Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Design Strength of Locally Buckling Stub-Lipped

Channel Columns
M. V. Anil Kumar1 and V. Kalyanaraman2
Downloaded from ascelibrary.org by CALIF INSTITUTE OF TECHNOLOGY on 05/14/13. Copyright ASCE. For personal use only; all rights reserved.

Abstract: The compressive strength of cold-formed steel (CFS) lipped channel (LC) members may be governed by yielding, local, distor-
tional, or overall buckling, and any possible interaction among these modes. The direct strength method (DSM) has been advanced recently
for evaluating the strength of CFS LC beams and columns. Although the DSM is an improvement over other methods in terms of simplicity
and accuracy, further improvement by minor modification to the DSM is possible if all the parameters that affect the strength of such members is
properly accounted for and understood. In this study, the DSM equations for evaluating the strength of members that fail after experiencing only
local buckling are discussed. The strength of such members, according to the DSM, is a function of only the ratio of yield strength of the section
(Py) to the elastic local buckling load (Pcr[) (also equal to the reciprocal of square of the nondimensional local buckling slenderness ratio, 1/l[2).
This study indicates that the relative area of the stocky elements of the cross section, which are less vulnerable to elastic local buckling, also
influences the strength of such members. Using the experimental results available in the literature and the finite-element analysis data generated
in this study, the behavior and strength of stub LC compression members experiencing only local buckling before failure by yielding are
evaluated. Simple modifications to the currently available DSM equations are suggested to more accurately evaluate the strength of such
members. DOI: 10.1061/(ASCE)ST.1943-541X.0000575. © 2012 American Society of Civil Engineers.
CE Database subject headings: Cold-formed steel; Columns; Beams; Buckling.
Author keywords: Cold-formed steel; Lipped channel; Stub Column; Local buckling; Direct strength method.

Introduction buckling disregard some of the parameters that affect their strength.
This conclusion is based on the available experimental results in the
Cold-formed steel (CFS) lipped channel (LC) compression mem- literature and the extensive data generated using finite-element
bers (Fig. 1) are vulnerable to local buckling (L), distortional analysis (FEA). Simple modifications to the DSM equations are
buckling (D), overall buckling (E), or an interaction of these modes suggested for a better strength evaluation of such members.
before the member fails by yielding. Studies on the interaction
between local and distortional buckling (LD interaction) (Kwon and
Hancock 1992; Yang and Hancock 2004; Kwon et al. 2005; Schafer Stub Column Strength after Local Buckling
2002; Silvestre et al. 2006) and the local, distortional, and overall
buckling (LDE interaction) (Kwon et al. 2009) have been reported Effective section methods [using the effective width and/or
in literature. thickness (EWT) at the ultimate load] [(AISI 2007) and ENV
The direct strength method (DSM) has been recently introduced 1993-1-3 (European Committee for Standardization [CEN] 2004)]
in the North American Specification (NAS) [American Iron and have been traditionally used to evaluate the strength of members
Steel Institute (AISI) 2007] and Australian/New Zealand Standard after local and/or distortional buckling. These methods are tedious,
AS/NZS 4600:2005 (Standards Australia 2005) as a simpler and iterative, and do not account for the interaction between the dif-
more accurate design method for the CFS beams and columns than ferent elements of the cross section in evaluating the elastic (local
the prevailing effective width method (EWM). In the DSM, and distortional) buckling strengths of the member. As a way of
equations to evaluate the strength of members failing by yielding addressing these shortcomings of the EWM, the DSM has been
after any one of the different independent buckling modes (L, D, or recently introduced in the NAS (AISI 2007) and AS/NZS 4600:2005
E) are also presented. This paper shows that the DSM equations for (Standards Australia 2005) for the design of CFS columns and
the strength of stub columns failing by yielding after only local beams based on the work of Schafer and Peköz (1998b), Hancock
et al. (1994), Schafer (2002), and Schafer (2008). The DSM equation
for the strength of the stub LC columns after only local buckling
1
Ph.D. Research Scholar, Structural Engineering Division, Dept. of Civil (referred to as DSM-LB equation hereafter) is a function of the ratio
Engineering, Indian Institute of Technology Madras, Chennai 600036, of the elastic local buckling load, Pcr[ (and not the critical elastic
India. buckling stress of the most slender hinged edge elements of the
2
Professor, Structural Engineering Division, Dept. of Civil Engineering, cross section, as in the EWM), to the gross area yield load, Py, which
Indian Institute of Technology Madras, Chennai 600036, India (correspond- is also the reciprocal of the square of nondimensional local buckling
ing author). E-mail: kalyan@iitm.ac.in
slenderness ratio, l[. The Pcr[ may be evaluated using the numerical
Note. This manuscript was submitted on August 13, 2011; approved on
January 30, 2012; published online on February 1, 2012. Discussion period
analysis software [FEA or CUFSM (Schafer and Adany 2006)] or
open until April 1, 2013; separate discussions must be submitted for the closed form equations [Appendix 1 of BS5950 Part5 (British
individual papers. This paper is part of the Journal of Structural Engi- Standards Institute 1987) and NAS (AISI 2007) commentary]. In the
neering, Vol. 138, No. 11, November 1, 2012. ©ASCE, ISSN 0733-9445/ DSM, the ultimate strength of the stub members fail after only local
2012/11-1291e1299/$25.00. buckling, Pu[,DSM, is given by

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / NOVEMBER 2012 / 1291

J. Struct. Eng. 2012.138:1291-1299.


Downloaded from ascelibrary.org by CALIF INSTITUTE OF TECHNOLOGY on 05/14/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Lipped channel

For l[ # 0:776; Pu[DSM ¼ Py ð1aÞ

"  0:4 # 0:4


Pcr[ Pcr[
For l[ . 0:776; Pu[DSM ¼ 1 2 0:15 Py
Py Py
ð1bÞ

From among the available test data on the LC stub compression


members in literature, only 32 test results (Lim 1985; Mulligan and
Peköz 1987; Weng and Peköz 1988; Hancock et al. 1994; Zaras and
Rhodes 1987), in which distortional buckling was not likely to have
occurred, were included in this study. This is based on the value of
the ratio of the elastic distortional buckling load, Pcrd, (computed
using generalized beam theory, which accounts for end-boundary
condition) to the ultimate strength after local buckling (Pcrd /Pu[DSM)
being greater than 3.20 [which is calculated by setting √(Pu[DSM / Fig. 2. Evaluation of DSM-LB equation for strength after local buckling
Pcr , 0.561)]. Such specimens (32 in number) also had their re-
spective postdistortional buckling, Pud, (AISI 2007) greater than the
postlocal buckling strength, Pu[DSM [Eqs. (1a) and (1b)]. Further, the The conceptual reasoning for these trends in the results can be
values of the nondimensional overall buckling slenderness ratio, described as follows.
le, of these specimens were less than 0.20 to avoid the interaction 1. The influence of the rotational compatibility at the web-flange
of the overall buckling with the local buckling. The mean (m) and junction in the postlocal buckling range may be different from
SD (s) of the ratio of the ultimate strength of these specimens that at the theoretical elastic local buckling stage (as reflected
from the tests (Putest) to that obtained using the DSM (PulDSM) in Pcr[). The rotational compatibility (the restraint provided by
[Eqs. (1a) and (1b)] are 1.011 and 0.109, respectively. The DSM the stocky element to the buckling element) is a function of the
predicts the strength of the CFS-LC compression members fail h/b ratio in the LC sections. Hereafter, this is referred to as
only after the local buckling is unsafe by as much as 17% and rotational compatibility effect. Results of a nonlinear FEA
conservatively by as much as 29% compared with the test results carried out under this study on plates with a long-aspect ratio
[Figs. 2(a and b)]. The strength of such members based on the and with two extreme hinged and fixed longitudinal bound-
DSM is conservative in the majority of the cases compared with aries, but having the same elastic local buckling strength (by
the test results. adjusting the respective plate width to thickness ratio appro-
The strength ratios (Putest/Pu[DSM) of all these specimens are priately), had ultimate strength Pu[ as different as 10% for
plotted with respect to the nondimensional local buckling slender- plates with higher slenderness (Kumar 2012).
ness ratio, l[, in Fig. 2(a) and with respect to the ratio of the height of 2. In the postlocal-buckling range, the reduction in the effective-
the web to the width of the flange, h/b, in Fig. 2(b). The following ness of the stocky plate element caused by the local buckling
trends are observed in these figures. would be more and more different from that of the slender plate
• Pu[DSM is generally unsafe with respect to Putest for lower values element, because the value of the h/b ratio changes more and
of l[ and is conservative for higher values of l[ [Fig. 2(a)]. more away from 1.0 (on the higher and lower side), even if
• Pu[DSM is generally unsafe with respect to Putest for values of h/ the elastic local buckling strength of the member, used in the
b closer to 1.0 and is conservative for values of h/b much higher DSM, is maintained. Hereafter, this effect is referred to as the
than and much lower than 1.0 [Fig. 2(b)]. influence of the postbuckling strength of stocky element
• The largest variation in the ratio of Pu[DSM/Putest (varying (PBSS) effect. An evaluation carried out under this study
between 0.8 and 1.2) is in the range of 1.0 , h/b , 2.0, wherein indicated that the PBSS effect can cause differences as high
the elastic local buckling strength, Pcr[, is closer to that corre- as 35% between the values of strength of the member after only
sponding to the hinged boundary conditions of the elements local buckling obtained from the DSM and theoretical values
[Fig. 2(b)]. (Kumar 2012).
• Pu[DSM of the LC compression members exhibits systematic Thus, although the DSM, which simplified the design of CFS
error compared with the available test results [Figs. 2(a and b)]. members, is an improvement over the EWM, it appears that further

1292 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / NOVEMBER 2012

J. Struct. Eng. 2012.138:1291-1299.


improvements to the accuracy of the method are possible, by a sys- considering both the material and geometric nonlinearities. The
tematic study of the DSM equations for evaluating the ultimate maximum amplitude of imperfections of 0.34 and 0.94 times the
strengths of CFS members failing only after any one of the different thickness of the elements, corresponding to the first local and dis-
buckling modes. Bambach (2010) also pointed out that single tortional buckling modes from the eigenvalue buckling analysis,
nondimensional slenderness (l[) or buckling stress (scr) would not respectively, were applied in the nonlinear analysis model. This was
be adequate to represent the strength of the section. This paper, based accomplished using the IMPERFECTION feature in ABAQUS
on a detailed evaluation of the parameters, suggests simple mod- (2007). These maximum imperfection amplitude values corresponded
ifications to the DSM to obtain a better accuracy for the ultimate to the average values of the measured imperfections reported by
strength of stub LC compression members failing by yielding after Schafer and Peköz (1998a). The overall bow imperfection was
only local buckling (here after referred to as Pu[). The available ex- neglected as not being relevant in the stub column analysis.
perimental data in literature, augmented by a large volume of nonlinear From the calibration study, comparison of the average stress at
Downloaded from ascelibrary.org by CALIF INSTITUTE OF TECHNOLOGY on 05/14/13. Copyright ASCE. For personal use only; all rights reserved.

FEA-based data generated in this study, is used for this purpose. the ultimate load (su[ 5 Pu[ /A), as obtained from FEA, corre-
sponding to the 10 stub column experimental results, chosen from
the literature, is presented in Table 1, along with the values from the
Details of the Finite-Element Analysis tests. The comparison is good, with the maximum difference in the
average stress at the ultimate load [(su[Exp/su[FEA 2 1) 3 100]
ABAQUS (2007) version 6.7 was used to generate strength data by being 9.7%. The mean (m) and the SD (s) of the percentage dif-
nonlinear FEA of the fixed-ended CFS-LC stub columns, which fail ference of all 10 results are 2.7 and 4.8, respectively. This validated
by yielding after only local buckling. Initially, to calibrate the FE the adequacy of FEA to generate data on Pu[. Using the FEA,
model, a few available experimental results of compression mem- a large volume of data on Pu[ was generated by systematically
bers failing after different buckling modes were analyzed. Sub- varying the relevant parameters. The details are presented in the
sequently, the FEA was used to systematically study the full range next section.
of parameters that may affect the value of Pu[ and generate data for
LC sections failing by yielding after local buckling only.
LC members were modeled using four nodded shell elements Multipoint Constraint in FEA to Prevent Distortional
with reduced integration (S4R elements) along the mid-thickness and Overall Buckling
of the plate elements of the cross section. The minimum number of
elements used along the width direction was eight for both the Theoretically, only local buckling before failure of the thin-walled
flanges and the web, whereas a minimum of three elements was compression member by yielding can be achieved by designing
used for the lips. To avoid stress concentration at ends because the members, such that the elastic overall buckling load, Pcre, and the
of the application of load and/or boundary conditions, three- elastic distortional buckling load, Pcrd, are much higher than Pu[.
dimensional (B33) beam elements with high-bending stiffness in However, the section dimension parameters affecting the Pcr[ and
the plane of the shell were provided, connecting all the nodes Pcrd are not that independent. Hence, designing experimental
along the edges of shell element at the two ends of the member. specimens with a sufficiently larger value of Pcrd/Pcr[ to avoid the
The same beam elements were used to connect every node at the distortional buckling before failure by yielding or local buckling
two ends of the member to the nodes at the centroid of the end cross over the full range of l[ values is difficult.
section. The axial load and/or deformation and end-boundary However, in FE modeling and analysis, initiation of only local
conditions were imposed at the node at centroid at the two end buckling of CFS compression members without either the distor-
sections. All of the beam elements were assigned relatively small tional or overall buckling before failure by yielding, can be achieved
values for axial stiffness to avoid restraint to Poisson effects at the through appropriate artificial constraints for suppressing both the
two ends. overall and distortional buckling deformations. The distortional
Fig. 3 shows a typical FE mesh used in the analysis, finalized after buckling was suppressed by using the artificial multipoint constraint
a convergence study. The nodes at the junction of the flange and web (MPC) feature in ABAQUS (2007). For this purpose, the flange-lip
are restrained from displacement out of plane to preclude overall junction was constrained to remain straight (in Fig. 3, setting the
buckling, because this study is concerned with the behavior of stub displacement in the y direction of all the nodes along the corner 1 and
columns. The analysis was performed in two stages. First, the ei- the corner 4 to be equal to the displacement in the y direction of the
genvalue buckling analysis was done using the geometry of the perfect corner 2 and the corner 3, respectively). To avoid the end-boundary
member, to identify the local and/or distortional buckling modes. influence on local buckling, the specimen length was kept larger than
Following this, nonlinear analysis was performed on the same model, three times the width of the largest plate element of the cross section.
The possible overall buckling deformation of the member in the x, y
directions and twisting of the member was suppressed by applying
artificial constraints (in Fig. 3, against displacements in both the x
and y directions at all the nodes along corner 1 over the entire length
of the member and in the x direction only at all the nodes over the
entire length along corner 3). These constraints prevent the overall
buckling, but allow the Poisson expansion of the web of the section.
To confirm that the overall buckling and the distortional buckling
restraints in the preceding model (1) do not affect the value of Pu[,
and (2) prevent the distortional and overall buckling in members,
which are also vulnerable for these modes of buckling before the
failure by yielding, the following studies were conducted.
A few sections with Pcrd/Pcr[ . 4.00 were analyzed using
Fig. 3. FEA model
ABAQUS (2007) with and without the MPC as previously discussed.
The details of the specimens and the FEA results of the specimens

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / NOVEMBER 2012 / 1293

J. Struct. Eng. 2012.138:1291-1299.


Table 1. Comparison of FE Results with Tests
su 5 Pu/A (MPa)

%
References Specimen l[ ld lc Test FEA Diff.
Talja (1990) C90 3 70 3 15 3 2.4 0.902 0.844 0.509 400.0 373.1 6.7
Talja (1990) C90 3 70 3 15 3 2.5 0.981 0.846 0.823 372.5 354.4 4.9
Kwon and Hancock (1992) CH1–5-800 2.714 3.548 0.402 147.8 153.3 23.7
Lim (1985) C2 1.723 1.661 0.241 108.8 114.4 25.2
Young and Rasmussen L36F0280 1.376 0.804 0.177 360.8 364.3 21.0
(1998)
Downloaded from ascelibrary.org by CALIF INSTITUTE OF TECHNOLOGY on 05/14/13. Copyright ASCE. For personal use only; all rights reserved.

Hancock et al. (1994) CH20–300 0.737 0.37 0.108 213.4 192.7 9.7
Lim (1985) C6 1.842 0.306 0.168 156.0 154.6 1.0
Weng and Peköz (1988) RFC13–2 0.566 0.526 0.135 376.8 354.6 5.9
Young and Rasmussen L48F1500 1.588 1.274 0.794 320.1 304.1 5.0
(1998)
Lau and Hancock (1990) CH20–1900 0.671 0.755 0.719 178.6 172.8 3.3
Mean (m) 2.7
SD (s) 4.8

Table 2. Influence of Restraints in FEA on the Strength of Members Failing Only after Local Buckling
Pu-FEA (kN)

S. No H (mm) b (mm) d (mm) t (mm) fy (MPa) ld Without restraint With restraint Percentage difference
1 79.31 78.41 19.89 0.893 297.6 0.495 38.7 39.8 22.8
2 79.85 81.71 17.22 1.194 230.1 0.502 52.3 53.8 22.8
3 151.41 81.94 17.30 2.578 230.1 0.537 60.7 60 1.2

with and without the restraints are presented in Table 2. The effect of Table 3. Influence of Restraint in FEA on Members Vulnerable for
the constraint has negligible impact on the Pu[. This analysis proves Distortional and Overall Buckling
that the overall and distortional buckling restraints do not appre- l[ d (mm) T (mm) Py (kN) Pu,FEA/Py
ciably influence the value of Pu[.
To ensure the adequacy of the MPC, as previously discussed, in 1.00 30.0 1.9120 172.4 0.741
preventing distortional and overall buckling in members that are also 25.0 1.9100 167.5 0.739
vulnerable to such buckling modes, LC members with web depth 20.0 1.9120 162.9 0.737
equal to 100 mm, flange width of 100 mm, lip depth varying between 15.0 1.9120 158.1 0.734
10 and 30 mm, and yield strength equal to 250 MPa were analyzed 10.0 1.9160 153.6 0.734
using ABAQUS (2007). The thickness was chosen in each case to 2.50 30.0 0.7585 68.4 0.395
obtain two values (1.00 and 2.50) of the nondimensional local 25.0 0.7584 66.5 0.397
buckling slenderness ratios, l[ (Table 3). The details of the member 20.0 0.7583 64.6 0.402
are given in Table 3, along with the FEA results under the artificial 15.0 0.7584 62.7 0.407
restraint. In the absence of or ineffectiveness of the artificial restraint, 10.0 0.7584 60.8 0.394
such members of constant section dimensions, except the lip size,
should have varying nondimensional ultimate strength from FEA
(PuFEA/Py) because the ultimate strength would be drastically af- evaluating the nondimensional strength, Pu[ /Py. One of the aims of
fected by the distortional buckling caused by the varying lip size. this study was to find the influence of parameters, other than the
However, as seen in Table 3, despite a large variation in the lip size, nondimensional elastic local buckling slenderness ratio, l[, such as
the PuFEA/Py remains nearly the same. This demonstrates that the the actual section dimensions, the yield strength of the material (fy),
artificial restraints are effective in preventing the distortional and and the web depth to flange width ratio (h/b) on the normalized ul-
overall buckling in members that are also vulnerable to such timate strength, Pu[ /Py, of the LC stub columns, which fail by yielding
buckling modes before failure. Fig. 4 shows that the load versus after only local buckling. For the different values of the studied
deformation behavior of the members with different lip sizes (d) is parameters, the thickness of the specimen (even if nonstandard) was
essentially the same as long as the other dimensions are kept the chosen by trial and error, to obtain results for different specific desired
same. The small variations are caused by the local buckling effects values of l[, based on the golden ratio search optimization algorithm
of the lip of varying d/t. linked to ABAQUS (2007), using the Python script.

Effect of Actual Section Dimensions


Evaluation of the Parameters Affecting Pu[
To study the effect of the actual dimensions of the cross section on
Currently, the nondimensional elastic local buckling slenderness ra- the nondimensional ultimate strength, Pu[/Py, of stub LC members
tio, l[, is the only parameter included in the DSM-LB equation for failing by yielding after only local buckling for a constant value of

1294 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / NOVEMBER 2012

J. Struct. Eng. 2012.138:1291-1299.


the yield stress equal 250 MPa, and equal size of the flange width and l[ alone, as assumed in the DSM. In this study, the values of h and
the web depth (h/b 5 1.0), and for each of the two different values of b were chosen so that the h/b values were 0.33, 0.5, 0.67, 1.0, 1.5,
l[ (1.0 and 2.0), LC sections with different dimensions were ana- 2.0, and 3.0. This was obtained by choosing the h in the range of
lyzed. The results of the ultimate strength obtained from FEA are 20e200 mm, b in the range of 30e200 mm, and d in the range of
presented in Table 4. For the LC stub columns that fail after only 7.5e18 mm. The material yield strength was taken equal to 250
local buckling, for a constant value of the nondimensional local MPa. The wall thickness of the specimen was chosen in the range of
buckling slenderness ratio, l[, the ratio of the ultimate strength to the 0.189e1.522 mm (even if nonstandard), so that l[ values were equal
yield strength (Pu[/Py) remains essentially the same for the speci- to 0.5, 1.0, 1.5, 2.0, 2.5, and 3.0. Any possible distortional and
mens with very different cross-section dimensions. This indicates overall buckling in the sections was prevented in the FEA by using
that the ratio of the ultimate strength to the yield strength (Pu[/Py) is the MPC as previously discussed. Among these, the specimens with
independent of the actual values of the section dimensions. h/b , 1.0 (b/h 5 1.5, 2.0, and 3.0) are not prequalified, as per NAS
Downloaded from ascelibrary.org by CALIF INSTITUTE OF TECHNOLOGY on 05/14/13. Copyright ASCE. For personal use only; all rights reserved.

(AISI 2007). All the other specimens are prequalified.


Effect of Yield Stress The variation of the ratio of the ultimate strength obtained from
FEA to that evaluated by the DSM (Pu[FEA/Pu[DSM) with respect to
To study if the normalized ultimate strength (Pu[/Py) is explicitly the h/b ratio is plotted in Fig. 5. The plots are made for different
affected by fy other than through l[, FEA was conducted on LC 38 3 distinct values of l[. The results show that there is systematic
38 3 7 members (Fig. 1). The yield strength varied from 165 to 400 variation in the ratio of the strengths from FEA to DSM for different
MPa. Corresponding to each value of the fy of the specimen, the h/b values, even when l[ is kept constant [similar to the trend in
section thickness was chosen appropriately (even if nonstandard) to Fig. 2(b)]. This factor is not recognized in the DSM-LB equation for
obtain two different specific values of l[ (equal to 1.0 and 2.0). The the strength, Pu[. The DSM underestimates the strength more and
results from the FEA are given in Table 5. As long as the non- more compared with FEA results, as the value of the l[ increases,
dimensional slenderness ratio l[ is kept constant, the value of the because the value of the h/b ratio increases from the value of 1.0 and,
nondimensional ultimate strength of the sections, Pu[/Py, is in- to a lesser extent, as the value of the h/b decreases below 1.0. Further,
dependent of the yield strength of the material. for values of l[ greater than 1.0, the DSM-LB equations evaluate the
strength to be higher than that from the FEA (unsafe side) in the
Effect of Web Height to Flange Width Ratio range of 0.5 , h/b , 1.5 and for l[ 5 1.0 in the range of h/b ,1.5.
The strength evaluated based on the DSM can be considered unsafe
It was already shown [Fig. 2(b)] that there is a systematic variation of as much as 17% compared with the FEA results (closer to h/b 5 1.0),
the Pu[/Py with respect to h/b, even for the same value of l[. This and as conservative as much as 40% for values of h/b around 3.0.
means that the h/b effect on Pu[/Py cannot be accounted for through Based on this parametric study, it can be concluded that in ad-
dition to l[, the h/b ratio is an important parameter that affects
the nondimensional ultimate strength (Pu[ /Py) of such members.

Table 5. Effect of the Yield Stress on Pu[/Py


l[ fy (MPa) t (mm) Py (kN) Pul-FEA/Py
1.00 165 0.5425 11.46 0.749
250 0.6700 21.44 0.746
300 0.7340 28.18 0.743
350 0.7930 35.51 0.741
400 0.7930 43.55 0.740
2.00 165 0.2705 5.71 0.496
250 0.3330 10.66 0.494
300 0.3650 14.02 0.494
350 0.3945 17.67 0.492
400 0.4220 21.61 0.491

Fig. 4. Load versus axial deformation of LC sections under only local


buckling before failure

Table 4. Effect of the Section Dimension on Pu[/Py


l[ Specimen Py (kN) Pul-FEA/Py
1.00 LC 38 3 38 3 7 3 0.670 21.44 0.746
LC 100 3 100 3 20 3 1.715 145.78 0.743
LC 150 3 150 3 25 3 2.575 321.88 0.739
2.00 LC 38 3 38 3 7 3 0.333 10.66 0.494
LC 100 3 100 3 20 3 0.852 72.42 0.475
LC 150 3 150 3 15 3 1.285 154.2 0.466
Fig. 5. Pul FEA/Pul DSM for different (h/b) or (b/h) and ll values
Note: Refer to Fig.1 for nomenclature of specimens.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / NOVEMBER 2012 / 1295

J. Struct. Eng. 2012.138:1291-1299.


Further, none of the other parameters of LC members affect the equations in Eq. (2b). To correspond to Region b, an upper bound,
nondimensional ultimate strength. In the next section, a detailed Pu[,max/Py, is introduced on the value of the nondimensional
analysis of the data is presented. Subsequently, simple modifications strength, Pu[ /Py [Eqs. (2c) and (2d)]. The resulting MDSM1 strength
to the DSM-LB equations [Eq. (1a) and (1b)] are suggested for more equations for stub LC members under compression are
accurate evaluation of the strength of such members, also accounting
Pu[
for the effect of h/b ratio. In Table 6, the normalized ultimate strength for l[ # 0:60; ¼ 1:00 ð2aÞ
of the stub LCs under compression failing, after only local buckling Py
as obtained from the FEA (Pu[FEA/Py), for a systematic variation
"  b #  b
of l[ and h/b, are given. These results also show that Pu[FEA/Py Pu[;max
Pu[ Pcr[ Pcr[
increases for the same value of l[, as the h/b ratio deviates (increases for l[ . 0:60; ¼ 1 2 a1 #
Py Py Py Py
or decreases) from 1.0. There are a few data that are an exception to
Downloaded from ascelibrary.org by CALIF INSTITUTE OF TECHNOLOGY on 05/14/13. Copyright ASCE. For personal use only; all rights reserved.

this (indicated as boldface in the Table 6). They correspond to ð2bÞ


cases wherein the stocky elements that do not experience local
buckling yield almost over the entire area before or at the ultimate where
load, Pu[.
a1 ¼ 0:27ð2 2 mÞ
Equations for the Ultimate Strength of LC Sections b ¼ 1 ð50 2 7mÞ $ 0:35
after Local Buckling 100
h i
Results presented in the previous sections clearly indicate that the m ¼ max h=b; ðb=hÞ0:6
normalized ultimate strength (Pu[/Py) of stub CFS-LC compression
member failing after only local buckling has to be a function of two
parameters, namely, h/b and l[. Hence, the DSM equations for the
ultimate strength of such members, which are based only on l[,
should be modified if a better accuracy in the strength evaluations is
desired. Two different alternative modifications to the DSM are
evaluated in the following sections, and the corresponding two
methods are referred to as MDSM1 and MDSM2. For this purpose,
in the place of coefficient 0.15 and the power 0.4 in the DSM
equations [Eq. (1b)], two variables, a and b, are introduced, re-
spectively, which are simple functions of the h/b ratio. These var-
iables, as a function of h/b, are evaluated based on regression
analysis using the FEA results.

Modified Direct Strength Methods


Fig. 6, based on the FEA results, shows that the nondimensional
ultimate compressive strength of stub LC, Pu[/Py, varies differently
in the different slenderness ratio regions (a, b, c), as given as follows.
• Region a: Both the stocky restraining elements and the slender
elements buckle locally before member failure by yielding.
• Region b: Only the stocky restraining elements yield over the
entire area, and the slender elements buckle locally before the
member failure by yielding.
• Region c: The member fails by yielding of the entire area of the
cross section without experiencing local buckling.
The DSM equation [Eq. (1a)] is retained in Region c. In Region a,
the DSM equation [Eq. (1b)] is modified introducing using two
parameters, a and b, as functions of h/b to obtain the MDSM1

Table 6. Effect of the h/b Ratio on Pu[/Py


Pul/Py

h/b→ 0.333 0.500 0.667 1.0 1.50 2.00 3.00


l [↓
0.5 0.991 0.988 0.989 0.989 0.995 0.988 0.990
1.0 0.774 0.789 0.791 0.746 0.861 0.859 0.852
1.5 0.637 0.65 0.632 0.587 0.671 0.763 0.769
2.0 0.55 0.558 0.513 0.478 0.537 0.624 0.709
2.5 0.496 0.471 0.454 0.400 0.449 0.528 0.603
3.0 0.452 0.405 0.393 0.369 0.397 0.468 0.513
3.5 0.408 0.362 0.352 0.330 0.357 0.415 0.455 Fig. 6. Comparison of Pul/Py from FEA with MDSM1

1296 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / NOVEMBER 2012

J. Struct. Eng. 2012.138:1291-1299.


"  0:15 # 0:15
Pu[;max P Pcr[
for h=b $ 1:00; ¼ 1 2 0:12 cr[
Py Py Py
ð2cÞ
"  0:32 # 0:32
Pu[;max P Pcr[
for h=b , 1:00; ¼ 1 2 0:20 cr[
Py Py Py
ð2dÞ

It should be noted that the value of a1 becomes negative over some


Downloaded from ascelibrary.org by CALIF INSTITUTE OF TECHNOLOGY on 05/14/13. Copyright ASCE. For personal use only; all rights reserved.

range of h/b. The specimens with b/h 5 1.5, 2.0, and 3.0 are not
prequalified, per NAS (AISI 2007). All the other specimens used in
the FEA are prequalified. In Fig. 6, the curves corresponding to the
MDSM1 [Eqs. (2a)e(2d)], the FEA results, and the DSM-LB equa-
tions are plotted for different discrete values of h/b. Fig. 6 shows that the
set of equations [Eqs. (2a)e(2d)] give values of the ultimate strength
(Pu-MDSM1/Py) that compare very well with the FEA results in all
three regions, namely, a, b, and c. The DSM results are closer to the
MDSM1 results only for h/b 5 1.5. Hence, the DSM results are on
the unsafe side for sections with 1.0 , h/b , 1.5. Fig. 7(b) shows that
the MDSM1 accurately evaluates the strength, even in the case of those
members that are not prequalified according to NAS (AISI 2007).
The MDSM1 equations [Eqs. (2a)e(2d)] can be further sim-
plified to obtain Eqs. (3a) and (3b), referred to as MDSM2 for a small
sacrifice in the accuracy of the strength compared with FEA results.
Only one parameter, a2, is introduced in this format to account for
the influence of h/b on Pu[/Py, instead of the coefficient 0.15 in DSM,
and the value of the power in the equation is equal to 0.4, as in DSM.
These equations have a slightly larger error (as reflected in the value
of SD in Table 7) compared with the FEA results. The MDSM2
equations may be written as

Pu[
for l[ # 0:60; ¼ 1:00 ð3aÞ
Py

for l[ . 0:60;
"  0:4 #  0:4
Pu [ Pcr[ Pcr[ Pu[;max ð3bÞ
¼ 1 2 a2 #
Py Py Py Py

where a2 5 0:7 2 0:4m and m 5 max½h=b; ðb=hÞ0:6 .


Because this study has been done only in the range of l[ , 3.5
and 0.33 # h/b # 3.0, the equations are recommended for use within
this practical dimensional range of CFS-LC sections.
Fig. 7. Comparison of theoretical strengths with test and FEA results

Comparison of MDSM Results with Experimental and


FEA Results
Resistance Factor for Reliability-Based Design Method
Comparison of the strength evaluated by the DSM [Eqs. (1a) and
(1b)], MDSM1 [Eqs. (2a)e(2d)], and MDSM2 [Eqs. (3a) and (3d)] The resistance factor, w, for the MDSM equations in the load and
with the test results available in literature and with the FEA results resistance factor design (LRFD) format is derived in this section. In
generated in this study, is presented in Figs. 7(aec) and Table 7. section F.1.1 of chapter F of NAS (AISI 2007), Eq. (4) is given for
Only the results of tests (32 in number) in which the strength of the this purpose
member is affected by only local buckling before failure are included   2b pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
VM2 1VF2 1CP VP2 1VQ2
in this comparison. Figs. 7(aec) and Table 7 clearly show that f ¼ Cf Mm Fm Pm e 0 ð4Þ
MDSM1 and MDSM2 evaluate the strength of CFS-LC stub
compression members better than DSM, as reflected by the lower where Cf 5 1.52, a calibration coefficient; Mm 5 1.10, a mean value
values of the SD and the maximum and minimum errors. Between of the material factor; Fm 5 1.0, a mean value of the fabrication
the MDSM1 and MDSM2, the accuracy of MDSM1 is slightly factor; VM 5 0.10, a coefficient of variation (COV) of the material;
superior. However, considering the simplicity of MDSM2 and its VF 5 0.05, a COV of the fabrication factor; Cp 5 correction factor
closer similarity to the DSM currently used in practice, MDSM2 may (given in Table 7); n 5 number of tests; and VQ 5 0.21, a COV of the
be acceptable for designs. load effect, are the values as given in NAS (AISI 2007). Target

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / NOVEMBER 2012 / 1297

J. Struct. Eng. 2012.138:1291-1299.


Table 7. Summary of Statistics of (Pu[ test or Pu[ FEA/Pu[ theory)

Test specimens FEA specimens Total specimens


Parameters
describing statistics DSM MDSM1 MDSM2 DSM MDSM1 MDSM2 DSM MDSM1 MDSM2
N 32 32 32 71 71 71 103 103 103
m 1.011 0.975 0.984 0.993 1.010 1.014 0.998 1.006 1.012
s 0.109 0.057 0.064 0.128 0.031 0.041 0.122 0.046 0.052
Vp 0.108 0.058 0.065 0.129 0.031 0.040 0.122 0.045 0.051
Max 1.291 1.116 1.165 1.398 1.081 1.090 1.398 1.175 1.209
Min 0.832 0.883 0.889 0.850 0.926 0.903 0.832 0.894 0.903
Cp 1.10 1.10 1.10 1.07 1.07 1.07 1.04 1.04 1.04
Downloaded from ascelibrary.org by CALIF INSTITUTE OF TECHNOLOGY on 05/14/13. Copyright ASCE. For personal use only; all rights reserved.

w 0.87 0.88 0.89 0.84 0.91 0.91 0.85 0.91 0.91

ultimate strength compared with DSM, unless 1.0 , h/b , 1.5. The
ratios of the strength evaluated (MDSM1/DSM and MDSM2/DSM)
are plotted in Figs. 8(a and b), with respect to the ratio of the element
widths (h/b or b/h) of the LC section. The LC compression members
for which the test results are reported in literature and the members
used in FEA in this study have been used for this evaluation. The
ratio of the width of the buckling element to the restraining ele-
ment increases ðh=b  1:0 or b=h  1:0Þ the design strength of
members based on MDSM1 and MDSM2 is higher than that from
the DSM. The ultimate strength results from MDSM may be as much
as 20e40% higher compared with the DSM results. Thus, more
economical designs could be obtained using MDSM compared with
DSM, in such ranges of h/b values, for essentially the same level of
target reliability.

Conclusions

The DSM evaluates the ultimate strength of stub LC compression


members failing after only local buckling as a function of only the
nondimensional local buckling slenderness ratio [l[ 5 √(Py /Pcr[)].
However, the nondimensional ultimate strength (Pu[ /Py) of these
members is also affected by the ratio of the width of the web to the
flange, h/b of the elements of the cross section. Hence, the DSM
equations for evaluating the strength of such members are less
accurate, and the results can be unsafe when h/b 5 1.0. Moreover,
the ultimate strength of such members evaluated by DSM are very
conservative for larger or smaller values of h/b. Simple mod-
ifications to the DSM equations are recommended in this study to
more accurately evaluate the strength of CFS-LC stub compression
members as affected only by local buckling. This could lead to more
economical designs with higher strengths (20e40% higher strength)
compared with the DSM for the same level of reliability, particularly
Fig. 8. Comparison of Pul from design methods when the width of the element initiating the buckling to the width of
the restraining element of the cross section increases, and the elastic
local buckling strength of the member is small.
reliability b0 of 2.5 is used as recommended. The parameters Pm and
VP are the mean (5m) and COVs (#0.065) of the column strength
ratios, respectively. The values for these as reported in Table 7 are References
used for the evaluation of the resistance factor, w. The resistance
factors, w, evaluated based on Eq. (4) are reported in Table 7 for the ABAQUS. (2007). ABAQUS standard user’s manual, Version 6.7, Dassault
DSM, MDSM1, and MDSM2. The values are equal to or higher Systèmes Simulia, Providence, RI.
(giving higher value of design strength) than the recommended value American Iron and Steel Institute (AISI). (2007). “North American speci-
fication for the design of cold-formed steel structural members.” AISI
of 0.85 by AISI (2007) for compression members; hence, the w value
S100e2007, Washington, DC.
of 0.85 can be safely used in both MDSM equations. Bambach, M. R. (2010). “Unified element and section approach to design of
Although the w value evaluated does not change significantly for cold-formed steel structures.” J. Struct. Eng., 136(4), 343e353.
all the three equations, the designs based on MDSM1 and MDSM2 British Standards Institute. (1987). “Structural use of steelwork in building.
are likely to be more economical, because as seen in Fig. 8, in most of Code of practice for design of cold formed thin gauge sections,” BS 5950-
the ranges of l[ and h/b, the DSM would give a higher value of the 5, London.

1298 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / NOVEMBER 2012

J. Struct. Eng. 2012.138:1291-1299.


European Committee for Standardization (CEN). (2004). “Design of steel Schafer, B. W., and Adany, S. (2006). “Buckling analysis of cold-formed
structures, part 1.3: Supplementary rules for cold-formed members and steel members using CUFSM: conventional and constrained finite strip
sheeting.” ENV 1993-1-3, Eurocode 3, Brussels, Belgium. methods.” Proc., 18th Int. Specialty Conf. on Cold-Formed Steel
Hancock, G. J., Kwon, Y. B., and Bernard, E. S. (1994). “Strength design Structures, Orlando, FL, 39e54.
curves for thin walled sections undergoing distortional buckling.” Schafer, B. W., and Peköz, T. (1998a). “Computational modeling of cold
J. Construct. Steel Res., 31(2e3), 169e186. formed steel: Characterizing geometrical imperfections and residual
Kumar, M. V. A. (2012). “Interaction of local, distortional and overall stresses.” J. Construct. Steel Res., 47(3), 193e210.
buckling in cold formed steel lipped channel compression mem- Schafer, B. W., and Peköz, T.(1998b). “Direct strength prediction of cold
bers.” Ph.D. thesis, Indian Institute of Technology Madras, Chennai, formed steel members using numerical elastic buckling solutions.”
India. Proc., 14th Int. Specialty Conf. on Cold-Formed Steel Structures,
Kwon, Y. B., and Hancock, G. J. (1992). “Strength design curves for thin- Univ. of Missouri-Rolla, St. Louis, 69e76.
walled sections undergoing local and distortional buckling.” J. Struct. Silvestre, B. W., Dinis, P. B., and Camotin, D. (2006). “Direct strength
Downloaded from ascelibrary.org by CALIF INSTITUTE OF TECHNOLOGY on 05/14/13. Copyright ASCE. For personal use only; all rights reserved.

Eng., 118(7), 1786e1803. method for lipped channel columns and beams affected by local-plate/
Kwon, Y. B., Kim, B. S., and Hancock, G. J. (2009). “Compression tests distortional interaction.” Proc., 18th Int. Specialty Conf. on Cold-
of high strength cold formed steel channels with buckling interaction.” Formed Steel Structures, Orlando, FL, 1e18.
J. Construct. Steel Res., 65(2), 278e289. Standards Australia. (2005). “Cold-formed steel structures.” AS/NZS4600:
Kwon, Y. B., Kim, N. K., and Kim, B. S. (2005). “A study on the direct 2005, Standards Australia/Standards New Zealand, Sydney, NSW,
strength method for compression members undergoing mixed mode Australia.
buckling.” Proc., Steel Structures ISSS’05, Korean Society of Steel Talja, A. (1990). “Design of the buckling resistance of compressed HSS
Construction, Seoul, Korea, 108e119. channels.” Research Note No. 1163, Technical Research Centre of
Lau, S. C. W., and Hancock, G. J. (1990). “Inelastic buckling of channel Finland, Vuorimiehentie, Espoo, Finland.
columns in the distortional buckling mode.” Thin-walled Struct., 10(1), Weng, C. C., and Peköz, T.(1988). “Compression tests of cold-formed steel
59e84. columns.” Proc., 9th Int. Specialty Conf. on Cold-Formed Steel
Lim, B. S. (1985). “Buckling behaviour of asymmetric edge stiffened Structures, St. Louis, MO, 1e20.
plates.” Ph.D. thesis, Univ. of Strthclyde, Glasgow, U.K. Yang, D., and Hancock, G. J. (2004). “Compression tests of high strength
Mulligan, G. P., and Peköz, T. (1987). “Local buckling interaction in cold steel channel columns with interaction between local and distortional
formed columns.” J. Struct. Eng., 113(3), 604e620. buckling.” J. Struct. Eng., 130(12), 1954e1963.
Schafer, B. W. (2002). “Local, distortional, and euler buckling in thin-walled Young, B., and Rasmussen, K. J. R. (1998). “Design of lipped channel
columns.” J. Struct. Eng., 128(3), 289e299. columns.” J. Struct. Eng., 124(2), 140e148.
Schafer, B. W. (2008). “Review: The direct strength method of cold- Zaras, J., and Rhodes, J. (1987). “Carefully controlled compression tests on
formed steel member design.” J. Construct. Steel Res., 64(7e8), thin-walled cold formed sections.” Applied solid mechanics, A. S. Tooth
766e778. and J. Spence, eds., Vol. 2, Elsevier, New York, 519e551.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / NOVEMBER 2012 / 1299

J. Struct. Eng. 2012.138:1291-1299.

You might also like