Improving

the accuracy of progressive collapse

risk assessment: Efficiency and contribution of
supplementary progressive collapse resisting
mechanisms
A r ticle i n Computer s & Struc tures No vember 201 5

Abstract
This paper investigates the structural response of RC framed buildings subjected to
accidental/abnormal loads (explosion, impact and other hazards). Several existing
national or international design codes (GSA 2003, DoD 2009, 2013) provide limited
guidelines for the assessment of progressive collapse resistance in the design process
and the alternate load path method is widely used in current structural design codes.
Since the progressive collapse is a dynamic and nonlinear event as it takes place in a
very short time frame and the structural components undergo nonlinear deformations
before failure, in this study the nonlinear dynamic analysis (NDA) is used. As it is
known, the NDA implies significant computational power and time costs. Thus, the
first objective of this paper is to determine the influence on the accuracy of results
when the number of structures bays is successively reduced and which is the
efficiency in saving run-time costs when simplified sub-structures are considered. In
order to resist abnormal loads, the structures could be able to develop resistance
mechanisms beyond the flexural behavior. The second objective of this paper is to
numerically identify the presence and contributions of such supplementary resisting
mechanisms. Important time savings, without affecting the results accuracy, are
obtained and compressive arch action supplementary resisting mechanism is
identified.
Keywords: progressive collapse, nonlinear, dynamic, accuracy, collapse resisting
mechanisms.

Introduction

Progressive collapse represents, according to GSA(2003) Guidelines [1], a situation

where local failure of a primary structural component leads to the collapse of
adjoining members which, in turn, leads to additional collapse. This topic captured

the attention of the research community after the catastrophic event that took place
in 1968 at Ronan Point apartment building. Due to an explosion generated by a gas
leak at 18th floor, a significant part (entire south-east corner) of a 22 story apartment
building collapsed. The domain still captures the interest of the researchers, mainly
due to the relatively recent progressive collapse events: A. Murrah Federal Building
in Oklahoma City, U.S. - 1995, World Trade Centre in New York City, U.S. - 2001
(Figure 1.a), Windsor Tower in Madrid, Spain - 2005 (Figure 1.b)

Figure 1: a) World Trade Centre, U.S. b) Windsor Tower, Spain.

Progressive collapse is a dynamic and nonlinear event as it takes place in a very
short time frame and the structural components undergo nonlinear deformations
before failure. Although the progressive collapse verdict can be obtained through
three types of structural analysis (linear static analysis - LSA, nonlinear static
analysis - NSA, nonlinear dynamic analysis - NDA), the most accurate results are
reported based on three dimensional (3D) nonlinear dynamic analyses which involve
a large number of structural members (beams, columns and slabs), and after
discretization, a large number of finite elements. Applying such a procedure can lead
to significant computational power and run-time requirements even for low-rise
buildings (3-story).
All mentioned analyses types (LSA, NSA and NDA) are proposed by the main
guidelines regarding progressive collapse risk assessment: GSA(2003) [1] and
DoD(2009) [2]. Based on the provisions of these two guidelines [1, 2], the
progressive collapse potential is assessed by considering only the flexural behavior
of the structures. Recent studies [3, 4, 5, 6] have shown that the dominant load
resisting mechanism in RC framed structures is represented by a combination of
several actions: Flexural Action (FA) currently considered in the current
progressive collapse analysis, Compressive Arch Action (CAA) developed in
beams which tends to elongate as they experience large flexural deformation,

cracking and yielding, Catenary Action (CA) acting only under very large
displacements and deformations and finally, Vierendeel Frame Action (VFA)
characterized by double curvature deformations of beams, slabs and columns.
Advanced analyses should consider not only the main (classic) progressive collapse
resisting mechanism (FA) but also the contribution of certain supplementary
mechanisms that can be mobilized sequentially to mitigate progressive collapse:
CAA, CA, VFA. These three mechanisms are still not incorporated in the current
design codes or guidelines and are not familiar to structural engineers.
This paper investigates the structural response of RC framed buildings subjected
to accidental/abnormal loads (explosion, impact, terrorist attack, and other hazards).
Based on the previous remarks the present study has two main objectives. The first
objective to determine the influence on the accuracy of results when the number of
structures bays is successively reduced and which is the efficiency in saving runtime costs when simplified sub-structures are considered in the NDA. The second
objective is to identify the presence and contributions of possible supplementary
resisting mechanisms (CAA, CA) acting beyond the beam flexural capacity.
In order to maximize the results accuracy, in this paper only nonlinear dynamic
analyses (NDA) are carried out and not the other accepted procedures (LS-linear
static and NS-nonlinear static), since the general opinion expressed in the technical
literature is that NDA provides the most accurate results [7, 8, 9].

Numerical models

The typical configuration for both analyzed RC structures (3 and 9 stories) consists
of three spans and five bays of 6.0m each, with a story height of 3.15m. The design
is made according to the provisions of the Romanian seismic code P100/1-2006 [10]
which are similar to those of Eurocode 8 [11], for a low seismic zone (ag = 0.08g).
Since the positive influence of the seismic design is reduced, the most unfavorable
response of the buildings will be obtained. The cross-sectional dimensions of the
linear structural elements (beams and columns) are summarized in Table 1. The slab
influence is considered by modeling the beams as T, respectively L - shaped cross
sections, according to the provisions of ACI318 [12].
Table 1: Cross-sectional dimensions of structural elements
Beam [mm]
Column
Structure
[mm]
Interior ( T )
Exterior ( L )
3story

9story

The structures are designed using a concrete class of C25/30 (fck = 25 N/mm2),
respectively a steel class of S500 (fyk = 500 N/mm2) for both longitudinal and
transverse reinforcement.

Analysis methods

In this study, the provisions of DoD(2009) [2] regarding the progressive collapse
risk assessment of RC framed structures are applied. According to these Guidelines
the verdict is obtained by applying several missing columns scenarios. Thus, an
exterior column from the short side, an exterior column from the long side, an
interior column or corner column is successively eliminated. These scenarios must
be studied for the ground floor, for a mid-height floor, respectively for the top floor.
This paper presents only the damage case of an exterior column from the building
short side, located at the ground floor.
DoD(2009) [2] specifies a strength increase factor of 1.25 for steel, respectively
1.50 for concrete, in accordance with ASCE-41 provisions [13]. The amplified
stress-strain curves (-) are presented in Figure 2.

Figure 2: Amplified stress-strain (-) curves for concrete and steel

As it was stated before, the progressive collapse potential can be accurately
evaluated based on a nonlinear dynamic analysis (NDA). For this type of analysis
the vertical load applied downward to the structure is determined according to
Eq. (1).
Load = 1.2 D + 0.5L

(1)

where: D represents the dead load (D=3.5 kN/m2, self-weight not included);
L represents the live load (L=2.0 kN/m2).
In order to induce the dynamic effect, the vertical support is removed in less than
1/10 of the period associated with the structural response mode for the vertical
motion of the bays above the removed column, as determined from the analytical
model with the column removed [2]. Thus, a removal time of 0.005 seconds is

specified. This period also correspond to the step size. In order to allow the structure
to reach an equilibrium state, a total time of 3 seconds is considered for the analyses
along with a 5% damping factor [7, 14].
Two structural analysis software (SAP2000 and Abaqus) based on the FE method
are used in order to achieve the assumed objectives for this study.

3.1 Plastic hinge concept (SAP2000)

The plastic hinge concept can be applied in multiple ways. In the most cases found
in literature [7, 15, 16], the nonlinear analysis is based on the moment plastic hinge
concept (M3-type). In this study, this plastic hinge concept is used and compared
with the fiber type plastic hinge concept (P-M2-M3 fiber type), both available in
SAP2000 structural analysis software.
First plastic hinge type (M3-type) is defined according to ASCE-41 [13]. The
original value corresponding to Collapse Prevention (CP) state is modified
according to DoD (2009) [2] provisions to accommodate the particular behavior
associated with progressive collapse phenomenon (Figure 3). M3-type plastic hinge
can accurately surprise the pure bending behavior of the structural elements.

Figure 3. M3 plastic hinge behavior [13]

The second plastic hinge type (P-M2-M3 fiber type) involves a process of
dividing the section in multiple longitudinal fibers. For each fiber that forms the
cross section, the material nonlinear stress-strain curve is used to define the axial relationship. Summing up the behavior of all the fibers in the cross section and
multiplying by the hinge length gives the axial force-deformation and biaxial
moment-rotation relationships [17]. The adopted hinge length is 0.5 of the beam
height, according to Park and Pauley [18]. P-M2-M3 fiber type plastic hinge is able
to consider both the flexural and compressive arch actions of structures.
Both types of plastic hinges described above are used combined with onedimensional (frame) finite elements. Each element is represented by a straight line
connecting two joints. The frame element considers six degrees of freedom at both
of its connected joints [17].

3.2 Distributed plasticity concept (Abaqus)

In this study, the distributed plasticity concept is applied by using the Concrete
Damage Plasticity (CDP) model for concrete, respectively by using the Plastic
option for steel, both available in Abaqus software. The CDP model and the Plastic
option involve the precise specification of the strain-stress (-) curves for concrete
and steel; these curves are obtained based on SR EN 1992-1-1 [19] provisions. This
concept is capable to surprise complex structural resisting mechanisms such as
flexural action (FA), compressive arch action (CAA) and catenary action (CA).
The structural elements (columns and beams) are modeled using solid finite
elements (C3D20R type with 20 integration points) while the reinforcement is
modeled using linear finite elements (T3D2 type with 2 integration points, one at
each end). During the analyses, the level of plastic deformations is verified at each
material integration point [20].
The accuracy of the distributed plasticity concept, available in Abaqus software,
was previously verified by a numerical calibration of a well-known experimental
study involving a reinforced concrete two-way slab subjected to gravity loads [21].
A good agreement between the numerical and experimental results was obtained.

Progressive collapse analysis

The progressive collapse potential is determined based on the DoD(2009) failure

criteria [2]. For a nonlinear (static or dynamic) analysis the maximum plastic
rotation is computed according to Eq. (2). The allowable plastic rotation value
specified for RC beams [2] is 0.063 rad (Table 6-7, [2]).
= /L

(2)

where: represents the plastic rotation;

is the corresponding vertical deflection;
L is the clear span length.

4.1 Accuracy and efficiency

As previously mentioned, the nonlinear dynamic analysis (NDA) offers results with
a high level of accuracy concerning the progressive collapse risk. The downside of
applying such an analysis is the significant computational power and run-time that
are necessary in obtaining the verdict. This is due to the high number of finite
elements required to model the entire structure.
The first objective of the current study is to investigate the maximum level of
model reduction in order to efficiently obtain results/verdicts with a satisfactory
level of accuracy. The reduction is made successively with respect to the
longitudinal number of bays. Thus, the original model with three spans and five bays
(Figure 4.a) is firstly reduced to a sub-model with three spans and two bays (Figure

4.b) and then to a sub-model with three spans and one bay (Figure 4.c). The stiffness
of the part removed from the model is simulated by imposing boundary conditions
to the translational degrees of freedom. This technique is applied for both three and
nine-story analyzed structures.

Figure 4. Model reduction: a) original model; b) 2-bay reduced model; c) 1-bay

reduced model (3-story structure).
The considered damage scenario is applied to each original model (3 and 9-story
structure), respectively to the corresponding sub-models (3x2 bays and 3x1 bays).
For the 3-story structure a maximum vertical displacement of 7.69 cm is obtained
when the original (3x5 bays) model is analyzed. A maximum vertical displacement
of 7.56 cm is obtained when the damage case is applied to the 3x2 bays sub-model.
The corresponding value for the 3x1 bays sub-model is 8.12 cm. The equivalent
rotations obtained based on the vertical displacement values are: 0.0137 rad (3x5
bays model), 0.0135 rad (3x2 bays sub-model) and 0.0145 rad (3x1 bays submodel). According to DoD(2009) [2] failure criterion the models analyses indicate
that the structure is adequate to resist progressive collapse. To reach this conclusion
a run-time of 746 minutes for the unreduced (3x5 bays) model, 322 minutes for the
first reduce model (3x2 bays), respectively 211 minutes for the second reduced
model (3x1 bays). The time-displacement curves for the analyzed models are
presented in Figure 5.
When the 9-story structure response (Figure 6) is investigated the peak value of
the vertical displacement is 3.39 cm for the original model (3x5 bays), 3.40 cm for
the first reduced model (3x2 bays), respectively 3.37 cm for the second reduce
model (3x1 bays). These displacements values lead to a rotation of 0.0063 rad,
0.00635 rad and 0.00629 rad. Similar to the previous case, the structure is capable to
resist progressive collapse when the exterior column on the short side is suddenly
removed. This verdict is obtained with different time costs: 4502 minutes for the
unreduced (3x5 bays) model, 795 minutes for the first reduce model (3x2 bays),
respectively 427 minutes for the second reduced model (3x1 bays). A significant
increase in the run-time duration of the analyses for the 9-story models is observed
compared to the 3-story models.

Figure 5. Time-displacement curves (NDA) and analyses time for the 3-story models

Figure 6. Time-displacement curves (NDA) and analyses time for the 9-story models

4.2 Progressive collapse resisting mechanisms

The verdict regarding the progressive collapse potential based on DoD (2009) [2]
Guidelines depicts only the bending behavior of the analyzed structures.
Experimental studies [3, 4] have shown that the structures have supplementary
resistance capacities that can be developed through other mechanisms: Compressive
Arch Action (CAA) developed in beams which tends to elongate as they
experience large flexural deformation, cracking and yielding, Catenary Action (CA)
acting only under very large displacements and deformations and finally,
Vierendeel Frame Action (VFA) characterized by double curvature deformations
of beams, slabs and columns.
Numerically, the flexural behavior of structures can be represented only as a pure
flexural action (FA) or more accurately as an improved mechanism (FA + CAA) by
considering the compressive force developed over the cross section of structural
elements. This study intends to numerically identify the appearance and influence of
CAA mechanism for both 3 and 9-story structures when the exterior short side
column is suddenly removed. NDA analyses are performed considering three
different ways to take into account the effect of large inelastic deformations that
occur during the abnormal loading of the structures: Moment Plastic Hinges (MPH)
of M3-type, Fiber Plastic Hinges (FPH) of P-M2-M3 fiber type, respectively
Distributed Plasticity Concept (DP).
4.2.1 3-story structure
When the plastic hinge model (MPH) is used in the nonlinear dynamic analysis the
building is not capable to resist progressive collapse. As may be seen in Figure 7,
the structure has not reached an equilibrium state, and the vertical displacement
continually increases.
For the FPH model, the maximum vertical displacement obtained is 4.78 cm
(Figure 7). This value corresponds to a rotation of 0.0085 rad. According to the DoD
(2009) [2] failure criterion, the structure is adequate to resist progressive collapse.
When the structure is analyzed by using the DP model, a maximum vertical
displacement of 7.69 cm is obtained, which is equivalent with a 0.0137 rad rotation.
Similar to the FPH based model, the structure does satisfy the progressive collapse
resistance requirements.
At a global level, the development of the Compressive Arch Action (CAA)
enhances the bending moment capacity of the structure. In order to better illustrate
the improved behavior of the building due to the CAA effect, a supplementary
analysis is performed. Thus, a load level corresponding to 85% of the DoD(2009)
loading level (eq. 1) is applied to the MPH based model (Figure 7). For this loading
level) the structure is able to reach an equilibrium state. The maximum rotation
obtained in this case is 0.007 rad, a value that meets the acceptance criterion [2].
Consequently, when the CAA mechanism developed by the structure is taken into
account (FPH and DP), an improved resisting capacity is revealed.

Figure 7. Time-displacement curves for 3-story structure: MPH vs. FPH vs. DP
4.2.2 9-story structure
For the second analyzed structure, all three plasticity models (MPH, FPH and DP)
used in the nonlinear dynamic analysis lead to a similar verdict:

Figure 8. Time-displacement curves for 9-story structure: MPH vs. FPH vs. DP

10

the building meets the progressive collapse resistance requirements, since no

rotation exceeds the DoD(2009)[2] criterion ( = 0.063 rad).
When MPH is used, the peak value of the vertical displacement is 2.63 cm. FPH
based model indicates a maximum vertical displacement of 2.96 cm, while DP based
model reveals a 3.39 cm displacement. These displacements correspond to rotations
of 0.0049 rad (MPH), 0.0055 rad (FPH), respectively 0.0063 rad (DP).
As may be seen in Figure 8 all three plasticity models (MPH, FPH, DP) indicate
that the 9-story structure is able to resist progressive collapse under the DoD(2009)
loading level described by Eq. 1. In order to evaluate the structures capacity to
develop CAA, three supplementary analyses (Figure 9) are performed. First step of
this process is to establish the ultimate resistance capacity of the structure using of
the MPH model. The analysis indicates a maximum loading level of approximately
112% of DoD(2009) loading level; this level is reached when the flexural capacity
of beams is considered( FA). The second step is to check if the FPH and DP models
could lead to an increased loading level. Thus, 130% of the DoD(2009) loading
level is applied to both models. In this case, the structure reaches an equilibrium
state. The increase in the resisting capacity of the structure is provided by the CAA
effect.

Figure 9. Time-displacement curves for 9-story structure: MPH vs. FPH vs. DP

Conclusions

This paper investigates the structural response of RC framed buildings subjected to

accidental/abnormal loads. Firstly, the study intends to determine the influence on

11

the accuracy of results when the number of structures bays is successively reduced.
Secondly, the presence and contributions of the CAA resisting mechanism is
evaluated. The conclusions drawn for the two main objectives previously described
are summarized as follows.

Objective 1:
Both analyzed structures (3 and 9-story) are successively reduced with respect to the
longitudinal number of bays (3x2 bays and 3x1 bays) and the exterior short side
column is eliminated. The results are compared with those obtained based on the
original model (3x5 bays).
For the 3-story structure the maximum vertical displacement obtained for the first
reduced model (3x2 bays) and for the second reduced model (3x1 bays) differ with
1.7 %, respectively with 5.6 % compared to the original (3x5 bays) model. A 43.2%,
respectively 28.8% from the run-time analysis for the unreduced model is necessary
in order to obtain the structural response of the reduced models (3x2 bays and 3x1
bays).
The maximum vertical displacement, for the 9-story structure, changes with 0.3%
for the first reduced model (3x2 bays), respectively with 0.6% for the second
reduced model (3x2 bays) compared to the original model. The analysis time
decreases with 82.4% and 90.5 % for 3x2 bays model, respectively 3x1 bays model
compared to the analysis time necessary for the 3x5 bays model.
Based on the differences obtained regarding the time costs of both analyzed
structures it may be stated that the significance of the run-time analysis economy
increases with the number of floors/bays augmentation while a good accuracy level
of the progressive collapse verdict is preserved (max. 5.6% differences).

Objective 2:
In this study, three different concepts of modeling the plastic behavior during
progressive collapse phenomenon are used: Moment Plastic Hinges (MPH) of M3
type, Fiber Plastic Hinges (FPH) of P-M2-M3 fiber type, respectively Distributed
Plasticity (DP).
As may be seen in Figure 7, a major difference regarding the progressive collapse
verdict of the 3-story structure under the DoD(2009) loading level is reported. The
MPH based model indicates that the structure is not able to resist progressive
collapse. On the other hand, the FPH and DP based models indicate the opposite
conclusion. For the 9-story structure, all plastic concepts used with the NDA lead to
the same verdict.

12

The enhanced resisting capacity of the structures revealed by the FPH and DP based
models is a consequence of the CAA mechanism due to the fact that these models
(FPH and DP) are capable to consider the axial forces developed over the cross
section of the beams. In order to emphasize the CAA mechanism effect, four
supplementary analyses have been performed.
For the 3-story structure, the MPH model reveals that the ultimate resisting
capacity, based only on the flexural capacity (FA) of beams, corresponds to 85% of
the DoD (2009) loading level (Figure 7). The FPH and DP models which may also
consider the CAA, show that the structure is capable to reach an equilibrium state
for the DoD (2009) standard loading level.
Similarly, for the 9-story structure, the MPH model shows that the ultimate
capacity is reached for 112% of the DoD (2009) loading level, while the FPH and
DP models indicate that the structure is able to reach an equilibrium state for an
increased loading level (130%).
Another significant mechanism that influences the progressive collapse resistance of
the RC structure is Catenary Action (CA) mechanism. Studies in progress
investigate the parameters (reinforcement ration, number of stories, etc.) that
influence the appearance and effect of such mechanism.

References
[1]

[2]
[3]

[4]

[5]

[6]

[7]

[8]

U.S. General Service Administration, Progressive Collapse Analysis and

Design Guideline for New Federal Office Buildings and Major Modernisation
Projects, Washington D.C., 119pp, June 2003.
Department of Defense, Design of Building to Resist Progressive Collapse,
Unified Facility Criteria, UFC-4-023-03, Washington D.C., 176pp, 2009.
W.-J. Yi, Q.-F. He, Y. Xiao, S. Kunnath, Experimental Study on Progressive
Collapse-Resistant Behavior of Reinforced Concrete Frame Structures, ACI
Structural Journal, Vol. 105, No.4, p.433-439, 2008.
Q.-F. He, W.-J. Yi, Experimental Study on Collapse-Resistant Behavior of
RC Beam-Column Sub-structure considering Catenary Action, 14th WCEE,
Beijing, China, 2008.
M. Sasani, S. Sagiroglu, Progressive Collapse of Reinforced Concrete
Structures: A Multihazard Perspective, ACI Structural Journal, Vol. 105,
No.1, p.96-103, 2008.
J. Yu, K. H. Tan, Analytical model for the capacity of compressive arch
action of reinforced concrete sub-assemblages, Magazine of Concrete
Research, Vol. 66, p.109-126, 2013.
M.-H. Tsai, B.-H. Liu, Investigation of Progressive Collapse Resistance and
Inelastic Response for an Earthquake-Resistant RC Building Subjected to
Column Failure, Engineering Structures. Vol. 30, p. 3619-3628, 2008.
S. Kokot, A. Anthoine, P. Negro, G. Solomos, Static and dynamic analysis of
a reinforced flat slab frame building for progressive collapse, Engineering
Structures, Vol. 40, p.205-217, 2012.

13

[9]

[10]
[11]

[12]
[13]
[14]
[15]

[16]

[17]
[18]
[19]

[20]
[21]

A. McKay, K. Marchand, M. Diaz, Alternate Path Method in Progressive

Collapse Analysis: Variation of Dynamic and Nonlinear Load Increase
Factors, Practice Periodical on Structural Design and Construction, Vol. 17,
p.152-160, 2012.
P100-1/2006, Seismic design code Part I: Design Rules for Buildings (in
Romanian), MTCT, Bucharest, 2006.
SR EN 1998 - 1/NA, "Eurocode 8: Design of Structures for Earthquake
Resistance Part 1: General Rules, Seismic Actions and Rules for Buildings"
(in Romanian), ASRO, Bucharest, 2008.
ACI 318-05, "Building Code Requirements for Structural Concrete",
American Concrete Institute, 2005.
American Society of Civil Engineers. ASCE / SEI 41-06, Seismic
Rehabilitation of Existing Building. Virginia, 2007.
Anil K. Chopra, Dynamics of Structures - Theory and Applications to
Earthquake Engineering, Prentice Hall, Englewood Cliffs, New Jersey, 1995.
J. Kim, T. Kim, Assessment of Progressive Collapse-Resisting Capacity of
Steel Moment Frames, Journal of Construction Steel Research, Vol. 65,
p. 169-179, 2009.
A. McKay, M. Gomez, K. Marchand, Non-Linear Dynamic Alternate Path
Analysis for Progressive Collapse: Detailed Procedures Using UFC 4-023-03.
34thDDESB Seminar. Oregon, 2010.
Computers and Structures, Inc. CSI Analysis Reference Manual (SAP2000).
Berkeley, 2011.
Park R, Paulay T. Reinforced concrete structures. New York: John Wiley&
Sons, 1975.
SR EN 1992-1-1: 2004, Eurocode 2: Design of Concrete Structures - Part 11: General Rules and Rules for Buildings (in Romanian), ASRO, Bucharest,
2004.
ABAQUS, ABAQUS - Users manual, version 6.11. Pawtucket, RI: Hibbit,
Karlsson & Sorenson, 2011.
M. Botez, L. Bredean, A. M. Ioani, Distributed Plasticity Concept in
Progressive Collapse Risk Assessment of RC Structures, C60 International
Conference - Tradition and Innovation - 60 Years of Constructions in
Transilvania, 2013.

14