PLAXIS

CONNECT Edition V20.04

User Defined Soil Models - PM4Silt: A Silt Plasticity model for Earthquake
Engineering
Last Updated: December 29, 2023
 Table of Contents
Chapter 1: Introduction ............................................................................................................ 3
1.1 Notation .......................................................................................................................................................................................4
1.2 Basic equations of elasto-plasticity ................................................................................................................................. 5
Chapter 2: Model Formulation ..................................................................................................7
2.1 Critical State Soil Mechanics framework ...................................................................................................................... 7
2.2 Bounding, Dilatancy and yield surfaces ......................................................................................................................12
2.2.1 Yield surface ................................................................................................................................................ 12
2.2.2 Dilatancy Surface .......................................................................................................................................13
2.2.3 Bounding Surface(BS) ............................................................................................................................. 13
2.3 Fabric-Dilatancy tensor ..................................................................................................................................................... 15
2.4 Elasticity ................................................................................................................................................................................... 15
2.5 Hardening rule .......................................................................................................................................................................16
2.6 Flow rule ...................................................................................................................................................................................17
2.6.1 Dilation: D<0 ............................................................................................................................................... 17
2.6.2 Contraction: D>0 ....................................................................................................................................... 18
2.7 Post shaking analyses ......................................................................................................................................................... 18
Chapter 3: Model parameters and state variables ...................................................................20
3.1 Model Parameters ................................................................................................................................................................ 20
3.1.1 Primary parameters .................................................................................................................................22
3.1.2 Secondary parameters ............................................................................................................................ 24
3.1.3 Post-shaking analysis parameters ..................................................................................................... 27
3.1.4 Alternative definition of a variable Undrained Shear Strength ............................................ 28
3.2 State variables ........................................................................................................................................................................28
Chapter 4: Model response in undrained monotonic and cyclic loading conditions ................. 32
4.1 Simulations of stress-controlled CDSS tests and CRR vs N curves ..................................................................34
4.2 Normalized shear modulus reduction and damping ratio curves from PLAXIS SoilTest simulations 38
Chapter 5: 1D site response analysis ........................................................................................ 41
5.1 Results ....................................................................................................................................................................................... 46
Chapter 6: References .............................................................................................................51

Appendices ......................................................................................................... 53
Appendix A: Dilatancy expressions ......................................................................................... 54
Appendix B: Anisotropic undrained strength for non-linear deformation analyses ..................56

PLAXIS 2 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
 Introduction
1
The seismic performance of civil and geotechnical structures during earthquakes significantly depends on the
undrained cyclic behavior of soils. Earth dams, river embankments, shallow foundations represent only a few
examples for which accumulation of shear strains and generation of high excess pore water pressures induced
by earthquakes can compromise the performance of the geotechnical systems. In both fine and coarse-grained
soils, significant deformations are associated with the effective stress reduction and the pore pressures
developed during ground motions. Specifically, in the case of low-plasticity silts and clays, the undrained
behavior is influenced by the plastic characteristics of the soil and the mechanical response can manifest a
transition between clay-like and sand-like behavior. Some differences between these two behavioral trends are
schematically depicted in Figure 1 (on page 3). While clays are characterized by a relatively unique
representation of the Normal Compression Locus (NCL), preserving also the parallelism between NCL and the
Critical State Locus (CSL), the yielding of sands is associated with an infinite number of NCL, depending on the
initial void ratio.

Figure 1: Schematic idealization of clay-like and sand-like behaviors(figure after Jefferies and Been, 2015)

Experimental evidence (Romero (1995 (on page 52)), Boulanger et al. (2006) (on page 51), Dahl et al. (2014)
(on page 51), Price et al. (2015) (on page 52), Price et al. (2017) (on page 52), Boulanger et al. (2016) (on
page 51)) shows that the mechanical behavior of silts is strongly influenced by the plastic characteristics of the
soil and that, for increasing values of the Plastic Index (PI), the tendency moves toward a clay-like behavior.
Sand-like and clay-like characteristics of the undrained cyclic loading response are emphasized in Figure 2 (on
page 4) where results of direct simple shear cyclic tests are shown for two normally consolidated silts having
different PI. The difference between the two mechanical responses is appreciable by observing the minimum
vertical effective stress reached at the end of the stress path and the evolution of the hysteresis loop which, for

PLAXIS 3 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Introduction
Notation

the non-plastic silt, are typical of sand-like behavior. Low-plasticity silts and clays can exhibit behaviors ranging
from sand-like in some aspects to clay-like in others (Boulanger et al. (2006) (on page 51), Boulanger et al.
(2018) (on page 51)). For instance, they can show stress-history normalized mechanical behavior (clay-like
aspect) but still accumulate significant levels of excess pore water pressures during cyclic simple shear tests, as
in sands (Dahl et al. (2014) (on page 51)), Boulanger et al. (2019b) (on page 51)).
To cope with these complex features, a constitutive model has been recently formulated in Boulanger et al.
(2018) (on page 51) which adapts a former model for sands (i.e. PM4Sand) to simulate the mechanical
behavior of low-plasticity silts and clays under monotonic and cyclic loadings. This constitutive model, known in
the literature as PM4Silt, has been implemented in PLAXIS 2D and is presented in the following sections. The
constitutive model formulation assumes the hypothesis of stress-history normalized undrained shear strength,
therefore PM4Silt is not suited for non-plastic silts. For these silts, the former PM4Sand model, also available in
PLAXIS 2D, should be preferred. After presenting the theoretical framework and explaining the meaning of the
model parameters, some numerical analyses are reported to show the performance of the PM4Silt model.

Figure 2: Cyclic stress-strain and stress path plots for two different normally consolidated silts resulting from
undrained cyclic direct simple shear tests (Boulanger et al. 2016)

1.1 Notation
PM4Silt is a constitutive model formulated for two-dimensional problems and, for this reason, the stress tensor
is defined as shown in Eq. [1]:

σ= (
σxx σxy
σxy σ yy ) Eq. [1]

whereas the mean effective stress is shown in Eq. [2]:

σxx + σ yy
p= 2
Eq. [2]

PLAXIS 4 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Introduction
Basic equations of elasto-plasticity

and the deviatoric stress tensor and the stress deviator are, respectively shown in Eq. [3].

s = σ - pI = (σxx − p
σxy
σxy
σ yy - p )
,q= 2s:s= 2⋅ |s| Eq. [3]

The symbol I represents the identity matrix ( I = δij , δij being the Kronecker symbol ) and the symbol “:” is used
to indicate the inner product between tensors (i.e. s : s = sij sij ) through which it is possible to define the
Euclidean norm | s | = s:s. To characterize the hardening mechanism and the plastic flow of the model, the
deviatoric stress ratio r and the related norm η (i.e the stress ratio) are defined respectively as in Eq. [4].

( )
σxx - p σxy
s p p
r= p
= ,η= 2⋅ |r| Eq. [4]
σxy σ yy - p
p p

Similar definitions apply also for the strain tensor ε which is expressed as in Eq. [5].

ε= (εxx εxy
εxy ε yy ) Eq. [5]

where the volumetric strain is defined as shown in Eq. [6].

εv = εxx + ε yy Eq. [6]

And the deviatoric strain tensor e ( Boulanger et al. (2018) (on page 51)) as shown in Eq. [7].
εv
εx - εxy
εv 3
e=ε- 3
I= Eq. [7]
εv
εxy εy - 3

It is important to note that the stress invariants p and q are not the same as those shown by PLAXIS SoilTest and
PLAXIS Output. If needed, these can be computed considering that p and q are defined as the center and the
diameter of the effective stress Mohr's circle, respectively. In what follows, all stress measures are considered as
effective and, for the sake of simplicity, the superscript " ' ", commonly used in soil mechanics notation, is
omitted. Furthermore, the symbol ''·" is used to denote incremental quantities.

1.2 Basic equations of elasto-plasticity

The main equations of the elasto-plastic theory are recalled hereafter:
1. Hypothesis of strain additivity: ε̇ = ε̇el + ε̇ pl .
2. Yield surface: f ( σ, α) = 0.
3. Elastic relationship between elastic strain rate and stress rate: σ̇ = De ⋅ ε̇el .
4. Flow rule: ε̇ pl = L R.
5. Hardening rule α̇ = L α.

As usual, ε̇el and ε̇ pl indicate the increments of the elastic and plastic parts of the strain tensor, the symbol L
stands for the plastic multiplier, R is the derivative of the plastic potential with respect to the current stress

PLAXIS 5 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Introduction
Basic equations of elasto-plasticity

state, De is the elastic stiffness matrix and α̇ is the rate of the hardening variable. denotes the Macaulay's
brackets.

The elastoplastic stiffness matrix Dep is obtained by enforcing the consistency conditions ḟ ( σ, α) = 0 to
calculate the plastic multiplier L and then by substituting this into the elastic relation used to define the stress
increments. This tensor, which relates the increments of stress and total strain (i.e. σ̇ = Dep ε̇), is defined as in
Eq. [8] :

Dep = De - L De R Eq. [8]

with the plastic multiplier expressed as shown in Eq. [9]

(De L)T
L = ε̇ Eq. [9]
+ (D R)L
e T
Kp

InEq. [9], L represents the gradient ∂ f / ∂ σ ij and K P the hardening modulus defined as shown in Eq. [10]:
Kp = - ( ∂∂ αf )α Eq. [10]

By substituting Eq. [9] into Eq. [8], it is possible to rewrite the expression of the elasto-plastic tensor in its final
form as shown in Eq. [11].

(De R)(De L)T

Dep = De - Eq. [11]
K + (D R) L
e T
p

PLAXIS 6 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
 Model Formulation
2
2.1 Critical State Soil Mechanics framework
PM4Silt is an elasto-plastic constitutive model formulated within the theory of Critical State Soil Mechanics
(CSSM) and can be used to simulate the cyclic response of silty soils. In this framework, the projection of the
Critical State Locus (CSL) in the p-q plane is defined by the Eq. [12]:
qcs = M ⋅ pcs Eq. [12]

M being the slope of the CSL in the p-q plane which is related to the constant volume friction angle φcv through
Eq. [13]:
M = 2 ⋅ sin (φcv ) Eq. [13]

The projection of the CSL in the e - ln( p ) is expressed as in Eq. [14]:

ecs = Γ - λ ⋅ ln 101.3 ( )
pcs
patm
Eq. [14]

where
patm = Atmospheric pressure.
λ = Slope of the CSL.
Γ = Void ratio at the reference mean stress p = 1kPa.
PM4Silt adopts the state parameter ξ (Been et al. (1985) (on page 51)) to distinguish states looser and denser
than the Critical State(CS), defined shown in Eq. [15],
ξ = e - ecs Eq. [15]
where
e = Current void ratio.
ecs = Void ratio on the CSL computed with the current mean pressure.

In PM4Silt the undrained strength at Critical State Su is given as an input and can be prescribed through two
alternative procedures 1) By assigning explicitly the value of Su and 2) By providing the undrained strength
ratio at CS as :
Su
Su,ratio = σvc
Eq. [16]

PLAXIS 7 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model Formulation
Critical State Soil Mechanics framework

σvc being the vertical effective stress at consolidation, usually assumed equal to the initial vertical effective
stress. According to Eq. [16], Su is then calculated as Su = σvc ⋅ Su,ratio . Consistently with other constitutive
models based on CSSM, Γ is not reported as a parameter but it is computed by using the undrained resistance at
CS. As a result, considering the relation between the stress deviator and the undrained strength (i.e.
qcs = 2 ⋅ Su ), the mean stress at critical stress is shown in Eq. [17]:

pcs = 2Su / M Eq. [17]

By combining Eq. [14] and Eq. [17] and by assuming a constant void ratio, the intercept Γ can be calculated as in
Eq. [18] and Eq. [19].

Γ = e0 + λ ⋅ ln 101.3 ( 2 Su
M patm ) Eq. [18]

or, equivalently, as

Γ = e0 + λ ⋅ ln 101.3 ( 2 σvc ⋅ Su,ratio

M patm ) Eq. [19]

Based on the hypothesis that ecs ≡ e 0 , this procedure enables to accommodate the CSL according to the
undrained strength Su and the initial void ratio which represents a further input of the model. Eq. [18] and Eq.
[19] are equivalent if a single element of soil is considered but they imply a different representation of the CSL in
the case of constant void ratio and variable vertical effective stresses. In this context, while prescribing a unique
value of Su,ratio involves different positions of the CSL and enables to consider the dependency of Su on the
initial vertical effective stress as shown in Figure 3 (on page 9), assigning Su as an input is equivalent to have
a unique representation of the CSL in the compressibility plane e - ln ( p ) as in Figure 3 (on page 9). As a
result, a unique value of the parameter Su implies a different initial state of the soil (normally or over-
consolidated in Figure 3 (on page 9), whereas, on the contrary, assigning Su,ratio is valid for a specific OCR
and a given Ko .

PLAXIS 8 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model Formulation
Critical State Soil Mechanics framework

Figure 3: Schematic representation of the CSL based on two different choices of the input parameters for the same
initial conditions: (a) case corresponding to a prescribed Su,ratio , (b) case corresponding to a prescribed Su . For
the sake of simplicity, in both figures, an isotropic initial stress state is considered (i.e pa and pb)

As the approach considered to accommodate the CSL in the compressibility plane as shown in Figure 3 (on page
9), is based on the constitutive framework characteristics of clay-like behavior, PM4Silt is not suited for purely
non-plastic silts. Although in the model the distinction between normally- and over-consolidated states cannot
be explicitly simulated due to the lack of a cap, in this framework the effect of the OCR on the undrained strength
as well as on the cyclic resistance by using Su,ratio should be considered in calibration. In clay-like materials, the
undrained strength ratio can be used to express Su as an analytical function of the initial consolidation pressure
and the intrinsic properties of the soil . For this purpose, we can define the following ratio, valid for normally
consolidated soils, as shown in Eq. [20].

α nc =
pcs
p0
= exp ( Γ− N
λ ) <1 Eq. [20]

and, by combiningEq. [20] and Eq. [12], it is possible to obtain the expression of the undrained strength for a
given soil and a given value of the initial stress as shown in Eq. [21].

Su =
qcs
2
=
M
2
pcs = ( M ⋅ α nc
2
)p o Eq. [21]

PLAXIS 9 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model Formulation
Critical State Soil Mechanics framework

The effect of having α nc < 1 is depicted in Figure 4 (on page 10) showing a qualitative stress path in undrained
conditions. The same logic can be applied to over-consolidated soils defining the ratio α oc = α nc OC R m, with m
expressed as a function of the compressibility properties of the soil, which gives Eq. [22],

Su =
qcs
2
=
M
2
pcs = ( M ⋅ α oc
2
)p curr Eq. [22]

As a result, by assuming an isotropic initial stress state (i.e. pcurr ≡ σ 'vc ), Critical State Soil Mechanics framework
(on page 7) (or Eq. [22]) can be rearranged to relate α and Su,ratio , that is shown in Eq. [23],
Su M
′
= 2
α = Su,ratio Eq. [23]
σvc

where
α ≡ αnc = For normally consolidated soils.
α≡α oc = For over-consolidated soils.

The relationship between OCR and the normalized undrained strength, deduced by using the analytical
framework of CSSM, is also depicted in Figure 5 (on page 11). It is important to remark that M is commonly
measured by means of triaxial compression tests and , consequently, for other types of stress paths, the Su,ratio
calculated with Eq. [23] should be decreased to account for a reduced value of M at CS.

Figure 4: Schematic representation of the effective stress paths of two samples of a clay during undrained triaxial
compression: (a) case of an initial normally consolidated state (b) case of an initial over-consolidated state

PLAXIS 10 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model Formulation
Critical State Soil Mechanics framework

1.0 0.40
(a)
0.35
0.8 TXC
0.30
DSS
0.6
0.25
TXE
0.4 0.20

0.15
0.2
(b)
0.10
1 2 4 6 8 10 0 20 40 60 80 100
PI [%]
Figure 5: Normalized undrained shear strength versus OCR in direct simple shear tests for the Boston Blue clay
(figure after Ladd et al. (1974)). (b) Differences in undrained strength for normally consolidated silts and clays with
different plasticity index (PI) (figure from Ladd et al. (2003)).

A further limitation of Eq. [23] is related to anisotropic consolidation processes which can have an influence on
the undrained strength ratio. Especially if an analysis of existing geotechnical structures is required, the
dependency of Su,ratio on K o should be accounted to obtain a realistic dependency of su on the initial vertical
effective stress. To overcome this limitation and account for a su that varies for each stress point, the method
proposed by Montgomery et al. (2014) (on page 52), extending the procedure described in Duncan et al.
(2005) (on page 51) for limit equilibrium analyses, has been proposed in the current implementation of
PM4Silt. This method consists of calculating the shear stress τ ff , K on the eventual failure plane at undrained
c
failure for general consolidation condition (with K c = σ1c / σ3c indicating the consolidation stress ratio through
a linear interpolation between the undrained strengths for consolidation at K c = 1 and K c = K f , i.e
τ ff ,( K =K -τ (
c f ) ff , K c =K f )
Su ≡ τ ff, Kc = τ ff ,( K =1) + Kf -1
(K c − 1) Eq. [24]
c

In this equation, K c is the consolidation stress ratio K c = σ1c / σ3c for generic conditions of consolidation
whereas K f is the highest possible K c ( f denoting failure). In Eq. [24], the undrained strength, although should
be Su = τ ff , K cos ( φcs ) , it has been defined as the stress on the failure plane as suggested in Boulanger
c /
(2019) (on page 51), where this method is used in nonlinear dynamic analysis of a compacted earth dam
performed with PM4Silt. The two shear stresses on the eventual failure planes (i.e. τ ff , K =1 and τ ff , K = K are
c c cf
computed respectively as in Eq. [25] and Eq. [26]:
τ ff , ( K =1) = d R + σ fc ⋅ tan (ψR ) Eq. [25]
c

τ ff , ( K
c
=K f ) = cc + σ fc ⋅ tan (φc ) Eq. [26]

where cc , ϕc and d R , ψR are parameters that can be computed from ICU Triaxial tests. To compute these
parameters no additional tests are required but only a different interpretation of the results. The two
parameters indicated with cc , ϕc are the intercepts and slopes of the effective stress failure envelope (usually
indicated as cc , ϕc ). In the context of the method, cc , ϕc are used to compute the undrained strength for the
case of consolidation at K c = K f as shown in Eq. [25]. The parameters d R , ψR can be obtained by plotting

PLAXIS 11 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model Formulation
Bounding, Dilatancy and yield surfaces

the undrained shear strength (considered as τ ff , ( K =1 ) versus the initial mean effective stress at consolidation
c
and evaluating the intercept and slope of the linear fit of the data. The computation of K c in Eq. [24] is
performed internally for every single stress point based on the initial effective stress. The four parameters
represent only an alternative way to prescribe a variable su throughout the domain. Additional mathematical
details about this approach are provided in Contracting behaviour (on page 55).

2.2 Bounding, Dilatancy and yield surfaces

PM4Silt incorporates the concepts of bounding and dilatancy stress ratio surfaces, consistently with the former
approach proposed in Dafalias et al, (2004) (on page 51). The bounding and dilatancy stress ratios indicated as
M b and M d respectively, depend on the current state of the material and converge to the stress ratio M when
the stress response approaches the CS. However, being PM4Silt defined for 2D problems, there is no
introduction of a dependency on the Lode angle in the plastic functions of the model.

2.2.1 Yield surface

The yield surface is formulated as a cone in the stress space, defined as shown in Eq. [27]:
1/2 1
f = (s - pα) : (s - pα) - pm = 0 Eq. [27]
2

The tensor α is the back-stress ratio tensor which represents the hardening variable of the model and defines
the position of the axis of the yield surface. m is the semi-amplitude of the cone and it is assumed to be constant
and equal to m=0.01 as in Figure 6 (on page 12). The hardening modulus and the elasto-plastic tensor are
calculated using the tensor n representing the deviatoric unit normal to the yield surface, i.e.
∂f r - α
n= ∂s
= 2 m
Eq. [28]

Critical state line M

m Elastic
range
m

p
Figure 6: Schematic representation of the yield surface formulated in PM4Silt

PLAXIS 12 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model Formulation
Bounding, Dilatancy and yield surfaces

2.2.2 Dilatancy Surface

The dilatancy stress ratio M d is expressed as Eq. [29]:

M d = M ⋅ exp n d ( ξ
λ
) Eq. [29]

n d being a positive parameter. According to the sign of ξ ,the DS lies below or above the CS line as in Figure 7
(on page 13), that is negative or positive values of ξ are associated with material states denser or looser than
CS, respectively. Analogously to the bounding, the dilatancy stress ratio is considered to define the dilatancy
back-stress ratio tensor in terms of
1
αd = 2
(M d - m)n Eq. [30]

Figure 7: Schematic representation of bounding and dilatancy stress ratios for a given value of the state parameter:
negative and positive ξ denote soils denser and looser than CS, respectively

2.2.3 Bounding Surface(BS)

The bounding stress ratio M b is defined differently for states looser ("wet" side) and denser ("dry" side) than
CS, i.e. negative or positive values of the state parameter, respectively. For the looser case, it is shown in Eq. [31].

(
M b = M ⋅ exp -n b,wet λ
ξ
) for ξ > 0 Eq. [31]

where n b,wet is a positive parameter of the model employed to have the bounding stress ratio smaller than the
CS (that is M b < M for ξ > 0 as seen in Figure 7 (on page 13)). On the "dry" side M b is expressed as:

( )
1 + CMb b,dry
n
Mb=M ⋅ p
for ξ < 0 Eq. [32]
+ CMb
pcs

PLAXIS 13 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model Formulation
Bounding, Dilatancy and yield surfaces

1
CMb = 1
Eq. [33]
( M
b,max
M
) n
b,dry
-1

M b,max = 2 ⋅ sin(φmax) Eq. [34]

n b,dry being a positive parameter of the model. φmax is a constant and is equal to φmax = 60 o . Eq. [31] and Eq.
[32] are used to define the image of the back-stress ratio tensor through Eq. [35].

αb =
1
2
(M b - m)n Eq. [35]

The above mentioned equations show that, for a given value of ξ, the DS and BS are represented by straight lines
in the p-q plane, having slopes corresponding to M d and M b, respectively. A representation of these surfaces
for a fixed value of the void ratio is shown in Figure 8 (on page 14). In this figure, it is possible to observe that
the DS, BS, and CS intersect each other at the specific value of the mean stress for which the state parameter
vanishes, corresponding to p ≡ p cs . The effect of the parameters n b,wet and n b,dry on the trend of the Bounding
line, respectively on the "wet" and "dry side", is also shown in Figure 8 (on page 14).

Figure 8: Schematic representation of the bounding and dilatancy stress ratios for a given void ratio

Consistently with former bounding surface models, the value of the initial back-stress ratio tensor is tracked at
every change in loading direction, identified by the condition ( α - αini ) :n < 0, and used to compute the
hardening modulus. Specifically, the values of αini are updated to the current α is assigned to αini . In this
manner, the back-stress ratio tensor is stored at the last two loading reversals of the cyclic loading. The model
app
defines also a tensor αini which is introduced to avoid unrealistic stiffness during small cycles of loading/
unloading.

PLAXIS 14 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model Formulation
Fabric-Dilatancy tensor

2.3 Fabric-Dilatancy tensor

In the model proposed in Dafalias et al.,(2014) (on page 51), the fabric-dilatancy tensor z was introduced to
take into account the effect of the previous straining on the mechanical response of sands. Correspondingly, the
incremental equation adopted in PM4Silt is defined as in Eq. [36]:

cz -ε̇ pl
v
ż = - zcum D (zmax n + z) Eq. [36]
1+ -1
2zmax

where
D = Dilatancy.
zmax = Parameter defining the range of values that z can attain.
Cz = parameter controlling the rate of evolution of z
zcum = cumulative value of the absolute changes in z which are computed
according to ż cum = | ż |

/
In Eq. [36], the Macauley's brackets in the ratio - ε̇ vpl D imply a variation of the fabric tensor only during
dilating behavior, being this the reason for this tensor to be denominated "fabric-dilatancy", although the same
ratio shows that as opposed to the model in (Dafalias et al., (2004) (on page 51)). ż depends only on the
modulus of the deviatoric plastic strain rate. By observing Eq. [36], it is also worth remarking that the rate of
evolution of z tends to vanish for large values of zcum (i.e zcum ≪ zmax ) , thus having the tensor z characterized
by constant values.

2.4 Elasticity
An isotropic hypo-elastic model is considered to simulate the elastic behavior of PM4Silt. The elastic shear
modulus depends on
1. The mean effective stress.
2. The stress ratio .
3. The fabric tensor and is defined as in Eq. [37].

1+( )zcum

( )nG zmax
p

( )
G = G0 patm patm
CSR zcum
Eq. [37]
1+ C
zmax GD

In this equation G0 and nG are two parameters related to the dependency of the small strain shear modulus on
the mean effective stress. CGD is a parameter controlling the shear stiffness degradation at large values of zcum
while CSR provides the dependency on the stress ratio as shown in Eq. [38],

PLAXIS 15 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model Formulation
Hardening rule

CSR = 1 - CSR,0 ⋅ ( )
M mSR
Mb
Eq. [38]

where mSR = 4 and CSR,0 = 0.5.

As the Poisson's ratio ν is given as an input parameter, the bulk modulus is calculated as shown in Eq. [39]:

K = (
2 1+ν
3 1 - 2ν
)G Eq. [39]

In the formulation of G, the dependency on the current void ratio, formerly proposed in Dafalias et al.,(2004) (on
page 51), has been replaced by introducing a dependency on the fabric-dilatancy tensor (i.e. by introducing the
variables zcum / zmax ) and the ratio M / M b. The advantage of using these complex dependencies in the
formulation of the shear modulus results in better modeling of the stress-strain response for soils susceptible to
flow liquefaction. In particular, the definition of the shear modulus has three main consequences:
1. The bulk modulus gradually decrease as zcum increases.
2. The decreasing trend of G after phase transformation involves a decreasing trend of the strain hardening and
3. The ability to better approximate the hysteretic stress-strain response during cyclic softening (Boulanger et
al., (2018) (on page 51)).
It is worth remarking that the stress component σzz has no influence on the plastic mechanisms of the model
therefore it is computed using a linear elastic relation and the hypothesis of plane-strain conditions, that is
σ̇ zz = νσ̇ xx + νσ̇ yy

2.5 Hardening rule

The yield surface of PM4Silt evolves according to a rotational hardening mechanism governed by an incremental
variation of the back-stress ratio tensor in line with the hardening rule defined in Dafalias et al. (2004) (on page
51), which is shown in Eq. [40].

α̇ = L h (αb - α) Eq. [40]

h being the hardening coefficient, a function of the plastic modulus Kp defined as in Eq. [41]
Kp
h = Eq. [41]
p ⋅ (α - α) : n
b

where

(αb - α) : n 0.5 CrevCkα

K p = h0 ⋅ G ⋅ ⋅
( )
Eq. [42]
(( app
) )
exp α - αini : n − 1 + Cγ1
1 + C Kp
zpeak
(αb - α) : n 1 - Czpk2
zmax

In this last expression, it appears ho which is a further parameter of the model as shown in Eq. [42]. Moreover,
Crev and Ckα contain a dependency on the back-stress ratio history and the fabric history during the loading
process, whereas z peak = max (|z| / 2, z peak ) is used to track the history of z and CΥ1 is fixed and equal to
ho/100. Further explanations and mathematical details about these variables can be found in Boulanger et al.
(2018) (on page 51) and Boulanger et al. (2017) (on page 51).

PLAXIS 16 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model Formulation
Flow rule

2.6 Flow rule

Plastic strain increments can be calculated by using the classical flow rule as shown in Eq. [43]:

ε̇pl = L R Eq. [43]

where
L = Plastic multiplier
R = Derivative of plastic potential which is expressed as shown in Eq. [44].
1
R=n+ 3
DI Eq. [44]

The direction of the plastic flow can be decomposed into two components defining the deviatoric and volumetric
plastic strain increments as shown inEq. [45] and Eq. [46].
2
ε̇pl
v = 3
L D Eq. [45]

ėpl = L n Eq. [46]

A synthetic description of the mathematical expressions implemented to define the dilatancy D for the two cases
of dilation and contraction is schematically depicted in Figure 9 (on page 17).

Figure 9: Schematic representation of the dilatancy function considered in PM4Silt (figure after Boulanger et al,
(2018))

2.6.1 Dilation: D<0

Dilation occurs every time (αdR − α):n < 0. In this case, the value of D is defined by comparing the following two
quantities as shown in Eq. [47] and Eq. [48].

PLAXIS 17 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model Formulation
Post shaking analyses

-z:n (αdR - α) : n
Drot = Ad ⋅ ⋅ CDR
Eq. [47]
2zmax

Dnon-rot = − Ad - (αd - α) : n Eq. [48]

where
αdR = Back-stress ratio tensor for the rotated dilatancy surface.
Ad = A function depending on the fabric tensor and the deviatoric unit tensor
n.

Both αdR and Ad are characterized by complex mathematical expressions which are reported in Dilatancy
expressions (on page 54)(Eq. [74] and Eq. [70]) . The definition of Ad implies the introduction of two further
parameters, Ad0 and Cε , enabling a better fit of the experimental data.

It is worth mentioning that in PM4Silt dilation occurs also for mean stresses p ≤ 2 pmin where pmin is an
internal parameter of the model and can be set in two ways as explained in Model Parameters (on page 20)(Eq.
[64] andEq. [65]). In this particular case, it is expressed as shown inEq. [49].

D = - 3.5Ado M b - M d ⋅ ( 2 pmin - p
pmin ) Eq. [49]

2.6.2 Contraction: D>0

In the case of contraction, it is expressed as shown in Eq. [50].

d
(
D = Adc α - αapp )
: n + Cin 2
⋅
(α - α) : n
C p,min Eq. [50]
in
( d
)
α - α : n + CD

The value of CD is usually assumed equal to 0.10, while the function Adc is defined as shown in Eq. [51].
Ad0 1+ z:n
Adc = hp
⋅ CdzCwet
Eq. [51]

where

h p = h p0 ⋅ exp − 0.7 + 0.2 3 − ( λ

)
ξ 2
Eq. [52]

h p0 and Ad0 being model parameters that can be evaluated during the calibration process to better fit the
experimental trend of the Cyclic Resistance Ratio (CRR). The mathematical expressions of the variables
Cin , C p,min , Cdz and Cwet are reported in Dilating behaviour (on page 54) and in Boulanger et al. (2018)
(on page 51)).

2.7 Post shaking analyses

Although PM4Silt is not suited for simulating the consolidation process due to the lack of a cap, it uses a
pragmatic approach to simulate consolidation phenomena after strong shaking in cyclically softened silts/clays

PLAXIS 18 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model Formulation
Post shaking analyses

without underestimating the volumetric strains. The approach considers a reduction of the stiffness as shown in
Eq. [53].
Gpost-shaking = F consolG Eq. [53]

where the factor F consol is defined as shown in Eq. [54],

(
F consol = 1 - 1 −
Gc,min
G ) 1−
M curr 0.25

Md
Eq. [54]

being
p 1
Gc,min = 8( λ ) ⋅
( )
zcum
Eq. [55]
1 + (C Gconsol - 1) ⋅
zcum + zmax

CGconsol being a parameter controlling the amount of shear modulus degradation when zcum tends to large
values. If zcum is small, Gc,min , corresponds to the shear modulus connected to the one-dimensional
recompression stiffness calculated through p and λ. As a result, the expression of F consol will return values close
to Gc,min if the loading is within the dilatancy surface ( M cur ≈ M d ) (Boulanger et al. (2018) (on page 51)).
Furthermore, to perform stability analyses after strong earthquakes and to account for the reduction of soil
undrained resistance (for its rate dependency or also in case of degradation or remolding phenomena), it is
possible to reduce the undrained strength by using a reduction coefficient F Su , which is a further parameter of
the model. This parameter does not act directly on su but it shifts the CSL as shown in Eq. [56].
p
ecs = Γ - λ ⋅ ln F Su ⋅ (1kPa)
Eq. [56]

Further details regarding the use of the parameters F consol and F Su for post-shaking analyses, i.e. when the flag
Post-shake is selected equal to 1, will be given in the next section.

PLAXIS 19 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
 Model parameters and state variables
3
PM4Silt is implemented for PLAXIS 2D as a User-defined Soil Model. In this section, a detailed description of the
model parameters and state variables is presented.

3.1 Model Parameters

The input parameters of PM4Silt can be grouped into three categories: primary, secondary, and optional:
• Primary input parameters: Su ( or Su, ratio ) , G0 , h p0 .
• Secondary input parameters:
n G , h 0 , e0 , λ, φcv , n b,wet , n b,dry , n d , Ado , τu,max , zmax , cz , cε , CGD , C K α f , ν.
• Optional input parameters
1. Post-shake reconsolidation: Post Shake, F su and CG consol .
2. Alternative definition of Su : ψR , d R and φc , cc .

To simplify the calibration process and give the order of magnitude of the secondary parameters, reference
values have been provided in Boulanger et al. (2018) (on page 51) which can be used as a first-tentative during
the process or retained in case the required experimental data are not available. If a zero value is assigned in the
PLAXIS 2D Input user interface, the reference values proposed in Boulanger et al.(2018) (on page 51) will be
automatically used internally.
While primary and secondary parameters are used in every kind of analysis, the two optional parameters F su
and CG consol are used only when Post shake is equal to 1. Furthermore, as explained in Post shaking analyses
(on page 18), the last four parameters can be used to consider the variability of Su in numerical analyses where
K 0 is not constant.

The entire set of parameters is depicted in Figure 10 (on page 21). All the parameters and state variables are
initialized based on the effective stress state at the beginning of the first calculation phase in which PM4Silt is
employed. If the model is used in different phases of a given analysis, by default the initialization is performed
only at the first phase in which PM4Silt is used therefore, if a reinitialization of the state variables (and the
computation of Su ) is required, the user can select the option Reset state variables in the Phases window of the
PLAXIS Input program.

PLAXIS 20 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model parameters and state variables
Model Parameters

Figure 10: Parameters of the User-defined Soil Model in the Material data set window

PLAXIS 21 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model parameters and state variables
Model Parameters

3.1.1 Primary parameters

Undrained strength at critical state (or undrained strength ratio): Su or Su,ratio

The parameter Su is used to determine the position of the CSL in the compressibility plane (as shown in Figure 3
(on page 9)). Su can be entered by the user or can be initialized based on the value of the undrained strength
ratio Su,ratio . Only one parameter, that is either Su or Su,ratio , should be specified by the user and in case both
values are different than zero, only Su will be used.

When Su,ratio is assigned as an input, Su is then computed from the vertical effective stress at the beginning of
the first phase of a given analysis in which PM4Silt is used. If the model is used in different phases, the vertical
effective stress considered in the computation of Su can be reset, as already mentioned, by selecting Reset state
variables in the Phases window of the PLAXIS Input program. To select the appropriate value for Su (or the
equivalent Su,ratio ), the user must consider, as clarified in Boulanger et al. (2018) (on page 51), that

• Although silts and clays can exhibit a peak undrained shear strength, the value of Su at critical state should
be estimated to assign the input parameter.
• The undrained strength is intrinsically characterized by a rate dependency (e.g. Sheahan et al., (1996) (on
page 52)). Due to the high rate of loading happening during seismic events, an increase of Su should be
considered for dynamic analyses. For this reason, as laboratory experiments are commonly performed by
applying the loading through slower processes, the calibrated value of Su needs to be increased about
20-40% to make it coherent with earthquake loadings (Boulanger et al., (2007) (on page 51)).
The undrained strength Su can be determined through different methods:

• Laboratory tests, e.g. consolidated undrained triaxial test or DSS tests.

• In situ tests, typically cone penetration or vane shear tests .
• Empirical correlations between the undrained strength ratio and OCR.
In boundary values problems where the effective stress is not homogeneous, the choice of initializing Su by
using Su,ratio allows one to consider a dependency of Su on the confinement, which is valid for homogeneous
values of OCR and K 0 ( and homogeneous effective stress path). Also, the variability of Su in the domain can be
considered by defining multiple layers. In some cases, as for the analysis of embankments, more complex
distributions of the undrained strength resulting from different consolidation stress states can be considered
using the method described in Critical State Soil Mechanics framework (on page 7).

Initial shear modulus coefficient: G0

G0 controls the small strain shear modulus in Eq. [37] which, at the state initialization, is shown in Eq. [57]:

ρ nG
G = G0 ⋅ ptm ⋅ ρatm
CSR Eq. [57]

PLAXIS 22 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model parameters and state variables
Model Parameters

The shear modulus coefficient G0 is a constant dimensionless parameter that can be assessed by knowing the
small strain shear modulus at a reference mean effective stress. The elastic shear modulus should be calibrated
to fit estimated or measured shear wave velocities V s according to Eq. [58]:

G = ρ (V S )2 Eq. [58]
where
ρ = Saturated density.
Two approaches to determine G0 are reported in Boulanger et al.(2019) (on page 51).

Contraction rate parameter: h po

This parameter has an influence on the dilatancy D in the case of contracting behavior (see Eq. [51]). The higher
the value assigned to h po , the lower the corresponding value of D. For this reason and the effect on the rate of
reduction of p in undrained conditions. h po has a strong influence on the number of cycles required to trigger
the cyclic resistance. This parameter has no direct physical meaning and therefore its evaluation requires an
iterative procedure. The user should tweak the value of h po until it is possible to match the cyclic strength
curves representing the relationship between the Cyclic Resistance Ratio (CRR) and the number of uniform
loading cycles required to cause a 3% shear strain (or any other preferred criterion) under undrained cyclic
loading conditions. Examples of h po calibration are reported in Boulanger et al, (2019) (on page 51).

The link existing among the cyclic resistance, the undrained strength, and the OCR, valid in the silts and clays
exhibiting a stress-history normalized behavior, allows estimating the CRR of these soil deposits through three
different approaches Boulanger et al, (2007) (on page 51).
1. CRR evaluation from cyclic laboratory testing.
2. CRR evaluation from a measured Su profile: When a direct evaluation through experimental laboratory tests
is not available, empirical relationships have been proposed to provide a reference value of CRR for
earthquakes of magnitude equal to 7.5, CRR M =7.5 , for instance, the empirical relationship is shown in Eq.
[59].

CRR M =7.5 = C2D ⋅ ( )

τcyc
Su N =30
⋅
Su
′
σvc
⋅ Kα Eq. [59]

C2D is a correction factor for two-dimensional versus one-dimensional cyclic loading, usually assumed as
0.96. K α is the static shear stress ratio correction factor to approximate the effect of initial static shear

stresses, ( )
τcyc
Su N =30
is the ratio of cyclic stress τcyc to monotonic undrained resistance Su required to trigger
the peak shear strain of 3% with 30 uniform cycles of loading. Despite many other factors (e.g. aging or OCR)

influencing its value, it is suggested that ( )

τcyc
Su N =30
is assumed equal to 0.83 with a variation of ±15% for
natural clay-like soils subjected to DSS loading conditions (Boulanger et al. (2004) (on page 51) and

Boulanger et al. (2007) (on page 51)). After replacing the values for the coefficients C2D and ( )
τcyc
Su N =30
,
the CRR M =7.5 for fine-grained soils can be expressed as shown in Eq. [60] :

CRR M =7.5 = 0.8 ⋅ ( )

Su
'
σvc
⋅ Kα Eq. [60]

PLAXIS 23 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model parameters and state variables
Model Parameters

The scaling to earthquakes of magnitude different than 7.5 can be finally done through
CRR ( M ) = MSF ( M ) ⋅ CRR M =7.5 , where MSF is a magnitude scaling factor that can be determined from
Figure 11 (on page 24).
3. CRR evaluation from a consolidation stress history profile: By combining the previous equation with the
classic relationship that links the normalized monotonic undrained resistance with the OCR, the CRR M =7.5
can also be estimated through Eq. [61].

CRR M =7.5 = 0.8 ⋅ S ⋅ OCR m ⋅ K α Eq. [61]

Figure 11: Magnitude Scaling Factor (MSF) as a function of the earthquake magnitude M

S and m being coefficients to be estimated empirically. As also recalled in Boulanger et al. (2004) (on page
51) for homogenous, low-plasticity, high-plasticity, and sedimentary clays, Ladd (1991) (on page 51)
proposed 0.22 and 0.8 for S and m, respectively. With these assumptions, Eq. [61] can be rewritten as in Eq.
[62].

CRR M =7.5 = 0.18 ⋅ OCR 0.8 ⋅ K α Eq. [62]

3.1.2 Secondary parameters

Atmospheric pressure (default value 101.3 kPa): p atm

It represents the atmospheric pressure in the unit set being used for the analysis. When set to zero the value of
101.3 is selected by default. If a unit system different from kPa is used, the correct value must be explicitly
provided in input.

Shear modulus exponent (default value 0.75): η G

η G controls how the shear modulus varies with the confining stress. This parameter can be determined together
with G0 to fit the values of V S or G

PLAXIS 24 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model parameters and state variables
Model Parameters

Plastic modulus ratio (default value 0.5):h 0

This parameter is used to increase or reduce the hardening modulus K p (Refer Eq. [42]). The effect of h 0 on the
predicted mechanical behavior can be easily understood by examining the equations of the stiffness matrix as
shown in Eq. [11], according to which increasing values of K P bring the elastic-plastic stiffness closer to the
elastic one. Therefore, higher values of K p reduce the degradation of the elastic stiffness with the development
of plastic strains. For this reason, h 0 influences

1. The stiffness on the shear level which modifies the trend of monotonic stress-strain behavior.
2. The shape of the shear modulus degradation curve.
3. The damping ratio curve.
Furthermore, increasing values of h 0 tend to move the shear modulus reduction curve towards the right part of
the plot, consistently with what is observed for materials with increasing PI. The parameter h 0 can be used to
adjust the dependency of the secant shear modulus and damping ratio on the shear strain level (Boulanger et al.
(2019a) (on page 51)). In Boulanger et al. (2019b) (on page 51) it is suggested to calibrate this parameter
against either monotonic or cyclic tests, in accord with the specific case at hand.

Initial void ratio (default value 0.9): e0

Initial void ratio of a soil layer.

Slope of the CSL (default value 0.6): λ

The slope of the critical state line in ln(p)-e space.

Friction angle at CS (default value 32o): φcv

Bounding surface parameter (default value 0.8): n b,wet

This parameter influences the peak value of the undrained resistance Su and it should be chosen to match the
trend of behavior observed in monotonic undrained shear tests. The upper and lower limits are 1.0 and 0.1,
respectively.

Bounding surface parameter (default value 0.5): n b,dry

This parameter influences peak friction angles for states denser than CS and modifies the shape of the BS on the
dry side as seen inFigure 8 (on page 14). Due to the constraint of no-intersection between the current state and
the BS, the shape of the bounding on the dry side influences the trend of the effective stress path for states
denser than CS.

Dilatancy surface parameter (default value 0.3): n d

This parameter influences the transition between contractive and dilative mechanical response.

Dilatancy parameter (default value 0.8): Ado

This parameter affects the dilatancy D.

PLAXIS 25 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model parameters and state variables
Model Parameters

Maximum excess pore pressure ratio (default value 0): ru,max

Δp
ru,max = po
Eq. [63]

where
Δ p = The maximum possible reduction effective stresses
po = The initial mean effective stresses

ru,max affects the value of pmin which is an important reference value of the mean effective stress appearing in
several equations of the model (see for instance Eq. [49]). In PM4Silt, the maximum reduction of p during
undrained cyclic loadings is predefined and, if desired, it can be regulated through the parameter ru,max . pmin
is computed internally and it is based on the initial effective stress and Su,ratio when the 0 value is assigned to
ru,max , that is

pmin =
pcs
8
=
1 Su
4 M
( ) Eq. [64]

or it is computed explicitly according to the value of ru,max when explicitly assigned as an input , i.e.
po
pmin = (1 - ru,max ) 2
Eq. [65]

The value 2 pmin represents the minimum mean effective stress reachable during undrained cyclic loadings. To
estimate ru,max it is important to account for the different definition of ru with respect to the classical one used
σvo − σv
Δ u
to interpret laboratory test results. For instance, for direct simple shear tests ru = σvo
= σvo
. It is worth
σv - σv Δu
noting that in the numerical simulation of simple shear tests the ru,max coincides with po
= po
.

Maximum value of fabric tensor (default value 0.0): zmax

This parameter determines the maximum value of the fabric tensor zij . If the input value is equal to 0, this
parameter is automatically computed based on the value of the Su,ratio , i.e.

zmax = 10 for Su,ratio ≤ 0.25

zmax = 40 Su,ratio for 0.25 < Su,ratio ≤ 0.50 Eq. [66]
zmax = 20 for Su,ratio > 0.50
zmax starts influencing the evolution of the dilatancy fabric tensor only after that zcum reaches the value 2zmax .
Higher values of zmax tend to increase the maximum value of the excess pore pressure while reducing the width
of the hysteresis loop and the cyclic strength, steepen the CRR-cycles curve, and increase the rate of strain
accumulation.

Fabric growth parameter(default value 100): Cz

This parameter controls the evolution of the fabric dilatancy tensor zij and has a strong influence on the CRR
curves. As shown in Eq. [36], a reduction of C z results into a decrease in the growth rate of the fabric-dilatancy
tensor and the corresponding zcum . The fabric tensor remains equal to zero until the stress state reaches the DS.

PLAXIS 26 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model parameters and state variables
Model Parameters

As a consequence, C z will not influence the response before the stress state reaches the DS. In the case of
dilatant behavior, when it is necessary to reduce the level of cumulated strains, it is preferable to modify C z
rather than varying h p0 . This parameter can also be used to adjust some portions of the effective stress path and
the stress-strain relation.

Strain accumulation parameter(default value 0): Cε

This plays a role in D for the case of dilating behavior, appearing at the denominator of Ad (see Eq. [70] in
Dilating behaviour (on page 54)). When this parameter is prescribed equal to 0, the default value is computed
internally as
Cε = 0.5 + 1.2 Su, ratio - 0.25 ≤ 1.3 Eq. [67]

Analogous to the role of C z , the parameter Cε has an influence on the mechanical behavior predicted by the
model only after that the stress state reaches the DS. Specifically, it reduces the tendency to dilate for stress
paths with a negative increment of mean stress and decreases the stiffness of the material, hence enabling faster
development of strains.

Degradation factor (default value 3): CGD

CGD affects the degradation of the shear modulus G when zcum becomes larger than zmax . Consequently, this
parameter influences the response when the amount of shear strain is very large and allows to account for the
stress-strain behavior of liquefied soils as the maximum degradation reaches a factor of 1 / C GD (see Boulanger
et al. (2018) (on page 51)).

Plastic modulus factor (default value 4): C K α f

As remarked by Boulanger et al. (2018) (on page 51), C K α f can adjust the effect that the initial static shear
stress has on plastic modulus and hence on cyclic strength. The effect of this parameter is small for states looser
than CS and becomes more important for states denser than CS.

Poisson's ratio (default value 0.3): ν

In the case of one-dimensional consolidation, as the model is not able to predict plastic behavior, K 0 can be
computed using the elastic relation K 0 = ν / ( 1 - ν ) .

3.1.3 Post-shaking analysis parameters

The three post-shaking analysis parameters are as follows:

1. Post shake: Post-shake reconsolidation flag (default value 0).
2. FSu- Post-shake strength reduction coefficient (default value 1.0).
3. GCconsol - post-shake shear modulus degradation parameter (default value 2.0).

The user can use the Post shake flag to perform post-shake reconsolidation analyses by assigning to it a value
equal to 1 (this option is inactive if Post-shake=0). In the case of Post-shake=1 a reduction of stiffness and
resistance is performed according to the equations presented in Post shaking analyses (on page 18). In this case,

PLAXIS 27 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model parameters and state variables
State variables

two additional parameters must be specified in the input, FSu and GCconsol . These parameters allow the user to
model a reduction of the undrained resistance and/or the stiffness which is sometimes found in practice. The
user should create a copy of the material and assign the value 1 to Post shake and the estimated values for FSu
and GCconsol .

3.1.4 Alternative definition of a variable Undrained Shear Strength

Parameters to estimate Su in presence of different consolidation conditions from results of

isotropically consolidated undrained triaxial tests :ϕc , cc , ψR , d R
These parameters are aimed to prescribe the undrained strength variable with the consolidated stress state.
This method is an alternative to prescribing either Su or Su,ratio and is automatically activated when both ϕc
and ψR are different than zero, regardless of the value of Su or Su,ratio . In this manner, the method proposed by
Montgomery et. al (2014) (on page 52), applied also to PM4Silt by Boulanger (2019) (on page 51), described
in Critical State Soil Mechanics framework (on page 7) is considered to account for a variable Su in presence of
different conditions of consolidation in the domain.

3.2 State variables

In PLAXIS Output the user can plot a list of variables related to the model Figure 12 (on page 31). The list
includes the state variables and additional variables are included to facilitate the interpretation of results. In
addition, also parameters depending on the initial effective stress, i.e. the small strain bulk and shear moduli, are
provided in the list of state variables. Here follows a description of these variables.

State Variables Description

Su,ratio Undrained strength ratio.

Su Undrained strength.

ru,max Maximum excess pore pressure ratio, as provided in input

2p min represents the minimum value of the current mean effective

stress, i.e. the following restriction must be satisfied pcur ≥ 2 ⋅ pmin . It
pmin
can be estimated based on ru,max or pcs . For additional details see
Flow rule (on page 17).

zmax Parameter influencing the maximum value of the fabric-dilatancy

tensor.

Cε Strain accumulation parameter.

PLAXIS 28 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model parameters and state variables
State variables

State Variables Description

M Critical state stress ratio.

Γ Void ratio corresponding to 1 kPa on the CSL.

pcs Mean effective stress at the critical state.

ξ State parameter.

e Current void ratio.

M cur Current value of the stress ratio.

K Current bulk modulus.

G Current shear modulus.

The static shear stress ratio is the ratio between the initial shear stress
and the initial vertical effective stress. αstatic is shown in output as a
value different than zero in case of a customized distribution of the
τxy,o undrained resistance which is provided by using the parameters
αstatic = ϕc , cc , ψR and d R . These parameters have an influence on the
σv,o
undrained resistance. As recalled by Eq. [59], the undrained cyclic
resistance depends on αstatic , therefore plotting this variable before the
dynamic calculation can be useful to better calibrate the cyclic
resistance.

Ratios between the initial principal effective stresses and the initial
effective stress, i.e. K c = σ1 / σ2 and K 0 = σh / σv , respectively.
Analogously to αstatic , K c and K 0 are characterized by non-zero values
K c and K 0
when the undrained resistance is distributed according to the trend
prescribed through the parameters ϕc , cc , ψR and d R . They influence
the computation of Su according to Eq. [24].

σv,o Initial vertical effective stress.

p - p
o
Current value of the ratio po
. The sign of ru indicates a dilative or
ru
contractive response. In undrained DSS tests, the numerator coincides
with Δu when the condition Δσh = Δσv = Δu is reached.

Maximum value attained by ru during a prescribed loading history. The

ru,Extreme
maximum value this variable can reach is ru, Limit .

γ
2 MaxExtreme
= ( 2
)
εxx + ε yy 2
2
+ εxy Maximum value of the Mohr's circle radius in the strain space.

PLAXIS 29 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model parameters and state variables
State variables

State Variables Description

Integer which can be either 0 or 1. In the case of a unit value, BCI

indicates a violation of the BS according to the prescribed value of K 0 .
BCI (Bounding Correction Indicator)
In this case, the initial stress is internally reinitialized to satisfy a
consistent prescription of the bounding.

γ / 2 Max = ( 2
)
εxx + ε yy 2
2
+ εxy
Current value of the Mohr's circle radius in the strain space.

Md Dilatancy surface (DS) stress ratio.

Mb Bounding surface (BS) stress ratio.

D Dilatancy.

α xx Component xx of the back-stress ratio tensor.

α yy Component yy of the back-stress ratio tensor.

α xy Component xy of the back-stress ratio tensor.

Value of ru calculated with the minimum value of the mean effective

stress. ru, Limit is defined equal to ru,max when ru,max is given as an
ru, Limit = ( p o - 2p min ) / po input to control the value of pmin otherwise, it is the value computed
according to Eq. [64]. While in sands this limit value is usually equal to
1, in silts generally can range between 0.8 and 1.

Indicator for the reduction of the mean effective stress. For contracting
behavior, the range of values for Ipr is limited between 0 and 1. In the
Ipr = ru,extreme / ru, Limit
case of dilating behavior, Ipr can be negative and indicates the distance
with respect to ru, Limit .

τxy,ratio = τxy / σv0 Normalized value of the shear stress with respect to the initial vertical
effective stress.

τxy,ratio, Extreme Maximum value of τxy,ratio during the loading history.

It is worth noting that many other variables are exposed in the user interface, although these will not be
described in this section as they are used for the numerical integration of the model.
All the parameters of PM4Silt included between the output state variables and σvo , αstatic . K o and K c and BCI all
remain constant after the initialization . The distributions of αstatic , K o and K c are useful for checking the values
of Su automatically computed based on the initial vertical effective stress and initial K o through the four
optional parameters. For this reason, these are shown only in this particular case. All the other quantities are
updated at each step of the calculation.

PLAXIS 30 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model parameters and state variables
State variables

Figure 12: State variables of PM4Silt

PLAXIS 31 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
 Model response in undrained monotonic and cyclic
4
loading conditions
To demonstrate the performance of the model, three monotonic undrained DSS tests have been solved through
the PLAXIS SoilTest facility by using the calibrations proposed in Boulanger et al. (2018) (on page 51) and
reported in Table 1 (on page 32). For monotonic tests, the main difference in three calibrations consists of the
different values of Su,ratio .

Table 1: Summary of the calibrations. Secondary parameters are set to zero, therefore at their default values.

Calibration
Primary parameters Units
A B C

Su,ratio 0.25 0.5 0.75 -

G0 588 776 913 -

h po 20 50 60 -

Secondary
Calibration (Default values) Units
parameters

patm 101.3 [kPa]

nG 0.75 -

h0 0.5 -

e0 0.9 -

λ 0.06 -

φcv 32 [o]

n b,wet 0.8 -

n b,dry 0.5 -

PLAXIS 32 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model response in undrained monotonic and cyclic loading conditions

Secondary
Calibration (Default values) Units
parameters

nd 0.3 -

Ado 0.8 -

ru,max 0(pmin = pcs / 8) -

zmax ( / σvc') ≤ 20
10 ≤ 40 Su -

Cz 100 -

Cε (
0.5 ≤ 1.5Su / σvc' + 0.2 ) ≤ 1.3 -

CGD 3.0 -

Ckaf 4.0 -

ν 0.3 -

Parameters for Post-

shaking and
alternative definition Calibration (Default values) Units
Su

Post shake 0 -

CG consol Not used when Post Shake = 0 -

F Su 1.0 -

ΨR Not used when 0 (Su or Su,ratio is specified) [o]

dR Not used when 0 (Su or Su,ratio is specified) [kPa]

ϕc Not used when 0 (Su or Su,ratio is specified) [o]

dc Not used when 0 (Su or Su,ratio is specified) [kPa]

Figure 13 (on page 34) shows results obtained in three simulations of strain-controlled undrained DSS tests
'
performed with the same initial vertical effective stress σ y0 and K 0 , while assuming values in correspondence
to the three calibrations in Table 1 (on page 32). Specifically, it is shown that increasing values of Su,ratio
determine an increase of the undrained strength (i.e. higher values of Su ) therefore the position of the BS tends
to move in the upper part of the stress space (Figure 13 (on page 34)). As shown in (Figure 13 (on page 34)),
different Su result in three different values of pcs , corresponding to the intersection between BS and CSL,

PLAXIS 33 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model response in undrained monotonic and cyclic loading conditions
Simulations of stress-controlled CDSS tests and CRR vs N curves

consistently with the stress path direction. In the analytical framework of the model Γ is defined as a function of
pcs and , due to this reason, differences in Γ cause three initial state parameters for the same initial effective
stress state. The different positions of the CSL corresponding to the calibrations A, B, and C can be observed in
Figure 13 (on page 34) showing the undrained stress path and the resulting mean effective stress at the critical
state. The contracting and dilating characteristics of the undrained DSS stress path can be observed in Figure 13
(on page 34) and Figure 13 (on page 34).

Figure 13: Monotonic undrained DSS performed with the parameters A,B and C. The same initial effective stress
'
σvc = 101.3kPa and K O = 0.5 apply. The red, green, and blue lines correspond to the results obtained with the sets
of parameters C,B and A respectively.

PLAXIS 34 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model response in undrained monotonic and cyclic loading conditions
Simulations of stress-controlled CDSS tests and CRR vs N curves

4.1 Simulations of stress-controlled CDSS tests and CRR vs N curves

A series of undrained Cyclic Direct Simple Shear (CDSS) tests have been performed by using the parameters
reported in Table 1 (on page 32) and by loading the sample with a given increment of shear stress (i.e. a given
Cyclic Stress Ratio (CSR) defined as Δτ xy / σ' v0 ). In Figure 14 (on page 35) and Figure 15 (on page 36), the
stress-strain behavior resulting from cyclic undrained DSS is shown to emphasize the evolution of different
cyclic responses obtained by selecting the calibrations A and C of Table 1 (on page 32). Although the two tests
'
are performed with the same σ y0 and K 0 , they are loaded with different shear stress increments, therefore
having different CSR corresponding to CSR=0.175 (Δτ xy = 17.73) and CSR=0.70 (Δτ xy = 70.91) for the
calibration A and C, respectively. From Figure 14 (on page 35), it is possible to observe as the soil response is
characterized by an initial contracting behavior and it needs many cycles to reach the cyclic mobility. To better
visualize the cycles which cumulate a shear strain larger than 3%, the stress-strain behavior has been plotted
with two different colors, green and orange, denoting cycles with shear strain smaller and larger than 3%,
respectively. A different scenario is depicted in Figure 15 (on page 36) where the stress path starts dilating
from the beginning of the loading although showing an overall reduction of the mean effective stress. Although
the two stress paths are characterized by a different loading (i.e. different Δτ xy / Su,ratio ), they show the ability
of the model to simulate two opposite trends of constitutive behavior and their different effect on triggering
mechanisms of cyclic mobility. It is worth noting that the two stress paths are also characterized by two different
values of the minimum mean effective stress which is explained by the dependency of Su on pmin as reported in
Eq. [64].

Figure 14: Undrained cyclic DSS test performed for normally consolidated silt, calibration A. In this test, the
increment of shear stress is equal to Δτ xy = 17.73 kPa

PLAXIS 35 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model response in undrained monotonic and cyclic loading conditions
Simulations of stress-controlled CDSS tests and CRR vs N curves

Calibration C
Test #5

Initial state

0.03

Figure 15: Undrained cyclic DSS test performed for normally consolidated silt, calibration C. In this test, the
increment of shear stress is equal to Δτ xy = 70.91 kPa.

In soil dynamics, it is common practice to characterize the cyclic resistance of soils by relating the CSR with the
number of cycles required to reach a given value of total shear strain. These curves, generally referred to as CSR-
N curves, are plotted in Figure 16 (on page 37) for the three calibrations reported in Table 1 (on page 32),
considering a single amplitude maximum shear strain of 3%. To verify the proposed implementation of PM4Silt,
the results for the three calibrated parameters are also compared with the plots reported in Boulanger et al.
(2018) (on page 51), showing a good agreement between the two implementations of the model. Further
details about this plot are given in Table 2 (on page 37) in which the entire set of initial conditions are
presented along with the corresponding values of the number of cycles employed for each undrained CDSS to
reach γxy = 3% . From Figure 16 (on page 37) it is possible to notice that for the case N=1 the CSR tends to be
equal to the prescribed value of Su,ratio . However, for the calibration A and N=1, the CSR tends to be slightly
higher than the monotonic undrained resistance at CS which is explained by the fact that the initial assumption
of nb,wet = 0.8 allows reaching a peak resistance slightly higher than 0.25 (see Figure 13 (on page 34)). The
trend of the CSR curves is mainly influenced by the parameters h po , Cε , zmax , and Cz .

As remarked in Model Parameters (on page 20), h po and Cε modify the rate of the volumetric plastic strain for
contracting and dilating behavior, respectively. An increase of h po attenuates the reduction of p in undrained
conditions (because D decreases) and therefore it increases the number of cycles required to reach 3% of shear
strain. As a result, for increasing values of h po the cyclic resistance curve tends to be stretched towards the right
part of the plot. On the contrary, when Cε is increased, the dilatant behavior of the soil is attenuating (i.e. D
decreases), thus requiring less cycles to reach 3% of shear strain.

PLAXIS 36 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model response in undrained monotonic and cyclic loading conditions
Simulations of stress-controlled CDSS tests and CRR vs N curves

Figure 16: CSR-N curves for calibrations A,B,C (data after Boulanger et al. (2018))

Table 2: List of initial conditions to simulate cyclic undrained DSS tests to obtain the CSR plotted in Figure 16. In
green are marked the tests shown in Fig.15 and 16, respectively

Number of
' Δτxy CSR / Su,ratio Number of
Test σ y0 CSR K0 steps per
cycles
quarter

[KPa] [KPa] - - -

1 101.3 13.93 0.1375 0.55 0.5 93.75 1000

2 101.3 15.95 0.1575 0.63 0.5 48.25 1000

A 3 101.3 17.73 0.175 0.7 0.5 28.75 1000

4 101.3 20.26 0.2 0.8 0.5 14.75 1000

5 101.3 22.29 0.22 0.88 0.5 8.75 1000

PLAXIS 37 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model response in undrained monotonic and cyclic loading conditions
Normalized shear modulus reduction and damping ratio curves from PLAXIS SoilTest simulations

Number of
' Δτxy CSR / Su,ratio Number of
Test σ y0 CSR K0 steps per
cycles
quarter

1 101.3 27.86 0.275 0.55 0.5 53.75 1000

2 101.3 31.91 0.315 0.63 0.5 19.25 1000

B 3 101.3 35.455 0.35 0.7 0.5 9.25 1000

4 101.3 40.52 0.4 0.8 0.5 3.75 1000

5 101.3 48.624 0.48 0.96 0.5 1.25 1000

1 101.3 41.79 0.4125 0.55 0.5 69.75 1000

2 101.3 47.86 0.4725 0.63 0.5 11.75 1000

C 3 101.3 53.1825 0.525 0.7 0.5 5.75 1000

4 101.3 60.78 0.6 0.8 0.5 2.75 1000

5 101.3 70.91 0.7 0.93 0.5 1.25 1000

4.2 Normalized shear modulus reduction and damping ratio curves from
PLAXIS SoilTest simulations
Strain-controlled undrained CDSS tests are often performed to characterize the trend of the normalized secant
shear modulus and equivalent damping ratio with respect to the maximum cyclic shear strain amplitude. To
obtain these curves, several strain-controlled CDSS tests are performed by enforcing at each cycle of loading a
given increment of shear strain Δγ xy which will be increased at the next cycle. Following this procedure, it is
possible to calculate the Equivalent Secant Shear Modulus (ESSM) and Damping Ratio (DR) at each cycle, thus
having the entire evolution of these variables in correspondence to the applied shear strain. An example of the
shear modulus decay and equivalent damping ratio is shown in Figure 17 (on page 39) by using the calibration
C reported in Table 1 (on page 32), and the following initial conditions: | σ yy | = 101.3 kPa, K o = 0.5.

PLAXIS 38 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model response in undrained monotonic and cyclic loading conditions
Normalized shear modulus reduction and damping ratio curves from PLAXIS SoilTest simulations

Figure 17: Undrained strain-controlled CDSS tests for calibration C: (a) Shear modulus reduction curves, (b) stress-
strain curves, (c) equivalent damping ratio curves, (d) effective stress paths.

The results required to plot Figure 17 (on page 39) can be obtained by running several simulations in the tab
"CDSS" of the PLAXIS SoilTest facility or preferably using the tab General through which a general stress path
can be solved. An example of how solving an undrained cyclic DSS test within the tab General is shown in Figure
18 (on page 40) where the horizontal stresses σxx and σzz have been assigned as
| σxx | = | σzz | = 50.15 kPa to prescribe K o = 0.5. As the test is controlled by enforcing the shear strain, a
loop of loading is completed after the following three phases:
• Phase 1: Δγxy represents the first increment of shear strain applied to the sample;
• Phase 2: It is applied - 2Δγxy to enforce the opposite amount of cumulated shear strain;
• Phase 3:Δγxy is finally applied to reach the stage of zero total shear strain and complete a loop.

After these three phases, a further loop composed of three new phases can be defined to simulate multiple cycles
and describe the full trend of shear modulus decay and damping ratio.

PLAXIS 39 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Model response in undrained monotonic and cyclic loading conditions
Normalized shear modulus reduction and damping ratio curves from PLAXIS SoilTest simulations

Figure 18: PLAXIS SoilTest General tab to perform undrained strain-controlled CDSS tests: shear modulus reduction
curve and equivalent stress-strain response for default calibration C

PLAXIS 40 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
 1D site response analysis
5
This section discusses the results of the seismic analysis of a soil deposit using PM4Silt to illustrate some
features of the model and how to use it in PLAXIS 2D. The use of some output state variables to interpret the
results and some practical indications for setting up the analysis are also provided. In this analysis, the set of
parameters A presented in Table 1 (on page 32) is considered to model the mechanical behavior of a normally
consolidated silty clay stratum (yellow layer of Figure 19 (on page 42)) which extends to a depth of 29m and is
delimited at the bottom by an elastic bedrock (green layer of Figure 19 (on page 42)). The groundwater table is
located two meters below the ground surface. The input motion chosen for the analysis is the time history of
horizontal acceleration shown in Figure 20 (on page 42). At the base of the model, a compliant base boundary
condition has been assigned to simulate the presence of an indefinite extension of the bedrock while along the
vertical sides, tied degree boundary conditions are applied.

Note:
1. In Figure 19 (on page 42) the mesh discretization was obtained by setting 1 to the coarseness factor on the
sides of the column and choosing a medium mesh in PLAXIS Input.
2. in Figure 20 (on page 42) the filtered input record is separately provided through the file
Acceleration_input_record.txt, showing a peak acceleration of -3.416 m/s2.

PLAXIS 41 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
1D site response analysis

Figure 19: Soil column model for the site response analysis of a silty clay deposit. Initial and boundary conditions
and the resulting mesh discretization

Figure 20: Filtered and unfiltered time history of the input horizontal acceleration

PLAXIS 42 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
1D site response analysis

The accelerogram in Figure 20 (on page 42) has been obtained by scaling with a coefficient 2 a real recorded
time history for which the maximum acceleration was 0.169g. This modification allows the model to achieve the
mobilization of high shear strain levels in the soil column and to illustrate some specific features like the use of
the state variables provided in output.

Figure 21: . Horizontal time history acceleration: filtered and unfiltered (a) Fourier amplitude spectrum, (b) Pseudo
Spectral Accelerations (PSA).

To avoid an alteration of the numerical results due to high frequencies, a Butterworth-type low-pass filter of
order 8 with a cutoff frequency of 14Hz has been applied to remove frequencies higher than 18Hz. The final
displacement drift at 60s is around 0.03m and, despite this being relatively small, the drift correction option has
been selected in the analyses. The corrected and uncorrected signals are shown in both Figure 20 (on page 42)
and Figure 21 (on page 43) and the unaltered characteristics of the two signals are further shown in Figure 21
(on page 43) by means of the PSA of the two accelerograms.
The measurement unit chosen for the analysis are [m], [kN], and [day] and the gravity acceleration is set to 9.81
m/s2. While the parameters used for PM4Silt are shown in Table 1 (on page 32) (i.e calibration A), all the other
properties of the soil stratum and the elastic bedrock are reported in Table 3 (on page 43). It is worth noting
that, due to the use of kPa as the default measurement unit, the atmospheric pressure can be set to 0.

Table 3: Additional input parameters for the two layers

CALIBRATION
INPUT PARAMETERS: Silty clay
UNIT
stratum
A

Drainage type Undrained A -

γunsat - unsaturated unit weight 13.786 kN/m3

γsat - saturated unit weight 18.432 kN/m3

α - Rayleigh damping coefficient 0.01256 -

β - Rayleigh damping coefficient 0.3185x10-3 -

e0 - initial void ratio 0.9 -

PLAXIS 43 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
1D site response analysis

CALIBRATION
INPUT PARAMETERS: Elastic
UNIT
bedrock
Elastic bedrock

Drainage type Undrained A

γunsat - unsaturated unit weight 20 kN/m3

γsat - saturated unit weight 20 kN/m3

α - Rayleigh damping coefficient 0.01256 -

β - Rayleigh damping coefficient 0.3185x10-3 -

e0 - initial void ratio 0.5 -

E- Young's modulus 4x106 kPa

ν- Poisson's ratio 0.2 -

The saturated and dry density assumed for the soil are computed considering a void ratio and a specific gravity
equal to e0 = 0.9 and Gs = 2.67, respectively. A small amount of Rayleigh damping is considered to prevent
numerical noises in the results and is recommended especially when a high level of strain is expected.
Specifically, the Rayleigh damping coefficients α and β are calibrated to have a small variability of the critical
damping ratio (DR) for the frequencies characteristic of the earthquake as well as the natural frequency of the
deposit. In this manner, the value of the DR can be approximated as a constant during the solution of the
problem. In this application, the Rayleigh damping coefficients α and β are given in Table 3 (on page 43) and
they correspond to a minimum DR equal to ξ0 = 0.2% for a frequency f 0 = 1Hz. As shown in Figure 22 (on page
44), this frequency corresponds to a minimum of the DR function and it is close to the natural frequency of the
deposit estimated with the initial elastic shear stiffness, i.e. approximately equal to f = 1.24Hz.

Figure 22: Plot of the DR as a function of the frequency

PLAXIS 44 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
1D site response analysis

As shown in Figure 23 (on page 45), the analysis is performed in two phases: (i) initial phase, calculation type
"K0 procedure", and (ii) Phase 1, calculation type "Dynamic". The first phase is used only to initialize the stress
state with a K0=0.5, whereas the second is employed to solve the dynamic problem.

Figure 23: PLAXIS Input screenshots of the phase settings for Initial phase and Phase 1

PLAXIS 45 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
1D site response analysis
Results

Although in the K0 procedure the equilibrium equations are not solved, in this specific example the distribution
of the effective stress is also balanced with the gravity loading, therefore it would not be necessary to reset the
state variables at the beginning of the dynamic phase. As a matter of fact, parameters and state variables
depending on the effective stress state are computed at the beginning of the first phase in which PM4Silt is used
or when a reset of state variables is requested (see Model Parameters (on page 20)).
If PM4Silt is used with the calculation type "Gravity loading procedure", where the initial effective stress is
considered only as a first tentative, a reset of the state variables must be done before solving the dynamic
problem thus avoiding the use of wrong parameters and state variables. The incorrect initialization could also
generate potential numerical issues due to unrealistic values of the state variables. In some circumstances, it can
therefore be useful to perform the static calculation type "Gravity loading procedure" with a simpler constitutive
model (for example elastic perfectly plastic) for the soil to achieve an equilibrated effective stress distribution
and then inserting a plastic nil-phase before the dynamic phase in which the PM4Silt material is assigned to the
soil deposit. Following this approach, it is then not necessary to reset the state variables before performing the
dynamic analysis. After the change of constitutive models or after a reset of the state variables, the BCI state
variable can be plotted to check if the initial effective stress state was violating the bounding surface and
therefore an internal and automatic correction of the stress state has been performed to restore the consistency
with the bounding surface.
Some of the state variables remain constant after the initialization (see State variables (on page 28)). All the
other quantities are updated at each step of the calculation. Therefore, for example, a check of the initial stiffness
distribution should be done in the plastic nil-phase or at the first step of the dynamic phase. The distribution of G
and K can be plotted to double check the calibration of the parameters G0 and n G . The dominant frequency 1.24
Hz of the deposit has been computed by exporting, through a cross-section, the initial elastic stiffness, and by
computing, using the saturated density, the shear wave velocity profile, and the equivalent value of the shear
∑ h i / V s,i
i
wave velocity of the deposit as V s,eq = H
, H being the height of the deposit and h i and V s,i
corresponding to the thickness and the associated value of the shear wave velocity at the specific depth,
respectively.
As shown in Figure 23 (on page 45), the number of steps is set equal to the number of points of the input
accelerogram to have the same definition of input and output time histories. Before running the analyses, two
nodes have been selected in correspondence to the top and bottom of the model to be able to retrieve the entire
history of accelerations.
The accelerations in Figure 20 (on page 42) have been considered as outcrop motions and applied with a scaling
coefficient equal to 0.5 on the line displacement at the bottom of the model. Considering that the input was
scaled by a factor of 2, it is evident that the results would have been the same directly applying the original
natural accelerogram. The choice has been done to remark that, when a compliant base is used, only the upward
component of the motion should be applied at the base, being equal to 0.5 the outcrop motion in the case of a
record performed on a stiff bedrock.

PLAXIS 46 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
1D site response analysis
Results

5.1 Results
Figure 24 (on page 47) shows the accelerations predicted at the top and the base of the column on the
preselected nodes. A de-amplification of the peak acceleration is observed. The amplification of the harmonic
with a frequency smaller than 1.24Hz (see the Fourier's spectrum in Figure 25 (on page 47)) is justified by an
expected degradation of the stiffness. Although the input has been filtered, harmonics with frequencies higher
than 18Hz are still generated during the computation, particularly in the top node. However, their influence on
the results is moderate, as shown by the acceleration at the top, confirming the benefic effects of the Rayleigh
damping and the filtering of the input.

Figure 24: Output of the horizontal acceleration history at the base and the top of the column

Figure 25: Fourier (left) and Pseudo Acceleration Spectra (right) of the accelerations at the top and the base of the
column

Figure 26 (on page 48) shows the distribution of the maximum accelerations and displacements reached
during the first 30s of the analysis along a vertical cross-section in correspondence to the mid-line. An overall
de-amplification is observed, which is in accord with the dissipative response of silts.

PLAXIS 47 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
1D site response analysis
Results

30.0 30.0

t = 30 s t = 30 s

25.0 25.0

20.0 20.0
y (m)

y (m)
15.0 15.0

10.0 10.0

5.0 5.0

0.0 0.0
0.00 0.50 1.00 1.50 2.00 0.00 0.05 0.10 0.15 0.20 0.25
Extreme |ax| (g) Extreme |ux| (m)

Figure 26: Cross-section profiles of the maximum horizontal accelerations and displacements after 30s

In Figure 27 (on page 48) the cross-section of the state variables γ / 2 MaxExtreme , ru, Extreme and
τxy,ratio,Extreme are shown at the instant of Figure 26 (on page 48). In 2D boundary values problems, the iso-
lines of γ / 2 MaxExtreme would represent the maximum deviatoric strains although, in the conditions of the
analysis (undrained and one-dimensional), γ / 2 MaxExtreme coincides with εxy, Extreme . The results show
localization of shear strains higher than 3% at 5m of depth associated to high values of ru, Extreme

30.0 30.0 30.0

t=30s t=30s t=30s

25.0 25.0 25.0

20.0 20.0 20.0

y (m)

y (m)
y (m)

15.0 15.0 15.0

10.0 10.0 10.0

5.0 5.0 5.0

0.0 0.0 0.0

0.00 2.00 4.00 6.00 0.00 0.20 0.40 0.60 0.80 1.00 0.00 0.20 0.40 0.60 0.80 1.00
gMax, Extreme (%) t xy, ratio ,Extreme (-) ru,Extreme (-)

Figure 27: Cross-section profiles of the three state variables after 30s

PLAXIS 48 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
1D site response analysis
Results

Looking at what is reported in Figure 28 (on page 49), the Ipr close to one indicates that the highest value of
ru,Extreme in Figure 27 (on page 48)coincides with ru, Limit ( as shown in Figure 29 (on page 50),
ru, Limit = 0.842). Therefore, in the portion of the layer where the shear strains are greater than 3%, the
minimum admissible mean effective stress has been reached . Note that ru,max in Figure 29 (on page 50) is
equal to zero because the pmin used to compute ru, Limit has been determined based on Su,ratio .

Figure 28: Index of reduction of p and cyclic mobility ("Liquefied points") distributions

PLAXIS 49 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
1D site response analysis
Results

Figure 29: Distributions of ru, Limit and ru,max for the calibration A ( Table 1)

The results reported in the previous figures show that even if the level of the mean effective stress does not
approach zero (the soil still retains ~80% of the initial mean effective stress), high levels of shear strains are
mobilized and high levels of excess pore water pressure develop, indicating permanent damage and post shaking
settlements in the deposit. The significant damage is quite localized for the case illustrated in this example,
although worse conditions could be activated with initial static shear stresses, which also reduces the cyclic
strength of the material (see Eq. [59]). These results obtained for a lithostatic and very schematic condition
indicate the importance of modeling the cyclic mobility and cyclic softening of clay-like soils and their
peculiarities compared to the sand-like liquefaction (see for instance Tutorial 3 of the PLAXIS PM4Sand manual).

PLAXIS 50 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
 References
6
1. Been, K. and Jefferies, M.G., 1985. A state parameter for sands. Géotechnique, 35(2), pp.99-112.
2. Boulanger, R.W. and Idriss, I.M., 2004. Evaluating the potential for liquefaction or cyclic failure of silts and
clays (p. 131). Davis, CA: Center for Geotechnical Modeling.
3. Boulanger, R.W. and Idriss, I.M., 2006. Liquefaction susceptibility criteria for silts and clays. Journal of
geotechnical and geoenvironmental engineering, 132(11), pp.1413-1426.
4. Boulanger, R.W. and Idriss, I.M., 2007. Evaluation of cyclic softening in silts and clays. Journal of geotechnical
and geoenvironmental engineering, 133(6), pp.641-652.
5. Boulanger, R.W. and Ziotopoulou, K., 2017. PM4Sand (Version 3.1): A sand plasticity model for earthquake
engineering applications. Center for Geotechnical Modeling Report No. UCD/CGM-17/01, Department of Civil
and Environmental Engineering, University of California, Davis, CA.
6. Boulanger, R.W., Moug, D.M., Munter, S.K., Price, A.B. and DeJong, J.T., 2016. Evaluating liquefaction and
lateral spreading in interbedded sand, silt, and clay deposits using the cone penetrometer. Australian
Geomechanics Society Geotechnical and Geophysical Site Characterisation 5, B. M. Lehane, H. Acosta-
Martinez, and R. Kelly, eds., Sydney, Australia, ISBN 978-0-9946261-2-7.
7. Boulanger, R.W. and Ziotopoulou, K., 2018. PM4Silt (Version 1): A silt plasticity model for earthquake
engineering applications. Report No. UCD/CGM-18/01, Center for Geotechnical Modeling, Department of Civil
and Environmental Engineering, University of California, Davis, CA.
8. Boulanger, R.W. and Wijewickreme, D., 2019, October. Calibration of a constitutive model for the cyclic
loading response of Fraser River Delta Silt. In Earthquake Geotechnical Engineering for Protection and
Development of Environment and Constructions: Proceedings of the 7th International Conference on
Earthquake Geotechnical Engineering,(ICEGE 2019), June 17-20, 2019, Rome, Italy (p. 121). CRC Press.
9. Boulanger, R.W. and Ziotopoulou, K., 2019. A constitutive model for clays and plastic silts in plane-strain
earthquake engineering applications. Soil Dynamics and Earthquake Engineering, 127, p.105832.
10. Boulanger, R.W., 2019. Nonlinear Dynamic Analyses of Austrian Dam in the 1989 Loma Prieta Earthquake.
Journal of Geotechnical and Geoenvironmental Engineering, 145(11), p.05019011.
11. Dafalias, Y.F. and Manzari, M.T., 2004. Simple plasticity sand model accounting for fabric change effects.
Journal of Engineering mechanics, 130(6), pp.622-634.
12. Dahl, K.R., DeJong, J.T., Boulanger, R.W., Pyke, R. and Wahl, D., 2014. Characterization of an alluvial silt and
clay deposit for monotonic, cyclic, and post-cyclic behavior. Canadian Geotechnical Journal, 51(4), pp.
432-440.
13. Duncan, J.M., Wright., 2005. Soil strength and slope stability. New York Wiley.
14. Jefferies, M.G. and Been, K., 2015. Soil liquefaction: a critical state approach. CRC press.
15. Ladd, C.C. and Foott, R., 1974. New design procedure for stability of soft clays. Journal of Geotechnical and
Geoenvironmental Engineering, 100(Proc Paper 10064).
16. Ladd, C.C., 1991 Stability evaluation during staged construction: 22th Terzaghi Lecture” Journal of
Geotechnical Engineering ASCE (117 (4) 537-615.
17. Ladd, C.C. and DeGroot, D.J., 2004. Recommended practice for soft ground site characterization: Arthur
Casagrande Lecture. Massachusetts Institute of Technology.

PLAXIS 51 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
References

18. Montgomery, J., Boulanger, R.W., Armstrong, R.J. and Malvick, E.J., 2014. Anisotropic Undrained Shear
Strength Parameters for Nonlinear Deformation Analyses of Embankment Dams. In Geo-Congress 2014: Geo-
characterization and Modeling for Sustainability (pp. 1294-1306).
19. Price, A.B., Boulanger, R.W., DeJong, J.T., Parra Bastidas, A.M. and Moug, D., 2015, November. Cyclic strengths
and simulated CPT penetration resistances in intermediate soils. In 6th International Conference on
Earthquake Geotechnical Engineering (Vol. 14).
20. Price, A.B., DeJong, J.T. and Boulanger, R.W., 2017. Cyclic loading response of silt with multiple loading events.
Journal of Geotechnical and Geoenvironmental Engineering, 143(10), p.04017080.
21. Romero, S. (1995). "The behavior of silt as clay content is increased." MS thesis, University of California,
Davis.
22. Sheahan, T.C., Ladd, C.C. and Germaine, J.T., 1996. Rate-dependent undrained shear behavior of saturated
clay. Journal of Geotechnical Engineering, 122(2), pp.99-108.

PLAXIS 52 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
 Appendices

PLAXIS 53 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
 Dilatancy expressions
A
A.1 Dilating behaviour
Dnon-rot and Drot in Eq.47 and Eq.48 are selected according to the following statement, if

Dnon-rot < Drot Eq. [68]

then D = Dnon-rot else if ( Dnon-rot ≥ Drot ) then

M b - M cur
D = Dnon-rot + ( Drot - Dnon-rot ) ⋅ ,
Eq. [69]
M b - M cur + 0.01
where
M cur = The current stress ratio η.

The parameter Ad in the equations of Drot and Dnon-rot is expressed as in Eq. [70],
Ado (C zin2)

( )( )(
Ad = 2
Eq. [70]
zcum -z : n 3
1- Cε )2(C pzp )(C zin1)(C zin2) + 1
zmax 2 ⋅ z peak

where
Ado and C ε = Model parameters which for silts and clays should be selected
in the range of 0.8 - 1.2 and 0.5 - 1.3, respectively.
The other variables are defined as

( |
C zin1 = 1 - exp - 2 ⋅
zin : n - z : n
zmax |) Eq. [71]

(
C zin2 = 1 + Czin1 ⋅
zcum - z peak
3z
max
)/ 1 + 3Czin1 ⋅ ( zcum - z peak
3z
max
) Eq. [72]

/ ( )
p 5
C pzp = 1 1 + 2.5 pzp
Eq. [73]

In the last equation pzp is defined as the mean stress at the time in which the product p ⋅ | z | reaches its
maximum value. To define the DF in case of soil dilation is necessary to define also the back-stress ratio tensor
for the rotated dilatancy, that is

M dR - m Md
αdR = n, M dR = Crot 1
Eq. [74]
2

PLAXIS 54 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Dilatancy expressions

where
M dR = A rotated stress ratio (i.e. a reduced inclination of M dR ).
Crot1 is equal to
-z : n
Crot 1 = 1 + 2 ⋅ ⋅( 1- C zin1 )Eq. [75]
2 ⋅ zmax

To ensure that the model will be dilative for mean stresses smaller than 2 pmin , an additional constraint on the
DF is enforced, i.e.
if ( p ≤ 2 pmin and D < Dmin ) then :

D = - 3.5Ado M b - M d ⋅ ( 2 pmin - p
pmin ) Eq. [76]

A.2 Contracting behaviour

The DF for contracting is expressed through the variables Cin , C p,min , Cdz and Cwet whose expressions are
2 z:n
Cin = ⋅ zmax
Eq. [77]
2

Cdz = 1 - 2 ⋅ C rot 2 ⋅ ( z peak

zmax ) (
⋅
zmax
zmax + zcum ⋅ C
rot 2
) Eq. [78]

z peak
Crot 2 = 1 - Eq. [79]
zcum + ( zmax
100
)
1 Cw1 4
Cwet = ≤1 1 ξ 2
1 1 Cwet 1 = 1 + Cwet 2 = 1 + ⋅ Eq.
Cwet 1
+
Cwet 2 (α b
- α) : n
Cw2 λ

[80]
where
Cw1 = Parameters equal to 0.02.
Cw2 = Parameters equal to 0.1.

The value of C p,min depends on the value of the mean stress according to the following conditions:

C p,min = 0 if ( p ≤ 2pmin ) Eq. [81]

C p,min = 1 if ( p > 8 pmin ) Eq. [82]

p - 2 ⋅ p min
otherwise C p,min = 6 ⋅ p min
Eq. [83]

PLAXIS 55 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
 Anisotropic undrained strength for non-linear
B
deformation analyses
In the current implementation of PM4Silt an undrained strength Su variable for every stress point can be
considered. The method proposed by Montgomery et al. (2014) (on page 52) has been implemented. The
sequence of computation performed in every single stress point is briefly recalled. A more detailed description
of this method and examples of application can be found in Montgomery et al. (2014) (on page 52) and
Boulanger (2019) (on page 51).
The undrained shear stress at failure is equated to the undrained resistance and computed as shown in Eq. [84],
τ ff ,
( K c =K f ) - τ ff , ( K c =1)
Su ≡ τ ff , Kc = τ ff , ( K
c
=1) + Kf -1 ( K c - 1) Eq. [84]

where
Kc = The consolidation stress ratio (i.e K c = σ1c / σ3c , with σ1c and σ3c the
maximum and minimum principal stresses at consolidation).
Kf = The stress ratio at failure (i.e K f = σ1f / σ3f , with σ1f and σ3f , the
maximum and minimum principal stresses at failure).
The shear stress on the eventual failure planes τ ff for consolidation at K c = 1 and K c = K f is computed as
shown in Eq. [85] and Eq. [86],
τ ff , K c =1 = d R + σ fc ⋅ tan (ψR ) Eq. [85]

τ ff , ( K
c
=K f ) = cc + σ fc ⋅ tan (ϕc ) Eq. [86]

where
cc , ϕc , d R and ψR = The parameters to be calibrated via ICUTX tests, while K c
and K f are calculated internally from the principal
stresses at consolidation and failure, corresponding to
1

σ1c = ( σv - σh
2
)+ ( σv - σh 2
2
+ ) τvh
2
Eq. [87]

σ3c = ( σv - σh
2
)- ( σv - σh 2
2
+ ) τvh
2
Eq. [88]

and

PLAXIS 56 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering
Anisotropic undrained strength for non-linear deformation analyses

τ ff - c' τ ff
σ1 f = tan (ϕ' )
+ τ ff tan (ϕ ' ) + Eq. [89]
cos (ϕ' ),

τ ff - c' τ ff
σ3 f = + τ ff tan (ϕ ' ) - , Eq. [90]
tan (ϕ' ) cos (ϕ ' )

respectively. In the previous equations, the parameters c' and ϕ' coincide with ϕc and cc , respectively, as
explained in Critical State Soil Mechanics framework (on page 7).

PLAXIS 57 User Defined Soil Models - PM4Silt: A Silt Plasticity

model for Earthquake Engineering