Experimental Study on Ultimate Strength of Steel Tube Column Filled with Reactive Powder Concrete

Composite concrete Filled Tubular Steel (CFT) members, which have excellent deformability due to the well-known confined and constrained interaction between steel tube and concrete, have largely been utilized as bridge piers or columns in high-rise buildings, resulting in increased strength and decreased column size. This study examined the experimental performance of steel tube columns filled with reactive powder concrete (RPC) under axial compression. Three sets of columns were used in the experiment, each with variations in shape (square, rectangular, and circular), length-to-diameter ratio, and compressive strength of the RPC. The first set consisted of five columns, while the second and third sets each had seven columns with three different lengths (750 mm, 600 mm, and 450 mm) and two different compressive strengths (54 and 92 MPa). A new numerical model was developed to calculate the ultimate failure load of the columns by considering factors such as the yield strength of steel, the compressive strength of concrete, the column shape, and the ratio of concrete to steel. This model was validated by comparing the results obtained from the experiments to those predicted by the model, as well as by designing equations from various codes. The results showed that the proposed numerical model accurately predicted the ultimate failure load for columns filled with different types of concrete, especially for RPC, while maintaining conservatism compared to the ACI, AISC, and EN codes equations.


Introduction
Concrete-filled steel tubes (CFST) have grown in popularity in seismic zones in recent years. The composite effect associated with these members, which is caused by the interaction between the steel tube and the concrete core, can improve both the ductility and capacity of the member [1]. Many researchers have studied the behavior of steel tube columns filled with normal concrete, such as Stephen et al. [2], Uenaka et al. [3], Yang & Han [4], Liao et al. [5], and Almamoori et al. [6]. However, it appears that little attention has gone into studying the behavior of a composite member subjected to axial compression with high or ultra-high-strength concrete. Experiments were carried out [7,8] on concentrically and eccentrically loaded circular, square, and rectangular CFST stub columns filled by high-strength selfcompacting concrete with a length-to-diameter (width) ratio of 3. The test findings revealed that there were substantial variations between partially loaded CFST stub columns and the corresponding fully loaded composite columns due to the presence of partially compressive forces. In recent years, structural engineers have been interested in the performance of CFST under axial compression loads with different concrete types. However, design codes for composite columns have not adequately addressed this issue. Moreover, the design formula was proposed by Uenaka et al. [3] based on the yield strengths of the tubes and the filled concrete cylinder. In recent years, a novel group of concrete known as reactive powder concrete (RPC) has recently gotten a lot of interest all over the world due to its superior mechanical properties, such as high strength, high ductility, high durability, low shrinkage, and high resistance to corrosion and abrasion. Many researchers have been studying the properties of RPC, such as Hassooni & Al-Zaidee [9] and Muteb & Hasan [10]. The authors showed that the increase in steel fiber improves the compressive and rupture stresses of the developed concrete. Other researchers studied the effect of RPC or UHPC on the behavior of composite sections, such as Luo et al. [11], Hoang et al. [12], and Mi et al. [13]. Based on their results, the strength and hoop coefficient of the developed column increased due to the growth in the ultimate strength of the RPC core, which highlighted an improvement in the ductility of developed columns. Limit studies on the behavior of steel tube columns filled with RPC under axial compression utilizing various cross-sections can be seen in the above preview. As a result, the focus of this work is on utilizing RPC to experimentally and analytically investigate the CFST's performance.

Research Significance
The purpose of the current study was to experimentally investigate the effect of using RPC on the behavior of CFST under axial compression loads. This is done by changing the column shape, the compressive strength of the concrete, and the length-to-diameter ratio. In addition to the experimental work, a numerical model has been developed to compute the failure load of the developed CFST columns.

General Description
The behavior of CFST using RPC while exposed to concentric axial compression stress was investigated using nineteen steel columns in this study. High-quality steel with a yield stress of 363 MPa was employed in this research. The details of the test columns are shown in Table 1 and Figure 1.

Materials
A CRESTA 42.5R cement was used in this study. This type of Portland Limestone cement agrees to ASTM C150/C150M-22 [14]. Locally available fine sand with a maximum particle size of 600μm (0.6mm) was used to produce the RPC. Fine aggregate testing was carried out in accordance with [15]. Master Glenium 51's commercial brands of additive (superplasticizer) were used. It's a non-hazardous third-generation superplasticizer produced by BASF-The Chemical Company. The steel fibers used in this study were straight steel wire, which has an aspect ratio (length to diameter) of about 59. ConMix Company's microsilica fume was employed as a mineral additive in the combination. Based on earlier research done by Muteb and Hasan [10], the mixes were employed to achieve maximum compressive strength after trial mixes. The details of the mixes are shown in Table 2. Following that, all specimens were cast and cured inside curing tanks at 60 C o for twenty-eight days.

Instruments and Procedure
After 28 days, all of the columns were removed from the curing water tank, dried, cleaned, and painted. Typical testing equipment with a 2000 kN capacity was used to provide axial compression force on the columns supported by a stiff, thick plate (20 mm). The plate is used to transfer the axial load from the machine to the column. Three Linear Variation Displacement Transducers (LVDT) sensors with a 0.001 precision were used. The first one records the axial deformation, while the second measures the lateral deformation of the developed CSFT columns, as shown in Figure 2.
When each tested column's resistance decreased (the loading curve levelled), the loading was terminated. Figure 3 illustrates the failure modes of the tested CFST columns. At first, all columns were in the elastic stage, then moved to the elastic-plastic phase. Local yielding began to appear at the ends of each column as the axial load increased due to stress concentration. All columns collapsed with local buckling at the ends during the last phases of loading; these results were comparable to those found by Luo et al. [11]. In general, the RPC that filled steel tube columns increased the compressive strength, delaying the failure of all columns. Also, the circular shape is more resistant to failure compared to square and rectangular ones. The results also showed that the length-to-diameter ratio was less effective compared to other variables.

Load-Deflection Curves
With various weights, the vertical displacement of the tested columns was recorded and drawn. The effects of three factors on load-deflection behavior were investigated as follows:

Length to Diameter Ratio Effect
The effects of the length-to-diameter ratio on the behavior of load-vertical displacement are shown in Figure 4 and Table 3. It can be seen that there is a significant improvement in the behavior of all filled columns compared with reference columns (without filling using RPC). The column filled with RPC (compressive strength of 92 MPa) is stiffer in behavior compared to a column filled with concrete (compressive strength of 54 MPa). This is due to the stiffer behavior of RPC containing microsteel fiber. And it can be observed that the presence of steel fiber makes the behavior of the curve stiffer for shorter columns and the stage after ultimate load more ductile for intermediate circular columns, which corresponds with the results of Hoang et al. [12]. The failure load for all shapes increased as the compressive strength of concrete increased. However, the failure load for the square and rectangular columns is less than the failure load for the circular columns when using the same concrete.

Compressive Strength Effect
The influence of compressive strength on vertical displacement behavior is seen in Figure 5. It can be noticed that the compressive strength has a significant effect on the behavior of each column. The columns filled with 92 MPa compressive strength concrete performed better than the columns filled with 54 MPa concrete by 62.5%, 54%, and 16.5% for the square, rectangular, and circular columns, respectively. From that, it can be concluded that the columns with higher compressive strength exhibit stiffer behavior; these results correspond to those found by Hoang et al. [12].

Lateral Displacement
The relationships between the axial load and lateral displacements at mid-height for all columns are shown in Figures  6 to 8. Small values of lateral displacements in all specimens are noticed and the lateral displacements in the x-direction are smaller than the displacements in the y-direction. This was referred to as the position of local-buckling occurred far from the place of the LVDT, which measured this displacement. In most situations, the columns behaved similarly in terms of lateral displacement in the x and y directions. Besides, the compressive strength of the concrete used as well as the column shape have an insignificant effect on the lateral displacement at mid-height of each column as clearly noticed in Figure 9 for square, rectangular and circular shape.

New Model Suggestion
Many equations such as the equations found in ANSI/AISC 360-22 [16], Eurocode 4 [17], and ACI 318 [18] codes are used to determine the strength of concrete-filled columns. Besides, many researchers suggested equations to calculate the compressive strength of concrete-filled columns like [3]. The common questions used to calculate the strength of the circular concrete-filled columns are listed as follows: The model developed in this research to predict the compressive load of concrete-filled tubular steel column is as follows: where is the ultimate axial load, is the shape factor, 1 2 are variables coefficients. Regression analysis was done to estimate the impact of each parameter on the performance of the developed model with a coefficient of determination (R 2 ) equal to 0.982 and the resulted values are:  Table 4 shows the difference between the experimental and theoretical results for all columns. It can be seen that the results of failure load for EN code are higher than the results of AISC and ACI code. Comparing the results of the current research with the results from the aforementioned codes, this study's proposed equation offers a better prediction of the compressive strength than the prediction offered by the codes equations when used to predict the compressive strength for square and circular cross-section. However, this study equation offers a less accurate prediction of the compressive strength for rectangular cross-sections columns compared to different codes equations. It can be concluded from these results that the effect of the cross-section of the column on its performance was clear. The suggested equation, based on R 2 , was able to predict 98% of the change in compressive strength depending on the factors employed.  Figure 10-a shows the normal probability plot of the residuals which verifies the assumption of the normal distribution of the residuals. The normal probability plot of the residuals generally follows a straight line. There is no evidence of abnormality, outliers, or unidentified variables. The residuals versus fits plot are used to confirm that the residuals are distributed randomly and have a constant variance. Ideally, the points would fall randomly on both sides of zero, with no discernible patterns. The data appears to be scattered randomly about zero. There is no indication that the residual value is influenced by the fitted values as seen in Figure 10-b. The residuals are not related to one another, as seen in Figure (10-d). When displayed in time order, independent residuals show no trends or patterns. Patterns in the points might suggest that residuals that are near each other are associated and hence are not independent. The residuals should appear to fall randomly around the centreline.

Conclusion
This study has investigated the performance of steel tube columns filled with reactive powder concrete (RPC) under axial compression with different cross sections. The collected test data from nineteen specimens was analyzed and compared with previous studies and design codes. The following conclusions can be drawn from the present study.
Due to the high compressive strength and presence of steel fiber, the composite effect caused by the interaction between the steel tube and the concrete core can improve both the ductility and capacity of the member. On the other hand, the failure mode of the columns filled with RPC differs from that of the columns filled with normal concrete. As a result, local buckling and local yielding appeared at the ends of the column due to some stress concentration at these locations. The behavior of the columns in terms of load-vertical displacement is stiffer with the increase in compressive strength due to the use of RPC with micro-steel fiber. The maximum ratios of increase in displacement resistance with increase in compressive strength were 62.5%, 54%, and 16.5% for square, rectangular, and circular columns, respectively. Also, with respect to the effect of shape, circular columns have a better vertical displacement compared with other shapes, as the displacement resistance increased by 28.5% and 20% for rectangular and square columns, respectively. While the lateral displacement was unaffected by the compressive strength or shape of the columns, this is attributed to the confinement effect. In addition to these main findings in the part of experimental results, there are some concluded remarks about the theoretically proposed equation, which, despite its limitations to pinned boundary conditions and concentrated axial load. Furthermore, it provided an accurate prediction while remaining conservative compared with the underestimated results offered by the ACI, AISC, and EN codes equations. In details, the enhancement of concrete capacity within the section represented by the parameter k2 converges the results of the proposed equation with the values obtained from the experimental test, and the shape factor β is very sensitive and affects the ultimate capacity of the CFST column.

Data Availability Statement
The data presented in this study are available in the article.