This specimen comprised bare steel sections having the same end plate connection as SFB3 but without the compound plate at the bottom flange. The M-<|> curves as indicated in Fig. 4-10 showed a linear behaviour at the early stages of loading. The moment
against deflection (Fig. 4-11 ) also showed a linear behaviour at the early stages of loading. At a moment o f 21 kNm a slight gap was observed between the column flange and the end plate. As moment increased this gap became wider. At 51 kNm noticeable bending was observed in the end plates. Although specimen TEST4 had the same connection arrangement as SFB3, the calculated moment o f resistance was slightly lower at 36.6 kNm, due to different value o f material properties (see Table 4-3).
It was decided to end the test when the applied moment was about 65 kNm at about 55 mrad to avoid the possibility o f the bolts fracturing. Deformation of the end plates was the principal cause of failure. No local buckling o f flanges and web was observed. Plate 4-6 indicated how the end plate had deformed.
4.8.5 Behaviour of TESTS
Compared to the previous specimen the connections in specimen TESTS were stiffer as larger bolts and thicker end plates were used. From the M-<|> curves (Fig. 4-12) it was observed that initially the rotation increased linearly as the applied moment increased. At the beginning of the test the right hand beam gave a negative rotation reading. It was thought that initial curvature o f the end plate o f the connection on the right hand beam contributed to this (see Plate 4-7). The moment-deflection curves (Fig. 4-13) also indicated that deflection increased linearly as the applied load increased.
The calculated moment resistance was 66.1 kNm. It was higher than the calculated moment of resistance for SFB2 (62.9 kNm) although both had the same end plate arrangement. Again this was due to the slight difference in the measured material properties. The predicted mode of failure was mode 2 (yielding + bolt failure). At the early stage of the test the nuts were tack welded to the end plate as a precautionary
measure against catastrophic bolt failure. It was decided to stop the test at about 96 kNm to avoid the danger o f the bolt fracturing; the calculated bolt force was than approximately 250 kN. At this juncture the rotation had reached just 25 mrad.
4.9 Assessment of Test Results
The average values o f M-<|> curves for all the specimen are illustrated in Fig. 4-14a to Fig. 4-14e inclusive. Table 4-4 shows the summary o f the mode of failure calculated based on the basis o f Annex J o f EC3[4-8]. The value o f the tensile strength of the bolt was obtained by reference to the values determined through the experiment by Godley and Needham[4-9], The comparisons indicated that the ultimate tensile strength of bolts were 28% above the nominal value.
The predicted and the experimental resistance moments (MRi) are given in Table 4-5, along with the failure modes. The predicted values are based on measured material properties. Annex J o f Eurocode 3 [4-8] were used to predict the resistance moments of the connections, ignoring the concrete. The proposals by Jaspart[4-10] formed the basis by which the prediction for the experimental values were derived. The experimental moment o f resistance o f each connection was defined as the moment at one third of the initial stiffness of the test. This stiffness may be obtained from the initial response of the specimen or more conveniently, from the unloading-reloading response at a later stage of the test.
The failure modes during the tests were identical to those predicted by Annex J. However the calculations show that Annex J of EC3[4-8] significantly underestimates the moment reached in all the tests. Two values o f Mm from the tests for SFB1, SFB2, and SFB3 are given in Table 4-5. This is because there are two different values of “initial” rotational stiffness 5y/. One is the response before the mesh fractures. The
second is from the unloading-reloading response after the mesh had fractured. Except for SFB3 (left) all the experimental resistance moments were greater than the calculated values.
None o f the specimens (except SFB1 left) showed a peak in the M-<)> curve. No test exhibited a true plastic plateau. This behaviour is typical of steel end plate joints in which brittle modes of failure have been prevented. In SFB1 (left), both fracture o f the mesh and bolt stripping had occurred.
As previously mentioned, initial rotational stiffness o f a connection can be determined from the unloading parts o f the M-<|> curves in a test. These are recorded in Table 4-6. The two values result from the notable difference observed for all the encased specimens before and after the mesh fractured. It is found from Table 4-6 that after the mesh fractured, and the concrete had cracked significantly, the initial stiffness value is similar to that given by the corresponding bare steel test.
Table 4-6 also shows the values of rotational stiffness at various stages o f loading. There were two calculated values for initial stiffness based on ENV 1993-1-1/pr A2[4- 11]; one by considering the stiffened column web due the effect o f encasement; the other by including the column flexibility.
Effects o f encasement
The concrete encasement effects can be clearly seen from Fig. 4-15 and Fig. 4-16 for comparison between SFB3 and TEST4 and comparison between SFB2 and TESTS respectively. At low rotation there is a much higher initial rotational stiffness for the encasement specimens. However, after the concrete had cracked and the first mesh fractured the stiffnesses were comparable to the corresponding bare steel value. From