Shear behaviour of open beam-to-tubular column angle connections

Transcription

1 Shear behaviour of open beam-to-tubular column angle connections Y. Liu, C. Málaga-Chuquitaype & A. Y. Elghazouli Department of Civil and Environmental Engineering, Imperial College London, UK ABSTRACT: Whilst the moment-rotation behaviour of typical connection configurations has been extensively examined in previous studies, there is a relative lack of information on the performance under more generalised loading conditions, particularly in relation to semi-rigid connections to tubular columns under shear loads. To this end, this paper deals with the shear behaviour of open beam-to-tubular column angle connections by means of experimental and numerical investigations. The experimental set-up, connection configurations and material properties are first introduced, followed by an overview of the results and observations from the tests on both blind-bolted angle connections and combined channel/angle details. The main behavioural patterns are identified and their effect on the connection performance is discussed. Subsequently, detailed three-dimensional finite element models are constructed by means of the nonlinear program ABAQUS, and their results are validated against the experimental data. It is shown that the FE model can accurately and efficiently simulate the overall behaviour of bolted angle connections. The experimental and numerical results presented in this paper highlight the main inelastic response characteristics of these forms of connection under shear loading conditions. 1 INTRODUCTION Hollow structural sections (HSS) represent an effective choice as column members due to both their structural efficiency and architectural appeal. Nevertheless, the difficulties associated with the lack of access for the installation of conventional bolts have often resulted in the under-exploitation of these merits. This situation is aggravated by the relative lack of research and design guidance for bolted angle connections with tubes. Most of the research on open-beam-to-tubular-column connections to date has focused on fully-rigid fully-welded details (Cao et al. 1998, Kosteski & Packer 3), and current European standards (CEN 5) incorporate rules for determining the stiffness and resistance of semirigid connections that do not extend to connections incorporating tubular columns. The practical and economical merits offered by bolted angle connections between open beams and open columns have prompted the development of semi-rigid open beam-to-tubular column connection alternatives such as the flowdrill process (Banks 1993) and special bolts with sleeves designed to expand inside the tube (Huck International Inc. 199, Lindapter International Ltd. 1995). A simpler blindbolt design is that presented by the Hollo-bolt, proposed by Lindapter International (1995). In particular, the wide availability and ease of use of the Hollo-bolt have motivated a number of experimental studies on Hollo-bolted connections subjected mainly to bending and direct tension (Barnett et al. 1, Elghazouli et al. 9). Barnet et al. (1) performed a review of different blind-bolting alternatives and carried out an experimental study on blindbolted T-stubs and connections using Hollo-bolts. More recently, Elghazouli et al. (9) performed an experimental investigation into the monotonic and cyclic behaviour of Hollo-bolted beam-to-tubular column connections. It was shown that the grade of the Hollo-bolt, coupled with the gauge distance between the Hollo-bolt and beam flange, have a most notable effect on the flexural response of this type of connection. Nevertheless, information on the behaviour of blind-bolted connections under other loading conditions is still limited, particularly for angle connections subjected to shear loads. Another alternative for bolted connections between open beams and tubular columns is that offered by combined channel/angle configurations. This type of connection incorporates a channel section which is shop-welded at the legs to the face of the column hence allowing the use of any conventional bolted detail between the channel and the beam. Despite its versatility, there is a dearth of experimental studies on such reverse channel configurations. Málaga-Chuquitaype & Elghazouli (21) carried out an experimental study into the flexural behaviour of combined channel/angle connections under monotonic and cyclic loading. It was observed that the flexibility of the reverse channel component has a direct influence on both the initial rotational stiffness and moment capacity of the connection. Nevertheless, as with blind-bolted details, there is a lack of information on the behaviour of combined channel/angle connections under other forms of loading conditions such as shear loads. Many studies have been carried out on the numerical simulation of semi-rigid connections incorporating conventional bolts (Kishi et al. 1, Citipitioglu et al. 2, Pirmoz et al. 8). However, detailed finite element models for the simulation of blindbolted or reverse channel joints with angles are still

2 limited. Recently, Wang et al. (21) developed theoretical and numerical models using ANSYS (3) to investigate the tension behaviour of Hollo-bolted T-stubs. However, similar detailed models for the simulation of blind-bolted or reverse channel joints with angles are still lacking. This paper examines the shear behaviour of semirigid angle connections between open beams and tubular columns through experimental and numerical investigations. Three shear tests on blind-bolted angle connections and three tests on combined channel/angle details are described. The main behavioural patterns are identified and their effect on the connection performance is discussed. Subsequently, detailed three-dimensional finite element models are constructed by means of the nonlinear program ABAQUS, and their results are validated against the experimental data. Good agreement between experimental and numerical results is obtained. (a) Schematic representation of test set-up. 2 TESTING ARRANGMENT AND SPECIMEN DETAILS 2.1 Experimental set-up The set-up used for testing the beam-to-tubular column connections in shear is shown in Figure 1. A column section of 8 mm height was used in all tests. The column was fixed at both ends by welding it to a 1 mm plate clamped at the base and bolting it to a lateral bracing at the top, as depicted in Figure 1. A hydraulic actuator operating in displacement control was used to apply vertical deformations at the head of the bolts connecting the beam and top angle. The force imposed by the actuator was transmitted through a heavy-section I-beam and a bearing plate of 3 mm thickness in order to achieve an even distribution of stresses over the head of the top angle bolts, as shown in Figure 1. In the case of the web angle connections, a 3 mm plate was placed on the top beam flange in line with the bolts connecting the beam web side of the web angles. The displacement at the head of the top angle bolts (for top and seat angle connections) and at the top beam flange (in the case of web angle connections) was gradually increased up to a displacement corresponding to the connection failure or until the capacity of the actuator was reached (at around 6 kn). The applied vertical displacement and corresponding vertical force were recorded by the load cell and transducer incorporated within the actuator. A more detailed description of the test set-up can be found in Liu (212). (b) General view of test arrangement. Figure 1. Test set-up of bolted connections (dimensions in mm). 2.2 Specimen details A total of six open beam-to-tubular column angle connections were studied: three blind-bolted connections and three reverse channel details. A summary of the test series is given in Table 1, which includes the geometric details of the connection as well as the column and beam sizes. Figure 2 depicts the connection configurations studied (Type A used in Hollobolted specimens, and Types B and C utilized in reverse channel configurations). The reference employed for the specimens is expressed by Pt-Nx-Ay-T for top and seat angle specimens, whereas Pt-Nx-Ay- W is utilized for specimens incorporating web angle components. In this nomenclature, P represents the connection type (B stands for the blind-bolted angle connection and C stands for combined channel/angle connection), t is the thickness of the column or channel in mm, x is the thickness of angle in mm, and y expresses the gauge distance between the centre of the bolt hole in the horizontal angle leg and column flange in mm (i.e. the value of a as indicated in Figure 2).

3 Table 1. Summary of the test specimens. Reference Type Angle Blind-bolted angle connections Reverse channel component Dimension (mm) (as shown in Figure 2) a b c d e f g h i j k B1-N8-A45-T A L B1-N8-A5-T A L B1-N15-A5-T A L Combined channel/angle connections C1-N8-A45-T B L 75 8 C6.3-N15-A5-T B L C1-N8-A5-W C L SHS SHS SHS (a) Type A. It is important to note that Square Hollow Sections SHS 15x15x1 were used as columns for all blind-bolted specimens whereas SHS xx1 columns were employed in combined channel/angle details. Also, Universal Beams UB 35x12x25 and Grade 1.9 M16 standard bolts were employed in all cases. The grade of the steel used for beams, columns and angles was S275. Table 2 presents the mean yield stress values and ultimate strength for the angle, beam and column components as obtained from at least three coupon tests in each case. (b) Type B. Table 2: Material properties of connection elements. Yield stress Ultimate stress Element (N/mm 2 ) (N/mm 2 ) Beam UB Column SHS Column SHS Angle L Angle L Hollo-bolt sleeve* *Obtained from the mean of three hardness tests (c) Type C. Figure 2. Connection configurations. Preliminary numerical studies carried out as part of the present research (Liu 212) have shown that the beam web stiffness plays a vital role in the shear performance of the connection. Buckling of the beam web can prevent the connection from reaching its maximum shear capacity and initiate complex joint interactions. Therefore, web stiffeners were used in order to limit undesirable beam failure modes, as shown in Figure 2. This enabled a better understanding of the basic mechanisms of bolt, angle and column face failure under shear loading. To this end, the beam web was stiffened in top and seat angle connections (Figure 2a and Figure 2b) by means of four 1 mm steel plates (two on each side) welded perpendicular to the web along the full beam

4 depth. Similarly, in the case of reverse channel connections with web angles (Specimen C1-A8-A5- W), the beam web was thickened by welding 1 mm-thick plates at each side as depicted in Figure 2c. Following the recommendations of Elghazouli et al. (9), Grade 1.9 M16 Hollo-bolts were used for the blind-bolted configurations. Also, as suggested by Málaga-Chuquitaype & Elghazouli (21), the reverse channel components used for combined channel/angle connections were obtained from SHS by longitudinal cutting. Table 1 summarizes the dimensions of the SHS from which the reverse channels were obtained. Fillet welding with thickness 1 mm was used to connect the column and the reverse channel throughout the length of channel component. 3 EXPERIMENTAL RESULTS AND OBSERVATIONS 3.1 Blind-bolted angle connections Figure 3a presents the shear force-displacement relationships obtained for the three blind-bolted angle connection specimens studied. The influence of the angle horizontal gauge distance (distance a in Figure 2) can be examined by comparing the results of specimens B1-N8-A45-T and B1-N8-A5-T with a = 45 mm and a = 5 mm, respectively. It can be observed from Figure 3a that the initial stiffness of Specimen B1-N8-A45-T is nearly 25% larger than that of Specimen B1-N8-A5-T due to the stiffer angle component (shorter horizontal gauge distance). However, after yielding occurs, at a load of around 122 kn, the stiffer vertical leg of Specimen B1-N15-A5-T (i.e. shorter distance d in Figure 2 and Table 1) seems to offset these initial stiffening effects. This counter-balancing effect is particularly evident for displacements greater than 6 mm. The results of Specimens B1-N8-A5-T and B1-N15-A5-T in Figure 3a, where the angle was varied from L 75 8 in the former to L in the latter, can be compared to illustrate the influence of angel thickness. As expected, the thickness of the angle has a direct effect on the connection response, including the stiffness, capacity, and deformation modes. The initial stiffness of Specimen B1-N15-A5-T with the thicker angle is around 6% larger than that of Specimen B1-N8- A5-T, while its capacity at a displacement of 8 mm is also kn higher than the corresponding capacity of Specimen B1-N8-A5-T. 3.2 Combined channel/angle connections Figure 3b presents the shear force-displacement relationships for the three combined channel/angle connection specimens. The effects of the angle thickness and reverse channel thickness on the shear response of combined channel/angle connections can be explored by comparing the experimental results of Specimens C1-N8-A45-T and C6.3-N15-A5-T. It can be observed from Figure 3b that the stiffness of Specimen C1-N8-A45-T is 37% higher than that of Specimen C6.3-N15-A5-T. This lower initial stiffness of Specimen C6.3-N15-A5-T relative to Specimen C1-N8-A45-T is attributed to the accumulation of bearing deformation in the relatively thinner channel as well as the increased horizontal gauge distance (dimension a in Figure 2). It is important to note that the shear force-displacement response of Specimens C1-N8-A45-T and C6.3-N15- A5-T are similar despite the significant differences in their deformation modes, with the deformation of the reverse channel-bolt assemblage dominating the response of Specimen C6.3-N15-A5-T. Additionally, bolt slip occurred almost instantaneously at a load level of kn in Specimen C1-N8-A45-T whereas a more gradual slip was observed in Specimen C6.3-N15-A5-T. Both specimens failed at a displacement of around 11 mm due to shear fracture of the bolts connecting the bottom angle and the channel components The response of double web angle connections under shear was studied with reference to Specimen C1-N8-A45-T. As illustrated in Figure 3b, the web angle configuration offers significantly higher stiffness and post-yield capacity when compared with top and seat angle details. 3.3 Ductility The experimental results show that blind-bolted angle connections provided significant shear-ductility levels exceeding 16 mm of vertical displacement without significant deterioration in stiffness or capacity. This high ductility levels can be attributed to the Hollo-bolt insert. Indeed, higher ductility has been documented for Hollo-bolts in comparison to standard bolts when subjected to single and double shear loading (Liu 212). Conversely, the combined channel/angle connection Specimens C1-N8-A45-T and C6.3-N15-A5-T failed by shear fracture of the standard bolts at around 11 mm of vertical displacement. The results also suggest that the addition of a flexible reverse channel component can enhance the ductility of the connections at the expense of some reduction in stiffness and capacity. Also, the ductility capacity of connections incorporating web angle/channel components (i.e. Specimen C1-N8- A5-W) can be improved due to the increased number of bolts in shear.

5 6 5 4 B1-N8-A45-T B1-N8-A5-T B1-N15-A5-T (a) Blind-bolted angle connections. 6 (a) Connection. (b) Angle. 5 4 C1-N8-A45-T C6.3-N15-A5-T C1-N8-A5-W (b) Combined channel/angle connections. Figure 3. Shear force-displacement relationships. 4 NUMERICAL SIMULATION 4.1 Modelling details Bolt in shear Bolt in shear Three-dimensional (3D) finite element models were developed using the FE software ABAQUS 6.7 (3). These models make use of eight-node brick solid elements of Type C3D8I, as shown in Figure 4. Special attention was given to the faithful representation of the geometric and mechanical characteristics of the bolts including the shank, sleeve, head and nut, as illustrated in Figure 4. The stress-strain relationships for the material of all the connection components were defined by a tri-linear kinematic hardening rule with an elastic modulus of 21 GPa and Poisson s ratio.3. The models consider the experimentally obtained yield stress and ultimate strength values for the angle, beam and column components, as illustrated in Table 2. (c) Standard bolt. (d) Hollo-bolt. Figure 4. FE representation of connection components. The contact phenomena between each pair of interacting surfaces was taken into account by defining hard and friction surface interaction properties. In the tangential direction, a friction coefficient of.3 was defined according to previous experimental studies, while in the normal direction, hard contact pressure-over closure relationship was employed, but the separation was allowed. The more flexible surface was chosen as the slave surface, while the more rigid area was assigned as master. Moreover, slippage between the bolt and hole surfaces was considered by means of the contact definition. Bolt pretension in standard and Hollo-bolts was introduced by means of two loading steps before external loading. Firstly, pretension forces of 11 kn were applied. The second step involved removing the applied pretension while simultaneously fixing the bolt length as its deformed value. The boundary conditions and loading in the numerical analyses followed the conditions and loading methods employed during the tests and described previously. A number of mesh sensitivity studies were carried out in order to arrive at an optimum representation which involves a comparatively finer mesh for the angles, bolts and the contact areas within the beams and columns, whereas a relatively coarser mesh was employed elsewhere. The dimensions of the adopted mesh ranged between 6 mm within the refined region, and up to mm within the coarser region.

6 4.2 Validation Specimens B1-N8-A45-T, C1-N8-A45-T and C1-N8-A5-W (web angle) are used herein as typical examples for validation purposes. The shear force-displacement relationships, obtained from the experimental results as well as FE simulation, are compared in Figure 5. It can be noted from Figure 5 that reasonable estimates of stiffness and capacity are obtained in all cases. The differences between the experimental results and numerical simulations observed (between.2 mm and 1.2 mm of vertical displacement) for Specimen B1-N8-A45-T in Figure 5a can be attributed to bolt slippage. Importantly, in the FE simulation, slippage of the bolt occurs almost instantaneously once the friction forces are overcome, whereas a more gradual slip displacement was observed during the test. Similarly, good agreement can be observed between the capacity estimations of the FE model and the experimental values within a difference of ±1%. On the other hand, the overestimation of postyield stiffness for Specimen C1-N8-A45-T in Figure 5b is attributed to differences in the evolution of bolt slippage displacement between the FE models and the experimental results. The same is true for Specimen C1-N8-A5-W where the multiple slip paths in the web angle components initiated distinguishable short slip-only deformations at load levels of and 18 kn approximately, while a more gradual accumulation of deformation was observed during the test. Figure 6 illustrates the deformation pattern observed in the top angle of Specimen B1-N8-A45-T at a vertical displacement of 16 mm and compares it with the FE model prediction at the same displacement level. It is evident from this figure that the deformation and plastic mechanism are replicated by the proposed FE model. Despite some differences, the FE estimations were found to correlate well with experimental results in the small as well as large levels of displacement demands. This good agreement with the experimental results confirms that the detailed FE models are able to simulate the shear response of bolted angle connections with reasonable accuracy Experiment Numerical model (a) Specimen B1-N8-A45-T Experiment Numerical model (b) Specimen C1-N8-A45-T Experiment Numerical model (c) Specimen C1-N8-A5-W. Figure 5. Comparison of experimental and numerical shear force-displacement response.

7 (a) Experiment mode. (b) Numerical mode. Figure 6. FE representation of connection components. 5 CONCLUDING REMARKS This paper has examined the shear behaviour of bolted angle connections between tubular columns and open beams. An experimental programme comprising three shear tests on blind-bolted angle connections and three tests on combined channel/angle connections has been described. A discussion on the salient response characteristics such as stiffness, capacity and failure mechanism has been presented. From the experimental results, it can be concluded that the inelastic mechanisms exhibited by the connections examined herein under shear loads primarily involve the interaction between the angle components and the column/channel face assemblage. As expected, it has been shown that the angle horizontal gauge distance between the bolt centre and the column face has a direct influence on the initial stiffness with only minor influence on the shear capacity. Importantly, the thickness of the angle is directly proportional to the connection capacity and initial stiffness. Similarly, the thickness of the reverse channel stands in direct relationship to the connection capacity and stiffness. Besides, it was observed that the overall resistance and stiffness of the connections with double web angels can be significantly increased in comparison with top and seat angle connections. In terms of ductility, blind-bolted angle connections can provide significant shear-ductility levels exceeding 18 mm of vertical displacement without signs of failure. Conversely, the combined channel/angle connection specimens with top and seat angles exhibited less ductility, failing at around 11 mm of vertical displacement. The results also show that the addition of a flexible reverse channel component may enhance the ductility of the connections at the expense of some reduction in stiffness and capacity. In the case of combined channel/angle connections under shear loads, the ductility development is typically limited by the brittle fracture of the bolts in shear rather than the failure of the other constituent components. Finally, the addition of web angle components was observed to delay the shear failure of the connection due to the increased redundancy offered by the higher number of standard bolts resisting the shear action. Continuum FE models were proposed which include a number of advanced modelling features such as loading reversal, contact phenomena, bolt slippage definition and bolt pretension application. Models were validated against available experimental results and were found to accurately and efficiently simulate the overall behaviour of bolted angle connections. In particular, the stiffness, load capacity and ultimate failure modes are all very well predicted. Moreover, the detailed characteristics of the models render the observation of key response parameters relatively straightforward, including the stress distribution and deformation patterns. 6 ACKNOWLEDGEMENTS The support of Tata Steel Tubes, particularly that of Mr T. Mustard, for the research described in this paper is gratefully acknowledged. The authors would also like to thank the technical staff of the Structures Laboratories at Imperial College London, especially Mr T. Stickland, for their assistance with the experimental work. Additionally, the first author would like to acknowledge the grant provided by The Chinese Scholarship Council and The UK Department for Innovation, Universities & Skills, for her doctoral research studies through a UK/China Scholarship for Excellence. REFERENCES ABAQUS 3. ABAQUS Theory Manual,Version 6.7, Hibbit, Karlsson and Sorensen Inc. ANSYS 3. ANSYS Multiphisics 1., Canonsbury, Pennsylvania Inc. Banks, G Flowdrilling for tubular structures. Proceedings of the Fifth International Symposium on Tubular Structures. UK. Barnett, T., W. Tizani & D. Nethercot 1. The practice of blind bolting connections to structural hollow

8 sections: A review. Steel and Composite Structures 1(1): Cao, J., J. Packer & N. Kosteski Design guidelines for longitudinal plate to HSS connections. Journal of Structural Engineering 124(7): CEN 5. EN , Eurocode 3: Design of steel structures, Part 1-8: Design of joints. Citipitioglu, A. M., R. M. Haj-Ali & D. W. White 2. Refined 3D finite element modeling of partiallyrestrained connections including slip. Journal of Constructional Steel Research 58(5-8): Elghazouli, A. Y., C. Málaga-Chuquitaype, J. M. Castro & A. H. Orton 9. Experimental monotonic and cyclic behaviour of blind-bolted angle connections. Engineering Structures 31(11): Huck International Inc Industrial fastening system. Arizona (USA). Kishi, N., A. Ahmed, N. Yabuki & W. F. Chen 1. Nonlinear finite element analysis of top and seat-angle with double web-angle connections. Structural Engineering and Mechanics 12(2): Kosteski, N. & J. A. Packer 3. Longitudinal plate and through plate-to-hollow structural section welded connections. Journal of Structural Engineering 129(4): Lindapter International Ltd Type HB hollo-bolt for blind connection to structural steel and structural tubes. UK, Lindapter International Ltd. Liu, Y Behaviour of steel connections under extreme loading conditions, PhD Thesis, Department of Civil and Environmental Engineering, Imperial College London, UK. Málaga-Chuquitaype, C. & A. Y. Elghazouli 21. Behaviour of combined channel/angle connections to tubular columns under monotonic and cyclic loading. Engineering Structures 32(6): Pirmoz, A., A. S. Daryan, A. Mazaheri & H. E. Darbandi 8. Behavior of bolted angle connections subjected to combined shear force and moment. Journal of Constructional Steel Research 64(4): Wang, Z. Y., W. Tizani & Q. Y. Wang 21. Strength and initial stiffness of a blind-bolt connection based on the T-stub model. Engineering Structures 32(9):