Engineering Structures 132 (2017) 808–821

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Behavior of two new moment resisting precast beam to column

connections subjected to lateral loading
Saeed Bahrami a, Morteza Madhkhan a, Fatemeh Shirmohammadi b,⇑, Nima Nazemi c
a
Department of Civil Engineering, Isfahan University of Technology, Isfahan, Iran
b
Walter P Moore and Associates Inc., Kansas City, MO, USA
c
Department of Civil Engineering, University of Memphis, Memphis, TN, USA

a r t i c l e i n f o a b s t r a c t

Article history: In this study, two new moment resisting connections of beams to precast concrete columns under lateral
Received 10 April 2016 load were analyzed via application of nonlinear finite element analysis software ABAQUS 6.10. The pre-
Revised 24 November 2016 cast connections considered were the beam-column connections in which precast beams is connected to
Accepted 26 November 2016
continuous precast column with corbel using (i) inverted E (bolted connection) and (ii) box section
(welded connection). Connection responses associated with lateral resistance, lateral stiffness, ductility,
and energy dissipation were compared to a reference monolithic connection. Achieved lateral resistance,
Keywords:
lateral stiffness and ductility of the proposed connections was approximately 98%, 80% and 80% of the
Finite elements
Precast concrete
equivalent monolithic connection, respectively. The effect of axial load on column and compressive
Energy dissipation strength of concrete on behavior of the connections were studied. The analytical results show that the
Lateral stiffness performance of two proposed precast connections were close to the performance of corresponding mono-
Ductility lithic connection.
Lateral resistance Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction sustainability and economical performance [2]. In 1989, Dolan and

Advantages of precast concretes have increased worldwide
usage of precast concrete structures. Beam to column connections
in frame systems affect significantly the constructability, stability,
strength and flexibility of structure. Furthermore, connections play
an important role in the dissipation of energy and redistribution of
loads as the structure is loaded. Also, the creation of totally rigid
connections of beams to columns in precast structures is very dif-
ficult and time-consuming, there by negating advantages gained
using precast features Studies have shown that if a semi rigid con-
nection stiffness is more than 80% of an equivalent monolithic con-
nection stiffness, seismic behavior of the system will not
significant change [1]. Therefore, properly designed connections
could develop expected mechanical features and demonstrate a
low-cost fast-construction system for multi-storey buildings.
In 1987, the research and development department of the Precast
Concrete Institution (PCI) conducted a laboratory investigation of 16
samples of precast connections (eight simple and eight moment
resisting connections). The aim of the project was to obtain connec-
tions behavior such as resistance, ductility, energy dissipation,
⇑ Corresponding author.
E-mail addresses: s_bahrami@cv.iut.ac.ir (S. Bahrami), madhkhan@cc.iut.ac.ir
(M. Madhkhan), fateme.shirmohammadi85@gmail.com (F. Shirmohammadi).
Pessiki created laboratory models with scales of one-half and con-
ducted an analytical study of connection in order to demonstrate
that a computer model can be an appropriate and acceptable
method for analysis of precast concrete connection behavior [3]. Bull
and Park created laboratory specimens in order to evaluate seismic
behavior of one type precast moment resisting beam to column con-
nection in New Zealand [4]. This connection was made by placing a
U-shaped precast concrete beam at the joint, and the connection was
completed using in situ concrete and slabs. The connection has been
used to make frame structures with small height. They adopted for-
mulations for connection design also. French et al. had looked into
the issue of moving the connections away from the column face
[5]. In their research, the connections were relocated to the beam
span at a distance away from the column faces. The models com-
prised a precast reinforced concrete column and a precast partially
prestressed beam. With such a frame configuration, the beam-
column joint core, in which the reinforcement details are compli-
cated, can be prefabricated precisely under factory conditions. The
reinforcement continuity will further enhance the integrity of the
joint and prevent premature failure. Most importantly, the coincid-
ing condition between the inherent plastic hinge locations and the
connection regions can be avoided. Also researchers such as
Restrepo et al. and Khoo et al. proposed beam to beam connections
[6,7]. The primary variable in the tests was the connection detail

http://dx.doi.org/10.1016/j.engstruct.2016.11.060
0141-0296/Ó 2016 Elsevier Ltd. All rights reserved.
 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821
between the beam and beam. For frames adopting mid-span beam-
to-beam connections, Park noted that the precast components of
this system can be very heavy and difficult to transport due to their
relatively large dimensions. This transportation difficulty would
then hinder the choice of precasting the cruciform members for
frames with long beams. Parastesh and Khaloo conducted experi-
ments on one type precast concrete connections with a scale of 0.4
under different bar ratio and stirrup distances in the beam. The pri-
mary objective of their research was to develop moment resistant
connection of precast beam to precast column for zones with high
seismicity. In these experiments, cruciform specimens were made
of continuous columns and beams that were separated. A gap was
created in the columns to provide enough bearing area for sitting
the beams and transferring the construction loads before in-situ
concrete becomes structural. The cross-section of the beams was
U-shaped at near connection, while the rest of cross-section was
rectangular. Forces resulting from flexure occurred due to overlap-
ping of protruding bars from columns and beams buried with
in situ concrete [8,9]. The test results showed that the connection
can be successfully designed and constructed to emulate cast-in-
place construction. But, on site, the placement of the beams on col-
umn need for using formwork and temporary vertical supports for
the beams. Shariatmadar and Zamani studied three interior precast
concrete connections and one monolithic reference connection
(MO). In two proposed precast connections (PC1, PC2), column
was discontinuous at connections. After placing the beams at the
axe of the column gap and entering the add-bars in beam and col-
umn free gap, 100 mm top of precast beams, and free spaces of col-
umn and beam were filled simultaneously with cast-in-place
concrete. In the third connection (PC3), the column was continuous
and the seated of the beam was created by welding available pend-
ing plates in the beams and column [10]. Fabrication of PC3 speci-
men was easier from the other precast connection. Discontinuity
of columns at each story level slow down construction time of build-
ings of systems PC1 and PC2. Elliott et al. tested four semi rigid con-
nections. Supports for beams are provided by means of steel corbel
or solid section on each side of the columns which also transferred
shear forces. The top bars passed from the columns and provided
connection continuity. This can lead to a low-cost fast-
construction system for multi-storey buildings, where multiple sto-
ries can be constructed at once. [11,12]. Connections were made
using proprietary pinned jointed billet, cleat and welded plate con-
nectors, to form cruciform assemblages subjected to sway and grav-
ity load. They were not purposefully intended as moment resisting
frame. In some cases 200 mm deep slab and high tensile reinforce-
ment completed the connections. They reported moment resistance
and flexural stiffness gained from each of the connections. Also
equations were presented to calculate effective length factor for col-
umn in semi rigid sway frame and percentage of rigidity of connec-
tions. The Elliot et al. connections achieved 10–48%, and 8–40% of
strength and stiffness of the corresponding monolithic connection,
respectively. SAFECAST project, with the aim of validating multi-
story precast concrete experimental models, performed 3 pseudo
dynamic seismic tests on full scale three story structural models.
Hollow core section beams connected to column capitals serving
as beams provided equivalent mechanical properties to that of T or
I shape, offering economic advantage by increasing floor area. Con-
nections are embedded inside the structural elements using steel
connectors and bolts. Finally, any remaining gaps are filled with
grout to finalize the process of forming the connection. Emulative
beam to column joints were reported to be satisfactory in terms of
stiffness and inter-story drift; yet, the structural responses differ
from a equivalent rigid joint and the connection behavior is catego-
rized as semi-rigid [13,14]. Choi et al. performed experimental test
on 4 precast beam to column assemblies and a monolithic assembly,
in order to achieve structural continuity, high shear deformation
809
demand, strength, stiffness and drift requirements along with intro-
ducing ease of erection and reliability [15]. Connections were simply
formed by connecting steel elements, embedded inside the struc-
tural members that connect the beam and column using bolts; and
finally, fiber reinforced concrete in the joints slightly increase the
tensile strength and shear toughness. Assemblies are laterally
loaded with reversing load pattern increasing the drift up to 5%. In
terms of strength, connection were able to maintain over 75% of
the strength at 3.5%. In connections where the plastic hinge was
designed to be away from the beam-column interface (Outside
type), higher stiffness reduction was observed, while the connec-
tions in which the plastic hinge location was near the column face
(inside type) resembled more with the monolithic counterpart.
The outside type, compared to other specimen, dissipated more
energy and responded more ductile.
The presented manuscript provides the preliminary numerical
study for proposing two new beam-to-column connections in pre-
cast structures. The results of the presented study have been used
to construct 0.6-scale experimental specimens that are explained
in detail in Refs. [23] and [24]. As discussed in Ref. [24], the exper-
imental results are in a good agreement with numerical results
presented here. The proposed connections are designed not only
to satisfy the required mechanical properties, but also minimize
construction time and labor work by providing simple construction
methods. During the construction period, semi precast beams were
placed on corbels, partly embedded inside the continuous columns.
This will remove the need for shoring and temporary bracing to
hold he precast unit in place and during the erection of the precast
components. Furthermore, the hidden corbel addresses the archi-
tectural drawback of ordinary concrete corbels. These goals were
also among the objectives of SAFECAST project. The proposed con-
nections in this study shows better performance in comparison to
Elliot connections, while not scarifying the ease of construction.
2. Proposed two precast connections and reference monolithic
connection
Design and performance of completely rigid connections in pre-
cast concrete structures is difficult and time-consuming. For some
connections, the need for framing or large volumes of in-situ con-
crete undermines the inherent beneficial features of prefabrica-
tions. Therefore, the industry of precast concrete structures seeks
connections with easy quick installation that provide required
mechanical features.
Two moment-resisting precast connections are proposed and
their performance is assessed in the presented study. The mono-
lithic reference specimen (MC) was designed for comparison the
behavior of cast-in-place connection with precast connections.
Fig. 1 illustrates details of the interior moment-resisting connec-
tions for precast concrete frames before cast-in-place concreting. In
both proposed connection systems, the prefabricated concrete col-
umns are cast continuously in elevation with steel corbel embed-
ded in the connection core to connect beam elements. Four
vertical bars are welded to the corbel in the connection zone of
the precast columns to provide adequate shear strength and stabil-
ity during the installation and prevent the slip of corbel during lat-
eral loading. The steel corbel provide enough bearing area for sitting
the precast concrete beams and transferring the construction loads
before in-situ concrete becomes structural. Consequently, in the
proposed systems, there is no need for using formwork and tempo-
rary vertical supports for beam and slab elements. This can lead to a
low-cost fast-construction system for multi-storey buildings,
where multiple stories can be constructed at once.
In both connections, the precast beam is placed on the steel
corbel, and steel bars below the beam were connected to the steel
810 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821

Fig. 1. Schematic interior precast connection.

corbel using bolts or welding and provide continuity. Two top con-
tinuous bars of beams passed through the holes in the columns and
develop negative moment. Cast-in-place concreting provides con-
nection continuity. More details on of the two proposed connec-
tions are provided in the following sections.
2.1. First precast connection (Specimen PC-1)
In the first proposed precast connection (PC-1), the precast con-
crete beam was placed on the embedded steel corbel in the contin-
uous column, and bottom threaded bars of the beam were
tightened between the grooves of the corbel by two nuts and steel
gaskets with thicknesses of 10 mm. The empty space of the con-
nection area was filled with expandable grout. After grouting,
two top bars were passed through two holes in the column. Those
holes were also grouted, and the connection was completed by slab
concreting. The schematic of this connection is illustrated in. The
details of the first beam-to-column connection is shown in Figs. 2
and 3.

(a)Precast column (b)Precast beam

(c)Beam to column connection before and after installation using two bolts

(d) precast beam to continuous column after installation

Fig. 2. Connection schema of first precast connection (specimen PC-1).
 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821 811

Fig. 3. Beam to column connection details of specimen PC-1.

Fig. 4. Beam to column connection details of specimen PC-2.

812 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821
2.2. Second precast connection (Specimen PC-2)
In the second proposed connection (PC-2), a steel box section of
size 120
120
10 mm embedded in the column was used as a
corbel to connect the beam and joint core. Bars below the beam
transferred their force to columns through welding to a channel
cross-section and then through channel-shaped welding to corbel.
The connection details and schematic of the specimen PC-2 are
shown in Figs. 4 and 5, respectively. A steel box section of size
150
150
12 mm embedded in the column projects a distance
of 150 mm from the column face, was used as a corbel to connect
the beam and joint core. An inverted channel section with 12 mm
thickness, 100 mm depth and length of 300 mm that 180 mm
length of it embedded into beam end, anchored into beam by weld-
ing of Bottom longitudinal rebars on each flange. Length of afore-
mentioned rebars and welding length were 2350 mm and
130 mm, respectively. On site, the precast beams were seated on
the box section in each side of the column. After adjustment of
beams on corbel, the channel flanges of the beams were welded
to the webs of corbel at both site. In fact positive moment must
have been transferred between beam and column by welding
rebars to plates and steel plates together. Then top rebars of beams
with length of 4900 mm were passed through the pending stirrup
of the beams and two holes in the column. Those holes were
grouted and free spaces on top of the beams and between precast
elements were filled with cast-in-place concrete. All the reinforce-
ment and tensile strength and the concrete compressive strength
in PC-2 were equal to those of PC-1. This connection ensured the
continuity of the beams bottom reinforcement without any lap
splicing of reinforcement.
After welding the channel to the box, the empty space was filled
with expandable high-strength grout, completing the connection.
Similar to the previous connection, two bars with diameters of
25 mm were passed from the columns, then the holes were filled
with grout, and connection was completed by slab concreting.
(a) Precast column
(c) Welding of inverted channel section to corbel
Fig. 5. Schematic of second precast connection (specimen PC-2).
2.3. Monolithic reference connection (Specimen MC)
For comparing the result a cast-in-place a monolithic reference
connection (specimen MC) was designed and modeled according to
ACI 318-2008. The beam section was 400 mm by 500 mm, the col-
umn section was 400 mm by 400 mm, longitudinal top bar and
bottom bar of beams and columns exactly were similar to speci-
mens PC-1 and PC-2. The specimens were designed according to
the strong column - weak beam design concept so that the inelastic
damage of the column did not occur. The beam span was 5.0 m and
the height of the column was 3.0 m.
3. Loading and boundary conditions
The two proposed PC-1 and PC-2 connections and monolithic
connection MC were loaded in lateral form and constant axial force
on the column. The lateral force was applied to the system step by
step to specimens fail in positive direction or negative direction.
The column was supported by pined connection at its base and free
on top. Roller supported was modeled to end of beams. Various
axial loads was applied to the columns at each specimens. The
boundary conditions of specimens are shown in Fig. 6.
4. Numerical modeling
4.1. Material properties
To model the behavior of the two proposed moment-resisting
connections a finite element model was developed using ABAQUS
software. The beams have 400
500 mm rectangular cross-
section and 2.5 m length. The columns have square cross-section
with dimensions of 400 mm. The longitudinal reinforcement of
the beam and column was deformed bars of Grade 420, while
the beam stirrups and column transverse ties were applied with
(b) Precast beam-to-column connection
 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821
Fig. 6. Boundary condition and loading of specimens.
Fig. 7. Proposed stress-strain model for reinforcing steel by Mander et al. [20,21].
Grade 300 bars. The compressive strength of concrete and grout
used for all specimens was 25 and 45 MPa respectively. Steel
plates, U sections, and box sections used in the constructions were
of st 37 with f y ¼ 240 MPa and f u ¼ 360 MPa in which f y and f u are
Fig. 8. Modelling of bottom bar connection to the corble using coupling nodal forces at PC-1 specimen.
813
Fig. 9. Elliot et al. welded connection [12].
yield and ultimate strengths respectively. The weld material is of
E60 electrode having as strengths f y ¼ 350 MPa and
f u ¼ 430 MPa. Modulus of elasticity and Poisson’s ratio are consid-
ered as 200 GPa and 0.3 for steel and 24 GPa and 0.2 for concrete,
respectively.
Concrete material is modeled with ABAQUS predefined concrete
damage plasticity. In this approach, damage parameter is defined
as a function of strain, providing a multiplier that reduces the mod-
ulus of elasticity as more stress, and consequently strain, is intro-
duced. Damage parameter is separately defined for tension and
compression. For the elastic portion of the concrete deformations,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0
modulus of elasticity is considered to be 4700 f c ðMPaÞ suggested
by ACI-318-2011 [16]. The constitutive relationships, i.e. the stress
814 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821

Fig. 10. Mesh of connection TW1-C.

Fig. 11. Cracking distribution in numerical model compared to the experimental specimen connection TW1-C.

100
strain relationship, in the nonlinear part is defined by modified
Hognestad quadratic curve. This constitutive stress-strain model
Connector Moment (kN.m)

considers two major mechanisms of rupture in concrete materials:

75
(a) tensile cracking and (b) compressive fracture. The compressive
behavior has a parabolic function up to the maximum compressive
50
stress. After ultimate compressive point, the post cracking behav-
ior is taken into account by specifying a post linear stress-strain
relation up to the maximum strain. The tensile behavior is linear
25 initially followed by a strain softening after the ultimate tensile
Analytical point.
TW1-C__BM2 Modeling the cracks in the concrete material due to its rapid
0 development is more difficult its observation in controlled envi-
0 0.0025 0.005 0.0075 0.01 ronment is nearly impossible. Thus, a few assumptions are made;
qffiffiffiffi
Relative Rotation (rad) 0
up to f r ¼ 0:7 f c , which is the modulus of rupture, concrete
Fig. 12. Moment-relative rotation curve of Elliot et al. connection TW1-C [11].

160 14

140 12
Dissipated Energy (kN.m)

120
Lateral Load (kN)

10
100
8
80
6
60
4
40 MC MC
PC-1 2 PC-1
20
PC-2 PC-2
0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Displacement (mm) Displacement (mm)
(a) Lateral load capacity (b) Energy dissipation
0
Fig. 13. Lateral load-displacement and energy dissipation curves of specimens under compressive load of 0:2f c Ag .
 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821 815

180 16

160 14

Dissipated Energy (kN.m)

140 12

Lateral Load (kN)

120
10
100
8
80
6
60
40 4 MC
MC
PC-1 2 PC-1
20
PC-2 PC-2
0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Displacement (mm) Displacement (mm)
(a) Lateral load capacity (b) Energy dissipation
0
Fig. 14. Lateral load-displacement and energy dissipation response of specimens under compressive load of 0:5f c Ag .

140 10

Dissipated Energy (kN.m)

120
8
Lateral Load (kN)

100

80 6

60 4
40
MC MC
2
20 PC-1 PC-1
PC-2 PC-2
0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Displacement (mm) Displacement (mm)
(a) Lateral load capacity (b) Energy dissipation
Fig. 15. Lateral load and energy dissipation versus displacement of specimens under tensile load of 0:5f y As .

0
Fig. 16. Cracking distribution in specimen PC-1 under compressive axial force of 0:2f c Ag at drift 1%.
816 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821

responds with the same manner as in compression; however
Young’s modulus in tension and compression, due to presence of
microcracks is slightly different, but it is neglected for the sake of
simplicity. Beyond the strain corresponding to fr, stress is assumed
to be linearly decreasing with strain, up to approximately 10–12
times the rupture strain.
The monotonic behavior of longitudinal steel was modeled
using Mander et al. model [20]. The model of Mander et al.
(1984) was developed as a result of many tension and compression
coupon tests. This model, which takes into account elastic behav-
ior, yield plateau, and strain hardening of steel material, has three
main regions, as shown in Fig. 7. The first region is a linear function
with slope equal to steel’s modulus of elasticity; the region ends at
the yield point with stress equal to yield stress of steel. The second
region simulates yield plateau, and the third region is an ascending
curve up to the maximum strength of steel, simulating the strain
hardening region of steel behavior. The post-ultimate stress region
is not considered in the Mander et al. model [21].
4.2. Finite element modeling
It is possible to more thoroughly evaluate the stresses and
deformations in a structure using the FE analysis than can be done
experimentally. The nonlinear analysis results in a better its load-
ing to fracture. In the present study, the specimens were analyzed
using the ABAQUS software. Three-dimensional (3D) elements
were applied to simulate the concrete and steel plates, while rein-
forcing bars were modeled as truss elements. ABAQUS 6.10 pro-
vides 6 types of three-dimensional (3D) stress/displacement
elements for modeling concrete, including the 4-node linear tetra-
hedron, the 6-node linear triangular prism, the 8-node linear brick,
the 10-node quadratic tetrahedron, the 15-node quadratic triangle

0
Fig. 17. Cracking distribution in specimen PC-2 under compressive axial force of 0:2f c Ag at drift 1%.

0
Fig. 18. Cracking distribution in specimen PC-1 under compressive axial force of 0:5f c Ag at drift 1%.
 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821 817

and the 20-node quadratic brick. To cover the concrete behavioral
properties and minimizing problematic convergence of 3D models,
the 8-node linear brick element, titled C3D8R was chosen for 3D
modeling. The suffix ‘‘R” at the title of the element indicates the
reduction of integration points to decrease the required run time.
This element has three degree of freedom at each node.
In ABAQUS, rebar can be specified as smeared layers in mem-
brane, shell, or surface elements or they can be included in contin-
uum elements by embedding rebar surface or membrane elements
into continuum element. Here, steel bars were modeled by truss
elements and interaction between bars and concrete were speci-
fied using embedded capability in the software. This displacement
capability interpolates the node related to the bar, with the node
related to the concrete element. This feature can simulate the
interaction of rebar and concrete to an acceptable limitation [17].
In connection PC1, tensile force from the bottom bars are trans-
ferred to the corbel via a nuts and a circular annulus plate. In the
model, the far end node of the bar is coupled to the face of an annu-
lus object, so that the sum of the forces on the annulus plate is
equal to the force of the bar. The plate is in contact with the inner
face of the corbel to transmit the forces. Fig. 8, schematically,
shows the coupling of the nodes on the plate face and the bar.
Adhesion between concrete surfaces with steel sheets and surfaces
between precast and in situ concretes was disregarded as insignif-
icant. Only tangential behavior between surfaces was taken into
account. The average effective coefficient of static friction varied
between concrete surfaces was 0.57 for dry interface to 0.7 for
wet interface [18,19]. Proposed friction between the steel and con-
crete surfaces was 0.7.
Welds connections in steel parts of the connection are modeled
3 dimensionally in order to observe the stresses in the welds. This
gives us the advantage of observing the stress in welds and adjust-
ing the sizes and locations in order to minimize the stress concen-
tration. Concrete to concrete or steel interfaces are considered to
interact only tangentially, since the normal contact strength is
reasonably small. The concrete element was the linear 8-node

0
Fig. 19. Cracking distribution in specimen PC-2 under compressive axial force of 0:5f c Ag at drift 1%.

Fig. 20. Cracking distribution in specimen PC-1 under tensile axial force of 0:5f y As at drift 1%.
818 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821

element with dimensions of 15 mm. Elements related to linear 20000

two-point rebar with dimensions measuring less than half of the MC PC-1 PC-2

Lateral stiffness (kN/m)

concrete element were also considered. Numerical integration
used in the software was the Gauss method, and the method of 15000
nonlinear system solution of the software was the Newton Raph-
son method. A maximum limit of 50 iterations was used for the
convergence and the tolerance was taken as 0.002. From the anal- 10000
yses it was observed that the convergence generally occurred in
less than 7 iterations. Initial investigations on some specimens
5000
showed the effect of geometric nonlinearity did not much vary
the results since extremely large displacements did not occur.
Hence, the geometric nonlinearity was neglected in the analysis. 0
0.2Agf'c 0.5Agf'c 0.5Asfy
Axial Load
5. Verification of modeling
Fig. 22. Lateral stiffness of specimens under various axial loads.
For verification of the developed finite element model, the
Elliott et al.’s welded plate connection (Fig. 9) was analyzed
[11,12]. In a proposed Elliott et al. connection, the narrow mild 175

Maximum Lateral Load (kN)

steel plate by 25 mm thick, 500 mm long and 100 mm deep was
anchored in the beam by welding two bent bars on each side of MC PC-1 PC-2
the plate. A solid section 100
100 mm cast into column projects
a distance of 90 mm from the beam bottom. The level of the billet 150
top was at 190 mm from the bottom of the beam. After a double
sided fillet site weld was made to the narrow plate, 80 mm long
and 20 mm thickness, a structural grade cast in situ infill concrete, 125
completed the joint at the end of the beam (TW1-C). The centroid
of the weld was therefore at 200 mm from the beam bottom. The
connection contains two precast-insitu interfaces as indicated in
Fig. 9. Vertical load using hand pumped hydraulic jacks on precast 100
0.2Agf'c 0.5Agf'c 0.5Asfy
beam end were applied incrementally. Beam to column rotations
were measured using four linear displacement transducers as Axial Load
known distances from the column face. The slope of vertical load Fig. 23. Lateral resistance Capacity of beams under various axial load levels.
vs displacement graph measured at 90–300 mm from the column
face. For verification, a total of 4700 C3D8R elements and 2300
T3D3 elements each with a size of about 30 and 15 mm, respec- sequence. Fig. 12 shows the numerical and experimental connector
tively, were used in the FE model (Fig. 10). Cracking distribution moment vs relative rotation graph of Elliot et al. connection TW1-
in analytical model and the Elliott et al.’s specimen’s cracking pat- C. In comparison with the resulted from the testing program, the
tern are shown in Fig. 11. As it can be seen in Fig. 11, Shear crack, analytical model offers a good result. The two curves almost over-
compressive cracks and gap opening at the top of beam to column lap over the elastic region and nearby the maximum displacement.
face under vertical load at experimental observation were The face column moment that correspond to the maximum drift
observed at analytical model. The monotonic force-displacement differ by only 1.69 kN m. Moreover, the analytical model shares
curve was obtained after applying the monotonic loading almost the same initial stiffness as the testing specimen, the differ-

Fig. 21. Cracking distribution in specimen PC-2 under tensile axial force of 0:5f y As at drift 1%.
 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821 819

ences being lower than 3.5%. In general, however, the comparison page or compressive concrete strain reaching ultimate strain of
of the numerical results with the experimental data shows that the concrete.
numerical model could capture quite accurately the overall behav- Figs. 13 and 14 show load-displacement curves and energy dis-
0
ior of the beam-to-column connection and the distribution of the sipation of the connections under compressive axial load of 0:2f c Ag
forces along the connection bars. 0
and 0:5f c Ag , respectively. According to this figures, the ultimate
displacement of monolithic concrete is greater than the ultimate
displacement of precast connections. Energy dissipation of the sys-
6. Modeling results
tem is equal to the area under the load-displacement graph.
Results indicated that lateral strength and energy dissipation of
The lateral force was applied to the column top step by step
the specimens improved when the axial load of the column
until the specimen fails. The ultimate lateral displacement of the
increased. As shown in Fig. 15 changing compression load of
column is defined to be corresponding to bar fracture or bar slip-

Table 1
Numerical results from FE model.

Specimen Axial load P max (kN) dy (mm) du (mm)) l S Pmax

PMC
lmax
lMC
S
SMC

0
MC 0:2f c Ag 145 14 104 7.4 15,480 1 1 1
PC-1 144 16 89 5.6 12,627 0.99 0.75 0.81
PC-2 145 14.5 92 6.3 13,625 1 0.85 0.88
MC 163 14.5 103 7.1 17,543 1 1 1
0
PC-1 0:5f c Ag 161 16 87 5.4 14,042 0.99 0.75 0.80
PC-2 158 14.5 90 6.2 14,130 1 0.97 0.81
MC 118 17.5 101 5.8 9600 1 1 1
PC-1 0:5f y As 120 18.5 86 4.6 9087 1.02 0.80 0.95
PC-2 118 18.5 86 4.6 9407 1 0.80 0.98

160 16
140 14
Dissipated Energy (kN.m)

120 12
Lateral Load (kN)

100 10
80 8
60 6
40 fc=25 MPa 4 fc=25 MPa
20 fc=30 MPa fc=30 MPa
2
fc=35 MPa fc=35 MPa
0
0 20 40 60 80 100 120 0
0 20 40 60 80 100 120
Displacement (mm) Displacement (mm)
(a) Lateral load capacity (b) Energy dissipation
0
Fig. 24. (a) Lateral load and energy dissipation versus displacement of specimen MC under axial load of 0:2f c Ag .

160 12

140
10
Dissipated Energy (kN.m)

120
Lateral Load (kN)

8
100

80 6

60
4
40 fc=25 MPa
fc=25 MPa
fc=30 MPa 2 fc=30 MPa
20
fc=35 MPa fc=35 MPa
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Displacement (mm) Displacement (mm)
(a) Lateral load capacity (b) Energy dissipation
0
Fig. 25. (a) Lateral load and energy dissipation versus displacement of specimen PC-1 under axial load of 0:2f c Ag .
820 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821
160
140
120
Lateral Load (kN)
100
80
60
40 fc=25 MPa
fc=30 MPa
20
fc=35 MPa
0 0
0 20 40 60 80
Displacement (mm)
(a) Lateral load capacity
Fig. 26. (a) Lateral load and energy dissipation versus displacement of specimen PC-2 under axial load of 0:2f c Ag .
0
0:2f c Ag to tensile load of 0:5As f y significantly decreased the lateral
resistance, energy dissipation and ultimate deformation capacity
by 22%, 32% and 7%, respectively.
The first flexural cracks were observed in the beam at 0.25%
drift ratio in all the specimens and flexural yielding occurred at
around 1.0% drift level. The cracking distribution in specimens
PC-1 and PC-2 under lateral load and compressive axial force of
0 0
0:2f c Ag , 0:5f c Ag and tensile axial force of 0:5As f y on column are
plotted in Figs. 16–21, respectively.
Figs. 16 and 17 show the crack development in PC-1 and PC-2
0
specimens, respectively under compression load of. 0:2f c Ag .
Although, the strength of the specimen PC-1 reached its yield load,
flexural cracks did not significantly propagate in the beam. This
result indicates that the yielding of the longitudinal rebars in the
beam concentrate to the beam-corbel joint. In PC-2, more flexural
cracks at bottom of beam at vicinity of beam-corbel joint can be
developed.
Crack distribution in PC-1 and PC-2 specimens under compres-
0 from this study:
sion load of. 0:5f c Ag (Figs. 18 and 19) are very similar to the corre-
0
sponding specimen under axial load of 0:2f c Ag . Increasing
compressive axial load decreases the tensile stress of the beam-
column connection. The longitudinal reinforcement of the column
yields before beam’s reinforcement. The deformation capacity and
flexural strength of the connection is decreased, consequently.
Initial lateral stiffness (S) of the connection is equal to the initial
slope of load-displacement curve. The changes in elastic stiffness
and maximum lateral resistance of connections are shown in
Figs. 22 and 23, respectively. As it can be seen in Fig. 22, the initial
stiffness of the monolithic concrete is greater than the precast con-
nections. Increasing the compression axial load, the initial stiffness
decreases and increasing the tensile axial force results in decrease
in initial stiffness.
The displacement ductility factor (l) is the ratio of the ultimate
deformation (Du ) to the initial yield deformation (Dy ). It is not easy
to determine yield points for the specimens directly from the lat-
eral load-displacement curves. For each specimen, the load-
displacement curve was used to define the yield and maximum
displacements according to Park’s criteria [22] for equivalent
elastoplastic energy absorption. Maximum lateral strength (P max ),
yield and ultimate displacement, initial stiffness and displacement
ductility factor of the precast connections to the monolithic refer-
ence connection under various axial load levels on column are
compared in Table 1. Results show the new proposed specimens
can gain at least 99%, 80% and 75% of the Maximum lateral
strength, initial stiffness and displacement ductility factor of the
monolithic reference specimen, respectively.
14
Dissipated Energy (kN.m)
12
10
8
6
4
fc=25 MPa
2 fc=30 MPa
fc=35 MPa
100 0 20 40 60 80 100
Displacement (mm)
(b) Energy dissipation
0
The effect of concrete compressive strength on the seismic
behavior of the connections were investigated. In Figs. 24–26
load-displacement curve of specimens MC, PC-1 and PC-2 are plot-
ted under various concrete compressive strength when the column
0
in under 0:2f c Ag compression axial load. As these figures, increas-
ing the compressive strength of concrete from 25 MPa to 35 MPa
(40% increase), lateral strength, ultimate displacement and energy
dissipation capacity of the specimens increase only 2%, 3% and 4%
respectively. However, the initial stiffness did not changed.
7. Conclusion
The present study investigated the performance of two pro-
posed precast connections of precast beam to column connections.
The proposed connections were analyzed under lateral load and
various levels of axial force using nonlinear finite element model
developed using ABAQUS. The following conclusions were drawn
 According to axial load of columns, the proposed connections
showed lateral resistance between 120 and 167 kN, about 98%
of lateral resistance of the reference monolithic connection in
each level of axial load.
 Stiffness ratio of the precast connections to monolithic connec-
tions ranged between 0.8 and 0.9 depending on the axial load
level of the column.
 The difference of lateral stiffness and resistance of the proposed
connections compared to an reference monolithic connection
could be attributed to the difference of force transfer mecha-
nism and position of the connection source (corbel presence,
manner of force transfer of bars, grouted connection source,
etc.).
 Ductility and energy dissipation of the proposed connections
were measured at approximately70–80 percent of an equiva-
lent monolithic connection. The final strain standard for the
end of the analysis was considered to be 0.0038.
 Lateral stiffness, ductility, and energy dissipation of the connec-
tions increased with increased compressive load of the column;
however, sensitivity of the precast connections was less than
the monolithic connections.
 Connections weakened when tensile load was applied. There-
fore, the decreasing effect of tensile load should be considered
when designing structures and their connections.
 Increasing the compressive strength of concrete results in
higher ultimate displacement, lateral resistance and energy dis-
sipation of connections.
 S. Bahrami et al. / Engineering Structures 132 (2017) 808–821
 Regardless of the limitations of software ABAQUS, it can simu-
late the mechanical behavior of concrete materials. Although
modeling many parameters in the software is not possible,
study results offered an appropriate and acceptable estimation
of the proposed connections behavior.
References
[1] Sucuoglu H. Effect of connection rigidity on seismic response of precast
concrete frames. PCI J 1995;40(1):94–103.
[2] Dolan CW, Stanton JF, Anderson RG. Moment resistant connections and simple
connections. PCI J 1987;32(2):62–74.
[3] Dolan CW, Pessiki SP. Model testing of precast concrete connections. PCI J
1989;34(2):84–103.
[4] Bull DK, Park R. Seismic resistance of frames incorporating precast prestressed
concrete beam shells. PCI J 1986;31(4):54–93.
[5] French CW, Amu O, Tarzikhan C. Connections between precast elements-
failure outside connection region. J Struct Eng 1989;115(2):316–40.
[6] Restrepo JI, Park R, Buchanan AH. Tests on connections of earthquake resisting
precast reinforced concrete perimeter frames of buildings. PCI J 1995;40
(4):44–61.
[7] Khoo JH, Li B, Yip WK. Tests on precast concrete frames with connections
constructed away from column faces. ACI Struct J 2006;103(3):18–27.
[8] Khaloo AR, Parastesh H. Cyclic loading response of simple moment-resisting
precast concrete beam-column connection. ACI Struct J 2003;100(4).
[9] Parastesh H, Hajirasouliha I, Ramezani R. A new ductile moment-resisting
connection for precast concrete frames in seismic regions: an experimental
investigation. Eng Struct 2014;70:144–57.
[10] Shariatmadar H. An investigation of seismic response of precast concrete beam
to column connections: experimental study. Asian J Civ Eng-Build Housing
2014;15(6):849–67.
821
[11] Elliot K, Davies G, Ferreira M, Gorgun H, Mahdi AA. Can precast Concrete
structures be designed as semi-rigid frames? Part 1-The Experimental
Evidence. Struct Eng 2003;81(16):14–27.
[12] Görgün H. Semi-rigid behavior of connections in precast concrete structures.
Doctoral dissertation. University of Nottingham, Department of Civil
Engineering; 1997.
[13] Negro P. Pseudodynamic tests on a full-scale 3-storey precast concrete
building: global response. Eng Struct 2013;57:594–608.
[14] Dionysios A. Pseudodynamic tests on a full-scale 3-storey precast concrete
building: behavior of the mechanical connections and floor diaphragms. Eng
Struct 2013;57:609–27.
[15] Choi H. Development and testing of precast concrete beam-to-column
connections. Eng Struct 2013;56:1820–35.
[16] ACI Committee. Building code requirements for reinforced concrete and
commentary, ACI 318–11/ACI 318R-11. Detroit: American Concrete Institute;
2011.
[17] ABAQUS (2010)/Theory Manual 6.10.
[18] Rabbat BG, Russell HG. Friction coefficient of steel on concrete or grout. J Struct
Eng 1985;111(3):505–15.
[19] Lee YH, Joo YT, Lee T, Ha DH. Mechanical properties of constitutive parameters
in steel–concrete interface. Eng Struct 2011;33(4):1277–90.
[20] Mander JB, Priestley MJN, Park R. Seismic design of bridge piers, University of
Canterbury, New Zealand. Res Rep 1984;84(2).
[21] Shirmohammadi F. Effect of load pattern and history on performance of
reinforced concrete columns. Doctoral dissertation. Kansas State University;
2015.
[22] Park R. Evaluation of ductility of structures and structural assemblages from
laboratory testing. Bull New Zealand Nat Soc Earthq Eng 1989;22(3):155–66.
[23] Bahrami S, Madhkhan M. Behavior under cyclic loading of a new moment
resisting beam to continuous column connection for precast structures. PCI J
2016 [submitted for publication].
[24] Bahrami S, Madhkhan M. Experimental performance of a new precast beam to
column connection using hidden corbel. Asian J Civ Eng 2016 [submitted for
publication].