In the past several decades, CFST have been widely used in seismic regions, due to their excellent earthquake-resisting proper- ties (Elremaily and Azizinamini 2002; Han and Yang 2005;  Han et al. 2003, 2006; Varma et al. 2002). Considering the behavioral difference between stainless steel and carbon steel, there  is  a need to understand the seismic behavior of CFSST columns before they can be safely used in earthquake-prone zones. However, research on CFSST columns under cyclic lateral loading is still missing, which indicates a need for further research in this area. Consequently, this paper presents tests on 10 CFSST columns conducted by applying cyclic lateral loading while keeping the ax- ial load constant. The main objectives of this research are threefold:

(1) to report a series of new tests on CFSST columns, investigating the influence of axial load level, cross-sectional type, and infilled concrete type on the seismic behavior of CFSST columns; (2) to compare the hysteretic performance of stainless steel composite specimens with that of conventional carbon  steel counterparts; and (3) to compare the measured ultimate strength and flexural

(a) (b)

Fig. 1. Cross sections of CFSST columns: (a) circular cross   section;

(b) square cross section

   Fig. 2. Indicative stainless steel and carbon steel σ − ε  relations   

stiffness with predictions from several existing codes for carbon steel CFST columns.

Experimental Program

General

Ten CFSST column specimens were tested. Test parameters were the cross-sectional type (circular  and  square),  axial  load level (n ¼ 0.02, 0.3, and 0.6), and infilled concrete type [recycled aggre- gate concrete (RAC) and normal concrete]. The axial load level (n) in this paper is defined as  follows:

 N0

determined by using the finite-element (FE) model developed previously by Han et al. (2013), where the measured material properties in the current tests were used in the calculations. An axial load level (n) of 0.3 or 0.6 reflects the actual load level for columns in real buildings. And, to investigate the flexural behavior of the composite columns, a very small axial load level of 0.02 was chosen for two specimens. The application of a small axial load was to ensure the specimens were properly supported. Table 1 summarizes the detailed information of the test speci- mens, where D and B are the overall diameter of a circular section and width of a square section, respectively; and t is the wall thick- ness of the steel tube. The specimen labels listed in Table 1   were

designated by using the following  rules:

• The initial character C or S means a circular or square section, respectively;

• The following character N or R denotes that the infilled concrete is normal or recycled aggregate concrete, respectively;  and

• The  last  number  0,  3,  or  6  stands  for  the  axial  load level

n ¼ 0.02, 0.3, or 0.6, respectively.

Due to the limitation of the test setup, only small-scale speci- mens could be tested. Therefore, the diameter D and width B of the cross section was selected as 120 mm and the thickness of the steel tube was selected as 4.0 mm for all specimens. Further research is required to clarify any size effect on the column   behavior.

Material Properties

Commercially available cold-formed grade American Iron and Steel Institute (AISI) 304 austenitic stainless steel tubes were used to fabricate the CFSST specimens. A series of tensile coupon tests was conducted to obtain the material properties of the stainless steel in the finished cold-rolled condition. The coupons were cut in the longitudinal direction of the tubes at locations opposite to the seam weld. Only the mechanical properties of the flat part of the square cross section were obtained. The highly worked corners are ex- pected to have higher yield stress and ultimate tensile strength, which can be predicted using models available in the literature (Tao et al. 2011; Uy et al. 2011). The results indicated that the stain- less steel showed obvious nonlinear stress–strain characteristics. The three basic Ramberg–Osgood parameters, i.e., initial elastic modulus E0, 0.2% proof stress σ0.2, and strain-hardening exponent