Professional Documents
Culture Documents
www.elsevier.com/locate/jcsr
Abstract
This paper describes the deformation limit and ultimate strength of welded T-joints in cold-
formed RHS sections. Both web buckling failure mode and chord flange failure mode are
investigated. The strength at a certain deformation (chord flange indentation) limit can be
regarded as the ultimate strength of a T-joint. The deformation limit mainly depends on the
ratio b (=b1/bo). Based on the test results of T-joints in cold-formed RHS sections, the defor-
mation limit is found as 3%bo for 0.6ⱕbⱕ0.8 or 2gⱕ15, and 1%bo for 0.3ⱕb⬍0.6 and 2g⬎15.
The ultimate strength so determined is compared with the existing design formulae. Proposed
formulae for ultimate strength of web buckling failure and chord flange failure are given.
2000 Elsevier Science Ltd. All rights reserved.
Keywords: Cold-formed steel; Deformation limit; Hollow sections; Ultimate strength; Welded joints
1. Introduction
0143-974X/00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 3 - 9 7 4 X ( 9 9 ) 0 0 0 6 3 - 2
150 X.-L. Zhao / Journal of Constructional Steel Research 53 (2000) 149–165
Nomenclature
Ns section capacity of web buckling
Pmax peak load
Pult ultimate load
P1%bo load at deformation of 1%bo
P3%bo load at deformation of 3%bo
PCIDECT predicted capacity using CIDECT model
Pkato predicted capacity using Kato model
Pzh1 predicted capacity using the modified Kato model
Pzh2 predicted capacity using the membrane mechanism model
bo chord width
b1 branch width
fy yield stress of RHS
ho chord depth
h1 branch width
rext external corner radius of RHS
to chord thickness
t1 branch thickness
ac reduction factor for web buckling capacity
b b1/bo
2g bo/to
was proposed by Lu et al. [20] to cover all types of welded tubular joints. It can be
summarised as:
앫 for a joint which has an obvious peak load at a deformation around 3%bo, the
X.-L. Zhao / Journal of Constructional Steel Research 53 (2000) 149–165 151
peak load or the load at 3%bo deformation is considered to be the ultimate load,
where bo is the width of the chord member as shown in Fig. 1;
앫 for a joint which does not have a pronounced peak load, the ultimate deformation
limit depends on the ratio of the load at 3%bo to the load at 1%bo. If the ratio is
greater than 1.5, the deformation limit is 1%bo, i.e. serviceability is in control.
The ultimate strength is taken as 1.5 times the load at 1%bo. If the ratio is less
than 1.5, the deformation limit is 3%bo, i.e. strength is in control. The ultimate
strength is taken as the load at 3%bo;
앫 a validity range of b (=b1/bo) and 2g (=bo/to) is given to determine whether the
design is governed by serviceability or by strength.
The proposal [20] was mainly based on tests of hot-rolled sections. There is a need
to verify the proposed deformation limit for welded T-joints in cold-formed RHS sec-
tions.
This paper describes the verification of the deformation limit using the test results
of welded T-joints in cold-formed RHS sections [2,7]. Both web buckling failure
mode and chord flange failure mode are investigated. Based on the test results, the
deformation limit is found to be 3%bo for 0.6ⱕbⱕ0.8 or 2gⱕ15 and 1%bo for
0.3ⱕb⬍0.6 and 2g⬎15. The ultimate strength of the web buckling is compared with
the existing design formulae given by Packer [21], Packer et al. [22] and Zhang et
al. [23]. The ultimate strength of chord flange failure is compared with the capacities
of T-joints determined using CIDECT design formula [22], the Kato model [24], the
modified Kato model [7] and the membrane mechanism model [19]. Proposed formu-
lae for the ultimate strength of web buckling failure and chord flange failure are
given, where the corners of cold-formed RHS sections are taken into account.
2. Experimental investigation
Tests on T-joints in cold-formed RHS sections were performed by Zhao and Han-
cock [7] in Australia and Kato and Nishiyama [2] in Japan. There are three main
failure modes, namely web buckling failure, chord flange failure and branch local
buckling failure. The failure mode of branch local buckling is similar to that observed
in a stub column test. This failure mode is not discussed in this paper since local
buckling can be prevented by using a plate slenderness which is lower than the plate
yield slenderness limit of RHS sections as reported by Zhao and Hancock [25] and
Hancock and Zhao [26]. A clear peak load is normally found for a web buckling
failure mode. The chord flange failure usually has a post-yield response due to the
effect of membrane forces in the chord and strain hardening of the material [19].
The section dimensions and material properties are summarised in Tables 1 and
2 for web buckling failure and for chord flange failure, respectively. The following
152 X.-L. Zhao / Journal of Constructional Steel Research 53 (2000) 149–165
Table 1
Dimensions and material properties (specimens with web buckling failure)
Table 2
Dimensions and material properties (specimens with chord flange failure)
For a joint which has a peak load (Pmax) at a deformation smaller than 3%bo the
peak load is considered to be the ultimate load (Pult), as shown in Fig. 2(a). For a
joint which has a peak load (Pmax) at a deformation larger than 3%bo the load at the
deformation limit 3%bo is considered to be the ultimate load (Pult), as shown in Fig.
2(b). The ultimate loads so determined are shown in Table 3 for T-joints with a web
buckling failure mode. The ultimate load (Pult) is compared with the peak load (Pmax)
obtained in the test. A mean ratio of 0.998 is reached with a small COV (coefficient
of variation) of 0.003. It can be concluded that for the web buckling failure mode
the 3%bo deformation limit for the ultimate strength proposed by Lu et al. [20]
applies to T-joints in cold-formed RHS sections with 0.8ⱕbⱕ1.0.
Table 3
Deformation limit and ultimate strength (web buckling failure)
From Lu et al. [20], the ultimate deformation limit depends on the ratio of the
ultimate load (P3%bo) to the serviceability load (P1%bo). If the ratio is less than 1.5,
the ultimate deformation limit is 3%bo, i.e. the strength is in control. The ultimate
strength is taken as P3%bo, as shown in Fig. 3(a). If the ratio is greater than 1.5, the
앫 for 0.6ⱕbⱕ0.8 or 2gⱕ15, strength is in control, i.e. the deformation limit is 3%bo
and Pult=P3%bo;
앫 for 0.3ⱕb⬍0.6 and 2g⬎15, serviceability is in control, i.e. the deformation limit
is 1%bo and Pult=1.5·P1%bo.
This agrees with the experimental observation that larger deformation is obtained
for joints with smaller b due to the effect of membrane forces in the chord mem-
ber [2,27].
Table 4
Deformation limit and ultimate strength (chord flange failure)
The existing formulae for web buckling of RHS T-joints are summarised in
Appendix A. The details can be found in Packer et al. [22], Packer [21] and Zhang
et al. [23]. The experimental ultimate strength (Pult) determined in Section 3 is com-
pared with predicted ultimate strength using the existing formulae. The comparison
is presented in Table 5. When the ratio is less than 1.0, it means that the formula
underestimates the capacity. When the ratio is larger than 1.0, it means that the
formula overestimates the capacity.
It can be seen that the CIDECT formula [22] underestimates the web buckling
strength. The formula by Zhang et al. [23] underestimates the web buckling strength
for b=1.0 and overestimates the web buckling strength for b⬍1.0. The formula by
Packer [21] gives the best prediction with a mean ratio of 1.056 and a COV of 0.181.
The following aspects are not considered in the existing formulae:
앫 rounded corners of cold-formed RHS sections in calculating the flat web depth;
앫 effect of b ratio which represents, to some extent, the influence of load eccen-
tricity; or
앫 effect of (ho⫺2rext)/to which represents, to some extent, the influence of column
slenderness.
In this paper the web buckling of RHS sections is treated as a column buckling
problem. The proposed model is shown in Fig. 1. The column length is assumed to
be (ho⫺2rext) where ho is the overall depth of the chord member and rext is the
X.-L. Zhao / Journal of Constructional Steel Research 53 (2000) 149–165 157
Table 5
Comparison of ultimate strength (web buckling failure)
external corner radius. This assumption is the same as that used in previous research
on web buckling of RHS sections under bearing forces [28–30]. The column area
is (h1+5rext)·to where h1 is the overall depth of branch member and to is the web
thickness. The column buckling strength can be expressed as
Pweb−buckling⫽ac·Ns (1)
where ac is a reduction factor and Ns is the section capacity, i.e.
Ns⫽2(h1⫹5rext)·to·fy (2)
in which fy is the yield stress of the chord member.
The reduction factor (ac) depends on the value of b and (ho⫺2rext)/to. The external
radius of corners (rext) is taken as 2.5t when the thickness of the tube is larger than
3 mm, otherwise the external radius of corners is taken to be twice the thickness
[31]. The value of b represents the effect of load eccentricity while the value of
(ho⫺2rext)/to represents the effect of column slenderness. The expression of ac can
be calibrated using the test results (Pult) in Table 3.
The ratio Pult/Ns (=ac) versus (ho⫺2rext)/to is plotted in Fig. 5 for specimens with
web buckling failure. It seems that the ratio Pult/Ns (=ac) is about 0.7 for b=1. The
ratio Pult/Ns (=ac) decreases as (ho⫺2rext)/to increases for 0.8⬍b⬍0.9. The simple
regression lines are plotted in Fig. 6. They are expressed as
ac⫽0.7 for b⫽1.0 (3)
158 X.-L. Zhao / Journal of Constructional Steel Research 53 (2000) 149–165
The existing formulae for chord flange failure of RHS T-joints are summarised
in Appendix B. They are all based on yield line mechanisms. The details can be
found in Packer et al. [22] for the CIDECT model, in Kato and Nishiyama [24] for
X.-L. Zhao / Journal of Constructional Steel Research 53 (2000) 149–165 159
Fig. 7. Experimental ultimate strength (Pult) versus proposed web buckling strength (ac·Ns).
the Kato model, in Zhao and Hancock [7] for the modified Kato model and in Zhao
and Hancock [19] for the membrane mechanism model. The experimental ultimate
strength (Pult) determined in Section 3 is compared with the predicted ultimate
strength using the existing formulae. The comparison is presented in Table 6.
It can be seen that the CIDECT model (PCIDECT) underestimates the chord flange
strength. The reasons for the conservatism of the CIDECT model are the effect of
membrane forces in the chord and the effect of local thickening of the corners on
the negative plastic moments along the corners [32]. The Kato model (Pkato) overesti-
mates the chord flange strength. This is most likely due to consideration of hinges
on the centre of the corners, which did not occur as a result of much higher yield
stress in the corners. Better results are obtained from the membrane mechanism
model (Pzh2). The modified Kato model (Pzh1) gives the best results. The predicted
strength (Pzh1) using the modified Kato model is plotted in Fig. 8 against the experi-
mental ultimate strength (Pult). Good agreement is obtained with a mean ratio
(Pzh1/Pult) of 1.049 and a COV of 0.091. The positions of hinges in the modified
Kato model are at the top of the web adjacent to the corners rather than at the centre
of the corners. It seems that the corners of cold-formed RHS sections should be
considered in predicting both web buckling strength and chord flange strength. The
modified Kato model is proposed to predict the chord flange strength.
160 X.-L. Zhao / Journal of Constructional Steel Research 53 (2000) 149–165
Table 6
Comparison of ultimate strength (chord flange failure)
Specimen Pult (kN) PCIDECT Pkato Pzh1 (kN) Pzh2 (kN) PCIDECT/Pult Pkato/Pult Pzh1/Pult Pzh2/Pult
label (kN) (kN)
S1B1C21 410 291 577 386 303 0.710 1.407 0.941 0.739
S1B1C22 137 125 208 166 141 0.912 1.518 1.212 1.029
S1B1C23 63 50.6 76.4 67.1 65.5 0.803 1.213 1.065 1.040
Kato5 76 45 126 80.3 65.6 0.592 1.658 1.057 0.863
Kato11 195 144 230 182 182 0.738 1.179 0.933 0.933
Kato12 255 185 301 234 246 0.725 1.180 0.918 0.965
Kato13 297 244 499 357 336 0.822 1.680 1.202 1.131
Kato15 569 389 781 564 483 0.684 1.373 0.991 0.849
Kato16 112 101 131 118 127 0.902 1.170 1.054 1.134
Kato17 158 131 171 152 182 0.829 1.082 0.962 1.152
Kato19 59 55.4 63.1 60.9 95.8 0.939 1.069 1.032 1.624
Kato21 305 263 333 303 340 0.862 1.092 0.993 1.115
Kato25 83 81.2 88.7 86.7 128 0.978 1.069 1.045 1.542
Kato26 97 101 131 118 127 1.041 1.351 1.216 1.309
Kato27 193 149 246 191 189 0.772 1.275 0.990 0.979
Kato32 216 212 242 233 239 0.981 1.120 1.079 1.106
Kato33 264 263 333 303 340 0.996 1.261 1.148 1.288
MEAN 0.840 1.276 1.049 1.106
COV 0.151 0.154 0.091 0.211
7. Conclusions
For the web buckling failure mode the 3%bo deformation limit for the ultimate
strength proposed by Lu et al. [20] applies to T-joints in cold-formed RHS sections
with 0.8ⱕbⱕ1.0.
For the chord flange failure mode:
앫 for 0.6ⱕbⱕ0.8 or 2gⱕ15, strength is in control, i.e. the deformation limit is 3%bo
and Pult=P3%bo;
앫 for 0.3ⱕb⬍0.6 and 2g⬎15, serviceability is in control, i.e. the deformation limit
is 1%bo and Pult=1.5·P1%bo.
A proposed web buckling formula (see Eq. (1)) has been given, which considers
the rounded corners of cold-formed RHS sections, the effect of b ratio and the effect
of (ho⫺2rext)/to. The section capacity Ns is given in Eq. (2). The reduction factor ac
is given in Fig. 6 or in Eqs. (3) and (4).
The modified Kato model has been found to give the best prediction for chord
flange failure.
It can be concluded that the corners of cold-formed RHS sections should be con-
sidered in predicting both web buckling strength and chord flange strength.
X.-L. Zhao / Journal of Constructional Steel Research 53 (2000) 149–165 161
Fig. 8. Experimental ultimate strength (Pult) versus proposed chord flange strength (Pzh1).
For b=1.0
PCIDECT⫽fy·to·(2h1⫹10to)
For bⱕ0.85
PCIDECT⫽ 冉
fy·t2o h1
1−b bo
冑
· 2 ⫹4 1−b 冊
For 0.85⬍b⬍1.0 use linear interpolation.
Ppacker⫽fy·b0.3 1.7
冋
o ·to · 3.8⫹10.75 冉 冊册
b1+h1
2bo
2
162 X.-L. Zhao / Journal of Constructional Steel Research 53 (2000) 149–165
Pzhang⫽2fy·to·h1e·k1
in which
k1⫽1.75⫺0.030·ho/to when ho/toⱕ25
k1⫽1.40⫺0.016·ho/to when ho/to⬎25
h1e⫽h1·k2
k2⫽(0.7·ho/h1)0.7 when h1/hoⱕ0.7
k2⫽(0.7·ho/h1)0.2 when h1/ho⬎0.7
PCIDECT⫽ 冉
fy·t2o h1
1−b bo
· 2 ⫹4 1−b 冑 冊
where b=b1/bo.
Pkato⫽
1−·
fy·t2o
b1+2·to
bo−1.88·to
冢
· 2·
h1+2·to
bo−1.88·to
⫹4 冪1−·b −1.88·t
b1+2t·o
o o 冣
B.3. Formula by Zhao and Hancock [7] (modified Kato model)
Pzh1⫽
1−·
fy·t2o
b1+2·to
bo
2·
冢
h1+2·to
bo
⫹4 冪1−·
b1+2·to
bo 冣
X.-L. Zhao / Journal of Constructional Steel Research 53 (2000) 149–165 163
冘
9
Pzh2⫽Pm⫹ kiPi
i⫽3,5
Pm⫽2·Sy·sinay
Sy⫽0.5·(bo⫹b1)·to·fy
冪1−·(1+e )
1
sinay⫽ 2
y
Table 7
Terms ki and Pi
Yield line type no. Number of yield Contribution of each ith type yield line (Pi)
(i) lines (ki)
3 2
P3⫽Mp· 冉冊
h1
n
5 2
P5⫽Mp· 冉 冊冉 冊
h1
·
n 1−e
1
冉冊
冢 冣
6 4 ho 2 2
e2 · +K
bo
冉冊
P6⫽Mp·
n
K·
bo
冉冊 冉 冊
冢 冉冊 冣
7 4 h1 2 2
(1−e)2 +K
bo e
P7⫽Mp·
n 1−e
K·
bo
8 4
冢 冉 冊冣 冉 冊
2
e ho
P8⫽Mp· ·
n bo
K
bo
9 2
P9⫽Mp· 2K冉 冉 冊 冊冉 冊
bo h1
n
⫹
e
n 1−e
164 X.-L. Zhao / Journal of Constructional Steel Research 53 (2000) 149–165
ey⫽fy/E
where E is taken as 200,000 MPa.
The terms ki and Pi are given in Table 7 in which
fy·t2o
M p⫽
4
e⫽ 冉冊n
ho
·b
bo·(1−b)
n⫽
2
b⫽b1/bo
冑
K⫽0.5·D21·D3⫹0.5D1· D21·D23+4·D2
冑 冪1+e
1−e
D1⫽ 1−b·
D2⫽ 冉 冊冉 冊
e·ho ho
n
·
bo
D3⫽ 冉 冊冉 冊冑
bo 1+b
to
·
2
· (1+ey)2−1
References
[1] Mouty J. Calus des changes ultimes des assemblages soudes de profils creux carres et rectangularies.
Construction Metallique 1976;2:37–58.
[2] Kato B, Nishiyama I. The static strength of R.R.-joints with large b/b ratio. CIDECT prog. 5y,
Department of Architecture, Faculty of Engineering, University of Tokyo (Tokyo, Japan), 1979.
[3] Stark JWB, Soetens F. Welded connections in cold-formed sections. In: Proceedings of the 5th
International Specialty Conference on Cold-Formed Steel Structures, St. Louis (MI, USA), 1980.
[4] Packer JA, Davies G, Coutie M. Ultimate strength of gapped joints in RHS trusses. J Struct Div,
ASCE 1982;ST2:411–31.
[5] Panjehshahi E. The behaviour of RHS tee joints under axial load and bending moment. Master
Thesis, Nottingham University, UK, 1983.
[6] Labed A. Membrane action in steel hollow section welded joints. Master Thesis, Nottingham Univer-
sity, UK, 1989.
[7] Zhao XL, Hancock GJ. T-joints in rectangular hollow sections subject to combined actions. J Struct
Engng, ASCE 1991;117(8):2258–77.
[8] Lu LH, Puthli RS, Wardenier J. The static strength of uniplanar tubular X-joints loaded by in-plane
and out-of-plane bending. Stevin Report 25.6.91.28/A1, Delft Univ. Tech., 1991.
[9] van der Vegte GJ, Lu LH, Puthli RS, Wardenier J. The ultimate strength and stiffness of uniplanar
tubular steel X-joints loaded by in-plane bending. In: Proceedings of the 1st World Conference on
Construction Steel Design, Mexico, 1992.
X.-L. Zhao / Journal of Constructional Steel Research 53 (2000) 149–165 165
[10] de Winkel GD, Rink HD, Puthli RS, Wardenier J. In: The behaviour and the static strength of
unstiffened I-beam to circular column connections under multiplanar in-plane bending moments.
Proceedings of the 3rd International Offshore and Polar Engineering Conference, ISOPE-93, Singa-
pore, 1993.
[11] de Winkel GD, Rink HD, Puthli RS, Wardenier J. In: The behaviour and static strength of plate
to circular column connections under multiplanar axial loadings. Tubular structures V. London: E&
FN Spon, 1993:703–11.
[12] de Winkel GD, Wardenier J. In: Parametric study on the static behaviour of I-beams to tubular
column connections under in-plane bending moments. Tubular structures VI. Rotterdam: Balkema,
1994:317–24.
[13] Yu Y, Liu DK, Puthli RS, Wardenier J. In: Numerical investigation into the static behaviour of
multiplanar welded T-joints. Tubular structures V. London: E&FN Spon, 1993:732–40.
[14] Yu Y, Wardenier J. In: Influence of the types of welds on the static strength of RHS T- and X-
joints loaded in compression. Tubular structures VI. Rotterdam: Balkema, 1994:597–605.
[15] Lu LH, Puthli RS, Wardenier J. In: Ultimate deformation criteria for uniplanar connections between
I-beams and RHS columns under in-plane bending. Proceedings of the 4th International Offshore
and Polar Engineering Conference, ISOPE-94, Osaka (Japan), 1994.
[16] Mouty J. In: Theoretical prediction of welded joint strength. Proceedings of the International Sym-
posium on Hollow Structural Sections, Toronto (Canada), 1977.
[17] Yura JA, Zettlemoyer N, Edwards IF. Ultimate capacity equations for tubular joints. OTC Proc
1980;1:3690.
[18] Korol RM, Mirza FA. Finite element analysis of RHS T-joints. J Struct Engng, ASCE
1982;108(9):2081–98.
[19] Zhao XL, Hancock GJ. Plastic mechanism analysis of T-joints in RHS under concentrated force. J
Singapore Struct Steel Soc 1991;2(1):31–44.
[20] Lu LH, de Winkel GD, Yu Y, Wardenier J. In: Deformation limit for the ultimate strength of
hollow section joints. Tubular structures VI. Rotterdam: Balkema, 1994:341–7.
[21] Packer JA. Web crippling of rectangular hollow sections. J Struct Engng, ASCE
1984;110(10):2357–73.
[22] Packer JA, Wardenier J, Kurobane Y, Dutta D, Yeomans N. Design guide for RHS joints under
predominantly static loading. Köln (Germany): Verlag TÜV Rheinland, 1992.
[23] Zhang ZL et al. Nonlinear FEM analysis and experimental study of ultimate capacity of welded
RHS joints. In: Proceedings of the 3rd International Symposium on Tubular Structures, Lappeenranta
(Finland), 1989.
[24] Kato B, Nishiyama I. T-joints made of rectangular tubes. In: Proceedings of the 5th International
Conference on Cold-Formed Steel Structures, St. Louis (MI, USA), 1980.
[25] Zhao XL, Hancock GJ. Tests to determine plate slenderness limits for cold-formed rectangular hol-
low sections of grade C450. J Australian Inst Steel Construct 1991;25(4):2–16.
[26] Hancock GJ, Zhao XL. Research into the strength of cold-formed tubular sections. J Construct Steel
Res 1992;123:55–72.
[27] Zhao XL. The behaviour of cold-formed RHS beams under combined actions. Ph.D Thesis, The
University of Sydney, Sydney, 1992.
[28] Zhao XL, Hancock GJ. Square and rectangular hollow sections subject to combined actions. J Struct
Engng, ASCE 1992;118(3):648–68.
[29] Zhao XL, Hancock GJ. Square and rectangular hollow sections under transverse end bearing force.
J Struct Engng, ASCE 1995;121(11):1565–73.
[30] Zhao XL, Hancock GJ, Sully RM. Design of tubular members and connections using amendment
No. 3 to AS4100. J Australian Inst Steel Construct 1996;30(4):2–15.
[31] AISC Design capacity tables for structural steel hollow sections. Sydney (Australia): Australian
Institute of Steel Construction, 1992.
[32] CIDECT The strength and behaviour of statically loaded welded connections in structural hollow
sections, monograph No. 6. Corby (UK): CIDECT, 1986.