). State of the art in the assessment of aging effects on soil
liquefaction. Soil Dynamics and Earthquake Engineering, 125, 105658.
Green, R.A., Bradshaw, A., & Baxter, C.D.P. (2022). Accounting for Intrinsic Soil Properties and
State Variables on Liquefaction Triggering in Simplified Procedures. Journal of Geotechnical
and Geoenvironmental Engineering, 148(7), 04022056.
Hayati, H. & Andrus, R.D. (2009). Updated Liquefaction Resistance Correlation Factors for Aged
Sands. Journal of Geotechnical and Geoenvironmental Engineering, 135(11), 1683-1692.
Saftner, D.A., Green, R.A., & Hryciw, R.D. (2015). Use of Explosives to Investigate Liquefaction
Resistance of Aged Sand Deposits. Engineering Geology, 199, 140-147.
Seed, H.B. (1979). Soil liquefaction and cyclic mobility evaluation for level ground during
earthquakes. Journal of the Geotechnical Engineering Division, ASCE, 105(GT2), 201–255.
Youd, T.L. & Hoose, S.N. (1977). Liquefaction susceptibility and geologic setting. Proc. 6th
World Conference on Earthquake Engineering, 3, 2189-2194.

104
CONSEQUENCE-BASED SUSCEPTIBILITY INCORPORATING COMPRESSIBILITY

Scott M. Olson
University of Illinois at Urbana-Champaign, Illinois, USA
[email protected]

THEME 3: A FUTURE NEW LIQUEFACTION SUSCEPTIBILITY MODEL

Traditionally, liquefaction susceptibility has focused solely on material-specific criteria,

including composition (gradation, plasticity, etc.), state (density and effective vertical stress; often
as reflected in an overburden stress-normalized penetration resistance), depositional environment,
and historical performance. However, disparate methods have developed to examine nonplastic
and plastic soils and to examine level-ground liquefaction and flow liquefaction.

This extended abstract outlines a new, universal, consistent consequence-based approach to

assess liquefaction susceptibility. When combined with the work presented by Prof. Kevin Franke,
this deterministic approach can be extended to performance-based (probabilistic) design for static
and seismic loading conditions.

PROMPT 5: CONSEQUENCE-BASED LIQUEFACTION SUSCEPTIBILITY

Although not a comprehensive list, the chief consequences of static and seismic liquefaction
are settlement, lateral spreading, and flow failure. Liquefaction flow failure generally is limited to
sloping ground, while lateral spreading can occur in mildly sloping ground or level-ground incised
by a river, stream, or other depression. Settlement can occur in all conditions (i.e., level and sloping
ground) and can occur in much denser soils than the former two consequences.

As noted above, disparate methods have been developed to evaluate the liquefaction
susceptibility of nonplastic and plastic soils and to examine level and sloping ground conditions.
However, nearly all of the material-specific criteria above can be combined using critical state soil
mechanics (CSSM) concepts. The position and slope of the critical state line (CSL) in void ratio
(e) – effective stress (σ') space is a function of the gradation and plasticity (i.e., compressibility)
of a soil. The susceptibility of a soil to various consequences (settlement, lateral spreading, and
flow failure) depends on soil density and geostructure geometry (or “depositional environment”).
Figure 1 schematically illustrates this concept. The CSL for a given soil tends to be curved in e –
σ' space and represents a boundary between soils that are contractive and dilative at large shear
strain (Fig. 1a). Although flow failure may occur at soil states slightly below the CSL, the CSL
generally provides a reasonable separation between states for a given soil that are and are not
susceptible to flow liquefaction (Fig. 1a). As discussed by Jefferies and Been (2016), soils with
denser states are susceptible to lateral spreading and liquefaction-induced settlements. Thus,
similar boundaries in e – σ' space could be developed for a given soil to identify states that are
susceptible to other consequences (Fig. 1b).

As defining in situ void ratio remains difficult, knowing that void ratio (at a given effective
stress) is inversely related to normalized penetration resistance (e.g., normalized CPT tip
resistance, qc1/pa) allows one to revise the e – σ' axes to become qc1/pa – σ' axes and define
consequence-based liquefaction susceptibility boundaries in this space (Fig. 1c).

105
 Normalized penetration
resistance (e.g., qc1/pa)

Susceptible to
Contractive flow failure

Effective stress (σ')

Susceptible to
Void ratio (e)

Void ratio (e)

CSL Susceptible to flow failure
lateral spread
Susceptible to
Dilative
Susceptible to liq- lateral spread
induced settlement
Susceptible to liq-
Effective stress (σ') Effective stress (σ') induced settlement

(a) (b) (c)

Figure 1. Schematic illustrations of: (a) critical state line (CSL) in e – σ' space; (b) consequence-
based liquefaction susceptibility in e – σ' space; and (c) consequence-based liquefaction
susceptibility in qc1/pa – σ' space.

Case histories corresponding to different liquefaction consequences can be used to develop the
boundaries illustrated schematically in Figure 1(c). For this purpose, I have initially utilized the
CPT-based case history datasets from Saye et al. (2021) for level-ground liquefaction, Olson and
Johnson (2008) for well-documented lateral spreads, and Olson and Stark (2002) for flow failures.
Using these datasets, Figure 2 presents consequence-based liquefaction susceptibility boundaries
for: (a) level-ground settlement; (b) lateral spreading; and (c) liquefaction flow failure.

It is widely known that soil compressibility affects penetration resistance, and as shown by
Ishihara (1993), increasing plasticity (generally for PI > ~10; which typically corresponds to
increasing compressibility) increases liquefaction resistance. As such, the susceptibility
boundaries presented in Figure 2 represent limiting boundaries corresponding to low-
compressibility, non-plastic soils. As compressibility increases, these boundaries should shift left.

Olson (2009) proposed a tentative adjustment for the flow liquefaction boundary as a function
of the slope of the CSL in e – log σ' space, termed λ10. Generally, λ10 can be used as a proxy for
soil compressibility. Similarly, the soil behavior type index, Δ Q (Saye et al. 2021), is inversely
related to soil compressibility. With Kevin Franke and Steve Saye, I currently am developing an
interim adjustment to qc1/pa as a function of ΔQ. Based on these concepts, Figure 3 presents
tentative consequence-based, liquefaction susceptibility/compressibility boundaries for: (a) level-
ground settlement using a ΔQ-based adjustment; (b) lateral spreading using a ΔQ-based adjustment
(possible – under investigation); and (c) flow liquefaction using the using a λ10-based adjustment.
As noted, these boundaries required further investigation and the λ10-based adjustment for flow
liquefaction needs to be reconciled with the ΔQ-based adjustment. This work is ongoing.

CONCLUDING REMARKS AND RECOMMENDATIONS

While further study is needed to validate these concepts, the tentative liquefaction
susceptibility boundaries considering compressibility presented herein represent a new universal,
consistent concept to evaluate susceptibility to the various consequences of liquefaction including
level-ground settlement, lateral spreading, and flow failure.

106
 qc1/pa qc1/pa qc1/pa
0 50 100 150 200 0 50 100 150 200 0 50 100 150 200
0 0 0
Not Not Not
susceptible50 susceptible50 susceptible
50

Effective vertical stress, 'vo (kPa)

Effective vertical stress, 'vo (kPa)

Effective vertical stress, 'vo (kPa)

100 100 100

150 150 150

200 200 200

Susceptible to Susceptible to
250 liq-induced 250 lateral spread 250
settlement
300 300 300

350 350 350

(a) (b) Susceptible to (c)

400 400 400 flow failure

Figure 2. Tentative liquefaction susceptibility boundaries for: (a) level-ground settlement; (b)
lateral spreading; and (c) liquefaction flow failure.

qc1/pa qc1/pa qc1/pa

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200
0 0 0
Not Not Not
50 susceptible50 susceptible50 susceptible

? ?
?
Effective vertical stress, 'vo (kPa)

Effective vertical stress, 'vo (kPa)

Effective vertical stress, 'vo (kPa)

100 100 100

?
150 150 ? ? 150

200 Q ≈ 25 32 50 90 250 200 ? 200

? ?

Susceptible to Q ≈ 32(?) 50(?) 90(?) 250(?)

250 liq-induced 250 250
settlement λ ≈ 0.17 0.06 0.03
300 300 Susceptible to 300
Q > 90> 90
deltaQ
lateral spread
60 < DQ
60> < 90
Q>90
350 32 < DQ
32> < 60 350
Q>60
350
Q << 32
DQ 32 (b) Susceptible to (c)
(a)
400 400 400 flow failure

Figure 3. Tentative liquefaction susceptibility boundaries incorporating compressibility for: (a)

level-ground settlement; (b) lateral spreading; and (c) liquefaction flow failure.

ACKNOWLEDGEMENTS

The author acknowledges and appreciates the detailed discussions on this topic with Mr. Steve
Saye and Profs. Kevin Franke, Bret Lingwall, and Armin Stuedlein.

107
REFERENCES

Ishihara, K. (1993). Liquefaction and flow failure during earthquakes. Geotechnique, 43(3), 351-
415.
Jefferies, M. and Been, K. (2016). Soil Liquefaction, A Critical State Approach, 2nd Edition. CRC
Press, Boca Raton, FL, 676p.
Olson, S.M. (2009). Strength ratio approach for liquefaction analysis of tailings dams. Proc.,
University of Minnesota 57th Annual Geotechnical Eng. Conf., Minneapolis, MN, Feb., 37-46.
Olson, S.M. and Stark, T.D. (2002). Liquefied strength ratio from liquefaction case histories.
Canadian Geotechnical Journal, 39, 629-647.
Olson, S.M. and Johnson, C.I. (2008). Use of liquefied strength ratios for analysis of lateral
spreads. J. Geotech. Geoenviron. Eng., 134(8), 1035-1049.
Saye, S.R., Olson, S.M., and Franke, K.W. (2021). Common-origin approach to assess level-
ground liquefaction susceptibility and triggering in CPT-compatible soils using ΔQ. J. Geotech.
Geoenviron. Eng., 147(7), 04021046.

108
 THE IMPACT OF INCORPORATING LIQUEFACTION SUSCEPTIBILITY INTO A
PERFORMANCE-BASED DESIGN FRAMEWORK

Kevin W. Franke
Brigham Young University, Provo, Utah, USA
[email protected]
INTRODUCTION

Not all soils are susceptible to liquefaction and its effects. When evaluating liquefaction
hazard, it is important to have reliable criteria to predict if the occurrence of liquefaction is even
possible given a particular soil. I have been involved in numerous consulting projects where the
susceptibility to liquefaction initiation quite literally was the determinant factor in the
installation/non-installation of $10s of millions in engineering ground improvement. Like
liquefaction initiation itself, liquefaction susceptibility has traditionally been assessed and
assigned in a binary manner: either the soil is susceptible, or it is not. Furthermore, developed
methods to assess liquefaction susceptibility from observed empirical case histories from the field
are arguably more of an assessment of the soil’s susceptibility to ground surface disruption than a
true assessment of liquefaction triggering susceptibility because there has historically been no
practical way (outside of recorded ground motion frequency analysis, trenching, or “geoslicing”)
to assess if liquefaction occurred subterraneously and did not manifest itself at the ground surface.

Few phenomena related to hazards and engineering, however, are truly binary, and liquefaction
is no exception. For example, the recent inclusion of fine-grained “liquefaction” triggering case
histories (and some likely cyclic softening case histories) in the Saye et al. (2021) Common-Origin
liquefaction triggering method shows empirical evidence that a gradual gradient from sand-like
liquefaction behavior to clay-like cyclic softening behavior likely exists (see Figure 1).
Susceptibility to liquefaction based on Q, a CPT-based parameter related to soil type and
compressibility, currently shows an approximate boundary at Q of about 20. However, if more
cyclic softening case histories were available and plotted at Q values less than 20, perhaps the
mCRR line (i.e., the blue boundary line between “Surface Manifestation” and “No Surface
Manifestation”) might extend higher. Therefore, to limit the consideration of susceptibility to
seismic-induced ground surface disruption to only sand-like soils may be a risky proposition.

LIQUEFACTION SUSCEPTIBILITY AND PERFORMANCE-BASED DESIGN

The idea of performance-based design is to predict as closely as possible the actual

performance of the soil (or infrastructure) during the specified design life or exposure time. Such
an effort requires the consideration and incorporation as many of the uncertainties as possible that
are associated with the seismic loading, the soil’s response to that seismic loading, and the
infrastructure’s response to the soil’s response. Thus, confinement of any aspect of the seismic
loading, soil response, and/or infrastructure into convenient “bins” like “sand-like” or “clay-like”
may not necessarily be conducive to accurately predicting the actual performance of the system.

109
 Figure 1. CPT-based liquefaction case histories from the Common-Origin method plotted
against their respective values of Q (modified from Saye et al. 2021).

In a performance-based framework such as that developed by PEER, susceptibility becomes

an additional conditionality in the hazard integrals. For example, the Kramer and Mayfield (2007)
formulation for the hazard curve for Nreq, defined as the SPT resistance needed to prevent
liquefaction triggering, can be modified to include liquefaction susceptibility as:
N M w N amax

N = 
  P  N req

 N req | amaxi , m j , P  Suscept = " yes "amax ,m j  P  Suscept = " yes " (1)
req i
j =1 i =1

Due to space limitations, detailed description of the terms in Equation (1) will not be presented
here, nor is such description necessary. The term of interest for this discussion is the susceptibility
term P  Suscept = " yes " . To use such a term would require the use of a probabilistic susceptibility
relationship for liquefaction triggering, which does not currently exist according to my knowledge.
Previous researchers have only developed and experimented with various heuristic approximations
of the probability of susceptibility to liquefaction triggering (e.g., Huang 2008).

Additionally, and as will be discussed by Prof. S.M. Olson in this workshop, there also
arguably exists susceptibility criteria for various effects of soil liquefaction (i.e., strength loss and
ground deformation). For example, it is possible for certain soils to experience liquefaction
triggering and seismic strength loss, but not lateral spread displacement, even if the
geomorphological and geometric conditions required for lateral spread are present (Youd et al.
2009). Performance-based formulations for computing hazard curves for these various effects
would also include conditional susceptibility terms in the hazard integrals similar to Equation (1).

110
WORKSHOP RESPONSES TO PROMPTS

The brief discussion presented above could apply to several of the PEER Susceptibility
prompts. However, I will focus my attention here on only one prompt that I believe applies
significantly to the incorporation of susceptibility into a performance-based design framework.

What are the consequences of incorrectly assessing liquefaction susceptibility,

and under what conditions are these consequences most acute?

Due to the lack of available probabilistic susceptibility relationships for liquefaction triggering
and its effects, most applications of performance-based liquefaction hazard assessment that I am
aware of apply the consideration of susceptibility in a deterministic “Go/No-Go” manner prior to
implementing the hazard integrals. However, what would be the effect if probabilistic
susceptibility relationships existed and could be incorporated into the hazard integrals themselves,
as suggested in Equation (1)?

This question can be indirectly explored by looking at the recent Bray and Macedo (2020)
model for Newmark sliding block displacements. Many engineering practitioners today use such
models with liquefied soil conditions to develop estimates of lateral spread displacements at bridge
abutments. While most models have traditionally recommended computing probabilities of
exceedance (i.e., fragility relationships) using a formulation such as P  D  D   , Bray and
Macedo recommend computing fragility relationships with their model using the formulation of
P  D  D | P  D  0 P  D  0 . In this formulation, P  D  0 is defined in Bray and Macedo
(2020) and is akin to a probabilistic susceptibility relationship. For a simple demonstration using
Ky=0.3, Mw=7.5, and a displacement to be exceeded of D*=3cm, the fragility curves from the Bray
and Macedo (2020) rigid sliding block model as a function of PGA for both the traditional fragility
calculation and the susceptibility-included fragility calculation are presented in Figure 2. As can
be seen in Figure 2, neglecting the susceptibility results in substantially larger probabilities of
exceedance being computed. This effect repeated millions of times in the performance-based
hazard integrals will have a substantial impact and reduction in the final displacement hazard
curves that are predicted. This impact could easily be much more substantial to the analysis results
than the impact from modifying other terms in the triggering models such as K or MSF, which is
where many of us researchers have historically spent much of our time and focus.

111
Figure 2. Demonstrative fragility curves from the Bray and Macedo (2020) model. The blue line
neglects susceptibility to displacement, and the orange line includes the susceptibility to
displacement.

CONCLUDING REMARKS AND RECOMMENDATIONS

Probabilistic relationships for quantifying susceptibility to liquefaction triggering, cyclic

softening, and their effects are needed for implementation in a performance-based design
framework. Inclusion of such relationships in a performance-based framework will substantially
alter and reduce the computed hazards. I recommend a community collaborative effort to develop
probabilistic susceptibility models for triggering, cyclic softening, and their effects.

REFERENCES

Bray, J.D. & Macedo, J. (2020). Procedure for Estimating Shear-Induced Seismic Slope
Displacement for Shallow Crustal Earthquakes. J. Geotech. Geoenviron. Eng., 145(12),
04019106.
Huang, Y.-M. (2008). Performance-Based Design and Evaluation for Liquefaction-Related
Seismic Hazards. Ph.D. Dissertation, Univ. Washington, Dept. Civil & Env. Eng., 355p.
Kramer, S.L. & Mayfield, R.T. (2007). Return Period of Soil Liquefaction. J. Geotech.
Geoenviron. Eng., 133(7), 802-813.
Saye, S.R., Olson, S.M., & Franke, K.W. (2021). Common-Origin Approach to Assess Level-
Ground Liquefaction Susceptibility and Triggering in CPT-Compatible Soils Using Q. J.
Geotech. Geoenviron. Eng., 147(7), 04021046.
Youd, T. L., DeDen, D.W., Bray, J.D., Sancio, R., Cetin, K.O., & Gerber, T.M. (2009). Zero-
Displacement Lateral Spreads, 1999 Kocaeli, Turkey, Earthquake. J. Geotech. Geoenviron.
Eng., 135(1), 46-61.

112
 PROBABILISTIC MODELS FOR SEISMIC SOIL LIQUEFACTION SUSCEPTIBILITY

Kemal Onder Cetin, Makbule Ilgac, and Gizem Can

Middle East Technical University, Ankara, Turkiye
[email protected], [email protected], [email protected]

THE STATE OF THE ART IN LIQUEFACTION SUSCEPTIBILITY ASSESSMENTS

Seismic soil liquefaction is commonly defined as significant reduction in shear strength and
stiffness due to increase in pore water pressure. Consistent with this definition, susceptibility to
liquefaction is defined as the state of being likely to experience significant shear strength and
stiffness losses, triggered by increase in pore water pressure. Ideally, susceptibility criteria are
expected to be independent of liquefaction triggering parameters of intensity and duration of
loading, and density state of soils; and more specifically linked to soils’ intrinsic characteristics.
A good and pioneering example of it is given by Tsuchida (1970), where a set of grain size
distribution boundaries, identifying “the most liquefiable” and “potentially liquefiable” fully
saturated coarse-grained soils was proposed. Unfortunately, the boundaries were subjectively and
deterministically defined, with limited to no reference to confidence levels of the proposed
boundaries as given in Figure 1 (a).

a)
a) Tsuchida (1970) b) Seed et al. (2003) c) Bray and Sancio (2006)
30
Liquefiable
Cyclic Non-liquefiable
Mobility
Potential

Mean-1
20
Mean boundary

PI (%) Mean+1

Cyclic
Liquefaction
10 Potential

0
0.0 .5 1.0 1.5 2.0 2.5
LI

Figure 6.6. Probabilistic liquefaction susceptibility criterion proposed by Cetin

and Bilge (2014)

d) Ishihara (1996) e) Boulanger and Idriss (2006) f) Cetin and Bilge (2014)

Figure 1. Liquefaction susceptibility criteria for a) clean, coarse-grained soils, and b-f) fines
containing soils and their mixtures.

As shown in Figure 1 (b) through (f), for the susceptibility assessment of mixtures of varying
percentage of fines with sands and gravels, Ishihara (1996), Seed et al. (2003), Bray and Sancio
(2006), and Boulanger and Idriss (2006), Cetin and Bilge (2014) proposed criteria based on

113
consistency limits and fines content, based on field performance or laboratory test data. Ishihara
(1996) proposed that up to the PI value of 10, cyclic resistance ratio (CRR) exhibits a constant
trend, followed by a linear increase with increasing PI>10. The susceptibility criteria by Seed et
al. (2003), and Bray and Sancio (2006) were developed mostly based on joint evaluation of field
performance case histories and the results of cyclic laboratory tests performed on undisturbed soil
samples retrieved from Adapazari after 1999 Kocaeli earthquake. As part of their cyclic testing
program, the stress and density states were chosen to simulate the field conditions in Adapazari
during 1999 Kocaeli earthquake. Hence, the recommendations based on these are more correctly
classified as Adapazari-Kocaeli earthquake criteria, and a good example of a hybrid susceptibility-
triggering assessments. Boulanger and Idriss (2006) suggested that CRR of soils beyond PI > 4 %
sharply increases. Due to reference to CRR in their proposed relationships, Ishihara (1996), and
Boulanger and Idriss (2006) criteria are classified as liquefaction triggering screening criteria
rather than a susceptibility one. Cetin and Bilge (2014) proposed a probability-based susceptibility
criterion based on cyclic triaxial tests performed on a wide range of high quality “undisturbed”
fine-grained soil specimens. Liquefaction susceptibility was judged with the onset of banana
shaped stress-strain cycles. Accordingly, the probability of susceptibility to liquefaction triggering
of fine - grained soils is assessed as given in Equation 1, where LI and Φ are the liquidity index
of fine-grained soils and standard normal cumulative distribution function, respectively.

𝐿𝐼 − 0.578 ⋅ 𝑙𝑛( 𝑃𝐼) + 0.940 (1)

𝑃[𝐿𝑖𝑞 − 𝑠𝑢𝑠𝑐𝑒𝑝𝑡𝑖𝑏𝑖𝑙𝑖𝑡𝑦] = Φ [ ]
0.101

This criterion is independent of the cyclic loading intensity and duration and refers to two
intrinsic properties of fine-grained soils: LI and PI. Hence, it fulfills the requirements of a true
“susceptibility” criterion; and moreover, expresses the susceptibility boundary in a probabilistic
sense, addressing the uncertainty of the problem. The major drawback of it is that it is developed
based on cyclic laboratory test data only, and common to all laboratory-based recommendations,
calibration with field performance case histories is needed.

RELIABILITY-BASED SUSCEPTIBILITY MODELS

As outlined in the previous section, an ideal liquefaction susceptibility assessment framework

should i) refer to intrinsic characteristics of soils, ii) be independent of liquefaction triggering
parameters (intensity of shaking, duration, relative density state, etc.), iii) address the uncertain
nature of susceptibility assessments (i.e.: probability-based), iv) benefit from both laboratory and
field case history data (i.e.: a verified and calibrated model). With the aim of fulfilling these
requirements, SPT and CPT-based liquefaction triggering case histories, documented as part of
Next Generation Liquefaction database (https://nextgenerationliquefaction.org/) were studied. The
grain size distribution curves of the critical soil layers from liquefied, or none-liquefied, or
marginally-liquefied SPT-based case histories, were compiled, as given in Figure 2(a). The median
liquefaction susceptibility grain size distribution curve, along with its standard deviation was
probabilistically assessed benefitting from the maximum likelihood framework. The resulting
median, and plus and minus 1,2,3 sigma bands, along with confidence intervals are shown on
Figure 2(b) as compared with the recommendations of Tsuchida (1970).

114
 Figure 2. a) Grain size distribution curves of susceptible, coarse-grained soils from SPT
database, b) the proposed probabilistic boundaries for susceptibility assessments.

A similar exercise was performed CPT-based liquefaction triggering case histories. The
median soil behavior index Ic, along with its standard deviation were probabilistically assessed
benefitting from the maximum likelihood framework. In Figure 3(a) through (c), the resulting
database and the Ic boundaries corresponding to different confidence levels are also comparatively
shown with CPT-based soil classification boundaries of Robertson (2010), and Cetin and Ozan
(2009).

Figure 3. CPT-soil classification-based liquefaction susceptibility boundary curves.

CONCLUDING REMARKS AND RECOMMENDATIONS

Most of currently available liquefaction susceptibility boundaries were subjectively and

deterministically defined, with limited to no reference to confidence levels of the proposed
boundaries. Also, some of them refer to triggering parameter of CRR; hence, better to be called as
screening criteria, which combining both susceptibility and triggering assessments. Within the
scope of this study, a set of probability-based screening boundaries were recommended for coarse-
and fine-grained soils. The recommended probabilistic boundaries were expressed as probabilistic

115
confidence intervals in the % fines by mass vs. particle size (D), and CPT q vs R f domains. Fine
grained soils with Ic>2.6 are concluded to be not susceptible to soil liquefaction with more than
99% confidence. Moreover, fine grained soils with PI > 12% were judged to be again not
susceptible to liquefaction with confidence levels of 99 %. Based on these, a flow chart scheme
is proposed to assess the susceptibility of soil mixtures with varying % of fines, which could not
be presented herein due to page limitations.

ACKNOWLEDGEMENTS

Some of the data presented herein is compiled as part of the NGL project, which is
acknowledged and appreciated.

REFERENCES

Boulanger, R.W. & Idriss, I.M. (2006). Liquefaction susceptibility criteria for silts and clays.
Journal of Geotechnical Geoenvironmental Engineering, ASCE, 132(11), 1413-1426.
Bray, J. D. & Sancio, R. B. (2006). Assessment of liquefaction susceptibility of fine-grained soils.
Journal of Geotechnical and Geoenvironmental Eng., 132(9), 1165-1177.
Cetin, K. O. & Ozan, C. (2009). CPT-based probabilistic soil characterization and classification.
ASCE Journal of Geotech. and Geoenvir. Eng. 135(1):84-107.
Cetin, K. O. & Bilge H. T. (2014). Recent advances in soil liquefaction engineering. Perspectives
on European Earthquake Engineering and Seismology: Volume 1, edited by A. Ansal,
Springer.
H. Tsuchida, “Prediction and Countermeasure against Liquefaction in Sand Deposits.” Abstract
of the Seminar of the Port and Harbour Research Institute. Ministry of Transport
Ishihara, K. (1996). Soil Behaviour in Earthquake Geotechnics. ISSN 0953-3222,
Robertson, P. K. (2010). Soil behaviour type from the CPT: an update. 2nd International
Symposium on Cone Penetration Testing, USA.
Seed, R. B., Cetin, K. O., Moss, R. E. S., Kammerer, A. M., Wu, J., Pestana, J. M., Riemer, M. F.,
Sancio, R. B., Bray, R. B., Kayen, R. E. & Faris, A. (2003). Recent advances in soil
liquefaction engineering: a unified and consistent framework. Report No. EERC 2003-06,
Earthquake Engineering Research Center, University of California, Berkeley.
Tsuchida, H. (19710). Prediction and Countermeasure against Liquefaction in Sand Deposits.
Abstract of the Seminar of the Port and Harbour Research Institute. Ministry of Transport,
Yokosuka, Japan, pp. 3.1-3.33 (In Japanese).

116
 INCORPORATING THE SPECTRUM OF SOIL BEHAVIORS DIRECTLY INTO
SYSTEMS LEVEL TRIGGERING AND CONSEQUENCE MODELS

Shideh Dashti
University of Colorado Boulder, Boulder, CO, USA
[email protected]

THE CURRENT STATE OF PRACTICE AND LIMITATIONS

Current procedures for assessing soil susceptibility to seismically-induced liquefaction focus

on identifying the transition point in soil behavior from sand-like to clay-like based on plasticity
index and water content. These guidelines were developed based on laboratory tests, backed by
observations of surface manifestation (typically in the form of sand boils or ejecta) in prior case
histories. The subsequent evaluation of liquefaction triggering and consequence followed by
design of mitigation typically ignore the presence of soil layers that are judged unsusceptible at
the first step. Nevertheless, these so-called “unsusceptible” soils may still experience cyclic
softening as well as excessive lateral and vertical deformations resembling those in liquefied
deposits. Alternatively, with their lower permeability, such layers may strongly influence the
redistribution and net generation of excess pore pressures in other susceptible or sand-like layers.
Not considering such “unsusceptible” layers in the profile at the systems level may lead to notable
underpredictions of the likelihood of liquefaction triggering or propagation of damage at the site
and hence, inadequate planning of mitigation strategies.

OPTIONS FOR FUTURE SUSCEPTIBILITY MODELS

As the community reimagines the next generation of susceptibility criteria, it might be

advantageous to move past a binary judgement of whether a soil is susceptible to seismic
liquefaction in the classical sense. In place of categorizing soil behavior as either sand- or clay-
like, a spectrum of behaviors can be considered, focusing on quantification of cyclic softening and
fluid flow at a systems level. Boundary value centrifuge experiments under controlled conditions
followed by 2D and 3D, fully-coupled, effective stress analyses conducted collectively and in a
coordinated and open environment can shed light on the spectrum of behaviors in layered granular
soils with varying fines contents and plasticity levels. The results can and must be compared with
the existing case history database in terms of manifestation of damage, particularly where in
addition to detailed in-situ testing, disturbed and undisturbed samples were obtained. The
generated experimental, numerical, and empirical database can directly feed into triggering and
more importantly consequence procedures that include the entire spectrum of soil types and
profiles at a systems level, without the binary exclusion of clay-like soils.

117
 EXPERIMENTAL EVALUATION OF LIQUEFACTION-INDUCED FOUNDATION
SETTLEMENT CAUSED BY SAND EJECTA

Ramin Motamed
University of Nevada, Reno, NV, USA
[email protected]

1G SHAKE TABLE TESTING TO STUDY LIQUEFACTION-INDUCED SEDIMENT

EJECTA

Liquefaction-induced sediment ejecta and its role in the free-field and building settlements
were briefly discussed during the 2016 US-Japan-NZ Liquefaction Workshop at UC Berkeley
(Bray et al. 2017a). Bray et al. (2017b) describe the three key mechanisms that control liquefaction-
induced building settlement: (1) Shear-induced deformation; (2) Volumetric-induced deformation,
and (3) Ground loss due to ejecta. Although there have been simplified procedures developed to
estimate the first two mechanisms such as Bray and Macedo (2017) for the shear-induced and
Ishihara and Yoshimine (1992) for the volumetric-induced components of settlement, there is a
lack of simplified procedure to quantify the last mechanism which is attributed to the loss of ground
beneath a building’s foundation due to the formation of “sediment ejecta”. As a result, this abstract
contributes to this workshop toward the theme of (1) the current state-of-the-practice and its
limitations.

Contrary to the prior experimental studies using centrifuge tests, 1g shake table testing can
reliably produce ejecta at different scales. Figure 1 presents liquefaction-induced sediment ejecta
examples from 1g shake table tests at different scales, in which the height of model grounds ranged
from 20 to 290 cm.

H=75cm

H=20cm H=51cm H=290cm H=290cm

Small Moderate Large
Height of Model
Figure 1. Observed ejecta in 1g shake table tests.

This abstract presents the preliminary results obtained based on a series of scaled shake table
tests at UNR to explore the significance of sediment ejecta on liquefaction-induced foundation
settlement. It highlights the limitations of current methods for estimating liquefaction-induced
settlement and the missing component of sand ejects which can yield unreliable predictions of soil-
foundation system response.

RESPONSE TO PROMPT 1. WHAT IS THE STATE-OF-THE-ART?

The current practice lacks a simplified procedure to quantify the loss of ground beneath a
building’s foundation due to the formation of “sediment ejecta”. Although the occurrence of sand

118
ejecta has been reported in many past earthquakes, the first systematic survey of sand ejecta was
conducted by Bardet and Kapuskar (1991) in the Marina District of San Francisco after the 1989
Loma Prieta earthquake. In recent work, we have collected and analyzed 56 cases of observed
ejecta near residential buildings or commercial structures during 20 earthquakes since the 1989
Loma Prieta earthquake to develop some statistical understanding of ejecta occurrence (Buhl &
Motamed 2020). The effort generated some insight on the ejecta occurrence, though it didn’t yield
any substantial findings on the relation to the observed foundation settlements mainly as a result
of insufficient information about the sub-surface soil conditions, key soil properties, earthquake
ground motions, and pore water pressure responses. Therefore, this abstract suggests a laboratory-
based approach to reproduce this phenomenon by conducting scaled shake table tests that will
produce significant sediment ejecta, thus enabling its effects to be studied.

Over the past several years, we have performed a series of exploratory mid-scale shake table
tests at UNR on liquefaction-induced model building settlements to generate preliminary data
which are briefly presented hereafter. For example, Figure 1 demonstrates the capabilities of 1g
shake table tests to reliably produce ejecta-induced building settlements in models with heights
ranging from 20 to 290 cm. In addition, according to the histogram presented in Figure 2(a), the
measured ejecta volumes were mainly smaller than 10 cm³ and larger ones were rarely observed.
The smaller ejecta volumes were more commonly observed especially in the free-field and
adjacent to the foundation. Figure 2(b) shows the observed correlation between the volume of
ejecta and the tilt of the foundation indicating a direct correlation.

(a) (b)
Figure 2. (a) Distribution of recorded ejecta with their distance categories free-field, adjacent
and near the foundation, (b) differential ejecta volume versus tilt of the foundation (Buhl et al.
2021).

CONCLUDING REMARKS AND RECOMMENDATIONS

The preliminary work presented in this abstract suggests the following findings on the effects
of sand ejecta on liquefaction-induced foundation settlement: (1) the use of field observations to
study the significance of sand ejecta requires a more rigorous measurement of the ejecta foundation
which can be incorporated in databases such as NGL when documenting future earthquakes, (2)

119
there is a need for complementary laboratory tests using 1g shake table in the area of surface
manifestation of liquefaction and sand ejecta.

ACKNOWLEDGEMENTS

The author would like to acknowledge the support of multiple sponsors including PEER
(Pacific Earthquake Engineering Research Center), DFI (Deep Foundations Institute), and Ram
Jack to conduct the presented shake table tests.

REFERENCES

Bardet, J. P. & Kapuskar, M. (1991). The Liquefaction Sand Boils in the San Francisco Marina
District During the 1989 Loma Prieta Earthquake. International Conferences on Recent
Advances in Geotechnical Earthquake Engineering and Soil Dynamics. 19.
Bray, J. D., & Macedo, J. (2017). 6th Ishihara lecture: Simplified procedure for estimating
liquefaction-induced building settlement. Soil Dynamics and Earthquake Engineering, 102,
215-231. doi:10.1016/j.soildyn.2017.08.026.
Bray, J. D., Boulanger, R. W., Cubrinovski, M., Tokimatsu, K., Kramer, S. L., O’Rourke, T.,
Rathje, E., Green, R. A., Robertson, P. K., & Beyzaei, C. Z. (2017a). U.S.—New Zealand—
Japan International Workshop, Liquefaction-Induced Ground Movement Effects, University
of California, Berkeley, California, 2-4 November 2016, PEER Report 2017-02. Pacific
Earthquake Engineering Research Center, University of California, Berkeley, CA.
Bray, J.D., Markham, C.S., & Cubrinovski, M. (2017b). Liquefaction Assessments at Shallow
Foundation Building Sites in the Central Business District of Christchurch, New Zealand. Soil
Dynamics and Earthquake Engineering J., V. 92, 153-164,
http://dx.doi.org/10.1016/j.soildyn.2016.09.049.
Buhl, M., & Motamed, R., (2020). Collection and Statistical Analysis of Case History Data on
Liquefaction-Induced Soil Ejecta near Buildings during Past Earthquakes. Geo-Congress
2020: Vision, Insight, Outlook, Minneapolis MN, February 25-28, 2020, GSP 318, pp. 334-
345.
Buhl, M., Jahed Orang, M., & Motamed, R., (2021). An Exploratory Study on the Effects of
Liquefaction-Induced Sand Ejecta on Foundation Settlements based on Moderate-Scale Shake
Table Tests. Geo-Extreme 2021, Savannah GA, August 15-18, 2021, GSP 329, pp. 288-297.
Ishihara, K & Yoshimine, M (1992). Evaluation of Settlements in Deposits Following Liquefaction
during Earthquakes. Soils and Foundations, 32(1), 173-188.

120
 METRICS FOR USE IN EVALUATING LIQUEFACTION POTENTIAL AND
CONSEQUENCES UNDER SEISMIC LOADING CONDITIONS

I. M. Idriss
University of California, Davis, CA, USA
[email protected]

ABSTRACT

Since the early 1960s, when systematic approaches for evaluating liquefaction during
earthquakes were initiated, the emphasis has been on using the shear stress for both demand and
capacity when evaluating the liquefaction potential during earthquakes. This was driven by the
fact that the only information regarding capacity available at that time was from triaxial cyclic
tests, which had been started by H. Bolton Seed and Kenneth L. Lee at the University of California
at Berkeley1. The first analysis of liquefaction was for the Niigata site in 1965/1966, in which the
demand was estimated in terms of shear stresses induced during shaking using recently developed
site response procedures. Up to that point there was no mention of acceleration (PGA), at the
ground surface or at any depth.

Therefore, the metric for what has been called "the stress approach" for evaluating the
liquefaction potential during earthquakes is shear stress and not PGA.

The PGA at the ground surface came into the picture, circa 1967, when developing a means to
estimate the shear stress (and not the acceleration) at a given depth so that, again, the demand can
be compared to the capacity, which was still being measured in either cyclic triaxial or cyclic direct
simple shear tests. Only the surface PGA was involved; never the PGA at any depth. That lead to
the development of the Seed-Idriss simplified liquefaction evaluation procedure, which
necessitated "invoking" the use of a "stress reduction factor, rd". That allowed estimating the shear
stress induced by shaking at the depth where liquefaction was estimated to have been triggered for
the then few available2 case histories having or not having surface evidence of liquefaction.

Dobry and colleagues in 1981 proposed the use of shear strain as an alternative metric. While
fundamentally strain is a "superior" metric than stress, it was difficult to estimate the strain level
at the depth at which liquefaction had or had not been triggered for the case histories available
then.3 Other metrics have been proposed over the years for evaluating liquefaction triggering or its
consequences, the earliest of which is the Arias Intensity (AI), then the cumulative absolute
velocity (CAV), and more recently the Housner spectral intensity, using the pseudo relative
velocity or the pseudo absolute acceleration, among others.
The other aspect to consider in this regard is the fact that the natural case histories noted earlier
have increased significantly in the past 25 or so years and have been augmented by additional
"case histories" obtained through testing of physical models, particularly in the centrifuge.

1
References are not included in this Abstract but will be included in the presentation.
2
In 1967, there were only 23 cases with observed surface evidence and only 12 cases with no observed evidence of
liquefaction.
3
By the early 1980s, the number of case histories had more than quadrupled.

121
Except for a few, the available natural case histories have no recording below the ground surface
and many have no recordings within a few kilometers. The physical models are typically well
instrumented vertically and laterally.

Whatever metric is to be used, it is essential that there be means to estimate the value of that metric
at the depth at which liquefaction is considered to have been triggered, or not triggered.

Therefore, it would be useful to examine the recordings available from downhole arrays, such as
the array at Port Island (Figure 1a) and the array at Treasure Island (Figure 1b), where recordings
were obtained at 4 and 6 depths, respectively. Many other arrays are also available from
California, Japan, Alaska etc. It is hoped that colleagues who have completed relevant centrifuge
tests will make a number of their case histories available.

The values of the metrics listed above – shear stress, shear strain, AI, CAV, spectral intensity etc.
– will be calculated from as many arrays as possible prior to the start of the Workshop in
September. Shear strain and shear stress will be computed from array recordings using the
procedure introduced by Zeghal and colleagues in 1995 when the vertical spacing is adequate and
if a strain-compatible shear modulus can be reasonably estimated.

Plots, observations, summaries, conclusions and recommendations will be presented at the

Workshop.

The intent of this effort is to gain insight about how these metrics vary with depth and with other
parameters to facilitate their use in interpreting case histories and in developing procedures for
forward evaluations of liquefaction potential and consequences.
Shear Wave Velocity, V S - m/sec Shear Wave Velocity, Vs -- m/sec
0 200 400 600 800
200 250 300 350 0
0 0 Sand (Fill)
Fill (gravel)

Fill (sand with gravel) 10

10 10
Clay
(Young Bay Mud)
20
20 Clay 20
Depth Below Ground Surface - m

30
Depth Below Ground Surface - m

Sand
Sand
30 30
Sand with Gravel 40
Clay
40 40 (Old Bay Mud)
50

50 50
Sand
60

60 60
Clay 70

Measured Shear Wave Velocities Sand

70 70 (Kokusho & Matsumoto, 1997) 80
Clay (Old Bay Mud)

80 Sand with Gravel 80 90 Weathered Rock

Strong Motion
Instrument Rock Half-Space
VS = 1,830 m/s Strong Motion
90 90 100 Instrument

(a) Port Island (b) Treasure Island

Figure 1. Soil profile, measured shear wave velocities and depths at which strong motion
instruments had been installed at the (a) Port Island Site and (b) Treasure Island Site.

122
 TOWARD IMPROVED ASSEMSSMENTS OF LIQUEFACTION SUSCEPTIBILITY
AND SEVERITY

Kohji Tokimatsu
Tokyo Soil Research Co. Ltd., Meguro-ku, Tokyo, JAPAN
[email protected]

INTRODUCTION

This extended abstract addresses partly “1. The current state-of-the practice and its limitations”
and partly “3. Options for future susceptibility models that could be used.” More specifically, this
abstract discusses issues somehow related to: PROMPTS 2, 5, and 7.

RESPONSE TO PROMPT 5: How should the next generation of liquefaction

susceptibility models be formulated to advance the state-of-the-art?

The current liquefaction susceptibility evaluations are based on field case histories of
occurrence or nonoccurrence of soil liquefaction. Those case histories, therefore, sometimes lack
information regarding the degree/severity of soil liquefaction which is more important for seismic
design of soil structure systems. Moreover, any data from those case histories plotted far above
the boundary line separating occurrence and nonoccurrence in the correlation between cyclic stress
ratio and either normalized SPT N-value, CPT resistance, or shear wave velocity, do not play an
important role; and only those close to the marginal condition control the position of the boundary.

The next generation of liquefaction case history datasets should, therefore, include information
regarding not only the occurrence or nonoccurrence of soil liquefaction but also the degree/severity
of soil liquefaction such as, for example, liquefaction-induced ground settlements for level
grounds, permanent displacements for inclined grounds, and settlement or tilting of buildings.
This enables one to establish a more advanced field performance estimate with emphasis placed
not only on susceptibility but also on degree/severity of liquefaction. This also enables all field
case histories much effective, regardless of their positions relative to the boundary line. An attempt
along the line but using centrifugal experiment data can be found elsewhere, e.g., for estimating
liquefaction-induced settlement and tilting of buildings with spread foundations on sandy deposits
(Tokimatsu 2019).

RESPONSE TO PROMPTS 2 & 7: What are the consequences of incorrectly

assessing liquefaction susceptibility, and under what conditions are these
consequences most acute? Can you describe a case where liquefaction
susceptibility was assessed using methods beyond those typically applied, given
the importance of the project and consequences of liquefaction?

We sometimes assess liquefaction susceptibility of Pleistocene sandy deposits given the

importance of the project, although liquefaction is unlikely, generally because their SPT N-values
are high due to long-term stress-strain history and aging effects. Nonetheless, they occasionally
show N-values lower than thought, leading to unexpected estimation where soil liquefaction is
likely during very strong shaking if based only on any of the current SPT based correlations. In
those cases, we tried to use Vs in addition to SPT for liquefaction evaluation and to make a

123
comprehensive decision regarding the susceptibility and consequences, as Vs can reflect somehow
the stress-strain effects on liquefaction resistance (Tokimatsu & Uchida 1990).

Uchida et al. (2019) compiled recent geotechnical and geophysical field tests in Japan and
examined the relation between SPT and Vs, and showed that Pleistocene sands have Vs about 10-
70% higher than Holocene sands for the same SPT N-value, e.g., 180-420 m/s versus 160-250 m/s,
for N=20. The increase in Vs of Pleistocene sands suggests that the shear wave velocity may indeed
reflect the long-term stress-strain history and aging effects more than the SPT N-value. The larger
variation in Vs of Pleistocene sand, at the same time, indicates that the effects of the long-term
stress-strain history effects may significantly vary depending on local site conditions. Although
further studies are needed to confirm this tendency, a new liquefaction susceptibility method using
both SPT N-value and Vs may be useful to somehow compensate for the stress-strain history
effects not fully reflected in the SPT N-value.

It is important to note that not only incorrectly assessing liquefaction susceptibility but also
inexperienced dynamic response analysis may lead to a wrong estimate of design strong motions
and spectra occasionally far different from those actually expected for structures and buildings to
be constructed at the site. It is therefore useful to create a website compiling worldwide downhole
strong motion datasets including soil liquefaction and cyclic mobility, if not exists, which can
make available to anyone who would like to enhance his skill for estimating design strong motions
and spectra for liquefiable sites.

CONCLUDING REMARKS AND RECOMMENDATIONS

The next generation of liquefaction case history datasets should include information regarding
not only the occurrence or non-occurrence of soil liquefaction but also the degree/severity of soil
liquefaction in order to establish a more advanced and comprehensive field performance estimate
with emphasis placed not only on susceptibility but also on consequences of liquefaction. In order
to enhance the reliability in liquefaction estimates particularly for sands having experienced long-
term aging effects, a hybrid use of both SPT N-value/CPT resistance and Vs may be useful to
compensate somehow for the resulting increase in liquefaction resistance not fully reflected in the
penetration resistance. It is desirable to create a website compiling worldwide downhole strong
motion datasets including soil liquefaction and cyclic mobility, which can make available to
anyone who would like to enhance his skill for estimating design strong motions and spectra for
liquefiable sites.

REFERENCES

Tokimatsu, K., (2019). A simplified procedure to estimate liquefaction-induced settlements of

buildings, Earthquake Geotechnical Engineering for Protection and Development of
Environment and Constructions, CRC Press, 5297-5304.
Tokimatsu, K., & Uchida, A. (1990). Correlation between Liquefaction Resistance and Shear
Wave Velocity. Soils and Foundations, 30(2), 33-42.
Uchida, A., Tokimatsu, K., & Tsujimoto, K. (2019). A consideration on shear wave velocity
estimated by N-value. AIJ Journal of Technology and Design, 25(59), 119-122.

124
 CHALLENGES AND LIMITATIONS OF LIQUEFACTION ASSESSMENT IN THE
PACIFIC NORTHWEST

Nason McCullough, Donald Anderson, Sean Shin, Maddie Heidari, Menzer Pehlivan
Jacobs Engineering Group
[email protected], [email protected]

Seismic hazard assessments in the Pacific Northwest present several challenges in practice
when evaluating native soil deposits for the potential and effects of liquefaction. There are many
sites where deep (greater than 80 feet), loose silty and/or sandy soils have the potential to liquefy
and affect the performance of our civil infrastructure. Though pore pressure generation or
liquefaction of these deeper deposits may have minimal surface manifestations, pore pressure
generation and liquefaction may play a key role in the assessment of the performance of deep
shafts or driven piles at the deeper depths. Not only do these deep soil deposits have the potential
to affect the performance of structures due to pore-pressure generation, such as loss in capacity
and post-seismic settlement, but these deeper deposits present challenges in sampling and
minimizing disturbance when laboratory testing is required to confidently evaluate pore pressure
generation.

This presentation summarizes Jacobs’ experience in the Pacific Northwest when evaluating
pore-pressure generation and liquefaction of deep sandy and silty deposits to assess their impact
on the seismic performance of structures. For our projects, we regularly test soil samples in Cyclic
Direct Simple Shear (CDSS) to evaluate the performance and calibrate CDSS results to advanced
numerical models (such as PM4Sand and PM4Silt in FLAC) used to estimate the performance of
structures. This presentation will present a summary of our findings from select projects.

The presentation will also summarize the challenges we have encountered in performing this
work, including sampling and minimizing the disturbance of samples, confirming quality
samples prior to testing, evaluating stress histories, discussion on stress- versus strain-based
testing, and the calibration of advanced constitutive models. One example is in regards to the
occurrence of deep clean sand deposits, which being a clen sand precludes typical relatively
undisturbed sampling methods. Therefore, in order to calibrate numerical models for these
deposits, more reliance is given to empirical charts, which themselves were not intended for
depths greater than 50 feet. We will share our experiences of doing these calibrations, as well as
present on the similarities, as well as differences, that we have observed in the performance of
silty sands compared to clean in regards to pore-pressure generation, post-cyclic strength, and
post-cyclic volumetric strain, and the uncertainties with these values.

125
 DYNAMIC BEHAVIOR OF THE TREASURE ISLAND NATURAL SHOALS

Uri Eliahu, Shah Vahdani, Pedro Espinosa, Stefanos Papadopulos, David Teague and
Christopher Stouffer
ENGEO Incorporated, San Francisco, CA, USA
[email protected], [email protected], [email protected],
[email protected], [email protected], [email protected]

BENEFITS OF DETAILED CHARACTERIZATION OF GEOLOGICAL UNITS

Treasure Island is located in the central San Francisco Bay, immediately north of Yerba Buena
Island, between the active San Andreas and Hayward faults. Treasure Island was constructed by
placing hydraulic sand fill within a perimeter of rock dikes. The hydraulic fill consists of loose to
medium-dense sand and its dynamic behavior is captured well by simplified conventional
analytical methods. The fill was placed over a natural shoal deposit consisting of varying layers of
silty to clayey sand with interbedded lenses of highly plastic clay. Standard-of-practice post-vibro-
compaction CPTs demonstrated that the fill can readily be densified. The underlying shoal, which
was of the same origin as the hydraulic fill, was not densified by high-energy vibro-compaction.
This study was undertaken to further investigate the dynamic behavior of the shoal deposit.

The objectives of this study are: (1) to evaluate the potential for pore-pressure generation and
characterize the stress-strain response in the shoal deposit and the resulting lateral movements, and
(2) to estimate the magnitude of settlement which may occur due to post-cyclic reconsolidation of
the shoal deposit. The results of this study demonstrate that the dynamic behavior of the shoal
differs from that of the fill and is controlled by variations in clay content, unique soil structure due
to the site-specific depositional environment, and biological activity.

Field testing included vibro-compaction using Direct Power Compaction (DPC) in an

instrumented test section. Geologic characterization of the shoal included detailed logging (SEM,
optical microscopy, thin sections and descriptions of sedimentary structures) of continuous
samples recovered using Dames-and-Moore sampling equipment. Laboratory analyses on high-
quality samples included cyclic direct simple shear testing, constant-rate-of-strain consolidation
testing, and triaxial testing with shear-wave-velocity measurements. Evaluations included seismic
site-response analyses, lateral deformation analyses using two-dimensional finite-element and
finite-difference models, and comparisons with observed seismic performance of similar sites
during past earthquakes.

The study found that the shoal deposit possesses commonly overlooked characteristics such as
clay cementation, irregularly distributed clay inclusions, a variable amount of silty to clayey fines
forming a matrix between sand grains, and unique structure caused by depositional processes and
biological activity (i.e., bioturbation). Figure 1 shows a detailed geological logging of the shoal
sample.

126
Figure 1. Detail from about 40.75 feet in the lower shoal unit showing variation in fines filling
pores. Some areas contain significant fines (upper left) while other regions contain few fines in
pores (lower left).

Independent 2D dynamic simulations were performed with the finite-difference program

FLAC and the finite-element program PLAXIS, resulting in very similar and consistent results.
The shoal is found to be more resistant to seismically induced lateral deformation than would be
predicted by simplified models. The work included a comprehensive program to select seven input
ground motions representative of the relative contributions of faulting scenarios determined from
the probabilistic hazard assessment for this site.

Based on site-specific field investigation programs, full-scale field vibro-compaction testing,

laboratory testing, geological interpretations, dynamic numerical analyses using PLAXIS and
FLAC, empirical data, and review of earthquake effects on similar waterfront sites during past
earthquakes, the estimated lateral movement would be insignificant at distances exceeding
approximately 250 feet from the shoreline.

Additional testing and evaluations

On-going development at Treasure Island has allowed further testing of the dynamic behavior
of the shoals. The Treasure Island ground improvement program included surcharging to
compress soft bay deposits which underlie the shoal. These soils are locally known as Young Bay
Mud (YBM). Following the surcharge program the project team investigated whether further,
incidental improvement of the shoal was achieved. The same sampling techniques were used as
in the original shoal study. Additional cyclic simple shear testing was also carried out. Figure 2
shows the results of the cyclic testing plotted against number of cycles to initiate liquefaction.

As shown on Figure 2, in general the surcharge improved the resistance of the shoal against
excess pore built-up. In addition, using the results of site-response analysis to calculate cyclic
stress ratios (CSR), instead of the empirically based standard of practice CSR, it was demonstrated
that the shoal will not undergo widespread softening and volumetric strain under the Maximum
Considered Earthquake (MCE)

127
 Figure 2. Resulting CSR/CRR for Shoal Samples Before and After Surcharge.

CONCLUDING REMARKS AND RECOMMENDATIONS

Based on our findings at Treasure Island, we believe that the industry will be well served by
not over-relying on CPTs and CPT-based software alone to determine the performance of a site
under seismic loading. On projects where excessive conservatism creates a substantial financial
burden, we recommend introducing very detailed geological characterization of potentially
liquefiable materials, cyclic laboratory testing of materials which can be sampled, and site-
response analysis calibrated with historical data. These evaluations, although more expensive and
time consuming than the current state of practice, can improve the financial viability of projects
and create designs that allow resources to be redirected to sustainable construction.

ACKNOWLEDGEMENTS

Special thank you to Treasure Island Development Group for supporting technical excellence
and advancement, and to our technical contributors Dr. Michael Beaty, GE, Dr. Steve Dickenson,
PE, Dr. Juan Pestana, UC Berkeley, Dr. Michael Riemer, UC Berkeley, Dr. Jonathan Bray, UC
Berkeley, Dr. Nicholas Sitar, UC Berkeley.

REFERENCES

ENGEO (2015). Geotechnical Characterization Report (GCR), Treasure Island Major Phase 1
Sub-Phase 1B, 1C, 1E Infrastructure Improvements, San Francisco, California, Project No.
7091.000.000, December 22, 2015.
ENGEO (2015). Geotechnical Data Report (GDR), Treasure Island Major Phase 1 Sub-Phase 1B,
1C, 1E Infrastructure Improvements, San Francisco, California, Project No. 7091.000.000,
December 22, 2015.

128
LIMITATIONS OF THE SIMPLIFIED PROCEDURES AND PRACTICE IN THE PNW: A
REVIEW OF THE CURRENT STATE OF THE PRACTICE AND RECENT CASE
HISTORIES

Jason D. Bock, PE and Scott M. Schlechter, PE, GE, D.PE

GRI, Tigard, Oregon, United States
[email protected], [email protected]

THE CONCERN

The last 20+ years have shown extensive growth in liquefaction assessment following the 1996
NCEER and 1998 NCEER/NSF workshops on liquefaction evaluation of soils. These workshops
paved the way for the development of many tools practicing engineers now use to evaluate
liquefaction. These techniques are firmly founded on a review of case histories from numerous
sites both with and without liquefaction, and the simplified procedures and software tools have
provided many engineers with a quick tool to estimate liquefaction potential. However, regions of
the country such as the Pacific Northwest have areas where many, if not most, of the soils, fall
outside the range of the sand and silty sand case histories used to develop the simplified
procedures. As described further below, newer procedures are available for fine-grained soils.
However, due to budget/time-constrained projects and/or lack of familiarity, many engineers do
not perform adequate testing (both in situ and laboratory) to better characterize and estimate soil
behaviors. Using sand-based procedures in early stages of a project has had significant impacts on
project budgeting and outcomes, particularly for projects where conceptual or preliminary
engineering efforts drive decisions for environmental permitting constraints. The paragraph below
provides additional detail on the issues related to fine-grained soils.

The simplified procedures are based on liquefaction assessment utilizing either corrected
standard penetration test blow counts, cone penetration test (CPT) tip resistance, or shear wave
velocities to evaluate a soils resistance to liquefaction. This estimated liquefaction resistance is
known as the cyclic resistance ratio (CRR) and is compared to the level of earthquake-induced
loading, known as the cyclic stress ratio (CSR). While this method of estimating CRR has shown
value and provides reasonable results for cohesionless soils, it ignores the nuances of fine-grained
soil behavior, such as the effects of plasticity and overconsolidation (OCR). Several guidelines
have been developed (Idriss & Boulanger 2008; Anderson et al. 2007), which aim to provide
insights into soil characterization and provide methodologies for estimating fine-grained soil CRR
while incorporating the effects of OCR. These guidelines set much of the framework needed for
incorporating more advanced testing such as cyclic direct simple shear testing into our liquefaction
evaluations and provide reasonable lower bounds to CRR trends in fine-grained soils.

RESPONSE TO PROMPT(S)

What are the consequences of incorrectly assessing liquefaction susceptibility,

and under what conditions are these consequences most acute?

Following the development of the simplified methods, many conceptual-level evaluations now
include screening-level evaluations based on worst-case sand-like behavior. While resulting in safe

129
designs, this approach increasingly drives early decision-making, often resulting in substantially
more expensive designs that could have potentially been avoided with the early adoption of more
robust in situ and laboratory testing. The consequences of these conditions are most acute in large
infrastructure projects where conceptual or preliminary level engineering changes permitting
approaches or structure type decisions are commonly made early.

What data resources would help improve understanding?

The large majority of the data currently publicly available comes from areas outside of
Oregon/Washington. To alleviate some of the inherent conservatism in liquefaction evaluations,
regional trends in fine-grained soil behavior and access to a database of regional information could
greatly benefit the geotechnical community. In this regard, the Oregon Department of
Transportation, with support from Portland State University and local practitioners, is currently
developing a database of cyclic testing for northwest silt soils. The database will include
preliminary dataset trends that help form the framework for a better picture of silt behavior.

CASE HISTORIES

Willamette Water Supply Program – Raw Water Facility

The Willamette Water Supply Program (WWSP) is currently in the process of design and
construction of a new resilient water supply for several communities near Portland, Oregon,
including those served by the Tualatin Valley Water District, the City of Beaverton, and the City
of Hillsboro. The project includes the construction of new water treatment plants, tanks, pipelines,
and improvements to an existing water intake system in Wilsonville known as the Raw Water
Facilities (RWF).

A conceptual-level evaluation of RWF was initially completed and included a

liquefaction/lateral spreading evaluation of the slope located between the river and the existing
structures (clear well, pumps, etc.). Previous explorations at the site indicated the soils are
predominantly Willamette Silt (WS) soils underlain by stiff, higher plasticity silts and clays of the
Troutdale Formation. Conceptual-level liquefaction evaluations considered the Willamette Silt as
“sand-like” and employed the use of simplified procedures. Results of the analyses indicated the
WS soils were susceptible to liquefaction and low factors of safety for slope stability, indicating
flow failure type behavior and significant lateral displacements (greater than 10 feet).

The next project phase completed by the final design team employed the use of cyclic direct
simple shear testing (CDSS) and the SHANSEP framework (Ladd & Foote 1974; Idriss &
Boulanger 2008) as well as Finite Difference Modeling. Results of this next phase of analysis and
testing indicated the silt soils would undergo limited cyclic softening and much smaller strain
levels than estimated using the simplified procedures. Similar to other case histories in the area,
the resulting deformations were closer to an order of magnitude lower than previously estimated.

Bonneville Power Administration Transmission Line Crossing Project

The Bonneville Power Administration (BPA) is tasked with managing power generation on 31
dams in the Columbia River Basin and provides power to local utilities such as Portland General
Electric (PGE). Overall, BPA represents approximately 28% of the power generation within the

130
northwest. As part of the agency’s seismic resiliency goals, a concept-level evaluation of multiple
existing transmission line towers crossing the Columbia and Willamette Rivers in Portland and
Troutdale was completed. The various project sites included a wide variation in soils ranging from
stiff, high plasticity clays to loose, low fine content sands. Preliminary evaluations utilizing CPT-
based simplified procedures characterized the relatively low PI silt as having “sand-like” behavior
and high susceptibility to liquefaction and large lateral spreading displacements.

Following the initial screening, a more advanced design phase was contracted to further
evaluate the liquefaction and lateral spreading risk as well as provide mitigation alternatives. To
evaluate the behavior of the silt and clay soils, the project team performed a robust CDSS testing
program. The results of this testing program allowed for the development of project-specific CRR
trends for the silt and clay soils as well as site-specific correlations between plasticity and Ic. This
additional work to accurately characterize the soil behavior indicated the silt and clay soils were
typically not susceptible to liquefaction but would undergo cyclic softening (Beaty et al. 2014).
Utilization of the cyclic testing resulted in an overall reduction in the need for ground improvement
in the fine-grained soils.

CONCLUDING REMARKS AND RECOMMENDATIONS

As our knowledge of liquefaction and cyclic softening continues to grow, we recommend that
the engineering community is strongly encouraged to actively adopt new methodologies and
techniques for fine-grained soils. We recommend the engineering practice focuses first on
adequately characterizing cyclic behavior of soils at early stages of the project if these behaviors
are likely to drive decision making. The current practice of using a sand-like screening approach
often leads to expensive decisions that can often be avoided. As a secondary recommendation,
development of regional data trends and correlations could greatly benefit the local engineering
practices as well as reduce uncertainty for owners.

REFERENCES
Anderson, D. L., Byrne, P. M., DeVall, R. H., Naesgaard, E., & Wijewickreme, D. (2007). Task
force report, geotechnical design guidelines for buildings on liquefiable sites in accordance
with NBC 2005 for greater Vancouver region. Greater Vancouver Liquefaction Task Force.
Beaty, M., Schlechter, S., Greenfield, M., Bock, J., Dickenson, S., Kempner Jr., L., & 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.
Idriss, I. M., & Boulanger, R. W. (2008). Soil liquefaction during earthquakes: Earthquake
Engineering Research Institute, Oakland, California.
Ladd, C.C. & Foote, R. (1974). A new design procedure for stability of soft clays. Journal of the
Geotechnical Engineering Division. ASCE, Vol. 100, No. GT7. pp. 763-786.

131
 LIQUEFACTION SUSCEPTIBILITY OF A LOW PLASTICITY SILTY SOIL UTLIZING
CYCLIC DIRECT SIMPLE SHEAR TESTING

Samuel S. Sideras
Shannon & Wilson, Inc, Portland, Oregon, USA
[email protected]

CYCLIC TESTING FOR LIQUEFACTION EVALUATIONS IN TRANSITIONAL SOILS

The liquefaction susceptibility of transitional and intermediate soils is commonly evaluated

based on soil index properties (e.g. Boulanger & Idriss 2006; Bray & Sancio 2006) which generally
classify a soil’s dynamic behavior and potential for pore pressure-induced strength loss as
liquefiable (sand-like) or nonliquefiable (clay-like). However, as noted by Armstrong and Malvick
(2016), the differences in the soil indices and soil behavior thresholds utilized by the various soil
index test-based methods can lead to inconsistent evaluations of liquefaction susceptibility in
typical practice. In addition, the available case history and laboratory database is limited with
respect to transitional soils.

This paper describes a comprehensive laboratory testing program that was utilized to evaluate
the liquefaction susceptibility and the dynamic behavior of a low plasticity silty soil for a major
infrastructure project in the Portland Oregon area. The following sections detail the soil index
test-based liquefaction susceptibility evaluations typically performed in practice and the results of
the cyclic laboratory tests. The results of the laboratory testing program illustrate the benefits of
site-specific cyclic laboratory testing for transitional soils that are not well defined in literature.

SOIL INDEX TEST-BASED LIQUEFACTION SUSCEPTIBILITY EVALUATIONS

Soil index tests including Atterberg Limit, water content, and fines content testing were
performed on 47 select samples obtained from the transitional soils encountered at the site using
Standard Penetration Test split-spoon and Shelby-tube samplers. The index test results for the
transitional soils included natural moisture contents that ranged from 29 to 51 percent (average of
42 percent), plasticity indices that ranged from nonplastic to 19 (average of four), and fines
contents that ranged from 35 to 83 percent (average of 61 percent). The soil was generally
characterized as a low plasticity sandy silt to silty sand.

A summary of the index test data is provided in Figure 1 which plots the plasticity index as a
function of the ratio of the water content to the liquid limit. Also included on the plot are zones
delineating the liquefaction susceptibility criteria thresholds of Boulanger and Idriss (2006) and
Bray and Sancio (2006). Based on the criteria of Boulanger and Idriss (2006) approximately 80
percent of the tested samples would classify as “sand-like” and can be evaluated with typical
simplified semi-empirical-based liquefaction methods. Per Bray and Sancio (2006),
approximately 98 percent of the samples would be susceptible to liquefaction and strength loss
due to generation of excess pore pressures during cyclic loading.

132
Figure 1. Summary of soil index test data and Boulanger and Idriss (2006) and Bray and Sancio
(2006) liquefaction-susceptibility criteria

Both index test-based methods typically used in practice suggest that the transitional soils at
the site are predominantly composed of liquefaction-susceptible soils. However, both Boulanger
and Idriss (2006) and Bray and Sancio (2006) note that index test-based procedures should be
considered as general screening guides and should be supplemented with cyclic laboratory testing
of field samples if the seismic behavior of the soils in question are a critical aspect of the project
design. Cone penetration tests (CPTs) were used to evaluate the resistance to liquefaction using
the semi-empirical liquefaction triggering framework of Boulanger and Idriss (2015). The soils
had an average normalized corrected CPT tip resistance, q c1ncs, of approximately 85, indicating a
low resistance to liquefaction. Given the prevalent nature of the transitional soils at the site and to
reduce the uncertainty in seismic deformation analyses performed for the project, a cyclic
laboratory testing program was performed as described in the following section.

CYCLIC LABORATORY TEST RESULTS

The cyclic laboratory testing program included harmonic stress-controlled cyclic direct simple
tests (CDSS) performed on select Shelby-tube samples. The index properties of the Shelby-tube
samples selected for CDSS testing included samples with natural moisture contents that ranged
from 35 to 51 percent (average of 43 percent), plasticity indices that ranged from nonplastic to ten
(average of three), and fines contents that ranged from 44 to 83 percent (average of 64 percent).
The samples used in the CDSS test program were obtained between depths of 20 to 90 feet below
the ground surface. Tests were performed at confining pressures ranging from approximately one
to three atmospheres with applied cyclic stress ratios (CSRs) between 0.15 and 0.3.

A summary of the CDSS test results is provided in Figure 2 which plots the CSR as a function
of the number of cycles (N) to reach a single amplitude shear strain of three percent for each test
that reached the given peak shear strain threshold. Also included in Figure 2 is the CSR vs N
relationship for clean sands derived from Boulanger and Idriss (2015) assuming a qc1ncs of 85 and
overburden effective stresses ranging from 1 to 3 atmospheres. The site-specific CDSS program
shows the soil has significantly more resistance to strain accumulation under cyclic loading than

133
implied from semi-empirical methods developed for clean sands. The site-specific cyclic
laboratory testing program indicated that the seismic behavior of the transitional soils at the site
were best represented by cyclic softening type evaluations (e.g. Boulanger & Idriss 2007) as
opposed to the clean sand liquefaction analyses framework as suggested by the index test-based
liquefaction susceptibility evaluations.

Figure 2. CSR vs number of uniform cycles to 3 percent shear strain from CDSS tests with
comparison to Boulanger and Idriss (2015) clean-sand semi-empirical relationship

CONCLUDING REMARKS AND RECOMMENDATIONS

This paper described the use of cyclic laboratory testing to characterize the dynamic behavior
of a transitional low plasticity silty soil in practice. This example highlights the known limitations
of index-test based liquefaction susceptibility evaluations currently used in practice and the utility
of performing site-specific cyclic testing in soils that are not well defined in literature. The
practicing geotechnical community would benefit from additional research in the cyclic response
of transitional and intermediate soils.

ACKNOWLEDGEMENTS

The presented CDSS tests were performed by Oregon State and Portland State Universities.

REFERENCES

Armstrong, R.J. & Malvick, E.J. (2016). Practical considerations in the use of liquefaction
susceptibility criteria. Earthquake Spectra, 32(3), 1941-1950.
Boulanger, R.W. & Idriss, I.M. (2006). Liquefaction susceptibility criteria for silts and clays.
Journal of Geotechnical and Geoenvironmental Engineering, 132(11), 1413-1426.
Boulanger, R.W. & Idriss, I.M. (2007). Evaluation of cyclic softening in silts and clays. Journal
of Geotechnical and Geoenvironmental Engineering, 133(6), 641-652.
Boulanger, R.W. & Idriss, I.M. (2015) CPT-based liquefaction triggering procedure. Journal of
Geotechnical and Geoenvironmental Engineering, 142(2), 04015065.
Bray, J.D. & Sancio, R.B. (2006). Assessment of liquefaction susceptibility of fine-grained soils.
Journal of Geotechnical and Geoenvironmental Engineering, 132(9), 1165-1177.

134
 LIQUEFACTION SUSCEPTIBILITY OF GRAYS HARBOR SILTS

Matthew Gibson
Clarity Engineering, LLC, Vashon, Washington, USA
[email protected]

INTRODUCTION

A pontoon casting basing was constructed in 2010 in the town of Aberdeen, Washington as
part of the SR 520 Floating Bridge Replacement project. The project site is located within the
Aberdeen tidelands on the north shore of Grays Harbor near the lower reach of the Chehalis River
at latitude 46.9648, longitude -123.8337. Subsurface conditions along the river are typical of a
fluvial depositional environment with embedded sand channels amongst silt and clayey silt
deposits. Given the liquefaction and lateral spread hazard form a Cascadia Subduction Zone
earthquake, an extensive geotechnical exploration program was conducted yielding borehole/CPT
pairs, index testing, and cyclic direct simple shear tests. This data set is presented here to
demonstrate the performance of various methods to assess liquefaction susceptibility and cyclic
mobility of silts with plasticity indices ranging from 11 to 94 and soil behavior type index (Ic)
greater than 3.0. In addition, the data set illustrates boundaries for transitional behavior that is
dependent on PI, CSR, and shear strain cycles.

SITE LOCATION AND SUBSURFACE CONDITIONS

32 borings and 26 CPTs were performed at the site resulting in 27 boring/CPT pairings (Landau
2009; SW 2011). The subsurface soils consisted of 10 to 15 feet of fill underlain by about 90 to
110 feet of very soft to stiff silt with embedded sandy stream channels occurring at various
elevations and is further underlain by very dense gravels and sandstone. Laboratory tests consisted
of moisture contents, sieve analyses, hydrometers, Atterberg limits, 1D consolidation tests,
undrained unconsolidated and consolidated triaxial tests, direct simple shear tests, cyclic direct
simple shear (CDSS) tests, vane shear, pressure meter tests, and shear wave velocity testing.

LIQUEFACTION SUSCEPTIBILITY ASSESSMENT

Simplified methods

Cone penetration test data was interpreted using procedures outlined by Robertson (2016)
resulting in the derived parameters normalized tip resistance (Qtn), normalized friction ratio (Fr),
and soil behavior type index (Ic). A comparison of plasticity index versus Ic for paired borings and
CPTs was made where no trend is observed between, however, per the Bray and Sancio (2006)
liquefaction susceptibility criteria, some silts are classified as potentially liquefiable while others
require further testing (See Figure 1A). The Boulanger and Idriss (2006) susceptibility method
indicates all soils would exhibit clay-like behavior. Robertson’s (2016) large strain soil behavior
descriptors indicates that all silts at the site would exhibit fine-grained contractive behavior (see
Figure 1B).

135
CDSS testing

A CDSS test program was performed to further assess the liquefaction susceptibility of the
silts (See Table 1). Cyclic stress strain curves were interpreted at 10, 20, and 30 cycles to evaluate
trends in excess pore pressure ratio (Ru) and maximum cyclic shear strain amplitude versus
plasticity index and cyclic stress ratio (See Figure 2).

Figure 1. A) Silt cyclic susceptibility to liquefaction on cyclic softening per Bray and Sancio
(2006), and B) large strain behavior index per Robertson (2016).

Table 1. Summary of CDSS test program

Sample Depth (m) PI LL PL MC Ic CSRs

H-08p-09, S-15 13.95 11 45 34 45 3.02 0.12, 0.2, 0.24
H-18-09, S-15 15.42 14 41 27 40 2.93 0.25, 0.3, 0.34
H-07-09, S-15 15.4 17 42 25 41 3.08 0.25, 0.3, 0.34
H-29p-09, S-13 11.95 23 59 36 59 3.06 0.2, 0.3, 0.34
H-16-09, S-13 12.25 94 173 79 173 3.80 0.29, 0.34, 0.39

For a CSR demand of 0.3 and 15 cycles, the Bray and Sancio (2006) liquefaction susceptibility
method was able to identify silts susceptible to classic sand-like liquefaction and silts transitioning
towards cyclic mobility. However, as CSR and shear strain cycles are increased, higher plasticity
silts begin to show liquefiable behavior (Figure 2A and 2C). This suggests that existing
susceptibility models may contain a level of inherent triggering criteria based on the limits of
empirical data. The data also suggests a relationship between PI and a threshold CSR at which Ru
and shear strains rapidly increase and liquefiable behavior is exhibited (Figure 2B and 2D).

CONCLUDING REMARKS AND RECOMMENDATIONS

This data set confirms that CPT based liquefaction susceptibility and cyclic mobility
evaluations based solely on Ic are potentially unconservative especially for moderate to large
magnitude earthquakes. However, the CPT was able to identify potentially liquefiable silts based

136
on the large strain behavior index. Further research to expand empirical data in susceptibility
models, susceptibility models based on large strain behavior, and CSR/PI threshold behavior is
recommended. It is also recommended that practitioners and researchers using CPT data evaluate
initial liquefaction susceptibility consider large-strain, contractive and transitional boundaries in
addition to Ic criteria. CPT data plotting in the translational and contractive regions signify that in-
situ sampling for Atterberg Limits and/or CDSS testing should be performed. This is especially
important for small to medium sized projects in which CPT is the first and sometimes last
investigation tool used to determine soil behavior during site-specific seismic hazards.

Figure 2. Results of CDSS test program.

ACKNOWLEDGEMENTS

All site data was collected by exploration programs executed by Landau Associates and
Shannon & Wilson, Inc. for The Washington Department of Transportation. The author
appreciates discussions with Claire Gibson that improved this manuscript.

REFERENCES

Boulanger, R. W., and I. M. Idriss. (2006). Liquefaction Susceptibility Criteria for Silts and Clays.
Journal of Geotechnical and Geoenvironmental Engineering, 132 (11): 1413–1426.
Bray, J. D., & R. B. Sancio. (2006). Assessment of the Liquefaction Susceptibility of Fine-Grained
Soils. J. Geotech. Geoenviron. Eng., 132 (9): 1165–1177.

137
Landau Associates. (2009). Geotechnical Data Report, Volume I SR 520 Pontoon Construction
Design-Build Project Aberdeen Log Yard Aberdeen, Washington, August 17, 2009.
Robertson, P. K. (2016). Cone penetration test (CPT)-based soil behaviour type (SBT)
classification system — an update. Canadian Geotechnical Journal, 53 (12): 1910–1927.
Robertson, P. K. (2022). Evaluation of flow liquefaction and liquefied strength using the cone
penetration test: an update. Can. Geotech. J., 59 (4): 620–624.
Shannon & Wilson, Inc. (2011). Released for Construction, Geotechnical Engineering
Recommendations, State Route (SR) 520. Pontoon Casting Facility Report, Aberdeen,
Washington, February 18, 2011.

138
THE IMPACTS OF ANALYZING DEEP SAND AND TRANSITIONAL SOIL PROFILES
WITH STATE OF THE PRACTICE METHODS

Brice Exley
Haley & Aldrich, Seattle, WA, U.S.A.
[email protected]

STATE-OF-THE-PRACTICE METHODS, AND THEIR LIMITATIONS

When assessing the potential for liquefaction triggering, most engineers rely solely on stress
based simplified methods such as Boulanger and Idriss (2014). These methods are then applied
broadly to sands of any fines content. If fine grained soils are considered at all, the penetration
resistance based simplified methods are also applied to low plasticity fine grained soils that may
qualify as “sand-like” using simplified screening methods such as Boulanger and Idriss (2006).
Correspondingly, when relying on cone penetration tests, the application of sand-like behavior
often occurs up to a soil type behavior index, Ic, of 2.6 (Robertson & Write 1998). Treating these
transitional soils as sand-like may under predict the cyclic resistance of these soils, therefore over
predicting the resulting impacts of cyclic loading on the soil profile.

If an engineer does identify that a transitional soil may have a cyclic behavior that is distinctly
different than a sand deposit, it is often difficult to reliably characterize the monotonic shear
strength of the soils which would facilitate the use of methods such as Jana and Stuedlein (2021).
The impact of the potential mischaracterization of the cyclic behavior of these soils is magnified
when they are part of a deep soil profile (e.g. 100+ feet of sand) where the cumulative effects of
predicted liquefaction can be quite large and expensive to mitigate.

These deep soil profiles are particularly problematic to the practicing engineer as the simplified
methods do not currently lead them to consider either the system response or explicitly capture the
horizontal variability of a soil deposit, which may play a significant impact on the seismic
performance of the site (Cubrinovski et al. 2019).

These limitations in combination with compounding factors of safety and hazard levels will
often leave an engineer with a design soil profile that is generally predicted to liquefy until a very
dense bearing layer is encountered. While partial saturation has been documented to significantly
increase the cyclic resistance of soils, there is limited precedence or documentation on the long-
term reliability of maintaining partial saturation below a water table. Engineers therefore do not
account for this behavior in practice. This often leaves the engineer with needing to provide
recommendations for deep foundations or significant quantities of ground improvement that may
not be necessary.

Deep foundations of an economical size can quickly become unfeasible due to the requirement
in the International Building Code (ICC 2017) to treat “deep foundation elements standing
unbraced in air, water or fluid soils” as columns without lateral support until they are 5 feet into
stiff soil or 10 feet into soft soil. A soil deposit that that has a thick continuous layer of predicted
liquefaction can quickly be controlled by buckling. Even if a project is not subject to the IBC, the
current state-of-the-practice often results in excessive deep foundation lengths depending on how
liquefaction induced downdrag is handled.

139
 The use of ground improvement can be problematic as well. Terminating ground improvement
in a liquefiable soil profile is often met with resistance from both building officials and engineers
who are uncertain on how to analyze the performance of such a system. The use of economical
ground improvement is further impacted by the loss of the assumption of strain compatibility for
discrete elements such as aggregate piers, and the resulting cost associated with jet grouting or soil
mixing.

RESPONSE TO PROMPT(S)

Have you experienced a case where the determination of susceptibility proved to

be pivotal, and what were the considerations associated with the application of
typical (i.e., state-of-the-practice) susceptibility procedures?

The susceptibility of a site to liquefaction has been pivotal on many projects I have participated
in. These are often impacted by deep interbedded alluvial soil profiles where the assumed behavior
of transitional soils can be critical. For the transitional soils we will at times evaluate them with a
combination of the CPT and advanced laboratory testing resulting in cyclic resistance profiling
based on yield stress ratio profiles (YSR) and the monotonic undrained shear strength. In these
cases, the CRR is evaluated using a power law as presented by Jana and Stuedlein (2021) or
Dickenson et al. (2021) and as shown in Figure 1 for a low plasticity fine grained soil deposit.
However, even if clay-like behavior is identified, determining an accurate profile of the YSR is
problematic in silts.

0.30

0.25

0.20
CSR/YSR^0.7

0.15

0.10

0.05

0.00
1.00 10.00 100.00
Number of Uniform Cycles, N

Cycles to 3% Power Law fit, a = 0.22, b = -0.15

Figure 1. Cyclic Resistance Curve Varying CSR with N for γ= 3%. These are not corrected for
rate effects.

For generally deep profiles, typical susceptibility procedures are often applied with some
acknowledgement about their limited applicability for depths greater than about 80 feet. However,
they are generally still used over the full soil profile.

140
Can you describe a case where liquefaction susceptibility was assessed using
methods beyond those typically applied, given the importance of the project and
consequences of liquefaction?

We have used both 1D and 2D site response analyses with models such as PM4SAND and
PM4SILT to capture the system response of a site. However, the use of 1D analyses in soft or
loose soil deposits is often met with concerns of potentially overdamping due to an incorrect
assumption of horizontally infinite and continuous layers.

CONCLUDING REMARKS AND RECOMMENDATIONS

The application of sand-like liquefaction susceptibility criteria to low plasticity fine grained
soils can often result in an underestimation of the cyclic resistance and post-cyclic strength of the
soil. When coupled with deep profiles that have a significant system response this may result in a
design profile consisting of a large degree of false positives, significantly increasing construction
costs. This has a cascading impact on all of the resulting analyses involving the impacts of
liquefaction. With further refinement of the applicability of transitional soil behavior to the CPT
and improved methods of characterizing the yield stress ratio in these soils a reliable estimation of
the CRR may be readily estimated without site specific advanced testing. However, the system
response of a soil profile remains a challenge for practitioners to address reliably and efficiently.
Development of simplified methods that can capture the system response or more explicit guidance
on the appropriate application of equivalent one-dimensional analyses using soil models intended
to capture this behavior would be beneficial. Finally, further study on the long-term reliability of
partial saturation in various environments may support increased cyclic resistance ratios for use in
design.

ACKNOWLEDGEMENTS

While no direct research funding has attributed to these observations, several private
organizations have patiently worked with project teams over several projects to address many of
the limitations presented herein.

REFERENCES

Boulanger, R. W. & Idriss, I. M. (2014). CPT and SPT Based Liquefaction Triggering Procedures.
Report NO. UCD/CGM-14/01. University of California at Davis.
Cubrinovski, M., Rhodes, A., Ntritsos, N., & Van Ballegooy, S., (2019). System response of
liquefiable deposits. Soil Dynamics and Earthquake Engineering, 124, 212-229.
Dickenson, S.E., Khosravifar, A., Beaty, M., Bock, J., Moug, D., Schlechter, S., & Six, J. 2021.
Cyclic and post-cyclic behavior of silt-rich, transitional soils of the Pacific Northwest; a
database for geo-professionals in practice and research. Salem, OR: Oregon Department of
Transportation.
International Code Council 2017. 2018 International Building Code.
Jana, A., & Stuedlein, A.W. (2021). Monotonic, cyclic, and postcyclic responses of an alluvial
plastic silt deposit. Journal of Geotechnical and Geoenvironmental Engineering, 147(3).
Robertson, P. K., & C. E. Wride. 1998. Evaluating cyclic liquefaction potential using the cone
penetration test. Canadian Geotech. J. 35 (3): 442–459.

141
 CHALLENGES OF LIQUEFACTION ASSESSMENT AT CALIFORNIA’S DAMS

Erik J. Malvick
California Department of Water Resources, Division of Safety of Dams, Sacramento, CA, USA
[email protected]

INTRODUCTION

The California Division of Safety of Dams (DSOD) regulates over 1240 dams in California.
The program was established in 1929 to protect life and property with respect to dam safety. Our
authority extends to dams owned by individuals, companies, utilities, local governments, and the
State. Over 70-percent of the dams regulated are over 50 years old. The total population at risk for
all dams exceeds 5,000,000. Given the high risk, DSOD relies on independent evaluations to
provide dual verification of dam safety. Eighty-percent of the dams are earthfill, which makes,
liquefaction assessments critical to dam safety.

While liquefaction evaluation techniques continue to evolve and improve, DSOD finds there
to be challenges that limit the confidence in our reviews. The most critical issues relate to geology
and materials at our dams, specifically the presence of gravel or soils (soils with moderate fines
contents, 20–50 percent or moderate plasticity, plasticity indices 12–20). The challenges lead to
inconsistent results and contrasting results that can have significant impacts on the final
determination for a dam. The most critical issue is the potential bias that can occur due to gravels
when they are at a high percentage or too large and impact in-situ measurements. Conversely, it is
uncertain what the potential maybe for over-conservatism with intermediate soils that may not be
susceptible to any strength loss.

The discussion that follows focuses on these topics and their impact on dam safety. While
specific case histories are not presented, DSOD has numerous to support most scenarios described.
This will be closed with recommendations towards goals and research that can help achieve more
consistency in practice related to liquefaction.

DETAILED RESPONSES

Most common drivers of challenges related to gravel

Gravel is a significant challenge for liquefaction evaluations at California’s dams. Methods for
characterizing gravels in-situ were standardized starting with the Becker Penetration Test (BPT)
work by Harder and Seed (1986) and have continued progress to modern instrumented BPTs
(iBPT) (e.g., DeJong et al. 2017 and Ghafgazi et al. 2017). While progress should lead to
consistency in techniques, DSOD find evaluations can be inconsistent due to an expectation that
early methods should give the same results as new methods. This leads to hesitation in practice
and lack of consensus regarding BPT tests overall and variable conclusions when multiple
techniques are considered. However, the inconsistencies are likely indicative of the complication
of characterizing gravels versus sands, which have generally been more studied and understood.

Specifically, we see limited trust in the applicability of BPTs because the core correlations are
derived from and tied directly to sand while the methods are intended for gravel. It is understood
this limitation was born of necessity given the limited availability of case histories of gravel

142
liquefaction. In practice, however, the BPT becomes an extrapolation when used with gravels since
a reliable correlation with Standard Penetration Tests (SPTs) are not likely with gravel presence.
Thus, BPTs may be limited in their ability to capture the real impact of particle size, gravel content,
coefficient of uniformity (Cu), and deposition processes that may impact their behavior. Each of
those factors may not only affect the in-situ resistance (e.g., CRR) but also the loads needed to
trigger liquefaction (e.g., CSR).

Additional challenges using BPT are tied to practicality. Accessibility at dam sites can be a
challenge for BPTs related to rig size, road size, dam height, and more. Further, the availability of
BPTs can impact schedule and costs. Thus, engineers often rely on extrapolative techniques, such
as SPTs measuring blows-per-inch, often without consideration for gravel size or deposition. Cone
Penetrometer Tests are often used with visibly damaged probes. Statistics are often relied on to
average results and justify the inclusion of all data without thought to what one extreme outlier
might have on results. DSOD has seen SPT data used where a shoe has been blocked or recovery
is zero despite high blow-counts. SPTs have been used in 60-percent gravel materials and even in
cobbles. This leads to widely inconsistent interpretations of liquefaction susceptibility. One
evaluation of a hydraulic fill dam had about 25-percent gravel with some SPT blow counts (N1-
60,cs) exceeding 60 with many below 20. The consultant concluded liquefaction was not an issue
because the mean N1-60,cs was greater than 30. However, it is hard for DSOD to trust a hydraulic
fill to be that competent given 75-percent of the material was finer than gravel.

The challenge comes down to limited guidance regarding the use of samplers versus particle
size. In one extreme, a dam recently explored with IBPTs was known to be constructed with
gravels approaching cobble sized.

Challenges related to intermediate soils

DSOD also faces lesser challenges with intermediate soils, soils generally thought not to be
liquefiable but not clearly in the realm of cyclic softening. Malvick et al. (2014) documented
DSOD’s protocols for evaluating these “transition” soils for strength loss susceptibility,
liquefaction or cyclic softening. Liquefaction and cyclic softening literature were compared to
identify areas in common and those where interpretations would vary. As a regulator, it is critical
to recognize all valid procedures. Our process focuses on identifying if soil behaviors are clay-like
or sand-like and fines-controlled or coarse-controlled. Given the mission to protecting life and
property, the absence of literature explicitly noting conditions where strength loss may not occur
or how the behavior from liquefaction to cyclic softening can transition, a conservative approach
is taken. Yet, these soils can have low N1-60,cs, yet quality samples show high density with Cu > 4.
We speculate these materials are not susceptible strength loss and in one case concluded so.

Impact on Dam Safety

The consequences of incorrectly assessing liquefaction susceptibility can be catastrophic at

dams. The Lower San Fernando Dam incident in 1971 shows us what liquefaction can do at a dam.
While there were no consequences, the incident highlights California’s major cities and the dams
that serve them are often in the state’s most seismically active regions. DSOD has focused re-
evaluations of dams for seismic loads and liquefaction as engineering methods improve and
evolve. In some cases, dams considered safe historically have been found not to be. Yet gravel

143
bias, especially when SPTs are misapplied, can lead to extreme risks and long-drawn-out processes
to reach consensus on what needs to be done with a dam. Mitigation and repairs can be costly as
$100 million or more repairs become more common.

RECOMMENDATIONS

It is recognized that the Next Generation Liquefaction project is intended to mimic the Next
Generation Attenuation (NGA) project used for ground motion parameters. DSOD sees merit in
the objective but does have concerns if there is intent for multiple models to be developed towards
the same purpose. For instance, there may be more value in one liquefaction susceptibility and
triggering model contributed by all potential authors to exemplify a full consensus. With the NGA,
DSOD sees inconsistencies related to varying choices of models, applicability, and other biases.
With liquefaction, the contrast of methods had been problematic while there may be more
acceptance and consistency with a consensus such as that of Youd and Idriss (1997).

In addition, there is a need for supplementary guidance, clarifications, and consensus when
dealing with gravel influenced and intermediate soils. Research may be needed or compiled to
clarify the conditions where gravelly or intermediate soils can liquefy or lose strength.
Consideration may be needed for the impact of Cu, soil matrix, sampler validity, sample quality,
and exclusion when strength loss may not need to be considered at all. For practitioners, one
inclusive reference may provide an advantage as the number of engineers and projects grow
resulting from ageing infrastructure, continued advances in engineering knowledge, and retirement
of our respected experts. In dam safety, consistency can make a huge difference in correctly
mitigating risks or potentially save a dam owner significant costs.

REFERENCES

DeJong, J.T., Ghafghazi, M., Sturm, A.P., Wilson, D.W., den Dulk, J., Armstrong, R.J., Perez, A.,
& Davis, C.A. (2017). Instrumented Becker Penetration Test, I: Equipment, Operation, and
Performance. ASCE J. Geotechnical and Geoenvironmental Engineering, 143(9).
Ghafghazi M., DeJong J.T., Sturm A.P., & Temple C.E. (2017). Instrumented Becker Penetration
Test, II: iBPT- SPT correlation for liquefaction assessment in gravelly soils. ASCE J.
Geotechnical and Geoenvironmental Engineering, 143(9).
Harder, L. F., Jr., & Seed, H. B. (1986). Determination of Penetration Resistance for Coarse-
grained Soils Using the Becker Hammer Drill. Coll. of Engrg., U of Calif., Berkeley, Calif.
Rep. No. UCB/EERC-86/06.
Malvick, E. J., Armstrong, R. J., & Martin, K. M. (2014). Estimating the dynamic shear strength
of soils for dam seismic safety analyses. The Journal of Dam Safety, 12(3), 9-20.
Youd, T. L., & Idriss, I. M. (eds.) (1997). Proceedings of the NCEER Workshop on Evaluation of
Liquefaction Resistance of Soils. Nat. Center for Earthquake Engrg. Research., Rep. No.
NCEER-97-0022.

144
 EVALUATING LIQUEFACTION SUSCEPTIBILITY FOR NUCLEAR POWER PLANT
SITES

Thomas Weaver
U.S. Nuclear Regulatory Commission, Rockville, MD, USA
[email protected]
The Nuclear Regulatory Commission (NRC) was established by the Energy Reorganization
Act of 1974 with the authority for all licensing and related regulatory functions associated with
civilian uses of nuclear materials and facilities. There are currently 93 operating electrical
generating nuclear power reactors at 55 sites in the United States, and according to the Nuclear
Energy Institute, the Nuclear Regulatory Commission (NRC) could receive 12 or more license
applications for advanced reactors per year as early as 2025. Liquefaction susceptibility
evaluations are performed for nuclear power plant sites when assessing the possible effects of
ground shaking on facility foundations.

10 C.F.R. §100.23 (1997) requires a power plant license applicant to evaluate the site for
liquefaction potential. This regulation does not provide details on how to assess liquefaction
susceptibility, liquefaction triggering, or consequences associated with liquefaction. Guidance to
licensees and applicants on implementing NRC’s regulations, techniques used by the NRC staff in
evaluating specific problems or postulated accidents, and data needed by the staff in its review of
applications for permits or licenses are documented in Regulatory Guides. Specific guidance for
evaluating liquefaction at a nuclear power plant site is found in Regulatory Guide 1.198,
“Procedures and Criteria for Assessing Seismic Soil Liquefaction at Nuclear Power Plant Sites,”
(US NRC 2003). The current regulatory position states that the applicant should identify soils that
might liquefy and that screening for liquefaction potential should include assessing susceptibility
by asking the following question: “Are potentially liquefiable soils present?” Although the
regulatory position does not describe specific susceptibility criteria, the Regulatory Guide
Discussion Section includes information on susceptibility criteria. Specifically, the discussion in
the guide states the following.

• Cohesive soils with fines content greater than 30 percent and fines that either (1) are
classified as clays based on the Unified Soil Classification system or (2) have a Plasticity
Index (PI) greater than 30 percent should generally not be considered susceptible to
liquefaction.

• Sands that have dual Unified Soil Classification system designations such as CL-ML, SM-
SC, or GM-GC are potentially liquefiable (Youd 1998).

• Other designations involving the “C” description, if the clay content is greater than 15
percent by weight and the liquid limit is greater than 35 percent and occurs at natural water
contents lower than 90 percent (Wang, 1979), can be considered nonliquefiable.

The discussion also notes that gravelly soils are potentially vulnerable to liquefaction and that
most liquefaction risk is associated with Holocene deposits and uncompacted fills; however, a few
cases of liquefaction have been observed in Pleistocene and Pre-Pleistocene deposits. Some soils
considered nonliquefiable according to the criteria presented in the regulatory discussion may be

145
susceptible to liquefaction according to criteria proposed by Seed et al. (2003) and Bray and Sancio
(2006). Work is in progress to update Regulatory Guide 1.198. This update will result in
modifications to the discussion on liquefaction susceptibility.

There is potential for the NRC to support liquefaction susceptibility research. The NRC
provides research grants to develop a workforce capable of supporting the design, construction,
operation, and regulation of nuclear facilities. Recently, the emphasis of research supported by
grants has been activities relevant to civilian advanced nuclear reactors. Research areas of interests
include evaluating technical gaps and major uncertainties in assessing risk for advanced reactors
and characterizing low frequency, high consequence natural hazards for advanced nuclear
application. Some key information on the grants program from 2022 includes the following. The
notice of funding opportunity for research and development grants was issued in February 2022
with a closing date in April 2022. These grants have a project period of performance of three years
and an award ceiling of $500,000. Grant applications can be submitted by U.S. public or private
institutions of higher education, and all graduate students, faculty, principal investigators, and co-
principal investigators must meet U.S. citizenship requirements.

REFERENCES
Bray, J. D., & Sancio, R. B. (2006). Assessment of the Liquefaction Susceptibility of Fine-Grained
Soils. Journal of Geotechnical and Geoenvironmental Engineering, 132(9), 1165–1177.
https://doi.org/10.1061/(ASCE)1090-0241(2006)132:9(1165)
Seed, R. B., Cetin, K. O., Moss, R. E. S., Kammerer, A. M., Wu, J., Pestana, J. M., Riemer, M. F.,
Sancio, R. B., Bray, J. D., Kayen, R. E., & Faris, A. (2003, April 30). Recent Advances in Soil
Liquefaction Engineering: A Unified and Consistent Framework. 26th Annual ASCE Los
Angeles Geotechnical Spring Seminar. Long Beach, California.
US NRC. (2003). Regulatory Guide 1.198 Procedures and Criteria for Assessing Seismic Soil
Liquefaction at Nuclear Power Plant Sites. US NRC.
https://www.nrc.gov/docs/ML0332/ML033280143.pdf
Wang, W. S. (1979). Some Findings on Soil Liquefaction. Water Conservancy and Hydroelectric
Power Scientific Research Institute.
Youd, T. L. (1998). Screening Guide for Rapid Assessment of Liquefaction Hazard at Highway
Bridge Sites (MCEER-98-0005; p. 92). Multidisciplinary Center for Earthquake Engineering
Research. State University of New York at Buffalo.

146
 INCORPORATING UNCERTAINTY IN SUSCEPTIBILITY CRITERIA INTO
PROBABILISTIC LIQUEFACTION HAZARD ANALYSIS

Andrew J. Makdisi
U.S. Geological Survey, Golden, CO, USA
[email protected]

INTRODUCTION

Most conventional approaches for assessing liquefaction triggering hazards generally rely on
simplified procedures that involve identifying liquefaction susceptible layers and calculating a
factor of safety against liquefaction (FSL) in each layer. Such procedures utilize deterministic semi-
empirical models for standard penetration test (SPT), cone penetrometer test (CPT), or shear wave
velocity (Vs)-based subsurface data. This general approach largely neglects considerable
uncertainties in ground shaking, as well as aleatory variabilities and epistemic uncertainties
inherent to liquefaction susceptibility and triggering prediction. A more robust methodology
known as probabilistic liquefaction hazard analysis (PLHA), integrates the full ground motion
hazard space with probabilistic forms of liquefaction triggering models (e.g., Boulanger & Idriss
2012) to compute of FSL profiles with consistent return periods (e.g., Kramer 2008). Multiple
PLHA computational platforms have been developed over the years, with the computational
framework from Makdisi (2021) serving as the basis for a new Liquefaction Hazard Tool under
development at the U.S. Geologic Survey (USGS).

Despite substantial improvements in recent years to the availability of seismic hazard data and
probabilistic triggering and effects models, the issue of incorporating uncertainty in characterizing
liquefaction susceptibility remains a challenge. Most compositional susceptibility criteria (i.e.,
whether the soil exhibits sand-like behavior) currently in use are presented as deterministic bounds
based on in situ or laboratory test data; similarly, determination of soil saturation is often based on
a single groundwater level from in situ testing. As a result, the same types of binary decisions must
be made in PLHA as in more conventional methods. With the expansion and availability of field
and laboratory data pertaining to liquefaction through resources such as the Next Generation
Liquefaction (NGL) project, an improved set of susceptibility models may be possible for CPT,
SPT, and Vs-based applications. Presented here is a brief discussion on how probabilistic
susceptibility modeling can be accommodated in PLHA calculations, as well as how the use of
multiple models can be leveraged within a logic tree to improve the representation of epistemic
uncertainty in liquefaction hazard analysis.

USE OF PROBABILITY OF SUSCEPTIBILITY IN LIQUEFACTION ANALYSIS

The PLHA framework is based on calculating the annualized rate of non-exceedance of a

given factor of safety value, Λ𝐹𝑆𝐿 , in a given soil layer, as follows:

𝑁𝑚 𝑁𝑝𝑔𝑎

Λ𝐹𝑆𝐿 (𝑓𝑠𝐿 ) = ∑ ∑ P[𝐹𝑆𝐿 < 𝑓𝑠𝐿 |𝑠𝑢𝑠𝑐, 𝑃𝐺𝐴𝑖 , 𝑀𝑤,𝑗 ] ∙ P[𝑠𝑢𝑠𝑐] ∙ Δλ𝑝𝑔𝑎𝑖 ,𝑚𝑤,𝑗
(1)
𝑗=1 𝑖=1
where PGA and Mw are peak ground acceleration and magnitude, respectively, and are utilized to
represent the peak cyclic stress and the effects of loading duration on incremental pore pressure

147
rise. Δλ𝑝𝑔𝑎𝑖 ,𝑚𝑤,𝑗 is the incremental joint annualized rate of exceedance of PGA and Mw, which can
be obtained by disaggregating the PGA hazard curve at a range of return periods. The probability
of susceptibility term P[𝑠𝑢𝑠𝑐] involves the joint probability of the soil layer (1) exhibiting sand-
like behavior and (2) existing below the groundwater table. However, current compositional and
saturation susceptibility criteria are largely deterministic, and therefore P[𝑠𝑢𝑠𝑐] is generally
implied to be either 0 or 1. Nevertheless, Equation (1) is formulated to accommodate uncertainties
in susceptibility as probabilistic criteria are further developed and improved upon. The full PLHA
calculation is repeated for a wide range of FSL values, yielding a soil profile of FSL non-exceedance
hazard curves, from which FSL profiles of uniform return periods can be extracted to estimate the
consequences of liquefaction.

For the purposes of determining soil saturation, it is important to note that groundwater levels
can vary over both seasonal and longer-term temporal scales. Shallower soil layers may see higher
probabilities of saturation during periods of intense rainfall or due to sea-level rise, and lower
probabilities occur during periods of drought. Greenfield and Grant (2020) found that groundwater
level uncertainties can be important contributors to uncertainty in liquefaction potential at the
regional scale, and that groundwater levels can be modeled as a normally distributed random
variable. Groundwater estimates can also come from multiple sources, and site-specific
measurements from SPT, CPT, or well data can be supplemented, where available, with regional-
scale groundwater models to form a more complete picture of the saturation likelihood in a given
soil layer.

Although most compositional criteria are not currently formulated or presented

probabilistically, they do acknowledge a degree of uncertainty in the process. For SPT data, Bray
and Sancio (2006) identified a zone of “moderately susceptible” materials in their laboratory test-
based criteria, whereas Idriss and Boulanger (2008) supplemented their more conservative
plasticity index-based (PI) recommendations with a transition zone between sand- and clay-like
soils. Using visual approximation, Kramer (2008) quantified these transition zones and presented
both criteria in terms of a susceptibility index that varies smoothly between 0 and 1. Such
modifications allow for (1) a pseudo-probabilistic representation of the susceptibility that can serve
as a starting point for more rigorous statistical models in the future, and (2) the incorporation of
multiple models in the characterization of susceptibility criteria via a weighted logic tree, capturing
a potentially important source of epistemic uncertainty in liquefaction hazard analysis.

Figure 1 introduces how such a set of susceptibility models could be implemented, along with
multiple subsequent models for liquefaction triggering, in a logic tree similar to the approach
utilized extensively in the USGS National Seismic Hazard Model (NSHM). FSL hazard curves can
be computed for each logic tree branch, and the overall mean non-exceedance rate of a given FSL
value is computed as the weighted mean of all branches. This framework also allows for FSL hazard
curve fractiles to be calculated to provide more information about the uncertainties surrounding
the PLHA calculation. Logic trees are an important tool for uncertainty quantification and are as
yet underutilized in liquefaction hazard analysis. Although the current slate of liquefaction models
limits our ability to thoroughly quantify epistemic uncertainty in liquefaction-related problems,
implementing the logic tree approach when possible is nonetheless important to both motivate and
accommodate future expansion in liquefaction model availability in any PLHA framework.

148
 Figure 1. Example logic tree for computing liquefaction triggering hazard curves.

CONCLUDING REMARKS AND RECOMMENDATIONS

The PLHA framework presents an important opportunity to advance liquefaction hazard

assessment in engineering practice. To make PLHA calculations more widely available to users,
the USGS is developing a computational platform that reflects the best available data and models
for capturing important aspects of uncertainty in both (1) ground shaking hazards and (2)
liquefaction triggering and effects prediction. At the same time, there is clear and equally important
need to formally quantify and incorporate uncertainties in liquefaction susceptibility in the PLHA
framework. A simple modification to the classic PLHA equation was presented that accounts for
the probability of susceptibility of a given soil layer, and a potential logic tree framework
highlighting the need to capture both aleatory and epistemic uncertainty in liquefaction
susceptibility characterization was discussed. The expansion of available data and models presents
a clear roadmap for developing a new class of susceptibility models that can meet these needs.

REFERENCES

Boulanger, R. W., & Idriss, I. M. (2012). Probabilistic Standard Penetration Test–Based

Liquefaction–Triggering Procedure. Journal of Geotechnical and Geoenvironmental
Engineering, 138(10), 1185–1195.
Bray, J. D., & Sancio, R. B. (2006). Assessment of the liquefaction susceptibility of fine-grained
soils. Journal of Geotechnical and Geoenvironmental Engineering, 132(9), 1165–1177.
Idriss, I. M., & Boulanger, R. W. (2008). Soil liquefaction during earthquakes. Earthquake
Engineering Research Institute, 237 pp.
Greenfield, M.W. & Grant, A. (2020). Probabilistic regional-scale liquefaction triggering
modeling using 3D Gaussian processes. Soil Dynamics and Earthquake Engineering, 134,
106159.
Kramer, S. L. (2008). Evaluation of liquefaction hazards in Washington state (No. WA-RD
668.1). Washington (State). Dept. of Transportation. Office of Research and Library
Services, 329 pp.
Makdisi, A. J. (2021). Liquefaction-targeted ground motion parameters. Doctoral Dissertation,
University of Washington, 363 pp.

149
 REGIONAL LIQUEFACTION SUSCEPTIBILITY ASSESSMENTS: DATA
COLLECTION NEEDS AND A FOCUS ON THE CENTRAL AND EASTERN U.S.

Christine Z. Beyzaei
National Institute of Standards and Technology, Gaithersburg, MD, USA
[email protected]

CURRENT STATE-OF-THE-PRACTICE AND LIMITATIONS

Liquefaction susceptibility assessments can be performed at the site-specific or regional scale.

Site-specific assessments typically utilize quantitative methods that rely on geotechnical data from
laboratory testing of soil samples retrieved in-situ (e.g., Bray & Sancio 2006; Boulanger & Idriss
2006). Regional assessments are generally more qualitative in nature and are based on factors that
influence susceptibility such as groundwater level, geology and depositional environment, age of
deposit, and fabric, or are based on applying site-specific methodologies at a regional scale (e.g.,
Youd & Perkins 1978; FEMA 2020).

Regional liquefaction susceptibility assessments are critical to improving the seismic resilience
of communities and infrastructure systems across the country. However, there exists a technical
gap between the detailed, quantitative site-specific studies carried out using laboratory testing data,
and qualitative, often proxy-based, regional studies. This extended abstract discusses two aspects
of regional susceptibility assessments: data collection needs to advance or improve existing
assessment methods, and limitations in applying existing state-of-the-practice methods, with a
focus on the Central and Eastern U.S. (CEUS).

DATA RESOURCES TO HELP IMPROVE UNDERSTANDING

Liquefaction susceptibility assessment methods are derived primarily from post-earthquake

field case histories and laboratory testing. Case histories require an observation, a ground motion
recording or estimate, and geotechnical data. However, immediately after an earthquake it is not
feasible to visit and photograph the entire affected area in person, due to time and safety
constraints. Developing the next generation of liquefaction susceptibility assessment methods,
including quantitative regional assessment methods, requires regional data collection and selection
of meaningful case history sites for detailed investigations. Extensive high-resolution aerial
photography after an earthquake event can provide essential data needed to meet these goals.
Consider the 2010-2011 Canterbury earthquake sequence (CES), which led to unprecedented
research advances. Within two days after major earthquake events in the CES, the New Zealand
government collected extensive aerial imagery throughout Christchurch and the surrounding
communities. The imagery was then made publicly available, accessible via Google Earth, and
maintained for over a decade following the events (NZGD 2022). Figure 1 shows the extent of
aerial imagery coverage and the imagery resolution at the ground scale. Extensive regional
coverage with high-resolution aerial imagery has enabled research investigations of CES post-
earthquake observations to continue to this day.

In addition to providing regional coverage, extensive aerial imagery allows researchers to

“revisit” sites years later and select critical, impactful case history sites for further investigation
and collection of quantitative geotechnical data. This is especially important for selecting sites that

150
perform well (i.e., where no ground failure is observed). For example, to investigate observations
of “no liquefaction” at silty soil sites in Christchurch, over 30 candidate sites were initially selected
and then narrowed down to 8 sites for development of detailed case histories (Beyzaei et al. 2018).
High resolution aerial images across the region enabled a large pool of candidate sites from which
the most impactful sites could be selected for further investigations. Aerial imagery of this extent
ensures that the ephemeral data from post-event observations are not lost and can later be used for
either regional or site-specific quantitative analysis.

(C)

(A) (B)

Figure 1. Extent of aerial imagery commissioned by the New Zealand Ministry of Civil Defence
and Emergency Management: (A) 4 Sept 2010 Darfield earthquake, acquired on 5 Sept 2010;
and (B) 22 Feb 2011 Christchurch earthquake, acquired on 24 Feb 2011 (NZGD 2022). (C) Inset
showing resolution of imagery at the ground level.

CHALLENGES WITH CURRENT METHODS AND POTENTIAL CONSEQUENCES

Existing methods and proposed frameworks for regional liquefaction susceptibility

assessments are typically based on examples from the Western U.S. and other areas of high seismic
hazard. However, there are several challenges in applying existing assessment methods to the
CEUS or other areas of low to moderate seismic hazard: 1) limited regional data availability (i.e.,
publicly available subsurface geotechnical and groundwater data), 2) practitioner and stakeholder
liquefaction hazard awareness, and 3) fewer earthquake events leading to the perception of
liquefaction hazard not being a “local” issue.

Improving practitioner and stakeholder awareness of liquefaction hazards and existing

liquefaction susceptibility assessment methods should be a primary goal, alongside research, to
advance technical knowledge and assessment models. Improving awareness and addressing the
perception of liquefaction not being a “local” issue increases the opportunities for stakeholder buy-
in and higher quality assessments in seismic or multi-hazard resilience planning.

151
 It is critical to get stakeholders onboard now to start planning for climate change impacts
affecting liquefaction susceptibility. Sea level rise has the potential to create larger or new areas
of susceptible soils, with studies demonstrating the effects of sea level rise on liquefaction
vulnerability for the Bay Area in California (USGS 2022) and Charleston, South Carolina (Ghanat
2021). A potential acute consequence of climate change impacts on liquefaction susceptibility is
that a region becomes more vulnerable to liquefaction hazard due to sea level rise. In areas of low
to moderate seismic hazard, if there is a lack of awareness or view that liquefaction is not a “local”
problem, communities may not have mitigation strategies in place and are then unprepared when
an earthquake occurs and damage ensues.

CONCLUDING REMARKS AND RECOMMENDATIONS

The next generation of liquefaction susceptibility models should close the gap between current
state-of-practice quantitative site-specific methods and qualitative regional methods. Extensive
aerial photography is key during post-earthquake reconnaissance and will allow for selection of
impactful case histories in the years after an event. Limited regional data availability and
practitioner and stakeholder awareness in low-to-moderate seismicity areas are challenges related
to the use of existing methods.

ACKNOWLEDGEMENTS

Any opinions, findings, and conclusions or recommendations expressed in this extended

abstract are those of the author and do not necessarily reflect the views of NIST.

REFERENCES

Beyzaei, C. Z., Bray, J. D., Cubrinovski, M., Riemer, M., & Stringer, M.E. (2018). Laboratory-
based characterization of shallow silty soils in southwest Christchurch. Soil Dynamics and
Earthquake Engineering, 110(2018), 93-109.
Boulanger, R. W. & Idriss, I. M. (2006). Liquefaction susceptibility criteria for silts and clays.
Journal of Geotechnical and Geoenvironmental Engineering, 132(11), 1413-1426.
Bray, J. D. & Sancio, R. B. (2006). Assessment of the liquefaction susceptibility of fine-grained
soils. Journal of Geotechnical and Geoenvironmental Engineering, 132(9), 1165-1177.
FEMA (2020). Hazus earthquake model technical manual: Hazus 4.2 SP3. Federal Emergency
Management Agency (FEMA), October.
Ghanat, S. T. (2021). Impact of projected climate-driven sea level rise on liquefaction vulnerability
in Charleston, SC. EERI Annual Meeting, virtual poster session.
NZGD (2022). Canterbury Maps, Collated Investigation Data: Aerial Photography. New Zealand
Geotechnical Database, http://www.nzgd.org.nz [accessed March 2022].
USGS (2022). Liquefaction and sea-level rise. United States Geological Survey, story map created
by T. Poitras, A. Grant, A. Wein, K. Knudsen, K. Befus, M. Erdman, K. Petersen,
https://www.usgs.gov/tools/liquefaction-and-sea-level-rise [accessed July 2022].
Youd, T. L. & Perkins, D. M. (1978). Mapping liquefaction-induced ground failure potential.
Journal of the Geotechnical Engineering Division, 104(No. GT4), 433-446.

152
 SHEAR WAVE VELOCITY SEISMIC SOIL LIQUEFACTION TRIGGERING
ANALYSIS–2022 UPDATE

Robert Kayen1, Makbule Ilgac2, Kemal O. Cetin3, Clinton Wood4, & Robb E.S. Moss5
1U.S.G.S., Menlo Park, CA 94720, and Department of Civil and Environmental Engineering,
University of California Berkeley, 94720, [email protected]
2Department of Civil and Environmental Engineering, University of California Berkeley,

[email protected]
3Dept of Civil Engineering, Middle East Technical University Ankara, [email protected]
4Dept of Civil Engineering, University of Arkansas, 72701, Arkansas, [email protected]
5Dept. of Civil and Env. Eng., California Polytechnic State University, [email protected]

THE CURRENT STATE-OF-THE-PRACTICE FOR VS TRIGGERING ANALYSIS

The shear wave velocity (Vs) of soil is a means to assess the triggering of seismic soil
liquefaction. Since the early 1990s, the size and quality of these data sets have grown enormously.
Based on these expanding data sets, we have actively worked to develop probabilistic models for
seismic soil liquefaction triggering using Bayesian analysis and system reliability methods.
Critical elements of any update of these triggering analyses involve building a modern case-history
catalog of legacy cases (e.g., Kayen et al. 2013 - KEA13) and inclusion of new case histories such
as the 2011 M9.0 Tohoku earthquake, 2010-2011 New Zealand-Canterbury earthquake sequence,
and other earthquake events less prominent in terms of Vs measurement.

Our current 2022 efforts to update the Vs -based database require characterization of the
mean and distributions of all relevant load and capacity parameters and understanding the
locations, details, and statistics of each case. The updated database currently consists of 537 case
histories (Table 1; Figure 1). When using surface wave data, we have computed averaged velocity
profiles of the critical layers using multiple inversion methods applied to the dispersion data. We
base the “Critical-layer” susceptibility determination on the analysis of nearby standard
penetration test (SPT), cone penetration test (CPT) data, geologic textual information, and the
characteristics of the Vs -profile. We digitized Vs profiles from the literature for legacy cases not
collected by the authors. Considerable effort has gone into improving the selection of unit weight
needed for computing effective and total stresses. One of the benefits of Vs analyses is that other
parameters such as Vs12m and Vs30m are available to model the mass participation parameter rd and
site-specific response analyses. Toward this objective, we developed a standard protocol for
processing data to minimize bias and measure the uncertainties of each input parameter.

153
 Table 1. Distribution of earthquake events within the Vs database
# of # of
Earthquake Earthquake
sites sites
1906 San Francisco 2 1989 Loma Prieta 49
1948 Fukui 11 1993 Hokkaido-Nansei-Oki 27
1964 Niigata 9 1993/1994 Kushiro-Oki/Kushiro 8
1968 Tokachi-Oki 4 1995 Hyogo-Nambu 83
1973 Miyagi-Ken Oki 11 1999 Chi-Chi, Taiwan 14
1975 Haicheng 5 1999 Izmit Earthquake 1
1976 Tangshan 24 2000 Tottori Seibu 3
1978 Miyagi-Ken Oki 8 2001 Geiyo-Hiroshima 5
1979 Imperial Valley 7 2002 Denali Fault 9
1980 Mid-Chiba 2 2005 Sanriku Minami 11
1981 Westmorland 7 2003 Tokachi-Oki 10
1983 Borah Peak 19 2003 Tokachi-Oki Aftershock 1
1983 Nihonhai-Chubu 8 2007 Niigata Chuetsu Oki 2
1983 Nihonhai-Chubu Aftershock 2 2008 Achaia-Elia 2
1986 Chiba-Ibaragi-Kenkyo 2 2010 Darfield 61
1986 Lotung Sequence 5 2010 Jiasian 1
1987 Chiba-Toho-Oki 1 2011 Christchurch 61
1987 Edgecumbe 2 2011 Tohoku 35
1987 Elmore Ranch 7 2011 Tohoku Aftershock 10
1987 Superstition Hills 7 2014 Napa Valley 1
Sum of Vs-based case history data: 537

This dataset is used to assess the likelihood that a site tips towards liquefaction or away
based on prior data and observations to estimate probabilities based on Bayesian inference. We
establish a limit state function with initial conditions of mean values of the entire dataset and assess
the likelihood of liquefaction occurrence by contrasting the individual site parameters with the
prior dataset. An example of a limit state function for shear wave velocity assessment was
presented in KEA13, KEA15 and is shown below (Equation 1). The parameters (e.g., amax, Vs, Mw,
σ'v, rd, etc.) are measured properties of load or capacity, and the Theta’s (𝛩 1, 𝛩2, etc.) are the
coefficients modeled through Bayesian analysis.
𝛩
𝑔𝑉𝑠1 = 𝛩1 𝑉𝑠12 + 𝛩3𝑙𝑛(𝐶𝑆𝑅) + 𝛩4 𝑙𝑛(𝑀𝑤 ) + 𝛩5 𝑙𝑛(𝜎’𝑣 ) + 𝛩6 𝐹𝐶 + 𝜀 (1)

154
 PL=50 %
0.7 0.7
Marginal
Marginal
0.6 0.6 Yes
Yes
0.5 0.5 No

CSRσ'v=100 kPa,Mw=7.5
No Kayen et
CSRσ'v,Mw

0.4 0.4
al. (2013)

0.3 0.3

0.2 0.2

0.1 0.1

0.0 0.0
0 100 200 300 0 100 200 300
Vs1 (m/s) Vs1 (m/s)

(a) (b)
Figure 1. 2022 Vs-based case history database (Effective stress-normalized velocity, Vs1 versus
magnitude and effective stress normalized cyclic stress ratio, CSRσ'v,Mw domain), b) on Vs1 versus
CSRσ'v=100 kPa,Mw=7.5 along with Kayen et al. (2013) Fifty-percent Probability of Liquefaction,
PL=50% curve. (dots, circles, and triangles represent liquefied, non-liquefied, and marginal sites,
respectively.)

ACKNOWLEDGEMENTS

This research was funded by the TUBITAK 2214-A International Research Fellowship
Program for Ph.D. students. Financial support was provided through the United States Geological
Survey (USGS), Pacific Earthquake Engineering Research (PEER) Center’s Lifelines Program,
and Research Center for Urban Safety and Security, Kobe University.

REFERENCES
Kayen, R., Moss, R. E. S., Thompson, E. M., Seed, R. B., Cetin, K. O., Der Kiureghian, A.,
Tanaka, Y. & Tokimatsu, K. (2013). Shear-wave velocity–based probabilistic and deterministic
assessment of seismic soil liquefaction potential. Journal of Geotechnical and Geoenvironmental
Engineering, 139(3), 407.

155
APPENDIX C: PRE-WORKSHOP POLL
RESULTS
Question 1 – Please indicate your primary roll:

25

20
Frequency

15

10

Question 2 – Please indicate your years of experience:

20

18

16

14
Frequency

12

10

0
0 to 5 5 to 10 10 to 15 15 to 20 20 to 25 25 to 30 30 to 35 35 to 40 40+
Years of Experience

156
Question 3 – Please distribute 100 points to the following quantities according to their
influence on a soil’s liquefaction susceptibility, with zero corresponding to no influence and
greater points corresponding to greater influence:
100

Mean
80 Maximum
Standard Deviation
Number of Points

60

40

20

Question 4 – How do you rate the reliability of cone penetration test-based liquefaction-
susceptibility interpretations, e.g., using the Soil Behavior Type Index, Ic? (choose one)
16

14

12

10
Frequency

0
Highly reliable Reliable Reliable in certain Reliable in certain Not generally Poor; do not trust
soils where soils where fines reliable
previous cyclic contents have
test data have also been
been obtained measured

157
Question 5 - Above what fines content do you consider soil behavior to be controlled by the
characteristics of fines (please report as percent)?

12

10

8
Frequency

0
10 to 15 15 to 20 20 to 25 25 to 30 30 to 35 35 to 40 40 to 45
Fines Content (%)

Question 6 – At what minimum fines content do you believe good quality undisturbed
samples can be obtained (please report as percent)?

158
Question 7 – How do you judge the susceptibility of interbedded soils?

1. Having an understanding of the depositional history, age and historical

performance is of outmost importance.
2. Depends on the thickness of the interbeds.
3. Carefully.
4. Depends on thickness of beds.
5. Combination of grain-size and Atterberg characteristics, blow counts,
strength, and stress history of soil profile are used. Where the surrounding
soils have lower strength and have physical characteristics of susceptible soils
then they are considered susceptible and softening and strength loss are
expected.
6. It depends, but typically would consider the susceptibility of the different
interbedded layers (if samples can be obtained from individual layers) and the
nature of the overall interbedding sequence at the site with depth and relative
to the groundwater table/zone of partial saturation.
7. I would evaluate susceptibility based on soil characteristics independent of
interbedding; however, I believe interbedding would have a significant impact
on system response.
8. Depends on the depositional and stratigraphic conditions (layer
thickness/length), and consequences of failure. If applicable, use in-situ
testing tools and assess susceptibility using engineering correlations.
9. Through use of CPT measurements and SPTs, if possible.
10. Cyclic shear.
11. With and without corrections, i.e. sensitivity to modeling.
12. If we simply talk about susceptibility (intrinsic soil characteristics which do
not have anything to do with the current soil conditions (e.g., Dr or water
content are not descriptive of susceptibility), I do not see any differences in
evaluating susceptibility of a single layer in a homogeneous profile vs that of
a layer within an interbedded soil. My opinion is different if we talk about
thinly-interbedded soils. For these kind of soils, I do not know whether we can
consider a portion of the soil as the mixture of soils part of various
interbedded layers.
13. Focus on what appears to be the most continuous soil units, work with
geologists to get an understanding of the geologic environment (i.e. 3D
continuity, is an interbedded unit an old meandering stream channel),
understand the gradation of soil materials (fines and gravel content) and how

159
 they differ from and might be influenced by other beds (will other beds
liquefy, help dissipate pressure, etc.).
14. Need to start with high-quality continuous sampling to determine thickness
and frequency of interbedding. Then, most likely need to perform effective
stress modeling of the layered system. Or, live with conservative predictions
(i.e., over-prediction of susceptibility and consequences) based on CPT.
15. Careful sampling and cyclic testing of clearly differing interbeds without
mixing of differentiable units; specimens must be uniform w/r/t the interbed in
question. How the "system" of interbeds responds is a question of triggering.
16. "Undisturbed" samples to see the stratification then effective stress modeling.
17. Based on the spatial continuity of susceptible layers and evaluation of
dominant characteristics of the deposit.
18. Depends on how thinly interbedded.... potentially the same as non interbedded
soils.
19. Evaluate the thickness of each soil layer to identify the predominant soil unit
and drainage paths to assess the overall behavior of the interbedded soils.
20. Based on the thicknesses of the respective liquefiable and nonliquefiable
layers.
21. I have not had a chance to do this in practice, but I think I would do cyclic lab
testing if at all possible.
22. I view susceptibility as a material characteristic, so I would tend to judge the
susceptibility of the individual materials separately. The interbeddedness
would affect triggering to some degree and consequences to probably a
greater degree.
23. They are susceptible but different than clean saturated sands.
24. I generally don't expect highly interbedded soils to liquefy in an appreciable
manner, at least not such that we would see surface manifestation effects.
25. CPT with layer adjustments when layers are thick enough for representative
assessment, otherwise generalized, "homogenized" soil approximation with
sensitivity evaluation.
26. I focus on the weakest susceptible layers that are over 0.5m thick.
27. I think you need to first define "susceptibility." If you have interbedded soils,
some layers will be susceptible and others won't be susceptible. I would look
at the PI of the various layers as a preliminary way to assess their
susceptibility.
28. I would judge susceptibility using the granular fraction and consider the
interbedding in separate analyses of manifestation.

160
 29. Use CPT, get undisturbed or at least sonic samples, consider the layer
thickness, relative presence of fine-grained soil layers against coarse-grained,
if possible do some cyclic shear testing.
30. Interbedded soils are often mixed together during sampling. If this does not
occur, I would recommend testing the individual soil types within the sample
separately.
31. It depends on the nature of the interbedding, and the assumed "homogenous"
behavior of the soil. If the interbedded materials should be liquefiable in other
conditions, or if the interbedding wouldn't significant improve the drainage
characteristics we treat the soils as liquefiable.

Question 8 – What are the main limitations of the state-of-the-practice with respect to
liquefaction susceptibility assessment?

1. Assessment of transitional fine-grained soils.

Lack of methods to assess liquefaction susceptibility for soils that are not
quartz sands (in particular fine-grained soils); and, lack of methods for post-
liquefaction hazard assessment (volumetric strain, shear strength, lateral
deformation) for fine-grained soils.
2. Too reliant on CPTs alone.
3. Having a consensus on the approach. Lack of a unified theory that links pore-
pressure generation, liquefaction, strength/stiffness reduction, and volumetric
strain for all soil (sands, silts, clays). Susceptibility and affects of liquefaction
at depths greater than 60-80 ft.
4. Soil layer variability and geologic details.
5. Lack of site specific testing, large variations in state of practice, conservative
use of sand trends regardless of soil type.
6. For silts, it can be challenging to have enough explorations, basic sampling
testing, and cyclic testing to provide a 3D interpretation of the site such that
the extent of susceptible soils can be distinguished. For sands with high fines
or plastic fines, it can be challenging to obtain samples for cyclic testing so
cyclic behavior can be observed.
7. Hard boundaries for soils we know to have transitional behavior. Relatively
small number of in-situ samples that can reasonably be collected relative to
the size of a site.
8. We have not quantified uncertainty in our susceptibility evaluation
procedures.
9. The community's understanding of the effects of non-zero mean static shear
stress, effect of fines, ageing effects.

161
10. High initial static shear stress and overburden pressures.
11. Research that established susceptibility assessment is generally on "ideal"
sands.
12. Complex mixtures (fills); silty soils, relatively wide "gray zone" of soils with
transitional behaviour, and lack of comprehensive measures (combination of
parameters).
13. The main issue is that we have a boolean approach: (1) soil is susceptible to
liquefaction, (2) soil is not susceptible to liquefaction (in which case we
expect soil as behaving as a clay-like material and cyclic softening might
become an issue that is dealt with separately). Also, there is some confusion
(no general agreement) on what "susceptibility" means: soil intrinsic
characteristics vs. soil characteristics + current condition. With latter being a
mix of susceptibility and triggering.
14. Lack of clarity in dealing with transitional soils and identifying where soils
are fines-controlled vs. not. Additionally, lack of clarity on specifically how
gradation (coefficient of uniformity) might impact susceptibility.
15. The simplified CPT, SPT and Vs procedures all work quite well for young,
clean sand deposits. There are still a lot of questions about susceptibility of
interbedded soils, soils with low-to-moderate plasticity, soils with micro-
structure due to aging or slight cementation, etc.
16. The source data underpinning the two most common liquefaction
susceptibility criteria (B&I 2006; B&S 2006) is relatively low. Since
mineralogy and deposition environment/time-dependent processes may affect
susceptibility, the databases needed to test these criteria need to be expanded
significantly.
17. Delineation of fines-controlled is broader in my opinion than current
thresholds typically used in practice. Characterization of deep deposits.
18. (1) Different lab criterion used by different people, (2) Inefficiency of field
measurements (e.g., Ic) in relation to lab-based criterion.
19. The lack of case histories for interbedded soils, transitional soils and
potentially liquefiable soils at greater depth to validate the mechanics-based
approach developed based on centrifuge and laboratory testing.
20. Lack of understanding of the role of compressibility.
21. Currently there's little consensus about what is meant by liquefaction
susceptibility and each model inherently means something different. There
are also a lot of uncertainties that are relatively undefined at present.
22. Cost of sampling to obtain plasticity index. In cases where only CPT is used,
lack of certainty about relationship between parameters like Ic, IB, etc. and

162
 susceptibility and obtaining reliable values of those parameters for thin layers
or interbedded profiles.
23. So far a lot of what we know is based on clean saturated sands. Only recently
there has been serious work using intermediate soils.
24. We don't have a good sense of the uncertainties inherent to susceptibility
characterization, we don't have that many different criteria, and CPT-based
criteria are quite simplistic.
25. (1) Assessment of Transitional Soils, especially with respect to cyclic/post-
cyclic behavior at the benchmark, Reference Shear Strain of 3.0% to 3.75%
developed for sand-like soil, and (2) Unresolved uncertainties in Ru
estimation at depth greater than roughly 75 ft (i.e., influence of confining
stress, aging, etc.).
26. Linking lab testing to field testing.
27. The ambiguity in what is being referred to as "susceptibility" is the most
significant limitation of current susceptibility criteria.
28. Over-reliance on two legacy methods. Poor understanding of the proper
definition of "susceptibility."
29. Not considering how difficult it is to get the void ratio right but still estimating
state parameter to high accuracy, not accounting for particle fabric, not
accounting for principal stress rotation.
30. It's not clear what susceptibility means sometimes. In my opinion,
susceptibility is a compositional feature of soil. Soil that is not susceptible to
liquefaction will not liquefy no matter how strongly it is shaken. Soil that is
susceptible to liquefaction will liquefy if shaken strongly enough. I think there
may be too much reliance upon soil behavior type index, Ic, on assessing
susceptibility without adequate consideration given to the uncertainty in the
relationship between Ic and plasticity characteristics.
31. Capturing the system response of the soils, and behavior of liquefied soils on
structures at depth.

Question 9 – What soil conditions are particularly problematic for diagnosing liquefaction
susceptibility, given the state-of-the-practice?

1. Transitional fine-grained soils.

2. Thinly interbedded soils with clay content in the low teens.
3. Soils deeper than 60-80 ft. Silty soil, especially low/non plastic silt.
4. Non-ideal soils (e.g., non-quartz, soils with mica, aged soils).
5. Silts.

163
6. Fluvial environments that have varying deposition resulting in complex
distributions of PI and fines content that exploration tools such as CPT cannot
distinguish.
7. Interbedded soils, partially saturated soils, deposits with fines contents in the
transitional range of 15-35% and plasticity indices in the range of 5-15.
8. Mid-range fines content (20 to 50%) with plasticity indices on the order of 8
to 12.
9. Intermediate Soils.
10. Soils with smaller PIs (4-7), soil with gravel contents.
11. Sands with micro-structure and cementation that may undergo cyclic
softening rather than flow liquefaction.
12. Thinly interbedded deposits, crushable soils, soils at the threshold between
susceptible and not-susceptible to liquefaction.
13. Intermediate plasticity soils and thinly-interbedded soils.
14. The most difficult soils conditions we see related to soils with gravel contents
greater than around 10% and with appreciable size. They make gathering
samples difficult. We especially find soils that are broadly graded with F/S/G
contents nearly evenly distributed (33,33,33) or even if gravels are lower but
sized bigger. It again gets to what defines the matrix and limited
understanding of gravel liquefaction, especially as gravel contents and size
increase (and Cu perhaps gets lower). Currently, it seems "susceptibility" is
more on focusing on whether they may trigger or not... i.e. assume they are
susceptible and then determine whether they might trigger, which has its own
issues.
15. As noted above; interbedded soils, soils with low-to-moderate plasticity, and
soils with micro-structure due to aging or slight cementation.
16. Clayey sands and low plasticity silts.
17. Low plasticity silts, low-plasticity fines with FC of 20-45%, finely
interbedded deposits, soils at high confinements.
18. Those near the sand-like vs. clay-like boundary, and highly interbedded soils.
19. Transitional and interbedded soils, especially at depth deeper than 60 to 80
feet.
20. Layered nonplastic to low plasticity soils.
21. Transitional soils, highly interbedded and loosely characterized (i.e., sites with
only CPT data that rely entirely on correlations with CPT data to determine
susceptibility, which have significant uncertainty).

164
22. Soils of intermediate plasticity. Profiles with thin or interbedded layers
(difficult to obtain reliable CPT-based susceptibility parameters).
23. Intermediate soils.
24. Interbedded and transitional soils.
25. Soils that are not similar to the clean sands on which most of the lab-based
research has been based. On project applications, regional "unique" soils are
routinely problematic due to inherent differences in composition, structure,
and cyclic behavior. Examples include; (i) high carbonate, shell-rich or
coralline sand, (ii) mine tailings, (iii) Transitional, Intermediate soil, and (iv)
sand-gravel mixtures.
26. Silty interbedded layers.
27. Clayey sands with low PI.
28. Materials of high fines content and marginal plasticity, PI =4-12.
29. Gravels, silts, medium density soils.
30. Sand with non-susceptible plastic fines is a tough one. I'm talking about sand-
dominated matrix soils here, so less than 35% fines. Another huge problem is
the fines correction applied to susceptible fine-grained soils. But this is not a
susceptibility problem, but rather an assessment of whether a susceptible fine-
grained soil will or will not liquefy under a specified loading condition.
31. Capturing non-homogenous soil profiles, particularly when the soil profile is
very deep. Additionally, assessing whether partial saturation is a long term
reliable marker for an increased CRR.

165
Question 10 – Which factors should susceptibility be related to (check all that apply)?

Degree of saturation

Insitu soil state (density, effective stress) conditions.

Depositional environment (e.g., marine, lacustrine,

fluvial, etc)

Soil behavior characteristics (e.g., parameters related

to shape of hysteresis loops, pore pressure ratio ru,
parameters related to whether undrained strength
normalizes by effective stress).

Soil material characteristics (e.g., fines content,

plasticity, gradation, angularity).

0 10 20 30 40
Frequency

Question 11 – Should susceptibility be designated in a binary (yes/no) manner or

probabilistically?
32

28

24

20
Frequency

16

12

0
Binary manner Probabilistic manner

166
Question 12 – Should geologic information be quantified and used in liquefaction
susceptibility assessment?
32

28

24

20
Frequency

16

12

0
Yes No

Question 13 – How can geologic information be used in liquefaction susceptibility

assessment?

1. Evaluation of stratigraphic continuity and variability to assess the extent of

liquefiable soils and liquefaction hazard; and guiding site exploration
programs.
2. To assess if additional advanced sampling and laboratory testing would be
required to understand liquefaction potential.
3. Depositional environment.
4. Generally no unless site specific testing is incorporated.
5. Known and commonly encountered geologic formations could be researched
with high quality testing not feasible for single projects.
6. Best used in regional assessments or areas where in-situ testing is not
available or feasible. Type and age of the deposit can be used, and through
new case histories with improved regional data collection perhaps more
granularity on "type" of deposit (or types of deposits in different parts of the
world) can be achieved.
7. I anticipate that geologic information can be incorporated into a statistical
regression relationship or qualitatively.
8. By including geologic information the community would have a stronger
understanding depositional conditions, mean layer thicknesses/lengths,
paleoliquefaction, age.

167
9. The age of the deposit, how many earthquakes the deposit has experienced,
depositional environment in terms of small layers and energy of deposition
and mineralogy.
10. Old, cemented sands are less likely to undergo liquefaction.
11. Source, formation and structure of the deposit (macro and micro
stratification).
12. Not sure.
13. Geologic information is useful for understanding the geomorphology for a
site. In practice, consultants and engineers are often limited by resources to
2D models of their sites. Incorporating and understanding how the geology of
a site into their models may help where units might appear to be discontinuous
but could be a meandering stream bed. Geologic processes and general
understanding can be utilized stochastically, and in combination of data over a
site to understand what the statistics might mean. I'd note that straight up
probabilistic based evaluations are often impacted by biases resulting from
poor data, which we often see in gravel particles impacting data.
14. Generally, to improve liquefaction hazard maps; to understand how
mineralogy affects susceptibility; to date soil deposits and assess
aging/geochemical processes.
15. In a similar manner as different fault-mechanisms are used in crustal GMPE's
as a binary term that increases or decreases the probability of liquefaction
susceptibility given other relevant soil information.
16. As a very preliminary screening tool, or to bin field-lab susceptibility
correlations on geologic units... I don't think it's of much quantitative use, per
se.
17. Geologic information should be used to assist in judging if the liquefaction
analysis results are reasonable.
18. Quantifying depositional energy.
19. I am not currently sure how this information can or should be used, but it
seems reasonable to me that if it could be used then it might help constrain
some of the uncertainty currently in susceptibility models.
20. Not sure, other than by crude correlation to nearby tested material. Factors
like age are triggering factors, to me.
21. Depositional history can have an effect on liquefaction susceptibility.
22. Essentially as a "prior", i.e. before we even poke holes in the ground. This
could be achieved using geospatial models for probability of susceptibility
(similar to Zhu et al.'s global model for liquefaction probability), and could be

168
 a useful way to blend regional and site-specific data in assessing liquefaction
potential.
23. Empirically-based scaling factors have been used in several regions to account
for the influence of various geologic influencing parameters (e.g., age,
cementation, depositional environment). This has been a worthwhile start;
however, quantification by way of correlation with geotechnical and
geophysical parameters seems warranted.
24. Depositional environment and ageing are key in understanding a soil's
susceptibility.
25. Sampling will always be limited. Knowing the depositional environment of
the profile allows you to extrapolate the properties of the sample laterally and
vertically.
26. Depositional environment could be useful - soils deposited in quiet
environments (lakes, bays) are likely fine-grained and non-susceptible. Soils
deposited in relatively rapid-flow fluvial environments are more likely to be
liquefiable.
27. Determine age, layering, expected fabric.
28. Paleoliquefaction studies could be used to assess whether a marginally
susceptible soil has exhibited evidence of liquefaction in past earthquakes.
Geology may also be used to assess conditions where soils may be
susceptible, not-susceptible, or marginal.

Question 14 – Should advanced laboratory testing (e.g., cyclic simple shear, cyclic triaxial)
be used to aid liquefaction susceptibility assessment?
35

30

25
Frequency

20

15

10

0
Yes No

169
Question 15 – How should advanced laboratory testing (e.g., test type: monotonic, cyclic;
specimen types: reconstituted, intact) be used in assessment of liquefaction susceptibility?

1. Need to know OCR. DSS for fine grained soils.

2. Evaluating cyclic strength and hazard for soils where high-quality samples can
be obtained.
3. As long as we can have "undisturbed" sampling, cyclic testing on the material
in question will provide a better understanding of behavior.
4. Test results provide insight in terms of the shear stress vs. shear strain
response of soil and the movement of water within the soil deposit.
5. Intact monotonic and cyclic DSS.
6. Cyclic testing of undisturbed samples can be used to establish cyclic behavior
for the range of possible ground motions.
7. Cyclic testing should be used in conjunction with field observations to
advance state of knowledge and assessment models, and also used in practice
for soils that are not well understood or analyzed using currently available
methods. Intact specimens are far preferable to reconstituted specimens for
this type of testing.
8. A combination of monotonic and cyclic testing can be used to demonstrate
whether the soil behavior will be sand like vs clay like. I think this type of
testing is applicable for critical infrastructure projects and research efforts to
develop more simplified relationships for use in general geotechnical
engineering practice.
9. Laboratory testing can help check or refine correlations for site specific
conditions. For in-situ conditions that are on the border line of liquefaction,
testing can help confirm predicted soil behaviors.
10. Intact samples are preferred but reconstituted might be useful. Cyclic tests
(DSS) would be most useful.
11. Cyclic simple shear.
12. Details of stress-strain behaviour can help to discriminate between
susceptible, not-susceptible and transitional soils. Intact specimens are the
best choice; reconstituted could be also useful for susceptibility assessment (it
depends on the soil characteristics).
13. They would help designing a model. I do not expect models needing advanced
laboratory testing to define susceptibility as this would defeat the purpose of
having a simple screening criterion/criteria.
14. I will say that advanced laboratory testing should be used to help supplement
research data. In practice, most consultants and projects are resource limited

170
 to monotonic testing, while cyclic softening to fill in research gaps at an
academic level could help bridge and relate susceptibility to more accessible
test types in practice (or other parameters). In practice, it is also extremely
challenging to get intact samples, yet those are of most concern at dam sites.
15. Given that we cannot sample and place "undisturbed" specimens of
liquefiable-type soils in testing devices, cyclic laboratory testing is primarily
valuable for parametric studies and/or for calibrating cyclic response of
constitutive models. I do not believe we can accurately evaluate in-situ
liquefaction in absolute terms from laboratory testing.
16. In my opinion, the single most reliable means to assess liquefaction
susceptibility is to perform cyclic testing. Since cyclic triaxial testing provides
the wrong stress path and is biased in extension, the cyclic direct simple shear
test is the most appropriate, accessible laboratory test which can establish the
potential for transient loss of shear stiffness and excess pore pressure
generation potential. Such tests should be accompanied by grain size analysis,
Atterberg limits, water content, and overconsolidation ratio in order to
understand how the specimen(s) maps to the deposit.
17. Cyclic tests on intact specimens when possible. Reconstituted cyclic tests
may be OK to identify general behavior with an appropriately extensive lab
program.
18. The key is not whether testing would be useful, but how to incentivize/require
it on routine (or not so routine) projects. If very few people are willing to pay
for it, this discussion isn't very purposeful. The incentive should be worked
out before the specific test details.
19. The advanced lab testing should be used to inform the susceptibility of
transitional and sandy soils at depth greater than 40 feet.
20. Reconstituted first to develop a framework and then undisturbed to validate
the framework.
21. I think there are aspects to be learned from all types listed in the question.
Pros and cons for each specimen type and lab type, but together they may
provide sufficient insights to assess susceptibility in challenging soils.
22. Get the best samples possible with the soil type of interest. If good intact
samples are not possible, reconstitute in manner that approximates actual
depositional processes. Examine shapes of hysteresis loops, highest ru
achieved, tendency for dilation upon phase transformation, rate of stiffening
upon dilation.
23. Absolutely much more testing using different lab techniques, following
different stress paths.

171
24. First, the use of the term "advanced laboratory testing" requires explicit
definition. What specific types of cyclic tests are being referred to as
advanced? For example, TX and CycDSS are not considered "advanced"
methods of testing and should be treated as routine supplements to a well-
planned geotech investigation for soil-types that can be sample with minimal
disturbance or for which testing of reconstituted specimens can provide
representative behavior. For sand-like soil, lab testing would continued to be
viewed as a supplement to, not in lieu of, standard simplified procedures.
Also, cyclic testing may be required for the calibration of constitutive models
such as PM4Sand. This latter consideration is too often overlooked in
practice.
25. By isolating specific variables and assessing their relative contribution.
26. Liquefaction triggering inherently involves the breakdown of the soil skeleton
and the transfer of the overburden stress to the pore fluid. As the PI of the soil
increases, bonded water fills a larger percentage of the voids, preventing the
soil skeleton from collapsing (i.e., bonded water results in the soil to behave
more viscously - clay-like behavior) when subjected to cyclic loading. So,
cyclic tests are useful in assessing the liquefaction susceptibility of a soil.
27. Identify metrics that indicate soil behavior, which could be max ru, relative
tangent moduli at small and large strains, or degree to which undrained
strength normalizes. Related those metrics to susceptibilty, through research
in the NGL project.
28. Consider all of the above and decide on what is the best approach.
29. Atterberg limits should be performed first. Monotonic strength and
consolidation testing can be performed to assess whether the normal
consolidation line and critical state line are straight and parallel, which is an
indication of clay-like behavior. Cyclic testing can also be used to assess
whether the soil reaches ru=1, or stabilizes at a lower value. Cyclic testing is
very important for evaluating strength loss potential for non-susceptible soils.
I think of them as being susceptible to cyclic softening rather than to
liquefaction.

172
Question 16 – How should such tests be interpreted in terms of the potential for
susceptibility?

1. Used to supplement simplified methods.

2. Evaluating CRR, excess porewater pressure generation, and soil stress-strain
behavior.
3. Should start with similar interpretations of similar cyclic testing, and include a
degree of judgement to the results to modify similar interpretations.
4. If the response of a soil is similar to that of a clean sand that we agree is
susceptible to liquefaction it should be considered as susceptible to
liquefaction.
5. Ru, strain accumulation, shape of hysteretic loops.
6. Test should be interpreted relative to the range of common design ground
motion levels and the ability of those motions to "change" the soil behavior
such that "liquefied/reduced strength" properties should be used in analyses
instead of "static" properties.
7. Development of strains and pore water pressure vs. number of cycles.
Frequency of loading is also a critical consideration for soils in the transitional
fines/PI range that may have rate effects.
8. Interpretation can be challenging. For example, I would have judged both
soils in Figure 10 from Boulanger and Idriss (2006) as soils that are
susceptible to liquefaction based on their stress-strain behavior; whereas,
Boulanger and Idriss (2006) characterized the silt with a PI=10.5 as behaving
as a clay. This indicates that stress path results may be needed as additional
support for justifying the interpretation of the soil behavior.
9. Check correlations, develop site specific correlations, confirm borderline
cases of susceptibility.
10. Either Ru approximately 1.0 or single axial strain of 2.5%.
11. Hysteresis loops and excess pore pressure.
12. Level of ru, i.e. residual effective stress levels; characteristics of stress-strain
curve (loop);
13. Not sure yet.
14. The quality of the samples is first that should be looked at. How
representative are the materials versus what is in the field. Have some
percentage of gravels been loss. Confidence in density. From a broader
perspective, if a suite of soils are tested, it is best to test them to a similar
playing field and perhaps measure their susceptibility as a comparison to

173
 known liquefiable soils, specifically with respect to both excess pore pressure
and shear strains.
15. The hysteresis loop at large strains is the single most important piece of
information which can definitively establish the susceptibility to liquefaction.
16. Pore pressure generation, stiffness degradation characteristics with ru,
hysteretic behavior, and post-seismic tests should be evaluated and compared
with established literature.
17. Test results should be interpreted in ways to correlate to the semi-empirical
method for top 40 to 60 feet and inform the limitation of the semi-empirical
methods for transitional soils and soils deeper than 40 feet.
18. Whether particular consequences can be developed (within a reasonable
number of cycles) regardless of the loading intensity.
19. Currently unsure.
20. See last response.
21. Can use # of cycles to reach a certain level of deformation or pwp ratio.
22. With all the usual caveats, i.e. that samples can be disturbed, that they only
represent an extremely small proportion of the soil deposit in question. I think
there's a lot of utility in generating site-specific FC-Ic correlations using lab
data.
23. Interpretation of the lab tests for use in practice will likely be soil specific in
the sense that the applicability of the lab data to field conditions requires
judgement on the influence of a host of considerations that have been well
addressed in the literature (e.g., specimen disturbance, stress path, shear strain
localization, 1D and 2D loading considerations, etc.).
24. Excess pore pressure and shear strain potential for a given number of cycles in
stress-based tests is a common approach.
25. Liquefaction triggering is not the issue of concern, but rather the
consequences of liquefaction triggering is of concern. I would use laboratory
tests to assess the soil properties after specified cyclic loading and use these
properties to assess the consequences of the cyclic loading.
26. See previous response.
27. Rate of degradation of stiffness, rate of generation of excess pore water
pressure, stiffness criteria instead of strain criteria.
28. I answered that in the previous cell.
29. Significant stiffness reduction (near zero) under the anticipated seismic loads.

174
Question 17 – What information should next generation liquefaction susceptibility models
provide to the user that current models aren’t providing now?

1. OCR correlations.
2. We need to start moving towards modeling of liquefaction using more
advanced models (PM4Sand/Silt) that can be calibrated with cyclic testing to
help understand the behavior and build a database that can be access by the
engineering community.
3. Depends on the NGL model.
4. Calibration with cyclic testing. Inclusion of new recently made empirical
data/laboratory testing.
5. Models should use consistent measures of susceptibility. Since susceptibility
is often yes/no, multiple criteria may be warranted. Or varying degrees of
strength loss, strain potential, settlement potential, etc... should be provided
whether deemed susceptible or not. This could help engineers with nuance
project specific questions related to consequences, client risk tolerance, and
life safety vs cost.
6. Better understanding of transitional soil response and system response of a
deposit that may have partial saturation or interlayering; probabilistic
estimates and integration with triggering models and consequences.
7. Next generation models should provide the probability for susceptibility
coupled with guidance on how to move forward when there is some moderate
probability of susceptibility. For example, if there is a 20% probability that
the soil is susceptible to liquefaction, should advanced laboratory testing be
used to establish cyclic behavior?
8. Better information about sensitivity of inputs. Would additional information
reduce susceptibility? Reduced uncertainty and stronger more accurate
predictions.
9. Overburden and static shear stresses.
10. CRR increase due to cementation or lamination.
11. Three classes: susceptible, transitional, not-susceptible.
12. Probability of liquefaction for sure and a more careful definition of
susceptibility that only considers intrinsic characteristics of soil.
13. The models would be best to come together into one cohesive model (or suite
of models) that interrelate to each other so that the practicing world can have
confidence that this is an agreed upon model. More technically, the next
generation models need to more clearly address how fines content and
gradation impact susceptibility, especially as fines contents exceed 20%

175
 towards 50% and PI's range between say 5 and 20. Currently, the models are
more binary in nature suggesting soils are no longer liquefiable at some
boundary depending on the model you look at. Perhaps something more
probabilistic representing the reality that the behavior probably gradually
transitions from liquefaction susceptibility (towards cyclic-softening
susceptible)... AND perhaps a clear identification of a zone where neither is a
possibility. From a theoretical perspective it seems logical that what we call
transitional soils, especially those that tend to be well-graded, are not
susceptible to liquefaction. Yet, literature rarely tells us that materials are
NOT susceptible.
14. Consequences of liquefaction; better estimates of settlement and post-
liquefaction shear strength.
15. A probabilistic liquefaction susceptibility model, material characteristics
based, that is pegged to cyclic hysteresis behavior observed in the laboratory.
16. Probabilistic evaluation, quantifiable means to account for geologic
characteristics,
17. A probability of susceptibility; ensemble predictions of susceptibility.
18. Quantitative measure of variability and uncertainties associated with
liquefaction to inform the probabilistic liquefaction susceptibility.
19. Susceptibility related to consequences of liquefaction.
20. Whatever the output from a model, a mean (or median) and an uncertainty
term from a probabilistic approach is desirable.
21. Probability of susceptibility.
22. Probability of susceptibility, increased emphasis on using CPT data.
23. A probability of susceptibility.
24. The triggering models should specify the applicable soil types.
25. Probability of liquefaction conditioned on different types of information
(simple index tests or Ic vs more advanced indices).
26. Deformation potential as opposed to susceptibility.
27. 1. A clear definition of what susceptibility means; and 2. Probability of
susceptibility, and a framework for incorporating it into analysis.
28. System response context, and drainage impacts due to layering.

176
Question 18 – Should NGL develop a database specifically for the study of susceptibility?
35

30

25
Frequency

20

15

10

0
Yes No

Question 19 – How can new technologies be applied to the problem of susceptibility

assessment?

1. Existing tools can be used to characterize those factors that affect soil
susceptibility.
2. Develop guidelines for use of cyclic testing data.
3. The state of practice would need to see the technology proven out by rigorous,
research level testing. It is otherwise difficult to "sell" either to an owner or
peer reviewer.
4. The same properties relevant to susceptibility are also typically relevant to
triggering and consequences, so an integrated NGL database will likely enable
more robust understanding of the data and development of models. New
technologies allow for better collection of large datasets and should be used
for transparent dissemination of data. Using new technologies for more
efficient collection and publication of geotechnical in-situ data should also be
a priority. Drone (or flight) imagery collection over large areas will also
enable better coverage for collecting post-earthquake observations of
liquefaction and no-liquefaction which can then be used to select sites for
detailed investigations.
5. New data analysis methods such as machine learning may be useful.
6. The use of instrumented BPT and SPT looks to be very promising.
7. Use of CSS testing to develop CRR for a soil.
8. Recovery of high-quality samples, index testing and cyclic testing of soils.
9. Not sure.

177
10. I think to the extent more reliable sampling techniques (again thinking about
gravel) can be used to get a better grip on gradation, that might help.
Potentially using modern equipment to more closely assess real sites where
liquefaction has occurred can be useful.
11. Experimentally: The development of a cheap, downhole (borehole?) based in-
situ cyclic testing apparatus could improve the ability to establish liquefaction
susceptibility. Analytically: Perhaps machine learning could be used to parse
out trends/material characteristics which give rise to certain hysteretic
features."
12. More downhole arrays specifically targeting transitional soils and deep
potentially liquefiable deposits.
13. Information sharing including advanced lab testing and centrifuge test results
for soils with similar characteristics.
14. Unsure.
15. Unsure.
16. NMR techniques (e.g. Vista Clara) can distinguish between free and "bound"
water - seems like bound water content could be related to plasticity.
17. There's a lot more data now, which is great, and we're a lot better as a field at
formulating things probabilistically and developing models with more
statistical rigor than we used to be. Not sure that qualifies as "new
technologies" but that's the advantage I think we have now... being more
quantitative and less reliant on engineering rules of thumb.
18. A very good question and one that I eagerly look forward to discussing with
those who are actively applying "new technologies" to dynamic soil behavior
(in the lab and in situ).
19. I don't have a good answer for this.
20. I don't know what new technologies are being referred to here. I think
undisturbed sampling and cyclic testing is the best way to assess
susceptibility.
21. Develop a database with: (1) lab index tests; (2) Ic values from CPT; (2)
advanced testing. Perform advanced regression, perhaps including machine
learning, to relate susceptibility to different indicators.
22. Find ways to get spatial variability as well as quantify particulate structure.
23. Methods like machine learning, logistic regression, etc. can be brought to bear
to gain insights into this classification problem.

178
Question 20 – What do you view as the most significant challenge to advancing liquefaction
susceptibility assessment?

1. The degree of relative difficulty and cost of the more advanced analysis versus
the state-of-the-practices, CPT only evaluations.
2. Thin layering and soil variability.
3. Lack of adopting new methods research and/or willingness to do something
different.
4. In general, clients do not understand the topic and can't differentiate between a
consultant who thinks attention to susceptibility of certain soils (e.g. silts) is
necessary and another consultant that does not. So advancement must come
from the academic side. Advancement has a better chance if susceptible soils
can be differentiated by their associated consequences of strength loss, strain
potential, settlement potential, etc. since many believe that for some soils
susceptible doesn't necessarily present a life safety problem which is the key
issue to address.
5. Development of community consensus models, or adopting a framework for
probabilistic analysis using multiple methods such as was done for ground
motions GMPEs. Having the data necessary to use the most current models
effectively.
6. I believe the most significant challenge is collecting the data needed to
develop new models.
7. Developing or using new procedures/ideas that deviate from the established
liquefaction susceptibility assessment tools that are precedent. To reduce
uncertainty and advance liquefaction assessment we should ask if our current
tools still the best? Does greater parameterization produce better assessment,
and why cannot our established approaches capture the information
introduced with this parameterization?
8. Not having sites with recorded ground motions. Not having subsurface
information for sites that didn't liquefy. The lack of sites with large
overburden stress and high initial static shear stresses.
9. Sample quality.
10. Complexity/uniqueness of soil composition, insufficient high-quality
sampling and then testing in the laboratory, and subtle differences in soil
behaviour.
11. Lack of reliable laboratory data and of a consistent definition of susceptibility.
12. Numerous unknowns regarding gravel is one issue. Another issue that
impacts dams in my daily work relates to the complexity associated with
deposition or engineered fill where it can be difficult to identify what is the

179
 matrix or not or what is controlling. For instance, it is common in practice to
try and subtract gravel influence for assessing susceptibility (and triggering).
It is difficult to do this when layers are "well-mixed". Similarly, most
liquefaction case histories tend to be in relatively flat sites under limited shear
stresses. It is unclear how much of what we base on liquefaction
susceptibility is incorporating higher shear stresses. Finally, for dam projects
in general, it is challenging to leap from in-situ conditions prior to the
construction of a dam to post-reservoir-filling when saturation levels increase
and further any geologic properties may be altered (e.g. loss of cementation in
older units due to water presence).
13. Settling on a clear definition of liquefaction susceptibility. Although we might
consider a probabilistic treatment as convenient, this definition ought to be
binary for maximum clarity given that liquefaction can either occur given
sufficient loading intensity and duration, or not.
14. Cost and logistics of sampling and testing intact samples with fines contents
less than about 50% across a range of PIs, clay fractions, and geologic and
depositional environments.
15. The paucity of all the required data types available to researchers (invasive
field samples; non invasive field measurements; lab index testing; lab cyclic
testing).
16. The lack of consensus among researchers and practitioners on the approach to
assess liquefaction susceptibility of transitional soils and interbedded soils,
specifically on how the results of the advanced lab testing results should be
incorporated in the analysis.
17. Defining a consistent definition of susceptibility that incorporates soil
compressibility and consequences of liquefaction.
18. Unsure.
19. In practice, getting budgets for sufficient drilling, high-quality sampling, and
laboratory testing. In research, developing reliable methods for measuring
parameters that control susceptibility and making them economical.
20. Utilizing multiple CPT-based criteria in a way that allows us to capture both
aleatory variabilities and epistemic uncertainties.
21. Coupled assessment of the "susceptibility and consequences" of reaching a
given Ru value, or perhaps a Reference Accumulated Cyclic Shear Strain.
Alternatively stated; what might the susceptibility of a soil be to significant
strength reduction or large strain accumulation given the amplitude and
duration of the cyclic loading? Many outstanding and useful procedures exist
for uncouple assessment of liquefaction related behavior. Practice-oriented

180
 tools such as PM4Sand / PM4Silt provide a bridge from the lab and
empirically-based procedures to a coupled framework.
22. Money. A large suite of full scale field tests at various fully characterized and
instrumented sites with a shaker or blasting could provide a robust data base
for pinning things down.
23. Being able to assess susceptibility in a way that is economically feasible for
typical engineering projects.
24. Resistance to change in practice and over-reliance on legacy models.
25. Reluctance to deviate from conventional approaches.
26. Clearly defining it.
27. Guidance documentation and code requirements.

181
 PEER Workshop on Liquefaction Susceptibility

APPENDIX D: WORKSHOP AGENDA

Agenda

Wednesday, 7 September
6:30 PM Welcome Reception and Ice Breaker
- 8:30 PM Location: The Vue, 517 SW 2nd St, Corvallis, OR 97330
Shuttles from Hilton Garden Inn to venue begin 6:15 PM
Note: venue located 1 mile (20 min. walk) east from Hilton Garden Inn
Day 1: Thursday, 8 September
7:30 AM Workshop Opening
- 9:00 AM MU Horizon Room

7:30 AM Check-in, Breakfast

8:30 AM Armin Stuedlein, Welcome, Agenda, and Workshop Objectives
8:35 AM Scott Ashford (Dean, College of Engineering), Welcome Address
8:40 AM Steven Kramer, Overview and Summary of Pre-Workshop Poll

9:00 AM Session 1: State-of-Practice and Limitations

- 12:30 PM MU Horizon Room

9:00 AM Part A: Presentations (12 min talks, 3 min short Q&A)

Önder Çetin, Probabilistic Models for Seismic Soil Liquefaction Susceptibility
Thomas Weaver, Evaluating Liquefaction Susceptibility for Nuclear Power Plant
Sites
Erik Malvick, Challenges of Liquefaction Assessment at California’s Dams
Uri Eliahu & Pedro Espinosa, Dynamic Behavior of the Treasure Island Natural
Shoals

10:00 AM 15-minute break

10:15 AM Part B: Presentations (12 min talks, 3 min short Q&A)

Sam Sideras, Liquefaction Susceptibility of a Low Plasticity Silty Soil Utilizing
Cyclic Direct Simple Shear Testing
Matt Gibson, Liquefaction Susceptibility of Grays Harbor Silts
Brice Exley, The Impacts of Analyzing Deep Sand and Transitional Soil Profiles
with State of the Practice Methods

11:00 AM Breakout Session, Report, and Discussion

Part A: Outdoor Breakout: Group Activity
11:30 AM Part B: MU Horizon Room: Report on Breakout
12:00 PM Part C: MU Horizon Room: Discussion

12:30 PM Lunch

182
1:30 PM Session 2: Where Do We Want to Be in 5 – 10 Years?
- 5:00 PM Model Development, Resource Needs/Gaps
MU Horizon Room

1:30 PM Part A: Presentations (12 min talks, 3 min short Q&A)

Shideh Dashti, Incorporating the Spectrum of Soil Behaviors Directly into
Systems Level Triggering and Consequence Models
Jonathan Bray, Liquefaction of Silty Soil
Dharma Wijewickreme, Particle Fabric Imaging for Understanding Shear
Response of Silts

2:15 PM 15-minute break

2:30 PM Part B: Presentations (12 min talks, 3 min short Q&A)

Laurie Baise, Geospatial Models for Liquefaction Susceptibility
Christine Beyzaei, Regional Liquefaction Susceptibility Assessments: Data
Collection Needs and a Focus on the Central and Eastern U.S.
Andrew Makdisi, Incorporating Uncertainty in Susceptibility Criteria into
Probabilistic Liquefaction Hazard Analysis

3:15 PM Breakout Session and Discussion

Part A: Outdoor Breakout: Individual Activity

3:45 PM 15-minute Break (& compilation of breakout responses)

4:00 PM Part B: MU Horizon Room: Report and Discussion

4:45 PM T. Matthew Evans, Day 1 Observations and Closing Remarks

5:00 PM Group Photograph, and adjourn to Reception:
NHERI Coastal Wave/Surge and Tsunami Center, O.H. Hinsdale Laboratory

7:00 PM Participants shuttled to Hilton Garden Inn or downtown for dining on their own

183
 PEER Workshop on Liquefaction Susceptibility

Day 2: Friday, 9 September

7:30 AM Breakfast
- 8:30 AM MU Horizon Room

8:30 AM Session 3: Opportunities for Synthesizing Laboratory- and Field-based

- 12:00 PM Observations, Consensus Recommendations
8:30 AM Part A: Presentations (12 min talks, 3 min short Q&A)
Scott Olson, Consequence-Based Susceptibility Incorporating Compressibility
Diane Moug, Relating Cyclic Behavior to CPT Data for Intermediate Fine-
Grained Soils
Brett Maurer, CPT-Based Probabilistic Prediction of Liquefaction Susceptibility

9:15 AM 15-minute break

9:30 AM Part B: Presentations (12 min talks, 3 min short Q&A)

Ross Boulanger, Susceptibility Criteria for Selecting Engineering Procedures
Scott Brandenberg & Jonathan Stewart, Cyclic Behavior of Low Plasticity
Fine-Grained Soils of Varying Salinity, and Cyclic Failure due to Dynamic Soil-
Structure Interaction
Armin Stuedlein & T. Matthew Evans, Linking Hysteretic Behavior to
Liquefaction Susceptibility

10:15 AM Breakout Session, Report, and Discussion

Part A: Outdoor Breakout: Group Activity
10:45 AM Part B: MU Horizon Room: Report on Breakout
11:15 PM Part C: MU Horizon Room: Discussion

11:45 PM Workshop Closing

I.M. Idriss, Observations on Workshop Discussions
Armin Stuedlein, Closing Remarks

12:00 PM Box Lunch (Shuttles to/from Hotel and Eugene Airport begin)

1:00 PM Workshop Organizers Draft Report

311 Kearney Hall

184
185
APPENDIX E: SESSION 1 PRESENTATIONS
This electronic appendix contains seven presentations from Session 1, titled “State-Of-Practice
and Limitations,” which may be found on the PEER report webpage, including:

• K. Önder Çetin: Probabilistic Models for Seismic Soil Liquefaction Susceptibility

• Thomas Weaver: Evaluating Liquefaction Susceptibility for Nuclear Power Plant

Sites

• Erik Malvick: Challenges of Liquefaction Assessment at California’s Dams

• Pedro Espinosa: Dynamic Behavior of the Treasure Island Natural Shoals

• Sam Sideras: Liquefaction Susceptibility of a Low Plasticity Silty Soil Utilizing

Cyclic Direct Simple Shear Testing

• Matt Gibson: Liquefaction Susceptibility of Grays Harbor Silts

• Brice Exley: The Impacts of Analyzing Deep Sand and Transitional Soil Profiles
with State of the Practice Methods

186
187
APPENDIX F: SESSION 2 PRESENTATIONS
This electronic appendix contains six presentations from Session 2, titled “Where Do We Want to
be in 5 to 10 Years? Model Development, Resource Needs and Gaps,” which may be found on the
PEER report webpage, including:

• Shideh Dashti: Incorporating the Spectrum of Soil Behaviors Directly into Systems
Level Triggering and Consequence Models

• Jonathan Bray: Liquefaction of Silty Soil

• Dharma Wijewickreme: Particle Fabric Imaging for Understanding Shear Response

of Silts

• Laurie Baise: Geospatial Models for Liquefaction Susceptibility

• Christine Beyzaei: Regional Liquefaction Susceptibility Assessments: Data

Collection Needs and a Focus on the Central and Eastern U.S.

• Andrew Makdisi: Incorporating Uncertainty in Susceptibility Criteria into

Probabilistic Liquefaction Hazard Analysis

188
189
APPENDIX G: SESSION 3 PRESENTATIONS
This electronic appendix contains six presentations from Session 3, titled “Opportunities for
Synthesizing Laboratory- and Field-based Observations, Consensus Recommendations,” which
may be found on the PEER report webpage, including:

• Scott Olson: Consequence-Based Susceptibility Incorporating Compressibility

• Diane Moug: Relating Cyclic Behavior to CPT Data for Intermediate Fine-Grained
Soils

• Brett Maurer: CPT-Based Probabilistic Prediction of Liquefaction Susceptibility

• Ross Boulanger: Susceptibility Criteria for Selecting Engineering Procedures

• Scott Brandenberg: Cyclic Behavior of Low Plasticity Fine-Grained Soils of

Varying Salinity, and Cyclic Failure due to Dynamic Soil-Structure Interaction

• Armin Stuedlein: Linking Hysteretic Behavior to Liquefaction Susceptibility

190
 Disclaimer

The opinions, findings, and conclusions or recommendations

expressed in this publication are those of the author(s)
and do not necessarily reflect the views of the study
sponsor(s), the Pacific Earthquake Engineering Research
Center, or the Regents of the University of California.
The Pacific Earthquake Engineering Research Center (PEER) is a multi-institutional research and education center
with headquarters at the University of California, Berkeley. Investigators from over 20 universities, several consulting
companies, and researchers at various state and federal government agencies contribute to research programs
focused on performance-based earthquake engineering.

These research programs aim to identify and reduce the risks from major earthquakes to life safety and to the economy
by including research in a wide variety of disciplines including structural and geotechnical engineering, geology/
seismology, lifelines, transportation, architecture, economics, risk management, and public policy.

PEER is supported by federal, state, local, and regional agencies, together with industry partners.

PEER Core Institutions

University of California, Berkeley (Lead Institution)

California Institute of Technology
Oregon State University
Stanford University
University of California, Davis
University of California, Irvine
University of California, Los Angeles
University of California, San Diego
University of Nevada, Reno
University of Southern California
University of Washington

Pacific Earthquake Engineering Research Center

University of California, Berkeley
325 Davis Hall, Mail Code 1792
Berkeley, CA 94720-1792
Tel: 510-642-3437
Email: [email protected]

ISSN 2770-8314
https://doi.org/10.55461/BPSK6314