0

$ 5
" O w

P,

eta

.r?

5 2 03

.;

VI

8 1s
u f f o
" 0 -

Z Z S
\5.

$2

W Q -

150

Beam Columns

Art. 3 . 3

Ch. 3

Beam Column with Distributed Lateral Loi

which, in view of (3.8) and (3.13), can be written as

Mmax

- Ql , PQ13
T -

1
48EI I - (PIP,,)

Fig. 3-4 Beam column with laterally

distributed load.

'L-.----A!

solving the governing clifl'esential 'equation. T o illustrate an altc

mcthod ol'nnalysis, wc shall now ubsc thc K;~ylcigh-Ritz
mcthod.
-"
-.. ..-..--.,.
/In
a Leam column, bending and axlaT compression usually 1
simultaneously. However, the bending deformations can be assume
independent of the axial deformations as long as deformations in
remain small. The analysis of a beam column by the energy method i
fore similar to the analysis of an axially loaded member. That is, the er
axial compression is omitted and only bending energy is considel
Article 2.2).
The strain energy that is stored in the member as it bends is

Simplifying the tcrm inside the bracket givcs

"

...I.

_
i
-

from which

*
-

"
4

*
*
%3

The 1eft:hand fact& in (3.16) is the maximum moment that would exist if no
axial force were present. Thus., letting

and the potential energy of the external loads is

the maximum moment in the beam column can be given as
Mm,x

= Mo

I - (O.ISP/P,,)
1 - (PlPC,)

(3.18)
Thus the total energy In thc system is

Equation (3.18) shows that the effect

-- - -of axial
---compression
..-- - .- .- on the..b.e -..
n d-~ n g
moment
is
very
similar
t
o
the
eiTEst
that
an
axial
load
has
on
the
cicfleclio~i.
----.- ---, - Like the deflection, the moment that exists in the absence of axiaicompressiol)
is amplified by the presence of an axial load. Tt is also interesting to notc the
similarity between the amplification factor for nloment and thecorrespond~ng
amplification factor for deflection.

i
I

0-1-I / = T o satisl) the boundary conditions, Ihc deflection JJ is assumed t o be

y

-- 6 sin nx
I

3.3 BEAM COLUMN WITH DISTRIBUTED

LATERAL LOAD

where 6 i b the midspan deflection. Substrtution of this expression in

(3.19) gives

Let us now consider the case of a simply supported member bent by a

uniformly distributed lateral load w and a set of axial forces P,as shown in
Fig. 3-4. As before, we assume that the material obeys Hooke's law, that
deformations remain small, and that the member is restrained against lateral
buckling. In Article 3.2 the investigation was carried out by setting up arid

i
i
I

nX

152

1.

CI?.3

Beam Colunrns

T o evaluate (3.21), we make use of tile following definite integrals:

t%.med

P. Since
shape for y was not exact, the deflection given by (3.26)
However, it K s T e e n shown by Timoshenko and
is o 3 ~ n -ximatyon.
b e r e (Ref. 1.2), who solved thc problcm rigorously, that the approximate
solution differs from the exact answer only slightly3 w
-The maximum moment in the member is

-7
-=--:,&+
~/jrJ

--

p,

y -

211~51
[I .I- v = E16'n4
.I
- - P6?n"
-4 1 ' n
41

c ~ j W
d

kia(r/w

For
-6 must vanish. That is,

7?
C . . .

M,,,,,

?II!/
7~

--l v"l L -k p&

(3.27)

'

I n ;iew of (3.25) and (3.26) this cxpression can be written as

1v12

Mn"x = 7

d(U -1- V ) - E16n4

$8
21'

from which

----

(3.22)

.y.
tlie system to be in equilibrium the derivative of U 4- Y with respcct to

2, EtPJ-yl

753

Equation (3.26) gives the maximum deflection of a simply supported beam

that is bent simultaneously by a distr~butedtransverse load 14, and axial forces

{:sin2yd.y-

Thus Eq. (3.21) becomes

Beam Column with Distributed Lateral Load

Art. 3.3

-' ~PIYI(

1
38481 1 -- (PIPc,)

P&n2 21

.
Simplifying the term inside the brackets, one obtains

If the numerator and denominator of (3.23).are multiplied by 5/3S4EI, onc

obtains

from which

whicli reduces to

'
y

The term outside the brackets is the maximum mornent that would exist if no
axial force were present. If one lets

or very nearly to
(3.25) can be written in the form
4

,,

The ieft-hand factor in this r e l a t i o ~is~ the detlcctiol~th;\t would csist if the
lateral load 1.i: were acting by itself. Thus we let

The maximum deflection of the beam column given by Eq. (3.26) and the
maximum moment given by Eq. (3.30) are thus both e&al-tXhe Froduct of
two terms, the maximum deflktion or moment that would exist if only lateral
lord were present a w m p l i f i c a t i o n f a ~ a c c o u n t for
s the effect of
the axial load. ~ L a 1st perhaps most significant in these relations is their
similarity to the corresponding expressions for deflectio~l and moment
obtained previously for a cancentratcd lateral load. It is at least partially due

and rewrite (3.24) as

n a
<

2:

g. g
2
-.
0"
J
-l

Elp

g.:
3

O"

S-I
2 V).

:2

2 z

'=, a

.:

0 E
3 Y

V)

%;
E

z'
3

V)

-.

% 5

zF
2

0
O

5. 3
62

z- g
0
. V)

750

Beam Columns

Ch. 3

Substitution of these results in Eq. (3.34) gives

Effect of Axial Load on Bending Stiffness

Art. 3.4

157

In a simr~armanner, the ~-otationat end B of the member can be obtained.

Setting x , - 1 in Eq. (3.37) gives
- - k / c s ~ k l )Mn
+ ~ (l klcotkl).

6,=$f(l

J/

from which
from which
-1- cos k l cos k x
sin k l

8" = M ~ ' (1 - kl csc kl)

(3.36!

or

Making use of the identity

o,, :=

where K,$ / , and

cos(a - P ) = sin a sin

P -+ cos a

cos /?

A4

$9

.-I-

+ kZEIl'
AM I( 1 - kl cot kl)

M
x!9n

4, are defined

(3.43)

4'

in (3.40)
/'

Solving Eqs. (3.39) and (3.43) for M A and M,, one obtains

and multiplying the numerator and denominator of the second term i n c:\cil
parenthesis of (3.36) by I. one obtains
k l cos k(1 -- x)]
sin kl
-

-,,-

ED(,
- klcos k x )
PI

sin kl

(3.37)

The end rotation at A is obtairied by setting s =

: 0. Thus
M

= z4(lX-l

8,

cot kl) -i- -M

*(I
PI

- kl csc I<()

Letting
(-3.38)

Jf,numerator and denominator i n (3.38) are lnultiplicd by I, and k T 1 is

substituted for P, one obtains

8,

where

M I
A(1
k212EI

=-

MI
kl cot kl! -I- &/(I

and

(3.44) and (3.45) can be put into the same form as (3.31). That is,
MA = 7 ( a n 6 , -I- ar6,)

- k l csc kl)

L-'

(3.47)

,'

Equations (3,47)'and (3.48) give the relation between end moments M,,
and MBand end rotations 8, and 8, for a member subject to both bending ant1
axial compression. Let us now
. . ..consider
. ...
Tor the, s a y ,fy-p~-~f~,men~be~~h
rela~~nbe_tw-een-the_<rid
... monlents and a rc!ativc .joint
displacement,_..
A.,_,.
._._...__.~___I_.____~___
The deformation, shoivri?n Fig. 3-6, is brought about by dtsplac~ngthe
ends of the member relative to one another a distance A and keeping the end
rotations 8, and 8 , at zero. As a consequence of this deformation, negative
nloments are illduced a t both ends of the member. Equating internal and

.
758

..

Beam Columns

759

Effect o f Axial Load on Bending Stiffness

Art. 3.4

Furthermore, the condition dy/dx -- 0 at x = l'leads to

( I -I- cos klj - k sin k l +

A -

or

1 --

Mr k(sin2 kl f cos2 kl) - k cos k l

-pL-

sin kl +

sin kl

0.53)

Making use of the identity sin2 a 4- cos2 a ==: 1 and multiplying the numerator and denominator of t.he first two terms inside the bracket of (3.53) by l,
one pbtains

A =:- --@(-klcscltl
1

PI

-- kl.cotkl -!- 2)

from which
external moments, a t a distance s from the origin, gives

A
I

"?.

-.;a 4,)

If numerator and denominator are now mul~ipliedby

A

lqhere k 2 =PIE(. The solutio11 to Eq. (3.49) is

"

M
-P

" '

a,

"1

(a. -i. a,)--

)i

=a

M cos kl -I- 1
4 ~ sin kl

!!![-sink\-(,
P
s ~ nk l

and

sin kl
.

-1- cos k/) -t- cos kx

Equation (3.56) is the slope-deflection

equation ......
adjusted
.- . .- -- to include the.
_._I______.~__.__,____._
In
Chnptc1.'4,
c.e
shall~~J+w,t.h&-$quati.on
-..
can
erect
. .. . ."of
. ....axial
... .. .-*. -colnpression.
. .. ..
be used ..to dcterrnine the c;ilic;il load-of a fi.arnewOrk;-'Sd
that thc cquation
... .
. -ciln bc used eIlicic,\tly, numerical valucs of a,, a n d a, calculated by Winter,
Hsu, Koo, a n d Loh (J!ef. 3.2) arc given in the Appendix.
. T o see what Eq. (3.0)
tells us about Ihe erect of axial co~~~pressiorl
on the
bending stirincss ol'a flexural member, let us consider the variation ofa,, with
k l plotted in Fig. 3-7. Thc quantity a, can be shown to be given by
.I.-

Substitution of 1,hese results in Eq. (3.50) gives

9,
9

M 42- $L
h2
K 4,, -i4,

'

one gets

Equation (3.55) gives the relation,between cnd moments and relative jointdisplacement when an axial. .load
Combining this expression with
. . is preserlt.
. ,. . . , ..
'C5-41
)'.
. - ."
.
, we obtain the complcte stiffness relation for a member that is subject
,
to both bending and axial compression. That is,

:.:7d letting y = A at x = 1 gives

A --

I-...

4,+- $,

or, in view of (3.46),

From the condition that. y =-. 0 at x = 0 one obtains

13.54)

;
;

-1
LA
(3.5:)
4-2s
- - 1 ]-l.,.t
.

"
/I

-1- cos k / ! -- k sin /i.s -1.. -

-1-

((3,:iZj

' 160

Beam Columns

Art. 3.5

Ch. 3

Failure of Beam Columns

161

greater than 4.49, a, is negative, which means that the moment and the
rotation are oppositely directed. Values of kl in excess of 4.49 correspond to
members that are elastically restrained by other members at the end to which
the rotation is applied. Another way of putting the same idea is to say that
Icl < 4.49 corresponds to cases where 8, is induced by the adjacent member
and kl > 4.49 to cases where 8, is resisted by the adjacent member.

3.5 FAILURE OF BEAM COLUMNS

Fig. 3-7 Variation of bentling stilfness with ratio of axirrl loatf lo c:ills:il

load.

Up to this point in our study of beam columns, we did not concern ourselves
with the suhiect of failure, and it was therefore possible to limit the analysis to
elastic behavior. Now, however, we are specifically interested in determining
the failure load, and, since failure involves yielding, it becomes necessary to
introduce the complexities of inelastic behavior into the investigation. When
studying the behavior of columns i n Chapter I , it was pointed out that problems which involve inelastic bending do not possess closed-form solutions.
They must either be solved numerically, which entails lengthy and tirneconsuming calculations, or approximate answers must be sought by making
simplifying assumptions, In this article we shall study the failure of beam
columns using the latter of these two approaches to the problem.
Let us consider the simply supported, symmetrically loaded member
shown in Fig. 3-8a. The member is simultaneously bent and compressed by

if 8, and A are set equal to zero in Eq. (3.56). Thus a, is proportional to the
moment M A that is needed to maintain a rotation 8, when 0, =: A :- 0. In
other words, a, is a measure of the bending stiffiless of the member. If (he
quantrly kl is rewritten in the form

where Pt is ihe Euler load, it beconies evident that kl is a measure of the ratio
of the axial load to the Euler load. T h k u r v e in Fig. 3-7 thus gives the vari3tion of the bending stiffness with the ratio of axial load to critical load.
When kl = 0, that is, when there is no axial load, a; = 4. This value of
a, is used in routine structllral analysis where the eA':ct of axial conzpression
on the bending stiffness is neslected. Betweel? Icl = 0 and kl -- 4.49, a,,
decreases as kl incrcascs. The hending s[ili'ncss is tllus rccluccd by a n incrcnsc
in the magnitude of thc axial load. A t I;/ :. 4.41! or P =-: 2.04Pc, a, = 0 . ' 1 1 1 ~
reason the bending stiffness vanisllcs ;IL this loadis that the n~emherwhich we
have considered up to this point is in elyect hinged at one support and fixed
at the, other, and therefore has a critical load of P -- 2.04Pt. For values of kl

tI

T
h

1Fig. 3-8
Jezek.

t~ - i

ldcalized hccun c o l u m ~ lof

(b)

a,

"7

uyb
Slroin
!c)

equal end couples M and axial forces P. It has been demonstrated by Jezek
(Refs. 3.3 and 3.4) that a closed-form solution for the load-?rf:ct.m c h m -

162

Beam Columns

Ch. 3

teristics, beyond theproportional limit, can be obtained provided tl~efollo~ving

assumptions are introduced.

Failure of Beam Columrrs

Art. 3.5

the curvature can be written as

d2v
SnZsin nx
;,
a=--

(3.61)

IZ

I . The cross section of the member is rectangular (Fig. 3-8b).

2. The material is an ideal elastic-plastic material (Fig. 3-8c).
3. The bending deflection of the member takes thc form 01'
sinewave.

;L

Ir;~lf-

The reason inelastic bending is difficult to analyze is -that the relation of

stress to strain varies in a complicated manner both along the 111elnberand
across the section, once the proportional limit has been exceeded. It is with
this problem in mind that Jezek introduces the foregoing assumptions.
Assumption 3 makes it possible to predict the behavior of the entiremember
from a consideration of the stresses a t only a single cross section, and assumptions 1 and 2 greatly simplify the manner in which stress and strain vary at
that one section. In addition to these major idealizations, the following
assumptions are made :
4. Defc>r~l~ations
arc finite but still small e ~ l o u gso
l ~ that thc cu~-v;~iurc
c:111
bc approxirnatcd
the second derivative.
5. The member is initially straight.
6. Bending takes place about the major principal axis.

::'the coordinate axes are taken as indicated in Fig. 3-8a, t l ~ ccxtcrnnl

bending moment a t a distance s from the origin is

1m

.' Substitution of (3.61) into (3.54) givcs

M

TIX
-+ Py = EI-STIZ
sin IZ
I

which reduces to
d

M -1.

p J ~ 6E16n2
-

(3.62)

I"

a t midspan.
Assuming that M is proportional to P,'we introduce the notation

and rewrile Eq. (3.62) in thc form

1 6)

or

Re

SEIz2
I2

-1 6 ) ==SP,

(3.64)

where P, = n2E//12is the Euler load of the member.' If both sides of (3.64)
are divided by the depth h and the terms rearranged, one obtains
'

This expression isYalid regardless of whether the elastic limit of' the n1;iteriaI
has been exceeded or not. The characteristics of the internal resisting moment
do, however, depend on the state of stress in the member. As long as Flooke's
law remains valid, the internal moment is given by the well-known relation

T o determine the relation between load and deflection up to the proportional limit, we equate (3.58) t o (3.57). Thus

where 0, = P,/bh is the Euler stress and a, = Plblz is the average axial stress.
As long as stresses remain elastic, Eq. (3.65) gives the correct loaddeflection relationship for the member. T o determine the load at which Eq.
(3.65) becomes invalid, one must consider the maximum stress in theme'mber.
That latter is

If the deflection is now assumed to be of the f o r n ~

or, substituting Pc for ll1 iind o, for Plbh,

nx

y = 6 sin -

I.,:

164

Beam Columns

Failure of Beam Columns

Art. 3.5

165

The distances c, d, and f are defined in Fig. 3-9a; and a, and a, are, respectively, the yield stress acting at the extreme fiber on the concave side of the
member and the tensile stress acting at the extreme fiber on the convex side.
The internal moment is obtained by taking the moment of all the forces
about the centroidal axis. ~ h 6 s

The elastic load-deflection relation given by (3.65) becomes inviilid wlicn

a,., as given by (3.67) equals the yield stress.

Of the three expressions, (3.57), (3.58), and (3.60), used to determine Lllc
elastic load-deflection relation, only (3.58), the moment-curvature relation,
must be revised when the elastic limit is exceeded. T o determine tlie inelastic
moment-curvature expression tliat is used in place of (3.58) in the illelastic
range, let us consider the stress distributions depicted in Fig. 3-9. As indicatcd,

Noting that f + c d = h, Eqs. (3.68) and (3.69) can be solved fo; c. After
some, fairly involved algebraic manipulations, which are not reproduced
here, one obtains

~ r o n ' Fig.
;
3-921 it is evident that

where p is the radius of curvature.

Stress distribution
for small e = M / P
(0)

Stress distribution
for large e = M/P
.(b)

Fig. 3-9 Stress distribution for beam colunin in inelastic range. (Adapted
from Ref. 1.12.)

two different distributions of stress are possible. If the ratio e = MIP is

relatively small, yielding occurs only or1 the concave side of the inember prior
to failure. This case is depicted in Fig. 3-9a. On the other hand, if e is relatively
large, both the convex as well as the cponcave side of the member will liave
started to yield before the maximum load is reached (see Fig. 3-9b). To
simplify the analysis, we shall restrict g,urselves to small values of e and thus
limit our concern to the stress distribution in Fig. 3-9a.
Equilibrium of forces in the s direction gives

Thus
.

from which

Finally, substituting the expression for cgiven in (3.70) into (3.73) leads to

This is the inelastic moment-curvature relation that must be used in place of

Eq. (3.58) once the stresses have excqeded the proportional linit.
In view of(3.57), (3.61), and (3.63), the curvature and moment at midspan
are given by

.which, r..?ter dividing both sides by bli, can be written as

72
Oqq '. 672
and

= P(e

+ 8)

166

Failure of ~ ' e a m
Columns

Art.3.5

Clr. 3

Beam Columns

Table 3-1

Substitution of these expressions into (3.74) leads to

0o

Since the Euler stress can be expressed as

a2EI x2EhZ
us=-=AI

1212

Eq. (3.75) can also be written in the form

)
]
T [ 2 (a,

(6

-Z

I
54 a,

a,

(3.76)

Load -dellcction data-for beam column

.,urnax

00

(ksi)

8/11

(ksi )

(ksi)

61h

0.036
0.080
0.137
0.212
0.314
0.463
0.710
1.150

6.4
14
23
34

8.0
8.5
9.0
9.1
9.0

0.21
0.24
,0.30
, 0.35
0.40

6
8
10
12
14
16

767

(b) Inelastic Range Eq. (3.76)

(a) Elastic range Eqs. (3.65)and (3.67)

3 .i:>;,;..

Equation (3.76) gives the load-deflection relation in the inelastic range. It can
be used from the onset ofyieldi;lg up to failure, provided failure occurs before
yielding commences on the convex side of the member.
With the aid of Eqs. (3.65) and (3.76) it is possiblc lo obtiiin lllc snli1.c
load-deflection curve, from thc beginning ol'lo;~dingto failurc, Sorilny mcmber that falls within the limit;ltio~lsoutlincd at thc start of thc ntlalysis.
As an example, let us consider a simply supported, rectangular, steel
beam column with the following dilnensions and properties:
length I = 120 in.
radius ofgyration r = 1 in.
ratio of moment to axial load e = 1.15 in.
yield stress a, = 34 ksi
modulus of elasticity E = 30 x lo3 ksi
Based on this. data

and

Ch= n2E-.- 20.6 ksi

(J/r)2

The load-deflection data for the elastic range, obtained using Eq. (3.65),
are given in columns I and 2 of Table 3.1~1.Corresponding to each set of
values for a, and d/lr listed in the table, the maximum stress has been determined using Eq. (3.67). The latter is given in column 3 of the table. It is evident
stl.ess i n the 111embcrrcnchcs 34 ksi. lhc
from these data that the i~~axirnuln

yield Stress, at approximately a, = 8 ksi., As a result, Eq. (3.76): the inelastic

load-deflection relation,must be usedto obtain deflections for axial stresses
in excess of 8 ksi. The load-deflection data for the inelastic range, obtained
using Eq. (3.76), a r e listed in Tablc 3- l b..
The entire load-deflection curvc, including both the elastic and the
inelastic portions, is plotted in Fig. 3-10. The solid line represents the actual
behavior of the member. A dashed line denoting the invalid part of theelastic
curve is also included for comparison. Up to a, = 8 ksi, the material obeys
Hooke's law and the deformations are relatively small. However, as soon as
yielding spreads beyond a, == 8 ksi, there occurs a noticeable decrease in the
stiffness of the member.This decrease builds up fairly rapidly until at approximately a, = 9.1 ksi the member is no longer able to resist an increase in load.
In other words, a, = 9.1 ksi represents the maximum load that the member
can support.
The results obtained here for a. rectangular section and for a perfect
elastic-plastic material are typical of the behavior exhibited by other shapes
and other materials. However, the determination of the maximum load for
most-other sections andmaterials involves considerably more effort than was
required to analyze the rectangular section with t h e perfect elastic-plastic
material. I n the majority of instances, closed-form solutions' of the type
presented here are out of the question, and numerical methods are the only
means available for obtaining the maximum load. One such numerical solution for the maximum load of a structural-steel 1 beam is given by Galambos
..
and Ketter (Ref. 3.5).
In view of the fact that the determination of the maximum load of a beam
column is invariably cot~lplcxand time consuming, the laad at which yielding

168

Beam Columns

Ch. 3

Design o f B e a h Columns-Interaction

A r t . 3.6

Equation

169

member, the calculations would hnic been evcnmore coinplex and fAr more
time consl~rning.The impracticability of obtaining the collapse load of a beam
colunln by purely theoretical procedures and the need for an empirical design
formula are thus self-evident.
When a member is subject t o a combined loading, such as bending and
axial compression, an interaction equation provides a convenient way of
approximating the ultimate strength (Ref. 3.6). Knowing the strength of the
member in both pure compression and pure bending and knowing that the
less compression and less bending when both of these loads
if.either is acting by itself, one con estimate how much
bendhig and compression can be resisted-if both are present. Such an approximation can then be verified experimentally.
To develop an interaction equation for combined 'bending ,and axial
compressiorl, let us introduce the ratios PIP, and MJM,,,where

P = axial load acting on the member at failure when both axial com-

pression and bending are present . .

= ultimate load of the member when only axial compression is
present, that is, the buckling load of the member
M = niaxi~nurn.primary bending moment acting on the -member at
.fi~ilurewhen both bcncling end axial compression ex.ist; this
cxcludcs thc amplification in thc morncnt duc to prcscncc of the
axial loatl
M,, = ulti~natebending morncnt when only bending exists, that is, the
plastic moment of the section

.'h

0.2

04

06

0.8

10

S
h

Fig. 3-10 Load-deflection curve for beam column.

begins has often been used in place of the maximum load as the limit of
structural usefulness. The load corresponding t o initial yielding is an attractive design criteria. because it is relatively eilsy to obtain and it gives a
conser :dtive estinlate of the actual collapse load. However, it does have the
disadvantage of being often too conservative. Fortunately, there has been
developed a n alternative semiempirical design criterion that is both ;~cc.tirate
and relatively easy t o use. This design &iterion, the interaction equallon, is
considered in the following article.

3.6 DESIGN OF BEAM COLUMNS

l NTERACTION EQUATION

Let us now calculate the above ratios for the rectangular beam column
ilnalyzed in Article 3 5. For that member the axial stress at failure was found
to be 0, -- 9.1 ksi and the Eulcr stress is a,, =: 20.6 ksi. .Thus

P -- 3
P, a,

--

9.1 -- 0.44
20.6-

The ratio of the niaxi~numprimary momcnt at failure to the plastic moment

of the section can bc written i n the form

In Article 3.5 the collapse load of a beam column was calculatcd. T o sin~plify
the analysis as much as possible, a. very idealized member was chosen, a
rectangular section made out of a perfect elastic-plastic material. Nevcrtiicless,
fairly lengthy and complicated calculations were needed t o obtain the desired
result. Had we attempted t o determine the maxiinun~load for sc)nlc olliel-

Substitution of elh

0.33, a,

9. I , and

a, = 34 gives

..

m-

.
170

'

Beam Columfrs

..

. .

..

References

Clt. 3

Ch. 3

The ratios PIP, = 0.44 and cl4/M,, = 0.35 give the ~liaximum\?i~lucsol'
P and h.1 that tlie recta~igularbeam column with e / h == 0.33 a~?slyzcclin

line
I

'

M . 1.0
--P -1 P u ' A*I,,
,

, .

i
1

Interaction equation for beam column.

giaph whose ordinate is PIP,, and whosc abscissa is MIM,,. C:~lcul:itio~i::

similar to those leading to point .4 II;I\T bccn cnrricd out by .Ic;r.ck (1tcl.s. 3.3
and 3.4) for rectangular beam columns with various values of e!l;, :ind tlic
results thus obtained i r e tabulated by Bleich (Ref. 1.12). Points 13 and C in
t h e figure have been plotted using these data. I'n addition to these results,
depicted by square points, the figure includes tllree circular points giving ihc
maxiniuni loads for structural-stecl I beams and two tri:i~igular poin~sthat
correspond to aluminum-alloy tubes. The data used to plot the circular points
were obtained by Galambos and Ketter (Ref. 3.5) using a numerical Integration nietllod, and the failure loads represented by the triangular points were
obtained experimentally by Clark (Ref. 3.7);
The lack of scatter exhibited by the failure loads plotted in Fig. 3-1 1
indicates that in all probability a single analytical expression can be found
which will predict themaximum load for a variety of different beam columns. .
Such a relation, which is also simple enough to be useful in routine engineering, will now be developed.
I t isfairly obvious that PIP, = I .O when MlM, = 0 and that MIM, == 1.0
5-h:q P,/P. = 0. The desired CUNe must therefore pass through the points
61, '2:): (0, 1). The simplest expression that satisfies this criterion is the straight

177

'

'

Article 3.5 can resist. In Fig. 3-1 1 this result is shown plotted as point /I O I I ;I

Fig. 3-1!

(3.77)

depicted by the dashed line in Fig. 3-1 I. All the theoretically and experhentally obtained failure loads included in the figure fall below this curve. It can
therefore be concluded that Eq. (3.77) gives an unconservativtestimate of t h e
maximum strength of beam columns and is not a satisfactory design criterion.
The reason for the discrepallcy between Eq, (3.77) and.the actual failure
,
loads is that M, in the equation, is only the primary part ofthe total moment
that sits on the member. In other words, M does not include-the secondary
nloment produced by the product of thc axial load and the lateral deflection.
It was shown in Article 3.2 that the preseocc of an axial load amplifies the
primarybending moment roughly by the ratio 1/[1 (PIP,)]. If this factor is
incorporated into Eq. (3.77), one obtains

This relation is shown plotted as a solid line in Fig. 3-1 I . It is evident that Eq.
(3.78) agrees much better with the actual failure loads than did the stiaight
line and that Eq. (3.78) appears to offer n satispdctory design criterion.
Although agreement has been shown to exist between Eq. (3.78) and only
a limited number of cascs, Eq. (3.78) is actui~llyable to predict the ultimate
load for n largc varicly of situations. Thc ccluation is applicable to I beams as
wcll as rectangular sections. and to aluminuni as wcll as steel. Furthermore, it
makes no difference whether the primary nlomcnt is due to eccentric axial
loading o r to transverse loads or to a combination of the two. The only
restriction is that the maximum momcnl occur at or near the center of the
beam. Equation (3.78) is still applicable if this condition is not satisfied.
However, a suitable factor must be introduced in the moment term of the
equation (Rcf. 3.8). Jn view of the fact that Eq. (3.78) is both simple to apply
and renlarkably accurate for a large number of different situations, it is used
extensively as a design criterion for beam columns.
References

3.1 J. I. PARCEL
and R. B. B. MOORMAN,
Analysis of Sfalically indererminate
Strrrctrrres (New York: John Wiley & Sons, Inc., 1955).
3.2 G . WINTER,
P. T. HSU,B. KOO,and M. H. LOH, "Buckling of Trusses and
Rigid Frames," Cornell University Ellginrering E-rperimental Station Bulletin,
No. 36, Ithaca, N.Y., 1948.

172

Ch. 3

Beam Columns

Ch. 3

Problems

3.3

P3-3 is given by the implicit relation

K. JEZEK," N a h e r ~ ~ ~ ~ g s b e r e c hder
n ~ ~Tragkraft
~lg
exzentrisch gcdriickter
~tahlstbbe,"Drr Sialrlborr, Vol. 8, 1935.

3.4 K. JEZEK,
"Die Tragfiihigkeit axial gedriicktel. und n~lfBiegi~ngbconspruchIcrStahlstabe," Der Stal:lbatr, Vol. 9, 1936.
Noting that

3.5 T. V. GALAMBOS
and R. L. KETTER,
"Columns Under Combined Bending and
Thrust," Traiisactioiis, ASCE, Vol. 120, 1955.

- P + P(" -t u m , x ) c
uy= umax
- A
I

and E. I. RYDER,"Stress Ratios." Avio/iorr, ~01.'36,No. 6, 1937.

3.6 F. R. SHANLEN:
3.7 J. W. CLARK,
"Ec~c~itric:~IIy
Loiided Columns," ~crtrsac/io~rs,
A S C E , Vol. 120.
1955.
3.8 W. MCGUIRE,Steel S ~ I - I I C I I ~(Englewood
~~S
Cliffs, N.J.: Prentice-Hall, Inc..
1968).

Problems

3.1 Obtain expressions for the niaxin~unldeflection and maximum monicnt of a

beam column whose ends arc built in and that is loaded with a concentrated
load a t midspan as shown in Fig. P3-1.

3.2

~ e t e r r n i n cthe niaxinium moment for ;I beiilii column that is bent in n reverse

curve as shown.in Fig. P3-2, when
.

(a) PIP, = 0.2

(b) PIP, = 0.8
where P, = nZEIIL2. In view o f the foregoing results, what can be concluded
regarding the maximum moment in a beam column with reverse curvature?
3.3 T h e load P a t which yielding commences in the bean1 column shown in Fig.

1-

*I

Fig. P3-3

II

derive the relation.

The terms used in the relations are defined as follows:
A = cross-sectional area
d
c = distance from neutral axis t o extreme fiber
r = radius of gyration
a, = yield stress
= maximum stress