casualties in dry bulk shipping (1963–1996)

Marine Policy 26 (2002) 437–450

Casualties in dry bulk shipping (1963–1996)

Stephen E. Robertsa,*, Peter B. Marlowb

aUnit of Health-Care Epidemiology, Department of Public Health, University of Oxford, Old Road, Oxford OX3 7LF, UKbCardiff Business School, Cardiff University, Colum Drive, Cardiff CF10 3EU, UK

Received 24 May 2002; accepted 9 June 2002

Abstract

In recent years, bulk carriers have been identified with high risks of catastrophic structural failure and foundering, and with heavy

loss of human life. This study, based upon Lloyd’s of London casualty records, identified four risk factors that had significant,

independent effect upon the likelihood of a bulk carrier foundering. The risk of foundering increased with the age of the ship, and

was related to the ship’s flag of registration. Most importantly, however, increased risks were found for heavy cargoes of iron ore

and scrap steel or iron, and for trading routes to the Far East and from Europe to North America. Additional safety measures, in

particular regarding ship design and high-risk trades, may well be necessary to reduce the high casualty rates.

r 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Bulk carriers; Foundering; Risk factors; Formal safety assessment

1. Introduction

‘‘God must have been a ship owner. He placed theraw materials far from where they were needed andcovered two thirds of the earth with water.’’ [ErlingNaess]

Bulk shipping has been used for many years to reducethe cost of sea transport and the transport of bulkcargoes is a vital component of international trade. Suchtrades require a sufficient volume of cargo suitable forbulk handling and hence justify a tailored shippingoperation. The five major dry bulk cargoes are coal, ironore, grain, bauxite and alumina, and phosphate rock,and each year the trade in bulks increases. Table 1 showsthe volume of world seaborne trade for 1999, and drybulk carrier vessels account for over 25% of the worldfleet (Table 2).

Bulk carriers have gained notoriety in merchantshipping in recent years on account of their elevatedcasualty records. While all merchant shipping is subjectto certain casualty risks such as collisions with othervessels, fires and explosions on board, and grounding onrocks, large bulk ships have become increasinglyassociated with high risks of catastrophic structural

failure and foundering. The casualty levels and the lossof human life, which typically affect the entire crewswhen large bulk ships suddenly break into two or floodafter loss of shell plating, and sink in minutes or evenseconds, have become far from acceptable.

In many countries and not least Britain, whichsuffered the sudden disappearance of the largest Britishship ever lost at sea (the 200,000 tonnes of displacementore/bulk/oil carrier MV Derbyshire) with all 42 crew andtwo supernumerary wives on board during a typhoonoff Japan in 1980, disasters involving large bulk shipshave been subject to considerable media coverage.In recent years, four other bulk carriers—includingthe sister ship of the Derbyshire, the MV Kowloon

Bridge—also foundered after structural failure in severeweather conditions when registered under the British‘‘second register’’ flags of Gibraltar and Hong Kong.Such disasters, which do not affect other types oflarge merchant ship on anywhere near the same scale,have become too numerous to be attributed simply to‘‘the forces of nature’’.

Several classification societies have carried out tech-nical investigations aimed at establishing the physicalcauses of bulk carrier failures. One study concluded thata primary cause of a high proportion of losses was thefailure of the sides of bulk carriers, caused by acombination of corrosion, physical damage sustainedduring cargo operations, and ship design [3]. This

*Corresponding author. Tel.: +44-1865-226745; fax: +44-1865-

226993.

E-mail address: [email protected] (S.E. Roberts).

0308-597X/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.

PII: S 0 3 0 8 - 5 9 7 X ( 0 2 ) 0 0 0 2 4 - 6

investigation revealed that the majority of ships lostwere over 20 years old and most were carrying iron orewhen they foundered. A second investigation similarlyreported that corrosion or metal fatigue was a majorcausal factor [4] and loss of shell plating has also beenidentified as a common occurrence in bulk carrierfailures [30].

Two more recent reports have also highlighted theincreased risks associated with ageing ships and heavycargoes such as iron ore and steel [5,6]. The latter alsoestimated that if the world bulk carrier fleet could bemaintained at ‘‘under 10 year old standards’’ total losseswould be reduced by 80%. The former concluded thatcorrosion and cracking within the cargo spaces were theprincipal contributing factors in many bulk carrierfailures, and also indicated that other factors—such asover-stressing due to incorrect cargo loading anddamage to hold side structures during cargo dischargeoperations—had also played a part. The InternationalMaritime Organisation (IMO), recognising thatsome losses had occurred due to improper loadingand unloading, issued a code of practice for theseoperations [7].

Following the structural failure of several bulk shipsdeparting from Australian ports during a relativelyshort period of time, between 1989 and 1991, Australia’sHouse of Representatives Standing Committee onTransport, Communications and Infrastructure(HORSCOTCI) investigated the commercial and reg-

ulatory environment of the market in which bulk shipswere operating [8]. They concluded that an underlyingfactor behind the increase in casualties, at that time, wasan ageing world fleet of bulk ships; which had resultedlargely from low freight rates and a disincentive for shipowners to replace ageing tonnage.

The Australian Bureau of Transport and Commu-nications Economics subsequently undertook an analy-sis of failures in 50 large bulk ships—including ninewhich had departed from Australian ports—that suf-fered structural failure between October 1989 andDecember 1991. They identified ships carrying a cargoof iron ore, ageing vessels, flag states with high lossratios and trading routes from, respectively, WesternAustralia to Northern Europe, South Africa to Asia,and Brazil to Asia as independent factors associatedwith increased risk of structural failure [9].1

Most of the lives lost from bulk carrier casualtiesbetween 1978 and 1997 have been identified as havingoccurred in storm force conditions, hurricanes ortyphoons [6]. A separate analysis of bulk ship casualtiesbetween 1990 and 1997, however, reported that only aminority (19%) of the 637 seafaring lives lost were dueto the sinking of a bulk carrier that was directlyattributable to adverse weather [11].

Although the problem of ship safety is a globalconcern it has been suggested that regional differenceshave arisen [12]. The Australian government’s commit-ment to rigorous and regular Port State Controlinspections has prompted a reaction among the AsiaPacific maritime nations. This has improved maritimesafety in that part of the world, and has led to afundamental change in the trading patterns of ageingand lower quality bulk tonnage. It was also argued thatage is an indicator of quality and that older vessels aresubject to speculative buying and selling in the second-hand market. This change of ownership may lead to achange of name and perhaps also a change of shipmanagement. Such changes will usually occur ‘‘in thetwilight years of a vessel’s life and it can be assumed thatit will accelerate the decline of the vessel’s quality and itsoverall safety standards’’.

Tamvakis and Thanopoulou [13] reported that inter-views with brokers suggest that age is a reasonablysignificant criterion used by charterers when evaluatingvessel quality, while the significance of flag or manage-ment is less important. Their analysis of the case forquality uses age as the most prominent factor but alsoincludes flag as a proxy for quality. They concluded that‘‘there are still no clear indications of strong pecuniaryincentives to owners for increasing fleet quality andimproving safety in the dry bulk carrier market through

Table 1

World seaborne trade, 1999

Product % of world trade

Oil 36.9

Dry bulk

Ore 8.0

Coal 9.4

Grain 4.1

Bauxite and alumina 1.0

Phosphate rock 0.6 (Total dry bulk=23.1)

Others (bulk & general cargo) 40.0

Total 100.0

Source: Fearnleys Review, 2000, and World Bulk Trades, 1999 [1,2].

Table 2

World merchant fleet by tonnage category (December 1999)

Type of ship % of world fleet

Oil tanker 28.3

Dry bulk carrier 25.6

General cargo 10.3

Container ship 10.2

Gas carrier 3.3

Ro-Ro ship 4.6

Others, e.g. dredging, fishing 17.7

Total 100.0

Source: Lloyd’s Register of Shipping (2000) [29].

1The loss ratios used in this study were based upon calculations

made by the Institute of London Underwriters for the period 1987–

1991 [10].

S.E. Roberts, P.B. Marlow / Marine Policy 26 (2002) 437–450438

vessel replacement’’, despite indications of a geographi-cal redistribution of larger vessels as noted by McCon-ville and Timmermann [12].

2. Formal safety assessment

Clearly shipping is a business where hazards occurand risks are present. The likelihood (probability) of ahazard developing into an undesirable outcome (con-sequence) is the focus of safety management and safetyregulation. Society attempts to reduce or even eliminatethe risks associated with an activity, and formal safetyassessment (FSA) is a tool that will develop acomprehensive risk management system for (the inter-national shipping) industry. It allows a systematic andproactive view to be taken of ship safety and enablesinformed decisions to be taken on the objective analysisof risk. The IMO has defined FSA as a ‘‘structured andsystematic methodology, aimed at enhancing maritimesafety, including protection of life, health, the marineenvironment and property by using risk and cost–benefitassessments’’. This risk-based approach to safety at seaevolved from Lord Carver’s 1992 Report [14] and wasproposed to the IMO in 1993, since when it has beendeveloped as a tool to assist regulators. The processof FSA consists of the problem definition plus fivesteps [15]:

1. Identification of hazards;2. Assessment of risks associated with the hazards;3. Consideration of alternative ways of managing the

risks identified;4. Cost–benefit assessment of alternative risk manage-

ment options;5. Decisions on which options to select/adopt.

3. Aims of study

Given this background the main aim of the presentstudy was to investigate the effects of various riskfactors on the structural failure or foundering of bulkcarriers. In contrast to the analysis provided by most ofthe previous investigations, a major objective of thisstudy was to use a longer time period of over 30 years. Itwas expected that this larger study, in addition toproviding a useful basis for comparison with previousfindings, would allow for a more extensive investigationinto bulk ship failures. For example, a larger studywould enable a more detailed analysis of the differenttrading routes, cargoes being transported and flag statesbeing used.

Of major importance in this study, also, is the use ofmultivariate logistic regression modelling to assess howthe various factors simultaneously affected the risk of

bulk ship failures, and further to identify the main riskfactors which had underlying and independent effectupon these shipping casualties. These findings should beparticularly useful to those undertaking any FSA ofbulk shipping.

4. Materials and methods

All bulk ship casualties between 1963 and 1996 wereidentified from records published by Lloyd’s MaritimeInformation Service [16]. These casualties included bulkcarriers which foundered or disappeared, grounded onrocks, were involved in collisions or contacts, had fire orexplosions on board or were declared as total con-structive losses as a consequence of various otherdamages by Lloyd’s of London underwriters.

Since one of the aims of this study was to investigatethe effects of risk factors on bulk carrier failures, whichwere en voyage, casualties to bulk carriers at the dockwere excluded. Casualties involving bulk ships, whichwere being towed away to be scrapped, and othercarriers which were no longer in active service, weresimilarly excluded from the analysis. Also, bulk carrierswhich were destroyed, damaged or detained as a resultof military conflict were excluded while casualtiesinvolving bulk ships which were at anchorage wereassumed to be on voyage, unless otherwise specified inLloyd’s records, and included.2

A total of 355 casualties to large bulk ships, whichsatisfied these inclusion criteria, were identified fromLloyd’s records. Thirty-eight casualties resulted fromcollisions and contacts, 69 from fires and explosions onboard, 105 from ships grounding, while 18 were due tovarious other damages. The remaining 125 shipsfoundered or disappeared, either as a result of cata-strophic structural failure or often as a likely conse-quence of structural failure. It is these 125 casualtieswhich provide the basis of the following analysis on thefoundering (or structural failure) of bulk carriers.

In order to establish accurately the role of risk factors,such as various cargoes being carried, trading routesbeing sailed or flag states of registration, ideally it wouldbe necessary to have recorded details of every voyage(world-wide between 1963 and 1996) for those bulkcarriers which were not casualties as well as for thosewhich were. Unfortunately, it is not feasible to obtaininformation on every bulk ship voyage over this longstudy period. Therefore, in order to assess the effects ofthe risk factors upon the foundering of the 125 bulkships, it is necessary to find an alternative basis forcomparative reference.

2Most of the shipping casualties caused by military conflict occurred

in the Persian Gulf during the Iraqi–Iranian war of the early and mid-

1980s.

S.E. Roberts, P.B. Marlow / Marine Policy 26 (2002) 437–450 439

One means is to use the information records on the230 bulk carriers which did not founder, but werecasualties due to other causes, as a basis for comparison.For example, if a cargo of iron ore or a ship over 15years old are considered to be major risk factors forstructural failure and foundering, there is no reason tobelieve that this cargo or an ageing vessel would have astrong influence on whether the ship was involved in acollision with another ship, had an explosion in theengine room, or even grounded on rocks during storms.3

Hence, the voyage data on the 230 casualties from othercauses are used as the basis for reference when modellingthe effects of risk factors on bulk ship failure in the nextsection.

The multivariate logistic regression models wereconstructed using a parsimonious, stepwise refinementapproach in order to identify the model of best fit,comprising only of those risk factors with significant,independent effect upon the bulk ship casualties. The tenmain risk factors investigated in this multivariateanalysis were trading route, type of cargo being carried,flag state, age, size and type of bulk ship, country andtime period in which the vessel was built, and the monthof year and time period in which the disaster occurred.

5. Results

The numbers of casualties involving bulk carriers,which occurred during each year of the study period,1963–1996, are illustrated for each of the five maincauses of disaster (foundering, collision and contact, fireand explosion, grounding and other causes) in Fig. 1.During the 1960s and early 1970s, when the world fleetof bulk ships was still small, the incidence of casualtieswas relatively low. The frequency of disasters increasedsharply during 1973 and continued to increase, with anexpanding world fleet, up until the mid-1980s. While1991 has been the worst year for bulk ship casualties(and foundering), thereafter the casualty levels appear tohave declined gently.

The locations of the 125 bulk carriers that founderedor disappeared are illustrated in Fig. 2.4 This map of theworld shows that large numbers of bulk ships founderedin the North West Pacific off Japan, in the South ChinaSea, in the Bay of Biscay and the North Atlantic, and in

the Arabian Sea. These locations largely refer to theregions of the world in which typhoons, hurricanes orsevere tropical cyclones most frequently occur.5 Inparticular, 22 of the bulk carriers (18%) sank in theNorth East Pacific and most of these had been destinedfor ports in Japan.

It was recorded in Lloyd’s casualty records that nineof the 125 bulk carriers (7%) which foundered ordisappeared were known to have suddenly broken intotwo. A further 32 bulk carriers (23%) were reported tohave foundered after flooding—often affecting theforemost hold(s)—through cracks in the hull, shellplating or holds. One ship foundered after her cargoof timber shifted in heavy seas, four foundered afterflooding of the engine room, one after flooding in thetunnel, four more were reported as having flooded afterhatch cover failure and another after leakage through adamaged keel duct. A further 16 sank after flooding ofholds—of unspecified cause—while it was reported that18 others also suffered catastrophic flooding but inunknown or unrecorded locations. The causes of theremaining 40 ships foundering were not known or notreported in Lloyd’s records although most of thesedisappeared suddenly in typhoons, hurricanes or stormsas a likely consequence of structural failure.

Almost 40% of the 125 bulk ships that founderedwere carrying a cargo of iron ore, while a further 29%were also loaded with heavy cargoes of various othermetals and ores (Table 3). The USA and Canada werethe most frequent countries of departure, Japan andChina were the most common destinations, while themost recurrent trading routes were from North Americato North East Asia (the USA to Japan and Canada toJapan); the MV Derbyshire had been loaded with ironore from Canada to Japan. Over 70% of the 125 bulkcarriers which foundered were over 15 years of age, mosthad been built in Japan, and a majority had beenregistered with flags of convenience and second registers.The weather conditions were storms, gales and/or heavyseas in most cases, while typhoons or hurricanes werereported in 6% of the cases.

The total of 355 casualties (all causes), identified fromLloyd’s records between 1963 and 1996, resulted in therecorded loss of 2038 seafarers (Table 4).

Most of these fatalities (79%) were lost through the125 bulk ships which foundered. The casualty rates forbulk ships foundering (per 10,000 ship-years) tended to

3Although the use of the 230 other casualties is an imperfect means

of reference for comparing the effects of risk factors among the 125

bulk ships which foundered, because of the lack of world-wide voyage

data over the long study period, it offers one means—probably the best

available—for attempting to model the effects of various risk factors

on the structural failure of bulk carriers.4The locations of the bulk carriers which disappeared or foundered

with no subsequent trace of the sunken wreck are based on their last

reported position or the location of wreckage or debris subsequently

identified from the wreck.

5During the 22 year period 1968/69–1989/90 a total of 988 typhoons

(hurricanes or severe tropical cyclones as they are variously labelled in

different regions of the world to denote sustained wind speeds in excess

of 65 knots) occurred worldwide [17]. These most frequently occurred

in the North West Pacific basin (352 or 36%), followed by the North

East Pacific basin (20%), the Atlantic basin, including the North

Atlantic Ocean, the Gulf of Mexico and the Caribbean Sea (12%) and

the North Indian basin, including the Arabian Sea and the Bay of

Bengal (10%).


be highest for flags of convenience such as St. Vincent,Malta and the Bahamas, and other (non-OECD) foreignfleets such as South Korea, and were usually lower formany of the OECD member state flags. Similarly, therates for all casualties were again highest for flags ofconvenience and other (non-OECD) foreign fleets suchas South Korea, Malta, St. Vincent and Cyprus. Most ofthe identified 2038 seafaring fatalities (58%) alsooccurred under flags of convenience and secondregisters.6

Logistic regression modelling was then applied toassess the effects of the various risk factors on the bulkcarrier casualties. The response variable distinguishedbetween, on the one hand, the 125 ships which founderedand, on the other, the 230 casualties from other causes.The results of the bivariate regression analysis of the tenmain risk factors are summarised in Table 5.

Four of the 10 risk factors considered here, whenfitted individually, were found to have a significanteffect upon the bulk carrier casualties. The two mostimportant were the type of cargo being carried and thetrading route. Relative to the reference category for typeof cargo, ‘‘all other cargoes’’, cargoes of scrap iron andsteel, and iron ore were found to be significantlyassociated with the increased likelihood of a ship

0

5

10

15

20

25

30

1963

1965

1968

1971

1973

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

Year

Num

ber

of c

asua

lties

Other

Grounding

Fire & explosion

Collision & contact

Foundered

Fig. 1. The annual numbers of casualties involving bulk carriers by type of casualty (1963–1996).

Fig. 2. World map showing the location of the 125 bulk carriers when they foundered (1963–1996).

6The categorisation of a flag of convenience ship has been taken

from a list of flag of convenience shipping registries, dated 16th June

1997, as compiled by the International Transport Federation [19].


Table 3

Summary statistics for risk factors in relation to the 125 bulk carriers

which foundered (1963–1996)

Risk factor No. (%)

Cargo carried

Iron ore 48 (38)

Other metals and ores 15 (12)

Scrap iron and steel 12 (10)

Steel products 9 (7)

Coal 9 (7)

Timber 5 (4)

Cement 4 (3)

Grain 4 (3)

Other cargoes 19 (15)

Country of departure

USA 19 (15)

Canada 14 (11)

Brazil 12 (10)

Australia 11 (9)

South Africa 7 (6)

India 7 (6)

Philippines 4 (3)

Chile 4 (3)

Belgium 3 (2)

Norway 3 (2)

South Korea 3 (2)

Venezuela 3 (2)

Other countries 35 (28)

Country of destination

Japan 25 (20)

China 12 (10)

USA 11 (9)

Germany 9 (7)

South Korea 8 (6)

Belgium 4 (3)

India 4 (3)

Taiwan 4 (3)

UK 4 (3)

Greece 3 (2)

France 3 (2)

Netherlands 3 (2)

Spain 3 (2)


Trading route

USA–Japan 7 (6)

Canada–Japan 6 (5)

South Africa–China 4 (3)

USA–South Korea 4 (3)

Canada–USA 3 (2)

Australia–Japan 2 (2)

Australia–Netherlands 2 (2)

Australia–UK 2 (2)

Brazil–China 2 (2)

Brazil–Germany 2 (2)

Canada–USA 2 (2)

Chile–Japan 2 (2)

Greece–USA 2 (2)

India–Italy 2 (2)

Norway–Germany 2 (2)

Table 3 (continued)

Risk factor No. (%)

Philippines–Japan 2 (2)

Other trading routes 79 (63)

Flag state

Panama 25 (20)

Greece 17 (14)

Liberia 15 (12)

South Korea 11 (9)

Cyprus 10 (8)

Malta 5 (4)

Bahamas 3 (2)

China (PRC) 3 (2)

Hong Kong 3 (2)

India 3 (2)

Italy 3 (2)

Philippines 3 (2)

St. Vincent 3 (2)

Turkey 3 (2)


Weather conditions

Typhoon/hurricane 8 (6)

Storms/gales/heavy seas 84 (67)

Fine/calm seas 2 (2)

Not known 31 (25)

Age of ship

0–4 years 2 (2)

5–9 years 10 (8)

10–14 years 23 (18)

15–19 years 40 (32)

20–24 years 42 (34)

25+years 8 (6)

Size of ship

o10,000 grt 18 (14)

10,000–14,999 grt 39 (31)

15,000–19,999 grt 16 (13)

20,000–29,999 grt 15 (12)

30,000–49,999 grt 18 (14)

50,000+grt 19 (15)

Type of bulk ship

Bulk carrier 104 (83)

Ore carrier 15 (12)

Ore/oil carrier 4 (3)

Ore/bulk/oil carrier 2 (2)

Country of build

Japan 68 (53)

UK 15 (12)

Germany 9 (7)

Norway 7 (6)

Sweden 5 (4)

Italy 4 (3)

USA 4 (3)

Denmark 2 (2)

Russia 2 (2)



foundering. A cargo of scrap was found to increasethe risk of foundering (as opposed to a casualty fromany other cause) by a factor of 8.1, while the increasedodds were 4.3 for iron ore. Increased but non-significant

odds were identified for steel products and coal (both1.9) and other metals and ores (1.8), while ships inballast were found to be at significantly reduced risk offoundering.

Table 4

Casualty rates by flag state: for bulk carriers which foundered and for all casualties to bulk carriers, 1963–1996

Rank Flag Type

of flagaSize of fleet

(average no. of bulk

carriers, 1968–1996)b

Foundered All casualties Reported total

crew lost

No. Casualty rate per

10,000 ship-years

(1968–1996)b

No. Casualty rate per

10,000 ship-years

(1968–1996)b

1 South Korea Other 92.5 11 (41.0) 18 (67.1) 58

2 St. Vincent FOC 26.8 3 (38.6) 5 (64.4) 0

3 Malta FOC 72.3 5 (23.8) 14 (66.8) 110

4 Bahamas FOC 47.4 3 (21.8) 4 (29.1) 20

5 Turkey OECD 49.5 3 (20.9) 7 (48.7) 19

6 Cyprus FOC 189.5 10 (18.2) 36 (65.5) 125

7 Panama FOC 501.7 25 (17.2) 63 (43.3) 267

8 Hong Kong FOCc/2nd Reg. 70.2 3 (14.7) 4 (19.6) 54

9 Greece OECD 490.9 17 (11.9) 72 (49.9) 181

10 Italy OECD 96.1 3 (10.8) 6 (21.5) 64

11 India Other 97.1 3 (10.7) 9 (32.0) 153

12 Philippines FOCc 108.3 3 (9.6) 5 (15.9) 49

13 Spain OECD 36.7 1 (9.4) 3 (28.2) 4

14 Yugoslavia Other 40.3 1 (8.6) 4 (34.2) 31

15 Liberia FOC 668.9 15 (7.7) 52 (26.3) 483

16 China (PRC) Other 140.7 3 (7.4) 3 (7.4) 32

17 Singapore FOCc 65.1 1 (5.3) 2 (10.6) 33

18 USA OECD 148.5 2 (4.6) 6 (11.6) 71

19 Norway OECD 177.7 2 (3.9) 6 (11.6) 73

20 UK OECD 139.5 1 (2.5) 3 (7.4) 42

21 Japan OECD 381.3 1 (0.9) 2 (1.8) 0

22 Australia OECD 26.6 0 (0.0) 2 (25.9) 7

23 Brazil Other 53.3 0 (0.0) 4 (25.9) 0

24 France OECD 35.9 0 (0.0) 2 (19.2) 7

25 Taiwan Other 38.7 0 (0.0) 2 (17.8) 0

26 Canada OECD 96.1 0 (0.0) 3 (10.8) 0

27 Poland Other 65.5 0 (0.0) 1 (5.3) 0

Other small flagsc

Argentina Other 12.8 0 c 1 c 0

Bermuda FOC/2nd Reg. 11.1 0 c 1 c 33

Chile Other 7.9 1 c 2 c 30

Georgia Other 0.7 1 c 1 c 3

Gibraltar FOC/2nd Reg. 4.7 1 c 1 c 2

GDR Other 13.8 1 c 2 c 0

Indonesia Other 9.7 0 c 1 c 0

Israel Other 7.6 1 c 1 c 24

Maldive Islands Other 1.5 0 c 1 c 11

Mexico Other 3.8 1 c 2 c 1

Netherlands OECD 20.4 0 c 1 c 1

Thailand Other 4.0 2 c 2 c 50

Vanuatu FOC 13.5 1 c 1 c 0

Total 125 355 2038

Flag states are ranked by casualty rate for ships which foundered and then by casualty rate for all casualties.aFOC=flag of convenience; FOC*=partial flag of convenience; OECD=OECD member state flag; 2nd Reg.=2nd. Register flag; other=other

foreign flag. OECD countries are defined by membership of the OECD as at the mid-year of the study period, 1980.bThe annual world fleet statistics used here [18] do not distinguish bulk carriers by flag state before 1968. Hence the fleet sizes and the casualty rates

quoted here refer to the 29-year period 1968–1996.cCasualty rates are provided only for those flags with an average number of registered bulk carriers in excess of 25 per year, or a total of 725 ship

years, during 1968–1996. Casualty rates for smaller fleets are unduly sensitive to single casualties.


Table 5

Bivariate logistic regression analysis of individual risk factors affecting casualties to bulk carriers (1963–1996)

Risk factor Odds ratio 95% Confidence

interval

(1) Cargo (w2 ¼ 77:8 on 8 d.f.; po0:001)All other cargoes Reference category

Scrap iron and steel 8.14** (2.25–29.5)

Iron ore 4.34*** (2.02–9.32)

Steel products 1.88 (0.66–5.36)

Coal 1.88 (0.66–5.36)

Other metals and ores 1.77 (0.72–4.33)

Fertilisers and chemicals 0.85 (0.26–2.75)

Grain and other foods 0.51 (0.18–1.48)

Ballast 0.06* (0.01–0.45)

(2) Trading route (w2 ¼ 54:4 on 15 d.f.; po0:001)All other trading routes Reference category

Australia–Europe 18.93** (2.22–161)

South Africa–North East Asia 16.22* (1.87–141)

North America–North East Asia 5.41*** (2.25–13.0)

Europe–North America 4.33* (1.30–14.4)

Europe–North East Asia 3.60 (0.76–17.2)

North America–Europe 2.40 (0.84–6.87)

South Asia–North East Asia 2.70 (0.92–7.92)

South America–North East Asia 1.80 (0.66–4.89)

South America–Europe 1.35 (0.46–3.96)

South Asia–South Asia 1.35 (0.38–4.86)

South Asia–Europe 1.20 (0.34–4.23)

North America–North America 1.04 (0.34–3.19)

Europe–Europe 0.61 (0.21–1.79)

South America–North America 0.30 (0.04–2.48)

Europe–South Asia 0.19 (0.02–1.54)

(3) Flag state (w2 ¼ 17:2 on 6 d.f.; p ¼ 0:009)OECD countries Reference category

Other FOC and 2nd Registers 5.08*** (1.96–13.2)

Other non-OECD countries 2.32* (1.17–4.60)

Panama 1.78 (0.93–3.42)

Malta 1.51 (0.47–4.84)

Liberia 1.10 (0.53–2.27)

Cyprus 1.04 (0.45–2.41)

(4) Time period (w2 ¼ 11:9 on 4 d.f.; p ¼ 0:018)1963–1974 Reference category

1975–1979 0.67 (0.24–1.88)

1980–1984 1.29 (0.51–3.27)

1985–1989 1.66 (0.66–4.21)

1990–1996 2.16 (0.87–5.32)

(5) Type of bulk ship (w2 ¼ 7:69 on 3 d.f.; not significant)

(6) Country of build (w2 ¼ 8:80 on 4 d.f.; not significant)

(7) Year built (w2 ¼ 7:76 on 5 d.f.; not significant)

(8) Size of ship (w2 ¼ 6:31 on 5 d.f.; not significant)

(9) Month of year (w2 ¼ 9:23 on 11 d.f.; not significant)

(10) Age of ship (w2 ¼ 5:16 on 5 d.f.; not significant)

Risk factors listed in order of decreasing significance.

Notes: * denotes a significant odds ratio at the 5% level, ** at the 1% level and *** at the 0.1% level.


For illustrative purposes, Fig. 3 shows the percentagecauses of all casualties for each of the main types ofcargo. Roughly 70% of the casualties involving shipsthat were carrying cargoes of scrap and iron ore refer toa foundering. By contrast, this was true for only about40% of the bulk carriers which were carrying steelproducts, other metals and ores, and coal, while for theships carrying other typically less dense cargoes of grain,chemicals, fertilisers and ballast the proportions ofcasualties due to a sinking were lower again.

Trading route was also found to be stronglyassociated with increased risks of bulk ships foundering.Significantly high risks were found, when contrastedwith the reference category of all other routes, for tradesfrom, respectively, Australia to Europe, South Africa toNorth East Asia, North America to North East Asia,and from Europe to North America. The high levels offoundering, as opposed to other types of casualty, onthese trading routes are all too apparent in Fig. 4.Increased, but non-significant, risks of foundering werealso found for other trades to North East Asia—fromEurope, South Asia and South America—and illustratethe high risk of failure of bulk carriers bound for theNorth East Asian countries of Japan (in particular),China, South Korea and Taiwan.7

The third risk factor that was found to be associatedwith increased risk of foundering was the flag state of

ship registration. Bulk carriers registered under otherflags of convenience (excluding Panama, Liberia,Cyprus and Malta) and the second registers, alongwith those flying other non-OECD state flags, weresimilarly at increased risk of sinking when contrastedwith those ships registered with OECD countries (seealso Fig. 5). A fourth risk factor, the time period inwhich the disaster occurred, also had a significant effect,with an increased rate of bulk ships foundering duringthe 1980s and 1990s. The other six factors consideredhere offered no significant impact on the type of bulkship casualty.

The respective risk factors were then analysedsimultaneously to determine the regression model ofbest fit, containing only those terms with significant,independent effect upon the bulk carrier casualties. Thefinal model was comprised of the four risk factors—typeof cargo, trading route, flag state and the age of the bulkship (see Fig. 6 for illustration of the relationshipbetween type of casualty and age of ship).

Details of the multivariate model of best fit arepresented in Table 6. Significantly increased, andindependent, risks of foundering were identifiedfor heavy cargoes of iron ore (odds ratio=8.6), scrapiron and steel (10.9), the trading routes from Europeto North America (21.0), from North America toNorth East Asia (8.2) and from South Asia to NorthEast Asia (4.3), and for ships registered with othernon-OECD countries (3.8). The age of the bulk carrier,which was not significant when assessed on its own,was—when considered in conjunction with the cargo,trade and flag state—an important risk factor. Vesselsaged over 15 were found to be at significantly increasedrisk of foundering, while this risk became inflated with

0% 20% 40% 60% 80% 100%

Ballast (49)

Grain & other foods (38)

Fertilisers & chemicals(21)

Other cargoes (52)

Other metals & ores (38)

Coal (22)

Steel products (22)

Iron ore (78)

Scrap iron & steel (16)

Car

go

Percent

Foundered

Collision & contact

Fire & explosion

Grounding

Other

Fig. 3. Percentage cause of all casualties to bulk carriers by cargo being transported (with number of casualties in brackets, 1963–1996).

7For the purposes of this study, North East Asia has been defined as

the ‘‘Far East’’ countries of Japan, China (PRC), North and South

Korea, Taiwan, Hong Kong and North East Russia. South Asia refers

to the rest of Asia. Also, North America has been defined to include

Canada, the USA, Mexico, Newfoundland, Greenland and Bermuda,

while South America refers to the remainder of the (South and

Central) American continent.


increasing age; for bulk ships aged between 15 and 19the increased odds were 4.4 relative to those less than 5years old, the increase was 5.3 for carriers aged between20 and 24, and 6.8 for those aged 25 and over.

Importantly, since the time period in which thecasualty occurred had no significant additional effecton the likelihood of foundering, the above findings arebroadly consistent over the entire study period, 1963–1996. Similarly, since the time period in which the vesselwas built also offered no additional explanation,changes in bulk carrier design over time would seemto have no further significant bearing on the findingsfrom this model.

6. Discussion

This study, based upon Lloyd’s of London casualtyrecords, identified a total of 125 bulk carriers thatfoundered, often as a consequence of structural failure,during the 36-year period 1963–1996. Most of theseships were carrying cargoes of iron ore or other heavycargoes; most were aged over 15 years, while tradingroutes to North East Asia were particularly recurrent.An analysis of casualty rates revealed that bulk carriersregistered with many of the flags of convenience andother non-OECD states were more likely to founderthan those in OECD fleets.

0% 20% 40% 60% 80% 100%

Europe - South Asia (15)

South America - North America (10)

Europe - Europe (27)

All other trades (100)

North America - North America (18)

South Asia - Europe (13)

South Asia - South Asia (12)

South America - Europe (18)

South America - NE Asia (20)

North America - Europe (17)

South Asia - NE Asia (16)

Europe - NE Asia (7)

Europe - North America (13)

North America - NE Asia (30)

South Africa - NE Asia (7)

Australia - Europe (8)

Tra

de

Percent

Foundered

Collision & contact

Fire & explosion

Grounding

Other

Fig. 4. Percentage cause of all casualties to bulk carriers by trading route (with number of casualties in brackets, 1963–1996).

0% 20% 40% 60% 80% 100%

OECD countries (109)

Cyprus (36)

Liberia (52)

Malta (14)

Panama (63)

Other non-OECDcountries (58)

Other FOC & 2ndregisters (23)

Fla

g

Percent

Foundered

Collision & contact

Fire & explosion

Grounding

Other

Fig. 5. Percentage cause of all casualties to bulk carriers by flag state (with number of casualties in brackets, 1963–1996).


Most importantly, multivariate logistic regressionmodelling identified four risk factors which hadsignificant, and independent, influence on the likelihoodof a bulk carrier foundering. These four factors—cargo,trading route, age of ship and flag state—broadlysupport previous findings from a shorter 27-monthAustralian study [9]. Heavy cargoes of iron ore, scrapsteel or iron, trading routes from Europe to NorthAmerica, North America to North East Asia and SouthAsia to North East Asia, ageing bulk carriers in excessof 15 years and registration with other non-OECDcountries were all found to significantly increase theodds of ships sinking in contrast to casualties due toother causes.8

Importantly the increased risks of foundering, asidentified in the multivariate model, for these particularcargoes, trades, flags and age groups of bulk carriersoperate independently of each other. For example, anageing bulk ship bound for North East Asia with acargo of iron ore from North America would be atincreased risk of foundering on all three counts; firstly,due to the trading route being sailed, secondly the typeof cargo being transported and also because of the ageof the vessel.

The detrimental effects of two of these four riskfactors—cargoes of iron ore and ageing ships—on thelikelihood of structural failure have been observed inseveral previous studies or technical investigationsconducted by classification societies.9 In particular theeffects of corrosion over time in older bulk carriers,together with the high stresses placed upon the hullstructures by an especially dense cargo, such as iron ore,have been documented as major causal factors in thestructural failure of many large bulk ships.

Recent regulations, developed by the Maritime SafetyCommittee of the International Maritime Organisation,have attempted to address the findings for these two riskfactors. As from July 1, 1999, bulk carriers over 10 yearsold (and in excess of 150m) were no longer permitted totransport heavy cargoes without a Special Survey underthe Enhanced Survey Programme [22].

Previous technical investigations have also high-lighted damage to hull structures sustained during cargooperations as a contributory factor leading to thestructural failure of the sides of many bulk carriers.These, too, have been addressed by the recent publica-tion of recommendation guidelines for cargo-handlingoperations [5,7,23,24] and by amendments to theInternational Convention for the Safety of Life at Sea[22]. Additional amendments, relating to bulk carriers in

0% 20% 40% 60% 80% 100%

0 - 4 years (19)

5 - 9 years (35)

10 - 14 years (63)

15 - 19 years (113)

20 - 24 years (101)

25+ years (24)

Age

of s

hip

Percent

Foundered

Collision & contact

Fire & explosion

Grounding

Other

Fig. 6. Percentage cause of all casualties to bulk carriers by age of ship (with number of casualties in brackets, 1963–1996).

8 It should be noted that while the reference category of casualties

from all other causes is an imperfect means for modelling the effects of

the risk factors on the foundering of dry bulk ships, because of the

unavailability of complete voyage data over the 36-year study period,

it is probably the best available. Nonetheless, the effect of the age of

the ship is likely to be underestimated in these models. In particular,

older ships are more likely to be written off as total constructive losses

and scrapped, following damages to the vessel, than newer ships where

there is a greater incentive to effect repairs and return the ship to

service.

9These include: Lloyd’s Register of Shipping [3]; Corkhill [20];

Bureau of Transport and Communications Economics [9]; Intercargo

[11] and Stopford [6]. Also, the importance of age in bulk carrier

failures led the Consultant Director of Intercargo to propose

widespread scrapping of older, non-commercially viable ships; as a

means of reducing seafaring fatalities and also to provide aid to

developing countries who need the type of scrap from broken up ships

to help economic growth [21].


excess of 150m in length, concern strengtheningbulkheads—in particular the bulkhead between holds 1and 210—and the hull bottom to withstand hold-floodedconditions, and also the use of monitoring equipment todetect stresses in the ship’s structure.

Despite these recent regulations and guidelines, thesafety of dry bulk shipping remains dubious as bulkcarriers continue to suffer structural failure. In parti-cular, concern continues to centre around theoreticalweaknesses in the design of bulk carriers and theirability to withstand abnormal waves. For example, LordDonaldson concluded in his report into the loss of theDerbyshire that little had been done to address the long-advocated, but unproven, flaws in the design of bulkcarriers and the bridge class vessels [25]. Lord Donald-son has also questioned the ability of bulk carriers to

Table 6

The model of best fit (Multivariate logistic regression analysis of risk factors affecting casualties to bulk ships (1963–1996))

Risk factor b Standard error (b) Odds ratio 95% confidence interval

Intercept �3.53 (0.86)

Cargo

Scrap iron and steel 2.39 (0.88) 10.90** (1.94–61.3)

Iron ore 2.16 (0.55) 8.64*** (2.95–25.3)

Other metals and ores 1.07 (0.57) 2.91 (0.95–8.89)

Steel products 0.88 (0.67) 2.42 (0.65–9.03)

Coal 0.66 (0.67) 1.94 (0.52–7.15)

Fertilisers and chemicals �0.54 (0.76) 0.59 (0.13–2.59)

Grain and other foods �0.97 (0.67) 0.38 (0.10–1.40)

Ballast �3.24 (1.25) 0.04** (0.01–0.45)

Trading route

Europe–North America 3.05 (0.91) 21.04*** (3.57–124)

North America–North East Asia 2.10 (0.59) 8.16*** (2.59–25.7)

South Africa–North East Asia 1.99 (1.34) 7.34 (0.54–101)

Australia–Europe 1.76 (1.26) 5.83 (0.50–68.5)

South Asia–North East Asia 1.45 (0.68) 4.27* (1.12–16.3)

South Asia–South Asia 1.11 (0.80) 3.04 (0.63–14.7)

Europe–North East Asia 1.01 (0.93) 2.74 (0.45–16.8)

North America–Europe 0.89 (0.74) 2.44 (0.57–10.4)

North America–North America 0.39 (0.71) 1.47 (0.37–5.93)

South America–North East Asia 0.10 (0.69) 1.11 (0.29–4.25)

Europe–Europe �0.44 (0.72) 0.65 (0.16–2.64)

South Asia–Europe �0.52 (0.80) 0.60 (0.12–2.88)

South America–Europe �0.56 (0.73) 0.57 (0.14–2.39)

South America–North America �1.70 (1.24) 0.18 (0.02–2.07)

Europe–South Asia �1.90 (1.16) 0.15 (0.02–1.45)

Flag State

Other non-OECD countries 1.32 (0.49) 3.76** (1.43–9.88)

Other FOC and 2nd Registers 1.08 (0.64) 2.93 (0.83–10.4)

Panama 0.76 (0.47) 2.14 (0.85–5.42)

Malta 0.29 (1.06) 1.34 (0.17–10.7)

Cyprus 0.12 (0.58) 1.12 (0.36–3.52)

Liberia �0.29 (0.52) 0.75 (0.27–2.06)

Age of ship

25 years + 1.91 (0.91) 6.77* (1.13–40.4)

20–24 years 1.68 (0.74) 5.35* (1.25–22.8)

15–19 years 1.48 (0.73) 4.38* (1.05–18.4)

10–14 years 0.82 (0.75) 2.26 (0.52–9.91)

5–9 years 0.57 (0.81) 1.76 (0.36–8.55)

Notes: w2 ¼ 145:1 on 34 d.f.; po0:001:* denotes a significant odds ratio at the 5% level, ** at the 1% level and *** at the 0.1% level.

The reference category is a bulk ship aged under 5 years old, registered with an OECD member state, carrying an ‘‘other’’ cargo on an ‘‘other’’ route.

10Technical investigations have revealed that many bulk carriers

sank after the bulkhead between holds 1 and 2 was breached. A study

by the US Marine Administration found that while a typical bulk

carrier should have been able to survive all one-hold flooding, the

flooding of any two holds would have disastrous consequences.


withstand abnormal waves, and the implications for theLoad Line convention [26].11 This may well be reflectedin the finding from this study that certain trading routes(most notably to North East Asia), where abnormalwaves and typhoons are more liable, exert an additional,important effect on the likelihood of bulk carriersfoundering.12

The evidence compiled from this study, along withthat from the similar Australian investigation [9],indicates that additional risk factors, other than oldships and cargoes of iron ore, come into play in affectingthe foundering of bulk ships. Modern tonnage, builtsince 1980, has used a significantly higher percentage ofhigh-tensile steel which has a greater propensity tofracture, although few of these ships would have reached15 years of age by the end of the study period in 1996.This has led to a shorter life expectancy for such vesselsas witnessed by their low second-hand values.

The Australian ‘‘Ships of Shame’’ report [8] identifieda method for determining the highest risk vessels basedon a points system relating to age, ship type andthe number of changes in name, class, and flag.Arguments rage over the relevance of the flag but thereport notes that a ship registered in a flag state with ahigh casualty rate had the same risk of failure as a ship 5years older registered under a flag state with a lowcasualty rate.

The casualty levels and patterns involving bulkcarriers demand close monitoring in the future. Addi-tional regulation, regarding the design of bulk carriersand criteria required for ships to undertake high-risktrades (for example, to Japan and the Far East), maywell be necessary if bulk ship casualties and theconsequent loss of life are to be brought into line withthe more acceptable levels associated with other sectorsof merchant shipping. The analysis contained in thispaper should help to provide an objective identificationof the relevant factors to be considered when under-taking a cost benefit analysis of proposed risk reductionmeasures in the context of FSA.

Acknowledgements

The authors are grateful to Professor Richard Gossfor helpful comments on an earlier draft of this paper.

References

[1] Fearnleys. Review, Fearnley A/S, Oslo, Norway, 2000.

[2] Fearnleys. World Bulk Trades, Fearnley A/S, Oslo, Norway,

2000.

[3] Lloyd’s Register of Shipping. Bulk carriers—the safety issues.

London: Lloyd’s Register of Shipping, 1991.

[4] American Bureau of Shipping. Submission by the American

Bureau of Shipping to the House of Representatives Standing

Committee on Transport, Communications and Infrastructure

Inquiry into Ship Safety, HORSCOTCI Submission no. 39.

Canberra, Australia: HORSCOTCI, 1992.

[5] International Association of Classification Societies. Bulk car-

riers: guidance and information on bulk cargo loading and

discharging to reduce the likelihood of over-stressing the hull

structure. London: International Association of Classification

Societies, 1997.

[6] Stopford M. Bulk carriers – in the dock or in the dry dock? The

Baltic, July/August 1998. p. 66–68.

[7] International Maritime Organisation, BLU Code. Code of

practice for the safe loading and unloading of bulk carriers:

1998 edition (IMO-266E). London: International Maritime

Organisation, 1998.

[8] HORSCOTCI. Ships of shame, inquiry into ship safety.

Canberra, Australia: Australian Government Publishing Service,

1992.

[9] Bureau of Transport and Communications Economics. Structural

failure of large bulk ships. Report 85. Canberra, Australia:

Australian Government Publishing Service, 1994.

[10] Commission of European Communities. A common policy on

safe seas. Communication from the Commission, COM(93)66

final, 24 February, 1993. Brussels, Belgium: Commission of the

European Communities, 1993.

[11] Intercargo. Analysis of total loss and fatality statistics for dry

bulk carriers, 1990–1997. London: Intercargo, 1997.

[12] McConville J, Timmermann KW. An analysis of the quality and

redistribution of dry capesize tonnage. Maritime Policy and

Management 1996;23(1):45–53.

[13] Tamvakis M, Thanopoulou HA. Does quality pay? The case of

the dry bulk market. Transportation Research Part E: Logistics

and Transportation Review 2000;36(4):297–307.

[14] Carver M. Safety aspects of ship design and technology. House of

Lords Committee Report. London: HMSO, 1992.

[15] Peachey JH. Overview of FSA: overview of the five steps of FSA.

Maritime Safety Committee 65th Session: Seminar on Formal

Safety Assessment. London: IMO, 1995.

[16] Hooke N. Maritime casualties, 1963–1996: 2nd ed. London:

Lloyd’s of London Press, 1997.

[17] Landsea CW. FAQ: hurricanes, typhoons and tropical cyclones.

Miami, USA: NOAA AOML/Hurricane Research Division,

1996.

[18] Lloyd’s Register of Shipping. World fleet statistics (annually,

1968–1996). London: Lloyd’s Register of Shipping, 1969–1997.

[19] International Transport Federation. List of flags of convenience.

17th June 1997. London: International Transport Federation,

1997.

[20] Corkhill M. Going down? Hazardous Cargo Bulletin

1991;12(11):5.

[21] Farthing B. Scrapping: the self-help solution. Marine Engineers

Review, October 1992.

[22] International Maritime Organisation. International convention

for the safety of life at sea (SOLAS, 1999 edition) (IMO-158E).

London: International Maritime Organisation, 1999.

[23] International Chamber of Shipping. Ship shore safety checklist

for loading and unloading dry bulk cargo carriers. London:

International Chamber of Shipping, 1996.

11Another study provides evidence that bulk carriers are vulnerable

to structural failure because of inadequate design detail, the use of

construction materials without guaranteed fracture toughness and the

serious limitations of current inspection practices both at the

construction stage and with the bulk carrier in service [27]. See also

Grey [28].12While the North West Pacific is the area of the world in which

typhoons, hurricanes or severe cyclone systems occur most frequently,

they are also recognized as being the most severe in this region in terms

of central pressure and wind speeds.


[24] International Association of Classification Societies. Bulk car-

riers—handle with care. London: International Association of

Classification Societies, 1998.

[25] Department of the Environment, Transport and the Regions.

M.V. Derbyshire Surveys: UK/EC Assessors’ Report. Department

of the Environment, Transport and the Regions, London, 1998.

[26] Lloyd’s List. Donaldson urges load line rethink (Modern bulk

carrier design provides insufficient strength against abnormal

waves, necessitating urgent review of the International Load

Line convention, said Lord Donaldson at the annual dinner on

the UK Chamber of Shipping), Lloyd’s List, 26 October 1995.

[27] Jubb J. Structural failures of bulk carriers. Proceedings of the

Institution of Mechanical Engineers, Part E—Journal of Process

Mechanical Engineering 1995;209(E2):83–91.

[28] Grey M. Welding expert criticises bulker design details. Lloyd’s

Casualty Week, 3 February 1995.

[29] Lloyd’s Register of Shipping. World fleet statistics, 1999. London:

Lloyd’s Register of Shipping, 2000.

[30] Nippon Kaiji Kyokai. Study report on bulk carrier losses. Tokyo,

Japan: Nippon Kaiji Kyokai, 1992.


casualties in dry bulk shipping (1963–1996)

Documents