Hydro-Québec's Embankment Dams

Seismic Assessment Progressive Approach

Verret D.1 and Péloquin É.1

1
Hydro-Québec Production, Montréal, Québec, CANADA

E-mails: Verret.Daniel@hydro.qc.ca and Peloquin.Eric@hydro.qc.ca

ABSTRACT: This paper presents Hydro-Québec's progressive approach for assessing and verifying
the seismic assessment of embankment dams and their foundation. This approach is based on 5
analysis levels (Levels 0, I, II, III, IV). It includes linear-equivalent analysis and non-linear analysis in
total stresses or effective stresses. The seismic safety assessment methodology is described by
detailing the general principle that a seismic assessment should progress to a higher level of analysis if
the previous level’s result indicates an inadequate seismic safety margin. The verification of the
potential shear strength loss by liquefaction or cyclic softening in the methodology and the analysis
progression according to the results of this verification is explained. A summary table listing the
number and types of seismic analyses carried out following this progressive approach is also
presented.

1 Introduction
For several years, Hydro-Québec has worked to develop and improve their embankment dams
seismic safety assessment in accordance with state-of-the-art practices. In 2001, Hydro-
Québec adopted an in-house seismic parameters selection guide based on seismological work
supplied by the Canadian Geological Survey. It details the seismic hazard approach and
provides peak bedrock acceleration (PRA), pseudospectral acceleration (PSA) and the
horizontal seismic coefficient (kh). In 2003, Hydro-Québec introduced guidelines for dam
seismic safety assessment which detailed the progressive approach methodology for non-
liquefiable soils. Since then, many efforts have been continuously deployed, including
research projects and site specific studies, to improve the embankment dams seismic
assessment progressive approach. This paper presents an up-to-date seismic analysis
progressive approach and summarizes the principal case studies conducted while following
this approach.

2 Methodology for Assessing Seismic Safety

The overall methodology for assessing the seismic safety of hydraulic structures like
embankment dams requires consideration of the following three steps:
1. Selecting the maximum design earthquake (MDE) and characteristics of the associated site
seismic hazards
2. Calculating the structural response of the system when subjected to the selected seismic
excitation
3. Interpreting results to quantify the safety margin
Each of these steps can be carried out with a variable degree of sophistication. Step 1 is
documented in the previously mentioned seismic parameter selection guide. For steps 2 and 3,
the progressive approach is followed.

621
3 Hydro-Québec's Seismic Analysis Progressive Approach
A progressive approach was adopted by Hydro-Québec to assess and verify embankment
dams and their foundation seismic safety. This approach is based on 5 analysis levels
(Levels 0, I, II, III, IV). The degree of conservatism decreases with the progression in these
analysis levels.
The general principle of the progressive approach adopted is that seismic assessment should
move to a higher level of analysis if the results of the previous level indicate an inadequate
seismic safety margin. The incremental methodology by analysis levels and each analysis
methodology are inspired by the worldwide practice for seismic assessment of dams (ICOLD,
CDA, USBR, FERC, and the FEMA). This methodology also includes a progressive
liquefaction susceptibility assessment approach integrated in the seismic safety analysis.
Figure 1 presents a flowchart of the application of the seismic analysis and the liquefaction
assessment methodologies for dams.

Figure 1: Embankment Dam and Foundation Seismic Analysis

622
Based on 5 levels of analysis (Levels 0, I, II, III, IV), the following approach is recommended
to assess and verify the seismic safety assessment of overburden or embankment dams:
Level 0: Preliminary assessment of the seismic safety of the structure: screening and
decision-making on seismic safety
Level I: Pseudostatic analysis (seismic coefficient method)
Level II: Analysis of permanent deformations (simplified methods)
Level III: Linear-equivalent or non-linear dynamic total stresses analysis for
permanent deformations assessment
Level IV: Non-linear dynamic effective stresses analysis for permanent deformations
assessment
In this process, soils that lose more than 15% of their static shear strength by liquefaction or
by cyclic softening are excluded from pseudostatic and conventional pseudodynamic type
analyses. For those soils, dynamic analyses are conducted, deformations are evaluated
(usually with non-linear analysis) with or without liquefaction (partial to complete
liquefaction).

4 Main Characteristics of Each Analysis Levels

4.1 Level 0: Preliminary Assessment of the Seismic Safety
The seismic safety analysis begins with a preliminary assessment which includes a pre-
selection of technical criteria (geotechnical, geometrical and seismological) and a decision-
making process. The first step of the progressive approach also includes verification of the
susceptibility for shear strength loss of the embankment dam and its overburden foundation
by liquefaction or cyclic softening.
This first phase of the earthquake assessment for earthworks of a hydroelectric facility aims to
establish a pre-selection of dams or dikes that are exposed to a certain risk and then select
those which should be retained for a more detailed analysis. This evaluation does not include
any extensive calculation; it is completed by evaluating the technical file (design report, such
as build drawings, field and laboratory investigations) and characteristics of the dam and its
foundation and by applying pre-selection criteria. It takes into account the local and regional
geological and seismological information, the seismic zoning and the seismic parameters
(Peak ground acceleration (PGA), PSA and seismic coefficients).
It also includes screening and seismic safety decision making using a preliminary assessment
of the susceptibility to shear strength loss by liquefaction or cyclic softening. The
methodology is based on the soil type: Seed method is used for granular soils (based on the
Geological Survey of Canada (GSC) seismic hazard, PGA, PRA, PSA, rd coefficient, free
field conditions, etc.) and cyclic softening or mobility criteria (function of parameters
calculated from Atterberg limits and the undrained shear strength) for coherent and plastic
soils (Bray and Sancio, 2006, Boulanger et Idriss, 2007). It also takes into account the seismic
zoning, the ground motion seismic parameters, and the local and regional seismicity.
If there is no risk of potential shear strength loss by liquefaction or cyclic softening, it is
necessary to continue the progressive approach with the next level analysis (when a result
indicating an adequate seismic safety margin is obtained, the process is generally stopped).
For cohesive soils susceptible to cyclic softening, the progressive approach of the seismic
analysis will consist of carrying out a Level III analysis.

623
For granular materials, if the Level 0 analysis confirms a liquefaction susceptibility
(SF < 1.0), the next step of the progressive approach should be to carry out a soil response
Level I analysis to evaluate more precisely (less conservatism) the CSR. However, it could
also be to directly evaluate the stability against slip failure in post-seismic conditions with the
residual strength of liquefied or softened soils and verify whether or not the safety factor with
these reduced shear strengths is equal or higher than 1.0. If the safety factor satisfies this
condition, an estimation of the post-liquefaction settlement analysis must confirm if the
freeboard is sufficient to enclose the reservoir. If the safety factor from the post-seismic
analysis is lower than 1.0, two options are available: 1) to conduct directly remedial work or
2) to invest using a more sophisticated dynamic analysis employing Level III.

4.2 Level I: Pseudostatic Analysis (Constant Horizontal Seismic Coefficient Method)

The pseudostatic method is the central element of Level I analysis if it is judged that the
foundation and fills are not likely to lose more than 15% of their shear strength during an
earthquake. With the pseudostatic method, the earthquake effects are represented by a
constant horizontal and/or vertical acceleration (the seismic excitation is assimilated to an
equivalent static force). These pseudostatic accelerations induce inertial forces which act
through the centroid of the failure soil mass in a limit equilibrium analysis. The safety factor
result of a pseudostatic stability analysis must be greater than 1.0 using the seismic coefficient
of the Hydro-Québec guide. The horizontal seismic coefficient (kh) is set at 50% of the PRA.
The choice of this criterion is based on Hynes-Griffin and Franklin’s (1984) research. By
applying the Newmark sliding rigid block analysis, Hynes-Griffin and Franklin (1984)
concluded that the peak deformation would not exceed 1 m if the pseudostatic safety factor is
at least 1.0 and the seismic coefficient is set at 50% of the PRA (kh=0.5*PRA/g).
If the pseudostatic analysis safety factor is less than 1.0, it is necessary to proceed to a Level
II deformation analysis.
The pseudo-static method is applicable only if the materials are not susceptible to liquefaction
or cyclic mobility. This means that the materials should not generate excess pore pressures
nor lose more than 15% of their shear strength during an earthquake.
Liquefaction Susceptibility and Cycling Softening
If there is susceptibility for shear strength loss either by liquefaction or by cycling softening
based on the Level 0 assessment, the next analysis of the seismic stability of the dam will
consist of conducting a soil response dynamic analysis in order to evaluate more precisely the
CSR. The aim of this higher level of analysis is to confirm the liquefaction susceptibility of
granular soils or the cycling softening of cohesive soils.
After completing the soil response dynamic analysis, for granular soils, if no risk of
liquefaction susceptibility is confirmed, a conventional pseudostatic analysis is conducted and
if required, the next level of seismic analysis must be completed. If there is an identified
cyclic softening susceptibility for cohesive soils, even after obtaining a less conservative CSR
when compared to previously calculated CRR (Boulanger and Idriss, 2007) , it is necessary to
continue and perform a Level II analysis.
Limitation of the Pseudostatic Analysis Method for a Seismic Assessment
The work of a special international expert committee mandated by Hydro-Québec to study
potential projects involving dams to be built on sensitive clays in the 1980's highlighted an
unrealistic assessment of seismic stability with the pseudostatic method in the case of dikes
founded on natural soft clay with a quasi-constant low undrained shear strength (su) value

624
with depth, particularly when the slopes are relatively flat with deep critical surface failures
based on the stability analysis.
Wilson and Marsal (1979) reported that Seed (1967) had noted a limitation of the pseudostatic
method in three case studies involving embankment dams constructed on a soft and sensitive
clay foundation. Experimental data from these sites exposed shallow surface failures instead
of the critical deep surface failure obtained by the pseudostatic method which reached the
bottom clay deposit base with no influence of the shear strength values.
Wu et Al. (1991) carried out a parametric study of the pseudostatic limit equilibrium method
to evaluate the seismic stability of clay slopes. Their work shows that the safety factor
decreases gradually while the failure surface becomes deeper and deeper during the analysis
and the final critical failure surface reaches the bottom of the clay deposit as described by
Wilson and Marsal (1979). Then the minimal safety factor for this type of analysis for a soft
clay foundation reaches an unrealistic low value.
In practice, the effect of adding berms or adopting milder slopes should increase the seismic
stability of an embankment founded on a clay foundation. However, because of the
pseudostatic method limitations for a dam founded on a soft clay deposits, the addition of
seismic stabilizer features make the safety factor even lower than the same dam without
berms and having steeper slopes.
Based on experience and literature, no pseudostatic analysis should be performed for cohesive
soils, at least for soft clay.

4.3 Level II: Analysis of Permanent Deformations (Pseudodynamic Simplified Methods)

The pseudodynamic method consists of calculating the dam crest displacement. For this type
of analysis the vibrational character of the seismic excitation is considered.
This type of simplified permanent deformation methods refers to the Newmark method (1965)
and the Makdisi and Seed method (1978). The Jansen, Bureau and Swaisgood empirical
methods for settlement assessment are also included in this level of analysis.
These simplified permanent deformation methods require knowledge of the crest acceleration
or potential instable volume of soil (block) acceleration which will initiate instability, called
yield acceleration (ay). The coefficient ay represents the seismic coefficient corresponding to
the pseudostatic acceleration required to induce instability of a soil mass (safety coefficient
of 1.0 in a limit equilibrium analysis). It represents the acceleration beyond which permanent
deformations occur if the earthquakes ground motion induced acceleration exceeds this ay
reference value. This coefficient (ay) can also be established by performing a deformation
dynamic analysis. At the technical level, the relative soil mass (or block) movement
(interpreted as deformations) estimation consists of integrating the relative acceleration twice
(area between the earthquakes ground motion induced acceleration and the ay reference). For
most simplified permanent deformation methods, simplified relations are a function of the
ratio between ay and the maximum earthquakes ground motion induced acceleration (amax). If
deformations are significant with simplified methods, it is recommended to proceed with a
more rigorous method. The Federal Energy Regulatory Commission (FERC, 1991) considers
that maximum acceptable settlement with this method should not exceed 0.60 m.
For cohesive soils, due to the limitation of the pseudostatic limit equilibrium method for soft
clay cohesive soils, if a Newmark type analysis is performed, the yield acceleration has to be
calculated by a deformation Level III dynamic analysis. if the Level 1 analysis has indicated a
cyclic mobility susceptibility (SF < 1.0), the next step of the seismic stability analysis will
consist of conducting laboratory cyclic tests in order to verify the cyclic mobility

625
susceptibility and to quantify the residual cyclic shear strength. The next step will be to
perform a Level III analysis considering the residual cyclic shear strength. If no cyclic
mobility softening is anticipated (SF > 1.0), a Level III analysis is perform with the intact
cyclic shear strength (could be reduced based on engineer judgement).
For granular materials, if the Level 1 analysis has indicated a liquefaction susceptibility
(SF < 1.0), the next step of the seismic stability could be to evaluate directly the stability
against slip failure in post-seismic conditions with residual strength of liquefied or softened
soils and to verify if the safety factor with these strength losses is equal to or higher than 1.0.
Following this analysis, a post-liquefaction reconsolidation settlement analysis must be
conducted to ensure the reservoir closure. At the end of this analysis, if the safety factor of the
post liquefaction stability analysis is still lower than 1.0 or if the calculated settlements
exceeds the freeboard required for reservoir closure, remedial work must be conducted.
Deformation dynamic Analysis: Levels III and IV
If the safety margin is still insufficient, it is possible to apply one of the dynamic methods
where the vibrational character of the seismic excitation is fully considered with various
levels of sophistication according to the target analysis (linear-equivalent, non-linear in total
stresses, non-linear in effective stresses, all with various behavior laws and in 2D or 3D).

4.4 Level III: Linear-Equivalent or Non-Linear Dynamic Total Stresses Analysis for
Permanent Deformations Assessment
This level refers to a dynamic analysis by finite elements or finite differences methods. It
could be a linear equivalent method of analysis or non–linear time domain method of analysis.
For a homogenous clay embankment founded on a soft clay foundation, the Level 3 analysis
follows the published Waba dam permanent deformations dynamic analysis which integrates
a Newmark type analysis (GEO-SLOPE International Ltd, 2005).
If it is anticipated that significant pore pressures occur, it is preferable to conduct a Level IV
analysis.
At this time, Hydro-Québec has completed some 2D permanent deformations dynamic total
stresses analyses and considers itself still in appropriation with this type of numerical
techniques.

4.5 Level IV Non-Linear Dynamic Effective Stresses Analysis for Permanent

Deformations Assessment
Dynamic non-linear analysis in effective stresses using finite elements or finite differences
method is proposed when it is anticipated that significant pore pressures could occur during a
prospective earthquakes. A dynamic effective stresses analysis is considered a highly
sophisticated analysis which requires several geotechnical laboratory and field test results.
This type of analysis could be coupled with UBCSand or another behavior law which takes
into account as accurately as possible the generation of excess pore-pressures and the shear
strength loss anticipated during an earthquake (MDE).
This type of dynamic analysis is still the focus of many university research projects. Hydro-
Québec has not yet carried out this type of analysis for embankment dams seismic safety
assessments.

626
5 Case Studies
Table 1 lists the seismic analyses that were conducted following the Hydro-Québec's
progressive approach. It excludes those trivial analyses that requested only a standard
pseudostatic analysis after completing the Level 0 analysis.
At this time, no Level IV analysis has been performed and only 3 cases requested a Level III
analysis. Based on this table, for those seismic studies which requested a higher Level
analysis than a standard pseudostatic analysis, the seismic stability was confirmed in more
than 50% of dams or dikes studied after completing a Level I analysis.

6 Conclusion
The embankment dams and foundation seismic progressive approach provides a valuable
engineering tool to study methodologically the seismic stability of dams according to state-
of-the-art practices. This methodology allows a gradation in the analysis according to local
seismicity and site conditions.
Hydro-Québec is continuing its historic commitment to research development in this field of
expertise so as to continuously improve its practices.

627
 Table 1: Summary Table Listing Seismic Safety Analyses Following the Progressive Approach
Dam and Dike Foundation
Case Crest Seismic Zone and Dynamic Analysis and Other
Max. Height Analysis Level
Studies Section Slope UP/DW Length Deposits Thickness PRA Analyses
(m)
(m)
0-3 m (peat)
Central till core with sand Peat followed by 1D and cyclic laboratory test
11.9 11H:1V 560 30 m (clay) 0 - I - II
shoulders and large berms clay on till (clay)
5-35 m (till)
Sand (cut off 25 m (sand), 10-25
Homogeneous earthfill dike (till) 1D and settlement analysis
6.9 3H:1V 1150 zone) and/or clay m (clay) and 5-20 0-I 1 (k=0.025)
A with a cut off (sand)
on till m (till) PRA=0.042g
Zoned rockfill dam with vertical 2H:1V on till Silty sand (till) or
22.2 641 <5m 0-I Pseudostatic analysis
till clay core 1.7H:1V on roc rock
Zoned rockfill dam with vertical
18 1.7H:1V 140 Rock --- 0-I Pseudostatic analysis
till clay core
Homogenous fill (silt), with 3.5H:1V / 2 (k=0.05) PRA=
B 9.1 580 Silt 20 m 0 - I - II - III 1D/2D
upstream till core 3.0H:1V 0.07g
4.0H:1V/
Homogenous fill 27.0 44 Silty sand (till) 10 m 0-I Post-liquefaction analysis
4.1H:1V 2 (k=0.10)
C
4.7H:1V/ PRA=0.20 g
Homogenous fill 9.0 492 Silty sand (till) 6-8 m 0-I Pseudostatic analysis
3.9 H:1V
Central clay core, sand filter with 31 (including
2H:1V (with 0 - 22 m (sand), 40 3 (k=0.10)
sand and gravel for shoulders and deep cut off 1000 Sand and clay 0-I 1D and pseudsStatic analysis
berm on clay) - 160 m (clay) PRA=0.14g
berms (founded on clay) trench)
D
1D and pseudostatic analysis
Central clay core, sand filter with 20 (including a 15 m (clay), 75 m 3 (k=0.10)
2.5H:1V 1190 Clay and sand 0 - I - II and cyclic laboratory test
sand and gravel shoulders key trench (sand) PRA=0.14g
(clay)
Homogenous fill, silt, sand and 5H:1V /
14.3 70 1D and Post-liquefaction
gravel 4.5H:1V 3 (k=0.10)
E Rock --- 0 - I - II
Homogenous fill, silt, sand and 3.6H:1V / PRA=0.14 g
18.3 50 1D and pseudo-static analysis
gravel 4.0H:1V
5-35 m (clay), 0-5 3 (k=0.15) 1D/2D and cyclic laboratory
F Homogenous clay fill 8 3H:1V / 2H:1V 45000 Clay and till 0 - I - II - III
m (till) PRA=0.32 g test (clay)
1.5H to 4H:1V
3 (k=0.10) 1D/2D
G Homogenous till fill 6 / 1335 Till 4m 0 - I - II - III
PRA=0.14 g Post-liquefaction analysis
2.5H:1V
Homogenous fill, sand, and 2.1H:1V/ Sand with silt 3 (k=0.10)
H 8 150 0-8 m 0 - II Post-liquefaction analysis
gravel with silt 2.7H:1V (Peat locally) PRA=0.14 g
Central till core with sand Silty sand and 12 m (sand), 30 m /
13.7 2H:1V 1800
shoulders clay and sand 5-30 m (clay)
4 (k= 0.15)
I > 10H:1V 0-I 1D
Central till core with sand for 30 m (clay), 5-30 m PRA = 0.3 g
26 (2H:1V with 900 Clay and sand
shoulders and large berms (sand)
large berms)

628
7 References
[1] ICOLD (International Commission on Large Dams). (2017). Bulletin, Preprint 167,
Regulation of Dam Safety: An overview of current practice worldwide.
[2] USBR (United States Bureau of Reclamation). (2015). Design Standards No. 13:
Embankment Dams, Chapter 13 - "Seismic Analysis and Design ", Department of
Interior, Bureau of Reclamation, Dam Safety Office, Denver, Colorado.
[3] Idriss IM, Boulanger RW. (2008). Soil Liquefaction During Earthquakes. Monograph
MNO-12, Earthquake Engineering Research Institute, Oakland, CA, 261 pp.
[4] CDA (Canadian Dam Association). (2007). Edition 2013, Dam Safety Guidelines.
Engineering Guidelines for the Evaluation of Hydropower Projects.
[5] Boulanger RW, Idriss IM (2007). Evaluation of cyclic softening in silts and clays, ASCE,
Journal of Geotechnical and Geoenvironmental Engineering, 133(6), 641-652.
[6] Bray J, Sancio R. (2006). Assessment of the Liquefaction Susceptibility of Fine-Grained
Soils, Journal of Geotechnical and Geoenvironmental Engineering, Vol. 132, No. 9:
pp. 1165-1177.
[7] FERC (Federal Energy Regulatory Commission). (2006). Engineering Guidelines for the
Evaluation of Hydropower Projects, Chapter 4 – Embankment Dams (Revised Draft
Version).
[8] Boulanger RW, Idriss IM. (2006). Liquefaction Susceptibility Criteria for Silts and Clays.
Journal of geotechnical and geoenvironmental engineering, 132(11), 1413-1426.
[9] FEMA (Federal Emergency Management Agency). (2005). Federal Guidelines for Dam
Safety: Earthquake Analyses and Design of Dams, FEMA 65.
[10] GEO-SLOPE International Ltd, (2005). Newmark Deformation Analysis, Canadian Dam
Association, Calgary, Alberta, Canada, 2005.
[11] GEO-SLOPE International Ltd, (2005). Waba Dam Permanent Deformation due to an
Earthquake, Canadian Dam Association, Calgary, Alberta, Canada.
[12] Swaisgood JR. (1998). Seismically-Induced Deformation of Embankment Dams. 6th U.S.
National Conference on Earthquake Engineering, Seattle, Wash, U.S.A.
[13] Kramer SJ. (1996). Geotechnical Earthquake Engineering. Prentice-Hall, 651 p.
[14] FERC (Federal Energy Regulatory Commission). (1991). Engineering Guidelines for
Evaluation of Hydro-Power Projects, Office of HydroPower Licensing, FERC 0-119-2.
[15] Wu XY, Law KT, Selvadural APS. (1991). An Examination of the Pseudo-Static Limit
Equilibrium Method for the Dynamic Stability Analysis of Slopes. In Proceedings of the
44th Canadian geotechnical conference, Calgary, 1961-1968.
[16] Jansen RB. (1988). "Advanced Dam Engineering for Design, Construction and
Rehabilitation", Van Nostrand Reinhold, H.Y., 811 p.
[17] Dobry R, Vucetic M. (1987). Dynamic Properties and Seismic Response of Soft Days
Deposits: State-of-the-Art Paper. Proc. Int. Symposium on Geotechnical Engineering of
Soft Soils, Mexico, 1987.
[18] Jansen RB. (1990). "Estimation of Embankment Dam Settlement Caused by Earthquake",
Int. Water Power and Dam Construction, 42(12), 35-40.

629
[19] USBR (United States Bureau of Reclamation). (1987). Design of Small Dams. United
States Department of Interior, Bureau of Reclamation, Third edition.
[20] Hynes-Griffin M, Franklin A. (1984). Rationalizing the Seismic Coefficient Method.
U.S., Army Corp of Engineers, WES, Miscellaneous paper GL-84-13.
[21] SEBJ (Société d’énergie de la Baie-James). (1983). Report of the Committee of
Specialists on Sensitive Clay of the NBR Complex, Final report - Volume 3, annexes 2
and 5- Report of the embankment stability subcommittee and Report of the slope stability
subcommittee.
[22] Seed HB. (1979). "Considerations in the Earthquake Resistant. Design of Earth and
Rockfill Dams", 19th Rankine Lecture, Geotechnique, 29(3).
[23] Wilson SD, Marsal RJ. (1979). Current Trends in Design and Construction of
Embankment Dams. New York, NY: ASCE.
[24] Makdisi FI, Seed HB. (1978). Simplified Procedure for Estimating Dam and
Embankment Earthquake-Induced Deformations. Journ. of the Geotech. Eng. Division.
Vol. 104, GT. 7, July. Page.
[25] Makdisi FI, Seed HB. (1977). A Simplified Procedure for Estimating Earthquake Induced
Deformations in Dams and Embankments. Report UCB/EERC-77/19, August.
[26] Newmark NM. (1965). The 5th Rankine Lecture: Effects of Earthquakes on Dams and
Embankments, Geotechnique, Vol. 5, No. 2.

630