Professional Documents
Culture Documents
LCF Fatigue Design Presentation Rev
LCF Fatigue Design Presentation Rev
ASSESSMENT FOR
OFFSHORE STRUCTURES
UNDER SEISMIC LOADING
1
contents
The engineering problem
Issues to consider low cycle fatigue
Solution by generalised Masing’s approach
Concept of relative notch SCF
Numeric implementation
Examples
Sensitivity study
Conclusion
2
The engineering problem
Many of the existing platforms from our offshore
operators are in seismic active zones
New risk assessment requires quantification of
failure and more rigorous justification of structural
integrity for seismic events which may not be
considered at the design stage
Each of the platforms have hundreds of hot spots
for potential failure by fatigue or fracture so
detailed FE analysis for each hot spot in the
platform is not practical
3
Seismic Analysis for Structure
with Non-linear Foundation
One ABAQUS
non-linear
analysis run of
a beam model
takes about 8
hours for a 40
sec seismic
event
4
Low Cycle Fatigue
Under seismic loading, hot spots in the structure will experience
reversals of high level stress or strain and low cycle fatigue
(LCF)
Damage has been observed in steel structures under
earthquake by LCF, e.g. Northridge earthquake
Normal S-N curves in offshore steel standards deal only with
fatigue of medium to low stress and high cycle
Other strain based fatigue analysis approaches ignore sequence
effect and hence ignore true strain development
A methodology is needed to assess LCF for large offshore
structures under random loading efficiently and consistently
5
Issues to be considered
6
Masing’s Hypothesis:
Masing suggested that the stress-strain curve for reversed straining
MEASURED FROM THE POINT OF REVERSAL would be the cyclic
curve scaled by a factor of 2. It has been applied to a number of
situation for strain related studies.
e e e e p
s 2 2 2
e e s
2 2E
1
e p s n'
2 2k'
Hence,
1
e e s s n'
2 2 E 2k'
7
Masing’s Hypothesis
s, s/2
Reversed
s-e curve
1
2e 2s s n'
2
2 2E 2k'
1
s s n'
e 2
Cyclic E 2k'
s-e curve
1
e s s n'
2 2E 2k'
8
e, e/2
Problem in using classic Masings
rule for random nominal stress
history
Classic Masing
400
Classic Masing
300
200
Stress strain
100
loops produced
from classic
0
-0.40% -0.20% 0.00% 0.20% 0.40% 0.60%
Masing’s rule
-100
-200
-300
show spurious
-400
high stresses
-500
-600
9
Generalised Masing rule
At each unloading path or reloading path,
the stress strain curve follows a track in
stress-strain space directed to a point
called the fiducial point, amount to energy
dissipation through the nonlinear property
of the material
Instead of a scale factor of 2 in Classic
Masing, the generalised Masing rule uses a
changing scale factor, controlled by only
one point in the loop, to control the cyclic
loading/unloading loop.
Although developed for soil under seismic
condition, principle of energy dissipation
is applicable for steel as well
It offers simplicity and stability in numeric
implementation. But it reduces the effect
from extreme ratcheting. For a mean level
of fatigue under random loading, those
extremes may be ignored.
10
SCF for offshore weld
11
Relative notch stress SCF for
different weld class
Kn = stress range So of smooth specimen / stress range So of notched specimen
HSE S-N Curve HSE GN in seawater with adequate
4thEdition in air protection
ID
So at 107 Kn So at 107 Kn
B 0.64P 83 1.00 61 1.351
C 0.76P 70 1.19 52 1.604
D 1.00P 53 1.57 39 2.111
E 1.14P 47 1.77 34 2.407
F 1.34P 40 2.08 29 2.828
F2 1.52P 35 2.37 26 3.209
G 1.83P 29 2.86 21 3.861
W 2.54P 21 3.95 15 5.363
T' 67 1.24 53 1.563
12
Comparison of notch stress SCF
K t SCF K n K 97
K97 = fatigue strength with mean curve / fatigue strength with design curve
For the B curve of smooth specimen, K97 = 1.233
13
Strain concentration in the
inelastic range
Stress-Strain at Notches
Kt K s Ke
2
Neuber’s rule:
e
Glinka’s modification: ( K t S ) 2
s de
2E 0
14
SCF in the inelastic range
15
Cyclic strain analysis flow chart
Obtain nominal stress history from axial, in plane bending and
out of plane bending from existing analysis in SAFJAC
16
Example of strain history loops -
1 Mudline Acceleration at Leg A1 - Y Direction
Seismic
25
20
acceleration
15
2
10
Acceleration, m/s
5
0
history
-5 0 5 10 15 20 25 30 35 40
-10
-15
800
-20
Stress (Mpa)
-25
600
Tim e, s
400
200
Stress 0
Strain
Strain
-4.0% -2.0% 0.0% 2.0% 4.0% 6.0% 8.0% 10.0% 12.0% 14.0%
-200
loops -400
-600
17
Example of strain history loops -
2
Nominal stress history
800.00
Nominal stress
600.00
400.00
Nominal stress (MPa)
Stress (Mpa)
200.00
1000
0.00 800
0 5 10 15 20 25 30 35
-200.00
600
-400.00
400
-600.00
time (s)
200
True strain history True Strain History -6.0% -4.0% Strain -2.0%
0
0.0% 2.0% 4.0% 6.0% 8.0%
8.0% -200
6.0% -400
4.0%
2.0%
time step
-600 Stress
Strain
Strain
-800
0.0%
1 1001 2001 3001
loops
-2.0% -1000
-4.0%
-6.0%
18
Example of strain history loops -
3
Stress (Mpa)
600
400
200
0
-16.0% -14.0% -12.0% -10.0% -8.0% -6.0% -4.0% -2.0% 0.0% 2.0% 4.0% 6.0%
Strain
-200
5.0%
-600
Stress
Strain
time step
0.0%
1 1001 2001 3001
Strain
-800
loops
-5.0%
-10.0%
-15.0%
-20.0%
19
Example of fatigue damage for a
platform for an earthquake analysis
20
Benefit from cyclic strain analysis
21
Sensitivity study - nominal stress
Strain-based fatigue
1000 Sy = 345 Mpa
4.1787
damage sum is
Sy = 245 Mpa y = 1.0788x
y, Damage (x)/Damage(x=1)
22
Sensitivity study - Yield stress
2
Fatigue damage is sensitive
to yield stress but depends
on stress amplitude.
Damage (Sy =245)
Damage(Sy=345)/
23
Conclusion
A method has been established for strain based fatigue
analysis for offshore structure with generalised Masing’s
rule and approximate notch factors based on weld classes
This method is fully integrated to normal structural
analysis programs in a seamless manner, enabling speedy
analysis for large number of hot spots in a structure
The results from analysis include damage ratios as well as
maximum/minimum true strain for integrity assessment
From sensitivity analysis for the current structure:
– The ratio of damage is proportional to ratio of nominal
stress to the power of 4
– Damage is sensitive to yield stress
24
Comparison of e-Nf curves
1
Grade 350
0.1
Grade 245
e
Program default
0.01
0.001
1 10 100 1000 10000 100000 1000000 10000000
Nf
25
FE model for notch stress analysis
26
Comparison of VM notch stresses
Neuber's VM
600.0
ESED VM
FE 3x0.5x1 NoWeld
FE 3x0.5x1
400.0
Von Mises stress (Mpa)
300.0
200.0
100.0
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2
P / PL
27
Comparison of VM notch strain
12.0%
Neuber's Eeq
ESED Eeq
10.0%
Neuber's Eeq using s1 only
FE 3x0.5x1
6.0%
4.0%
2.0%
0.0%
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2
P / PL
28
Stress distribution around the weld
toe when the applied stress on the
brace= 100 MPa
700
8 Inc_VM
8 Inc_VM (1x2)
8 Inc_VM (0.5x1)
8 Inc_VM (4X4) Noweld
600 8 Inc._Pmax (0.5x1) Noweld
400
300
200
-8 -4 0 4 8 12 16 20 24 28 32
Distance from weld toe (mm)
29
30
Ratio of Neuber Prediction against FE result
20%
40%
60%
80%
100%
120%
140%
0%
Von Mises
equivalent
stress
Von Mises
equivalent
strain
Max
P / PL= 0.7
principal
stress
Max
principal
strain
Von Mises
General comparison
equivalent
stress
Von Mises
equivalent
strain
Principal stress only
Max
P / PL= 1.4
principal
HS extension with Ktq= Kt
stress
HS extension with Ktq= 0.88 Kt
Max
principal
strain