Vous êtes sur la page 1sur 8

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 106-S55

Progressive Collapse Resistance of Axially-Restrained


Frame Beams
by Youpo Su, Ying Tian, and Xiaosheng Song

Twelve specimens representing reinforced concrete frame beams simulated the removal of the middle column at the first story.
were tested to investigate their gravity load-carrying capacity Beam catenary action was identified in the test, and the failure
against progressive collapse. In these tests, the beams within the of the specimen was controlled by the rupture of beam longi-
frame subassemblies were restrained longitudinally against axial tudinal reinforcement. Orton11 studied the technique of
deformation. The tests indicated that the compressive arch action
due to longitudinal restraint can significantly enhance the flexural
using carbon fiber-reinforced polymer to strengthen beams
strength of a beam subjected to vertical loads. The compressive without continuity of bottom reinforcement that makes a
arch action was observed to be a function of flexural reinforcement reinforced concrete frame vulnerable to progressive collapse.
ratio and ratio of beam span to depth. The test results validated an Beams are critical structural components of a reinforced
analytical model that has considered the axial restraining effects concrete frame to resist progressive collapse when a bearing
on beam loading capacity. The application of compressive arch column fails. The two collinear beams framing into the failed
effect to the prevention of progressive collapse is discussed. column become a single structural component resisting the
redistributed gravity load that causes significantly increased
Keywords: beam(s); catenary action; compressive arch action; progressive bending moment and shear demands in the beam. Unless
collapse; reinforced concrete. sufficient strength is provided, the two-bay beam spanning
over the failed column cannot retain the local failure, and a
INTRODUCTION progressive collapse may result. Note that the pure bending
Abnormal loading events such as explosions, vehicle capacity that generally fits the test data of simply supported
collisions, and foundation failure are not considered in an beams has been used to define the flexural strength of a
ordinary structural design. The local damage caused by such reinforced concrete beam in the routine design and analysis
low-probability loads, however, may result in a chain reaction of structures. Under gravity loading, however, compressive
of structural element failure that leads to partial or even full arch action and tensile catenary action develop in a reinforced
collapse of a structure. The collapses of the Ronan Point concrete flexural member that is restrained axially at the
Tower in London in 1968 and the Alfred P. Murrah Federal boundaries, as is the case with frame beams. Compressive
Building in Oklahoma City in 1995 have demonstrated the axial force develops, as observed in tests,10-12 when the
disastrous consequences of a progressive collapse. To mitigate longitudinal expansion due to concrete cracking is confined
the risk of progressive collapse due to an abnormal loading by the adjacent structural components. Accordingly, beams
event, a structure must accommodate the initial local damage or slabs with top cracking due to negative bending at the
and develop an alternative load-path to sustain the redistributed supports and bottom cracking due to positive bending at the
loads. Two federal guidelines, GSA 20031 and DOD 2005,2 midspan act as a compressive arch subjected to both bending
adopted this strategy and proposed threat independent-analysis and compression. Tests12 showed that the load capacity of
procedures to assess the potential of progressive collapse of a one-way slabs due to the compressive arch action can be 30
structure following the notional removal of major load- to 100% higher than that determined from the yield-line
bearing elements. collapse mechanism approach. In addition, analytical studies
Despite notable analytical studies,3-6 very limited experimental indicated that even a partially restrained one-way slab may
data exist as the basis of assessing progressive collapse resistance achieve considerable extra strength as a result of the
of reinforced concrete frame structures. Sasani et al.7 conducted compressive axial force.13 Compared with the studies of
an in-place test of a reinforced concrete building with one- slabs, however, there is a lack of experimental evidence for
way floor slabs supported by transverse frames. The structure compressive arch effects for reinforced concrete beams.
was instrumented and its performance following the removal of
an exterior bearing column in the first story by explosion was RESEARCH SIGNIFICANCE
studied. Analysis using conventional line elements was This study experimentally investigated the gravity load
conducted as a complementary study. Sasani and Sagiroglu8 capacity of axially restrained reinforced concrete beams and
and Sasani9 similarly examined the dynamic response and the critical parameters affecting the compressive arch action.
the potential of progressive collapse of another reinforced The study validated the effectiveness of a mechanical model
concrete frame building where one corner column and one addressing the compressive arch effects. Taking into account
adjacent exterior column were simultaneously demolished the beneficial compressive arch effects on flexural capacity
by explosion. Although the two aforementioned structures
were deficient in structural integrity as required by the more
recent design codes, load redistribution was achieved and no ACI Structural Journal, V. 106, No. 5, September-October 2009.
progressive collapse was observed. Yi et al.10 carried out MS No. S-2008-022.R2 received November 25, 2008, and reviewed under Institute
publication policies. Copyright 2009, American Concrete Institute. All rights reserved,
laboratory testing of a 1/3-scale four-bay and three-story including the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including authors closure, if any, will be published in the July-
planar frame specimen subjected to static loading that August 2010 ACI Structural Journal if the discussion is received by March 1, 2010.

600 ACI Structural Journal/September-October 2009


Youpo Su is a Professor of civil engineering and Director of the Hebei Earthquake
Engineering Research Center at Hebei Polytechnic University, Tangshan, China.

ACI member Ying Tian is an Assistant Professor of civil engineering at the University
of Nevada, Las Vegas, NV. He received his PhD from the University of Texas at Austin,
Austin, TX. He is a member of ACI Committee 369, Seismic Repair and Rehabilitation;
and Joint ACI-ASCE Committees 421, Design of Reinforced Concrete Slabs, and
445, Shear and Torsion. His research interests include experimental investigation
and numerical modeling of reinforced concrete members and systems.

Xiaosheng Song is a Lecturer of civil engineering at Hebei Polytechnic University.

of beams would lead to an improved understanding of the


load redistribution characteristics of a reinforced concrete Fig. 1Test subassembly and reinforcement layout. (Note:
frame structure in the scenario of losing a bearing column 1 mm = 0.0394 in.)
and thus lead to a more accurate estimate of the risk of
progressive collapse of the structure. Table 1Specimen properties
ln , fcu , Longitudinal
EXPERIMENTAL PROGRAM b x h, mm MPa reinforcement and ratio
Test variables Test mm (in.) (in.) ln /h (psi) Top Bottom Ties
The experimental program consisted of testing 12 reduced- 150 x 300 1225 4.08 32.3 212, 212,
(4680) = 0.55% = 0.55% 8 at 100
A1
scale specimens. Each specimen represented a two-bay beam (5.9 x 11.8) (48)
resulting from the removal of a supporting column. The tests 150 x 300 1225 35.3 312, 312,
A2 8 at 80
(5.9 x 11.8) (48) 4.08 (5120) = 0.83% = 0.83%
of one-way slabs indicated that the compressive arch action
150 x 300 1225 4.08 39.0 314, 314,
due to longitudinal restraint at the slab boundaries is a function A3 8 at 80
(5.9 x 11.8) (48) (5660) = 1.13% = 1.13%
of flexural reinforcement ratio and the ratio of span to depth.12
150 x 300 1225 4.08 28.8 212, 114,
(4180) = 0.55% = 0.38% 8 at 100
In addition, the instantaneous removal of a bearing column A4
(5.9 x 11.8) (48)
results in a rapid load redistribution at the beams bridging over the 150 x 300 1225 33.1 312, 212,
A5 8 at 80
removed column. It was of interest to confirm the development (5.9 x 11.8) (48) 4.08 (4800) = 0.83% = 0.55%
of compressive arch action in a frame beam subjected to fast 150 x 300 1225 4.08 35.8 314, 214,
A6 8 at 80
loading. Therefore, three series of tests were conducted to (5.9 x 11.8) (48) (5190) = 1.13% = 0.75%
study the following parameters affecting the compressive 150 x 300 1975 6.58 23.2 314, 314,
(3360) = 1.13% = 1.13% 8 at 100
B1
arch effect: 1) flexural reinforcement ratio (A-series); 2) ratio of (5.9 x 11.8) (78)
beam span to depth (B-series); and 3) loading rate (C-series). 150 x 300 2725 24.1 314, 314,
(5.9 x 11.8) (107) 9.08 (3500) = 1.13% = 1.13% 8 at 120
B2

150 x 300 2725 26.4 314, 214,


Test subassemblies (5.9 x 11.8) (107) 9.08 (3830) = 1.13% = 0.75% 8 at 120
B3
As shown in Fig. 1, each test subassembly consisted of two 100 x 200 1225 6.12 19.9 212, 212,
doubly reinforced beams connected with a column stub at C1 8 at 80
(3.9 x 7.9) (48) (2890) = 1.30% = 1.30%
the center and two short columns at the edges where the 100 x 200 1225 6.12 21.0 212, 212,
C2 8 at 80
rotational and longitudinal restraints on beams were (3.9 x 7.9) (48) (3050) = 1.30% = 1.30%
applied. The center column represented the removed column C3
100 x 200 1225 20.4 212, 212,
8 at 80
and was 250 mm (9.8 in.) square for all specimens. The edge (3.9 x 7.9) (48) 6.12 (2960) = 1.30% = 1.30%
columns had an enlarged size for ease of being anchored into
the test setup. 150 mm (5.9 in.) cubes from the same batch of concrete as the
In the A-series tests, beams varied in flexural reinforcement specimen. Because the specimens were doubly reinforced,
ratios and measured 150 mm (5.9 in.) wide, 300 mm (11.8 in.) concrete strength had negligible influence on the beam pure
deep, and 1225 mm (48 in.) long for the clear span. The clear flexural capacity. The wide range of concrete strength,
span ln is defined as the distance from the face of the center however, enabled the examination of the effects of beam axial
column to an edge column. Because deeper girders with stiffness, generally assumed as a function of the concrete
sufficient stiffness directly over the removed column may be strength, on the compressive arch action. Deformed steel
required to ensure an adequate alternative load path,14 beams bars with diameters of 12 and 14 mm (0.47 and 0.55 in.) were
with lower span-depth ratio ln/h were considered in this used for beam flexural reinforcement. The longitudinal
study and included in the A-series tests. In the B-series tests, reinforcement was anchored into the edge columns with
beams had the same cross section as in the A-series but with hooks and met the ACI 318-0515 code requirements for
different span lengths. The beams of C-series specimens development length. Smooth bars with a diameter of 8 mm
were identically reinforced and were 100 mm (3.9 in.) in (0.31 in.) were used for ties. All specimens had a concrete
width, 200 mm (7.9 in.) in depth, and 1350 mm (53 in.) in clear cover of 20 mm (0.79 in.) for the longitudinal bars.
length. Because of the limitation of loading rate imposed by Because the tests focused on the compressive arch effects,
the loading equipment, specimens with smaller scale were continuous longitudinal bars at the center column and hoops
included in the C-series tests. The geometry of the test specimens with close spacing were used in the beams to avoid any premature
is summarized in Table 1. failure due to reinforcement discontinuity or shear distress.
All specimens were constructed using ready mixed Figure 1 illustrates the typical reinforcement layout of the
normalweight concrete. The concrete compressive strength specimens. The measured concrete strength and reinforcing
fcu varied from 19.9 to 39 MPa (2890 to 5660 psi). The details for each specimen are given in Table 1. Table 2
concrete strength was measured at the commencement of a provides the measured material properties of reinforcement
test and was determined by averaging the values of three based on the average of three coupons.

ACI Structural Journal/September-October 2009 601


Table 2Reinforcement properties
Diameter, Yield strength, Ultimate strength, Elongation,
Steel type mm (in.) MPa (ksi) MPa (ksi) %
8 8 (0.31) 290 (42) 455 (66) 33
12 12 (0.47) 350 (51) 540 (78) 26
Fig. 2Schematic illustration of test setup. 14 14 (0.55) 340 (49) 535 (78) 27

pauses for inspection, a constant loading rate of 5 mm/min


(0.197 in./min) was maintained until the failure of specimens.
For the C-series tests, the loading rate was controlled at 0.2, 2,
and 20 mm/s (0.008, 0.079, and 0.787 in./s) for Specimens
C1, C2, and C3, respectively. It is noted that the average rate
of vertical displacement at the columns immediately following
their sudden removal by explosion in the field tests described in
References 7 and 8 did not exceed 90 mm/s (3.54 in./s),
which can be converted into a lower value in a test of specimen
with reduced scale. Thus, if the C-series specimens are assumed
to have a 1/4-scale, the 20 mm/s (0.787 in./s) loading rate
applied on Specimen C3 likely simulated the displacement
speed at the removed column of a frame structure.
Fig. 3Support details. During the tests, the vertical load P and deflection at the
center column stub were measured by a built-in load cell
within the actuator and displacement transducers. The horizontal
reaction N and vertical reaction F shown in Fig. 2 were
measured by load cells embedded in the struts that connected
the steel socket and the support. The measured deflection
and forces F, N, and P allowed for the determination of beam
negative bending moment M at the supports and positive
bending moment M at the midspan according to the equilibrium
of specimens evaluated at their deformed positions.
Steel tensile strains were measured at selected locations in
the potential plastic hinge zones of the beams at the supports
and the center column. In addition, horizontal displacement
and rotation of the edge columns were measured to determine
the actual rigidity of the support. The horizontal rigidity of the
Fig. 4Test setup. support, defined as the force to produce a unit displacement
in the horizontal direction, was approximately 1000 kN/mm
(5710 kip/in.). The rotational stiffness of the support,
Test setup and measurements defined as the bending moment required to produce a unit rota-
Figure 2 shows the experimental scheme adopted in this tion, was approximately 17,500 kN-m/rad (12,910 k-ft/rad).
study. The test scheme permitted the simulation of vertical,
longitudinal, and rotational restraints imposed at the beam
EXPERIMENTAL RESULTS AND DISCUSSION
ends by the rest of a structure. A test setup was designed and Vertical load and horizontal reaction
constructed to achieve the restraining effects at the beam versus deflection
boundaries. Before a test was conducted, the edge columns The overall behavior of the specimens is described in
of the specimen were anchored into the steel sockets (shown Fig. 5 to 9 by the response of vertical load P and horizontal
in Fig. 3) by steel plates and high-strength bolts. Each socket reaction N versus the center deflection normalized by the
was connected with the rigid supports on a steel test bed by beam depth h. Compressive reaction force is assigned to be
a pin located at a distance of 150 mm (5.9 in.) from the beam negative and presented in the lower half of a plot. Figures 5
end. The pin connections allowed the sockets to rotate and and 6 present the test data for the A-series specimens with
horizontally displace temporarily. The specimen was therefore and without symmetrical flexural reinforcement, respectively.
simply supported at this stage. Then the socket at each end of The response of the B-series specimens is shown together with
the specimen was further connected to the support by a Specimens A3 and A6 for comparison purposes in Fig. 7 and 8.
vertical strut and a horizontal strut to achieve axial and Figure 9 shows the response of the C-series specimens
rotational restraints at the beam ends after assembling. Two subjected to different loading rates. The load when yielding
roller bearings were used on the side faces of the center occurred at the supports Py , and the peak load due to
column stub to prevent the specimen from twisting and compressive arch actions Pcu are also shown in Fig. 5 to 8.
exhibiting out-of-plane movement during testing. Figure 4 The test data reported herein do not account for the beam
shows the test setup together with an installed specimen bending effect due to the self-weight of specimens.
prior to a test. Under the applied load, flexural cracking took place first
Gravity loading was simulated by applying downward at the interface of the beam and the center column stub. The
displacements at the center column stub through a servo- cracking loads ranged from 24 to 30 kN (5.4 to 6.7 kips) for
controlled actuator reacting against a loading frame. For the A-series specimens. For the B-series specimens, due to
A- and B-series tests, except for the initial loading and some the larger span and thus higher effects of specimen self-

602 ACI Structural Journal/September-October 2009


Fig. 5Vertical load and horizontal reaction force versus Fig. 7Vertical load and horizontal reaction force versus
normalized center deflection (A1, A2, and A3 with symmetrical normalized center deflection (B1, B2, and A3 with symmetrical
reinforcement). reinforcement).

Fig. 6Vertical load and horizontal reaction force versus Fig. 8Vertical load and horizontal reaction force versus
normalized center deflection (A4, A5, and A6 with normalized center deflection (B3 and A6 with unsymmetrical
unsymmetrical reinforcement). reinforcement).

weight, the cracking loads were reduced to 9.9 to 13 kN


(2.2 to 2.0 kips). The first cracking was soon followed by
the cracking of beam concrete at the supports. The horizontal
reaction forces were generally negligible during the initial
loading but increased notably after flexural cracking
occurred at both midspan and supports, indicating the onset
of compressive arch action. Because of the mobilization of
axial restraint, concrete cracking did not cause a significant
decrease in stiffness and most specimens presented an
approximately linear load-deflection response until the flexural
yielding of the beams.
First yielding of the beams occurred at the midspan of the
specimens due to positive bending and then at the supports
due to negative bending, which resulted in pronounced stiffness
degradation. Following the yielding at the supports, the peak
load Pcu was achieved at a normalized deflection ranging Fig. 9Vertical load and horizontal reaction force versus
between /h = 0.16 and 0.34. Even though Pcu had been normalized center deflection (C1, C2, and C3 subjective to
reached, the horizontal reaction force could keep increasing. different loading rates).
The maximum compressive reaction force Nmax measured in
a test was reached when concrete crushed at the midspan at indicated by the decrease of the measured compressive axial
a value of /h between 0.29 and 0.50. Because of the unexpected force, diminished the axial restraining effects provided at the
in-plane rotation of the center column stub, concrete crushing supports. It is noted that, for A-series specimens with a lower
occurred earlier at /h = 0.20 in Specimen A2. Concrete ln/h (= 4.08), concrete crushing generally caused a rapid drop
crushing reduced the axial stiffness of the beams and thus, as of the applied vertical load and horizontal reaction force.

ACI Structural Journal/September-October 2009 603


This process, however, was fairly gradual for B-series specimens greater compressive arch capacity Pcu, the effects of loading
with higher ln/h (= 6.58 and 9.08). rate considered in the tests can be neglected.
With the increased center deflection, most specimens, Neither shear distress nor anchorage failure was observed
except for A5, developed tensile axial force, indicating a in the three series of tests. Table 3 summarizes the test results
transition from the compressive arch action to the tensile for each specimen including the loads at cracking and
catenary action. The development of catenary action was yielding, peak load under compressive arch action, load at
generally accompanied by the increase of vertical load. Prior fracture of flexural reinforcement, peak axial reaction force,
to failure (represented by the terminating points of the curves and measured center deflections at several characteristic
in Fig. 5 to 9), the specimens have exhibited significant loading stages.
deflection mainly due to the concentrated inelastic deformation
at the supports and the midspan. The final failure of all specimens Effects of compressive arch action on beam
was announced by the fracture of bottom reinforcement at the load-carrying capacity
interface of beam and center column stub. The failure mode The measured loading capacities Pcu due to compressive
of Specimen B3 is shown as an example in Fig. 10. The arch effects for the A- and B-series specimens and Specimen C1
values of /h at failure ranged from 0.66 for Specimen A3 to were compared with analytical predictions using the classical
1.44 for Specimen B3. The relative deflection /ln that can be plastic collapse mechanism approach and using the formulations
taken as a function of the plastic hinge rotation, however, was suggested by Park.13 The loading capacity Pyu according to the
limited to close values between 0.16 and 0.20 radians. It is conventional plastic analysis approach was determined at the
noted that the ultimate load under catenary action Ptu formation of a collapse mechanism due to hinging at both the
measured at the fracture of longitudinal reinforcement was lower midspan and the supports of a specimen. The ultimate
than the compressive arch capacity Pcu for the A-series bending capacity without considering the existence of axial
specimens with shorter span and was 20 to 45% higher for restraining force was used to define the flexural strength of a
B-series specimens with longer span. The C-series specimens plastic hinge. No strain hardening was considered for the
(ln/h = 6.12) subjected to three different loading rates longitudinal reinforcement. The data series with hollow
presented similar behavior to the B-series specimens. markers in Fig. 11 shows the comparison between Pcu and
Compressive arch action was observed in all tests. Even Pyu. The vertical distance from the diagonal line of the figure
though a higher loading rate resulted in higher cracking load,
larger compressive horizontal reaction force, and slightly

Fig. 10Failure mode of Specimen B3. Fig. 11Comparison of measured and calculated results.

Table 3Experimental results


Compressive arch action Tensile arch action
Load at yielding Maximum compressive Load at fracture
Cracking at supports, Peak load Deflection at Horizontal reaction horizontal reaction Deflection at of bottom bar Deflection at
Test load, kN (kip) kN (kip) Pcu , kN (kip) Pcu, mm (in.) at Pcu, kN (kip) Nmax, kN (kip) Nmax , mm (in.) Ptu, kN (kip) Ptu, mm (in.)
(1) (2) (3) (4) (5) (7) (8) (9) (10) (11)
A1 26 (5.84) 117 (26.3) 168 (37.8) 48 (1.89) 281 (63.2) 388 (87.2) 87.8 (3.46) 93.1 (20.9) 202 (7.95)
A2 30 (6.74) 148 (33.3) 221 (49.7) 56.4 (2.22) 318 (71.5) 324 (72.8) 59.3 (2.33) 140 (31.5) 205 (8.07)
A3 29 (6.52) 152 (34.2) 246 (55.3) 76.4 (3.01) 296 (66.5) 305 (68.6) 87.8 (3.46) 178 (40.0) 199 (7.83)
A4 24 (5.40) 82.3 (18.5) 147 (33.0) 65 (2.56) 309 (69.5) 344 (77.3) 93.5 (3.68) 45.9 (10.3) 238 (9.37)
A5 29 (6.52) 129 (29.0) 198 (44.5) 70.7 (2.78) 340 (76.4) 393 (88.3) 108 (4.25) 58.1 (13.1) 227 (8.94)
A6 27 (6.07) 153 (34.4) 226 (50.8) 69.2 (2.72) 177 (39.8) 191 (42.9) 89.1 (3.51) 144 (32.4) 209 (8.23)
B1 13 (2.92) 105 (23.6) 125 (28.1) 100 (3.94) 211 (47.4) 225 (50.6) 146 (5.75) 150 (33.7) 389 (15.3)
B2 10 (2.25) 73.2 (16.5) 82.9 (18.6) 102 (4.02) 190 (42.7) 210 (47.2) 125 (4.92) 121 (27.2) 255 (10.0)
B3 9.9 (2.23) 65.0 (14.6) 74.7 (16.8) 85.5 (3.37) 172 (38.7) 210 (47.2) 150 (5.91) 90.2 (20.3) 431 (17.0)
C1 8.0 (1.80) 48.2 (10.8) 60.9 (13.7) 33.7 (1.33) 91.6 (20.6) 108 (24.3) 62.5 (2.46) 65.7 (14.8) 227 (8.94)
C2 9.1 (2.05) 64.9 (14.6) 33.5 (1.32) 96.4 (21.7) 117 (26.3) 65.4 (2.57) 77.6 (17.4) 250 (9.84)
C3 10.2 (2.29) 68.6 (15.4) 28.7 (1.13) 108 (24.3) 134 (30.1) 60.0 (2.36) 54.4 (12.2) 201 (7.91)

604 ACI Structural Journal/September-October 2009


to the data points clearly indicates the remarkably underestimated
loading capacities of the specimens based on the conventional
approach. The strength enhancement factor due to the
compressive arch effect, defined as the ratio of Pcu to Pyu,
ranged from 1.53 for Specimen B2 to 2.63 for Specimen A4.
Park13 proposed an analytical model for one-way slabs
longitudinally restrained at the slab boundaries. Parks
formulations were modified herein to be used for axially
restrained beams. The load capacity due to compressive arch
effect for the specimens tested in this study was determined as the
maximum value of P computed using the following equation

1
P = ------- 0.85f c 1 bh --- 1 ----1- + --- ( 1 3 ) +
h
l n 2 2 4 Fig. 12Normalized horizontal reaction force, applied
load, and bending moments at support and midspan versus
2
l n - 2
2
l n
--------- ( 1 1 ) total + ----- 2 ----1- + --------- 1 ----1- total normalized center deflection (Specimen B3).
8h 2 h 2
(1)
2 4 2
1 l n 2 ( T T Cs + C s ) It is noted that the choice of S for practical situations may
------------------ total ------------------------------------------ + ( Cs + C s )
h
2 3.4f c b not have a pronounced effect on the calculated load capacity.
The longitudinal restraint at the beam ends is provided by the
h h surrounding structural components including the columns as
--- d --- + ( T + T ) d --- + --- l n
2 2 2 2 well as the floor slab. Based on Parks model, the effects of
lateral rigidity on Pcu is a function of the ratio Sln/(bhEc) that
represents the relative axial rigidity of the support to the
where b is the beam width; h is the beam depth; fc is the axial stiffness of the beam. It was found that, even if only
concrete cylinder compressive strength assumed as f c = 0.8fcu; 20% of the actual support rigidity in the tests was considered,
1 (determined according to ACI 318-0515) is the ratio of the the computed loading capacity Pcu of the specimens was
depth of concrete equivalent stress block to the depth of reduced by less than 10%.
neutral axis; is the ratio of the distance from a plastic hinge
at the midspan to the nearest support to 2ln; T and T are Effects of axial restraint on internal forces
tensile resultant forces of steel at the midspan and supports; Although it is well recognized that the existence of axial
Cs and Cs are compressive resultant forces of steel at the force in a column can increase its flexural capacity, the role
midspan and supports; T, T, Cs, and Cs are calculated using of axial force developed in a longitudinally restrained beam
steel yield strength; d is the beam effective depth; d is the is complex. To illustrate this situation, Fig. 12 presents in the
distance from the centroid of compressive reinforcement to same plot the response of vertical load P, horizontal reaction N,
the concrete compression surface; is the beam self-weight bending moment at the supports M, and moment at the
per unit length of beam; and total is the total axial strain due midspan M as the center deflection increased in Specimen B3
to beam axial deformation and support longitudinal until its failure. For convenience of comparison, P was
displacement, and is determined as normalized by Pyu, the load at the formation of a plastic
collapse mechanism without considering the compressive arch
-------- h T T C s + C s C s T effects. N was normalized by the absolute value of the
- + ------- 0.85f c 1 --- --- ---------------------------------------
1 b
- + ---------------
hE c l n S 2 4 1.75f c 1 b b maximum compressive axial force Nmax. M and M determined
total = ----------------------------------------------------------------------------------------------------------------------------------------------
- (2)
0.85f c 1 l n 1
2
from test data were normalized by the corresponding pure
- --------- + ------
b
1 + -----------------------------
hE c l n s bending capacity M0 and M0 calculated without considering
the interaction between axial force and moment. P, N, M,
where S is support rigidity in the horizontal direction, and Ec is and M satisfy
concrete elastic modulus determined according to ACI 318-05.15
The solution provided by Eq. (1) and (2) is implicit because P = 2(M + M N Mg)/ln (3)
deflection must be increased gradually to determine the peak
value of P. It is noted that due to the assumptions made in the where Mg denotes the bending moment caused by the specimen
model regarding the state of stress for steel and concrete, the self-weight. Figure12 indicates that compressive arch action
P- relation defined by Eq. (1) and (2) cannot be used to can significantly improve the flexural capacity of a beam
describe the entire load-deflection response of a specimen. section. Compared with M0 and M0, the beam flexural
Equations (1) and (2) were adopted to estimate Pcu , the capacity of Specimen B3 at the supports and the midspan
load-carrying capacity of the specimens due to compressive was increased by 66% and 150%, respectively. However, a
arch action. The measured horizontal rigidity of the support, two-fold effect of the compressive axial force exists. As indicated
approximately 1000 kN/mm (5710 kip/in.), was used to define by Eq. (3), the axial force tends to reduce the load capacity by
S. Comparison of calculated versus measured capacity is shown means of P- effect. As a result, Pcu was reached well before
by solid dots in Fig. 11. It can be seen that good agreement the maximum bending moments at the supports and the
was achieved between the test results and the analytical midspan were achieved. Thus, the overall strength enhancement
results based on Parks model. The average ratio of calculated effect, measured by the peak value of P/Pyu , was less than the
to measured capacity is 0.95 with a standard deviation of 0.04. average of the peak values of M/M0 and M/M0 that represented

ACI Structural Journal/September-October 2009 605


midspan of all other unsymmetrically reinforced specimens
decreased by less than 5% when the maximum negative
bending was achieved at the supports. Accordingly, an
average of the tensile reinforcement ratio evaluated at the
supports and the midspan was used to examine the effects of
flexural reinforcement ratio. As shown in Fig. 13(a), the
strength enhancement factor was observed to increase in
an approximately linear manner from 1.91 to 2.63 as the
reinforcement ratio decreased from 1.13 to 0.46%.
The influence of span length on the compressive arch
action is illustrated in Fig. 13(b) by comparing the ln/h and
the strength enhancement factor for five specimens. The
solid dots denote the symmetrically reinforced specimens,
Fig. 13Effect of reinforcement ratio and ratio of beam A3, B1, and B2 (reinforcement ratio = 1.13%). The hollow
span to depth: (a) effect of flexural reinforcement ratio; and dots denote the unsymmetrically reinforced specimens, A6
(b) effect of ratio of beam span to depth. and B3 (average reinforcement ratio = 0.9%). Figure 13(b)
indicates that the strength enhancement factor that
measures the beneficial compressive arch effect on beam
gravity load capacity decreases as the ln/h increases. The
lowest strength enhancement factor, however, could still
reach a value of 1.54 that occurred in Specimen B2 with a
ln/h of 9.08.

Load capacity under catenary action


The specimen load-carrying capacity under catenary
action, Ptu, was estimated using a simple model schematically
shown in Fig. 14. The model was established according to
equilibrium at the center column stub where the specimens
failed in fracture of the bottom reinforcement. Ptu was
assumed to be resisted only by the vertical component of Ntu,
the tensile force developed in the bottom bars. The resistance
provided by the top bars at the midspan and bottoms at the
supports was neglected. Ntu was determined from the area
and tensile strength of the bottom bars. It was assumed that
the line of action of Ntu extended from the bottom bar at the
Fig. 14Modeling for catenary action and comparison of center column stub to the top bars at the support. The load
measured and calculated results for loading capacity under capacity due to catenary action was calculated as Ptu = 2Ntusin,
catenary action. where defines the orientation of Ntu and was determined
based on: 1) the beam clear span length, and 2) the summation
of measured vertical displacement of the center column stub
the flexural capacity enhancement at the supports and the at failure and the vertical distance between the top and
midspan. It is noted that, prior to failure, the specimen could bottom reinforcing bars. Figure 14 shows the comparison of
still resist a significant amount of bending moment at the calculated to measured load capacity. The figure indicates
critical sections. Therefore, a double curvature deformed that, although the simple model accurately predicted Ptu for
shape was maintained in the beams until failure when the some specimens, a significant discrepancy exists between
bottom reinforcement at the midspan fractured under measured and calculated results for other specimens.
catenary action.
Use of compressive arch action and catenary
Effects of reinforcement ratio and span-depth action in progressive collapse prevention
ratio on compressive arch effect The tests conducted in this study indicated that, if adequate
As indicated in Eq. (3), the compressive arch capacity Pcu transverse reinforcement and continuous bottom reinforcement
depends on beam flexural capacity, compressive axial force, are provided in a frame beam, gravity load-carrying capacity
and beam deflection that are functions of flexural reinforce- can be considerably higher than that conventionally
ment ratio and the ratio of beam span to depth. Given that the assumed. The extra loading capacity due to compressive
analytical model formulated in Eq. (1) and (2) is complex for arch action is suggested to be considered in analyses so that
practical applications, the influences of beam reinforcement the performance of a structure following the loss of a bearing
ratio and the ln/h on compressive arch effect were examined column can be more accurately simulated. The nonlinear
separately using test data. finite element model6 employing fiber discretization for
Figure 13(a) plots the strength enhancement factor against reinforced concrete beams likely serves this purpose because
beam flexural reinforcement ratio for A-series specimens this model can capture the compressive axial force developed in
(ln/h = 4.08). Both the top bars at supports and the bottom a longitudinally restrained beam. The line elements
bars at midspan had yielded in all tests before Pcu was conventionally used in practice for modeling frame beams,
reached. In addition, similar to the behavior of Specimen B3 however, must be modified so that the compressive arch
shown in Fig. 12, the positive bending moment at the effect can be approximately addressed. For this purpose, a

606 ACI Structural Journal/September-October 2009


strength enhancement factor may be introduced to modify the compressive arch effect still could be developed but the
the definition of beam flexural strength. The strength effects of loading speed can be neglected. The study indicated
enhancement factor defined in Eq. (4) was empirically that the beam gravity load-carrying capacity under compressive
derived from a regression analysis of test data shown in Fig. 13 arch action can be predicted by an analytical model with
and can be used for beams with flexural reinforcement ratio less sufficient accuracy. The compressive arch effects are
than 1.3% (the largest reinforcement ratio for the specimens suggested to be incorporated into the assessment of vulnerability
tested in this study). to progressive collapse of reinforced concrete frame structures.
To take advantage of the compressive arch action, beams
with deeper sections and lower longitudinal reinforcement
ratios should be used in a structural design, if other structural

= 2.4 1 --- 1 + ------------
5
(4) components such as the adjacent beams, columns, or slabs
3 2
l---n can provide an effective longitudinal restraint.
h
ACKNOWLEDGMENTS
The tests of B- and C-series specimens indicated that The authors gratefully acknowledge the fund provided by the Natural
Science Foundation of China (No. 50478114) for the work presented in this paper.
considerable deflection must be reached in a beam to
develop a Ptu higher than Pcu. The ultimate deformation of a
beam, however, largely depends on the elongation of tensile REFERENCES
1. GSA, Progressive Collapse Analysis and Design Guidelines for New
reinforcement at the critical sections. Thus, unless sufficient Federal Office Buildings and Major Modernization Projects, U.S. General
steel elongation has been specified and unless an appropriate Services Administration, Washington, DC, 2003.
strength reduction factor is applied to address the uncertainty 2. DOD, Design of Building to Resist Progressive Collapse, Unified
involved in predicting the catenary action capacity, caution Facility Criteria, UFC 4-023-03, U.S. Department of Defense, Washington,
DC, 2005.
should be exercised to count on the catenary action in a beam 3. Baldridge, S. M., and Humay, F. K., Preventing Progressive Collapse
to resist progressive collapse. In addition, although Ptu was in Concrete Buildings, Concrete International, V. 25, No. 11, Nov. 2003,
22 to 49% larger than Pcu for B-series specimens, Ptu was pp. 73-79.
close to or even less than Pcu for C-series specimens 4. Hayes, J. R. Jr.; Woodson, S. C.; Pekelnicky, R. G.; Poland, C. D.;
Corley, W. G.; and Sozen, M., Can Strengthening for Earthquake Improve
subjected to faster loading. Therefore, instead of defining the Blast and Progressive Collapse Resistance? Journal of Structural
limit state of a progressive collapse at the fracture of tensile Engineering, ASCE, V. 131, No. 8, 2005, pp. 1157-1177.
reinforcement under catenary action, compressive arch 5. Sasani, M., and Sagiroglu, S., Progressive Collapse of Reinforced
capacity may have the potential to be used as the design basis Concrete Structures: A Multihazard Perspective, ACI Structural Journal,
for progressive collapse prevention. It should be noted that, V. 105, No. 1, Jan.-Feb. 2008, pp. 96-103.
6. Bao, Y.; Kunnath, S. K.; El-Tawil, S., and Lew, H. S., Macromodel-
if the compressive arch action is taken into account, the Based Simulation of Progressive Collapse: RC Frame Structures, Journal
vertical displacement at the removed column must be of Structural Engineering, ASCE, V. 134, No. 7, 2008, pp. 1079-1091.
controlled within an acceptable level. The test data shown in 7. Sasani, M.; Bazan, M.; and Sagiroglu, S., Experimental and Analytical
Fig. 5 to 9 indicate that, if the vertical deflection at the Progressive Collapse Evaluation of an Actual Reinforced Concrete Structure,
ACI Structural Journal, V. 104, No. 6, Nov.-Dec. 2007, pp. 731-739.
removed column can be limited to 0.3h, the compressive 8. Sasani, M., and Sagiroglu, S., Progressive Collapse Resistance of
arch effects can be considered with sufficient confidence. Hotel San Diego, Journal of Structural Engineering, ASCE, V. 134, No. 3,
Consistent with the design philosophy of the GSA guidelines,1 2008, pp. 478-488.
the aforementioned deformation limitation could also reduce 9. Sasani, M., Response of a Reinforced Concrete Infilled-Frame Structure
the extent of damage of a structure by arresting the failure to Removal of Two Adjacent Columns, Engineering Structures, V. 30, No. 9,
2008, pp. 2478-2491.
progression of structural components. 10. Yi, W.; He, Q.; Xiao, Y.; and Kunnath, S. K., Experimental Study
on Progressive Collapse-Resistant Behavior of Reinforced Concrete Frame
SUMMARY AND CONCLUSIONS Structures, ACI Structural Journal, V. 105, No. 4, July-Aug. 2008, pp. 433-439.
An experimental study was carried out to investigate the 11. Orton, O. L., Development of a CFRP System to Provide Continuity
in Existing Reinforced Concrete Buildings Vulnerable to Progressive
gravity load-carrying capacity of reinforced concrete frame Collapse, PhD dissertation, University of Texas at Austin, Austin, TX,
beams when a supporting column is removed due to 2007, 363 pp.
abnormal loading events. The tests confirmed the strength 12. Leslie, K. G., and Edward, J. R., Membrane Action in Partially
enhancement effect of compressive arch action on beam Restrained Slabs, ACI Structural Journal, V. 85, No. 4, July-Aug. 1988,
pp. 365-373.
flexural capacity. For the specimens tested in this study, the 13. Park, R., and Gamble, W. L., Reinforced Concrete Slabs, second
compressive arch action resulting from axial restraint edition, John Wiley & Sons, 2000, 716 pp.
contributed 50 to 160% extra loading capacity beyond the 14. Marjanishvili, S. M., Progressive Analysis Procedure for Progressive
capacity estimated without considering axial restraining forces. Collapse, Journal of Performance of Constructed Facilities, ASCE, V. 18,
No. 2, 2004, pp. 79-85.
The tests indicated that the effects of compressive arch 15. ACI Committee 318, Building Code Requirements for Structural
action increased with decreased beam span-depth ratio and Concrete (ACI 318-05) and Commentary (318R-05), American Concrete
decreased flexural reinforcement ratio. Under fast loading, Institute, Farmington Hills, MI, 2005, 430 pp.

ACI Structural Journal/September-October 2009 607

Vous aimerez peut-être aussi