Anda di halaman 1dari 10

Effect of Beam Growth on Reinforced Concrete Frames

Jubum Kim1; John Stanton2; and Gregory MacRae3

Abstract: In well-designed reinforced concrete frames subjected to earthquake shaking, flexural yielding and cracking occur at the ends
of the beams as they undergo large lateral displacements. As the cracks form, the horizontal distance between column centerlines
increases. This is known as ‘‘beam growth’’ and it causes an increase in member demands that is not generally taken into account in
Downloaded from ascelibrary.org by Northwestern University Library on 03/25/15. Copyright ASCE. For personal use only; all rights reserved.

analysis used for building design. This paper describes the mechanism of beam growth, observations of it from previous experimental
studies, the development and calibration of analytical models considering and ignoring beam growth, and static and dynamic inelastic
frame analyses to quantify beam growth effects. Changes in beam axial forces, beam flexural strength, and column moments and shears
are quantified for several frame configurations. It is shown that the column demand due to beam growth tends to be greatest at the first
story level and this demand tends to be most severe in frames with deep beams and with many bays. For the four-, eight-, and ten-bay
frames used in the study, pushover analyses showed increases in column shear demands due to beam growth of 52, 76, and 86%,
respectively. Dynamic loading to design-level earthquake records caused a 23% increase in the four-bay reference frame.
DOI: 10.1061/共ASCE兲0733-9445共2004兲130:9共1333兲
CE Database subject headings: Beams; Concrete, reinforced; Building frames; Seismic response; Seismic design.

Introduction Fenwick and Davidson 1995兲, and with multiple columns 共e.g.,
Zerbe and Durrani 1989; Sakata and Wada 1992; Fenwick and
In seismic regions of the United States, reinforced concrete mo- Megget 1993; Restrepo-Posada et al. 1993; McBride et al. 1996兲.
ment frames are often used to resist the lateral loads on buildings. Some tests have also included slabs 共e.g., Zerbe and Durrani
Such frames are almost invariably analyzed using models in 1990; Qi and Pantazopoulou 1991; McBride et al. 1996; Lau and
which the beam is represented by a flexural line element. The Fenwick 2002兲. Fenwick and Davidson 共1995兲 have shown that
computed axial forces in these beams are often low, while their for beams with flexural hinges subject to cyclic loading, particles
axial stiffnesses are high, so the change in beam length is negli- of crushed concrete may fall into the cracks, preventing their
gible. However, in a real frame, the beams start to crack as the closing fully, and causing permanent elongation.
frame moves laterally, causing the distance between column cen- Fenwick and Davidson 共1995兲 proposed a simple analytical
terlines to increase as shown in Fig. 1. There, flexural cracking is model for beam growth, as shown in Fig. 3. At the end of each
modeled as being concentrated at the beam ends and the distance beam, and at potential hinge locations along the span, a composite
between column centerlines increases by the sum of the crack element was placed that consisted of one concrete element and
widths at mid-height of the beam. This phenomenon is known as one steel element on the top and bottom of the section. A six-
‘‘beam growth’’ 共e.g., Fenwick and Fong 1979; Fenwick and story, three-bay frame was designed for the approximate equiva-
Davidson 1995兲. Greater beam growth occurs with greater beam lent of UBC Seismic Zone 4 共ICBO 1997兲 and was analyzed with
depths and story drift ratios. In a frame subject to large lateral and without the beam growth elements. The spans were relatively
displacements, beam growth effects push the exterior columns long and gravity moments in the beam were comparable to the
outwards, causing an increase in column curvature demand on seismic moments, so positive moment hinges formed within the
one side of the frame, but a reduction on the other, as shown in span. The hinges yielded only in one direction, and the exterior
Fig. 2. columns were pushed outwards, especially in the lower stories.
The existence of beam growth has been confirmed experimen- The frame deformations increased progressively with cycling.
tally. Beam–column subassemblage tests have been carried out on This behavior is particularly undesirable, but can be shown to
specimens with a single column 共e.g., Zerbe and Durrani 1989; occur only when gravity moments are large.
In the United States, the frames of most structures in seismic
1
Assistant Professor, Dept. of Civil and Environmental Engineering, regions have deep beams that carry relatively little gravity load,
Clarkson Univ., Potsdam, NY 13699.
2
Professor, Dept. of Civil and Environmental Engineering, Univ. of
Washington, Seattle, WA 98195-2700.
3
Associate Professor, Dept. of Civil and Environmental Engineering,
Univ. of Washington, Seattle, WA 98195-2700.
Note. Associate Editor: Yan Xiao. Discussion open until February 1,
2005. Separate discussions must be submitted for individual papers. To
extend the closing date by one month, a written request must be filed with
the ASCE Managing Editor. The manuscript for this paper was submitted
for review and possible publication on December 2, 2002; approved on
September 23, 2003. This paper is part of the Journal of Structural
Fig. 1. Effects of beam growth: 共a兲 initial condition and 共b兲 idealized
Engineering, Vol. 130, No. 9, September 1, 2004. ©ASCE, ISSN 0733-
displaced shape
9445/2004/9-1333–1342/$18.00.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2004 / 1333

J. Struct. Eng. 2004.130:1333-1342.


Fig. 2. Frame sway including beam growth

so flexural hinges are likely to occur only at the beam ends. Each
hinge is likely to rotate in both directions, so beam growth will
Downloaded from ascelibrary.org by Northwestern University Library on 03/25/15. Copyright ASCE. For personal use only; all rights reserved.

not accumulate as found by Fenwick and Davidson. It will there-


fore be lower. However beam growth effects may still cause large
increases in the forces induced in the columns and joints. Studies
to quantify these changes in demands have not been performed.
There is therefore a need to establish the likely effects of beam
growth on typical reinforced concrete 共RC兲 seismic frames in the
United States, so that their consequences can be evaluated.
In this paper, an analytical model is developed to represent
beam growth. It is calibrated against experimental results and
then used to quantify beam growth effects in a set of RC moment
frames, each with different characteristics. From frame static
pushover and dynamic analyses, the frame behavior is investi-
gated and column demands, beam demands, and joint demands
are quantified.

Fig. 4. Beam–column interface models: 共a兲 reinforced concrete


Analytical Modeling and Verification frame including beam growth model and 共b兲 reinforced concrete
frame excluding beam growth model
The computer model shown in Fig. 4共a兲 was used to simulate the
beam–column joint region. A similar model had originally been
developed to represent the gap opening and beam growth behav-
ior at beam–column joints in precast systems in which the rota- simulating the behavior of cracked concrete and the beam growth
tions are concentrated in a single crack at the interface 共Kim that accompanies cracking. As the beam rocks relative to the col-
2002兲, and it was modified for use in this study. In this reinforced umn, one face 共top or bottom兲 lifts off, while the other experi-
concrete frame including beam growth 共RFIBG兲 model, the open- ences compression. The model is thus capable of capturing beam
ing at each beam–column interface represents the opening from growth at the interface as the frame sways. Although in the inter-
all of the nearby cracks in the beam. The model was constructed ests of clarity, the figure shows only two nodal pairs on each side
in DRAIN-2DX 共Prakash et al. 1993兲 and it uses elements from of the beam, the model uses nine gap elements, equally spaced, in
the standard library. order to replicate the gradual lift-off that occurs in practice. The
Nodes c1-7 and b1-8 are associated with the column and model differs in this regard from Fenwick and Davidson’s two-
beam, respectively. In the column, Node c7 acts as a master node, strut model 共1995兲. Shear is transferred across the interface by
and Nodes c1–c6 are slaved to it so that all seven move together providing very stiff vertical springs at Nodes b5–c5 and b6 –c6,
as a rigid body. Beam Nodes b2 and b4 are slaved to b6 in the while the horizontal displacements and the rotations of the nodes
same way. Joint deformations are not modeled. Moments are
transferred between the beam and column by horizontal tension
and compression between node pairs such as b2–c2. Two parallel
sets of elements connect the nodes in each pair. One is an inelastic
truss element that simulates the reinforcement and resists tension
or compression. The other is a gap element that has inelastic
properties in compression but has no tension strength, thereby

Fig. 5. Test configuration of indeterminate frame 共Zerbe and Durrani


Fig. 3. Fenwick and Davidson’s 共1995兲 two-strut model 1989兲

1334 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2004

J. Struct. Eng. 2004.130:1333-1342.


Downloaded from ascelibrary.org by Northwestern University Library on 03/25/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Measured and computed force–displacement relationships of Fig. 7. Measured and computed beam growth for Zerbe and Durra-
Zerbe and Durrani’s frames ni’s indeterminate frame

are independent. The beam and column elements are assumed to


remain elastic, while the column base is assumed to be elasto- the RFEBG model, the frame resistance at any displacement was
plastic. twice the resistance of the one exterior specimen plus the resis-
A second, conventional model was also developed for the tance of the interior specimen, thereby representing simply the
study, to provide comparable response but without the beam sum of the strengths of the individual parts of the frame. In the
growth. It is shown in Fig. 4共b兲 and was called the reinforced test specimen, the beam growth is partially restrained, compres-
concrete frame excluding beam growth 共RFEBG兲 model. Its pri- sive forces are induced in the beams, which therefore become
mary feature is an inelastic rotational spring at each end of the stronger in flexure, and the frame resistance increases. The
beam element. The spring’s properties were chosen to have the RFIBG model replicates this behavior, and matches the measured
same moment-rotation characteristics as the whole joint in the base shear within 4% at all drifts. The RFEBG model follows the
RFIBG model when there is no axial force in the beam, as is main trends of the test data of the sum of the determinate frames.
typically assumed in structural analysis. The moment capacity However, because its inelastic action was modeled by bilinear
was thus assumed to be independent of axial force. This is differ- 共rotational兲 elements, whereas the test data was curvilinear, the
ent from the RFIBG model in which the axial force–moment match was less accurate than between the RFIBG model predic-
interaction is included. tions and the indeterminate frame response. The flexibility of the
The RFEBG and RFIBG models were verified against Zerbe test rig was accounted for in the comparisons.
and Durrani’s 共1989兲 test data, because test details were readily The computed beam growth values from the RFIBG model are
available. Those authors tested the indeterminate two-bay frame compared with the measured ones in Fig. 7. The model replicates
shown in Fig. 5 and also two statically determinate subassem- the primary features of the measured response. During the initial
blages that were identical to the interior and exterior beam– section of the curve, the gaps at the beam–column interface are
column joints of the indeterminate frame. Both analytical models closed, essentially no beam growth occurs, and drift is accommo-
were used to represent the indeterminate frame. dated exclusively through beam and column deformation. Once
The element properties were directly related to the physical the gaps start to open, beam growth is approximately linearly
properties of the system. Priestley’s empirical equation 共Priestley related to drift.
et al. 1996兲 was used for the plastic hinge length of the beam. The
truss element for the reinforcing steel was bilinear and its prop-
erties were selected to provide the appropriate force– Frames Designed
displacement characteristics. The strain-hardening ratio of the
steel was 0.02. The properties of the concrete gap elements were A suite of 15 five-story buildings was designed and analyzed in
assigned based on the plastic hinge length and an elasto–perfectly order to evaluate the effect of beam growth and its dependence on
plastic stress–strain curve with a ‘‘yield’’ strain of 0.003 at the the number of bays and the beam depth. The number of bays was
compressive strength f c⬘ . Cumulative cyclic effects were not ac- 2, 4, 6, 8, or 10 and beam depths were 914, 1,219, and 1,524 mm
counted for in the model. 共36, 48, and 60 in.兲 as summarized in Table 1. Each frame was
The measured and computed force–displacement responses named according to the number of stories and bays, and the beam
are compared in Fig. 6, where the measured values represented depth. For example, the reference frame, RF0504-48 shown in
the envelope of cyclic loading. Since measured and computed Fig. 8, was a five-story, four-bay frame, with 48 in. 共1219 mm兲
values were similar during cyclic loading, the effects of crushed deep beams over the height. The bay width and story height were
concrete falling into the cracks appeared not to be significant. In assumed to be the same for all the frames.

Table 1. Reinforced Concrete Moment Frames Considered


Number of bays 2 4 6 8 10
h b ⫽914 mm 共36 in.兲 RF0502-36 RF0504-36 RF0506-36 RF0508-36 RF0510-36
h b ⫽1,219 mm 共48 in.兲 RF0502-48 RF0504-48 RF0506-48 RF0508-48 RF0510-48
h b ⫽1,524 mm 共60 in.兲 RF0502-60 RF0504-60 RF0506-60 RF0508-60 RF0510-60

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2004 / 1335

J. Struct. Eng. 2004.130:1333-1342.


Table 2. Member Properties for Frame RF0504-48
Property Value
Column
Size 914⫻914 mm2
Compressive strength, f c⬘ 55 MPa
Assumed moment of inertia 0.60I g a
Yield moment at base 3,390 kN m
Beam
Width⫻depth 609⫻1219 mm2
Compressive strength, f c⬘ 34 MPa
Assumed moment of inertiab 0.30I g a
Downloaded from ascelibrary.org by Northwestern University Library on 03/25/15. Copyright ASCE. For personal use only; all rights reserved.

Mild steel barsb,c 8-D30 共8-#9兲


Elastic modulus, E s 200 GPa
Yield strength, f sy 469 MPa
Strain-hardening ratio 0.02
Fig. 8. Five-story, four-bay reference frame 共RF0504-48兲
Plastic hinge length, L ph 533 mm
Distance to the cover 89 mm
Yield momentb 2,576 kN m
The prototype structures were designed for zones of high seis- a
Gross sectional moment of inertia.
micity 共Seismic Zone 4兲 in accordance to the UBC 共ICBO 1997兲 b
Average value, equal to value at third story.
assuming standard occupancy, type D 共stiff兲 soil profiles and no c
Each at top and bottom.
near-fault effects. For the four-bay structure, the seismic mass at
each floor was assumed to be 910 t 共2,000 kips兲. For frames with
different numbers of bays, the story mass was linearly related to yielding at the base at about 0.5% roof drift ratio. At 2% roof drift
the number of bays. The same size members were used over the ratio, the base shear predicted by the RFIBG model was 3%
frame height. The columns were elastic, except at the base. higher than that of the RFEBG model.
As part of the design, the frame was initially analyzed with Both models predicted the same initial deformations, but the
UBC equivalent static lateral forces ignoring beam growth ef- magnitude and distribution of the displacements began to differ
fects. This caused beam flexural demands to vary approximately after cracking had started. Fig. 10 shows the horizontal displace-
linearly in the ratio of 3:1 from the bottom to the top of the ments at all floors and at all columns at 2% roof drift ratio. The
building. Design moment capacities associated with these de- first story drift ratio at the east end of the frame 共Column E兲 was
mands were provided. The stiffnesses of the beams were then 3.0%, while that at the west end 共Column A兲 was 1.5% in the
changed to vary in the same way as the moment demands, since it RFIBG model. At the same roof drift level, the first story drift
has been argued 共Paulay 2000; Priestley and Kowalsky 2000兲 that ratio in the RFEBG model was 2.4% for all columns and was
cracked beam flexural stiffness is proportional to strength. The close to that of the central column in the RFIBG model.
sum of the beam stiffnesses remained unchanged. The change in In the RFIBG model, the beams at the lower floors experi-
stiffness distribution did not affect the roof yield displacement. enced large axial forces as shown in Fig. 11, especially at the first
Details of the material and member properties used in the frame floor. These axial forces, and the beam moments, increased across
are summarized in Table 2. P-delta effects were not considered in the building from west to east, as the frame swayed to the east.
these analyses. It was assumed that slabs either did not exist, or The moment in the east bay beam increased by 24% due to the
were detailed in such a way as to have little effect on the frame restraint of beam growth. At the second floor, the east end beam
behavior. was in tension in the RFIBG model, thereby increasing the beam
growth at the second floor. The figure also indicated that the beam
moments at the upper floors differed little between the RFEBG
Frame Behavior and RFIBG models.

The behavior of the reference frame 共RF0504-48兲 is first de-


scribed to illustrate the general trends of behavior. The effects of
the number of bays and the beam depth are then discussed.

Static Pushover Analysis


A lateral load, inversely triangular and distributed in a ratio of 1:2
for the exterior and interior columns at each story, was applied to
the frame. It increased until the frame roof drift ratio reached 2%,
where roof drift ratio was defined as the displacement of the
center column at the top floor divided by the building height. Fig.
9 shows the base shear force versus roof drift ratio for both the
RFIBG and RFEBG models. Restraint of beam growth by the
columns increased the axial force and thus the resisting moments
Fig. 9. Base shear versus roof drift ratio for RF0504-48 under static
in the beams, which slightly increased the base shear needed to
loading
produce a given drift. This was evident once the columns started

1336 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2004

J. Struct. Eng. 2004.130:1333-1342.


Downloaded from ascelibrary.org by Northwestern University Library on 03/25/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Comparison of column displacements for RF0504-48 at 2%


roof drift ratio under static loading

Fig. 12. Column moment and shear for RF0504-48 at 2% roof drift
ratio under static loading: 共a兲 reinforced concrete frame including
beam growth model and 共b兲 reinforced concrete frame excluding
beam growth model 共units: kN, m兲

Column demands are compared in Fig. 12. The differences


between the predictions of the two models are most acute in the
first two stories, and are caused by both the expansion of the
building and the change in beam moments with axial force. The
maximum shear on the columns at the first story occurred in
Column C 共center兲 in the RFEBG model, whereas it was in Col-
umn E 共east兲 in the RFIBG model. Column E carried 26 and 17%
of the total base shear in the RFIBG and RFEBG models, respec-
tively, indicating that beam growth imposed a greater shear de-
mand on the outermost columns. The peak moment demand on
Column E also increased by 104% as a result of beam growth.
This large moment demand on the outermost columns is likely to
cause premature yielding that is not predicted by analyses that do
not account for beam growth.

Dynamic Analysis
Fig. 11. Beam moments and axial forces 共in parentheses兲 for The dynamic characteristics of the reference frame 共RF0504-48兲
RF0504-48 at 2% roof drift ratio under static loading: 共a兲 reinforced were studied using the nine earthquake motions discussed below.
concrete frame including beam growth model and 共b兲 reinforced con-
Horizontal floor mass was lumped equally at the mid-span of each
crete frame excluding beam growth model 共units: kN, m兲; tension is
beam. Small translational and rotational masses were placed at
positive
the beam–column interfaces to minimize spurious high accelera-

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2004 / 1337

J. Struct. Eng. 2004.130:1333-1342.


Table 3. Earthquake Ground Motions 共Somerville 1997兲
Probability of exceedence Name 共SAC name兲 Earthquake record Magnitude Scaling factor Peak ground acceleration 共g)
2% la02a 共LA21兲 1995 Kobe 6.9 1.15 1.28
in 50 years la02b 共LA24兲 1989 Loma Prieta 7.0 0.82 0.47
共la02 series兲 la02c 共LA26兲 1994 Northridge 6.7 1.29 0.94

10% la10a 共LA02兲 1940 Imperial Valley, El Centro 6.9 2.01 0.68
in 50 years la10a 共LA05兲 1979 Imperial Valley, Array #06 6.5 0.84 0.30
共la10 series兲 la10c 共LA18兲 1994 Northridge, Sylmar 6.7 0.99 0.82

50% la50a 共LA42兲 1979 Coyote Lake 5.7 2.28 0.33


in 50 years la50b 共LA47兲 1992 Landers 7.3 2.63 0.34
Downloaded from ascelibrary.org by Northwestern University Library on 03/25/15. Copyright ASCE. For personal use only; all rights reserved.

共la50 series兲 la50c 共LA57兲 1971 San Fernando 6.5 1.3 0.25

tion demands there 共Kim 2002兲. Rayleigh damping was used with second story, but are not shown. It may be seen that the beam
5% of critical damping in the first and the fifth modes. The stiff- growth at the first floor was less than that at the second floor for
ness component was proportional to initial stiffness. the same drift ratio, implying that the first floor beams elongate
The ground motions were obtained from the FEMA/SAC steel less and attract more axial force. This finding is consistent with
project 共Somerville 1997兲. Three sets of earthquake motions for the larger beam moment strengths, and consequently higher base
Los Angeles, each with a different probability of exceedance, shears, found in Fig. 14.
were chosen and they are summarized in Table 3. They were Fenwick and Davidson 共1995兲 have suggested that beam
developed by taking recorded ground motions and scaling their growth is proportional to beam depth h b and to the number of
amplitudes so that they matched, in a weighted average sense, a bays n b . A beam growth coefficient ␤ can be defined by
target spectrum at periods of 0.3, 1.0, 2.0, and 4.0 s. The scaling

factors and the peak ground accelerations after scaling are given ␤⫽ (1)
in Table 3 共Somerville 1997兲. The la02 series 共la02a, la02b, and n b h b 共 ␪⫺␪ o 兲
la02c兲, with a probability of exceedance of 2% in 50 years, may where ⌬⫽beam growth at a floor; ␪⫽story drift ratio; and ␪ o
be regarded as strong ground motions, and the la50 series as weak ⫽threshold drift ratio beyond which beam growth occurs
ground motions. The spectral accelerations of the moderate (⫽0.5%). The physical interpretation of beta is that ␤ multiplied
ground motions 共la10 series兲 were, on average, the closest match
to the UBC Zone 4 design spectrum, but significant differences
existed at individual periods 共Kim 2002兲.
Fig. 13 shows the envelope of maximum story drift ratios at
each floor. The two models gave similar story drifts in response to
small and moderate earthquakes 共la50 or la10 series兲 but the
RFIBG showed similar but slightly greater center column drifts
under the stronger la02 series ground motions. These story drift
ratios exceeded 2% in the lower stories. The computed column
forces followed the same pattern; they were similar in the two
models for the low and moderate earthquakes, but the RFIBG
model gave larger moments and shears under the stronger la02
ground motions than did the RFEBG model.
Fig. 14 shows the maximum base shear as a function of maxi-
mum first story drift ratio for the two models. This is done for
both static and dynamic loading. In the dynamic case, the maxi-
mum base shear and drift did not necessarily occur at the same
time. The fact that the maximum dynamic shears are different
from the static shears is a result of the dynamic load distribution
being different from the equivalent static load distribution. This
effect is commonly referred to as ‘‘dynamic magnification’’ of
forces 共Paulay and Priestley 1992兲. The RFIBG shears are con-
sistently greater than the RFEBG shears at first story drift ratios
greater than about 0.5%. The increase in base shear is mostly a
result of higher beam moments that are caused by the higher
beam axial forces induced by beam growth.
Beam growth is plotted against story drift in Fig. 15. In all the
analyses, the beam growth became significant at drift ratios
greater than 0.5% and thereafter increased almost linearly with
drift. For static pushover analyses, beam growth at the first floor
is plotted against the first story drift and the growth at the second
floor is plotted against the second story drift. For floors and sto-
Fig. 13. Maximum story drift ratios under earthquake loading
ries further up the structure, the lines were similar to that for the

1338 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2004

J. Struct. Eng. 2004.130:1333-1342.


Table 4. Average Ratios of Maximum Member Forces in Reinforced
Concrete Frame Including Beam Growth Model to Those in Rein-
forced Concrete Frame Excluding Beam Growth Model at First Story
for Frame RF0504-48 at 2% Roof Drift Ratio from Static Loading
Shear Moment
Loading Central Outermost Central Outermost
type column columns column columns
la02 series 1.19 1.54 1.28 1.91
la10 series 1.08 1.23 1.14 1.46
la50 series 0.98 1.03 0.98 1.03
Downloaded from ascelibrary.org by Northwestern University Library on 03/25/15. Copyright ASCE. For personal use only; all rights reserved.

story outer column increased by 23 and 46%, for shear and mo-
ment, respectively, under the moderate la10 series that approxi-
mately matched with the UBC design spectrum.

Fig. 14. Maximum base shear versus maximum first story drift ratio Effects of Beam Depth and Number of Bays
for static and dynamic analyses
The effects on beam growth of beam depth and the number of
bays were studied here by analyzing frames with the RFIBG
by beam height is approximately twice the distance between the model under static pushover analysis. The geometric effects are
neutral axis and the mid-height of the beam. Note that the drift discussed first, followed by an evaluation of the forces.
ratios are defined here at the center column. The data from Fig. 15 Fig. 16共a兲 shows the value of ␤ in Eq. 共1兲 plotted against
for the first floor under static conditions suggest a value of ap- number of bays n b for conditions at the first floor when the roof
proximately 2/3 for the coefficient ␤. drift ratio was 2%. If the beam growth were directly proportional
For the dynamic analyses, the peak beam growth at each floor to n b and h b , all the points would have lain on a single, horizon-
is plotted against the maximum story drift at the same floor. It tal line on the plot. As the number of bays increases, however, the
generally occurred at the second floor level for the same reasons beams experience increasingly higher compression forces and
that the static growth is greater at this level. In almost all cases, elongate less. Consequently, ␤ decreases with n b . Fig. 16共b兲
and particularly at high drift ratios, the peak dynamic beam
growths are smaller than the comparable static growths. The
smaller beam growth leads to larger axial forces and moment
strengths in the beams, which is in agreement with the trend
shown in Fig. 14 where base shears are larger under dynamic than
static conditions.
The two models led to different shear and moment distribu-
tions among the columns. The largest shears and moments usually
occurred at the central column in the RFEBG model, whereas
they occurred at the outermost column in the RFIBG model. The
ratios of maximum shears in the RFIBG model to those in the
RFEBG model at the central and outermost columns at the first
story are summarized in Table 4, where they are averaged for the
la02, la10, and la50 series, respectively. The demands on the first

Fig. 16. Normalized beam growth at first story at 2% roof drift ratio
Fig. 15. Maximum beam growth versus maximum story drift ratio at under static loading against: 共a兲 number of bays n b and 共b兲 beam
all stories for static and dynamic analyses depth h b

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2004 / 1339

J. Struct. Eng. 2004.130:1333-1342.


Downloaded from ascelibrary.org by Northwestern University Library on 03/25/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 17. Base shear at 2% roof drift ratio under static loading

shows the same data, plotted against the beam depth h b , as the
independent variable. It demonstrates that ␤ increases with h b ,
rather than remains constant. This is because the compressed re-
gion in contact with the column occupies a smaller fraction of the
total beam area, as the beam depth increases. If the axial force
remained constant, the compressed area would remain constant.
However, the axial force increases somewhat with beam depth,
but it increases less rapidly than does the beam depth itself, so the
result is still that the neutral axis depth becomes a smaller fraction
of the beam depth.
The base shear also increases with both beam depth and num-
Fig. 19. Distribution of first story column shear at 2% roof drift ratio
ber of bays, as illustrated in Fig. 17. For the frames with 1219 under static loading: 共a兲 frames with different number of bays and 共b兲
mm 共48 in.兲 deep beams, the base shear forces in the RFIBG frames with different beam depths
model were then divided by those in the RFEBG model, as shown
in Fig. 18. The increase in strength ratios can therefore be attrib-
uted solely to beam growth effects; the restraint in beam growth column shears for the frames with different beam depths. In the
induces larger compressive forces and higher flexural strength in RFIBG model, the shear was always highest at the outer east
the beams, and thus larger base shears. column except for the RF0504-36 frame, in which the moment in
The shear forces at the first story columns were almost con- the interior column was higher. As the beam depth decreased, so
stant for the RFEBG model but they varied almost linearly from did the beam growth, and the frame behaved more like the con-
west to east in the RFIBG model. Fig. 19共a兲 shows that, while the ventional RFEBG model, wherein the column shear is larger at
minimum column shear, which occurred in the west columns, was the interior columns.
almost the same in all frames, the maximum value on the east of
the frame increased with the number of bays. This is consistent
with the finding in Fig. 18 that the average column shear in- Discussion
creased for more bays due to the restraint of beam growth. A
comparison of shear forces at the east end columns from the In addition to the effects that beam growth has on the frame
RFIBG model to those from the RFEBG model shows increases deformation, and on the beam and column demands, it may also
due to beam growth of 40, 52, 64, 76, and 86% for the two-, four-, have other effects as discussed below.
six-, eight-, and ten-bay frames, respectively. Fig. 19共b兲 shows the The additional axial force in the beam that results from beam
growth increases the joint shear demand, and may induce prema-
ture joint shear failure. This is particularly true at the lower floors.
Joint shear at column D under static loading was investigated,
since the beam axial forces and moments were largest there. The
joint shear force in RF0504-48 at 2% roof drift was found to be
6,672 kN as illustrated in Fig. 20共b兲, when the beam growth was
not accounted for. However, it increased to 8,580 kN as the ef-
fects of beam growth were considered in Fig. 20共a兲. The increase
in joint shear demand was therefore 29%. Joint shear often con-
trols the column size, so this finding is important.
Beam growth effects are not likely to directly cause a soft-
story mechanism because increases in column shear in different
columns act in different directions. However, due to increased
column flexural demands, exterior column hinging may occur as
the frame sways back and forth. As the strength of these columns
Fig. 18. Normalized base shear at 2% roof drift ratio under static
degrades, interior columns may be required to sustain greater de-
loading
mands, possibly leading to undesirable performance. Further stud-

1340 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2004

J. Struct. Eng. 2004.130:1333-1342.


lower floors, where the story drift ratios are larger. It be-
comes important after the drift reaches some threshold value,
which was a roof drift ratio of approximately 0.5% in this
study. The average beam growth at each bay in the four-bay
reference frame was approximately 2/3 of the beam depth
times the drift.
3. Inclusion of beam growth in the analysis significantly
changes the distribution of moments and shears throughout
the frame. Peak joint shears also increase. The column mo-
ments and shears increase most in the columns on the lead-
ing side of the frame at the first story level. At 2% roof drift
under the static pushover analysis, the increases in column
Downloaded from ascelibrary.org by Northwestern University Library on 03/25/15. Copyright ASCE. For personal use only; all rights reserved.

shear and joint shear were 52 and 29%, respectively, on the


outer columns of the four-bay reference frame. In the ten-bay
frame the column shear increased by 86%.
4. If beam growth is ignored in the frame design, then demands
may be underestimated, possibly resulting in undesirable
frame behavior and local damage.
5. Frames in which slabs restrain beam growth require further
study.

References

Fenwick, R., and Davidson, B. 共1995兲. ‘‘Elongation in ductile seismic-


resistant reinforced concrete frames.’’ Proc., Tom Paulay Symp.,
Farmington Hills, Mich., American Concrete Institute, SP 157-7,
Fig. 20. Forces contributing to joint shear demands on column D of 143–170.
RF0504-48 at 2% roof drift ratio under static loading: 共a兲 reinforced Fenwick, R., and Fong, A. 共1979兲. ‘‘The behavior of reinforced concrete
concrete frame including beam growth model and 共b兲 reinforced con- beams under cyclic loading.’’ Research Rep. No. 176, Dept. of Civil
crete frame excluding beam growth model Engineering, Univ. of Auckland, Auckland, New Zealand.
Fenwick, R. C., and Megget, L. M. 共1993兲. ‘‘Elongation and load deflec-
tion characteristics of reinforced concrete members containing plastic
hinges.’’ Bull. NZ National Soc. Earthquake Eng., 26共1兲, 28 – 41.
International Conference of Building Officials 共ICBO兲. 共1997兲. Uniform
ies, in which column yielding and strength degradation can occur
building code, Whittier, Calif.
above the base, may be useful in quantifying this effect.
Kim, J. 共2002兲. ‘‘Behavior of hybrid frames under seismic loading.’’ PhD
For frames in which slab effects are significant, the behavior
thesis, Univ. of Washington, Seattle.
will be different from that described above. This is because, while Lau, D., and Fenwick, R. 共2002兲. ‘‘The influence of precast prestressed
the slab is intact, it restricts gap opening at the beam ends. Lau flooring components on the seismic performance of reinforced con-
and Fenwick 共2002兲 have shown that after a number of loading crete perimeter frames.’’ SESOC J., 14共2兲, 17–26.
cycles, cracking and deterioration of the slab adjacent to the col- McBride, A. P., Fenwick, R. C., and Davidson, B. J. 共1996兲. ‘‘The influ-
umn occur such that the expected beam growth does occur. How- ence of slabs on the lateral cyclic behaviour of ductile concrete
ever, further studies are required to determine the increase in col- frames.’’ Research Rep. No. 566, School of Engineering, Univ. of
umn moment demand in a frame with a slab as a result of gap Auckland, Auckland, New Zealand.
opening. When precast floor systems are used in which the floor Paulay, T. 共2000兲. ‘‘A review of some structural principles relevant to a
panels run parallel to the perimeter frame beams, beam growth seismic design strategy for ductile systems.’’ Presented to the Canter-
may cause the units to slide off their seating if the seat length is bury Structures Group, New Zealand.
not sufficient. Paulay, T., and Priestley, M. J. N. 共1992兲. Seismic design of reinforced
concrete and masonry buildings, Wiley, New York.
Prakash, V., Powell, G. H., and Campbell, S. 共1993兲. ‘‘DRAIN-2DX base
Conclusions program description and user guide: Version 1.10.’’ Rep. No. UCB/
SEMM-93/17&18, Dept. of Civil Engineering, Univ. of California,
Berkeley, Calif.
Analyses of a number of RC frames were carried out to evaluate
Priestley, M. J. N., and Kowalsky, M. J. 共2000兲. ‘‘Direct displacement-
the effects of beam growth on the behavior. Major findings are as
based seismic design of concrete buildings.’’ Bull. NZ Soc. Earth-
follows: quake Eng., 33共4兲, 421– 444.
1. The analytical connection model, in which a series of contact Priestley, M. J. N., Seible, F., and Calvi, G. M. 共1996兲. Seismic design
elements was used to model the concrete behavior at the and retrofit of bridges, Wiley, New York.
beam–column interface, reproduced the beam growth behav- Qi, X., and Pantazopoulou, S. J. 共1991兲. ‘‘Response of RC frame under
ior found in experiments by others. This model provides a lateral loads.’’ J. Struct. Eng., 117共4兲, 1167–1188.
link between axial and rotational behavior at the beam– Restrepo-Posada, J. I., Park, R., and Buchanan, A. H. 共1993兲. ‘‘Behaviour
column interface that is not present in conventional analyses. of connections between precast concrete elements.’’ Civil Engineering
2. Beam growth increases with greater beam depth, number of Research Rep. No. 93-3, Univ. of Canterbury, Canterbury, New
bays in the frame, and story drift ratio. It is larger at the Zealand.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2004 / 1341

J. Struct. Eng. 2004.130:1333-1342.


Sakata, H., and Wada, A. 共1992兲. ‘‘Elasto-plastic behavior of one- Zerbe, H. E., and Durrani, A. J. 共1989兲. ‘‘Seismic response of connections
twentieth scale RC frame.’’ Proc., 10th World Conf. Earthquake En- in two-bay R/C frame subassemblies.’’ J. Struct. Eng., 115共11兲, 2829–
gineering, Vol. 6, Balkema, Rotterdam, The Netherlands, 3335–3340. 2844.
Somerville, P. G. 共1997兲. ‘‘Development of ground motion time histories Zerbe, H. E., and Durrani, A. J. 共1990兲. ‘‘Seismic response of connections
for phase 2 of the FEMA/SAC steel project.’’ Technical Rep. SAC in two-bay reinforced concrete frame subassemblies with a floor
(Series) No. BD-97/04, SAC Joint Venture, Sacramento, Calif. slab.’’ ACI Struct. J., 87共4兲, 406 – 415.
Downloaded from ascelibrary.org by Northwestern University Library on 03/25/15. Copyright ASCE. For personal use only; all rights reserved.

1342 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / SEPTEMBER 2004

J. Struct. Eng. 2004.130:1333-1342.

Anda mungkin juga menyukai