Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Critical Buckling Loads of Semi-Rigid Steel Frames

Lei Xu
Assistant Professor
Department of Civil Engineering, University of Waterloo
Waterloo, Ontario, Canada N2L 3G1

ABSTRACT

The problem of determining critical elastic buckling loads of steel frames under
variable load patterns is discussed in this paper. In light of concepts of storey-
based buckling and end-fixity factors to characterize the lateral sway buckling of
frames and the semi-rigid behaviour of beam-to-column connections,
respectively, the problem for determining the critical buckling loads is presented
as a minimization and maximization problem with subject to stability constraints
and is solved by a linear programming method.

INTRODUCTION

The assessment of column strength in framed structures are commonly carried out by
evaluating the effective-length factor of columns in conjunction with the alignment charts in
the current ASD or LRFD Specifications (AISC, 1, 2). The concept of the effective length
factor is considered to be an essential part of many analysis procedures and is valid for
ideal structures. However, several assumptions on the buckling modelling of the frame were
made in developing the alignment charts. When the assumptions are violated, the use of
alignment charts results in erroneous effective-length factors.

The concept of storey-based buckling for unbraced frames is established based on the fact that
an individual column cannot fail by lateral sway buckling without all of other columns in the
same storey also buckling in the same sway mode (Yura, 3). Therefore, lateral sway buckling of
unbraced frames is a storey phenomenon. Various procedures of evaluating the stability of the
frame based on this concept have been proposed (LeMessurier 4, Lui 5, Aritizabal-Ochoa, 6,
Chong-Siat-Moy, 7).

Beam-to-column connections play an important role in the resistance of structural frames to


loads and maintain the stability of the structure. In the current practice of stability analysis of
steel-framed building structures, the actual behaviour of connections is generally simplified to
the two idealized extremes of either fully-rigid behaviour or ideally-pinned behaviour. Although
the adoption of such idealized joint behaviour simplifies the stability analysis, it by no means

117
represents the actual behaviour of the structure. Therefore, the predicted response of the idealized
structure may be quite unrealistic compared to that of the actual structure. This is because most
connections used in current practice actually exhibit semi-rigid deformation behaviour that can
contribute substantially to the stability of the structure as well as to the distribution of member force.
Numerous experimental investigations on connection behaviour have clearly demonstrated that
a pinned connection possesses a certain amount of rotational stiffness, while a rigid connection
possesses some degree of flexibility. Neglecting realistic connection behaviour may lead to
unrealistic predictions of the response and strength of structures, and therefore, to
approximations in design. Therefore, beam-to-column connections should be treated as semi-
rigid connections in the stability analysis of steel frames.

Over the years, considerable efforts have been made to incorporate the connection behaviour
into the determination of stability strength of semi-rigid frames (Bjorhovde, 8; Chen and Lui, 9;
Xu, 10; Kishi et al, 11; Christopher and Bjorhovde, 12). However, all of the foregoing studies
that have been carried out so far are under the assumption that the frame is subjected to
proportional loading. The variable loading case, which has taken into consideration the volatility
of loads and is more closely related to actual situations in practice, is left unsolved due to the
complexity of the problem. The main objective of this paper is to investigate the elastic in-plane
buckling characteristics of semi-rigid unbraced frames under variable loading.

END-FIXITY FACTORS OF SEMI-RIGID MEMBER

To incorporate semi-rigid connection behaviour into frame stability analysis, the effects of
connection flexibility are modelled by attaching rotational springs at the two ends of a beam-
column member as shown in Figure 1. A so-called end-fixity factor r defines the stiffness of the
connection to the attached beam-column member (Monforton and Wu, 13),
αj 1
rj = = ; ( j =1, 2) (1)
φj 3EI
1+
RjL

where R1 and R2 are the rotational stiffnesses of the spring connections at ends 1 and 2; L and I
are the length and cross-section moment of inertia of the beam-column member, respectively;
and E is Young’s modulus.
V1 V2
M2
N1 R1 R2
@ @
M1 1 E, I 2 N2
L

Fig. 1. Semi-rigid beam-column member

For pinned connections, the rotational stiffness of the connection is idealized as zero and thus
the value of the end-fixity factor is zero (rj = 0). For rigid connections, the end-fixity factor has a
value of one (rj = 1), because the connection rotational stiffness is taken to be infinite. A semi-
rigid connection has an end-fixity factor between zero and one (0 < rj < 1). By using the end-
fixity factor, different member-end restraint conditions, such as rigid-pinned, rigid-semi-rigid, and
pinned-semi-rigid, are readily modelled simply by setting the end-fixity factors at the two ends of
the member to appropriate values. Therefore, the proposed analysis method is comprehensive

118
regardless of member end-rotational conditions and can be applied to the analysis of frames
with any combination of pinned, rigid and semi-rigid connections.
The end-fixity factor also simplifies the analysis procedure for semi-rigid framed structures. The
formulations of stiffness matrices for both first and second-order analysis, member end-
reactions, span deflections, and effective-length factors of beam-columns can all be expressed
in terms of the end-fixity factors (Xu, 14). The end-fixity factor has further value in design
because it provides a physical interpretation of the extent of rigidity available in a connection. It
also provides designers with a convenient way to compare the structural responses of a
member with semi-rigid connections to those of one with rigid or pinned connections.

By Eq. (1), the relationship between the end-fixity factor and the connection stiffness is
nonlinear, as shown in Figure 2. It is also clear that the relationship between the connection
stiffness and the end-fixity factor is almost linear when the connection is relatively flexible with a
value of the end-fixity factor between 0.0 and 0.5. However, as the end-fixity factor approaches
1.0, the required increase of connection stiffness becomes substantial. Therefore, designers
should keep in mind that with a certain percentage increase in the end-fixity factor, the
corresponding increment in connection stiffness may be quite different depending on whether
the connection is relatively flexible or rigid. Gerstle (15) reports that the stiffness ratio, RL/EI for
“rigid” connections, ranges from 10 to 50 in typical building design, which implies that the end-
fixity factor ranges from 0.77 to 0.94.
Connection stiffness ratio (RL/EI)

60
55
50
45
40
35
30
25
20
15
10
5
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

End-fixity factor (r )

Fig. 2. Relationship between connection stiffness and end-fixity factor

The end-fixity factor is a better indicator of how connections affect the structural behaviour than
the connection stiffness. The latter has little direct meaning in analysis. Figure 3 illustrates the
variation of mid-span moments, end-reaction moments, and mid-span deflection of the beam
with respect to the connection stiffness ratio, RL/EI, for a uniformly loaded semi-rigid beam with
identical connection stiffnesses at both ends. Figure 4 shows the same information in terms of
the end-fixity factor, r. Figure 3 shows that the relationships of the moments and the deflection
to connection stiffness ratio of the beam are highly nonlinear when the ratio is between 0 and
20, and almost linear when the ratio is between 20 and 50. On the other hand, as shown in
Figure 4, the relationships of the moments and the deflection to the end-fixity factor, for its full
range of zero to one, are approximately linear. Thus, it is advantageous to characterize semi-
rigid behaviour using the end-fixity factor rather than the connection stiffness.

119
Deflection (5WL4/384EI)
1.6
end m om ent 1
1.4
mid-span deflection
Moment (WL2/12)

1.2 m id-s pan m om ent 0.8


1.0
0.6
0.8
0.6 0.4
0.4
0.2
0.2
0.0 0
0 20 40 60 0 20 40 60
Ratio RL/EI Ratio RL/EI

Fig. 3. Moment and deflection of a uniformly loaded beam vs. ratio RL/EI

1.6 Deflection (5WL4/384EI)


end moment
1.4 1.0 mid−span deflection
1.2 mid-span moment
Moment (WL2/12)

0.8
1
0.8 0.6
0.6 0.4
0.4
0.2
0.2
0 0.0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
End-fixity factor r End-fixity factor r

Fig. 4. Moments and Deflection of a uniformly loaded beam vs. end-fixity factor

It is observed from Figure 2 that when the connection stiffness is large, very significant changes
in stiffness produce only very small changes in the end-fixity factor. Consequently, from Figure
4, such change has a negligible influence on both moments and deflection of the beam.
Conversely, from Figure 2 with low values of connection stiffness, small increases in the
stiffness result in appreciable increases in the end-fixity factor. Therefore, as Figure 4 shows
there is a considerable effect on the bending moments and the deflection. Thus, in practice,
when a real pinned connection has some stiffness, a considerable restraining moment may
develop to the benefit of the structure. At the other extreme, attempting to achieve further
increase in connection stiffness beyond that of a nearly rigid connection is inefficient and costly
because it involves only a small change in the end-fixity factor. Consequently, it has little effect
on the response of structure.

120
LATERAL STIFFNESS OF A SEMI-RIGID COLUMN

Shown in Figure 5 is an axially loaded semi-rigid column in an unbraced frame, in which EIi /Li is
the flexural stiffness of the column and Pi is the column axial load. The end-fixity factors, rl and
ru, define the rotational restraints provided by the connected beams at the lower and upper
joints, respectively. Let
2
Pi Li
φi = = π Pi / Pe (2)
EIi

The lateral stiffness of the column can be expressed as (Xu et al, 16)
12EIi
Si = βi (φi ,rl ,ru ) 3
(3)
Li

where βi(φi, rl, ru) is the modification factor of the lateral stiffness that accounts for the effects of
axial force and column end rotational restraints.
3
φi a1φi cos φi + a2 sin φi
βi (φi ,rl ,ru ) = (4)
12 18rl ru − a3 cos φi + a4φi sin φi
where
a1 = 3[rl (1− ru ) + ru (1− rl )] (5a)

a2 = 9rl ru −(1− rl )(1− ru )φ2 (5b)

a3 =18rl ru + [3rl (1− ru ) + 3ru (1− rl )]φ2 (5c)

a 4 = −9r l ru + 3r l (1− ru ) + 3ru (1− r l ) + (1− ru )(1− r l )φ 2 (5d)

θu
P P
@

ru u
@

Mu

EI
L

θl

l
@

rl

P Ml
Figure 5. Lateral buckling of axially loaded semi-rigid column

121
It is difficult to evaluate the column critical buckling load due to the transcendental relationship
between βi and φi in Eq. (4), especially in a multicolumn-unbraced frame. Thus, a simpler
approximation of Eq. (4) can be obtained by its first-order approximation of Taylor series
expansion as
βi (φi ,rl ,ru ) = βi 0 (rl ,ru ) − βi 1(rl ,ru )φi2 (6)
where
(rl + ru + rl ru )
βi 0 (rl ,ru ) = lim βi (φi ,rl ,ru ) = (7)
φ→0 4 − rl ru
2
40 + 8( rl + ru2 ) + rl ru (rl + ru + 3rl ru − 34)
βi 1(rl ,ru ) = (8)
30( 4 − rl ru )2
The Taylor series expansion, Eq. (6), provides a satisfactory approximation of Eq. (4) in the
evaluation of the critical buckling load of the column (Xu et al, 16).

STOREY-BASED BUCKLING OF UNBRACED SEMI-RIGID FRAMES

Unlike the alignment chart method, which ignores the fact that columns in a storey of the frame
will restrain each other in resisting buckling, the interaction among the columns due to the fact
that stronger columns brace the weaker columns in the sidesway buckling is taken into account
in storey-based buckling. The condition for the multicolumn storey-based buckling in a lateral
sway mode is that the total lateral stiffness of the storey vanishes. For the proportional loading
case, the stability equation becomes
n n
 EI i Pi 
∑i =1
Si =12 ∑ L
i =1
3
i
β0 i −
Li
β1i λ cr  = 0

(9)

where n is the number of columns in a storey, Pi is the axial force due to the specified load of
column i, and λ is the critical load multiplier. Because the load pattern is predefined in the case
of proportional loading; therefore, only one critical load multiplier, λ, is need to be determined as
n
EIi
∑L
i =1
3
β0 i
λ= n
i
(10)
Psi
∑ i =1 Li
β1i

In the case of variable loading, the conventional assumption of proportional loading is


abandoned. Therefore, different load patterns cause the frame to buckle at different levels of
critical loads. Because the magnitude of each individual load can vary independently in order to
capture the worst load case in variable loading; therefore, there are n critical load multipliers, λi
(i = 1, 2, …, n), which need to be determined. Thus, the stability equation for variable loading
becomes
n
 EI i Pi 
∑ L
i =1
3
i
β0i −
Li
β1i λ i  = 0

(11)

122
CRITICAL BUCKLING LOADS OF FRAMES UNDER VARIABLE LOADING

Eq. (11) involves n variables; therefore, there is no unique solution for critical loads. Among
those critical loads, the one associated with the minimum and the maximum magnitudes of total
loads define the lower and upper bounds of the critical loads, respectively. The load patterns
associated with the minimum and maximum critical loads are the most critical and most
favourable load patterns for the elastic buckling of the frame. The lower and upper bounds of
the buckling loads and their associated load patterns have clearly characterized lateral buckling
of the frame under extreme conditions, which is crucial to evaluate the buckling strength of the
frame. The problem for determining the upper and lower bounds of critical buckling loads can be
presented as a minimization and maximization problem,
n
Maximize
Minimize
: Z= ∑λ P
i =1
i i (12a)

n
 EIi Pi 
Subject to: ∑ L
i =1
3
i
β0 i −
Li
β1i λ i  = 0

(12b)

π2EIi
Pli ≤ λ i Pi ≤ Pui = 2
; (i = 1, 2, …n) (12c)
Li
where the objective function Z corresponds to either the upper or the lower elastic buckling load
of the frame, and Eq. (12b) defines the storey-based lateral buckling condition of the frame. Eq.
(12c) imposes a constraint on each individual load such that the magnitude of the load can not
exceed its upper-bound, Euler buckling load, and should be greater than its lower-bound due to
the specified dead load. The problem stated in Eqs. (12) is a linear programming problem and
can be solved by the simplex method for determining either the maximum or the minimum
critical buckling loads and associated load patterns.

EXAMPLE

Shown in Fig. 6 is a simple semi-rigid unbraced steel frame. The moments of inertia for columns
1 and 2 are I1 = 45.2 × 106 mm4 and I2 = 104 × 106 mm4, respectively. The moment of inertia of
the beam is I3 = 617 × 106 mm4. The end-fixity factors for column bases are r1 and r2 while the
end-fixity factors associated with the left- and right-ends of the beams are r3 and r4, respectively.
Young’s Modulus of steel is taken as E = 200,000 MPa.

P1 kN P2 kN

r3 3 r4
@ @
W10 × 45
W8 × 31

W21 × 68
4877

1 2
r1 r2
@ @

7315

Figure 6. Semi-rigid unbraced steel frame

123
First, considering the cases that column bases are pinned with r1 = r2 = 0 and r1 = r2 = 0.1,
respectively, let the end-fixity factors associated with the beam (r3 = r4) vary from zero to one.
The corresponding upper and lower bounds of the critical buckling loads under variable loading,
Pmax and Pmin, are evaluated from Eqs. (12) and illustrated in Figure 7. The buckling loads
obtained from Eq. (10) under proportional loading with P1 = P2 are also plotted in Figure 7. The
maximum differences between the upper and lower bounds of the critical buckling loads for the
two cases are 4.55% and 4.29%, respectively. However, the corresponding maximum
differences between the buckling load under proportional loading and the minimum critical
buckling load under variable loading for the two cases are only 1.47% and 1.57%, respectively.

It is observed from Figure 7 that for flexible beam-to-column connections (r ≤ 0.3), an increase
in the end-fixity factor of the beam would result in considerable increase in the critical buckling
loads of the frame. However, for nearly rigid connections (r ≥ 0.7), further increase in the end-
fixity factor has trivial effect on the critical buckling loads. Considering the relationship between
the end-fixity factor and connection stiffness shown in Fig. 2, it is concluded that it is important
and economical to consider the semi-rigid behaviour of connections in the design of unbraced
steel frames with flexible beam-to-column connections.

Upper curve: Variable Loading Pmax


4000
Mid curve: Proportional P
Lower curve: Variable Loading Pmin
r1 = r2 = 0.1
Critical Loads of Frame (kN)

3000

r1 = r2 = 0
2000

Upper curve: Variable Loading Pmax


Mid curve: Proportional P
1000 Lower curve: Variable Loading Pmin

0
0 0.2 0.4 0.6 0.8 1
End-Fixity Factor of Beam (r3= r4)
Figure 7. Critical loads of frame vs. end-fixity factor of beam

Second, considering that column 1 has rigid connections at both the lower and upper ends while
column 2 is a lean-on column with pinned connections at its ends, the end-fixity associated with
columns and the beam are r1 = 1.0, r2 = 0, r3 = 1.0 and r4 = 0. Based on Eqs. (12), the problem
of solving the upper and lower bounds of critical loads under variable loading can be expressed
as
Maximize / Minimize: P1 + P2
Subject to: 0.239P1 + 0.205P2 + 845.645 = 0

124
0 ≤ P1 ≤ 3751 kN
0 ≤ P2 ≤ 8632 kN

Because only two variables are involved in the above problem, graphical solutions are available
as that shown in Figure 8. The magnitudes and the maximum differences of the upper and lower
bounds of critical loads are listed in Table 1. The maximum difference between the bounds of
critical loads is substantial with a value of 16.59%. However, considering the semi-rigid
behaviour of connections, with r = 0.8 and 0.2 are adopted for rigid and pinned connections,
respectively, the maximum difference of the critical loads is reduced to 5.5%.

Connection End-fixity Maximum Critical Minimum Critical  


Factor Buckling Load (kN) Buckling Load (kN) ∑
 Pi −
 max

min
Pi 


∑P
min
i

Pinned Rigid P1 P2 ∑P
max
i P1 P2 ∑P
min
i (%)

r=0 r = 1.0 0 4125 4125 3538 0 3538 16.59


r = 0.2 r = 0.8 0 5429 5429 3751 1395 5146 5.50

Table 1: Critical Loads of Simple Frame with Lean-on Column

P2 kN
10000 Pe2 = π2EI2/L22
= 8632 kN

8000
Pe1 = π2EI1/L12
6000 = 3751 kN
max. Z =Pmax = 4125 0.8
4000 r = 0.2, 0.8
0.228P1+0.212P2+1151=0
2000 0.239P1+0.205P2+845.6=0
r = 0, 1.0
Min. Z = Pmin = 3538
0
0 2000 4000 6000 P1 kN

Figure 8. Graphical solutions of critical loads

125
CONCLUSION

Based on the concept of storey-based buckling, the stability of unbraced semi-rigid steel frames
under variable loading is investigated. The maximum and minimum critical buckling loads under
variable loading characterize the buckling strengths of the frame under extreme conditions;
therefore, it is of importance for frame design. The benefit of considering semi-rigid behaviour of
beam-to-column connections in the stability analysis of unbraced steel frames, particularly for
those with flexible connections, is demonstrated. The fact that the maximum difference between
the maximum and minimum critical buckling loads can be substantial for the frames with lean-on
columns may suggest that traditional proportional loading approach may not be adequate to
assess the load capacity for such frames. Therefore, the variable loading approach is
recommended for the frames with lean-on columns.

ACKNOWLEDGMENT

This work was funded by a grant from the Natural Science and Engineering Research Council of
Canada.

NOTATIONS

E = Young’s modulus
I = moment of inertia of member
L = member length
M = flexural moment
Pi = axial force
Pe = Euler buckling load
r = end-fixity factor
rl, ru = end-fixity factor of lower and upper end of column
Si = lateral stiffness of semi-rigid column
Rj = connection stiffness
Z = objective function
β = modification factor of column lateral stiffness
λ = critical load multiplier

REFERENCES

1. AISC, (1989). Manual of Steel Construction-Allowable Stress Design. American Institute of


Steel Construction, 9th Ed., Chicago.
2. AISC, (1994). Manual of Steel Construction-Load and Resistance Factor Design (LRFD)
(1994). American Institute of Steel Construction, 2st Ed., Chicago, III.
3. Yura, J.A. (1971). “The Effective Length of Column in Unbraced Frame.” Engineering
Journal, AISC, 2, 37-42.
4. LeMessurier, W.J. (1977). “A Practical Method of Second Order Analysis, Part 2-Rigid
Frame.” Engineering Journal, AISC, 2, 49-67.
5. Lui, E.M. (1992), “A Novel Approach for K Factor Determination”, AISC, Engineering
Journal, 4, 150-159.
6. Aristizabal-Ochoa, J.D. (1997). “Storey Stability of Braced, Partially Braced, and Unbraced
Frames: Classical Approach.” Journal of Structural Engineering, ASCE, 123(6), 799-806.
7. Cheong-Siat-Moy F. (1999). “An Improved K-Factor Formula.” Journal of Structural
Engineering, ASCE, 125(2), 169-174.
8. Bjorhovde, R. (1984). “Effect of End Restraint on Column Strength-Practical Applications.”

126
Engineering Journal, AISC, 1, 1-13.
9. Chen, W.F. and Lui, E.M. (1985). “Stability Design Criteria for steel members and frames in
the United States.” Journal of Construction of Steel Research, 5, 31-74.
10. Xu, L. (1994). “Optimal Design of Steel Frameworks with Semi-Rigid Connections.” Ph.D.
Department of Civil Engineering, University of Waterloo, Canada.
11. Kishi, N., Chen, W.F. and Goto, Y. (1997). “Effective Length Factor of Columns in Semi-
Rigid and Unbraced Frames”, Journal of Structural Engineering, ASCE, 123(3), 313-320.
12. Christopher, J.E. and Bjorhovde, R. (1999). “Semi-Rigid Frame Design Methods for
Practicing Engineers.” Engineering Journal, AISC, 36(1), 12-28.
13. Monforton, G. R. and Wu, T. S. (1963). “Matrix Analysis of Semi-Rigidly Connected
Frames.” Journal of Structural Division, ASCE, 89, ST6, 13-42
14. Xu, L. (2000). “Second-Order Analysis for Semi-rigid Steel Frame Design”, to be published
in Canadian Journal of Civil Engineering.
15. Gerstle, K.H. (1988). “Effect of Connection on Frames.” Journal of Construction Steel
Research, 10, 241-267.
16. Xu, L., Liu, Y. and Chen, J. (2000). “Stability of Unbraced Frames under Non-Proportional
Loading.” Submitted for publication in Structural Engineering and Mechanics.

127

You might also like