Professional Documents
Culture Documents
Multi Surface Model Aniso Brittle Mat
Multi Surface Model Aniso Brittle Mat
Quasi-brittle Materials
Paulo B. Lourenço1 and Jan G. Rots2
Abstract. A novel yield criterion capable of modelling the a few numerical implementations and calculations
softening behaviour of anisotropic materials under plane have actually been carried out. An example is given in
stress conditions is presented. Individual yield criteria are [1] where the implementation of an elastic-perfectly-
considered for tension and compression, according to two plastic Hill criterion is fully treated. In principle,
different failure mechanisms. The former is associated with
hardening behaviour could be simulated with the
a localised fracture process, denoted by cracking of the
material, and, the latter, is associated with a more fraction model [2] but not much effort has been done
distributed fracture process which is usually termed as in this direction. More recent attempts are given in
crushing of the material. The model is capable of [3], where linear tensorial hardening is included in the
reproducing independent (in the sense of completely Hill criterion, and [4], where linear hardening is
diverse) elastic and inelastic behaviour along a prescribed included in a modified (pressure dependent) Von
set of material axes. The energy-based regularisation Mises to fit either the uniaxial tensile or the com-
technique resorts then to four different fracture energies pressive behaviour.
(two in tension and two in compression). While problems can be acute in the implementation
of isotropic plasticity models, they can become even
1 INTRODUCTION more pronounced for anisotropic models where
algebraic simplifications are hardly possible. The
The difficulties in accurately modelling the behaviour present article represents, thus, a step further in the
of orthotropic materials are, usually, quite strong. This formulation of anisotropic plasticity models.
is due, not only, to the fact that comprehensive Individual yield criteria are considered for tension and
experimental results (including pre- and post-peak compression, according to different failure
behaviour) are generally lacking, but also to intrinsic mechanisms, one in tension and the other in
difficulties in the formulation of orthotropic inelastic compression. This represents an extension of
behaviour. It is noted that a representation of an conventional formulations for isotropic quasi-brittle
orthotropic yield surface solely in terms of principal materials to describe orthotropic behaviour. The
stresses is not possible. For plane stress situations, proposed yield surface combines the advantages of
which is the case here, a graphical representation in modern plasticity concepts with a powerful
terms of the full stress vector in a predefined set of representation of anisotropic material behaviour,
material axes (σx, σy, τxy) is necessary. which includes different hardening/softening
In this article, the theory of plasticity is utilised to behaviour along each material axis. The behaviour of
combine anisotropic elastic behaviour with the model is demonstrated by means of single element
anisotropic inelastic behaviour. Even if many tests and a comparison between numerical results and
anisotropic plasticity models have been proposed from experimental results for the case of masonry shear
1 Delft University of Technology, Faculty of Civil walls. The comparison shows good agreement both
Engineering, The Netherlands. Presently back at the for ductile and brittle failure modes.
Department of Civil Engineering, School of Engineering,
University of Minho, Azurém, P-4800 Guimarães,
Portugal 2 PROPOSED YIELD SURFACE
2 Delft University of Technology, Faculty of Civil Engi-
Different approaches for the conception of a yield
neering, The Netherlands. Also at TNO Building and
Construction Research, P.O. Box 49, 2600 AA Delft, The
surface can be used. One approach is to describe the
Netherlands material behaviour with a single yield criterion. The
Hoffman criterion is quite flexible and attractive to
© 1996 P.B. Lourenço and J.G. Rots use, see e.g. [5], but yields a non-acceptable
ECCOMAS 96. representation of quasi-brittle materials, see [6], with
Published in 1996 by John Wiley & Sons, Ltd. very poor fit of the experimental values. A single
purely theoretical and experimental standpoints, only
criterion fit of experimental data would lead to an tensile strength even if in the perpendicular direction
extremely complex yield surface with a mixed harden- damage has already occurred. A solution for this
ing/softening rule in order to describe properly the problem seems quite complex, see e.g. [8]. Therefore,
inelastic behaviour. It is believed that this approach is the scalar κt measures the amount of softening
practically non-feasible. Thus, a different approach simultaneously in the two material axes, even though
will be adopted. Formulations of isotropic quasi- the model still incorporates two different fracture
brittle materials behaviour consider, generally, differ- energies.
ent inelastic criteria for tension and compression. In The expression for the Rankine criterion, cf.
the present study, an extension of [7], where this eq. (1), can be rewritten as
approach is utilised for concrete with a Rankine and a
Drucker-Prager criterion, will be presented. In order (σ x − σ t (κ t )) + (σ y − σ t (κ t ))
to model orthotropic material behaviour, a Hill type f1 = +
2
criterion for compression and a Rankine type criterion
2
for tension, see Fig. 1, are proposed. Note that the (σ x − σ t (κ t )) − (σ y − σ t (κ t ))
word type is used here because the yield surfaces + τ xy2
2
adopted are close to the original yield criteria. (2)
Nevertheless, they represent solely a fit of experimen-
tal results. where coupling exists between the stress components
and the yield value. Setting forth a Rankine type
criterion for an orthotropic material, with different
yield values σ tx ( κ t ) and σ ty ( κ t ) along the x, y
directions is now straightforward if eq. (2) is modified
to
(σ x − σ tx (κ t )) + (σ y − σ ty (κ t ))
f1 = +
2
(3)
2
(σ x − σ tx (κ t )) − (σ y − σ ty (κ t ))
Figure 1. Proposed composite yield surface with iso-shear +α τ xy2
2
stress lines. Different strength values for tension and
compression along each material axis
where the parameter α, which controls the shear stress
contribution to failure, reads
3 FORMULATION OF THE MODEL
The most relevant aspects of the proposed yield f tx f ty
α= (4)
criteria are given next. For a complete description of τ
2
u, t
the model the reader is referred to [6].
Here, ftx, fty and τu,t are, respectively, the uniaxial
tensile strengths in the x, y directions and the pure
3.1 A Rankine type criterion shear strength. Note that the material axes are now
An adequate formulation of the Rankine criterion is fixed with respect to a specific frame of reference and
given by a single function, which is governed by the it shall be assumed that all stresses and strains for the
first principal stress and one yield value σ t which elastoplastic algorithm are given in the material
reference axes.
describes the softening behaviour of the material as,
Eq. (3) can be recast in a matrix form as
see [7],
T 1 T
f 1 = ( 1 2 {ξ} [ Pt ]{ξ}) 2 + 1 2 {π} {ξ} (5)
2 (1)
σ x +σ y
σ x −σ y
f1 = + +τ xy2 −σ t (κ t )
where the projection matrix [Pt] reads
2 2
where the scalar κt controls the amount of softening. 12 − 12 0
The assumption of isotropic softening is not [Pt ] = − 1 2 12 0
completely valid for quasi-brittle materials such as
0 0 2 (6)
concrete or masonry which can be loaded up to the
κ⋅ t = ε⋅1 = + 1 2 ( ε⋅ x − ε⋅ y ) + ( γ⋅ xy )
p p p 2 p 2
(12)
2
σ cy (κ c )
2 β 0
σ cx (κ c )
σ cx (κ c )
[Pc ] = β 2 0
σ cy (κ c )
0 0 2γ
(17)
the yield value σ c is given by
σ c ( κ c ) = σ cx ( κ c ) σ cy ( κ c ) (18)
Example 1
Example 2
Ex Ey νxy Gxy
2460 N/mm2 5460 N/mm2 0.18 1130 N/mm2
Figure 3. Possible behaviour of the model along the
material axes for two different sets of material parameters. Table 2. Rankine material parameters (α = 1.73)
ftx fty Gfx Gfy
4 EXAMPLES 0.28 N/mm2 0.05 N/mm2 0.02 N/mm 0.02 N/mm
The performance of the anisotropic continuum model Table 3. Hill material parameters (β = -1.05, γ = 1.20)
is validated next by a comparison with experimental
fcx fcy Gfcx Gfcy
results in hollow clay brick masonry shear walls [9].
These experiments are well suited for the validation of 1.87 N/mm2 7.61 N/mm2 5.0 N/mm 10.0 N/mm
the model because most of the parameters necessary
to characterise the model are available from biaxial The first wall analysed (W1) is subjected to an
tests. initial vertical load p of 0.61 N/mm2 and shows a very
Fig. 4 shows the geometry of the walls, which ductile response with tensile and shear failure along
consist of a masonry panel of 3600 × 2000 × 150 mm3 the diagonal stepped cracks [9]. The comparison
and two flanges of 150 × 2000 × 600 mm3. Additional between numerical and experimental load-
boundary conditions are given by two concrete slabs displacement diagrams, for wall W1, is given in
placed in the top and bottom of the specimen. Fig. 5. Good agreement is found. The low initial
Initially, the wall is subjected to a vertical load p vertical load combined with the confinement provided
uniformly distributed over the length of the wall. This by the flanges and the top concrete slab yields an
is followed by the application of a horizontal force F extremely ductile behaviour. The unloading found at d
on the top slab causing a horizontal displacement d. A equal to 2.0 mm is due to the mode I crack opening of
regular mesh of 24 × 15 4-noded quadrilaterals is used the left flange. The behaviour of the wall is depicted
for the panel and 2 × 15 cross diagonal patches of 3- in Fig. 6 in terms of the deformed mesh at ultimate
noded triangles are used for each flange. The analyses stage, where the centre node of the crossed diagonal
are carried out with indirect displacement control with patch of the flanges is not shown in order to obtain a