Rangkuman Text Book (Tugas 4) untuk Mata Kuliah Pelat & Cangkang

Chapter 6
Structural Analysis with the Finite Element Method
Linear Static
Vol.2 – Beam, Plates and Shells

1
LIST URUTAN & DAFTAR KELAS (PELAT & CANGKANG 2021) :
No.​ Nama ​ NPM​ Dari Hal​ -​ Sampai Hal​
1​ Daniel Alvin Giovanni S. ​ 2006545591​ 291 -​ 294
2​ Sarah Fatihah Nugroho ​ 2006546253​ 295 -​ 298
3​ Imam Taufik Rahmadi ​ 2006545931​ 299 -​ 302
4​ Agus Mudo Prasetyo ​ 2006545414​ 303 -​ 306
5​ Arthur Reginald ​ 1706035870​ 307 -​ 311
6​ Imam Nakhrowi ​ 2006545925​ 312 -​ 316
7​ Alfetra Henoch Tandita ​ 1706036021​ 317 -​ 321
8​ Bias January Parmadi ​ 2006545572​ 322 -​ 326
9​ Vivin Novi Adryana​ 2006546373​ 327 -​ 331
10​ Asri Puspita Sari Sidabutar ​ 2006545534​ 332 -​ 336
11​ Susilo Widyatmoko​ 2006493871​ 337 -​ 341
12​ Annisa Ayu Wulandari​ 2006545515​ 342 -​ 346
13​ Nur Hafidz Hidayat ​ 2006493650​ 347 -​ 351
14​ Ryan Sulistian ​ 1706070330​ 352 -​ 356
15​ Vania Callista ​ 1706035984​ 357 -​ 361
16​ Erriman Manru ​ 2006545673​ 362 -​ 366
17​ Madelestin ​ 1706020515​ 367 -​ 371
18​ Ida Royana Tambunan​ 2006545912​ 372 -​ 376
19​ Syaiful Rachman ​ 2006546303​ 377 -​ 381
2
 CHAPTER 6

THICK/THIN PLATES
REISSNER-MINDLIN THEORY

3
 INTRODUCTION
• The so called Reissner-Mindlin plate theory assumes that the normal to the plate do not remain
orthogonal to the mid–plane after deformation, thus allowing for transverse shear deformation
effects. This allows us to use Co continuous elements.
• Unfortunately, some difficulties arise when Reissner-Mindlin elements are used for thin plate
situations due to the excessive influence of the transverse shear deformation terms.
• Reissner-Mindlin plate elements can be taken as the starting point for deriving C o continuous
thin plate elements by adequately constraining the transverse shear deformation to be zero at
selected element points.
• Reissner-Mindlin plate theory is very adequate for studying composite laminate plates for
which shear deformation effects are important.
• The simplicity of Reissner-Mindlin plate elements and their versatility for analysis of thick and
thin plates with homogeneous and composite material have contributed to their popularity for
practical applications.

Hal 291-294 | Daniel Alvin Giovanni S. (2006545591) 4

 REISSNER-MINDLIN PLATE THEORY
Reissner-Mindlin plate bending theory shares the first three assumptions of Kirchhoff plate theory:

1. In the points belonging to the middle plane (z = 0)

In other words, the points on the middle plane only move vertically.
2. The points along a normal to the middle plane have the same vertical displacement (i.e. the
thickness does not change during deformation).
3. The normal stress σz is negligible (plane stress assumption).

The fourth assumption on the rotation of the normal is different and reads as follows:

4. A straight line normal to the undeformed middle plane remains straight but not necessarily
orthogonal to the middle plane after deformation (Figure 6.1).

Hal 291-294 | Daniel Alvin Giovanni S. (2006545591) 5

 REISSNER-MINDLIN PLATE THEORY

Hal 291-294 | Daniel Alvin Giovanni S. (2006545591) 6

 REISSNER-MINDLIN PLATE THEORY
Displacement Field

identical to
Kirchhoff theory

The displacement vector is

Assumption 4 allows us to express the rotation of the normal on the plane xz and yz as

The rotations θx and θy can not be computed in terms of the deflection only and, therefore, are
treated as independent variables.

Hal 291-294 | Daniel Alvin Giovanni S. (2006545591) 7

 REISSNER-MINDLIN PLATE THEORY
Strain and Stress Fields

substitute

The non-orthogonality of the normal

vector induces non zero transverse shear
strains γxz and γyz, which coincide (in
absolute value) with the rotations ɸx
and ɸy.

Hal 291-294 | Daniel Alvin Giovanni S. (2006545591) 8

 REISSNER-MINDLIN PLATE THEORY
Strain and Stress Fields
The strain vector containing the significant strains is therefore

The strain vector of Eq.(6.6) can be expressed as

where

is the generalized strain vector.

Hal 291-294 | Daniel Alvin Giovanni S. (2006545591) 9

 The generalized bending strain vector and the transverse shear strain vector

The strain transformation matrix S in Eq.(6.7) is

Note:
• As usual in plate theory σz has been excluded from Eq.
(6.11) due to the plane stress assumption (σz = 0)
• The sign convention for the stresses

The stress vector is

stresses due to
pure bending

stresses due to
transverse shear
effects

Hal 295-298 | Sarah Fatihah Nugroho (2006546253) 10

 6.2.3 Stress-Strain Relationship
Material : homogeneous and isothermal conditions only

Starting from the constitutive equation of 3D elasticity and using the plane stress assumption (σz = 0), we can find a relationship between the
non–zero stresses and strains. For an orthotropic material with orthotropy axes 1,2,3 with 3 = z and satisfying the condition of plane
anisotropy (i.e. the plane 1,2 is a plane of material symmetry [BD4]) we can write

the bending and transverse shear stresses and strains in the principal orthotropy axes

If isotropy exists in the direction 1 (i.e. on the plane 2−z, Figure 6.3) as for fiber composites with fibers in the direction 1 covered by a matrix,
then G13 = G12

Hal 295-298 | Sarah Fatihah Nugroho (2006546253) 11

 The following relationships hold [BC,BD4,CMPW] (6.15) Combining Eqs.(6.12) and (6.15) gives the constitutive
equations for the bending and transverse shear stresses in
global axes as

The inverse relationship can be obtained simply by replacing β by −β in Eqs.

(6.15)
(6.17)

(6.16)

Hal 295-298 | Sarah Fatihah Nugroho (2006546253) 12

 Isotropic material  simply recovered for β = 0

Substituting Eq.(6.7) into (6.17) gives the relationship between the stresses at a
point across the thickness and the generalized strains as

In practice this implies modifying the shear constitutive relationship of Eq.(6.17a)

as

Hal 295-298 | Sarah Fatihah Nugroho (2006546253) 13

Hal 299-302 | Imam Taufik Rahmadi (2006545931) 14
Hal 299-302 | Imam Taufik Rahmadi (2006545931) 15
Hal 299-302 | Imam Taufik Rahmadi (2006545931) 16
Hal 299-302 | Imam Taufik Rahmadi (2006545931) 17
Hal 299-302 | Imam Taufik Rahmadi (2006545931) 18
Hal 299-302 | Imam Taufik Rahmadi (2006545931) 19
 This splitting provides a more economical way for computing the element stiffness
matrix. It also helps explaining the behaviour of the element for thin plate situations

Hal 303-306 | Agus Mudo Prasetyo (2006545414) 20

 Global equivalent nodal force vector

equivalent nodal force vector

Hal 303-306 | Agus Mudo Prasetyo (2006545414) 21

Hal 303-306 | Agus Mudo Prasetyo (2006545414) 22
 • The mesh density was higher near the plate edges in order to capture
the boundary layer
• The condition θs = 0 along an inclined axis requires transforming
the cartesian rotations to the boundary axes

The boundary conditions on the simply supported inclined side AB in the plate of Figure 6.8 are w' = θx' = 0 . The displacement transformation for all
nodes belonging to the inclined boundary line AB is written as:

• The stiffness matrix of the element adjacent to the

boundary line AB is transformed as

Hal 307-311 | Arthur Reginald (1706035870) 23

 Reissner–Mindlin plate elements suffer from the same drawbacks as Timoshenko beam elements, i.e. the numerical solution “locks” for thin
plate situations.
This defect can be observed following the same procedure as for beams (Section 2.5). Let us consider an isotropic plate of constant thickness
under nodal point loads. The global equilibrium equations are written as

the material properties and the thickness are constant

ak =

Hal 307-311 | Arthur Reginald (1706035870) 24

 • Singularity of Ks can be achieved by reduced integration, similarly as for Timoshenko beam elements
• The singularity of Ks can be anticipated by evaluating the constraint index (CI) of each element. The CI is obtained by applying Eq.(2.50) for a single quadrilateral plate
element with two adjacent edges clamped and the other two edges free. This gives (for s = 2)
j = number of free DOFs
s = number of independent relationships at each integration point
p = number of integration points

• CI ≥ 4 ensure the singularity of Ks for any mesh

• CI < 0 indicate that Ks is not singular and invalidate the element

• CI close to zero indicates that the element is not reliable and that some situations where singularity is not satisfied can be found (leading to locking)

Hal 307-311 | Arthur Reginald (1706035870) 25

 • An element has a mechanism when it can adopt a deformed shape compatible with the boundary conditions, without consuming internal work (strain energy). For this
reason mechanisms are also called zero energymodes.

• An individual element, free of external constraints, has the standard rigid body mechanisms of translation and rotation which disappear by prescribing the boundary
conditions, a plate element has three mechanisms: the deflection and two rotations (Figure 6.10).

• These mechanisms can be identified by computing the zero eigenvalues of the stiffness matrix for an unconstrained element. Each zero eigenvalue corresponds to a
mechanism whose shape is given by the corresponding eigenmode.

• Reduced integration can induce additional zero eigenvalues in the element stiffness matrix and hence originate new mechanisms in addition to the rigid body motions.

• These new mechanisms can or can not propagate themselves within a mesh. This depends on their compatibility with adjacent elements and with the boundary conditions.

• Consequently, the singularity of K(e)s must always be verified together with the existence of spurious mechanisms in the global element stiffness matrix K(e)

Hal 307-311 | Arthur Reginald (1706035870) 26

Hal 307-311 | Arthur Reginald (1706035870) 27
Hal 312-316 | Imam Nakhrowi (2006545925) 28
 Cont.

Hal 312-316 | Imam Nakhrowi (2006545925) 29

 Cont.

Hal 312-316 | Imam Nakhrowi (2006545925) 30

Hal 312-316 | Imam Nakhrowi (2006545925) 31
 Cont.

Hal 312-316 | Imam Nakhrowi (2006545925) 32

Hal 312-316 | Imam Nakhrowi (2006545925) 33
Hal 312-316 | Imam Nakhrowi (2006545925) 34
 Q = quadrilateral
H = hierarchical
9 = 9 nodal Small "spring"

Eliminated by
multiply diagonal
stiffness by (1+e)
e = 10–2 t/L

Deformed
Shapes

QL9 = nine-noded lagrange plate quadrilateral

QS8 = eight-noded serendipity plate quadrilateral
QH9 = nine-noded hierarchical plate quadrilateral
QHG = generalization of the 9-noded hierarchical plate quadrilateral
QHET = nine-noded heterosis plate quadrilateral
F = full integration ; S = selective integration

Hal 317-321 | Alfetra Henoch Tandita (1706036021) 35

 Q = quadrilateral
H = hierarchical
G = generalized

QH9
Difference : N9

Hal 317-321 | Alfetra Henoch Tandita (1706036021) 36

 e1 = e2 = e3 = 0 --------> QL9
e1 e2 e3 very large ----> QS8
e1 e2 very large ----------> QH9

Eliminated by e1 = e2 = e3 = 0,004

Hal 317-321 | Alfetra Henoch Tandita (1706036021) 37

 Q = quadrilateral
HET = heterosis

Displacement field θx θy w
Ni L = shape function Lagrangian
Ni S = shape function serendipity

Hal 317-321 | Alfetra Henoch Tandita (1706036021) 38

 e1 = e2 = 0, e3 very large ------> QHG

Hal 317-321 | Alfetra Henoch Tandita (1706036021) 39

 QS12 ---> QS8 ---> 0 spurious mechanism
QL16 ----> QL9 ----> 1 spurious mechanism

Hal 317-321 | Alfetra Henoch Tandita (1706036021) 40

Hal 317-321 | Alfetra Henoch Tandita (1706036021) 41
Hal 317-321 | Alfetra Henoch Tandita (1706036021) 42
Hal 317-321 | Alfetra Henoch Tandita (1706036021) 43
Hal 322-326 | Bias January Parmadi (2006545572) 44
Hal 322-326 | Bias January Parmadi (2006545572) 45
Hal 322-326 | Bias January Parmadi (2006545572) 46
Hal 322-326 | Bias January Parmadi (2006545572) 47
Hal 322-326 | Bias January Parmadi (2006545572) 48
Hal 322-326 | Bias January Parmadi (2006545572) 49
Hal 322-326 | Bias January Parmadi (2006545572) 50
Hal 322-326 | Bias January Parmadi (2006545572) 51
Hal 322-326 | Bias January Parmadi (2006545572) 52
Hal 322-326 | Bias January Parmadi (2006545572) 53
Hal 322-326 | Bias January Parmadi (2006545572) 54
Hal 322-326 | Bias January Parmadi (2006545572) 55
Hal 322-326 | Bias January Parmadi (2006545572) 56
Hal 322-326 | Bias January Parmadi (2006545572) 57
Hal 322-326 | Bias January Parmadi (2006545572) 58
Hal 327-331 | Vivin Novi Adryana (2006546373) 59
Hal 327-331 | Vivin Novi Adryana (2006546373) 60
Hal 327-331 | Vivin Novi Adryana (2006546373) 61
Hal 327-331 | Vivin Novi Adryana (2006546373) 62
 The global stiffness matrix is integrated with a full 3 × 3 quadrature which
eliminates any spurious mechanisms [DL,HH2,HH4].
This element is termed QQQQ-S (for Serendipity plate Quadrilateral with
Quadratic deflection, Quadratic rotations and Quadratic transverse shear strain
fields). Conditions (6.61) are satisfied for meshes of more than 2 × 2 elements
(Figure 6.17).

Hal 332-336 | Asri Puspita Sari Sidabutar (2006545534) 63

6.7.3 9-noded Lagrange plate quadrilateral with assumed quadratic transverse shear strain field (QQQQ-L)

Hal 332-336 | Asri Puspita Sari Sidabutar (2006545534) 64

 6.7.4 Sixteen DOFs plate quadrilateral (QLQL)
The interpolating fields are the following.

A full 3 × 3 quadrature is used for integrating all the stiffness

matrix terms and this eliminates the spurious mechanisms. The
element satisfies conditions (6.61) for meshes with more than one
element (Figure 6.18).
The element is termed QQQQ-L (for Lagrange plate Quadrilateral
with Quadratic deflection, Quadratic rotations and Quadratic
trasverse shear strain fields). This element was simultaneously
developed by Hinton and Huang [HH3,HH4] and Bathe and Dvorkin
[BD2]. Donea y Lamain [DL] presented a different version of the
element based in the direct derivation of expressions (6.100) in
cartesian coordinates. The performance of the QQQQ-L element is
superior to the QQQQ-S.

Hal 332-336 | Asri Puspita Sari Sidabutar (2006545534) 65

65
3. The transverse shear strains are assumed to vary linearly as in Eq.(6.86). The substitute transverse shear strain matrix is found as explained
for the QLLL element (Section 6.7.1). A 2 × 2 quadrature is used for all the terms of the stiffness matrix.

6.7.5 4-noded plate quadrilateral of Tessler-Hughes

Hal 332-336 | Asri Puspita Sari Sidabutar (2006545534) 66

66
Hal 337-341 | Susilo Widyatmoko (2006493871)
Hal 337-341 | Susilo Widyatmoko (2006493871)
Hal 337-341 | Susilo Widyatmoko (2006493871)
Hal 337-341 | Susilo Widyatmoko (2006493871)
Hal 337-341 | Susilo Widyatmoko (2006493871)
Hal 337-341 | Susilo Widyatmoko (2006493871)
Hal 337-341 | Susilo Widyatmoko (2006493871)
Hal 337-341 | Susilo Widyatmoko (2006493871)
Hal 337-341 | Susilo Widyatmoko (2006493871)
Hal 337-341 | Susilo Widyatmoko (2006493871)
Hal 337-341 | Susilo Widyatmoko (2006493871)
Hal 337-341 | Susilo Widyatmoko (2006493871)
Hal 337-341 | Susilo Widyatmoko (2006493871)
 The vector of Nodal Variables is :

Hal 342-346 | Annisa Ayu Wulandari (2006545515) 80

Hal 342-346 | Annisa Ayu Wulandari (2006545515) 81
Hal 342-346 | Annisa Ayu Wulandari (2006545515) 82
Hal 342-346 | Annisa Ayu Wulandari (2006545515) 83
 6.10 REISSNER-MINDLIN PLATE ELEMENTS BASED ON LINKED INTERPOLATIONS
A different class of Reissner-Mindlin plate elements can θx = ∂w/∂x
be derived by using an interpolation for the deflection field This favours satisfying the thin plate conditions in the thin limit
of one order higher than for the rotations. θy = ∂w/∂y
 For the 3-noded linked triangle and the 4-noded linked quadrilateral we require that

1. The deflection field along a side is defined by the rotations at the end side nodes only, in order to guarantee C◦ continuity;
2. The rotation terms must introduce quadratic expansions into the deflection field;
3. The transverse shear strains are constant along the sides.
 The following interpolation satisfies above conditions

Ni = the shape functions of the original element

ns = is the number of nodes along the element sides
lij = the length of side ij
θsi and θsj = the rotations along the tangential directions to
the k-th side with nodes i and j
Ni = quadratic shape functions vanishing at the corner nodes.

Hal 347-351 | Nur Hafidz Hidayat (2006493650) 84

 For 3-noded triangles,
For 4-noded quadrilaterals,
 The tangential rotations θsi are expressed in terms of the cartesian components at each node. For side ij
where Øij = the angle that side ij forms with the x axis.

yields the deflection field in terms of the deflection and the two cartesian rotations at the corner nodes
Subtitusikan in Eq (6.131a)
 For a 4-noded rectangle Eq.(6.131a) takes the following form for side 12 with η = 1

 Eqs.(6.134) and (6.131b) guarantee a constant shear distribution along the side. Assuming a rectangular shape with s = ξ = x gives for side 12

Hal 347-351 | Nur Hafidz Hidayat (2006493650) 85

 6.11 DISCRETE–KIRCHHOFF PLATE ELEMENTS

A family of thin plate elements can be derived by imposing the Kirchhoff

constraints (γxz = γyz = 0) at selected points within a Reissner-Mindlin element
so that the transverse shear strain energy is effectively zero. The so-called
Discrete-Kirchhoff (DK) plate elements were originally proposed by Wempner et
at. [WOK], Stricklin et al. [SHTG] and Dhatt [Dh,Dh2] as early as in 1968-70 as
a way for overcoming the C 1 continuity limitations of Kirchhoff plate theory.
Several DK plate and shell elements were subsequently developed
[BD5,6,BRI,Cr,Cr3,DMM], the most successful ones being the 3-noded DK
triangle (DKT) detailed in the next section and the “semi-loof” shell element
[IA,Ir2] (Section 7.12.5). The derivation of DK elements can be viewed as a
particular class of assumed transverse shear strain techniques, leading to the
vanishing of the transverse shearstrain energy over the element. A state of the art
on DK plate elements is presented in [BD3].
6.11.1 3-noded DK plate triangle (DKT)
The starting point is the 6-noded Reissner-Mindlin triangle of Figure 6.26a under the following constraints:

1. The rotations θx and θy vary quadraticaly over the element (12 DOFs).
2. The deflection varies as a cubic Hermite polynomial along each side ij in terms of After transformation, this gives a
total of 9 DOFs
3. A linear variation for the normal rotation θn is imposed along each side.
4. The conditions of zero transverse shear strain are imposed:

5. Only the contribution of the bending terms is taken into account for computing the element si ffness matrix, i.e. K (e)= Kb (e)

Hal 347-351 | Nur Hafidz Hidayat (2006493650) 86

  The rotation field is finally expressed in terms of the standard nine nodal DOFs as

is the nodal displacement vector

 The shape functions Nxi and Nyi are shown in Box 6.1.

Hal 347-351 | Nur Hafidz Hidayat (2006493650) 87

  Eq.(6.136) allows us to obtain the bending strain matrix B b from which the element sti ffness matrixcan be exactlycomputed using 3 point quadrature.
The bending moments are sampled at the quadrature points. Batoz [Bat,BD5] derived an explicit form for the sti ffness matrix of the DKT element.
 First the matrix is evaluated in the local axes ¯x, ¯y of Figure 6.27

 Matrices Q and S are shown in Figure 6.28.

 The global stiffness matrix is obtained by

 The bending moments are obtained at any element point by

where N3 is the linear shape function matrix for the 3-noded triangle [Bat,BD5].
Figure 6.28 shows the explicit form for the local stiffness matrix of the DKT element.

Hal 347-351 | Nur Hafidz Hidayat (2006493650) 88

Hal 352-356 | Ryan Sulistian (1706070330) 89
Hal 352-356 | Ryan Sulistian (1706070330) 90
Hal 352-356 | Ryan Sulistian (1706070330) 91
Hal 352-356 | Ryan Sulistian (1706070330) 92
 PATCH TESTS FOR REISSNER-MINDLIN PLATE ELEMENTS
Patch Test B
ABSENCE OF SPURIOUS NODES AND RIGID BODY MOTIONS
• Displacement field

Where a,b, and c are arbitrary numbers

• Curvatures and transverse shear strain must be zero everywhere in the patch

PURE BENDING STATE

• Displacement field

Where a,b, and c are arbitrary numbers

• Constant curvature field and zero transverse shear strain field must be
obtained

Hal 357-361 | Vania Callista (1706035984) 93

 PATCH TESTS FOR REISSNER-MINDLIN PLATE ELEMENTS
LESS SEVERE PATCH TEST
• Displacement field

Where a,b, and c are arbitrary numbers

• α = 0 in order to make Ks a null matrix element

CONSTANT TRANSVERSE SHEAR STRAIN FIELD AND ZERO CURVATURE FIELD

• Displacement field

Where a,b, and c are arbitrary numbers

Patch Test C
• Stability of element can be obtained by computing the rank of the stiffness matrix in element
patches with the minimum number of DOFs prescribed
Hal 357-361 | Vania Callista (1706035984) 94
 EXAMPLES
PERFORMANCE OF SOME PLATE ELEMENTS BASED ON ASSUMED TRANSVERSE SHEAR STRAIN FIELDS

Questions

• a/t ratio 10 (thick)

• a/t ratio 100 (thin)
• Load: uniform load and a central point load
• Elements studied: QLLL, QLQL, TLQL, and TLLL
• FEM results for the deflection at the center of the middle plane obtained using a mesh of 40x40x6 8-noded
hexahedra
• For the point load case the analytical value of the deflection under the load given by thick plate theory is
infinity, so, results for thick plates are compared at the mid-point along a central line in this case

Hal 357-361 | Vania Callista (1706035984) 95

 Answers

Hal 357-361 | Vania Callista (1706035984) 96

 EXAMPLES
SIMPLE SUPPORTED PLATE UNDER UNIFORM LOAD-ADAPTIVE SOLUTION

Questions

Hal 357-361 | Vania Callista (1706035984) 97

 Figure 6.32 shows results for the same SS squate plate Table 6.3 shows results for the deflection and the bending
obtained with the TLLL element for the SS (hard) case moment at the center for thick and thin SS and clamped
under uniform loading. The results plotted are the central circular plates obtained with the TLLL element. The
deflection and the central bending moment. The performance of this simple element is remarkably good
performance is good for thick and thin situations and with less than 10% error versus the reference solution for
results are less sensitive to the mesh orientation than for meshes of 64 elements in all cases.
the TLQL element.
Hal 362-366 | Erriman Manru (2006545673) 98
 Figure 6.33 shows results for a moderately thick clamped circular plate using the TLQL, QLLL and QLQL elements.
Graphics show the convergence of the deflection and the bending moment at the center toward the analytical solution
(Appendix H and [BD5,TW]) and the distribution of the radial bending moment and the radial shear force along a
radial line. All elements behave well. Note the accuracy of the QLLL element.

Hal 362-366 | Erriman Manru (2006545673) 99

 Table 6.4 shows the convergence of the central
deflection at the two free corners in cantilever skew
plates under uniform load and different skew angles.
Results obtained with the TLLL element are shown
as well as those obtained with the DRM triangle
[ZTPO,ZT2] (an equivalent of the TLQL of Section
6.8.2) and the EL1 triangle of Figure 6.25b for
comparison purposes. The performance of the TLLL
element is again noticeable.

Hal 362-366 | Erriman Manru (2006545673) 100

 Figure 6.34 shows the geometry and material properties of the plate and the initial unstructured mesh of 68 TLQL
elements. The weak support conditions yield a zero torque M xy along the supporting sides. This originates a boundary
layer for the torque Mxy in the vecinity of these sides [OCK]. The percentage of admissible global error in the energy
norm is η = 5%.We have taken p = 1 and d = 2 in the expression for the mesh refinement parameter β (e) for each AMR
strategy (Section 9.9.4 of [On4]). The computed value of the global error parameter for the initial mesh is ξ g = 2.7074.
The aim is to reduce this parameter to a value close to unity by adaptive mesh refinement.

Hal 362-366 | Erriman Manru (2006545673) 101

 Figure 6.34 shows the sequence of the refined
meshes obtained using two mesh optimality
criteria: one based on the equidistribution of the
global error (Section 8.9.4.1 of [On4]) and the
based in the equidistribution of the error density
in the mesh (Section 8.9.4.2 of [On4]). The
number of elements and the value of ξg for each
mesh are also shown. The criterion based on the
equidistribution of the error density leads to
meshes with a larger number of elements. Its
advantage, however, is that it captures very well
the boundary layer for the torque Mxy (Figure
6.35).

Hal 362-366 | Erriman Manru (2006545673) 102

 6.14.3 Effect of shear deformation in a plate simply
supported at three edges under a line load acting on the
free edge
The plate is displayed in Figure 6.36, where the
geometry, the boundary conditions and the uniformly
distributed loading along the edge is also shown. The
edge loading induces high transverse shear stresses and
strains near the edge, which play an important role in
total deflection value for thick situations. This example
was analyzed in [On] using 20 Reissner-Mindlin linear
strip elements (Chapter 11) and in [Bl] using a mesh of
10 × 10 QLLL plate elements. In both cases half of the
plate was studied due to symmetry, with a finer mesh in
the vicinity of the free edge. Both experimental [AR]
and analytical results (based on Kirchhoff thin plate
theory) [TW] are available.

Hal 362-366 | Erriman Manru (2006545673) 103

Hal 362-366 | Erriman Manru (2006545673) 104
 Thin Plate Solution
(Kirchoff) > Thick
Plate (Reissner-
Mindlin)

• Differences <<<< as
moving away from
the edge and center

Hal 367-371 | Madelestin (1706020515) 105

Hal 367-371 | Madelestin (1706020515) 106
Hal 367-371 | Madelestin (1706020515) 107
Hal 367-371 | Madelestin (1706020515) 108
Hal 367-371 | Madelestin (1706020515) 109
Hal 372-376 | Ida Royana Tambunan (2006545912) 110
 The tangential rotation is expressed in terms of the nodal
deflection and shear angle variables as

Hal 372-376 | Ida Royana Tambunan (2006545912) 111

Hal 372-376 | Ida Royana Tambunan (2006545912) 112
Hal 372-376 | Ida Royana Tambunan (2006545912) 113
Hal 372-376 | Ida Royana Tambunan (2006545912) 114
Hal 377-381 | Syaiful Rachman (2006546303) 115
Hal 377-381 | Syaiful Rachman (2006546303) 116
Hal 377-381 | Syaiful Rachman (2006546303) 117
  Kirchhoff plate element have obviously limitations to reproduce the behaviour of thick plates
 Reissner-Mindlin plate elements are superior as they are applicable to thick and thin situation

Hal 377-381 | Syaiful Rachman (2006546303) 118

Hal 377-381 | Syaiful Rachman (2006546303) 119
 Important difference between thick and thin plates where point loads are involved:
 In the thin case the deflection w remains finite under the load.
 However, transverse shear deformation effects lead to an infinite displacements under the
load for thick plates (as indeed 3D elasticity theory predicts).
 In finite element approximations one always predicts a finite displacement at point load
locations with the magnitude increasing without limit as a mesh is refined near the point
load
 Thus, it is meaningless to compare the deflections at point load locations.
 It is recommended to compare the total strain energy for such situations [ZT2].

Hal 377-381 | Syaiful Rachman (2006546303) 120

 6.17 CONCLUSIONS
• Reissner-Mindlin plate theory is the basis for systematically deriving C0 continuous plate elements
which include transverse shear strain effects
• Practically, the only drawback is the appearance of shear locking for thin plate situations
 The reduced integration of the transverse shear stiffness terms is a simple and efficient procedure
for eliminating shear locking, although it can introduce spurious mechanisms which can pollute
the solution in some cases
 The assumed transverse shear strain technique is a more consistent approach for designing robust
locking-free plate elements (note: good performance of the lower order QLLL and TLLL)

• Reissner-Mindlin plate elements are also the starting point for deriving Discrete-Kirchoff thin
plate elements, while keeping all the features of the C0 continuous formulation
• The merits of the Reissner-Mindlin plate formulation will show clearer when dealing with composite
plates and shells in the subsequent chapters

Hal 377-381 | Syaiful Rachman (2006546303) 121

TERIMA
KASIH

122