Behaviour and Design of Connections for Demountable Steel and Composite Structures

ISTRUC-00122; No of Pages 12 Structures xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Structures journal homepage: www.elsevier.com/locate/structures

Behaviour and Design of Connections for Demountable Steel and Composite Structures Brian Uy a,⁎, Vipulkumar Patel a,b, Dongxu Li a, Farhad Aslani a,c a b c

Centre for Infrastructure Engineering and Safety, School of Civil and Environmental Engineering, The University of New South Wales, Sydney, NSW, 2052, Australia School of Engineering and Mathematical Sciences, College of Science, Health and Engineering, La Trobe University, PO Box 199, Bendigo, VIC 3552, Australia School of Civil, Environmental and Mining Engineering, The University of Western Australia, Crawley, WA 6009, Australia

a r t i c l e

i n f o

Article history: Received 4 April 2016 Received in revised form 14 June 2016 Accepted 16 June 2016 Available online xxxx Keywords: Composite structures Demountability Steel structures Structural design

a b s t r a c t This paper provides an overview of demountable connections and elements for steel and composite structures. Demountable connections are generally assembled from prefabricated structural members which can be reused after dismantling. This paper investigates innovative connections which enable steel and concrete framed structures to be made demountable. Demountable connections considered in this paper include beam-beam connections, column-column connections and beam-slab connectors. The finite element models for the nonlinear inelastic analysis of coped beams, column-column connections and composite beams are developed based on nonlinear finite element analysis techniques using the commercial finite element program, ABAQUS. The behaviours for coped steel beams and composite beams are then compared with the corresponding experimental results. The finite element models are used to investigate the effects of material and geometric properties on the behaviour of demountable connections. The finite element models can efficiently predict the plastic damage in the coped steel beam used in beam-beam connections. It is demonstrated that the coped beam can be made demountable up to a load of about 50% of the ultimate load which is greater than typical service loads. The computational results indicate that increasing the sleeve length and reinforcement ratio increases the stiffness and ultimate tensile strength of demountable column-column connections. © 2016 The Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.

1. Introduction The concepts of demountable buildings and prefabricated construction can be traced back two millennia to the bible [1]. More recently in Australia, the Green Building Council of Australia has provided guidelines for best practice in the use of structural steel and concrete in construction projects. Furthermore, more recent discussion has centred on construction and demolition waste where demountable buildings which allow for the reuse of materials. Current estimates in Australia have determined that approximately 40% of landfill waste can be directly attributed to building and construction. Methods for lowering this rate can be achieved through changes in construction materials, methods of construction and demolition. This paper presents the concept of innovative connectors between steel and concrete elements that allow structures to be made demountable. Current Australian practice in steel building construction encourage steps that structural designers can take to maximise the potential for re-using steel buildings including using bolted connections in preference to welded joints and ensuring easy access to connections [2]. Recent examples of this concept include the Olympic Stadium project in Sydney completed in 2000 (Fig. 1). The end stands of this stadium were made ⁎ Corresponding author. E-mail address: [email protected] (B. Uy).

demountable using innovative blind bolts and the structural steel was then reused to upgrade the Wollongong Stadium 80 km south of Sydney. In addition to the economic and environmental benefits that are promoted by demountable buildings, the proper design of connections that ensures buildings are able to be systematically disassembled also promotes safety in design. The concept of demountable structures changes the traditional construction technology paradigm which often does not consider the reuse of structural members at the end of their service life. Demountable structures are generally able to be easily erected and dismantled, and are capable of adaption to different situations. Structural steel is one of the most promising materials for allowing structures to be made demountable. Bolted connections can also be utilised to promote demountability in commercial composite buildings. The erection procedures can be used in the reverse order for the efficient dismantling of demountable structures. For example, temporary supports used to erect the structure can be utilised for dismantling the structures. This procedure may prevent structural members from being bent, distorted or overstressed during dismantling. The deformation in a structural member may occur during its service life. The deformed members can be clearly marked for identification after dismantling a structure. These members could be repaired before reusing them. Demountable structures are generally limited to short-term use such as travelling shows, carnival structures and school classrooms. However, they have

http://dx.doi.org/10.1016/j.istruc.2016.06.005 2352-0124/© 2016 The Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.

Please cite this article as: Uy B, et al, Behaviour and Design of Connections for Demountable Steel and Composite Structures, Structures (2016), http://dx.doi.org/10.1016/j.istruc.2016.06.005

2

B. Uy et al. / Structures xxx (2016) xxx–xxx

Fig. 1. Olympic stadium Sydney, Australia.

not been implemented in larger structures for long-term use. This paper presents the details of specific connections which may permit steel and steel-concrete framed structures to be made demountable. One of the major practical/technical impediments of the reuse of steel is the ability to render structures demountable and thus this paper highlights the development of innovative connectors for both beams and columns that promotes the concepts of demountability. Therefore, this paper explores the behaviour of bolted-connectors for composite beams and columns which allow structures to be made demountable. Demountable connections considered in this paper include the following: • Beam-beam connections, • Column-column connections; and • Beam-slab connectors.

Finite element analysis is conducted to compare demountable structural elements with traditional structural elements. The deterioration rate of structural steels over their lifespan should be considered in the structural design for reusing the components. New structures with reused steel could potentially be unsafe without accounting for the deterioration rate into the project design. Advanced structural health monitoring methods could provide information about the strains and deformations over a demountable structure's life.

Fig. 2. Beam-beam connection with web-side plates.

requirement for the design model of the steel connections was specified in Clause 9.1.3 of AS4100-1998 [3]. The design method, capacity tables and detailing parameters for the web side plate beam-beam connections commonly used in Australia are given by Hogan and Munter [8]. This paper extends the design guidelines developed by Hogan and Munter [8] to predict the demountable behaviour of coped steel beams used in steel beam-beam connections. When the coped beams are restrained by lateral torsional buckling, the following geometric conditions must be satisfied for the single web coped beam [8]: dc ¼ ≤ 0:5D

ð1Þ

D 900 8 for ≤ qffiffiffiffiffiffiffiffi D > > t wb < 730  106 D f yw c rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  3ffi 3 D 900 > > D N qffiffiffiffiffiffiffiffi for : f yw t wb t wb f

ð2Þ

yw

in which D denotes the beam depth, dc represents the cope depth, c is the cope length, twb is the thickness of the secondary beam web and fyw denotes the yield strength of the secondary beam web.

2. Beam-beam connections 2.2. Finite element analysis 2.1. Basic concept The web-side plates are used for connecting the primary beams to secondary beams in the structural steel framed connections. Fig. 2 depicts a beam-beam connection using web side plates. The web-side plates are either cut from Grade 250 plate. The steel beam is generally cut and the holes are drilled in the coped region. These manufacturing operations should comply with the limitations of maximum tolerance given in Section 14 of Australian Standard AS 4100-1998 [3], Eurocode 3 [4], Canadian Standard CSA-S16-09 [5], American Standard AISCLRFD [6], and Japanese Standard AIJ [7]. The bolt hole diameter is maintained 2 mm larger than the nominal bolt diameter. The bolts, M20 or M24, can be utilised in the web-side plate beam-beam connection with a minimum of 2 bolt rows to a maximum of 9 bolt rows and a maximum of 2 columns of bolts. The web-side plates are welded to the primary beam web with 6 mm or 8 mm fillet welds including the minimum fillet weld size of 0.75ti, in which ti represents the web-side plate thickness. The fillet weld should be continued on both sides of the plate but not continued across the top and bottom of the web side plate. A minimum 20 mm clearance is allowed between the primary steel beams and coped secondary steel beams as shown in Fig. 2. The

The finite element program, ABAQUS [9], was employed to study the demountable behaviour of coped steel beams used in the web side plate beam-beam connections. The object oriented Python script was developed for the finite element analysis. The developed Python script was utilised to automate the simulation including the creation of the finite element model, assigning of materials, creation of analysis procedure, application of loads, assigning of boundary conditions and collection of the results from the final analysis. The user-defined Python script incrementally generates many finite element models for the parametric study [10]. Steel beams are generally not perfectly straight but have small initial geometric imperfections, which are induced during the manufacturing and construction processes. These initial geometric imperfections may reduce the stiffness and ultimate strengths of coped steel beams. The initial geometric imperfections are generally assigned by perturbations in the steel beam. An imperfection was defined as the linear superposition of buckling modes which are used as input in the subsequent static analysis to assign an imperfection in the coped steel beam by adding the predicted buckling modes to the real geometry. The first buckling mode, as shown in Fig. 3, was assumed to define the model critical

Please cite this article as: Uy B, et al, Behaviour and Design of Connections for Demountable Steel and Composite Structures, Structures (2016), http://dx.doi.org/10.1016/j.istruc.2016.06.005

B. Uy et al. / Structures xxx (2016) xxx–xxx

3

Fig. 5. A typical stress-strain relationship for structural steels. Fig. 3. First buckling mode obtained from the eigenvalue buckling analysis.

imperfection. The models created for both analyses, eigenvalue buckling analysis and static analysis, should be identically defined in terms of the geometry and mesh size. The maximum initial geometric imperfection used in the nonlinear analysis was assumed as 1 mm which has been suggested by Yam et al. [11]. Fig. 4 depicts a finite element model for coped steel beams used in demountable beam-beam connections. The applied load was modelled by selecting a concentric force in the analysis. The lateral deflection of the coped beam was restrained at three positions as shown in Fig. 4. Material and nonlinear analyses on coped steel beams incorporating the initial geometric imperfections were conducted. A four-node shell element (S4R) with reduced integration was utilised in the present study. A 20 × 20 mm mesh was considered in the nonlinear analyses and it was judged to be economic to yield an accurate simulation for the use in the structural analysis. 2.2.1. Stress-strain relationship for structural steel In the present study, a typical tri-linear stress-strain curve was utilised for simulating the material behaviour of structural steel in tension and compression as shown in Fig. 5. The three stage material constitutive model developed from tension coupon tests assumes that the structural steel follows the same stress-strain curve in compression and tension. The same stress-strain curve in tension and compression was incorporated in the current finite element model. In Fig. 5, σs represents the steel stress, εs denotes the steel strain, fy is the yield strength of the steel, εy represents the yield strain of steel, εt is the strain hardening, εu is the ultimate strain and fu is the ultimate tensile strength of steel material. The steel hardening strain (εt) is taken as 0.005 in the finite element model.

Fig. 4. Finite element model of coped steel beams for demountable beam-beam connections.

2.2.2. Verification of the finite element model for coped steel beams The accuracy of the developed finite element model is examined by comparing the predicted results with existing experimental results. The ultimate strengths and applied load-deflection curves of the coped steel beam were considered in the comparative study. The material properties and geometry of the coped steel beams tested by Yam et al. [11] are given in Table 1. The cope depth-to-beam depth (dc/D) ratios varied between 0.05 and 0.3. Specimens adopted a hot-rolled beam section of UB406 × 140 × 39. The yield strength of the steel beams was 313 MPa. The steel beam web end is welded perpendicular to the plate which was connected to the wall. This single end plate was modelled by applying fixed boundary conditions at the beam web. The boundary conditions minimised the plates' in-plane rotational stiffness for the simulation of a simply supported boundary condition at the coped beam end. The steel beam was laterally restrained by bracing at the applied load, reaction and coped end close to the compression flange. Experimental studies conducted by Yam et al. [11] indicated that the steel beams possess initial geometric imperfections although these were not measured. However, Yam et al. [11] considered the effects of the initial geometric imperfections in their finite element analysis. The initial geometric imperfection of 1 mm, 0.25 mm and 0.2 mm was used for specimens 406d005, 406d01 and 406d03, respectively. The Python script developed was utilised to determine the ultimate strengths of the tested steel beams. The test results reported that the coped steel beams failed by local web buckling at the coped region (Fig. 6). The test results also indicated that increasing the cope depth-to-beam depth (dc/D) ratios decreases the ultimate strength of the coped steel beam. This is mainly due to the fact that a coped region with a larger cope depth-to-beam depth (dc/D) ratio has a lower area in the steel web, undergoes local buckling earlier, which decreases the ultimate strength of the coped steel beam. The experimental and computational ultimate strengths of coped steel beams are given in Table 1, in which Pexp denotes the experimental ultimate strength and PFEM represents the ultimate strength determined by the finite element model. It appears from Table 1 that the computed ultimate strength of the tested steel beams agrees well with the corresponding experimental results. The mean ultimate strength obtained from the finite element model is 97% of the experimental ultimate strength with a coefficient of variation of 0.02 and a standard deviation of 0.02. The applied load-deflection curves for the coped steel beams obtained from the finite element model were compared with experimental responses given by Yam et al. [11]. Fig. 7 depicts the applied loaddeflection responses for specimen 406d005 obtained from the tests conducted by Yam et al. [11] and predicted by the finite element model. It can be seen from Fig. 7 that the applied load-deflection response predicted by the finite element model agrees reasonably well with the experimental data. The initial stiffness predicted by the finite element model is slightly higher than the initial stiffness obtained from the experiments up to a loading level of 150 kN. This is due to

Please cite this article as: Uy B, et al, Behaviour and Design of Connections for Demountable Steel and Composite Structures, Structures (2016), http://dx.doi.org/10.1016/j.istruc.2016.06.005

4

B. Uy et al. / Structures xxx (2016) xxx–xxx

Table 1 Ultimate loads of coped beams for demountable beam-beam connections. Specimens

D (mm)

c (mm)

dc (mm)

dc/D

fy (MPa)

E (GPa)

PFEM (kN)

PFEM (kN)

P FEM P ;exp

Ref.

406d005 406d01 406d03 Mean Standard deviation (SD) Coefficient of variation (COV)

398 398 398

342.9 342.9 342.9

19.9 39.8 119.4

0.05 0.1 0.3

343 343 343

216.6 216.6 216.6

248.8 241.6 167.9

243.5 240.0 160.3

0.98 0.99 0.95 0.97 0.02 0.02

[11]

the uncertainty of the steel material properties as the average steel strength was utilised in the finite element model. The computational and experimental applied load-deflection responses for specimen 406d01 tested by Yam et al. [11] are shown in Fig. 8. It can be observed from Fig. 8 that the initial stiffness determined by the finite element model is almost the same as the initial stiffness obtained from the tests up to the ultimate strength. After attaining the ultimate strength, the stiffness of the tested specimen deviates from the computational predictions. Fig. 9 illustrates a comparison between the computational and experimental applied load-deflection responses for specimen 406d03. The figure shows that the initial stiffness of the coped beam predicted by the finite element model agrees reasonably well with the experimental results. The computational prediction in the post-peak range of the applied load-deflection curves is higher than the experimental results. The applied load-lateral deflection curve obtained from the finite element model for specimen 406d03 tested by Yam et al. [11] is presented in Fig. 10. It appears that the finite element model provides a reasonable prediction for the experimental applied load-deflection curves. The initial stiffness of the coped beam is predicted to be slightly higher than the tested beams. This is likely due to the uncertainty of the actual steel material properties as the average steel yield strength was utilised in the finite element model. The comparative study demonstrates that the finite element model accurately predicts the ultimate strength, applied loaddeflection curves and applied load-lateral deflection curves for the coped steel beams. 2.2.3. Parametric study for coped steel beams The behaviour of a coped steel beam can be characterised by the vertical load-deflection curve. A parametric study was conducted to examine the effects of the cope depth-to-beam depth ratio, aspect ratio and span length-to-beam depth ratio on the behaviour of coped steel beams utilised in web-side plate beam-beam connections. The initial geometric imperfection of the coped steel beams was assumed as

Fig. 6. Failure modes of coped steel beams.

1.0 mm in the following parametric study [11]. Only one parameter was changed at a time to investigate its individual effect. 2.2.3.1. Effects of cope depth-to-beam depth ratio. The web local buckling of a coped steel beam depends on its cope depth-to-beam depth (dc/D) ratio. The finite element model was utilised to examine the effects of cope depth-to-beam depth (dc/D) ratio on the fundamental behaviour of coped steel beams. An Australian Universal steel beam with a crosssection of 610UB125 was considered [12]. The typical cope depth-tobeam depth (dc/D) ratios were considered by changing the cope depth of the steel beam while the beam depth was unchanged. The effects of the cope depth-to-beam depth (dc/D) ratio on the applied load-deflection curves for the coped steel beam are depicted in Fig. 11. It can be seen that by increasing the cope depth-to-beam depth (dc/D) ratio of the steel beam slightly decreases their initial stiffness. However, increasing the cope depth-to-beam depth (dc/D) ratio significantly decreases the ultimate strength of coped steel beams. This is due to the fact that a beam section with a larger cope depth-tobeam depth (dc/D) ratio has a smaller area in the beam web and it may be subject to premature local buckling which decreases the ultimate strength of the coped steel beam. This is also due to the fact that the coped region reduces the bending and torsional stiffness. A coped corner in the steel beams exhibits a high stress concentration from the geometric discontinuity which leads to premature local buckling. Fig. 12 presents the effects of cope details for the steel beams on the normalized lateral deflection. Each curve in Fig. 12 was normalized to the maximum deflection at the mid-length where the load is applied on the steel beams. The cope depth-to-beam depth ratio dc/D = 0 was used in the finite element model to predict the lateral deflection of an uncoped steel beam along its length. It can be observed from Fig. 12 that the lateral deflection of a coped steel beam along its length is higher than that of an uncoped steel beam.

Fig. 7. Comparison of predicted and experimental applied load-deflection curves for specimen 406d005.

Please cite this article as: Uy B, et al, Behaviour and Design of Connections for Demountable Steel and Composite Structures, Structures (2016), http://dx.doi.org/10.1016/j.istruc.2016.06.005

B. Uy et al. / Structures xxx (2016) xxx–xxx

5

Fig. 8. Comparison of predicted and experimental applied load-deflection curves for specimen 406d01.

Fig. 10. Comparison of predicted and experimental applied load-lateral deflection curves for specimen 406d03.

2.2.3.2. Effects of aspect ratio. The local buckling strength of the coped steel beam is affected by the aspect ratio (c/ho), in which ho represents the cope height (Fig. 12). To investigate the effects of aspect ratio on the behaviour of the coped steel beam, an Australian Universal steel beam with a cross-section of 610UB113 [12] was simulated using the finite element model. The aspect ratios (c/ho) were determined as 0, 1 and 2 by varying the cope length (c) while maintaining the same cope height (ho = 303.5 mm). Fig. 13 shows the applied axial load-deflection curves for the coped steel beams with various aspect ratio (c/ho). It illustrates that the ultimate strength of a steel beam is significantly affected by the aspect ratio (c/ho). It can also be seen that the initial stiffness of the steel beams reduces with an increase in the aspect ratio (c/ho). In addition, increasing aspect ratio (c/ho) significantly decreases its ultimate strength. This is attributed to the fact that an increase in the aspect ratio (c/ho) decreases the steel area at the cope region which significantly reduces the local buckling strength. Furthermore, the vertical deflection at the maximum load of the steel beam decreases with an increase in the aspect ratio (c/ho). The effects of the aspect ratio (c/ho) ranging from 0 to 2 on the ultimate applied loads of the steel beams are demonstrated in Fig. 14, in which Pu represents the ultimate applied loads and Poa

denotes the ultimate applied load for the steel beam without a cope (c/ho = 0). It would appear from Fig. 14 that increasing the aspect ratio (c/ho) between 0.5 and 2.0 causes a considerable reduction in the ultimate applied load of the steel beam. The ultimate applied load ratio is 0.93, 0.49 and 0.39 for the steel beams with an aspect ratio (c/ho) of 0.5, 1.5 and 2.0, respectively.

2.2.3.3. Effects of span length-to-beam depth ratio. The fundamental behaviour of a coped steel beam is influenced by the span length-tobeam depth ratio (L/D). An Australian Universal steel beam with a cross-section of 410UB53.7 [12] was adopted in the parametric study. The cope depth-to-beam depth ratio (dc/D) of the section was 0.1. The aspect ratio (c/ho) was taken as 0.6 in the finite element analysis. The effects of the span length was examined by changing the span length (L) while maintaining the same beam depth. The relationship between the ultimate applied load and span lengthto-beam depth ratio (L/D) ranging from 4 to 25 predicted by the finite element model is depicted in Fig. 15, in which Pos represents the ultimate applied load for the coped steel beam with a span length-tobeam depth ratio of 4. It can be seen from Fig. 15 that increasing the

Fig. 9. Comparison of predicted and experimental applied load-deflection curves for specimen 406d03.

Fig. 11. Effects of the cope depth-to-beam depth (dc/D) on the applied load-deflection curve for coped beams.

Please cite this article as: Uy B, et al, Behaviour and Design of Connections for Demountable Steel and Composite Structures, Structures (2016), http://dx.doi.org/10.1016/j.istruc.2016.06.005

6

B. Uy et al. / Structures xxx (2016) xxx–xxx

Fig. 14. Effects of the aspect ratio (c/ho) on the ultimate applied load.

Fig. 12. Effects of the coped details on the normalized lateral deflection for coped beams.

span length-to-beam depth ratio (L/D) significantly reduces the ultimate applied load. This implies that the ultimate strengths of the coped steel beam can significantly be improved by using shorter spans. The predicted ultimate applied load ratio for the span lengthto-beam depth ratio of 10, 20 and 25 is 0.8, 0.3 and 0.2, respectively.

coped beam is presented in Fig. 16. It can be observed from Fig. 16 that the coped beam can be demounted up to a load 125 kN where the PEEQ is predicted to be zero in the coped region, which is about 50% of the ultimate load and greater than typical service loads which are usually less than 40% of the ultimate load.

3. Column-column connections 3.1. Design

2.2.4. Equivalent plastic strain (PEEQ) The demountability of a beam-beam connection depends on the elasticity of the coped beam. The elasticity of a coped beam is characterised by the ability to sustain elastic deformation without undergoing significant plastic deformation. The demountability of a beam-beam connection cannot be achieved for the beam with large plastic deformation. The finite element analysis of a coped beam in demountable beam-beam connection was conducted for predicting the amount of plastic deformation in the coped region. It should be noted that the elastic deformation can be represented by zero equivalent plastic strain (PEEQ). The applied load-deflection curve for this

Fig. 13. Effects of the aspect ratio (c/ho) on the applied load-deflection curve for coped steel beams.

Fig. 17 presents the proposed column-column splice connections for concrete-filled steel tubular (CFST) columns. The proposed connection allows CFST columns to be connected at any location throughout the height. These CFST columns have a pre-welded sleeve plate extending past the base of the column. Blind bolts are utilised to connect the sleeve plate with CFST columns passing through pre-drilled holes. The wet concrete could be pumped from the drilled holes or another opening. Reinforcing bars are placed inside the concrete core to achieve additional compressive and tensile capacity. Furthermore, to ensure the discontinuity of the concrete and reinforcement and subsequently easily dismantling of this connection, a baseplate was welded to the bottom of the top tube. Bolts can be removed as the bolt head is arranged on

Fig. 15. Effects of span length-to-beam depth ratio L/D on the ultimate applied loads.

Please cite this article as: Uy B, et al, Behaviour and Design of Connections for Demountable Steel and Composite Structures, Structures (2016), http://dx.doi.org/10.1016/j.istruc.2016.06.005

B. Uy et al. / Structures xxx (2016) xxx–xxx

7

Fig. 18. Composite steel-concrete frame removing column under blast load.

where Nc represents the ultimate axial strength under axial compression and η reflects the factor which accounts for the accidental damage and it is expressed by η¼

Fig. 16. Applied load-deflection with demountability for coped steel beam.

the outside of the steel tube. The connection must resist the applied tension load and axial compression. This paper presents the design and analysis of demountable connections under axial tension. Some guidance on this connection under axial compression has been provided by Li et al. [13]. The column-column connection may be subjected to axial tension when a bottom column is removed after exhibiting accidental damage or through sever earthquake loading arising in load reversal on frames [14]. Fig. 18 illustrates a typical column-column connection in a composite steel-concrete frame. It can be seen from the figure that if the column is removed from the frame after experiencing the damage, this then reverses the loading on the connection from compressive to tensile. The design equation given in AS 1170-2002 [15] for determining the ultimate tensile strength of a column-column connection is expressed by. T c ¼ ηNc

ð3Þ

N1 No

ð4Þ

in which No denotes the axial compression before the accidental damage. An empirical equation for determining this axial compression (No) of a column without accidental damage is given in AS 1170-2002 [15] as follows: No ¼ 1:2G þ 1:5Q

ð5Þ

in which G represents the dead load in kPa and Q is taken as 3.0 kPa [15]. In Eq. (4), N1represents the axial compression after the column exhibits the accidental damage such as in a blast, fire or earthquake. The axial compression (N1) can be determined by the equation given in AS 1170-2002 [15] and is given by N1 ¼ G þ ψQ

ð6Þ

in which ψ represents the factor for live load after the accidental damage. The factor ψ is taken as 0.3 for cyclic loading and 0.4 for fire [15]. It is found that the reduction factor η given in Eq. (4) is determined as 0.51 and 0.54 for cyclic loading and fire, respectively. It should be noted that column-column connections are therefore required to resist a tensile load up to approximately half of the ultimate compressive strength of the column. The reinforcement is therefore embedded in the concrete to increase the ultimate tensile strength of the column-column connections. Fig. 17 depicts the reinforcement in the column-column connection. This reinforcement should not be welded to the base plate to allow the connection to be made demountable. This concept provides discontinuous reinforcement at the connections instead of continuous reinforcement found in most conventional connections in composite or concrete framed buildings. 3.2. Finite element model

Fig. 17. CFST column-column splice demountable connection.

The behaviour of column-column connections subjected to axial tension is affected by the sleeve plate length, bolt position and reinforcement ratio. The finite element model using ABAQUS [9] has been developed to investigate the effects of these important parameters on the axial load-strain curves of column-column connections. The following parameters were held constant in the present parametric study. A CFST column with a square cross-section of 250 × 250 mm was considered. The thickness of the steel tube was 4 mm so that its depth-tothickness ratio was 62.5. The yield and tensile strengths of the steel tube were 350 MPa and 430 MPa, respectively while its Young's modulus was 200 GPa. Normal strength concrete with a compressive strength of 32 MPa was filled into the steel tube. M20 blind bolts were utilised for

Please cite this article as: Uy B, et al, Behaviour and Design of Connections for Demountable Steel and Composite Structures, Structures (2016), http://dx.doi.org/10.1016/j.istruc.2016.06.005

8

B. Uy et al. / Structures xxx (2016) xxx–xxx

composing the connection. The yield and tensile strength of the reinforcement was 500 MPa and 690 MPa, respectively. The Young's modulus of the steel reinforcement was 200 GPa. The reinforcement ratio was taken as 0.8% in the analysis. The base plate was 5 mm thick whilst the sleeve plates were 10 mm thick. 3.2.1. Effects of sleeve length-to-connection length ratio The finite element model was employed to investigate the effects of sleeve length-to-connection length ratio (Ls/L) on the axial load-strain curves for column-column connections under axial tension. The sleeve length-to-connection length ratios of 0.2, 0.4 and 0.6 were considered by changing the sleeve length of the connection while maintaining the same connection length (L = 1500 mm). The column-column connections with different sleeve lengths are presented in Fig. 19. The effect of the sleeve length-to-connection length ratio on the axial load-strain curves for the column-column connection under axial tension is depicted in Fig. 20. It can be observed from Fig. 20 that both the initial axial stiffness and the ultimate tensile strength of the connection tend to increase with an increase in the sleeve length-to-connection length ratio. This is mainly due to the steel contribution in the connection increases as the sleeve length-to-connection ratio increases. The ultimate tensile strength is increased by 1% and 7%, respectively, when the sleeve length-to-connection length ratio increases from 0.2 to 0.4 and 0.8. 3.2.2. Effects of bolt position The finite element model developed was utilised to examine the effects of bolt position on the axial load-strain curves of column-column connections. Fig. 21 shows the column-column connections with two different bolt positions. In Fig. 21, e represents the bolt position from the middle of connection and L represents the total length of connection. The effects of bolt position on the axial load-strain curves of the connection are illustrated in Fig. 22. It can be observed from Fig. 22 that the initial stiffness of the connection with a centrally bolted position with e/(L/2) ratio of 0.10 is less than that of the connection with an end bolted position with e/(L/2) of 0.74. The ultimate tensile strength of the connection with a centrally bolted position is also lower than that of the connection with an end bolted position. 3.2.3. Effects of reinforcement ratio The effects of reinforcement ratio on the axial load-strain curves of the column-column connection with different reinforcement ratio (p) were studied by the finite element model. The reinforcement ratio (p) is defined as the ratio between the cross-sectional area of reinforcement As (clause 3.4.3.2 and 13.2.2 in AS 3600-2009) and cross-sectional area of a column Ag [16]. The minimum reinforcement ratio (p) is varied

Fig. 19. Column-column connections with different sleeve length (Ls).

Fig. 20. Effects of sleeve length-to-connection length ratio (Ls/L) on the axial load-strain curve for connection under axial tension.

between 0.01Ag and 0.04Ag which was specified in AS 3600-2009, Clause 10.7.1 [16]. The column-column connection shown in Fig. 19(a) was utilised to examine the effects of reinforcement ratio. Fig. 23 depicts the effects of reinforcement ratio (p) on the axial loadstrain curves of the column-column connections under axial tension. It can be seen from Fig. 23 that the initial axial stiffness of the demountable connections is slightly affected by the reinforcement ratio. The initial axial stiffness increases with an increase in the reinforcement ratio. However, the ultimate axial strength of the demountable connections increases with an increase in the reinforcement ratio (p). The ultimate tensile strength is increased by 9% and 21%, respectively, when the reinforcement ratio (p) in the demountable connection increases from 0.8% to 1.4% and 2.1%. 4. Beam-slab connectors 4.1. Basic concept Steel-concrete composite beams are heavily utilised for the construction of high rise steel residential buildings and bridges. The shear connectors are used for providing the interaction between the steel beam and concrete slab. The slip between the concrete slab and steel beam is prevented by the shear connectors which are generally welded to the top flange of a steel beam. The slabs can be characterised as solid slabs and composite slabs which use the steel sheeting. Composite slabs with profiled steel sheeting have a number of advantages over solid slabs. Rapid construction is obtained with the profiled steel sheeting which provides a working platform. This construction technique allows the structural erection to speed up from fabrication to installation by speeding up each process significantly. This technique also reduces the

Fig. 21. Column-column connection with different bolt position.

Please cite this article as: Uy B, et al, Behaviour and Design of Connections for Demountable Steel and Composite Structures, Structures (2016), http://dx.doi.org/10.1016/j.istruc.2016.06.005

B. Uy et al. / Structures xxx (2016) xxx–xxx

9

Fig. 24. Demountable composite beams with in situ concrete slab on metal decking.

Fig. 22. Effects of bolt position on the axial load-strain curves for column-column connection under axial tension.

constructional costs because it completely eliminates the need for formwork support frames for the concrete. While profiled composite slabs provide time benefits, they also provide a safer construction method. Different shear connectors such as bolted and welded studs can be utilised for composite beams. The use of bolted shear connectors instead of headed studs allows composite steel-concrete beams to be made demountable. A three-dimensional half model of a composite beam was developed for the numerical analysis using the ABAQUS finite element code [9,17]. The corresponding finite element model was also developed for composite beams with headed studs. The finite element analysis was conducted by utilising the ABAQUS/Implicit static general procedure. Material stress-strain relationships, surface-to-surface interactions, constraints and boundary conditions were assigned to accurately simulate the push test specimens.

The model incorporates a bolted connector with 20 mm diameter and 128 mm length. The profiled steel sheeting was 1 mm thick. Both eight node brick (C3D8R) and quadratic brick elements (C3D20R) were employed to mesh the shear connectors, slab and beam. A four node doubly curved thin shell element (S4R) was utilised to model the steel sheeting. The reinforcing bars were simulated with two-node linear three dimensional truss elements (T3D2). 4.3. Interactions, boundary conditions and loading Surface-to-surface contact between individual surfaces was created by defining the interaction relationship. The finite sliding technique was assumed for the interaction between the slab, bolt, sheeting and beam. It is also assumed that the surfaces do not penetrate so these surfaces are constrained by ‘Hard Contact’ in the normal direction. A penalty friction formulation was selected to model the tangential behaviour of the surface-to-surface contacts. A friction factor of 0.25 was used in the present finite element model. 4.4. Material constitutive models

4.2. Geometry, element type and mesh Steel concrete composite beams with profiled steel sheeting are illustrated in Fig. 24. A concrete slab was used with a dimension of 600 mm width, 150 mm depth and 600 mm length. Reinforcing bars of 12 mm diameter were utilised in the model. The composite beams consist of four bolts embedded in the concrete slab. Symmetry for the composite beam is assumed at the centre of the steel beam web along the longitudinal direction.

Fig. 23. Effects of reinforcement ratio on the axial load-strain curves for column-column connection under axial tension.

4.4.1. Structural steel The structural steel generally follows the same stress-strain curves under tension and compression. Pathirana [18] conducted tensile tests on coupons cut obtained from the web and flange of steel beams, reinforcing bars and shear connectors. Test results indicated that structural steel follows an elastic stress-strain relationship up to the yield strength which is followed by strain hardening before failure. The two-stage stress-strain curve for steel suggested by Pathirana [18] is employed in the finite element model. A two-stage linear stress-strain relationship for steel under compression is shown in Fig. 25.

Fig. 25. Stress-strain curve for profiled steel sheeting, bolts, steel beam and studs.

Please cite this article as: Uy B, et al, Behaviour and Design of Connections for Demountable Steel and Composite Structures, Structures (2016), http://dx.doi.org/10.1016/j.istruc.2016.06.005

10

B. Uy et al. / Structures xxx (2016) xxx–xxx

load-slip curves obtained from the finite element model are compared with experimental results given by Mirza et al. [20]. Fig. 27 shows the comparison of the experimental and finite element model applied load-slip curves for a composite beam. It can be seen from Fig. 27 that the applied load-slip curves obtained from the finite element model agree well with the experimental results. The initial stiffness of the applied load-slip curves is identical up to the ultimate load. However, the experimental curve deviates from the computational results after 2 mm slip. This is due to the fact that the microcracks observed in the concrete resulting the load reduction after 2 mm [21], while the finite element analysis does not consider the microcracks in modelling of concrete element. 4.6. Results and discussion

Fig. 26. Idealised stress-strain curves for concrete in composite beam.

4.4.2. Concrete The material behaviour of concrete in a composite beam is modelled as an elastic-plastic stress-strain relationship with strain softening. Fig. 26 presents the stress-strain curve for concrete in a composite beam. The concrete stress for the stress-strain curve is determined from the equations proposed by Carreira and Chu [19] as   εc ε0c σc ¼  γ γ−1 þ εε0c 0

f cγ

ð7Þ

c

in which σc represents the concrete compressive stress, εc is the concrete compressive strain, fc′ denotes the cylinder compressive strength of the concrete, εc′ is the strain corresponding to fc′ and γ is given by  0 3  f  γ ¼  c  þ 1:55 32:4

4.7. Parametric study ð8Þ

4.5. Verification The finite element model for the composite beam is verified by comparing the predicted results with experimental data. The applied

Fig. 27. Comparison of experimental and finite element results.

The typical failure mode of the concrete in the composite beam is depicted in Fig. 28. It can be observed from the figure that the bolt rotates and the inclination of the embedded nut pushes the concrete. This results in a prying out force which is responsible for concrete pull out of the AJAX bolt. Consequently, the failure mode is concrete failure as the removal of the bolt begins before yielding of the shank. Fig. 29 illustrates the comparison of the applied load-slip curves for welded stud and bolted connectors. It can be observed from the figure that a bolt connector exhibits lower stiffness at serviceability loads compared with a welded stud. The bolt connector initially experiences a slip of 1 mm at low loads because the oversized hole for easily demounting the composite beams. The bolt connector was simulated to sit on the steel flange with no pretension, however in practice after fastening, the bolt would experience some tensile forces which would result in a higher initial stiffness until full bearing in the hole is achieved. Once closure in the bolt-to-hole clearance has been achieved the stiffness of the bolt connection increases linearly up to a load level of 210 kN. The bolt exhibits nonlinear behaviour compared with the welded stud as shown in Fig. 29.

The performance of demountable composite beams is typically characterised by considering the slip between the concrete slab and the steel beam. The slip can be determined as the displacement at the steel beam and concrete slab interface [22]. The slip between the steel beam and the composite slab is due mainly to the local concrete crushing and shear connector bending [23]. The performance of demountable composite beams is affected by concrete compressive strengths, steel sheeting thickness and shear connector positions. It is expensive and time consuming to experimentally evaluate the effects of every parameter on the performance of demountable composite beams. In this case, the finite element analysis treatment provides a cost effective option for determining the performance of demountable composite beams. In the present parametric study, bolt connectors are inserted with a 1 mm clearance in the oversized holes. Therefore, the initial 1 mm slip can be observed before the bolt contact with the surface of the hole. 4.7.1. Effects of concrete compressive strengths The effects of concrete compressive strengths on the applied loadslip curves for demountable composite beams were studied. Fig. 30 illustrates the applied load-slip curves with different concrete compressive strengths. It can be observed from the figure that ultimate strengths of demountable composite beams increase as the concrete compressive strengths increase. The initial stiffness of the composite beams with profiled steel sheeting increases as the concrete compressive strength increases. Increasing the concrete compressive strength from 25 MPa to 32 MPa results in an increase of the ultimate strength by 20%.

Please cite this article as: Uy B, et al, Behaviour and Design of Connections for Demountable Steel and Composite Structures, Structures (2016), http://dx.doi.org/10.1016/j.istruc.2016.06.005

B. Uy et al. / Structures xxx (2016) xxx–xxx

11

Fig. 28. Demountable composite beam with bolted connectors.

4.7.2. Effects of profiled slab and solid slab The trapezoidal profiled steel sheeting are increasingly used in highrise composite buildings because they allow large spans without propping and plywood formwork. The components of the composite beam with trapezoidal profiled steel sheeting can easily be separated for subsequent reuse. This makes the composite beam demountable at the end of its service life. The stiffness, strength and ductility of the composite beams with profiled steel sheeting was studied experimentally and numerically by Mirza et al. [20,21]. The finite element model is employed to study the effects of the profiled slab and solid slab on the performance of composite beams. The effect of the profiled slab and solid slab on the applied load-slip curves for composite beams is depicted in Fig. 31. It can be observed that both initial stiffness and ultimate strength of the solid slab are higher than that of the profiled slab. The ductility of solid slab is also higher than that of the trapezoidal profiled slab. This is due to the solid slab failed in stud fracture while the failure mode of trapezoidal profiled slab is concrete failure [20].

1 mm slip is noted with the use of bolted shear connectors. This is because the oversized hole arrangement is initially provided to achieve demountability in composite beams. The initial stiffness (kint) of bolted and welded connectors are depicted in Fig. 29, in which the initial stiffness (kint) is determined as a slope of the load-slip response in the elastic range. The initial stiffness of the composite beams with bolted connectors is lower than that of the composite beams with welded connectors, which agree with the research findings of Ban et al. [24]. This is due to the movements of the bolted connectors in oversized holes provided in the steel beam flange. The bolt-to-hole clearance causes a slip plateau in the load-slip response of bolted connectors until the bolt stands against the hole. The stiffness of composite beams is sharply changed at a load level of 450 kN which indicates the initiation of the steel beam yielding. It can also be seen from Fig. 24 that the ratio of initial stiffness for bolted and headed connectors is 0.3.

4.7.3. Effects of shear connectors (headed vs bolted) The effect of shear connectors on the performance of demountable composite beams is examined by using the finite element model. Fig. 29 shows the effects of shear connectors on the applied load-slip curves for demountable composite beams with steel sheeting. It can be seen from Fig. 29 that the use of bolted shear connectors tends to increase the ultimate strengths of composite beams. This is due to the fact that the yield strength of bolted shear connectors is higher than that of headed studs. It can also be observed from the figure that the initial

This paper is concerned with the adequacy of demountable connections proposed for steel and composite structures. The finite element analysis of coped beams, column-column connections and composite beams was discussed to explore the demountability of steel and composite structures. For beam-beam connections, plastic deformation in the vicinity of the coped web was characterised using the equivalent plastic strain (PEEQ). The plastic damage in the coped region was determined from the finite element model. It was found that the beam-beam connections can be dismantled until the coped region maintains its

Fig. 29. Effects of connectors on the load-slip curves for composite beams.

Fig. 30. Effects of concrete compressive strengths on the applied load-slip curves for composite beams.

5. Conclusions

Please cite this article as: Uy B, et al, Behaviour and Design of Connections for Demountable Steel and Composite Structures, Structures (2016), http://dx.doi.org/10.1016/j.istruc.2016.06.005

12

B. Uy et al. / Structures xxx (2016) xxx–xxx

References

Fig. 31. Effects of slab types on the applied load-slip curves for demountable composite beams.

elasticity without any plastic damage. The coped beam can be demounted up to about 50% of the ultimate load and greater than typical service loads which are usually less than 40% of the ultimate load. A design procedure for demountable column-column connections under axial tension was also discussed. Parametric studies were conducted to examine the effects of sleeve length, bolt position and reinforcement ratio on the performance of demountable column-column connections under axial tension. It was demonstrated that an increase in the sleeve length and reinforcement ratio increases the ultimate tensile strengths of a column-column connection. Finally, a finite element model was developed for the modelling of a demountable composite beam. The effects of concrete compressive strength, sheeting thickness and connector types on the behaviour of a composite beam were investigated using the developed finite element model. It was shown that the strength of composite beams increases with increasing concrete compressive strength and sheeting thickness.

Acknowledgements The research described in this paper is financially supported by the Australian Research Council (ARC) under its Discovery Scheme (Project No.: DP140102134). This financial support is gratefully acknowledged.

[1] Teutsh M. Building in the bible-biblical buildings. Concr Precast Plant Technol 2001; 67(12):12–21. [2] Green Building Council of Australia. Green star overview. April 2009 [URL] www. gbca.org.au. [3] AS 4100. Steel structures, Sydney, New South Wales, Australia: Standard Australia; 1998. [4] EN 1993-1-8:2005. Eurocode 3: design of steel structures – part 1–8: design of joints. Brussels, Belgium: European Committee for Standardization; 2005. [5] Canadian Standards Association (CSA). CAN/CSA-S16-09 limit states design of steel structures. Toronto, ON, Canada: Canadian Standards Association; 2009. [6] AISC-LRFD. Load and resistance factor design specification for structural steel buildings. Chicago, II, USA: American Institute of Steel Construction; 2010. [7] Architectural Institute of Japan (AIJ). Standard for limit state design of steel structures; 1990. [8] Hogan TJ, Munter SA. Connection design guide 3-simple connections, design guide 3: web side plate connections. Australian Steel Institute; 2007. [9] ABAQUS user's manual version 12.1. Pawtucket, R.I.: Hibbit. Karlsson and Sorenson; 2012. [10] Puri GM. Python scripts for Abaqus learn by example. Dassault Systèmes Simulia Corporation; 2011. [11] Yam MCH, Lam ACC, Lu VP, Cheng JJR. Local web buckling strength of coped steel I beams. J Struct Eng 2003;129(1):3–11. [12] Standards Australia/Standards New Zealand (AS/NZS). AS·NZS3679.1-2010 structural steel – hot-rolled bars and sections. Sydney: AS & Wellington: NZS; 2010. [13] Li D, Uy B, Patel VI. Behaviour of demountable CFST column-column connections. In: B U, Lam D, editors. The 2015 International Conference on Steel and Composite Structures; August 25-29 2015 [Korea]. [14] Eurocode 1. Actions on structures, Part 1–7: accidental actions. DDENV 1991-1-7. European Committee for Standardisation (CEN); 1991. [15] AS/NZS 1170. Structural design actions part 0: general principles. Standards Australia Limited/Standards New Zealand; 2002. [16] AS 3600-2009. Concrete structures. Sydney, New South Wales, Australia: Standard Australia; 2009. [17] Wen R, Akbas B, Sutchiewcharn N, Shen J. Inelastic behaviors of steel shear tab connections. Struct Des Tall Special Build 2014;23:929–46. [18] Pathirana SW, Uy B, Mirza O, Zhu X. Strengthening of existing composite steelconcrete beams utilising bolted shear connectors and welded studs. J Constr Steel Res 2015;114:417–30. [19] Carreira DJ, Chu KH. Stress-strain relationship for plain concrete in compression. ACI J 1985;82(6):797–804. [20] Mirza O, Uy B, Han L. Time-dependent analysis of composite steel-concrete beams with innovative deep trapezoidal decks. Steel & Composite Structures-Proceedings of the 4th International Conference, 2010. [21] Mirza O, Uy B. Numerical modelling of short and long term behaviour of headed stud shear connector in composite steel-concrete beams. Incorporating sustainable practice in mechanics of structures and materials. CRC press; 2010. p. 91–6. [22] Kwon G, Engelhardt MD, Klingner RE. Behavior of post-installed shear connectors under static and fatigue loading. J Constr Steel Res 2010;66(4):532–41. [23] Viest IM, Colaco JP, Furlong RW, Griffs LG, Leon RT, L. W. Composite construction: design for Buildings. New York, United States: McGraw-Hill; 1997. [24] Ban H, Uy B, Pathirana SW, Henderson I, Mirza O, Zhu X. Time-dependent behaviour of composite beams with blind bolts under sustained loads. J Constr Steel Res 2015; 112:196–207.

Please cite this article as: Uy B, et al, Behaviour and Design of Connections for Demountable Steel and Composite Structures, Structures (2016), http://dx.doi.org/10.1016/j.istruc.2016.06.005