Understanding and classifying local,

distortional and global buckling in open

thin-walled members
 Motivation and challenges
Modal definitions based on mechanics
Implementation
Examples
Thin-walled members
 What are the buckling modes?
member or global buckling EI 2
Pcr
KL 2

plate or local buckling

E t
2 2

f cr k 2

12 1 b
other cross-section buckling modes?
distortional buckling?
stiffener buckling?
Buckling solutions by the finite strip method
Discretize any thin-walled
cross-section that is regular
y along its length
The cross-section strips
are governed by simple
mechanics
membrane: plane stress
finite strip bending: thin plate theory
Development similar to FE
m y
Ym sin All modes are captured
a
 Typical modes in a thin-walled beam

400

350

300

250

Mcr
200
BUCKLING CURVE BUCKLING CURVE

150 500 500

400 400
load factor

100
load factor

300 300

200 BUCKLING CURVE

50 200

100 5.0,172.76 5.0,172.76

100
20.0,133.65 500
20.0,133.65
0

Lcr
0
0 0 1 0 1
2 3 2400 3
10 10 0 10 1 10 2 3

load factor
10 10 10 10
half-wavelength 300
half-wavelength

200

100 5.0,172.76
20.0,133.65

0
0 1 2 3
10 10 10 10
half-wavelength

local buckling distortional buckling lateral-torsional buckling

 Why bother? modes strength
1

Elastic buckling
Yield
Global
0.8
Local
Distortional
strength

0.6

0.4

0.2

0
0 0.5 1 1.5 2 2.5 3

slenderness
Whats wrong with what we do now?
 What mode is it?

Local LTB
 Are our definitions workable?
Distortional buckling. A mode of buckling
involving change in cross-sectional shape,
excluding local buckling

Not much better than

you know it when you see it

definition from the Australian/New Zealand CFS standard,

the North American CFS Spec., and the recently agreed
upon joint AISC/AISI terminology
 We cant effectively use FEM
We need FEM methods to solve the type of general
stability problems people want to solve today
tool of first choice
general boundary conditions
handles changes along the length, e.g., holes in the section

30 nodes in a cross-section
100 nodes along the length
5 DOF elements
15,000 DOF
15,000 buckling modes, oy!

Modal identification in FEM is a disaster

 Generalized Beam Theory (GBT)
GBT is an enriched beam element that performs its
solution in a modal basis instead of the usual nodal
DOF basis, i.e., the modes are the DOF

GBT begins with a traditional beam element and then

adds modes to the deformation field, first Vlasov
warping, then modes with more general warping
distributions, and finally plate like modes within flat
portions of the section

GBT was first developed by Schardt (1989) then

extended by Davies et al. (1994), and more recently
by Camotim and Silvestre (2002, ...)
 Generalized Beam Theory
Advantages
modes look right
can focus on individual modes or subsets of modes
can identify modes within a more general GBT analysis
Disadvantages
development is unconventional/non-trivial,
results in the mechanics being partially obscured
not widely available for use in programs
Extension to general purpose FE awkward

We seek to identify the key mechanical assumptions of

GBT and then implement in, FSM, FEM, to enable
these methods to perform GBT-like modal solutions.
GBT inspired modal definitions
 G D L O
modes modes modes modes
#1 xy = 0, x = 0, v is linear Yes Yes Yes No
#2 y 0 Yes Yes No -
#3 y = 0 Yes No - -

Global modes are those

deformation patterns that
satisfy all three criteria.
 G D L O
modes modes modes modes
#1 xy = 0, x = 0, v is linear Yes Yes Yes No
#2 y 0 Yes Yes No -
#3 y = 0 Yes No - -

#1 membrane strains:
xy = 0, membrane shear
strains are zero,
x = 0, membrane transverse
strains are zero, and
v = f(x), long. displacements
are linear in x within an
element.
 G D L O
modes modes modes modes
#1 xy = 0, x = 0, v is linear Yes Yes Yes No
#2 y 0 Yes Yes No -
#3 y = 0 Yes No - -

#2 warping:
y 0,
longitudinal membrane
strains/displacements are
non-zero along the length.
 G D L O
modes modes modes modes
#1 xy = 0, x = 0, v is linear Yes Yes Yes No
#2 y 0 Yes Yes No -
#3 y = 0 Yes No - -

#3 transverse flexure:
y = 0,
no flexure in the transverse
direction. (cross-section
remains rigid!)
 G D L O
modes modes modes modes
#1 xy = 0, x = 0, v is linear Yes Yes Yes No
#2 y 0 Yes Yes No -
#3 y = 0 Yes No - -

Distortional modes are those

deformation patterns that
satisfy criteria #1 and #2,
but do not satisfy criterion
#3 (i.e., transverse flexure
occurs).
 G D L O
modes modes modes modes
#1 xy = 0, x = 0, v is linear Yes Yes Yes No
#2 y 0 Yes Yes No -
#3 y = 0 Yes No - -

Local modes are those

deformation patterns that
satisfy criterion #1, but do
not satisfy criterion #2 (i.e.,
no longitudinal warping
occurs) while criterion #3
is irrelevant.
 G D L O
modes modes modes modes
#1 xy = 0, x = 0, v is linear Yes Yes Yes No
#2 y 0 Yes Yes No -
#3 y = 0 Yes No - -

Other modes (membrane

modes ) do not satisfy
criterion #1. Note, other
modes typically do not
exist in GBT, but must
exist in FSM or FEM due
to the inclusion of DOF for
the membrane.
 example of
implementation into FSM
 Constrained deformation fields
x x u1 my
FSM membrane disp. fields: u ( x , y) 1 sin
b b u 2 a

x x v1 my
v( x , y) 1 cos
b b v 2 a

u u u 1 u 2 my
a GBT criterion is x 0 so x sin 0
x x b a

u1 1 0 0
v u
therefore 1 0
u 2 1
1
0
0
0 v1 or d Rd r
v 2
v 2 0 0 1
 impact of constrained deformation field

general FSM

constrained FSM
 Modal decomposition
Begin with our standard stability (eigen) problem
K ed K gd
Now introduce a set of constraints consistent with a
desired modal definition, this is embodied in R
K e Rdr K g Rdr
Pre-multiply by RT and we create a new, reduced
stability problem that is in a space with restricted
degree of freedom, if we choose R appropriately we
can reduce down to as little as one modal DOF
R T K e Rdr R T K g Rdr
K er d r K grd r
examples
 lipped channel in compression
typical CFS section
50mm
20mm
Buckling modes include
local,
200mm

distortional, and
P
global
Distortional mode is
t=1.5mm
indistinct in a classical
FSM analysis
 classical finite strip solution

1,2
all-mode
1 global
0,8 dist.
(Pcr/Py)

local
0,6

0,4

0,2

0
10 100 1000 10000
buckling length (mm)
 modal decomposition

1,2
all-mode
1 global
0,8 dist.
(Pcr/Py)

local
0,6

0,4

0,2

0
10 100 1000 10000
buckling length (mm)
 modal identification
100
80 global
modes (%)

60 dist
40 local
20 other
0
1,2 10 100 1000
buckling length (mm)
10000
all-mode
1 global
0,8 dist.
(Pcr/Py)

local
0,6

0,4

0,2

0
10 100 1000 10000
buckling length (mm)
 I-beam cross-section
80mm textbook I-beam
tf=10mm Buckling modes include
local (FLB, WLB),
200mm

distortional?, and
M
global (LTB)
tw=2mm If the flange/web
juncture translates is it
distortional?
 classical finite strip solution

20 all-mode
global
15 local
(Pcr/Py)

10

0
10 100 1000 10000
buckling length (mm)
 modal decomposition

20 all-mode
global
15 local
(Pcr/Py)

10

0
10 100 1000 10000
buckling length (mm)
 6

Pcr/
modal identification
4
2
0
100 10 100 1000 10000
buckling length (mm)
80
modes (%)

global
60 dist
40 local
20 other
0
10 100 1000 10000
all-mode
20 buckl. length (mm)
global
15 local
(Pcr/Py)

10

0
10 100 1000 10000
buckling length (mm)
 concluding thoughts
Cross-section buckling modes are integral to
understanding thin-walled members
Current methods fail to provide adequate solutions
Inspired by GBT,
mechanics-based definitions of the modes are possible
Formal modal definitions enable
Modal decomposition (focus on a given mode)
Modal identification (figure out what you have)
within conventional numerical methods, FSM, FEM..
The ability to turn on or turn off certain mechanical
behavior within an analysis can provide unique insights

Much work remains, and definitions are not perfect

 acknowledgments
Thomas Cholnoky Foundation
Hungarian Scientific Research Fund
U.S., National Science Foundation
 varying lip angle in a lipped channel
lip angle from 0 to 90
120mm
Where is the local
q
distortional transition?
200mm

P ?

t=1mm
 q
classical finite strip solution
(a) q = 0
0,25
(b) = 18
0,2 (c) = 36
(d) = 54
Pcr/Py

0,15
(e) = 72
0,1 (f) = 90
0,05

0
10 100 1000 10000
buckling length (mm)

Local? Distortional? L=700mm, q=54-90

Local? Distortional? L=170mm, q=0-36
 (a) q = 0
0,25
(b) = 18
0,4
0,2 (c) = 36
(a)
0,3
(b) (d) = 54
Pcr/Py
Pcr/Py
0,15
0,2 (e) = 72
0,1 (f) = 90
0,1
0,05
0
0 10 100 1000 10000

100 10 100 1000 10000

(a) buckling length (mm)
80
q0
modes (%)

60 global
dist
40
local
20 other
0
10 100 1000 10000
100
q18 80
(b)
modes (%)

60 global
dist
40 local
20 other
0
10 100 1000 10000
buckl. length (mm)
What mode is it?

?
 lipped channel with a web stiffener
modified CFS section
50mm
20mm
Buckling modes include
local,
200mm

2 distortional, and
P
global
20mm x 4.5mm
Distortional mode for the
t=1.5mm
web stiffener and edge
stiffener?
 classical finite strip solution

0,6
all-mode
global
0,4 dist.
(Pcr/Py)

local

0,2

0
10 100 1000 10000
buckling length (mm)
 modal decomposition

0,6
all-mode
global
0,4 dist.
(Pcr/Py)

local

0,2

0
10 100 1000 10000
buckling length (mm)
 0,2

Pcr
0,1
modal identification
0
100 10 100 1000 10000
(e) buckling length (mm)
80
modes (%)

global
60 dist
40 local
20 other
0
10 100 1000 10000
0,6 buckl. length (mm)
all-mode
global
0,4 dist.
(Pcr/Py)

local

0,2

0
10 100 1000 10000
buckling length (mm)
Coordinate System
 FSM Ke = Kem + Keb
Membrane (plane stress) K t B D BdA
T

x x u1
u 1 Ym
b b u2
x x v1 a '
v 1 Ym
b b v 2 m

m y
Ym sin
a

x u x


y v y Bd N 'd
u y v x
xy
 FSM Ke = Kem + Keb
Thin plate bending K t B D BdA
T

w1
3x 2 2 x 3 2x x 2 3x 2 2 x 3 x 2 x q 1
w Ym 1 2 3 x 1 2 3 x 2
b b b b2 b b b b w2
q 2
m y
Ym sin
a

2w
2
x x2
w
y 2 B d
2 y
xy w
xy

 FSM Ke = Kem + Keb
K t B D BdA
T

Membrane (plane stress)

x x u1
u 1 Ym
b b u2

x x v1 a '
v 1 Ym
b b v 2 m

m y
Ym sin
a

w1
3x 2 2 x 3 2x x 2 3x 2 2 x 3 x 2 x q 1
w Ym 1 2 3 x 1 2 2 3 x 2
b b b b b b b b w2
q 2
 FSM Solution
Ke
Kg
Eigen solution
FSM has all the cross-section modes in there
with just a simple plate bending and membrane
strip
 Classical FSM
Capable of providing complete solution for all
buckling modes of a thin-walled member
Elements follow simple mechanics
membrane
u,v, linear shape functions
plane stress conditions
bending
w, cubic beam shape function
thin plate theory
Drawbacks: special boundary conditions, no variation
along the length, cannot decompose, nor help identify
mechanics-based buckling modes
 Are our definitions workable?
Local buckling. A mode of buckling involving plate
flexure alone without transverse deformation of the
line or lines of intersection of adjoining plates.
Distortional buckling. A mode of buckling involving
change in cross-sectional shape, excluding local
buckling
Flexural-torsional buckling. A mode of buckling in
which compression members can bend and twist
simultaneously without change of cross-sectional
shape.

* definitions from the Australian/New Zealand CFS standard

 finite strip method
Capable of providing complete solution for all
buckling modes of a thin-walled member
Elements follow simple mechanics
bending
w, cubic beam shape function
thin plate theory
membrane
u,v, linear shape functions
plane stress conditions
Drawbacks: special boundary conditions, no variation
along the length, cannot decompose, nor help identify
mechanics-based buckling modes
 Special purpose FSM can fail too

0,8
(Pcr/Py)

0,6

0,4

0,2

0
10 100 1000 10000
buckling length (mm)
 Experiments on cold-formed steel columns

267 columns , b = 2.5, f = 0.84