11- Softening-Hardening Material Model
Experimental evidence indicates that the plastic deformation in soils starts from the early stages of loading.
To capture such a behavior in a constitutive model the typical elasto-perfect plastic models are not adequate.
To simulate such behavior constitutive models that utilize a hardening law after initial yielding are required.
The Softening-Hardening Model in RS2 and RS3 has been developed to meet the abovementioned need.
The model can utilize up to three yield surfaces that includes deviatoric (shear), volumetric (cap) and
tension cut off. The yield surfaces and hardening characteristics of this model are illustrated in Figure 11.1
in p-q plane and Figures 11.2 and11.3 show the yield surfaces in 3D stress space. Based on the formulations
of this model it is apt to say that this model has three different mechanisms, i.e. deviatoric, volumetric and
tension cutoff. The model is designed to be very flexible with its numerous options and formulations. It can
take advantage of the various options of elastic behavior that were presented in the Chapter 2. In its simplest
form this model can be equivalent to Mohr-Coulomb model, with activation of a few options can replace
the Duncan-Change model or the ChSoil model, and by taking advantage of more options, including
nonlinear elasticity, it can be equivalent to the Hardening Soil, Double yield and CySoil model.
The formulations of these three mechanisms, definition of yield surfaces and their corresponding plastic
potential and hardening law are presented in the following sections (Pietruszczak 2010).
q q
1 1
Mf Mf
Yield Surface Yield Surface
Hardening Hardening
pc p pc p
(a) (b)
Figure 11.1 - The yield surfaces of the Softenign-Hardening model; a) Deviatoric yield surface
(red) and the vertical cap (green); b) Deviatoric yield surface (red) and elliptical cap (blue)
Figure 11.2- Yield surface of Softening-Hardening model with vertical cap in 3D stress space
Figure 11.3- Yield surface of Softening-Hardening model with elliptical cap in 3D stress space
11.1- Deviatoric Hardening Mechanism
The deviatoric mechanism is the core of this model and its yield surface is very similar to the yield surface
of the Mohr-Coulomb model.
The equation for the Mohr-Coulomb yield surface using the (𝑝𝑝, 𝑞𝑞, 𝜃𝜃) invariants is given by
𝑐𝑐
𝐹𝐹𝑠𝑠 = 𝑞𝑞 + 𝑀𝑀 �𝑝𝑝 − tan 𝜑𝜑� = 0 (11.1)
where
3 sin 𝜑𝜑
𝑀𝑀 = (11.2)
√3 cos 𝜃𝜃−sin 𝜃𝜃 sin 𝜑𝜑
The hardening for this yield surface is considered for the mobilized friction angle (and cohesion) and it is
attributed to plastic distortion. The equations above are rewritten as:
𝑐𝑐 3 sin 𝜑𝜑
𝐹𝐹𝑠𝑠 = 𝑞𝑞 + 𝑀𝑀 �𝑝𝑝 − tan 𝜑𝜑 � = 0 , 𝑀𝑀 = (11.3)
𝑓𝑓 √3 cos 𝜃𝜃−sin 𝜃𝜃 sin 𝜑𝜑
In above 𝜑𝜑𝑓𝑓 is the ultimate/failure friction angle and 𝑐𝑐 and 𝜑𝜑 are the mobilized cohesion and friction angle
respectively.
There are two types of hardening law considered for this model. The first one uses a relationship between
tan 𝜑𝜑 and the deviatoric plastic strain presented in equation 11.4.
𝑝𝑝
𝜀𝜀𝑞𝑞
tan 𝜑𝜑 = tan 𝜑𝜑𝑓𝑓 𝜀𝜀𝑝𝑝 +𝐴𝐴 (11.4)
𝑞𝑞
𝑝𝑝
where 𝜀𝜀𝑞𝑞 is the deviatoric plastic strain that is generated by the deviatoric mechanism only, and 𝐴𝐴 is a
positive and constant hardening parameter.
The second hardening law uses custom tabular piecewise linear values for the mobilized friction angle and
𝑝𝑝 𝑝𝑝
cohesion with deviatoric plastic strain �𝜀𝜀𝑞𝑞 , 𝜑𝜑� & �𝜀𝜀𝑞𝑞 , 𝑐𝑐� .
The first hardening law is quite simple and uses only one additional parameter on top of what is required
for a simple Mohr-Coulumb material model. The second one gives the user the flexibility to define any
kind of hardening/softening rule that meets their material modelling needs.
Two options are considered for the plastic potential of the deviatoric hardening yield surface. The first one
is very similar to that of the Mohr-Coulomb model where a dilation angle (𝜓𝜓𝑓𝑓 ) is defined and controls the
dilation tendency of the material. The only difference here is that the defined dilation angle is the dilation
angle at failure, where the mobilized friction angle reaches its ultimate value (𝜑𝜑 = 𝜑𝜑𝑓𝑓 ). The actual dilation
angle used during plastic flow is proportional to the ratio of the mobilized friction angle to its failure value.
The plastic potential function for this option is given in equation 6.
𝜑𝜑
𝜓𝜓 = 𝜓𝜓𝑓𝑓 𝜑𝜑 (11.5)
𝑓𝑓
3 sin 𝜓𝜓
𝑄𝑄𝑠𝑠 = 𝑞𝑞 + 𝑀𝑀𝜓𝜓 𝑝𝑝 = 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 , 𝑀𝑀𝜓𝜓 = (11.6)
√3 cos 𝜃𝜃−sin 𝜃𝜃 sin 𝜓𝜓
The other option for plastic potential is to define a compaction-dilation angle in such a way that where the
mobilized friction angle (𝜑𝜑) is less than this angle (𝜓𝜓) the volumetric plastic strain is positive (compaction)
and when the mobilized friction angle is greater than 𝜓𝜓 the volumetric plastic strain is negative (dilation).
The plastic potential function for this case is:
−𝑝𝑝 3 sin 𝜓𝜓
𝑄𝑄𝑠𝑠 = 𝑞𝑞 − 𝑀𝑀𝜓𝜓 𝑝𝑝𝑝𝑝𝑝𝑝 � 𝑝𝑝 � = 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 , 𝑀𝑀 = (11.7)
𝑐𝑐 √3 cos 𝜃𝜃−sin 𝜃𝜃 sin 𝜓𝜓
When used solely, this mechanism, can be a good alternative to the Duncan-Chang model and does not
have its short fallings. This is also a good alternative for the ChSoil model, and using the various options
for nonlinear elasticity and the flexibility of the tabular hardening rule gives lots of flexibilities to the
behaviors this mechanism can simulate. The tabular hardening option can also be used to simulate a
softening behavior. Note that simulating the softening behavior as a material behavior is not recommended
since it will make the finite element solution results unreliable and mesh dependent. The apparent softening
behavior that is observed in laboratory tests is to be simulated, for example, by taking advantage of more
complex formulations that can deal with strain localization shear banding.
11.2- Volumetric Hardening Mechanism
The main role of the volumetric mechanism (cap) is to close the elastic domain in space (𝑝𝑝 − 𝑞𝑞) on the
hydrostatic (𝑝𝑝) axis and simulate the densification/compaction of the material. Addition of the the cap yield
is optional in this model but when activated there are two options for the it, vertical cap and elliptical cap.
The yield surface of the vertical cap is defined as follows.
𝐹𝐹𝑐𝑐 = 𝑝𝑝 + 𝑝𝑝𝑐𝑐 = 0 (11.8)
where 𝑝𝑝𝑐𝑐 is the location of the intersection of this yield surface with the 𝑝𝑝 axis. With activation of the
vertical cap this model is equivalent to Double-Yield model.
The elliptical cap is very similar to the yield surface of the modified Cam-Clay model with an offset to
consider the cohesion that was defined in the deviatoric mechanism.
2
𝑞𝑞 𝑐𝑐 3 sin 𝜑𝜑𝑓𝑓
𝐹𝐹𝑐𝑐 = �𝑀𝑀 � + �𝑝𝑝 − 𝑡𝑡𝑡𝑡𝑡𝑡𝜑𝜑 � (𝑝𝑝 + 𝑝𝑝𝑐𝑐 ) = 0 , 𝑀𝑀𝑓𝑓 = (11.9)
𝑓𝑓 𝑓𝑓 √3 cos 𝜃𝜃−sin 𝜃𝜃 sin 𝜑𝜑𝑓𝑓
Activation of the elliptical cap makes this model similar to the Hardening Soil and CySoil models. The
difference between the cap in this model with the cap in the Hardening Soil and CySoil is that the apex of
this cap tracks the ultimate shear yield surface. This cap is similar to the yield surface of modified Cam
Clay model in that at ultimate shear failure the cap will be at critical state condition and will not generate
any compaction.
The hardening for these yield surfaces is considered for 𝑝𝑝𝑐𝑐 and it is attributed to volumetric plastic strain.
The built in function for the hardening follows the same hardening law as in the Modified Cam-Clay model:
𝑝𝑝
Δ𝜀𝜀𝑣𝑣
(𝑝𝑝𝑐𝑐 )𝑛𝑛+1 = (𝑝𝑝𝑐𝑐 )𝑛𝑛 exp � � (11.10)
𝜆𝜆
𝑝𝑝
where 𝑛𝑛 is the step number, 𝜀𝜀𝑣𝑣 is the volumetric plastic strain, and 𝜆𝜆 is the difference between the slope of
normal consolidation line and the swelling line. This hardening rule is similar to the hardening rule of Cam
Clay model.
Tabular hardening law which uses custom tabular piecewise linear values for 𝑝𝑝𝑐𝑐 versus volumetric plastic
𝑝𝑝
strain �𝜀𝜀𝑣𝑣 , 𝑝𝑝𝑐𝑐 � is also available for this model. This option will give the flexibility to use any other
hardening law for the cap, e.g. the hardening rules used in Hardening Soil or CySoil models
The flow rule is associated for this yield surface.
11.3- Tension Cut off
This mechanism is to incorporate the tensile strength of the material to this model. In this mechanism the
minor principal stress is limited to the tensile strength of the material. The flow rule is associated and the
mechanism has no hardening.
FT = σ 1 − T = 0 (11.11)
In above T is the tensile strength of the material.
11.4- Examples
Figure 11.4 and 11.5 shows the experimental and numerical results of drained triaxial tests on loose and
dense Hostun sand respectively. The experimental results of drained triaxial test on dense and loose Hostun
sand are depicted from Schanz and Vermeer (1996). All the triaxial tests start from an initial hydrostatic
confinement of 300 kPa. The numerical results are obtained by using only the deviatoric hardening
mechanism with compaction-dilation option for the plastic flow and a linear elastic behavior. The hardening
rule for the loose Hostun sand uses the form that is presented in Equation 11.4. The Dense sand is simulated
once by using the same hardening rule as in Equation 11.4 and also by using a tabular function to capture
the softening behavior. The model parameters are presented next to the graphs. Stress paths of the drained
tests include variations of axial stress and volumetric strain with increasing axial strain, variation of
deviatoric stress with deviatoric strain and the stress path in p-q plane. The ultimate yield surface of the
deviatoric mechanism is also shown in the p-q plane. The simulated behavior captures the hardening
behavior and also the gradual compaction of loose sand. In the case of dense sand, the initial compaction
that follows with dilation is fully captured with the help of the appropriate choice of plastic flow option.
The hardening of dense sand is captured by both options of the hardening rules that are used here, but only
the tabular hardening option can capture softening behavior.
Figure 11.6 shows the experimental and numerical results of an undrianed triaxial test on loose Banding
sand. The experimental results of undrained triaxial test on Banding sand are depicted Castro (1969). The
triaxial test starts at the initial confinement of 400 kPa. The generation of excess pore water pressure in
loose sands under undrained conditions can lead to static liquefaction and total loss of strength. This
phenomenon can be observed in Figure 6. The simulation results presented in this figure are in good
agreement with the experimental results.
E = 180 MPa
ν = 0.35
ϕ f = 35o
ψ = 33.5o
A = 0.0038
Figure 11.4. Stress paths of drained triaxial tests on loose Hostun sand
E = 300 MPa
ν = 0.35
ϕ f = 43.5o
ψ = 30o
A = 0.0036
Figure 11.5. Stress paths of drained triaxial tests on dense Hostun sand
Figure 11.5. Stress paths of undrained triaxial tests on Banding sand
References
Castro, G. (1969). Liquefaction of sands, Harvard Soil Mechanics Series, No.81, Pierce Hall.
Pietruszczak, S. (2010). Fundamentals of Plasticity in Geomechanics. CRC Press.
Schanz, T. and Vermeer, P.A. (1996). Angles of friction and dilatancy of sand, Géotechnique 46(1): 145-
151.
