DeepEX 2016 – Non Linear Analysis Manual

NON LINEAR ANALYSIS MANUAL

DeepEX software program (Version 2016)

Version 13.0.0.1

Issued: 6-April-2016

Deep Excavation LLC

www.deepexcavation.com

Deep Excavation LLC Page 1

 DeepEX 2016 – Non Linear Analysis Manual
Table of Contents

1. INTRODUCTION ......................................................................................................................................... 4
2. GENERAL ANALYSIS METHODS AND SOFTWARE INTRODUCTION............................................................ 4
3. MODEL IN THE NON-LINEAR ANALYSIS ENGINE ....................................................................................... 6
4. STAGES OF THE NON-LINEAR ANALYSIS.................................................................................................... 8
4.1 Initial stage .......................................................................................................................................... 8
4.2 Typical excavation stage ..................................................................................................................... 8
4.3 Typical backfill stage ........................................................................................................................... 9
4.4 Support installation stage ................................................................................................................... 9
4.5 External loads and restraints on the wall ......................................................................................... 10
5. SOIL MODELS IN DEEPEX ......................................................................................................................... 11
5.1 Introduction ...................................................................................................................................... 11
5.2 Soil properties ................................................................................................................................... 11
5.3 Soil model – General issues .............................................................................................................. 14
5.4 The soil model for clays..................................................................................................................... 17
5.4.1 Failure condition ............................................................................................................................ 18
5.4.2 Drained behavior of Clays .............................................................................................................. 19
5.4.2 Undrained behavior of Clays .......................................................................................................... 20
5.4.3 Transaction between clay conditions ............................................................................................ 21
5.4.4 Simplified undrained behavior of Clays ......................................................................................... 22
5.4.5 Initializing the clay behavior .......................................................................................................... 22
5.4.6 Summarization ............................................................................................................................... 23
6. WATER STRESSES IN DEEPEX................................................................................................................... 24
6.1 Introduction ...................................................................................................................................... 24
6.2 Water pressures in steady state seepage conditions ....................................................................... 24
6.3 Pore pressure distributions in existence of one undrained soil layer .............................................. 27
6.4 Dredge line stability and lining option in DeepEX............................................................................. 29
6.4.1 Soil improvement option in DeepEX .............................................................................................. 30
6.4.2 The lining option of DeepEX........................................................................................................... 31
6.5 Tabular pore pressures profiles in DeepEX ....................................................................................... 32
7. ESTIMATION OF SOIL PROPERTIES .......................................................................................................... 34

Deep Excavation LLC Page 2

 DeepEX 2016 – Non Linear Analysis Manual
7.1 Introduction ...................................................................................................................................... 34
7.2 Soil properties table .......................................................................................................................... 35
7.3 Remarks on field tests ....................................................................................................................... 36
7.4 Correlations for granular soils........................................................................................................... 38
7.4.1 Resistance properties..................................................................................................................... 38
7.4.2 Stiffness properties ........................................................................................................................ 38
7.5 Correlations for cohesive soils .......................................................................................................... 39
7.5.1 Undrained resistance properties ................................................................................................... 39
7.5.2 Drained resistance properties........................................................................................................ 40
7.5.3 Undrained stiffness properties ...................................................................................................... 40
7.5.4 Drained stiffness properties........................................................................................................... 40
7.6 Thrust coefficients ............................................................................................................................ 40
7.6.1 At –rest coefficient......................................................................................................................... 40
7.6.2 Active thrust coefficient ................................................................................................................. 40
7.6.3 Passive thrust coefficient ............................................................................................................... 41
7.7 Reference permeability values ......................................................................................................... 44
8. ESTIMATION OF VERTICAL SETTLEMENTS IN DEEPEX ............................................................................ 45
9. STRIP FOUNDATION MODELING IN DEEPEX ........................................................................................... 47
10. MODELING A BERM IN DEEPEX ............................................................................................................. 50
11. DEEPEX: RESULT ASSEDDMENT ............................................................................................................ 52
12. SAFETY FACTORS IN DEEPEX ................................................................................................................. 56
13. REFERENCES .......................................................................................................................................... 58

Deep Excavation LLC Page 3

 DeepEX 2016 – Non Linear Analysis Manual
1. INTRODUCTION
This document summarizes the theory used in the nonlinear engine of DeepEX. DeepEX users
should know that using the software without proper skill and background in soil mechanics could
result in an unrealistic and unreliable design. Users should understand and accept the suggested
criteria for estimation of soil parameters, as well as the design methods included in this manual.
Deep Excavation LLC will have no responsibility on any consequence and loss of money, health
and goodwill due to the use of the program.

2. GENERAL ANALYSIS METHODS AND SOFTWARE INTRODUCTION

DeepEX is capable of analyzing braced excavations with “conventional” limit-equilibrium-
methods and beam on elastic foundations (i.e. the traditional NONLINEAR engine). An excavation
can be analyzed in one of the following sequences:
a) LEM: Conventional analysis only, which for reasons of common language is called “Limit
Equilibrium”.
b) NL: Non-linear analysis only, with active and passive stage dependent soil springs (Winkler
method)
c) Combined “Conventional”-Non-linear Analysis: 1st Conventional analysis with
traditional safety factors stored in memory. Once the traditional analysis is completed,
then the non-linear analysis is launched.

The Non Linear Engine of this the program is based on the soil model Roberto Nova published in
1987. DeepEX is a finite element program which considers both soil, water, and structural
elements.

DeepEX comes packed with many scientific methods, design standards, extensive databases of
supports and additional modules that cover a huge part of deep excavation design.

DeepEX is used by more than one thousand engineers and contractors worldwide.
Representative projects include:

1. Westside Subway Expansion in Los Angeles, CA, USA

Deep Excavation LLC Page 4

 DeepEX 2016 – Non Linear Analysis Manual
2. Hudson Yards Redevelopment in New York, NY, USA

3. City University, New York, NY, USA

4. New Apple Headquarters, Cupertino City, CA, USA

The following chapters describe the methods and capabilities included in the non-linear engine of
DeepEX.

Deep Excavation LLC Page 5

 DeepEX 2016 – Non Linear Analysis Manual
3. MODEL IN THE NON-LINEAR ANALYSIS ENGINE
DeepEX has a non-linear finite element engine for the analysis of the behavior of flexible earth
retaining structures during all the intermediate steps of a braced excavation. The actual problem
is reduced to a 2D plane problem, in which a unit width slice of the wall is considered (Figure 1).

The beam on elastoplastic foundations approach is adopted in modelling soil-wall interaction.

The retaining wall is modelled by beam elements with transversal bending stiffness EJ and the
soil is modelled with active and passive independent elastoplastic springs at each node.

Figure 1: Problem simplification in non-linear analysis.

Figure 2: Soil, anchor and flexible wall modeled with spring and beam elements in DeepEX.

According to the Winkler model, the behavior of every soil spring is independent from the
behavior of adjacent elements. The actual interaction among different soil regions is totally

Deep Excavation LLC Page 6

 DeepEX 2016 – Non Linear Analysis Manual
dependent on tthe retaining wall. The real progress of an excavation process is reproduced in all
the intermediate steps, by means of a STATIC INCREMENTAL analysis. Due to the elastoplastic
behavior of the soil elements, every step in general depends on the solution at the previous steps.
The solution at each analysis step is obtained by means of a Newton-Raphson iterative scheme
(Bathe, 1996).

The non-linear engine of DeepEX only computes the lateral behavior of the retaining wall. At each
node, only the lateral displacement and out-of-plane rotation (about the X axis) are activated as
independent degrees of freedom.

Moreover, the vertical stress distribution in the soil not influenced by the lateral deformations in
the soil itself. At each depth, the vertical stress is an independent variable that is calculated by
means of the usual assumption of geostatic distribution.

Deep Excavation LLC Page 7

 DeepEX 2016 – Non Linear Analysis Manual
4. STAGES OF THE NON-LINEAR ANALYSIS
Each non-linear analysis stage can differ from the other stages due to changes in excavation
depths, water elevations, different active structural elements, and different soil properties. All
finite elements can be activated and later removed during the analysis. Each activated element
is assumed to be strain free. If there is any prestressing force, it is transferred to the adjacent
elements. The following paragraphs will describe the different stages during the non-linear
analysis of an excavation supported by a flexible wall.

4.1 Initial stage

As with real conditions, the non-linear analysis in DeepEX starts capturing the initial at-rest lateral
and vertical soil stress conditions. To capture at-rest conditions in the initial stage the excavation
and water elevations should be the same on each wall side (driving and resisting) at stage 0. In
addition, no external load or tieback anchor should be activated in this initial stage.

In this initial stage we should expect zero lateral displacements, moment and shear forces on the
wall. Lateral soil element stresses should capture the at-rest lateral stress distribution, related to
the vertical stresses by the at rest Ko coefficient. The lateral and vertical effects due to strip
loadings are added to the geostatic stresses. The non-linear analysis engine assumes that the wall
is wished in place (no effect on soil stresses).

4.2 Typical excavation stage

During a typical excavation stage, the non-linear engine removes all soil elements above the
excavation level. This action apparently modifies the equilibrium configuration of the previous
stage. As a result, the engine tries to reach a new balanced configuration by iteration, converging
to a new deformed configuration model, unless failure conditions are reached. Removing
excavated soil causes the vertical stress of the remaining soil spring elements in the excavation
to be reduced.

The maximum horizontal stress limit of the remaining elements (passive condition) readjusts
accordingly, so stress redistribution is required in order to return within the plasticity boundaries.
A water table change could affect the overall wall stability significantly, so, such an effect should
be modelled carefully.

In some cases, the natural soil in the excavation is improved by technical methods like
jetgrouting. These improvements upgrade the stiffness and resistance of the soil, so they should
be taken into account by employing a soil change command.

Deep Excavation LLC Page 8

 DeepEX 2016 – Non Linear Analysis Manual
4.3 Typical backfill stage
When a backfill operation is performed, i.e. the excavation level is raised, the stress field of soil
spring elements is readjusted. The effective vertical stress in computed assuming geostatic
conditions and including the contribution of any external surcharges. The effective horizontal
stress is calculated by multiplying the vertical stress including the uniform surcharge contribution
(but not the contribution of strip loadings), with K0. Finally, the water pressure is computed as in
any other wall portion.

If backfilling was performed to restore the ground surface to the original elevation (after an
excavation procedure), then after the iteration process, the effective horizontal stress will likely
differ from the at-rest conditions.

4.4 Support installation stage

It is recommended to install active ground anchor supports in an independent stage where no
other modifications take place. The tieback installation (i.e. activation) stage should be preceded
by another stage where the excavation subgrade should be just below the anchor elevation on
the wall (typically 2ft or 0.5m), as shown in Figure 3.

Excavation stage Anchor installation stage

Figure 3: Ground anchor installation stage and the preceding excavation stage.

Deep Excavation LLC Page 9

 DeepEX 2016 – Non Linear Analysis Manual
It is recommended that a tieback prestress force is applied only in the installation stage. This is
the support reaction for this stage. In all following stages, the anchor reaction force due to wall
deflection is calculated, using a stiffness:
𝐴
𝐾=𝐸∙
𝐿
Where:
E: is the strand Young modulus
A: is the anchor section area per unit length
L: is the length of the deformable part of the anchor.
As presented in Figure 3, L=Lfree + Lfixed*n, where n<1 in most cases

If an initial prestress force is not assigned, then the support is modelled as a passive anchor.
Similar behavior is assumed during other support types installation (slabs, struts, rakers).

4.5 External loads and restraints on the wall

In the nonlinear engine the user can assign directly on the walls external linear or distributed
lateral forces, external moments, or displacements. This could assist in simulating specific
situations, like the presence of a temporary load cantilever fixed to the wall. The vertical loads
are not treated as external loads on the wall. They affect the vertical stress distribution in the
soil.

In addition, the user can include fixed supports and prescribed displacements and rotation in any
position along the wall.

Deep Excavation LLC Page 10

 DeepEX 2016 – Non Linear Analysis Manual
5. SOIL MODELS IN DEEPEX

5.1 Introduction
The interaction between soil and a retaining support system (wall and supports) is a complex
geotechnical problem that can be solved by making certain approximations. The two-dimensional
plane strain finite element model is the most commonly used approach in solving braced
excavations as it allows for a precise soil parameter description. Finite element software
programs using this method allow the inclusion of quite complex and realistic models.

The non-linear engine of DeepEX adopts a numerical method, in which the soil is modeled by an
array of active and passive Winkler springs. In DeepEX, the soil springs can capture the most
observed soil behavior aspects. Soil spring stiffness does not only depend on the soil properties,
but also on the wall geometry and the wall flexibility (Jamiolkowski and Pasqualini, 1979). DeepEX
requires the definition of the usual shear resistance and elasticity parameters, leading to quite
complex and complete modeling features. As always, the reliability of obtained results using
DeepEX depends on the accuracy of the defined parameters.

The following sections present the physical meaning of the soil parameters and model in the non-
linear engine of DeepEX.

5.2 Soil properties

Lateral pressure thrust coefficients are the at-rest coefficient K0, the active coefficient KA and the
passive coefficient KP. The at-rest coefficient affects the initial stress state in the soil prior to any
excavation or construction activities. K0 is used in the calculation of the effective horizontal stress
σ'h according to the following equation:

σ′ℎ = 𝐾0 ∙ σ′𝑣 , where:

σ'h is the effective horizontal stress
σ'v is the effective vertical stress
K0 is the at-rest lateral earth pressure coefficient

K0 depends on the soil resistance, through the effective friction angle φ', and on the geological
history:

K 0 = 𝐾0 𝑁𝐶 ∙ (𝑂𝐶𝑅)𝑚 , where:
𝐾0 𝑁𝐶 = 1 − 𝑠𝑖𝑛𝜑′ is the normal at-rest coefficient (OCR=1).
OCR is the overconsolidation ratio
m is an empirical parameter, usually ranging between 0.4 and 0.7

Deep Excavation LLC Page 11

 DeepEX 2016 – Non Linear Analysis Manual
The active and passive thrust coefficients for frictionless walls when the ground surfaces are
horizontal can be calculated according to Rankine using:
𝜑′
K𝐴 = 𝑡𝑎𝑛2 (45° − )
2
𝜑′
K 𝑃 = 𝑡𝑎𝑛2 (45° + )
2

Though, in DeepEX we can take into account the wall friction angle δ and the soil surface sloping.
In this case, selected values in NAVFAC (1976) or by Caquot and Kerisel (1948) or by Lancellota
are recommended.

Extreme effective horizontal stress limits are given by the following equations:

σ′ℎ = 𝐾𝐴 ∙ σ′ 𝑣 − 2 ∙ 𝑐′ ∙ √𝐾𝐴 , min. stress for active conditions always greater than zero
σ′ℎ = 𝐾𝑃 ∙ σ′ 𝑣 + 2 ∙ 𝑐′ ∙ √𝐾𝑃 , max. stress for passive conditions

c' is the drained cohesion

Soil compressibility parameters affect the calculated spring stiffness. The stiffness k per unit
length of a Winkler spring array is calculated by the following equation:

𝑘 = 𝐸/𝐿 , where
E is the soil stiffness modulus
L is a scale length.

In DeepEX, lumped springs are generated at finite distance Δ, thus the stiffness of each spring is:

𝐸𝛥
𝑘=
𝐿

The parameter Δ depends on the finite element mesh density. The parameter L is automatically
selected by the software and represents a characteristic length which is different between
resisting and driving soil regions.

2 𝜑′
For driving soil (active zone): L𝐴 = 3 𝜆𝐴 ∙ tan (45° − )
2
2 𝜑′
For resisting soil (passive zone): L𝑃 = 𝜆𝑃 ∙ tan (45° + )
3 2
𝜆𝐴 = min⁡{𝑙⁡, 2𝐻}
𝜆𝑃 = min⁡{𝑙 − 𝐻⁡, 𝐻} , where:
l is the total wall height
H is the current excavation depth

Deep Excavation LLC Page 12

 DeepEX 2016 – Non Linear Analysis Manual
The soil modulus E is frequently related to the effective mean stress 𝜌 = (σ′ ℎ + σ′ 𝑣 )/2 using:

𝐸 = 𝑅(𝜌/𝜌𝑎 )𝑛 , where:

𝜌𝑎 is the atmospheric pressure

R and n are experimental parameters.

Setting n=0, we obtain a constant modulus, whilst setting n=0, typical modulus variation is
obtained for normally consolidated soils. R is different between virgin compression and loading-
reloading paths.

Reference values for R and n are reported by Janbu (1963). Such parameters vary within a very
wide range: for a sand, n may be between 0.2 and 1.0 and R between 8 and 200 MPa. Since the
initial stress state is not isotropic, the virgin compression soil stiffness is currently less that the
measured stiffness in a drained isotropically consolidated triaxial test.

If n=0, R modulus in virgin compression can be identified with the usual Young modulus. The
unloading-reloading modulus is currently 3 to 10 times higher than virgin modulus for clays,
whereas it is usually 1.5 to 3 higher for sands.

Deep Excavation LLC Page 13

 DeepEX 2016 – Non Linear Analysis Manual
5.3 Soil model – General issues
The vertical and horizontal effective stress components (σ′ 𝑣 ⁡end⁡σ′ ℎ respectively) are defined as
the principal stresses. The yield function and a hardening rule are also defined in the principal
stress plane (Figure 4).

Figure 4: The stress plane for a drained soil element.

The following three phase conditions are possible:

 Elastic phase: The soil element behaves elastically. This state corresponds with an
unloading-reloading condition in the soil. The stress is less or equal to a previous stress
level. In Figure 4, this area is marked with UL-RL. The highest element stiffness is assigned
to the element.
 Hardening phase: In this phase, the stress increases beyond the maximum stress level
that was reached in a previous stage. During this phase, a strain hardening behavior is
assigned and the element stiffness is still not zero. In Figure 4, this area is marked with PC
(Primary Compression).
 Yielding: The minimum (active) or maximum (passive) horizontal stress is reached, so the
element behaves as a plastic material.

The effective vertical stress at each depth is computed based on the ground surface, on the
surcharge an on the water level, in order to set up the initial stress state at the beginning of the
analysis. The horizontal stress is then recovered using the at rest coefficient K0. The contributions
due to point loads are then added. To establish the initial element phase (if the element is in UR-
LR or PC phase), the overconsolidation ratio OCR the normally consolidated at rest coefficient
K0NC are used as follows:

Deep Excavation LLC Page 14

 DeepEX 2016 – Non Linear Analysis Manual

We assume that maximum past stresses are:

σ′𝑣,𝑚𝑎𝑥 = 𝑂𝐶𝑅 ∙ σ′ 𝑣,𝑠𝑡𝑎𝑔𝑒⁡1

σ′ℎ,𝑚𝑎𝑥 = 𝐾0 𝑁𝐶 ∙ σ′ 𝑣,𝑚𝑎𝑥

If both initial stress components are below these limits, then the element is initially in the UL-RL
conditions, otherwise, it is in virgin compression state.

In following stages, the vertical stress is calculated based on the current excavation layout,
surcharge, and seepage conditions. The horizontal stress is updated by calculating the stress
increments due to the element incremental deformation, by means of the elastic soil properties.
The incremental stress is iteratively updated to meet the yield conditions based on the current
vertical stress value. The element pore pressure is then added to the effective lateral pressure,
so that the total lateral pressure is calculated.

The relationship between σ′ 𝑣 ⁡end⁡σ′ ℎ depends on the current element state. Initially, an
element is represented by the at-rest coefficient K0. At yielding,⁡σ′ 𝑣 ⁡end⁡σ′ ℎ are constrained to
meet the yield condition. In a stress path internal to the elastic domain, corresponding with null
incremental lateral deformation, the incremental horizontal effective stress Δσ’h is related to the
incremental vertical effective stress Δσ’v depending on K0NC , σ’v,max end σ’h,max.

If Δσ’v > 0 and the stress point represents a normally consolidated state on the elastic domain
bounded by the σ′ 𝑣 = σ′ 𝑣,𝑚𝑎𝑥 and σ′ ℎ = σ′ ℎ,𝑚𝑎𝑥 , then:
Δσ′ℎ = 𝐾0 𝑁𝐶 ∙ 𝛥σ′ 𝑣

For an overconsolidated which is reloaded or in an uploading stress path, Δσ′ℎ is calculated by a

non-linear equation, which for an oedometric condition, meets the following equation:

σ′ℎ = 𝐾0 𝑁𝐶 ∙ (σ′ 𝑣,𝑚𝑎𝑥 /σ′ 𝑣 )𝑛 ∙ σ′ 𝑣 or σ′ℎ = 𝐾0 𝑁𝐶 ∙ (𝑂𝐶𝑅)𝑛 ∙ σ′ 𝑣

Deep Excavation LLC Page 15

 DeepEX 2016 – Non Linear Analysis Manual

Figure 5: Oedometric condition.

Figure 6 presents the stress path in the σ′ 𝑣 − σ′ ℎ and in the 𝛿 − σ′ ℎ planes, where δ = the lateral
soil deformation. Point 1 corresponds with at-rest conditions for a granular overconsolidated soil.
This point is internal to the boundary between the elastic and the virgin compression regions.
Subsequent points track the lateral stress state evolution at σ’v=constant: by compressing the
element, the virgin compression boundary is firstly encountered (pt 2); from 1 to 2 the elastic
(unloading-reloading) stiffness is used, whereas from pt 2 on, the virgin compression stiffness is
adopted until the yield (passive) limit is reached. Point 4 is reached by the development of plastic
strain. The path from 4 to 5 represents an unloading path (with elastic stiffness) whereas from 5
to 6 a reloading path.

Figure 6: Stress paths for a granular overconsolidated soil.

In Figure 7, point 1 represents an at rest stress state for a normally consolidated granular
material: in fact such point establishes the boundary between elastic (unloading-reloading)
region and the virgin compression zone. Subsequent points no. 2 and 3 represent a stress
evolution at s’v=const. towards passive conditions (point 3). The path from 3 to 4 is related to
vertical stress reduction along with a lateral strain release towards active limit conditions in 4.

Deep Excavation LLC Page 16

 DeepEX 2016 – Non Linear Analysis Manual

Figure 7: Stress paths for a granular normal consolidated soil.

5.4 The soil model for clays

The soil model for clays describes the limit conditions based on effective resistance parameters
only, and allows the simulation of both drained and undrained conditions, as well as the
transition between them.

In drained conditions, this model is very similar to the model for granular soils. The only
difference lays in the apparent cohesion parameter c’, which is dependent on the
preconsolidation, whereas for granular soils is a fixed user input value.

In undrained conditions, both effective stress path (ESP) and total stress path (TSP) are computed
and the ESP is monitored to check limit conditions. The ESP evolution is highly affected by the
imposed constraint on the volumetric deformation which must be zero in undrained conditions.
Since both ESP and TSP can be computed, the pore pressure change within the saturated soil in
undrained condition can be easily computed as well.

Deep Excavation LLC Page 17

 DeepEX 2016 – Non Linear Analysis Manual
5.4.1 Failure condition
Figure 8 presents the limit conditions of the clay model. The elastic domain depends on four
thrust coefficients (KA,CV, KA,peak, KP,CV, KP,peak) and on the reconsolidation level.

Figure 8: Clay model – limit conditions.

The points A and P on the limit state lines σ′ℎ = 𝐾𝐴,𝐶𝑉 ∙ σ′ 𝑣 and σ′ℎ = 𝐾𝑃,𝐶𝑉 ∙ σ′ 𝑣 are defined by
σ′ 𝑣,𝑚𝑎𝑥 and σ′ ℎ,𝑚𝑎𝑥 respectively. The points A’ and P’ on axis are defined as intersections of two
lines from A and P with slope equal to KA,peak and KP,peak respectively. The segments O-A’ and O-
P’ represent the cohesion. As σ′ 𝑣,𝑚𝑎𝑥 and σ′ ℎ,𝑚𝑎𝑥 increase, points A and P move and the elastic
domain expands irreversibly.

The limit state lines represent the ultimate conditions related to large deformations. KA,CV and
KP,CV are active and passive thrust coefficients respectively, depending on the critical friction angle
φ’CV and on the friction between the soil and the wall. KA,peak and KP,peak are active and passive
thrust coefficients respectively, depending on a friction angle φ’peak < φ’CV and on the friction
between the soil and the wall (Figure 9).

Deep Excavation LLC Page 18

 DeepEX 2016 – Non Linear Analysis Manual

Figure 9: Mohr limit condition for OC clays.

From this model we can assume that a clay resistance can be described by φ’ CV, which depends
on the plasticity index. The non-linear engine of DeepEX in this case, calculates the peak friction
angle according to the following equation:

tan⁡(φ′ 𝑝𝑒𝑎𝑘 ) = tan⁡(φ′ 𝐶𝑉 )/1.5

The thrust coefficients are calculated internally using Rankine formulas.

5.4.2 Drained behavior of Clays

The clay model in case of drained behavior is similar to the model for granular soil. The apparent
cohesion varies along with stress. Figure 10 presents the drained soil stiffness calculation paths
based on effective modulus according to user’s input.

Deep Excavation LLC Page 19

 DeepEX 2016 – Non Linear Analysis Manual

Figure 10: Drained clay model stress paths.

5.4.2 Undrained behavior of Clays

In undrained clay conditions, the zero volume change condition is imposed. We assume initial
drained conditions according to the following:
σh,0=K0(σv,0 – u) + u + strip foundations effects

In the subsequent stages, if the lateral deformation is freezed, any total vertical stress increment
produces an equal total lateral stress increment, due to a water pressure increment. On the other
hand, when incremental lateral deformations are allowed, total lateral stress increments occur
(with zero total vertical stress increment). In the elastic domain it must be ensured that zero
incremental volumetric strain constraint is assured, so: Δσ’v + Δσ’h =0

Such conditions dictated the effective stress path slope within the elastic domain. When the
elastic domain boundary is reached, different behaviors are possible, depending on the stress
path. In such cases, plastic strains are computed assuming an associated flow rule. The water
pressure is computed by subtracting the effective stress from the total stress and we assume that
soil permeability is zero.

Deep Excavation LLC Page 20

 DeepEX 2016 – Non Linear Analysis Manual

Figure 11: Undrained clay model stress paths.

The undrained shear strength Su is not required to model the undrained behavior, since such
parameter is implicitly derived from the effective constitutive model. It is however possible to
assign an external value for Su, thus prescribing an additional check on the total shear. In this
case, both effective and total stress paths are monitored by DeepEX. The load increment is
stopped as soon as the first one of the above reaches its relevant boundary. The undrained soil
stiffness is calculated based on effective modulus.

5.4.3 Transaction between clay conditions

Switching from undrained to drained behavior:

During undrained clay condition, the pore pressure varies because of soil shear deformation.
When switching to drained conditions, the pore pressures are recalculated based on the current
uncoupled water table conditions. Excess pore pressure due to soil deformation is dissipated, In
addition, just effective stress component is included in the constitutive equations.

Switching from drained to undrained behavior:

In the stage that this change happens, DeepEX calculates the new horizontal soil stresses in
undrained condition by adding the water pressure at the end of the previous stage, u, to the
horizontal effective stress at the end of the previous step. During the iterative process, the
horizontal soil stress in undrained conditions may vary due to incremental soil displacements.

Deep Excavation LLC Page 21

 DeepEX 2016 – Non Linear Analysis Manual
5.4.4 Simplified undrained behavior of Clays
It is possible to perform a simplified undrained analysis in DeepEX by just providing the values for
undrained shear strength Su and the undrained modulus Eu. In this case, the effective stress
failure conditions will be neglected and DeepEX will monitor only the total stress. When using
this simplified model, it is not possible to pass to undrained conditions, since the pore pressure
evolution is lost during the undrained phase.

5.4.5 Initializing the clay behavior

The initial soil conditions should be restored as precisely as possible, since the past conditions
affects both drained and undrained behavior. In overconsolidated clays, a realistic OCR profile
has to be restored initially. The following options are available:
a. A unique OCR value greater than 1 has to be defined for each overconsolidated clay layer.
b. Initially we define an OCR=1. In this case, in the first stage, drained conditions are
activated and an external surcharge is imposed to the model, in order to simulate the
overconsolidation ratio. In the following stage, we maintain drained conditions and we
remove the surcharge. Finally, we add a new stage, which is the actual initial stage of the
model, in which the clay is in either drained or undrained conditions, according to the
expected behavior during the excavation process. Figure 12 describes this approach.
The second approach is recommended, since the obtained OCR model is more realistic.

Stage 1 Stage 2 Stage 3

Figure 12: Recovering initial OCR profile in clay layer.

Deep Excavation LLC Page 22

 DeepEX 2016 – Non Linear Analysis Manual
5.4.6 Summarization
The features of the available soil models are available in the following table:

Soil Model
Simplified model
Features Granular soil
Clay model for undrained
model
behavior
Drained conditions Yes Yes No
Undrained conditions No Yes Yes
From drained to undrained No Yes Yes
From undrained to drained No Yes No
Resistance parameters (drained) c' , φ’ φ’CV,φ’peak No
Resistance parameters (undrained) No φ’CV,φ’peak,Su Su
Flexibility parameters (drained) Eur,Evc Eur,Evc No
Flexibility parameters (undrained) No Eur,Evc Eu
Resistance parameter modifications
Yes No Yes
during analysis
Flexibility parameter modifications
Yes Yes Yes
during analysis
Pore pressure calculations Yes Yes No
Permeability Yes No No
Granular soil modeling (sands, gravels) Yes Yes No
Clay modeling (drained conditions) Simplified Yes No
Clay modeling (undrained conditions) No Yes Simplified
Cemented sand modelling Yes No No
Modelling of improved soil behavior Recommended
Simulation of weak rock behavior by
Recommended
unconfined compressive strength

Deep Excavation LLC Page 23

 DeepEX 2016 – Non Linear Analysis Manual
6. WATER STRESSES IN DEEPEX

6.1 Introduction
The non-linear engine of DeepEX assumes that the submerged soil is fully saturated. It is assumed
that the soil stress and deformation do not affect the pore water pressure distribution. In the
submerged undrained clay soil layers, the pore pressure is undefined, unless the clay model is
selected. The soil consolidation effects are not considered in the analysis.

The following models of water analysis are considered:

1. Hydrostatic soil pressures, without water flow
2. Steady state seepage conditions, represented by a water within a porous medium

The following paragraphs will analyze the simplified flow options.

6.2 Water pressures in steady state seepage conditions

Figure 13 presents the pore pressures and the hydraulic gradients calculation during the steady
state seepage conditions.

Figure 13: Steady state seepage scheme in DeepEX.

As shown in Figure 13, a vertical flow path is assumed. The overall path length L is calculated as
the minimum path length adjacent to the wall (neglecting the wall thickness). DeepEX assumes
that the lining effects of the wall finish at elevation Z = Zbalance. The flow path length is minimized
based on this approximation.

Deep Excavation LLC Page 24

 DeepEX 2016 – Non Linear Analysis Manual
Let:
DH: Overall head loss (DH=DHW in Figure 13)
v: Flow velocity
Ki: Permeability coefficient of the i-th soil layer
Li: Thickness of the i-th soil layer
DHi: Partial head loss within i-th soil layer

By invoking the Darcy law, the continuity equation and assuming a constant hydraulic gradient
inside each soil layer, we can say that:

𝐷𝐻𝑖
𝑣 = 𝐾𝑖 and 𝐷𝐻 = ∑𝑗 𝐷𝐻𝑗
𝐿𝑖

By removing the unknown v, we can calculate the i-th partial head loss from the following
equation:
𝐿𝑖
𝐾𝑖
𝐷𝐻𝑖 = 𝐷𝐻 ∙
𝐿
∑𝑗 𝑗
𝐾𝑗
Such sum has to be extended over all the crossed soil layers in both uphill and downhill regions.
Once each partial head loss has been computed, the pore pressure distribution is given by
applying the Bernoulli theorem:

𝑢 = 𝑢(𝑍) = 𝛾𝑤 ∙ (𝑍𝑊𝑇 − 𝑍) − ∑ 𝐷𝐻𝑘

𝑘
In this equation, the sum but be extended over all crossed soil layers up to the current position.

If an undrained conditions soil layer is encountered along the assumed flow path, the flow is
actually stopped and hydrostatic conditions occur.

NOTES:
 Such scheme represents an acceptable solution of the real problem, since it is on the safe
side. The hydraulic gradients are overestimated thus overestimating the seepage drag
forces that usually represent the real danger since downhill quick conditions are
concerned.
 The conservativeness in estimation of the seepage forces justify the slightly
underestimated pore pressure distribution of this method in comparison to other more
precise approaches.
 When the permeability is omitted in the definition of the property of a soil layer, DeepEX
assigns a very low number for such parameter, assuming that such layer is practically
impervious. However, if such a default value is assigned to all the soil layers (when the
permeability is always omitted), a seepage in a homogeneous medium is actually
reproduced, because DeepEX activates the flow anyway.
 It is important to preserve the representative permeability ratios among the various soil
layers.

Deep Excavation LLC Page 25

 DeepEX 2016 – Non Linear Analysis Manual
 The balance elevation can by directly assigned by user, modifying the DeepEX assumption
corresponding with the lowest wall depth. By placing the Zbalance at a very low level, an
uncoupled pore pressure distribution between uphill and downhill soil is obtained, like
when an impervious soil exists at the wall toe. By placing Zbalance at a lower elevation
than the wall toe, the flow path is artificially stretched, recovering a desired pressure
distribution that differs from the default scheme.
 For the soil layers that are below the balance elevation, the pore pressure is calculated by
setting the balance level to the layer position.

IMPORTANT: The water weight must be defined for all soil layers, else, unreasonable and
unpredictable results might be obtained in DeepEX.

Figure 14 presents some typical cases that can be modeled with DeepEX.

Figure 14: Typical cases that can be designed with DeepEX.

Deep Excavation LLC Page 26

 DeepEX 2016 – Non Linear Analysis Manual
6.3 Pore pressure distributions in existence of one undrained soil layer
As mentioned in the previous paragraph, if an undrained conditions soil layer is encountered
along the assumed flow path, the flow is actually stopped and hydrostatic conditions occur.
a. Case in which only one undrained soil region interferes with the possible flow path. In this
case, the undrained behavior is assigned to only one soil layer, in only one side of the wall.
The undrained layer acts as an impervious boundary between soil regions that
communicate with either the uphill or the downhill phreatic levels. Figure 15 presents
some examples of such cases.

Figure 15: Typical cases with only one undrained region.

In case 1, uphill region 3 directly communicates with the downhill water table; in case 2,
on the contrary, downhill region 2 communicates with the uphill water table; in case 3,
regions 3, 4 and 5 are linked to the downhill water table; in particular, in region 3 a null
pore pressure distribution is assumed. In case 4 the same hydrostatic condition hold for
any soil region.

In the undrained regions, the pore pressures are undefined.

Deep Excavation LLC Page 27

 DeepEX 2016 – Non Linear Analysis Manual

b. Case in which several undrained regions interfere with the assumed seepage path. In this
case some drained regions may exist, but they do not communicate either with the uphill
or the downhill water table. In the first analysis step (Step 0), DeepEX assigns to the
unconnected drained regions a mean water table between uphill and downhill level. In
any subsequent step (Step i with i>0), DeepEX assigns to these regions the pore pressures
of the previous step (Figure 16). According to this approximation, it is strongly
recommended to prescribe initial balanced hydraulic conditions.

Figure 16: Typical cases with several undrained regions.

Deep Excavation LLC Page 28

 DeepEX 2016 – Non Linear Analysis Manual
c. Case in which some undrained regions exist, though they are not interfering with the
assumed flow path (Figure 17). In this case the steady seepage is not stopped by the
undrained impervious barrier.

Figure 17: Case with undrained layer that is not interfering with flow path.

6.4 Dredge line stability and lining option in DeepEX

The following options are available in DeepEX, in order to model some lining effects at the dredge
line inside an opencut excavation, so that dewatering operations can be performed:
a. A soil improvement can be modelled which highly reduces the permeability of a soil
mass inside the excavation.
b. Use of the Lining option of DeepEX.
The following paragraphs describe these two options.

Deep Excavation LLC Page 29

 DeepEX 2016 – Non Linear Analysis Manual
6.4.1 Soil improvement option in DeepEX
The dewatering inside an excavation is sometimes activated after improving a certain soil mass
at the dredge line by cement grouting or by other techniques like jetgrouting. Such a treated soil
mass receives an improvement in terms of permeability reduction, shear strength and elastic
modulus increase.

In DeepEX, we can simulate this improvement operation by changing the natural soil properties
(permeability, cohesion and stiffness) in accordance with the foreseen improved properties. The
dewatering operation can be modelled by simply assigning a value for the head loss parameter
DZWT>0

If the downhill water table is lowered under the lower surface of the improved mass (Figure 18 –
Case A), a seepage flow will be activated in the natural soil only; no water pressures will occur at
the improved mass base, and the stability will be not reduced. The soil under the improved mass
may contribute to the wall stability unless quick conditions are reached in such zone.

If the downhill water head is higher than the lower improved soil mass elevation (Figure 18 –
Case B), the head loss will be basically dissipated inside the improved soil; in the natural soil, the
uphill water head will act and the seepage forces will be negligible. Such a situation is only
possible if the uplift water pressure resultant at the improved soil mass base does not exceed the
total weight of the improved soil mass.

Such stability condition can be expressed by the following condition:

𝛾𝑡 ∙ ℎ𝑡 ≥ 𝛾𝑤 ∙ 𝑧𝑡 , where:

γt: Total unit weight of the improved soil

ht: Thickness of the improved soil
γ w: Water unit weight
zt: Distance from the uphill water table to the bottom of the improved soil layer.

If the condition above is not met, additional downwards load must be added. This can be done
by adding a surcharge and/or by vertical ground anchors. In DeepEX, an equivalent surcharge qs
at the dredge line level can be added. The surcharge qs shall be at least equal to 𝛾𝑤 ∙ 𝑧𝑡 − 𝛾𝑡 ∙ ℎ𝑡 .
Figure 18 presents the soil improvement option.

Deep Excavation LLC Page 30

 DeepEX 2016 – Non Linear Analysis Manual

Figure 18: Improving soil for dewatering.

6.4.2 The lining option of DeepEX

When the Lining Option is selected, DeepEX removes the pore pressure part of the element stress
for those soil elements above the lining level and neglects the water weight above the ground
level when computing the total vertical stress in the remaining elements under the excavation
level at the excavation side. In order this solution to be realistic, an adequate surcharge qs is
applied at the excavation level, with respect to the uphill water table. Figure 19 presents the
lining option.

Figure 19: Lining option.

Deep Excavation LLC Page 31

 DeepEX 2016 – Non Linear Analysis Manual
6.5 Tabular pore pressures profiles in DeepEX
Uphill and downhill pore pressure profiles can be assigned by tabular values along the wall depth,
as an alternative to the default procedure described in the previous paragraphs. At each side, the
phreatic level is assumed corresponding with maximum elevation in pore pressure profile
description. Below such phreatic elevation, saturated soil is assumed. Finally, we assume that out
of the pressure definition range, the pore pressure is zero. Figure 20 presents these assumptions.

Figure 20: Tabular pore pressure profiles.

IMPORTANT: In this mode, user is responsible to input values consistent to physical problem,
since DeepEX does not perform any checks on these values.

The pore pressure definition by tabular values prevails on standard water description. If same
tabular pore pressures should be used in many steps, user should explicitly assign such values at
each step. If in any step description, no tabular values are given, standard water description is
restored.

Tabular description may be useful to model special situations such as:

 Definition of pore pressures profiles computed by different algorithms, such as finite
difference flownets.
 Modelling of special water conditions that cannot be reproduced by standard
procedure.

Deep Excavation LLC Page 32

 DeepEX 2016 – Non Linear Analysis Manual

Figure 21: Special situations modeled with tabular description.

Deep Excavation LLC Page 33

 DeepEX 2016 – Non Linear Analysis Manual
7. ESTIMATION OF SOIL PROPERTIES

7.1 Introduction
DeepEX offers a set of estimation tools that can help user to estimate the required soil properties.
These correlations are intentionally limited to soils or weak rocks which currently represent most
of the situations where a flexible wall is required. The proposed estimation tools have been
selected amongst various available methods and should be used to initially guess values for soil
parameters. The proposed estimated values should always be and adjusted to the actual problem
according to user’s experience.

Most of such correlations are based on field tests: for granular, soils such method of investigation
currently represent the most common and feasible technology employed, however, for cohesive
soils, an adequate laboratory test campaign is more advisable. As for the proposed correlations
for elastic moduli, most of the available studies aim at an evaluation of suitable elastic properties
for foundation problems, which obviously quite differ from the problem of a retaining wall, in
particular in the fact that, in the latter case, lower deformation are currently expected with
respect to the previous one. Nevertheless, such correlations, yet not perfectly suitable for the
estimation of elastic properties in a retaining wall problem, may as well assist user in the
evaluation of such parameters.

Finally, the correlations are currently presented in a graphic way, should reveal the complexities
underneath and suggest a great care in their usage. When relevant doubts still remains in the
evaluation of some parameters, a sensitivity analysis is advised, thus studying the dependency of
relevant results on the variation of most doubtful parameters. By this approach, a quite reliable
depiction of the problem under investigation is currently obtained.

Deep Excavation LLC Page 34

 DeepEX 2016 – Non Linear Analysis Manual
7.2 Soil properties table
The following table presents the symbols of the various soil properties, included in DeepEX.
Symbol Description
c' Drained cohesion
Su Undrained shear strength
φ' Friction angle
φ'CV Constant volume friction angle
φ'peak Peak friction angle
φ'u Undrained friction angle
σ’h Effective horizontal stress
σ’v Effective vertical stress
σh Total horizontal stress
σv Total vertical stress
Relative density
𝐷𝑟 = (𝑒𝑚𝑎𝑥 − 𝑒)/(𝑒𝑚𝑎𝑥 − 𝑒𝑚𝑖𝑛 )
Granular soils are usually subdivided as follows:
Dr(%) – Description
Dr 0-15 – Very loose
15-35 - Loose
35-65 - Medium
65-85 - Dense
85-100 – Very dense
e void ratio defined as the ratio of the void volume to the solid volume
emin Minimum void ratio corresponding to the densest state
emax Maximum void ratio corresponding to the loosest state
E, Evc, Eur, Eu Young’s modulus
G Shear modulus
G0 Shear modulus at very small strains
Ka Active thrust coefficient
Kp Passive thrust coefficient
Normalized blow representing the SPT blow count measured by a SPT
N60
device with an energy ratio=60% at an overburden pressure of 1 atm
Nk Cone factor
NSPT SPT blow count
OCR Overconsolidation ratio
pa Atmospheric pressure
pl Plasticity index: pl=wl-wp
wl Liquid limit (see Atterberg limits definition)
wp Plastic limit (see Atterberg limits definition)
water content defined as the ratio of water weight to the
w
solid weight

Deep Excavation LLC Page 35

 DeepEX 2016 – Non Linear Analysis Manual
Symbol Description
δ Friction angle between soil and wall
SPT Standard Penetration Test
CPT Cone Penetration Test
qC Cone penetration resistance
u Pore pressure
η Energy ratio of an SPT device

7.3 Remarks on field tests

Most of the correlations proposed by DeepEX are based on penetrometric tests. When using
dynamic penetrometric tests, explicit reference is made to the SPT test carried out according to
the ASTM procedure. The normalized blow count N60 (equivalent blow count that would be
measured by a 60% performing equipment at a depth corresponding with a effective vertical
pressure of 1 atm) requested by some tools is defined as:

𝑁60 = (𝜂⁄60) ∙ 𝐶𝑁 ∙ 𝑁𝑆𝑃𝑇 , where:

η: sampling device energy ratio (as a percentage between 0 and 100)
𝐶𝑁 : overburden correction factor
NSPT: sampled blow count (no. of blows to obtain 1 ft penetration)

Figure 22: Effective vertical stress – overburden correction factor diagrams.

Deep Excavation LLC Page 36

 DeepEX 2016 – Non Linear Analysis Manual
Several different dynamic penetration procedures which actually differ from the SPT test are
available and very frequently used. However the results given by such procedures should not
directly used to enter the correlations based on the SPT results, unless information is available
that can somehow correlate the used non-standard procedure to SPT. For non-standard dynamic
penetration procedures used in Italy, such kind of information is available after Cestari (1990).

Along with the blow count profile, a borehole log is currently necessary as well as granulometric
tests etc. Using the proposed correlation without a good knowledge of the soil nature may bring
to very unreliable conclusions.

As for the static cone penetration test (CPT), reference is made to the cone resistance qc that
must be measured by a standard procedure. CPT tests are usual for clays and loose sands.
Literature relationship between qc and Nspt (Figure 23) are available to set up a cross reference
between such different results. It may be worthwhile using this relationship to compare the
results given by a correlation based either on Nspt or on q c.

Figure 23: qc/NSPT to particle size diagram.

No explicit reference is made here to other field test technology like, for example, the
pressuremeter, which may provide engineer with even more valuable information than the
penetrometric tests. User is free to select the desired testing method and directly enter the soil
parameters without using the correlations proposed by DeepEX.

Deep Excavation LLC Page 37

 DeepEX 2016 – Non Linear Analysis Manual
7.4 Correlations for granular soils
DeepEX offers some estimation tools for the granular soil properties correlation based on
Standard Penetration Test (SPT) and Cone Penetration Test (CPT – used only for loose sand soils).

7.4.1 Resistance properties

Friction angle φ’ value is estimated in DeepEX through the following procedure:
a. Calculation of Dr:
 If SPT is available, then N60 is used for the estimation of Dr. Correlations after Cubrinovsky
& Ishihara (1999) are implemented which include a dependency with the relevant grain
size by means of emax-emin.
 If CPT is available, then qc is used for the estimation of Dr. Correlations for NC and OC
sands are given according to Baldi et al. (1988).

b. Calculation of φ’peak: DeepEX offers correlations after Bolton (1986) or Schmertmann

(1977) for the estimation of φ’peak.

c. Assessment for the design value of φ’: The design value for friction angle φ’ lays between
the values of φ’CV and φ’peak. As an initial estimation, an 1.5 safety factor is applied to
(φ’peak - φ’CV), thus:
𝜑 ′ = 𝜑 ′ 𝐶𝑉 + (𝜑𝑝𝑒𝑎𝑘 − 𝜑𝐶𝑉 )/1.5

7.4.2 Stiffness properties

In DeepEX there are available tools for the estimation of the Elastic moduli values (E vc, Eur)
according to NSPT or qc:
 𝐸 = 𝐶1 ∙ 𝑁𝑆𝑃𝑇 + 𝐶2
 𝐸 = 𝑎 ∙ 𝑞𝑐

Values for α are recovered in agreement with experimental data. Hence two different approaches
are used to compute the elastic modulus, they produce very similar predictions for such values.

Most of the available correlations give a prediction of a secant elastic modulus Es to be used in
the calculation of foundations settlements. By definition, the secant modulus depends on the
strain level at which it has been defined (Figure 24): hence the values given by such correlations
are defined at typical strain level for foundations (about the 30%-50% of the strains at failure),
usually higher than the typical strain values for the problems dealt with by DeepEX.

From Figure 25, we note that soil stiffness currently decreases when the strain is increased: it can
be therefore concluded that the secant moduli obtained by experimental correlations tuned for
foundations currently underestimate the secant moduli to be used in a retaining wall problem.
In other words, we can say that the correlations included in DeepEX just give a quite coarse
approximation of the moduli and very likely such values represent an underestimate of the real
values. A similar conclusion has been reached by back analysis studies on real excavation cases
(see, for example, Fenelli & Pagano, 1999).

Deep Excavation LLC Page 38

 DeepEX 2016 – Non Linear Analysis Manual

Figure 24: Secant modulus definition.

Figure 25: Secant shear modulus G decay with strain increase (Viggiani & Atkinson, 1995).
If stress dependent moduli are used, in the form 𝐸 = 𝑅 ∙ (𝑝′ /𝑝𝑎 )𝑛 , DeepEX presents some initial
estimated values of the R parameters for sand or gravels, related to D r (values from Lancellotta,
1988) to be used taking p’ = σ’h , with n varying from 0.4 to 0.6. The Eur/Evc, ratio is related to
OCR. For low OCR values, a fixed value equal to 1.60 is suggested. In general, OCR varies along
with the analysis however DeepEX assumes that Eur/Evc is kept constant equal to the initial value
assigned by the user.

7.5 Correlations for cohesive soils

DeepEX offers some estimation tools for the cohesive soil properties correlation based on Cone
Penetration Test (CPT).

7.5.1 Undrained resistance properties

The undrained shear strength can be calculated based on qc, according to the following equation:

𝑆𝑢 = (𝑞𝑐 − 𝜎𝑣 )/𝑁𝑘 , where:

𝑁𝑘 : Cone factor between 10 and 20. Some studies show that it depends on the plasticity index
pl. Initially, we assume Nk = 15.

An alternative linear relationship between Su and σ’v is given for NC clays.

The undrained friction angle φu is zero for a saturated soil.

Deep Excavation LLC Page 39

 DeepEX 2016 – Non Linear Analysis Manual
7.5.2 Drained resistance properties
φ'p can be calculated according to the following equation:
tan (φ′ p ) = tan(φ′ CV ) /1.5

7.5.3 Undrained stiffness properties

The elastic modulus Eu is related to Su by a linear relationship:
Eu/Su=k

The factor k decreases as long as OCR and pl increase. DeepEX includes the correlation after
Duncan & Buchigani (1976), related to secant moduli at 50% of the strain at failure (referred to
as Eu50).

7.5.4 Drained stiffness properties

If the undrained modulus Eu is known, an initial estimation of the drained modulus may be
represented by the 80% of Eu. DeepEX also includes some correlations with qc (after Bowles,
1988) as well as some indications for variable moduli 𝐸 = 𝑅 ∙ (𝑝′ /𝑝𝑎 )𝑛
In the latter case, for NC clays, typical values of R are calculated by = 𝑅 ∙ (𝜎 ′ 𝑣 /𝑝𝑎 ) , which
describes a linearly varying modulus with depth.

7.6 Thrust coefficients

7.6.1 At –rest coefficient

The at-rest coefficient for normal consolidated soils can be calculated using the following
equations:
 Sands: 𝐾0 𝑁𝐶 = 1 − sin⁡(𝜑′𝑝𝑒𝑎𝑘 )
 Clays: 𝐾0 𝑁𝐶 = 1 − sin⁡(𝜑′𝐶𝑉 )
The at-rest coefficient for over consolidated soils can be calculated using the following equation:
𝐾0 = 𝐾0 𝑁𝐶 ∙ 𝑂𝐶𝑅 𝑛 , with n=0.5 for both sands and clays.

7.6.2 Active thrust coefficient

The active thrust coefficient Ka depends on φ’, on δ and on the uphill soil slope. DeepEX includes
a tool to calculate Ka based on Coulomb formulas, which give a reasonable estimation for
practical purposes, with respect to other formulas, like the one by Caquot & Kerisel, 1948. The
friction angle δ between the soil and the wall slightly affects Ka and it is advised to neglect such
effect. Figure 27 presents a diagram that can be used to compute Ka accounting for a sloped soil.
For this purposes, the diagrams in appendix G of EUROCODE 7, Part 1 can be used as well.
For undrained behavior, where φu=0, DeepEX automatically sets Ka= Kp= 1.

Deep Excavation LLC Page 40

 DeepEX 2016 – Non Linear Analysis Manual
7.6.3 Passive thrust coefficient
Kp mostly affects the results especially in a cantilever or a single anchor wall analysis. Like Ka, it is
related to φ’, on δ and on the uphill soil slope. Applicable values may be estimated using the
diagram of Figure 27 or directly by the tool provided by DeepEX. It’s worthwhile mentioning that
such values, based on logarithmic failure surfaces, show a good agreement with experimental
data. In addition, they are more conservative than the values predicted by the Coulomb
equations based on planar slip surfaces.

Figure 6-5 proposes a comparison between the two approaches (logarithmic or planar surfaces)
is for a horizontal ground surface. When the friction angle is greater than 30° and δ/φ’>0,
Coulomb values may be significantly not conservative. Additional references for Kp calculation
can also be found in appendix G of EUROCODE 7, Part 1, where the suggested procedure
produces similar values to Caquot & Kerisel theory. All the available approaches are approximate
because not all the soil behavior components are included. For this reason, a great care is
recommended when selecting the value for Kp for design purposes.

Figure 26: Passive thrust coefficient – Friction angle diagram (Caquot-Kerisel and Coulomb).

Deep Excavation LLC Page 41

 DeepEX 2016 – Non Linear Analysis Manual

Figure 27: Ka and Kp values (after Navfac (1986) DM 7.02).

Deep Excavation LLC Page 42

 DeepEX 2016 – Non Linear Analysis Manual
Table 1: Reference values for δ after Navfac (1986).
Friction
Friction
Interface Materials angle δ
factor, tanδ
(deg)
Mass concrete on the following foundation materials:
Clean sound rock............................................................................. 0.7 35
Clean gravel, gravel-sand mixtures, coarse sand............................ 0.55 to 0.6 29 to 31
Clean fine to medium sand, silty medium to coarse sand, silty or
clayey gravel................................................................................... 0.45 to 0.55 24 to 29
Clean fine sand, silty or clayey fine to medium sand...................... 0.35 to 0.45 19 to 24
Fine sandy silt, nonplastic silt.......................................................... 0.3 to 0.35 17 to 19
Very stiff and hard residual or preconsolidated clay...................... 0.4 to 0.5 22 to 26
Medium stiff and stiff clay and silty clay......................................... 0.3 to 0.35 17 to 19
(Masonry on foundation materials has same friction factors.)
Steel sheet piles against the following soils:
Clean gravel, gravel-sand mixtures, well-graded rock fill with
spalls.............................................................................................. 0.4 22
Clean sand, silty sand-gravel mixture, single size hard rock fill..... 0.3 17
Silty sand, gravel or sand mixed with silt or clay 0.25 14
Fine sandy silt, nonplastic silt........................................................ 0.2 11
Formed concrete or concrete sheet piling against the
following soils:
Clean gravel, gravel-sand mixture, well-graded rock fill with
spalls............................................................................................... 0.4 to 0.5 22 to 26
Clean sand, silty sand-gravel mixture, single size hard rock
fill.................................................................................................... 0.3 to 0.4 17 to 22
Silty sand, gravel or sand mixed with silt or clay 0.3 17
Fine sandy silt, nonplastic silt........................................................ 0.25 14
Various structural materials:
Masonry on masonry, igneous and metamorphic rocks:
Dressed soft rock on dressed soft rock........................................... 0.7 35
Dressed hard rock on dressed soft rock.......................................... 0.65 33
Dressed hard rock on dressed hard rock........................................ 0.55 29
Masonry on wood (cross grain)...................................................... 0.5 26
Steel on steel at sheet pile interlocks............................................. 0.3 17
Interface Materials (Cohesion) Adhesion ca (psf)
Very soft cohesive soil ( 0 - 250 psf) 0 – 250
Soft cohesive soil ( 250 - 500 psf) 250 – 500
Medium stiff cohesive soil ( 500 - 1000 psf) 500 – 750
Stiff cohesive soil (1000 - 2000 psf) 750 – 950
Very stiff cohesive soil (2000 - 4000 psf) 950 – 1000

Deep Excavation LLC Page 43

 DeepEX 2016 – Non Linear Analysis Manual
7.7 Reference permeability values
In DeepEX it is required to assign permeability for each soil layer. This value is required in the
pore pressure calculation. Table 2 presents some typical permeability values for several soil
types.
Table 2: Typical permeability values for soil types.
Soil Type Permeability
(cm/sec)
Clean gravel 10-1
Sand and gravel 10-2
Clean sand 10-3
Silty sand 10-4
Silt 10-5
Clay 10-8

It is useful sometimes to redefine the permeability in a certain soil region to model, for example,
some soil improvement by cement grouting or by the jetgrouting technique. The obtained
permeability after improvement obviously depends on the technology with respect to the in situ
soil nature. it is usually quite difficult to predict in advance the improved soil permeability,
whereas an a posteriori measurement is currently needed. However, just for a very first guess
estimate, it can be anticipated that a well done jetgrouting improvement in a gravely soil reduces
the natural soil permeability by two orders of magnitude at least.

Deep Excavation LLC Page 44

 DeepEX 2016 – Non Linear Analysis Manual
8. ESTIMATION OF VERTICAL SETTLEMENTS IN DEEPEX

DeepEX analyses the lateral soil-wall interaction. However, sometimes it is necessary to make an
estimation of the foundation settlements induced by the opencut excavation. DeepEX gives some
valuable information to indirectly estimate such results. Based on lateral wall deformations, an
estimate of vertical displacements can be attempted using some proposed methods like those
reported in Bransby & Milligan (1975), which are based on experimental results and small scale
models.

Figure 28 presents an approach to calculate the vertical settlements in granular soils. According
to Figure 28, vertical settlements in both downhill and uphill soil region can be evaluated
calculating the soil volumes subject to lateral movements.

Figure 28: Calculation of the vertical settlements in granular soil.

Calculate the area A of the ABC triangle based on the lateral displacements already calculated in
DeepEX (linearly interpolate between zero displacement point C and maximum deflection point
B). Assume that the soil beyond linear boundaries CD e CE remains undeformed. If deformation
developed at constant volume, the uphill wedge would preserve its original area: thus, assuming
a straight deformed surface between B’ and D, constant volume condition would require that
area of BB'D = area of ABC = A, thus:

𝐵𝐵 ′ = 2 ∙ 𝐴/𝜆𝑚

Similar assumption could be made for EGC wedge in the downhill soil. However, the soil, during
deformation, currently increases its volume (dilatancy). Calling ψ be the angle of dilatancy,
between the volumetric strain 𝑣 = (𝜀𝑣 + 𝜀ℎ )/2 and maximum shear strain |𝛾| = |(𝜀𝑣 − 𝜀ℎ )/2|
a simple equation holds:

Deep Excavation LLC Page 45

 DeepEX 2016 – Non Linear Analysis Manual
𝑣 = −|𝛾|𝑡𝑎𝑛𝜓 → 𝜀𝑣 = −𝜀ℎ (1 − 𝑡𝑎𝑛𝜓)/(1 + 𝑡𝑎𝑛𝜓)

Assuming a linear deformed shape, the average uphill lateral deformation can be calculated by:

𝜀ℎ 𝑀 = 𝛿𝑚 /𝜆𝑚 = constant, then: 𝜀𝑣 𝑀 = −𝜀ℎ 𝑀 (1 − 𝑡𝑎𝑛𝜓)/(1 + 𝑡𝑎𝑛𝜓)

Recalling that the settlements was assumed to linearly vary between B and D, in B, the maximum
settlement can be calculated by:

𝑢𝑚 = 𝜀𝑣 𝑀 ∙ 𝜆𝑚

Whereas in D, a null settlement will be calculated.

As mentioned above, ψ is the dilatancy at critical state. We assume that:

𝜓 = 𝜑 − 𝜑𝐶𝑉 , in which 𝜑𝐶𝑉 is the constant volume friction angle (typically 30 deg for a sand). ψ
cannot have a negative value.

This engineering approach cannot be generalized: when the soil-structure interaction is complex,
a simple solution scheme like the one by DeepEX could be unsuitable to produce a reasonable
prediction of the real behavior; therefore and a more complete analysis tool should be selected.

Deep Excavation LLC Page 46

 DeepEX 2016 – Non Linear Analysis Manual
9. STRIP FOUNDATION MODELING IN DEEPEX

The effective vertical stress σ’v is calculated using the soil layer weights, the applied surcharge on
the ground and the water table. The deformation of the soil mass is not taken into account.
Initially, the lateral effective stress is calculated using the following equation:

𝜎′ℎ = 𝐾0 ∙ 𝜎′𝑣 , where K0 is the at-rest coefficient. In the subsequent stages, σ’v changes
according to the lateral wall deflection. This approximation is not accurate enough when the
initial soil stress cannot be described accurately, i.e. when a concentrated surcharge exists near
next to the wall. In this case, both the vertical and horizontal stress distribution becomes more
complex than the simple geostatic stress field. The following approaches can be used in DeepEX
in order to include such effects:

1. If the loaded area is quite far from the wall and the applied surcharge is not too high, a
uniform equivalent surcharge can be prescribed (Figure 29).

Figure 29: Equivalent uniform surcharge.

2. If the loaded area is close to the wall or the applied surcharge is high, it can be assumed
that the concentrated load produces an additional stress distribution in the soil mass that
can be computed by means of the Theory of Elasticity (see Timoshenko & Goodier (1970)).
This additional stress field is added to the geostatic field (which depends on K0), to set up
the initial stress condition for the analysis.

When using the second approach, initially in all soil layers, both in uphill and downhill side, the
lateral effective stress is calculated using the following equation:

𝜎′ℎ = 𝐾0 ∙ 𝜎 ′ 𝑣 + 𝛥𝜎′ℎ,𝑖𝑛𝑖𝑡

Deep Excavation LLC Page 47

 DeepEX 2016 – Non Linear Analysis Manual
In the previous equation, 𝜎 ′ 𝑣 does not account for increment due to the foundation load. 𝛥𝜎′ℎ,𝑖𝑛𝑖𝑡 is
calculated by the elastic equation from the half-space solution with uniform strip load at the free surface
(Figure 30). Such value of 𝜎 ′ ℎ is just used to establish initial stress condition for each soil layer. The lateral
stress may later change due to the wall deflections and the 𝛥𝜎′ℎ,𝑖𝑛𝑖𝑡 part may disappear.

Figure 30: Half space solution for plain strain condition.

Figure 31: Calculation of 𝛥𝜎′𝑣 .

Deep Excavation LLC Page 48

 DeepEX 2016 – Non Linear Analysis Manual
With reference to Figure 31, if Z > ZF, no additional vertical stress is considered. Otherwise, a
stress increment 𝛥𝜎′𝑣,1 is calculated as follows:
If 𝑍 > 𝑍𝐹 − 𝐷𝑌 ∙ tan⁡(𝛽) then 𝛥𝜎′𝑣,1 = 0 , else 𝛥𝜎′𝑣,1 is calculated using the following equation:
𝐷𝑌+𝐵 𝑄𝐹(𝑦)
𝛥𝜎′𝑣,1 = ∫𝐷𝑌 𝐿(𝑦)
𝑑𝑦 , where:

𝑄𝐹(𝑦) = 𝑄𝐹 if 𝑦 ≥ (𝑍𝐹 − 𝑍)/tan⁡(𝛽)

𝑄𝐹(𝑦) = 0 if 𝑦 < (𝑍𝐹 − 𝑍)/tan⁡(𝛽)
𝐿(𝑦) = 𝑦 + (𝑍𝐹 − 𝑍)/tan⁡(𝛽)

The integral is calculated by subdividing the integration range into 100 segments. The obtained
value is compared with 𝜎 ′ 𝑣 given by formula in Figure 30 and the maximum is selected:
𝛥𝜎′𝑣 = max⁡(𝛥𝜎 ′ 𝑣,1 , 𝜎 ′ 𝑣 )
This stress increment decreases with depth.

This formulation assumes a uniform pressure distribution on the loaded area, suitable for the
modelling of a very flexible foundation. If the foundation is rigid, the actual pressure distribution
is given in Figure 32. In this case, such not uniform distribution may be approached by subdividing
the loaded area into several uniformly loaded sub-areas.

Figure 32: Pressure distribution under a rigid foundation.

Deep Excavation LLC Page 49

 DeepEX 2016 – Non Linear Analysis Manual
10. MODELING A BERM IN DEEPEX

The dredge line inside the excavation is sometimes modelled to provide a limited soil mass
adjacent to the wall, higher than the general excavation depth. Such a soil mass is called “berm”
and may be included in the calculation during some intermediate steps. Input data to DeepEX
may be given according to one of the following criteria:
 Method 1: An equivalent horizontal dredge line is assumed (Figure 33).
 Method 2: A uniform distributed load equal to the berm weight divided by the passive
wedge extension L is applied at the excavation level (Figure 34).
 Method 3: An equivalent sloped dredge line is assumed. Excavation level can be raised up
to point A (Figure 35), but passive thrust coefficient must be reduced to account for
unfavorable sloping.

Figure 33: Berm in DeepEX, Model 1.

Deep Excavation LLC Page 50

 DeepEX 2016 – Non Linear Analysis Manual

Figure 34: Berm in DeepEX, Model 2.

Figure 35: Berm in DeepEX, Model 3.

Deep Excavation LLC Page 51

 DeepEX 2016 – Non Linear Analysis Manual
11. DEEPEX: RESULT ASSEDDMENT

At each analysis step, the following results are both printed out on the output file and stored in
the result database:
1. NODAL DISPLACEMENT COMPONENTS: i.e. total lateral y displacement and total x
rotation.
2. FINITE ELEMENT RESULTS: the issued results depend on each element type
3. REACTIONS AT FIXED NODAL POINTS: lateral reaction (force per unit out-of-plane depth)
and moment (moment per unit out-of-plane depth)

Element available results are listed in the following.

 TRUSS ELEMENT
1. FORCE : ELEMENT FORCE PER UNIT OUT-OF-PLANE DEPTH (+VE TENSION)
2. STRESS : ELEMENT STRESS

Figure 36: Beam element results.

 BEAM ELEMENT (see Figure 36)

1. VA : SHEAR AT FIRST EDGE
2. VB : SHEAR AT LAST EDGE
3. MA : MOMENT AT FIRST EDGE
4. MB : MOMENT AT LAST EDGE
(all of them, per unit out-of-plane depth)

 ELPL ELEMENT (ELASTOPLASTIC SUPPORT)

1. FORCE : ELEMENT FORCE PER UNIT OUT-OFPLANE DEPTH (+VE TENSION)
2. PLASTIC: PLASTIC STRAIN

Deep Excavation LLC Page 52

 DeepEX 2016 – Non Linear Analysis Manual
 WIRE ELEMENT (GROUND ANCHOR)
FORCE : ELEMENT FORCE PER UNIT OUT-OF-PLANE DEPTH (+VE TENSION)

 CELAS ELEMENT (ELASTIC SUPPORT)

1. FORCE : ELEMENT FORCE PER UNIT OUT-OF-PLANE DEPTH (+VE TENSION)
2. MOMENT : ELEMENT MOMENT PER UNIT OUT-OF-PLANE DEPTH

 SLAB ELEMENT (A SLAB BETWEEN TWO WALLS)

1. VA : SHEAR AT FIRST EDGE
2. VB : SHEAR AT LAST EDGE
3. MA : MOMENT AT FIRST EDGE
4. MB : MOMENT AT LAST EDGE
5. AXIAL : AXIAL FORCE
(all of them, per unit out-of-plane depth)

 NODAL VARIABLES
1. YDISPL : LATERAL DISPLACEMENT
2. XROT : ROTATION
3. YREACT : HORIZONTAL REACTION
4. XMOMREAC: MOMENT REACTION

As a first check, it is suggested to carefully assess the initial step results. In this step lateral
deflections should be zero; hence, in the soil elements, if no strip foundations have been
prescribed, the at-rest geostatic conditions should be found:

𝜎′ℎ
( ) = ⁡ 𝐾0
𝜎′𝑣 𝑆𝑇𝐴𝐺𝐸⁡0
It may be worthwhile controlling the resultant of the stresses in downhill or uphill soil elements.
Such resulting thrusts are automatically issued in the final report file. Figure 37 presents the
stress distribution between elevations ZA and ZB.

Figure 37: Stress distribution between elevations ZA and ZB.

Deep Excavation LLC Page 53

 DeepEX 2016 – Non Linear Analysis Manual
The following table presents the calculated items:

ITEM EQUATION DESCRIPTION

𝑍𝐵
True Effective Effective stress resultant over
𝑆 = ∫ 𝜎′ℎ 𝑑𝑧
Thrust 𝑍𝐴 all the soil elements in a group
𝑍𝐵
Pore pressure resultant over all
Water Thrust 𝑊 = ∫ 𝑢𝑑𝑧
𝑍𝐴 the soil elements in this group

True Total Thrust 𝑇 =𝑆+𝑊 Overall thrust on the wall

𝑍𝐵 Minimum thrust from this soil
Minimum Allowable
𝐴 = ∫ (𝐾𝐴 ∙ 𝜎 ′ 𝑣 − 2 ∙ 𝑐′ ∙ √𝐾𝐴 )𝑑𝑧 region, if active conditions are
Thrust 𝑍𝐴 fully developed
Maximum thrust that can be
𝑍𝐵
Maximum Allowable resisted by this soil region, if
𝑃 = ∫ (𝐾𝑃 ∙ 𝜎 ′ 𝑣 + 2 ∙ 𝑐′ ∙ √𝐾𝑃 )𝑑𝑧
Thrust 𝑍𝐴 passive conditions are fully
developed
Ratio between the passive
Maximum/True 𝑃 thrust and current effective
Ratio 𝑆 thrust: it can be used as a
safety factor
The actual effective thrust is
Passive Thrust 𝑆 represented by the percentage
100 ∙
Percentage 𝑃 of the maximum available
resistance
Ratio between the current
True/Minimum 𝑆
effective thrust and minimum
Ratio 𝐴 soil resistance

If failure conditions are almost reached, as in a limit analysis, soil element pressures and wall
bending moments can be compared with hand calculated distributions by traditional methods
(see Bowles (1988)). Pore pressures can be easily reproduced by hand calculations too, because
the simple formulas reported in this manual are used by the program as well.

Frequent checks between expected results and DeepEX results are encouraged. The
postprocessing features included in DeepEX allow an extensive result assessment in both
graphical and numerical depictions.

Deep Excavation LLC Page 54

 DeepEX 2016 – Non Linear Analysis Manual
At the end of each analysis, a summary report is printed out, where information about
convergence is outlined. If a not converged solution is detected at a certain step, a very clear
warning message is issued. A not converged solution means that, for that step, no satisfactory
equilibrium condition has been reached, using the maximum number of iteration allowed. Note
that DeepEX does not stop if non convergence is reached since a converged (well balanced)
solution may be recovered in the next step: however the results for the non-converged solution
step are unreliable and are not released to the User.

When a not converging solution is detected, user may try the following actions:
1. Increase the maximum number of iterations (note that such remedy is rarely effective, it
may be successful in cases where a very flexible wall is inserted in a very stiff soil)
2. Change the solution strategy by, for example, subdividing an excavation step into more
intermediate excavations
3. Check relevant soil parameters (like Kp etc.) and eventually slightly modify them.

When clear failure conditions are reached, a non-positive definite stiffness matrix is found by
DeepEX. In this case the software suddenly stops. When NON CONVERGENCE or FAILURE
conditions are reached, a true collapse of the wall is very likely to have been encountered. In this
case, the input data must be carefully revised since the wall design may require some
modifications.

Finally, as mentioned above, soil parameters are currently affected by great uncertainty, thus,
even very satisfactory solution results should be very carefully assessed and discussed with
criticism; sensitivity analyses on most relevant and/or uncertain soil parameters are strongly
advised, especially when new or non-familiar soil conditions are to be dealt with.

Deep Excavation LLC Page 55

 DeepEX 2016 – Non Linear Analysis Manual
12. SAFETY FACTORS IN DEEPEX

DeepEX is an engineering tool in which the final design of the retaining wall arrangement is
obtained by a progressive refinement of an initial assumed trial configuration. It provides user
with all the information for a precise assessment on the adequacy of an initial model and any
refined embedment.

Since the estimated soil parameters are accurate, the information given by the non-linear engine
of DeepEX are in general more accurate than those available by traditional design methods based
on limit equilibrium concepts. Such methods, for example, do not give reliable information about
the lateral movements of the wall. On the other hand, limit equilibrium methods currently issue
some safety factor of the wall, a parameter that is not directly given by the non-linear engine.
For simple retaining walls, a safety factor may be defined in several ways (see Bowles, 1988).
Here follow some examples:
 Calculate the embedment length of a cantilever wall, based on a reduced passive thrust
coefficient Kp divided by a safety factor and then use such length as the final length
 Calculate the minimum length using an unreduced Kp and then prescribe an increased
embedment.

In the case of walls with several supports, the safety factor in the traditional sense (the ratio of
the true embedment to the minimum embedment) may be meaningless.

For very simple wall schemes (cantilever or single propped walls) a special procedure is included
in the non-linear engine of DeepEX to calculate a safety factor in the traditional sense as follows:
1. The wall is analyzed in all the realistic excavation phases, taking into account the real
(unreduced) embedment length.
2. From the last step on, additional steps are solved in which the dredge line is kept constant
at the final design depth, but the wall embedment is progressively reduced (upon request,
DeepEX automatically removes all the finite elements below a prescribed depth: User
must activate the Find Safety Factor option in dialog box where analysis steps are defined
and, at each step, input the cut depth in the Zcut field)
3. Corresponding with the minimum embedment depth, a solution is no long possible which
meets both equilibrium and plasticity requirements (DeepEX may fail to converge or may
compute a converged solution with unreasonably high deformations)

This way, the User may recognize the minimum wall configuration required for equilibrium.
Collapse conditions may be reached in many alternative ways like:
 increasing a surcharge
 decreasing the soil resistance parameters (increasing Ka and reducing Kp and cohesion).

The resulting safety factors will be different depending on the way they have been computed.
It’s worth comparing the results by different methods.
When the wall has two supports at least, no failure is possible only due to limit conditions in the
soil. In such cases, to use the limit equilibrium approach like the one outlined above, it would be

Deep Excavation LLC Page 56

 DeepEX 2016 – Non Linear Analysis Manual
necessary to include limit conditions for props as well. A different approach is more meaningful
for such situations. The safety of the wall may be effectively related to a stability number given
by the following ratio:

𝑃𝑎𝑠𝑠𝑖𝑣𝑒⁡𝑇ℎ𝑟𝑢𝑠𝑡⁡𝑅𝑒𝑠𝑢𝑙𝑡𝑎𝑛𝑡
𝑇𝑟𝑢𝑒⁡𝑇ℎ𝑟𝑢𝑠𝑡⁡𝑅𝑒𝑠𝑢𝑙𝑡𝑎𝑛𝑡

This ratio should be calculated for the downhill soil region below the excavation level. Such ratio
may be revised as a safety factor with respect to limit (passive) condition in the embedded toe.
If an unfactored Kp coefficient has been included in the analysis, acceptable values for such ratio
may fall within 2 and 3, depending on the problem. If a reduced Kp value has been used in DeepEX,
values between 1.5 and 2 may be fairly acceptable.

In the final report, DeepEX issues such ratio at every analysis step. In the final judgement on the
wall safety, the calculation of the safety factor above is not sufficient. A global stability check of
the wall plus the soil within a potential slip surface should be performed by means of traditional
methods (Bishop (1955), Fellenius (1936), Morgenstern & Price (1965), etc. or complex numerical
analyses. (Cundall & Board (1988)).

Figure 38: Failure mechanism not detectable by the non-linear engine of DeepEX.

A careful assessment of the ground anchor safety as well as the stability of struts, if any, should
be included as well in the final stability report.

When water table lowering has to be dealt with, additional considerations must be explicitly
included about:
 stability of the excavation against uplift;
 safety factor with respect to quick conditions (DeepEX gives very useful information for
this aspect)
 adequacy of water lowering devices with respect of fine removal from soil composition
etc.

Deep Excavation LLC Page 57

 DeepEX 2016 – Non Linear Analysis Manual
13. REFERENCES

BALDI G., BELLOTTI R., GHIONNA V., JAMIOLKOWSKI M., PASQUALINI E. (1981) “Cone Resistance
in Dry NC and OC sands”, ASCE Symp. on Cone Penetration Testing and Experience, St. Louis.
BARLA G., BECCI B., COLOMBO A., NOVA R., PEDUZZI R. (1988) “A method for the analysis and
design of flexible retaining structures. Application to a strutted excavation” , Proceedings of the
Sixth Conference on Numerical Methods in Geomechanics, Innsbruck, Vol. 3, pp. 1635-1640.
BATHE, K.J. (1996), “Finite element procedure”, Prentice Hall
BECCI, B., NOVA, R. (1987) “Un metodo di calcolo automatico per il progetto di paratie”, Rivista
Italiana di Geotecnica, 1, 33-47
BISHOP, A. W. (1955) “The use of the slip circle in the stability analysis of slopes”, Geotechnique,
5, pp. 7-17
BOLTON, M.D., (1986) “The strength and dilatancy of sands”, Geotechnique36, 1, 65-78.
BOWLES J.E. (1988),”Foundation Analysis and design”, 4th ed. McGraw-Hill
BRANSBY P.L., MILLIGAN G.W.E. (1975) “Soil Deformations near Cantilever Sheet Pile Walls”,
Geotechnique 25, 2, 175-195.
CAQUOT A. & KERISEL J. (1948) “Tables for the Calculation of Passive Pressure, Active Pressure
and Bearing Capacity of Foundations”, Gautiers-Villars, Paris
CAQUOT A., KERISEL J., ABSI E., (1973) “Tables de butée et de poussée”, Gautiers-Villars, Paris
CESTARI F. (1990) “Prove geotecniche in sito”, Geo-Graph
CESTELLI-GUIDI C. (1984) “Geotecnica e tecnica delle fondazioni”, Hoepli
CUNDALL P., BOARD M. (1988) “A microcomputer program for modelling large-strain plasticity
problems”, Proceedings of the Sixth Conference on Numerical Methods in Geomechanics,
Innsbruck, Vol. 3, pp. 2101-2108.
CUBRINOVSKY M.C. & ISHIHARA K, (1999) “Empirical correlations between SPT N-value and
relative density for sandy soils”, Soils and Foudations 39, 5, 61-71
D’APPOLONIA D.J., D’APPOLONIA E., BRISETTE R.F. (1970) “Discussion on settlements of spread
footings in sand”, ASCE J. SMFD 96.
COMITATO EUROPEO DI NORMAZIONE (CEN) (1994) “ENV 1997 - EUROCODICE 7 – Progettazione
geotecnica” Parte 1: Regole generali (Norma UNI ENV 1997-1, aprile 1997).
DUNCAN J. M. & BUCHIGANI A. L., (1976) “An Engineering Manual for Settlements Studies”, Dept.
of Civil Engineering, Univ. of California, Berkeley
FELLENIUS, W. (1936) “Calculation of the stability of earth dams”, Proceedings of the Second
Congress in Large Dams, 4, pp. 445-464
FENELLI G.B. & PAGANO L., (1999) “Computing Top-Beam Effects in Retaining Walls”, Journal of
Geotechnical and Geoenvironmental Engineering, ASCE, 125, 8, 665-672
FLEMING W.G.K., WELTMAN A.J., RANDOLPH M.F., ELSON W.K., (1992) “Piling Engineering”, 2nd
ed., Blackie
JAMIOLKOWSKI M., LANCELLOTTA R., MARCHETTI S., NOVA R., PASQUALINI E. (1979) “Design
parameters for soft clays” 7 ECSMFE, Brighton, 5,27-57
JAMIOLKOWSKI M., PASQUALINI E. (1979) “Introduzione ai diversi metodi di calcolo dei
diaframmi con riferimento ai parametri geotecnici che vi intervengono e alla loro determinazione
sperimentale” Atti Istituto Scienza delle Costruzioni, Politecnico di Torino, n. 451

Deep Excavation LLC Page 58

 DeepEX 2016 – Non Linear Analysis Manual
JANBU N. (1963) “Soil compressibility as determined by oedometer and triaxial tests” 3 ECSMFE,
Wiesbaden, 1, 19-25
LADD C.C., FOOT R., ISHIHARA K., SCALLOSSER F, POULOS H.G, (1977) “Stress-deformation and
Strength characteristics” State-of-the-art volume 9 ICSMFE, Tokyo, 421-494
LAMBE T.W., WHITMAN R.V. (1969) “Soil Mechanics”, John Wiley & Sons, Inc.
LANCELLOTTA R., (1988) “Geotecnica”, Zanichelli
LIAO, S. C. and WHITMAN, R.V. (1986). "Overburden correction factors for SPT in sand" Journal
of Geotechnical Engineering, ASCE, Vol 112, No. 3, pp.373-377
MORGENSTERN, N. R., PRICE, V. E., (1965) “The analysis of the stability of general slip surfaces”,
Geotechnique, 15, 70-93
NAVFAC (1986) “Design Manual D: M 7.02 -Foundations and earth structures”, USA
NOVA, R., (1978) “Geotecnica”, ed. CLUP, Milano
NOVA, R., BECCI, B. (1987) “A method for analysis and design for flexible retaining structures”,
Proc. Conf. INTERACTIONS SOLS-STRUCTURES, pp. 657-664
RANDOLPH M. F., (1981) “The response of flexible piles to lateral loading” , Geotechnique 31, 2,
247-259.
SCHMERTMANN J.H. (1977) “Interpreting the Dynamics of Standard Penetration Test“, Univ. of
Florida, Gainsville (USA)
SCOTT R.F. (1981) “Foundation analysis”, Prentice Hall
TIMOSHENKO, S. P., GOODIER, J. N. (1970), “Theory of Elasticity”, 3th ed. McGraw-Hill
VIGGIANI G. & ATKINSON J. H., (1995) “Stiffness of fine-grained soil at very small strains”,
Geotechnique, 45, 2, 249-265

Deep Excavation LLC Page 59