Keywords

1 Introduction

Concrete filled composite (CFST) columns made of normal-strength materials have proven their suitability under seismic loading in engineering practice over the last decades, and their high resistance has been confirmed in several research projects e.g. [1,2,3,4]. A further development of these type of composite columns are concrete filled double steel tubes (CFDST) - columns with high strength building materials. These types of columns consist of an outer and an inner steel tube. The gap between the outer and inner tube as well as the inner tube itself are both filled with concrete. In comparison to conventional composite columns the external dimensions can be reduced while maintaining the same load-bearing capacity and due to the prefabrication in factories the time required to erect individual floors is minimized while the manufacturing tolerances meet the standards for steel structures.

The load bearing behaviour of these innovative columns with high performance materials has already been investigated at ambient temperatures and in case of fire [5,6,7]. On the other hand, CFDST-columns with high performance materials are not yet implemented in standards [8, 9] and, in particular, the seismic performance has not been comprehensively investigated so far.

In this paper, the experimental results of three test series of CFDST-columns and one test series of CFST-columns under cyclic loading are presented. The columns were tested under bi-axial loading at the Ruhr-Universität Bochum. The cyclic performance of the composite columns is evaluated in terms of hysteric behaviour, ductility and energy dissipation.

2 Experimental Program

2.1 Test Specimen

In this study a total of eight specimens were tested with a total length of 1,500 mm, Table 1. Six specimens were manufactured as CFDST-columns (no. 1 to 6) with normal and high-performance materials, two specimens were manufactured as CFST-columns (no. 7 and 8) with normal strength materials. These columns were used as reference to evaluate the performance of the CFDST-columns. The outer tube had a diameter of 273.0 mm and a wall thickness of 8.8 mm. Only normal strength steel was used. The inner tube had the same wall thickness and a diameter of 177.8 mm. Both grade S355 and S770 steels were used.

Table 1. Properties of the tested specimens
Full size table

The columns were manufactured from steel tubes of the same batch. For each batch three tensile specimens were tested. The compression strengths of the concrete specimens were tested at the same day as the tests of the columns using three cube specimen per batch. Table 1 shows the mean values of the material characterization test results and the dimensions of each column.

2.2 Test Setup and Measurement Equipment

The test setup was arranged in a 20 MN universal testing machine (Fig. 1). A lateral actuator (1,000 kN/±125 mm capacity) was fixed in horizontal orientation with hinges at both ends (cardan joints) and centered to the intermediate plate. A sliding plate (stainless steel) and a greased UHPWE-layer was added beneath the tested specimens. Between the foot of the columns and the intermediate plate, a bolt joint was installed to enable rotation during the horizontal movement. The head of the specimens is centered in a stiffened steel quiver (2 × HEB400), filled with grout mortar and connected to the crosshead of the machine. The length of the quiver clamping was l = 500 mm. The free length of the specimen was 1,000 mm.

Fig. 1.
figure 1

(a) Test setup, (b) schematic setup, (c) view of the DIC – system

Full size image

The displacement and rotation of the test specimens and the setup components were measured by means of digital image correlation (DIC) (Fig. 1c). On each tube two strain gauges were installed 50 mm outside of the quiver clamping.

2.3 Loading Protocol

Fig. 2.
figure 2

Monotonic and cyclic test curves of specimen no. 1 and 2

Full size image

The tests were performed according to the loading protocol presented in [10]. An overview of the loading parameters is given in Table 2. Figure 2 illustrates with the axial load time-histories of specimens no. 1 (monotonic) and 2 (cyclic) the loading protocol. The specimens were first continuously loaded by an axial load (i.) up to NTest = 0.3 Npl, which is the axial load limit in columns design according to [8].

Table 2. Loading parameters
Full size table

The plastic resistance Npl was calculated according to Eurocode 4 [9] with the mean values of the material tests listed in Table 1. From this point on forward, the axial load N was kept constant in load control and the lateral displacement Δa (ii.) was applied displacement-controlled. For each parameter set of the cross-section dimensions two loading types for the lateral displacement Δa were considered – (a) a monotonic test (Fig. 2a) and (b) a cyclic test (Fig. 2b). In the monotonic load test (a), the specimens were continuously loaded up to the maximum lateral actuator displacement Δa of ±125 mm. The monotonic-test was used to determine the displacement Δy corresponding to the yield load Fy as input value for the definition of the cyclic loading protocol. In the cyclic loading part (b) the loading was performed with incremental increase of the displacement Δy until the maximum actuator displacement of Δa = ±125 mm was reached (see Fig. 2b).

3 Results

3.1 Accuracy of the Test Facility

The quality of the specimens mounting in the test apparatus was examined. Figure 2 shows, as an example, additional to the control (primary) parameters of the tests (continuous lines) the evolution of the measured (secondary) parameters, like the lateral load F (red) and the lateral specimen displacement Δ (blue) (dotted lines) over the course of the test time. The lateral load axis was 548 mm under the foot of the specimens (Fig. 1d). Therefore, the actual specimen displacement Δ result smaller than the actuator displacement Δa (eg. Fig. 2a, ΔSpecimen,no.1,max = −96.89 mm). The comparison between the primary and secondary parameter curves illustrates an immediate load application without any slippage in the setup.

3.2 Hysteretic Behaviour

The hysteretic behaviour of the columns is analysed in terms of lateral load F vs. displacement Δ curves, where the events yielding, reaching the ultimate load Fmax and reaching 80% of Fmax are separately marked (Fig. 3). The corresponding values are presented in Table 3. The characteristic phases of the hysteric loops are plotted in red (elastic), blue (non-linear behaviour) and green (post ultimate range).

Fig. 3.
figure 3

Lateral load F vs. displacement Δ

Full size image

Generally, the CFDST showed good resemblance to the CFST in the shape of hysteresis curves (Fig. 3a to c vs. 3d), implying that the CFDST-column type had a global cyclic response akin to that of the CFST-type. All specimens showed a symmetrical and ductile behaviour during the cyclic loading. In addition, the CFDST-columns (no. 2, 4, 6) exhibited a fuller hysteretic curve compared to the CFST (no. 8), indicating a better energy dissipation capacity (Fig. 4).

Table 3. Lateral displacements and ultimate lateral load
Full size table

Each column displayed material softening under repeated loading at the same displacement. The use of a high-strength inner tube increases the stiffness and lateral load capacity Fmax (no. 2 and 4) compared to specimen no. 6 (S355 inner tube) or specimen no. 8 (CFST). While at the same time the influence of the high-strength concrete had less influence on the lateral load capacity by comparing specimen no. 2 and 4 (see Table 3).

Fig. 4.
figure 4

Envelope curves and strain distributions for specific events

Full size image

The differences in cyclic behaviour of the CFDST and the CFST is analysed in terms of lateral load F versus lateral displacement Δ envelope curves of specimen no. 4 and 8 (Fig. 4). The lateral stiffness and ultimate lateral load Fmax were positively affected by the inner steel tube. The comparison of the envelope curves displays an increase in stiffness and of the ultimate load Fmax. The post ultimate behaviours of both types can be regarded as highly ductile, while no. 4 exceeds the ductility of no. 8 despite the high strength core materials. The envelope curves (Fig. 4) describe a mostly symmetrical behaviour under cyclic loading with slight deviations resulting from geometric imperfections and after the initial buckling of the specimens.

The measured strain distribution of key events (I–V) during the cyclic tests are displayed in Fig. 4. Event (I) marks the point when the axial load NTest was applied (F = 0). The effect of the geometric imperfections on the strain distribution is shown by inclination the in the distribution on each tube. For the CFDST a difference in stain was measured between the inner and outer tube indicating a gap between the inner tube and base plate of the clamping. In event (II) the outer steel tube reached its yield limit of εyield, S355 = ±1,974 µm/m. Because of the higher axial load NTest, of specimen no. 4, the yield limit was reached already at a displacement of ΔII,no.4 = −6.77/+7.09 mm compared to corresponding displacement of specimen no. 8 ΔII,no.8 = −8.41/+7.33 mm. After the yielding of the outer tube the load shifts to the inner steel tube (no. 4), which is still elastic at this point of the loading (II). At event (III) the maximum concrete tension strain of 100 µm/m was reached. This point was calculated indirectly by strain gauge measurements on the tubes. As shown at (I) the axial loads lead to different compression strain levels between the two specimens. Due to the bending moment applied resulting from the lateral displacement, the CFST-column reached event (III) before its tube began to yield (II). For the no. 4 the behaviour was reversed. With further loading the F over Δ curve of the CFDST-specimen begins to slowly show a non-linear behaviour until the inner tube starts to yield (IV) at a strain of εyield,S770 = −4,180 µm/m. Shortly after event (IV) the maximum lateral load Fmax (V) is reached. The CFST (no. 8) had no additional inner tube and could no redistribute the load, a bent in the envelope curve can be observed between point (II) and (III) until the maximum load Fmax,no.8 is reached (V).

3.3 Comparison and Evaluation

The specimen’s ductility µ is defined as the ratio between yield displacement Δy and the lateral displacement Δ80% Fmax at 80% of the ultimate lateral load Fmax. Figure 5a presents the cyclic ductility µ for the cyclic tested columns for both load directions. The values of all columns resulted between 4.43 and 5.72. The overall ductility µ of the CDFST-columns is comparable or exceeds the ductility µ of the CFST-specimen and can be regarded as highly satisfactory. According to [4] the ductility of columns can be rated as “very good”, if the value µ is >5.0.

The dissipated energy per cycle (EDC) is defined as the area of the F vs. Δ – curve shown in Fig. 3. Figure 5b shows the actual EDC (dotted lines) as well as the normalized EDC related to the CFST-column (no. 8) (continuous lines). The lowest amount of energy is dissipated by the CFST-column (no. 8). The largest amount of energy was dissipated by CFDST-column (no. 6), with up to 3.2 times the energy of no. 8 at a displacement ratio of 2.0. The EDC values of specimens no. 2 and no. 4 were between the extremes of the normal strength CFST and CFDST-column. The reason for that was the higher yield limit of the S770 inner tube. The results show that normal strength core materials are more beneficial for energy dissipation compared to the high-strength core materials. Nevertheless, all CFDST-specimens independent from there material configuration exceeds the dissipation capacity of tested normal strength CFST-specimen.

Fig. 5.
figure 5

(a) Ductility µ and (b) EDC of the columns

Full size image

4 Conclusion

In this experimental study six CFSDT-columns with normal and high-strength materials and two CFST-columns with normal strength materials were tested. The following conclusions can be drawn regarding the performance of the normal and high strength CFDST-columns compared to normal strength CFST-columns: I) The axial and lateral load bearing capacity of normal and high strength CFDST-columns is higher than the capacity of the tested normal strength CFST-column. II) The ductility µ of the tested CFDST-columns is on the same level or surpasses the CFST-column despite the use of the high-performance materials. III) The additional steel tube of the CFDST-columns leads to an increased energy dissipation capacity compared to the CFST-column.