Behaviour and design of hollow and concrete-filled spiral welded steel tube columns subjected to axial compression

Journal of Constructional Steel Research 128 (2017) 261–288

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Behaviour and design of hollow and concrete-filled spiral welded steel tube columns subjected to axial compression Farhad Aslani a,b,⁎, Brian Uy a,c, James Hur a, Paolo Carino a a b c

Centre for Infrastructure Engineering and Safety, School of Civil and Environmental Engineering, The University of New South Wales, Sydney, NSW 2052, Australia School of Civil, Environmental and Mining Engineering, The University of Western Australia, Crawley, WA 6009, Australia School of Civil Engineering, The University of Sydney, Sydney NSW 2006, Australia

a r t i c l e

i n f o

Article history: Received 25 July 2016 Received in revised form 26 August 2016 Accepted 29 August 2016 Available online xxxx Keywords: Spiral welded tube Longitudinal welded tube Finite element model Concrete-filled steel tube columns

a b s t r a c t Spiral welded tube (SWT) structures have found worldwide application in pipeline construction, wind turbine towers, foundation piles, and columns in tall buildings. However, the understanding of their fundamental behaviour is still insufficient and efficient analysis and design methods have not been precisely developed owing to the lack of experimental and numerical research on these types of structures. A distinct advantage of SWT is their streamlined manufacturing process, so that today large diameter SWT can be economically produced. Due to the application of SWT as structural members being relatively new, this paper presents an investigation into the behaviour of hollow and concrete-filled steel SWT columns when subjected to axial compressive loading. Parameters of particular interest affecting the strength and failure modes include the weld's spiral geometry and initial imperfections from the production process. To evaluate the behaviour of SWT columns, an accurately developed finite element model (FEM) which incorporates the effects of initial local imperfections and residual stresses using the commercial finite element program ABAQUS has been prepared. The FEM buckling behaviour of SWT is compared with that of longitudinally welded tubes (LWTs). Experimental laboratory testing is carried out on twenty columns under displacement-controlled loading conditions in order to calibrate and verify the accuracy of the model results. Furthermore, a design model is proposed for circular concrete-filled steel tube columns. In addition, comparisons with the prediction of axial load capacity using the proposed design model, Australian Standards, Eurocode, and American Institute of Steel Construction code provisions for hollow and concrete-filled SWT and LWT columns is also carried out. © 2016 Elsevier Ltd. All rights reserved.

1. Introduction Steel tubes are widely used in many industrial applications, and are usually distinguished by their method of production. They can be produced either seamless (with diameter from 21 to 406 mm) or with seam method by longitudinal welding (with diameter from 10 to 1630 mm) or spiral welding (with diameter from 160 to 3000 mm) from rolled strip or thick plate [1]. Welded tubes are produced by bending metal strips (skelp) or plates into the form of a tube by roll forming and welding the seam by various welding processes. Currently around two thirds of the steel tubes produced in the world are accounted for by welding processes. There are two types of welded tubes, longitudinally welded tube (LWT) and spirally welded tube (SWT) [2]. LWTs are manufactured from steel plate with only one weld seam joining the two edges of the rolled plate. SWTs are manufactured by helical rolling of the steel coils. In contrast to LWT production where each ⁎ Corresponding author at: School of Civil, Environmental and Mining Engineering, The University of Western Australia, Crawley, WA 6009, Australia. E-mail address: [email protected] (F. Aslani).

http://dx.doi.org/10.1016/j.jcsr.2016.08.023 0143-974X/© 2016 Elsevier Ltd. All rights reserved.

tube diameter requires a certain plate width, spiral tube production is characterised by the fact that various tube diameters can be manufactured from a single strip or plate width [3]. Total global SWT and LWT production distribution was 37% and 63% in 2007, respectively, as shown in Fig. 1 [4]. The most common structural application of a cylindrical shell with helical features is the SWT, first used at the end of the 19th century in water transmission pipelines [5–6]. SWTs were originally manufactured by riveting together properly bent plates until progress in welding technology allowed for the efficient tandem arc welding process [5] (see Fig. 2). The process of producing SWT has been progressively improved, so that today large diameter SWTs can be economically produced. These tubes are used for pipeline construction, wind turbine towers, foundation piles, load-bearing members in combined walls, and columns in tall buildings [2,7]. SWTs provide significant benefits over traditional longitudinal and butt-welded tubes, for the following reasons: (a) SWTs are a cost-effective solution compared with other manufacturing processes, (b) SWTs can be manufactured in 30 m lengths with outside diameters from 160 to 3000 mm and wall thicknesses ranging from 2 to 30 mm [8],

262

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

capacity by using the Australian Standards, Eurocodes, American Institute of Steel Construction code provisions, and the proposed design model for hollow and concrete-filled SWT and LWT columns is also carried out. 2. Experiments This section outlines the test program undertaken which consists of hollow and concrete-filled SWT and LWT column tests and associated material property tests. The test set-up for the columns will be described and the results will then be presented. A general review and description of the failure modes will then be presented.

Fig. 1. SWTs and LWTs consumption [4].

and (c) continuous or very long tubular members may be constructed efficiently both in a factory and onsite from compact coils of metal strip, reducing the need for costly transport of long structural members [9]. SWTs can be used as structural columns in buildings. These SWT columns will be more effective especially if they are filled with concrete. Concrete filled steel tubes (CFSTs) are stiff and strong in axial compression, and have substantial bending resistance which has been proven in previous studies [10]. The combined action between the steel tube and the concrete infill provides a higher resistance in tension (steel tube), compression (concrete filled), and reduces local and global instabilities. These properties make them ideal for bridge piers, foundation caissons and piles, as well as columns in multi-storey buildings subjected to both gravity and extreme loadings, such as earthquake, impact, and blast [11]. In this study, SWTs have been used in CFST columns and the research has been conducted with an eye towards characterising the engineering properties of SWTs and resulting design recommendations. Available code provisions do not currently address the design of CFST incorporating SWTs and this study attempts to remedy this situation. Moreover, a finite element model (FEM) which includes the effects of initial local imperfections and residual stresses using the commercial program ABAQUS has been implemented to evaluate the behaviour of SWT and LWT columns. Experimental laboratory tests are carried out on ten hollow SWT and LWT columns and ten concrete-filled SWT and LWT columns under axial compression loading conditions in order to calibrate and verify the accuracy of the model results. Moreover, a design model incorporating a concrete confining pressure approach is proposed for predicting the ultimate axial strength of circular CFST columns. Furthermore, comparisons with the prediction of axial load

Hot-rolled material coil Extensive rolling Skelp end welding Forming rolls

Flying cut-off

Inside & outside tandem welding

Fig. 2. Schematic illustration of the SWT forming process.

2.1. Material properties 2.1.1. Tensile coupon tests One of the crucial concerns for the SWT is the possible influence of material anisotropy on the bending resistance of the resulting fabricated tube. The rolling process used to form the SWT causes the steel to develop anisotropic characteristics [12]. The longitudinal yield strength is generally lower than in the transverse direction of the steel. For LWT, the orientation of anisotropy remains the same as that of the steel plate from which it is formed. As a result, the highest yield strength occurs in the direction around the circumference of the cylinder, which is termed the hoop strength. This also means that LWTs are weaker along the longitudinal axis. However, this situation is not the same for SWT. The longitudinal strength follows the spiral with the transverse yield strength perpendicular the weld. Thus, the axial direction along the length of the cylinder is stronger compared with that of a LWT. On the basis of material anisotropy directions, SWT has the potential to outperform LWT when subject to axial compressive loading. Heiberg et al. [12] presented findings of bending stiffness being greater in SWT than LWT, and proposed that this might also be attributed to SWTs having a higher Young's modulus in the axial direction. Also, Sadowski et al. [13] have conducted tensile tests on four SWTs with outer diameters of 820 mm (two had a nominal thickness of 8 mm and two of 11 mm) at four different orientations parallel to the tube axis, transverse to the tube axis, parallel to the strip axis and transverse to the strip axis. They have concluded that SWTs exhibit no significant differences for different orientations, and therefore the stress-strain characteristics of a SWT may thus be treated as isotropic. In this study, two tensile coupon test scenarios have been considered. To determine the stress-strain characteristics of the SWT and LWT steel plate in tension, six 400 mm tensile coupons were produced from the virgin SWT and LWT steel plate and six 400 mm tensile coupons were extracted from SWT in three different orientations relative to the tube axes and the spiral directions (Parallel to the tube axis (denoted V), transverse to the tube axis (S), parallel to the weld axis (P)) and tested in an Instron uniaxial testing machine. Pertinent data for these test coupons is provided in Table 1. Six tests were conducted for virgin SWT and LWT steel plates with a mean value of yield stress of 288 N/mm2 and 277 N/mm2 being established, respectively. The tests revealed an increase in stress after yielding and the mean ultimate stress of the SWT and LWT steel plate in tension was determined to be 298 N/mm2 and 308 N/mm2, respectively. Stress-strain diagrams are provided in Fig. 3 and the failure mode of the tensile coupons is illustrated in Fig. 4. The ductility can also be observed both by the pronounced necking of the specimens, which resulted in ultimate strains in excess of 1441 and 1440 microstrain (με) for SWT and LWT, respectively. Another six tests which were conducted for six 400 mm tensile coupons were extracted from SWT and the results show that there is no significant difference between the yield stress for the V, P, and S orientations. Furthermore, these results concur with the conclusions of Heiberg et al. [12] and Sadowski et al. [13].

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

263

Table 1 Tensile coupon test results for SWT and LWT. Tensile coupon test results - virgin SWT and LWT plates Specimen Yield stress, fy (N/mm2)

Ultimate stress, σu Young's modulus, (N/mm2) Es (N/mm2)

Yield strain (με)

SWT-1 SWT-2 SWT-3 Mean SD

291 287 286 288 2.41

302 293 298 298 4.41

199,651 196,814 203,104 199,856 3149.98

1455 1458 1408 1441 28.15

LWT-1 LWT-2 LWT-3 Mean SD

259 291 281 277 16.10

297 316 312 308 9.97

178,316 205,050 193,809 192,392 13,423.30

1453 1418 1447 1440 19.01

Tensile coupon test results - SWT plates extracted from tube Specimen Yield stress, fy (N/mm2)

Ultimate stress, σu Young's modulus, (N/mm2) Es (N/mm2)

Yield strain (με)

SWT-P1a SWT-P2 Mean SD

270 272 271 1.37

318 321 319 2.27

151,407 180,045 165,726 20,250.08

1787 1513 1650 193.34

SWT-V1 SWT-V2 Mean SD

272 274 273 2.04

320 314 317 4.27

155,793 152,451 154,122 2363.35

1743 1801 1772 40.38

SWT-S1 SWT-S2 Mean SD

265 261 263 3.22

312 316 314 3.07

147,365 125,354 136,359 15,563.51

1801 2081 1941 197.95

a Parallel to the tube axis (denoted V), transverse to the tube axis (S), parallel to the weld axis (P).

2.1.2. Concrete cylinder tests Using self-compacting concrete (SCC) produces several benefits over conventional concrete and it is especially useful for concrete-filled steel tubular structures [14]. Hence, SCC has been used in this study. The SCC mix was provided with a compressive strength at 28 days of approximately 40 MPa. Provided SCC has been satisfied, all fresh property tests requirements. SCC cylinders with 100 mm × 200 mm dimensions were cast and tested to allow the characteristic compressive strength of the concrete to be determined. In total, nine cylinders were tested at ages 7, 28, and 65 days (i.e. testing age of columns) and results are summarised in Table 2. The mean compressive strengths of the columns at the time of testing were estimated at 50 N/mm2. The stress-strain behaviour of the specimens under compressive load has been experimentally evaluated by conducting uniaxial compressive load tests. The axial and circumferential strains were recorded by means of extensometers/strain gauges mounted on the specimen. The stress-strain curves of SCC at age 28 days have been captured, as shown in Fig. 5.

2.2. Column tests This section of this study involved a general description of the hollow and concrete-filled SWT and LWT column experimental test programme. SWT and LWT columns were welded using the MIG process with a welding current of 160 A and a voltage of 25 V. The welding electrode size was 1.2 in diameter. The shielding gas used in this experimental work was Argoshield Light having 3% oxygen + 5% carbon dioxide + 92% argon with a flow rate of 15–17 l per minute. The wire feed rate was automatically adjusted by the welding machine based on the current and voltage selected. All dimensions and parameters were precisely measured for the purposes of producing an accurate representation of initial local imperfections, finite element models, and

Fig. 3. Tensile coupon tests for (a) virgin SWT plates, (b) virgin LWT plates, and (c) SWT plates extracted from tube.

undertaking theoretical calculations. The measured parameters included length of the tube (L), outer most diameter at three sections along the tube (D1 - top, D2 - middle, and D3 - bottom), thickness of the weld (tw), width of the weld (dweld), and in the case of the SWTs, the pitch of the weld seam (Distance to 1 Revolution), as shown in Fig. 6. It should be noted that all specimens displayed some initial geometric imperfections, in the form of ovalisation, variations in the weld

264

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

Fig. 4. Failure modes of tensile coupons for (a) SWT and (b) LWT.

thickness and tube length. The effects of this source of error were mitigated by taking the average of several measurements along the tube and averaging these values. The final dimensions of all twenty specimens are summarised in Tables 3 and 4. The test specimens considered were hollow SWT and LWT columns and concrete-filled SWT and LWT columns. The hollow SWT and LWT set consisted of ten columns and the composite SWT and LWT section set consisted of ten columns giving a total of twenty columns which are summarised in Table 5. The SWT and LWT sections were all fabricated using spiral and longitudinal welds of average 4.1 mm and 3.0 mm throat thickness (see Tables 3 and 4) along the full length of the columns. The SWT and LWT columns were fabricated from four equal diameter steel plates (i.e. 102 mm, 152 mm, 203 mm, and 254 mm). The nominal diameters of the columns are shown in Table 5. SWT and LWT nominal plate thicknesses were equal to 1.90 mm. The weld patterns in SWT and LWT fabricated columns are shown in Fig. 7. Moreover,

the non-dimensional plate slenderness [(D/t) (fy/250)] was determined as shown in Table 5. The SWT and LWT columns were also fabricated with end stiffeners to ensure that localised bearing failure was prevented at the ends. The aim of this was to avoid failure occurring at the column ends that would adversely affect the test results at low stresses. After the columns were fabricated, six stiffeners were welded to each end of the columns.

Table 2 Compressive strength of concrete. f′c (N/mm2) at 7

f′c (N/mm2) at 28 f′c (N/mm2) at testing age of columns

Cylinder number

days

days

1 2 3 Mean SD

23.17 23.9 22.47 23.18 0.72

39.89 39.85 36.6 38.78 1.89

49.93 48.88 51.2 50.00 1.16

Fig. 5. Compressive stress-strain curve for self-compacting concrete at 28 days.

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288 Stiffeners

265

Stiffeners

D3

D3

dweld Dist. to 1 Rev

D2

L

D1

D2

D1

LWT

SWT

Fig. 6. Initial dimension measurements layout.

The stiffeners were designed so that local buckling would not occur in the stiffeners. Details of both the cross-section and end stiffeners are illustrated in Fig. 6. After the welding was completed the box columns were positioned in a casting bay and SCC was poured inside.

The SWT and LWT columns had a different plate slenderness limit, the local buckling stress was calculated for the hollow columns and the composite columns to establish if any of the experimental columns would be affected by the onset of local buckling prior to full yield of tests

Table 3 Initial dimension measurements of SWT and LWT for hollow columns.

Table 4 Initial dimension measurements of SWT and LWT for concrete-filled columns.

D2 D3 tweld dweld Distance L D1 (mm) (mm) (mm) (mm) (mm) (mm) to 1 Rev. (mm)

Average D (mm)

SWT H-SWT102-S H-SWT152-S H-SWT203-S H-SWT254-S H-SWT203-L

300 450 600 752 1400

103.1 153.0 205.0 254.5 203.0

103.5 152.5 – – –

103.0 152.5 203.5 249.0 203.0

4 5 4 4 4

10 10 10 10 10

260 380 330 320 335

103.05 152.75 204.25 251.75 203.00

LWT H-LWT102-S H-LWT152-S H-LWT203-S H-LWT254-S H-LWT203-L

300 450 600 750 1400

101.5 152.0 204.0 256.5 203.0

102.0 151.5 – – –

102.0 151.0 204.0 256.5 205.0

3 3 3 3 3

5 5 5 5 5

– – – – –

101.75 151.50 204.00 256.50 204.00

D2 D3 tweld dweld Distance L D1 (mm) (mm) (mm) (mm) (mm) (mm) to 1 Rev. (mm)

Average D (mm)

SWT C-SWT102-S C-SWT152-S C-SWT203-S C-SWT254-S C-SWT203-L

300 450 600 747 1400

103.0 152.5 204.5 254.0 203.0

103.5 152.5 – – –

103.0 152.0 203.5 250.0 203.5

4.5 4 4 4 4

10 10 10 10 10

260 380 335 315 335

103.00 152.25 204.00 252.00 203.25

LWT C-LWT102-S C-LWT152-S C-LWT203-S C-LWT254-S C-LWT203-L

300 450 600 750 1400

101.5 153.0 204.0 255.0 203.0

101.7 152.0 – – –

101.0 151.5 204.0 255.0 204.0

3 3.5 3 3 3

5 5 5 5 5

– – – – –

101.25 152.25 204.00 255.00 203.50

266

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

Table 5 Experimental tests details. Specimen

D (mm)

t (mm)

D/t

L (mm)

f′c (N/mm2)

fy (N/mm2)

Column test type

Plate slenderness

Yield slenderness limit

H-SWT102-S H-SWT152-S H-SWT203-S H-SWT254-S H-SWT203-L H-LWT102-S H-LWT152-S H-LWT203-S H-LWT254-S H-LWT203-L C-SWT102-S C-SWT152-S C-SWT203-S C-SWT254-S C-SWT203-L C-LWT102-S C-LWT152-S C-LWT203-S C-LWT254-S C-LWT203-L

103.05 152.75 204.25 251.75 203.00 101.75 151.50 204.00 256.50 204.00 103.00 152.25 204.00 252.00 203.25 101.25 152.25 204.00 255.00 203.50

1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90 1.90

54.2 80.4 107.5 132.5 106.8 53.6 79.7 107.4 135.0 107.4 54.2 80.1 107.4 132.6 107.0 53.3 80.1 107.4 134.2 107.1

300 450 600 750 1400 300 450 600 750 1400 300 450 600 750 1400 300 450 600 750 1400

– – – – – – – – – – 50 50 50 50 50 50 50 50 50 50

288 288 288 288 288 277 277 277 277 277 288 288 288 288 288 277 277 277 277 277

Hollow Hollow Hollow Hollow Hollow Hollow Hollow Hollow Hollow Hollow Composite Composite Composite Composite Composite Composite Composite Composite Composite Composite

62.5 92.6 123.8 152.6 123.1 59.3 88.3 119.0 149.6 119.0 62.5 92.3 123.7 152.8 123.2 59.0 88.8 119.0 148.7 118.7

840.0 840.0 840.0 840.0 840.0 840.7 840.7 840.7 840.7 840.7 1454.9 1454.9 1454.9 1454.9 1454.9 1456.2 1456.2 1456.2 1456.2 1456.2

developing. The local buckling slenderness limits for hollow and concrete-filled circular steel tubes can be determined by using the local buckling coefficients derived by Bradford et al. [15–16], where the local buckling stress is determined from Eqs. (1) and (2), respectively. The elastic local buckling stress for a hollow circular steel tube section is given by:   2Es t σ ol ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  ffi D 3 1−v2

vs = 0.30. If the nominal yield strength of SWT and LWT (fy) is taken as 288 N/mm2 and 277 N/mm2, respectively, the yield slenderness limits for hollow and composite columns are therefore calculated using Eqs. (3) and (4), respectively. D 2Es ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  ffi t f y 3 1−v2s

ð3Þ

D 2Es ¼ pffiffiffiffiffiffiffiffiffiffiffiffi t f y 1−v2s

ð4Þ

ð1Þ

s

The elastic local buckling stress for concrete-filled circular steel tubes is given by:   2Es t σ ol ¼ qffiffiffiffiffiffiffiffiffiffiffiffi D 1−v2s

ð2Þ

where Es is the steel elastic modulus, vs is the steel Poisson's ratio, D is the outside diameter of the section, and t is the thickness of the section. The elastic modulus of the SWT and LWT were taken as 199,856 N/ mm2 and 192,392 N/mm2, respectively and the Poisson's ratio of steel,

The elastic local buckling stress is determined and this allows one to determine the post-buckling behaviour. The post-local buckling behaviour is best characterised using an effective width principle. The effective width is given by Eq. (3). De ¼α D

sffiffiffiffiffiffiffi σ ol fy

ð5Þ

where De is the effective diameter of the section and α is the

Outside Weld

Weld

5mm

10mm

Weld Interface

Steel Tube

1mm 2mm

Inside Weld

4mm

(a)

(b) Fig. 7. Weld pattern in (a) SWT and (b) LWT.

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

LSCT2

Table 6 Local buckling effect and experimental squash load of columns.

267

σol (N/mm2)

De/D

Class

Nu,exp (kN)

σlat (N/mm2)

H-SWT102-S H-SWT152-S H-SWT203-S H-SWT254-S H-SWT203-L H-LWT102-S H-LWT152-S H-LWT203-S H-LWT254-S H-LWT203-L C-SWT102-S C-SWT152-S C-SWT203-S C-SWT254-S C-SWT203-L C-LWT102-S C-LWT152-S C-LWT203-S C-LWT254-S C-LWT203-L

54.2 80.4 107.5 132.5 106.8 53.6 79.7 107.4 135.0 107.4 54.2 80.1 107.4 132.6 107.0 53.3 80.1 107.4 134.2 107.1

2699.54 1821.20 1362.00 1105.02 1370.38 2625.47 1761.83 1319.20 1054.32 1319.20 8102.56 5481.53 4091.00 3311.76 4106.09 7876.42 5285.49 3957.61 3162.97 3957.61

1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00

Compact Compact Compact Compact Compact Compact Compact Compact Compact Compact Compact Compact Compact Compact Compact Compact Compact Compact Compact Compact

183.36 221.93 274.50 320.40 296.80 174.00 256.70 285.20 352.70 282.70 609.81 1044.92 1788.00 2525.00 1698.00 555.29 1100.39 1827.00 2733.00 1719.00

– – – – – – – – – – 3.77 1.46 0.22 0.22 0.23 4.79 2.45 0.85 0.11 0.86

SG3 L/2

D/t

SG2

L

Specimen

SG6

SG1

imperfection parameter that was taken to be 1.0 and 0.65 for hollow and composite columns, respectively [16]. In Table 6, details of the local buckling stress and post-local buckling effective width ratios are presented.

SG5

L/4

SG4

2.3.1. Test equipment The experimental test series was carried out in a 2000 kN capacity Instron uniaxial testing machine and a large load actuator machine with a capacity of 5000 kN at The University of New South Wales. During the initial dimension measurements, it was noted that all the ends of the tubes were ragged or uneven and it was determined that this would lead to eccentric loading hence resulting in additional moments and thus inaccuracy of the experimental results. Two methods were proposed; either cutting or grinding the edges such that they were smooth

LSCT1

2.3. Experimental set up

Fig. 9. The location of strain gauges and LSCTs short columns.

but also level or to cap each end with flat plates to evenly distribute the axial load to the tube. Zimmerman et al. [17] demonstrated that welding a plate to the ends of hollow tubes was a viable method to transfer loads

Fig. 8. General view of short and long columns test setup.

268

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

to the specimens. Based on this, the hollow SWT and LWT columns for this study were appropriately grinded and a 10 mm thick steel plate was tack welded onto both ends. However, for the concrete-filled SWT and LWT columns, plaster was used to ensure the compressive load was uniformly distributed over the whole top and bottom surface [18]. This method was utilised for this experiment using 100 N/mm2 plaster and 16 mm thick plates. All columns were tested under displacement control so that the full displacement to failure and the resultant ductility could be observed. Moreover, this allowed more extensive observation of the post-local buckling reserve of strength of the steel plates and the unloading path was monitored in greater detail. Fig. 8 shows the general test setup view for short and long columns that were used throughout the set of tests. 2.3.2. Strain gauge locations The placement of the strain gauges was an important consideration in the design of the experiments. Strain gauges were used to monitor the strain distribution along the cross-section of the columns. Six strain gauges were used per specimen and they provided more reliable information on longitudinal strains. One transverse strain gauge and five vertical strain gauges were placed on the short and long SWT and LWT columns. The transverse strain gauge was used to monitor the confinement of the concrete provided by the steel and vertical strain gauges monitored the distribution of vertical strains to allow the monitoring of the buckling behaviour of the plate elements. The strain gauges used were general-purpose uniaxial foil strain gauges with a gauge length of 5 mm and temperature compensated for steel. Figs. 9 and 10 show the configurations of strain gauge locations that were used throughout the set of tests.

2.3.3. Linear strain conversion transducer locations Linear strain conversion transducers (LSCTs) were used to measure the axial shortening of short and long columns and also to measure the lateral displacement of the long columns. Two LSCTs in the top and bottom of short and long columns have been used for measuring the axial shortening and six LSCTs on two sides of the long columns have been used for measuring the lateral displacement. Figs. 8–10 show the configurations of LSCT locations that were used throughout the set of tests. 3. Experimental results This section outlines the results for failure loads and discusses the pertinent failure modes for each of the specimens. For the hollow and concrete-filled SWT and LWT columns, load–axial shortening measurements were all recorded. 3.1. Failure loads The maximum loads attained by all hollow and concrete-filled SWT and LWT columns are summarised in Table 6, where the maximum applied axial force is denoted by Nu,exp. 3.2. Load–axial shortening The axial (load-shortening) response of each column was recorded and these results were useful in being able to establish the location at which yielding took place and the onset of ultimate failure, which was generally characterised by concrete crushing and softening. Fig. 11

L/2

L/2

L/4

LSCT8 LSCT1

LSCT4

LSCT1

SG3

SG2

SG6

SG5

LSCT2

LSCT2

SG5

LSCT5

SG1

LSCT3

LSCT3

LSCT6

LSCT7

SG4

Front view

Back view Fig. 10. The location of strain gauges and LSCTs long columns.

SG6

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

Fig. 11. Load-axial shortening curves for hollow and concrete-filled SWT and LWT columns.

269

270

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

Fig. 12. Load-average longitudinal and lateral strain curves for hollow SWT columns (a) H-SWT102-S, (b) H-SWT152-S, (c) H-SWT203-S, (d) H-SWT203-L, and (e) H-SWT254-S.

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

271

Fig. 13. Load-average longitudinal and lateral strain curves for hollow LWT columns (a) H-LWT102-S, (b) H-LWT152-S, (c) H-LWT203-S, (d) H-LWT203-L, and (e) H-LWT254-S.

272

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

Fig. 14. Load-average longitudinal and lateral strain curves for concrete-filled SWT columns (a) C-SWT102-S, (b) C-SWT152-S, (c) C-SWT203-S, (d) C-SWT203-L, and (e) C-SWT254-S.

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

273

Fig. 15. Load-average longitudinal and lateral strain curves for concrete-filled LWT columns (a) C-LWT102-S, (b) C-LWT152-S, (c) C-LWT203-S, (d) C-LWT203-L, and (e) C-LWT254-S.

274

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

Fig. 16. Lateral displacement curves for long hollow and concrete-filled SWT and LWT columns (a) H-SWT203-L, (b) H-LWT203-L, (c) C-SWT203-L, and (d) C-LWT203-L.

illustrates axial (load-shortening) results for the hollow and concretefilled SWT and LWT columns tested. For the fabricated hollow sections, one can see that the composite sections have a larger stiffness, as well as reaching a larger ultimate load capacity. Furthermore, the apparent ductility of the hollow and concrete-filled SWT and LWT columns is presented to be quite acceptable with a significant post peak reserve of strength being revealed. 3.3. Load–strain The load-average longitudinal strain curves for the hollow and concrete-filled SWT and LWT columns are useful in establishing the onset of yield as well as highlighting local buckling on the compression faces which was inelastic in all the columns tested. Additionally, the load-average lateral strain curves and load-average lateral displacement curves were used to monitor the confinement of the concrete provided by the short and long concrete-filled SWT and LWT columns. The load-average longitudinal strain, load-average lateral strain, and load-average lateral displacement curves were plotted for each of the tests and these are

collectively shown in Figs. 12–16 for the hollow and concrete-filled SWT and LWT columns. The average transverse and vertical strains (i.e. εt and εv) were calculated, to observe the Poisson's ratio of the steel during loading. This analysis allowed the effect of confinement during elastic loading to be examined. The recorded load-transverse/vertical strains (εt / εv) illustrate that no confinement exists during the elastic range, as the Poisson's ratio of the concrete infill is lower than that of the steel box. Confinement of the concrete infill is not achieved until higher stages of loading when the Poisson's ratio of the concrete increases (as shown in Fig.17). The confinement of the concrete by the circular section was determined by Eq. (6). This equation determines the confining pressure by summing the transverse stresses on the plate elements (as shown in Fig. 18) [19]. The calculated lateral confining pressure (σlat) at yield load is summarised in Table 6. σ lat ¼

2t σ trans D−2t

ð6Þ

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

275

4.1. Model creation

Fig. 17. Typical load-transverse-to-vertical strain curves for short concrete-filled SWT and LWT columns.

where σtrans is the transverse stress in the steel (=Es εtrans), D represents the diameter of the circular section, t is the thickness of the steel plate, Es is the steel elastic modulus, and εtrans is the transverse strain in the steel.

3.4. Failure modes Different failure modes of hollow and composite columns were observed for the different column lengths. The main failure modes associated with these columns involved local buckling. None of the specimens showed any signs of weld fracture in the welding lines of SWT and LWT columns during the tests. All these failure modes are highlighted in Tables 7 and 8 for hollow and concrete-filled SWT and LWT columns, respectively. All columns were tested in pure compression and therefore failure was fundamentally compressive in nature. Failure commenced by compressive yield since all plate sections were compact. Once yielding began in the concrete-filled SWT and LWT columns, concrete crushing and inelastic local plate buckling of the plate elements generally followed this. The majority of local buckling failures of the SWT and LWT columns were due to a radial expansion of the steel tube in the inelastic range. Also, columns with large D/t ratios exhibited more local buckling, with higher apparent distortions, compared with the sections with smaller D/t ratios. This behaviour accounted for the differences observed in the strain-hardening and strain-softening characteristics of the steel tube. Moreover, among the concrete-filled SWT columns, local buckling occurred partially along the spiral weld line close to the centre of the column. In the C-SWT254-S column as shown in Table 8, if local buckling occurred in the spiral weld line, it had no great impact on the overall axial compressive behaviour of the column.

The steel tube was modelled with solid elements using the ‘extrusion’ tool and was reshaped using the ‘cut’ tool to make space for the weld to fit into the steel tube. There were two weld shapes used for the models with one representing the SWT weld and the other representing the LWT weld. The different weld shapes are a result of the slight difference in the manufacturing process of these two tubes, as shown in Figs. 7 and 20. The LWT weld has a straight seam weld and was modelled with an ideal straight base, despite the presence of rough lumps on the experimental samples. Based on the assumption that the steel tube-weld bond would be perfect, the SWT weld was modelled as a single part with the combined dimensions of the outer and inner welds. This effectively eliminated the need to define an interaction at the weld interface and minimised the computational time. Once the weld was aligned correctly into the cut seam of the steel tube, the concrete infill was modelled. The curved geometry of the tube posed a challenge as ABAQUS utilised a discretisation method in which a smooth continuous shape (such as the circle of the tube's ends) was replaced by a series of straight lines connected in the general shape of a circle. When the concrete was modelled as a circle and assembled together with the tube, the discretisation would cause areas of overlap and gaps along the steel-concrete boundary. In order to accurately represent the concrete as a component that tightly fits into the tube, which acts as formwork during its setting period, the concrete was modelled with a ‘cut/merge’ tool. This not only allowed a tight fit but no re-alignment of the concrete part was required. 4.2. Restraints and loading conditions To observe the buckling behaviour of the hollow and concrete-filled SWT and LWT columns under axial compression loading, it was necessary to impose the appropriate boundary conditions. The bottom end of the columns was fully fixed with zero degrees of freedom, whilst the top end was given a single degree of freedom for translation along the z axis. The end columns were planar during the analysis with these boundary conditions and therefore it was not necessary to take into account the end plates or stiffeners in the model. Furthermore, to ensure that the individual parts would be correctly assembled into a single structure, any separation needed to be restricted between the steel tube to concrete, weld to concrete and steel tube to weld surface boundary layer. The steel tube to weld boundary was assumed to be perfectly bonded and was thus modelled with a “tie” constraint, which would

σtrans

σlat

4. Finite element analysis The FEM program ABAQUS [20] was used in this study to develop an accurate FEM model for predicting the behaviour of hollow and concrete-filled SWT and LWT columns under axial compression. For the purposes of accurate FE analysis, element type, element mesh, boundary condition, steel tube–concrete interface, material properties for steel tube and confined concrete, initial imperfections and residual stresses must all be considered. FEM models of short and long SWT and LWT columns are depicted in Fig. 19.

t

D

σtrans Fig. 18. Confining pressure pattern in the circular steel section.

276

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

Table 7 Failure modes of hollow SWT and LWT columns.

Short SWT column D = 152 mm H-SWT152-S

Short SWT column D = 203 mm H-SWT203-S

Short SWT column D = 254 mm H-SWT254-S

Long SWT column D = 203 mm H-SWT203-L

Short LWT column D = 102 mm H-LWT102-S

Short LWT column D = 152 mm H-LWT152-S

Short LWT column D = 203 mm H-LWT203-S

Short LWT column D = 254 mm H-LWT254-S

Long LWT column D = 203 mm H-LWT203-L

LWT

SWT

Short SWT column D = 102 mm H-SWT102-S

eliminate any movement along the weld surface with respect to the contacting surface of the steel tube. The steel tube to concrete and weld to concrete surface boundaries could not be assumed to be

perfectly bonded and were defined using a contact interaction property which was simulated by the Coulomb friction model, with a “hard” normal contact and a conservative friction coefficient value of 0.25 [21].

Table 8 Failure modes of concrete-filled SWT and LWT columns.

Short SWT column D = 152 mm C-SWT152-S

Short SWT column D = 203 mm C-SWT203-S

Short SWT column D = 254 mm C-SWT254-S

Long SWT column D = 203 mm C-SWT203-L

Short LWT column D = 102 mm C-LWT102-S

Short LWT column D = 152 mm C-LWT152-S

Short LWT column D = 203 mm C-LWT203-S

Short LWT column D = 254 mm C-LWT254-S

Long LWT column D = 203 mm C-LWT203-L

LWT

SWT

Short SWT column D = 102 mm C-SWT102-S

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

277

Fig. 19. FEM models of short and long (a) SWT and (b) LWT columns in ABAQUS.

Moreover, to minimise the computation time, a “rigid body” constraint was applied to the top loading surface such that the whole surface would replicate the loading conditions of the reference point (located at the centre of the top plane). The load was modelled onto the constrained reference point as a controlled displacement value (the whole top surface would be subjected to a fixed vertical deflection). After conducting multiple trial tests to determine the most appropriate displacement-length ratio such that a certain degree of ultimate failure was reached, a final ratio of L/30 was applied to each model. 4.3. Mesh refinement and mesh type selection Several FEM models used in past research conducted by Scott and Finneran [22] and Zimmerman et al. [17] utilised 3D eight-node linear hexahedral solid elements with reduced integration (C3D8R) to model the steel tube and the weld. This particular mesh type provided reduced computational time as well as improved accuracy compared with the conventional eight-node linear iso-parametric element (C3D8). However, a study conducted by Brown [23] found that the eight-node linear solid element with incompatible modes (C3D8I) produced results of similar accuracy to the C3D8R elements. Despite this, the author recommended the use of C3D8I elements based on the fact that this element type was less sensitive to the mesh thickness than the C3D8R type. Furthermore, C3D8I elements were found to be very effective in displaying local instabilities like the buckling of the steel tube in areas with high compressive stresses. To verify which element type was the most appropriate for this analysis, a basic element type test was carried out. The result of a single test on the model of H-SWT102-S has been observed and the C3D8I model displays an improved distribution of stress across the whole tube and also produces a more realistic buckling deformation pattern. Based on these tests, it was concluded that C3D8I elements would be used to model all parts over the C3D8R elements. In theory, analytical results would converge towards a single solution as the mesh density is increased (smaller mesh size), however this would also lead to higher computational times even for the most simplest case. Although the use of C3D8I elements would minimise

the effects of mesh thickness on the accuracy of the results, a mesh sensitivity test similar to the test conducted by Kulkarni [24] was undertaken. An example of the mesh sensitivity test, where the von Mises stresses are compared, based on the H-SWT102-S sample is shown in Table 9. It is clear that as the mesh seed size is reduced from 12 to 11 there is a change of 5.5% in the von Mises stress and a 2.6% increase in the ultimate load. However when the mesh seed size is reduced further by 1 unit, there is a negligible 0.72% change in the von Mises stress and almost no change in the ultimate load capacity of the H-SWT102-S. Thus, it was assumed that convergence was reached and a mesh seed size of 10 was used for the final analysis of this particular model. This test was repeated for all hollow and concrete-filled SWT and LWT column models to obtain an appropriate mesh density that would produce accurate results. 4.4. Influence of initial geometric imperfections and residual stresses 4.4.1. Initial geometric imperfections SWTs are produced in accordance with the dimensional and tolerance requirements of EN 10219-2 [25], API 5L [26], ISO 3183 [27], and AS 1579 [28] specifications, as shown in Table 10. These dimensional and tolerance specifications are used by most of the SWT manufacturers and they are subsequently used for comparison with the available geometric tolerance experimental results of SWTs. Moreover, the use of narrower raw material like SWTs compared with one-seam LWTs means closer wall-thickness tolerances and the process itself does not rely on individual dies for every diameter. Tubes can be produced based on an inside diameter, outside diameter, or on any agreed diameter which also allows a wide variety in choosing the most economical diameter for a certain project. The SWT diameter tolerance is small, particularly with regard to ovality; and SWT has an inherent straightness due to its axial symmetry. No calibration process (expanding or roll sizing) is necessary. End and body tolerance of a joint are more or less the same. Such tolerances relate directly to the design of tubes, in which the out-of-roundness of the tube becomes a controlling factor

278

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

(a)

(b)

Fig. 20. Modelled weld patterns for (a) SWT and (b) LWT in ABAQUS.

in the determination of the wall thickness. It is apparent that an increasing ovality of a tube reduces its capacity to bear external stress. As initial imperfections are considered in the present model, an eigenvalue buckling analysis was first carried out to provide the lowest buckling mode to be used as the shape of the initial imperfection in following the load-deflection nonlinear analysis as shown in Figs. 21–22. The ultimate strength and post-buckling behaviour of hollow and concrete-filled SWT and LWT columns was then achieved from the nonlinear load-deflection analysis. The magnitude of the initial imperfections Table 9 Mesh sensitivity results of H-SWT102-S column. Mesh seed size

Max reaction force (kN)

Von Mises stress (N/mm2)

10 11 12

159 159 155

775.4 781.0 740.6

for circular steel tubes was introduced by means of a quantified imperfection value of 0.5t and 0.003D by Mahmoud et al. [29] and Wheeler and Pircher [30], respectively. To incorporate the effects of geometric imperfections primarily due to ovalisation, both imperfection values proposed by Mahmoud et al. [29] and Wheeler and Pircher [30] were considered. It was found that due to the small absolute wall thickness of SWT and LWT columns (i.e. 1.90 mm), the 0.5t formulation which equates to imperfections of 0.79 mm and 0.95 mm caused mesh volume difficulties (either zero or negative mesh volume) and thus a value of 0.003D was adopted for the analysis. 4.4.2. Residual stresses Sectional residual stress have a significant influence on the mechanical performance of steel structural members, especially the ultimate capacity of columns caused by premature yielding and loss of axial and flexural stiffness. The quality of SWT is directly influenced by forming

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

279

Table 10 Geometric tolerances of EN 10219-2, API 5L, ISO 3183, and AS 1579 specifications. Specifications

Outside diameter (D)

Wall thickness (t)

Straightness

Out-of-roundness

Mass

Maximum weld bead height

EN 10219-2 [25]

±1% Max. ±10 mm If D ≤ 1422 mm: ±0.5% If D N 1422 mm: ≤4 mm Max. ±10 mm ±0.5%

±10% Max. ±2 mm If b15 mm: ±10% If ≥15 mm: ±1.5% Max. ±2 mm –

0.2% of total length

±2%

±6%

0.2% of total length

If D/t ≤ 75: ±1.5% If D/t N 75: by agreement If D b 1422 mm: ≤15 mm ±1%

+10% −3.5%

t ≤ 14.2 mm: 3.5 mm t N 14.2 mm: 4.8 mm t ≤ 13 mm: 3.5 mm t N 13 mm: 4.8 mm t N 14.2 mm: 4.8 mm t b 12 mm: 1.5 mm t ≥ 12 mm: 3.0 mm

API 5L [26] ISO 3183 [27] AS 1579 [28]

0.2% of total length

and welding input parameters. During the production of SWT, the plate or strip material goes through bending, forming, and welding. These mechanisms interact to produce a complex three-dimensional residual stress field. On the other hand, even relatively small unbalanced states of residual stress at some locations could lead to stress corrosion cracking or corrosion fatigue fracture to steel tubes in the presence of a corrosive medium. Therefore, a common problem faced by SWT manufacturers is the control of process input parameters to obtain a good tube quality with minimal detrimental residual stresses [31–32]. Vasilikis and Karamanos [33] considered residual stresses in the FEM models for SWTs by average stress at 0.2% plastic strain and this distribution was applied directly to each tube as an initial stress state. Hence, in their FEM model the weld was not considered as a separate part of the model and it was considered as an initial stress loading to the steel tube. However, in this study and as reported by Eltaher et al. [34], the weld for the SWT and LWT columns is considered as a separate solid element and it is perfectly bonded with the steel tube.

4.5. Material properties for the steel tube and confined concrete 4.5.1. Steel tube and weld Steel material properties specified in ABAQUS included the elastic modulus of the SWT and LWT and Poisson's ratio (vs) taken as

+10% −3.5%

199,856 N/mm2 and 192,392 N/mm2 (as presented in Table 1) and of 0.3, respectively. In this FE model the steel is assumed to behave as an elastic–plastic material with strain hardening in compression. For the elastic-plastic model with linear hardening, the strain hardening modulus was taken as Es/200 [35]. The weld material was differentiated from the steel by having a yield strength 10% greater than that of the steel tube, based on past finite element modelling conducted by Susilo [36]. This accounts for welds having characteristically higher strength than that of the parent metal material that arises largely due to quenching [37].

4.5.2. Confined concrete The steel tube provides confinement to the concrete infill in CFST columns under axial compression which increases the strength and ductility of the concrete infill. The confinement effect depends on the diameter-to-thickness ratio of the steel tube and material properties. By using the FEM, strength improving at the state of triaxial loading can be achieved by the definition of the yielding surface, and the description of the plastic behaviour coming from the equivalent stress–strain relationships of the concrete infill [38]. The damage plasticity model defined in ABAQUS is used in the analysis. Since CFST columns are under monotonic axial compression, damage variables were not defined. Consequently, concrete nonlinearity was modelled as plasticity only.

Fig. 21. Typical first buckling mode shapes for short and long LWT columns.

280

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

Fig. 22. Typical first buckling mode shapes for short and long SWT columns.

Concrete material parameters required to be defined included elastic modulus (Ec), Poisson's ratio (νc), flow potential eccentricity (e), viscosity parameter (μ) (i.e. it is used for the visco-plastic regularization of the

concrete constitutive equations and the default value is 0.0), dilation angle (ψ), shape factor for yield surface (Kc), the ratio of initial equibiaxial compressive yield stress to initial uniaxial compressive

Table 11 FEM failure modes of hollow SWT and LWT columns.

Short SWT column D = 152 mm H-SWT152-S

Short SWT column D = 203 mm H-SWT203-S

Short SWT column D = 254 mm H-SWT254-S

Long SWT column D = 203 mm H-SWT203-L

Short LWT column D = 102 mm H-LWT102-S

Short LWT column D = 152 mm H-LWT152-S

Short LWT column D = 203 mm H-LWT203-S

Short LWT column D = 254 mm H-LWT254-S

Long LWT column D = 203 mm H-LWT203-L

LWT

SWT

Short SWT column D = 102 mm H-SWT102-S

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

281

Table 12 FEM failure modes of concrete-filled SWT and LWT columns.

Short SWT column D = 152 mm C-SWT152-S

Short SWT column D = 203 mm C-SWT203-S

Short SWT column D = 254 mm C-SWT254-S

Long SWT column D = 203 mm C-SWT203-L

Short LWT column D = 102 mm C-LWT102-S

Short LWT column D = 152 mm C-LWT152-S

Short LWT column D = 203 mm C-LWT203-S

Short LWT column D = 254 mm C-LWT254-S

Long LWT column D = 203 mm C-LWT203-L

LWT

SWT

Short SWT column D = 102 mm C-SWT102-S

yield stress (fb0/f′c), and equivalent stress-strain relationship for compression and tension sides, respectively. qffiffiffiffiffi 0 The elastic modulus of the concrete was calculated as Ec ¼ 3320 f c þ6900 (N/mm2), which was suggested by ACI [39], where f′c is in N/ mm2 and Poisson's ratio of νc = 0.20 is adopted. Default values of 0.1 and 0.0 were used for the flow potential eccentricity and viscosity parameter, respectively. These two parameters have no significant influence on the prediction accuracy.

The ψ, Kc, fb0/f′c can be determined as follows [35]: ψ ¼ 40

ð7Þ

 0 −0:075 f b0 0 ¼ 1:5 f c fc

ð8Þ

5:5  0 0:075 5 þ 2 fc

ð9Þ

Kc ¼

The compressive equivalent stress-strain curve proposed by Han et al. [38] was used to simulate the material behaviour of confined concrete in circular CFST columns. The Han et al. [38] equivalent stress– strain model, which is suitable for the FEA using ABAQUS software, is described as follows:    2 σc ε ε − 0 ¼2 εo εo fc

if

  ε σc εo  2  ε  0 ¼ fc β0 εεo −1 þ εo 0:2

ε o ¼ εc þ 800ξ

ε ≤1 εo

if

ε N1 εo

 10−6

 0 ε c ¼ 1300 þ 12:5f c  10−6 qffiffiffiffiffi 0 ½0:25þðξ−0:5Þ7  f c β0 ¼ 2:36  10−5 ≥0:12 2

Fig. 23. Ductility Index of concrete-filled SWT and LWT columns.

ξ¼

As f y Ac f ck

ð10Þ

ð11Þ

ð12Þ ð13Þ

ð14Þ ð15Þ

282

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

Fig. 24. Comparison between FEM predictions and measured axial load – axial shortening curves for hollow SWT and LWT columns.

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

283

Fig. 24 (continued).

in which σc is the longitudinal compressive stress of the concrete, ε is the longitudinal compressive strain of the concrete, εo is the strain at f′ c, As and Ac are the cross-sectional areas of the steel and concrete, respectively, fy is the yield strength of the steel, fck is the characteristic strength of the concrete, fck equals to 0.67 fcu for normal strength concrete, fcu is the cube strength of the concrete, and f′c is the cylinder strength of the concrete. The tensile behaviour is assumed to be linear until the tensile qffiffiffiffiffi 0 strength is reached, which is taken as 0:56 f c (N/mm2) suggested by ACI [39]. Beyond this tensile strength, the tensile response is represented by means of a fracture energy approach defined by FIP [40] and Bažant and Becq-Giraudon [41] as follows:   f 0 0:7 2 c N=m G f ¼ 0:0469d max −0:5d max þ 26 10

ð16Þ

where f′c is in N/mm2 and dmax is the maximum coarse aggregate size (in mm). Furthermore, if dmax was not reported in a reference, it was taken to be 20 mm.

5. Discussion of the results 5.1. Comparison of experimental and FEM failure modes In all hollow SWT and LWT columns, no noticeable signs of failure were detected prior to the specimens reaching their predicted ultimate load capacities. In the post-peak range, the structural failure processes became obvious. The mode of failure for the hollow columns tested was local buckling of the steel tube wall. The failure of the hollow LWT columns was characterised by the formation of a continuous concentric local buckle around the tube diameter as shown in Table 7. The appearance of such failure modes was observed to occur only as the ultimate axial capacity was reached. Also, hollow SWT columns were found to buckle ideally via folding of the steel tube running parallel to the weld seam. This folding action was matched in the experimental tests and FEM results as shown in Table 11. Also, results show that the angle between the loading direction and rolling direction in SWT columns (i.e. the angle of the spiral weld) will have negligible influence on the axial compression behaviour of the tubes. Concrete-filled SWT and LWT columns were found to fail in a concrete shear failure mode. The ultimate load capacity was linked

284

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

Fig. 25. Comparison between FEM predictions and measured axial load – axial shortening curves for concrete-filled SWT and LWT columns.

with the formation of a classical shear plane at an oblique angle within the concrete infill for all concrete-filled SWT and LWT columns tested. The initial signs of failure were shown by the

presence of small lateral bulging on the exterior of the SWT and LWT corresponding with the achievement of the ultimate load capacity. As specimens were loaded well into the post-peak

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

285

Fig. 25 (continued).

range, the angle of inclination became clearly noticeable with local buckling and Lüders bands appearing on the SWT and LWT exterior. These inelastic failure mechanisms were detected in the SWT and LWT until well into the post ultimate phase. Experimental test results not only confirmed this buckling shape but also verified the

increase in concrete strength due to the confinement effect. The general buckled shape produced in the FEM is remarkably similar to the buckling in the actual experiment. The failure mode predicted by the present model also agrees well with that observed in the tests as shown in Tables 11 and 12.

Table 13 Comparison between FEM, design codes predictions and experimental ultimate strengths for hollow columns. Specimen

D (mm)

D/t

Nu,exp (kN)

Nu,exp/Nu,FEM

Nu,exp/Nu,AS4100

Nu,exp/Nu,EC3

Nu,exp/Nu,AISC

H-SWT102-S H-SWT152-S H-SWT203-S H-SWT254-S H-SWT203-L H-LWT102-S H-LWT152-S H-LWT203-S H-LWT254-S H-LWT203-L Mean SD

103.05 152.75 204.25 251.75 203.00 101.75 151.50 204.00 256.50 204.00

65.2 96.7 129.3 159.3 128.5 53.6 79.7 107.4 135.0 107.4

183.36 221.93 274.50 320.40 296.80 174.00 256.70 285.20 352.70 282.70

1.01 1.01 0.99 1.00 1.03 1.01 1.03 0.99 1.00 1.01 1.01 0.01

1.27 1.20 1.28 1.34 1.40 1.05 1.08 1.03 1.13 1.03 1.18 0.13

1.26 1.03 0.95 0.90 1.04 1.05 1.04 0.85 0.84 0.85 0.98 0.13

1.27 1.03 0.95 0.90 1.05 1.06 1.04 0.86 0.84 0.86 0.99 0.13

286

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

Table 14 Comparison between FEM, design codes predictions and experimental ultimate strengths for concrete-filled columns. Specimen

D (mm)

D/t

Nu,exp (kN)

Nu,exp/Nu,FEM

Nu,exp/Nu,AS5100.6

Nu,exp/Nu,EC4

Nu,exp/Nu,AISC

Nu,exp/Nu,proposed design model

C-SWT102-S C-SWT152-S C-SWT203-S C-SWT254-S C-SWT203-L C-LWT102-S C-LWT152-S C-LWT203-S C-LWT254-S C-LWT203-L Mean SD

103.00 152.25 204.00 252.00 203.25 101.25 152.25 204.00 255.00 203.50

65.2 96.4 129.1 159.5 128.6 53.3 80.1 107.4 134.2 107.1

609.81 1044.92 1788.00 2525.00 1698.00 555.29 1100.39 1827.00 2733.00 1719.00

1.02 1.02 1.02 1.02 0.99 1.04 1.03 1.01 1.01 0.99 1.02 0.02

0.98 0.92 0.90 0.85 1.01 1.04 0.97 0.92 0.90 1.01 0.95 0.06

1.02 0.92 0.93 0.89 0.95 0.97 0.95 0.94 0.93 0.94 0.94 0.04

1.28 1.09 1.09 1.04 1.05 1.15 1.12 1.09 1.08 1.03 1.10 0.07

1.09 0.96 0.99 0.97 0.95 0.98 0.98 0.99 1.00 0.94 0.99 0.04

5.2. Concrete-filled SWT and LWT columns ductility To quantify the various ductile responses observed from the concrete-filled SWT and LWT columns load-displacement curves, a Ductility Index (DI) as adopted by Clarke [50] was applied. The Ductility Index is derived as a singular numeric ratio relating to the difference in deformation of the specimen in both loading and unloading curve branches for a particular proportion of ultimate axial capacity. DI is defined by (Δ2 − Δ1 / Δ1), where Δ1 is the axial displacement on the ascending branch of the curve corresponding to 90% ultimate load and Δ2 is the corresponding value on the descending curve. The DI results for all composite specimens are represented in Fig. 23. The results indicate that the degree of ductility is primarily dependent on the concrete strength and the quantity of the confining steel (δs = Asfy / Nu, where As is the area of the steel tube cross section, fy is the yield stress of the steel tube, Nu is the nominal section capacity). Fig. 23 shows that the ductility of concrete-filled SWT and LWT columns are almost similar.

5.3. Comparison of test strengths with FEM, Australian Standards, Eurocode, and American Institute of Steel Construction predictions Comparisons between the presented test results and the calculated results were carried out to verify the FE model. Figs. 24 and 25 show comparisons between the FEM predictions and the measured axial (load–shortening) results for hollow and concrete-filled SWT and LWT columns, respectively. The ultimate strengths predicted by the FEM and code provision predictions are compared with the experimental results in Tables 13 and 14 for hollow and concrete-filled SWT and LWT columns, respectively. Good agreement is observed with a mean value of 1.01 and 1.02 and standard deviations of 0.01 and 0.02 for Nu,exp/Nu,FEM of hollow and concrete-filled SWT and LWT columns. It can be observed that the present model simulates the nonlinear responses of hollow and concrete-filled SWT and LWT columns reasonably well. The test results Nu,exp for hollow SWT and LWT columns were compared with the predictions of axial load capacity by using the Australian Standards AS4100 [42], Eurocode 3 (EC3) [43], and AISC [44] models, as shown in Table 13. The mean value for Nu,exp/Nu,AS4100, Nu,exp/Nu,EC3 and Nu,exp/Nu,AISC are 1.18, 0.98 and 0.99, respectively. These results illustrate that the models for circular steel columns in AS4100, EC3, and AISC are conservative and thus appropriate for designers to use. Furthermore, the test results Nu,exp for concrete-filled SWT and LWT columns were compared with the predictions of axial load capacity by using the Australian Standards AS5100.6 [45], Eurocode 4 (EC4) [46], and AISC [44] models. The results presented in Table 14 show the effects of increasing the section slenderness on the squash load. The squash load is shown to decrease when the section slenderness increases and this issue is associated with the local buckling effect of the component steel plates of the section. The mean value for Nu,exp/Nu,AS5100.6, Nu,exp/Nu,EC4 and Nu,exp/ Nu,AISC are 0.95, 0.94 and 1.10, respectively. These results illustrate that

the models for composite circular columns in AS5100.6, EC4, and AISC are conservative and thus appropriate for designers to use. 5.4. Proposed design model The ultimate axial strength of circular CFST columns subjected to axial compression depends on the material and geometric properties. It also relies on the concrete confinement offered by the encased steel tube [47]. A design equation is proposed by Tang et al. [48] for predicting the ultimate axial strength of axially loaded circular CFST short columns. The design equation is given as follows: Nu;proposed design model ¼ γ s f y As þ σ 0cc Ac

ð17Þ

In the Tang et al. [48] method, the change of the Poisson ratio of concrete and steel with column loading is investigated. An empirical factor, β, is introduced for this purpose and subsequently the lateral pressure at the peak load is given by: 0

σ 0cc ¼ γc f c þ mβ

2t f D−2t y

ð18Þ

β ¼ ve −vs

ð19Þ

ve ¼ 0:2312 þ 0:3582v0e −0:1524 þ 4:8430v0e

! 0 fc fy !2

! 0 0 fc f −9:1690 c fy fy

 3  2 D D v0e ¼ 0:881  10−6 −2:580  10−4 þ 1:953 t   t D  10−2 þ 0:4011 t  −0:1 D γs ¼ 1:458 t

0:9≤γ s ≤1:1

γc ¼ 1:85ðD−2t Þ−0:135

0:85 ≤γc ≤1:0

ð20Þ

ð21Þ

ð22Þ ð23Þ

where σ′cc is the maximum compressive strength of confined concrete, Ac is the area of the concrete cross section, As is the area of the steel tube cross section, ve and vs are the Poisson ratios of a steel tube with and without filled-in concrete, respectively, vs is taken to be equal to 0.50 at the maximum strength point, f′c/fy ranging from 0.04 to 0.20 where most of the practically feasible columns are found within, γs and γc are the strength reduction factor for the steel tube and concrete, respectively which is given by Liang and Fragomeni [49], D is the diameter of the circular steel tube, t is the wall thickness of the steel tube, fy is the yield stress of the steel tube, f′c is the compressive strength of the concrete, and m is an empirical coefficient and is assumed to be 4.0.

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

Table 14 provides a comparison between the design model, codes and experimental results for the evaluation of existing guidelines of CFST columns. The proposed design model provides an accurate prediction of the ultimate axial strength for axially loaded circular CFST columns. The ratio of the mean ultimate axial strengths predicted by the proposed design model to the experimental results was 0.99. 6. Conclusions In conclusion, experimental and numerical investigations have been conducted in order to provide a better understanding of the behaviour of hollow and concrete-filled SWT and LWT columns when subjected to axial compression. The methods utilised to analyse the specimens involved experimental laboratory testing, finite element modelling, and a study of the suitability of international code provisions. The following conclusions can be drawn based on the results of this study: (a) The experimental laboratory testing of the specimens under displacement-controlled concentric axial compressive loading is amongst the first of its kind to be conducted. The experimental tests have provided load-displacement curves, ultimate axial strengths, qualitative identification of failure modes, and transverse and longitudinal strain data. The results suggest that SWT are as good as LWT in terms of strength. (b) Comprehensive refinement of simulations conducted using the commercial finite element program ABAQUS has produced a reliable FEM model for the nonlinear analysis of hollow and concrete-filled SWT and LWT columns. The FEM model accounts for the unique properties of the steel and weld materials, unique spiral geometry of the SWTs and initial imperfections and residual stresses. Verifying the FEM results against the measured experimental behaviour has illustrated that the model provides a sound, conservative prediction of the ultimate strength. (c) The measured experimental ultimate axial strengths have been used to demonstrate the suitability of Australian Standards, Eurocode, and American Institute of Steel Construction code provisions for the design of hollow and concrete-filled SWT and LWT columns under axial compression. The results have illustrated that Australian Standards, Eurocode, and American Institute of Steel Construction models are conservative and appropriate for designing hollow and concrete-filled SWT and LWT columns. (d) A design model for predicting the ultimate axial strengths was proposed for concrete-filled SWT and LWT columns. The proposed design model yields accurate predictions for the ultimate axial strengths of axially loaded composite sections. (e) The experimental results presented in this study could be expanded upon with further testing in the future. It is crucial that this testing be carried out in order to increase understanding as to how SWTs and LWTs would perform in construction (e.g. large diameter CFST columns, and high strength CFST columns). (f) Both analytical and experimental parametric studies are required to fully understand the behaviour of these complex structures under different loading conditions. The information in this study will be valuable in the development of new or modification of existing design codes, for the purpose of creating an easily available and reliable means for professional engineering practice. (g) Future studies on SWT columns under both bending and compression loading and cyclic loading are essential.

Acknowledgements The authors would like to acknowledge the Australian Research Council Discovery Projects Grants Scheme for their financial support of

287

this project under DP120101944 titled “The behaviour and design of composite columns coupling the benefits of high-strength steel and high-strength concrete for large scale infrastructure”. The experimental work was carried out in the structures laboratory of the School of Civil and Environmental Engineering at the University of New South Wales. The assistance of all the laboratory staff is acknowledged here.

References [1] K.H. Brensing, B. Sommer, Steel tube and pipe manufacturing processes, 2003 (64 p). [2] F.M. Knoop, B. Sommer, Manufacturing and use of spiral welded pipes for high pressure service-state of the artProceedings of IPC2004, International Pipeline Conference, Calgary, Canada 2004, pp. 1–9. [3] L. Richard, Spiral Welded Pipe Piles for Structures in Southeastern Louisiana(Master of Science Thesis) University of New Orleans, 2004. [4] Hatch Beddows, Global overview of line pipeAmerican Metals Markets, 2nd Annual Pipe & Tube Conference, 2008. [5] M. Hanada, H. Takeda, H. Fukushima, I. Koizumi, Y. Noguchi, Development of high speed submerged arc welding in spiral pipe mill, Transactions ISIJ 26 (1986) 433–438. [6] T. Burstall, Bulk Water Pipelines, Thomas Telford Ltd, 1997. [7] H. Liu, Pipeline Engineering, CRC Press, 2003 (448 p). [8] ArcelorMittal, Spirally welded steel pipes, Corporate Brochure, Projects Europe, Foundation Solutions, The Netherlands, 2011. [9] J. Lancaster, Handbook of Structural Welding: Processes, Materials and Methods Used in the Welding of Major Structures, Pipelines and Process Plants, Woodhead Publishing, 1997. [10] D. Goode, D. Lam, Concrete-filled steel tube columns—tests compared with Eurocode 4Proceedings of the Composite Construction in Steel and Concrete Conference VI, Colorado, US 2008, pp. 317–325. [11] D.E. Lehman, C.W. Roeder, Foundation connections for circular concrete-filled tubes, J. Constr. Steel Res. 78 (2012) 212–225. [12] G. Heiberg, A. Eltaher, P. Sharma, P. Jukes, M. Viteri, Spiral wound linepipe for offshore applicationsOffshore Technology Conference, Houston, Texas, 2011. [13] A. Sadowski, J. Rotter, T. Reinke, T. Ummenhofer, Analysis of variance of tensile tests from spiral welded carbon steel tubes, Constr. Build. Mater. 75 (2015) 208–212. [14] L.H. Han, G.H. Yao, X.L. Zhao, Tests and calculations for hollow structural steel (HSS) stub columns filled with self-consolidating concrete, J. Constr. Steel Res. 61 (9) (2005) 1241–1269. [15] M.A. Bradford, H.Y. Loh, B. Uy, Slenderness limits for CHS sectionsNinth International Symposium on Tubular Structures 2001, pp. 377–381 (Düsseldorf). [16] M.A. Bradford, H.Y. Loh, B. Uy, Slenderness limits for filled circular steel tubes, J. Constr. Steel Res. 58 (2002) 243–252. [17] T. Zimmerman, C. Timms, J. Xie, J. Asante, Buckling resistance of large diameter spiral welded linepipeProceedings of IPC2004 International Pipeline Conference, IPC2004-364, Calgary, Alberta, Canada, 2004. [18] B. Uy, Local and post-local buckling of concrete filled steel welded box columns, J. Constr. Steel Res. 47 (1998) 47–72. [19] K.A.S. Susantha, H. Ge, T. Usami, Uniaxial stress–strain relationship of concrete confined by various shaped steel tubes, Eng. Struct. 23 (10) (2001) 1331–1347. [20] ABAQUS, ABAQUS Standard User's Manual, Version 6.12, Dassault Systèmes Corp., Providence, RI (USA), 2012. [21] E. Ellobody, Numerical modelling of fibre reinforced concrete-filled stainless steel tubular columns, Thin-Walled Struct. 63 (2013) 1–12. [22] C. Scott, S. Finneran, Conceptual design of a spiral wound pipeline with manufactured residual stressesASME Pressure Vessels and Piping Conference, 2013. [23] J. Brown, Characterization of MSC/NASTRAN & MSC/ABAQUS Elements for Turbine Engine Blade Frequency Analysis, Air Force Research Lab, 1997. [24] M. Kulkarni, Finite Element Analysis of 2-D Representative Volume Element(Postgraduate Thesis) Mechanical Engineering, Michigan Technological University, 2012. [25] EN 10219-2, Cold Formed Welded Structural Hollow Sections of Non-alloy and Fine Grain Steels-Part 2: Tolerances, Dimensions and Sectional Properties, European Committee for Standardization (CEN), 2006. [26] API 5L, Specification for Linepipe, American Petroleum Institute, 2004. [27] ISO 3183, Petroleum and Natural Gas Industries – Steel Pipe for Pipeline Transportation Systems, 2007. [28] AS 1579, Arc-welded Steel Pipes and Fittings for Water and Wastewater, Standards Australia, 2001. [29] A. Mahmoud, S. Torabian, A. Jay, A. Myers, E. Smith, B.W. Schafer, Modeling protocols for elastic buckling and collapse analysis of spirally welded circular hollow thin-walled sections, Annual Stability ConferenceStructural Stability Research Council, 2015. [30] A. Wheeler, M. Pircher, Measured imperfections in six thin-walled steel tubes, J. Constr. Steel Res. 59 (2003) 1385–1395. [31] F. Aslani, B. Uy, S. Hicks, W.H. Kang, Spiral welded tubes – imperfections, residual stresses, and buckling characteristicsThe Eighth International Conference on Advances in Steel Structures, July 21–24, Lisbon, Portugal, 2015. [32] A.M. Arif, A.S. Al-Omari, A.K. Sheikh, Y. Al-Nassar, Estimation of residual stress in spiral welded pipe – regression and numerical modelProceedings of the ASME 2012 International Mechanical Engineering Congress & Exposition, IMECE2012-87703, Houston, Texas, USA (9-15/11), 2012.

288

F. Aslani et al. / Journal of Constructional Steel Research 128 (2017) 261–288

[33] D. Vasilikis, S.A. Karamanos, Bending Resistance of Steel Tubes in CombiWalls Task 4.2: Modeling of the Spiral-Welded Pipe. Research Report, University of Thessaly, 2014. [34] A. Eltaher, S. Jafri, P. Jukes, G. Heiberg, Advanced finite element analysis for qualification of spiral welded pipe for offshore applicationInternational Offshore and Polar Engineering Conference, 2012. [35] F. Aslani, B. Uy, Z. Tao, F. Mashiri, Behaviour and design of composite columns incorporating compact high-strength steel plates, J. Constr. Steel Res. 107 (2015) 94–110. [36] A. Susilo, Scripting for Modeling Local Buckling of a Spiral Welded Pipe(Masters thesis) Memorial University of Newfoundland, 2014. [37] R. Narayanan, V. Kalyanaraman, A.R. Santhakumar, S. Seetharaman, S.R. Satish Kumar, S. Arul Jayachandran, R. Senthil, Welds-Static and Fatigue Strength – II, vol. 2(31), Institute for Steel Development & Growth Teaching Material, 2011 1–27. [38] L.H. Han, G.H. Yao, Z. Tao, Performance of concrete-filled thin-walled steel tubes under pure torsion, Thin-Walled Struct. 45 (1) (2007) 24–36. [39] ACI-318, Building Code Requirements for Reinforced Concrete, ACI, Detroit, MI, 2008. [40] FIP, CEB-FIP Model Code 1990, Thomas Telford Ltd., London, 1993. [41] Z.P. Bažant, E. Becq-Giraudon, Statistical prediction of fracture parameters of concrete and implications for choice of testing standard, Cem. Concr. Res. 32 (4) (2002) 529–556.

[42] Standards Australia AS 4100-1998/Amdt1-2012, Australian Standard: Steel Structure, 2012. [43] Eurocode 3, Design of Steel Structures – Part 1-1: General Rules and Rules for Buildings. BS EN 1993-1-1, British Standards Institution, London, UK, 2005. [44] American Institute of Steel Construction, Load and Resistance Factor Design Specification for Structural Steel Buildings(Chicago) 2005. [45] Standards Australia AS5100.6-2004, Bridge Design, Part 6: Steel and Composite Construction, 2004. [46] Eurocode 4, Design of Composite Steel and Concrete Structures, Part 1.1: General Rules and Rules for Building. BS EN 1994-1-1, 2004 (London, UK; 2004). [47] F. Aslani, B. Uy, Z. Tao, F. Mashiri, Predicting the axial load capacity of high-strength concrete filled steel tubular columns, Steel Compos. Struct. 19 (4) (2015) 967–993. [48] J. Tang, S. Hino, I. Kuroda, T. Ohta, Modeling of stress–strain relationships for steel and concrete in concrete filled circular steel tubular columns, Steel Constr. Eng. 3 (11) (1996) 35–46. [49] Q.Q. Liang, S. Fragomeni, Nonlinear analysis of circular concrete-filled steel tubular short columns under axial loading, J. Constr. Steel Res. 65 (12) (2009) 2186–2196. [50] W.S. Clark, Axial load capacity of circular steel tube columns filled with high strength concrete, Victoria University of Technology, Research Master thesis, 1994.