Solid Elements

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Solid Elements

The Finite Element Library of the MIDAS Family Programs includes the following Solid Elements: 4-
node tetrahedron, 6-node pentahedron, and 8-node hexahedron shown in Fig. 1. The finite element
formulation of all element types is based on the isoparametric procedure (the element geometry and
displacements are interpolated in the same way).

w6 v6

v4
ζ
w4
6 u6
w5
u4 4
η u5
5
v5 w3 v3
ζ
v4 w1 v1 ξ
w4 3 u3
u4 w2
4 v2
u1 1 w8 v8 w7 v7
η 2 u2
u8 ζ u7
w3 w5
v3 8 w6 7
u5 η
ξ 6 u6
w1 v5 5
v1 u3 v6
w2
3 w4 v4 ξ w3
v2 v3
u1 1 u1
w1 w2
2 u2 4 3 u3
Z u1 v2
Y
1 2 u2
v1
X

Figure 1 3D Solid Finite Elements

The nodal degrees of freedom (DOF’s) are illustrated in Fig. 1. The element geometry and displacement
field are defined in terms of nodal coordinated and DOF’s by the following functions:

 n
 n

 x = ∑ fi (ξ ,η , ζ ) xi  u = ∑ f i (ξ ,η , ζ ) ui
 i =1
 i =1

 n
 n

 y = ∑ fi (ξ ,η , ζ ) yi  v = ∑ f i (ξ ,η , ζ ) vi ⇒ u = fq
 i =1  i =1
 n  n
 z = ∑ fi (ξ ,η , ζ ) zi  w = ∑ f i (ξ ,η , ζ ) wi
 i =1  i =1

where
u i , vi , w i = displacements in direction of global X, Y, Z− axes, respectively, at i − th node
xi , yi , zi = x, y, z − coordinates at i − th node
f i (ξ , η , ζ ) = interpolation function related to i − th node and defined in the natural coordinate
system of the element with variables ξ ,η , ζ ( − 1 ≤ ξ ≤ +1 , −1 ≤ η ≤ +1 , and
−1 ≤ ζ ≤ + 1 )
ξ i ,η i , ζ i = natural coordinates of the i − th node of the element
n = number of nodes in element

The nodal interpolation functions f i (ξ ,η , ζ ) are of the following form:

For 4-node tetrahedron element

1
 2 (1 + ζ ) for i = 4

1
fi (ξ ,η , ζ ) =  (1 + ξ )( 1 −ζ ) for i = 3
4
1
 8 (1 + ξiξ ) (1− η ) (1 − ζ ) for i =1,2

For 6-node pentahedron (wedge) element

1
 4 (1 + ξ )(1 + ζ iζ ) for i = 3, 6
f i (ξ , η , ζ ) = 
 1 (1 + ξ ξ ) (1 + η η ) (1 + ζ ζ ) for i = 1, 2, 4,5
 8 i i i

For 8-node hexahedron (brick) element

1
f i (ξ , η , ζ ) = (1 + ξiξ ) (1 + ηiη ) (1 + ζ iζ ) for i = 1, 2,3,.....,8
8
Now, the strains at any point within element domain are expressed in terms of nodal displacements as,

 ∂u   ∂ 
 ∂x   ∂x 0 0
   
 ∂v   0 ∂
0
 ε x   ∂y   ∂y 
ε     
 y   ∂w   0 0
∂
u 
 ε z   ∂z   ∂z   
 = =   v  ⇒ ε = du = dfq = Bq
γ xy   ∂u + ∂v   ∂ ∂
0   w
γ yz   ∂y ∂x   ∂y ∂x 
     
γ zx   ∂v + ∂w   0 ∂ ∂
 ∂z ∂y   ∂z ∂y 
   
 ∂w + ∂u   ∂ 0
∂
 ∂x ∂z   ∂z ∂x 
where
B = strain-displacement matrix
q = vector of nodal displacements
Then, the stress-strain relation become,

σ = Dε = DBq

where, D is the elasticity matrix defining mechanical properties of the material. For a linear isotropic
material D matrix takes the following form:

1 − ν ν ν 0 0 0 
 1 −ν ν 0 0 0 

 1 −ν 0 0 0 
 
 1 − 2ν
D=
E 0 0 
(1 + ν )(1 − 2ν )  2 

1 − 2ν
 Sym. 0 
 2 
 1 − 2ν 
 
 2 

in which
E = Young modulus
ν = Poisson’s ratio
Accordingly, the stiffness matrix and force vectors for a typical isoparametric 3D solid element are
defined by the following integrals:

K = ∫ BT DBdV
V

p b = ∫ f T bd Ω

p 0 = ∫ BT D ε 0 d Ω

where
K = stiffness matrix
V = volume of the element
A = surface area of the element
pb = nodal force vector due to distributed body forces b
p0 = nodal force vector due to initial strain ε0
Ω = range of the integration

So far we have demonstrated the standard isoparametric formulation procedure, which is identical for all
3D solids elements. It should be noted the 4-node tetrahedron and 6-node wedge elements presented here
are in fact the degenerated forms of the 8-node hexahedron (brick). Formulation of these elements by
collapsing the 8-node hexahedra allows evaluating the above integrals for all types of 3D solid elements
by use of the same standard 3D numerical integration procedure, based on the Gauss-Legendre quadrature.
Thus the numerical integration formulas used for 3D elements are:

l m n
K = ∑∑∑ WW
i jWk B (ξ i ,η k , ζ k ) DB (ξ i ,η k , ζ k ) J (ξ i ,η k , ζ k )
T

i =1 j =1 k =1
l m n
pb = ∑∑∑ WW
i jWk f (ξ i ,η k , ζ k ) b (ξ i , η k , ζ k ) J (ξ i ,η k , ζ k )
T

i =1 j =1 k =1
l m n
p 0 = ∑∑∑ WW
i jWk B (ξ i ,η k , ζ k ) Dε 0 (ξ i ,η k , ζ k ) J (ξ i ,η k , ζ k )
T

i =1 j =1 k =1

where:
Wi = weighting factor of i -th integration point
ξi ,ηi , ζ i = natural coordinates of i -th integration point

J (ξ i , η j , ζ k ) = determinant of the Jacobian matrix

l , m, n = number of integration points in direction of ξ , η , and ζ , respectively.

The appropriate order of numerical integration and corresponding locations of integration points used in
the 3D solid elements are shown in Table bellow.
Table Gauss-Legendre Integration for 3D Solid Elements

Integration
order Location of integration points

ζ ζ ζ η
6 7
4 8
η 7 8
4 7 5
8 6
7 8
η 5 5 5 6
2×2×2 6
3
5 6 4 ξ
4
3 ξ
3
3 3
4 3 ξ 1
4 1 2
1
2 1 2 2
1 1
2 2

ζ η
ζ
6 7
26 27 8
η 25
11 5
4 12 6
9 7
5
3×3×3 7 10 8
5 4 4 ξ
3 6 ξ 5 6
4 3 1 3
1 2 3
2 1 2
1 2

Here it should be noted that the compatible 8-node hexahedron element based on standard isoparametric
formulation does not produce accurate results in many cases for both displacements and stresses. In order
to improve the performance of this element are added so called the incompatible displacements modes.
Taking into account the incompatible modes, the displacement field is defined as,

 8

 u = ∑ fi (ξ ,η , ζ )i ui + α1 P1 + α 4 P2 + α 7 P3
 i =1

 8

 v = ∑ fi (ξ ,η , ζ )i vi + α 2 P1 + α 5 P2 + α 8 P3
 i =1
 8


w = ∑i =1
f i (ξ ,η , ζ ) wi + α 3 P1 + α 6 P2 + α 9 P3

where, P1 , P2 , P3 are the extra shape functions related to the incompatible modes expressed as,

P1 (ξ ,η , ζ ) = 1 − ξ 2 , P2 (ξ ,η , ζ ) = 1 − η 2 and P3 (ξ ,η , ζ ) = 1 − ζ 2

and α i (i = 1, 2,3,...,9) are so called “nodeless extra DOF’s”. Note that the extra shape functions
permit a parabolic deformation along the element edge and improve the element stiffness performance.

Here it should be noted that the final stiffness matrix is obtained by static condensation of the
incompatible modes in the same way as in case of 4-node quadrilateral 2D plane element.

You might also like