Sie sind auf Seite 1von 18

Seismic performance of

moment resisting frames


with flexible joints
A. Osman, A. Ghobarah and R. M. Korol
Department of Civil Engineering and Engineering Mechanics, McMaster University,
Hamilton, Ontario, Canada, L8S 4L7
(Received April 1992)

An analytical study was conducted to investigate the effect of joint


flexibility, in particular extended end-plate joints, on the response of
moment resisting frames. Eight storey frames having extended endplate joints and with joints detailed according to different design
criteria, were analysed for different earthquake records using a stepby-ste p time history analysis. Realistic models developed from an
extensive experimental research program were used to simulate the
joint response. An evaluation of the frames' performance in terms of
lateral deflection, inter-storey shear, drift, and the inelastic rotational
demands imposed upon the beams, columns and joints was carried
out. It is concluded that joint behaviour is an important consideration
in a frame's response. Also, its performance can be much improved
by properly designing and detailing the joints.

Keywords: seismic,

steel,

frames,

joints,

end-plate,

design,

behaviour, analysis

Moment resisting frames (MRFs) are widely used as


lateral support systems for low-rise to medium-rise steel
buildings, particularly in seismic regions. Their versatility, unobstructed spaces between columns and ease
of erection mean they are favoured by architects and
structural engineers alike. However, in the analysis and
design of these frames, simplifications are often made
regarding the behaviour of beam-to-column joints.
Joints are usually idealized either as ideally pinned or as
fully rigid. Although these idealizations simplify the
analysis and the design procedure, the predicted
response may not be realistic as all joints in reality
behave in a semi-rigid manner.
Experimental research on the cyclic response of
joints, together with the availability of analysis programs, provide tools for predicting the real response of
MRFs. Also, they provide a means for evaluating the
influence of the design procedure and detailing practice
on likely frame response.
In general, the joints' semi-rigid behaviour arises
from two main sources: connection flexibility and panel
zone shear deformation. The connection is defined as the
medium that connects the beam to the column flanges,
while the column web and the column flanges adjacent
to the connection are referred to as the panel zone. In
extended end-plate joints, the connection refers to the
end-plate, the bolts, and the column flange as shown in
Figure 1. The column web and the column flange within

the area of the beam continuity plates are referred to as


the panel zone.
In the past decade, a large number of studies have
been conducted to investigate the influence of joint flexibility on the frame's response. Several researchers have
focused on examining the effect of connection semi-rigid
action on frame behaviour 1-3, while others have
investigated the role of panel zone deformation on frame
performance 45. However, no attempt has been made to
include the flexibilities of both elements, i.e. the connection and the panel zone, in the analysis of MRFs,
which is expected to be the case with bolted joints. The
analysis of each source of joint flexibility separately was
favoured by researchers for several reasons. Firstly,
considering the panel zone deformation and connection
flexibility in the analysis is expected to complicate the
problem since computationally additional degrees of
freedom have to be introduced to the structural model.
Theoretically, at least, the effect of the connection media
on the response of panel zones cannot be ignored.
Secondly, this task necessitates the development of
accurate models that are capable of simulating the
behaviour of both the connection and the panel zone,
which in turn requires conducting elaborate experimental investigations.
Recently, a research programme was conducted to
investigate the effect of the flexibility inherent in
extended end-plate joints on the seismic performance of

0141-0296/93/020117-18
1 9 9 3 B u t t e r w o r t h - H e i n e m a n n Ltd

Eng. Struct. 1993, Vol. 15, No 2

117

Seismic performance of moment resisting frames: A. Osman et al.

Design of frames
Selection of members
Column flange

Panel zone

Figure 1 Typical extended end-plate joint

moment resisting frames. In the experimental phase of


this programme, several full scale beam-to-column
subassemblages were tested under controlled cyclic
displacement 6-8. The objectives were to examine the
behaviour of such a joint type under cyclic loading conditions, evaluate the seismic performance of the joint
and its individual components, and propose design
criteria for detailing and proportioning the joint. Meanwhile, the analytical phase of the study involved the
development of a model capable of accurately predicting
the behaviour of such a joint 9~.
In this study, these models will be incorporated in a
nonlinear dynamic analysis program to examine the
effect of joint behaviour on the static and seismic
response of MRFs and to evaluate the adequacy of proposed design criteria. The study involves the analysis of
a part of a perimeter framing system for an eight-storey
building. Unlike conventional approaches, the analysis
was carried out by permitting both the connection and
the panel zone to deform. The connections were
designed and detailed according to the design criterion
proposed by Korol et al. 7. Several design criteria were
proposed for detailing the panel zones. The effects of
these criteria on the seismic response of MRFs in both
the serviceability and ultimate limit states were
investigated. The frames' performances in the ultimate
state were evaluated in terms of storey shear, storey
drift, plastic hinging and inelastic rotation demands
experienced by the columns, beams, and joints. The
study includes three main sections. The first part briefly
describes the procedure followed for designing and
detailing the frames and their components. The second
part deals with the approach used to model the joints'
cyclic behaviour. The third section presents some of the
results for the static and the seismic response of such
frames and an evaluation of their performance.

118

Eng. Struct. 1993, Vol. 15, No 2

The structure considered for analysis in this study is an


eight-storey all steel, three-bay by four-bay rectangular
office building as shown in Figure 2, assumed to be
located at Vancouver, Canada. The structure consists of
noncomposite floors, leaning interior columns, and a
perimeter framing system with MRF in the E-W direction and braced frames in the N-S direction. Since the
interior columns were leaning columns designed to carry
only gravity loads, all the lateral forces in the E-W
direction have to be resisted by the two exterior rigid
frames. The analysed frame is the MRF along axis A.
The building's structural members were sized to support
both gravity and lateral loads in accordance with the
minimum requirements provided by the National
Building Code of Canada (NBCC) ~. The roof level
members were designed to sustain vertical loads due to
self weight, and a uniformly distributed snow load of
1.76 kPa (Vancouver, Canada). The mechanical floor
members were sized for 2 kPa permanent equipment
load and 3.6 kPa live load (LL) in addition to the floor
dead load. All other floors were designed for 2.4 kPa
(LL) in addition to their dead load and 1 kPa to account
for the weight of partitions and ceiling fixtures. The
loads of the exterior windows and the core walls were
taken as 1.65 kPa and 2 kPa over their vertical surfaces,
respectively.
Two types of lateral loads were considered in the
design of the frame, wind and earthquake. For wind
loads, a 0.44 kPa average overall wind pressure was
determined corresponding to the reference data
applicable for Vancouver. For earthquake loads, the
base shear was calculated according to the formula given
by the NBCC ~. The frame was assumed to be a ductile
moment resisting frame with a modification factor, R,
equal to 4. The structure is located in a region of high
seismicity with a design peak horizontal ground
acceleration equal to 0.23 g for 10% probability of
exceedance in 50 years. The base shear was distributed
through the height of the frames in accordance with the
procedure described in the NBCC. The calculated unfactored base shears for the frame due to wind and earthquake were computed as 878kN and 1267 kN,
respectively.
The analysis and design of the frame was carried out
in an iterative process assuming infinitely rigid joints
and based on centre line-to-centre line dimensions, an
approach likely to be followed in real practice. It
included lateral sway effects (P-6) and as is customary,
the column sections were changed at every two floors
and the girders of any particular floor assigned to have
the same size.
Initially, the members were selected to satisfy strength
requirements. Then, the design was checked for drift
limitations. The NBCC recommends that the interstorey
drift should not exceed either h/500 or O.02h/R, where
h is the storey height, when the building is subjected to
specified wind load or earthquake load, respectively.
Analysis shows that the drift requirements govern the
design. Consequently, the beam sections were increased
to satisfy the drift limitations.
After refining the design, the frames were checked for
the strong column-weak beam philosophy using the for-

Seismic performance of moment resisting frames: A. Osman et al.


Roof

Mech
floor

W310x60

W310x60

W610x125
:r

W610x125 X

W610x12S "~
o

g.

LI.

ol

Office
floors ' 4

W610x12_5

W610xt25 ~=

W610x125 ~=

W610x195

W610x195

w61ox19s

W610x195

W610x195 ~=

W610x195 ~=

0~

w61ox2,,

W610x241

W610x2q1

W610x2ql

~=

W610x2ql ~:

W610x241

W610x241

=~

WG10x241

U.

Ground
floor

U.

'
L.

_L

9000mm-L

U.

13 raci ng

9000mm ,

9000mm

. 9000mm

a
Figure 2

II

I
9000mm

I-4

xi..- L

9000mm

==

=~

W610x241

U,.

8 Storey office building. (a), deviation; (b), plan

mula given by the uniform building code (UBC) j2.


Although this check is not required by the code, it is
believed that frames proportioned according to this concept possess a superior performance. Also; the frame
members were checked for being class 1 sections,
especially the beams. This requirement is in accordance
with the seismic steel specification 13. The resulting
beam and column sizes for the design are shown in
Figure 2. The member sizes are shown using the Canadian metric designation, where the first letter indicates
the member shape ('W' indicates rolled wide flange section and 'WWF' indicates welded wide flange section)
followed by nominal depth of the member in millimetres

times the mass of the member per unit length in


kilograms per metre.

Design of connections
In detailing the connections, the design criterion proposed by Korol et al. 7 was adopted. In this criterion,
the connections are detailed to have adequate resistance
to transmit a moment of 1.25 Mb, where Mb is the
beam plastic moment. Consequently, the connectiOn
rotation can be limited and the inelastic action in the connection can be avoided. Details of the connections for
the analysed frame are listed in Table 1.

Table 1 End-plate design for 8-storey frame, (All bolts are A 4 9 0 )

Floor
level

Dimensions
length width
(mm)

Thickness
(mm)

Bolt 9
diameter
(in)

Column flange
stiffened

7/8
1 1/8
1 1/8
1 1/2
1 1/2
1 1/2
1 1/2
1 1/2

Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes

7/8
1 1/8
1 1/8
1 1/2
1 1/2
1 1/2
1 1/2
1 1/2

No
No
Yes
Yes
Yes
Yes
Yes
No

Exterior connections
8
7
6
5
4
3
2
1

480
822
822
822
822
822
822
970

x
x
x
x
x
x
x
x

230
260
260
380
380
380
380
400

25
40
40
45
45
50
50
45
Interior connections

8
7
6
5
4
3
2
1

480
822
822
822
822
822
822
970

x
x
x
x
x
x
x
x

23
260
260
380
380
380
380
400

25
4O
40
45
45
50
50
45

Eng. Struct. 1993, Vol. 15, No 2

119

Seismic performance of m o m e n t resisting frames: A. Osman et al.

while the applied shear force Vbp arising from Mh~ and

Design of panel zones


Current seismic steel design specifications, UBC and
CAN3-S16-M891213 provide special provisions for
detailing the panel zones in ductile MRFs. These provisions are formulated based on the conclusions drawn
from the experimental investigation conducted by
Krawinkler 14 on fully welded joints. No attempt was
made to verify the applicability of these provisions when
other connecting media rather than fully welded connections are used. Recently, the experimental investigation
conducted by Osman et al. 15 sho W ed that these provisions adopted by the codes tend to underestimate the
panel shear strength when extended end-plate joints are
employed. This underestimation results from ignoring
the participation of the end-plate in resisting the panel
zone deformation. Furthermore, it was concluded that
the performance of beam-to-column subassemblages can
be improved substantially by allowing both the beam and
the joint panel zone to jointly participate in inelastic
action. On the othdr hand, severe yielding of the panel
zones can adversely affect the performance of the frame.
Ideally, they should be detailed such that they participate
in the inelastic action without failure or without
experiencing excessive deformation. Consequently, an
important question was raised. How should the panel
zones in extended end-plate joints be detailed? To
answer this question, alternative criteria for detailing the
panel zones in such a joint type were proposed. The
following explains each of the proposed criteria.

"Mr,2 is given by

Mhl

Mb2

V~'r = (1-.95~/bi + 0~i95db: - /4,

(2)

In the above equation, ay is the yield stress of steel, t,,


is the thickness of the column web panel including the
thickness of doubler plates if any, dc is the column
depth, and Hc is the column shear which is equal to
r.Mbi/L,.. The flexural strength of the ith beam framing
into the joint is Mbi and L, is the storey height.
Equating V,. and Vbp results in a design criterion having
relatively strong panels with thick doubler plates. As
such, only limited inelastic action is expected to take
place in the joint panels.

Design 2
In this case, the joint panels are detailed to have shear
strength, Vm, equal to the overstrength shear force
delivered by the beams framing into them. Anticipating
that the beams' overstrength flange forces will be 25 %
higher than that delivered by their plastic moment
capacities, the panels thicknesses can be determined
according to the following equation
1.25V~,~, = Vp,

(3)

where the panel overstrength, augmented principally by


the column flanges and end-plates, is given by m

Design 1
In this design criterion, the joint panels are detailed to
yield under the same moment as the adjacent beam. This
can be accomplished by detailing the panel so that its
yield shear resistance, Vy, is equal to the shear force
delivered by the beams framing into the panel when they
develop their plastic moment capacity, Mb~ and Mb2.
Reference is made to Figure 3(a).
The panel yield resistance, V,, is given by

V, = 0.55o, d,t,,

(1)

2Ma
dt, i + db2

2Mjl

= 55a"d"t'" +

+ M~"i"d~ G,,t,w(O.95d,.) 1
6EI~,,,, "

(4)

Mr and My2 are the flexural strengths of the column


flanges and the end-plate combined thickness assuming

L;

//c

b
de
Figure 3

120

Internal forces at interior joint

Eng. Struct. 1993, Vol. 15, No 2

Seismic performance of moment resisting frames: A. Osman et al.


complete contact, E is the modulus of elasticity and G,,
is the strain hardening shear modulus of the column
web. M~i, is the lesser of Mp and Mrz, while I ~ , is the
associated moment of inertia.
As can be observed, the first term in equation (4)
represents the shear strength of the panel at yield. The
term in square brackets represents the additional
resistance provided through web strain hardening and by
column flange and end-plate bending. Figure 3(b) shows
the flange bending force system involved. This additional resistance represents approximately 25 % to 40%
of the panel shear strength, Vp,. Since the overstrength
factor for the framing beams is 25 %, it can be concluded
that in such a design criterion, the joint panels would be
expected to yield prior to or with the adjoining beams.
Compared with the design 1 criterion, this design is
expected to yield panels of smaller thickness.

Design 3
In this design, the panel thicknesses were reduced further. The panels were detailed to have a shear strength
equal to the shear force delivered by the beams framing
into the panels when they develop their plastic moment
capacity. Consequently

vp.~ = vbp

(5)

This design criterion is equivalent in concept to the


design criterion adopted by the code. It ensures that the
panels will yield and experience several strain reversals
prior to beam yielding. Thus, efficient participation
from the panels in the inelastic action is guaranteed.
Also, it represents a capacity design approach in which
the panel constitutes the weakest element in the frame.
These design criteria were then applied to the analysed
frames. In the design process, the column web thickness
requirements were checked according to each of the
previously discussed design criteria and if found
unsatisfied, doubler plates rounded to the closest
available plate thickness were used to stiffen the panel
zone. The resulting doubler plate thicknesses as computed for the frames' joints are listed in Table 2.
Table 2

Floor
level

Doubler plate thicknesses for MRFs

Design 1

Design 2

Design 3

Exterior columns
8
7
6
5
4
3
2
1

None
2 PL@
2 PL@
2 PL@
2 PL@
2 PL@
2 PL@
2 PL@

10
10
16
16
16
16
14

8
7
6
5
4
3
2
1

None
2 PL@
2 PL@
2 PL@
2 PL@
2 PL@
2 PL@
2 PL@

20
20
25
25
30
30
30

None
2 PL@ 6
2 PL@ 6
2 PL@ 10
2 PL@ 10
2 PL@ 12
2 PL@ 12
2 PL@ 8
Interior columns
None
2 PL@
2 PL@
2 PL@
2 PL@
2 PL@
2 PL@
2 PL@

10
10
22
22
25
25
25

None
1 PL@
1 PL@
2 PL@
2 PL@
2 PL@
2 PL@
1 PL@

6
6
6
6
6
6
8

None
2 PL@
2 PL@
2 PL@
2 PL@
2 PL@
2 PL@
2 PL@

10
10
16
16
20
20
18

Cases studied
In this study, several frames were analysed. The dimensions of the frames were kept identical and similar
beams and columns sections were employed. However,
their joints were detailed differently according to the
previously proposed design criteria. The cases studied
included:

Frame 1. In this frame, all the joints were assumed to


be infinitely rigid. In practice this represents the case
when fully welded connections are used to join beams to
columns and the panel zones are heavily stiffened. Consequently, inelastic action is expected to take place only
in beams and columns.
F~ame 2. In this frame only the connections were permitted to deform, while the panel zones were assumed
to be infinitely rigid. This represents the case when flexible end-plate connections are used to join the beams to
the columns and the panel zones are heavily stiffened.
Frame 3. In this case, both the connections and the
panel zones were permitted to deform. While identical
connections to those employed in frame 2 were used, the
panel zones were detailed in accordance with the first
panel zone proposed design criterion. Comparing the
response of this frame with the response of frame 2,
shows the effect of adopting design 1 for detailing the
panel zones on the MRFs response.
Frames 4 and 5. These frames were identical to frame
3 with the exception that their panels were detailed
according to design 2 and design 3 criteria, respectively.
Thus, frame 5 represented the frame with the weakest
panels in this study.

Modelling of the frames' joints


As discussed, a joint's flexibility arises from two
sources, the connection rotation and the panel zone
distortion. In extended end-plate joints, the connection
rotation results from the deformation of three components. These are the end-plate deformation in the tension region, Aee, deformation of the column flange,
At/, and cripphng or yield deformation, Acw, of the
column web in the compression zone for the case of a
column without beam flange continuity plates. The
various deformations are shown in Figure 4. Furthermore, the panel zone distortion results from pure shear
deformation of the column web accompanied by bending
of the end-plate and the column flanges as shown in
Figure 5. Examining these figures indicates that the
nature of the forces causing deformation in each component is different. In addition, each component possess
different strength and deformation capacity. Consequently, the use of a global model to simulate the joint
response is difficult. Therefore, it was decided to use
two different models, one to represent the connection
moment-rotation relationship and the other to represent
the panel zone shear deformation.

Connection model
To simulate the connection behaviour, the model
developed and verified by Osman et al. 9 was adopted.

Eng. Struct. 1993, Vol. 15, No 2

121

Seismic performance of moment resisting frames: A. Osman

et

al.
i

A. 1
~W

~bl~TEL;-

Figure 4

Storey drift due to connection fleibility

I
I

I!

Hcl2

-I
I

IU
6
P

Lbl2
Figure 5

122

-I-

Panel zone shear d e f o r m a t i o n

Eng. Struct. 1993, Vol. 15, No 2

dc

-I-

Lb'12

"l

Aep+ Acf

AC W

Z~ep+Acf

Seismic performance of moment resisting frames: A. Osman et al.

The connection initial stiffness, K~, can be evaluated


by calculating the deflections of the connection
individual components (Ap, Acy, and Acw) at a beam's
flange levels in terms of the applied beam flange force,
P. Then, by dividing these deflections by the beam
depth, the connection rotation can be determined. To
determine the deflection of the end-plate at the tension
region, Aep, the end-plate was idealized at a T-stub and
the modified slope deflection method developed by
Azizinamini et al. 16 was used to solve for the deflection
(Figure 7). For the column flange, a similar procedure
was used to calculate Ac: if it were stiffened. In the case
of unstiffened column flanges, the deflection at the tension side, A:, was evaluated by idealizing the column
flange as an infinitely long cantilever plate subjected to
equal point concentrated loads at bolt locations as shown
in Figure 8. In the compression region, the deflection
due to web yielding, A~,, was obtained by idealizing
the effective compression zone of the web as a rectangular plate of dimensions d~d', as shown in Figure
9. The dimensions d" is equal to d c - 2k and d~ is equal
to (t~: + 6k + 2tep + 2twf).
The plastic moment capacity of the connection, M e,
in this model was taken as the smallest moment required
to cause yielding to any individual component of the
connection. Yield line mechanisms developed by Packer
and Morris ~7 were used to evaluate the beam moment
required to yield either the end-plate or the column
flange. Also, the beam flange load required to yield the
column web as predicted by Hendrick and Murray ]8,
was used to evaluate the beam moment that would cause
yielding to the column web in the compression zone.
The post elastic stiffness, Ko, was taken in this model
as a percent of the initial stif(ness, Ki. Tests conducted
on full scale specimens show that the connections
experienced post-elastic stiffness in the range of 10% to

This model is simple to use, efficient computationally


and can approximate t h e key aspects of the connection
behaviour. According to this model, the momentrotation relationship for each connection can be approximated by a bilinear piece-line curve as shown in Figure
6. The curve is completely defined by three parameters:
(1) the connection initial stiffness, Ki; (2) the connection post-elastic stiffness, Kp; and (3) the connection
plastic moment, Mp~. These parameters can be determined analytically for any connection provided that its
geometrical and mechanical properties are known.
Moment

M
pc

ZMpc

~ J

Figure 6 Proposed connection behaviour response model

,*-14
b1

b2

T
(7

a
Figure 7 End-plate model. (a) T-stub model; (b),

0
~ep

idealization

Eng~ Struct. 1993, Vol. 15, No 2

123

Seismic performance of moment resisting frames: A. Osman et al.

P/4
4-

-4PI4

i=
iI
11
d PI4
II qIi
II
-4II
Ij PI4

1...~ PI2
f'-I~ PI2

II

tt
1I

It
il

||

II
I|

I!

2
Figure 8

Idealization of unstiffened column flange

20 % of the initial stiffness. The accuracy of the model


was verified by comparing its predictions with the
experimental results. Close agreement between the
model predictions and the experimental measurements,
for one of the specimens is shown in Figure 10.
Panel zone

model

Based on the experimental investigation, it was concluded that the panel zone response can best be approximated by a trilinear curve as shown in Figure 11. The
curve consists of three segments that represent the panel
elastic response, transition behaviour and strain hardening. Five parameters are required to define the complete
behaviour for each panel zone. The parameters are, (1)
the initial stiffness of the panel, Ke; (2) the panel
plastic moment, Mp,,; (3) the panel stiffness in the transition zone, Kr; (4) the panel zone strength, M,,,2; and

400

300

....=.:.~. -.:

2OO
oo '.:,'';/-;r::":::'-"
/
0

0.0000

er[menta] +ve rotation

............ ~xpde~/tm~nta! - v e rotation


=
0.0050

Rotation
Figure 10

I
0,0100
(rad)

0.0150

Experimental versus predicted moment rotation

relationship

Moment l

K - line

K - line

I|

|I

|
!
!

|
|
I

lit;
i

-~r

c
Figure 9

124

Compression region of column flange

Eng. S t r u c t .

1993,

Vol. 15, No 2

Ke

2tdpnl

Figure 1 1 Proposed model to simulate panel zone behaviour

(5) the panel zone strain hardening stiffness, Ks. These


parameters can be calculated analytically for any panel
as described by Osman and Ghobarah t. A summary of
the procedure can be found in the Appendix. The
accuracy of the model in predicting the panel response
can be illustrated by comparing the experimental

Seismic performance of moment resisting frames: A. Osman et al.


600

q00

200

'E
Z

E
0
:E
-200

-400

Experimental
Predicted
-600
-0.060

-0.040

-0.020

0.000

0.020

0.0q0

0.060

Shear s t r a i n ( r a d }

Figure 12 Comparison between experimental and predicted panel response

behaviour of the tested panels with the model predictions. As can be seen from Figure 12, good agreement
exists between the model prediction and experimental
results.
Following the design of the frames and the detailing
of their joints in accordance with the proposed criteria,
the parameters that define the behaviour of each component were analytically determined and incorporated as
input to the program.

Analysis of moment resisting frames

Computer modelling
In analysing the MRFs, the nonlinear dynamic program,
Drain-2D tg, was used. The available two-dimensional
beam-column element was used to represent the beams
and the columns. Plastic hinges were assumed to take
place only in concentrated points at the element ends.
For the beams, the yield surfaces were considered to be
affected only by bending moment. As such, beam yield
strength was limited to Zay, where Z is the plastic
modulus of the beam section and ay is the steel yield
stress." For the columns, the following axial forcebending moment interaction relationship given by the
CAN3-S16-1VI89 was adopted as limiting a column's
strength
C / + 0.85 Myx < 1.0 where Mf~ < 1.0

Cy

Mp

Mp

(6)

In inequality (6), C/and M/x are the factored axial


load and bending moment about the major axis due to
gravity and lateral forces, while Cy and Mp are the axial
yield compressive resistance, Ao,., and column plastic
moment resistance, respectively. For the joints, bilinear
and trilinear springs were used to model the connection
and the joint panel zone behaviour, respectively.
In the dynamic analysis, theframes were subjected to
gravity loads that included specified dead and live loads.
The live loads were subjected to tributary area reduction
as permitted by the NBCC. Masses of the tributary floor
area, wall units and glazing were all assumed to be
lumped to the column-beam joints and were associated
with the horizontal displacement of the floors only. In
addition, 25% of the snow load on the roof and 100%
of permanent equipment loading in the mechanical floor
were included. The analysis was carried out using the
clear spans for the beams and columns. A damping coefficient proportional to both the mass and the tangent
stiffness was used. A damping ratio of 3 % was assigned
to the first two modes.

Static and dynamic characteristics of MRFs


To examine the effect of joint flexibility on the static
behaviour of the MRFs in the serviceability state, the
frames were subjected to the code specified lateral earthquake loads. The resulting frame lateral displacements
are shown in Figure 13. As can be observed, frame 2,
with flexible connections, experienced a 9 % increase in
the roof lateral deflection compared to frame 1 with
infinitely rigid joints. On the other hand, frames 3, 4 and

Eng. Struct. 1993, Vol. 15, No 2

125

Seismic performance of moment resisting frames: A. Osman et al.


tion periods by at most 15 %. However, in most cases the
increase was even smaller.

Nonlinear dynamic analysis of MRFs

>

~,

Code limit
Frame 1
Frame 2
= Frame 3
; Frame 4

- ..........

=
v
.

50

Figure 13

~-

"

Frame

100
150
Lateral deflection (ram)

200

250

Static elastic lateral d e f l e c t i o n for d i f f e r e n t frames

5 experienced 19.3%, 23.3% and 28.4% increase in


their roofs lateral displacements, respectively, compared to frame 2. Since these frames, i.e. 3, 4 and 5,
have identical connections to frame 2, the increase in
their lateral deflection is attributed only to panel zone
flexibility.
In practice, designers normally use centre-line to
centre-line dimensions to compensate for the effect of
joint flexibility. However, based on previous results it
seems that neglecting the effect of joint flexibility in the
analysis may lead to underestimating the lateral deflection by up to 40% which cannot be compensated for by
using centre-line to centre-line dimensions.
With regard to the frames' dynamic characteristics,
Table 3 shows the first seven vibration periods for each
frame. As can be seen, the incorporation of joint flexibility in the analysis tends to increase the frames' vibra-

Table 3

Three different earthquake records were selected to


assess the seismic performance of the eight-storey
frames in the ultimate state. The records are, the N-S
component of the 1977 Bucharest earthquake with peak
horizontal ground acceleration of 0.205 g, the N-S component of the 1985 Mexico earthquake (station-SCT)
with peak ground acceleration of 0.17 g, and the E-W
component of the Near Coast of Honshu 1968 earthquake with peak ground acceleration of 0.2 g. The time
histories for these records are shown in Figure 14. In
selecting these records certain aspects were considered:
(1) all the records have a peak horizontal ground
acceleration less than 0.23 g which is the design peak
acceleration for the analysed frames: (2) their predominant periods are close to those of the frames in order to
produce severe inelastic action to the members and
joints. The 3% damping linear elastic pseudoacceleration spectra for the selected records with the
periods for the first three modes are shown in Figure 15
for the analysed frames. Such frames would be expected
to experience severe inelastic deformatior,, particularly
from the Mexican event.
The frames were then subjected to these records.
Figures 16 and 17 show the floor displacement
envelopes for the analysed frames. As can be observed,
the response of all frames was dominated by their first
modes which is expected for low and medium-rise
frames. Comparing the response of frame 1, with
infinitely rigid connections, with that of frame 2,
employing flexible connection (Figure 16), reveals that
the latter experienced an increase in its roof's lateral
deflection compared to frame 1. This increase in roof
lateral deflection ranged from as low as 3 % during the
Bucharest event up to as high as 15% for the Mexico
event. It is also important to note that frame 2
experienced larger deflections than frame 1 on one side
and lesser deflections in the opposite direction during
the Mexico and Honshu events. This is attributed to the
fact that once the structure experienced severe inelastic
action due to a strong pulse in one direction, it starts
vibrating after that around a new shifted datum, the socalled 'crawling effect'. With regard to frames 3, 4 and
5 with flexible joint panels, Figure 17 shows that they
all experienced higher lateral deflections compared to

V i b r a t i o n periods for MRFs


V i b r a t i o n periods (s)

Modes

Frame 1

Frame 2

Frame 3

Frame 4

Frame 5

1
2
3
4
5
6
7

2.170
0.823
0.489
0.359
0.267
0.196
0.155

2.279
0.850
0.512
0.375
0.283
0.210
0.168

2.412
0.893
0.532
0.385
0.289
0.214
0.169

2.431
0.906
0.536
0.388
0.289
0.214
0.170

2.490
0.927
0.544
0.391
0.291
0.215
0.170

126

Eng. Struct. 1993, Vol. 15, No 2

Seismic performance of moment resisting frames: A. Osman et

03f

al.

0.2

0.2
0o
to

0.1

o:

u
<
<~ -0.1
-0,1

_0.2 I

-0,2
2

8
10
Time (s)

12

14

16

18

10

20

30

120

lq0

q0
T i m e (s)

50

60

70

80

0.400

r,o

o.ooo
u
<

-o.qo0
,,.
T i m e (s)

100

Figure 14 Time histories. (a) Bucharest earthquake; (b), Mexico earthquake; (c), Honshu earthquake

Bucharest
Mexico
-

Honshu

Frame

Frame 2
Frame 3

Mode 3
(-

.2

u
u
to

$
o.

/\

Frame q
Frame 5

LJl

Mode 1

Mode 2

I!ii

i
I
I
!

'!!

I
I

I
I
I
I
\

iiii
01
0

I
3

1
Period

.,

I
q

]
5

(s)

Figure 15 Linear elastic spectra for selected events

Eng. S t r u c t .

1 9 9 3 , V o l . 15, N o 2

127

Seismic performance of moment resisting frames: A. Osman et al.

- Frame 1
......... Frame 2

,'
,'

,I

-500

500

1000 -1000

-500

Deflection (mm)

//

:
-1000

500

Deflection (ram}

1000 -1000

, ,
-500

Jl

0
500
Deflection (mm)

1000

Figure 16 Effect of connection flexibility or frames' lateral deflection. (a), Bucharest earthquake; (b), Honshu earthquake; (c), Mexico

earthquake

-~ 7
1";'6

1"7"6

3
Frame
...... Frame
~

-1000

-500
0
500
Deflection (mm)

1000 -1000

-500
0
500
Deflection (ram)

Frame

1000 -1000

-500

~12

500
Deflection (ram)

10100

Figure 17 Effect of panel zone design on frames' lateral deflection. (a), Bucharest earthquake; (b),Honshu earthquake; (c), Mexico earth-

quake

frame 2 with infinitely rigid joint panels. Again, these


deflections were higher in one direction and lower in the
other.
In addition to the frames' lateral deflections,
envelopes for the frames' storey shear and storey drift
were calculated. Figures 18 and 19 show these results
for the analysed frames due to the considered events.
Examining these results indicates that the more the
frame becomes flexible, the less it will attract seismic
forces. They also show that with the exception of the
Honshu event, the frames experienced severe storey
drifts that exceeded the 2% limit anticipated by the
code It for frames at the ultimate state. Such high storey
drifts can result in severe overall sway and instability
problems.

128

Eng. Struct. 1993, Vol. 15, No 2

The energy dissipation mechanisms and plastic hinging throughout the frames were examined. Figures
2 0 - 2 4 , show the plastic hinging patterns and the maximum plastic rotations experienced by each element in
the analysed frames due to the Bucharest earthquake.
Similar results were obtained for the other events.
Examination of these results shows that allowing the
panel zones to participate in dissipating earthquake
energy, significantly reduces the rotation demand
imposed on other frame components. Also,' it tends to
distribute the hysteretic energy more uniformly over the
frames height preventing any damage concentration.
With regard to individual components, Figures 25 and
26 show the ductility demands imposed on the beams

Seismic performance of moment resisting frames: A. Osman et el.

..........

IL

-4000

-20'00

I
0

2000

I .

Frame 2

,1[I

4000 -4000

,~1

-2000

Storey shear (kN}

~!

2000

4000 -4000

Storey shear (kN}

-2o'00

2000

4000

Storey shear (kN}

Figure 18 Storey shear envelopes for MRFs. (a), Bucharest; (b), Mexico; (c), Honshu

ill l

ill

I~:

Ili

Itil,

;I

M,-2-iJ':'

I! ii! ,,

I~

"~

t,,l

~ 4

I1 ;

~ 4

~g
~ !

!ii

1 rJ,~
i

-0,040
a

! I

-0.020

0.000
storey drift

tJJ
0.020

i: !1!

,11 I,li
,
0.040 -0.040
-0.020

-0

l,' IJ i ..........
I~il

.....
I=== . . . . . . . .

li~ - -

lill;

,F,a o

i[

Frame ]

Frame
Frame

Ii

Frames

!i! I~
i~!1!

!',i~

iC

0.000
0.020
Storey d r i f t

0.040 -0,040
C

0.000

-0.020

0.020

0.040

Storey d r i f t

Figure 19 Storey drift envelopes for MRFs. (a), Bucharest; (b), Mexico; (c), Honshu

Eng. Struct.

1993,

Vol.

15,

No 2

129

Seismic performance of moment resisting frames: A. Osman et al.

EO3
m
z4"
d~

~ R Z,

{:~

~3
O

"t3
o
o

o
0

T,p
.,qp
Nog

~!

83

-p

o
o

'4

o
~3

~
0

E
~0q~

~,~

g,

,~

-F

0
(3
"(3
c
c~

E
o3

~D
o

o
I

~
I

iL

t~
0

L~
0

0
I

0
0

N ~ I P_

_i I

o3

0
D

0
0
o

l)

o
0

:}

h ~
0

0
0
0

~g
g~

m
Q.
~o

0
0

"i

r=

E
-=

1 1
~q~

0
0

0
I

o
o

130

Eng. Struct. 1 9 9 3 , Vol. 15, No 2

o3
.Q
I

~2

0
0

0
o

E
E~

Seismic performance of moment resisting frames: A. Osman et al.


-0.0059

0.0014

-0.0053

0.0008

0.0055

-0.0007

0.0007 " ' "


0 0 0 2 ~ 0.0008

-0.0027 0.0007

-0.0139

0"0139f
)0.0113

-0.0006~
-0~1,~0.0126

-0.0035

0.0011

-o.o . . ~

o.oo.~.~32

0.00-'~51 10.004q

-0.0005~

-0.0158 0.0121
;,;
0.0009

-0.0157

0.0023

0,00"~61 0.0032

0.00-~33

=0.0009

-0.006C

~0.0040

0.0040

-0.0042

-0.0042

0.0021~

0.0014 ~

O.O02q~

0.0041

0.0037

-o.oo6~

-o.oo6L

-0.0042

-0.0042

-0.0006

-0.0034

-0.0031

0.0002

-0.0009

-0.0032

-0.0030

-0.0054 0.0018

-0.0066

-0.0037

-0.0037

w,v

-0.0014

0.0011

-0.0035

-0.0012

0.0008

-0.0026

I-O.OOO6 -o.oo5L 2.oo14

0.0114

0.0129

0.0129

0.0047

0.0034

0.0117

Figure 22 Maximum beam joint plastic rotations for frame 3 due to Bucharest event (in radians). (a), beams and columns; (b), panel zones

0.0125/

-0.0007

-0.0065 0.0018
v|v

-0.0058

-0.0026 0.0005

o.o1421~ 7

-o.o02L 0.0003
w w
o.0.138f

-0.0064

-0.0077 0.0065

1.0075

0.0016

0.0021

-0.0042

).0043

0.0007:
o.oo-~'~

0.0004.

-0.00"~'~

-0.0032 ~0.0033
wlw

-0.0049

-0.0001 0.0002
';

-0.0009

-0.0018A 0.0007

-0.0027 0.0006
;

).0101

-0.0012

0.0053

0.0055

~0.0056

0.0043

0062

-0.0137

0131

-0.0047

0073

-O.OOq7

0073

-0.0039

-0.0038

0037

0.0040

-0.0038

-0.0037

0038

0.0066

-0.0046

-0.0045

.o067

0.0137,,,~ '1

-0.0077

0.0026

.0167

-0.0043

( )0.0074

0.0020"
o.oo-~

0.0023 -0.00132

-0.0017

4,

Figure 23

- o ~ 0.0030
00oo3"
ooo-~o2

-0. 0004 0.0003

0.0081

0.0019
w v

()

O. 0072

0.0033

-0.0048

(
(

0.0101

-0.001

.0138

-0.0047

()

()

0.0089
-IR

b
M a x i m u m j o i n t plastic r o t a t i o n s f o r f r a m e 4 due t o B u c h a r e s t e v e n t (in radians). (a), b e a m s and c o l u m n s ; (b), panel zones

Eng. Struct. 1 9 9 3 , V o l

15, No 2

131

Seismic performance of moment resisting frames: A. Osman et al.


-0.0056 0.0026

-0.0062 0.0023

-0.0006

v,v

-0.0012 0.0011
L0.0068

0.0050
,0.0121

-0.0028 0.0002
I

:-0.0107

50.oo62

L0.0117

-0.0066A | A 0.0055

0.0022 0.0092
~.
0.0007
II

-0.001G

0.0001
t

-0.0008

0.0002

-0.0006
A

0.0054

()

(J,-0.0123

0.0234

!-0.0119

( 5 '''
0.0117

O.

0063

0.0067

0070

-0.0027

0.0044

<

-0.0046

,4

-0.0046

-0.0104

-0. 0045

0.0043

00047

f.0043

(~0121

()~0.0123

(~.0209j
0.0124

"

C)
0.0102

0.0046

()

~.0045

( 0.0049

()
0.0110

0.0061

0.0052

i'0.0063
%

-0.0017

Figure 24

Maximum joint plastic rotations for frame 5 due to Bucharest event (in radians). (a), beams and columns; (b), panel zones

and the joint panel zones in the different designs due to


the three events. As can be noted, with the exception of
the first and second floor beams in frames 1 and 2 for
the Mexico event, all the ductility demands were within
the component supply capacities.

Bucharest earthquake
II Mexico earthquake
A Honshu earthquake
Design
. . . . . . . . . . Design ~
". . . . . . .
Design 3

, ~
I
(/ ~"i ~ _ _

Conclusions

The results of an analytical study that was concluded to


investigate the role of joint flexibility, particularly
extended end-plate joints, on the static and dynamic
behaviour of MRF have been presented. They highlight
the importance of incorporating joint flexibility into an
analysis. Also, they provide information regarding the
influence of the proposed design criteria on frame
response. On the basis of the described investigations,
the following conclusions may be drawn:

. - ' . ; f X,
El"&

Ifll

I / ,
0

,7 4

(1)

'%"-,,,.

(2)
OI

Figure 25

132

6
8
Panel ductility

10

Panel zone ductility demands

Eng. Struct. 1993, Vol. 15, No 2

12

14

Bolted connections, particularly extended endplate connections, contribute to a frame's lateral


deflection even when they are detailed to respond
elastically. This contribution ranged from a low of
3% to a high of 15% over the fully rigid connection case.
The design of the joint panel zone greatly affects
the dynamic response of MRFs. Frames with
relatively thin panel zones attract smaller seismic
forces and may experience higher lateral drifts
when compared to those with strong joint panels.

Seismic performance of moment resisting frames: A. Osman et al.


9

Frame 1

........

Frame 2

....

Frame 3
Frame q

Frame 5

IiS
q

o 41
,7

4
It.

0
0

a
Figure 26

Rotation d u c t i l i t y

q
(UO)

Rotation d u c t i l i t y

(~0)

Rotation d u c t i l i t y

(la o)

Beam rotational ductility demands. (a) Bucharest; (b), Mexico; (c), Honshu

(3)

The energy dissipation mechanisms and the plastic


hinging patterns for the frames can be substantially altered as a result of joint flexibility. Frames
with thin joint panel zones tend to confine most of
the inelastic action to such panels, while those
with thick panel zones impose high demands on
their beams and columns.
(4) Allowing the panel zone to participate in
dissipating the earthquake input energy seems to
improve the frame response in some aspects, since
it distributes the hysteretic energy more uniformly
over frame height thus avoiding damage concentration. In addition, such participation allows
relaxation of some of the restrictions imposed on
selecting the beam and column sections to delay
the onset of both local and lateral torsional buckling. As a consequence, such design may result in
more economical solutions. On the other hand,
flexible panels tend to increase a frame's lateral
deflections which may lead to overall instability.
(5) Based on the analytical investig-ation conducted,
detailing the panel zones in extended end-plate
joints according to design 2 criteria appears to be
the most appropriate approach. In this criterion,
the panels are detailed to yield prior to or with the
adjacent beams. This ensures the participation of
the panel zones in the inelastic action while controlling its distortion.

References
1 Goto, Y. and Chen, W. F. 'On the computer-based design analysis
of flexibly jointed frames', J. Const. Steel Res. 1987, 8, 203-231
2 Youssef-Agha, W., Aktan, H. M. and Olowokere, O. D. 'Seismic
response of low-rise steel frames', J. Struct. Eng., ASCE, 1989, 115,
594 - 607
3 Sivakumaran, K. S. "Seismic response of multi-storey steel building
with flexible connections', Engng. Struct. 1988, 10, 239-248
4 Krawinkler, H. and Mohasseb, S. 'Effects of panel zone deformations on seismic response', J. Const. Steel Res. 1987, 8, 233-250
5 Popov, E. P., Tsai, K. and Engeihardt, M. D. 'On seismic steel joints
and connections' Engng. Struct. 1989, 11, 148-162

6 Ghobarah, A., Osman, A. and Korol, R. M. 'Behaviour of extended


end-plate connections under cyclic loading', Engng. Struct. 1990, 12,
15 - 2 7
7 Korol, R. M., Ghobarah, A. and Osman, A. 'Extended end-plate
connections under cyclic loading: behaviour and design's J. Const.
Steel Res. 1990, 16, 253-280
8 Ghobarah, A., Korol, R. M. and Osman, A. 'Cyclic behaviour of
extended end-plate joints', J . Struct. Eng. ASCE 1992, 118 (5),
1333-1353
9 Osman, A., Ghobarah, A. and Korol, R. M. 'Moment-rotation relationship for extended end-plate connections', Proc. 1990 CSCE Conf.
IV, 517-536
10 Osman, A. and Ghobarah, A. 'Behaviour of panel zones in extended
end-plate joints', Proc. 10 WCEE, Madrid, 1992 Vol 6,
pp 3175-3180
11 National Building Code of Canada, National Research Council,
Ottawa, Ontario, Canada, 1990
12 Uniform Building Code, International Conference of Building
Officials, California, USA, 1988
13 Steel Structures for building (Limit States Design), CAN3-SI6-M89,
Canadian Standard Association, Rexdale, Ontario, Canada, 1989
14 Krawinkler, H. 'Shear in beam-column joints in seismic design of
steel frames', Engng J., AISC, 1978, 15 (3), 82-91
15 Osman, A., Korol, R. M. and Ghobarah, A. 'Bolted beam-to-column
subassemblages under repeated loading', Proc. sixth Canadian Conf.
Earthquake Engng, Toronto, 1991, pp 687-694
16 Azizinamini, A., Bradburn, J. H. and Radziminski, J. B. 'Initial stiffness of semi-rigid steel beam-to-column connections', J. Const. Steel
Res., 1987, 8, 7 1 - 8 0
17 Packer, J. A. and Morris, L. J. 'A limit state design method for the
tension region of bolted beam-to-column connections', Struct, Engng,
1977, 55, 446-458
18 Hendrick, A. and Murray, T. M. 'Column web compression strength
at end-plate connections', Engng. J., AISC, 1984, 3, 161 - 169
19 Kannan, A. E. and Powell, G. H. 'Drain-2D, A general purpose
computer program for dynamic analysis of inelastic plane structures',
Rep. EERC-73/6, University of California, Berkeley, 1973

Appendix
The applied moment versus shear strain relationship for
the panel zones is characterized by elastic behaviour
followed by a transition zone and then a strain hardening
zone.

Eng. Struct. 1993, Vol. 15, No 2

133

Seismic performance of moment resisting frames: A. Osman et al.

Elastic range

moment at this state, Mpn~, is given by

In this range, the panel zone deformation is resisted.


mainly by elastic shear stiffness of the panel and bending
stiffness of the column in the panel area. The initial stiffness of the panel, K,,, is given by

K,, = 1/Gtcw(dc _ t,,y) + d~/12El

(A1)

Here, t4 is the column flange thickness, G is the shear


modulus, I is the column's inertia and ot is a factor added
to the equation to account for the effect of column shear
and is equal to db/L,., where L~ is the column height.
The elastic range is normally extended to the point
where the column web at the panel area has yielded. The
yield moment is given by

Mp.i = 0.55dctc~oydb

(A2)

Mp,2 = Mp,, + 2(ovZ41 + oyZ42)


+ (YT -- %)Gs!t,',,,(d,- t,j)d#

(A3)

where G,, is defined as the plastic shear modulus, Zcj~is


the plastic modulus of the ith column flange and the
adjoining end-plate as one unit, and ~'r is the shear
deformation of the panel at the formation of the sway
mechanism in the column flanges and end-plates. Also,
3'7 is given by
7r =

a,,Z,-ymi,db
6E!~,,i,

(A4)

Consequently, Kr is given by
KT -- M~,,2 - Mr.,
(3'T -

(AS)

%)

Transition range

Strain hardening range

In this range, the panel deformation is resisted mainly by


the flexural stiffness of both the column flanges and the
end-plates and by the web strain hardening. It may be
extended up to the formation of a plastic mechanism in
the column flanges and the end-plate. The limiting

The stiffness at the strain hardening range is given by

134

Eng. Struct. 1993, Vol. 15, No 2

~S!

K,. = ~

K,.

(AS)

Das könnte Ihnen auch gefallen