D S M C - F S D: Irect Trength Ethod OF OLD Ormed Teel Esign

DRAFT - JANUARY 7, 2002
Design Manual for
DIRECT STRENGTH METHOD

OF
COLD-FORMED STEEL DESIGN
prepared by: Ben Schafer, Ph.D.
for: American Iron and Steel Institute - Committee on Specifications
Contents:
SPECIFICATION
Applicability
Elastic buckling
Columns
Beams
COMMENTARY
Applicability
Elastic buckling
Columns
Beams
EXAMPLES
Strength and stiffness
Elastic Buckling
CUFSM SOFTWARE
TUTORIALS
PRESENTATIONS (AND ADDITIONAL MATERIALS)
DRAFT – December 6, 2002
DRAFT – DECEMBER 6, 2002
SPECIFICATION

OF
Contents:
Preface..................................................................................................... 1
1.1 General Provisions ...................................................................... 2
1.1.1 Applicability ....................................................................... 2
1.1.2 Elastic Buckling .................................................................. 5
1.1.3 Deflection Determination.................................................... 5
1.2 Columns ...................................................................................... 6
1.2.1 Flexural, Torsional, or Flexural-Torsional Buckling.......... 6
1.2.2 Local Buckling.................................................................... 6
1.2.3 Distortional Buckling.......................................................... 6
1.3 Beams.......................................................................................... 7
1.3.1 Lateral-Torsional Buckling................................................. 7
1.3.2 Local Buckling.................................................................... 7
1.3.3 Distortional Buckling.......................................................... 7
Appendix 1: Design of Cold-Formed Steel Structural

Members by the Direct Strength Method
Preface
Users of the main Specification are aware of the complexity involved in current design
calculations for cold-formed steel members. A new design method: Direct Strength, has been
created that aims to alleviate the current complexity, ease calculation, provide a more robust and
flexible design procedure, and integrate with available, established, numerical methods.
Advantages of the Direct Strength Method of design include
everyday design improvements:
• no effective properties for strength,
• no element calculations,
• no iteration for beams (webs),
• gross properties of the section used for strength calculations,
theoretical improvements:
• interaction of elements in local buckling (e.g., web/flange interaction) is accounted for,
• distortional buckling is explicitly treated in the design process,
• reduction in systematic error in portions of the main Specification,
improvements in applicability and scope:
• is applicable to wider group of cross-sections than the main Specification,
• provides rational analysis procedure for sections not previously covered,
• allows and encourages greater cross-section optimization,
important philosophical changes:
• numerical methods and rational analysis brought to everyday design, and
• integrates known behavior into a straightforward design procedure.
Use of the Direct Strength Method requires (1) determination of the elastic buckling behavior of
the member and (2) using that information in a series of ultimate strength curves to predict the
strength. The commentary to this Appendix provides full details of rational analysis methods,
both traditional hand solutions and numerical solutions, that may be used to accurately calculate
the elastic buckling behavior necessary for step 1. A freely available program, developed in part
with AISI funds: CUFSM (www.ce.jhu.edu/bschafer/cufsm) complete with tutorials and
examples is available to aid in the elastic buckling calculation. This appendix provides the
ultimate strength curves along with appropriate safety and resistance factors necessary for the
strength prediction of step 2.
The Direct Strength Method is verified with same data and care as the main Specification, and
thus no loss of reliability is inherent in its use. The procedure employs the same underlying
empirical assumptions as the effective width method used in the main Specification: ultimate
strength is a function of elastic buckling stress (or load) and the yield stress (or load) of the
material. The method has been extensively explored for beams and columns only. The
applicability of the provided provisions is detailed in the General Provisions of this Appendix.
Appendix 1 - 1
1.1 General Provisions
1.1.1 Applicability
The provisions of this Appendix are applicable for determination of the nominal axial (Pn) and
flexural (Mn) strength of cold-formed steel members. Sections 1.2 and 1.3 present a method
applicable to all cold-formed steel beams and columns consistent with the rational analysis
clause of Section A1.1(b) of the main Specification. Those members meeting the geometric and
material limitations of Section 1.1.1.1 for columns and 1.1.1.2 for beams have been pre-qualified
for use, and calibrated Ω and φ factors apply.
Currently, the Direct Strength method provides no explicit provisions for members in: shear,
combined bending and shear, web crippling, combined bending and web crippling, or combined
axial load and bending (beam-columns). Further, no provisions are given for structural
assemblies or connections and joints. As detailed in A1.1 of the main Specification, the
provisions of the main Specification, when applicable, shall be used for all cases listed above.
For inapplicable members or situations, obvious extensions to the Direct Strength method may
exist (e.g. in shear, in certain structural assemblies, etc.). Users who choose to employ such
extensions to the Direct Strength method are subject to the same provisions as any other rational
analysis procedure as detailed in Section A1.1(b) of the main Specification: (1) applicable
provisions of the main Specification must be followed when they exist, (2) reduced φ and
increased Ω factors are employed for strength when rational analysis is conducted.
Appendix 1 - 2
1.1.1.1 Pre-qualified Columns

Unperforated columns that fall within the geometric limitations given below have been pre-
qualified and may use the higher φ and Ω factors of section 1.2. Dimensional limitations listed in
gray and italicized are not mandatory. All given dimensions are out-to-out.
Lipped Channel 34 < h/t < 472
b 18 < b/t < 159
θ 4 < d/t < 33
0.7 < h/b < 5.0
h 0.05 < d/b < 0.41
θ ~ 90 deg.
d 340 < E/Fy < 1020
(29 ksi < Fy < 86 ksi) or (200 MPa < Fy < 593 MPa)
Lipped Channel with 105 < h/t < 489
Web Stiffener(s) 68 < b/t < 160
6 < d/t < 33
d
1.3 < h/b < 2.7
i
d/b ~ 0.20
h’ 0.4 < di/d < 2.8 (one or two intermediate stiffeners)
340 < E/Fy < 820
Zee (Zed) 76 < h/t < 137
b 30 < b/t < 56
d θ
0 < d/t < 36
h 1.5 < h/b < 2.7
0.00 < d/b < 0.73
θ ~ 50 deg.
E/Fy ~ 590
(Fy ~ 50 ksi) or (Fy ~ 345 MPa)
Rack Upright 34 < h/t < 51
12 < b/t < 22
5 < d/t < 8
2.1 < h/b < 2.9
1.6 < b2/d < 2.0 (b2=small outstand parallel to b)
d2/d ~ 0.3 (d2=second lip parallel to d)
E/Fy ~ 343
(Fy ~ 86 ksi) or (Fy ~ 593 MPa)
Hat 33 < h/t < 50
12 < b/t < 20
4 < d/t < 6
1.0 < h/b < 1.2
d/b ~ 0.13
428 < E/Fy < 952
Appendix 1 - 3
1.1.1.2 Pre-qualified Beams

Unperforated beams that fall within the geometric limitations given below have been pre-
qualified and may use the higher φ and Ω factors of section 1.3. Dimensional limitations listed in
gray and italicized are not mandatory. All dimensions are out-to-out. All beams are bent about a
horizontal axis as shown in the figures.
Channels 43 < h/t < 321
b 12 < b/t < 75
θ 0 < d/t < 34
1.5 < h/b < 17.0
h 0.0 < d/b < 0.70
44 deg. < θ < 90 deg.
d 421 < E/Fy < 983
Lipped Channels with Web 200 < h/t < 358
Stiffener 29 < b/t < 58
14 < d/t < 17
d
5.5 < h/b < 11.7
i
0.27 < d/b < 0.56

h’
θ ~ 90 deg.
578 < E/Fy < 670
Zee (Zed) 114 < h/t < 183
b 45 < b/t < 71
d θ
10 < d/t < 16
h 2.5 < h/b < 4.1
0.15 < d/b < 0.34
36 deg. < θ < 90 deg.
440 < E/Fy < 921
Hats (Decks) 37 < h/t < 97
92 < bc/t < 467
0 < ds/t < 26
0.14 < h/bc < 0.87
0.44 < bc/2bt < 2.7
0<n<4
492 < E/Fy < 656
Trapezoids (Decks) 51 < h/t < 203
61 < bc/t < 231
0.42 < (h/sinθh)/bc < 1.91
0.55 < bc/2bt < 1.69
0 < nc < 2 (compression flange stiffener)
0 < nw < 2 (web stiffener/fold)
0 < nt < 2(tension flange stiffener)
52deg. < θh < 84deg.
310 < E/Fy < 686
Appendix 1 - 4
1.1.2 Elastic Buckling

Elastic buckling of cold-formed steel members under flexure or axial load may involve three
distinct buckling modes: local, distortional, and overall. Rational analysis is required for
determination of the elastic buckling loads and or moments used in the Direct Strength method,
namely, for columns Pcrl, Pcrd, and Pcre of 1.2, and for beams Mcrl, Mcrd, and Mcre of 1.3. For a
given beam or column all three modes may not exist. In this case, the non-existent mode should
be ignored in the calculations of sections 1.2 and 1.3.
The rational analysis method(s) employed for determination of the elastic buckling loads and or
moments shall provide the solution for the member as a whole. Compatibility and equilibrium
between the elements that comprise a member must be included, or if neglected in part or in full,
must be done so in a conservative fashion for the entire member. The rational analysis model
employed should be able to reproduce classic benchmark problems (at a minimum a simply
supported plate in pure compression and pure bending) with no more than a 1% over-estimation
of the exact buckling solution. The same methods used to produce benchmark solutions (e.g., the
element sizes and density for the finite element or finite strip method) must also be employed in
actual solutions on members.
The commentary to this Appendix discusses a variety of applicable rational analysis methods for
elastic buckling prediction. Two methods which meet the criteria of this section are discussed in
full detail: (1) a conservative extension of conventional hand solutions, and (2) a numerical
implementation of the finite strip method. Complete formulas for the former method are
provided in the Appendix. For the latter method a free program complete with tutorials and
examples is referenced and discussed in the Appendix. Elastic buckling is a well-defined and
repeatable calculation; a variety of other methods: finite element, boundary element, generalized
beam theory, can provide a reliable solution. Any methods meeting the criteria of this section are
acceptable as a rational analysis method for elastic buckling prediction.
1.1.3 Deflection Determination

The bending deflection at any moment (Ma) due to service loads, may be determined by reducing
the gross moment of inertia, Ig, to an effective moment of inertia for deflection, per:
Ieff = Ig(Mna/Ma) (Eq. 1.1.3-1)
where: Mna = Mn of section 1.3, determined with My replaced by Ma
Ma = moment due to service loads on member of interest (Ma < My)
The axial deflection at any service load ( Pa) may be determined by reducing the gross area, Ag,
to an effective area for deflection, per:
Aeff = Ag(Pna/Pa) (Eq. 1.1.3-2)
where: Pna = Pn of section 1.2, determined with Py replaced by Pa
Pa = service load of interest (Pa < Py)
Appendix 1 - 5
1.2 Columns
The nominal axial strength, Pn, is the minimum of Pne, Pnl and Pnd as given below.
For columns meeting the geometric criteria of section 1.1.1.1, φ and Ω are as follows:
USA and Mexico Canada
Ωc (ASD) φc (LRFD) φc (LSD)
1.80 0.85 0.80
For all other columns, standard rational analysis values are used: 0.9φ and 1.1Ω, as follows:
Ωc (ASD) φc (LRFD) φc (LSD)
2.00 0.75 0.70
1.2.1 Flexural, Torsional, or Flexural-Torsional Buckling

The nominal axial strength, Pne, for flexural, torsional, or flexural-torsional buckling is
for λ c ≤ 1.5 ( 2
)
Pne = 0.658 λc Py (Eq. 1.2.1-1)
 0.877 
for λc > 1.5 Pne =  2 Py (Eq. 1.2.1-2)
 λc 
where λc= Py Pcre (Eq. 1.2.1-3)
Py = AgFy (Eq. 1.2.1-4)
Pcre = Minimum of the critical elastic column buckling load in flexural,
torsional, or flexural-torsional buckling
1.2.2 Local Buckling

The nominal axial strength, Pnl, for local buckling is
for λl ≤ 0.776 Pnl = Pne (Eq. 1.2.2-1)
 P 
0.4
 P 
0.4
for λl > 0.776 Pnl = 1 − 0.15 crl   crl  Pne (Eq. 1.2.2-2)
  Pne   Pne 
 
where λl= Pne Pcrl (Eq. 1.2.2-3)
Pcrl = Critical elastic local column buckling load
1.2.3 Distortional Buckling

The nominal axial strength, Pnd, for distortional buckling is
for λd ≤ 0.561 Pnd = Py (Eq. 1.2.3-1)
  Pcrd   Pcrd 
0.6 0.6

for λd > 0.561 Pnd = 1 − 0.25    P (Eq. 1.2.3-2)
 P   P  y
  y   y 
where λd= Py Pcrd (Eq. 1.2.3-3)
Pcrd = Critical elastic distortional column buckling load
Appendix 1 - 6
1.3 Beams
The nominal flexural strength, Mn, is the minimum of Mne, Mnl and Mnd as given below.
For beams meeting the geometric criteria of section 1.1.1.2, φ and Ω are as follows:
Ωb (ASD) φb (LRFD) φb (LSD)
1.67 0.90 0.85
For all other beams, standard rational analysis values are used: 0.9φ and 1.1Ω, as follows:
Ωb (ASD) φb (LRFD) φb (LSD)
1.85 0.80 0.75
1.3.1 Lateral-Torsional Buckling

The nominal flexural strength, Mne, for lateral-torsional buckling is
for Mcre < 0.56My Mne = Mcre (Eq. 1.3.1-1)
for 2.78My>Mcre>0.56My
10 M
(
Mne = 109 M y 1 − 36 M crey ) (Eq. 1.3.1-2)
for Mcre > 2.78My Mne = My (Eq. 1.3.1-3)
where
My = SgFy , where Sg is referenced to the extreme fiber in first yield (Eq. 1.3.1-4)
Mcre = Critical elastic lateral-torsional buckling moment

The nominal flexural strength, Mnl, for local buckling is
for λl ≤ 0.776 Mnl = Mne (Eq. 1.3.2-1)
 M 
0.4
 M 
0.4
for λl > 0.776 Mnl = 1 − 0.15 crl   crl  M ne (Eq. 1.3.2-2)

  M ne   M ne 
 
where λl= M ne M crl (Eq. 1.3.2-3)
Mcrl = Critical elastic local buckling moment

The nominal flexural strength, Mnd, for distortional buckling is
for λd ≤ 0.673 Mnd = My (Eq. 1.3.3-1)
  M crd   M crd 
0.5 0.5

for λd > 0.673 Mnd = 1 − 0.22    My (Eq. 1.3.3-2)
 M   M 
  y   y 
where λd= M y M crd (Eq. 1.3.3-3)
Mcrd = Critical elastic distortional buckling moment
Appendix 1 - 7
DRAFT – DECEMBER 6, 2002
COMMENTARY

OF
Contents:
Preface..................................................................................................... 1
1.1 General Provisions ...................................................................... 1
1.1.1 Applicability ....................................................................... 1
1.1.1.1 Pre-qualified Columns .................................................... 2
1.1.1.2 Pre-qualified Beams........................................................ 4
1.1.2 Elastic Buckling .................................................................. 6
1.1.2.1 Elastic Buckling Numerical Solutions ............................ 7
1.1.2.2 Elastic Buckling Hand Solutions .................................. 20
1.1.3 Deflection Determination.................................................. 30
1.2 Columns .................................................................................... 30
1.2.1 Flexural, Torsional, or Flexural-Torsional Buckling........ 33
1.2.2 Local Buckling.................................................................. 33
1.2.3 Distortional Buckling........................................................ 33
1.3 Beams........................................................................................ 34
1.3.1 Lateral-Torsional Buckling............................................... 36
1.3.2 Local Buckling.................................................................. 36
1.3.3 Distortional Buckling........................................................ 37
References............................................................................................. 37
DRAFT – December 6, 2002 – COMMENTARY: Specification for the Direct Strength Method
COMMENTARY Appendix 1: Design of Cold-Formed Steel

Structural Members by the Direct Strength Method
Preface
The complexity of the main specification for cold-formed steel design is immediately apparent to
all who use it. This complexity is justified by the rather advanced and optimized mechanical
behavior that occurs in thin-walled cold-formed steel beams and columns. However, much of the
complexity is related to performing detailed elastic buckling calculations, by hand, for the large
variety of different buckling modes that exist in cold-formed steel members (much of Chapters B
and C of the main Specification are elastic buckling calculations).
As members become more optimized the elastic buckling calculations become more
complicated. However, elastic buckling of thin-walled members may be readily completed using
numerical methods that are verifiable and robust. Therefore, a design method which employs
accurate elastic buckling calculations determined by rational analysis has a chance of eliminating
much of the complexity of current design procedures: the Direct Strength Method of this
appendix is an attempt at such a method.
Currently, the procedures, verification, and calibration for the Direct Strength method have only
been completed for beams and columns, but work continues on extending this method to all
cold-formed members in all loading cases. This commentary to Appendix 1 provides
• rational analysis procedures and guidelines for elastic buckling,
• summary tables of all beams and columns examined, and
• discussion and justification of the strength expressions employed.
The bulk of the commentary (section 1.1.2) is devoted to providing procedures and guidelines
for elastic buckling calculation. The Specification allows rational analysis for elastic buckling
calculation, and thus no specific methods are prescribed. This commentary provides complete
hand methods for elastic buckling calculation that are applicable to conventional cold-formed
steel members. Additionally, the specification provides background, discussion, examples,
references and freely available software for elastic buckling calculation via numerical methods.
The remaining commentary is organized to parallel the structure of Appendix 1.
1.1 General Provisions
1.1.1 Applicability
Section A1.1 of the main Specification states that rational analysis methods are allowed when
the composition or configuration of a component is such that calculations cannot be made in
accordance with the main Specification. The provisions of Appendix 1 provide a calibrated
rational analysis method for strength prediction of beams and columns which cannot be
completed by the main Specification. The reliability of the method is insured by using the
calibrated safety and resistance factors within set geometric limits only, and conservative safety
and resistance factors for all other configurations.
Appendix 1 - 1
What about those members which ostensibly can be calculated by the main Specification? Can
Appendix 1 be used to determine the strength of such members? Yes. Appendix 1 is unique
when compared to the typical rational analysis methods of section A1.1(b), because it is
explicitly a part of the actual Specification. Therefore, the methods of Appendix 1 are available
for all cold-formed steel beams and columns.
The applicability of Appendix 1 to all beams and columns implies that in some situations
competing methods may exist for strength determination of a member: the main Specification,
and Appendix 1. In this situation there is no “preferred method.” Either method may be used to
determine the strength [resistance]. The fact that one method may give a greater or lower
strength prediction in a given situation does not imply an increased accuracy for either method.
The calibrated safety and resistance factors are designed to insure that both methods reach their
target reliability, which for members is a β of 2.5.
The method of Appendix 1 provides solutions for beams and columns, but suggests that these
solutions may be combined with the regular provisions of the main Specification to cover other
cases: shear, beam-columns, etc. This notion is based on the fact that the overall reliability of
Appendix 1 is the same as that of the main Specification, even though individual members may
have different predictions in strength. The Direct Strength Method has been successfully applied
to typical purlins, where the provisions of this Appendix are used for the bending capacity, and
then augmented by Shear, and Shear + Bending calculations, in-line with the main Specification
(Quispe and Hancock 2002). Beam-columns may be conservatively examined using the
provisions of the main Specificaton, but new rational analysis methods are also available
(Schafer 2002b).
1.1.1.1 Pre-qualified Columns

An extensive amount of testing has been performed on concentrically loaded, pin-ended, cold-
formed steel columns (Kwon and Hancock 1992, Lau and Hancock 1987, Loughlan 1979, Miller
and Peköz 1994, Mulligan 1983, Polyzois et al. 1993, Thomasson 1978). Data from these
researchers was compiled and used for calibration of the Direct Strength Method. The geometric
limitations listed in the Specification are based on these experiments. A summary of the
geometry used by each researcher follows:
b b
θ d θ di
h h
h’
d
(a) lipped channel (b) lipped zed (c) lipped channel with web stiffener
and additional lip to lip restraints
(d) lipped channel with web stiffener (e) hat (f) rack sections
Appendix 1 - 2
Figure 1.1-1 Cross-section of columns used for calibration
Table 1.1-1 Geometry of lipped channel columns

h/t b/t d/t h/b d/b fy num
reference max min max min max min max min max min max min
Loughlan (1979) 322 91 80 30 33 11 5.0 1.6 0.41 0.36 35 34 33
Miller and Pekoz (1994) 170 46 38 18 8 5 4.6 2.5 0.32 0.19 53 44 19
Mulligan (1983) 207 93 93 64 16 14 2.9 1.0 0.22 0.16 35 32 13
Mulligan (1983) Stubs 353 65 100 33 22 7 3.9 0.7 0.23 0.16 60 29 24
Thomasson (1978) 472 207 159 69 32 14 3.0 3.0 0.23 0.19 69 42 13
totals 472 46 159 18 33 5 5.0 0.7 0.41 0.16 69 29 102
more lipped channels (subset of U. Sydney tests which fails in distortional mode)
Lau and Hancock (1988) 50 34 37 25 6 4 1.4 1.3 0.18 0.16 69 31 8
Kwon and Hancock (1992) 111 110 83 82 6 4 1.3 1.3 0.08 0.05 86 86 5
totals 111 34 83 25 6 4 1.4 1.3 0.18 0.05 86 31 13
Table 1.1-2 Geometry of lipped zed columns

Polyzois et al. (1993) 137 76 56.1 30 36.2 0 2.74 1.5 0.73 0 50 50 85
Table 1.1-3 Geometry of lipped channel with a web stiffener columns

(a) Additional restraints from lip to lip
h/b h/t b/t d/t
max min max min max min max min count
Thomasson (1978) 3.1 3.0 489 205 160 68 33 14 46
di/d h'/t
max min max min count
no intermediate web stiffeners - - - - 14
one intermediate web stiffener 0.94 0.39 222 91 16
two intermediate web stiffeners 0.94 0.47 145 57 16
(b) Standard section
lipped channel with a web stiffener, width (ws), and depth (ds) all steel fy = 86 ksi
h/t b/t d/t h/b ds/d ws/b num
Kwon and Hancock (1992) 110 105 82 82 13 6 1.3 1.3 2.8 1.4 0.1 0.1 6
Table 1.1-4 Geometry of hat columns

Lau and Hancock (1988) 50 33 43 34 6 4 1.2 1.0 0.14 0.13 69 31 4
(subset of Lau and Hancock data which failed in distortional mode)
Table 1.1-5 Geometry of rack section columns

lipped channel with an outstand (width b2)
h/t b/t d/t h/b b2/d fy num
Lau and Hancock (1988) 46 34 20 12 8 5 2.7 2.3 2.0 1.6 69 31 9
lipped channel with an outstand (width b2) and a return lip (width d2)
h/t b/t d/t h/b b2/d d2/d num
Lau and Hancock (1988) 51 34 22 12 7 6 2.9 2.1 2.0 1.6 0.3 0.3 8
(subset of Lau and Hancock data which failed in distortional mode)
Appendix 1 - 3
It is the intent of this Specification that as more cross-sections are verified for use in the Direct
Strength method that these tables and sections would be augmented with new information. Those
companies with proprietary sections may wish to perform their own testing and follow the details
of Chapter F of the main Specification to independently justify the use of higher φ and lower Ω
factors for a particular cross-section. Alternatively, members which fall outside the scope of this
testing may still use the method, but with the reduced φ and increased Ω factors consistent with
any rational analysis method.
1.1.1.2 Pre-qualified Beams

An extensive amount of testing has been performed on laterally braced beams (Cohen 1987,
Ellifritt et al. 1997, LaBoube and Yu 1978, Moreyara 1993, Phung and Yu 1978, Rogers 1995,
Schardt and Schrade 1982, Schuster 1992, Shan 1994, Willis and Wallace 1990) and on hats and
decks (Acharya 1997, Bernard 1993, Desmond 1977, Höglund 1980, König 1978, Papazian et al.
1994). Data from these researchers was compiled and used for calibration of the Direct Strength
Method. The geometric limitations listed in the Specification are based on the experiments
performed by these researchers. A summary of the geometry used by each researcher follows
(beams are bent about the horizontal axis, as oriented in the figures below):
Cohen (1987) Ellifritt et al. (1997) Laboube and Yu (1978)
I II B MB
Moreyra (1993) Rogers (1995) Schardt and Schrade (1982)
Schuster (1992) Shan et al. (1994) Willis and Wallace (1990)
Figure 1.1-2 Cross-section and testing details of of lipped channel and zed beams used for calibration
Table 1.1-6 Geometry of lipped channel beams

section Max of Min of Max Min of Max Min of Max of Min Max of Min Max Min Max Min Max Min
name n
id h/t h/t of b/t b/t of d/t d/t h/b of h/b d/b of d/b of q of q of t of t of fy of fy
C Cohen (1987) 128 78 55 33 16 9 2.38 2.34 0.43 0.20 90 89 0.107 0.064 61 54 14
Ellifritt et. al. (1997) 126 113 48 47 14 11 2.63 2.36 0.29 0.24 90 90 0.072 0.066 70 60 5
LaBoube and Yu (1978) 269 77 75 28 15 11 8.33 1.58 0.44 0.17 90 90 0.052 0.046 54 34 32
Moreyra (1993) 124 117 36 33 16 12 3.54 3.45 0.46 0.35 90 90 0.073 0.069 64 57 9
Phung and Yu (1978) 321 210 32 31 17 16 10.38 6.47 0.55 0.50 90 90 0.042 0.040 51 51 6
Rogers (1995) 228 53 61 15 34 3 13.72 1.49 0.70 0.16 90 90 0.076 0.042 61 44 49
Schuster (1992) 168 166 34 33 11 10 5.10 4.95 0.33 0.29 90 90 0.048 0.048 48 39 5
Shan (1994) 256 43 58 19 20 6 7.34 1.50 0.41 0.24 90 90 0.073 0.031 65 30 29
Willis and Wallace (1990) 131 126 40 38 17 14 3.46 3.18 0.46 0.35 89 87 0.063 0.061 60 54 4
C Total 321 43 75 15 34 3 13.72 1.49 0.70 0.16 90 87 0.107 0.031 70 30 153
Table 1.1-7 Geometry of unlipped channel beams

name n
C0 Rogers (1995) 225 53 59 12 0 0 17.36 1.67 0.00 0.00 90 90 0.076 0.042 61 44 10
Appendix 1 - 4
Table 1.1-8 Geometry of lipped channel with edge stiffener at an angle (q) beams
name n
Cs Cohen (1987) 128 78 55 33 16 9 2.38 2.34 0.43 0.20 62 44 0.107 0.064 61 54 22
Table 1.1-9 Geometry of lipped channel with a web stiffener beams

name n
Cw1 Phung and Yu (1978) 358 200 58 29 17 14 11.70 5.54 0.56 0.27 90 90 0.042 0.037 51 44 42
Table 1.1-10 Geometry of liped zed beams

name n
Z Ellifritt et. al. (1997) 139 114 39 31 16 13 3.71 3.54 0.43 0.38 41 36 0.072 0.060 67 61 5
Schardt and Schrade (1982) 180 178 70 50 10 10 3.58 2.56 0.20 0.14 61 45 0.040 0.040 39 37 8
Willis and Wallace (1990) 129 125 40 34 13 11 3.77 3.20 0.40 0.28 55 51 0.080 0.062 59 52 6
Z Total 180 114 70 31 16 10 3.77 2.56 0.43 0.14 61 36 0.080 0.040 67 37 19
Zr Schardt and Schrade (1982) 183 179 71 45 16 10 4.06 2.55 0.34 0.15 90 90 0.040 0.039 54 32 29
Zr Total 183 179 71 45 16 10 4.06 2.55 0.34 0.15 90 90 0.040 0.039 54 32 29
(a) Acharya/Papazian sections (b) Desmond sections (c) König sections

Figure 1.1-3 Cross-section of hat section beams used in calibration
Table 1.1-11 Geometry of hat section beams

web comp. flange stiffener aspect ratio neutral axis subelement depth/width stiffener yield stress thickness
95 h/t bc/t ds/t h/bc bc/(2bt) bc/[(n+1)t] ds/(bc/(n+1)) n fy t
members max min max min max min max min max min max min max min max min max min max min
Acharya et al. 93 38 393 95 25 6 0.87 0.14 2.40 0.44 79 32 0.35 0.11 4 2 50 45 0.058 0.023
Papazian.. 93 37 398 92 25 10 0.45 0.15 2.47 0.75 80 46 0.37 0.20 4 1 50 45 0.057 0.022
22 h/t bc/t ds/t h/bc bct/(2btt+bptbp) bc/[(n+1)t] ds/(bc/(n+1)) n fy t
Desmond et al. 97 68 327 104 24 5 0.83 0.29 0.76 0.48 163 52 0.31 0.05 1 1 46 40 0.069 0.056
Konig et al. 94 56 467 167 26 0 0.34 0.20 2.72 1.34 467 59 0.23 0.00 2 0 60 53 0.070 0.025
Appendix 1 - 5
(a) Bernard deck (b) Höglund deck

Figure 1.1-4 Cross-section of trapezoidal decks used in calibration
Table 1.1-12 Geometry of trapezoidal deck beams

web comp. flange stiffener aspect ratio neutral axis subelement depth/width stiffener yield stress thickness web angle
27 h/t bc/t ds/t (h/sinθh)/bc bc/(2bt) (bc-ws)/(2t) ds/((bc-ws)/2) n fy t θh
members max min max min max min max min max min max min max min max min max min max min max min
Bernard et al. 94 94 231 77 16 1 1.25 0.42 1.64 0.55 81 61 0.26 0.01 1 0 95 95 0.023 0.023 78 78
web comp. flange web angle aspect ratio neutral axis tens. flange web comp. flange yield stress thickness
98 h/t bc/t θh (h/sinθh)/bc bc/(2bt) stiffeners stiffeners stiffeners fy t
Hoglund 203 51 191 61 84 52 1.91 0.70 1.69 0.60 2 0 2 0 2 0 61 43 0.045 0.017
It is the intent of this Specification that as more cross-sections are verified for use in the Direct
Strength Method that these tables and sections would be augmented with new information.
Those companies with proprietary sections may wish to perform their own testing and follow the
details of Chapter F of the main Specification to independently justify the use of higher φ and
lower Ω factors for a particular cross-section. Alternatively, members which fall outside the
scope of this testing may still use the method of Appendix 1, but with the reduced φ and
increased Ω factors consistent with any rational analysis method.
1.1.2 Elastic Buckling

Thin-walled cold-formed steel members have multiple elastic buckling modes that are important
for design. What is elastic buckling? Elastic buckling is a phenomenon that occurs when the
change in energy associated with out-of-plane deformation response to an in-plane load is equal
to the change in energy for in-plane response to the same in-plane load. The elastic buckling load
is the load in which the equilibrium of the member is neutral between the two alternative states:
buckled and straight. Using finite strip analysis the elastic buckling response of seven different
cold-formed steel members under a variety of different in-plane loads, both flexural and
compression, are summarized in Figure 1.1-5 through Figure 1.1-11.
The figures demonstrate that there are at least three relevant modes for elastic buckling response
of cold-formed steel members: local, distortional, and global. All modes are not guaranteed to
occur in all members, but many common members: lipped channels and zeds for example, do
include all three modes. Design of cold-formed steel members must include consideration of all
relevant buckling modes.
Traditionally, the main Specification has only specifically addressed local and global buckling
modes. Additionally, the main Specification’s approach to local buckling is to conceptualize the
member as a collection of “elements” and investigate local buckling of each element separately.
Appendix 1 - 6
The method of this Appendix provides a means to incorporate local, distortional and global
buckling (all relevant modes) into the design process. Further, all buckling modes are determined
for the member as a whole rather than element by element. This insures compatibility and
equilibrium are maintained at element junctures. Consideration of interaction amongst elements
in local buckling is a key difference between the elastic bucking calculations of the main
Specification and those of this Appendix.
Consider, as an example, the lipped channel shown in pure compression in Figure 1.1-5. This
member is the same as design example I-8 from the AISI (1996) design manual. The member’s
local elastic buckling load from the finite strip analysis (see Figure 1.1-5) is:
Pcrl = 0.13·50.42 = 6.55 kips .
The column has a gross area (Ag) of 0.917 in2, implying a member local buckling stress of
fcrl = Pcrl/Ag = 7.15 ksi .
The main Specification determines a ‘k’ value for each element, then fcr, and finally the effective
width. The centerline dimensions (ignoring corner radii) are h = 8.94 in., b = 2.94 in., d = 0.47
in., and t = 0.06 in. For this example, lets consider just the fcr of each element as determined from
the main Specification:
lip: k = 0.43, fcrl-lip= 0.43[π2E/(12(1-µ2))](t/d)2 = 186.8 ksi
flange: k = 4, fcrl-flange= 4.0[π2E/(12(1-µ2))](t/b)2 = 44.4 ksi
web: k = 4, fcrl-web= 4.0[π2E/(12(1-µ2))](t/h)2 = 4.8 ksi
Each element predicts a different buckling stress, even though the member is a connected group
of “elements”. These differences in the buckling stress are ignored in the main Specification. The
high flange and lip buckling stresses have little relevance given how low the web buckling stress
is. Comparisons to the distortional buckling stress (fcrd) using k from B4.2 of the main
Specification do no better (see Schafer and Peköz 1999 for beams and Schafer 2002 for
columns). The finite strip analysis, which includes the interaction amongst the elements, shows
that the flange aids the web significantly in local buckling, increasing the web buckling stress by
50% over the simply supported value of 4.8 ksi, but the buckling stress in the flange and lip are
much reduced due to the same interaction.
Whether calculated numerically, or by hand, the Direct Strength Method relies on an accurate
determination of the member elastic buckling stress. Thus, using different elastic buckling
stresses for each element, and ignoring distortional buckling does not suffice in this approach.
On one hand this makes the calculations required greater. On the other hand, elastic buckling is
the key basic concept for cold-formed steel structures – effort spent on this calculation is
rewarded with more efficient members and structures. Further, numerical analysis methods are
now freely available for completing this calculation.
1.1.2.1 Elastic Buckling Numerical Solutions

A variety of numerical methods: finite element, finite differences, boundary element, generalized
beam theory, finite strip analysis, and others may provide accurate elastic buckling solutions for
cold-formed steel beams and columns.
The Appendix states the basic requirements for these numerical methods: accuracy in classic
problems should be no more than 1% greater than exact solutions and compatibility and
Appendix 1 - 7
equilibrium must be maintained, or if neglected done so conservatively. The final comment

refers primarily to hand techniques such as the element method for local buckling prediction –
this method ignores compatibility and equilibrium of the elements, but by using a minimum
solution from all the elements, does so in a conservative fashion.
Traditional finite element analysis using thin plate or shell elements may be used for elastic
buckling prediction. Due to the common practice of using polynomial shape functions, the
number of elements required for reasonable accuracy can be significant. Finite element analysis
books such as Cook (1989) and Zienkiewicz and Taylor (1989, 1991) explain the basic theory;
while a number of commercial implementations, including but not limited to: ABAQUS,
MSC/NASTRAN, MARC, ANSYS, and STAGS can all provide accurate elastic buckling
answers if implemented with care. Note, in finite element implementations a large number of
“higher modes” may have to be investigated in order to reveal the three fundamental modes of
interest for a given member.
Finite difference solutions for plate stability are implemented in Harik et al. (1991) and others.
The approach is feasible, but it is not known if commercial or academic software is currently
available for this method. The boundary element method may also be used for elastic stability
(Elzein 1991). Examples demonstrate that the method can provide sufficient accuracy, but it is
not known if commercial or academic software is available for this method. Generalized beam
theory, developed by Schardt (1989), extended by Davies et al. (1994) and implemented by
Davies and Jiang (1996, 1998), and others has been shown to be a useful tool for elastic stability
analysis of cold-formed steel members. The ability to separate the different buckling modes
makes the method especially amenable to design methods. Generalized beam theory codes have
been developed in England with Davies’ group and in Portugal with Camotim’s group, but
details of their distribution to the general population are unknown.
Finite strip analysis is a specialized variant of the finite element method. For elastic stability of
cold-formed steel structures it is one of the most efficient and popular methods. Cheung and
Tham (1998) explains the basic theory while Hancock et al. (2001) and Schafer (1997) provide
specific details for stability analysis with this method. Hancock and his researchers (see Hancock
et al. 2001 for full references and descriptions) pioneered the use of finite strip analysis for
stability of cold-formed steel members and convincingly demonstrated the important potential of
finite strip analysis in both cold-formed steel design and behavior. Hancock’s research group
distributes the commercial program THIN-WALL for finite strip analysis
(www.civil.usyd.edu.au/case/THINWALL.htm). Schafer’s research group has also used finite
strip analysis extensively and make available the academic program, CUFSM, for finite strip
analysis (www.ce.jhu.edu/bschafer/cufsm). Also, RSG Software’s program, CFS, performs finite
strip analysis (www.rsgsoftware.com) in addition to strength calculations by the AISI
Specification.
The Direct Strength method of this appendix emphasizes the use of finite strip analysis for
elastic buckling determination. Finite strip analysis is a general tool that provides accurate elastic
buckling solutions with a minimum of effort and time. Finite strip analysis, as implemented in
conventional programs, does have limitations, the two most important ones are
• the model assumes the ends of the member are simply supported, and
• the cross-section may not vary along its length (uniform properties longitudinally).
Appendix 1 - 8
These limitations preclude some analysis from readily being used with the finite strip method,
but despite these limitations the tool is useful, and a major advance over plate buckling solutions
and plate buckling coefficients “k’s” that only partially account for the important stability
behavior of cold-formed steel members.
The American Iron and Steel Institute has sponsored research that, in part, has lead to the
development of the freely available program, CUFSM, which employs the finite strip method for
elastic buckling determination of any cold-formed steel cross-section. The program is available
at www.ce.jhu.edu/bschafer/cufsm and runs on any PC with Windows 9x, NT, 2000, XP. A
complete set of tutorials and examples are available online at the same address. Information
available online includes:
• Introduction and Common Questions
What is CUFSM? What is Finite Strip?
What are the system requirements? What is a buckling curve?
Why would I use CUFSM? What is a buckling mode?
Why do 2 versions of CUFSM exist? What is the half-wavelength?
Why would I use the Matlab version? What is the load factor?
CUFSM is free? What are Mcr and Pcr?
How can I use finite strip in design?
• Installation • Advanced Functions
• Overview of program features Boundary conditions
Constraints
• Tutorials Springs
Manipulating a Cee section in bending
Multiple materials
SSMA Cee in Compression (Pcrl and Pcrd) Orthotropic material
LGSI Zee in Bending (Mcrl and Mcrd) • Advanced Ideas
• Advanced Matlab Defining buckling modes
scripting Why define buckling modes?
parametric studies Understanding higher modes
Utilizing higher modes
• Appendix: Finite Strip Theory
Handling indistinct modes
Solution accuracy
The information is designed to allow someone with no experience with this type of numerical
analysis to become a proficient and conscientious user of the tool. The software and its source
code are freely available in an attempt to insure that the solution is not a “black box” for the user.
Additionally, hand methods provided in this commentary may be used as a sanity check on any
analysis to insure solutions are at least within expected bounds.
1.1.2.1.1 Numerical, Finite strip analysis notes on Pcrl, Mcrl

Prediction of local buckling modes for a variety of different members in compression and
bending using the finite strip analysis program CUFSM are demonstrated in Figure 1.1-5 through
Figure 1.1-11. In the finite strip method members are loaded with a reference stress distribution:
pure compression for finding Pcr pure bending for finding Mcr. The reference stress distributions
are shown in the figures. Determination of the buckling mode requires consideration of the half-
wavelength and the mode shape.
Appendix 1 - 9
Half-wavelength
Local buckling minima occur at half-wavelengths that are less than the largest characteristic
dimension of the member under compressive stresses. For the examples of Figure 1.1-5 through
Figure 1.1-11, this length has been demarcated with a short vertical dashed line. For instance, the
largest outside dimension for the lipped channel of Figure 1.1-5 is 9 in., therefore the cutoff (and
the vertical line) for local buckling is at 9 in. Minima in the buckling curve shown in the Figures
that fall at half-wavelengths less than this length may generally be considered as local buckling
modes. Buckling modes occurring at longer lengths are either distortional or global in nature.
The criteria of limiting the half-wavelength for local bucking to less than the largest outside
dimension under compressive stresses is based on the following. Local buckling of a simply
supported plate in pure compression occurs in square waves, i.e., it has a half-wavelength that is
equal to the plate width (the largest outside dimension). If any stress gradient exists on the plate,
or any beneficial restraint is provided to the edges of the plate by other elements, the critical
half-wavelength will be less than the width of the plate. Therefore, local buckling, with the
potential for stable post-buckling response, is assumed to occur only when the critical half-
wavelength is less than the largest potential “plate” (i.e., outside dimension with compressive
stresses applied) in a member.
Mode shape
Local buckling involves significant distortion of the cross-section, but this distortion involves
only rotation, not translation at the fold lines of the member. The mode shapes for members with
edge stiffened flanges such as those of Figure 1.1-5 and Figure 1.1-6 provide a direct comparison
between the difference between local buckling and distortional bucking. Note the behavior at the
flange/lip junction – for local buckling only rotation occurs, for distortional buckling translation
occurs.
Complications
Local buckling may be indistinct from distortional buckling in some members, for example
Figure 1.1-9 shows a simple angle that may be considered as local buckling by the main
Specification, but is considered distortional buckling here, because of the half-wavelength of the
mode, and the characteristics of the mode shape. By the definitions of this Appendix no local
buckling mode exists for this member. Local buckling may be at half-wavelengths much less
than the characteristic dimension if intermediate stiffeners are in place, or if the element
undergoes large tension and small compressive stress. For instance Figure 1.1-15 demonstrates
local buckling in negative bending that exhibits this trait.
Additional examples, and tutorials are available online at www.ce.jhu.edu/bschafer/cufsm.
1.1.2.1.2 Numerical, finite strip analysis notes on Pcrd, Mcrd

Prediction of distortional buckling modes for a variety of different members in compression and
pure compression for finding Pcr, pure bending for finding Mcr. The reference stress distributions
wavelength and the mode shape.
Appendix 1 - 10
Half-wavelength
Distortional buckling occurs at a half-wavelength intermediate to local and global buckling
modes, as shown in the figures. The half-wavelength is typically several times larger than the
largest characteristic dimension of the member. The half-wavelength is highly dependent on both
the loading and the geometry.
Mode shape
Distortional buckling involves both translation and rotation at the fold line of a member.
Distortional buckling involves distortion of one portion of the cross-section and predominately
rigid response of a second portion. For instance, the edge stiffened flanges of Figure 1.1-5 and
Figure 1.1-6 are primarily responding as one rigid piece while the web is distorting.
Complications
Distortional buckling may be indistinct (without a minimum) even when local buckling and long
half-wavelength (global) buckling are clear - Figure 1.1-5 and Figure 1.1-6 show this basic
behavior. For some members distortional buckling may not occur, as in Figure 1.1-8. (Actually, a
close examination of higher modes shows the distortional mode in this example, but the low
values of the global torsional buckling mode dominate the response at both intermediate and
long half-wavelengths.)
Further examples, and discussion of handling indistinct buckling modes, understanding higher
modes, etc. are all available online at www.ce.jhu.edu/bschafer/cufsm.
1.1.2.1.3 Numerical, finite strip analysis notes on Pcre, Mcre

Prediction of global buckling modes for a variety of different members in compression and
pure compression for finding Pcr, pure bending for finding Mcr. The reference stress distributions
wavelength and the mode shape. Global bucking modes for columns include: flexure, torsion and
flexural-torsional buckling. For beams bent about their strong-axis, lateral-torsional buckling is
the global buckling mode of interest.
Half-wavelength
Global (or “Euler”) buckling modes: flexural, torsional, or flexural-torsional for columns, lateral-
torsional for beams, occur as the minimum mode at long half-wavelengths.
Mode Shape
Global buckling modes involve translation (flexure) and/or rotation (torsion) of the entire cross-
section. No distortion exists in any of the elements in the long half-wavelength buckling modes.
Complications
Flexure and distortional buckling may interact at relatively long half- wavelengths making it
difficult to determine long column modes at certain intermediate to long lengths. Finite strip
analysis assumes simply supported ends. When long column end conditions are not simply
supported, or when they are dissimilar for flexure and torsion, higher modes may need to be
considered, or classical long column calculations performed.
Appendix 1 - 11
Higher modes for the lipped channel of Figure 1.1-5 are given in Figure 1.1-12. Based on the
boundary conditions, the effective length, KL, for a given mode can be determined. With KL
known then Pcre for that mode may be read directly from the finite strip at a half-wavelength of
KL by using the curve that corresponds to the appropriate mode. Mixed flexural and torsional
boundary conditions may not be directly treated. Further, in some cases a mode of interest may
occur at too high a load factor for practical consideration. In these cases, the classical long
column calculations, such as given in section 1.1.2.2.3 of this Appendix may be used.
Further examples and discussion are available online at www.ce.jhu.edu/bschafer/cufsm.
Appendix 1 - 12
1.5
AISI96 Ex. 1−8
My=135.91kip−in.
1
Mcr / My
Lateral−torsional
Local Mcr/My=0.67 Distortional M cr/My=0.50

0.5
0
0 1 2 3
10 10 10 10
half−wavelength (in.)
0.4
AISI96 Ex. 1−8
0.35
Py=50.42kips
0.3
Flexural−torsional
0.25
Pcr / Py
0.2
0.15 Local Pcr/Py=0.13 Distortional Pcr/Py=0.20
0.1
0.05
0
0 1 2 3
10 10 10 10
Figure 1.1-5 Finite Strip Analysis of 9CS3x060 of AISI ’96 Example I-8 in Bending and Compression
Appendix 1 - 13
1.5
AISI96 Ex. 1−10
My=112.82kip−in.
1
Mcr / My
Lateral−torsional
Local Mcr/My=0.86 Distortional M cr/My=0.65

0.5
0
0 1 2 3
10 10 10 10
0.5
AISI96 Ex. 1−10

0.45
0.4
Py=46.55kips
0.35
0.3
Flexural
Pcr / Py
0.25
Distortional Pcr/Py=0.27
0.2
Local Pcr/Py=0.16
0.15
0.1
0.05
0
0 1 2 3
10 10 10 10
Figure 1.1-6 Finite Strip Analysis of 8Z2.5x060 of AISI ’96 Example I-10 in Bending and Compression
Appendix 1 - 14
AISI96 Ex. 1−9
2.5
My=20.78kip−in.
2
Mcr / My
1.5
Lateral−torsional
Distortional M cr/My=1.44
1
0.5
0
0 1 2 3
10 10 10 10
Figure 1.1-7 Finite Strip Analysis of 5.5CU1.25x057 of AISI ’96 Example I-9 in Bending
AISI96 Ex. 1−11

0.9
0.8
Py=28.17kips
0.7
0.6
Pcr / Py
0.5
Torsional
0.4
Local Pcr/Py=0.49
0.3
0.2
0.1
0
0 1 2 3
10 10 10 10
Figure 1.1-8 Finite Strip Analysis of 4LS4x060 of AISI ’96 Example I-11 in Compression
Appendix 1 - 15
1.5
AISI96 Ex. 1−12
My=2.12kip−in.
1
Mcr / My
Lateral−torsional
0.5
0
0 1 2 3
10 10 10 10
3.5
AISI96 Ex. 1−12
My=2.12kip−in.
2.5
2
Mcr / My
1.5
Lateral−torsional
0.5
0
0 1 2 3
10 10 10 10
Figure 1.1-9 Finite Strip Analysis of 2LU2x060 of AISI ’96 Example I-12 in Positive and Negative Bending
Appendix 1 - 16
AISI96 Ex. 1−13

7
My=87.00kip−in.
5
Mcr / My
Lateral−torsional
3
Local Mcr/My=3.20
0
0 1 2 3
10 10 10 10
AISI96 Ex. 1−13

4.5 Py=95.50kips
3.5
3
Pcr / Py
2.5
Flexural−torsional
2 Local Pcr/Py=2.41
1.5
0.5
0
0 1 2 3
10 10 10 10
Figure 1.1-10 Finite Strip Analysis of 3HU4.5x135 of AISI ’96 Example I-13 in Bending and Compression
Appendix 1 - 17
1.5
AISI96 Approx. Ex. 1−14
My=21.35kip−in.
Mcr / My
0.5
Lateral−torsional
Local Mcr/My=0.42
0
0 1 2 3
10 10 10 10
1.5
AISI96 Approx. Ex. 1−14
My=21.35kip−in.
1
Mcr / My
0.5
Local Mcr/My=0.60 Lateral−torsional
0
0 1 2 3
10 10 10 10
Figure 1.1-11 Finite Strip Analysis of Wall Section of AISI ’96 Example I-14 in Positive and Negative
Bending
Appendix 1 - 18
0.9
AISI96 Ex. 1−8
0.8
Py=50.42kips Flexural−torsional (2)

0.7
0.6
Pcr / Py
0.5
0.4
0.3
Flexural
0.2 Flexural−torsional (1)
0.1
0
50 100 150 200 250 300 350 400 450 500 550 600
Figure 1.1-12 Finite strip analysis of lipped channel of AISI ’96 example I-8 showing long half-wavelength
and higher buckling modes for examination of global buckling behavior
Appendix 1 - 19
1.1.2.2 Elastic Buckling Hand Solutions
1.1.2.2.1 Local Buckling by Hand: Pcrl, Mcrl

Two hand solution methods for local buckling prediction are provided: (1.1.2.2.1.1) the element
method and (1.1.2.2.1.2) the semi-empirical interaction method. The element method provides a
simple lowerbound solution with wide applicability, while the semi-empircal interaction method
provides a more accurate local buckling solution which includes the interaction of any two
elements (e.g., the web and flange), but has limited applicability.
The element method (1.1.2.2.1.1) employs plate buckling solutions for the elastic local buckling
stress in a manner somewhat similar to Chapter B of the main Specification. An appropriate plate
buckling coefficient, or k value, must be found for each element making up a member. Each k
value implies a buckling stress for a given element. Thus, the member buckling stress remains
unknown as each element provides a different prediction. In Chapter B of the main Specification
this fact is ignored, and along with it compatibility and equilibrium of the elements. The element
method takes a conservative approach to this problem and assumes the minimum buckling stress
of all the elements governs.
The element method for local buckling prediction may be excessively conservative. The semi-
empirical interaction method (1.1.2.2.1.2) provides a more accurate local buckling prediction by
including the equilibrium and compatibility of any two elements, e.g. the flange and the web.
This interaction may be quite important, for example, Figure 1.1-13 shows the local plate
buckling coefficient for a uniformly compressed flange (b), as influenced by the depth of the
web (h), and the type of stress gradient on the web (ξ). Ignoring interaction, the flange plate
buckling coefficient k = 4, with interaction, Figure 1.1-13 demonstrates the appropriate k.
ξ=3
4.5 ξ=2 local buckling
ξ = 1.4 finite strip analysis
ξ=1
4
plate buckling coefficient for the flange kf
3.5 ξ=3
2.5
ξ = 2, pure bending
1.5
1
ξ = 1.4
ξ=1
0.5
0
1 2 3 4 5 6 7 8 9 10
ratio of web height to flange width (h/b)
Figure 1.1-13 Influence of interaction of two elements (flange and web) on local buckling solution
Appendix 1 - 20
1.1.2.2.1.1 Hand: Element Method for Local Buckling

The element method is the simplest and most conservative approximation of the elastic
local buckling load or moment for a member. To determine local buckling by the element
method: assume all elements of the member are simply supported, calculate the buckling stress
of each element, and take the minimum of the predicted element buckling stresses as the
buckling stress of the member. The element method provides a lower bound approximation to
the actual local buckling stress.
For columns,
Pcrl = Agfcrl (C-1.1.2-1)
Ag = gross area
fcrl = local buckling stress (see below)
For beams,
Mcrl = Sgfcrl (C-1.1.2-2)
Sg = gross section modulus to the extreme compression fiber
fcrl = local buckling stress at the extreme compression fiber (see below)
fcrl may be found as the minimum of the plate buckling solution of all elements:
2
π2E  t 
f crl = k   (C-1.1.2-3)
( )
12 1 − µ 2  w 
E = Young’s Modulus
µ = Poisson’s ratio
t = element thickness
w = element flat width
k = element (plate) buckling coefficient (see below)
Stiffened or edge stiffened element under stress gradient1:

k = 0.5ξ 3 + 4ξ 2 + 4 (C-1.1.2-4)
ξ=(f1-f2)/f1 (C-1.1.2-5)
Stiffened or edge stiffened element in pure compression (f1 = f2, ξ = 0):
k=4 (C-1.1.2-6)
Stiffened or edge stiffened element in pure bending (f1 = -f2, ξ = 2):
k =24 (C-1.1.2-7)
Unstiffened element2 under stress gradient:
k = 1.4ξ 2 − 0.25ξ + 0.425 for 0 ≤ ξ ≤ 1.1 (C-1.1.2-8)
k = 13ξ − 65.5ξ + 131ξ − 80 for 1.1 < ξ ≤ 2
3 2
(C-1.1.2-9)
ξ=(f1-f2)/f1 (C-1.1.2-10)
Unstiffened element2 in pure compression (f1 = f2, ξ = 0):
k = 0.425 (C-1.1.2-11)
where f1 and f2 are the stresses on opposite ends of the element, f1>f2,
compression stresses are positive and tension stresses are negative.
Appendix 1 - 21
(1) The expression given for the plate buckling coefficient under a stress gradient is slightly
different than that currently used in the main Specification; the given expression is accurate over
a greater range of stress gradient values and was first presented in Schafer and Peköz (1999).
(2) The larger compressive stress (f1) is assumed to be at the simply supported end of the
unstiffened element. The main Specification provides no k values for this case; the expression
given was first presented in Schafer and Peköz (1999). If the half-wavelength for buckling of the
member in this mode is greater than the largest outside width of the member, then “local
buckling” of an unstiffened element should be considered as distortional buckling of the
member. Thus, the formulas for unstiffened elements given above are most applicable to lips
(serving as edge stiffeners) and small outstands. For unlipped channels or other members in
which the entire flange is an unstiffened element, buckling of the unstiffened element is
distortional buckling of the member. Rational analysis using a numerical method, such as finite
strip, provides a more meaningful answer in these cases.
Since each element of a member has a different width, w, one must compare fcr, not k, to
determine the element with the minimum stress. Further, for elements under a stress gradient,
since buckling stress is typically determined from the flat portion of element, the predicted
buckling stress may have to be converted to the extreme compression fiber to compare against
other elements.
Note, stiffened elements and edge stiffened elements have the same plate buckling coefficients
(k) because in local buckling an edge stiffened element behaves as a stiffened element. The
mode of buckling, in which the edge stiffener does not fully support the element is termed
distortional buckling, and should be calculated using methods given for distortional buckling in
the following section.
Sub-elements: those elements between a corner and an intermediate stiffener, or between

intermediate stiffeners, should be treated as stiffened elements - for the purposes of calculating
the local buckling stress. If the distance between the corner and the intermediate stiffener is bs,
then bs is the plate width, w, for use in C-1.1.2-3 above. If the sub-element, bs, is under a stress
gradient, then the stresses f1 and f2 are the stresses at the ends of bs, for use in C-1.1.2-5 above.
For an intermediately stiffened element, buckling of the entire element, with the stiffener
included, is a form of distortional buckling, and should be determined as such. Numerical
analysis methods, such as finite strip analysis, provide the most accurate elastic buckling values
in such cases.
Commentary Table C-B2-1 provides additional long plate buckling solutions for k. Upperbound
approximations to the local buckling stress may be found by using the fixed boundary condition
solutions, instead of the simply supported solutions, as given in the table. Use of the fixed
boundary condition solutions for the elements is generally inappropriate for use in the Direct
Strength method (this Appendix) as it is difficult to estimate the degree to which the actual
member elastic buckling stress has been unconservatively predicted. Therefore, the simply
supported k values, given above, are suggested for general use in the element method for elastic
buckling prediction.
Appendix 1 - 22
Predictions of the half-wavelength for a buckling mode are not required for strength
determination. However, for comparisons to numerical analysis, or to provide more insight in a
given situation, knowing the half-wavelength can be useful. The approximate half-wavelength
for local buckling of a stiffened or edge stiffened element, based on expressions determined in
Schafer (1997), are:
L = βw (C-1.1.2-12)
β = −0.06ξ + 0.58ξ − 1.97ξ + 2.40ξ + 0.04 for ξ > 1
4 3 2
(C-1.1.2-13)
β = 1.0 for ξ ≤ 1 (C-1.1.2-14)
ξ = (f1 - f2)/f1 (C-1.1.2-15)
L = half-wavelength for local buckling
w = element flat width
f1 and f2 are the stresses on opposite ends of the element, f1>f2,
1.1.2.2.1.2 Hand: Interaction Method for Local Buckling

In many cases the element method for local buckling determination gives excessively
conservative predictions of the member buckling stress. For limited cases hand solutions are
available that improve upon the traditional plate buckling solutions. These solutions consider the
local buckling stress of any two connected elements. For instance, a typical lipped channel,
consisting of a stiffened element (the web) an edge stiffened element (the flange) and an
unstiffened element (the lip) – may be considered by finding the local buckling stress for the
flange/web combination and the flange/lip combination and taking the minimum of these two
values (as opposed to finding individual values for the flange, web, and lip and taking the
minimum of those three values as is done in the element method). Specifically:
For columns,
Pcrl = Agfcrl (C-1.1.2-16)
For beams,
Mcrl = Sgfcrl (C-1.1.2-17)
where fcrl may be found as the minimum of the plate buckling solutions given below:
For any two stiffened or edge stiffened elements of uniform thickness, t, connected together, one
of width, h, the second element of width, b, and both elements loaded in pure compression, the
local buckling stress of the two combined elements is (Schafer 2002):
2
π2 E  t 
f crl =k   (C-1.1.2-18)
( )
12 1 − µ 2  b 
( )
k = 2 − (b h ) ⋅ 4(b h )
0.4 2
if h/b ≥ 1 (C-1.1.2-19)
k = (2 − (h b ) )⋅ 4
0.2
if h/b < 1 (C-1.1.2-20)
For any two stiffened or edge stiffened elements of uniform thickness, t, connected together, one
of width, h, the second element of width, b, with element h under a stress gradient (ξ) and
Appendix 1 - 23
element b loaded in pure compression, the local buckling stress of the two combined elements
is (Schafer 2001):
2
π2 E  t 
f crl = k w   (C-1.1.2-21)
( )
12 1 − µ 2  h 
2
 k 2wss − k m k wss  b  b k1

kw =  
 h  + k m for < (C-1.1.2-22)
 k1   h k wss
2
h b k1
k w = k1   for ≥ (C-1.1.2-23)
b h k wss
k wss = 0.5ξ3 + 4ξ 2 + 4 (C-1.1.2-24)
k m = 7.7ξ + 4 2
(C-1.1.2-25)
k1 = −0.08(ξ − 3) + 4.72
2
(C-1.1.2-26)
ξ = (f1 − f 2 ) f1 (C-1.1.2-27)
f1 and f2 are the stresses on opposite ends of element h, f1>f2,
For an edge stiffened element of width, b, in which the edge stiffener is a simple lip (unstiffened
element) of width, d, (but any angle θ) and element b is under pure compression, and element
d is under a stress gradient, the local buckling stress of the two combined elements is (Schafer
and Peköz 1999):
2
π2 E  t 
f crl =k   (C-1.1.2-28)
( )
12 1 − µ 2  b 
k = (8.55ξ − 11.07 )(d b ) + (− 1.59ξ + 3.95)(d b ) + 4 for ξ ≤ 1 and d/b ≤ 0.6 (C-1.1.2-29)
2
ξ = (f1 − f 2 ) f1 (C-1.1.2-30)
f1 and f2 are the stresses on opposite ends of element d, f1>f2,
f1 is the stress at the end of element d, connected to b.
The above formulas allow for an accurate calculation of the local buckling stress of lipped
channels and zeds. They are applicable to other cases as well and may be compared (combined)
with individual element solutions in order to find the minimum buckling stress. The formulas
given here are not intended to preclude the creation, or use of additional local buckling formulas
that use rational analysis to arrive at the critical elastic local buckling load or moment.
1.1.2.2.2 Distortional Buckling by Hand: Pcrd, Mcrd

Available hand methods for distortional buckling include: members with edge stiffened flanges
and elements with intermediate stiffener(s). The first case, members with edge stiffened flanges,
includes lipped channels and zeds in compression or bending, and applies to any case where the
flange of a member connected to a flat web of width, h, may buckle in the distortional mode. The
second case, elements with intermediate stiffener(s), include hats and decks with intermediate
stiffeners in the flanges or in the webs, and applies to any stiffened element with single or
multiple intermediate stiffener(s) under compression or bending.
Appendix 1 - 24
The available hand formulas for members with edge stiffened flanges calculate distortional
buckling as a member phenomena; however, the available formulas for elements with
intermediate stiffeners only calculate distortional buckling as an element phenomena. Therefore,
the hand solutions of 1.1.2.2.2.1 are directly applicable in the Direct Strength Method, while the
hand solutions of 1.1.2.2.2.2 and 1.1.2.2.2.3 may require additional care. Distortional buckling
formula require a significant amount of hand calculation to provide an accurate solution.
Numerical analysis, such as the finite strip method, provides a useful alternative.
For elements with intermediate stiffener(s) (sections 1.1.2.2.2.2 and 1.1.2.2.2.3) member
distortional buckling will be well approximated by distortional buckling of the element only if
the element governs the distortional buckling mode of the member. This implies that any other
element connected to the element with an intermediate stiffener does not require a net rotational
or translational restraint prior to buckling. This situation may be approximated by determining
the local bucking stress of any connected elements and insuring that the local bucking stress of
any connected element is greater than the distortional buckling stress of the element with an
intermediate stiffener. If this criteria cannot be met, then numerical analysis should be employed
to properly account for the interaction of the elements in the member distortional buckling.
1.1.2.2.2.1 Hand: Distortional Buckling of Members with Edge Stiffened Flanges

Based on Schafer and Peköz (1999) for beams and Schafer (2002) for columns, distortional
buckling of members with edge stiffened flanges may be predicted by:
For columns,
Pcrd =Agfcrd (C-1.1.2-31)
Ag = gross area of the member
fcrd = distortional buckling stress (see below)
For beams,
Mcrd = Sgfcrd (C-1.1.2-32)
Sg = gross section modulus to the extreme compression fiber
fcrd = distortional buckling stress at the extreme compression fiber (see below)
where:
k φfe + k φwe
f crd = ~ ~ (C-1.1.2-33)
k φfg + k φwg
L = min(L cr , L m )
Flange Rotational Stiffness:

4 2  2
 EI (x − h )2 + EC − E I xyf (x − h )2  +  π  GJ
(k φf )e =  Lπ  (C-1.1.2-34)
    L
xf o x wf o x f
I yf

2  2  
(~k )
φf g
 π  
=   A f  (x o − h x )

 I yf 

2  I xyf 
 I xyf 
− 2 y o (x o − h x )
 I yf 
 + h 2x + y o2  + I xf + I yf 
 (C-1.1.2-35)
L        
   
Flexural Member: Critical Length and Web Rotational Stiffness
Appendix 1 - 25
( )
1
 4π 4 h 1 − ν 2  2  4 4  4
L cr =   I (x − h )2 + C − I xyf (x − h )2  + π h  (C-1.1.2-36)
 t3  xf o x wf
I yf
o x
 720 
   
Et 3  3  π  19h  π  h 3 
2 4
k φwe = +  +  (C-1.1.2-37)
( )
12 1 − ν 2  h  L  60  L  240 
 2 2 
 (45360(1 − ξ ) + 62160 ) L  + 448π 2 +  h  (53 + 3(1 − ξ ))π 4 
web web
~ htπ 2  h L 
k φwg =  2 4  (C-1.1.2-38)
13440  4 2 L  L 
 π + 28 π   + 420  
 h h 
Compression Member: Critical Length and Web Rotational Stiffness

( )
1
 6π 4 h 1 − ν 2  2 
4
L cr =  I (x − h )2 + C − I xyf (x − h )2   (C-1.1.2-39)
 t3  xf o x wf
I yf
o x

  
Et 3
k φwe = (C-1.1.2-40)
(
6h 1 − ν 2 )
32
~  π  th
k φwg =   (C-1.1.2-41)
 L  60
G = Shear Modulus
ν = Poisson’s Ratio
Lm = Distance between restraints which restrict distortional buckling
t = plate thickness
h = web depth
ξweb = (f1- f2)/f1 = stress gradient in the web, where f1 and f2 are the stresses at the
opposite ends of the web, f1>f2, compression positive, tension negative, calculated on
the basis of the gross section, (e.g., pure symmetrical bending, f1=-f2, ξweb = 2)
Af, Ixf, Iyf, Ixyf, Cwf, Jf, xo, and hx, are properties of the compression flange (flange and
edge stiffener) about x, y axes. Where the x, y axes are located at the centroid of the
flange, with the x-axis measured positive to the right from the centroid, and the y-axis
positive down from the centroid. For a flange with a simple lip stiffener (i.e, the flange of
a typical Cee or Zee) appropriate expressions from Schafer (1997) are given below.
Af = cross-sectional area of the flange
Ixf = x-axis moment of inertia of the flange
Iyf = y-axis moment of inertia of the flange
Ixyf = product of the moment of inertia of the flange
Cwf = warping torsion constant of the flange
Jf = St. Venant torsion constant of the flange
xo = x distance from the flange/web junction to the centroid of the flange
hx = x distance from the centroid of the flange to the shear center of the flange
Appendix 1 - 26
b b
θ d θ
h h
A f = (b + d )t A f = (b + d )t
3 3 3 3
Jf = 1
3 bt + 3 dt
1 Jf = 1
3 bt + 3 dt
1
I xf =
(
t t b + 4bd + t 2 bd + d 4
2 2 3
) I xf =
(
t t b + 4bd − 4bd 3 cos 2 (θ) + t 2 bd + d 4 − d 4 cos 2 (θ)
2 2 3
)
12(b + d ) 12(b + d )
I yf =
(
t b 4 + 4db 3 ) I yf =
(
t b 4 + 4db 3 + 6d 2 b 2 cos(θ) + 4d 3b cos 2 (θ) + d 4 cos 2 (θ ) )
12(b + d ) 12(b + d )
tb 2 d 2 tbd 2 sin (θ)(b + d cos(θ ))
I xyf = I xyf =
4(b + d ) 4(b + d )
I of = 1
3 tb
3
+ 112 bt 3 + 1 3 td 3 I of = 1
3 tb 3 + 112 bt 3 + 1
3 td
3
C wf = 0 C wf = 0
b2 b 2 − d 2 cos(θ)
xo = xo =
2(b + d ) 2(b + d )
− d2 − d 2 sin (θ)
h y = yo = h y = yo =
2(b + d ) 2(b + d )
hx =
− b 2 + 2db( ) hx =
(
− b 2 + 2db + d 2 cos(θ) )
2(b + d ) 2(b + d )
b = xo − hx b = xo − h x
Figure 1.1-14 Geometric flange properties for lipped channel and lipped zeds
Similar distortional buckling hand formulas for columns are available in Lau and Hancock
(1987) and for beams in Hancock et al. (1996) and Hancock (1997).
For flanges with unusual edge stiffeners (compound lips, etc.) the properties of the flange must
be determined (no explicit formula are given). However, in some cases explicit formulas may be
found in the literature (e.g., Bambach et al. 1998).
Distortional buckling of members with unstiffened flanges (unlipped channels, etc) may be
approximated by setting d to 0 in the expressions of Figure 1.1-14.
1.1.2.2.2.2 Hand: Distortional buckling of a stiffened element with an intermediate stiffener under
a stress gradient
For flexural members with single intermediate stiffeners in the web, explicit formula for
distortional buckling of the element are available based on the work of Schafer (1997).
Appendix 1 - 27
A
f1 f1 x
c
f2 f2
a AA
A
y
Figure 1.1-15 Stiffened element with an intermediate stiffener under a stress gradient
Sg = gross section modulus to the largest compressive edge of the element (f1 end)
fcrd = distortional buckling stress at the largest compressive edge of the element (f1 end)
2
π2E  t 
f crd = kd   (C-1.1.2-43)
( )
12 1 − µ 2  b 
b = element (web) width
kd = element (plate) buckling coefficient for distortional buckling
− r − r 2 − 4qs
kd = (C-1.1.2-44)
2q
(
q = β 4 BG + D 2 − E
2
) (C-1.1.2-45)
r = β 2 (2 γG + BH + DI ) (C-1.1.2-46)
s = 2 γH + J (C-1.1.2-47)
α = c/b (C-1.1.2-48)
γ = (EIs)/(bDpl), where, Dpl = (Et3)/(12(1-µ2)) (C-1.1.2-49)
δ = As/(bt) (C-1.1.2-50)
ξ = (f1 – f2)/f1 (C-1.1.2-51)
1
  παξ   γ  4 
4
β =  sin 2     + βns 

3
(C-1.1.2-52)
  1.2   0.17ξ  
for ξ > 1 β ns = −0.06ξ + 0.58ξ − 1.97ξ 2 + 2.40ξ + 0.04
4 3
(C-1.1.2-53)
for ξ ≤ 1 βns = 1.0 (C-1.1.2-54)
B = −2δ(1 − ξα ) (C-1.1.2-55)
D = −(1 − ξ 2 ) (C-1.1.2-56)
E = − (16ξ ) (9π 2 ) (C-1.1.2-57)
G = D(S11 + S 22 ) − 2ES12 (C-1.1.2-58)
H = C 2S11 + C1S22 (C-1.1.2-59)
I = C1 + C 2 (C-1.1.2-60)
J = C1C 2 (C-1.1.2-61)
Appendix 1 - 28
C1 = (1 + β 2 )
2
(C-1.1.2-62)
(
C 2 = 1 + 4β 2 2
) (C-1.1.2-63)
S11 = sin 2 (πα ) (C-1.1.2-64)
S22 = sin 2 (2πα ) (C-1.1.2-65)
S12 = sin (πα )sin (2πα ) (C-1.1.2-66)
where:
c = distance from f1 edge of the element to the centerline of the stiffener
Is = moment of inertia of the stiffener1
As = area of the stiffener1
f1, f2 = stresses on opposite ends of element h, f1>f2, compression stresses
are positive and tension stresses are negative.
(1) The stiffener moment of inertia may conservatively be taken as the moment of inertia about
the stiffener centroidal axis, parallel to the axis of the main element, b. Schafer (1997) provides a
more general approach for determining stiffener area and moment of inertia for use in the
formulas above. However, for highly complicated stiffener geometries numerical analysis
provides a more accurate and efficient result.
1.1.2.2.2.3 Hand: Distortional buckling of a stiffened element with intermediate stiffener(s) under
pure compression
The distortional buckling stress of an element under pure compression with intermediate
stiffener(s) may be approximated as:
Columns,
Pcrd =Agfcrd (C-1.1.2-67)
fcrd = distortional buckling stress of the intermediately stiffened element (see below)
Beams,
Sg = gross section modulus to the stress in the intermediately stiffened element
fcrd = distortional buckling stress of the intermediately stiffened element (see below)
2
π2E  t 
f crd = kd   (C-1.1.2-69)
( )
12 1 − µ 2  b 
b = element width
kd = element buckling coefficient for distortional buckling, Spec. Eq. B5.1.2-2
Main specification section B5 provides the elastic buckling solution (both local and distortional)
for elements with intermediate stiffeners in pure compression. The commentary to section B5 of
Appendix 1 - 29
the main specification provides the development of kd, further details can be found in Schafer
and Peköz (1998).
1.1.2.2.3 Euler buckling by Hand: Pcre, Mcre

The long half-wavelength “Euler” buckling mode for beams and columns is calculated with
some care in the main Specification. Therefore, for both beams and columns, extensive hand
expressions are already available and may be used for hand calculation of Pcre and Mcre. See the
commentary to main Specification sections C4 and C3 for additional details.
For columns,
Pcre =Agfcre (C-1.1.2-70)
fcre = minimum of the elastic critical flexural, torsional, or flexural-torsional buckling
stress. fcre is equal to Fe of Section C4 of the main Specification. The hand methods
presented in Specification sections C4.1 through C4.4 provide all necessary
formula. Note, C4.4 specifically addresses the long-standing practice that Fe (or fcre)
may be calculated by rational analysis. Rational analysis hand solutions to long
column buckling are available - see the commentary for main Specification section
C4.4 as well as Yu (2001) or Hancock et al. (2001). The hand calculations may be
quite lengthy, particular if member properties xo and Cw are unknown.
For beams,
Mcre = Sgfcre (C-1.1.2-71)
Sg = gross section modulus to the extreme fiber in compression
fcre = elastic critical lateral-torsional buckling stress. fcre is equal to Fe of main
Specification section C3.1.2.1. for open cross-section members and C3.1.2.2 for
closed cross-section members. Hand solutions are well established for doubly- and
singly-symmetric sections, but not so for point symmetric sections (Zeds). Fe of
point-symmetric sections is taken as ½ of the value for doubly-symmetric sections.
Rational numerical analysis may be desirable in cases where a more exacting
solution is required.
1.1.3 Deflection Determination

The provisions of this Appendix use a simplified approach to deflection calculations that assume
the moment of inertia of the section for deflection calculations is linearly proportional to the
strength of the section, determined at the allowable stress of interest. This approximation avoids
lengthy effective section calculations for deflection determination.
1.2 Columns
Commentary section C4 provides a complete discussion on the behavior of cold-formed columns
as it relates to the main Specification. This commentary addresses the specific issues raised by
the use of the Direct Strength Method of Appendix 1 for the design of cold-formed columns.
The thin-walled nature of cold-formed columns complicates behavior and design. Elastic
buckling analysis (e.g., Figure 1.1-5 through Figure 1.1-11) reveals at least three buckling
modes: local, distortional, and Euler (flexural, torsional, or flexural-torsional) that must be
Appendix 1 - 30
considered in design. Therefore, in addition to usual considerations for steel columns: material
non-linearity (e.g., yielding), imperfections, and residual stresses; the individual role and
potential for interaction of buckling modes must also be considered. The Direct Strength Method
of this Appendix emerged through the combination of more refined methods for local and
distortional buckling prediction, improved understanding of the post-buckling strength and
imperfection sensitivity in distortional failures, and the relatively large amount of available
experimental data.
Fully effective or compact columns are generally well predicted by conventional column curves
(AISC (2001), Galambos (1998), etc.). Therefore, the long column strength Pne, follows the same
practice as the main Specification and uses the current AISC (2001) curves for strength
prediction. The main Specification provides the long column strength in terms of a stress, Fn (Eq.
C4-2 and C4-3). In the Direct Strength method this is converted from a stress to a strength by
multiplying times the gross area, Ag, resulting in the formulas for Pne given in Appendix 1.
The main Specification calculates the column strength by multiplying the long column stress, Fn,
times the effective area, Ae, calculated at Fn. This accounts for local buckling reductions in the
actual column strength (i.e., local-global interaction). The Direct Strength Method chooses to
break this calculation into two parts: the long column strength without any reduction for local
buckling (Pne) and the long column strength considering local-global interaction (Pnl).
The strength curves for local and distortional buckling of a fully braced column are presented in
Figure 1.2-1. The curves are presented as a function of slenderness – which in this case refers to
slenderness in the local or distortional mode – as opposed to traditional long column slenderness.
For non-slender (a.k.a., fully effective, or compact) columns Pcr>>Py and the cross-section is
stable and the capacity is equal to the squash load, Py of the column. Slightly more slender
columns behave inelastically, with Pn<Pcr and Pn<Py. Slender cross-sections behave primarily
elastically, but elastic behavior in a thin-walled column implies post-buckling capacity, with
Pn>Pcr. Practical cold-formed steel column sizes exist in all three ranges: fully effective, Pn=Py;
inelastic Pn<Pcr and Pn<Py; and post-buckling Pn>Pcr and Pn<Py. Figure 1.2-1 also shows that at
the if the local and distortional slenderness is the same, the distortional mode is assumed to have
a reduced post-buckling capacity compared with the local mode.
The development and calibration of the Direct Strength provisions for columns is reported in
Schafer (2000, 2002). The reliability of the column provisions was determined using the test data
of section 1.1.1.1 and the provisions of Chapter F of the main Specification. Based on a target
reliability, β, of 2.5 a resistance factor, φ, of 0.84 was calculated for all the investigated columns.
Based on this information the safety and resistance factors of section 1.2 were determined for the
pre-qualified members. Since the range of pre-qualified members is relatively large extensions of
the Direct Strength Method to geometries outside the pre-qualified set is allowed. However,
given the uncertain nature of this extension reduced resistance factors and increased safety
factors are applied in that case.
The provisions of Appendix 1, applied to the columns of section 1.1.1.1, are summarized in
Figure 1.2-2 below. The controlling strength is either by section 1.2.2 which considers local
buckling interaction with long column buckling, or by section 1.2.3 which considers the
Appendix 1 - 31
distortional mode alone. The controlling strength (minimum predicted of the two modes) is
highlighted for the examined members by the choice of marker. Overall performance of the
method can be judged by examination of Figure 1.2-2. Scatter exists throughout the data set, but
the trends in strength are clearly shown, as is the fact that for the same slenderness distortional
buckling systematically has a reduced capacity. The scatter shown in the data is similar to that of
the main Specification, and in fact since the main Specification has no rules for distortional
buckling the Direct Strength Method actually provides better agreement than the main
Specification when compared with this test database for many members.
1.5
Distortional: Eq. 1.2.3-2
Local: Eq. 1.2.2-2
1 Elastic
inelastic region
Pn
Py
0.5
post-buckling reserve
0
0 1 2 3 4 5 6 7 8
slenderness = Py Pcr
Figure 1.2-1 Direct Strength Method column curves for a fully braced column
Appendix 1 - 32
1.5
distortional
local
 Ptest  Distortional: Eq. 1.2.3-2

 
 P  1
 y d Local: Eq. 1.2.2-2
or
 Ptest 
 
 Pne  l
0.5
0
0 1 2 3 4 5 6 7 8
λ d = Py Pcrd or λ l = Pne Pcrl
Figure 1.2-2 Direct Strength Method for concentrically loaded pin-ended columns
1.2.1 Flexural, Torsional, or Flexural-Torsional Buckling

As discussed in detail above, the strength expressions for long wavelength buckling of columns
follow directly from section C4 of the main Specification. These provisions are identical to those
used for compact section hot-rolled columns in the AISC Specification (2001) and are fully
discussed in the commentary to section C4. The axial capacity Pne, calculated in this section
represents the upperbound capacity for a given column. Actual column strength is determined by
considering reductions that may occur due to local buckling, and performing a separate check on
the distortional mode.
See section 1.1.2 for information on rational analysis methods for calculation of Pcre.

The expression selected for local buckling of columns is shown in Figure 1.2-1 and Figure 1.2-2
and discussed above. The potential for local-global interaction is presumed, thus the column
strength in local buckling is limited to a maximum of the long column strength, Pne.
See section 1.1.2 for information on rational analysis methods for calculation of Pcrl.

The expression selected for distortional buckling of columns is shown in Figure 1.2-1 and Figure
1.2-2 and discussed above. Based on experimental test data and on the success of the
Australian/New Zealand code (see Hancock et al. 2001 for discussion and Hancock et al. 1994
for further details) the distortional buckling calculation is completed on Py instead of Pne. This
Appendix 1 - 33
presumes that distortional buckling failures are independent of long-column behavior – i.e., little
if any distortional-global interaction exists.
See section 1.1.2 for information on rational analysis methods for calculation of Pcrd.
1.3 Beams
Commentary section C3 provides a complete discussion on the behavior of cold-formed beams
as it relates to the main Specification. This commentary addresses the specific issues raised by
the use of the Direct Strength Method of Appendix 1 for the design of cold-formed beams.
The thin-walled nature of cold-formed beams complicates behavior and design. Elastic buckling
analysis (e.g., Figure 1.1-5 through Figure 1.1-11) reveals at least three buckling modes: local,
distortional, and lateral-torsional buckling (for members in strong-axis bending) that must be
considered in design. The Direct Strength Method of this Appendix emerged through the
combination of more refined methods for local and distortional buckling prediction, improved
understanding of the post-buckling strength and imperfection sensitivity in distortional failures,
and the relatively large amount of available experimental data.
The lateral-torsional buckling strength, Mne, follows the same practice as the main Specification.
The main Specification provides the lateral-torsional buckling strength in terms of a stress, Fc
(Eq. C3.1.2.1-2, -3, and -4 ). In the Direct Strength method this is converted from a stress to a
moment by multiplying times the gross section modulus, Sg, resulting in the formulas for Mne
given in Appendix 1.
For beams that are not fully braced and locally unstable the main Specification calculates the
beam strength by multiplying the predicted stress for failure in lateral-buckling, Fc, times the
effective section modulus, Sc, determined at stress Fc. This accounts for local buckling
reductions in the lateral-torsional beam strength (i.e., local-global interaction). The Direct
Strength Method chooses to break this calculation into two parts: the lateral-torsional beam
strength without any reduction for local buckling (Mne) and the strength considering local-global
interaction (Mnl).
The strength curves for local and distortional buckling of a fully braced beam are presented in
Figure 1.3-1. The curves are presented as a function of slenderness. For non-slender (a.k.a., fully
effective, compact, or locally stable) columns Mcr>>My and the cross-section is stable and the
capacity is equal to the yield moment, My of the beam. Inelastic reserve capacity is currently not
considered in the Direct Strength Method. Slightly more slender beams behave inelastically, with
Mn<Mcr and Mn<My. Slender beams behave primarily elastically, but elastic behavior in a thin-
walled beam implies post-buckling capacity, with Mn>Mcr. Practical cold-formed steel beams
exist in all three ranges: fully effective, Mn=My; inelastic Mn<Mcr and Mn<My; and post-buckling
Mn>Mcr and Mn<My. Figure 1.3-1 also shows that at the if the local and distortional slenderness
is the same, the distortional mode is assumed to have a reduced post-buckling capacity compared
with the local mode.
Appendix 1 - 34
The reliability of the beam provisions was determined using the test data of section 1.1.1.2 and
the provisions of Chapter F of the main Specification. Based on a target reliability, β, of 2.5 a
resistance factor, φ, of 0.90 was calculated for all the investigated beams. Based on this
information the safety and resistance factors of section 1.3 were determined for the pre-qualified
members. Since the range of pre-qualified members is relatively large extensions of the Direct
Strength Method to geometries outside the pre-qualified set is allowed. However, given the
uncertain nature of this extension reduced resistance factors and increased safety factors are
applied in that case.
The provisions of Appendix 1, applied to the beams of section 1.1.1.2, are summarized in Figure
1.3-2 below. The controlling strength is either by section 1.3.2 which considers local buckling
interaction with lateral-torsional buckling, or by section 1.3.3 which considers the distortional
mode alone. The controlling strength (minimum predicted of the two modes) is highlighted for
the examined members by the choice of marker. Overall performance of the method can be
judged by examination of Figure 1.3-2. The scatter shown in the data is similar to that of the
main Specification, and in fact since the main Specification has no rules for distortional buckling
the Direct Strength Method actually provides better agreement than the main Specification when
compared with this test database for many members.
1.5
Local: Eq. 1.3.2-2
1 Elastic Buckling
inelastic region
Mn
My
0.5 post-buckling reserve
0
0 1 2 3 4 5
slenderness = M y M cr
Figure 1.3-1 Direct Strength method beam strength curves for a laterally braced beam
Appendix 1 - 35
1.5
Distortional
Local

1
Local: Eq. 1.3.2-2
M test
My
0.5
0
0 1 2 3 4 5
λmax = M y M cr
Figure 1.3-2 Direct Strength Method for Laterally Braced Beams
1.3.1 Lateral-Torsional Buckling

As discussed in detail above, the strength expressions for lateral-torsional buckling of beams
follows directly from section C3 of the main Specification and are fully discussed in the
commentary to section C3. The bending capacity Mne, calculated in this section represents the
upperbound capacity for a given beam. Actual beam strength is determined by considering
reductions that may occur due to local buckling, and performing a separate check on the
distortional mode.
See section 1.1.2 for information on rational analysis methods for calculation of Mcre.

The expression selected for local buckling of beams is shown in Figure 1.3-1 and Figure 1.3-2
and discussed above. The use of the Direct Strength Method for local buckling and the
development of the empirical strength expression is given in Schafer and Peköz 1998b. The
potential for local-global interaction is presumed, thus the beam strength in local buckling is
limited to a maximum of the lateral-torsional beam strength, Mne. For fully braced beams the
maximum Mne value is the yield moment, My.
See section 1.1.2 for information on rational analysis methods for calculation of Mcrl.
Appendix 1 - 36

The expression selected for distortional buckling of beams is shown in Figure 1.3-1 and Figure
1.3-2 and discussed above. Based on experimental test data and on the success of the
Australian/New Zealand code (see Hancock 2001 for discussion) the distortional buckling
calculation is completed on My instead of Mne. This presumes that distortional buckling failures
are independent of lateral-torsional buckling behavior – i.e., little if any distortional-global
interaction exists.
See section 1.1.2 for information on rational analysis methods for calculation of Mcrd.
References
Acharya, V.V. and R.M. Schuster (1998), “Bending Tests of Hat Section with Multiple
Longitudinal Stiffeners,” Proceedings of the Fourteenth International Specialty Conference
on Cold-Formed Steel Structures, University of Missouri-Rolla, Rolla, MO, October, 1998.
AISC (2001). Manual of Steel Construction: Load and Resistance Factor Design 3rd Ed.
American Institute of Steel Construction, Chicago, IL.
AISI (1996). Cold-Formed Steel Design Manual. American Iron and Steel Institute, Washington,
D.C.
Bambach, M.R., Merrick, J.T., Hancock, G.J. (1998). “Distortional Buckling Formulae for Thin
Walled Channel and Z-Sections with Return Lips.” Proceedings of the 14th International
Specialty Conference on Cold-Formed Steel Structures, St. Louis, Missouri, 21-38.
Bernard, E.S. (1993). “Flexural Behavior of Cold-Formed Profiled Steel Decking.” Ph.D. Thesis,
University of Sydney, Australia.
Cheung, Y.K., Tham, L.G. (1998). Finite Strip Method. CRC Press.
Cohen, J. M. (1987). “Local Buckling Behavior of Plate Elements.” Department of Structural
Engineering Report, Cornell University, Ithaca, New York.
Cook, R.D., Malkus, D.S., Plesha, M.E. (1989). Concepts and Applications of Finite Element
Analysis. John Wiley & Sons, 3rd Ed.
Davies, J.M., Jiang, C. (1996). “Design of Thin-Walled Beams for Distortional Buckling.”
Proceedings of the Thirteenth International Specialty Conference on Cold-Formed Steel
Structures, 141-154, St. Louis, Missouri.
Davies, J.M., Jiang, C., Ungureanu, V. (1998). “Buckling Mode Interaction in Cold-Formed
Steel Columns and Beams.” Proceedings of the 14th International Specialty Conference on
Cold-Formed Steel Structures, St. Louis, Missouri, 53-68.
Davies, J.M., Leach, P., Heinz, D. “Second-Order Generalised Beam Theory.” J. of Const. Steel
Res., Elsevier, 31 (2-3) 221-242.
Desmond, T.P. (1977). “The Behavior and Design of Thin-Walled Compression Elements with
Longitudinal Stiffeners.” Ph.D. Thesis, Cornell University, Ithaca, New York.
Ellifritt, D., Glover, B., Hren, J. (1997) “Distortional Buckling of Channels and Zees Not
Attached to Sheathing.” Report for the American Iron and Steel Institute, Washington D.C.
Elzein, A. (1991). Plate Stability by Boundary Element Method. Springer-Verlag, New York.
Appendix 1 - 37
Galambos, T.V. (1998). Guide to Stability Design Criteria for Metal Structures. John Wiley &
Sons, 5th Ed.
Hancock, G.J. (1997). “Design for Distortional Buckling of Flexural Members.” Thin-Walled
Structures, 27(1), 3-12, Elsevier Science Ltd.
Hancock, G.J., Kwon, Y.B., Bernard, E.S. (1994). “Strength Design Curves for Thin-Walled
Sections Undergoing Distortional Buckling.” J. of Const. Steel Res., Elsevier, 31(2-3), 169-
186.
Hancock, G.J., Murray, T.M., Ellifritt, D.S. (2001). Cold-Formed Steel Structures to the AISI
Specification. Marcell-Dekker, New York, New York.
Hancock, G.J., Rogers, C.A., Schuster, R.M. (1996). “Comparison of the Distortional Buckling
Method for Flexural Members with Tests.” Proceedings of the Thirteenth International
Specialty Conference on Cold-Formed Steel Structures, 125-140, St. Louis, MO.
Harik, I.E., Liu, X., Ekambaram, R. (1991). “Elastic stability of plates with varying rigidities.”
Computers and Structures, 38 (2) 161-168
Höglund, T. (1980). “Design of Trapezoidal Sheeting Provided with Stiffeners in the Flanges
and Webs.” Swedish Council for Building Research, Stockholm, Sweden, D28:1980.
König, J. (1978). “Transversally Loaded Thin-Walled C-Shaped Panels With Intermediate
Stiffeners.” Swedish Council for Building Research, Stockholm, Sweden, D7:1978.
Kwon, Y.B., and Hancock, G.J., (1992) “Strength Tests of Cold-Formed Channel Sections
undergoing Local and Distortional Buckling.” J. of Struct. Eng., ASCE, 117(2), 1786 – 1803.
LaBoube, R.A., Yu, W. (1978). “Structural Behavior of Beam Webs Subjected to Bending
Stress.” Civil Engineering Study Structural Series, 78-1, Department of Civil Engineering,
University of Missouri-Rolla, Rolla, Missouri.
Lau, S.C.W., and Hancock, G.J (1987). ”Distortional Buckling Formulas for Channel Columns.”
J. of Struct. Eng., ASCE, 113(5), 1063 – 1078.
Loughlan, J. (1979). “Mode Interaction in Lipped Channel Columns under Concentric or
Eccentric Loading.” Ph.D. Thesis. University of Strathclyde, Glasgow.
Miller, T.H., Peköz, T. (1994). “Load-Eccentricity Effects on Cold-Formed Steel Lipped-
Channel Columns.” J. of Struct. Eng., ASCE, 120(3), 805-823.
Moreyra, M.E. (1993). “The Behavior of Cold-Formed Lipped Channels under Bending.” M.S.
Thesis, Cornell University, Ithaca, New York.
Mulligan, G.P. (1983). “The Influence of Local Buckling on the Structural Behavior of Singly-
Symmetric Cold-Formed Steel Columns.” Ph.D. Thesis. Cornell University. Ithaca, New
York.
Papazian, R.P., Schuster, R.M., Sommerstein, M. (1994). "Multiple Stiffened Deck Profiles."
Proceedings of the Twelfth International Specialty Conference on Cold-Formed Steel
Structures, University of Missouri-Rolla, 217-228.
Phung, N., Yu, W.W. (1978). "Structural Behavior of Longitudinally Reinforced Beam Webs."
Civil Engineering Study Structural Series, Department of Civil Engineering, 78-6, University
of Missouri-Rolla.
Appendix 1 - 38
Polyzois, D., Charnvarnichborikarn, P. (1993). “Web-Flange Interaction in Cold-Formed Steel

Z-Section Columns”. J. of Struct. Eng., ASCE, 119(9), 2607-2628.
Quispe, L., Hancock, G.J. (2002). “Direct Strength Method for the Design of Purlins.”
Proceedings of the 16th International Specialty Conference on Recent Research and
Developments in Cold-Formed Steel Design and Construction, Orlando, FL. 561-572.
Rogers, C.A. (1995). "Interaction Buckling of Flange, Edge Stiffener and Web of C-Sections in
Bending." M.S. Thesis, University of Waterloo, Ontario, Canada.
Schafer, B.W. (1997). “Cold-Formed Steel Behavior and Design: Analytical and Numerical
Modeling of Elements and Members with Longitudinal Stiffeners,” Ph.D. Thesis, Cornell
University, Ithaca, New York..
Schafer, B.W. (2000). “Distortional Buckling of Cold-Formed Steel Columns: Final Report.”
Sponsored by the American Iron and Steel Institute, Washington, D.C.
Schafer, B.W. (2001) “Progress Report 2: Test Verification of the Effect of Stress Gradient on
Webs of Cee and Zee Sections” submitted to the AISI and MBMA.(July 2001)
Schafer, B.W. (2002). “Local, Distortional, and Euler Buckling in Thin-walled Columns.”
scheduled for publication March, ASCE Journal of Structural Engineering.
Schafer, B.W. (2002b). “Progress on the Direct Strength Method.” Proceedings of the 16th
International Specialty Conference on Recent Research and Developments in Cold-Formed
Steel Design and Construction, Orlando, FL. 647-662.
Schafer, B.W., Peköz, T. (1998). “Cold-Formed Steel Members with Multiple Longitudinal
Intermediate Stiffeners in the Compression Flange.” Journal of Structural Engineering.
124(10).
Schafer, B.W., Peköz, T. (1998b). “Direct Strength Prediction of Cold-Formed Steel Members
using Numerical Elastic Buckling Solutions.” Fourteenth International Specialty Conference
on Cold-Formed Steel Structures. St. Louis, Missouri.
Schafer, B.W., Peköz, T. (1999). “Laterally Braced Cold-Formed Steel Flexural Members with
Edge Stiffened Flanges.” J. of Struct. Eng.. 125(2).
Schardt, R. (1989). Verallgemeinerte Technische Biegetheorie [Generalized Beam Theory].
Springer-Verlag, Berlin.
Schardt, R. Schrade, W. (1982). “Kaltprofil-Pfetten.” Institut Für Statik, Technische Hochschule
Darmstadt, Bericht Nr. 1, Darmstadt.
Schuster, R.M. (1992). “Testing of Perforated C-Stud Sections in Bending.” Report for the
Canadian Sheet Steel Building Institute, University of Waterloo, Waterloo Ontario.
Shan, M., LaBoube, R.A., Yu, W. (1994). “Behavior of Web Elements with Openings Subjected
to Bending, Shear and the Combination of Bending and Shear.” Civil Engineering Study
Structural Series, 94-2, Department of Civil Engineering, University of Missouri-Rolla,
Rolla, Missouri.
Thomasson, P. (1978). “Thin-walled C-shaped Panels in Axial Compression”. Swedish Council
for Building Research. D1:1978, Stockholm, Sweden.
Appendix 1 - 39
Willis, C.T., Wallace, B. (1990). “Behavior of Cold-Formed Steel Purlins under Gravity
Loading.” Journal of Structural Engineering, ASCE. 116(8).
Yu, W.W. (2000). Cold-Formed Steel Design. John Wiley & Sons, Inc.
Zienkiewicz, O.C., Taylor, R.L. (1989). The Finite Element Method: Volume 1 Basic
Formulations and Linear Problems. McGraw Hill, 4th Ed.
Zienkiewicz, O.C., Taylor, R.L. (1991). The Finite Element Method: Volume 2 Solid and Fluid
Mechanics Dynamics and Non-linearity. McGraw Hill, 4th Ed.
Appendix 1 - 40
EXAMPLES

OF
Contents:
STRENGTH AND STIFFNESS

1. C-Section with lips: Bending, Compression (AISI ’96 Ex. I-8)
2. C-Section without lips: Bending (AISI ’96 Ex. I-9)
3. Z-Section with lips: Bending, Comp., Deflections (AISI ’96 Ex. I-10)
4. L-Section with lips: Compression (AISI ’96 Ex. I-11)
5. L-Section without lips: (+) & (-) Bending (AISI ’96 Ex. I-12)
6. H-Section: Bending, Compression (AISI ’96 Ex. I-13)
7. Wall Section: (+) & (-) Bending, Deflections (AISI ’96 Ex. I-14)
ELASTIC BUCKLING
1. C-Section With Lips - Axial: Elastic Buckling Load by Hand
2. C-Section With Lips – Bending: Elastic Buckling Moment by Hand
DRAFT Direct Strength Examples December 20, 2001
DIRECT STRENGTH EXAMPLE I-8 C-SECTION WITH LIPS

Based on AISI'96 Ex. I-8
Given:
1. Section 9CS3x060, yield stress 55 ksi, and finite strip analysis results (as shown below)
Required:
1. Bending capacity for fully braced member
2. Compression capacity at a uniform compressive stress of 37.67 ksi
1. Computation of bending capacity for a fully braced member

Finite strip analysis of section in pure bending results:
Inputs from the finite strip analysis include:

My 135.91 . kip . in
M crl 0.67 . M y M crl = 91.06 kip . in
M crd 0.50 . M y M crd = 67.955 kip . in
per 1.3, Mn is the minimum of Mne, Mnl, Mnd. For a fully braced member lateral-torsional buckling will
not occur and thus Mne = My, Mnl and Mnd must still be checked.
M ne My M ne = 135.91 kip . in (fully braced)

Local buckling check per 1.3.2
M ne
λl λ l = 1.222 (subscript "l" = "l") (Eq. 1.3.2-3)
M crl
M nl M ne if λ l 0.776 (Eq. 1.3.2-1)

(Eq. 1.3.2-2)
0.4 0.4
M crl M crl
1 0.15 . .M
ne if λ l > 0.776
M ne M ne
M nl = 100.995 kip . in
Distortional buckling check per 1.3.3
My
λd λ d = 1.414 (Eq. 1.3.3-3)
M crd
M nd M y if λ d 0.673 (Eq. 1.3.3-1)

(Eq. 1.3.3-2)
0.5 0.5
M crd M crd
1 0.22 . .M
y if λ d > 0.673
My My
M nd = 81.153 kip . in
Predicted bending capacity per 1.3

Mn min M ne M nl M nd M n = 81 kip . in
compare vs. AISI 1996 Ex. 1-8
1.75 . in . 55 . ksi M nAISI96 = 96.25 kip . in

3
M nAISI96
The predicted capacity in local buckling (Mnl) is greater than AISI '96, but due to inclusion of a
check on distortional buckling the Direct Strength method predicts lower capacity than AISI '96
for this member. Members with small lip lengths, as in this example, have systematically lower
strength predictions than AISI '96, Schafer (2000) provides an extensive discussion on this issue,
as it relates to columns. In short, the Direct Strength Method specifically encourages longer lip
lengths than AISI '96.
The geometry of this section fall within the "pre-qualified" beams of 1.1.1.2 and the lower φ and
higher Ω of section 1.3 may therefore be used.
Schafer, B.W. (2000). "Distortional Buckling of Cold-Formed Steel Columns." Final Report to the
American Iron and Steel Institute, Washington DC, August, 2000.
summary materials without design examples and industry impact discussion may be found in
Schafer, B.W. (2002). "Local, Distortional, and Euler Buckling in Thin-Walled Columns" ASCE,
Journal of Structural Engineering, (in press - scheduled for March)

Finite strip analysis of section in pure compression results:

50.42 . kip P y . ( 55 . ksi )
1 2
Py Ag A g = 0.917 in
P crl 0.13 . P y P crl = 6.555 kip
P crd 0.20 . P y P crd = 10.084 kip
per 1.2, Pn is the minimum of Pne, Pnl, Pnd. In AISI '96 effective area calculations for columns which
interact with long column buckling are determined at a given stress (Fn ). A similar procedure is done
in the Direct Strength method by limiting Pnl to Pne at the same stress (Fn ), e.g.
P ne A g . 37.67 . ksi P ne = 34.533 kip

P ne
λl λ l = 2.295 (subscript "l" = "l") (Eq. 1.2.2-3)
P crl
P nl P ne if λ l 0.776 (Eq. 1.3.2-1)

(Eq. 1.3.2-2)
0.4 0.4
P crl P crl
1 0.15 . .P
ne if λ l > 0.776
P ne P ne
P nl = 16.394 kip
Py
λd λ d = 2.236 (Eq. 1.2.3-3)
P crd
P nd P y if λ d 0.561 (Eq. 1.2.3-1)

(Eq. 1.2.3-2)
0.4 0.4
P crd P crd
1 0.25 . .P
y if λ d > 0.561
Py Py
P nd = 23.008 kip

Pn min P ne P nl P nd P n = 16.4 kip
0.468 . in . 37.67 . ksi

2
P nAISI96 P nAISI96 = 17.6 kip
The predicted capacities by the two methods are nearly equal in this case. For shorter column
sthe uniform compressive stress (Fn) would increase - AISI '96 woudl continue to predict
increased capacity, the Direct Strength method would also predict increased capacity up to 23
kips, and would then be limited by distortional buckling. No explicit distortional buckling check
exists in AISI '96.
The geometry of this sections fall within the "pre-qualiufied" columns of section 1.1.1.1 and
therefore may use the lower φ and higher Ω listed in section 1.2 for pre-qualified members.
DIRECT STRENGTH EXAMPLE I-9 C-SECTION WITHOUT LIPS

Given:
1. Section 5.5CU1.25x057, yield stress 33 ksi, and finite strip analysis results (as shown below)
Required:


My 20.78 . kip . in
M ne My M ne = 20.78 kip . in
Members with unstiffened flanges have a distortional buckling mode, but no local buckling mode, in
such a case local buckling may be ignored by setting Mnl = Mne (its maximum obtainable value).
M nl M ne M nl = 20.78 kip . in
My
λd λ d = 0.833 (Eq. 1.3.3-3)
M crd
M nd M y if λ d 0.673 (Eq. 1.3.3-1)

(Eq. 1.3.3-2)
0.5 0.5
M crd M crd
1 0.22 . .M
y if λ d > 0.673
My My
M nd = 18.353 kip . in

Mn min M ne M nl M nd M n = 18.4 kip . in
.580 . in . 33 . ksi M nAISI96 = 19.1 kip . in

3
M nAISI96
The results of the two methods are comparable. As the finite strip analysis shows,
lateral-torsional buckling is likely to govern the behavior even for relatively short span lengths.
The geometry of this section is "prequalified" per 1.1.1.2 and may use the lower φ and higher Ω
factors for pre-qualified members listed in section 1.3.
DIRECT STRENGTH EXAMPLE I-10 Z-SECTION WITH LIPS

Given:
1. Section 8Z2.5x060, yield stress 55 ksi, and finite strip analysis results (as shown below)
Required:
2. Effective moment of inertia at a service moment of 56.3 kip-in.


My 112.82 . kip . in
M ne
λl λ l = 1.078 (subscript "l" is "l") (Eq. 1.3.2-3)
M crl
M nl M ne if λ l 0.776 (Eq. 1.3.2-1)

(Eq. 1.3.2-2)
0.4 0.4
M crl M crl
1 0.15 . .M
ne if λ l > 0.776
M ne M ne
M nl = 91.215 kip . in
My
λd λ d = 1.24 (Eq. 1.3.3-3)
M crd
M nd M y if λ d 0.673 (Eq. 1.3.3-1)

(Eq. 1.3.3-2)
0.5 0.5
M crd M crd
1 0.22 . .M
y if λ d > 0.673
My My
M nd = 74.825 kip . in


3
M nAISI96
The predicted capacity in local buckling is nearly the same as AISI '96, but due to inclusion of
distortional buckling the Direct Strength method predicts significantly lower capacity. If the
member is attached to sheating, sheathing, or other restraint that restricts the distortional
buckling mode then the capacity may increase to a maximum of Mnl. Longer lip lengths would
also specifically increase the distortional buckling capacity.
The geometry of this Z is not "pre-qualified" based on currently published data. All criteria of
1.1.1.2 are met, except b/t is slightly too small (42 vs. a min of 45). Judgment suggests that lower
b/t's should be adequate and pre-qualified φ and Ω could be applied in this case.
Predicted moment of inertia for deflection calculation per 1.1.3
Ma 56.3 . kip . in this Ma is 60% of MnAISI96 .
per 1.1.3, Mna is the minimum of Mne, Mnl, Mnd. With My set to Ma . For a fully braced member
lateral-torsional buckling will not occur and thus Mne = Ma , find Mnl and Mnd.
My Ma M a = 56.3 kip . in
Local buckling capacity per 1.3.2 at Ma
M ne
M crl
M nl M ne if λ l 0.776 (Eq. 1.3.2-1)

(Eq. 1.3.2-2)
0.4 0.4
M crl M crl
1 0.15 . .M
ne if λ l > 0.776
M ne M ne
M nl = 56.3 kip . in
Distortional buckling capacity per 1.3.3 at Ma
My
λd λ d = 0.876 (Eq. 1.3.3-3)
M crd
M nd M y if λ d 0.673 (Eq. 1.3.3-1)

(Eq. 1.3.3-2)
0.5 0.5
M crd M crd
1 0.22 . .M
y if λ d > 0.673
My My
M nd = 48.121 kip . in
Predicted bending capacity per 1.3 at Ma

M na min M ne M nl M nd M na = 48.121 kip . in
Reduced moment of inertia
per Eq. 1.1.3-1, the effecitve moment of inertia is determined as the ratio of Mna/Ma
M na
= 0.855
Ma
8.144 . in
4
Ig based on finite strip model, vs. 8.146 calculated in AISI'96 Ex. 1-10
M na
I g.
4
I eff I eff = 6.96 in
Ma
AISI 1996 predicts fully effective at Ma
The predicted effective moment of inertia in local buckling is the same as AISI '96 (fully effective),
but again due to inclusion of distortional buckling the Direct Strength method predicts a reduced
inertia at the service moment. If calculations are performed at 60% of the Direct Strength Mn ,
then the section is predicted to be 92% effective instead of only 86%.


46.55 . kip P y . ( 55 . ksi )
1 2
Py Ag A g = 0.846 in (vs. 0.847 in2 of AISI '96 Ex. 1-10)
P crl 0.16 . P y P crl = 7.448 kip
P crd 0.27 . P y P crd = 12.569 kip
P ne A g . 26.2 . ksi P ne = 22.175 kip

P ne
P crl
P nl P ne if λ l 0.776 (Eq. 1.3.2-1)

(Eq. 1.3.2-2)
0.4 0.4
P crl P crl
1 0.15 . .P
ne if λ l > 0.776
P ne P ne
P nl = 12.943 kip
Py
λd λ d = 1.925 (Eq. 1.2.3-3)
P crd
P nd P y if λ d 0.561 (Eq. 1.2.3-1)

(Eq. 1.2.3-2)
0.4 0.4
P crd P crd
1 0.25 . .P
y if λ d > 0.561
Py Py
P nd = 23.489 kip

0.601 . in . 26.2 . ksi

2
Predicted capacity is less than AISI '96. In this case, the uniform compressive stress of 26.2 ksi
implies a relatively long column. The Direct Strength method predicts local buckling will control,
since distortional buckling is presumed to not interact with long column buckling. The introduction
of a small longitudinal stiffener in the web would increase Pcrl markedly and along with it the
compression capacity of this member.
The geometry of this column is pre-qualified per the criteria of 1.1.1.1 and may use the
pre-qualified φ and Ω listed in section 1.2
DIRECT STRENGTH EXAMPLE I-11 L-SECTION WITH LIPS

Given:
1. Section 4LS4x060, yield stress 55 ksi, and finite strip analysis results (as shown below)
Required:


28.17 . kip P y . ( 55 . ksi )
1 2
Py Ag A g = 0.512 in
P crl 0.49 . P y P crl = 13.803 kip
P ne A g . 14.89 . ksi P ne = 7.626 kip

The finite strip analysis results reveal that distoritonal buckling does not play a role in the response of
this memeber, only local and overall. Therefore, distortional buckling may be assumed equal to Py.
P nd Py
P ne
P crl
P nl P ne if λ l 0.776 (Eq. 1.3.2-1)

(Eq. 1.3.2-2)
0.4 0.4
P crl P crl
1 0.15 . .P
ne if λ l > 0.776
P ne P ne
P nl = 7.626 kip
Predicted compression capacity per 1.3

0.381 . in . 14.89 . ksi

2
The Direct Strength Method predicts no reduction due to local buckling at this stress, while AISI
1996 predicts a 26% reduction in strength. Here the difference between the two methods is quite
marked due to the edge stiffener expressions of B4.2 which partially account for distortional
buckling, but in this case appear to overly penalize the capacity. For higher compressive stresses
(shorter columns) local will interact with the long-wavelength buckling mode (note λl=0.743 and
reductions in Pne start at λl=0.776 so the section is just barely fully effective at this stress.)
The geometry of this section is not pre-qualified per 1.1.1.1 and thus the more conservative φ and
Ω of section 1.2 must be used.
DIRECT STRENGTH EXAMPLE I-12 L-SECTION WITHOUT LIPS

Given:
1. Section 2LU2x060, yield stress 33 ksi, and finite strip analysis results (as shown below)
Required:
1. Bending capacity, fully braced, bent about horizontal axis, top (outstand) in compression
2. Bending capacity, fully braced, bent about horizontal axis, bottom in compression
1. Bending capacity, fully braced, bent about horizontal axis, top (outstand) in compression

My 2.12 . kip . in
as the finite strip analysis demonstrates, there is no local buckling for an L without lips, the local
buckling capacity can be set to the yield moment capacity to ignore local bucking, Mnl=My.
M nl My
My
λd λ d = 0.985 (Eq. 1.3.3-3)
M crd
M nd M y if λ d 0.673 (Eq. 1.3.3-1)

(Eq. 1.3.3-2)
0.5 0.5
M crd M crd
1 0.22 . .M
y if λ d > 0.673
My My
M nd = 1.671 kip . in


3
M nAISI96
AISI 1996 prediction is markedly more conservative, as the k value used is far more conservative
than the actual buckling of the angle. As the finite strip analysis suggests, the bracing will have to
be nearly continuous to provide this strength, as lateral-torstional buckling occurs for even
relatively short span lengths - but the above analysis assume the angle is fully braced.
2. Bending capacity, fully braced, bent about horizontal axis, bottom in compression

My 2.12 . kip . in
as the finite strip analysis demonstrates, there is no local buckling for an L without lips, the local
buckling capacity can be set to the yield moment capacity to ignore local bucking, Mnl=My.
M nl My
My
λd λ d = 0.679 (Eq. 1.3.3-3)
M crd
M nd M y if λ d 0.673 (Eq. 1.3.3-1)

(Eq. 1.3.3-2)
0.5 0.5
M crd M crd
1 0.22 . .M
y if λ d > 0.673
My My
M nd = 2.111 kip . in


3
M nAISI96
Both methods predict essentially the same bending capacity for the bottom in compression.
The geometry of this member is not pre-qualified for bending or compression. The more
conservative φ and Ω factors must be applied in design.
DIRECT STRENGTH EXAMPLE I-13 H-SECTION

Based on AISI'96 Ex. I-13*
Given:
1. Section 3HU4.5x135, yield stress 55 ksi, and finite strip analysis results (as shown below)
Required:
1. Bending capacity, compression on top flange
2. Compression capacity at a uniform compressive stress of 55 ksi
1. Computation of Bending capacity for a fully braced member, compression on top flange
Inputs from the finite strip analysis include: *Geometry used in analysis is the same as that
used in the calculations for AISI 1996 Ex. 1-6 and
My 87.00 . kip . in 1-13, but the Figures in the AISI 1996 examples
manual are wrong, they show incorrect inner
M crl 3.20 . M y M crl = 278.4 kip . in radius and incorrect tension flange width.
M ne My M ne = 87 kip . in
For an H-section with the stiffened element flange in compression distortional buckling does not
occur (as shown above), set Mnd=My, to ignore distortional buckling
M nd My
M ne
M crl
M nl M ne if λ l 0.776 (Eq. 1.3.2-1)

(Eq. 1.3.2-2)
0.4 0.4
M crl M crl
1 0.15 . .M
ne if λ l > 0.776
M ne M ne
M nl = 87 kip . in


3
M nAISI96
Direct Strength predicts a fully effective member, while MnAISI96 based on initiation of first yield
predicts a slight reduction from My. The inelastic reserve capacity calculations have not been
extended for use in the Direct Strength Method.


95.50 . kip P y . ( 55 . ksi )
1 2
Py Ag A g = 1.736 in (vs. 1.737 in Ex. 1-6)
P crl 2.41 . P y P crl = 230.155 kip
P ne A g . 55.0 . ksi P ne = 95.5 kip

As shown above, in this section distortional buckling does not control, thinner columns of a similar
cross-section maybe different, set Pnd=Py to ignore distortional buckling.
P nd Py
P ne
P crl
P nl P ne if λ l 0.776 (Eq. 1.3.2-1)

(Eq. 1.3.2-2)
0.4 0.4
P crl P crl
1 0.15 . .P
ne if λ l > 0.776
P ne P ne
P nl = 95.5 kip

Pn min P ne P nl P nd P n = 96 kip
1.73 . in . 55.0 . ksi

2
P nAISI96 P nAISI96 = 95 kip
The predicted capacities by the two methods are nearly equal in this case. The section is
predicted to be fully effective at a compressive stress of 55.0 ksi.
The geometry of this member is not pre-qualified as a beam or a column, the w/t ratios of the
elements are generally stockier than those published. Since all w/t ratios are lower than the
published data it is left to the engineer's judgment if it is acceptable to use the pre-qualified φ and
Ω values for this member.
DIRECT STRENGTH EXAMPLE I-14 WALL PANEL SECTION

Based on AISI'96 Ex. I-14*
Given:
1. Wall panel section, with yield stress 50 ksi, and finite strip analysis results (as shown below)
Required:
1. Bending capacity for positive bending
2. Moment of inertia for deflection calculation at Ma = 7.87kip-in.
3. Bending capacity for negative bending
1. Computation of Bending capacity for positive bending


My 21.35 . kip . in * The stiffener geometry in the pan of this wall
panel is approximated from AISI '96 Ex. I-14,
M crl 0.42 . M y M crl = 8.967 kip . in as a result the moment of inertia of this
example and AISI '96 Ex. I-14 are not exactly
the same.
M ne
M crl
M nl M ne if λ l 0.776 (Eq. 1.3.2-1)

(Eq. 1.3.2-2)
0.4 0.4
M crl M crl
1 0.15 . .M
ne if λ l > 0.776
M ne M ne
M nl = 13.49 kip . in
My
λd λ d = 1.213 (Eq. 1.3.3-3)
M crd
M nd M y if λ d 0.673 (Eq. 1.3.3-1)

(Eq. 1.3.3-2)
0.5 0.5
M crd M crd
1 0.22 . .M
y if λ d > 0.673
My My
M nd = 14.412 kip . in


3
M nAISI96
Results are close for the bending capacity by the two methods.

Predicted moment of inertia for deflection calculation per 1.1.3
Ma 7.87 . kip . in Ma is the service load, D/L ratio of 1/5 and Mn =MnAISI96
per 1.1.3, Mna is the minimum of Mne, Mnl, Mnd. With My set to Ma . For a fully braced member
lateral-torsional buckling will not occur and thus Mne = Ma , find Mnl and Mnd.
My Ma M a = 7.87 kip . in
Local buckling capacity per 1.3.2 at Ma
M ne
M crl
M nl M ne if λ l 0.776 (Eq. 1.3.2-1)

(Eq. 1.3.2-2)
0.4 0.4
M crl M crl
1 0.15 . .M
ne if λ l > 0.776
M ne M ne
M nl = 6.981 kip . in
Distortional buckling capacity per 1.3.3 at Ma
My
λd λ d = 0.736 (Eq. 1.3.3-3)
M crd
M nd M y if λ d 0.673 (Eq. 1.3.3-1)

(Eq. 1.3.3-2)
0.5 0.5
M crd M crd
1 0.22 . .M
y if λ d > 0.673
My My
M nd = 7.495 kip . in
Predicted bending capacity per 1.3 at Ma
M na
min M ne M nl M nd M na = 6.981 kip . in
Reduced moment of inertia
per Eq. 1.1.3-1, the effective moment of inertia is determined as the ratio of Mna/Ma
M na based on finite strip model, vs. 0.442 in4 calculated in

0.455 . in
4
= 0.887 Ig AISI'96 Ex. I-14, this model is only an approximation -
Ma
different stiffeners used, etc.
M na
I g.
4
I eff I eff = 0.404 in
Ma
0.407 . in
4
I effAISI96
The predicted effective moment of inertia is nearly the same.
3. Computation of Bending capacity for negative bending


My 21.35 . kip . in
M ne
M crl
M nl M ne if λ l 0.776 (Eq. 1.3.2-1)

(Eq. 1.3.2-2)
0.4 0.4
M crl M crl
1 0.15 . .M
ne if λ l > 0.776
M ne M ne
M nl = 15.276 kip . in
My
λd λ d = 1.387 (Eq. 1.3.3-3)
M crd
M nd M y if λ d 0.673 (Eq. 1.3.3-1)

(Eq. 1.3.3-2)
0.5 0.5
M crd M crd
1 0.22 . .M
y if λ d > 0.673
My My
M nd = 12.953 kip . in


3
M nAISI96
In this case distortional buckling of the pan controls the Direct Strength calculation and yields a
lower prediction than AISI '96 - the Direct Strength geometry is only an approximation of the
actual geometry, but is close enough for simple comparisons. Note, that the local buckling
strength is approximately equal to the AISI '96 prediction.
Elastic Buckling Direct Strength Examples Dec. 20, 2001
ELASTIC BUCKLING EXAMPLE 1: C-SECTION WITH LIPS

Determination of Pcrl, Pcrd, Pcre by HAND for used in the Direct Strength Method
Given
1. Column Dimensions and Properties
Required
1. Pcrl by the element method of Commentary 1.1.2.1.1.1
2. Pcrl by the semi-emprical method of Commentary 1.1.2.1.1.2
3. Pcrd by the formulas of Commentary 1.1.2.1.2.1
4. Pcre by the formulas of main Specification C4
Column Dimensions
centerline dimensions of AISI '96 example I-8 are used (inclusion of
corner radii may be best for formal calc's.) This example is
demonstrative only
h 8.94 . in Kx 1
b
y d θ
b 2.94 . in Ky 1
h
d 0.47 . in Kt 0.5
x
t 0.06 . in
Lx 20 . ft
E 29500 . ksi
Ly 1 . ft
ν 0.3
Fy 55 . ksi Lt 1 . ft
π
θ 90 . canned formula used for Cw requires θ=90, modify as needed.
180
Glossary of Variables:
h = web height Kx = x-axis effective length
b = flange width Ky = y axis effective length
d = lip length
Kt = torsion effective length
θ = lip angle (radians)
t = thickness Lx = x-axis unbraced length
E = Young's modulus Ly = y-axis unbraced length
ν = Poisson's ratio Lt = torsion unbraced length
Fy = yield stress
Preliminary work: Gross Property Calculations
E 10 1 2
Material Property: G G = 7.823 10 kg m s
2.( 1 ν)
Gross Section Properties:

This is a series of "canned" formulas for gross property calculations of a lipped channel.
They do not apply to other cross-section geometry. The Cw formula is from Yu,
Cold-Formed Steel Design.
t .( h 2.b 2.d)
2
Ag A g = 0.946 in
1. . 3 1 3 1 3
2. .b.t 2. .d.t
3 4
J ht J = 1.135 10 in
3 3 3
1.
y cg h y cg = 4.47 in
2
1 . 3. 1. . . 2 2 . 3. 1. . . 2 1. . 3
d .t .h
2 4
Ix h t bth d t dth bt I x = 11.635 in
12 2 3 2 6
b.( b 2.d)
x cg x cg = 0.724 in
h 2.b 2.d
2.d)
2
1 . .3 2. . 3 1. . 3 (b
2.d.t .b ( h.t 2.b.t 2.d.t ) .b .
2 2 4
Iy ht tb dt I y = 1.009 in
2.b 2.d)
12 3 6 2
(h
b.t .( b 2.d) b.t . . . 2
3.b.h 8.d
2 3
xo 6dh x o = 1.898 in
Ag 12 . I x
m xo x cg m = 1.174 in
x cg . A g . h b2
2
C wterm1 . m
2
m .b C wterm1 = 736.376 in
6
t 3
I x.m
2
Ag
C wterm2 . m2 . h3 b d ( 2.d
2. 2.
3.h) .( 2.h 4.d) C wterm2 = 170.811 in
6
3.t t
m .d
2
C wterm3 . 8 . b2 . d 2.m .( 2.d.( d h) b.( 2.d 3.h) ) C wterm3 = 14.26 in
6
3
b .h . m .h
2 2 2 4
( 3.d b) .( 4.d 6.d
2 3 6
C wterm4 h) C wterm4 = 3.064 10 in
6 4
2
t . 6
Cw C wterm1 C wterm2 C wterm3 C wterm4 C w = 15.065 in
Ag
1. Pcrll by the element method of Commentary 1.1.2.1.1.1
Pcrl = Ag fcrl in the element method fcrl is determined for each separate element and the
minimum is used determination of element buckling stresses follow.
1(a) Flange Local Buckling:

Classical solution for a simply supported plate in pure compression is employed.
k flange 4
π .E
2 2
f cr_flange k flange . . t f cr_flange = 44.419 ksi
12 . 1
2 b
ν
1(b) Web Local Buckling:

Classical solution for a simply supported plate in pure compression is employed.
k web 4
π .E
2 2
f cr_web k web . . t f cr_web = 4.804 ksi
12 . 1
2 h
ν
1(c) Lip Local Buckling:

Classical solution for a plate simply supported on three sides and free along one edge
is employed.
k lip 0.43
π .E
2 2
f cr_lip k lip . . t f cr_lip = 186.842 ksi
12 . 1
2 d
ν
Finally, we may conclude that
f crl min f cr_flange f cr_web f cr_lip f crl = 4.804 ksi
P crl A g . f crl P crl = 4.542 kip
In this case you can see that the web is anticipated to drive the buckling of the member. The
high buckling stresses predicted for the flange and lip will not occur because of the buckling of
the web; however ignoring web/flange interaction underestimates the web buckling stress
considerably, so actual member buckling stress will be higher than the element methods
prediction.
2. Pcrll by the semi-empirical interaction method of Commentary 1.1.2.1.1.2

Pcrl = Ag fcrl in this method fcrl is determined for each separate pair of elements and the
minimum is used, determination of element-pair buckling stresses follow.
2(a) Flange / Lip Local Buckling

The expression given for flange/lip local buckling is a function of the stress gradient on the lip. This
stress gradient is defined in terms of ξ=(f1 -f2 )/f1 where f1 and f2 are the stresses at the edges of
the lip (compression positive) for pure compression f l=f2 and ξ=0
ξ 0 lip is in pure compression
2
d d
k flange_lip ( 8.55 . ξ 11.07 ) . ( 1.59 ξ 3.95 ) . 4 k flange_lip = 4.349
b b
d
note, d/b<0.6 and ξ<1 = 0.16 OK ξ=0 OK
b
π .E
2 2
f cr_flange_lip k flange_lip. . t f cr_flange_lip = 48.289 ksi
12 . 1
2 b
ν
2(b) Flange / Web Local Buckling
0.4 2
b .4. b h
k flange_web 2 if 1 k flange_web = 0.588
h h b
0.2
h .4 h
2 if <1
b b
note, that if we wanted kweb_flange, so that the k is referenced to h instead of b we simply multiply by
(h/b)2 . h
2
k web_flange k flange_web . k web_flange = 5.436
b
with this we can see the classic result of considering local buckling interaction. k for the flange has
been drastically reduced, because in this case flange buckling is driven by the web, but k for the
web has been increased because the flange provides some positive benefit in local buckling. In
reality all three elements interact, and numerical methods can provide the best answer.
π .E
2 2
f cr_flange_web k flange_web. . t f cr_flange_web = 6.529 ksi
12 . 1
2 b
ν
f crl2 min f cr_flange_web f cr_flange_lip f crl2 = 6.529 ksi
P crl2 A g . f crl2 P crl2 = 6.174 kip
Local buckling comparison:

Element method: P crl = 4.542 kip
Interaction method: P crl2 = 6.174 kip
Finite strip analysis: P crl3 6.555 . kip
3. Pcrd by the formulas of commentary 1.1.2.1.2.1
The hand methods for distortional buckling prediction require that section properties of the
isolated flange be calculated. The expressions here are only applicable for simple lips. More
complicated flanges would follow the same procedures, but new expressions would be required.
hx xo
S Material Properties:
W/F
hy yo
C x E
G
2.( 1 ν)
Properties of the Flange Only:

d) .t
2
Af (b A f = 0.205 in
1. . 3 1. . 3 4 4
Jf bt dt J f = 2.455 10 in
3 3
t . t .b 4.b.d 4 . b . d . cos( θ ) t .b.d d . cos( θ )

2 2 3 3 2 2 4 4 2
d 3 4
I xf I xf = 1.915 10 in
12 . ( b d )
t. b 4.d.b 6 . d . b . cos( θ ) 4 . d . b . cos( θ ) d . cos( θ )

4 3 2 2 3 2 4 2
4
I yf I yf = 0.18 in
12 . ( b d )
t . b . d . sin( θ ) . ( b d . cos( θ ) )
2
I xyf 3 4
4.( b d) I xyf = 8.399 10 in
t .b b.t t .d
3 3 3
I of 4
3 12 3 I of = 0.51 in
d . cos( θ )
2 2
b x distance from the centroid to
x of x of = 1.267 in
2.( b d) the shear center.
d . sin( θ )
2
y distance from the centroid to
y of
2.( b d) the shear center. y of = 0.032 in
2.d.b d . cos( θ )
2 2
b x distance from the centroid
h xf to the web/flange juncture. h xf = 1.673 in
2 (b d)
d . sin( θ )
2
y distance from the centroid
h yf h yf = 0.032 in
2.( b d) to the web/flange juncture.
0 . in
6 6
C wf C wf = 0 in
3. (continued) Pcrd by the formulas of commentary 1.1.2.1.2.1
Pcrd = Ag fcrd where determination of fcrd follows:
Determine the critical half-wavelength at which distortional buckling occurs:
1
2 4
6.π .h. 1
4 2 I xyf
ν . I . x 2 . x 2
L cr xf of h xf C wf of h xf L cr = 23.198 in
3 I yf
t
If bracing is provided that restricts the distortional mode at some length less than
Lcr, then this length should be used in place of Lcr.
Determine the elastic and "geometric" rotational spring stiffness of the flange:
4 2 2
π I xyf π
k φfe . E .I . x h xf
2
E .C E. . x h xf
2 .G .J
xf of wf of f
L cr I yf L cr
k φfe = 0.182 kip
2 2
π I xyf I xyf
k φfg . A . x h xf
2.
2 . y of . x of h xf . h xf
2
y of
2
I xf I yf
f of
L cr I yf I yf
2
k φfg = 0.014 in
Determine the elastic and "geometric" rotational spring stiffness from the web:
E .t
3
k φwe k φwe = 0.131 kip
6.h. 1
2
ν
π . 3 2
k φwg .t h k φwg = 0.013 in
2
L cr 60
Determine the distortional buckling stress:
k φfe k φwe
f crd
k φfg k φwg
f crd = 11.545 ksi
P crd A g . f crd P crd = 10.917 kip
Distortionanl buckling comparison:

Element method: P crd = 10.917 kip
Finite strip analysis: P crd2 10.084 . kip
4. Pcre by formulas of the main Specification section C4
Elastic Flexural Buckling about the x-axis:
Ix Slenderness is:
rx r x = 3.508 in
K x.L x
Ag
= 68.421
π .E
2
F ex F ex = 62.193 ksi rx
K x.L x
2
rx
Elastic Flexural Buckling about the y-axis:
Iy Slenderness is:
ry r y = 1.033 in
K y .L y
Ag
= 11.618
π .E
2
3
F ey F ey = 2.157 10 ksi ry
K y .L y
2
ry
Elastic Flexural-torsional buckling
2 2 2
σ ex F ex ro rx ry xo r o = 4.12 in
π .E .C w
2 2
1 xo
σt . G.J β 1 σ t = 7.592 10
3
ksi
A g.r o K t .L t
2 2 ro
β = 0.788
1 .
4 . β . σ ex . σ t
2
F et σ ex σt σ ex σt F et = 62.084 ksi
2.β
The controlling long-wavelength buckling load is the minimum:

Fe min F ex F ey F et F e = 62.084 ksi
mode_is "x-axis flexure" if F e F ex mode_is = "flexural torsional"
"y-axis flexure" if F e F ey
"flexural-torsional" if F e F et
P cre A g.F e P cre = 58.707 kip

A member fully braced in weak-axis and torsion, such as AISI '96 example III-1 would only use the
strong-axis flexure value..
ELASTIC BUCKLING EXAMPLE 2: C-SECTION WITH LIPS

Determination of Mcrl, Mcrd by HAND for used in the Direct Strength Method
Given
1. Beam Dimensions and Properties
Required
1. Mcrl by the element method of Commentary 1.1.2.1.1.1
2. Mcrl by the semi-emprical method of Commentary 1.1.2.1.1.2
3. Mcrd by the formulas of Commentary 1.1.2.1.2.1
Column Dimensions
centerline dimensions of AISI '96 example I-8 are used (inclusion of
corner radii may be best for formal calc's.) This example is
demonstrative only
h 8.94 . in
b
y d θ
b 2.94 . in
h
d 0.47 . in
x
t 0.06 . in
E 29500 . ksi
ν 0.3
Fy 55 . ksi
π
θ 90 . canned formula used for Cw requires θ=90, modify as needed.
180
Glossary of Variables:
h = web height
b = flange width
d = lip length
θ = lip angle (radians)
t = thickness
E = Young's modulus
ν = Poisson's ratio
Fy = yield stress
Preliminary work: Gross Property Calculations
E 10 1 2
Material Property: G G = 7.823 10 kg m s
2.( 1 ν)
Gross Section Properties:

This is a series of "canned" formulas for gross property calculations of a lipped channel.
They do not apply to other cross-section geometry. The Cw formula is from Yu,
Cold-Formed Steel Design.
t .( h 2.b 2.d)
2
Ag A g = 0.946 in
1. . 3 1 3 1 3
2. .b.t 2. .d.t
3 4
J ht J = 1.135 10 in
3 3 3
1.
y cg h y cg = 4.47 in
2
1 . 3. 1. . . 2 2 . 3. 1. . . 2 1. . 3
d .t .h
2 4
Ix h t bth d t dth bt I x = 11.635 in
12 2 3 2 6
b.( b 2.d)
x cg x cg = 0.724 in
h 2.b 2.d
2.d)
2
1 . .3 2. . 3 1. . 3 (b
2.d.t .b ( h.t 2.b.t 2.d.t ) .b .
2 2 4
Iy ht tb dt I y = 1.009 in
2.b 2.d)
12 3 6 2
(h
b.t .( b 2.d) b.t . . . 2
3.b.h 8.d
2 3
xo 6dh x o = 1.898 in
Ag 12 . I x
m xo x cg m = 1.174 in
x cg . A g . h b2
2
C wterm1 . m
2
m .b C wterm1 = 736.376 in
6
t 3
I x.m
2
Ag
C wterm2 . m2 . h3 b d ( 2.d
2. 2.
3.h) .( 2.h 4.d) C wterm2 = 170.811 in
6
3.t t
m .d
2
C wterm3 . 8 . b2 . d 2.m .( 2.d.( d h) b.( 2.d 3.h) ) C wterm3 = 14.26 in
6
3
b .h . m .h
2 2 2 4
( 3.d b) .( 4.d 6.d
2 3 6
C wterm4 h) C wterm4 = 3.064 10 in
6 4
2
t . 6
Cw C wterm1 C wterm2 C wterm3 C wterm4 C w = 15.065 in
Ag
Ix 3
S gt S gt = 2.603 in
y cg
1. Mcrll by the element method of Commentary 1.1.2.1.1.1
Mcrl = Sg fcrl in the element method fcrl is determined for each separate element and the
minimum is used determination of element buckling stresses follow.
1(a) Flange Local Buckling:
k flange 4
π .E
2 2
f cr_flange k flange . . t f cr_flange = 44.419 ksi
12 . 1
2 b
ν
1(b) Web Local Buckling:

f1 Fy f2 F y f1 and f2 are the stresses at the ends of the element of interest f1 is
the stress at the compression flange, f2 at the tension flange.
f1 f2
ξ ξ=2
f1
0.5 . ξ 4.ξ
3 2
k web 4 k web = 24
π .E
2 2
f cr_web k web . . t f cr_web = 28.823 ksi
12 . 1
2 h
ν
1(c) Lip Local Buckling:

d
f1 Fy f2 F y. 1 f 2 = 49.217 ksi
0.5 . h
f1 f2
ξ ξ = 0.105 ξ is so close to 0, that little benefit is gained from considering the
f1 stress gradient..
1.4 . ξ 0.25 . ξ
2
k lip 0.425 k lip = 0.414
π .E
2 2
f cr_lip k lip . . t f cr_lip = 179.973 ksi
12 . 1
2 d
ν

All stresses were calculated relative to the compression flange, so we may directly compare:
f crl min f cr_flange f cr_web f cr_lip f crl = 28.823 ksi
M crl S gt . f crl M crl = 75.02 kip . in
In this case you can see that the web is anticipated to drive the buckling of the member. The
high buckling stresses predicted for the flange and lip will not occur because of the buckling of
the web.
2. Mcrll by the semi-empirical interaction method of Commentary 1.1.2.1.1.2

Mcrl = Ag fcrl in this method fcrl is determined for each separate pair of elements and the
minimum is used, determination of element-pair buckling stresses follow.
2(a) Flange / Lip Local Buckling
The expression given for flange/lip local buckling is a function of the stress gradient on the lip. This
stress gradient is defined in terms of ξ=(f1 -f2 )/f1 where f1 and f2 are the stresses at the edges of
the lip (compression positive)
ξ = 0.105 from the calculations of 1(c)

2
d d
k flange_lip ( 8.55 . ξ 11.07 ) . ( 1.59 ξ 3.95 ) . 4 k flange_lip = 4.345
b b
d
note, d/b<0.6 and ξ<1 = 0.16 OK ξ = 0.105 OK
b
π .E
2 2
f cr_flange_lip k flange_lip. . t f cr_flange_lip = 48.248 ksi
12 . 1
2 b
ν
2(b) Flange / Web Local Buckling
f1 f2
f1 Fy f2 Fy stress gradient across h ξ ξ=2
f1
0.08 . ( ξ 7.7 . ξ
2 2
k1 3) 4.72 k 1 = 4.64 km 4 k m = 34.8
0.5 . ξ 4.ξ
3 2
k wss 4 k wss = 24
k m . k wss b
2
k wss 2
b k1
kw . km if < k w = 28.759
k1 h h k wss
h
2
b k1
k 1. if
b h k wss
π .E
2 2
f cr_web_flange k w. . t f cr_web_flange = 34.538 ksi
12 . 1
2 h
ν
note, that if we wanted kflange_web, so that the k is referenced to b instead of h we simply multiply by
(b/h)2 . b
2
k flange_web k w. k flange_web = 3.11
h
π .E
2 2
f cr_flange_web k flange_web. . t f cr_flange_web = 34.538 ksi
12 . 1
2 b
ν
f crl2 min f cr_flange_web f cr_flange_lip f crl2 = 34.538 ksi
M crl2 S gt . f crl2 M crl2 = 89.895 kip . in

Local buckling comparison:

Element method: M crl = 75.02 kip . in
Interaction method: M crl2 = 89.895 kip . in
Finite strip analysis: M crl3 91.06 . kip . in
3. Mcrd by the formulas of commentary 1.1.2.1.2.1
The hand methods for distortional buckling prediction require that section properties of the
isolated flange be calculated. The expressions here are only applicable for simple lips. More
complicated flanges would follow the same procedures, but new expressions would be required.
hx xo
S Material Properties:
W/F
hy yo
C x E
G
2.( 1 ν)
Properties of the Flange Only:

d) .t
2
Af (b A f = 0.205 in
1. . 3 1. . 3 4 4
Jf bt dt J f = 2.455 10 in
3 3
t . t .b 4.b.d 4 . b . d . cos( θ ) t .b.d d . cos( θ )

2 2 3 3 2 2 4 4 2
d 3 4
I xf I xf = 1.915 10 in
12 . ( b d )
t. b 4.d.b 6 . d . b . cos( θ ) 4 . d . b . cos( θ ) d . cos( θ )

4 3 2 2 3 2 4 2
4
I yf I yf = 0.18 in
12 . ( b d )
t . b . d . sin( θ ) . ( b d . cos( θ ) )
2
I xyf 3 4
4.( b d) I xyf = 8.399 10 in
t .b b.t t .d
3 3 3
I of 4
3 12 3 I of = 0.51 in
d . cos( θ )
2 2
b x distance from the centroid to
x of x of = 1.267 in
2.( b d) the shear center.
d . sin( θ )
2
y distance from the centroid to
y of
2.( b d) the shear center. y of = 0.032 in
2.d.b d . cos( θ )
2 2
b x distance from the centroid
h xf to the web/flange juncture. h xf = 1.673 in
2 (b d)
d . sin( θ )
2
y distance from the centroid
h yf h yf = 0.032 in
2.( b d) to the web/flange juncture.
0 . in
6 6
C wf C wf = 0 in
3. (continued) Mcrd by the formulas of commentary 1.1.2.1.2.1
Mcrd = Sg fcrd where determination of fcrd follows:
Determine the critical half-wavelength at which distortional buckling occurs:
1
2 4
4.π .h. 1 π .h
4 2 I xyf 4 4
ν . I . x 2 . x 2
L cr xf of h xf C wf of h xf L cr = 20.985 in
3 I yf 720
t
If bracing is provided that restricts the distortional mode at some length less than
Lcr, then this length should be used in place of Lcr.
Determine the elastic and "geometric" rotational spring stiffness of the flange:
4 2 2
π I xyf π
k φfe . E .I . x h xf
2
E .C E. . x h xf
2 .G .J
xf of wf of f
L cr I yf L cr
k φfe = 0.257 kip
2 2
π I xyf I xyf
k φfg . A . x h xf
2.
2 . y of . x of h xf . h xf
2
y of
2
I xf I yf
f of
L cr I yf I yf
2
k φfg = 0.017 in
ξ web ξ
Determine the elastic and "geometric" rotational spring stiffness from the web:
from 2(b)
E .t
3 2 4 3
. 3 π . 19 . h π . h
k φwe k φwe = 0.234 kip
12 . 1
2 h L cr 60 L cr 240
ν
2
L cr h
2
45360 . 1 62160 . . 53 3. 1 .π4
2
ξ web 448 π ξ web
π .h.t .
2 h L cr
k φwg
13440 2 4
L cr L cr
28 . π . 420 .
4 2
π
h h
Determine the distortional buckling stress:
3 2
k φfe k φwe k φwg = 2.683 10 in
f crd
k φfg k φwg
f crd = 24.913 ksi
M crd S gt . f crd M crd = 64.843 kip . in
Distortionanl buckling comparison:

Formulas above: M crd = 64.843 kip . in
Finite strip analysis: M crd2 67.955 . kip . in
CUFSM SOFTWARE

OF
Contents:
online home: www.ce.jhu.edu/bschafer/cufsm

SOFTWARE
INSTALLATION
FREQUENTLY ASKED QUESTIONS
CUFSM? What is CUFSM?
What are the system requirements?
Why would I use CUFSM?
Why do 2 versions of CUFSM exist?
Why would I use the Matlab version?
CUFSM is free?
Finite Strip? What is Finite Strip?
What is a buckling curve?
What is a buckling mode?
What is the half-wavelength?
What is the load factor?
What are Mcr and Pcr?
How can I use finite strip results in design?
OVERVIEW
please go to
www.ce.jhu.edu/bschafer/cufsm
the online home of CUFSM to download the software.
CUFSM2.5
What is the critical elastic buckling stress of a thin-walled beam or column? Find out by using our
free finite strip software for the elastic buckling calculation of thin-walled structures. Recently
updated, CUFSM v2.5 is now available for download (12.07.2001). The latest version includes
• streamlined user interface
• templates for the input of common shapes (C's and Z's)
• simplified interface for applying loads
• external springs and constraint equations
• compare results of multiple (up to 9) runs
• run all the graphics without Matlab in a standalone version
Extensive tutorials and instructions are available online and in the Design Manual for Direct
Strength Design of Cold-Formed Steel. The finite strip method implemented in CUFSM is at the
core of improving and modernizing the design of thin-walled structures. Understanding the uses of
CUFSM and calculating buckling loads and moments for arbitrary thin-walled members is the first
step towards numerical design of thin-walled members.
Disclaimer: Every possible effort has been made to develop CUFSM as sound engineering software. However the
developers expressly disclaim all liability for damages of any sort whether direct, indirect or consequential, arising out
of the use, reference to or reliance on these programs or any of their components.
CUFSM is freeware, please provide a reference to the author (Ben Schafer) and note the version you are using if you
use this code in any form, otherwise feel free to incorporate or distribute this code how you see fit.
Note, the materials developed to support CUFSM and the program CUFSM were both developed, in part, through the
sponsorship of research projects with the American Iron and Steel Institute (www.steel.org)
CUFSM2.5 Installation
• Standalone version
• Matlab version
problems? email: schafer@jhu.edu

CUFSM2.5
Standalone CUFSM Installation (1 of 2)

1. Download
cufsm2p5_setup.exe
2. Double-click the setup program and follow the
on-screen instructions

CUFSM2.5
Matlab CUFSM installation

1. Download cufsm2p5.zip
2. Unzip the files into the directory of your choice
(e.g., c:\My Documents\cufsm2)
3. Start Matlab
4. At the Matlab prompt type
a. >>cd ‘c:\My Documents\cufsm2’
b. >>cufsm2
5. CUFSM2 will run when you type b. above

CUFSM2.5
Introductory Questions
• CUFSM?
– What is CUFSM?
– What are the system requirements?
– Why would I use CUFSM?
– Why do 2 versions of CUFSM exist?
– Why would I use the Matlab version?
– CUFSM is free?
• Finite Strip?
– What is Finite Strip?
– What is a buckling curve?
– What is a buckling mode?
– What is the half-wavelength?
– What is the load factor?
– What are Mcr and Pcr?
– How can I use finite strip results in design?
What is CUFSM?
• Software for exploring elastic buckling behavior.
• CUFSM calculates the buckling stress and
buckling mode of arbitrarily shaped, simply
supported, thin-walled members.
• CUFSM was originally written to support
research on the behavior and design of cold-
formed steel members with a variety of different
types of longitudinal stiffeners.
• CUFSM is freely available and distributed.
What are the system requirements?
• For the standalone version a PC with any flavor* of
Windows is required
• For the Matlab version, any machine with Matlab v6 will
work - this means most flavors of Unix and all PCs are
supported - but not Macs* at this time.
• The faster your machine the faster your analysis will
run, but finite strip is an efficient solution to a
complicated problem, and typical analyses, on even
older 486 or Pentium machines, take less than a minute
to complete.
* (version 2.5 has not been tested on Windows XP at this time) .
**(Rumors are that Matlab will be ported to OS X, then CUFSM will run on a Mac.)
Why would I use CUFSM?
• To explore and better understand elastic buckling
behavior of thin-walled members.
• To accurately determine the elastic buckling stress of a
thin-walled section of arbitrary cross-section.
• Design and hand methods that are traditionally used for
“plate” structures often ignore compatibility at plate
junctures and typically provide no means to calculate a
variety of important buckling modes (e.g., distortional
buckling). CUFSM allows all elastic buckling modes of a
structure to be quantified and examined.
• To determine inputs such as Pcr and Mcr for the Direct
Strength Method of design.
Why do 2 versions of CUFSM exist?
• The standalone version is for those users who do not
have Matlab and only require access to the available
features in the graphical version of CUFSM.
• The Matlab version of CUFSM is for all users who have
Matlab available. CUFSM was originally coded in
Matlab and thus using CUFSM in Matlab provides
greater flexibility.
• Matlab exists on platforms other than Windows PCs
(including Linux) and thus the Matlab version of CUFSM
runs on many more platforms than the PC standalone
version.
Why would I use the Matlab version?
• The Matlab version allows you to directly access and
modify the source code itself.
• Parameter studies may be easily completed using the
CUFSM subroutines from within your own Matlab
program (see Advanced Functions - Matlab).
• Matlab is available on many more platforms and allows
you to use CUFSM on high-powered workstations and
other machines (e.g., Linux).
• Far more flexibility is available with the Matlab version
since any subroutine, from the graphics, to the actual
computations may be separately accessed.
CUFSM is free?
• Yes.
• CUFSM is distributed as FREEWARE.
• If you use CUFSM in design work or in scientific
studies I ask that you please provide a reference
to my work and cite the version of CUFSM you
are using for the work.
• Standard disclaimers apply: Although all attempts have
been made to insure CUFSM is reliable, the responsibility for use
of the program rests solely on the user.
What is Finite Strip?
• Finite strip is a specialized version of the finite element method.
finite element finite strip
• In the finite strip method, element shape functions use polynomials in the
transverse direction, but trigonometric functions in the longitudinal direction.
Judicious choice of the longitudinal shape function allows a single element, a
“strip” to be used.
• Classical finite strip, as implemented in CUFSM, uses a single half sine wave
(sin(πx/a)) for the longitudinal direction. See Appendix (1) Theory for more
details on the finite strip method.
What is a buckling curve?
• The tutorials and later examples cover this in detail, but for now,
the buckling curve is the primary result from a finite strip
analysis. A typical buckling curve is shown below. The minima of
this curve are of special interest as the indicate the critical half-
wavelength and load factor for a given buckling mode.
What is a buckling mode?
• The buckling mode is the shape
that a member buckles into, for
2D representation of the local
example shown here is the local buckling mode shape (only the
buckling mode of a Zee in bending. undeformed and the maximum
deformed cross-sections are shown)
More precisely, a buckling mode for a Zee in pure bending.
represents a secondary deformed
shape that has the same potential
energy as the primary deformation,
simple bending in this case.
3D representation of the local

buckling mode shape. Local buckling
occurs at short half-wavelengths
(repeats in short intervals) as shown
here.
applied load on a Zee
What is the half-wavelength?
• The finite strip method assumes longitudinal
deformation occurs in 1/2 a sine wave (a half-
wavelength)
• The half-wavelength is the length of the 1/2 sine wave
assumed in the analysis
• Analysis is performed for systematically increasing half-
wavelengths to determine the buckling behavior (mode
shape and load factor) of a member
• Note: half-wavelength is not strictly equal to the
unbraced length, as any identified mode may repeat
itself multiple times within a given unbraced length
What is the load factor?
• To perform finite strip analysis the member is loaded with a reference stress
distribution
• the load factor times the reference stress distribution is equal to the buckling
stress
• For example, consider a member with a stress distribution of 1.0 ksi at every
location (node) on the member. Assume that after analysis a local buckling
load factor of 15.4 is identified. What is the local buckling stress? The local
buckling stress is 1.0 ksi x 15.4 = 15.4 ksi
• For another example, consider a member that is loaded with a reference
stress distribution that is equal to the moment that causes first yield in a
member, My. Assume after analysis a distortional buckling load factor of 0.5 is
identified. This implies that distortional buckling occurs at 0.5My. If first yield is
in compression, and the yield stress is denoted, fy, then the distortional
buckling stress is 0.5fy.
• More precisely: the load factor is the eigenvalue of the relevant eigenvalue
buckling problem, and the buckling mode is the eigenvector.
How can I use CUFSM results in design?
• Use the Direct Strength Method
• The Direct Strength Method requires that you
know the elastic buckling load (Pcr) or elastic
buckling moment (Mcr) for your member. CUFSM
provides a means to calculate these values for
any arbitrary cross-section.
• Most design codes for thin-walled structures rely
on simplified plate buckling coefficients, or “k”
values - CUFSM may be used to determine far
more accurate “k” values than used in design.
CUFSM2.5
What are Mcr and Pcr?

• They are the elastic buckling moment and the elastic
buckling load.
• They are inputs in the Direct Strength Method.
• For typical open thin-walled shapes, such as cold-
formed steel Cees, Zees or hats, three critical
loads/moments exist
– Pcrl/Mcrl: Elastic critical local buckling load/moment
– Pcrd/Mcrd: Elastic critical distortional buckling load/moment
– Pcre/Mcre: Elastic critical Euler buckling load/moment
• Multiple modes (e.g. flexural, torsional, and flexural-
torsional may exist for Pcre).
CUFSM2.5
CUFSM Overview
• Main
• Input
• Properties
• Post
• Compare
1 2 3 4
Load and Save files as 1. Enter the Geometry Print, sends the current
desired. To compare 2. Apply the desired load/stress distribution screen to the default
more than one analysis, 3. Analyze the member printer.
save several different 4. Post-process and recover modes and critical loads
Copy sends the current
files after performing
screen to the clipboard
analysis and load them
in bitmap format.
into the Comparison
post-processor (start this Reset starts the program
by pressing Compare). over and clears all
entered data.
Exit, leave CUFSM
Define as many materials as you like here.
This is the input page.

•An example is built-in to CUFSM (see
Directly enter, or cut Tutorial 1) and that example always comes up
and paste in,the when CUFSM is started.
geometry of your
member here. •Tutorial 1 demonstrates the basic functionality
of this page and many of the different plotting
options.
•Tutorial 2 shows how to enter a member from
scratch and how the Double Elem. button may
be used to improve your model.
Define the elements here. •Tutorial 3 shows how to use the C/Z template
Each element has an to input a member.
individual thickness and
material that you select.
Define the half-wavelengths that your member will be analyzed at here.
You can model any external springs that are attached You can model any equation constraints for your
to your member here. mode here.
Simple member properties are calculated and
given above. These are used below to
determine stress distributions on the member.
By defining a Tutorials 2 and 3

maximum, or yield show how to use this
stress, loads (P) and section effectively for
moments (M) may be simple Cee and Zee
determined. Any of members.
these P or M can be
used to generate a
reference stress
distribution on the
member you create in
the Input page.
All plotting of the buckling
mode shapes is controlled to Tutorials 1,2 and 3
the left. The key buttons are show how to
the arrows that control the effectively interpret
half-wavelength. 2D and 3D and manipulate the
plots of the mode shapes are post-processing page.
available as well as a plot of
the stress distribution. Values
associated with the currently
shown plot are given above
the plot.
Full control over the buckling curve given below is

available using the controls to the left. The numerical
results for the first mode may be written to a text
output file, if desired. The “red circle” on the plot
shows where you are at on the plot, at all times.
CUFSM2.5
The user has complete

control over which
buckling mode to view -
select the type of plot, the
half-wavelength, the mode,
and which file you want to
view!
Multiple files are

loaded for post-
processing at the same
time, see list to the left.
You decide which

loaded files should
show in the buckling
curves to the right.
TUTORIALS

OF
Contents:
TUTORIAL 1 - LEARN THE BASICS AND INTERPRET RESULTS

TUTORIAL 2 - BUILD A MODEL FROM SCRATCH AND FIND PCR
TUTORIAL 3 - BUILD A MODEL USING THE TEMPLATE AND FIND MCR
CUFSM2.5
Tutorial 1
• Default Cee section in bending
• Objective
To introduce CUFSM and the finite strip method and gain a
rudimentary understanding of how to perform an analysis and
interpret the results.
• A the end of the tutorial you should be able to
– enter simple geometry
– enter loads (stresses)
– perform a finite strip analysis
– manipulate the post-processor
– identify and understand different buckling modes
SELECT
Cross-section geometry is
entered by filling out the nodes
3. SELECT
and the elements, e.g., node 6 is
at coordinate 0.0,6.0 **separate
your entries by single spaces**
1. UNCHECK
We will discuss more about all
2. CHECK those 1’s after the nodal
coordinates and the last column
in the Nodes section later on.
You can always press the ‘?’
buttons if you want to learn
more now.
Let’s take a look at the
elements. (follow the arrows)
Elements define how the
3. SELECT geometry is connected, how
thick the member is, and what
material a particular element is
composed of.
1. UNCHECK For example element 5,
connects nodes 5 and 6 together,
has a thickness of 0.04 in. and
2. CHECK uses material “100” - material
100 is defined above in the
Material Properties Section.
Let’s take a look at the loading.
(follow the arrows)
Each node has a “stress”
3. SELECT assigned to it. Our analysis will
give a “buckling load factor”
that is a multiplier times the
inputted stresses.
In this case the stresses amount
to a pure bending case with
fy=50 ksi.
1. UNCHECK Let’s take a look at the stress
2. CHECK distribution. (follow the arrows)
1 2
select 1, use the

robust solver,
analysis will
proceed, then The stress distribution (the
select 2 loading) is clearly shown to be
pure bending.
Note the “Lengths” below.
These are the half-wavelengths
that we will analyze. Each half-
wavelength has a different
buckling load factor.
Let’s Analyze and then go to the
Post processor to view the
results.
2 select 1, 4 times, This screen shows what “Post”
until the red circle looks like after you analyze.
below moves to 5.0, The buckling mode for the first
then select 2, to data point is shown above.
1 view the buckling Select different half-
mode wavelengths using the arrow
buttons above and plot the
different mode shapes. The
minima of the buckling curve
below identify important
locations to examine.
“red circle”= where you are at

2 select 1, until the red The local buckling mode is
circle moves to 60.0, shown to the left. The mode
then select 2, to repeats at a half-wavelength of
view the distortional 5.0 in. (See summary above plot
1 buckling mode and numbers below in the
buckling curve). The buckling
load factor is 0.17 for local
note, the scale of the
buckling. This means elastic
buckling mode is
critical local buckling occurs at
arbitrary! 1 or -1 are
0.17 times the stress distribution
equally valid, as is 0.5 or
entered - remember the stress
4, or any other convenient
magnitudes from before?
multiplier.
3
The distortional buckling mode
follow 1,2,3 to take a
is shown to the left. The mode
look at the buckling
1 2 repeats at a half-wavelength of
modes in 3D
60.0 in. The buckling load
factor is 0.31. This means
elastic critical distortional
buckling occurs at 0.31 times
the stress distribution entered.
2 this is distortional
buckling, select 1
and go back to 5.0,
then select 2, let’s
look at local
1 buckling first
The lengths that are The 3D plot to the left shows
analyzed in the plot local buckling at a half-
below are selected wavelength of 5.0 in. Note, the
by the user. Let’s 2D plot presented earlier shows
add some points in the maximum cross-section
the circled sections deflected shape only.
below to smooth
out our plot. - CUFSM finite strip analysis
Select input - assumes a single half sine wave
in the longitudinal direction (as
shown). Return to a half-
wavelength of 60 in. to see the
distortional buckling mode.
2 3
select 2, use the

robust solver,
analysis will
proceed, then
select 3
1. Add the additional “lengths below”. All the

half-wavelengths that are entered below are
analyzed. If you are only concerned about a
particular range of half-wavelength (e.g., local
buckling) then you may remove some lengths.
The analysis results for all of Q: What would happen if I
the lengths are shown below. assumed the member always
buckled in two half sine waves
Q: Why is the load factor at a
(i.e a full sine wave) instead of
half-wavelength of 10 in.
one?
greater than at 5.0 in?
A: You would see the same
A: Because the analysis always
buckling curve, but it would be
assumes buckling is in a single
translated to the right. For
half sine wave, this may not be
example - local buckling has a
how the actual member buckles.
minimum at 5.0 in. for one half
sine wave, and will have a
minimum at (2)(5.0) = 10.0 in.
for two half sine waves.
See next slide for result

(1) (2)
This is the same buckling curve This is a specially constructed
we have been looking at for the curve, in which two half sine
Cee in bending. Note the familiar waves have been assumed
minimums: local at 5.0 in., and throughout the analysis of the Cee
distortional at 60.0 in. in bending. Note now that local
occurs at (2)(5.0) = 10.0 in., and
This curve, like all other finite
distortional at (2)(60.0) = 120.0 in.
strip analysis generated by
CUFSM assumes a single half In a real structure the buckling
sine wave in the longitudinal mode is free to form any number
direction. of half sine waves - therefore only
the minimums of the first curve
are of primary interest.
Q: What would 3 half sine waves look like? 4?

CUFSM2.5
Tutorial 1: Conclusion
• Default Cee section in bending
• Objective
To introduce CUFSM and the finite strip method and gain a
rudimentary understanding of how to perform an analysis and
interpret results.
– enter geometry
– enter loads (stresses)
– perform a finite strip analysis
– manipulate the post-processor
– identify and understand different buckling modes
CUFSM2.5
Tutorial 2
• SSMA Cee in Compression: 600S200-33 Fy = 50ksi
• Objective
To model a typical Cee stud in compression and determine
the elastic critical local buckling load (Pcrl)and elastic critical
distortional buckling load (Pcrd).
– enter material, nodes, elements, and lengths from scratch
– apply a reference load P, or M as desired
– interpret a simple buckling curve
– identify local and distortional buckling in a simple member
– determine Pcrl and Pcrd
2. SELECT 1. SELECT
This screen shows the default
section that appears when you
enter the Input screen for the
first time. In our case we do not
want to use this section so we
need to start from scratch in
order to enter our 600S200-33
member.
Highlight each section: Material

Properties, Nodes, Elements,
Lengths and delete the current
values.
Here we will enter in the material properties, in this case
they will be for steel: E=29500 ksi, ν=0.3
Here we need to enter in the node numbers and the

coordinates that define the geometry. We just need to
enter in the corner nodes (we will ignore the corner radii
in this example).
Here we enter in the elements. We need to give the

connectivity of each element (what nodes are used to
make the element) the thickness of each element and a
number that refers back to the material being used.
Finally we will need to enter in the half-wavelengths that

we wish to do the analysis at.
Now enter in the material properties as shown to the left.
Let’s define material #1.
CUFSM allows you to define orthotropic materials, but
in our case we are just using a simple isotropic material.
Therefore Ex = Ey and νx = νy.
For isotropic steel:
E=29500 ksi, ν=0.3, G=E/(2(1+ν))=11346 ksi
If our cross-section has multiple material types we could
define a new material number and add a row to the
material properties definition. That is not necessary in
this case.
Let’s start with the geometry next.
Remember a 600S200-033 Cee section has:
6 in. web
2 in. flange
0.62 in. lips
0.0346 in. thickness
Now enter in the nodes and elements to define the
bottom flange as shown to the left. Select Update
Plot to see the results.
The nodes include a node number, followed by the x
and z coordinate followed by 4, “1’s” followed by
1.0. The 4 “1’s” indicate that there is no external
longitudinal restraint at those nodes - for normal
**separate your entries by spaces** member analysis this is always the case. The final
We are using simple outside 1.0 is the stress input on that node, we use 1.0, but
dimensions, o.k. for this example. Lip any value would do, because we are going to change lip
= 0.62 in., flange = 2.00 in. this input later.
bottom flange
The element definition requires you to enter the

element number, then its connectivity, then the
element thickness, and finally the mat#, where 1 refers
to the material we defined above.
let’s finish the nodes and the elements…

Select Twice
Note, use of the
double elem. Enter the last of the nodes and
button is not elements and select Update Plot. The
reversible, (“no model is nearly complete, but we
undo!”). You need to consider a technical issue:
may want to how many elements do I need to get a
save the model good solution?
before doubling
the elements. Four elements in any “flat” in
compression will provide a nicely
converged answer. Even two
elements does well, but 1 is too few.
Press Double Elem. two times to
increase the discretization of your
member.
After entering the lengths,
select Properties to define
the loading.
Now we need to define the lengths.
“Evenly” spacing the lengths in
logspace as done below is a
reasonable first estimate.
For local buckling the half-
wavelength of interest is close to the
maximum dimension of the member
(6 in. in this case). Distortional
buckling is usually 2 to 8 times that
length, and interest in the longer
lengths depends on the application.
enter in
let’s complete the loading.
lengths as
shown
Maximize the screen, if you can’t see the cursor.

relevant axes, origin,
etc. are all shown on
the cross-section.
Basic properties of the cross-section are

shown above. The area, centroid, moments
of inertia etc. should be what you expect,
otherwise you may have made a mistake
entering in the data..
Finite strip analysis requires that you enter in a reference

longitudinal stress. The buckling load factor output is a multiplier
times this reference stress. The tools to the left make entering in
the reference stress easier.
For example,
enter in 50 for fy
Select calculate P and M
Uncheck P
Select Generate Stress Using checked P and M
Go back to the input page
to see the result of
generating stress using the
“M” you checked.
The loads are generated based on the fy you select. So, the
generated P is the squash or yield load (Py) for this section. The
M is the moment that causes first yield (My) etc. Based on the
loads you check off, a stress distribution is generated.
Note, for this

symmetric section
the maximum and
minimum stresses
are equal to the
inputted fy.
Return to properties to
remove this bending stress
and define a compressive
stress on the stud instead. Here we can see that the
generated stress has placed a
pure bending stress gradient on
our member, note the entries in
“Nodes” to the right and the
values shown on the plot.
1 2
select 1, use the

robust solver,
analysis will
proceed, then
select 2
Now our reference load is

Enter the yield Py, the squash load. So, if
stress, calculate the buckling load factor is
the P and M 0.5 then elastic critical
values, and buckling is at 0.5Py. You
generate a pure can load with any reference
compression stresses that are convenient
stress. for your application.
Maximum stress = 1.0, or fy
are often convenient
choices.
change the half-
wavelength to 5 The local buckling mode is
and hit Plot Mode shown to the right. Note,
that there is no translation
at the folds, only rotation.
The load factor is 0.10, so
elastic critical local
buckling (Pcrl) occurs at
0.10Py in this member.
We do not have enough

lengths! Go back to input,
add more lengths between 10
and 30 and re-analyze (see
Tutorial 1 for adding lengths)
change the half- Distortional buckling is
wavelength to 26 identified at a half-
and hit Plot Mode wavelength of 26 in. The
elastic critical distortional
buckling load Pcrd=0.32Py
What exists at longer

half-wavelengths, for
example, 300 in.? Change
the half-wavelength and
select Plot Mode
What if?
At 300 in. the lowest
What happens if the
buckling mode, is
member is thicker?
weak-axis flexural
Save these results as
buckling of the
600S200-033, change
column, as shown to
to a 600S200-097 with
the left.
a t=0.1017 using the
Input page, reanalyze
and save the results as
600S200-097. Then
use the compare
button to look at the
two analyses.
The comparison post-processor
allows you to examine up to 8
different runs at the same time.
Useful when comparing
different loading, geometry, or all key info.
other changes. summarized here, in
this case we are looking
at local buckling of
600S200-033
note, the thickness difference in

the elements when you change
between File 1 an File 2.
Remember the reference
loads were equal to Py,
but the Py of the two
members are not the same
Distortional because the area is not the
Local same…
CUFSM2.5
Tutorial 2: Conclusion
• SSMA Cee in Compression: 600S200-33 Fy = 50ksi
• Objective
To model a typical Cee stud in compression and determine
the elastic critical local buckling load (Pcrl)and elastic critical
distortional buckling load (Pcrd).
– determine Pcrl and Pcrd
CUFSM2.5
Tutorial 3
• LGSI Zee in Bending: Z 12 x 2.5 14g, Fy = 50ksi
• Objective
To model a typical Zee purlin or girt in bending and determine
the elastic critical local buckling moment (Mcrl)and elastic
critical distortional buckling moment (Mcrd).
– OR use the C and Z template to enter a geometry
– determine Mcrl and Mcrd
2. SELECT 1. SELECT
Select
Note, we could
enter the
geometry node
by node as in This screen shows the default
Tutorial #2, but section that appears when you
in this case, let’s enter the Input screen for the
use the template first time. In our case we do not
instead. want to use this section so we
need to start from scratch in
order to enter our LGSI Z
12x2.5 14g purlin.
Select C/Z template

This is the default template that comes up when
you select the template button. Note that all
dimensions are centerline dimensions - e.g., h is
the flat distance, not the out-to-out distance as is
typically listed in product catalogs, etc.
Enter in all the appropriate dimensions and select
Submit to input.
The centerline dimensions for an LGSI Z 12 x 2.5

14g member are shown to the right. Enter in these
dimensions and then press Update Plot. When
complete select Submit to Input.
Note, the material is assumed to be steel, but two
units systems are supported. Geometry other than
the typical Cee or Zee can be entered.
The template automatically selects an adequate
number of elements.
The template automatically selects lengths to be
analyzed as well.
This is the model
generated by the
template. It can be
still be modified as
desired. The default
loading is 1.0 on
every node, let’s go
to the properties page
and apply a pure
bending stress
distribution.
Note that the principal coordinate system is not in line
with the global x,z coordinate system, as expected.
Enter a yield stress,

calculate P and M, Switch to restrained
uncheck P and bending and re-
examine the generated calculate the stress
stress distribution. As distribution.
shown to the right, it
reflects unsymmetric
bending.
1 2
select 1, use the

robust solver,
analysis will
proceed, then
select 2
This is the yield moment, My, the

buckling load factor results will be in
terms of My=192 kip-in.
The stress distribution to

the right would be
applicable for a laterally
braced beam, and is
typically assumed in
cold-formed steel design
codes. Note that the
flanges are different
sizes and in this case the
wider flange has been
placed in compression.
This screen shows the post-
processing page that will
come up when you select
Post. Note, the two minima
in the plot: local and
distortional buckling.
Clean up the curve and

change the half-
wavelength to show local
buckling.
Local buckling results are
shown here. Mcrl=0.66My
and the buckling mode
shape is as given to the
right.
Change the half-

wavelength to examine
distortional buckling.
Distortional buckling
results are shown here.
Mcrd=0.70My and the
buckling mode shape is as
given to the right.
CUFSM2.5
Tutorial 3
• LGSI Zee in Bending: Z 12 x 2.5 14g Fy = 50ksi
• Objective
To model a typical Zee purlin or girt in bending and determine
the elastic critical local buckling moment (Mcrl)and elastic
critical distortional buckling moment (Mcrd).
– OR use the C and Z template to enter a geometry
– determine Mcrl and Mcrd
PRESENTATIONS

OF
Contents:
ADVANCED FUNCTIONS - BOUNDARY CONDITIONS, SPRINGS, CONSTRAINTS..

ADVANCED IDEAS - DEFINING BUCKLING MODES, HIGHER MODES...
ADVANCED MATLAB - PARAMETER STUDIES
THEORY - BACKGROUND AND STIFFNESS MATRICES
CUFSM2.5
CUFSM Advanced Functions

• Boundary conditions
• Constraints
• Springs
• Multiple materials
• Orthotropic Material
Boundary conditions
• Longitudinal boundary conditions (fixity) can be set in
the finite strip model
• Modeling classic problems requires using this feature
– simply supported plate
– fixed plate
• Special cases may exist where artificial boundary
conditions are added in an analysis to examine a
particular buckling mode in exclusion of other modes
(see Advanced Ideas for more on this)
• Symmetry and anti-symmetry conditions may be
modeled by modifying the boundary conditions
Boundary conditions continued
• How to
• Simply supported plate example
• Fixed-free plate example
• Flange only model
• Symmetry model on a hat in bending example
How to (boundary conditions)
These columns of ones set the boundary conditions for
the model. A 1 implies that the degree of freedom is free
along its longitudinal edge. All models are simply
supported at the ends due to the choice of shape function
in the finite strip method.
For models of members these always remain 1, however
if longitudinal restraint should be modeled then the
appropriate degree of freedom (direction) should be
changed from a 1 to a 0.
z
q
x
y
Simply supported plate (boundary
Simply supported plate in pure compression
conditions)
Plate is 10 in. wide and t = 0.10 in., material is steel.
The x and z degree of freedom at node 1 have been
supported by changing the appropriate 1’s to 0’s.
The z degree of freedom at node 5 has been supported by
changing the appropriate 1 to 0.
Green boxes appear at 1 and 5 to indicate some boundary
conditions have been changed at this node.
Simply supported plate (boundary
conditions)
Theory → f cr
( ) =k
( )
π2E ætö
ç ÷ =4
12 1 − ν è ø
2
b
2
π 2 29500 æ 0.1 ö
12 1 − 0.3 è
2
ç
10 . 0 ø
2
÷ = 10.66ksi
Input reference stress is 1.0 ksi. So in this case the load factor is equal to the
buckling stress in ksi, i.e., 10.67 ksi. versus 10.66 ksi by hand.
Fixed-free plate (boundary
Fixed-free plate in pure compression
conditions)
The x, z and q (θ) degree of freedom at node 1 have been
supported by changing the appropriate 1’s to 0’s.
Green boxes appear at 1 to indicate some boundary
conditions have been changed at this node.
Fixed-free plate (boundary
conditions)
2 2
π2E ætö π 2 29500 æ 0.1 ö
Theory → f cr =k ç ÷ = 1.277 ÷ = 3.40ksi
( ) ( )
12 1 − ν 2 è b ø
ç
12 1 − 0.32 è 10.0 ø
Input reference stress is 1.0 ksi. So in this case the load factor is equal to the
buckling stress in ksi, i.e., 3.42 ksi. versus 3.40 ksi by hand.
Flange only model (boundary
Isolated flange in pure compression
conditions)
Lip is 2 in. long and the same material and thickness
The x, z and q (θ) degree of freedom at node 1 have been
supported by changing the appropriate 1’s to 0’s. So, the
left end is “built-in” or “fixed”.
Flange only model (boundary
conditions)
fixed
Adding the lip stiffener increases the buckling stress significantly.

Adding the lip stiffeners introduces the possibility of two modes,
one local, one distortional.
Local
Distortional
Symmetry model on a hat in bending
Hat in bending - full model
(boundary conditions)
The hat is 2 x 4 x 10 in.
Pure bending is applied as the reference load.
The reference compressive stress for the top flange is 1.0
ksi which results in -1.75 tension for the bottom flange
Hat in bending - half model

The hat is 2 x 4 x 10 in.
Pure bending is applied as the
reference load.
The reference compressive stress for
the top flange is 1.0 ksi which results
in -1.75 tension for the bottom flange.
Symmetry conditions are enforced at
mid-width of the top flange, note the
degrees of freedom changed to 0 at
node 11 in the Nodes list to the left.
full model local buckling stress in compression = 15.11 ksi

half model using symmetry local
buckling stress in compression =
15.11 ksi
Constraints
• You may write an equation constraint: this enforces the
deflection (rotation) of one node to be a function of the
deflection (rotation) of a second node.
• Modeling external attachments may be aided by using
this feature
– an external bar that forces two nodes to have the same
translation but leaves them otherwise free
– a brace connecting two members (you can model multiple
members in CUFSM)
• Special cases may exist where artificial equation
constraints are added in an analysis to examine a
particular buckling mode in exclusion of other modes
(see Advanced Ideas for more on this)
Constraints continued
• How to
• Connected lips in a member
• Multiple connected members
How to (constraints)
Equation Constraints are determined by defining the
degree of freedom of 1 node in terms of another.
For example, the expression below in Constraints says
At node 1, set degree of freedom 2 equal to 1.0 times
node 10, degree of freedom 2:
w1=1.0w10
You can enter as many constraints as you like, but once
you use a degree of freedom on the left hand side of the
equation it is eliminated and can not be used again.
Symbols appear on the nodes that you have written
constraint equations on, as shown in this plot for nodes 1
and nodes 10.
Connected lips in a member
(constraints)
Constraints example 1
Use the default member
Change the loading to pure compression
Constrain the ends of the lips, nodes 1 and 10
to have the same vertical displacement
Compare against analysis which does not
have this constraint.
Connected lips in a member
(constraints)
Loading is pure
compression with a The two lips have the
reference stress of 1.0, same vertical
the two results show the displacement. Anti-
influence of the symmetric distortional
constraint on the buckling results.
solution.
distortional with
the constraints on
the lips
typical
local is distortional
the same buckling
Multiple connected members
(constraints)
Multiple Member Equation Constraint Example
Two members are placed toe-to-toe.
Geometry is the default Cee section in CUFSM. The
loading is pure compression.
In this example only the top lips are connected, say
for example because of an unusual access situation.
Equation constraints are written, as shown below to
force that x, z and q of nodes 10 and 20 are identical.
Multiple connected members
(constraints)
Local buckling is top lips are
not affected by the connected.
constraint, but This has an
distortional influence on
buckling and long distortional
wavelength buckling, as
buckling is… shown.
weak-axis flexural
buckling occurs in
the model with the
lips attached at the
top.
local and distortional

buckling for a single
member.
flexural-torsional buckling occurs in

the single isolated member
Springs
• External springs may be attached to any node.
• Modeling continuous restraint may use this feature
– Continuous sheeting attached to a bending member might be
considered as springs
– Sheathing or other materials attached to compression
members might be considered as springs
• Springs may be modeled as a constant value, or as
varying with the length of the model (i.e. a foundation)
Springs
• How to
• Sheeting attached to a purlin
• Spring verification problem
How to (springs)
Springs are determined by defining the node where a
spring occurs, what degree of freedom the spring acts
in, the stiffness of the spring, and whether or not the
spring is a constant value (e.g. force/length) or a
foundation spring (e.g. (force/length)/length).
Constant springs use kflag=0, foundations use kflag=1.
You can enter as many springs as you like.
The springs always go to “ground”. Therefore they
cannot be used to connect two members.
Springs appear in the picture of your model once you
define them.
Springs are modeled as providing a continuous
contribution along the length.
Sheeting attached to a purlin
(springs)
Purlin with a sheeting “spring” example
Use the LGSI Z 12 x 2.5 14g model from Tutorial 3
The applied bending stress is restrained bending about
the geometric axis with fy=50 ksi. (first yield is in
tension in this model as the flange widths are slightly
different sizes)
Assume a spring of k = 1.0 (kip/in.)/in. exists in the
vertical direction at mid-width of the compression
flange. (Ignore, in this case, rotational stiffness
contributions from the sheeting, etc.)
See Springs below for the definition of the vertical
spring.
Sheeting attached to a purlin
(springs)
The buckling curve below
shows the results of an
The buckling mode to the
left shows distortional
analysis without the springs buckling with the spring in
(1) and analysis with the place. Note, the “star”
spring (2). Note that the denotes the existence of the
spring has greatly increased spring in the model.
the distortional buckling
stress.
Example for demonstrative purposes only - actual sheeting may have much lower
stiffness, and other factors may be considered in the analysis.
Spring verification
400
Eule r k=0
CUFS M k=0
350 Eule r k=0.001 kip/in./in.
CUFS M k=0.001 kip/in./in.
300
buckling load P cr (kip)
E = 29500 ks i
250
Iwe a k = 106.736 in. 4
200 P cr = [π2 EI/(L2 )](1+kL4 /(π4 EI))
150
100
50
0
200 400 600 800 1000 1200 1400 1600
le ngth (in.)
Multiple materials
• Multiple materials may be used in a single
CUFSM model
• Explicitly modeling attachments that are of
different materials may use this feature
• Some unusual geometry changes may be
modeled by changing the material properties
explicit sheathing modeling
0.25 in. thick sheet E=1/10Esteel, see mat# 200
perfect connection at
mid-width between stud
and sheathing done by
constraints.
Toe-to-toe studs with 1-sided Sheathing
Use a pair of the default CUFSM Cee sections and
connect them to a 0.25 in. sheathing on one flange
only. The sheathing should have E=1/10Esteel
Note, the use of a second material and the
constraints that are added to model the connection.
Toe-to-toe studs with 1-sided Sheathing
Material numbers are shown using the material#
check-off in the plotting section.
The loading is pure compression on the studs, and
no stress on the sheathing.
Local buckling is
not affected by the
sheathing, but
distortional
buckling and long
wavelength
buckling is…
weak-axis flexural
buckling occurs in
the model with the
sheathing
local and distortional

buckling for a single
member.
flexural-torsional buckling occurs in

a single isolated member
Orthotropic Material
• Orthotropic materials may be used in CUFSM
• Plastics, composites, or highly worked metals
may benefit from using this feature
1/2 G, SS Plate
Orthotropic Material Example
Simply supported plate where Gxy is 1/2Gisotropic
Low G modulus are typical concerns with some
modern plastics and other materials. Also, some
sheathing materials may be modeled orthotropically.
CUFSM2.5
1/2 G, SS Plate
2 2
π2 E ætö π 2 29500 æ 0.1 ö
Isotropic Theory → f cr =k ç ÷ =4 ÷ = 10.66ksi
( )
12 1 − ν 2 è b ø ( )ç
12 1 − 0.32 è 10.0 ø
vs. 8.80 ksi when Gxy = 1/2Gisotropic
CUFSM2.5
Advanced Ideas and Examples

• Defining buckling modes
• Why define buckling modes?
• Understanding higher modes
• Utilizing higher modes
• Handling Indistinct modes
• Solution Accuracy
Defining Buckling Modes
• For the majority of open-section thin-walled
members the relevant buckling modes can be
broken into 3 groups:
– Local
– Distortional
– Long
• Defining these buckling modes relies on an
understanding of the role of the buckling
– mode (shape), and
– half-wavelength
Local Buckling
Half-Wavelength
Local
Local buckling minima occur at half- rotation only.
wavelengths that are less than the
largest characteristic dimension of the
member. In the example to the right,
this implies the minimum must be at 9 in.
half-wavelengths < 9 in.
Mode Shape
Local buckling involves ONLY rotation only.
rotation, NOT translation at the fold
lines of the member. Local buckling
involves distortion of the cross-
section.
Complications
Local may be indistinct from
Local minima Distortional minima in Long column
distortional buckling in some
in this region this region response in this
members.
region.
Local buckling may be at half-
wavelengths much less than the
characteristic dimension if
intermediate stiffeners are in place, or
if the element undergoes large tension
and small compressive stress.
Local buckling half-wavelength criteria
• Local buckling of a simply supported plate in pure
compression occurs in square waves, i.e., it has a half-
wavelength that is equal to the plate width.
• If any stress gradient exists on the plate, or any
beneficial restraint is provided to the edges of the plate,
the critical half-wavelength (mode 1 minimum) will be at
a half-wavelength less than the plate width.
• Therefore, local buckling, with the potential for stable
post-buckling response, is assumed to occur only when
the critical half-wavelength is less than the largest
potential “plate” in a member. If the half-wavelength is
longer - the mode is not local buckling.
Distortional
Half-Wavelength
Distortional translation
Distortional buckling occurs at a half-
wavelength intermediate to local and
Long mode buckling. The half-
wavelength is typically several times
larger than the largest characteristic 9 in.
dimension of the member.
Mode Shape
Distortional buckling involves BOTH
translation AND rotation at the fold translation
line of a member. Distortional
buckling involves distortion of a
portion of the cross-section and
predominately rigid response of
another portion.
Complications Local minima Distortional minima in Long column
Distortional buckling may be indistinct
region.
(without a minimum) even when local
buckling and long half-wavelength
buckling are clear.
The half-wavelength for distortional
buckling is highly dependent on the
loading and the geometry.
Long / “Euler”
Half-Wavelength
Long
The traditional “Euler” long column
buckling modes: flexural, torsional,
flexural-torsional occur as the
minimum mode at long half-
wavelengths.
Mode Shape
Long buckling modes involve
translation (flexure) and/or rotation
(torsion) of the entire cross-section.
No distortion exists in any of the
elements in the long buckling modes.
Complications
Flexure and distortional buckling may
interact at relatively long half-
Local minima Distortional minima in Long column
wavelengths making it difficult to
determine long column modes at
region.
certain intermediate to long lengths.
Finite strip analysis assumes simply
supported ends. When long column
end conditions are not simply
supported, or when they are dissimilar
for flexure and torsion, higher modes
may need to be considered, or classical
long column calculations performed.
Why define buckling modes?
• CUFSM and the finite strip analysis provide only the
elastic critical response of a member
• elastic critical buckling is a good input for design, but it
is not the design itself - thin-walled members have
important post-buckling behavior that is not considered
in this elastic buckling analysis
• Engineers have found that different failure
characteristics and strength exist in the different
buckling modes - thus design rules have been
developed that are unique for each mode. To use these
design rules the different definitions of the elastic
buckling modes are necessary.
Understanding Higher Modes
• Consider classic long column bucking. For thin-walled
members this generally includes the possibility of
flexure (weak and strong direction), torsion, and
flexural-torsional buckling
• Assume for a given unbraced length that flexural-
torsional buckling has the lowest stress, i.e., it is the 1st
mode. This implies that the other modes are higher
modes: the 2nd, the 3rd and so on.
• Now, if long column buckling has a 2nd (and a 3rd) mode
then it should stand to reason that local and distortional
buckling have higher modes as well. In fact many higher
modes exist and can be viewed using CUFSM.
Mode 1 - Long Mode 1 - Long
Flexural-torsional
buckling.
Note, red circle below
indicates where the
buckling mode is
determined.
distortional flexural-
local torsional 2
flexural
flexural-torsional
Mode 2 - Long Mode 2 - Long
Weak-axis flexural
buckling
local torsional 2
flexural
flexural-torsional
Mode 1 - Distortional Mode 1 - Distortional
Symmetric distortional
buckling
local torsional 2
flexural
flexural-torsional
Mode 2 - Distortional Mode 2 - Distortional
Anti-symmetric
distortional bukling
local torsional 2
flexural
flexural-torsional
Mode 1 - Local Mode 1 - Local
Local buckling
local torsional 2
flexural
flexural-torsional
Mode 2 - Local Mode 2 - Local
Local buckling with anti-
symmetric local web
buckling
local torsional 2
flexural
flexural-torsional
Utilizing Higher Modes
• Knowledge of higher mode response benefits
– Long column buckling determination when effective
length (KL) is different for different buckling modes
– examination of indistinct buckling modes and
understanding of switching between buckling modes
as a function of half-wavelength
– determination of member response if restraints were
in place (e.g., connecting the lips of the members
would change distortional buckling to anti-symmetric
distortional buckling)
“KL” and Higher modes
Pcr1 flexural
Pcr2
flexural-torsional
say KL for flexural-torsional is here
and say KL for weak-axis flexural is here
At a given L (half-wavelength) flexural-torsional is lower than weak-axis flexure, but considering the bracing
situation given, and examining KL (the effective “pin-pin” length) we find that Pcr2<Pcr1 and weak-axis flexure
would control for the imagined column end conditions. In this case Pcre = Pcr2.
Handling Indistinct Modes
• Examples
– Local and distortional combine
– No distinct distortional mode
Local and distortional combine
3
Local
1 2 3
No local mode exists in (1), unlipped channels and members with small stiffeners may have distortional only!
No distinct
Nomode
distortional distinct distortional mode
Consider the SSMA 600S200 - 033 of
Tutorial 2 with a slightly reduced lip
length (lip length = 0.46 in.)
The analysis results are given to the
right.
Distortional buckling clearly occurs in
the first mode as is shown at a half-
wavelength of 19.3 in. However, no
distinct minimum exists for
distortional buckling, so why not use
one of the lower values to the left?
How can one determine where local
buckling ends and distortional
buckling begins in this case?
Boundary Conditions and
No distinct distortional
Equation Constraints mode
Model 1 - Base model
Model 2 - Pins are enforced at all fold
lines. This allows local buckling, but
retards all other modes - thus curve 2
uniquely describes local buckling.
Model 3 - Equation constraints are
enforced such that the rotation at the
flange/lip juncture must equal the
rotation at the flange/web juncture - as
is the case in distortional buckling.
These constraints provide a minimum
in distortional buckling as shown to
the right.
The minimum bounding curves of 2
and 3 provide distinct boundaries
between local buckling and
distortional buckling of the member.
Solution Accuracy
• Number of elements
• Number of lengths
Number of elements
(simply supported plate 10 in. x 0.1 in., E=29500ksi, ν=0.3)
• Pure compression • Pure bending

11.4 74
11.3 Finite Strip 72 Finite Strip
buckling stress
Theory
buckling stress
11.2 Theory
70
11.1
11 68
10.9 66
10.8
64
10.7
10.6 62
1 2 3 4 1 2 3 4 5 6 7 8
number of elements number of elements
Conclusion:
At least 2 elements are needed in the compression region of any member flat for reasonable accuracy. This
can generally be insured by always having at least 4 elements in any flat portion of a member.
CUFSM2.5
Number of lengths
Consider the results for the
default C in bending with only a
few half-wavelengths and with
100 evenly spaced points.
The first analysis results are
clearly inadequate. The
minimums are not identified with
confidence due to the poor
resolution of the measured
response. However, in this case
errors in the estimated half-
wavelength are much greater
than the errors in the load factor.
The second analysis is superior
to the first, but is finer than
required. Since the minimums are
of primary interest an efficient
analysis will use more half-
wavelengths near these areas.
CUFSM2.5
CUFSM and Matlab

• The Matlab version of CUFSM allows much greater
flexibility than the standalone version.
• Within the Graphical User Interface you can use
mathematics, anything you could enter on the command
line in Matlab you can use in the GUI examples:
– entering 1:1:10 in the Lengths section will be interpreted as
1 2 3 4 5 6 7 8 9 10
– to evenly space 30 points in log space between 100 and 103
enter logspace(0,3,30) in the Lengths section
– to shift a member over 2 in. just add +2 to x the nodal
coordinates of all nodes
– etc.
CUFSM and Matlab
• The real power of the Matlab version is the
ability to access all the features of the program
from within your own programs (m-files.)
• In your m-files you can call the CUFSM routines
to do your pre- and post-processing as well as
perform the analysis directly and perform
parametric studies, optimization, etc..
Useful CUFSM Matlab functions
• Plotting and Post-Processing
– crossect.m: plot the cross-section, node numbers, springs, etc.
– strespic.m: show the stress distribution
– thecurve.m: single model, buckling curve plot
– thecurve2.m: multiple model, buckling curve plot
– dispshap.m: plot a 2D buckling mode shape
– dispshap3d.m: plot a 3D buckling mode shape
• Model Building
– templatecalc.m: Generate finite strip model from centerline dimensions of a C or Z
– doubler.m: double the number of elements in a model
– grosprop.m: calculate the properties (A, I, etc.) for a model
– yieldMP: given fy calculate the yield loads and moments for a model
– stresgen.m: given a load and/or moment calculate nodal stresses for a model
• Analysis
– strip.m: Perform finite strip calculations for a model
Matlab Example
• An example of performing a parametric study in Matlab
using your own files is presented in the following slides.
• The example m-file is: example_parameter_study.m
• The example
– set’s up the CUFSM inputs for a Cee section in compression
– performs analyses while varying the lip length
– saves all results
– plots the initial cross-sections used in the parametric study
– plots the buckling curves from the study
– plots the local and distortional buckling modes from the study
Parameter Study Example
The dimensions of the model are defined to the left.

Setup and perform analysis
The nodes and elements are defined just as they would be in the
graphical version of CUFSM, but now we use variables instead
of numbers. Most importantly, since d is an array we can perform
a loop and systematically vary the lip length ‘d’.
Analysis and saved data
this one line performs the analysis.

Plot sections and buckling curves
this plot is generated directly from the m-file!
Plot mode shapes
CUFSM2.5

excerpted from Chapter 2 of Schafer, B.W. (1998), Cold-Formed Steel Behavior and Design: Analytical and
Numerical Modeling of Elements and Members with Longitudinal Stiffeners, Ph.D. Thesis.
2. Elastic Buckling Solution Methods

for Cold-Formed Steel Elements and
Members
Cold-formed steel elements and members are often quite slender. it is not uncommon for elastic
buckling stresses to be much lower than the material yield stress. Therefore, buckling often
dominates the behavior. Although elastic buckling does not fully describe the ultimate behavior it
can be an important part of characterizing that behavior. Careful definitions of elastic buckling
shall be left to others, here an engineering approach is taken: elastic buckling refers to the
determination of a load in which the member is in equilibrium both in a straight and slightly bent
configuration. Such a loading and the corresponding bent configuration, form the buckling load
(stress) and buckling mode for the member.
Calculation of the elastic buckling stress and modes can be completed in a variety of ways. Three
common methods: finite element analysis, finite strip analysis, and classical Fourier Series
solutions are presented here. These techniques are by no means the only possible solution
methods. For example, infrequently used techniques such as the boundary element method can be
employed successfully (Elzein 1991).
Presentation of the finite element, finite strip, and Fourier Series methods are given in detail
because these methods are used throughout this work, and each have their own particular
advantages. Elastic buckling calculations via the finite element method are advantageous for
unusual geometry or varying boundary conditions along the length. In addition, the finite element
elastic buckling solution is a useful tool for generating imperfection configurations when more
general nonlinear analysis is to be performed. The finite strip method allows for an efficient
solution of problems which may have complex geometry in their cross-section, but are simple
along the length. The finite strip method lends itself to parameter studies due to its simple
requirements on input and the speed of the solution. The classical Fourier Series solutions are
useful because they have the potential to yield closed-form solutions, which are advantageous for
design approximations.
2.1 FINITE ELEMENT ANALYSIS

{section removed in this excerpt}
2
2.2 FINITE STRIP ANALYSIS

The finite strip method was originally developed by Y.K. Cheung. An excellent summary of the
method, and the theory behind it, can be found in his book (Cheung 1976). The use of the finite
strip method for understanding and predicting the behavior of hot-rolled steel members, and
cold-formed steel members has been greatly extended by G. Hancock. Hancock used the stiffness
matrices derived in Cheung’s book, and with some modification, created BFINST - a computer
program for solution of the elastic buckling problem of open thin-walled members via finite
strip. His early work in the field on I-Beams (Hancock 1977, 1978) led to the acceptance and
understanding of the use of the finite strip method. More recently, in a book on cold-formed steel
design, the use of the finite strip method as a design aid is explicitly shown (Hancock 1994).
Finite strip analysis provides a convenient and efficient way to determine the elastic buckling
stress and corresponding modes. In this section, an introduction to the basic differences between
the finite element method and the finite strip method is provided. This is followed by the initial
stiffness and geometric stiffness matrices for a finite strip. These matrices are presented in their
entirety because they are employed in a program written by the author. Details of the program
(CUSTRIP) and verification problems can be found in the Appendix.
2.2.1 Introduction and Comparison to Finite Element Method

The finite strip method can be considered as a specialization of the finite element method. The
basic methodology and theory of the two methods is identical. Shape functions are used to define
the displacement field in terms of nodal degrees of freedom. Strain is defined in terms of the
nodal degrees of freedom, since strain is a function of the displacement field. With strain defined,
and a constitutive relationship known, (i.e., the stress-strain relationship) stiffness coefficients for
the nodal degrees of freedom can be derived.
The only difference between the finite element and finite strip methods is the discretization of a
member. The finite strip method, is so named, because only a single element (strip) is used to
model the longitudinal direction. The result of the differing discretization assumptions are
depicted in Figure 2.1.
finite element finite strip

Figure 2.1 Finite Element and Finite Strip Discretization
3
The advantages and accuracy of the finite strip solution is dependent on a judicious choice of the
shape function for the longitudinal displacement field. Using the finite strip method, the total
number of equations needed for the solution are greatly reduced from that of a typical finite
element solution.
2.2.2 Initial Stiffness Matrix for Plates

The stiffness matrix presented in this section follows directly from the work presented by Cheung
(1976). The presentation here is also similar to the work of Hancock (1977) and Mulligan (1983).
The strip (element) and its degrees of freedom are shown in Figure 2.2.
a y
x
v1 v2 z
θ1 θ2
u1 u2
w1 w2
Figure 2.2 Strip, Degrees of Freedom
The standard definition of an initial stiffness matrix is apparent from:
{ F } = [ K ]{d }
or, expanded to explicitly show the nodal forces, nodal degrees of freedom, and the initial
stiffness submatrices: [Kuv] (plane stress) and [Kwθ] (bending):
ì Fu1 ü é 0 0 0 0ù ì u1 ü
ïF ï ê ï ï
0ú ï v1 ï
ï v1 ï
ï Fu2 ï
ê é
ê êë
Kuv ùúû 0
0
0
0
0
0
ú
0ú ï u2 ï
ï ï ê úï ï
ï Fv2 ï ê 0 0 0 0ú ï v 2 ï
íF ý= í ý
ï w1 ï ê0 0 0 0 ú ï w1 ï
ê0 0 ú
ï Mθ1 ï
ï ï ê
ê0 0
0
0
0
0
é
êë K ù ú ïθ 1 ï
wθ úû ú ïw ï
ï Fw2 ï ï 2ï
ï Mθ ï ê0 0 0 0 úï ï
î 2þ ë û îθ 2 þ
The initial stiffness matrix may be expressed as:

4
[ K ] = ò [ B]T [ E ][ B]dV or ò [ N '] [ E ][ N ']dV

T
Where: [B] or [N’] is the appropriate derivatives of the shape functions [N]. [N], the shape
function, is defined from (u v w)T = [N]{d}. Where, (u v w)T is the displacement field and {d} is a
vector of the nodal degrees of freedom.
For an orthotropic plate, and assuming no variation in the thickness (t) of the strip, [K] may be
expressed as:
[ K ] = t ò [ B]T [ D][ B]dA
The plate rigidities are defined as:
é Dx D1 0 ù
ê ú
[ D] = ê D1 Dy 0 ú
ê0 0 Dxy úû
ë
Ext 3 Eyt 3
Dx = Dy =
(
12 1 − ν x ν y ) (
12 1 − ν x ν y )
Gt 3 ν y Ext 3 ν x Eyt 3
Dxy = D1 = = .
12 (
12 1 − ν x ν y ) (
12 1 − ν x ν y )
As mentioned before, the key to the finite strip method is the selection of the shape functions [N].
The finite strip solution used here employs a polynomial in the transverse direction and a
harmonic function in the longitudinal direction. In this derivation, the longitudinal direction is
assumed to take the form of a half sine wave. This is consistent with the boundary condition of
simply supported ends. The advantage of such an assumption is that the integrals used in
forming the stiffness matrix decouple, and the solution is simplified. Other boundary conditions
result in far more complex stiffness matrices, see Cheung (1976).
The derivation of the initial stiffness matrix is in two completely decoupled parts. A pure plane
stress condition is assumed for the in plane u and v degrees of freedom. The w and θ degrees of
freedom are derived using classical small deflection plate theory to arrive at the bending initial
stiffness matrix. The two matrices [Kuv] and [Kwθ] are combined to form the total initial stiffness
matrix.
5
2.2.2.1 Plane Stress Initial Stiffness Matrix [Kuv]

The shape functions for use in determining the in plane stiffness matrix are:
éæ xö æ x ö ù ì u1 ü éæ xö æ x ö ù ì v1 ü a '
u = êç 1 − ÷ ç ÷ ú í ýYm v = êç 1 − ÷ ç ÷ úí ý Y
ëè b ø è b ø û îu2 þ ëè b ø è b ø û îv 2 þ mπ m
æ mπy ö
Ym = sinç ÷
è a ø
The expressions can be put in the general form [N] such that:
ì u1 ü
ïv ï
ìu ü ï 1ï
í ý [ ]í ý = [ N ]{d }
= N
îv þ ïu2 ï
ïîv2 ïþ
With the shape functions in this form, the strain-displacement matrix [B] can be written in terms
of derivatives of [N]:
ìεx ü ì ∂u ∂x ü
ï ï ï ï
íεy ý = í ∂v ∂y ý = [ B]{d } = [ N ']{d }
ïγ ï ï∂u ∂y + ∂v ∂x ï
î xy þ î þ
Using these definitions, and performing the necessary substitutions in to the expression for the
stiffness matrix presented before, the explicit plane stress stiffness matrix for an element, or strip
is given in Figure 2.3.
é æ aE1 abk m2 G ö ù
ê ç + ÷ symmetric ú
ê è 2b 6 ø ú
êæ ak mν x E 2 ak m G ö æ abk m2 E 2 aG ö ú
êç − ÷ ç + ÷ ú
ê è 4 4 ø è 6 2b ø ú
[ K uv ] = t ê æ aE abk 2 G ö ú
m æ ak mν x E 2 ak m G ö æ aE1 abk m2 G ö
ê çç − 1 + ÷÷ ç− − ÷ ç + ÷ ú
ê è 2b 12 ø è 4 4 ø è 2b 6 ø ú
2
ê ak ν E ak G æ abk m2 E 2 aG ö æ ak mν x E 2 ak m G ö æ abk m E 2 aG ö ú
êæç m x 2 + m ö÷ ç − ÷ ç− + ÷ ç + ÷ú
êëè 4 4 ø è 12 2b ø è 4 4 ø è 6 2b ø úû
mπ Ex Ey
where: k m = E1 = E2 =
a 1 − ν xν y 1 − ν xν y
Figure 2.3 Plane Stress Initial Stiffness Matrix

é æ 13ab 4 12a 2 ö ù
ê ç 70 k m Dy + 5b k m Dxy ÷ ú
êç ÷ symmetric ú
ê çç + 6a k 2 D + 6a D ÷÷ ú
m 1 x
êè 5b b3 ø ú
ê æ 3a a ö æ ab 3 4 4ab 2 ö ú
ê ç k m2 D1 + k m2 Dxy ÷ ç k m Dy + k m Dxy ÷ ú
ê ç 5 5 ÷ ç 210 15 ÷ ú
ê ç 3a 11ab 2 4 ÷ ç 2ab 2 2a ÷ ú
k m Dy ÷ ç + k m D1 + Dx ÷ ú
ê çè + 2 Dx +
b 420 ø è 15 b ø
ê 2
ú
12a 2 ö æ 13ab 4 a ö æ 13ab 4 12a 2 ö
[ K wθ ] = ê æ 9ab 4 ú
ê ç 140 k m D y − 5b k m Dxy ÷ ç k m D y − k m2 Dxy ÷ ç k m Dy + k m Dxy ÷ ú
÷ ç 840 5 ÷ ç 70 5b ÷
êç ú
ç 6a
êç − k D − 2 6 a ç a 2 3a ÷ çç + 6a k 2 D + 6a D ÷÷ ú
m 1 3
Dx ÷÷ ç − k m D1 − 2 Dx ÷ m 1 x
êè 5b b ø è 10 b ø è 5b b3 ø ú
3 2 3
êæ 13ab 2 a ö æ 3ab 4 ab 2 ö æ 11ab 4 a ö æ ab 4 4ab 2 öú
êç − k m4 D y + k m2 Dxy ÷ ç− k m Dy − k m Dxy ÷ ç− k m D y − k m2 Dxy ÷ ç k m Dy + k m Dxy ÷ ú
êç 840 5 ÷ ç 840 15 ÷ ç 420 5 ÷ ç 210 15 ÷ú
êç a 2 3a ÷ ç ab 2 a ÷ ç 3a 2 3a ÷ ç 2ab 2 2a ÷ú
êçè + k m D1 + 2 Dx ÷ø ç − k m D1 + Dx ÷ ç − k m D1 − 2 Dx ÷ ç + k m D1 + Dx ÷ ú
ë 10 b è 30 b ø è 5 b ø è 15 b øû
mπ
where: k m =
a
Figure 2.4 Bending Initial Stiffness Matrix

é70( 3T1 + T2 ) 0 70(T1 + T2 ) 0 0 0 0 0 ù
ê ú
ê 70(3T1 + T2 ) 0 70(T1 + T2 ) 0 0 0 0 ú
ê 70(T1 + 3T2 ) 0 0 0 0 0 ú
ê ú
ê 70( T1 + 3T2 ) 0 0 0 0 ú
g
[K ] =C
ê 8(30T1 + 9T2 ) 2b(15T1 + 7T2 ) 54(T1 + T2 ) −2b(7T1 + 6T2 ) ú
ê ú
ê b 2 (5T1 + 3T2 ) 2b(6T1 + 7T2 ) −3b 2 (T1 + T2 ) ú
ê symmetric 24( 3T1 + 10T2 ) −2b(7T1 + 15T2 )ú
ê ú
êë b 2 ( 3T1 + 5T2 ) úû
2
where: C = b( mπ ) 1680a
Figure 2.5 Geometric Stiffness Matrix

9
2.2.2.2 Bending Initial Stiffness Matrix [Kwθ]

The shape functions for use in determining the bending stiffness matrix are:
ì w1 ü
éæ 3 x 2x ö
2 3
æ 2x x ö 2
æ 3x2
2x ö 3
æx 2
x ö ù ïïθ ïï
w = Ym êç 1 − 2 + 3 ÷ xç1 − + ÷ ç 2 − 3 ÷ xç 2 − ÷ úí 1 ý
ëè b b ø è b b2 ø è b b ø èb b ø û ï w2 ï
îïθ 2 þï
æ mπy ö
Ym = sinç ÷
è a ø
With the shape functions in this form, the strain-displacement matrix [B] can be written in terms
of derivatives of [N]:
ì ∂ 2w ü
ï− 2 ï
ï ∂x2 ï
∂ w
{ε } = ïí− 2 ïý = [ B]{d } = [ N ']{d }
ï ∂2 y ï
ï ∂ w ï
ï ∂x∂y ï
î þ
The explicit bending stiffness matrix for an element, or strip is given in Figure 2.4.
2.2.3 Geometric Stiffness Matrix for Plates

The geometric stiffness matrix for a plate strip subjected to linearly varying edge traction can be
determined by either directly considering the higher order strain terms, or equivalently by
forming the potential energy due to in-plane forces. The latter is the method selected for use here.
Consider a strip with linear edge traction as shown in Figure 2.6. The tractions correspond to
linear edge stresses f1 and f2 via T1 = f1t and T2 = f2t.
10
a y
x
T1 T2
Figure 2.6 Strip with Edge Traction
The expression for the potential energy (U) due to the in-plane forces is:
x ö æ é ∂u ö
a b T
1 æ ∂v ∂w ù ì ∂u ∂v ∂w ü
U = ò ò ç T1 − ( T1 − T2 ) ÷ çç ê í ý ÷ dxdy
2 0 0è b ø è ë ∂y ∂y ∂y úû î ∂y ∂y ∂y þ ÷
ø
The derivatives in the expression for U, may be expressed in terms of the nodal degrees of
freedom {d}. The matrix resulting from differentiating the shape functions in this case is called
[G].
T
ì ∂u ∂v ∂w ü
í ý = [G ]{d }
î ∂y ∂y ∂y þ
The potential energy may now be expressed in terms of {d} and a matrix known as the geometric
stiffness [Kg].
1
U=
2
[ ]
{d } T K g {d }
a b
æ xö T
[ ]
K g = ò ò ç T1 − (T1 − T2 ) ÷ [G ] [G ]dxdy
è bø
0 0
The definition of [Kg] as a stiffness matrix follows directly from the energy derivation of an
initial stiffness matrix (see Cook 1989 or Ugural and Fenster 1995 for energy derivation of an
initial stiffness matrix, or Crisfield 1991 for more discussion of geometric stiffness matrices). For
the problem at hand, [Kg] now takes the explicit form given in Figure 2.5.
11
2.2.4 Finite Strip Solution Methods

In the previous three sections explicit matrices are given for the initial stiffness and geometric
stiffness of a discrete finite strip. For a member composed of multiple strips the contribution of
each strip must be formed into a global initial stiffness and geometric stiffness. Thus:
# strips # strips
[ K ] = å [ k ]n
n =1
[ ] å [k ]
and K g =
n =1
g n
The summation implies proper coordinate transformations and correct addition of the stiffness
terms in the global coordinates and degrees of freedom. The elastic buckling problem is a
standard eigenvalue problem of the following form:
[ K ]{d } = λ[ K g ]{d }
Where the eigenvalues λ, are the buckling load, and the eigenvectors the buckling modes.
Solution of such an eigenvalue problem may readily be solved in such programs as MATLAB
(MathWorks Inc., 1996).
Both [K] and [Kg] are a function of the length, a. Therefore, the elastic buckling stress and the
corresponding buckling modes are also a function of a. The problem can be solved for several
lengths, a, and thus a complete picture of the elastic buckling stress and modes can be
determined. The minima of such a curve could be considered as the critical buckling loads and
modes for the member. The program CUSTRIP generates such curves, for more information on
this program see the Appendix.
2.3 CLASSICAL SOLUTION: FOURIER SERIES METHOD FOR A

SS PLATE
{section removed in this excerpt}
2.4 COMPARISON OF FINITE ELEMENT, FINITE STRIP, AND

FOURIER SERIES SOLUTION
In the previous theoretical discussion of the finite element, finite strip, and Fourier Series
methods little insight is given on qualitative interpretation of the results and direct comparison of
the different methods. In order to address that issue two examples are examined. The first
concerns a channel section with a longitudinal web stiffener in bending. The finite strip method
and finite element method are compared and contrasted for this example. The second example
consists of an isolated simply supported plate with a longitudinal stiffener in bending. All three
methods are presented for this example, with particular emphasis given to the Fourier Series
approach.
12
2.4.1 Example: Channel Section with a Longitudinal Web Stiffener

The selected problem is a C-section with a longitudinal web stiffener, under pure bending. The
member geometry is from series of tests conducted at the University of Missouri-Rolla, the
section label is B-1a-1. (Phung et. al. 1978). The inset of Figure 2.7 shows the geometry of the
section. The observed failure mode is web-buckling and flange yielding.
2.4.1.1 Finite Strip Analysis

A finite strip analysis is conducted on UMR section B-1a-1. The section is loaded under pure
bending and analyzed at different lengths to determine the lowest buckling mode at each length.
Figure 2.7 shows the section geometry and the finite strip analysis results. The curve shown in
Figure 2.7 is typical of finite strip results. In this case, at half-wavelengths (½λ) greater than
approximately 65 in. lateral buckling occurs at a lower load than local web buckling.
The minima indicate the lowest load factor for a particular mode. Examining the finite strip
results for half-wavelengths less than 65 inches, three distinct minima exist: local flange buckling
at a ½λ = 1.60”, and a load factor = 1.53 (i.e., Mcr = 1.53My); local web buckling at a ½λ = 8.63”
and a load factor = 0.69; and distortional Buckling at a ½λ = 21.60” and a load factor = 0.71.
2.4.1.2 Finite Element Analysis

FEM Comparison 1:
The elastic buckling finite element analysis is conducted using ABAQUS. The analysis consists
of evaluating UMR B-1a-1 as simply supported at the end cross-sections, and unbraced over a
length of 50 inches. Consistent nodal loads are derived for the end sections such that a pure
bending moment equal to the yield moment is applied as a reference load.
Figure 2.9 - Figure 2.12 show the predicted buckling mode shapes for the three minima of the
finite strip analysis. The results indicate that both the finite element analysis and finite strip
analysis predict similar buckling load factors and mode shapes. Both methods report a minimum
load factor/eigenvalue of 0.69 (Mcr = 0.69My) consisting primarily of local web buckling. Both
methods also report local flange buckling occurring at a minimum load factor/eigenvalue of 1.53.
Differences do exist between the finite element analysis results and the finite strip results. In
Figure 2.11 the buckling mode from the finite element analysis at a load factor/eigenvalue of
1.33 is shown. In Figure 2.11 the mode shape in the compression flange is mixed between a
longer wavelength distortional buckling and short wavelength local buckling in the compression
flange. The result is that one distinct half-wavelength does not characterize this mode. Mixed
wavelength modes are common in the finite element analysis but impossible in the given finite
strip formulation.
FEM Comparison 2:
To further highlight the difference between the two methods, and in order to provide a more
direct comparison between the finite element and finite strip buckling analysis, a second test is
conducted. In this test the actual member length studied in the finite element analysis is varied.
13
This is similar to the method employed by the finite strip analysis for finding the first mode.
Figure 2.8 shows the results for the two methods.
The curve shows the difference between analyzing the structure at different lengths and analyzing
at different half-wavelengths. For instance, the lowest mode appears to require a larger physical
length to develop than is indicated by the finite strip half-wavelength. In addition, the finite
element analysis indicates that the lowest mode always decreases with the structure length - this
is not so with the first mode half-wavelength curve of a finite strip analysis. The primary
conclusion drawn from this analysis is the "up" part of a first mode half-wavelength curve should
be considered fictitious. The lowest previous mode will still govern in the physical structure as
the length is increased.
The box on the final point of the finite element curve of Figure 2.8 indicates the member length
employed for FEM Comparison 1.
14
2.5
2.0
M / My 1.5
1.0
0.5
0.0
0 10 20 30 40 50 60
½λ (in.)
Figure 2.7 UMR B-1a-1 Finite Strip Results
2.5
2.0 Finite Strip

Finite Element
1.5
M / My
1.0
0.5
0.0
0 10 20 30 40 50 60
½λ or L (in.)
Figure 2.8 Direct Comparison of finite element and finite strip

Figure 2.9 Local Web Buckling Figure 2.11 Mixed-Mode Buckling
Figure 2.10 Distortional Buckling Figure 2.12 Local Flange Buckling

16
2.4.2 Example: SS Plate with a Longitudinal Stiffener in Pure Bending

In order to compare the results of the introduced analytical methods, and to understand better
how to interpret the results an example problem has been selected for study by all three methods.
The selected problem is shown in Figure 2.13. The loading is ξ = 2, or equal and opposite
stresses.
For the Fourier analysis six transverse terms and 4 separate longitudinal sine terms are employed.
The analysis is conducted using CUPLATE a program written in MATLAB, CUPLATE is
discussed in the Appendix. The results are presented in a k vs. β or “buckling curve” as seen in
Figure 2.14. The curves show that for small β (short plates) local buckling occurs. However as β
increases the overall buckling of the plate controls. For reference, local buckling of a plate
without a stiffener would result in a plate buckling coefficient (k) of approximately 24.
The example problem may also be examined using finite strip and finite element techniques. The
finite element solution is forced to mimic the methods used by the finite strip and Fourier Series
solution by varying the length modeled and then plotting the first eigenmode for each length. In
addition, the solutions for the buckling load are in terms of stress, or ratio of stress to fy. These
values are transformed into plate buckling coefficients. The finite strip solution, finite element
solution using ABAQUS, and the minimum of the Fourier Series solution are all shown in Figure
2.15. A minimum curve similar to the Fourier Series could be generated using finite strip as well,
but this is not done. Thus, the upwards branch of the finite strip curve after β > 1.5 is not a cause
for concern. All three methods yield minimum values within 5% of one another.
A 10mm
f1 f1 x
c=40mm
b=200mm 2mm
f1 f1 1mm
a AA
A
y
Figure 2.13 Geometry of Example Problem
160 160
140 140
120 120
100 100
k
k
80 80
60 60
40 40
local Finite Strip
20 overall 20 Finite Element
Fourier Series
0 0
0 1 2 3 4 5 0 1 2 3 4 5
β β
Figure 2.14 Fourier Analysis Solution Figure 2.15 Comparison of Buckling Solutions
18
2.4.3 Conclusions and Comments about Comparison

The primary conclusion of the two studied examples is that all three methods are
adequate for characterizing the elastic buckling stress. However, the finite element
implementation is the most robust of the three. The finite element procedure allows for
varying support conditions at the ends and along the length. In addition, the geometry
does not have to be regular along the length.
The finite strip implementation provides a useful tool for examining the behavior of cold-
formed steel members. In particular, since the solution is efficient, parameter studies
examining the elastic buckling stress proceed quickly. The finite strip method also
properly accounts for the restraint between elements. As a result, it provides a way to
examine the elastic buckling interaction of various elements, such as the web and flange
of a bending member.
The classical Fourier Series solution only has limited use for solving general problems.
As implemented it only applies for specialized boundary conditions. However, as a
method for design it can be ideal. In particular, the method can be used to generate
closed-form equations for the elastic buckling stress in special situations. It is in this
capacity that its use is the most important.

D S M C - F S D: Irect Trength Ethod OF OLD Ormed Teel Esign

Uploaded by

Copyright:

Available Formats

You might also like

D S M C - F S D: Irect Trength Ethod OF OLD Ormed Teel Esign

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

D S M C - F S D: Irect Trength Ethod OF OLD Ormed Teel Esign

Uploaded by

Copyright:

Available Formats

DRAFT - JANUARY 7, 2002

Design Manual for

DIRECT STRENGTH METHOD

DRAFT – DECEMBER 6, 2002

DIRECT STRENGTH METHOD

Appendix 1: Design of Cold-Formed Steel Structural

1.1 General Provisions

1.1.1.1 Pre-qualified Columns

1.1.1.2 Pre-qualified Beams

0.27 < d/b < 0.56

1.1.2 Elastic Buckling

1.1.3 Deflection Determination

1.2.1 Flexural, Torsional, or Flexural-Torsional Buckling

1.2.2 Local Buckling

1.2.3 Distortional Buckling

1.3.1 Lateral-Torsional Buckling

1.3.2 Local Buckling

for λl > 0.776 Mnl = 1 − 0.15 crl   crl  M ne (Eq. 1.3.2-2)

1.3.3 Distortional Buckling

DIRECT STRENGTH METHOD

COMMENTARY Appendix 1: Design of Cold-Formed Steel

The remaining commentary is organized to parallel the structure of Appendix 1.

1.1 General Provisions

1.1.1.1 Pre-qualified Columns

Figure 1.1-1 Cross-section of columns used for calibration

Table 1.1-1 Geometry of lipped channel columns

Table 1.1-2 Geometry of lipped zed columns

Table 1.1-3 Geometry of lipped channel with a web stiffener columns

Table 1.1-4 Geometry of hat columns

Table 1.1-5 Geometry of rack section columns

1.1.1.2 Pre-qualified Beams

Cohen (1987) Ellifritt et al. (1997) Laboube and Yu (1978)

Schuster (1992) Shan et al. (1994) Willis and Wallace (1990)

Table 1.1-6 Geometry of lipped channel beams

Table 1.1-7 Geometry of unlipped channel beams

Table 1.1-9 Geometry of lipped channel with a web stiffener beams

Table 1.1-10 Geometry of liped zed beams

(a) Acharya/Papazian sections (b) Desmond sections (c) König sections

Table 1.1-11 Geometry of hat section beams

(a) Bernard deck (b) Höglund deck

Table 1.1-12 Geometry of trapezoidal deck beams

1.1.2 Elastic Buckling

1.1.2.1 Elastic Buckling Numerical Solutions

equilibrium must be maintained, or if neglected done so conservatively. The final comment

1.1.2.1.1 Numerical, Finite strip analysis notes on Pcrl, Mcrl

Additional examples, and tutorials are available online at www.ce.jhu.edu/bschafer/cufsm.

1.1.2.1.2 Numerical, finite strip analysis notes on Pcrd, Mcrd

1.1.2.1.3 Numerical, finite strip analysis notes on Pcre, Mcre

Further examples and discussion are available online at www.ce.jhu.edu/bschafer/cufsm.

AISI96 Ex. 1−8

Local Mcr/My=0.67 Distortional M cr/My=0.50

AISI96 Ex. 1−8

0.15 Local Pcr/Py=0.13 Distortional Pcr/Py=0.20

AISI96 Ex. 1−10

Local Mcr/My=0.86 Distortional M cr/My=0.65

AISI96 Ex. 1−10

AISI96 Ex. 1−9

AISI96 Ex. 1−11