ICCM214 28-3 th July, Cambridge, England Finite Element Analysis of sustainable and deonstrutable semi-rigid beam-toolumn omposite joints *Abdolreza Ataei 1, Mark A. Bradford 2 1,2 Centre for Infrastruture Engineering & Safety, Shool of Civil & Environmental Engineering, UNSW Australia, Sydney, NSW 252, Australia Corresponding author: a.ataei@unsw.edu.au In this paper, an innovative sustainable semi-rigid beam-to-olumn omposite joint with deonstrutable bolted shear onnetors is modelled by using the general purpose software ABAQUS. The strutural mehanis of this joint onsidered in the paper requires areful onsideration, in order to apture the response aurately using omputational tehniques as the interations of the various omponents is omplex. Three laboratory test speimens having sustainable and deonstrutable semi-rigid beam-to-olumn joints have been tested and the results are used for validation of the finite element model. Preast green onrete (GC) slabs having redued CO 2 emissions during their manufature were attahed ompositely to the steel beam via pre-tensioned bolted shear onnetors, with the omposite beam being onneted to H-setion olumns using a flush end plate with two rows of bolts. The experimental testing was full-sized, with all the omponents being of the same size as would be met in pratie. The numerial model simulates the omposite beam-to-olumn onnetion under hogging moment and inludes the nonlinear material properties of all onstitutive materials of the omposite joint. For validation of the omputational proedure, the results of the numerial modelling are ompared in the paper with the experimental results, with good agreement being demonstrated. Keywords: FE modelling, Semi-rigid omposite joint, deonstrutablility, Preast, Connetions Introdution The traditional flush end plate semi-rigid omposite onnetion is one of the best hoies for onneting a omposite beam to a olumn. This kind of onnetion has several advantages suh as its ease of onstrution as well as being eonomial ompared to a rigid onnetion. Apart from these benefits, the rigidity in this onnetion an allow for adequate moment distribution in the frames. These omposite onnetions have higher initial stiffness and moment apaity as well as rotational apaity ompared with steel onnetions, owing to the ontribution of the reinforing bars loated in the slab. The indued tensile fores are resisted by the top bolts and the reinforing bars and ompressive fores are resisted by the steel beam. The reinforing bars ontribute signifiantly to the strength and stiffness of the onnetion. Traditional omposite systems utilise onrete derived from Portland ement, whih is one of the largest global soures of CO 2 emissions. Moreover, the traditional omposite floor systems suh as a solid reinfored onrete slab or profiled metal deking floor systems are ommon systems in omposite strutures. For typial onstrution praties for these types of systems, onrete asting, profiled steel deking plaing and onventional reinforing detailing are undertaken on-site, whih is time onsuming and labour intensive, and whih an inrease the ost of onstrution, and they an lead to quality redutions in the onstrution industry. Combining preast GC slabs having redued emissions during their manufature with steel elements by using a deonstrutable shear onnetion may solve these problems and onerns assoiated with traditional omposite strutures. Pre-tensioned high strength bolts installed through holes in preast GC slabs into pre-drilled holes in the steel beam produe a omposite flooring system that an be deonstruted at the end of the life of the struture (Bradford and Pi 212a,b, 213; Rowe and Bradford 213; Ataei and Bradford 213; Lee and Bradford 213). Marshall et al. (1971) appear to be the first researhers to have reported the use of bolted shear onnetion, but the ontext of the usage is not lear. Twelve push tests using high strength bolts as shear onnetors were arried out and reported by Dallam (1968). In these set of tests, the bolts were embedded in the onrete slab and pre-tensioned by the turn-of-nut method after the onrete had aged 28 days. He pointed out 1
that high strength bolts displayed a higher apaity and ultimate strength than stud shear onnetors. Six full-sale simply supported omposite beams with high-strength bolted shear onnetors were tested by Dallam and Harpster (1968), but the bolted shear onnetors were embedded in the onrete slabs. Based on this, they onluded that pre-tensioned high strength bolts provide a very rigid onnetion between the steel beam and onrete slab at servie loads, and a reserve apaity suffiient to develop the ultimate moment apaity of the fully omposite setion is attainable. A series of tests was onduted on three types of 22-mm diameter post-installed shear onnetors under stati and fatigue loading by Kwon et al. (21). It was onluded that bolted shear onnetors exhibited signifiantly higher fatigue strengths than stud shear onnetors. Five fullsale non-omposite beams were onstruted to investigate the retrofitting of the bridge beams by Kwon et al. (211). The reinfored onrete slabs were attahed ompositely to the steel girder via post-installed onnetors in four beams. It was onluded that the strength and stiffness of the nonomposite bridge girder an be improved signifiantly. In the tests onduted by Kwon et al. (21, 211), the bolts were embedded in the onrete or grout. Lee and Bradford (213) onduted two series of push-out tests to obtain the behaviour of the post-installed pre-tensioned bolted shear onnetors. The first and the seond series of this experimental study inluded five and four pushout speimens, respetively. All speimens in the first series and two speimens in the seond series were onstruted by using post-installed pre-tensioned bolted shear onnetors. The major differenes between the first and seond series were size of the preast slab, the reinforement and the number of bolts. These studies did not fous on the testing and modelling of deonstrutable and sustainable semi-rigid flush end plate omposite joints. In order to provide a robust and effiient means for modelling sustainable semi-rigid beam-toolumn omposite onnetions with deonstrutable bolted shear onnetors, the present paper presents a three-dimensional modelling using ABAQUS software. Three speimens having sustainable and deonstrutable semi-rigid beam-to-olumn joints have been tested and the results are used for validation of the finite element model. Preast GC slabs are attahed ompositely to the steel beam via pre-tensioned bolted shear onnetors and the omposite beam is onneted to H- setion olumns using a flush end plate with two rows of bolts. The model simulates a omposite beam-to-olumn onnetion under hogging moment and it inludes the non-linear material properties of all onstitutive materials of the omposite joint. Almost all omponents were modelled as being of the same size as in the experimental tests, inluding the steel beam, steel olumn, flush end plate and bolts in the onnetion region. For validation of the model, the results of the numerial modelling are ompared with the experimental test results and good agreement is ahieved. The modelling is shown to provide an effiaious tehnique for onduting parametri studies, so as to develop design guidane in this novel appliation in omposite onstrution. Finite Element Model Material modelling For the ABAQUS modelling, the atual stress-strain urves for the materials used an be determined from the material tests. Material tests for the bolts, steel beam and olumn, reinforing bars, reinfored onrete and bolted shear onnetors were onduted and the results were used for the FE model. Von Mises plastiity was used to model the all the strutural steel parts as an elastiplasti material with hardening in both tension and in ompression. The relationship of the strain and stress for all strutural steel parts adapted in the model is illustrated in Fig. 1. The load-slip relationship and the points defining the relationship between the load and slip of the bolted shear onnetors adopted in the FE modelling are shown in Fig. 1. Conrete in ompression and tension was represented using the damaged plastiity model in ABAQUS. For onrete under uniaxial ompression, the formulation of Carreira and Chu (1985) that is ommonly used in numerial modelling was adopted (Fig. 2) as E f / 1 / 35 f 35 f, (1) 2
Stress (MPa) where 2, 3 f / 32 4 155 (2) is a fator whih ontrols the urvature of the stress-strain relationship and f is the mean ompressive ylinder strength of the onrete in units of MPa. In order to model the onrete in tension, the tensile stress was assumed to inrease linearly to 1 of its ompressive strength. After raking the onrete, the stress delines to zero at strain of about 1 times the failure strain, as an be seen in Fig. 2. 1, 8 6 4 2 Steel beam and olumn Reinforement Bolts 5 1 15 2 Strain () Fig. 1. Stress-strain relationship adapted in FE modelling; Reinforing bars, bolts and steel beam and olumn, Axial onnetors / f 1.8.6.4.2 1.6 2.4 4. 6. / f 1 / 1 2 3 4 / f : ompressive strength t / f t 1.8.6.4.2 f t : tensile Linear elasti Linear strength 2 4 / 6 8 1 t softening Fig. 2 Outline of the normalised uniaxial stress-strain relationship for onrete under ompression (Carreira and Chu, 1985) tension. 3
Element Type and modelling strategy The type and size of elements sometimes have a signifiant effet on the results, and so the determination of the element type and size is one of the important issues in FEM. A finer mesh leads to a better result. However, the finer mesh may leads to omputational time problems. Three-dimensional solid elements were used to model the bare and omposite onnetions. Exept for the reinforing bars, all omponents are modelled by 8-node solid elements (C3D8R) with a redued integration sheme whih prevents shear loking, redues omputational time, and provides the required auray (Bursi and Jaspart 1998; Bathe 1996; Cook et al. 22). For modelling the steel reinforement, a two-node linear truss element (T3D2) is used. The reinforement was embedded into the slab, with the slab being the host region and the bars being an embedded region. This tehnique onnets these two different omponents and prevents slip between them. A typial FE model of a omposite joint is shown in Fig. 3. Fig. 3. Finite element model of a omposite joint; with showing onrete slab without showing onrete slab Contat Modelling There are various omponents in omposite onnetions that interat with eah other, and the results of the FE analysis depend on the aurate modelling of the ontat interation between these omponents. Experimental results show that there is no separation between the head of the bolt and the flush end-plate, nor between the nut of the bolt and the inner fae of the steel olumn. Therefore, the TIE option was used for onneting these omponents whih provides full interation between the bolts, nut and bolt head. The TIE option was also used for onneting the steel beam to the flush end-plate, beause these two omponents are welded together. In order to simulate the interation between the top flange of the steel beam and the lower part of the preast onrete slab, surfae-to-surfae ontat interation using a penalty method with a oeffiient of 4
frition of 25 was adopted, in whih the top flange and the onrete slab were onsidered as master and slave surfaes, respetively. Bolted shear onnetor model Modelling the interfae between the onrete slab and shear onnetors is one of the main issues in the FE modelling of the omposite beam and joint. In this researh, the strength and the stiffness harateristis of the pre-tensioned bolted shear onnetors determined from the push tests onduted by authors (Fig. 1) are used for modelling of the onnetion between the onrete slab nodes and steel flange nodes, and so an axial onnetor model was used to model the interfae slip. These onnetors were loated at the same positions where bolts were plaed on the speimen. A shemati diagram of the axial onnetor model is shown in Fig. 4. The frature of the bolter shear onnetors was assumed to our at the ultimate slip (15 mm) obtained from the push-out tests. Load Appliation and boundary onditions Figure 4: Axial onnetor model for Composite Joint 1 The loading was applied in two steps. First, the pretension was applied to the bolts loated in onnetion and the joints were then loaded, at whih state the bolts were subjeted to the pretension as an be seen in Fig. 5. Mirza and Uy (211) have pointed out that Riks tehnique is needed to apture any unloading in the non-linear analysis, and so the GENERAL method and modified RIKS method were used for the first step and seond step respetively. A stati onentrated load was applied at the entre of the steel olumn, as was done in the tests. The initial inrement plays a vital role in the onvergene of the modelling whih is why this parameter will be adjusted if a onvergene problem ours during the modelling. Fig. 5. Stress distribution at the first step of loadings Bolt Flush end plate. 5
In FE modelling, orret representations of the boundary onditions are essential sine slightly different boundary onditions an produe signifiantly different results. For the simulation in this paper, the boundary onditions were taken as being exatly the same as in the tests, with the olumn being allowed to move in the vertial diretion and with the flanking ends of the beams having roller supports. Experimental study Three beam-to-olumn joints were designed and onstruted in a ruiform arrangement to simulate the internal joint in a semi rigid frame. The details of beam-to-olumn joint speimens are summarised in Table 1. A navel methodology of shear onnetion was adopted by using pretensioned bolted shear onnetors to attah the preast onrete slabs to the top flange of steel beams. Speimens 1 and 2 were designed as a omposite joint (CJ) and Speimen 3, whih is a non-omposite joint, was designed as a ontrol test speimen to ompare against the omposite joint tests. All beam-to-olumn joint speimens onsist of a steel beam of 46 UB 821 and steel olumn of 25 UC 895. A 12 mm flush end plate welded at the end of steel beam used and onneted to the flange of the olumn by using 4 M24 grade 88 bolts. Stiffener plates were welded to the olumn web at the level of the bottom and top flanges of beam to prevent bending of the olumn flanges in tension and failure of the olumn web in ompression. The geometri and design details for Composite Joints are presented in Fig. 6. Table 1: The details of beam-to-olumn joint speimens. Speimen Beam Column T ep B d R N B s T s N s (mm) (mm) (mm) (Mpa) CJ1 46 UB 82.1 25 UC 89.5 12 M24 6N16 6M2 525 12 1 CJ2 46 UB 82.1 25 UC 89.5 12 M24 6N16 6M2 525 12 1 SJ3 46 UB 82.1 25 UC 89.5 12 M24 6N16 6M2 N.A. N.A. N.A. Notes: T ep =end plate thikness; B d =bolt diameter in onnetion region; R=Reinforing; N = no. of bolts as shear onnetors per beam; B s = shear onnetor spaing; T s =slab thikness; N s =No. of slab units. 331 M2 Bolts Washer 6x6x6 mm GC Preast Slab Grade 8.8 275 125 26 125 275 Reation 525 525 12 12 525 525 Reation A Steel beam (46 UB 82.1) 6N16 effetive N1 @15 or 2 12 N1 @16 83 195 85 34 11 422 256 422 M2 Grade 8.8 Flush End plate 51x22x12 4 M24 Bolts Grade 8.8 6 N16 effetive N1 @15 or 2 85 Stiffener 2X(22x12x16 mm) End plate 4 M24 bolts 51X22X12 mm Grade 8.8 Load Steel olumn Stiffener 25 UC 89.5 4X(22x12x16 mm) Elevation View A 125 78 1 78 H-setion Column 25 UC 89.5 331 8 85 85 8 2 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 2 16 16 19 11 19 16 16 N1 @15 Dut 3x276 H-setion Column 25 UC 89.5 N1 @15 6N16 effetive Plane View(Reinforing Bars) 6
A Plate 11X8x8mm Pre-tensioned bar M2 Bolts Washer 6x6x6 mm 331 GC Preast Slab 275 125 26 125 275 Reation 525 525 12 12 525 525 Reation Plasti pipe 6N16 effetive 11 422 256 422 M2 Grade 8.8 195 6 N16 effetive Plasti pipe 12 Anhor Head and Wedges Steel beam (46 UB 82.1) N1 @15 or 2 N1 @16 83 85 34 Flush End plate 51x22x12 4 M24 Bolts Grade 8.8 N1 @15 Stiffener 2X(22x12x16 mm) End plate 51X22X12 mm Steel olumn 25 UC 89.5 Load Elevation View 4 M24 bolts Grade 8.8 Stiffener 4X(22x12x16 mm) A 85 125 78 1 78 Setion A-A H-setion Column 25 UC 89.5 331 8 85 85 8 2 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 2 Anhor Head and Wedges 6N16 effetive 16 16 19 11 19 Plate 11X8x8mm 16 16 N1 @15 Dut 3x276 H-setion Column 25 UC 89.5 Plane View(Reinforing Bars) N1 @15 Validation of the FEM Figure 6: Details of joints: CJ1; CJ2. The results of the ABAQUS-based FE modelling are ompared herein with the experimental results obtained from experimental study to investigate the auray and reliability of the FEM, as well as to validate the FEM. Three beam-to-olumn joint speimens were tested in the Heavy Strutures Researh Laboratory at the University of New South Wales and the results were used for validation of the FE model. These full-sale semi-rigid flush end plate beam-to-olumn omposite joint tests with deonstrutable bolted shear onnetors were onduted under symmetrial loading to evaluate the strutural harateristis of these new omposite joints. In this omparison, the load versus defletion response was modelled, and the results are given in Figs. 7 to 7 for the two omposite onnetions tested and in Fig. 7() for the onnetion with a bare steel beam. Table 2 shows the omparison of the FE modelling results and experimental tests. It an be seen that the FE model an predit the ultimate load and defletion of all speimens aurately. CJ1 onsists of one unit of the preast onrete panels attahed to the top flanges of steel beam by using pre-tensioned bolted onnetors. The longitudinal reinforement ratio for this slab was 9%. Six effetive longitudinal N16 reinforing bars were distributed and plaed in the top layer of the preast onrete slab. A maximum load of about 55 kn at the defletion of about 55 mm was reorded for this omposite joint and the speimen behaved non-linearly before failure. The load versus defletion response was modelled, and the results are given in Fig. 7 for CJ1. It an be seen that the agreement is very good. 7
Load (kn) Load (kn) Load (kn) 6 5 4 3 2 1 Bolt Frature CJ 1 FEM 2 4 6 8 Defletion (mm) 6 5 4 3 2 1 Bolt Frature CJ 2 FEM 1 2 3 4 5 6 Defletion (mm) 25 2 15 1 5 Bolt Frature SJ 3 FEM 1 2 3 4 5 6 Defletion (mm) () Fig. 7. Comparison of FE model and tests for load-defletion response; CJ1, CJ2, () SJ3 Table 2: Comparison of FE model and tests. P ult 8 D ult Test FEM Test FEM CJ1 55 549 55 48.5 CJ2 498 49 54 56 SJ3 25 217.5 52 53.8 Notes: P ult = ultimate load, D ult = ultimate defletion CJ2 was similar to CJ1 exept that CJ2 onsisted of two separated preast onrete panels attahed to the top flanges of steel beam. N16 Longitudinal reinforing bars were not plaed in the top layer of the slabs before onrete asting. In order to onnet two preast onrete panel together a pretensioning proedure was used. The N16 reinforing bars were plaed into the prepared holes by using plasti tube and then stressed to about 1% of their axial load apaity before testing. A maximum load of about 498 kn at the defletion of about 54 mm was reorded for this omposite joint and the speimen behaved non-linearly before failure. The load versus defletion response was modelled, and the results are given in Fig. 7 for CJ2. It an be seen that the agreement is very good. SJ3 was designed as the referene and ontrol test speimen. The steel joint (SJ3) is similar to the two omposite joints exept that SJ3 is non-omposite and preast onrete slab was not attahed on the top flange of the steel beam. A maximum load of about 25 kn at the defletion of about 52 mm was reorded for this speimen and it behaved non-linearly before the failure. The load versus
defletion response was modelled, and the results are given in Fig. 7() for SJ3. It an be seen that the agreement is very good. Failure modes All speimens failed by frature of the M24 bolts due to the resulting tension fores. The FE model an predit aurately the failure mode of the omposite onnetion. Fig. 8 shows the strain distribution for the bolt in the mode of failure for CJ1. As an be seen in Fig. 8, the maximum strains in the shank of the bolt are almost same as the strain apaity of the bolt obtained from the tensile test on the bolts. Table 3 shows a omparison of the failure modes between the FE modelling and experimental testing. It an be seen that the FE model an predit the failure mode of all speimens aurately. The FEM an also predit aurately the plasti deformation and bending of the flush end plate, as illustrated in Fig. 9. Fig. 8. Bolt failure at the failure mode; Test (CJ1), FEM Test (CJ 1 ) FEM Conlusions Fig. 9. Deformation and bending of the end plate; Test (CJ1), FEM (in Pa) Table 3: Comparison of the failure mode. CJ1 CJ2 SJ3 Test BF BF BF FE Model BF BF BF Note: BF: Bolt failure This paper has desribed a numerial modelling of semi-rigid flush end plate beam-to-olumn omposite joint tests with deonstrutable bolted shear onnetors. Three speimens with sustainable and deonstrutable semi-rigid beam-to-olumn joints have been tested and the results 9
were used for validation of the finite element model. Preast GC slabs were attahed ompositely to the steel beam via pre-tensioned bolted shear onnetors and the omposite beam was onneted to H-setion olumns using flush end plates with two rows of bolts. The model simulates a omposite beam-to-olumn onnetion under hogging moment and inludes non-linear material properties of all onstitutive materials of the omposite joint. For validation of the model, the results of the numerial modelling were ompared with experimental test results and good agreement was ahieved. The modelling was shown to provide an effiaious tehnique for onduting parametri studies, so as to develop design guidane in this novel appliation in omposite onstrution. Aknowledgements The work reported in this paper was undertaken with the finanial support of the Australian Researh Counil through an Australian Laureate Fellowship (FL1163) awarded to the seond author. Referenes Ataei A. and Bradford M.A. (213) FE Modelling of sustainable semi-rigid flush end plate omposite joints with deonstrutable bolted shear onnetors, International Conferene on Composite Constrution CCVII, Palm Cove, Queensland. Bradford M.A. and Pi Y.-L. (212a) Numerial modelling of deonstrutable omposite beams with bolted shear onnetors, Conferene on Numerial Modeling Strategies for Sustainable Conrete Strutures, Aix-en-Provene, Frane, II-2, 1-8. Bradford M.A. and Pi Y.-L. (212b) Numerial modelling of omposite steel-onrete beams for life-yle deonstrutability, 1st International Conferene on Performane-Based and Life-Cyle Strutural Engineering, Hong Kong, 12-19. Bradford M.A. and Pi Y.-L. (213) Nonlinear elasti-plasti analysis of omposite members of high-strength steel and geopolymer onrete, Computer Modeling in Engineering and Sienes 232: pp. 1-27. Bursi OS, Jaspart JP. Benhmarks for finite element modelling of bolted steel onnetions. Journal of Construtional Steel Researh 1998;43(1-3);17-42. Bathe KJ. Finite element proedures. Enwood Cliffs (NJ): Prentie-Hall; 1996. Cook RD, Malkus DS, Plesha ME, Witt RJ. Conepts and appliation of finite element analysis. 4thedition John Wiley & Sons In;22. CEN. Euroode 4: Design of Composite Steel and Conrete Strutures. Part 1.1: General Rules and Rules for Buildings (pren 1994-11:24). Dallam L.N. (1968) Pushout tests with high strength bolt shear onnetors, Report for Missouri State Highway Department, Department of Civil Engineering, University of Missouri-Columbia, Missouri, USA. Dallam L.N. and Harpster J.L. (1968) Composite beam tests with high-strength bolt shear onnetors, Report for Missouri State Highway Department, Department of Civil Engineering, University of Missouri-Columbia, Missouri, USA. Kwon G., Engelhardt, M.D. and Klinger, R.E. (21) Behavior of post-installed shear onnetors under stati and fatigue loading. Journal of Construtional Steel Researh, Vol. 66, pp. 532-541. Kwon G., Engelhardt, M.D. and Klinger, R.E. (211) Experimental behavior of bridge beams retrofitted with postinstalled shear onnetors. Journal of Bridge Engineering, ASCE 16: 536-545. Lee S.S.M. and Bradford M.A. (213) Sustainable omposite beams with deonstrutable shear onnetors, 5th International Conferene on Strutural Engineering, Mehanis and Computation, Cape Town, South Afria. Marshall W.T., Nelson H.M. and Banerjee, H.K. (1971) An experimental study of the use of high-strength frition-grip bolts as shear onnetors in omposite beams, The Strutural Engineer, Vol. 49, pp. 171-178. Mirza, O. & Uy, B. (211), Behaviour of omposite beam-olumn flush end-plate onnetions subjeted to lowprobability, high-onsequene loading. Engineering Strutures 211; 33(2):642-662. Oehlers, D.J. and Bradford, M.A. (1995) Steel and Conrete Composite Strutural Members: Fundamental Behaviour, Pergamon, Oxford. Rowe M. and Bradford M.A. (213) Partial shear interation in deonstrutable omposite steel-onrete omposite beams with bolted shear onnetors, International Conferene on Design, Fabriation and Eonomy of Welded Strutures, Miskol, Hungary, pp. 585-59. 1