Load versus  Displacement  Hysteretic Curves

Based on the displacements measured along the specimen, it was found that the measured deflection curves are generally in the shape of a half-sine wave. Figs. 6(a and b) show typical sets of deflection curves for a circular section and a square section respectively, where the measured lateral displacements are marked by solid circles and the sinusoids with the same deflections at the midspan are shown by dashed curves. It can be seen that generally the mea- sured deflection curves of CFSST columns are in accordance with the half-sine wave. This comparison indicates that the deflection curve of a CFSST member may be assumed as a half-sine wave in future simplified numerical analysis.

Fig. 5. Failure modes of tested specimens: (a) all specimens after testing; (b) typical failure modes (CN-3); (c) typical failure modes (SN-3)

seen even for specimens under a relatively small axial load level. Compared with the square section, a circular steel tube tends to offer better confinement to its  concrete  core.  For  this  reason, the circular CFSST columns have plumper hysteretic loops than the square ones. The axial load level has significant influence on the shape of the hysteretic curves, particularly in the unloading stage. As the axial load level increases, the slope of the postpeak curve becomes much steeper, the hysteretic loop becomes less plump, and the energy dissipation capacity decreases significantly. In general,

(a)                                                         the features of CFSST columns under constant axial load and cycli- cally increasing flexural loading are quite similar to those of the

conventional CFST columns (Han and Yang 2005; Han et al. 2006), except that the CFSST columns seem to show higher deformation ability whether the axial load level is low or   high.

Fig.  6.  Lateral  deflections  along  the  length  of  typical specimens:

(a) CN-3; (b) SN-3

The recorded lateral load (P) versus midspan lateral displace- ment (Δ) hysteretic curves for all specimens are shown in Fig. 7. The hysteretic loops of the P − Δ curves of CFSST columns are very plump under cyclic loading and no obvious pinching effect is

P −Δ Envelope Curves

The lateral load (P) versus lateral displacement (Δ) envelope curves of all specimens are shown in Fig. 8, where the curves were obtained by connecting the peak point of each loading increment on the hysteretic curves. Table 1 gives the ultimate lateral loads (Pue) obtained in all tests, calculated as the average values of the positive and negative maximum lateral loads obtained from the P − Δ re- sponses. For some specimens, such as CN-0, the lateral load P kept increasing without any descending stage for its P − Δ envelope curve. In this case, the ultimate strength (Pue) was determined as the lateral load corresponding to the time that the maximum tensile strain of the steel tube attained 10,000 με (Han et al. 2006).

It was found from Fig. 8 that both the ultimate strength and duc- tility of the tested CFSST columns decreased with the increase  of

© ASCE 04016219-5 J. Struct. Eng.

150 150 150

75 75 75

0 0