Flexural behavior of precast concrete-filled steel tubes connected with high-performance concrete joints

Ten slender PC columns, each with a circular cross-section and a diameter of 100 mm, were constructed and subjected to monotonic flexural loading until failure. To assess the bending capacity of these slender PC columns, precast NC was used to fill circular steel tubes on the left and right parts of the column, as illustrated in Figure 1. Various types of intermediate concrete connections were tested following the concept introduced in the study of Hamoda et al. [7]. Additionally, a standard slender RC column, measuring 750 mm in total length without intermediate connections and steel tubes, was also tested, and it served as a control specimen.

Figure 1

Geometric and reinforcement details: (a) reinforcement details for beam B0 and (b) details of the connection zone (units: mm).

Similarly to previously published studies, e.g., Hamoda et al. [7,8], two primary parameters were examined within the test program: the type of HPC used for the connection (NC, ECC, and UHFRC) and the development length of the embedded steel bars or connecting concrete joint (L). Three different embedment lengths – 150, 200, and 300 mm – were investigated, as summarized in Table 1. The columns with intermediate connections were categorized into three groups based on the connection type (G1, G2, and G3). In the naming convention for the tested columns, “N” represents NC, “E” stands for ECC, and “H” indicates UHFRC. The subsequent number (L1, L2, and L3) denotes the development length of the embedded steel bars, which are 150, 200, and 300 mm, respectively.

The specimens were produced using ready-mix NC, ECC, and UHFRC, with their respective mix proportions detailed in Table 2. As indicated in Table 2, ECC requires a higher amount of high-range water reducer (HRWR) compared to UHFRC. Therefore, from an economic standpoint, UHFRC is a more favorable choice for filling the joint connections of precast columns.

To characterize the properties of the materials used in the test program, both uniaxial tension and uniaxial compression tests were conducted. Compression tests were performed using 150 mm × 300 mm cylinders on the same day of experimentation (Figure 2(c)). Uniaxial tension tests were conducted using dog bone samples, following the specifications and recommendations of the ACI [43], as shown in Figure 2. The idealized and measured compressive and tensile stress–strain responses of NC, ECC, and UHFRC are depicted in Figure 3(a) and (b), respectively.

Figure 2

Uniaxial tensile and compressive tests for concrete: (a) concrete specimen dimensions for tension, (b) test of the concrete specimen under tension, and (c) test of the concrete specimen under compression.

Figure 3

Constitutive stress–strain law for all concrete specimens: (a) under compression and (b) under tension.

Additionally, tensile tests were carried out on coupons to evaluate the mechanical properties of the GSS and steel bars used. The recorded tensile stress strain responses of the steel bars and GSS are presented in Figure 4, along with idealized multilinear uniaxial tension laws [7].

Figure 4

Actual and idealized stress strain law for steel elements.

As illustrated in Figure 5(a), circular steel tubes were constructed using GSS secured with three steel bolts. These tubes were then filled with NC to create two precast CFST panels for each column to be tested. The precast CFST panels had a height of 300 mm, with additional lengths for the connections (L _d) of 150, 200, and 300 mm.

To build the full-length columns, two 300 mm long precast CFST panels were assembled in opposite directions, leaving a gap (L _d) for casting NC, ECC, or UHFRC inside, as shown in Figure 5(a) and (b). For example, combining two 300 mm precast CFST panels with L _d values of 150, 200, and 300 mm resulted in total column lengths of 750, 800, and 900 mm, respectively. The connection between the precast CFST columns and the NC/ECC/UHFRC infill was achieved using a cylindrical formwork, as depicted in Figure 5(b), into which the flowable NC, ECC, or UHFRC material could be poured.

Figure 5

Preparation of connected beams: (a) preparation of pre-casted panels and (b) connecting of panels and casting of the connection zone.

The test program described in this study was conducted at the testing laboratory of Kafrelsheikh University in Egypt. The circular precast slender CFST columns were tested in four-point bending using a 1,000 kN hydraulic jack securely fixed to a stiff steel frame, as shown in Figure 5(a). The loads were applied via loading plates, as illustrated in Figure 6(b). The distance between the supports and the loading plates was kept constant across all tests to allow for a consistent comparison of the flexural capacity of the tested columns.

Figure 6

Test set-up and details of the instrumentations: (a) schematic of tested connected beam and (b) image of the beam during the test (units: mm).

A linear variable displacement transducer was strategically positioned vertically to measure mid-span deflection, as depicted in Figure 6. Additionally, two PI-shaped displacement transducers were employed to measure the crack opening displacement on the surface of the concrete column. These transducers were connected to a system that automatically collected recorded data during the experimental process.

The control specimen B0 exhibited a failure mode characterized by flexural cracking and crushing in the compressive zone, as shown in Figure 7. This mode of failure, involving a combination of flexural cracks and compressive failure, aligns with the typical behavior observed in RC structures under bending stress, e.g., Hamoda et al. [7]. For the precast CFST columns with NC, engineered cementitious composite (ECC), and ultra-high-performance fiber-reinforced concrete (UHFRC) connections, distinct failure patterns were identified. In CFST columns with NC connections, failure was initiated by a diagonal crack at the junction between the NC joint and the CFST part (Figure 8), a phenomenon also noted in studies where material discontinuities lead to stress concentrations [7]. Additionally, horizontal cracking was observed in the column with a 200 mm NC connection at the ultimate loading stage.

Figure 7

Failure mode of master beam (B0).

Figure 8

Failure modes of Group G1: (a) Beam N-L1, (b) Beam N-L2, and (c) Beam N-L3.

In the columns with ECC connections, separation between the ECC joint and the CFST part was evident in those with longer connection lengths of 200 and 300 mm (Figure 9b and c). This type of debonding or interface failure between dissimilar materials has been widely reported in the literature, particularly in composite systems under high flexural or tensile loads [7]. Interestingly, the column with the shortest connection length exhibited a diagonal crack similar to those observed in the NC-connected columns, further supporting the hypothesis that shorter joints can experience stress concentrations at the interface.

Figure 9

Failure modes of Group G2: (a) Beam E-L1, (b) Beam E-L2, and (c) Beam E-L3.

The columns with UHFRC connections showed a combination of diagonal and horizontal cracks, along with crushing in the compressive zone (Figure 10). This mode of failure, involving both cracking and crushing, is often associated with the high strength and stiffness of UHFRC, which can resist tensile cracking but may eventually experience compression failure at ultimate loads [7].

Figure 10

Failure modes of Group G3: (a) Beam H-L1, (b) Beam H-L2, and (c) Beam H-L3.

Table 3 summarizes the evaluated values of crack loading (P _cr) and ultimate loading (P _u), along with the corresponding vertical displacements (Δ_cr and Δ_Pu).

The load–deflection response of all investigated columns is depicted in Figure 11. The measured mid-span deflections at the cracking force and the ultimate force are listed in Table 3. Generally, all columns exhibited linear behavior up to the point of crack development. As shown in Figure 11 and Table 3, columns with UHFRC connections demonstrated the highest stiffness compared to those with ECC and NC connections. The columns with NC connections displayed a relatively lower bearing capacity than the reference columns B0, whereas the columns with UHFRC and ECC connections showed greater bearing capacity than the reference columns B0, as illustrated in Figure 11.

Figure 11

Load–deflection relationships for tested connected beams: (a) Group G1, (b) Group G2, and (c) Group G3.

The higher stiffness in UHFRC-connected columns can be attributed to the material’s higher tensile strength and fiber reinforcement, which limit crack development and deflection. UHFRC’s improved crack resistance delays the reduction in stiffness, resulting in smaller deflections compared to ECC and NC connections. In contrast, NC connections exhibit larger deflections due to the brittleness and lower tensile strength of NC, which cause earlier crack formation and a quicker reduction in stiffness. ECC connections show a more balanced behavior, as the material’s strain-hardening properties and multiple micro-cracking mechanisms help distribute the load more effectively, resulting in greater ductility and controlled deflection.

All columns transitioned into the plastic phase and exhibited hardening until the ultimate load was reached, as shown in Figure 11. Notably, the columns with UHFRC displayed more ductile behavior compared to those with NC, while the columns with ECC exhibited the most ductile behavior among the three types of connections investigated in this study.

The absorbed energy (E) of all columns was calculated as the total area under the load–displacement curve following the approach commonly used in the literature (e.g., Hamoda et al. [28,44,45]). A comparison of the calculated energy E values for all tested columns is shown in Table 3. The evaluated absorbed energy of the columns with NC connections was relatively lower than that of the reference column B0. In contrast, columns with ECC connections demonstrated higher energy absorption compared to the reference column B0. For all column types, the absorbed energy increased with the length of the embedded connection (Figure 12).

Figure 12

Absorbed energy (E) for all beams (kN mm).

Incorporating ECC with strain-hardening behavior and UHFRC with crack-bridging effects in the intermediate joint, along with the GCSS tubes, led to significant improvements in energy absorption over the control NC column. For instance, as shown in Table 3, the energy absorption of columns E-L3 and H-L3 increased by 69 and 70%, respectively, compared to the control column B0.

The recorded elastic stiffness values (K) for all columns are summarized in Table 3. These values are defined by the linear response slope of the load versus mid-span deflection relationship. All precast CFST slender columns with NC connections exhibited lower elastic stiffness compared to the control specimen B0. However, CFST columns with UHFRC connections showed higher elastic stiffness than the control specimen B0, as shown in Figure 13. For example, the column with a 300 mm embedded length exhibited a 110% increase in elastic stiffness. The elastic stiffness for all column types increased with the length of the embedded connection, as shown in Table 3.

Figure 13

Elastic index (K) for all beams (kN/mm).

A summary of the recorded loads at which the first crack (P _cr) occurs, as well as the ultimate loads (P _u) for all tested columns, is presented in Table 3. The recorded cracking force in the CFST precast columns with NC connections was lower than that of the control specimen B0, whereas the recorded cracking force for the columns with UHFRC exceeded that of the control specimen.

For example, columns with UHFRC joints, specifically FC-L2 and FC-L3, exhibited crack loads 28 and 47% higher than that of the control column B0, respectively. For columns with UHPECC joints, the crack loads of H-L1, H-L2, and H-L3 were 7, 12, and 18% higher than the crack load of the reference column B0, respectively.

It is evident that the delay in the onset of the first crack in all CFST columns tested correlates with an increase in the development length of the longitudinal bars, which in turn increases the characteristic cracking load (P _cr). This trend is clearly demonstrated in Table 3. For instance, increasing the development length from 200 to 300 mm in the precast columns with ECC joints resulted in a 13% increase in the cracking load. Similarly, for columns with UHFRC joints, the cracking force increased by 5% when the development length was increased from 150 to 200 mm, and again when it was increased from 200 to 300 mm.

This observed trend aligns with experimental results documented in the study of Elwood and Eberhard [9] for circular PC columns with HPC interconnect. Among the three types of joints investigated, columns with UHFRC joints exhibited the highest crack load compared to those with NC and ECC joints.

From Table 3, it is clear that increasing the embedment length leads to relatively higher load-bearing capacities. For columns with NC joints, for example, increasing the embedment length from 150 to 200 mm and then to 300 mm led to incremental increases in load-bearing capacities of 5 and 6%, respectively. Similarly, increases of 8 and 6% were observed for columns with ECC joints when the embedment length was increased from 150 to 200 mm and then to 300 mm. For columns with UHFRC joints, increases of 5 and 4% were observed.