Metocean Extreme and Operating Conditions 3.5 Joint Events 61
Part A | 3.5
Princeton Ocean Model (POM). Nowcasts have been
archived back to 2010. Other regional models are also
available: RTOFS for the North Atlantic, MERCATOR for Mediterranean [3.104], and BLUElink for
Australia [3.105]. These models assimilate satellite observations within their domain and take their boundary
conditions from larger-scale global models.
When utilizing the archive data sets from 3-D models, one must keep in mind the weaknesses described
earlier in this section. More specifically, these archived
products will not adequately resolve the peak current
during tropical cyclones. Nor will they be able to reliably replicate historical mesoscale features, though they
may be able to reproduce the statistics of those features
(e.g., reproduce the histogram of speed). In summary, if
there are energetic ocean current processes with length
scales of less than 100 km affecting the site of interest,
model archives should only be used with caution. At the
very least, several months (preferably much more) of
local measurements should be obtained and used to validate and calibrate the model before relying on model
results.
3.5 Joint Events
Most ships and offshore facilities are designed to withstand a load with a specific return interval of n years,
e.g., the 100-year event. For many decades, offshore
designers assumed that the n-y event was created by the
simultaneous occurrence of the n-y wind, n-y wave, and
n-y current (i. e., the so-called n-y independent events),
all aligned in the same direction. However, about
30 years ago, metocean researchers started collecting
detailed measurements during major storm events and
realized that the peaks of winds, waves, and currents,
in fact, did not occur simultaneously in direction or
time and they began developing various techniques
to account for this fact. The more popular ones are
described next.
3.5.1 Response-Based Analysis
The simplest and perhaps most accurate way of estimating the n-y response is to feed a time series of wind,
wave, current into a response model of the facility and
then do an extreme analysis of a key response variable. For example, if a structural engineer designing
the legs in an offshore jacket for the 100-y overturning
moment (OTM). In the response-based approach, the
structural engineer would first develop a fairly simple
response function whose input variables include wind,
wave, and current and whose output is the OTM. Second, the metocean time series is fed into the response
function resulting in a time series of OTM. Third, a peak
over threshold (POT) analysis is done, as described in
Sect. 3.7.2 and the n-y OTM calculated. Finally, a set
of winds, waves and currents that produce the OTM is
found and used for detailed analysis.
Ewans [3.106] describes the application of the
response-based approach to pipeline stability. Heideman et al. [3.107] provide one of the earliest examples
of the approach and show that for a jacket-type structure in the North Sea, one can combine the 100-y wind
and wave with an equivalent current that is 0.25 times
the 100-y current to reach the 100-y OTM. Their case is
perhaps on the extreme end of potential savings as it is
situated in the North Sea where storm winds and waves
are weakly correlated to the extreme current. Nevertheless, even in regions dominated by hurricanes where
the metocean variables are highly correlated, ANSI/
API [3.1] recommends that the 100-y wave can be combined with 0.95 of the 100-y wind speed and 0.75 of the
100-y surface current speed. A further 3% reduction of
the current and wind is allowed if directionality is considered.
The response-based approach is versatile and can
apply to the calculation of extreme loads like base
shear or OTM in a jacket, extreme responses like
the n-y heave in a ship, or operating conditions like
the marginal probability distribution of pitch and roll.
ANSI/API [3.1] recommends the response-based analysis as the preferred alternative. Part of the reason
for the rise in popularity of response-based analysis is
the increase in computer power, which has made the
repetitive solution of fairly complex response functions
feasible. Another enabling technology has been the advent of long-duration hindcast datasets of simultaneous
wind, wave, and current time series derived from numerical models.
That said, the downside of response-based analysis
is the need for a response model with sufficient complexity to accurately reflect the critical response of the
facility yet with sufficient computational efficiency to
run many thousands of times. Developing such response
models can be daunting for complex floating systems
like TLPs (tension-leg platform) or spars. Of course
there are shortcuts in the analysis that can reduce the
computational requirement yet still preserve accurate
results. For instance, in the case of calculating extreme
loads, the metocean time series can be truncated into
a much smaller set of events that only considers the
stronger storms. Obviously this approach does not
work as well for developing operational criteria.
Part A | 3.5
Princeton Ocean Model (POM). Nowcasts have been
archived back to 2010. Other regional models are also
available: RTOFS for the North Atlantic, MERCATOR for Mediterranean [3.104], and BLUElink for
Australia [3.105]. These models assimilate satellite observations within their domain and take their boundary
conditions from larger-scale global models.
When utilizing the archive data sets from 3-D models, one must keep in mind the weaknesses described
earlier in this section. More specifically, these archived
products will not adequately resolve the peak current
during tropical cyclones. Nor will they be able to reliably replicate historical mesoscale features, though they
may be able to reproduce the statistics of those features
(e.g., reproduce the histogram of speed). In summary, if
there are energetic ocean current processes with length
scales of less than 100 km affecting the site of interest,
model archives should only be used with caution. At the
very least, several months (preferably much more) of
local measurements should be obtained and used to validate and calibrate the model before relying on model
results.
3.5 Joint Events
Most ships and offshore facilities are designed to withstand a load with a specific return interval of n years,
e.g., the 100-year event. For many decades, offshore
designers assumed that the n-y event was created by the
simultaneous occurrence of the n-y wind, n-y wave, and
n-y current (i. e., the so-called n-y independent events),
all aligned in the same direction. However, about
30 years ago, metocean researchers started collecting
detailed measurements during major storm events and
realized that the peaks of winds, waves, and currents,
in fact, did not occur simultaneously in direction or
time and they began developing various techniques
to account for this fact. The more popular ones are
described next.
3.5.1 Response-Based Analysis
The simplest and perhaps most accurate way of estimating the n-y response is to feed a time series of wind,
wave, current into a response model of the facility and
then do an extreme analysis of a key response variable. For example, if a structural engineer designing
the legs in an offshore jacket for the 100-y overturning
moment (OTM). In the response-based approach, the
structural engineer would first develop a fairly simple
response function whose input variables include wind,
wave, and current and whose output is the OTM. Second, the metocean time series is fed into the response
function resulting in a time series of OTM. Third, a peak
over threshold (POT) analysis is done, as described in
Sect. 3.7.2 and the n-y OTM calculated. Finally, a set
of winds, waves and currents that produce the OTM is
found and used for detailed analysis.
Ewans [3.106] describes the application of the
response-based approach to pipeline stability. Heideman et al. [3.107] provide one of the earliest examples
of the approach and show that for a jacket-type structure in the North Sea, one can combine the 100-y wind
and wave with an equivalent current that is 0.25 times
the 100-y current to reach the 100-y OTM. Their case is
perhaps on the extreme end of potential savings as it is
situated in the North Sea where storm winds and waves
are weakly correlated to the extreme current. Nevertheless, even in regions dominated by hurricanes where
the metocean variables are highly correlated, ANSI/
API [3.1] recommends that the 100-y wave can be combined with 0.95 of the 100-y wind speed and 0.75 of the
100-y surface current speed. A further 3% reduction of
the current and wind is allowed if directionality is considered.
The response-based approach is versatile and can
apply to the calculation of extreme loads like base
shear or OTM in a jacket, extreme responses like
the n-y heave in a ship, or operating conditions like
the marginal probability distribution of pitch and roll.
ANSI/API [3.1] recommends the response-based analysis as the preferred alternative. Part of the reason
for the rise in popularity of response-based analysis is
the increase in computer power, which has made the
repetitive solution of fairly complex response functions
feasible. Another enabling technology has been the advent of long-duration hindcast datasets of simultaneous
wind, wave, and current time series derived from numerical models.
That said, the downside of response-based analysis
is the need for a response model with sufficient complexity to accurately reflect the critical response of the
facility yet with sufficient computational efficiency to
run many thousands of times. Developing such response
models can be daunting for complex floating systems
like TLPs (tension-leg platform) or spars. Of course
there are shortcuts in the analysis that can reduce the
computational requirement yet still preserve accurate
results. For instance, in the case of calculating extreme
loads, the metocean time series can be truncated into
a much smaller set of events that only considers the
stronger storms. Obviously this approach does not
work as well for developing operational criteria.