net/publication/315615330
CITATIONS READS
2 413
2 authors, including:
Michael H. Beaty
Beaty Engineering LLC
24 PUBLICATIONS 246 CITATIONS
SEE PROFILE
All content following this page was uploaded by Michael H. Beaty on 24 March 2017.
Ross W. Boulanger1
Michael H. Beaty2
ABSTRACT
INTRODUCTION
The increased use of NDAs has, however, been accompanied by concerns regarding the
repeatability of results. There are examples where different engineers have analyzed the
same embankment dam, with apparently similar representations of the site
characterization and seismic loading conditions, and produced results that have ranged
from being in reasonable agreement to being so dissimilar they would lead to different
decisions on the nature of any remedial actions. Evaluating these differences is
challenging due to the complexity of the analyses as well as the somewhat unpredictable
1
Professor, Department of Civil & Environmental Engineering, University of California, Davis, CA 95616,
rwboulanger@ucdavis.edu
2
Principal Engineer, Beaty Engineering LLC, Beaverton, OR, 97007, mhb@beatyeng.com
Reviewers and regulators can have difficulties evaluating the suitability of a particular
analysis because of wide variations in the available analysis methods, the procedural
details of calibrating the constitutive models and executing the analysis, and the level of
documentation provided in engineering reports. Increasing the confidence that owners,
regulators, and engineers can have in the standard practice and performance of NDA
studies will require improved guidance (or protocols) for use of these analysis methods
and higher standards for documentation of the methods and results. These improvements
will facilitate the necessary internal or external review process and can lead to improved
insights and benefits for all involved.
Other factors affecting the quality and value of an NDA study include the organization of
the team (necessary range of skills; effective communications; review process) and the
reasonableness of the budget and schedule. The effort spent in performing a high quality
NDA study can add value over the long-term for dam owners by, for example, reducing
undue levels of conservatism and avoiding schedule delays that could develop if the
NDA results are later questioned by external reviewers or regulatory agencies. Improved
guidance on NDA documentation requirements or expectations will accordingly help
facilitate better definition of budgets and schedules for NDA studies.
The purpose of this paper is to provide a checklist of questions that a reviewer of an NDA
study should consider asking as part of any internal or external review process. The
questions posed herein are not exhaustive, but rather represent the authors' perspectives
on many of the details that can significantly affect NDA results. Not all of the aspects
presented below will be relevant or significant to all projects, but most should be
considered or addressed in engineering analyses and reports representing the highest
levels of practice.
The questions are grouped into twelve categories, spanning from numerical details to the
judgments drawn from the NDA results. The technical issues behind each question are
not discussed, as the scope of issues covered in an NDA study must draw from a large
body of technical literature. Similarly, opinions on the best technical approaches are not
provided because of length restrictions for providing context on each issue. The focus is
instead on suggesting what should be addressed in reports summarizing NDA studies,
with the aim that the first step toward advancing the standard of practice is improving the
transparency with which actual practices are documented. It is the authors' hope that the
checklist provided herein will help both consumers and producers of NDA studies for
critical facilities.
REVIEW QUESTIONS
Questions that should be considered when reviewing an NDA study are presented in
twelve categories. The questions are intended to encompass a range of issues that can
The engineering effort involved in addressing these questions is not large for a well-
executed analysis study given the increased value it provides. In most cases, the answers
involve brief statements documenting the basis of the work that was done and transparent
presentation of intermediate results which were likely already available. Much of the
information can be easily obtained if the need for transparent documentation is
anticipated at the beginning of the analysis process. The inclusion of such details,
however, can determine if the consumers of the analyses can independently evaluate the
results or exercise their own judgment regarding the insights they provide.
The process of addressing these questions, like the process of an NDA study, is likely to
be iterative. NDA studies frequently benefit from a cycle of updates or refinements to the
input parameters based on the insights gained as the study progresses. For example, the
results of the initial analyses may indicate that certain parameters have a greater influence
on the computed response than originally anticipated, such that additional effort to refine
the estimates of those parameters is warranted. In fact, addressing the following questions
as the NDA study progresses can help identify when such a cycle of refinement may or
may not be warranted. Thus, the order of the following questions should not be viewed as
implying a linear process for either the NDA study or its review.
Have the potential failure modes associated with seismic loading been summarized or
discussed?
Have the engineering behaviors important to those failure modes been adequately
identified? Are the limitations in the NDA model's ability to simulate the key engineering
behaviors acknowledged?
• The choice of a simpler versus more complex analysis model should be justified
based on the degree to which the key behaviors can be modeled and the degree to
which a more refined analysis may or may not affect final decisions regarding risk
mitigation efforts or remedial works.
• Limitations in the numerical model's ability to simulate certain behaviors should
be acknowledged and how they may affect the performance assessment should be
discussed. For example, current two-dimensional models cannot address most
cracking mechanisms and so crack depths must be estimated by other means.
Have the constitutive models and numerical modeling procedures been fully described
and validated?
Has the basis for all material parameters been described and related to the site
characterization?
• The input parameters for all components of the numerical model should be listed
in tables, including all geotechnical materials, structural elements, and any
interface elements.
• The basis for all soil and rock constitutive model parameters should be discussed
with reference to the site characterization data. A geologically based description
Have the key material properties and their uncertainties been identified and adequately
addressed?
• Key parameters affecting the response, and the relative magnitude of these
effects, should be identified through sensitivity studies.
• The adequacy of the site characterization for support of the analysis models
should be discussed. For example, the potential for additional explorations to
affect final decisions should be addressed based on the geologic setting, available
site characterization data, the likelihood of additional explorations to reduce
uncertainty in the input parameters, and the sensitivity of the analysis results to
reasonable variations in the estimated material properties.
Are the constitutive models able to approximate the important soil behaviors for both the
static and dynamic phases of the analysis? Have the strengths and limitations of the
constitutive model been assessed and documented?
Have the modeling approaches for the static and dynamic phases of the analyses been
documented and are the approaches reasonable?
Are the selected ground motions consistent with both the seismic hazard evaluation and
the numerical model?
• The static stress state (e.g., σ′yy, σ′xx, τxy), pore pressure (u), coefficient of lateral
earth pressure Ko (where Ko = σ′xx/σ′yy), and static shear stress ratio α (where α =
τxy/σ′yy) should be documented with contour plots.
• Equipotential plots for hydraulic head should also be provided as an aid in
evaluating the steady state seepage solution. The pore pressure head distribution
should be compared to available piezometric measurements if available.
• The initial stress state, as documented on the contour plots, should be consistent
with the anticipated variations across material zones and beneath slopes (e.g., no
unreasonable arching of stress between zones, no areas of excessively high Ko, no
unreasonable artifacts of modeling procedure, no random fluctuations). Simplified
approaches for static analysis have been known to introduce local stress
concentrations that are not realistic.
• If the initial pore pressure conditions were imposed assuming a ground water
table and hydrostatic conditions, the potential impacts of this approximation on
the dynamic response should be discussed.
• Contour plots of total unit weight should be provided to ensure the proper mass is
included both above and below the phreatic surface. If elements above the
phreatic surface fully desaturate during a steady-state seepage analysis, some
analysis programs will compute unit weights for these elements that are less than
typical moist unit weights.
Is the estimated dynamic response reasonable, and has it been presented in sufficient
detail to allow for this assessment?
• The following information should be provided and assessed for at least one input
motion:
o Acceleration time series for a few key points (e.g., crest, downstream face,
upstream face, toes, top of rock, base of model).
o Acceleration response spectra for the above points.
o Displacement time series for the above points.
o Spectral amplification ratios for key points, such as the crest relative to the
bedrock outcrop motion. Alternatively, Fourier amplification spectra can
be used to evaluate predominant response frequencies.
o Deformed shapes with contours of shear strains, excess pore water
pressure ratios, and displacements. Deformed shapes with displacement
vectors can also be valuable.
o For cases with undrained strength loss (e.g., residual strength or remolded
strength), contour plots of the final strength distribution.
o Displacement, moment, and shear envelopes for embedded beam-type
structures.
o Stress-paths and stress-strain response for representative elements.
• Response summaries should be provided for each set of modeling conditions and
suite of ground motions (e.g., peak crest acceleration and displacements, ground
surface displacement profile, peak response of embedded structural elements).
• The full suite of ground motions should be evaluated using best-estimate input
parameters.
• Significant differences in the character of the estimated response over the range of
input motions and modeling conditions should be identified and assessed.
• The magnitude of base bending should be described if a compliant base is used
for both the x and y directions.
• If remedial measures such as ground remediation are included in the analysis, the
response of these measures as well as their impact to the dam response should be
clearly described.
• Comparisons of the estimated responses to field observations from past
earthquakes and, where possible, to the results from other applicable analysis
techniques (e.g., two-dimensional equivalent linear response analyses) should be
provided.
Is the post-dynamic stability analysis reasonable, and has it been sufficiently documented
to permit this assessment?
• The following information should be provided and assessed for the same input
ground motion(s), nodal points, and elements detailed in Aspect 8:
Have sufficient parametric analyses been performed to identify the behaviors and
parameters of most importance to the computed responses?
Have the uncertainties and limitations in the numerical model been identified and
addressed?
Has the case been made, in plain language, for the conclusions and recommendations
drawn from the results of deformation analyses?
Nonlinear deformation analyses (NDAs) are a valuable tool for evaluating soil and soil-
structure systems affected by liquefaction, cyclic softening, and/or large seismic loads. A
high-quality NDA study can provide more realistic assessments of deformation
mechanisms and potential failure modes than are possible with simplified equivalent-
static or equivalent linear analysis methods.
A suggested checklist of questions was provided for participants and reviewers of NDA
studies. The questions are not exhaustive but represent the authors' perspectives on many
of the details that can significantly affect NDA results for embankment dams. These
details should be addressed or considered in engineering analyses and reports
representing the highest levels of practice. An additional objective was to foster
discussion on items that should be addressed in engineering reports summarizing NDA
studies. It is hoped that advancing the standard of practice for documentation will help
advance the standard of practice for NDA studies and ultimately benefit both consumers
and producers of NDA studies.
ACKNOWLEDGMENTS
The opinions expressed in this paper draw from the practices of numerous colleagues
whom the authors have been fortunate to work or interact with over many years. The
authors also appreciate the valuable comments and suggestions provided by Dr. Rambod
Hadidi and Mr. Andrew Dinsick during their review of the draft paper.
REFERENCES
Beaty, M. H., and Byrne, P. M. (2011). UBCSAND constitutive model: version 904aR.
Documentation report: UBCSAND constitutive model on Itasca UDM Web Site,
February.
Beaty, M.H. and Perlea, V.G. (2011). Several Observations on Advanced Analyses with
Liquefiable Materials. In Proc., 31st USSD Annual Meeting and Conference, United
States Society on Dams, San Diego, April 11 – 15, 2011. (pp. 1369-1397).
Boulanger, R. W., and Ziotopoulou, K. (2015). "PM4Sand (Version 3): A sand plasticity
model for earthquake engineering applications." Report No. UCD/CGM-15/01, Center
for Geotechnical Modeling, Department of Civil and Environmental Engineering,
University of California, Davis, CA, 112 pp.