Anda di halaman 1dari 10

MODELING PROCEDURES FOR PANEL ZONE DEFORMATIONS IN

MOMENT RESISTING FRAMES



Finley A. Charney, Virginia Tech, U.S.A.
William M. Downs, Simpson Strong Tie, Inc., U.S.A.


ABSTRACT

Elastic and inelastic deformations in the panel zone of the beam-column joint
region of moment resisting frames are responsible for a very significant
portion of the lateral flexibility of such systems. This paper provides a brief
theoretical basis for computing panel zone deformations, and compares
results obtained from two simple mechanical models to each-other and to
those obtained using detailed finite element analysis. It is shown that the
simplest mechanical model, referred to as the Scissors model, produces
results that are comparable to those obtained from the more complex
mechanical model, and also correlates well with the results computed from
the detailed finite element model.


INTRODUCTION

The influence of panel zone deformations on the flexibility of steel moment resisting frames
is very significant. This is true for elastic response, and particularly for inelastic response
when yielding occurs in the panel zone. Structural analysis should always include such
deformations.

While the state of stress in the panel zone is extremely complex, the sources of deformation
can be divided into three parts: axial, flexural, and shear. For low to medium rise frames,
axial deformations are negligible, flexural deformations are minor but significant, and shear
deformations are dominant. This paper concentrates on the shear component of panel
zone deformation. See Downs (1) for a detailed discussion on modeling approaches for
axial and flexural deformations within the panel zone.

Mathematical modeling procedures for panel zone deformation are typically based on
simple mechanical analogs which consist of an assemblage of rigid links and rotational
springs. The principal challenge in the derivation of such models is the development of the
transformations from shear in the panel zone to rotation in the springs of the analog. Two
mechanical models were studied in the research reported herein. These are the
Krawinkler Model (2) and the Scissors Model. When properly used, the results obtained
from these models are essentially identical, even though the kinematics of the Krawinkler
model is significantly different than that of the Scissors model. Unfortunately the Scissors
model is often misused in practice because analysts tend to compute spring properties that
were derived for the Krawinkler model and use them in the Scissors model. A complete
description of the mechanical models is presented in the next section of this paper.

Results obtained from structures implementing the simple mechanical models were
compared with those computed from a detailed finite element model of a full beam-column
subassemblage. The detailed model was created using ABAQUS (3). Both elastic and
inelastic analysis was performed. It was found that good correlation was obtained between
Connections in Steel Structures V - Amsterdam - June 3-4, 2004 121
the mechanical models and the finite element model. The results from the finite element
model were also compared to experimental results obtained during the SAC Project (4, 5)
and reasonably good correlation was obtained. The detailed analysis is briefly described in
the second main part of this paper.


MECHANICS OF BEAM-COLUMN JOINT

A typical interior beam-column subassemblage of a moment resisting frame is shown in
figure 1. The subassemblage is in equilibrium under the loads shown if the moments at the
mid-span of the girders and mid-height of the columns are zero. It is assumed that size and
span of the girders on either side of the column are same, and that a single column section
is used over the full height. The girders are welded to the column flanges. A doubler plate
may be used to reinforce the panel zone.

Terms and represent the ratios of the effective depth of the column to the span length,
and the effective depth of the girder to the column height, respectively. The effective depth
of a section is defined as the distance between the centers of the flanges. Use of these
terms in lieu of the actual physical dimensions greatly simplifies the derivation of the
properties of the models.
L
L
H H
V
C
V
C
V
C
/ H
V
C
/ H
L
L
H H
V
C
V
C
V
C
/ H
V
C
/ H
L
L
H H
V
C
V
C
V
C
/ H
V
C
/ H
C
V H
L
C
V H
L
L
L
H H
V
C
V
C
V
C
/ H
V
C
/ H
L
L
H H
V
C
V
C
V
C
/ H
V
C
/ H
L
L
H H
V
C
V
C
V
C
/ H
V
C
/ H
C
V H
L
C
V H
L
C
V H
L
C
V H
L


Figure 1. Typical interior beam-column subassemblage.


Total subassemblage drift

The total drift in the subassemblage, , is defined as the lateral displacement of the top of
the column with respect to the bottom of the column under the load V
C
. Following the
procedure described by Charney (6), this drift may be divided into three components, one
for the column, one for the girder, and one for the panel zone.


P G C
+ + = (1)

The column and girder displacement components are due to axial, flexural and shear
deformations occurring in the clear span region of the respective sections. The panel
contribution to displacement may also be divided into axial, flexural, and shear components:


PV PF PA P
+ + = (2)

122 Connections in Steel Structures V - Amsterdam - June 3-4, 2004
As stated earlier, this paper concentrates on the development of the panel zone shear
component of subassemblage displacement. It should be noted that this component
includes localized bending in the flanges of the column, but bending through the depth of
the panel is represented by
PF
.

Panel zone participation in total subassemblage drift

If it is assumed that the moment in the girder at the face of the column is resisted entirely by
the flanges of the girder, it can be shown by simple statics that the horizontal shear force in
the panel zone is

) 1 (
=
C
P
V
V (3)

The corresponding shear stress in the panel is


P
C
P
H V


=
) 1 (
(4)

This shear stress is uniform throughout the volume of the panel zone. The term
P
, which
represents the volume of the panel zone, appears repeatedly in the following derivations.

To determine the panel zone contribution to subassemblage drift, equal and opposite unit
virtual forces are applied in lieu of the actual column shears V
C
. The shear stress in the
panel due to the unit virtual shear force is


P
H


=
) 1 (
1

(5)

The contribution of panel zone shear strain to subassemblage drift is obtained by integrating
the product of the real strains and the virtual stresses over the volume of the panel. The
uniformity of stress and strain over the volume of the panel simplifies the integration.


P
C
V
P
PV
G
H V
dV
G

= =

2 2
1
) 1 (
(6)

The Krawinkler model and the Scissors model must have the same panel zone shear
contribution to displacement as given by equation 6.

Panel zone shear strength

Research performed by Krawinkler (2) has shown that the strength of the panel zone
consists of two components; shear in the panel itself, and flexure in the column flanges.
The larger of these components is the panel zone shear, which is resisted by the web of the
column acting in unison with the doubler plate, if present. If it is assumed that the yield
stress in shear is 3 / 1 0.6 times the uniaxial yield stress and that the column and doubler
plate are made from the same material, the yield strength of the panel in shear is


H
F
Lt F V
P y
P y YP


= =
6 . 0
6 . 0 (7)

Connections in Steel Structures V - Amsterdam - June 3-4, 2004 123
The second component of strength arises from flexural yielding of the flanges of the column.
This phenomenon, which is most significant for W14 and W18 columns with very thick
flanges, has been observed from tests (2), and may be computed using the principle of
virtual displacements. The computed shear strength due to column flange yielding in the
joint region is

H
t b F
V
Cf Cf y
YF

2
8 1. =
(8)

where the 1.8 multiplier is a calibration factor based on test results.

Force-deformation response

The assumed force-deformation behavior of the beam column joint is illustrated in figure 2.
In the figure the deformation is the racking displacement over the height of the panel. The
moment-rotation aspect of figure 2 is used later.

Shear Displacement
Spring Rotation

Y
4
Y
4
Y
Shear, V
Moment, M
V
YP
M
YP
V
YF
M
YF
Panel
Flange
Total
V
YP
M
YP
H
H

Y

Figure 2. Force-deformation relationship for beam-column joint.

The total response is equal to the sum of the response of the panel and the column flanges.
Following Krawinkler (2) it is assumed that the flange component yields at four times the
yield deformation of the panel component. It should be noted that figure 2 shows that the
flange component of the resistance is effective immediately upon loading. Krawinkler
assumes that this component of resistance does not occur until the panel yields in shear.
We have used the relationship shown in the figure as it simplifies the implementation of the
mechanical models without compromising accuracy.

The Krawinkler and Scissors models must be proportioned such that yielding is consistent
with figure 2.

The Krawinkler model

The Krawinkler model is shown in figure 3. The model consists of four rigid links connected
at the corners by rotational springs. The springs at the lower left and upper right corners
have no stiffness, and thereby act as true hinges. The spring at the upper left is used to
represent panel zone shear resistance, and the spring at the lower right is used to represent
column flange bending resistance. A total of twelve nodes are required for the model (there
are two nodes at each corner). The number of degrees of freedom in the model depends
124 Connections in Steel Structures V - Amsterdam - June 3-4, 2004
on the use of nodal constraints or slaving. The minimum number of DOF required to model
the panel is four in a planar structure. The maximum is twenty-eight.

Rigid Link
Real
Hinge
Rotational Spring
for Panel Shear
Rotational Spring
for Column Flange
Bending


Figure 3. The Krawinkler model.

The properties of the springs in the Krawinkler model are easily computed in terms of the
physical properties. Looking at only the panel spring, for example, the moment in the spring
is equal to the panel shear times the height of the panel. (See the diagram at the right of
figure 2.) The rotation in the spring is equal to the shear displacement in the panel divided
by the panel height. Hence,

H V M
P K P
=
,
(9)


P
P
P
P
K P
G
H V
H Lt G
H V

= =

1
,
(10)

Note that the K subscript in the above expressions refers to the Krawinkler model.

The stiffness of the rotational spring representing the panel in the Krawinkler model is the
moment divided by the rotation;


P
K P
K P
K P
G
M
S = =
,
,
,

(11)

The yield moment in the spring is simply the panel shear strength times the height of the
panel. Using equation 7,


PZ Y YP K YP
F H V M = = 6 . 0
,
(12)

As seen in figure 2, the stiffness of the flange bending component of the Krawinkler model
is equal to the yield moment in the flange bending component divided by 4.0 times the yield
rotation of the panel component. The yield rotation of the spring representing the panel
component is


G
F
K
M
Y
K YP
K YP
K YP
6 . 0
,
,
,
= = (13)

The yield moment is equal to the yield strength times the panel height;
Connections in Steel Structures V - Amsterdam - June 3-4, 2004 125


2
8 1 ) )( ( .
, Cf Cf Y YF K YF
t b F H V M = =
(14)

and the resulting stiffness is


G t b
M
S
Cf Cf
K YP
K YF
K F
2
75 0
4
) )( ( .
,
,
,
= =

(15)

In summary, Expressions 11 and 12 and 14 and 15 are all that are needed to model the
panel spring and the flange spring, respectively, in the Krawinkler model. If desired, a strain
hardening component may be added.

The Scissors model

The Scissors model is shown in figure 4. This model derives its name from the fact that the
model acts as a scissors, with a single hinge in the center. Only two nodes are required to
model the joint if rigid end zones are used for the column and girder regions inside the
panel zone. The model has four degrees of freedom. As with the Krawinkler model, one
rotational spring is used to represent the panel component and the other is used to
represent the flange component of behavior.

The properties of the Scissors model are determined in terms of those derived previously
for the Krawinkler model. First, consider the displacement participation factor for panel
shear as derived in equation 6. Noting that the denominator of this equation is the same as
the panel spring stiffness for the Krawinkler model, equation 6 may be rewritten as


K P
C
P
S
H V
,
2 2
) 1 (
= (16)

For the Scissors model, the moment in the spring under the column shear V
C
is simply V
C
H.
If the Scissors spring has a stiffness S
P,S
, the rotation in the spring is V
C
H/S
P,S
. The drift over
the height of the column is the rotation times the height, thus for the Scissors model,


S P
C
Scissors P
S
H V
,
2
,
= (17)

As this displacement must be identical to that given in equation 16, it is evident that the
relationship between the Krawinkler spring and the Scissors spring is as follows:


2
,
,
) 1 (
=
K P
S P
S
S (18)

Similarly, when the moment in the Krawinkler spring is V
P
H, the moment in the Scissors
spring is V
C
H. Using equations 3 and 9

) 1 (
,
= = H V H V M
C P K P
(19)


) 1 (
,
,

=
K P
S P
M
M (20)
126 Connections in Steel Structures V - Amsterdam - June 3-4, 2004
The relationships given by equations 18 and 20 hold also for the column flange components
of the models:

2
,
,
) 1 (
=
K F
S F
S
S (21)


) 1 (
,
,

=
K F
S F
M
M (22)

Rotational Spring
For Column Flange
Bending
Boundary of
Panel Zone
Real
Hinge
Rigid
Link
Rotational Spring
For Panel Shear
Rotational Spring
For Column Flange
Bending
Boundary of
Panel Zone
Real
Hinge
Rigid
Link
Rotational Spring
For Panel Shear


Figure 4. The Scissors model.

As an example, consider the case where and are 0.1 and 0.2, respectively, the Scissor
spring must be approximately twice as stiff and 1.43 times stronger than the Krawinkler
spring. Many analysts erroneously use the springs derived for the Krawinkler model in the
Scissors model. This will produce models that are more flexible than the true structure, and
that prematurely yield in the panel zone regions.

Comparisons between the Krawinkler and Scissors models

One should note from equations 18 and 20 that while the properties of the Scissors models
are dependent on the quantities and , those of the Krawinkler model are not. Since it
was explicitly assumed that the columns and girders on both sides of the joint are of equal
height and span, and these terms are reflected in and , the Scissors model may not be
used when this condition is violated. There is no such restriction on the use of the
Krawinkler model.

The deformed shape of the Krawinkler and Scissors models are shown in figure 5. In this
figure all of the deformation is assumed to be in the panel, with the girder and column rigid.
The most striking difference in the behavior between the two models is the offset in the
centrelines of the columns and girders in the Krawinkler model, which are not present in the
Scissors model.

A series of analyses were carried out using DRAIN-2DX (7) to determine the effect of the
kinematic differences on the pushover response of a series of assemblages and planar
frames which had yielding in the panel zone and at the ends of the girders. A variety of
girder spans were used, but the column height remained constant. Analysis was performed
with and without gravity load, and with and without P-Delta effects. For simple
subassemblages analyzed using the Krawinkler and the Scissors models, the pushover
responses were identical. For structures created by assembling subassemblages into a
rectilinear frame, but with real hinges at the midspan of the girders and midheight of the
columns, the pushover responses were again identical. Minor differences in the pushover
responses were obtained when the midspan/midheight hinges were removed. It was
Connections in Steel Structures V - Amsterdam - June 3-4, 2004 127
concluded, therefore, that the Scissors model, when properly used, is generally as effective
for analysis as is the Krawinkler model, given the approximations in the derivations and the
uncertainties involved in the analysis.


Offsets


Figure 5. Kinematics of Krawinkler model (left) and Scissors model (right).


COMPARISON WITH DETAILED FINITE ELEMENT ANALYSIS AND SAC RESEARCH

To evaluate the effectiveness of the elastic modeling techniques developed above, a series
of comparisons was performed using test results provided by Ricles (5) from Phase II of the
SAC Steel Project. The properties used for one of the test specimens are provided in Table
1. Additionally, the SAC subassemblage was modeled using ABAQUS. The ABAQUS
model of the subassemblage is shown in figure 6.

Table 1. Lehigh test C1: Geometric and material properties.
Yield Stress (ksi)
Member Size Length (in.) Grade
Mill Certs. Coupon Test
Girder W36x150 354 A572 Grade 50 57
56.7 flange
62.9 web
Column W14x398 156 A572 Grade 50 54
53.2 flange
52.2 web
Continuity Plate N/A N/A N/A N/A
Doubler Plate (2) @ A572 Grade 50 57 57.1



Figure 6. ABAQUS model of SAC subassemblage.
128 Connections in Steel Structures V - Amsterdam - June 3-4, 2004
The results of the analysis showed very good agreement between the Krawinkler, Scissors,
and ABAQUS models, and the SAC experimental results. See Downs (1) for a detailed
discussion of modeling techniques used in the ABAQUS analysis.


CONCLUSIONS

Simplified mechanical models such as the Krawinkler model and the Scissors model are
extremely effective in representing both elastic and inelastic panel zone deformations in
steel frame structures. While the Krawinkler model is considerably more complex and has
significant kinematic differences with the Scissors model, the results obtained using the two
models are essentially identical. The results from the simple mechanical models also
correlate well with more advanced finite element analysis, and with experimental results.

The Scissors model, however, is limited to use in frames with equal bays widths and equal
story heights. The Krawinkler model has no such restriction. Very significant errors will
occur in the analysis if the properties derived for the Krawinkler model are used in a
Scissors model. Such inconsistent use of the models was found in several references
reviewed by the authors.


NOTATION

ratio of effective depth of column to span length
ratio of effective depth of girder to column height

Y
yield displacement of panel
lateral displacement over height H

C
column contribution to displacement

G
girder contribution to displacement

P
panel zone contribution to displacement

PA
axial contribution to
P

PF
flexural contribution to
P

PV
shear contribution to
P

P,K
rotation in Krawinkler spring representing panel resistance

YP,K
yield rotation in Krawinkler spring representing panel resistance

P
shear stress in panel zone due to column shear force V
C

1
shear stress in panel zone due to unit column shear force
t
P
panel zone thickness, equal to column web thickness + doubler plate thickness
t
Cf
thickness of column flange
b
Cf
width of column flange
F
y
yield stress of steel used in column and doubler plate
G shear modulus of steel
H height of column from center of story above to center of story below
L length of girder from center of bay to center of bay
M
F,K
moment in Krawinkler spring due to column flange bending resistance
M
P,K
moment in Krawinkler spring due to panel shear resistance
M
YF,K
yield moment in Krawinkler spring due to column flange bending resistance
M
YP,K
yield moment in Krawinkler spring due to panel shear resistance
M
YF,S
yield moment in Scissors spring due to column flange bending resistance
M
YP,S
yield moment in Scissors spring due to panel shear resistance
S
P,K
stiffness of Krawinkler spring due to panel shear resistance
S
F,K
stiffness of Krawinkler spring due to column flange bending resistance
S
P,S
stiffness of Scissors spring due to panel shear resistance
S
F,S
stiffness of Scissors spring due to column flange bending resistance
Connections in Steel Structures V - Amsterdam - June 3-4, 2004 129
V
C
average shear force in columns above and below the joint
V
P
horizontal shear force in panel zone
P
volume of panel zone = LHt
P


REFERENCES

(1) Downs, William M, (2002). Modeling and Behavior of the Beam/Column Joint Region
of Steel Moment Resisting Frames, M.S.Thesis, Department of Civil and Environmental
Engineering, Virginia Tech, Blacksburg, Virginia.
(2) Krawinkler, H., (1978), Shear in Beam-Column Joints in Seismic Design of Frames,
Engineering Journal, v15, n3, American Institute of Steel Construction, Chicago,
Illinois.
(3) Hibbit, Karlson, and Sorensen, (2001). ABAQUS Users Manual, Verson 6.2.
(4) FEMA (2000). Recommended Seismic Design Criteria for New Steel Moment Frame
Buildings, FEMA-350. Federal Emergency Management Agency, Washington D.C.
(5) Ricles, J. M., (2002). Inelastic Cyclic Testing of Welded Unreinforced Moment
Connections Journal of Structural Engineering, ASCE, v128, n4.
(6) Charney, Finley A., (1993). "Economy of Steel Frame Buildings Through Identification
of Structural Behavior", Proceedings of the Spring 1993 AISC Steel Construction
Conference, Orlando, Florida.
(7) Prakesh, V. and Powell, G. H., (1993). DRAIN 2D-X Users Guide, University of
California, Berkeley, California.
130 Connections in Steel Structures V - Amsterdam - June 3-4, 2004

Anda mungkin juga menyukai