Design validation & optimization of offshore foundations (Wout Weijtjens, VUB)
Foundation monitoring on offshore wind farms for O&M optimization and life time extension (Wout Weijtjens, VUB)
3. Our
Op-wind
Team
Prof.
dr.
ir.
Christof
Devriendt
christof.devriendt@vub.ac.be
ir.
Nymfa
Noppe
nymfa.noppe@vub.ac.be
ir.
Tim
Verbelen
,m.verbelen@vub.ac.be
MEng.
Alexandros
Iliopoulos
Alexandros.iliopoulos@vub.ac.be
dr.
Rasoul
Shirzadeh
Has
leT
the
department
dr.ir.
Wout
Weijtjens
wout.weijtjens@vub.ac.be
4. Problem
descrip,on
Offshore
wind
turbines
are
pushing
boundaries
-‐
Monopile
design
remains
dominant
(
>
90%
)
-‐
Growing
mismatch
between
lab
and
real-‐life
-‐
Diameters
of
7-‐8
m
-‐
6-‐7
MW
machine
on
Monopile
-‐
1200
ton
weight
Meanwhile
the
designs
remain
fa-gue
driven
5. Problem
descrip,on
There
is
a
known
mismatch
between
design
assump,ons
and
real
life
condi,ons
which
directly
affect
fa-gue
life!
-‐
Uncertain,es
in
the
soil
models
6. Problem
descrip,on
There
is
a
known
mismatch
between
design
assump,ons
and
real
life
condi,ons
which
directly
affect
fa-gue
life!
-‐
Uncertain,es
in
the
soil
models
E.g.
PISA
project
in
UK
(led
by
DONG
energy)
7. Problem
descrip,on
There
is
a
known
mismatch
between
design
assump,ons
and
real
life
condi,ons
which
directly
affect
fa-gue
life!
-‐
Uncertain,es
in
the
soil
models
-‐
Unknown
damping
behavior
8. Problem
descrip,on
There
is
a
known
mismatch
between
design
assump,ons
and
real
life
condi,ons
which
directly
affect
fa-gue
life!
-‐
Uncertain,es
in
the
soil
models
-‐
Unknown
damping
behavior
-‐
Hydrodynamic
loads
on
very
large
cylinders
E.g.
WiFi
JIP
project
in
NL
9. Problem
descrip,on
There
is
a
known
mismatch
between
design
assump,ons
and
real
life
condi,ons
which
directly
affect
fa-gue
life!
-‐
Uncertain,es
in
the
soil
models
-‐
Unknown
damping
behavior
-‐
Hydrodynamic
loads
on
very
large
cylinders
E.g.
WiFi
JIP
project
in
NL
10. Problem
descrip,on
What
with
already
built
farms?
!
uncertain,es
have
led
to
overdesign
Poten,al
room
for
life
,me
extension
or
repowering
11. Resonance
frequency
shiT
Operator
/
Developer
driven
ques-on:
Were
the
as
designed
frequency
values
correct?
WP5
12. Resonance
frequency
shiT
Operator
/
Developer
driven
ques-on:
Design
valida-on
:
Were
the
as
designed
frequency
values
correct?
WP5
13. Resonance
frequency
shiT
Operator
/
Developer
driven
ques-on:
Were
the
as
designed
frequency
values
correct?
As designed frequencies for first mode differ with about
5-10%!
WP5
14. Resonance
frequency
shiT
Operator
/
Developer
driven
Stiffer than designed is not Necessarily a good thing
Direct impact on design of Northwind and Nobelwind -> New O&O
Most likely cause > Soil!
WP5
15. O&M
:
Scour
assessment
Use
design
valida,on
set-‐up
To
check
if
the
resonance
Frequencies
are
s,ll
within
an
Acceptable
range
!
Currently
overcompensated
by
the
s-ffer
soil
WP7
16. Blue: May 2012
Green: November 2013
Red: model
500 1000 1500 2000 2500
1.38
1.4
1.42
1.44
1.46
1.48
1.5
1.52
ResonanceFrequency(SS2)(Hz)
Time
500 1000 1500 2000 2500
−0.04
−0.02
0
0.02
0.04
0.06
0.08
PredictionError(Hz)
Time
Training Use relative error as a damage feature
Residual
error
leaves
3σ
boundary
Prediction
O&M
Monitoring
Resonance
Frequencies
SOIL STIFFENING?
Methodology
to
monitor
scour
by
tracking
the
resonance
frequencies
WP7
17. Damping
values
2012
Damping
Es-ma-on
of
an
Offshore
Wind
Turbine
on
a
Monopile
Founda-on
-‐
Determined
the
damping
ra,o
of
the
turbine’s
first
modes
aTer
an
overspeed
stop
-‐
Use
of
ambient
vibra,ons
to
determine
the
damping
ra,o
FA
:
1.04%*
SS
:
1.25%*
FA
:
1.05%*
*
With
Tuned
Mass
Damper
turned
off
WP5
18. Damping
values
2013
Monitoring
of
Resonant
Frequencies
and
Damping
Values
of
a
Parked
Offshore
Wind
Turbine
on
a
Monopile
Founda-on
A
fully
automated
method
to
track
the
first
five
resonance
frequencies
and
damping
ra,os
con,nuously
Resonance
frequencies
and
damping
ra-os
depend
on
the
ambient
condi-ons
(e.g.
Wind
speed,
Tidal
level)
WP5
19.
Damping
values
2013
Monitoring
of
Resonant
Frequencies
and
Damping
Values
of
a
Parked
Offshore
Wind
Turbine
on
a
Monopile
Founda-on
A
fully
automated
method
to
track
the
first
five
resonance
frequencies
and
damping
ra,os
con,nuously
Resonance
frequencies
and
damping
ra-os
depend
on
the
ambient
condi-ons
(e.g.
Wind
speed,
Tidal
level)
WP5
20. Damping
values
2014
Classifying
Resonant
Frequencies
and
Damping
Values
of
a
Fully
Opera-onal
Offshore
Wind
Turbine
Resonance
frequencies
and
Damping
ra-os
can
shiM
significantly
between
opera-onal
condi-ons
!
An
es-mate
for
each
opera-onal
condi-on
WP5
21. Damping
values
2015
Automated
monitoring
of
five
offshore
wind
turbines
in
the
Belgian
North
Sea
Mul,
algorithm
technique
:
FDD,
SSI-‐COV,
polymax
Consistent
behavior
between
turbines
and
over
-me
WP5
22. Classifying
the
data
Use
par,al
SCADA-‐data
to
classify
each
ten-‐minute
data
set
into
different
opera-onal
cases
Get
results
case-‐by-‐case
Case
defini,ons
can
be
updated
for
different
applica,ons
WP5
23. Cau,on
with
Harmonics
The
original
algorithm
will
falsely
iden-fy
harmonics
as
if
they
are
structural
!
Track
harmonics
and
reject
unreliable
results
WP5
25. Damping
Analysis
of
FA1
(e.g.
Using
the
SCADA-‐Windspeed)
(
considering
only
produc6on
cases)
WP5
26. Damping
Analysis
of
SS1
(e.g.
Using
the
SCADA-‐Windspeed)
(
considering
only
produc6on
cases)
WP5
27. Modeling
Damping
Greatest
disadvantage
of
damping
determined
with
OMA
:
Only
total
damping
es-mated
-‐>
Has
almost
no
value
for
current
design
code!
Accounted
for
in
,me
domain
simula,on
(Aero-‐
and
Hydro-‐dynamic
loads)
Accounted
for
in
Structural
model
Best
Es,mate
-‐>
Need
for
full
scale
measurements!
OMA
Design
WP5
28. Methodology
to
decompose
in
different
damping
contribu-ons
Mixed
method
that
combines
Simula,ons
and
Measurements
29. Modeling
damping
• Loads
are
modeled
correctly?
• Soil
model
already
ques,oned
• How
to
update
your
damping
correctly
• Based
on
solely
accelera,ons
might
be
insufficient
• Impact
on
induced
strain
of
different
damping
distribu,ons?
-‐>
Fa-gue
WP5
30. Fa,gue
monitoring
• Issues
revealed
in
modeling
fa,gue
loads
• Mo,va,on
for
direct
measurements
and
monitoring
of
fa,gue
life!
WP7
31. Fa,gue
monitoring
Hot
spot
Monitoring
Stress
Monitoring
Fa,gue
Life
Es,ma,on
Farm
wide
Fa,gue
Life
Con,nuous
measurements
of
the
strains
-‐
Bending
moments
at
measurement
loca-on
-‐ Comparison
with
design
loads
-‐
Local
stress
cycles
WP7
33. Fa,gue
monitoring
Hot
spot
Monitoring
Stress
Monitoring
Fa,gue
Life
Es,ma,on
Farm
wide
Fa,gue
Life
Virtual
sensing
-‐
Stress
cycles
at
Fa-gue
cri-cal
loca-on
-‐
e.g.
as
defined
by
design
-‐
Stress
cycles
at
any
loca-on
-‐
Determine
the
actual
cri-cal
loca-on
WP7
34. Fa,gue
monitoring
Hot
spot
Monitoring
Stress
Monitoring
Fa,gue
Life
Es,ma,on
Farm
wide
Fa,gue
Life
Rainflow
coun,ng
on
Virtual
stress
cycles
Assess
damage
with
-‐
Appropriate
S/N-‐Curves
-‐
Stress
concentra9on
factors
Assess
impact
of
Corrosion
regime
>
O&M
Support
The closer you get,!
the faster you fatigue!
WP7
35. Fa,gue
monitoring
Hot
spot
Monitoring
Stress
Monitoring
Fa,gue
Life
Es,ma,on
Farm
wide
Fa,gue
Life
But
rainflow
coun,ng
only
gives
accumulated
damage
in
the
past!
To
extrapolate
in
9me
we
need
to
know
the
link
between
Damage
and
:
-‐
Opera9onal
condi9ons
-‐
Environmental
condi9ons
-‐
Excep9onal
load
cases
WP7
36. Fa,gue
monitoring
Hot
spot
Monitoring
Stress
Monitoring
Fa,gue
Life
Es,ma,on
Farm
wide
Fa,gue
Life
Farm
wide
fa,gue
assessment
-‐
Extrapolate
the
results
from
the
FLEET
LEADERS
to
the
en-re
farm
-‐ Turbulence
and
Turbine
interac-ons
-‐
Site
specific
correc-on
factors
-‐
Different
resonance
frequencies
-‐
Damping
values
WP7
37. What
is
next?
Fleet
leader
concept
Instrumented
Fleet
leaders
To
predict
fa,gue
life
in
the
farm
Preliminary
work
and
IWT
project
proposal
Nymfa
Noppe
WP7
38. What
is
next?
O&O
Nobelwind
Verifying
the
lessons
learned
from
Northwind
&
Belwind
-‐
New
design
approaches
were
applied
at
Nobelwind
e.g.
advanced
soil
modeling
(
Construc,on
starts
Q1
2016
)
-‐
Monitoring
campaign
on
three
wind
turbines
-‐
Unique
setup
with
strain
gauges
on
the
monopile