Stability of battened

columns with and without

taper
J. C. Ermopoulos, S. S. loannidis and A. N. Kounadis

Department of Civil Engineering, National Technical University, Athens, Greece

(Received March 1990)

In this investigation built-up columns with a linear variation of depth

under various support conditions are analysed as rigid-jointed
frameworks. A rigorous analysis for determining critical loads allows
formulae to be established for buckling load estimates of practical
importance. The individual and coupling effects on the critical loads
are assessed for a variety of parameters such as: degree of nonuni-
fortuity, number of panels as well as stiffness and length ratios of
component members. The proposed solution technique is demon-
strated with examples.

Keywords: stability, battened columns, tapered build-up members,

support conditions, critical buckling load

Frameworks with built-up members of varying cross

section are frequently encountered in engineering prac-
tice. The earliest study on this problem was by
Engesser ~ who established approximate formulae for
the buckling load of battened and laced columns.
Bleich" and Timoshenko and Gere ~ presented more
accurate and understandable stability analyses fi)r
nonuniform battened and laced columns. Nevertheless
such built-up columns were not considered as frame-
works of a high degree of indeterminacy. Indeed,
1 I
various simplifying assumptions regarding the buckled I I
configuration, the slope continuity of the chords, and the l
I I
I
degree of end fixity of the lacing bars were made which z 1 i
permitted approximate formulae for establishing the in- II
ternal forces of the framework. More accurate studies
dealing with the stability analysis 4'~ and the post-
,I]
buckling response 6 of simple frames having latticed
members with linear depth variation have been
presented.
In this investigation tapered built-up members are
analysed as rigid-jointed frameworks with particular Figure 1 Members with batten plates of varying cross-section
emphasis on battened columns. The main objectives of
this paper are to use a linear stability analysis to
establish critical loads of tapered battened columns; to S t a t e m e n t o f the p r o b l e m
discuss the effect of various parameters on the critical
loads; and to avoid the aforementioned intractable The axial load-carrying capacity of tapered members
analysis by proposing approximate but reliable formulae having a linearly varying depth with a constant cross-
for predicting the load-carrying capacity of such sectional area along their axis is studied. Such members
members. with batten plates are shown in Figure 1. In this case the
Post-buckling effects on the critical load are not moment of inertia of the cross-section varies as a second
accounted for, however, the inclusion of such effects power of the distance from the point of intersection of
could be justified in view of some residual post-buckling the two chord axes, as follows (Figure 2)
strength in the members under consideration. In this
case one can employ a reliable simplified post-buckling l(x) = Io (!)
analysis 7.8
0141-0296/91/030237-05
,c! 1991 B u u c r w o r l h - H ¢ i n e m a n n Ltd

Eng. Struct. 1991, Vol. 13, July 237

 Stability of b a t t e n e d columns with and w i t h o u t taper: J. C. Ermopoulos et al.

2P
Mathematical analysis
Consider the built up column shown in Figure 2a which
is composed of N - 1 equal panels of length 10cos ~,.
The column, being symmetric with respect to its
I[,f~.l, I M,-I',t
longitudinal axis. is subjected to an axial compressive
load 2P. The detailed geometry and the sign convention
, -- 7-----:: , (-4. of the internal forces are shown in Figure 2. The slope
deflection method which includes the effect of axial
~-~ hm -r M"' °I ' Mr'/. t '
forces is used below.
I ~ ---~ t';'"
':.~---
I V,. C,,'
The bending moments and the shearing force cor-
22J23 ,,I_ ___..I,
i t responding to the (i - 1. i) chord segment with length
/, have the following expressions 9
Vd,.r
.i i
.-IN-,, k----3 v,..,
' ~ 1 I
M,-i, - Io oL,,O,_,+ (x:O,
ii,
' L I

I 2P lI
CL l 6,_ I + 6, ]
+ (~,, + o,j) - - ) ~ - ........]
, I

2El, [
',I/ M,.i_, = io a, Oi + o40, _, (3)
1- 'i'o
a b 6,_ i -- ~, ]
+ (or,, + c9) l0 ]
Figure 2 G e o m e t r y and sign c o n v e n t i o n of a partially fixed-
hinged n o - n u n i f o r m c o l u m n
V, M,_ I., + M,.i- I P 6,_ t - 8,
/. cos ~o 10

where at, x are distances from the origin as shown in where

Figure 2a. a n d / . is the moment of inertia at the bottom
of the member (x = o0. Such a law corresponds with
sufficient accuracy to the case of a tapered member con- , , , (x/= 2(e,~- "
¢i)
sisting of fi)ur angles or two channels (or double tees)
connected by lacing bars or batten plates. The joints for
. . . . . . . . . .
the latter case are considered to be rigidly connected.
Hence the battened member is a highly redundant I - B,.cot B,. sin ~,.
(4)
framework. IJ;
The analysis that follows refers to members under an
axial compressive load P with the lower end hinged and i
the upper end partially fixed, as shown in Figure 2. It
is also assumed that the batten plates are subjected only + : -nL-'v
to bending since their axial load is negligible before ",,1 \2 + U
buckling. The axial deformation of the members is not
taken into account. The geometrical description, internal
forces and sign convention are also shown in Figure 2. and /3~ =/3-'/cos~o being constant along the length of
The deformed state will be established by using a linear each chord. Equilibrium of moments at each joint i' and
stability analysis in terms of 0, and 6,. where 0, and 6, i yield 2N equations, while the equilibrium of shearing
are the angle of rotation and the lateral deflection respec- forces as the free-body diagram of Figure 2b shows,
tively of the ith joint (i = 0. I . . . . . N - 1). yields N - 2 equations.
The following auxiliary quantities are also used These equations can be written in the matrix form as
follows

L, h,,, 1 ~, PI:
#= .n= - [= . = - (2)
I, I ot El, KT = 0 (5)

where i,, and /h are the moments of inertia of each where 6 = 811o, while the matrices A (of order
chord and each batten plate respectively; h,. the depth 2 N x 2 N ) , B(of order ( N - 2 ) x ( N - 2 ) ) and K (of
of the cross-section at the middle of the member: I is the order 2N x (N - 2)) are functions of the quantities a,,,
length of the member, and E is the modulus of elasticity. a / (equations (4)), #. n. [ (equations (2)), and K r is the

238 Eng. Struct. 1991, Vol. 13, July

 Stability of battened columns with and without taper: J. C. Ermopoulos et al.

transpose of matrix K. The vanishing of the (stability)

determinant associated with equation (5) leads to the
overall buckling equation of the member from which the
successive buckling loads are numerically evaluated.
Using this equation one can treat the case of a simply
.°°t
supported column by setting c = 0, as well as the case .=o,
of the propped cantilever column (fixed-hinged column)
by setting c - oo (implying 00 = 00' = 0), where c =
rotational spring constant (Figure 2a).

Numerical results and discussion 200] '

With the aid of numerical evaluation of the buckling
equation one can obtain critical (buckling) loads cor- n=03
responding to various geometrical configurations of
simply supported and fixed-hinged columns. A variety
of numerical results are given in graphical form.
Figures 3a and b, correspond to a simply supported n=05
beam and a proposed beam respectively, they show the IOC
variation of the dimensionless critical load/~, versus #
for various values of n ( = 0 . 1 , 0.3, 0.5), [(=0, 10) and
f o r N = 10. el0 :35
, I
5.0
__ 1
75
I
iO
The critical loads corresponding to nonuniform col- 7
unms ([ ~ 0) are higher than those of the uniform col- Figure 4 Critical loads ~ , versus I: for # = 1. N = 10 and
umns ([ = 0), for the segments of the curves/3~., versus various values of n. / ' = 0 corresponds to a uniform column.
which are located below the dotted lines in both plots. (-- ), fixed-hinged; ( - - - ) , hinged-hinged
For the usual case of a simply supported column with
n = 0. I and/,t = 0.5, the critical load of the uniform col-
umn is greater by about 10% than the critical load of the
corresponding nonuniform column. Note that such a dif- a ~ 7
fcrence decreases substantially with an increase in n.

40o a --7=o
•

F ,t0~,~"n =OI 400 ."/#7 .z .:/ I j 'l

300
...: /i '
2OO 200 iJ" "
/" t g-22~_°.-o~ ,oo
,0o

o"
oo,
-,
. . ~ ..... P . . . . . . ~ ~

'
- ~ - -

,
'
,
Ib
I# ....

........~................... .
,oo~b _~. o.o, ~"i " o

/ IF
r- ~ ' °.,o /- ..~i'" l / ":
=. ' ~ ~=,o

"" 200"" / "" "..0 n=03 " "'

. ,.o
°o o, o 5 ~ oI~ ,

Figure 3 Critical loads/~, for a simply supported (a) and a fixed- Figure 5 Critical loads /~'~, for simply supported (a) and fixed-
hinged (b) column, as f u n c t i o n s of p,, n and f hinged (b) columns, as functions of/~, /'and n

Eng. Struct. 1991, Vol. 13, July 239

 Stability of battened columns with and w i t h o u t taper: J. C. Ermopoulos et al.

/- n--QI
0 5 10 4 /"rT n

'~V" ~:-~ --::- ...... L

: ; t , / /Z/_/ , 5j l_

\
0t
-5O0 [ ~ I0
1500

r/ 'ooor' x~!..... 0:o~.oo~

5000
k -_. ~.(,~_ 50

- , ~ O O r - 0 I=-:-,--"--~---,'----:'.'-,-
0 5 IO
7
a -I00~- /I/ n ==0.1
/// --- 0.3

0.~,,.j~._~,~ . _ r _ . _ . . . _ ~. . . . . . . . . . . . . .-,

.,oo~..... ~ ~ ~o~ ~ - ~ ~ = =
___2l ,'. . . . . z50o- z o ~ ~
'~' '~ I0
-soup I I / , ' ,, ' _ zooo-\ ~ I1 -/ / , 5.
///:? ;:o:~°' \ ..o, o ,o

I///,..o, oo,\ :..o, ,

"7/ _,ooo
,°.°
,poe-\\
\\,..:.,:i~
~~ - 50
,~-\/.? //.'
/.\x(7 I/?

.,~oo ~°°--7+-
Oo,,,
~ - ';,~-...... :,~ .,ooI
Figure 6 C o e f f i c i e n t s ao, bo a n d co as f u n c t i o n s o f I~ n a n d N c o r r e s p o n d i n g ; a. to s i m p l y s u p p o r t e d columns; a n d b, to f i x e d -
hinge columns. ( ..... ), N = 5; ( - - - ) , N = 10; ( ), N = 15

For the case of a fixed-hinged column, the maximum noted that this figure has been taken from Figure 5 in
difference between the critical load of the nonuniform order to show the difference of the critical loads between
column and that of the uniform one, occurs for n = O. ! the cases of a pinned-pinned (Figure 5a) and a pinned-
and # = O. 1 and reaches 20%. Contrary to the previous fixed (Figure 5b) column.
case, the critical load of the nonuniform column is A three-dimensional graphical repremntation of/3~,
greater than that of the uniform column. with respect to # a n d / , corresponding to a simply sup-
It is also worth observing that the difference in the ported column and a fixed-hinged column is given for
critical loads corresponding to a simply supported various values o f n(=O. I, 0.3, 0.5), in Figures 5a and
column and a fixed-hinged column for the same 5b respectively. In these plots the projection o f the
geometrical configurations is much less than the cur- curves (corresponding to n = O. 1, 0.3, 0.5) o f the plane
responding difference for colunms with solid cross- [ = O, on the plane / = 10 gives the intersection points F,
sections. Indeed for solid cross-sections obtained for C, D and E, as shown in Figure 3. Note that such a pro-
# -- up and n - 0 (in the case [ = O) such a difference jection has not been included in Figures 5a and 5b, in
tends to I/0.7" = 2, while for battened members it is order to avoid any confusion. From these figures it is
much less as is shown in Figure 4. From this figure, one evident that the effect of nonuniformity on the critical
can see the variation of/3~., versus [ for N = 10, /x = 1 load is more pronounced for the fixed-hinged column
and various values o f n ( = O . l , 0.3, 0.5). It should be than the corresponding simply supported one.

240 Eng. Struct. 1991, Vol. 13, July

 Stability of battened columns with and without taper: J. C. Ermopoulos et al.

a the critical l o a d / ~ , versus N, /~, n and [ for a simply

400 i n=O.I supported column as well as for a fixed-hinged column,
as shown in Figures 7a and 7b respectively. In both
,0.=1 columns the increase in the number of panels implies an
3OO increase in the critical load. Such an increase is more
pronounced for slender columns (i.e. n = O. 1) than it is
200 /,o l -i= io for short ones (n = 0.5) regardless of the degree of
n=05 nonuniformity of the cross-section. This is because
100 n=O.I the latter columns behave like slender frames.
l.t-=0 I
n=05
0 Oi I I I I l" I I I I I i Conclusions
IO 15
N The most important conclusions of this investigation
based on two types of columns are as follows. First,
b there are ranges of variation of the stiffness and length
/ n=O.I
400 ratio parameters for which the nonuniformity of the
cross-section for both types of columns implies an in-
300 crease in the load-carying capacity. Such an increase of
the load-carrying capacity may reach 10% for the simply
-]=lO
supported column and 20% for the fixed-hinged column.
:~00 n=05 Second, it should also be noted that the difference in the
critical loads between these two types of columns is
I00 B=OI much less than the corresponding difference of critical
loads of columns with solid cross-sections. Third, for
O l I l I I , I t t i I l large values of the ratio of the moments of inertia/~ and
5 l0 15 for very small values of the length ratio n both types of
N
nonunifi~rm colunms behave as those of solid cross-
Figure 7 Critical loads /Jc2,, for simply s u p p o r t e d (a), and fixed- section. Finally, the cumbersome numerical evaluation
hinged (b) columns, as f u n c t i o n s of N, n and /~ for establishing the exact load-carrying capacity of both
types of colunms can be avoided by using a simple
approximate but very reliable fi~tmula.
A reliable fi~rmula fi~r establishing the critical load
B~r of both types of columns is given as a function of References
the parameters #, n, / a n d N ( = 5 , !0, 15), as follows
I Engcsscr, E. "Uber dic knickfestigkeit yon Rahmcnstiibcn Zcntralblat!
B~:~= ao#" + bo# + co (6) dcr Bauver-waltung, 1909, 29, 136
2 Bleich, F. "Buckling xlrenglh of racial .s'lructures" McGraw-Hill B~mk
Co.. Inc.. New York, N.Y.. 1952
where the dimensionless coefficients ao, b0 and co are 3 Timoshenko, S. P. and Gore, J.M. "'lheorv of elastic stabilio'"
taken from the nomograms presented in Figure 6a McGraw-Hill. New York, 1961. p. 66
(simply supported column) and Figure 6b (fixed-hinged 4 ErmoIxmlos, I.Ch. and Kounadis. A. A. "Buckling analysis of simple
column), for various values of [ and n. I'ramcs with varying built-up sections" Proc. Conf Mechani~w.
HSTAM, Athens. 1983
For instance, for a simply supported column cor- 5 Ermop~mlos. I.Ch. and Kounadis. A. N. "Stability of frames with
responding to /~ = 0 . 2 , [ = 4, n = 0.1 and N = 15, tapered built-up members" J. Slnwt. I='ngng.. ASCE 1985. I I I (STg).
using Figure (m we find a u = - 2 3 0 , b, = 800, co = 22. 1979 - 1992
Using equation (6) we obtained B~, = 172.8 which 6 Kounudis. A. N. and Ermolmmlos I.Ch. 'Post-buckling analysis of a
slightly differs from the value of the exact critical load simple frame with varying stiffness" Acta Meclmnica 1984. 54,
9 5 - IO5
B~,= 165.4. Similarly: for a propped column cor- 7 Kounadis. A. A. 'An efficient simplified approach for the non-linear
responding to # = 0.4 ! = 6, n = 0.3 and N = 10, using buckling analysis of frames" AL4A J 1984. 23, 1254- 1259
Figure 6b we find: a, = - 8 0 , b0 = 230, Cn = 24, and 8 Kounadis. A. N. "Efficiency and accuracy of linearizcd pqt}sl-buckling
therefore B~, = 103.2 while the exact value is analyses of frames based on elastica" Int. J. Solids and Structures.
1988, 24. 1097-1112
B~, = 106.6
9 Winter. G.. Hsu. P. T.. K{~. B. and l..~lh. M. H. 'Buckling ol'trusscs
Considering columns with given length (1) and and rigid frames'. Cornell University Engineering Experimental
moment of inertia (/,) one can study the dependence of Station Bulletin, No. 36. 1948

Eng. Struct. 1991, Vol. 13, July 241