November 2021

Seepage and Slope Stability Modeling for Embankment Dams

A Guidance Document for Planning, Interpreting, Verifying, and Reporting Results
Prepared for:
State of Montana, Department of Natural Resources and Conservation 1424 9th Avenue
Helena, MT 59620

Prepared by:
AECOM
7595 Technology Way
Denver, CO 80237 aecom.com

The material in this publication has been prepared in accordance with generally
recognized engineering principles and practices, and is for general information
only. The information presented should not be used without first securing competent
advice from qualified professionals with respect to its suitability for any general
or specific application. No reference made in this publication constitutes an
endorsement or warranty thereof by AECOM. Anyone using the information presented in
this publication assumes all liability arising from such use.

Acknowledgements

This document is intended to provide practical guidance for dam owners and
engineers on seepage and slope stability modeling of embankment dams, particularly
small embankment dams. This document is not intended to be an all-inclusive guide
for completing seepage and slope stability analyses for embankment dams. In many
instances, the document directs readers to other references that provide more
detailed information. In addition, an extensive list of references on the topic is
provided at the end of this document.
This document was prepared by AECOM, under contract to the State of Montana,
Department of Natural Resources and Conservation (DNRC). The work was authored by
Julie Heitland, P.E. and Harry Donaghy, P.E., and was reviewed by Ms. Jennifer
Williams, P.E. of AECOM and Ms. Michele Lemieux, P.E. of Montana DNRC.

Table of Contents

Acronyms and Abbreviations vi

1. Introduction 1-1
2. Seepage in Dams 2-1
2.1 What Is Seepage? 2-1
2.2 Seepage-Related Issues 2-2
3. Seepage Analysis of Dams 3-1
3.1 Seepage Analysis Principles 3-1
3.1.1 Darcy’s Law 3-1
3.1.2 Hydraulic Conductivity 3-2
3.1.3 Laplace Equation 3-4
3.2 Seepage Analysis Methods 3-6
3.2.1 Graphical Methods 3-6
3.2.2 Analogs (Physical Models) 3-8
3.2.3 Numerical Models 3-9
3.2.4 General Guidelines for Selecting a Seepage Analysis Method 3-12
3.3 When Is a Seepage Analysis Warranted? 3-13
3.3.1 What Is the Objective of the Seepage Analysis? 3-13
3.3.2 Can the Objective Be Met Adequately Without a Seepage Analysis? 3-14
3.3.3 Will a Seepage Analysis Accurately Capture the Objective? 3-15
3.3.4 Is There Time to Complete a Seepage Analysis (Not an Emergency)? 3-15
3.4 Planning for a Seepage Analysis 3-16
3.4.1 Modeling Approach 3-16
3.4.2 Minimum Data Requirements 3-21
3.4.3 Model Setup Considerations 3-25
3.4.4 Model Calibration Considerations 3-28
3.4.5 Model Convergence Considerations 3-29
3.5 Interpreting, Verifying, and Reporting Results 3-30
3.5.1 Interpreting and Verifying Results 3-30
3.5.2 Reporting Results 3-37
4. Slope Instability of Dams 4-1
4.1 What Is Slope Stability? 4-1
4.2 Slope Instability-Related Issues 4-2
5. Slope Stability Analysis of Dams 5-1
5.1 Slope Stability Analysis Principles 5-1
5.1.1 Stress-Strain Behavior 5-2
5.1.2 Undrained and Drained Conditions 5-3
5.1.3 Total and Effective Stresses 5-4
5.2 Slope Stability Analysis Methods 5-4
5.2.1 Limit Equilibrium Methods 5-5
5.2.2 Finite Element Method 5-10
5.2.3 Slope Stability Charts 5-11
5.3 When Is a Slope Stability Analysis Warranted? 5-11
5.4 Planning for a Slope Stability Analysis 5-13
5.4.1 Modeling Approach 5-13
5.4.2 Minimum Data Requirements 5-19
5.4.3 Model Slip Surface Considerations 5-31
5.4.4 Model Sensitivity Considerations 5-34
5.4.5 Model Convergence Considerations 5-38
5.5 Interpreting, Verifying, and Reporting Results 5-39
5.5.1 Interpreting and Verifying Results 5-39
5.5.2 Reporting Results 5-42
6. Conclusion 6-1
7. References and Additional Reading 7-1

Figures
Figure 2-1: Seepage Emanating from Downstream Toe of Embankment Dam (Photo
Courtesy of the DNRC) 2-1
Figure 2-2: Example of an Internal Erosion Pathway through the Foundation of an
Embankment Dam (Fell et al. 2008) 2-2
Figure 2-3: Example of Uplift Resulting in Blowout at the Downstream Toe of an
Embankment 2-3
Figure 2-4: Example of Increased Pore Pressures in an Embankment Dam Resulting in
Slope Instability 2-3
Figure 3-1: Schematic of Darcy’s Law (FEMA 2015b) 3-2
Figure 3-2: Schematic of Heterogeneous and Anisotropic Foundation 3-4
Figure 3-3: Schematic of (a) 2D Seepage Flow beneath Concrete Dam and (b) Steady-
State Flow across an Element of Foundation Soil (USACE 1993) 3-5
Figure 3-4: Flow Net for Seepage beneath Sheet Pile Wall in Permeable Foundation
(FEMA 2015b) 3-7
Figure 3-5: Flow Net for Unconfined Seepage through Homogeneous Embankment
(FEMA 2015b) 3-7
Figure 3-6: Graphically Constructed Phreatic Surface through Homogenous Embankment
(Adapted from Reclamation 2014) 3-8
Figure 3-7: Use of 2D Conducting Paper to Evaluate (a) Equipotential Lines and (b)
Flow
Lines (USACE 1993) 3-9
Figure 3-8: Numerical Seepage Model Results from the SEEP/W Program for Zoned
Embankment (Heitland et al. 2020) 3-9
Figure 3-9: Transient Seepage Analysis for Drawdown through Zoned Embankment
(Reclamation 2014) 3-17
Figure 3-10: Examples of When 3D Seepage Modeling Becomes Beneficial: (a) Irregular
Bedrock Foundation; (b) Convex Bend in Dam Alignment; (c) Complex Model Geometry
(Heitland et al. 2020) 3-19
Figure 3-11: Typical Volumetric Water Content Functions of Soils (Reclamation 2014)
3-20

Figure 3-12: Example Saturated / Unsaturated Hydraulic Conductivity Functions of

Embankment Soil (Reclamation 2014) 3-20
Figure 3-13: Cross Section Orientation for Numerical Seepage Modeling (Heitland et
al.
2020) 3-25
Figure 3-14: Typical Seepage Model Boundary Conditions from the SEEP/W Program
(Heitland et al. 2020) 3-27
Figure 3-15: Seepage Model Results from the SEEP/W Program Using 10-foot Mesh Size
(Approximately 1,000 Elements) (Heitland et al. 2020) 3-28
Figure 3-16: Seepage Model Results from the SEEP/W Program Using 5-foot Mesh Size
(Approximately 4,000 Elements) (Heitland et al. 2020) 3-28
Figure 3-17: Example Seepage Model Using SEEP/W (Heitland et al. 2020) 3-32
Figure 3-18: Seepage Model Results from the SEEP/W Program Showing Total Head
Contours and Flow Paths (Flow Net Simulation) (Heitland et al. 2020) 3-32
Figure 3-19: Seepage Model Results from the SEEP/W Program Showing Pore Water
Pressure Contours (Heitland et al. 2020) 3-33
Figure 3-20: Seepage Model Results Showing Flow Vectors (SEEP/W) (Heitland et al.
2020) 3-33
Figure 3-21: Seepage Model Results from the SEEP/W Program Showing Flow Section
(Graphing Tool) – Foundation Alluvium (Heitland et al. 2020) 3-34
Figure 3-22: Seepage Model Results from the SEEP/W Program Showing Flow Section
(Graphing Tool) – Foundation Bedrock (Heitland et al. 2020)3-35
Figure 3-23: Seepage Model Results from the SEEP/W Program Showing Horizontal
Seepage Gradients (Graphing Tool) (Heitland et al. 2020) 3-36
Figure 3-24: Seepage Model Results from the SEEP/W Program Showing Vertical (Exit)
Gradients (Graphing Tool) (Heitland et al. 2020) 3-36
Figure 3-25: Seepage Model Results from the SEEP/W Program Showing Vertical (Exit)
Seepage Gradient Contours (Heitland et al. 2020) 3-37
Figure 4-1: Slope Instability on Downstream Slope of Embankment Dam (FEMA 2016)
4-1
Figure 4-2: Types of Slope Instability Failures of Embankment Dams (Reclamation
1988) 4-2
Figure 4-3: Severe Longitudinal Crack on Downstream Embankment Slope (AECOM
2013) 4-3
Figure 4-4: Seepage Exiting Downstream Embankment Slope (AECOM 2013) 4-4
Figure 4-5: Slope Failure on Downstream Embankment Slope (AECOM 2013) 4-4
Figure 5-1: Embankment Slope and Potential Slip Surface (USACE 2003) 5-1
Figure 5-2: Generalized Stress-Strain Curve (Duncan et al. 2014) 5-2
Figure 5-3: Schematic of (a) Procedure of Slices and (b) Typical Forces on an
Individual
Slice (USACE 2003) 5-6
Figure 5-4: Sensitivity of Various Slope Stability Limit Equilibrium Analysis
Methods on
Factor of Safety (France and Winckler 2010) 5-9
Figure 5-5: Potential Sequence of Loading during Life of Embankment Dam 5-14
Figure 5-6: Mohr-Coulomb Shear Strength Envelope for Coarse-Grained Soils (Duncan
et
al. 2014) 5-22
Figure 5-7: Curved Shear Strength Envelope with Linear Interpretation and Apparent
Cohesion (c’) (AECOM 2015) 5-22
Figure 5-8: Drained Mohr-Coulomb Shear Strength Envelop for Clays (Duncan et al.
2014) 5-24

Figure 5-9: Drained Shear Strength of Stiff-Fissured Clays (Duncan et al. 2014)
5-25
Figure 5-10: Anisotropy Effects for Clays – (a) Stress Orientations at Failure and
(b) Undrained Shear Strength Anisotropy of Clays and Shales – UU Triaxial
Tests (Duncan et al. 2014) 5-26
Figure 5-11: Variation of Su/’v with OCR for Clays, measured in Anisotropically
Consolidated DSS Tests (Duncan et al. 2014) 5-27
Figure 5-12: Typical Piezometer Data Plot 5-30
Figure 5-13: Common Slip Surface Configurations (USACE 2003) 5-32
Figure 5-14: Examples of Local, Intermediate, and Global Slip Surfaces for Slope
Stability Modeling 5-33
Figure 5-15: Slope and Slip Surface with Tension Crack (Duncan et al. 2014) 5-34
Figure 5-16: Example 1 – Slope Stability Model for Homogeneous Embankment with Toe
Drain (France and Winckler 2010) 5-35
Figure 5-17: Example 2 – Slope Stability Model for Zoned Embankment with Chimney
and Blanket Drains (France and Winckler 2010) 5-35
Figure 5-18: Slope Stability Model Results for Example 1, Base Case Condition
(France
and Winckler 2010) 5-36
Figure 5-19: Slope Stability Model Results for Example 2, Base Case Condition
(France
and Winckler 2010) 5-36
Figure 5-20: Phreatic Surfaces Corresponding to Increasing Anisotropy Ratios and
Slope Stability Model Results for Variations in Phreatic Surface (France and
Winckler 2010) 5-38
Figure 5-21: Example Slope Stability Model Using UTEXAS4 (France and Winckler 2010)
5-40
Figure 5-22: Slope Stability Model Results Showing Factors of Safety for Trial Slip
Surfaces (France and Winckler 2010) 5-41

Tables
Table 3-1: Typical Permeability Ranges by Soil Type (Cedergren 1989) 3-3
Table 3-2: Typical Computer Programs for Modeling Seepage (FEMA 2015b) 3-11
Table 3-3: Guidelines for Selecting a Seepage Analysis Method (Adapted from FEMA
2015b) 3-12
Table 3-4: Examples That May or May Not Warrant a Numerical Seepage Model (Heitland
et al. 2020) 3-13
Table 3-5: Minimum Data Requirements for Seepage Modeling 3-21
Table 3-6: Typical Seepage Model Boundary Conditions (Heitland et al. 2020) 3-26
Table 3-7: Key Seepage Model Output Features for Checking the Validity of Results
(Heitland et al. 2020) 3-30
Table 5-1: Typical Coefficient of Consolidation Ranges by Soil Type (Duncan et al.
1990) 5-4
Table 5-2: Comparison of Slope Stability Limit Equilibrium Analysis Methods Using
Procedure of Slices 5-8
Table 5-3: Typical Computer Programs for Modeling Slope Stability Using Limit
Equilibrium Methods 5-9
Table 5-4: Typical Factor of Safety Criteria 5-18
Table 5-5: Minimum Data Requirements for Slope Stability Modeling 5-20

Table 5-6: Slope Stability Model Results for Sensitivity Case – Variations in Unit
Weight
(France and Winckler 2010) 5-37
Table 5-7: Slope Stability Model Results for Sensitivity Case – Variations in
Strength
(France and Winckler 2010) 5-37
Table 5-8: Summary of Slope Stability Model Results for Variations in Phreatic
Surface
(France and Winckler 2010) 5-38

Acronyms and Abbreviations

1D One-dimensional
2D Two-dimensional
3D Three-dimensional
ASDSO Association of State Dam Safety Officials
CD Consolidated-drained
cm/s Centimeters per second
CPT Cone penetrometer test
CU Consolidated-undrained
DNRC Montana Department of Natural Resources and Conservation
DS Direct shear
DSS Direct simple shear
EAP Emergency Action Plan
FEMA Federal Emergency Management Agency
FS Factor of safety against sliding
ft2/day Square foot per day
ft3/sec/ft Cubic foot per second per foot
gpm/ft Gallons per minute per foot
H:V Horizontal to vertical
IDF Inflow design flood
O&M Operations and maintenance
OCR Overconsolidation ratio
SHANSEP Stress History and Normalized Soil Engineering Properties
SPT Standard penetration test
Su Undrained shear strength
UC Unconfined compression
UU Unconsolidated-undrained

1. Introduction
Seepage and slope stability modeling are often proposed as part of an embankment
dam evaluation. However, the modeling objective is often not well defined, and the
model output may be nothing more than an expensive, colorful graphic without much
insight into the dam’s seepage- or stability-related issues, if not appropriately
modeled, interpreted, and documented in an analysis report.
Nonetheless, seepage and slope stability modeling can help engineers, regulators,
and owners better understand how seepage and stability may influence the
performance of an embankment dam and provide the information needed to guide future
dam safety actions. The following are some questions engineers, regulators, and
owners should be asking before proceeding with a seepage or slope stability model:
• What should engineers consider when proposing a seepage or slope stability
model to a dam owner or regulator?
• What should dam owners and regulators consider when reviewing an engineering
proposal that involves a seepage or slope stability model?
• How do engineers develop an efficient model to achieve the desired objective?
• What are the minimum data needs for a reliable model? How much effort will
the modeling take?
• How should the results be interpreted? How does one check for reasonableness
of the results?
• What if two-dimensional models cannot adequately represent actual conditions?
The purpose of this guidance document is to provide a basic understanding of the
standard of practice in preparing seepage and slope stability analyses of
embankment dams. This document provides tips, tools, and guidance for planning the
analyses, including modeling considerations, as well as interpreting, verifying,
and reporting the results. Basic seepage and slope stability concepts are
summarized throughout the document with references to additional publications that
elaborate on the concepts. This document does not provide guidance on how to
perform seepage and slope stability modeling but presents some basic modeling
considerations.
The content of this guidance document is intended for:
• Entry-level to senior-level dam safety professionals and engineers,
• Dam owners,
• Dam regulators, and
• All other members in the dam safety community with interest in seepage and
slope stability modeling of dams.
Information presented in this guidance document related to seepage analysis and
modeling considerations was primarily adapted from two partnering technical papers
prepared for the Dam Safety 2020 National Conference of the Association of State
Dam Safety Officials (ASDSO) titled Dam Seepage Models – Tools, Rules & Guidance
(From a Regulatory
Perspective) [Lemieux 2020] and Seepage Models – Tips, Tools & Guidance (From an
Engineer’s Perspective) [Heitland et al. 2020]. These papers were the
steppingstones to this more comprehensive guidance document that has been prepared
in conjunction with AECOM and the Montana Department of Natural Resources and
Conservation (DNRC).

2. Seepage in Dams
2.1 What Is Seepage?
Seepage is the flow of water through the porous space within a soil or rock mass.
In embankment dams, seepage can occur through the embankment, foundation,
abutments, or along embankment penetrations. This includes flow through a large
area of soil or concentrated flow along defects, such as cracks, loose lifts, rock
discontinuities (e.g., fractures and joints), and other pathways. The reservoir is
generally the largest source of water for seepage, but it may also come from
groundwater sources. Figure 2-1 shows an example of seepage emanating from the
downstream toe of an embankment dam.

Figure 2-1: Seepage Emanating from Downstream Toe of Embankment Dam (Photo Courtesy
of the DNRC)

Seepage and leakage occur to some degree at all embankment dams and is not
necessarily a problem if it is identified, monitored, evaluated, and controlled.
Seepage can become a dam safety concern if it is not controlled and results in
internal erosion, excess uplift pressures, or instability. Some factors that can
lead to uncontrolled seepage include the following (FEMA 2015b):

• Hidden construction defects that originated during design and construction

• Large unprecedented seismic or climatic events
• Deterioration of one or more seepage control features
• Gradual deterioration of the embankment or foundation from past seepage
• Unabated animal burrow or tree growth activity
2.2 Seepage-Related Issues
This guidance document discusses the planning and interpretation of seepage
analyses, which can provide insight into seepage-related issues. Several studies
have been conducted on the failure of embankment dams (e.g., Richards and Reddy
2007; Foster et al. 2000). These studies consistently showed that seepage-related
failures of dams comprise about one-half of all documented dam failures.
Identifying potential seepage-related issues pertinent to a site prior to
conducting a seepage analysis will help in establishing the objective, approach,
and data needs of the analysis. The following is a list of some of the seepage-
related issues that might be informed by seepage analysis:
• Internal erosion
• Uplift
• Increased pore pressures and slope instability
• Water loss (not a dam safety issue, but maintenance, nuisance, and economic
issues)
Internal erosion occurs when seepage causes detachment and migration of soil
particles. This soil movement can damage a dam’s earthen embankment, foundation, or
abutments. Internal erosion can develop in many forms and through many pathways.
Figure 2-2 depicts an example of an internal erosion pathway.

Figure 2-2: Example of an Internal Erosion Pathway through the Foundation of an

Embankment Dam (Fell et al. 2008)
Several factors influence the susceptibility of a dam and its foundation to
internal erosion, one of which is the magnitude of seepage forces that would be
expected to occur under different loading conditions. Seepage analyses can be used
to evaluate seepage gradients, velocity, and direction at various locations under
anticipated loading conditions. For information on internal erosion pathways and
mechanisms, refer to additional references in Section 7 (AECOM 2016; FEMA 2015a;
FEMA 2016; Fell et al. 2008).

Figure 2-3), can cause damage to

concrete spillway chutes, and can cause instability in concrete structures or their
foundations. Uplift pressures can be estimated by seepage analyses.

Figure 2-3: Example of Uplift Resulting in Blowout at the Downstream Toe of an

Embankment

Figure 2-4 depicts slope instability in an embankment. Pore pressures in the

embankment and foundation of a dam can be estimated from seepage analyses.

Figure 2-4: Example of Increased Pore Pressures in an Embankment Dam Resulting in

Slope Instability

3. Seepage Analysis of Dams

A seepage analysis is a computational method that models the conditions of an
embankment dam to estimate seepage characteristics through, beneath, and/or around
the embankment. A seepage analysis can provide an understanding of the following:
• Seepage pathways,
• Seepage flow rates and velocities,
• Seepage gradients,
• Total head, pressure head, and pore water pressures, and
• Saturation.
There are a variety of seepage analysis methods for evaluating embankment dams that
range from simple graphical approaches completed by hand to more complex numerical
modeling using computer programs. Selecting an appropriate analysis method will
depend on the objective of the seepage analysis and complexity of the situation.
Common applications of a seepage analysis include quantifying the performance of
the dam under current, expected future, and/or extreme conditions; evaluating
observed seepage conditions; or evaluating and comparing seepage control design
alternatives.
The results of a seepage analysis can be used for the following:
• Evaluating internal erosion potential,
• Evaluating slope stability implications, and
• Designing seepage control systems, including collection (filters and drains,
toe drains, etc.) and reduction (low permeability blankets, cutoff walls, etc.).
3.1 Seepage Analysis Principles
The theoretical principles that govern the movement of energy through conducting
media (such as electricity or heat) similarly apply to the movement of water
through soils. Water moves from a higher energy state to a lower energy state. The
differential in energy is the amount of energy required to overcome the soil’s
resistance to the flow of water. The theory used to understand and evaluate the
seepage response of embankment dams is based on Darcy’s Law and the Laplace
equation. This section provides a basic overview of these principles for seepage
analysis. For more information on seepage analysis principles beyond that discussed
below, refer to additional references in Section 7 (FEMA 2015b; Reclamation 2014;
USACE 1993; Cedergren 1989).
3.1.1 Darcy’s Law
Similar to Ohm’s Law, which governs the flow of electricity, Henry Darcy derived an
empirical formula in 1856 to explain the behavior of seepage through saturated
soils. Darcy’s Law states that the amount of flow is directly proportional to the
hydraulic gradient. Darcy’s Law is expressed by the equation:

𝑞 =

𝑘𝛥ℎ𝐴

= 𝑘𝑖𝐴 = 𝑣𝐴

Where: q = Rate of Seepage

k = Hydraulic Conductivity or Permeability
Δh = Head Loss
A = Cross-Sectional Area Normal to Direction of Flow L = Length of Seepage Path
i = Hydraulic Gradient = Δh/L v = Discharge Velocity = ki
Figure 3-1 is a schematic illustrating the concept of Darcy’s Law, where a
prismatic soil sample is exposed to a head of water on the left side and a smaller
head on the right side. This results in water flowing through the soil sample at a
rate directly proportional to the hydraulic gradient.

Figure 3-1: Schematic of Darcy’s Law (FEMA 2015b)

Limitations to Darcy’s Law include the following (FEMA 2015b):

• Darcy’s Law is only applicable to laminar, steady-state flow through
saturated, homogeneous soils.

• Darcy’s Law is not applicable to flow through defects, such as cracks or rock
fractures and joints.
3.1.2 Hydraulic Conductivity
As Darcy’s Law states, the amount of flow is directly proportional to the hydraulic
gradient. The constant relating the flow to the hydraulic gradient is the hydraulic
conductivity (or permeability). Permeability is the ability of water to seep or
flow through void spaces in soil. A high permeability material will pass more flow
under the same gradient that a low permeability material will pass. Permeability is
one of the most highly variable material properties in

geotechnical engineering. Table 3-1 provides ranges of permeability values for a

variety of soil types. This is a rough guideline which should only be used to
compare permeability estimates as a reality check and not to be used directly in a
seepage analysis.
Table 3-1: Typical Permeability Ranges by Soil Type (Cedergren 1989)

Soil Type Permeability, k (cm/s)

Clays 1x10-7 to 1x10-9
Very Fine Sands, Silts, Mixtures of Sand Silt and Clay 1x10-7 to 1x10-3
Clean Sand, Clean Sand and Gravel Mixtures 1x10-3 to 1
Clean Gravel 1 to 1x102

Key factors affecting permeability include the following:

• Grain size and distribution,
• Soil structure,
• Density,
• Discontinuities/stratification, and
• Viscosity of fluid (not typically of consequence for embankment dams since
the fluid is normally water).
Permeability is also heterogeneous (i.e., varies with location) and anisotropic
(i.e., varies with direction of flow). For example, clay may have a low
permeability, but fissures or desiccation cracks can provide a preferential flow
path for water. In addition, alluvial soil deposits can lead to a higher
permeability in the horizontal direction than the vertical direction. Figure 3-2 is
a schematic illustrating heterogeneity and anisotropy.

Figure 3-2: Schematic of Heterogeneous and Anisotropic Foundation

Figure 3-2 depicts a heterogenous alluvium foundation, sitting below an embankment

dam, and overlaying bedrock. The alluvium foundation consists of a majority of
silty sand, however there are seams of clay and gravel. This is a heterogenous
foundation. Further, the gravel seam is anisotropic, meaning it has a higher
hydraulic conductivity in the horizontal direction, than in the vertical direction.
Finally, permeability varies with the degree of saturation. As the degree of
saturation decreases, the permeability can decrease by orders of magnitude.
Although there are several field and laboratory methods for estimating
permeability, the effects of heterogeneity and anisotropy are difficult to identify
and model with precision. Thus, selection of permeability values should be
considered to give order-of-magnitude levels of accuracy.
3.1.3 Laplace Equation
The Laplace equation is a partial differential equation that describes the flow of
water through homogeneous, isotropic soils. It is used in two- and three-
dimensional (2D and 3D) seepage analyses. The equation assumes that the quantity of
water entering an element must be equal to the amount leaving the element (i.e.,
steady-state flow). The Laplace equation can be arranged in terms of gradients and
permeabilities by applying Darcy’s Law. The Laplace equation for 2D and 3D flow is
expressed by:

2D: 𝑘

𝛿𝛿2ℎ

𝑥 𝛿𝛿𝑥2

𝛿𝛿2ℎ

𝑧 𝛿𝛿𝑧2

3D: 𝑘

𝛿𝛿2ℎ + 𝑘
𝛿𝛿2ℎ + 𝑘

𝛿𝛿2ℎ = 0

𝑥 𝛿𝛿𝑥2

𝑦 𝛿𝛿𝑦2

𝑧 𝛿𝛿𝑧2

Where: k = Permeability in the x, y, and z Directions h = Total Head

In two dimensions, the solution of the Laplace equation represents a family of
curves that describe seepage flow. This family of curves is known as a flow net.
Flow nets are presented in Section 3.2.1.1.
The Laplace equation for flow of water through soils requires the following
assumptions (FEMA 2015b):
• The soils are homogeneous and isotropic (when soil is anisotropic, a
transformation technique is used as discussed in USACE [1993]).
• The soils are saturated.
• The soil structure and water are incompressible (i.e., no change in void
space).
• Flow is laminar.
• Darcy’s Law is valid.
The Laplace equation is the mathematical basis used in seepage analyses. The
Laplace equation in conjunction with specific boundary conditions and soil
properties can be used to define seepage pathways, flow quantities, gradients, and
pressures. Figure 3-3 is a schematic illustrating the concept of the Laplace
equation.

Figure 3-3: Schematic of (a) 2D Seepage Flow beneath Concrete Dam and (b) Steady-
State Flow across an Element of Foundation Soil (USACE 1993)

3.2 Seepage Analysis Methods

As mentioned earlier, there are a variety of seepage analysis methods for
evaluating embankment dams that range from simple graphical approaches completed by
hand to more complex numerical modeling using computer programs. Most methods
incorporate Darcy’s Law and involve solving the Laplace equation. Typical seepage
analysis methods include the following:
• Graphical Methods
– Flow Nets
– Graphical Construction of Phreatic Surface
• Analogs (Physical Models)
• Numerical Models
Generally, it is best to start with the simplest and least expensive method before
proceeding to the more complex and costly method. However, modern and robust
computing capabilities have made the use of numerical modeling the current standard
of practice. This section provides a basic overview of the typical seepage analysis
methods. General guidelines for selecting an appropriate seepage analysis method
are provided at the end of this section.
3.2.1 Graphical Methods
Analyzing the seepage response of embankment dams using 2D graphical methods is the
oldest approach and includes flow nets or graphically constructing the phreatic
surface.
3.2.1.1 Flow Nets
Flow nets are a graphical solution of the Laplace equation that includes a family
of curves that describe seepage flow. A flow net is constructed by hand and
consists of two sets of orthogonal (i.e., intersecting at right angles) curves
referred to as flow lines and equipotential lines. Flow lines represent seepage
paths through the soil. The space between two flow lines is a flow channel, and
each channel represents equal quantities of flow. Equipotential lines intersect the
flow lines at right angles. Equipotential lines show the location of points within
the soil that have the same piezometric head and represent equal pressure drops
along the flow net. Figure 3-4 illustrates a flow net solution for seepage under a
sheet pile wall in a permeable foundation that is assumed to consist of a
homogeneous, isotropic soil. Figure 3-5 illustrates a flow net for unconfined
seepage through a homogeneous embankment.

Figure 3-4: Flow Net for Seepage beneath Sheet Pile Wall in Permeable Foundation
(FEMA 2015b)

Figure 3-5: Flow Net for Unconfined Seepage through Homogeneous Embankment (FEMA
2015b)

Flow nets are a practical and versatile method for evaluating seepage and have
historically been used to analyze 2D seepage problems. Flow nets are relatively
fast to create, easy to draw for simple cases, inexpensive, and provide insight
into seepage flow characteristics and quantities. However, flow nets take practice
and experience to draw accurately, require a fair amount of simplification to
geometry and material properties, are difficult to draw for complicated geometries
and multiple permeabilities, and are no longer commonly used. With practice, flow
nets can be a valuable tool for evaluating seepage in dams and can also be used to
help verify numerical solutions. As discussed later, modelers should attempt to
draw a conceptual flow net in advance of modeling, as it will help guide thinking
and aid in model set up. For more information on flow nets, refer to additional
references in Section 7 (Cedergren 1989; FEMA 2015b; NRCS 1973; NRCS 1979;
Reclamation 2014; USACE 1993).
3.2.1.2 Graphical Construction of Phreatic Surface
The upper line of seepage (i.e., flow line) through an embankment dam is known as
the phreatic surface and represents a line of zero pressure. The phreatic surface
through an embankment

can be graphically constructed using a procedure described by Casagrande (1937).

Figure 3-6 illustrates the graphical construction of a phreatic surface for a
homogeneous embankment.

Figure 3-6: Graphically Constructed Phreatic Surface through Homogenous Embankment

(Adapted from Reclamation 2014)

Graphically constructing the phreatic surface is a relatively fast, easy to follow,

and inexpensive approach for defining the phreatic surface for slope stability
analyses or estimating seepage quantities. This approach for constructing the
phreatic surface is fast, repeatable, and may also be used as a starting point for
construction of a flow net (NRCS 1979). However, it has limited applicability, less
versatility than flow nets, and is no longer commonly used. Graphical construction
of the phreatic surface is primarily limited to evaluating drainage alternatives
for homogeneous embankments on relatively impervious foundations. For more
information on graphically constructing the phreatic surface, refer to additional
references in Section 7 (Casagrande 1937; FEMA 2015b; NRCS 1979; Reclamation 2014).
3.2.2 Analogs (Physical Models)
Before numerical models used computer programs, analogs were used to evaluate
seepage flow. “Analogs” is a term for physical models that are analogous to the
flow of water through porous media. Such physical models include electrical analog
models and viscous fluid models. These models simulate the flow of seepage because
the physics are governed by the same principles as flow. Figure 3-7 illustrates the
use of an electric analog to evaluate seepage conditions beneath a concrete dam.

Figure 3-7: Use of 2D Conducting Paper to Evaluate (a) Equipotential Lines and (b)
Flow Lines (USACE 1993)

Analogs can evaluate a variety of seepage problems and indicate the reaction of a
system to a change in condition (e.g., change in head, geometry). However, analogs
are rarely used and can be time-consuming and costly to calibrate depending on the
type of model. For more information on analogs (physical models), refer to
additional references in Section 7 (USACE 1993).
3.2.3 Numerical Models
The most common approach and current standard of practice for analyzing the seepage
response of embankment dams is using 2D and 3D numerical models, provided that a
robust analysis using modern computing capabilities is warranted.
3.2.3.1 General
Numerical models use computer programs to run finite element analyses that
mathematically approximate the Laplace equation in complex flow conditions. In a
numerical model, the geometry is discretized into small (i.e., finite) elements
that form a grid. Each element intersection is called a node. The nodes represent a
continuum through the entire model. The model uses a series of equations to
approximate the Laplace equation. For example, if the grid consists of N elements,
there will be N equations and N unknowns to solve. Figure 3-8 illustrates a
numerical seepage model for a zoned embankment.

Figure 3-8: Numerical Seepage Model Results from the SEEP/W Program for Zoned
Embankment (Heitland et al. 2020)

There are many benefits to numerical modeling, which include the following:
• Numerical models can properly characterize permeability and evaluate flow
through both saturated and unsaturated soils. Characterizing unsaturated flow is a
limitation of flow nets.
• Numerical models are easier to use for complex situations (e.g., complex
embankment geometry or foundation stratigraphy).
• Both steady-state and transient (or time-dependent) flow can be modeled.
• Both 2D and 3D problems can be modeled.
• Zones where seepage gradients or velocities are high can be more accurately
modeled by varying the size of the discrete elements.
• A variety of boundary conditions can be modeled.
• Most numerical models have graphical results that can be visually checked for
reasonableness.
• Numerical models provide results (e.g., seepage flow rates, velocities,
gradients, pressures) at any location (i.e., element) within the model.
• Results can be easily used and input into slope stability computer programs.
Some limitations to numerical modeling include the following:
• Numerical models are only as good as the modeler’s understanding of the input
and ability to interpret the results.
• Modeling requires practice and training to understand the sensitivities of
the model.
• Numerical models are susceptible to convergence issues.
• Numerical models will often run without error and produce professional-
looking results that can be invalid or produce results that do not make sense. It
takes knowledge of seepage principles and experience to properly interpret and
verify the results.
• Modeling can be time-consuming and costly.
For more information on numerical models, refer to additional references in Section
7 (FEMA 2015b; GEO-SLOPE 2012; Heitland et al. 2020; Reclamation 2014; USACE 1993).
3.2.3.2 Computer Programs
Typical computer programs for modeling 2D and 3D seepage according to the Federal
Emergency Management Agency (FEMA) are summarized in Table 3-2, along with their
modeling capabilities, benefits, and limitations. The most commonly used computer
program for seepage modeling in dams is SEEP/W developed by GEO-SLOPE, as part of
their GeoStudio suite.

Seepage and Slope Stability Modeling for Embankment Dams

Table 3-2: Typical Computer Programs for Modeling Seepage (FEMA 2015b)
Computer Program
Method of Modeling
Modeling Capabilities
Benefits
Limitations
SEEP/W
(GeoStudio) 2D and finite element 3D finite element
capabilities were added in 2019 Groundwater, pore water pressures, seepage flow
quantities, velocities, gradients, and uplift pressures. • User friendly,
good quality graphics.
• Seepage pressures from SEEP/W can be imported into SLOPE/W for slope
stability analysis. • User friendly nature results in the program
frequently being misused by novice analysts, which can result in unrealistic
results.
• Only models laminar flow through homogeneous media. Concentrated flow, such
as through bedrock features, cannot be accurately modeled.
MODFLOW 2D and 3D finite element Groundwater flow in aquifers; can
evaluate well performance. • Several user interfaces available from
commercial and non-commercial sources. • Non-orthogonal anisotropies not
allowed.
FEFLOW 2D and 3D finite element Groundwater and flow through porous and
fractured media. • Can model complex geologic features. • More complex
models are more expensive to run, require greater expertise, and take longer to
learn how to use.
FRACMAN 3D finite element Flow through fracture networks in bedrock.
• Requires detailed geologic information inputs.
FRAC 1D discrete boundary Matrix and fracture flow through porous media and
fractured rock. • Works with MODFLOW to model flow through fractured rock.
• Requires detailed geologic information inputs.

In some situations, there may be merit for dam owners or regulators to develop a
simplified seepage model to assist in making decisions. While seepage models can be
valuable in understanding the performance of an embankment dam, the cost of
purchasing a computer program can rarely be financially justified for owners and
regulators. A new affordable option that is available is GEO-SLOPE’s Basic SEEP/W,
which is a trimmed down version of SEEP/W that is well suited for owner and
regulatory needs. Limitations of the basic version include a coarse finite element
grid and the inability to model transient flow. However, owners and regulators are
not generally going to perform a rapid drawdown analysis or toe drain design.
Rather, the owner or regulator is likely attempting to understand the embankment
and foundation, general flow patterns, missing information, and most importantly,
whether a more refined seepage model may be warranted. The complex seepage modeling
is often best left to the consulting engineer.
3.2.4 General Guidelines for Selecting a Seepage Analysis Method
Deciding which seepage analysis method to use for the evaluation of an embankment
dam is based on the objective of the analysis and complexity of the situation. Some
methods will be appropriate for some situations, while others will not. General
guidelines for selecting an appropriate seepage analysis method are summarized in
Table 3-3.
Table 3-3: Guidelines for Selecting a Seepage Analysis Method (Adapted from FEMA
2015b)

Situation Typical Investigations Suggested Analysis Methods

Homogeneous embankment, impervious foundation, 2D steady- state Phreatic surface,
pore water pressures, seepage forces (stability), seepage control alternatives.
• Flow Net
• Graphical Construction of Phreatic Surface
• Numerical Model
Zoned embankment, impervious foundation, 2D steady-state Phreatic surface, pore
water pressures, seepage forces (stability), seepage control alternatives. •
Flow Net
• Numerical Model
Homogeneous embankment, uniform pervious foundation, 2D steady-state Phreatic
surface, pore water pressures, seepage forces (stability). • Flow Net
• Numerical Model
Seepage flow quantities, exit gradient, material property variations, seepage
control alternatives. • Numerical Model
Zoned embankment, pervious foundation, 2D steady-state Same as above. •
Numerical Model
Uniform pervious abutment, 3D steady-state Phreatic surface, seepage flow
quantities. • Plan View Flow Net1
• Numerical Model
Heterogeneous pervious foundation and abutments, 3D steady-state Phreatic surface,
seepage flow quantities, exit gradient, material property variations, seepage
control alternatives. • Numerical Model
2D transient flow, steady boundary conditions Tracking saturation, time to
steady- state. • Transient Flow Net2
• Numerical Model
Non-steady 2D flow, saturated/unsaturated, zone or homogeneous foundation,
transient boundary conditions, 2D transient state First fill, flood cycle,
cyclic operation, moisture content and pore pressure changes, effects of
precipitation and evaporation. • Numerical Model

1 Plan view flow net described in Section 3.4.1.3.

2 Transient flow net is described in Cedergren (1989).

3.3 When Is a Seepage Analysis Warranted?

A seepage analysis of an embankment dam can be used to gain a holistic
understanding of the seepage regime through the embankment and its foundation.
Seepage modeling can also provide valuable insight into how seepage may influence
the performance of the embankment dam. However, seepage modeling can be time-
consuming and costly and may not provide any additional insight into seepage-
related issues. Key considerations in evaluating the cost- effectiveness and
necessity for a numeric seepage model include the objective of the analysis, the
amount of available information, and the complexity of the seepage regime. To
identify when a seepage analysis may be warranted, consider the following
questions:
• What is the objective of the seepage analysis?
• Can the objective be met adequately without a seepage analysis?
• Will a seepage analysis accurately capture the objective?
• Is there time to complete a seepage analysis (not an emergency)?
This section provides guidance in answering these questions to evaluate whether a
seepage analysis may be warranted.
3.3.1 What Is the Objective of the Seepage Analysis?
This important question must be answered in detail before initiating a seepage
analysis so that the analysis method and approach can be tailored to the specific
objective. The objective of the seepage analysis will ultimately inform whether a
model is warranted. The objective should be clear, well-defined, and consider the
intended audience for which the objective will be explained.
When establishing the objective of a seepage analysis, it is important to have a
basic understanding of instances when a numerical seepage model may or may not be
warranted for an embankment dam. A few examples that may or may not warrant a
seepage model are summarized in Table 3-4.
Table 3-4: Examples That May or May Not Warrant a Numerical Seepage Model (Heitland
et al. 2020)
Examples That May Warrant a Seepage Model Examples That May Not Warrant a Seepage
Model
Evaluating an embankment under a future reservoir operating or flood loading
condition it has not yet experienced. Site with limited information on the
embankment zoning, foundation conditions (including geology and stratigraphy), and
material properties.
Evaluating an embankment with a complex geometry or foundation stratigraphy and/or
a site with complex geology. Well-instrumented site (e.g., piezometers) for which
current data can be used to evaluate seepage response rather than a model.
Evaluating a non-hydrostatic condition (i.e., total head is not constant with
depth). A site that is not sensitive to seepage performance, which can be
demonstrated using conservative assumptions.
Site configuration that limits accuracy of 2D simplified assumptions. Sensitivity
analyses (e.g., slope stability) can be performed for a potential range of pore
water pressures to compensate for uncertainty.

When describing the need for a seepage model, the description must be written for
both a technical and non-technical audience, limiting undefined jargon. Depending
on the non-technical audience, it may be necessary to supplement the objective
description with hand drawings or more detailed information. The importance of this
cannot be understated. The objective must be clearly defined for a potentially wide
audience, including inexperienced dam owners, regulators, elected officials, and
board/commission members, as well as peers and colleagues.
As an example, consider the following objective: “A seepage model will help with
the design of a toe drain.” This is a simply stated objective with no clear
explanation of why or how the seepage model will help with the design of the toe
drain. Instead, the objective should be elaborated as follows: “A seepage model is
needed to understand the seepage pressures and gradients in the alluvial sand and
gravel foundation layer and verify this layer is the likely primary water bearing
layer. The model will be used to understand how these pressures and gradients
change with distance from the reservoir and the potential for internal erosion of
the foundation materials in this layer to initiate. It is suspected that the deeper
bedrock is low permeability (tight) and not transmitting significant water; the
model will help confirm this assumption. The model will also be used to confirm
that a toe drain will be effective in intercepting flow through the water bearing
layer and provide guidance as to where the drain should be located and its
approximate depth. The model will be useful in evaluating performance of the
proposed drain during a flood event that raises the reservoir level 3 feet.”
3.3.2 Can the Objective Be Met Adequately Without a Seepage Analysis?
In some cases, the specified objective can be met without a seepage analysis if the
embankment dam has adequate geotechnical information (including boring logs) and is
well instrumented with piezometers in the embankment and foundation. Key factors in
making this determination include the following:
• Piezometers are properly located and isolated in zones or layers of interest.
Piezometric measurements can often be used to evaluate the seepage response and
phreatic surface in an embankment dam rather than a model.
• The reservoir level does not fluctuate significantly on an annual basis such
that steady- state flow conditions can be assumed. It is more difficult to estimate
seepage response from measured data if transient flow conditions exist.
• There is no need to evaluate the embankment dam under a future operating
condition (e.g., higher reservoir pool level due to dam raise) or extreme loading
condition (e.g., flood pool level). If there are questions about the embankment
performance under future operating or extreme loading conditions, a model will be
necessary.
• The dam owner has a limited budget. This is especially important if there is
concern that a seepage model may not produce useful results, or if there is a known
issue that may make calibration challenging. In general, dam owners prefer to spend
their money on tangible items—something they can defend to their board or
commission.
While the objective may not always warrant a seepage analysis, using a numerical
seepage model to develop embankment cross sections and plot the piezometer
locations and measured water levels can be valuable in understanding the embankment
geometry, internal zoning, foundation contact and stratigraphy, and seepage regime,
even if the model is never run. Hand drawing a rough flow net or drawing a water
level contour map (in the case of a 3D model) is highly recommended, as it helps to
conceptualize flow through the system and is important for efficient calibration.
Developing a conceptual model representing the hydrological and hydrogeological
conditions of the dam site before the actual modeling begins provides

opportunity to evaluate the effort that would be required and expected outcomes
from the model.
3.3.3 Will a Seepage Analysis Accurately Capture the Objective?
Seepage is sensitive to small localized variations of permeability, defects,
anomalies, and fissures that are difficult to identify and model with precision,
such that a high level of accuracy can be difficult to achieve in a seepage
analysis. A seepage analysis should be considered to give an order of magnitude
level of accuracy that is dependent on the estimated permeabilities for the
embankment and foundation materials. Therefore, sufficient data on the embankment
geometry, internal zoning, foundation contact and stratigraphy, and material
properties is warranted for developing effective seepage models. In some cases,
additional site exploration and investigation may be required to obtain the
necessary data to perform a seepage analysis. Further information on the minimum
data requirements for performing a seepage analysis is presented in Section 3.4.2.
A couple of specific conditions that are difficult to accurately capture in a
seepage analysis include embankment defects (e.g., cracks or construction flaws)
and geologic defects (e.g., rock discontinuities), which are often the root cause
of seepage issues. For these conditions, direct observation and monitoring are the
best methods of evaluation.
3.3.4 Is There Time to Complete a Seepage Analysis (Not an Emergency)?
A sudden, unexpected failure of an embankment dam due to seepage is unlikely to
occur if the following conditions are true (FEMA 2015b):
• The embankment has been properly designed and constructed to the current
standard of practice.
• The embankment has been maintained properly.
• Inspections are routinely performed by qualified personnel.
• An adequate amount of instrumentation is installed, monitored, and evaluated
on a timely basis.
• Dam safety repairs are made as conditions dictate.
However, there are instances where these conditions may not hold true, and sudden
changes in seepage conditions may require emergency action rather than an analysis
of the seepage conditions. Signs of seepage conditions that may indicate imminent
danger include the following:
• A whirlpool in the reservoir, particularly above the upstream embankment
slope.
• Sudden or increased cloudy or muddy seepage.
• Frequent sand boils.
• Sudden sloughs on the downstream embankment slope.
• Sinkholes on the embankment.
• Cloudy discharge observed adjacent to an embedded structure (e.g., outlet
conduit, spillway wall).
• Abrupt changes in piezometric water levels or seepage flow rates.

If any of these conditions are observed, a seepage analysis may not be warranted,
and corrective action should be taken quickly under the guidance of an experienced
dam safety engineer. The type of corrective action will vary for each of the above
conditions and should be outlined in the dam’s Emergency Action Plan (EAP). An EAP
is a formal document that identifies potential emergency conditions at a dam and
specifies actions to be followed by the dam owner to minimize loss of life and
property damage. For more information on EAPs, refer to additional references in
Section 7 (FEMA 2013).
If it is determined that a seepage analysis is warranted, then the following
sections should be consulted for guidance on planning a seepage analysis, as well
as interpreting, verifying, and reporting the results.
3.4 Planning for a Seepage Analysis
While seepage modeling can be beneficial in understanding the design and
performance of embankment dams, it should also be understood that all models are
wrong in some context. Models are merely an oversimplification or interpretation of
reality and should be treated as such. A seepage model is only as good as the
modeler’s understanding of the inputs, how the inputs are used in the model, and
the modeler’s ability to interpret the results. Thus, modelers must be fluent in
the model inputs, outputs, and seepage theory and recognize the model sensitivities
to get the most out of a seepage model.
This section provides tips, tools, and guidance on planning for a seepage analysis
and focuses specifically on numerical modeling using computer programs, which is
the most commonly used method for evaluating the seepage response of embankment
dams. Other seepage analysis methods are introduced in Section 3.2, with references
to additional publications for further information on these methods.
Planning for a numerical seepage analysis involves defining the modeling approach
and minimum data requirements. The planning process should also include
considerations for model setup, calibration, and convergence. Guidance is provided
below for each of these topics.
Guidance for the interpretation, verification, and reporting of results is provided
in Section 3.5.
3.4.1 Modeling Approach
The first step in performing a numerical seepage analysis involves defining the
modeling approach, which includes considerations for the following:
• Geometry,
• Steady-state versus transient seepage analysis,
• 2D plane strain versus 3D modeling, and
• Saturated only versus saturated/unsaturated material model.
3.4.1.1 Geometry
The geometry of both the dam site and embankment should be taken into consideration
in seepage modeling. In a 2D analysis, cross sections should be selected at
locations where critical seepage conditions are expected and seepage results are
required. This will typically include the maximum embankment section at a minimum.
In a 3D analysis, an entire dam and surrounding area can be included in the model.

3.4.1.2 Steady-State versus Transient Analysis

Seepage through an embankment dam can be analyzed under steady-state or transient
flow conditions. A steady-state seepage analysis represents the long-term operating
condition of a dam. In a steady-state model, internal pore water pressures and flow
conditions are computed for a given set of boundary conditions and are assumed to
be steady (i.e., unchanging). This condition is typically evaluated with the
reservoir at the normal operating pool level and is the most commonly analyzed
condition.
A transient seepage analysis is one that is always changing. In a transient model,
both initial and future boundary conditions must be specified to evaluate how long
it takes for the embankment materials to respond to the given set of boundary
conditions. Typical transient analysis scenarios include evaluating the wetting
front rate through an embankment during the first reservoir fill after
construction, the maximum reservoir drawdown rate to meet stability or other
requirements of the applicable regulatory agency, the annual pore water pressure
regime through an embankment that experiences yearly reservoir fluctuations, and/or
the effect of flood loading (i.e., how long it will take to saturate the embankment
and reach a steady-state condition). An example transient analysis is presented in
Figure 3-9, which illustrates a rapid drawdown scenario though a zoned embankment
dam. A transient seepage analysis may be completed in conjunction with a slope
stability analysis to evaluate safe drawdown rates.

Figure 3-9: Transient Seepage Analysis for Drawdown through Zoned Embankment
(Reclamation 2014)

Irrigation reservoirs that experience fluctuations in water levels annually may not
be well represented with a steady-state seepage model. Typically, there is a lag
between changes in

reservoir water level and corresponding responses in piezometers. When modeling

irrigation reservoirs, look carefully at lags in piezometric responses as the
reservoir fills, reaches full pool, and starts to draw down. If a significant lag
is obvious, it may be necessary to perform a transient seepage analysis to obtain
meaningful results.
For more information on steady-state and transient seepage analyses, refer to
additional references in Section 7 (Reclamation 2014).
3.4.1.3 Cross Section versus Plan View Models
Seepage analysis for embankment dams is most commonly modeled using 2D cross
sections. However, in some cases a seepage analysis may be modeled in plan view. A
plan view seepage model evaluates 2D groundwater flow using the same principles
described in Section 3.2.1.1, Flow Nets. Although not commonly used, plan view
seepage models have been used to evaluate groundwater flow around abutments or
cutoff walls. More commonly, plan view seepage analyses are used in the selection
and design of dewatering systems or relief wells.
A plan view analysis is intended to model groundwater flow through confined
aquifers, so application to unconfined problems must be conducted with caution. For
more information on plan view models, refer to additional references in Section 7
(GEO-SLOPE 2012; NRCS 1979).
3.4.1.4 2D versus 3D Modeling
Historically, seepage analyses have primarily comprised 2D modeling because of the
complexity of 3D modeling and limited computer programs capable of 3D modeling.
However, seepage analyses have recently been expanding into the 3D realm as the
benefits of 3D modeling are becoming clearer and computer programs with 3D
capability are becoming more prevalent.
While 3D seepage modeling can be valuable under certain conditions, 3D modeling is
much more rigorous than 2D and therefore requires careful consideration to ensure
3D modeling is the appropriate approach. 3D modeling is very time consuming,
requires a much greater level of effort and expertise, and is more costly.
Conditions in which 3D seepage modeling becomes beneficial include the following:
• There are significant 3D cross-valley effects along the embankment dam
alignment (e.g., narrow “v”-shaped valley profile, irregular/uneven or sloping
foundation or abutment surface, pervious foundation or abutment) – Figure 3-10(a).
• A relatively long dam (or levee) that has a convex bend in the embankment.
The bend serves as a point where seepage may converge from multiple directions –
Figure 3-10(b).
• Complex embankment geometry (e.g., discrete seepage control features such as
filters/drains, toe drain, relief wells) and/or foundation geology (e.g., bedrock
discontinuities, faults) – Figure 3-10(c).

Figure 3-10: Examples of When 3D Seepage Modeling Becomes Beneficial: (a) Irregular
Bedrock Foundation; (b) Convex Bend in Dam Alignment; (c) Complex Model Geometry
(Heitland et al. 2020)

In all cases, there must be adequate information available to justify the expense
of a 3D model. If lacking boundary condition input or geotechnical/geological data,
a 3D model may not be warranted. It is recommended that any 3D model be informed by
2D models.
3.4.1.5 Saturated Only versus Saturated/Unsaturated Material Model
After the seepage analysis type (i.e., steady-state or transient and 2D or 3D) is
established, the type of material model (i.e., saturated only or
saturated/unsaturated) will need to be specified. A saturated/unsaturated material
model should be selected to properly characterize permeability and evaluate flow
through both saturated and unsaturated soils. This material model requires
specifying hydraulic conductivity (permeability) and volumetric water content
functions.
The hydraulic conductivity function describes the ability of the soil to transport
water under both saturated and unsaturated conditions. The volumetric water content
function describes the portion or volume of the pore spaces within a soil that
remains water-filled as the soil drains. In a saturated soil, all the pore spaces
between the soil particles are filled with water. In an unsaturated soil, the pore
spaces are filled with air, becoming non-conductive conduits to flow. When the pore
spaces become air-filled as the soil drains, the pore water pressures decrease
rapidly and become increasingly more negative, which in turn quickly reduces the
permeability of the soil. Thus, permeability becomes a function of negative pore
water pressure when a soil is unsaturated, and this is captured by defining the
hydraulic conductivity and volumetric water

content functions. Error! Reference source not found. illustrate the relationship
between the soil water content and hydraulic conductivity. For details on how to
define and build the hydraulic conductivity and volumetric water content functions,
refer to additional references in Section 7 (GEO-SLOPE 2012; Fredlund 1998;
Reclamation 2014).

Figure 3-11: Typical Volumetric Water Content Functions of Soils (Reclamation 2014)

Figure 3-12: Example Saturated / Unsaturated Hydraulic Conductivity Functions of

Embankment Soil (Reclamation 2014)
In most cases, a saturated/unsaturated material model should be specified because
embankments typically have both saturated and unsaturated zones below and above the
anticipated phreatic surface. This material model should also be selected if there
is any uncertainty in whether specific materials comprising the embankment and/or
foundation will

remain unsaturated. A saturated-only material model should only be used when it is

known that the materials will remain saturated under the given set of boundary
conditions. For a coarse- mesh, simplified model, a saturated-only analysis could
be used as an initial estimate of pore water pressures. However, a
saturated/unsaturated analysis is recommended for almost all engineering design
analyses.
3.4.2 Minimum Data Requirements
Seepage models can be adjusted based on very little data until the results look
like what the modeler hoped they would, but “garbage-in, garbage-out” makes those
models a waste of time. The minimum data needed to develop an efficient and
reliable seepage model for an embankment dam are summarized in Table 3-5, along
with the purpose of the data and types of reference documents where the data can be
obtained. These minimum data requirements are discussed further below.
Table 3-5: Minimum Data Requirements for Seepage Modeling

Data Category Data Requirements Typical Data Sources

Model Geometry • Embankment Geometry and Internal Zoning.
• Foundation Contact and Stratigraphy. • Topographic or Lidar Surveys
• As-Built Construction Drawings
• Design Drawings
• Design or Construction Reports
• Geologic and Geotechnical Investigation Reports
Material Properties • Permeabilities for Embankment and Foundation
Materials. • Geotechnical Investigation and Data Reports
• Construction Reports
• Published Data
Calibration Data (for Existing Dams) • Reservoir Levels, Piezometer Data,
and Weir Flow Data. • Instrumentation Records
• Inspection Reports
• Geotechnical Investigation and Data Reports

3.4.2.1 Model Geometry

Seepage models are typically developed for one or more embankment cross sections
along the dam alignment. To develop the model geometry, there must be sufficient
available data on the embankment geometry and internal zoning, as well as the
foundation contact and stratigraphy. A detailed model geometry will delineate the
various embankment zones (core, shells, filters, drains, etc.), foundation layers,
and any other seepage control systems (toe drains, low permeability blankets,
cutoff walls, etc.).
The best data source for defining the embankment geometry and internal zoning of an
existing embankment dam is typically as-built construction drawings. Ideally, the
external geometry (i.e., embankment crest, upstream and downstream slopes, and
downstream ground surface) should be defined by a recent topographic or lidar
survey. When construction drawings and recent surveys are not available, design
drawings and/or design or construction reports can be used for defining the model.
However, the modeler should be aware that the as-built and/or current embankment
condition may differ from the design condition. Information from geologic and
geotechnical investigation reports can also be valuable in verifying the internal
zoning and/or variations in the embankment materials. For a new embankment dam that
has yet to be

constructed, the embankment geometry and internal zoning is typically defined using
design drawings, but the design can be adjusted based on the results of the slope
stability analysis.
The best data sources for defining the foundation contact and stratigraphy are
typically geologic and geotechnical investigation reports.
3.4.2.2 Material Properties
Seepage modeling requires assigning material properties to the embankment and
foundation materials. These material properties include permeability and the
resulting anisotropy ratio (i.e., the ratio of horizontal to vertical
permeability). Similar to the model geometry, there must be sufficient available
data on the embankment and foundation materials to estimate the material
properties. Typically, seepage properties are estimated using data collected from
geotechnical investigations and/or published data. Information on the materials may
also be available in construction reports.
Permeability values can be estimated from laboratory tests (e.g., constant head,
falling head, or flexible wall permeameter), field tests (e.g., borehole soil
permeability tests or rock packer tests), published tables of values, and/or
empirical equations, which most often relate permeability to material gradation and
void ratio (Cedergren 1989; NRCS 2009; Reclamation 2014). Laboratory and field
tests represent the most reliable estimates of permeability.
Published tables of values and empirical equations should be used with caution in
regard to the accuracy of the estimated permeability values and should consider the
specific material for which the empirical correlation is applicable. Most empirical
correlations are applicable to granular materials and become unrepresentative for
materials with high fines contents. Seepage models using material properties based
only on published tables of values or empirical correlations should consider a
range of potential permeabilities assigned to the various materials by performing
sensitivity analyses to represent uncertainty. At a minimum, estimating reasonable
seepage material properties requires adequate geotechnical/geological data on the
embankment and foundation materials, including:
• Soil or rock type,
• Gradations,
• Density,
• Stratification/discontinuities, and
• Compaction procedures.
For more information on seepage material properties and laboratory and field
permeability tests, refer to additional references in Section 7 (Reclamation 2014;
USACE 1993).
For saturated/unsaturated material models, hydraulic conductivity and volumetric
water content functions need to be specified. Depending on the level of accuracy
required in seepage modeling, specific functions can be developed for the materials
using soil-water retention properties. These properties can be measured in a
laboratory or in the field (e.g., advanced tensiometers). However, measuring soil-
water retention properties in the laboratory or field can be expensive. Soil-water
retention properties can be estimated from grain-size distributions and compared to
a database of existing test results (Fredlund 1998). For a screening-level
solution, typical functions available in computer programs (e.g., SEEP/W) or
published data can be used. In this case, sensitivity analyses on the functions
should be performed to understand the effect of their variation and represent the
uncertainty due to a lack of specific test data.

3.4.2.3 Calibration Data

For existing dams, seepage models should be calibrated if possible. Developing a
plan to calibrate the model is an important step in the data review process.
Calibrating a seepage model requires adequate records on reservoir levels,
piezometer data, and/or weir flow data. Reservoir levels in conjunction with
piezometer data can be used to calibrate the phreatic surface or potentiometric
surfaces modeled within an embankment. Weir flow data can be used to compare actual
measured seepage flow rates with estimated seepage flow rates. Caution should be
used when relying on weir flow data to calibrate the seepage model, as weirs can
collect water from other groundwater sources, or seepage may not be collected by
the weir.
Calibration data including reservoir levels, piezometer data, and weir flow data
can typically be found in instrumentation records and/or inspection reports.
Geotechnical investigation reports can also be reviewed to evaluate if groundwater
was encountered during test hole explorations. The water levels measured in test
holes can also be used to supplement piezometer data for calibrating the modeled
phreatic surface.
3.4.2.4 Desktop Review
A desktop review of all available information for an embankment dam should be
completed to inform whether there is sufficient data on the embankment geometry,
internal zoning, foundation contact and stratigraphy, and material properties to
perform the seepage analysis. The desktop review can also identify potential data
gaps. FEMA (2015b) presents the following checklist for guidance on conducting a
desktop review:
Other questions to consider when evaluating whether there is sufficient data to
develop a seepage model include the following:
• Are the embankment geometry, internal zoning, and foundation contact and
stratigraphy understood and characterized well enough to develop a representative
model? There must be adequate geotechnical/geological data to understand the
embankment and foundation materials, as well as their material properties and
spatial variability.
• Are construction documents available that define compaction procedures, which
can also influence material properties (specifically anisotropy ratio)?
• Are there quality reservoir level measurements? Verify that the dam tender is
taking accurate measurements. There must be consistent and plentiful data from a
variety of reservoir levels.

• If there are existing piezometers, are they properly isolated within zones or
layers of interest? If a piezometer’s influence zone (screened interval plus filter
pack) spans more than one zone or layer, it may be of limited use for model
calibration.
• Is there hydraulic information on the downstream conditions (i.e., downstream
piezometer measurements or tailwater levels) to establish downstream boundary
conditions?
• Are there unusual flow patterns from abutments or other issues that could be
obstacles in calibrating a model in a reasonable time period?
• Will there be too much guessing? The type of information needed will depend
on the objective of the seepage analysis. For example, if the objective is to
understand the seepage regime through an embankment during spring filling and only
sporadic reservoir level measurements are available, it may be best to
conscientiously collect a full year of data before attempting a seepage model.
The following is an example of how a modeler may plan for an analysis by performing
a data review and comparing the available data to the data required for meeting the
seepage analysis objective:
Problem – After a recent flood event, shallow sloughing was observed on the
downstream slope of a homogeneous embankment. It was decided that a slope stability
analysis of the embankment should be performed. The embankment includes a
downstream toe drain.
Objective – Complete a seepage analysis to evaluate the phreatic surface for use in
a slope stability analysis. The stability analysis will inform the risk of dam
failure by slope instability.
Data Requirements – The seepage model will require an accurate 2D cross section to
be developed, for which the surface topography, embankment zoning, and foundation
boundary need to be understood. The location and details of the toe drain are also
significant to the model development. In addition to model geometry considerations,
the material properties (specifically permeability) of the embankment and toe drain
materials are needed. The foundation is considered to be impervious. Finally,
reservoir operations and instrumentation monitoring data will be used to develop
boundary conditions and calibrate the model.
Results of Desktop Review:
 Present-condition and preconstruction topographic surveys of the site and
construction drawings were identified and will be used to develop the model
geometry (including defining the foundation boundary and toe drain).
 An operations and maintenance (O&M) manual, instrumentation monitoring
records, and inspection report were also identified that include reservoir
operations (normal operating pool level, flood pool level, etc.), as well as
seepage weir data and the approximate location of the slough. This information will
be used to develop boundary conditions and calibrate the model. Boundary conditions
and model calibration are discussed further in Sections 3.4.3.2 and 3.4.4,
respectively.
A geotechnical data report with boring logs and material properties was not
identified in the desktop review, so a data gap exists for the seepage properties.
Permeability values could be reasonably estimated using the seepage weir data and
location of the slough to calibrate the model. However, a geotechnical

investigation may be warranted to obtain material properties for the slope

stability analysis. Therefore, the modeler must decide if additional exploration is
warranted prior to developing the seepage model.
For information on performing geotechnical investigations in embankments, including
typical drilling and sampling methods, refer to additional references in Section 7
(AECOM 2014a; FEMA 2015b).
3.4.3 Model Setup Considerations
Three critical considerations for setting up a seepage model include cross section
orientation, boundary conditions, and finite element mesh size and are discussed
further below.
3.4.3.1 Cross Section Orientation
When performing a 2D seepage analysis, the cross section orientation must be
selected carefully to limit 3D effects. A 2D cross section assumes all seepage flow
is parallel to the section. Provided below are tips for consideration.
• Tip 1 – Orient the cross section parallel to the anticipated direction of
seepage flow, as shown by the “correct” cross section in Figure 3-13(a). Figure 3-
13 shows total head contours in plan view along a dam alignment. The “correct”
cross section is oriented perpendicular to the total head contours, so that flow
remains parallel to the section. The “incorrect” cross section intersects the total
head contours at an angle such that seepage would be flowing out of the model in a
third dimension. If flow is oblique to the section, results will not be a good
match to reality.
• Tip 2 – Consider a bent section for cases where seepage flow may not be
perpendicular to the dam alignment, as shown in Figure 3-13(b). The bent section
remains perpendicular to the total head contours.
• Tip 3 – If calibrating the model using piezometer data, it would be prudent
to select a cross section at or near piezometers.

Figure 3-13: Cross Section Orientation for Numerical Seepage Modeling (Heitland et
al. 2020)

3.4.3.2 Boundary Conditions

Boundary conditions must be carefully defined along the exterior boundaries of the
seepage model, as the results are a direct response to the boundary conditions.
Boundary conditions define where flow enters and exits the model, the upstream and
downstream water levels, and even the direction of flow. Table 3-6 summarizes the
typical boundary conditions assigned to a seepage model. These boundary conditions
are also shown in Figure 3-14. Provided below are tips for consideration.
• Tip 1 – Input relatively long seepage models to avoid boundary effects at the
vertical edges of the model, which can result in incorrect estimates of total head.
The rule of thumb is to extend the model both upstream and downstream by a length
equivalent to at least the width of the embankment section from the upstream to
downstream toe.
• Tip 2 – Consider potential impacts of a no-flow boundary condition at the
base of the model for an embankment founded on deep pervious materials. If the base
of the model is too shallow with a no-flow boundary condition assigned, it can
result in inaccurate estimates of total flow through the pervious foundation. For
seepage models with deep, pervious foundations, the rule of thumb is to extend the
base of the model below the base of the embankment to a depth between one and two
times the height of the embankment.
• Tip 3 – In addition to the exterior model boundaries, boundary conditions (or
nodes) can be applied to features within the model. For example, a zero-pressure
node can be assigned to a discrete toe drain so that the phreatic surface
intersects the toe drain.
Table 3-6: Typical Seepage Model Boundary Conditions (Heitland et al. 2020)
Model Boundary Boundary Condition
Upstream Vertical Edge of Model
Upstream Ground Surface Below Reservoir Level Upstream Embankment Slope •
Total head boundary equivalent to the elevation of the reservoir level
Embankment Crest
Downstream Embankment Slope Downstream Ground Surface • Potential seepage face
boundary (i.e., where seepage can exit the model)
Downstream Vertical Edge of Model • Total head boundary equivalent to either
the elevation of the tailwater level (if applicable) or the elevation of the
natural groundwater level
• If there is no tailwater or no downstream piezometer to measure the
groundwater level, the total head boundary is often assumed to be equivalent to a
groundwater level at the downstream ground surface elevation. If the model results
indicate this assumption is influencing the seepage patterns at the dam, then the
extent of the model domain should be extended farther downstream.
Base of Model • No-flow boundary

Figure 3-14: Typical Seepage Model Boundary Conditions from the SEEP/W Program
(Heitland et al. 2020)

Information to support boundary conditions can typically be found in O&M manuals

(e.g., annual reservoir fluctuations, normal operating pool level, flood pool
level), instrumentation monitoring reports (e.g., piezometer and seepage weir
data), or inspection reports.
3.4.3.3 Finite Element Mesh Size
The mesh size is dependent on the specific seepage model. Typically, a larger mesh
size is specified for a large model geometry and a smaller mesh size is specified
for a small model geometry. A common mistake is thinking that the more finite
elements there are, the more accurate the results will be. Assigning a very small
mesh size to a large model geometry can cause the model to run unreasonably slow,
making it difficult to obtain results. Provided below are tips for consideration.
• Tip 1 – Start with a large mesh size (fewer elements) and verify that the
results make sense. Once verified, reduce the mesh size as appropriate to better
define the seepage regime. This will require several iterations to understand the
model sensitivity to mesh size. Figure 3-15 and Figure 3-16 illustrate a comparison
between two different mesh sizes, with Figure 3-15 showing model results using a
10-foot mesh size and Figure 3-16 showing results using a 5-foot mesh size. The
larger mesh size results in an irregular phreatic surface near the embankment core
contact with the downstream shell, as depicted by the red circle in Figure 3-15.
When the mesh size is reduced, as shown in Figure 3-16, the phreatic surface
smooths out and becomes more reasonable.
• Tip 2 – If working with a large, complex model in which a smaller mesh size
may not be reasonable, consider reducing the mesh size at discrete seepage control
features (e.g., filters/drains, toe drains, relief wells), or other specific
regions of interest, to better capture the trends surrounding these narrow/smaller
features, which typically have higher permeability values.

Figure 3-15: Seepage Model Results from the SEEP/W Program Using 10-foot Mesh Size
(Approximately 1,000 Elements) (Heitland et al. 2020)

Figure 3-16: Seepage Model Results from the SEEP/W Program Using 5-foot Mesh Size
(Approximately 4,000 Elements) (Heitland et al. 2020)

3.4.4 Model Calibration Considerations

Once the model setup is complete, it is important to calibrate the model based on
past performance of the dam and piezometer data (if available). Model calibration
is an iterative process performed by changing model parameters to match the results
with observed or measured conditions. Most commonly, material properties are used
as the calibration parameters. Therefore, permeability and anisotropy ratio become
the main calibration parameters. In saturated/unsaturated analyses, unsaturated
parameters can also be used for calibration. Provided below are tips for
consideration.
• Tip 1 – Start with the most representative values of permeability and
anisotropy ratio for the materials and perform the simulation. If the results
(phreatic surface) deviate from the observed data, change the permeability of the
material in the area where the deviation is significant.
• Tip 2 – Change one parameter at a time.

For most cases, the phreatic surface from the model results is in good agreement
with measured results when the resulting phreatic surface is within a few feet of
the measured results. However, the final decision on whether the variance between
the model results and measured conditions is considered acceptable is site specific
and dependent on how accurate the model results need to be for the intended
purpose. The accuracy of the calibration will be reflective of the accuracy of any
subsequent result using the model.
If the past performance and/or piezometer data are unavailable (e.g., if the dam is
new), sensitivity analyses can be performed by using higher and lower input
parameters to understand the effect of their variation on the model results.
It is important to acknowledge that calibration can be the most time-consuming and
costly part of modeling. Thus, it is critical to have a plan in place and
communicate with the dam owner and/or regulator before modeling begins. The plan
should consider the following:
• Description of how the calibration will proceed (i.e., the piezometer data
and associated reservoir levels that will be applied to the model).
• Description of what will be considered reasonable calibration and the amount
of time (i.e., cost) it could take to achieve reasonable calibration.
• When efforts to calibrate a model should cease if a reasonable model is not
obtained.
• The parameters that will be calibrated.
• How hydraulic conductivity and volumetric water content functions will be
estimated.
• Potential problems that could make calibration difficult, such as flow from
an abutment.
• Method to keep track of calibration runs.
• The possibility that the model cannot be calibrated. It is important that
this be explained to the dam owner or regulator. It is worthwhile to also note the
benefits of the model even if it cannot be calibrated.
• The possibility of using sensitivity analyses for situations where
calibration is not possible (e.g., new dam construction).
3.4.5 Model Convergence Considerations
Model convergence issues occur when the model cannot obtain a solution for one or
more elements within the seepage model during the simulation. The residual error of
the solution is higher than the specified value in a non-converged model.
Convergence issues are generally experienced when the model geometry and boundary
conditions are complex and the soil property functions are highly non-linear. An
important consideration to overcome convergence issues is to start with a simple
model. There is no one solution that fits all situations of non- convergence.
However, the following tips may be helpful:
• Tip 1 – Simplify the model if it makes sense to do so.
• Tip 2 – Increase or decrease the mesh size.
• Tip 3 – Increase the number of simulation iterations.
• Tip 4 – Decrease the non-linearity of the material property functions (e.g.,
hydraulic conductivity function) if it makes sense to do so.

3.5 Interpreting, Verifying, and Reporting Results

Seepage modeling is complex and requires practice and experience. Models will often
run without error and produce professional-looking results but can be invalid.
Successful runs should not be confused with accurate results. Interpreting and
verifying the results are not always intuitive to the novice user. Vetting errors
and understanding the sensitivity of a model to the potential range of each input
parameter should be a priority before using the results. This often requires
knowledge gained through personal trial and error experience. Consider seeking
guidance from experienced engineers and modelers. If not available in-house,
program vendors and developers often offer limited technical support for vetting
specific problems.
Specific situations may warrant hiring an expert to perform the seepage modeling.
It is easier to create models for steady-state conditions than for transient
conditions, but most reservoirs fluctuate during normal operation, and piezometric
measurements often lag behind actual reservoir levels. Thus, a transient seepage
model may be needed. This adds an additional level of complexity to both developing
and calibrating a model. Another factor to consider is the time and expense to
learn or relearn the modeling program if it is not used frequently. In these cases,
consider answering the question of when a seepage analysis is warranted, but then
contract with an expert to perform the modeling and interpret and verify the
results.
There may also be merit to getting a secondary review from an expert, as some
regulatory agencies may not have the modeling experience to catch problems.
The following sections provide tips, tools, and guidance on how to interpret and
verify seepage analysis results, as well as report the results.
3.5.1 Interpreting and Verifying Results
After the seepage model is run and results are obtained, it is important to examine
the various output features, as these are visual tools that help users understand
and verify the model results. Table 3-7 summarizes the key output features
(specific to the SEEP/W computer program) and associated tips for checking the
validity of the model results. Although the table is specific to the SEEP/W
computer program, the output features and concepts are similar among seepage
modeling computer programs.
Table 3-7: Key Seepage Model Output Features for Checking the Validity of Results
(Heitland et al. 2020)

Output Feature Description Tips/Considerations

Total Head Contours (1) • Depict where total head values (i.e., pressure head
plus elevation) are the same.
• Indicator of the direction of seepage flow. • Total head contours
should decrease from upstream to downstream.
• Farthest upstream contour should be equivalent to the elevation of the
reservoir level.
Flow Paths (1) • Show individual water particles traveling within the flow
regime from upstream to downstream under a steady-state condition. • Flow
paths should intersect total head contours at right angles (or at least close to
right angles) for homogeneous sections.
• Flow paths may cross above the phreatic surface since water can flow from the
saturated to the unsaturated zone, and vice versa.

Output Feature Description Tips/Considerations

Phreatic Surface • Transition from positive to negative pore water pressures.
• Boundary between saturated and unsaturated flow. • Phreatic surface
should be the line of zero pressure.
• Undulations or irregularities in the phreatic surface may be an indication
that the mesh size needs to be reduced.
Pore Water Pressure Contours • Depict where pore water pressure values are the
same. • Pore water pressure contours should become increasingly more positive
below the phreatic surface and increasingly more negative above the phreatic
surface.
• Useful tool for verifying the model is properly computing the saturated and
unsaturated zones.
Flow Vectors • Depict the direction and magnitude of seepage flow. •
Larger vectors (arrows) indicate higher seepage flow velocities.
• Smaller vectors indicate lower seepage flow velocities.
• Understanding where high and low permeability zones are in the model should
help the user judge whether the relative flow direction and velocity magnitude make
sense.
Flow Sections (2016 and Newer
Versions of SEEP/W Use Subdomain Graphing Tool) • A defined section of the
model in which unit seepage flow quantity across the section is computed. •
Useful tool for evaluating the seepage flow rate through a specific material
region of interest.
Horizontal and Vertical Seepage Gradients • Change in total head (i.e., head
loss) over the length of the flow path.
• Two visual options: Option 1 defines an area of the model to compute average
seepage gradients and Option 2 looks at seepage gradient contours. • Option
1 is a useful tool for evaluating minimum, maximum, and average horizontal seepage
gradients and vertical (exit) gradients through a specific area of interest.
• Option 2 is useful as a check to Option
1. Author preference is to evaluate gradients using Option 1 because point
anomalies (i.e., gradient spikes) can be more difficult to define using Option 2.
As the contour interval decreases, the gradient point anomaly tends to increase
with no set end value. Refer also to the example discussion below.
Note:
(1) Total head contours and flow paths can be used to approximate the flow net.
Total head contours are equivalent to equipotential lines in a flow net. However,
flow paths are not the same as flow lines in a flow net. Thus, the addition of flow
paths to the total head contours can only simulate the flow net. In a flow net, the
amount of flow between each flow line (referred to as flow channels) must be
equivalent. In a seepage model, flow paths can be drawn at any point within the
flow regime of the model such that the flow between flow lines will not always be
equivalent.

An example seepage model using the SEEP/W computer program is presented in Figure
3-17 and is used here to provide additional guidance on interpreting results. The
example model is a zoned earthfill embankment dam consisting of a sandy clay core
with relatively low permeability and a core trench that extends to the top of
bedrock. The core is supported by upstream and downstream pervious sand shells that
are founded on slightly less pervious alluvium.

Figure 3-17: Example Seepage Model Using SEEP/W (Heitland et al. 2020)

The validation and interpretation of the seepage model results (shown in Figure 3-
18 through Figure 3-25) are summarized as follows:
• Figure 3-18 illustrates the resulting flow net simulation, with the total
head contours decreasing from upstream to downstream and the flow paths
intersecting the total head contours. As depicted by the red rectangle in Figure 3-
18, there are flow paths that cross above the phreatic surface near the downstream
embankment toe. This indicates the presence of water in the pore spaces of the
shell material near the toe such that the material is not completely unsaturated
(or saturated) and is indicative of upward flow.

Figure 3-18: Seepage Model Results from the SEEP/W Program Showing Total Head
Contours and Flow Paths (Flow Net Simulation) (Heitland et al. 2020)

• Pore water pressure contours are shown in Figure 3-19. The contours become
increasingly more positive below the phreatic surface and increasingly more
negative

above the phreatic surface, with the phreatic surface (dashed blue line)
representing the line of zero pressure.
• Review of the phreatic surface shown in Figure 3-18 and Figure 3-19 indicates
the embankment core is serving as an adequate seepage barrier, reducing the
phreatic surface through the core. Downstream of the core, the phreatic surface
extends approximately along the foundation contact to the downstream toe as you
would expect with a pervious shell founded on a slightly less permeable foundation.

Figure 3-19: Seepage Model Results from the SEEP/W Program Showing Pore Water
Pressure Contours (Heitland et al. 2020)

• Flow vectors are shown in Figure 3-20 and indicate that most of the seepage
flows below the embankment core trench through the foundation bedrock and then back
up through the foundation alluvium. This should be expected because the alluvium
has a higher permeability than the bedrock. Based on the size of the flow vectors,
the highest seepage flow velocities occur through the alluvium underlying the
downstream shell. As seen in the closeup of the red rectangle shown in Figure 3-20,
the flow vectors near the downstream embankment toe are oriented slightly upward,
indicating upward flow consistent with the flow paths.

Figure 3-20: Seepage Model Results Showing Flow Vectors (SEEP/W) (Heitland et al.
2020)

• Flow sections through the foundation alluvium and bedrock are shown in Figure
3-21 and Figure 3-22, respectively. In SEEP/W, the graphing tool can be used to
evaluate the seepage flow rate at a specified section through a material region by
plotting water rate on the vertical axis and the time step on the horizontal axis
(which is zero for a steady-state analysis). Review of the resulting seepage flow
quantities through the alluvium and bedrock verifies that the location of highest
flow is through the alluvium, which is consistent with the flow vectors.

Figure 3-21: Seepage Model Results from the SEEP/W Program Showing Flow Section
(Graphing Tool) – Foundation Alluvium (Heitland et al. 2020)

Figure 3-22: Seepage Model Results from the SEEP/W Program Showing Flow Section
(Graphing Tool) – Foundation Bedrock (Heitland et al. 2020)

• Horizontal seepage gradients through the downstream foundation alluvium and

vertical (exit) seepage gradients at the downstream toe area are shown in Figure 3-
23 and
Figure 3-24, respectively. Like the flow sections, the graphing tool in SEEP/W can
be used to evaluate the seepage gradients at a specified area by plotting gradient
(horizontal or vertical) on the vertical axis and the model distance along the
specified area on the horizontal axis. The resulting plot can be used to estimate
the minimum, maximum, and average gradients through the specified area, which can
further be used for evaluating internal erosion potential. As depicted in Figure 3-
24, a vertical gradient point anomaly (spike) is identified and a good indication
of where to potentially locate a toe drain. In addition to the graphing tool,
seepage gradients can be viewed as contours. Figure 3-25 shows the vertical
gradient contours at the downstream toe area for comparison to the plot shown in
Figure 3-24. The estimated average vertical gradient using the plot shown in Figure
3-24 is consistent with the contour shown at the downstream toe in Figure 3-25.
Both figures show the point anomaly representing a potential location for a toe
drain. It should be noted that as the contour interval decreases, the vertical
gradients at the location of the point anomaly will continue to increase. Thus, it
is strongly recommended to evaluate gradients using the graphing tool, as point
anomalies can be more difficult to define using only the contours.

Figure 3-23: Seepage Model Results from the SEEP/W Program Showing Horizontal
Seepage Gradients (Graphing Tool) (Heitland et al. 2020)

Figure 3-24: Seepage Model Results from the SEEP/W Program Showing Vertical (Exit)
Gradients (Graphing Tool) (Heitland et al. 2020)

Figure 3-25: Seepage Model Results from the SEEP/W Program Showing Vertical (Exit)
Seepage Gradient Contours (Heitland et al. 2020)

For complex seepage models, it is recommended that an experienced engineer with a

strong understanding of seepage principles and numerical modeling computer programs
perform the seepage analysis and interpret and verify the results. In some cases,
well-drawn flow nets of similar designs can be referenced to help verify the
general seepage conditions from a model’s output.
3.5.2 Reporting Results
Upon completion of a seepage analysis, the analysis should be adequately documented
in a report for submittal to the dam owner and/or regulator. The report should
include sufficient detail for the reader to understand the purpose of the analysis,
how the analysis was performed, the results of the analysis, and recommendations.
Seepage analysis reports should include a discussion on the following topics:
• Purpose – Clearly define the objective of the seepage analysis. Refer to
Section 3.3.1.
• Methodology – Describe the seepage analysis method and approach (steady-state
or transient, 2D or 3D, computer program used, etc.). Refer to Section 3.4.1.
• Model Geometry – Summarize the embankment geometry and internal zoning and
the foundation stratigraphy modeled in the seepage analysis. Refer to Section
3.4.2.1.
• Material Properties – Summarize the seepage material properties (e.g.,
permeability and anisotropy ratio) assigned to the embankment and foundation
materials and explain how they were developed. Refer to Section 3.4.2.2.
• Boundary Conditions – Summarize the boundary conditions assigned to the
exterior boundaries of the seepage model and their basis. Refer to Section 3.4.3.2.
• Model Calibration – If applicable, describe how the model was calibrated
based on past performance of the dam and piezometer data (if available). Refer to
Section 3.4.4.
• Results – Provide summary tables and figures of the results and summarize the
interpretation of the results. This may also include a discussion on the
uncertainties of the

evaluation and the results of sensitivity analyses to evaluate the uncertainties.

Refer to Section 3.5.1.
• Recommendations – If applicable, provide recommendations for action (e.g.,
increased monitoring, reservoir storage restrictions, remediation) or additional
investigations based on the results. The results may also be used for alternative
design recommendations.

4. Slope Instability of Dams

4.1 What Is Slope Stability?
Slope stability is the ability of a slope to resist the driving forces tending to
move earth materials downslope. The slopes of an embankment dam can move downward
and outward under the force of gravity. Instability occurs when there is an
imbalance of driving and resisting forces.
Generally, upstream earth embankment slopes should be no steeper than 3H:1V
(horizontal to vertical), and downstream earth embankment slopes no steeper than
2H:1V. The stability of an embankment can be adversely affected by excessive
stresses on the crest or slopes, sudden addition or loss of water in the reservoir,
changes in internal pore pressures, or loss of materials due to erosion (both
internal, such as piping or internal erosion, and external, such as surface
erosion). Slides, slumps, slips, cracking, excessive settlement, and other
deformations are forms of instability. Figure 4-1 shows an example of slope
instability on the downstream slope of an embankment dam.

Figure 4-1: Slope Instability on Downstream Slope of Embankment Dam (FEMA 2016)

All embankment dams in service deform and settle under self-weight and imposed
loads. Deformations occur as a response to the weight of a dam and routine
operations of the reservoir, including reservoir drawdown and flooding. Excessive
deformations occur when movements exceed tolerable limits.

4.2 Slope Instability-Related Issues

Slope instability-related failures of embankment dams are rare, comprising about 6
percent of all dam failures (Foster et al. 2000). Although not as common as
seepage-related failures (discussed in Section 2.2), failures due to slope
instability can be sudden and catastrophic, often with little to no warning signs.
For this reason, it is important to evaluate the factor of safety against slope
instability under all expected loading conditions through analysis.
This guidance document discusses the planning and interpretation of slope stability
analyses, which can provide insight into slope instability-related issues.
Identifying and understanding potential slope instability-related issues prior to
conducting a slope stability analysis will ultimately help in establishing the
objective of the stability analysis.
For new embankment dams, slope instability can occur during construction by
exceeding the strength of the foundation or new embankment fill. For existing
embankment dams, slope instability is typically triggered by some change in
condition, such as the occurrence of a rapid drawdown, flood, or seismic event,
which can increase the driving forces and decrease the resisting forces. A change
in seepage or settlement over time can also result in instability. Figure 4-2
illustrates various types of slope instability failures of embankment dams.

Figure 4-2: Types of Slope Instability Failures of Embankment Dams (Reclamation

1988)

Slope instability failures under construction or operational conditions are often

the result of design deficiencies, poor construction, and/or neglected remediation
actions. Under seismic conditions, slope instability failures are often the result
of seismic-induced deformations and/or strength loss caused by liquefaction (sand-
like soils) or cyclic softening (clay-like soils).
Excessive deformations may lead to the dam being overtopped, or differential
settlement may cause transverse cracks allowing rapid flow of water through the
embankment, resulting in dam failure due to internal erosion or elevated pore
pressures.

Regular visual inspection is the best tool for dam owners and engineers to assess
the safety of an embankment dam when it comes to both seepage- and slope
instability-related issues.
Visual observations or indicators related to potential seepage-related issues are
discussed in Section 2.2. Visual indicators related to potential slope instability-
related issues may include the following (AECOM 2013):
• Longitudinal cracks on the embankment crest or slopes – Figure 4-3.
• Wet areas or seepage on the downstream embankment slope or toe, indicating an
adverse internal phreatic surface within the embankment – Figure 4-4. The
relationship between reservoir level and seepage quantity and quality should be
established and used to compare successive observations.
• An apparent embankment slope failure, slump, or scarp – Figure 4-5.
• Erosion or sloughing of the downstream embankment slope, which results in
oversteepening of the overall slope.
• Bulges at or downstream of the embankment toe.
• Depressions or sinkholes in the embankment crest or slopes.
• Displaced riprap (on the upstream embankment slope), crest station markers,
or fence lines, indicating movement.
• Changes in the appearance of the normal reservoir waterline against the
upstream embankment slope at multiple water levels.

Figure 4-3: Severe Longitudinal Crack on Downstream Embankment Slope (AECOM 2013)

Figure 4-4: Seepage Exiting Downstream Embankment Slope (AECOM 2013)

Figure 4-5: Slope Failure on Downstream Embankment Slope (AECOM 2013)

5. Slope Stability Analysis of Dams

A slope stability analysis is a computational method that models the stability
conditions of an embankment dam and results in a factor of safety against sliding
(i.e., slope instability). The stability or instability of a mass of soil depends
on its weight, the external forces acting on it, and the shear strengths and pore
water pressures along a slip surface, such as that shown in
Figure 5-1. A slip surface is an assumed surface along which sliding may occur when
the forces causing instability of a soil mass exceed the shear strength of the soil
(i.e., shear resistance). If the shear resistance of the soil along the slip
surface exceeds that necessary to provide equilibrium, the mass is stable. If the
shear resistance is insufficient, the mass is unstable. Thus, the factor of safety

𝑆
against sliding (FS) is expressed by the equation:

𝐹𝑆 =
𝜏
Where: S = Available Shear Strength
 = Equilibrium Shear Stress

Figure 5-1: Embankment Slope and Potential Slip Surface (USACE 2003)

The most commonly used and accepted slope stability analysis methods for evaluating
embankment dams are limit equilibrium methods, which are often implemented in
computer programs.
The results of a slope stability analysis can be used for the following:
• Verifying slopes meet minimum factor of safety criteria established by
federal and/or state regulatory agencies,
• Evaluating piezometer thresholds to maintain stability, and
• Designing slope stabilization measures (e.g., berm, buttress, slope
flattening).
5.1 Slope Stability Analysis Principles
The theoretical principles that govern the equilibrium of a soil mass and that are
used to understand and evaluate the slope stability of embankment dams are based on
soil mechanics. The concepts of stress-strain behavior, undrained and drained
conditions, and total and

effective stresses are of fundamental importance in the mechanical behavior of

soils. This section provides a basic overview of the principles of soil mechanics
for slope stability analysis and was primarily adapted from AECOM (2015) and France
et al. (2015). Selection of appropriate shear strengths for stability analysis is
presented in Section 5.4.1.2. For more information on slope stability analysis
principles beyond that discussed below, refer to additional references in Section 7
(Duncan et al. 2014; Reclamation 2011; USACE 2003).
5.1.1 Stress-Strain Behavior
Understanding the stress-strain behavior of different soils is fundamental to
understanding how to characterize and test shear strength. Soils do not generally
exhibit significant tensile shear strength; they fail in shear under compression or
extension loading. Soil shear strength depends on the following:
• Types of soil particles and mineralogy,
• Consolidation pressure,
• Drainage allowed,
• Stress history (including overconsolidation), and
• Stress paths.
For slope stability analysis, the assumption is usually made that the stress-strain
curves of all soils involved reach a maximum shear stress, which then remains
constant with further strains, as shown by the generalized stress-strain curve in
Figure 5-2. If a material shows minor change in shear stress with increasing shear
strain or deformation, that material is said to have a ductile behavior and is not
sensitive to deformation. However, if a material shows lower shear stresses with
increased deformation, that material is called brittle or sensitive because of its
softening behavior under strain. For a brittle or sensitive material, the peak
shear strength versus the
post-peak shear strength is compared and discussed as peak-post-peak behavior.

Figure 5-2: Generalized Stress-Strain Curve (Duncan et al. 2014)

The primary factors controlling the shape of the stress-strain curve of soils
include the following:
• Soil type,
• Initial structure and state of particle arrangement, and
• Method of loading.
5.1.2 Undrained and Drained Conditions
When saturated or partially saturated soils are loaded in shear, they tend to
change in volume. Loose sands or normally consolidated clays tend to decrease in
volume (contract), while dense sands or overconsolidated clays tend to increase in
volume (dilate). If the loading is applied slowly enough, pore water will flow into
or out of the soil mass, the volume of the soil mass will change, and pore water
pressures will not change. However, if the loading is applied more quickly than
drainage can occur, pore water pressures will be generated within the soil mass.
Positive pore pressures will generate in loose sands or normally consolidated clays
due to the tendency to compress, while negative pore pressures will generate in
dense sands or overconsolidated clays due to the tendency to expand. Coarse-grained
soils (sands and gravels) have relatively high permeabilities and sufficient
drainage capacity to prevent pore water pressures from changing for most loadings
(with the exception of seismic loading, which is beyond the scope of this guidance
document), while fine-grained soils (clays and silts) have low permeabilities and
can develop excess pore water pressures during some static loading conditions.
Undrained conditions occur when loading is applied more rapidly than soil can
drain. Under undrained conditions, water cannot flow into or out of the soil in the
length of time the loading is applied. As a result, pore water pressures increase
or decrease in response to changes in load, as described above. Drained conditions
occur when loading is applied slowly enough relative to the permeability of the
soil that pore water can drain. Pore water pressures do not change under drained
loading conditions because water can move into or out of the soil freely in
response to changes in load.
Hence, whether a particular loading should be considered undrained or drained is
dependent on rate of loading, soil permeability, and the distance over which
drainage must occur to prevent pore water pressure changes. Duncan et al. (1990)
provide a logical basis for estimating the degree of drainage to evaluate whether a
material will behave in a drained or undrained manner during rapid drawdown. This
basis can be extended to other possible loading conditions to evaluate whether the
loading would be drained or undrained by using the dimensionless time factor (T),

𝑇 = 𝐶𝑣𝑡/𝐷2
which is expressed by the equation:

Where: Cv = Coefficient of Consolidation t = Loading Time

D = Length of Drainage Path
Table 5-1 summarizes typical ranges of coefficient of consolidation values for
various soil types.

Table 5-1: Typical Coefficient of Consolidation Ranges by Soil Type (Duncan et al.
1990)

Soil Type Coefficient of Consolidation, Cv (ft2/day)

Coarse Sand >10,000
Fine Sand 100 to 10,000
Silty Sand 10 to 1,000
Silt 0.5 to 100
Compacted Clay 0.05 to 5
Soft Clay <0.2

If the value of T exceeds 3.0, it is reasonable to treat the material as drained.

If T is less than 0.01, it is reasonable to treat the material as undrained. If T
is between these two limits, both possibilities should be considered.
Alternatively, soils having a permeability greater than approximately 1x10-3
centimeters per second (cm/s) can be considered to be free-draining under static
loading, as a general rule of thumb (Duncan et al. 2014). Although conditions can
be intermediate between undrained and drained, loading conditions are almost always
modeled as either one or the other. In some cases, when it is not clear whether the
loading conditions are undrained or drained, both cases are considered in the
analysis.
5.1.3 Total and Effective Stresses
Total stresses within a soil mass include both stresses resulting from forces
transmitted through inter-particle contacts and pore water pressures. Effective
stresses within a soil mass include only stresses resulting from the forces
transmitted through inter-particle contacts. At any given location, the effective
stress equals the total stress minus the pore water pressure.
Soil shear strengths can be defined as a function of either total stresses or
effective stresses. Effective stress methods should always be used for drained
loading conditions. For undrained loading, the analyst needs to choose between
total stress and effective stress methods. Total stress methods are used when it is
easier to predict the shear strength during undrained loading than it is to predict
the pore water pressures during undrained loading, which is almost always the case.
Soil shear strengths are always governed by effective stresses, or inter-particle
forces, regardless of loading condition. Total stress strength characterizations
are simply used in those cases where pore water pressure responses cannot be easily
predicted, and the undrained strength can be more easily predicted. The pore water
pressure is implicit in the selected total stress strength; the pore pressure is
whatever value is necessary to produce an effective stress state that results in
the predicted strength.
5.2 Slope Stability Analysis Methods
The most commonly used and accepted slope stability analysis methods for evaluating
embankment dams are limit equilibrium methods, which are often implemented in
computer programs. Other slope stability analysis methods include the finite
element method and simple slope stability charts. Robust computing capabilities
have made the use of stability modeling with complex geometries and shear strength
characterizations the current standard of practice. This section provides a basic
overview of the typical slope stability analysis methods, focusing specifically on
limit equilibrium methods.

5.2.1 Limit Equilibrium Methods

Limit equilibrium methods of analysis is the oldest form of numerical slope
stability evaluation. The history of slope stability analyses is described by GEO-
SLOPE (2012) starting in 1936, when Fellenius introduced the Ordinary or Swedish
method of slices. In the mid-1950s, Janbu and Bishop developed advances in this
method. The advent of computers in the 1960s led to more rigorous methods, such as
those developed by Morgenstern and Price and Spencer. One of the reasons limit
equilibrium methods were adopted so readily is that solutions could be obtained by
hand calculations. The introduction of powerful, personal computers in the early
1980s led to the development of commercial computer programs based on limit
equilibrium methods.
5.2.1.1 General
Limit equilibrium methods use one or more of the equations of static equilibrium
applied to the soil mass bounded below by an assumed slip surface and above by the
surface of the slope. The three equations of static equilibrium include: (1)
equilibrium of forces in the vertical direction (∑Fy = 0), (2) equilibrium of
forces in the horizontal direction (∑Fx = 0), and (3) equilibrium of moments about
any point (∑M = 0). Two different procedures can be used to satisfy static
equilibrium: a single free-body procedure or a series of individual vertical slices
making up the total free body (termed the procedure of slices). The single free-
body procedure is relatively simple to use and includes infinite slope, logarithmic
spiral, and Swedish slip circle methods.
This subsection focuses on the procedure of slices, which is the most common
approach used for the stability analysis of embankment dams.
In the procedure of slices, the soil mass above the slip surface is subdivided into
a finite number of vertical slices. The number of slices depends on the slope
geometry and soil profile.
Figure 5-3 is a schematic illustrating the procedure of slices and typical forces
acting on an individual slice.

Figure 5-3: Schematic of (a) Procedure of Slices and (b) Typical Forces on an
Individual Slice (USACE 2003)

Except for the weight of the slice, all the forces, locations of the forces, and
factor of safety are unknowns and must be calculated in a way that satisfies static
equilibrium. There are more unknowns than the number of equilibrium equations.
Therefore, assumptions must be made to achieve a statically determinate solution.
Several limit equilibrium methods using the procedure of slices have been developed
over time and include the following:
• Ordinary Method of Slices
• Simplified Bishop Method
• Modified Swedish Method
• Simplified Janbu Method
• Lowe-Karafiath Method
• Morgenstern-Price Method
• Spencer’s Method
Each method subscribes to a different set of assumptions to achieve a balance of
equations and unknowns and satisfy static equilibrium. Each method also differs
with regard to which equilibrium equations are satisfied. For example, the Ordinary
Method of Slices, Simplified Bishop, Modified Swedish, Simplified Janbu, and Lowe-
Karafiath Methods do not satisfy all static equilibrium equations. The Ordinary
Method of Slices only satisfies overall moment equilibrium about the center of the
circle. The Simplified Bishop Method satisfies vertical force equilibrium for each
slice as well as overall moment equilibrium about the center of the circle.

The Modified Swedish, Simplified Janbu, and Lowe-Karafiath Methods are “force
equilibrium” procedures that satisfy vertical and horizontal force equilibrium for
each slice but ignore moment equilibrium. Conversely, the Morgenstern-Price Method
and Spencer’s Method satisfy all static equilibrium equations. Methods that satisfy
static equilibrium fully are referred to as “complete” equilibrium methods.
Complete equilibrium methods have generally been more accurate than those that do
not satisfy complete static equilibrium and therefore are preferable to
“incomplete” methods. However, the incomplete methods are often sufficiently
accurate and useful for many practical applications, including hand calculations
and preliminary analyses.
Primary limitations of the limit equilibrium methods include the following:
• The factor of safety is assumed to be constant along the slip surface.
Although the factor of safety may not in fact be the same at all points on the slip
surface, the average factor of safety computed by assuming that the value is
constant provides a valid measure of stability for slopes in ductile (nonbrittle)
soils. For slopes in brittle soils, the factor of safety computed assuming the
value is the same at all points on the slip surface may be higher than the actual
value.
• The stress-strain behavior of soils is not explicitly accounted for. If the
shear strength is fully mobilized at any point on the slip surface, the soil fails
locally. If the soil has brittle stress-strain characteristics so that the strength
drops once the peak strength is mobilized, the stress at that point of failure is
reduced, and stresses are transferred to adjacent points, which in turn may then
fail. In extreme cases, this may lead to progressive failure and collapse of the
slope. If soils possess brittle stress-strain characteristics with relatively low
residual shear strengths compared to the peak strengths, reduced strengths and/or
higher factors of safety may be required for stability.
• The initial stress distribution within the slope is not explicitly accounted
for. Limit equilibrium methods aim to provide static equilibrium for each slice and
make the factor of safety the same for each slice. These inherent concepts and
assumptions mean that it is not always possible to obtain realistic stress
distributions along the slip surface or within the potential sliding mass.
For more information on the various limit equilibrium methods, refer to additional
references in Section 7 (Duncan et al. 2014; GEO-SLOPE 2012; USACE 2003).
5.2.1.2 Selection of Method
A comparison of the slope stability limit equilibrium analysis methods using the
procedure of slices is summarized in Table 5-2 and can be helpful in selecting a
suitable method for stability analysis. As discussed above, some limit equilibrium
methods satisfy all static equilibrium equations, while others satisfy one or two
of the equations. Some methods are restricted to circular slip surfaces, while
others can evaluate both circular and noncircular slip surfaces.
Some methods are more rigorous and require the aid of a computer program, while
others can be used without the aid of a computer program and are convenient for
checking results obtained from a computer program. Spencer’s Method is a rigorous
method and considered the standard of practice when it comes to detailed
evaluations of embankment dams.

Seepage and Slope Stability Modeling for Embankment Dams

Table 5-2: Comparison of Slope Stability Limit Equilibrium Analysis Methods Using
Procedure of Slices

Feature Ordinary Method of Slices Simplified Bishop Method Modified

Swedish Method Simplified Janbu Method Lowe- Karafiath Method
Morgenstern- Price Method
Spencer’s Method
Accuracy X X X X
Satisfies Vertical Force Equilibrium X X X X X X
Satisfies Horizontal Force Equilibrium X X X X X
Satisfies Moment Equilibrium X X X X
Circular Slip Surfaces X X X X X X X
Noncircular Slip Surfaces X X X X X
Suitable for Hand Calculations X X X X X

France and Winckler (2010) evaluated the sensitivity of various limit equilibrium
analysis methods on the calculated factor of safety for a critical slip surface
using the example slope stability model presented in Figure 5-4. For the four limit
equilibrium analysis methods shown in the figure, the variation in the calculated
factors of safety between analysis methods is relatively small and within the
typical level of accuracy of stability evaluations.

Figure 5-4: Sensitivity of Various Slope Stability Limit Equilibrium Analysis

Methods on Factor of Safety (France and Winckler 2010)

5.2.1.3 Computer Programs

Typical computer programs for modeling 2D and 3D limit equilibrium slope stability
are summarized in Table 5-3. Computer programs are user friendly and can model a
wide variety of slope geometries, soil profiles, soil shear strengths, pore water
pressures, and external loads. Most programs also have capabilities for
automatically searching for the most critical slip surface with the lowest factor
of safety and can handle both circular and noncircular slip surfaces.
Table 5-3: Typical Computer Programs for Modeling Slope Stability Using Limit
Equilibrium Methods
Computer Program
Method of Modeling
SLOPE/W
(GeoStudio) 2D and 3D
• Ordinary Method of Slices
• Simplified Bishop Method
• Modified Swedish Method
• Simplified Janbu Method
• Lowe-Karafiath Method
• Morgenstern-Price Method
• Spencer’s Method

Computer Program
Method of Modeling
UTEXAS4 2D
• Simplified Bishop Method
• Modified Swedish Method
• Lowe-Karafiath Method
• Spencer’s Method
SLIDE 2D and 3D
• Ordinary Method of Slices
• Simplified Bishop Method
• Modified Swedish Method
• Simplified Janbu Method
• Lowe-Karafiath Method
• Spencer’s Method
PLAXIS LE 2D and 3D
• Over 15 classic procedure of slices methods

In some situations, there may be merit for dam owners or regulators to develop a
simplified slope stability model to assist in making decisions. While slope
stability models can be valuable in understanding the performance of an embankment
dam, the cost of purchasing a computer program can rarely be financially justified
for owners and regulators. A new affordable option is GEO-SLOPE’s “Basic SLOPE/W,”
which is a trimmed down version of “SLOPE/W” that is well suited for owner and
regulatory needs. Limitations of the basic version include the inability to model a
staged rapid drawdown analysis, limited options for soil strength modeling, and the
use of only one piezometric line. However, owners and regulators are not generally
going to perform a rapid drawdown analysis or use undrained strength functions.
Rather, the owner or regulator is likely attempting to understand the embankment
and foundation, the drained stability, missing information, and most importantly,
whether a more refined stability model may be warranted.
The complex stability modeling is often best left to the consulting engineer.
5.2.2 Finite Element Method
While limit equilibrium methods are capable of providing an accurate index of slope
stability, the calculated stress distributions are not necessarily representative
of actual field conditions. This is because limit equilibrium methods essentially
make assumptions to convert a statically indeterminate problem into a statically
determinate one. The finite element method can be used to compute the stresses and
displacements caused by applied loads. However, it does not provide a value for the
overall factor of safety without additional processing of the computed stresses.
The basis for the finite element method is the representation of a body or a
structure by an assembly of finite elements. These elements are interconnected at
nodal points to form a finite element model. Solutions are obtained in terms of
displacements at these nodal points and average stresses in the elements. This
procedure can account for various types of stress-strain behavior, heterogeneous
conditions, irregular geometry, and complex boundary conditions, as well as time-
dependent loading. The computed shear stresses are compared to the corresponding
shear strengths to evaluate the factor of safety on an element-by-element basis.
This information is used to assess an average factor of safety along an assumed
slip surface by

taking an average of the calculated factor of safety values for the elements along
the shear surface. Similarly, potentially critical shear zones are identified by
connecting the elements with low factor of safety values.
In general, slope stability analyses using limit equilibrium methods and the finite
element method calculate similar factors of safety for a slip surface. For more
information on the finite element method, refer to additional references in Section
7 (GEO-SLOPE 2012; USACE 2003).
5.2.3 Slope Stability Charts
Slope stability charts provide a means for rapid analysis of slope stability by
estimating the factor of safety for various types of slopes and soil conditions.
Stability charts rely on dimensionless relationships between factor of safety and
other parameters that describe the slope geometry, soil shear strengths, and pore
water pressures. They are useful for preliminary estimates of stability or checking
detailed analyses. However, chart solutions should never be used as the only means
of analyzing stability. For more information on example slope stability charts and
procedures for using the charts, refer to additional references in Section 7
(Duncan et al. 2014; USACE 2003).
5.3 When Is a Slope Stability Analysis Warranted?
Design of new embankment dams and the more common scenario of reviewing the
conditions of existing embankment dams should, as general practice, include
evaluating the slope stability of the embankment structure. A slope stability
analysis of an embankment dam is a criteria- based evaluation to assess whether the
embankment is stable under various loading conditions and meets minimum factor of
safety criteria established by federal and/or state regulatory agencies. Because
slope stability is a criteria-based evaluation, identifying when a slope stability
analysis may be warranted is typically much more straightforward than identifying
when a seepage analysis may be warranted. Seepage models can be adjusted based on
very little data until the results look like what the modeler hoped they would.
Therefore, it takes a more critical assessment when it comes to identifying whether
a seepage analysis may be warranted, as discussed in Section 3.3. Assessing whether
a slope stability analysis may be warranted is more general and can be related to
specific triggers. Triggers that signify a stability analysis will be required may
include the following (AECOM 2013; Reclamation 1988):
• Designing a new embankment dam.
• Raising an existing embankment dam.
• Construction of a berm to address stability issues.
• Potential reclassification of a dam to high hazard.
• Deterioration of existing conditions (e.g., oversteepening of the embankment
slopes for any reason).
• Visual observations (e.g., longitudinal cracking along the embankment crest
or slopes, scarps, toe bulges) indicate that instability may be developing.
• Surface measurement points and/or internal instrumentation (e.g.,
inclinometers) indicate movement.
• Internal instrumentation (e.g., piezometers) indicate excessive pore water
pressures in the embankment and/or foundation.

• Reassurance is needed that a latent, undetected issue has not developed;

indicators of such an issue may include embankments with steep slopes (steeper than
2H:1V), soft foundation conditions, high phreatic surface within the embankment
and/or foundation, seepage at the slope or toe, observed scarp or bulge, or
depression/sinkhole formation.
• Review of design and construction records indicate the presence of previously
unrecognized but potentially harmful geologic conditions.
• Gradual loss of strength in foundation clay shales or overconsolidated clays
due to swelling.
• Embankment is experiencing an unusually high and perhaps sustained reservoir
level.
• Embankment is anticipated to experience an unusually severe drawdown of the
reservoir. The severity of drawdown can be in terms of a more rapid rate or to a
lower level than it has experienced before.
• Embankment has been exposed to a prolonged dry period followed by rain. The
dry period can cause desiccation cracks to develop in some dams; subsequent rain
can fill the cracks with water and precipitate slides.
• There is a need for a dynamic deformation analysis.
• There is no slope stability analysis for the embankment dam, stability under
specific loading conditions have not been evaluated, or the characterization of
specific loading conditions have changed.
In some cases, a slope stability analysis may already exist for an embankment dam,
and an assessment must be made as to whether the existing analysis is sufficient in
capturing the stability of the current structure. Triggers warranting an additional
stability analysis may include the following (Reclamation 1988):
• Existing analysis is not in agreement with the current accepted standard-of-
practice methodologies.
• Existing conditions have deteriorated.
• Hazard potential of the dam has increased.
• Embankment has been or will be subjected to loading conditions more severe
than designed for.
• Assumed design or analysis parameters cannot be satisfactorily justified.
If satisfactory behavior of an existing embankment dam is observed under loading
conditions that are not expected to be exceeded during the life of the structure,
then a slope stability analysis may not be warranted. This is provided that adverse
changes in the physical condition of the embankment do not occur.
Other considerations when assessing the need for a slope stability analysis should
include the following:
• Clearly define the objective of the analysis. Refer to Section 3.3.1 for
guidance on developing a clear, well-defined objective. Supplementing objectives
with hand drawings can be helpful.
• List the benefits of performing a slope stability analysis and the
consequences of not performing an analysis.

It is important to involve dam owners and regulators early in the decision to

complete an analysis. What is important to the analyst may vary from that of the
owner or regulator. Sometimes the results of an analysis will not justify the cost.
The results of the analysis should aim to provide tangible benefits to the project.
If it is determined that a slope stability analysis is warranted, then consult the
following sections for guidance for on planning a stability analysis, as well as
interpreting, verifying, and reporting the results.
5.4 Planning for a Slope Stability Analysis
Similar to seepage modeling, slope stability modeling can be beneficial in
understanding the design and performance of embankment dams. However, a stability
model is only as good as the modeler’s understanding of the inputs, how the inputs
are used in the model, and the modeler’s ability to interpret the results. Thus,
modelers must be fluent in the model inputs, outputs, and soil mechanics and
recognize the model sensitivities to get the most out of a stability model.
This section provides tips, tools, and guidance on planning for a slope stability
analysis and focuses specifically on modeling using limit equilibrium methods
implemented in computer programs, which are the most commonly used methods for
evaluating the stability of embankment dams. Other slope stability analysis methods
are introduced in Section 5.2, with references to additional publications for
further information on these methods.
Planning for a slope stability analysis involves defining the modeling approach and
minimum data requirements. The planning process should also include general
modeling considerations. Guidance is provided below for each of these topics.
Guidance for interpreting, verifying, and reporting results is provided in Section
5.5.
5.4.1 Modeling Approach
The first step in performing a limit equilibrium slope stability analysis involves
defining the modeling approach, which includes considerations for the following:
• Geometry,
• Loading conditions,
• Factor of safety criteria,
• 2D plane strain versus 3D modeling,
• Piezometer threshold analysis, and
• Back analysis.
5.4.1.1 Geometry
Similar to seepage modeling, the geometry of both the dam site and embankment
should be taken into consideration in slope stability modeling. In a 2D analysis,
cross sections should be selected at locations where critical stability conditions
are expected and stability results are required. This will typically include the
maximum embankment section, at a minimum, as well as other locations along the dam
alignment that may have steep embankment slopes, a narrow embankment crest, or more
adverse/variable foundation or embankment conditions. In a 3D analysis, an entire
dam can be included in the model.

5.4.1.2 Loading Conditions

Loading conditions are the external loads that apply stresses on an embankment dam.
It is important to understand the potential loading conditions for which an
embankment dam should be evaluated. Shear strengths in embankment and foundation
soils need to be characterized differently for various loading conditions that can
occur during the life of the dam. As a result, the slope stability of an embankment
dam varies depending on the particular loading condition. A potential sequence of
loading conditions during the life of an embankment dam is shown in Figure 5-6.
Typical loading conditions include during-and-end-of-construction, steady-state,
rapid drawdown, flood, and post-earthquake. Each of these loading conditions is
discussed further below, including selection of appropriate shear strengths. Refer
to applicable state or federal agency regulations for specific requirements related
to when each loading condition should be evaluated. For more information on slope
stability loading conditions beyond that discussed below, refer to additional
references in Section 7 (AECOM 2015; Duncan et al. 2014; France et al. 2015;
Reclamation 2011; USACE 2003).

Figure 5-5: Potential Sequence of Loading during Life of Embankment Dam

During-and-End-of-Construction
The during-and-end-of-construction loading condition represents the stability of an
embankment dam at specified stages while it is being constructed. Construction may
include the initial construction of the dam or additional dam modifications. This
loading condition assumes there is no water stored in the reservoir and no phreatic
surface present within the embankment during and at the end of construction.
Embankment and foundation materials are evaluated using either drained or
unconsolidated- undrained shear strengths depending on the saturation and
permeability of the soil. Fine-

grained (cohesive) soils generally have low permeability such that little drainage
occurs during construction. Coarse-grained (cohesionless) soils have relatively
high permeability and are typically free draining. Increased load induced by the
placement and compaction of embankment fill during construction may generate excess
pore water pressures in the low permeability embankment and foundation materials,
thereby developing undrained strength conditions in these materials. Therefore,
unconsolidated-undrained shear strengths are assigned to the low permeability
materials, and drained shear strengths are assigned to the free-draining materials.
The during-and-end-of-construction loading condition should also be evaluated when
unconsolidated-undrained shear strengths are estimated to be less than drained
shear strengths (contractive soils).
Both the upstream and downstream embankment slopes are evaluated under the during-
and- end-of-construction loading condition.
Steady-State
The steady-state (drained) loading condition represents the long-term stability of
an embankment dam under normal operating reservoir conditions. This loading
condition assumes pore water pressures within the embankment have reached their
steady-state seepage condition with no excess pore water pressures remaining from
construction, elevated reservoir levels, or other new loading, and all materials
are assumed to be fully consolidated under the embankment load. For this loading
condition, the phreatic surface and internal piezometric conditions correspond to
long-term, normal operating conditions with the reservoir level conservatively
modeled at the maximum normal pool level, or service/principal spillway crest
elevation. All embankment and foundation materials are assigned drained shear
strengths related to effective stresses.
Both the upstream and downstream embankment slopes are typically evaluated under
the steady-state loading condition. Often, the factor of safety for the downstream
slope is the most critical case. The upstream slope generally has a higher factor
of safety than the downstream slope due to the stabilizing effect of water pressure
on the upstream slope from reservoir storage, providing a buttressing effect on the
embankment. However, the factor of safety of the upstream slope can be low if the
slope is very steep or there is something unusual about the embankment zonation or
foundation conditions. In this case, analysis of the upstream slope is warranted.
If annual pool levels are often at partial pool conditions, a pool level below the
maximum normal pool should be evaluated for upstream slope stability, as it may be
the more critical case.
Rapid Drawdown
The rapid drawdown loading condition represents the stability of an embankment dam
under an assumed instantaneous (i.e., rapid) lowering of the reservoir level from a
steady-state pool level, typically taken as the maximum normal pool, to the lowest
outlet elevation, removing the buttressing effect of the reservoir on the upstream
slope. This loading condition assumes the fine-grained, low permeability embankment
and foundation materials below the steady-state phreatic surface are saturated to
steady-state conditions prior to drawdown and remain saturated after drawdown.
During rapid drawdown of the reservoir, the rate of unloading on the upstream slope
may occur rapidly enough that pore water pressures do not have time to dissipate
within the saturated, low permeability materials. The excess pore water pressures
in these materials therefore develop undrained strength conditions.
The state of practice for evaluating the rapid drawdown loading condition is to use
a three-stage slope stability analysis described by Duncan et al. (1990). This
method evaluates appropriate shear strength conditions for the materials depending
on the stress conditions prior to, during,

and after drawdown. The first stage of the analysis calculates the effective
stresses for the existing steady-state seepage condition (i.e., steady-state
phreatic surface) of the embankment prior to drawdown. In the first stage, the
phreatic surface and internal piezometric conditions correspond to long-term,
normal operating conditions with the reservoir level conservatively modeled at the
maximum normal pool level, and all embankment and foundation materials are assigned
drained (effective stress) shear strengths.
The second stage of the analysis calculates the effective stresses for the phreatic
surface after drawdown. In the second stage, the phreatic surface is modeled at the
lowest outlet elevation, and all saturated, low permeability embankment and
foundation materials that cannot drain as the reservoir is lowered are assigned
undrained shear strengths based on the effective stresses before drawdown, as
calculated in the first stage. Coarser, free-draining materials having a
permeability greater than 1x10-3 cm/s are typically assigned drained shear
strengths (Duncan et al. [2014]).
The third stage of the analysis compares the calculated undrained shear strength of
the low permeability materials based on the steady-state phreatic surface prior to
drawdown (calculated in the first stage) with the calculated drained shear strength
of these materials based on the phreatic surface after drawdown (calculated in the
second stage). If the undrained shear strength calculated from the first stage is
greater than the drained shear strength from the second stage at any point along
the slip surface that passes through the low permeability materials, the drained
shear strength is used at that location for the third stage analysis. The third
stage uses drained shear strengths wherever the undrained shear strength is greater
to prevent the analysis from relying on the development of negative pore water
pressures for stability. The factor of safety is calculated using the lower of
either the undrained or drained strengths for the saturated, low permeability
materials from the third stage with the drained strengths for all other materials
and reservoir pressures on the upstream embankment slope from the second stage.
Only the upstream embankment slope is evaluated under the rapid drawdown loading
condition since this loading condition does not affect the stability of the
downstream slope.
Flood
The flood loading condition represents the stability of an embankment dam under a
raising of the reservoir level from the steady-state, maximum normal pool level to
a flood pool level resulting from a hydrologic or operational event. Inflow
produced by hydrologic events can cause the reservoir behind an embankment to rise
to levels higher than the normal pool level typically considered for the steady-
state loading condition. The higher reservoir level increases the water pressure
acting on the upstream slope. Depending on how quickly the reservoir rises and the
permeabilities of the embankment and foundation materials, piezometric pressures
within the embankment and foundation may increase.
Traditionally, the stability of an embankment dam under the flood loading condition
has been analyzed by estimating the steady-state seepage condition that would be
expected to develop if the higher reservoir level were in place long enough to
allow steady-state seepage to fully develop. The loading condition was evaluated
using piezometric pressures estimated for the steady-state seepage condition with
the higher reservoir level in conjunction with drained shear strengths assigned to
all materials. It was recognized that, in many cases, the permeabilities of the
materials involved were low enough that steady-state seepage likely would not
develop during flood loading, but the approach was believed to be conservative.
However, if this conservative approach with elevated piezometric pressures and
drained strengths indicates an unsatisfactory factor of safety, it may be
appropriate to consider an alternate approach.

With modern access to computer programs that make transient seepage analyses
easier, some analysts have begun to use transient seepage analyses to estimate
phreatic conditions expected to develop during the estimated duration of the flood,
and then use those seepage conditions in conjunction with drained strengths in the
analysis. There are two potential problems with this approach. First, the accuracy
of transient seepage analyses is subject to significant uncertainty because of
variations in permeability and stratigraphy. The uncertainties in transient
analyses are even greater because of additional parameters that must be estimated
for the analyses. Second, as was noted in the discussion of the rapid drawdown
loading condition, if the flood load is applied more quickly than water can flow in
or out of some of the materials, pore water pressures will generate in those
materials in response to shear loads, and these shear-induced pore water pressures
are not considered in transient seepage analyses.
For most large embankment dams, the increase in reservoir level is relatively small
compared to the normal reservoir depth, and the changes in shear loads may not be
large. But for smaller embankment dams or flood management dams, the increase in
reservoir level may be significant, and the undrained loading condition may need to
be considered.
Therefore, more recent analysis of the flood loading condition assumes that the
duration of the hydrologic event would not be sufficient to develop steady-state
seepage in the fine-grained, low-permeability embankment and foundation materials.
Pore water pressures within the embankment are initially taken as those developed
under steady-state seepage with the reservoir level at the maximum normal pool. The
rate of increased loading on the upstream slope resulting from the higher reservoir
level during the hydrologic event may occur rapidly enough to generate excess pore
water pressures in the saturated, low-permeability materials, thereby developing
undrained strength conditions in these materials.
A flood loading condition should therefore be evaluated using a two-stage slope
stability analysis to evaluate appropriate shear strength conditions for the
materials depending on the stress conditions prior to and during the flood. The
two-stage analysis is similar to the first two stages of the rapid drawdown loading
condition described above, in which effective stresses for the steady-state seepage
condition (i.e., steady-state phreatic surface) of the embankment prior to the
flood are calculated in the first stage of the analysis and effective stresses for
the phreatic surface during the flood are calculated in the second stage of the
analysis. The factor of safety is computed using undrained shear strengths assigned
to the saturated, low-permeability embankment and foundation materials based on the
effective stresses prior to the flood (calculated in the first stage) and drained
shear strengths assigned to the coarser, free-draining materials based on the
effective stresses during the flood (calculated in the second stage), as well as
reservoir pressures on the upstream embankment slope from the second stage.
Both the upstream and downstream embankment slopes are typically evaluated under
the flood loading condition.
Post-Earthquake
The post-earthquake loading condition represents the stability of an embankment dam
at the end of shaking resulting from a seismic event. Rapid shaking and
accelerations produced by seismic events can induce cyclic loading within the
embankment and foundation. This cyclic loading may generate excess pore water
pressures in the saturated, fine-grained, low- permeability embankment and
foundation materials, thereby developing undrained strength conditions in these
materials. Cyclic loading may also cause strength loss (i.e., residual strength
conditions) due to liquefaction of saturated, loose, sand-like materials or cyclic
softening of saturated, soft, clay-like materials. Thus, this loading condition
often requires a liquefaction- triggering and/or cyclic-softening analysis of
loose/soft embankment and foundation materials to evaluate the liquefaction or
cyclic softening potential of these materials.

Embankment and foundation materials that are not susceptible to liquefaction or

cyclic softening can be assigned drained or undrained shear strengths related to
steady-state effective stresses prior to the earthquake. Undrained shear strengths
are assigned to the saturated, low- permeability materials, and drained shear
strengths are assigned to the coarser, free-draining materials. Conversely,
embankment and foundation materials that are found to be susceptible to
liquefaction or cyclic softening are assigned residual shear strengths. For more
information on liquefaction and cyclic softening evaluations, refer to additional
references in Section 7 (Idriss and Boulanger 2008).
Both the upstream and downstream embankment slopes are typically evaluated under
the post- earthquake loading condition. If the stability analysis under the post-
earthquake loading condition indicates an unsatisfactory factor of safety, a
numerical dynamic deformation analysis may be warranted to estimate the embankment
deformations resulting from an earthquake for comparison against tolerable
deformations. Dynamic deformation analyses are highly specialized and beyond the
scope of this guidance document.
Further guidance for evaluating whether an embankment dam requires a post-
earthquake stability analysis is provided in Technical Note 5 – Simplified Seismic
Analysis Procedure for Montana Dams (HDR Engineering 2020), which lays out a
simplified stepwise path using recent industry publications. Although the ground
motions used in this reference are specific to Montana, the procedures are widely
applicable.
5.4.1.3 Factor of Safety Criteria
The analyzed stability of a slope is expressed as a factor of safety. A factor of
safety greater than 1 indicates the estimated driving forces are less than the
resisting forces. However, due to inherent uncertainties in the behavior and
characterization of earth materials that compose embankment dams, regulations and
good practice require a factor of safety greater than 1 for the various loading
conditions. Therefore, it is important to establish factor of safety acceptance
criteria for slope stability modeling. Typical factor of safety criteria for slope
stability analyses based on Reclamation (2011) and/or USACE (2003) guidelines are
summarized in Table 5-4.
Criteria may differ among various state and federal agencies.
Table 5-4: Typical Factor of Safety Criteria
Loading Condition Minimum Factor of Safety
During-or-End-of-Construction 1.3
Steady- State (Reservoir at maximum normal pool level) 1.5
Rapid Reservoir drawdown from maximum normal pool level to inactive pool (lowest
1.3
Drawdown outlet elevation)
Reservoir drawdown from inflow design flood (IDF) pool to inactive pool
(lowest 1.1
outlet elevation)
Reservoir drawdown from IDF pool to active pool (maximum normal pool level)
1.2
Flood (Reservoir at IDF pool level) 1.2 / 1.4 (1)
Post-Earthquake 1.2 to 1.3 (2)
Notes:
(1) The Reclamation (2011) and USACE (2003) guidelines differ in their minimum
factor of safety for the flood loading condition. Reclamation recommends a minimum
factor of safety of 1.2 for the condition in which steady-state seepage develops
with the reservoir at the IDF pool level. USACE recommends a minimum factor of
safety of 1.4 with the reservoir at the IDF pool level but uses internal pore water
pressures taken as those developed under the steady-state seepage condition with
the reservoir at the maximum normal pool level. If the IDF pool level is

anticipated to be a relatively small change from the maximum normal pool level in
comparison to the dam height and the flood loading duration will not be sufficient
to develop steady-state seepage in the low permeability embankment and foundation
materials at the surcharge level, the USACE guidelines are considered more
representative.
(2) The Reclamation (2011) and USACE (2003) guidelines do not present a minimum
factor of safety for the post- earthquake loading condition. A factor of safety of
1.2 to 1.3 is consistent with guidance provided by FEMA (2005) and with standard of
practice and should be selected based on consideration of uncertainties. This range
of factors of safety is considered to be representative of limited post-earthquake
deformations.

5.4.1.4 2D versus 3D Modeling

Historically, slope stability analyses have primarily comprised 2D modeling because
of the complexity of 3D modeling and limited computer programs capable of 3D
modeling. However, similar to seepage analyses, slope stability analyses have
recently been expanding into the 3D realm as the benefits of 3D modeling are
becoming clearer and computer programs with 3D capability are becoming more
prevalent.
While 3D slope stability modeling can be valuable under certain conditions, 3D
modeling is much more rigorous than 2D and therefore requires careful consideration
to ensure 3D modeling is the appropriate approach. 3D modeling is more time-
consuming, requires a greater level of effort and expertise, and is more costly
than 2D modeling. However, 3D slope stability modeling becomes beneficial for
conditions in which there are significant 3D cross-valley effects along the
embankment dam alignment (e.g., narrow “v”-shaped valley profile) or when
evaluating the abutment-embankment contact area (i.e., groin).
5.4.1.5 Piezometer Threshold Analysis
In some cases, it may be beneficial to perform a piezometer threshold analysis to
provide a basis for establishing critical levels for piezometers within an
embankment dam. A piezometer threshold analysis is conducted by incrementally
raising the phreatic surface through the embankment at locations of piezometers
until the factor of safety for the critical slip surface is reduced to a target
factor of safety. The target factor of safety may be selected to be consistent with
the acceptance criteria, or a lower value may be selected to evaluate the critical
piezometer threshold levels that may warrant action.
5.4.1.6 Back Analysis
When an embankment slope fails due to instability, a back analysis can be performed
to provide useful information into the conditions of the slope at the time of
failure. A back analysis is conducted using the resulting slip surface and an
assumed factor of safety of 1.0 to back- calculate the unit weights and maximum
potential shear strengths of the soils and/or pore water pressure conditions at the
time of failure. The back-calculated conditions can be used to inform the design of
slope stabilization measures. For some cases, such as with brittle soil, the factor
of safety could be significantly less than 1 during the failure. For more
information on assessing the conditions of a failed slope through back analysis,
refer to additional references in Section 7 (Duncan et al. 2014).
5.4.2 Minimum Data Requirements
Slope stability models require accurate data to obtain reliable results.
Understanding the minimum data requirements that go into a stability model is
valuable in terms of identifying where data gaps may exist to inform the need for
additional investigation and/or studies prior to performing the modeling. The
minimum data requirements to develop an efficient and reliable slope stability
model for an embankment dam are summarized in Table 5-5, along with typical

data sources where the data needs can be found. These minimum data requirements are
discussed further below.
Table 5-5: Minimum Data Requirements for Slope Stability Modeling
Data Category Data Requirements Typical Data Sources
Model Geometry • Embankment geometry and internal zoning.
• Foundation contact and stratigraphy. • Topographic or Lidar Surveys
• As-Built Construction Drawings
• Design Drawings
• Design or Construction Reports
• Geologic and Geotechnical Investigation Reports
Material Properties • Unit weights and shear strengths for embankment and
foundation materials. • Geotechnical Investigation and Data Reports
• Construction Reports
• Published Data
Phreatic Surface • Pore water pressures in embankment and foundation. •
Piezometer Data / Piezometric Measurements
• Seepage Analyses
5.4.2.1 Model Geometry
Slope stability models are typically developed for one or more embankment cross
sections along the dam alignment. To develop the model geometry, there must be
sufficient available data on the embankment geometry and internal zoning, as well
as the foundation contact and stratigraphy. A detailed model geometry will
delineate the various embankment zones (core, shells, filters, drains, etc.),
foundation layers, and any slope stabilization measures if applicable (berm,
buttress, slope flattening, etc.).
The best data source for defining the embankment geometry and internal zoning of an
existing embankment dam is typically as-built construction drawings. Ideally, the
external geometry (i.e., embankment crest, upstream and downstream slopes, and
downstream ground surface) should be defined by a recent topographic or lidar
survey. When construction drawings and recent surveys are not available, design
drawings and/or design or construction reports can be used for defining the model.
However, the modeler should be aware that the as-built and/or current embankment
condition may differ from the design condition. Information from geologic and
geotechnical investigation reports can also be valuable in verifying the internal
zoning and variations in the embankment materials. For a new embankment dam that
has yet to be constructed, the embankment geometry and internal zoning are
typically defined using design drawings. The design can then be adjusted
accordingly based on the results of the slope stability analysis.
The best data source for defining the foundation contact and stratigraphy is
typically geologic and geotechnical investigation reports. Identifying and modeling
weak seams within the foundation is important because weak seams will typically
control the critical slip surface.
5.4.2.2 Material Properties
Slope stability modeling requires assigning material properties to the embankment
and foundation materials. These material properties include unit weight and shear
strength. Similar to the model geometry, there must be sufficient available data on
the embankment and foundation materials to estimate the material properties.
Typically, a comprehensive material characterization of the embankment and
foundation materials is performed prior to the slope

stability analysis to estimate the stability properties using data collected from
geotechnical investigations and/or published data. Information on the materials may
also be available in construction reports.
There are three types of unit weights: dry unit weight, total (or moist) unit
weight, and saturated unit weight. Saturated unit weights are assigned to materials
below the phreatic surface, while total unit weights are assigned to materials
above the phreatic surface. Often, total unit weights are assigned to all materials
in a slope stability model for simplification because stability analyses are not
highly sensitive to unit weight, as discussed further in Section 5.4.4.1. Dry unit
weights are never assigned to materials in a slope stability model because it is
uncommon for soils to be completely dry.
Total and saturated unit weights can be estimated from moisture content and dry
unit weight laboratory tests, Proctor compaction laboratory tests, published tables
of values, or empirical correlations (typically with blow counts).
Shear strength is the single most influential factor in a slope stability analysis
aside from embankment geometry, yet it is also the most complex to characterize.
After the loading conditions are selected (Section 5.4.1.2), appropriate shear
strengths must be estimated for the embankment and foundation materials. Shear
strengths can be estimated from laboratory tests, field tests, and/or various
empirical correlations. It is unrealistic to obtain samples that represent the
entire range of materials in the field, and soils will behave differently in the
laboratory than in the field. Therefore, experience and engineering judgment play a
major role in shear strength selection. Good practice is to perform sensitivity
analyses for a potential range of shear strengths to compensate for uncertainty.
Characterizing the shear strength of soils is dependent on both the type of soil
and whether the soil displays drained or undrained behavior under a particular
loading condition. Provided below is a description of the shear strengths typically
evaluated for coarse-grained, cohesionless soils (sands and gravels) and fine-
grained cohesive soils (clays and silts), including the corresponding laboratory
and field testing to measure strengths and empirical correlations to estimate
strengths. The description below focuses on the characterization of drained and
undrained shear strengths and was adapted from France et al. 2015. For more
information on shear strength characterization, refer to additional references in
Section 7 (France et al. 2015; USACE 2003). The characterization of residual shear
strengths for potentially liquefiable (sand- like) or cyclic softened (clay-like)
soils is beyond the scope of this guidance document. For information on the
evaluation of residual shear strengths, refer to additional references in Section 7
(Idriss and Boulanger 2008; HDR Engineering 2020).
Coarse-Grained Soils (Sands and Gravels)
Coarse-grained, or granular, soils (sands and gravels) are typically free-draining
and defined by drained shear strengths, except for very rapid loading (e.g.,
seismic loading, which is beyond the scope of this guidance document). Coarse-
grained soils have relatively high permeabilities and sufficient drainage capacity
to prevent pore water pressures from changing under most loading conditions.
Characterizing the drained shear strength of coarse-grained soils involves
evaluating or estimating the effective stress friction angle (’). The Mohr-Coulomb
shear strength envelope for coarse-grained soils goes through the origin of stress,
as illustrated in Figure 5-6, and thus the effective stress cohesion (c’) is zero.
Coarse-grained soils are therefore also often referred to as cohesionless soils.
Although the effective stress cohesion is zero, the strength envelope is often
curved, as illustrated in Figure 5-7 for the dense soil example. For mathematical
simplicity, the

strength envelope may be approximated as linear over the normal stress range of
interest for the analysis, which may result in an “apparent” effective stress
cohesion, as shown in
Figure 5-7. It is important to understand that this is a mathematical convenience,
and not a true property of the soil.

Figure 5-6: Mohr-Coulomb Shear Strength Envelope for Coarse-Grained Soils (Duncan
et al. 2014)

Figure 5-7: Curved Shear Strength Envelope with Linear Interpretation and Apparent
Cohesion (c’) (AECOM 2015)

Typical factors affecting values of ’ for coarse-grained soils include relative

density, confining pressure, angularity, and gradation. Values of ’ increase as
the soil relative density increases, confining pressures decrease, particle
angularity increases, and the soil gradation becomes broader (i.e., a wider range
of particle sizes are included).
Laboratory tests used to measure values of ’ for coarse-grained soils include the
consolidated- drained (CD) triaxial shear test, consolidated-undrained (CU’)
triaxial shear test with pore pressure measurements, and the direct shear (DS)
test. It is difficult, however, to obtain undisturbed samples of in-place granular
soils or reconstitute the structure of the natural deposits. Laboratory tests are
often used to estimate ’ for coarse-grained soils that will be placed during
construction, such as embankment soils, while empirical correlations are typically
used to select strengths for in-place cohesionless soils, such as foundations or
existing embankments, as discussed further below. Even when laboratory tests are
completed, the

results should be checked against expected values based on relative density,

gradation, and/or blow count data. The presence of large particles (e.g., scalped
samples or rockfill) may make laboratory test results misleading or impractical.
Field tests, including the standard penetration test (SPT), cone penetrometer test
(CPT), and shear wave velocity measurements, are often used to estimate values of
’ for in-place, coarse- grained soils through the application of empirical
correlations. Values of ’ for in-place coarse- grained soils can also be estimated
using empirical correlations to relative density and confining pressure. Common
empirical shear strength correlations are presented in Duncan et al. (2014), which
is one of the more comprehensive references for shear strength characterization.
Fine-Grained Soils (Clays and Silts)
Fine-grained soils (clays and silts) are generally defined by undrained shear
strengths for short- term loading conditions and drained shear strengths for long-
term loading conditions. These soils generally have low permeabilities and can
develop excess pore water pressures during some static loading conditions. Given
adequate drainage and time, pore water pressures will eventually dissipate in fine-
grained soils.
Clays – Characterizing the shear strength of clays is complex and can be quite
different for the various loading conditions, as discussed in Section 5.4.1.2. The
drained shear strength of clays can be expressed in terms of effective stress (’,
c’) strength parameters. The undrained shear strength can be expressed in terms of
total stress (, c) strength parameters or in terms of undrained strength (Su).
Different forms of characterization can be used for Su, for example, constant Su,
Su as a function of effective confining stress, and Su as a function of depth.
The overconsolidation ratio (OCR) of clays also has an impact on shear strength and
is defined as the ratio of the maximum pre-consolidation pressure of a soil mass to
the current consolidation pressure of that soil mass. For normally consolidated to
lightly overconsolidated clays, both undrained and drained shear strengths are of
interest. When normally consolidated to lightly overconsolidated clays are loaded
in shear, they tend to compress and generate positive pore water pressures
(contract), thereby resulting in an undrained shear strength that is less than the
drained shear strength. Hence, the lower undrained shear strength must be used when
analyzing undrained loading conditions. In contrast, for heavily overconsolidated
clays, drained shear strengths are of most interest because these clays tend to
expand (dilate) when loaded in shear and therefore generate negative pore water
pressures when sheared undrained. The negative pore water pressures result in an
undrained shear strength that is greater than the drained shear strength. As
discussed in Section 5.4.1.2 with respect to the rapid drawdown loading condition,
higher strengths resulting from negative pore water pressures are not normally used
in stability analyses because the negative pore water pressures cannot be relied
upon in the field. For heavily overconsolidated soils subjected to undrained
loading, the drained shear strength parameters are used, practically capping the
undrained strength as being no greater than the drained strength.
A typical drained Mohr-Coulomb shear strength envelope for clays is presented in
Figure 5-8. For normally consolidated clays, the Mohr-Coulomb strength envelope
goes through the origin of stress and the effective stress cohesion (c’) is equal
to zero. For overconsolidated clays, the Mohr-Coulomb strength envelope is
generally curved in the low stress range but still goes through the origin, such
that the effective stress cohesion is equal to zero. Similar to coarse- grained
soils, the strength envelope may be approximated as linear over the normal stress
range of interest for the analysis, which may result in an “apparent” effective
stress cohesion. Again, the strength envelopes with intercepts (shown in Figure 5-
8) are a mathematical convenience.

The higher strengths depicted in the lower stress ranges for the overconsolidated
clays in Figure 5-8 are the peak strengths for these soils. In most cases, the
drained stress-strain behavior for overconsolidated clays exhibits a pronounced
peak, followed by a drop to a much lower post-peak, remolded, or fully softened
strength. This reduction to post-peak strength occurs at relatively modest strains.
The remolded or fully softened strength is the same strength exhibited by a
normally consolidated specimen of the same clay. For the reasons cited below for
stiff-fissured clay, as well as to guard against progressive failure, it is
recommended that for drained loading, the fully softened strength be used for
saturated clays in the embankment and foundation.

Figure 5-8: Drained Mohr-Coulomb Shear Strength Envelop for Clays (Duncan et al.
2014)

Drained Mohr-Coulomb shear strength envelopes for stiff-fissured clays are

presented in Figure 5-9. Stiff-fissured clays are heavily overconsolidated clays
that are typically stiff and contain fissures. These clays can exhibit peak, fully
softened, and residual drained shear strengths. An undisturbed clay specimen tested
in the field or laboratory will exhibit a peak drained shear strength around a
shear displacement of 0.1 to 0.25 inch, or 3 to 6 millimeters (Duncan et al.,
2014). As displacement continues beyond the peak, the shearing resistance of the
specimen decreases to a residual value (at a displacement of about 10 inches, or
250 millimeters), as shown in Figure 5-9.
A slickensided surface along the failure plane is generally formed when the
specimen reaches its residual shear strength. The clay will exhibit a fully
softened drained shear strength if the same specimen is remolded and mixed with
enough water to raise its water content to the liquid limit. The fully softened
strength is lower than the undisturbed peak strength. As the remolded specimen is
sheared and displacement continues beyond the fully softened strength, the shearing
resistance of the specimen decreases to the same residual value as the undisturbed
specimen. The undisturbed peak shear strength is generally not used to evaluate the
stability of slopes composed of stiff-fissured clays. Shear strengths that can be
mobilized in the field are generally less than in the laboratory since more
softening and swelling occurs in stiff-fissured clays in the field over long
periods of time. The fully softened shear strength is generally more appropriate to
account for swelling and softening of the clay. However, if a failure has occurred
and a slickensided failure surface has developed, only the residual shear strength
can be mobilized to resist sliding and should be used in stability analyses.

Figure 5-9: Drained Shear Strength of Stiff-Fissured Clays (Duncan et al. 2014)

Other factors affecting clay shear strengths include anisotropy and strain rate.
The undrained shear strength of clays varies with the orientation of the principal
stresses at failure and with the orientation of the failure plane. Figure 5-10(a)
illustrates principal stress orientations at failure around a shear surface (i.e.,
slip surface). The undrained shear strength varies along the shear surface. Figure
5-10(b) shows variations in undrained shear strengths with orientation of the
applied stress from unconsolidated-undrained (UU) triaxial tests on two normally
consolidated clays and two heavily overconsolidated clay shales. Furthermore,
undrained shear strengths evaluated through laboratory testing can sometimes be
overestimated due to higher strain rates used to fail the specimen compared to
those in the field.

Figure 5-10: Anisotropy Effects for Clays – (a) Stress Orientations at Failure and
(b) Undrained Shear Strength Anisotropy of Clays and Shales – UU Triaxial Tests
(Duncan et al. 2014)

Laboratory tests used to measure the undrained shear strength (, c, or Su) of
clays include the unconfined compression (UC) test (Su), UU triaxial shear test (,
c, or Su), CU’/CU triaxial shear test with or without pore pressure measurements
(, c, or Su), and the direct simple shear (DSS) test (Su). Sample disturbance can
reduce the undrained shear strength measured in laboratory tests. This effect may
be reduced if the sample is consolidated to the same confining pressure it was
consolidated to in the field. The SHANSEP (Stress History and Normalized Soil
Engineering Properties) method can also be used to compensate for sample
disturbance and is a common approach used to estimate the undrained shear strength
of clays. As described by Ladd and Foot (1974) and Ladd et al. (1977), the method
involves consolidating clay samples to effective stresses that are greater than the
in-situ stresses and interpreting measured strengths in terms of an undrained shear
strength ratio (Su/’v). Figure 5-11 shows the variation of Su/’v with OCR for six
clays.

Figure 5-11: Variation of Su/’v with OCR for Clays, measured in Anisotropically
Consolidated DSS Tests (Duncan et al. 2014)
The equation used to evaluate Su/’v for clays that normalize under the SHANSEP
procedure is expressed by:

𝑆𝑢
𝜎′𝑣

= 𝑆(𝑂𝐶𝑅)𝑚

Where: ’v = Effective Vertical Consolidation Pressure before Load Is Applied

S = Undrained Strength Ratio for Clay in Normally Consolidated Condition OCR =
Overconsolidation Ratio
m = Empirical Exponent
In the absence of site-specific data for the clay, Jamiolkowski et al. (1985)
proposed that a value of 0.23 could be used for S and 0.8 for m. Shear strength
ratios developed using typical CU’ triaxial tests (where the samples are not
consolidated to effective stresses that are greater than the in-situ stresses and
backed off to known OCR values) will result in linear strength envelopes for clay
samples that are normally consolidated at the time of shearing and curved envelops
for clay samples that are overconsolidated at the time of shearing. Again, strength
envelopes with intercepts are a mathematical convenience. It is important to note
that undrained strength assignments using undrained shear strength ratios should be
made using the steady-state effective stresses before the load is applied. This can
be accomplished using a staged analysis approach, where the total stress analysis
is completed for the second stage applied loads and the undrained strength of the
material is calculated based on stresses in the first stage (before

loading). As discussed in Section 5.4.1.2, the analysis procedure for the rapid
drawdown loading condition by Duncan et al. (1990) uses this approach to assign
undrained shear strengths.
A field test used for direct measurement of Su for clays is the vane shear test,
which has been successfully used for measuring the undrained shear strength of soft
to medium-stiff clays.
Limitations of the vane shear test are that it can be affected by sand lenses and
seams, and the raw undrained shear strength measured from the test requires an
empirical correction factor that varies with plasticity index and accounts for
anisotropy and strain rate effects. The data that provide the basis for the
correction factor are widely scattered, and therefore, vane strengths should not be
viewed as precise. The pocket penetrometer test and Torvane test can be used to
obtain quick, approximate measurements of undrained shear strength in the field or
laboratory. However, the pocket penetrometer and Torvane tests are relatively crude
and should be considered as only rough indications of shear strength.
The laboratory test most commonly used to measure the drained shear strength (’,
c’) of clays is the CU’ triaxial shear test. The CU’ triaxial test is more
practical than the CD triaxial test because the strain rates required for a CD test
are typically extremely slow, requiring an impractically long test time. In
addition, the CU’ test can be used to obtain both undrained (total stress) and
drained (effective stress) shear strength parameters. However, the CD triaxial test
and DS test can also be used if the long test times can be accommodated. For stiff-
fissured clays, laboratory tests are performed on remolded specimens to evaluate
fully softened and/or residual drained shear strengths. The DS test is commonly
used to measure the fully softened shear strength of stiff-fissured clays.
Torsional ring shear tests are most suitable for estimating the residual shear
strength of stiff-fissured clays because the test can measure shear stresses over
any magnitude of displacement through continuous rotation. For more information on
laboratory and field shear strength tests, refer to additional references in
Section 7 (AECOM 2014).
Laboratory tests provide the best strength data for clays, and reasonably
undisturbed samples of these soils can typically be obtained. However, various
empirical correlations developed to estimate strengths for clays may be sufficient
in some cases. It is recommended that these methods and correlations be used with
caution because the behavior, and hence strength characterization, for clays is
typically complex and may not be appropriately captured by the correlations. A few
of the more common empirical shear strength correlations are presented in Duncan et
al. (2014). It may also be useful to use correlations as a check to validate the
results of laboratory tests.
Silts – Silts have an interesting soil particle composition, as they can behave
similar to either fine sands or clays. When the term “fine-grained” is used, it
almost always includes silts in this category. But silts themselves can be divided
into two general categories: non-plastic and plastic. Non-plastic silts behave more
like fine sands, while plastic silts behave more like clays. Evaluating whether a
unit of silt is non-plastic or plastic can be achieved using the Atterberg Limit
laboratory test.
Since silts can have a wide range of permeabilities, it can be difficult to predict
if these soils will display drained or undrained behavior under various loading
conditions. It is common to characterize silts using both drained and undrained
shear strengths, similar to clays. For drained conditions, the shear strength of
silts can be characterized by an effective stress friction angle (’) with an
assumed effective stress cohesion (c’) equal to zero. For undrained conditions, the
shear strength of silts can be expressed using total stress strength parameters (,
c) or in terms of undrained strength (Su). Similar to clays, there are different
forms of

characterization that can be used for Su, for example, constant Su, Su as a
function of effective confining stress, and Su as a function of depth.
Laboratory tests used to measure values of ’ for silts include the CD and CU’
triaxial shear tests and the DS test. Laboratory tests used to measure the
undrained shear strength (, c, or Su) of silts include the UC test (Su), UU and
CU’/CU triaxial shear tests (, c, or Su), and the DSS test (Su). Similar to
coarse-grained soils, it can be difficult to obtain quality undisturbed samples of
silts in the field, particularly non-plastic or very low plasticity silts.
Strength behaviors of silts have not been as widely studied as those of sands or
clays. As a result of this general lack of research and compilation of data, very
few empirical correlations exist for predicting shear strength values for silts.
Empirical shear strength correlations that are available for silts are often
regionally specific and developed with relatively limited data sets. It would be
prudent to incorporate a level of conservatism when using these correlations for
silts.
Similar to sands and clays, SPT, CPT, and shear wave field tests can be used for
empirical shear strength correlations of silts. Empirical correlations using
results of field tests for sands can generally be applied to estimate shear
strengths of non-plastic silts. Shear strengths of plastic silts can generally be
estimated from empirical correlations using results of field tests for clays.
5.4.2.3 Phreatic Surface
In addition to material properties, slope stability modeling requires inputting a
phreatic surface through the embankment dam, which involves an evaluation of pore
water pressures through the embankment and foundation. The evaluated phreatic
surface is typically associated with the steady-state loading condition. In cases
where the pore pressures vary significantly between the embankment and foundation,
it may be acceptable to apply two separate piezometric surfaces, one assigned to
the embankment (phreatic surface) and one assigned to the foundation
(potentiometric surface). Because instability in embankment dams is often preceded
by seepage problems, it is essential to understand and capture the seepage
conditions occurring through the embankment and foundation in the stability model.
Pore water pressures and the resulting phreatic surface through the embankment can
be estimated from piezometer and/or monitoring well data or seepage analyses.
When piezometer and/or monitoring well data are used to estimate pore water
pressures and the resulting phreatic surface, it is beneficial to plot the
available piezometric measurements with measurements of reservoir level, as shown
in Figure 5-12, to evaluate the fluctuation in piezometric water levels with
reservoir filling and drawdown cycles. The phreatic surface is typically estimated
using the maximum water levels recorded in the piezometers, which should generally
correspond to the maximum normal reservoir level. It may also be helpful to plot
the approximate phreatic surface (and potentiometric surface if present) on a cross
section of the embankment before evaluating the stability model to guide critical
thinking about how the embankment may perform.

Figure 5-12: Typical Piezometer Data Plot

For irrigation reservoirs that experience fluctuations in water levels annually,

the piezometric response to reservoir level must be carefully evaluated before
assuming a steady-state phreatic surface. Typically, there is a lag between
reservoir water level fluctuations and piezometer response. When modeling
irrigation reservoirs, carefully look at lags in piezometric responses as the
reservoir fills, reaches full pool, and starts to draw down. A 2-week lag is
common. If the reservoir is not at full pool at least this long, assuming a steady-
state phreatic surface may not be fully representative but is usually conservative.
When seepage analyses are used to estimate pore water pressures and the resulting
phreatic surface, the results can be directly input into the stability model.
Seepage analyses are discussed further in Section 3 of this guidance document.
Seepage analyses for slope stability modeling are typically only performed when
current, sufficient, and reliable piezometer data are not available to evaluate
seepage response in existing embankment dams, or for new embankment dams in the
design process or recently constructed dams for which steady-state conditions with
sufficient piezometric measurements have not yet been developed.
For cases when piezometer data are not available or when a seepage analysis is too
costly to perform, it may be appropriate to approximate the phreatic surface based
on engineering judgment and experience with similar embankment configurations. For
examples on how to draw typical steady state phreatic surfaces, see Section
3.2.1.2, and for additional reading, see Section 7 (Cedergren 1989; NRCS 1979; NRCS
1973). Sensitivity analyses can also be performed for a potential range of pore
pressures to compensate for uncertainty.
5.4.2.4 Desktop Review
A desktop review of all available information for an embankment dam should be
completed to determine whether there are sufficient data on the embankment
geometry, internal zoning, foundation contact and stratigraphy, and material
properties to perform the slope stability

analysis. The desktop review can also identify potential data gaps. FEMA (2015b)
presents the following checklist for guidance on conducting a desktop review:

If data gaps are identified, additional investigations or studies may be warranted.

For information on performing geotechnical investigations in embankments, including
typical drilling and sampling methods, refer to additional references in Section 7
(AECOM 2014a, 2014b, 2014c, 2015; FEMA 2015b).
5.4.3 Model Slip Surface Considerations
In a limit equilibrium slope stability model, the factor of safety is calculated
for a trial slip surface and is assumed to be constant along the slip surface. A
number of trial slip surfaces must be evaluated to identify the critical slip
surface, which is the surface resulting in the lowest (or minimum) calculated
factor of safety. Critical considerations for evaluating trial slip surfaces in a
slope stability model include the slip surface configuration and location, as well
as tension cracks. These considerations are discussed further below.
5.4.3.1 Slip Surface Configuration and Location
Evaluating the configuration (or shape) and location of the critical slip surface
corresponding to the minimum factor of safety is one of the greatest challenges in
slope stability modeling and requires a trial procedure. The slip surface
configuration and location will depend on the embankment geometry and internal
zoning, foundation stratigraphy, material characteristics (including shear strength
and anisotropy), and pore water pressure conditions.
Two common slip surface configurations for slope stability modeling are circular
and noncircular slip surfaces, as shown in Figure 5-13. Circular slip surfaces are
defined by an arc of a circle that cuts through the slope and are typically
applicable for evaluating relatively homogeneous or zoned embankments founded on
thick soil deposits. Noncircular slip surfaces are defined by connected linear
segments that cut through the slope and are typically more applicable for
evaluating zoned embankments on layered foundations, particularly with weak seams.
A wedge slip surface is a type of noncircular surface defined by three linear
segments and may be appropriate for evaluating a relatively long, horizontal weak
seam bounded by stronger materials. Due to modern computing capabilities, stability
modeling using noncircular slip surfaces is very common and useful for complex
geometries.

Figure 5-13: Common Slip Surface Configurations (USACE 2003)

The critical slip surface location is related to zones of relatively weaker

materials and/or higher pore water pressures. Automatic search procedures used in
computer programs can be performed for both circular and noncircular slip surfaces
and aid in locating the critical slip surface corresponding to the minimum factor
of safety. However, considerable judgment must be exercised to ensure that the most
critical slip surface has been located. Details of the search procedures vary among
computer programs and should be understood prior to the analysis.
Most search procedures require an initial estimate of the starting location of the
slip surface. For circular slip surfaces, the starting location is typically
defined by the x and y coordinates of the center of the circle and the radius of
the circle. For noncircular slip surfaces, the starting location is defined by the
x and y coordinates for each point on the slip surface.
Provided below are tips for consideration.
• Tip 1 – Conduct automatic searches for trial slip surfaces at several
starting locations to fully explore the soil profile and detect multiple local
minimums. Multiple trial slip surfaces can be evaluated to verify that the global
minimum factor of safety has been found.

Consider evaluating slip surfaces that pass through the embankment only and also
those that pass through both the embankment and foundation.
• Tip 2 – Estimate the starting location of noncircular slip surfaces by first
evaluating circular slip surfaces or by identifying locations of weak materials or
seams.
• Tip 3 – Evaluate a range of local, intermediate, and global slip surfaces, as
shown in Figure 5-14. The factor of safety can vary between local slip surfaces
that pass through a limited portion of the embankment slope or toe area and global
slip surfaces that pass through the embankment crest and encompass the full slope
and toe area.
• Tip 4 – Examine slip surfaces and factors of safety that are not minimums.
The slip surface with the minimum factor of safety is not always the one of
greatest interest. Often, shallow infinite slip surfaces have a lower factor of
safety than deeper slip surfaces but are not considered to have a global stability
impact or impact the safety of dams. Failure along a shallow surface may consist of
material raveling downslope and presenting merely a maintenance issue, whereas
failure along a deeper surface can have more severe consequences.

Figure 5-14: Examples of Local, Intermediate, and Global Slip Surfaces for Slope
Stability Modeling

5.4.3.2 Tension Cracks

Steep slip surface angles at the entrance and exit of the evaluated slope can cause
tensile forces to develop at the interfaces between slices and on the bottom of
slices at the entrance and exit areas. These tensile forces can sometimes cause
model convergence issues or numerical problems in calculating the factor of safety.
As discussed in Section 5.1.1, soils do not generally exhibit significant tensile
shear strength and thus cannot withstand tension. Therefore, the presence of
tension in slope stability models is considered unrealistic. Tension can imply
tensile strength that does not exist and will result in a calculated factor of
safety that is too high. The adverse effects of tension can be eliminated from the
calculations of stability modeling by introducing a tension crack. A tension crack
terminates the slip surface at the edge of a slice at

an appropriate depth below the ground surface, as shown in Figure 5-15. The depth
of a tension crack can be estimated from simple equations derived from earth
pressure theory. In most cases, only an approximation of the depth of tension crack
is needed. Typically, a tension crack is introduced into the slope stability model,
and the depth is varied until convergence is satisfied. For more information on
tension cracks and calculating the depth of a tension crack, refer to additional
references in Section 7 (Duncan et al. 2014).

Figure 5-15: Slope and Slip Surface with Tension Crack (Duncan et al. 2014)

5.4.4 Model Sensitivity Considerations

Slope stability modeling is influenced by several factors, including variations in
unit weight, shear strength, and phreatic surface conditions. Modelers should
understand which factors have greater impacts on the results of slope stability
analyses so that more effort can be spent estimating the most influential factors.
The sensitivities of variations in unit weight, shear strength, and phreatic
surface conditions were evaluated by France and Winckler (2010) and are discussed
below.
5.4.4.1 Unit Weight and Shear Strength
The effects of variations in unit weight and shear strength on the results of slope
stability analyses were evaluated using two example stability models presented in
Figure 5-16 and Figure 5-17 for a base case condition and sensitivity case. The
base case condition used the materials properties (i.e., unit weights and drained
effective stress shear strengths) shown in Figure 5-16 and Figure 5-17 and a
phreatic surface estimated using an anisotropic permeability ratio (i.e., ratio of
horizontal to vertical permeability) of 4(kh):1(kv). The slope stability results
for the two examples under the base case condition are presented in Figure 5-18 and
Figure 5-19, which indicate calculated minimum factors of safety for a global
(i.e., deep-seated) slip surface of 1.91 and 1.66, respectively.

Figure 5-16: Example 1 – Slope Stability Model for Homogeneous Embankment with Toe
Drain (France and Winckler 2010)

Figure 5-17: Example 2 – Slope Stability Model for Zoned Embankment with Chimney
and Blanket Drains (France and Winckler 2010)
Figure 5-18: Slope Stability Model Results for Example 1, Base Case Condition
(France and Winckler 2010)

Figure 5-19: Slope Stability Model Results for Example 2, Base Case Condition
(France and Winckler 2010)

The sensitivity case varied the unit weights and shear strengths of all the
materials independently by ±5%, ±10%, and ±20%. The variations in strengths were
applied to both the tangent of the effective stress friction angle (tan’) and the
effective stress cohesion (c’). The slope stability model results for the two
examples under the sensitivity case incorporating independent variations in unit
weight and shear strength are summarized in Table 5-6 and Table 5-7, respectively.

Table 5-6: Slope Stability Model Results for Sensitivity Case – Variations in Unit
Weight (France and Winckler 2010)

Variation in Unit Weight Minimum Calculated Factor of Safety

Example 1 Example 2
-20% 1.84 1.62
-10% 1.88 1.64
-5% 1.89 1.65
Base Case 1.91 1.66
+5% 1.92 1.66
+10% 1.93 1.67
+20% 1.95 1.67

Table 5-7: Slope Stability Model Results for Sensitivity Case – Variations in
Strength (France and Winckler 2010)

Variation in Shear Strength Minimum Calculated Factor of Safety

Example 1 Example 2
-20% 1.53 1.33
-10% 1.72 1.49
-5% 1.81 1.58
Base Case 1.91 1.66
+5% 2.00 1.74
+10% 2.10 1.82
+20% 2.29 1.99

Based on the results, the following observations were noted:

• For variations in shear strength, changes in the minimum factors of safety
are approximately proportional to changes in strength.
• For variations in unit weight, changes in the minimum factors of safety are
much less than proportional to changes in unit weight.
Therefore, the results of slope stability analyses are significantly more sensitive
to variations in shear strength than unit weight, and more time should be spent
estimating the strengths of the materials. It may also be important to approach the
selection of shear strengths for stability analyses in a conservative manner.
Alternatively, sensitivity analyses can be performed for a potential range of shear
strengths to compensate for uncertainty.
5.4.4.2 Phreatic Surface
Similar to the unit weight and shear strength, the effect of variations in the
phreatic surface (i.e., internal water levels) on the results of slope stability
analyses were evaluated using the example stability model presented in Figure 5-16
above. Phreatic surfaces were estimated using increasing anisotropy ratios (i.e.,
ratios of horizontal to vertical permeability) of 1, 4, and 9. The phreatic
surfaces corresponding to these three anisotropy ratios are illustrated in Figure
5-20.

The slope stability model results for the example under variations in the phreatic
surface are summarized in Table 5-8. The results indicate that the factor of safety
significantly decreases under higher phreatic surfaces associated with increasing
anisotropy ratios. Therefore, the results of slope stability analyses are sensitive
to variations in the phreatic surface, and time should be spent estimating the
phreatic surface from available piezometer data or seepage analyses. It may also be
important to approach the estimation of the phreatic surface for stability analyses
in a conservative manner. Alternatively, sensitivity analyses can be performed for
a potential range of pore water pressures (i.e., phreatic surface conditions) to
compensate for uncertainty and/or a piezometer threshold analysis can be performed,
as discussed in Section 5.4.1.5.

Figure 5-20: Phreatic Surfaces Corresponding to Increasing Anisotropy Ratios and

Slope Stability Model Results for Variations in Phreatic Surface (France and
Winckler 2010)

Table 5-8: Summary of Slope Stability Model Results for Variations in Phreatic
Surface (France and Winckler 2010)

Anisotropy Ratio, kh/kv Minimum Calculated Factor of Safety

1 1.98
4 1.91
9 1.71

5.4.5 Model Convergence Considerations

For limit equilibrium slope stability analyses, model convergence issues occur when
the model cannot obtain moment or force equilibrium. Convergence issues are
commonly due to high shear strength contrasts between zones, sharp corners in the
material or slip surface geometry, and point loads. Starting simple is a good way
to overcome convergence issues. There is no one solution that fits all situations
of non-convergence. However, the following tips may be helpful:

• Tip 1 – Simplify the model if it makes sense to do so.

• Tip 2 – Reduce high shear strength contrasts.
• Tip 3 – Remove sharp corners.
• Tip 4 – Distribute point loads.
• Tip 5 – Try different slip surface shapes and search methods.
If convergence issues do occur, an error message will typically display, or the
model will not run. However, on very rare occasions, the authors have found that
some models will run with convergence issues. Red flag warnings include an
unreasonably low factor of safety and lack of slip surfaces propagating through the
weakest material.
5.5 Interpreting, Verifying, and Reporting Results
Although not as complex as seepage modeling, slope stability modeling is still
complex and requires practice and experience. Models will often run without error
and produce professional- looking results but can be invalid. Successful runs
should not be confused with accurate results. Interpreting and verifying the
results are not always intuitive to the novice user. Vetting errors and
understanding the sensitivity of a model to the potential range of each input
parameter should be a priority before using the results. This often requires
knowledge gained through personal trial and error. Consider seeking guidance from
experienced engineers and modelers. If not available in-house, program vendors and
developers often offer limited technical support for vetting specific problems.
Specific situations may warrant hiring an expert to perform the slope stability
modeling. It is easier to create models for the steady-state loading condition than
other conditions, but most embankments require an evaluation of all loading
conditions. Shear strengths need to be characterized differently for the various
loading conditions. Shear strength is the single most influential parameter in a
slope stability analysis, yet also the most complex to characterize, especially for
undrained strengths. Some specific situations that may warrant hiring an expert are
as follows:
• Transient conditions,
• Complex model geometry with high shear strength contrast,
• Material strength variability within the core or foundation,
• High consequences (i.e., model is used to direct a significant design
decision), or
• Modeling reinforcements or point loads.
There may also be merit in getting a secondary review from an expert, as some
regulatory agencies may not have the modeling experience to catch problems.
The following sections provide tips, tools, and guidance on how to interpret and
verify slope stability analysis results, as well as report the results.
5.5.1 Interpreting and Verifying Results
After the slope stability model is run and results are obtained, it is important to
examine the calculated minimum factor of safety and corresponding critical slip
surface to understand and verify the model results. Tips for checking the validity
of the model results include the following:

• Research the computer program being used for stability modeling to understand
the theory, methodology, assumptions, and weaknesses inherent to the program.
Having adequate knowledge of the program will allow for better interpretation of
the results.
• Plot multiple trial slip surfaces together on the analyzed embankment cross
section to verify the critical slip surface is reasonable. The configuration and
location of the critical slip surface is dependent on zones of weaker materials.
For a zoned embankment with a core material that is weaker than the shell material,
the critical slip surface is likely to pass through the core. For an embankment on
a layered foundation with a weak seam, the critical slip surface is likely to be a
noncircular surface that passes through the weak seam.
• Compare the results to past performance of the dam. If the calculated minimum
factor of safety for the steady-state loading condition is less than 1.0 but the
embankment has historically performed well without stability issues, the results
may be invalid and may warrant further evaluation of model inputs, such as material
properties (e.g., shear strengths) and/or the phreatic surface.
• Use a second computer program, slope stability charts, and/or detailed hand
calculations to verify similar results.
• Perform sensitivity analyses to understand the sensitivity of the model to a
potential range of each input parameter. Sensitivity analyses are also useful for
evaluating uncertainties in shear strengths, pore water pressures, etc.
An example slope stability model using the UTEXAS4 computer program is presented in
Figure 5-21 and is used here for providing additional guidance in interpreting
results. The
example model is for a homogeneous embankment dam founded on relatively strong
sandstone bedrock with two weak interbedded horizontal clay seams. The phreatic
surface through the embankment intersects with a toe drain at the downstream
portion of the embankment.

Figure 5-21: Example Slope Stability Model Using UTEXAS4 (France and Winckler 2010)
The validation and interpretation of the slope stability model results (shown in
Figure 5-22) are summarized as follows:
• Figure 5-22 illustrates the resulting factors of safety for trial slip
surfaces using three different search methods: (1) circular search, (2) noncircular
search through the upper weak clay seam, and (3) noncircular search through the
lower weak clay seam.
• The circular search resulted in an intermediate slip surface confined to the
embankment, with a calculated factor of safety of 1.78. The noncircular searches
resulted in global slip surfaces passing through each of the weak clay seams, with
significantly lower calculated factors of safety of 1.21 through the upper weak
seam and 1.11 through the lower weak seam. Thus, the critical slip surface
corresponding to the minimum calculated factor of safety is the global noncircular
slip surface that passes through the lower weak seam. This should be expected and
is reasonable since the lower clay seam is weaker than the upper clay seam, as well
as the other embankment and foundation materials. Furthermore, the noncircular slip
surfaces have more adverse pore water pressures compared to the circular slip
surface. The noncircular surfaces are completely below the phreatic surface, while
the circular surface is mostly above the phreatic surface.

Figure 5-22: Slope Stability Model Results Showing Factors of Safety for Trial Slip
Surfaces (France and Winckler 2010)

The slope stability model input and output (results) files should also be
critically reviewed to verify that the following conditions are true for the
critical slip surface (with spot checks for surfaces other than the critical
surface):
• There are no outstanding errors or warnings.
• Model geometry is correct.
• Materials are assigned the correct material properties.
• Pore pressure is calculated correctly.
• Load conditions are being executed correctly.
• Shear stress for each slice is greater than or equal to zero, and no shear
stress is unreasonably high (relative to the other slices). Shear stress should
never be negative.

• The summation of horizontal forces, vertical forces, and moments are

practically zero (i.e., summation will never be exactly zero).
• The slice calculations are using the correct material to estimate shear
strength.
5.5.2 Reporting Results
Upon completion of a slope stability analysis, the analysis should be adequately
documented in a report for submittal to the dam owner and/or regulator. The report
should include sufficient detail for the reader to understand the purpose of the
analysis, how the analysis was performed, the results of the analysis, and
recommendations. Slope stability analysis reports should include a discussion on
the following topics:
• Purpose – Clearly define the objective of the slope stability analysis. Refer
to Section 5.3.
• Methodology – Describe the stability analysis method and approach (e.g.,
loading conditions, factor of safety criteria, 2D or 3D, computer program used).
Refer to Section 5.4.1.
• Model Geometry – Summarize the embankment geometry and internal zoning and
the foundation stratigraphy modeled in the stability analysis. Refer to Section
5.4.2.1.
• Material Properties – Summarize the stability material properties (e.g., unit
weight and shear strength) assigned to the embankment and foundation materials and
explain how they were evaluated. Refer to Section 5.4.2.2.
• Pore Water Pressures – Summarize the pore water pressure conditions used in
the analysis and describe the source of the data (e.g. phreatic surface from
piezometer data, pore pressures imported from seepage analysis). Refer to Section
5.4.2.3.
• Results – Provide summary tables and figures of the results and summarize the
interpretation of the results. This may also include a discussion on the
uncertainties of the evaluation and the results of sensitivity analyses to evaluate
the uncertainties. Refer to Section 5.5.1.
• Recommendations – If applicable, provide recommendations for action (e.g.,
increased monitoring, reservoir storage restrictions, remediation) or additional
investigations based on the results. The results may also be used for alternative
design recommendations.

6. Conclusion
Seepage and slope stability modeling are often proposed as part of an embankment
dam evaluation. These analyses can be valuable to a project when the objective is
clearly defined, alternatives are considered, models are planned and executed
properly, and results are interpreted and reported correctly.
This guidance document describes the standard of practice for seepage and slope
stability analyses of embankment dams and provides tips, tools, and guidance on
planning, modeling considerations, and interpreting, verifying, and reporting the
results.
Basic seepage and slope stability concepts are summarized throughout the document
to give the reader an understanding of how best to approach performing efficient,
effective, and reliable seepage and stability models. References to additional
publications that elaborate on these concepts are included in Section 7.

7. References and Additional Reading

AECOM. “Embankment Dam Slope Stability 101,” Western Dam Engineering Technical
Note, Volume 1, Issue 3, November 2013 (sponsored by Colorado Division of Water
Resources, Montana Department of Natural Resources, and Wyoming State Engineer’s
Office).
http://dnrc.mt.gov/divisions/water/operations/docs/dam-safety/technical-
references/western_dam_engineering_technote-vol1issue3.pdf
AECOM. “Poking the Bear: Drilling and Sampling for Embankment Dams,” Western Dam
Engineering Technical Note, Volume 2, Issue 1, April 2014a (sponsored by Colorado
Division of Water Resources, Montana Department of Natural Resources, and Wyoming
State Engineer’s Office).
http://dnrc.mt.gov/divisions/water/operations/docs/dam-safety/technical-
references/western_dam_engineering_technote-vol2issue1.pdf
AECOM. “Soil Characterization (Part 1) – Here’s the Dirt,” Western Dam Engineering
Technical Note, Volume 2, Issue 2, July 2014b (sponsored by Colorado Division of
Water Resources, Montana Department of Natural Resources, and Wyoming State
Engineer’s Office).
http://dnrc.mt.gov/divisions/water/operations/docs/dam-safety/technical-
references/western_dam_engineering_technote-vol2issue2.pdf
AECOM. “Soil Characterization (Part 2) – Laboratory and Field Shear Strength
Testing,” Western Dam Engineering Technical Note, Volume 2, Issue 3, October 2014c
(sponsored by Colorado Division of Water Resources, Montana Department of Natural
Resources, and Wyoming State Engineer’s Office).
http://dnrc.mt.gov/divisions/water/operations/docs/dam-safety/technical-
references/western_dam_engineering_technotevol2issue3.pdf
AECOM. “Soil Characterization (Part 3) – Shear Strength Characterization for Slope
Stability Analyses,” Western Dam Engineering Technical Note, Volume 3, Issue 1,
February 2015 (sponsored by Colorado Division of Water Resources, Montana
Department of Natural Resources, and Wyoming State Engineer’s Office).
http://dnrc.mt.gov/divisions/water/operations/dam-safety/technical-references-and-
links/WesternDamEngineering_Issue01_Vol03_FINAL.2.26.2015.pdf
AECOM. “Internal Erosion: Issues Just Below the Surface,” Western Dam Engineering
Technical Note, Volume 4, Issue 2, August 2016 (sponsored by Colorado Division of
Water Resources, Montana Department of Natural Resources, and Wyoming State
Engineer’s Office).
http://dnrc.mt.gov/divisions/water/operations/docs/dam-safety/other-dam-safety-
references/western-dam-news-
letter/WesternDamEngineering_Issue02_Vol04_FINAL.PDF/view
Casagrande, A. "Seepage Through Dams,” Contribution to Soil Mechanics 1925-1940,
Boston Society of Civil Engineers: 295, 1937.

Cedergren, H.R. Seepage, Drainage and Flow Nets, Third Edition, John Wiley & Sons,
Inc., 1989.
Duncan, Michael J., Wright, Steven G., and Brandon, Thomas L. Soil Strength and
Slope Stability, Second Edition, John Wiley & Sons, Inc., 2014.
Duncan, Michael J., Wright, Steven G., and Wong, Kai S. “Slope Stability During
Rapid Drawdown,” Proceedings of H. Bolton Seed Memorial Symposium, Volume 2, 1990.
Federal Emergency Management Agency (FEMA). Earthquake analyses and Design of Dams,
Federal Guidelines for Dam Safety, FEMA P-65, May 2005.
https://www.fema.gov/sites/default/files/2020-08/fema_dam-safety_earthquake-
analysis_P-65.pdf
FEMA. Federal Guidelines for Dam Safety, Emergency Action Planning for Dams, FEMA
64, July 2013.
https://www.fema.gov/sites/default/files/2020-08/eap_federal_guidelines_fema_p-
64.pdf
FEMA. Evaluation and Monitoring of Seepage and Internal Erosion, Interagency
Committee on Dam Safety (ICODS), FEMA P-1032, May 2015a.
https://www.fema.gov/sites/default/files/2020-
08/fema_p1032_eval_monitoring_seepage_internal_erosion.pdf
FEMA. Training Aids for Dam Safety, Module Q: Evaluation of Seepage Conditions
(Second Edition), September 2015b.
FEMA. Pocket Safety Guide for Dams and Impoundments, FEMA P-911, October 2016.
https://www.fema.gov/sites/default/files/2020-
08/fema_911_pocket_safety_guide_dams_impoundments_2016.pdf
Fell, R., M.A. Foster, J. Cyganiewicz, G.L. Sills, N.D. Vroman, and R.R. Davidson.
2008. Risk Analysis for Dam Safety: A Unified Method for Estimating Probabilities
of Failure of Embankment Dams by Internal Erosion and Piping, URS Australia,
Sydney, Australia.
Foster, M., Fell, R., and Spannagle, M. “The statistics of embankment dam failures
and accidents,” Canadian Geotechnical Journal, Volume 37, Issue 5, pp. 1000-1024
October 1, 2000.
France, John W., Williams, Jennifer L., Winckler, Christina J.C., and Adams,
Tiffany E. “Soil Shear Strength Selection for Stability Analysis – Practical
Guidance,” Dam Safety 2015, National Conference of the Association of State Dam
Safety Officials (ASDSO). https://damsafety.org/content/soil-shear-strength-
selection-stability-analysis-practical- guidance
France, John W. and Winckler, Christina J.C. “Embankment Slope Stability Analysis –
What is Important and What is Not?, Dam Safety 2010, National Conference of the
Association of State Dam Safety Officials (ASDSO).
https://damsafety.org/content/embankment-slope-stability-analysis-what-important-
and- what-not

Fredlund, M.D. “Unsaturated Seepage Modeling Made Easy,” Geotechnical News, Bitech
Publishers, June 1998.
https://www.researchgate.net/publication/237451982_Unsaturated_Seepage_Modeling_
Made_Easy
GEO-SLOPE International Ltd (GEO-SLOPE). Seepage Modeling with SEEP/W, An
Engineering Methodology, November 2012 Edition.
http://downloads.geo- slope.com/geostudioresources/8/0/6/books/seep%20modeling.pdf?
v=8.0.7.6129
HDR Engineering. “Technical Note 5 – Simplified Seismic Analysis Procedure for
Montana Dams,” Montana Department of Natural Resources and Conservation (DNRC) Dam
Safety Technical Notes, November 30, 2020.
http://dnrc.mt.gov/divisions/water/operations/dam-safety/technical-notes
Heitland, Julie; Williams, Jennifer; and Wanninayake, Ajitha. “Seepage Models –
Tips, Tools & Guidance (From an Engineer’s Perspective,” Dam Safety 2020, National
Conference of the Association of State Dam Safety Officials (ASDSO).
https://damsafety.org/reference/seepage-models-%E2%80%93-tips-tools-guidance-
engineer%E2%80%99s-perspective
Idriss, I.M. and R.W. Boulanger. Soil Liquefaction During Earthquakes, Earthquake
Engineering Research Institute, Monograph MNO-12, Oakland, CA, 2008.
Jamiolkowski, M., Ladd, C.C., Germaine, J.T., Lancelotta, R., “New Developments in
Field and Laboratory Testing of Soils.” Proceedings of the Eleventh International
Conference on Soil Mechanics and Foundation Engineering, San Francisco, CA, 1985,
pp. 57-153.
Ladd, C.C. and Foot, R. “New design procedure for stability of soft clays, American
Society of Civil Engineers (ASCE) Journal of the Geotechnical Engineering Division,
Volume 100, No. GT7, pp. 763-786, 1974.
Ladd, C.C., Foot, R., Ishihara, K., Schlosser, F., and Poulos, H.G. “Stress-
deformation and strength characteristics, International Conference on Soil
Mechanics and Foundation Engineering, Volume 2, pp. 421-494, 1977.
Lemieux, Michele. “Dam Seepage Models – Tools, Rules & Guidance (From a Regulatory
Perspective,” Dam Safety 2020, National Conference of the Association of State Dam
Safety Officials (ASDSO).
https://damsafety.org/reference/dam-seepage-models-tools-rules-guidance-regulatory-
perspective
Richards, K.S. and Reddy, K.R. “Critical appraisal of piping phenomena in earth
dams,” Bulletin Engineering Geology Environment, Volume 66, pp. 381-402, 2007.
U.S. Army Corps of Engineers (USACE). Engineer Manual (EM) 1110-2-1901, Seepage
Analysis and Control for Dams, April 30, 1993.
https://www.publications.usace.army.mil/Portals/76/Publications/EngineerManuals/
EM_1 110-2-1901.pdf?ver=INPQyoyQDHXDtlSpfJ9fXw%3d%3d

USACE. Engineer Manual (EM) 1110-2-1902, Slope Stability, October 31, 2003.
https://www.publications.usace.army.mil/Portals/76/Publications/EngineerManuals/
EM_1 110-2-1902.pdf?ver=E1mrnP_5qHsVXEiXzeka-Q%3d%3d
U.S. Department of Agriculture, Natural Resources Conservation Service (NRCS). Soil
Mechanics Note No. 5, Flow Net Construction and Use, October 1973.
https://www.arcc.osmre.gov/resources/impoundments/USDA-SoilMechanicsNote5-
FlowNetConstructionandUse1973a.pdf
U.S. Department of Agriculture, NRCS. Soil Mechanics Note No. 7, The Mechanics of
Seepage Analyses, October 1979.
https://directives.sc.egov.usda.gov/OpenNonWebContent.aspx?content=18467.wba
U.S. Department of Agriculture, NRCS. Part 651, Agricultural Waste Management Filed
Handbook, “Chapter 10: Agricultural Waste Management System Component Design,”
August 2009.
U.S. Department of the Interior, Bureau of Reclamation (Reclamation). Training Aids
for Dam Safety, Module: Evaluation of Embankment Dam Stability and Deformation,
1988. https://damfailures.org/wp-content/uploads/2015/06/Evaluation-of-Embankment-
Dam- Stability-and-Deformation.pdf
U.S. Department of the Interior, Reclamation. Design Standards No. 13, Embankment
Dams, “Chapter 4: Static Slope Stability,” October 2011.
https://damfailures.org/wp-content/uploads/2018/09/Design-Standards-No.-13-Ch-4.pdf
U.S. Department of the Interior, Reclamation. Design Standards No. 13, Embankment
Dams, “Chapter 8: Seepage,” January 2014.
https://www.usbr.gov/tsc/techreferences/designstandards-datacollectionguides/
finalds- pdfs/DS13-8.pdf

About AECOM
AECOM is the world’s trusted infrastructure consulting firm, delivering
professional services throughout the project lifecycle — from planning, design and
engineering to program and construction management. On projects spanning
transportation, buildings, water, new energy and the environment, our public- and
private-sector clients trust us to solve their most complex challenges. Our teams
are driven by a common purpose to deliver a better world through our unrivaled
technical expertise and innovation, a culture of equity, diversity and inclusion,
and a commitment to environmental, social and governance priorities. AECOM is a
ForŁune 500 firm and its Professional Services business had revenue of $13.2
billion in
fiscal year 2020. See how we are delivering sustainable legacies for generations to
come at aecom.com and @AECOM.
aecom.com

©2021 AECOM. All Rights Reserved.