Sie sind auf Seite 1von 10

Marine Heavy Transport & Lift IV, 29-30 October, London, UK

AN OPERATIONAL BASED METHOD FOR DETERMINING THE DESIGN SEA STATES


OF HEAVY TRANSPORT VESSELS
M Deelen and J B de Jonge, Dockwise Shipping BV, The Netherlands

SUMMARY
Since May 2008 Dockwise gathers weather and ship motion data. The Octopus motion and weather database includes
now casted weather data and motion data along the sailed route. Up to 1.6 million nautical miles of data is currently
available. This research describes a suitable method to implement Octopus weather data into the design process resulting
in more consistent operational margin over the whole range of design wave heights. The basic aim is to increase the
margin at lower design sea states <5.0 [m] and to reduce the margin for higher design sea states. This Operational Based
Method (OBM) uses a Joint Probability Density Function combining a marginal 3-parameter Weibull distribution for the
significant wave height with a conditional log-normal distribution for the zero crossing wave period to extrapolate the
wave data to the 10-year return design contour. Herewith, the weather routing capacity of the captain is included.

NOMENCLATURE 1. INTRODUCTION

BWA Bad Weather Avoidance This paper describes the development of a new
CoGH Cape of Good Hope calculation algorithm which reflects the conclusions of
CSM Cargo Securing Manual the data collected with the Octopus motion monitoring
DW Dockwise project on the Dockwise fleet of vessels. It is shown
Dosuite Dockwise design wave program where the Octopus data can be implemented in the design
GWS Global Wave Statistics process and how the data can be used to evaluate the
HMT Heavy Marine Transport design sea states. This Operational Based Method leads
HTV Heavy Transport Vessel to a more accurate and representative design wave
JPDF Joint Probability Density Function calculation, when compared to the standard Dosuite
LoNoWe Log Normal Weibull method [1].
OBM Operational Based Method The limiting design accelerations in the Heavy
POT Peak over Threshold Marine Transport (HMT) are often based on the severe
RAO Response Amplitude Operator weather conditions around the Cape of Good Hope
SPOS Ship Performance Optimization (CoGH), though depending on the route of the vessel.
System This particular region is area 90 in the Global Wave
WFE Wave Forecast Error Statistics (GWS). Therefore the results presented in this
paper are based on area 90, although the method is also
ai, bi Distribution fitting parameters (-) applicable to all other GWS areas, depending on the
α, β, γ Weibull distribution parameters (-) amount of sailing time available.
d Water depth (m) The operational margin is defined as the
E Mean value (-) difference between the design wave height and the now
g Gravitational constant (m/s2) casted wave height along the route of the vessel.
Hmax Maximum wave height (m) Regarding the results shown in this paper, focus is given
Hs Significant wave height (m) to reducing the operational margin of the higher design
Number of observations within class wave heights.
number i (-)
Sum of total number of observations in 2. BACKGROUND INFORMATION
preceding class, plus one (-)
Total observations (-) Whereas the OBM is meant to replace the Dosuite
Cumulative probability (-) method for calculating the design sea states an overview
Ss Wave steepness (-) is given in [1], [2] on the methods used within Dockwise
Tz Zero-up crossing period (s) to calculate the design wave height for heavy marine
µ Mean of natural logarithm of zero transports.
crossing period (ln[s]) Various papers already include a detailed
σ Standard deviation of natural logarithm description of the Octopus system of which one can refer
of zero crossing period (ln[s]) to [3] and [4]. From the Octopus data a weather database
λ Wave length (m) has been built. In here, the operational insights and
Wave frequency (rad/s) conclusions of the motion monitoring program are
described. The Operational Based Method is based on the
data collected by the Octopus system. Detailed
information of the data can be found in [2], [3] and [5].

© 2014: The Royal Institution of Naval Architects


Marine Heavy Transport & Lift IV, 29-30 October, London, UK

Probability of occurrence [%]

Figure 1: Wave scatter diagrams Octopus vs. Dosuite 3.3 FROM SCATTER DIAGRAM TO DESIGN
WAVE HEIGHT
This paper focuses on how to implement this Octopus
weather data in the design process for heavy transport First, a short description of the determination of the
engineering. Dosuite design wave height is given. Hereafter, the
difference with the OBM design wave height
3. IMPLEMENTATION OF OCTOPUS DATA determination is explained.
Ref. [1] and [2] describe that Dosuite uses a
3.1 OCTOPUS SCATTER DIAGRAMS Gumbel cumulative probability distribution function for
the distribution of the wave height classes in the wave
The time records of the SPOS-NOW wave data, as scatter diagrams of GWS. In formula:
described in [5], are converted into wave scatter
diagrams. This conversion makes the comparison with
( )
the GWS scatter diagrams, as used in Dosuite, much √ (Eq.1)
easier. Furthermore, both methods use scatter diagrams
as basis.
In Figure 1 the Octopus wave scatter diagrams of area 90 The Gumbel formula was chosen because it should give
are visualized. For comparison, the yearly GWS wave the probability of exceedance of a certain wave height
scatter diagram is also added. The difference between the without ever reaching a probability of non-exceedance of
conventional design wave climate and the observed wave one.
climate is clearly visible, especially in wave height. In the design wave height calculation of Dosuite the
The Octopus wave scatter diagram confirms the probability of exceedance of the 3-hour design wave
conclusions in [2] and [5], in which it was obtained that height is calculated, given the season and transit time
the highest wave encountered is about 5.5 [m]. through the area. To account for calm periods, the
calculation is corrected for waves smaller than 4.0 [m].
3.2 IMPLEMENTING OCTOPUS SCATTER Together with Eq.1, the Dosuite method results
DIAGRAMS in a design wave height which is always lower than the
maximum observed wave height in the GWS diagram.
The Octopus data could be implemented in the design Whereas the new operational design method makes use
process as a new weather database, as indicated by the of the weather database built up from the sailing vessels,
dotted line in Figure 2. This is the most obvious location, it means it is entirely based on weather which is already
as the Octopus wave database is converted into wave encountered.
scatter diagrams and both Dosuite and OBM use scatter For a design method, it is often required to design for sea
diagrams as basis. Because the weather database is states which have a very small probability of encounter.
changed, both methods calculate different design sea If the Dosuite method would be used for obtaining the
states. Note that the calculation of the non-environmental design sea states and only the weather database is
part of the design values like RAOs, response spectrum, changed, the resulting design sea states will be
spectral moments, etc. will not be changed. unrealistically low.

© 2014: The Royal Institution of Naval Architects


Marine Heavy Transport & Lift IV, 29-30 October, London, UK

Standard types of joint distribution functions include,


amongst others, the Bivariate Log-Normal Distribution
[7], Fang and Hogben’s Distribution [8] and the Bivariate
Weibull Distribution [9]. But neither of them gave a
good fit to the wave data.
However, one distribution obtains a good fit to
the data. The particular distribution is the Marginal 3-
parameter Weibull distribution for the significant wave
height and a conditional log-normal distribution for the
zero-up crossing wave period. This distribution is
referred to as the LoNoWe method [6].
For evaluating the fit of the distributions, the
wave scatter diagram is converted into a logarithmic
contour plot. So, the logarithm is taken of the numbers in
the wave scatter diagram, before fitting the contour lines.
The advantage here is that more emphasis is given to the
rarer observations (higher sea states).
The used technique for comparing the data and
the fit are contour plots of the normalized deviations
between the data and the fitted distribution. Based on [6],
the Octopus data is described by the LoNoWe method for
the zero-up crossing wave period to describe the long
term wave data.

4.2 JPDF - OPERATIONAL BASED DESIGN


METHOD
Figure 2: Schematic overview of HMT design process
The question is now if the Octopus wave data can be
Therefore, it is required to extrapolate the wave height described by the LoNoWe method, as no ‘extreme’
data to obtain sea states with lower probability of waves are included in the Octopus wave scatter diagrams
occurrence. and the shape of the contour lines between the Octopus
wave scatter diagram and the GWS scatter diagram are
4. OPERATIONAL BASED METHOD much different.
This paragraph shows the mathematical description of
Whereas the wave scatter diagrams consists of two the LoNoWe method. Next paragraphs show the fitted
variables (wave height and wave period), two distributions to the significant wave height, the fitted
distributions are required in order to make extrapolation distributions to the zero-up crossing wave period and the
possible. The probability of a certain sea state in a wave resulting design sea states obtained from the Operational
scatter diagram can be described by a joint probability Based Method.
density function (JPDF).
First, these are analysed by DW of which several The LoNoWe method can be mathematically expressed
standard JPDFs are available. This is described in 4.1. as Eq.2, together with Eq.3 and Eq.4:
Secondly, paragraph 4.2 shows the mathematical
description of the JPDF as used in the OBM. Thirdly, the ( ( ) )
distribution of the wave height is investigated. Various [ { }]
JPDFs use the Weibull distribution for the wave height. √
As last one, the wave period distribution is dealt with.

4.1 JOINT PROBABILITY DENSITY ( ) { ( ) }


FUNCTIONS

In the past, research is performed to compare standard


types of (joint) distributions of the significant wave (Eq.2)
height and zero-up-crossing period for the long term
wave climate. Where:
In [6] it is stated that there is no theoretical reason why
any particular distribution function should fit the wave ( ) (Eq.3)
data best. The evaluation of the different distribution
functions is therefore done on empirical basis, with focus
( ) (Eq.4)
towards the more extreme sea states.

© 2014: The Royal Institution of Naval Architects


Marine Heavy Transport & Lift IV, 29-30 October, London, UK

And , are the parameters of the marginal Weibull


distribution.
Above formula will act as the basis for the new,
operational based design method to extrapolate the
Octopus wave data to the design sea states.

4.3 EXTRAPOLATION OF OCTOPUS WAVE


HEIGHT DATA

Extrapolation to the 10-year return period is used as the


Dosuite method is comparable with the 10-year return
values of SAFETRANS, which is generally accepted in
heavy transport engineering.
When the Weibull distribution is used for
extrapolating the Octopus wave data (wave height), some
remarkable differences appear when comparing the
Octopus wave data and the GWS wave data. The tail of
the Octopus wave height data clearly deviates to lower
probabilities of occurrences then calculated by the
Weibull distribution. This demonstrates the direct effect
of weather routing by the Dockwise masters.
The explanation is found in the fact that the
encountering of higher waves (up to 5.5 [m]) is actively
suppressed by means of human interaction, such that
these higher waves are rarely encountered.
For areas around the equator, the tail deviation
from the Weibull distribution is not seen. This is mainly Figure 4: Various tail distributions of Octopus wave data
due to absence of high waves in this region and as a
result weather routing is not used. This implies a non-reliable fit and therefore not
According to Figure 3 it is assumed that the Octopus favourable.
wave height can be described by a Weibull distribution, Also, it was assumed that the tail is exponentially
except the tail of the distribution. It is interesting to find distributed. It can be seen that assuming an exponential
a distribution solely for the tail, to extrapolate to higher tail of the distribution leads to an extrapolated wave
design wave heights and to even possibly find an height which is very close to the highest now casted
asymptote in the Octopus wave height. wave. This preliminary research shows that the
Several distributions have been tried by behaviour of the tail is very difficult to describe.
evaluating the fit and taking into account the tail A safer, but slightly more conservative,
deviation [10]. The tail is defined as waves higher than approach is to fit the wave data with a Weibull
4.5 [m]. For instance, the tail of the distribution was distribution with a lower Peak-over-Threshold (POT)
assumed Weibull distributed, with different Weibull value of 3.3 [m], or a probability of exceedance of 10%
parameters as found from the lower wave heights. The and lower. More data is thus taken into account which
assumed Weibull distributed upper tail will give the leads to less influence of the tail deviation. See Figure 4,
results obtained in Figure 4, top. In here, the 95% bottom.
confidence bounds are plotted where one can see that Keep in mind that, although the distribution of the tail is
these confidence bounds are widely spread. not specifically accounted for, the wave data contains the
effect of bad weather avoidance.
Next paragraph describes the distribution and
extrapolation of the Octopus wave period data in more
detail.

4.4 EXTRAPOLATION OF OCTOPUS WAVE


PERIOD DATA

The extrapolation of the wave period is done with a


conditional Log-Normal function, in which the term
conditional refers to a certain wave period per wave
Figure 3: Distribution of Octopus significant wave height height bin.

© 2014: The Royal Institution of Naval Architects


Marine Heavy Transport & Lift IV, 29-30 October, London, UK

Figure 5: Mean of ln(Tz) of Octopus and GWS wave data

Figure 6: Standard deviation of ln(Tz) of Octopus and The fitting to the mean of ln(Tz) gives a good fit for the
GWS wave data Octopus wave data. Some other functions have also been
tried to fit the data, but none of them provided a better
For each bin of the significant wave height, the fit.
parameters and of the conditional Log-Normal
distribution of the zero-up crossing wave period are 4.4 (b) Standard deviation of ln(Tz) as function of Hs
estimated.
The modelling of the parameters is done by empirical The result of fitting the standard deviations is shown in
regression functions. First, the fitting of the -parameter Figure 6 for the Octopus wave data and the GWS wave
is shown. data. The DNV recommended fit of the standard
deviation of ln(Tz) as a function of Hs is given by Eq.6.
4.4 (a) Mean value of ln(Tz) as function of Hs
( ) (Eq.6)
The method’s recommended fit of the mean value of
ln(Tz) as a function of Hs is given by Eq.3. The result of This formula is in good agreement with Eq.4 with the
the fitting with this formula for the Octopus wave data is original b1 parameter set at 0.07.
visible in Figure 5. For the Octopus wave data the dotted red line represents
For clarity, the fit to the GWS is also included. The the methods recommended fit. This represents in Eq.7.
fitting of the GWS data has been done with the DNV However, the scatter of the Octopus standard
recommended practice formula: deviation data is bigger than the scatter of the mean
ln(Tz) data. Therefore, it is more complicated to fit this
( ) (Eq.5) data.
Several distributions have been fitted, which are listed
Eq.5 is a simplification of Eq.3 because the original below (for clarity, the recommended fit is also included
parameter is set to 0.70 for all areas. in the form of Eq.7):

© 2014: The Royal Institution of Naval Architects


Marine Heavy Transport & Lift IV, 29-30 October, London, UK

( ) (Eq.7) (Eq.13)

( ) (Eq.8)
(onl fitted for m) Eq.13 is valid for non-significant wave heights.
Throughout this paper, the significant wave height is
( ) (Eq.9) used. The wave steepness is then written as:

(Eq.14)
( ) (Eq.10)

As can be seen in Figure 6, the -distribution will


predict a standard deviation of zero of most of the The limiting values for is for and
extremer Octopus wave data. It is expected that the for . Between the boundaries linear interpolation
resulting contour lines of the JPDF, above the wave is applied.
height where the standard deviation is zero, are absent. With use of Eq.14 the physical limitation of the waves is
For a design method, this is not favourable. Therefore, in visualized in the wave scatter diagrams by the dash-dot
the fourth distribution, denoted , the first term of Eq.7 line in Figure 7.
is removed. The result will be that at the higher The upper bound is determined by the contour line,
significant wave heights (say, Hs > 6 meters), the calculated by the LoNoWe method.
standard deviation is still nonzero.
The second distribution, , is more 4.6 OBM DESIGN SEA STATES
conservative as it will result in a higher standard
deviation at the higher significant wave heights. Finally, A model is built to calculate the sea states, following
the distribution is a compromise between the from the contour lines of the JPDF. The model is able to
conservative fit and the fit. Whereas the fit is handle all POT values of the significant wave height. For
slightly more conservative than the fit, the fit is the standard deviation of the wave period per significant
used as the base case. wave height, three distributions are pre-fitted ( ).
Now both the marginal Weibull distribution for The model is able to calculate both the yearly
the significant wave height and the conditional and seasonal design sea states. The output of the model
distribution for the zero-up-crossing period are known are the resulting JPDF contour lines, together with the
and the contour lines can be visualized according to Eq.2. used POT value of the Weibull distribution for the
significant wave height as well as the used distributions
4.5 UPPER AND LOWER BOUNDS OF WAVE for the wave period.
PERIOD The contour lines of the Octopus wave data
provided by the JPDF result in the design sea states of
For each period, the wave height has a maximum value. the Operational Based Method (OBM). This paragraph
This is because a wave has some physical limitations like shows these design sea states.
the maximum wave steepness. If the wave steepness The base case of this method is a POT
becomes too large, the waves start to break. This percentage of 10% for the Weibull fit for the significant
determines the lower bound of the wave period. The
maximum wave height, Hmax, for a periodic wave is
determined by the fact that the particle velocity in the
crest cannot be larger than the forward speed of the
wave. Hmax can be approximated by:

(Eq.11)
( )

For deep water ( ), above Eq.11 reduces to Eq.12:

(Eq.12)

In which is defined as the wave steepness .


The wave period and the wavelength are related via the
dispersion relation ( ( )), in which
. Rewriting the dispersion relation, for deep water,
gives an expression for the wavelength: Figure 7: Yearly OBM design sea states

© 2014: The Royal Institution of Naval Architects


Marine Heavy Transport & Lift IV, 29-30 October, London, UK

wave height, together with the standard deviation


formula as given in Eq.9.
The nomenclature used for the base case is ’10-
3’, as a reference to the POT 10% fit for the significant
wave height and the standard deviation fit number three
for the wave period.
The yearly JPDF contour lines are shown in Figure 7.
Sensitivity of the results is obtained in paragraph 6.
Whereas the waves stored in the database are
now casted waves, it could be that waves were lower or
higher in reality. To account for this limitation, a wave
forecast error (WFE) is introduced. Currently, the WFE
is set at 1 [m] for the wave height and 1 [s] for the wave
period. Herewith, the design sea states for area 90 are
calculated as the squares on the solid line in Figure 7.

5. COMPARISON OBM WITH DOSUITE


Figure 9: OBM and Dosuite design wave heights,
5.1 DESIGN SEA STATES OBM VS DOSUITE
including WFE
To place the OBM design sea states in perspective,
differences in design wave heights are expected.
Figure 8 shows both the yearly OBM design sea states as
well as the Dosuite design sea states. The latter
5.2 SEASONAL DESIGN WAVE HEIGHTS OBM
represents the highest design sea state being obtained
VS DOSUITE
from the observed Cargo Securing Manuals (CSM),
namely a design sea state with a significant wave height
The result of comparing the Octopus seasonal design
of 10.17 [m] and zero-crossing wave periods of 8.70,
wave height with the Dosuite design wave height (both
9.70, 10.70 and 11.70 [s]. Depending on the season, the
weather routing included and excluded) is obtained in
vessel characteristics and the vessel route, Figure 8 could
Figure 9. For the Dosuite design wave height, it is
vary.
assumed that the transit time in area 90 (CoGH) is 61
One must know that the results being obtained
hours, which is about 2.5 days, this is a common transit
in this chapter are compared to particular results of the
time for most voyages. Figure 9 shows that a significant
Dosuite method only. Solely voyages with non-reduced
difference in design wave height is obtained when the
wave heights (so no sea state relaxation and no short trip
Octopus wave database is used. It is seen that the new
scenario) were used. The reason is that this leads to a
operational design method has its largest reduction in
clear reduction in design wave height. When the OBM
design wave heights for voyages sailed in winter time
method is compared to voyages which are calculated
and the smallest reduction in design wave heights for
with sea state relaxation in Dosuite, no significant
voyages sailed in autumn.

Season Season Database coverage


[%]
Autumn Mar-May 39.8%
Winter Jun-Aug 20.3 %
Spring Sep-Nov 17.9 %
Summer Dec-Feb 22.0 %

Table 1: Seasonal database coverage area 90

From the seasonal Octopus wave scatter diagrams, it is


seen that the Dockwise vessels not equally spend time in
each season. Table 1 shows the amount of seasonal data
stored in the database for area 90.
The 10-year return design wave height of the winter
season is 4.84 [m] and the 10-year return design wave
height of the summer season is 4.63 [m]. When
comparing the maximum now casted wave height during
Figure 8: Yearly OBM design sea states vs. Dosuite the voyage with the design wave height of the OBM, it
design sea states becomes clear that the design wave height (without

© 2014: The Royal Institution of Naval Architects


Marine Heavy Transport & Lift IV, 29-30 October, London, UK

WFE) is close to the maximum now casted wave height As can be seen in Table 2 and 3, both the
for above mentioned seasons. transverse- and vertical design accelerations differ only
Because the trend in severity of the seasonal scatter (maximum) 1-2% when comparing the 50% POT value
diagrams of area 90, compared to the GWS scatter with the 10% POT value.
diagrams of area 90, is not yet well understood and the
yearly scatter diagrams contain more data it is advised to 6.2 FITTED STANDARD DEVIATION
use the yearly OBM for doing design calculations for
voyages that will pass the CoGH. Especially the fitted standard deviation has a big
influence on the resulting contour lines of the Octopus
6. SENSITIVITY ANALYSIS wave data, as can be seen in Figure 10. The different
design sea states obtained from the 10-year return
Within the development of the new design method, contour lines are visualized in here.
several assumptions have been made for the OBM. For
instance, the significant wave height is fitted with a POT
value of 10%. But different POT values could result in
different design wave heights. Furthermore, the standard
deviation of the wave period could be fitted in various
ways. Finally, the sensitivity of the wave forecast error is
shown.
To better show the influence of above mentioned
variables relative to the assumed base case, the design
accelerations of 22 voyages are calculated from the
design wave heights.

6.1 POT VALUE WEIBULL DISTRIBUTION

This paragraph shows the difference in design


acceleration when the POT value of the Weibull
distribution to the significant wave height is changed.
The POT 50% fit will be compared with the POT 10%
fit. Above 50% the data is assumed not to be Weibull
distributed. Below 10%, some seasons showed no Figure 10: Variation in design sea states with different
convergence while fitting. distribution fittings of the standard deviation
Difference The design accelerations are calculated for the transverse
[50-3] vs [10-3] direction as well as the vertical direction for the yearly
Minimum 0.31% wave data. For each voyage the design accelerations are
Mean 0.59% calculated according to the design sea states, which in
Maximum 0.89% turn are calculated according different underlying
distributions of the standard deviation of the wave
Table 2: Sensitivity of transverse OBM design periods.
accelerations, [50-3] vs [10-3] The results obtained are calculated according to
the base case, which includes the standard deviation
number ‘3’ (Eq.9). Therefore, the design accelerations
Difference obtained from the standard deviation number ‘2’ (Eq.8)
[50-3] vs [10-3] and ‘4’ (Eq.10) are compared with this base case.
Minimum -1.23% The vertical design accelerations are generally
Mean -0.73% within 2% of each other, as can be seen in Table 5. The
Maximum 1.23% main reason for this is that the natural heave period is
approximately 10 seconds for the Dockwise vessels.
Table 3: Sensitivity of vertical OBM design Figure 10 shows that the design contours lie almost on
accelerations, [50-3] vs [10-3] top of each other around this wave period. The wave
spectra constructed from these design sea states are
One has to recall Figure 3 to obtain the fitted Weibull therefore almost similar.
distribution. For the transverse design accelerations the
The obtained difference in design acceleration is quite differences are generally less than ± 5%, see Table 4. For
trivial when one is comparing the observed design wave two voyages the transverse differences are up to +15%.
heights. Minor differences are caused by the different tail These large differences occur due to the fact that the
behaviour, when compared to the Weibull distribution. periods of the design sea states will be higher when the
The tail gains influence when the POT value is lower.

© 2014: The Royal Institution of Naval Architects


Marine Heavy Transport & Lift IV, 29-30 October, London, UK

standard deviation fit number ‘2’ is used. Likewise, the Difference


wave spectrum is constructed with a larger period. [10-3-WFE] vs [10-3]
Because 2 voyages have their natural roll period Minimum 10.09%
around 24 seconds, design wave periods of the standard Mean 14.12%
deviation number ‘2’ fit will be more close to the natural Maximum 26.64%
roll period of the vessel which will give an increase in
the response of the vessel. For vessels which have its
natural roll period lower than 20 seconds, or around 27 Table 6: Sensitivity of transverse OBM design
seconds or more, almost no difference is obtained when accelerations, [10-3-WFE] vs [10-3]
comparing the wave spectra.
Although the difference might be significant, it Difference
must be noted that the contour lines obtained by using [10-3-WFE] vs [10-3]
the standard deviation number ‘2’ are conservative. Minimum 16.61%
Tucker [11] already stated that the variation in wave Mean 19.22%
period narrows down when the sea heights get more Maximum 20.19%
extreme. This theory is supported in Figure 5, where it is
shown that the wave periods indeed will only slightly
increase when the wave height is increasing. Table 7: Sensitivity of vertical OBM design
accelerations, [10-3-WFE] vs [10-3]

Difference Difference
[10-2] vs [10-3] [10-4] vs [10-3] 7. CONCLUSIONS & RECOMMENDATIONS
Minimum 0.00% -4.67%
Mean 2.66% -1.44% 7.1 CONCLUSIONS
Maximum 14.96% 0.00%
An operational based design method, which takes the
Table 4: Sensitivity of transverse OBM design avoidance of bad weather into account, is developed and
accelerations, [10-2] and [10-4] vs [10-3] analysed through modelling, calculations and
comparisons with measurements. The following
conclusions are drawn on basis of the results and are
Difference Difference valid for all DW vessels and for GWS area 90 only:
[10-2] vs [10-3] [10-4] vs [10-3]
Minimum 0.60% -1.90%  For the higher design wave heights the operational
Mean 1.11% -0.83% margin, defined as difference between the design
Maximum 1.88% 0.00% wave height and the now casted wave height along
the route of the vessel, is decreased.
Table 5: Sensitivity of vertical OBM design
accelerations, [10-2] and [10-4] vs [10-3]  The significant wave height of the Octopus wave
data is Weibull distributed, except the tail of the
6.3 WAVE FORECAST ERROR distribution. The seasons in which the highest
waves have been encountered show a clear
To account for the WFE, it is assumed that both the deviation from the assumed Weibull distributed
maximum error of 1 [m] in wave height and 1 [s] in wave significant wave height data towards lower
period will occur at the same time. probabilities of exceedance. This is a direct effect of
A sensitivity analysis is performed when the wave the weather routing capacity of the Dockwise
forecast error is taken into account. The difference in captains.
yearly transverse design acceleration when compared to
the base case (POT value 10% for the Weibull  When the waves will grow in amplitude, the wave
distribution of the significant wave height, standard periods will hardly increase. The relation between
deviation fit number ‘3’ for the wave period) is the mean, taken of the natural logarithm of the wave
calculated. Tables 6 and 7 shows these differences. period per wave height, and the significant wave
Clearly, the wave forecast error has a large influence on height is very well described by the
the resulting design accelerations. function ( ) . This function
Recently a new research is started to investigate the local shows only a minor increase in wave period when a
wave forecast error and to account for the error in the vessel will encounter higher waves.
design program.

© 2014: The Royal Institution of Naval Architects


Marine Heavy Transport & Lift IV, 29-30 October, London, UK

7.2 RECOMMENDATIONS 4. L.J.M. Adegeest, ‘Response based Weather-


Routing and Operation Planning of Heavy
 To increase the amount of data used when doing the Transport Vessels’, RINA Marine Heav
analysis it is advised to use the yearly OBM design Transport & Lift II, London, UK, p123-131,
acceleration when doing the future design 2008.
calculations for voyages around the Cape of Good
Hope. 5. J.B. de Jonge and O.A.J. Peters, ‘Operational
margin from weather and motion database for
 In order to increase the applicability of the heav transport vessels’, OMAE2011-49811,
Operational Based Method, more areas have to be 2012.
investigated and included in the OBM. Other severe
areas are for instance area 16 (East-Atlantic Ocean), 6. Mathisen J., Bitner-Gregersen E., ‘Joint
area 85 (West of CoGH) and area 11 (North Sea). distributions for significant wave height and
This research includes only area 90. In this manner, wave zero-up-crossing period’, Appl Ocean Res,
the OBM could also be used for voyages which will 1990, Vol. 12, No. 2
not pass the Cape of Good Hope.
7. Ochi M.K., ‘Wave statistics for the Design of
 In this research, mainly the effect of reducing the Ships and Ocean Structures’, Trans. SNAME,
design accelerations has been investigated, and Vol.86, 1978
thereby reducing the operational margin. Only
preliminary research has been done in increasing 8. Fang Z.S., Hogben N., ‘Anal sis and Prediction
the design accelerations to increase the operational of Long Term Probability Distribution of Wave
margin. This is especially the case for voyages Heights and Periods’, Technical Report of the
which will not pass the Cape of Good Hope and National Maritime Institute, London, 1982
will have a design wave height of about 4.5 meters.
More research should be done into achieving 9. S., A Course in Ocean Engineering, Veritas
sufficient operational margin for all voyages. Research Report 85-2028, HØvik, 1985

 The wave forecast error which is currently included 10. M. Deelen, ‘An operational based method for
is a conservative assumption. Sensitivity analysis determining the design sea state of heavy
shows that this value has a large influence in the transport vessels’, MSc thesis, Delft University
resulting design wave height. Furthermore, the of Technology, 24-05-2013.
WFE might be different when only a specific area is
investigated. Knowing the error per area will 11. Tucker M.J., ‘Waves in ocean engineering,
improve the accuracy of data. measurement, anal sis, interpretation’. London:
Ellis, Horwoord, 1991.
 The amount of data present in the database near
area 90, the Cape of Good Hope, includes about one 9. AUTHORS BIOGRAPHY
year of continuous measurements. To increase the
reliability of the data, it is advised to keep Michael Deelen holds the current position of transport
continuously monitoring the wave environment and engineer at Dockwise. He is responsible for support to
ship motions. answers and alternatives on engineering issues of clients
and Dockwise projects. This will include feasibility
8. REFERENCES studies, design of solutions and project support in the
execution phase.
1. DOCKWISE GC.ENG101, ‘Routing and Design
Environmental Conditions’, Guidelines and Jan de Jonge works as a senior marine engineer at
Criteria, November 2008. Dockwise and is involved in all kind of hydromechanics
topics like ship motion behaviour, float over analysis,
2. M.J.A. van Exsel and J.B. de Jonge, mooring analysis, environment and model testing. In
‘Developments in heav transport design addition daily work involves leading a fleet wise motion
calculations’, Marine Heav Transport & Lift monitoring project, processing and analysing of the data.
III, 2012.

3. O.A.J. Peters and L.J.M. Adegeest, ‘Motion


Monitoring and Decision Support during Heavy
Transport’, OMAE2010-21143, 2010.

© 2014: The Royal Institution of Naval Architects

Das könnte Ihnen auch gefallen