280 Y. Shi and J. Jing The end plate connection can fail with the following failure modes: (i)Fracture of bolts in tension zone; (ii)end plate yielding; and (iii)compression between the bolted plates diminishing. The moment resistance and failure modes may vary with the connection configuration. In this study, the compression C between the contact surfaces and the bolt tension T at the tension zone versus applied bending moment were obtained around each bolt. It should be noted that when there is no bending moment applied, the compression C and the bolt tension T is equal, that is C = T. It is also noted that with the increase of bending moment, the bolt tension T will increase, but compression C will reduce. The variations of the tension T and compression C with applied moment for different connections were shown in Fig. 5b-~5f. It can be seen that for the flush end plate(Fig. 5b), the higher tension was developed in the first row of bolts even the applied moment is small and the connection fails with bolt fracture. The maximum moment resistance of the flush end plate connection is much less than that of the corresponding extended end plate connection(Fig. 5d). Fig. 5c to 5e compares the effects of end plate thickness. When the thin plate is applied(Fig. 5c, t = 0.5d, where d is the bolt diameter), the end plate deformation and prying force is significant and connection fails with end plate yielding. When medium thickness of end plate is applied(Fig. 5d, t = d), the connection resistance largely depends on the tension resistance of the second row of bolts, where the maximum tension force is developed. However, if the extended part is stiffened(Fig. 5f), the maximum tension can be developed in both the first and the second rows of bolts and the higher moment resistance is achieved. When the thick plate is applied(Fig. 5e), the compression between the bored plates diminishes very quickly and the connection may fail by the separation of the bolted plates. However, it is difficult or impossible to calculate the bolt tension T and/or compression C by a simple method. From the compression C and bolt tension T, the force Nt produced by bending moment and resisted by each bolt can be given by Nt = T- C. The distribution of Nt among the bolt rows are compared under the maximum bending moment(Table 1 and Fig. 6). It is noted that for the flush end plate connection(Fig. 6a), the tension force distribution is similar to traditional design model. For the extended end plate connection, the tension force between the beam flanges more or less distributed triangularly, but on the extended part, the tension force Ntl varies with the plate thickness and stiffening. The maximum Nt appears at the second row of bolts instead of the first row of bolts(Fig. 6b). Since the actual tension force on the extended part is far less than that calculated from the traditional design model(Fig. 2), the end plate yielding may more likely happen around the second row of bolts, rather than the first row of bolts as predicted by the current code of practice (CECS102:98, 1998). Even the extended part is stiffened(Fig. 6c), the tension force on the first row unlikely exceeds the force on the first row, unless the end plate is extremely thick. Ntl (a)Flush (b)Unstiffened (c)Stiffended (a)Unstiffened (b)Stiffened Figure 6: Bolt force distributions Figure 7: Proposed design model Design Moment Resistance of End Plate Connections TABLE 1 BOLT TENSION UNDER MAXIMUM BENDING MOMENT 281 Extended End-plate Unstiffened, t = 10mm Unstiffened, t=20mm Unstiffened, t=40mm Stiffened, t=20mm First row Ntl(kN) 26.3 97.5 105.3 137.9 Second row No(kN) 42.2 139.0 130.9 141.2 Ntl/Nt2 0.62 0.70 0.81 0.98 DESIGN PROPOSAL From the above analysis, itis concluded that the traditional tension distribution model may not be applicable to the extended end plate connections. The necessary revision is proposed and recommended in this paper. Since the tension distribution Art for the flush end plate connection is close to the traditional design model, the force resisted by any row of bolts can still be calculated by Eq.(1), while for the extended end plate connections, the following design procedures were proposed" (1)When the extended part unstiffened, the bolt force distribution can be assumed as Fig. 7a and the second row of bolts is supposed subject to the maximum tension, and tension force on any row in the tension zone can be given by Nt2 = m YlY2~l 2 + Z Yi 2 +Y,,Y,-1 i=2 Utl = Nt2~:l/~: 2 Nt~ = Nt2 Y,/Y2 where ~:1 = (ha + h2 - Ya )/ha and ~z = Y2/h2 y, distance from the ith bolt row to the center of bolt group; M bending moment applied on the connection; n number of bolt rows; m number of bolt columns; hi, h/ length of the extended part and the distance from the beam flange to the center of bolt group respectively. (2) When the extended part stiffened, the bolt force distribution can be assumed as Fig. 7b and the tension force on the bolt rows adjacent to the beam flange in tension is equal. Tension force on any row in the tension zone can be given by Nt2 "- m YlY2 + E Yi 2 +Y.Y,-1 (3a) i=2 Ntl = Nt2 (3b) N~ = Nt2 Yi/Y2 (3e) (2a) (2b) (2c) 5O 5O u u L i,i i *-i4- I I 1 .4. 4- I I I I i. i ~ . I I I~ ~- [ oo [ Figure 8. .Io I r o Example 282 Y. Shi and J. Jing The proposed design model by this paper is compared with the traditional design model, taking a typical extended end plate connection as an example(Fig 8). The applied bending moment is M = 160kNm, and the design results are listed in Table 2. It can be seen that because the maximum tension Art is assumed on the first row of bolts in the current code of practice, the larger flexible moment generated in the end plate happens on the extended part. Therefore, thicker end plate is required to prevent the connection failing in end plate yielding mode. However, if the bolt tension force is calculated by the proposed method where larger tension force appears on the second row of bolts instead of the first row, the moment generated in the end plate is reduced significantly. As the results, less thick end plate is required. TABLE 2 DESIGN RESULTS Bolt Tension(kN) First row(Nt0 Second row(Nt2) Bolt Parameters End Plate Thickness(mm) Current Code 95.2 60.3 Grade 8.8, M22 27 This Paper 60.0 91.2 Grade 8.8, M22 22 CONCLUSION A 3D finite element model was established in this paper to investigate the end plate connection behaviour. The analysis model is verified by the test results. Based on the comparison results obtained from analyzing some typical connections, it is concluded that the traditional design model may not be applicable to the end plate connection design. A revised model is proposed for end plate connection design and a formula is derived for evaluating the bolt tension distribution. REFERENCE Brown, D. G., Fewster, M. C., Hughes, A. F. and Owens G. W.(1996). A New Industry Standard for Moment Connection in Steelwork. The Structural Engineer 74:20, 335 - 342. CECS102:98 (1998). Technical Specification for Light Gauge Steel Structure of Low Rise Buildings with Portal Frames. Association of China Engineering Construction Standard, Beijing. Gebbeken, N., Rothert, H. and Binder B.(1994). On the Numerical Analysis of End-plate Connections. Journal of Constructional Steel Research 30, 177 - 196. Jenkins, W. M., Tong, C. S. and Prescott, A. T. (1986). Moment-transmitting Endplate Connections in Steel Construction, and a Proposed Basis for Flush Endplate Design. The Structural Engineer 64A:5, 121 - 132. JGJ82 91 (1992). Specification for Design and Construction of High Strength Bolt in Steel Structures. Ministry of Construction, Beijing. Sherbourne, A. N. and Bahaari, M. R.(1994). 3d Simulation of End-plate Bolted Connections. Journal of Structural Engineering 120:11, 3122 - 3136. THREADED BAR COMPRESSION STIFFENING FOR MOMENT CONNECTIONS T. F. Nip 1 and J. O. Surtees 1 1 School of Civil Engineering, University of Leeds, Leeds LS2 9JT, UK. ABSTRACT A new means of providing local column reinforcement in the compression zone of high moment capacity end plate connections, using threaded bars, has been developed. Conventional welded plate stiffening is difficult to fabricate by automatic processes and is particularly expensive when used on site in structural upgrading schemes. In the new approach, a system of threaded bars is locked against inner flange faces of the column to transmit the horizontal compression force from incoming beams. Joint performance has been studied in a programme of tests on simple compression specimens and full scale beam/column/beam joints. The influences of concrete encasement and steel .sleeving on threaded rod capacity have also been investigated. In a typical connection using high strength threaded bar only, enhancements of column web bearing capacity in the order of 300% have been demonstrated. The validity of using simple compression tests to represent the compression zone of beam/column joints is examined by comparison with full scale connection tests. It is concluded that serviceability and ultimate conditions can be met fully by the proposed form of connection. Experience of fabricating and assembling the test specimens indicates that significant savings are possible in comparison with the true cost of providing welded stiffeners. KEYWORDS Steel frames, moment connections, bolted joints INTRODUCTION Beam-to-column end plate connections have been used extensively in multi-storey construction for resisting moments from wind and gravitational loading. UK and European design codes now provide the necessary 'continuous construction' background for member and connection design but magnitude of bending moment which may be transmitted by conventional end-plate connections is somewhat limited. This arises from the inherent incapacity of typical column flanges to resist normal force, even when reinforced locally by conventional welded stiffening. Demands for longer clear spans and lower 283 284 T.F. Nip and J.O. Surtees floor construction depths have almost outstripped the capability of such connections. For instance, the maximum moment resistance of connections appropriate to a beam of depth 650mm is likely to be only 25% of the moment capacity of the beam. Moment capacity may be increased by increasing either the bolt diameter or the total number of bolts in the tension cluster. The former of these has been explored at Leeds University, using backing angles to reinforce the column flange (Grogan and Surtees (1995) and Grogan and Surtees (1999)). The second approach entails increasing the number of bolt rows or the number of bolts per row. Several investig-ators have examined these options (Grundy et al (1980), Murray and Kukreti (1988) and Murray (1988)). The second option was explored without recourse to column stiffening but has been investigated more recently at Leeds University in a Science and Engineering Research Council (now EPSRC) supported project (Surtees and Yeung (1996)). In the latter investigation, which applied particularly to double-sided connections, local bending of the column flange was reduced by linking beam tension flange forces across the column via socketed couplers placed between opposing tension bolts. In tests on full-scale specimens, improvements in moment capacity up to 200% were observed, compared with less than 40% when using conventional welded reinforcement in the tension zone. The work reviewed above has focused on tension zone stiffening and a principal feature has been the use of bolted forms of stiffening. Recently at Leeds University, the potential use of bolted compression zone stiffening was examined in depth and this paper presents details of tests on a particular form of stiffening element developed in that investigation, namely, the threaded bar stiffener. USE OF THREADED BAR COMPRESSION ZONE STIFFENING The term threaded bar is used to describe continuously threaded stock material readily available in various diameters and material grades at lengths up to 3m. It is usually cut to precise shorter lengths in the manufacturing process. In the present context, plain round bar with minimum threading to satisfy installation requirements is equally acceptable, though not necessarily cheaper than threaded bar. Two forms of threaded bar compression stiffening element were used in the tests. The first consists of a short threaded bar with end nuts which fits between, rather than passes through, the column flanges. Blind flange holes may be used to locate the stiffener or, alternatively, thin punched retaining plates may be suspended from nearby end plate fixing bolts. The second form also has internal nuts but passes through the flanges to engage outer nuts and is therefore able to act as tension zone stiffening in the event of moment reversal. Figure 1: Forms of threaded bar compression stiffening. Threaded Bar Compression Stiffening for Moment Connections 285 This latter form must be positioned with appropriate installation clearance from beam flange surfaces. Both forms may be combined to provide compression zone stiffening in non-reversing or partially reversing moment connections. Figure l(a) shows a possible configuration for a non-reversing moment connection. Four threaded bar stiffeners are placed in line with the compression flange. On the tension side, a larger number of (off-line) stiffeners is necessary because of weaker participation of the web and presence of prying. In the full moment reversal connection shown in Figure 1 (b), tension zone requirements determine stiffener provisions at both flange levels. TEST PROGRAMME The primary objectives of the tests were (i) to verify the feasibility of threaded bar stiffening and (ii) to investigate the force distribution in the connection elements in order to establish design guidelines. To this end, the following aspects were studied in the tests: 9 threaded bar properties 9 bar configuration and size effects 9 influence of concrete encasement Because of the large number of tests, all involving heavy sections, it was decided to confine testing mainly to compression zone specimens. Some tests on full connection specimens were, however, carried out for comparison purposes. Compression tests Compression zone check calculations invariably require that column web stiffening be provided. When this is done, the available compression zone capacity is usually much in excess of requirements and connection failure occurs elsewhere. In devising a new form of stiffening, the possibility of a more balanced provision for compression and tension zone failure might be allowed to advantage. However, most specimens tested in the present series had strengths and stiffnesses well in excess of what might be termed a minimum compression zone performance. A full description of the isolated compression zone test specimens is given in Table 1. A threaded bar diameter of 24mm was used for most of the tests in recognition of the fabrication industry preference for M20 and M24 bolts. Larger sizes of threaded bar were tested both to examine their efficiency as concentrated compression stiffening and to detect potential assembly difficulties. In all cases except CB2 the threaded bar stiffening was prepared from plain material, leaving a small unthreaded central portion to accommodate ER strain gauges for direct force measurement. Test CB2 used commercial threaded bar. All bars were calibrated prior to testing and the characteristics of the two forms were compared for control purposes. The compressive test load was restricted to 4110kN maximum by the capacity of the loading frame. The full connection tests described below used UB533x210x101kg (Grade 50) beam material throughout, corresponding to a nominal maximum compression flange force of 1730kN. Rather than represent the beam flange thickness correctly in the test rig and thereby limit the maximum test load, a 40x40x215 steel block and curtailed 25 thick end plate was used to load each side of the specimen. In this way, the true capacity of the stiffened column web was measured, rather than that of the beam flange. Dial gauges, LVDT deflection gauges and ER strain gauges were used to measure displacements and strain distributions in the test specimens. The spread of yield was monitored using heat applied resin coatings on the columns. A typical set up is shown in Figure 3. 286 T.F. Nip and J.O. Surtees TABLE 1 DESCRIPTION OF ISOLATED COMPRESSION ZONE SPECIMENS Test Column description Size Length (mm) CB 1 UC 254x254x73 700 CB2 UC 254x254x73 460 CB3 UC 254x254x73 460 CB4 UC 254x254x89 700 CB5 UC 305x305x118 700 CB6 UC 305x305x118 700 CB7 UC 305x305x118 700 CB8 UC 305x305x118 700 CB9 UC 305x305x118 700 CB 10 UC 305x305x 118 700 CC4 UC 305x305x118 700 UC 305x305x 118 700 Stiffener details Dimension Type (Fig. 2) ~24 HTS bars A M24 grade 8.8 A threaded bar ~24 HTS bars A ~24 HTS bars B ~24 HTS bars B ~24 HTS bars C ~24 HTS bars C ~30 HTS bars A(HP) ~t~45 mild steel D(HP) ~24 HTS bars B ~24 HTS bars A ~24 HTS bars A Further reinforcement 25 mm thick backing plates 1.5" (0.25" thick ) HTS tubes Grade C60 concrete Grade C20 concrete HP: with Hanging Plate Figure 2: Threaded bar configurations in Table 1 Threaded Bar Compression Stiffen&g.for Moment Connections 287 Figure 3: Typical setup for isolated compression test Cruciform tests Two full size beam/column/beam connections, FB2 and FB4, were tested. The direct compression tests excluded interaction from nearby shear or tension zone forces and the whole-connection tests allowed the effect of these omissions to be studied. FB2 was an over-stiffened specimen with 8 threaded bars in the compression zone whereas FB4, with only 4 threaded bars placed directly against the incoming beam flanges, was marginally under-stiffened in relation to the tension capacity of the connection. Specimen FB2 was the whole-connection equivalent of CB5. Both specimens were constructed from UC305x305x118kg/m, UB533x210x101kg/m and 705x305x35 endplate (see Figure 4). Figure 4: Test specimen FST4 288 RESULTS T.F. Nip and J.O. Surtees Table 2 summarises the isolated compression zone test results. Column web bearing capacity was increased substantially by the stiffening. Specimen CB5, which would be typical for beam depths of up to 600mm, was able to sustain a force equivalent to this order of beam plastic moment capacity using 8 M24 bars. The highest resistance recorded in the series was 4110kN, which would satisfy the required compression zone bearing capacity for all connections up to a beam depth of 900mm. The general pattern of behaviour was similar in all the tests. Inter-crossing shear yield lines inclined at 45 ~ to the horizontal first occurred in the centre of the web panel. Heavy yielding then occurred at the flange/web junction near to the load and support points. As this spread into the web, the threaded bars absorbed an increasing proportion of the load and eventually buckled after substantial yielding. In the uncased Type A specimens (see Figure 2), a sidesway buckling mode occurred. In the remaining uncased specimens, stiffeners buckled without sidesway and were partially restrained at their ends. Test CB 1 CB2 TABLE 2 SUMMARY OF ISOLATED COMPRESSION ZONE TEST RESULTS Nominal web bearing capacity F~w (~) 507 507 CB3 ,j 507 CB4 651 n CB5 1063 || CB6 1063 1063 Failure load Fc Fc + Few (1~) 1575 1425 1425 2250 2780 Failure mode 2400 3200 3.107 21811 2.811 3.456 2.615 2.258 3.010 CB7 |l CB8 1063 3308 3.112 ,, CB9 1063 2515 2.366 || CB 10 1063 4000 3.763 CC4 1063 1063 4110+ 3500 CC5 3.866 3.293 Web sidesway Web sidesway Web sidesway Stiffeners yielding Stiffeners yielding Stiffeners yielding Stiffeners yielding Web sidesway Stiffeners yielding Web sidesway Concrete cracking Concrete cracking + capacity of loading rig reached before failure of specimen The relationship between applied load and flange to flange displacement is shown in Figure 5. The compressibility of the connection was higher in the case of slender off-line stiffeners but was still well below an acceptable maximum value. Use of internal backing plates for the above case improved stiffness and strength significantly. In case of in-line stiffeners, concrete encasement prevented buckling and enabled them to develop their full yield capacity supplemented by the compression resistance of the concrete. The results in Table 2 show significant improvements in this respect. It was established for in-line stiffeners that the total bearing resistance could be taken as the nominal web bearing capacity plus the compressive capacity of the stiffeners based on standard column design procedures with effective length equal to actual length and cross-section based on tensile stress area. For off-line stiffeners, the total bearing resistance is dependent on stiffener position and relative material strength. Threaded Bar Compression Stiffening for Moment Connections 289 Figure 5" Deformation stiffeners in isolated compression zone tests (305x305xl 18kg/m UC) Figure 6: Moment rotation relationship for whole-connection tests . connections, bolted joints INTRODUCTION Beam-to-column end plate connections have been used extensively in multi-storey construction for resisting moments from wind and gravitational loading. UK and. of behaviour was similar in all the tests. Inter-crossing shear yield lines inclined at 45 ~ to the horizontal first occurred in the centre of the web panel. Heavy yielding then occurred at the. positioned with appropriate installation clearance from beam flange surfaces. Both forms may be combined to provide compression zone stiffening in non-reversing or partially reversing moment connections.