An investigation of the block shear strength of coped beams with a welded. clip angles connection Part I: Experimental study

Transcription

1 An investigation of the block shear strength of coped beams with a welded clip angles connection Part I: Experimental study Michael C. H. Yam a*, Y. C. Zhong b, Angus C. C. Lam b, V. P. Iu b a Department of Building and Real Estate, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, China b Department of Civil and Environmental Engineering, University of Macau, Macau, China Abstract The ends of a coped beam are commonly connected to the web of a girder by double clip angles. The clip angles may either be bolted or welded to the web of the beam. One of the potential modes for the failure of the clip angles connection is the block shear of the beam web material. To investigate the strength and the behavior of the block shear of coped beams with welded end connections, ten full-scale coped beam tests were conducted. The test parameters included the aspect ratio of the clip angles, the web shear and tension area around the clip angles, the web thickness, beam section depth, cope length, and connection position. The test results indicated that the specimens failed, developing either tension fractures of the web near the bottom of the clip angles or local web buckling near the end of the cope. Although the final failure mode of the six specimens was local web buckling, it was observed during the tests that these specimens exhibited a significant deformation of the block shear type prior to reaching their final failure mode. No shear fracture was observed in all of the tests. A comparison between the ultimate loads in the test and the predictions using the current design equations indicates that the current design standards such as the AISC-LRFD, CSA-S16-01, Eurocode 3, BS5950-1:2000, AIJ and GB50017, are inconsistent in

2 predicting the block shear strength of coped beams with welded end connections. The analytical study of the strength of the test specimens using the finite element method, a parametric study, and a proposed design model for designing block shears for coped beams with welded clip angles are included in a companion paper. Keywords: Block shear tests, Coped beams, Welded clip angles, Tension fracture, Local web buckling *Corresponding author: Tel. (852) ; fax: (852) address: bsmyam@polyu.edu.hk (Michael C. H. Yam) 1. Introduction In steel construction, beams are often coped (cut away) at the flanges to provide clearance for the framing beams or to maintain the same elevation for intersecting beams. The cope can be at the top (Fig. 1), the bottom, or both flanges near the connection in order to facilitate construction. The ends of the coped beam are commonly connected to the web of the girder by double clip angles. The clip angles may be either bolted or welded to the web of the coped beam. One of the potential modes of the failure of the coped beam with a clip angles connection is the block shear of the beam web material. Block shear is a phenomenon of rupturing or tearing, where a block of material is torn out by a combination of tensile and shear failure (Birkemoe and Gilmor [1]). Figure 2 shows the potential block shear failure mode in a coped beam at the shear connection. The web block can be partially or entirely torn out from the remaining part of the beam. Block shear is usually associated with bolted details, because a reduced area is present in such cases. However, block shear may also be a potential failure mode in welded connections depending on the details of the connection. The block 2

3 shear failure mode appears in different types of bolted structural members such as gusset plates, coped beams, angles, or tee-sections (Hardash and Bjorhovde [2]; Epstein [3]; Epstein and Stamberg [4], Epstein and McGinnis [5], Franchuk et al. [6], Orbison et al. [7], etc.). The block shear failure mode of coped beams was first identified by Birkemoe and Gilmor [1]. In their tests, the coped beam failed by the tearing of the web as a block of material at the shear connection. The authors suggested that the failure model of block shear was provided by a combination of tensile and shear stresses acting over their respective areas (across plane AA and plane BB, respectively) as shown in Fig. 3. Yura et al. [8] conducted a series of twelve coped and uncoped beam tests with bolted double clip angle connections. Among eight coped beam tests, three exhibited failure of the block shear type. The specimens with two lines of bolts had a lower capacity than desired. A further investigation by Ricles and Yura [9] was carried out to examine the block shear failure mode of coped beams for bolted connections. The test parameters included the end and edge distance, bolt arrangement, and the type of holes. All seven coped beam specimens failed in the block shear mode and the web buckled at the cope in four of them. Based on the test and the finite element analysis results, the authors suggested assuming shear yielding for the vertical side of the web and tensile fracture for the horizontal side (perpendicular to the applied reaction force at the coped end) in evaluating the block shear strength of the connection. It was further noted that the shear yielding that occurred along the gross vertical area of the web was based on the experimental observation. This implied that the connection capacity was the sum of a triangular normal stress on the net area of the tension face and shear yielding on the gross shear area as shown in Fig. 4. The proposed capacity equation is as follows: 3

4 P= 0.5F A + 0.6F A (1) u nt y gv where: P is the ultimate connection capacity; F u is the tensile strength; F y is the yield strength; A nt is the net tension area, and A gv is the gross shear area. Aalberg and Larsen [10] examined the behavior of coped I-beams with bolted end connections fabricated from high-strength steel and normal structural steel. Identical failure modes were observed for both normal and high-strength steels except that the connection ductility was reduced for the high-strength steel specimens. A comprehensive review of block shear issues was conducted by Kulak and Grondin [11], [12]. They revisited the block shear failure mode in different cases of coped beams, gusset plates, and angles. The review showed that the failure modes were significantly different in two important categories, namely: gusset plates and the web of coped beams. Taylor [13] discussed the issue of the lack of experimental data for the block shear strength of coped beams with welded connections. Most recently, Franchuk et al. [6], [14] conducted an extensive experimental program consisting of seventeen tests to investigate the block shear behavior of coped beams with bolted connections. The test results indicated that magnitudes of tension and shear areas significantly affect connection capacity. The test results also substantiated the previous experimental observation that shear yielding on the gross (vertical) area should be used in the block 4

5 shear design of coped beams. The latest research was performed by Topkaya [15]. A finite element parametric study on the block shear failure of steel tension members was carried out to develop a block shear capacity prediction equation. The parametric study was conducted to identify the important parameters, such as ultimate-to-yield ratio, connection length, and boundary conditions. Based on the above discussion, it can be seen that all of the available experimental and analytical studies on the block shear of coped beams concentrated on bolted end connections. Therefore, the main objective of this study is to provide experimental data for the block shear strength and behavior of coped beams with welded clip angle connections. The evaluation of the ultimate strength of the test specimens using current design codes (AISC-LRFD [16], CSA-S16-01 [17], Eurcode 3 ENV [18], BS5950-1:2000[19], AIJ [20] and GB50017 [21]) will also be presented. 2. Experimental program 2.1 Test Specimens The purpose of the experimental program was to examine the block shear failure strength and behavior of a coped beam web with a welded clip angles connection. A total of ten full-scale tests were conducted in the experimental program. The test parameters included connection geometry and cope details such as the aspect ratio of the clip angles, the tension and shear area of the web block, web thickness, beam section depth, cope length, and connection position. The test was conducted individually at each end of the 3.3 m long test beam. The schematics and details of the specimens are shown in Figs. 5, 6, and 7. These five test beams were fabricated from three different section sizes, including universal beam UB406x140x46, UB457x191x74, and UB356x171x67 (SCI [22]). Grade 43 steel conforming to BS 4360 [23] was used 5

6 for the beams. All of the double clip angles were fabricated by 16 mm steel plates conforming to BS 4360 [23] Grade 50 to provide the required connection dimensions. The angles were designed to provide enough strength for the connections and, at the same time, to minimize the in-plane rotational stiffness in order to simulate a simply supported boundary condition. The double angles were welded to the web of the beams. The nominal and measured dimensions of the beam sections and the connection details are shown in Table 1 and Table 2, respectively. These tables should be read in conjunction with Fig. 5. The test beams were each designated by a letter (A through E), and each end connection was designated by a number, 1 or 2, respectively. Each specimen was also designated according to the beam type, and the testing variable was assigned to facilitate the identification. For example, A1-406r3 represents beam A, connection 1, and UB406x140x46 section with an aspect ratio of around 3. Figure 6 illustrates the typical overall design dimensions of the coped beam specimens and the clip angles, and Fig. 7 presents the connection details of all of the specimens. As summarized in Table 3, the test parameters included the aspect ratio of the clip angles, the tension and shear area of the web block, web thickness, beam section depth, cope length, and connection position. Only one parameter was varied in each group of tests. The specimens with various configurations were carefully designed to fail in the block shear mode by the current design practice. Note that for the aspect ratio series, the design capacities of the related tests were nearly identical for the purpose of comparison. The top flange was coped for all specimens. In general, the cope dimensions were fixed, with the cope depth extending 30 mm below the top flange and the cope length extending 50 mm away from the end of the clip angles except for the specimens that were employed to study the effect of cope length. 6

7 In order to obtain the material properties, tension coupon tests were carried out. Tension coupons were prepared from the web and the flange of the test beams and then tested according to the ASTM A370 standard [24]. An extensometer with a 50 mm gauge length was used to measure the strain in the coupon and the load was obtained as a read-out of the testing machine. Pairs of strain gauges were mounted on both faces of the coupons in order to determine the modulus of elasticity (E) of the steel in the elastic range. The static readings were taken on the yield plateau, along the strain hardening curve, and near the ultimate tensile strength. 2.2 Test setup, instrumentation, and test procedures The test setup is shown schematically in Fig. 8. The beam specimen with welded clip angle connections was installed to the column by 24 mm diameter bolts. Washers were also placed between the outstanding leg of the angles and the supporting column in all tests to minimize the end rotational stiffness. The bolts were then tightened to a snug-tight condition. Load was applied to the beam vertically by a 890 kn hydraulic cylinder. The load position was chosen to produce block shear failure in the connection and to avoid either shear or bending failure of the test beam. The chosen load position also has little effect on the stress distribution in the coped region. The distance, L, from the load position to the reaction end was 510 mm (1.3D), 600 mm (1.3D), and 550mm (1.5D) for the test beams Beam406, Beam457, and Beam356, respectively, as illustrated in Fig. 8. Roller assemblies were used at the load position and the support so that both rotation and longitudinal movements were allowed. Lateral bracings were provided at the load position and the support to prevent lateral movements of the test beam. Additional lateral bracings were provided at the top flange of the beam near the cope end to improve the lateral stability of the connection. Stiffeners were welded on both 7

8 sides of the web at the load position in order to avoid local web crippling. Load cells were used to record the applied load and the support reaction. Linear variable differential transformers (LVDT) were employed to measure the displacements. The locations of the LVDTs are typically shown in Fig. 9. To measure the main features, LVDTs were placed separately at the bottom of the beam underneath the load point, and at the top and the bottom of the beam at the connection. The displacement of the clip angle at the column was also recorded. The overall deflection was determined by taking the difference between the displacement under the applied load and the displacement of the clip angles, which excluded the displacements due to bolt bearing and slippage. In addition, the deformation between the vertical displacements measured from below and above the connection provided a good indication of the significant tensile yielding or sudden fracture of the beam web. Another LVDT was set at the top flange near the loading position to detect lateral movements. The rotation of the angles and the tested beam end were also recorded by LVDTs. In addition, an inclinometer was mounted on the web at the centroid of coped section to monitor end rotation. Longitudinal strain gauges and rosettes were extensively mounted around the connection to measure the strain distribution at the critical area, as shown in Fig. 10. More details regarding the instrumentation can be found in Zhong et al. (2004). All readings were collected by a data acquisition system. Before the test, the specimen was whitewashed to make it possible to detect the yielding process, and 50 mm square grid lines were drawn to show the web deformations. In general, the test procedures were similar for all of the specimens. The loading process was divided into two stages: load control in the beginning and stroke control when nonlinear behavior commenced. Load was applied incrementally and a smaller 8

9 load increment was used to capture the nonlinear behavior in the inelastic stage of the tests. During the test, it was necessary to hold the loading at regular intervals for static equilibrium so that the specimens could redistribute the stress and deform completely, therefore making it possible to obtain the reading on static load. Readings of load cells, strain gauges, and LVDTs were taken continually during the loading process. The yielding pattern and process were recorded in detail. The test was terminated when the maximum load was reached and the load decreased significantly. After the test on one end was completed, the other end of the beam was installed to the column, and a new test was then conducted. 3. Test results 3.1 Material test results The tension coupon test results for all specimens are summarized in Table 4, and the results for clip angles are included as well. Since the block shear failure was mostly associated with the beam web, the mean values of the coupons from the web of test beams were used as the basic material properties for the specimens. The average static yield strength for the specimens ranged from MPa to MPa, whereas the average static ultimate strength ranged from MPa to MPa. Although Grade 50 steel was assumed in the design of the clip angles, it was believed that Grade 43 steel was used instead for the clip angles based on the tension coupon test results as illustrated in Table 4. The material properties for S275 and S335 (EN :2004) are also included in the table for comparison purpose. 3.2 General The test results are summarized in Table 5, where the static values of the ultimate load are reported, and the connection reactions and moments were calculated from 9

10 static equilibrium. For simplicity, only the letter and the first number in the specimen designation were used to identify the specimens (for example, specimen A1 instead of specimen A1-406r3). The moment developed in the connection was found to be small compared with the yield moment of the beam. In general, there were two kinds of failure modes in the tests: the block shear of the beam web with tension fractures underneath the clip angles and local web buckling near the end of the cope. Two specimens failed in the block shear mode while six specimens ultimately failed in local web buckling near the cope due to combined shear and bending. Although the final failure mode of the six specimens was web local buckling, it was observed during the tests that these specimens exhibited a significant block shear type deformation prior to reaching their final failure mode. In the block shear cases, necking of the tension area in the web underneath the clip angles was observed before the web fractured abruptly. The reduction in web thickness in the tension region is also included in Table 5. A tensile crack developed in the web along the bottom of the welded clip angles rather than a complete web block tear-out. No signs of shear fracture along the vertical plane were observed. As to the remaining two specimens, the test loads exceeded the capacity of the testing frame; hence, these two tests were terminated at the safe load of the test setup. The photographs of typical failed specimens at the ultimate stage are shown in Fig. 11. Figure 12 contains an illustration of typical local web buckling failure near the cope end. In general, a similar yielding pattern was observed for the specimens. Yield lines were usually initiated in the web either underneath the welded clip angles at the extreme beam end or near the cope end. Eventually, the yield lines extended through the web block and could be obviously observed below the welded clip angle and across the cope end. These yield lines indicated that significant deformation had occurred in 10

11 these regions. Yield lines could also be found in the shear area near the welded clip angles in most cases. The results showed that the web plate buckled to various extents at the cope end even though the flange near the cope was braced against lateral movement. Severe compression and shear yielding was observed near the cope end before the specimens reached the ultimate loads. This indicated that local web buckling was also a potential failure mode for coped beams with welded clip angle connections. This failure mode was also observed by Ricles and Yura [9] in their coped beam tests. They pointed out that the high horizontal compressive stresses in the web could cause the web to buckle at the cope. 3.3 Load deflection behavior A typical applied load versus deflection curves are shown in Fig. 13. Plotted on the horizontal axis is the net deflection at the load point, which was determined by taking the difference between the displacement under the applied load and the displacement of the clip angles; hence, this net deflection excluded the displacements due to bolt bearing and slippage. Thus, the overall behavior of the test beams can be presented. Generally, similar behavior was observed of the specimens with regard to the two kinds of failure modes. As the load was applied, yielding first appeared in the web underneath the welded clip angles or near the cope end. Nonlinear behavior then commenced. When the load continued to increase, shear yielding developed gradually along the vertical area near the welded angles. High flexural stresses in the upper region of the web increased and excessive deformation developed. The web block deformed significantly before the web plate near the cope end became distorted or buckled inelastically. Subsequently, for specimens C2 and E1, tension fractures 11

12 developed abruptly underneath the clip angles, and the load therefore decreased significantly. Continued loading caused further opening of the cracks. As to specimens A1, A2, B1, B2, D1, and E2, at the ultimate load level severe compression as well as shearing caused local web buckling at the cope end, and the load subsequently descended. 3.4 Strain distribution Strain distributions are presented in the critical area along the web block as indicated by the locations of strain gauges and rosettes shown in the insets of Figs. 14 and 15. The strain distributions are plotted within the elastic stage, against the same load levels of 70 kn and 183 kn. Only the strain distributions for specimen E1 are presented since there is no significant difference in the strain distributions of other specimens. The vertical longitudinal strain distribution along the horizontal plane underneath the welded clip angles (the tension area) for specimen E1 is plotted in Fig. 14. As expected, the strain was largest at the beam end and decreased along the length of the tension area. An almost linear strain distribution was found within the elastic range. This kind of distribution reflected the load eccentricity on the connection that caused the rotation of the web block. The line of action of the vertical reaction force at the beam end was outside the centroid of the web block and consequently induced this eccentricity. However, at same load level, the positions of the neutral axis for the longitudinal strains were close to each other for specimens with the same beam sizes. For the cases with a large area of tension, the longitudinal strains beneath the corner of the clip angle indicated slight compression at the beginning of loading. It was believed that the compressive strain would shift to tension as the load increased. 12

13 Vertical shear strain distributions along the vertical plane near the welded clip angles (the shear area) are also examined. Typically, the result for specimen E1 is shown in Fig. 15, in which the dash line indicates the shape of the clip angle. It can be seen from the figure that the shear stress distribution determined by three rosettes was uneven. It differed from the shear distribution calculated by the shear formula for the reduced beam section. The stress concentration around the top and the bottom corners of the clip angle was believed to have some influence on the reading on strains measured at these positions. In general, the shear strain distributions of the other specimens were similar. Since the number of rosettes in the shear area was limited, the shear strain profile obtained might not represent the entire shear strain distribution in this area for the test specimen. Nevertheless, the shear strain values measured at the critical locations provide the data for the calibration of the finite element model. It will be seen that the measured shear strains are consistent with the numerical ones. The calibration and discussion of the stress/strain distribution of the finite element model of the test specimens can be found Yam et al. [25] and Zhong [26]. The local bending effect around the web block was caused by the connection rotation. The distribution of flexural strains in the shear area can be illustrated by the longitudinal strains in the horizontal direction (marked with No. 7, 10, and 13 in the inset of Fig. 16). The distribution features are similar among all specimens. Take for example specimen E1 as shown in Fig. 16. High flexural compressive strains developed in the beam web near the cope, while tensile strains developed near the bottom of the clip angle. It was believed that the high compression as well as the shearing that developed in the compression region would cause early yielding of the web plate; and that, consequently, at a high load significant yielding would induce inelastic web buckling at the cope end. 13

14 4. Discussion of the test results 4.1 Failure mode Previous studies focused on the block shear failure of coped beam web with bolted connections. For bolted end connections, a reduction in the cross-sectional area due to the presence of holes has adverse effects on the strength of the connections. The concentration of stress at the bolt holes initiated fracturing, and contributed to the block shear failure mode. For welded end connections, since there is no reduction in gross section area due to bolt holes, the behavior and failure mode of the connection would be different. As described before, two kinds of failure modes were observed in the tests, namely; block shear failure and local web buckling at the cope end. Block shear failure was exhibited in a partial tear-out of the web block with a tension fracture underneath the clip angles. The tension fracture was triggered at the extreme end of the beam web, and necking of the web plate was observed before a sudden rupture. Shear yielding rather than shear fracture was observed along the vertical plane of the shear area. Shear ultimate strength was difficult to achieve since the shear deformation was relatively small compared with the tension deformation. Hence, tension fracture was reached prior to shear fracture due to the significant deformation in the area of tension. Another potential failure mode was local web buckling near the end of the cope. High compressive stresses and shear stresses were localized in the web near the cope because of the combination of bending and shear in the reduced section; hence resulting in extensive yielding and consequently inducing local web buckling in that region. As mentioned before, lateral bracing at the top flange near the cope was provided to improve the lateral stability of the test beams. However, various degrees of web 14

15 buckling were still observed in most cases, and six specimens among them ultimately failed in the local web buckling mode. In fact, in previous studies, such as those by Birkemoe and Gilmor [1] and Ricles and Yura [9], local web buckling was also observed in cope beam tests for block shear with bolted connections. For those specimens that failed due to local web buckling, excessive deformation developed around the web block at a high load due to significant yielding in the tension and shear areas. However, the deformation was insufficient to induce fracturing even though the ultimate tensile stress was attained in the area of tension. On the other hand, severe compression and shear yielding accumulated near the cope end, thus eventually resulting in local web buckling. Nevertheless, observations showed that these specimens exhibited a deformation of the block shear type prior to reaching their ultimate loads with local web buckling failure. 4.2 Effect of aspect ratio Aspect ratio has been defined as the ratio of vertical shear area (b ) to the horizontal tension area (a) of the web block, as indicated in Fig. 5. Note that the design capacities of the related specimens are nearly identical for the purpose of comparison. The aspect ratio was examined in two series of tests. In the first series, specimens A1, A2, and B2 had an aspect ratio of 3.6, 2.3, and 1.4, respectively. Their final failure mode was local web buckling, as shown in Table 5. For specimens A2 and B2, the aspect ratio varied from 2.3 to 1.4 while the capacity of A2 was 12% higher than that of B2. Although specimen A1 had the largest aspect ratio, it was worth noting that the double clip angles used in this specimen was different from those of the other specimens due to a fabrication error. This had an adverse effect on the capacity of the specimen. Since the bolted leg of the clip angle was fabricated shorter for the arrangement of the bolts, the boundary condition of the pin support that was simulated by the clip angle was 15

16 influenced to some extent. This meant that the position of the pin support for the beam specimen was therefore lowered, while the rotational stiffness of the joint was reduced. Since the negative moment that developed in the connection at the beam end was due to a slight fixity in connection, the negative moment of A1 in the connection was smaller than that of A2 and B2, as shown in Table 5. Hence, it was believed that the reduction in the capacity of specimen A1, which failed because of the local web buckling, was attributable to the influence of the clip angle. The results of the finite element analyses that will be discussed in the companion paper (Yam et al. [25]) also indicates that specimen A1 would have a larger capacity than specimen A2 by using double clip angles similar to those of the others. For specimens C1 and D1 in the second series, the aspect ratio was 3.8 and 1.6, respectively. Although the ultimate failure of C1 did not occur before the test was terminated at the safe load of the test setup, it was believed that C1 would have a much larger capacity than D1 from the trend of the load deflection curve, which can be observed in Fig. 17. In addition, further observations of the yield lines also led to an interpretation of the effect of the aspect ratio on the deformation of the web. For specimen D1, many yield lines were observed in the beam web near the bottom of the clip angle as shown in Fig. 11d. This might indicate that the web material in this region had been subjected to excessive deformation caused by local bending due to the connection rotation. On the other hand, for specimen C1, there were significantly fewer yield lines, as shown in Fig. 11b. Nevertheless, it can be seen that as the aspect ratio decreased, local bending of the web near the clip angle due to the connection rotation was more significant. In fact, the local bending would cause high compressive stress to the web material near the end of the cope corner and hence might lead to a failure in the local buckling of the web. In addition, the local bending also induced high tensile stress in the web near the bottom 16

17 edge of the clip angle, which might initiate fractures of the web block similar to those seen in specimens C2 and E1. Based on the above discussion, it is believed that the block shear capacity of the connection decreased as the aspect ratio decreased. As the aspect ratio decreased (increase in a), the loading eccentricity between the welded clip angle and the support (bolted end) at the column face increased, hence generating more local bending of the web in the vicinity of the clip angle. Since the aspect ratio is associated with the ratio of the shear area and the tension area, it is necessary to examine these two aspects individually to reveal their relationship and the effect on the strength of the connection. 4.3 Effect of shear area and tension area Specimens C1 and C2 were identical in all aspects except for the shear area. The length of the shear area (b) of specimen C1 and C2 was 170 mm and 120 mm, respectively. As shown in Fig. 18, the connection stiffness of C1 was higher than that of C2. Specimen C2 failed due to block shear with tension fractures. However, the strength of C1 exceeded the test setup capacity, so it was certainly larger than that of C2. According to the test data, C1 was estimated to be at least 10% larger in capacity than C2. As expected, the connection capacity was improved by increasing shear area. For specimens C2 and D1, the length of the tension area (a) varied from 50 mm to 90 mm, while the shear area of the specimens was held at a constant value of 120 mm in length. As shown in Table 5, the capacities of these two specimens were very close, even though there was a significant difference in the tension area. Observed during these two tests, the webs buckled similarly before their final failure. However, specimen C2 eventually failed due to block shear with tension fractures, while specimen D1 failed due to local web buckling. The yielding pattern of specimen D1 17

18 showed that there was excessive deformation in the web around the clip angle caused by significant connection rotation. The rotation was believed to have been produced by the eccentricity of the reaction force on the web block, while the eccentricity was associated with the area of tension. Hence, this indicated that increasing the area of tension led to an increase in loading eccentricity and therefore generated more bending stress to the edge of the beam web region near the clip angles and the cope end. This increase in bending stress counteracted the increase in the connection strength due to the additional area of tension and initiated either the earlier local web buckling in the end of the cope or web tension fractures near the bottom of the clip angles. Therefore, the capacity of specimen D1 did not increase considerably even though the area of tension increased by 80%. 4.4 Effect of web thickness Specimens B1, C2, and E1 were fabricated from three different beam sections with identical connection dimensions. The nominal beam web thickness (t w ) was 6.8 mm for specimen B1, and 9.1 mm for specimens C2 and E1. As shown in Table 5, when the web thickness of the specimens decreased the ultimate loads decreased; however, the decrease was not directly proportional to the change in the thickness of the web. This is illustrated in Fig. 19, where the original reaction versus web deformation curve of specimen B1 was modified proportionally by multiplying a scale factor (9.1/6.8) to account for the difference in thickness from the original one. It can be seen from this figure that the modified ultimate load of specimen B1 (527 kn) was 14% less than that of specimen E1 (601 kn), even though the two specimens had a similar ultimate tensile strength (F u ). The figure also shows that the behavior of specimens C2 and E1, which had the same web thickness, was nearly identical. For thin web specimen B1, its weak 18

19 axis flexural stiffness was lower than that of the other two specimens and therefore increased the susceptibility of the beam web to local buckling. Subsequently, like specimens C2 and E1, B1 ultimately buckled rather than failed by block shear with tension fractures. 4.5 Effect of Beam Section Depth The effect of beam section depth can also be observed in Fig. 19. For specimens C2 and E1, the web thicknesses and the dimensions of the connections were identical while the depth of the beam section varied from 457 mm to 363 mm. The results showed that these two specimens had almost the same amount of connection ductility. Specimen C2, with a section depth that was 26% deeper, only produced a 5% increase in capacity regardless of the slight difference in the material tensile strength. In addition, the failure modes (tension fractures) of these two specimens were similar, implying that block shear capacity was not affected by section depth. 4.6 Effect of Cope Length The cope length (c) varied from 150 mm to 80 mm for specimens D1 and D2, respectively, and the other details for these two specimens were identical. Specimen D2 had the smallest cope length of all of the specimens, in which the uncoped top flange extended close to the end of the beam and beyond the region of the shear path near the welded clip angle. As shown in Fig. 20, it can be seen from the comparison of specimens D1 and D2 that cope length had a significant effect on connection capacity. Connection D2 was much stiffer than D1. The web plate did not buckle and only a small deformation was observed in specimen D2 at the maximum load of the test frame. Therefore, the capacity of D2 was certainly larger than that of D1. Hence, for the specimen with a shorter cope length the connection was strengthened by the uncoped 19

20 top flange; the susceptibility to local web buckling near the cope end was therefore reduced. In other words, increasing the cope length increased the possibility of instability and therefore led to a decrease in the connection capacity. 4.7 Effect of Connection Position For specimens E1 and E2, the connected position (p) of the welded clip angles varied from 20 mm to 45 mm (the vertical distance from the clip angle to the top edge of the beam web as shown in Fig. 5) while other details were identical. Similar load deflection behavior and a slight difference in capacity were observed between these two cases. The capacity of E2 was 3% less than that of E1, even if the vertical shear path (b ) of E2 was longer. In addition, E2 ultimately failed in local web buckling mode rather than in the block shear with tension fractures as E1 did. This may indicate that the final failure mode of the specimen was affected by the connected position of the welded clip angles. Since high compressive and shear stresses developed in the beam web between the top of the clip angle and the cope end, it was believed that a local web buckling failure would likely occur in the case where the clip angles was placed at a lower position. 5. Comparison of Test Results with Current Design Methods 5.1 General The block shear design equations prescribed by the current standards of CSA-S16-01 [17], AISC LRFD [16], Eurocode 3 ENV [18], BS :2000 [19], AIJ [20] and GB50017 [21] were used to evaluate the block shear strength of the test specimens. However, it should be noted that the block shear design equations for coped beams were mainly associated with bolted end connections. The failure mechanism in welded end connections may be different from that of bolted end connections since, in 20

21 the former, there is no reduction in gross section area due to bolt holes. Nevertheless, these equations are also used for welded end connections. Consequently, for the test specimens in this study, the block shear failure was checked around the periphery of the welded connections. As mentioned above, although the final failure mode of six specimens was local web buckling, it was observed during the tests that these specimens exhibited a significant deformation of the block shear type prior to reaching their final failure mode. Therefore, it is believed that the use of the block shear design equations for the specimens that failed in local web buckling would provide an indication of the strength of the connection. 5.2 Current Design Standards CSA-S16-01 (2001) The CSA-S16-01 [17] standard provides two equations (Eq. (2) and Eq. (3)) to evaluate the block shear strength of connections. P = φ (0.5F A F A ) (2) r u nt y gv P = φ (0.5F A F A ) (3) r u nt u nv where: P r is the factored ultimate connection capacity; φ is the resistance factor; F u is the tensile strength; F y is the yield strength; A nt is the net tension area; A gv is the gross shear area; and 21

22 A nv is the net shear area. As suggested by Kulak and Grondin [12], the contribution of the area of tension is reduced by one-half, which implies that a triangular stress block with a maximum stress of F u is assumed. Equation (2) assumes that the block shear capacity is determined by a non-uniform stress distribution with an average stress of 0.5F u at a net area of tension, along with the shear yielding stress at the gross shear area; while Eq. (3) assumes that the shear contribution is the shear rupture at the net shear area. The lesser of above two equations should be used as the block shear capacity. A comparison of the previous test data on coped beams with bolted end connections shows that this procedure provides conservative results with an average test-to-predicted ratio of In the current study, since no shear fracture was observed for the failed specimens in this study, only Eq. (2) (the lesser one) was used to evaluate the block shear strength of the test specimens. Without the reduction in bolt holes when calculating area, the gross area and the net area in Eq. (2) were identical AISC LRFD (1999) For the AISC LRFD [16] standard, again, two equations (Eq. (4) and Eq. (5)) are provided: If FA 0.6FA : u nt u nv P = φ( F A F A ) φ ( F A F A ) (4) r u nt y gv u nt u nv If FA < 0.6FA : u nt u nv P = φ( F A F A ) φ ( F A F A ) (5) r y gt u nv u nt u nv 22

23 where: A gt is the gross tension area and the other symbols have been defined above. When the ultimate rupture strength at the net section is used on one segment, yielding at the gross section shall be used on the perpendicular segment. The design equations incorporate two possible modes of failures: the rupturing of the net tension area combined with yielding of the gross shear area, or yielding of the gross tension area combined with rupture of the net shear area. The upper limit is the sum of tension rupture on the net tension area and shear rupture on the net shear area. It should be noted that the first failure model is reasonable and supported by test observations. Experimental results in which block shear is evident tend to exhibit a failure mode similar to that described by Eq. (4); however, the qualifying condition listed in the specification makes it rarely control the design. Thus, Eq. (5) will commonly govern in practice. Kulak and Grondin [12] points out that the possibility of attaining the shear ultimate strength in combination with the tension yield strength seems unlikely due to the small ductility in tension as compared with shear, and there is insufficient tensile ductility to permit the occurrence of shear fractures. Nevertheless, the use of Eq. (4) and Eq. (5) does not provide good predictions of the available test results and produces a high degree of variation. In particular, for the two-line bolted connections, the equations are for the most part not conservative. In general, they do not accurately represent the failure modes observed in tests. The AISC LRFD specification [16] also indicates that the block shear failure mode should also be checked around the periphery of welded connections. For the current study, the larger one of Eq. (4) and Eq. (5) was used to evaluate the block shear strength of the test specimens. 23

24 5.2.3 Eurocode 3 ENV For Eurocode 3 [18], a single equation can be derived from a series of equations according to the specifications for block shear on coped beams as follows: P = F ( L kd ) t + F A r γ 0 3 u gt ot w 3 y gv M, (6) where: γ M 0 is the partial safety factor for resistance (a vale of 1.0 is used in this study); L gt is the length of the gross tension area; d ot is the hole size for tension area; and k is the coefficient for tension area as follows: for a single line of bolts: k = 0.5 ; for two lines of bolts: k = 2.5. The equation combines the reduced normal stress acting over the tension area with shear yielding acting over the gross shear area. Franchuk et al. [6] indicated that the reduction coefficient of 1 3 for the normal stress is considered similar to the 0.5 used in CSA-S16-01, although this value appears to be derived from the von Mises yield criterion. It is found that the equation provides a very conservative prediction even for two-line bolted connections BS :2000 In this updated BS :2000 standard [19], block shear for bolted connections is covered as follows: 24

25 P= 0.6 p K ( L kd ) t p Lt, (7) y e t t w y v w where: P is the ultimate connection capacity p y is the design strength K e is the effective net area coefficient t is the web thickness L t is the length of the tension area L v is the length of the shear area k is defined above, and D t is the hole size for the tension area For general purposes regardless of steel grade, tensile strength. Therefore, Eq. (7) becomes: K e U s =, where U s is the minimum 1.2 p y P= 0.5 U ( L kd ) t p A (8) s t t w y gv Disregarding the partial factor and the different definitions for design strength, this equation is very similar to that of Eurocode 3, except that 0.5 instead of 1 3 is used again for the reduction of the tension contribution. Compared with Eq. (2), there is a slight difference in the calculation of the net tension area. Equation (8) is believed to produce conservative results for two-line bolted connections in coped beams. For welded end connection, if the design strength and the minimum tensile strength in Eq. (8) is substituted by F y and F u respectively, an equation the same as Eq. (2) can be deduced. This means that for a welded end connection, the prediction by BS

26 1:2000 [19] is exactly identical to that provided by the CSA-S16-01 standard [17]. Thus, the evaluation results for BS :2000 [19] will not be repeated here Standard for limit state design of structures (AIJ 1990), Architectural Institute of Japan, The AIJ (1990) [20] provides the following conservative procedure: 1 Pr = φ ( FuAnt + FyAnv) (9) 3 1 Pr = φ ( FyAnt + FuAnv) (10) 3 where the symbols are similar to the ones defined before. The lesser of the equations is used for the connection capacity. It combines tensile and shear stresses acting on both net areas. The results show that these equations give more conservative predictions, but with a high degree of variation. In addition, these equations cannot represent the actual failure mechanism for block shear in coped beams. In contrast with the equations of AISC LRFD [19] (Eq. (4) and Eq. (5)), there is only a negligible difference in the coefficient. In fact, the AISC LRFD standard [19] uses the larger of these two equations instead of the smaller Chinese Standard GB50017 (2003) Based on the results of the gusset plates test, a block shear equation is derived and added to the latest version of the Chinese standard for the design of steel structures (GB [21]). From the commentary in the code, the following explicit equation can be derived for coped beams: 26

27 1 P= FuAnt + FuAnv, (11) 3 where the symbols are similar to the ones defined before. This equation assumes that tensile rupture on the net tension area and shear rupture on the net shear area appear simultaneously. The average test-to-predicted ratio for Eq. (11) on the results of the gusset plates test provided in the code GB50017 [21] is The code also shows that the equation is applicable for both bolted and welded connections. It is interesting to note, neglecting the partial factor, that this model is almost the same as that of the former Canadian standard CAN/CSA-S [17], which is expressed as: P = 0.85 φ ( F A F A ) (12) r u nt u nv 5.3 Design Strength of Test Specimens Based on Current Design Standards The ultimate strengths of the test specimens were compared to those predicted by all of the above design equations from different design standards for block shear. The results are summarized in Table 6. The predicted capacities were based on the measured dimensions of the specimens, including the measured weld sizes and the measured material properties. All resistance factors are taken as 1.0. The test-topredicted ratio (or termed professional ratio) of those failed specimens are listed in Table 6. As described before, for the six specimens that ultimately failed due to local buckling of the web rather than block shear, the observation from test results showed that the deformed block did appear prior to ultimate buckling of the web. Therefore, qualitative comparisons of these specimens may still be made using the current design 27

28 standards. For the specimens that failed by block shear with tension fracture, the average test-to-predicted ratios were 1.49 and 1.06 based on CSA-S16-01 [17] and AISC LRFD [16], respectively. This may indicate that the CSA-S16-01 [17] design model may be too conservative for block shear in coped beams with a welded clip angles connection. At the same time, it is important to note that the average test-topredicted ratios for those specimens that failed due to local buckling of the web were 1.26 and 0.88, as estimated by the CSA-S16-01 [17] and the AISC LRFD [16] block shear equations, respectively. Although the test results were not conclusive, it is believed that current North American design standards are inconsistent to allow predictions of the block shear of coped beams with welded end connections. The approach of Eurocode 3 [18] produced very similar results in capacity predictions as compared to CSA-S16-01 [17]. It was found that the capacity predicted by Eurocode 3 [18] was about 3% larger than that of CSA-S16-01 [17], which could be attributed to the slightly different coefficients used in the related equations. The average test-to-predicted ratios were 1.45 and 1.22 for the block shear cases and the local web buckling cases, respectively. Besides, as mentioned before, predictions by BS :2000 [19] were exactly identical to those by the CSA-S16-01 standard [17]. Consequently, it can be concluded that, in general, the current European and British design standards are also conservative for block shear in coped beams with a welded clip angles connection. AIJ standard [20] did not provide a consistent prediction for both failure modes, and generally overestimated the capacity of the specimens that failed due to local buckling of the web. The average test-to-predicted ratios were 1.15 and 0.96 for the block shear cases and the local web buckling cases, respectively. Nevertheless, the AIJ standard [20] provides a safer estimation than AISC LRFD [16] because the former 28

29 adopts the smaller value of the two equations, while the latter uses the larger one. However, the procedures of both the AIJ standard [20] and AISC LRFD [16] could not accurately represent the actual failure modes observed in the tests, since no fractures were found in the shear area. It should be pointed out that GB50017 [21] overestimated the capacities of all specimens. The average test-to-predicted ratios were 0.96 and 0.80 for the block shear cases and the local web buckling cases, respectively. Hence, this standard produced non-conservative estimates for the block shear strength of coped beams with a welded clip angles connection. In addition, it was unlikely that the ultimate tensile strength of the tension area and the ultimate shear strength of the shear area would be achieved simultaneously; hence, the model did not reflect the actual failure mechanism. Generally, the predictions using these design equations exhibited a vast degree of variation with regard to the test data. For example, these methods often produced large overestimations for specimens with large areas of tension, such as specimens B2 and D1. Furthermore, in all cases, the predictions were less conservative for the specimens that failed due to local buckling of the web, even though a significant deformation of the block shear type occurred before the specimens reached their final failure mode. Therefore, it is concluded that the existing design standards are inconsistent in predicting the bock shear of cope beams with welded end connections. 6. Summary and Conclusions To investigate the block shear strength and behavior of coped beams with welded end connections, ten full-scale coped beam tests were conducted. The failure mechanism in coped beams with welded end connections was different from that with bolted end connections, since there was no reduction in the web section area due to bolt 29

30 holes. The test results showed that only two of the ten specimens failed by block shear mode with tension fractures. Tension fractures occurred abruptly in the beam web underneath the clip angles. No shear fractures were observed and no tear-out type of block shear occurred. Local web buckling was also a potential failure mode for coped beams with welded end connections. High compressive stresses and shear stresses were localized in the web near the end of the cope due to the combination of bending and shear in the reduced section, hence resulting in extensive yielding at the cope and inducing local web buckling. However, these specimens exhibited significant deformation of the block shear type prior to reaching their final failure mode. The test parameters examined included the aspect ratio of the clip angles, the tension and shear area of the web block, web thickness, beam section depth, cope length, and connection position. The results showed that the connection capacity increased as aspect ratio and shear area increased. A large area of tension would increase the loading eccentricity and hence generate more bending moments to the edge of the beam web region near the angle. A thin beam web and long cope length increased the susceptibility to local web buckling. The depth of the beam section and connected position did not greatly affect the connection capacity. The design equations from current standards were used to evaluate the capacity of the specimens. It was found that the existing design standards did not provide consistent predictions of the block shear strength of coped beams with welded end connections. In addition, the rules provided by the standards could not accurately reflect the failure mode observed in the tests. For a better understanding of the connection behavior and the failure mechanism, non-linear finite element analyses (FEA) were carried out and presented in a companion paper (Yam et al. 2005), which 30

31 included a parametric study and a proposed design equation for the block shear strength of coped beams with welded clip angles connections. 7. Acknowledgments The authors would like to express their gratitude to the Research Committee of the University of Macau for providing financial support for this project. The assistance of the technical staff of the Structural Laboratory at the University of Macau, Macau, China, is also acknowledged. 8. References [1] Birkemoe, PC, and Gilmor, MI. Behavior of bearing critical double-angle beam connections, Engineering Journal, AISC, 1978; Vol. 15, No. 4, pp [2] Hardash, S, and Bjorhovde R. New design criteria for gusset plates in tension, Engineering journal, AISC, 1985; Vol. 22, No. 2, second quarter. [3] Epstein, HI. An experimental study of block shear failure of angles in tension, Engineering journal, AISC 1992; Vol. 29, No. 2, second quarter. [4] Epstein, HI, and Stamberg, H. Block shear and net section capacities of structural tees in tension: Test results and code implications, Engineering Journal, 2002; Vol. 39, No. 4, Fourth Quarter, pp [5] Epstein, HI, and McGinnis, MJ. Finite element modeling of block shear in structural tees, Computers and Structures 2000; Vol. 77, No. 5, pp

32 [6] Franchuk, CR, Driver, RG, and Grondin, GY. Experimental investigation of block shear failure in coped steel beams, Canadian Journal of Civil Engineering, 2003; Vol. 30, No. 5, pp [7] Orbison, JG, Wagner, ME, and Fritz, WP. Tension plane behavior in single-row bolted connections subject to block shear, Journal of Constructional Steel Research, 1999; 49: [8] Yura, JA, Birkemoe, PC, and Ricles, JM. Beam web shear connections: an experimental study, Journal of the Structural Division, 1982; ASCE, Vol. 108, No. ST2, pp [9] Ricles, JM, and Yura, JA. Strength of double-row bolted-web connections, Journal of Structural Engineering, 1983; ASCE, Vol. 109, No. 12, pp [10] Aalberg, A. and Larsen, PK. Strength and ductility of bolted connections in normal and high strength steels, In Proceedings of the Seventh International Symposium on Structural Failure and Plasticity, Melbourne, Australia, [11] Kulak, GL, and Grondin, GY. Block shear failure in steel members a review of design practice, In Proceedings of Fourth International workshop on Connections in Steel Structures IV: Steel Connections in the New Millennium, Roanoke, Va., October Edited by Leon, RT, and Easterling, WS., AISC, Chicago, Ill. pp , [12] Kulak, GL, and Grondin, GY. AISC LRFD rules for block shear in bolted connections a review, Engineering Journal, AISC, 2001, 38(4): See also Errata and Addendum, Engineering Journal, AISC, 39(2). 32

33 [13] Taylor, JC. More work is required, In Proceedings of Fourth International workshop on Connections in Steel Structures IV: Steel Connections in the New Millennium, Roanoke, Va, Edited by Leon, R T, and Easterling, WS., AISC, Chicago, Ill. pp 33-40, [14] Franchuk, CR, Driver, RG, and Grondin, GY. Block shear failure of coped steel beams, Proceedings of 4th Structural Specialty Conference of the Canadian Society for Civil Engineering, Montreal, Quebec, Canada, [15] Topkaya, C. A finite element parametric study on block shear failure of steel tension members, Journal of Constructional Steel Research, 2004; Vol. 60, Issue 11, pp [16] American Institute of Steel Construction (AISC). Load and resistance factor design specification for structural steel buildings, Chicago, IL, USA, [17] Canadian Standards Association. (CSA). CAN/CSA-S16-01 Limit states design of steel structures. CSA, Toronto, ON, Canada, [18] European Committee for Standardisation (ECS) (1992). Eurocode 3: design of steel structures Part 1.1: general rules and rules for buildings. ENV , Brussels, Belgium, [19] British Standards Institution (BSI) BS 5950: Structural use of steelwork in building Part 1: Code of practice for design rolled and welded sections, [20] Architectural Institute of Japan (AIJ). Standard for limit state design of steel structures, [21] Ministry of Construction of PR China (2003). Code for design of steel structures (GB ), Chinese Planning Press., Beijing, China, 2003 (in Chinese). 33

34 [22] Steel Construction Institute, Steelwork design guide to BS5950: Part 1:1990. Volume 1: Section properties, member capacities, 4 th edition, [23] British Standards Institution (BSI) (1990). BS 4360: Specifications for weldable structural steels, London [24] American Society for Testing and Materials (ASTM). Standard test methods and definitions for mechanical testing of steel products, ASTM Standard A370-02, Philadelphia, Pa, [25] Yam, CHM, Zhong, YCJ, Lam, CCA, and Iu, VP. An investigation of the block shear strength of coped beams with a welded clip angles connection, Part II: Analytical Study, Journal of Constructional Steel Research (companion paper). [26] Zhong, YCJ. Investigation of block shear of coped beams with welded clip angles connection. MSc thesis, University of Macau, Macau,

35 Table 1 Cross-sectional dimensions of the test beams Beam Designation Beam Serial t w (mm) T (mm) D (mm) B (mm) Note A, B (Beam406) UB406x140x Nominal Measured C, D (Beam457) UB457x191x Nominal Measured E (Beam356) UB356x171x Nominal Measured D tw B T 35

36 Table 2 Connection dimensions of the test specimens Designation A1-406r3 A2-406r2 B1-406t B2-406r1 C1-457R3 C2-457T D1-457R1 D2-457L E1-356T E2-356P Connected Connection Cope Cope Weld Length Position Length Depth Size a b p c d c s Note (mm) (mm) (mm) (mm) (mm) (mm) Nominal Measured Nominal Measured Nominal Measured Nominal Measured Nominal Measured Nominal Measured Nominal Measured Nominal Measured Nominal Measured Nominal Measured c dc p b b' a s s 36

37 Table 3 Summary of the test parameters Test Parameter Specimen Designation Nominal Value Aspect ratio (b'/a) A1-406r3, A2-406r2, B2-406r1 3.6, 2.3, 1.4 C1-457R3, D1-457R1 3.8, 1.6 Shear area, b (mm) C1-457R3, C2-457T 170, 120 Tension area, a (mm) C2-457T, D1-457R1 50, 90 Web thickness, t w (mm) B1-406t, C2-457T, E1-356T 6.8, 9.1, 9.1 Beam depth C2-457T, E1-356T UB457, UB356 Cope length, c (mm) D1-457R1, D2-457L 150, 80 Connection position, p (mm) E1-356T, E2-356P 20, 45 37

38 Table 4 Summary of the tension coupon test results Coupon Specimen Elastic Static Yield, Modulus F y (MPa) (MPa) Static Ultimate Strength, F u (MPa) Final Elongation (%) Beam406 (Beam A, B) Flange Flange Average Web Web Average Beam457 (Beam C, D) Flange Flange Average Web Web Average Beam356 (Beam E) Flange Web Web Average Clip Angle Angle Angle Average Steel Grade Material properties according to EN :2004 Elastic Modulus (MPa) Minimum Yield, F y (MPa) Tensile Strength, F u (MPa) Minimum Percentage Elongation (%) S S

39 Table 5 Summary of the test results Specimen Ultimate Load (kn) Ultimate Reaction (kn) Connection Moment* (kn. m) Web Thickness Reduction + (%) Failure Mode A1-406r Buckling A2-406r Buckling B1-406t Buckling B2-406r Buckling C1-457R Exceeded** C2-457T Fracture D1-457R Buckling D2-457L Exceeded** E1-356T Fracture E2-356P Buckling Note: * A negative moment indicates tension in the top flange. ** Failure did not take place at the maximum applied load. + Reduction in web thickness near the bottom of the clip angles. 39

40 Table 6 Summary of the predicted capacity and test capacity of the specimens Failure Mode F y F u Test Ultimate Reaction CSA-S16-01 (2001) Eq. (2) AISC LRFD (1999) Larger of Eq. (4) or Eq. (5) Predicted Test Eurocode 3 (1992) Eq. (6) AIJ (1990) Lesser of Eq. (9) or Eq. (10) Predicted Test GB50017 (2003) Eq. (11) Specimen (MPa) (MPa) Predicted Test Predicted Test Predicted Test Capacity Predicted Capacity Predicted Capacity Predicted Capacity Predicted Capacity Predicted A1-406r3 Buckling A2-406r2 Buckling B1-406t Buckling B2-406r1 Buckling D1-457R1 Buckling E2-356P Buckling Average C2-457T Fracture E1-356T Fracture Average C1-457R3 Exceeded D2-457L Exceeded Note: All resistance factors are taken as 1.0. The units of forces are in kn. 40

41 Cope Bolted connection Welded connection Figure. 1 Schematic of a bolted or welded connection on a coped beam 41

42 Cope Potential block shear of failure web material mode tearing off Bolted end connection Cope weld Potential block shear of failure web material mode tearing off Welded end connection Figure. 2 Potential block shear failure in a coped beam Note: The block shear failure generally consists of shear yielding on the gross area of the shear face and tensile rupture along the net area of the tension face 42

43 B A B A Rupture Surface Figure. 3 Block shear model of failure identified by Birkemoe and Gilmor (1978) 43

44 0.6Fy Fu Figure 4 Block shear model proposed by Ricles and Yura (1983) 44

45 c dc p D tw a s s b b' B T Figure 5 Schematics of the test specimen 45

46 E1 356T E2 356P Beam E (UB 356x171x67) 26 Dia. holes for M24 bolt Note: All dimensions are in millimeters Details of typical clip angle Details of A1 clip angle Figure 6 Overall design dimensions of the test specimens 46

47 A1-406r3 A2-406r2 B1-406t B2-406r1 C1-457R3 C2-457T D1-457R1 D2-457L E1-356T E2-356P Note: All dimensions are in millimeters Figure 7 Connection details of all of the test specimens 47

48 Reaction (Q) L Applied Load (P) Test Beam 2700mm 3300mm Lateral Bracing Support D Beam L (mm) A 510 B 510 C 600 D 600 E 550 Test Frame Load Cell (890kN) Roller Assemblies Hydraulic Cylinder (890 kn) Lateral Bracing Lateral Bracing Test Beam Supporting Column Test Connection Bracing Column Roller Assemblies Stub Support Load Cell (330kN) Concrete Strong Floor Figure 8 Schematic of the test setup 48

49 LVDT 5 LVDT 3 Applied Load (P) LVDT 8 (lateral) Reaction (Q) LVDT 4 LVDT 7 LVDT 6 LVDT 2 LVDT 1 Figure 9 Layout of LVDTs 49

50 Applied Load (P) Reaction (Q) (Back side) Figure 10 Layout of strain gauges 50

51 (a) Specimen B2 (b) Specimen C1 (c) Specimen C2 (d) Specimen D1 51

52 (e) Specimen E1 Figure 11 Photos of failed specimens B2, C1, C2, D1, and E1 52

53 Top flange Beam web buckled Clip angle Bottom flange (a) Specimen A1 (b) Specimen B2 Figure 12 Photos of the buckled shapes of specimens A1 and B1 53