Professional Documents
Culture Documents
Buckling Behavior and Design of Concentrically Loaded T-Section Aluminum Alloy Columns
Buckling Behavior and Design of Concentrically Loaded T-Section Aluminum Alloy Columns
Buckling Behavior and Design of Concentrically Loaded T-Section Aluminum Alloy Columns
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
A R T I C L E I N F O A B S T R A C T
Keywords: Tees are utilized as chord members in a number of arrangements of aluminum alloy trusses. The purpose of this
Aluminum alloy T-section columns study is to investigate the buckling behavior and design methods of concentrically loaded T-section aluminum
Local buckling alloy columns. Using column tests and finite element analysis, this study analyzed the buckling modes and ul
Eurocode 9
timate strengths of high-strength aluminum alloy T-columns under axial compression and validated the nu
Torsional buckling
Flexural-torsional buckling
merical model against the experimental results. Comparisons with the current Eurocode 9 and Direct Strength
Method (DSM) design rules were also conducted. Fixed- and pin-ended columns exhibit a quite different buckling
behavior. The failure modes of flexural buckling about the two principal axes could be treated with sufficient
safety and accuracy using the column design rules in Eurocode 9 and DSM. The effects of the effective centroid
shift due to the local buckling of slender parts could be ignored. The Eurocode 9 and DSM leads to inaccurate and
conservative results for fixed-ended columns failing by local-torsional buckling and flexural–torsional buckling.
The paper recommends suggestions based on the DSM approach that improve the design strength predictions for
aluminum alloy T-section columns.
* Corresponding author.
E-mail address: zhangqilin@tongji.edu.cn (Q. Zhang).
https://doi.org/10.1016/j.engstruct.2022.114221
Received 8 September 2021; Received in revised form 11 March 2022; Accepted 31 March 2022
Available online 5 April 2022
0141-0296/© 2022 Elsevier Ltd. All rights reserved.
L. Yuan and Q. Zhang Engineering Structures 260 (2022) 114221
of rectangular hollow columns accurately. On the reviewed papers, it Fig. 2. Geometry and notation of T-section.
can be seen that there is a lack of experimental information on the
buckling behavior of T-columns. Besides, aluminum columns’ flexur
accordance with the Chinese Standard for Tensile testing for Metallic
al–torsional buckling and local buckling behavior have not been thor
materials [19] (Fig. 3). The tensile coupons were tested in Tinius Olsen
oughly investigated.
150ST (maximum test load of 150 kN) with a displacement rate of 2
Experimental investigations have been conducted on eleven T-sec
mm/min. Fig. 4 plots the stress–strain curves for coupons extracted from
tion columns made from high-strength aluminum alloy 7075-T6. The
Section DS2. The average values of the elastic modulus E, the proof stress
program comprised fifteen tensile coupon tests, seven fixed-ended col
(f0.1 and f0.2), the ultimate strength (fu), the ultimate strain (εu), and the
umn tests, and four pin-ended column tests. Initial geometric imper
exponent n in the Ramberg-Osgood law [20] for each section are sum
fections of the specimens were measured before the compression column
marized in Table 3. The average values of f0.2 and fu are 521.4 MPa and
tests. The experimental results were used for the validation of finite
588.7 MPa, respectively.
element (FE) models. Moreover, a series of parametric results were
generated. The design provisions given in EC9 and DSM were assessed
by the ultimate strengths of T-section columns obtained from the tests 2.3. Test setup and instrumentation
and FE analysis.
Compression tests of aluminum alloy 7075-T6 T-columns were con
2. Experimental investigation ducted under fixed and pinned ends to investigate their buckling
behavior. A hydraulic jack with a capacity of 2000 kN was used for
2.1. Test overview applying compression load onto the specimens. The anchor device
consisted of a hardened endplate and four profiled stiffening plates. Two
Five cross-sections were selected after a careful study of the cross- pairs of bolts were used to adjust the spacing of the stiffing plates to
sectional behavior, as given in Table 1. The sectional geometric pa clamp the specimen (Fig. 5(a)). The anchor device constrained end
rameters of the T-section are shown in Fig. 2. The plate slenderness of twisting and warping of the column ends. For rigidly-fixed tests, the
the five cross-sections is classified as Class 4 according to the slenderness endplate was directly connected to the load plate, which could introduce
limits for flat outstand parts defined in EC9 [14]. The geometric di the applied load to the specimens. For pinned tests, the endplate was
mensions of the tested specimens were measured and presented in hinged connected to the load plate through knife edges (Fig. 5(b)), and
Table 2. “L” is the column length. The test specimens are labeled such thus the specimen was free to rotate about two principal axes at both
that the bearing condition, the section ID, and the nominal length are ends. Table 4 shows the boundary conditions in the tests. Fig. 6 shows
shown. For example, the label “F-SS1-450” could be interpreted as fol the test set-up for the rigidly fixed and pin-ended specimens.
lows: i) “F” refers to the fixed-ended boundary condition; ii) “SS1” refers Each specimen was carefully aligned geometrically, so that the cross-
to the cross-section given in Table 1; iii) the number “450” is the nom sectional centroid and centroid of the bearing plate coincided. The x-axis
inal length of the specimen in millimeters. Besides, the initial geometric and y-axis of the cross-section of the specimen were parallel to the
imperfections of all specimens were measured before compression tests. adjacent and far-away knife edges, respectively. In this case, the corre
The maximum values of the initial torsional twist (tanβ) and crooked sponding effective buckling lengths (Lx and Ly) in both directions are
ness parallel to the y-axis (dV) and x-axis (dH) were given in Table 2. depicted in Fig. 6 and given in Table 2. Two draw-wire displacement
sensors (DS1 and DS2) were positioned on the bottom endplate and
extended to the top endplate to measure the end shortening and rotation
2.2. Material testing of the specimen. Four LVDTs were placed at the mid-height of the
specimen to record the specimen’s lateral displacements. Besides, four
Fifteen tensile coupon tests were conducted to determine the 7075- strain gauges (SGs) were attached at a distance of 5 mm from the tips of
T6 alloy properties of the specimens. For each cross-section, three cou the web and flange, and two SGs were attached at the center of the
pons were extracted from the center of the web plate and flange plate flange and web at the mid-height to record the structural behavior with
along the longitudinal direction. The dimensions of the coupons were in progressing load increments. After the specimen was aligned in the
testing machine, the test was started with an initial load of 1/15 of the
Table 1 predicted ultimate load capacity of the column. This was done to pre
Cross-section dimensions, geometric properties of selected cross-sections. serve the alignment established at the beginning of the test. Further
Section Cross-section dimensions Plate slenderness loads were applied at a rate of 6.9 MPa/min [21], and the corresponding
ID deflections, strains and loads were recorded instantly.
bf tf H tw(mm) A βw = βf =
(mm) (mm) (mm) (mm2) hw/tw b/tf
2
L. Yuan and Q. Zhang Engineering Structures 260 (2022) 114221
Table 2
Measured geometries and initial imperfections of the T-section specimens.
Specimen bf (mm) H (mm) tf (mm) tw (mm) L (mm) Lx (mm) Ly (mm) tanβ (10− 2) dH (mm) dV (mm) Pu,test (kN)
F-SS1-450 75.4 41.09 6.53 6.49 451.0 – – 1.3 0.15 0.13 364
F-SS1-600 75.3 41.09 6.47 6.47 600.0 – – 1.8 0.32 0.08 322
F-DS1-550 102.3 76.07 6.49 6.55 549.5 – – 0.8 0.15 0.38 417
F-DS2-500 75.2 75.15 9.58 9.58 500.5 – – 1.0 0.28 0.32 719
F- DS2-750 75.2 75.10 9.60 9.54 750.0 – – 1.8 0.33 0.42 657
F-DS2-1000 74.5 74.91 9.38 9.54 1001.0 – – 2.8 0.69 0.52 556
F-DS3-1000 75.0 75.13 6.51 6.46 1000.5 – – 3.0 0.78 0.27 305
P-SS1-350 75.1 41.19 6.50 6.49 350.2 478.1 580.8 0.5 0.22 0.13 227
P-SS2-350 102.4 54.12 6.41 6.48 350.4 478.3 581.0 0.7 0.16 0.08 387
P-SS2-550 102.4 54.17 6.51 6.55 549.7 681.5 780.3 1.0 0.56 0.09 213
P-SS2-750 102.2 54.15 6.51 6.62 751.0 884.4 981.6 1.3 0.62 0.36 158
Table 4
Boundary conditions adopted in experiments.
Displacement in Rotation about Warping
x, y, z axes x, y, z axes
Fig. 3. Details of the tensile coupon tests. ux uy uz θx θy θz θw
barely noticeable for loads up to the ultimate load, where it became most
evident (Fig. 7(a)). When the applied load was close to the peak load,
LVDT-1 and LVDT-2 moved towards the opposite direction, while LVDT-
4 moved longer than LVDT-3 since it was located farther from the shear
center. For the cross-section DS2, when the member length was 500 mm
(F-DS2-500), failure was initiated by local buckling of the flange and
web. At the ultimate load, the specimen failed because of the combi
nation of local buckling and torsional buckling (Fig. 7(b)). For F-DS2-
Fig. 4. Experimental stress–strain curves of coupons extracted from the Sec 750, F-DS2-1000 and F-DS3-1000, all these specimens failed by flexural
tion DS2. buckling about the symmetry axis (Fig. 7(c, d, e)). Two flexual buckling
directions about the asymmetry axis (x-axis) were observed from the
tests, namely “FBx-D” and “FBx-U”, indicating that the failed column
Table 3
buckled away from and towards the flange, respectively. The specimens
Measured material properties of T-section specimens.
F-SS1-450 and F-SS1-600 failed in the FBx-D mode (Fig. 7(f, g)), while
Section label E (MPa) f0.1 (MPa) f0.2 (MPa) fu (MPa) εu (%) n
the four specimens in the pin-ended tests failed in the FBx-U mode
SS1 74,700 466.6 475.3 553.4 12.8 24.4 (Fig. 7(h - k)). Fig. 9 compares the longitudinal strain distribution of the
SS2 74,082 537.9 554.6 610.1 15.4 37.2 cross-section at the mid-height for the two flexural buckling directions.
DS1 71,500 453.8 461.2 538.7 14.6 24.0
DS2 75,300 556.1 564.0 627.8 11.0 35.5
For the specimen F-SS1-450, the distribution of longitudinal strain
DS3 74,500 547.5 552.2 613.5 14.0 35.9 demonstrated a uniform compression into the cross-section in the early
Average 74,016 512.4 521.4 588.7 13.6 31.4 loading phase. When the ultimate load was reached, the flange and web
were compressed. In the descending post-buckling path, the induced
“downward” bending generated additional compression in the flange
and tensile strain in the web tip (Fig. 9(a)). Conversely, for the specimen
3
L. Yuan and Q. Zhang Engineering Structures 260 (2022) 114221
Fig. 7. Failure mode of tested specimens (TB: torsional buckling, FBx: flexural buckling about x-axis, FBy: flexural buckling about y-axis, LB: local buckling, U:
upward, D: downward).
4
L. Yuan and Q. Zhang Engineering Structures 260 (2022) 114221
Fig. 8. The specimen F-DS1-550 (a) load–displacement curve and (b) load-torsional rotation curve.
Fig. 9. The strain distribution at the mid-height for the specimen (a) F-SS1-450 and (b) P-SS1-350.
3. Numerical modeling
5
L. Yuan and Q. Zhang Engineering Structures 260 (2022) 114221
6
L. Yuan and Q. Zhang Engineering Structures 260 (2022) 114221
Table 5
Cross-section geometry and buckling modes.
ID bf (mm) H (mm) tf, tw (mm) βw βf Fixed-ended columns Pin-ended columns
LT (mm) Modes L < LT Modes L > LT LT (mm) Modes L < LT Modes L > LT
iii) The shape of the local mode involved in the critical buckling are shown in Eqs. (4)-(6). Pne is the nominal axial strength for flexural,
mode depends mainly on the flange slenderness relative to the torsional, torsional-flexural for columns. Pcre is the critical elastic
web. As shown in Fig. 15(d.2, e.2), modes 5 and 6 participate in buckling load for the corresponding global buckling mode. Pnl is the
the mixed local-torsional-flexural modes for pinned columns nominal axial force for local buckling. Pcrl is the critical elastic local
T76.2–76.2–5 and T114.3–76.2–8, respectively. For the former buckling load. The value of Pcrl could not be determined from the
cross-section, the flange can provide clamped rotational restraint signature curves directly since the signature curves of the T-columns
on the web along the adjoined edge since the flange’s slenderness analyzed in this paper do not exhibit local minima corresponding to
is much less than the web slenderness (βw = 2βf). As a result, the local buckling. In this paper, Pcrl is calculated by using the revised
flange exhibits a rigid-body rotation about the adjoined edge buckling coefficient k influenced by the element interaction, which was
while the web exhibits transverse bending (mode 5). However, derived by Bleich [26] for the web plate in T-sections, see Eq. (7). The
for the latter cross-section, since the flange’s slenderness is interaction coefficient ζ can be computed by Eq. (8).
slightly less than the web slenderness (βw = 1.3βf), the flange
Pn = min(Pne , Pnl ) (4)
provides boundary conditions between simply-supported and
fixed along the adjoined edges. As a result, the flange and web ⎧
⎨ (0.658λ2c )P for λ ⩽1.5
exhibit both torsion and transverse bending, but the flange has a
(5)
y c
Pne = ( )
smaller rotation than the web (mode 6). ⎩ 0.877/λ2c Py for λc > 1.5
7
L. Yuan and Q. Zhang Engineering Structures 260 (2022) 114221
Fig. 15. Signature curves, modal participation diagrams for T-sections in parametric analysis.
there might be a shift in the neutral axis. However, aluminum extrusions strengths predicted by EC9 and DSM for columns failing by LTB and FTB.
usually do not contain very slender plate elements because the extrusion It is found that the design strength predictions are particularly conser
process is generally unsuited for producing them. Based on the analyzed vative for fixed-ended columns failing by LTB and FTB. The mean and
columns, it is concluded that the effects of effective centroid shift can be CoV of the Pu,FE/Pu,EC ratio is 1.57/0.22 and 1.62/0.28 for LTB and FTB
ignored for the design of concentrially loaded columns. columns, respectively. The mean and CoV of the Pu,FE/Pu,DSM ratio is
Fig. 17 compares the numerical strengths and the nominal axial 1.32/0.08 and 1.41/0.03, indicating a slight improvement of the mean
8
L. Yuan and Q. Zhang Engineering Structures 260 (2022) 114221
Table 6 mode plays a more critical role than the flexural buckling mode in the
Examination for higher-order bifurcation loads of local buckling. post-buckling range. The ultimate load can be determined with the
ID BCs L Critical (First) Higher-order local buckling mode nominal axial force for local buckling in DSM (Eq. (6)) and the critical
(mm) buckling mode elastic local buckling load Pcrl is replaced by the critical torsional
Shape Load Order Shape Bifurcation buckling load. When a column with Pcrft < Pcrt,∞, its representative point
(kN) load (kN) falls in domain 2 of the signature curve, and the globally flexural
S1 Fixed 200 3þ4 475.2 5 9 844.2
buckling mode about the symmetry axis is decisive in the post-buckling
S1 Pinned 125 3þ4 192.7 4 9 519.5 behavior. The ultimate load can be calculated by employing the current
S2 Fixed 150 4 887.1 3 8 1185.9 DSM provisions for concentrically loaded T-columns bending about the
S2 Pinned 150 3þ4 339.3 6 8 776.6 symmetry axis. As the member’s length increases, the post-buckling
behavior exhibits local-dominant mode for short to moderate lengths
and globally flexural-dominant mode for moderate to long lengths. The
and a slightly lower CoV compared to the current EC9 procedure. For
movement from “local” to “global” is taken into account by the sug
pin-ended columns failing by FTB, the comparisons between the nu
gested approach. Fig. 17(a, b, c) shows the comparison of FE results
merical strengths to the code predictions are shown in the last row of
against the results predicted by the suggested design approaches. The
Table 7. The mean and CoV of the ratio Pu,FE/Pu,EC are 1.21 and 0.09,
values of Pm, CoV and reliability index β of the ratio Pu,FE/Pu,pred for LTB
respectively for such columns. The DSM predictions, in this case are
and FTB columns are found to be 1.05, 0.06, 2.72; and 1.01, 0.05, 2.55
quite close with their EC9 counterparts.
respectively (see Table 7). It indicates that the suggested design
approach leads to quite accurate and economic predictions for torsional-
4.4. Modified DSM approach
involved buckling ultimate loads of fixed-ended columns. Finally, the
suggestions for the application of DSM to aluminum alloy concentrically
In this section, a modified DSM strength prediction model is pro
loaded tees are summarized in Table 8.
posed to improve the prediction accuracy for fixed-ended columns
failing by FTB and LTB. Fig. 18 depicts a typical signature curve of a
5. Conclusions
fixed-ended T-section. The asymptotic value of the critical pure torsional
buckling load for very long columns is calculated by Eq. (9) and defined
The buckling behavior of aluminum alloy T-columns under axial
as Pcrt,∞.
compression were investigated in this study. Eleven column tests were
⃒
GJ + π2 EIw /LT2 ⃒⃒ GJ conducted under fixed-ended and pin-ended conditions. A finite element
Pcrt,∞ = = 2 (9)
2 2
r0 + y 0 ⃒
LT =∞ r 2
0 + y0
model was developed and validated against the experimental results.
Using the validated FE model, four typical cross-sections were selected
where G is the shear modulus of elasticity. LT is the effective length in the parametric study. The buckling behavior and ultimate loads of
against torsion. J is the torsional constant.r02 = i2x + i2y + y20 , where ix these sections over a wide range of lengths were determined through the
and iy are the radii of gyration about the x-axis and y-axis, respectively. GBTUL analysis and numerical simulations. The generated data were
y0 is the shear center coordinate (see Fig. 2). then compared with the current Eurocode 9 [14] and DSM [18] design
By comparing the critical FTB load Pcrft with the value Pcrt,∞, two methods. Based on the results analyzed in this work, we conclude the
domains in the Pcrft vs. L curve are divided. When a column with Pcrft > following:
Pcrt,∞, its representative point falls in domain 1, and local/torsional
Fig. 16. Comparison of the FB ultimate strengths with the design rules in (a) EC9 and (b) DSM.
Table 7
Means, CoVs and relability indexes of the numerical-to-predicted ultimate strength ratios provided by EC9, DSM and proposed design suggestions.
Group Pu/Pu,EC Pu/Pu,DSM Pu/Pu,pred
9
L. Yuan and Q. Zhang Engineering Structures 260 (2022) 114221
Fig. 17. Comparison of FEA and predicted strengths for columns failing by LT and FT buckling.
Pcrx < Pcrx > Pcrft, Pcryt > Pcrx > Pcrx < Pcrft Pcrx > Pcrft The authors declare that they have no known competing financial
Pcrft Pcrt,∞ Pcrft, interests or personal relationships that could have appeared to influence
Pcryt < the work reported in this paper.
Pcrt,∞
10
L. Yuan and Q. Zhang Engineering Structures 260 (2022) 114221
References [11] Sena Cardoso F, Rasmussen KJR. Behavior and design of concentrically loaded T-
section steel columns. J Struct Eng 2014;140(7):04014039. https://doi.org/
10.1061/(ASCE)ST.1943-541X.0000969.
[1] Adeoti GO, Fan F, Wang YJ, Zhai XM. Stability of 6082–T6 aluminium alloy
[12] Dinis PB, Camotim D, Silvestre N. On the local and global buckling behaviour of
columns with H-section and rectangular hollow sections. Thin-Walled Struct 2015;
angle, T-section and cruciform thin-walled members. Thin-Walled Struct 2010;48
89:1–16. https://doi.org/10.1016/j.tws.2014.12.002.
(10–11):786–97. https://doi.org/10.1016/j.tws.2010.04.012.
[2] Wang YQ, Yuan HX, Chang T, Du XX, Yu M. Compressive buckling strength of
[13] Taras A, Kugler P, Unterweger H. On the behaviour and Eurocode design of T-
extruded aluminium alloy I-section columns with fixed-pinned end conditions.
section columns, beams and beam-columns with slender webs. J Constr Steel Res
Thin-Walled Struct 2017;119:396–403. https://doi.org/10.1016/j.
2017;129:250–62. https://doi.org/10.1016/j.jcsr.2016.10.017.
tws.2017.06.034.
[14] Eurocode 9: Design of aluminum structures – Part 1-1: General structural rules.
[3] Wang YJ, Fan F, Lin SB. Experimental investigation on the stability of aluminium
EN1999-1-1. European Committee for Standardization; 2007.
alloy 6082 circular tubes in axial compression. Thin-Walled Struct 2015;89:54–66.
[15] Su MN, Young B, Gardner L. The continuous strength method for the design of
https://doi.org/10.1016/j.tws.2014.11.017.
aluminium alloy structural elements. Eng Struct 2016;122:338–48. https://doi.
[4] Wang ZX, Wang YQ, Sojeong J, Ouyang YW. Experimental investigation and
org/10.1016/j.engstruct.2016.04.040.
parametric analysis on overall buckling behavior of large-section aluminum alloy
[16] Piluso V, Pisapia A. Interactive plastic local buckling of box-shaped aluminium
columns under axial compression. Thin-Walled Struct 2018;122:585–96. https://
members under uniform compression. Thin-Walled Struct 2021;164:107828.
doi.org/10.1016/j.tws.2017.11.003.
https://doi.org/10.1016/j.tws.2021.107828.
[5] Zhao Y, Zhai X, Wang J. Buckling behaviors and ultimate strength of 6082–T6
[17] Zhu JH, Young B. Aluminum alloy tubular columns—Part II: Parametric study and
aluminum alloy columns with square and circular hollow sections under eccentric
design using direct strength method. Thin-Walled Struct 2006;44(9):969–85.
compression–Part I: Experiments and finite element modeling. Thin-Walled Struct
https://doi.org/10.1016/j.tws.2006.08.012.
2019;143:106207. https://doi.org/10.1016/j.tws.2019.106207.
[18] North American Specification for the design of cold-formed steel structural
[6] Yuan L, Zhang QL, Ouyang YW. Experimental investigation and design method of
members. Washington, DC: American Iron and Steel Institute; 2001.
the flexural buckling resistance of high-strength aluminum alloy H-columns.
[19] GB/T 228.1-2010. Metallic materials-tensile testing- Part 1: Method of test at room
Structures 2021;35:1339–49. https://doi.org/10.1016/j.istruc.2021.11.013.
temperature. Beijing: Standards Press of China; 2010.
[7] Wang YQ, Wang ZX, Hu XG, Han JK, Xing HJ. Experimental study and parametric
[20] Ramberg W, Osgood WR. Description of stress-strain curves by three parameters.
analysis on the stability behavior of 7A04 high-strength aluminum alloy angle
Technical note 902. Washington, DC: National Advisory Committee for
columns under axial compression. Thin-Walled Struct 2016;108:305–20. https://
Aeronautics; 1943.
doi.org/10.1016/j.tws.2016.08.029.
[21] Ziemian RD, editor. Guide to Stability Design Criteria for Metal Structures, 6th ed.
[8] Hu XG, Cheng YF, et al. Tests on domestic 703 high-strength aluminum alloy angle
New York: John Wiley & Sons, Inc.; 2010.
members in axial compression. J Tongji Univ (Nat Sci) 2020;48:10.
[22] GB 5237.1-2008. Aluminum alloy extruded profiles for architecture – Part 1: Mill
[9] Wang Z, Wang Y, Yun X, Gardner L, Hu X. Experimental and numerical study of
finish profiles. Beijing: Standards Press of China; 2008.
fixed-ended high-strength aluminum alloy angle-section columns. J Struct Eng
[23] Mazzolani FM. Aluminium alloy structures. 2nd ed. London: E&FN Spon; 1995.
2020;146(10):04020206. https://doi.org/10.1061/(ASCE)ST.1943-
[24] Bebiano R, Camotim D, Gonçalves R. GBTUL 2.0 – a second-generation code for the
541X.0002773.
GBT-based buckling and vibration analysis of thin-walled members. Thin-Walled
[10] Zhu JH, Li ZQ, Su MN, Young B. Behaviour of aluminium alloy plain and lipped
Struct 2018;124:235–53. https://doi.org/10.1016/j.tws.2017.12.002.
channel columns. Thin-Walled Struct 2019;135:306–16. https://doi.org/10.1016/
[25] Aluminum Design Manual. Washington, DC: The Aluminum Association; 2020.
j.tws.2018.11.010.
[26] Bleich F. Buckling strength of metal structures. NewYork, NY: McGraw-Hill; 1952.
11