Cost Project C1 "Semi-Rigid Behaviour": Points of Reference
Cost Project C1 "Semi-Rigid Behaviour": Points of Reference
Cost Project C1 "Semi-Rigid Behaviour": Points of Reference
"SEMI-RIGID BEHAVIOUR"
C1/WD/92-0
DELEGATION: ITALY DATE: October 28-30, 1992
KEY WORDS: Rotation capacity, steel members, local buckling, seismic design.
ABSTRACT:
In this paper a new method for evaluating rotation capacity of beams and beam-columns
is presented. The method can be defined as semi-empirical because it is based on the
theoretical evaluation of the moment versus curvature relationship and on an empirical
relation for calculating the critical stress leading to the local buckling of the compressed
flange of H cross-sections. This empirical relation takes into account both the longitu-
dinal stress gradient due to the bending moment variation and the influence of the steel
mechanical properties.
The results obtained by means of the proposed method are compared with the
experimental data available in technical literature.
The proposed formulations can be used in order to evaluate the available ductility of the
member plus connection as a whole.
1 - INTRODUCTION
In limit design of structures it is postulated that plastic hinges have a sufficient rotation
capacity. Therefore it is clear that the cross-section of members have to satisfy precise geome-
trical requisites in order to provide plastic deformations until the collapse mechanism of the
structure is reached without loosing its load carrying capacity. The knowledge of the member
rotation capacity is an unquestionable need in all cases in which the computation of the available
ductility of the member plus connection as a whole is requested.
The rotation capacity of steel members is undermined by the occurrence of local buckling of
the plate elements which constitute the member cross-section and, if torsional restraints are not
provided, by the occurrence of lateral torsional buckling.
In order to design cross-sections able to provide sufficient rotational capacity, the local
buckling phenomenon has to be controlled. In particular the occurrence of local buckling in
elastic range has to be absolutely avoided; therefore, geometrical properties of the cross-sections
have to guarantee the attainment of buckling in plastic range.
For these reasons modern codes provide limitations to the width-to-thickness ratios of the
cross-section of members. A very important concept in design of steel structures, which has been
introduced for the first time in Eurocode 3 [1], is represented by the subdivision of the structural
sections into different behavioural classes (fig.1):
- class 1: plastic sections, i.e. characterized by the capability to develop a plastic hinge with an
high rotation capacity;
- class 2: compact sections, i.e. sections able to provide their maximum plastic flexural strength
but with a limited deformation capacity;
- class 3: semi-compact sections, for
which the bending moment leading
M SECTION CLASSES to first yielding can be attained but,
Mp 1 - ductile due to local buckling phenomena,
Mu 2 - compact
plastic redistribution is not possible;
1 3 - semi-compact
Mp - class 4: slender sections, in which
2 4 - slender
1
the total elastic flexural strength can-
My 3
not be developed due to the prema-
Mp
4
ture attainment of local buckling of
the compressed parts of the section,
so that failure occurs in elastic range.
R It is clear that the parameter go-
1 u max verning the above behaviours and,
/ y
y y
therefore, which defines the class of
the structural sections is the width-
to-thickness ratio b/t of the compres-
sed plates which constitute the sec-
Fig.1 - Cross-section behavioural classes
tion.
In [2,3,4] it has been pointed out that the theoretical and experimental backgrounds of b/t
limitations for defining classes of structural sections are not completely assessed. In particular
width-to-thickness ratio limitations are prescribed independently for flange and web. It seems
that such independent limitations are unreasonable because, obviously, flange is restrained by
the web and web is restrained by the flange.
The beams and columns of rigid frames subjected to horizontal forces have to withstand
double curvature bending which can be simulated by an assembly of configurations of cantilever
beams. Moreover, rotation capacity of cantilever beams can be compared to the one of centrally
loaded beams which are usually adopted as test specimen (fig.2). Rotation capacity can be defined
as the ratio between the plastic rotation at the collapse state to the elastic one (fig.1):
p u y u (1)
R = = = 1
y y y
This definition leads to a cautelative evaluation of the rotation capacity because it neglects
the deformation capacity due to the post-buckling behaviour which is represented by the
softening branch of the moment versus curvature relationship (fig.1).
In order to evaluate rotation capacity of steel members, different methods have been proposed.
They can be divided into three groups:
- theoretical methods;
- semi-empirical methods;
- empirical methods.
Theoretical methods are based upon the approximated theoretical evaluation of the moment
versus curvature relationship of the member cross-section and upon the theoretical analysis of
buckling phenomena.
Semi-empirical methods differ from theoretical ones due to the fact that local buckling
phenomena are taken into account by means of relations provided by experimental evidence. To
this scope stub column tests have been performed.
Empirical methods are based upon the statistical analysis of experimental data of full-scale
member tests. Therefore, they can take into account by means of experimental evidence both
local and global buckling phenomena.
In [5] a detailed analysis of dif-
ferent methods belonging to the
above groups is presented and the
F
corresponding results are discussed
on the base of experimental eviden-
L L ce. The examination of the different
approaches has pointed out that the
M evaluation of the critical stress lea-
ding to the local buckling of the
v compressed flange represents al-
ways the crucial step. As the local
buckling phenomenon has to be re-
v
stricted to the plastic range and the
F/2
restraining conditions of the com-
pressed flange are not easy to defi-
ne, theoretical evaluations of the
Fig.2 - Structural schemes adopted for test specimen critical stress are very difficult.
Therefore, as confirmed by the comparison with the experimental data, semi-empirical methods
are preferable [5].
The semi-empirical method proposed by B.Kato [2,3,4] is based on a simplified stress-strain
behavioural relationship of the material. Moreover, it doesnt take into account that, due to the
bending moment variation, the compressed flange is subjected to a longitudinal stress gradient.
For these reasons, a new semi-empirical method is proposed in order to improve the prediction
of the rotation capacity of beams and beam-columns in steel frames subjected to horizontal forces.
2 - THEORETICAL BASES
Moment-curvature relationship
Semi-empirical methods require the evaluation of the moment versus curvature relationship.
For actual H cross-sections, this relation depends upon the thickness of flanges and web and is
influenced by the distribution of residual stresses. In approximate way the theoretical evaluation
of the moment versus curvature law can
be performed with reference to an ideal
arctg Eh
y two-flange section. This approach,
which is generally accepted in technical
literature [6,7,8], has been used by B.
h y arctg E Kato [2,3,4] also for analyzing the defor-
y h mation capacity of square hollow sec-
tions and circular hollow sections. In all
cases the equivalent ideal two-flange
section has the same area of the actual
y
one, while its depth is computed by im-
posing the equivalence of plastic mo-
ment for both sections. Finally it has to
be taken into account that in the theory
Fig.3 - Stress-strain relationship adopted in the developed in [2,3,4] the rotation due to
theoretical analysis the part of the member which remains in
elastic range has been neglected by as-
suming a rigid-plastic-hardening or a ri-
gid-hardening law for the stress-strain relationship of the material.
In this section the moment-curvature relationship of an ideal two-flange section is presented
by assuming an elastic-plastic flow-strain hardening behaviour of the material (fig.3).
Let:
- No the external axial load;
- M the external bending moment;
- y the yield strain;
- h the strain corresponding to the beginning of the strain hardening;
- E the elastic modulus;
- Eh the hardening modulus;
- y the yield stress;
- cr = sy the critical stress due to the local buckling of the compressed flange or to
lateral-torsional buckling (with the limitation s u/y, being u the ultimate stress of the
material);
- o = y the stress due to the external axial load;
- A the area of the cross-section;
- he the depth of the two-flange section;
Ahe2
- Ie = the moment of inertia of the two-flange section.
4
Moreover, the following positions are made:
(2)
= y
y he
and:
(3)
= h
h h
e
Furthermore, the introduction of the reduced modulus of elasticity:
2EE (4)
Er = h
E+ Eh
allows a convenient simplification of the moment versus curvature relationship.
Two possibilities can occur:
a) the critical bending moment (s ) Mp is obtained when the lower flange is still in elastic
range;
b) the critical bending moment (s ) Mp is obtained when the lower flange is in plastic flow or
in hardening range.
It can be shown that the first situation is verified if the following condition is satisfied:
s 1 (5)
2
(s- )Mp
arctg ErIe
(1- )Mp
arctg EI e
1 2 cr
arctg ErIe
(1- )Mp
arctg EI e
1 2 3 4 cr
in the opposite case the lower flange is in plastic flow or strain hardening range when the critical
bending moment is attained.
In table 1 the moment versus curvature relationships are summarized for the three fundamental
cases: (s1) 2 (fig.4), (s1) 2 (fig.5) and = 0.
Moreover it is useful to note that by assuming h = y in the relations provided in table 1 the
moment versus curvature relationship for the case of elastic-strain hardening behaviour is
obtained.
L (s 1) Mp (6)
u = (1 )2 y + (s 1) (1 2) y + h +
s 2Er Ie
s 1
Case:
2
By integrating the curvature diagram, the following equation is obtained (fig.7):
L Mp (s 2 1)2 Mp (7)
u = [ (1+ 2 ) 2 (s 1) ] + 2 (s 1) + 2 (s 1) +
s y h E I 2E I
r e h e
z
(1- )M p
(1- )M p
(1+ )M p z
(s- )M p
(s- )Mp
M M
(1- )L L (1- - )L L L
z z
1
1
2
2
3
cr
cr
Fig.6 - Evaluation of the ultimate rotation Fig.7 - Evaluation of the ultimate rotation
for >(s-1)/2 for <(s-1)/2
Case: = 0
The relation for evaluating ultimate rotation of beams is obtained by means of equation (7)
with the condition = 0 which provides:
L (s 1)2 Mp (8)
u = + 2 (s 1) +
s y h 2E I
h e
1 h s 1 h h E I (12)
h
R = (1 ) + (1 + + (s 1) 1
Er Ie
2)
s he 1 he y he
s 1
Case:
2
h
R =
1
[1+ 2 2 (s 1) ]
h
+ 2 h (s 1) +
E I
4 (s 1) + (s 2 1)2
E I
1
(13)
(s )(1 ) h h E I E I
e y e r e h e
Case: = 0
The equations for evaluating the rotation capacity of beams are obtained by assuming
= 0 in relation (13).
Finally it is useful to point out the general validity of the above equations, from which in
particular the case of elastic-strain hardening behaviour can be obtained by assuming h = y
and the case of rigid-strain-hardening behaviour by assuming h = 0.
d (16)
w = w
y
tw
are the slenderness parameters of the flange and of the web respectively (bf is the width of the
flange, dw is the depth of the web, tf is the thickness of the flange and tw is the thickness of the
web).
Relation (14) can be adopted independently from the grade of steel. Moreover, more accurate
relations which take into account the effects due to the mechanical properties of steel, are given
by the same Author for H-sections and with reference to steel grades currently adopted in Japan
[4]. It has to be pointed out that the given experimental relations for evaluating the critical stress
leading to the local buckling of the compressed flange are based upon stub column tests in which
webs are uniformly compressed, while webs in beams and beam-columns are only partially
compressed. This difference can be taken into account by introducing the effective width of
the web which represents the compressed part of the web when the section is fully yielded. With
reference to fig.8, it results:
Nw = N = A y = (1 2 ) Aw y (17)
therefore:
Aw A (18)
=
2 Aw
and:
1 A (19)
dw = 1+ dw dw
e 2 Aw
The main difficulties in the theoretical evaluation of the rotation capacity arise from the
necessity to compute the critical
stress leading to the local buc-
kling of the compressed flange.
y y The value of this critical stress is
influenced by the restraining ac-
dw y tion exerted by the web which,
dwe depending upon the stress state
1 bf E h (20)
= 0.546321 + 1.632533 2f + 0.062124 2w 0.602125 + 0.001471 + 0.007766
s L Eh y
The proposed relation (20) is more complete than the ones provided in [2,3] because it includes
the influence of the slenderness of the flange and of the web as well as the effects due to the
stress gradient and the influence of the mechanical properties of the material.
Moreover, it seems that the simplified model (ideal two flange section), which has been
assumed for evaluating the ultimate rotation, has to be rationally kept also in the computation of
the first yielding rotation. As a consequence, the proposed method is based upon the use of
equations (12) and (13) for evaluating rotation capacity and, obviously, upon the computation
of the critical stress leading to the local buckling of the compressed flange by means of equation
(20); the control of the conditions leading to the lateral-torsional buckling of the member is made
according to the following section.
Lateral-torsional buckling
The buckling stress ratio s can be computed by means of the relations presented in the previous
section if local buckling occurs before flexural-torsional buckling. This condition is verified
when the distance between torsional restraints is sufficiently small in order to prevent the lateral
displacement of the compressed flange. In the opposite case, the buckling stress ratio has to be
computed as the minimum value between the one leading to local buckling and the one leading
to lateral-torsional buckling.
This last value can be computed, taking into account moment-gradient, by means of the
relations which provide the ultimate equivalent uniform bending moment for lateral-torsional
buckling and the relations correlating end moments to the equivalent uniform bending moment.
For these formulations reference can be made to Eurocode 3 [1].
16
fluence of the part of the member which remains
UNSAFE
in elastic range and the influence of the stress
+20%
14
gradient and of the mechanical properties of the
PREDICTED VALUE
12
+30%
material, seems to be able to provide a considerable
-20%
10
reduction of the scatters together with sufficiently
conservative results [5].
8 -30% Taking into account the complexity of the phe-
6 nomenon and the advantages due to the use of an
equation in closed form, the approximation ob-
4
tained by means of this method can be considered
SAFE
2 satisfactory. Regarding to the obtained degree of
0
approximation, it is useful to point out that some
0 2 4 6 8 10 12 14 16
researchers [10] have performed a numerical simu-
EXPERIMENTAL VALUE
lation of the experimental tests of Lukey and
Adams [11] by using the finite element method and
Fig.9 - Comparison between predicted adopting nine node Lagrangian shell finite ele-
and experimental values of the
ments with 2x2 Gauss integration and a bounding
rotation capacity
surface model based on the linear flow theory, but
the obtained scatters between predicted and expe-
rimental values of the rotation capacity varied from -40% to +60%.
Finally, it is useful to point out that none of the available simplified procedures nor the
definition itself of the rotation capacity are able to take into account that, in the actual case of
members belonging to a real structure, the distance between the inflection point and the location
of the maximum bending moment changes during the process of loading.
5 - REFERENCES
[1] Commission of the European Communities: Eurocode 3: Design of Steel Structures, 1992.
[2] B. Kato: Rotation Capacity of Steel Members subject to Local Buckling, 9th World Conference on Earthquake
Engineering, Vol.IV, paper 6-2-3, August 2-9, Tokyo-Kyoto, 1988.
[3] B. Kato: Rotation Capacity of H-Section Members as Determined by Local Buckling, Journal of Construc-
tional Steel Research, pp.95-109, N.13, 1989.
[4] B. Kato: Deformation Capacity of Steel Structures, Journal of Constructional Steel Research, pp.33-94, N.17,
1990.
[5] V. Piluso: Inelastic behaviour of seismic-resistant steel frames, PhD Thesis, University of Naples, Italy,
February 1992 (in italian).
[6] M.G. Lay, T.V. Galambos: Inelastic Beams Under Moment Gradient, Journal of Structural Division, ASCE,
vol.93, February, 1967.
[7] M.G. Lay, T.V. Galambos: Inelastic Beams Under Uniform Moment, Journal of Structural Division, ASCE,
vol.91, December, 1965.
[8] A.R. Kemp: Interaction of Plastic Local and Lateral Buckling, Journal of Structural Engineering, ASCE,
vol.111, October, 1985.
[9] U. Kuhlmann: Definition of Flange Slenderness Limits on the Basis of Rotation Capacity Values, Journal of
Constructional Steel Research, pp.21-40, 1989.
[10] G. Greshick, D.W. White, W. McGuire, J.F. Abel: Toward the prediction of flexural ductility of wide-flange
beams for seismic design, Proceedings of Fourth U.S.National Conference on Earthquake Engineering, Palm
Springs, California, Vol.2, pp.107-115, May 20-24, 1990.
[11] A.F. Lukey, P.F. Adams: Rotation capacity of beams under moment gradient, Welding Research Council
Bulletin, N.142, pp.1-19, July 1969.