Beruflich Dokumente
Kultur Dokumente
of Marine Structures
Chapter 2:
Buckling of Bars and Frames
by
Professor Jørgen Amdahl
MTS-2014.16.02
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 2 of 68
CONTENTS
z, w
y Mi
N w N
x
x
l
For
Fornontrivial
nontrivial solution, we must
solution, we musthave
have
kl = nπ, , n n =1,1,2,......
kl n 2,......
whichimplies
which implies that
that the smallest
smallestvalue
valueofofk kisisatatn n= =1,1,
and is given
and as ask k =. πThis
is given yields
. This yields
l l
π 2 EI
N = N cr = N E = (2.5)
l2
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 4 of 68
Stable path
wo/l
0 0
wo/l
0.8
0.8
Figure Figure
2.2 Linear And
2.2 Linear AndNon-linear Buckling
Non-linear Buckling Solution.
Solution.
ds
φ
dx dy
dφ
x
Figure 2.3 Exact Definition of Curvature.
Figure 2.3 Exact Definition of Curvature.
In order to make the derivations valid for finite deflection a more accurate curvature expression
must be introduced. That will be done using Figure 2.3.
In order to make the derivations valid for finite deflection a more accurate curvature expression
must be introduced. That will
The exact definition be donereads
of curvature using Figure 2.3.
1 d
= =
The exact definition of curvature reads (2.6)
ds
1 dφ
κ= = (2.6)
ρ ds
where dφ is the change in angle, and ds is the arc length. Using Figure 2.3 we can write,
ds 2 = dx 2 + dw2 (2.7)
dw
tan φ = (2.8)
dx
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 5 of 68
dφ dx
κ= (2.9)
dx ds
1
ds ⎛ ⎛ dw ⎞ ⎞ 2
2
= ⎜1 + ⎜ ⎟ ⎟ (2.10)
dx ⎜⎝ ⎝ dx ⎠ ⎟⎠
2
1 w
dφ = d 2 dx (2.11)
cos φ
2
dx
which implies that,
dφ 2
d w 1 2
d w
= cos 2φ 2 =
dx dx 1 + tan 2φ dx 2
1 2
d w (2.12)
= 2 2
⎛ dw ⎞ dx
1+ ⎜ ⎟
⎝ dx ⎠
d 2w
κ= dx 2 (2.13)
3
⎛ ⎛ dw ⎞ 2
⎞ 2
⎜⎜ 1 + ⎜ ⎟ ⎟⎟
⎝ ⎝ dx ⎠ ⎠
For small deflections, this expression approaches the linear expression, given as
2
w
κ=d 2
(2.14)
dx
If the exact curvature expression is introduced in Equation (2.2), the differential equation
becomes
3
w ,xx + k 2(1 + w ,2x ) 2 w = 0 (2.15)
The exact solution for the above equation has been presented by Timoshenko. An approximate
solution is given by
2
N π2⎛w ⎞
= 1+ ⎜ o ⎟ (2.16)
NE 8 ⎝ l ⎠
This solution is traced in Figure 2.2. The capacity increases for large deflections.
It is a fairly easy task to calculate the buckling length for other idealized boundary conditions.
In all cases it is found convenient to relate the buckling strength of the actual compressed
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 6 of 68
member with length l to the buckling strength of an equivalent pin-ended member with length
le. The length le is termed the effective length of the actual member.
The critical load of the actual member can now be written as
π 2 EI
N cr = 2 (2.17)
le
The ratio between the effective length and actual member length is denoted by K so that
le = Kl (2.18)
The factor K for some idealised boundary conditions are given in Figure 2.4 Because idealised
boundary conditions seldom are attained for actual structures, values recommended for design
are slightly conservative.
Figure 2.4 Effective length factors for some idealized boundary conditions.
The effective length can physically be interpreted as the distance between the inflection points
for the member. For several of the cases shown in Figure 2.4 it is an easy task to determine the
inflection points by visual inspection.
y w0
N w0i N
x
U H = −EI2 0∫w, xx w i , xx dx
x i x EI ⎜ ⎟ w o w oi 2
2 2 2 ⎝ ll ⎠ 2
0 (2.21)
The critical load is obtained from
1 l the principle of minimum 1 potential energy,
2 δ (U + H ) = 0 ,
2 2 2 2
which gives, H N w, x w i , x dx N w o w oi
20 4 2 2l 2
1 ⎛π ⎞ 1 ⎛π ⎞
EI ⎜ ⎟ ⋅ 2(wo − woi ) − N ⎜ ⎟ ⋅ ⋅ 2 wo = 0
The critical load is obtained from2 the⎝ principle
⎠ ⎠ 2 energy, U H
2 2 ⎝ potential
of minimum (2.22)0 ,
which gives, π EI wo − woi =
2
⎛ w ⎞
⇒N= N E⎜⎜ 1 − oi ⎟⎟
4
1 ⎝ wo ⎠
2
1 wo
or, EI 2 wo woi N 2 wo 0
2 1
2 2 2
wo = woi2 EI N
(2.22)
(2.23)
N 1 − wo woi N E 1 woi
N E wo wo
or,
where the second term can be considered as an amplification factor.
1
wo woi (2.23)
In this case there is not such a true N
1 buckling problem. The column starts to deflect immediately.
The capacity is always smaller than N NE (which is approached asymptotically), Figure 2.6. The
larger the initial imperfection the more non-linear the response becomes.
where the second term can be considered as an amplification factor.
In this case there is not such a true buckling problem. The column starts to deflect immediately.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 8 of 68
N
NE
Increasing woi
w o wo wo
1 2
Figure 2.6 Load-Displacement Relationship for Initially Imperfect Column.
External moment: Me = −N w ( x)
Internal moment: M i = − EI ( w, xx − wi , xx )
Equilibrium yields,
EI ( w, xx − wi , xx ) + Nw = 0
(2.24)
⇒ w, xx +k 2 w = wi , xx
The total solution to Equation (2.24) is given by the sum of the homogeneous solution, wh, and
the particular solution, wp.
• Homogeneous solution:
wh = A sin kl = 0
kl = π (for N = NE )
A=0 otherwise.
• Particular solution:
πx
Assume wi = wio sin , then the particular solution can be written as
l
πx
w p = C sin (
l
2.25)
1 πx
wN
Then, the total solution (for N < =Ew p = be written
) can sin
wioas, (2.28)
N l
1−
1 NE x
w wp sin (2.28)
N wio 1
1
and the bending moment NE
1 1 x
and the bending moment M = Nw = Nwio sin π (2.29)
N l
11 − x
M Nw Nwio sin
NE (2.29)
N
is amplified accordingly. 1
NE
isThe
amplified accordingly.
deformations start to grow immediately when imperfections are present. The axial force
approaches NE asymptotically for large deformations.
The deformations start to grow immediately when imperfections are present. The axial force
approaches NE asymptotically for large deformations.
N N
x
Previously, the differential equation was established considering moment equilibrium as,
EI w, xx −
Nw = 0 (2.30)
M
Mi e
This lateral force must balance the true lateral force, i.e.
πx
EIw ,xxxx + Nw ,xx = qo sin
l (2.32)
q πx
⇒ w ,xxxx + k w ,xx = o sin
2
EI l
The total solution is given by w = wh + w p . The homogeneous solution, wh, is as for a pure axial
force, and the particular solution is assumed to be
πx
w p = C sin (2.33)
l
which, when inserted to Equation (2.33), gives
4 2
⎛π ⎞ ⎛π ⎞ q
C ⎜ ⎟ − k2⎜ ⎟ C = o (2.34)
⎝l⎠ ⎝l⎠ EI
4
q ⎛l⎞ 1 1
⇒C = o ⎜ ⎟ = wo (2.35)
EI ⎝ π ⎠ 1 − N 1−
N
NE NE
where wo is the deformation at mid-span when N = 0.
The bending moment can be determined from equilibrium considerations or from the
relationship:
2
⎛l⎞ 1 1
M = EIw, xx = q0 ⎜ ⎟ = M0
⎝ π ⎠ 1− N 1−
N
NE NE
We find the same amplification factor for sinusoidal lateral loading as the one obtained for
sinusoidal initial imperfection. This is always the case if the first order moment has exactly the
same shape as the second order (additional bending moment). In all other cases, this is not
exactly correct.
In most cases, however, the amplification factor may be used with good accuracy for more
general load cases. This is shown for some simple load cases in Figure 2.8, and given in Table
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 11 of 68
2.1. The actual buckling length has to be used in assessing the amplification factor.
is always good for a maximum deflection (~2-3% deviation), but is somewhat poorer for
maximum bending moment. Despite this it is widely used. Although the discrepancy can be
significant for large axial force, it should be recalled that the total stress is dominated by the
mean compressive stress, so that the discrepancy for the total stress becomes significantly
smaller.
8 8
Load Case 4
6 6
Amplification Factor
Amplification Factor
4 4
Load Case 1 & 3
2 2
0 0
0 0.4 0.8 1 0 0.4 0.8 1
N/NE N/NE
(a) Amplification Factors for Mid-span Deflection. (b) Amplification Factors for Mid-span Moment.
Figure 2.8 True versus Simplified Amplification Factor for Some Load and Boundary Conditions.
π N
2 π EI l 2
2
3 ⎛ tan a ⎞ 24 ⎛ 1 a2 ⎞ 24 ⎛ tan a − a ⎞ 24 ⎛ a a⎞
f wmax ⎜ − 1⎟ − − 1 ⎟ a 3 ⎜⎝ ⎟ 3 ⎜
tan − ⎟
a2 ⎝ a ⎠ 5a 4 ⎜⎝ cos a 2 ⎠ 2 2⎠ a ⎝ 2 2⎠
tan a 2 ⎛ 1 ⎞ a 2 ⎛ 1 ⎞
⎜ − 1⎟ tan ⎜ − 1⎟
fm a a 2 ⎝ cos a ⎠ 2 a 2 ⎝ cos a ⎠
a 2
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 12 of 68
2.4Design of Beam-Columns
π 2
EI π 2E i 2
σE = 2
= 2 (2.37)
Al k lk
I lk
i= and λ = (2.38)
A i
π 2E
σE = (2.39)
λ2
The slenderness that gives σ E = σ Y is denoted as λY, i.e.
π 2
E
σY = (2.40)
λ 2
Y
The normalized Euler stress is plotted in Figure 2.9. It is seen that the Euler stress exceeds the
yield stress for a reduced slenderness ratio of λ < 1. This is unreasonable, therefore a transition
curve must be introduced as indicated. The transition curve, which is generally on the basis of
experiments and numerical analyses, may vary from one code to another.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 13 of 68
σE
Euler
σY
Proportionality limit
0.5
1.0 2 λ
inelastic elastic
σ cr = 1 λ≥ 2
2
σY λ
σ cr = 1 − 1 2 λ≤ 2
λ
σY 4
The ECCS design curve is based on the first yield criterion for a column with a sinusoidal
imperfection. The total stress is composed of one contribution from pure axial
compression and one from bending due to initial imperfection with amplification.
1 σ x Awio =
σx+ σY (2.42)
1− σ x W
σE
where σx is the axial stress, A is the cross-sectional area, wio is the initial imperfection, W
is the elastic modulus, and σY is the yield stress. This expression can also be formulated as
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 14 of 68
σx + 1 ⎛ σ x Awi 0 ⎞
⎜ ⎟ =1 (2.43)
σY 1− σ x λ 2 ⎝σY W ⎠
σY
Based on a large number of tests, numerical simulations and curve fitting, ECCS substituted the
term
A
wio = µ = α ( λ − λo ) (2.44)
W
where α and λo are the coefficients depending on the cross-sectional shape and the fabrication
method. They represent both the real imperfections and the residual stresses. α governs the
degree of reduction compared to the Euler buckling curve, and λo represents the cut-off limit by
which the critical stress is equal to the yield stress. The factors are determined such that the
initial stress shall fit the results of numerical simulation and experiments.
σ cr φ − φ 2 − λ 2
= r
σY λ2
where (2.45)
φ=
1
2
(
1 + α ( λ − λ0 ) + λ 2 )
The various design curves are shown in Figure 2.10. . They can be calculated numerically by
using the coefficients given in Table 2.2.
Remark:
By rearranging Equation (2.43), the equivalent imperfection can also be formulated as
wio 1 σ Y i ⎛⎜ λo ⎞⎟ (2.47)
= α 1− ⎟
l π E z o ⎜⎝ λ ⎠
where zo is the distance from the neutral axis to the extreme fibre of the cross-section and i is the
radius of gyration. For a thin-walled tube with σY = 250 MPa, the expression becomes
wi ⎛ λ ⎞
= 0.0015 ⎜ 1 − o ⎟ (2.48)
l ⎝ λ ⎠
Hence, the equivalent imperfection approaches asymptotically 0.0015 times the member length. This
coincides also with the typical tolerance requirements for member out-of-straightness given in the
codes.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 15 of 68
The classification of the various profiles is performed according to Figure 2.11. It is seen that
hot-formed tubes with circular or rectangular cross-section are assigned to column curve a, i.e.
the curve with the highest critical stress. Rolled I-profiles can be assigned to any curve
depending on the height/width ratio and the buckling axis. The differentiation between narrow
and wide flanges is due to the difference in the residual stresses.
For built up I-profiles, there is distinguished between rolled flanges and flame cut flanges. Flame
cut flanges get tensile residual stresses along plate edges which are beneficial with respect to
buckling about weak axis. Stress relief by heat treatment of I-profiles and box-section upgrades
the cross-sections to curve a for buckling about strong axis.
Table 2.2 Numerical Values of λo and α.
Curve λo α
a 0.2 0.20
b 0.2 0.35
c 0.2 0.50
d 0.2 0.65
e 0.6 0.35
e
1
Critical stress ratio σc / σY
0.8 a Euler
b
c
0.6
d
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3
2.5Capacity of Beam-Columns
Beam-columns are characterized by simultaneous action of axial force and lateral load. The
capacity under combined loading is often presented in the form of interaction diagrams.
For pure axial compression the critical load should be equal to the buckling force, and for pure
bending the bending moment should approach the plastic bending moment, Mp, or first yield
moment, MY, (provided that local buckling does not take place).
N M max
+ ≤1 (2.49)
N cr M Y or M p
where Ncr is the characteristic buckling strength from the column curve, and Mmax is the
maximum moment including amplification due to axial compression.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 17 of 68
N
Np
1
l/i=0
20
40
60
M
1 Mp
Figure 2.12 Interaction Diagram for Beam Columns.
N N
Mo Mo
For the beam-column shown in Figure 2.13, the exact amplification factor is given by
1
M max = M o (2.50)
π N
cos
2 NE
Using Taylor’s expansion, we can write
2 4 2
π N 1 ⎛π ⎞ N 1 ⎛π ⎞ ⎛ N ⎞
cos = 1− ⎜ ⎟ + ⎜ ⎟ ⎜
2 NE 2 ⎝ 2 ⎠ N E 24 ⎝ 2 ⎠ ⎝ N E ⎟⎠
(2.51)
N
≈1− [for N << N E ]
NE
implying that the expression for the maximum bending moment can be rewritten as
1
M max ≈ M o N
(2.52)
1−
NE
N Mo
+ ≤1 (2.53)
N cr ⎛ 1 − N ⎞
⎜ ⎟MY
⎝ NE ⎠
For other non-symmetric bending moments it is not straight-forward to determine the exact
location of the critical section. Instead, Mo is substituted by an equivalent moment, Meq = CMMo.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 18 of 68
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames
N CM M o Page 18 of 69
+ ≤1
N cr ⎛ N ⎞
N ⎜ 1C−M M o ⎟ M Y 1
⎝ NE ⎠ (2.54)
N cr N
1
⎛ E MY M 2 ⎞
C M ∈ N⎜ 0.6 − 0.4 , 0.4 ⎟ (2.54)
⎝ M1 ⎠
CM 0.6 0.4 M 2 , 0.4
where M 1 = M o , is the larger end moment. OnMstress
1
form,
where M 1 M o , is the larger end moment. On stress form,
σ x + C Mσ b W ≤1 (2.55)
σx cr (1C−Mσ bσWE ) σ Y Z (2.55)
1
cr 1 E Y Z
where W and Z are the elastic and plastic section modulus, respectively. The variation of CM is
shown in and
where W Figure 2.14,
Z are theand for and
elastic otherplastic
load conditions, they are
section modulus, given in the
respectively. Thecodes (e.g.of
variation AISC).
CM is
shown in Figure 2.14, and for other load conditions, they are given in the codes (e.g. AISC).
CM
1.0
0.4 M1 M2 M1 M2
N N
-1 0 1 M2/M1
Figure
Figure2.14
2.14Variation of CofM C
Variation Coefficients.
M Coefficients.
ItIt should
should also
also be
be checked
checkedthat
thatthe
thestate
stateofof
force does
force notnot
does violate the the
violate plastic interaction
plastic function
interaction function
for combined loading.
for combined loading.
AISC
AISC N M
N M 1 for I-profiles
Np +1.18M p ≤ 1 for I-profiles
Np 1.18M p
M N
M cos π N 0 for thin-walled tubes.
Mp − cos2 N p ≤ 0 for thin-walled tubes.
Mp 2 Np
where N p Y A , is the plastic axial force.
where N p = σ Y A , is the plastic axial force.
ECCS - NPD
A fictitious eccentricity, e* ( wio), is introduced. It represents the combined effect of
imperfections, residual stresses, etc. and is determined such that the extreme case of pure column
buckling is included, (i.e. N Ncr for Mo = 0). A first yield criterion is used.
*
N cr N cre 1 (2.56)
Np 1 N cr N E M Y
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 19 of 68
According to ISO 19902 /7/ brace members in offshore structures are checked with respect
to:
There is significant interaction between different failure modes. This interaction must be
accounted for. In addition, for deep water platforms, the resistance will be influenced by the
effect of hydrostatic pressure.
Short tubular members subjected to axial compression will fail either by material yielding or
local buckling, depending on the diameter-to-thickness (D/t) ratio. Tubular members with low
D/t ratios are generally not subject to local buckling under axial compression and can be
designed on the basis of material yielding, i.e., the local buckling stress may be considered
equal to the yield strength. However, as the D/t ratio increases, the elastic local buckling
strength decreases, and the tubular should be checked for local buckling.
The local elastic buckling strength for shells subjected to axial compression is
t
f xe = C x E (2.56)
r
in which t is wall thickness of the member and r is the radius (refer section. Theoretically Cx
should be equal to 0.6 (refer Chapter 5. However, shells are very sensitive to imperfections.
In order to comply with tolerance limits for fabrication, a reduced value of Cx = 0.3 is used.
In the elasto-plastic range the characteristic local buckling resistance is determined from:
f cl
= 1.0 λ ≤ 0.412
fy
f cl
= 1.047 − 0.274λ 2 0.412 ≤ λ ≤ 1.382 (2.57)
fy
f cl = f ce 1.382 ≤ λ
f xe
in which λ =
fy
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 20 of 68
A comparison between test data and the characteristic local buckling strength equation, is
shown in Figure 2.15. The developed equations have a bias of 1.065, the standard deviation of
0.073, and the coefficient of variation of 0.068.
The elastic local buckling stress represents one-half of the theoretical local buckling stress
computed using classical small-deflection theory. The reduction accounts for the detrimental
effect of geometric imperfections. Based on the test data the reduction is considered to be
conservative for cylinders with t ≥ 6 mm and D/t < 120. Offshore platform members typically
fall within these dimensional limits. For thinner cylinders and cylinders with higher D/t ratios,
larger imperfection reduction factors may be required. Reference is made to Section 5:
Buckling of Cylindrical Shells
Figure 2.16 shows that the local buckling effect starts form D/t > 60, but is very for practical
unstiffened tubular members.
1"
0.8"
0.6"
0.4"
0.2"
0"
0" 20" 40" 60" 80" 100" 120" 140"
Diameter/thickness"ra=o"
2.6.1.2 Bending
The behaviour of tubular members subjected to bending can be grouped into three categories,
refer Figure 2.17;
i. High rotational capacity; the bending moment may attain the fully plastic bending
moment and maintain this level significant rotation. The failure mode is ductile,
exhibiting very gradual decay of the resistance - A2(24)
ii. Intermediate rotational capacity; the bending moment may still attain the fully plastic
bending moment, but cannot maintain this level for significant rotation. The failure
mode is semi-ductile failure mode, exhibiting gradual strength decay - A5(48).
iii. Low rotational capacity; the maximum bending moment may be no more than the
yield moment or even less. Little post-ultimate ductility exists, the resistance exhibits
a rapid decay - A8(96).
Figure 2.17 Typical normalized moment-rotation curves for cylinders for various D/t-ratios
(MPS is plastic bending moment)
The governing parameter for the bending ultimate bending moment resistance is the
“slenderness” parameter:
fy
α= = λ eq
2
(2.58)
t
E
D
The resistance is defined by the following set of equations for the three categories:
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 22 of 68
fm Z fyD
= ≤ 0.0517 (2.59)
fy W Et
fm ⎛ ⎛ f D ⎞⎞ ⎛ Z ⎞ fyD
= ⎜ 1.13 − 2.58 ⎜ y ⎟ ⎟ ⎜ ⎟ 0.0517 < ≤ 0.1034 (2.60)
fy ⎝ ⎝ Et ⎠ ⎠ ⎝ W ⎠ Et
fm ⎛ ⎛ f D ⎞⎞ ⎛ Z ⎞ fyD fy
= ⎜ 0.94 − 0.76 ⎜ y ⎟ ⎟ ⎜ ⎟ 0.1034 < ≤ 120 (2.61)
fy ⎝ ⎝ Et ⎠ ⎠ ⎝ W ⎠ Et E
where
W = elastic section modulus
π ⎡⎣ D − ( D − 2t ) ⎤⎦
4 4
=
32 D
Z = plastic section modulus
1 3
= ⎡⎣ D − ( D − 2t )3 ⎤⎦
6
fy
For > 0.170 the tubular is a class 4 cross section and may behave as a shell, refer Section
f cle
5.
Note that the allowable bending stress may exceed the yield strength. This is because the
stresses are determined by elastic analysis and elastic section modulus. As plastic bending
may be accepted, it is necessary to use an equivalent bending stress which exceeds the yield
stress.
The allowable bending stress according to ISO19902/Norsok N-004 is plotted in Figure 2.18
versus the slenderness α, along with results from experiments. The design curve represents a
lower fractile in the distribution of the test results. On the average the tests results are 10%
higher than the design curve. The bending stress corresponding to first yield and fully plastic
bending moment for a thin-walled section are indicated. Figure 2.20 shows the same curve
versus D/t ratio for a yield stress of 355 MPa.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 23 of 68
Plastic thin-walled
Yield
Figure 2.18 Normalised allowable bending stress for cylinders - comparison of test data and
design equations (from ISO19902)
1.6"
1.4"
Allowable(bending(stress(fm/fY%
1.2"
1"
0.8"
0.6"
0.4"
0.2"
0"
0" 20" 40" 60" 80" 100" 120"
Diameter/thickness(ra6o(D/t(
Figure 2.19 Normalised allowable bending stress for cylinders versus D/t-ratio fy = 355 MPa
pr
σx = (2.62)
2t
pr
σh = (2.63)
t
Ring stiffeners affect the ring stresses at the position of the rings. If distance between the
rings is smaller than a certain value, the ring stress midway between the rings is also affected.
Otherwise it may be taken equal to the stress for an un-stiffened tube. Reference is made to
section 5.
The longitudinal stress does not cause contribute to elastic buckling. This is explained as once
the tubular should deform laterally, the increase/decrease of wall surface will cause the
hydrostatic pressure to restore the tube. However, both stress components will affect elasto-
plastic column buckling through the material part.
Figure 2.20 Stresses in closed, unstiffened circular cross-sections for external hydrostatic
pressure
The compressive stresses make the tubular susceptible to elastic or inelastic local buckling of
the shell wall between the restraints. Once initiated, the collapse will tend to flatten the
member from one end to the other.
Ring-stiffened members are subject to local buckling of the shell wall between rings, while
the rings remain essentially circular. However, the rings may rotate or warp out of their plane.
Ring-stiffened tubular members are also subject to general instability, which occurs when the
rings and shell wall buckle simultaneously at the critical load. It is desirable to provide rings
with sufficient residual strength to prevent general instability. Reference is made to Section 5,
Buckling of Cylindrical Shells.
As shown in Section 5.4.4 long cylinders fail by ovalisation with with 2 waves in the
circumferential direction, i.e. n = 2. The elastic hoop buckling strength, fhe, is given by
2
⎛t⎞
f he = 0.275E ⎜ ⎟ (2.64)
⎝r⎠
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 25 of 68
⎛ t⎞
f he = 2Ch E ⎜ ⎟ (2.65)
⎝ D⎠
where Ch = 0.55t/D. In order to account for possible out-of roundness up to 1 percent, the
elastic hoop buckling stress is taken as 0.8 of the theoretical value from classical small
deflection theory. This gives, Ch=0.44t/D.
The out-of roundness in % is defined as
D max − D min
(2.66)
0.01 D nom
where Dmax and Dmin are the maximum and minimum of any measured diameter at a cross
section and Dnom the nominal diameter.
For shorter cylinders the Ch value increases. In the ISO code Ch is expressed as functions of
the parameter µ, defined as
L D
µ= = 0.72 Z Z = 1.9µ2 (2.67)
D t
where Z is the Batdorf parameter, refer Section 5, and L is length of tubular between
stiffening rings, diaphragms, or end connections
Ch = 0.44 t/D for µ ≥ 1.6D/t
= 0.44 t/D + 0.21 (D/t)3/µ4 for 0.825D/t ≤ µ < 1.6D/t
= 0.737/(µ - 0.579) for 1.5 ≤ µ < 0.825D/t
= 0.80 for µ < 1.5
For short cylinders, the buckling coefficient is equal to 80% of the buckling coefficient of a
plane, wide plate, i.e 0.80 x 1= 0.80.
The elastic buckling stress is modified for plasticity as follows:
fh
=1 λ < 0.64
fy
f h 0.7
= 0.64 < λ < 1.35 (2.68)
f y λ 0.8
f h = f he 1.35 < λ
fy
where λ =
f he
For ring-stiffened members, Eq. 2.82 gives the hoop buckling strength of the shell wall
between the rings.
The design curve for hoop buckling strength is compared with test results in Figure 2.21. The
bias is 14% with a coefficient of variation of 0.12.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 26 of 68
Figure 2.21 Hoop buckling strength of fabricated cylinders subjected to hydrostatic pressure-
test data compared with design equation (from ISO19902)
Hoop+buckling+stress+versus+diameter6thickness+ra7o+
1"
Normalised+hopp+buckling+stress+
Elas-c"buckling"stress"f_he/f_Y"
0.8"
Cri-cal"buckling"stress"f_h/f_Y"
0.6"
0.4"
0.2"
0"
0" 20" 40" 60" 80" 100" 120" 140"
Diameter6thickness+ra7o+
Dmax − Dmin
= 1 − 0.2 = 1 − 0.2 OOR
0.01 Dnom
The characteristic axial compressive resistance for tubular member is determined on the basis
of Johnson-Ostenfeldt expression. For large slendernesses the critical stress is equal to the
Euler buckling stress, except for a 10% reduction
fc 0.9
= for λ > 1.34 (2.70)
fy λ 2
where
fy π 2 EI
λ= , fE =
(kl ) A
2
fE
k is the effective buckling length factor.
For stockier members modification for plasticity is carried out using the conventional
Johnson-Ostenfeldt correction with a 10% increase:
fc
= 1.0 − 0.28λ 2 for λ ≤ 1.34 (2.71)
fy
Hydrostatic pressure causes a hoop stress σh= pr/t and axial stress σx = pr/2t in closed
cylinders. This presence of these stresses reduces the stress available for elasto-plastic column
buckling. The column buckling stress, fch , which accounts for the effect of hydrostatic
pressure is calculated through the following interaction equation:
The following interaction is adopted:
2 −1
⎛ f ch ⎞ ⎛ f c 2σ x ⎞ f ch σ x ⎛ σ x ⎞ ⎡ 2σ ⎤
⎜ ⎟ −⎜ − ⎟ + ⎜ − 1⎟ = 0, λ < 1.34 ⎢(1 − x )⎥ (2.72)
⎝ f cl ⎠ ⎝ f cl f cl ⎠ f cl f cl ⎝ f cl ⎠ ⎣ f cl ⎦
where fcl is the equivalent yield stress accounting for local buckling and fc is the basic column
buckling strength .
Solved with respect to fch this becomes:
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 28 of 68
−1
f ch 1 2σ σ ⎡ 2σ ⎤
= [ξ − x + ξ 2 + 1.12 λ 2 x ], λ < 1.34 ⎢(1 − x )⎥ (2.73)
f cl 2 f cl f cl ⎣ f cl ⎦
where ξ = 1 − 0.28λ 2
For reduced slendernesses exceeding the above value there is no reduction due to hydrostatic
pressure.
The interaction relationships are plotted in Figure 2.23 for various column slendernesses.
The figure should be interpreted as follows:
The values on the ordinate show the column buckling strength versus reduced
slenderness ration with no hydrostatic pressure. The values are normalised with the
strength against local buckling, which for large diameter/thickness-ratios may be lower
than the yield stress.
For stocky columns the reduction is linearly dependent on the capped end axial stress,
which is also normalised against the local buckling strength. For slender columns, the
buckling strength is almost constant; it is in the elastic domain except at very large water
depth, hence the reduction in effective yield strength due to the capped end stresses has
no effect.
In the same diagram the water depth corresponding to the normalised capped end
stress is shown for various D/t-ratios for a yield stress of 355 MPa. In practice it is the
“left” part of the interaction diagram that is activated, because hoop buckling limits the
use of high D/t –ratios for large water depths (refer Section 2.6.2.3). The approximate
possible range of the various D/t-ratios is shown in the diagram. It is indicated in the
diagram how the reduction in buckling strength can be read for various reduced
slenderness ratios for D/t = 30 for a water depth of 300 m.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 29 of 68
lamda=1.5
0.5 250
D/t = 30
0.4 D/t = 40 200
D/t = 50
0.3 150
D/t = 60
0.2 100
0.1 50
0 0
0 0.2 0.4 0.6 0.8 1
Sigma_x/f_cl
Figure 2.23 Interaction between hydrostatic pressure and column buckling. Water depth
calculated for fy = 355 MPa
Hydrostatic pressure causes a hoop stress σh = pr/t. This presence of this stress reduces the
stress available for elasto-plastic bending. The bending stress, fmh , which accounts for the
effect of hydrostatic pressure is calculated through the following interaction equation:
2 2η
⎛ f mh ⎞ ⎛ f mh ⎞ ⎛ σh ⎞ ⎛ σh ⎞
⎜ ⎟ + 2v ⎜ ⎟⎜ ⎟ + ⎜ ⎟ = 0
f
⎝ m ⎠ ⎝ m ⎠ ⎝ fh ⎠ ⎝ fh ⎠
f
(2.74)
fh
η = 5−4
fy
where fh is the hoop buckling stress and fm is the critical bending stress with no hydrostatic
pressure. Solved for fmh this becomes
2 2η
f mh ⎛σ ⎞ ⎛σ ⎞ ⎛σ ⎞
= 1 + 0.09 ⎜ h ⎟ − ⎜ h ⎟ − 0.3 ⎜ h ⎟ (2.75)
fm ⎝ fh ⎠ ⎝ fh ⎠ ⎝ fh ⎠
The interaction is plotted in Figure 2.24. The plots show that the interaction is weak when the
hoop buckling resistance is small, but increases with increasing resistance fh/fy. For fh/fy =1 the
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 30 of 68
interaction is close to the von Mises criterion for combined compression and tension, which is
natural as tensile side in bending will be the critical location..
Figure 2.25 shows that the reduction in bending strength starts to become significant for
typical D/t-ratios < 45) when the water depth is larger than 100 m. For large water depths, the
rather thick-walled tubulars must be used.
1.2
f_h/f_y =0.2
1 f_h/f_y =0.4
f_h/f_y =0.6
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1 1.2
Sigma_x/f_h
Figure 2.24 Interaction between local bending strength and hydrostatic pressure
300
250
Waterdepth (m)
200
10 % 100 %
150
5%
100
2%
50
0
20 30 40 50 60 70 80 90 100
Diameter/thickness
Figure 2.25 Reduction in bending strength as a function of water depth and D/t-ratio
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 31 of 68
2.6.3 Failure of beam column accounting for local buckling and hydrostatic pressure
Tubular members will have to be checked as beam columns both in compression and tension.
In addition local buckling/yielding failure have to be considered. The equations in the
following include the effect of hydrostatic pressure. If hydrostatic pressure is not included,
then the column buckling strength, fc, and bending strength, fm, with no effect of hydrostatic
pressure should be used.
For compressive members three checks must be performed:
i. Beam column failure (Perry-Robertson type formulation)
0.5
⎡⎛ 2
⎞ ⎛ ⎞ ⎤
2
⎢ ⎥
⎜ ⎟
σa 1 ⎢⎜ Cmyσ my ⎟ ⎜ Cmzσ mz ⎟ ⎥
+ ⎢ +⎜ ⎟ = 1.0 (2.76)
f ch f mh ⎢⎜ σa ⎟ ⎜ σ a ⎟ ⎥⎥
1− 1−
⎢⎜⎝ f Ey ⎟⎠ ⎜⎝ f Ez ⎟⎠ ⎥
⎣ ⎦
The first term represents the utilisation with respect to column buckling. σa is the net axial
stress (capped end stress σx = pr/2t is not included). The effect of the capped end stress is
included through interaction in the resistance to column buckling, fch. The second term
represents the utilisation with respect to bending. The bending term includes the
magnification factor. The allowable bending stress includes the effect of hoop stress caused
by hydrostatic pressure.
σa + σ x σ my
2
+ σ mz
2
+ = 1.0 (2.77)
f cl f mh
The local buckling check is based on linear interaction between axial compression and
bending. The denominator in the first term is the local buckling resistance without hydrostatic
pressure. The effect of hydrostatic pressure is accounted for in the load effect (nominator)
term. Note that the bending stress shall not be magnified in this check.
For relatively thin-walled members (and thus only relevant intermediate water depths),
interaction between elastic local buckling in axial direction and hoop direction must be carried
out:
σ max − 0.5 f he σ h 2
+ ( ) = 1.0
f xe − 0.5 f he f he
where (2.78)
σ max = σ a + σ x + σ m
is the maximum compressive stress in the member
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 32 of 68
σa −σx σ my
2
+ σ mz
2
+ = 1.0 (2.79)
fth f mh
Here the capped end axial stress must subtracted from the tensile axial stress.
The effective yield stress is reduced in the same manner as the maximum bending stress because
of hydrostatic pressure (refer Section 2.6.2.3):
2 2
fth ⎛σ ⎞ ⎛σ ⎞ ⎛σ ⎞
= 1 + 0.09 ⎜ h ⎟ − ⎜ h ⎟ − 0.3 ⎜ h ⎟ (2.80)
fy ⎝ fh ⎠ ⎝ fh ⎠ ⎝ fh ⎠
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 33 of 68
So far, buckling of individual beam-columns with idealized boundary conditions have been
considered. In practice, however, a given component is an integral part of a complete frame
system. The interaction formulas developed for individual beam-columns are also used for
braces in frames. The system analysis is reduced to a determination of the effective buckling
lengths of the individual components.
One method to determine the effective buckling length is to use alignment charts. For offshore
structures the method is mostly relevant for laterally unbraced members. The use of alignment
charts is based on the following assumptions:-
• elastic behaviour,
• all columns buckle simultaneously,
N
• the relative utilization of all members, , is equal.
π EI / 2
2
Within the system, the individual components are distinguished between the compression
members (columns) and the bending members providing restraint, and denoted, respectively, by
the subscripts c and b. Then, the following factor, which measures the relative stiffness between
the compression and the restraining members, is defined.
∑ lI c
G= c
(2.81)
∑α lI bb
where l is the member length, and I is the moment of inertia. α is a rotational factor depending
on the boundary constraint of the restraining member at far end. It is distinguished between
sidesway prevented case and sidesway permitted case, (see Figure 2.28).
The factors are easily derived from simple beam theory accounting for actual boundary
conditions .
Beam bending
Joint flexibility
Node
Cj
Cb
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 34 of 68
Depending upon the geometry of the joint, all restraining beams may not be considered as
perfectly rigid. Accordingly the stiffness of these members will be reduced. The stiffness of
the resisting beams, Cb, and the joint, Cj, can be considered as two springs in series, (see Figure
2.26). Hence, the flexibilities can be added so that
1 1 1
*
= + (2.82)
C Cb C j
yielding,
1
C* = C b = Cb ⋅ β (2.83)
1+ C b
Cj
The β-factor, which represents the softening of the beam stiffness due to joint flexibility, is
introduced into the G-factor,
∑ lI c
G= c
(2.84)
∑αβ lI bb
and the factor Cb is defined as follows
⎧ EI b
⎪⎪ 2α l : base case for sidesway prevented
Cb = ⎨ (2.85)
⎪6α EI b : base case for sidesway permitted.
⎪⎩ l
The joint rotational stiffness Cj is given by parametric equations. For a T-joint subjected to in-
plane bending the stiffness is calculated according to the following formula:
r
2.35−1.5
⎛T ⎞ R
C j = 0.43ERH 3 ⎜ − 0.01⎟ (2.86)
⎝R ⎠
buckling control
For other joints it is conservative to use the above relations to assess local shell stiffness for
buckling control
N N N
k = 2EI/l k = 4EI/l
N N N
k = 6EI/l k = 4EI/l
The determination of the effective buckling length is based upon the G-factor calculated at each
end of the considered member. Then a straight line is drawn between the values on the g-factor
axis. Then the effective buckling length factor can be read as the intersection of this line with the
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 36 of 68
K-axis. In the example shown in Figure 2.29, GA=5 and GB = 0.5 yields K=0.8 in the sidesawy
prevented case.
Another way to determine the effective buckling length is to first derive the stiffness matrix
for a beam including p-δ (geometric effect) effects on element level.
Consider the beam shown in Figure 2.30. The internal moment at any section, X = x, and the
external moment acting on the beam are given, respectively, as
M i = −EIw, xx
(2.89)
M e = M A + QA X − N ( w − wA )
Mi + Me = 0
(2.90)
⇒EIw, xx − M A − Q A X + N (w − wA) = 0
1
w, xx + k 2 w = (M A + Q A x + NwA) (2.91)
EI
N
where, k 2 = .
EI
w
QB
QA MB
MA
θB
j N
N
j θ
A
w(x) wB
wA
X
X=x
Figure 2.30 Beam Element With End Forces.
The total solution to the differential Equation (2.87) is given by the sum of the homogeneous
and the particular solutions, which can be written as,
w( x ) = wh + w p
1 (2.92)
= C1 sin kx + C 2 cos kx + (M A + Q A x + NwA)
N
By means of these boundary conditions together with the equilibrium conditions, the
following relationships can be established: -
6EI 4EI 2EI
M A = 2 φ 2 (wB − w A ) + 3 φ 3 θ A + φ θB
4
6EI EI 4EI
M B = 2 φ 2 (wB − w A ) + 2 φ 4 θ A + φ θB (2.94)
3
12EI 6EI 6EI
Q A = − Q B = 3 φ 5 (w A − wB ) − 2 φ 2 θ A − 2 φ 2 θ B
which can be written in a matrix form as,
Compression Tension
β β ⎫ π
,N E = π
2
φ1 = φ1 = N EI
tan β ⎬ β=
tanh β ⎭ 2 NE l
2
1 β 2
1 β 2
φ2 = φ2 =
3 (1 − φ 1) 3 (φ 1 − 1)
1 3 1 3
φ 3 = φ1 + φ 2 φ 3 = φ1 + φ 2
4 4 4 4
1 3 1 3
φ 4 = − φ1 + φ 2 φ4 = − φ1 + φ 2
2 2 2 2
φ 5 = φ1 φ 2 φ 5 = φ1φ 2
It is seen that each term in the matrix is identical to the linear term (for pure bending) except
for the φ-functions, which contain the influence of axial force. These functions are expressed
in terms of the axial force normalized versus the Euler buckling force. The Euler buckling
force is a characteristic force independent of the actual boundary conditions for the element.
The φ-functions are sketched in the diagram shown in Figure 2.31.
For zero axial force they all become equal to unity, as they should. When the axial force
increases, all stability functions decrease, except for φ4. For certain values of φ they change
sign. This represents the potential instability points for the element.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 39 of 68
For negative axial force (tension), the stability functions increase. Therefore, the element
becomes stiffer when it is subjected to tension.
For certain values of the axial force the displacements may increase infinitely with no change
in the external forces (end moments and shear forces), i.e.
ΔS =KV= 0 (2.96)
In that case, the nontrivial solution is given by K = 0 , which means that the determinant of
the stiffness matrix, K, should be equal to zero.
The critical load for several well-known cases may be derived by considering subsets of the
matrix given in Equation (2.95).
4 ! !
1
! !
! -value
0
-5 -4 -3 -2 -1 0 1 2 3 4 5
-1
! !
-2
! !
-3
-4
! !
-5
-6
Axial force ! ! ! !! !! E
Example 1.
N A B N
Consider a simply supported beam shown in
l Figure 2.32. The boundary conditions are
wA = wB = 0 . Then,
Figure 2.32 A simply Supported Beam.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 40 of 68
2EI ⎡2 φ 3 φ 4 ⎤ 1
K= K = 0,⇒ 4 φ 32 − φ 42 = 0;∴φ3 = φ4
⎢⎣ φ 4 2 φ 3⎥⎦ 2
Example 2.
N A B
wA = wB = θ B = 0
l
Figure 2.33
4EI
K= φ K = 0,⇒ ForN = 2 N E
3
1
This means that k = = 0.7 .
2
Example 3.
N
A B
wB = θ B = 0
l
Figure 2.34
2EI ⎡ 6 φ 5 − 3 φ 2⎤
K= 3 ⎢ ⎥ K = 0⇒ 12l 2φ3φ5 − 9l 2φ22 = 0
⎣- 3 φ 2 2 φ 3 ⎦
2
Example 4.
N
A B
wB = θ A = θ B = 0
l
Figure 2.35
12EI
K= 3
φ5 K = 0⇒ φ5 = 0ForN = N E
From the above examples, it is seen that the stiffness matrix contains all the necessary
information to predict the critical load, with respect to buckling, depending on the boundary
conditions.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 41 of 68
Example 5.
Buckling of a Member with Rotational Springs.
Consider the beam shown in Figure 2.36 which is subjected to end compression. It is assumed that
the adjacent structure restrains the beam against rotation. This restraint is represented by the
rotational spring with stiffness, c
c c EI
N EI N c =η
θA θB = -
A θA B
4EI 2EI
MA=
i
φ 3θ A + φ θB
4
4EI ⎛ 1 ⎞
= ⎜ φ 3 − φ 4 ⎟θ A
⎝ 2 ⎠
Further,
1 1
φ 3 − φ 4 = φ1
2 2
Hence,
EI
MA=2
i
φ θA
1
The spring moment,
EI
M A = Cθ A = η θA
e
Equilibrium yields,
2EI ⎛ 1 ⎞
M = M Ai + M Ae ⇒M = ⎜ φ 1 + η ⎟θ A
⎝ 2 ⎠
The stiffness becomes singular for,
1
f = φ1 + η = 0 or η = −2φ1
2
This indicates buckling. The solution for different values of η is shown in Figure 2.37.
The solution to the buckling problem for various types of flexible boundary conditions have
been prepared in diagrams as shown in Figure 2.37. The solution to the problem considered
above can be found in the lower left diagram.
It is seen that a rotational spring with stiffness equal to the bending stiffness of the beam
raises the critical load by 37 %.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 42 of 68
φ1
+1
0
1.37 1.67 2.12 2.44 ρ = P PE
η=1 1 2 3
β =π ρ 2
η=2
-1
η=4
-2
-3 η=6
In general, the solution of the buckling problem for a beam leads to a non-linear equation.
This must be solved by iterative procedures. For structural problems, a combined incremental-
iterative procedure based on the Newton-Raphson method is often used. The Newton-
Raphson method solves the equation,
f (x ) = 0 (2.97)
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 43 of 68
f( xn )
xn+1 = xn − (2.98)
f ' ( xn )
where,
df
f ' (x) = (2.99)
dx
β
f = 2 φ1 + η = 2 +η = 0
tan β
(2.100)
df ⎛ β ⎞
= 2⎜⎜ cot β − 2 ⎟⎟
dβ ⎝ sin β ⎠
Table 2.3
2
N ⎛ 2 ⎞
ρ= = ⎜⎜ ⎟⎟ = 2.12 (2.101)
N E ⎝ πβ ⎠
Example 6
Buckling of X-braces.
X-braces are commonly used in jackets and jack-up platforms. The braces carry most of the
lateral force on the platform. For a platform with single X-braces as the one shown in Figure
2.39 the lateral force caused by waves, current and wind will be carried by compression in
one of the braces and tension in the other. These forces are of equal magnitude. In addition
there will be a contribution in compression due to compression of the legs from the topside
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 44 of 68
weight. Consequently the total force in the compressive brace will be larger than the force in
the tension brace. If the environmental forces are assumed to come in from the left in Figure
2.39 force N1 will be in compression and force N2 will be in tension.
Waves
N1 N2
wind
current
N2 N1
Out-of-plane buckling is critical. There is interaction between the two braces because the
tension brace acts as a support for the compression brace. With reference to Figure 2.40 this
support is modelled as an equivalent spring with stiffness KN2. In addition, the legs and any
other braces framing into the ends of the X-braces provide a rotational restraint, which is
modelled by a rotational spring with stiffness C. For simplicity it is assumed equal at all four
supports.
Remark:
For out-of-plane bending the rotational resistance of the adjacent members will consist
of one contribution from bending and one contribution from torsion. The total stiffness is
found by vectorial summation of these stiffnesses. For a thin-walled tube the stiffness in
torsion and bending may be taken as
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 45 of 68
aEI αEπr 3t
Cbending = =
l l
EI t E ⋅ 2πr 3t 1
Ctorsion = = = C
l 2(1 + ν )l α (1 + ν ) bending
The compression brace may fail in two modes as shown in Figure 2.40:
• symmetric buckling
• symmetric buckling.
If the tension brace is not sufficiently strong the symmetric buckling mode governs. At some
limiting value Klim, the tension brace resists lateral deflection and the buckling mode changes
into the asymmetric mode.
C θA wB C
N2 N2
A B
/2
The effective spring stiffness of the tension brace can be determined by means of the model
shown in Figure 2.41. For symmetry reasons only one half of the system needs to be
analyzed.
Using Equation (2.95) the stiffness of the left half of the tension brace system can be written
as
⎡ EI EI ⎤
⎢ 4 φ3 (N 2 ) + C 6 φ (N 2 ) ⎥
⎡M A ⎤ (l 2 ) (l 2)2 2
⎡θ A ⎤ ⎡0 ⎤
⎢Q ⎥ = ⎢ ⎥⎢ ⎥ = ⎢ ⎥ (2.102)
⎣ B ⎦ ⎢ 6 EI φ2 (N 2 ) 12 EI φ5 (N 2 )⎥ ⎣ wB ⎦ ⎣Q N 2 ⎦
⎢⎣ (l 2 )2 (l 2)3 ⎥⎦
Using the first equation the rotation at end A can be eliminated so that the following
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 46 of 68
expression is obtained for the contact force between the two braces
⎛ ⎞
⎜ ⎟
EI ⎜ 3 1
QN 2 = 96 3 φ5 (N 2 ) − φ2 (N 2 ) ⎟w = K w
2
(2.103)
l ⎜ C ⎟ B N2 B
4 φ ( N ) +
⎜ 3 2
8EI l ⎟⎠
⎝
The expression in the bracket represents the stiffness of the tension brace as felt by the
compression brace. It is seen that for N1 = 0 it condenses into the well-known relationship for
EI
a beam with a concentrated load at mid-span : 2QN2 =0 = 48 3 wB .
l
Symmetric buckling:
Symmetric mode
C θA C
wB N1
N1
/2 KN2
The governing equations for the left half of the column read:
⎡ EI EI ⎤
⎢ 4 φ3 (N1 ) + C 6 φ (N 1 ) ⎥
⎡M A ⎤ (l 2) (l 2)2 2
⎡θ A ⎤
⎢Q ⎥ = ⎢ ⎥ ⎢ ⎥ = Kv (2.104)
⎣ B ⎦ ⎢ 6 EI φ2 (N1 ) 12 EI φ5 (N 2 ) + K N ⎥ ⎣ wB ⎦
⎢⎣ (l 2 )2 (l 2)3 2
⎥⎦
The critical load is obtained when the determinant becomes singular or:
2
⎛ EI ⎞⎛ EI ⎞ ⎛ EI ⎞
⎜⎜ 4 φ3 (N1 ) + C ⎟⎟⎜⎜12 φ (N1 ) + K N2 ⎟⎟ − 36⎜⎜ ⎟⎟ φ2 (N1 )2 = 0 (2.105)
⎝ (l 2) ⎠⎝ (l 2)
3 5
⎠ ⎝l 2⎠
⇓
⎛ ⎞
⎜ ⎟ (2.106)
⎛ C ⎞⎜ 3 1
4⎜⎜ φ3 (N 1 ) + ⎟⎟ φ5 (N 1 ) + φ5 (N 2 ) − φ2 (N 2 ) ⎟ − 3φ 2 = 0
2
8 EI l ⎠⎜ C ⎟ 2
⎝ 4 φ ( N ) +
⎜ 3 2
8 EI l ⎟⎠
⎝
Asymmetric buckling:
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 47 of 68
C C
N1 θA θB N1
l/2
In this case the right end of the half beam is pinned so that the critical load is obtained when
EI EI
4 φ3 (N1 ) + C 2 φ4 (N1 )
l 2 l 2
det =0 (2.107)
EI EI
2 φ4 (N1 ) 4 φ3 (N1 )
l 2 l 2
or
⎛ ⎞
⎜ ⎟
C ⎟
4 φ3 (N1 ) +
⎜ φ3 (N1 ) − φ4 (N1 )2 = 0 (2.108)
⎜ EI ⎟
⎜ 8
⎝ l 2 ⎟⎠
This is the same problem studied by means of G-functions for non-sway mode. It corresponds
to a case with zero rotational stiffness at right end.
The critical loads as given by Equations (2.106) and (2.108) may conveniently be solved by
means of an iterative procedure, e.g. bi-section method. The solutions for symmetric and
asymmetric buckling are shown in Figure 2.44 as a function of the non-dimensional rotation
C
stiffness ρ = . The symmetric mode depends upon the axial force in the tension brace, so
EI l
that N2/N1 is introduced as a parameter (N2 is positive in compression, negative in tension).
The left axis with ρ = 0 corresponds to pinned ends, the right axis with 100/ρ = 0 corresponds
to fixed ends.
Some of the calculated results are easily recognised. For pinned ends and N2/N1= 1 both
braces are in compression and there is no stiffening effect from N2. The curve applies to a
single brace, hence K = 1 for pinned conditions and K = 0.5 for fixed ends. For N2/N1 = 0 the
bending energy is doubled, hence the effective length is reduced by a factor of 1 2 to
K =1 2 ≈ 0.71 for pinned ends to K = 0.5 2 ≈ 0.35 for fixed ends.
By increasing tensile force, N2, the stiffness of the tension brace increases due to the rapid
increase of the stability function φ5 in tension, confer Figure 2.31. Consequently, the effective
length for the compression brace is reduced. However, the decrease in the effective length for
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 48 of 68
symmetric buckling is limited by the onset of asymmetric buckling, when the tension brace
becomes too strong to be displaced. While symmetric buckling governs regardless of the
rotation stiffness when N2 = 0, asymmetric buckling is always the governing if N2 < -0.6
1.2
Symmetric mode
N2/N1 =
1 1.0
0.5
0.8
0.2
0
K-factor
-0.2
0.6
-0.5
0.4 -1.0
Asymmetric mode
0.2
Figure 2.44 Buckling factor versus rotational end restraint for X-braces
For typical X-braces the force distribution is such that the asymmetric buckling mode will be
the governing failure mode. Depending upon the rotational restraint the effective K-factor
should therefor be in the range of 0.5 to 0.35, or 1.0 to 0.7 if the longer segment of X-brace
(here:half brace) is used. Based on results from experimental investigations a factor of 0.8 is
supported.
For braces without lateral support, for example compression braces in K-braced frames,
experimental observations advocate a factor of 0.7.
The effective length factors, K, and moment reduction factors, CM, for jackets members
recommended by API (also proposed as ISO-standard) in lieu of rational analysis are listed in
Table 2.4.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 50 of 68
Table 2.4. Effective length and moment reduction factors for jacket members according to
API regulations.
Structural element K CM
Superstructure Legs
Braced 1.0 0.85
Portal (unbraced) K 0.85
Jacket Braces
Primary Diagonals and Horizontals 0.7 (b) or (c)
K-braces 0.7 (c)
X-braces (longer segment length) 0.8 (c)
2.9Torsional Buckling
Previously, flexural (column) buckling has been considered. However, for short bars, with
open thin-walled sections, where the torsional rigidity is low, failure may occur as pure
twisting or by a combination of bending and twisting. This depends on the location of the
shear centre (S.C) relative to the centre of gravity (C.G). If SC and CG do not coincide, a
combined flexural-torsional buckling will occur, (see Figure 2.45).
SC
CG (a) Torsional
Buckling.
SC
CG
(b) Flexural-torsional
Buckling.
dφ
Tφ = GI t (2.109)
dx
where G is the shear modulus, It is the torsional moment of inertia, and φ is the torsion
angle. All cross-sections free to warp, i.e. no stress due to restrained warping.
• Closed sections
T
T
t
4 F2
It = (2.110)
ds
∫t
where F is the enclosed area. For a tubular section, the torsional moment of inertia will be
4 ⋅ (π r 2 )
2
I t = 2πr =2πr t
3
(2.111)
• Open sections
The torsional moment of inertia for open sections is given by,
1
It =
3
∑ bi ti3 (2.112)
where bi and ti are the width and thickness of the subpoints of which the section is
composed.
bi
ti
T
T
z T y, v x
h/2
x
Q
M
A h/2
Q
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 53 of 68
Section A-A
The beam shown in Figure 2.48 is subjected to a torsional moment, T, at the right end and
is restrained against torsion at the left end. It is seen that the torsion of the profile induces
a bending moment in the flanges. Considering the top flange, this bending moment is
given by,
M f = − EI f v ,xx (2.113)
where,
1
If =
3
t f bf (2.114)
12
is the in-plane bending stiffness of the top flange. With the sign convention used, the shear
force can be written as,
dM
Qf = − = EI f v ,xxx (2.115)
dx
Now, with reference to Figure 2.48, the lateral deflection of the top flange, v, is related to
the torsional angle, φ, for small rotations through
h
v=− φ (2.116)
2
Hence,
h
Qf = − EI f φ ,xxx (2.117)
2
The resisting moment on the beam, from restrained warping, to the right of section A-A is,
2
h
T w = −h Q f = EI f φ ,xxx (2.118)
2
The resisting moment from Saint-Venant torsion is,
T φ = −GI tφ , x (2.119)
T + T w + Tφ = 0
(2.120)
⇒ T = GI tφ , x − EI wφ , xxx
2
h
Iw = If (2.121)
2
The solution to Equation (2.120) is presented in the lecture notes “Statikk for Marine
Konstruksjon”.
z' P
z P(y, z)
CG CG
zo y'
SC
y w φ
SC
v
yo
Figure 2.49 Torsional Buckling.
l l l l
1 1 1 1
U = ∫ EI y w,2xx dx + ∫ EI z v,2xx dx + ∫ GI tφ ,2x dx − ∫ EI wφ , xxx φ , x dx (2.122)
20 20 20 20
where the last term represents the strain energy in warping. Using integration by parts, we can
write
l l l
− ∫ φ ,xxx φ ,x dx = −[φ ,xx φ , ] + ∫ φ ,xx φ ,xx dx = ∫ φ ,2xx dx
l
x 0
(2.123)
0 0 0
on the condition that either φ,xx or φ,x vanish at the integration boundaries. This is usually the
case for frequently occurring support conditions.
To establish the potential energy of the external load, it is necessary to consider the potential
of each fibre in the cross-section. The deformation components of a point P(y, z) are given by,
v(y, z) = v − zφ
(2.124)
w(y, z) = w + yφ
where v and w are the displacement components of the shear centre, and φ is the rotation
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 55 of 68
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. angle.
Buckling of Bars
Then, and Frames
the potential energy of the external load is, Page 56 of 69
where yo and zo are the coordinates of the centre of gravity, CG, and Ip is the polar moment of
where yo and zo are Therefore,
inertia. the coordinates of the centre
the potential of gravity,
energy CG, and Iload
of the external p is the
canpolar momentas,of
be written
inertia. Therefore, the potential energy of the external load can be written as,
H
N
[
H =N− ∫2 v ,2x +2 w ,22x + i22p φ ,2x − 2 zo v ,x φ ,x + 2 yo w ,x φ ,x dx
2v ,x w ,x i p ,x 2 zo v ,x ,x 2 yo w ,x ,x dx
] (2.138)
(2.127)
2 0 0
where,
where,
Ip
ip ip
=
Ip
(2.139) (2.128)
A A
is is
thethe polar
polar radius
radius of inertia
of inertia about
about the shear
the shear centre.
centre.
Example
Example
Consider
Considera simply supported
a simply column
supported having
column a fork
having a type
fork of support
type as shown
of support in Figure
as shown 2.50. 2.50.
in Figure
Boundary conditions:
• Boundary conditions:
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 56 of 68
⎧v = v, xx = 0
At x = 0, ⎪
⎨w = w, xx = 0
⎪φ = φ , = 0
⎩ xx
• Displacement functions:
πx
v = C1 sin
l
πx
w = C 2 sin
l
πx
φ = C3 sin
l
When the displacement functions are introduced, the total potential energy can be written as,
Π =U + H
(2.129)
= π 2 [C 12 (N Ez − N ) + C 22 (N Ey − N ) + C 32 i 2p (N E φ − N ) + 2 C 1 C 3 Nzo − 2 C 2 C 3 Nyo ]
2
4
where,
π 2 EI z π 2 EI y
N Ez = 2
; N Ey = 2
1⎛ π 2
EI w ⎞ GI t π EI w
2
N Eφ = 2 ⎜ GI t + ⎟ ⎯⎯→ σEφ = + 2
ip ⎝
2
⎠ or Ip I p
∂Π δΠ ∂Π
δΠ = δ C1 + δ C2 + δ C3 = 0 (2.130)
∂ c1 ∂ c2 ∂ C3
For this condition to be generally satisfied, each term has to be equal to zero. This yields three
equations, with unknown quantities C1, C2, and C3, which can be arranged in a matrix form.
To satisfy the singularity requirement for the system, the determinant must be equal to zero,
i.e.
(N Ez − N ) 0 Nzo
det 0 (N Ey − N ) − Nyo = 0 (2.131)
− Nyo i p (N Eφ − N )
2
Nzo
22 2 2
N yo
(N Ez − N )(N Ey − N )(N Eφ − N ) − (N Ez − N ) 2 − (N Ey − N ) N 2 zo = 0 (2.132)
ip ip
Special Cases:
y o = zo = 0
(2.133)
⇒ (N Ez − N )(N Ey − N )(N Eφ − N ) = 0
Double symmetric cross-sections buckle in either of the two bending modes, or by torsion
about the shear centre. A combined torsional/bending buckling does not occur.
The buckling mode is governed by the lower value of NEz, Ney, and NEφ.
N Eφ < N Ey , N Ez ⇒σ Eφ = N Eφ A (2.134)
reduced slenderness,
σY
λφ = (2.135)
σ Eφ
z
⎡ N yo ⎤
2 2
zo = 0 ⇒ (N Ez − N )⎢(N Ey − N )(N Eφ − N ) − 2 ⎥ = 0 (2.136)
yo = 0 ip ⎦
⎣
SC CG
y
y0
Figure 2.51 A Channel Section.
Consider a channel section shown in Figure 2.51. The possible solutions are,
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 58 of 68
N Eyφ N Eyφ
Various combinations of and satsifying this interaction function are plotted in
N Ey N Eφ
Figure 2.52 for various values of the k parameter.The line N Eφ = N Ey splits the area into two
domains and the diagram is symmetric about this line: Below the line the torsional buckling
strength is larger than the strong axis buckling strength, above the line the weak axis buckling
strength is greater than the torsional buckling strength.
The figure shows the influence of interaction. If the torsional buckling strength and the strong
axis buckling strength differ considerably, the interaction is moderate (see e.g. the intersection
with the line N Eφ = 4N Ey . It is seen that the maximum reduction amounts to 50% and is
obtained for k = 0 and N Eφ = N Ey . For k-values in the range of 0.7-0.9 the reduction caused
by interaction is fairly small if the relative buckling strength for torsion and strong axis
buckling is larger than two.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 59 of 68
1
0.9 k =1.0
0.8 0.9
0.7 0.8
NEy =4NEφ 0.7
0.6
NEyφ /NEφ
0
0 0.2 0.4 0.6 0.8 1
NEyφ/NEy
The flexural buckling strength increases with reduced buckling length, whereas the St.Venant
contribution to the torsional buckling strength is independent of the buckling length. Hence,
relatively speaking, short members are more prone to torsional buckling. This is shown for
channel sections in Figure 2.53, where the relative buckling strength for strong axis and –
torsional buckling strength is depicted. The same buckling length is assumed for strong axis
buckling and warping. The figure shows also the k-factor for the various profiles.
It is observed that fairly equal buckling strengths is possible over some interval, which
implies that interaction should be considered. However, as the moment of inertia about the
weak axis for most profiles is several times smaller than that about the strong axis, weak axis
buckling will be critical, unless the profile is supported such that the weak axis buckling
length is smaller.
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling Relative
of Bars elastic
and Frames
buckling strength for strong axis - and torsional buckling ofPage 60 of 68
channel sections
4
200x80x4 k=0,77
Relative buckling stress: σEφ/σEz or
100x100x4 k=0.34
2
1.5
0.5
Torsionally weak Flexurally weak
0
0 0.5 1 1.5 2 2.5 3
Reduced slenderness ratio-strong axis buckling
Figure 2.53 Relative buckling strength for strong axis – and torsional buckling of channel
sections. Yield stress: 250 MPa.
z
y0C3
C3
C2
CG
y
y0 (1+C2/(y0C3)
In this case the stiffener is forced to rotate about the web toe when it undergoes torsional
buckling. The displacement of the shear centre is given by,
v = − hφ
(2.140)
w= 0
and the displacement of any arbitrary point in the shear centred axis system is given by,
TMR4205 Buckling and Ultimate Strength of Marine Structures
TMR4205 Buckling
2. Buckling of Barsand
andUltimate
Frames Strength of Marine Structures Page 62 of 69
2. Buckling of Bars and Frames Page 61 of 68
v y, z v z h z
v (wy,y,z )z= v y− zφ = −(h + z )φ (2.152)
(2.141)
w( y, z ) = yφ
z, w
y,v SC
The potential energy of the external load can then be written as,
The potential energy of the external load can then be written as,
1
( )
2 2
⎡ (h z )2 y 2 dA ⎤ , x 2dx
2
H 1
H = − 2∫0σ x ⎢ ∫A h + z + y dA⎥φ , x dx
x
2 0 ⎣A ⎦ (2.153)
1 2 1 2 2 (2.142)
12 x I p , dx
1 N i p2 , 2dx
x x
= − ∫0σ x I p φ , x dx = − N i p0∫ φ , x dx
2 2
20 2 0
where the polar moment of inertia, Ip, is evaluated about the weld toe,
where the polar moment of inertia, Ip, is evaluated about the weld toe,
2
Ip r 2 dA h z y 2 dA (2.154)
I p = ∫ r dA = ∫ (h + z ) + y dA
A 2 A
( 2 2
) (2.143)
A A
Using the same boundary conditions as before, the displacement shape functions are given by,
Using the same boundary conditions as before, the displacement shape functions are given by,
x x
v C1 sin h C 3 sin
πx πx
v w= C01 sin = −h C 3 sin (2.155)
w = 0C 3 sin x (2.144)
πx
φ = C 3 sin
By comparing with the general (4.125),
it can easily be seen that the following total potential
energy is obtained,
By comparing with the general (2.140), it is seen that the following total potential energy is
2
obtained, 1 2 2 2
U H 2 h N Ez i p N E N C3 0 (2.156)
4
2
π 1( 2
Π =U + H = h N Ez + i p (N Eφ − N ))C 3 = 0
2 2
which yields, (2.145)
4 22
N h N Ez (2.157)
NE NE 2
which yields, ip
2
= N E =is Nlocatedh N Ez
For L-profiles, the shearN centre Eφ + (2.146)
at2 the web/top flange intersection. This yields a
ip
For L-profiles, the shear centre is located at the web/top flange intersection. This yields a
negligible warping stiffness. Hence,
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 62 of 68
1
N Eφ = 2 GI t (2.147)
ip
(see the general equation in Section 2.9.2). Therefore, the critical load comes out to be,
N E = GI t + π Eh I z
2 2
σE = 2
(2.148)
A Ip Ip
From Figure 2.54 it is seen that the plate has to rotate when the stiffener rotate about the web
toe. If the plate is little utilised with respect to buckling it is reasonable to assume that it
exerts some restraint on the rotation. Hence, the critical stress will be increased compared to
equation 2.148.On the other hand, if the plate fails first this is likely to speed up torsional
buckling. These phenomena are still not understood in detail, but in NORSOK N-004 the
effect is accounted for by multiplying the St. Venant contribution with a factor β so that
GI t π Eh I z
2 2
σE = β + 2 (2.149)
Ip l Ip
β is given by
3C + 0.2
β= (2.150)
C + 0.2
3
h⎛ t ⎞
C = ⎜⎜ ⎟⎟ (1 − η) (2.151)
s ⎝ tw ⎠
where h is the web height of the stiffener, s is the stiffener spacing, t is plate thickness and tw is
web thickness. η is the utilisation with respect to local buckling of the plate.
For the L-profile shown in Figure 2.55, the following relationships hold true:-
I p = h 2 ⎜ Af + Aw ⎟and I t = (t 2f Af + tw2 Aw )
⎛ 1 ⎞ 1
(2.152)
⎝ 3 ⎠ 3
Therefore,
1 ⎡ ⎛ t f ⎞ ⎤ ⎧⎪ ⎛ t w ⎞ ⎫⎪
2 2
1 π 2 EI z
σE = + β
⎢ Aw ⎜ ⎟ A f ⎥ ⎨ ⎜ ⎟ ⎬
G + (2.153)
Aw + 3 A f ⎣⎢ ⎝ t w ⎠ ⎦⎥ ⎪⎩ ⎝ h ⎠ ⎪⎭ ⎛ Aw + A ⎞ l
2
⎜ f ⎟
⎝ 3 ⎠
where,
1
t f b + Aw y + A f (e − y )
2 2
Iz =
3
12
A f Aw ⎡ (2.154)
1 eA f ⎤
= Af b + e f
2 2
, ⎢For y = ⎥
12 A f + Aw ⎣ Aw + A f ⎦
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling
TMR4205 of Bars
Buckling and and Frames
Ultimate Strength of Marine Structures Page 64 of 69
2. Buckling of Bars and Frames e = eccentricity of top flange centroid Page 63 of 68
Af b
tf
e
h
tw
Aw
In obtaining
In obtaining the critical
the critical stress
stress due due to non-linear
to non-linear material
material behaviour,column
behaviour, columncurve
curve ee isis used,
used,
refer Section
refer Section 2.4.1.2.4.1. Restrained
Restrained lateral-flexural
lateral-flexural buckling
buckling strengthis isequal
strength equaltotothe
theyield
yield strength
strength
if if
or
2.8 orY σ y y 0.6 (2.166)
σE ≥ 2.8
E
σY ⎯ ⎯→ λ = ≤ E0.6 (2.155)
σE
If the restrained torsional buckling strength of the stiffener is less than the yield strength, this
If the isrestrained
taken intotorsional
account bybuckling strength
replacing of strength
the yield the stiffener is less
with the than the
torsional yield strength
buckling strength,inthis
the
is taken into equations
design account byforreplacing the yield
stiffened plates andstrength withSection
shells, Refer the torsional buckling
3 and Section 5. strength in the
design equations for stiffened plates and shells, Refer Section 3 and Section 5.
z, z s
Mo
σo
+
A ys zo
SC
y
CG
A
Mo -
v( y, z ) = v − {z − z c }φ
(2.156)
w( y, z ) = 0 + yφ
where v is the displacement of the shear centre and zc is the distance form the centre of gravity
(CG) to the shear centre (SC). The axial stress for an arbitrary point is given by
σ = Mo z (2.157)
Iy
(Note: Compressive stress is defined positive)
Taking into account the linear variation of the axial stress, the potential energy of the external
load takes the following form,
[ ]
l
1 Moz
H =− ∫ ∫ (v ,x − (z − z c )φ ,x )2 + ( yφ ,x )2 dAdx
2 0 A Iy
(2.158)
1 M ⎡ ⎤
l
( ( ) ( ) )
= − ∫ o ⎢ ∫ v 2x z − 2 z 2 − z z c v ,x φ ,x + z 3 − 2 z 2 z c + z z c2 + zy 2 φ ,2x dA⎥ dx
2 0 Iy ⎣ A ⎦
∫z dA = I y
2
(2.159)
( ( ) )
l
1 Mo
H =− ∫
2 0 Iy
− 2 I y v, x φ , x + ∫ z 3 dA + ∫ z y 2 dA − 2 I y z c φ , 2x dx (2.160)
2 M (− v , φ , x + γφ ,2x )dx
1
2∫
H =− o x
(2.161)
0
γ =
1
2Iy
(∫ z dA + ∫ z y dA − z )
A
3
A
2
c (2.162)
l
1 1l 1l
U= ∫ EI z v ,xx + ∫ GI t φ ,x dx + 2∫
φ ,2xx dx
2 2
EI w (2.164)
20 20 0
πx
v = C 1 sin
l
w=0 (2.165)
πx
φ = C 3 sin
l
Comparing with Equation (2.140), it is seen that the total potential energy comes out to be
Π =U + H =
π2
4l 2
(C N2
1 Ez ( )
+ C 32 i p2 N Eφ − 2γ M o + 2 C 1C 3 M o ) (2.166)
δΠ = 0 (2.167)
yields
⎡N E Mo ⎤
det ⎢
⎢⎣ M o (i N 2
p Eφ
− 2γ M o ⎥⎥
⎦
=0
) (2.168)
M o + 2γ N Ez M o − i p N Eφ N Ez = 0
2 2
(2.169)
with a solution,
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 66 of 68
M o = N Ez (γ 2
+ C2 − γ ) (2.170)
where,
i p2 N Eφ IW ⎛l⎞ IT
2
C = = +⎜ ⎟ (2.171)
2
N Ez I z ⎝ π ⎠ 2(1 + ν ) I z
If the top flange is in tension the bending stress reverses sign and the solution takes the
following form
2
(
M o = N Ez γ + C + γ
2
) (2.173)
For double symmetrical section it is easily seen that γ = 0. This yields the following critical
stress
M o zo π 2 EI z z o
σ EV = σ o = = 2
⋅C (2.175)
Iy l Iy
Elasto-plastic buckling is obtained from curve e according to ECCS (refer Section 2.4.1),
where the reduced slenderness is calculated as,
σY
λV = (2.176)
σ EV
Lateral torsional buckling is not governing if λ V < 0.6 , (σ EV = σ Y ) . If λ V > 0.6 the critical
stress is obtained from column curve e.
If the top flange is restrained against lateral displacement (the top flange may for example be
a part of a continuous plate field) the beam will buckle about the web toe at the plate flange.
The bending and torsional deformation is coupled such that the lateral displacement at the plate
flange is zero. This yields the following displacement functions:
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 67 of 68
πx
v = (z 0 − z c )φ = (z 0 − z c )C 3 sin
l
w=0 (2.177)
πx
φ = C 3 sin
l
Introducing these displacement functions into the load potential the following critical load is
obtained
M o = N Ez
1
2γ
(
(z 0 − z c )2 + C 2 ) (2.178)
i.e. the distance to the plate flange, z0, replaces the distance to the shear centre, zc.
Literature
/1/ Eurocode 3: design of steel strucxtures . Part1.1 General rules and rules for buildings DD
ENV 1993-1-1
/3/ Det Norske Veritas: Classification note 30.1: Buckling Strength Analysis, 1992 and 1995.
/5/ Norwegian Petroleum Directorate - Regulations relating to loadbearing structures, NPD 1993
/6/ American Institute of Steel Construction (AISC), Manual of Steel Construction- Load and
resistance Factor design, Vol1, 2nd ed., 1994
/7/ ISO 19902:2007 Petroleum and natural gas industries -- Fixed steel offshore structures
TMR4205 Buckling and Ultimate Strength of Marine Structures
2. Buckling of Bars and Frames Page 68 of 68
INDEX
A J
allowable stress · 21 joint flexibility
amplification factor · 7, 9, 10, 11, 17, 22 effect of - buckling · 35
B L
Beam Columns · 17 lateral-torsional buckling · 22, 64
bifurcation
point · 4
N
C nontrivial solution
to differential equation · 3, 39
column buckling · 18
column curve · 15, 16, 58, 64, 67
critical stress · 14, 15, 19, 21, 22, 63, 64, 67 O
out-of-straightness · 14, 19
D
determinant · 39, 47, 57 P
particular solution
E differential equation · 8, 10
potential energy · 7, 55, 56, 57, 62, 65, 66
effective buckling length · 34, 36
elasto-plastic material · 12
equivalent bending stress · 20 R
equivalent imperfection · 14, 19
equivalent spring · 45 reduced slenderness ratio · 12
Equivalent Stress · 21 residual stresses
Euler buckling stress · 12 imperfection · 12
F S
flexural-torsional buckling · 52 section modulus · 18
sidesway
prevented /permitted buckling · 22, 34
H Stiffness Matrix · 37
strain energy · 7, 55, 66
homogeneous
solution to differential equation · 8, 10, 38
T
I Torsional Buckling · 52, 55, 58, 61, 64, 65
transition curve · 12
Imperfection
geometric · 8
initial distortions · 12, 59 W
interaction formula · 19, 21, 22
warping · 52, 54, 55, 63