You are on page 1of 10

University of Alexandria Dept.

of Naval Architecture & Marine Engineering


Faculty of Engineering Instructor: H. W. Leheta

CHAPTER 4 INTRODUCTION TO THE FINITE ELEMENT


METHOD (FEM)

The basic concept of the finite element method was originally developed mainly from
matrix stiffness analysis, where the structure is represented as an assemblage of
structural elements connected at a discrete number of nodes.
In FEM, a continuum structure (including plates and shells) is divided into a number
of artificial “finite” elements. The internal displacements of the elements must be
related to the nodal displacements. The problem then becomes a discrete instead of
a continuous problem.
The finite element representation must satisfy everywhere conditions of equilibrium
and compatibility.
FEM is very broad and powerful and has a great variety of applications-both
structural and nonstructural.
A number of general purpose finite element computer programs, e.g. NASTRAN,
ANSYS, ABAQUS, etc. for structural applications are available.

4.1 Dividing structure into finite elements

This is what we call structural idealization or modeling. The accuracy of the method
increases with the number of elements used, however the computer time is also
increased and hence the cost.

Types of structural elements used in marine structures:

1-Truss (bar) element (2-D or 3-D)


2- Beam element (2-D or 3-D)
3- Membrane (plane stress) element: this is a thin plate subjected to in-plane forces.
Usually, the material is isotropic but it can be orthotropic.
4- Plain strain element: like the previous element except that it may carry stress
normal to the element plane.
5- Bending plate element: it is a thin plate element subjected to bending, both
isotropic and orthotropic elements may be used.
6- Thin shell element: to represent singly or doubly curved thin shells with both
stretching and bending.
7- Axisymmetric thin shell element: used to represent the axisymmetric (possessing
axial symmetry in cylindrical coordinates) shell structure in which both bending and
in-plane forces occur.
8- The 3-D solid element: the most common are the tetrahedron (four corner nodes)
and the hexahedron (eight corner nodes).
9- Thick shell element: a 3-D solid element with reduced dimension in the thickness
direction.

4-1
University of Alexandria Dept. of Naval Architecture & Marine Engineering
Faculty of Engineering Instructor: H. W. Leheta

4-2
University of Alexandria Dept. of Naval Architecture & Marine Engineering
Faculty of Engineering Instructor: H. W. Leheta

4.2 Use of displacement functions to derive the element local matrix

4.2.1 Beam element (flexure only)

Let us consider a flexure only beam element:

v1 v2

θ1 θ2
1 2

 v1 
v ' 
dv  
θ= = v' δ =  1
dx v 2 
v' 2 

STEP 1: SELECT SUITABLE DISPLACEMENT FUNCTION:

A displacement function is a function that defines the state of displacement at all


points within the element in terms of the nodal degrees of freedom. Assume:
v (x ) = C1 + C2 .x + C3 .x 2 + C4 .x 3
which is cubic polynomial with 4 coefficients since we have four degrees of freedom.

In matrix form:
v (x ) = H (x ).C
where: H (x ) = 1 x { x 2
x 3
} and
C1 
C 
 
C =  2
C3 
C4 

v' (x ) = C2 + 2C3 x + 3C4 x 2

We define δ(x), the generalized displacement vector:


C1 
 
 v (x )  1 x x 2 x 3  C2 
δ(x ) =  = 2 
. 
v' (x ) 0 1 2 x 3 x  C3 
C4 

STEP 2: RELATE GENERAL DISPLACEMENTS WITHIN THE ELEMENT TO


NODAL DISPLACEMENTS

At node 1, x = 0 hence v1 = C1 and v’1 = C2

4-3
University of Alexandria Dept. of Naval Architecture & Marine Engineering
Faculty of Engineering Instructor: H. W. Leheta

At node 2, x = L hence
v 2 = C1 + C2 L +C 3 L2 + C4 L3

v' 2 = C2 + 2C3 L + 3C4 L2


In matrix form:
 v1  1 0 0 0  C1 
v'  0 0  C2 
 1  1 0
 = ⋅ 
 v 2  1 L L2 L3  C3 
v' 2  0 
1 2 L 3 L2  C4 

or δ = A ⋅ C ⇒ C = A −1 ⋅ δ

 1 0 0 0 
 0 1 0 0 
 3 2 3 1
A −1
= − 2 − − 
 L L L L
 2 1 2 1 
− 3
 L3 L2 L L2 

v (x ) = H (x ) ⋅ C = H (x ) ⋅ A −1 ⋅ δ = N(x ) ⋅ δ

N(x) is known as the shape function:

 1 0 0 0 
 0 1 0 0 
 3
N (x ) = H (x ) ⋅ A −1
=1 x[ x 2
]
x 3 ⋅ − 2 −
2 3 1
− 
 L L L2 L
 2 1 2 1 
 L3 − 3
L2 L L2 

 x2 x3 x2 x3 x2 x3 x2 x3 
= 1 − 3 2 + 2 3 x −2 + 3 − 2 − + 
 L L L L2 L2 L3 L L2 

STEP 3: EXPRESS THE INTERNAL DEFORMATION IN TERMS OF THE NODAL


DISPLACEMENTS

Curvature ψ =d2v/dx2 =v’’ (for small deflections)

v' ' (x ) = 2C3 + 6C4 x

In matrix form:

4-4
University of Alexandria Dept. of Naval Architecture & Marine Engineering
Faculty of Engineering Instructor: H. W. Leheta

C1 
C 
 
v' ' (x ) = [0 0 2 6 x ] ⋅  2  = [0 0 2 6 x ] ⋅ A −1 ⋅ δ = B ⋅ δ
C3 
C4 

 1 0 0 0 
 0 1 0 0 
 3  6 12 x 2 6x 
B = [0 0 2 6 x ] ⋅ − 2
2 3 1 4 6x 6 12 x
− −  = − 2 + 3 − + − 3 − +
 L L L2 L  L L L L2 L2 L L L2 
 2 1 2 1 
 L3 − 3
L2 L L2 

STEP 4: EXPRESS THE INTERNAL FORCE IN TERMS OF THE NODAL


DISPLACEMENTS USING THE ELEMENT’S LAW OF ELASTIC BEHAVIOR

Force-deformation relationship:

M( x )
v' ' (x ) = ⇒ M (x ) = EI .v' ' ( x ) = EI.B.δ is the internal bending moment
EI

 v1 
v ' 
 6 12 x 4 6x 6 12 x 2 6x   
∴ M ( x ) = EI  - 2 + 3 − + − 3 − + 2 ⋅ 1
 L L L L2 L2 L L L  v 2 
v' 2 

At the nodes:

 6 4 6 2   v1 
 ( ) − L2 − −  v ' 

M 0
 = EI ⋅  6 L L2 L  1
M (L ) 2 6 4  v 2 
 − 2  
 L2 L L L  v' 2 

STEP 5: OBTAIN THE ELEMENT STIFFNESS MATRIX ke BY RELATING NODAL


FORCES TO NODAL DISPLACEMENTS

Principle of virtual work:

f = vector of nodal forces


δ =vector of nodal displacements

Impose virtual displacements on the beam δ :

4-5
University of Alexandria Dept. of Naval Architecture & Marine Engineering
Faculty of Engineering Instructor: H. W. Leheta

 v1   fy 1 
  M 
v' 1   
δ =   , virtual displacements f =  1  , real forces
v 2   fy 2 
v' 2  M2 
 

T
External V.W.: W E = δ .f = v 1 .fy1 + v' 1 .M1 + v 2 .fy 2 + v' 2 .M2

M (x )
L L
Internal V.W.: W I = ∫ .M (x ).dx = ∫ v' ' ( x ).M (x ).dx
0 EI 0

v' ' (x ) = B ⋅ δ M ( x ) = EI ⋅ B ⋅ δ

v ' ' ( x ) ⋅ M ( x ) = δ ⋅ B T ⋅ EI ⋅ B ⋅ δ
T

L L
W I = ∫ δ ⋅ B ⋅ EI ⋅ B ⋅ δ ⋅ dx = δ ∫B
T T
T T
⋅EI ⋅ B ⋅ dx ⋅ δ = W E ⇒
0 0

L
f = ∫ B T ⋅ EI ⋅ B ⋅ dx . δ but f = ke ⋅ δ
0

 6 12 x 
− L2 + L3 
 4 6x 
L  − + 
 L L2   6 12 x 2 6x 
L
4 6x 6 12 x
∴ k = ∫ B .EI ⋅ B ⋅ dx = EI ⋅ ∫ 
e T
⋅ − + 3 − + − 3 − + ⋅ dx
0 
6 12 x   L2 L L L2 L2 L L L2 
0 − 
 L L 
2 3

 2 6x 
 − L + L2 

General form:

ke = ∫B ⋅ D ⋅ B ⋅ dV
T

ve

where
ve is the volume of the element
B is the strain matrix, such that ε = B ⋅ δ
D is the material stiffness matrix, such that σ = D ⋅ ε

4-6
University of Alexandria Dept. of Naval Architecture & Marine Engineering
Faculty of Engineering Instructor: H. W. Leheta

4.2.2 Constant stress triangle CST (plane stress)

Consider a flat triangular-shaped element of constant thickness t and isotropic


material properties. Almost any two-dimensional shape can be represented by an
assemblage of triangles.
v3
y
3 u3 v(x,y)
v2

u(x,y)
v1 u2
2
1 u1
x

f = ke ⋅ δ

The same five steps used for the beam element can be used to derive the stiffness
matrix for a triangular element.

STEP 1: SELECT SUITABLE DISPLACEMENT FUNCTION

u1  fx1 
v  f 
 1  y1 
u  fx 
δ =  2 f = 2
v 2  fy 2 
u3  fx 3 
   
v 3  fy 3 

u (x , y ) = C1 + C2 x + C3 y v ( x, y ) = C 4 + C 5 x + C 6 y

𝐶𝐶1
⎡𝐶𝐶 ⎤
⎢ 2⎥
𝑢𝑢 1 𝑥𝑥 𝑦𝑦 0 0 0 ⎢𝐶𝐶3 ⎥
𝛅𝛅(x, y) = � � = � �.
𝑣𝑣 0 0 0 1 𝑥𝑥 𝑦𝑦 ⎢𝐶𝐶4 ⎥
⎢𝐶𝐶5 ⎥
⎣𝐶𝐶6 ⎦
or

δ(x,y)=H(x,y).C

4-7
University of Alexandria Dept. of Naval Architecture & Marine Engineering
Faculty of Engineering Instructor: H. W. Leheta

STEP 2: RELATE THE GENERAL DISPLACEMENT WITHIN THE ELEMENT TO


THE NODAL DISPLACEMENTS

𝜹𝜹𝟏𝟏 = 𝜹𝜹(𝑥𝑥𝟏𝟏 , 𝑦𝑦1 ) = 𝐇𝐇(𝑥𝑥1 , 𝑦𝑦1 ). 𝐂𝐂

1 𝑥𝑥1 𝑦𝑦1 0 0 0
=� � 𝐂𝐂
0 0 0 1 𝑥𝑥1 𝑦𝑦1

𝜹𝜹𝟐𝟐 = 𝜹𝜹(𝑥𝑥𝟐𝟐 , 𝑦𝑦2 ) = 𝐇𝐇(𝑥𝑥2 , 𝑦𝑦2 ). 𝐂𝐂

1 𝑥𝑥2 𝑦𝑦2 0 0 0
=� � 𝐂𝐂
0 0 0 1 𝑥𝑥2 𝑦𝑦2

𝜹𝜹𝟑𝟑 = 𝜹𝜹(𝑥𝑥𝟑𝟑 , 𝑦𝑦3 ) = 𝐇𝐇(𝑥𝑥3 , 𝑦𝑦3 ). 𝐂𝐂

1 𝑥𝑥3 𝑦𝑦3 0 0 0
=� � 𝐂𝐂
0 0 0 1 𝑥𝑥3 𝑦𝑦3

Therefore

𝜹𝜹𝟏𝟏
𝜹𝜹 = �𝜹𝜹𝟐𝟐 �
𝜹𝜹𝟑𝟑

𝜹𝜹 = 𝑨𝑨𝑨𝑨

𝑪𝑪 = 𝑨𝑨−𝟏𝟏 𝜹𝜹

𝜹𝜹(𝑥𝑥, 𝑦𝑦) = 𝑯𝑯𝑨𝑨−𝟏𝟏 𝜹𝜹

STEP 3: EXPRESS THE INTERNAL DEFORMATION (STRAIN) IN TERMS OF THE


NODAL DISPLACEMENTS

ε x 
  ∂u ∂v ∂u ∂v
ε( x , y ) = ε y  εx = εy = γ = +
γ  ∂x ∂y ∂y ∂x
 

𝜕𝜕
𝜀𝜀𝑥𝑥 = (𝐶𝐶 + 𝐶𝐶2 𝑥𝑥 + 𝐶𝐶3 𝑦𝑦) = 𝐶𝐶2
𝜕𝜕𝜕𝜕 1
𝜕𝜕
𝜀𝜀𝑦𝑦 = (𝐶𝐶 + 𝐶𝐶5 𝑥𝑥 + 𝐶𝐶6 𝑦𝑦) = 𝐶𝐶6
𝜕𝜕𝜕𝜕 4

𝜕𝜕 𝜕𝜕
𝛾𝛾 = (𝐶𝐶1 + 𝐶𝐶2 𝑥𝑥 + 𝐶𝐶3 𝑦𝑦) + (𝐶𝐶 + 𝐶𝐶5 𝑥𝑥 + 𝐶𝐶6 𝑦𝑦) = 𝐶𝐶3 + 𝐶𝐶5
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 4

4-8
University of Alexandria Dept. of Naval Architecture & Marine Engineering
Faculty of Engineering Instructor: H. W. Leheta

Hence

𝐶𝐶1
⎧𝐶𝐶 ⎫
𝜀𝜀𝑥𝑥 2
𝐶𝐶2 0 1 0 0 0 0 ⎪𝐶𝐶 ⎪
𝜺𝜺(𝑥𝑥, 𝑦𝑦) = �𝜀𝜀𝑦𝑦 � = � 𝐶𝐶6 � = �0 0 0 0 0 1� 3
𝐶𝐶
𝛾𝛾 𝐶𝐶3 + 𝐶𝐶5 0 0 1 0 1 0 ⎨ 4⎬
⎪𝐶𝐶5 ⎪
⎩𝐶𝐶6 ⎭

𝜺𝜺(𝑥𝑥, 𝑦𝑦) = 𝑮𝑮𝑮𝑮 = 𝑮𝑮𝑨𝑨−𝟏𝟏 𝜹𝜹

A strain coefficient matrix is defined:

𝑩𝑩 = 𝑮𝑮𝑨𝑨−𝟏𝟏

𝜺𝜺(𝑥𝑥, 𝑦𝑦) = 𝑩𝑩. 𝜹𝜹

STEP 4: EXPRESS THE INTERNAL FORCE (STRESS) IN TERMS OF THE


NODAL DISPLACEMENTS, USING THE ELEMENT’S LAW OF ELASTIC
BEHAVIOR

σ x 
 
σ = σ y 
τ 
 

For plane stress, the relationship between stress and strain is:

𝜎𝜎𝑥𝑥 𝜈𝜈 𝜎𝜎𝑦𝑦
𝜀𝜀𝑥𝑥 = −
𝐸𝐸 𝐸𝐸
−𝜈𝜈𝜎𝜎𝑥𝑥 𝜎𝜎𝑦𝑦
𝜀𝜀𝑦𝑦 = +
𝐸𝐸 𝐸𝐸
𝜏𝜏 2(1 + 𝜈𝜈)
𝛾𝛾 = = 𝜏𝜏
𝐺𝐺 𝐸𝐸

In matrix notation, the relationship is:

1 1 −𝜈𝜈 0 𝜎𝜎𝑥𝑥
𝜺𝜺(𝑥𝑥, 𝑦𝑦) = �−𝜈𝜈 1 0 𝜎𝜎
� � 𝑦𝑦 �
𝐸𝐸 0 0 2(1 + 𝜈𝜈) 𝜏𝜏

4-9
University of Alexandria Dept. of Naval Architecture & Marine Engineering
Faculty of Engineering Instructor: H. W. Leheta

The inverse of the relationship is:

1 𝜈𝜈 0 𝜀𝜀𝑥𝑥
𝐸𝐸 𝜈𝜈 1 0
𝝈𝝈(𝑥𝑥, 𝑦𝑦) = � 𝜀𝜀
1 − 𝜈𝜈 2 1 − 𝜈𝜈 � � 𝑦𝑦 �
0 0 𝛾𝛾
2

or

𝝈𝝈(𝑥𝑥, 𝑦𝑦) = 𝑫𝑫𝜺𝜺(𝑥𝑥, 𝑦𝑦)

 
1 ν 0 
E
where D= ν 1 0 
1 −ν 2  1 −ν 
0 0 
 2 

STEP 5: OBTAIN THE ELEMENT STIFFNESS MATRIX BY RELATING NODAL


FORCES TO NODAL DISPLACEMENTS

𝒌𝒌𝒆𝒆 = � 𝑩𝑩𝑻𝑻 . 𝑫𝑫. 𝑩𝑩. 𝑑𝑑𝑣𝑣𝑣𝑣𝑣𝑣


𝑣𝑣𝑣𝑣𝑣𝑣

𝒌𝒌𝒆𝒆 =BT.D.B.A123.t

∫ 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = 𝐴𝐴123 . 𝑡𝑡

A123 = the area of the triangle.

Other important elements include: the linear strain rectangle (LSR) element and the
constant shear stress rectangle (CSSR) element.

For more on the FEM, refer to the book:

Owen F. Hughes and Jeom Kee Paik, Ship Structural Analysis and Design, SNAME,
2010.

4-10

You might also like