PERGAMON Computers and Structures 68 (1998) 555±560

Generation of shape functions for straight beam elements

Charles E. Augarde *
Department of Engineering Science, University of Oxford, Parks Road, Oxford, U.K.
Received 17 June 1997; received in revised form 14 February 1998

Abstract

Straight beam ®nite elements with greater than two nodes are used for edge sti€ening in plane stress analyses and
elsewhere. It is often necessary to match the number of nodes on the edge sti€ener to the number on a whole plane
stress element side. Beam elements employ shape functions which are recognised to be level one Hermitian
polynomials. An alternative to the commonly adopted method for determining these shape functions is given in this
note, using a formula widely reported in mathematical texts which has hitherto not been applied to this task in the
®nite element literature. The procedure derives shape functions for beams entirely from the set of Lagrangian
interpolating polynomials. Examples are given for the derivation of functions for a three and four-noded beam
element. # 1998 Elsevier Science Ltd. All rights reserved.

Keywords: Finite elements; Beams; Hermitian interpolation; Shape functions

1. Introduction node. With this amendment, beam elements are usually

Analysis of structures using the ®nite element
method is well established. Many formulations exist
for complex elements but simple elements remain pop-
ular since they are usually well-tested and easy to im-
plement into an analysis program. Two-dimensional
plane stress analysis, for thin structures subject to in-
plane loading, may employ continuum elements, such
as the ®fteen-node triangle, having a large number of
nodes along a side. Where edge sti€ening is required,
beam elements can be connected to continua edge
nodes. There is then a requirement for formulations of
beam elements having more than two nodes.
Conventional two-dimensional beam elements have
two degrees of freedom at each node: one lateral dis-
placement and one rotation. Unless the structure is
loaded entirely laterally, axial sti€ness must also be in-
corporated, by an additional degree of freedom at each
* Author to whom correspondence should be addressed.
referred to as frame elements. The axial e€ects are
uncoupled for straight beams and the determination of
suitable shape functions is straightforward [1].
The author has recently contributed to the develop-
ment of a complex three-dimensional ®nite element
model at the Department of Engineering Science,
Oxford University to study the damage accruing to
surface structures from adjacent tunnelling [2].
Modelling was also undertaken in two-dimensions (to
validate the complex model) where tunnel linings were
represented with simple beam elements. The procedure
outlined in this note was used in the implementation of
these beam elements into an existing in-house ®nite el-
ement code (OXFEM).
In this note we are concerned with the generation of
the shape functions which interpolate the lateral displa-
cements along beam elements having more than two
nodes. Bernoulli-Euler beam theory is assumed where
transverse shear deformation is zero. While most ®nite
element texts describe the simple two-noded beam [1, 4]
few explain how more complex elements may be
formulated [3].

0045-7949/98/$19.00 # 1998 Elsevier Science Ltd. All rights reserved.

PII: S 0 0 4 5 - 7 9 4 9 ( 9 8 ) 0 0 0 7 1 - 6
556 C.E. Augarde / Computers and Structures 68 (1998) 555±560

2. Standard Procedure where,

X ˆ f1 x x2 x3 g & a T ˆ fa1 a2 l a3 l2 a4 l3 g: …5†
The beam element of Fig. 1(a) is de®ned by two
nodes, a distance l apart, along the element local x- Taking Eq. (4) and its ®rst derivative with respect to x,
axis. A common convention, and that adopted here, is four further equations can be formed, one for each
to make the nodal coordinates non-dimensional, that degree of freedom:
is
d ˆ Aa
a …6†
x
x ˆ ; …1†
l where
2 3
so that the two-noded beam has nodes at x=0, +1. 1 0 0 0
Using shape functions, lateral displacement w(x) of 60 1 0 07
6
Aˆ4 7 …7†
the two-noded beam element of Fig. 1(a) is 1 1 1 15
0 1 2 3
w…
x† ˆ Nd …2†
Solving for a in Eq. (6) and substituting into Eqs. (4)
where
and (2) gives
N ˆ fN1 N2 N3 N4 g & dT ˆ fw1 y1 w2 y2 g: …3†
N ˆ XAÿ1 …8†
Nj,(j = 1,4) are the bending shape functions and wi, yi,
The three-node beam of Fig. 1(b), of overall length 2 l,
(i = 1,2) are the displacements and rotations at the
has nodes at x= ÿ 1, 0, +1. The same procedure as
nodes. Determination of the former, using the method
above yields
to be found in many ®nite element texts [4, 5], proceeds
by ®rst writing w(x) as an n-termed polynomial with N ˆ fN1 N2 N3 N4 N5 N6 g; …9†
unknown coecients, n being the number of degrees of
freedom in the element dT ˆ fw1 y1 w2 y2 w3 y3 g; …10†
w…
x† ˆ Xa
a …4†
X ˆ f1 x x2 x3 x4 x5 g …11†

a T ˆ fa1 a2 l a3 l2 a4 l3 a5 l4 a6 l5 g …12†
and to solve for N requires the inversion of
2 3
1 ÿ1 1 ÿ1 1 ÿ1
6 0 1 ÿ2 3 ÿ4 5 7
6 7
61 0 0 0 0 0 7
Aˆ6 60 1
7: …13†
6 0 0 0 0 7
7
41 1 1 1 1 1 5
0 1 2 3 4 5

3. Hermitian Interpolation

The shape functions in Eq. (2) are Hermitian poly-

Fig. 1. Beam elements showing bending degrees of freedom.
nomials since the displacement w(x) is interpolated
from nodal rotations as well as nodal displacements.
This contrasts with Lagrangian interpolation, used for
continuum elements' shape functions and for the axial
e€ects in frame elements. Considering small displace-
ments, the nodal rotations are the ®rst derivatives of
the unknown real displacement function at the nodes
thus ful®lling the de®nition of Hermitian interpolation.
This property allows an alternative procedure to be
used to determine the shape functions to that outlined
above.
 C.E. Augarde / Computers and Structures 68 (1998) 555±560 557

The generation of Hermitian (or Hermite) poly- code. The former are required for continuum elements
nomials from Lagrangian interpolation polynomials is and the derivatives are required for isoparametric el-
described in many mathematical texts [6±8]. Despite ements. The use of this procedure also provides a sys-
the clear understanding that bending shape functions tematic approach to allow simpler coding.
are equivalent to Hermitian polynomials the technique
described below has not, to the author's knowledge,
been linked to the generation of those shape functions. 4. Examples
One-dimensional interpolation is required for
straight beam elements. The single dimension is along 4.1. Three-node beam
the element centreline, de®ned as the x-axis. One-
dimensional Hermitian interpolation for an unknown, A three-node element with six bending degrees of
w(x) proceeds as [6, 7] freedom and a total length of 2 l is shown in Fig. 1(b).
This element has nodes at x=(ÿ1, 0, 1). The axial
nnod 
X    r 
dw dw degrees of freedom are omitted from this element as in
w…x† ˆ Hr0i wi ‡ Hr1i ‡ . . . Hrri …14†
iˆ1
dx i dxr i the derivations above. Shape functions for these
degrees of freedom are the Lagrangian polynomials of
where Hrji is a Hermite polynomial of level r, relating order 2. From the preceding section, it is clear that
to node i and to derivative order j of w. The sum is these are also required for the derivation of the bend-
over the number of nodes, nnod where values of w and ing shape functions.
its derivatives are available. The level of the poly- The Lagrangian polynomials are
nomial indicates the highest order derivative used in
…x ÿ x2 †…x ÿ x3 †
the interpolation. L1 …x† ˆ …21†
…x1 ÿ x2 †…x1 ÿ x3 †
Comparison of Eqs. (2) and (14) reveals that the
bending shape functions are level one Hermitian poly-
…x ÿ x1 †…x ÿ x2 †
nomials as follows: L2 …x† ˆ …22†
…x2 ÿ x3 †…x2 ÿ x1 †
N1 ˆ H101 N2 ˆ H111 N3 ˆ H102 N4 ˆ H112 : …15†
…x ÿ x1 †…x ÿ x2 †
Level one Hermitian polynomials are derived from L3 …x† ˆ : …23†
…x3 ÿ x1 †…x3 ÿ x2 †
Lagrangian polynomials by the following formulae:
Rewriting in terms of the non-dimensional coordinate
H10i ˆ ‰1 ÿ 2…x ÿ xi †L 0i …xi †Š‰Li …x†Š2 …16†
x and substituting for values of x at nodes [i.e. (x1, x2,
x3) = (ÿ1, 0, 1)], gives
H11i ˆ …x ÿ xi †‰Li …x†Š2 …17†
x
L1 …
x† ˆ …x ÿ 1† …24†
where Li(x) is the one-dimensional Langrangian poly- 2
nomial of degree (nnod ÿ 1) calculated at node i, given
by L2 …
x† ˆ ÿ
x…
x ‡ 1† …25†
Y
nnod
x ÿ xj x
Li …x† ˆ …18† L3 …
x† ˆ …x ‡ 1†: …26†
x
jˆ1;j6ˆi i
ÿ xj 2
The derivatives are
and L'i(x) is its ®rst derivative with respect to x. A
polynomial of order (nnod  2 ÿ 1) is required to in- 1
L 01 …
x† ˆ …2
x ÿ 1† …27†
terpolate over nnod points, each contributing two 2l
values. Inspection of Eqs. (16) and (17) shows that the
Hermite polynomials are of the correct order for in- 1
L 02 …
x† ˆ ÿ …2
x ‡ 1† …28†
terpolation. l
Re-writing Eqs. (16) and (17) in terms of the non-
dimensional coordinate, x 1
L 03 …
x† ˆ …2
x ‡ 1†: …29†
2l
H10i ˆ ‰1 ÿ 2l…
xÿ xi †L 0i … x†Š2
xi †Š‰Li … …19†
The bending shape functions are equivalent to the fol-
lowing Hermite polynomials:
H11i ˆ l… x†Š2 :
x ÿ xi †‰Li … …20†
N2 ˆ H101 N3 ˆ H111 N5 ˆ H102 …30†
The advantage of a derivation based on the
Lagrangian polynomials and their ®rst derivatives is
N6 ˆ H112 N8 ˆ H103 N9 ˆ H113 : …31†
that these are already likely to be present in a program
558 C.E. Augarde / Computers and Structures 68 (1998) 555±560

Note that the ®rst four are not the same as those in Similarly,
Eq. (15) since each set is based on di€erent order hx i2 l
Lagrangian polynomials. H111 ˆ l…
x ‡ 1† … x ÿ 1† ˆ ‰x2 ÿ x3 ÿ x4 ‡ x5 Š …33†
2 4
From Eqs. (24) and (27) and with substitution for x1
we obtain   
1
   H102 ˆ 1 ÿ 2l
x ÿ ‰ÿ x ‡ 1†Š2 ˆ 1 ÿ 2
x… x2 ‡ x4
3 hx i2
l
H101 ˆ 1 ÿ 2l…
x ‡ 1† ÿ …
x ÿ 1†
2l 2 …34†
5 1 3
x2 ÿ x3 ÿ x4 ‡ x5 :
ˆ …32†
4 2 4 H112 ˆ l
x‰ÿ x ‡ 1†Š2 ˆ l‰
x… x3 ‡ x5 Š
x ÿ 2 …35†

Fig. 2ÐCaption overleaf

 C.E. Augarde / Computers and Structures 68 (1998) 555±560 559

Fig. 2. (a) Shape functions for a three-node beam element. (b) Shape functions for a three-node beam element.

  h i2 lated to derivatives have slopes of 1 at the associated

3 x
H103 ˆ 1 ÿ 2l…
x ÿ 1† …
x ‡ 1† node.
2l 2
5 1 3 4.2. Four-node beam
x2 ‡ x3 ÿ x4 ÿ x5
ˆ …36†
4 2 4
Fig. 1c shows a beam with four nodes having the
hx i2 l
H113 x ÿ 1† …
ˆ l… x2 ÿ x3 ‡ x4 ‡ x5 Š:
x ‡ 1† ˆ ‰ÿ same overall length as the three-node beam. The non-
2 4 dimensional coordinates (x) of the nodes are ÿ1, ÿ1/3,
…37† +1/3 and +1. Eight shape functions are required, de-
rived from the following Lagrangian polynomials and
The suitability of the above as shape functions can be
derivatives:
demonstrated by plotting each polynomial (Fig. 2).  
The polynomials related to displacements have value 1 9 1 x 2 3
L1 …
x† ˆ ÿ ‡ ‡ x ÿ x …38†
at the associated node and zero elsewhere. Those re- 16 9 9
560 C.E. Augarde / Computers and Structures 68 (1998) 555±560

1
N1 ˆ ‰13 ÿ 15x ÿ 243x2 ‡ 281x3
512
‡ 1215 x4 ÿ 1413x5 ÿ 729x6 ‡ 891
x7 Š …46†

l
N2 ˆ x2 ‡ 19
‰1 ÿ x ÿ 19 x3
256
‡ 99x4 ÿ 99x5 ÿ 81
x6 ‡ 81
x7 Š: …47†

Plotting these two functions [Fig. 3(a), (b)] shows that

they also satisfy the basic requirement of shape func-
tions as outlined above.

5. Conclusion

This work presents an alternative derivation of

bending shape functions for simple beam elements, for
implementation of many-noded straight beam elements
within a ®nite element analysis code. While the el-
ements described are simple, the theory will be of inter-
est to developers of other C1 continuous elements such
as rectangular plates.

Acknowledgements

Fig. 3. Shape functions for a four-node beam element. This work was carried out as part of an EPSRC
funded project examining numerical modelling of tun-
 
27 1 x2 nelling at the Department of Engineering Science,
L2 …
x† ˆ ÿ x ÿ ‡ x3 …39† University of Oxford, UK under the guidance of Dr
16 3 3
H.J. Burd. The author would like to acknowledge the
  help and contribution of Dr Burd to this work.
27 1 x2
L3 …
x† ˆ ‡ x ÿ ÿ x3 …40†
16 3 3
  References
9 1 x
L4 …
x† ˆ ÿ ÿ ‡ x2 ‡ x3 …41†
16 9 9 [1] Astley RJ. Finite Elements in Solids and Structures.
London: Chapman & Hall, 1992.
 
9 1 [2] Augarde CE. Numerical modelling of tunnelling pro-
L 01 …
x† ˆ ‡ 2 x2
x ÿ 3 …42† cesses for assessment of damage to buildings. D.Phil
16l 9
thesis. University of Oxford, 1997.
  [3] Dawe DJ. Matrix and Finite Element Displacement
27 2 Analysis of Structures. Oxford: Clarendon Press, 1984.
L 02 …
x† ˆ x2
ÿ1 ÿ x ‡ 3 …43†
16l 3 [4] Cook RD. Concepts and Applications of Finite Element
Analysis. 2nd ed. Chichester: John Wiley & Sons, 1981.
  [5] Mohr GA. Finite Elements for Solids, Fluids and
27 2
L 03 …
x† ˆ 1 ÿ x ÿ 3
x2
…44† Optimisation. Oxford: Oxford University Press, 1992.
16l 3
[6] Jacques I, Judd C. Numerical Analysis. London,
  Chapman & Hall, 1987.
9 1 [7] Morris JL. Computational Methods in Elementary
L 04 …
x† ˆ ÿ ‡ 2 x2 :
x ‡ 3 …45†
16l 9 Numerical Analysis. Chichester: John Wiley & Sons,
1983.
The ®rst two shape functions, relating to lateral displa- [8] Spanier J, Oldham KB. An Atlas of Functions. London:
cement and rotation at node 1, can then be derived as Hemisphere, 1987.