Static and seismic experiment for bolted-welded joint in modularized prefabricated steel structure

Journal of Constructional Steel Research 115 (2015) 417–433

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Static and seismic experiment for bolted-welded joint in modularized prefabricated steel structure X.C. Liu ⁎, S.H. Pu, A.L. Zhang, A.X. Xu, Z. Ni, Y. Sun, L. Ma Beijing Engineering Research Center of High-rise and Large-span Prestressed Steel Structures, Beijing University of Technology, Beijing 100024, China

a r t i c l e

i n f o

Article history: Received 5 June 2015 Received in revised form 23 July 2015 Accepted 27 August 2015 Available online xxxx Keywords: Prefabricated steel structure Bolted-welded truss–column joint Static performance Seismic capacity Experimental study Simplified calculation

a b s t r a c t This study proposes a type of bolted-welded beam–column joint for modularized prefabricated multi-rise and high-rise steel structures. The components in the same module are welded in the factory, and the modules are quickly assembled using the proposed joint at site. The static performance, hysteretic performance, skeleton curves, ductile performance, energy dissipation capacity, rotation capacity and stiffness degradation patterns of four joints are obtained by model experiment and finite element analyses, and the effect of thicknesses of the chords and web members on the static and seismic performance of the joint as well as the effect of welding quality are investigated. The results show that due to the presence of the bolted connecting parts, the proposed joints maintain relatively good seismic performance including ductile performance, energy dissipation capacity and plastic rotation capacity, and good static bearing capacity after the welding seams fracture, so they can be used in structures of seismic zones. Reducing the thicknesses of the chord and the web members can significantly decrease the load-bearing capacity of the joint; however, this decrease is not proportional to the decrease of the cross-sectional area. In addition, reducing the thicknesses of the chord and web members has no significant impact on the ductile performance and energy dissipation capacity of the joint. Simplified computation formulas for load-bearing capacity of the joint were proposed and the computation results get along well with the experimental results. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction Modularized prefabricated steel structures are consistent with the definition of green buildings that is given in the Assessment Standard for Green Building [1], and have an extensive application prospects. The standardized and modularized design and fabrication of members of modularized prefabricated steel structures are realized [2–3]. Steel structures have excellent machinability and are suitable for industrial production; they are lightweight and suitable for transportation, and are suitable to being connected by high-strength bolts, which makes them mostly suitable for prefabricated buildings [4–8]. Beam–column connecting is the core technology of prefabricated steel structures. When designing beam–column joints, it is necessary to fully consider the assemblability on site as well as to ensure the mechanical properties including strength, stiffness and ductility. This study proposes a new type of bolted-welded joint for modularized prefabricated multi-rise and high-rise steel structures. The proposed joint is used to connect two close modules and can facilitate the rapid assembly of modules on site. As shown in Fig. 1, the proposed joint consists of one column base, upper column with a flange, lower column with a flange, one piece of single-angle truss welded to the column base, one piece of singleangle truss bolted to the vertical connecting plate and upper and lower ⁎ Corresponding author at: Beijing University of Technology, Beijing 100024, China. E-mail address: [email protected] (X.C. Liu).

http://dx.doi.org/10.1016/j.jcsr.2015.08.036 0143-974X/© 2015 Elsevier Ltd. All rights reserved.

cover plates, two joint cover plates, two joint flitches, and one vertical connecting plate. Each joint cover plate and the corresponding flange form one component that is named as cover plate flange that is welded to the end of the column in the factory. The column base is formed by a short column and two flanges which are welded together in the factory. The vertical connecting plate is welded to the column base at the side of the short column, and on the surface of the upper and lower flanges. The truss beam consists of two pieces of single-angle truss beams. As shown in Fig. 1, one piece of single-angle truss beam is welded onto the column base and the vertical connecting plate in the factory, and as shown in Fig. 2, this welded part is placed in one module; the other piece of single-angle truss beam, which is placed in another module, is bolted to the vertical connecting plate with joint flitch and bolted to the cover plate with the upper and lower chords. The two pieces of single-angle truss beam are spliced by the bolt at the intersection joint of the chord and web members, thus a double-angle truss beam is formed. On the construction site, the two close modules are connected at the beam– column joints using bolts as well as at each intersection joint of web members and chords using bolts, so that the two close modules are merged together with the splicing of the two piece of single-angle truss. The carried out researches have primarily focused on the basic theory analysis, experimental studies, and design and construction methods for treelike joints, dog bone joints, joints with a cantilever segment, joints with an opening in the web plate, bolted joints with long and circular holes, joints with reinforced haunch or ribbed plates, joints with

418

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

Fig. 1. Structural diagram of the joint. (a) Exploded view of components. (b) Assembly drawing. (c) Assembled joint.

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

419

Fig. 2. Fabrication diagram. (a) Detail drawing of mainboard module. (b) Module splicing.

reinforced flanges, and honeycomb-like frame beam–column joints [9–18]. The literatures retrieved show that the joint that is proposed in this paper has never been described before. We previously conducted a preliminary finite element analysis on it, but did not conducted detail theory analysis and model test [19]. This study conducts full-scale model tests and finite element analysis on four bolted-welded beam– column joints with flange-spliced column base for modularized prefabricated steel structures, and simplified computation formulas are derived. The angle steel truss is widely used in the roof of light weight steel structure, but it is less used in multi-rise building, especially seldom used in high-rise building, because the stress states and requirements are different. The joint of truss and column is normally pined for roof structure without resisting bending moment, but the joint of multistory structure is normally rigid connected or half-rigid connected. For the proposed bolted-welded joint, the web members extend to the limb back of the chord and are welded to both limbs of the chord angle, and are strengthened by the subplate, instead of being connected with gusset plate in light weight steel structure. Subhash [20] designed a four-floor steel structure using angle steel trusses as early as 1994 and investigated the hysteretic performance of single beams under low cyclic loading; however, because the truss beams had excessively thin web members, no out-of-plane constraints and joint connecting problems, the web members prematurely became elastically unstable, which

resulted in relatively poor seismic performance. Hisham improved the steel structure that was designed by Subhash using X-diagonal configuration, and the resulting steel structure exhibited relatively good energy dissipation capacity [21]. Several researchers have placed special energy dissipation devices on the truss beams of structures, and the modified structures exhibited good seismic performance; for example, a 9-floor structure that was equipped with energy dissipation devices on its truss beams exhibited good seismic performance [22–25]. We improve the structure by adjusting the cross-section size and the length of the chords and web members as well as the connecting method of members to avoid elastic instability, which is a relatively cheaper and easier way. 2. Composition and engineering application of the structural system 2.1. Composition of the structural system This study proposes a new modularized prefabricated high-rise angle steel truss beam rectangular hollow section column structure, which mainly includes a prefabricated truss mainboard and a prefabricated flange column. The prefabricated truss mainboard includes a lattice truss beam with angle steel chords and web members, column base and a floor slab. As shown in Fig. 2(a), the trusses and column bases are welded together in the factory, and the floor slab which is made up of reinforced concrete is casted in the factory and connected with truss

420

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

Fig. 4. Details of the test specimens. (a) Elevation drawing of the bolted side. (b) Elevation drawing of the welded side. (c) Planar drawing.

Fig. 3. Two-floor experimental structure. (a) Bolted-welded joints. (b) Hoisting of a truss mainboard modular. (c) Assembled two-story experimental structure.

by studs. The truss mainboard is the assembly module for water, heating, electricity and other utilities. The prefabricated truss mainboard modules are spliced together via single-angle truss at the edge of module and beam-column joints on site to form the beam and floor system. Prefabricated flange columns are connected to the column base which is located in the mainboard through its flange joints to form a multilayer steel frame structure. The mainboards and flange columns are

both prefabricated in the factory and are assembled using bolts on site. In Fig. 2(b), the modules shown in different colors are the different mainboard modules, the modules are integral lifted and connected using the joints that are investigated in this paper to form the beam, column and floor system. 2.2. Engineering application We built a two-floor experimental building. As shown in Fig. 3(a), it includes full-welded beam–column joints in one module that is the right side joint [26], bolted-welded beam–column joints for connecting two modules on site which is the upper and lower side joints and fullbolted joints for connecting two modules which is the left side joint

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433 Table 1 Number and dimension of the specimens. Number

Chord/mm

Web/mm

Test type

Axillary plate

SH1-J SH1-N SH2-J SH2-N

2L75 × 8 2L75 × 8 2L75 × 6 2L75 × 6

2L45 × 6 2L45 × 6 2L45 × 5 2L45 × 5

Static Quasi-static Static Quasi-static

No No No No

[27]. The structure of the experimental building is the first and second stories of a 15-story building which is designed according to the level 8 seismic fortification standard. As shown in Fig. 3(b), the 12 mainboards and columns are lifted one by one, and it took only eight hours to assemble the building on site. Fig. 3 (c) shows a photograph of the building after construction was completed. The completed building has closelyconnected and non-loosened components, which shows the building structure is very stable. The system has also been used in a public building in Yantai China. The building has a total area of 25,546 m2, a height of 42.75 m and includes nine floors above the ground and one below the ground [27]. 3. Test design 3.1. Specimen design Based on the mechanical characteristics of beam–column joints under horizontal and vertical loading conditions, the two ends of the box column were rigidly connected, and the end of the truss beam was subjected to monotonic and cyclic loads. To study the effect of dimensions of the structural steel that is used as the truss beam on the

421

performance of the beam–column joint, test specimens with different thicknesses of the chord and web members were fabricated with all of the other parameters the same. Q235B steel was used for all of the components. Fig. 4 shows the dimensions of the welded-bolted beam–column joint specimen, and Table 1 lists the numbers and cross-sectional dimension of the specimens. 3.2. Loading scheme This set of experiments was conducted at the Engineering Structure Experimental Center of Beijing University of Technology. Fig. 5 shows the test device. Each specimen was placed vertically in the test device. The box column at each end of each specimen was tied to the ground reaction wall. Horizontal sliding of each box column was restricted by a horizontal stopper bolt. The loading point was located at the end of the truss beam. A 50 t hydraulic jack was used to provide the loading. In the static loading test, each specimen was subjected to monotonic tensile loading so that the impact of the torsion of each specimen at the loading point that was caused by the eccentricity of the spherical hinge at the end of the jack could be eliminated. In the quasi-static loading test, the hydraulic jack applied cyclic loading to each specimen; in addition, two angle steel bars were welded onto the chord members of the truss beam of each specimen just below the fastener of the loading device to prevent the fastener from coming off and sliding down. 3.3. Measurement scheme Based on the joint design and the finite element analysis results, measuring points were set on the specimens as shown in Fig. 6. Strain

Fig. 5. Experimental setup. (a) Sketch of the test loading device. (b) Photograph of the test setup.

422

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

Fig. 6. Measuring point layout of test model. (a) Front view. (b) Back view. (c) Left side view. (d) Right side view.

gauges were placed at the limb tips and limb backs of the web and chord angle steel on the same cross section, the measuring points on the chord members are denoted as LXn, the measuring points on the web members are denoted as LFn, the measuring points on the columns, column base and vertical joint plate are denoted as Zn, and the measuring points on the cover plate are denoted as Gn; only the numbers n are shown in Fig. 6. The stress and strain of each specimen during the test process were measured; the axial force and moment on the cross section can be calculated based on the stress. A horizontal displacement meter 1 was placed at the loading end of the truss beam of each specimen to measure the displacement of the end of the beam. Horizontal and vertical displacement meters 2, 3 and 4 were placed at the locations where the truss beam and the columns intersected to measure the displacements of the chord members that were caused by the moment at the end of the beam as well as the overall rigid horizontal displacement of the test model. Vertical displacement meters 5 and 6 were placed at

the flange joints to measure the relative sliding displacements of the left and right flanges. 3.4. Loading system Three loading termination criteria were used for the test model: (1) the specimen was suddenly damaged, and the members suddenly lost their stability, which resulted in the specimen no longer carrying the load; (2) the total rotation angle of the beam end θ was greater than 0.06 rad, that is the beam end displacement Δ = ±43.8 mm; and (3) after reaching the ultimate load Pu, then the load P at the beam end loading point decreased to below 80% of Pu. In the static loading test, monotonic loading with a load of 1 kN at each stage was applied to each specimen until the specimen failed. In the quasi-static loading test, the yield load fy and yield displacement Δy of a specimen were determined based on the static loading test results

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

423

of another specimen with the same cross-sectional dimensions as the specimen. Throughout the loading process, the loading conditions were controlled by the displacement of each specimen; Δy/3 was applied at each displacement stage during the elastic stage, which was cycled once, and Δy/2 was applied at each displacement stage during the elastic–plastic stage, which was cycled twice until the specimen failed [28].

4. Test process analysis Specimen SH1-J underwent a relatively complete loading process; when the displacement at the loading point was 6 mm, the load–displacement curve exhibited an insignificant turning point, and few measuring points yielded. When the displacement at the loading point reached 9.3 mm, the load–displacement curve exhibited a significant turning point, and the displacement began to acceleratedly increase; however, a visual inspection detected no significant deformation in any of the components. With an increasing load, micro-cracks appeared at the welding seam between tension chord member B and the column base flange. These micro-cracks continued to develop; however, the load–displacement curve did not decrease suddenly, and the load could be applied to the specimen continuously. When the load reached 250 kN, significant torsional deformation occurred to web member A, and concave deformation occurred to the end of the compression chord member A that was welded with angle steel near the joint. When the load reached 270 kN, the load decreased suddenly to 260 kN accompanied by a small sound, and the welding seams fracture partially. Afterwards, the applied load continued to increase; when the displacement at the loading point was 50 mm, web member A became significantly deformed, and the welding seam between tension chord member B and the column base flange completely fractured. Fig. 7 shows the failure mode of the specimen. Specimen SH1-N underwent a relatively complete loading process. When the displacement at the loading point reached + 6 mm during the first cycle, the curve exhibited a slight turning point, and the specimen initially exhibited elastic–plastic properties. When a displacement of approximately −6 mm was applied during the first cycle, the load– displacement curve also exhibited a slight turning point. When the displacement at the loading point reached +12 mm during the first cycle, cracks appeared at the welding seam on the angle steel of chord member B near the joint end, but the welding seam did not completely fracture. When the displacement at the loading point reached − 12 mm during the first cycle, cracks appeared at the welding seam on the angle steel of chord member A near the joint end, and a small sound was heard; the load–displacement curve began to fluctuate, and the load decreased suddenly by small level. When the displacement at the loading point reached 12 mm during the second cycle, compression web member A began to exhibit slight deformation; with the increasing displacement, the deformation gradually increased. When the displacement at the loading point reached −15 mm during the first cycle, the welding seam between welded single-angle of chord member A and column base flange completely fractured, and the bolted single-angle of chord member B, which was bolted to the cover plate, exhibited slight bending deformation near the joint cover plate. When the displacement at the loading point reached − 15 mm during the second cycle, compressed web member B exhibited slight deformation. With further loading, this deformation continued and increased. When the applied load reached 266 kN, the web members became significantly deformed; the welding seam between welded single-angle of chord member B and column base flange fractured and the vertical connecting plate partially fractured subsequently with a very loud sound; because no damage occurred to the bolted side of the angle truss beam, the entire truss beam twisted sideways toward the bolted side, the limb back expanded outward, and the test was terminated. However, neither the bolted connection between the cover plate and the chord slid nor the

Fig. 7. Failure mode of specimen SH1-J. (a) Bending-torsional deformation of web member A. (b) Bending deformation of chord member A. (c) Overall deformation of the specimen. (d) Fractured welding seam of the specimen.

chords fracture during the load process. Fig. 8 shows the failure mode of the specimen. Specimen SH2-J also underwent a relatively complete loading process. When the displacement at the loading point reached 5.1 mm, the load–displacement curve exhibited a slight turning point, and some measuring points on the specimen yielded; however, the specimen exhibited no significant deformation. When the load reached 210 kN, compression web member A exhibited significant deformation; afterwards, the bolted angle steel of the compression chord member A exhibited concave deformation at the location near the joint cover plate. As the loading process continued, these two types of deformation continued to develop. Eventually, due to the significant deformation of the web and chord members, the loading process was terminated. No

424

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

Fig. 8. Failure mode of specimen SH1-N. (a) Bending-torsional deformation of web member A. (b) Bending-torsional deformation of web member B. (c) Deformation at the end of a chord member. (d) Overall deformation of the specimen. (e) Fracturing of the welding seam on the welding angle steel. (f) Partial fracturing of the vertical connecting plate.

welding seams fractured during the loading process. Fig. 9 shows the failure mode of the specimen. Specimen SH2-N also underwent a relatively complete loading process. When the displacement at the loading point reached + 5.5 mm, the load–displacement curve exhibited a slight turning point. When the displacement at the loading point reached − 5.5 mm, the load–displacement curve also exhibited a slight turning point; some points on the specimen entered yielding state. When the displacement at the loading point reached +15 mm during the first cycle, web member A exhibited significant deformation, the welding seam between welded single-angle of chord member B and column base flange fractured. When the displacement at the loading point reached −15 mm during the first cycle, the welding seam between welded single-angle of chord member A and column base flange fractured. With the continuation of the loading process, the deformation continued to develop; in addition, the bolted single-angle of chord members A and B, which were bolted to the cover plates, exhibited convex or concave deformation to varying degrees at the location near the joint cover plates. Because no damage occurred to the bolted side of the specimen, the entire truss beam twisted sideways toward the bolted side, and the limb back expanded outward. When the displacement at the loading point reached −36 mm during the first cycle, the weld seam between web member A and chord member A partially fractured, and web member A exhibited significant torsional buckling deformation. Other locations also exhibited very large deformations, and the test was terminated. However, neither the bolted connection between the cover plate and the chord slid nor the

chords fracture during the load process. Fig. 10 shows the failure mode of the specimen. 5. Test results and analysis 5.1. Main performance The data of the entire testing process were processed, and the yield load Py, ultimate load Pu, ultimate failure load P0.85, yield displacement at the loading point Δy, ultimate displacement Δu and ultimate rotational angle failure load P0.05 of each specimen were obtained. The test load– displacement curves show that after the joint reached its ultimate bearing capacity, the load did not significantly decrease to P0.85. In this instance, the corresponding load when the bearing capacity does not decrease significantly and the test was stopped was denoted as Pl, and the corresponding displacement was the ultimate displacement Δu. Table 2 lists the data of specimens SH1-J and SH2-J, and Table 3 lists the data of specimens SH1-N and SH2-N. 5.2. Load–displacement curves The load–displacement curve of the static loading test can reflect the property variations of the beam–column joint, such as the static bearing capacity and the rotational stiffness with the load on the beam end. Fig. 11 shows the curve for the load at the loading point on the beam end and the displacement of the beam end.

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

425

Fig. 9. Failure mode of specimen SH2-J. (a) Bending-torsional deformation of web member A. (b) Deformation at the end of a chord member. (c) Overall deformation of the specimen.

Fig. 11 and Table 2 demonstrate that the fracturing of the welding seam between chord of the truss beam and the column flange did not result in significant decrease in the bearing capacity and large fluctuations in the load–displacement curves of the bolted-welded joint; this was because the angle steel that was bolted to the cover plate could continue to carry the load after the welding seam between the welded angle steel and the column base flange fractured. Table 2 indicates that the bolted-welded joint still exhibited relatively large bearing capacity when there was a relatively large displacement at the loading point. Elastic–plastic failure was the final failure mode of each specimen; each specimen exhibited significant plastic deformation before failure, and thus, the failure of each specimen is a typical ductile failure. 5.3. Hysteretic performance The P–Δ hysteretic curve that is obtained from the quasi-static loading test can reflect the seismic performance of the joint, such as the elastic–plastic performance, ductile performance, stiffness and energy dissipation capacity; thus, the P–Δ hysteretic curve is the main basis for the evaluation of seismic performance [29]. Fig. 12(a) shows that due to the eccentricity of the loading point caused by the sudden fracturing of the welding seam between the angle steel and the column base flange as well as the local tensile failure of the vertical connecting plate, the hysteretic curve of specimen SH1-N is not very plump but it does not exhibit a significant decreasing trend. In addition, the energy dissipation capacity of specimen SH1-N is lower than that from the finite element simulation, and specimen SH1-N mainly relied on the

Fig. 10. Failure mode of specimen SH2-N. (a) Bending-torsional deformation of web member A. (b) Bending-torsional deformation of web member B. (c) Deformation at the end of a chord member. (d) Outward expansion of a chord member. (e) Overall deformation of the specimen.

426

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

Table 2 Main performance of the specimens by static test. Specimen

Py/kN

Pu/kN

P0.85/Pl

Δy/mm

Δu/mm

Δu/Δy

P0.05/kN

SH1-J SH2-J

162.99 133.20

296.39 240.21

/289.77 /239.74

6.00 4.99

50.24 43.02

8.37 8.62

285.39 235.92

angle steel that was bolted to the cover plate to dissipate energy. The finite element simulation results show that when no fracture failure occurs in the welding seam, the hysteretic curve of specimen SH1-N is plump and it is an apparent spindle shape. In addition, the ultimate bearing capacity of specimen SH1-N by FEA is slightly higher than the test value, which indicates that the fracturing of the welding seam did not significantly decrease the ultimate bearing capacity of the specimen and that the bolted part had a very high bearing capacity, but the fracturing of the welding seam resulted in a significant decrease in the energy dissipation capacity of the joint. Fig. 12(b) shows that the test curve of specimen SH2-N is plump, exhibits an apparent spindle shape, and matches the finite element simulation relatively well, which is mainly because even though the welding seam between the angle steel and the column base flange fractured during the loading process, it did not result in eccentricity of the loading point, and the energy dissipation capacity of the bolted part of the joint was fully utilized. The vertical connecting plate did not fracture during the loading process, and the test had relatively good continuity. Even though the area of the component section of specimen SH2-N was less than that of specimen SH1-N and the ultimate bearing capacity of specimen SH2-N was less than that of specimen SH1-N, the energy dissipation capacity of specimen SH2-N was significantly higher than that of specimen SH1-N. The test values shown in Fig. 12 are slightly less than the finite element simulation values, which is mainly because the finite element model did not consider the initial defects of the component, and the second-order effect on the structural deformation caused by initial defects makes the stiffness and bearing capacity of the structure decreased slightly. 5.4. Skeleton curves A skeleton curve is formed by connecting the peak points of each cycle on a hysteretic curve and reflects the yield load and ultimate bearing capacity of a specimen. Fig. 13 shows the skeleton curves of specimens SH1-N and SH2-N. The skeleton curves of the two joints are essentially straight lines and coincide at the elastic stage, and the stiffness of specimen SH1-N is slightly higher than that of specimen SH2N. The ultimate bearing capacity of specimen SH1-N is higher than that of specimen SH2-N; however, due to the fracturing of the welding seam between the angle steel and column base flange as well as the eccentricity due to loading, the skeleton curve of specimen SH1-N has no descending part and exhibits less significant plastic deformation. The plastic deformation of specimen SH2-N is very significant; its skeleton curve has a significant descending part. The trends of this type of joint in the positive and negative directions are essentially the same. The ultimate loads in the two directions are also similar, which reflects that the bolted-welded joint has the same mechanical and seismic performance in the positive and negative directions. Even though the web members are not symmetrical along the center line of the truss, the mechanical performance of the web members is nearly symmetrical; thus,

Fig. 11. Load–displacement curve.

the web members have similar tensional and compressional bearing capacities.

5.5. Ductile performance and energy dissipation capacity The ductile performance of a component refers to its non-elastic deformability when the bearing capacity of the component does not degrade significantly. The ductile performance of a component can be calculated based on the hysteretic curve of the component. The displacement ductility ratio μ is Δu/Δy, where Δu represents the maximum horizontal displacement when the specimen fails [30,31]. The energy dissipation capacity of a specimen is described using the area surrounded by the hysteretic loop of one cycle of the hysteretic curve of the specimen. The energy dissipation capacity is an important index that reflects the seismic performance of a specimen. This study describes the energy dissipation capacity using the equivalent viscous damping coefficient he [32] calculated by the last hysteretic loop. Table 4 lists the displacement ductility ratios and equivalent viscous damping coefficients he of the two beam–column joints. Table 2 shows that the corresponding ductility coefficients of static loading test specimens SH1-J and SH2-J are 8.37 and 8.62, respectively. Both coefficients are greater than 3.5; thus, specimens SH1-J and SH2-J both exhibit relatively good ductile performance. Table 4 shows that the ductility of specimen SH1-N is inferior to that of specimen SH2-N; thus, the twisting of specimen SH1-N that occurred after the welding seam fractured was extremely detrimental to the plastic development of the joint and resulted in a change in the load transmission path and rapid collapse of the load transmission system of the specimen. Because specimen SH2-N did not twist, it exhibited good ductile development. Hence, even though the welded part and bolted part of the boltedwelded joint are designed and calculated according to the current Code for Design of Steel Structures [33] and have similar design strengths, the strength of the welded part during the test were not the same as the strength of the bolted part; the strength of the bolted part was significantly higher than that of the welded part. In particular, after the load exceeded the design strength, the welding seam gradually began to fail under hysteretic loading, but the bolted connection was not affected by the cycling load.

Table 3 Primary performance indicators of the specimens of the quasi-static tests. Specimen

SH1-N SH2-N

Δy/mm

Py/kN

Δu/mm

Pu/kN

P0.85 or Pl/kN

Positive

Negative

Positive

Negative

Positive

Negative

Positive

Negative

Positive

Negative

168.09 134.06

172.51 144.32

6.00 4.26

7.01 4.76

269.10 234.34

291.93 230.39

29.02 31.07

28.42 30.71

262.29 199.19

266.12 195.83

Note: tension of the jack is set as positive direction and compression is set as negative direction.

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

427

Table 4 Displacement ductility coefficient of the specimens under quasi-static loading. μ/μ0

Specimen number

μ Positive

Negative

μ0 Positive

Negative

Positive

Negative

SH1-N SH2-N

4.84 7.29

4.05 6.45

6.50 7.44

6.62 6.71

0.74 0.98

0.61 0.96

he

0.23 0.36

Note: μ0 is the simulated value. μ is the test value.

Table 5 Rotation angle of specimens during quasi-static loading. Specimen number

SH1-N SH2-N

θy

θu

θp

Positive

Negative

Positive

Negative

Positive

Negative

0.040 0.043

0.039 0.042

0.008 0.006

0.009 0.007

0.032 0.037

0.030 0.035

minimum requirement of the seismic design 0.03 rad [34]; thus, these two dimensional specimen can be used in high seismic regions. The two dimensional specimens had similar rotational capacities in the positive and negative directions, without being affected by asymmetry of the truss beam. 5.7. Stiffness degradation

The ultimate rotation angle and plastic angle of a joint are important indexes for evaluating the rotation capacity of the joint. Table 5 lists the test values of the ultimate rotation angles θu, elastic rotation angles θy and plastic rotation angles θP of the quasi-static specimens under the effects of positive and negative loading. θu = Δu/l, θy = Δy/l, θp = θu − θy where Δu represents the ultimate displacement of the loading point, and l represents the distance between the loading point and the centroid of the rotating cross-section, where 730 mm is the value of l in this paper. Table 5 shows that the welding seams between the end of the chord member and column-base flange had a relatively small impact on the rotation angle of the joint; after the welding seams fractured, the bolted part could still play a role in carrying the load and deforming. The plastic rotation angles of the two dimensional specimens in the positive and negative directions are both greater than the

Stiffness degradation is an important index for studying and calculating the seismic performance of a structure. The slope of the line that connects a point on the skeleton curve of a specimen obtained from the quasi-static loading test to the origin of the coordinate system is defined as the equivalent stiffness. Stiffness degradation can be expressed using the equivalent stiffness degradation coefficient, which is defined as the ratio of the equivalent stiffness to the maximum stiffness of a specimen; stiffness degradation curve is shown in Fig. 14. Fig. 14 shows that for the bolted-welded joint, the two specimens were in the elastic state in the horizontal straight line and that the equivalent stiffness of the two specimens remained essentially the same. When the displacement at the loading point reached ±5 mm to ±7 mm, both specimens entered the yield state, the stiffness degradation coefficient curves of the two specimens began to decrease suddenly, which shows that the stiffness of joint started to degrade. Fig. 14 and Table 3 indicate that both specimens were in the elastic–plastic stage at this point, but neither specimen had reached its ultimate bearing capacity. At this stage, due to the impact of the eccentricity, the stiffness degradation rates of specimen SH1-N in the positive and negative directions were both greater than those of specimen SH2-N. When the displacement reached ±16 mm, the stiffness degradation coefficient curves of the two specimens began to level off, and the changes in the stiffness degradation rates of the two specimens slowed. In addition, the specimens reached their ultimate bearing capacities at this stage, after which the load

Fig. 13. Skeleton curves.

Fig. 14. Stiffness degradation curve.

Fig. 12. Load–displacement hysteric curve. (a) P–Δ hysteretic curve of specimen SH1-N. (b) P–Δ hysteretic curve of specimen SH2-N.

5.6. Rotation capacity

428

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

gradually decreased instead of increasing with further loading. At this time, due to the impact of the eccentricity, specimen SH1-N twisted to a relatively large degree and exhibited unstable deformation; in addition, the stiffness degradation rate of specimen SH1-N in the positive direction was slower than that of specimen SH2-N. The test stiffness degradation rate of specimen SH1-N is faster than its simulated value, which is mainly because the finite element simulation does not consider the impacts of the initial defects, which results in relatively ideal results. The test values are consistent with the simulation values, which verifies the stiffness degradation patterns of these two dimensional joints.

5.8. Strain analysis The strain analysis on the measuring points of the specimens in the static loading test was conducted, as shown in Fig. 15, in which the two horizontal dash lines denote the yield load of specimens SH1-J and SH2J with a value of 162.99 kN and 133.20 kN, respectively. The ordinate is the load when the measuring point reaches the yield strain. It is evident from Fig. 15 that the first yielding location of both specimens is located at web A that is under compression. The location is on the tips at the end of the angle steel close to chord B. When the overall yield occurred in both specimens, most measuring points on the specimens did not yield. A strain analysis on the measuring points at the end section of the chord near the joint and the key section of the web during the quasistatic tests was conducted as shown in Figs. 16 and 17. It is evident from Fig. 16 that for specimen SH1-N, when loading in the positive direction, the displacement at the loading point is 6.00 mm, which is the yield displacement, measuring points No. 8 and No. 16 at the end of angle steel of web A and measuring point No. 10 at the web B yield, but the chord does not yield. At the ultimate displacement of 29.02 mm, measuring

points No. 12, No. 21 and No. 22 at the end of the chord yield and the web yields as well. When loading in the negative direction at the yield displacement of 7.01 mm, measuring points No. 8 and No. 16 at the end of angle steel of web A and measuring point No. 10 at web B yield, while the chord does not. At the ultimate displacement of 28.42 mm, the chord yields as well. It is evident from Fig. 17 that for specimen SH2-N, when loading occurs in the positive direction, the displacement at the loading point is 4.26 mm, which is the yield displacement, measuring points No. 8 and No. 16 at the end of angle steel of web A and measuring point No. 2 at the web B yield, but the chord does not yield. At the ultimate displacement of 31.07 mm, the chord yields as well. When loading in the negative direction at the yield displacement of 4.76 mm, measuring point No. 8 at the end of angle steel of web A and measuring point No. 12 at the chord A yield. At the ultimate displacement of 30.71 mm, the chord yields as well.

6. Finite element analysis 6.1. Analysis model A finite element analysis was conducted on the four specimens using the finite element software ABAQUS. Because of the irregularity of the shape, the C3D10 tetrahedral element was used for the truss beam, and the C3D8R hexahedral element was used for other regular-shaped parts. The parts that were connected using high-strength bolts were defined to be in a contact relationship. To speed up the FMA, the bolts were not built in the analysis model, the pretensions of the high-strength bolts were simulated using the equivalent force method, and the pretensions were applied on the cover plate and chord. The friction coefficient of the bolt-connected contact surfaces between cover plate and

Fig. 15. Yield load of each measuring point of the specimens under static loading. (a) Load when the measuring points on the chord yield. (b) Load when the measuring points on the web yield.

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

429

Fig. 16. Strain distribution of specimen H1-N. (a) Strain distribution during positive loading. (b) Strain distribution during negative loading.

chord is 0.303, which was the average value of three friction test models as 0.29, 0.30 and 0.32 respectively according to the Technical specification for high strength bolt connections of steel structures [35], as shown in Fig. 18. The boundary conditions and load were applied according to the actual status in the test, as shown in Fig. 19. Three tests were conducted on each type of component to determine its material properties. A total of five sets of material property tests on 15 specimens were conducted. All of the average material test values were used as material constitutive relations for the finite element

analysis and to process the test data. The materials were simulated based on the bilinear kinematic hardening rule. The yield strength was 294 MPa, the elastic modulus was 2.06 × 105 MPa, and the Mises yield criterion was used. 6.2. Analysis results The failure mechanism of the components in the finite element calculation was similar to that during the test; the component underwent

Fig. 17. Strain distribution of specimen H2-N. (a) Strain distribution during positive loading. (b) Strain distribution during negative loading.

430

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

that is the stiffness degraded relatively slowly during the simulation process. However, the trends of the stiffness degradation coefficient curves obtained from the simulation and tests are the same. 7. Simplified calculation of joint The joint calculation assumes that the bolted connection and the welded connection bear the external force on the truss beam together. The welded single-angle truss is welded to the column base with equivalent strength, the calculation sketch is shown in Fig. 21 (a), all of the tension is carried by the welding seam on the chord with the center of the welding seam being the same as that of the chord, and all of the shear force is carried by the bolts on the web plate. For the bolted single-angle truss, as shown in Fig. 21 (b), all of the tension is carried by the bolts on the chord members, and all of the shear force is carried by the bolts on the web plate. The mechanical and FE analyses show that the prying action by cover plate on the chord surface is very small, so it is neglected in the calculation sketch. Grade S10.9 M24 bolts are used. Four bolts are placed on each chord member, and three bolts are placed on the vertical connecting plate and joint flitches. The design shear capacity of one M24 bolt is: Nbv ¼ n f μp ¼ 1  0:303  225 ¼ 68:2 kN:

ð1Þ

The bearing capacity at the end of the truss supported by bolts is: P1 ¼ 4Nbv  h=l1 ¼ 4  68:2  0:3=0:645 ¼ 126:9 kN:

ð2Þ

The bearing capacity at the end of the truss supported by welding seams is: For SH1 :

P2 ¼ Af y h0 =l1 ¼ 1:15  294  ð0:3−2  0:0215Þ=0:645 ¼ 134:7 kN

For SH2 :

P2 ¼ Af y h0 =l1 ¼ 0:880  294  ð0:3−2  0:0207Þ=0:645 ¼ 103:8 kN:

ð3Þ The bearing capacity at the end of the truss supported by welding seams and bolts is: For SH1 : For SH2 :

Pc ¼ P1 þ P2 ¼ 261:6 Pc ¼ P1 þ P2 ¼ 230:7:

ð4Þ

The web bolts are 3 Grade S10.9 M20 bolts with the shearing capacity: V ¼ 3Nbv ¼ 3n f μp ¼ 3  2  0:303  155 ¼ 281:8 kN: Fig. 18. Friction coefficient test setup. (a) Picture of the test models. (b) Test setup.

complete elastic and elastic–plastic development process. The main failure deformation of the specimens with high-quality welding seams in the finite element analysis was the same as that during the test; the deformation occurred at the ends of the compressed chord members and web members. Fig. 20 shows the typical failure modes. Figs. 11 and 12 show the load–displacement curves of the four specimens from the finite element analysis. The simulated load–displacement curves are consistent with the load–displacement curves that were obtained from the tests; however, the finite element simulation values are relatively ideal, and the ultimate bearing capacities, ultimate displacements and areas of the hysteretic curves obtained from the finite element simulations are slightly greater than those obtained from the test, which is mainly due to the fact that the simulation did not consider the impacts of initial defects and the fracturing of the welding seams. Figs. 13 and 14 also show that the stiffness degradation coefficients obtained from the simulation are slightly higher than the values obtained from the tests;

ð5Þ

The shearing capacity of the joint is 281.8 kN; it is large than the test load 269 kN and 230 kN at the end of truss, so the web bolts can resist the load with a large margin. The yielding loads of the SH1 and SH2 were 261.6 and 230.7 kN by formula (4); during the test, the minimum broken loads of hysteretic test in the positive and negative directions of SH1 and SH2 were 269 and 230 kN from Table 3, respectively; under these loads, the bolted connection did not slide but the welding seams fracture, thus, the calculated results of the proposed simplified calculating Eqs. (1) to (5) meet the models test results well, which verifies the calculating equations. 8. Conclusions This study conducts monotonic static loading and cyclic loading tests and finite element analysis of bolted-welded beam–column joints in modularized prefabricated steel structures. The main conclusions are summarized as follows: (1) The welding seams between the ends of the chord members and the flange and the joint vertical connecting plate on the welded side significantly affected the failure mode and various mechanical properties of the beam–column joint. Sudden fractures of the

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

Fig. 19. Finite element model.

Fig. 20. Typical failure modes. (a) Failure deformation of specimen SH1. (b) Failure deformation of specimen SH2.

431

432

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433

Notation The following symbols are used in this paper: A fy l l1 h h0 P1 P2 Pc

section area of chord member; yield strength of the test material; distance between welding seam and the loading point of beam end; distance between vertical bolt line and the loading point of beam end; height of the truss; distance between center line of the upper chord and that of the lower chord, the effective height of truss; calculation load-carrying capacity of the loading point of bolted single-angle truss calculated by Eq. (2); calculation load-carrying capacity of the loading point of welded single-angle truss by Eq. (3); calculation load-carrying capacity of the loading point of beam end calculated by Eq. (4);

Acknowledgments The writers gratefully acknowledge the support for this work, which was funded by the National Natural Science Foundation of China (51278010). Fig. 21. Calculation sketch of joint. (a) Welded single-angle truss. (b) Bolted single-angle truss.

Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.jcsr.2015.08.036.

(2)

(3)

(4)

(5)

(6)

welding seams resulted in a decrease in the plastic deformability of the joint. However, due to the relatively high strength and plastic deformation capacity of the bolted side, the joint could still keep better function under the effect of the splicing bolts on the truss beam and the vertical connecting plate after the welding seams fractured, which resulted in only less decrease in the plastic deformability of the joint. The full-strength welding seams between the ends of the chord members and the flanges could satisfy the static design requirements for strength. To better meet the seismic energy dissipation requirements, it is recommended that axillary plate be added at the ends of the chord members to increase the lengths of the welding seams to ensure that they will not fracture under hysteretic load. The bolted-welded joints exhibited good seismic performance, such as good rotation capacity, ductile performance and energy dissipation capacity. The presence of the bolted connection of the truss beam ensured that the energy dissipation capacity would not decrease significantly due to the brittle fracture of the welding seams. The plastic rotation angles of the joints were all greater than 0.03 rad, which met the structural seismic requirements. The mechanical properties of the bolted-welded joints including rotational stiffness and ultimate bearing capacity in the positive and negative directions were similar and symmetrical. The fatigue failure of the welding seams under cyclic load resulted in the ultimate load of the joint being lower under hysteretic load than under static load. When the welding seams fractured suddenly under static or hysteric load, the sudden change in the bearing capacity of the joint was relatively small due to the presence of the bolted connection. The results of the simplified calculation formulas proposed in the paper gets along well with that of the model test. The formulas can be used to describe the strength of joint under static load and cyclic load.

References [1] GB/T 50378, Assessment Standard for Green Building, China Building Industry Press, Beijing, China, 2013 (in Chinese). [2] A.L. Zhang, X.C. Liu, The new development of industrial assembly high-rise steel structure system in China, Proceedings of Tenth Pacific Structural Steel Conference, Singapore October 8-11 2013, pp. 976–981. [3] A.L. Zhang, The key issues of system innovation, drawing up standard and industrialization for modularized prefabricated high-rise steel structures, Ind. Constr. 44 (08) (2014) 1–6 (in Chinese). [4] L. Jaillon, C.S. Poon, The evolution of prefabricated residential building systems in Hong Kong: a review of the public and the private sector, Autom. Constr. 18 (3) (2009) 239–248. [5] T.N. Dao, J.W. van de Lindt, Seismic performance of an innovative light-frame coldformed steel frame for midrise construction, J. Struct. Eng. 139 (2013) 837–848. [6] A.L. Zhang, T.T. Hu, X.C. Liu, The classification and comparative analysis of the matching external wall for the prefabricated steel structure residence, Ind. Constr. 44 (08) (2014) 23–26 (in Chinese). [7] W. Wang, Y.Y. Chen, Y.C. Yu, L.W. Tong, J.X. Yang, D.W. Liu, F. Kenji, Floor-by-floor assembled steel braced structures for prefabricated buildings, J. Build. Struct. 42 (10) (2012) 48–52 (in Chinese). [8] X.X. Zha, L.L. Wang, S.T. Zhong, The method of constructing multi-storied used shipping container buildings and the deduction of practical formula about structural security, J. Build. Struct. 40 (6) (2010) 462–465 (in Chinese). [9] M. Gerami, H. Saberi, V. Saberi, A.S. Daryan, Cyclic behavior of bolted connections with different arrangement of bolts, J. Constr. Steel Res. 67 (4) (2011) 690–705. [10] Q.S. Kent Yu, C.M. Uang, J. Cross, Seismic rehabilitation design of steel moment connection with welded haunch, J. Struct. Eng. 126 (1) (2000) 69–78. [11] K. Michael Mcmullin, A. Astaneh-Asl, Steel semirigid column-tree moment resisting frame seismic behavior, J. Struct. Eng. 129 (9) (2003) 1243–1249. [12] C.C. Chen, C.C. Lin, Seismic performance of steel beam-to-column moment connections with tapered beam flanges, Eng. Struct. 48 (2013) 588–601. [13] S. Maleki, M. Tabbakhha, Numerical study of slotted-web-reduced-flange moment connection, J. Constr. Steel Res. 69 (1) (2012) 1–7. [14] R.L. Ma, Y. Yang, Q.S. Chen, L.X. Lu, Seismic performance testing study on high strength bolt connections with slotted holes, J. Build. Struct. 30 (1) (2009) 101–106 (in Chinese). [15] Q.S. Yang, A seismic connection of steel moment-resisting frame with opening on beam web, China Saf. Sci. J. 15 (2) (2005) 45–50 (in Chinese). [16] Y.J. Shi, G. Shi, Y.Q. Wang, A simplified calculation method for moment–rotation curve of semi-rigid end-plate connections, China Civ. Eng. J. 39 (3) (2006) 19–23 (in Chinese). [17] Y. Wang, S. Feng, Y.T. Wang, Experimental study on hysteretic behavior for rigidreinforced connections, China Civ. Eng. J. 44 (5) (2011) 57–68 (in Chinese).

X.C. Liu et al. / Journal of Constructional Steel Research 115 (2015) 417–433 [18] L.G. Jia, H.C. Li, Y.H. Wu, Experimental study of the behavior of beam–column connections of cellular steel frames under low-cyclic reversed loading, China Civ. Eng. J. 45 (1) (2012) 61–68 (in Chinese). [19] X.C. Liu, A.X. Xu, Z. Ni, A.L. Zhang, Analysis of the limit bearing capacity and seismic performance on typical joint of truss–beam and column in fabricated high-rise steel structure, Ind. Constr. 44 (08) (2014) 23–26 (in Chinese). [20] S.C. Goel, A.M. Itani, Seismic behavior of open-web truss-moment frame, J. Struct. Eng. 120 (6) (1994) 1763–1780. [21] H.S. Basha, S.C. Goel, Special truss moment frames with Vierendeel middle panel, Eng. Struct. 17 (5) (1995) 352–358. [22] G. Pekcan, C. Linke, A. Itani, Damage avoidance design of special truss moment frames with energy dissipating devices, J. Constr. Steel Res. 65 (6) (2009) 1374–1384. [23] A. Longo, R. Montuori, V. Piluso, Theory of plastic mechanism control of dissipative truss moment frames, Eng. Struct. 37 (4) (2012) 63–75. [24] N. Wongpakdee, S. Leelataviwat, S.C. Goel, W.C. Liao, Performance-based design and collapse evaluation of Buckling Restrained Knee Braced Truss Moment Frames, Eng. Struct. 60 (2) (2014) 23–31. [25] G. Pekcan, A.M. Itani, C. Linke, Enhancing seismic resilience using truss girder frame systems with supplemental devices, J. Constr. Steel Res. 94 (3) (2014) 23–32. [26] X.C. Liu, A.X. Xu, A.L. Zhang, Z. Ni, H.X. Wang, L. Wu, Static and seismic experiment for welded joints in modularized prefabricated steel structure, J. Constr. Steel Res. 112 (9) (2015) 183–195.

433

[27] X.C. Liu, A.X. Xu, A.L. Zhang, Z. Ni, The static performance analysis and experimental research of the joint in modular prefabricated high-rise steel structure, Ind. Constr. 44 (08) (2014) 27–34 (in Chinese). [28] ANSI/AISC 341, Seismic Provisions for Structural Steel Buildings, American Institute of Steel Construction, Chicago, Illinois, U.S.A., 2005 [29] A.L. Zhang, J. Yu, M. Xu, Experimental research on steel specially shaped columns with cruciform section under cyclic loading, J. Build. Struct. 31 (02) (2010) 11–19 (in Chinese). [30] C.D. Stoakes, L.A. Fahnestock, Cyclic flexural testing of concentrically braced frame beam–column connections, J. Struct. Eng. 137 (7) (2011) 739–747. [31] B. Guo, Y.L. Guo, F. Liu, G.M. Li, C.Y. Chi, Research on cyclic behavior of welded and bolted steel frames, J. Build. Struct. 27 (2) (2006) 47–56 (in Chinese). [32] L.M. Li, Z.H. Chen, N. Li, Experimental study on seismic capability of diaphragmthrough style beam–column joint, J. Earthq. Eng. Eng. Vib. 27 (1) (2007) 46–53 (in Chinese). [33] GB50017, Code for Design of Steel Structures, China Planning Press, Beijing, China, 2003 (in Chinese). [34] FEMA-350, Recommended Seismic Design Criteria for new Steel Moment-Frame Buildings, SCA Joint Venture, Federal Emergency Management Agency, Washington DC, USA, 2000. [35] JGJ82-2011, Technical Specification for High Strength Bolt Connections of Steel Structure, China Building Industry Press, Beijing, China, 2011 (in Chinese).