Você está na página 1de 12

Safety and Reliability for Managing Risk Guedes Soares & Zio (eds)

2006 Taylor & Francis Group, London, ISBN 0-415-41620-5

Longitudinal strength reliability of a tanker accidentally grounded


R.M. Lus, A.P. Teixeira & C. Guedes Soares
Unit of Marine Technology and Engineering, Technical University of Lisbon, Instituto Superior Tcnico,
Lisboa, Portugal

ABSTRACT: This paper presents a reliability analysis of a damaged Suezmax double hull tanker, using a first
order reliability method. The accidental grounding being analyzed is assumed to be centered on the keel, which
is the worst possible scenario. The ultimate strength of the damaged tanker, to be used in the formulation, is
calculated by means of a specific structural code, in which the damage is simulated by the removal of the damaged
elements. Different sizes of the damage are analyzed and a relationship is established with the reliability index
related to the ultimate strength. The wave induced loads are calculated from long-term distributions related to
Global Wave Statistics. The still water loads are defined from the ship loading manual. A parametric study about
the impact of the ultimate bending moment of the ship on the reliability index is then performed.

INTRODUCTION

The safety of ships at sea has always been an important aspect of ship operations. Accidental damages of
ships can occur in any number of ways being the two
most concerning ones the collision with other ships
and grounding on rocky seabed.
Many studies about the impact of damage in the ship
ultimate strength have been made. Paik et al. (1998)
developed a fast method to assessing the collapse of
the hull girder in the damaged condition using the formulation of the American Bureau of Shipping. Gordo
& Guedes Soares (2000), have studied the ability of
simplified methods for the calculation of the vertical ultimate bending moment to predict the ultimate
longitudinal strength of damaged ships. But, so far
few authors have published studies about the reliability
assessment of damaged ships (Fang & Das, 2004a).
Therefore it is the objective of this study to present
a possible methodology to calculate the reliability of
the damage ship.
For the method to be fast and useful, during ship
operations, it becomes necessary to establish a relationship between the ship reliability this and the effects
of the damage in the ultimate bending moment of the
hull girder. And, it is also necessary that the loads of
the ship can be taken from the known load conditions
of the ship in the intact condition.
Three load conditions (LC) where considered for
this study full load (FL), ballast load (BL) and partial load (PL), they are the most common ones. The
lightship condition should also have been considered,
because the ship can be towed, from the accident

location to dry-dock, in this condition. This would


allow analyzing the advantages or disadvantages of
the possible transfer of cargo before the voyage. But,
the lack of information about this condition made
it impossible to perform calculations for the wave
loading.

DAMAGE & SHIP CHARACTERISTICS

2.1 Ship characteristics


The ship analyzed here is a Suezmax double hull tanker
that fulfills all modern rules of the classification societies for this vessel type. The main particulars of the
vessel are presented in Table 1. The operational profile
of the ship is presented in Table 2.
Using a simplified approach it was considered
that each of the reinforcements on the ship has an
associated plate, therefore, the ship section can be
divided into separate elements, composed by one profile and the surrounding plate. The aspect of the
section, divided in reinforced plate elements, is presented in Figure 1 the bilge elements do not have
reinforcements.

1499

Table 1.

Ship particulars.

Length PP (m)
Breadth (m)
Depth (m)
Draught (m)

270.0
48.2
23.0
17.1

Table 2.

Ship operational profile.

Loading condition
Fraction of ship life
Voyage duration
(days)

Harbor
15%

FL
35%
23.5

BL
35%
23.5

PL
35%
2.0

Figure 2. Damage dimensions.

Figure 1. Ship section divided in reinforced plate elements


(intact section).

2.2 Damage characteristics


The damage considered for this paper is accidental
grounding around the keel area, which is believed to
be the worst possible scenario. Considering that the
ship section is composed of several elements than the
elements around the keel within the damaged area are
considered damaged.
This paper, adopted the simplified approach to deal
with the damage that is recommended by most classification societies. Its basic assumption is that the
strength of the damaged section can be found by simply
removing the damaged elements from the intact section and recalculating its ultimate bending moment.
This is a conservative approach since some of the
damaged elements might still give some contribution for the strength. However the possibility of crack
propagation along the ship hull is disregarded, that
is an important field of study for a damaged ship,
but it is outside the parameters set for this particular
study.
The definition of the extension of the damage used
in this paper is based on the ABS (1995) recommendations. But, while these assume that the penetration
of the damage can be reduced near its transversal
ends, in this study the penetration is assumed constant, like is shown in Figure 2. The damaged area is
therefore rectangular with a height of 3/4H (where
H is the double bottom height) while the height
for major damage is H . The width is equal to B/6
(were B is the breadth), which was varied 20%. The
aspect of the sections after the elements have been
removed is presented in Figure 3 for both types of
penetration.

Figure 3. Damaged hull sections (top: damage; bottom:


major damage).

ULTIMATE STRENGTH

The ultimate strength of the ship was calculated


using the HULLCOLL program that uses the method
developed in Gordo et al. (1996).
For the calculations the net scantlings were used,
like advised by the Classification Societies. In any case
the use of the corrosion thickness should never be used
in the case of accident, because there is no idea, at the
time of the accident, of the condition of the hull.
Considering the damage defined before one has
seven different conditions with different ultimate
bending moment capacities, which are presented in
Table 3.

1500

Table 3.

Ultimate vertical bending moment.

Intact
Damage 20%
Damage
Damage +20%
Major Damage 20%
Major Damage
Major Damage +20%

Table 4.

Sagging
(MNm)

Hogging
(MNm)

10400
10240
10210
10180
10220
10180
10090

12740
12190
12080
11910
11780
11510
11210

Residual strength indexes.

Intact
Damage 20%
Damage
Damage +20%
Major Damage 20%
Major Damage
Major Damage +20%

Sagging

Hogging

1.000
0.985
0.982
0.979
0.983
0.979
0.970

1.000
0.957
0.948
0.935
0.925
0.903
0.880

Figure 4. Diagram of the RIF in function of the width of


both the damage (GCD) and of the major damage (GCMD)
in sagging.

With the ultimate collapse defined one can use the


residual strength index (RIF) as defined by Fang &
Das (2004b) and presented in Equation 1.

where Mu,dam is the ultimate moment of the damaged


section and Mu,int is the ultimate moment of the intact
section.
This index is useful for the parametric study of
the reliability index (). The RIF obtained for each
damaged section is presented in Table 4.
There is a higher loss of strength in hogging than
in sagging, as the damage dimensions increase, this
in spite of the ultimate bending moment being bigger
for hogging. This is due to the position of the damage.
The elements removed from the bottom of the ship lead
to an increase in the height of the neutral axis, which
increases the strength of the ship in sagging (opposite
effect of the loss of area), but this same effect leads to
a decrease of the strength in hogging. This is so much
so that, in sagging, with Major Damage the ship has
a RIF equal to the one that has with Damage +20%.
Figures 4 and 5 present a diagram of the RIF values for sagging and hogging, respectively. While in
hogging the variation of the ultimate strength remains
mainly linear. In sagging, for deeper penetration, a
non-linear variation can be observed.
The uncertainty in the ultimate strength calculations
( u ) to use is based on the background work for the
joint tanker project (ABS et al., 2005) it has a normal

Figure 5. Diagram of the RIF in function of the width of


both the damage (GCD) and of the major damage (GCMD)
in hogging.

distribution with mean 1.05 and standard deviation of


0.1. This uncertainty includes all uncertainties in the
model, even from the material properties.

4 APPLIED LOADS
4.1 Still-water load Effects
The still water load effects were calculated for the
intact section only. The effects of the damage on
the load distribution are accounted for by means of
the load coefficient of the still water bending moment
(Kus ).
The still-water bending moment is assumed to vary
monotonically from departure to arrival, according to a
uniform distribution. Guedes Soares et al. (1996) proposed a Gaussian distribution for the still-water load

1501

model. Soares & Dogliani (2000) come up with a way


to calculate the parameters for the Gaussian distribution of the still-water moments from the ship loading
manual. The approach followed in this study is the one
presented in ABS et al. (2005), where the distribution
of the still water loads in one voyage is a Gaussian with
mean (sw ) equal to 0.7 times the maximum load and
a standard deviation ( sw ) of 0.2 times the maximum
load. The maximum load is defined as the maximum,
from the load manual, of the voyages in a given load
condition c.
Guedes Soares & Moan (1988) performed a statistical analysis where the distribution of the load extremes
for a given time period (Tc ) that the ship spends in
a given load condition can be obtained by assuming
independence between the voyages, such relationship
is shown in Equation 2.

Table 5.

Still-water distribution parameters for hogging.


Gaussian (1 week)

Gumbel (1 year)

LC

sw

sw

se

se

FL
BL
PL

55.1
2709.0
2452.8

15.7
774.0
700.8

75.6
3714.9
3893.7

14.0
686.5
410.3

Table 6.

Still-water distribution parameters for sagging.


Gaussian (1 week)

Gumbel (1 year)

LC

sw

sw

se

se

FL
BL
PL

1834.1
96.3
1593.5

524.0
27.5
455.3

2515.1
132.1
2529.5

464.9
24.4
266.5

where nc is the number of voyages in time period Tc .


For one year, according to the operational profile
presented in Table 2, the number of voyages is 5.4 for
FL and BL and 27.4 for PL.
The distribution of the extreme values over Tc , can
then be approximated as a Gumbel law with parameters
xe * and , given by Equations 3 and 4 (Teixeira &
Guedes Soares, 1998), respectively.

where Fs is the cumulative probability function of the


Gaussian distribution and fs is its probability density
function.
The mean value of the Gumbel distribution (se )
and the standard deviation ( se ) were calculated using
Equations 5 and 6.

In the case of a damaged ship it makes sense to


analyze the reliability for a single voyage, from the
moment of the accident to dry-dock, in this study such
voyage was considered to last one week. In this scenario, from Equation 2, the distribution of the extremes
is equal to the distribution for one voyage. So, for
the reliability calculations considering 1 week the
Gaussian distribution is used instead of the Gumbel
distribution.
Table 5 and Table 6 present the values for the stillwater bending moments distribution in hogging and

Figure 6. Global wave statistics ocean areas for the North


Atlantic.

sagging, respectively, the negative values represent


opposite loading (i.e. in the hogging situation the negative values imply sagging loadings and vice-versa).
The uncertainty, in the modeling of the still-water
bending moment ( sw ), is based on ABS et al. (2005)
it is assumed to have a Normal distribution with mean
1.00 and standard deviation 0.1.

4.2 Wave induced load effects


The wave induced load effects depend of the sea zone
where the ship is operating. Two different ocean zones
were considered in this study, the North Atlantic sea
areas 8, 9 and 10 (ATLN Fig. 6), as recommended
by IACS (2000), and the European coastal areas 27,
28 and 30 (ECA, author designation Fig. 7), which
is a more logical zone for an accidental grounding to
occur. The calculations are made using the scatter diagrams covering these areas given by the Global Wave
Statistics (Hogben et al., 1986).

1502

Table 7. Stochastic model of extreme wave induced load for


the North Atlantic.
FL
Weibull
w
336.5
k
0.912
Gumbel 1 year
n
1.57E+06
mean
6199.1
St. Dev.
476.1
Figure 7. Global wave statistics coastal areas of Europe.
Full load

FL

Weibull fit

Q(x)

1.0E-03
1.0E-05
1.0E-07
1.0E-09
1.0E-11
0.00E+00

3.00E+03

6.00E+03

9.00E+03

PL

325.6
0.892

320.1
0.874

1.57E+06
6410.9
503.7

6.76E+06
6258.6
533.7

Table 8. Stochastic model of extreme wave induced load for


the coast of Europe.

1.0E+01
1.0E-01

BL

1.20E+04

VBM (MNm)

Figure 8. Weibull fit for the full load condition in the ECA.

Weibull
w
406.7
k
0.947
Gumbel 1 year
n
1.57E+06
mean
6727.4
St. Dev.
497.6
Gumbel 1 week
n
8.64E+04
mean
5291.0
St. Dev.
491.3

BL

PL

380.3
0.918

377.6
0.896

1.57E+06
6888.4
526.0

6.76E+06
6858.5
570.4

8.64E+04
5375.8
515.4

8.64E+04
5696.2
559.5

The probability distribution in the given areas at a


random point in time is fitted by a Weibull model like
presented in Equation 7.

where w is the scale parameter and k is the shape


parameter.
In Figure 8 is presented an example of the fit of the
Weibull distribution for the ship in full load condition
in the ECA.
This is however not enough, because what one is
interested in, is the distribution of the maximum amplitudes of the wave (because they induce the highest
loads) in n cycles during the ship voyage. According
to Guedes Soares (1984) the distribution of extreme
values in time T can be obtained as a Gumbel law,
presented in Equation 8.

Where xn and are the parameters of the Gumbel


distribution.
The parameters of the Gumbel distribution can be
calculated from the Weibull parameters and from the
mean number of cycles n in the time T that the voyage
takes, by means of Equations 9 and 10.

In the North Atlantic the average wave period is


7 seconds. With this in mind and considering two
possible situations for the damaged ship, one year
of operations and a voyage of one week to dry-dock
(a more logical approach), the stochastic models of
extreme wave induced bending moments are calculated. The resulting values are presented in Tables 7
and 8 respectively.
In wave load bending moments there are two types
of uncertainty. The first is with respect to the linear
results (w , Guedes Soares & Moan, 1991), the second
is with respect to the non-linear effects (nl , Guedes
Soares, 1996). The models adopted for the uncertainties in the still-water bending moment are the same as
used in ABS et al. (2005) they are both assumed to have
a Normal distribution with mean 1.00 and standard
deviation 0.1.
4.3 Load combination
The reliability assessment depends on the the combination of both loads, which is usually less than the
sum of the two maximum that can occur at any time.
For the intact ship there are many previous studies like

1503

the one of Guedes Soares (1992). However this study


covers damaged sections, which may lead to different
combinations. Equation 11 presents the method used
to combine the loads.

where Kus is the still-water load combination coefficient and Kuw is the wave load combination coefficient.
Regarding the still-water load coefficient, this is
usually considered equal to 1.0 for the intact ship
(Guedes Soares & Teixeira, 2000). The ABS (1995)
guide for damaged ships recommends that for hogging
the value used should be 1.1, while for sagging should
be 0.9. However the difference between the still-water
loads in the intact and damaged conditions is not so
clear. The ship gains weight (loads sea water) while
losing buoyancy. Depending on the longitudinal position and extension of the damage and of the initial load
condition then, one cannot implicitly say if it increases
or decreases. Also, if one considers that the initial load
condition might be opposite to the resistant bending
moment (e.g. if one is studying the ability of the ship to
resist in sagging, but the still-water load being induced
is opposite, then decreasing it will reduce the ability of
the ship to resist the loading) than there is subject for
an entire new study. In this study in order to simplify
matters only the increase of the load is considered in
both hogging and sagging. Three values were used, 1.0,
1.1 and 1.5, the first was used to calculate the reliability for one year operation of the intact ship, while the
two others are used to analyze the impact of still-water
load on the reliability.
The wave load coefficient depends on the load
condition of the ship. However, Teixeira & Guedes
Soares (2005) and Guedes Soares & Parunov (2006)
indicate that, systematic calculations have shown that
the values of the load combination do not change
significantly for different amplitudes of the load variables within the same operational profiles. So for the
wave load combination coefficient the values adopted,
depend only of the load condition and are the following, 0.92 for full load, 0.91 for the ballast load and 0.80
for the partial load. These values are used in all analysis, but also used was the value 0.5 recommended in
ABS (1995), which is applied to one year of operations
in the North Atlantic, it should give an approximation
to the loading in the damage condition. The reliability
index obtained with this last value will be compared to
the ones obtained for a voyage of one week in coastal
areas.
5

RELIABILITY ANALYSIS

5.1 Formulation
The reliability formulation adopted for this study
is based on the approaches found in design codes.

Classification Societies have adopted limit state functions with variables having their nominal values based
on the ship characteristics. However in this study, while
following the same basic concept for the limit state
function, random variables are used.
The limit state function for the failure of the hull
girder under vertical bending moment used in this
study is presented in Equation 12.

where u , nl and w are the modeling uncertainties


of the ultimate strength, non-linear wave load effects
and linear wave load effects, respectively, Kus and Kuw
are the load combination coefficients for the still-water
and wave loads, respectively, Mu is the ultimate vertical
bending moment, Ms is the still-water vertical bending
moment, Mw is the wave bending moment and RIF is
the reliability index.
The reliability index depends of the variables which,
in turn, depend on several assumptions made for the
conditions in which the ship operates. These are presented in Table 9, and a short description of each
condition follows:
(1) ATLN1 This is the conventional process to
calculate the reliability index, considering ship
operations for one year in ATLN zone.
(2) ATLN2 For this condition it is assumed that the
still water bending moment increases because of
the damage. By comparing ATLN1 to this one it
is possible to analyze the impact of the still-water
load on the reliability index.
(3) Rules This condition follows the load combination presented in ABS (1995), only the worst
possible scenario was used: increased still-water
bending moment.
(4) ECA1 This condition considers that the ship
operates in coastal areas (this is a more likely scenario for a damaged ship) for one year. Comparing
this with ATLN2 it allows to analyze the influence
of the area being considered in the reliability index.
(5) ECA2 In this condition it was considered that
the ship makes a single voyage, lasting one week,
Table 9. Conditions in which the reliability index is
calculated.
Kuw

Name

Sea
Zone

Time

Kus

FL

BL

PL

ATLN1
ATLN2
Rules
ECA1
ECA2
ECA3

ATLN
ATLN
ATLN
ECA
ECA
ECA

1 year
1 year
1 year
1 year
1 week*
1 week*

1.0
1.1
1.0
1.1
1.1
1.5

0.92
0.92
0.50
0.92
0.92
0.92

0.91
0.91
0.50
0.91
0.91
0.91

0.80
0.80
0.50
0.80
0.80
0.80

1504

Only 1 voyage is done during this time.

Assuming that the three load conditions are independent and mutually exclusive, the global reliability
index can be found, for one year of operations, by
Equation 13.

where 1 is the inverse of the standard normal probability distribution function and PF BL , PF FL and PF PL
are the probability of failure in the different load
conditions.
For one week the global reliability index has no
meaning because there is only one voyage at a given
load condition.The use of the RIF allows establishing a
relationship between the damage sustained by the ship
and the reliability, = f (RIF), and, consequently, with
the probability of failure, PF = f (RIF). From these
relationships is then possible to extract the reliability
for a given damage.

should remember that the difference between ATLN1


and ATLN2 is the still-water bending moment, which
is higher in the second case. The difference between
ATLN2 and ECA1 is the sea zone being considered,
North Atlantic and coast of Europe, respectively, the
wave load bending moments of the second are higher
than of the first. It is clear from the figures that in
hogging the still-water bending moment reduce the
reliability as much as the wave loads. In sagging the
wave loads show a higher impact on the reduction of
the reliability.
It can also be seen that, while in general the
reliability in sagging is smaller than the values for
hogging, the smallest value is obtained for hogging
ATLN1

ATLN2

ECA1

2.30
2.10
1.90
1.70
1.50

probably to dry-dock for repairs. This is the most


likely scenario for a damaged ship.
(6) ECA3 This condition is similar to the previous
one but has a greater still-water load for analyzing
its impact in the reliability of the damaged ship in
this type of conditions.

1.30
1.10
0.90

5.2 Results of the reliability calculations

3.0E-02

2.70

2.5E-02

2.50

1.10

2.0E-02

2.30

0.90

1.5E-02

2.10

1.0E-02

1.90

5.0E-03

1.70

GCMD+20%

GCMD

GCMD-20%

Figure 9. Reliability index () and the probability of failure


(PF) vs. the RIF, for ECA2 conditions in hogging.

GCD

GCMD

1.50
1.00

RIF

Intact

0.50

GCMD+20%

0.98

0.70
GCMD-20%

0.95

1.30

GCD+20%

0.93

1.50

GCD-20%

0.90

GCD+20%

1.70

2.90

0.88

ECA1

1.90

Reliability index

3.5E-02

0.85

ATLN2

2.10

3.10

0.83

GCD
ATLN1

2.30

4.0E-02

0.0E+00
0.80

GCD-20%

Figure 10. Global reliability index for ATLN1, ATLN2 and


ECA1 in hogging.

PF

Probability of failure

0.50

Intact

0.70

The calculations for the reliability were performed


using the computer program COMREL (Gollwitzer et
al., 1988), for the six set of conditions presented before
in both hogging and sagging.
Figure 9 shows an example of the relationship
between the RIF and both the reliability index and
the probability of failure. It is clear that the reliability index is almost linearly dependent of the ultimate
moment of the damaged ship.
Figures 10 and 11 present the reliability indexes for
conditions ATLN1, ATLN2 and ECA1 in hogging and
sagging respectively (GCD Grounding Center Damage, GCMD Grounding Center Major Damage). One

Figure 11. Global reliability index for ATLN1, ATLN2 and


ECA1 in sagging.

1505

ECA1

ECA2

ECA3

ECA1
3.50

3.00

3.00

2.50

2.50

2.00

2.00

ECA2

ECA3

MD+20%

GCMD

GCMD-20%

Figure 13. Reliability index for ECA1, ECA2 and ECA3,


for the ship with full load in sagging.
ECA2 - FL

ECA2 - BL

Rules

ECA - avg.

ECA2 - PL

5.00

4.50

4.00

3.50

3.00

2.50

GCMD+20%

GCMD

GCMD-20%

GCD+20%

GCD

GCD-20%

2.00
Intact

(GCMD + 20%). This happens, in spite of the ultimate


bending moment being bigger for hogging, because the
RIF is smaller than for sagging with the same damage.
Like it was seen before, this is due to the location of
the damage. This clearly indicates that, concerning a
damaged ship, one cannot analyze solely the ship in
sagging. Depending on the location and extension of
the damage the lowest reliability might be found in
hogging.
Figure 12 presents the reliability index for the
ECA1, ECA2 and ECA3 situations for the ship with
ballast load in hogging. The ballast load is the load
condition that presents the smallest values for the reliability in hogging. The difference from ECA1 to ECA2
is that one considers one voyage with the duration of
one week, for the second, and one year of operation, for
the first. The higher reliability of ECA2 is explained
by the fact that the wave bending moment is the dominant load. So if the ship is exposed to the same wave
load period for a shorter time, the probability of failure decreases. All of this proofs the common sense
of sending a ship to dry-dock in the shorter period
possible after being damaged. The difference between
ECA2 and ECA3 is the value of the still-water bending moment which is higher for the second. Like it was
seen before the reliability index is very sensible to this
moment while in hogging.
Figure 13 shows the reliability index for the ECA1,
ECA2 and ECA3 for the ship with full load in sagging. The full load is the load condition that presents
the smallest values for the reliability in sagging. The
results for sagging confirm what was seen for hogging,
but the impact of the still-water bending moment on
the reliability index is smaller in sagging.
In Figures 14 and 15 the global reliability index
obtained in the Rules condition is compared with the

GCD+20%

GCMD+20%

Figure 12. Reliability index for ECA1, ECA2 and ECA3,


for the ship with ballast load in hogging.

GCD

0.00
Intact

0.00
GCMD

0.50

GCMD-20%

0.50

GCD+20%

1.00

GCD

1.00

GCD-20%

1.50

Intact

1.50

GCD-20%

3.50

Figure 14. Results for the reliability index obtained with the
Rules and for ECA2 conditions in hogging.

reliability index obtained for each load condition in


ECA2 conditions. The Rules condition uses the coefficients presented in ABS (1995) to combine the loads.
These coefficients present a simplified manner of
calculating the load in the damage condition using
the known wave and still-water bending moments
obtained for one year of operations in the North
Atlantic. The ECA2 conditions combine a realistic scenario of operations (one voyage) with a combination
coefficient for the still-water loads similar to the one
used in the Rules conditions. In hogging the Rules
conditions evaluate well with the values obtained for
ECA2 in partial load but underestimate the damage
effect of the ship in ballast load which has the smallest

1506

ECA2 - FL

ECA2 - BL

Rules

ECA - avg.

Table 11.

ECA2 - PL

5.00

ECA2

4.50

FL

BL

PL

FL

BL

PL

4.86
4.68
4.64
4.58
4.54
4.44
4.33

3.03
2.78
2.73
2.65
2.59
2.46
2.32

3.47
3.23
3.18
3.10
3.04
2.91
2.77

4.87
4.69
4.65
4.59
4.55
4.45
4.34

2.24
1.99
1.94
1.86
1.80
1.67
1.53

2.71
2.46
2.41
2.33
2.27
2.14
1.99

2.69
2.60
2.58
2.57
2.59
2.57
2.52

3.98
3.92
3.90
3.89
3.91
3.89
3.85

3.12
3.04
3.02
3.01
3.03
3.01
2.96

2.10
2.01
1.99
1.98
2.00
1.98
1.92

4.00
3.93
3.92
3.91
3.93
3.91
3.87

2.61
2.52
2.51
2.49
2.51
2.49
2.44

4.00

Reliability indexes for one voyage.

3.50

3.00

2.50

GCMD+20%

GCMD

GCMD-20%

GCD+20%

GCD

GCD-20%

Intact

2.00

Figure 15. Results for the reliability index obtained with the
Rules and for ECA2 conditions in sagging.
Table 10.

Global reliability indexes.

Hogging
Intact
GCD20%
GCD
GCD+20%
GCMD20%
GCMD
GCMD+20%
Sagging
Intact
GCD20%
GCD
GCD+20%
GCMD20%
GCMD
GCMD+20%

ATLN1

ATLN2

Rules

ECA1

2.12
1.84
1.78
1.69
1.61
1.46
1.28

1.87
1.58
1.52
1.42
1.35
1.18
0.99

3.33
3.11
3.06
2.99
2.94
2.81
2.68

1.58
1.27
1.21
1.10
1.02
0.85
0.64

1.77
1.67
1.65
1.63
1.66
1.63
1.57

1.58
1.48
1.46
1.44
1.46
1.44
1.38

3.58
3.49
3.48
3.46
3.48
3.46
3.42

1.18
1.07
1.05
1.03
1.06
1.03
0.96

Hogging
Intact
GCD20%
GCD
GCD+20%
GCMD20%
GCMD
GCMD+20%
Sagging
Intact
GCD20%
GCD
GCD+20%
GCMD20%
GCMD
GCMD+20%

Table 12. Probability of failure considering one year of


operations scenario.
ATLN1
(102 )

ATLN2
(102 )

Rules
(103 )

ECA1
(101 )

Sagging

1.70
3.31
3.76
4.59
5.34
7.26
10.10
ATLN1
(102 )

3.08
5.71
6.44
7.74
8.89
11.80
16.00
ATLN2
(102 )

0.44
0.94
1.09
1.39
1.67
2.44
3.74
Rules
(104 )

0.57
1.02
1.14
1.35
1.53
1.99
2.62
ECA1
(101 )

Intact
GCD20%
GCD
GCD+20%
GCMD20%
GCMD
GCMD+20%

3.82
4.75
4.94
5.14
4.87
5.14
5.79

5.67
6.98
7.25
7.53
7.16
7.53
8.42

1.74
2.37
2.51
2.65
2.46
2.65
3.15

1.18
1.42
1.47
1.52
1.45
1.52
1.67

Hogging
Intact
GCD20%
GCD
GCD+20%
GCMD20%
GCMD
GCMD+20%

values for the reliability. It gives however more conservative values than for the average of ECA2 (avg.),
with a mean difference of 13.3%. The Rules condition
for sagging do not evaluate well with any load conditions, underestimating the effect of the damage for
both the partial and full loads. It also underestimates
the average of ECA2, with a mean of 9.8%.
Tables 10 and 11 present the results for the reliability
index while Tables 12 and 13 present the results for the
probability of failure. From these values it becomes
clear that the effect of increasing the still-water load
can not be ignored. It is clear that in ballast load and
partial load the ship presents a higher reliability for
hogging than for sagging while the opposite can be
verified for the full load. Such information should help

ECA3

the ship operator to decide if it is more reliable to


change the ship load before the voyage to dry-dock.
For this purpose a study of the reliability of the ship
in the lightweight condition should also be performed.
This was not done here because of the lack of data
available.
6

CONCLUSIONS

The reliability analysis of a Suezmax tanker accidentally grounded was performed. The study was

1507

Table 13.

Probability of failure considering one voyage.


ECA2

ECA3

Hogging

FL (106 )

BL (103 )

PL (103 )

FL (106 )

BL (102 )

PL (103 )

Intact
GCD20%
GCD
GCD+20%
GCMD20%
GCMD
GCMD+20%

0.58
1.44
1.72
2.28
2.83
4.44
7.34
ECA2

1.23
2.69
3.14
3.98
4.76
6.88
10.30

1.50
3.32
3.89
4.95
5.95
8.68
13.10

0.56
1.38
1.65
2.18
2.71
4.26
7.01
ECA3

1.27
2.33
2.62
3.14
3.59
4.73
6.36

3.40
6.94
7.98
9.88
11.60
16.10
23.00

Sagging

FL (103 )

BL (105 )

PL (103 )

FL (102 )

BL (105 )

PL (103 )

Intact
GCD20%
GCD
GCD+20%
GCMD20%
GCMD
GCMD+20%

3.63
4.69
4.91
5.14
4.84
5.14
5.90

3.45
4.54
4.77
5.02
4.69
5.02
5.84

0.90
1.19
1.25
1.32
1.23
1.32
1.53

1.80
2.24
2.33
2.42
2.30
2.42
2.72

3.19
4.20
4.41
4.64
4.34
4.64
5.40

4.57
5.87
6.14
6.42
6.05
6.42
7.36

successful on finding a way to link the influence of


the damage in the ultimate moment of the ship with
the reliability, using the residual strength index (RIF).
A clear dependency between the RIF and the reliability
index was proven.
The loading of the ship was defined based on the
extremes that the ship could find during operation for
both still-water loads and wave loads.
The conventional process of calculation (IACS,
2000) was compared with one voyage of one week
to dry-dock thru European coastal areas.
It was shown that the load combination coefficient
for the still-water (Kus ) influences the reliability considerably and should be the target of future studies.
Methods should be developed to extract its value from
the load distribution of damaged ships.
It was also shown that in spite of the reliability being
lower in the intact condition for sagging; in the damaged condition it is possible to find lower values for
hogging. It all depends on the location and size of the
damaged areas. In the future new studies, considering
damage in other areas of the ship should be performed.
The reliability index values for the load conditions
can help decide in which one should the ship make its
voyage to dry-dock, however the lightweight condition
should also be studied, which was not done here due
to lack of data.
ACKNOWLEDGEMENTS
This paper has been prepared within the project
MARSTRUCT Network of Excellence on Marine
Structures, (www.mar.ist.utl.pt/marstruct/), which is

being funded by the European Union through the


Growth program under contract TNE3-CT-2003506141.
The authors are grateful to Ricardo Pascoal and
Joko Parunov for their help in providing data for the
study in this paper.
REFERENCES
ABS 1995, Guide for Assessing Hull-Girder Residual
Strength for Tankers. American Bureau of Shipping.
ABS, DNV & Lloyds Register 2005. Common structural
rules for double hull oil tankers, March 2005, Background
document, Section 9 Design verification. American
Bureau of Shipping, Det Norske Veritas and Lloyds
Register, Version: 1.
Fang, C. & Das, P.K. 2004a. Survivability and reliability
of damaged ships after collision and grounding. Ocean
Egineering, 32: 293307.
Fang, C. & Das, P.K. 2004b. Hull girder ultimate strength
of damaged ship. 9th Symposium on Practical Design
of Ships and Other Floating Structures, LuebeckTravemuende, Germany.
Gollwitzer, S., Abdo, T. and Rackwitz, R. 1988. FORM
Program Manual. Munich.
Gordo, J.M. & Guedes Soares, C. 2000. Residual strength of
damaged ship hulls. 9th International Congress of International Maritime Association of the Mediterranean, Ischia,
Italy.
Gordo, J.M., Guedes Soares, C. and Faulkner, D. 1996.
Approximate Assessment of the Ultimate Longitudinal
Strength of the Hull Girder, Journal of Ship Research,
4(1): 6069.
Guedes Soares, C. & Dogliani, M. 2000. Probabilistic modeling of time-varying still-water load effects in tankers.
Marine Structures, 13(2): 129143.

1508

Guedes Soares, C. & Moan, T. 1988. Statistical analysis of


still water load effects in ship structures. Transactions
of the Society of Naval Architects and Marine Engineers
(SNAME), New York, 96(4): 129156.
Guedes Soares, C. & Parunov, J. 2006. Structural reliability of a suezmax oil tanker designed according to new
joint tanker project rules. Proceedings of the 25th International Conference on Offshore Mechanics and Artic
Engineering, (OMAE2006), ASME, New York, Paper
no. 92650.
Guedes Soares, C. & Teixeira, A.P. 2000. Structural reliability
of two bulk carriers designs. Marine Structures, 13(2):
107128.
Guedes Soares, C. 1984. Probabilistic models for load effects
in ship structures. Division of Marine Structures, The
Norwegian Institute of Technology, Report UR-84-38.
Guedes Soares, C. 1992. Combination of primary load effects
in ship structures. Probabilistic Engineering Mechanics,
7: 103111.
Guedes Soares, C. 1996. On the definition of rule requirements for wave induced vertical bending moments.
Marine Structures, 9: 409425.
Guedes Soares, C. and Moan, T. 1991. Model uncertainty
in the long-term distribution of wave induced bending
moments for fatigue design of ship structures. Marine
Structures, 4: 295315.

Guedes Soares, C., Dogliani, M., stergaard, C.,


Parmentier, G. & Penderson, P.T. 1996. Reliability based
ship structural design. Transactions of the Society of Naval
Architects and Marine Engineers (SNAME), New York,
104: 357389.
Hogben, N., Da Cunha, L.F. & Olivier, H.N. 1986. Global
Wave Statistics. British Marine Technology, London, Publishing Urwin Brothers Limited.
IACS. 2000. Recommendation No. 34. Standard Wave Data,
Rev. 1.
Paik, J.K., Thayamballi, A.K. & Yang, S.H. 1998. Residual
strength assessment of ships after collision and Grounding. Marine Technology, 35: 3854.
Teixeira, A.P. & Guedes Soares, C. 1998. On the reliability
of ship structures in different coastal waters. Shiraishi,
Shinozuca & Wen (ed.), Structural Safety and Reliability,
A. A. Balkema, pp. 20732076.
Teixeira, A.P. & Guedes Soares, C. 2005. Assessment of
partial safety factors for the longitudinal strength of
tankers. Guedes Soares, Garbatov & Fonseca (ed.), Maritime Transportation and Exploitation of Ocean and
Coastal Resources, London: Taylor & Francis Group,
pp. 16011609.

1509

Você também pode gostar