Flexural - Torsional Buckling Analysis of Thin Walled Columns Using The Fourier Series Method
Flexural - Torsional Buckling Analysis of Thin Walled Columns Using The Fourier Series Method
d 2v d 2w dv dw
( Mly ez N x ) ( Mlz e y N x ) Vz Vy
dx2x dx2x dx dx
H wVw d
C zVy C yVz t( x) (2.3)
Cw dx
where Mlz , Mly moments caused by transverse loads only.
z( y z dA
2
Hy z2 )dA I yy 2
A A
y( y y dA
2
Hz z2 )dA I yy 2
A A
2(w0 w )( y yz dA
2 2
Hw z )dA I yz
A A
I zz H y I yz H z
Cy 2
2ez
I yy I zz I yz
I yy H z I yz H y
Cz 2
2e y
I yy I zz I yz
J = St Venant torsional stiffness of the section
E = Youngs modulus of elasticity
G = shear modulus
Iyy, Izz = moments of inertia
Cw = warping constant
Iyz = product of inertia
ey, ez = coordinates of the shear center
Vy, Vz = shear forces
qy, qz = transverse loads
Nx = axial load
v(x),w(x) are transverse displacements
x = twist rotational displacement
x = longitudinal axial coordinate, yz is the plane of the cross section
A = area of the cross section of the column.
The governing equilibrium equations for the generalized column buckling problem is represented by the system of three
simultaneous differential equations in the three unknown displacements v(x), w(x) and (x). For columns with prismatic
cross-sections, the elasticity properties (GJ and E) as well as the inertial and geometrical properties are constant; but the load
coefficients (Mly, Mlz, Vy, Vz, Ww, Vw) are variables depending on the axial longitudinal coordinate (x). The system of
governing differential equations thus have variable coefficients, rendering them difficult to solve mathematically.
However, simplifications of the system of governing differential equations can be obtained by using special characteristics of
different types of column problems. A simplification of the governing equilibrium equations can be obtained if the yz
coordinates are principal coordinates. Then Iyz = 0, and the governing differential equations become [16][13]
d 4v d 2v d 2 d
EI zz 4
Nx 2
( Mly N x ez ) 2
Vzy qy (2.4)
dx dx dx dx
d 4w d 2w d 2 d
EI yy 4
Nx 2
( Mlz N x e y ) 2
Vy qz (2.5)
dx dx dx dx
d 4w I N H W d 2 d 2v
ECw 4
GJ E x C z Mlz C y Mly w w 2 ( Mly ez N x ) 2
dx A Cw dx dx
d 2w dv dw H V d
( Mlz e y N x ) 2
Vz Vy C zVy C yVz w w t( x) (2.6)
dx dx dx Cw dx
If the member is free of transverse loads qy, qz become zero; and the transverse moments are constant, then the shear forces
Vy, Vz are also zero. For member (columns) subjected only to axial compressive load Nx acting through the centroid of the
section, the moments due to the transverse loads vanish, the applied torque vanishes; and if the load is applied such that the
bimoment vanishes, the system of differential equations become [15][16]
d 4v d 2v d 2
EI zz Nx N x ez 0 (2.7)
dx 4 dx2 dx2
d 4w d 2w d 2
EI yy Nx N x ey 0 (2.8)
dx 4 dx2 dx2
d 4 I E N x d 2 d 2v d 2w
EC w GJ N x e z N x e y 0 (2.9)
dx 4 A dx2 dx2 dx2
The system of differential Equations (2.7), (2.8) and (2.9) represent the governing equilibrium equations for a particular case
of the generalized elastic column buckling problem presented by Equations (2.1), (2.2) and (2.3) when I yz 0,
qy qz 0, Vy z 0, Mly Mlz 0, Ww 0 .
III. APPLICATION OF THE FOURIER SERIES Suitable Fourier series representations of the three
METHOD unknown displacement functions that satisfy apriori the
We seek to apply the Fourier series method to solve the above boundary conditions are given by
system of differential Equations (2.7), (2.8) and (2.9)
mx
representing the governing-equilibrium equations for the v( x ) = vm sin l
elastic column buckling problem for the case when the m 1
ends at x = 0, x = lare on pinned supports.
mx
For pinned ends at x = 0, x = l, the relevant boundary w( x ) = wm sin l
(3.2)
conditions are m 1
v ( x 0) 0 v( x 0) 0
mx
w ( x 0) 0 w( x 0) 0
( x ) = m sin l
m 1
( x 0 ) 0 ( x 0) 0 (3.1) where vm, wm and m are infinite number of unknown
v( x l ) 0 v( x l ) 0 coefficients of the Fourier sine series representations of
w( x l ) w( x l ) 0 the unknown displacement functions v(x), w(x) and (x)
that we seek to determine. If Equation (3.2) represent
( x l ) ( x l ) 0 solutions of the governing Equations (2.7), (2.8) and
(2.9), then
d4 mx d2 mx d2 mx
EI zz
dx 4 m 1
v m sin
l
+ N x
dx2 m 1
v m sin
l
+ N x e z
dx2 m 1
m sin 0
l
(3.3)
d4 mx d2 mx d2 mx
EI yy
dx 4 m 1
wm sin
l
+ N x
dx2 m 1
wmsin
l
N e
x y
dx2 m 1
m sin 0
l
(3.4)
d4 mx IE N x d2 mx
ECw
dx 4 m 1
m sin
l GJ
A dx2 m 1
m sin
l
d2 mx d2 mx
+ N x ez 2
dx m 1
vm sin N x ey
l
dx2 m 1
wm sin
l
(3.5)
A m
2
P ECw GJ
IE l
2
m P I E
ECw GJ
l A
Hence we have
( Pzz N x ) 0 N x ez
0 ( Pyy N x ) N xey 0 (3.14)
P I E I E N x
N x ez N x ey
A A
By expansion of the determinant, we have
( Pyy N x ) N x ey 0 ( Pyy N x )
( Pzz N x ) IE N x ez 0 (3.15)
N x ey ( P N x ) N x ez N x ey
A
I
(Pzz N x ) (Pyy N x )( P N x ) E ( N x e y )2 N x ez (0 N x ez (Pyy N x )) 0 (3.16)
A
I
( Pzz N x )( Pyy N x )( Px N x ) E ( N x e y )2 ( Pzz N x ) ( N x ez )2 ( Pyy N x ) 0 (3.17)
A
A A
(Pzz N x )(Pyy N x )(P N x ) N x2e2y (Pzz N x ) N x2ez2 (Pyy N x ) 0 (3.18)
IE IE
But
I xx I yy
r02 e2y e2z
A
I E I xx I yy (e2x ez2 ) A
IE I xx I yy
e2y ez2 r02
A A
Hence,
e2y e2z
( Pzz N x )( Pyy N x )( P N x ) N x2 ( Pzz N x ) N x2 ( Pyy N x ) 0 (3.19)
r02 r02
e2y N x2ez2
( Pzz N x )( Pyy N x )( P N x ) N x2 2
( Pzz N x ) ( Pyy N x ) 0 (3.20)
r0 r02
This is the characteristic equation for determining the
buckling load of a column with an asymmetric cross-
section. The buckling equation is a third order polynomial
e2z
in Nx. Thus, it can be solved using the methods for ( Pzz N x )( Pyy N x )( P N x ) N x2 ( Pyy N x ) 0
solving polynomials to obtain the three roots Pcr , Pcr
r02
1 2 (3.21)
and Pcr . The smallest of the three critical buckling loads or,
3
will govern the buckling behaviour of the column. N 2e2
However, if the zz axis is the axis of symmetry of the ( Pyy N x ) ( Pzz N x )( P N x ) x2 z 0
r0
cross-section, ey = 0, and the characteristic buckling
equation simplifies to become (3.22)