You are on page 1of 14

Structures 54 (2023) 1336–1349

Contents lists available at ScienceDirect

Structures
journal homepage: www.elsevier.com/locate/structures

Dynamic progressive collapse response of multi-storey frame structures


with masonry infills
Fabio Di Trapani a, *, Antonio Pio Sberna a, Marilisa Di Benedetto a, Sofia Villar a,
Cristoforo Demartino b, Giuseppe Carlo Marano a
a
Dipartimento di Ingegneria Strutturale, Edile e Geotecnica, Politecnico di Torino, Corso Duca degli Abruzzi 24, Turin 10129, Italy
b
Dipartimento di Architettura, Roma Tre University, Via Aldo Manuzio, 68L, Rome, Italy

A R T I C L E I N F O A B S T R A C T

Keywords: The response of reinforced concrete (RC) frame building structures to exceptional events like explosions or
Infilled frames impacts causing the loss of a primary structural element depends on the robustness of the structural configu­
OpenSees ration. In the case of a sudden column loss, the resisting mechanism of the involved spans substantially changes,
Progressive collapse
while a progressive collapse mechanism can develop under the arising dynamic loads. The presence of masonry
Robustness
infill walls can substantially modify the overall response to a sudden column loss scenario. In this context, this
Dynamic
Reinforced concrete paper investigates the response of reinforced concrete frame structures undergoing instantaneous column losses
in order to assess the influence of masonry infills on the progressive collapse response under dynamic load
demands. A recently formulated equivalent-strut macro-modelling approach is employed to reproduce the me­
chanical interaction between infill walls and the frames. A ten-storey, six-bays 2D reinforced concrete frame is
selected as a case study structure considering different reinforcement layouts (seismic or non-seismic design) and
column loss scenarios (central column or corner column loss). The simulations are carried out using the fiber-
section beam/column elements available in the OpenSees software platform, as they can account for the arch­
ing mechanism developing in the post-cracked regime. The dynamic responses of the case-study tests to the
sudden column loss scenarios are assessed with and without the inclusion of masonry infills within the structural
models. Results demonstrate that masonry infills introduce a substantial modification of the resisting mechanism
and of the dynamic response, limiting the propagation of progressive collapse in most of the considered cases.

1. Introduction the most critical condition potentially inducing a progressive collapse


mechanism, can be represented by the loss of one, or more, columns due
Robustness-based design of structures and infrastructures, as well as to impacts, explosions, or advanced material degradation. The damage
progressive collapse assessment, are attracting the interest of an mechanism involves the spans connected to the removed column, which
increasing number of researchers and practitioners [1–4]. The funda­ must face the additional static and dynamic load, working with a
mental principle of robustness-based design refers that, the effect of different static configuration. In general, the arising resisting mecha­
accidental damage suffered by a structural system, must not be dispro­ nism of the beams converging to the removed columns evolves in three
portionate with respect to the cause that has generated it. Unfortunately, sequential phases: a) flexure; b) post-cracking arching effect; c) catenary
structural collapses that have occurred recently and in the past (e.g. effect [5–7]. The damage evolution can be arrested within one of these
Ronan Point (London, 1968), Capitán Arenas (Barcelona, 1972), three phases if the dynamic load demand is balanced by an adequate
Argentine Israelite Mutual Association (Buenos Aires, 1994), Murrah resistance supply. Otherwise, progressive collapse occurs. Several
Federal Building (Oklahoma, 1995), Sampoong Department Store experimental tests have been recently carried out to assess the pro­
(Seoul, 1995), World Trade Center (New York, 2001), Polcevera Viaduct gressive collapse performance of reinforced concrete structural sub-
(Genoa, 2018), Champlain Towers South (Miami, 2021)) have shown assemblages [7–8] and real structures [9–10]. Further experimental
that civil structures are not sufficiently redundant when accidental studies have been also accompanied by refined numerical in­
events occur. Referring to reinforced concrete frame building structures, terpretations [7,11–13], investigating the role of geometrical and

* Corresponding author.
E-mail address: fabio.ditrapani@polito.it (F. Di Trapani).

https://doi.org/10.1016/j.istruc.2023.04.108
Received 31 October 2022; Received in revised form 13 April 2023; Accepted 26 April 2023
Available online 7 June 2023
2352-0124/© 2023 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
F. Di Trapani et al. Structures 54 (2023) 1336–1349

mechanical parameters on the progressive collapse response of rein­ modification of the damage mechanism and of the dynamic response
forced concrete structures, as well as the role of the boundary condi­ due to the presence of infills, highlighting also that progressive can be
tions. Other studies focused on the numerical simulation of the arrested because of the contribution provided by masonry infills as
progressive collapse response of 2D and 3D case-study frame structures secondary structural elements.
subject to quasi-static and dynamic simulations [14–17]. In this context
a focus on strengthening and retrofitting techniques to mitigate pro­ 2. Modelling of the frame-infill system subject to a column loss
gressive collapse was also provided [18,19]. scenario
Simultaneously to the above-referenced studies, experimental and
numerical investigations have shown that the progressive collapse 2.1. Modelling of reinforced concrete frame elements
response of frame structures is significantly modified by masonry infill
walls, becoming much more complex [20–25]. Infills provide a notice­ A 2D perimetral RC frame subject to a progressive collapse scenario
able increase in resistance of the progressive collapse mechanism that is can be effectively modelled using nonlinear distributed plasticity fiber-
also accompanied by an increase in vertical stiffness [23]. Experimental section elements. Better results are found with the displacement-based
tests and numerical simulations have clearly shown that the masonry beam/column elements, after providing a proper discretization at the
enclosed by the frames, in the proximity of the corner regions, receives ends of the beams involved in the progressive collapse mechanism [26].
an additional confining action, forming a compression stress field that In the current study, the geometrical nonlinearity under large dis­
causes the migration of the plastic hinge toward the inner of the beam placements was also considered by adopting the corotational coordinate
[21–23] (Fig. 1). Also, the diagonal compression affects masonry infills, transformation. The fiber cross-sections of RC beams elements were
resulting in an additional contribution to the overall resistance. assembled by assigning different uniaxial stress–strain laws to the con­
Proposals for a simplified equivalent multi-strut macro-modelling of crete core and cover fibers to reproduce the effect of the confinement
masonry infills have been recently developed [22,23,25–27]. These from the stirrups. The Concrete02 material model (Fig. 2a) was used.
studies proposed modelling strategies to reproduce the experimental Confined peak and ultimate stresses (fcc0 and fccu) and strains (εcc0, εccu)
pushdown responses (vertical reaction vs. vertical displacement) of and the unconfined ones (fc0, fcu, εc0, εcu) are obtained using the confined
infilled frames. In particular, the model proposed by Di Trapani et al. concrete model by Razvi and Saatcoglu (1992) [30]. The hysteretic
[26], has been extensively validated for single-storey and multi-storey response of the Concrete02 model is regulated by the parameter λ
reinforced concrete frame structures subject to pushdown tests defined by Yassin [31], namely the ratio between the unloading slope at
[24,26,28]. However, in order to perform a rigorous capacity/demand the ultimate strain and the initial elastic slope. The parameter λ was
assessment under a sudden column loss scenario, the analysis needs to assumed as 0.1, as it is common for reinforced concrete. Given the large
be unavoidably performed in a dynamic regime. This is particularly displacement demand to the beams and the consequent relevant damage
important when dealing with infilled frames, as their major resistance [7,8,11–13], steel rebars are modelled to account for a potential fracture
does not automatically ensure a safer condition, because the stiffness in tension and buckling in compression. The Hysteretic material model
increment also provides an increase in the dynamic amplification factor, backbone curve (Fig. 2b) was calibrated to effectively simulate the
besides a modification of the resisting mechanism [23]. fracture of rebars in tension, in correspondence with the achievement of
Acknowledging this lack, the paper focuses on the influence of ma­ the ultimate tensile capacity (ft, εsu), and buckling in compression [32]
sonry infills on the dynamic response of reinforced concrete frames in correspondence of the buckling point (σ*, ε*) evaluated according to
subject to sudden column losses. A ten-storey, six-bay reinforced con­ Dhakal and Maekawa [33] (Fig. 2b). The hysteretic response of steel
crete frame structure is selected as a case study. A 2D sub-frame rebars was simply modelled by assuming the stiffness of the unloading
extracted is extracted from the structural layout and modelled with branch as equal to the elastic one, while the cycles were capped but the
the OpenSees software [29]. The time-history responses to the sudden backbone curve in Fig. 2b. All the beam elements were loaded with
column removals are compared with and without including masonry distributed loads simulating dead and live loads. Finally, beam-columns
infills, which are modelled according to the recently proposed equiva­ intersection zones were modelled as rigid links.
lent stut approach by Di Trapani et al. [26], which has been applied for
the first time to a case-study structure subject to a sudden column
removal scenario. The case study tests included different column loss 2.2. Modelling of masonry infills involved in the progressive collapse
scenarios, namely the removal of a central column and the removal of a mechanism
corner column and two reinforcement layouts of the RC frame, one
obtained from the application of seismic design forces and seismic The macro-element model by Di Trapani et al. [26] is used to
design rules, the other obtained from the application of gravity loads simulate the effect of masonry infills within the portion of the frame
only, to reproduce the design practice prior to the entry in force of involved in the column loss scenarios. Numerical tests carried out in
seismic design technical codes. Results showed a substantial [26] proved that this model is capable to reproduce the progressive
collapse response of infilled reinforced concrete frames up to the large

Fig. 1. Damage pattern at the end of two experimental progressive collapse tests on infilled frames: (a) Quian and Li (2017) Specimen WNL [21]; (b) Quian and Li
(2017) Specimen WNS [21].

1337
F. Di Trapani et al. Structures 54 (2023) 1336–1349

Fig. 2. Uniaxial stress-strain model adopted for: (a) Confined and unconfined concrete in tension and compression; (b) Steel rebars in tension and compression.

displacement field. The model provides three diagonal struts per bay connection node with the S2 struts. The coefficient αb (0.15≤ αb ≤0.5)
(Fig. 3). The two side struts (called S2 struts) have infinite stiffness and identifies the portion of the net length (lb) of the first and last storey
strength. These struts are used to properly reproduce the experimentally beams where to locate the intersection point with the S2 struts. The
observed damage mechanism [21,22] in which the masonry in the definition of this points identifies two parallel lines intersecting the
proximity of the left and right sides remains almost intact, generating beams at different locations. These further intersection points identify
compression fields forcing the plastic hinges of the beams to form to­ the position of the S2 struts from storey 2 to storey n-1.
wards the inner part of the bay. The inclination of the S2 struts is vari­ As regards S1 struts, they are arranged in such a way as to connect
able and can be evaluated as specified in the following. The central strut the intersection nodes of the S2 struts with the beams for each storey.
(called S1) is an inelastic fiber-section truss with no tensile resistance. The stress-strain response of S1 struts is modelled as a concrete-type
The latter phenomenologically simulates the transverse response of the material (Fig. 3). The Concrete02 material model was used. The latter
infill undergoing diagonal compression due to the vertical displacement. provides a parabolic ascending branch with linear softening and con­
For a multi-storey infilled frame having n floors, S2 struts position fol­ stant residual strength. The stress-strain model is defined by the pa­
lows two parallel lines crossing the frame from the first storey to the last rameters ̃f (peak-strength), ̃f (ultimate-strength), εmd0 (peak-
md0 mdu
one as depicted in Fig. 3. The parameter defining their slope is the
strain), εmdu (ultimate strain), where ̃f md0 and ̃f mdu are evaluated as:
product αblb, representing the distance from the end of the beam, of the

Fig. 3. Equivalent strut model schematization of a multy-storey infilled frame subject to the central column loss according to the model by Di Trapani et al. [26].

1338
F. Di Trapani et al. Structures 54 (2023) 1336–1349

Fig. 4. Reference simplified limit equilibrium schemes: (a) beams at a generic storey; (b) struts at a generic storey.

̃f md0 = ξ⋅fmd0 ̃f mdu = ξ⋅fmdu (1) ∑ns ̃ mj ⋅tj ⋅wj


E
Ks = λj ⋅sinθj (5)
in which ξ is the strength reduction coefficient (0.25≤ ξ ≤1), considering j=1
dj
a potential reduction of the diagonal compressive strengths fmd0 and fmdu √̅̅̅̅̅̅̅̅̅̅̅̅̅̅
as a function of the effective degree of interaction between the infill and In Eqs. (3) to (5) hc, is the net height of the column, ̃f m = fm1 ⋅fm2 , is
the frame. The coefficient ξ strictly depends on the relative strength and the conventional strength of the masonry, which considers the masonry
stiffness features of the infill-frame system. The diagonal compressive strengths (fm1 and fm2) along the two orthogonal directions, fvm, is the
strengths fmd0 and fmdu and the corresponding strains εmd0 and εmdu are shear strength of the masonry, Eci and Ici are the elastic modulus and the
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
evaluated according to the infill equivalent strut model by Di Trapani moment of inertia of the beams, E ̃mj = Em1,j ⋅Em2,j , is the conventional
et al. [34] as described in the Appendix. According to [34], the Young’s modulus of the masonry, which is obtained considering the
parameter λ [31] regulating the hysteretic response of the struts masonry Young’s moduli (Em1, and Em2,j) along the two orthogonal di­
modelled with the Concrete02 material is conveniently assumed as 0.08. rections, tj and wj are the thickness and the width of the generic equiv­
Parameters αb and ξ thus regulate the definition of the macro-model. alent strut, dj is its length, while θj is its angle of inclination. The
The latter can be evaluated by means of the procedure briefly recalled in definition of wj and θj is explained in the appendix. In Eqs. (4) and (5) the
the following [26]. The parameter ξ is obtained using the following index i = 1…nb refers to the generic beam and the index j = 1…ns refers
relationship: to the generic strut. Coefficients ψ i and λj are introduced to take into
account the influence of the lateral constraint degree on the actual
ξ = 0.212⋅e1.45κ (2) stiffness of the beams and of the struts to the progressive collapse
mechanism. In particular, ψ i refers to the flexural stiffness of the
where the parameter κ is calculated as:
boundary joints and λj to the lateral stiffness contribution of the frames
√̅̅̅̅̅√̅̅̅̅̅̅
outside the progressive collapse mechanisms. The following values can
Kf lb ̃f m
κ= (3) be assumed as a function of the recognized constraint degree:
Ks hc fvm
⎧ ⎧
⎨ 12 (rigid) ⎨1 (rigid)
where, with reference to Fig. 4a, Kf is the conventional vertical stiffness ψ i = 9 (semi − rigid) λj = 0.75 (semi − rigid) (6)
⎩ ⎩
from the frame side, namely the sum of the shear stiffnesses of the nb 6 (defomable) 0.5 0 (deformable)
beams involved in the mechanism, that is:
The parameter αb is calculated from the following expression:
nb
∑ E I
Kf = ψ i ⋅ ci3 bi (4) αb = − 0.09ρ3 + 0.54ρ2 − 1.05ρ + 0.86 (7)
i=1
lbi
where ρ is the frame-infill strength ratio, defined as:
With reference to Fig. 4b, Ks is the conventional vertical stiffness
contribution provided by the ns infills involved in the mechanism at all Rf
ρ= (8)
the stories: Rs

1339
F. Di Trapani et al. Structures 54 (2023) 1336–1349

where, with reference to Fig. 4a, Rf is the conventional vertical strength Table 1
Steel and concrete nominal mechanical parameters.
from the frame side, that is the sum of the vertical resistances of the nb
beams involved in the mechanism, namely: Concrete Steel

nb fc Ec fy ft Es εsu
∑ −
Mp,i +
+ Mp,i
Rf = (9) (MPa) (MPa) (MPa) (MPa) (MPa) (%)
lb
i=1 28 31,476 450 540 210,000 12
In Eq. (9) Mp,i
+
and Mp,i− are the conventional positive and negative
plastic moments at the end of the nb beams. The latter are evaluated as a 3. Case study tests
function of the axial force acting on the beams, which depends on the
boundary conditions and on the axial force due to arching action 3.1. Design details of the reference structures
developing after the first cracking. In order to estimate the axial force on
the beams without performing the analysis, the maximum expected axial A ten-storey six-bay reinforced concrete 2D frame is considered as a
force (Nb,max) is used, that is: reference structure for the case study tests. The frame is considered as a
Nb,max = vb,max fc bwb hwb (10) perimeter frame extracted from a major 3D frame structure whose
geometrical characteristics in the plan are illustrated in Fig. 5a. The
where fc is the concrete strength, bwb, and hwb the cross-section di­ interaxis length of each span of the extracted sub-frame is 6.3 m, while
mensions, and vb,max is the expected maximum dimensionless axial force, the interstorey height of each floor is 3.4 m (Fig. 5b). At every storey the
which can be estimated as: beams and the columns have a cross-section of 300x500 and 300x800
{ mm respectively. The mechanical properties of concrete nominal
− 8.27χ 2 + 2.74χ + 0.017 (rigid) compressive strength (fc), and Youngs’ modulus (Ec) and steel nominal
νb,max = (11)
− 5.79χ 2 + 1.36χ − 0.041 (semi − rigid) yielding strength (fy), rupture strength (ft), Youngs’ modulus (Es) and
ultimate strain capacity (εsu) are reported in Table 1.
in which χ=hwb/lb, is the beam height-to-length ratio, which is related to
In order to assess the influence of the seismic reinforcement detailing
the arching action developed by the beams in the post-cracked stage.
on the progressive collapse performance of the structure, the rein­
Finally, the parameter Rs is the conventional vertical resistance contri­
forcement has been designed twice, namely with and without consid­
bution of the struts. With reference to Fig. 4b, the latter results as the
ering seismic actions and seismic detailing. In the first case, the seismic
sum of the vertical resistances of the ns struts involved in the mechanism,
design was carried out according to the current Italian Techincal Code
so that:
(NTC 2018) [35]. In the second case, no earthquake loads were

ns considered, while only the vertical dead and live loads were applied.
Rs = ̃f mdo,j ⋅tj ⋅wj ⋅sinθj (12) This condition was introduced to simulate a possible reinforcement
scenario for the structure close to a design made prior to the entry in
j=1

It should be observed that this modelling approach has been vali­ force of seismic regulations. Reinforcement details of the beams are
dated with quasi-static pushdown tests while it is here used to reproduce reported in Table 2 and Table 3 for the seismically designed and non-
the inelastic dynamics response. The reliability of the response is how­ seismically designed frames respectively.
ever considered acceptable, firstly because the single material models (e.
g. those used of steel and concrete) have been extensively validated in 3.2. Case-study tests
the past with cyclic load patterns. Moreover, progressive collapse sce­
narios generally entail oscillations in the quasi-elastic field if the damage The above-described structures were analysed using a sudden col­
propagation is arrested, while in the case of collapse, the response is umn removal approach by performing a dynamic time history analysis.
approximately monotonic. Two main cases were considered, the removal of a central column and
the removal of a corner column. For both cases, the response was
assessed with and without the presence of infills within the bays

Fig. 5. Reference case-study structure and extracted sub-frame geometrical details: (a) plan of the overall structures; (b) 2D extracted sub-frame. Dimensions in mm.

1340
F. Di Trapani et al. Structures 54 (2023) 1336–1349

Table 2
Reinforcement details of the frame with seismic design (dimensions in mm).
Frame with seismic design

Section Stories 1–2-3–4 Stories 5–6-7 Stories 8–9-10

Top reinforcement Bottom reinforcement Top reinforcement Bottom reinforcement Top reinforcement Bottom reinforcement

Ends 8 Φ 16 4 Φ 16 6 Φ 16 3 Φ 16 4 Φ 16 3 Φ 16
Longitudinal
Midspan 3 Φ 16 3 Φ 16 3 Φ 16 3 Φ 16 3 Φ 16 3 Φ 16

Ends Φ8/110
Transverse
Midspan Φ 8/220

Table 3
Reinforcement details of the frame with no seismic design (dimensions in mm).
Frame with seismic design

Section Stories 1–2-3–4 Stories 5–6-7 Stories 8–9-10

Top reinforcement Bottom reinforcement Top reinforcement Bottom reinforcement Top reinforcement Bottom reinforcement

Ends 8 Φ 16 4 Φ 16 6 Φ 16 3 Φ 16 4 Φ 16 3 Φ 16
Longitudinal
Midspan 3 Φ 16 3 Φ 16 3 Φ 16 3 Φ 16 3 Φ 16 3 Φ 16

Ends Φ8/110
Transverse
Midspan Φ8/220

Table 4
Case-study tests.
Case study Infills Column removal Seismic Design

CEN-BF-S No Central Yes


CEN-IF-S Yes Central Yes
CEN-BF-NS No Central No
CEN-IF-NS Yes Central No
COR-BF-S No Corner Yes
COR -IF-S Yes Corner Yes
COR -BF-NS No Corner No
COR-IF-NS Yes Corner No

involved in the column loss scenario. A summary of the tests is provided 3.3. Definition of the macro-model for the infilled case studies
in Table 4. The nomenclature of each test is divided into three parts. The
first part denotes the removed column, central (CEN), or corner (COR). The definition of the equivalent strut macro-model for the IF cases
The second part denotes the presence of the infills in the model, namely was carried out according to the procedure described in the previous
if the bays are bare frames (BF) or infilled frames (IF). The third part section. For the central column loss case, it was reasonably assumed that
denotes the reinforcement design, that is seismic design (S) or non- the lateral stiffness provided by the frames at the left and right sides of
seismic (NS). A scheme of the main case studies investigated is also the bays involved in the mechanism, and so the lateral constraint degree,
illustrated in Fig. 6 was progressively reducing with the height of the frame [26]. In the
Masonry infills are supposed made of clay masonry hollow units with corner column loss case, only one bay is involved in the mechanism. The
a thickness of 300 mm. Mechanical properties of the infills are reported latter, has a low horizontal displacement restraint with respect to the
in Table 5. central bay case. Therefore, a lower constraint degree was assumed for

Table 5
Mechanical properties of the masonry constituting the infills.
f m1 f m2 ̃f f vm Em1 Em2 ̃m
E
m
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa) (MPa)

4.18 8.66 6.02 1.07 5032 6401 5675

Table 6
Lateral constraints assumed at the different storeys for the infilled cases studies.
Test reference Storeys 1–2-3–4 Storeys 5–6-7 Storeys 8–9-10

Left constraint Right constraint Left constraint Right constraint Left constraint Right constraint

CEN-IF-S Rigid Rigid Semi-Rigid Semi-Rigid Deformable Deformable


CEN-IF-NS Rigid Rigid Semi-Rigid Semi-Rigid Deformable Deformable
COR-IF-S – Deformable – Deformable – Deformable
COR-IF-NS – Deformable – Deformable – Deformable

1341
F. Di Trapani et al. Structures 54 (2023) 1336–1349

Table 7
Geometrical details of the infilled frames and of the S1 struts.
hc hc’ lb lb’ d bwc hwc bwb hwb t w
(mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm)

2900 3400 5500 6300 7159 400 800 300 500 300 706.3

Table 8
Parameters for the determination of the geometrical and mechanical properties of S1 and S2 struts.
Test reference θj Kf Ks κ ξ ̃f
md0
̃f
mdu
εmd0 εmdu Rf Rs ρ αb
(◦ ) (kN/m) (kN/m) - - (MPa) (MPa) - - (kN) (kN) - -

CEN-IF-S 38.82 109,964 1,632,365 0.22 0.29 0.47 0.12 0.000839 0.00794 2099 1116 1.88 0.19
CEN-IF-NS 38.82 95,775 1,421,737 0.22 0.29 0.47 0.12 0.000839 0.00794 1169 1116 1.05 0.25
COR-IF-S 38.82 54,982 816,182 0.22 0.29 0.47 0.12 0.000839 0.00794 1050 558 1.88 0.19
COR-IF-NS 38.82 47,887 710,868 0.22 0.29 0.47 0.12 0.000839 0.00794 584 558 1.05 0.25

all the storeys. The assumptions made for the lateral constraint degree 4. Dynamic progressive collapse analyses
for the considered case studies are summarized in Table 6. The corre­
sponding values of coefficients ψ j and λ j are obtained by Eq. (6). 4.1. Definition of the analysis
For all the considered cases, the infilled frames have the same
geometrical and mechanical properties, leading to the same widths for The dynamic progressive collapse analysis is made in OpenSees by
the equivalent struts (w) according to the model by [34] as modified by performing a time history analysis with a fictitious constant zero
[23]. Geometrical data of the infilled frames are reported in Table 7 external acceleration. During the time history analysis, the element
together with those of the equivalent S1 struts. The determination of the removal OpenSees command is activated to suddenly remove a column
coefficients αb and ξ, useful to define the inclination of S2 struts and the element. The definition of the nodal masses is performed from gravity
stress-strain response of S1 struts, follow the approach described by Eqs. loads as a function of the tributary lengths of influence (LT). Nodal
(1) to (11). The main parameters resulting from the application of the masses are reduced by 50% to simulate the same dynamic effect as the
procedure are reported in Table 8. A schematic view of the macro- distributed loads (Fig. 8). In this way, considering that the distributed
models defined for the central column loss and corner column loss sce­ load on the beams (q) for the infilled frame cases is 25 kN/m, the
narios is shown in Fig. 7.. equivalent mass (m) is equal to qLT/2g. As it can be observed from Fig. 8a
and b, LT is equal to l’b for the central column loss case and l’b/2 for the

Fig. 6. Schemes of the main analysed case studies: (a) bare frame central column loss; (b) bare frame corner column loss; (c) infilled frame corner column loss; (d)
infilled frame corner column loss.

1342
F. Di Trapani et al. Structures 54 (2023) 1336–1349

Fig. 7. Equivalent strut macro-models for: (a) central column loss mechanism; (b) corner column loss mechanism.

corner column loss case, so that one obtains m=qlb’/2g = 8.03 t and variation on the standing columns versus the displacement and the time,
m=qlb’/4g=4.01 t as equivalent masses respectively. As regards the bare and the vertical displacement versus time. The static pushdown curves
frames, a slightly lower load was considered (q=22 kN/m) to take into are also displayed in background in Fig. 9a and b. It can be firstly
account the effect of the openings on the partition walls. The resulting appraised that both for the seismically and the non-seismically designed
masses are 7.06 t and 3.53 t for the central and corner column loss cases frames, the contribution of the infills radically modified the responses.
respectively. The analyses are carried out for the case studies listed in In fact, while the removal of the central column caused the collapse of
Table 4, recording the variation of the sum of the base vertical load the central bays of the bare frames under large vertical displacements
reactions of the remaining columns and of the vertical displacement (Fig. 9e and f), the infilled frame bays underwent only an oscillation. In
over time. During the analysis, the variation of the internal forces in this latter case, the dynamic load demand stabilized around the value of
some critical cross-sections of the beams involved in the mechanism is the static load previously borne by the central column (Fig. 9c and d).
also recorded. The strength increment due to the infills could be appreciated only for
the non-seismically designed frame. This was approximately +25% of
the strength of the bare frame. For the case of the seismically designed
4.2. Analysis results: Central column loss scenario frame, the strength increment was not recognizable as the oscillations
were arrested in the quasi-elastic stage (Fig. 9a).
Results from the analysis of bare and infilled frame case study tests The vertical cracked stiffness increment due to the presence of infills
subject to the central column loss scenario are compared and discussed ranged from 2.7 to 4.5 times that of the bare frame for the NS and S cases
in this section. The four tests (CEN-BF-S, CEN-IF-S, CEN-BF-NS, CEN-IF- respectively. This highlights, as expected, a major influence of the infills
NS) also included the frames with and without seismic detailing. Results on the response of the non-seismically designed frames. As regards the
of the analyses are illustrated in Fig. 9 in terms of base axial load

LT qLT LT
m qLT
q g m
g

l’ b l’ b l’ b

Fig. 8. Definition of the computational scheme for the progressive collapse dynamic analysis for: a) central column loss scenario; b) corner column loss scenario.

1343
F. Di Trapani et al. Structures 54 (2023) 1336–1349

Fig. 9. Progressive collapse analysis results from the central column loss scenario: (a) load vs. displacement for BF-S and IF-S; (b) load vs. displacement for BF-NS and
IF-NS; (c) load vs. time for BF-S and IF-S; (d) load vs. time for BF-NS and IF-NS; (e) displacement vs. time for BF-S and IF-S; (f) displacement vs. time for BF-NS and IF-
NS. (The blu arrows denote the collapse).

collapse mechanisms, the divergence of the displacement curves in much more limited. Residual displacements at the end of the oscillations
Fig. 9e and f, clearly individuate the collapse of the bays connected at all were about 15 mm and 300 mm for the S and NS cases respectively. Both
storeys with the central column. For the infilled frames, the damage was displacements are consistent with a possible reparation intervention, as

Fig. 10. Deformed shapes at the same time frame of the analysis for and monitored cross-sections (C and L): (a) bare frame (CEN-BF-S); (b) infilled frame (CEN-IF-S).

1344
F. Di Trapani et al. Structures 54 (2023) 1336–1349

(a) (b) (c)

(d) (e) (f)


Fig. 11. Response of C and L sections during the central column loss scenario analysis: (a) bending moment vs. displacement for BF-S and IF-S; (b) bending moment
vs. displacement for IF-S; (c) axial-force/bending moments trajectories withing the resistance domain BF-S and IF-S; (d) bending moment vs. displacement for BF-NS
and IF-NS; (e) bending moment vs. displacement for IF-NS; (f) axial-force/bending moments trajectories withing the resistance domain BF-NS and IF-NS.

they result in 0.13% and 0.26% of the length of the bay. For the NS case, represented in background in Fig. 12a and b. It is noteworthy observing
a possible reparation intervention should be extended also to the infills, that, in this case, although the contribution of the infills was still rele­
as they achieved their peak resistance (Fig. 9b and d) and so significant vant, collapse of the corner bays could be avoided only for the frame
cracks could be expected. The magnified deformed shapes at the same with seismic design. For the NS-infilled corner bays, strength increment
time frame of the analysis are compared in Fig. 10 for bare (CEN-BF-S) due to infills was about +50% with respect to the bare frame (Fig. 12b
and infilled (CEN-IF-S) frames. The latters qualitatively illustrate the and d). Despite this, the displacement demand was high enough to bring
extent of the damage in the two cases. the infill S1 struts reaching the softening branch, where the consequent
Two reference control cross-sections were monitored during the strength reduction did not allow achieving a new equilibrium configu­
analysis. These are located at the ends of the beams involved in the ration possible (Fig. 12d).
mechanism and named C-section (central) and L- section (lateral) The collapses of the bare S and NS frames and of the infilled NS frame
(Fig. 10). The bending moments, and the axial-force/bending moment are evident from the divergence of the displacement curves in Fig. 12e
trajectories were monitored during the analyses in sections C and L. and f. It should be noted that for the IF-NS case, the collapse is delayed
Results are illustrated in Fig. 11. They highlight that the presence of the because of the presence of the infills until the resistance decay becomes
infills reduced the flexural demand on the beams, therefore, no signifi­ significant (Fig. 12b, d and f). As regards the infilled seismically
cant damage, in terms of inelastic displacement demand, was experi­ designed corner frame, the collapse was arrested because of the infills.
enced for the IF-S case (Fig. 11a and c), while limited damage could be The residual displacement at the end of the oscillations was about 20
recognized for the IF-NS. For the latter cases, the cross-sections had a mm, which is 0.33% the length of the bay. In this case, a reparation
reduced flexural capacity, especially section C, which was subjected to intervention is possible and should include also the reparation of the
the inversion of the bending moment sign (Fig. 11d – f). Also, the axial infills. The magnified deformed shapes of the seismically designed bare
force excursions due to the arching action affecting the BF cases were and infilled frames at the same time frame of the analysis are also
reduced by the presence of the infills because of a major redistribution of compared in Fig. 13. The latter qualitatively repented the magnitude of
the internal forces. Fig. 11a and d clearly highlighted that the effect of the damage mechanisms for the two cases.
the compressive axial force or the BF cases significantly reduced the The reference control cross-sections monitored during the analysis,
ductility capacity of the cross sections, causing a rapid loss of bearing are located at the ends of the first storey beam and named C-section
capacity. Therefore, infills played a double favourable role, both (central) and L- section (lateral) (Fig. 13). Results are illustrated in
reducing axial-force and bending moment demands. Fig. 14. They highlight also that, in this case, the presence of the infills
reduced the flexural demand on the beams. This was quite evident for
4.3. Analysis results: Corner column loss scenario the infilled S cases (Fig. 14a – c). Despite this reduction, failure in the NC
cases occurred both with and without the infills (Fig. 14d and e). It
The corner column loss scenario represents the most critical condi­ should be also observed that the compressive axial force on the beams is
tion because of the lower structural redundancy. Results of the analyses negligible (Fig. 14c and f) because of the reduced arching response of the
for the four tests (COR-BF-S, COR-IF-S, COR-BF-NS, COR-IF-NS) are re­ beams. This led to a more ductile response of the cross-section combined
ported in Fig. 12. Also in this case the static pushdown curves are with lower resistance (Fig. 14a and d). Concluding, the influence of

1345
F. Di Trapani et al. Structures 54 (2023) 1336–1349

Fig. 12. Progressive collapse analysis results from the central column loss scenario: (a) load vs. displacement for BF-S and IF-S; (b) load vs. displacement for BF-NS
and IF-NS; (c) load vs. time for BF-S and IF-S; (d) load vs. time for BF-NS and IF-NS; (c) displacement vs. time for BF-S and IF-S; (d) displacement vs. time for BF-NS
and IF-NS. (The blu and red arrows denote the collapse). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version
of this article.)

Fig. 13. Deformed shapes at the same time frame of the analysis and monitored cross-sections (C and L) for: (a) seismically designed bare frame (COR-BF-S); (b)
seismically designed infilled frame (COR-IF-S).

1346
F. Di Trapani et al. Structures 54 (2023) 1336–1349

Fig. 14. Response of C and L sections during the corner column loss scenario analysis: (a) bending moment vs. displacement for BF-S and IF-S; (b) bending moment
vs. displacement for IF-S; (c) axial-force/bending moments trajectories withing the resistance domain BF-S and IF-S; (d) bending moment vs. displacement for BF-NS
and IF-NS; (e) bending moment vs. displacement for IF-NS; (f) axial-force/bending moments trajectories withing the resistance domain BF-NS and IF-NS.

infills for the corner columns cases can be helpful to avoid progressive portion of the load, and so reducing the flexural demand to the
collapse only if the beam cross sections at the ends are equipped with beams.
adequate flexural overstrength. • The considered case studies have shown that the presence of ma­
sonry infills can arrest the damage propagation and avoid the acti­
5. Conclusions vation of the progressive collapse mechanism both for the seismically
designed and non-seismically designed frames. However, the most
Progressive collapse analysis of reinforced concrete frame structures critical column loss configurations, such as the corner column loss,
due to exceptional events inducing column losses deserves special care can induce severe cracking of the infill in the case of poor over­
in modelling and assessment, especially in consideration of the high strength of the beams. In these cases, as it occurred for the non-
mechanical and geometrical nonlinearity of the problem. Masonry infills seismically designed frame, the loss of bearing capacity of the
cannot be neglected in the modelling and assessment phases because of infills was too high and the progressive collapse mechanism could
their strong interaction with primary reinforced structures. The paper not be arrested.
investigated the dynamic progressive collapse response of infilled • For the cases in which the progressive collapse was avoided, the
frames by implementing a recently developed macro-element model for residual damage in terms of vertical displacement was limited, so,
the infills. Comparative assessments were carried out for a ten-storey localized repairing is possible.
reinforced concrete frame subjected to the sudden loss of a central col­ • Infill contribution cannot be neglected for the assessment of struc­
umn and of a corner column. The analyses were carried out with and tural robustness of RC frames. The development of new design
without including the mechanical contribution of masonry infill walls. methods for structurally redundant structures could effectively
Two reinforcement layouts of the frame were considered, namely a consider the infills as secondary elements activating in the case of
seismically designed reinforcement and a non-seismically designed column loss scenario.
reinforcement. Overall, eight case study tests were carried out. From the • This study investigated the effect of solid infills within a 2D frame.
observation of the results the following conclusions can be drawn: Further investigation is needed to assess the behaviour of 3D frame
including also infill with openings. Also.
• Masonry infills within the reinforced concrete frames involved in a
sudden column loss scenario strongly interact with primary struc­ Declaration of Competing Interest
tures, providing additional strength and stiffness and modifying the
progressive collapse mechanism. The authors declare that they have no known competing financial
• The influence of infills provides generally a favourable contribution interests or personal relationships that could have appeared to influence
in the case of a sudden column loss as they tend to sustain a big the work reported in this paper.

Appendix A. Evaluation of the equivalent strut reference properties according to [26]

To model the S1 strut cross-section dimensions and material properties, the procedure provided by Di Trapani et al. [26] can be applied. This

1347
F. Di Trapani et al. Structures 54 (2023) 1336–1349

approach is a modification of the procedure by Di Trapani et al. [34], originally developed to define the equivalent struts of the infills subject to
horizontal seismic loads. According to this model, the struts have the same thickness as the infills, while the width (w) of the strut is evaluated as:
hc c*
w= d (13)
lb λ* β*

√̅̅̅̅̅̅̅̅̅̅̅̅̅̅
where hc and lb are the interaxis length of the column and of the beam, and d is the length of the diagonal strut obtained as d = hc + lb . Parameters c*
and β* are evaluated by:
c* = 0.249 − 0.0116 ν + 0.567 ν2 ≅ 0.254
(14)
β* = 0.146 − 0.0073 ν + 0.126 ν2 ≅ 0.147

where ν is the Poisson’s ratio of the infill along the diagonal direction, which can be set as 0.1. The parameter λ* is evaluated by means of the formula
initially proposed by Papia et al. [36], where the positions of lb and hb are inverted to take in consideration the different load directions and using the
conventional Young’s modulus (E ̃m ) so that:
( )
̃ m tl′b l′b 2 1 Ab l′c
E
λ* = ′ 2 + (15)
Ec Ab lc 4 Ac lb

there the symbols Ab and Ac represent the area of the cross-sections of beams and columns, Ec is the elastic modulus of the concrete frame, while the
other parameters have been already defined.
The stress-strain response of S1 strut passes through the definition of the four parameters fmd0, fmdu, εmd0, εmdu. According to [34] these are obtained
as:

fmd0 = ̃f m ⋅26.9⋅α− 0.28


(16)

fmdu = fmd0 ⋅(0.043⋅β − 0.06) (17)

εmd0 = εm0 ⋅3.024⋅γ0.347 (18)

εmdu = εmd0 ⋅0.0184⋅δ− 1.166


(19)

where parameters α, β, γ, and δ can be obtained as:


̃f 2 ⋅w⋅t
α= m
0.2 (h /l )⋅λ*0.2
(20)
fvm c b

0.7
fmd0 ⋅w⋅t
β= 0.2
(21)
̃
Em d
⎛ ⎞
( 2
)
fmdu ⎝ Ec ⎠
γ= (22)
fmd0 ̃ 1.5
E m

̃ 0.20
δ=E m ⋅εmd0 (23)

in which εm0=0.0015, while and the influence of vertical loads is neglected and the position of lb and hb is inverted. For the case studies considered in
the paper, the values of parameters α, β, γ, δ and λ* are reported in Table 9.

Table 9
Parameters for the determination of the geometrical and mechanical properties of S1 strut.
α β γ δ λ*

9.482E + 06 7.331 0.00772 0.004726 7.944

Appendix B. Evaluation of the conventional angle of inclination (θj) of the S1 strut

θj is the angle of inclination of the S1 strut, which depends on αb (Fig. 3), which is still depending on θj. According to [26], the determination of θj
can be thus performed iteratively, or more simply, this can be assumed as the average between the maximum and minimum slopes (θmax and θmin
respectively) resulting by assuming αb=0.5 and αb=0.15, so that:

1348
F. Di Trapani et al. Structures 54 (2023) 1336–1349

θmax + θmin
θj = (24)
2
where supposing that all the n stories have the same interstorey height (h’c), it can be easily found that:
( ′ )
nhc
θmax = arctan ′ (25)
nlb − (n − 1)(0.5lb + 0.5hwc )
( ′ )
nhc
θmin = arctan ′ (26)
nlb − (n − 1)(0.15lb + 0.5hwc )

References [18] Kiakojouri F, De Biagi V, Chiaia B, Sheidaii MR. Strengthening and retrofitting
techniques to mitigate progressive collapse: A critical review and future research
agenda. Eng Struct 2022;262:114274.
[1] Adam JM, Parisi F, Sagasetac J, Lu X. Research and practice on progressive collapse
[19] Liu X-Y, Qin W-H, Xu Z-D, Xi Z, Zhang Z-C. Investigation on the progressive
and robustness of building structures in the 21st century. Eng Struct 2018;173:
collapse resistance of three-dimensional concrete frame structures reinforced by
122–49.
steel-FRP composite bar. J Build Eng 2022;59:105116.
[2] Kunnath SK, Bao Y, El-Tawil S. Advances in Computational Simulation of Gravity-
[20] Shan SD, Li S, Xu SY, Xie LL. Experimental study on the progressive collapse
Induced Disproportionate Collapse of RC Frame Buildings. J Struct Eng 2018;144
performance of RC frames with infill walls. Eng Struct 2016;111:80–92.
(2):03117003.
[21] Qian K, Li B. Effects of Masonry Infill Wall on the Performance of RC Frames to
[3] Marchand KA, Stevens DJ. Progressive Collapse Criteria and Design Approaches
Resist Progressive Collapse. J Struct Eng 2017;143(9):04017118.
Improvement. J Perform Constr Facil 2015;29(5):B4015004.
[22] Li S, Kose MM, Shan S, Sezen H. Modeling Methods for Collapse Analysis of
[4] Kiakojouri F, De Biagi V, Chiaia B, Sheidaii MR. Progressive collapse of framed
Reinforced Concrete Frames with Infill Walls. J Struct Eng 2019;145(4):04019011.
building structures: Current knowledge and future prospects. Eng Struct 2020;206:
[23] Di Trapani F, Giordano L, Mancini G. Progressive Collapse Response of Reinforced
110061.
Concrete Frame Structures with Masonry Infills. J Eng Mech 2020;146(3):
[5] Izzuddin BA, Vlassis AG, Elghazouli AY, Nethercot DA. Progressive collapse of
04020002.
multi-storey buildings due to sudden column loss. Part I: Simplified assessment
[24] Pang Bo, Wang F, Yang J, Zhang W, Huang X-H, Azim I. Evaluation on the
framework. Eng Struct 2008;30(5):1308–18.
progressive collapse resistance of infilled reinforced concrete frames based on
[6] Vlassis AG, Izzuddin BA, Elghazouli AY, Nethercot DA. Progressive collapse of
numerical and semi-analytical methods. Eng Struct 2022;267:114684.
multi-storey buildings due to sudden column loss. Part II: Application Eng Struct
[25] Shan S, Li S, Wang S. Effect of infill walls on mechanisms of steel frames against
2008;30(5):1424–38.
progressive collapse. J Constr Steel Res 2019;162:105720.
[7] Weng J, Lee CK, Tan KH, Lim NS. Damage assessment for reinforced concrete
[26] Di Trapani F, Tomaselli G, Cavaleri L, Bertagnoli G. Macroelement model for the
frames subject to progressive collapse. Eng Struct 2017;149:147–60.
progressive-collapse analysis of infilled frames. J Struct Eng 2021;147(6):
[8] Ren P, Li Y, Lu X, Guan H, Zhou Y. Experimental investigation of progressive
04021079. https://doi.org/10.1061/(ASCE)ST.1943-541X.0003014.
collapse resistance of one-way reinforced concrete beam–slab substructures under
[27] Qian K, Lan DQ, Li SK, Fu F. Effects of infill walls on load resistance of multi-story
middle-column-removal scenario. Eng Struct 2016;118:28–40.
RC frames to mitigate progressive collapse. Structures 2021;33:2534–45.
[9] Xiao Y, Kunnath S, Li FW, Zhao YB, Lew HS, Bao Y. Collapse test of three-story half-
[28] Feng D-C, Zhang M-X, Brunesi E, Parisi F, Yu J, Zhou Z. Investigation of 3D effects
scale reinforced concrete frame building. ACI Struct J 2015;112(4):429–38.
on dynamic progressive collapse resistance of RC structures considering slabs and
[10] Stathas N, Karakasis I, Strepelias E, Palios X, Bousias S, Fardis MN. Tests and
infill walls. J Build Eng 2022;54:104421.
analysis of RC building, with or without masonry infills, for instant column loss.
[29] McKenna F, Fenves GL, Scott MH. Open system for earthquake engineering
Eng Struct 2019;193:57–67.
simulation. Berkley: University of California; 2000.
[11] Yu J, Tan KH. Experimental and numerical investigation on progressive collapse
[30] Razvi SR, Saatcioglu M. Strength and Ductility of Confined Concrete. J Struct Eng
resistance of reinforced concrete beam column subassemblages. Eng Struct 2013;
1992;118(6):1590–607.
55:90–106.
[31] Yassin MHM. Nonlinear Analysis of Prestressed Concrete Structures under
[12] Pham XD, Tan KH, Yu J. A simplified approach to assess progressive collapse
Monotonic and Cycling Loads. Berkeley: University of California; 1994. PhD
resistance. Eng Struct 2015;101:45–57.
dissertation.
[13] Pham AT, Tan KH, Yu J. Numerical investigations on static and dynamic responses
[32] Ferrotto M, Cavaleri L, Di Trapani F. FE modeling of Partially Steel-Jacketed (PSJ)
of reinforced concrete sub-assemblages under progressive collapse. Eng Struct
RC columns using CDP model. Comput Concr 2018;22(2):143–52.
2017;149:2–20.
[33] Dhakal RP, Maekawa K. Modeling for Postyield Buckling of Reinforcement. J Struct
[14] Vinay M, Kodanda Rama Rao P, Dey S, Swaroop AHL, Sreenivasulu M,
Eng 2002;128(9):1139–47.
Venkateswara RK. Evaluation of progressive collapse behavior in reinforced
[34] Di Trapani F, Bertagnoli G, Ferrotto MF, Gino D. Empirical equations for the direct
concrete buildings. Structures 2022;45:1902–19.
definition of stress-strain laws for fiber-section based macro-modeling of infilled
[15] Lu JX, Wu H, Fang Q. Progressive collapse of Murrah Federal Building: Revisited.
frames. J Eng Mech 2018;144(11):04018101.
J Build Eng 2022;57:104939.
[35] Ministerial Decree 17/01/2018 (NT 2018). Nuove norme tecniche per le
[16] Brunesi E, Nascimbene R, Parisi F, Augenti N. Progressive collapse fragility of
costruzioni. 2018.
reinforced concrete framed structures through incremental dynamic analysis. Eng
[36] Papia M, Cavaleri L, Fossetti M. Infilled frames: developments in the evaluation of
Struct 2015;104:65–79.
the stiffening effect of infills. Struct Eng Mech 2003;16(6):675–93.
[17] Arshian AH, Morgenthal G. Three-dimensional progressive collapse analysis of
reinforced concrete frame structures subjected to sequential column removal. Eng
Struct 2017;132:87–97.

1349

You might also like