Buckling Load of Non-Uniform Columns: I RT I I
Buckling Load of Non-Uniform Columns: I RT I I
Buckling Load of Non-Uniform Columns: I RT I I
A 20 m column consists of three steel tubular sections welded together. The middle part
is 17 m long, has a diameter of 1036 mm and a wall thickness of 50 mm. The end parts
are 1.5 m long, have diameters of 1016 mm and wall thicknesses of 40 mm. Young’s
modulus of steel is 2.1⋅105 N/mm²
I = π R3t
(2 )
3
I m = π 1 1036 50 = 2183 ⋅107 mm 4
Since an exact formula for the buckling load does not exist, it is approximated using
potential energy.
πx
w = w sin .
l
⎛ 2π x ⎞
w = w 1 ⎜ 1 − cos ⎟,
2⎝ l ⎠
where w is the maximum deflection in the middle of the column.
The curvature κ is
d 2w
κ=− .
dx 2
M = EI κ .
l
dy 2
u = l − ∫ 1− ( ) dx .
dx
0
The latter formula can be derived by considering an infinitesimal part of the deformed
column. However, it cannot be evaluated easily. Using the Taylor series
1 − f = 1 − 1 f + O( f 2 ) we can approximate u accurately.
2
l
dy 2
u≈ 1
2 ∫ ( dx ) dx .
0
E pot = (13937 − 0.0001234 P) w2 .
E pot = (52006 − 0.0001234 P ) w2 .
The column is stable when the potential energy curve ( E pot versus w ) is bowl shaped. It
becomes unstable when P is such that the bottom of the “bowl” is flat. Therefore, it is
unstable when
d 2 E pot
= 0.
dw2
The pinned end result has been checked with a structural analysis program (geometrical
nonlinear analysis with Dr. Frame including a small lateral load). Pcr = 112000 kN,
which is almost the same.
The method gives good results when the estimated buckling shape agrees well with the
real buckling shape. If not, the method will overestimate the real buckling load.
This method works also for a cross-section that changes gradually over the length
(tapered sections).