Performance of T-Stub to CFT Joints Using Blind Bolts with Headed Anchors

The experimental program consisted of ten connection tests, with each configuration subjected to a monotonic tensile load. The objectives of the tests were (1) to establish the influence on connection behaviour when the tube is filled with concrete; (2) to compare the tensile performance of various blind bolted connections to that of conventional bolted connections; and (3) to investigate the connection behavior of anchored blind bolted connections by varying the principal joint parameters.

The testing matrix is summarized in Table 1, which outlines specimen index alongside the test variables. The test variables included type of fastener, addition of a concrete tube infill, strength of the concrete infill, spacing among bolts, and the class of connecting blind bolts. The test setup and the different configurations that were tested are shown schematically in Figs. 3 and 4, respectively.

The experimental setup simulated the idealized tension region of a typical connection designed to resist bending forces, with the critical force being the component of force acting at 90° to the tube face. Each specimen represented a double-sided joint constructed with T-stub sections and a total of eight bolts. The connections to the tube member were formed on opposite faces, with four bolts used on each face. The connections fastened with conventional bolt-nut-washer systems were designated “Type M” and were tested in combination with unfilled [Fig. 4(a)] (Barnett 2001) and CFT [Fig. 4(b)] (Ellison and Tizani 2004) sections. The connections fastened using the Hollo-bolt blind bolt were designated “Type HB” and were also tested in combination with unfilled [Fig. 4(c)] (Barnett 2001) and CFT members [Fig. 4(d)] (Ellison and Tizani 2004). “Type EHB” used Extended Hollo-bolt blind bolts connected to CFT sections [Fig. 4(e)].

All connecting bolts were tightened using a handheld torque wrench. Except those of property class 10.9 (designated Batch E), all test bolts were fastened at a tightening torque of $190\mathrm{N}\mathrm{m}$. Property class 10.9 bolts were tightened at $300\mathrm{N}\mathrm{m}$. All test connections had a tube width–to–thickness ratio ($b_o/t_o$) of 20 and a T-stub width–to–tube width ratio ($b_p/b_o$) of 1.1. The upper and lower T-stub sections, which had a 50-mm-thick flange, were designed as such to eliminate prying forces, allowing for an exclusive evaluation of interaction performance between the fastening system and the tube member.

To ensure that tube wall deformation was not affected by the open-end conditions, tube member length was arbitrary selected to be 900 mm. The joint was formed in the center of the length, which provided a distance of at least twice the tube width (i.e., $2b_o$) on either side of the joint. This was deemed suitable in terms of distributing the applied load without interference from the ends of the tube. To examine the effect of bending moment on the deflection of the tube face, and to investigate the performance of a group of EHB fasteners, the spacing between connection bolts was varied within practical ranges. The bolt gauge, which is defined as the transverse distance among bolt centerlines, varied from 90 to 120 mm. The bolt pitch, which is defined as the vertical distance between bolt rows, varied from 100 to 140 mm.

Connection tests were conducted under displacement control, using a 2-MN-capacity testing machine. The moveable crosshead was located on the upper side of the setup and the lower T-stub was fixed into position to allow the application of monotonic increasing load. Additional test instrumentation included displacement transducers and strain gauges, with the latter used on a limited number of EHB specimens only. Displacement transducers were used to monitor the displacement of the connected T-stubs relative to the tube face. Strain gauges were employed to measure strain along different directions in the underlying surface of the test part, with gauges mounted on the tube walls and on the embedded bolt shank. In particular, three strain gauges were mounted on the tube walls (S1, S2, and S3 in Fig. 3) and one on the bolt shank as close as possible to the end anchor (S4 in Fig. 3). The S1 and S3 gauges monitored the bending strain on the tube face in the $x$- and $y$-directions, respectively; S2 measured the axial strain on the tube side wall in the $z$-direction (at the mid-depth of the tube). Finally, S4 recorded the axial strain in the bolt shank in the $z$-direction, representing the development of mechanical anchorage due to resistance from the end anchor head. An actual connection specimen ready for testing is shown in Fig. 5.

The primary mechanical and geometrical properties of the different bolt groups used in the connection tests are summarized in Table 2, with each group being labeled with an alphabetic character to distinguish among the several bolt batches. All bolt shanks had a nominal (major) diameter of 16 mm. The bolts used to assemble the HB and M joint types had a total shank length of 120 mm, whereas the bolts used in the EHB joint types had a total shank length of 150 mm. The nominal grade and actual strength of concrete on the day of testing—for the CFT connections—are included in the test matrix in Table 1 next to the specimen indexes. The age of concrete was typically seven days on the day of testing.

To determine whether a concrete tube infill influences joint performance, connection tests were performed on unfilled and CFT joints for Type HB and Type M. The test results are graphed together for both configurations (i.e., without concrete and CFT) in Fig. 6 for Type HB and in Fig. 7 for Type M. The results are shown as normalized applied force versus average T-stub separation relative to the tube face.

The performance of the unfilled and CFT connections was exceptionally dissimilar in terms of stiffness, strength, ductility, and ultimate failure mode. For both tested joint types (i.e., HB and M), enhanced connection performance resulted when the tube was filled with concrete. Regarding connection stiffness, the force-separation curves achieved by the connections to CFT sections were significantly stiffer in comparison to those achieved by the unfilled connections. Connection ductility also increased for CFT connections with standard bolts (Type M). With respect to strength, the concrete infill increased the joint’s overall capacity. The strength of Type HB (CFT) increased by 50%, and the achievable strength of Type M (CFT) was higher by approximately 30% than that of the equivalent unfilled configuration.

The source of this additional strength was borne out by the failure modes observed during the tests. For example, the Type HB (without concrete) joint was seen to ultimately fail by the standard Hollo-bolt being pulled through the bolt hole in the tube because of the widening action of the conical expanded sleeve and the shearing off of the sleeve legs. The tests done on connections with standard bolts to unfilled tubes ultimately failed because of the stripping of the bolt threads, which occurred from secondary bending induced by excessive tube deformation. Indeed, for both configurations, local to the connection region, the walls of the unfilled tube underwent considerable deformation (Fig. 8).

In contrast, the ultimate failure mode of the joints with CFT sections was bolt shank fracture accompanied by negligible tube deformation (Fig. 9). Therefore, the enhanced connection behavior in the CFT case was attributed not just to the ability of the concrete to resist deformation of the tube walls; the tube infill also provided a favorable form of resistance by anchoring the connecting bolts. The concrete–bolt interaction improved the axial stiffness and strength of the connecting bolts, permitting their full tensile capacity to develop. The difference in response to applied force between unfilled and concrete-filled tubes is illustrated in Fig. 10, which highlights the desirable effects provided by the concrete as established for the tested configurations.

The performance of connections made with the standard Hollo-bolt was seen to be improved by the addition of a concrete infill, exhibiting equivalent strength and comparable ductility to those of the standard bolt connections. Similar to the conventional configuration (i.e., Type M), the strength of the connections to CFT made with Hollo-bolts was ultimately controlled by the capacity of their internal bolts. However, their connection stiffness was lower than that of the standard bolts. This is because the Hollo-bolt remained susceptible to being partially pulled through the hole prior to the development of the tensile capacity.

The lower tensile stiffness of connections made with the Hollo-bolt, even to tubes subsequently filled with concrete, means that the standard Hollo-bolt remains unsuitable for use in moment-resisting connections for the majority of configurations. However, recognizing the benefits gained by filling the tube with concrete (i.e., stiffening of tube walls and connecting bolts), a new blind bolt was designed for optimum interaction with the infill to achieve behavior comparable to that of conventional bolts. This new blind bolt—the Extended Hollo-bolt (EHB)—achieved optimum performance by maximizing mechanical anchorage within the readily available material. The EHB is a modified version of the standard Hollo-bolt, involving an extended internal, fully threaded bolt shank combined with an anchor head threaded onto its end—hence, “Extended.”

Fig. 11 presents the force-separation curves for connections to CFT made with (1) conventional bolts, (2) Hollo-bolts, and (3) EHBs. The results are shown in a normalized form, where the applied force for each connection is normalized relative to the maximum force reached in the tests (Table 3 lists actual failure loads). The results show all three connections to have an equivalent strength (equal to 1.0), which effectively equates with the strength of the connecting bolts because the ultimate failure mode for all tests was bolt fracture. As anticipated, the conventional connections exhibited the highest stiffness and the response of the joint formed using the EHB showed a favorable deviation toward the conventional data (i.e., Type M).

Knowing that the level of axial stiffness in conventional bolts is suitable for use in moment connections, achieving such stiffness characteristics indicates the suitability of the novel fastener in moment-resisting construction. The improved characteristics of EHB-anchored blind bolts are highlighted in Fig. 11, demonstrating that the axial stiffness of EHB is indeed improved by the modifications, although at the expense of some ductility. Fig. 11 shows that the ability of the EHB connection to absorb deformation is reduced compared to the other configurations. But, overall, the connection performance using EHBs is comparable to conventional bolting systems and therefore can be considered suitable for moment-resisting applications.

The remainder of this paper concerns the tests done on connections to CFT using the headed anchored EHB blind bolts, where a variation in the primary joint parameters was investigated to further understand connection behavior.

Two double-sided CFT joints, formed using EHBs, were tested to assess the influence of bolt pitch on connection behavior. One specimen had a bolt pitch of 100 mm and the other had a bolt pitch of 140 mm. The corresponding specimen indexes were EHB-8.8D-G120-P100-SHS10-C40 and EHB-8.8D-G120-P140-SHS10-C40. Considering tube size, the bolt pitches were selected based on practical distances commonly used in the design of moment connections.

Fig. 25 presents the normalized force-separation curves for the connection tests. Initially, the connection with the longer bolt pitch performed better than the one with the shorter pitch. However, beyond the level of yield force (0.8), the performance of the connections was very comparable, ultimately achieving an equivalent strength. Both joints failed because of bolt fracture.

For a closer examination of joint performance, Fig. 26 graphs the test results with respect to yield (${\delta }^y$) and ultimate separation (${\delta }^u$); joint ductility (${\delta }^{cd}$) was established similarly to the previous analysis, where ${\delta }^y$ was determined at a normalized force of 0.8 and ${\delta }^u$ at a normalized force of 1.0. Fig. 26 shows that the longer bolt pitch decreases the separation between the connecting members at the yielding force of the joint (evidenced by the reduction in ${\delta }^y$). It also increases the deformation capacity of the connection (evidenced by the increase in ${\delta }^{cd}$).

Bolts of property class 8.8 and 10.9 are most commonly applied in structural connections. The responses of EHB connections to CFT using these two property classes are compared in Fig. 27, where the applied force is shown normalized relative to the maximum force that was reached in the connection test which made use of grade-8.8 bolts. The corresponding specimen indexes were EHB-8.8D-G120-P140-SHS10-C40 and EHB-10.9E-G90-P140-SHS10-C40. Despite the different bolt gauges between the specimens, the comparison was considered valid because the variation in bolt gauge did not affect the characteristics of the joint (see the section “Influence of Bolt Gauge”) and the other primary joint parameters were not varied.

The response of the connection with grade-10.9 bolts was notably improved in comparison to that with grade-8.8 bolts in terms of stiffness and strength. Both joints ultimately failed because of bolt fracture. Assuming that the connection would develop the full tensile capacity of the grade-10.9 bolts, it was expected that the strength of the joint using these bolts would be higher. Indeed, the connection achieved the full capacity of the bolts and allowed for an increase of nearly 20% in capacity compared to the joint with grade-8.8 bolts. The theoretical ratio of the nominal strength of grade-10.9 bolts to that of grade-8.8 bolts is equal to 1.25 (1,000/800). The actual strength ratio measured for the bolts used in the tests was 1.23 (1,118/909). This is analogous to the increased capacity in Fig. 27, demonstrating consistency in the test results.

The enhanced stiffness of the connections that used bolts of property class 10.9 was attributed to the increased torque applied at their tightening stage (300 Nm instead of 190 Nm), which provided a higher clamping force between joint members that did not permit their separation at the early stages of loading. To highlight the influence of bolt property class, in Fig. 28 the yield (${\delta }^y$) and ultimate (${\delta }^u$) separation of the grade-8.8 connection is charted alongside the separation of the grade-10.9 connection at the same normalized force levels (0.8 and 1.0). The EHBs with internal grade-10.9 bolts resulted in much less separation at the corresponding forces.

The deformation capacity (${\delta }^{cd}$) is also shown in Fig. 28. To allow for a reasonable comparison of ductility, however, ${\delta }^{cd}$ for the grade-10.9 connection was established based on a yield separation determined at 0.9 of the maximum force reached in the test. The ratio of 0.9 was selected based on the ratio of bolt nominal yield to ultimate strength, which was equal to 0.9 for 10.9-grade bolts. It was identified that the ductility capacity of connections to CFT using EHBs with internal grade-10.9 bolts decreased compared to those with grade-8.8 bolts. This is because of the brittle mechanical properties of high-strength bolts (e.g., property class 10.9).