Professional Documents
Culture Documents
z x
As you recall, an isotropic body can have normal stresses acting on each surface: x, y, z When the only normal stress is x this causes a strain along the x- axis according to Hookes Law
x z z y
x =
x
E
x x
y
x x
+ z
x =
x
E
y
E
z
E
y
E +
z
E
y = z =
x
E
y
E
z
E
x
E
y
E
z
E
x =
y yx
xy = G xy = 2G xy
z x
yz = G yz = 2G yz zx = G zx = 2G zx
x z z y
xy
Stress-Strain Relationships
0 0 0 1 0 0 0 x x 1 1 0 0 0 y y 1 2 E z 0 0 0 0 0 z = 2 xy (1+)(12) xy 12 0 0 0 0 0 yz yz 2 zx zx 12 0 0 0 0 0 2
3D Stress-Strain Matrix
0 0 0 1 1 0 0 0 0 0 0 1 1 2 0 E 0 0 0 0 [ D] = 2 1 1 2 + ( )( ) 1 2 0 0 0 0 0 2 1 2 0 0 0 0 0 2 E Note : G = 2(1 +)
Strain-Displacement
u x = x v y = y u v xy = + y x u w xz = + z x
Two-dimensional Elements
1. 2. 3. 4. 5.
Thin 2D elements . Two coordinates to define position. Elements connected at common nodes and/or along common edges. Nodal compatibility enforced to obtain equilibrium equations Two basic types
1. 2. Plane stress Plane Strain
{} = [D]{ }
or {} = [D]1{}
Modeled as
Plane Stress
Plane Strain
Axisymmetric
Plane Stress
Plane Stress, is defined to be a state of stress in which the normal and shear stresses perpendicular to the x-z plane are zero, the ythickness is very small, and the constraints (Rx,Rz) and loads act only in the x-z plane and throughout the y-thickness.
F2 F1 x
z = 0 xz= 0, yz=0 thickness, y dimension, is very small compared to x and z dimensions Loads act only in the x-z plane and throughout the y-thickness
T
x
Stress-Strain Relationships
For Plane-Stress:
z = 0, xz= 0, yz=0
0 0
(1 2 ) 2
1 x 1 y 1 E 0 = 0 0 0 xy (1 + )(1 2 ) 0 0 0 0 0 0 0 0
0 0 0 0
(1 2 ) 2
0 0
0 x 0 y 0 z 0 xy 0 yz (1 2 ) zx 2
yz = 0 zx = 0 z = x + y 1
2 2 0 1 1 1 x x 2 2 E = 1 0 y y 1 1 (1+)(12) xy (12) xy 0 0 2
1 0
{ } = [ D]{ }
or { } = [ D]1{ }
Plane Strain
Plane Strain, is defined to be a state of strain in which the normal strain in the y-direction, y and the shear strains, xy and yz are zero. Note, the y-thickness of the body is very large, and constraints and loads act in x-z plane throughout thickness. z = 0, xz= 0, yz =0 The thickness, y-dimension of the body is y
r Fy
very large (infinite). Loads and constraints act only in the x-z plane through a unit y-thickness
r Fx
Forces are defined as force per unit ylength A plane stress state, where y is a very large value, does not approximate plane strain conditions!
Stress-Strain Relationships
For Plane-Strain: z = 0, xz= 0, yz =0
0 0 0 1 x x 1 0 0 0 y y 1 0 0 0 E z 0 (1 2) = 0 0 0 0 0 2 + ( 1 )( 1 2 ) xy xy (1 2) yz 0 0 0 0 0 0 2 (1 2) 0 xz 0 0 0 0 0 2
yz = 0 zx = 0 z = 0
1 0 x x E 1 0 y y = + ( 1 )( 1 2 ) ( 1 2 ) xy xy 0 0 2
1 0 E [ D] = 1 + 1 2 1 0 ( )( ) 1 2 0 0 2
{ } = [D]{} or {} = [D]1{ }
Stress-Strain Relationships Strain-Displacements Relationships Deformation-Displacement Relationships (Shape Functions) Minimum Energy Principle
Strain-Displacements Relationships
u x x v {} = y = y xy u v y + x
[B ] [D ] [B ]
V T
dV
{d } = {f }
[k ] = [B ] [D ] [B ]
V
dV
Constant Strain Triangular (CST) Element (3 nodes) Linear Strain Triangular (LST) Element (3 nodes) Four Node Rectangular Element (4 nodes) Four Node Quadrilateral Element (4 nodes)
- Four-node iso-parametric finite element is one of the most commonly used elements. - Eight unknowns: two displacements per each node. - Iso-parametric: the same interpolation method is used for displacement and geometry. - Mapping relation from physical element to reference element. - Numerical integration
Iso-parametric Mapping
Lagrange interpolation method (Shape functions)
Linear Functions
1. 2. 3. 4.
Ensures compatibility between elements. Displacements vary linearly along any line. Displacements vary linearly between nodes. Edge displacements are the same for adjacent elements if nodal displacements are equal.
Interpolation of Displacement
Displacement-strain relationship
Displacement-strain relationship
J is Jacobian matrix
Derivatives of shape functions with respect to coordinate directions are required. Since shape functions depend on and coordinates, chain rule of differentiation must be used
Numerical Integration
- Numerical integration evaluates the integrals involved in the element stiffness matrix and distributed force. - In the finite element literature, the Gauss quadrature is usually preferred because it requires fewer function evaluations as compared to other methods. - In the Gauss quadrature, the integrand is evaluated at predefined points (called Gauss points). The sum of this integrand values, multiplied by appropriate weights (called Gauss weight) gives an approximation to the integral:
Gauss quadrature
When choosing elements and creating meshes for FEA problems users must make sure that
Chosen mesh size and density are optimal for the problem (to save computational time) Chosen element types are appropriate for the analysis type performed (for accuracy) Element shapes do not result in near singular stiffness matrices Chosen elements and meshes can represent force distributions properly
Choose element types that are appropriate for the loading and stress conditions of the problem Make sure that the elements chosen capture all possible significant stresses that may result from the given loading, geometry, and boundary conditions
Slender beam; beam elements
Thick beam (shear present); quadrilateral plane stress or plane strain elements
Aspect Ratio
For a good mesh all elements must have a low aspect ratio Specifically
b h
b 24 h
where b and h are the longest and the shortest sides of an element, respectively
Element Shape
Worse
Better
Mesh Refinement
Finer meshing must be used in regions of expected high stress gradients (usually occur at discontinuities) Mesh refinement must be gradual with adjacent elements of not too dissimilar size Mesh refinement must balance accuracy with problem size
Discontinuities
In general different types of elements with different DOF at their nodes should not share global DOF (for example do not use a 3D beam element in conjunction with plane stress elements)
The approximations and discretizations generated by the FE method enforce some equilibrium and compatibility conditions but not others
Equilibrium of nodal forces and moments is always satisfied because of KU = F Compatibility is guaranteed at the nodes because of the way K is formed; i.e. the displacements of shared nodes on two elements are the same in the global frame in which the elements are assembled
Equilibrium-Compatibility (contd)
Equilibrium is usually not satisfied across inter-element boundaries; however discrepancies decline with mesh refinement
( 1) (2) x x ( 1) (2) y y
( 1) (2) xy xy
along this boundary Stresses at shared nodes are typically averaged over the elements sharing the node
Principal Stresses
2
x + y + 1 = 2
x y 2 + 2 xy
2
2 1
x + y x y 2 + 2 = xy 2 2 2 xy tan 2p = x y
P
2