Professional Documents
Culture Documents
Progressive Collapse Resistance of RC Beams
Progressive Collapse Resistance of RC Beams
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
a r t i c l e i n f o a b s t r a c t
Article history: The current guidelines for evaluating progressive collapse potential of existing and new buildings require
Received 7 October 2014 analyzing and evaluating the structure for the case of an instantaneous loss of a primary vertical support,
Revised 20 March 2015 such as a column. In this paper, the response of a continuous beam bridging over a lost column is eval-
Accepted 22 March 2015
uated. A series of beam models are developed by changing structural characteristics such as lateral load
Available online 8 April 2015
design type (ordinary vs. special frames), axial stiffness at the beam boundaries, steel yield stress, amount
of integrity reinforcement at bar cut-off locations and the beam span to study their effects on the perfor-
Keywords:
mance of the beam. The modeling technique used for the analyses of the beams has been validated by
Progressive collapse
Beam growth
comparing experimental and analytical results of an RC beam subjected to large deformations. Push-
Reinforcement detailing down analyses are carried out in order to study and characterize the full range of response and compare
Special frame the behavior of the beams. For each case, the behavior of the critical sections are evaluated and used to
Ordinary frame describe load transfer mechanisms. The effects of different structural characteristics on the performance
Integrity requirements of the beam to resist progressive collapse are discussed by comparing the results of the analyses.
Ó 2015 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2015.03.044
0141-0296/Ó 2015 Elsevier Ltd. All rights reserved.
62 E. Livingston et al. / Engineering Structures 95 (2015) 61–70
provided by the rest of the structure (mainly by the floor system), RC beam. The test consisted of imposing an increasing downward
an axial compressive force develops in the beam [3,17]. This axial displacement at the center column stub, at the location considered
compressive force can significantly affect the flexural behavior to be above the removed column. The applied force, vertical deflec-
and in turn the vertical load carrying capacity of the beam. In the tion and other response measures were recorded during the test.
case of loss of an interior column of a frame, the beam bridging A two-dimensional finite-element model of the RC beam speci-
over the lost column can experience three phases of response [3] men was developed using OpenSees. Each of the two beam spans
which are explained in the following sections. During the first was divided into seven segments and modeled using force (flexibil-
phase the beam flexural response provides the vertical load carry- ity) based nonlinear beam–column elements with fiber sections
ing capacity. The axial compressive force of the beam continues to and distributed plasticity. The center stub is modeled using two
increase during this phase (improving the flexural response elastic beam–column elements with high axial and flexural stiff-
through P–M interaction) until the maximum vertical load carrying ness relative to the adjacent nonlinear beam elements’ stiffness.
capacity is provided. This axial force is an important system-level Each nonlinear beam–column element is modeled using two
behavior affecting the beam response that should be accounted integration points. The concrete cover, concrete core and steel bars
for. In this paper we will evaluate the response of beams bridging are modeled using uniaxial nonlinear stress–strain relationships
over a lost column as part of a structural system. In order to study with model parameters based on material testing of concrete and
and characterize the full range of response and compare the behav- reinforcing steel samples.
ior of different beams, push-down analyses are carried out. The In order to model the effects of steel bar slip within the joint
effects of the following parameters on the performance of the (and supports), zero-length elements are used at the end nodes
beam to resist progressive collapse are studied: lateral load design of the beams. The zero-length element was defined by a bilinear
type (ordinary vs. special frames), axial stiffness at the beam moment–rotation relationship to model the rotation due to the
boundaries, steel yield stress, amount of integrity reinforcement base crack. This relationship was defined such that it allowed for
at bar cut-off location, and the beam span length. strains of the tensile steel fibers at the beam end sections to reach
the yield strain before yielding of the zero-length elements and the
2. Analytical model validation maximum rotation at the base cracks was consistent with the
opening and depth of the base crack measured during the test.
In order to validate the modeling technique for evaluating the The results are obtained using a displacement-controlled non-
large displacement response of RC beams using the computer pro- linear static pushdown analysis, which consisted of imposing an
gram OpenSees [9], analytical results are compared with experi- increasing downward displacement at the center node of the
mental data obtained from an RC beam subjected to large model, as well as the corresponding horizontal displacement and
deformations [3,17]. The test specimen was a 3/8th scaled model rotation of the center column stub, as monitored during the test.
of a continuous two-span RC beam, designed as part of an exterior The material nonlinearity is accounted for through the use of the
frame of a 7-story ordinary RC building where a middle exterior fiber sections and the nonlinear stress–strain relationships of the
column was instantaneously removed [3]. fibers. Geometric nonlinearity is included in the analysis through
Fig. 1 shows the dimensions and reinforcement details of the the use of the corotational geometric transformation in OpenSees.
scaled beam specimen. The RC beam specimen had concrete end Fig. 2 compares the analytical and experimental force–displace-
blocks that were securely attached to a steel reaction frame using ment relationships of the RC beam. A downward force is consid-
steel threaded rods in order to provide fixed-end conditions to the ered negative. As can be seen, the analytical model captures the
12" 12"
(305mm) (305mm)
6 3/4"
(170mm)
6'-6 3/4" 6'-6 3/4"
(2000mm) (2000mm)
14 1/2"
23 5/8" 23 5/8" 23 5/8" (368mm) B 23 5/8" A
2#3
(600mm) (2D10) (600mm) (600mm) (midspan) (600mm)
3'-8" 3#3 W1.4@2.7" 3#3 W1.4@2.7" 3#3
(1120mm) (3D10) (D4@70) (3D10) (D4@70) (3D10)
2#3 2#3
(2D10) 14 1/2" (2D10) B A
(368mm)
712"
(190mm)
114" 712"
114" (32mm) (190mm)
(32mm)
5#3 2#3 3
3
(5D10) 4"
(2D10) 4"
W1.4@2.7'' (19mm) 71" (19mm) 71"
2 W1.4@2.7'' 2
(Diam=0.135'') (Diam=0.135'')
(190mm) (190mm)
(D4@70mm) 2#3 (D4@70mm) 2#3
(2D10) (2D10)
3 3
4" 4"
(19mm) Section A - A (19mm) Section B - B
3. Structure characteristics
1 2 3 4 5 6 7
A
30'
9.14 (m)
30'
9.14 (m)
C
26' 26' 26' 26' 26' 26'
7.92 (m) 7.92 (m) 7.92 (m) 7.92 (m) 7.92 (m) 7.92 (m)
C3 24.5' C4 The concrete fibers of sections are divided into a core area
1 2 7.47 (m) 3 which is the confined portion of the section and a cover area which
8' 8' is the unconfined portion. The stress–strain relationship for the
3#9 Ordinary (3D29) 2#5 (2D16) 3#9 Ordinary (3D29) concrete is defined by a uniaxial material model. The cover con-
3#7 Special (3D22) 2#7 (2D22) 3#7 Special (3D22)
crete is defined by a compressive strength and strain of 5 ksi
(a) (34.5 MPa) and 0.002 and crushing strength and strain of 0.5
1 2 26' 3
(3.45 MPa) and 0.007 respectively. Consequently, the stress–strain
7.92(m) Removed relationships of core concrete for special and ordinary design cases
Column
are found using the modified Kent and Park model [10]. The crush-
C3 24.5' C4 ing strain is defined as 0.01 for the ordinary beam case and 0.02 for
1 2 7.47(m) 3 the special design case. The only exception is for the middle ele-
8' 8' ment in the special beam model, which has the same concrete core
4#7 Ordinary (4D22) 2#7 (2D22) 4#7 Ordinary (4D22) properties as that of the ordinary design.
2#7 Special (2D22) 2#7 (2D22) 2#7 Special (2D22)
Steel yield and ultimate strengths of 60 ksi (413.7 MPa) and
(b) 90 ksi (620.5 MPa) are used respectively. The effects of higher yield
1 2 26' 3 and ultimate strengths are also studied. Steel bar buckling is
7.92(m) Removed accounted for based on [20] and as described by Sagiroglu [14]
Column
and Sagiroglu and Sasani [15].
Fig. 4. Reinforcement detail of ordinary and special frames (a) Case 1, (b) Case 2
(response of sections 1–3 are evaluated). 5. Analytical results
Line of
1350 lb/ft
(19.7 kN/m) Symmetry
Fig. 6. External and internal forces for Case 1 ordinary and special beams. Top: bending moments; and bottom: axial force (divided by 5) and vertical force.
Fig. 5) decreases. At 3.52 in. (89 mm) of displacement at which the than the rate at which the P–D is increasing (in fact both P and
beam tensile (top) reinforcement at the face of the column adja- D are increasing). And therefore, because the P–D effect reduces
cent to the removed column (section 1, see Fig. 4a) yields. This is the shear carrying capacity of the beam faster than the increase
reflected in the bending moment of section 1 (see Fig. 4), i.e. due to higher beam end moments, the downward force starts to
M1-O shown in Fig. 6c, and in turn as a drop in the slope of the decrease as the vertical displacement increases beyond this point.
vertical force shown in Fig. 6d. At 9 in. (229 mm) of displacement the concrete cover at the top
As the beam deforms downwards and flexural cracks form and of section 3 reaches its compressive strength, followed at 9.8 in.
yielding occurs, the beam tends to grow in length. This tendency is (249 mm) by the core at the bottom of section 1 reaching its com-
constrained by the boundary conditions of the beam, which leads pressive strength. Soon after, the compressive steel in section 1
to the development of axial compressive force in the beam [17]. yields. The compressive axial force developed in the beam reaches
The axial (horizontal) force of the beam at the line of symmetry, its maximum value and starts to decrease at 10.02 in. (254 mm) of
where the beam rotation is zero, is shown in Fig. 6d. This axial displacement, and then the moment capacities developed at the
compressive force on the one hand improves the flexural capacity beam ends also begin to decrease.
of sections and in turn the vertical load carrying capacity of the At 19.42 in. (493 mm), the third phase of the response starts at
beam through the section axial–flexural interaction and on the the point where the upward vertical force required for equilibrium
other hand reduces the vertical load carrying capacity through begins to decrease. It is important to note that the regain of
the P–D effect. strength is achieved while the beam end moments are dropping
At 3.64 in. (93 mm) of displacement the beam reaches equilib- (see Fig. 6c) and the beam is still under a compressive axial force.
rium on its own (i.e. the vertical force drops to zero) and in order Again, the initiation of the third phase of response (similar to the
for the beam to continue displacing, a downward force is required. first phase) can be explained by the fact that the rate at which
At 3.9 in. (99 mm) the bottom reinforcement at the face of the the sum of the beam end moments is decreasing becomes less than
removed column (section 3) yields which leads to the yielding of the rate at which the P–D is decreasing. And therefore, because the
the section (M3-O in Fig. 6c). At 5.75 in. (146 mm) of displacement reduction of P–D leads to a larger vertical load carrying capacity of
the cover concrete at the bottom of section 1 reaches its peak com- the beam faster than the decrease in capacity due to lower beam
pressive strength and begins to crush. end moments, the downward force reaches its second local peak.
At 6.81 in. (173 mm) the beam begins to weaken, i.e. the mag- At 29.64 in. (753 mm), the axial force switches from compression
nitude of the downward (negative) force required to continue dis- into tension. This can be considered to be the start of catenary
placing the beam starts to decrease (the downward force reaches action.
its local peak value), which marks the initiation of the second
phase of the response. It is interesting to note that in this case, 5.2. Special frame Case 1 (medium axial stiffness/force)
the drop in the vertical load carrying capacity occurs in spite of
the fact that both beam end moments are still increasing (see The results for the special frame Case 1 analysis with medium
Fig. 6c). However, at this point, the rate at which the sum of the axial stiffness show a similar sequence of events to the ordinary
beam end moments (which contribute to developing the beam frame analysis for the same case. The steel rebars yield is almost
shear force) after tensile rebar yielding is increasing becomes less exactly in the same sequence as the ordinary case, albeit at
66 E. Livingston et al. / Engineering Structures 95 (2015) 61–70
(a) (c)
(b) (d)
Fig. 8. External and internal forces for ordinary beams Case 1 and Case 2. Top: bending moments; and bottom: axial force (divided by 5) and vertical force.
(see Fig. 10). The ordinary frame design requires that only a quarter compressive axial force at 12.43 in. (316 mm). Also, section 1
of the bottom reinforcement at the mid span needs to be continu- reaches its maximum negative bending moment right before the
ous to satisfy the integrity requirement. Due to this, the ordinary positive (bottom) reinforcement in the section yields in
frame design has 2#6 + 2#9 (2D19 + 2D29) at the bottom mid span compression.
for a total area of 2.88 in.2 (1858 mm2) for which 2#6 (2D19) At around 20.67 in. (525 mm) the vertical force required to
extends at the ends for a total area of 0.88 in.2 (568 mm2). The spe- reach equilibrium begins to decrease. That is, the vertical load car-
cial frame design requires that the positive moment capacity of a rying capacity increases. Similar to the discussion made for the
beam at the face of a column is at least half of that of the negative 26 ft (7.92 m) long ordinary beam, this occurs in spite of the fact
moment capacity [1]. The special frame design has 2#7 + 3#10 that the beam end bending moments reduces as the vertical dis-
(2D22 + 3D32) at the top ends for a total area of 5.01 in.2 placement increases at this point. At 44.22 in. (1123 mm) the ver-
(3232 mm2) and 4#7 (4D22) at the bottom ends for a total area tical force drops to zero (system is at equilibrium) and in order for
of 2.4 in.2 (1548 mm2). It is noted that due to the integrity require- the beam to continue displacing a downward force must be
ments explained above, the special frame design consists of a total applied. As the bottom core concrete in section 1 starts to crush
area at the bottom ends of almost three times that of the ordinary the negative moment capacity of the section goes down and the
frame design. This is important in analyzing which design behaves compressive rebars start to buckle.
better in terms of progressive collapse resistance following loss of a
column. 6.2. Special frame (medium axial stiffness) – 37 ft (11.28 m) span
6.1. Ordinary frame (medium axial stiffness) – 37 ft (11.28 m) span Fig. 11c and d show the moment and force responses of the 37 ft
(11.28 m) special frame beam along with those of ordinary frame
The load carrying capacity of the 37 ft (11.28 m) ordinary frame beam. Compared to the response of the ordinary frame beam, the
beam can be evaluated by examining Fig. 11c and d and by the positive moment of the special beam at section 3 (M3-S) is con-
additional description provided below. At 8.58 in. (218 mm) of dis- siderably larger. Due to higher compressive and transverse
placement the top rebars of section 1 (see Fig. 10a) yield. This is reinforcement in the special frame compared to the ordinary
followed by the cover concrete at the bottom of section 1 reaching frame, concrete crushing does not occur in the core of the special
its compressive strength at 9.92 in. (252 mm) of displacement. frame beam. The beam reaches equilibrium at around 36.95 in.
Shortly after, at 10.08 in. (256 mm) of displacement the upward (939 mm), which is almost 8 in. (203 mm) smaller than that for
force required for the beam equilibrium begins to increase, i.e. the ordinary frame for the same axial stiffness.
the vertical load carrying capacity of the beam drops. Similar to
the response of the frame with a 26 ft (7.92 m) span beam, this 6.3. Effect of beam span
occurs in spite the fact that the beam end moments are still
increasing. Note that section 3 reaches its maximum positive bend- As can be seen from Fig. 11 which compares the results of the
ing moment at 12.29 in. (312 mm), followed by the maximum cases for all three levels of axial stiffness for the 37 ft (11.28 m)
68 E. Livingston et al. / Engineering Structures 95 (2015) 61–70
(a) (c)
(b) (d)
Fig. 9. External and internal forces for ordinary and special beams under different yield strengths Case 1. Top: bending moments; and bottom: axial force (divided by 5) and
vertical force.
Fig. 10. Reinforcement detail of (a) ordinary and (b) special frames.
7. Conclusions
beam span, it can be concluded that beams of special frames have a
better performance than beams of ordinary frames for longer span In order to evaluate the collapse resistance of RC frame struc-
lengths. The vertical force–displacement curve of the special frame tures following loss of a middle column of the perimeter frame,
beams reaches equilibrium sooner than ordinary frame beams and the response of the beam bridging over the lost column is studied.
require larger downward forces to continue displacing the beam The important effects of the 3D structure on the beam response are
after equilibrium. By comparing these results to the results of accounted for by properly modeling the boundary conditions of the
E. Livingston et al. / Engineering Structures 95 (2015) 61–70 69
Fig. 11. External and internal forces for ordinary and special beams. Top: bending moments; and bottom: axial force (divided by 5) and vertical force [37 ft (11.28 m) span].
beam. The model captures the axial compressive force developed displacement increases during the third phase of response, the
in the beam, which is a combined result of RC beams’ tendency beam axial compressive force eventually turns to tensile and cate-
to grow due to cracking and yielding and constraint to this ten- nary action develops.
dency to elongation provided by the surrounding structure. A veri- The progressive collapse resistance of ordinary and special
fied analytical modeling technique is used to analyze various frame structures with 26 ft (7.92 m) spans following loss of a col-
beams bridging over a lost column with different structural charac- umn is evaluated and compared. While the beam of the special
teristics in order to characterize and evaluate the effects of these frame is more ductile than a corresponding ordinary beam, the
characteristics on the performance of the beams in resisting pro- results show that the ordinary beam demonstrates a higher resis-
gressive collapse. tance to progressive collapse following loss of a column. This is
It is discussed that two peaks of the vertical force–displacement true regardless of the level of axial compressive force developed
relationship of the beam bridging over the lost column divide the in the beam (or the beam axial boundary condition). In other
curve into three phases. Perhaps counter intuitively, while just words, for all cases and up to a vertical displacement of 4 ft
after the first peak the absolute values of beam end moments still (1.22 m), at any vertical displacement, the ordinary beam provides
increase, the vertical load carrying capacity of the beam decreases. a higher vertical load carrying capacity than the special beam and
Also after the second peak, while the absolute values of beam end comes to equilibrium with zero axial force at the removed column
moments decrease, the vertical load carrying capacity of the beam location at a smaller vertical displacement. One reason for such a
increases. These are explained below by accounting for the P–D behavior is that in progressive collapse analysis of frame structures
effect. following loss of a column, the constant load previously carried by
Although the beam end moments are increasing just after the the lost column needs to be redistributed, therefore the strength
first peak, it is shown that the rate of increase in shear due to and the vertical load carrying capacity of the structure are of pri-
the beam end moments is less than the rate of reduction in shear mary importance. However, under seismic ground motion the lat-
due to the P–D effect. Similarly, during the second phase of the ver- eral loads are a function of not only the severity of the ground
tical force–displacement relationship, as the vertical displacement motion, but also the strength of the structure, and therefore the
increases, the vertical load carrying capacity of the beam decreases. structure’s ductility and deformation capacity is of primary impor-
The second phase of the response ends at the second peak of the tance. This results in ordinary frame designs being stronger but
vertical force. Beyond this peak, while the beam end moments less ductile than a comparable special frame designed for seismic
(and therefore shear due to the beam end moments) drop, the total loads.
vertical load carrying capacity of the beam increases, which can For 37 ft (11.28 m) span beams considered in this study, how-
also be explained by the P–D effect. It is shown that at this point, ever, the above conclusion is not valid. This is due to the following
because the axial compressive force in the beam drops, the P–D two facts. For a given seismicity, the longer the span the larger the
effect due to this axial force drops at a rate faster than the drop ratio of the gravity to the seismic moments is. Therefore the differ-
in shear due to the beam end moments, and therefore the vertical ence between the beam top longitudinal reinforcement at the face
load carrying capacity of the beam increases. As the vertical of the columns of special and ordinary frame will be less.
70 E. Livingston et al. / Engineering Structures 95 (2015) 61–70
Furthermore, with similar design top end reinforcements, the more [6] GSA. Progressive collapse analysis and design guidelines for new federal office
buildings and major modernization projects. Washington, DC: U.S. General
stringent detailing requirements for special frames results in
Service Administration; 2003.
higher amounts of bottom reinforcement at the beam ends (and [7] Kokot S, Anthoine A, Negro P, Solomos G. Static and dynamic analysis of a
therefore higher positive moment capacity). reinforced concrete flat slab frame building for progressive collapse. Eng Struct
The higher top reinforcement beyond the rebar cut-off location 2012;40:205–17 [Elsevier].
[8] Livingston E, Sasani M. Integrity, robustness and progressive collapse
does not considerably improve the vertical load carrying capacity resistance of RC structures designed for different levels of seismic loads.
of the beam in the first two phases of the response, but leads to Boston, MA: Structures Congress; 2014.
a higher resistance in the third phase, where the catenary action [9] OpenSEES. Open system for earthquake engineering simulation. Berkeley, CA:
Pacific Earthquake Engineering Research Center, University of California.
becomes the primary resisting mechanism, because of a larger area <http://opensees.-berkeley.edu>; 2013.
of continuous reinforcement. It is also shown that a larger yield [10] Park R, Priestley MJN, Gill WD. Ductility of square-confined concrete columns.
stress of the rebars provide a higher vertical load carrying capacity. J Struct Div ASCE 1982;108(ST4):929–50.
[11] Qian K, Li B. Slab effects on response of reinforced concrete substructures after
It should be emphasized that the conclusions drawn here are loss of corner column. Struct J ACI 2012;109(6):845–56.
valid with the assumptions and the cases studied in this paper [12] Qian K, Li B. Performance of three-dimensional reinforced concrete beam-
and should not be generalized. Furthermore, this study evaluated column substructures under loss of a corner column scenario. J Struct Eng
ASCE 2013;139(4):584–94.
the axial and flexural response of beams only and did not consider [13] Sadek F, Main JA, Lew HS, Bao Y. Testing and analysis of steel and concrete
potential shear or other modes of element and section failure. beam–column assemblies under a column removal scenario. J Struct Eng ASCE
Lastly, neither a 3D redistribution of gravity loads nor effects of 2011;137(9):881–92.
[14] Sagiroglu S. Analytical and experimental evaluation of progressive collapse
Vierendeel frame action were considered.
resistance of reinforced concrete structures. PhD dissertation. Boston, MA:
Northeastern University; 2012.
[15] Sagiroglu S, Sasani M. Progressive collapse resisting mechanisms of reinforced
Acknowledgements concrete structures and effects of initial damage locations. J Struct Eng ASCE
2014;140(3):1–12 [04013073].
This material is based upon work in part supported by the [16] Sasani M, Kazemi A, Sagiroglu S, Forest S. Progressive collapse resistance of an
actual 11-story structure subjected to severe initial damage. J Struct Eng ASCE
National Science Foundation under Grant No. CMMI-0547503. 2011;137(9):893–902.
The review and comments provided by Justin A. Murray is [17] Sasani M, Kropelnicki J. Progressive collapse analysis of an RC structure. Struct
appreciated. Des Tall Spec Build 2008;17(4):757–72.
[18] Sasani M, Sagiroglu S. Progressive collapse resistance of hotel San Diego. J
Struct Eng ASCE 2008;134(3):478–88.
References [19] Su Y, Tian Y, Song X. Progressive collapse resistance of axially-restrained frame
beams. Struct J ACI 2009;106(5):600–7.
[20] Urmson C, Mander J. Local buckling analysis of longitudinal reinforcing bars. J
[1] ACI 318. Building code requirement for structural concrete. MI: American
Struct Eng ASCE 2012;138(1):62–71.
Concrete Institute; 2011.
[21] Yap SL, Li B. Experimental investigation of reinforced concrete exterior beam–
[2] ASCE/SEI 7. Minimum design loads for buildings and other structures. Reston,
column subassemblages for progressive collapse. Struct J ACI
VA: Structural Engineering Institute-American Society of Civil Engineers; 2010.
2011;108(5):542–52.
[3] Bazan M. Response of reinforced concrete elements and structures following
[22] Yi W, He Q, Xiao Y, Kunnath SK. Experimental study on progressive collapse-
loss of load bearing elements. PhD dissertation. Boston, MA: Northeastern
resistant behavior of reinforced concrete frame structures. Struct J ACI
University; 2008.
2008;105(4):433–9.
[4] Choi H, Kim J. Progressive collapse-resisting capacity of RC beam–column sub-
[23] Yi W, Kunnath SK, Zhang F, Xiao Y. Large-scale experimental evaluation of
assemblage. Mag Concr Res 2011;63(4):297–310.
building system response to sudden column removal. Structures Congress, SEI;
[5] DOD. Design of building to resist progressive collapse. Unified facility criteria.
2011. p. 2353–7.
UFC 4-023-03. Washington, DC: U.S. Department of Defense; 2010.