Influence of column web stiffening on the seismic behaviour of beam-tocolumn joints A.L. Ciutina & D. Dubina The Politehnica University of Timisoara, Romania ABSTRACT: The present paper summarises the experimental results on column web panels, carried out at the Politehnica University of Timisoara, Romania with the scope of investigating the influence of different column web stiffening solutions on the performance of beam-to-column joints of Resisting Frames. The response parameters, such as resistance, rigidity and ductility have been monitored. Five different types of panel web stiffening are compared in regard to a reference test. A comparison of the experimental tests to the mathematical models given by and EUROCODE 3 - Part 1.8 are also presented. 1 INTRODUCTION The lateral loads acting on MR Frames (e.g. wind and earthquake loads) introduces high shear forces into the Column Web Panels (CWP) of joints. This way, rather than forming plastic flexural hinges in beams or columns, the dissipation of input energy is located into the CWP. Depending also on the connection typology, the CWP can develop a part or even the entire rotation capacity of the joint. In case of internal joints, the shearing effect is predominant, due to opposite moments acting at the ends of the connected beams. The resulting web deformations have an important effect on the overall structural response, leading sometimes to important secondorder structural effects. The various experimental tests, conducted lately on CWP have revealed important observations regarding to their behaviour. A sheared CWP develops a maximum strength significantly greater than the yielding strength, due to its strain-hardening effect. However, large inter-story drifts are required for attaining the full-resistance of the CW. That is why their maximum shear capacity is not easily attained. On the other hand, in the inelastic range, the CWP show a very ductile behaviour, both for monotonic and cyclic loading. In case of cyclic loading, the hysteretic loops are very stable even for large deformations. et al. (1971) proposed a mathematical model for the equivalent non-linear springs representing the CWP behaviour for structural analysis, model that was adopted in several American Building Codes. A three-linear model has been derived for modelling the CWP behaviour in terms of shear force (V) and panel distortion (γ). The current provisions of the part 1.8 of the EUROCODE 3 () permit the calculation of the initial stiffness and the resistance of the CWP. For design purposes the maximum allowed column web stiffening is limited to a single supplementary web plate, with a maximum thickness equal to that of the column web. However, other studies (Dubina et al. ) have shown that in the case of cruciform cross-sectional columns, the entire shear area (the web and the two adjacent column flanges) should be considered in shear. In many cases, the beam-to-column joint resistance is dependent on the resistance of CWP. Consequently, the reinforcing of CWP should be necessary. Additional column web doubler plates can increase both strength and stiffness of the joint. This fact may influence also the ductility of the joint. For this purpose, an adequate design is needed for finding the best resistance/ductility ratios. Twelve experimental tests have been carried out at the Politehnica University of Timisoara, in order to investigate the influence of column web stiffening to the overall joint behaviour. Five different types of web stiffening are compared to a reference (unstiffened) case, as later described. For each case of stiffening, monotonic and cyclic loadings have been applied. GENERAL CHARACTHERISTICS OF A CWP ELEMENT The column web panel can be considered basically a rotational element that transfers moments between columns and beams framing into the specified joint. The boundary forces (see Figure 1) on a panel zone
Nb Mb Vb Nc Vc Vc1 Nc1 hc Mc Mc1 Vb1 Mb1 Panel Boundary Forces Nb1 hb IPE 36 beam HEB column IPE 36 beam Stiffening Equivalent Panel Shear Forces Figure 1. Boundary forces and shear forces of a CWP. can be transformed into an approximate equivalent shear force as follows: M M V V b1 b c1 c = = z M cwp z (1) where: M b1, M b, V c1, V c, are the boundary internal forces, and z - the level arm of the CWP. The properties that define the design resistance values of a panel joint are the rotational stiffness and the yield shear strength. The hysteretic rules and the rotation capacity are the main supplementary parameters that characterise the cyclic behaviour. By equation (1), the panel zone force versus the panel zone deformation can be transferred into the panel zone moment M cwp versus the panel zone distortion, γ. this case was of mm. - Two distanced plate specimens (CP-IIPD series). In this case, the two doubler plates are distanced from the column web, resulting a width of 6 mm. This solution is largely used in USA, but not recommended by the design guides in Europe. - In-filled concrete specimens (CP-C series) the web panel is stiffened by reinforced concrete. The steel stirrups (Φ=6mm) are welded on the column web and attached to longitudinal reinforcements: Φ 1 mm re-bars (see Figure ). All stiffened series have been compared to a reference series, having no CWP stiffening series CP-R. For all specimens, hot-rolled sections of IPE 36 and HEB have been used for beams and columns respectively. Figure shows the disposition of the stiffening elements, in regard to the column section. The beam-to-column connections have been made by means of full-penetration double bevel butt welds, with quality control. In the case of doubler plates, fillet welds have been used. The thickness for all the doubler plates is 1 mm. No continuity plates have been placed on columns. 3 SPECIMENS, TESTING SET-UP AND LOADING PROTOCOL The scope of the tests was to observe the influence of different types of stiffening, into the global behaviour of CWP. In this purpose, five typologies of stiffening have been tested, as follows: - One-side doubler plate specimens (CP-IP series) they represent the common European CWP stiffening method. The width of the doubler plate is of mm, with a thickness of 1 mm. - Two-sides plate specimens (CP-IIP series) composed of two identical doubler plates welded symmetrical to the column web. - Two larger plate specimens (CP-IIPL series) comprised two identical plates, but larger than those of CP-IIP series in order to cover the entire column web, until the root fillet. The width of the plates in Figure 3. Testing set-up. The testing set-up is presented in Figure 3. Statically, the joint is simply supported a t the two ends of connected beams and pinned at the column base. This way, symmetrical and opposite moments forms in beams, causing shearing of the CWP. No axial force was introduced into the column. The corresponding panel moment - M cw was computed at CP-R CP-IP CP-IIP CP-IIPL CP-IIPD CP-C None One plate Two plates Two plates Two plates Concrete larger distanced Figure. Typologies of CWP specimen series cross section through the columns.
the outer face of the column, starting from the top force, while the panel distortion - γ was calculated starting from the diagonal LVDT transducers placed on the web panel. For each series, two types of loading have been applied, monotonic (M) and cyclic (C) respectively. For both types of loadings, test were conducted by displacement control. In case of cyclic loading, the ECCS testing procedure (1986) was used. Prior tests, samples of material - concrete and steel (column web and flanges and doubler plates) - have been tested in order to establish the actual mechanical characteristics of the CWP elements. Table 1 gives the characteristic values of the tensile tests. For the design of joints S35 steel was considered. In case of concrete, a nominal compressive strength of 31.7 N/mm² has been obtained for C5/3design concrete. Table 1. Mechanical characteristics of steel elements. Element f y [N/mm²] f u [N/mm²] A [%] Column Web 31.75 451.78 35.45 Flange 6.47 45.33 3.91 Doubler plate 76.8 47.5 9.3 Stirrups Φ6 mm 68.33 538.87 7.6 4 TESTS RESULTS Table presents the main results given by the experimental tests. The following notations have been used: M max the maximum panel moment resistance; S j,ini the initial rigidity for monotonic curves, or for envelope curves in the case of cyclic specimens; γ y, M y the yielding panel distortion and resistance respectively, computed on the basis of ECCS procedure; γ max, - the maximum panel distortion, computed for a drop of % in the maximum moment M max ; S j,hard the strain hardening slope of the graphs, computed on envelopes in case Table. Principal results of the CWP tests. of cyclic tests; Energy the total cumulated energy in cyclic loading; No. Cycl. the number of cycles performed to failure. 4.1 Monotonic tests Monotonic Responses CP-IIPD-M CP-IIPL-M CP-IP-M CP-IIP-M CP-C-M CP-R-M rotation [rad.].5.1.15. Figure 4. Results of the monotonic tests. Main parameters describing the monotonic tests are given in Table while their graphical responses are given in Figure 4, under the form of M γ responses. As expected, the yielding and the maximum resistant moments increase from the reference specimen CP-R-M to the one having the maximum shear area CP-IIPD-M. This tendency is not so evident in the case of the initial rigidity - S j,ini, which is practically constant for R, IP and IIP monotonic specimens, but significantly greater for specimens with larger and distanced doubler plates. Also, the in-filled concrete increases the initial rigidity of the panel, but it has no significant implication on the resistant moment. However, Simoes & al (1999) proved that the total encasement of the steel column in concrete provides important increases of both initial rigidity and the moment resistance of CWP. By increasing the initial rigidity, and applying the ECCS procedure there has been noticed an increase on the values of the yielding moment. The strain-hardening rigidities are also higher for the specimens having the greater shear area. A very Specimen M max S j,ini [knm/rad] γ y [mrad] M y γ max * [mrad] S j,hard [knm/rad] Energy [knm rad] No. Cycles Poz Neg Poz Neg Poz Neg Poz Neg Poz Neg Poz Neg CP-R-M 19.14 3139 3.36 117.56 14.4** 931 --- --- CP-R-C 4.84 17.56 3188 3195 3.84 3.67 19.3 19.71 79.7 84 783 99 89.55 4 CP-C-M 44.76 49635.76 148.7 167.** 383 --- --- CP-C-C 19.4 9.89 5613 5194.91.88 158.73 16.97 74.8 84 179 191 971.4 44 CP-IP-M 79.3 3143 4.51 163.17 89.3** 948 --- --- CP-IP-C 6.58 68.51 3838 4674 4.13 3.37 175.84 178.93 54.5 56 111 1517 365.94 CP-IIP-M 33.8 3169 5.83 5.44 177.3 5 --- --- CP-IIP-C 37.57 314.49 4674 4158 5.4 4.8 17.74 1.9 68 7 117 177 458.4 CP-IIPL-M 413.47 4975 5.81 65 136.5 1574 --- --- CP-IIPL-C 387.14 39. 5713 4454 5.39 5.9 79.34 85.6 61 61 1758 595 387.5 19 CP-IIPD- M 43.5 58418 4.66 84.7 84.3 1459 --- --- CP-IIPD-C 49.48 431.43 6143 6355 4.75 4.78.18 311.79 46.4 46.1 3 11 338.75 1
small strain-hardening slope is observed in the case of concrete specimen. Although there have been observed very important distortions of the CWP (until.15 rad.), no failures occurred into panels. Practically, the actuator stroke limitation (approximately 4 cm) set the end of monotonic experiments. The only exception was for the case of CP-IIPL-M specimen, where the connection failed, but side-away of the CWP. Anyway, the monotonic tests proved a very good and stable monotonic behaviour, with panel rotations of more than 1 mrad. 4. Cyclic tests Table also presents the values of the main parameters for cyclic behaviour, both for positive and negative branches. Figure 5 shows the panel moment M versus the panel distortion γ curves, scaled at the same amplitudes. As it can be seen, all the specimens prove very stable cyclic behaviour, without degradation until failure. Generally, the same ascending tendency of the maximum resistance is noticed, as in the case of monotonic tests. However, a drop up to 1% of the maximum moments was observed for the cyclic specimens. On the other hand, the cyclic behaviour present a more rigid behaviour for the elastic part. But, this time the rigidity increases proportionally to the shearing area. So, major differences are recorded for the cases of CP-IP-C and CP-IIP-C specimens (both branches) in regard to the corresponding monotonic specimens. As a consequence of this increased stiffness, the yielding moment consequently increased. The concrete stiffened specimen presents also for cyclic loading a particular behaviour. Although it reported almost the same elastic rigidity as the monotonic specimen and a normal degradation of the maximum moment, it presented a visible cyclic degradation of the moment and plastic rigidity during the cycles at equal amplitude. This phenomenon was due to the degradation of the concrete in the first cycle of increased amplitude. Similar to the elastic rigidity, the hardening rigidity of cyclic tests increases with the stiffening area, as presented in Table ; Even if the values of positive and negatives branches are different, they are generally greater than those obtained in the monotonic tests, for CWP specimens doubled by steel plates. A reverse tendency could be observed in the case of the reference specimens. Mainly, two failure modes have been observed to cyclic specimens: - The ductile mode, observed for the case of reference and the CP-C-C specimens. For these cases, small cracks begun on the kinking zones, situated on the external parts of the beam flange-tocolumn welds. As the distortion amplitude increased, the cracks advanced on the entire column flange thickness and later continued into the column web, on lines parallel to column flanges. Meantime, a crack developed horizontally on a median line of the panel. The final crack pattern formed the H letter shape, as shown in Figure 6 - a. In addition to the steel cracks, concrete cracks were observed even from the beginning cycles in the case of concrete infilled specimen. They were formed horizontally along the boundaries of CWP. At higher values of distortion stirrup shear failures were noticed. - Brittle failures, observed for the other tests, excepting the CP-IIPD-C specimen. For these specimens, the initial cracks formed near the doubler plates at the level of column flanges. They rapidly propagated vertically, along the doubler plates, as shown in Figure 6 b, and produced the degradation of resistance. In case of CP-IIPD-C specimen, the failure occurred by brittle rupture of a beam flange, near the weld to the column flange. The maximum panel distortions followed generally the joint failure mode. The greatest CP-R-C -.1 -.5.5.1 - - - CP-IIP-C -.1 -.5.5.1 - - CP-C-C -.1 -.5.5.1 - - - CP-IIPL-C -.1 -.5.5.1 - - CP-IP-C -.1 -.5.5.1 - - - CP-IIPD-C -.1 -.5.5.1 - - - - - Figure 5. Panel moment panel distortion for cyclic tests.
(a) Figure 6. Ductile failure (a) and brittle failure (b) modes for cyclic specimens. distortion values (84 mrad) have been obtained in the case of CP-R and CP-C specimens. However, all the specimens proved a rotation greater than 35 mrad, value considered as safe by the modern antiseismic design codes. Also, it has to be noticed, that even in case of the stronger stiffened specimen, the panel zone properly worked and dissipated energy. The value of the cumulated energy depends on a several parameters, i.e. the number of cycles performed, their amplitude etc. The maximum dissipated energy was obtained for concrete in-filled specimen, followed by the reference specimen, due to their ductile behaviour and specific failure mode. In comparison, all the other specimens experienced a energy dissipation of about half. Although they had a greater resistance, their failure mode was brittle. Briefly, the cyclic tested specimens proved a very stable behaviour for all the cases, their resistances increasing proportional to the shear area. They also reported a good ductility and strain-hardening ratios. However the failure was brittle in the case of stiffened specimens. 5 COMPARISON TO THEORETICAL MODELS In this section, the results of the monotonic tests are compared to the results obtained using two theoretical models for the shear behaviour of a CWP. The s model and the design approach of EUROCODE 3 Part 1.8 have been considered. The Part 1-8 of EUROCODE 3 allows the calculation of the initial stiffness - S j,ini and of the resistant moment - M wp, for a single part of the connection as follows:.9 f A y, Rd vc M wp = z () 3.38EzAvc S j, ini = (3) β where: A vc - the shear area of the column (limited in the case of doubler plates to a single plate of the same thickness of the column web); f y,rd - the yielding resistance of the column web; z - the level arm of the panel zone; E - the Young s Modulus for steel, and β - the transformation parameter ( in the case of opposite moments formed on beams). On the other hand, s model allows a three-linear modelling of the whole column web panel as follows: f A y, Rd vc M 1 / y = z P P (4) y 3 3.1 f b t y, Rd cf cf M sh M + y 3 (b) = (5) S j e = GA z vc, (6) j, 1 1.4 Gb t cf cf S = (7) S j = G A z st vc, (8) where: M y, M sh - the moments corresponding to the yielding and to the strain-hardening of the panel; S j,e, S j,1, S j, the panel rigidities corresponding to elastic, post-elastic and strain-hardening branches; P, P y - the value of the column axial load and the yielding column axial load; b cf, t cf - the column flange width and thickness respectively; G, G st - the shear modulus and strain hardening shear modulus of column web. Although the formulae for the two different approaches starts from the same mathematical model, some differences exist i.e.: - the s approach for yielding moment takes into account the real axial load into the column, while the EC 3 considers a general coefficient of.9, corresponding to an axial load of.4 P y ; - the effective shear area taken into consideration is different for the two approaches. While EC3 considers the entire shear area of the hot-rolled profiles (including the root fillet), the model uses a reduced value of (h c -t cf )t wc ; - the height of the level arm used is also different for the two approaches: the EC 3 accounts for a height of the level arm of h b -t bf, while the level arm used by is equal to the beam height - h b. Table 3 presents the values of yielding moments and initial rigidities obtained by the two theoretical approaches and those obtained by monotonic tests (only steel specimens). The same comparison is shown graphically in Figure 7. It should be men-
5 1 CP-R 5 CP-R-M..4.6.8.1 CP-IIPL 5 1 5 CP-IP CP-IP-M..4.6.8.1 5 4 35 5 1 5 CP-IIP CP-IIP-M..4.6.8.1 CP-IIPD CP-IIPL-M 1 Rotire CP-IIPD-M..4.6.8.1..4.6.8.1 Figure 7 Graphical comparisons of monotonic results to the theoretical models tioned that in the case of EC3 calculations, the full shear area have been considered. Generally good agreement is found for the resistant moments M R, computed by the two approaches. On the other hand, the initial rigidity is systematically greater in case of numerical results, showing important differences to the experimental ones. It results, that in case of the stiffened specimens, a different approach for computing the initial rigidity should be used. In the post-elastic range, the s model proved to be safe until a high level of distortion of the CWP. Table 3. Comparison of experimental results to theoretical approaches. Spec. Resistance- M R Initial Stiffness S j,ini kn m kn m/rad Exp. EC3 Kraw. Exp. EC3 Kraw. CP-R 117.56 18.3 15.1 3139 4139 48776 CP-IP 163.17 164.34 148.19 3143 61311 7553 CP-IIP 5.44 3.78 191.17 3169 879 9331 CP-IIPL 65. 36.97 31.8 4975 963 11656 CP- IIPD 84.7 54.35 54. 58418 9573 1471 6 CONCLUSIONS important increases for the initial rigidity. The specimens having one (CP-IP) or two (CP-IIP) partial doubler plates have shown a brittle failure, for small gains in resistance. A more increased resistance capacity and initial rigidity was obtained for total web stiffening CP-IIPL specimen. The specimen CP-IIPD proved a very high initial rigidity, and a resistance greater than of the connected beams. Generally a linear relationship exists between the total shear area and the moment resistance of a CWP. Panel stiffening affects also the elastic rigidity of a joint, increasing non-linearly with shearing area. For design purposes, total doubler web depth plates are recommended. They provide an effective increase in shear resistance and rigidity, while they can undergo large inelastic distortions. For the cases where the CWP should be totally rigid, distanced doubler plates should be used. The two theoretical models considered as a base of comparison to the experimental values provided good agreements in what concern the moment resistance. However, important differences have been obtained for the initial rigidity. Also, on the case of EC3 approach, it is recommended that entire shear area (profile shear area plus entire area of the doubler plates) should be considered for the computation of shear resistance of a CWP. The objective of the experimental tests was to observe the influence of different stiffening solutions of the column web panel on the global behaviour of a beam-to-column joint. A comparison to two theoretical approaches is also presented. The behaviour of the CWP to seismic loads remains stable even when the panel stiffening area becomes important. All the specimens developed high values of plastic rotations, with important strain-hardening effects. The CP-R and CP-C specimens proved the best rotation capacities, energy dissipation and a ductile failure mode. Concrete stiffening of the panel is not so effective in what concern the panel resistance, but introduces REFFERENCES H, Bertero & V.Popov E. (1971) Inelastic behaviour of steel beam-to-column subassemblages. Report No. EERC 71/7, University of California, Berkley, CA; 1971 Dubina, D., Ciutina, A., & Stratan, A. (). Cyclic Tests on Bolted Steel and Composite Double-Sided Beam-to-Column Joints, International Journal of Steel & Composite Structures, Vol., No., Apr., pp.147-16. pr EN 1993 1 8 () EUROCODE 3: Part 1.8 Design of steel joints. CEN, European Committee for Standardisation, February Simões, R., Simões da Silva, L., & Cruz, P., (1999)
Experimental Models of End-Plate Beam-to-Column Composite Connections., Proc. nd European Conference on Steel Structures, EUROSTEEL.99, Praha, Czech Republic, 65-69