Specimen E1 attained the peak load at the midheight deflection, δc ¼ 3。63 mm, and  E4  with  stiffeners  attained  the  peak  load at δc ¼ 4。64 mm。 In E3 with mild steel, the peak load occurred at δc ¼ 4。42 mm。 In rectangular Specimens E2 and E5, the peak loads occurred at δc ¼ 8。53 and 8。96 mm, respectively。

 Axial load versus midheight lateral deflection relationship of specimens

Yield deflection δy was defined by the secant stiffness connect- ing the origin and 75% of the peak load, and maximum deflection δmax was defined as the postpeak deflection corresponding to 85% of the peak strength (Fig。 6)。 Ductility was defined as μ ¼ δmax=δy (Park 1988)。 In the square specimens, the ductility was the greatest in E4 with μ ¼ 4。97。 The ductility of E1 and E3 was μ ¼ 3。85 and 2。01, respectively。 In rectangular Specimens E2 and E5, the duc- tility was μ ¼ 1。97 and 2。71, respectively。

Failure Modes

Fig。 7 shows the failure modes of the specimens。 In E1, E2, and E4, local buckling occurred near the center of the columns。 On the other

hand, in E3 and E5, local buckling occurred in the upper region of the columns。 The local buckling initiated in the compressive flange and propagated to the neighboring  webs。

In E3 with mild steel, local buckling of the compression flange occurred at P ¼ 2,000 kN with yielding of the plate。 Afterwards, the tangential stiffness significantly decreased。 On the other hand, in E1 and E2 with high-strength steel, local buckling occurred in the slender flange plates at P ¼ 4,000 and 1,500 kN, respectively。 However, the load-carrying capacities continued to increase with- out significant stiffness degradation。 In E4 and E5 with vertical stiffeners, local buckling occurred at P ¼ 7,000 and 4,000 kN, re- spectively, which was close to the peak load。 As shown in Fig。 7, the buckling width and length were decreased by the effect of the stiffeners。 In E2, local buckling did not occur in the compact web plates。 In all the specimens, failure occurred due to crushing of the infill concrete, which was detected from the sound during testing。 In E1 and E4, at 77% of the peak load in the postpeak behavior, weld fracture occurred at the joint of the compressive flange and web plates。 This result indicates that careful attention is   required

when welding high-strength steel  plates。

Strains of Steel Tube

Fig。 8 shows the strain distributions at the midheight tube section of E4 and E5 with stiffeners。 In the case of specimens without stiff- eners, the strains were not accurately measured due to local buck- ling at early loading。 In Fig。 8, the horizontal axis refers to the distance from the center of the cross section。 Until 80% of the peak load, the strains were linearly distributed in both E4 and E5。 At the peak load, the strains showed nonlinear distribution due to local buckling and yielding。 However, the strains at the flange are appa- rently greater than 0。003, which is the ultimate compressive strain of ordinary concrete under axial compression。 This result indicates that the deformation capacity of the concrete infill increased due to

 Comparisons of test results and fiber model analysis: (a) E1; (b) E2 (c) E3; (d) E4; (e) E5; (f) stress-strain relationships of steel and concrete for fiber analysis

Table 2。 Summary of Test  Results

Stiffness Strength Ductility

Specimen δy

(mm) Ke

(kN=mm) δc

(mm)