Response of RPC-Filled Circular Steel Tube Columns under Monotonic and Cyclic Axial Loading

Results from mechanical tests on thirteen reactive powder concrete(RPC-) filled circular steel tube (RFCT) columns under monotonic and cyclic axial loading are presented in this paper. /e test variables include monotonic and cyclic loadings, confinement coefficient, and diameter of the steel tube. /e test results show that the envelope curves of specimens under cyclic loading were similar to the load-deformation curves of the specimens under monotonic loading. Confinement coefficient had a significant influence on the failure modes of RFCTcolumns.With an increase in confinement coefficient of 0.53 to 0.98, the failure mode transformed from shear failure to compressive failure for specimens under monotonic and cyclic loading. In the elastic stage, no confining effect was provided by the steel tube to the RPC since Poisson’s ratio of steel was larger than the transverse deformation coefficient of RPC. Beyond the elastic stage, the axial compressive strength and ultimate strain of RPC increased significantly due to the confining effect when compared to unconfined RPC. Stress of the steel tube and RPC was investigated by using an elastic-plastic analytical model. Before yielding of the steel tube, stress development in the tube was faster in the longitudinal direction than in the hoop direction. /e results of the experiment indicate that the compressive strength of RPC could be predicted by Mander’s model for confined concrete. Based on Mander’s model, an equation is extended to calculate the axial compressive strength of RFCT columns, and the predicted results are in good agreement with the test results. Based on comparative analysis of 180 RFCTcolumns axial compressive tests, the equation given by EC4 considering the confinement effect can be applied to predict the compressive strength of RFCT columns.


Introduction
Research interests on reactive powder concrete (RPC) has been growing in civil engineering due to its ultrahigh strength, excellent durability, and high toughness [1].For instance, compressive strength of RPC is 100-800 MPa and elasticity modulus is 40 GPa-60 GPa [2][3][4].Application of RPC has increased recently [5,6].However, the main disadvantages of RPC are mainly the lower ratio of ultimate strain to peak strain, which make it more brittle than normal strength concrete (NSC) or high-strength concrete (HSC).Further, due to more compactness, it is more susceptible to fire-induced spalling [7][8][9][10].Recently, an experimental research on fire resistance was performed on seven reinforced RPC columns by our research group.As shown in Figure 1, spalling occurred in all columns with 2% steel fibers.is indicates that the spalling of RPC is a huge challenge, and ways should be explored on making RPC structures safe against fire [9].
Pouring RPC into steel tube is capable of reducing the fire-induced spalling of RPC.Further, RPC-filled circular steel tube (RFCT) columns are expected to possess improved ductility and axial compressive capacity.Moreover, the seismic behavior of RFCT columns is improved when compared to the conventionally reinforced RPC or NSC columns. is is mainly because the confining effect provided by the steel tube enhances the strength and ultimate strain of RPC.erefore, application of RFCT columns is expected to be promising and better structural performance can be achieved [11].Currently, RPC-filled high-strength steel columns were applied in the construction of Techno Station in Tokyo, Japan.As compared with normal strength materials, the dimension of the column was reduced from 800 mm to 500 mm [12].
Understanding the behavior of columns under axial loading is the basic for the structural design [13,14].Researchers have investigated HSC-filled steel tube columns and RFCT columns.Liu et al. [15] investigated the axial compressive behavior of eighteen circular tube columns filled with HSC for which the compressive strength of concrete varied from 60 to 94 MPa. e external diameter of the columns ranged from 133 mm to 140 mm.Axial compressive strength of HSC-filled tube column was higher than that of normal concrete-filled tube column having the same size because the HSC core could carry larger external load when compared to normal concrete core.Qi et al. [16] investigated the behavior of circular tube columns filled with HSC having compressive strength of 76.9 MPa when subjected to axial loading.e external diameter and the thickness of the tube were, respectively, 210 mm and 3 mm.Test results showed that HSC-filled specimens exhibited greater axial load capacity when compared to normal concrete filled tube columns having same volumetric steel ratio.Confinement coefficient had a large influence on the failure mode of HSCfilled columns.
Lin et al. [17] investigated the behavior of short RFCT columns subjected to axial compression.e diameter and length of columns were 133 mm and 400 mm, respectively.Compressive strength of RPC infill varied from 109 MPa to 154 MPa, and thickness of steel tube varied from 3 mm to 12 mm.Experimental results showed that RFCT columns with higher confinement coefficient (confinement coefficient ξ � A s f y /(A c f c )) were more effective to prevent shear failure.Liu et al. [18] investigated RFCT columns subjected to axial compression.It was reported that RFCT columns exhibited lower confinement effect when compared to steel tubes filled with normal concrete.A total of 56 circular and square RFCT short columns under axial load with high steel strength up to 780 MPa and RPC with compressive strength up to 190 MPa were conducted by Xiong et al. [19].Further, the flexural behavior, moment-axial force interaction, and overall buckling resistances of RFCT columns subjected concentric and eccentric compression were investigated [20,21].Results showed that as compared with concentrically loaded columns, the eccentrically loaded columns had better ductility.
While some research has been conducted on the behavior of RFCT columns, no systematic experimental research exists on the behavior of RFCT columns subjected to cyclic axial compressive load.e cyclic loading test can be utilized to measure the unloading and reloading branches of columns [22,23].e response of RFCT columns can be quantified by the unloading and reloading stiffness process.In addition, the influence of confinement coefficient on the failure modes of RFCT columns is still unknown.e differences between RPCfilled steel tubes and NSC/HSC infilled steel tubes are that the compressive strength of RPC is higher than NSC/HSC, but the lateral deformation of RPC is lower than NSC/ HSC.Further, the failure mode of RPC-filled steel tubes is different with NSC/HSC at same confinement coefficient.erefore, it is essential to experimentally investigate the behavior of RFCT stub columns subjected to cyclic axial loading.
To overcome the abovementioned limitations and to extend the understanding on the behavior of RFCT under cyclic loading, this paper experimentally investigates short RFCT columns under monotonic and cyclic axial load with a relatively lager cross section, with the diameter of specimens ranging between 219 mm and 273 mm.A case analysis is done to reveal the effect of different factors, namely, the strength of RPC and steel tube and the diameter-thickness ratio.Finally, a comparative analysis of 180 RFCT columns axial compressive tests is conducted to propose a calculation formula of axial compressive strength of RFCT columns.

Test Specimens.
irteen RFCT columns were tested, out of which six columns were tested under cyclic axial compressive load, and seven columns were tested under monotonic axial load.As shown in Table 1, the columns were classified into two groups, namely, group-c for columns with cyclic loading and group-m for columns with monotonic loading.
e diameter of the seamless steel tube in the specimens was 219 mm, 245 mm, or 273 mm.In order to investigate the mechanical properties of RFCT columns subjected to axial load and to prevent significant flexural deformation, the length-to-diameter ratio was maintained as 3.While this ratio is relatively small when compared to the ratio in actual structural columns, a smaller ratio was chosen to eliminate any likely effect of column buckling in the experiments [24][25][26].e yield strength of tube steel (f y ) was determined by testing standard tensile strength specimens which were cut from each tube.e RPC infill in all tubes had the same mixture proportion and was subjected to identical steam curing regime.e RFCT column specimens were stored at room temperature for 48 hours after pouring  Shock and Vibration and then were cured under steam at 90 °C-95 °C for 72 hours.e strength of RPC was determined by testing 70.7 mm size cube specimens (f cu ), which were cured under the same condition as for RFCT columns.e cylinder strength of concrete was estimated as f c � 0.845f cu [27].e measured elastic modulus (Ec) of RPC was 45 GPa.A summary of the critical test parameters of the test columns is tabulated in Table 1.
(1) In the nomenclature of groups, group-c means specimens were subjected to cyclic compressive load, while group-m means specimens were subjected to monotonic compressive load.(2) In the nomenclature of specimens, for example c-219-8, the first letter "c" represents cyclic axial compression; the second number "219" represents the diameter D of the steel tube; and the third number "8" represents the thickness of the steel tube.
(3) Parameters in this table include the diameter of steel tube D; the thickness of steel tube t; the length of RFCT columns L; the yield strength of steel tube f y ; the 70.7 mm cube compressive strength of RPC f cu ; the cross-sectional area of steel tube A s ; the crosssectional area of RPC A c ; and the confinement coefficient of RFCT columns ξ(ξ � A s f y /A c f c ). (4) N ue is the axial compressive strength from experiment; N u1 is the calculated capacity from equation (8); and N u2 is the calculated capacity from equation (9).

Material Properties of RPC.
e ingredients and mixture ratio of RPC are listed in Table 2. Cement used was ordinary Portland cement (OPC) of strength class 52.5 MPa.Fine quartz sand, with SiO 2 content of more than 99.6%, was used with particle size ranging from 0.18 mm to 0.6 mm.Silica fume was used with SiO 2 content of 95% and particle size ranging from 0.1 μm to 0.2 μm.Moreover, slag with SiO 2 content of 34.90% and high consistency naphthalene water reducer (FDN) were used in the RPC mixture.Brass-coated steel fibers (elastic modulus and tensile strength of steel fibers were 200 GPa and 2250 MPa) with diameter of 0.22 mm and length of 13 mm on an average were embedded to improve the strength and ductility.

Test Setup and Instrumentation.
e RFCT column specimens were tested using a 10,000 kN hydraulic compressive testing machine, as illustrated in Figure 2. e RFCT columns in group-m were subjected to monotonic load.Load control with a load increment of 10% ultimate load was adopted until the load reached 70% of the estimated ultimate load.After that, a displacement control mode was adopted.In order to ensure that specimens were loaded concentrically, specimens were properly checked and leveled before testing.e rate of load control (R L ) was maintained to have a stress rate of 0.06 MPa/s, and the rate of displacement control (R d ) was maintained to have a strain rate of 10 με/s.
Specimens in group-c were tested following a cyclic loading program, as illustrated in Figure 3. Multistage loading was adopted.e increment in each step was 50% of the calculated capacity N u1 , and one loading cycle was applied for each step.After yielding, the loading procedure was changed to displacement control mode.
e incremental displacement in each step was 2Δ y (where Δ y is the displacement corresponding to the calculated capacity N u1 ), and two cycles were applied for each step.In order to ensure that specimens were loaded concentrically, the top surface was leveled by plastering before testing.e rates of load and displacement control for group-c specimens were same as for group-m specimens.
Six linear displacement transducers (LVDTs) were used to measure the axial deformation of the stub column specimens.Out of them, four LVDTs (range: ±25 mm, precision:0.01mm) were used to measure the deformation of the middle third of the stub columns, and two LVDTs (range: ±250 mm, precision:0.01mm) were used to measure the total deformation.Four longitudinal and four hoop strain gauges were placed vertically at the middle part of each column.Figure 2 illustrates the details of LVDTs and strain gauges.

Test Results of RFCT Columns
3.1.Failure Mode of RFCT Columns.Generally, all specimens in group-c showed similar failure process.e confinement coefficient had a significant influence on the failure modes of RFCT stub columns.For the confinement coefficient in the range of 0.53 to 0.98, the failure mode of specimens under cyclic loading was shear failure.e shape of steel tube was not changed when the load was under elastic range.RPC behaved as a homogeneous elastic material which has a small coefficient of lateral deformation.Figure 4 shows the failure mode of RFCT columns in group-c.It was observed that stub columns with the lower confinement coefficient (ξ < 0.8) exhibited severe share failure.On the shear slip plane, steel fibers were pulled out and the RPC was crushed.e shear failure in other stub columns (ξ ≥ 0.8) was relatively slight as shown in Figures 4(c) and 4(f ). is was because steel tube provided relatively stronger constraint in those columns.With an increase in confinement coefficient, the failure mode transformed from shear failure to compressive failure for specimens under monotonic load.
Specimens in group-m showed different failure modes when compared to group-c specimens.e failure mode of specimens with lower confinement coefficient (ξ < 0.8) was shear failure.However, when the confinement coefficient exceeded 0.9 (ξ ≥ 0.8), the specimens exhibited local buckling failure.It is concluded that with an increase in confinement coefficient, the restraining effect of steel tube increased which contributed in reducing shear failure. is observation indicates that if the confinement coefficient of stub columns is large enough, shear failure can be completely avoided.e confinement stress affecting the failure modes is mainly because the core RPC is under triaxial compressive in RPCT columns; with the increase of compressive stress, the hoop stress increases.en, the shear failure mode occurred at the section where shear capacity is lower [28].As compared with normal strength of concretefilled steel tubes, the shear failure is more susceptible to occur in high-strength concrete and RPC-filled steel tubes.However, if the confinement coefficient is higher enough, the hoop stress provided by the steel tube is higher, and the failure mode changed from shear failure to compressive failure.
e shear angle (β) of RPC core was measured and found to lie between 55 °and 70 °.As in the case of circular steel tube filled with normal strength concrete (NSC), shear angle β of RPC can be calculated by Mohr-Coulomb theory as where ϕ is the internal friction angle of (RPC) concrete.β of ordinary concrete prism is β � 58 ∘ ∼64 ∘ , and β of HSC-filled tube columns is β � 57 ∘ ∼65 ∘ [28].e test results demonstrated that the shear angle of RFCT columns satisfied Mohr-Coulomb theory.
e influence of shear angle of RPC on the strength of RFCT columns needs further research.Note.Water/binder ratio is the weight of water to the combined weight of cement, silica fume, and slag. 4 Shock and Vibration

Load-Axial Strain Relationship of RFCT Columns.
Load-axial strain relationships of RFCT columns are shown in Figure 5, where ε � Δ/L (Δ is the total axial deformation monitored by LVDTs and L is the length of the stub column specimens).All stub columns exhibited excellent ductility.
For the specimens under cyclic axial load, the loadingunloading curves basically coincided confirming that the unloading process was elastic.Furthermore, as compared with the specimens under monotonic loading with the same cross section, the envelope curves of specimens under cyclic loading showed a good agreement with the loaddeformation curves of the specimens under monotonic loading.With the increase of loading cycles, the deformation of specimens is increased.At the beginning of unloading stage, the axial load dropped sharply with little deformation restored.When unloaded completely, the residual deformation existed and accumulated with the development of plastic strain.Besides, the unloading stiffness and reloading stiffness remains similar as the loading stiffness.Figure 6 compares the envelope curves and unitary envelope curves of RFCT columns, where N μ is the peak load and ε μ is the strain corresponding to the peak load.For a given cross section and for a given loading program, the axial compressive strength of RFCT columns increased with an increase in the confinement coefficient and with a decrease in the diameter-to-thickness ratio.After the peak load, the compressive strength of specimens in group-c decreased gradually until the compressive strength reduced to 80%-90% of the peak load.
en, the bearing capacity of RFCT columns increased again, though slowly, exhibiting the strain-hardening phenomenon in the RFCT columns.

Strain Analysis of Steel Tube.
e envelope curves of load-strain relationship and the variation of the ratio of hoop strain and longitudinal strain are shown in Figure 7, where ε hc and ε vc are, respectively, the hoop strain and longitudinal stress at the midheight of steel tube in group-c and ε hm and ε vm are, respectively, the hoop strain and longitudinal stress of steel tube in group-m.e yield point of the RFCT columns is shown in Figure 7, and the specimens yielded before the peak load point.von Mises yield criterion was adopted to define the yield point of the stub specimens as where σ v and σ h are the longitudinal and hoop stresses of the steel tube and can be obtained by the following equations [15]:   Shock and Vibration -16000 -12000 -8000 -4000 0 4000 8000 12000 16000 -16000-12000 -8000 -4000 0 4000 8000 12000 16000  Shock and Vibration where E s and v s are the elastic modulus and Poisson's ratio of RFCT columns and ε v and ε h are the longitudinal and hoop strains of the steel tube, respectively.In the experiment, the longitudinal strain developed faster than the hoop strain in the elastic stage.e ε h /ε v ratio was in the range of 0.28 to 0.36 and was approximately equal to Poisson's ratio of steel. is means steel tubes did not provide con ning e ect to the RPC in the elastic stage.However, with an increase in the axial stress, the longitudinal strain was exceeded by the hoop strain because of the rapid lateral expansion of the RPC, and the value of ε h /ε v increased as well.
e con ning force was e ective only after the elastic stage.e lateral deformation coe cient (υ) of concrete varied with compressive strength as shown in Figure 8, where σ and f c are the external load and compressive strength of concrete, respectively.e lateral deformation coe cients varied with concrete strength as listed in Table 3. Results showed that the lateral deformation coe cient increased with an increase in the load level (σ varied from 0.2f c to 0.9f c ) [29,30].It was observed that when σ 0.2f c and σ 0.6f c , concrete was in the elastic stage; thus, three types concrete (NSC, HSC, and RPC) had similar coe cient, with the mean value of the coe cient as υ 0.204.However, when load level reached to 0.8f c , microcracks were formed and propagated in concrete.Since most cracks were parallel to the stress direction, the width and number of microcracks caused the evolution of lateral deformation.e coe cient υ increased the most for NSC, but the coe cient increased the least for RPC. is phenomenon became more obvious when σ f c .As shown in Figure 8(d), NSC had a large lateral deformation, and the coe cient was up to 0.57.As compared to NSC, the lateral deformation of HSC and RPC increased slowly.
e coe cient of HSC ranged from 0.26-0.36,and the coe cient of RPC was between 0.25 and 0.31. is is mainly because the development of microcracks in RPC is lower than that in HSC and NSC at the same load level [31]. is suggests that concrete with higher compressive strength has a lower lateral deformation coe cient.
e core concrete with a large lateral deformation coe cient can obtain more con ning e ect, and in that sense, con nement e ect of steel tube is not as pronounced in RPC in ll as with NSC in ll.

Stress-Strain Analysis of RFCT Columns.
Based on the elastic-plastic analytical method [15,28], a stress-strain analysis was conducted for RFCT column specimens.Plane stress-strain relationship of steel tube was adopted to evaluate the stress level of steel tube and RPC under axial load.Figure 9 illustrates the stress-strain analysis results of RFCT columns, including the envelope curves of load vs. stress in steel tube and the envelope curves of stress vs. strain in RPC.In the gure, σ vc , σ hc , and σ zm are, respectively, the longitudinal, hoop, and equivalent stresses of group-c specimens, and σ vm , σ hm , and σ zm are, respectively, the longitudinal, hoop, and equivalent stresses of group-m.Similarly, σ cc and σ cm are compressive stresses in RPC in group-c and group-m specimens, and ε is the strain of the stub columns.e equivalent stress σ z can be calculated by the following equation: Stress results show that steel tube yielded before the stub columns reached the peak pressure.Hoop stress grew slowly and remained at a low level, which implies that the yielding of

Shock and Vibration
steel tube was essentially caused by the growth of the longitudinal stress.After the yield point, the hoop stress gradually increased, while there was a decreasing trend for the longitudinal stress.is is because the transverse deformation of RPC increased sharply during plastic stage and the conning stress provided by the steel tube increased.erefore, HSC (Teng,[33]) RPC (Wang,[34]) RPC (Wu,[35]) RPC (Zeng,[36]) RPC (Ke, [30])  (Tian,[32]) NSC (Stemberk,[29]) HSC (Teng,[33]) RPC (Ke, [30]) [32]) NSC (Stemberk,[29]) HSC (Teng,[33]) RPC (Ke, [30])    12 Shock and Vibration the tube was stressed more in the transverse direction than in the axial direction, and the steel tube began to buckle locally.Given the same diameter of columns, the con nement coe cient increases with the decrease of diameter-thickness ratio, and the hoop stresses are increased.e peak stress and the strain at peak stress of RPC increased when surrounded by the steel tube.ere was a slightly descending segment for core RPC after the axial compressive strength exceeded the peak stress.It is concluded that if RPC is con ned by a steel tube, both the ultimate bearing capacity and the ductility will be improved even though di erent loading types showed varying degrees of improvement.e plastic stress is converted from the measured strains; the equation of measured strains and plastic stress was derived from Reference [37].As shown in Figure 9, compressive strength, as well as the ultimate strain, of con ned RPC under cyclic axial load was similar as that under monotonic axial load.

Compressive Strength of Con ned RPC.
Previous researchers [38] proposed a reasonable stress-strain model to analyze con ned concrete.is paper used Mander's model [38] to predict the compressive strength of con ned RPC of the RFCT columns as where f r is the con nement stress provided by the steel tube to the RPC and can be calculated by the following equation: where σ h is the hoop stress of the steel tube and should be calculated based on the longitudinal stress σ v and the von Mises yield criteria. Figure 10 compares the compressive strength of con ned RPC obtained from the experiment (f cc,e ) and equation ( 5) (f cc,c ).e results indicate that compressive strength of con ned RPC can be predicted by Mander's model.

Axial Compressive Strength of RFCT Columns
Compressive strength of RFCT columns in this experimental study was compared with the design strength as proposed by Eurocode 4(2012) [39].e EC4 equations for compressive strength of concrete-lled steel tube are as follows, in which equation ( 7) neglects the con ning e ect, while equation ( 8) considers the con ning e ect with the relative slenderness λ ≤ 0.5: where A s and A c are the cross-sectional areas of the steel tube and RPC and t and D are the thickness and diameter of the steel tube.η a and η c are given by the following expressions: where λ is the relative slenderness and is calculated as N cr is the elastic critical normal force and is calculated as Shock and Vibration 13 where where E s and E c are the elastic moduli of steel and RPC and I s and I c are the second moments of area of the steel and RPC sections.e results were compared in Table 1, which indicates that equation ( 7) is conservative to estimate compressive strength of RFCT columns in group-c and group-m.Equation (8) shows good agreement with RFCT columns under monotonic and cyclic loading.Furthermore, two di erent loading types (cyclic and monotonic) showed similar rules.e specimen with the minimum diameterthickness ratio (D/t 18.3) exhibited the maximum value of N ue /N u ratio (N ue /N u1 , N ue /N u2 ).For a given diameter of RFCT columns, the value of N ue /N u ratio increased with a decrease in diameter-thickness ratio, except for specimens with D 273 mm.When the diameter-thickness ratio is less than 25, it is e ective to increase the compressive strength of RFCT columns by reducing the diameter-thickness ratio.
e comparisons between axial compressive strength calculated by equations ( 7) and ( 8) and the 167 test results [17][18][19][40][41][42][43][44][45] as well as 13 test results in Table 1 are shown in Figures 11(a) and 11(b), respectively.e average of N ue /N u1 is 1.19 and standard deviation is 0.116, while the average of N ue / N u2 is 1.07 and standard deviation is 0.090.Results show that equation (7) can provide conservative calculation by 20% because the con nement e ect is not considered.While equation (8) considering the con nement e ect shows better agreement with test results, it is recommended that equation ( 8) can be applied to predict the compressive strength of RFCT columns.

Conclusions
is paper presented an experimental research of RFCT columns under monotonic and cyclic axial compressive loading.
An equation was proposed to predict the ultimate axial compressive strength of RFCT columns.e following are the main conclusions drawn from this experimental investigation: (1) Load-strain curves of specimens under cyclic loading were similar to the load-deformation curves of the specimens under monotonic load.(2) Con nement coe cient had a signi cant in uence on the failure modes and strength of RFCT stub columns.
(3) ere was a relatively small hoop stress in the steel tube before yielding.After attaining the peak load, RPC core in RFCT columns showed a slightly descending segment in the stress-strain curve.Mander's model could predict the axial compressive strength of confined RPC.(4) If the confinement effect is not considered, the EC4 method can provide conservative calculation by 20%.
It is recommended that equation (9) given by EC4 considering confinement effect can be applied to predict the compressive strength of RFCT columns.Longitudinal strain of RFCT column σ h :

Notations
Hoop stress of the steel σ v : Longitudinal stress of the steel σ z : Equivalent stress of steel tube α: Strength incremental coefficient λ: Relative slenderness.

Figure 1 :
Figure 1: Spalling of RPC columns in fire.

Figure 2 :Figure 3 :
Figure 2: Test setup and instrumentation of stub columns.

Figure 10 :
Figure 10: Comparison between test results and Mander's model.

Figure 11 :
Figure 11: Comparison of test results with design strength based on Eurocode 4. (a) Comparison of test results with prediction from equation (7).(b) Comparison of test results with prediction from equation (8).

Table 3 :
e lateral deformation coe cient of concrete with di erent strengths.
Note. f c is the compressive strength of concrete.