Safety Science 85 (2016) 282–292

Contents lists available at ScienceDirect

Safety Science
journal homepage: www.elsevier.com/locate/ssci

Statistical analysis of ship accidents and review of safety level

Eliopoulou Eleftheria a, Papanikolaou Apostolos b,⇑, Voulgarellis Markos a
a
School of Naval Architecture and Marine Engineering, Ship Design Laboratory, National Technical University of Athens, Greece
b
Ship Design Laboratory, National Technical University of Athens, Greece

a r t i c l e i n f o a b s t r a c t

Article history: Safety may be defined as an acceptable state of risk by society. In this respect, for assessing the current
Received 25 September 2015 safety level of ships, it is necessary to quantify the risk level of the operating world fleet, thus estimate
Received in revised form 18 December 2015 and assess the basic contributors to risk, namely the frequency of maritime accidents and the extent of
Accepted 1 February 2016
their consequences. The present investigation was motivated by earlier published work of Det Norske
Veritas (DNV, 2006), in which they were some alarming signals of worsening of the level of maritime
safety. A justified question therefore is whether and how the level of ship safety changed thereafter.
Keywords:
Recalling that a fundamental step of a Formal Safety Assessment of maritime assets is the investigation
Maritime safety
Risk assessment
of relevant casualty reports and the analysis of historical data, which characterise the maritime safety
Ship accident statistics performance in the studied period, the herein presented work deals with a systematic analysis of ship
Safety level accidents in the last decade as a way to evaluate the current level of safety for the majority of ship sub-
Formal Safety Assessment types present in the world merchant fleet and to conclude on the foreseeable future. The presented anal-
Maritime regulations ysis also includes a deeper investigation about possible relationships between accident rates and ship’s
age, which proved more complex than initially thought. The outcome of the present study indicates that
in the last decade although the frequencies of ship accidents generally increased, the safety level of var-
ious ship types did not significantly change, as the consequences of accidents remained in average at
about the same level.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction schemes which all contributed to a drastic reduction of accident fre-

The history of maritime transport is marked by ship accidents
with partly disastrous consequences on human lives and impact
on society and the marine environment. In response to these disas-
trous accidents, more and more new requirements and amend-
ments of existing regulations for the safe maritime transport
were introduced by the International Maritime Organisation
(IMO). A remarkable example in this respect is the improvement
of safety of tanker operations after year 1990, which is marked
by the catastrophic marine pollution accident of the Exxon Valdez1
tanker in Alaska in 1989; namely, the introduction of new regula-
tions and guidelines, of safety codes and improved crew training
⇑ Corresponding author.
E-mail address: papa@deslab.ntua.gr (P. Apostolos).
URL: http://www.naval.ntua.gr/sdl (P. Apostolos).
1
The Exxon Valdez was that largest oil spill in US waters until the Deepwater
Horizon offshore spill in 2010; the estimated volume of the spill is being disputed and
is ranging between 41,000 and 119,000 m3, impacting 2100 km of the Alaskan
shoreline; the associated oil spill costs exceeded 7 billion USD; acc. to the Interna-
tional Tanker Owners Pollution Federation (ITOPF), it was the ‘most expensive oil spill
in history’, even though not the largest one by the amount of spilt oil of tankers
(Eliopoulou et al., 2012).
quencies thereafter (Eliopoulou and Papanikolaou, 2007). They led to
an improvement of the maritime safety culture, with the shipping
industry adopting more and more a proactive approach to ship oper-
ation. Noting that an overwhelming part of marine accidents (more
than abt. 80%) is attributable to human factors, it may be concluded
that introduced measures addressing ship’s operation (active safety
measures) were very effective in the reduction of accident frequen-
cies, whereas measures affecting ship’s design and technology (pas-
sive safety measures) have more contributed to the mitigation of
accident consequences (for which achieved reduction and control
were not that significant). Fig. 1 presents a snapshot for the develop-
ment of navigational (collision, grounding, contact) accident fre-
quencies for AFRAMAX tankers from the late 70ties and up to the
early 2000. In the graph, we have marked the implementation of
some key regulations and safety codes that could be held responsible
for the declining trends of particular rates.
Although Fig. 1 pertains at first only to AFRAMAX tankers, rele-
vant trends may be qualitatively extended to all tanker categories
and even to other ship types, as most of the noted specific regula-
tions were applied to all ship types; exceptions apply to VETTING
and the 94 SOLAS ISM, which are applicable only to oil carrying
tankers. The essence of the above snapshot, however, capturing a

http://dx.doi.org/10.1016/j.ssci.2016.02.001
0925-7535/Ó 2016 Elsevier Ltd. All rights reserved.
 E. Eleftheria et al. / Safety Science 85 (2016) 282–292
Fig. 1. Timeline of navigational accident rates vs. introduced international maritime regulations, safety guidelines and codes (Eliopoulou and Papanikolaou, 2007).
period of 21/2 decades, is that the frequencies of navigational acci-
dents were reduced for the collisions and groundings by a factor of
close to 10, and to a lesser (but still significant) degree for the con-
tacts. However, in the post 2000 period, frequencies of occurrence
of serious accidents started increasing again, according to the pre-
sent study, which is at first alarming and subject of the present
paper.
Relatively late compared to other industries (land- and air
transport, energy/nuclear industry) and not simply responding to
maritime disasters, IMO introduced in year 2000, a formalised pro-
cedure for the assessment of ship safety, known from other safety
critical industry systems, namely the Formal Safety Assessment
process (FSA). FSA is a formalised risk assessment methodology
aimed at enhancing maritime safety by using risk analysis and cost
benefit assessment. FSA was originally developed (at least partly)
as response to the Piper Alpha disaster of 1988, when the particular
offshore platform exploded in the North Sea and 167 people lost
their lives. Fig. 2 shows schematically the major steps of FSA. Step
1 includes the identification of all possible hazards of the problem
under investigation leading to a list of relevant accident scenarios
with potential causes and outcomes. In Step 2, a risk analysis is
performed in order to evaluate all relevant risk factors. The step
includes the determination of the frequency rates of identified
major accident categories and of their consequences by statistical
analysis of historical data. In Step 3, various risk control options
(RCOs) are identified aiming at controlling the ensuing risk factors
and mitigating the consequences of accidents. The viability of
selected risk control options are examined by a cost benefit assess-
ment (CBA) determining the cost effectiveness of each alternative
risk control option (Step 4). In the final Step of FSA, recommenda-
tions for decision-making are given (Wang, 2001).
In 2002, IMO approved the guidelines on Formal Safety Assess-
ment (MSC/Circ.1023, 2002) in order to support decision making in
relation to the introduction of new regulations; they namely intro-
duced a process formalising the assessment of new regulations
using risk analysis and cost benefit assessment techniques
(Skjong, 2002, 2009). The particular Guidelines were amended by
MSC/Circ.1180-MEPC/Circ.474 in 2005 and further revised in
2013 (MSC-MEPC.2/Circ. 12, 2013). Based on this procedure, a ser-
ies of high level FSA studies were elaborated within the EU funded
research project SAFEDOR for the most important ship types,
namely Containerships (MSC 83/INF.8, 2007), LNG ships (MSC 83/
INF.3, 2007), Cruise ships (MSC 85/INF.2, 2008), RoPax ships
(MSC 85/INF.3, 2008) and Large Crude Oil tankers (MEPC 58/
283
INF.2, 2008). The accident statistics of the pre-mentioned FSA stud-
ies concerned mainly the time period 1990 (or 1993) to 2004,
depending on the availability of data in the EU funded project
SAFEDOR (2005–2009). For some ship types, it was considered nec-
essary to update the historical analysis of casualty data and conse-
quently the risk assessment of the particular FSA study. In this
respect, an update of the FSA for fully cellular containerships was
performed (Eliopoulou et al., 2013; Hamann et al., 2013). Focusing
on the passenger ships, two sequential thorough re-investigations
were done through the EU funded project GOALDS (Papanikolaou
et al., 2013) and the EMSA III project (Pagiaziti et al., 2015). Further
significant risk assessment case studies in the maritime field were
recently presented by Montewka et al. (2014) and Goerlandt and
Montewka (2015).
The herein presented investigation started as an update
(Papanikolaou et al., 2014) of similar research work conducted a
decade ago by DNV (DNV, 2006) and went deeper into relation-
ships of accident frequencies and consequences to ship age
(Voulgarellis, 2015). The objectives of the present paper are the
quantification of the risk level of the operating world fleet by sta-
tistical analysis of historical data and the assessment of the safety
level of all basic merchant ship types in terms of accidents’ occur-
rence, initial frequencies and basic consequences. For each ship
type, accidents occurred in the time period 2000–2012 are being
analysed and presented with respect to accident category, total
losses of ships and number of fatalities. Furthermore, the complex
relationship between accident rates and ship age is also investi-
gated and the various contributing factors discussed.
The paper is organised as following: After a general introduc-
tion to maritime safety, the interaction between marine accidents
and international regulatory developments at IMO and the intro-
duction of Formal Safety Assessment, we present a statistical anal-
ysis of historical casualty data, in which we elaborate on the
statistical sampling plan and operational world fleet at risk; a
review of the obtained major results of the analysis follows, in
which we first comment on the yearly frequencies of the recorded
serious accidents per accident and ship type for the period 2000–
2012; in the same frame, a conducted more specific analysis con-
siders a sampling plan of the ships categorised by their age and
studies the impact of ship’s age on the frequency of accidents,
which is not uniform; a similar parametric analysis follows for
the consequences of serious accidents in terms of total ship losses
and fatalities of People On Board (POB); a timeline analysis regard-
ing the sufficiency of the used statistical sample by use of Kendall’s
284 E. Eleftheria et al. / Safety Science 85 (2016) 282–292
Fig. 2. The five main steps of Formal Safety Assessment (MEPC 58/INF.2, 2008).
Tau test follows; finally, we discuss the overall outcome of this
study and conclude on the current level of safety of the world fleet.
2. Statistical analysis of casualty data
The source of the processed casualty data is the IHS Sea-webÒ
database and the investigation pertains to the generic ship types
as defined by the particular database; namely, General Cargo ships,
Bulk Carriers (DWT > 1500), Fishing vessels, Reefer ships, Ro-Ro
cargo ships, Car Carriers, LNG and LPG carriers, pure Passenger
and Cruise ships and Passenger Ro-Ro cargo vessels. The analysis
of casualty records, performed for the time period 2000–2012,
include those accidents associated with merchant cargo and pas-
senger ships, regardless ship size (minimum vessel size was
100 GT), however built after year 1980. Operational fleet at risk
has been calculated from source data of the Lloyd’s Register of
ShippingÒ database for all the afore-mentioned ship types.
The presented results referring to Cellular Containerships and
Large Crude Oil tankers (greater than 60,000 DWT) are coming
from two earlier completed research projects, namely CONTIOPT
(2011–2013) and the EU funded project SAFEDOR (2005–2009),
which were complemented by internally funded research of the
Ship Design Laboratory in later years (Eliopoulou et al., 2012).
2.1. Sampling plan
Casualty records were extracted from the IHS Sea-webÒ data-
base according to the definitions comprising the ‘‘generic ship
type”, as illustrated in Table 1. With respect to the accident cate-
gory (and ensuing type of hazards), the following accident cate-
gories, as defined in the particular database, were included in the
analysis:
Collision: striking or being struck by another ship, regardless of
whether under way, anchored or moored. This category does
not include striking under water wrecks.
Contact: striking or being struck by an external substance but
not another ship or the sea bottom.
Wrecked/stranded (includes grounding and bumping over bars,
etc.): includes ships reported hard and fast for an appreciable
period of time and cases reported touching sea bottom. This
category includes entanglement on under water wrecks.
Fire/explosion: where the fire and/or explosion is the first event
reported (except where first event is a hull/machinery failure
leading to fire/explosion).
Foundered: includes ships which sank as a result of heavy
weather, springing of leaks, breaking in two etc., and not as a
consequence of the other accident categories.
Hull/machinery damage: includes ships lost or damaged as a
result of hull/machinery damage or failure.
Missing vessel: the cases where after a reasonable period of time,
no news having been received of a ship and its fate being there-
fore undetermined.
War loss/damage during hostilities: this category is intended to
encompass damage or other incidents occasioned to ships by
hostile acts.
Miscellaneous: includes ships which have been lost or damaged
which, for want of sufficient information, or for other reasons,
cannot be classified.
The initial sampling plan of available accident records was fil-
tered to include only accidents of serious degree of severity,2 since
the ‘‘non serious” cases were very few in comparison to the ‘‘serious”
cases. The latter suggests significant underreporting for the ‘non seri-
ous’ cases which means that their statistical analysis would be erra-
tic. Moreover, in the ‘‘serious” cases, events with total losses of ships
are also included. In effect, the analysis was focused on serious acci-
dents associated with merchant passenger and cargo ships built after
year 19803 resulting to a sampling plan of 4572 serious accidents
within the studied period, as shown in Table 1.
2.2. Operational fleet at risk
Before proceeding to the determination of accident frequency
rates, as necessary for the conduct of a risk analysis, we need to
relate the number of accidents to the yearly operating fleet at risk.
2
Serious marine casualty is the casualty, which results to structural damage,
rendering the ship unseaworthy or to ship breakdown, actual total loss or any other
undefined situation resulting in damage or financial loss considered to be serious.
3
The exclusion of older ships supports the uniformity of the statistical sample in
terms of employed shipbuilding technology, even though (less radical) changes are
being observed also in the decades after 1980. Limiting the statistical sample to
shorter time periods generally leads to an increase of the uncertainty of statistical
metrics, due to the limited number of recorded accidents.
 E. Eleftheria et al. / Safety Science 85 (2016) 282–292
Table 1
Number of serious accidents per ship type, time period 2000–2012.
General Bulk Fishing Reefer Ro-Ro
Cargo Carriers vessels ships Cargo
3228 1609 456 210 184
Passenger Ro-Ro Cargo (RoPax)
888
Tables 2 and 3 present the operational fleet at risk per ship type for
the period from 2000 to 2012. For all ship types, except for Large
Crude Oil Tankers and Cellular Containerships for which relevant
data are coming from projects SAFEDOR (2005–2009) and
CONTIOPT (2011–2013), the annual operational fleet was calcu-
lated using raw data from Lloyd’s Register of ShippingÒ database.
3. Review of major results
In the following we analyse and review the major results of the
conducted statistical analysis in terms of frequencies of accident
occurrence for all accident categories and ship types, as well as
in terms of typical accident consequences. Presented accident fre-
quencies were calculated by dividing the total annual number of
registered accidents in a certain period by the number of ships
operating worldwide in the same period (Fleet at Risk). Where
overall accident frequencies are presented, the accident categories
‘‘War Loss/Hostilities”, ‘‘Miscellaneous”, and ‘‘Missing” were taken
into consideration; however, the consequences of these accidents
were not considered in the listed statistics. Finally note that the
presently conducted study on relationships between ship age and
frequencies of accident occurrence did not include Large Tankers
and Cellular Containerships due to the lack of data for the investi-
gated time period (see, however, an earlier related study by
Papanikolaou and Eliopoulou, 2008 on the impact of age on tanker
accidents).
3.1. Serious accidents
Table 4 presents the annual frequency of occurrence of serious
accidents by ship type. A peak value appears for all ship types in
the period of 2007–2008, whereas thereafter frequencies have sig-
nificantly decreased, depending on ship subtype. The observed
peak value in 2007–2008 may be attributed to following reasons:
 Increased traffic in a period of heat of the world economy, thus
increased frequency of navigational accidents.
 Reduced maintenance and delayed laid-up of aged ships in a
period of high transport demand.
 Change of practice in the recording of accidents by the database
provider (IHS Fairplay), due to organisational changes (transfer
from Lloyds Register to IHS Fairplay).
Commenting on the statistics of cargo ships fleet, lower fre-
quencies of serious accidents are observed for fishing vessels,4 with
overall averaged values of 5.40E 03 accidents/shipyear, LNG with
9.12E 03, LPG ships with 1.37E 02 and large crude oil tankers with
4
Note that in this study, we do consider only ocean going fishing vessels of Gross
Tonnage over 100 GT; it is acknowledged that the number of accidents and fatalities,
when considering additionally occupational risk (that means, the professional risk of a
fisherman, independently of the main hazards, considered in this study) and smaller
fishing vessels/boats, is much higher. Namely, during IMO-SLF49 (2006), the Food and
Agriculture Organization (FAO) informed the IMO Sub-Committee that the fishing
industry suffers in excess of 24,000 fatalities per year, the large majority of which were
occurring on small fishing vessels.
285
Car LPG LNG Fully Cellular Large Crude Oil,
Carriers ships ships Containerships up to 2010
194 140 21 1090 259
(Pure) passenger ships Cruise ships
356 217
1.28E 02. On the other side, Reefer ships, General cargo, Cellular
Containerships and Bulk Carriers have comparably much higher fre-
quencies, namely 1.94E 02, 2.73E 02, 2.42E 02 and 2.37E 02
respectively. Finally, highest overall frequency of accidents was
observed for Ro-Ro cargo ships (3.34E 02) and Car Carriers
(2.97E 02).
Among passenger ships, cruise vessels present the highest over-
all accident frequency (5.18E 02) and RoRo Passenger ships follow
(4.39E 02), as shown in Table 5. Both ship types exhibit much
higher values of frequencies in comparison to small passenger
ships (1.36E 02). Likewise to cargo ships, a considerable peak of
frequency value appears in the period of 2007–2010, and a consid-
erable drop thereafter.
Fig. 3 presents the overall frequency of accident’s occurrence by
accident category for cargo and passenger ships. The accident cat-
egory with the highest values of frequency of occurrence is hull/
machinery damage; this could be expected, because the particular
accident category has by definition a broad range and includes
accidents related to machinery failure (without leading to another
accident type, i.e. contact, grounding), to failures of hull fittings/
equipment and to hull structural failures that resulted in a
non-accidental way (so-called NASF: Non-Accidental Structural
Failures, e.g. caused by fatigue of the structure, improper construc-
tion/welding, etc.).
When focusing on the cargo ships (Fig. 3 left hand side) we
observe the following:
 Collision events: lowest frequency is observed for LNG ships and
Fishing vessels, whereas the highest values are related to Con-
tainerships, Car Carriers and RoRo Cargo ships, which are char-
acterised by more frequent operation in congested terminal
areas.
 Contact events: Fishing vessel and LPG ships exhibit lower fre-
quencies, whereas the highest values are given for Car Carriers
and RoRo Cargo ships, for which the same causes like for colli-
sions apply.
 Fire/explosion: lower frequencies are presented again by LNG
and Fishing vessels, whereas the highest values are given for
RoRo Cargo ships and Car Carriers, transporting partly sensitive
Ro-Ro cargo.
 In the accident category ‘‘Foundered”, General Cargo ships have
the highest frequency; this is related to the high average age for
this type of ships (see also later section), to which many small
coasters belong; LNG ships did not have any such accident.
 LNG ships and Fishing vessels have the lowest frequencies of
grounding events, whereas the highest frequency values are
coming from Bulk Carriers and General Cargo ships and are
related their reduced maneuverability and powering for safe
operation in adverse weather conditions (see, Papanikolaou
et al., 2015).
Among the passenger ships, cruise and Ro-Ro Passenger ships
exhibit significantly higher values of frequency of hull and machin-
ery damage, whereas the frequencies for other accident categories
are quite similar, except for the pure passenger ships, which are
286 E. Eleftheria et al. / Safety Science 85 (2016) 282–292

Table 2
Cargo ships, operational fleet at risk, time period 2000–2012.

Bulk Carriers Car Carriers Reefer ships Fishing ships General Cargo LNG ships LPG ships RoRo Cargo Large tankers Containershipsa
2000 3122 303 785 5337 7375 67 574 330 1554 2036.5
2001 3292 342 806 5629 7615 81 606 355 1538 2207.3
2002 3586 360 819 5977 7802 82 637 367 1567 2411.5
2003 3796 376 823 6233 8004 92 661 383 1636 2601.4
2004 3955 395 826 6408 8225 107 692 398 1663 2778.8
2005 4206 419 830 6554 8527 128 712 413 1756 3009.0
2006 4513 454 832 6684 8876 148 725 421 1855 3337.6
2007 4817 496 835 6770 9231 174 765 438 1977 3730.2
2008 5114 547 842 6824 9658 205 826 451 2114 4169.6
2009 5778 615 849 6907 10,127 257 908 461 2258 4447.1
2010 6888 677 855 6972 10,568 298 970 480 2373 4628.5
2011 8836 743 861 7054 10,988 324 1031 499 – 4809.3
2012 9919 801 854 7113 11,329 340 1086 512 – 4932.5
Total 67,822 6528 10,817 84,462 118,325 2303 10,193 5508 20,291 45,099
a
Calculated on a monthly base.

Table 3 Bulk Carriers appear to have an almost linear relationship

Passenger ships, operational fleet at risk, time period 2000–2012. between frequency of accidents and ship’s age. The frequency
(Pure) passenger ships Cruise ships Passenger RoRo increases, when ship age increases, reaching up to 3 + times higher
Cargo (RoPax) values for the older ships in comparison to the newbuildings. Car
2000 1653 222 1197 Carriers and Bulk Carriers exhibit considerably increased frequen-
2001 1728 242 1254 cies of accidents in ages over 20 years, when fatigue and other
2002 1794 263 1328 structural problems are getting gradually more serious.
2003 1865 283 1390
LNG ships present relatively low frequencies up to the age of
2004 1929 300 1457
2005 1995 313 1513 20 years, whereas the frequency of accidents pertaining to ships
2006 2038 324 1565 with age greater than 20 years has a 2 + times higher value.
2007 2072 338 1610 RoRo Cargo ships, as General Cargo and Car Carriers, exhibit sig-
2008 2120 355 1669 nificantly high frequencies in the age group of less than 5 years,
2009 2167 369 1730
2010 2220 380 1777
which indicates possible problems with the quality of some new
2011 2261 397 1834 built ships by low cost yards.
2012 2290 407 1889 Passenger ships do not present significant differences among
Total 26,132 4193 20,213 the different age groups, thus their accident occurrence can be con-
sidered as independent to ship age. Among them, cruise ships exhi-
bit higher frequency of accidents in the first 5 years, indicating
‘‘teething problems” of complex ship newbuildings. Pure Passenger
generally smaller in size and present overall much lower values ships exhibit the lowest frequency among all passenger ships, not-
(Fig. 3 right hand side). ing that here belong relatively small vessels of less complex con-
struction and operation.

3.1.1. Effect of ship’s age on accident frequencies

In order to study possible effects of ship’s age on accident fre-
quencies, we proceed with a breakdown of the sample ships into
five (5) age group categories, namely less than 5 years old, between
6 and 10 years, 11–15 years, 16–20 and over 20 years old. The fre-
quency of accident occurrence per group age and basic ship type is
presented in Fig. 4. In general, we may distinguish two categories
of ships, namely one with distinct high frequencies and the other
with much lower frequencies.
To the high frequency category belong General cargo ships, Bulk
Carriers, Car Carriers and RoRo Cargo ships. In the lower frequency
category the following ship subtypes belong: Fishing vessels, LNG,
LPG ships and Reefer ships. It is noted that the frequency values of
the first category of ships are roughly twice as large as those of the
second category in all group ages, except for the ship ages over
20 years, where the values for the majority of ship types better
converge.
Commenting on the particular ship types:
Fishing vessels exhibit the lowest overall frequencies and have
almost the same value for all age groups, thus the ship age seems
to be unrelated to the accident occurrence for this ship type. An
explanation to this may be the fact that fishing vessels are to a
large extend operated by ship masters, who happen to be also
(major) owners of the ship, thus are very familiar with the partic-
ular ship’s operation and how to respond to accident hazards.
3.2. Ship total losses
The analysis of ship total losses can provide some insight into
how often a catastrophic event happens. Within the studied time
period and accounting for all accident categories, including war
losses, miscellaneous and missing vessels, the frequencies of ship
total loss by ship type are as follows:
General cargo ships exhibit the highest frequency of ship total
losses, namely 2.98E 03 per shipyear. Fishing vessels and RoRo
Cargo ships follow with 1.63E 03. Reefer vessels, Bulk Carriers
and Car Carriers present reduced values in relation to the pre-
mentioned ship types, with 1.29E 03, 1.21E 03 and 1.07E 03
respectively. LPG ships have a very low frequency of ship total loss
of 5.89E 04, Containerships exhibit a frequency of 2.00E 04,
whereas LNG ships is the only ship type, along with the Large
Crude Oil tankers, that have no ship total loss in the studied period.
Commenting on the above results it is evident that General
Cargo ships, with the highest frequencies of founderings (see,
Table 6), are the most affected ships by total losses. This type of
accidents are to a great extend caused by adverse weather condi-
tions, for which the relatively small ship size (of coaster cargo
ships) and the inferior structural integrity of overaged ships may
lead to disastrous consequences. Fishing vessels are often trapped
in adverse weather conditions, when fishing, and may capsize/sink
 E. Eleftheria et al. / Safety Science 85 (2016) 282–292 287

due to flooding of their deck and/or internal spaces. Finally, Ro-Ro

1.55E 02
1.62E 02

3.51E 03

1.57E 03
3.66E 03
8.17E 03
1.68E 02
2.50E 02
cargo ships may suffer stability problems due to shift of Ro-Ro

0.00E+00
cargo, again in relation to adverse weather.

2012
Note that only two ship types exhibit total losses due to war or

–
hostilities, namely General cargo and Fishing vessels with
4.23E 05 and 2.37E 05 respectively. General cargo ships have

1.71E 02
1.22E 02
1.75E 02
3.38E 03

1.32E 02
1.36E 02
4.51E 02
2.60E 02
0.00E+00
also a notable value for ship total loss frequency (8.45E 06) in
2011

‘‘miscellaneous” accident events. Finally, General cargo ships and

–
Fishing vessels have a frequency of 2.54E 05 and 1.18E 05
respectively in the ‘‘missing” accident category, which are often
1.82E 02
1.73E 02
2.31E 02

2.73E 02
3.19E 03

6.97E 03

2.51E 02
1.22E 02
1.40E 02
0.00E+00 close to foundering accidents.
Tables 6 and 7 present the frequency of serious accidents that
2010

led to ship’s total loss per accident category pertaining to cargo

and passenger ships respectively.
Commenting on the frequencies of the various accident types,
02

02
03
03
03
02

02
02
02

02

General Cargo ships and Car Carriers exhibit the highest frequen-
3.54E

2.11E
7.53E
7.78E
6.61E
2.59E

2.81E
1.55E
3.20E

3.90E
2009

cies of ship total losses in collision events; Bulk Carriers, Fishing

vessels and Reefer ships have relatively low frequencies and the
remaining ship types present negligible values.
02
02
02
03
03
02
02
02
02
02

For the contact events, only General cargo ships and Fishing
3.93E
2.62E
6.76E
7.47E
9.76E
2.54E
3.92E
5.54E
3.65E
1.32E
2008

vessels have a notable value of frequency of ship total loss. For

the remaining ship types, no ship total loss due to contact events
was reported.
02

03
02

02
02
02
02
02
02

02

Relatively high frequency of fire/explosion events is presented

for RoRo Cargo ships and Fishing vessels. Reefer, LPG ships, General
3.78E

6.79E
1.15E

4.31E
5.25E
3.14E
2.73E
3.70E
5.04E

2.09E
2007

Cargo ships and Bulk Carriers exhibit here low frequency of ship’s
total loss, whereas the remaining ship types present negligible fre-
quency values.
02
02

03

02
02
02
02
02
02

02

Foundering cases present significantly high frequencies for the

2.43E
2.84E

4.19E

2.21E
2.88E
2.61E
2.97E
1.19E
3.08E

2.03E
2006

General Cargo ships in comparison to the other ship types. Fishing

vessels, RoRo Cargo ships, Reefer ships, Bulk Carriers and LPG ships
follow with lower frequency values.
02
02
03
02
02
02
02
02
02
02

hull/machinery failures present very low frequencies of ship’s

2.78E
2.86E
5.65E
1.56E
1.54E
1.81E
2.66E
2.13E
1.77E
2.50E

total loss for the majority of ship types, except for RoRo Cargo
2005

ships. Although the frequency of a serious hull/machinery failure

is significantly high (see Fig. 3), this accident type leads rarely to
ship’s total loss.
02
02
02
03
02

02
02
02
02
02

Finally, General Cargo ships, Reefer ships, Bulk Carriers and Car
2.12E
2.93E
3.54E
3.75E
3.74E

1.21E
2.26E
2.88E
1.68E
2.02E
2004

Carriers exhibit the highest frequencies of ship’s total loss in

grounding accidents. This is related to maneuverability problems
for this type of ships in adverse weather conditions, as elaborated
1.86E 02
2.24E 02
3.19E 02

7.56E 03
3.65E 03
2.35E 02
1.73E 02
9.17E 03
4.01E 03
0.00E+00

in the EU funded project SHOPERA (Papanikolaou et al., 2015).

2003

Concerning ships carrying passengers (Table 7), only pure Pas-

senger ships exhibit a notable frequency of ship’s total loss due
to collision events and hull & machinery failures, noting that this
1.55E 02
1.62E 02

3.51E 03

1.57E 03
3.66E 03
8.17E 03
9.95E 03
2.50E 02

passenger ship subtype refers to mostly small size shortsea ships.

0.00E+00

0.00E+00

For all other accident categories, frequencies of the various passen-

2002
Cargo vessels, annual frequency of serious accidents per shipyear.

ger ship subtypes are comparable.

High frequencies of total loss (2.38E 04) are observed for
Cruise vessels in contact events, which involve mostly side con-
1.71E 02
1.22E 02
1.75E 02
3.38E 03

1.32E 02
1.36E 02
4.51E 02
1.50E 02
0.00E+00

0.00E+00

tacts to fixed installations and piers. This is often caused by insuf-

2001

ficient maneuvrability of cruise ships in limited waters and under

the action of strong winds in relation to their large windage area.
Likewise, Cruise ships and RoRo Passenger ships exhibit high
1.82E 02
1.73E 02
2.31E 02
3.19E 03

6.97E 03

2.73E 02
1.33E 02
1.40E 02

1.09E 02
0.00E+00

frequencies of total losses in grounding events and this is attribu-

ted to the operational profile for this type of ships5 in limited
2000

waters and frequent port calls. In any case, the highest frequencies
for ship’s total loss of passenger ships in general are related to fire/
explosion events, followed by Foundered cases. Increased fire/explo-
Cellular containerships

sion frequencies are related to the inherent difficulty of effective fire

control in manifold and complex accommodation/public spaces of
Fishing vessels
General Cargo

Reefer vessels

Large tankers

passenger ships; foundering events are always related to adverse

Bulk Carriers
Car Carriers

RoRo Cargo
LNG ships
LPG ships

5
Table 4

With the most prominent cruise ship total losses due to grounding were in the
study period the Costa Concordia accident in January 2012 (32 fatalities) off the coast
of Isola del Giglio in Italy and the Sea Diamond at Santorini Island in Greece in 2007.
288 E. Eleftheria et al. / Safety Science 85 (2016) 282–292

Table 5
Passenger vessels, annual frequency of serious accidents per shipyear.

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Cruise vessels 3.15E 02 4.96E 02 4.56E 02 3.18E 02 3.33E 02 5.11E 02 5.25E 02 7.10E 02 8.17E 02 7.05E 02 9.47E 02 2.02E 02 2.70E 02
Passenger 3.63E 03 6.37E 03 5.02E 03 8.58E 03 9.85E 03 9.52E 03 6.38E 03 1.01E 02 1.56E 02 3.00E 02 2.57E 02 1.90E 02 1.92E 02
vessels
Passenger RoRo 1.84E 02 1.59E 02 1.51E 02 2.95E 02 3.98E 02 2.91E 02 2.30E 02 5.78E 02 6.53E 02 7.23E 02 7.71E 02 4.80E 02 5.03E 02
vessels

Cargo Vessels -Frequencies per Accident Type Passenger Vessels -Frequencies per Accident Type
General Cargo Bulk Carrier Car Carriers Fishing Vessels LNG
Cruise Vessels Passenger Vessels Passenger RoRo Vessels
LPG Reefer Vessels RoRo Cargo Cellu. Cont/ships Large Tankers
1.40E-02
3.00E-02
1.20E-02
2.50E-02
1.00E-02
2.00E-02
8.00E-03
1.50E-02
6.00E-03
1.00E-02
4.00E-03
2.00E-03 5.00E-03

0.00E+00 Wreck./Str 0.00E+00

Wreck./Stra
Hull/Mchy.

Hull/Mchy.
Collision

Collision
Fire/Expl.

Foundered

Fire/Expl.

Foundered
Contact

Contact
and.
Dam

Dam

nd.
Fig. 3. Frequency per accident category, per shipyear.

Cargo Vessels - Frequency of Accidents per Vessel Age Passenger Vessels - Frequency of Accidents per Vessel Age
General Cargo Bulk Carrier Car Carriers Fishing Vessels Cruise Vessels Passenger Vessels Passenger RoRo Vessels
LNG LPG Reefer Vessels RoRo Cargo 8.00E-02
7.00E-02 7.00E-02
6.00E-02 6.00E-02
5.00E-02 5.00E-02
4.00E-02 4.00E-02
3.00E-02 3.00E-02
2.00E-02 2.00E-02
1.00E-02 1.00E-02
0.00E+00 0.00E+00
less than 5 6 to 10 11 t o15 16 to 20 Above 20 less than 5 6 to 10 11 t o15 16 to 20 Above 20

Fig. 4. Frequency of accidents per ship type and age group.

weather conditions and often to badly maintained and operated 3.3. Fatalities
ships in less developed countries.
In the earlier Fig. 4 it was noticed that Bulk Carriers, RoRo Cargo, Table 8 presents the number of registered fatalities (as the sum
LNG ships and Reefer ships having an age over 20 years exhibit of ‘‘killed” and ‘‘missing” persons) per ship type. Within the studied
high frequencies of accidents in relation to the younger age groups. period, 4603 persons were lost during ship accidents of interest;
Focusing on ship total losses, as elaborated in Fig. 5, this particular from this number, about 65% are attributed to accidents of passen-
relationship is further strengthened (except for the LNG ships), ger ships (RoPax, Passenger and Cruise ship), whereas the
with older ships more prone to total loss events. However, focusing remained percentage is related to lost lives of crew of various cargo
on the ages up to 15 years old, some differences are visible looking ship types (noting that close to 22% of the total is assigned to Gen-
into the ship subtypes, with some ships exhibiting higher total loss eral Cargo ships).
rates in intermediate age groups. Significantly high value of total In Table 9, values for the Potential Loss of Life (PLL, fatalities per
losses is noticed for the younger Car Carriers, which can be only shipyear) are presented, which were deduced from the available
related ship losses in extreme weather conditions with possible statistical dataset (Historical Loss of Life). Even though the number
shift of Ro-Ro cargo. of People On Board (POB) is very small for cargo ships (only crew
Finally, focusing on ships carrying passengers (Fig. 5 right members), compared to passenger ships, it is very important to
hand side), there is a small increase of the total loss frequency highlight that the fatalities per shipyear for some cargo ships are
of cruise ships, when the ship age increases, having also negligi- comparable with those of passenger ships (!), indicating the
ble frequency in the first 5 years age group. Pure passenger ships increased risk of these ship types towards fatal accidents. Also,
exhibit a slight increase up to the age of 15 years; then the fre- the comparably high fatality rate of RoPax ships is noted. The latter
quency reduces, but picks up significantly at the age above is attributed to the fact that some RoPax ships in less developed
20 years. Finally, Ro-Ro Passenger ships with an age above countries are badly maintained and operated, while accidents of
20 years exhibit also significantly high values, compared to overloaded ships in bad weather conditions happen quite fre-
younger ships. quently, sometimes with fatal outcome.
 E. Eleftheria et al. / Safety Science 85 (2016) 282–292 289

Table 6
Cargo vessels, frequency of total losses by accident category.

Collision Contact Fire/explosion Foundered Hull/mchy. damage Wrecked/stranded

General Cargo 6.51E 04 4.23E 05 1.69E 04 1.33E 03 1.10E 04 6.08E 04
Bulk Carrier 2.36E 04 0.00E+00 1.18E 04 2.95E 04 8.85E 05 4.72E 04
Car Carriers 6.13E 04 0.00E+00 0.00E+00 0.00E+00 0.00E+00 4.60E 04
Fishing vessels 1.30E 04 3.55E 05 3.79E 04 7.70E 04 3.55E 05 2.49E 04
LNG ships 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00
LPG ships 0.00E+00 0.00E+00 1.96E 04 1.96E 04 0.00E+00 1.96E 04
Reefer vessels 9.24E 05 0.00E+00 2.77E 04 3.70E 04 0.00E+00 5.55E 04
RoRo Cargo 0.00E+00 0.00E+00 7.26E 04 5.45E 04 1.82E 04 1.82E 04
Cellular containerships 0.00E+00 0.00E+00 0.00E+00 8.87E 05 4.43E 05 6.65E 05
Large Crude Oil Tankers 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00

Table 7
Passenger ships, frequency of total losses by accident category.

Collision Contact Fire/explosion Foundered Hull/mchy. damage Wrecked/stranded

Cruise vessels 0.00E+00 2.38E 04 4.77E 04 0.00E+00 0.00E+00 2.38E 04
Passenger vessels 3.83E 05 3.83E 05 2.30E 04 4.21E 04 7.65E 05 1.15E 04
Passenger RoRo vessels 0.00E+00 0.00E+00 3.96E 04 4.45E 04 0.00E+00 2.47E 04

Another interesting ship type are Reefer ships, which also
exhibited quite high values of historical loss of life in year 2000
(2.42E 02) and in year 2006 (4.21E 02). Both values correspond
to two single accidents, namely the first one in year 2000 with
18 fatalities and the second one in year 2006, when the 3-years
old Reefer vessel ‘‘Heng Da 1” had a grounding event with total loss
and 35 fatalities.
Fatalities per shipyear distributed by accident category are pre-
sented in Fig. 6. Focusing on the cargo ships, the highest fatality
rates are coming from foundering events, whereas groundings, col-
lisions and fire/explosion follow. Contact events, hull/machinery
damages have all very low fatalities frequencies.
Looking closer into RoRo Passenger ships (Fig. 6 right hand
side), both foundering events along with grounding and fire/explo-
sion events are all contributing factors to the observed fatality rate
values and this is related to the operational profile and fitness of
some RoPax ships, operating in less developed countries and
sustaining a high transportation work in terms of volume. Cruise
vessels exhibit also relatively high fatalities in case of contact
events, which is assumed related to the frequent operation of these
ships in limited waters and terminals areas, while strong blowing
winds may affect their safe manoeuvring and docking.
Finally, Tables 10 and 11 present the fatality rates by ship type
and ship age group. Calculating the overall fatality rate for each age
group regardless the shiptype, the younger vessels (less than
5 years old) exhibit the highest fatality rate and this is herein
attributed to loss of a young reefer ship, as it was mentioned earlier
in the discussion about the Table 9 results. The remaining age
groups practically present a uniform relationship between fatality
rates and ship’s age, where the fatality rate generally increases
with the increase of ship’s age. For some ship types the rate of
increase at higher ages is steeper, but this not contradicting the
concluded general dependence rule.
Regarding ships carrying passengers, the age group of
11–15 years as well as the group above 20 years exhibit cumula-
tively high values of fatalities per shipyear, mainly because of

Cargo Vessels - Frequencies of Total Losses per Vessel Age Passenger Vessels - Frequencies of Total Losses
per Vessel Age
General Cargo Bulk Carrier Car Carriers Fishing Vessels
Cruise Vessels Passenger Vessels Passenger RoRo Vessels
LNG LPG Reefer Vessels RoRo Cargo
6.00E-03 3.00E-03

5.00E-03 2.50E-03

4.00E-03 2.00E-03

3.00E-03 1.50E-03

2.00E-03 1.00E-03

1.00E-03 5.00E-04

0.00E+00 0.00E+00
less than 5 6 to 10 11 t o15 16 to 20 Above 20 less than 5 6 to 10 11 t o15 16 to 20 Above 20

Fig. 5. Frequency of total losses by ship group age.

Table 8
Number of fatalities per ship type, time period 2000–2012.

General Cargo Bulk Carriers Fishing vessels Reefer ships Ro-Ro Cargo Car Carriers LPG ships LNG ships Cellular containerships Large Crude Oil, up to 2010
1007 239 174 64 13 16 19 0 34 18

Passenger Ro-Ro Cargo (RoPax) (Pure) passenger ships Cruise ships

2371 575 42
290 E. Eleftheria et al. / Safety Science 85 (2016) 282–292

Table 9
Fatalities per shipyear, time period 2000–2012.

Ship type Fatalities per shipyear Ship type Fatalities per shipyear
Serious cases, fatalities per shipyear
General Cargo 8.51E 03 Passenger Ro-Ro (RoPax) 1.17E 01
Reefer ships 5.92E 03 Passenger vessels 2.20E 02
Bulk Carriers 3.52E 03 Cruise vessels 1.00E 02
Car Carriers 2.45E 03
Ro-Ro Cargo 2.36E 03
Fishing vessels 2.06E 03
LPG ships 1.86E 03
Large Crude Oil 8.87E 04
Cellular containerships 7.54E 04
LNG ships 0.00E 00

Cargo Vessels - Fatalies per shipyear by Accident category Passenger Vessels - Fatalies per shipyear by
6.00E-03 Accident category
5.00E-03 6.00E-02
4.00E-03
3.00E-03
5.00E-02
2.00E-03
1.00E-03
0.00E+00 4.00E-02
Hull /
Fire / Wrecked /
Collision Contact Foundered Mchy. 3.00E-02
Explosion Stranded
Damage
General Cargo 1.78E-03 1.01E-04 3.47E-04 5.04E-03 5.92E-05 7.52E-04 2.00E-02
Bulk Carrier 1.92E-04 0.00E+00 5.90E-04 2.34E-03 8.85E-05 3.10E-04
Car Carriers 1.69E-03 0.00E+00 7.66E-04 0.00E+00 0.00E+00 0.00E+00 1.00E-02
Fishing Vessels 3.32E-04 1.18E-05 2.60E-04 1.01E-03 2.37E-05 2.49E-04
LNG 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00
Hull / Wrecked
LPG 0.00E+00 0.00E+00 1.18E-03 1.96E-04 4.91E-04 0.00E+00 Fire / Foundere
Collision Contact Mchy. /
Explosion d
Reefer Vessels 9.24E-05 0.00E+00 0.00E+00 2.59E-03 0.00E+00 3.24E-03 Damage Stranded
RoRo Cargo 0.00E+00 0.00E+00 7.26E-04 1.63E-03 0.00E+00 0.00E+00 Cruise Vessels 0.00E+00 8.35E-03 4.77E-04 0.00E+00 7.15E-04 4.77E-04
Cellular Containerships 8.87E-05 0.00E+00 3.99E-04 2.44E-04 0.00E+00 2.22E-05 Passenger Vessels 1.72E-03 0.00E+00 3.83E-05 2.02E-02 0.00E+00 0.00E+00
Large Tankers 4.44E-04 0.00E+00 4.44E-04 0.00E+00 0.00E+00 0.00E+00 RoRo Passenger 2.28E-03 4.95E-05 1.77E-02 5.62E-02 0.00E+00 4.11E-02

Fig. 6. Fatalities per shipyears by accident type.

Table 10
Cargo ships, fatalities per shipyear, time period 2000–2012 by group age.

General Cargo Bulk Carrier Car Carriers Fishing vessels LNG LPG Reefer vessels RoRo Cargo
Less than 5 3.24E 03 1.17E 03 4.92E 03 1.23E 03 0.00E+00 8.07E 04 4.87E 02 0.00E+00
6–10 2.54E 03 3.18E 03 0.00E+00 2.84E 03 0.00E+00 0.00E+00 0.00E+00 3.79E 03
11–15 4.95E 03 1.72E 03 0.00E+00 3.92E 03 0.00E+00 9.38E 04 7.50E 03 0.00E+00
16–20 1.01E 02 8.32E 03 1.07E 03 1.31E 03 0.00E+00 0.00E+00 2.38E 03 3.99E 03
Above 20 1.67E 02 6.44E 03 4.29E 03 1.20E 03 0.00E+00 8.62E 03 1.20E 03 3.57E 03

Ro-Ro Passenger ships, as shown in Table 11. Recalling the earlier Table 11
made comment regarding the unique operational conditions and Passenger ships, fatalities per shipyear, time period 2000–2012 by group age.
inferior fitness of Ro-Ro Passenger ships in less developed coun-
Cruise vessels Passenger vessels Passenger RoRo vessels
tries, we note that in the reporting period they were two disastrous
accidents with high number of fatalities, which decisively deter- Less than 5 9.99E 04 0.00E+00 5.13E 04
6–10 3.34E 02 4.29E 02 1.10E 02
mine the values of the shown statistics: the 970 fatalities of the 11–15 0.00E+00 1.68E 04 2.32E 01
‘‘Le Joola”, which capsized off the coast of Gambia in 2002 and 16–20 4.26E 03 7.81E 03 3.14E 02
831 fatalities for the ‘‘Princess of the Stars”, which was lost during Above 20 8.68E 03 5.60E 02 2.89E 01
typhoon Fengshen in Philippines in 2008.
methods possible, what may set off some interesting trends hidden
4. Assessment of uncertainty of timeline analysis in the casualty databases.
In the present study we followed a more conventional way,
A basic problem in the statistical analysis of historical ship acci- namely in order to conclude about the sufficiency of the statistical
dent data is the scarcity and sometime quality of data. Regarding sample in terms of the amount of available historical accident data
the quality of data: the analyst is requested to cross-check acciden- we conducted for each ship type, a Kendall’s Tau non-parametric
tal data through various independent sources and update them, as test of statistical significance regarding the timeline of observed
necessary. events. This allows us to rationally determine a possible increasing
A valuable approach, when investigating historical data of the trend in the number of future accidents. The results of the test are
present type, was presented by Kelangath et al. (2012); it intro- given in Table 12.
duces the use of data-driven Bayesian modelling in risk analysis From the above conducted test, we observe that as far as
and makes a comparison with different data-driven Bayesian the cargo ships fleet is concerned, we note an increasing trend of
 E. Eleftheria et al. / Safety Science 85 (2016) 282–292
Table 12
Kendall’s Tau test for accident timelines per ship type.
Ship type Correlation coefficient
Bulk Carriers .205
Car Carriers .051
Fishing vessels 0.692**
General Cargo 0.590**
LNG .082
LPG .256
RoRo Cargo .282
Reefer vessels .256
Cruise vessels .256
Passenger vessels 0.744**
RoPaX 0.590**
**
Correlation is significant at the 0.01 level (2-tailed).
accidents for the General Cargo ships and the Fishing Vessels.
Referring to ships carrying passengers, Pure Passenger ships and
Passenger Ro-Ro ships (RoPax) also present a statistically signifi-
cant increasing trend. For the remaining ship types, no significant
trend can be concluded from the available/processed data.
5. Conclusions
The present investigation illustrates current levels of risk and
safety for the most popular types of merchant ships in terms of fre-
quencies of ship accidents, of people onboard fatalities and of ship
total losses.
Frequencies related to the occurrence of serious accidents show,
in general, increased values in the last ten years (post 2000) of the
studied period (compared to DNV, 2006), of about 30%, depending
on ship type, except from LPG ships, where exhibited values con-
siderably decreased.
However, focusing on the fatalities per shipyear, the frequency
of loss of life was significantly reduced, except for Reefer vessels
and RoPax ships, for which the fatality rates increased; in these
cases, however, statistics were determined by individual, catas-
trophic type of accidents. This observation of declining fatalities
for the other ship types appears contradicting the generally
increased accident frequency values and indicates that recorded
accidents were obviously less severe in terms of consequences
and/or this was the result of an improved recording practice of ship
accidents by the database provider (less ‘underreporting’, Hassel
et al., 2011).
The investigation of relationships between ship’s age and acci-
dental rates as well as of consequences led to a variety of interest-
ing conclusions; namely, that it is not always straightforward that
an older ship simply suffers more accidents than a younger one, as
sometimes younger ships exhibit more accidents; thus, it is rather
a complicated relationship that is affected by a variety of parame-
ters, like ship type and quality of (new) shipbuilding, the quality
and extent of ship’s maintenance and ship operation (seamanship),
the time of major class surveys and possible change of ownership,
etc. However, clearly an older ship, especially over 20 years of age6
is more prone to total loss accidents.
Among passenger ships, the highest frequency of accidents is
observed for cruise ships, but in terms of fatalities Ro-Ro Passenger
ships are leaders. Among the various types of investigated cargo
ships, the Ro-Ro Cargo and Car Carriers exhibit the highest accident
frequencies, but the General Cargo ships have the highest number
of total losses and of fatalities.
Estimated values of Potential Loss of Life (PLL) by historical data
for RoPax ships present highest values in the last decade, namely
6
Traditionally, the economic life of a cargo ship is set at 20 years, whereas for
passenger ships this often increases to 30 years.
291
1.17E 01 pertaining to all studied accident categories, whereas
values for Cruise and Passenger ships are more or less comparable,
Sig. (2-tailed) namely 1.00E 02 and 2.20E 02 respectively. These data were con-
.329 firmed in a more recent risk analysis study performed in the frame-
.807 work of the EMSA III project (EMSA III, 2013) regarding ships larger
.001 than 80 m, it was shown that the PLL values of Cruise vessels are in
.005 the range of 4.20E 01–5.98E 02 (depending on the ship size),
.705
.222
whereas RoPax ships present PLL values in the range of 1.09E
.180 01–7.12E 01. It is noted that PLL statistics are greatly affected
.222 by individual disastrous accidents that (if they happen) decisively
.222 determine the statistical values.
.000
Recalling a basic definition of safety, namely that safety is a state
.005
of acceptable risk by society, we may herein conclude on the level of
safety of the world fleet in terms of the average level of risk of ship
accidents, thus quantitatively in terms of averaged accident fre-
quencies and of their consequences. In that respect we note an
increasing trend of the risk of accidents for some ship types, like
General Cargo Ships, (large) Fishing vessels, Passenger and Ro-Ro
Passenger ships, whereas for other ship types no significant
changes are observed. Considering the outcome of the DNV
(2006) study, it can be said that in the last decade the frequencies
of ship accidents generally increased (which may be disputed, if
considering the reduced underreporting, the extent of which is
unknown), but the safety level of various ship types did not signif-
icantly change, as the consequences of accidents remained in aver-
age at about the same level. Nevertheless the safety performance of
the various ship types differs and needs special care when aiming
at a continuous improvement of ship safety in the years to come.
Independently, considering that the causes for the overwhelming
majority of ship accidents (especially of the navigational type of
accidents) are related to human factors, which are to a great extent
independent of ship type, efforts for the improvement of maritime
safety need to include research on human factors, improved crew
training schemes and familiarisation with new technologies, crisis
management, etc., even though the quantification of possible
improvements will be difficult to establish in view of the nature
of the safety problem and the complexity of human factors effects
(Harrald et al., 1998).
References
CONTIOPT, Formal Safety Assessment and Multi-Objective Optimization of
Containerships. NTUA-Germanischer Lloyd Bilateral Project, 2011–2013.
DNV, 2006. Level of Safety of Ships in the Period 1999–2004. DNV Report.
Eliopoulou, E., Papanikolaou, A., 2007. Casualty analysis of large tankers. J. Mar. Sci.
Technol. 12 (4), 191–271, November 2007, Springer Japan, ISSN 0948-4280
(Print) 1437-8213 (Online).
Eliopoulou, E., Hamann, R., Papanikolaou, A., Golyshev, P., 2013. Casualty analysis of
cellular container ships. In: Proceedings of the 5th International Maritime
Conference on Design for Safety-4th Workshop on Risk-based Approaches in
the Marine Industries (IDFS 2013), 25–27 November, Shanghai, China.
Eliopoulou, E., Papanikolaou, A., Diamantis, P., Hamann, R., 2012. Analysis of tanker
casualties after OPA90. J. Eng. Marit. Environ., Part M 226 (4), November 2012,
ISSN 1475-0902.
EMSA III (2013–2015). Assessment of the acceptable and practicable risk level of
passenger ships related to damage stability. Project Funded by the European
Maritime Safety Agency.
Goerlandt, F., Montewka, J., 2015. A framework for risk analysis of maritime
transportation systems: a case study for oil spill from tankers in a ship–ship
collision. J. Saf. Sci. http://dx.doi.org/10.1016/j.ssci.2015.02.009.
Hamann, R., Papanikolaou, A., Eliopoulou, E., Golyshev, P., 2013. Assessment of
safety performance of container ships. In: Proceedings of the 5th International
Maritime Conference on Design for Safety-4th Workshop on Risk-based
Approaches in the Marine Industries (IDFS 2013), 25–27 November, Shanghai,
China.
Harrald, J.R., Mazzuchi, T.A., Spahn, J., Van Dorp, R., Merrick, J., Shrestha, S.,
Grabowski, M., 1998. Using system simulation to model the impact of human
error in a maritime system. Saf. Sci. 30, 235–247.
Hassel, M., Asbjørnslett, B.E., Hole, L.P., 2011. Underreporting of maritime accidents
to vessel accident databases. J. Accid. Anal. Prevent. 4 (6), 2053–2063.
Kelangath, S., Purnendu, D., Quigley, J., Hirdaris, S., 2012. Risk analysis of damaged
ships – a data-driven Bayesian approach. Ships Offshore Struct. 7 (3), 333–347.
http://dx.doi.org/10.1080/17445302.2011.592358.
292 E. Eleftheria et al. / Safety Science 85 (2016) 282–292
MEPC 58/INF.2, 2008. FSA – Crude Oil Tankers, Details of the Formal Safety
Assessment. Submitted by Denmark, IMO London, 4 July 2008.
Montewka, J., Ehlers, S., Goerlandt, F., Hinz, T., Tabri, K., Kujala, P., 2014. A
framework for risk assessment for maritime transportation systems – a case
study for open sea collisions involving RoPax vessels. J. Reliab. Eng. Syst. Saf.
http://dx.doi.org/10.1016/j.ress.2013.11.014.
MSC 83/INF.3, 2007. FSA – Liquefied Natural Gas (LNG) Carriers, Details of the
Formal Safety Assessment. Submitted by Denmark, IMO London, 3 July 2007.
MSC 83/INF.8, 2007. FSA – container vessels, Details of the Formal Safety
Assessment. Submitted by Denmark, IMO London, 3 July 2007.
MSC 85/INF.2, 2008. FSA – Cruise Ships, Details of the Formal Safety Assessment.
Submitted by Denmark, IMO London, 21 July 2008.
MSC 85/INF.3, 2008. FSA – RoPax Ships, Details of the Formal Safety Assessment.
Submitted by Denmark, IMO London, 21 July 2008.
MSC/Circ.1023, 2002. ’Guidelines for Formal Safety Assessment (FSA) for use in the
IMO Rule-Making Process. IMO, London, 5 April 2002.
MSC/Circ.1180-MEPC/Circ.474, 2005. Amendments to the Guidelines for Formal
Safety Assessment (FSA) for use in the IMO Rule-Making Process (MSC/
Circ.1023 – MEPC/Circ.392). IMO London, 25 August 2005.
MSC-MEPC.2/Circ.12, 2013. Revised Guidelines for Formal Safety Assessment (FSA)
for Use in the IMO Rule-Making Process. IMO London, 8 July 2013.
Pagiaziti, A., Maliaga, E., Eliopoulou, E., Zaraphonitis, G., Hamann, R., 2015. Statistics
of collision, grounding and contact accidents of passenger and container ships.
Proceedings of the 5th International Symposium on ship Operations,
Management and Economics (SOME), 28–29 May 2015, Athens.
Papanikolaou, A., Bitha, K., Eliopoulou, E., Ventikos, N.P., 2014. Statistical analysis of
ship accidents that occurred in the period 1990–2012 and assessment of safety
level of ship types. In: Proceedings of the Maritime Technology and Engineering
Conference (MARTECH), 15–17th October 2014, Lisboa, Portugal, pp. 227–233
(ISBN 978-1-138-02727-5).
Papanikolaou, A., Eliopoulou, E., 2008. Impact of ship age on tanker accidents. In:
Proceedings of the 2nd International Symposium on Ship Operations,
Management & Economics, SNAME Local Section, Athens.
Papanikolaou, A., Hamann, R., Lee, B.-S., Mains, C., Olufsen, O, Tvedt, E., Vassalos, D.,
Zaraphonitis, G., 2013. GOALDS – goal based damage stability of passenger
ships. In: Transactions of the 2013 Annual Conference of the Society of Naval
Architect and Marine Engineers (SNAME), Seattle.
Papanikolaou, A, Bitner-Gregersen, E., El Moctar, O., Guedes Soares, C., Reddy, R.,
Spengler, F., Shigunov, V., Zaraphonitis, G., 2015. Energy efficient safe ship
operation (SHOPERA). In: Transactions of the 2015 World Maritime Technology
Conference (WMTC2015), Society of Naval Architect and Marine Engineers
(SNAME), Providence.
Skjong, R., 2002. Risk Acceptance Criteria: current proposals and IMO positions.
Proceedings of the Surface Transport Technologies for Sustainable
Development, Valencia, Spain, 4–6 June 2002.
Skjong, R., 2009. Regulatory framework. In: Papanikolaou, A. (Ed.), Risk-Based Ship
Design, Methods, Tools and Applications. Springer-Verlag, Berlin, Heidelberg
(ISBN 978-3-540-89041-9).
SAFEDOR, 2005–2009. Design, Operation and Regulation for Safety. EU Project, FP6-
516278.
Voulgarellis, M., 2015. Statistical Analysis of Serious Accidents in the Period 2000–
2012 (Diploma Thesis). National Technical University of Athens, Ship Design
Laboratory, March 2015 (in Greek).
Wang, J., 2001. The current status and future aspects in formal ship safety
assessment. J. Saf. Sci. 38 (1), 19–30.