Professional Documents
Culture Documents
1 Introduction
Where are we headed?
1. Rectangular Elements
• Linear
• Quadratic (Quadrilateral)
2. Triangular Elements
• Linear
• Quadratic
3. Axisymmetric Elements
4. Isoparametric Elements
5. ANSYS Elements
1
2.1 Using Local Coordinates
Temperature Distribution
Substitute B.C.s into equation to create four equations with four unknowns. Solve equations
simultaneously to obtain b1 , b2 , b3 , and b4 .
b1 = Ti
1
b2 = (Tj − Ti )
ℓ
1
b3 = (Tn − Ti )
w
1
b4 = (Ti − Tj + Tm − Tn )
ℓw
Substituting into equation . . .
1 1 1
T (e) = Ti + (Tj − Ti ) x + (Tn − Ti ) y + (Ti − Tj + Tm − Tn ) xy
ℓ w ℓw
Algebraically rearranging to obtain shape functions
T (e) = Si Ti + Sj Tj + Sm Tm + Sn Tn
ψ (e) = Si ψi + Sj ψj + Sm ψm + Sn ψn
Or
ψi
[ ] ψj
ψ (e) = Si Sj Sm Sn
ψm
ψn
Relationship between Local coordinates (x and y) and the Natural coordinates (ξ and η).
2x 2y
ξ= −1 η= −1
ℓ w
2-D linear shape functions in natural coordinates
1 1
Si = (1 − ξ)(1 − η) Sj = (1 + ξ)(1 − η)
4 4
1 1
Sm = (1 + ξ)(1 + η) Sn = (1 − ξ)(1 + η)
4 4
2.2.1 Meeting Shape Function Properties
Notice shape function properties are retained for natural coordinates:
1. Si = 1 at Node i (i.e. ξ = −1, η = −1)
Sj = 1 at Node j (i.e. ξ = 1, η = −1)
Sm = 1 at Node m (i.e. ξ = 1, η = 1)
Sn = 1 at Node n (i.e. ξ = −1, η = 1)
3. Si + Sj + Sm + Sn = 1
A process similar to the Lagrange Functions can be used to generate these shape functions (see
textbook).
3
3 Quadratic Rectangular (Quadrilateral) Elements
The 8-node Quadratic Quadrilateral Element is basically a higher order version of the 2-D 4-node
quadrilateral element.
• For same number of elements, this element produces better nodal results than the linear
4-Node element.
• The approximate solution for this element in natural coordinates is:
ψ (e) = b1 + b2 ξ + b3 η + b4 ξη + b5 ξ 2 + b6 η 2 + b7 ξ 2 η + b8 ξη 2
• Wow!
• Using B.C.’s, solving for unknowns (b1 , b2 . . . b8 ), and determining the shape functions is a
huge task!!
Shape Functions
For corner nodes:
−1 −1
Si = (1 − ξ)(1 − η)(1 + ξ + η) Sj = (1 + ξ)(1 − η)(−1 + ξ − η)
4 4
1 −1
Sm = (1 + ξ)(1 + η)(−1 + ξ + η) Sn = (1 − ξ)(1 + η)(1 + ξ − η)
4 4
For midpoint nodes:
1 1
Sk = (1 − η)(1 − ξ 2 ) Sℓ = (1 + ξ)(1 − η 2 )
2 2
1 1
So = (1 + η)(1 − ξ 2 ) Sp = (1 − ξ)(1 − η 2 )
2 2
General element solution with shape functions and ψ notation
ψ (e) = Si ψi + Sj ψj + Sm ψm + Sn ψn + Sk ψk + Sℓ ψℓ + So ψo + Sp ψp
OR
ψi
ψj
ψ
m
[ ] ψn
ψ (e) = Si Sj Sm Sn Sk Sℓ So Sp
ψk
ψℓ
ψo
ψp
. . . and shape function properties still apply.
4
4 Linear Triangular Elements
Ex: Heat Transfer through a fin surrounded by fluid
• The linear triangular element is well suited for curved boundaries, more so than the linear
quadrilateral element.
• The approximate temperature distribution in triangular elements is:
T (e) = a1 + a2 X + a3 Y
T = Ti at X = Xi and Y = Yi
T = Tj at X = Xj and Y = Yj
T = Tk at X = Xk and Y = Yk
Substitute B.C.s into equation to create three equations with three unknowns. Solve equations
simultaneously to obtain a1 , a2 , and a3 .
1
a1 = [(Xj Yk − Xk Yj ) Ti + (Xk Yi − Xi Yk ) Tj + (Xi Yj − Xj Yi ) Tk ]
2A
1
a2 = [(Yj − Yk ) Ti + (Yk − Yi ) Tj + (Yi − Yj ) Tk ]
2A
1
a3 = [(Xk − Xj ) Ti + (Xi − Xk ) Tj + (Xj − Xi ) Tk ]
2A
5
where A ≡ Area of triangle element, comes from . . .
T (e) = Si Ti + Sj Tj + Sk Tk
where λ ≡ Lambda. The coordinates of P define what areas A1 , A2 , and A3 will be.
For triangle elements we use areas to define the natural (area) coordinates ξ, η,andλ:
A1 A2 A3
ξ= , η= , λ=
A A A
By observation we see that: ξ + η + λ = 1. Hey! This looks familiar! Through further investigation
we find that ξ, η,andλ are the shape functions.
Therefore:
ξ = Si η = Sj λ = Sk
where Si , Sj , Sk are defined previously (substitute for nodal coordinates for A, αi , βi , δi )
6
5 Quadratic Triangular Elements
The 6-Node Quadratic Triangular Element is basically a higher order version of the 2-D 3-Node
Triangular Element. For the same number of elements, this element produces better nodal results
than the linear 3-Node element.
ψ (e) = a1 + a2 X + a3 Y + a4 X 2 + a5 XY + a6 Y 2
There are 6 unknown variables (a1 , a2 , a3 , a4 , a5 , a6 ) so we need 6 B.C.’s (Ex: T at each of 6 nodes)
Developing the shape functions in natural coordinates through the same process we obtain:
Si = ξ (2ξ − 1)
Sj = η (2η − 1)
Sk = λ (2λ − 1) = 1 − 3 (2 + η) + 2 (ξ + η)2
Sℓ = 4ξη
Sm = 4ηλ = 4η (1 − ξ − η)
Sn = 4ξλ = 4ξ (1 − ξ − η)
Again, the general solution for the 6-node Quadratic Triangular Element is:
ψi
ψj
[ ]
(e) ψ k
ψ = Si Sj Sk Sℓ Sm Sn
ψℓ
ψ
m
ψn
The spatial variable, ψ, can be temperature, velocity, displacement, etc. This is a very commonly
used element.
7
6 Axisymmetric Elements
Axisymmetric elements are used in a special class of 3-D problems where geometry and loading (i.e.
temperature, velocity, etc.) are symmetrical about an axis. Typically, the axis of symmetry is
the z-axis and all the elements are “rotated” about it.
Notice that because of symmetry, we can use 2-D elements even though it is a 3-D problem.
8
Therefore shape functions now are:
1
Si = (αi + βi r + δi z)
2A
1
Sj = (αj + βj r + δj z)
2A
1
Sk = (αk + βk r + δk z)
2A
where:
αi = Rj Zk − Rk Zj βi = Zj − Zk δi = Rk − Rj
αj = Rk Zi − Ri Zk βj = Zk − Zi δj = Ri − Rk
αk = Ri Zj − Rj Zi βk = Zi − Zj δk = Rj − Ri
Let’s convert our shape functions from x-y coordinates to R-Z coordinates
9
The relationships between the x-y coordinates and the R-Z coordinates are:
r = Ri + x → x = r − Ri
z = Zi + y → y = z − Zi
Also:
ℓ = Rj − Ri and w = Zn − Zi
Substitute these relationships into the 2-D linear rectangular shape functions to obtain the axisym-
metric rectangular shape functions.
10
6.3 Example
Solution:
Ti
[ ]
(e) Tj
T = Si Sj Sm Sn
T
m
Tn
( )( ) ( )( )
Rj − r Zn − z 7 − 3.5 4−3
Si = = = 0.233
ℓ w 5 3
( )( ) ( )( )
r − Ri Zn − z 3.5 − 2 4−3
Sj = = = 0.100
ℓ w 5 3
( )( ) ( )( )
r − Ri z − Zi 3.5 − 2 3−1
Sm = = = 0.200
ℓ w 5 3
( )( ) ( )( )
Rj − r z − Zi 7 − 3.5 3−1
Sn = = = 0.467
ℓ w 5 3
Ti
[ ]
Tj
T (e) = Si Sj Sm Sn
Tm
Tn
◦
50 F
[ ] 40◦ F
T (e) = 0.233 0.100 0.200 0.467
55◦ F
◦
70 F
T (e) = (0.233) (50◦ F) + (0.100) (40◦ F) + (0.200) (55◦ F) + (0.200) (55◦ F) + (0.467) (70◦ F)
T (e) = 59.3◦ F
7 Isoparametric Elements
Recall from a previous lecture, we use a single set of parameters (i.e. shape functions) to define
the unknown variables displacements, velocity, temperature, etc., and use the same parameters
(i.e. shape functions) to express the geometry, then we are using an “Isoparametric” Formulation.
These types of elements are called Isoparametric Elements. (See Table 7.1 from Moaveni).
11
7.1 Quadrilateral Plane-Stress Element (Linear, Natural Coordinates)
U describes the nodal displacements in the Global Coordinates and u−v describe the displacements
within the element, x-y directions respectively.
u(e) = Si Uix + Sj Ujx + Sm Umx + Sn Unx
7.2 Elements
In matrix form:
Uix
U
iy
{ } [ ]
U jx
u S 0 Sj 0 Sm 0 Sn 0 Ujy
= i
v 0 Si 0 Sj 0 Sm 0 Sn Umx
Umy
Unx
Uny
Use the Natural Coordinate shape functions developed previously. These same shape functions can
also be used to describe the position of any point within the element.
x = Si xi + Sj xj + Sm xm + Sn xn
y = Si yi + Sj yj + Sm ym + Sn yn
12