Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Method and Computational Structural Dynamics

Finite Elements for Plates and Shells

Manish Shrikhande
[email protected]

Department of Earthquake Engineering

Indian Institute of Technology Roorkee

Based on “Shrikhande, M. Finite Element Method and Computational

Structural Dynamics, PHI Learning Pvt. Ltd., New Delhi, India.
2014. ISBN-13: 978-8120349957.”

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–1

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–1

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–1

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–1

I The plates and shells can be conceived as
Z two-dimensional extensions of the beams and arches,
respectively and are used to support the loads normal to
the surface.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–1

I The plates and shells can be conceived as
Z two-dimensional extensions of the beams and arches,
respectively and are used to support the loads normal to
the surface.
I A distinctive feature for plates and shells is that the
Y thickness dimension is much smaller than the other two
dimensions.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–1

I The plates and shells can be conceived as
Z two-dimensional extensions of the beams and arches,
respectively and are used to support the loads normal to
the surface.
I A distinctive feature for plates and shells is that the
Y thickness dimension is much smaller than the other two
dimensions.
I Plates differ from shells in supporting the normal loads
by predominantly flexural and some shear deformations,
X whereas for shells the in-plane or membrane action is
predominant.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–1

I The plates and shells can be conceived as
Z two-dimensional extensions of the beams and arches,
respectively and are used to support the loads normal to
the surface.
I A distinctive feature for plates and shells is that the
Y thickness dimension is much smaller than the other two
dimensions.
I Plates differ from shells in supporting the normal loads
by predominantly flexural and some shear deformations,
X whereas for shells the in-plane or membrane action is
predominant.
I Modeling is based on assumed kinematic constraints.
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–2

I Since the thickness is very small in comparison with other two dimensions, the out
of plane deformations are assumed to be invariant through the thickness.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–2

I Since the thickness is very small in comparison with other two dimensions, the out
of plane deformations are assumed to be invariant through the thickness.
I Allows development of a working model for the strain field with reference to the
out of plane deformation of the mid-surface (z = 0).

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–2

I Since the thickness is very small in comparison with other two dimensions, the out
of plane deformations are assumed to be invariant through the thickness.
I Allows development of a working model for the strain field with reference to the
out of plane deformation of the mid-surface (z = 0).
I The invariance of the deformation through the thickness effectively transforms the
stress analysis problem to a two-dimensional reference frame (XY -plane).

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–2

I Since the thickness is very small in comparison with other two dimensions, the out
of plane deformations are assumed to be invariant through the thickness.
I Allows development of a working model for the strain field with reference to the
out of plane deformation of the mid-surface (z = 0).
I The invariance of the deformation through the thickness effectively transforms the
stress analysis problem to a two-dimensional reference frame (XY -plane).
I Material filaments normal to the mid-surface (analogous to plane sections normal
to the neutral axis in an Euler-Bernoulli beam) extend upto ±h/2 from the
mid-surface and the bounding surface of the plate at +h/2 is referred to as the
top surface while the bounding surface at −h/2 is known as the bottom surface.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–3

I In the case of a plate, the thickness is small but not too small in comparison with
the other dimensions.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–3

I In the case of a plate, the thickness is small but not too small in comparison with
the other dimensions.
I An extremely thin entity acts like a fabric and has negligible flexural
resistance—an important aspect of the plate behaviour. The thickness to length
ratio t/L varies from 1/100 to 1/5, where L denotes the characteristic length.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–3

I In the case of a plate, the thickness is small but not too small in comparison with
the other dimensions.
I An extremely thin entity acts like a fabric and has negligible flexural
resistance—an important aspect of the plate behaviour. The thickness to length
ratio t/L varies from 1/100 to 1/5, where L denotes the characteristic length.
I Plate-like behaviour: the flexural deformation in response to the applied loads
normal to the mid-surface donot cause any significant changes in the dimensions
of the mid-surface — inextensional bending

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Geometry and Kinematics of Plates and Shells–3

I In the case of a plate, the thickness is small but not too small in comparison with
the other dimensions.
I An extremely thin entity acts like a fabric and has negligible flexural
resistance—an important aspect of the plate behaviour. The thickness to length
ratio t/L varies from 1/100 to 1/5, where L denotes the characteristic length.
I Plate-like behaviour: the flexural deformation in response to the applied loads
normal to the mid-surface donot cause any significant changes in the dimensions
of the mid-surface — inextensional bending
I Shell-like behaviour: when the mid-surface undergoes significant stretch (or,
contraction) — extensional bending.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–1

I The material normals to the mid-surface plane before deformation remain straight
and normal to the mid-surface after deformation.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–1

I The material normals to the mid-surface plane before deformation remain straight
and normal to the mid-surface after deformation.
I The transverse shear is assumed to be negligible—a condition that is closely
approximated by thin plates under small deformations, i.e., the normal
deformation of the mid-surface (w (x, y , 0)) is very small in comparison with the
plate thickness, h.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–1

I The material normals to the mid-surface plane before deformation remain straight
and normal to the mid-surface after deformation.
I The transverse shear is assumed to be negligible—a condition that is closely
approximated by thin plates under small deformations, i.e., the normal
deformation of the mid-surface (w (x, y , 0)) is very small in comparison with the
plate thickness, h.
I The variation of the normal displacement w (x, y ) with respect to the thickness
direction, z is considered negligible, i.e., w (x, y , z) = w (x, y ).

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–1

I The material normals to the mid-surface plane before deformation remain straight
and normal to the mid-surface after deformation.
I The transverse shear is assumed to be negligible—a condition that is closely
approximated by thin plates under small deformations, i.e., the normal
deformation of the mid-surface (w (x, y , 0)) is very small in comparison with the
plate thickness, h.
I The variation of the normal displacement w (x, y ) with respect to the thickness
direction, z is considered negligible, i.e., w (x, y , z) = w (x, y ).
I The plate thickness is uniform so that the three-dimensional stress effects are
negligible and the plate is symmetric in fabrication about its mid-surface.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–2

I The normal loads should be distributed over the plate surface area of dimension h,
or larger.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–2

I The normal loads should be distributed over the plate surface area of dimension h,
or larger.
I In accordance with the assumption of intextensional bending, the boundary
conditions are such that no significant extensions of the mid-surface can develop.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–2

I The normal loads should be distributed over the plate surface area of dimension h,
or larger.
I In accordance with the assumption of intextensional bending, the boundary
conditions are such that no significant extensions of the mid-surface can develop.
I The complete deformation field can be stated in terms of normal displacement of
the mid-surface, w (x, y ), and rotations of the mid-surface normals about X axis
(θx ) and about Y axis (θy ), which are related to the slope of the deformed
mid-surface by:
∂w ∂w
− θx = γyz = 0 and + θy = γzx = 0
∂y ∂x
γyz and γzx : the transverse shear strains in the YZ and ZX planes respectively.
These conditions are known as Kirchhoff constraints for bending of thin plates and
enforce vanishing transverse shear strains condition.
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–3

θy

Z, w

Ω w _
= θy Deformed
x
X

X θx
Undeformed

Γ Mid−surface profile at y=0.

Two-dimensional domain and mid-surface profile (the  represents a normal pointing

out of the plane while the ⊗ denotes a normal pointing into the plane)
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–4

I The displacement field at any point in the plate may be given by:
∂w ∂w
u(x, y , z) = zθy = −z ; v (x, y , z) = −zθx = −z ; w (x, y , z) = w (x, y , 0)
∂x ∂y

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–4

I The displacement field at any point in the plate may be given by:
∂w ∂w
u(x, y , z) = zθy = −z ; v (x, y , z) = −zθx = −z ; w (x, y , z) = w (x, y , 0)
∂x ∂y
I The strain field may be given as:
∂u ∂2w ∂v ∂2w ∂u ∂v ∂2w
xx = = −z 2 ; yy = = −z 2 ; γxy = + = −2z
∂x ∂x ∂y ∂y ∂y ∂x ∂x ∂y
∂w ∂v ∂w ∂u ∂w
zz = =0 ; γyz = + = 0 ; γzx = + =0
∂z ∂z ∂y ∂z ∂x

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–4

I The displacement field at any point in the plate may be given by:
∂w ∂w
u(x, y , z) = zθy = −z ; v (x, y , z) = −zθx = −z ; w (x, y , z) = w (x, y , 0)
∂x ∂y
I The strain field may be given as:
∂u ∂2w ∂v ∂2w ∂u ∂v ∂2w
xx = = −z 2 ; yy = = −z 2 ; γxy = + = −2z
∂x ∂x ∂y ∂y ∂y ∂x ∂x ∂y
∂w ∂v ∂w ∂u ∂w
zz = =0 ; γyz = + = 0 ; γzx = + =0
∂z ∂z ∂y ∂z ∂x
I The transverse shear strains are zero implying that the corresponding shear
stresses should also vanish for an isotropic and elastic plate. However, the
transverse shear stresses are necessary to maintain equilibrium with the normal
loads.
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–5

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–5

Myx + ∆ Myx Myx + ∆ Myx

Y Q y + ∆Q y
Myy + ∆ Myy Myy + ∆ Myy
Mxy Mxx τxz Q + Q Mxx Q x + ∆Q x
q x ∆ x
dy
Mxx + ∆ Mxx Mxy q Mxy + ∆ Mxy
Z Qx dy
σxx Mxy + ∆ Mxy
τxy Qx Mxx + ∆ Mxx
h
τyz σyy τyx X Qy Myy
Q y Myy
Myx
Myx
dx dx

Stress resultants in a plate supporting normal loads

(positive sense)

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–5

Myx + ∆ Myx Myx + ∆ Myx

Y Q y + ∆Q y
Myy + ∆ Myy Myy + ∆ Myy
Mxy Mxx τxz Q + Q Mxx Q x + ∆Q x
q x ∆ x
dy
Mxx + ∆ Mxx Mxy q Mxy + ∆ Mxy
Z Qx dy
σxx Mxy + ∆ Mxy
τxy Qx Mxx + ∆ Mxx
h
τyz σyy τyx X Qy Myy
Q y Myy
Myx
Myx
dx dx

Stress resultants in a plate supporting normal loads

(positive sense)

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–5

I The strain field corresponds
to a state of plane strain with
Myx + ∆ Myx Myx + ∆ Myx
Y Myy + ∆ Myy
Q y + ∆Q y
Myy + ∆ Myy
zz = 0, γyz = 0 and γzx = 0.
Mxy Mxx q τxz Q x + ∆Q x Mxx Q +
x ∆ xQ The plane stress σzz = 0
dy M q M + ∆ M
Z Qx
Mxx + ∆ Mxx
dy
xy xy xy condition is closer to practical
σxx Mxy + ∆ Mxy observations. Acceptable for
τxy Q x Mxx + ∆ Mxx
h
τyz σyy τyx X Qy M yy negligible Poisson’s effect in
Q y Myy
Myx
Myx an isotropic homogeneous
dx dx
medium.
Stress resultants in a plate supporting normal loads
(positive sense)

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–5

I The strain field corresponds
to a state of plane strain with
Myx + ∆ Myx Myx + ∆ Myx
Y Myy + ∆ Myy
Q y + ∆Q y
Myy + ∆ Myy
zz = 0, γyz = 0 and γzx = 0.
Mxy Mxx q τxz Q x + ∆Q x Mxx Q +
x ∆ xQ The plane stress σzz = 0
dy M q M + ∆ M
Z Qx
Mxx + ∆ Mxx
dy
xy xy xy condition is closer to practical
σxx Mxy + ∆ Mxy observations. Acceptable for
τxy Q x Mxx + ∆ Mxx
h
τyz σyy τyx X Qy M yy negligible Poisson’s effect in
Q y Myy
Myx
Myx an isotropic homogeneous
dx dx
medium.
Stress resultants in a plate supporting normal loads I The normal loading and
(positive sense) bending of plate leads to
distribution of stresses and
stress resultants in the plate.
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–5

I The stress resultants (per unit length) are related to the bending stresses as:
Z h/2 Z h/2
Mxx = −σxx z dz Myy = −σyy z dz
−h/2 −h/2
Z h/2 Z h/2
Mxy = −τxy z dz Myx = −τyx z dz
−h/2 −h/2

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–5

I The stress resultants (per unit length) are related to the bending stresses as:
Z h/2 Z h/2
Mxx = −σxx z dz Myy = −σyy z dz
−h/2 −h/2
Z h/2 Z h/2
Mxy = −τxy z dz Myx = −τyx z dz
−h/2 −h/2

I The moment equilibrium requires that Mxy = Myx (or, τxy = τyx ).

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–5

I The stress resultants (per unit length) are related to the bending stresses as:
Z h/2 Z h/2
Mxx = −σxx z dz Myy = −σyy z dz
−h/2 −h/2
Z h/2 Z h/2
Mxy = −τxy z dz Myx = −τyx z dz
−h/2 −h/2

I The moment equilibrium requires that Mxy = Myx (or, τxy = τyx ).
I The stresses are obviously not invariant through the thickness of the plate and
hence the plane stress assumption cannot be justified.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–5

I The stress resultants (per unit length) are related to the bending stresses as:
Z h/2 Z h/2
Mxx = −σxx z dz Myy = −σyy z dz
−h/2 −h/2
Z h/2 Z h/2
Mxy = −τxy z dz Myx = −τyx z dz
−h/2 −h/2

I The moment equilibrium requires that Mxy = Myx (or, τxy = τyx ).
I The stresses are obviously not invariant through the thickness of the plate and
hence the plane stress assumption cannot be justified.
I As a practical solution, we assume each lamina of the plate (i.e., the plane at
z = constant) to be in a state of plane stress.
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–6

I For an isotropic plate, we have:
       
σxx 1 ν 0 xx 1 ν 0 κxx
σyy  = E ν 1 0   yy = −
 zE 
ν 1 0   κyy 
1 − ν2 1−ν 1 − ν2 1−ν
τxy 0 0 2 γ xy 0 0 2 2κxy
∂ w 2
where, κxy (= ∂x∂y ) and similarly for κxx and κyy represent the curvatures of the
plate mid-surface. These curvatures—assumed to be uniform over the plate
thickness—can be related to the moments:
    
Mxx 1 ν 0 κxx
Myy  = D ν 1 0   κyy 
1−ν
Mxy 0 0 2 2κxy
Eh3
where, D = 12(1−ν 2 )
is known as the isotropic plate rigidity.
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–7

I The maximum stresses can be obtained as:
max,min 6Mxx max,min 6Myy max,min 6Mxy
σxx =± 2
; σyy = ± 2 ; τxy =± 2 (1)
h h h

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–7

I The maximum stresses can be obtained as:
max,min 6Mxx max,min 6Myy max,min 6Mxy
σxx =± 2
; σyy = ± 2 ; τxy =± 2 (1)
h h h
I These maximum/minimum values of the stresses occur at the plate surfaces.
These relations are useful in design calculations. The force equilibrium
considerations also dictate the existence of transverse shears varying parabolically
through the thickness similar to the shear stress distribution in Euler-Bernoulli
beam. The transverse shear stresses (τxz and τyz ) are maximum at the
mid-surface, vanish at the plate surface and vary through the thickness as:
4z 2 4z 2
   
max max
τxz = τxz 1− 2 and τyz = τyz 1− 2
h h

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–8

I The transverse shear forces (per unit length) may be given by:
Z h/2 Z h/2
2 max 2 max
Qx = τxz dz = τxz h and Qy = τyz dz = τyz h
−h/2 3 −h/2 3

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–8

I The transverse shear forces (per unit length) may be given by:
Z h/2 Z h/2
2 max 2 max
Qx = τxz dz = τxz h and Qy = τyz dz = τyz h
−h/2 3 −h/2 3

I The contribution of transverse shear is ignored in Kirchhoff plate bending theory

and therefore the transverse shear stresses should be very small in comparison to
the bending stresses.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–8

I The transverse shear forces (per unit length) may be given by:
Z h/2 Z h/2
2 max 2 max
Qx = τxz dz = τxz h and Qy = τyz dz = τyz h
−h/2 3 −h/2 3

I The contribution of transverse shear is ignored in Kirchhoff plate bending theory

and therefore the transverse shear stresses should be very small in comparison to
the bending stresses.
I The existence of finite transverse shears inspite of vanishing transverse shear
strains is reconciled by assuming infinite shear rigidity (G → ∞).

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–9

Equilibrium Equations

∂Qx ∂Qy ∂2w

+ + q = ρh 2
∂x ∂y ∂t
∂Mxx ∂Mxy ρh3 ∂ 2 θy ∂Mxy ∂Myy ρh3 ∂ 2 θx
+ + Qx = − and + + Qy =
∂x ∂y 12 ∂t 2 ∂x ∂y 12 ∂t 2
where, θx (= ∂w ∂w
∂y ) and θy (= − ∂x ) denote the rotation of material normals about X
and Y axes, respectively. Eliminating Qx and Qy from the above equilibrium equations
leads to the governing equation of motion as:

∂ 2 Mxx ∂ 2 Mxy ∂ 2 Myy ∂2w ρh3

 4
∂4w

∂ w
+2 + − q + ρh 2 − + =0
∂x 2 ∂x∂y ∂y 2 ∂t 12 ∂t 2 ∂x 2 ∂t 2 ∂y 2

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–10

Equilibrium Equations
In the operator form:
12 ∂ 2 w
 
−

Mxx 2 2
ρh3 T  h∂ 3 w∂t 
∇T LT Myy  − q − ∇  ∂t 2 ∂x  = 0
12 ∂3w
Mxy 2 ∂t ∂y

where, the operators ∇ and L are defined as:

∂2
 
0 0
 
1 ∂x22
∂  ∂
∇ ≡  ∂x and L ≡ 0 0

 ∂y 2
∂ ∂ ∂
∂y 0 ∂y ∂x

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–11

Equilibrium Equations

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–11

Equilibrium Equations

sY Y

Qn
Qx M xx M ns
n
x φ
dy X M xy M nn X
ds
x dx M yy
Qy
M xy

Stress resultants on a
plate edge at an arbitrary
orientation

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–11

Equilibrium Equations
The stress resultants on an arbitrarily inclined edge can be
related to the stress resultants along Cartesian directions as:
sY Y

Qn
M ns
Qn = Qx cos φ + Qy sin φ
Qx n M xx
x φ Mnn = Mxx cos2 φ + Myy sin2 φ + 2Mxy sin φ cos φ
dy X M xy M nn X
x dx
ds
M yy
Mns = (Myy − Mxx ) sin φ cos φ + Mxy (cos2 φ − sin2 φ)
Qy
M xy
and the cartesian derivatives may be related to the normal
Stress resultants on a
and tangential derivatives as:
plate edge at an arbitrary
orientation ∂ 
cos φ − sin φ
 ∂  
nx −ny
 ∂ 
∂x = ∂n = ∂n
∂ ∂ ∂
∂y sin φ cos φ ∂s
n y n x ∂s

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–12

Equilibrium Equations
The governing differential equation—in terms of the normal deflection—for the
bending of thin plates:

∂4w ∂4w ∂4w ∂2w ρh3

 4
∂4w
 
1 ∂ w
+ 2 + − q − ρh + + =0
∂x 4 ∂x 2 ∂y 2 ∂y 4 D ∂t 2 12 ∂t 2 ∂x 2 ∂t 2 ∂y 2
Eh 3
where, D = 12(1−ν 2 ) is the isotropic plate rigidity. The primary and secondary variables

of the problem can be identified from the weak form of the weighted residual
statement:
∂ 2 Mxy ∂ 2 Myy
 2
∂2w ρh3
 4
∂4w
ZZ 
∂ Mxx ∂ w
0= W +2 + − q + ρh 2 − + dΩ
Ω ∂x 2 ∂x∂y ∂y 2 ∂t 12 ∂t 2 ∂x 2 ∂t 2 ∂y 2

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–13

The Weak Form
The weak form can be developed by using divergence theorem as:
ρh3 ∂ 3 w ρh3 ∂ 3 w
Z     
0 = W Qx − nx + Qy − ny dΓ
Γ 12 ∂t 2 ∂x 12 ∂t 2 ∂y
Z  
∂W ∂W
− (Mxx nx + Mxy ny ) + (Mxy nx + Myy ny ) dΓ
Γ ∂x ∂y
ZZ  2
∂2W ∂2W

∂ W
+ Mxx + 2 Mxy + Myy dΩ
Ω ∂x 2 ∂x ∂y ∂y 2
ρh3 ∂W ∂ 3 w ∂W ∂ 3 w ∂2w
ZZ   ZZ  
+ + dΩ − W q − ρh 2 dΩ
12 Ω ∂x ∂t 2 ∂x ∂y ∂t 2 ∂y Ω ∂t
which leads to the symmetric bilinear form by substituting for moments in the domain
integrals from the moment-curvature relations.
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–14

The Weak Form

ρh3 ∂ 3 w ρh3 ∂ 3 w

Z    
0 = W Qx − nx + Qy − ny dΓ
Γ 12 ∂t 2 ∂x 12 ∂t 2 ∂y
Z  
∂W ∂W
− (Mxx nx + Mxy ny ) + (Mxy nx + Myy ny ) dΓ
Γ ∂x ∂y
ZZ  2  2 2 ∂2W ∂2w

∂ W ∂ w ∂ w
+D + ν + 2(1 − ν) ·
Ω ∂x 2 ∂x 2 ∂y 2 ∂x ∂y ∂x ∂y
2 2 2 3 ∂W ∂ 3 w ∂W ∂ 3 w
  ZZ  
∂ W ∂ w ∂ w ρh
+ ν 2 + dΩ + + dΩ
∂y 2 ∂x ∂y 2 12 Ω ∂x ∂t 2 ∂x ∂y ∂t 2 ∂y
∂2w
ZZ  
− W q − ρh 2 dΩ
Ω ∂t
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–15

The Weak Form
The cartesian derivatives in the boundary terms may be transformed into local normal
and tangential derivatives as:
∂W ∂W ∂W ∂W ∂W ∂W
= nx − ny and = ny + nx
∂x ∂n ∂s ∂y ∂n ∂s
and the weak form may be written as:
ρh3 ∂ 3 w
Z   Z Z
∂W ∂W
0 = W Qn − 2 ∂n
dΓ − · M nn dΓ − · Mns dΓ
Γ 12 ∂t Γ ∂n Γ ∂s
ZZ  2  2
∂2w ∂2W ∂2w

∂ W ∂ w
+D + ν + 2(1 − ν) ·
Ω ∂x 2 ∂x 2 ∂y 2 ∂x ∂y ∂x ∂y
2 2 2
 
∂ W ∂ w ∂ w
+ 2
ν 2 + dΩ
∂y ∂x ∂y 2
ρh3 ∂W ∂ 3 w ∂W ∂ 3Department ∂ 2 w of Technology Roorkee
ZZ   ZZ  
Manish Shrikhande [email protected] w of Earthquake Engineering Indian Institute
+Computational Structural Dynamics +
Finite Element Method and dΩ − W q − ρh dΩ
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–16

The Weak Form
or, in the operator form:
ρh3 ∂ 3 w
Z   Z Z
∂W ∂W
0 = W Qn − 2 ∂n
dΓ − · M nn dΓ − · Mns dΓ
Γ 12 ∂t Γ ∂n Γ ∂s
12 ∂ 2 w
 
2 2
ρh3 h ∂t
ZZ Z Z ZZ
T T  ∂3w 
+ (L∇W ) D̄(L∇w ) dΩ + (∇W )  ∂t 2 ∂x  dΩ − Wq dΩ
Ω 12 Ω ∂3w Ω
∂t 2 ∂y

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–16

The Weak Form
or, in the operator form:
ρh3 ∂ 3 w
Z   Z Z
∂W ∂W
0 = W Qn − 2 ∂n
dΓ − · M nn dΓ − · Mns dΓ
Γ 12 ∂t Γ ∂n Γ ∂s
12 ∂ 2 w
 
2 2
ρh3 h ∂t
ZZ Z Z ZZ
T T  ∂3w 
+ (L∇W ) D̄(L∇w ) dΩ + (∇W )  ∂t 2 ∂x  dΩ − Wq dΩ
Ω 12 Ω ∂3w Ω
∂t 2 ∂y

The constitutive matrix D̄ is given by:

 
1 ν 0
Eh3 ν
D̄ = 1 0 
2(1 − ν 2 ) 1−ν
0 0 2
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–16

The Weak Form
or, in the operator form:
ρh3 ∂ 3 w
Z   Z Z
∂W ∂W
0 = W Qn − 2 ∂n
dΓ − · M nn dΓ − · Mns dΓ
Γ 12 ∂t Γ ∂n Γ ∂s
12 ∂ 2 w
 
2 2
ρh3 h ∂t
ZZ Z Z ZZ
T T  ∂3w 
+ (L∇W ) D̄(L∇w ) dΩ + (∇W )  ∂t 2 ∂x  dΩ − Wq dΩ
Ω 12 Ω ∂3w Ω
∂t 2 ∂y

The constitutive matrix D̄ is given by: Three primary variables can be identified

1 ν 0
 from the boundary terms of the weak form,
Eh3 ν namely, the normal deflection (w ), and two
D̄ = 1 0 
2(1 − ν 2 ) 1−ν orthogonal rotations (∂w /∂n and ∂w /∂s).
0 0 2
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–17

Boundary Conditions
I Three pairs of primary and secondary variables are too many for a differential
equation whose highest derivative order is four. Ideally there should be only two
boundary terms instead of the three boundary terms.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–17

Boundary Conditions
I Three pairs of primary and secondary variables are too many for a differential
equation whose highest derivative order is four. Ideally there should be only two
boundary terms instead of the three boundary terms.
I The three boundary terms can be reduced to just two independent conditions.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–17

Boundary Conditions
I Three pairs of primary and secondary variables are too many for a differential
equation whose highest derivative order is four. Ideally there should be only two
boundary terms instead of the three boundary terms.
I The three boundary terms can be reduced to just two independent conditions.
I Assuming that the plate boundary is defined by a curve that joins two points,
namely A and B, the last boundary integral can be written as:
Z Z
∂W ∂Mns
· Mns dΓ = − W dΓ + [W · Mns ]AB
Γ ∂s Γ ∂s

where, [W · Mns ]AB represents the evaluation of limits as [W · Mns ]B − [W · Mns ]A ,

which vanishes for plates with smooth boundary such as circular disks.
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Kirchhoff’s Theory for Thin Plates–18

Boundary Conditions

I With this substitution, the boundary terms of the weak form can be rearranged as:

ρh3 ∂ 3 w
Z   Z
∂W
W Vn − 2
dΓ − · Mnn dΓ
Γ 12 ∂t ∂n Γ ∂n

where, Vn (= Qn + ∂W ∂s ) is the effective transverse shear and is referred to as

ns

Kirchhoff free edge condition and the corner uplift forces are assumed to vanish.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–1

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–1

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–1

I From the weak form, the primary variables for
interpolation may be identified as w , ∂w ∂w
∂x , and ∂y :
suggesting a C 1 continuous interpolation model.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–1

I From the weak form, the primary variables for
interpolation may be identified as w , ∂w ∂w
∂x , and ∂y :
suggesting a C 1 continuous interpolation model.
I The earliest functional finite element model for
Kirchhoff plate bending problem is given by a four-node
rectangle with three degrees of freedom (namely, w ,
∂w ∂w
∂x , and ∂y ) at each node.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–1

I From the weak form, the primary variables for
interpolation may be identified as w , ∂w ∂w
∂x , and ∂y :
suggesting a C 1 continuous interpolation model.
I The earliest functional finite element model for
Kirchhoff plate bending problem is given by a four-node
rectangle with three degrees of freedom (namely, w ,
∂w ∂w
∂x , and ∂y ) at each node.
I Thus, a 12-term approximation for the normal
displacement may be assumed as:

w (x, y ) = a0 + a1 x + a2 y + a3 x 2 + a4 xy + a5 y 2 + a6 x 3
+a7 x 2 y + a8 xy 2 + a9 y 3 + a10 x 3 y + a11 xy 3
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–1

I From the weak form, the primary variables for
interpolation may be identified as w , ∂w ∂w
∂x , and ∂y : Y
suggesting a C 1 continuous interpolation model.
y4
I The earliest functional finite element model for 4 3

Kirchhoff plate bending problem is given by a four-node

Ωe
rectangle with three degrees of freedom (namely, w ,
∂w ∂w
∂x , and ∂y ) at each node.
y
I Thus, a 12-term approximation for the normal 1 1 2

displacement may be assumed as: x1 x2 X

2
w (x, y ) = a0 + a1 x + a2 y + a3 x + a4 xy + a5 y + a6 x 2 3 A four-node rectangular
plate element
+a7 x 2 y + a8 xy 2 + a9 y 3 + a10 x 3 y + a11 xy 3
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–2

I The required interpolation functions may be obtained by using the method of PC−1 .

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–2

I The required interpolation functions may be obtained by using the method of PC−1 .
I The interpolated displacement field within the element is then given by:
w (e) (x, y ) = Pa = PC−1 w(e) = Nw(e)
where, w(e) (= [w1 , w,x1 , w,y 1 , . . . , w4 , w,x4 , w,y 4 ]T ) is the 12 × 1 vector of nodal degrees of
freedom for the rectangular element.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–2

I The required interpolation functions may be obtained by using the method of PC−1 .
I The interpolated displacement field within the element is then given by:
w (e) (x, y ) = Pa = PC−1 w(e) = Nw(e)
where, w(e) (= [w1 , w,x1 , w,y 1 , . . . , w4 , w,x4 , w,y 4 ]T ) is the 12 × 1 vector of nodal degrees of
freedom for the rectangular element.
I The curvatures required for the definition of strain field may be given in terms of the
interpolated displacement field as:
 2    2  
  ∂ ∂
κxx ∂x22 0 0 w 2 0 0 1
 κyy  =  ∂   ∂w   ∂x ∂2  ∂ 
 ∂y 2 0 0  ∂x =  ∂y 2 0 0  ∂x w
∂w ∂
2κxy 0 ∂ ∂
∂y 0 ∂ ∂
∂y
∂y ∂x ∂y ∂x

= (L∇)w (x, y ) = (L∇)Nw(e) = Bw(e)

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–3

I Curvatures can be related to moments through the moment-curvature relations:
    
Mxx 3 1 ν 0 κxx
Eh
Myy  = ν 1 0   κyy  = D̄Bw(e)
2(1 − ν 2 ) 1−ν
Mxy 0 0 2 2κxy

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–3

I Curvatures can be related to moments through the moment-curvature relations:
    
Mxx 3 1 ν 0 κxx
Eh
Myy  = ν 1 0   κyy  = D̄Bw(e)
2(1 − ν 2 ) 1−ν
Mxy 0 0 2 2κxy

I Substituting into the equation of motion:

2
ρh ∂∂tw2
 
3
∂3w 
(L∇)T D̄(L∇)Nw(e) + ∇T − ρh
12 ∂t 2 ∂x  − q = 0

3
∂3w
− ρh
12 ∂t 2 ∂y

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–4

I The weak form of the Galerkin weighted residual statement of the governing
equation of motion may be given for the element as:
  T  3
∂3w
Qn − ρh

Z 1 12 ∂t 2 ∂n
ZZ
0 =  ∂  N   −Mnn
∂n
 dΓ − NT q dΩ
Γ ∂ Ω
∂s −Mns
 12 
ρh3 h2
ZZ ZZ
∂ 
+ (L∇N)T D̄(L∇N)w(e) dΩ + (∇N)T  ∂x Nẅ(e) dΩ
Ω 12 Ω ∂
∂y

or, 0 = Q(e) − q(e) + K(e) w(e) + Mẅ(e)

where, ẅ(e) denotes the 12 × 1 vector of accelerations of the primary variables at
the four nodes.
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–5

I The element stiffness, mass matrices and the vectors of equivalent nodal loads are:
 T  3
∂3w
Qn − ρh
 
Z 1 12 ∂t 2 ∂n
ZZ
Q(e) =  ∂
∂n
 N  −Mnn  dΓ ; q(e) = NT q dΩ ;
Γ ∂ Ω
∂s −Mns
 12 
ρh3 h2
ZZ ZZ
∂ 
K(e) = BT D̄B dΩ ; M(e) = (∇N)T  ∂x N dΩ
Ω 12 Ω ∂
∂y

where, B(≡ L∇N) is the curvature-displacement matrix.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–6

I The problem with this four-node rectangular plate bending element is that its
interpolated displacement field does not ensure continuity of normal derivatives
along the element interfaces.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–6

I The problem with this four-node rectangular plate bending element is that its
interpolated displacement field does not ensure continuity of normal derivatives
along the element interfaces.
I Along the edge 1 − 4 where x = constant, the variation of w is given by a cubic
polynomial in y which is uniquely defined by the nodal variables w1 and w,y 1 at
node 1 and w4 and w,y 4 at node 4.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–6

I The problem with this four-node rectangular plate bending element is that its
interpolated displacement field does not ensure continuity of normal derivatives
along the element interfaces.
I Along the edge 1 − 4 where x = constant, the variation of w is given by a cubic
polynomial in y which is uniquely defined by the nodal variables w1 and w,y 1 at
node 1 and w4 and w,y 4 at node 4.
I The normal derivative (∂w /∂x for the edge 1 − 4) also varies as cubic polynomial
in y which is not uniquely represented by the two nodal values of the normal
derivative, namely, w,x1 and w,x4 .

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–6

I The problem with this four-node rectangular plate bending element is that its
interpolated displacement field does not ensure continuity of normal derivatives
along the element interfaces.
I Along the edge 1 − 4 where x = constant, the variation of w is given by a cubic
polynomial in y which is uniquely defined by the nodal variables w1 and w,y 1 at
node 1 and w4 and w,y 4 at node 4.
I The normal derivative (∂w /∂x for the edge 1 − 4) also varies as cubic polynomial
in y which is not uniquely represented by the two nodal values of the normal
derivative, namely, w,x1 and w,x4 .
I This leads to an incompatibility of the slope of the displacement field along
normal directions in adjacent elements.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–6

I The problem with this four-node rectangular plate bending element is that its
interpolated displacement field does not ensure continuity of normal derivatives
along the element interfaces.
I Along the edge 1 − 4 where x = constant, the variation of w is given by a cubic
polynomial in y which is uniquely defined by the nodal variables w1 and w,y 1 at
node 1 and w4 and w,y 4 at node 4.
I The normal derivative (∂w /∂x for the edge 1 − 4) also varies as cubic polynomial
in y which is not uniquely represented by the two nodal values of the normal
derivative, namely, w,x1 and w,x4 .
I This leads to an incompatibility of the slope of the displacement field along
normal directions in adjacent elements.
I Similar arguments show the incompatibility of slope ∂w /∂y along the edges 1 − 2
and 3 − 4.
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–7

∂ w 2
I The second partial derivative ( ∂x∂y ) at a corner node is non-unique when
evaluated along the edges sharing the node (e.g., the edges 1–4 and 1–2 for node
∂2w
1) because the variation of ∂x∂y along the edge 1–4 is defined by the primary
variables at nodes 1 and 4 only whereas its variation along the edge 1–2 is defined
by the values at nodes 1 and 2 only. Therefore, the variation of computed
derivative along these two edges is completely independent of each other.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–7

∂ w 2
I The second partial derivative ( ∂x∂y ) at a corner node is non-unique when
evaluated along the edges sharing the node (e.g., the edges 1–4 and 1–2 for node
∂2w
1) because the variation of ∂x∂y along the edge 1–4 is defined by the primary
variables at nodes 1 and 4 only whereas its variation along the edge 1–2 is defined
by the values at nodes 1 and 2 only. Therefore, the variation of computed
derivative along these two edges is completely independent of each other.
I The interpolated displacement field is thus incompatible and violates the basic
tenets of finite element approximation for monotonic convergence. Such elements
are known as nonconforming elements.

Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics
Geometry and Kinematics Kirchhoff’s Theory for Thin Plates Finite Element Model for Kirchhoff Plate Bending

Finite Element Model for Thin Plates–7

∂ w 2
I The second partial derivative ( ∂x∂y ) at a corner node is non-unique when
evaluated along the edges sharing the node (e.g., the edges 1–4 and 1–2 for node
∂2w
1) because the variation of ∂x∂y along the edge 1–4 is defined by the primary
variables at nodes 1 and 4 only whereas its variation along the edge 1–2 is defined
by the values at nodes 1 and 2 only. Therefore, the variation of computed
derivative along these two edges is completely independent of each other.
I The interpolated displacement field is thus incompatible and violates the basic
tenets of finite element approximation for monotonic convergence. Such elements
are known as nonconforming elements.
I The element can be shown to pass the patch test—for constant moment and
constant curvature states—and the convergence is assured as the finite element
mesh is refined. The issue of incomatibility of the normal derivative is insignificant
for small element sizes as the compatibility at the nodes is always guaranteed.
Manish Shrikhande [email protected] Department of Earthquake Engineering Indian Institute of Technology Roorkee
Finite Element Method and Computational Structural Dynamics