Journal of Constructional Steel Research 64 (2008) 418–428

www.elsevier.com/locate/jcsr

Punching shear strength and post-punching behavior of CFT column to RC

flat plate connections
Cheol-Ho Lee a,∗ , Jin-Won Kim a , Jin-Gyu Song b
a Department of Architectural Engineering, Seoul National University, Seoul, 151-742, Republic of Korea
b Department of Architectural Engineering, Chonnam National University, Kwangju, 500-757, Republic of Korea

Received 2 February 2007; accepted 16 August 2007

Abstract

This paper summarizes full-scale test results on CFT (concrete filled tube) column to RC (reinforced concrete) flat plate connections subjected
to gravity loading. CFT construction is gaining wide acceptance due to its various structural and constructional advantages. However, efficient
details for CFT column to RC flat plate connections have not been proposed yet. Based on the strategies that maximize economical field
construction, several connecting schemes were proposed and tested in this study. Test results showed that the proposed connections can exhibit
punching shear strength and connection stiffness exceeding those of RC flat plate counterparts. A semi-analytical procedure is presented to model
the behavior of CFT column to RC flat plate connections. The parameters needed to model the behavior from elastic to post-punching range are
calibrated based on the test data of this study. The application of the proposed model to progressive collapse analysis is also illustrated.
c 2007 Elsevier Ltd. All rights reserved.

Keywords: Composite construction; CFT column; Connection; Flat plates; Progressive collapse; Punching shear

1. Introduction connections and to show the application of the proposed model

to a simplified nonlinear static progressive collapse analysis.
The use of CFT column in tall buildings is rapidly
increasing due to its advantages on structural performance 2. Testing program
and construction process over conventional RC (reinforced
concrete) construction. In Korea, using RC flat plate system This study is a part of on-going research. Gravity load
in the basement and residential floors of tall buildings is tests conducted for typical interior joints are presented in the
often mandatory to reduce story height and to enable rapid following with focusing on punching behavior. Currently cyclic
construction. The use of flat plate floors for the basement lateral load test is under preparation and will be conducted as
parking areas also minimizes the amount of excavation so that the next phase study. The connection details proposed in this
total construction time and cost can effectively be reduced. study were developed based on the following considerations:
It is expected that a combination of CFT column and RC (i) Structural integrity and the ease of construction were the
flat plate (or replacing RC column with CFT column) can primary concerns. Normally, longitudinal bars of a slab within
further enable rapid construction of such floors. However, it the width of the column need to be welded to the surface of the
appears that the issues related to connecting CFT column to steel tube at the field to develop full strength. However, such
RC flat plate have not been fully addressed yet. The primary field welding was avoided in this study due to the difficulty
objective of this study was to propose efficient and robust in quality control and to reduce construction time. Instead,
connection details for CFT column to RC flat plate based on alternatives which ensure the development of full strength and
full-scale gravity load test. The second objective was to propose easy construction were explored.
a semi-analytical model of punching behavior of the proposed (ii) The CFT column–RC flat plate connection does not have
any structural element that can reliably transfer shear forces
∗ Corresponding author. Tel.: +82 2 880 8735; fax: +82 2 871 5518. between two nonhomogeneous materials (i.e. column faces
E-mail address: ceholee@snu.ac.kr (C.-H. Lee). and slabs) while, in conventional RC flat plate connections,

c 2007 Elsevier Ltd. All rights reserved.

0143-974X/$ - see front matter
doi:10.1016/j.jcsr.2007.08.003
 C.-H. Lee et al. / Journal of Constructional Steel Research 64 (2008) 418–428 419

a full penetration type (FP type), in which longitudinal bars

Notation completely pass the CFT column through the holes on both
sides of the steel box (Fig. 1(a)), (ii) a half penetration type
The following symbols are used in this paper: (HP type), where separate longitudinal bars pass through the
holes on each side of steel box, but stop at the inner face on
Asc sectional area of post-punching bars; the other side of steel box such that the lap splice between the
Asm minimum required area of post-punching bars; bars from each side is formed (Fig. 1(b) and (c)), and (iii) a
Ast sectional area of tension bars; hook anchorage type (HK type), where longitudinal bars are
a length of an edge of the slab; hook-anchored into the hole of the tension tab which is shop-
b0 length of the critical section for punching; welded to the outside of the column (Fig. 1(d)). For FP type,
D flexural rigidity of the slab; passing longitudinal bars through the holes on both faces by
d effective depth of the slab; pushing from one side may not be easy when the size of the
Ec modulus of elasticity of concrete; column is large. Guide conduits can be used temporarily to lead
Fy measured yield strength of rebars; the rebars to the right position. HP type was intended to make
f ck compressive strength of concrete; this job easier. It is expected that the strength of the lap splice
fy yield strength of post-punching bar; within the column can be improved, in addition to bonding with
h thickness of the slab; concrete, by extra contributions provided by the out-of-plane
L span length; bending stiffness of steel plates composing the steel box. When
l1 , l2 slab span lengths in each direction; rebars are anchored at the outside of the column using standard
K1 initial secant stiffness until punching failure; hooks and the steel box is built with relatively thin plates, the
K2 negative stiffness right after punching failure; plates can deform in the out-of-plane direction due to the bar-
K3 positive stiffness for a region after K 2 and before pulling force. In this case, stud anchors were attached to the
formation of local catenary mechanism; inner face of the column so that a complete load path can be
Kc vertical stiffness of interior slab in the idealized provided (see Fig. 1(d)). For punching shear transfer from slabs
two-spring serial system; to columns, two types of shear key were considered: (i) a tee
KE vertical stiffness of exterior slab in the idealized section (ST type) and (ii) a wide flange section (SH type).
two-spring serial system;
Ke elastic stiffness of a simply supported square slab 4. Design of test specimens
at its center;
Ps axial load at the first story column; Each specimen in Figs. 1 and 2 represents a slab–column
Pu punching shear strength; joint of a parking structure with a span length of 6 m long in
Pu,post post-punching residual strength; each direction, 200-mm-deep flat plates, and 400-mm-square
wmax vertical deflection at the center of a simply columns. The slab was designed using the ACI direct design
supported square slab; method [2] to resist design loads consisting of dead and live
wu design load per unit area; load (5.88 kN/m2 ) for indoor parking area according to AIK
αc kink angle of post-punching bar; 2000 code [4]. Design compressive strength of concrete for the
αt kink angle of tension bar; specimens was assumed as 27 MPa. The column strip of the flat
δbal maximum downward deflection at energy balance plate was reinforced using D16 bars (SD40, Fy = 400 MPa)
point; with a reinforcement ratio of 0.64%.
φ strength reduction factor; The wide flange shear key was sized by following the
v Poisson’s ratio of concrete. ACI design provisions for shear heads (ACI 318-05 [2],
Corley–Hawkins [6], and Wang and Salmon [10]). A wide
structural continuity is ensured by monolithic casting of flange section of H-100 × 100 × 6 × 8 (SS400 steel, Fy =
concrete. It was speculated that a certain type of shear key is 235 MPa) with 320 mm long was required to ensure punching
required to increase both structural continuity and redundancy shear strength comparable to that of RC flat plate counterpart
in the connection regions. The punching shear strength of (see Fig. 1(c)). On the other hand, the web thickness of the
connections using this shear key was aimed to be higher than tee shear key was sized considering the shear yield limit state
that of conventional RC flat plate connections. of the tee-web against factored gravity shear (AISC [5]). The
(iii) To obtain reliable results, full-scale specimens were length of the tee shear key was simply taken to be equal to
prepared and tested. the slab depth (or 200 mm long) such that punching failure
plane is within the shear key. A built-up tee section using
3. Connection details of test specimens 9-mm-thick steel plates (SS400 steel, Fy = 235 MPa) was
attached to the column because appropriate rolled section was
Based on aforementioned rationale, connection details as not available. All welding was made using the electrode of
following were proposed (see Figs. 1 and 2). For bending tensile strength FE X X = 490 MPa. All specimens, except for
moment transfer from slabs to columns, three types of the specimens with wide flange shear key, had a stud welded
anchoring method for longitudinal bars were considered: (i) on each outer surface of the steel box at the level of tensile
420 C.-H. Lee et al. / Journal of Constructional Steel Research 64 (2008) 418–428

(a) FP-ST.

(b) HP-ST.

(c) HP-SH.

Fig. 1. Specimen set #1 (shown upside down).

reinforcing bars to delay possible separation of concrete from with a thinner wall thickness of 20 mm was to examine whether
the surface of the steel box. The welded studs for delaying or not thinner wall sections would lead to less strength and
separation were not provided to the specimens with wide flange stiffness of the connection due to weaker out-of-plane bending
shear keys due to the space limitation near the tension flange stiffness of the wall. Rolled tubular sections, much cheaper
area. The wall thickness of the steel box was 40 mm in all than welded built-up sections, are available up to 20 mm in
specimens except for the specimen HK-ST which had a 20- Korea. To investigate the impact of variations in anchorage type
mm-thick wall. The reason for including the specimen HK-ST (i.e., FP, HP, or HK type) and shear key type (i.e., tee section or
 C.-H. Lee et al. / Journal of Constructional Steel Research 64 (2008) 418–428 421

(d) HK-ST.

Fig. 1. (continued)

(a) FPP-ST.

(b) FPP-SH.

Fig. 2. Specimen set #2.

wide flange section) on the punching shear strength, a total of In addition, to see the effects of post-punching bars on
five specimens (specimen set #1) were prepared and tested (see the behavior of the slab–column connection, another three
Table 1). specimens (specimen set #2) were prepared and tested (see
422 C.-H. Lee et al. / Journal of Constructional Steel Research 64 (2008) 418–428

Table 1
Test specimens

Set # Specimen Anchorage for Shear key type

moment transfer
Set #1 1 BM-RC (benchmark – –
RC specimen)
2 FP-ST Full penetration Tee section
3 HP-ST Half penetration Tee section
4 HP-SH Half penetration Wide flange
5 HK-ST Hook Tee section
Set #2 1 BM-RCP (benchmark – –
RC specimen with
post-punching rebar)
2 FPP-ST Full penetration Tee section Fig. 3. Test setup.
3 FPP-SH Full penetration Wide flange

Table 2
Compressive strength from concrete core test

Specimen Strength assumed Measured strength

in design (MPa) (MPa)
Set #1 BM-RC
FP-ST
HP-ST 28.95
HP-SH
27
HK-ST

Set #2 BM-RCP 23.02

FPP-ST 18.12
FPP-SH 17.08 Fig. 4. LVDT installation for measurement of vertical displacement.

strength assumed in design (see Table 2). This was caused by

Table 3 the low temperature during curing. All steel material properties
Tensile test results of steel plates and deformed bars satisfied the minimum requirements as shown in Table 3.
Material-thickness or diameter Set #1 Set #2
(mm) 5. Test setup and instrumentation
Fy Fu Fy Fu
(MPa) (MPa) (MPa) (MPa) The column–slab specimens were placed upside down as
SS400-6 332 443 312 424 shown in Fig. 3 and monotonic downward vertical loading was
SS400-8 305 435 296 404 applied to the top of the column to simulate gravity load acting
SS400-9 314 483 289 364 on the slabs. Four edges of the specimen slab were simply
SM490-20 357 533 – – supported and lateral movement of the edges was restrained
SM490-40 338 495 320 511
SD40-16 447 644 442 615
to simulate the inflection lines. Rubber pads were placed to
SD40-25 – – 433 615 ensure ideal contact between the bottom of the specimen edge
and the support. Vertical displacement was measured using four
LVDTs attached at the bottom of the column as shown in Fig. 4.
Table 1). The post-punching bars for the specimen set #2 were
Strain gauges were attached to the steel elements around the
designed according to the equation in the ACI 352.1R [3], or
connection. Loading was applied gradually by displacement-
per Eq. (1) in the below.
control and overall crack patterns were recorded.
0.5wu l1l2
Asm = , (1) 6. Test results
φ fy
where wu = design load per unit area, l1 and l2 = slab span Only global behavior is presented in this paper. Fig. 5
length in each direction, φ = 0.9, and f y = yield strength of shows typical crack patterns in the tension side (or the bottom
the post-punching bar. Two D25 bars with the yield strength face) of the slab after punching failure. Observed failure mode
of 400 MPa (SD40) were placed as post-punching bars in both of all specimens agreed to what was intended. The failure
directions (see Fig. 2). mode intended in design was: (i) the yielding of flexural
Measured strengths of concrete and steel material are reinforcement should precede the yielding of shear key, and
summarized in Tables 2 and 3, respectively. In the case of (ii) punching should be the final failure mode. Early separation
specimen set #2, the compressive strengths were lower than the of the slab concrete from the steel column surface was not
 C.-H. Lee et al. / Journal of Constructional Steel Research 64 (2008) 418–428 423

(a) BM-RCP. (b) FPP-ST.

(c) HP-SH.

Fig. 5. Crack patterns on the tension (bottom) side after punching failure.

Fig. 7. Comparison of punching shear strength ratios.

Fig. 6. Load–displacement curves. punching shear strength of an RC flat plate with no shear
reinforcement. The measured compressive strengths of the
observed, which may be attributed to the steel studs attached concrete in Table 2 were used in the normalization.
on the outer surfaces of the steel box. p
Vc = 0.33 f ck b0 d [MN]. (2)
7. Punching shear strength In the ACI code, the punching shear is defined as a function
of concrete compressive strength, f ck , the length of the critical
Fig. 6 shows the measured vertical load versus downward section for punching shear, b0 , which is affected by the size and
displacement relationships for all specimens. Normalized the shape of the column, and effective depth of the slab, d. The
punching shear strengths are compared in Fig. 7. The critical section for punching shear is assumed to be located d/2
normalization was based on Eq. (2), the ACI formula for the way from the surface of the column.
424 C.-H. Lee et al. / Journal of Constructional Steel Research 64 (2008) 418–428

Fig. 9. Local catenary action after punching failure.

Fig. 8. Comparison of secant stiffness.
than conventional RC flat plates. This also implies that, for
As shown in Fig. 7, observed punching shear capacity of deflection control of the proposed CFT column–flat plates,
all CFT specimens exceeded, or at least equal to, that of the the code provisions for conventional RC flat plates can be
RC benchmark specimens (BM-RC and BM-RCP). The thinner used with conservatism. The full penetration generally showed
wall specimen HK-ST which had less out-of-bending stiffness performance superior to the half penetration in both joint
of the wall also exhibited similar connection strength. Although stiffness and punching strength. Performance of the wide flange
test data is limited, this result implies that cheaper rolled shear key was slightly better than that of the tee shear key. Post-
sections, if available, may be used in lieu of costly welded built- punching bars enhanced joint stiffness by about 40% to 80%.
up sections. Specimens with post-punching bars showed higher
punching shear strength by about 20%–40% than ones without.
9. Semi-analytical modeling of punching behavior
This is because the post-punching bars reinforced the concrete
compression zone, thus delaying the crushing of concrete. The
In this section, a model to describe the local catenary action
variations in punching shear capacity due to the difference in
after punching failure is proposed and its relevant parameters
flexural anchorage and shear key were generally minor.
are calibrated based on the test data of this study. As shown
Upon punching, the load carrying capacity of all specimens
in Fig. 9, it is presumed that, after punching failure, tension
drastically decreased (see Fig. 6). Residual strength of
bars and post-punching bars become kinked at the damaged
specimen set #1 after punching failure was about one-
regions around the column and the gravity loads are resisted
sixth to one-third of the peak strength. However, in the
by the vertical force components of these kinked bars. Further,
case of specimens with post-punching bars, the capacity
assuming that punching failure occurs uniformly around the
after the punching failure gradually increased with increasing
column and ignoring the contribution of shear key to post-
displacement, and finally reached to about two-thirds of the
punching resistance, the kink angles of tension and post-
peak value. This behavior is associated with the development
punching bars can be back-calculated by using Eq. (3) and the
of local catenary action of longitudinal bars passing through
test results of this study (see Fig. 10):
the column, which is important from the perspective of
progressive collapse prevention. Modeling of this behavior will Pu,post = 4(Ast Fy sin αt + Asc Fy sin αc ), (3)
be discussed in next section.
where Pu,post = post-punching residual strength, Ast and
8. Joint stiffness Asc = sectional areas of the tension and post-punching bars,
respectively, Fy = measured yield strength of rebars, αt
Stiffness of joints significantly impacts on the deflection and αc = kink angles of tension and post-punching bars,
of structural members. Especially, in the design of flat plates, respectively.
controlling the vertical deflection in an appropriate level is one Calibrated kink angles of tension and post-punching bars are
of the critical design considerations. The secant stiffness of all summarized in Table 4. The kink angle of tension bars (αt ) was
specimens were evaluated from the load–displacement curves first estimated by inserting the test results of specimen set #1
(see Fig. 6), and compared in Fig. 8. As the load–displacement into Eq. (3) with taking the area of post-punching bars (Asc )
curves until punching failure are slightly bilinear, the secant as zero. Next, the kink angle of post-punching bars (αc ) was
stiffness is used as the representative quantity in comparison similarly estimated by using the test results of specimen set #2
of joint stiffness. As shown in Fig. 8, specimens without post- and previously obtained αt . As shown in Table 4, estimated
punching bars (set #1), except for specimen HK-ST which had kink angles for tension and post-punching bars were about
thinner wall (or 20 mm thick) and less flexural continuity (or 9 and 10 degrees, respectively. This means that tension bars
hook anchorage), exhibited 20%–60% higher joint stiffness also almost equally participate in resisting gravity loads by
than the specimen BM-RC. Similarly, specimens with post- the local catenary action. It should be mentioned here that
punching bars (set #2) showed 20%–25% higher stiffness than loading specimen up to the available limit of the actuator stroke
the specimen BM-RCP. This implies that CFT column–RC flat was often impossible due to the tilting of the column at large
plates using the proposed connection details will less deflect deformations after punching.
 C.-H. Lee et al. / Journal of Constructional Steel Research 64 (2008) 418–428 425

(a) Specimens without post-punching bars.

Fig. 11. Idealized load–displacement relationship.

For the punching shear strength, Pu , based on the test results

of this study, a value 14% higher than that computed by using
Eq. (2) is recommended as follows:

Pu = 1.14 × (0.33 f ck b0 d) [MN].

p
(4)

The initial secant stiffness, K 1 , was first calibrated relative

(b) Specimens with post-punching bars. to the flexural stiffness of the elastic slab. When a simply
supported square slab is subjected to a concentrated load, P,
Fig. 10. Calibration model for post-punching residual strength. at its center, vertical displacement at the center is given by Eq.
(5) (per Timoshenko and Krieger [9]). “Elastic” stiffness with
Table 4 respect to applied force and vertical displacement at the center,
Calibration of kink angles for tension and post-punching bars K e , can be expressed as Eq. (6).
Specimen Max. Ppost Pu,post Ast Fy Asc Fy αt αc
Pa 2
load (kN) (kN) (kN) (kN) wmax = (0.0116) (5)
(kN) D
BM-RC 560.36 84.57 – – 7.0 – P D 1
Ke = = 2× (6)
FP-ST 596.62 98.29 – – 8.0 – wmax a 0.0116
HP-ST 563.70 116.82 – – 9.6 –
HP-SH 578.00 135.24 – – 11.1 – Ec h3
175.17 D= , (7)
HK-ST 580.55 119.07 – – 9.8 – 12(1 − ν 2 )
BM-RCP 640.23 – – – –
FPP-ST 563.60 – 404.84 438.96 – 10.1 where a = the length of an edge of the slab, D = flexural
FPP-SH 626.61 – 431.20 – 10.3 rigidity of the slab, E c = modulus elasticity of concrete,
Average 9.1 10.2
h = thickness of the slab and ν = Poisson’s ratio of concrete
(=0.17). Next, two remaining stiffness parameters, K 2 and
Generally, it is very difficult to analytically describe the K 3 , were calibrated relative to the initial secant stiffness K 1 .
behavior of CFT column–flat plates from the elastic phase Table 5 summarizes the calibration results of the three stiffness
to punching and subsequent local catenary range since the parameters obtained by using the test results of specimen set
complicated behavior of the connection, and highly nonlinear #2. The average ratios of K 1 to K e , K 2 to K 1 , and K 3 to K 1 are
and large deformation should be accounted for. Thus, a semi- about 0.33, 3.5, and 0.1, respectively. The calibrated values for
analytical modeling was tried in this study. Relevant modeling each parameter are consistent, although the number of test data
parameters were extracted and estimated by calibration with the is limited. The ratios of K 2 /K 1 , which ranged from 3.1 to 3.9,
limited test results of this study. Fig. 11 shows an idealized showed a little scatter. However, the use of the average value of
load–displacement model motivated by the test results. The 3.5 seems acceptable since this ratio does not have significant
model parameters include: (1) punching shear strength, Pu , (2) influence.
initial secant stiffness until punching failure, K 1 , (3) a negative The residual strength, Pu,post , can be obtained using Eq.
stiffness right after punching failure, K 2 , (4) a positive stiffness (3). The kink angle may be assumed as 10 degrees for both
for a region after K 2 and before formation of local catenary tension and post-punching bars until further experimental data
mechanism, K 3 , and (5) residual strength Pu,post . is obtained. Fig. 11 is the summary of the proposed model.
426 C.-H. Lee et al. / Journal of Constructional Steel Research 64 (2008) 418–428

Table 5
Calibration of stiffness parameters

Specimen Ke K1 K2 K3 K 1 /K e K 2 /K 1 K 3 /K 1
(kN/mm) (kN/mm) (kN/mm) (kN/mm)
FPP-ST 131.32 42.20 130.68 4.39 0.32 3.10 0.1
FPP-SH 127.50 44.10 173.94 4.48 0.34 3.94 0.1

Fig. 14. Deflected profile of double-span flat plate.

Fig. 12. Global catenary action.

Fig. 15. Idealization of boundary conditions of double-span flat plate.

lost according to the column removal scenario, the flat plates

around the missing column will undergo flexural deformations
to transfer the vertical load (which was resisted by the missing
column) to the adjacent columns. Thus, an inflection line
which is the border line between the positive and negative
moment regions is formed in the double-span plate. In order
to approximately locate this inflection line, linear finite element
analysis was performed using the general purpose finite element
analysis program ABAQUS [1]. The plate was modeled using
C3D8I element in ABAQUS. A vertical concentrated load was
applied to its center with assuming the slab edge as being fixed.
Fig. 14 shows the deflection profile obtained from the analysis.
Analysis results showed that the inflection line is located
approximately at a distance of 2L/3 from each edge, where L
Fig. 13. Local catenary action.
is the span length. For a simplified modeling, the whole plate
10. Application to progressive collapse analysis is subdivided into a punching-interior region and an elastic-
exterior region as shown in Fig. 15. The exterior region of the
Low-redundant flat plates and precast concrete systems are slab is assumed to be elastic as a first approximation. Then, the
considered to have a weakness associated with progressive behavior of a double-span plate can be idealized as two-spring
collapse as compared with conventional moment resisting serial systeml as shown in Fig. 16. One spring represents the
frames. Figs. 12 and 13 illustrate two possible types of load contribution of interior slab (K c ), and the other represents the
resisting mechanism when a column at the first floor is lost contribution of the exterior slab (K E ). The value for K E can
due to an abnormal event such as explosions and fire. Fig. 12 be obtained by simple linear finite element analysis of the plate
shows the more desirable mechanism in which the gravity load excluding the interior slab portion with applying line load along
is resisted through the global catenary action of entire slabs. In the inflection line (see Fig. 17). K c is estimated by the modeling
this case, a sufficient rotation capacity at all yield lines should procedure presented above. Finally, combining the two-spring
be provided by appropriate detailing. However, if the punching properties gives the load–displacement relationship of a double-
failure should precede as shown in Fig. 13, progressive collapse span flat plate as shown in Fig. 18.
should be suppressed by the local catenary action through the The following is an example of application of the proposed
kinking deformation of the top and bottom bars passing through procedure to simplified nonlinear static progressive collapse
the columns. In the case where a column at the first floor is analysis:
 C.-H. Lee et al. / Journal of Constructional Steel Research 64 (2008) 418–428 427

Table 6
Calculated parameters

Ke KE K1 K2 K3 PU Pu,post
(kN/mm) (kN/mm) (kN/mm) (kN/mm) (kN/mm) (kN) (kN)
158.8 20.63 52.4 183.4 5.24 728.9 412.88

Fig. 16. Idealized two-spring serial system for double-span flat plate.

Fig. 19. Load–deflection curve of example double-span flat plate system.

span length is 6 m long in each direction. Total dead load is

7.84 kN/m2 including self-weights, finishing, partitions and
ceilings. Design live load is 5.88 kN/m2 . Compressive strength
of concrete and yield strength of reinforcing bars are 27 MPa
and 420 MPa, respectively. Two D16 bars and two D25 bars,
both penetrating the CFT column, are assumed to be placed as
the top and bottom bars, respectively, in each direction along
the whole stories. It is also assumed that the first story column
in typical interior span is missing suddenly.
(i) Estimation of the axial load at the first story column
Ps = (6 × 6)[7.84 + 0.25(5.88)] × 4 = 1340.64 kN. (8)
All of dead load and 25% of live load are accounted for in
above calculation per Department of Defense 2004 [7].
Fig. 17. Elastic exterior region. (ii) Modeling of double-span flat plate
The parameters needed for modeling the punching-interior
region are computed based on the procedure proposed in
this study (refer to Fig. 11) and the computed results are
summarized in Table 6. Fig. 19 presents the load–displacement
curve of the double-span flat plate in this example.
(iii) Estimation of maximum downward deflection
As suggested by Powell [8], the energy balance method is a
good technical basis for a simplified nonlinear static progressive
collapse analysis. At maximum downward deflection position,
the external work done by the force at the missing column (see
Fig. 20) should be equal to the energy absorbed by the double-
span plates (see Fig. 21), that is, the energy balance equation
as expressed in Eq. (9) should be satisfied. In computing the
external work on the left-hand side of Eq. (9), full value
of the column load should be used since the load is a kind
Fig. 18. Combined behavior of a double-span plate. of suddenly applied load. It is also assumed that all the
double-span plates above the missing-column undergo the same
A four-story building with 400-mm-square CFT columns downward deflection due to the kinematic constraint provided
and 200-mm-deep flat plate floors is used as an example. The by the columns. Therefore, the internal work on the right-hand
428 C.-H. Lee et al. / Journal of Constructional Steel Research 64 (2008) 418–428

design provisions for strength and stiffness in the ACI code can
be used for the proposed connections with conservatism.
(2) Specimens with post-punching bars exhibited higher
strength by 20%–40% as compared to specimens without such
bars. A combination of full penetration of flexural reinforcing
bars and wide flange shear key exhibited best strength. Similar
to the punching shear strength, joint stiffness of all CFT
specimens using the proposed details was higher than that of
the benchmark specimens. Specimens with post-punching bars
exhibited significantly increased joint stiffness by 40%–80% as
compared to specimens without such bars.
Fig. 20. External work done by the axial force at the first story column. (3) After punching failure, the kink angles of tension and
post-punching bars were calibrated as about 9–10 degrees,
respectively. This means that the tension bars almost equally
participate in resisting external load after punching.
(4) A semi-analytical procedure was presented to model
the behavior of CFT column to RC flat plate connections.
Parameters for the proposed model were calibrated using the
limited test data of this study, and its application to simplified
progressive collapse analysis was also illustrated.

Acknowledgements

Financial supports to the first author by the Korea

Research Foundation (Grant No. KRF-2004-041-D00757) and
Fig. 21. Internal work (or absorbed energy) by the CFT column–RC flat plate
system.
the Ministry of Construction and Transportation (Grant No. 05
R&D D02-01) are gratefully acknowledged.
side of Eq. (9) should be calculated by summing the absorbed
energy of all the double-span plates above the missing column. References
Solving simple polynomial equation is sufficient to obtain the [1] ABAQUS. ABAQUS user’s manual, version 6.4. Hibbit, Kalson &
balanced deflection (δbal ). Sorenson, Inc.; 1998.
[2] ACI Committee 318. Building code requirements for structural concrete
X Z δbal
Ps × δbal = Pci dδ. (9) (ACI 318-05) and commentary (ACI 318R-05). Detroit (MI): American
0 Concrete Institute (ACI); 2005.
i
[3] ACI-ASCE Committee 352. Recommendation for design of slab-column
In this example, the energy balance point is calculated as connections in monolithic reinforced concrete structures (ACI 352.1R-
112 mm. Considering the load–deformation curve in Fig. 6, it 89). ACI Structural Journal 1988;85(6):675–96.
is expected that the CFT column to flat plate connections of this [4] Architectural Institute of Korea. AIK2000 code: Standard design loads for
buildings. 2000 [in Korean].
study can exhibit this level of deformability.
[5] AISC. Manual of steel construction: Load and resistance factor design.
3rd ed. 2001.
11. Summary and conclusions [6] Corley WG, Hawkins NM. Shearhead reinforcement for slabs. ACI
Journal, Proceedings 1968;65:811–24.
A few promising connection details for CFT column to RC [7] Department of Defense. Design of buildings to resist progressive collapse.
flat plate system were proposed and tested. The experimental Unified Faculties Criteria UFC 4-023-03. 2004.
and associated analytical results of this study are summarized [8] Powell G. Collapse analysis made easy (more or less). In: Proceedings,
as follows. Los Angeles tall buildings structural design council annual meeting. Los
Angeles. 2003.
(1) As intended, all CFT specimens using the proposed
[9] Timoshenko SP, Krieger SW. Theory of plates and shells. McGraw-Hill
details showed satisfactory punching shear strength and Book co.; 1987. p. 141–3.
connection stiffness which generally exceed those of the [10] Wang CK, Salmon CG. Reinforced concrete design. Harper & Row
reinforced concrete benchmark specimens. This implies that the Publishers; 1979.