Anda di halaman 1dari 6

ARMA 15-311

Anisotropic Geomaterial Deformation Formulation for


the Combined Finite-Discrete Element Method in 2D
Rougier, E., Lei Z. and Knight E.E.
Los Alamos National Laboratory, Los Alamos, NM, USA
Munjiza, A.
University of London, London, UK

Copyright 2015 ARMA, American Rock Mechanics Association


This paper was prepared for presentation at the 49th US Rock Mechanics / Geomechanics Symposium held in San Francisco, CA, USA, 28 June-
1 July 2015.
This paper was selected for presentation at the symposium by an ARMA Technical Program Committee based on a technical and critical review of
the paper by a minimum of two technical reviewers. The material, as presented, does not necessarily reflect any position of ARMA, its officers, or
members. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of ARMA
is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 200 words; illustrations may not be copied. The
abstract must contain conspicuous acknowledgement of where and by whom the paper was presented.

ABSTRACT: In this paper, the Combined Finite-Discrete Element Method (FDEM) has been applied to analyze the deformation
of anisotropic geomaterials. In the most general case geomaterials are both non-homogeneous and non-isotropic. With the aim of
addressing anisotropic material problems, improved 2D FDEM formulations have been developed. These formulations feature the
unified hypo-hyper elastic approach combined with a multiplicative decomposition-based selective integration for volumetric and
shear deformation modes. This approach is significantly different from the co-rotational formulations typically encountered in
finite element codes. Unlike the co-rotational formulation, the multiplicative decomposition-based formulation naturally
decomposes deformation into translation, rotation, plastic stretches, elastic stretches, volumetric stretches, shear stretches, etc. This
approach can be implemented for a whole family of finite elements from solids to shells and membranes. This novel 2D FDEM
based material formulation was designed in such a way that the anisotropic properties of the solid can be specified in a cell by cell
basis, therefore enabling the user to seed these anisotropic properties following any type of spatial variation, for example, following
a curvilinear path. In addition, due to the selective integration, there are no problems with volumetric or shear locking with any
type of finite element employed.
FDEM software package. Through numerical examples
1. INTRODUCTION
the generalized anisotropic capabilities of the unified
In this work a unified constitutive approach for a 2D constitutive approach have been demonstrated and are
composite triangle has been developed for the Combined presented in this paper.
Finite-Discrete Element Method (FDEM) [1-3]. Since its
inception the FDEM has become a tool of choice for a
diverse field of practical engineering and scientific 2. ELEMENT GEOMETRY AND MATERIAL
simulations[4-9]. From the very initial idea of FDEM a PROPERTIES
large strain-large displacement formulation for the finite Each triangular finite element is sub-divided into four
element side of FDEM has been employed in its exact three-noded triangles, as shown in Fig. 1-a. For each of
multiplicative decomposition (as opposed to co- the three-noded triangles, separate local α and β material
rotational) formulation. In recent years this formulation axes are introduced and the geometry of the material
has been generalized through the concept of the so called base is calculated from deformation kinematics. The
Munjiza material element, which enables a pragmatic initial position of the material axes,
engineering approach to anisotropic constitutive law
formulations for both large displacements and large
strains in the context of the exact decomposition-based [α β ] = αα i βi 

β j 
(1)
format. This approach has been recently described in  j

detail in the book entitled “Large Strain Finite Element


are provided as an input.
Method: A Practical Course” by Munjiza et al. [10],
where also some novel concepts of selective integration Full integration over a finite element may lead to
have been proposed and applied to a whole family of locking, while reduced integration may lead to zero
finite elements. One of the elements proposed is the energy modes. In order to fix both of the above
composite triangle finite element in 2D. In this work, problems, selective integration is employed, different
this element has been implemented into an in-house constitutive components are evaluated at different
integration points. Stretches generated due to deviatoric ~ ~ ~
xc − ~
deformation are calculated for each sub-triangle at [~ ]
~  xa − xb
α β = ~ ~ ~
xb 
yb 
yc − ~
integration points G1 , G 2 , G 3 , and G 4 , while  ya − yb
  xa − xb b i  
−1
stretches generated due to volumetric deformation are
 xc − xb  a i
  ya − yb  
yc − yb  a j
calculated, for the whole composite triangle, at
b j 
integration point G 4 , see Fig. 1-a. In other words, four  
different infinitesimal deviatoric elements are defined where the values of the indexes a, b and c for each of the
for each sub-triangle, while a single infinitesimal volume sub-triangles of the composite triangle are shown in
element is defined for the composite triangle as a whole. Table 1.
The material formulation is implemented in such a way
Table 1. Indexes for each sub-triangle
that each composite triangle can be assigned a different
orientation for the material axis. This enables the Sub-triangle a b c
simulation of complex geologic structures, where the 1 2 1 6
material axes are not constant across the layers, see Fig. 2 3 2 4
1-b. 3 5 4 6
4 4 2 6
a) 5 The obtained four sets of base vectors [α β ] define
y four different generalized material elements. These are
G3 passed to the material package together with the
6 3 volumetric stretch (which is therefore the same for all
4 x four sub-triangles) resulting in a single sampling point
G4 for volumetric change and four sampling points for all
the other stretches.
G2
1 G1 4
2
1
2 4. MATERIAL LAW
b) 3 Very often the material law is defined in terms of a strain
energy function [11]. This approach is relatively difficult
to generalize to anisotropic materials. In contrast, the
unified constitutive approach is relatively easy to
generalize to anisotropic materials. In order to calculate
the stress due to the deformation kinematics described
above, the stretches of the volume, the edges and the
angle of the material element are calculated as described
in the following paragraphs. The volumetric stretch is
obtained as follows
v~
Fig. 1. a) The six-noded composite triangle finite element. b) sv = (3)
Graphic representation of a generic material axis orientation in v
a geologic environment.
where v is the initial volume of the generalized material
element,

3. DEFORMATION KINEMATICS α i αj


v = det  = αi β j − α j βi (4)
For each sub-triangle the previous and the current β i β j 
infinitesimal solid element is defined by
and v~ is the current volume of the generalized material
[αˆ βˆ ] =  yxˆˆ a − xˆb
− yˆ b
xˆc − xˆb 
yˆ c − yˆ b 
element
 a α~i α~ j  ~ ~ ~ ~
(2) v~ = det  ~ ~ = α i β j − α j βi (5)
  xa − xb
 xc − xb  a i
−1
b i   βi β j 
  ya − yb  
yc − yb  a j b j The linear stretches of the edges of the material element
 
are given by
~
α~ β mβ~
sα = ; sβ = (6)
α β mα~ ~ ~
~ ~ β P
β P
where the previous and the current lengths of the edges
of the material element are given by ~
~
α α
α = α = α i2 + α j2 ; β = β = βi 2 + β j2 mβ~α~
(7) ~
~ = α~ 2 + α~ 2 ; β~ = ~ ~ ~ ~ β
α~ = α β = β i2 + β j2 β
i j
~ mv ~ mα~β~
P P
The angular stretch is given by ~ ~
α α
ψ~
sψ = (8)
ψ
mβ~ mv
where the initial and the current angles of the material
element are given by
~ mα~
~
~⋅β β
 α ⋅β  α 
ψ = arcos   ; ψ~ = arcos  ~ ~  (9) mβ~α~
 αβ 
 αβ    ~ mα~β~
P
The logarithmic strains are obtained from the ~
generalized stretches as follows, α
Fig. 2. Components of the Munjiza stress tensor matrix.
ev = ln sv ; eα = ln sα
(10)
eβ = ln sβ ; eψ = ln sψ The Munjiza stress matrix does not change with
deformation (i.e., is invariant in respect to the
deformation function) and it is therefore stored within
5. INTERNAL FORCES the material package – the Cauchy stress tensor matrix is
calculated from the Munjiza stress tensor matrix; the
It is worth noting that stress is a tensorial quantity (a Cauchy matrix is not stored but is temporarily derived
tensor) that defines the state of internal forces (i.e., and used to calculate nodal forces. In this manner, any
internal forces through a given surface) at a given familiar deformation-dependent “stress updates” (such as
material point P. As such, the stress tensor represents a the Jaumann rate) are eliminated completely.
physical reality at point P and is uniquely defined, while
at the same time it can be represented by many different
stress tensor’s matrices. One of these is the Munjiza 6. NUMERICAL EXAMPLES
stress tensor matrix. The components of the Munjiza
The implementation of the unified material model is
matrix are given by
illustrated using a 2D square block with a circular
mv = M v ev ; mα = M α eα borehole in a center, as shown in Fig. 3.
(11)
mβ = M β eβ ; mψ = M ψ eψ

where M v , M α , M β , M ψ are the Munjiza elastic 1.0


constants. For a plane stress case, these constants are 20
given by y
νE E
Mv = ; M α = Mβ = M ψ = (12) x
(1 - ν )
2 (1 + ν )
A graphical representation of the Munjiza stress tensor 20
matrix is given in Fig. 2.
Fig. 3. General dimensions of the model (all dimensions in
meters).
A pressure pulse described by a Heaviside step function
was applied to the borehole at the beginning of the
simulation. The magnitude of the pressure applied was
1.0 MPa.
For the isotropic case the following material properties
were adopted:
mv = 0.0 GPa ; ma = 1.0 GPa
mβ = 1.0 GPa ; mψ = 1.0 GPa (13) a) b)

ρ = 1000.0 kg/m 3
For the anisotropic case the material properties used
were the same as the ones listed in Eq. (13) with the
following exception
mβ = 2.0 GPa (14)

Four different orientations for the material axes [α β]


c) d)
were used, as shown in Fig. 4. In all four cases the
Fig. 5. Wave propagation for the isotropic material model: a)
material axis α and β are orthogonal to each other. In horizontal, b) 45º, c) general #1, d) general #2.
Fig. 4-a the material axis α is collinear with the x-axis
across the whole model. The material axis orientation A snapshot of the wave propagation for the case with an
shown in Fig. 4-b is the same as the one shown in Fig. 4- anisotropic material is shown in Fig. 6. In this case, the
a, but rotated 45º counterclockwise. Fig. 4-c shows a shape of the wave propagation front is strongly affected
general orientation of the material axes that changes by the orientation of the material axes.
from finite element to finite element. The orientation of
the material axis shown in Fig. 4-d is the same as the one
shown in Fig. 4-c, but rotated 45º counterclockwise.

β
β α
α

a) b)
β
α α
a) b)

β α
α
β

c) d)
α α
Fig. 6. Wave propagation for the anisotropic material model:
β a) horizontal, b) 45º, c) general #1, d) general #2.
c) β
d)
Fig. 4. Orientation of the material axis α: a) horizontal, b) 45º, The results must not affected by the relative orientation
b) #2.
c) general #1, d) general of the material axis with respect to the mesh. This is
demonstrated by comparing Fig. 6-a against Fig. 6-b
A snapshot of the wave propagation for the case with an rotated 45º clockwise and Fig. 6-c against Fig. 6-d also
isotropic material is shown in Fig. 5. As expected, the rotated 45º clockwise. These comparisons are shown in
wave front preserves a circular shape, regardless of the Fig. 7 with the help of isolines. The red isolines
orientation of the material axes. correspond to the un-rotated results (Fig. 6-a and Fig. 6-
c), while the white isolines correspond to the rotated
results (Fig. 6-b and Fig. 6-d). It is evident that there is It is worth mentioning that this represents, in a sense, a
no change in the wave propagation front as the material whole family of approaches that can lead to deformation
axis are rotated. independent stress tensor matrices. One of these matrices
is the Munjiza stress tensor matrix.
In this work, it has been demonstrated that this approach
is well suited to describing anisotropic layered
geomaterials to the point that a whole tensor field of
material orientations is supplied to the input - leading to
different stress wave patterns.

REFERENCES
1. Munjiza, A. 1992. Discrete elements in transient
dynamics of fractured media. PhD thesis. Civ. Eng.
Dept. Swansea.
2. Munjiza, A. 2004. The Combined Finite-Discrete
Element Method, 1st ed. London; John Wiley and
Sons Ltd.
3. Munjiza A., Knight E.E., and Rougier E. 2012.
a) Computational Mechanics of Discontinua, 1st ed.
London, John Wiley and Sons.
4. Elmo, D. and Stead, D. 2010. An integrated
numerical modeling-discrete fracture network
approach applied to the characterization of rock
mass strength of naturally fractured pillars, Rock
Mech. Rock Eng. 43:3-19.
5. Lisjak, A., and Grasselli, G. 2010. Rock impact
modeling using FEM/DEM. In Discrete Element
Methods: 5th Int. Conference on Discrete Element
Methods, London - UK, 25 – 26 August 2010, eds.
A. Munjiza, 269-274.
6. Mahabadi, O.K., Grasselli, G. and Munjiza, A. 2009.
Numerical modelling of a Brazilian Disc test of
layered rocks using the combined finite-discrete
element method. In RockEng09: 3rd Canada-US
rock mechanics symposium, Toronto - Canada, 9 –
b) 15 May 2009, eds. C.S. Diederichs and Grasselli,
87–88.
Fig. 7. Wave propagation for the anisotropic material model: 7. Rougier E., Knight E.E., Broome, S.T., Sussman, A.,
a) horizontal, b) general. and Munjiza, A. 2014. Validation of a Three-
Dimensional Finite-Discrete Element Method using
Experimental Results of the Split Hopkinson
7. CONCLUSIONS Pressure Bar Test. Int. J. Rock Mech. Min. Sci. 70,
101–108.
Recently published generalized deformation kinematics
(Munjiza et al. [10]) opens a whole range of possibilities 8. Morris, J.P., Rubin, M.B., Block, G.I., and Bonner,
for developing material laws in the context of large- M.P. 2006. Simulations of Fracture and
strain large-displacements formulation for solid Fragmentation of Geologic Materials using
deformation - instead of using traditional approaches to Combined FEM/DEM Analysis. Int J Impact Eng.
defining material laws such as energy functions, a 33, 463-473.
pragmatic engineering approach for calculating stress
9. Morris, J.P., Johnson, S. 2009. Dynamic simulations
from stretches has been described in detail in [10].
of geologic materials using combined
FEM/DEM/SPH analysis. Geomech. Geoeng.
4(1):91-101.
10. Munjiza A., Rougier E., and Knight E.E. 2015.
Large Strain Finite Element Method: A Practical
Course, 1st ed. London. John Wiley and Sons.
11. Gurtin, M.E. 1981. An Introduction to Continuum
Mechanics. Acad. Press.

View publication stats