• The proposed model features the anisotropic damage evolution with regularization using

gradient-enhancement. The model has the ability to predict the anisotropic evolution of
damage under multi-axial proportional and non-proportional loading conditions.
• The concept of damage effect tensors is utilized in the formulation such that the damage can
be described in a general framework rather than an orthotropic one. Nevertheless, the
damage growth is evaluated in principal directions, which makes the computations relatively
easy. By transforming the principal damage components properly into the material
coordinate system, the formulation of anisotropic damage is achieved in a general
framework.
• The rate form of the damage evolution is derived from a simple exponential damage
evolution equation, this results in the evolution of principal damage components asymptotic
to unity, without taking any other measures. On the other hand, the rate of damage
evolution helps to capture any rotation of the principal damage direction with respect to the
strain directions.
• Moreover, to account for the tension-compression asymmetry in concrete, to separate
internal variables namely, equivalent strain in tension and and equivalent strain in
compression are used as the damage driving variables and are also the nonlocal variables
being regularized as part of graduent enhancement.
• "Biaxial Testing of Concrete under Static and Dynamic Loading" by Gergely and Lutz. This
study investigated the behavior of concrete under biaxial loading using both static and
dynamic loading conditions. The authors used a servo-hydraulic biaxial testing machine to
apply biaxial loading to concrete specimens and measured the strength and stiffness of the
material.
• "Biaxial Testing of Concrete with Various Stress Paths" by Riedel and Willam. This study
investigated the behavior of concrete under biaxial loading with various stress paths,
including equibiaxial tension and compression, and deviatoric tension and compression.
The authors used a biaxial testing machine with hydraulic actuators to apply the loading and
measured the material response using strain gauges.
• "Biaxial Testing of Reinforced Concrete Beams" by Kupfer and Hilsdorf. This study
investigated the behavior of reinforced concrete beams under biaxial loading. The authors
used a biaxial testing machine with hydraulic actuators to apply the loading and measured
the material response using strain gauges and displacement transducers.
• "Biaxial Testing of Concrete with Different Strengths" by Gopalaratnam and Reddy. This
study investigated the behavior of concrete with different strengths under biaxial loading.
The authors used a biaxial testing machine with hydraulic actuators to apply the loading and
measured the material response using strain gauges and displacement transducers.
• "Biaxial Testing of Concrete Cylinders and Prisms" by Bari and Bakhtiyarov. This study
investigated the behavior of concrete cylinders and prisms under biaxial loading. The
authors used a biaxial testing machine with hydraulic actuators to apply the loading and
measured the material response using strain gauges.

• Nonlocal
• The numerical analysis of ductile damage and failure in engineering materials is
often based on the micromechanical model of Gurson [1]. Numerical studies in the
context of the finite-element method demonstrate that, as with other such types of
local damage models, the numerical simulation of the initiation and propagation of
damage zones is not reliable and strongly mesh-dependent. The numerical
 problems concern the global load-displacement response as well as the onset, size,
and orientation of damage zones and thus to the reliability of the obtained results.
From a mathematical point of view, this problem is caused by the loss of ellipticity of
the set of partial differential equations determining (the rate of) the deformation
field. Physically it has to do with the fact that the effect of interaction in the
microstructure on the material behaviour has not been taken into account.
• When the gradients of state variables are used in the formulation, additional
degrees of freedom need to be implemented for Finite Element Analysis, for
instance the third-order stress tensor (conjugate to the gradient of deformation). In
integration based nonlocal models, a variable at a point is calculated as a weighted
average over a certain neighborhood of that point [36–40]. The weights that quantify
the intensity of the interaction between Gauss points is tabulated, so that each
Gauss point interacts with the Gauss points in its neighborhood. The size of the
neighborhood is determined by an internal length parameter.
•
•
• Enriched continuum damage models
•
• Classical continuum models cease to give a meaningful representation of the
mechanical behaviour of materials when the scale of observation approaches that
of the microstructure of the [Link] material can no longer be considered
homogeneous at this scale and descriptions in terms of the classical stresses and
strains thus break down. In particular, the typical dimension of the microstructure
(e.g., grain size, lattice distance, ®bre length etc.) manifests itself in the mechanical
response of the material, for instance in size e€ects. A similar situation arises in
larger components when the scale of variation of the relevant variables becomes
comparable to the microstructural scale. This may be the case for instance in wave
propagation analyses and at stress concentrators such as notches and cracks.
• The need to describe the microstructure-dominated behaviour in an averaged
sense, without modelling the full detail of the microstructure and the deformation
processes at the microscale, has led to the development of enriched continuum
models. In these models, nonstandard deformation and/or stress quantities account
for the in¯uence of the microstructure on deformation processes. Taking into
account long-range cohesive forces, which are neglected in the standard theory,
has led to the classical Toupin± Mindlin gradient theory of elasticity (Toupin, 1962;
Mindlin, 1965), see also Mindlin (1964) and Eringen (1966). In this theory ®rst,
second and possibly higher-order gradients of the strain and stress enter the
constitutive model in order to describe the e€ect of the microstructure. More
recently, some of these ideas have been extended to plasticity (Aifantis, 1984; Fleck
and Hutchinson, 1993; Steinmann, 1996). Following arguments similar to those of
Toupin and Mindlin, Kroner, Eringen, Edelen and others have arrived at nonlocal
formulations of elasticity (Kroner,
 1967; Eringen and Edelen, 1972). In the nonlocal
theory long-range interactions are taken into account by weighted spatial averages
of constitutive quantities. Both approaches introduce at least one length scale into
the continuum modelling, which is related to the distance at which spatial
interactions occur and thus to the scale of the microstructure of the material.
• Starting from the mid-1980s, nonlocal and gradient continua have regained interest
because of their potential as localisation limiters. When used to model degradation
of the mechanical properties of a material, standard continuum models tend to
predict a response in which the entire degradation process is concentrated in a
vanishing volume. Apart from often being physically unrealistic, this localisation
results in a perfectly brittle macroscopic fracture behaviour instead of the intended
 gradual loss of sti€ness and/or strength. In numerical analyses, the volume
contributing to the failure process is set by the spatial reso- lution of the
discretisation. As a result, analyses become sensitive to the discretisation (mesh
sensitivity) and converge to the nonphysical, perfectly brittle response in the limit of
an in®nite spatial resolution. In reality, the microprocesses which are responsible
for material damage usually extend over a band, the thickness of which is
comparable to the characteristic microstructural length (e.g., the average grain
size). This means that in a continuum mechanics description constitutive variables
show strong variations at the scale of the microstructure. The pathological
localisation of damage and deformation observed in standard continuum models is
a direct consequence of the inability of these models to describe variations at the
microstructural scale. This understanding has led to the use of nonlocal and
gradient formulations to avoid pathological localisation. Both approaches have been
successfully used to limit the localisation process to the microstructural scale.
•
• Discrete and smeared crack models
•
• The numerical simulation of fracture in quasi-brittle materials was initiated in the late
1960s. Two markedly different approaches werepioneered, namely the discrete
crack models [6] and the smeared crack models [7]. The discrete crack approach is
intuitively appealing: a crack is introduced as a discontinuity in the geometry.
Hence, the topology of the body is changing continuously when cracks propagate or
when more cracks nucleate. Indeed, the discrete crack model basically aims at
simulating the initiation and propagation of a limited number of (clearly) visible
cracks.
• On the other hand, the smeared crack model is based on the idea that in quasi-
brittle materials, due to the heterogeneity which can be observed at a mesoscopic
level, and the possible presence of reinforcement such as glass or polymer fibres,
or steel bars, many small cracks nucleate which link up only in a later stage to form
one or more dominant cracks. Since not each individual crack is resolved the
smeared crack model captures the deterioration process through a constitutive
relation
•