320 THE FINITE ELEMENT MODEL. J.N. Karadelis et al. The basic model consisted of the rectangular hollow section with the cap-plate fully integrated at one end and the cleat-plate on top (Figure: 1). As the structural member (tie) was symmetrical about a plane at right angles to its longitudinal axis, only half of the member was initially modelled. No modelling of the weld was present. Material properties such as a Young's Modulus value of 205000 N/mm 2 and a Poisson's Ratio of 0.27 were inserted in the program. Translational restraints were applied at the cut end and a negative pressure load of 1 kN at the cleat plate. The correct choice of element is very important in finite element analysis. A 3D, 4 node tetrahedral structural solid element (SOLID 72) with three translational and three rotational degrees of freedom (DOF) per node was chosen. This is described by ANSYS as a general purpose element particularly suited to automatic meshing of irregular volumes. A linear elastic stress analysis was performed using the unit load and the results were normalised for other load values. These results are shown plotted on the same axes with the experimental results. Utilising the experience obtained from the laboratory tests and the linear elastic analysis results, the model was divided into three substructures. The cap plate was allowed to undergo both, large displacements and material non-linearities, as previous experimental work with various cap-plate thicknesses demonstrated that when thin, the later undergoes excessive deflections under and near the beating of the cleat plate. The region of the SHS near the cap-plate was seen to develop excessive stress concentration. These stresses tended to exceed the yield stress of the material, hence non-linear material properties were attributed to it. Finally, the weld was treated linearly at this early stage of the investigation. A non-linear (multi-linear, elasto-plastic) analysis featuring isotropic hardening effects was performed. The large displacements option was kept open to accommodate possible non-linear effects of the cap-plate. The results are shown plotted in the next pages: RESULTS AND DISCUSSION Figure: 4 shows the variation of strain with load just below the interface of the cap-plate and the hollow section. This is a region of high stress concentration as predicted by the finite element model. In order to save space, strains at positions 13 and 17 as well as 16 and 18 (Figure: 1, diagram of test specimen) were averaged and plotted in pairs. Plots from the linear and non-linear analyses were superimposed for ease of comparison. The same procedure is repeated for strains developing at positions 19 and 21 and also 20 and 22, measured 300 mm below the cap plate on the SHS and presented in Figure: 5. The finite element analysis results are plotted with them. It can be seen that the strain values predicted by the finite element model are in good agreement with the corresponding results obtained in the laboratory. Figure: 6 shows the variation of displacements as measured at positions 5 and 8 and also 6 and 7. The agreement here cannot be considered as satisfactory as the one above. At the time of writing this paper the authors are investigating all the possibilities. It is anticipated that had the weld been modelled in a more rigorous manner, its contribution to the behaviour of the SHS connection would have been better represented. Another possibility currently under scrutiny is the reliability of the ageing DENISON machine on which some tests were carded out. The Behaviour of T-End Plate Connections to SHS. Part H Figure: 4 Variation of Experimental and Calculated Average Strain with Load Strain Gauges: (13+17), (16+18) 321 Figure: 5 Variation of Experimental and Calculated Average Strain with Load. Strain Gauges: (19+21), (20+22) Figure: 6 Variation of Average Displacement with Load. LVDTs: (5+8), (6+7) 322 J.N. Karadelis et al. CONCLUSIONS AND FUTURE WORK The behaviour of the SHS connections has been modelled using finite element analysis techniques. Satisfactory agreement was obtained between the experimental and the calculated strains. The deformation of the cap-plate as noted in the laboratory was predicted accurately by the FE-model. Stress concentrations were also predicted to develop directly under the cap-plate indicating symptoms of weld yielding, possible separation from the parent metal and therefore the necessity to model the weld more accurately. The finite element model is to be refined and calibrated. Efforts will be turned to the weld around the cap-plate and between the interface of the cap-plate and the cleat-plate. The improved finite element model will be used to carry out the appropriate parametric and sensitivity studies. When all parameters are investigated structural optimisation techniques will be used to optimise the connection and develop the appropriate design guides. Figure: 1 SHS connection, FE-model and line diagram of the test specimen. REFERENCES Kohnke P (editor), (1998), ANSYS 5.4, Theory Manual. Canonsburg, PA USA. Mase G E, 1970, Continuum Mechanics, McGraw-Hill book company. Saidani M, Omair M R, Karadelis J N (1999), Behaviour of End Plate Connection to Rectangular Hollow Section, Part I, Experimental Investigation, ICASS99' Hong Kong, PRC. Walz J E, Fulton R E, Cyrus N J (1978), Accuracy and Convergence of Finite Element Approximations., Proceedings, Second Conference on Matrix Methods in Structural Mechanics, Wright-Paterson Air Force Base, Ohio, USA. Zienkiewicz O C, Taylor R L (1994), The Finite Element Method, Vol. 1-2, 4 th Ed., McGraw-Hill. CYCLIC BEHAVIOUR OF BEAM-TO-COLUMN WELDED CONNECTIONS Elena Mele 1, Luis Calado 2, Antonello De Luca I 1 Structural Analysis and Design Department (DAPS), University of Naples "Federico II", P.le Tecchio 80, 80125 Naples, Italy. 2 Civil Engineering Department (DECivil), Instituto Superior Tecnico, Av. Rovisco Pais, 1096 Lisboa Codex, Lisbon, Portugal. ABSTRACT In this paper the results of an experimental program devoted to the assessment of the cyclic behaviour of full scale, European type, beam-column subassemblages with welded connections are presented. Six tests (five cyclic and one monotonic) have been carded out on three different series of specimens, encompassing a total of eighteen tests. The tests have evidenced the effect of column size and panel zone design on the cyclic behaviour and failure modes of the connections, as well as the dependency of the moment capacity and of the maximum and cumulative plastic rotation of the joint upon the applied loading history. KEYWORDS welded connections, cyclic tests, loading histories, rotation capacity, panel zone, failure modes. INTRODUCTION The confidence of structural engineering in welded moment resisting frames (WMRFs) was strongly compromised by the performances observed in the earthquakes of Northridge (1994) and Hyogoken- nanbu (1995). Following these earthquakes, extensive unexpected brittle connection damage were detected in several frames, thus discovering the alarming problem of the high seismic vulnerability of the welded steel framed structures. The brittle modes of failure occurred at the beam-to-column joints have been defined "unexpected", since the WMRF connections were usually considered as the ones characterised by the more stable and ductile behaviour, giving rise to large rotational capacity and energy dissipation. It should be underlined though that, as reported by (Bertero et AI., 1994), almost all the types of failures occurred as a result of the Northridge seismic shaking, had been observed in past experimental tests carded out in U.S.A., as well as in Japan and Europe. However the experimental behaviour of the welded connections appears highly and perhaps randomly variable. 323 324 E. Mele et al. Starting from these observations, significant research efforts have been undertaken in the United States (Mahin et AI., 1996; Malley, 1998), in Japan (Tanaka et AI., 1997; Nakashima et AI., 1998) and also in Europe (Mele et AI., 1997; Plumier et AI., 1998; Taucer et AI., 1998; Calado et AI., 1999), in order to enrich the experimental data base and to assess the major parameters affecting the cyclic behaviour of beam-to-column connections. In this context, a wide experimental program has been carried out at the Material and Structures Test Laboratory of the Instituto Superior T6cnico of Lisbon on different types (both welded and bolted) of beam-to-column connections. The experimental tests have been performed on specimens representative of frame structure beam-to-column joints close to the ones typical of European design practice (beams less deep than the ones adopted in the current US design of SMRFs), with the aim of defining the effect of the column size and of the PZ design on the connection behaviour, varying the applied loading history. Some preliminary experimental results on the welded connections have been presented in (Mele et AI., 1997). In this paper a complete overview on the experimental program carded out on welded connections is reported. In particular the experimental results are presented through hysteresis loops obtained in the increasing amplitude tests; further, the failure modes of the specimens are described, and the major factors affecting the cyclic behaviour and the rotation capacity are assessed. THE EXPERIMENTAL PROGRAM Aim The experimental program on welded beam-to-column connections presented in this paper was aimed at evaluating the effect of the column dimensiom and panel zone design on the cyclic behaviour, ultimate strength and deformation capacity of the welded connections, varying the applied loading history. Specimen geometry A total of 18 beam-to-column fully welded joints (3 series x 6 specimens) have been designed, fabricated and tested up to failure under different loading histories. The specimens, made of $235 JR steel, are T-shaped beam-column subassemblages, consisting of a 1000 mm long beam and a 1800 mm long column. In the three types of specimens, respectively appointed as BCC5, BCC6 and BCC8, the beam cross section is the same (IPE300), while the column cross section is varied, being respectively HE160B for the BCC5 series, HE200B for the BCC6 series, and HE240B for the BCC8 series. The section properties of beam and columns adopted in the three specimen types are reported in table 1. Height (ram) TABLE 1 BEAM AND COLUMN SECTION PROPERTIES Beam Section All specimens IPE 300 300 BCC5 HE160B 160 Column Section BCC6 HE200B 200 BCC8 HE240B 240 Width (mm) 150 160 200 240 t~ (mm) 7.1 9 8 10 tf(mm) 10.7 15 13 17 I (mm 4) 83356 x 103 24920 x 103 56960 x 103 112600 x 103 Wr (mm 3) 557 x 103 311X 10 3 570x 10 3 354 x 10 3 628 x 103 643 x 103 Wpl (mm 3) 938 x 10 3 1053 x 103 Cyclic Behaviour of Beam-To-Beam Welded Connections 325 Due to the relative cross-section dimensions of column and beam in the three series of connection specimens, the beam plastic modulus is respectively larger, approximately equal and smaller than the column plastic modulus for the BCC5, BCC6 and BCC8 series. In all the specimens, the beam flanges have been connected to the column flange by means of complete joint penetration (CJP) groove welds, while fillet welds have been applied between both sides of the beam web and the column flange. The continuity of the connection through the column has been ensured by horizontal 10 mm thick plate stiffeners, fillet welded to the column web and flanges. Material properties The structural steel used for the specimens (beam, colunm, stiffener plates) is $235 JR type. The basic monotonic stress-strain curve and the mechanical properties of the specimen steel components have been determined through coupon tension tests. The average values of material properties (yield and ultimate stress) for the beam and column flanges and web are provided in table 2. In the same table are also provided the plastic and ultimate flexural capacities (Mp=Wpl x fy, Mu Wpl x fy) of the beam and of the colunm, computed on the basis of the corresponding values of yield stress fly) and ultimate stress (fu) of the section flanges obtained from the tension tests. fy (MPa) fu (MPa) YR Mp (kNm) Mu (kNm) TABLE 2 AVERAGE VALUES OF MATERIAL PROPERTIES AND DERIVED FLEXURAL CAPACITIES. BCC5 BCC6 BCC8 Beam IPE300 flange web 274.8 305.5 404.6 412.6 1.47 1.35 166 234 Column HE160B flange web 323.1 395.6 460.2 490.1 1.42 1.24 118 157 Beam IPE300 flange Web 278.6 304.9 398.8 411.4 1.43 1.35 169 242 Column HE200B flange web 312.6 401.6 434.9 489.8 1.39 1.22 198 276 Beam IPE300 Flange web 292 300 445 450 1.53 1.50 183 280 Column HE240B Flange web 300 309 457 469 1.52 1.52 316 482 Experimental set-up, instrumentation plan and loading histories The test set-up, represented in figure 1, mainly consists in a foundation, a supporting girder, a reaction r.c. wall, a power jackscrew and a lateral frame. The power jackscrew (capacity 1000 kN, stroke + 400mm) is attached to a specific frame, pre-stressed against the reaction wall and designed to accommodate the screw backward movement. The specimen is connected to the supporting girder through two steel elements. The supporting girder is fastened to the reaction wall and to the foundation by means ofpre-stressed bars. An automatic testing technique was developed to allow computerised control of the power jackscrew, of the displacement and of all the transducers used to monitor the specimens during the testing process. Specimens have been imtrtmaented with electrical displacement transducers (LVDTs), for carefully recording the various phenomena occurring during the tests. The same arrangement of LVDTs has been adopted for the three specimen types. The typical instrumentation set-up is provided in figure 2. Each specimen type has been tested up to failure under several cyclic rotation histories. The complete set of loading histories is provided in table 3, where loadings are defined in terms of: applied beam tip displacement (d); applied beam tip displacement normalised to theoretical yield displacement dr (d/dy); interstory drift angle (d/H), i.e. d normalised to the distance between beam tip and column centreline H. 326 E. Mele et al. Figure 1: Experimental set-up Figure 2: Specimen instrumentation B BB C D E mon TABLE 3 LOADING HISTORIES BCC5 Ii BCC6 + 75 + 7.5 + 7.5 + 75 + 7.5 + 7.5 _+ 75 _+ 7.5 _+ 7.5 _+ 75 _+ 7.5 _+ 7.5 Stepwise Increasing (ECCS) Stepwise Increasing (ECCS) Monotonic Monotonic ii Bcc8 I _+37.5 +3.75 +3.75 I I Stepwise Increasing (ECCS) +37.5 1___3.75 I_+3.75 Monotonic EXPERIMENTAL RESULTS: GLOBAL BEHAVIOUR AND FAILURE MODES In the following the experimental results obtained in the test program are provided. In particular the cyclic behaviour and the failure modes observed for the three sets of specimens are descn'bed, and the moment rotation hysteresis loops obtained in the stepwise increasing amplitude cyclic tests are provided. In the moment rotation hysteresis loops hereafter presented, the rotation values have been calculated both as the "unprocessed" total rotation given by the applied interstory drift angle d/H, and as the beam rotation ~ obtained through the measured LVTDs displacements at the beam cross sections. Correspondingly, in the M-d/H experimental curves the moment is evaluated at the centreline of the column, while in the Mb-tlh, curve the moment is evaluated at the column flee. In figure 3 (a) the moment - total rotation (M-d/H) experimental curves resulting from the BCC5C, BCC6C and BCC8D tests (cyclic increasing stepwise amplitude) are plotted, while in figure 3 (b) both the corresponding moment - beam plastic rotation and the moment - panel zone rotation curves are plotted. The beam plastic rotation has been obtained through the measured displacements at the transducers 1 and 2 (see figure 2) by subtracting the contributions of the beam and column elastic rotations as well as of the panel zone distortion. Cyclic Behaviour of Beam-To-Beam Welded Connections 327 Figure 3 (a): Moment-global rotation curves Specimens BCC5 Figure 3 (b): Moment-beam plastic rotation and Moment-panel rotation curves As can be derived from the curves reported in figure 3 (a) and (b), and as demonstrated also in the other tests carried out in the experimental program, the cyclic behaviour of the specimen BCC5 is characterised by a great regularity and stability of the hysteresis loops up to failure, with no deterioration of stif~ess and strength properties. The very last (18 th) cycle presents a sudden and sharp reduction of strength, corresponding to the collapse of the specimen, which occurred due to fracture initiated in the beam flange and propagated also in the web. During the test, significant distortion of the joint panel zone has been observed, while not remarkable plastic deformation in the beam occurred. In table 4 a sunmm~ of the number of complete plastic cycles to collapse and the failure mode of the specimens is reported. TABLE 4 NUMBER OF PLASTIC CYCLES AND FAILURE MODES OF BCC5 SPECIMENS , ,.= !.o. +o,. ,.,,.r od. 1 . [ 5 :48 Fracture of the beam flange near the weld. I Crack on the beam flange d~ to the weld, propagated in the beam web ! 328 Specimens BCC6 E. Mele et al. Throughout the test program, two different kinds of cyclic behaviour have been observed for the BCC6 specimens. In some cases (tests C and D) the behaviour of the specimens is close to the behaviour observed for the BCC5 type, with almost no deterioration of the mechanical properties up to the last cycle, during which the collapse occurred. On the contrary, for the other tests (A, B and BB) a gradual reduction of the peak moment at increasing number of cycles is evident. In these eases, starting from the very first plastic cycles, local buckling of the beam flanges occurred, and a well defined plastic hinge has formed in the beam. In the specimens BCC6 the contribution of the panel zone deformation has not been as significant as in the BCC5 specimen type. The collapse of the specimens BCC6A and BCC6B was due to fracture of the beam flange in the buckled zone. The specimens BCC6BB, BCC6C and BCC6D failed due to fracture in the beam flange along or close to the weld line. In table 5 a smmnm~ of the number of complete plastic cycles to collapse and the failure mode of the specimens is reported. TABLE 5 NUMBER OF PLASTIC CYCLES AND FAILURE MODES OF BCC6 SPECIMENS Specimens BCC8 The hysteresis loops obtained from the tests on the BCC8 specimens (except the one obtained in the C test) show a gradual reduction of the peak moment starting from the second cycle, where the maximum value of the applied moment has been usually registered. This deterioration of the flexural strength of the connection is related to occurrence and spreading of local buckling in the beam flanges and web. A well defined plastic hinge in the beam has formed in all the tested specimens. In the test C, where the specimen has been subjected to a constant amplitude rotation history, equal to 7.5% rad, an unstable behaviour of the specimen has been observed, with multiple buckling occurred in the beam flanges starting from the first plastic cycle, and a sudden failure occurred at the third plastic cycle due to the fracture in the beam flange along the weld. In the specimens BCC8 the panel zone deformation has not been remarkable, and the plastic deformation mainly took place in the beam. The collapse of the specimens BCC8A and BCC8D was due to fracture of the beam flange in the buckled zone. In the tests B, C and E the collapse of the specimens occurred dueto fracture in the beam, starting along the weld or very close to the weld line. In table 6 a sunmam3r of the number of complete plastic cycles to collapse and the failure mode of the specimens is reported. TABLE 6 ~ER OF PLASTIC CYCLES AND FAILURE MODES OF BCC8 SPECIMENS Cyclic Behaviour of Beam-To-Beam Welded Connections COMPARISONS AND OBSERVATIONS 329 Panel zone and beam rotations. The contribution of the total (elastic + plastic) panel zone deformation to the global rotation of the specimens has been, throughout the experimental program: remarkable (in average equal to the 80% of the total imposed rotation) in the BCC5 specimens, having the smallest column section (HE160B); less significant (in average equal to the 65% of the total imposed rotation) for the BCC6 specimens, with intermediate column section (HE200B); minor (40-50 % of the applied rotation) in the BCC8 specimens, characterised by the largest column section (HE240B). Consistently, the plastic rotations registered in the beam have been minor for the BCC5 specimens, comparable to the panel zone rotations in the BCC6 specimens, larger for the BCC8 specimens. The values of the total rotation capacity, which, in the increasing amplitude test, reaches 0.064 rad for the BCC5 specimen, 0.053 rad for the BCC6 specimen and 0.046 rad (at maximum strength decrease not less than 90%) for the BCC8 specimen, correspond to low values of beam plastic rotations, respectively equal to 0.0057, 0.0175 and 0.0242 rad for the three specimens, thus confirming that large rotations can be experienced thanks to column web panel deformations. Effect of column size on the cyclic behaviour and failure mode The BCC5 specimens, even though able to experience high deformation levels, have shown brittle failure modes in all the cyclic tests, with hysteresis loops practically overlaid and no degradation of the flexural strength up to the very last cycle, where a sudden decay of the carrying capacity occurred due to fracture, generally developed in the proximity of the weld. On the contrary the BCC8 specimens have exhibited a typical ductile behaviour, with formation of a well defined plastic hinge in the beam starting from the first plastic cycles, and a gradual decrease of the peak moment at increasing number of cycles up to the collapse. The BCC6 specimens displayed a behaviour sometimes closer to the BCC5 ones (tests BCC6C and BCC6D), sometimes to the BCC8 ones (tests BCC6A, BCC6B and BCC6BB), depending on the applied loading sequence. Also with regard to the final collapse of the specimens, in the former cases it involved fracture in the beam starting at or close to the weld location, while in the latter cases it was due to the cracking in the buckled zones of the beam flanges. Effect of the loading history The different cyclic histories applied to the specimens have evidenced the dependence of the plastic deformation capacity on the loading histories. The cmnulated plastic rotations computed on the basis of the test data, result in highly variable values for the BCC5 specimens (d/~l,c,m=0.65-0.27 rad), while the BCC8 specimens show, except test C, similar values for all the tests (d~pl.cum=0.48-0.55 rad). In the tests B and BB, in which the BCC5 and BCC6 specimens have been subjected to the same loading history (constant amplitude rotation d/H=7,5%), the BCC5 specimens have shown the same failure mode and similar number of cycles to failure (test B: 5, test BB: 4). On the contrary the BCC6 specimens showed a different behaviour, since the BCC6B specimen experienced 11 plastic cycles and collapsed due to crack in the beam flange at the plastic hinge location, while the failure of the BCC6BB specimen occurred after 6 plastic cycles due to fracture in the beam flange along the weld, propagated also in the web. Similarly to the BCC5 specimens, the BCC8 specimens which have been subjected to the same loading history (BCCSB and BCCSE, constant cycle amplitude, d=37.5 ram) have shown the same collapse mode and close values of the number of plastic cycles (test B:16; test E:15). . Finally, the weld was treated linearly at this early stage of the investigation. A non-linear (multi-linear, elasto-plastic) analysis featuring isotropic hardening effects was performed. The. 1.52 316 482 Experimental set-up, instrumentation plan and loading histories The test set-up, represented in figure 1, mainly consists in a foundation, a supporting girder, a reaction r.c related to occurrence and spreading of local buckling in the beam flanges and web. A well defined plastic hinge in the beam has formed in all the tested specimens. In the test C, where the specimen