Number 10 Volume 19 October 2013 Journal of Engineering

Exact Stiffness Matrix for Nonprismatic Beams with Parabolic

Varying Depth
Dr.Musab Aied Qissab Al-Janabi
Lecturer/ Department of Civil Engineering
AL-Nahrain University

ABSTRACT
In this paper, an exact stiffness matrix and fixed-end load vector for nonprismatic beams having parabolic
varying depth are derived. The principle of strain energy is used in the derivation of the stiffness matrix.
The effect of both shear deformation and the coupling between axial force and the bending moment are
considered in the derivation of stiffness matrix. The fixed-end load vector for elements under uniformly
distributed or concentrated loads is also derived. The correctness of the derived matrices is verified by
numerical examples. It is found that the coupling effect between axial force and bending moment is
significant for elements having axial end restraint. It was found that the decrease in bending moment was
in the range of 31.72%-42.29% in case of including the effect of axial force for the studied case. For
midspan deflection, the decrease was 46.07% due to the effect of axial force generated at supports as a
result of axial restraint.

KEYWORDS: Stiffness; Parabolic; Shear deformation; Axial force; Beams

‫ﺍﻟﺨﻼﺻﺔ‬

‫ ﺗﻢ ﺍﺷﺘﻘﺎﻕ ﻣﺼﻔﻮﻓﺔ ﺍﻟﺠﺴﺎءﺓ ﻭ ﻣﺘﺠﻪ ﺣﻤﻞ ﺍﻟﻨﻬﺎﻳﺎﺕ ﺍﻟﻤﺜﺒﺘﺔ ﻟﻸﻋﺘﺎﺏ ﺍﻟﻐﻴﺮ ﻣﻮﺷﻮﺭﻳﺔ ﺫﺍﺕ ﺍﻟﻌﻤﻖ ﺍﻟﻤﺘﻐﻴﺮ ﻻﺧﻄﻴﺎ )ﻗﻄﻊ‬،‫ﻓﻲ ﻫﺬﺍ ﺍﻟﺒﺤﺚ‬
‫ ﺗﻢ ﺍﺳﺘﻌﻤﺎﻝ ﻣﺒﺪﺃ ﻁﺎﻗﺔ ﺍﻹﻧﻔﻌﺎﻝ ﻓﻲ ﺍﺷﺘﻘﺎﻕ ﻣﺼﻔﻮﻓﺔ ﺍﻟﺠﺴﺎءﺓ ﻣﻊ ﺍﻷﺧﺬ ﺑﻨﻈﺮ ﺍﻹﻋﺘﺒﺎﺭ ﺗﺄﺛﻴﺮ ﺗﺸﻮﻫﺎﺕ ﺍﻟﻘﺺ ﻭﺍﻟﺘﺄﺛﻴﺮ ﺍﻟﻤﺘﺒﺎﺩﻝ ﺑﻴﻦ‬.(‫ﻣﻜﺎﻓﻰء‬
‫ ﺗﻢ‬.‫ ﺗﻢ ﺍﺷﺘﻘﺎﻕ ﻣﺘﺠﻪ ﺣﻤﻞ ﺍﻟﻨﻬﺎﻳﺎﺕ ﺍﻟﻤﺜﺒﺘﺔ ﻟﻌﻨﺎﺻﺮ ﺫﺍﺕ ﺍﺣﻤﺎﻝ ﻣﻮﺯﻋﺔ ﺑﺈﻧﺘﻈﺎﻡ ﺍﻭ ﺃﺣﻤﺎﻝ ﻣﺮﻛﺰﺓ‬،‫ ﻛﺬﻟﻚ‬.‫ﺍﻟﻘﻮﺓ ﺍﻟﻤﺤﻮﺭﻳﺔ ﻭﻋﺰﻡ ﺍﻹﻧﺤﻨﺎء‬
‫ ﻣﻦ ﺧﻼﻝ ﺍﻟﻨﺘﺎﺋﺞ ﺍﻟﺘﻲ ﺗﻢ ﺍﻟﺤﺼﻮﻝ ﻋﻠﻴﻬﺎ ﻭﺟﺪ ﺑﺄﻥ ﺍﻟﺘﺄﺛﻴﺮ‬.‫ﺍﺧﺘﻴﺎﺭ ﻋﺪﺓ ﺍﻣﺜﻠﺔ ﺗﻄﺒﻴﻘﻴﺔ ﻟﻐﺮﺽ ﺍﺛﺒﺎﺕ ﺻﺤﺔ ﺍﻟﻤﺼﻔﻮﻓﺎﺕ ﺍﻟﺘﻲ ﺗﻢ ﺍﺷﺘﻘﺎﻗﻬﺎ‬
‫ ﻓﻘﺪ ﻟﻮﺣﻆ ﺗﻨﺎﻗﺼﺎ ﻓﻲ ﻣﻘﺪﺍﺭ ﻋﺰﻡ‬.‫ﺍﻟﻤﺘﺒﺎﺩﻝ ﻟﻠﻘﻮﺓ ﺍﻟﻤﺤﻮﺭﻳﺔ ﻣﻊ ﻋﺰﻡ ﺍﻹﻧﺤﻨﺎء ﺫﺍﺕ ﻓﺎﻋﻠﻴﺔ ﺑﺎﻟﻨﺴﺒﺔ ﻟﻸﻋﻀﺎء ﺫﺍﺕ ﺍﻟﻨﻬﺎﻳﺎﺕ ﺍﻟﻤﻘﻴﺪﺓ ﻣﺤﻮﺭﻳﺎ‬
‫ ﺃﻣﺎ ﺑﺎﻟﻨﺴﺒﺔ‬.‫ ﻓﻲ ﺣﺎﻝ ﺗﻢ ﺍﺩﺧﺎﻝ ﺗﺄﺛﻴﺮ ﺍﻟﻘﻮﺓ ﺍﻟﻤﺤﻮﺭﻳﺔ ﺑﺎﻟﻨﺴﺒﺔ ﻟﻠﺤﺎﻟﺔ ﺍﻟﺘﻲ ﺗﻢ ﺩﺭﺍﺳﺘﻬﺎ‬42.29% ‫ ﺍﻟﻰ‬31.72% ‫ﺍﻹﻧﺤﻨﺎء ﻭﺑﻨﺴﺐ ﺗﺘﺮﺍﻭﺡ ﻣﻦ‬
‫ ﻓﻲ ﻣﻨﺘﺼﻒ ﺍﻟﻔﻀﺎء ﻧﺘﻴﺠﺔ ﻟﺘﺄﺛﻴﺮ ﺍﻟﻘﻮﺓ ﺍﻟﻤﺤﻮﺭﻳﺔ ﺍﻟﻤﺘﻮﻟﺪﺓ ﻣﻦ ﺗﻘﻴﻴﺪ ﺍﻟﺤﺮﻛﺔ ﺍﻟﻤﺤﻮﺭﻳﺔ ﻟﻠﻤﺴﺎﻧﺪ‬46.07% ‫ ﻓﻘﺪ ﻛﺎﻥ ﺍﻟﺘﻨﺎﻗﺺ ﺑﻨﺴﺒﺔ‬،‫ﻟﻠﻬﻄﻮﻝ‬
.‫ﻣﻘﺎﺭﻧﺔ ﻣﻊ ﺍﻟﺤﺎﻟﺔ ﺍﻟﺘﻲ ﺗﻢ ﻓﻴﻬﺎ ﺍﻫﻤﺎﻝ ﺗﺄﺛﻴﺮﻫﺎ‬

‫ ﺍﻷﻋﺘﺎﺏ‬،‫ ﺍﻟﻘﻮﺓ ﺍﻟﻤﺤﻮﺭﻳﺔ‬،‫ﺗﺸﻮﻫﺎﺕ ﺍﻟﻘﺺ‬،‫ ﻗﻄﻊ ﻣﻜﺎﻓﻰء‬،‫ ﺍﻟﺠﺴﺎءﺓ‬:‫ﺍﻟﻜﻠﻤﺎﺕ ﺍﻟﺮﺋﻴﺴﻴﺔ‬

1212
Musab Aied Qissab Al-Janabi Exact Stiffness Matrix for Nonprismatic
Beams with Parabolic Varying Depth

INTRODUCTION
Members with variable depth are used in many
engineering structures such as highway bridges, PROBLEM STATEMENT
buildings, as well as in many mechanical
components and aerospace engineering structures. Consider a nonprismatic Euler- Bernoulli beam
In civil engineering construction, nonprismatic element of length L as shown in Fig. 1(a). The
members are frequently used to optimize material element is rectangular in cross-section and has a
distribution and stresses, increase the overall parabolic varying depth and constant width. Three
stability and stiffness, reduce the dead load degrees of freedom are assumed at each node.
positive moment and deflection, and sometimes to Only the deformation in the plane of the element
satisfy architectural requirements. Accordingly, and the bending moment about the centroidal
the analysis of structures having nonprismatic main axis are considered. The positive direction of
elements is of interest in structural, mechanical, displacements and forces are as shown in Fig.
and aerospace engineering. The analysis of 1(b).
nonprismatic members is covered in several
publications (e.g., Timoshenko and Young 1965; The stiffness components corresponding to the
AL-Gahtani 1996; Al-Gahtani and Khan 1998; degrees of freedom shown in Fig. 1(b) can be
Luo et al 2007). Most of the available publications obtained by using Castigliano's second theorem
deal with the analysis of tapered members only. (Boresi, A.P. and Schmidt, R.J. 2003), which
Some particular cases (e.g., Timoshenko and states that the deflection caused by an external
Young 1965; AL-Gahtani 1996) deal with the force is equal to the partial derivative of the strain
analysis of nonprismatic beams having parabolic energy (U) with respect to that force. The total
varying depth. However, these cases are limited to strain energy (U) for the element shown in Fig.
the analysis only (no stiffness matrix derivation) 1(a) including the strain energy caused by bending
of such type of members involving lengthy and moment, shear and axial forces can be given by
tedious calculations which are not applicable for 2 2
1 L Mx 1.2 L Q x
use in the analysis packages in which the analysis= U ∫ dx + ∫ dx
2E 0 I x 2Gb 0 h 2
is based on matrix operations. In addition, the x (1)
available analytical solutions do not consider the 2
1 L Px
effect of shear deformation and the axial force- + ∫ dx
bending moment interaction. The other 2Eb 0 hx
alternative publications deal with the numerical where Mx, Qx, Px, Ix, hx, are the bending moment,
methods of analysis such as the finite element shear force, axial force, moment of inertia and the
method (e.g., Bathe 1996) in which the member is depth of the element at the distance x respectively;
discretized to a number of elements and the b, E, G, are the width of the element, Young's and
stiffness matrices of the elements are assembled to shear modulus of elasticity, respectively. The
obtain the stiffness matrix for the whole member. bending moment Mx, shear force Qx, and the axial
The main disadvantage resulting from member force Px can be found from equilibrium as
discretization is the large number of input data follows
required even for simple structures.
The purpose of this paper is to present an 1
=Mx Pi h0 (cx 2 ) + Q i x − M i (2a)
exact stiffness matrix for nonprismatic beam 2
elements with parabolic varying depth including Q x = Qi (2b)
the effect of shear deformation and the axial Px = Pi (2c)
force-bending moment interaction. The
The moment of inertia and the depth of the
correctness of the derived stiffness matrix and the
element cross section at a distance x can be given
fixed-end load vector is examined through
by
numerical examples.
bh 3
I x = 0 (1 + cx 2 )3 =I 0 (1 + cx 2 )3 ;=
hx h0 (1 + cx 2 )
12

1213
 Number 10 Volume 19 October 2013 Journal of Engineering

(3) (6)
The partial derivative of the strain energy (U)
where h0=the minimum depth of the element (at with respect to Pi, Qi, and Mi can be given
the origin), c=(h1-h0)/(h0L2), and h1= the max- respectively as follows
imum depth of the element.
Substituting eqs. (2a), (2b), (2c), and (3) into
eq. (1) and after integrations, the following exact   h0 2  
expression for the strain energy can be given as ∂U 1  Pi  2I 0a0 + (a1 )  + 

= u=  2  (7)
∂Pi 2EI 0   
i

Q i ( h0 (a4 ) ) − M i ( h0 (a2 ) ) 
1  2 Pi 2 h0 2
=
U  Pi I 0 (a0 ) + (a1 ) − Pi M i h0 (a2 )
2EI 0  4
∂U 1  Pi ( h0 (a4 ) ) + Q i ( 2(a6 ) + 4EI 0 K v ) 
+ M i 2 (a3 ) + Pi Q i h0 (a4 ) − M i Q i (a5 ) = v=  
∂Q i
i

2EI 0  −M i (a5 ) 
+ Q i 2 (a6 + 2EI 0 K v )  (4) (8)


∂U 1
where θi =  −Pi ( h0 (a2 ) ) − Q i ( a5 ) + M i (2a3 ) 
=
∂M i 2EI 0
(9)
φ
a0 = 0 (5a) where ui, vi, and θi are the displacement
c A0
components in horizontal ,vertical directions, and
rotation angle at node (i) respectively.
1 3  1 1  1 3 
  φ0 − sin(2φ0 )  − sin φ0 cos φ0  (5b)
a1 = The stiffness coefficient (kij) of an element can
c 4  2 4  4  be defined as the force or moment at node (i)
required to induce a unit displacement or rotation
at node (j) with all other displacements equal to
1 1 1  zero. Therefore, eqs. (7), (8), and (9) will be used
=a2  2 φ0 − 8 sin(4φ0 )  (5c)
to derive the stiffness matrix of the element.
4 c  
Writing eqs. (7), (8), and (9) in a matrix form
1 3  1 1  1  yield the following
a3 =   φ0 + sin(2φ0 )  + cos φ0 sin φ0  (5d)
3
c 4  2 4  4 
 h0 2 
 1 (2I 0a0 + (a1 )) ( h0 (a4 ) ) − ( h0 (a2 ) ) 
  i 
1 1 1 P
=
a4  − +  (5e)  2

 2(1 + cL ) (1 + cL ) 2  ( h0 (a4 ) ) ( 2(a6 ) + 4EI 0 K v )
2 2 2
2c  −(a5 )   Q i 
 
 − ( h0 (a2 ) ) −(a5 ) (2a3 )  M i 

1  1   
=
a5 1 − 2 2
(5f)
2c  (1 + cL ) 
u i 
 
= 2EI 0 v i 
1 θ 
a6 = (a2 ) (5g)  i 
c
(10)
and
or
1.2 φ0
Kv = , φ0 = tan −1 ( c L ) ,and A 0 = bh0
2GA 0 c
[ D ]{F } = 2EI 0 {δ } ; {F } = 2EI 0 [ D ]−1 {δ }

1214
 Musab Aied Qissab Al-Janabi Exact Stiffness Matrix for Nonprismatic
Beams with Parabolic Varying Depth

(11) h1 − h0
k 61 =M j =k 11 ( ) + k 21L − k 31 (14f)
2
where

where
d 11 d 12 d 13 
[ D ] = d12 d 22 d 23  = ( )
λ d 11 d 33d 22 − d 232 − d 12 (d 33d 12 − d 13d 23 ) +
d 13 d 23 d 33  d 13 (d 23d 12 − d 13d 22 )
 h0 2  (15)
(2I 0 a0 + (a1 )) ( h0 (a4 ) ) − ( h0 (a2 ) ) 
 2 
=  ( h0 (a4 ) ) ( 2(a6 ) + 4EI 0 K v ) −(a5 )  FLEXURAL STIFFNESS
 
 − ( h0 (a2 ) ) −(a5 ) (2a3 )  Following the same procedure given for the
  derivation of axial stiffness, the flexural
(12) (translational and rotational) stiffness coefficients
can be obtained by applying a unit lateral
displacement or a unit rotation (with all other
displacements equal to zero) to obtain the
 Pi  u i  translational or rotational stiffness coefficients,
{F } =  Q i  
 ; {δ } = 2EI 0 v i

 (13) respectively.
M  θ 
 i   i 
Hence, by substituting ui=0, vi= 1, and θi = 0 in
the displacements vector δ , the translational
stiffness coefficients can be given as
AXIAL STIFFNESS
Applying a unit axial displacement at node (i) −2EI 0
= P=
k 12 (d 33d 12 − d 13d 23 ) (16a)
λ
with all other displacements equal to zero (i.e. put i

ui=1, vi= 0, and θi = 0 in the displacements vector

δ ), the stiffness coefficients corresponding to = Q=
k 22
2EI 0
(d 33d 11 − d 132 ) (16b)
λ
i
that displacement can be found by solving eq. (11)
for the column matrix F , hence
−2EI 0
k= =
M (d 23d 11 − d 13d 12 ) (16c)
λ
32 i
2EI 0
= P=
k 11 (d 33d 22 − d 23 )
2
(14a)
λ
i

−2EI 0
k=
21 Q= (d 33d 12 − d 13d 23 ) (14b) From equilibrium, the stiffness coefficients or
λ
i
the force vector at node (j) corresponding to the
2EI 0
unit lateral displacement at node (i) (i.e. ui=0, vi=
k= =
M (d 23d 12 − d 13d 22 ) (14c) 1, and θi = 0 ) can be given as
λ
31 i

From equilibrium, the force vector (or stiffness k 42 =

−k 12 =
Pj (16d)
coefficients) at node (j) corresponding to the unit
axial displacement at node (i) can be given as
k 52 = Q j = −k 22 (16e)

h1 − h0
k 41 = P j = −k 11 (14d) k 62 =M j =k 12 ( ) + k 22 L − k 32 (16f)
2

Similarly, the rotational stiffness coefficients can

k 51 = Q j = −k 21 (14e) be obtained by substituting ui=0, vi= 0, and

1215
 Number 10 Volume 19 October 2013 Journal of Engineering

θi = 1 in the displacement vector [δ ] and solving 2EI 0

=k 35 (d 23d 11 − d 13d 12 ) (19c)
eq. (11) as follows λ

2EI 0 −2EI 0
= P=
k 13 (d 23d 12 − d 13d 22 ) (17a)=k 45 (d 33d 12 − d 13d 23 ) (19d)
i
λ λ

−2EI 0 2EI 0
k=
23 Q= (d 23d 11 − d 13d 12 ) (17b) =k 55 (d 33d 11 − d 132 ) (19e)
i
λ λ

2EI 0 h1 − h0
k= =
M (d 22d 11 − d 12 2 ) (17c) =
k 65 k 15 ( ) + k 25 L − k 35 (19f)
λ
33 i 2

and from equilibrium h1 − h0

=
k 16 k 11 ( ) + k 21L − k 31 (20a)
2
k 43 = P j = −k 13 (17d)
h1 − h0
=
k 26 k 12 ( ) + k 22 L − k 32 (20b)
2
k 53 = Q j = −k 23 (17e)
h1 − h0
=
k 36 k 13 ( ) + k 23 L − k 33 (20c)
2
h1 − h0
k 63 =M j =k 13 ( ) + k 23 L − k 33 (17f)
2 h1 − h0
k 46 =
−k 11 ( ) − k 21L + k 31 (20d)
2
Taking advantage of the symmetry
characteristic in the stiffness matrix, and from h1 − h0
equilibrium requirements, the other coefficients of k 56 =
−k 12 ( ) − k 22 L + k 32 (20e)
2
the 6*6 stiffness matrix can be given as follows
h1 − h0
−2EI 0 =
k 66 k 16 ( ) + k 62 L − k 36 (20f)
=k 14 (d 33d 22 − d 23 ) 2
(18a) 2
λ
where λ is given by eq. (15)

2EI 0 Similar results can be obtained for the stiffness

=k 24 (d 33d 12 − d 13d 23 ) (18b)
λ coefficients given by eqs. (18a)-(20f) by using
the same procedure presented before. Therefore,
−2EI 0 substituting for Mx, Qx, and Px in the strain energy
=k 34 (d 23d 12 − d 13d 22 ) (18c)
λ expression (eq. (1)) interms of the nodal force
vector at node (j) and following the same previous
2EI 0 procedure will yield the same expressions given in
=k 44 (d 33d 22 − d 232 ) (18d)
λ eqs. (18a)-(20f).
The obtained stiffness coefficients can be written
−2EI 0 in a matrix form as
=k 54 (d 33d 12 − d 13d 23 ) (18e)
λ
 k 11 k 12 k 13 k 14 k 15 k 16 
h1 − h0  k 22 k 23 k 24 k 25 k 26 
=
k 64 k 14 ( ) + k 24 L − k 34 (18f) 
2  k k 34 k 35 k 36 
[ K ] =  sym . 33  (21)
2EI 0  k 44 k 45 k 46 
=k 15 (d 33d 12 − d 13d 23 ) (19a)  k 55 k 56 
λ  
 k 66 
−2EI 0
=k 25 (d 33d 11 − d 132 ) (19b)
λ For a beam element having an orientation as
shown in Fig. 2, the stiffness coefficients can be
1216
 Musab Aied Qissab Al-Janabi Exact Stiffness Matrix for Nonprismatic
Beams with Parabolic Varying Depth

obtained by the same previous procedure. The h1 − h0

k 35 =
−k 12 ( ) − k 22 L + k 32 (25b)
stiffness matrix for this case can be written 2
interms of the obtained coefficients (eqs. (14a)-
(20f)) of the above stiffness matrix as follows h1 − h0
=
k 36 k 13 ( ) + k 23 L − k 33 (25c)
2

2EI 0
=k 44 (d 33d 22 − d 232 ) (25d)
 k 11 k 12 k 13 k 14 k 15 k 16  λ
 
 k 22 k 23 k 24 k 25 k 26 
 k 36 
2EI 0
(22) = (d 33d 12 − d 13d 23 )
k 33 k 34 k 35 k 45 (25e)
 K  =   λ
 sym . k 44 k 45 k 46 
 k 55 k 56 
  =k 46
2EI 0
(d 23d 12 − d 13d 22 ) (25f)
 k 66  λ

in which 2EI 0
=k 55 (d 33d 11 − d 132 ) (26a)
λ
2EI 0
=k 11 (d 33d 22 − d 232 ) (23a)
λ =k 56
2EI 0
(d 23d 11 − d 13d 12 ) (26b)
λ
2EI 0
=k 12 (d 33d 12 − d 13d 23 ) (23b)
λ =k 66
2EI 0
(d 22d 11 − d 12 2 ) (26c)
λ
h1 − h0
=
k 13 k 14 ( ) + k 24 L − k 34 (23c)
2 For elements having no axial force-bending
moment coupling such as when the centroidal axis
−2EI 0 of the element is straight (i.e. the element is
=k 14 (d 33d 22 − d 232 ) (23d)
λ symmetric about centroidal axis), the obtained
coefficients can be modified by substituting the
−2EI 0 following values for the[ D ]matrix coefficients
=k 15 (d 33d 12 − d 13d 23 ) (23e)
λ (eq. (12)) such that

−2EI 0
=k 16 (d 23d 12 − d 13d 22 ) (23f)
λ
d 11 d 12 d 13  (2I 0 a0 ) 0 0 

=k 22
2EI 0
(d 33d 11 − d 132 ) (24a)
[ D=] d12 d 22 d 23=
  0
 ( 2(a6 ) + 4EI 0 K v ) −(a5 ) 
λ d 13 d 23 d 33   0 −(a5 ) (2a3 ) 

h1 − h0
=
k 23 k 12 ( ) + k 22 L − k 32 (24b)
2 (27)
−2EI 0 FIXED-END LOAD VECTOR DUE TO
=k 24 (d 33d 12 − d 13d 23 ) (24c)
λ UNIFORM LOAD
−2EI 0
=k 25 (d 33d 11 − d 132 ) (24d) Consider a nonprismatic beam element with a
λ parabolic varying depth under a uniform load q as
shown in Fig. 3(a). By using the principle of
−2EI 0
=k 26 (d 23d 11 − d 13d 12 ) (24e) superposition and knowing that the sum of all
λ displacement components in each direction at the
fixed end must be zero, the flexibility matrix
h1 − h0
=
k 33 k 16 ( ) + k 62 L − k 36 (24f) equation corresponding to node (i) can be written
2 as
h1 − h0
k 34 =
−k 14 ( ) − k 24 L + k 34 (25a)
2
1217
 Number 10 Volume 19 October 2013 Journal of Engineering

 1 h0 − h0 
 ( β1 ) (γ ) (γ 1 − β )   f 11 f 12 f 13 
 c 8c 
[ ] f 21 f 22 f 23 
2 c
   PFi  F =
 h0 (γ ) 1 −1  1   
(α ) 1 −   Q Fi   f 31 f 32 f 33 
 8c 3/2
4c  (1 + cL2 ) 2   

c 
 M Fi   1 − h0 
 −h −1  1  1   ( β1 )
h0
(γ ) (γ 1 − β ) 
 0 (α ) 1 −  (β )   c 8c 2 c 
 2 c 4c  (1 + cL2 ) 2  c   
−1  
=  0 (γ )
h 1 1
 h0  (α )  1 − 
 4 (α1 )  8c c 3/2
4c  (1 + cL2 ) 2  
 
   −h 
q  sin 4 φ0  −1  1  1
= 3/2   0 (α ) 1 −  (β ) 

c  8 c   2 c 4c  (1 + cL2 ) 2  c 
 −1 
 (α ) 
 2 
(35)
(28)
and solving eq. (28) for the unknown fixed end
where
reactions PFi , Q Fi , and M Fi yields
1 1
α
= φ0 − sin(4φ0 ) (29)
8 32 q  h0α1 sin 4 φ0
=PFi  3/2 (f 33f 22 − f 23f 32 ) − (f 33f 12 − f 32 f 13 )
ψ  4c 8c 2
1 31 1 
=β cos3 φ0 sin φ0 +  φ0 + sin(2φ0 )  (30) α 
4 42 4  − 3/2 (f 23f 12 − f 22 f 13 ) 
2c 
  (36a)
1 2
γ=
1 + − 2 
(31)
 (1 + cL2 ) 2 (1 + cL ) 
q  −h0α1 sin 4 φ0
=Q Fi  3/2 (f 33f 21 − f 31f 23 ) + (f 33f 11 − f 31f 13 )
ψ  4c 8c 2
3 1 1 
α1 =
 φ0 − sin(2φ0 ) + sin(4φ0 )  (32) α 
8 4 32  + 3/2 (f 23f 11 − f 21f 13 ) 
2c 
 I 0 h0 2  h2 (36b)
β1 = +  φ0 + 0 ( β − 2γ 1 ) (33)
 A0 4  4
 q  h0α1 sin 4 φ0
=
M Fi  3/2 (f 32 f 21 − f 31f 22 ) − (f 32 f 11 − f 31f 12 )
ψ  4c 8c 2
1 1 
γ1
= φ0 + sin(2φ0 )  (34)
2  4  α 
− 3/2 (f 22 f 11 − f 21f 12 ) 
2c 
Writing the flexibility matrix [ F ] in eq. (28) in (36c)
the form
where

ψ= f 11 (f 22 f 33 − f 32 f 23 ) − f 12 (f 33f 21 − f 31f 23 )
+ f 13 (f 32 f 21 − f 31f 22 )
(37)

The right hand side of eq. (28) represents the

free-end displacement vector at node (i) due to
applied load q.

1218
 Musab Aied Qissab Al-Janabi Exact Stiffness Matrix for Nonprismatic
Beams with Parabolic Varying Depth

From equilibrium, the fixed-end reactions at 1 31 1 

node (j) can be given as =β cos3 φ1 sin φ1 +  φ1 + sin(2φ1 )  ,
4 42 4 
PFj = −PFi , Q Fj= qL − Q Fi , and
φ1 = tan −1 ( c L1 )
 h −h  qL 2
(41)
=
M Fj PFi  1 0  + Q Fi − M Fi −
 2  2
(38) and all other constants are previously defined.

FIXED-END LOAD VECTOR DUE TO Solving eq. (39) yields the following expressions
for the unknowns fixed-end reactions at node (i)
CONCENTRATED LOAD
interms of the flexibility matrix coefficients
For a beam element loaded by a concentrated load
P
(P) at an arbitrary location defined by a distance =
PFi (δ1 (f 33f 22 − f 23f 32 ) − δ 2 (f 33f 12 − f 32 f 13 )
(L1) from the left support as shown in Fig. 3(b), ψ
the fixed-end load vector can be derived by using +δ 3 (f 23f 12 − f 22 f 13 ) )
the same procedure given before. The flexibility (42a)
matrix equation for this case can be given P
as Q Fi =( −δ1 (f 33f 21 − f 31f 23 ) + δ 2 (f 33f 11 − f 31f 13 )
ψ
 1 
 ( β1 )
h0
(γ )
− h0
(γ 1 − β )  −δ 3 (f 23f 11 − f 21f 13 ) )
 c 8c 2 c  (42b)
  P 
 h0 (γ ) 1 −1  1    Fi  P
 8c (α ) 1 −   Q Fi  =
M (δ1 (f 32 f 21 − f 31f 22 ) − δ 2 (f 32 f 11 − f 31f 12 )
4c  (1 + cL2 ) 2    ψ
Fi
c 3/2 
  M Fi 
 −h −1  1  1  + δ 3 (f 22 f 11 − f 21f 12 ) )
 0 (α ) 1 −  (β ) 
 2 c 
4c  (1 + cL2 ) 2  c (42c)
 δ1  And from equilibrium, the fixed-end reactions
 
= P δ 2  at node (j) can be given as
δ 
 3 PFj = −PFi , Q Fj= P − Q Fi ,and
(39)
 h −h 
=
M Fj PFi  1 0  + Q Fi − M Fi − P (L − L1 )
where  2 
(43)
− h0   1 1   1
=δ1  2 − +
8c   (1 + cL 2 ) (1 + cL 2 )   (1 + cL 2 ) 2 NUMERICAL EXAMPLES
  1 
 
−
1

(1 + cL12 ) 2 
+ 4 L1 c (
(φ1 + β ) − (φ0 + β ) 
 ) To verify the correctness of the derived matrices,
the following examples are considered.

(40a)
Example 1
1  1 1 
=δ2   (φ0 − φ1 ) − (sin(4φ0 ) − sin(4φ1 )  Consider the beam shown in Fig. 4 which has a
 2 
3/2
4c 8 single span of length, L=1units and fixed at both
 1 1  ends. The beam is carrying a uniformly distributed
+ L1 c  −
 (1 + cL2 ) 2 (1 + cL 2 ) 2   load, q=1. The depth of the beam is h0 =1units at

 1  the left end and increase parabolically to h1 =2
(40b) units at the right end and has a unit width, b=1
units. The beam was analyzed by Khan and Al-
−1   1 1  Gahtani (1995) by using the boundary integral
=δ3  4L1 c ( β − β ) +  −  
4c   (1 + cL 2 ) 2 (1 + cL2 ) 2  method (BIM). Using the same dimensionless
 1 
data adopted by Al-Gahtani and Khan, the beam is
(40c) reanalyzed by using the derived expressions for
1219
 Number 10 Volume 19 October 2013 Journal of Engineering

the fixed-end load vector. The results are including bending, shear, and axial strain energies
presented in Table 1. which is used to obtain the exact expressions for
the coefficients of the stiffness matrix. The
Example 2 correctness of the derived expressions is examined
through numerical examples. It is found that the
Consider a three-span continuous bridge girder derived stiffness matrices and the equivalent load
having a parabolic varying depth as shown in Fig. vector are efficient for the analysis of structures
5. The depth of the girder varies from h=2.5 units having members with parabolic varying depth.
at both ends and midspan to h=7.5 units at interior Furthermore, the derived matrices can be used in
supports. This problem has been analyzed by the structural analysis softwars as compared to the
Timoshenko and Young 1965 ; Al-Gahtani and available analytical solutions. The obtained results
Khan (1998).Using the same dimensionless data, show a significant effect for axial force-bending
the problem is reanalyzed by using the derived moment coupling in continuous beams with axial
stiffness matrices and the fixed- end load vector. restraint.
The analysis results (at nodes 1,2,3,4,and 5) are
presented in Table 2 together with those obtained REFERENCES
by Timoshenko and Young (1965) ; Al-Gahtani
and Khan (1998). It can be seen that when the Al-Gahtani, H. J. (1996). "Exact stiffnesses for
girder is restraint against horizontal (axial) tapered members" J. Struct. Eng., 122(10), 1234-
displacement at supports (i.e. all supports are 1239.
hinges), the results diverge significantly from that
obtained by other methods as given in the last Al-Gahtani, H. J., and Khan, M. S. (1998). "Exact
column of Table 2. This is due to the coupling analysis of nonprismatic beams" J. Eng. Mech.,
effect between the axial force generated from 124(11), 1290-1293.
axial restraint and the bending moment which
reduces the displacement, rotations, and bending Bathe, K. J., (1996). "Finite element procedures"
moments. Prentice-Hall, New Jersey.

Boresi, A. P., and Schmidt, R. J. (2003).

Example 3
"Advanced mechanics of materials" 6th Ed.,
Finally, consider the beam shown in Fig. 6 which Wiley, New York.
has a span of unit length and a depth varies
Khan, M. S., and Al-Gahtani, H. J. (1995).
parabolically from h=1.0 units at left end to h=2.0
"Analysis of continuous non-prismatic beams
units at the right end. The beam is supported at
using boundary procedures" The Fourth Saudi
the left end on a translational spring with a
Engineering Conference, V11, 137-145.
stiffness constant of K=10 , fixed at the right end
and carrying a concentrated load P=1.0 at the left Luo, Y., Xu, X., and Wu, F. (2007). "Accurate
end. The beam is analyzed by using the derived stiffness matrix for nonprismatic members" J.
stiffness matrix interms of the given Struct. Eng., 133(8), 1168-1175.
dimensionless data. The analysis results are
presented in Table 3 in which the third column Timoshenko, S. P., and Young, D. H. (1965).
show the results when the shear deformation is "Theory of structures" 2nd Ed., McGraw-Hill, New
considered. The results show a significant effect York.
for shear deformation. This is due to the large
translational stiffness relative to the rotational hx = depth of beam at any section x;
stiffness for this beam.
I0, Ix = moment of inertia of beam cross-section;
SUMMARY AND CONCLUSIONS
Kv = coefficient defined in eq. (6);
In this paper, an exact stiffness matrix and fixed-
end load vector for beams with parabolic varying [K],[ K ] = stiffness matrices;
depth are derived. An exact integrations were
carried out to obtain the strain energy equation L = length of beam;

1220
Musab Aied Qissab Al-Janabi Exact Stiffness Matrix for Nonprismatic
Beams with Parabolic Varying Depth

M, P, Q = bending moment, axial force, and shear θi ,θ j = rotational angles at nodes i, j;

force;
λ = variable defined in eq. (15);
U = total strain energy;
φ0 , φ1 = variables defined by eq. (6) and eq. (41)
u, v = displacements in X and Y directions;
respectively; and
α , β , γ , α1 , β1 , γ 1 = variables defined in eqs. (29)-
ψ = variable defined in eq. (37).
(34);

δ1 , δ 2 , δ 3 = variables defined in eqs. (40a)-(40c);

NOTATION
The following symbols are used in this paper:

A0 = minimum cross-sectional area of beam element;

b = width of beam cross-section;

c = depth variation variable;

[D] = matrix defined by eq. (12)

E = Young's modulus;

[F] = flexibility matrix defined by eq. (35);

G = shear modulus;

h0, h1 = minimum ,and maximum depth of beam respectively;

1221
 Number 10 Volume 19 October 2013 Journal of Engineering

Table 1. Fixed-End Actions

Variable BIMa Exactb

PF1 0.0000 0.0088
QF1 0.4267 0.4232
MF1 0.0568 0.0564
PF2 0.0000 0.0088
QF2 0.5733 0.5768
MF2 0.1301 0.1287
a
Khan and Al-Gahtani (1995)
b
Present Analysis

Table 2. Supports Reactions, Midspan Deflection, and Angles of rotation

Exactc Exactc
a b
Variable BIM Slop-Deflection (Free horizontal (Horizontal disp.
disp.) is restraint)
P1 0.000 0.000 0.000 -6.489
P2 0.000 0.000 0.000 98.978
P4 0.000 0.000 0.000 -128.116
P5 0.000 0.000 0.000 35.667
Q1 1.510 1.500 1.337 7.090
Q2 72.450 72.850 72.644 66.165
Q4 46.620 46.150 46.768 40.187
Q5 -12.580 -12.500 -12.732 -5.433
M2 -593.750 -594.000 -598.393 -408.562
M3 124.810 138.600 125.941 72.680
M4 -452.810 -453.000 -458.562 -283.633
v3 ˗ ˗ -30086.330 -16224.470
θ1 290.250 376.560 387.600 -59.462
θ2 -423.920 -551.880 -560.712 -212.144
θ4 615.500 800.640 809.900 433.410
θ5 -775.540 -1006.200 -1018.890 -502.100
a
Al-Gahtani and Khan (1998)
b
Timoshenko and Young (1965)
c
Present Analysis
v 3 =EI0 v 3 , θ1 =EI0 θ1 , θ 2 =EI0 θ 2 , θ 4 =EI0 θ 4 , θ5 =EI0 θ5

1222
Musab Aied Qissab Al-Janabi Exact Stiffness Matrix for Nonprismatic
Beams with Parabolic Varying Depth

Table 3. Displacements, rotation, and Support reactions

Exact Exact
Variable (no shear (with shear
deformation) deformation)
u1 -0.01577 -0.00776
v1 -0.04950 -0.07515
θ1 0.09460 0.04658
Q1 0.4950 0.7515
Q2 0.5050 0.2485
M2 -0.5050 -0.2485

hx

h0 X
x dx h1

(a)

Qi,vi
Qj,vj
Pi,ui i n.a
j
Mi,θi Pj,uj
Mj,θj
(b)
Fig. 1 . A beam element with parabolic varying depth:
(a) typical element; (b) degrees of freedom and nodal forces

1223
Number 10 Volume 19 October 2013 Journal of Engineering

Qj,vj
Qi,vi
n.a j Pj,uj
Pi,ui i Mj,θj
Mi,θi

Fig. 2. A beam element with parabolic varying depth

h0 i

j h1

(a)

P
L1

h0 i

j h1

(b)

Fig. 3. Beam with parabolic varying depth fixed at both ends under the action of:
(a) uniformly distributed load; (b) concentrated load

1224
Musab Aied Qissab Al-Janabi Exact Stiffness Matrix for Nonprismatic
Beams with Parabolic Varying Depth

q=1.0

h0=1.0
h1=2.0

L=1.0

Fig. 4. A Beam with parabolic varying depth (example 1)

q=1.0

2.5 1 2.5 2.5

7.5 3 7.5 5
2 4
36.0 72.0 36.0

Fig. 5. A three-span continuous bridge girder (example 2)

P=1.0

1.0 1

Ks=10 2 2.0

L=1.0

Fig. 6. A Beam with parabolic varying depth elastically supported at one end
and fixed at the other end (example 3)

1225