Upload
independent
View
1
Download
0
Embed Size (px)
Citation preview
Numerical modeling of bending of micropolar plates
Roman Kvasov, Lev Steinberg n
University of Puerto Rico at Mayagüez, Department of Mathematical Sciences, Call Box 9000, Mayagüez, Puerto Rico 00681-9018, USA
a r t i c l e i n f o
Article history:
Received 9 January 2013
Received in revised form
7 March 2013
Accepted 2 April 2013Available online 20 May 2013
Keywords:
Cosserat elasticity
Micropolar plates
Finite Element modeling
a b s t r a c t
In this paper we present the Finite Element modeling of the bending of micropolar elastic plates. Based
on our recently published enhanced mathematical model for Cosserat plate bending, we present the
micropolar plate field equations as an elliptic system of nine differential equations in terms of
the kinematic variables. The system includes an optimal value of the splitting parameter, which is the
minimizer of the micropolar plate stress energy. We present the efficient algorithm for the estimation of
the optimal value of this parameter and discuss the approximations of stress and couple stress
components. The numerical algorithm also includes the method for finding the unique solution of the
micropolar plate field equations corresponding to the optimal value of the splitting parameter.
We provide the results of the numerical modeling for the plates made of polyurethane foam used in
structural insulated panels. The comparison of the numerical values of the vertical deflection for the
square plate made of dense polyurethane foam with the analytical solution of the three-dimensional
micropolar elasticity confirms the high order of approximation of the three-dimensional (exact) solution.
The size effect of micropolar plate theory predicts that plates made of smaller thickness will be more
rigid than would be expected on the basis of the Reissner plate theory. We present the numerical results
for plates of different shapes, including shapes with rectangular holes, under different loads.
& 2013 Elsevier Ltd. All rights reserved.
1. Introduction
In the classical theory of elasticity the material particles of a
continuum body are assumed to have three degrees of freedom,
which are responsible for their macrodisplacement within the
body. The size effects of the particles and their mutual rotational
interactions are ignored. The surface loads are assumed to be
completely determined by the force vector, which results in the
symmetry of the stress tensor. Classical theory of elasticity is
widely used in engineering and is usually applied to the linear
elastic materials under small deformations (stainless steel, alumi-
nium, plastic, etc.). However, many modern materials possess
certain microstructure (cellular solids, pores, macromolecules,
fibers, grains, voids, etc.) and exhibit experimental behavior that
cannot be adequately described by the classical elasticity.
For example the influence of size effect on the deformation
reported in [21] can be only predicted by micropolar elasticity.
Micropolar theory of elasticity describes the influence of the
microstructure on the deformation of the material. The mate-
rial particles are allowed to have three additional degrees of
freedom, which represent their microrotations. The surface loads
are described by the force and moment vectors, which leads to
asymmetry of the stress tensor and the introduction of the couple
stress tensor. Such important engineering materials as chopped
fiber composites, platelet composites, particulate composites,
sandwich and grid structures, trusses and honeycombs are just a
few examples of the materials that exhibit nonclassical behavior.
Rocks, soils, concrete with sand, ferroelectric and phononic crys-
tals, polyfoams, human bones and some biological tissues are also
considered to be micropolar materials [24,1,18,5,19,2,27,21,17].
The development of the micropolar elasticity can be traced
back to 1887 when Voigt made an assumption that the surface
loads should be described by both force and moment vectors, thus
introducing the concept of couple stress. In 1909 Cosserat brothers
presented the equations of local balance of momenta for stress
and couple stress, and the expressions for surface tractions and
couples [7]. Eringen developed micromorphic and micropolar
theories of solids, fluids, memory-dependent media, microstretch
solids and fluids and solved several problems in these fields [11].
Eringen also introduced a theory of plates in the framework of
the linear micropolar elasticity [13]. The theory was based on a
technique of integration of the micropolar elasticity and assumes
no transverse variation of micropolar rotations. More remarks on
the history of the modeling of classic linear elastic plates can be
found in [23,30,31].
The micropolar plate theory based on the Reissner plate theory
was first proposed in [32] and in the final form has been
developed in [33]. The preliminary computations showed that
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/tws
Thin-Walled Structures
0263-8231/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tws.2013.04.001
n Corresponding author. Tel.: +1 7878324040; fax: +1 7872655454.
E-mail addresses: [email protected], [email protected] (L. Steinberg).
Thin-Walled Structures 69 (2013) 67–78
the precision of the micropolar plate theory [33] is compatible
with the precision of the Reissner elastic plate theory [10].
In this paper we present the Finite Element analysis of
micropolar plate bending. In Section 2 we make an overview of
the micropolar plate theory, its assumptions, expression for
micropolar plate stress energy and the bending system of
equations. We also present the field equations in terms of the
kinematic variables and derive the explicit expression for
the optimal value of the splitting parameter, which minimizes
the micropolar plate stress energy. In Section 3 we provide the
numerical analysis of the bending of micropolar plates. We
describe the Finite Element method used to solve the micropolar
field equations and discuss the approximations of stress and
couple stress components. We also provide the efficient algo-
rithm for the optimal value of the splitting parameter. The
comparison of the displacement for the square plate made of
dense polyurethane foam with the analytical solution of the
three-dimensional micropolar elasticity presented in [33] shows
the high order of approximation of the three-dimensional (exact)
solution. We also provide the Finite Element analysis of the plates
of different construction shapes and plates with rectangular
holes, under different loads.
2. Enhanced model for Cosserat plate bending
In this section we will reproduce the main micropolar plate
equations and the general results of the micropolar plate theory
presented in [33]. We will develop the micropolar plate field
equations in terms of the kinematic variables and discuss the
ellipticity of the obtained system. We will obtain the explicit
expression for the optimal value of the splitting parameter for
plates of arbitrary shape.
2.1. Micropolar plate equations
The equations of balance of momentum and moment of mome-
ntum of the linear theory of micropolar elasticity are respectively
given by
sji;j þ ρðf i−∂2ui=∂t
2Þ ¼ 0;
εijksjk þ μji;j þ ρðli−j∂2φi=∂t
2Þ ¼ 0;
where sij is the stress tensor, μij is the couple stress tensor, εijk is
the Levi-Civita symbol, ρ is the mass density, f is the body force, uiis the displacement vector, li is the body couple, j is the micro-
inertia and φi is the microrotation vector. In this theory ρ and j are
assumed to be constant [13,12].
For static case the equilibrium equations of the linear micro-
polar theory without body forces and body moments have the
following form:
sji;j ¼ 0; ð1Þ
εijksjk þ μji;j ¼ 0: ð2Þ
The constitutive equations of the linear theory are given in the
form [26]:
sij ¼ ðμþ αÞγij þ ðμ−αÞγji þ λδijγkk; ð3Þ
μij ¼ ðγ þ ϵÞχ ij þ ðγ−ϵÞχji þ βδijχkk; ð4Þ
where γij and χij are the micropolar strain and torsion tensors
respectively, μ and λ are the Lamé parameters, α, β, γ and ϵ are the
asymmetric elasticity constants.
The strain–displacement and torsion–rotation relations are
given as in [12]:
γij ¼ uj;i−εijkφk; ð5Þ
χij ¼ φj;i: ð6Þ
Let us consider the body B0 to be a plate of thickness
h and x3 ¼ 0 containing its middle plane P0. The sets T and B
are the top and bottom surfaces contained in the planes x3 ¼ h=2
and x3 ¼ −h=2 respectively and the curve Γ is the boundary
of the middle plane of the plate. We consider the vertical
load and the boundary conditions at the top and bottom of the
plate:
s33ðx1; x2;h=2Þ ¼ stðx1; x2Þ; s33ðx1; x2;−h=2Þ ¼ s
bðx1; x2Þ; ð7Þ
s3βðx1; x2;7h=2Þ ¼ 0; ð8Þ
μ33ðx1; x2;h=2Þ ¼ μtðx1; x2Þ; μ33ðx1; x2;−h=2Þ ¼ μbðx1; x2Þ; ð9Þ
μ3βðx1; x2;7h=2Þ ¼ 0; ð10Þ
where ðx1; x2Þ∈P0:
We briefly review our approach presented in [33]. As it is
assumed in the standard theory of plates, we use the following
approximation for the stress components:
sαβ ¼h
2ζmαβðx1; x2Þ; ð11Þ
where ζ¼ ð2=hÞx3; and α; β∈f1;2g. Based on (11) and by means of
the first two equations of stress equilibrium (1) written in the
component form
sjβ;j ¼ 0
we obtain for the shear stress components
s3β ¼ qβðx1; x2Þð1−ζ2Þ; ð12Þ
We use expression for the stress components [33]:
sβ3 ¼ qn
βðx1; x2Þð1−ζ2Þ þ q̂βðx1; x2Þ: ð13Þ
Substituting Eq. (13) in the remaining equilibrium differential
equation for stress
sj3;j ¼ 0
we obtain the expression for the transverse normal stress
s33 ¼ ζ 13ζ
2−1
! "
knðx1; x2Þ þ ζl
nðx1; x2Þ þmnðx1; x2Þ: ð14Þ
The next step is to accommodate approximations (14) to the
boundary (7). By direct substitution to (7) it easy to obtain that
s33 ¼−34 ð
13ζ
3−ζÞp1ðx1; x2Þ þ ζp2ðx1; x2Þ þ s0ðx1; x2Þ; ð15Þ
where
p1ðx1; x2Þ þ 2p2ðx1; x2Þ ¼ pðx1; x2Þ
We consider the parametric solution of the last equation in the
form:
p1ðx1; x2Þ ¼ ηpðx1; x2Þ;
p2ðx1; x2Þ ¼ð1−ηÞ
2pðx1; x2Þ
and η∈½0;1( is a parameter, which we call the splitting parameter.
This allows us to split the bending pressure on the plate pðx1; x2Þ
into two parts corresponding to different orders of stress approx-
imations. The optimal value of the splitting parameter shows the
contribution of different types of approximation in the plate
bending. This approach gives a more accurate description of the
mechanical phenomenon of bending.
R. Kvasov, L. Steinberg / Thin-Walled Structures 69 (2013) 67–7868
Note that for
pðx1; x2Þ ¼ stðx1; x2Þ−s
bðx1; x2Þ
and
s0ðx1; x2Þ ¼12ðs
tðx1; x2Þ þ sbðx1; x2ÞÞ
the expression (15) satisfies the boundary condition requirements.
Note that in the case of η¼ 1 expression (15) is identical to the
expression of s33 given in [32].
We use the following approximation for the couple stress
components [33]:
μαβ ¼ ð1−ζ2Þrαβðx1; x2Þ þ rnαβðx1; x2Þ: ð16Þ
and couple stress
μβ3 ¼ ζsnβðx1; x2Þ: ð17Þ
Note that the first two equations of (2) can be written in the
form
ϵβjksjk þ μjβ;j ¼ 0; ð18Þ
and substituting the couple stress (16) into (18) and taking into
account (12) and (13) we obtain the expression for the transverse
shear couple stress
μ3β ¼ ð13ζ3−ζÞsβðx1; x2Þ: ð19Þ
Substituting (19) into boundary (8) we obtain that
sβðx1; x2Þ ¼ 0;
i.e. the transverse shear couple stress [32]
μ3β ¼ 0: ð20Þ
We use the assumption from [32], i.e. μ33 is a first order
polynomial
μ33 ¼ ζbnðx1; x2Þ þ cnðx1; x2Þ: ð21Þ
which satisfies the remaining differential equation of the equili-
brium of angular momentum (2)
ϵ3jksjk þ μj3;j ¼ 0: ð22Þ
This assumption is also consistent with the equilibrium equa-
tion (22) and allows us to proceed as we did for the determination
of transverse loading stress (15) from the stress boundary condi-
tions. The boundary conditions (9) are sufficient to determine μ33,
which must be of the form [32]
μ33 ¼ ζvþ t; ð23Þ
where the function vðx1; x2Þ and tðx1; x2Þ are given as
vðx1; x2Þ ¼12ðμ
tðx1; x2Þ−μbðx1; x2ÞÞ;
tðx1; x2Þ ¼12ðμ
tðx1; x2Þ þ μbðx1; x2ÞÞ:
The choice of kinematic assumptions is based on simplicity and
their compatibility with the constitutive relationships of stress and
couple stress assumptions are the following [33]:
uα ¼h
2ζΨ αðx1; x2Þ; ð24Þ
u3 ¼wðx1; x2Þ þ ð1−ζ2Þwnðx1; x2Þ: ð25Þ
The terms Ψ αðx1; x2Þ in (24) represent the rotations in the
middle plane.
We also use the microrotation φα in the following form [33]:
φα ¼5
4Ω0
αðx1; x2Þð1−ζ2Þ þ Ω̂αðx1; x2Þ; ð26Þ
φ3 ¼5h
8ζ 1−
1
3ζ2
# $
Ω3ðx1; x2Þ: ð27Þ
The constitutive formulas motivate us to chose the forms (26)
and (27), which produce expressions for φα;β and φ3;3 similar to
what we have for couple stress approximations (16).
Thus, we summarize the resulting assumptions for stress sij
and couple stress μij components across the thickness [33]
sαβ ¼h
2ζmαβðx1; x2Þ; ð28Þ
s3β ¼ qβðx1; x2Þð1−ζ2Þ; ð29Þ
sβ3 ¼ qn
βðx1; x2Þð1−ζ2Þ þ q̂βðx1; x2Þ; ð30Þ
s33 ¼−3
4
1
3ζ3−ζ
# $
p1ðx1; x2Þ þ ζp2ðx1; x2Þ þ s0ðx1; x2Þ; ð31Þ
μαβ ¼ ð1−ζ2Þrαβðx1; x2Þ þ rnαβðx1; x2Þ; ð32Þ
μβ3 ¼h
2ζsnβðx1; x2Þ; ð33Þ
μ3β ¼1
3ζ3−ζ
# $
sβðx1; x2Þ; ð34Þ
μ3β ¼ 0; ð35Þ
μ33 ¼ ζvþ t; ð36Þ
where
pðx1; x2Þ ¼ stðx1; x2Þ−s
bðx1; x2Þ;
s0ðx1; x2Þ ¼1
2ðstðx1; x2Þ þ s
bðx1; x2ÞÞ;
p1ðx1; x2Þ ¼ ηpðx1; x2Þ;
p2ðx1; x2Þ ¼ð1−ηÞ
2pðx1; x2Þ;
vðx1; x2Þ ¼1
2ðμtðx1; x2Þ−μ
bðx1; x2ÞÞ;
tðx1; x2Þ ¼1
2ðμtðx1; x2Þ þ μbðx1; x2ÞÞ:
The kinematic assumptions are given as
uα ¼h
2ζΨ αðx1; x2Þ; ð37Þ
u3 ¼Wðx1; x2Þ þ ð1−ζ2ÞWnðx1; x2Þ; ð38Þ
φα ¼5
4Ω0
αðx1; x2Þð1−ζ2Þ þ Ω̂αðx1; x2Þ; ð39Þ
φ3 ¼5h
8ζ 1−
1
3ζ2
# $
Ω3ðx1; x2Þ: ð40Þ
We also introduce the set of kinematic variables defined as
UðηÞ ¼ ½Ψ α;W ;Ω3;Ω
0α ;W
n; Ω̂α(T ;
and the micropolar plate stress set
SðηÞ ¼ ½Mαβ ;Qα;Q
n
α; Q̂ α;Rαβ;Rn
αβ; Sn
β (;
where
Mαβ ¼h
2
# $2 Z 1
−1ζ3sαβ dζ3 ¼
h3
12mαβ ; ð41Þ
Qα ¼h
2
Z 1
−1s3α dζ3 ¼
2h
3qα; ð42Þ
Qn
α ¼h
2
Z 1
−1qn
αð1−ζ2Þ dζ3 ¼
2h
3qn
α; ð43Þ
R. Kvasov, L. Steinberg / Thin-Walled Structures 69 (2013) 67–78 69
Q̂ α ¼h
2
Z 1
−1q̂αð1−ζ
2Þ dζ3 ¼2h
3q̂α ð44Þ
Rαβ ¼h
2
Z 1
−1rαβð1−ζ
2Þ dζ3 ¼2h
3rαβ ; ð45Þ
Rn
αβ ¼h
2
Z 1
−1rnαβð1−ζ
2Þ dζ3 ¼2h
3rnαβ ; ð46Þ
Snα ¼h
2
# $2 Z 1
−1ζ3μα3 dζ3 ¼
h3
12snα; ð47Þ
here M11 and M22 are the bending moments, M12 and M21 the
twisting moments, Qα the shear forces, Qn
α; and Q̂ α the transverse
shear forces, R11;R22;Rn
11;Rn
22 the micropolar bending moments,
R12;R21;Rn
12;Rn
21 the micropolar twisting moments, Snα the micro-
polar couple moments, all defined per unit length. [33]
The components of the micropolar plate strain set
EðηÞ ¼ ½eαβ;ωβ;ω
n
a; ω̂α; τ3α; ταβ; τn
αβ(
are defined as in [33]
eαβ ¼3
h
Z 1
−1ζ3γαβ dζ3; ð48Þ
ωα ¼3
4
Z 1
−1γ3αð1−ζ
2Þ dζ3; ð49Þ
ωn
α ¼3
4
Z 1
−1γα3ð1−ζ
2Þ dζ3; ð50Þ
ω̂α ¼3
4
Z 1
−1γα3 dζ3; ð51Þ
τ3α ¼3
h
Z 1
−1ζ3χ3α dζ3; ð52Þ
ταβ ¼3
4
Z 1
−1χαβð1−ζ
2Þ dζ3; ð53Þ
τnαβ ¼3
4
Z 1
−1χαβ dζ3: ð54Þ
The density of the work done by the stress and couple stress
over the micropolar strain field defined as
WðηÞ ¼
h
2
Z 1
−1ðr ) γ þ μ ) χ Þ dζ3 ð55Þ
is equal to the micropolar plate stress energy density [33].
The expression for the density of the work is obtained by
substituting the stress and couple stress assumptions (28)–(36)
and performing the integration along the thickness of the
plate [33]
WðηÞ ¼ S
ðηÞ ) EðηÞ ¼Mαβeαβ þ Qαωα
þQn
αωn
α þ Q̂ αω̂α þ Rαβταβ þ Rn
αβτn
αβ þ Snατ3α: ð56Þ
The components of the micropolar plate strain set EðηÞ are
represented in terms of the components of the set of kinematic
variables UðηÞ by the following micropolar plate strain–displace-
ment relations1:
eαβ ¼Ψ β;α−ε3αβΩ3; ð57Þ
ωα ¼ Ψ α−ε3αβðΩ0β þ Ω̂βÞ; ð58Þ
ωn
α ¼W ;α þ4
5Wn
;α þ ε3αβðΩ0β þ Ω̂βÞ; ð59Þ
ω̂α ¼3
2W ;α þWn
;α þ ε3αβð5
4Ω0
β þ3
2Ω̂βÞ; ð60Þ
τ3α ¼Ω3;α; ð61Þ
ταβ ¼Ω0β;α þ Ω̂β;α; ð62Þ
τnαβ ¼5
4Ω0
β;α þ3
2Ω̂β;α; ð63Þ
The equilibrium and constitutive equations can be obtained in
accordance with the Generalized Hellinger–Prange–Reissner (HPR)
principle for the micropolar plate, which requires
δΘðs; ηÞ ¼ 0;
where Θ is a functional on A defined by
Θðs; ηÞ ¼ USK ðηÞ−
Z
P0
ðS ) E−p )W þ vΩ03Þ da
þ
Z
Γs
So ) ðU−UoÞ dsþ
Z
Γu
Sn ) U ds; ð64Þ
A is the set of all admissible states that satisfy the micropolar plate
strain–displacement and torsion–rotation relations (5) and (6),
p¼ ðp̂1; p̂2Þ, W ¼ ðW ;WnÞ and for every s¼ ½U ; E;S(∈A.
The optimization is equivalent to [33]
(A) The equilibrium system of equations for the plate bending
Mαβ;α−Qβ ¼ 0; ð65Þ
Qn
α;α þ p̂1 ¼ 0; ð66Þ
Rαβ;α þ ε3βγðQn
γ−Q γÞ ¼ 0; ð67Þ
ε3βγMβγ þ Snα;α ¼ 0; ð68Þ
Q̂ α;α þ p̂2 ¼ 0; ð69Þ
Rn
αβ;α þ ε3βγQ̂ γ ¼ 0; ð70Þ
where p̂1 ¼ ηp, p̂2 ¼23 ð1−ηÞp; and η is the splitting parameter;
(B) The zero variation of the stress energy with respect to the
splitting parameter
δUSK ðηÞ ¼ 0;
(C) The constitutive formulas are given in the following reverse
form2:
Mαα ¼h3μðλþ μÞ
3ðλþ 2μÞΨ α;α þ
λμh3
6ðλþ 2μÞΨ β;β þ
ð3p1 þ 5p2Þλh2
30ðλþ 2μÞ; ð71Þ
Mβα ¼ðμ−αÞh
3
12Ψ α;β þ
h3ðα þ μÞ
12Ψ β;α þ ð−1Þβ
αh3
6Ω3; ð72Þ
Rβα ¼5ðγ−ϵÞh
6Ω0
β;α þ5hðγ þ ϵÞ
6Ω0
α;β; ð73Þ
Rαα ¼10hγðβ þ γÞ
3ðβ þ 2γÞΩ0
α;α þ5hβγ
3ðβ þ 2γÞΩ0
β;β; ð74Þ
Rn
βα ¼2ðγ−ϵÞh
3Ω̂β;α þ
2ðγ þ ϵÞh
3Ω̂α;β; ð75Þ
Rn
αα ¼8γðγ þ βÞh
3ðβ þ 2γÞΩ̂α;α þ
4γβh
3ðβ þ 2γÞΩ̂β;β; ð76Þ
1 Eqs. (57) and (58) are the correct versions of the equations developed in [33]. 2 In the following formulas a subindex β¼ 1 iff α¼ 2 and β¼ 2 iff α¼ 1.
R. Kvasov, L. Steinberg / Thin-Walled Structures 69 (2013) 67–7870
Qα ¼5hðαþ μÞ
6Ψ α þ
5ðμ−αÞh
6W ;α þ
2ðμ−αÞh
3Wn
;α ð77Þ
þð−1Þβ5hα
3ðΩ0
β þ Ω̂βÞ; ð78Þ
Qn
α ¼5ðμ−αÞh
6Ψ α þ
5ðμ−αÞ2h
6ðμþ αÞW ;α þ
2ðμþ αÞh
3Wn
;α
þð−1Þα5hα
3ðΩ0
β þðμ−αÞ
ðμþ αÞΩ̂βÞ; ð79Þ
Q̂ α ¼8αμh
3ðμþ αÞW ;α þ ð−1Þα
8αμh
3ðμþ αÞΩ̂β; ð80Þ
Snα ¼5γϵh3
3ðγ þ ϵÞΩ3;α: ð81Þ
2.2. Micropolar plate field equations
In order to obtain the micropolar plate bending field equations
in terms of the kinematic variables, we substitute the constitutive
formulas in the reverse form (71)–(80) into the bending system of
Eqs. (65)–(70). The micropolar plate bending field equations can be
written in the following form:
LU ¼F ðηÞ ð82Þ
where
L¼
L11 L12 L13 L14 0 L16 kL13 0 L16
L12 L22 L23 L24 L16 0 kL23 L16 0
−L13 −L23 L33 0 L35 L36 L77 L38 L39
L41 L42 0 L44 0 0 0 0 0
0 −L16 −L38 0 L55 L56 −kL35 L58 0
L16 0 −L39 0 L56 L66 −kL36 0 L58
−L13 −L14 L73 0 L35 L36 L77 L78 L79
0 −L16 −L78 0 L85 L56 −kL35 L88 kL56
L16 0 −L79 0 L56 L55 −kL36 kL56 L99
2
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
4
3
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
5
;
U ¼ Ψ1; Ψ 2; W ; Ω3; Ω01; Ω0
2; Wn; Ω01; Ω0
2 (T;
h
F ðηÞ ¼ −3h2
λð3p1;1 þ 5p2;1Þ
3ðλþ 2μÞ; −
3h2λð3p1;2 þ 5p2;2Þ
30ðλþ 2μÞ; −p1; 0; 0; 0;
h2ð3p1 þ 4p2Þ
24; 0; 0
" #T
:
The operators Lij are defined as follows:
L11 ¼ c1∂2
∂x21þ c2
∂2
∂x22−c3; L12 ¼ ðc1−c2Þ
∂2
∂x1x2;
L13 ¼ c11∂
∂x1; L14 ¼ c12
∂
∂x2;
L16 ¼ c13; L17 ¼ kc11∂
∂x1; L22 ¼ c2
∂2
∂x21þ c1
∂2
∂x22−c3;
L23 ¼ c11∂
∂x2; L24 ¼ −c12
∂
∂x1;
L33 ¼ c4∂2
∂x21þ
∂2
∂x22
!
; L35 ¼−c13∂
∂x2; L36 ¼ c13
∂
∂x1;
L38 ¼−c10∂
∂x2; L39 ¼ c10
∂
∂x1;
L41 ¼−c12∂
∂x2; L42 ¼ c12
∂
∂x1; L44 ¼ c6
∂2
∂x21þ
∂2
∂x22
!
−2c12;
L55 ¼ c7∂2
∂x21þ c8
∂2
∂x22−2c13; L56 ¼ ðc7−c8Þ
∂2
∂x1x2; L58 ¼−c9;
L66 ¼ c8∂2
∂x21þ c7
∂2
∂x22−2c13; L73 ¼ c5
∂2
∂x21þ
∂2
∂x22
!
;
L77 ¼ c3∂2
∂x21þ
∂2
∂x22
!
;
L78 ¼ −c14∂
∂x2; L79 ¼ c14
∂
∂x1; L85 ¼ c7
∂2
∂x21þ c8
∂2
∂x22−2c13;
L88 ¼ kc7∂2
∂x21þ kc8
∂2
∂x22−c15; L99 ¼ kc8
∂2
∂x21þ kc7
∂2
∂x22−c15;
where
k¼4
5; c1 ¼
h3μðλþ μÞ
3ðλþ 2μÞ; c2 ¼
h3ðαþ μÞ
12;
c3 ¼5hðαþ μÞ
6; c4 ¼
5hðα−μÞ2
6ðα þ μÞ;
c5 ¼hð5α2 þ 6αμþ 5μ2Þ
6ðα þ μÞ; c6 ¼
h3γϵ
3ðγ þ ϵÞ;
c7 ¼10hγðβ þ γÞ
3ðβ þ 2γÞ; c8 ¼
5h γ þ ϵð Þ
6;
c9 ¼10hα2
3ðαþ μÞ; c10 ¼
5hαðα−μÞ
3ðα þ μÞ; c11 ¼
5hðα−μÞ
6; c12 ¼
h3α
6;
c13 ¼5hα
3; c14 ¼
hαð5αþ 3μÞ
3ðα þ μÞ; c15 ¼
2hαð5αþ 4μÞ
3ðα þ μÞ:
The parametric system (82) is an order two elliptic system of
nine partial differential equations, where L is a linear differential
operator acting on the set of kinematic variables U . The ellipticity
of the system (82) can be easily verified by calculating the
determinant of the principle symbol LðξÞ
det LðξÞ ¼256αμ
375ðαþ μÞc1c2c3c6c
27ðξ
21 þ ξ22Þ
9;
where ξ¼ ðξ1; ξ2Þ is an arbitrary vector from R2. For nonzero elastic
constants the determinant of LðξÞ is nonzero if and only if ξ≠0, and
thus the system of equations (82) is elliptic. The ellipticity
condition for the system suggests the invertibility of the principal
part of its symbol LðξÞ for all non-zero values of ξ [9].
The right-hand side, and therefore the solution U are the
functions of the splitting parameter η. The optimal value of the
parameter η corresponds to the minimum of elastic energy over
the micropolar strain field (55).
The system (82) should be complemented with the boundary
conditions. We describe the case of the hard simply supported
boundary conditions in the numerical analysis section.
2.3. Optimal value of the splitting parameter
As we have mentioned earlier, the equilibrium systems of
partial differential equations correspond to a state of the system
(82) where the minimum of the energy is reached. The optimiza-
tion of the splitting parameter appears as a result of the
Generalized Hellinger–Prange–Reissner (HPR) principle for the
micropolar (64). The bending system of Eqs. (65)–(70) depends
on the splitting parameter and therefore its solution is parametric.
The minimization procedure for the elastic energy allows us to
find the optimal value of this parameter, which corresponds to the
unique solution of the bending problem.
Taking advantage of the linearity of the system (82) we
consider two cases:
LU0 ¼F ð0Þ; ð83Þ
LU1 ¼F ð1Þ; ð84Þ
where U0 is a solution of the bending system of Eqs. (82) when
η¼ 0, and U1 when η¼ 1.
R. Kvasov, L. Steinberg / Thin-Walled Structures 69 (2013) 67–78 71
Let us define Uη as a linear combination of U0 and U1
Uη ¼ ð1−ηÞU0 þ ηU1: ð85Þ
Notice that since p1 ¼ ηp, p2 ¼1−η2
1 2
p, F ðηÞ ¼ ð1−ηÞF ð0Þ þ ηF ð1Þ
and therefore due to the linearity of system (82)
LUη ¼ Lðð1−ηÞU0 þ ηU1Þ ¼ ð1−ηÞLU0 þ ηLU1
¼ ð1−ηÞF ð0Þ þ ηF ð1Þ ¼F ðηÞ:
The set of kinematic variables Uη defined as (85) is a solution of
the micropolar plate bending system of Eqs. (82).
The optimal value of the splitting parameter η is a minimizer of
the density of the work WðηÞ done by the stress and couple stress
over the micropolar strain field. Let us obtain an explicit expres-
sion for the optimal value of the parameter η.
Let us represent each component of the stress set SðηÞ and strain
set EðηÞ as a linear combinations of the stress sets Sð0Þ and Sð1Þ, and
the strain sets Eð0Þ and E
ð1Þ respectively
SðηÞ ¼ ð1−ηÞSð0Þ þ ηSð1Þ
and
EðηÞ ¼ ð1−ηÞEð0Þ þ ηEð1Þ:
Therefore
WðηÞ ¼ S
ðηÞ ) EðηÞ ¼ ðð1−ηÞSð0Þ þ ηSð1ÞÞ ) ðð1−ηÞEð0Þ þ ηEð1ÞÞ:
The zero of the derivative ∂WðηÞ=∂η gives the optimal value η0 of
the splitting parameter η. Thus
∂WðηÞ
∂η¼ ð−Sð0Þ þ S
ð1ÞÞ ) ðð1−ηÞEð0Þ þ ηEð1ÞÞ þ ðð1−ηÞSð0Þ þ ηSð1ÞÞ
)ðEð1Þ−E
ð0ÞÞ;
and the optimal value is
η0 ¼2Wð00Þ
−Wð10Þ
−Wð01Þ
2ðWð11Þ þWð00Þ
−Wð10Þ
−Wð01ÞÞ
; ð86Þ
where
Wð00Þ ¼ S
ð0Þ ) Eð0Þ; ð87Þ
Wð01Þ ¼ S
ð0Þ ) Eð1Þ; ð88Þ
Wð10Þ ¼ S
ð1Þ ) Eð0Þ; ð89Þ
Wð11Þ ¼ S
ð1Þ ) Eð1Þ: ð90Þ
3. Numerical analysis
In this section we describe the Finite Element algorithm used to
solve the micropolar field equations. We develop the algorithm for
the calculation of the optimal value of the splitting parameter.
We compare the main kinematic variables for the square plate
made of dense polyurethane foam with the analytical solution of
the three-dimensional micropolar elasticity presented in [33] and
provide the analysis of the plates of different shapes.
3.1. Comparison of the analytical solutions
In our computations we consider plates made of polyurethane
foam—a material reported in the literature to be micropolar [21].
Insulation materials made from rigid polyurethane foam are used
in the construction of large industrial and agricultural buildings.
It is known that in the plaster supports in wooden structures and
bridging large spans between the top chords in flat roofing, the
insulation materials are exposed to bending stresses [15].
Therefore the prediction of the behavior of the polyurethane foam
under bending stress is extremely important. Another example of
the polyurethane foam plates can be found in the structural
insulated panels – a high-performance walling materials, where
the sides of the panel are made of cement, plywood, or oriented
strand board, and the middle part is made of highly dense
polyurethane foam. The structural insulated panels are widely
used in walls, floor slabs and roofs of houses and commercial
buildings.
In our calculations we will use the following technical para-
meters: Young's modulus E, Poisson's ratio ν, the characteristic
length for bending lb, the characteristic length for torsion lt, the
coupling number N.
The conversion formulas between the technical constants and
elastic parameters [17]
E¼μð3λþ 2μÞ
λþ μ; ν¼
λ
2ðλþ μÞ;
lb ¼1
2
ffiffiffiffiffiffiffiffiffiffiffi
γ þ ϵ
μ
r
; lt ¼
ffiffiffi
γ
μ
r
; N¼
ffiffiffiffiffiffiffiffiffiffiffiffi
α
μþ α
r
;
imply the following inverse conversion formulas:
λ¼Eν
2ν2 þ ν−1; μ¼
E
2ðνþ 1Þ; α¼
EN2
2ðνþ 1ÞðN2−1Þ
;
β¼Elt
2ðνþ 1Þ; γ ¼
El2t
2ðνþ 1Þ; ϵ¼
Eð4l2b−l2t Þ
2ðνþ 1Þ:
Let us consider a plate of thickness h¼0.1 m made of polyur-
ethane foam, and the ratio a=h varying from 5 to 30. The values of
the technical elastic parameters for the polyurethane foam are
reported in [21]: E¼ 299:5 MPa, ν¼ 0:44, lt ¼ 0:62 mm, lb ¼ 0:327,
N2 ¼ 0:04. These values correspond to the following values
of Lamé and asymmetric parameters: λ¼ 762:616 MPa, μ¼
103:993 MPa, α¼ 4:333 MPa, β¼ 39:975 MPa, γ ¼ 39:975 MPa,
ϵ¼ 4:505 MPa (the ratio β=γ is equal to 1 for the bending).
Let us first analyze the case of the square plate and compare the
analytical solution for the field equations (82) and the analytical
solution of the corresponding three-dimensional micropolar elas-
ticity problem presented in [33]. Let us consider the following
hard simply supported boundary conditions:
W ¼ 0; Wn ¼ 0; Ψ ) s¼ 0; n )Mn¼ 0; Sn) n¼ 0; ð91Þ
Q n) s¼ 0; Q̂ ) s¼ 0; s ) Rn¼ 0; s ) Rnn¼ 0: ð92Þ
where n and s are the normal and the tangent vectors to the
boundary ∂B0 respectively.
Let the body B0 be a square plate a* a of thickness
h: B0 ¼ ½0; a( * ½0; a( * ½−h=2;h=2(. The boundary of the body is
given by
G¼ G1∪G2
where
G1 ¼ fðx1; x2Þ : x1∈I; x2∈½0; a(g;
G2 ¼ fðx1; x2Þ : x1∈½0; a(; x2∈Ig;
and I¼ f0; ag is the set of two numbers: 0 and a.
For the square plate, the hard simply supported boundary
conditions (90) and (91) can be represented in the following
mixed Dirichlet–Neumann form:
G1 : W ¼ 0; Wn ¼ 0; Ψ 2 ¼ 0; Ω01 ¼ 0; Ω̂
0
1 ¼ 0;
G1 : Ω3 ¼ 0;∂Ψ 1
∂n¼ 0;
∂Ω02
∂n¼ 0;
∂Ω̂0
2
∂n¼ 0;
G2 : W ¼ 0; Wn ¼ 0; Ψ 1 ¼ 0; Ω02 ¼ 0; Ω̂
0
2 ¼ 0;
R. Kvasov, L. Steinberg / Thin-Walled Structures 69 (2013) 67–7872
G2 : Ω3 ¼ 0;∂Ψ2
∂n¼ 0;
∂Ω01
∂n¼ 0;
∂Ω̂0
1
∂n¼ 0:
The initial distribution of the pressure is sinusoidal and is given
in MPa units:
pðx1; x2Þ ¼ sinπx1a
1 2
sinπx2a
1 2
: ð93Þ
The analytical solution of the three-dimensional micropolar
bending problem is presented in [33]. The two-dimensional field
equations (82) can be solved by applying the method of separation
of variables. We obtain the kinematic variables in the following
form:
Ψ1 ¼ A1 cosπx1a
1 2
sinπx2a
1 2
;
Ψ2 ¼ A2 sinπx1a
1 2
cosπx2a
1 2
; ð94Þ
W ¼ A3 sinπx1a
1 2
sinπx2a
1 2
;
Ω03 ¼ A4 cos
πx1a
1 2
cosπx2a
1 2
; ð95Þ
Ω01 ¼ A5 sin
πx1a
1 2
cosπx2a
1 2
; ð96Þ
Ω02 ¼ A6 cos
πx1a
1 2
sinπx2a
1 2
; ð97Þ
Wn ¼ A7 sinπx1a
1 2
sinπx2a
1 2
;
Ω̂0
1 ¼ A8 sinπx1a
1 2
cosπx2a
1 2
; ð98Þ
Ω̂0
2 ¼ A9 cosπx1a
1 2
sinπx2a
1 2
:
We find Ai by substituting the expressions (96) into the system
of Eqs. (82) and solving the obtained system of nine linear
equations for Ai.
The comparison of the analytical solutions for the maximum
vertical deflection and the maximum microrotation of the two-
dimensional model versus the analytical solutions for three-
dimensional micropolar elasticity is provided in Tables 1 and 2.
The comparison of the numerical values of the kinematic
variables for the square plate made of dense polyurethane foam
with the analytical solution of the three-dimensional micropolar
elasticity confirms the high order of approximation of three-
dimensional (exact) solution. The relative error of order 1% for
the maximum vertical deflection is compatible with the precision
of the Reissner plate theory [10] (Fig. 1).
The comparison of the transverse variations of the components
of the displacement of the two-dimensional micropolar plate
theory versus three-dimensional micropolar elasticity theory is
given in Fig. 2.
3.2. Finite element analysis
Let us look for the solution of the micropolar plate bending
system of Eqs. (82) in the Hilbert space V defined as
V ¼ V1 * V2 * V3 * V4 * V5 * V6 * V7 * V8 * V9;
where V1 ¼ V6 ¼ V9 ¼ fv∈H1ðB0Þ; v¼ 0 on G1g, V2 ¼ V5 ¼ V8 ¼
fv∈H1ðB0Þ; v¼ 0 on G2g, V3 ¼ V7 ¼ fv∈H1ðB0Þ; v¼ 0 on Gg,
V4 ¼H1ðB0Þ, and H1ðB0Þ is a standard Hilbert space of func-
tions that are square-integrable together with their first partial
derivatives.
The weak formulation of the micropolar plate bending problem
is obtained as follows. Let V∈H1ðB0Þ9. By multiplying both sides of
(82) by V we have
VLU ¼ VF ðηÞ
and then integrating over the plate B0 we obtainZ
B0
VLU ds¼
Z
B0
VF ðηÞ ds:
Let us introduce the following notation:
aðU;VÞ ¼
Z
B0
VLU ds¼
Z
B0
V iLijU j ds;
f ðVÞ ¼
Z
B0
VF ðηÞ ds:
The corresponding weak problem is formulated as follows: find
all U∈V such that for all V∈H1ðB0Þ9
aðU;VÞ ¼ f ðVÞ: ð99Þ
Note that the form aðU;VÞ ¼R
B0V iLijU j ds is a summation of the
weak forms of the operators Lij. In order to obtain the expression
for the weak formulation (99) we observe that there are only three
types of linear operators present in the expression for aðU ;VÞ—
operators of order zero, one and two:
Lð0Þ ¼ 1: ð100Þ
Table 1
Comparison of the maximum vertical deflection u3 of the polyurethane foam square plate.
a=h 10 15 20 25 30
Optimal value of η 0.2383156 0.1280473 0.0795667 0.0548782 0.0407998
Micropolar plate model (m) 0.0236306 0.0622810 0.1205956 0.2011537 0.3076744
Micropolar 3D elasticity (m) 0.0237443 0.0624282 0.1207626 0.2013360 0.3078709
Relative error (%) 0.47 0.23 0.13 0.09 0.06
Table 2
Comparison of the maximum microrotation φ1 of the polyurethane foam square plate.
a=h 10 15 20 25 30
Optimal value of η 0.2383156 0.1280473 0.0795667 0.0548782 0.0407998
Micropolar Plate Model 0.0013365 0.0052564 0.0132366 0.0266257 0.0467707
Micropolar 3D Elasticity 0.0013430 0.0052686 0.0132545 0.0266490 0.0467995
Relative Error (%) 0.48 0.23 0.14 0.09 0.06
R. Kvasov, L. Steinberg / Thin-Walled Structures 69 (2013) 67–78 73
Lð1Þ ¼∂
∂xi: ð101Þ
Lð2Þ ¼∇ ) A∇: ð102Þ
where A¼ A11A21
A12A22
1 2
.
The weak forms of these operators are given as follows:
Z
B0
vðLð0ÞuÞ ds¼ c
Z
B0
vu da;
Fig. 1. Comparison of the vertical deflection for the square plate made of dense polyurethane foam: analytical solution of the proposed plate theory WP and the analytical
solution of the three-dimensional micropolar elasticity WE .
Fig. 2. The comparison of the analytical solutions of the transverse variation for the displacement components for the square plate made of polyurethane foam: solid red line
– the two-dimensional micropolar plate theory, dashed black line – three-dimensional micropolar elasticity ða=h¼ 30Þ.
R. Kvasov, L. Steinberg / Thin-Walled Structures 69 (2013) 67–7874
Z
B0
vðLð1ÞuÞ ds¼
Z
B0
v∂u
∂xida;
Z
B0
vðLð2ÞuÞ ds¼
Z
B0
∇ ) A∇uv da¼
Z
∂B0
ðA∇u ) nÞv dτ−
Z
B0
A∇u )∇v da;
where the weak form of the second order operator is obtained by
performing the corresponding integration by parts.
Note that all second order operators in the weak formulation
(99) can be represented in the form (102):
∂2
∂x21þ
∂2
∂x22¼∇ )
1 0
0 1
# $
∇;
∂2
∂x1x2¼∇ )
0 12
12 0
!
∇;
c1∂2
∂x21þ c2
∂2
∂x22¼∇ )
c1 0
0 c1
!
∇:
In the previous section we showed that the system (82)
represents an elliptic system of partial differential equations. It is
possible to prove that the operator L is in fact strongly elliptic. This
will imply the coerciveness of the operator L and its bilinear form
on V, which is important for the verification of the hypothesis of
Babuška–Lax–Milgram theorem (see [25,3] for details). The corre-
sponding convergence analysis of finite element approximations
of the elliptic systems and the analytic estimates on the domains
with corners are given in [8].
Theorem 1 (Strong ellipticity of the bending system of equations).
The micropolar plate bending system of Eqs. (82) is strongly elliptic.
Proof. Let ξ¼ ðξ1; ξ2Þ and υ¼ ðυ1; υ2; υ3; υ4; υ5; υ6; υ7; υ8; υ9Þ be some
arbitrary vectors from R2 and R9 respectively. Let us assume that
ξ¼ ðξ1; ξ2Þ≠0 and denote the principle symbol of the system of
equations (82) by LðξÞ. The bilinear form υLðξÞυT can be expressed as
υLðξÞυT ¼ l1 þ l2 þ l3 þ l4 þ l5;
where
l1 ¼ c1ðυ1ξ1 þ υ2ξ2Þ2 þ c2ðυ2ξ1−υ1ξ2Þ
2;
l2 ¼ c6υ24ðξ
21 þ ξ22Þ;
l3 ¼ c4υ23ðξ
21 þ ξ22Þ
2 þ4
5c3υ
27ðξ
21 þ ξ22Þ
2;
l4 ¼ c7ð45ðυ8ξ1 þ υ9ξ2Þ
2 þ ðυ5ξ1 þ υ6ξ2Þðυ8ξ1 þ υ9ξ2Þ þ ðυ5ξ1 þ υ6ξ2Þ2Þ;
l5 ¼ c8ð45ðυ9ξ1−υ8ξ2Þ
2 þ ðυ6ξ1−υ5ξ2Þðυ9ξ1−υ8ξ2Þ þ ðυ6ξ1−υ5ξ2Þ2Þ:
It is easy to check that l1, l2, and l3 are always nonnegative.
Indeed, since c1; c240 then l1≥0 for all ξ∈R2, and l1 ¼ 0 if and only
if υ1 ¼ υ2 ¼ 0. Since c640 then l2≥0 for all ξ∈R2, and l2 ¼ 0 if and
only if υ4 ¼ 0. Since c3; c440 then l3≥0 for all ξ∈R2, and l3 ¼ 0 if
Fig. 3. The cross section that contains the center of the micropolar square plate 2.0 m*2.0 m* 0.1 m made of polyurethane foam without load (left) and under the load
(right). The applied sinusoidal load is represented by the dashed black line.
R. Kvasov, L. Steinberg / Thin-Walled Structures 69 (2013) 67–78 75
and only if υ3 ¼ υ7 ¼ 0. The analysis of l4 and l5 is very similar
since they have quadratic form: 45 x
2 þ xyþ y2 with non-positive
discriminant. Since c740 and c840 then l4≥0 and l5≥0; l3 ¼ 0 if
and only if υ8 ¼ υ9 ¼ υ5 ¼ υ6 ¼ 0, and l4 ¼ 0 if and only if
υ8 ¼ υ9 ¼ υ5 ¼ υ6 ¼ 0. Since li≥0, then υLðξÞυT ¼ 0 if and only if all
li ¼ 0. The last is equivalent to υ¼ 0 and therefore we conclude
that the operator L is strongly elliptic.
The strong ellipticity of the operator L implies the coerciveness
of the bilinear form AðU;VÞ and therefore by Babuška–Lax–
Milgram theorem there exists a unique solution U∈V to the weak
(99).
The proposed minimization of the plate stress energy allows us
to introduce the following effective algorithm for the optimal
value of the splitting parameter η; which can be used for the plate
of an arbitrary shape.
Fig. 4. Hard simply supported square plate 2.0 m*2.0 m*0.1 m made of polyurethane foam: the initial mesh and the isometric view of the resulting vertical deflection
of the plate.
0.0 0.5 1.0 1.5 2.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.5 1.0 1.5 2.0
0.0
0.2
0.4
0.6
0.8
1.0
Fig. 5. Nonrectangular hard simply supported plate made of polyurethane foam: (a) the area where the sinusoidal load is applied, (b) initial mesh, (c) the density plot
and (d) the isometric view of the resulting vertical deflection of the plate.
R. Kvasov, L. Steinberg / Thin-Walled Structures 69 (2013) 67–7876
Algorithm 2 (Optimal value of the splitting parameter).
1. Solve the systems (82) and (83) for U0 and U1 respectively.
2. Calculate the components of the micropolar plate stress sets
Sð0Þ and S
ð1Þ from the sets of kinematic variables U0 and U1,
using the constitutive formulas in the reverse form (71)–(80).
3. Calculate the components of the micropolar plate strain sets.
Eð0Þ and E
ð1Þ from the sets of kinematic variables U0 and U1,
using the strain–displacement relations (57)–(63).
4. Find the work densities Wð00Þ, W
ð11Þ, Wð10Þ and W
ð01Þ by
substituting the micropolar plate stress and strain sets Sð0Þ,
Sð1Þ, Eð0Þ and E
ð1Þ into the definitions (86)–(89).
5. Substitute the values of the work densities Wð00Þ, Wð11Þ, Wð10Þ
and Wð01Þ into the expression for the optimal value of the
splitting parameter η0 (86).
The above algorithm can be efficiently used to find the solution
of the system of equations (82). Indeed, once the optimal value of
the splitting parameter η0 is found, the solution Uη0of the bending
system of Eqs. (82) is found from (85) as a linear combination of
the sets of kinematic variables U0 and U1
Uη0¼ ð1−η0ÞU0 þ η0U1:
The numerical modeling of the vertical deflection of the square
plate made of polyurethane foam is given in Figs. 3 and 4.
The distribution of the vertical deflection for nonrectangular
plates and plates with rectangular holes is given in Figs. 5–7.
The effect of the asymmetric part on the maximal vertical deflec-
tion of the plate made of polyurethane foam is shown in Table 3 and
Fig. 8 for a¼1.0 m and the ratio a=h varying from 10 to 30.
Fig. 8 provides information for comparison between micropolar
and classical elasticity. The energy of microrotation becomes a part
of the total elastic energy. This causes the redistribution of the
elastic energy depending on the value of the asymmetric part.
We perform computations for different levels of the asymmetric
microstructure by reducing the values of the elastic asymmetric
parameters. In the case when 0.1% of real values of asymmetric
parameters are used, i.e. the microstructure is almost irrelevant,
the solution of the proposed model converges to the correspond-
ing Reissner solution. Fig. 8 illustrates the size effect of micropolar
plate theory which predicts that plates of smaller thickness will be
more rigid than would be expected on the basis of the Reissner
plate theory. Similar experimental behavior was reported in [21]
for torsion and bending of cylindrical rods of a Cosserat solid.
4. Conclusion
In this paper we presented the Finite Element modeling of the
bending of micropolar elastic plates. Based on our enhanced math-
ematical model for Cosserat plate bending [33] we introduced the
micropolar plate field equations as an elliptic system of nine differ-
ential equations in terms of the kinematic variables. We have devel-
oped the efficient algorithm for finding the optimal value of the
splitting parameter and discussed the approximations of stress and
couple stress components. The numerical algorithm includes the
efficient method for finding the solution of the micropolar plate field
10 15 20 25 300.994
0.995
0.996
0.997
0.998
0.999
1.000
1.001
Deflection 3D Elasticity
Deflection Proposed Plate Theory
Fig. 6. Rectangular plate made of polyurethane foam with hard simply supported
rectangular holes: (a) the area where the sinusoidal load is applied, (b) initial mesh,
(c) the density plot and (d) the isometric view of the resulting vertical deflection of
the plate.
0.04 0.02 0.00 0.02 0.04
0.010
0.005
0.000
0.005
0.010
0.015
Thickness x3 m
Dis
pla
cem
entu
1m
0.04 0.02 0.00 0.02 0.04
0.015
0.010
0.005
0.000
0.005
0.010
0.015
Thickness x3 m
Dis
pla
cem
entu
2m
0.04 0.02 0.00 0.02 0.04
0.3072
0.3073
0.3074
0.3075
0.3076
0.3077
0.3078
Thickness x3 m
Dis
pla
cem
entu
3m
0.015
Fig. 7. Rectangular plate made of polyurethane foam with hard simply supported rectangular holes: (a) the area where the sinusoidal load is applied, (b) initial mesh,
(c) the density plot and (d) the isometric view of the resulting vertical deflection of the plate.
Table 3
Effect of the asymmetric part on the vertical deflection of the plate.
a=h 10 15 20 25 30
Micropolar Plate (m) 0.02363 0.04071 0.05738 0.07370 0.08980
Micropolar Plate 10 of asymmetric part (m) 0.06835 0.17631 0.31858 0.47802 0.64428
Micropolar Plate 1% of asymmetric part (m) 0.08509 0.26975 0.60504 1.11329 1.80130
Micropolar Plate 0.1% of asymmetric part (m) 0.08723 0.28491 0.66508 1.28470 2.19753
Reissner Solution (m) 0.08748 0.28670 0.67250 1.30707 2.25260
R. Kvasov, L. Steinberg / Thin-Walled Structures 69 (2013) 67–78 77
equations corresponding to the optimal value of the splitting para-
meter. The comparison of the numerical values of the kinematic
variables for the square plate made of the dense polyurethane foam
with the analytical solution of the three-dimensional micropolar
elasticity confirmed the high order of approximation of three-
dimensional solution. The size effect of micropolar plate theory
predicts that plates made of smaller thickness will be more rigid than
would be expected on the basis of the Reissner plate theory. The
numerical modeling of the bending nonrectangular plates and plates
with rectangular holes made of the polyurethane foam under different
loads have been provided.
Acknowledgments
We would like to thank the reviewers for useful suggestions
and Dr. Paul Castillo, Dr. Robert Acar and Dr. Frederick Just for
helpful discussions in the preparation of the paper.
References
[1] Altenbach H, Eremeyev VA. Thin-walled structures made of foams. In: H.Cellular and porous materials: modeling, testing, application, CISM Coursesand Lecture Notes, vol. 521. Springer; 2010. p. 167–242.
[2] Anderson W, Lakes R. Size effectes due to cosserat elasticity and surfasedamage in closed-cell polymenthacrylimide foam. Journal of Materials Science1994;29:6413–9.
[3] Babuška I. Error-bounds for finite element method. Numerische Mathematik1971;16:322–33.
[5] Chang T, Lin H, Wen-Tse C, Hsiao J. Engineering properties of lightweightaggregate concrete assessed by stress wave propagation methods. Cement andConcrete Composites 2006;28:57–68.
[7] Cosserat E, Cosserat F. Theorie des corps deformables; 1909.[8] Costabel M, Dauge M, Nicaise S. Analytic regularity for linear elliptic systems
in polygons and polyhedra; 2011.[9] Costabel M, Dauge M, Nicaise S. Corner singularities and analytic regularity for
linear elliptic systems; 2010.
[10] Donnell L, Beams. Plates and shells engineering societies monographs.
McGraw-Hill Inc.; 1976.[11] Eringen AC. Microcontinuum field theories. Springer; 1999.[12] Eringen AC. Theory of micropolar elasticity. Fracture 1968;2:621–729.[13] Eringen AC. Theory of micropolar plates. Journal of Applied Mathematics and
Physics (ZAMP) 1967;18:12–30.[15] Federation of European Rigid Polyurethane Foam Associations Report no. 1.
Thermal insulation materials made of rigid polyurethane foam (PUR/PIR)
BING; 2006.[17] Gauthier R, Jahsman W. A quest for micropolar elastic constants. Journal of
Applied Mechanics 1975;42:369–74.[18] Gerstle W., Sau N., Aguilera E., Micropolar peridynamic constitutive model for
concrete. In Transactions of the 19th International Conference of Structural
Mechanics in Reactor Technology (SMiRT 19). Toronto Canada 2007; August
12-17, SMiRT19-B02/1: 1–8.[19] Kumar R. Wave propagation in micropolar viscoelastic generalized thermo-
elastic solid. International Journal of Engineering Science 2000;38.[21] Lakes R. Experimental microelasticity of two porous solids. International
Journal of Solid Structures 1986;22(1):55–63.[23] Love AEH. A treatise on the mathematical theory of elasticity. New York:
Dover; 1986.[24] Merkel A, Tournat V, Gusev V. Experimental evidence of rotational elastic
waves in granular phononic crystals. Physical Review Letters 2011;107.[25] McLean W. Strongly elliptic systems and boundary integral equations. Cam-
bridge: Cambridge University Press; 2000.[26] Nowacki W. Theory of asymmetric elasticity. Oxford, New York: Pergamon
Press; 1986 (2007).[27] Purasinghe R, Tan S, Rencis J. Micropolar elasticity model the stress analysis of
human bones. In: Proceedings of the 11th annual international conference;
1989.[30] Reissner E. Reflections on the theory of elastic plates. Applied Mechanics
Reviews 1985;38:1453–64.[31] Rössle A, Bischoff M, Wendland W, Ramm E. On the mathematical foundation
of the (1,1,2)-plate model. International Journal of Solids and Structures
1999;36:2143–68.[32] Steinberg L. Deformation of micropolar plates of moderate thickness. Inter-
national Journal of Applied Mathematics and Mechanics 2010;6(17):1–24.[33] Steinberg L, Kvasov R. Enhanced mathematical model for Cosserat plate
bending. Thin-Walled Structures 63; 2013:51–62.
10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
a h
WMWR
Reissner Plate
Micropolar Plate with 0.1 asymmetric part
Micropolar Plate with 1 asymmetric part
Micropolar Plate with 10 asymmetric part
Micropolar Plate
Fig. 8. Comparison of the vertical deflections of the micropolar plate WM with Reissner plate WR , which illustrates the influence of the thickness of the micropolar plate.
R. Kvasov, L. Steinberg / Thin-Walled Structures 69 (2013) 67–7878