344

Section D

COLUMN DESIGN

J.W. Butterworth* and K.C.F. Spring**

This paper is the result of deliberations of the Society's

Study Group for the Seismic Design of STEEL STRUCTURES.

1. CONTENTS Unless noted otherwise, the column

section that is implied throughout is an I
2. INTRODUCTION or box section.
3. DUCTILITY
3. DUCTILITY
3.1 Available Ductility
3.2 Ductility Demand
3.1 Available Ductility in Columns
3.3 Recommendations
4. EFFECTIVE LENGTHS The ability of a rolled steel column
4.1 Introduction to dissipate energy in a stable flexural
4 . 2 Elastically Responding mode through a plastic hinge at one end has
Structures been demonstrated in a series of tests
4.3 Inelastically Responding reported in (1) . The stability of the pro-
Structures cess was shown to depend on a number of
4 . 4 Effect of Drift Limitation factors, particularly lateral torsional
4.5 Recommendations buckling, local buckling and the level of
axial load. The effect of axial load is to
5. LATERAL BUCKLING
amplify any tendency to lateral or local
5.1 Recommendations
buckling leading to accelerated strength
5.1.1 Elastically Responding
degradation.
Structures
5.1.2 Limited and Fully
Provided local buckling is prevented
Ductile Structures
by suitable limits on the cross-section
6. LOCAL BUCKLING geometry (see section 6 ) , it was found in
the tests described in (1) that good duc-
7. MOMENT-AXIAL LOAD INTERACTION tility under inelastic cyclic loading was
7 .1 Introduction obtained when the axial load to squash load
7.2 Axially Loaded Members ratio P/Py was less than 0.5. Strain-
7.3 Combined Axial Load and hardening reduces the area of steel needed
Moment to carry the axial load, resulting in a
7.3.1 At a Support small increase in flexural capacity and
7.3.2 Away from a Support better energy dissipation than might be
expected. Axial shortening occurs due to
8. SHEAR
accumulated strain and results in consider-
9. NOTATION able energy dissipation. Unfortunately
10 . REFERENCES this dissipation is not helpful in counter-
acting seismic energy input.
11. APPENDIX
2. INTRODUCTION Test results showed that with
P/P > 0.5 sharp drops in column strengths
y

occurred.
The following paper looks at the seis-
mic design of columns in braced and unbraced
3.2 Ductility Demand in Columns
structural steel frames. The design of such
columns will be considered for the case
The usual strong column-weak beam
where the column is to remain elastic and
design strategy forces plastic hinges to
also for the less usual case in which
form predominantly in the beams where
hysteretic energy dissipation is to take
reliably ductile behaviour can be readily
place in the column.
achieved. Capacity design procedures ensure
that storey sway mechanisms should not
In view of the small amount of
occur and that column plastic hinges will
research that has been conducted in New
be required to contribute only a very small
Zealand on steel column design, much of the
proportion of the total ductility demand of
following has been adopted from results of
the structure.
research done principally in the USA, Japan
and Europe.
Columns which are expected to hinge,
typically those at ground floor level, but
possibly at other levels depending on
vertical accelerations and mode of vibra-
Senior Lecturer, School of Engineering,
tion , should have their P/P and slenderness
University of Auckland, New Zealand.
ratios limited as suggested in section 3.2,
Partner, Brickell Moss & Partners,
Wellington. have cross-sections complying with

BULLETIN OF THE NEW ZEALAND NATIONAL SOCIETY FOR EARTHQUAKE ENGINEERING, Vol 18, No 4, December 1985
 345

section 6 and be laterally restrained in where P is the Euler load of the column
E

accordance with section 5. and P is the axial load in the column at

c r

the frame critical load.

Low rise structures with long span
and gravity dominated beams will generally 4.2 Elastically Responding Structures
require columns of high flexural capacity
which will have naturally low P/P ratios. Effective length factors of columns
The column hinging that is almost inevitable in elastic structures may be determined
in such structures is unlikely to create a most accurately by carrying out a stability
ductility demand that cannot be readily analysis of the structure and substituting
catered for. in the relationship above. Alternatively,
there are many approximate methods, especi-
3.3 Recommendations ally for regular framed structures, such as
those set out in appendix E of NZS 3404 in
The advisability of allowing plastic the form of charts. Various refinements
hinges to form in columns is also dependent and corrections to these charts are avail-
on slenderness ratio and the curvature able to allow for effects such as varying
induced by the terminal bending moments. column sizes and loads, flexible connections
In particular, care must be taken to avoid and column base rotations (2,3,4). The
hinges forming away from the ends at appendix to this paper contains an example
unrestrained interior points. Compliance illustrating the conservatism of the chart
with rule 10.5 of NZS 3404 should eliminate approach for a structure with varying column
potentially unstable columns due to this loads across a storey.
effect. The following recommendations are
therefore made: Note that if a second order analysis
F used to determine column actions then
Column ductility P/P ^ j_ 9 C C 1 >"£ Isr-lY - each column need
y z ca reater than unity, as to do
tively magnify the actions
< 0 3
Fully ductile 3 + 6 - >
- _pc idxng Structures
a n d K
T + 3 + X
ted ductile or
< 0 .7
fully ductile -a"' ss tnei e is a possibility
r

Limited ductile .1 + 3 - A of inelas J j c > ^sno^se =r factored loading

a n d <
1 + B + A and a certainty of it dur ing capacity design
(although not necessarily in the columns).
Elastic < 1.0 In either case it is nece ssary to ensure
that elastic or elasto-pi astic stability
effects do not prejudice the ability of the
structure to achieve the desired strength
Figures la and lb illustrate the state or energy dissipati ng mechanisms.
bounds to the regions outside which plastic
hinging should not be permitted to occur.
Effective length factors are the same
for both elastic and inelastic columns only
EFFECTIVE LENGTHS when the column acts as an independent mem-
ber with end conditions that may include
4 .1 Introduction
hinged, free or fully fixed, with or without
The concept of effective length was lateral sidesway restraint. When the column
introduced into elastic column theory as a is part of a continuous frame, the relation-
simple means of applying formulas developed ship between its rotational stiffness and
for pin-ended columns to columns with other the stiffness of adjacent restraining mem-
end conditions. In the case of an isolated bers is variable with respect to both column
column the effective length is readily load and the effective tangent modulus of
understood to be the length of a pin-ended the material. It follows that K c an no t be
column which has the same critical load, determined a priori as in the elastic range
cross-sectional and material properties as for which column stiffness varies only with
the column being studied. axial load.

Thus if KL is the effective length and Yura (5) has recommended the use of a
P is the critical load of the column, modified value of stiffness ratio G
cr '
TT EI 2 Z(Eml/L) ,
cr „ _ T ' columns
T(EI/L)
(KL) ' beams

When the column can not be effectively where E ^ is the tangent modulus of the
isolated from its parent structure the column material. The standard alignment
charts are then used in an iterative proce-
effective length is taken to mean the length
dure to obtain the inelastic K value for
of an equivalent pin-ended column whose
the column.
Euler load equals the axial force in the
real column at the moment when the structure
Le Messurier (6) has proposed an
as a whole reaches its critical load.
alternative and more accurate approach to
determine K factors for columns in sway
Rearranging the equation above
frames. His approach, in common with Yura's,
requires no more than a first order analysis
/PITP" of the frame.
 Figure la Fully Ductile

Figure lb Limited Ductility

 347

4.4 Effect of Drift Limitation by a factor of 1.3. Further research to

support this approach is needed.
Interstorey drift limitation imposes
a lateral shear stiffness constraint on a 5. LATERAL BUCKLING
frame for the purpose of limiting damage to
non-structural components. Teal (7) argues If a column has differing bending
that any reasonable control of drift will stiffnesses about each of its principal
also ensure structural stability (sidesway axis and bending moments are applied to the
mode), even in the inelastic range, and that stronger of the two axes, then it may not
real problems of instability due to p-delta be able to develop its in-plane rotational
effects in seismic test frames are few and and bending capacities before failure occurs
are connected only with large enough forces due to lateral torsional buckling.
to cause extreme inelastic response.
In order to maintain in-plane rotation
This view is supported by the draft and moment capacities, sufficient bracing
revision of the SEAOC code (8} in which it must be provided to prevent lateral deflec-
is proposed that frames designed to the tion and twisting occurring.
code drift limit for seismic zones 3 and 4
may be considered "braced against joint 5.1 Recommendations
translation". The effective length factor
K for columns in the plane of seismic ben- 5.1.1 Elastically responding structures
ding forces can therefore be taken as 1.0.
Lateral bracing should conform to the
A comparison of the "average" building requirements of NZS 3404 , chapter 5.
shear stiffnesses implied by various code
drift limits for a ductile, moment resisting 5.1.2 Limited and fully ductile structures
frame founded in non-flexible soils is
presented in table 1. Shear stiffness is The provisions of section 10.9 of
calculated in dimensionless form as NZS 3404 are intended to ensure that a mem-
ber does not buckle laterally before
shear stiffness = Vh/Wd,
achieving the strains of up to the strain-
hardening level (11). This should result
where V = seismic base shear, h = storey in the section being able to sustain
height, W = seismic weight, and d = inter- sufficient inelastic rotation to justify
storey drift. limited ductile categorisation.

For the fully ductile case it is

Code Period 1.2 sees 0 .5 sees
suggested that the rotation ratio R be
increased from the value of 10 used in
Zone A 7.5 15 .0
NZS 4203 plastic design to 24 in order to allow
Zone C 7.5 15 .0
adequate rotational ductility.
Zone 3 9.1 12 .2 If Ly denotes the length of column
UBC (9)
Zone 4 14.1 18.9 over which the compression flange is fully
yielded, then the spacing of restraints to
Zone 3 8.3 14 .9
SEAOC (8) the critical flange is governed by whether
Zone 4 11.1 19 .9
L is greater or less than 640 aa, where
a = 1.5//1 + R/8 and a = r //F^.
Y

Table 1. Dri ft limit-implied shear stiff-

nesses , (Vh/Wd) Section 10.9 of NZS 3404 allows L
to be calculated as the length of column
It can be seen from the table that over which the bending moment M £ 0.5 M p . C

computed deformations under the seismic In the presence of axial load the 0.85
loadings of NZS 420 3 would need to be factor used may be unconservative as it
increased by about 30 percent to give shear corresponds to the yield moment in a member
stiffnesses comparable with those of the with zero axial force and a typical I-beam
UBC or draft SEAOC code provisions. shape factor. When axial force is present
the compression flange yields at a lower
4.5 Recommendations moment which is given by

Whi1st greater availability of M y c = (0.85/1.18)M pc

efficient and reliable second order analysis

computer programs should result in a decline
in the need for approximate methods of Allowing a small margin between first yield
effective length determination, a short term and full yielding of the flange leads to a
need remains for simplified methods. suggested value of 0.75 M ^ as the basis for
c

Methods such as those of Yura (5), Wood (10) calculating L . For P/P ^< 0.15 the origi-
y

and Le Messurier (6) are suggested as nal value of 0.85 is satisfactory.

amongst the best available for the inelastic
cases, although further research is still Table 2 sets out the appropriate
needed if they are to be applied with restraint spacing for the fully and limited
confidence. ductile cases.

The adoption of K - 1 on the basis of 6. LOCAL BUCKLING

meeting the drift criterion of NZS 4203 is
recommended provided deformations are cal- The effects of local buckling on the
culated on the basis of an accurate struc- behaviour of columns have been illustrated
tural model (including joint and shear by a number of authors (12). In general
deformations, for example) and are multiplied the occurrence of local buckling in a
348
section will reduce its strength and its 7.3.2 Away from a support
capacity to absorb energy hysteretically.
The strength and rotation capacity at a The effect of axial compression on
plastic hinge in a column are strongly the moment capacity of a column shall be
dependent on the ability of the flanges and defined by
web to achieve strains of at least strain-
hardening level prior to the initiation of Bending about the major principal axis
local buckling. (i) For P / P 0.15, P / P + M / M p < 1.0
a c
<
a c X

Table 3 sets out the suggested limits (ii) For P / P a c * 0.15,

on cross-section geometry which should
ensure that the various member ductility
P/Pac < 1.0
levels are not impaired by local buckling. ac + (i - P / P o c x )M o x

The values generally follow Section C (13)

recommendations. For web geometry the axial
load effects impose more restrictive values Bending about both principal axes
than flexural loading. (i) For P / P < 0.15, a c

P/Pac + M / M + M /M < 1.0 x o x y p y

7. MOMENT-AXIAL LOAD INTERACTION

(ii) For P / P a c * 0.15
7.1 Introduction
c
mx x M
C m y M y

There is currently a high level of P/Pa c +' f, * 1.0

research and discussion in the area of ac P I
1 -p ^ - 1 M.
moment-axial load interaction in columns,
P
ocxr° x
ocyJ
especially in relation to the change to
limit states or LRFD codes in several coun-
tries. Chen and Lui (14) review the current 8. SHEAR
and proposed US design criteria and Chen
and Atsuta (15) provide an extensive back- In general shear is not a dominant
ground to the problem. action in the design of a column. However,
should a column mechanism be critical in
The linear interaction equations pro- determining the strength of a structure,
vided in NZS 3404, section 14.5, although then it may be necessary to investigate the
conservative in some situations are recom- interaction of bending, shear and axial
mended to be retained until the next major load. In such cases the interaction formula
revision of steel design code philosophy. of Neal (16) is recommended:
The equations given in NZS 3404 contain a
number of errors which have been corrected
in the version given below. (V/V ) n

M/M p + (P/P ) y + < 1.0

7.2 Axially Loaded Members [1-(P/P ) 3 y

The maximum load capacity of members NOTATION

not subjected to bending shall be
A F
_ s ac
ac " 0.6 b Outstand of flange beyond connection
to web (I columns), or clear distance
where F is determined from NZS 3404, between sides of box column
a c

chapter 6, for the appropriate effective d Clear depth of column web

length.
d Interstorey drift
7.3 Combined Axial Load and Moment E Elastic modulus in tension and
compression
7.3.1 At a support
E T Tangent modulus in compression
The effect of axial tension or com- F a c Maximum permissible compressive stress
pression on the moment capacity of the I- in a column in the absence of axial
member shall be defined by load
Bending about the major principal axis F y Yield stress of column material (in
MPa units when used in dimensionally
(i) For P/P y < 0.15, M / M p < 1.0 inconsistent expressions)
(ii) For P/P y * 0.15, P / P + M/(1.18 M
Y ) < 1.0 G Ratio of inframing beam stiffnesses
to column stiffness at beam-column
Bending about the minor principal axis joint
(i) For P / P < 0.4, M/Mp < 1.0
y
h Storey height
(ii) For P/P > 0.4, y
I Second moment of area about a princi-
( P / P ) + M/(1.18 M ) < 1.0
y
2
pal axis
K Factor by which column length must
Bending about both principal axes
be multiplied to give effective length
(i) For P/P < 0 . 1 5 , M / M
y + M /M x p x y p y < 1.0
L Actual length of a column
(ii) For P/P ^ 0.15,
y
L y Length of column over which compres-
P/P + M / ( 1 . 1 8 M
y x p x ) + M /(1.18 M y p y ) <1.0 sion flange is fully yielded
M Bending moment in column
 349

Calculated maximum moment capacity of Criteria", AISC Eng. J., Vol. 12,
c o l u m n in absence of axial load (may No. 2 (1975) .
be governed by lateral buckling or
plasticity) about the major axis (8) SEAOC, "Recommended Lateral Force
Requirements and Commentary", Draft
Fully plastic moment c a p a c i t y o f a
Revision, 19 84.
c o l u m n section in t h e a b s e n c e of
axial load
(9) "Uniform Building Code", International
F u l l y plastic m o m e n t c a p a c i t y o f a Conference of Building Officials,
c o l u m n s e c t i o n in the p r e s e n c e o f California, 1982.
axial load
M o m e n t to c a u s e first y i e l d in a
(10) Wood, R. H . , "Effective Lengths of
yc column in the p r e s e n c e o f axial load
Columns in M u l t i - s t o r e y Buildings",
in three parts, Struct. E n g . , Vol. 52,
Axial load in column Nos. 7, 8 and 9, July, August, Sept.,
Euler load o f p i n - e n d e d column 19 7 4 .

E l a s t i c c r i t i c a l load of column w i t h (11) Lay, M.G., "Source Book for the

length = e f f e c t i v e l e n g t h Australian Steei Structures Code ~
Squash load o f c o l u m n (A Fy)
AS1250", AISC, 1975.
s

Ratio of r o t a t i o n at a plastic h i n g e (12) Mitani, Isao, Makino, Minoru, Matsui,

to relative elastic r o t a t i o n o f far Chiaki, "Influence of Local Buckling
e n d of column segment containing the on Cyclic Behaviour of Steel Beam-
hinge Columns " , Proc. 6th World Conf. on
E a r t h q u a k e Eng., N e w Delhi, India ,
Radius of oy.; 3 ; or o f c r 0 5 s - se ~t icv>
Vol. 3, 1977, pp. 3175-3180.
a b o u t a pr.. r_i. ' dxis
F l a n g e thic;; • (13) Walpole , W.R. a n d Butcher , G . W. , "Beam
D e s i g n " , S e c t i o n C - N Z N S E E Study G r o u p
Web thickness
on the Seismic Design of Steel Structures ,
Seismic b a s e s h e a r force o n 3 s t r u c - B u l l . N Z N S E E , Vol. 18, N o . 4, Dec. 1985 .
ture
S h e a r force at p l a s t i c h i n g e in a
(14). Chen, W . F . , Lui , E.M., "Stability
column
Design C r i t e r i a for Steel Members
and Frames in the United States",
Shear force to c a u s e p l a s t i c s h e a r J. Construct. Steel Research, 5, 1985.
hinge in a b s e n c e of a x i a l l o a d or
bending m o m e n t (.15) Chen, W .F . , Atsuta, T . , "Theory of
Seismic weight of a structure Beam-Columns", 2 Vols., McGraw-Hill,
19 77 .
6 Ratio of s m a l l e r c o l u m n end m o m e n t to
l a r g e r , m e a s u r e d in t h e same (16) Neal, E.G., "The Effect of Shear and
rotational direction Normal F o r c e s on the Fully Plastic
Moment of an I Beam", J. Mech. Eng.
Normalisec s i e n d e r n e s s rat
Sci, , Vol. 3, 1961,

11. APPENDIX
10. REFERENCES
The f o l l o w i n g example illustrates the
potential c o n s e r v a t i s m of the code chart
(1) Popov, Egor P., Eertero, Vitelmo V.,
approach to effective length determination.
and Chandramouli , S , , "Ilys tere tic
Behaviour of Steel Columns", EERC
Report No. 75-11, September 1975 . Consider the frame shown below and
assume that adequate out-of-plane restraint
is provided. We wish to determine the
(2) Chu, K.H., and Chow, H.L., "Effective
effective length and column reliable loads
Column Length in Unsymmetrical Frames",
as governed by in-plane frame buckling.
IABSE, Vol. 29-1, 1969.
P/4
(3) De Falco, F., and Marino, F.J.,
"Column Stability in Type 2 Construc-
tion" , AISC Eng. J., Vol. 8, No. 2 C (rigid)
(April 1966).
5m
(4) Galambos, Theodore V., "Influence of
Partial Base Fixity on Frame Stability'
Trans. ASCE, Vol. 126, part 2, paper
3256, pp. 929-969 .
A and B
(5) Yura, J.A., "The Effective Length of
Columns in Unbraced Frames", AISC 4564000 m m 4

Eng. J., Vol. 8, No. 2(April 1971).

88.1 mm
(6) Le Messurier , W . J . , "A Practical 3604 kN
Method of Second Order Analysis",
AISC Eng. J., Vol. 14, No. 2 (19 7 2 ) . A stability analysis gives the critical
load of the frame as
(7) Teal, E.J., "Seismic Drift Control
P r r - 1434 kN
350

so K A = ^P /P r E C

= 1.59

and Kg = /P /0.25E P c r

= 3.17

For column A: KL/r = 9 0

giving P r e l i a b l e = 1362 kN

For column B: KL/r = 180

giving P r e l i a b l e = 274 kN

Column B governs, so maximum reliable

load on frame is
P = 4 x 274
= 1096 kN
T h e c o d e
chart approach to this
problem gives K = 2 for both columns.

The maximum reliable load in column A

and column B is

Reliable = 6 3 2 k N

Column A therefore governs now and the

maximum reliable load on the frame is

P = 632 kN

Table 2. Spacing of lateral restraints

Fully ductile Limited ductility

R 24 10

a 0.75 1.0

L y < 480 a L y < 640 a

placing of critical within or at one within or at one

flange restraint end of Ly end of Ly

adjacent restraint £ 720 a * 960 a

L y * 480 a L y ^ 640 a

spacing of critical
< 480 a < 640 a
flange restraints

Table 3. Section geometry limits

Member ductility Fully ductile Limited ductility Elastic

b/T (I column ^ 120//F^ * 136//F^ £ 256//F^

flanqe)
b/T (box flange) ^ 500//F^ £ 512/ZF^ ^ 560//F^

d/t (web) ^ 500//F^ ^ 512//F^ * 560/^