Está en la página 1de 36

Journal of Constructional Steel Research 58 (2002) 529–564

www.elsevier.com/locate/jcsr

Modeling of steel moment frames for seismic


loads
Douglas A. Foutch a,∗, Seung-Yul Yun b
a
University of Illinois at Urbana-Champaign, 3129 Newmark Lab, 205 N. Mathews, Urbana,
IL 61801, USA
b
University of Illinois at Urbana-Champaign, 3113 Newmark Lab, 205 N. Mathews, Urbana,
IL 61801, USA

Received 23 April 2001; received in revised form 15 August 2001; accepted 7 September 2001

Abstract

Simple elastic models based on centerline dimensions of beams and columns are widely
used for the design of steel moment resisting frames. However, for the performance prediction
and evaluation of these structures, different nonlinear models are being used to better simulate
their true behavior. Simple nonlinear modeling methods widely used as well as those with
more detailed modeling representations are investigated and compared.
A 9-story building and a 20-story building were designed for this study according to the
1997 NEHRP provisions. Different models for these structures were developed and analyzed
statically and dynamically. The models investigated involved the use of centerline dimensions
of elements or clear length dimensions, nonlinear springs for the beam connections, and linear
or nonlinear springs for the panel zones. A second group of models also incorporated the
fracturing behavior of beam connections to simulate the pre-Northridge connection behavior.
Two suites of ground motions were used for the dynamic analysis: typical California and near
fault ground motions. The differences in structural responses among different models for both
suites of motions are investigated.
According to static pushover analyses with roof displacement controlled, the benefit of the
increase in capacity that results from the detailed models is consistently observed for both the
9-story and 20-story buildings. When the models were excited by different ground motions
from each suite, the median responses of the more detailed models showed an increase in
capacity and a decrease in demand as expected. However, due to the randomness inherent in
the ground motions, variations were also observed. Overall, the model which incorporates
clear length dimensions between beams and columns, panel zones and an equivalent gravity


Corresponding author.
E-mail address: d-foutch@uiuc.edu (D.A. Foutch).

0143-974X/02/$ - see front matter  2002 Published by Elsevier Science Ltd.


PII: S 0 1 4 3 - 9 7 4 X ( 0 1 ) 0 0 0 7 8 - 5
530 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

bay without composite action from the slab seems to be a practical model with appropriate
accuracy.  2002 Published by Elsevier Science Ltd.

Keywords: Seismic analysis; Steel moment frames; Steel buildings; Earthquake response; Non-linear
analysis; Elastic analysis

1. Introduction

The engineer’s ability to model buildings has increased quickly over the past sev-
eral years with the development of advanced analysis programs and the competition
among software developers. In fact, our ability to model structural behavior probably
exceeds our ability to fully understand the observed behavior.
The first structural analysis programs that were developed in the early 1960s could
handle only linear prismatic beam and column members with fully restrained or
pinned joints and centerline dimensions. Programs in use today have a number of
elements that model material and geometric nonlinearities, rigid or partially
restrained connections, and flexible foundations and diaphragms. This paper will
cover commonly used modeling procedures for steel moment frames.
A word of caution is required. Although the modeling procedures described herein
are quite detailed and match measured behavior very well, it must be remembered
that this is still greatly simplified from the case of a real building which has cladding,
partitions, mechanical equipment, stairways and many other discounted attributes. A
real building might have irregularities and flexible foundations that are important
but not included here. It must be remembered that the calculations that follow are
only estimates of actual behavior.
A 9-story and a 20-story building were designed in accordance with the 1997
NEHRP provision for this study. Different models for those structures were
developed and analyzed statically as well as dynamically. Two suites of ground
motions were used for the dynamic analyses: typical California and near fault ground
motions. The comparisons of computed structural responses for the different models
are investigated.

2. Design of 9-story and 20-story buildings according to the 1997 NEHRP


provisions

The plan and elevation views of the buildings are given in Fig. 1. The buildings
were designed for a site in downtown Los Angeles where SS is 1.61g and S1 is 1.15g.
The perimeters of the buildings were designed as special moment frames so the
response reduction factor of R=8 was used. According to the 1997 NEHRP pro-
visions, the base shears corresponding to the 9-story and 20-story building were 300
and 244 kips, respectively. The approximate period equation prescribed in the pro-
vision was used to check for strength as well as drift requirements. Drift requirements
governed the design for both of the buildings. The section members assigned for
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 531

Fig. 1. Plan view and elevation view of 9-story and 20-story building.

each of the buildings are listed in Table 1. Box sections were used for the corner
columns of the 20-story frame since they needed to resist bi-axial bending from
lateral loadings. Doubler plates were inserted at the middle story panel zones of the
interior columns to satisfy the shear requirement as shown in the table.
The new element in the DRAIN-2DX program developed by Foutch and Shi [1]
was used to model the nonlinear behavior of the beam connections as well as panel
zones. Detailed descriptions of the nonlinear springs used for the beams and panel
zones will follow in a later section.
Six different models of the buildings were investigated. The first model used cent-
erline dimensions with nonlinear springs for yielding of beams as well as a leaning
column attached to the moment resisting frame to correctly account for the P⫺⌬
effect for the building. This model is denoted as M1-WO. The next three models
used clear length dimensions with nonlinear springs to model the panel zones as
well the beams. The first model of the three is similar to M1-WO but used the clear
532 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

Table 1
Sections assigned for the 9-story and 20-story buildings

Story w14 Story w24


Columns Doubler plate Beam Columns Doubler plate Beam

Exterior Interior Exterior Interior Exterior Interior Exterior Interior

9 w14×342 w14×398 0 0 w21×62 20 15×15×0.5 w24×207 0 0 w18×46


8 w14×342 w14×398 0 0 w27×94 19 15×15×0.5 w24×207 0 0 w24×55
7 w14×398 w14×455 0 0 w33×118 18 15×15×0.75 w24×250 0 0 w27×84
6 w14×398 w14×455 0 0 w33×118 17 15×15×0.75 w24×250 0 0.125 w30×108
5 w14×455 w14×550 0 0 w36×150 16 15×15×1.0 w24×279 0 0 w30×108
4 w14×455 w14×550 0 0 w36×150 15 15×15×1.0 w24×279 0 0.25 w33×118
3 w14×550 w14×550 0 0 w36×150 14 15×15×1.0 w24×279 0 0.375 w33×118
2 w14×550 w14×550 0 0 w40×183 13 15×15×1.0 w24×335 0 0.125 w33×118
1 w14×550 w14×605 0 0 w40×183 12 15×15×1.0 w24×335 0 0.25 w36×135
0 w14×550 w14×605 0 0 w40×183 11 15×15×1.0 w24×335 0 0.25 w36×135
10 15×15×1.25 w24×408 0 0 w36×135
9 15×15×1.25 w24×408 0 0 w36×135
8 15×15×1.25 w24×408 0 0 w36×135
7 15×15×1.25 w24×408 0 0 w36×135
6 15×15×1.25 w24×408 0 0.125 w36×135
5 15×15×1.25 w24×408 0 0.125 w36×135
4 15×15×2.0 w24×492 0 0 w36×135
3 15×15×2.0 w24×492 0 0 w33×118
2 15×15×2.0 w24×492 0 0 w33×118
1 15×15×2.0 w24×492 0 0 w33×118
0 15×15×2.0 w24×492 0 0 w33×118
⫺1 15×15×2.0 w24×492 0 0 w14×22

length of beams and columns with the flexibility of the panel zones modeled into
the joint. This is denoted as M2-WO. The second of the three includes one bay of
the frame model that represents all of the interior gravity columns but with simple
connection properties assumed for the beam springs and is denoted as M2-SC. The
last of the three is identical to the second model, but with resistance from the com-
posite slab on top of the beam in the gravity frames modeled into the beam springs,
and this is referred to as M2-Comp. The last two models of the six models are
identical to the M2-WO and M2-SC but fracturing behavior of the beam connections
is incorporated into the models. For those connections, when the plastic moment is
reached, the strength of the beam connection drops down to 10% of the plastic
moment capacity. The periods of each model for the 9-story and 20-story moment
frames are listed in Table 2. The model with the equivalent gravity bay frame with
rotational resistance from the slab is the stiffest since the contribution from the conti-
nuity of the interior columns and rotational strength of the beam connection is
included. It is interesting to note that M2-WO is stiffer than M1-WO. This is due
to the fact that the M2 model uses clear lengths of the beams and columns that make
this structure stiffer even though a flexible panel zone is also included. When clear
length models and centerline models are pushed statically using displacement control,
the demands for elements for the clear length model will be larger. The natural
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 533

Table 2
1st and 2nd mode of each model

M1 M2
WO WO SC Comp WO-frac SC-frac Comp-fr

9-story
T1, (s) 2.49 2.38 2.35 2.30 2.38 2.35 2.30
T2, (s) 0.89 0.85 0.83 0.81 0.85 0.83 0.81
20-story
T1, (s) 3.77 3.48 3.45 3.41 3.47 3.45 3.41
T2, (s) 1.30 1.20 1.20 1.19 1.20 1.20 1.19

periods for these models are remarkably similar and will respond about the same
for response to dynamic motions if linear elastic behavior is assumed.
First, drift demands using static pushover analyses and then dynamic analyses
using both suites of ground motions were investigated. Finally, dynamic drift
capacities of the models were calculated using Increment Dynamic Analysis (IDA)
which will be described later in this paper.

3. Ground motions

Two different suites of accelerograms were used for the study. The first suite of
accelerograms represents the typical ground motions for the LA site. The second
suite represents near fault ground motions. Each of the typical ground motions in
the first suite was scaled in a least square manner to match the 2% in 50-year hazard
spectra of the site at periods of 1.0, 2.0 s and 4.0 s. The descriptions of the ground
motions with their scaling factors are given in Table 3. The scale factors range from
1.72 to 1.87. A different scaling method was used for the second suite of ground
motions since those ground motions were generated specifically to represent the 2%

Table 3
Description of typical ground motions

Name Ground motion name Scale factor used

EQ01 Taft (1952) 1.72


EQ02 Castica (1971) 1.87
EQ03 Imperial Valley (1979) 1.83
EQ04 Pacoima Dame (1971) 1.83
EQ05 Northridge (1994) 1.85
EQ06 El Centro (1940) 1.85
EQ07 San Fernando (1971) 1.39
EQ08 Mammoth Lakes (1980) 1.82
EQ09 Morgan Hill (1984) 1.87
EQ10 North Palm Spring (1986) 1.81
534 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

Table 4
Description of near fault ground motions

Name Ground motion name Scale factor used

LF21 Kobe (1995) 0.65


LF23 Loma Prieta (1989) 0.65
LF25 Northridge (1994) 0.65
LF27 Northridge (1994) 0.65
LF29 Tabas (1974) 0.65
LF31 Elysian Park (Simulated) 0.65
LF33 Elysian Park (Simulated) 0.65
LF35 Elysian Park (Simulated) 0.65
LF37 Palos Verdes (Simulated) 0.65
LF39 Palos Verdes (Simulated) 0.65

in 50-year hazard level. They are the normal component of the LA 2% in 50-year
hazard level ground motions developed by Somerville et al. [2] for the SAC Phase
II project. The ground motions were scaled to minimize the error for the median
response of the ground motions. The scaling factor for this suite of ground motions
came out to be 0.65. The descriptions of the ground motions with their scaling factors
are given in Table 4. The scaled response spectra of the both suites of ground motions
are shown in Figs. 2 and 3. It was interesting to notice that the spectral accelerations
in the short period range (less than 1.0 s) for the typical ground motions were high
compared to those for the near fault motions whereas some of the near fault motions
possess bumps in the period region of 0.7–1.8 s. Therefore, the effects of higher
modes for the typical motions and the pulse motion for the near fault motions should
be examined for the calculated responses. Because of the large spectral accelerations

Fig. 2. Scaled response spectra for EQ01–EQ10 and their median spectra.
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 535

Fig. 3. Scaled response spectra for LF21, LF23, LF25, ..., LF39 and their median spectra.

at longer periods, it should be expected that the near-fault motions on average would
affect the 9-story and 20-story buildings more than for shorter structures.

4. Description of systems

4.1. Linear centerline models

When designing new buildings or evaluating existing or damaged buildings two


acceptance criteria must be checked: member strength and building stiffness (drift).
For new steel moment frame buildings the drift limitation always governs in high
seismic regions.
Research done by Krawinkler [3] has shown that a linear elastic model using
centerline dimensions is acceptable for design of special moment frames. The beam
moments may be checked at the location in the beam where it intersects the column
flange. Even though this model gives adequate results for design, it will not always
give good estimates of the distribution of shears, moments and axial forces through-
out the building under dynamic loads. The panel zones must be modeled explicitly
for frames with weak panel zones.

4.2. Elastic models with panel zones included

The next increase in reality is to include the panel zone behavior in the model.
The panel zone is the region in the column web defined by the extension of the
beam flange lines into the column as shown in Fig. 4. The simplest way to model
the panel zone for linear analysis is referred to as the scissors model also shown in
Fig. 5. The beams and column are modeled with a rigid link through the panel zone
region and a hinge in the beam is placed at the intersection of the beam and column
536 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

Fig. 4. Definition of panel zone.

Fig. 5. Scissors model for panel zone modeling.

centerlines. A rotational spring with stiffness kθ is then used to tie the beam and
column together. The rigid links stiffen the structure but the panel zone spring adds
flexibility. The net result is that this building model is usually stiffer than the center-
line model. Since it is stiffer it will help in satisfying the drift design criteria. It will
also give better estimates of shears, moments and axial forces in the members. Most
finite element programs currently used by engineers for seismic analysis have this
feature. The equations for determining the stiffness of the panel zone spring are
based on the yield properties of the panel zone. The yielding property of the panel
zone is
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 537

Fy
gy ⫽ ⫽ qy
冑3G
(1)

My ⫽ Vy·db ⫽ 0.55Fydct·db (2)


where,

Fy=the yield strength of the panel zone


G=the shear modulus=E / 2·(1 ⫹ n)
dc=depth of column
t=thickness of panel zone which is the thickness of the web of the column
plus the thickness of the doubler plates if they are utilized.
db=depth of beam
ν=Poisson’s ratio=0.3

So, the stiffness of the panel becomes


My
Kq ⫽ (3)
qy

4.3. Nonlinear centerline models

Models that allow yielding in the beams and columns are much more realistic
than linear models. Although nonlinear models are not required for design of new
buildings, they are very useful for evaluating existing and damaged buildings [4].
Most commonly used programs model this behavior by including a nonlinear flexural
spring at the ends of elastic beam and column members. The springs should be
assigned a very high stiffness compared to that of the beam or column. However,
the spring yields at the plastic moment capacity of the member. The correct structure
stiffness is maintained because it comes from the actual members rather than from
the spring. This model is shown schematically in Fig. 6 and is referred to in this
paper as M1-WO.
The spring is rigid until the plastic moment of the member is reached. After yield-
ing a post-yield stiffness is assigned to the spring that represents the strain hardening
behavior of the member. A strain hardening coefficient, α, is assigned to the spring
after yielding. A value of α equal to 0.03 is a reasonable choice. The spring behavior
and member plus spring behavior are shown in Fig. 6. The value of α equal to 0.03
is a good choice for calculating story drift angles out to about 3–4%. After this,
local flange buckling will begin to occur that causes α to gradually decrease to zero
and then it can become negative with larger drifts. Most programs will not allow a
negative value of α. For calculating building behavior beyond 4%, it is best to choose
a strain-hardening factor of zero.
For performance evaluation, the expected values of the yield strengths of the steels
should be used. Expected yield strengths of commonly used steels are given in Table
5 [5,6].
538 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

Fig. 6. Centerline model with nonlinear elements.

Table 5
12
Expected and lower bound material properties for structural steel of various grades [5,6]

Yield strength Tensile strength


Material Year of Lower bound Expected Lower bound Expected
specification construction

ASTM, A36 1961–1990


Group 1 41 51 60 70
Group 2 39 47 58 67
Group 3 36 46 58 68
Group 4 34 44 60 71
Group 5 39 47 68 80
ASTM, A572 1961–
Group 1 47 58 62 75
Group 2 48 58 64 75
Group 3 50 57 67 77
Group 4 49 57 70 81
Group 5 50 55 79 84
A36 and dual 1990–1999
grade 50
Group 1 48 55 66 73
Group 2 48 58 67 75
Group 3 52 57 72 76
Group 4 50 54 71 76

1
Lower bound values for material are mean⫺2 standard deviation values from statistical data. Expected
values for material are mean values from statistical data.
2
For wide flange shapes, indicated values are representative of material extracted from the web of the
section. For flange, reduce indicated values by 5%.
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 539

4.4. Nonlinear models with panel zones

Most of the pioneering work on nonlinear panel zone modeling has been perfor-
med by Krawinkler. His state-of-the-art report [3] provides a good discussion of this
topic and includes references to his earlier work [7,8]. Two methods of modeling
the nonlinear behavior of frames with yielding beams, columns and panel zones are
available. One procedure is based on the scissors model shown in Fig. 5. The panel
zone springs as well as the springs at the ends of the members are nonlinear. The
behavior of the member spring is exactly the same as described in the previous
section. The panel zone spring is assigned a stiffness of
My
Kq ⫽ (4)
qy
where
My ⫽ Vy·db ⫽ 0.55Fydct·db (5)
Fy
q y ⫽ gy ⫽
冑3G
(6)

In most cases, panel zones have a steeper post yield stiffness. Therefore, a value
of α equal to 0.06 is a reasonable value to use.
A better model is shown in Fig. 7. This model holds the full dimension of the
panel zone with rigid links and controls the deformation of the panel zone using two
bilinear springs that simulate a tri-linear behavior. With this, the large strength differ-
ence between the real behavior and the model is reduced.
The first slope post yield is steep and represents the behavior between the time
that yielding is initiated and the full plastic capacity is reached. After the plastic

Fig. 7. Panel zone modeling.


540 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

capacity is reached a small slope (2%) or zero slope may be used. This is shown in
Fig. 8.
Since yielding in the beams, columns and panel zones is represented well by this
model, the actual distribution of yielding throughout the structure will be represented
well. For design of new special moment frames, the panel zones yield first. But,
because of the steep strain-hardening slope for the panel zones, the beams will yield
shortly thereafter. This model is referred to as M2-WO.
Fig. 1 shows the 20-story building that was designed according to the 1997
NEHRP Provisions. This building will be used for comparing the different models
described here. The results of pushover analyses for the buildings with the specified
distribution of lateral forces required for new design in the 1997 NEHRP provisions
are shown in Fig. 9. ‘M1’ in the figure is the modeling case with the centerline
dimensions, whereas, ‘M2’ is for the model based on clear lengths plus panel zones.
M2 also includes the modeling of the panel zones. The panel zone is modeled with
tri-linear model spring and the full dimension of the member for the analysis. As
can be seen, the M2 model is initially a little stiffer than the M1 model. The M1
model with P⫺⌬ gives the lowest strength. Care should be taken when plotting the
roof drift ratio versus the total base shear. The roof drift ratio can be misleading
because it is incapable of capturing the local drift concentration. A good example
of this case can be seen for this building that is pushed to about 4% of global drift
with P⫺⌬ effects. The concentration of plastic deformations around the 3rd level
was the controlling factor. Fig. 10 shows the plot of global roof drift ratio, top story
drift ratio, and 3rd level story drift ratio versus total base shear. Global roof drift
ratio is defined as the roof displacement divided by the total height of the building.
Top story drift ratio is the story drift divided by the height of the story. The global
drift ratio shows the averaged drift ratio over the whole height. This pushover plot
reaches a peak at a drift of about 0.02 and rapidly has an increasing negative slope

Fig. 8. Panel zone load–deformation behavior.


D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 541

Fig. 9. Comparison of modeling for 1997 NEHRP 20-story building.

Fig. 10. Comparison between global drift ratio vs story drift ratios for 20-story building.

after that. When each story drift ratio is plotted the 3rd level concentration of plastic
deformation is very noticeable as shown in Fig. 10. Note, however, that this does
not reach a peak until almost 0.04 drift and then slowly becomes negative. The top
story drift actually remains elastic throughout the entire loading sequence. A plot of
displaced shapes of the building with increasing roof displacement is shown in Fig.
11. The story level where the tangential slope is small indicates a large change in
drift ratio. The concentration of plastic deformation can clearly be seen in Fig. 12
where the story drift ratio for each story level with increasing lateral load is plotted.
These results indicate that any nonlinear static procedure that relies on global roof
drift for a static pushover analysis is highly questionable.
542 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

Fig. 11. Displaced shape from static pushover analysis for 20-story building.

Fig. 12. Story drift ratio from static pushover analysis for 20-story building.

5. Description of components

5.1. Nonlinear springs for beams, columns, and panel zones

5.1.1. Reduced beam section connection


For new buildings, reduced beam sections that are also referred to as dog-bone
members were used for the analysis. They exhibit very good hysteretic behavior
with stable loops and good energy dissipation. Tests were performed by Venti and
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 543

Engelhardt [9]. A typical case of the hysteretic behavior is shown in Fig. 13. This
test used a w14×398 column member and w36×150 beam section. Both members
have a nominal yield strength of 50 ksi. A model for the analysis using the DRAIN-
2DX program is shown in Fig. 13. The expected yield strength of 57.6 ksi was used
in the modeling. The behavior of the member was modeled using a tri-linear model.
The model simulated the specimen behavior very well. The ratio between the beam
plastic moments to the first yielding point as well as the second moment value were
calculated and used for determining the yielding properties of the other member
sizes. Seventy-four percent of the plastic moment of the beam was used as the first
yield moment for both positive and negative moments. For the second yield moment
value, factors of 132% of the first yielding moment for the positive side and 120%
of that for the negative side of the connection were used. The rotational value for
the second yielding moment of 0.03 radians for the positive side and 0.017 radians
for the negative side were used for the protocol model. The rotational values that
are proportional to the plastic section modulus were assigned for the other beam
sections. The strength degradation ratio that is the drop of the strength at each new
plastic excursion was assigned a value of 0.83. This value was fixed for all member
sizes although in reality, there would be variations from member to member. The
drift demand is not significantly affected by the choice of this ratio. Differences in
drift demand calculations would not vary by more than 2 or 3% because of this
difference. An illustration of the yielding values for the protocol member (w36×150)
and the 6th level beam in the 9-story building (w33×118) are shown in Fig. 14. The
plastic moments for the members are 33,750 (k-in) for w36×150 and 23,904 (k-in)
for the w33×118 member.

5.1.2. Fracturing beam connection


Fracturing beam connections were incorporated into the model to simulate the
behavior observed for the pre-Northridge buildings. The measured and the modeled
hysteresis behavior of the connection are shown in Fig. 15. The new element in

Fig. 13. Measured [9] and model [4] of moment–rotation behavior of RBS connection.
544 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

Fig. 14. Illustration of yielding values for w36×150 (protocol) and w33×118 (6th and 7th level of 9-
story building).

Fig. 15. Measured [13] and model [4] of moment–rotation behavior of fracturing connection.

DRAIN-2DX developed by Foutch and Shi [1] was used to model the strength drop
in the connection after fracture. In the positive rotation case, the strength was mod-
eled to drop to 10% of the original strength of the connection when the plastic
moment was reached just like the measured response. In the negative rotation side,
the loss of strength at about 0.04 radians is observed for the measured behavior. For
negative moment the crack in the bottom flange closes so typical bilinear-type of
behavior occurs out to about 0.04 radians when the top flange fractures. However,
due to limitations of the element, a gradual decrease in strength was modeled into the
connection. Therefore, the connection arrives at about zero strength at 0.04 radians to
simulate the fracture of the connection. Significant increase in demand as well as
decrease in capacity of structural response is expected.
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 545

Fig. 16. Illustration of simple connection in gravity frames.

5.1.3. Simple connections in gravity frames


The gravity frames are usually thought of as frames with no resistance to the
lateral load since the beam flanges are not connected to the column flanges. The
frame is sometimes modeled with pinned connections to capture the P-Delta effect
due to additional gravity load from the interior frames. However, according to the
experimental results from Liu and Astaneh-Asl [10] the resistance not only exists
but sometimes is significant due to the additional resistance occurring when a com-
pression force in the composite floor slab is connected by a tension force in the
shear tab. Additional resistance is encountered when the flanges of the beam come
in contact with the column. An illustration of the connection is shown in Fig. 16.
Fig. 17 shows a typical case where the shear tab with concrete slab on top of the
beam resists lateral load for many cycles of motion. This is a case with w18×35
beam connected to the w14×90 column with shear tab and concrete on top. Minimum
reinforcement was used for the slabs. The moment–rotation behavior of the connec-
tion was modeled with a nonlinear spring that drops in strength at specified rotations.

Fig. 17. Measured [10] and model [4] of moment–rotation behavior of simple beam in gravity frame.
546 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

The model of the connection is shown in Fig. 18. A portion of the beam stiffness
was used for the stiffness of the connection since it will not be like the rigid cases.
The proportion was determined to be 25% of the stiffness of the beam. Also, the
connections cannot be expected to develop the full plastic moment capacity. The
maximum moment for the positive moment was taken as 38% of Mp and that for
the negative side as 11% because these values resulted in good matches between
experiment and analysis. The fact that the positive side develops higher moment is
attributed to the compressive resistance of concrete slab on top of the girder bearing
against the column. The tensile strength of the slab cannot be expected to help much
since minimum reinforcement is used. The rotation at which the strength drops is
assigned a value of 0.045 radians for the positive side and 0.05 radians for the
negative side of the connection. The drop in strength was assigned a value of 53%
for the positive and 89% for the negative side. Those rotational values for the other
sections were calculated using the disproportional value to the depth of the beams.
Again gradual degradation of strength was modeled using 0.97 as the strength degra-
dation factor. As will be seen later in this paper, the resistance from the gravity
frame is significant. However, most of the contribution is not from the composite
connection but from the flexural resistance from the continuous columns acting in
conjunction with the rigid floor slabs. According to the report by Yun and Foutch
[11], the differences in responses between the models with the simple connection
are negligible as long as the continuity of the gravity frame columns is modeled.
This is due to the fact that the connections lose strength at very early stages of the
ground motions leaving only the columns to resist the lateral load. Fig. 18 shows
an illustration of the yielding properties of a protocol connection and the connection
from a typical floor of the 9-story building.

5.2. Other modeling attributes

Another feature that should be included for analysis of tall buildings or shorter
buildings taken out to large drifts is the P⫺⌬ effect. When the structure is displaced

Fig. 18. Illustration of yielding properties for w18×35 (protocol) and w16×26 beams in gravity frames
(typical beam for 9-story building).
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 547

Fig. 19. Modeling interior columns for P⫺⌬ effect only.

laterally the gravity forces acting through the displacement causes additional over-
turning moments to develop in the structure. For a perimeter frame building this can
be a very significant effect since the perimeter frames must carry the overturning
moments of the entire building including the gravity frames.
One way to do this is to provide a dummy column in the model that carries the
gravity loads in the building not directly carried by the moment frame. The column
is connected to the moment frame using rigid links with hinges at each end as shown
in Fig. 19. The columns are hinged at both top and bottom. By doing this, only
additional overturning moment from the lateral displacement will be induced. The
columns will not help carry any of the lateral loads since they are pinned. However
in reality, the interior columns do help the moment frames since the columns are
not connected with a hinge and some resistance exists for the shear tab connection
in the beams due to the slab on top. An additional bay that has the equivalent proper-
ties for all of the interior frames can be used as shown in Fig. 20. The columns and

Fig. 20. Modeling interior columns for P⫺⌬ effect and resistance from the equivalent interior bay.
548 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

beams will have the equivalent stiffness and strength for the corresponding stories
of the interior gravity frames. The beam springs used for the gravity frame have the
hysteresis behavior described in Fig. 17. The contribution of the equivalent gravity
bay comes from both the flexural resistance of the columns and well as those from
the beam springs used. However, since the strengths of the beam springs are very
small compared to the moment frame springs, most of them will yield at a very early
stage of the excitation. Modeling parameters for these gravity frames connections
are given in Yun and Foutch [11].

6. Static analysis

Static pushover analyses using both 9-story and 20-story building models were
performed using the lateral force distribution calculated from the 1997 NEHRP Pro-
visions. The approximate period from the provisions was used to obtain the total
base shear for each building. The lateral force distribution coefficient is defined in
1997 NEHRP Provision as
Wx·hxk
Cvx ⫽
冘 wi·hik
(7)

Models were pushed to 5% global drift ratio with roof displacement controlled.
For the models with fracturing connections, the structures were unable to sustain 5%
global drift ratio so the analysis had to be stopped earlier. The static pushover plots
for both of the buildings are shown in Figs. 21 and 22. The model with the equivalent
gravity bay and slab effect shows the highest capacity and that with centerline model
exhibiting the lowest capacity for both 9-story and 20-story buildings. The strength
difference between the two is 18% for the 9-story building and 25% for the 20-story

Fig. 21. Result from static pushover analysis for the 9-story building models.
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 549

Fig. 22. Result from static pushover analysis for the 20-story building models.

building which is significant. Little difference in strength is observed among the


models with fracturing beam connections. This is due to the fact that the resistance
lost when reaching the plastic moment of the beam connection is large. So when
one joint fractures, the other joints have to share the additional force and this causes
the other joints to fracture almost simultaneously. The difference in behavior between
those with connections that fracture and those that do not is substantial. This is a
strong indication that models used to evaluate existing buildings with pre-Northridge
connections must include the effects of fracture if the results are to be meaningful.
This can be done more simply than indicated here [4]. The percent of lateral resist-
ance from the equivalent gravity bay is very small compared to the main lateral
resisting frame. The higher effect of P⫺⌬ is very noticeable for the 20-story frame
for which the load–deformation response becomes negative at a much smaller drift.
It should be pointed out that the response in the negative tangential stiffness region
is not realistic. This part of the curve exists because a displacement controlled static
pushover method was used. Therefore if collapse is defined as the drift angle at zero
tangential slope, the drift capacity of the 9-story and 20-story models would be about
3.5 and 1.7%, respectively. As will be verified later in the dynamic analysis section,
the capacity obtained from static pushover is very conservative. The reduction in
lateral force due to elongation of fundamental period and cyclic nature of the ground
motions will help the structure to sustain larger drifts in dynamic response.

7. Dynamic analysis

Nonlinear dynamic analyses using both suites of scaled ground motions described
earlier have been used for drift demand as well as drift capacity calculations.
550 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

7.1. Dynamic drift demands

The median of the maximum inter-story drifts from each scaled ground motion
was calculated for all of the models described. All of the maximum drift values as
well as the median responses for the 9-story and 20-story frames are listed in Tables
6 and 7, respectively. Results for the typical ground motions are listed on the left
whereas those for the near fault ground motions are listed on the right side of the
table. The maximum drift demands for each earthquake are plotted in Figs. 23 and 24.
The median drifts for the 9-story models without fracture are all about 0.032 for
the ordinary motions and 0.043 for the near-fault motions. The drift demands for
the M2 model with composite action in the gravity frames were about 10% smaller
than the other three models in both cases, but this is not significant. The median
drift demands for the frames with connection fracture were significantly larger than
for frames without. The median values were about 0.056 for the frames with fractured
connections for both sets of ground motions. The frames with fracturing connections
collapsed during two of the ordinary and three of the near fault motions.
The median drift demands for the 20-story building models without fracture were
slightly smaller than for the 9-story buildings for both sets of ground motions. How-
ever, the accuracy is overstated in these figures so one could say that the drift
demands averaged 0.03 for all 9- and 20-story buildings without fracture. There was
a very significant difference between the 20-story models with fracturing connec-
tions. Model M2-WO-fr model collapsed during eight ordinary ground motions and
six near fault motions, but M2-SC-fr did not collapse for any of the accelerograms.
This implies that the gravity frames should be included in models if pre-Northridge
connections are used in the buildings.
Figs. 25 and 26 show maximum inter-story drifts over the height of the building
for the M1-WO and the M2-Comp model excited by near fault ground motions.
Although median drifts are similar, the M2-Comp model had a smaller standard
deviation over the height of the frame. In addition, the M2-Comp 20-story model
sustained significantly smaller maximum drift demands for the one or two extreme
ground motions.
The effect of incorporating the contribution from the equivalent gravity bay is
more noticeable for the 20-story frame models due to larger P⫺⌬ effects. The model
with fracturing beam connections but without the equivalent gravity bay collapsed
for many ground motions of both suits as mentioned above. However, only one
ground motion induced collapse for the model with fracturing beam connections and
the equivalent gravity bay.

7.2. Dynamic drift capacities

7.2.1. Incremental Dynamic Analysis (IDA)


The Incremental Dynamic Analysis (IDA) procedure was used to determine the
capacity of the frames. Median responses were calculated using a similar procedure
as for the statistical drift demand calculations. It is important to note that the analyti-
cal model used for determining the global drift demand reproduces the major features
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564
Table 6
Drift demands for the 9-story building for ordinary ground motions (left) and near fault ground motions (right)

M1-WO M2-WO M2-SC M2-Comp M2-WO- M2-SC-fr M1-WO M2-WO M2-SC M2-Comp M2-WO-fr M2-SC-fr
fr

Eq. 1 0.0507 0.0430 0.0430 0.0379 0.1037 collapse If21 0.0498 0.0542 0.0537 0.0524 0.0539 0.0570
Eq. 2 0.0292 0.0350 0.0389 0.0318 0.0436 0.0407 If23 0.0277 0.0284 0.0286 0.0286 0.0379 0.0365
Eq. 3 0.0443 0.0445 0.0514 0.0509 0.0539 0.0629 If25 0.0359 0.0361 0.0358 0.0325 0.0400 0.0383
Eq. 4 0.0268 0.0264 0.0266 0.0255 0.0286 0.0300 If27 0.0390 0.0411 0.0393 0.0376 0.0377 0.0443
Eq. 5 0.0240 0.0221 0.0267 0.0260 0.0466 0.0469 If29 0.0289 0.0300 0.0301 0.0245 0.0268 0.0347
Eq. 6 0.0279 0.0287 0.0330 0.0293 0.0719 0.0556 If31 0.0331 0.0346 0.0346 0.0252 0.0540 0.0630
Eq. 7 0.0303 0.0306 0.0290 0.0281 0.0878 collapse If33 0.0440 0.0450 0.0446 0.0448 0.0480 0.0658
Eq. 8 0.0402 0.0461 0.0431 0.0308 0.0605 0.0604 If35 0.0685 0.0596 0.0553 0.0529 collapse 0.3258
Eq. 9 0.0191 0.0207 0.0234 0.0217 0.0327 0.0345 If37 0.0759 0.0689 0.0672 0.0615 collapse collapse
Eq. 10 0.0233 0.0230 0.0244 0.0258 0.0299 0.0386 If39 0.0345 0.0356 0.0359 0.0361 0.0406 0.0442
δ 0.3135 0.2946 0.2773 0.2657 0.4322 0.2935 δ 0.3644 0.3036 0.2885 0.3147 0.2852 0.3544
Xm 0.0301 0.0307 0.0327 0.0297 0.0509 0.0469 Xm 0.0409 0.0414 0.0408 0.0377 0.0406 0.0442

551
552
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564
Table 7
Drift demands for the 20-story building for ordinary ground motions (left) and near fault ground motions (right)

M1-WO M2-WO M2-SC M2-Comp M2-WO- M2-SC-fr M1-WO M2-WO M2-SC M2-Comp M2-WO-fr M2-SC-fr
fr

Eq. 1 0.0379 0.0258 0.0273 0.0261 collapse 0.0383 If21 0.0282 0.0257 0.0259 0.0257 collapse 0.0483
Eq. 2 0.0341 0.0277 0.0274 0.0262 collapse 0.0312 If23 0.0241 0.0239 0.0217 0.0215 0.0321 0.0289
Eq. 3 0.0372 0.0401 0.0412 0.0421 collapse 0.0352 If25 0.0278 0.0276 0.0283 0.0266 0.0341 0.0352
Eq. 4 0.0188 0.0167 0.0168 0.0167 3.0468 0.0245 If27 0.0211 0.0185 0.0171 0.0169 collapse 0.0188
Eq. 5 0.0218 0.0255 0.0292 0.0280 0.0260 0.0287 If29 0.0211 0.0164 0.0188 0.0198 collapse 0.0224
Eq. 6 0.0175 0.0165 0.0165 0.0161 2.2844 0.0262 If31 0.0209 0.0193 0.0189 0.0193 0.0239 0.0236
Eq. 7 0.0752 0.0577 0.0433 0.0369 collapse 0.0467 If33 0.0357 0.0300 0.0269 0.0262 0.0427 0.0367
Eq. 8 0.0311 0.0368 0.0428 0.0454 collapse 0.0322 If35 0.0865 0.0859 0.0724 0.0711 21.5200 collapse
Eq. 9 0.0211 0.0194 0.0213 0.0198 0.0248 0.0265 If37 0.0395 0.0384 0.0358 0.0353 collapse 0.0393
Eq. 10 0.0160 0.0180 0.0156 0.0156 collapse 0.0181 If39 0.0223 0.0199 0.0180 0.0177 collapse 0.0243
δ 0.5276 0.4395 0.3766 0.3853 collapse 0.2557 δ 0.5624 0.6099 0.5408 0.5346 collapse 0.3074
Xm 0.0270 0.0258 0.0262 0.0253 collapse 0.0298 Xm 0.0279 0.0254 0.0245 0.0243 collapse 0.0289
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 553

Fig. 23. Drift demands for the 9-story building for ordinary ground motions (left) and near fault ground
motions (right).

Fig. 24. Drift demands for the 20-story building for ordinary ground motions (left) and near fault ground
motions (right).

Fig. 25. Drift demands for the 9-story M1-WO (left) and M2-Comp (right) excited by near fault ground
motions (right).
554 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

Fig. 26. Drift demands for the of 20-story M1-WO (left) and M2-Comp (right) excited by near fault
ground motions (right).

of the measured response such as sudden loss of strength or pinching. This means that
the measured hysteresis behavior must be modeled as closely as possible. Modeling
requirements are given in an earlier part of this paper. It should be noted that based
on the SAC research the connection that reaches a plastic rotation of 0.03 without
significant loss of strength and 0.05 without complete loss of strength should have
a median global drift capacity of 0.09 or greater for both 9- and 20-story models
for L.A.-type ground motions. This can be thought of as the lower bound behavior
of a connection that satisfies the AISC test protocol. Including the gravity columns
in the model helps to stabilize the building at large drifts. If the computer program
is capable of handling complex moment–rotation behavior, the moment developed
in gravity frames through the columns composite beam action can be included.
The global stability limit is determined using the Incremental Dynamic Analysis
(IDA) technique developed by Cornell and his associates [12, 15]. The procedure
that was used to perform this analysis is as follows:

1. Choose a suite of ten to twenty accelerograms representative of the site and hazard
level. The SAC project developed typical accelerograms for Los Angeles, Seattle
and Boston sites [2]. These might be appropriate for similar sites.
2. Perform an elastic time history analysis of the building for one of the accelerog-
rams. Plot the point on a graph whose vertical axis is the spectral ordinate for
the accelerogram at the first period of the building and the horizontal axis is the
maximum calculated drift at any story. Draw a straight line from the origin of
the axis to this point. The slope of this line is referred to as the elastic slope for
the accelerogram. Calculate the elastic slope for the rest of the accelerograms
using the same procedure and then calculate the median slope. The slope of this
median line is referred to as the elastic slope, Se (see Fig. 27).
3. Perform a nonlinear time history analysis of the building subjected to one of the
accelerograms. Plot this point of maximum drift on the graph. Call this point ⌬1.
4. Increase the amplitude of the accelerogram and repeat step 3. This may be done
by multiplying the accelerogram by a constant that increases the spectral ordinates
of the accelerogram by 0.1g. Plot this point as ⌬2. Draw a straight line between
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 555

points ⌬1 and ⌬2. If the slope of this line is less than 0.2 Se then ⌬1 is the global
drift limit. This can be thought of as the point at which the inelastic drifts are
increasing at 5 times the rate of elastic drifts.
5. Repeat step 4 until the straight-line slope between consecutive points ⌬i and ⌬i+1,
is less than 0.2 Se. When this condition is reached, ⌬i is the global drift capacity
for this accelerogram.
6. Choose another accelerogram and repeat steps 3 through 5. Do this for each acce-
lerogram. The median capacity for global collapse is the median value of the
calculated set of drift limits. An example for one accelerogram for an L.A. site
for a 20-story weak-column OMF building is shown in Fig. 27. The open circles
represent the IDA calculations for an accelerogram where the 0.2 Se slope determ-
ined the capacity. The point ⌬7 would be considered to be the drift capacity.

For the SAC project, the upper bound on the drift capacity was assumed to be
0.10. It was believed that the analytical results for drift greater than 0.10 would
not be reliable [14]. The issue of the safety of the occupants was paramount in
this design.

7.2.2. Calculated capacity


The IDAs were performed according to the procedures described in the previous
section. The median drifts as well as the spectral acceleration capacities were calcu-
lated for each suit of ground motions described earlier. A strain-hardening ratio of
0.03 was used for all of the analyses in this study. The increment of ground motion
intensity used was 0.2g for the cases without fracturing connections and 0.1g for the
models with fracturing connections since those models are expected to collapse at
earlier intensities of ground motions. However, the 20-story model with fracturing
connections without the equivalent gravity bay collapsed even at 0.1g for two ordi-
nary ground motions and for one near fault ground motion. The lateral force that 9-
story and 20-story building was designed for is 0.066g and 0.042g, respectively. A

Fig. 27. Definition of collapse from IDA analyses for two ground motions.
556 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

smaller increment of 0.02g was used for some IDA analyses. It was found that this
20-story model reaches incipient collapse at 0.08g that is only a little greater than
the design level. The ground motion increment must be small enough so that drift
increment is relatively small for each step. The values given above should be con-
sidered as upper bounds. The use of a larger increment would usually result in a
smaller drift capacity and larger variation of the capacity. Therefore, it would give
conservative results.
The individual drift capacities for each 9-story model along with the median values
are given in Table 8. The capacities for the 20-story buildings are given in Table 9.
Plots of the drift capacities are shown in Figs. 28 and 29.
For the 9-story models without fracture, the median capacities for the ordinary
California ground motions were all about the same. For the near-fault motions, the
model with the gravity frames had greater capacities. The drift capacities for the 9-
story models with fracturing connections had much smaller capacities than the mod-
els without fracture. For instance the capacities for M2-SC and M2-SC-fr were 0.13
and 0.08, respectively, for the standard motions and 0.18 and 0.06, respectively, for
the near fault motions. There was not a significant difference between the two models
with fractures.
The differences in capacity among the various models were much greater for the
20-story models when subjected to the ordinary accelerograms. Models M1-WO and
M2-WO had drift capacities of 0.07 and 0.05, respectively, while models M2-SC
and M2-Comp had capacities of 0.09 and 0.10, respectively. For the near-fault
motions, the four models without fracture had comparable capacities of about 0.10.
Again, the capacities for the model with fracture were significantly smaller than for
those without fracture. The capacities for M2-WO-fr and M2-SC-fr were 0.02 and
0.05, respectively, for the ordinary motions and 0.03 and 0.05 for the near-fault
motions. In addition, the capacities for the fracturing models without the gravity bay
were significantly smaller than those with the gravity bay.
It should be pointed out once again that this is a numerical exercise where relative
capacities are compared. One should not expect that a real building would be able
to resist drift levels of 0.20 without collapse. This is why the SAC project placed
an upper limit of drift capacity of 0.10.
The SAC project computed uncertainties and confidence levels in terms of story
drifts. This was chosen because it is a quantity that is calculated by the designer as
a regular part of the design process. Another quantity that could be used is the
collapse limit based on spectral amplitude. These are also a natural result of the IDA
analyses. The median strengths for the 0- and 20-story buildings are given in Tables
10 and 11 respectively. The spectral acceleration capacities for each ground motions
are shown in Figs. 30 and 31.
The spectral acceleration capacities vary greatly for the ordinary ground motions
but very little for the near-fault motions. Spectral acceleration capacities were sug-
gested by some researchers because it was believed that there would be less scatter.
This is clearly not so. The same conclusions that were drawn from the drift capacities
would also be derived from the spectral acceleration capacities.
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564
Table 8
Drift capacities for the 9-story building for ordinary ground motions (left) and near fault ground motions (right)

M1-WO M2-WO M2-SC M2-Comp M2-WO-fr M2-SC-fr M1-WO M2-WO M2-SC M2- M2-WO-fr M2-SC-fr
Comp

Eq. 1 0.0590 0.0555 0.0542 0.1049 0.0351 0.0655 If21 0.1531 0.1908 0.2315 0.2086 0.0545 0.0722
Eq. 2 0.1043 0.1009 0.1561 0.1235 0.0664 0.0803 If23 0.1936 0.2180 0.2341 0.1941 0.0625 0.0426
Eq. 3 0.1410 0.1630 0.1332 0.0858 0.0638 0.0836 If25 0.0884 0.1477 0.1329 0.1271 0.0547 0.0490
Eq. 4 0.2240 0.2051 0.2313 0.2583 0.0680 0.0744 If27 0.1662 0.1701 0.1558 0.1935 0.0621 0.0866
Eq. 5 0.1236 0.1081 0.1156 0.1334 0.0736 0.0838 If29 0.1161 0.1406 0.1957 0.1666 0.0697 0.0310
Eq. 6 0.0516 0.0638 0.0843 0.0531 0.0665 0.0953 If31 0.1192 0.1498 0.2317 0.1857 0.0873 0.1069
Eq. 7 0.0526 0.0775 0.0842 0.0759 0.0444 0.0385 If33 0.1431 0.1685 0.1191 0.1506 0.0700 0.0777
Eq. 8 0.1312 0.1289 0.1466 0.1314 0.0637 0.1082 If35 0.1525 0.1152 0.1549 0.1368 0.0788 0.0709
Eq. 9 0.1870 0.1983 0.2052 0.1779 0.0994 0.1208 If37 0.2111 0.2159 0.1405 0.2492 0.0387 0.0389
Eq. 10 0.1324 0.1650 0.1590 0.1638 0.0561 0.0731 If39 0.2014 0.1936 0.2569 0.2262 0.0609 0.0593
δ 0.4486 0.4108 0.3876 0.4306 0.2651 0.2718 δ 0.2538 0.1946 0.2670 0.2100 0.2105 0.3617
Xm 0.1091 0.1164 0.1271 0.1192 0.0615 0.0794 Xm 0.1496 0.1678 0.1788 0.1798 0.0625 0.0595

557
558
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564
Table 9
Drift capacities for the 20-story building for ordinary ground motions (left) and near fault ground motions (right)

M1-WO M2-WO M2-SC M2-Comp M2-WO- M2-SC-fr M1-WO M2-WO M2-SC M2-Comp M2-WO-fr M2-SC-fr
fr

Eq. 1 0.0792 0.0337 0.0676 0.0691 0.0092 0.0352 If21 0.0618 0.0463 0.0635 0.0685 0.0401 0.0468
Eq. 2 0.0779 0.0180 0.0835 0.0884 0.0094 0.0401 If23 0.1364 0.1386 0.1058 0.1068 0.0415 0.0657
Eq. 3 0.0586 0.0508 0.0580 0.1396 0.0116 0.0454 If25 0.0469 0.1184 0.0906 0.0853 0.0485 0.0413
Eq. 4 0.0973 0.0533 0.1645 0.1514 0.0095 0.0373 If27 0.1540 0.1442 0.0956 0.0942 0.0067 0.0424
Eq. 5 0.0683 0.0876 0.0789 0.0785 0.0318 0.0550 If29 0.0844 0.0346 0.0853 0.0697 0.0242 0.0325
Eq. 6 0.0540 0.0402 0.1094 0.1083 0.0261 0.0433 If31 0.1363 0.0659 0.1222 0.1146 0.0497 0.0680
Eq. 7 0.0276 0.0211 0.0924 0.0870 0.0109 0.0586 If33 0.1106 0.0816 0.1475 0.1345 0.0659 0.0313
Eq. 8 0.0528 0.0412 0.0885 0.0887 0.0162 0.0389 If35 0.1216 0.1236 0.1000 0.1004 0.0541 0.0418
Eq. 9 0.0839 0.0655 0.1114 0.0909 0.0316 0.0731 If37 0.0540 0.0901 0.0715 0.0680 0.0091 0.0507
Eq. 10 0.0926 0.0805 0.1188 0.1153 0.0196 0.0467 If39 0.1092 0.0963 0.1401 0.1403 0.0230 0.0435
δ 0.3004 0.4499 0.3063 0.2581 0.4921 0.2445 δ 0.3601 0.3858 0.2643 0.2631 0.5090 0.2594
Xm 0.0662 0.0455 0.0928 0.0984 0.0156 0.0460 Xm 0.0951 0.0872 0.0987 0.0949 0.0319 0.0449
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 559

Fig. 28. Drift capacities for the 9-story building for ordinary ground motions (left) and near fault ground
motions (right).

Fig. 29. Drift capacities for the 20-story building for ordinary ground motions (left) and near fault ground
motions (right).

8. Summary and conclusions

Six models of a 9-story building and six of a 20-story building were developed
for this study. The two buildings were designed to comply with the 1997 NEHRP
provisions. Four of the models for each building had ductile connections modeled
after test results for a reduced beam section connection. The other two models for
each building had connections that could fracture in a similar fashion to a fully
welded pre-Northridge connection. The effects of gravity frames were included in
three models for each building. Each model was subjected to a suite of 10 standard
California accelerograms and one of 10 near fault motions.
Results of the study suggest the following conclusions:

1. A bare-frame model using centerline dimensions is more flexible and weaker than
all other models. This model is conservative for use in design of new buildings but
is not recommended for performance evaluation of existing or damaged buildings.
2. Models which include the interior gravity frames had smaller drift demands and
greater drift capacities than those without. The effects of including composite
560
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564
Table 10
Spectral acceleration capacities for the 9-story building for ordinary ground motions (left) and near fault ground motions (right)

M1-WO M2-WO M2-SC M2-Comp M2-WO-fr M2-SC-fr M1-WO M2-WO M2-SC M2- M2-WO-fr M2-SC-fr
Comp

Eq. 1 0.60 0.60 0.60 0.80 0.40 0.50 If21 1.60 1.60 1.80 1.80 0.50 0.60
Eq. 2 1.00 1.00 1.20 1.20 0.90 0.90 If23 1.20 1.40 1.40 1.40 0.70 0.60
Eq. 3 1.00 1.00 1.00 1.00 0.90 0.80 If25 0.80 1.00 1.00 1.00 0.50 0.40
Eq. 4 2.00 2.00 2.00 2.20 0.80 0.80 If27 1.60 1.80 2.00 2.20 0.90 0.90
Eq. 5 1.40 1.40 1.40 1.60 0.80 0.80 If29 1.60 1.60 1.80 1.80 0.80 0.70
Eq. 6 1.20 1.40 1.40 1.40 0.90 1.00 If31 1.60 1.80 2.00 2.00 0.80 0.80
Eq. 7 0.80 0.80 0.80 0.80 0.40 0.40 If33 1.40 1.60 1.60 1.80 0.80 0.80
Eq. 8 0.80 1.00 1.00 1.00 0.60 0.80 If35 0.80 0.80 1.00 1.00 0.50 0.40
Eq. 9 3.60 4.00 4.60 4.60 1.30 1.40 If37 1.00 1.00 0.80 1.20 0.30 0.30
Eq. 10 1.80 2.00 2.00 2.20 1.00 1.00 If39 2.00 2.20 2.60 2.60 0.80 0.80
δ 0.57 0.59 0.65 0.62 0.33 0.32 δ 0.29 0.29 0.34 0.31 0.29 0.32
Xm 1.20 1.27 1.30 1.39 0.76 0.80 Xm 1.31 1.42 1.51 1.60 0.63 0.60
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564
Table 11
Spectral acceleration capacities for the 20-story building for ordinary ground motions (left) and near fault ground motions (right)

M1-WO M2-WO M2-SC M2-Comp M2-WO-fr M2-SC-fr M1-WO M2-WO M2-SC M2- M2-WO-fr M2-SC-fr
Comp

Eq. 1 0.60 0.40 0.60 0.60 0.10 0.50 If21 1.00 0.60 1.20 1.20 0.20 0.80
Eq. 2 0.80 0.20 0.80 1.00 0.10 0.70 If23 0.80 0.80 0.80 0.80 0.40 0.50
Eq. 3 0.60 0.40 0.60 0.80 0.08 0.60 If25 0.40 0.60 0.60 0.60 0.40 0.40
Eq. 4 1.40 1.00 1.80 1.80 0.20 0.90 If27 0.80 0.80 0.80 0.80 0.10 0.60
Eq. 5 1.60 1.20 2.00 2.00 0.40 0.90 If29 0.60 0.60 0.80 0.80 0.30 0.40
Eq. 6 0.80 0.60 1.00 1.00 0.30 0.80 If31 0.80 0.80 1.00 1.00 0.50 0.60
Eq. 7 0.20 0.20 0.40 0.40 0.08 0.40 If33 0.60 0.60 0.80 0.80 0.40 0.30
Eq. 8 0.80 0.40 0.80 0.80 0.10 0.70 If35 0.40 0.40 0.40 0.40 0.20 0.20
Eq. 9 2.80 2.40 3.20 3.20 0.70 1.50 If37 0.40 0.60 0.60 0.60 0.08 0.40
Eq. 10 1.00 1.00 1.20 1.20 0.20 0.70 If39 0.80 0.80 1.00 1.00 0.30 0.50
δ 0.62 0.74 0.63 0.60 0.75 0.38 δ 0.31 0.20 0.28 0.28 0.46 0.35
Xm 0.87 0.59 1.02 1.07 0.17 0.72 Xm 0.63 0.65 0.77 0.77 0.26 0.44

561
562 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

Fig. 30. Spectral acceleration capacities for the 9-story building for ordinary ground motions (left) and
near fault ground motions (right).

Fig. 31. Spectral acceleration capacities for the 20-story building for ordinary ground motions (left) and
near fault ground motions (right).

action in the gravity frame model were not significant. Therefore, it is rec-
ommended that the columns in the gravity frames be included for evaluation and
performance prediction of steel moment frame buildings.
3. The frame models that had fracturing connections had significantly larger drift
demands and smaller drift capacities than those with ductile connections. This is
true for both static and dynamic analyses.
4. The effects of the gravity frames had significantly greater effect for frames with
pre-Northridge connections. In many cases they made the difference between col-
lapse or survival. It is highly recommended that the gravity frames be included
in models used for evaluation of existing or damaged pre-Northridge buildings.
5. Several of the models with pre-Northridge connections collapsed for some acceler-
ograms that represented the 2% in 50 year hazard level. So, it should not be
surprising if some existing steel moment frame buildings collapse if subjected to
comparable ground motions.
D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564 563

Acknowledgements

This research was funded by the Federal Emergency Management Agency through
a grant to the SAC Joint Venture. The SAC Joint Venture is composed of the Struc-
tural Engineers Association of California (SEAOC), the Applied Technology Council
(ATC) and the California Universities for Research in Earthquake Engineering
(CUREE). This support is gratefully acknowledged. Any results, findings and con-
clusions are solely those of the authors and do not necessarily represent those of
the sponsors.

References

[1] Foutch DA, Shi S. Connection element (type 10) for DRAIN-2DX. Internal Report. Urbana (IL):
University of Illinois at Urbana-Champaign, 1997.
[2] Somerville P, Smith N, Puntamurthula S, Sun J. Development of ground motion time histories
for phase 2 of the FEMA/SAC steel project. Background Document, Report No. SAC/BD-97/04,
October 1997.
[3] Krawinkler H. The state-of-the-art report on system performance of moment resisting steel frames
subjected to earthquake ground shaking. FEMA 355c. Washington, DC: Federal Emergency Manage-
ment Agency, 2000.
[4] Lee K, Foutch DA. Performance prediction and evaluation of steel moment frames for seismic loads.
SAC Background Document SAC/BD-00/25. Richmond, CA: SAC Joint Venture, 2000.
[5] Barsom JM, Frank K. State of art report on materials and fracture. FEMA 355a. Washington, DC:
Federal Emergency Management Agency, 2000.
[6] Roeder C. The state-of-the-art report on connection performance. FEMA 355d. Washington, DC:
Federal Emergency Management Agency, 2000.
[7] Krawinkler H, Bertero VV, Popov EP. Inelastic behavior of steel beam-to-column subassemblages.
Report No. UCB/EERC-71/7, University of CA at Berkeley: Earthquake Engineering Research
Center (EERC), 1971.
[8] Krawinkler H, Mohasseb S. Effect of panel zone deformations on seismic response. J Constr Steel
Res 1987;8:233–50.
[9] Venti M, Engelhardt MD. Brief report of steel moment connection test, specimen DBBW (Dog
Bone-Bolted Web). Internal SAC phase 2 report, 1999.
[10] Liu J, Astaneh-Asl A. Cyclic tests on simple connections including effects of the slab. SAC Report
No. SAC/BD-00/03. 2000.
[11] Yun SY, Foutch DA. Performance prediction and evaluation of low ductility steel moment frames
for seismic loads. SAC Background Document SAC/BD-00/26, Richmond, CA: SAC Joint Ven-
ture. 2000.
[12] Luco N, Cornell CA. Effects of random connection fractures on the demands and reliability for a
3-story pre-Northridge SMRF structure. In: Proceedings of the 6th U.S. Nat. Conf. on Earthquake
Engineering, Seattle, WA, May 31–June 4, 1998. Oakland, CA: Earthquake Engineering Research
Institute; 1998.
[13] Goel SC, Stojadinovic B. Test results of Pre-Northridge, simulated filed welding, weld fracture,
small plastic rotation. SAC test results. Ann Arbor, MI: University of Michigan, 1999.
[14] Barsom JM. Failure analysis of welded beam to column connections. SAC Report No. SAC/BD-
99/23, 1999.
[15] Foutch DA. The state-of-the-art report on performance prediction and evaluation of moment resisting
steel frame structures. FEMA 355f, Washington, DC: Federal Emergency Management Agency,
2000.
[16] Luco N, Cornell CA. Seismic drift demands for two SMRF structures with brittle connections. In:
564 D.A. Foutch, S.-Y. Yun / Journal of Constructional Steel Research 58 (2002) 529–564

Proceedings of the 6th U.S. Nat. Conf. on Earthquake Engineering, Seattle, WA, May 31–June 4,
1998. Oakland, CA: Earthquake Engineering Research Institute; 1998.
[17] Shi S. Evaluation of connection fracture and hysteresis type on the seismic response of steel build-
ings. PhD dissertation, University of Illinois, October 1997.

También podría gustarte