Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
A R T I C L E I N F O A B S T R A C T
Keywords: An innovative high-performance precast beam-to-column joint for multi-storey framed structures is introduced.
Prefabricated high-strength composite columns The proposed solution consists of coupling between Composite Steel Truss Concrete (CSTC) beams with a
Experimental investigation concrete base and High-Strength Concrete (HSC) columns. This paper focuses mainly on the static performance
Finite element analysis of prefabricated HSC columns, including experimental and numerical investigations. The finite element model of
Strength domain
the structural system is calibrated on experimental static testing evidence. On the basis of the data collected, the
Analytical approach
finite element model is used to evaluate the strength domain of members and optimize the related coupling
system between columns. Aiming to limit time-consuming analyses and provide practical rules for design, an
analytical simplified procedure is introduced to compute the assembly’s strength domain. An analytical approach
is adopted to estimate the reference design strength for the considered composite columns, accounting for the
partial safety factor imposed by codes and allowing for estimation of the limit on the number of storeys for
frames adopting the proposed precast technology.
1. Introduction Zealand and Japan during the 80s and consists of a capacity-design-
approach applied to precast structures. Connections are designed to
The use of prefabricated concrete components and their related restore the performance of the equivalent monolithic system [8]. De-
coupling systems in seismic engineering constitutes a subject of wide tails of different types of connections have been discussed extensively
and deep interest among researchers, practitioners and manufacturers by Park [9] and FIB Bulletin 43 [10] and more recently by Bhatt and
all over the world. As demonstrated by a large number of studies con- Kirk [11], Seckin and Fu [12], and French et al. [13,14].
ducted in numerous countries, especially in Japan, New Zealand and Opposite to emulative is the dry joint precast approach, which
the United States since the early 1980s and, in relatively more recent consists of a moving inelastic response from members to connections.
times, in Italy. These are detailed to be weaker than the precast elements and are in-
The reference layout for traditional precast structures is the 1- to 3- tended for locations of inelastic deformations [15]. As consequence, the
storey gravity-resisting frames assembled by monolithic columns fixed precast members should not be detailed for ductility and should remain
at the base and free at the top, with pinned beams resting on corbels, elastic during seismic action. A major advantage of a dry-jointed con-
sometimes strengthened with a shear panel. nection system is the reduction of scheduling conflicts between con-
In the mentioned countries, both academia and research have pro- struction phases related to in-place concrete casting and strengthening.
vided viable alternatives to this typical precast layout. Proposed solu- This is just a finishing phase and is not required to get structural
tions range from an emulative approach, namely, moment-resisting strength and stability during the assembly process.
frames assembled with trough monolithic precast elements joined to- Conversely, in Italy and more generally in Europe, the typical pre-
gether with cast-in-place concrete, to a dry-connection approach, cast skeleton structure has long been adopted, and this same solution is
namely, frame connection based on a rocking effect, joined together still widely adopted at present [16]. Considerable research effort in
through tendons or damping devices [1–7]. Europe has been devoted in the last decade to investigation of the
Cast-in-place emulative technology is mainly developed in New seismic performance of those types of industrial buildings, as tested by
⁎
Corresponding author.
E-mail addresses: yufanhuang@fzu.edu.cn (Y. Huang), enrico.mazzarolo@gmail.com (E. Mazzarolo), bruno@fzu.edu.cn (B. Briseghella), zordan@bolinaingegneria.com (T. Zordan),
a.chen@tongji.edu.cn (A. Chen).
https://doi.org/10.1016/j.engstruct.2018.01.012
Received 14 February 2017; Received in revised form 8 November 2017; Accepted 5 January 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
Y. Huang et al. Engineering Structures 159 (2018) 227–244
the project PRECAST STRUCTURES EC8, which concluded in early released specific instructions for the design of these components, which
2007 after 4 years of activity. The project SAFECAST, aiming to design are fully comparable to steel-concrete composite elements, removing
dry connections between members and study their contributions to the any limitations for their employment in seismic regions [22]. Among
structural global behaviour [17–19]. the benefits provided by CSTC beam technology, some are here re-
A key issue of research on precast concrete still remains, that is, the called: high bearing capacity, high mounting speed thanks to un-
possibility of applying the typical benefits of the pre-fabrication in areas propped erection, and limited costs owing to a high prefabrication
characterized by medium to high seismic intensity. Not only to low-rise level. Furthermore, as a concrete base section is adopted for CSTC
industrial or commercial structures but also to multi-storey frames for beams, good fire strength is also provided.
public and strategic buildings, such as schools, hospitals and many On-site column assembly of CSCT beams is straightforward: the
others, as well as to high-rise residential premises [20,21]. asymmetrical layout of vertical steel plates belonging to SCJs enables
The research activity presented in this paper aims to provide a po- convenient beam accommodation (Fig. 4), drastically reducing con-
sitive contribution to this topic through the proposal of an innovative struction tolerance issues and increasing construction speed. During the
hybrid high-performance precast beam-to-column joint for multi-storey temporary phase, CSCT beams rest in a simple support scheme without
framed structures, characterized by a modular assembly process that is the need for temporary scaffolding (Fig. 5a). To restore beams’ long-
less dependent on the construction phases. itudinal continuity through the joint, specifically designed lateral Steel
To verify the effectiveness of the proposed precast technology, the Lattice Girders (SLGs, highlighted in red1 in Fig. 5b) are finally pro-
conducted research activity consists of two complementary approaches. vided. Finally after placement of the floor slabs (Fig. 5c), the concrete is
First is an experimental phase that aims to improve knowledge of the casted to make the assembly monolithic (Fig. 5d).
behaviour and resisting mechanisms of the joint, owing to an in- The proposed precast system can be classified as a hybrid solution
novative layout and a lack of analogous reference experiences from the since it adopts both wet and dry connection approaches: cast-in-place
literature. Afterwards, a numerical and analytical study phase, to concrete to provide structural continuity to the horizontal beams and
evaluate the components’ strength domains, optimizes structural com- dry bolted-joints for the columns. This strategy guarantees the overall
ponents and defines the design rules. ease of construction with reduced tolerance problems and self-bearing
This paper focuses on the static performance of the system. Both the capacity during construction phases, with a consequent reduction of
experimental and numerical results reported address evaluating the scheduling conflicts between construction phases and favourable con-
static axial capability of High Strength Concrete (HSC) columns. sequent economic impacts on construction time and costs. Fig. 6 shows
one parking garage where this type of prefabricated high-strength
2. Description of the system composite column is used. The authors have also investigated the cyclic
behaviour of CSTC beams connected by HSC columns through experi-
The major advantage of dry-jointed connection systems over mental and numerical simulation methods [23]. The results showed
monolithic ones is the reduction of scheduling conflicts between con- that, despite the absence of lateral reinforcement, the beam-to-column
struction phases related to in-place concrete casting and strengthening. joint specimens showed an acceptable ductility.
This is just a finishing phase and is not required to obtain structural The proposed joint layout appears extremely tidy, avoiding re-
strength and stability during the assembly process. Hence, the challenge inforcement congestion typical of reinforced concrete frames or other
faced during the design phase of the new precast system consisted of precast solutions; the modularity of the precast components allows a
making the assembly process less dependent on the construction phases low-storey frame to be assembled with the same ease as a taller one.
than typical emulative precast technologies, taking inspiration from the The total self-bearing capacity and mounting ease of the proposed
dry-jointed system approach, without surrendering ease of assembly solution make rapid all-weather erection possible, even by unskilled
and structural performance. labours, contributing to the final cost reductions. The quality of the
Fig. 1 shows the manufacturing process of high strength composite adopted materials and smartness of the final joint layout make this
columns. Concrete columns are made by centrifuged HSC class C75/90, solution suited not only to multi-storey industrial or commercial
casted through a process that leads to the typical ovoid section. HSC structures and multi-storey parking garages but also to multi-storey
allows for high bearing capacity with limited overall section dimensions frames for municipal facilities and strategic building, such as schools,
(330 by 550 mm) and consequent maximization of the commercial or hospitals and many others, and even high-rise residential structures in
saleable space. The precast process of these members is highly stan- areas of medium to high seismic intensity. Despite the high quality of
dardized and automated, thus containing manufacturing costs, though the base materials, the final cost is comparable to that of equivalent RC
the base materials are of high quality. As a first step, the single-storey structures thanks to the high automation level during the manu-
column-skeleton is assembled by welding two end-flanges (steel grade facturing precast process and the on-site mounting ease.
S355) at both ends on 8 longitudinal rebars (steel grade B450)
(Fig. 1a–d) and coiling up ϕ6 stirrups to complete the reinforcement
cage (Fig. 1e). 3. Experimental tests
Short column elements are easier to transport, handle and raise. The
best compromise is found in the adoption of single- or double-storey 3.1. Test specimen design
columns (Fig. 2), with voids at the inter-storey level. Such portions of
the column are referred to as Steel Core Joints (SCJs) and allow beams The innovative layout of the presented precast system and the lack
to rest in the temporary phase. Bolted connections at the two ends of of similar reference experiences in the literature required experimental
the column provide structural continuity for the vertical columns and tests to appraise the resisting mechanisms of the proposed HSC system.
complete the layout of the presented precast columns. Moreover, all It is very important to investigate the response of the joint in all the
horizontal flanges present a central hole (120 mm in diameter) designed phases and particularly understand the contribution of each structural
for the HSC centrifuged casting process, which is later performed on element to the joint response. A detailed finite element model simula-
formwork specifically designed for this purpose (Fig. 3). tion performed at the beginning of the research confirmed that it was
The horizontal structural element of the proposed joint is con- necessary to understand better the flow of vertical forces between the
stituted by a Steel Truss Concrete (CSCT) beam with a concrete base. concrete column and the steel flanges of the joint, and that a great
Such a precast solution is introduced in Italy in the 1970s, while it is
currently widely adopted and is spreading rapidly even outside Italian 1
For interpretation of color in Figs. 5 and 21, the reader is referred to the web version
borders. Moreover, the Italian Superior Work Council has recently of this article.
228
Y. Huang et al. Engineering Structures 159 (2018) 227–244
Fig. 8 shows the test setup, which is located at the State Key La-
boratory for Disaster Reduction in Civil Engineering in Tongji
229
Y. Huang et al. Engineering Structures 159 (2018) 227–244
University, China. Each specimen is positioned at the centre of a rigid plates of the SCJs become the weak component of the assembly. The
frame, and axial load on the top column is applied by an electro-hy- structural responses from different specimens are rather similar despite
draulic jack with a maximum capacity equal to 20,000 kN. the different geometries of the steel-core joints and reinforcement.
Several unloadings and reloadings between 3000 and 5000 kN are The yielding force is estimated as the force level corresponding to a
used to settle the composite column and reduce any possible gap at the residual stiffness equal to 15% of the initial one. Experimental values
connection interface. The load is finally increased until failure. The load range between 5294 and 5933 kN, while the corresponding yielding
speed application is limited to 5 kN/s. displacements range between 1.28 and 1.77 mm. The average yielding
Several sensors are positioned on specimens to capture strains force equals 5620 kN, which agrees well with the analytical estimation
during testing. Eight and 6 strain gauges are placed on the column of the axial strength provided by the vertical plates obtained through
surface 200 mm from the central joint to record concrete strains in the Eq. (1).
axial and transversal direction, respectively (Fig. 9a). Twelve strain-
Fy = Asteel ·f y = 210 × 40 × 2 × 345 = 5796 kN (1)
gauges are stuck on the steel-core joint, 8 of them to monitor axial
strains on the vertical steel plate, and the remaining on the lower steel
flange to monitor in-plane strain (Fig. 9b). Finally, to measure the axial
deformation of the steel-core joint, there are two Linear Variable Dif- 3.4. Strain gauges on concrete
ferential Transformers (LVDTs) on both column faces (Fig. 10): the first
between the top horizontal steel flange and top concrete column and The responses retrieved by these sensors are similar among different
the latter between the two concrete columns including the steel joint. samples. For the sake of brevity, only partial results are reported.
Typical records for horizontal strains for specimens A1 and B1 are
3.3. Force-displacement curves reported in Fig. 14. They experience deformations for both the upper
and lower columns, always less than 300 µm, suggesting that the con-
Load vs. steel-core-joint deformation relationships are reported in crete response in the transversal column direction is perfectly elastic
Fig. 11 The results obtained from different specimens are similar de- during the entire test. This is visually confirmed by the absence of any
spite the different geometries. In all cases, splitting failure of grouted vertical cracking in the column.
concrete at the core-joint level occurs when the active load reaches Typical records for vertical strains for specimens A1 and B1 are
5000 kN (Fig. 12). Good deformation capabilities are observed before reported in Fig. 15. Due to the asymmetrical layout of the vertical steel
failure, with the latter achieved by buckling of vertical steel plates plate in SCJs, a non-uniform axial strain field is present over their
(Fig. 13). section. In particularly, sensors V10 and V22, placed nearer to the
Because the inter-storey-core concrete is split out, the vertical steel vertical steel plates, experience the highest compressive deformation,
230
Y. Huang et al. Engineering Structures 159 (2018) 227–244
HSC column
CSCT beam
SCJ
a)
b)
c) d)
Fig. 5. Beam-to-column joint assembly operations.
ranging from 1000 to 1650 µm. Concrete strains tend to diminish for that HSC columns perform elastically during static testing and are thus
adjoining sensors belonging to the same alignment. Axial strains at the far from having their maximum bearing capacity exploited.
centre of the column face range between 500 and 750 µm. The re-
maining gauges located in the arch-shaped side of the column record 3.5. Vertical strain on the vertical steel plate
still lower deformation, never exceeding 500 µm. Global force vs. strain
trends appear to be nearly linearly elastic for all sensors until failure. Considerable plastic deformation should be expected from these
The highest recorded strain is equal to 1650 µm, well below the peak- sensors, as the vertical steel plates are the weak component of the
strength-strain Ɛc0 = 2800 µm. composite column. Considering that the average deformation of a steel-
The absence of cracking also in the horizontal direction confirms core joint is nearly 10 mm and that the steel plates are 380 mm long, the
231
Y. Huang et al. Engineering Structures 159 (2018) 227–244
Table 1 justified by two motivations. On the one hand, the deformation results
Geometrical properties of composite-column specimens. in some sensors being severely damaged after concrete splitting failure.
For specimens A3 and B1, only two sensors worked continuously, while
Specimen A [mm] b [mm] c [mm] Rebar [mm]
data provided by the others are clearly corrupted and thus disregarded.
A1 65 30 35 30 On the other hand, the buckling failure achieved by specimens might
A2 45 25 30 30 have favoured a premature separation of several sensors from steel
A3 30 20 25 30
surfaces. Specimen A1, for example, failed after reaching an active force
B1 50 25 30 20
B2 35 20 25 20
level equal to 6950 kN, while the gauges stopped recording at 6000 kN.
B3 25 15 20 20 A similar behaviour is shown even by sensors applied to B1.
232
Y. Huang et al. Engineering Structures 159 (2018) 227–244
recorded in the HSC column, well below the peak limit. In the joint specimens. Fig. 17 exhibits the numerical model in the ABAQUS soft-
zone, adequate confinements should be provided to obtain the max- ware. The numerical model consists of a full tri-dimensional re-
imum structural performance of the structural system. This aspect is presentation of the composite column geometry. There are a total of
further investigated in the next paragraphs. 27,578 nodes and 20,080 finite elements with 19,456 solid, 192 planar
and 432 linear elements.
The 8-node hexahedral solid element (C3D8R) is adopted for the
4. Numerical investigations entire model, with the exception of the spiral stirrups inside the
column, which are modelled by linear truss elements (B31). A regular
4.1. Finite-Element numerical modelling mesh is ensured by the mapped meshing subroutine. The average mesh
size for concrete is 40 mm. The circumference of the longitudinal
The possibility of fitting experimental results via numerical simu- column reinforcement is split 8 times to accurately reproduce the cir-
lation is of great interest since it allows for the obtainment of reference cular section geometry. The steel joint is meshed with a finer mesh with
finite element (FE) models, suitable for a wider study of the proposed an average edge size of approximately 20 mm. Vertical steel plates are
technology, as a substitute for experimental real scale testing. An ex- split 4 times in the thickness direction to more accurately capture
perimental survey of the static performance of the composite column second-order effects. The concrete cast to fill the core joint is not
reported is adopted for this purpose, as the column’s geometry is rather modelled owing to the impossibility of simulating sudden concrete
simple and collected data provide a comprehensive picture of specimen splitting failure via the adopted concrete plasticity model. All transla-
response during testing. Numerical analyses are conducted using tional degrees of freedom (DOFs) are restrained at the bottom plane
ABAQUS V.6.10 FE software [24], which could be implemented as an level. The imposed force is applied at a specific reference point located
alternative to costly and time consuming experimental tests, and allow in the centre of mass of the upper rigid plane. A monotonic increasing
an extensive parametric investigation of composite joints and possible axial load is applied.
design optimizations [25]. As expected, during the tests the concrete in
the joint was immediately subject to brittle failure and its contribution
to the resistance was negligible. This aspect confirmed the importance 4.2. Material mechanical proprieties
of confinement in the SCJ. It is very important to investigate the steel
flanges and the distribution of the stresses in the columns, so it was For concrete class C25/30, both compressive strength (Rc) and
decided not to consider the concrete in the joint in the numerical elastic modulus (Ec) are measured in the laboratory, by testing concrete
analysis performed. cube with edge dimension 150 mm. Elastic modulus is evaluated mea-
Six different models are defined, as many as there are tested suring secant stiffness after 5 time loading and unloading procedure at
233
Y. Huang et al. Engineering Structures 159 (2018) 227–244
8 8 8
force (kN) x10 3
4 4 4
2 2 2
A1 A2 A3
0 0 0
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
displacement (mm) displacement (mm) displacement (mm)
8 8 8
force (kN) x10 3
4 4 4
2 2 2
B1 B2 B3
0 0 0
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
displacement (mm) displacement (mm) displacement (mm)
υ = 0.2 (7)
Fig. 12. Inter-storey cast-in-place concrete failure.
where fcm is average compressive strength, Ecm is average elastic mod-
50% of maximum strength. Cylindrical strength (fc) is estimated as ulus, fctm is average concrete tensile strength, εc,0 is unconfined concrete
0.83Rc. Totally, three specimens are tested. According to GB/T 50,081- strain at peak stress fcm, εc,50 is unconfined concrete strain at 50% fcm
2002 [26], curing of specimen is made at the controlled temperature of residual strength, and ν is Poisson’s coefficient for concrete in elastic
20 °C ± 3 °C and relative humidity higher than 90% for 28 days. phase.
Loading speed during testing is limited to 0.3 MPa/s. For precast con- For fracture energy value, reference is made to CEB-FIB Model Code
crete classes C30/37 and C75/90, characteristic compressive strength 2010 [28] formulation, see Eq. (8):
fck is assumed as a reference value (30 and 75 MPa, respectively). Gf = 73(fcm )0.18 (8)
Missing parameters are estimated adopting EN 1992-1-1 [27] provi-
sions, see Eqs. (2)–(7): Table 2 summarizes the main mechanical parameters for adopted
concrete classes. Complete stress-strain relationship for concrete is de-
fcm = fck + 8 MPa (2) fined adopting the Légeron and Paultre model [29], which defines a
compressive concrete model suitable for both Normal Strength Concrete
f 0.3
Ecm = 22,000 ⎛ cm ⎞ ⎜ ⎟ (NSC) and HSC.
⎝ 10 ⎠ (3) Different components are adopted in the proposed beam-to-column
234
Y. Huang et al. Engineering Structures 159 (2018) 227–244
joint, each characterized by specific material grades. Components are elastically until the end of testing.
listed below, together with their corresponding steel grade and chosen
reference code: (1) lattice girders, steel-core-joint and CSCT beam truss, 5. Strength domain estimation
steel S355J0, according to EN 10025 [30]; (2) column reinforcements
and stirrups; steel B450C, according to EN 1993-1-1 [31]; (3) bolts; It is stated as during the static experimental tests, the core-concrete
steel grade 10.9, according to ISO 898-1:1999 [32]. Material properties was subjected to splitting failure, when approaching an axial load level
of steel are listed in Table 3. almost equal to 5000 kN, due to absence of lateral confinement. Such a
Estimation of yield plateau length in the σ-ε curve and the corre- failure could be prevented by confining action provided by flooring
sponding strain limit εL is shown in Fig. 18, which estimated through a system. In addition, specific reinforcement could be adopted, especially
formulation reported in SINTAP BS/23 [33], and the Briseghella model for lateral joints. This aspect confirmed the importance of confinement
[34] is assumed to define the complete stress-strain relationship: in the SCJ. The experiment tests considering only axial load and a poor
σ (ε ) = E ·ε ε ⩽ εy confinement of the concrete in the core, showed the basic behaviour of
σ (ε ) = f y εy < ε ⩽ εL the joint and its components. In a following step of the research the
P
contribution of the confinement was numerically investigated, con-
εu − ε
σ (ε ) = fu −(f y −fu ) εL < ε ⩽ εu sidering also different load conditions that could introduce bending
εu − εy (9) moments. In this way it was possible to understand better the response
where P is calculated by Eq.10, of the joint and investigate the possible applications to real cases. In
order to investigate the maximum performances available from com-
εu−εL
P = Epl posite-column system, it is assumed that adequate confinement is pro-
σu−σy (10) vided to core-joint. The reference resistance domain of HSC column and
composite core-joint taken as separate are defined. The developed
4.3. Numerical analysis model calibration analyses are performed imposing increasing rotation at the top of the FE
model (corresponding to a concentrate bending action) at different
The first aspect to take into consideration when checking the cap- axial load levels, to get the complete M-N strength domain.
ability of an FE model to reproduce an experimental response is the
global quantities that summarize the structural behaviour of the tested 5.1. HSC columns
specimens. The total applied force vs. axial deformation of the steel-
core joint is particularly suited for this purpose (Fig. 19). The HSC column FE model is derived from reference models
Beyond the excellent correspondence between the experimental and adopted for static experimental test simulation. The model represents
numerical trends suggested by these values, attention should be paid to an HSC column stub with a length of 0.79 m, (Fig. 21). Longitudinal
the amplified (by 5 times) deformed shape of the numerical specimen at rebars (in blue) and concrete are explicitly solidly modelled through
failure, as shown in Fig. 20. With only an exception for model A1, all hexahedral elements (C3D8R), while confining spiral stirrups (in red)
the remaining models clearly evidence a failure condition associated are modelled with embedded linear truss elements (B31). A perfect
with vertical steel plate buckling, analogous to the experimental re- bond between the concrete and reinforcement is supposed. At the top
sponse. This confirms the robustness of the adopted numerical ap- and bottom of the model, two rigid planes are meshed with quad-
proach and that it is able to account properly for the second-order ef- rilateral elements (R3D4). These allow for redistribution of the
fect. Concrete columns attained limited damage levels, corresponding boundary conditions over the entire column section, avoiding stress
to experimental evidence, thus confirming that HSC columns performed concentration at the column’s ends. The so-defined model consists of
235
Y. Huang et al. Engineering Structures 159 (2018) 227–244
10,375 nodes and 7440 solid, 192 planar and 216 linear elements, for a Table 2
total of 7848 finite elements. Material properties of concretes.
The base is assumed as fixed, while two load steps are considered at
Grade fcm (Mpa) fctm (Mpa) Ecm (Mpa) εc,0 (‰) εc,50 (‰) Gf (N/ ν
the top column’s end. During the first step, an axial force is applied. mm)
Several numerical simulations are performed with an axial load level
ranging between 0 and Nmax with a step of 2000 kN. The second step C25/30 28.7 2.8 32,200 2 3.5 133 0.2
consists of an imposed increasing rotation. Because this step is dis- C30/37 38 3.3 32,800 2 3.5 140 0.2
C75/90 83 5.7 41,500 2.8 2.8 161 0.2
placement controlled, the softening branch of the global structural re-
sponse can be monitored. Fig. 22 shows the numerical response in terms
of M-χ (reaction moment vs. curvature) curves at different axial load of axial load level. It should be noted that the estimated maximum axial
levels. Considering that the model section property is constant, the strength capacity for the HSC column is approximately 15,000 kN, more
curvature can be estimated by simply dividing the imposed rotation than double the maximum axial force level attained during experi-
value by the specimen’s height, as in Eq. (11). mental testing.
χ = φ/h (11)
where φ is the top rotation and h is the height of the specimen. 5.2. Core joints
Nmax is evaluated by increasing the axial force level until con-
vergence is no longer attained. This procedure allows an accurate es- The core joint is also modelled separately to obtain its response in
timation of the M-N (bending moment vs. axial force) strength domain terms of strength and stiffness as a function of the acting axial load
for the HSC column to be obtained (Fig. 23). Such domain results are level. The FE model is reported in Fig. 24. It consists of two vertical
fundamental for all subsequent numerical analyses since they represent steel plates made of structural steel S355, with section dimensions of
the reference limit strength domain for the composite-column as- 210 by 40 mm, embedded in normal strength concrete (C25/30).
sembly. In Fig. 23, the section bending stiffness is plotted as a function Concrete and vertical plates are both modelled through hexahedral
Fig. 17. FE model: (a) full model (b) HSC column mesh (c) horizontal steel plate mesh (d) steel-core-joint mesh.
236
Y. Huang et al. Engineering Structures 159 (2018) 227–244
Table 3
Material properties of steels.
Grade tk (Mpa) fy (Mpa) ft (Mpa) εy (%) εL (%) εu (%) E (Mpa) Epl (Mpa) ν
237
Y. Huang et al. Engineering Structures 159 (2018) 227–244
fully develop. In models 20/30 and 20/40, the weak element becomes 6. Analytical approach for design
the lower flange. Plastic hinges on this element develop in proximity of
the vertical steel plate as a result of excessive bending provided by bolts 6.1. Provisions for HSC column bending strength
in tension, while the upper flange still performs elastically. For all these
models, the behaviour of connection could be associated with Mode 2 In addition to numerical modelling, an important aspect is the
failure, as described in EN 1993-1-8 [35], characterized by both the identification of simplified analytical approaches to model composite-
flanges and bolts yielding. In the remaining models, the flanges are stiff column behaviour, particularly for the design phase. A first issue to
and strong enough to provide elastic performance until failure. Such consider is the possibility of obtaining the strength domain of the
behaviour recalls the rigid mechanism (Mode 3 failure) reported in EN composite column’s components taken separately. Related to this aspect
1993-1-8 [35], characterized by the bolts yielding. is correctly accounting for concrete strength, particularly when dealing
In Fig. 28, from the strength point of view, increasing the flange with HSC. A reference formulation to evaluate the expected column
thicknesses leads to a general improvement of composite-column per- compressive strength under pure axial compression is reported in ACI
formances. Models 30/30, 30/40 and 40/40 fit the reference domain 318 [36]:
reasonably well, with the exception of very low axial load levels
P0 = αfc (Ag −As ) + f y As α = 0.85 (13)
(≈0 kN). This implies that a bolted connection could represent a weak
component of the assembly when limited axial loading is provided. where P0 = the pure axial load capacity of columns; fc = the concrete
Additionally, the results of equivalent stiffness of composite column are compressive strength; fy = the steel yielding stress; Ag = the gross
sensibly affected by low axial load levels, independently from the cross-sectional area; and As = the area of longitudinal steel.
connection layout. According to these observations, it is possible to A correction coefficient for concrete strength α = 0.85 should be
identify model 30/30 as the optimum specimen among the considered adopted to design independently from the concrete grade. It accounts
ones, as further increasing the flange thickness does not provide any for the difference between the concrete strength provided by cylindrical
appreciable performance improvement in terms of strength or in terms concrete specimens and that provided by real-scale column samples.
of stiffness. The correct design of high strength reinforced concrete sections under
Fig. 21. Column FE model: (a) full model (b) embedded reinforcement (c) mesh section.
238
Y. Huang et al. Engineering Structures 159 (2018) 227–244
Fig. 22. Moment vs. curvature: FE solution under different model and axial load conditions.
Fig. 23. M-N strength domain for the HSC column: FE solution. Fig. 25. M-N domain comparison.
combined axial and bending solicitations requires the same additional an analytical approach, a specific computer program is developed in
considerations, particularly to account for the different behaviours of MATLAB [38]. The implemented subroutine uses a layered re-
HSC compared to NSC samples. presentation of the section, each layer being separated into a confined
In ACI 318 [36], the reduction factor α used for compressed core layer and an unconfined cover layer with the corresponding ma-
members is substituted by the equivalent rectangular stress-block cor- terial properties and characterized by specific average dimensions b(y)
rection factor η = 0.85. For the depth of the rectangular compression and distance y from the neutral axis (Fig. 30). The confined concrete
block λ (Fig. 29), the following equation is assumed in Eq. (14). material properties account for the effectiveness of the confinement
reinforcement and the yield strength of the transverse steel reinforce-
λ = 0.85−0.05 × (fc −40)/6.9 ⩽ 0.85 (14)
ment by the Légeron and Paultre model [29]. The longitudinal re-
Li et al. [37] indicated that the ratios of the experimental flexural inforcement is also represented by layers, with the appropriate material
strength to that calculated according to ACI formulations are less than 1 properties. The program calculates the moment-curvature response by
when considering HSC columns subjected to high axial load levels. A an incremental iterative procedure based on the plane section hy-
further coefficient η should be considered for the design of HSC mem- pothesis and fulfilling of axial and bending solicitations equilibrium
bers subjected to axial-bending solicitations in EN 1992-1-1 [27], which conditions.
reports specific formulations for concrete strength lower or higher than N
x
50 MPa (Eq. (15)). N= ∫0 σc (y )·b (y )·dy + ∑ σs,i·As,i
i=1 (16)
η = 1 fck ⩽ 50 MPa
η = 1−(fck −50)/200 fck > 50 MPa N
(15) x
M=− ∫0 σc (y )·b (y )·y·dy− ∑ Zi·yi
To define the strength domain of general layout RC sections through i=1 (17)
Fig. 24. Core-joint FE model: (a) full FE model (b) embedded steel plates (c) section mesh.
239
Y. Huang et al. Engineering Structures 159 (2018) 227–244
Fig. 26. Double flange-composite-column FE model with a 4-bolt layout: (a) full model (b) inside steel plates and rebars (c) top view of bolt modelling (d) upward view of bolt modelling.
First, a certain axial load level N is fixed. Second, the εc value is either the ultimate confined concrete strain εccu, ultimate steel strain εsu
assumed at the top edge of the concrete section. As a first tentative or 80% of the residual bending strength [39]. Critical points, such as
assumption, the neutral axis is located at a middle height. The neutral yielding of longitudinal reinforcement, onset of cover spalling, and
axis is then moved iteratively upwards and downwards through a bi- ultimate curvature, are also stored. Repeating the process for different
section method until axial load equilibrium is achieved (Eq. (16)). The axial load levels, the complete M-N strength domain is defined.
resulting bending moment is then calculated (Eq. (17)). Afterwards, a The HSC column strength domains, obtained from numerical
new increased value for εc is assumed, and the iterative process starts (Fig. 23) and analytical approaches, are plotted together and compared
again. In this manner, the complete M-χ curve is analytically defined in Fig. 31. The coefficient α is assumed to be unity, as this reduction
for a certain axial load level. Failure corresponds to the attainment of factor is disregarded during numerical analyses. To reproduce the
Fig. 27. Plastic strain at the column-to-joint interface: 4-bolt flanged connection; deformation scale factor of 10×.
240
Y. Huang et al. Engineering Structures 159 (2018) 227–244
(a) Comparison of the strength ratio (b) Comparison of the stiffness ratio
Fig. 28. Global performance comparison of the composite column with a double-flange 4-bolt connection.
Fig. 31. Numerical vs. analytical estimation of the strength domain for the HSC column.
The same approach is applied to obtain the strength domain for the
design of the composite column, adopting a safety factor of the base
material required by codes. The concrete strength reduction coefficient
Fig. 30. RC section discretization for the MATLAB subroutine. α and η are also considered. The first one accounts for differences be-
tween cylindrical concrete strength and real-size specimen strength and
strength domain when both axial and bending conditions are con- is assumed equal to 0.85. HSC base material properties and corre-
sidered, two values for η are defined: 0.85 and 0.95. An analytical sponding partial safety factors are reported in Table 4, taking as re-
formulation with η = 0.95 better agrees with the numerical results for ference provisions reported in Ref. [27] and Italian Code NTC 2008
axial loading lower than 8000 kN. η = 0.85 leads to better fitting of the [40].
resistance domain for higher axial load levels. For axial loading lower The maximum axial bearing capacity of the HSC column is com-
than 4000 kN, the strength domain is well captured by both coeffi- puted from Eq. (18), in which coefficient η is obtained from Eq. (15).
cients, as the role played by concrete is secondary.
The analytical results confirm the importance of introducing a
strength reduction for HSC. The fact that numerical simulation captures Table 4
evidence similar to that of the experimental test reported in the lit- Material properties and partial safety factors of the HSC base.
erature concerning the lower performance of HSC columns compared to Concrete Steel rebars
NSC ones is an additional confirmation of the goodness of the adopted
numerical approach. For a safe analytical estimation of the HSC column Grade fck (Mpa) γc Grade Fyk (Mpa) γs
strength domain, η = 0.85 should be assumed, which agrees with Eq.
C75/90 75 1.5 B450C 450 1.15
(15). In proximity of a pure axial loading condition, the numerical
241
Y. Huang et al. Engineering Structures 159 (2018) 227–244
Table 5 Table 6
Material properties and partial safety factors of the composite-core-joint base. Static load amplification factor.
Grade fck (MPa) γc Grade Fyk (MPa) γs Structural dead load (G1) – 1.3
Non-structural dead load (G2) – 1.5
C25/30 25 1.5 S355 345 1.05 Live load (crowd) (Q1) – 1.5
Wind load (Q2) 0.6 1.5
f fyk
Pmax = αη γck (Ac −As ) + As
c γs L = γG1 G1 + γG2 G2 + γQ1 Q1 + ∑ ψ0j Qj
(21)
= 0.85 × 0.875 ×
75
1.5 ( 158,000−8 ×
π × 302
4 ) +8×
450
1.15
×
π × 302
4
j> 1
= 7878 kN (18) Ref. [27] makes provisions (Eq. (22)) to estimate single columns’
eccentricity to account for second-order effects due to possible frame
In a similar manner, the reference strength domain for the compo- elevation irregularities:
site core joint is estimated. Material properties and corresponding θi = θ0 αh αm
safety coefficients are reported in Table 5. 2
αh = 2/ ℓ ; 3
⩽ αh ⩽ 1
The confining pressure acting on the core joint is assumed equal to
2.5 MPa (half of the value applied for numerical simulations). αm = 0.5 × (1−1/ m) (22)
According to Ref. [40], the confined concrete strength is equal to:
where θ0 = the reference base eccentricity, taken as equal to 1/200;
fcck = fck × (1125 + 2.5 × 2.5/ fck ) = 34 MPa αh = the reduction factor to account for building height; αm = the re-
(19)
duction factor to account for the number of vertical elements; ℓ = the
The maximum axial bearing capacity of the composite core joint is total frame’s height; and m = the number of vertical elements con-
computed in Eq. (20), which is based on Eq. (15) for the considered sidered.
concrete grade (C25/30), and the coefficient η is taken as unity. Considering an inter-storey span equal to 3.2 m, the reference ec-
centricity is estimated as 7.5 mm, rounded up to 10 mm. A regular plain
fcck fy frame numerical model is built in SAP2000 [41]. Each storey in the
Pmax = αη (Ac −As ) + As
γc γs model is horizontally shifted with respect to the lower one to provide
34 345 elevation irregularities, and a nonlinear geometry analysis is performed
= 0.85 × 1 × 1.5
(158,000−210 × 40 × 2) + 1.05
× 210 × 40 × 2
(20) to account for second-order effects. The maximum allowable number of
= 8240 kN
storeys is equal to 10 and 8 for frames with bays spanning 6 and 8 m
The complete strength design domain is plotted in Fig. 32, together long, respectively.
with the reference strength domain provided by the numerical core- The force distributions in terms of axial forces and bending mo-
joint solution. The HSC column domain is completely enveloped by the ments computed for these frames are reported in Figs. 33 and 34, re-
core-joint domain, confirming the possibility of classifying it as a strong spectively. The worst solicitations’ combinations for the two frame
component of the assembly. typologies are plotted together with the reference strength domain in
The analytical composite column’s design strength domain is Fig. 35. It should be noted that the strength domain limit is nearly at-
adopted as a reference point to provide an estimation of the maximum tained. The identified maximum number of storeys should be con-
number of storeys for frames adopting the proposed precast system. sidered as a conservative lower limit, as taller frames are feasible
Loads considered for this analysis are reported by Ref. [40]. Wind ef- considering lower live loads of those assumed for the current analyses.
fects are also considered to account for additional bending moment The results illustrate that the proposed analytical approach can be used
acting on the base column at ground level. A reference wind speed in design practice. The analytical approach represents a viable alter-
equal to 27 m/s is assumed (an average value over the Italian territory). native to time-consuming numerical analyses and allows for effective
The equivalent wind load is estimated as 0.5 kN/m2, increased by a strength domain estimation for design practice.
specific wind profile factor ce(z) as specified by Ref. [40] prescriptions.
The verification process makes reference to the Ultimate Limit State 7. Conclusions
(ULS) conditions in Eq. (21). The load’s amplification factors γ sug-
gested by Ref. [40] are reported in Table 6. A compressive investigation has been carried out to analyse the
static performance of innovative prefabricated high strength composite
columns by means of experimental and numerical methods. The fol-
lowing conclusions are summarized based on the investigation results
from this paper.
(1). The axial load bearing capacity of the composite column and the
influence of the steel-core-joint geometry on static performances
were investigated through 6 different specimens. For all tests,
failure is achieved as a consequence of core-joint buckling after
concrete splitting, thus providing a reference limit strength to be
used for the design of a temporary phase, equal to almost 6000 kN.
As a consequence, particularly in the design phase, this limit
strength should be taken for exterior joints, where the confinement
action provided by the flooring system is limited.
(2). The structural responses from different samples are rather similar,
despite the different geometries of steel-core joints and the re-
Fig. 32. Design strength domain for the composite column.
inforcement. This implies that a reduced steel flange thickness
242
Y. Huang et al. Engineering Structures 159 (2018) 227–244
243
Y. Huang et al. Engineering Structures 159 (2018) 227–244
244