Composite Structures 185 (2018) 286–303

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Experimental and non-linear numerical investigation of the novel MARK

detachable mechanical joints with laminated plates for composite precast
beam-column joint
⁎
J.D. Nzabonimpa, Won-Kee Hong , Jisoon Kim
Department of Architectural Engineering, Kyung Hee University, Yongin 446-701, Republic of Korea

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

Keywords: The aim of this study is to provide experimental and non-linear numerical investigation of the mechanical
Beam-column moment connection connections with a pair of steel and concrete filler plates laminated to transfer axial loads and moments cross the
Steel-concrete composite precast frames column joints for precast concrete frames. The extended beam and column plates are, then, interconnected by
Extended end plate high-strength bolts. The proposed connection for both steel-concrete composite precast frames and reinforced
Strain rate
concrete precast frames under the condition of fully or partially restrained moment connections is to replace
Plastic analysis
conventional monolithic cast-in-place concrete frames. The structural behavior based on the rate of strain in-
Nonlinear finite element analysis with concrete
damaged plasticity crease of the novel mechanical beam-column joints with laminated metal and concrete plates including the
extended endplates, rebar, and steel flanges was numerically investigated to understand the structural perfor-
mance of the structural elements comprising the proposed mechanical joints for frames. The numerical structural
performance considering concrete damaged plasticity was in good agreement with the test observations at
concrete, rebar, and steel flanges in terms of strains, load-displacement relationship, and the influence of the
endplates on the structural elements of the proposed joints; numerical data were calibrated to test data in
microscopic level to numerically present the real-world behavior with strain evolution of the proposed beam
joints.

1. Introduction connections, depending mainly on the diameter of the bolts and the
endplate thickness [7,8]. Also, bolt failure depends slightly on the
1.1. Conventional cast-in-place concrete/steel-concrete composite frame endplate thickness, which controls the magnitude of the bending effect.
In their analytical investigation, Muresan and Balc [9] pointed out that
A conventional concrete cast requires pour forms, which have been endplates were likely to experience large deformations around the
inevitable in the construction of conventional concrete/steel-concrete tension flange. Recently, Bai et al. [10] introduced a revision to the
composite frames. Traditional precast concrete frames require cast-in- conventional design method for a bolted endplate connection and
place concrete joints to form a moment connection, as shown in Fig. 1 suggested useful design equations for the bending moment in bolts and
[1]. The joint concrete must be cured before further erection of precast endplates.
frames proceeds.
2. Research significance
1.2. Conventional steel connections
2.1. The joint details for fully/partially restrained moment connections;
Conventional extended endplates are used only for transferring beam-to-column mechanical connections for concrete/steel-concrete
moments through steel members [2]; however, studies related to ex- composite frame allowed for rapid erection
tended endplates for steel-concrete composite precast frames and re-
inforced concrete precast members are largely absent from the litera- A concrete filler plate placed between extended metal endplates and
ture. Previous studies investigated the behavior of extended endplates column plates embedded in the column hides nuts and provides an-
subjected to either monotonic or cyclic loadings [3–6]. Those studies chorage to threaded ends of rebars. The plates were, then, connected by
found that such connections fall into either semi-rigid or pinned high-strength bolts to transfer axial loads and moments. The structural

⁎
Corresponding author.
E-mail address: jean@khu.ac.kr (W.-K. Hong).

https://doi.org/10.1016/j.compstruct.2017.11.024
Received 21 October 2017; Accepted 7 November 2017
Available online 08 November 2017
0263-8223/ © 2017 Elsevier Ltd. All rights reserved.
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 1. Typical precast beam-column joints with cast-in-

place moment connection [1].

Fig. 2. Joint and fastener details for precast steel-concrete

composite [11].

capacity of the moment connections depends on the stiffness of the extended endplate connections with concrete filler plates by subjecting
beam endplates, which should be determined to enable the transfer of them to monotonic and cyclic seismic loads to discuss the contributions
full or partial axial loads and the moment at the joints. In this study, the of endplate thickness and the concrete filler plates to the precast frame
extended endplates widely used in steel-frame structures for the con- connections which can act as either fully rigid or semi-rigid joints. The
ventional bolted beam-to-column connections were modified for the influences of endplate stiffness on the design parameters, bolt diameter,
implementation in mechanical joints with concrete filler plates for use number of bolt rows and columns, bolt spacing, bolt grade, stiffeners,
in both steel-concrete precast frames and reinforced concrete precast and column and beam sizes were also explored using a nonlinear finite
frames. Figs. 2 (a) and (b) [11] depict details of the mechanical moment element analysis considering concrete damaged plasticity. The use of
connections with an extended endplate for column-beam joint as- beam endplates and plates embedded in columns to form beam-column
sembly. The primary application of these extended endplates is to joints were proposed to provide moment connections for the column-
provide a fully or partially restrained moment capacity between the beam joint assembly, as shown in Fig. 3 (a). Figs. 3 (b) to (d) exhibited
connected members. The proposed mechanical connection for the the full scale erection test for assembly, demonstrating the efficiency of
beam-to-column joint of a steel-concrete precast composite allows for an assemblage of precast frames compared with conventional precast
rapid erection and does not require the additional expense of fire- construction. Nuts are threaded on the rebar ends located in counter-
proofing because the steel section is encased in concrete. The proposed bores in the filler plates, which are thick enough to completely hide and
novel extended endplate connection can also be implemented in fast- offer anchorage to the nuts. The new joint system consists of steel
track modular offsite construction for reinforced concrete buildings and endplates at the ends of beams and plates on the face of columns bolted
steel-concrete composite buildings, accelerating construction by repla- together. The plates and bolts are designed to transfer force couples
cing the conventional steel-concrete frames that requires pour-forms. (tension and compression) to create moment connections. Additional
Now, the rapid erection of precast frames with verified seismic per- structural elements for the new joints include couplers, threaded rebar,
formance is possible. A significant number of experimental and analy- and anchor rebar. The erection of the proposed frame connection is as
tical studies were conducted to evaluate the seismic behavior of our follows; (1) lift beams, (2) place beams between column plates, (3)

287
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 3. Detachable mechanical joints with lami-

nated plates for composite precast beam-column
joint [11].

connect beam and column plates by bolts. As shown in Fig. 2 (a) [11], to provide connections between the beam plates and the column plates,
the top rebar of the beams is anchored on the rear face of the beam as shown in Fig. 4. The important design parameters to transfer tensile
endplates using threaded ends and nuts. Steel sections for composite forces from the beams to the anchor rebar in the columns are the
beams are welded to the endplates. The anchor rebar in a column unit is stiffness of the endplates and the size of the bolts, which allow moment
connected to the beam endplates using high-strength bolts via couplers resistance at the joints. Prying action was precluded by endplates with

288
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 4. Fabrication of a beam-column joint for testing

[11].

Fig. 5. Instrumented specimens with laminated beam end plates [11].

289
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Table 1 thickness of 45 mm, and the plate thickness of B3 is 20 mm. The plate
Properties of specimens. thickness of B4 is 16 mm, and B6 is a control specimen without an
endplate, being instead a cast-in-place monolithic beam-column joint.
Category Size Material properties
Fig. 5 (e) and (f) illustrate specimen specification and loading protocol
Beam section 300 × 330 mm Concrete fc′ = 21 MPa [11]. Figs. 6–8 [11] demonstrate the hysteretic load-displacement re-
Column section 500 × 500 mm Concrete fc′ = 21 MPa lationships for all beam-column specimens. The beam endplates with 16
Endplate for B2 500 × 530 × 45 mm SM490, Fy = 325 MPa and 20 mm thickness were deformed regardless of the type of filler
Endplate for B3 500 × 530 × 20 mm SM490, Fy = 325 MPa plates, as shown in Table 2 [11] and Figs. 6 and 8 [11]. These endplates
Endplate for B4 500 × 530 × 16 mm SM490, Fy = 325 MPa
successfully transferred tensile forces from the beams to the columns as
Endplate for B5 500 × 530 × 20 mm SM490, Fy = 325 MPa
B6 (no endplate) 300 × 330 mm Concrete fc′ = 21 MPa
partially restrained moment connections. However, the beam endplates
Column plate 500 × 530 × 10 mm SM490, Fy = 325 MPa with 45 mm thickness transferred tension forces from the beam to the
Beam rebar 4-HD25 fy = 500 MPa column, forming a joint rigid moment connection, as shown in Spe-
Column rebar 4-HD25 fy = 500 MPa cimen B2. We used the strains measured from the four test specimens
Hoops and stirrups HD10@200 SS400, fy = 235 MPa subjected to cyclic loadings to evaluate the structural behavior of the
Steel section for all specimens H-200x100x4.5x7 SM490, Fy = 325 MPa
proposed beam-to-column joint with concrete filler plates, including the
Steel section for column H-250 × 250 × 9 × 14 SM490, Fy = 325 MPa
Nut M22-F10T fy = 900 MPa deformations and stresses of the beam endplates and the failure mode of
High-strength bolts M20-F10T fy = 900 MPa the frame joints with endplates of various stiffness. The structural be-
Concrete filler for B3 and B4 500 × 530 × 22 mm Concrete fc′ = 21 MPa havior of the proposed beam-to-column joint with concrete filler plates,
Metal filler plate for B5 500 × 530 × 22 mm SM490, Fy = 325 MPa including the deformations and stresses of the beam endplates and the
failure mode of the frame joints with endplates having various stiffness
was evaluated based on the strains measured from the four test speci-
sufficient stiffness, resisting the coupled forces of tension and com-
mens subjected to cyclic loadings. The influence of the beam endplate
pression and contributing to moment transfer through the joints.
deformations on the flexural capacity of the specimens was also nu-
merically investigated to compare with the strain test data retrieved
2.2. Stiffness of endplates from the reinforcing bars/steel and beam endplates. The adequately
designed joint plates demonstrated a fully restrained moment connec-
The plates and bolts shown in Figs. 2 and 4 [11] were designed tion equivalent to that found in conventional steel-concrete composite
based on a nonlinear finite element analysis. The influence of the moment frames.
stiffness of the metal plates on the strains of the structural components,
rebar, steel flange, and concrete, was evaluated experimentally to un-
derstand the load paths as the over-turning moments progressed. 3.2. Control Specimen (B6) and Specimen B2

3. Experimental investigations Specimen B6 was a control cast-in-place specimen with a monolithic

beam-column joint comprising a steel and reinforced concrete frame.
3.1. Fabrication of test specimens and test set-up Specimen B6 demonstrated a maximum load capacity of 158 kN at a
stroke of 90 mm (174 kN, 90 mm), as presented by the authors [11].
Beam-column joints for testing were fabricated as can be seen in Compressive failure modes were observed as loadings increased to the
Fig. 5 [11]. The photos exhibit the couplers and high-strength bolts that maximum load limit state [11]. Specimen B2 presented a flexural ca-
connect the column plates to the threaded ends of the anchor rebar and pacity similar to that of the conventional concrete frame without joint
beam endplates, respectively. The couplers were tack-welded to the plates because the endplates were stiff enough to transfer the tensions
column plates to ensure the locations of the anchor rebar. All the di- exerted from the rebar to the column. The steel beams and reinforcing
mensions and material properties used in the experimental investiga- bars were fracturedby the tension before structural degradation oc-
tion are provided in Table 1. Four test specimens equipped with gauges curred in the joint plates, demonstrating that Specimen B2 had failure
are presented in Fig. 5. In that photo, the beam endplates in B2 have a modes similar to those of the monolithic beam-column joint. The plate

Fig. 6. Load-displacement relationships; Specimen B3 [11].

290
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 7. Failure modes for specimen B3 [11].

Fig. 8. Load-displacement relationships

and failure modes for Specimen B4 [11].

Table 2 Fig. 7 [11], which demonstrates failure modes to understand the load
Test summary [11]. paths as the over-turning moments increased. Specimen B3, which was
fabricated with a 20 mm endplate and a concrete filler plate between
Specimen (Plate thickness) Deformation Prying action
the beam and column plate, demonstrated a maximum load capacity of
B2 (45 mm) Embedded nut No deformation No (Fully restrained) 84 kN at a stroke of 105 mm (85 kN, 75 mm), as shown in Fig. 6 [11].
B3 (20 mm) Concrete filler plate 14–18 mm Yes (Partially Rapid load reduction was observed after reaching the maximum load
restrained)
due to the slippage of the nuts off the threaded ends of the rebar, which
B4 (16 mm) Concrete filler plate 15–20 mm Yes (Partially
restrained) began at the stroke of 135 mm, as shown in Fig. 6 [11]. Significant
B5 (20 mm) Metal filler plate 13–17 mm Yes (Partially failures began to appear at the interface between the endplates and
restrained) concrete beam at the stroke of 120 mm–135 mm, causing the displaced
B6 Control monolithic – – nuts to collide with the concrete filler plate. The forces holding the
specimen
rebar decreased in the vicinity of that stroke, causing the reduction of
load resisting capacity. Failure modes at the stroke of 135 mm and
embedded in the column had no noticeable surface damage [11]. The 150 mm led to significant separation between the interface and the
nuts with no damage did not cause strength degradation to the joint. exposed threaded end of the rebar (Fig. 6 [11]) until the test was ter-
Fig. 9 [11] compares the experimental hysteretic load displacement minated with sudden drops of loads. The use of nuts with enough head
relationships in the plates of 45 mm (B2) and 16 mm (B4) with those of length and nuts holes large enough in diameter are recommended to
the control beam (B6); the influence of the stiffness of the beam end- prevent such slippage and the beating of nuts against the filler plates.
plates on the load transfer at the joints is demonstrated by the large The two specimens, one with a steel filler plate and one with a concrete
strength reduction of the beam with the 16-mm plate (B4). The load filler plate, represented by Legends 4 and 2 in Fig. 9 [11], respectively,
displacement relationships in the control specimen (B6) were similar to have beam endplates of the same thickness. The load bearing capacity
those of Specimen B2, which was fabricated with a plate stiff enough to of the specimen with the steel filler plate is slightly higher than the one
suppress deformation. with the concrete filler plate until the stroke of 100 mm. However, the
beam-column joint with the concrete filler plate demonstrated better
ductility, as shown in Fig. 9 [11]. The concrete filler plate demonstrated
3.3. Specimen B3 bearing failures in the final failure modes due to compression, but, in
general, the filler plate was not completely damaged and successfully
The evolution of degradation at a stroke of 150 mm is presented in protected the nuts as described in Section 6.2. The deformation of the

291
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 9. Control beam (B6) vs. beams with an extended plate [11].

Fig. 10. Proposed contact elements between

metal and concrete for the plate and bolts de-
sign for B3; failure mode matched at the end of
the test, Fig. 7.

Fig. 11. Concrete.

endplate of the disassembled specimen was measured after testing, as the gradual deformation of the plate before the strength was entirely
shown in the final failure modes of Fig. 7 (b), in which the deformation lost when the steel flange fractured. Additional interior bolts, however,
of the beam plate was 14–18 mm at a stroke of 150 mm. Specimen B3 might be required to reduce the unsupported length between the ex-
was one of the specimens that did not provide enough stiffness for a terior bolts and decrease the deformation of the plates.
rigid connection. Fig. 10 shows the plate deformed by as much as
18 mm, calculated based on the nonlinear finite element analysis with
3.4. Specimen B4
concrete damaged plasticity. The load-displacement relationship ex-
hibited ductile behavior because the inelastic energy dissipated during
Specimen B4, fabricated with a 16-mm endplate and a concrete

292
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 12. Rebar.

Fig. 13. Steel section.

Fig. 14. Metal plate.

filler plate between the beam and column plates, exhibited a maximum specimen’s significant reduction in flexural strength. The unexpected
load capacity of 72 kN in a stroke range of 225 mm (−73 kN, 150 mm), distortions of the nuts and threaded ends of the rebar can be prevented
as depicted in Fig. 8 [11]. A constant load-bearing capacity in this by minimizing the twisted distortion of the plates or providing enough
stroke range was observed because the structural degradation with gaps between the nuts and the holes in the plates. Fig. 9 [11] compares
energy dissipation was concentrated on the beam plate; the deforma- the hysteretic envelope of the load deflection relationship of the control
tion of the beam plate gradually increased to 15–20 mm at a stroke of beam (B6), a conventional rigid joint, with that of the beams with ex-
150 mm. The concrete specimen did not show noticeable deterioration tended endplates with sufficient and insufficient stiffness. Arrow #1,
or strains because all the damage was concentrated on the plate, which representing Specimen B4 assembled with a beam endplate of 16 mm
experienced extensive deformation, as observed in Fig. 8 [11]. The and concrete filler, indicates the strength reduction of the moment re-
strains of the beam endplates of Specimens B3 and B4 were greater than sistance. Specimen B3 (20 mm, concrete filler) with endplates of in-
those of B2 (Fig. 14), whereas the strains of the rebar and steel section sufficient stiffness showed similar strength reduction to Specimen B4;
of B3 and B4 were smaller than those of B2 and the control specimen the load bearing capacity of the specimen with a 20-mm plate was
(Figs. 12 and 13(b)). Nuts were displaced during the transfer of the slightly greater than that of the one with a 16-mm endplate. The slips of
tension from the rebar because the deformed plate pushed the nuts off the nuts began at the stroke of 165 mm and were accompanied by the
the rebar threads, as shown in Fig. 8 [11]. Specimen B4 demonstrated separation of the interface between the beam endplate and the em-
the least flexural strength among all the test specimens because the bedded concrete plate, as shown in Fig. 8 [11]. The nut fracture con-
endplate (only 16 mm) had the least stiffness. Fig. 8 [11] shows the tinued, and the nut slipped off the threaded ends of the rebar com-
plate deformation, as large as 15–20 mm, which was responsible for the pletely at the stroke of 225 mm. Eventually, the rebar could no longer

293
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 15. 3D FE model for Specimen B3.

Fig. 16. Interaction properties; Surface-to-surface contact.

exert tension onto the metal beam endplate, causing a decrease in the flanges, and concrete beams was discussed.
load resisting capacity. The test was terminated with a sudden drop in
load shortly after the nuts fractured at the 200 mm stroke. The low rate
of strain increase of the rebar measured from B4 (Fig. 12) made the nuts 4.1. Load-displacement relationship
in Specimen B4 less susceptible to failure. The concrete filler plate of B4
showed fracture modes similar to those of B3, which was crushed by the The load-displacement relationship is identified in Fig. 9 [11],
compression, but in general, the filler plate was not completely da- where load and displacement were marked at the concrete strain of
maged. 0.002 and 0.003. The displacement, at which the nuts were displaced
from the threaded end of the rebar, is also marked in Curve B3. B2
4. Microscopic observation of strains and rates of strain increase showed no noticeable deterioration, with no rapid drop of loads and no
stiffness degradation up to the stroke of 100 mm. Specimen B2 showed
The structural behavior of the proposed precast beam-column joints only compressive failure modes for the concrete and tensile failure
was investigated based on strains measured during testing modes for the rebar/steel section as loading increased to the maximum
(Figs. 11–14). In this section, the influence of the stiffness of the metal load limit state.
plates on the strains of the structural components including rebar, steel

294
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 17. Constitutive relations with proposed damaged plasticity of concrete [14].

4.2. Influence of metal and concrete plates on the rates of the strain increase endplates as the applied loads increased, limiting the contribution of
of concrete the rebar to the flexural capacity of B3 and B4, as shown in Fig. 12
(a) and (b). B3 and B4 demonstrated very low rates of strain increase
A larger contribution from concrete was exhibited by Specimens B2 (Fig. 12 (b)), causing a rotational rigid body mode after 40 and 60 mm
and B6 (control) than by Specimens B3 and B4; the beam endplates of strokes, respectively.
B3 and B4 were deformed, preventing the concrete from being fully
activated, as shown in Fig. 11 (a) and (b). The concrete of B3 con- 4.4. Influence of the stiffness of plates on the rate of strain increase of the
tributed to the resistance against loads up to the concrete strain of steel sections
0.003, whereas B2 and B6 reached strains of 0.008 and 0.0045, re-
spectively. The strains reached by B2 and B6 were the largest, followed The rate of strain increase of the steel flanges, which follows trends
by those of B3 and B4, with a significant difference between them, as similar to those of the rebar, is shown in Fig. 13 (a) and (b). The flange
shown in Fig. 11 (b). The concrete of Specimen B4 was the least acti- strains (0.0023) reached by Specimens B2 and B6 were larger than
vated because of the concentration of the deformation on the thin beam those (0.0019) of B3 and B4 because their metal plates deformed due to
endplates, causing the concrete to rotate with respect to the base their insufficient stiffness, as shown in Fig. 13 (b).
without strain changes, similar to rigid body rotation. The concrete The steel strain of B3 and B4 was lower than that of B2 and B6,
strains were measured up to a stroke of only 75 mm due to gauge limiting the contribution of the steel flanges to resistance against loads
malfunction. in B3 and B4. The results indicate that the steel flanges were not acti-
vated enough to contribute to the flexural capacity of the frame in B3
4.3. Influence of metal and concrete plates on the rates of the strain increase and B4. On the other hand, strains of 0.0025–0.0023 were reached by
of rebar the rebar and steel flanges, respectively, of B2, implying that the
structural elements of the steel-concrete composite frames are activated
Significant deformation of beam endplates occurred in Specimens when the metal plates offer sufficient stiffness.
B3 and B4, whereas the metal plates of B2 did not undergo any no-
ticeable deformation, as shown in Table 2 [11]. The rebar strains ob- 4.5. Strength of extended endplate and concrete filler plates
served in Specimens B2 and B6 were larger than those of B3 and B4
because the strain energy was absorbed by the metal plates of B3 and As shown in Fig. 14 (b), the metal beam endplates of B3 and B4
B4, as shown in Fig. 12 (a) and (b); the results indicate that rebar strain reached similar strains, whereas the plate of B2 reached only 0.0013,
was not fully activated in B3 and B4. The rebar strain of the control indicating that the 45-mm endplates of B2 did not undergo noticeable
specimen reached 0.0035, and the measured rebar strains of B4 and B3 deformation. In fact, endplates of B2 drive the full stress activation of
reached 0.0025 and 0.0018, respectively. The loads increased to the the other structural elements (concrete, rebar, and steel sections, as
maximum load capacity of 83.4 kN at a stroke of 105 mm (83.4 kN, shown in Figs. 11 (b), 13 (b) and 14 (b), respectively). The rates of
105 mm), with a positive load direction for Specimen B3 followed by a strain increase of the beam endplates of B3 and B4 are greater than
rapid load reduction. The initial rate of the strain increase was similar those of B2 because the plates without sufficient stiffness are more
among all specimens. However, the rates of the strain increase of B3 vulnerable to deformation, as shown in Fig. 14 (b). The metal beam
and B4 (Fig. 12 (b)) rapidly decreased from a strain of 0.001 at a 40-mm endplates of B4 did not contribute to transferring forces from the rebar
stroke and 0.0017 at a 60 mm stoke, respectively, implying that both and steel flanges to the columns as much as they did in B3 because the
the rebar and beam endplates of B3 and B4 contributed to the resistance beam endplate of B4 deformed more than that of B3, as verified by the
of the loads when the loads were small in the beginning of the loading greater rate of strain increase of B4 (16 mm endplate) compared with
application. However, the deformation concentrated on the beam that of B3 (20 mm endplate), as shown in Fig. 14 (b).

295
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 18. Contribution of the metal plates to the flexural capacity of the beams, numerical.

5. Computational evaluation of the structural performance relationship, eccentricity, and viscosity. The influence of the strength of
the metal beam endplates and concrete filler plates on the flexural ca-
5.1. Mathematical model pacity of the connection was numerically investigated. The proposed
nonlinear numerical models also provided failure modes for the metal
The proposed novel composite precast concrete frames jointed by beam endplates and concrete filler plates. The presented analytical
metal beam endplates and concrete filler plates were explored using a model was found very efficient and accurate in evaluating not only the
nonlinear numerical analysis based on concrete damaged plasticity. The failure mode of the connection but also the stress and strain fields of the
material properties used to simulate the finite element (FE) model are metal plates and concrete fillers along the column-beam interface. The
identical to those used to design the test specimens (Table 1). The proposed mechanical joint exhibited structural behavior similar to that
proposed joints consist of a pair of metal plates with a concrete filler seen with steel filler plates.
plate between them, requiring that the numerical models account for
surface-to-surface contact between the metal plates and the concrete
filler plates, which are contacted with nuts threaded on the rebar ends. 5.2. Mesh and modeling of contact elements
The parameters were calibrated by the test results in this study; they
include dilation angle, concrete damage factor, pressure-overclosure Generating meshes is a crucial consideration in FE analyses. Clearly,
a good mesh leads to reliable results, whereas bad meshes provide

296
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 19. Locations of the microscopic strains for the comparison of test and numerical investigation for B2 and B3.

Table 3
Microscopic strains of selected rebar (fy = 500 MPa, yield strain: 0.0025) and steel sections (Fy = 325 MPa, yield strain: 0.001625).

Specimens Experimental investigation Computational investigation Rebar status

(a) Rebar strains at a concrete strain of 0.0025

B2 (plate t = 45 mm) 0.0025 0.00252 Yielded

B3 (plate t = 20 mm) 0.0013 0.00174 Not yielded
B4 (plate t = 16 mm) Strain gauge malfunctioned 0.00161 Not yielded
B5 (plate t = 20 mm) 0.00180 0.00203 Not yielded
B6 (control: no plate) 0.00250 0.00242 Yielded

Specimens (b) H-steel strains at a concrete strain of 0.0025 H-steel status

B2 (plate t = 45 mm) 0.00241 0.00201 Yielded

B3 (plate t = 20 mm) 0.0011 0.00103 Not yielded
B4 (plate t = 16 mm) Strain gauge malfunctioned 0.00073 Not yielded
B5 (plate t = 20 mm) 0.00131 0.00112 Not yielded
B6 (control: no plate) 0.00247 0.00213 Yielded

Fig. 20. Relationships of strain-stress and strain-stroke for concrete filler plate for B3, numerical analysis.

warnings and numerical errors, resulting in inaccurate results. The z directions). A total of 157,176 elements and 183,450 nodes for Spe-
ABAQUS [12] library provides a wide variety of elements, including 3D cimen B3 were modeled, as indicated in Fig. 15.
solid elements, 1D elements, and 2D elements. However, before Fig. 16 illustrates the assigned interaction properties for two sur-
choosing elements for a particular FE model, experienced ABAQUS faces that interact (master surface and slave surface). In the proposed
users carefully examine the problem domain. In the analysis of the mechanical joint, a total of six interactions (Int-1 to Int-6) was defined
proposed mechanical joint, 3D solid elements of type (C3D8) were used based on tangential and normal behavior. The tangential behavior was
with three degrees of freedom at each node (translations in the x, y, and set to be frictionless, and the normal behavior was defined using the

297
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 21. Stress-strain relationships, numerical analysis.

penalty method (standard) for contact enforcement and hard contact 5.3. Damaged plasticity model for concrete with constitutive relation
pressure-overclosure. To control the movement of bodies in contact, the
tracking approach was used to allow small slides between the bolt The input values for the constitutive relationship of concrete were
shank and the plates’ holes (Int-5), whereas finite sliding was assigned established based the on strain-stress curve proposed by Kent-Park [13]
the remaining interactions (Int-1 to Int-4, and Int-6). In most cases, a and assigned the damage factor (dt ) for compression starting from the
small sliding approach (Int-5) is selected to adjust contact surfaces with end of the linear point corresponding to 0.4fc′, as indicated in Fig. 17
nonmatching meshes. In this case, a kinematic constraint was issued to (a) and (b). In tension, the damage factor (dt ) was introduced in Fig. 17
prevent nodes on the slave surface (plate) from penetrating the master (c) and (d) when the concrete reaches its maximum tensile stress (σt0) .
surface (bolt). Large deformations were restrained in Int-5; only a little The maximum tensile stress of concrete was taken as 2.1 MPa, and the
sliding between the plates’ holes and bolts was allowed. maximum compressive stress of concrete used for all specimens was
21 MPa. The input values for the assumed constitutive model of con-
crete in both compression and tension are represented in Fig. 17 [14].

298
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 22. Strain-stroke relationships; test data vs. numerical analysis.

Fig. 23. Calibration of the proposed model with test data for B3.

299
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 24. Stress-strain evolution of the joint.

5.4. Influence of the stiffness of the metal plates on the load paths validated Figs. 20–22.
by numerical and test data based on microscopic strain measurements

5.4.1. Contribution of the metal plates to the flexural capacity of the beams, 5.4.2. Strain-stroke relationships, numerical analysis
numerical analysis Fig. 20 (a) and (b) display the relationships of strain-stress and
Specimen B3 (20 mm), which had endplates of insufficient stiffness, strain-stroke for the concrete filler plate. The strain of concrete plate
lost substantial strength, with noticeable plate deformation. The plates increased rapidly at the stroke of 17 mm, as shown in Fig. 20 (b); the
were unable to transfer tension forces between the beam and column in deformation of the beam endplate also increased at a stroke similar to
a fully restrained moment connection. However, the 45-mm beam that shown in Fig. 18 (b), with damage to the concrete filler plate that
plates did successfully transfer loads between the beam and column, propagated rapidly. However, a rapid increase in strain in the metal
creating a rigid joint for moment connection, as shown in Specimen B2. filler plate was not found; thus, the beam endplate with a metal filler
The inelastic energy was dissipated at the plates of Specimen B3, with plate was less deformed than the one with a concrete filler plate after
substantial deformation, rather than at the rebar. The stress-strain re- the stroke of 17 mm, as shown in Fig. 18 (b); the deformation of the
lationships calculated based on the proposed nonlinear model (im- plates between the two specimens (B3 and B5) was similar before
plemented in ABAQUS [12]) are presented in Fig. 18 (a). The strains 17 mm stroke. Fig. 21 depicts the numerical evaluation of stress-strain
and stresses of Specimen B3 for the selected structural components are relationships for selected elements of Specimens B2 and B3. It is clear
plotted up to a concrete strain of 0.003. Locations of strain comparison that Specimen B2, which was designed with a 45-mm plate, did not
for test and numerical investigation are shown in Fig. 19. The tensile exhibit noticeable deformation; the beam endplate and the bolts con-
strains of the beam endplates for B2 and B3 yielded at the strokes of necting the beam endplate to the column plate in Specimen B2 ex-
30 mm and 9 mm, respectively. The plate deformation of Specimen B2 hibited lower strains than those in B3, as shown Fig. 21 (d) and (e). The
does not vary significantly, regardless of the stroke. However, the de- strains of the rebar, steel section, and bolts connected the plates were
formation of B3 increases substantially as the stroke increases, as shown larger in Specimen B2 than those recorded in Specimens B3 and B4, as
in Fig. 18 (b); these results demonstrate that the metal plates of B2 shown Fig. 21 (a), (b) and (c), indicating that those elements were
contributed to the flexural capacity of the beams, fully activating the activated enough to offer strength similar to that in a monolithic beam-
rebar and steel sections. For Specimens B2, B3, B4, and B6 (Table 3), column joint (Specimen B6).
good correlation was exhibited between the measured strains of the Fig. 22 (a) demonstrates that rates of the strain increase of the rebar
tensile rebar and the H-steel section (measured from the gauge) and the vs. stroke were similar up to the stroke of 12 mm for Specimens B2 and
strains calculated numerically at a concrete strain of 0.0025. It is no- B3/B5. However, after the stroke of 12 mm, the strains in Specimen B2
teworthy that a close correlation was found between the experimental increased rapidly, indicating that the rebar is more activated in B2 than
and analytical investigations; this comparison indicates that numerical in B3/B5 (Fig. 22 (a)). Similar trends for the steel section were found as
analysis can provide an attractive alternative to high-cost experimental shown in Fig. 22 (b). The rebar and steel sections of Specimens B2 and
investigations, allowing accurate design of complex composite struc- B3/B5 were activated similarly with small loads at the initial phase of
tural systems. It is also worth noting that the observed strains of the load application, as described in Section 4.3 (Fig. 12). The rates of the
rebar (0.0025) and H-steel section (0.0024) of specimens B2 and B6 strain increase of the metal plates for Specimens B2 and B3/B5 were
were greater than the strains of the rebar (0.0013) and H-steel section compared as shown in Fig. 22 (c). The endplate in Specimens B3 and B5
(0.0011) of B3 because the metal plates with insufficient stiffness in B3 underwent severe deformation with high rates of the strain increase,
and B4 deformed. Table 3 presents direct comparison of strains calcu- failing to provide a rigid joint. A good match for B2 (Fig. 22 (c)) was
lated numerically with those retrieved from test data, yielding accep- observed, indicating that the numerical model accurately predicted the
table reliability and accuracy representing real-world behavior in mi- structural behavior of the endplate when the rates of the strain increase
croscopic level. Computational data in Table 3 were calibrated to test observed from the experimental investigation were compared with
data to predict the structural behavior of the proposed column joints in those obtained numerically.

300
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Fig. 25. Failure mode of concrete filler plate.

301
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

6. Strain evolution in the structural elements of the proposed model of the novel mechanical precast frame connection was also based
composite joint on surface-to-surface contact elements, which were especially suitable
for the metal and concrete filler plates in contact with concrete and the
6.1. Influence of metal plate on strain-stress relationships of structural nuts threaded on the rebar ends. A new contact element between metal
components and concrete with continuum damage modeling approach of successive
initiation of cracks was introduced.
The damaged joint with a concrete filler plate exhibited no notice- The computational observation was validated by performing ex-
able reduction in flexural load bearing capacity compared to the spe- periments to explore the behavior of steel-concrete precast frame joints
cimen with a metal filler plate, as shown in the curves of Legends 3 and with metal beam endplates and concrete filler plates. The nonlinear
6 in Fig. 23, where the damage variables were implemented only in the numerical analysis based on the proposed concrete damaged plasticity
concrete filler plate, providing similar flexural capacities for the two was performed to explore the influence of the laminated metal and
types of joints. Points A and B indicate strains on the load-displacement concrete plates on the seismic performance for precast concrete beam-
relationship corresponding to the compressive concrete strain of 0.003 column joints; they included the failure modes with deformation of the
and 0.008. In Fig. 24, the FE meshes of Specimen B3 representing the metal plates, strains/stresses and flexural capacity of the bolted frame
strains and stresses for the H-steel section, beam endplate, bolts, joint. The computed structural behavior in terms of the microscopic
nominal moments, and displacements of the composite beam are strains was well matched with the test data. The strains, rates of strain
plotted at their designed and ultimate limit states corresponding to increase of the structural elements including laminated metal plates
concrete strains of 0.003 and 0.008. The bolts and beam endplate with were numerically predicted to show how the strains of structural
1.2 mm deformation yielded with a strain of 2εy, whereas the steel components, the connecting metal plates, rebar, steel flanges, and
section remained elastic at the designed limit state; the mechanical concrete, were influenced by the stiffness of the concrete filler plates.
connection of B3 can thus be described as a partially restrained moment The seismic behavior of the proposed joints demonstrated low
joint. nonlinear rates of strain increase for rebar and steel flanges assembled
with extended endplates of insufficient stiffness. The structural beha-
6.2. Failure mode of concrete filler plate vior of the precast frames was identified based on the strains measured
from testing Specimens B3 and B4, with 20 mm and 16 mm plates, re-
The observed failure mode of the concrete filler plate is presented in spectively. The strains in B3 and B4 for the tensile rebar and steel
Fig. 25. The bearing failure was dominant due to the cyclic overturning sections were compared to investigate the influence of metal plates with
loads. The damages and failure modes of the concrete filler plate are concrete filler plates on the flexural capacity of the beam-column me-
also reproduced in Fig. 25 using an FE analysis based on the parameters chanical joints. The results demonstrated that the rates of strain in-
introduced in the legends of Fig. 23, enabling an accurate structural crease in the concrete, rebar, and steel flanges of the precast beam-
evaluation of the mechanical joints and deformation of the concrete column frame fabricated with 20 mm beam endplates were greater than
filler plates. The damage of the concrete filler plate was most severe in those of the precast beam-column frame fabricated with 16 mm beam
the area where compression was concentrated through the rebar. The endplates, indicating that the structural elements in thin plates were
deformation of the concrete plates predicted by the FE analysis was less activated. It is worth noting that the beam endplates of the precast
compared with the experimental results (Fig. 24). The design para- beam-column frame fabricated with both 16 mm and 20 mm-thick
meters for the concrete filler plate in this study were implemented in beam endplates demonstrated larger strains and rates of the strain in-
the nonlinear FE analysis to explore the cracks propagated through the crease than those of the joint with sufficient plate stiffness from a
plates by crushing. In Fig. 25, the failure modes of the concrete filler thickness of 45 mm, indicating that deformation was concentrated on
plate in depth were evaluated at concrete strains of 0.003, 0.008, and the beam endplates with the 20 and 16 mm thickness as the inelastic
the end of the test to compare the numerical damage evolution with the strain energy was dissipated in the plates. However, higher strains and
experimentally observed test failure modes. The compressive crushing rates of strain increase of strain increase were observed for the rebar,
pattern observed during testing did not affect the overall flexural steel sections, and concrete in the specimen (B2) with endplates of
strength of the proposed joint, as also shown in the curves indicated by sufficient stiffness (45 mm thickness) than in the joints with insufficient
Legends 3, 4, and 5 of Fig. 23. The influence of the bearing strength of plate stiffness (20- or 16 mm thickness), for which lower rates of strain
the concrete filler plates on the flexural capacity of the proposed increase for the rebar and steel flanges were identified.
column-beam joints was similar to that of the metal filler plates. The Both the experimental and analytical investigations showed that the
nonlinear numerical analysis with concrete damaged plasticity was steel-concrete precast beam-column joints with concrete filler plates
implemented to explore the influence of joint loadings on the stress offered sufficient structural capability in transferring loadings from
concentration, compressive deformation, and bearing capacity of the beams to columns through a proposed connection. Numerical analysis
concrete filler plates to reliably and economically design hybrid com- and experimental investigation also demonstrated that concrete filler
posite concrete frames with concrete and metal plates. plates offer stiffness similar to that of metal filler plates, providing
flexural capacity similar to that of mechanical joints with steel filler
7. Conclusions plates.

Although extensive studies have investigated the behavior of bolted Acknowledgements

endplates, most of them explored the structural behavior of beam
endplates for steel frames. Beam endplates designed for steel-concrete This work was also supported by Basic Science Research Program
precast members have been notably absent. An advantage of the me- through the National Research Foundation of Korea (NRF) funded by
chanical joint introduced in this study is the application of a precast the Ministry of Education (2016R1D1A1A02937558).
concrete based moment-resisting joint with concrete filler plates to
provide assembly speed similar to that of a steel frame. In this study, References
parameters which were identified for the numerical investigation to
study the behavior of a precast beam-to-column frame connected by [1] Nzabonimpa JD, Hong WK, Kim J. Mechanical connections of the precast concrete
laminated plates consisting of mechanical metal and concrete filler columns with detachable metal plates. Struct Des Tall Spec Build 2017.
[2] AISC. AISC 358–05, Prequalified connections for special and intermediate steel
plates included dilation angle, concrete damage, eccentricity, clearance moment frames for seismic applications, American Institute of Steel Construction,
of the contact surfaces, and penalty stiffness. The nonlinear numerical

302
J.D. Nzabonimpa et al. Composite Structures 185 (2018) 286–303

Chicago, IL; 2005. mechanical beam-column joints of precast concrete-based frames. Comput Struct
[3] Mashaly Elsayed, El-Heweity Mohamed, Abou-Elfath Hamdy, Osman Mohamed. 2017;189:31–48.
Behavior of four-bolt extended end-plate connection subjected to lateral loading.
Alexandria Eng J 2011;50(1):79–90. Mr. J.D. Nzabonimpa is currently enrolled as a Ph.D. candidate in the Department of
[4] Bahaari Mohammed R, Sherbourne Archibald N. Behavior of eight-bolt large ca- Architectural Engineering at Kyung Hee University, Republic of Korea. His research in-
pacity endplate connections. Comput Struct 2000;77(3):315–25. terest includes precast composites structures.
[5] Díaz Concepción, Victoria Mariano, Martí Pascual, Querin Osvaldo M. FE model of
beam-to-column extended end-plate joints. J Constr Steel Res
2011;67(10):1578–90. Dr. Won-Kee Hong is a professor of Architectural Engineering at Kyung Hee University.
[6] Ju SH, Fan CY, Wu GH. Three-dimensional finite elements of steel bolted connec- Dr. Hong received his master’s and Ph.D. degrees from UCLA, and he worked for
tions. Eng Struct 2004;26(3):403–13. Englelkirk and Hart, Inc. (USA), Nihhon Sekkei (Japan) and Samsung Engineering and
[7] Mashaly Elsayed, El-Heweity Mohamed, Abou-Elfath Hamdy, Osman Mohamed. Construction Company (Korea) before joining Kyung Hee University (Korea). He also has
a professional engineering license from both Korea and the USA. Dr. Hong has more than
Finite element analysis of beam-to-column joints in steel frames under cyclic
loading. Alexandria Eng J 2011;50(1):91–104. 30 years of professional experience in structural engineering. His research interests in-
clude a new approach to construction technologies based on value engineering with
[8] Wang J, Chen L. Experimental investigation of extended end plate joints to con-
crete-filled steel tubular columns. J Constr Steel Res 2012;79:56–70. hybrid composite structures. He provided many useful solutions to issues in current
[9] Muresan, I.C. and Balc, R., Finite element analysis of an extended end-plate con- structural design and construction technologies as a result of his research that combines
nection using the T-stub approach. In AIP Conference Proceedings (Vol. 1648, No. 1, structural engineering with construction technologies. He is the author of numerous pa-
p. 850091). AIP Publishing; 2015. pers and patents both in Korea and the USA. Currently, Dr. Hong is developing new
connections that can be used with various types of frames including hybrid steel–concrete
[10] Bai R, Chan SL, Hao JP. Improved design of extended end-plate connection allowing
for prying effects. J Constr Steel Res 2015;113:13–27. precast composite frames (SMART frames), precast frames and steel frames. These con-
[11] Nzabonimpa JD, Hong WK, Park SC. Experimental investigation of dry mechanical nections would contribute to the modular construction of heavy plant structures and
beam-column joints for precast concrete based frames. Struct Des Tall Spec Build buildings as well.
2017;26(1).
[12] ABAQUS Analysis User's Manual 6.14-2. Dassault Systèmes Simulia Corp., Jisoon Kim received his Master degree from Department of Architectural Engineering at
Providence, RI, USA. Kyung Hee University, Republic of Korea. His research interest includes precast compo-
[13] Park R, Paulay T. Reinforced concrete structures. John Wiley &Sons, Inc; 1975. sites structures.
[14] Nzabonimpa JD, Hong WK, Kim J. Nonlinear finite element model for the novel

303