Investigations of seismic behavior of hybrid connections

Sudhakar Apparao Kulkarni and Bing Li

Precast concrete technology is popular due to its inherent advantages, such as ease of construction, cost effectiveness, and quality. However, the catastrophic failure of some precast concrete buildings during earthquakes has drawn attention to the importance of connection design. Due to the scarcity of land and urbanization, precast concrete construction has been used extensively in Singapore since the 1980s due to its minimal site disturbance and storage-area requirements. It has also gained popularity in countries such as the United States, Canada, Japan, Turkey, and Malaysia. Although many of these areas are low- to moderate-seismicity regions, it is essential that the precast concrete structures built within them are capable of resisting lateral loads. To investigate the behavior of hybrid (steel-concrete) connections for precast concrete frames, a series of experiments were conducted at Nanyang Technological University (NTU) in Singapore. In the first phase, pilot studies on the structural behavior of precast concrete beams and columns were completed. The second phase consisted of hybrid beam-column joints with different connection alternatives. The innovative connections used in these investigations included steel sections of different configurations for the beam-column and column-column fasteners. From the significant amount of the experimental data gathered and the subsequent finite-element (FE) method studies performed, design principles are being refined to verify the ability of these connections to resist the lateral loading. This paper focuses on the experimental and FE method investigations of hybrid steel-concrete beam-column joints subjected to seismic loading. Four prototype specimens of beam-column joints with slabs were tested under reversed cyclic loading to evaluate their strength, ductility, and energy-dissipation capacity. Two were cast-in-place concrete specimens, and two were precast concrete specimens constructed with the proposed hybrid connections. All of the precast concrete components were assembled, and the connections were encased with cast-in-place concrete.

Editors quick points

n Connection design is an important consideration for the successful application of precast concrete technology in terms of achieving constructible and stable structures. n This paper focuses on the experimental and finite-elementmethod investigations of hybrid steel-concrete beam-column joints subjected to seismic loading. n Test results showed that the major drawback of the hybrid connection was the discontinuity in the bottom reinforcement.

PCI Journal | Wi n t e r 2009

67

The specimens were tested under lateral cyclic loading, and the joint behavior was experimentally studied. By comparing the experimental observations with the numerical results, the FE models (FEMs) were validated. Using the FEMs, the complex behavior of the hybrid connections was further investigated by varying the main parameters influencing joint performance.

the findings of researchers since the 1970s. These include Stanton, Anderson, Dolan, and McCleary in the areas of moment-resistant connections and simple connections.5 The strength of members with dapped ends was introduced by Mattock and Theryo,6 while Shaikh and Yi investigated the behavior and strength of welded headed studs.7 These are important studies about precast concrete technology, and their results have been beneficial to the industry. In current practice, little attention has been given to the connection of component parts, whereas more effort has been devoted to the design of individual precast concrete elements. The connection details are not taken into consideration until the completion of the overall structural system. Connection design is, however, one of the most important considerations in the successful application of precast

Literature survey
The Structural Use of Concrete. Code of Practice for Design and Construction BS 81101 is the major design code used in Singapore. However, BS 8110 does not yet fully cover the code specifications related to the usage of precast concrete elements. Precast concrete construction technical references, such as PCI manuals and handbooks, have been used as supplementary references in addition to the local design guides.24 These manuals and handbooks consist of

963 mm

574 mm

574 mm

963 mm

963 mm

963 mm

1700 mm

600 1700 mm mm Specimen MJ1 A A B

1850 mm

300 mm

B 1850 mm Specimen MJ2 A A B Fillet weld 800 mm 1850 mm 300 1850 mm mm Specimen CJ2 600 mm B Inverted T-beam 317 mm x 167 mm x 20 mm

963 mm

574 mm

963 mm

Fillet weld

Inverted T-beam 317 mm x 167 mm x 20 mm

1700 mm

600 1700 mm mm Specimen CJ1

600 mm 300 mm

Main bars: 10M 22M

Links: 10M at 150 mm center-on-center

Section A-A (specimen MJ1/CJ1)

Main bars: ten 13M 300 Main bars: 1000 mm Transverse bars: mm 10M 22M 10M at 200 mm 127 mm Links: center-on-center 447 mm 10M at 150 mm Ties: 10M at 100 mm center-on-center center-on-center Main bars: three 22M 300 mm Section A-A Section B-B (specimen MJ2/CJ2)

Figure 1. These diagrams illustrate the typical sectional views of the test specimens. Note: 10M = no. 3; 13M = no. 4; 22M = no. 7. 1 mm = 0.0394 in. 68 W int e r 2 0 0 9 | PCI Journal

963 mm

800 mm

574 mm

963 mm

200 300
90 90
10-mm-thick plate

Connection plate G8.8 M22 hexagon bolts

10
10 G8.8 M22 hexagon bolts

318 18 165

55
B

90

55 Precast concrete beam

317 mm x 167 mm at 76 kg/m

Precast concrete column Beam-to-column connection details Section B-B showing plate connection details
178 482 178 UC 254 mm x 254 mm x 89 kg/m (cut) 15-mm-thick steel plate 482 15-mm-thick steel plate UC 254 mm x 254 mm x 89 kg/m (cut) Ten M22 hexagon bolts

Ten M22 hexagon bolts

Sectional plan showing connection details at joint core for specimen CJ1

Plan showing connection details at joint core for specimen CJ2

Figure 2. The sections and plan show the connection details of the precast concrete specimens. Note: All measurements are in millimeters. UC = universal column = 10 in. 10 in. and weighs 60 lb/ft. M22 = 7/8 in. 1 mm = 0.0394 in.; 1 m = 3.281 ft; 1 kg = 2.2 lb.

concrete technology in terms of achieving constructability and stability. The design and detailing of the connection influence the strength, stability, ductility, and load redistribution within the structure. However, the amount of testing information available on the seismic behavior of hybrid steel-concrete structures and the cyclic performance of connections in the inelastic range is limited, and this precludes the establishment of a prescriptive code or design guidelines for the seismic design of these structures.812 The catastrophic failure of some precast concrete buildings during earthquakes has drawn attention to the importance of connection design. A hybrid joint system consisting of steel beams and reinforced concrete columns was recently investigated to understand its seismic performance. This hybrid connection experimental study proved that the inelastic cyclic

response of hybrid reinforced concrete columnsteel beam joints maintained strength at large levels of story drifts with no significant loss in stiffness.13 In a different study, it was found that the amount of joint stirrups can be reduced or even eliminated if band plates or fiber-reinforced concrete is used in the joint.14 An experimental study on the welded connections consisting of steel beams and reinforced concrete columns encased with steel plates proved to be satisfactory in terms of strength and ductility under large inelastic deformation reversals, such as strong earthquakes.15 It may be noted that although a few research studies on hybrid beamcolumn connections with cast-in-place concrete specimens are available, literature related to hybrid precast concrete joints is scarce.

PCI Journal | Wi n t e r 2009

69

Description of experimental program

Details of test specimens The emphasis of the test was to serve as a pilot study to ascertain the feasibility of the proposed precast concrete connections, identify the possible drawbacks, and propose some new recommendations in the design. Four specimens were tested, two of which were cast-in-place concrete specimens, MJ1 and MJ2, and two of which were made from precast concrete components, CJ1 and CJ2. The specimens were replicas of the critical joint regions in moment-resisting frames that were designed and constructed according to BS 8110. A cast-in-place concrete slab was incorporated into all specimens to serve as the top flange to the beams. Figure 1 shows the details of all specimens. Specimen MJ1 had beams that were framed for flexure about the major axis of the column, while specimen MJ2 had beams that were framed for flexure about the minor axis of the column. The same dimensions and configurations were used for specimens CJ1 and CJ2. The rectanTable 1. Material test results

gular column simulated two different structural combinations: one was strong columnweak beam and another was weak columnstrong beam. Both were tested to evaluate how the connection details of the different systems influenced the strength of joints. Figure 2 shows the proposed simple precast concrete connection details. Two 10-mm-thick (0.4 in.) steel plates connected the precast concrete beam to the precast concrete column with the aid of six M22 ( 7/8 in.) bolts. The column contained a protruding T-section that was 254 mm 254 mm (10 in. 10 in.) and weighed 89 kg/m (60 lb/ft). The T-section projected 200 mm (4 in.) from the face of the column and was welded to the steel cage in joint area. An inverted T-beam section (317 mm 167mm 20 mm [12.5 in. 6.6 in. 0.8 in.] and weighing 76 kg/m [51.2 lb/ft]) was embedded in the beam by welding the flange to the bottom reinforcement of the beam. The purpose of the protruding section was to facilitate field construction.

Reinforcing bars Bar size 22M 13M 10M Yield strength, MPa 495 485 340 Concrete sections Specimen MJ1 MJ2 CJ1 CJ2 All Compressive strength, MPa 38.2 38.9 37.9 37.9 37.3 Steel sections Type T-section 317 mm 167 mm 20 mm at 76 kg/m 10-mm-thick plate Yield strength, MPa 330 327 Ultimate strength, MPa 473 524 Elongation, % 31 26 Tensile strength, MPa 3.3 3.4 3.4 3.4 3.3 Slabs Beams and columns Structure type Ultimate strength, MPa 580 550 400 Elongation, % 17 20 24

Note: CJ1 = a precast concrete specimen that had beams that were framed for flexure about the major axis of the column; CJ2 = a precast concrete specimen that had beams that were framed for flexure about the minor axis of the column; MJ1 = a cast-in-place concrete specimen that had beams that were framed for flexure about the major axis of the column; MJ2 = a cast-in-place concrete specimen that had beams that were framed for flexure about the minor axis of the column. 22M = no. 7; 13M = no. 4; 10M = no. 3. 1 mm = 0.0394 in.; 1 m = 3.281 ft; 1 kg = 2.2 lb; 1 MPa = 0.145 ksi.
70 W int e r 2 0 0 9 | PCI Journal

Specimen construction Specimens were designed using BS 8110 and BS 5950 codes of practice. During the design stage, concrete with a compressive strength of 40 MPa (5800 psi) and maximum aggregate size of 20 mm (0.79 in.) was targeted in all specimens.
16

Table 2. Grout properties Description Sand content, % Temperature, oC Expansion, % Mini-cone slump, mm Compressive strength, MPa Elastic modulus, MPa Poissons ratio Value 50 24.2 4 140 63.5 23,300 0.23

Table 1 reports the material test results for different specimens. The slump value of the concrete measured was 75 mm 25 mm (3 in. 1 in.). The reinforcing bars used for the specimens were 10M, 13M, and 22M (no. 3, no. 4, and no. 7), and had yield strengths of 340 MPa, 485 MPa, and 495 MPa (49.3 ksi, 70.3 ksi, and 71.8 ksi), respectively.

Note: 1 mm = 0.0394 in.; 1 MPa = 0.145 ksi; F = (C 1.8) + 32.

Precast concrete column Precast concrete column

Assembly of precast concrete beams and column

Specimen after casting slab

Figure 3. Typical precast concrete specimens are assembled in the sequence shown.

Transfer beam

Loading frame

3694 mm

Specimen

Load cell Support Hydraulic jack

Figure 4. The sketch indicates a typical test setup. Note: 1 mm = 0.0394 in. PCI Journal | Wi n t e r 2009 71

Load controlled 8 6 Displacement ductility factor 4 2 0 -2 -4 -6 -8 0 1 2 3 4 5 6 0.5Pi 0.75Pi

Displacement controlled

crete sections. The precast concrete column had 125 mm (5 in.) of discontinuity in the concrete section at the elevation of the top of the precast concrete beams, which resulted in a continuous cast-in-place concrete slab at the columns locations. Table 2 presents the properties of the grout. The cast-inplace and precast concrete specimens were cast vertically, simulating the on-site conditions. Figure 3 presents the assembly of a typical precast concrete specimen. Test setup Each unit was subjected to quasi-static load reversals that simulated earthquake loading. Figure 4 shows the test setup. A reversible horizontal load was applied to the tops of the columns using a double-acting, 89.9-kip-capacity (400 kN) hydraulic jack. The bottom of the column was pinned to a strong floor, and the beam ends were connected to this strong floor by steel links, which permitted rotation and free horizontal movement of the beam but not vertical movement, thus providing the vertical reactions to the beams. No axial load was applied to the specimens because zero axial compression is the most critical case for joint-core shear strength. Loading arrangement All test specimens were loaded under quasi-static simulated seismic loading (Fig. 5). The first two cycles were controlled by load, and the remainder of the tests were controlled by displacement. In all tests, two cycles of horizontal loading up to 0.5Pi and 0.75Pi were initially applied, where Pi is the horizontal load at the top of the column associated with the theoretical flexure strength Mi being reached at the critical sections of the members. The theoretical flexure strength Mi is calculated using the conventional compressive stress block for the concrete with an extreme fiber concrete compressive strain of 0.0035, the measured concrete compressive cylinder strength, and the steel yield strength from tests. The yield displacement y for all test specimens was calculated using the stiffness at the interstory horizontal displacement measured at 0.75Pi, extrapolated linearly to Pi (Fig. 6). The applied cyclic loading in the inelastic range was controlled by displacement. The test specimens were subjected to two cycles of loading to the displacement ductility factors (DFs) of 1, 2, 3, and 4, where DF is /y and is the interstory horizontal displacement of the test specimen. Instrumentation The specimens were extensively equipped with measuring devices to understand their behavior under cyclic loading. The lateral displacement of the column top was measured

10

11

12

Cycle number

Figure 5. The graph shows cyclic loading and displacement history applied to the test specimens. Note: Pi = ideal story horizontal load strength.

Horizontal load +Pi +0.75Pi +0.5Pi y1 y2 Horizontal displacement 0.5Pi 0.75Pi Pi y = (y1 + y2)/2

Figure 6. The diagram defines yield displacement. Note: Pi = ideal story horizontal load strength; = interstory horizontal displacement; y1 = maximum interstory horizontal displacement in a negative cycle; y2 = maximum interstory horizontal displacement in a positive cycle; y = column top horizontal displacement.

Steel components consisted of T-beam sections with dimensions of 317 mm 167 mm 20 mm (12.5 in. 6.6 in. 0.8 in.) and weighing 76 kg/m (51.2 lb/ft) and universal-column sections with dimensions of 254 mm 254 mm (10 in. 10 in.) and weighing 89 kg/m (60 lb/ ft), and 16-mm-thick and 10-mm-thick (0.6 in. and 0.4 in.) plates were used to fabricate the test specimens. Average values of the strengths of the steel sections were obtained from coupon tests. Table 1 summarizes the material test results of concrete, steel bars, and steel sections. Concrete and cement grout were used to fill the connection gaps between fabricated steel sections and the precast con-

72

W int e r 2 0 0 9 | PCI Journal

Story shear force, kip

Story shear force, kN

Story shear force, kN

100 0 -100 -200 -3.0 -2.0 -1.0 0.0 1.0 Ductility factor 2.0 3.0 Specimen MJ1

22.5 0.0

100 0 -100 -200 -3.0 -2.0 -1.0 0.0 1.0 Ductility factor 2.0 3.0 Specimen MJ2

22.5 0.0

-22.5 -45.0

-22.5 -45.0

Story shear force, kN

Story shear force, kip

200 100 0 -100 Specimen CJ1 -200 -3.0 -2.0 -1.0 0.0 1.0 Ductility factor 2.0 3.0

45.0 22.5 0.0

200 Story shear force, kN 100 0 -100 -200 -3.0 -2.0 -1.0 0.0 1.0 Ductility factor 2.0 3.0 Specimen CJ2

45.0 22.5 0.0 -22.5 -45.0 Story shear force, kip

-22.5 -45.0

Figure 7. These diagrams present hysteresis loops for each specimen.

using a displacement transducer, while a range of transducers were installed within the beam-column joint region to measure the flexural and shear deformations. Electrical strain gauges were installed in beam and column reinforcement to observe the deformations throughout the test.

tial pinching behavior in their hysteresis loops beginning from the fourth and third loading cycles, respectively. The possibility of a bolt connection failure was ruled out because there was no sudden or brittle failure. Yielding and local buckling of the inserted steel beam or steel plate were not expected because the excess flexibility in the connections caused low stress that was transferred from the columns to beams. However, at higher loading cyclescorresponding to the seventh cycle in specimen CJ1 and the eighth cycle in specimen CJ2the yielding of plates was observed. Figure 8 shows the cracking patterns of specimens at the end of the test. Figure 9 shows specimens MJ1 and CJ1 with cracks marked on them. Plastic hinges formed at the top portion of all beams where the reinforcement was found to have yielded. The discontinuity in the bottom reinforcement of the beams in precast concrete specimens caused their top reinforcement to be stressed more than their cast-in-place counterparts. During the loading stage, the bottom of the beam was alternately in tension and compression, due to the change in loading directions. In the tension region, the initial flexural cracks appeared and propagated rapidly with the increasing

Experimental results and discussion

As depicted in Fig. 7, both specimens CJ1 and CJ2 exhibited structural behavior similar to that of specimens MJ1 and MJ2 in the initial stage of loading. However, the specimen types behaved differently after cracking. The initial two loading cycles were elastic, and the remaining cycles exhibited inelastic behavior. Specimens MJ1 and MJ2 generally showed a stable energy-dissipating capacity with cracks all over the specimens. Significant pinching was observed in both of the specimens after the second loading cycle. These specimens behaved in a ductile manner. The levels of energy dissipation and strength achieved in specimens CJ1 and CJ2 were less than those of their castin-place concrete specimens regardless of the orientation of the columns. Specimens CJ1 and CJ2 exhibited a substan-

PCI Journal | Wi n t e r 2009

Story shear force, kip

200

45.0

200

45.0

73

MJ1

MJ2

CJ1

CJ2

Figure 8. The test specimens developed cracking patterns as shown.

load, forming vertical cracks between the beam and connection region. This might be due to the lack of aggregate interlocking. With the absence of continuity in the bottom reinforcement, the vertical cracks propagated more rapidly, causing the beams to lose their ability to resist the moment. Figure 10 shows the decomposition percentages of the total lateral displacement measured at the top of the column at the peak of each load cycle. Lateral displacement contributed by beam flexural and beam fixed-end rotation was dominant in specimens with a strong columnweak beam combination, such as specimens MJ1 and CJ1. Alternatively, column flexural and column fixed-end rotation significantly affected the total lateral displacement of specimen MJ2. This confirmed that the flexural strength of specimen MJ2 column was inferior to that of its beams. However, this situation was not found in specimen CJ2 even though its column strength was similar to that of specimen MJ2. The excess flexibility in the beam-column connections caused a major percentage of the lateral displacement to be contributed mainly through the beam flexural resistances. This accounted for the dominant lateral displacements in the composite specimens CJ1 and CJ2. The contribution of lateral displacement from shear distortion was not significant in all specimens because the joint-core region was strong and rigid.

Summary of test observations

All specimens generally showed satisfactory results in their strength, ductility, and energy-dissipation capacity. Due to the discontinuous bottom reinforcement in the precast concrete specimens, the strength development was limited and only matched with those of their cast-in-place concrete counterparts. The top reinforcement of the precast concrete specimens was stressed to a greater level compared with the cast-inplace concrete specimens. The discontinuity in the precast concrete specimens bottom reinforcement might have caused an incomplete stress transfer, and hence, more stress was sustained by the top reinforcement. This might be due to the yielding of some of the bottom bolts or slippage of fasteners to some degree during cyclic loading. In the precast concrete specimens, the high flexibility in the beam-column joint caused by the discontinuity on bottom reinforcement was observed by the excess rotation in the beams. With the same level of load, less rotation was observed in the cast-in-place concrete specimens. The precast concrete specimens, however, achieved consistent hysteresis loops throughout the test and behaved similarly to the cast-in-place concrete specimens in energy dissipation.

74

W int e r 2 0 0 9 | PCI Journal

Specimen MJ1

Specimen CJ1
Figure 9. The test specimens show their cracking patterns. PCI Journal | Wi n t e r 2009 75

100 80 60 40 20 0 -20 -40 -60 -80 -100

DF 075

DF 1.5

DF 05

DF 1.0

DF 2.0 Column fixed-end Shear deformation

100 80 60 40 20 0 -20 -40 -60 -80 -100

DF 075

DF 1.5

Displacements, %

Displacements, %

DF 05

DF 1.0

DF 2.0 Column fixed-end Shear deformation

Beam flexure Beam fixed-end Column flexure

Beam flexure Beam fixed-end Column flexure

Specimen MJ1
DF 075 DF 1.5

Specimen MJ2
DF 075 DF 1.5

100 80 60 40 20 0 -20 -40 -60 -80 -100

DF 05

DF 1.0

DF 2.0 Column fixed-end Shear deformation

100 80 60 40 20 0 -20 -40 -60 -80 -100

Displacements, %

Displacements, %

DF 05

DF 1.0

DF 2.0 Column fixed-end Shear deformation

Beam flexure Beam fixed-end Column flexure

Beam flexure Beam fixed-end Column flexure

Specimen CJ1
Figure 10. The graphs indicate the decomposition of lateral displacement of the specimens.

Specimen CJ2

The decomposition of lateral displacement revealed the dominance of beam flexural and beam fixed-end rotation contributions in specimens MJ1 and CJ1, which had strong columnweak beam combinations. Although column flexural and column fixed-end rotation significantly affected the total lateral displacement of specimen MJ2, the same observation was not found in specimen CJ2, even though its column had strength similar to specimen MJ2. The discontinuity of reinforcement in the bottoms of beams of the precast concrete connections is the critical part that requires improvement. Due to the unique nature of the test specimens, the effects of several critical variables influenc-

ing the joint performance (such as the connection-plate thickness and the absence of axial load) were not clearly understood. To provide more information about the behavior of hybrid connections, the following sections explain the development of two-dimensional (2-D) FEMs and detailed parametric studies.

Finite-element analysis
Mesh Figure 11 shows the FEMs of the specimens in their deformed shapes. The concrete was modeled with four-node

76

W int e r 2 0 0 9 | PCI Journal

Specimen MJ1

Specimen MJ2

Specimen CJ1

Specimen CJ2

Figure 11. The diagrams show the finite-element models of the specimens in deformed shapes.

isoparametric elements, while two-node truss elements were used for modeling reinforcing bars. The connections investigated consisted of elements with different materials and configurations. Steel plates embedded inside the beams and connection region were simulated as 2-D elements for the web region. The steel-plate elements modeled were assigned material properties of steel. For specimens CJ1 and CJ2, the main reinforcement was discontinued at the bottom. The concrete filled in the beam-column connection region after fastening the plates was neglected in the analysis. Material modeling In this analysis, a constant stress cutoff criterion was applied for cracking. In this approach, once the maximum principal tensile stress reaches the tensile strength independent of other principal stresses, a crack is initiated perpendicular to the principal stress. The orientation of the crack is then stored, and the material response perpendicular to the crack is determined by a stress-strain relationship for the cracked material volume. Additional cracks may appear at the same location, but their formation to the existing cracks is greater than 15 deg. However, if the angle is less than that, the secondary cracks are assumed not to exist even when the tensile stress

reaches the fracture envelope.17 The fracture energy GF and the concrete tensile strength ft were used to calculate the ultimate crack opening value wu. To simulate the softening effect of the concrete in tension after cracking, a bilinear tension stress-strain curve was used in which the ultimate strain of the concrete at crackcr ing u was taken as 0.001. The value is based on the assumption that the strain softening after failure reduces the stress linearly to zero at a total strain of about 10 times the strain at failure of concrete in tension, which is typically 0.0001. The uniaxial tensile strength of concrete used in the analysis was determined according to CEB-FIP Model Code 1900: Design Code18 in Eq. (1).
ft = 0.30 f c'

( )

2/3

(1)

where
f c'

= concrete compressive strength

The response of the concrete in compression is taken into account by an elastic-plastic model. The elastic state of stress was limited by a Drucker-Prager yield surface. Isotropic hardening with an associated flow rule was used after yielding of the surface had occurred.

PCI Journal | Wi n t e r 2009

77

Story shear force, kN

Story shear force, kip

Story shear force, kN

100 0 -100 -200 -3.0 -2.0 Experimental FEM -1.0 0.0 1.0 Ductility factor 2.0 3.0

22.5 0.0 -22.5 -45.0

100 0 -100 -200 -3.0 -2.0 Experimental FEM -1.0 0.0 1.0 Ductility factor 2.0

22.5 0.0

-22.5 -45.0 3.0

Story shear force, kN

Story shear force, kip

Story shear force, kN

100 0 -100 -200 -3.0 -2.0 Experimental FEM -1.0 0.0 1.0 Ductility factor 2.0 3.0

22.5 0.0 -22.5 -45.0

100 0 -100 -200 -3.0 -2.0 Experimental FEM -1.0 0.0 1.0 Ductility factor 2.0 3.0

22.5 0.0

-22.5 -45.0

Figure 12. The graphs indicate the comparison of story shear force versus ductility factors of the specimens. Note: FEM = finite-element model.

The DIANA software evaluates the yield surface using the current state of stress, the angle of internal friction , and the cohesion c. Per the recommendations of the DIANA software manual, the angle of internal friction in concrete can be approximated to be 30 deg. The cohesion c used in the analysis is given by Eq. (2).
p ct = fc' uniaxial

The elasto-plastic material law was used for the reinforced bars. These are modeled with the DIANA options of separate truss elements. In this case, the reinforcement elements had separate degrees of freedom. The strength and stiffness of the concrete elements were increased in the direction of the embedded reinforcement. The option assumed perfect bond between the reinforcement and the surrounding concrete. At key locations where bond slip was likely to occur, the bars were also modeled with interface elements, appropriately accounting for the interaction between the reinforcement and the concrete. Steel plates were modeled as 2-D plane stress elements, and the material properties of the steel were assigned to them. The von Mises yield criterion with isotropic strain hardening and an associated flow rule was used to describe the constitutive behavior of steel plates.

) 1 sin 2cos

(2)

where
p uniaxial

= plastic strain in the direction of the uniaxial compression stress

p f c' uniaxial = hardening or softening parameter as a function of the plastic strain in the direction of the uniaxial compression stress

Standard uniaxial tests on concrete cylinders were used to define the stress-strain relations up to the peak stress. CEBFIP model code recommendations were used to evaluate the post-peak behavior of the concrete using cylinder compression strength tests. A Poissons ratio of 0.15 was used in the analysis.

Verification of FEM
To verify the FEM, the numerical results were compared with the experimental observations. Figure 12 shows that for specimen MJ1, the FEM seems to have predicted a good response with respect to the experimental observations. The specimen achieved a displacement ductility factor of about

78

W int e r 2 0 0 9 | PCI Journal

Story shear force, kip

200

45.0

200

45.0

Story shear force, kip

200

45.0

200

45.0

2.0, and pinching was observed in the loops. The FEM loops were thin and similar to the experimental results. Story shear forces achieved in the FEM loops were somewhat less than their experimental results. The specimens global deformation of the joint corresponding to a DF of 2.0 is given in Fig. 11. As shown by the figure, the major part of the total displacement observed at the column top was contributed by the deformation of the columns. Figure 12 also presents a comparison of the FEM and experimental results for specimen MJ2. The shapes of a few initial loops of the FEM results showed good agreement with the experimental observations. Both the FEM and the experimental loops were thin, depicting a lower level of energy dissipation compared with that of specimen MJ1. Although experimental loops showed pinching after a DF of 0.5, the pinching observed in the FEM loops was at a DF of 1.0. An extensive joint-core deformation and a large horizontal displacement of the joint frame can also be seen in Fig. 11. Both the top and bottom parts of the column showed excessive deformations. Figure 12 shows the comparison of the hysteresis loops for the FEM and experimental results of specimen CJ1. The energy dissipation observed both in the FEM and experimental hysteresis loops was good. The area confined within the FEM hysteresis loops was large, and the loops

agreed fairly well with the respective experimental results. Pinching was clearly noticed in the FEM loops at a DF of about 1.5 and greater. The highest-story shear force carried in the experimental and FEM results also matched well for the loadings in the positive direction. Figure 11 presents the deformed shape of the specimen. An excessive deformation of the joint core and the surrounding part can be seen from the figure. A large rotation of the column due to a narrow neck-type joint connection can also be observed. Figure 12 shows the comparison of the FEM and experimental hysteresis loop results of specimen CJ2. Although initial loops had a good agreement with the experimental values, the last few loops predicted the story shears somewhat lower than their experimental counterparts. Similar to the experimental observations, the FEM loops also had large displacements. Observed energy dissipation was slightly less when compared with that of specimen CJ1. Pinching in the hysteresis loops of both the experimental and FEM results was observed. Initial FEM loops showed a greater value of story shear compared with experimental results. Figure 11 shows the deformed shape of the joint core for a DF of 2.0. A large deformation of joint core can be seen from the figure, which agreed with the experimental observations. The specimen experienced large horizontal displacements due to its greater flexibility.

200 Story shear force, kN 100 0 -100 -200 -3.0 -2.0 -1.0 0.0 Without axial load Axial load = 0.1f'c Ag Axial load = 0.2f'c Ag Axial load = 0.3f'c Ag Axial load = 0.4f'c Ag 1.0 2.0 3.0

45.0 22.5 0.0 -22.5 -45.5 Story shear force, kip

Ductility factor
Figure 13. Story shears are compared under different axial load levels for specimen CJ1. Note: Ag = gross area of a member; = compressive strength of concrete. 79

PCI Journal | Wi n t e r 2009

Comparison of the FEM and experimental results of all specimens showed that the lateral-load-displacement hysteresis loops obtained from the FEM were similar to their experimental results. The energy-dissipating capacities at different loading cycles and the ultimate ductility capacities also correlated well with their experimental counterparts. From the observations and predictions, the application of FEM technique can, therefore, be further extended to study joint performance by varying parameters.

In this study, the influence of axial loading on the seismic behavior of hybrid joints was investigated. Different axial loading levels in addition to the same loading histories as those used in the analysis of specimens without axial loading were applied, and the story shear force versus horizontal displacement loops were studied. Figure 13 shows the comparison of the ductility factors versus the story shears for specimen CJ1. As the axial' ' load ratios were increased to 0.1 f c Ag and 0.2 f c Ag, the specimen showed an increase in the story shears by about 8% and 11%, respectively. A further increase in ' axial load ratio to a value of 0.3 f c Ag witnessed only a marginal increase in the story shear. However, a reduction in story shear occurred as the axial load ratio was ' enhanced beyond 0.4 f c Ag. There was a decrease in the ultimate number of cycles with the increase in column axial loads. Figure 14 shows specimen CJ2, which also showed a similar increase in the story shears with the application of column axial load. Story shear was enhanced by about 7% ' ' and 8% for the axial load ratios of 0.1 f c Ag and 0.2 f c Ag, respectively. It may be explained that within certain limits of column axial load, there would be reduction in the resultant stresses in the column due to the combined bending and axial loads.

Parametric studies
To understand the response of certain key factors affecting joint behavior, the FE method was used to conduct certain key parametric studies. The following sections present the influence of the column axial load, plate thickness, and beams bottom reinforcement continuity on the performance of the joint core. Column axial load effect Axial loading is a critical parameter in the study of beamcolumn joints, but its effect on seismic behavior of beamcolumn joints has not been fully understood. Previous investigations have shown that axial force is beneficial to joint shear resistance.19

200 Story shear force, kN 100 0 -100 -200 -3.0 -2.0 -1.0 0.0 Without axial load Axial load = 0.1f'c Ag Axial load = 0.2f'c Ag Axial load = 0.3f'c Ag Axial load = 0.4f'c Ag 1.0 2.0 3.0

45.0 22.5 0.0 -22.5 -45.5 Story shear force, kip

Ductility factor
Figure 14. Story shears are compared under different axial load levels for specimen CJ2. Note: Ag = gross area of a member; 80 W int e r 2 0 0 9 | PCI Journal = compressive strength of concrete.

200 Story shear force, kN 100 0 -100 -200 -3.0 -2.0 -1.0 0.0

45.0 22.5 0.0 -22.5 -45.5 Story shear force, kip

Plate thickness = 10 mm Plate thickness = 12 mm Plate thickness = 14 mm Plate thickness = 16 mm 1.0 2.0 3.0

Ductility factor
Figure 15. The graph shows the comparison of story shears versus ductility factors for specimen CJ1. Note: 1 mm = 0.0394 in.

The decrease in critical stresses at column edges helped in substantially lessening the flexural cracking of columns and moderately reducing the joint deformation. These effects contributed to the enhancement of story shears of the joint under the influence of column axial loads. Further increase in axial load ratios, however, reduced the story shears due to comparatively excess values of bending stresses. The ultimate load cycles also decreased when the ' axial-load ratio was increased beyond 0.3 f c Ag. Therefore, ' ' the axial-load ratios from 0 to 0.2 f c Ag and 0 to 0.1 f c Ag were beneficial to the joint behavior of strong column weak beam and weak columnstrong beam systems, respectively. Increases in axial load ratio beyond these values proved to be detrimental to the joints performance. The reason for a lower axial load ratio in the case of the weak column strong beam system was due to small column width. Even with lower column-load eccentricity values, the bending stresses developed were substantially higher. Effect of connection-plate thickness Similar to the experimental observations, the FEMs also showed the yielding of connection plates in the bottoms of beams, corresponding to the sixth and the seventh cycles, respectively, in specimens CJ1 and CJ2. In this study, the connection-plate thicknesses of specimens CJ1 and CJ2

were varied, and the possible influence on joint behavior was investigated. Figures 15 and 16 show the ductility factor versus the story shear forces of the hybrid specimens under different plate thicknesses. Figure 15 shows that the story shear in specimen CJ1 increased about 5% as the thickness of connection plate was increased from 10 mm (0.4 in.) to 12 mm (0.5 in.). An improvement in energy dissipation was also observed from the analysis. Further increase in story shear of about 9% and 12% was noticed with 14-mm-thick and 16-mm-thick (0.55 in. and 0.63 in.) plates, respectively. The specimen also showed higher energy dissipation with enhanced plate thickness. When the plate thickness was increased beyond 16 mm, no appreciable change in energy dissipation was observed, though some initial cycles showed a higher value of story shears. Although the concrete in the connection region was neglected in the analysis, the energy dissipation obtained was comparable to the experimental studies. The reason for this may be that the concrete filled later was not effectively bonded to the plates. Specimen CJ2 also showed good energy dissipation as observed from the loops in Fig. 16. Higher-story shears corresponding to increasing plate thicknesses were observed. As the plate thickness was increased from 10 mm to 12

PCI Journal | Wi n t e r 2009

81

200 Story shear force, kN 100 0 -100 -200 -3.0 -2.0 -1.0 0.0

45.0 22.5 0.0 -22.5 -45.5 Story shear force, kip

(3)

Plate thickness = 10 mm Plate thickness = 12 mm Plate thickness = 14 mm Plate thickness = 16 mm 1.0 2.0 3.0

Ductility factor
Figure 16. The graph shows the comparison of story shears versus ductility factors for specimen CJ2. Note: 1 mm = 0.0394 in.

mm (0.4 in. to 0.5 in.), an increase of 4% in story shear was obtained at a DF of 2.6. A further increase of 8% and 10% in story shears occurred for plate thicknesses of 14 mm (0.55 in.) and 16 mm (0.63 in.), respectively. Besides increased energy dissipation, a lower level in the maximum principal compressive stresses was noticeable in the connection region of plates. With the increase in plate thicknesses, the story shears carried and the energy dissipations of the joint were increased. A lower stress state was generated in the specimens with thicker connection plates. Effect of beam-bottom reinforcement continuity In the experimental studies, to facilitate proper connections of precast concrete members at beam-column joints, the beams bottom reinforcement was discontinued in the connection region (Fig. 1). The precast concrete beams were connected to the column using steel plates and necessary fixtures. Because of reinforcement discontinuity, the strength and ductility of the joint were substantially affected. In this study, an effort was made to investigate the influence of the beam-bottom reinforcement continuity. The continuity was assumed to be maintained by extending

the truss elements, inside the column, that were previously terminated at the beam faces. In addition, relatively thinner-sized plate elements (about half of the normal plate thickness) were added in the bottom of the beams adjacent to the joint region to maintain continuity. The FE analysis was performed by varying the percentage of beam-bottom reinforcement in the connection region. Figure 17 shows the ductility factor versus the story shear forces for specimen CJ1. By varying the reinforcement from 0.5% to 1.0% of the gross area of the specimen Ag, its effect on the joint performance in terms of story shears was studied. The reinforcement ratios considered were greater than the minimum longitudinal reinforcement for beams specified by NZS 310120 in Eq. (3).

min =
where

4 fy

f c'

min = minimum reinforcement ratio

fy = yield strength of steel reinforcement

The specimen showed increases of about 11% and 14% in story shears as the reinforcement was increased by 0.5%Ag

82

W int e r 2 0 0 9 | PCI Journal

and 0.75%Ag, respectively. The energy dissipation also exhibited a drastic improvement. A further increase in reinforcement to 1.0Ag negligibly enhanced the story shear by about 1.5%. Figure 18 explains the ductility factor versus story-shear forces for specimen CJ2. Increases in story shears of about 10% and 13% were noticed when the reinforcement ratio was enhanced to 0.5%Ag and 0.75%Ag, respectively, accompanied by obvious improvement in the energy dissipation. With further increase in reinforcement to 1.0%Ag, there was no substantial increase in the story shears. The FE analysis results also showed that the stress distribution in the connection region observed was of a lower level in magnitude and smooth in variation. The continuation of beam-bottom reinforcement up to a value of 0.75%Ag substantially improved the strength and ductility of the specimens. A lower stress level with smooth distribution was clearly seen in the connection region. However, reinforcement enhancement beyond 0.75%Ag did not have a significant influence on the joints performance.

1.

Design the beam and column elements following the standard code provisions. In this case, BS 8110 was used. Resolve the bending moment in the beam section into equal and opposite forces, forming a couple separated by the corresponding distance at the centroid of top and bottom reinforcement. Following BS 5950, calculate the area of steel required for the flange and choose a suitable steel section providing an area about 20% in excess. Select the web size satisfying the condition that shear capacity Pv of the web section should be greater than the shear force at the section Fv. According to BS 5950, Pv is calculated as Pv = 0.6pyAv where py = design strength of a steel section Av = shear area of a member

2.

3.

Simplified analysis procedure

4. In order to get a clear understanding of the hybrid steel connections with precast concrete members, the following steps are recommended for their analysis:

Use values of the bending moment and the shear force at the face of the column to design the bolted connection and to decide the size of bolts and their pitch and the size of the cover plates.

200 Story shear force, kN 100 0 -100 -200 -3.0 -2.0 -1.0

45.0 22.5 0.0 -22.5 -45.5 Story shear force, kip

Without reinforcement Reinforcement area = 0.5% Ag Reinforcement area = 0.75% Ag Reinforcement area = 1.0% Ag

0.0

1.0

2.0

3.0

Ductility factor
Figure 17. The graph indicates the effect of beam bottom reinforcement continuity for specimen CJ1. Note: Ag = gross area of a member. PCI Journal | Wi n t e r 2009 83

200 Story shear force, kN 100 0 -100 -200 -3.0 -2.0 -1.0

45.0 22.5 0.0 -22.5 -45.5 Story shear force, kip

Without reinforcement Reinforcement area = 0.5% Ag Reinforcement area = 0.75% Ag Reinforcement area = 1.0% Ag 0.0 1.0 2.0 3.0

Ductility factor
Figure 18. The graph indicates the effect of beam bottom reinforcement continuity for specimen CJ2. Note: Ag = gross area of a member.

5.

Size the steel plates, which are in the joint-core region, following the criteria of axial load-carrying capacity, lateral buckling, and connection ease. A minimum of 10 mm (0.39 in.) plate thickness is preferred. Use a 2-D FE technique, as described in the FiniteElement Analysis section, to verify the designed model for its performance. Assign proper material laws and properties and boundary conditions to conduct the analysis of the FEM. Using a nonlinear procedure, the FEM is analyzed applying reversible cyclic loading. The performance of the connection is studied in terms of the story shear versus the displacement hysteresis loops, the energydissipating capacity, and pinching behavior. Verify the stress distribution around the joint region to know any abrupt change in values and/or yielding of materials. Repeat steps 2 through 7 by modifying the size of the connection plate if the analysis results are not satisfactory.

6.

tion was the discontinuity in the bottom reinforcement, which reduced the flexural strength of the specimens. The nonlinear FE analysis successfully modeled the hybrid steel-concrete beam-column connections. The FE analysis results were compared with the experimental results and good agreement between the two was achieved. Based on the experimental observations and FE analysis results of this study, several conclusions were drawn: Precast concrete specimens can achieve consistent hysteresis loops throughout cyclic loading and behave well compared with cast-in-place concrete specimens. The top reinforcement of precast concrete specimens is stressed to a higher level than cast-in-place concrete top reinforcement during seismic loading, but with an increase in the plate thickness of the hybrid connection, a lower state of stress level can be achieved in the precast concrete connection. The high flexibility in the beam-column joint of the precast concrete specimens caused by the discontinuity of the bottom reinforcement causes higher rotation of the beams compared with the cast-in-place concrete specimens when subjected to seismic loading. With the increase in connection-plate thickness, this high flexibility in the precast concrete specimen is easily corrected.

7.

8.

Conclusion
All tested specimens showed satisfactory results in strength, ductility, and energy-dissipation capacity. Test results showed that the major drawback of the connec-

84

W int e r 2 0 0 9 | PCI Journal

For hybrid connections with a strong columnweak ' beam system, an axial load between 0 and 0.2 f c Ag enhances performance. For connections with a weak columnstrong beam system, the axial load marginally ' helps with an axial load ranging from 0 to 0.1 f c Ag. Any increase in axial load value beyond these limits may be detrimental to the performance of the joints. With the increase in plate thicknesses, a hybrid connection is capable of carrying the required story shears and the energy dissipations of the joint increases. A lower stress state is generated in precast concrete specimens with thicker connection plates. The continuation of beam-bottom reinforcement up to a value of 0.75%Ag substantially improves the strength and ductility of the precast concrete specimens. A lower stress level with smooth distribution was clearly seen in the connection region. However, reinforcement enhancement beyond 0.75%Ag did not have a significant influence on the joints performance.

6.

Mattock, A. H., and T. S. Theryo. 1986. Strength of Members with Dapped Ends. PCI Journal, V. 31, No. 5 (SeptemberOctober): pp. 5875. Shaikh, A. F., and W. Yi. 1985. In-Place Strength of Welded Headed Studs. PCI Journal, V. 30, No. 2 (MarchApril): pp. 5681. Applied Technology Council (ATC). 1981. ATC-8 Proceedings of a Workshop on Design of Prefabricated Concrete Buildings for Earthquake Loads. Berkeley, CA: ATC. Dolan, C., J. Stanton, and R. Anderson. 1987. Moment Resistant Connections and Simple Connections. PCI Journal, V. 32, No. 2 (MarchApril): pp. 6274.

7.

8.

9.

10. Englekirk, R. 1987. Concepts for the Development of Earthquake Resistant Ductile Frames of Precast Concrete. PCI Journal, V. 32, No. 1 (JanuaryFebruary): pp. 3048. 11. Englekirk, R. Overview of ATC Seminar of Prefabricated Concrete Buildings for Earthquake Loads. PCI Journal, V. 27, No. 1 (JanuaryFebruary): pp. 8097. 12. Hawkins, N., and R. Englekirk. 1987. U.S.Japan Seminar on Precast Concrete Construction in Seismic Zones. PCI Journal, V. 32, No. 2 (MarchApril): pp. 7585. 13. Adajar, J. C., T. Kanakubo, M. Nonogami, N. Kayashima, Y. Sonobe, and M. Fusisawa. 2004. An Analysis of the Behavior of Hybrid Steel Beam-RC Column Connection. Paper no. 2398 presented at the 13th World Conference on Earthquake Engineering, Vancouver, B.C., Canada. 14. Parra-Montesinos, G. J., and J. W. Wight. 2000. Seismic Response of Exterior RC Column-to-Beam Steel Beam Connections. Journal of Structural Engineering, V. 126, No. 10 (October): pp. 11131121. 15. Fargier-Gabaldon, L. B., and G. J. Parra-Montesinos. 2006. Behavior of Reinforced Concrete Column-Steel Beam Roof Level T-Connections under Displacement Reversals. Journal of Structural Engineering, V. 132, No. 7 (July): pp. 10411051. 16. BSI. 2000. Structural Use of Steelwork in Building. Code of Practice for Design. Rolled and Welded Sections. BS 5950. London, UK: BSI. 17. Hajime, O., and M. Kohichi. 1991. Nonlinear Analysis and Constitutive Models of Reinforced Concrete. Tokyo, Japan: Gihodo Shuppan Co. Ltd.

Acknowledgments
The experimental work was performed at Nanyang Technological University in Singapore. Support by the Building and Construction Authority of Singapore is gratefully acknowledged. Any opinions, findings, and conclusions expressed in this paper are those of the writers and do not necessarily reflect the views of Building and Construction Authority of Singapore. The authors extend thanks to the reviewers and PCI editors for the constructive suggestions to improve the paper.

References
1. Committee B/525/2. 1997. Structural Use of Concrete. Code of Practice for Design and Construction. BS 8110. London, UK: British Standards Institution (BSI). PCI Connection Details Committee. 1973. Design and Typical Details of Connections for Precast Prestressed Concrete. 1st ed. Chicago, IL: PCI. PCI Industry Handbook Committee. 1971. PCI Design Handbook: Precast and Prestressed Concrete. 1st ed. Chicago, IL: PCI. PCI Industry Handbook Committee. 1985. PCI Design Handbook: Precast and Prestressed Concrete. 3rd ed. Chicago, IL: PCI. Stanton, J. F., R. G. Anderson, C. W. Dolan, and D. E. McCleary. 1986. Moment Resistant Connections and Simple Connections. PCI Specially Funded Research and Development Program project no. 1/4. Chicago, IL: PCI.

2.

3.

4.

5.

PCI Journal | Wi n t e r 2009

85

18. Comite Euro-International du Beton (CEB). 1993. CEB-FIP Model Code 1990: Design Code. Lausanne, Switzerland: Thomas Telford. 19. Paulay, T. 1989. Equilibrium Criteria for Reinforced Concrete Beam-Column Joints. ACI Structural Journal, V. 86, No. 6 (NovemberDecember): pp. 635643. 20. New Zealand Standards. 1995. Concrete Structures StandardThe Design of Concrete Structures. 3101. New Zealand: New Zealand Standards.

= angle of internal friction

Notation
Ag Av c
f c'

= gross area of a member = shear area of a member = cohesion = compressive strength of concrete = tensile strength of concrete = yield strength of steel = shear force at a section = fracture energy of concrete = theoretical flexural strength (moment) = ideal story horizontal load strength = shear capacity of a web section = design strength of a steel section = ultimate crack opening = interstory horizontal displacement = column top horizontal displacement = maximum interstory horizontal displacement in a negative cycle = maximum interstory horizontal displacement in a positive cycle = ultimate strain in concrete

ft fy Fv GF Mi Pi Pv Py wu y y1 y2
cr u

p uniaxial = plastic strain in uniaxial stress direction

min

= minimum reinforcement ratio

86

W int e r 2 0 0 9 | PCI Journal

About the authors

Sudhakar A. Kulkarni, PhD, is a senior structural engineer at Maunsell Consultants (Singapore) Pte. Ltd. in Singapore.

Bing Li, PhD, is an associate professor for the School of Civil and Environmental Engineering at Nanyang Technological University in Singapore.

FE method were performed to provide a better understanding of the joints behavior. The FE models developed were calibrated with the experimental results by verifying the load displacement behavior, global deformation shapes, and stress patterns in detail. The FE method results showed that global behavior of the joints could be simulated to correlate with the experimental observations. In addition, the influence of key parameterssuch as the connection-plate thickness, axial load, and beam bottom reinforcement continuityaffecting the joints performance were explored with the developed models.

Keywords
Ductility factor, FEM, finite-element analysis, hybrid connection, seismic, story shear.

Synopsis
This paper presents experimental and finite-element (FE) method investigations on hybrid (steel-concrete) beam-to-column connections conducted under reversed cyclic loading. Four beam-column joints with slab were tested to explore their viability for quasistatic seismic loading. Experimental findings of two specimens with the proposed hybrid connections and their cast-in-place connection (monolithic) counterparts are discussed. Because of the inherent complexity in hybrid connections and unique features of the test specimens, however, these experimental findings were insufficient. Therefore, numerical investigations using the

Review policy
This paper was reviewed in accordance with the Precast/Prestressed Concrete Institutes peer-review process.

Reader comments
Please address any reader comments to PCI Journal editor-in-chief Emily Lorenz at elorenz@pci.org or Precast/Prestressed Concrete Institute, c/o PCI Journal, 209 W. Jackson Blvd., Suite 500, Chicago, IL 60606. J

PCI Journal | Wi n t e r 2009

87