You are on page 1of 14

Mathematics http://mms.sagepub.

com/
and Mechanics of Solids

On the derivation of constitutive relations for Cosserat continuum description of structural elements
and materials with microstructure
A. Riahi and J.H. Curran
Mathematics and Mechanics of Solids 2011 16: 716 originally published online 13 April 2011
DOI: 10.1177/1081286510387862

The online version of this article can be found at:


http://mms.sagepub.com/content/16/7/716

Published by:

http://www.sagepublications.com

Additional services and information for Mathematics and Mechanics of Solids can be found at:

Email Alerts: http://mms.sagepub.com/cgi/alerts

Subscriptions: http://mms.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://mms.sagepub.com/content/16/7/716.refs.html

>> Version of Record - Sep 14, 2011

OnlineFirst Version of Record - Apr 13, 2011

What is This?

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


Article

Mathematics and Mechanics of Solids


16(7): 716–728

On the derivation of constitutive ©The Author(s) 2011


Reprints and permission:
sagepub.co.uk/journalsPermissions.nav
relations for Cosserat continuum DOI: 10.1177/1081286510387862
mms.sagepub.com
description of structural elements
and materials with microstructure

A Riahi
Rocscience Inc., Toronto, Canada

JH Curran
R.M. Smith Professor Emeritus, Department of Civil Engineering, University of Toronto, Canada

Received 3 February 2010; accepted 6 August 2010

Abstract
In this paper we discuss the mechanics of the Cosserat continuum and its relevance to the description of structural
elements and materials with microstructure. It shows how the standard beam and plate formulations can be derived
as reduced forms of the generalized Cosserat continuum. Furthermore, the Cosserat description of materials with
microstructure and the procedure to determine the constitutive equations for such materials are described.

Keywords
cosserat continuum, microstructure, plate, three-dimensional beam

1. Introduction
Generalized continuum theories are increasingly applied to engineering problems concerning materials with
microstructure. However, it seems that the limits and potential applications of such theories are not fully rec-
ognized. The major challenge in the application of Cosserat theory is defining the constitutive parameters that
describe the deformation response of a particular material or structure. We intend to show, in simple mathemat-
ics, how a generalized continuum, and in particular a Cosserat continuum, can be used to develop formulations
for structural elements, such as beams and plates. We will also discuss how the constitutive equations for more
complex behavior arising in materials with microstructure can be determined. We hope that through discussions
focused on the generality of Cosserat continuum theory and the physical relevance of constitutive parameters, in
the future Cosserat continuum theory will become a more widely used framework in computational mechanics.
The Cosserat continuum theory, proposed by the Cosserat brothers at the beginning of the last century [1],
belongs to a group of enhanced or generalized continuum theories referred to as gradient continuum theories.
However, some early attempts at more generalized descriptions of continua can be traced back to the work of
Voigt [2].

Corresponding author:
A Riahi, Rocscience Inc., 31 Balsam Avenue, Toronto, Ontario M4E 3B5, Canada
Email: azadeh.riahi@rocscience.com

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


Riahi and Curran 717

Macro volume x̂2 d3


x̂2 p Micro
volume ο Ω d2

po po d1

x̂1
x̂1
x̂3
x̂3

(a) (b)
Figure 1. (a) Macro and micro volumes and the relations between the macro and micro displacements;  represents the macro
volume, while O represents the micro volume (adapted from Pasternak E and Mühlhaus [4]). (b) A 3D Cosserat point, a point
surrounded by a finite region. The deformation of the finite region around, Po , is described by a number of director vectors, di .

In the 1960s, the theoretical aspects of gradient continuum theories were the focus of much attention [3].
The underlying idea of these theories is that for a material with microstructure, as shown in Figure 1(a), the
motion of each material point is determined by macro displacements, as well as micro displacements, which
reflect the presence of a microstructure. The displacement of a point P of the macro volume , which is located
within the micro body O, is represented by two vectors:

(1) the macro position vector, r, defines the position of the reference point, P◦ , on the micro body, o, with
respect to the origin of the coordinate system;
(2) the micro position vector, also called the director vector, d, represents the position of a point, P, with
respect to the reference point, P◦ .

Using a Taylor’s series, the micro displacement vector can be expanded in the vicinity of point P◦ [4]. This
results in the introduction of new kinematic variables in the equations of motion. The Cosserat continuum
theory, which associates three displacement and three rotational degrees of freedom with each material point,
can be considered as the simplest form of the gradient theories.
Gradient continuum theories necessitate new kinetic measures, which are work conjugates to the additional
deformation measures. Despite the mathematical elegance of generalized continuum theories, due to the lack of
physical interpretation of the constitutive parameters linking the kinetic and kinematic terms, their application
has remained limited.
New interest in the application of Cosserat theory was initiated by the work of Ericksen and Truesdell [5],
Naghdi [6, 7], Green et al. [8, 9], and Naghdi and Rubin [10] on the formulations for rods, plates, and shells
within a framework referred to as the theory of directed media. In this approach, the deformation of a three-
dimensional (3D) body is described by a displacement vector and a number of additional deformable director
vectors (see Figure 1(b)). Variants of this approach, referred to as theory of Cosserat shells (surfaces), Cosserat
rods (curves), and Cosserat points [11] have been developed for the formulation of structural elements where
the director vectors become physically interpretable. It can be argued that the theory of Cosserat points [11–13]
represents the most general form of a directed media in which a small volume that is essentially a point is
surrounded by some small but finite region. In general, in each of these approaches, the first director vector
represents the position of the point P◦ on the neutral axis of a rod, neutral surface of a shell, or centroid of a
brick structure. Depending on the geometry of the structure under consideration, a number of director vectors
are utilized to represent the geometry and deformation of the region in the vicinity of P◦ .
Application of Cosserat theory to the continuum description of materials with microstructure was initiated
in the mid 1980s. In the Cosserat description of these materials, the motion of each point is represented by a
displacement vector and a Cosserat rotation vector. The rotation components represent the rigid body rotation
of the micro volume surrounding the point. Other types of deformation at the micro level, such as stretching,
are disregarded.
The physical link between kinetic and kinematic variables in the Cosserat description of materials with
microstructures was first made by Mühlhaus [14] and Mühlhaus and Vardoulakis [15] for particulate media.

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


718 Mathematics and Mechanics of Solids 16(7)

Later, the Cosserat theory was applied to the behavior of layered and blocky materials in a two-dimensional
(2D) framework [16, 17].
In this paper, we focus on the derivation of two different types of structural elements and materials with
microstructure from the general Cosserat continuum description. We begin by showing that by imposing par-
ticular geometric assumptions, the standard formulation of 3D beams and plates can be recovered from the
general 3D Cosserat continuum theory. Subsequently, we discuss the procedure to determine the constitutive
parameters of Cosserat materials with periodic microstructure.
In Section 2, the fundamentals of the Cosserat continuum, Cosserat rotations, and measures of strain
and stress are discussed. Section 3 presents the procedure for deriving the constitutive relations for Cosserat
continua. Section 4 provides a brief conclusion to this paper.
Throughout this text, bold-faced notation represents a vector or tensor; a single sub-index refers to compo-
nents of vectors and double sub-indices denote components of second-rank tensors or matrices. x̂i (i = 1, 2, 3)
represent the Cartesian coordinate system, and (.),i is the partial derivative with respect to the Cartesian
coordinate x̂i . The discussions presented in this paper assume a small deformation framework.

2. Cosserat continuum
The Cosserat theory provides an enhanced mathematical description of the mechanics of a deformable body
by considering additional rotational degrees of freedom at each material point. The Cosserat rotation is
independent from the material rotation, defined as
1
wi = eijk uk,j , (1)
2
where w is the axial vector of the second-rank anti-symmetric material rotation tensor and e is the third-rank
Levi-Civita permutation tensor.
In the 3D framework, each material point is associated with three translational degrees of freedom, ui , and
three (Cosserat) rotational degrees of freedom, θic , about the coordinate axes.

2.1. Governing equations, micropolar stress, and micropolar couple stress


The micropolar or Cosserat theory assumes that micromoments exist at each point of the continuum.
Equilibrium of forces and moments in the Cosserat continuum is expressed by [18]
σij,i + bj = 0, (2)
mk + μkj,j + ekij σij = 0, (3)
where b is the body force, m is the body couple moment, and σ and μ are the Cosserat stress and Cosserat
couple stress (or moment stress), respectively. The stress tensor, σ , is analogous to the Cauchy stress of the
classical continuum. In addition, the stress vector or stress traction and the couple stress vector or moment
traction are defined by
tσ = σ · n and tm = μ · n, (4)
where n is the normal to the surface in the current or spatial configuration.
Figure 2 shows the stress and couple stress measures acting on the characteristic volume of a 3D Cosserat
continuum. The first subscript of the stress tensor refers to the direction of the normal vector to the surface on
which the stress acts and the second subscript refers to the direction that the stress acts. The first subscript of
the moment stress refers to the axis about which it causes rotation, while the second denotes the surface on
which it acts.
In the absence of body moment and when couple stress terms are self-equilibrated, the condition of symme-
try of the Cauchy stress and its work-conjugate strain measure is retrieved, and the Cosserat continuum reduces
to the classical continuum.

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


Riahi and Curran 719

σ 22 x̂2 μ22

σ 23 σ 21 μ12
x̂1 μ32
σ 12 x̂3 μ21
σ 32 μ23
σ 11 μ13 μ11
σ 31 σ μ31
σ 33 13
μ33

Figure 2. 3D representation of stress and couple stress measures.

2.2. Infinitesimal and finite Cosserat rotations


Cosserat rotation is defined as the rotation of a rigid triad at each material point, which rotates independently
from the material triad. The representation of micropolar rotation in its most general form is given by [19]
Rc = exp(spn(θ c )), (5)
where θ c = θic ei is the axial vector of rotation and defines the axis of rotation with rotation angle θ c = θ c .
The skew symmetric tensor associated with the axial vector is expressed by
⎛ ⎞
0 −θ3c θ2c
spnθ c = e · θ c = ⎝ θ3c 0 −θ1c ⎠ , (6)
−θ2 θ1c
c
0
where e is the permutation symbol and · represents the dot-product operation.
The mathematical definition of the rotation tensor, Rc , is
sin (θ c ) 1 − cos (θ c ) c
Rc = exp(spn(θ c )) = cos (θ c ) I + spn(θ c
) + θ ⊗ θ c. (7)
θc (θ c )2
The above formula represents the finite rotation of a point of the generalized continuum. In most cases, the
terms of Equation (7) are not closed form. One exception is when the rotation vector θ c coincides with one of
the coordinate directions.
From Equation (7) in a small rotation framework, the rotation matrix, Rc , is approximated by
⎛ ⎞
1 −θ3c θ2c
Rc ∼= I + spn(θ ) = ⎝ θ3c 1 −θ1c ⎠ . (8)
−θ2 θ1c
c
1

2.3. Micropolar strain and curvature


A comprehensive study of the micropolar theory of finite rotations and finite strains has been conducted by
Steinmann [19]. The work-conjugate strain measure to the stress, σij , can be shown to be
γij = uj,i − eijk θkc . (9)
In the Cosserat continuum, in addition to γij , the variation in the Cosserat rotations is the additional kinematic
variable, referred to as curvature. Curvature enters into the equations as the work-conjugate measure to the
micromoment tensor. Curvature, which is a third-order anti-symmetric tensor, can be reduced to a second-order
tensor as
1 
κls = elij Rcki Rckj,s . (10)
2
By substituting Rc from Equation (8) into the above expression, and disregarding any higher-order terms of
rotation, the expression for the second-order cuvature tensor in a small deformation framework becomes
⎛ ⎞ ⎛ ⎞
κ11 κ12 κ13 −θ1,1 −θ1,2 −θ1,3
κ = ⎝ κ21 κ22 κ23 ⎠ = ⎝ −θ2,1 −θ2,2 −θ2,3 ⎠ or κij = −θi,j c
. (11)
κ31 κ32 κ33 −θ3,1 −θ3,2 −θ3,3

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


720 Mathematics and Mechanics of Solids 16(7)

d1
∂P x3
x2 ∂u3 / ∂x1 − θ 2c

d2
x1
do
Cosserat rigid triad
do
(a) (b) (c)

Figure 3. (a) A Cosserat rod. (b) A beam element and the rigid triad representing Cosserat rotation. (c) Deformed geometry of a
beam in the x̂1 x̂3 plane.

In the above, the curvature measures θi,j for i = j represent a twisting mechanism, while the curvature θi,j for
i = j represents bending.

3. Constitutive equations
In the Cosserat continuum, in addition to stresses and strains, micromoments and curvatures form a new
kinetic–kinematic work-conjugate pair in the governing equations. The material constitutive relations there-
fore must define the stress–strain and micromoment–curvature relations. The constitutive parameters should
be determined based on geometrical and mechanical considerations for the particular structural element or
microstructure under consideration.
In this section we illustrate the generality of the Cosserat continuum description, and the relevance of its
mathematics to various types of mechanical behavior. The first section focuses on deriving the formulation
of beam and plate structures from the general Cosserat continuum description. The Cosserat formulation of
materials with a periodic microstructure will be subsequently discussed.

3.1. Reduced form of Cosserat continuum and the representation of structural elements
3.1.1. Three-dimensional beam A rod component is a 3D flexible body with a small cross-section relative to its
length (Figure 3(a)). There are two approaches to derive equations of a Cosserat beam. Using a direct approach
similar to that of theory of Cosserat rods [11], the deformation of a beam is characterized by the displacement
of the neutral axis and stretch and rotation of the director vectors representing the deformation of the beam
cross-section.
In this paper we focus on an alternative approach based on reduced forms of the 3D Cosserat elastic contin-
uum. The idea is to consider a general 3D Cosserat continuum and reduce the dimensions of the cross-section
to a point. Cosserat translational degrees of freedom represent the displacement of a point on the reference line.
The Cosserat rotation, θic , represents the local microstructural deformation, in this case the rotation of a rigid
cross-section of the beam. A through-the-thickness integration is performed to obtain the additional Cosserat
parameters.
For the sake of simplicity in the rest of this work, we assume that the local coordinates of the beam coincide
with the global coordinate system.
Cosserat rotations represent the rotation of the cross-section about the axes of the coordinate system,
therefore (see Figure 3(c))
θ2c = u1,3 and θ3c = −u1,2 . (12)
Since the Cosserat rotation is assumed to remain constant on each cross-section (see Figures 3(b) and (c)):

κi2 = −θi,2
c
and κi3 = −θi,3
c
. (13)

In addition,
γ22 = u2,2 = 0 and γ33 = u3,3 = 0. (14)

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


Riahi and Curran 721

Equation (13) enforces that the cross-section remains planar. In addition, substituting Equation (12) into
Equation (9) leads to
γ21 = u1,2 + θ3c = 0 and γ31 = u1,3 − θ2c = 0. (15)
Using Equation (9), the remaining strain components for a 3D beam are

γ11 = u1,1 (16)

and
γ12 = u2,1 − θ3c , γ13 = u3,1 + θ2c . (17)
Equation (16) represents the extension of the reference line (neutral axis) of the beam. The strain measures, γ12
and γ13 , expressed by Equation (17), represent the difference between the Cosserat rotation and the deformation
of the reference line, due to the transverse shear deformation of the cross-section.
Considering Equation (11), the remaining curvature components are

κ21 = −θ2,1 , κ31 = −θ3,1 , and κ11 = −θ1,1 . (18)

The constitutive coefficients are determined from the elastic properties of the material comprising the beam
and the geometry of the cross-section. For a thin beam comprised of isotropic elastic material with Young’s
modulus of elasticity, E, the axial stress due to the extension of the neutral axis is

σ11 = Eε11 . (19)

In addition, the shear stresses are


σ12 = Gγ12 , σ13 = Gγ13 , (20)
where G = E/2(1 + υ) is the shear coefficient and υ is the Poisson’s ratio.
The Cosserat couple stresses are moments per unit area μij = Mij /Aj . By considering the variation of normal
stresses due to the extension of beam fibers, the bending moment acting on the beam cross-section is obtained
as  
M21 = E x̂3 θ2,1 x̂3 dx̂3 or M21 = EI21 k21 and M31 = E x̂2 θ3,1
c c
x̂2 dx̂2 or M31 = EI31 k31 , (21)
A A
where x̂3 θ2,1c
represents the extension of a fiber located at x̂3 from the neutral axis as a result of bending about
x̂2 , and Iij (i = j) represents the second moment of area of the cross-section normal to x̂j about x̂i .
For a beam of uniform rectangular cross-section with a height, h and width, b, Equation (21) reduces to

bh3 c hb3 c
μ21 = M21 /bh with M21 = E θ2,1 and μ31 = M31 /bh with M31 = E θ . (22)
12 12 3,1
In order to determine the coefficient relating the last remaining work-conjugate pair, m11 and k11 , the problem of
pure torsion of a straight beam about its axis must be considered. In this case, θ1,1
c
represents the relative rotation
of the neighboring cross-sectional planes along x̂1 . The moment (torque) M11 can be obtained by considering
the variation of shear stresses over the cross-section to get

μ11 = M11 /A1 with M11 = GI11 θ1,1


c
, (23)

where I11 represents the polar moment of area of the beam cross-section.
Compared to the general 3D continuum shown in Figure 1, the non-zero kinetic variables acting on a
beam are σ11 , σ12 , σ13 , μ11 , μ21 , and μ31 . The relations between kinematic and kinetic variables expressed by
Equations (19) and (20) are similar to the conventional constitutive equations of a 3D continuum. Equations
(21)–(23) show the well-known coupling between geometry of the beam element and the material properties.
In general, θ2c and θ3c are independent variables which, similar to the Timoshenko beam theory, stipulate that
the cross-section plane does not remain perpendicular to the neutral axis. When θ2c = −u3,1 and θ3c = u2,1 , the
Cosserat rotation becomes equal to the material rotation, and the cross-sectional plane remains perpendicular

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


722 Mathematics and Mechanics of Solids 16(7)

to the reference axis (see Figure 3(c)). In addition, if the cross-section of the beam is homogenous, θ1c is equal
to the material rotation w1 = 1/2(u2,3 − u3,2 ).
Compared to the theory of Cosserat rods, the above derivation represents the simple case in which the angle
between the two directors remains constant and the directors do not stretch. The two Cosserat rotations in this
section then represent the rotation of the director vectors. A more general formulation can be developed by
removing the restrictive assumption of rigidity of the cross-section. Such treatments require additional degrees
of freedom associated with the micro body deformations.

3.1.2. Plates Shells are structures in which the thickness is much smaller than the other two dimensions. A plate
is a flat shell with constant thickness that can be reduced to a 2D (surface) structure.
Similar to beams, a plate formulation can be developed using a direct approach, or alternatively by integrat-
ing through the thickness, reducing the 3D Cosserat continuum to a surface. A comprehensive review of direct
and through-the-thickness approaches for deriving the constitutive parameters for Cosserat plates is presented
in Altenbach and Eremeyev [20, 21].
Assuming that the normal to the plate is along x̂3 , the 3D Cosserat continuum is reduced to a surface where
ui represent the translational displacements of points on the surface, and θic is taken to represent the free rotation
of the rigid cross-section of a plate about the x̂i and is constant on the cross-section. Therefore

θ1c = −u2,3 and θ2c = u1,3 , (24)

where θic is considered positive according to the right-hand rule.


Similar to the discussions presented in the previous section, it can be deduced that

κi3 = −θi,3
c
(25)

and
γ3i = ui,3 = 0. (26)
The remaining kinematic components for a plate structure are

γ11 = u1,1 , γ22 = u2,2 , γ12 = u2,1 − θ3c , γ21 = u1,2 + θ3c , (27)

where γ11 and γ22 represent the extension of the reference axes of the plate x̂i (i = 1, 2), and γ12 and γ21
represent in-plane shear deformation of the surface. Together the kinematic variables of Equation (27) represent
membrane mechanisms. In addition

γ13 = u3,1 + θ2c and γ23 = u3,2 − θ1c (28)

represent the transverse shear deformation of the cross-sectional planes, and

κ21 = −θ2,1
c
and κ12 = −θ1,2
c
(29)

represent bending.
For a homogeneous plate, θ3c = 1/2 (u1,2 − u2,1 ) is not an independent variable. As a result, κ31 and κ32 can
be disregarded.
The kinetic–kinematic constitutive relations for a thin plate comprised of isotropic elastic material with
Young’s modulus, E, and shear modulus, G, can be expressed as

σ11 = Eε11 , σ22 = Eε22 , σ12 = Gγ12 , σ21 = Gγ21 . (30)

The stresses due to the shear deformation of cross-sectional planes are

σ13 = Gε13 and σ23 = Gε23 . (31)

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


Riahi and Curran 723

(a) (b) (c) (d) (e)

Figure 4. Characteristic volumes for 3D Cosserat materials: (a) layered material; (b) and (c) material with beam-like microstructure;
(d) blocky material; (e) particulate material.

The relations between moments and curvatures are derived by considering the extension of material fibers due
to the corresponding curvature, similar to Equation (21), by applying a through-the-thickness integration (also
see Altenbach and Eremeyev [20]). In plates, however, the Poisson effect must be considered.
For a homogeneous, isotropic plate, the relations are
 +h/2  +h/2
E
M12 = (x̂3 θ2,1 )x̂3 dx̂3 + υ (x̂3 θ1,2 )x̂3 dx̂3 = B(κ12 + υκ21 ) and M21 = B(κ21 + υκ12 ), (32)
1 − υ 2 −h/2 −h/2

where υ is the Poisson’s ratio of the material comprising the layers, B is the bending stiffness of the plate, and
Mij is the bending moment per unit length.
The twisting couple stresses are related to their corresponding curvature measures through
 +h/2  +h/2
B(1 − υ) B(1 − υ)
M11 = G (x̂3 θ1,1 )x̂3 dx̂3 = k11 and M22 = G (x̂3 θ2,2 )x̂3 dx̂3 = k22 . (33)
−h/2 2 −h/2 2
For a rectangular cross-section with a thickness, h, B = Eh3 /12(1 − υ 2 ), the Cosserat moment stresses are
μij = Mij /h.
Compared to the general case shown in Figure 2, the non-zero kinetic variables reduce to σ11 , σ22 , σ12 , σ21 ,
σ13 , σ23 , μ11 , μ21 , μ22 , and μ12 . The formulation presented above is similar to the Mindlin–Reissner plate for-
mulation. More comprehensive formulations with enhanced constitutive parameters reflecting the effects of
the plate thickness relative to characteristic length under bending (see Eringen and Microcontinuum [22]) are
presented in Altenbach and Eremeyev [20, 21]. When θ1c = u3,2 and θ2c = −u3,1 , the Cosserat rotations become
equal to the material rotations and the plate formulation becomes identical to Kirchhoff’s plate bending theory.

3.2. Three-dimensional Cosserat continuum and the representation of a material with periodic microstructure
The difference between the classical and the Cosserat continuum is that each point of a Cosserat continuum can
describe certain geometric properties associated with its microstructure (Figure 4). The conservation of linear
and angular momentum equations (Equations (2) and (3)) describe the deformation of a characteristic volume
that encompasses a representative sample of the microstructure (see Figure 4). The size of this characteristic
volume is much smaller than the macro-volume characteristic dimension, but is much larger than the micro-
volume characteristic dimension (see Figure 1).
In a 3D volume (as opposed to a reduced form of 3D volume), all nine components of strain, expressed by
Equation (9), are preserved. The stress–strain elasticity matrix is of the following form:
⎡ ⎤
E1111 E1122 E1133 0 0 0 0 0 0
⎢ E2211 E2222 E2233 0 0 0 0 0 0 ⎥
⎢ ⎥
⎢ 3311
E E 3322 E 3333 0 0 0 0 0 0 ⎥
⎢ ⎥
⎢ 0 0 0 G2323 G2332 0 0 0 0 ⎥
⎢ ⎥
σ = Dε or σij = Dijkl εkl with Deq = ⎢ 0 0 0 G3223 G3232 0 0 0 0 ⎥,
⎢ ⎥
⎢ 0 0 0 0 0 G1313 G1331 0 0 ⎥
⎢ 0 0 ⎥
⎢ 0 0 0 0 G3113 G3131 0 ⎥
⎣ 0 0 0 0 0 0 0 G1212 G1221 ⎦
0 0 0 0 0 0 0 G2112 G2121
(34)

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


724 Mathematics and Mechanics of Solids 16(7)

The micromoment–curvature constitutive relations, however, must be based on the mechanics of the microstruc-
ture under consideration.
In the Cosserat description of materials with microstructure, the heterogeneous characteristic volume is
replaced by a homogenized equivalent continuum. The discrete material is comprised of two fundamental
mechanical components: the material comprising each microstructure and the interfaces between these struc-
tures. These components interact similarly to a number of mechanical springs. Determining the Cosserat
constitutive parameters for materials with microstructure requires: (i) a homogenization procedure; and (ii)
consideration of the mechanics of the microstructure.
Using a homogenization procedure we can determine the equivalent (effective) elastic properties of the
characteristic volume. The fundamental assumption in the homogenization technique is that the microstructure
follows a periodic pattern with a random or structured assembly. The elastic properties of the homogenized
continuum are determined based on the elastic properties of the material comprising the microstructure, the
mechanical characteristics of contacting surfaces, and other considerations, such as orientation of the interfaces
and presence of voids [4].
The interaction condition between the components of the microstructure is defined by material parameters
representing the mechanical behavior of the contacting surfaces, and is often represented by stiffness coeffi-
cients kn , ks , and kθ , representing the normal, shear, and rotational stiffnesses of the interfaces, respectively.
Two limit cases of the shear stiffness parameter ks are of particular interest.
As ks → ∞, the jump in the displacement field in the tangential direction of the interface disappears. In
this case, the material behaves as a conventional solid.
As ks → 0, no shear stress is transferred across the contacting interfaces. In this case, each component of
the microstructure behaves as a single independent structure.
In the following sections, constitutive equations for the Cosserat description of layered and beam-like
materials will be discussed.

3.2.1. Cosserat material with layered microstructure The Cosserat constitutive parameters for a medium with a layered
microstructure are derived considering the mechanical model of a stack of interacting plates. The mechanical
response of each layer is similar to that of the plate discussed in Section 3.1.2.
We assume that each layer consists of a homogenous, isotropic, elastic material. Similar to the mechanical
representation of a single plate, for materials with plate-like microstructure, θ1c and θ2c represent the rotation of
the cross-section of each layer. In addition, for layers consisting of homogeneous material, θ3c is equivalent to the
material rotation, and is disregarded as an independent degree of freedom. Finally, the two shear components,
γ23 and γ13 , and the non-zero curvatures are expressed by Equations (28) and (29).
In contrast to a single plate, in the Cosserat continuum description of layered materials, the shear stresses
developed on planes parallel to the layers are not necessarily zero. The magnitude of the shear stresses at the
interface between two layers is proportional to the relative displacements of the adjacent surfaces, Ft ∝ urelative
t .
We assume that these tangential forces are a linear function of the relative displacement with the proportionality
coefficient of ks (force/length).
In addition, each layer is comprised of a homogenous, isotropic, elastic material with shear coefficient,
G, and a thickness of h. The equivalent elasticity coefficients of the characteristic volume are determined by
inverting the compliance matrix D = C−1 . Considering that the interfaces and intact material interact in a
similar manner to a system of springs assembled in series, it can be shown that the equivalent continuum
compliance matrix coefficients are [23]

eqv eqv eqv


C1111 = C2222 = 1/E, C3333 = 1/E + 1/hkn3 , (35)

eqv eqv eqv eqv eqv


C1122 = C2211 = C1133 = C3311 = C2233 = C3322 = −υ/E,
eq eq eq eq eq eq eq eq
C3131 = C3113 = C3232 = C3223 = (1/G + 1/hks3 ), and C1212 = C1221 = C2112 = C2121 = 1/G.

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


Riahi and Curran 725

σ 33 σ 32
σ 31
σ 13
σ 12 μ12 μ21
σ 21 σ 11 μ11
σ 22 σ μ22
23

Figure 5. Relevant stress and couple stress components for a 3D Cosserat continuum with layered microstructure.

The elasticity coefficients are obtained by finding the inverse of the compliance matrix. For details of the normal
coefficients refer to Riahi and Curran [23]. The shear coefficients are
eq eq eq eq
G3131 = G3113 = G3232 = G3223 = Ghks3 /(G + hks3 ). (36)

The shear stresses that develop on planes parallel to the layer surfaces depend on the equivalent shear coefficient
of the material. They are expressed by
eq eq
σ31 = G3131 (γ13 + γ31 ) and σ32 = G3232 (γ23 + γ32 ). (37)

However, it can be argued that the stresses occurring across the layer thickness, σ13 and σ23 , consist of two parts,
one of which is in equilibrium with their conjugate shear component, σ31 and σ32 , and the other, a contribution
from the shear deformation of the cross-sections of layers, γ13 and γ23 . These stresses therefore depend on the
shear modulus of the material comprising the individual layers, G, and are expressed by [24]
eq eq
σ13 = G3131 (γ13 + γ31 ) + Gγ13 and σ23 = G3232 (γ23 + γ32 ) + Gγ23 . (38)

The difference between the conjugate stresses is equilibrated by micromoments that develop in the material due
to the presence of a plate-like microstructure.
In the limit case when the layers are non-interacting (ks → 0), the bending stiffness of the equivalent
continuum is equal to the sum of the bending stiffnesses of the individual layers defined by Equation (32),
which is naturally taken into account by integration over the area. When ks → ∞, the discontinuity in the
displacement at the interface between layers vanishes. In this case, the Cosserat rotations become equal to the
material rotations, and the asymmetrical part of the stress tensor disappears. Therefore, the bending stiffness
of the overall cross-section rather than the individual layers, controls the behavior. The second term in the
following equation represents the interaction conditions between the layers:
 
Eh2 G − Geq
B =eq
, (39)
12(1 − υ 2 ) G + Geq

where υ represent the Poisson’s ratio for the intact material. Kinetic variables of a 3D Cosserat continuum with
a layered microstructure are presented in Figure 5.

3.2.2. Cosserat material with beam-like microstructure The characteristic volume for a Cosserat material with extruded
beam microstructure is shown in Figures 4(b) and (c), where it is assumed that x̂1 coincides with the longitudinal
axis of the beams.
The equivalent properties are determined based on the geometry of the microstructure under consideration.
In this work we assume the micro beams are comprised of homogeneous, isotropic, elastic material, with a
rectangular cross-section of height, h, and width, b. We assume edge-to-edge contacts are formed between the
micro beams without any openings between them. The assembly shown in Figure 6 results in an equivalent
orthotropic material in the x̂2 x̂3 plane. The equivalent material coefficients are therefore
eq eq eq
C1111 = C1111
intact
= 1/E, C2222 = (1/E + 1/bkn2 ) and C3333 = (1/E + 1/hkn3 ), (40)

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


726 Mathematics and Mechanics of Solids 16(7)

b
(i-1,j+1) (i,j+1) (i+1,j+1) (i-1,j+1) (i+1,j+1)

a a
(i-1,j) (i,j) (i+1,j) (i-2,j) (i,j) (i+2,j)

(i-1,j-1) (i,j-1) (i+1,j-1) (i-1,j-1) (i,j-1)

(a) b (b)

ωm um
b
x3 n
2a
xm xn ωn un l

x1 x2
(d) (e)
(c)

Figure 6. In-plane geometric representation of an assembly of beams with: (a) rectangular cross-section; (b) circular cross-section;
(c) force transmission inside a blocky medium with edge-to-edge contacts without relative rotation (adapted from Sulem and Mühlhaus
[25]); (d) relative motion between two particles; (e) force transmission inside a granular medium (adapted from Zhang, et al. [27]).

where kni represents the normal stiffness on the interface surface normal to x̂i . The full compliance matrix
can be obtained similar to the approach discussed in section 3.2.2. It can be shown that the equivalent shear
coefficients are

G3131 = G3113 = (1/G + 1/hks3 )−1 and G2121 = G2112 = (1/G + 1/bks2 )−1 .
eq eq eq eq
(41)

Therefore the two shear stresses, σ31 and σ21 , are expressed by
eq eq
σ31 = G3131 (γ13 + γ31 ) and σ32 = G3232 (γ23 + γ32 ). (42)

The shear stresses acting on the beam cross-section are given by


eq eq
σ13 = G3131 (γ13 + γ31 ) + Gγ13 and σ12 = G2121 (γ12 + γ21 ) + Gγ12 . (43)

The bending stiffness takes into account the cross-section geometry and the interaction conditions at the beam
interfaces. For a beam with a rectangular cross-section,
 eq   eq 
eq Eh2 G − G3131 eq Eb2 G − G2121
B21 = eq and B31 = eq , (44)
12(1 − υ 2 ) G + G3131 12(1 − υ 2 ) G + G2121
where υ represents the effective Poisson’s ratio. In comparison with Equation (22), the bending stiffnesses
(Equation (22)) are expressed per unit area and Poisson effects are considered. These modifications reflect the
fact that these equations are written for a 3D continuum.
eq
In the case where the micro beams do not interact (e.g. ks3 → 0), then the associated I eq (e.g. I21 ) reduce
to the stiffness per unit area of the individual beams comprising the microstructure. When the micro beams
are fully bonded (ksi → ∞), the stiffness contribution due to presence of layers, expressed by Equation (44),
becomes zero.
The constitutive parameters for the x̂2 x̂3 plane (the plane normal to the beam axis) must be determined by
considering the assembly, geometry, and relative rotation of the microstructure. Figure 6 shows three possible
in-plane geometric configurations for a structured assembly of beam cross-sections, with edge-to-edge contacts
between the micro beams.
Limited attempts have been made to define the kinetic–kinematic relations of blocky materials in two
dimensions. Based on mechanical considerations, Mühlhaus [16] proposed constitutive relations for 2D blocky
materials. Sulem and Mühlhaus [25] have identified the constitutive parameters by comparing the differential
equations of the Cosserat continuum with the discrete equations obtained from an assembly of discrete blocks
(Figure 6(a)), and assuming inter-block forces acting at six points, as shown in Figure 6(c).

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


Riahi and Curran 727

Constitutive behavior is greatly affected by the packing and geometry of the particles. An early application
of the Cosserat continuum theory was in the representation of a particulate material in two dimensions (Figure
6(b)). The constitutive coefficients are deduced by considering the relative rotation and translation of the centers
of adjacent particles, and depend on the particle dimensions and inter-particle properties, namely the normal,
tangential, and rotational stiffnesses, the particle radius, and the package density [4, 14, 15, 26, 27]. Therefore

i = ui − ui + eijk (wj rk − wj rk ) and


umn m n m mc n nc
(45)
θimn = θim − θin , (46)

where umn i and θimn are the relative displacement and rotation of particles m and n with contact point l, and
nc mc
rk and rk are vectors joining the center of each particle to the contact point.
Determining the constitutive parameters for particle and block interaction in the 2D framework requires an
improved representation of: (i) the shear coefficients of the equivalent continuum approach by considering the
specific assembly of blocks; (ii) the interaction condition between blocks; and (iii) opening–closure conditions.
Finally, in extending the 2D constitutive equation to 3D, the effects of μ11,1 must also be considered.

4. Conclusion
The assumption that each point of the material can experience a rotation that is independent of the material
rotation enables the Cosserat continuum theory to describe the deformation of material points that can represent
the geometric properties of the material microstructure.
Our objective with this paper was to show the applicability of the 3D Cosserat continuum theory as a more
general framework for formulating computational mechanics problems. The major challenge with the Cosserat
approach is determining a physically meaningful link between the kinetic and kinematic variables.
The paper discussed some common techniques for determining the constitutive equations of Cosserat con-
tinua. Specifically, the Cosserat constitutive parameters relevant to the standard formulation of 3D beam and
plates were discussed. It was shown that the formulation of Timoshenko beams and Mindlin–Reissner plates
can be recovered by taking the Cosserat rotation as the rotation of the rigid cross-section of these compo-
nents. The Cosserat description of materials with microstructure was further discussed, with an emphasis on
homogenization techniques and consideration of the mechanics of the microstructures.
It is the authors’ opinion that one of the most attractive applications of Cosserat continuum theory is in
the far-field analysis of engineering-scale problems with a discrete nature. The challenge remains in defin-
ing the constitutive relations for the particular case under consideration. Other approaches, such as the theory
of directed media or comparison of the Cosserat governing equations to the discrete formulation, can pro-
vide insight in determining the constitutive parameters. Future research should therefore focus on developing
mechanically based constitutive laws for more complicated mechanics, such as those arising in materials with
general blocky structures.

Funding
The authors gratefully acknowledge the financial support provided by the Natural Sciences and Engineering Research Council of Canada
through an IRDF (A. Riahi) and Discovery grant (J. Curran).

Conflict of interest
None declared.

Dedication
Dedicated to Mike Carroll on occasion of his 75th birthday, with fond memories of discussions at the Egg Shop and handball
matches.

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013


728 Mathematics and Mechanics of Solids 16(7)

References
[1] Cosserat, E, and Cosserat, F. Théorie des corps deformable. Paris: Hermann, 1909.
[2] Voigt, W. Theoretische Studien über die Elasticitätsverhältnisse der Krystalle [Theoretical studies of the elastic behaviour of
crystals]. Abhandlungen Gesellschaft Wissenschaften Gottingen 1887; 34.
[3] Mindlin, RD. Micro-structure in linear elasticity. Arch Ration Mech Anal 1964; 16: 51–78.
[4] Pasternak, E, and Mühlhaus, HB. Generalized homogenization procedure for granular materials. J Eng Math 2005; 52: 199–229.
[5] Ericksen, JL, and Truesdell, C. Exact theory of stress and strain in rods and shells. Arch Ration Mech Anal 1958; 1: 295–323.
[6] Naghdi, PM. The theory of plates and shells. In: Truesdell C (ed.) S. Flugges Handbuch der physik, 2. Berlin: Springer, 1972,
pp.425–640.
[7] Naghdi, PM. Finite deformation of elastic rods and shells. In: Proceedings of the IUTAM Symposium Finite Elasticity, Bethlehem,
PA, 1982, pp.17–103.
[8] Green, AE, Naghdi, PM, and Wenner, ML. On the theory of rods, I. Derivation from the three-dimensional equations. Roy Soc
London 1974; A337: 451–483.
[9] Green, AE, Naghdi, PM, and Wenner, ML. On the theory of rods, II. Developments by direct approach. Roy Soc London 1974;
A337: 485–507.
[10] Naghdi, PM, and Rubin, MB. Constrained theories of rods. J Elast 1984; 14: 343–361.
[11] Rubin, MB. Cosserat theories: shells, rods and points. In: Solid mechanics and its applications, 79. The Netherlands: Kluwer,
2000.
[12] Nadler, B, and Rubin, MB. A new 3-D finite element for nonlinear elasticity using the theory of a Cosserat point. Int J Solid
Struct 2003; 40: 4585–4614.
[13] Nadler, B, and Rubin, MB. Determination of hourglass coefficients in the theory of a Cosserat point for nonlinear elastic beams.
Int J Solid Struct 2003; 40: 6163–6188.
[14] Mühlhaus, HB. Scherfugenanalyse bei Granularem Material im Rahmen der Cosserat Theorie. Ingenieur Archiv 1986; 56:
389–399.
[15] Mühlhaus, HB, and Vardoulakis, I. The thickness of shear bands in granular materials. Geotechnique 1987; 37: 271–283.
[16] Mühlhaus, HB. Continuum models for layered and blocky materials. In: Comprehensive rock mechanics. Oxford: Pergamon
Press, 1993.
[17] Mühlhaus, HB. A relative gradient method for laminated materials. In: Continuum models for materials with microstructure.
Ch. 13. Toronto: Wiley, 1995.
[18] Truesdell, C, and Toupin, R. The classical field theories. Handbuch der Physik 1960; 3: 226–793.
[19] Steinmann, P. A micropolar theory of finite deformation and finite rotation multiplicative elasto-plasticity. Int J Solid Struct 1994;
1: 1063–1084.
[20] Altenbach, H, and Eremeyev, VA. On the linear theory of micropolar plates. Z Angew Math Mech 2009; 89: 242–256.
[21] Altenbach, H, and Eremeyev, VA. On the theory of plates based on the Cosserat approach. In: Maugin, A, and Metrikine, AV.
(eds) Mechanics of Generalized Continua, Advances in Mechanics and Mathematics, 21, Ch. 3 2010 : 27-35. New York :Springer.
[22] Eringen, AC. Microcontinuum field theory. I. Foundations and solids. New York: Springer, 1999.
[23] Riahi, A, and Curran, JH. Full 3D finite element Cosserat formulation with application in layered structures. Appl Math Model
2009; 33: 3450–3464.
[24] Zvolinski, NV, and Shkhinek, KN. Continuum model for a laminar elastic medium. Solid Mech 1984; 19(1): 1–9.
[25] Sulem, J, and Mühlhaus, HB. A continuum model for periodic two-dimensional block structures. Mech Cohesive Frictional
Mater 1997; 2: 31–46.
[26] Zhang, X, Jeffery, RG, Mai, YW. A micromechanics-based Cosserat-type model for dense particulate solids. Z Angew Math Phys
2006; 57: 682–707.
[27] Chang, CS, and Ma, L. Elastic material constants for isotropic granular solids with particle rotation. Int J Solid Struct 1992; 29:
1001–1018.

Downloaded from mms.sagepub.com at BOSTON UNIV on March 2, 2013

You might also like