MECHANICS

RESEARCH COMMUNICATIONS

Mechanics Research Communications 32 (2005) 692–703

www.elsevier.com/locate/mechrescom

A new technique for large deflection analysis

of non-prismatic cantilever beams
Mohammad Dado a, Samir Al-Sadder b,*

a
Department of Mechanical Engineering, Faculty of Engineering, University of Jordan, Amman 11943, Jordan
b
Department of Civil Engineering, Faculty of Engineering, Hashemite University, Zarqa 13115, Jordan

Available online 10 February 2005

Abstract

This paper studies the very large deflection behavior of prismatic and non-prismatic cantilever beams subjected to
various types of loadings. The formulation is based on representing the angle of rotation of the beam by a polynomial
on the position variable along the deflected beam axis. The coefficients of the polynomial are obtained by minimizing
the integral of the residual error of the governing differential equation and by applying the beamÕs boundary conditions.
Several numerical examples are presented covering prismatic and non-prismatic cantilever beams subjected to uniform,
non-uniform distributed loads and tip concentrated loadings in vertical and horizontal directions. The loads considered
in this study are restricted to the non-follower type loads. Cases with different loadings and geometries are compared
with MSC/NASTRAN computer package. However, for some very large deflection case, the MSC/NASTRAN failed
to predict the deflected shape due to divergence problems.

2005 Elsevier Ltd. All rights reserved.

Keywords: MSC/NASTRAN; Non-prismatic cantilever; Slender cantilever beams; Very large deflection

1. Introduction

Closed form solutions of large deflection problems for cantilever beams with general loading conditions
using elliptic integrals are not available. If a non-prismatic beam is considered, the complexity of the prob-
lem becomes much greater. In this situation a numerical scheme is the only approach available. Available
numerical approaches such as finite elements, finite differences and numerical integration are cumbersome
due to cost, difficulties in setting up the proper model and the divergence of solution. This paper presents a
general numerical scheme that yields accurate results for prismatic and non-prismatic cantilever beams

*
Corresponding author. Tel.: +962 795048993.
E-mail address: [email protected] (S. Al-Sadder).

0093-6413/$ - see front matter
2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mechrescom.2005.01.004
 M. Dado, S. Al-Sadder / Mechanics Research Communications 32 (2005) 692–703 693

subjected to various types of loadings. The present scheme is based on the integrated least square error
of the non-linear governing differential equation in which the angle of rotation is presented by a
polynomial.
Most of the research that deals with large deflection problems has used four different approaches. The
first approach is based on elliptic integral formulation (e.g. Barten, 1945; Bisshop and Drucker, 1945;
Timoshenko and Gere, 1961; Lau, 1974; Mattiasson, 1981; Chucheepsakul et al., 1994; Bona and Zelenika,
1997; Wang et al., 1997; Chucheepsakul et al., 1999; Coffin and Bloom, 1999; Ohtsuki and Ellyin, 2001).
This approach is tedious and only suitable for simple loading cases. For example, it cannot solve any
non-prismatic or prismatic beam with simple uniformly distributed load in the vertical or horizontal direc-
tions. The second approach uses numerical integration with iterative shooting techniques (e.g. Freeman,
1946; Conway, 1947; Holden, 1972; Wang and Watson, 1980; Wang, 1981; Watson and Wang, 1981; Wang
and Watson, 1982; Watson and Wang, 1983; Mau, 1990; Wang and Kitipornchai, 1992; Lee and Oh, 2000;
Lee, 2001; Magnusson et al., 2001). However, it is suitable only for beams subjected to loads producing
moderate deflections and it fails in cases incorporating very large deflection.
The third approach utilizes incremental finite element method in connection with Newton–Rhapson
iteration techniques for solving elastic problems (e.g. Schmidt, 1977; Golley, 1984; Kooi, 1985; Golley,
1997). This method requires the use of expensive commercial packages and the generation of a very fine
mesh requiring huge computational time. In addition, the method may experience divergence problems
in very large deflection cases and it requires special types of numerical techniques such as arc-length
method. Moreover, an experience user of this program is needed to setup the model and the solution
method in the proper form.
In the fourth approach, the incremental finite differences method in connection with Newton–Rhapson
iteration techniques is used (e.g. Kooi and Kuipers, 1984; Saje and Srpcic, 1985; Srpcic and Saje, 1986).
This method requires a very large number of nodes for accurate results and it is prone to divergence in very
large deflection cases.
This paper studies the very large deflection behavior of prismatic and non-prismatic cantilever beams
subjected to various types of loadings. The formulation is based on representing the angle of rotation by
a polynomial and substituting it into the governing differential equation and boundary conditions. The gov-
erning differential equation is set in the residual form and the integrated residual error over the deflected
beam axis is minimized. Representative numerical examples are studied covering prismatic and
non-prismatic cantilever beams subjected to uniform, non-uniform distributed loads and tip concentrated
loadings in vertical and horizontal directions. Some of the results were compared with MSC/NASTRAN
for Windows 95 (1995) computer package. However, for very large deflection cases the comparisons cannot
be made due to the divergence of NASTRAN commercial package.

2. Formulation of basic differential equations

Consider an inextensible non-prismatic slender cantilever beam of length L and a variable flexural stiff-
ness EI(s) subjected to variable distributed loads in global x, y-axes as shown in Fig. 1. Where s is the
curved coordinate along the deflected axis of the beam. qx(s) and qy(s) are the horizontal and vertical dis-
tributed loads and Fx, Fy and Me are concentrated tip loads and moment. By definition, these loads are of
non-follower type.
Consider the right segment of the beam as a free body diagram as shown in Fig. 2 where the length of the
segment is equal to (L
s). Horizontal and vertical equilibriums of this segment gives:
Z s¼L
H ðsÞ ¼ qx ðsÞ ds þ F x ð1Þ
s¼s
694 M. Dado, S. Al-Sadder / Mechanics Research Communications 32 (2005) 692–703

y Fy
Me

Fx

qy (s)
L

qx (s)

s
x

Fig. 1. Non-prismatic cantilever beam subjected to general loading conditions.

y Fy
Me

Fx

qy (s)

qx (s)
M (s)
H (s)
V (s) M (s)
V (s)
s H (s)

Fig. 2. Free body diagram of a beam segment.

Z s¼L
V ðsÞ ¼ qy ðsÞ ds þ F y ð2Þ
s¼s

Consider an infinitesimal element of length ds of the beam as shown in Fig. 3. The internal forces that act
on this element are the horizontal force H, the vertical force V and the bending moment M. x and y are the
coordinates of a point on this beam. Angle h represents the angle of rotation of the beam with respect to the
positive x-axis as shown in Fig. 3.
Moment equilibrium equation of the infinitesimal element yields:
dM
ds þ V dx
H dy ¼ 0 ð3Þ
ds
or
dM dy dx
¼H
V ð4Þ
ds ds ds
But from geometrical relationship of the infinitesimal element (Fig. 3):
dy dx
¼ sin h; ¼ cos h ð5Þ
ds ds
 M. Dado, S. Al-Sadder / Mechanics Research Communications 32 (2005) 692–703 695

y dV
V+ ds
ds
dM
M+ ds
ds
ds
qy(s) dH
H+ ds
ds

dy

M
θ qx(s)

H
V dx

Fig. 3. Internal and external forces acting on an infinitesimal element (ds) in global coordinate.

Then
dM
¼ H sin h
V cos h ð6Þ
ds
Substituting Eqs. (1) and (2) into Eq. (6), results in
Z s¼L  Z s¼L 
dM
¼
qy ðsÞ ds þ F y cos h þ qx ðsÞ ds þ F x sin h ð7Þ
ds s¼s s¼s

From Euler–Bernoulli law, which states that local bending moment is proportional to local curvature;
dh
MðsÞ ¼ EIðsÞ ð8Þ
ds
Differentiating Eq. (8) once with respect to s, leads to
dM d2 h dIðsÞ dh
¼ EIðsÞ 2 þ E ð9Þ
ds ds ds ds
Substituting Eq. (9) into Eq. (7), results in
Z s¼L  Z s¼L 
d2 h dIðsÞ dh
EIðsÞ 2 þ E þ qy ðsÞ ds þ F y cos h
qx ðsÞ ds þ F x sin h ¼ 0 ð10Þ
ds ds ds s¼s s¼s

Eq. (10) is the differential equation governing the large deflection behavior of non-prismatic cantilever
beams subjected to various types of loadings.

3. Method of solution

The numerical solution scheme for Eq. (10) is based on approximating the angle of rotation (h) along the
deflected beam axis by a polynomial on the non-dimensional parameter
s

s ¼ ð11Þ
L
The polynomial approximation is defined as
X
i¼N
hð
sÞ ¼ ai
si ð12Þ
i¼0
696 M. Dado, S. Al-Sadder / Mechanics Research Communications 32 (2005) 692–703

where ai(i = 0, 1, 2, . . . N) are unknown coefficients to be determined by the solution procedure and N is the
order of the polynomial approximating the angle of rotation. Substituting the approximation of h(s) and its
derivates into Eq. (10) leads to a residual error e(s) defined as
Z 1  !
X N
dIð
sÞ X N XN
i
2 i
1 i
eð
sÞ ¼ EIð
sÞ iði
1Þai
s þ E iai
s þ qy ð
sÞ d
s þ F y cos ai
s
i¼0
d
s i¼0
s i¼0
Z 1  !
XN
i

qx ð
sÞ d
s þ F x sin ai
s ð13Þ

s i¼0

where eð
sÞ is the error at position
s along the deflected axis of the beam. The aim of the procedure is to find
the optimum values of ai that minimize the integrated error over the deflected beam axis. The integrated or
total residual error is defined as
Z 1
TRE ¼ ðeð
sÞÞ2 d~
s ð14Þ
0

The polynomial coefficients are obtained by minimizing the total residual error TRE by setting oTRE
oai
¼ 0,
(i = 0, 1, 2, . . . , N) which in turn leads to the following set of simultaneous equations on ai:
Z 1
oeð
sÞ
eð
sÞ d
s ¼ 0 i ¼ 0; 1; 2; . . . ; N ð15Þ
0 oai

The boundary conditions for the cantilever beam are as follows:

(1) Zero slope at the fixed end;

hð0Þ ¼ 0 ð16Þ
(2) Zero deflection at the fixed end;
yð0Þ ¼ 0 ð17Þ
(3) End moment at the free moment is equal to the applied external moment;

EIð1Þ dh
Mð1Þ ¼ ¼ Me ð18Þ
L d
s 
s¼1

Since the internal shear equation is used in expressing the governing differential equation, the shear
boundary conditions should not be included here. Writing these conditions in terms of the approximated
angle of rotation, leads to the following relations;

ao ¼ 0 ð19Þ

X
N
M eL
iai
si
1 ¼ ð20Þ
i¼0
EIð1Þ

where Eqs. (19) and (20) correspond to Eqs. (16) and (18), respectively. Eq. (17) is clearly satisfied by
observing Eq. (5).
 M. Dado, S. Al-Sadder / Mechanics Research Communications 32 (2005) 692–703 697

The different load components are define in a non-dimensional form as,

qx ð
sÞL3
qx ð
sÞ ¼


EIð0Þ
qy ð
sÞL3
qy ð
sÞ ¼


EIð0Þ
F x L2 ð21Þ
Fx ¼
EIð0Þ
F y L2
Fy ¼
EIð0Þ
M eL
Me ¼
EIð0Þ
Eqs. (15), (19) and (20) are solved simultaneously using Newton–Raphson iterative method for the
unknown parameters ai (i = 0, 1, 2, . . . N). The form of these equation allows the exact formulation for
the residual vector and the Jacobian matrix so that numerical differentiation is not used. However, a numer-
ical integration scheme using 16 points Gaussian Quadrature is utilized in performing the required
integrals.

4. Large displacement finite element analysis using MSC/NASTRAN

As a check for the results of the present study, a large displacement finite element analysis using the
multi-purpose computer program MSC/NASTRAN is performed by dividing the cantilever beam into
sufficient number of two-dimensional beam elements depending on beam geometry and loading conditions.
The following fundamental parameters are introduced to the program:

(1) Number of loading increments = 50,

(2) Maximum number of iterations during each increment = 25,
(3) Number of iterations before updating tangent stiffness matrix = 15,
(4) Method of updating tangent stiffness matrix = iterative,
(5) Tolerance for displacement convergence criterion = 1*10
4,
(6) Solution strategy = conventional Newton–Raphson iteration method.

5. Numerical examples and discussions

Several representative examples are presented in this section. These examples demonstrate the robustness
of the present technique in dealing with very large deflection problems. As stated earlier, these examples
were also solved using MSC/NASTRAN for comparison. Unfortunately, MSC/NASTRAN had failed
in predicting cases with very large deflections. The running time for MSC/NASTRAN examples was more
than one hour on the average while the present technique required only few seconds using the same com-
puting facility, namely, P4-2.4 GHz-PC.
Example 1. In this example, a prismatic slender cantilever beam is subjected to uniformly distributed load
in the vertical direction as shown in Fig. 4. Several intensity values of the load are considered. Fig. 5 shows
the deflection shapes for the loads
qy ð
sÞ ¼
4,
10,
20,
40,
100. This example was solved using MSC/
698 M. Dado, S. Al-Sadder / Mechanics Research Communications 32 (2005) 692–703

qy(s)

x
constant EI
s
L

Fig. 4. Prismatic cantilever beam subjected to uniformly distributed load in y-direction.

0.1

0
Prsent Study
MSC/NASTRAN
-0.1

-0.2

-0.3 qy = −4

-0.4
y/L

-0.5

-0.6

-0.7 -10

-0.8
-20
-0.9 -40
-100
-1
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x/L

Fig. 5. Deflection shapes for Example 1.

NASTRAN with 100 two-dimensional beam elements. Both results up to q
y ð
sÞ ¼
20 are in excellent
agreement. However, for
qy ð
sÞ ¼
40 and
100, the MSC/NASTRAN failed in predicting results due to
divergence problem. The results shown in Fig. 5 emphasize the stability of the present technique especially
qy ð
sÞ ¼
100 where the beam is almost aligned with the load for a large portion of the beam
for the case of

length. The order of the polynomial for the case of
qy ð
sÞ ¼
100 is (N = 15) and the total residual error
TRE as defined in Eq. (14) is equal to (2 · 10
4).

Example 2. This example is similar to Example (1) except that the beam is non-prismatic but tapered
as shown in Fig. 6. Fig. 7 shows the deflection shapes for the loads
qy ð
sÞ ¼
4;
6;
8;

10;
12;
14;
16;
20;
30;
40;
100. This example was solved using MSC/NASTRAN with 200
two-dimensional beam elements. Both results up to
qy ð
sÞ ¼
12 are in excellent agreement. However,
qy ð
sÞ ¼
14 to
100, the MSC/NASTRAN failed in predicting results due to divergence problem. Even
for

with the presence of tapered cross-section the results as shown in Fig. 7 are stable and reach the physical
limit of the deflection shape. The order of the polynomial for the case of
qy ð
sÞ ¼
100 is (N = 15) and the
total residual error is (1.5 · 10
4).
 M. Dado, S. Al-Sadder / Mechanics Research Communications 32 (2005) 692–703 699

qy(s)

A
x
A Variable EI
s
L

h
h1 h2
b
Section A-A

Fig. 6. Tapered cantilever beam subjected to uniformly distributed load in the y-direction.

0
Present Study
-0.1 MSC/NASTRAN

-0.2

-0.3

-0.4
qy = −4
y/L

-0.5

-0.6
-6
-0.7
-8
-10
-0.8 -12
-14
-16
-20
-0.9 -30
-40
-100
-1
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x/L

Fig. 7. Deflection shapes for Example 2.

Example 3. The loading case for this example is a linearly distributed load q
y ð
sÞ in the vertical direction
and a uniformly distributed load
qx ð
sÞ in the horizontal direction and the cantilever beam is a prismatic
as shown in Fig. 8. The maximum intensity of the linearly distributed vertical load
qy ð
sÞ was held at a con-
stant value of
5, while the uniformly distributed horizontal load
qx ð
sÞ was given the values

1,
5,
10,
15,
20.
30,
50,
80,
100,
120. Fig. 9 shows the deflection shapes for the different load-
ing combinations. This example was solved using MSC/NASTRAN with 100 two-dimensional beam ele-
qx ð
sÞ ¼
20 are in excellent agreement. However, for
qx ð
sÞ ¼
30 to
120, the
ments. Both results up to

MSC/NASTRAN failed in predicting results due to divergence problem. For the case of
qx ð
sÞ ¼
120,
the order of the polynomial is (N = 15) and the total residual error equals to (1.4 · 10
3).
700 M. Dado, S. Al-Sadder / Mechanics Research Communications 32 (2005) 692–703

qy(s)
y

x
s Constant EI qx(s)
L

Fig. 8. Prismatic cantilever beam subjected to linearly distributed load in the y-direction and uniformly distributed load in the
x-direction.

0
qy = −5
qx = −1
-0.2 -120
-100
-80
-0.4
-50
y/L

-0.6 -30 -5

-20
-0.8
Present Study
-15 -10
MSC/NASTRAN
-1
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
x/L

Fig. 9. Deflection shapes for Example 3.

Example 4. A tip concentrated loads in the vertical and the horizontal directions: F y , F x along with a tip
concentrated moment M e were applied on a prismatic cantilever beam as shown in Fig. 10. Several different
combination of these loads were examined. These loads were selected in such away to produce deflection
shape with an inflection point. Fig. 11 shows these load combinations and the resulting deflection shapes.
As in the previous examples the MSC/NASTRAN was limited in its deflection shapes prediction. For the
case of F x ¼
7, F y ¼ 14 and M e ¼
7, the order of the polynomial is (N = 12) and the total residual error
equals to (6.6 · 10
5).

Example 5. A tapered cantilever beam with several taper ratios is considered. The load consist of a uni-
qy ð
sÞ and a tip concentrated moment M e as shown in Fig. 12. The loads
formly distributed vertical load

were held constant while the taper ratio was given the values 0.2, 0.3, 0.4,. 0.6, 0.8 and 1. Fig. 13 illustrates
the large deflection beam behavior for these different taper ratios. This example was solved using MSC/
NASTRAN with 200 two-dimensional beam elements. Both results were in excellent agreement for taper
ratios from 0.4 to 1. However, for taper ratios 0.2 and 0.3, the MSC/NASTRAN failed in predicting results

y
Fy

Fx
x
Constant EI
s Me
L

Fig. 10. Prismatic cantilever beam subjected to tip concentrated forces and tip moment.
 M. Dado, S. Al-Sadder / Mechanics Research Communications 32 (2005) 692–703 701

0.9
Present Study
MSC/NASTRAN
0.8

0.7

0.6 -4
-6 8
12 -4
0.5 -7 -6
y/L

14
-7 -2
0.4
4
-2
0.3

0.2

0.1 Fx = – 1
Fy = 2
0 Me = – 1

-0.1
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x/L

Fig. 11. Deflection shapes for Example 4.

qy(s)

A Me
x
A Variable EI
s
L

h
h1 h2
b
Section A-A

Fig. 12. Tapered cantilever beam subjected to uniformly distributed load in the y-direction and to tip concentrated moment.

due to divergence problem. It is observed that for a taper ratio of 0.3 the beam made half a loop in the tip
region and two loops for the case of taper ratio 0.2. This is due to low flexural stiffness at the tip. For the
case of taper ratio = 0.2, the order of the polynomial is (N = 12) and the total residual error equals to
(8 · 10
6). The accurate prediction of this complicated deflection behavior demonstrates the strength of
the present method.
702 M. Dado, S. Al-Sadder / Mechanics Research Communications 32 (2005) 692–703

0.1

Present Study
-0.1 MSC/NASTRAN

-0.2

-0.3

-0.4
y/L

-0.5
h2 /h1 =0.2
-0.6
0.3
-0.7 q = –50 0.4
y

Me = 1 1.0
-0.8 0.8 0.6

-0.9

-1
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x/L

Fig. 13. Deflection shapes for Example 5.

6. Conclusion

A robust and stable numerical scheme for the solution of very large deflection problem for prismatic and
non-prismatic slender cantilever beams is presented in this paper. The method is based on approximating
the angle of rotation of the deflected beam axis by an Nth order polynomial. The governing differential
equation along with the beams boundary conditions are used to evaluate the polynomial coefficients. Sev-
eral examples ranging from moderately large deflection to very large deflection were considered. It was
shown that the presented scheme handles extreme cases with high accuracy and solution stability while
the finite element method as depicted by MSC/NASTRAN failed in predicting these extreme cases. A rec-
ommended extension for this work is the application of this scheme on beams with different boundary con-
ditions such as: pinned-pinned, fixed-pinned, and fixed-fixed. In addition, continuous beams may also be
considered using the same scheme.

References

Barten, H.J., 1945. On the deflection of a cantilever beam. Quart. J. Appl. Math. 3, 275.
Bisshop, K.E., Drucker, D.C., 1945. Large deflections of cantilever beams. Quart. J. Appl. Math. 3, 272–275.
Bona, F., Zelenika, S., 1997. A generalized elastica-type approach to the analysis of large displacements of spring-strips. Proc. Instn.
Mech. Engrs. Part C 21, 509–517.
Chucheepsakul, S., Buncharoen, S., Wang, C.M., 1994. Large deflection of beams under moment gradient. ASCE J. Eng. Mech. 120,
1848.
Chucheepsakul, S., Wang, C.M., He, X.Q., Monprapussorn, T., 1999. Double curvature bending of variable-arc-length elastica. J.
Appl. Mech. 66, 87–94.
Coffin, D.W., Bloom, F., 1999. Elastica solution for the hygrothermal buckling of a beam. Int. J. Non-lin. Mech. 34, 935.
Conway, H.D., 1947. Large deflection of simply supported beams. Philos. Mag. 38, 905.
Freeman, J.G., 1946. Mathematical theory of deflection of beam. Philos. Mag. 37, 551.
Golley, B.W., 1984. The finite element solution of a class of elastica problems. Comput. Meth. Appl. Mech. Eng. 46, 159–168.
Golley, B.W., 1997. The solution of open and closed elasticas using intrinsic coordinate finite elements. Comput. Meth. Appl. Mech.
Eng. 146, 127–134.
 M. Dado, S. Al-Sadder / Mechanics Research Communications 32 (2005) 692–703 703

Holden, J.T., 1972. On the finite deflections of thin beams. Int. J. Solids Struct. 8, 1051–1055.
Kooi, B.W., Kuipers, M., 1984. A unilateral contact problem with the heavy elastica. Int. J. Non-lin. Mech. 19, 309–321.
Kooi, B.W., 1985. A unilateral contact problem with the heavy elastica solve by use of finite element. Int. J. Non-lin. Mech. 21, 95–103.
Lau, J.H., 1974. Large deflections of beams with combined loads. ASCE J. Eng. Mech. Div. 12, 140.
Lee, B.K., Oh, S.J., 2000. Elastica and buckling load of simple tapered columns with constant volume. Int. J. Solid Struct. 37, 2507–
2518.
Lee, K., 2001. Post-buckling of uniform cantilever column under a combined load. Int. J. Mech. Non-lin. Mech. 36, 813–816.
Magnusson, A., Ristinmaa, M., Ljung, C., 2001. Behavior of the extensible elastica column. Int. J. Solid Struct. 38, 8441–8457.
Mattiasson, K., 1981. Numerical results from large deflection beam and frame problems analysis by means of elliptic integrals. Int. J.
Num. Meth. Eng. 16, 145.
Mau, S.T., 1990. Elastica solution of braced struts. J. Eng. Mech. 116, 688.
MSC/NASTRAN for Windows 95, 1995. McNeal-Schwendler Corporation, CA, USA.
Ohtsuki, A., Ellyin, F., 2001. Analytical approach to large deformation problems of frame structures (in case of a square frame with
rigid joints). JSME Int. J. 44, 89–93.
Saje, M., Srpcic, S., 1985. Large deformations of in-plane beams. Int. J. Solids Struct. 21, 1181.
Schmidt, W.F., 1977. Nonlinear bending of beams using the finite element method. Int. J. Comput. Struct. 8, 153.
Srpcic, S., Saje, M., 1986. Large deformations of thin curved plane beam of constant initial curvature. Int. J. Mech. Sci. 28, 275.
Timoshenko, S.P., Gere, J.M., 1961. Theory of Elastic Stability. McGraw-Hill, NY.
Wang, C.M., Kitipornchai, S., 1992. Shooting-optimization technique for large deflection analysis of structural members. Eng. Struct.
14, 231–240.
Wang, C.Y., Watson, L.T., 1980. On large deformations of C-shaped springs. Int. J. Mech. Sci. 22, 395–400.
Wang, C.Y., 1981. Large deformations of a heavy cantilever. Quart. J. Appl. Math. 39, 261–273.
Wang, C.Y., Watson, L.T., 1982. The elastic catenary. Int. J. Mech. Sci. 24, 349–357.
Wang, C.M., Lam, K.Y., He, X.Q., Chucheepsakul, S., 1997. Large deformation of an end supported beam subjected to a point load.
Int. J. Mech. Non-lin. Mech. 32, 63–72.
Watson, L.T., Wang, C.Y., 1981. Hanging an elastic ring. Int. J. Mech. Sci. 23, 161–167.
Watson, L.T., Wang, C.Y., 1983. Periodically supported heavy elastica sheet. ASCE J. Eng. Mech. Div. 109, 811.