Professional Documents
Culture Documents
1 Introduction in the previous research by the authors of Ref. [9]. It seems that
the prediction based on the solution of Rusch and Lindner [10] is
Rectangular tubes are widely used as structural members in
reasonable compared with the results obtained from FEM. How-
numerous applications such as automobiles, trains, vehicles, air-
ever, in the solution of Rusch and Lindner [10], the plate is sup-
craft, offshore structure, etc. The primary collapses of thin-walled
ported at only one longitudinal edge with the other longitudinal
rectangular tubes are axial and bending collapses. Structural
edge being free. This boundary condition is obviously different
designers are often required to estimate the collapse load of rec-
from that of the tube web.
tangular tubes under axial and bending load.
This paper complements the previous work by addressing the
Although the real loading is seldom pure axial or bending, pure
postbuckling strength of the web under stress gradients. That is,
axial and bending collapses of square thin-walled tubes have been
apart from the solution based on the effective width concept given
subjected to extensive studies (for example, Refs. [1–7]). The
by Rusch and Lindner [10], in this paper the postbuckling strength
studies on collapse of tubes under pure axial or bending load give
of the web under stress gradients is also calculated using the effec-
fundamental understanding to collapse mechanism of the tubes
tive width concept given in AS/NZS 4600 standard [12] and NAS
under combined bending and compression.
[13] for a plate compressed with stress gradient and supported at
A commonly used method for predicting the collapse load or
both two longitudinal edges. The stress distributions on the tube
the maximum moment of rectangular tubes subjected to pure
web after buckling obtained from AS/NZS 4600 standard [12] and
bending was proposed by Kecman [5]. According to Kecman’s
NAS [13], and obtained from the solution of Rusch and Lindner
study, there are two types of collapses. The first type is a collapse
[10] are investigated by comparing with FEM numerical results.
due to buckling at the compression flange, and the second type is
The validity of the collapse load estimation is checked by the
a collapse due to plastic yielding at the flanges. However, when
results of FEM numerical analysis.
the web is wider than the flange, collapse due to buckling at the
Moreover, in Kecman’s method [5] the condition for the cross-
compression web may occur.
sectional fully plastic yielding of tubes under bending is depend-
In two recent papers [8,9], the authors investigated the maxi-
ent on the flange slenderness only. However, it is very likely that
mum moment at collapse of rectangular tubes subjected to bend-
the web slenderness also affects the cross-sectional fully plastic
ing. From these investigations, it is found that the collapse may
yielding when the web is wider. Therefore, in the present paper a
also occur due to the buckling of web in a rectangular tube under
new method is also proposed to predict the maximum moment
bending, when the tube has a cross section with a large aspect
considering the effect of web slenderness on the cross-sectional
ratio of web to flange b/a [8]. In order to evaluate the collapse
fully plastic yielding.
load of such tubes under bending, the effective width concept,
given by Rusch and Lindner [10], based on the study of Brune
[11], for a plate compressed with stress gradient was used to cal-
2 Numerical Analysis Method
culate the postbuckling strength of the web under stress gradients
In the present paper, thin-walled rectangular tubes under pure
1
Corresponding author.
bending are considered, and the FEM analysis software MSC.MARC
Manuscript received October 11, 2015; final manuscript received December 1, [14] is used in numerical simulation for obtaining the maximum
2015; published online December 21, 2015. Assoc. Editor: Nick Aravas. bending moment of tubes under bending. Figure 1 shows the
Fig. 2 Schematic representation of axial stress distribution used in Kecman’s method: (a)
Case 1: rbuc–a < rs; (b) Case 2: rs £ rbuc–a < 2rs; and (c) Case 3: rbuc–a 2rs
Fig. 4 Axial stress distribution on cross section at the maximum moment: (a) t 5 0.4 mm, a 5 50 mm, b 5 100 mm; (b)
t 5 0.5 mm, a 5 20 mm, b 5 100 mm; and (c) t 5 1.2 mm, a 5 50 mm, b 5 150 mm
be is given by
be ¼ q b (14)
1
q¼ (15)
k
which is proposed by von Karman et al. [20]. The following for-
mula for q:
1 0:22
q¼ 1 (16)
k k
is also proposed by Winter [21] and is well used for design
Fig. 6 Plate subjected to compression and bending: (a) ana- specifications. The reason of Eq. (15) modified to Eq. (16) in
lyzed model and (b) axial compressive stress rx distribution on actual design is mainly due to the fact that the maximum load
E–E cross section in (a) capacity of a buckling plate is reduced greatly by imperfec-
8 tions when the buckling stress is close to the yield stress [22].
> be Therefore, Eq. (15) is desirable for the present study because
>
> be1 ¼
>
< 3w the influence of imperfections is not taken into consideration
( (9) here. Moreover, in order to consider continuity of the load
>
> be =2 when w 0:236 capability of a web with k ¼ 1, for which elastic buckling does
>
: be2 ¼ b b
>
when w > 0:236 not occur because rbuc–b ¼ rs, we apply Eq. (15) to the present
e e1
study.
In addition, be1 þ be2 shall not exceed the compression portion of Equation (9) is applied to the webs investigated in Figs. 4(a)
the web. and 4(b) to determine the corresponding effective width; the stress
Here, be is given in Eq. (14) later, and w is ratio of f1 and f2 . f1 distributions on the web based on the obtained effective width
and f2 are web stresses shown in Fig. 6(b) using Eq. (9) are shown in Figs. 7(a) and 8(a). In Fig. 7(a), the
stress distribution obtained using Eq. (9) is qualitatively corre-
f2 sponding with the redistribution of the compression stress after
w¼ (10) buckling obtained from the FEM numerical analysis. However,
f1
in Fig. 8(a), even though there is a fall of the compression
k is defined by stress in the compression portion of the web after buckling as
shown by the FEM analysis, the stress distribution obtained
rffiffiffiffiffiffiffiffiffiffiffiffiffi from Eq. (9) looks like a straight line, because the effective
rs widths be1 and be2 determined by Eq. (9) satisfy the following
k¼ (11)
rbucb equation:
Fig. 7 Stress distribution of web when the ultimate load is reached for the tube used in
Fig. 4(a): (a) comparison with Eq. (9) and (b) comparison with Eq. (18)
In fact, when the effective width is determined using Eq. (9), According to the solution published in Ref. [10] the effective
there are many instances in which Eq. (17) is satisfied. Figure 9 widths be1 and be2 are given by
shows various possible values of buckling stress of web, for which 8
Eq. (17) is satisfied, for various assumed stress ratios w by solid >
< be1 ¼ be be2
line, as evaluated in Eq. (9) with q defined in Eq. (15). In Fig. 9, be2 0:226 (18)
the dashed line shows the corresponding result if q was calculated >
: ¼ 2
using Eq. (16); it is seen from the dashed line that even if Eq. (16) b k
is used for q the instances in which Eq. (17) is satisfied still exist.
For these instances, the redistribution of the compression stress where
after buckling cannot be expressed by the effective width obtained
from Eq. (9); this means that there is a possibility of giving a too be q
¼ (19)
large load capability of web from Eq. (9). Therefore, here as a b 1w
comparison we also use another solution given by Rusch and
Lindner [10] to evaluate the effective width of web after buckling Here, k and q are calculated by Eqs. (11) and (15), respectively,
[9]. The solution in Ref. [10] is given for the same plate shown in the buckling stress rbuc–b is determined by Eq. (12) with kb deter-
Fig. 6(a) but with one of the two longitudinal edges BC mined as follows:
being free. Although the free boundary condition at the longitudi-
nal edge BC is different from the actual situation of web constitut- kb ¼ 1:7 5w þ 17:1w2 (20)
ing the tube, the effect of the boundary condition at the edge BC
is assumed to be small, because the edge BC is under tension Figures 7(b) and 8(b) show the comparisons of stress distribu-
stress. tions on the web obtained from FEM and from Eq. (18) for the
tubes used in Figs. 4(a) and 4(b), from which it is seen that the
redistribution of stress after web buckling can be approximately
expressed using Eq. (18). Comparing (a) and (b) in Fig. 7, it is
seen that for the stress distribution on the web in the tube used in
Fig. 4(a), Eq. (18) is inferior in accuracy to Eq. (9). However, as
shown in Fig. 8, which shows the stress distributions on the web
for the tube used in Fig. 4(b), although the fall of the compression
stress in the compression portion of the web after buckling is not
expressed by the solution obtained from Eq. (9), it is expressed by
the solution from Eq. (18). In fact, it is seen from Eq. (19) that for
the stress distribution on the web as obtained from Eq. (18) the
length of be1 þ be2 is always smaller than the compression portion
of the web.
(
rbuca 2rs where teb is the web thickness for which the elastic buckling stress
(23) is equal to the yielding stress and is given by
rs < rbucb < 2rs
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffirffiffiffiffiffi
or 12ð1 2 Þ rs
teb ¼ b (29)
( kb p 2 E
rs < rbuca < 2rs
(24) Furthermore, we assume that this technique can also be used to
rs < rbucb < 2rs evaluate the maximum moment in the case when Eq. (24) is satis-
fied. That is, Mmax is determined from the smaller one from both
the stress on cross section is expressed by Case 2 shown in Fig. values of bs given in Eq. (26) and in Eq. (28). Validity of this
2(b). This fact can be confirmed from Fig. 4(c) for which Eq. (23) assumption can be understood from Fig. 10 shown later, in which
is satisfied. for the tube with t/a ¼ 0.016 Eq. (24) is satisfied: rbuc–a ¼ 1.23rs
It is seen from the cross-sectional stress distribution shown and rbuc–b ¼ 1.39rs. Therefore, when using Eq. (25) to determine
in Fig. 4(c) that the maximum moment in this case is depend- Mmax for Case 2, the value of bs is calculated using Eq. (26) if
ent on the plastic yielding region in the web. Denoting the
length of this plastic yielding region by bs (see Fig. 2(b)), the tea > teb (30)
Fig. 10 Flow chart of a new method proposed in the present study for predicting the
maximum moment of tubes under pure bending
for Case 4 :
Mmax w 2
1 w 3
for Case 5 :
Mmax w 2
1 w 3
(35)
¼ b þ d12 d22 þ b þ d13 d23 þ 2ab
rs t 2 3b
Fig. 11 Prediction of the maximum bending moment Mmax for In Figs. 11, 12, and 13, the maximum moment predicted by the
rectangular tubes with b/a 5 1 present method is compared with that obtained from the FEM
numerical analysis. As a result of the prediction method proposed
in the present paper, both the analyses using Eq. (9) and using
and is calculated using Eq. (28) if Eq. (18) for calculating the effective width are shown, respec-
tively; the former is denoted as “method-1” and the latter is
tea < teb (31) denoted as “method-2.”
The case number of the collapse corresponding to each thick-
Using tea and teb the condition of Eq. (21) can be rewritten as ness is also shown in the figures. In Case 2, there are two possible
pffiffiffi pffiffiffi subcases: (1) tea > teb as shown in Figs. 10 and 12 and (2) tea < teb
t 2tea and t 2teb (32)
as shown in Fig. 13; the maximum moment is determined by Eq.
(26) for the former and by Eq. (28) for the latter. As shown in
3.4 Estimation of Collapse Load for Thin-Walled Rectan-
these figures, Eqs. (26) and (28) give good prediction to the corre-
gular Tubes Under Bending. The present method for predicting
sponding subcase, respectively.
the maximum moment of tubes under pure bending is summarized
For Case 4 and Case 5, although each result obtained from the
and is shown by a flow chart in Fig. 10, from which it is seen that
method-1 and the method-2 is approximately in agreement with
both the possible buckling at web and the effect of web slender-
the analysis of FEM, it is found that there is a gap in the results
ness on the cross-sectional fully plastic yielding are taken into
between method-1 and method-2. When the buckling stress of the
account in the method. In the flow chart rbuc–b,1 and rbuc–b,2 are
web rbuc–b is close to the yielding stress rs the method-1 gives a
the buckling stress of web assuming the stress ratio w to be
too large prediction as compared with the FEM analysis, reflecting
b y1 ae þ b the fact that be1 þ be2 given by Eq. (9) may be equal to the com-
w¼ ¼ (33) pression portion of the web as shown in Fig. 9. However, for small
y1 aþb
Fig. 12 Prediction of the maximum bending moment Mmax for Fig. 13 Prediction of the maximum bending moment Mmax for
rectangular tubes with b/a 5 2 rectangular tubes with b/a 5 3