82 Finite Element Analysis

5.5 FINDING SHAPE FUNCTIONS USING LAGRANGE POLYNOMIALS

If only continuity of basic unknown (displacement) is to be satisfied, Lagrange polynomials can be used to
derive shape functions. Lagrange polynomial in one dimension is defined by
n
b g ∏ xx −− xx
Lk x =
k
m

m
…(5.37)
m=1
m≠ k

Thus, if n = 5 and k = 3,

b g b xb x−−xxgbgbxx −− xx gbgb xx −−xx gbgbxx− −x xg g

L3 x =
3 1
1

3
2

2 3
4

4 3
5

Obviously equation 5.37 takes the value equal to zero at all points except at point k. At point k its value
is unity.
This is exactly the property required for the interpolation functions. Hence Lagrange Polynomial can be
straight way used as shape functions for one dimensional problems. The following example illustrates it.
Example 5.16: Using Lagrange polynomial find shape functions for
(i) Two noded bar element
(ii) Three noded bar element and
(iii) Five noded bar element
Plot the variation of shape functions.
Solution: (i) Two noded bar element
The typical two noded bar elements is shown in Fig. 5.17 (a)
General Lagrange Polynomial is

Typical element

Variation of N1

Variation of N2
(a)

Fig. 5.17 (a) Two noded bar element

Lk =
b x − x gb x − x g ... d x − x id x − x i ... b x − x g
1 2 k −1 k +1 n

b x − x gb x − x g ... d x − x id x − x i ... b x − x g
k 1 k 2 k k −1 k k +1 k n

Now n = 2. Hence when k = 1

 Shape Functions 83

x − x2 x −x N
∴ N1 = L1 = = 2 , same as found earlier
x1 − x2 x 2 − x1 1× 8
When k = 2,
x − x1
N 2 = L2 = , same as found earlier variation of N1 and N2 is also shown in Fig. 5.17 (a).
x2 − x1
(ii) For Three Noded Element
n = 3. Hence when k = 1,

b x − x gb x − x g
2 3
N1 = L1 =
b x − x gb x − x g
1 2 1 3

When k = 2

b x − x gb x − x g
1 3
N 2 = L2 =
b x − x gb x − x g
2 1 2 3

b x − x gb x − x g
1 2
and N 3 = L3 =
b x − x gb x − x g
3 1 3 2

1 3 2
Typical 3 noded bar element

Variation of N1

Variation of N2

Variation of N3

Fig. 5.17 (contd) (b) Three noded bar element

The typical element and the variation of its shape functions are shown in Fig. 5.17 (b).
(iii) For Five Noded Element
n=5
b x − x gb x − x gb x − x gb x − x g
2 3 4 5
N1 = L1 =
b gb gb gb
x1 − x2 x1 − x 3 x1 − x4 x1 − x5 g
84 Finite Element Analysis

b x − x gb x − x gb x − x gb x − x g
1 3 4 5
N 2 = L2 =
b x − x gb x − x gb x − x gb x − x g
2 1 2 3 2 4 2 5

b x − x gb x − x gb x − x gb x − x g
1 2 4 5
N 3 = L3 =
b x − x gb x − x gb x − x gb x − x g
3 1 3 2 3 4 3 5

b x − x gb x − x gb x − x gb x − x g
1 2 3 5
N 4 = L4 =
b x − x gb x − x gb x − x gb x − x g
4 1 4 2 4 3 4 5

b x − x gb x − x gb x − x gb x − x g
1 2 3 4
N 5 = L5 =
b x − x gb x − x gb x − x gb x − x g
5 1 5 2 5 3 5 4
The five noded bar element and variation of its shape functions are shown in Fig. 5.17 (c)

1 3 4 5 2

(a) Typical noded element

Variation of N1

Variation of N3

Variation of N4

Variation of N5

Variation of N2

Fig. 5.17 (contd) (c) Typical 5 noded bar element

Lagrange Polynomial Approach for Two Dimensional Elements

Although Lagrangian interpolation functions are for only one dimension, we may extend the concept to two
and three dimensions by forming the product of the functions which hold good for the individual one
dimensional coordinate directions i.e.,
 Shape Functions 85

bg bg
N1 = L1 ξ L1 η …(5.38)
Thus for 4 noded rectangular element shown in Fig. 5.18,

bg bg
N1 = L1 ξ L1 η =
ξ − ξ 2 η − η4
ξ 1 − ξ 2 η1 − η 4

=
bξ − 1gη −1 1
b 1
gb
= ξ −1 η −1 = 1−ξ 1−η g b gb g
−1 − 1 −1 − 1 4 4

4 (–1, 1) 3 (1, 1)

1 (–1, –1) 2 (1, –1)

Fig. 5. 18 Typical 4 noded rectangular element

bg bg
N 2 = L2 ξ L2 η =
ξ − ξ1 η − η3
ξ 2 − ξ1 η2 − η3

cξ − b−1gh bη − 1g = bξ + 1gbη − 1g = 1 b1 + ξgb1 − ηg

=
1 − b −1g −1 − 1 b−4g 4

bg bg
N 3 = L3 ξ L3 η =
ξ − ξ 4 η − η2
ξ 3 − ξ4 η3 − η 2

ξ +1 η+1 1+ξ 1+η b gb g

=
b g b g
1 − −1 1 − −1 = 4

bg bg
N 4 = L4 ξ L4 η =
ξ − ξ 3 η − η1
ξ 4 − ξ 3 η 4 − η1

bξ − 1g η + 1 = bξ − 1gbη + 1g = b1 − ξgb1 + ηg
−1 − b1g 1 − b −1g
=
−4 4

Thus Ni =
b1 + ξξ gb1 + ηη g
i i
4
Example 5.17:: Using Lagrange functions write shape functions for the nine noded rectangular element sshown
in Fig. 5.19.
86 Finite Element Analysis

8
7 9
(0, 1)
(–1, 1) (1, 1)

4
(–1, 0) 5 (0, 0) 6
(1, 0)

1 2 3
(–1, –1) (0, –1) (1, –1)

Fig. 5.19 Typical nine noded rectangular element

Solution: The natural coordinates of various nodes are as shown in the figure. For the Co continuity element
in two dimensions,
N i = Li ξ Li ηbg bg
where Li refers to Langrangian function at node i. In this case there are 3 nodes in each direction. Hence n =
3 in Lagrange function

b gb
g b gb g
ξ − ξ2 ξ − ξ3 η − η4 η − η7
b
N1 =
gb gb gb g
ξ 1 − ξ 2 ξ 1 − ξ 3 η1 − η 4 η1 − η 7

bξ − 0gbξ − 1g bη − 0gbη − 1g = ξ bξ − 1g η bη − 1g
=
b−1 − 0gb−1 − 1g b−1 − 0gb−1 − 1g 4

bξ − ξ gbξ − ξ g bη − η gbη − η g
1 3 5 8
N = 2
bξ − ξ gbξ − ξ g bη − η gbη − η g
2 1 2 3 2 5 2 8

bξ + 1gbξ − 1g bη − 0gbη − 1g = bξ + 1gbξ − 1gη bη − 1g

=
b0 + 1gb0 − 1g b−1 − 0gb−1 − 1g b−2g
bξ − ξ gbξ − ξ g bη − η gbη − η g
1 2 6 9
N =
3
bξ − ξ gbξ − ξ g bη − η gbη − η g
3 1 3 2 3 6 3 9

bξ + 1gbξ − 0g bη − 0gbη − 1g = bξ + 1g ξ η bη − 1g
=
b1 + 1gb1 − 0g b−1 − 0gb−1 − 1g 4

bξ − ξ gbξ − ξ g bη − η gbη − η g
5 6 1 7
N = 4
bξ − ξ gbξ − ξ g bη − η gbη − η g
4 5 4 6 4 1 4 7
 Shape Functions 87

bξ − 0gbξ − 1g bη + 1gbη − 1g = ξ bξ − 1gbη + 1gbη − 1g

=
b gb g b0 + 1gb0 − 1g
−1 − 0 −1 − 1 −2

bξ − ξ gbξ − ξ g bη − η gbη − η g
4 6 2 8
N5 =
bξ − ξ gbξ − ξ g bη − η gbη − η g
5 4 5 6 5 2 5 8

bξ + 1gbξ − 1g bη + 1gbη − 1g = bξ + 1gbξ − 1gbη + 1gbη − 1g

=
b0 + 1gb0 − 1g b0 + 1gb0 − 1g 1

bξ − ξ gbξ − ξ g bη − η gbη − η g
4 5 3 9
N6 =
bξ − ξ gbξ − ξ g bη − η gbη − η g
6 4 6 5 6 3 6 9

bξ + 1gbξ − 0g bη + 1gbη − 1g = bξ + 1g ξ bη + 1gbη − 1g

=
b1 + 1gb1 − 0g b0 + 1gb0 − 1g −2

bξ − ξ gbξ − ξ g bη − η gbη − η g
8 9 1 4
N7 =
b − ξ gbξ − ξ g bη − η gbη − η g
ξ7 8 7 9 7 1 7 4

bξ − 0gbξ − 1g bη + 0gbη − 0g = ξ bξ − 1gbη + 1g η

=
b−1 − 0gb −1 − 1g b1 + 1gb1 − 0g 4

bξ − ξ gbξ − ξ g bη − η gbη − η g
7 9 2 5
N8 =
bξ − ξ gbξ − ξ g bη − η gbη − η g
8 7 8 9 8 2 8 5

bξ + 1gbξ − 1g bη + 1gbη − 0g = bξ + 1gbξ − 1gbη + 1gbηg

=
b0 + 1gb0 − 1g b1 + 1gb1 − 0g −2

bξ − ξ gbξ − ξ g bη − η gbη − η g
7 8 3 6
N9 =
bξ − ξ gbξ − ξ g bη − η gbη − η g
9 7 9 8 9 3 9 6

bξ + 1gbξ − 0g bη + 1gbη − 0g = bξ + 1g ξ bη + 1gbηg

=
b1 + 1gb1 − 0g b1 + 1gb1 − 0g 4
Thus in this case, for corner nodes,

Ni =
1
4
b gb
ξη ξ + ξ i η + η i g …(5.41)

For nodes 2 and 8 where ξ = 0 ;

Ni =
bξ + 1gbξ − 1g η bη + η g i
−2
88 Finite Element Analysis

For nodes 4 and 6, where η i = 0

Ni =
b
ξ ξ + ξi g bη − 1gbη + 1g
−2
and for central node,

N5 =
bξ + 1gbξ − 1g bη + 1gbη − 1g
4

Example 5.18: Using Langrange functions, derive shape function for hexahedron (brick) element.
Solution: Typical hexahedron element is shown in Fig. 5.20.
The coordinates for various nodal points are
1 (1, –1, –1), 2 (1, 1, –1), 3 (1, 1, 1), 4 (1, –1, 1), 5 (–1, –1, –1), 6 (–1, 1, –1), 7 (–1, 1, 1), 8 (–1, –1, 1)
In general, shape function by Lagrange function for three dimensional case is given by
bg bg bg
N i = Li ξ Li η Li ζ
8

4
3

5
6

z
1
2

Fig. 5.20 Typical hexahedron element

bξ − ξ g 5 η − η2 ζ − ζ 4 ξ +1 η −1 ζ −1b gb gb
ξ +1 η−1 ζ −1g b gb gb g
∴ N1 =
ξ1 − ξ 5 η1 − η 2 ζ 1 − ζ 4
=
b gb
1 + 1 −1 − 1 −1 − 1
=
gb8 g
bg bg bg
N 2 = L2 ξ L2 η L2 ζ

bξ − ξ g bη − η g bζ − ζ g
6 1 3
=
bξ − ξ g bη − η g bζ − ζ g
2 6 2 1 2 3
 Shape Functions 89

bξ + 1gbη + 1gbζ − 1g = bξ + 1gbη + 1gbζ − 1g

=
b1 + 1gb1 + 1gb−1 − 1g −8

N3 = L bξ g L bηg L bζ g
3 3 3

=
bξ − ξ g η − η ζ − ζ
7 4 2
ξ 3 − ξ 7 η3 − η4 ζ 3 − ζ 2

=
ξ + 1η + 1ζ + 1
=
b
ξ +1 η+1 ζ +1 gb gb g
1+1 1+1 1+1 8

bg bg bg
N 4 = L4 ξ L4 η L4 ζ =
ξ − ξ 8 η − η3 ζ − ζ 1
ξ 4 − ξ8 η4 − η3 ζ 4 − ζ 1

=
ξ +1 η −1 ζ +1
=
b
ξ +1 η −1 ζ +1 gb gb g
1 + 1 −1 − 1 1 + 1 −8
bg bg bg
N 5 = L5 ξ L5 η L5 ζ

=
bξ − ξ g η − η ζ − ζ
1 6 8
=
bξ − 1g η − 1 ζ − 1 = bξ − 1gbη − 1gbζ − 1g
ξ 5 − ξ1 η5 − η6 ζ 5 − ζ 8 −1 − 1 −1 − 1 −1 − 1 −8
bg bg bg
N 6 = L6 ξ L6 η L6 ζ

=
ξ − ξ 2 η − η5 ζ − ζ 7
=
ξ −1 η+1 ζ −1
=
ξ −1 η +1 ζ −1 b gb gb g
ξ 6 − ξ 2 η 6 − η5 ζ 6 − ζ 7 −1 − 1 1 + 1 −1 − 1 8
bg bg bg
N 7 = L7 ξ L7 η L7 ζ

=
ξ − ξ 3 η − η8 ζ − ζ 6
=
ξ − 1 η + 1ζ + 1
=
ξ −1 η +1 ζ +1 b gb gb g
ξ 7 − ξ 3 η 7 − η8 ζ 8 − ζ 6 −1 − 1 1 + 1 1 + 1 −8

bg bg bg
N 8 = L8 ξ L8 η L8 ζ

=
ξ − ξ 4 η − η7 ζ − ζ 5
=
b
ξ −1 η −1 ζ +1
=
gb
ξ −1 η−1 ζ +1g b gb gb g
ξ 8 − ξ 4 η8 − η 7 ζ 8 − ζ 5 −1 − 1 −1 − 1 1 + 1 8
In general it may be noted that,

Ni =
1
8
b
1 + ξξ i g b1 + ηη g b1 + ζζ g
i i …(5.42)

5.6 SHAPE FUNCTIONS FOR SERENDIPITY FAMILY ELEMENTS

Figure
g 5.21 shows Serendipity family elements. These elements may be called as boundary node family
elements also. In these elements nodes are only on the boundaries. Zienkiewcz called them as ‘Serendip
family’ elements by referring to the famous princess of Serendip noted for chance discoveries. The terms
linear, quadratic, cubic and quartic are used since the variation of shape functions about a boundary is of that
order. The shape functions are found from the consideration that Ni = 1 for ith node and is zero when referred
90 Finite Element Analysis

to any other node. Discovery of these elements clubbed with isoparametric concept (explained in ch.13) has
made major break through in the finite element analysis. In this article derivation of shape functions for this
family of elements is presented, through examples.

(a) Linear (b) Quadratic

(c) Cubic (d) Quartic

Fig. 5.21 Serendipity family elements

Example 5.19: Using ‘serendipity concept’ derive shape functions for 4 noded rectangular element.
Solution: Figure 5.22 shows the typical element in natural coordinate system.

4 (–1, 1) =1 3 (1, 1)

1 (–1, –1) 2 (1, –1)

Fig. 5.22 Typical serendipity linear element

N1 has to satisfy the conditions

(a) along ξ = 1, N1 = 0.
(b) along η = 1, N1 = 0
and (c) at ξ = − 1, η = − 1 N1 = 1.
 Shape Functions 91

Hence let
b gb g
N1 = C 1 − ξ 1 − η , where C is arbitrary constant.
Conditions (a) and (b) are satisfied.
Condition (c) gives,

b gb g
1 = C 1 + 1 1 + 1 or C =
1
4

∴ N1 =
b1 − ξgb1 − ηg
4
On the same lines we can get,

N2 =
b1 + ξ gb1 − ηg
4

N3 =
b1 + ξgb1 + ηg
4

and N4 =
b1 − ξgb1 − ηg
4
These are same as given in equations 5.26 and 5.40, which were derived from different approaches.
Example 5.20: Using serendipity concept find shape functions for quadratic serendipity family element.
Solution: Figure 5.23 shows a typical element of this type.
1 0
4 (–1, 1) 7 (0, 1)
3 (1, 1)

1 0

8 ( –1, 0)
0 6 (1.0, 0)

1 0

1 (–1, –1) 2 (1, –1)

5 (0, –1)
1 0

Fig. 5.23 Typical serendipity quadratic element

The conditions to be satisfied by N1 are,

(a) along line ξ = 1, N1 = 0
(b) along line η = 1, N1 = 0
(c) along line 5–8, N1 = 0
92 Finite Element Analysis

i.e., ξ + η + 1 = 0.
(d) At ξ = − 1 and η = −1 , N1 = 1
Hence let
b gb
N1 = C 1 − ξ 1 − η 1 + ξ + ηgb g
It satisfies the requirements a, b, c. In other words it ensues N1 = 0 at nodes 2, 3, … 8.
From condition (d),

b gb gb
1= C 1+ 1 1+ 1 1− 1−1 ∴C = − g 1
4
∴ For the corner node

N1 = −
b1 − ξgb1 − ηgb1 + ξ + ηg
4
Similarly we can show that

N2 = −
b1 + ξgb1 − ηgb1 − ξ + ηg
4

N3 = −
b1 + ξ gb1 + ηgb1 − ξ − ηg
4

and N4 = −
b1 − ξgb1 + ηgb1 + ξ − ηg
4
For mid-side node 5, the conditions to be satisfied are
(a) Along ξ = 1 N5 = 0
(b) Along η = 1 N5 = 0
(c) Along ξ = − 1 N5 = 0
(d) At node 5 where ξ = 0 , η = − 1 , N5 = 1

b gb gb
∴ Let N 5 = C 1 − ξ 1 − η 1 + ξ g
This form satisfies N1 = 0 at all nodes other than node 5. From the condition ‘d’ we get,
1 = C (1 – 0) (1 + 1) (1 + 0)
1
∴C=
2

∴ N5 =
1
b gb
1− ξ 1−η 1+ ξ = gb g e
1 − ξ2 1 − η jb g
2 2
Similarly it may be shown that

N6 =
b1 + ξg e1 − η j 2
…(5.44a)
2
 Shape Functions 93

N7 =
e1 − ξ j b1 + ηg
2
…(5.44b)
2

and N8 =
b1 − ξg e1 − η j
2

…(5.44c)
2